{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 3.0,
  "eval_steps": 500,
  "global_step": 230403,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.0,
      "grad_norm": 3.333552122116089,
      "learning_rate": 0.0005999999999721121,
      "loss": 10.4494,
      "step": 1
    },
    {
      "epoch": 0.0,
      "grad_norm": 3.0125420093536377,
      "learning_rate": 0.0005999999998884487,
      "loss": 9.6611,
      "step": 2
    },
    {
      "epoch": 0.0,
      "grad_norm": 2.6661605834960938,
      "learning_rate": 0.0005999999997490094,
      "loss": 9.1093,
      "step": 3
    },
    {
      "epoch": 0.0,
      "grad_norm": 2.2665178775787354,
      "learning_rate": 0.0005999999995537946,
      "loss": 8.7966,
      "step": 4
    },
    {
      "epoch": 0.0,
      "grad_norm": 2.258434534072876,
      "learning_rate": 0.0005999999993028042,
      "loss": 8.358,
      "step": 5
    },
    {
      "epoch": 0.0,
      "grad_norm": 2.2307076454162598,
      "learning_rate": 0.000599999998996038,
      "loss": 7.9822,
      "step": 6
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.9850260019302368,
      "learning_rate": 0.0005999999986334962,
      "loss": 7.7379,
      "step": 7
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.9015101194381714,
      "learning_rate": 0.0005999999982151787,
      "loss": 7.464,
      "step": 8
    },
    {
      "epoch": 0.0,
      "grad_norm": 2.131283760070801,
      "learning_rate": 0.0005999999977410855,
      "loss": 6.7768,
      "step": 9
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.7408180236816406,
      "learning_rate": 0.0005999999972112169,
      "loss": 6.8904,
      "step": 10
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.938135027885437,
      "learning_rate": 0.0005999999966255724,
      "loss": 6.3359,
      "step": 11
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.1402738094329834,
      "learning_rate": 0.0005999999959841522,
      "loss": 6.5391,
      "step": 12
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.4963619709014893,
      "learning_rate": 0.0005999999952869564,
      "loss": 5.8945,
      "step": 13
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.167161226272583,
      "learning_rate": 0.0005999999945339849,
      "loss": 6.1773,
      "step": 14
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.1107890605926514,
      "learning_rate": 0.0005999999937252379,
      "loss": 5.9208,
      "step": 15
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.0124188661575317,
      "learning_rate": 0.0005999999928607151,
      "loss": 5.8527,
      "step": 16
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.5532121658325195,
      "learning_rate": 0.0005999999919404167,
      "loss": 6.0626,
      "step": 17
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.7631114721298218,
      "learning_rate": 0.0005999999909643426,
      "loss": 5.9585,
      "step": 18
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.9131847023963928,
      "learning_rate": 0.0005999999899324929,
      "loss": 5.9013,
      "step": 19
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.874756932258606,
      "learning_rate": 0.0005999999888448674,
      "loss": 5.7454,
      "step": 20
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.9496322274208069,
      "learning_rate": 0.0005999999877014663,
      "loss": 6.0503,
      "step": 21
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.0668278932571411,
      "learning_rate": 0.0005999999865022896,
      "loss": 5.5091,
      "step": 22
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.9393265843391418,
      "learning_rate": 0.0005999999852473372,
      "loss": 5.9098,
      "step": 23
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.7112236618995667,
      "learning_rate": 0.0005999999839366092,
      "loss": 5.6226,
      "step": 24
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.0036046504974365,
      "learning_rate": 0.0005999999825701055,
      "loss": 5.6878,
      "step": 25
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.8598136305809021,
      "learning_rate": 0.0005999999811478261,
      "loss": 5.878,
      "step": 26
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.8182787895202637,
      "learning_rate": 0.000599999979669771,
      "loss": 5.6233,
      "step": 27
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.9063557982444763,
      "learning_rate": 0.0005999999781359403,
      "loss": 5.5198,
      "step": 28
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.9363874197006226,
      "learning_rate": 0.000599999976546334,
      "loss": 5.3355,
      "step": 29
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.8684234619140625,
      "learning_rate": 0.000599999974900952,
      "loss": 5.3375,
      "step": 30
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.2803339958190918,
      "learning_rate": 0.0005999999731997943,
      "loss": 5.1944,
      "step": 31
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.8992094397544861,
      "learning_rate": 0.000599999971442861,
      "loss": 5.3419,
      "step": 32
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.0520493984222412,
      "learning_rate": 0.0005999999696301519,
      "loss": 5.2319,
      "step": 33
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.2816475629806519,
      "learning_rate": 0.0005999999677616674,
      "loss": 5.4843,
      "step": 34
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.17536199092865,
      "learning_rate": 0.000599999965837407,
      "loss": 5.7219,
      "step": 35
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.8263903260231018,
      "learning_rate": 0.000599999963857371,
      "loss": 5.503,
      "step": 36
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.1010477542877197,
      "learning_rate": 0.0005999999618215595,
      "loss": 5.3992,
      "step": 37
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.9924551844596863,
      "learning_rate": 0.0005999999597299722,
      "loss": 5.4081,
      "step": 38
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.9753313660621643,
      "learning_rate": 0.0005999999575826093,
      "loss": 5.288,
      "step": 39
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.4642199277877808,
      "learning_rate": 0.0005999999553794707,
      "loss": 5.5393,
      "step": 40
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.0110118389129639,
      "learning_rate": 0.0005999999531205564,
      "loss": 5.1807,
      "step": 41
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.8805031776428223,
      "learning_rate": 0.0005999999508058666,
      "loss": 5.2338,
      "step": 42
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.6441406607627869,
      "learning_rate": 0.000599999948435401,
      "loss": 5.3066,
      "step": 43
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.7667922973632812,
      "learning_rate": 0.0005999999460091598,
      "loss": 5.1773,
      "step": 44
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.0281957387924194,
      "learning_rate": 0.0005999999435271431,
      "loss": 5.1125,
      "step": 45
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.8171626329421997,
      "learning_rate": 0.0005999999409893505,
      "loss": 4.8928,
      "step": 46
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.7831050157546997,
      "learning_rate": 0.0005999999383957823,
      "loss": 5.1374,
      "step": 47
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.8651368021965027,
      "learning_rate": 0.0005999999357464385,
      "loss": 5.0475,
      "step": 48
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.8686742186546326,
      "learning_rate": 0.000599999933041319,
      "loss": 5.0763,
      "step": 49
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.7638999819755554,
      "learning_rate": 0.0005999999302804239,
      "loss": 5.0373,
      "step": 50
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.7715578079223633,
      "learning_rate": 0.0005999999274637532,
      "loss": 4.9814,
      "step": 51
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.9218789339065552,
      "learning_rate": 0.0005999999245913068,
      "loss": 4.7862,
      "step": 52
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.9035757184028625,
      "learning_rate": 0.0005999999216630848,
      "loss": 5.0168,
      "step": 53
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.7926275730133057,
      "learning_rate": 0.000599999918679087,
      "loss": 4.9168,
      "step": 54
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.8810080885887146,
      "learning_rate": 0.0005999999156393138,
      "loss": 4.9919,
      "step": 55
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.9466562867164612,
      "learning_rate": 0.0005999999125437647,
      "loss": 5.0432,
      "step": 56
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.2732425928115845,
      "learning_rate": 0.0005999999093924401,
      "loss": 4.7769,
      "step": 57
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.7813524007797241,
      "learning_rate": 0.0005999999061853398,
      "loss": 5.0463,
      "step": 58
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.7570122480392456,
      "learning_rate": 0.0005999999029224638,
      "loss": 5.0313,
      "step": 59
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.9066333770751953,
      "learning_rate": 0.0005999998996038123,
      "loss": 4.7012,
      "step": 60
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.0237386226654053,
      "learning_rate": 0.000599999896229385,
      "loss": 5.0439,
      "step": 61
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.8569740653038025,
      "learning_rate": 0.0005999998927991822,
      "loss": 4.6381,
      "step": 62
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.9928956627845764,
      "learning_rate": 0.0005999998893132037,
      "loss": 5.028,
      "step": 63
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.7721015810966492,
      "learning_rate": 0.0005999998857714495,
      "loss": 4.9611,
      "step": 64
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.750993013381958,
      "learning_rate": 0.0005999998821739197,
      "loss": 4.9725,
      "step": 65
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.7299620509147644,
      "learning_rate": 0.0005999998785206143,
      "loss": 5.1691,
      "step": 66
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.8687590956687927,
      "learning_rate": 0.0005999998748115331,
      "loss": 4.5921,
      "step": 67
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.8385632634162903,
      "learning_rate": 0.0005999998710466765,
      "loss": 4.9595,
      "step": 68
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.7929795980453491,
      "learning_rate": 0.0005999998672260442,
      "loss": 4.6858,
      "step": 69
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.7431840896606445,
      "learning_rate": 0.0005999998633496361,
      "loss": 5.0343,
      "step": 70
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.7247875332832336,
      "learning_rate": 0.0005999998594174525,
      "loss": 4.9526,
      "step": 71
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.902977466583252,
      "learning_rate": 0.0005999998554294932,
      "loss": 4.7576,
      "step": 72
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.7638068795204163,
      "learning_rate": 0.0005999998513857583,
      "loss": 5.0271,
      "step": 73
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.712199866771698,
      "learning_rate": 0.0005999998472862478,
      "loss": 4.7281,
      "step": 74
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.6579356789588928,
      "learning_rate": 0.0005999998431309616,
      "loss": 4.7431,
      "step": 75
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.687681257724762,
      "learning_rate": 0.0005999998389198997,
      "loss": 4.6848,
      "step": 76
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.6668212413787842,
      "learning_rate": 0.0005999998346530623,
      "loss": 4.6858,
      "step": 77
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.7390393614768982,
      "learning_rate": 0.0005999998303304493,
      "loss": 4.8809,
      "step": 78
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.7560957670211792,
      "learning_rate": 0.0005999998259520606,
      "loss": 4.7483,
      "step": 79
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.8142073750495911,
      "learning_rate": 0.0005999998215178963,
      "loss": 4.5482,
      "step": 80
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.7179496884346008,
      "learning_rate": 0.0005999998170279563,
      "loss": 4.4757,
      "step": 81
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.6329760551452637,
      "learning_rate": 0.0005999998124822407,
      "loss": 4.7223,
      "step": 82
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.0728511810302734,
      "learning_rate": 0.0005999998078807494,
      "loss": 4.6706,
      "step": 83
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.7416367530822754,
      "learning_rate": 0.0005999998032234826,
      "loss": 4.6683,
      "step": 84
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.8202375769615173,
      "learning_rate": 0.0005999997985104401,
      "loss": 4.8167,
      "step": 85
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.9597409963607788,
      "learning_rate": 0.0005999997937416221,
      "loss": 4.7757,
      "step": 86
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.8239755034446716,
      "learning_rate": 0.0005999997889170283,
      "loss": 4.545,
      "step": 87
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.8220269680023193,
      "learning_rate": 0.000599999784036659,
      "loss": 4.9843,
      "step": 88
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.8184282779693604,
      "learning_rate": 0.000599999779100514,
      "loss": 4.6641,
      "step": 89
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.8332849144935608,
      "learning_rate": 0.0005999997741085935,
      "loss": 4.6073,
      "step": 90
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.7201291918754578,
      "learning_rate": 0.0005999997690608971,
      "loss": 4.851,
      "step": 91
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.6965822577476501,
      "learning_rate": 0.0005999997639574254,
      "loss": 4.7383,
      "step": 92
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.8310244083404541,
      "learning_rate": 0.0005999997587981779,
      "loss": 4.6532,
      "step": 93
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.7917149066925049,
      "learning_rate": 0.0005999997535831548,
      "loss": 4.5295,
      "step": 94
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.8012035489082336,
      "learning_rate": 0.0005999997483123561,
      "loss": 4.3412,
      "step": 95
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.7652168869972229,
      "learning_rate": 0.0005999997429857819,
      "loss": 4.5907,
      "step": 96
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.7819420099258423,
      "learning_rate": 0.0005999997376034318,
      "loss": 4.6278,
      "step": 97
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.663823127746582,
      "learning_rate": 0.0005999997321653063,
      "loss": 4.3942,
      "step": 98
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.7135194540023804,
      "learning_rate": 0.0005999997266714053,
      "loss": 4.4949,
      "step": 99
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.0346275568008423,
      "learning_rate": 0.0005999997211217285,
      "loss": 4.4113,
      "step": 100
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.7200731635093689,
      "learning_rate": 0.0005999997155162762,
      "loss": 4.8211,
      "step": 101
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.7358216047286987,
      "learning_rate": 0.0005999997098550482,
      "loss": 4.3419,
      "step": 102
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.7837105989456177,
      "learning_rate": 0.0005999997041380445,
      "loss": 4.4106,
      "step": 103
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.7734938859939575,
      "learning_rate": 0.0005999996983652653,
      "loss": 4.5234,
      "step": 104
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.7373958826065063,
      "learning_rate": 0.0005999996925367105,
      "loss": 4.596,
      "step": 105
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.7748330235481262,
      "learning_rate": 0.0005999996866523801,
      "loss": 4.4755,
      "step": 106
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.860731840133667,
      "learning_rate": 0.0005999996807122741,
      "loss": 4.7049,
      "step": 107
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.7402271032333374,
      "learning_rate": 0.0005999996747163924,
      "loss": 4.6659,
      "step": 108
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.8090092539787292,
      "learning_rate": 0.0005999996686647353,
      "loss": 4.5649,
      "step": 109
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.7566320896148682,
      "learning_rate": 0.0005999996625573025,
      "loss": 4.7587,
      "step": 110
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.7610787749290466,
      "learning_rate": 0.0005999996563940941,
      "loss": 4.5039,
      "step": 111
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.7685642242431641,
      "learning_rate": 0.00059999965017511,
      "loss": 4.3545,
      "step": 112
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.8931743502616882,
      "learning_rate": 0.0005999996439003504,
      "loss": 4.5626,
      "step": 113
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.8313314318656921,
      "learning_rate": 0.0005999996375698151,
      "loss": 4.2116,
      "step": 114
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.7205467224121094,
      "learning_rate": 0.0005999996311835044,
      "loss": 4.5657,
      "step": 115
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.7530926465988159,
      "learning_rate": 0.0005999996247414179,
      "loss": 4.7273,
      "step": 116
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.8243007659912109,
      "learning_rate": 0.0005999996182435559,
      "loss": 4.2392,
      "step": 117
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.7954682111740112,
      "learning_rate": 0.0005999996116899184,
      "loss": 4.3932,
      "step": 118
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.666382372379303,
      "learning_rate": 0.0005999996050805051,
      "loss": 4.8211,
      "step": 119
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.8117617964744568,
      "learning_rate": 0.0005999995984153164,
      "loss": 4.7143,
      "step": 120
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.7416430711746216,
      "learning_rate": 0.000599999591694352,
      "loss": 4.3619,
      "step": 121
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.7674328088760376,
      "learning_rate": 0.0005999995849176121,
      "loss": 4.7796,
      "step": 122
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.8263363838195801,
      "learning_rate": 0.0005999995780850965,
      "loss": 4.7101,
      "step": 123
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.7288904190063477,
      "learning_rate": 0.0005999995711968054,
      "loss": 4.7295,
      "step": 124
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.8394801020622253,
      "learning_rate": 0.0005999995642527388,
      "loss": 4.2279,
      "step": 125
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.8499646782875061,
      "learning_rate": 0.0005999995572528964,
      "loss": 4.5696,
      "step": 126
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.7309108972549438,
      "learning_rate": 0.0005999995501972786,
      "loss": 4.4376,
      "step": 127
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.7102392315864563,
      "learning_rate": 0.0005999995430858851,
      "loss": 4.5335,
      "step": 128
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.9125083684921265,
      "learning_rate": 0.0005999995359187161,
      "loss": 4.2661,
      "step": 129
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.7575920224189758,
      "learning_rate": 0.0005999995286957716,
      "loss": 4.4227,
      "step": 130
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.7038982510566711,
      "learning_rate": 0.0005999995214170514,
      "loss": 4.4793,
      "step": 131
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.8566485643386841,
      "learning_rate": 0.0005999995140825556,
      "loss": 4.318,
      "step": 132
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.7665771245956421,
      "learning_rate": 0.0005999995066922843,
      "loss": 4.323,
      "step": 133
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.770192563533783,
      "learning_rate": 0.0005999994992462374,
      "loss": 4.464,
      "step": 134
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.8051774501800537,
      "learning_rate": 0.000599999491744415,
      "loss": 4.2779,
      "step": 135
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.98851078748703,
      "learning_rate": 0.0005999994841868169,
      "loss": 4.3226,
      "step": 136
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.6750397682189941,
      "learning_rate": 0.0005999994765734433,
      "loss": 4.5229,
      "step": 137
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.0412522554397583,
      "learning_rate": 0.0005999994689042941,
      "loss": 4.0347,
      "step": 138
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.7410924434661865,
      "learning_rate": 0.0005999994611793695,
      "loss": 4.2102,
      "step": 139
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.9931344985961914,
      "learning_rate": 0.0005999994533986691,
      "loss": 4.163,
      "step": 140
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.7699488401412964,
      "learning_rate": 0.0005999994455621933,
      "loss": 4.3001,
      "step": 141
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.7801388502120972,
      "learning_rate": 0.0005999994376699418,
      "loss": 4.2232,
      "step": 142
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.7909395098686218,
      "learning_rate": 0.0005999994297219149,
      "loss": 4.627,
      "step": 143
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.7093692421913147,
      "learning_rate": 0.0005999994217181124,
      "loss": 4.5127,
      "step": 144
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.9182920455932617,
      "learning_rate": 0.0005999994136585343,
      "loss": 4.316,
      "step": 145
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.8089715242385864,
      "learning_rate": 0.0005999994055431807,
      "loss": 4.4281,
      "step": 146
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.7842017412185669,
      "learning_rate": 0.0005999993973720514,
      "loss": 4.2821,
      "step": 147
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.8799046874046326,
      "learning_rate": 0.0005999993891451467,
      "loss": 4.1099,
      "step": 148
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.7562128305435181,
      "learning_rate": 0.0005999993808624664,
      "loss": 4.1675,
      "step": 149
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.7839937210083008,
      "learning_rate": 0.0005999993725240106,
      "loss": 4.28,
      "step": 150
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.787366509437561,
      "learning_rate": 0.0005999993641297792,
      "loss": 4.4061,
      "step": 151
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.8872716426849365,
      "learning_rate": 0.0005999993556797724,
      "loss": 4.5024,
      "step": 152
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.7903965711593628,
      "learning_rate": 0.0005999993471739899,
      "loss": 4.1439,
      "step": 153
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.7767014503479004,
      "learning_rate": 0.0005999993386124319,
      "loss": 4.144,
      "step": 154
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.9595321416854858,
      "learning_rate": 0.0005999993299950982,
      "loss": 4.267,
      "step": 155
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.8256951570510864,
      "learning_rate": 0.0005999993213219892,
      "loss": 4.2313,
      "step": 156
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.735489010810852,
      "learning_rate": 0.0005999993125931045,
      "loss": 4.3421,
      "step": 157
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.7574724555015564,
      "learning_rate": 0.0005999993038084444,
      "loss": 4.5054,
      "step": 158
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.8281086087226868,
      "learning_rate": 0.0005999992949680087,
      "loss": 4.4342,
      "step": 159
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.7908215522766113,
      "learning_rate": 0.0005999992860717975,
      "loss": 4.4727,
      "step": 160
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.8033297061920166,
      "learning_rate": 0.0005999992771198108,
      "loss": 4.1816,
      "step": 161
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.8187950849533081,
      "learning_rate": 0.0005999992681120486,
      "loss": 3.8764,
      "step": 162
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.7656470537185669,
      "learning_rate": 0.0005999992590485107,
      "loss": 4.5305,
      "step": 163
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.8067736029624939,
      "learning_rate": 0.0005999992499291973,
      "loss": 4.4553,
      "step": 164
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.8471931219100952,
      "learning_rate": 0.0005999992407541084,
      "loss": 4.0909,
      "step": 165
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.9011847972869873,
      "learning_rate": 0.0005999992315232441,
      "loss": 4.001,
      "step": 166
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.6745538115501404,
      "learning_rate": 0.0005999992222366042,
      "loss": 4.3251,
      "step": 167
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.708890974521637,
      "learning_rate": 0.0005999992128941888,
      "loss": 4.1073,
      "step": 168
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.7389917969703674,
      "learning_rate": 0.0005999992034959978,
      "loss": 4.3223,
      "step": 169
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.7193297743797302,
      "learning_rate": 0.0005999991940420313,
      "loss": 4.1784,
      "step": 170
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.9975218772888184,
      "learning_rate": 0.0005999991845322893,
      "loss": 4.5262,
      "step": 171
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.7643300294876099,
      "learning_rate": 0.0005999991749667719,
      "loss": 4.1415,
      "step": 172
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.7935953736305237,
      "learning_rate": 0.0005999991653454789,
      "loss": 4.2138,
      "step": 173
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.71882164478302,
      "learning_rate": 0.0005999991556684104,
      "loss": 4.3205,
      "step": 174
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.7693049907684326,
      "learning_rate": 0.0005999991459355664,
      "loss": 4.2405,
      "step": 175
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.7514294981956482,
      "learning_rate": 0.000599999136146947,
      "loss": 4.4423,
      "step": 176
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.8154879212379456,
      "learning_rate": 0.000599999126302552,
      "loss": 4.1658,
      "step": 177
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.8674474358558655,
      "learning_rate": 0.0005999991164023814,
      "loss": 4.1979,
      "step": 178
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.7639039158821106,
      "learning_rate": 0.0005999991064464354,
      "loss": 4.3907,
      "step": 179
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.8473486304283142,
      "learning_rate": 0.0005999990964347139,
      "loss": 4.3083,
      "step": 180
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.7374647259712219,
      "learning_rate": 0.0005999990863672169,
      "loss": 4.1341,
      "step": 181
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.8613146543502808,
      "learning_rate": 0.0005999990762439445,
      "loss": 4.1869,
      "step": 182
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.772564709186554,
      "learning_rate": 0.0005999990660648964,
      "loss": 4.5549,
      "step": 183
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.7726752758026123,
      "learning_rate": 0.0005999990558300729,
      "loss": 3.9745,
      "step": 184
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.8878213763237,
      "learning_rate": 0.000599999045539474,
      "loss": 3.9536,
      "step": 185
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.131377100944519,
      "learning_rate": 0.0005999990351930996,
      "loss": 4.249,
      "step": 186
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.771456778049469,
      "learning_rate": 0.0005999990247909496,
      "loss": 4.0005,
      "step": 187
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.0678298473358154,
      "learning_rate": 0.0005999990143330243,
      "loss": 4.1223,
      "step": 188
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.7170795202255249,
      "learning_rate": 0.0005999990038193234,
      "loss": 4.1916,
      "step": 189
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.8546355962753296,
      "learning_rate": 0.000599998993249847,
      "loss": 4.0031,
      "step": 190
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.7311908006668091,
      "learning_rate": 0.0005999989826245951,
      "loss": 4.1031,
      "step": 191
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.7769582271575928,
      "learning_rate": 0.0005999989719435679,
      "loss": 4.1298,
      "step": 192
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.7440436482429504,
      "learning_rate": 0.000599998961206765,
      "loss": 4.5072,
      "step": 193
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.7252722978591919,
      "learning_rate": 0.0005999989504141867,
      "loss": 4.2883,
      "step": 194
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.8554401993751526,
      "learning_rate": 0.000599998939565833,
      "loss": 4.0917,
      "step": 195
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.7796382904052734,
      "learning_rate": 0.0005999989286617039,
      "loss": 3.9819,
      "step": 196
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.7006290555000305,
      "learning_rate": 0.0005999989177017992,
      "loss": 4.3074,
      "step": 197
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.9136213660240173,
      "learning_rate": 0.000599998906686119,
      "loss": 4.0748,
      "step": 198
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.7587248682975769,
      "learning_rate": 0.0005999988956146636,
      "loss": 4.0868,
      "step": 199
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.8188908100128174,
      "learning_rate": 0.0005999988844874325,
      "loss": 4.0865,
      "step": 200
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.9946153163909912,
      "learning_rate": 0.0005999988733044261,
      "loss": 3.8526,
      "step": 201
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.7716606855392456,
      "learning_rate": 0.000599998862065644,
      "loss": 4.214,
      "step": 202
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.8663153648376465,
      "learning_rate": 0.0005999988507710866,
      "loss": 4.0626,
      "step": 203
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.7735663652420044,
      "learning_rate": 0.0005999988394207537,
      "loss": 3.9767,
      "step": 204
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.7176697254180908,
      "learning_rate": 0.0005999988280146455,
      "loss": 3.9202,
      "step": 205
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.8405921459197998,
      "learning_rate": 0.0005999988165527618,
      "loss": 3.9496,
      "step": 206
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.9574785232543945,
      "learning_rate": 0.0005999988050351026,
      "loss": 4.0072,
      "step": 207
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.8448621034622192,
      "learning_rate": 0.000599998793461668,
      "loss": 3.9732,
      "step": 208
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.0649324655532837,
      "learning_rate": 0.000599998781832458,
      "loss": 3.7756,
      "step": 209
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.7856062650680542,
      "learning_rate": 0.0005999987701474725,
      "loss": 3.996,
      "step": 210
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.9608957171440125,
      "learning_rate": 0.0005999987584067115,
      "loss": 3.9272,
      "step": 211
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.7891876101493835,
      "learning_rate": 0.0005999987466101752,
      "loss": 4.2277,
      "step": 212
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.9935110211372375,
      "learning_rate": 0.0005999987347578633,
      "loss": 4.0094,
      "step": 213
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.9191291928291321,
      "learning_rate": 0.0005999987228497762,
      "loss": 3.9878,
      "step": 214
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.8609839677810669,
      "learning_rate": 0.0005999987108859135,
      "loss": 3.8923,
      "step": 215
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.8229352831840515,
      "learning_rate": 0.0005999986988662754,
      "loss": 3.8771,
      "step": 216
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.846924364566803,
      "learning_rate": 0.0005999986867908619,
      "loss": 4.0059,
      "step": 217
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.7673807144165039,
      "learning_rate": 0.0005999986746596731,
      "loss": 4.2627,
      "step": 218
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.9532973170280457,
      "learning_rate": 0.0005999986624727088,
      "loss": 4.0677,
      "step": 219
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.8173874020576477,
      "learning_rate": 0.000599998650229969,
      "loss": 3.8127,
      "step": 220
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.8483481407165527,
      "learning_rate": 0.0005999986379314539,
      "loss": 4.0488,
      "step": 221
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.8377381563186646,
      "learning_rate": 0.0005999986255771632,
      "loss": 4.0478,
      "step": 222
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.8230250477790833,
      "learning_rate": 0.0005999986131670973,
      "loss": 4.0358,
      "step": 223
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.8938882350921631,
      "learning_rate": 0.0005999986007012559,
      "loss": 4.0823,
      "step": 224
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.8089737892150879,
      "learning_rate": 0.0005999985881796392,
      "loss": 3.529,
      "step": 225
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.9510349035263062,
      "learning_rate": 0.000599998575602247,
      "loss": 4.0584,
      "step": 226
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.7390662431716919,
      "learning_rate": 0.0005999985629690795,
      "loss": 4.1496,
      "step": 227
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.9014082551002502,
      "learning_rate": 0.0005999985502801365,
      "loss": 3.8433,
      "step": 228
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.7124630808830261,
      "learning_rate": 0.0005999985375354181,
      "loss": 4.0519,
      "step": 229
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.091424584388733,
      "learning_rate": 0.0005999985247349244,
      "loss": 3.986,
      "step": 230
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.7602275013923645,
      "learning_rate": 0.0005999985118786552,
      "loss": 4.0946,
      "step": 231
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.9239240288734436,
      "learning_rate": 0.0005999984989666107,
      "loss": 3.8782,
      "step": 232
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.7362231016159058,
      "learning_rate": 0.0005999984859987907,
      "loss": 4.1638,
      "step": 233
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.9121220111846924,
      "learning_rate": 0.0005999984729751955,
      "loss": 4.062,
      "step": 234
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.7567428946495056,
      "learning_rate": 0.0005999984598958248,
      "loss": 4.2326,
      "step": 235
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.0525256395339966,
      "learning_rate": 0.0005999984467606788,
      "loss": 3.8297,
      "step": 236
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.7714121341705322,
      "learning_rate": 0.0005999984335697573,
      "loss": 4.2193,
      "step": 237
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.7656370401382446,
      "learning_rate": 0.0005999984203230606,
      "loss": 3.9494,
      "step": 238
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.6706433892250061,
      "learning_rate": 0.0005999984070205884,
      "loss": 4.1056,
      "step": 239
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.8330416083335876,
      "learning_rate": 0.0005999983936623408,
      "loss": 3.8755,
      "step": 240
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.748309850692749,
      "learning_rate": 0.000599998380248318,
      "loss": 4.2586,
      "step": 241
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.0891811847686768,
      "learning_rate": 0.0005999983667785196,
      "loss": 4.1471,
      "step": 242
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.8477956056594849,
      "learning_rate": 0.000599998353252946,
      "loss": 4.05,
      "step": 243
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.110763430595398,
      "learning_rate": 0.0005999983396715971,
      "loss": 3.8906,
      "step": 244
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.0184152126312256,
      "learning_rate": 0.0005999983260344728,
      "loss": 3.9448,
      "step": 245
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.3893234729766846,
      "learning_rate": 0.0005999983123415731,
      "loss": 3.9975,
      "step": 246
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.7482725977897644,
      "learning_rate": 0.000599998298592898,
      "loss": 3.6385,
      "step": 247
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.7943481206893921,
      "learning_rate": 0.0005999982847884477,
      "loss": 4.0492,
      "step": 248
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.7221561074256897,
      "learning_rate": 0.0005999982709282219,
      "loss": 3.6964,
      "step": 249
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.0153038501739502,
      "learning_rate": 0.0005999982570122209,
      "loss": 3.8891,
      "step": 250
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.6832186579704285,
      "learning_rate": 0.0005999982430404445,
      "loss": 4.0163,
      "step": 251
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.7646830677986145,
      "learning_rate": 0.0005999982290128928,
      "loss": 4.2014,
      "step": 252
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.7842199802398682,
      "learning_rate": 0.0005999982149295657,
      "loss": 3.9892,
      "step": 253
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.8045599460601807,
      "learning_rate": 0.0005999982007904633,
      "loss": 4.0738,
      "step": 254
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.8179710507392883,
      "learning_rate": 0.0005999981865955856,
      "loss": 3.6867,
      "step": 255
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.813129186630249,
      "learning_rate": 0.0005999981723449325,
      "loss": 4.1183,
      "step": 256
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.8819712996482849,
      "learning_rate": 0.0005999981580385041,
      "loss": 3.6523,
      "step": 257
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.6587932109832764,
      "learning_rate": 0.0005999981436763004,
      "loss": 3.8462,
      "step": 258
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.1874860525131226,
      "learning_rate": 0.0005999981292583214,
      "loss": 3.7143,
      "step": 259
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.9212256669998169,
      "learning_rate": 0.0005999981147845671,
      "loss": 3.5998,
      "step": 260
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.0413103103637695,
      "learning_rate": 0.0005999981002550374,
      "loss": 3.9064,
      "step": 261
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.9262663125991821,
      "learning_rate": 0.0005999980856697324,
      "loss": 3.7613,
      "step": 262
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.9080905914306641,
      "learning_rate": 0.0005999980710286523,
      "loss": 4.2497,
      "step": 263
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.9112027287483215,
      "learning_rate": 0.0005999980563317967,
      "loss": 3.9914,
      "step": 264
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.8161675333976746,
      "learning_rate": 0.0005999980415791658,
      "loss": 3.9959,
      "step": 265
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.8202124238014221,
      "learning_rate": 0.0005999980267707597,
      "loss": 4.3425,
      "step": 266
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.8519237637519836,
      "learning_rate": 0.0005999980119065782,
      "loss": 4.0895,
      "step": 267
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.6978029012680054,
      "learning_rate": 0.0005999979969866214,
      "loss": 4.0256,
      "step": 268
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.7663387060165405,
      "learning_rate": 0.0005999979820108893,
      "loss": 3.9053,
      "step": 269
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.713042676448822,
      "learning_rate": 0.000599997966979382,
      "loss": 4.1687,
      "step": 270
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.0364189147949219,
      "learning_rate": 0.0005999979518920994,
      "loss": 4.1294,
      "step": 271
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.8039187788963318,
      "learning_rate": 0.0005999979367490415,
      "loss": 3.9559,
      "step": 272
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.7119114398956299,
      "learning_rate": 0.0005999979215502083,
      "loss": 3.8975,
      "step": 273
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.853126049041748,
      "learning_rate": 0.0005999979062955999,
      "loss": 4.0797,
      "step": 274
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.8641420006752014,
      "learning_rate": 0.0005999978909852162,
      "loss": 3.8625,
      "step": 275
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.7527245283126831,
      "learning_rate": 0.0005999978756190572,
      "loss": 4.0649,
      "step": 276
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.9469287395477295,
      "learning_rate": 0.000599997860197123,
      "loss": 3.8055,
      "step": 277
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.9605068564414978,
      "learning_rate": 0.0005999978447194133,
      "loss": 3.9229,
      "step": 278
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.7984328866004944,
      "learning_rate": 0.0005999978291859285,
      "loss": 3.662,
      "step": 279
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.8527603149414062,
      "learning_rate": 0.0005999978135966685,
      "loss": 3.9469,
      "step": 280
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.7995588183403015,
      "learning_rate": 0.0005999977979516333,
      "loss": 4.1256,
      "step": 281
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.7414883375167847,
      "learning_rate": 0.0005999977822508226,
      "loss": 3.9734,
      "step": 282
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.875241219997406,
      "learning_rate": 0.0005999977664942368,
      "loss": 4.0136,
      "step": 283
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.7620292901992798,
      "learning_rate": 0.0005999977506818758,
      "loss": 4.0074,
      "step": 284
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.8181167840957642,
      "learning_rate": 0.0005999977348137395,
      "loss": 3.9534,
      "step": 285
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.7822670340538025,
      "learning_rate": 0.0005999977188898279,
      "loss": 3.9229,
      "step": 286
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.9908952713012695,
      "learning_rate": 0.0005999977029101411,
      "loss": 3.7652,
      "step": 287
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.732564389705658,
      "learning_rate": 0.0005999976868746791,
      "loss": 4.0016,
      "step": 288
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.8784609436988831,
      "learning_rate": 0.0005999976707834418,
      "loss": 3.9222,
      "step": 289
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.8418092131614685,
      "learning_rate": 0.0005999976546364293,
      "loss": 3.7851,
      "step": 290
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.7377402186393738,
      "learning_rate": 0.0005999976384336416,
      "loss": 3.8977,
      "step": 291
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.7650694847106934,
      "learning_rate": 0.0005999976221750787,
      "loss": 3.9808,
      "step": 292
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.8385749459266663,
      "learning_rate": 0.0005999976058607406,
      "loss": 3.7989,
      "step": 293
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.7984728217124939,
      "learning_rate": 0.0005999975894906272,
      "loss": 3.7798,
      "step": 294
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.854062020778656,
      "learning_rate": 0.0005999975730647386,
      "loss": 3.9629,
      "step": 295
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.8632822036743164,
      "learning_rate": 0.0005999975565830748,
      "loss": 3.9141,
      "step": 296
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.7573164105415344,
      "learning_rate": 0.0005999975400456357,
      "loss": 3.2957,
      "step": 297
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.7972484827041626,
      "learning_rate": 0.0005999975234524215,
      "loss": 3.8319,
      "step": 298
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.7826972603797913,
      "learning_rate": 0.0005999975068034321,
      "loss": 3.8478,
      "step": 299
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.7916067838668823,
      "learning_rate": 0.0005999974900986675,
      "loss": 3.7277,
      "step": 300
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.7413652539253235,
      "learning_rate": 0.0005999974733381277,
      "loss": 3.9409,
      "step": 301
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.7667040228843689,
      "learning_rate": 0.0005999974565218127,
      "loss": 3.8751,
      "step": 302
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.8275686502456665,
      "learning_rate": 0.0005999974396497225,
      "loss": 3.9546,
      "step": 303
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.7988513112068176,
      "learning_rate": 0.0005999974227218571,
      "loss": 3.8617,
      "step": 304
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.7592062950134277,
      "learning_rate": 0.0005999974057382166,
      "loss": 3.8938,
      "step": 305
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.8250828981399536,
      "learning_rate": 0.0005999973886988008,
      "loss": 4.254,
      "step": 306
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.9370415806770325,
      "learning_rate": 0.0005999973716036098,
      "loss": 3.8241,
      "step": 307
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.8374484777450562,
      "learning_rate": 0.0005999973544526437,
      "loss": 3.7645,
      "step": 308
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.8091897964477539,
      "learning_rate": 0.0005999973372459024,
      "loss": 3.8512,
      "step": 309
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.7614434361457825,
      "learning_rate": 0.0005999973199833859,
      "loss": 3.8882,
      "step": 310
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.9586446285247803,
      "learning_rate": 0.0005999973026650944,
      "loss": 3.7711,
      "step": 311
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.7224688529968262,
      "learning_rate": 0.0005999972852910277,
      "loss": 3.7228,
      "step": 312
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.7942538261413574,
      "learning_rate": 0.0005999972678611857,
      "loss": 3.8023,
      "step": 313
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.7562121152877808,
      "learning_rate": 0.0005999972503755686,
      "loss": 3.9084,
      "step": 314
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.6715697050094604,
      "learning_rate": 0.0005999972328341764,
      "loss": 3.8767,
      "step": 315
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.7766149640083313,
      "learning_rate": 0.0005999972152370091,
      "loss": 3.8714,
      "step": 316
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.8340709805488586,
      "learning_rate": 0.0005999971975840665,
      "loss": 4.0641,
      "step": 317
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.8251932263374329,
      "learning_rate": 0.0005999971798753488,
      "loss": 3.5047,
      "step": 318
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.8136237263679504,
      "learning_rate": 0.000599997162110856,
      "loss": 3.7408,
      "step": 319
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.8365110158920288,
      "learning_rate": 0.0005999971442905881,
      "loss": 3.6432,
      "step": 320
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.7311885356903076,
      "learning_rate": 0.0005999971264145451,
      "loss": 3.7998,
      "step": 321
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.8833811283111572,
      "learning_rate": 0.0005999971084827268,
      "loss": 3.8156,
      "step": 322
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.7664169073104858,
      "learning_rate": 0.0005999970904951334,
      "loss": 3.7276,
      "step": 323
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.8380867838859558,
      "learning_rate": 0.0005999970724517651,
      "loss": 3.9853,
      "step": 324
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.8069133162498474,
      "learning_rate": 0.0005999970543526215,
      "loss": 4.1316,
      "step": 325
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.7328994870185852,
      "learning_rate": 0.0005999970361977027,
      "loss": 4.0979,
      "step": 326
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.8523638248443604,
      "learning_rate": 0.000599997017987009,
      "loss": 3.6221,
      "step": 327
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.752654492855072,
      "learning_rate": 0.0005999969997205402,
      "loss": 3.817,
      "step": 328
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.8595878481864929,
      "learning_rate": 0.0005999969813982961,
      "loss": 4.1093,
      "step": 329
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.8009824156761169,
      "learning_rate": 0.000599996963020277,
      "loss": 3.8201,
      "step": 330
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.7829863429069519,
      "learning_rate": 0.0005999969445864827,
      "loss": 3.6963,
      "step": 331
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.7743197083473206,
      "learning_rate": 0.0005999969260969135,
      "loss": 3.6129,
      "step": 332
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.0758576393127441,
      "learning_rate": 0.000599996907551569,
      "loss": 3.8515,
      "step": 333
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.2601102590560913,
      "learning_rate": 0.0005999968889504496,
      "loss": 3.593,
      "step": 334
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.9484147429466248,
      "learning_rate": 0.000599996870293555,
      "loss": 3.4145,
      "step": 335
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.02406907081604,
      "learning_rate": 0.0005999968515808853,
      "loss": 3.7773,
      "step": 336
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.116937518119812,
      "learning_rate": 0.0005999968328124406,
      "loss": 3.6571,
      "step": 337
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.2057361602783203,
      "learning_rate": 0.0005999968139882209,
      "loss": 3.8514,
      "step": 338
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.8934628963470459,
      "learning_rate": 0.000599996795108226,
      "loss": 3.5395,
      "step": 339
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.8088117837905884,
      "learning_rate": 0.000599996776172456,
      "loss": 3.941,
      "step": 340
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.1190721988677979,
      "learning_rate": 0.0005999967571809111,
      "loss": 4.0012,
      "step": 341
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.9041601419448853,
      "learning_rate": 0.0005999967381335909,
      "loss": 3.7472,
      "step": 342
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.813223123550415,
      "learning_rate": 0.0005999967190304958,
      "loss": 3.7391,
      "step": 343
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.8350166082382202,
      "learning_rate": 0.0005999966998716256,
      "loss": 3.7628,
      "step": 344
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.7902444005012512,
      "learning_rate": 0.0005999966806569804,
      "loss": 3.947,
      "step": 345
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.8452714085578918,
      "learning_rate": 0.0005999966613865602,
      "loss": 3.4809,
      "step": 346
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.8953971266746521,
      "learning_rate": 0.0005999966420603648,
      "loss": 3.7414,
      "step": 347
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.7923241853713989,
      "learning_rate": 0.0005999966226783945,
      "loss": 3.7222,
      "step": 348
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.741895854473114,
      "learning_rate": 0.0005999966032406492,
      "loss": 3.8374,
      "step": 349
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.9918361306190491,
      "learning_rate": 0.0005999965837471286,
      "loss": 3.9397,
      "step": 350
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.8582453727722168,
      "learning_rate": 0.0005999965641978333,
      "loss": 3.7296,
      "step": 351
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.9850597381591797,
      "learning_rate": 0.0005999965445927628,
      "loss": 3.6227,
      "step": 352
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.7966710925102234,
      "learning_rate": 0.0005999965249319174,
      "loss": 3.9418,
      "step": 353
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.4104052782058716,
      "learning_rate": 0.0005999965052152969,
      "loss": 3.6746,
      "step": 354
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.7288115620613098,
      "learning_rate": 0.0005999964854429014,
      "loss": 3.9069,
      "step": 355
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.8794954419136047,
      "learning_rate": 0.0005999964656147308,
      "loss": 3.6724,
      "step": 356
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.1800158023834229,
      "learning_rate": 0.0005999964457307853,
      "loss": 3.8622,
      "step": 357
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.8757210969924927,
      "learning_rate": 0.0005999964257910647,
      "loss": 3.6982,
      "step": 358
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.9500802755355835,
      "learning_rate": 0.0005999964057955692,
      "loss": 3.6769,
      "step": 359
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.753205418586731,
      "learning_rate": 0.0005999963857442987,
      "loss": 3.7279,
      "step": 360
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.7589293718338013,
      "learning_rate": 0.0005999963656372532,
      "loss": 3.4468,
      "step": 361
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.888909637928009,
      "learning_rate": 0.0005999963454744327,
      "loss": 3.91,
      "step": 362
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.8411450386047363,
      "learning_rate": 0.0005999963252558371,
      "loss": 3.8926,
      "step": 363
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.9326704144477844,
      "learning_rate": 0.0005999963049814666,
      "loss": 3.7358,
      "step": 364
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.9271461963653564,
      "learning_rate": 0.0005999962846513213,
      "loss": 3.5749,
      "step": 365
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.8107596635818481,
      "learning_rate": 0.0005999962642654008,
      "loss": 3.8761,
      "step": 366
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.9131306409835815,
      "learning_rate": 0.0005999962438237055,
      "loss": 3.7349,
      "step": 367
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.8247781991958618,
      "learning_rate": 0.0005999962233262351,
      "loss": 3.566,
      "step": 368
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.8403633236885071,
      "learning_rate": 0.0005999962027729898,
      "loss": 3.5404,
      "step": 369
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.786941409111023,
      "learning_rate": 0.0005999961821639696,
      "loss": 3.8145,
      "step": 370
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.7587957382202148,
      "learning_rate": 0.0005999961614991743,
      "loss": 3.9315,
      "step": 371
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.8550770282745361,
      "learning_rate": 0.0005999961407786042,
      "loss": 3.8403,
      "step": 372
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.8719338178634644,
      "learning_rate": 0.000599996120002259,
      "loss": 3.7824,
      "step": 373
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.7530505061149597,
      "learning_rate": 0.000599996099170139,
      "loss": 3.652,
      "step": 374
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.9939268231391907,
      "learning_rate": 0.000599996078282244,
      "loss": 3.7269,
      "step": 375
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.8265697360038757,
      "learning_rate": 0.0005999960573385741,
      "loss": 3.6934,
      "step": 376
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.9210556149482727,
      "learning_rate": 0.0005999960363391292,
      "loss": 3.4884,
      "step": 377
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.7605462670326233,
      "learning_rate": 0.0005999960152839094,
      "loss": 3.5169,
      "step": 378
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.0550994873046875,
      "learning_rate": 0.0005999959941729148,
      "loss": 3.7197,
      "step": 379
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.829499363899231,
      "learning_rate": 0.000599995973006145,
      "loss": 3.8554,
      "step": 380
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.7981818914413452,
      "learning_rate": 0.0005999959517836005,
      "loss": 3.8077,
      "step": 381
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.8845255374908447,
      "learning_rate": 0.0005999959305052811,
      "loss": 3.5118,
      "step": 382
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.7828789353370667,
      "learning_rate": 0.0005999959091711867,
      "loss": 3.4766,
      "step": 383
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.878169596195221,
      "learning_rate": 0.0005999958877813174,
      "loss": 3.7767,
      "step": 384
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.8058963418006897,
      "learning_rate": 0.0005999958663356733,
      "loss": 3.4015,
      "step": 385
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.9147908687591553,
      "learning_rate": 0.0005999958448342543,
      "loss": 4.0314,
      "step": 386
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.9118794798851013,
      "learning_rate": 0.0005999958232770603,
      "loss": 3.7199,
      "step": 387
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.0075081586837769,
      "learning_rate": 0.0005999958016640915,
      "loss": 3.7279,
      "step": 388
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.8901219367980957,
      "learning_rate": 0.0005999957799953478,
      "loss": 3.9284,
      "step": 389
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.8271854519844055,
      "learning_rate": 0.0005999957582708292,
      "loss": 3.4827,
      "step": 390
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.307999610900879,
      "learning_rate": 0.0005999957364905357,
      "loss": 3.7311,
      "step": 391
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.8893154859542847,
      "learning_rate": 0.0005999957146544674,
      "loss": 3.8998,
      "step": 392
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.8755370378494263,
      "learning_rate": 0.0005999956927626243,
      "loss": 3.6244,
      "step": 393
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.8464742302894592,
      "learning_rate": 0.0005999956708150062,
      "loss": 3.7593,
      "step": 394
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.846300482749939,
      "learning_rate": 0.0005999956488116132,
      "loss": 3.8366,
      "step": 395
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.0124757289886475,
      "learning_rate": 0.0005999956267524455,
      "loss": 3.7045,
      "step": 396
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.7593119144439697,
      "learning_rate": 0.0005999956046375028,
      "loss": 3.7782,
      "step": 397
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.8873269557952881,
      "learning_rate": 0.0005999955824667854,
      "loss": 3.8599,
      "step": 398
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.7938565611839294,
      "learning_rate": 0.000599995560240293,
      "loss": 3.8148,
      "step": 399
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.7284159064292908,
      "learning_rate": 0.0005999955379580259,
      "loss": 3.7529,
      "step": 400
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.7894511818885803,
      "learning_rate": 0.000599995515619984,
      "loss": 3.7489,
      "step": 401
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.7982641458511353,
      "learning_rate": 0.0005999954932261671,
      "loss": 3.7271,
      "step": 402
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.7279492020606995,
      "learning_rate": 0.0005999954707765755,
      "loss": 3.5125,
      "step": 403
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.720663845539093,
      "learning_rate": 0.000599995448271209,
      "loss": 3.6575,
      "step": 404
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.7680631279945374,
      "learning_rate": 0.0005999954257100678,
      "loss": 3.6058,
      "step": 405
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.7662671804428101,
      "learning_rate": 0.0005999954030931517,
      "loss": 3.6059,
      "step": 406
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.781071126461029,
      "learning_rate": 0.0005999953804204607,
      "loss": 3.5807,
      "step": 407
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.7409411668777466,
      "learning_rate": 0.000599995357691995,
      "loss": 3.7459,
      "step": 408
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.7960779666900635,
      "learning_rate": 0.0005999953349077546,
      "loss": 3.8467,
      "step": 409
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.7565190196037292,
      "learning_rate": 0.0005999953120677392,
      "loss": 3.9305,
      "step": 410
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.7831207513809204,
      "learning_rate": 0.0005999952891719493,
      "loss": 3.7241,
      "step": 411
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.9333382248878479,
      "learning_rate": 0.0005999952662203844,
      "loss": 3.5932,
      "step": 412
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.7413186430931091,
      "learning_rate": 0.0005999952432130447,
      "loss": 3.5792,
      "step": 413
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.8361467719078064,
      "learning_rate": 0.0005999952201499302,
      "loss": 3.7104,
      "step": 414
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.8094792366027832,
      "learning_rate": 0.000599995197031041,
      "loss": 3.7898,
      "step": 415
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.8415712118148804,
      "learning_rate": 0.0005999951738563771,
      "loss": 3.5405,
      "step": 416
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.7576629519462585,
      "learning_rate": 0.0005999951506259383,
      "loss": 3.4549,
      "step": 417
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.7692533135414124,
      "learning_rate": 0.0005999951273397248,
      "loss": 3.5408,
      "step": 418
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.8644924163818359,
      "learning_rate": 0.0005999951039977365,
      "loss": 3.7295,
      "step": 419
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.1317819356918335,
      "learning_rate": 0.0005999950805999735,
      "loss": 3.6114,
      "step": 420
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.770646870136261,
      "learning_rate": 0.0005999950571464358,
      "loss": 3.5293,
      "step": 421
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.5251834392547607,
      "learning_rate": 0.0005999950336371232,
      "loss": 3.6775,
      "step": 422
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.8126622438430786,
      "learning_rate": 0.0005999950100720361,
      "loss": 3.6203,
      "step": 423
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.1604665517807007,
      "learning_rate": 0.000599994986451174,
      "loss": 3.5964,
      "step": 424
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.8971484899520874,
      "learning_rate": 0.0005999949627745374,
      "loss": 3.8157,
      "step": 425
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.1843843460083008,
      "learning_rate": 0.0005999949390421258,
      "loss": 3.673,
      "step": 426
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.9539657235145569,
      "learning_rate": 0.0005999949152539398,
      "loss": 3.6842,
      "step": 427
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.8402116298675537,
      "learning_rate": 0.0005999948914099788,
      "loss": 3.5329,
      "step": 428
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.8994260430335999,
      "learning_rate": 0.0005999948675102433,
      "loss": 3.384,
      "step": 429
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.8572306036949158,
      "learning_rate": 0.0005999948435547329,
      "loss": 3.8649,
      "step": 430
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.9629313945770264,
      "learning_rate": 0.0005999948195434479,
      "loss": 3.4402,
      "step": 431
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.8295775055885315,
      "learning_rate": 0.0005999947954763883,
      "loss": 3.5932,
      "step": 432
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.8208048939704895,
      "learning_rate": 0.0005999947713535539,
      "loss": 3.6218,
      "step": 433
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.8195542097091675,
      "learning_rate": 0.0005999947471749447,
      "loss": 3.6888,
      "step": 434
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.8548824191093445,
      "learning_rate": 0.000599994722940561,
      "loss": 3.4665,
      "step": 435
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.9219136834144592,
      "learning_rate": 0.0005999946986504025,
      "loss": 3.8647,
      "step": 436
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.023024082183838,
      "learning_rate": 0.0005999946743044693,
      "loss": 3.6138,
      "step": 437
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.8213322758674622,
      "learning_rate": 0.0005999946499027616,
      "loss": 3.812,
      "step": 438
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.9470049142837524,
      "learning_rate": 0.0005999946254452791,
      "loss": 3.4677,
      "step": 439
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.8388521075248718,
      "learning_rate": 0.0005999946009320219,
      "loss": 3.4416,
      "step": 440
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.7315393090248108,
      "learning_rate": 0.0005999945763629902,
      "loss": 3.6981,
      "step": 441
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.1449923515319824,
      "learning_rate": 0.0005999945517381837,
      "loss": 3.5115,
      "step": 442
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.8009401559829712,
      "learning_rate": 0.0005999945270576026,
      "loss": 3.5912,
      "step": 443
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.9234108328819275,
      "learning_rate": 0.0005999945023212468,
      "loss": 3.6897,
      "step": 444
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.0602726936340332,
      "learning_rate": 0.0005999944775291165,
      "loss": 3.6758,
      "step": 445
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.7882094979286194,
      "learning_rate": 0.0005999944526812114,
      "loss": 3.7757,
      "step": 446
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.11075758934021,
      "learning_rate": 0.0005999944277775318,
      "loss": 3.7561,
      "step": 447
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.9040165543556213,
      "learning_rate": 0.0005999944028180775,
      "loss": 3.714,
      "step": 448
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.2047001123428345,
      "learning_rate": 0.0005999943778028485,
      "loss": 3.5859,
      "step": 449
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.1528342962265015,
      "learning_rate": 0.0005999943527318451,
      "loss": 3.1302,
      "step": 450
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.8241983652114868,
      "learning_rate": 0.000599994327605067,
      "loss": 3.7709,
      "step": 451
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.7803109288215637,
      "learning_rate": 0.0005999943024225142,
      "loss": 3.4905,
      "step": 452
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.9820301532745361,
      "learning_rate": 0.0005999942771841868,
      "loss": 3.4507,
      "step": 453
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.8873933553695679,
      "learning_rate": 0.0005999942518900848,
      "loss": 3.8286,
      "step": 454
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.8788480162620544,
      "learning_rate": 0.0005999942265402083,
      "loss": 3.7387,
      "step": 455
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.8301413059234619,
      "learning_rate": 0.0005999942011345572,
      "loss": 3.4307,
      "step": 456
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.7974580526351929,
      "learning_rate": 0.0005999941756731315,
      "loss": 3.4182,
      "step": 457
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.944500744342804,
      "learning_rate": 0.0005999941501559312,
      "loss": 3.459,
      "step": 458
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.7782037854194641,
      "learning_rate": 0.0005999941245829563,
      "loss": 3.4898,
      "step": 459
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.7418212890625,
      "learning_rate": 0.0005999940989542069,
      "loss": 3.6051,
      "step": 460
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.7881678938865662,
      "learning_rate": 0.0005999940732696829,
      "loss": 3.569,
      "step": 461
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.9475841522216797,
      "learning_rate": 0.0005999940475293844,
      "loss": 3.5019,
      "step": 462
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.7840602397918701,
      "learning_rate": 0.0005999940217333112,
      "loss": 3.6412,
      "step": 463
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.8868883848190308,
      "learning_rate": 0.0005999939958814635,
      "loss": 3.4711,
      "step": 464
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.7657698392868042,
      "learning_rate": 0.0005999939699738413,
      "loss": 3.335,
      "step": 465
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.9719642400741577,
      "learning_rate": 0.0005999939440104445,
      "loss": 3.4288,
      "step": 466
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.9327822327613831,
      "learning_rate": 0.0005999939179912731,
      "loss": 3.5675,
      "step": 467
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.8279078006744385,
      "learning_rate": 0.0005999938919163274,
      "loss": 3.431,
      "step": 468
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.070550799369812,
      "learning_rate": 0.000599993865785607,
      "loss": 3.6465,
      "step": 469
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.8005127310752869,
      "learning_rate": 0.0005999938395991122,
      "loss": 3.7541,
      "step": 470
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.8457826375961304,
      "learning_rate": 0.0005999938133568428,
      "loss": 3.7065,
      "step": 471
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.9083381295204163,
      "learning_rate": 0.0005999937870587988,
      "loss": 3.4788,
      "step": 472
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.7672153115272522,
      "learning_rate": 0.0005999937607049804,
      "loss": 3.5329,
      "step": 473
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.8186494708061218,
      "learning_rate": 0.0005999937342953875,
      "loss": 3.5824,
      "step": 474
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.8081429600715637,
      "learning_rate": 0.0005999937078300199,
      "loss": 3.1802,
      "step": 475
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.0677380561828613,
      "learning_rate": 0.0005999936813088781,
      "loss": 3.4417,
      "step": 476
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.8462650179862976,
      "learning_rate": 0.0005999936547319617,
      "loss": 3.6237,
      "step": 477
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.95313560962677,
      "learning_rate": 0.0005999936280992708,
      "loss": 3.8707,
      "step": 478
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.9214100241661072,
      "learning_rate": 0.0005999936014108054,
      "loss": 3.7132,
      "step": 479
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.778019368648529,
      "learning_rate": 0.0005999935746665656,
      "loss": 3.5739,
      "step": 480
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.8912267088890076,
      "learning_rate": 0.0005999935478665513,
      "loss": 3.4943,
      "step": 481
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.8711246848106384,
      "learning_rate": 0.0005999935210107625,
      "loss": 3.5913,
      "step": 482
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.7874966263771057,
      "learning_rate": 0.0005999934940991993,
      "loss": 3.6903,
      "step": 483
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.2002983093261719,
      "learning_rate": 0.0005999934671318617,
      "loss": 3.3461,
      "step": 484
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.7593053579330444,
      "learning_rate": 0.0005999934401087495,
      "loss": 3.5793,
      "step": 485
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.9493660926818848,
      "learning_rate": 0.0005999934130298629,
      "loss": 3.4193,
      "step": 486
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.7637120485305786,
      "learning_rate": 0.0005999933858952018,
      "loss": 3.6371,
      "step": 487
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.798506498336792,
      "learning_rate": 0.0005999933587047665,
      "loss": 3.4996,
      "step": 488
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.841910719871521,
      "learning_rate": 0.0005999933314585565,
      "loss": 3.712,
      "step": 489
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.8655360341072083,
      "learning_rate": 0.0005999933041565723,
      "loss": 3.503,
      "step": 490
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.7094971537590027,
      "learning_rate": 0.0005999932767988135,
      "loss": 3.0864,
      "step": 491
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.8512486815452576,
      "learning_rate": 0.0005999932493852804,
      "loss": 3.6313,
      "step": 492
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.8355221748352051,
      "learning_rate": 0.0005999932219159728,
      "loss": 3.6555,
      "step": 493
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.795727014541626,
      "learning_rate": 0.0005999931943908909,
      "loss": 3.5134,
      "step": 494
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.0118874311447144,
      "learning_rate": 0.0005999931668100346,
      "loss": 3.4216,
      "step": 495
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.7971856594085693,
      "learning_rate": 0.0005999931391734039,
      "loss": 3.625,
      "step": 496
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.1987416744232178,
      "learning_rate": 0.0005999931114809987,
      "loss": 3.6331,
      "step": 497
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.8251052498817444,
      "learning_rate": 0.0005999930837328192,
      "loss": 3.3303,
      "step": 498
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.9577553272247314,
      "learning_rate": 0.0005999930559288653,
      "loss": 3.6365,
      "step": 499
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.106835126876831,
      "learning_rate": 0.0005999930280691371,
      "loss": 3.3479,
      "step": 500
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.8702870607376099,
      "learning_rate": 0.0005999930001536344,
      "loss": 3.6584,
      "step": 501
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.447774887084961,
      "learning_rate": 0.0005999929721823576,
      "loss": 3.8739,
      "step": 502
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.7970075011253357,
      "learning_rate": 0.0005999929441553063,
      "loss": 3.4632,
      "step": 503
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.053102970123291,
      "learning_rate": 0.0005999929160724805,
      "loss": 3.2237,
      "step": 504
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.8477090001106262,
      "learning_rate": 0.0005999928879338804,
      "loss": 3.5317,
      "step": 505
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.7163652777671814,
      "learning_rate": 0.0005999928597395061,
      "loss": 3.5759,
      "step": 506
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.863470733165741,
      "learning_rate": 0.0005999928314893574,
      "loss": 3.4721,
      "step": 507
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.801381528377533,
      "learning_rate": 0.0005999928031834343,
      "loss": 3.8178,
      "step": 508
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.7340736985206604,
      "learning_rate": 0.000599992774821737,
      "loss": 3.6247,
      "step": 509
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.1413087844848633,
      "learning_rate": 0.0005999927464042654,
      "loss": 3.628,
      "step": 510
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.065171241760254,
      "learning_rate": 0.0005999927179310194,
      "loss": 3.5443,
      "step": 511
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.7443844079971313,
      "learning_rate": 0.000599992689401999,
      "loss": 3.5804,
      "step": 512
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.9166836738586426,
      "learning_rate": 0.0005999926608172044,
      "loss": 3.4278,
      "step": 513
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.8475005030632019,
      "learning_rate": 0.0005999926321766355,
      "loss": 3.2714,
      "step": 514
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.8289783596992493,
      "learning_rate": 0.0005999926034802923,
      "loss": 3.743,
      "step": 515
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.9269516468048096,
      "learning_rate": 0.0005999925747281748,
      "loss": 3.4073,
      "step": 516
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.2235773801803589,
      "learning_rate": 0.000599992545920283,
      "loss": 3.6187,
      "step": 517
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.8380818367004395,
      "learning_rate": 0.0005999925170566171,
      "loss": 3.2662,
      "step": 518
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.046067237854004,
      "learning_rate": 0.0005999924881371767,
      "loss": 3.449,
      "step": 519
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.0084965229034424,
      "learning_rate": 0.0005999924591619621,
      "loss": 3.6383,
      "step": 520
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.9088554978370667,
      "learning_rate": 0.0005999924301309733,
      "loss": 3.2897,
      "step": 521
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.7583843469619751,
      "learning_rate": 0.0005999924010442101,
      "loss": 3.6388,
      "step": 522
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.7995028495788574,
      "learning_rate": 0.0005999923719016728,
      "loss": 3.5119,
      "step": 523
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.8535855412483215,
      "learning_rate": 0.0005999923427033612,
      "loss": 3.3261,
      "step": 524
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.8672518730163574,
      "learning_rate": 0.0005999923134492754,
      "loss": 3.5938,
      "step": 525
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.7281264066696167,
      "learning_rate": 0.0005999922841394153,
      "loss": 3.5026,
      "step": 526
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.7841817140579224,
      "learning_rate": 0.000599992254773781,
      "loss": 3.3918,
      "step": 527
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.8626765608787537,
      "learning_rate": 0.0005999922253523724,
      "loss": 3.4721,
      "step": 528
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.7345212697982788,
      "learning_rate": 0.0005999921958751897,
      "loss": 3.6145,
      "step": 529
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.7931383848190308,
      "learning_rate": 0.0005999921663422328,
      "loss": 3.8229,
      "step": 530
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.72552490234375,
      "learning_rate": 0.0005999921367535016,
      "loss": 3.5417,
      "step": 531
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.7026363611221313,
      "learning_rate": 0.0005999921071089962,
      "loss": 3.5306,
      "step": 532
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.7684736251831055,
      "learning_rate": 0.0005999920774087166,
      "loss": 3.6783,
      "step": 533
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.8254563212394714,
      "learning_rate": 0.0005999920476526629,
      "loss": 3.7499,
      "step": 534
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.8695827722549438,
      "learning_rate": 0.000599992017840835,
      "loss": 3.5431,
      "step": 535
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.779861569404602,
      "learning_rate": 0.0005999919879732329,
      "loss": 3.6447,
      "step": 536
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.9077920317649841,
      "learning_rate": 0.0005999919580498567,
      "loss": 3.7315,
      "step": 537
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.7548225522041321,
      "learning_rate": 0.0005999919280707061,
      "loss": 3.5868,
      "step": 538
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.7461755275726318,
      "learning_rate": 0.0005999918980357816,
      "loss": 3.5371,
      "step": 539
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.0242420434951782,
      "learning_rate": 0.0005999918679450828,
      "loss": 3.336,
      "step": 540
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.8568832874298096,
      "learning_rate": 0.0005999918377986098,
      "loss": 3.8114,
      "step": 541
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.8512759208679199,
      "learning_rate": 0.0005999918075963629,
      "loss": 3.2665,
      "step": 542
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.9493868350982666,
      "learning_rate": 0.0005999917773383416,
      "loss": 3.6932,
      "step": 543
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.7613044381141663,
      "learning_rate": 0.0005999917470245462,
      "loss": 3.2439,
      "step": 544
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.0672211647033691,
      "learning_rate": 0.0005999917166549768,
      "loss": 3.2311,
      "step": 545
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.819365918636322,
      "learning_rate": 0.0005999916862296333,
      "loss": 3.4616,
      "step": 546
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.8369011878967285,
      "learning_rate": 0.0005999916557485155,
      "loss": 3.6726,
      "step": 547
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.1985048055648804,
      "learning_rate": 0.0005999916252116237,
      "loss": 3.4746,
      "step": 548
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.2318602800369263,
      "learning_rate": 0.0005999915946189577,
      "loss": 3.5316,
      "step": 549
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.1507594585418701,
      "learning_rate": 0.0005999915639705178,
      "loss": 3.6927,
      "step": 550
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.9877436757087708,
      "learning_rate": 0.0005999915332663036,
      "loss": 3.0357,
      "step": 551
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.5629380941390991,
      "learning_rate": 0.0005999915025063153,
      "loss": 3.289,
      "step": 552
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.105570912361145,
      "learning_rate": 0.0005999914716905532,
      "loss": 3.33,
      "step": 553
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.9267014265060425,
      "learning_rate": 0.0005999914408190167,
      "loss": 3.2883,
      "step": 554
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.9353382587432861,
      "learning_rate": 0.0005999914098917063,
      "loss": 3.4177,
      "step": 555
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.036714792251587,
      "learning_rate": 0.0005999913789086218,
      "loss": 3.6023,
      "step": 556
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.0274107456207275,
      "learning_rate": 0.0005999913478697633,
      "loss": 3.3428,
      "step": 557
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.1122069358825684,
      "learning_rate": 0.0005999913167751307,
      "loss": 3.8948,
      "step": 558
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.0080984830856323,
      "learning_rate": 0.000599991285624724,
      "loss": 3.3422,
      "step": 559
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.7892789840698242,
      "learning_rate": 0.0005999912544185432,
      "loss": 3.4034,
      "step": 560
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.312290906906128,
      "learning_rate": 0.0005999912231565885,
      "loss": 3.3876,
      "step": 561
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.11339271068573,
      "learning_rate": 0.0005999911918388597,
      "loss": 3.5412,
      "step": 562
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.0692470073699951,
      "learning_rate": 0.0005999911604653568,
      "loss": 3.5554,
      "step": 563
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.9016618728637695,
      "learning_rate": 0.0005999911290360801,
      "loss": 3.454,
      "step": 564
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.8238917589187622,
      "learning_rate": 0.0005999910975510292,
      "loss": 3.6162,
      "step": 565
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.7549861073493958,
      "learning_rate": 0.0005999910660102044,
      "loss": 3.4143,
      "step": 566
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.861735463142395,
      "learning_rate": 0.0005999910344136056,
      "loss": 3.5282,
      "step": 567
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.9319385290145874,
      "learning_rate": 0.0005999910027612328,
      "loss": 3.5412,
      "step": 568
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.758252739906311,
      "learning_rate": 0.0005999909710530859,
      "loss": 3.5157,
      "step": 569
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.7930407524108887,
      "learning_rate": 0.000599990939289165,
      "loss": 3.5427,
      "step": 570
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.7458273768424988,
      "learning_rate": 0.0005999909074694702,
      "loss": 3.6591,
      "step": 571
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.7970849871635437,
      "learning_rate": 0.0005999908755940015,
      "loss": 3.5354,
      "step": 572
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.8071842789649963,
      "learning_rate": 0.0005999908436627587,
      "loss": 3.6511,
      "step": 573
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.8131169676780701,
      "learning_rate": 0.0005999908116757421,
      "loss": 3.4721,
      "step": 574
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.875365138053894,
      "learning_rate": 0.0005999907796329514,
      "loss": 3.2533,
      "step": 575
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.8255504369735718,
      "learning_rate": 0.0005999907475343868,
      "loss": 3.543,
      "step": 576
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.004029393196106,
      "learning_rate": 0.0005999907153800483,
      "loss": 3.1752,
      "step": 577
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.798754870891571,
      "learning_rate": 0.0005999906831699359,
      "loss": 3.4412,
      "step": 578
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.8280819654464722,
      "learning_rate": 0.0005999906509040493,
      "loss": 3.5771,
      "step": 579
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.8711402416229248,
      "learning_rate": 0.0005999906185823891,
      "loss": 3.6514,
      "step": 580
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.8323184847831726,
      "learning_rate": 0.0005999905862049548,
      "loss": 3.382,
      "step": 581
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.3318605422973633,
      "learning_rate": 0.0005999905537717466,
      "loss": 3.4413,
      "step": 582
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.8614515066146851,
      "learning_rate": 0.0005999905212827646,
      "loss": 3.7557,
      "step": 583
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.2018440961837769,
      "learning_rate": 0.0005999904887380086,
      "loss": 3.5166,
      "step": 584
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.9310529828071594,
      "learning_rate": 0.0005999904561374787,
      "loss": 3.6003,
      "step": 585
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.0775172710418701,
      "learning_rate": 0.0005999904234811749,
      "loss": 3.8222,
      "step": 586
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.0684548616409302,
      "learning_rate": 0.0005999903907690973,
      "loss": 3.5359,
      "step": 587
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.8897414207458496,
      "learning_rate": 0.0005999903580012458,
      "loss": 3.6652,
      "step": 588
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.8535715341567993,
      "learning_rate": 0.0005999903251776204,
      "loss": 3.2128,
      "step": 589
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.818472683429718,
      "learning_rate": 0.0005999902922982212,
      "loss": 3.3632,
      "step": 590
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.8770492076873779,
      "learning_rate": 0.0005999902593630481,
      "loss": 3.5581,
      "step": 591
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.1060638427734375,
      "learning_rate": 0.0005999902263721011,
      "loss": 3.6796,
      "step": 592
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.7245755195617676,
      "learning_rate": 0.0005999901933253802,
      "loss": 3.2618,
      "step": 593
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.7876566648483276,
      "learning_rate": 0.0005999901602228856,
      "loss": 3.6634,
      "step": 594
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.9993135333061218,
      "learning_rate": 0.0005999901270646172,
      "loss": 3.2501,
      "step": 595
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.8253893256187439,
      "learning_rate": 0.0005999900938505749,
      "loss": 3.5744,
      "step": 596
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.8575803637504578,
      "learning_rate": 0.0005999900605807587,
      "loss": 3.5498,
      "step": 597
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.9513584971427917,
      "learning_rate": 0.0005999900272551688,
      "loss": 3.5574,
      "step": 598
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.8615128993988037,
      "learning_rate": 0.000599989993873805,
      "loss": 3.5256,
      "step": 599
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.910697340965271,
      "learning_rate": 0.0005999899604366675,
      "loss": 3.5193,
      "step": 600
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.0285612344741821,
      "learning_rate": 0.0005999899269437561,
      "loss": 3.4061,
      "step": 601
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.1850478649139404,
      "learning_rate": 0.000599989893395071,
      "loss": 3.2984,
      "step": 602
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.8815431594848633,
      "learning_rate": 0.0005999898597906121,
      "loss": 3.6032,
      "step": 603
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.8975661396980286,
      "learning_rate": 0.0005999898261303794,
      "loss": 3.4771,
      "step": 604
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.757195234298706,
      "learning_rate": 0.0005999897924143729,
      "loss": 3.5097,
      "step": 605
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.966744065284729,
      "learning_rate": 0.0005999897586425926,
      "loss": 3.3165,
      "step": 606
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.8303767442703247,
      "learning_rate": 0.0005999897248150387,
      "loss": 3.8319,
      "step": 607
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.8378311991691589,
      "learning_rate": 0.0005999896909317109,
      "loss": 3.4908,
      "step": 608
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.9601596593856812,
      "learning_rate": 0.0005999896569926094,
      "loss": 3.3905,
      "step": 609
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.8744773268699646,
      "learning_rate": 0.0005999896229977342,
      "loss": 3.2739,
      "step": 610
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.2184559106826782,
      "learning_rate": 0.0005999895889470852,
      "loss": 3.1805,
      "step": 611
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.7622731924057007,
      "learning_rate": 0.0005999895548406624,
      "loss": 3.5121,
      "step": 612
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.2169854640960693,
      "learning_rate": 0.000599989520678466,
      "loss": 3.154,
      "step": 613
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.0930895805358887,
      "learning_rate": 0.000599989486460496,
      "loss": 3.4856,
      "step": 614
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.8971847891807556,
      "learning_rate": 0.0005999894521867521,
      "loss": 3.5673,
      "step": 615
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.1158446073532104,
      "learning_rate": 0.0005999894178572346,
      "loss": 3.2888,
      "step": 616
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.7854042053222656,
      "learning_rate": 0.0005999893834719433,
      "loss": 3.4101,
      "step": 617
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.8417444825172424,
      "learning_rate": 0.0005999893490308784,
      "loss": 3.6953,
      "step": 618
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.0599336624145508,
      "learning_rate": 0.0005999893145340398,
      "loss": 3.5159,
      "step": 619
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.8014999032020569,
      "learning_rate": 0.0005999892799814275,
      "loss": 3.3427,
      "step": 620
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.146329641342163,
      "learning_rate": 0.0005999892453730415,
      "loss": 3.4663,
      "step": 621
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.0870041847229004,
      "learning_rate": 0.000599989210708882,
      "loss": 3.3837,
      "step": 622
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.7977124452590942,
      "learning_rate": 0.0005999891759889487,
      "loss": 3.4093,
      "step": 623
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.0889734029769897,
      "learning_rate": 0.0005999891412132417,
      "loss": 3.2065,
      "step": 624
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.8392372727394104,
      "learning_rate": 0.0005999891063817611,
      "loss": 3.4064,
      "step": 625
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.7126564383506775,
      "learning_rate": 0.0005999890714945069,
      "loss": 3.3532,
      "step": 626
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.8350976705551147,
      "learning_rate": 0.0005999890365514791,
      "loss": 3.3594,
      "step": 627
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.9223216772079468,
      "learning_rate": 0.0005999890015526776,
      "loss": 3.2089,
      "step": 628
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.7886050343513489,
      "learning_rate": 0.0005999889664981026,
      "loss": 3.3675,
      "step": 629
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.7799350619316101,
      "learning_rate": 0.0005999889313877539,
      "loss": 3.1585,
      "step": 630
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.7520298361778259,
      "learning_rate": 0.0005999888962216316,
      "loss": 3.4539,
      "step": 631
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.7680748701095581,
      "learning_rate": 0.0005999888609997357,
      "loss": 3.5733,
      "step": 632
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.7863929867744446,
      "learning_rate": 0.0005999888257220662,
      "loss": 3.531,
      "step": 633
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.8166525959968567,
      "learning_rate": 0.0005999887903886231,
      "loss": 3.3324,
      "step": 634
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.7944643497467041,
      "learning_rate": 0.0005999887549994065,
      "loss": 3.5543,
      "step": 635
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.7804011702537537,
      "learning_rate": 0.0005999887195544161,
      "loss": 3.5342,
      "step": 636
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.8716467022895813,
      "learning_rate": 0.0005999886840536523,
      "loss": 3.2005,
      "step": 637
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.9455381035804749,
      "learning_rate": 0.000599988648497115,
      "loss": 3.3864,
      "step": 638
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.1551458835601807,
      "learning_rate": 0.0005999886128848041,
      "loss": 3.4173,
      "step": 639
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.9207316637039185,
      "learning_rate": 0.0005999885772167197,
      "loss": 3.6745,
      "step": 640
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.011630892753601,
      "learning_rate": 0.0005999885414928617,
      "loss": 3.3486,
      "step": 641
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.7127177119255066,
      "learning_rate": 0.0005999885057132303,
      "loss": 3.2299,
      "step": 642
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.061045527458191,
      "learning_rate": 0.0005999884698778252,
      "loss": 3.3902,
      "step": 643
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.8790097236633301,
      "learning_rate": 0.0005999884339866465,
      "loss": 3.3875,
      "step": 644
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.7489104270935059,
      "learning_rate": 0.0005999883980396945,
      "loss": 3.4065,
      "step": 645
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.8663876056671143,
      "learning_rate": 0.000599988362036969,
      "loss": 3.7543,
      "step": 646
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.8435371518135071,
      "learning_rate": 0.0005999883259784698,
      "loss": 3.3367,
      "step": 647
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.0195229053497314,
      "learning_rate": 0.0005999882898641973,
      "loss": 3.2559,
      "step": 648
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.812505841255188,
      "learning_rate": 0.0005999882536941512,
      "loss": 3.4544,
      "step": 649
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.8585780262947083,
      "learning_rate": 0.0005999882174683316,
      "loss": 3.5572,
      "step": 650
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.8765333890914917,
      "learning_rate": 0.0005999881811867387,
      "loss": 3.4072,
      "step": 651
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.7615121603012085,
      "learning_rate": 0.0005999881448493721,
      "loss": 3.5031,
      "step": 652
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.1103681325912476,
      "learning_rate": 0.0005999881084562321,
      "loss": 3.4584,
      "step": 653
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.8299665451049805,
      "learning_rate": 0.0005999880720073188,
      "loss": 3.5001,
      "step": 654
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.9612872004508972,
      "learning_rate": 0.000599988035502632,
      "loss": 3.2681,
      "step": 655
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.6742570400238037,
      "learning_rate": 0.0005999879989421717,
      "loss": 3.4597,
      "step": 656
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.9059084057807922,
      "learning_rate": 0.000599987962325938,
      "loss": 3.5999,
      "step": 657
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.7879665493965149,
      "learning_rate": 0.0005999879256539309,
      "loss": 3.3557,
      "step": 658
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.943993866443634,
      "learning_rate": 0.0005999878889261503,
      "loss": 3.1882,
      "step": 659
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.7308316826820374,
      "learning_rate": 0.0005999878521425964,
      "loss": 3.5851,
      "step": 660
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.7961083054542542,
      "learning_rate": 0.000599987815303269,
      "loss": 3.5208,
      "step": 661
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.8542512059211731,
      "learning_rate": 0.0005999877784081682,
      "loss": 3.3938,
      "step": 662
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.755682647228241,
      "learning_rate": 0.0005999877414572941,
      "loss": 3.4757,
      "step": 663
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.8829278349876404,
      "learning_rate": 0.0005999877044506464,
      "loss": 3.2846,
      "step": 664
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.8717522025108337,
      "learning_rate": 0.0005999876673882256,
      "loss": 3.386,
      "step": 665
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.8942267894744873,
      "learning_rate": 0.0005999876302700313,
      "loss": 3.2878,
      "step": 666
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.8135119676589966,
      "learning_rate": 0.0005999875930960638,
      "loss": 3.5343,
      "step": 667
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.7700066566467285,
      "learning_rate": 0.0005999875558663227,
      "loss": 3.3353,
      "step": 668
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.8145167827606201,
      "learning_rate": 0.0005999875185808082,
      "loss": 3.5688,
      "step": 669
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.8949995040893555,
      "learning_rate": 0.0005999874812395206,
      "loss": 3.5847,
      "step": 670
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.7354756593704224,
      "learning_rate": 0.0005999874438424596,
      "loss": 3.3659,
      "step": 671
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.7411088943481445,
      "learning_rate": 0.0005999874063896253,
      "loss": 3.4139,
      "step": 672
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.7916101217269897,
      "learning_rate": 0.0005999873688810176,
      "loss": 3.3141,
      "step": 673
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.7791243195533752,
      "learning_rate": 0.0005999873313166365,
      "loss": 3.4582,
      "step": 674
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.9098423719406128,
      "learning_rate": 0.0005999872936964822,
      "loss": 3.2412,
      "step": 675
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.8043963313102722,
      "learning_rate": 0.0005999872560205547,
      "loss": 3.3149,
      "step": 676
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.7996683120727539,
      "learning_rate": 0.0005999872182888537,
      "loss": 3.2843,
      "step": 677
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.7907979488372803,
      "learning_rate": 0.0005999871805013795,
      "loss": 3.2176,
      "step": 678
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.8069727420806885,
      "learning_rate": 0.0005999871426581321,
      "loss": 3.0664,
      "step": 679
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.8115625381469727,
      "learning_rate": 0.0005999871047591114,
      "loss": 3.512,
      "step": 680
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.824285626411438,
      "learning_rate": 0.0005999870668043174,
      "loss": 3.3283,
      "step": 681
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.8240658044815063,
      "learning_rate": 0.00059998702879375,
      "loss": 3.3516,
      "step": 682
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.8088521957397461,
      "learning_rate": 0.0005999869907274096,
      "loss": 3.1889,
      "step": 683
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.7754963040351868,
      "learning_rate": 0.0005999869526052958,
      "loss": 3.4082,
      "step": 684
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.9048473834991455,
      "learning_rate": 0.0005999869144274088,
      "loss": 3.3609,
      "step": 685
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.7378559708595276,
      "learning_rate": 0.0005999868761937486,
      "loss": 3.3231,
      "step": 686
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.9863017201423645,
      "learning_rate": 0.0005999868379043151,
      "loss": 3.6687,
      "step": 687
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.6879296898841858,
      "learning_rate": 0.0005999867995591085,
      "loss": 3.1957,
      "step": 688
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.1461498737335205,
      "learning_rate": 0.0005999867611581287,
      "loss": 3.3054,
      "step": 689
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.7632550597190857,
      "learning_rate": 0.0005999867227013756,
      "loss": 3.4449,
      "step": 690
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.8535746932029724,
      "learning_rate": 0.0005999866841888493,
      "loss": 3.4523,
      "step": 691
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.741168737411499,
      "learning_rate": 0.0005999866456205498,
      "loss": 3.2162,
      "step": 692
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.8711710572242737,
      "learning_rate": 0.0005999866069964772,
      "loss": 3.4197,
      "step": 693
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.7853380441665649,
      "learning_rate": 0.0005999865683166314,
      "loss": 3.3136,
      "step": 694
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.7544312477111816,
      "learning_rate": 0.0005999865295810125,
      "loss": 3.5798,
      "step": 695
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.7794734835624695,
      "learning_rate": 0.0005999864907896203,
      "loss": 3.2542,
      "step": 696
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.8529940247535706,
      "learning_rate": 0.0005999864519424551,
      "loss": 3.4602,
      "step": 697
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.796694278717041,
      "learning_rate": 0.0005999864130395166,
      "loss": 3.0806,
      "step": 698
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.8461068868637085,
      "learning_rate": 0.000599986374080805,
      "loss": 3.2838,
      "step": 699
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.9021090865135193,
      "learning_rate": 0.0005999863350663202,
      "loss": 3.2494,
      "step": 700
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.8209679126739502,
      "learning_rate": 0.0005999862959960626,
      "loss": 3.3528,
      "step": 701
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.755278468132019,
      "learning_rate": 0.0005999862568700316,
      "loss": 3.2971,
      "step": 702
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.8889586329460144,
      "learning_rate": 0.0005999862176882275,
      "loss": 3.3629,
      "step": 703
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.8623082637786865,
      "learning_rate": 0.0005999861784506504,
      "loss": 3.6733,
      "step": 704
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.8678564429283142,
      "learning_rate": 0.0005999861391573002,
      "loss": 3.4507,
      "step": 705
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.8210625052452087,
      "learning_rate": 0.0005999860998081769,
      "loss": 3.4233,
      "step": 706
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.8222780227661133,
      "learning_rate": 0.0005999860604032805,
      "loss": 3.4395,
      "step": 707
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.9021074771881104,
      "learning_rate": 0.000599986020942611,
      "loss": 3.5727,
      "step": 708
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.785944402217865,
      "learning_rate": 0.0005999859814261685,
      "loss": 3.2129,
      "step": 709
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.9776563048362732,
      "learning_rate": 0.0005999859418539528,
      "loss": 3.1816,
      "step": 710
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.8119636178016663,
      "learning_rate": 0.0005999859022259642,
      "loss": 3.1939,
      "step": 711
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.9023457169532776,
      "learning_rate": 0.0005999858625422026,
      "loss": 3.2681,
      "step": 712
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.8898686766624451,
      "learning_rate": 0.0005999858228026678,
      "loss": 3.5318,
      "step": 713
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.846539318561554,
      "learning_rate": 0.0005999857830073602,
      "loss": 3.41,
      "step": 714
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.8138471245765686,
      "learning_rate": 0.0005999857431562793,
      "loss": 3.4087,
      "step": 715
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.8342599272727966,
      "learning_rate": 0.0005999857032494255,
      "loss": 3.5439,
      "step": 716
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.8145731091499329,
      "learning_rate": 0.0005999856632867988,
      "loss": 3.6398,
      "step": 717
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.8120128512382507,
      "learning_rate": 0.0005999856232683991,
      "loss": 3.2804,
      "step": 718
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.759366512298584,
      "learning_rate": 0.0005999855831942262,
      "loss": 3.2651,
      "step": 719
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.7144477963447571,
      "learning_rate": 0.0005999855430642804,
      "loss": 3.4524,
      "step": 720
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.8309988379478455,
      "learning_rate": 0.0005999855028785617,
      "loss": 3.4575,
      "step": 721
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.8910622000694275,
      "learning_rate": 0.0005999854626370701,
      "loss": 3.3792,
      "step": 722
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.8517363667488098,
      "learning_rate": 0.0005999854223398054,
      "loss": 3.1278,
      "step": 723
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.754740297794342,
      "learning_rate": 0.0005999853819867678,
      "loss": 3.4277,
      "step": 724
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.0368021726608276,
      "learning_rate": 0.0005999853415779573,
      "loss": 3.4771,
      "step": 725
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.8015038371086121,
      "learning_rate": 0.0005999853011133739,
      "loss": 3.4362,
      "step": 726
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.8309590220451355,
      "learning_rate": 0.0005999852605930174,
      "loss": 3.3464,
      "step": 727
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.4098455905914307,
      "learning_rate": 0.000599985220016888,
      "loss": 3.5036,
      "step": 728
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.9395402669906616,
      "learning_rate": 0.0005999851793849857,
      "loss": 3.1746,
      "step": 729
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.1658068895339966,
      "learning_rate": 0.0005999851386973105,
      "loss": 3.437,
      "step": 730
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.8590440154075623,
      "learning_rate": 0.0005999850979538625,
      "loss": 3.3333,
      "step": 731
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.877025842666626,
      "learning_rate": 0.0005999850571546415,
      "loss": 3.2781,
      "step": 732
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.9276472926139832,
      "learning_rate": 0.0005999850162996478,
      "loss": 3.4689,
      "step": 733
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.8943889141082764,
      "learning_rate": 0.000599984975388881,
      "loss": 3.3456,
      "step": 734
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.894435465335846,
      "learning_rate": 0.0005999849344223415,
      "loss": 3.4187,
      "step": 735
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.8328893184661865,
      "learning_rate": 0.000599984893400029,
      "loss": 3.1837,
      "step": 736
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.858498215675354,
      "learning_rate": 0.0005999848523219436,
      "loss": 3.2978,
      "step": 737
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.8917897343635559,
      "learning_rate": 0.0005999848111880856,
      "loss": 3.5523,
      "step": 738
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.0036474466323853,
      "learning_rate": 0.0005999847699984546,
      "loss": 3.1829,
      "step": 739
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.7422719597816467,
      "learning_rate": 0.0005999847287530507,
      "loss": 3.3129,
      "step": 740
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.171730399131775,
      "learning_rate": 0.0005999846874518741,
      "loss": 3.5475,
      "step": 741
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.8029626607894897,
      "learning_rate": 0.0005999846460949247,
      "loss": 3.0603,
      "step": 742
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.8554865121841431,
      "learning_rate": 0.0005999846046822023,
      "loss": 3.1229,
      "step": 743
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.073297381401062,
      "learning_rate": 0.0005999845632137072,
      "loss": 3.4033,
      "step": 744
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.8882404565811157,
      "learning_rate": 0.0005999845216894394,
      "loss": 3.5824,
      "step": 745
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.9747998714447021,
      "learning_rate": 0.0005999844801093988,
      "loss": 3.3225,
      "step": 746
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.8060855269432068,
      "learning_rate": 0.0005999844384735854,
      "loss": 3.5225,
      "step": 747
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.9199317693710327,
      "learning_rate": 0.0005999843967819992,
      "loss": 3.3594,
      "step": 748
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.8478732109069824,
      "learning_rate": 0.0005999843550346402,
      "loss": 3.4365,
      "step": 749
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.8949321508407593,
      "learning_rate": 0.0005999843132315085,
      "loss": 3.5656,
      "step": 750
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.8184410929679871,
      "learning_rate": 0.000599984271372604,
      "loss": 3.2006,
      "step": 751
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.8938193321228027,
      "learning_rate": 0.0005999842294579269,
      "loss": 3.4848,
      "step": 752
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.2647660970687866,
      "learning_rate": 0.000599984187487477,
      "loss": 3.3456,
      "step": 753
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.7307713031768799,
      "learning_rate": 0.0005999841454612542,
      "loss": 3.5022,
      "step": 754
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.0275088548660278,
      "learning_rate": 0.0005999841033792588,
      "loss": 3.0511,
      "step": 755
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.9624373912811279,
      "learning_rate": 0.0005999840612414907,
      "loss": 3.2248,
      "step": 756
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.7257282733917236,
      "learning_rate": 0.0005999840190479501,
      "loss": 3.0836,
      "step": 757
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.9335401654243469,
      "learning_rate": 0.0005999839767986367,
      "loss": 3.3753,
      "step": 758
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.8170746564865112,
      "learning_rate": 0.0005999839344935505,
      "loss": 3.604,
      "step": 759
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.8183690905570984,
      "learning_rate": 0.0005999838921326916,
      "loss": 3.3703,
      "step": 760
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.8618467450141907,
      "learning_rate": 0.0005999838497160601,
      "loss": 3.3115,
      "step": 761
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.7993667125701904,
      "learning_rate": 0.000599983807243656,
      "loss": 3.3307,
      "step": 762
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.792827308177948,
      "learning_rate": 0.0005999837647154792,
      "loss": 3.2018,
      "step": 763
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.853661298751831,
      "learning_rate": 0.0005999837221315297,
      "loss": 3.5814,
      "step": 764
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.871378481388092,
      "learning_rate": 0.0005999836794918076,
      "loss": 3.5378,
      "step": 765
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.8352946639060974,
      "learning_rate": 0.000599983636796313,
      "loss": 3.341,
      "step": 766
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.8937336206436157,
      "learning_rate": 0.0005999835940450456,
      "loss": 3.2598,
      "step": 767
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.079317331314087,
      "learning_rate": 0.0005999835512380057,
      "loss": 3.0648,
      "step": 768
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.9059333205223083,
      "learning_rate": 0.0005999835083751931,
      "loss": 3.5012,
      "step": 769
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.1738885641098022,
      "learning_rate": 0.0005999834654566079,
      "loss": 3.2762,
      "step": 770
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.7429060935974121,
      "learning_rate": 0.0005999834224822503,
      "loss": 3.0523,
      "step": 771
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.050021767616272,
      "learning_rate": 0.0005999833794521199,
      "loss": 3.1093,
      "step": 772
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.8014231324195862,
      "learning_rate": 0.000599983336366217,
      "loss": 3.3141,
      "step": 773
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.8100259304046631,
      "learning_rate": 0.0005999832932245415,
      "loss": 3.42,
      "step": 774
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.791104257106781,
      "learning_rate": 0.0005999832500270935,
      "loss": 3.3181,
      "step": 775
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.7583730816841125,
      "learning_rate": 0.0005999832067738729,
      "loss": 3.3237,
      "step": 776
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.7855584025382996,
      "learning_rate": 0.0005999831634648799,
      "loss": 3.4065,
      "step": 777
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.9040716290473938,
      "learning_rate": 0.0005999831201001142,
      "loss": 3.1451,
      "step": 778
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.8186565041542053,
      "learning_rate": 0.0005999830766795761,
      "loss": 3.3068,
      "step": 779
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.8966705799102783,
      "learning_rate": 0.0005999830332032653,
      "loss": 3.334,
      "step": 780
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.9440838694572449,
      "learning_rate": 0.0005999829896711821,
      "loss": 3.4555,
      "step": 781
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.7755387425422668,
      "learning_rate": 0.0005999829460833265,
      "loss": 3.66,
      "step": 782
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.9743005037307739,
      "learning_rate": 0.0005999829024396984,
      "loss": 3.2713,
      "step": 783
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.7744138240814209,
      "learning_rate": 0.0005999828587402976,
      "loss": 3.3704,
      "step": 784
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.8642399311065674,
      "learning_rate": 0.0005999828149851245,
      "loss": 3.1362,
      "step": 785
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.0280178785324097,
      "learning_rate": 0.0005999827711741789,
      "loss": 3.2036,
      "step": 786
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.8084916472434998,
      "learning_rate": 0.0005999827273074609,
      "loss": 3.1876,
      "step": 787
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.8680306673049927,
      "learning_rate": 0.0005999826833849703,
      "loss": 3.3436,
      "step": 788
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.9047846794128418,
      "learning_rate": 0.0005999826394067074,
      "loss": 3.2996,
      "step": 789
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.0264270305633545,
      "learning_rate": 0.0005999825953726719,
      "loss": 3.3062,
      "step": 790
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.9794740080833435,
      "learning_rate": 0.0005999825512828641,
      "loss": 3.1758,
      "step": 791
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.0372703075408936,
      "learning_rate": 0.0005999825071372839,
      "loss": 3.2022,
      "step": 792
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.9320720434188843,
      "learning_rate": 0.0005999824629359313,
      "loss": 3.3716,
      "step": 793
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.8474474549293518,
      "learning_rate": 0.0005999824186788062,
      "loss": 3.1923,
      "step": 794
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.9650818705558777,
      "learning_rate": 0.0005999823743659088,
      "loss": 3.3987,
      "step": 795
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.8994830846786499,
      "learning_rate": 0.0005999823299972389,
      "loss": 3.2399,
      "step": 796
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.7566675543785095,
      "learning_rate": 0.0005999822855727967,
      "loss": 3.211,
      "step": 797
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.0106488466262817,
      "learning_rate": 0.0005999822410925821,
      "loss": 3.3039,
      "step": 798
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.821566641330719,
      "learning_rate": 0.0005999821965565952,
      "loss": 3.2155,
      "step": 799
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.8570018410682678,
      "learning_rate": 0.0005999821519648359,
      "loss": 3.2606,
      "step": 800
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.8725783824920654,
      "learning_rate": 0.0005999821073173042,
      "loss": 3.5404,
      "step": 801
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.8091997504234314,
      "learning_rate": 0.0005999820626140002,
      "loss": 3.6528,
      "step": 802
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.1548793315887451,
      "learning_rate": 0.0005999820178549239,
      "loss": 3.3631,
      "step": 803
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.8277431130409241,
      "learning_rate": 0.0005999819730400754,
      "loss": 3.408,
      "step": 804
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.8793411254882812,
      "learning_rate": 0.0005999819281694544,
      "loss": 3.3645,
      "step": 805
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.7964050769805908,
      "learning_rate": 0.0005999818832430611,
      "loss": 3.2091,
      "step": 806
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.8662764430046082,
      "learning_rate": 0.0005999818382608956,
      "loss": 3.1855,
      "step": 807
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.8610309958457947,
      "learning_rate": 0.0005999817932229578,
      "loss": 3.3675,
      "step": 808
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.7929859161376953,
      "learning_rate": 0.0005999817481292478,
      "loss": 3.4367,
      "step": 809
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.7899400591850281,
      "learning_rate": 0.0005999817029797653,
      "loss": 3.4544,
      "step": 810
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.8667285442352295,
      "learning_rate": 0.0005999816577745107,
      "loss": 3.4068,
      "step": 811
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.7746016383171082,
      "learning_rate": 0.0005999816125134838,
      "loss": 3.2477,
      "step": 812
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.7665903568267822,
      "learning_rate": 0.0005999815671966847,
      "loss": 2.846,
      "step": 813
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.7905123233795166,
      "learning_rate": 0.0005999815218241133,
      "loss": 3.0903,
      "step": 814
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.7763956785202026,
      "learning_rate": 0.0005999814763957697,
      "loss": 3.1368,
      "step": 815
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.7339471578598022,
      "learning_rate": 0.0005999814309116539,
      "loss": 3.3631,
      "step": 816
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.7810478806495667,
      "learning_rate": 0.0005999813853717659,
      "loss": 3.4407,
      "step": 817
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.7550322413444519,
      "learning_rate": 0.0005999813397761057,
      "loss": 3.1647,
      "step": 818
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.7281199097633362,
      "learning_rate": 0.0005999812941246733,
      "loss": 2.8103,
      "step": 819
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.7636454701423645,
      "learning_rate": 0.0005999812484174686,
      "loss": 3.5674,
      "step": 820
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.792745053768158,
      "learning_rate": 0.0005999812026544919,
      "loss": 3.3081,
      "step": 821
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.7439790368080139,
      "learning_rate": 0.0005999811568357429,
      "loss": 3.3834,
      "step": 822
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.9040713906288147,
      "learning_rate": 0.0005999811109612218,
      "loss": 3.2005,
      "step": 823
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.7880575656890869,
      "learning_rate": 0.0005999810650309286,
      "loss": 3.2711,
      "step": 824
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.7583553791046143,
      "learning_rate": 0.0005999810190448632,
      "loss": 2.9447,
      "step": 825
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.0028287172317505,
      "learning_rate": 0.0005999809730030257,
      "loss": 2.9,
      "step": 826
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.7782334685325623,
      "learning_rate": 0.000599980926905416,
      "loss": 3.5678,
      "step": 827
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.8022214770317078,
      "learning_rate": 0.0005999808807520342,
      "loss": 3.433,
      "step": 828
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.7236291170120239,
      "learning_rate": 0.0005999808345428804,
      "loss": 3.1812,
      "step": 829
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.7988166809082031,
      "learning_rate": 0.0005999807882779544,
      "loss": 3.3484,
      "step": 830
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.7723302841186523,
      "learning_rate": 0.0005999807419572562,
      "loss": 3.434,
      "step": 831
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.8560715317726135,
      "learning_rate": 0.0005999806955807861,
      "loss": 3.5048,
      "step": 832
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.7039713859558105,
      "learning_rate": 0.0005999806491485439,
      "loss": 3.3395,
      "step": 833
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.860588788986206,
      "learning_rate": 0.0005999806026605295,
      "loss": 3.2926,
      "step": 834
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.7624431848526001,
      "learning_rate": 0.0005999805561167432,
      "loss": 3.0672,
      "step": 835
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.1814032793045044,
      "learning_rate": 0.0005999805095171848,
      "loss": 3.1159,
      "step": 836
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.8567785620689392,
      "learning_rate": 0.0005999804628618544,
      "loss": 3.1227,
      "step": 837
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.0165398120880127,
      "learning_rate": 0.0005999804161507519,
      "loss": 3.2275,
      "step": 838
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.8041603565216064,
      "learning_rate": 0.0005999803693838774,
      "loss": 3.3752,
      "step": 839
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.80474454164505,
      "learning_rate": 0.0005999803225612308,
      "loss": 3.3519,
      "step": 840
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.9454123377799988,
      "learning_rate": 0.0005999802756828124,
      "loss": 3.2991,
      "step": 841
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.0272222757339478,
      "learning_rate": 0.0005999802287486219,
      "loss": 3.2229,
      "step": 842
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.9342088103294373,
      "learning_rate": 0.0005999801817586594,
      "loss": 3.1576,
      "step": 843
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.8539234399795532,
      "learning_rate": 0.0005999801347129249,
      "loss": 3.0556,
      "step": 844
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.3631280660629272,
      "learning_rate": 0.0005999800876114185,
      "loss": 3.3969,
      "step": 845
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.799173891544342,
      "learning_rate": 0.0005999800404541401,
      "loss": 3.1005,
      "step": 846
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.3325204849243164,
      "learning_rate": 0.0005999799932410897,
      "loss": 3.2726,
      "step": 847
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.7651214003562927,
      "learning_rate": 0.0005999799459722674,
      "loss": 3.1349,
      "step": 848
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.9402390122413635,
      "learning_rate": 0.0005999798986476731,
      "loss": 3.3189,
      "step": 849
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.7697799801826477,
      "learning_rate": 0.000599979851267307,
      "loss": 3.2775,
      "step": 850
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.8653131723403931,
      "learning_rate": 0.000599979803831169,
      "loss": 3.2634,
      "step": 851
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.1483381986618042,
      "learning_rate": 0.000599979756339259,
      "loss": 3.4562,
      "step": 852
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.0148472785949707,
      "learning_rate": 0.0005999797087915771,
      "loss": 3.2195,
      "step": 853
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.8002171516418457,
      "learning_rate": 0.0005999796611881233,
      "loss": 3.2479,
      "step": 854
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.782416820526123,
      "learning_rate": 0.0005999796135288977,
      "loss": 3.1046,
      "step": 855
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.7983565926551819,
      "learning_rate": 0.0005999795658139001,
      "loss": 3.2262,
      "step": 856
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.7673105597496033,
      "learning_rate": 0.0005999795180431307,
      "loss": 3.273,
      "step": 857
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.8335174322128296,
      "learning_rate": 0.0005999794702165895,
      "loss": 3.0428,
      "step": 858
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.7781415581703186,
      "learning_rate": 0.0005999794223342764,
      "loss": 3.1615,
      "step": 859
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.8781360983848572,
      "learning_rate": 0.0005999793743961915,
      "loss": 3.3431,
      "step": 860
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.722184419631958,
      "learning_rate": 0.0005999793264023347,
      "loss": 3.255,
      "step": 861
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.7857291102409363,
      "learning_rate": 0.0005999792783527062,
      "loss": 3.352,
      "step": 862
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.837360143661499,
      "learning_rate": 0.0005999792302473058,
      "loss": 3.5004,
      "step": 863
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.7732536792755127,
      "learning_rate": 0.0005999791820861335,
      "loss": 3.3093,
      "step": 864
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.8052021861076355,
      "learning_rate": 0.0005999791338691895,
      "loss": 3.4684,
      "step": 865
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.833842933177948,
      "learning_rate": 0.0005999790855964739,
      "loss": 3.0877,
      "step": 866
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.8405336737632751,
      "learning_rate": 0.0005999790372679863,
      "loss": 3.2679,
      "step": 867
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.8179133534431458,
      "learning_rate": 0.000599978988883727,
      "loss": 3.1609,
      "step": 868
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.7780095338821411,
      "learning_rate": 0.0005999789404436959,
      "loss": 3.3557,
      "step": 869
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.7876889109611511,
      "learning_rate": 0.0005999788919478931,
      "loss": 3.1878,
      "step": 870
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.7700024843215942,
      "learning_rate": 0.0005999788433963186,
      "loss": 3.2032,
      "step": 871
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.9838472008705139,
      "learning_rate": 0.0005999787947889723,
      "loss": 3.2283,
      "step": 872
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.8543108105659485,
      "learning_rate": 0.0005999787461258543,
      "loss": 3.4052,
      "step": 873
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.9238500595092773,
      "learning_rate": 0.0005999786974069646,
      "loss": 3.2455,
      "step": 874
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.7652746438980103,
      "learning_rate": 0.0005999786486323032,
      "loss": 3.1499,
      "step": 875
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.9613887667655945,
      "learning_rate": 0.0005999785998018701,
      "loss": 3.2634,
      "step": 876
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.8797415494918823,
      "learning_rate": 0.0005999785509156653,
      "loss": 3.2021,
      "step": 877
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.8232781291007996,
      "learning_rate": 0.0005999785019736888,
      "loss": 3.3051,
      "step": 878
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.9357144236564636,
      "learning_rate": 0.0005999784529759406,
      "loss": 3.3177,
      "step": 879
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.9409860968589783,
      "learning_rate": 0.0005999784039224209,
      "loss": 3.2413,
      "step": 880
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.741877019405365,
      "learning_rate": 0.0005999783548131294,
      "loss": 3.1583,
      "step": 881
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.8516069650650024,
      "learning_rate": 0.0005999783056480664,
      "loss": 3.023,
      "step": 882
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.7395252585411072,
      "learning_rate": 0.0005999782564272317,
      "loss": 3.2332,
      "step": 883
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.7904126644134521,
      "learning_rate": 0.0005999782071506254,
      "loss": 3.0208,
      "step": 884
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.7280486226081848,
      "learning_rate": 0.0005999781578182474,
      "loss": 3.5139,
      "step": 885
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.7957977652549744,
      "learning_rate": 0.0005999781084300979,
      "loss": 3.2402,
      "step": 886
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.7666311860084534,
      "learning_rate": 0.0005999780589861768,
      "loss": 3.2309,
      "step": 887
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.7503247857093811,
      "learning_rate": 0.0005999780094864841,
      "loss": 3.4372,
      "step": 888
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.7919597029685974,
      "learning_rate": 0.0005999779599310198,
      "loss": 3.4303,
      "step": 889
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.9152701497077942,
      "learning_rate": 0.0005999779103197839,
      "loss": 3.0531,
      "step": 890
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.17238187789917,
      "learning_rate": 0.0005999778606527765,
      "loss": 2.8773,
      "step": 891
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.8534356951713562,
      "learning_rate": 0.0005999778109299975,
      "loss": 3.7238,
      "step": 892
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.0786784887313843,
      "learning_rate": 0.000599977761151447,
      "loss": 3.2053,
      "step": 893
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.8469918966293335,
      "learning_rate": 0.000599977711317125,
      "loss": 3.1357,
      "step": 894
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.000461220741272,
      "learning_rate": 0.0005999776614270314,
      "loss": 3.2897,
      "step": 895
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.7569755911827087,
      "learning_rate": 0.0005999776114811664,
      "loss": 3.2771,
      "step": 896
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.7928051948547363,
      "learning_rate": 0.0005999775614795299,
      "loss": 3.2365,
      "step": 897
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.149909257888794,
      "learning_rate": 0.0005999775114221219,
      "loss": 3.4547,
      "step": 898
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.9179600477218628,
      "learning_rate": 0.0005999774613089423,
      "loss": 3.5341,
      "step": 899
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.9396262168884277,
      "learning_rate": 0.0005999774111399913,
      "loss": 3.092,
      "step": 900
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.8275272846221924,
      "learning_rate": 0.0005999773609152688,
      "loss": 3.3443,
      "step": 901
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.9419363141059875,
      "learning_rate": 0.0005999773106347749,
      "loss": 3.412,
      "step": 902
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.9949929118156433,
      "learning_rate": 0.0005999772602985097,
      "loss": 3.4685,
      "step": 903
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.9364804625511169,
      "learning_rate": 0.0005999772099064728,
      "loss": 3.3355,
      "step": 904
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.823532223701477,
      "learning_rate": 0.0005999771594586646,
      "loss": 3.4766,
      "step": 905
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.8039122819900513,
      "learning_rate": 0.000599977108955085,
      "loss": 3.5103,
      "step": 906
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.8246879577636719,
      "learning_rate": 0.0005999770583957339,
      "loss": 3.3926,
      "step": 907
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.7773407697677612,
      "learning_rate": 0.0005999770077806115,
      "loss": 3.1797,
      "step": 908
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.8230880498886108,
      "learning_rate": 0.0005999769571097177,
      "loss": 3.1499,
      "step": 909
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.8533528447151184,
      "learning_rate": 0.0005999769063830524,
      "loss": 3.201,
      "step": 910
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.8195852041244507,
      "learning_rate": 0.0005999768556006159,
      "loss": 3.2741,
      "step": 911
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.7543277740478516,
      "learning_rate": 0.0005999768047624079,
      "loss": 3.1567,
      "step": 912
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.734049379825592,
      "learning_rate": 0.0005999767538684286,
      "loss": 3.3849,
      "step": 913
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.91982501745224,
      "learning_rate": 0.000599976702918678,
      "loss": 3.4453,
      "step": 914
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.745570719242096,
      "learning_rate": 0.0005999766519131561,
      "loss": 3.3616,
      "step": 915
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.77304607629776,
      "learning_rate": 0.0005999766008518627,
      "loss": 3.3152,
      "step": 916
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.8050276041030884,
      "learning_rate": 0.0005999765497347982,
      "loss": 3.2017,
      "step": 917
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.8315979242324829,
      "learning_rate": 0.0005999764985619622,
      "loss": 3.1504,
      "step": 918
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.7340912222862244,
      "learning_rate": 0.000599976447333355,
      "loss": 3.1238,
      "step": 919
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.7896406054496765,
      "learning_rate": 0.0005999763960489766,
      "loss": 3.2541,
      "step": 920
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.8371152877807617,
      "learning_rate": 0.0005999763447088269,
      "loss": 3.302,
      "step": 921
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.320132851600647,
      "learning_rate": 0.0005999762933129059,
      "loss": 3.3192,
      "step": 922
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.755148708820343,
      "learning_rate": 0.0005999762418612135,
      "loss": 3.1898,
      "step": 923
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.8962346315383911,
      "learning_rate": 0.00059997619035375,
      "loss": 3.294,
      "step": 924
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.7845365405082703,
      "learning_rate": 0.0005999761387905152,
      "loss": 3.219,
      "step": 925
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.8395890593528748,
      "learning_rate": 0.0005999760871715093,
      "loss": 3.0859,
      "step": 926
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.1017383337020874,
      "learning_rate": 0.0005999760354967322,
      "loss": 3.1794,
      "step": 927
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.9495001435279846,
      "learning_rate": 0.0005999759837661837,
      "loss": 3.3505,
      "step": 928
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.8175073862075806,
      "learning_rate": 0.0005999759319798641,
      "loss": 3.3059,
      "step": 929
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.7522081732749939,
      "learning_rate": 0.0005999758801377733,
      "loss": 3.0807,
      "step": 930
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.9382736682891846,
      "learning_rate": 0.0005999758282399113,
      "loss": 3.0371,
      "step": 931
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.7736467719078064,
      "learning_rate": 0.0005999757762862782,
      "loss": 2.9666,
      "step": 932
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.9363311529159546,
      "learning_rate": 0.0005999757242768738,
      "loss": 3.2678,
      "step": 933
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.8351675868034363,
      "learning_rate": 0.0005999756722116985,
      "loss": 3.3175,
      "step": 934
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.8093544244766235,
      "learning_rate": 0.0005999756200907518,
      "loss": 3.365,
      "step": 935
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.8454013466835022,
      "learning_rate": 0.0005999755679140341,
      "loss": 3.0995,
      "step": 936
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.8721874356269836,
      "learning_rate": 0.0005999755156815453,
      "loss": 3.1481,
      "step": 937
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.9051503539085388,
      "learning_rate": 0.0005999754633932854,
      "loss": 3.2205,
      "step": 938
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.8022783994674683,
      "learning_rate": 0.0005999754110492543,
      "loss": 3.2506,
      "step": 939
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.8098456263542175,
      "learning_rate": 0.0005999753586494521,
      "loss": 3.3427,
      "step": 940
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.0588864088058472,
      "learning_rate": 0.0005999753061938789,
      "loss": 3.2742,
      "step": 941
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.810872495174408,
      "learning_rate": 0.0005999752536825346,
      "loss": 3.3245,
      "step": 942
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.7351536154747009,
      "learning_rate": 0.0005999752011154193,
      "loss": 3.3074,
      "step": 943
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.0267196893692017,
      "learning_rate": 0.0005999751484925328,
      "loss": 3.1871,
      "step": 944
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.7646951079368591,
      "learning_rate": 0.0005999750958138754,
      "loss": 3.3696,
      "step": 945
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.911105215549469,
      "learning_rate": 0.000599975043079447,
      "loss": 3.4375,
      "step": 946
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.8438509702682495,
      "learning_rate": 0.0005999749902892475,
      "loss": 3.3997,
      "step": 947
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.7844448089599609,
      "learning_rate": 0.000599974937443277,
      "loss": 3.2008,
      "step": 948
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.7930577993392944,
      "learning_rate": 0.0005999748845415354,
      "loss": 3.1705,
      "step": 949
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.8296140432357788,
      "learning_rate": 0.0005999748315840229,
      "loss": 3.1228,
      "step": 950
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.7420133948326111,
      "learning_rate": 0.0005999747785707394,
      "loss": 3.2924,
      "step": 951
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.7674648761749268,
      "learning_rate": 0.000599974725501685,
      "loss": 2.9717,
      "step": 952
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.7592647671699524,
      "learning_rate": 0.0005999746723768595,
      "loss": 3.2449,
      "step": 953
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.8280392289161682,
      "learning_rate": 0.0005999746191962631,
      "loss": 3.1273,
      "step": 954
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.744523823261261,
      "learning_rate": 0.0005999745659598958,
      "loss": 2.9715,
      "step": 955
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.8949483633041382,
      "learning_rate": 0.0005999745126677575,
      "loss": 2.8226,
      "step": 956
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.8055506944656372,
      "learning_rate": 0.0005999744593198483,
      "loss": 3.3655,
      "step": 957
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.8287325501441956,
      "learning_rate": 0.0005999744059161682,
      "loss": 3.386,
      "step": 958
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.8468881845474243,
      "learning_rate": 0.0005999743524567172,
      "loss": 3.2552,
      "step": 959
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.803646445274353,
      "learning_rate": 0.0005999742989414952,
      "loss": 3.2279,
      "step": 960
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.7658111453056335,
      "learning_rate": 0.0005999742453705025,
      "loss": 3.646,
      "step": 961
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.1433824300765991,
      "learning_rate": 0.0005999741917437388,
      "loss": 3.1497,
      "step": 962
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.7451978921890259,
      "learning_rate": 0.0005999741380612043,
      "loss": 3.1786,
      "step": 963
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.9449644684791565,
      "learning_rate": 0.0005999740843228988,
      "loss": 3.2743,
      "step": 964
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.9123852849006653,
      "learning_rate": 0.0005999740305288226,
      "loss": 3.117,
      "step": 965
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.1655330657958984,
      "learning_rate": 0.0005999739766789756,
      "loss": 3.2334,
      "step": 966
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.9363653063774109,
      "learning_rate": 0.0005999739227733577,
      "loss": 3.2247,
      "step": 967
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.7191228866577148,
      "learning_rate": 0.0005999738688119689,
      "loss": 3.0193,
      "step": 968
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.2715610265731812,
      "learning_rate": 0.0005999738147948095,
      "loss": 3.0768,
      "step": 969
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.6953585147857666,
      "learning_rate": 0.0005999737607218791,
      "loss": 2.9978,
      "step": 970
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.7894326448440552,
      "learning_rate": 0.000599973706593178,
      "loss": 3.2833,
      "step": 971
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.8002513647079468,
      "learning_rate": 0.0005999736524087061,
      "loss": 3.1605,
      "step": 972
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.8323729038238525,
      "learning_rate": 0.0005999735981684636,
      "loss": 3.3804,
      "step": 973
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.7737404704093933,
      "learning_rate": 0.0005999735438724502,
      "loss": 3.0634,
      "step": 974
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.0668045282363892,
      "learning_rate": 0.000599973489520666,
      "loss": 3.0501,
      "step": 975
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.8560475707054138,
      "learning_rate": 0.0005999734351131112,
      "loss": 3.1911,
      "step": 976
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.8539261221885681,
      "learning_rate": 0.0005999733806497856,
      "loss": 3.0787,
      "step": 977
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.7208546996116638,
      "learning_rate": 0.0005999733261306893,
      "loss": 3.0815,
      "step": 978
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.7949949502944946,
      "learning_rate": 0.0005999732715558224,
      "loss": 3.2914,
      "step": 979
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.0123343467712402,
      "learning_rate": 0.0005999732169251847,
      "loss": 3.2651,
      "step": 980
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.7942947149276733,
      "learning_rate": 0.0005999731622387763,
      "loss": 3.1001,
      "step": 981
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.0998497009277344,
      "learning_rate": 0.0005999731074965972,
      "loss": 3.248,
      "step": 982
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.6929829120635986,
      "learning_rate": 0.0005999730526986475,
      "loss": 3.3534,
      "step": 983
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.7850948572158813,
      "learning_rate": 0.0005999729978449271,
      "loss": 3.185,
      "step": 984
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.7910817861557007,
      "learning_rate": 0.000599972942935436,
      "loss": 3.173,
      "step": 985
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.7938651442527771,
      "learning_rate": 0.0005999728879701744,
      "loss": 3.2571,
      "step": 986
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.7618404030799866,
      "learning_rate": 0.0005999728329491421,
      "loss": 3.3346,
      "step": 987
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.0399407148361206,
      "learning_rate": 0.0005999727778723393,
      "loss": 3.3465,
      "step": 988
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.7477458119392395,
      "learning_rate": 0.0005999727227397658,
      "loss": 3.3197,
      "step": 989
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.8651089072227478,
      "learning_rate": 0.0005999726675514217,
      "loss": 3.3435,
      "step": 990
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.7483588457107544,
      "learning_rate": 0.0005999726123073071,
      "loss": 3.2168,
      "step": 991
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.8895077705383301,
      "learning_rate": 0.0005999725570074218,
      "loss": 3.4159,
      "step": 992
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.70838862657547,
      "learning_rate": 0.0005999725016517661,
      "loss": 3.2084,
      "step": 993
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.7113087177276611,
      "learning_rate": 0.0005999724462403398,
      "loss": 3.2786,
      "step": 994
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.8079417943954468,
      "learning_rate": 0.0005999723907731428,
      "loss": 3.1876,
      "step": 995
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.8162917494773865,
      "learning_rate": 0.0005999723352501754,
      "loss": 3.2637,
      "step": 996
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.8508199453353882,
      "learning_rate": 0.0005999722796714375,
      "loss": 3.314,
      "step": 997
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.7570256590843201,
      "learning_rate": 0.0005999722240369291,
      "loss": 3.2727,
      "step": 998
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.8637495636940002,
      "learning_rate": 0.0005999721683466501,
      "loss": 3.3213,
      "step": 999
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.8875898122787476,
      "learning_rate": 0.0005999721126006008,
      "loss": 3.2964,
      "step": 1000
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.8888888955116272,
      "learning_rate": 0.0005999720567987809,
      "loss": 3.2095,
      "step": 1001
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.9117664098739624,
      "learning_rate": 0.0005999720009411905,
      "loss": 3.2336,
      "step": 1002
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.8036960959434509,
      "learning_rate": 0.0005999719450278296,
      "loss": 3.0975,
      "step": 1003
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.7982900142669678,
      "learning_rate": 0.0005999718890586983,
      "loss": 3.0721,
      "step": 1004
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.9107368588447571,
      "learning_rate": 0.0005999718330337966,
      "loss": 2.9866,
      "step": 1005
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.9616687297821045,
      "learning_rate": 0.0005999717769531245,
      "loss": 3.1131,
      "step": 1006
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.9235131740570068,
      "learning_rate": 0.000599971720816682,
      "loss": 3.1375,
      "step": 1007
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.0693320035934448,
      "learning_rate": 0.000599971664624469,
      "loss": 3.3017,
      "step": 1008
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.7139890789985657,
      "learning_rate": 0.0005999716083764857,
      "loss": 3.1581,
      "step": 1009
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.1308497190475464,
      "learning_rate": 0.000599971552072732,
      "loss": 2.9762,
      "step": 1010
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.9742271304130554,
      "learning_rate": 0.0005999714957132078,
      "loss": 2.9227,
      "step": 1011
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.8929553627967834,
      "learning_rate": 0.0005999714392979133,
      "loss": 3.2373,
      "step": 1012
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.065039873123169,
      "learning_rate": 0.0005999713828268485,
      "loss": 3.2495,
      "step": 1013
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.8397495150566101,
      "learning_rate": 0.0005999713263000133,
      "loss": 3.4009,
      "step": 1014
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.7782397270202637,
      "learning_rate": 0.0005999712697174078,
      "loss": 3.2867,
      "step": 1015
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.039871335029602,
      "learning_rate": 0.000599971213079032,
      "loss": 3.1669,
      "step": 1016
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.8527975678443909,
      "learning_rate": 0.0005999711563848859,
      "loss": 3.4778,
      "step": 1017
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.9281208515167236,
      "learning_rate": 0.0005999710996349695,
      "loss": 3.2084,
      "step": 1018
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.8596192002296448,
      "learning_rate": 0.0005999710428292828,
      "loss": 3.2944,
      "step": 1019
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.7145438194274902,
      "learning_rate": 0.0005999709859678257,
      "loss": 3.2145,
      "step": 1020
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.9110297560691833,
      "learning_rate": 0.0005999709290505985,
      "loss": 3.3455,
      "step": 1021
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.2129672765731812,
      "learning_rate": 0.0005999708720776009,
      "loss": 3.1231,
      "step": 1022
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.845855176448822,
      "learning_rate": 0.0005999708150488331,
      "loss": 3.2452,
      "step": 1023
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.2743784189224243,
      "learning_rate": 0.0005999707579642952,
      "loss": 3.3961,
      "step": 1024
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.8357268571853638,
      "learning_rate": 0.0005999707008239869,
      "loss": 3.0207,
      "step": 1025
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.0586267709732056,
      "learning_rate": 0.0005999706436279085,
      "loss": 3.3595,
      "step": 1026
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.8415248394012451,
      "learning_rate": 0.0005999705863760598,
      "loss": 2.8374,
      "step": 1027
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.7049532532691956,
      "learning_rate": 0.000599970529068441,
      "loss": 3.2651,
      "step": 1028
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.8434281349182129,
      "learning_rate": 0.0005999704717050519,
      "loss": 2.9709,
      "step": 1029
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.7931018471717834,
      "learning_rate": 0.0005999704142858927,
      "loss": 3.4521,
      "step": 1030
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.7711025476455688,
      "learning_rate": 0.0005999703568109633,
      "loss": 3.2821,
      "step": 1031
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.8913404941558838,
      "learning_rate": 0.0005999702992802638,
      "loss": 3.2638,
      "step": 1032
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.137822151184082,
      "learning_rate": 0.0005999702416937941,
      "loss": 3.5526,
      "step": 1033
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.7401607632637024,
      "learning_rate": 0.0005999701840515543,
      "loss": 3.3368,
      "step": 1034
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.906863808631897,
      "learning_rate": 0.0005999701263535444,
      "loss": 3.345,
      "step": 1035
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.8230988383293152,
      "learning_rate": 0.0005999700685997644,
      "loss": 3.2055,
      "step": 1036
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.13247811794281,
      "learning_rate": 0.0005999700107902142,
      "loss": 3.3716,
      "step": 1037
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.8463082313537598,
      "learning_rate": 0.000599969952924894,
      "loss": 3.306,
      "step": 1038
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.0613385438919067,
      "learning_rate": 0.0005999698950038037,
      "loss": 3.0394,
      "step": 1039
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.7503798604011536,
      "learning_rate": 0.0005999698370269434,
      "loss": 3.3584,
      "step": 1040
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.9245538711547852,
      "learning_rate": 0.000599969778994313,
      "loss": 3.3815,
      "step": 1041
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.9340380430221558,
      "learning_rate": 0.0005999697209059125,
      "loss": 3.0874,
      "step": 1042
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.7950480580329895,
      "learning_rate": 0.0005999696627617421,
      "loss": 2.9819,
      "step": 1043
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.9342819452285767,
      "learning_rate": 0.0005999696045618016,
      "loss": 3.2342,
      "step": 1044
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.97747802734375,
      "learning_rate": 0.0005999695463060911,
      "loss": 3.1744,
      "step": 1045
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.9325199723243713,
      "learning_rate": 0.0005999694879946104,
      "loss": 3.1765,
      "step": 1046
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.30983567237854,
      "learning_rate": 0.00059996942962736,
      "loss": 3.1973,
      "step": 1047
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.7985244989395142,
      "learning_rate": 0.0005999693712043395,
      "loss": 3.0764,
      "step": 1048
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.771816074848175,
      "learning_rate": 0.000599969312725549,
      "loss": 3.1579,
      "step": 1049
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.8829121589660645,
      "learning_rate": 0.0005999692541909886,
      "loss": 3.2733,
      "step": 1050
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.9923681616783142,
      "learning_rate": 0.0005999691956006582,
      "loss": 3.0581,
      "step": 1051
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.8298034071922302,
      "learning_rate": 0.0005999691369545579,
      "loss": 3.1887,
      "step": 1052
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.9441575407981873,
      "learning_rate": 0.0005999690782526878,
      "loss": 3.2838,
      "step": 1053
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.231813669204712,
      "learning_rate": 0.0005999690194950476,
      "loss": 3.2265,
      "step": 1054
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.8083035349845886,
      "learning_rate": 0.0005999689606816375,
      "loss": 3.1902,
      "step": 1055
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.8884376287460327,
      "learning_rate": 0.0005999689018124576,
      "loss": 3.3561,
      "step": 1056
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.0242387056350708,
      "learning_rate": 0.0005999688428875078,
      "loss": 3.1584,
      "step": 1057
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.9227923154830933,
      "learning_rate": 0.0005999687839067881,
      "loss": 3.3312,
      "step": 1058
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.7554488778114319,
      "learning_rate": 0.0005999687248702985,
      "loss": 3.0384,
      "step": 1059
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.0374337434768677,
      "learning_rate": 0.0005999686657780391,
      "loss": 2.8733,
      "step": 1060
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.823466956615448,
      "learning_rate": 0.00059996860663001,
      "loss": 3.2523,
      "step": 1061
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.7275535464286804,
      "learning_rate": 0.0005999685474262109,
      "loss": 3.0216,
      "step": 1062
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.8510707020759583,
      "learning_rate": 0.0005999684881666421,
      "loss": 3.1785,
      "step": 1063
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.8917899131774902,
      "learning_rate": 0.0005999684288513033,
      "loss": 2.8969,
      "step": 1064
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.8587782382965088,
      "learning_rate": 0.0005999683694801948,
      "loss": 3.086,
      "step": 1065
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.8957409858703613,
      "learning_rate": 0.0005999683100533167,
      "loss": 3.0636,
      "step": 1066
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.8099112510681152,
      "learning_rate": 0.0005999682505706686,
      "loss": 3.1503,
      "step": 1067
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.7475759983062744,
      "learning_rate": 0.0005999681910322509,
      "loss": 2.9836,
      "step": 1068
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.8968919515609741,
      "learning_rate": 0.0005999681314380632,
      "loss": 3.294,
      "step": 1069
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.7861024141311646,
      "learning_rate": 0.000599968071788106,
      "loss": 3.2411,
      "step": 1070
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.7743328213691711,
      "learning_rate": 0.0005999680120823789,
      "loss": 3.4172,
      "step": 1071
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.7286455631256104,
      "learning_rate": 0.0005999679523208823,
      "loss": 3.1161,
      "step": 1072
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.186294436454773,
      "learning_rate": 0.0005999678925036159,
      "loss": 3.043,
      "step": 1073
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.8022221922874451,
      "learning_rate": 0.0005999678326305798,
      "loss": 2.986,
      "step": 1074
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.4452162981033325,
      "learning_rate": 0.000599967772701774,
      "loss": 3.0092,
      "step": 1075
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.9518951773643494,
      "learning_rate": 0.0005999677127171985,
      "loss": 2.8525,
      "step": 1076
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.9676651358604431,
      "learning_rate": 0.0005999676526768534,
      "loss": 3.0055,
      "step": 1077
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.8458954095840454,
      "learning_rate": 0.0005999675925807386,
      "loss": 3.1981,
      "step": 1078
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.8877878785133362,
      "learning_rate": 0.0005999675324288542,
      "loss": 3.0428,
      "step": 1079
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.744545042514801,
      "learning_rate": 0.0005999674722212002,
      "loss": 3.327,
      "step": 1080
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.9303997755050659,
      "learning_rate": 0.0005999674119577765,
      "loss": 3.3824,
      "step": 1081
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.7365705966949463,
      "learning_rate": 0.0005999673516385832,
      "loss": 3.2044,
      "step": 1082
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.7762224078178406,
      "learning_rate": 0.0005999672912636204,
      "loss": 3.0289,
      "step": 1083
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.7530665397644043,
      "learning_rate": 0.000599967230832888,
      "loss": 3.1007,
      "step": 1084
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.7415132522583008,
      "learning_rate": 0.000599967170346386,
      "loss": 3.3053,
      "step": 1085
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.8976336121559143,
      "learning_rate": 0.0005999671098041144,
      "loss": 3.5643,
      "step": 1086
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.7471707463264465,
      "learning_rate": 0.0005999670492060733,
      "loss": 2.7876,
      "step": 1087
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.7541609406471252,
      "learning_rate": 0.0005999669885522627,
      "loss": 3.3514,
      "step": 1088
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.8504964113235474,
      "learning_rate": 0.0005999669278426825,
      "loss": 3.0283,
      "step": 1089
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.7986454963684082,
      "learning_rate": 0.0005999668670773327,
      "loss": 3.2912,
      "step": 1090
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.7862951755523682,
      "learning_rate": 0.0005999668062562135,
      "loss": 3.2828,
      "step": 1091
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.7656617760658264,
      "learning_rate": 0.000599966745379325,
      "loss": 3.2632,
      "step": 1092
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.9661446213722229,
      "learning_rate": 0.0005999666844466667,
      "loss": 3.4327,
      "step": 1093
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.7249305844306946,
      "learning_rate": 0.0005999666234582391,
      "loss": 3.2017,
      "step": 1094
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.893046498298645,
      "learning_rate": 0.000599966562414042,
      "loss": 2.8909,
      "step": 1095
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.747602105140686,
      "learning_rate": 0.0005999665013140755,
      "loss": 3.0232,
      "step": 1096
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.9594663381576538,
      "learning_rate": 0.0005999664401583395,
      "loss": 2.9737,
      "step": 1097
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.7281781435012817,
      "learning_rate": 0.0005999663789468342,
      "loss": 2.9006,
      "step": 1098
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.9128285646438599,
      "learning_rate": 0.0005999663176795594,
      "loss": 3.0489,
      "step": 1099
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.7814525365829468,
      "learning_rate": 0.0005999662563565151,
      "loss": 3.2447,
      "step": 1100
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.7785247564315796,
      "learning_rate": 0.0005999661949777015,
      "loss": 2.9808,
      "step": 1101
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.8611516952514648,
      "learning_rate": 0.0005999661335431186,
      "loss": 3.2982,
      "step": 1102
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.9337912797927856,
      "learning_rate": 0.0005999660720527662,
      "loss": 3.1704,
      "step": 1103
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.7416540384292603,
      "learning_rate": 0.0005999660105066444,
      "loss": 3.0363,
      "step": 1104
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.7599864602088928,
      "learning_rate": 0.0005999659489047534,
      "loss": 3.1203,
      "step": 1105
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.7450264692306519,
      "learning_rate": 0.0005999658872470931,
      "loss": 2.933,
      "step": 1106
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.7615723013877869,
      "learning_rate": 0.0005999658255336634,
      "loss": 3.432,
      "step": 1107
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.7043272256851196,
      "learning_rate": 0.0005999657637644644,
      "loss": 3.1481,
      "step": 1108
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.916090726852417,
      "learning_rate": 0.0005999657019394962,
      "loss": 3.2226,
      "step": 1109
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.7984136939048767,
      "learning_rate": 0.0005999656400587585,
      "loss": 3.1717,
      "step": 1110
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.7991804480552673,
      "learning_rate": 0.0005999655781222516,
      "loss": 3.1973,
      "step": 1111
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.9701079726219177,
      "learning_rate": 0.0005999655161299755,
      "loss": 3.1195,
      "step": 1112
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.9655210375785828,
      "learning_rate": 0.0005999654540819302,
      "loss": 3.1086,
      "step": 1113
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.8417152762413025,
      "learning_rate": 0.0005999653919781156,
      "loss": 3.1367,
      "step": 1114
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.7660375237464905,
      "learning_rate": 0.0005999653298185317,
      "loss": 3.4394,
      "step": 1115
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.0324054956436157,
      "learning_rate": 0.0005999652676031787,
      "loss": 3.1873,
      "step": 1116
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.7769565582275391,
      "learning_rate": 0.0005999652053320564,
      "loss": 3.178,
      "step": 1117
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.0605767965316772,
      "learning_rate": 0.0005999651430051649,
      "loss": 3.1983,
      "step": 1118
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.7969644069671631,
      "learning_rate": 0.0005999650806225043,
      "loss": 3.0867,
      "step": 1119
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.9952507019042969,
      "learning_rate": 0.0005999650181840745,
      "loss": 3.2703,
      "step": 1120
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.9452176690101624,
      "learning_rate": 0.0005999649556898755,
      "loss": 3.3363,
      "step": 1121
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.9965848326683044,
      "learning_rate": 0.0005999648931399073,
      "loss": 3.3786,
      "step": 1122
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.741960883140564,
      "learning_rate": 0.0005999648305341701,
      "loss": 3.2807,
      "step": 1123
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.8655504584312439,
      "learning_rate": 0.0005999647678726637,
      "loss": 3.2769,
      "step": 1124
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.7500970959663391,
      "learning_rate": 0.0005999647051553882,
      "loss": 2.8701,
      "step": 1125
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.7620172500610352,
      "learning_rate": 0.0005999646423823437,
      "loss": 3.3753,
      "step": 1126
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.7470266819000244,
      "learning_rate": 0.00059996457955353,
      "loss": 3.107,
      "step": 1127
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.7812240123748779,
      "learning_rate": 0.0005999645166689472,
      "loss": 3.3875,
      "step": 1128
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.9735129475593567,
      "learning_rate": 0.0005999644537285953,
      "loss": 3.1482,
      "step": 1129
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.7803528904914856,
      "learning_rate": 0.0005999643907324745,
      "loss": 3.045,
      "step": 1130
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.9471489191055298,
      "learning_rate": 0.0005999643276805846,
      "loss": 3.4245,
      "step": 1131
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.7451218962669373,
      "learning_rate": 0.0005999642645729257,
      "loss": 3.3632,
      "step": 1132
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.1972061395645142,
      "learning_rate": 0.0005999642014094976,
      "loss": 3.0368,
      "step": 1133
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.7065299153327942,
      "learning_rate": 0.0005999641381903006,
      "loss": 3.0903,
      "step": 1134
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.9648718237876892,
      "learning_rate": 0.0005999640749153347,
      "loss": 3.0462,
      "step": 1135
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.8870189785957336,
      "learning_rate": 0.0005999640115845997,
      "loss": 3.1746,
      "step": 1136
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.8566933274269104,
      "learning_rate": 0.0005999639481980958,
      "loss": 3.1763,
      "step": 1137
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.8868460655212402,
      "learning_rate": 0.0005999638847558229,
      "loss": 3.166,
      "step": 1138
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.7874283790588379,
      "learning_rate": 0.0005999638212577811,
      "loss": 3.014,
      "step": 1139
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.9819658398628235,
      "learning_rate": 0.0005999637577039703,
      "loss": 3.0374,
      "step": 1140
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.7300655245780945,
      "learning_rate": 0.0005999636940943906,
      "loss": 3.0275,
      "step": 1141
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.855124294757843,
      "learning_rate": 0.000599963630429042,
      "loss": 2.8375,
      "step": 1142
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.9029626846313477,
      "learning_rate": 0.0005999635667079244,
      "loss": 3.1867,
      "step": 1143
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.7677856683731079,
      "learning_rate": 0.0005999635029310381,
      "loss": 3.289,
      "step": 1144
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.9405258297920227,
      "learning_rate": 0.0005999634390983828,
      "loss": 3.0275,
      "step": 1145
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.7418372631072998,
      "learning_rate": 0.0005999633752099586,
      "loss": 3.2562,
      "step": 1146
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.8302958607673645,
      "learning_rate": 0.0005999633112657657,
      "loss": 3.2555,
      "step": 1147
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.9565913677215576,
      "learning_rate": 0.0005999632472658038,
      "loss": 3.1772,
      "step": 1148
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.9732460975646973,
      "learning_rate": 0.0005999631832100731,
      "loss": 3.0852,
      "step": 1149
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.2889302968978882,
      "learning_rate": 0.0005999631190985737,
      "loss": 3.3041,
      "step": 1150
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.9608477354049683,
      "learning_rate": 0.0005999630549313054,
      "loss": 3.15,
      "step": 1151
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.7605515122413635,
      "learning_rate": 0.0005999629907082681,
      "loss": 3.036,
      "step": 1152
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.9011639952659607,
      "learning_rate": 0.0005999629264294623,
      "loss": 2.9203,
      "step": 1153
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.9218935966491699,
      "learning_rate": 0.0005999628620948877,
      "loss": 3.023,
      "step": 1154
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8198060989379883,
      "learning_rate": 0.0005999627977045443,
      "loss": 3.0154,
      "step": 1155
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.9794735908508301,
      "learning_rate": 0.0005999627332584322,
      "loss": 3.1267,
      "step": 1156
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.723556637763977,
      "learning_rate": 0.0005999626687565513,
      "loss": 3.0972,
      "step": 1157
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8665453791618347,
      "learning_rate": 0.0005999626041989016,
      "loss": 3.2725,
      "step": 1158
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.1164849996566772,
      "learning_rate": 0.0005999625395854833,
      "loss": 2.7803,
      "step": 1159
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7816228270530701,
      "learning_rate": 0.0005999624749162963,
      "loss": 3.3267,
      "step": 1160
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.9926425218582153,
      "learning_rate": 0.0005999624101913406,
      "loss": 3.1839,
      "step": 1161
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8045907616615295,
      "learning_rate": 0.0005999623454106162,
      "loss": 3.1679,
      "step": 1162
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8245301246643066,
      "learning_rate": 0.0005999622805741232,
      "loss": 3.3304,
      "step": 1163
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8521836996078491,
      "learning_rate": 0.0005999622156818616,
      "loss": 2.9913,
      "step": 1164
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7745464444160461,
      "learning_rate": 0.0005999621507338312,
      "loss": 3.2751,
      "step": 1165
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7974501252174377,
      "learning_rate": 0.0005999620857300322,
      "loss": 3.3368,
      "step": 1166
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.750714898109436,
      "learning_rate": 0.0005999620206704647,
      "loss": 3.0789,
      "step": 1167
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.2495253086090088,
      "learning_rate": 0.0005999619555551285,
      "loss": 2.7119,
      "step": 1168
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.892277717590332,
      "learning_rate": 0.0005999618903840238,
      "loss": 2.9989,
      "step": 1169
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7610248923301697,
      "learning_rate": 0.0005999618251571504,
      "loss": 3.355,
      "step": 1170
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7956582903862,
      "learning_rate": 0.0005999617598745086,
      "loss": 2.9823,
      "step": 1171
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7471626400947571,
      "learning_rate": 0.0005999616945360981,
      "loss": 3.068,
      "step": 1172
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8737244009971619,
      "learning_rate": 0.0005999616291419191,
      "loss": 3.092,
      "step": 1173
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.9097580909729004,
      "learning_rate": 0.0005999615636919716,
      "loss": 3.0829,
      "step": 1174
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8397306203842163,
      "learning_rate": 0.0005999614981862556,
      "loss": 3.2774,
      "step": 1175
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.764947772026062,
      "learning_rate": 0.000599961432624771,
      "loss": 3.0337,
      "step": 1176
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7741892337799072,
      "learning_rate": 0.0005999613670075179,
      "loss": 3.0767,
      "step": 1177
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7725731730461121,
      "learning_rate": 0.0005999613013344964,
      "loss": 3.4439,
      "step": 1178
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.1513020992279053,
      "learning_rate": 0.0005999612356057065,
      "loss": 3.2467,
      "step": 1179
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7585788369178772,
      "learning_rate": 0.000599961169821148,
      "loss": 3.0989,
      "step": 1180
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7457687854766846,
      "learning_rate": 0.0005999611039808211,
      "loss": 3.2422,
      "step": 1181
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.9627360105514526,
      "learning_rate": 0.0005999610380847258,
      "loss": 3.2258,
      "step": 1182
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8841953873634338,
      "learning_rate": 0.000599960972132862,
      "loss": 3.2156,
      "step": 1183
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7833623290061951,
      "learning_rate": 0.0005999609061252298,
      "loss": 2.973,
      "step": 1184
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8285154104232788,
      "learning_rate": 0.0005999608400618294,
      "loss": 2.8281,
      "step": 1185
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8539541959762573,
      "learning_rate": 0.0005999607739426605,
      "loss": 3.0827,
      "step": 1186
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8661805987358093,
      "learning_rate": 0.0005999607077677232,
      "loss": 3.1867,
      "step": 1187
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8207781314849854,
      "learning_rate": 0.0005999606415370175,
      "loss": 3.1308,
      "step": 1188
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7639138102531433,
      "learning_rate": 0.0005999605752505436,
      "loss": 3.1969,
      "step": 1189
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.0922297239303589,
      "learning_rate": 0.0005999605089083012,
      "loss": 2.9403,
      "step": 1190
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7280257940292358,
      "learning_rate": 0.0005999604425102905,
      "loss": 3.2332,
      "step": 1191
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.3094793558120728,
      "learning_rate": 0.0005999603760565116,
      "loss": 3.2717,
      "step": 1192
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7754924893379211,
      "learning_rate": 0.0005999603095469643,
      "loss": 3.2511,
      "step": 1193
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.0374292135238647,
      "learning_rate": 0.0005999602429816488,
      "loss": 3.1622,
      "step": 1194
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7215610146522522,
      "learning_rate": 0.000599960176360565,
      "loss": 3.041,
      "step": 1195
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8848457336425781,
      "learning_rate": 0.0005999601096837129,
      "loss": 2.9986,
      "step": 1196
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.9979071021080017,
      "learning_rate": 0.0005999600429510926,
      "loss": 2.9815,
      "step": 1197
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.9445819854736328,
      "learning_rate": 0.0005999599761627041,
      "loss": 3.3057,
      "step": 1198
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8852051496505737,
      "learning_rate": 0.0005999599093185473,
      "loss": 3.1791,
      "step": 1199
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.746296226978302,
      "learning_rate": 0.0005999598424186222,
      "loss": 3.0338,
      "step": 1200
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8421093821525574,
      "learning_rate": 0.0005999597754629292,
      "loss": 3.3233,
      "step": 1201
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.9872077107429504,
      "learning_rate": 0.0005999597084514678,
      "loss": 3.0993,
      "step": 1202
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8288213014602661,
      "learning_rate": 0.0005999596413842382,
      "loss": 3.0958,
      "step": 1203
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.890599250793457,
      "learning_rate": 0.0005999595742612405,
      "loss": 3.2858,
      "step": 1204
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8327617049217224,
      "learning_rate": 0.0005999595070824747,
      "loss": 3.0565,
      "step": 1205
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.9131150245666504,
      "learning_rate": 0.0005999594398479408,
      "loss": 3.1636,
      "step": 1206
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.712181568145752,
      "learning_rate": 0.0005999593725576387,
      "loss": 3.0161,
      "step": 1207
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.88016676902771,
      "learning_rate": 0.0005999593052115684,
      "loss": 3.189,
      "step": 1208
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7965489029884338,
      "learning_rate": 0.0005999592378097301,
      "loss": 3.2133,
      "step": 1209
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.1803970336914062,
      "learning_rate": 0.0005999591703521237,
      "loss": 2.9713,
      "step": 1210
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8296916484832764,
      "learning_rate": 0.0005999591028387493,
      "loss": 3.0566,
      "step": 1211
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7283281087875366,
      "learning_rate": 0.0005999590352696067,
      "loss": 3.077,
      "step": 1212
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8249499201774597,
      "learning_rate": 0.0005999589676446962,
      "loss": 3.3031,
      "step": 1213
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.0447947978973389,
      "learning_rate": 0.0005999588999640176,
      "loss": 2.8347,
      "step": 1214
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7949090003967285,
      "learning_rate": 0.000599958832227571,
      "loss": 3.2144,
      "step": 1215
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7635548710823059,
      "learning_rate": 0.0005999587644353563,
      "loss": 3.3136,
      "step": 1216
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8541753888130188,
      "learning_rate": 0.0005999586965873738,
      "loss": 2.9432,
      "step": 1217
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7517316937446594,
      "learning_rate": 0.0005999586286836231,
      "loss": 3.1508,
      "step": 1218
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.9065231680870056,
      "learning_rate": 0.0005999585607241046,
      "loss": 3.1263,
      "step": 1219
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.1717177629470825,
      "learning_rate": 0.000599958492708818,
      "loss": 3.0759,
      "step": 1220
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.9596511721611023,
      "learning_rate": 0.0005999584246377636,
      "loss": 3.0384,
      "step": 1221
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.9735271334648132,
      "learning_rate": 0.0005999583565109411,
      "loss": 2.9893,
      "step": 1222
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.9542409181594849,
      "learning_rate": 0.0005999582883283507,
      "loss": 3.1808,
      "step": 1223
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.108068823814392,
      "learning_rate": 0.0005999582200899926,
      "loss": 3.1137,
      "step": 1224
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.9304248690605164,
      "learning_rate": 0.0005999581517958664,
      "loss": 3.2331,
      "step": 1225
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8621933460235596,
      "learning_rate": 0.0005999580834459723,
      "loss": 3.0952,
      "step": 1226
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.0258021354675293,
      "learning_rate": 0.0005999580150403105,
      "loss": 3.1574,
      "step": 1227
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7642295956611633,
      "learning_rate": 0.0005999579465788806,
      "loss": 3.0516,
      "step": 1228
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.9197223782539368,
      "learning_rate": 0.0005999578780616832,
      "loss": 3.0727,
      "step": 1229
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.0629197359085083,
      "learning_rate": 0.0005999578094887177,
      "loss": 3.2603,
      "step": 1230
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8761627674102783,
      "learning_rate": 0.0005999577408599844,
      "loss": 3.182,
      "step": 1231
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.3680663108825684,
      "learning_rate": 0.0005999576721754834,
      "loss": 3.007,
      "step": 1232
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8354089856147766,
      "learning_rate": 0.0005999576034352145,
      "loss": 3.1858,
      "step": 1233
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8120739459991455,
      "learning_rate": 0.0005999575346391779,
      "loss": 2.9962,
      "step": 1234
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.1183561086654663,
      "learning_rate": 0.0005999574657873735,
      "loss": 3.2597,
      "step": 1235
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8778237104415894,
      "learning_rate": 0.0005999573968798014,
      "loss": 3.1047,
      "step": 1236
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8410761952400208,
      "learning_rate": 0.0005999573279164615,
      "loss": 3.2219,
      "step": 1237
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.9072355031967163,
      "learning_rate": 0.0005999572588973539,
      "loss": 3.1492,
      "step": 1238
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8566491603851318,
      "learning_rate": 0.0005999571898224785,
      "loss": 3.283,
      "step": 1239
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8023739457130432,
      "learning_rate": 0.0005999571206918355,
      "loss": 2.8753,
      "step": 1240
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8397207260131836,
      "learning_rate": 0.0005999570515054247,
      "loss": 3.0994,
      "step": 1241
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7604035139083862,
      "learning_rate": 0.0005999569822632463,
      "loss": 3.2327,
      "step": 1242
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8146730661392212,
      "learning_rate": 0.0005999569129653003,
      "loss": 2.9227,
      "step": 1243
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7887133359909058,
      "learning_rate": 0.0005999568436115865,
      "loss": 3.1927,
      "step": 1244
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7777902483940125,
      "learning_rate": 0.0005999567742021051,
      "loss": 3.1009,
      "step": 1245
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.852233350276947,
      "learning_rate": 0.0005999567047368561,
      "loss": 3.302,
      "step": 1246
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8602052927017212,
      "learning_rate": 0.0005999566352158396,
      "loss": 3.5025,
      "step": 1247
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8677137494087219,
      "learning_rate": 0.0005999565656390554,
      "loss": 2.8431,
      "step": 1248
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7839433550834656,
      "learning_rate": 0.0005999564960065036,
      "loss": 3.2684,
      "step": 1249
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8847722411155701,
      "learning_rate": 0.0005999564263181841,
      "loss": 2.9517,
      "step": 1250
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.0416761636734009,
      "learning_rate": 0.0005999563565740972,
      "loss": 3.1207,
      "step": 1251
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7844823598861694,
      "learning_rate": 0.0005999562867742428,
      "loss": 3.0944,
      "step": 1252
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.0076103210449219,
      "learning_rate": 0.0005999562169186208,
      "loss": 3.0676,
      "step": 1253
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8904355764389038,
      "learning_rate": 0.0005999561470072313,
      "loss": 3.2171,
      "step": 1254
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7748270630836487,
      "learning_rate": 0.0005999560770400742,
      "loss": 2.9974,
      "step": 1255
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.1650474071502686,
      "learning_rate": 0.0005999560070171496,
      "loss": 3.0934,
      "step": 1256
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.079647421836853,
      "learning_rate": 0.0005999559369384576,
      "loss": 3.1508,
      "step": 1257
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7662287354469299,
      "learning_rate": 0.0005999558668039982,
      "loss": 3.1475,
      "step": 1258
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.9390916228294373,
      "learning_rate": 0.0005999557966137711,
      "loss": 3.0785,
      "step": 1259
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8950755596160889,
      "learning_rate": 0.0005999557263677767,
      "loss": 3.237,
      "step": 1260
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7772256135940552,
      "learning_rate": 0.0005999556560660149,
      "loss": 3.2503,
      "step": 1261
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.9420457482337952,
      "learning_rate": 0.0005999555857084856,
      "loss": 3.0888,
      "step": 1262
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.2155040502548218,
      "learning_rate": 0.000599955515295189,
      "loss": 3.3408,
      "step": 1263
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8072608113288879,
      "learning_rate": 0.0005999554448261249,
      "loss": 3.2678,
      "step": 1264
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.0359784364700317,
      "learning_rate": 0.0005999553743012935,
      "loss": 3.0762,
      "step": 1265
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.2039530277252197,
      "learning_rate": 0.0005999553037206948,
      "loss": 3.0331,
      "step": 1266
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7763333320617676,
      "learning_rate": 0.0005999552330843285,
      "loss": 2.9317,
      "step": 1267
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8549830913543701,
      "learning_rate": 0.0005999551623921951,
      "loss": 3.0586,
      "step": 1268
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.9751284718513489,
      "learning_rate": 0.0005999550916442943,
      "loss": 3.0394,
      "step": 1269
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7355383038520813,
      "learning_rate": 0.0005999550208406261,
      "loss": 2.973,
      "step": 1270
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.9636980891227722,
      "learning_rate": 0.0005999549499811907,
      "loss": 3.0426,
      "step": 1271
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7821171283721924,
      "learning_rate": 0.000599954879065988,
      "loss": 3.1191,
      "step": 1272
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7884069681167603,
      "learning_rate": 0.000599954808095018,
      "loss": 2.7866,
      "step": 1273
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.835060179233551,
      "learning_rate": 0.0005999547370682807,
      "loss": 3.0006,
      "step": 1274
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8041236996650696,
      "learning_rate": 0.0005999546659857762,
      "loss": 2.9981,
      "step": 1275
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.6930768489837646,
      "learning_rate": 0.0005999545948475045,
      "loss": 3.03,
      "step": 1276
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8630000948905945,
      "learning_rate": 0.0005999545236534655,
      "loss": 3.3329,
      "step": 1277
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8143981099128723,
      "learning_rate": 0.0005999544524036593,
      "loss": 3.0973,
      "step": 1278
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.6743859648704529,
      "learning_rate": 0.000599954381098086,
      "loss": 2.7895,
      "step": 1279
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7551833391189575,
      "learning_rate": 0.0005999543097367455,
      "loss": 3.1114,
      "step": 1280
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7695192098617554,
      "learning_rate": 0.0005999542383196378,
      "loss": 3.0787,
      "step": 1281
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7790603637695312,
      "learning_rate": 0.0005999541668467628,
      "loss": 3.059,
      "step": 1282
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7571256756782532,
      "learning_rate": 0.0005999540953181209,
      "loss": 3.113,
      "step": 1283
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8794574737548828,
      "learning_rate": 0.0005999540237337117,
      "loss": 3.1753,
      "step": 1284
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7403808236122131,
      "learning_rate": 0.0005999539520935355,
      "loss": 3.1197,
      "step": 1285
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.1124075651168823,
      "learning_rate": 0.0005999538803975923,
      "loss": 3.0538,
      "step": 1286
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.807080864906311,
      "learning_rate": 0.0005999538086458818,
      "loss": 2.9465,
      "step": 1287
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.9060982465744019,
      "learning_rate": 0.0005999537368384042,
      "loss": 2.979,
      "step": 1288
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7782241702079773,
      "learning_rate": 0.0005999536649751596,
      "loss": 2.9883,
      "step": 1289
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7020358443260193,
      "learning_rate": 0.000599953593056148,
      "loss": 3.0571,
      "step": 1290
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8774701356887817,
      "learning_rate": 0.0005999535210813693,
      "loss": 2.9345,
      "step": 1291
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8052990436553955,
      "learning_rate": 0.0005999534490508236,
      "loss": 3.0696,
      "step": 1292
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7866107225418091,
      "learning_rate": 0.000599953376964511,
      "loss": 3.1266,
      "step": 1293
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7869583964347839,
      "learning_rate": 0.0005999533048224314,
      "loss": 3.3295,
      "step": 1294
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7890189290046692,
      "learning_rate": 0.0005999532326245846,
      "loss": 3.1088,
      "step": 1295
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8663771748542786,
      "learning_rate": 0.0005999531603709709,
      "loss": 3.1386,
      "step": 1296
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.791856586933136,
      "learning_rate": 0.0005999530880615904,
      "loss": 3.0379,
      "step": 1297
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8098143339157104,
      "learning_rate": 0.0005999530156964428,
      "loss": 2.9973,
      "step": 1298
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8160732388496399,
      "learning_rate": 0.0005999529432755284,
      "loss": 3.1164,
      "step": 1299
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7552885413169861,
      "learning_rate": 0.000599952870798847,
      "loss": 3.0985,
      "step": 1300
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8049165606498718,
      "learning_rate": 0.0005999527982663988,
      "loss": 3.0753,
      "step": 1301
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.9521748423576355,
      "learning_rate": 0.0005999527256781836,
      "loss": 2.906,
      "step": 1302
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7448561787605286,
      "learning_rate": 0.0005999526530342015,
      "loss": 3.074,
      "step": 1303
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8196133971214294,
      "learning_rate": 0.0005999525803344526,
      "loss": 3.5107,
      "step": 1304
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.988564670085907,
      "learning_rate": 0.000599952507578937,
      "loss": 3.0699,
      "step": 1305
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7978960275650024,
      "learning_rate": 0.0005999524347676543,
      "loss": 3.2286,
      "step": 1306
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8604587316513062,
      "learning_rate": 0.000599952361900605,
      "loss": 3.0306,
      "step": 1307
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8203248381614685,
      "learning_rate": 0.0005999522889777888,
      "loss": 3.0896,
      "step": 1308
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8082907795906067,
      "learning_rate": 0.0005999522159992058,
      "loss": 3.131,
      "step": 1309
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.6997441649436951,
      "learning_rate": 0.0005999521429648559,
      "loss": 3.2626,
      "step": 1310
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8269714117050171,
      "learning_rate": 0.0005999520698747395,
      "loss": 3.1238,
      "step": 1311
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.0387473106384277,
      "learning_rate": 0.0005999519967288562,
      "loss": 3.0048,
      "step": 1312
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7649436593055725,
      "learning_rate": 0.0005999519235272062,
      "loss": 2.9975,
      "step": 1313
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.4218225479125977,
      "learning_rate": 0.0005999518502697894,
      "loss": 3.3552,
      "step": 1314
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8433598279953003,
      "learning_rate": 0.000599951776956606,
      "loss": 3.0252,
      "step": 1315
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.957028329372406,
      "learning_rate": 0.0005999517035876557,
      "loss": 3.1517,
      "step": 1316
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.425141453742981,
      "learning_rate": 0.0005999516301629389,
      "loss": 3.1885,
      "step": 1317
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8055659532546997,
      "learning_rate": 0.0005999515566824555,
      "loss": 2.9978,
      "step": 1318
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.817449152469635,
      "learning_rate": 0.0005999514831462053,
      "loss": 3.0529,
      "step": 1319
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.9673850536346436,
      "learning_rate": 0.0005999514095541885,
      "loss": 3.2205,
      "step": 1320
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8118335604667664,
      "learning_rate": 0.000599951335906405,
      "loss": 3.0361,
      "step": 1321
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8975399732589722,
      "learning_rate": 0.000599951262202855,
      "loss": 3.1003,
      "step": 1322
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8348236680030823,
      "learning_rate": 0.0005999511884435383,
      "loss": 2.9744,
      "step": 1323
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7988766431808472,
      "learning_rate": 0.000599951114628455,
      "loss": 2.8851,
      "step": 1324
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.9972968697547913,
      "learning_rate": 0.0005999510407576052,
      "loss": 2.949,
      "step": 1325
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7282366752624512,
      "learning_rate": 0.0005999509668309889,
      "loss": 3.1454,
      "step": 1326
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8217811584472656,
      "learning_rate": 0.000599950892848606,
      "loss": 3.2894,
      "step": 1327
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.766775906085968,
      "learning_rate": 0.0005999508188104564,
      "loss": 2.96,
      "step": 1328
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8857177495956421,
      "learning_rate": 0.0005999507447165404,
      "loss": 3.2429,
      "step": 1329
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7516603469848633,
      "learning_rate": 0.000599950670566858,
      "loss": 3.0084,
      "step": 1330
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.9424275755882263,
      "learning_rate": 0.0005999505963614091,
      "loss": 3.128,
      "step": 1331
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.941084623336792,
      "learning_rate": 0.0005999505221001936,
      "loss": 3.086,
      "step": 1332
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.9024030566215515,
      "learning_rate": 0.0005999504477832117,
      "loss": 3.0663,
      "step": 1333
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.4124737977981567,
      "learning_rate": 0.0005999503734104633,
      "loss": 3.1837,
      "step": 1334
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8850489258766174,
      "learning_rate": 0.0005999502989819485,
      "loss": 3.1342,
      "step": 1335
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8867300152778625,
      "learning_rate": 0.0005999502244976672,
      "loss": 3.0966,
      "step": 1336
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.9923032522201538,
      "learning_rate": 0.0005999501499576196,
      "loss": 3.0036,
      "step": 1337
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.747525691986084,
      "learning_rate": 0.0005999500753618056,
      "loss": 2.9978,
      "step": 1338
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.894270122051239,
      "learning_rate": 0.0005999500007102252,
      "loss": 3.2264,
      "step": 1339
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.0846078395843506,
      "learning_rate": 0.0005999499260028783,
      "loss": 3.1571,
      "step": 1340
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7230873703956604,
      "learning_rate": 0.0005999498512397652,
      "loss": 2.9254,
      "step": 1341
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8654605746269226,
      "learning_rate": 0.0005999497764208858,
      "loss": 3.047,
      "step": 1342
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.0180740356445312,
      "learning_rate": 0.00059994970154624,
      "loss": 2.9508,
      "step": 1343
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7898190021514893,
      "learning_rate": 0.0005999496266158279,
      "loss": 3.0171,
      "step": 1344
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.0032747983932495,
      "learning_rate": 0.0005999495516296494,
      "loss": 3.2411,
      "step": 1345
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.0814449787139893,
      "learning_rate": 0.0005999494765877048,
      "loss": 3.0853,
      "step": 1346
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7296811938285828,
      "learning_rate": 0.0005999494014899938,
      "loss": 2.9457,
      "step": 1347
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8258246183395386,
      "learning_rate": 0.0005999493263365166,
      "loss": 2.845,
      "step": 1348
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.1068089008331299,
      "learning_rate": 0.0005999492511272732,
      "loss": 3.3527,
      "step": 1349
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7723930478096008,
      "learning_rate": 0.0005999491758622635,
      "loss": 3.2629,
      "step": 1350
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8568795919418335,
      "learning_rate": 0.0005999491005414876,
      "loss": 2.9094,
      "step": 1351
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.831814169883728,
      "learning_rate": 0.0005999490251649454,
      "loss": 3.3687,
      "step": 1352
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8166736364364624,
      "learning_rate": 0.0005999489497326373,
      "loss": 3.0221,
      "step": 1353
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7416441440582275,
      "learning_rate": 0.0005999488742445627,
      "loss": 2.9562,
      "step": 1354
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7871399521827698,
      "learning_rate": 0.0005999487987007221,
      "loss": 3.0796,
      "step": 1355
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.806056022644043,
      "learning_rate": 0.0005999487231011154,
      "loss": 3.4336,
      "step": 1356
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7720430493354797,
      "learning_rate": 0.0005999486474457425,
      "loss": 3.3389,
      "step": 1357
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7475869655609131,
      "learning_rate": 0.0005999485717346035,
      "loss": 3.2981,
      "step": 1358
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8358380794525146,
      "learning_rate": 0.0005999484959676983,
      "loss": 2.8776,
      "step": 1359
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7198745012283325,
      "learning_rate": 0.0005999484201450272,
      "loss": 2.9808,
      "step": 1360
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7155094742774963,
      "learning_rate": 0.00059994834426659,
      "loss": 3.2077,
      "step": 1361
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8922838568687439,
      "learning_rate": 0.0005999482683323866,
      "loss": 3.1085,
      "step": 1362
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8066011071205139,
      "learning_rate": 0.0005999481923424172,
      "loss": 2.9467,
      "step": 1363
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7893433570861816,
      "learning_rate": 0.0005999481162966818,
      "loss": 2.9355,
      "step": 1364
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.095184326171875,
      "learning_rate": 0.0005999480401951804,
      "loss": 3.1159,
      "step": 1365
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.74529629945755,
      "learning_rate": 0.000599947964037913,
      "loss": 2.8147,
      "step": 1366
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8139212131500244,
      "learning_rate": 0.0005999478878248796,
      "loss": 2.8611,
      "step": 1367
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.0028043985366821,
      "learning_rate": 0.0005999478115560802,
      "loss": 3.2512,
      "step": 1368
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7584793567657471,
      "learning_rate": 0.0005999477352315149,
      "loss": 2.8459,
      "step": 1369
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.2132645845413208,
      "learning_rate": 0.0005999476588511837,
      "loss": 3.1812,
      "step": 1370
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7991018891334534,
      "learning_rate": 0.0005999475824150864,
      "loss": 3.0228,
      "step": 1371
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7921807765960693,
      "learning_rate": 0.0005999475059232232,
      "loss": 3.1145,
      "step": 1372
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8330631256103516,
      "learning_rate": 0.0005999474293755942,
      "loss": 2.7866,
      "step": 1373
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7002035975456238,
      "learning_rate": 0.0005999473527721992,
      "loss": 3.1607,
      "step": 1374
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7577264904975891,
      "learning_rate": 0.0005999472761130384,
      "loss": 3.2506,
      "step": 1375
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8269782066345215,
      "learning_rate": 0.0005999471993981117,
      "loss": 2.8854,
      "step": 1376
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7259654402732849,
      "learning_rate": 0.0005999471226274193,
      "loss": 2.9169,
      "step": 1377
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8116534948348999,
      "learning_rate": 0.0005999470458009609,
      "loss": 3.0561,
      "step": 1378
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8215710520744324,
      "learning_rate": 0.0005999469689187367,
      "loss": 3.0312,
      "step": 1379
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.9524262547492981,
      "learning_rate": 0.0005999468919807468,
      "loss": 3.1155,
      "step": 1380
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8153765797615051,
      "learning_rate": 0.000599946814986991,
      "loss": 3.2772,
      "step": 1381
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7376987934112549,
      "learning_rate": 0.0005999467379374696,
      "loss": 3.084,
      "step": 1382
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.9080320000648499,
      "learning_rate": 0.0005999466608321821,
      "loss": 3.0921,
      "step": 1383
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.761940062046051,
      "learning_rate": 0.0005999465836711292,
      "loss": 3.1246,
      "step": 1384
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.9839903712272644,
      "learning_rate": 0.0005999465064543104,
      "loss": 3.2377,
      "step": 1385
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7678260207176208,
      "learning_rate": 0.0005999464291817259,
      "loss": 2.8439,
      "step": 1386
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7122710943222046,
      "learning_rate": 0.0005999463518533757,
      "loss": 2.9406,
      "step": 1387
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8977315425872803,
      "learning_rate": 0.0005999462744692599,
      "loss": 3.123,
      "step": 1388
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.756363034248352,
      "learning_rate": 0.0005999461970293782,
      "loss": 2.905,
      "step": 1389
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.9311947226524353,
      "learning_rate": 0.0005999461195337311,
      "loss": 3.1511,
      "step": 1390
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8360684514045715,
      "learning_rate": 0.0005999460419823183,
      "loss": 3.1805,
      "step": 1391
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8850170373916626,
      "learning_rate": 0.0005999459643751397,
      "loss": 3.0687,
      "step": 1392
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.808295726776123,
      "learning_rate": 0.0005999458867121957,
      "loss": 3.0412,
      "step": 1393
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8546037673950195,
      "learning_rate": 0.000599945808993486,
      "loss": 2.9001,
      "step": 1394
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7530057430267334,
      "learning_rate": 0.0005999457312190106,
      "loss": 2.846,
      "step": 1395
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8662375211715698,
      "learning_rate": 0.0005999456533887698,
      "loss": 3.512,
      "step": 1396
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7890413999557495,
      "learning_rate": 0.0005999455755027633,
      "loss": 3.0224,
      "step": 1397
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8237061500549316,
      "learning_rate": 0.0005999454975609914,
      "loss": 2.9483,
      "step": 1398
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8326157331466675,
      "learning_rate": 0.0005999454195634539,
      "loss": 2.9996,
      "step": 1399
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.6546998620033264,
      "learning_rate": 0.0005999453415101508,
      "loss": 2.9826,
      "step": 1400
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7489100694656372,
      "learning_rate": 0.0005999452634010824,
      "loss": 2.8677,
      "step": 1401
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.9111331105232239,
      "learning_rate": 0.0005999451852362484,
      "loss": 3.117,
      "step": 1402
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.2000938653945923,
      "learning_rate": 0.0005999451070156489,
      "loss": 3.1572,
      "step": 1403
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.9529575705528259,
      "learning_rate": 0.0005999450287392839,
      "loss": 2.8822,
      "step": 1404
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7088766098022461,
      "learning_rate": 0.0005999449504071535,
      "loss": 3.2383,
      "step": 1405
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.9518072605133057,
      "learning_rate": 0.0005999448720192577,
      "loss": 3.0682,
      "step": 1406
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.6904155015945435,
      "learning_rate": 0.0005999447935755966,
      "loss": 2.9514,
      "step": 1407
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8610660433769226,
      "learning_rate": 0.00059994471507617,
      "loss": 2.8686,
      "step": 1408
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7311219573020935,
      "learning_rate": 0.0005999446365209779,
      "loss": 3.0644,
      "step": 1409
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.9166873097419739,
      "learning_rate": 0.0005999445579100205,
      "loss": 3.1773,
      "step": 1410
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7019796371459961,
      "learning_rate": 0.0005999444792432979,
      "loss": 2.9973,
      "step": 1411
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7032312750816345,
      "learning_rate": 0.0005999444005208098,
      "loss": 2.8779,
      "step": 1412
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7672300934791565,
      "learning_rate": 0.0005999443217425564,
      "loss": 3.0467,
      "step": 1413
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7846834063529968,
      "learning_rate": 0.0005999442429085376,
      "loss": 3.2069,
      "step": 1414
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8190429210662842,
      "learning_rate": 0.0005999441640187536,
      "loss": 2.9268,
      "step": 1415
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7111059427261353,
      "learning_rate": 0.0005999440850732044,
      "loss": 2.9416,
      "step": 1416
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.9198386669158936,
      "learning_rate": 0.0005999440060718899,
      "loss": 3.1421,
      "step": 1417
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8325636386871338,
      "learning_rate": 0.0005999439270148101,
      "loss": 3.1259,
      "step": 1418
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7515615224838257,
      "learning_rate": 0.000599943847901965,
      "loss": 3.2002,
      "step": 1419
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.701876163482666,
      "learning_rate": 0.0005999437687333548,
      "loss": 2.8419,
      "step": 1420
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.9248743653297424,
      "learning_rate": 0.0005999436895089792,
      "loss": 3.1174,
      "step": 1421
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7632994651794434,
      "learning_rate": 0.0005999436102288386,
      "loss": 3.0097,
      "step": 1422
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8650075197219849,
      "learning_rate": 0.0005999435308929328,
      "loss": 3.0631,
      "step": 1423
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7565599083900452,
      "learning_rate": 0.0005999434515012618,
      "loss": 3.1615,
      "step": 1424
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.0977505445480347,
      "learning_rate": 0.0005999433720538256,
      "loss": 3.2707,
      "step": 1425
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7666118144989014,
      "learning_rate": 0.0005999432925506243,
      "loss": 2.9368,
      "step": 1426
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8948745131492615,
      "learning_rate": 0.0005999432129916579,
      "loss": 2.9328,
      "step": 1427
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8831580281257629,
      "learning_rate": 0.0005999431333769264,
      "loss": 3.2439,
      "step": 1428
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8540193438529968,
      "learning_rate": 0.0005999430537064297,
      "loss": 3.1448,
      "step": 1429
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8336238861083984,
      "learning_rate": 0.0005999429739801681,
      "loss": 3.0468,
      "step": 1430
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7629430294036865,
      "learning_rate": 0.0005999428941981413,
      "loss": 2.9994,
      "step": 1431
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.798897922039032,
      "learning_rate": 0.0005999428143603496,
      "loss": 3.3137,
      "step": 1432
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.0468724966049194,
      "learning_rate": 0.0005999427344667927,
      "loss": 3.0681,
      "step": 1433
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8243779540061951,
      "learning_rate": 0.0005999426545174708,
      "loss": 3.3199,
      "step": 1434
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.9798418283462524,
      "learning_rate": 0.0005999425745123841,
      "loss": 3.1949,
      "step": 1435
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.9021846055984497,
      "learning_rate": 0.0005999424944515322,
      "loss": 3.0741,
      "step": 1436
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.3241934776306152,
      "learning_rate": 0.0005999424143349155,
      "loss": 3.086,
      "step": 1437
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.9409375786781311,
      "learning_rate": 0.0005999423341625337,
      "loss": 2.9248,
      "step": 1438
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.1320298910140991,
      "learning_rate": 0.000599942253934387,
      "loss": 3.074,
      "step": 1439
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7761962413787842,
      "learning_rate": 0.0005999421736504753,
      "loss": 3.0432,
      "step": 1440
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7789764404296875,
      "learning_rate": 0.0005999420933107988,
      "loss": 3.2616,
      "step": 1441
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8417879343032837,
      "learning_rate": 0.0005999420129153574,
      "loss": 2.9827,
      "step": 1442
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7519844174385071,
      "learning_rate": 0.000599941932464151,
      "loss": 3.0782,
      "step": 1443
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7705081105232239,
      "learning_rate": 0.0005999418519571799,
      "loss": 3.2599,
      "step": 1444
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.6888156533241272,
      "learning_rate": 0.0005999417713944439,
      "loss": 3.1246,
      "step": 1445
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.0745296478271484,
      "learning_rate": 0.000599941690775943,
      "loss": 2.8842,
      "step": 1446
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8585619926452637,
      "learning_rate": 0.0005999416101016773,
      "loss": 3.0164,
      "step": 1447
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8808173537254333,
      "learning_rate": 0.0005999415293716468,
      "loss": 3.2609,
      "step": 1448
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.9160615801811218,
      "learning_rate": 0.0005999414485858515,
      "loss": 2.9746,
      "step": 1449
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7048495411872864,
      "learning_rate": 0.0005999413677442914,
      "loss": 2.8897,
      "step": 1450
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8145028352737427,
      "learning_rate": 0.0005999412868469666,
      "loss": 3.2075,
      "step": 1451
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.9298685789108276,
      "learning_rate": 0.000599941205893877,
      "loss": 2.9647,
      "step": 1452
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7862663269042969,
      "learning_rate": 0.0005999411248850227,
      "loss": 3.0537,
      "step": 1453
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7564418315887451,
      "learning_rate": 0.0005999410438204037,
      "loss": 2.8152,
      "step": 1454
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.0457022190093994,
      "learning_rate": 0.0005999409627000199,
      "loss": 2.9866,
      "step": 1455
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.9500264525413513,
      "learning_rate": 0.0005999408815238716,
      "loss": 3.1806,
      "step": 1456
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7552927136421204,
      "learning_rate": 0.0005999408002919584,
      "loss": 2.9306,
      "step": 1457
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7720341682434082,
      "learning_rate": 0.0005999407190042808,
      "loss": 3.6407,
      "step": 1458
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8658562898635864,
      "learning_rate": 0.0005999406376608384,
      "loss": 3.1038,
      "step": 1459
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7261186838150024,
      "learning_rate": 0.0005999405562616314,
      "loss": 3.1126,
      "step": 1460
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.2149784564971924,
      "learning_rate": 0.0005999404748066597,
      "loss": 2.9326,
      "step": 1461
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8427217602729797,
      "learning_rate": 0.0005999403932959235,
      "loss": 3.1331,
      "step": 1462
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.9121522307395935,
      "learning_rate": 0.0005999403117294227,
      "loss": 3.1253,
      "step": 1463
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8336421251296997,
      "learning_rate": 0.0005999402301071573,
      "loss": 3.1802,
      "step": 1464
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7661100029945374,
      "learning_rate": 0.0005999401484291274,
      "loss": 2.9923,
      "step": 1465
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8426101803779602,
      "learning_rate": 0.0005999400666953329,
      "loss": 3.1831,
      "step": 1466
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.9059759378433228,
      "learning_rate": 0.000599939984905774,
      "loss": 2.8459,
      "step": 1467
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.1826893091201782,
      "learning_rate": 0.0005999399030604505,
      "loss": 3.0483,
      "step": 1468
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8336043357849121,
      "learning_rate": 0.0005999398211593625,
      "loss": 3.013,
      "step": 1469
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8496936559677124,
      "learning_rate": 0.00059993973920251,
      "loss": 3.2445,
      "step": 1470
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7833266854286194,
      "learning_rate": 0.0005999396571898932,
      "loss": 3.0601,
      "step": 1471
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7857226729393005,
      "learning_rate": 0.0005999395751215118,
      "loss": 2.9988,
      "step": 1472
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7595179677009583,
      "learning_rate": 0.0005999394929973661,
      "loss": 3.3454,
      "step": 1473
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7916429042816162,
      "learning_rate": 0.0005999394108174559,
      "loss": 3.0593,
      "step": 1474
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8605343103408813,
      "learning_rate": 0.0005999393285817813,
      "loss": 3.0221,
      "step": 1475
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7738061547279358,
      "learning_rate": 0.0005999392462903424,
      "loss": 2.6625,
      "step": 1476
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8265538215637207,
      "learning_rate": 0.000599939163943139,
      "loss": 2.7387,
      "step": 1477
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.9268002510070801,
      "learning_rate": 0.0005999390815401714,
      "loss": 2.9568,
      "step": 1478
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7741271257400513,
      "learning_rate": 0.0005999389990814393,
      "loss": 2.9802,
      "step": 1479
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.1658806800842285,
      "learning_rate": 0.000599938916566943,
      "loss": 3.2098,
      "step": 1480
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7888190746307373,
      "learning_rate": 0.0005999388339966823,
      "loss": 3.0014,
      "step": 1481
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8807920813560486,
      "learning_rate": 0.0005999387513706574,
      "loss": 2.8887,
      "step": 1482
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.9295074939727783,
      "learning_rate": 0.0005999386686888681,
      "loss": 3.2981,
      "step": 1483
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.744819164276123,
      "learning_rate": 0.0005999385859513147,
      "loss": 3.115,
      "step": 1484
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.0396437644958496,
      "learning_rate": 0.000599938503157997,
      "loss": 2.9219,
      "step": 1485
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7639064192771912,
      "learning_rate": 0.0005999384203089149,
      "loss": 2.9662,
      "step": 1486
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8513237237930298,
      "learning_rate": 0.0005999383374040687,
      "loss": 3.1654,
      "step": 1487
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7526600360870361,
      "learning_rate": 0.0005999382544434584,
      "loss": 2.9256,
      "step": 1488
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7404652833938599,
      "learning_rate": 0.000599938171427084,
      "loss": 2.9312,
      "step": 1489
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8356943130493164,
      "learning_rate": 0.0005999380883549452,
      "loss": 2.8374,
      "step": 1490
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7958910465240479,
      "learning_rate": 0.0005999380052270424,
      "loss": 2.9161,
      "step": 1491
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.834613025188446,
      "learning_rate": 0.0005999379220433753,
      "loss": 3.1004,
      "step": 1492
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7763651609420776,
      "learning_rate": 0.0005999378388039442,
      "loss": 2.9382,
      "step": 1493
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7740482687950134,
      "learning_rate": 0.0005999377555087489,
      "loss": 3.1736,
      "step": 1494
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8166682124137878,
      "learning_rate": 0.0005999376721577896,
      "loss": 3.0994,
      "step": 1495
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7295792698860168,
      "learning_rate": 0.0005999375887510663,
      "loss": 2.9671,
      "step": 1496
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7932668924331665,
      "learning_rate": 0.0005999375052885788,
      "loss": 3.0541,
      "step": 1497
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7679215669631958,
      "learning_rate": 0.0005999374217703273,
      "loss": 2.9452,
      "step": 1498
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8505843877792358,
      "learning_rate": 0.0005999373381963117,
      "loss": 2.9604,
      "step": 1499
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.6807980537414551,
      "learning_rate": 0.0005999372545665322,
      "loss": 2.8719,
      "step": 1500
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8033422231674194,
      "learning_rate": 0.0005999371708809887,
      "loss": 2.9502,
      "step": 1501
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.9215672612190247,
      "learning_rate": 0.0005999370871396811,
      "loss": 2.8638,
      "step": 1502
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.0854122638702393,
      "learning_rate": 0.0005999370033426095,
      "loss": 2.751,
      "step": 1503
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.6791413426399231,
      "learning_rate": 0.0005999369194897742,
      "loss": 3.1491,
      "step": 1504
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7670359015464783,
      "learning_rate": 0.0005999368355811748,
      "loss": 2.9037,
      "step": 1505
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7009246349334717,
      "learning_rate": 0.0005999367516168116,
      "loss": 2.9997,
      "step": 1506
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8460661768913269,
      "learning_rate": 0.0005999366675966842,
      "loss": 3.1843,
      "step": 1507
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7709742188453674,
      "learning_rate": 0.0005999365835207932,
      "loss": 2.9276,
      "step": 1508
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.96630859375,
      "learning_rate": 0.0005999364993891382,
      "loss": 3.139,
      "step": 1509
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8528409600257874,
      "learning_rate": 0.0005999364152017195,
      "loss": 2.9369,
      "step": 1510
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8075109720230103,
      "learning_rate": 0.0005999363309585368,
      "loss": 3.1543,
      "step": 1511
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8343510627746582,
      "learning_rate": 0.0005999362466595902,
      "loss": 3.1304,
      "step": 1512
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7643350958824158,
      "learning_rate": 0.00059993616230488,
      "loss": 3.0442,
      "step": 1513
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8302398920059204,
      "learning_rate": 0.0005999360778944059,
      "loss": 2.9274,
      "step": 1514
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7881988286972046,
      "learning_rate": 0.0005999359934281681,
      "loss": 3.1936,
      "step": 1515
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.826488196849823,
      "learning_rate": 0.0005999359089061664,
      "loss": 3.0463,
      "step": 1516
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.793880045413971,
      "learning_rate": 0.000599935824328401,
      "loss": 2.9946,
      "step": 1517
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8871197700500488,
      "learning_rate": 0.000599935739694872,
      "loss": 2.8546,
      "step": 1518
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8232203125953674,
      "learning_rate": 0.0005999356550055791,
      "loss": 3.1742,
      "step": 1519
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8853620290756226,
      "learning_rate": 0.0005999355702605226,
      "loss": 2.9774,
      "step": 1520
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7880744338035583,
      "learning_rate": 0.0005999354854597024,
      "loss": 3.1669,
      "step": 1521
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8643462061882019,
      "learning_rate": 0.0005999354006031185,
      "loss": 3.1495,
      "step": 1522
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7818781137466431,
      "learning_rate": 0.000599935315690771,
      "loss": 2.9654,
      "step": 1523
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8049386143684387,
      "learning_rate": 0.0005999352307226597,
      "loss": 3.0638,
      "step": 1524
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7884074449539185,
      "learning_rate": 0.000599935145698785,
      "loss": 2.8135,
      "step": 1525
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.9171607494354248,
      "learning_rate": 0.0005999350606191465,
      "loss": 3.0345,
      "step": 1526
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.6825299263000488,
      "learning_rate": 0.0005999349754837447,
      "loss": 2.933,
      "step": 1527
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8597666025161743,
      "learning_rate": 0.000599934890292579,
      "loss": 2.9806,
      "step": 1528
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.0093027353286743,
      "learning_rate": 0.0005999348050456499,
      "loss": 2.9331,
      "step": 1529
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8416862487792969,
      "learning_rate": 0.0005999347197429573,
      "loss": 2.9805,
      "step": 1530
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.1285697221755981,
      "learning_rate": 0.0005999346343845011,
      "loss": 3.1523,
      "step": 1531
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.37631356716156,
      "learning_rate": 0.0005999345489702814,
      "loss": 2.963,
      "step": 1532
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.131454348564148,
      "learning_rate": 0.0005999344635002981,
      "loss": 3.0638,
      "step": 1533
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.2630599737167358,
      "learning_rate": 0.0005999343779745516,
      "loss": 3.1806,
      "step": 1534
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.9598047733306885,
      "learning_rate": 0.0005999342923930414,
      "loss": 3.2213,
      "step": 1535
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8304891586303711,
      "learning_rate": 0.0005999342067557679,
      "loss": 2.9347,
      "step": 1536
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.0710079669952393,
      "learning_rate": 0.0005999341210627308,
      "loss": 2.88,
      "step": 1537
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.892662763595581,
      "learning_rate": 0.0005999340353139304,
      "loss": 3.0156,
      "step": 1538
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7408807873725891,
      "learning_rate": 0.0005999339495093666,
      "loss": 2.9345,
      "step": 1539
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8894253373146057,
      "learning_rate": 0.0005999338636490394,
      "loss": 3.0621,
      "step": 1540
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8503592014312744,
      "learning_rate": 0.0005999337777329489,
      "loss": 3.0554,
      "step": 1541
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8050125241279602,
      "learning_rate": 0.0005999336917610949,
      "loss": 3.0751,
      "step": 1542
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7165399789810181,
      "learning_rate": 0.0005999336057334777,
      "loss": 3.1676,
      "step": 1543
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8100335597991943,
      "learning_rate": 0.0005999335196500971,
      "loss": 2.9434,
      "step": 1544
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8695785403251648,
      "learning_rate": 0.0005999334335109531,
      "loss": 2.9003,
      "step": 1545
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8069173693656921,
      "learning_rate": 0.000599933347316046,
      "loss": 2.9952,
      "step": 1546
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.0141714811325073,
      "learning_rate": 0.0005999332610653757,
      "loss": 2.8949,
      "step": 1547
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.755062997341156,
      "learning_rate": 0.0005999331747589419,
      "loss": 3.0499,
      "step": 1548
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7973271012306213,
      "learning_rate": 0.000599933088396745,
      "loss": 3.0825,
      "step": 1549
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.9314206838607788,
      "learning_rate": 0.0005999330019787848,
      "loss": 2.8776,
      "step": 1550
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7301434874534607,
      "learning_rate": 0.0005999329155050615,
      "loss": 3.0746,
      "step": 1551
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8316800594329834,
      "learning_rate": 0.0005999328289755749,
      "loss": 3.2777,
      "step": 1552
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.9497971534729004,
      "learning_rate": 0.0005999327423903253,
      "loss": 2.9066,
      "step": 1553
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.811190664768219,
      "learning_rate": 0.0005999326557493123,
      "loss": 3.1347,
      "step": 1554
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7696388959884644,
      "learning_rate": 0.0005999325690525363,
      "loss": 3.1874,
      "step": 1555
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7092611789703369,
      "learning_rate": 0.0005999324822999972,
      "loss": 3.0365,
      "step": 1556
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7706802487373352,
      "learning_rate": 0.0005999323954916949,
      "loss": 2.982,
      "step": 1557
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7606653571128845,
      "learning_rate": 0.0005999323086276295,
      "loss": 2.7582,
      "step": 1558
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8450019955635071,
      "learning_rate": 0.0005999322217078011,
      "loss": 2.9418,
      "step": 1559
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7873334288597107,
      "learning_rate": 0.0005999321347322096,
      "loss": 2.984,
      "step": 1560
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7381137609481812,
      "learning_rate": 0.000599932047700855,
      "loss": 3.0109,
      "step": 1561
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.6982640624046326,
      "learning_rate": 0.0005999319606137375,
      "loss": 3.2807,
      "step": 1562
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7771971821784973,
      "learning_rate": 0.0005999318734708569,
      "loss": 2.9785,
      "step": 1563
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7880092263221741,
      "learning_rate": 0.0005999317862722134,
      "loss": 3.0916,
      "step": 1564
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.9223207831382751,
      "learning_rate": 0.0005999316990178067,
      "loss": 2.7168,
      "step": 1565
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8742111325263977,
      "learning_rate": 0.0005999316117076373,
      "loss": 3.1367,
      "step": 1566
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.9274669289588928,
      "learning_rate": 0.0005999315243417048,
      "loss": 2.8544,
      "step": 1567
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.903902530670166,
      "learning_rate": 0.0005999314369200094,
      "loss": 3.2081,
      "step": 1568
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8950431942939758,
      "learning_rate": 0.000599931349442551,
      "loss": 3.1387,
      "step": 1569
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.9289902448654175,
      "learning_rate": 0.0005999312619093299,
      "loss": 2.9065,
      "step": 1570
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8276894092559814,
      "learning_rate": 0.0005999311743203458,
      "loss": 3.0408,
      "step": 1571
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.991442084312439,
      "learning_rate": 0.0005999310866755987,
      "loss": 2.7685,
      "step": 1572
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.830954909324646,
      "learning_rate": 0.0005999309989750889,
      "loss": 3.0746,
      "step": 1573
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.9615793228149414,
      "learning_rate": 0.0005999309112188163,
      "loss": 3.1619,
      "step": 1574
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7711109519004822,
      "learning_rate": 0.0005999308234067807,
      "loss": 3.0566,
      "step": 1575
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.9034106135368347,
      "learning_rate": 0.0005999307355389825,
      "loss": 2.9117,
      "step": 1576
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8018738627433777,
      "learning_rate": 0.0005999306476154214,
      "loss": 3.096,
      "step": 1577
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7411562204360962,
      "learning_rate": 0.0005999305596360976,
      "loss": 3.3948,
      "step": 1578
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7817267179489136,
      "learning_rate": 0.0005999304716010111,
      "loss": 3.1911,
      "step": 1579
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.9737333059310913,
      "learning_rate": 0.0005999303835101618,
      "loss": 3.1293,
      "step": 1580
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8945849537849426,
      "learning_rate": 0.0005999302953635497,
      "loss": 3.0706,
      "step": 1581
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8327854871749878,
      "learning_rate": 0.000599930207161175,
      "loss": 2.9948,
      "step": 1582
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8443467617034912,
      "learning_rate": 0.0005999301189030375,
      "loss": 2.617,
      "step": 1583
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7501587867736816,
      "learning_rate": 0.0005999300305891375,
      "loss": 3.1381,
      "step": 1584
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8081191182136536,
      "learning_rate": 0.0005999299422194748,
      "loss": 2.8648,
      "step": 1585
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7877047061920166,
      "learning_rate": 0.0005999298537940494,
      "loss": 3.0759,
      "step": 1586
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7042189240455627,
      "learning_rate": 0.0005999297653128613,
      "loss": 3.1039,
      "step": 1587
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7529239058494568,
      "learning_rate": 0.0005999296767759106,
      "loss": 3.0357,
      "step": 1588
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8870639801025391,
      "learning_rate": 0.0005999295881831975,
      "loss": 2.8804,
      "step": 1589
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.904926598072052,
      "learning_rate": 0.0005999294995347217,
      "loss": 2.889,
      "step": 1590
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8940550088882446,
      "learning_rate": 0.0005999294108304834,
      "loss": 2.9964,
      "step": 1591
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.9921493530273438,
      "learning_rate": 0.0005999293220704824,
      "loss": 3.1274,
      "step": 1592
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.810862123966217,
      "learning_rate": 0.0005999292332547191,
      "loss": 3.0075,
      "step": 1593
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.113916039466858,
      "learning_rate": 0.0005999291443831931,
      "loss": 2.9939,
      "step": 1594
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.836197555065155,
      "learning_rate": 0.0005999290554559049,
      "loss": 2.6702,
      "step": 1595
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8152380585670471,
      "learning_rate": 0.0005999289664728539,
      "loss": 2.6606,
      "step": 1596
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8907443881034851,
      "learning_rate": 0.0005999288774340407,
      "loss": 3.0043,
      "step": 1597
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8470190167427063,
      "learning_rate": 0.0005999287883394649,
      "loss": 3.2189,
      "step": 1598
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.054215908050537,
      "learning_rate": 0.0005999286991891266,
      "loss": 3.1657,
      "step": 1599
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7664937973022461,
      "learning_rate": 0.0005999286099830261,
      "loss": 3.1525,
      "step": 1600
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8381310105323792,
      "learning_rate": 0.0005999285207211631,
      "loss": 3.0573,
      "step": 1601
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7976519465446472,
      "learning_rate": 0.0005999284314035377,
      "loss": 3.1933,
      "step": 1602
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8035401105880737,
      "learning_rate": 0.00059992834203015,
      "loss": 3.1639,
      "step": 1603
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.6841264367103577,
      "learning_rate": 0.000599928252601,
      "loss": 2.8915,
      "step": 1604
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.9165793061256409,
      "learning_rate": 0.0005999281631160876,
      "loss": 3.4836,
      "step": 1605
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.9300179481506348,
      "learning_rate": 0.0005999280735754129,
      "loss": 2.9699,
      "step": 1606
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7641578316688538,
      "learning_rate": 0.0005999279839789759,
      "loss": 2.8396,
      "step": 1607
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7917534708976746,
      "learning_rate": 0.0005999278943267767,
      "loss": 2.8577,
      "step": 1608
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8176587820053101,
      "learning_rate": 0.0005999278046188151,
      "loss": 2.8511,
      "step": 1609
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8524308204650879,
      "learning_rate": 0.0005999277148550913,
      "loss": 2.957,
      "step": 1610
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.6482193470001221,
      "learning_rate": 0.0005999276250356055,
      "loss": 2.9132,
      "step": 1611
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.9349744915962219,
      "learning_rate": 0.0005999275351603573,
      "loss": 2.7877,
      "step": 1612
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7093232870101929,
      "learning_rate": 0.000599927445229347,
      "loss": 2.833,
      "step": 1613
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.9018207788467407,
      "learning_rate": 0.0005999273552425744,
      "loss": 2.8419,
      "step": 1614
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.728020429611206,
      "learning_rate": 0.0005999272652000397,
      "loss": 3.0105,
      "step": 1615
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7762616276741028,
      "learning_rate": 0.0005999271751017429,
      "loss": 3.0128,
      "step": 1616
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7884990572929382,
      "learning_rate": 0.0005999270849476839,
      "loss": 2.7868,
      "step": 1617
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.6921452283859253,
      "learning_rate": 0.0005999269947378628,
      "loss": 2.9305,
      "step": 1618
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7575113773345947,
      "learning_rate": 0.0005999269044722797,
      "loss": 3.1378,
      "step": 1619
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8149704933166504,
      "learning_rate": 0.0005999268141509344,
      "loss": 3.057,
      "step": 1620
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8533684611320496,
      "learning_rate": 0.0005999267237738271,
      "loss": 2.6839,
      "step": 1621
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7903765439987183,
      "learning_rate": 0.0005999266333409578,
      "loss": 3.0753,
      "step": 1622
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7908089756965637,
      "learning_rate": 0.0005999265428523264,
      "loss": 2.9275,
      "step": 1623
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.801975429058075,
      "learning_rate": 0.000599926452307933,
      "loss": 3.335,
      "step": 1624
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8846064209938049,
      "learning_rate": 0.0005999263617077777,
      "loss": 2.9236,
      "step": 1625
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8065542578697205,
      "learning_rate": 0.0005999262710518604,
      "loss": 3.094,
      "step": 1626
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.9200791716575623,
      "learning_rate": 0.0005999261803401811,
      "loss": 3.1687,
      "step": 1627
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.747754693031311,
      "learning_rate": 0.00059992608957274,
      "loss": 3.1479,
      "step": 1628
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8907409310340881,
      "learning_rate": 0.0005999259987495368,
      "loss": 2.8432,
      "step": 1629
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7733085751533508,
      "learning_rate": 0.0005999259078705718,
      "loss": 2.9997,
      "step": 1630
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7396509051322937,
      "learning_rate": 0.0005999258169358449,
      "loss": 2.9439,
      "step": 1631
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7793645858764648,
      "learning_rate": 0.0005999257259453561,
      "loss": 2.9233,
      "step": 1632
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8412473201751709,
      "learning_rate": 0.0005999256348991054,
      "loss": 3.1653,
      "step": 1633
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8660628199577332,
      "learning_rate": 0.0005999255437970929,
      "loss": 2.9218,
      "step": 1634
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7319201231002808,
      "learning_rate": 0.0005999254526393187,
      "loss": 2.842,
      "step": 1635
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7686276435852051,
      "learning_rate": 0.0005999253614257826,
      "loss": 3.0688,
      "step": 1636
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8067184090614319,
      "learning_rate": 0.0005999252701564846,
      "loss": 3.1842,
      "step": 1637
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8194860219955444,
      "learning_rate": 0.000599925178831425,
      "loss": 2.8227,
      "step": 1638
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8092554807662964,
      "learning_rate": 0.0005999250874506036,
      "loss": 3.1231,
      "step": 1639
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8935441970825195,
      "learning_rate": 0.0005999249960140204,
      "loss": 3.1455,
      "step": 1640
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7736596465110779,
      "learning_rate": 0.0005999249045216755,
      "loss": 2.7905,
      "step": 1641
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.043200135231018,
      "learning_rate": 0.0005999248129735689,
      "loss": 2.9858,
      "step": 1642
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7631661891937256,
      "learning_rate": 0.0005999247213697008,
      "loss": 3.0037,
      "step": 1643
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8573158383369446,
      "learning_rate": 0.0005999246297100708,
      "loss": 3.282,
      "step": 1644
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.9808943867683411,
      "learning_rate": 0.0005999245379946792,
      "loss": 3.0866,
      "step": 1645
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.0674535036087036,
      "learning_rate": 0.0005999244462235261,
      "loss": 2.9824,
      "step": 1646
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.116046667098999,
      "learning_rate": 0.0005999243543966113,
      "loss": 3.0708,
      "step": 1647
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.9291520714759827,
      "learning_rate": 0.0005999242625139348,
      "loss": 3.061,
      "step": 1648
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.827045202255249,
      "learning_rate": 0.0005999241705754968,
      "loss": 2.951,
      "step": 1649
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7627332806587219,
      "learning_rate": 0.0005999240785812972,
      "loss": 2.9576,
      "step": 1650
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7945206165313721,
      "learning_rate": 0.000599923986531336,
      "loss": 3.0634,
      "step": 1651
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7087850570678711,
      "learning_rate": 0.0005999238944256134,
      "loss": 2.9198,
      "step": 1652
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8133813738822937,
      "learning_rate": 0.0005999238022641293,
      "loss": 3.0231,
      "step": 1653
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8937743902206421,
      "learning_rate": 0.0005999237100468836,
      "loss": 2.8738,
      "step": 1654
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7043418884277344,
      "learning_rate": 0.0005999236177738765,
      "loss": 2.9473,
      "step": 1655
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7216187119483948,
      "learning_rate": 0.0005999235254451079,
      "loss": 3.1244,
      "step": 1656
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.69992995262146,
      "learning_rate": 0.0005999234330605779,
      "loss": 2.7102,
      "step": 1657
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7657837867736816,
      "learning_rate": 0.0005999233406202864,
      "loss": 3.2711,
      "step": 1658
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.6793688535690308,
      "learning_rate": 0.0005999232481242335,
      "loss": 3.1928,
      "step": 1659
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7345349788665771,
      "learning_rate": 0.0005999231555724192,
      "loss": 2.8948,
      "step": 1660
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7521305680274963,
      "learning_rate": 0.0005999230629648436,
      "loss": 3.0759,
      "step": 1661
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.9806193709373474,
      "learning_rate": 0.0005999229703015066,
      "loss": 3.0442,
      "step": 1662
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8122025728225708,
      "learning_rate": 0.0005999228775824082,
      "loss": 3.0877,
      "step": 1663
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.0162218809127808,
      "learning_rate": 0.0005999227848075485,
      "loss": 2.9369,
      "step": 1664
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.0011696815490723,
      "learning_rate": 0.0005999226919769276,
      "loss": 3.1429,
      "step": 1665
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.0163570642471313,
      "learning_rate": 0.0005999225990905453,
      "loss": 3.0968,
      "step": 1666
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8713245391845703,
      "learning_rate": 0.0005999225061484017,
      "loss": 3.2063,
      "step": 1667
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8259766697883606,
      "learning_rate": 0.000599922413150497,
      "loss": 2.7138,
      "step": 1668
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8657404184341431,
      "learning_rate": 0.0005999223200968309,
      "loss": 3.1117,
      "step": 1669
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7202569246292114,
      "learning_rate": 0.0005999222269874036,
      "loss": 3.053,
      "step": 1670
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.6952456831932068,
      "learning_rate": 0.0005999221338222153,
      "loss": 3.1029,
      "step": 1671
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7218242287635803,
      "learning_rate": 0.0005999220406012655,
      "loss": 2.8691,
      "step": 1672
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8007674813270569,
      "learning_rate": 0.0005999219473245547,
      "loss": 3.0251,
      "step": 1673
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7204271554946899,
      "learning_rate": 0.0005999218539920828,
      "loss": 2.8982,
      "step": 1674
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8400158882141113,
      "learning_rate": 0.0005999217606038497,
      "loss": 2.7156,
      "step": 1675
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.802024781703949,
      "learning_rate": 0.0005999216671598555,
      "loss": 2.9427,
      "step": 1676
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.870357096195221,
      "learning_rate": 0.0005999215736601002,
      "loss": 3.1481,
      "step": 1677
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7901319861412048,
      "learning_rate": 0.0005999214801045838,
      "loss": 3.3662,
      "step": 1678
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8179908394813538,
      "learning_rate": 0.0005999213864933062,
      "loss": 3.0331,
      "step": 1679
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7172346711158752,
      "learning_rate": 0.0005999212928262677,
      "loss": 2.9712,
      "step": 1680
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.2598453760147095,
      "learning_rate": 0.0005999211991034682,
      "loss": 2.8623,
      "step": 1681
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7269269824028015,
      "learning_rate": 0.0005999211053249077,
      "loss": 3.155,
      "step": 1682
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7068845629692078,
      "learning_rate": 0.0005999210114905862,
      "loss": 3.173,
      "step": 1683
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8849838972091675,
      "learning_rate": 0.0005999209176005035,
      "loss": 2.9417,
      "step": 1684
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7129116058349609,
      "learning_rate": 0.0005999208236546601,
      "loss": 3.0652,
      "step": 1685
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7798347473144531,
      "learning_rate": 0.0005999207296530556,
      "loss": 2.9064,
      "step": 1686
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7593429088592529,
      "learning_rate": 0.0005999206355956903,
      "loss": 3.002,
      "step": 1687
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.6823811531066895,
      "learning_rate": 0.0005999205414825641,
      "loss": 2.8064,
      "step": 1688
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7298662662506104,
      "learning_rate": 0.0005999204473136769,
      "loss": 2.8119,
      "step": 1689
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7466630339622498,
      "learning_rate": 0.0005999203530890289,
      "loss": 2.7255,
      "step": 1690
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7684394717216492,
      "learning_rate": 0.0005999202588086199,
      "loss": 3.0924,
      "step": 1691
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8996901512145996,
      "learning_rate": 0.0005999201644724503,
      "loss": 2.7317,
      "step": 1692
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.6958341598510742,
      "learning_rate": 0.0005999200700805197,
      "loss": 3.1383,
      "step": 1693
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8268656134605408,
      "learning_rate": 0.0005999199756328284,
      "loss": 2.8324,
      "step": 1694
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8047569990158081,
      "learning_rate": 0.0005999198811293762,
      "loss": 3.2138,
      "step": 1695
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.1504390239715576,
      "learning_rate": 0.0005999197865701634,
      "loss": 3.0947,
      "step": 1696
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7936536073684692,
      "learning_rate": 0.0005999196919551897,
      "loss": 2.8928,
      "step": 1697
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8840271830558777,
      "learning_rate": 0.0005999195972844554,
      "loss": 2.9557,
      "step": 1698
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8387977480888367,
      "learning_rate": 0.0005999195025579603,
      "loss": 3.158,
      "step": 1699
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7441126704216003,
      "learning_rate": 0.0005999194077757045,
      "loss": 3.0233,
      "step": 1700
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.4479403495788574,
      "learning_rate": 0.000599919312937688,
      "loss": 3.0466,
      "step": 1701
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7619702219963074,
      "learning_rate": 0.0005999192180439109,
      "loss": 2.8305,
      "step": 1702
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8586984276771545,
      "learning_rate": 0.0005999191230943732,
      "loss": 2.9785,
      "step": 1703
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8150102496147156,
      "learning_rate": 0.0005999190280890748,
      "loss": 2.9963,
      "step": 1704
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7931879162788391,
      "learning_rate": 0.0005999189330280157,
      "loss": 3.138,
      "step": 1705
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.9399305582046509,
      "learning_rate": 0.0005999188379111961,
      "loss": 3.0693,
      "step": 1706
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8207175731658936,
      "learning_rate": 0.0005999187427386159,
      "loss": 2.9082,
      "step": 1707
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8417760729789734,
      "learning_rate": 0.0005999186475102752,
      "loss": 2.9165,
      "step": 1708
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.863400936126709,
      "learning_rate": 0.0005999185522261739,
      "loss": 3.1624,
      "step": 1709
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8355720639228821,
      "learning_rate": 0.0005999184568863122,
      "loss": 2.8151,
      "step": 1710
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8899855613708496,
      "learning_rate": 0.0005999183614906899,
      "loss": 2.775,
      "step": 1711
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7036992311477661,
      "learning_rate": 0.0005999182660393071,
      "loss": 3.0082,
      "step": 1712
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.9880885481834412,
      "learning_rate": 0.000599918170532164,
      "loss": 2.9466,
      "step": 1713
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8782994747161865,
      "learning_rate": 0.0005999180749692602,
      "loss": 2.9725,
      "step": 1714
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8336834907531738,
      "learning_rate": 0.0005999179793505961,
      "loss": 2.7431,
      "step": 1715
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.906253457069397,
      "learning_rate": 0.0005999178836761715,
      "loss": 3.1101,
      "step": 1716
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8358325958251953,
      "learning_rate": 0.0005999177879459866,
      "loss": 2.7401,
      "step": 1717
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7656639218330383,
      "learning_rate": 0.0005999176921600414,
      "loss": 2.9422,
      "step": 1718
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7857221364974976,
      "learning_rate": 0.0005999175963183357,
      "loss": 2.9703,
      "step": 1719
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.6830969452857971,
      "learning_rate": 0.0005999175004208696,
      "loss": 2.9575,
      "step": 1720
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7712628245353699,
      "learning_rate": 0.0005999174044676433,
      "loss": 2.7461,
      "step": 1721
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7705033421516418,
      "learning_rate": 0.0005999173084586567,
      "loss": 2.8412,
      "step": 1722
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.805978000164032,
      "learning_rate": 0.0005999172123939097,
      "loss": 3.3022,
      "step": 1723
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8336998820304871,
      "learning_rate": 0.0005999171162734025,
      "loss": 3.0513,
      "step": 1724
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7456144094467163,
      "learning_rate": 0.000599917020097135,
      "loss": 2.8319,
      "step": 1725
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.9278692007064819,
      "learning_rate": 0.0005999169238651074,
      "loss": 3.0958,
      "step": 1726
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8068827986717224,
      "learning_rate": 0.0005999168275773194,
      "loss": 3.124,
      "step": 1727
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7789250016212463,
      "learning_rate": 0.0005999167312337714,
      "loss": 3.0262,
      "step": 1728
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7789266705513,
      "learning_rate": 0.000599916634834463,
      "loss": 3.1944,
      "step": 1729
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7563624382019043,
      "learning_rate": 0.0005999165383793946,
      "loss": 2.7672,
      "step": 1730
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8252482414245605,
      "learning_rate": 0.0005999164418685661,
      "loss": 2.8187,
      "step": 1731
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7145665884017944,
      "learning_rate": 0.0005999163453019774,
      "loss": 3.2403,
      "step": 1732
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.0013691186904907,
      "learning_rate": 0.0005999162486796285,
      "loss": 2.9143,
      "step": 1733
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8078101873397827,
      "learning_rate": 0.0005999161520015196,
      "loss": 2.8312,
      "step": 1734
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.03291654586792,
      "learning_rate": 0.0005999160552676505,
      "loss": 2.7745,
      "step": 1735
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.9929647445678711,
      "learning_rate": 0.0005999159584780215,
      "loss": 2.8395,
      "step": 1736
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.9059405326843262,
      "learning_rate": 0.0005999158616326323,
      "loss": 2.8361,
      "step": 1737
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.717201292514801,
      "learning_rate": 0.0005999157647314832,
      "loss": 3.0358,
      "step": 1738
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7869920134544373,
      "learning_rate": 0.000599915667774574,
      "loss": 3.0394,
      "step": 1739
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7369383573532104,
      "learning_rate": 0.000599915570761905,
      "loss": 2.9879,
      "step": 1740
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7020668387413025,
      "learning_rate": 0.000599915473693476,
      "loss": 2.9272,
      "step": 1741
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7987614870071411,
      "learning_rate": 0.000599915376569287,
      "loss": 3.0791,
      "step": 1742
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8604391813278198,
      "learning_rate": 0.000599915279389338,
      "loss": 3.0078,
      "step": 1743
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.943852424621582,
      "learning_rate": 0.0005999151821536292,
      "loss": 2.8676,
      "step": 1744
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7437036633491516,
      "learning_rate": 0.0005999150848621604,
      "loss": 2.8443,
      "step": 1745
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7370374798774719,
      "learning_rate": 0.0005999149875149317,
      "loss": 3.0113,
      "step": 1746
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.96298748254776,
      "learning_rate": 0.0005999148901119432,
      "loss": 3.0242,
      "step": 1747
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7777007818222046,
      "learning_rate": 0.000599914792653195,
      "loss": 3.0745,
      "step": 1748
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7619503736495972,
      "learning_rate": 0.0005999146951386868,
      "loss": 2.779,
      "step": 1749
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7230318188667297,
      "learning_rate": 0.0005999145975684189,
      "loss": 2.9018,
      "step": 1750
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.001947283744812,
      "learning_rate": 0.0005999144999423911,
      "loss": 2.8412,
      "step": 1751
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7177121639251709,
      "learning_rate": 0.0005999144022606036,
      "loss": 2.8108,
      "step": 1752
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8301251530647278,
      "learning_rate": 0.0005999143045230564,
      "loss": 2.8203,
      "step": 1753
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7876956462860107,
      "learning_rate": 0.0005999142067297493,
      "loss": 3.0458,
      "step": 1754
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.9019865989685059,
      "learning_rate": 0.0005999141088806827,
      "loss": 2.8798,
      "step": 1755
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.809235155582428,
      "learning_rate": 0.0005999140109758563,
      "loss": 2.851,
      "step": 1756
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.772678792476654,
      "learning_rate": 0.0005999139130152703,
      "loss": 3.3154,
      "step": 1757
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8533087968826294,
      "learning_rate": 0.0005999138149989246,
      "loss": 2.8609,
      "step": 1758
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8404032588005066,
      "learning_rate": 0.0005999137169268191,
      "loss": 2.8983,
      "step": 1759
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.6885297298431396,
      "learning_rate": 0.0005999136187989542,
      "loss": 2.8652,
      "step": 1760
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7276596426963806,
      "learning_rate": 0.0005999135206153296,
      "loss": 2.8075,
      "step": 1761
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8100686073303223,
      "learning_rate": 0.0005999134223759455,
      "loss": 2.9901,
      "step": 1762
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.0039805173873901,
      "learning_rate": 0.0005999133240808017,
      "loss": 2.9958,
      "step": 1763
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7800149917602539,
      "learning_rate": 0.0005999132257298984,
      "loss": 2.8131,
      "step": 1764
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.9721164703369141,
      "learning_rate": 0.0005999131273232356,
      "loss": 2.9733,
      "step": 1765
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7286327481269836,
      "learning_rate": 0.0005999130288608132,
      "loss": 2.7299,
      "step": 1766
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.9031243920326233,
      "learning_rate": 0.0005999129303426315,
      "loss": 3.022,
      "step": 1767
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7645567059516907,
      "learning_rate": 0.0005999128317686901,
      "loss": 3.0017,
      "step": 1768
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8694065809249878,
      "learning_rate": 0.0005999127331389894,
      "loss": 3.0056,
      "step": 1769
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7477595806121826,
      "learning_rate": 0.0005999126344535292,
      "loss": 3.048,
      "step": 1770
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7896435260772705,
      "learning_rate": 0.0005999125357123097,
      "loss": 2.9127,
      "step": 1771
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7358728647232056,
      "learning_rate": 0.0005999124369153306,
      "loss": 3.0522,
      "step": 1772
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8283519148826599,
      "learning_rate": 0.0005999123380625922,
      "loss": 3.0195,
      "step": 1773
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7299193739891052,
      "learning_rate": 0.0005999122391540944,
      "loss": 3.1365,
      "step": 1774
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7303270697593689,
      "learning_rate": 0.0005999121401898374,
      "loss": 3.0844,
      "step": 1775
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7235165238380432,
      "learning_rate": 0.0005999120411698209,
      "loss": 2.9297,
      "step": 1776
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.6640509366989136,
      "learning_rate": 0.0005999119420940452,
      "loss": 2.8483,
      "step": 1777
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.745580792427063,
      "learning_rate": 0.0005999118429625101,
      "loss": 3.1747,
      "step": 1778
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8574270606040955,
      "learning_rate": 0.0005999117437752159,
      "loss": 3.0675,
      "step": 1779
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7616317272186279,
      "learning_rate": 0.0005999116445321623,
      "loss": 2.8443,
      "step": 1780
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.807663083076477,
      "learning_rate": 0.0005999115452333494,
      "loss": 2.7451,
      "step": 1781
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8919357061386108,
      "learning_rate": 0.0005999114458787774,
      "loss": 3.0027,
      "step": 1782
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7591288685798645,
      "learning_rate": 0.0005999113464684462,
      "loss": 2.7143,
      "step": 1783
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7417558431625366,
      "learning_rate": 0.0005999112470023558,
      "loss": 2.7744,
      "step": 1784
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7980374097824097,
      "learning_rate": 0.0005999111474805063,
      "loss": 3.0007,
      "step": 1785
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7966530919075012,
      "learning_rate": 0.0005999110479028975,
      "loss": 2.9028,
      "step": 1786
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.6967571973800659,
      "learning_rate": 0.0005999109482695297,
      "loss": 3.0739,
      "step": 1787
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.873458743095398,
      "learning_rate": 0.0005999108485804028,
      "loss": 3.0062,
      "step": 1788
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7327402234077454,
      "learning_rate": 0.0005999107488355169,
      "loss": 2.8768,
      "step": 1789
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7926539182662964,
      "learning_rate": 0.0005999106490348717,
      "loss": 3.1069,
      "step": 1790
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8274723887443542,
      "learning_rate": 0.0005999105491784675,
      "loss": 3.0179,
      "step": 1791
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7608490586280823,
      "learning_rate": 0.0005999104492663043,
      "loss": 2.8761,
      "step": 1792
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.930854082107544,
      "learning_rate": 0.0005999103492983823,
      "loss": 2.7704,
      "step": 1793
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7855798006057739,
      "learning_rate": 0.0005999102492747011,
      "loss": 2.7942,
      "step": 1794
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7921724915504456,
      "learning_rate": 0.0005999101491952608,
      "loss": 2.7808,
      "step": 1795
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7498272657394409,
      "learning_rate": 0.0005999100490600617,
      "loss": 2.9498,
      "step": 1796
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.9572819471359253,
      "learning_rate": 0.0005999099488691037,
      "loss": 3.0483,
      "step": 1797
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.721319317817688,
      "learning_rate": 0.0005999098486223867,
      "loss": 2.8787,
      "step": 1798
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8388048410415649,
      "learning_rate": 0.0005999097483199108,
      "loss": 3.2155,
      "step": 1799
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7842227816581726,
      "learning_rate": 0.0005999096479616761,
      "loss": 2.8503,
      "step": 1800
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7854987978935242,
      "learning_rate": 0.0005999095475476825,
      "loss": 3.1941,
      "step": 1801
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.0180909633636475,
      "learning_rate": 0.00059990944707793,
      "loss": 2.9511,
      "step": 1802
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7663751840591431,
      "learning_rate": 0.0005999093465524186,
      "loss": 2.9716,
      "step": 1803
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8791009187698364,
      "learning_rate": 0.0005999092459711485,
      "loss": 2.8729,
      "step": 1804
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.833001434803009,
      "learning_rate": 0.0005999091453341197,
      "loss": 2.8633,
      "step": 1805
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8455013632774353,
      "learning_rate": 0.000599909044641332,
      "loss": 2.8569,
      "step": 1806
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7594231963157654,
      "learning_rate": 0.0005999089438927856,
      "loss": 2.9371,
      "step": 1807
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8564037680625916,
      "learning_rate": 0.0005999088430884803,
      "loss": 3.3489,
      "step": 1808
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8644270896911621,
      "learning_rate": 0.0005999087422284165,
      "loss": 2.8805,
      "step": 1809
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7589526772499084,
      "learning_rate": 0.000599908641312594,
      "loss": 3.0549,
      "step": 1810
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.800071120262146,
      "learning_rate": 0.0005999085403410127,
      "loss": 2.8435,
      "step": 1811
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8930865526199341,
      "learning_rate": 0.0005999084393136728,
      "loss": 2.94,
      "step": 1812
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8513879179954529,
      "learning_rate": 0.0005999083382305743,
      "loss": 3.1897,
      "step": 1813
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7969205379486084,
      "learning_rate": 0.000599908237091717,
      "loss": 3.0041,
      "step": 1814
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7984876036643982,
      "learning_rate": 0.0005999081358971013,
      "loss": 2.9513,
      "step": 1815
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.313926100730896,
      "learning_rate": 0.000599908034646727,
      "loss": 2.9265,
      "step": 1816
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.770166277885437,
      "learning_rate": 0.000599907933340594,
      "loss": 3.0639,
      "step": 1817
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8442248106002808,
      "learning_rate": 0.0005999078319787026,
      "loss": 3.0593,
      "step": 1818
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7570866346359253,
      "learning_rate": 0.0005999077305610526,
      "loss": 2.9431,
      "step": 1819
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7632806301116943,
      "learning_rate": 0.0005999076290876441,
      "loss": 2.9815,
      "step": 1820
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.0097205638885498,
      "learning_rate": 0.0005999075275584772,
      "loss": 2.7518,
      "step": 1821
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.6850093603134155,
      "learning_rate": 0.0005999074259735517,
      "loss": 2.8814,
      "step": 1822
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8382006287574768,
      "learning_rate": 0.0005999073243328677,
      "loss": 2.7558,
      "step": 1823
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.6378855109214783,
      "learning_rate": 0.0005999072226364255,
      "loss": 2.7373,
      "step": 1824
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.9341282844543457,
      "learning_rate": 0.0005999071208842246,
      "loss": 2.854,
      "step": 1825
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.6848804354667664,
      "learning_rate": 0.0005999070190762655,
      "loss": 3.1642,
      "step": 1826
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8443750739097595,
      "learning_rate": 0.000599906917212548,
      "loss": 3.0587,
      "step": 1827
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8220899105072021,
      "learning_rate": 0.0005999068152930721,
      "loss": 2.7639,
      "step": 1828
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8639240264892578,
      "learning_rate": 0.0005999067133178378,
      "loss": 3.1011,
      "step": 1829
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8375234007835388,
      "learning_rate": 0.0005999066112868453,
      "loss": 2.9878,
      "step": 1830
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.1461044549942017,
      "learning_rate": 0.0005999065092000945,
      "loss": 3.1816,
      "step": 1831
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8253985643386841,
      "learning_rate": 0.0005999064070575853,
      "loss": 2.9577,
      "step": 1832
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8659101724624634,
      "learning_rate": 0.000599906304859318,
      "loss": 3.1152,
      "step": 1833
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.0396479368209839,
      "learning_rate": 0.0005999062026052923,
      "loss": 2.9746,
      "step": 1834
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7840397953987122,
      "learning_rate": 0.0005999061002955085,
      "loss": 2.8846,
      "step": 1835
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.016115665435791,
      "learning_rate": 0.0005999059979299664,
      "loss": 2.7366,
      "step": 1836
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7540351152420044,
      "learning_rate": 0.0005999058955086661,
      "loss": 3.1187,
      "step": 1837
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8543146252632141,
      "learning_rate": 0.0005999057930316078,
      "loss": 3.0246,
      "step": 1838
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.1693131923675537,
      "learning_rate": 0.0005999056904987913,
      "loss": 2.9957,
      "step": 1839
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7264663577079773,
      "learning_rate": 0.0005999055879102165,
      "loss": 3.0752,
      "step": 1840
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.070868968963623,
      "learning_rate": 0.0005999054852658838,
      "loss": 2.9602,
      "step": 1841
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.9509969353675842,
      "learning_rate": 0.0005999053825657928,
      "loss": 3.1767,
      "step": 1842
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7996395826339722,
      "learning_rate": 0.0005999052798099438,
      "loss": 3.1575,
      "step": 1843
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.9735598564147949,
      "learning_rate": 0.0005999051769983369,
      "loss": 2.8278,
      "step": 1844
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.800588071346283,
      "learning_rate": 0.0005999050741309718,
      "loss": 2.7285,
      "step": 1845
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7643049359321594,
      "learning_rate": 0.0005999049712078487,
      "loss": 2.9624,
      "step": 1846
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.9432118535041809,
      "learning_rate": 0.0005999048682289676,
      "loss": 3.2503,
      "step": 1847
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8805199861526489,
      "learning_rate": 0.0005999047651943286,
      "loss": 2.8122,
      "step": 1848
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7539815902709961,
      "learning_rate": 0.0005999046621039316,
      "loss": 3.2653,
      "step": 1849
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.1435625553131104,
      "learning_rate": 0.0005999045589577767,
      "loss": 2.9672,
      "step": 1850
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.9933125376701355,
      "learning_rate": 0.0005999044557558639,
      "loss": 3.018,
      "step": 1851
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.939182698726654,
      "learning_rate": 0.0005999043524981931,
      "loss": 2.8839,
      "step": 1852
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.9891266226768494,
      "learning_rate": 0.0005999042491847646,
      "loss": 2.864,
      "step": 1853
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.007082462310791,
      "learning_rate": 0.000599904145815578,
      "loss": 2.8575,
      "step": 1854
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8855592608451843,
      "learning_rate": 0.0005999040423906337,
      "loss": 2.8753,
      "step": 1855
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.9971553683280945,
      "learning_rate": 0.0005999039389099315,
      "loss": 2.7895,
      "step": 1856
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.9781344532966614,
      "learning_rate": 0.0005999038353734716,
      "loss": 3.0577,
      "step": 1857
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7649887800216675,
      "learning_rate": 0.0005999037317812538,
      "loss": 2.8882,
      "step": 1858
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.0683656930923462,
      "learning_rate": 0.0005999036281332783,
      "loss": 2.8278,
      "step": 1859
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.9053491950035095,
      "learning_rate": 0.0005999035244295451,
      "loss": 2.8975,
      "step": 1860
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7000234127044678,
      "learning_rate": 0.0005999034206700541,
      "loss": 2.7196,
      "step": 1861
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8345761895179749,
      "learning_rate": 0.0005999033168548055,
      "loss": 3.1178,
      "step": 1862
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.0018519163131714,
      "learning_rate": 0.000599903212983799,
      "loss": 2.8921,
      "step": 1863
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7812990546226501,
      "learning_rate": 0.0005999031090570349,
      "loss": 3.1137,
      "step": 1864
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.9691644310951233,
      "learning_rate": 0.0005999030050745133,
      "loss": 3.0708,
      "step": 1865
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7524101734161377,
      "learning_rate": 0.000599902901036234,
      "loss": 2.8131,
      "step": 1866
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7541730403900146,
      "learning_rate": 0.000599902796942197,
      "loss": 2.8055,
      "step": 1867
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8485041856765747,
      "learning_rate": 0.0005999026927924026,
      "loss": 2.8168,
      "step": 1868
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7813301682472229,
      "learning_rate": 0.0005999025885868504,
      "loss": 2.9089,
      "step": 1869
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7110317945480347,
      "learning_rate": 0.0005999024843255409,
      "loss": 2.8044,
      "step": 1870
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8489604592323303,
      "learning_rate": 0.0005999023800084736,
      "loss": 2.9252,
      "step": 1871
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8339295387268066,
      "learning_rate": 0.000599902275635649,
      "loss": 2.9469,
      "step": 1872
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7870714068412781,
      "learning_rate": 0.0005999021712070668,
      "loss": 3.0782,
      "step": 1873
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7890341877937317,
      "learning_rate": 0.0005999020667227271,
      "loss": 3.0944,
      "step": 1874
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8278881311416626,
      "learning_rate": 0.0005999019621826299,
      "loss": 2.8875,
      "step": 1875
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7570728063583374,
      "learning_rate": 0.0005999018575867754,
      "loss": 2.8503,
      "step": 1876
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7297350764274597,
      "learning_rate": 0.0005999017529351636,
      "loss": 3.0332,
      "step": 1877
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8787208199501038,
      "learning_rate": 0.000599901648227794,
      "loss": 2.8161,
      "step": 1878
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7952726483345032,
      "learning_rate": 0.0005999015434646674,
      "loss": 3.2897,
      "step": 1879
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7836579084396362,
      "learning_rate": 0.0005999014386457832,
      "loss": 3.0254,
      "step": 1880
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8619025945663452,
      "learning_rate": 0.0005999013337711418,
      "loss": 2.9141,
      "step": 1881
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7626659870147705,
      "learning_rate": 0.000599901228840743,
      "loss": 2.8432,
      "step": 1882
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8495050668716431,
      "learning_rate": 0.000599901123854587,
      "loss": 2.859,
      "step": 1883
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.091750144958496,
      "learning_rate": 0.0005999010188126737,
      "loss": 3.0301,
      "step": 1884
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7022618651390076,
      "learning_rate": 0.0005999009137150031,
      "loss": 2.9437,
      "step": 1885
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.9475826025009155,
      "learning_rate": 0.0005999008085615752,
      "loss": 2.9222,
      "step": 1886
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.0313130617141724,
      "learning_rate": 0.0005999007033523901,
      "loss": 2.7512,
      "step": 1887
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8928608894348145,
      "learning_rate": 0.000599900598087448,
      "loss": 2.7101,
      "step": 1888
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.2155355215072632,
      "learning_rate": 0.0005999004927667484,
      "loss": 3.0997,
      "step": 1889
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8632717728614807,
      "learning_rate": 0.0005999003873902918,
      "loss": 2.7947,
      "step": 1890
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.830479621887207,
      "learning_rate": 0.0005999002819580781,
      "loss": 2.9985,
      "step": 1891
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.2278459072113037,
      "learning_rate": 0.0005999001764701073,
      "loss": 2.7505,
      "step": 1892
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.0330722332000732,
      "learning_rate": 0.0005999000709263792,
      "loss": 2.8239,
      "step": 1893
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7471146583557129,
      "learning_rate": 0.0005998999653268942,
      "loss": 2.9614,
      "step": 1894
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.9129323363304138,
      "learning_rate": 0.0005998998596716521,
      "loss": 3.2508,
      "step": 1895
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.1669656038284302,
      "learning_rate": 0.0005998997539606529,
      "loss": 3.0251,
      "step": 1896
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.9262046217918396,
      "learning_rate": 0.0005998996481938968,
      "loss": 2.7065,
      "step": 1897
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.896630585193634,
      "learning_rate": 0.0005998995423713835,
      "loss": 3.0367,
      "step": 1898
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.9399252533912659,
      "learning_rate": 0.0005998994364931133,
      "loss": 2.91,
      "step": 1899
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.874342679977417,
      "learning_rate": 0.0005998993305590862,
      "loss": 3.0099,
      "step": 1900
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.72823166847229,
      "learning_rate": 0.000599899224569302,
      "loss": 2.9009,
      "step": 1901
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7529273629188538,
      "learning_rate": 0.000599899118523761,
      "loss": 2.9796,
      "step": 1902
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8697402477264404,
      "learning_rate": 0.0005998990124224631,
      "loss": 2.934,
      "step": 1903
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8677735328674316,
      "learning_rate": 0.0005998989062654083,
      "loss": 2.8106,
      "step": 1904
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7359393239021301,
      "learning_rate": 0.0005998988000525966,
      "loss": 2.9999,
      "step": 1905
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.9966592788696289,
      "learning_rate": 0.000599898693784028,
      "loss": 2.951,
      "step": 1906
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.2058982849121094,
      "learning_rate": 0.0005998985874597027,
      "loss": 2.954,
      "step": 1907
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.6935309171676636,
      "learning_rate": 0.0005998984810796204,
      "loss": 2.9869,
      "step": 1908
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.9642782807350159,
      "learning_rate": 0.0005998983746437816,
      "loss": 3.0147,
      "step": 1909
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.4748727083206177,
      "learning_rate": 0.0005998982681521858,
      "loss": 2.881,
      "step": 1910
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7504557371139526,
      "learning_rate": 0.0005998981616048333,
      "loss": 2.9092,
      "step": 1911
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.0148998498916626,
      "learning_rate": 0.0005998980550017242,
      "loss": 2.9359,
      "step": 1912
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7885136604309082,
      "learning_rate": 0.0005998979483428582,
      "loss": 2.7285,
      "step": 1913
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8948175311088562,
      "learning_rate": 0.0005998978416282356,
      "loss": 3.1501,
      "step": 1914
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.73993980884552,
      "learning_rate": 0.0005998977348578563,
      "loss": 2.8838,
      "step": 1915
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8431190252304077,
      "learning_rate": 0.0005998976280317204,
      "loss": 2.8838,
      "step": 1916
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8802720308303833,
      "learning_rate": 0.0005998975211498278,
      "loss": 2.831,
      "step": 1917
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.6952444911003113,
      "learning_rate": 0.0005998974142121786,
      "loss": 2.9132,
      "step": 1918
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.6826069951057434,
      "learning_rate": 0.0005998973072187729,
      "loss": 2.9644,
      "step": 1919
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8546233177185059,
      "learning_rate": 0.0005998972001696105,
      "loss": 3.1802,
      "step": 1920
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7723889946937561,
      "learning_rate": 0.0005998970930646916,
      "loss": 2.9192,
      "step": 1921
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7371392846107483,
      "learning_rate": 0.0005998969859040162,
      "loss": 2.9789,
      "step": 1922
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.8080343008041382,
      "learning_rate": 0.0005998968786875843,
      "loss": 2.9317,
      "step": 1923
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7681041359901428,
      "learning_rate": 0.0005998967714153959,
      "loss": 2.8365,
      "step": 1924
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.820220410823822,
      "learning_rate": 0.0005998966640874511,
      "loss": 2.9869,
      "step": 1925
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.8177988529205322,
      "learning_rate": 0.0005998965567037497,
      "loss": 3.1602,
      "step": 1926
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.9406993389129639,
      "learning_rate": 0.0005998964492642919,
      "loss": 2.7217,
      "step": 1927
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7390802502632141,
      "learning_rate": 0.0005998963417690778,
      "loss": 3.0459,
      "step": 1928
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7940952777862549,
      "learning_rate": 0.0005998962342181072,
      "loss": 3.0858,
      "step": 1929
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7677420377731323,
      "learning_rate": 0.0005998961266113803,
      "loss": 2.8643,
      "step": 1930
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7686255574226379,
      "learning_rate": 0.000599896018948897,
      "loss": 2.8465,
      "step": 1931
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7704368233680725,
      "learning_rate": 0.0005998959112306574,
      "loss": 2.8586,
      "step": 1932
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.8650120496749878,
      "learning_rate": 0.0005998958034566614,
      "loss": 2.654,
      "step": 1933
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.8566858172416687,
      "learning_rate": 0.0005998956956269092,
      "loss": 3.0568,
      "step": 1934
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7659881711006165,
      "learning_rate": 0.0005998955877414007,
      "loss": 2.9518,
      "step": 1935
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.087336540222168,
      "learning_rate": 0.000599895479800136,
      "loss": 3.0693,
      "step": 1936
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.8040370345115662,
      "learning_rate": 0.000599895371803115,
      "loss": 2.9559,
      "step": 1937
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7845446467399597,
      "learning_rate": 0.0005998952637503378,
      "loss": 3.227,
      "step": 1938
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.8548194766044617,
      "learning_rate": 0.0005998951556418045,
      "loss": 3.0882,
      "step": 1939
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.799647331237793,
      "learning_rate": 0.0005998950474775149,
      "loss": 2.9507,
      "step": 1940
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7937114238739014,
      "learning_rate": 0.0005998949392574692,
      "loss": 2.8474,
      "step": 1941
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7320470213890076,
      "learning_rate": 0.0005998948309816674,
      "loss": 3.0889,
      "step": 1942
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.0003215074539185,
      "learning_rate": 0.0005998947226501095,
      "loss": 3.1638,
      "step": 1943
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7624526023864746,
      "learning_rate": 0.0005998946142627954,
      "loss": 3.0387,
      "step": 1944
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7480859756469727,
      "learning_rate": 0.0005998945058197253,
      "loss": 3.0042,
      "step": 1945
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.8570588231086731,
      "learning_rate": 0.0005998943973208992,
      "loss": 2.8346,
      "step": 1946
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.8000507354736328,
      "learning_rate": 0.0005998942887663171,
      "loss": 2.6358,
      "step": 1947
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.8094562888145447,
      "learning_rate": 0.0005998941801559789,
      "loss": 2.7609,
      "step": 1948
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.9301173686981201,
      "learning_rate": 0.0005998940714898847,
      "loss": 3.1906,
      "step": 1949
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.755216658115387,
      "learning_rate": 0.0005998939627680346,
      "loss": 2.7069,
      "step": 1950
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7627948522567749,
      "learning_rate": 0.0005998938539904284,
      "loss": 2.5931,
      "step": 1951
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.8082130551338196,
      "learning_rate": 0.0005998937451570665,
      "loss": 2.9462,
      "step": 1952
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.991457462310791,
      "learning_rate": 0.0005998936362679485,
      "loss": 2.9469,
      "step": 1953
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7976449131965637,
      "learning_rate": 0.0005998935273230747,
      "loss": 3.1703,
      "step": 1954
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.2869062423706055,
      "learning_rate": 0.000599893418322445,
      "loss": 2.7702,
      "step": 1955
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.8073232769966125,
      "learning_rate": 0.0005998933092660595,
      "loss": 2.8816,
      "step": 1956
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.8691274523735046,
      "learning_rate": 0.0005998932001539182,
      "loss": 2.8454,
      "step": 1957
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.9411946535110474,
      "learning_rate": 0.0005998930909860209,
      "loss": 2.9899,
      "step": 1958
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.6677420139312744,
      "learning_rate": 0.000599892981762368,
      "loss": 2.8002,
      "step": 1959
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7148891091346741,
      "learning_rate": 0.0005998928724829592,
      "loss": 2.6987,
      "step": 1960
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.990506112575531,
      "learning_rate": 0.0005998927631477948,
      "loss": 2.8812,
      "step": 1961
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.6488837003707886,
      "learning_rate": 0.0005998926537568747,
      "loss": 3.0648,
      "step": 1962
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.823282778263092,
      "learning_rate": 0.0005998925443101988,
      "loss": 2.9658,
      "step": 1963
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7116197943687439,
      "learning_rate": 0.0005998924348077671,
      "loss": 2.8562,
      "step": 1964
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7354429364204407,
      "learning_rate": 0.0005998923252495799,
      "loss": 2.8586,
      "step": 1965
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.724783182144165,
      "learning_rate": 0.0005998922156356369,
      "loss": 2.9869,
      "step": 1966
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.9144281148910522,
      "learning_rate": 0.0005998921059659385,
      "loss": 3.0683,
      "step": 1967
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7296857237815857,
      "learning_rate": 0.0005998919962404844,
      "loss": 3.0603,
      "step": 1968
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.8071166276931763,
      "learning_rate": 0.0005998918864592747,
      "loss": 2.7643,
      "step": 1969
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.8452547788619995,
      "learning_rate": 0.0005998917766223094,
      "loss": 2.8011,
      "step": 1970
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7604900002479553,
      "learning_rate": 0.0005998916667295886,
      "loss": 2.7189,
      "step": 1971
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7500594854354858,
      "learning_rate": 0.0005998915567811123,
      "loss": 2.8552,
      "step": 1972
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.9254981875419617,
      "learning_rate": 0.0005998914467768804,
      "loss": 3.0295,
      "step": 1973
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7688036561012268,
      "learning_rate": 0.0005998913367168932,
      "loss": 2.661,
      "step": 1974
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.8424748778343201,
      "learning_rate": 0.0005998912266011505,
      "loss": 2.9679,
      "step": 1975
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.0351704359054565,
      "learning_rate": 0.0005998911164296522,
      "loss": 2.8783,
      "step": 1976
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7552047967910767,
      "learning_rate": 0.0005998910062023986,
      "loss": 2.7613,
      "step": 1977
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7888008952140808,
      "learning_rate": 0.0005998908959193895,
      "loss": 2.9246,
      "step": 1978
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.8663063645362854,
      "learning_rate": 0.0005998907855806251,
      "loss": 2.8858,
      "step": 1979
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.8002691864967346,
      "learning_rate": 0.0005998906751861054,
      "loss": 2.7644,
      "step": 1980
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.859498918056488,
      "learning_rate": 0.0005998905647358303,
      "loss": 2.7453,
      "step": 1981
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7233223915100098,
      "learning_rate": 0.0005998904542297999,
      "loss": 2.9065,
      "step": 1982
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7907515168190002,
      "learning_rate": 0.0005998903436680142,
      "loss": 3.0046,
      "step": 1983
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.8468871116638184,
      "learning_rate": 0.0005998902330504732,
      "loss": 3.0182,
      "step": 1984
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.8048920035362244,
      "learning_rate": 0.0005998901223771769,
      "loss": 2.8327,
      "step": 1985
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7062415480613708,
      "learning_rate": 0.0005998900116481254,
      "loss": 2.9308,
      "step": 1986
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.8679699301719666,
      "learning_rate": 0.0005998899008633188,
      "loss": 2.9712,
      "step": 1987
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.6499254107475281,
      "learning_rate": 0.000599889790022757,
      "loss": 2.9521,
      "step": 1988
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7514140605926514,
      "learning_rate": 0.0005998896791264399,
      "loss": 3.3396,
      "step": 1989
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7303810119628906,
      "learning_rate": 0.0005998895681743676,
      "loss": 3.1252,
      "step": 1990
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.923772931098938,
      "learning_rate": 0.0005998894571665403,
      "loss": 2.9114,
      "step": 1991
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7018480896949768,
      "learning_rate": 0.0005998893461029579,
      "loss": 3.0891,
      "step": 1992
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.903429388999939,
      "learning_rate": 0.0005998892349836203,
      "loss": 3.2499,
      "step": 1993
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.9229446053504944,
      "learning_rate": 0.0005998891238085277,
      "loss": 2.9594,
      "step": 1994
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.8496990203857422,
      "learning_rate": 0.00059988901257768,
      "loss": 2.6904,
      "step": 1995
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.037636399269104,
      "learning_rate": 0.0005998889012910774,
      "loss": 2.8786,
      "step": 1996
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7493432760238647,
      "learning_rate": 0.0005998887899487196,
      "loss": 2.8052,
      "step": 1997
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.990203857421875,
      "learning_rate": 0.000599888678550607,
      "loss": 2.8591,
      "step": 1998
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7958164811134338,
      "learning_rate": 0.0005998885670967393,
      "loss": 2.9348,
      "step": 1999
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7715801000595093,
      "learning_rate": 0.0005998884555871167,
      "loss": 3.0144,
      "step": 2000
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.8215363621711731,
      "learning_rate": 0.0005998883440217393,
      "loss": 3.0923,
      "step": 2001
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.1568928956985474,
      "learning_rate": 0.0005998882324006068,
      "loss": 2.9871,
      "step": 2002
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.958536684513092,
      "learning_rate": 0.0005998881207237195,
      "loss": 2.9988,
      "step": 2003
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.3212751150131226,
      "learning_rate": 0.0005998880089910773,
      "loss": 2.965,
      "step": 2004
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7565361857414246,
      "learning_rate": 0.0005998878972026803,
      "loss": 2.7977,
      "step": 2005
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.9343516826629639,
      "learning_rate": 0.0005998877853585284,
      "loss": 2.7395,
      "step": 2006
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.8902658224105835,
      "learning_rate": 0.0005998876734586218,
      "loss": 2.9076,
      "step": 2007
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7091558575630188,
      "learning_rate": 0.0005998875615029604,
      "loss": 2.9518,
      "step": 2008
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.8543210625648499,
      "learning_rate": 0.0005998874494915443,
      "loss": 3.14,
      "step": 2009
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.9303277134895325,
      "learning_rate": 0.0005998873374243734,
      "loss": 3.1592,
      "step": 2010
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7173624038696289,
      "learning_rate": 0.0005998872253014478,
      "loss": 3.0651,
      "step": 2011
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.9587964415550232,
      "learning_rate": 0.0005998871131227675,
      "loss": 2.893,
      "step": 2012
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7058107852935791,
      "learning_rate": 0.0005998870008883324,
      "loss": 2.7873,
      "step": 2013
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7082306742668152,
      "learning_rate": 0.0005998868885981428,
      "loss": 2.7665,
      "step": 2014
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.8080980777740479,
      "learning_rate": 0.0005998867762521985,
      "loss": 2.7509,
      "step": 2015
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.6729606986045837,
      "learning_rate": 0.0005998866638504996,
      "loss": 3.0385,
      "step": 2016
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.6905429363250732,
      "learning_rate": 0.0005998865513930461,
      "loss": 2.8712,
      "step": 2017
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7294578552246094,
      "learning_rate": 0.0005998864388798381,
      "loss": 2.8957,
      "step": 2018
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.816547155380249,
      "learning_rate": 0.0005998863263108755,
      "loss": 2.8413,
      "step": 2019
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7767800092697144,
      "learning_rate": 0.0005998862136861583,
      "loss": 2.704,
      "step": 2020
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7146404981613159,
      "learning_rate": 0.0005998861010056867,
      "loss": 2.8144,
      "step": 2021
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7913563847541809,
      "learning_rate": 0.0005998859882694605,
      "loss": 2.8464,
      "step": 2022
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.9452394247055054,
      "learning_rate": 0.00059988587547748,
      "loss": 3.1187,
      "step": 2023
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.8372299075126648,
      "learning_rate": 0.0005998857626297449,
      "loss": 2.7329,
      "step": 2024
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.9365825057029724,
      "learning_rate": 0.0005998856497262554,
      "loss": 2.8323,
      "step": 2025
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7830345034599304,
      "learning_rate": 0.0005998855367670116,
      "loss": 2.8733,
      "step": 2026
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7951073050498962,
      "learning_rate": 0.0005998854237520134,
      "loss": 2.986,
      "step": 2027
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.9805779457092285,
      "learning_rate": 0.0005998853106812608,
      "loss": 2.6596,
      "step": 2028
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7413759827613831,
      "learning_rate": 0.0005998851975547538,
      "loss": 2.8771,
      "step": 2029
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7622909545898438,
      "learning_rate": 0.0005998850843724925,
      "loss": 2.683,
      "step": 2030
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.936707615852356,
      "learning_rate": 0.0005998849711344768,
      "loss": 2.8028,
      "step": 2031
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7455191016197205,
      "learning_rate": 0.0005998848578407071,
      "loss": 2.9237,
      "step": 2032
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7537509799003601,
      "learning_rate": 0.000599884744491183,
      "loss": 3.0733,
      "step": 2033
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.8010852932929993,
      "learning_rate": 0.0005998846310859046,
      "loss": 2.7845,
      "step": 2034
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7086145281791687,
      "learning_rate": 0.000599884517624872,
      "loss": 2.658,
      "step": 2035
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7833667397499084,
      "learning_rate": 0.0005998844041080853,
      "loss": 2.7618,
      "step": 2036
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.8226791620254517,
      "learning_rate": 0.0005998842905355444,
      "loss": 3.1058,
      "step": 2037
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.8426749110221863,
      "learning_rate": 0.0005998841769072493,
      "loss": 2.9681,
      "step": 2038
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.822858989238739,
      "learning_rate": 0.0005998840632232001,
      "loss": 2.7962,
      "step": 2039
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.8265706896781921,
      "learning_rate": 0.0005998839494833968,
      "loss": 2.8839,
      "step": 2040
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7738246917724609,
      "learning_rate": 0.0005998838356878394,
      "loss": 2.8059,
      "step": 2041
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.839146614074707,
      "learning_rate": 0.000599883721836528,
      "loss": 2.8994,
      "step": 2042
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.9677287340164185,
      "learning_rate": 0.0005998836079294625,
      "loss": 3.0636,
      "step": 2043
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7885153889656067,
      "learning_rate": 0.0005998834939666429,
      "loss": 3.0095,
      "step": 2044
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7993105053901672,
      "learning_rate": 0.0005998833799480695,
      "loss": 2.7839,
      "step": 2045
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7245429158210754,
      "learning_rate": 0.0005998832658737419,
      "loss": 2.8851,
      "step": 2046
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7224494218826294,
      "learning_rate": 0.0005998831517436604,
      "loss": 2.7654,
      "step": 2047
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7369645833969116,
      "learning_rate": 0.0005998830375578251,
      "loss": 3.2164,
      "step": 2048
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7819618582725525,
      "learning_rate": 0.0005998829233162357,
      "loss": 3.0279,
      "step": 2049
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7937363982200623,
      "learning_rate": 0.0005998828090188925,
      "loss": 2.9249,
      "step": 2050
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.8195468783378601,
      "learning_rate": 0.0005998826946657954,
      "loss": 2.8765,
      "step": 2051
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7043704390525818,
      "learning_rate": 0.0005998825802569444,
      "loss": 2.8114,
      "step": 2052
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.6886816620826721,
      "learning_rate": 0.0005998824657923397,
      "loss": 2.6616,
      "step": 2053
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7168062925338745,
      "learning_rate": 0.0005998823512719812,
      "loss": 3.1686,
      "step": 2054
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7629327774047852,
      "learning_rate": 0.0005998822366958688,
      "loss": 2.9651,
      "step": 2055
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7590177655220032,
      "learning_rate": 0.0005998821220640026,
      "loss": 3.0861,
      "step": 2056
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7464116811752319,
      "learning_rate": 0.0005998820073763826,
      "loss": 3.0933,
      "step": 2057
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.9484373927116394,
      "learning_rate": 0.0005998818926330091,
      "loss": 2.9851,
      "step": 2058
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7656694650650024,
      "learning_rate": 0.0005998817778338817,
      "loss": 2.8299,
      "step": 2059
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.9145796895027161,
      "learning_rate": 0.0005998816629790007,
      "loss": 3.1346,
      "step": 2060
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.8254315257072449,
      "learning_rate": 0.0005998815480683661,
      "loss": 3.2502,
      "step": 2061
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7743854522705078,
      "learning_rate": 0.0005998814331019777,
      "loss": 2.9231,
      "step": 2062
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.32979154586792,
      "learning_rate": 0.0005998813180798358,
      "loss": 2.8233,
      "step": 2063
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.062041163444519,
      "learning_rate": 0.0005998812030019403,
      "loss": 2.8849,
      "step": 2064
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7918543219566345,
      "learning_rate": 0.0005998810878682912,
      "loss": 2.9561,
      "step": 2065
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.103550910949707,
      "learning_rate": 0.0005998809726788885,
      "loss": 2.8552,
      "step": 2066
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.9391119480133057,
      "learning_rate": 0.0005998808574337324,
      "loss": 2.8558,
      "step": 2067
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.8348532915115356,
      "learning_rate": 0.0005998807421328227,
      "loss": 2.8529,
      "step": 2068
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.9928792119026184,
      "learning_rate": 0.0005998806267761595,
      "loss": 3.1399,
      "step": 2069
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7946481704711914,
      "learning_rate": 0.0005998805113637428,
      "loss": 3.098,
      "step": 2070
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7734258770942688,
      "learning_rate": 0.0005998803958955727,
      "loss": 2.9518,
      "step": 2071
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.9742555022239685,
      "learning_rate": 0.0005998802803716491,
      "loss": 2.9226,
      "step": 2072
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7823353409767151,
      "learning_rate": 0.0005998801647919722,
      "loss": 2.8517,
      "step": 2073
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7856823801994324,
      "learning_rate": 0.0005998800491565419,
      "loss": 2.9538,
      "step": 2074
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7710857391357422,
      "learning_rate": 0.0005998799334653582,
      "loss": 2.9559,
      "step": 2075
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7239710688591003,
      "learning_rate": 0.0005998798177184212,
      "loss": 2.9972,
      "step": 2076
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7751869559288025,
      "learning_rate": 0.0005998797019157308,
      "loss": 3.1677,
      "step": 2077
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.753287672996521,
      "learning_rate": 0.0005998795860572871,
      "loss": 2.8253,
      "step": 2078
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.9243274331092834,
      "learning_rate": 0.0005998794701430903,
      "loss": 2.7389,
      "step": 2079
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.8762354850769043,
      "learning_rate": 0.00059987935417314,
      "loss": 2.9523,
      "step": 2080
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7186788320541382,
      "learning_rate": 0.0005998792381474367,
      "loss": 3.1361,
      "step": 2081
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.780532956123352,
      "learning_rate": 0.00059987912206598,
      "loss": 2.8916,
      "step": 2082
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.810183584690094,
      "learning_rate": 0.0005998790059287703,
      "loss": 2.7676,
      "step": 2083
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.712039589881897,
      "learning_rate": 0.0005998788897358073,
      "loss": 2.6575,
      "step": 2084
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.8589749336242676,
      "learning_rate": 0.0005998787734870911,
      "loss": 3.1872,
      "step": 2085
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7787895798683167,
      "learning_rate": 0.0005998786571826219,
      "loss": 3.068,
      "step": 2086
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7395980954170227,
      "learning_rate": 0.0005998785408223994,
      "loss": 3.0264,
      "step": 2087
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.9524626135826111,
      "learning_rate": 0.000599878424406424,
      "loss": 2.9138,
      "step": 2088
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.852493405342102,
      "learning_rate": 0.0005998783079346955,
      "loss": 3.0041,
      "step": 2089
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.813162624835968,
      "learning_rate": 0.0005998781914072141,
      "loss": 2.7617,
      "step": 2090
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7630054950714111,
      "learning_rate": 0.0005998780748239795,
      "loss": 2.7786,
      "step": 2091
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.0955212116241455,
      "learning_rate": 0.0005998779581849919,
      "loss": 2.8187,
      "step": 2092
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7470128536224365,
      "learning_rate": 0.0005998778414902515,
      "loss": 2.98,
      "step": 2093
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7398212552070618,
      "learning_rate": 0.000599877724739758,
      "loss": 2.9273,
      "step": 2094
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7614160776138306,
      "learning_rate": 0.0005998776079335117,
      "loss": 3.0136,
      "step": 2095
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7955130934715271,
      "learning_rate": 0.0005998774910715123,
      "loss": 2.759,
      "step": 2096
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.793015718460083,
      "learning_rate": 0.00059987737415376,
      "loss": 2.7216,
      "step": 2097
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7168318629264832,
      "learning_rate": 0.000599877257180255,
      "loss": 2.7397,
      "step": 2098
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.729984700679779,
      "learning_rate": 0.0005998771401509971,
      "loss": 2.7041,
      "step": 2099
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.9768990278244019,
      "learning_rate": 0.0005998770230659864,
      "loss": 2.859,
      "step": 2100
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7718768119812012,
      "learning_rate": 0.0005998769059252229,
      "loss": 2.9861,
      "step": 2101
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7868697643280029,
      "learning_rate": 0.0005998767887287066,
      "loss": 2.9075,
      "step": 2102
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.760439395904541,
      "learning_rate": 0.0005998766714764376,
      "loss": 3.0867,
      "step": 2103
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7116325497627258,
      "learning_rate": 0.0005998765541684157,
      "loss": 2.9244,
      "step": 2104
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7211179137229919,
      "learning_rate": 0.0005998764368046413,
      "loss": 2.8337,
      "step": 2105
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7061159014701843,
      "learning_rate": 0.0005998763193851141,
      "loss": 2.5656,
      "step": 2106
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.739380419254303,
      "learning_rate": 0.0005998762019098343,
      "loss": 3.2239,
      "step": 2107
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7409529089927673,
      "learning_rate": 0.0005998760843788017,
      "loss": 2.7519,
      "step": 2108
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.8887012004852295,
      "learning_rate": 0.0005998759667920166,
      "loss": 2.9683,
      "step": 2109
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7030766606330872,
      "learning_rate": 0.000599875849149479,
      "loss": 2.6782,
      "step": 2110
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7777583003044128,
      "learning_rate": 0.0005998757314511886,
      "loss": 2.7652,
      "step": 2111
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7343490123748779,
      "learning_rate": 0.0005998756136971457,
      "loss": 2.9765,
      "step": 2112
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7514509558677673,
      "learning_rate": 0.0005998754958873503,
      "loss": 2.9062,
      "step": 2113
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.854041576385498,
      "learning_rate": 0.0005998753780218024,
      "loss": 3.1218,
      "step": 2114
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7630205750465393,
      "learning_rate": 0.000599875260100502,
      "loss": 3.0585,
      "step": 2115
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.8120003938674927,
      "learning_rate": 0.0005998751421234491,
      "loss": 3.0757,
      "step": 2116
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.8253029584884644,
      "learning_rate": 0.0005998750240906437,
      "loss": 2.8626,
      "step": 2117
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7212100028991699,
      "learning_rate": 0.000599874906002086,
      "loss": 2.9307,
      "step": 2118
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7057375907897949,
      "learning_rate": 0.0005998747878577759,
      "loss": 2.531,
      "step": 2119
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7424182891845703,
      "learning_rate": 0.0005998746696577133,
      "loss": 2.8729,
      "step": 2120
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7361471056938171,
      "learning_rate": 0.0005998745514018983,
      "loss": 2.7573,
      "step": 2121
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7897217273712158,
      "learning_rate": 0.0005998744330903311,
      "loss": 2.8474,
      "step": 2122
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.843806266784668,
      "learning_rate": 0.0005998743147230116,
      "loss": 3.0029,
      "step": 2123
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.6793617606163025,
      "learning_rate": 0.0005998741962999397,
      "loss": 2.6286,
      "step": 2124
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7780630588531494,
      "learning_rate": 0.0005998740778211156,
      "loss": 3.0273,
      "step": 2125
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.785753071308136,
      "learning_rate": 0.0005998739592865392,
      "loss": 2.9891,
      "step": 2126
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.6810691356658936,
      "learning_rate": 0.0005998738406962106,
      "loss": 2.812,
      "step": 2127
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7154274582862854,
      "learning_rate": 0.0005998737220501298,
      "loss": 2.8134,
      "step": 2128
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7680500149726868,
      "learning_rate": 0.0005998736033482967,
      "loss": 3.081,
      "step": 2129
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7364724278450012,
      "learning_rate": 0.0005998734845907114,
      "loss": 2.9277,
      "step": 2130
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.733765184879303,
      "learning_rate": 0.0005998733657773742,
      "loss": 2.6942,
      "step": 2131
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7500525116920471,
      "learning_rate": 0.0005998732469082847,
      "loss": 3.2136,
      "step": 2132
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7736193537712097,
      "learning_rate": 0.0005998731279834432,
      "loss": 2.9902,
      "step": 2133
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7646517753601074,
      "learning_rate": 0.0005998730090028496,
      "loss": 2.9616,
      "step": 2134
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.8983700275421143,
      "learning_rate": 0.0005998728899665039,
      "loss": 2.7236,
      "step": 2135
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.700266420841217,
      "learning_rate": 0.0005998727708744062,
      "loss": 2.7999,
      "step": 2136
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.9927412867546082,
      "learning_rate": 0.0005998726517265565,
      "loss": 2.7852,
      "step": 2137
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.8010129332542419,
      "learning_rate": 0.0005998725325229549,
      "loss": 2.8448,
      "step": 2138
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.8738046884536743,
      "learning_rate": 0.0005998724132636012,
      "loss": 2.8963,
      "step": 2139
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7753413915634155,
      "learning_rate": 0.0005998722939484956,
      "loss": 2.9125,
      "step": 2140
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.8237431049346924,
      "learning_rate": 0.0005998721745776381,
      "loss": 2.7993,
      "step": 2141
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.9406463503837585,
      "learning_rate": 0.0005998720551510287,
      "loss": 2.8593,
      "step": 2142
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.731626033782959,
      "learning_rate": 0.0005998719356686674,
      "loss": 2.7771,
      "step": 2143
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.8312716484069824,
      "learning_rate": 0.0005998718161305543,
      "loss": 2.8927,
      "step": 2144
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.8047170639038086,
      "learning_rate": 0.0005998716965366893,
      "loss": 2.9672,
      "step": 2145
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.8034612536430359,
      "learning_rate": 0.0005998715768870725,
      "loss": 2.7561,
      "step": 2146
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7715107202529907,
      "learning_rate": 0.0005998714571817038,
      "loss": 2.8406,
      "step": 2147
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.8159104585647583,
      "learning_rate": 0.0005998713374205835,
      "loss": 2.9353,
      "step": 2148
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7838863730430603,
      "learning_rate": 0.0005998712176037115,
      "loss": 3.0037,
      "step": 2149
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.9163872003555298,
      "learning_rate": 0.0005998710977310876,
      "loss": 2.8456,
      "step": 2150
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.8677927255630493,
      "learning_rate": 0.0005998709778027121,
      "loss": 2.6844,
      "step": 2151
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.0663254261016846,
      "learning_rate": 0.000599870857818585,
      "loss": 2.8991,
      "step": 2152
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.8458361625671387,
      "learning_rate": 0.0005998707377787063,
      "loss": 2.832,
      "step": 2153
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.0529775619506836,
      "learning_rate": 0.0005998706176830758,
      "loss": 2.8912,
      "step": 2154
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.819962203502655,
      "learning_rate": 0.0005998704975316938,
      "loss": 2.9449,
      "step": 2155
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.8511109352111816,
      "learning_rate": 0.00059987037732456,
      "loss": 2.9514,
      "step": 2156
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.8761009573936462,
      "learning_rate": 0.0005998702570616749,
      "loss": 2.973,
      "step": 2157
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.8081458210945129,
      "learning_rate": 0.0005998701367430381,
      "loss": 2.9783,
      "step": 2158
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.1107066869735718,
      "learning_rate": 0.0005998700163686498,
      "loss": 2.8218,
      "step": 2159
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.6996334791183472,
      "learning_rate": 0.0005998698959385101,
      "loss": 2.5368,
      "step": 2160
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.9336655139923096,
      "learning_rate": 0.0005998697754526188,
      "loss": 2.9772,
      "step": 2161
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7101180553436279,
      "learning_rate": 0.0005998696549109761,
      "loss": 2.734,
      "step": 2162
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.0569939613342285,
      "learning_rate": 0.000599869534313582,
      "loss": 2.7277,
      "step": 2163
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.832057535648346,
      "learning_rate": 0.0005998694136604365,
      "loss": 2.7949,
      "step": 2164
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7680583596229553,
      "learning_rate": 0.0005998692929515396,
      "loss": 3.0963,
      "step": 2165
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7596874237060547,
      "learning_rate": 0.0005998691721868913,
      "loss": 3.1467,
      "step": 2166
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7623439431190491,
      "learning_rate": 0.0005998690513664917,
      "loss": 2.9903,
      "step": 2167
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7453188300132751,
      "learning_rate": 0.0005998689304903407,
      "loss": 3.0495,
      "step": 2168
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.8083946704864502,
      "learning_rate": 0.0005998688095584386,
      "loss": 2.8724,
      "step": 2169
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7401097416877747,
      "learning_rate": 0.000599868688570785,
      "loss": 2.7523,
      "step": 2170
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7895482778549194,
      "learning_rate": 0.0005998685675273803,
      "loss": 3.1318,
      "step": 2171
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7513300180435181,
      "learning_rate": 0.0005998684464282243,
      "loss": 3.0864,
      "step": 2172
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7946348786354065,
      "learning_rate": 0.0005998683252733171,
      "loss": 2.9218,
      "step": 2173
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7802730798721313,
      "learning_rate": 0.0005998682040626587,
      "loss": 2.7947,
      "step": 2174
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7378263473510742,
      "learning_rate": 0.0005998680827962492,
      "loss": 2.9494,
      "step": 2175
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.8644872903823853,
      "learning_rate": 0.0005998679614740885,
      "loss": 2.9338,
      "step": 2176
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7453209757804871,
      "learning_rate": 0.0005998678400961768,
      "loss": 3.1315,
      "step": 2177
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7459641098976135,
      "learning_rate": 0.000599867718662514,
      "loss": 3.0194,
      "step": 2178
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7703120112419128,
      "learning_rate": 0.0005998675971731,
      "loss": 2.8042,
      "step": 2179
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7748092412948608,
      "learning_rate": 0.0005998674756279351,
      "loss": 2.8279,
      "step": 2180
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.9259583950042725,
      "learning_rate": 0.0005998673540270191,
      "loss": 2.9008,
      "step": 2181
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.8613226413726807,
      "learning_rate": 0.000599867232370352,
      "loss": 2.9707,
      "step": 2182
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.8570031523704529,
      "learning_rate": 0.000599867110657934,
      "loss": 2.8877,
      "step": 2183
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.2120418548583984,
      "learning_rate": 0.0005998669888897651,
      "loss": 2.9931,
      "step": 2184
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.8412429094314575,
      "learning_rate": 0.0005998668670658452,
      "loss": 2.8926,
      "step": 2185
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.8791328072547913,
      "learning_rate": 0.0005998667451861744,
      "loss": 2.9732,
      "step": 2186
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.8673864006996155,
      "learning_rate": 0.0005998666232507528,
      "loss": 2.8695,
      "step": 2187
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.9333680868148804,
      "learning_rate": 0.0005998665012595802,
      "loss": 3.0643,
      "step": 2188
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.9605239629745483,
      "learning_rate": 0.0005998663792126568,
      "loss": 2.9534,
      "step": 2189
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7084693312644958,
      "learning_rate": 0.0005998662571099826,
      "loss": 2.8392,
      "step": 2190
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.8638198971748352,
      "learning_rate": 0.0005998661349515576,
      "loss": 3.0473,
      "step": 2191
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.6958181262016296,
      "learning_rate": 0.0005998660127373818,
      "loss": 2.9712,
      "step": 2192
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7434493899345398,
      "learning_rate": 0.0005998658904674553,
      "loss": 2.9916,
      "step": 2193
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7223270535469055,
      "learning_rate": 0.000599865768141778,
      "loss": 2.8295,
      "step": 2194
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7886364459991455,
      "learning_rate": 0.0005998656457603501,
      "loss": 3.0193,
      "step": 2195
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.8021574020385742,
      "learning_rate": 0.0005998655233231715,
      "loss": 2.9748,
      "step": 2196
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.812633752822876,
      "learning_rate": 0.0005998654008302421,
      "loss": 3.0195,
      "step": 2197
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.9375292658805847,
      "learning_rate": 0.0005998652782815622,
      "loss": 2.6785,
      "step": 2198
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7362523674964905,
      "learning_rate": 0.0005998651556771315,
      "loss": 2.8483,
      "step": 2199
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7501875162124634,
      "learning_rate": 0.0005998650330169504,
      "loss": 3.073,
      "step": 2200
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7373062372207642,
      "learning_rate": 0.0005998649103010187,
      "loss": 2.8945,
      "step": 2201
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7872507572174072,
      "learning_rate": 0.0005998647875293364,
      "loss": 2.6715,
      "step": 2202
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7064765095710754,
      "learning_rate": 0.0005998646647019036,
      "loss": 2.9128,
      "step": 2203
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7819071412086487,
      "learning_rate": 0.0005998645418187204,
      "loss": 2.8537,
      "step": 2204
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7725965976715088,
      "learning_rate": 0.0005998644188797866,
      "loss": 3.0257,
      "step": 2205
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7424801588058472,
      "learning_rate": 0.0005998642958851023,
      "loss": 3.019,
      "step": 2206
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7807191014289856,
      "learning_rate": 0.0005998641728346677,
      "loss": 2.822,
      "step": 2207
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.6975749731063843,
      "learning_rate": 0.0005998640497284826,
      "loss": 2.9384,
      "step": 2208
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7773790955543518,
      "learning_rate": 0.0005998639265665471,
      "loss": 2.9438,
      "step": 2209
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7759329676628113,
      "learning_rate": 0.0005998638033488612,
      "loss": 3.011,
      "step": 2210
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7540960907936096,
      "learning_rate": 0.0005998636800754252,
      "loss": 2.9802,
      "step": 2211
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7485987544059753,
      "learning_rate": 0.0005998635567462386,
      "loss": 2.9211,
      "step": 2212
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7385587692260742,
      "learning_rate": 0.0005998634333613018,
      "loss": 2.9572,
      "step": 2213
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.8219103813171387,
      "learning_rate": 0.0005998633099206147,
      "loss": 2.8993,
      "step": 2214
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.8083028793334961,
      "learning_rate": 0.0005998631864241775,
      "loss": 3.0502,
      "step": 2215
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7482160329818726,
      "learning_rate": 0.0005998630628719899,
      "loss": 2.8819,
      "step": 2216
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7279016375541687,
      "learning_rate": 0.0005998629392640522,
      "loss": 2.8338,
      "step": 2217
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7518342733383179,
      "learning_rate": 0.0005998628156003643,
      "loss": 2.7915,
      "step": 2218
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7925392985343933,
      "learning_rate": 0.0005998626918809262,
      "loss": 3.0725,
      "step": 2219
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7853243947029114,
      "learning_rate": 0.0005998625681057381,
      "loss": 2.7795,
      "step": 2220
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.6792628169059753,
      "learning_rate": 0.0005998624442747996,
      "loss": 2.9034,
      "step": 2221
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.732712984085083,
      "learning_rate": 0.0005998623203881112,
      "loss": 3.0054,
      "step": 2222
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7334317564964294,
      "learning_rate": 0.0005998621964456728,
      "loss": 2.7182,
      "step": 2223
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.847946286201477,
      "learning_rate": 0.0005998620724474843,
      "loss": 2.7947,
      "step": 2224
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.774483323097229,
      "learning_rate": 0.0005998619483935457,
      "loss": 2.8883,
      "step": 2225
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7527991533279419,
      "learning_rate": 0.0005998618242838571,
      "loss": 2.9911,
      "step": 2226
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.743852436542511,
      "learning_rate": 0.0005998617001184187,
      "loss": 2.7828,
      "step": 2227
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.8369457125663757,
      "learning_rate": 0.0005998615758972302,
      "loss": 2.6337,
      "step": 2228
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.8201175928115845,
      "learning_rate": 0.0005998614516202918,
      "loss": 2.7096,
      "step": 2229
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7539345026016235,
      "learning_rate": 0.0005998613272876035,
      "loss": 3.0773,
      "step": 2230
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.6876362562179565,
      "learning_rate": 0.0005998612028991654,
      "loss": 2.7897,
      "step": 2231
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7370336055755615,
      "learning_rate": 0.0005998610784549774,
      "loss": 2.7875,
      "step": 2232
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7817413210868835,
      "learning_rate": 0.0005998609539550394,
      "loss": 3.0133,
      "step": 2233
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.8312408924102783,
      "learning_rate": 0.0005998608293993518,
      "loss": 2.9253,
      "step": 2234
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7041609883308411,
      "learning_rate": 0.0005998607047879144,
      "loss": 2.7486,
      "step": 2235
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7850757837295532,
      "learning_rate": 0.0005998605801207271,
      "loss": 2.9649,
      "step": 2236
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7520643472671509,
      "learning_rate": 0.0005998604553977901,
      "loss": 3.0044,
      "step": 2237
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7588242292404175,
      "learning_rate": 0.0005998603306191034,
      "loss": 3.0424,
      "step": 2238
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7710031270980835,
      "learning_rate": 0.000599860205784667,
      "loss": 3.063,
      "step": 2239
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.9691678881645203,
      "learning_rate": 0.0005998600808944809,
      "loss": 2.7551,
      "step": 2240
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.785449743270874,
      "learning_rate": 0.0005998599559485453,
      "loss": 2.7844,
      "step": 2241
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7395849227905273,
      "learning_rate": 0.0005998598309468599,
      "loss": 2.9464,
      "step": 2242
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.8870104551315308,
      "learning_rate": 0.000599859705889425,
      "loss": 2.9275,
      "step": 2243
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7338253259658813,
      "learning_rate": 0.0005998595807762405,
      "loss": 2.9217,
      "step": 2244
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.9412561058998108,
      "learning_rate": 0.0005998594556073063,
      "loss": 2.7597,
      "step": 2245
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.6768505573272705,
      "learning_rate": 0.0005998593303826228,
      "loss": 2.8375,
      "step": 2246
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.929733157157898,
      "learning_rate": 0.0005998592051021897,
      "loss": 2.6874,
      "step": 2247
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.8590959310531616,
      "learning_rate": 0.0005998590797660071,
      "loss": 2.7438,
      "step": 2248
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7057855725288391,
      "learning_rate": 0.0005998589543740751,
      "loss": 3.091,
      "step": 2249
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7248392105102539,
      "learning_rate": 0.0005998588289263934,
      "loss": 2.8451,
      "step": 2250
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7512311935424805,
      "learning_rate": 0.0005998587034229627,
      "loss": 3.0016,
      "step": 2251
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.766473650932312,
      "learning_rate": 0.0005998585778637822,
      "loss": 3.0861,
      "step": 2252
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7659862041473389,
      "learning_rate": 0.0005998584522488526,
      "loss": 2.7635,
      "step": 2253
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.8032305836677551,
      "learning_rate": 0.0005998583265781736,
      "loss": 3.0086,
      "step": 2254
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7365007400512695,
      "learning_rate": 0.0005998582008517452,
      "loss": 3.0203,
      "step": 2255
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7971333861351013,
      "learning_rate": 0.0005998580750695676,
      "loss": 2.8103,
      "step": 2256
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.6963884234428406,
      "learning_rate": 0.0005998579492316405,
      "loss": 3.0986,
      "step": 2257
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.682475745677948,
      "learning_rate": 0.0005998578233379644,
      "loss": 2.6985,
      "step": 2258
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7497090101242065,
      "learning_rate": 0.000599857697388539,
      "loss": 2.9294,
      "step": 2259
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.6987460851669312,
      "learning_rate": 0.0005998575713833643,
      "loss": 2.8033,
      "step": 2260
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.6826688647270203,
      "learning_rate": 0.0005998574453224406,
      "loss": 2.8828,
      "step": 2261
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7145518660545349,
      "learning_rate": 0.0005998573192057676,
      "loss": 2.9242,
      "step": 2262
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.8627458810806274,
      "learning_rate": 0.0005998571930333455,
      "loss": 2.8684,
      "step": 2263
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.8001700043678284,
      "learning_rate": 0.0005998570668051743,
      "loss": 2.866,
      "step": 2264
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.8281999826431274,
      "learning_rate": 0.000599856940521254,
      "loss": 2.8807,
      "step": 2265
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.6781601309776306,
      "learning_rate": 0.0005998568141815846,
      "loss": 2.7505,
      "step": 2266
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.98170405626297,
      "learning_rate": 0.0005998566877861662,
      "loss": 2.9135,
      "step": 2267
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7855561375617981,
      "learning_rate": 0.0005998565613349988,
      "loss": 2.8308,
      "step": 2268
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7369747161865234,
      "learning_rate": 0.0005998564348280825,
      "loss": 2.803,
      "step": 2269
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.1096278429031372,
      "learning_rate": 0.000599856308265417,
      "loss": 2.8752,
      "step": 2270
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.6879796385765076,
      "learning_rate": 0.0005998561816470026,
      "loss": 2.6792,
      "step": 2271
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.1724144220352173,
      "learning_rate": 0.0005998560549728394,
      "loss": 2.8568,
      "step": 2272
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7918099164962769,
      "learning_rate": 0.0005998559282429272,
      "loss": 2.6707,
      "step": 2273
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7714400887489319,
      "learning_rate": 0.0005998558014572662,
      "loss": 2.8864,
      "step": 2274
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.9156453013420105,
      "learning_rate": 0.0005998556746158563,
      "loss": 2.816,
      "step": 2275
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.8097001314163208,
      "learning_rate": 0.0005998555477186975,
      "loss": 2.9513,
      "step": 2276
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.1482927799224854,
      "learning_rate": 0.00059985542076579,
      "loss": 2.9718,
      "step": 2277
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.866282045841217,
      "learning_rate": 0.0005998552937571337,
      "loss": 2.5672,
      "step": 2278
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.0526798963546753,
      "learning_rate": 0.0005998551666927286,
      "loss": 3.048,
      "step": 2279
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.802226185798645,
      "learning_rate": 0.0005998550395725749,
      "loss": 2.7584,
      "step": 2280
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.0500012636184692,
      "learning_rate": 0.0005998549123966722,
      "loss": 2.944,
      "step": 2281
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.8515597581863403,
      "learning_rate": 0.000599854785165021,
      "loss": 2.7573,
      "step": 2282
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.9499987959861755,
      "learning_rate": 0.0005998546578776212,
      "loss": 2.991,
      "step": 2283
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.8618533611297607,
      "learning_rate": 0.0005998545305344727,
      "loss": 3.0435,
      "step": 2284
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.6398425698280334,
      "learning_rate": 0.0005998544031355755,
      "loss": 2.6206,
      "step": 2285
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.880530059337616,
      "learning_rate": 0.0005998542756809298,
      "loss": 2.8548,
      "step": 2286
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7209012508392334,
      "learning_rate": 0.0005998541481705354,
      "loss": 2.9561,
      "step": 2287
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.8306952118873596,
      "learning_rate": 0.0005998540206043927,
      "loss": 2.7294,
      "step": 2288
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.8277239203453064,
      "learning_rate": 0.0005998538929825013,
      "loss": 2.8316,
      "step": 2289
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.8030091524124146,
      "learning_rate": 0.0005998537653048614,
      "loss": 3.0483,
      "step": 2290
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.8911025524139404,
      "learning_rate": 0.0005998536375714732,
      "loss": 2.7995,
      "step": 2291
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.062931776046753,
      "learning_rate": 0.0005998535097823363,
      "loss": 2.9439,
      "step": 2292
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7045261263847351,
      "learning_rate": 0.0005998533819374511,
      "loss": 2.8087,
      "step": 2293
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.886135995388031,
      "learning_rate": 0.0005998532540368174,
      "loss": 2.8164,
      "step": 2294
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7464367151260376,
      "learning_rate": 0.0005998531260804355,
      "loss": 3.1545,
      "step": 2295
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7644054889678955,
      "learning_rate": 0.0005998529980683051,
      "loss": 2.6256,
      "step": 2296
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.8784204721450806,
      "learning_rate": 0.0005998528700004264,
      "loss": 3.1454,
      "step": 2297
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7353897094726562,
      "learning_rate": 0.0005998527418767994,
      "loss": 2.8158,
      "step": 2298
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.8684032559394836,
      "learning_rate": 0.0005998526136974242,
      "loss": 2.6783,
      "step": 2299
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7528250813484192,
      "learning_rate": 0.0005998524854623006,
      "loss": 2.6126,
      "step": 2300
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.8394548892974854,
      "learning_rate": 0.0005998523571714287,
      "loss": 2.9407,
      "step": 2301
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7145257592201233,
      "learning_rate": 0.0005998522288248087,
      "loss": 2.9323,
      "step": 2302
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.8929355144500732,
      "learning_rate": 0.0005998521004224405,
      "loss": 2.948,
      "step": 2303
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.8145368695259094,
      "learning_rate": 0.0005998519719643242,
      "loss": 2.7966,
      "step": 2304
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.001183271408081,
      "learning_rate": 0.0005998518434504597,
      "loss": 2.9215,
      "step": 2305
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.742191731929779,
      "learning_rate": 0.0005998517148808471,
      "loss": 2.7324,
      "step": 2306
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.0163025856018066,
      "learning_rate": 0.0005998515862554864,
      "loss": 3.1199,
      "step": 2307
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.9120702147483826,
      "learning_rate": 0.0005998514575743776,
      "loss": 2.6268,
      "step": 2308
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7047255635261536,
      "learning_rate": 0.0005998513288375208,
      "loss": 2.6298,
      "step": 2309
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.9496548771858215,
      "learning_rate": 0.000599851200044916,
      "loss": 3.0528,
      "step": 2310
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.9414237141609192,
      "learning_rate": 0.0005998510711965631,
      "loss": 3.0056,
      "step": 2311
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7818029522895813,
      "learning_rate": 0.0005998509422924622,
      "loss": 3.0769,
      "step": 2312
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.1099581718444824,
      "learning_rate": 0.0005998508133326136,
      "loss": 2.7464,
      "step": 2313
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.8610965609550476,
      "learning_rate": 0.0005998506843170169,
      "loss": 2.7578,
      "step": 2314
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.798303484916687,
      "learning_rate": 0.0005998505552456722,
      "loss": 2.759,
      "step": 2315
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.0361562967300415,
      "learning_rate": 0.0005998504261185797,
      "loss": 2.9784,
      "step": 2316
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.8708915114402771,
      "learning_rate": 0.0005998502969357394,
      "loss": 2.8159,
      "step": 2317
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7406286001205444,
      "learning_rate": 0.0005998501676971513,
      "loss": 2.963,
      "step": 2318
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.8186811804771423,
      "learning_rate": 0.0005998500384028153,
      "loss": 2.6108,
      "step": 2319
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.1717573404312134,
      "learning_rate": 0.0005998499090527315,
      "loss": 2.8557,
      "step": 2320
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7300366163253784,
      "learning_rate": 0.0005998497796469,
      "loss": 3.0253,
      "step": 2321
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7483167052268982,
      "learning_rate": 0.0005998496501853208,
      "loss": 2.8508,
      "step": 2322
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.770897388458252,
      "learning_rate": 0.0005998495206679938,
      "loss": 2.9259,
      "step": 2323
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.6702894568443298,
      "learning_rate": 0.0005998493910949192,
      "loss": 2.7403,
      "step": 2324
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7148359417915344,
      "learning_rate": 0.0005998492614660969,
      "loss": 2.8476,
      "step": 2325
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7925755977630615,
      "learning_rate": 0.0005998491317815269,
      "loss": 2.8418,
      "step": 2326
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.9236418008804321,
      "learning_rate": 0.0005998490020412094,
      "loss": 2.9166,
      "step": 2327
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7100194096565247,
      "learning_rate": 0.0005998488722451442,
      "loss": 2.632,
      "step": 2328
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.8061372637748718,
      "learning_rate": 0.0005998487423933315,
      "loss": 2.836,
      "step": 2329
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7611356973648071,
      "learning_rate": 0.0005998486124857714,
      "loss": 2.8672,
      "step": 2330
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7807998061180115,
      "learning_rate": 0.0005998484825224637,
      "loss": 2.9718,
      "step": 2331
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7426683306694031,
      "learning_rate": 0.0005998483525034083,
      "loss": 3.0357,
      "step": 2332
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.774451732635498,
      "learning_rate": 0.0005998482224286055,
      "loss": 2.9465,
      "step": 2333
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.8227857947349548,
      "learning_rate": 0.0005998480922980554,
      "loss": 3.0123,
      "step": 2334
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.731563150882721,
      "learning_rate": 0.0005998479621117578,
      "loss": 2.9669,
      "step": 2335
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7273282408714294,
      "learning_rate": 0.0005998478318697129,
      "loss": 2.9493,
      "step": 2336
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.8235801458358765,
      "learning_rate": 0.0005998477015719204,
      "loss": 2.67,
      "step": 2337
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7950217723846436,
      "learning_rate": 0.0005998475712183808,
      "loss": 2.8848,
      "step": 2338
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7686046361923218,
      "learning_rate": 0.0005998474408090938,
      "loss": 2.7079,
      "step": 2339
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.900385320186615,
      "learning_rate": 0.0005998473103440594,
      "loss": 3.2949,
      "step": 2340
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7082587480545044,
      "learning_rate": 0.0005998471798232777,
      "loss": 2.868,
      "step": 2341
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.8369800448417664,
      "learning_rate": 0.000599847049246749,
      "loss": 2.8661,
      "step": 2342
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7731757760047913,
      "learning_rate": 0.0005998469186144729,
      "loss": 2.8598,
      "step": 2343
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.8849717974662781,
      "learning_rate": 0.0005998467879264496,
      "loss": 2.8745,
      "step": 2344
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7425972819328308,
      "learning_rate": 0.0005998466571826791,
      "loss": 2.8926,
      "step": 2345
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7311619520187378,
      "learning_rate": 0.0005998465263831615,
      "loss": 2.6505,
      "step": 2346
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7287903428077698,
      "learning_rate": 0.0005998463955278967,
      "loss": 2.6861,
      "step": 2347
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7717069983482361,
      "learning_rate": 0.0005998462646168848,
      "loss": 3.0763,
      "step": 2348
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7191064953804016,
      "learning_rate": 0.0005998461336501259,
      "loss": 3.1895,
      "step": 2349
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.8731751441955566,
      "learning_rate": 0.00059984600262762,
      "loss": 3.0048,
      "step": 2350
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7734737992286682,
      "learning_rate": 0.000599845871549367,
      "loss": 2.5639,
      "step": 2351
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.8399993777275085,
      "learning_rate": 0.0005998457404153669,
      "loss": 2.8288,
      "step": 2352
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.9529133439064026,
      "learning_rate": 0.0005998456092256199,
      "loss": 2.9531,
      "step": 2353
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.8587525486946106,
      "learning_rate": 0.000599845477980126,
      "loss": 3.0815,
      "step": 2354
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7836793065071106,
      "learning_rate": 0.000599845346678885,
      "loss": 2.9414,
      "step": 2355
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.0095248222351074,
      "learning_rate": 0.0005998452153218972,
      "loss": 2.9218,
      "step": 2356
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.8416150212287903,
      "learning_rate": 0.0005998450839091625,
      "loss": 2.7711,
      "step": 2357
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.8776577115058899,
      "learning_rate": 0.000599844952440681,
      "loss": 2.8801,
      "step": 2358
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.9919650554656982,
      "learning_rate": 0.0005998448209164525,
      "loss": 2.8007,
      "step": 2359
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7088320255279541,
      "learning_rate": 0.0005998446893364773,
      "loss": 2.9049,
      "step": 2360
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.8875607252120972,
      "learning_rate": 0.0005998445577007554,
      "loss": 2.8609,
      "step": 2361
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.8312252163887024,
      "learning_rate": 0.0005998444260092866,
      "loss": 2.7983,
      "step": 2362
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7140122652053833,
      "learning_rate": 0.0005998442942620711,
      "loss": 2.719,
      "step": 2363
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.0328558683395386,
      "learning_rate": 0.0005998441624591089,
      "loss": 2.9803,
      "step": 2364
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7731636762619019,
      "learning_rate": 0.0005998440306004,
      "loss": 2.9284,
      "step": 2365
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.823769211769104,
      "learning_rate": 0.0005998438986859444,
      "loss": 2.9173,
      "step": 2366
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.0342984199523926,
      "learning_rate": 0.0005998437667157422,
      "loss": 2.8083,
      "step": 2367
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.704213559627533,
      "learning_rate": 0.0005998436346897934,
      "loss": 2.9861,
      "step": 2368
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.8889617323875427,
      "learning_rate": 0.000599843502608098,
      "loss": 2.6191,
      "step": 2369
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7719780802726746,
      "learning_rate": 0.000599843370470656,
      "loss": 2.961,
      "step": 2370
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7445523142814636,
      "learning_rate": 0.0005998432382774675,
      "loss": 2.9128,
      "step": 2371
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.8932211995124817,
      "learning_rate": 0.0005998431060285325,
      "loss": 2.7921,
      "step": 2372
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.9075191617012024,
      "learning_rate": 0.0005998429737238509,
      "loss": 2.5535,
      "step": 2373
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.6858242154121399,
      "learning_rate": 0.0005998428413634229,
      "loss": 2.7247,
      "step": 2374
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.107664942741394,
      "learning_rate": 0.0005998427089472484,
      "loss": 2.6067,
      "step": 2375
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.253530740737915,
      "learning_rate": 0.0005998425764753276,
      "loss": 2.6811,
      "step": 2376
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7553577423095703,
      "learning_rate": 0.0005998424439476604,
      "loss": 2.9905,
      "step": 2377
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.1492921113967896,
      "learning_rate": 0.0005998423113642466,
      "loss": 2.8831,
      "step": 2378
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7737727165222168,
      "learning_rate": 0.0005998421787250867,
      "loss": 3.199,
      "step": 2379
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7955979704856873,
      "learning_rate": 0.0005998420460301804,
      "loss": 2.7505,
      "step": 2380
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.9427530169487,
      "learning_rate": 0.0005998419132795278,
      "loss": 2.6148,
      "step": 2381
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.8329724073410034,
      "learning_rate": 0.0005998417804731289,
      "loss": 2.9494,
      "step": 2382
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.8670951128005981,
      "learning_rate": 0.0005998416476109838,
      "loss": 3.0263,
      "step": 2383
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.6781741380691528,
      "learning_rate": 0.0005998415146930924,
      "loss": 2.8305,
      "step": 2384
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.803211510181427,
      "learning_rate": 0.0005998413817194549,
      "loss": 2.8101,
      "step": 2385
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7657666802406311,
      "learning_rate": 0.0005998412486900712,
      "loss": 2.8912,
      "step": 2386
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7484264373779297,
      "learning_rate": 0.0005998411156049414,
      "loss": 3.2068,
      "step": 2387
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.8414695858955383,
      "learning_rate": 0.0005998409824640654,
      "loss": 2.7435,
      "step": 2388
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.8683329820632935,
      "learning_rate": 0.0005998408492674433,
      "loss": 2.9312,
      "step": 2389
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.6502180099487305,
      "learning_rate": 0.0005998407160150752,
      "loss": 3.0019,
      "step": 2390
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7694963216781616,
      "learning_rate": 0.000599840582706961,
      "loss": 2.9762,
      "step": 2391
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7739273905754089,
      "learning_rate": 0.0005998404493431007,
      "loss": 2.9414,
      "step": 2392
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7625907063484192,
      "learning_rate": 0.0005998403159234945,
      "loss": 2.8147,
      "step": 2393
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.8206735253334045,
      "learning_rate": 0.0005998401824481423,
      "loss": 2.7629,
      "step": 2394
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.875849187374115,
      "learning_rate": 0.0005998400489170442,
      "loss": 2.863,
      "step": 2395
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.77391517162323,
      "learning_rate": 0.0005998399153302001,
      "loss": 2.9984,
      "step": 2396
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.9989989399909973,
      "learning_rate": 0.0005998397816876101,
      "loss": 3.0258,
      "step": 2397
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7105626463890076,
      "learning_rate": 0.0005998396479892742,
      "loss": 2.6991,
      "step": 2398
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.822279691696167,
      "learning_rate": 0.0005998395142351925,
      "loss": 2.5973,
      "step": 2399
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7257847189903259,
      "learning_rate": 0.000599839380425365,
      "loss": 2.8818,
      "step": 2400
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7532752156257629,
      "learning_rate": 0.0005998392465597917,
      "loss": 2.9624,
      "step": 2401
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7714336514472961,
      "learning_rate": 0.0005998391126384726,
      "loss": 3.1343,
      "step": 2402
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.832524836063385,
      "learning_rate": 0.0005998389786614078,
      "loss": 2.833,
      "step": 2403
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7762848734855652,
      "learning_rate": 0.0005998388446285971,
      "loss": 2.8391,
      "step": 2404
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7170887589454651,
      "learning_rate": 0.0005998387105400409,
      "loss": 3.015,
      "step": 2405
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.8796876072883606,
      "learning_rate": 0.0005998385763957389,
      "loss": 2.8916,
      "step": 2406
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7173342108726501,
      "learning_rate": 0.0005998384421956913,
      "loss": 2.6613,
      "step": 2407
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.9465100765228271,
      "learning_rate": 0.0005998383079398981,
      "loss": 2.7645,
      "step": 2408
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.8029413819313049,
      "learning_rate": 0.0005998381736283592,
      "loss": 3.0564,
      "step": 2409
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7370434999465942,
      "learning_rate": 0.0005998380392610749,
      "loss": 3.076,
      "step": 2410
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.8199747204780579,
      "learning_rate": 0.0005998379048380449,
      "loss": 2.9075,
      "step": 2411
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7599244117736816,
      "learning_rate": 0.0005998377703592694,
      "loss": 2.7736,
      "step": 2412
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7684251666069031,
      "learning_rate": 0.0005998376358247484,
      "loss": 2.7466,
      "step": 2413
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.9347872138023376,
      "learning_rate": 0.0005998375012344819,
      "loss": 2.6543,
      "step": 2414
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.8779689073562622,
      "learning_rate": 0.00059983736658847,
      "loss": 2.9278,
      "step": 2415
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.8176613450050354,
      "learning_rate": 0.0005998372318867126,
      "loss": 2.8613,
      "step": 2416
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.9781022071838379,
      "learning_rate": 0.0005998370971292098,
      "loss": 2.5768,
      "step": 2417
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.8209463953971863,
      "learning_rate": 0.0005998369623159618,
      "loss": 2.8745,
      "step": 2418
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.9451572895050049,
      "learning_rate": 0.0005998368274469683,
      "loss": 3.0057,
      "step": 2419
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.8492891192436218,
      "learning_rate": 0.0005998366925222295,
      "loss": 2.8993,
      "step": 2420
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7159214615821838,
      "learning_rate": 0.0005998365575417454,
      "loss": 3.0567,
      "step": 2421
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.8622539639472961,
      "learning_rate": 0.000599836422505516,
      "loss": 2.6717,
      "step": 2422
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7630833983421326,
      "learning_rate": 0.0005998362874135414,
      "loss": 2.8247,
      "step": 2423
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7784586548805237,
      "learning_rate": 0.0005998361522658214,
      "loss": 2.8911,
      "step": 2424
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.8904475569725037,
      "learning_rate": 0.0005998360170623565,
      "loss": 3.1072,
      "step": 2425
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.6960905194282532,
      "learning_rate": 0.0005998358818031462,
      "loss": 2.7931,
      "step": 2426
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.701079785823822,
      "learning_rate": 0.0005998357464881908,
      "loss": 2.8776,
      "step": 2427
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7923266291618347,
      "learning_rate": 0.0005998356111174903,
      "loss": 2.9915,
      "step": 2428
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.789756715297699,
      "learning_rate": 0.0005998354756910448,
      "loss": 2.8271,
      "step": 2429
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.71023029088974,
      "learning_rate": 0.0005998353402088541,
      "loss": 2.7056,
      "step": 2430
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.9586816430091858,
      "learning_rate": 0.0005998352046709183,
      "loss": 2.9823,
      "step": 2431
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7439730167388916,
      "learning_rate": 0.0005998350690772376,
      "loss": 2.7977,
      "step": 2432
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.6694998741149902,
      "learning_rate": 0.0005998349334278118,
      "loss": 2.8939,
      "step": 2433
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7960318326950073,
      "learning_rate": 0.000599834797722641,
      "loss": 3.0016,
      "step": 2434
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.9510160684585571,
      "learning_rate": 0.0005998346619617254,
      "loss": 2.514,
      "step": 2435
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7119179964065552,
      "learning_rate": 0.0005998345261450647,
      "loss": 2.9957,
      "step": 2436
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.6876430511474609,
      "learning_rate": 0.0005998343902726593,
      "loss": 3.0246,
      "step": 2437
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.0144445896148682,
      "learning_rate": 0.0005998342543445089,
      "loss": 2.7165,
      "step": 2438
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.811950147151947,
      "learning_rate": 0.0005998341183606137,
      "loss": 2.8497,
      "step": 2439
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.989866316318512,
      "learning_rate": 0.0005998339823209737,
      "loss": 2.7169,
      "step": 2440
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.8944609761238098,
      "learning_rate": 0.0005998338462255889,
      "loss": 3.01,
      "step": 2441
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.8019275665283203,
      "learning_rate": 0.0005998337100744592,
      "loss": 2.8346,
      "step": 2442
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.0327564477920532,
      "learning_rate": 0.0005998335738675849,
      "loss": 2.9789,
      "step": 2443
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7681899070739746,
      "learning_rate": 0.0005998334376049657,
      "loss": 2.8355,
      "step": 2444
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7875962853431702,
      "learning_rate": 0.000599833301286602,
      "loss": 2.8518,
      "step": 2445
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.790354311466217,
      "learning_rate": 0.0005998331649124936,
      "loss": 2.8163,
      "step": 2446
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.8199900388717651,
      "learning_rate": 0.0005998330284826405,
      "loss": 2.8082,
      "step": 2447
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.8471972346305847,
      "learning_rate": 0.0005998328919970428,
      "loss": 2.9767,
      "step": 2448
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.8407652378082275,
      "learning_rate": 0.0005998327554557004,
      "loss": 2.8717,
      "step": 2449
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7852345705032349,
      "learning_rate": 0.0005998326188586136,
      "loss": 3.1093,
      "step": 2450
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.680059015750885,
      "learning_rate": 0.0005998324822057822,
      "loss": 2.6665,
      "step": 2451
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.6612427830696106,
      "learning_rate": 0.0005998323454972063,
      "loss": 2.8885,
      "step": 2452
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7271769642829895,
      "learning_rate": 0.0005998322087328858,
      "loss": 2.8155,
      "step": 2453
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7404655814170837,
      "learning_rate": 0.000599832071912821,
      "loss": 2.8745,
      "step": 2454
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.8080680966377258,
      "learning_rate": 0.0005998319350370116,
      "loss": 2.7321,
      "step": 2455
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7970831990242004,
      "learning_rate": 0.0005998317981054578,
      "loss": 2.8053,
      "step": 2456
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.763114333152771,
      "learning_rate": 0.0005998316611181597,
      "loss": 2.737,
      "step": 2457
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.9069132804870605,
      "learning_rate": 0.0005998315240751172,
      "loss": 2.9324,
      "step": 2458
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.6921560168266296,
      "learning_rate": 0.0005998313869763303,
      "loss": 2.9423,
      "step": 2459
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7394306659698486,
      "learning_rate": 0.0005998312498217992,
      "loss": 2.949,
      "step": 2460
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.705560028553009,
      "learning_rate": 0.0005998311126115237,
      "loss": 3.0492,
      "step": 2461
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7354623079299927,
      "learning_rate": 0.000599830975345504,
      "loss": 2.8281,
      "step": 2462
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.6635681390762329,
      "learning_rate": 0.00059983083802374,
      "loss": 2.8642,
      "step": 2463
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.24271559715271,
      "learning_rate": 0.0005998307006462319,
      "loss": 2.8715,
      "step": 2464
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.6966912746429443,
      "learning_rate": 0.0005998305632129795,
      "loss": 2.9081,
      "step": 2465
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7546772360801697,
      "learning_rate": 0.0005998304257239831,
      "loss": 2.737,
      "step": 2466
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7470026016235352,
      "learning_rate": 0.0005998302881792424,
      "loss": 2.5254,
      "step": 2467
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7391142249107361,
      "learning_rate": 0.0005998301505787577,
      "loss": 3.1157,
      "step": 2468
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.720024824142456,
      "learning_rate": 0.0005998300129225287,
      "loss": 3.0586,
      "step": 2469
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.8291887640953064,
      "learning_rate": 0.0005998298752105559,
      "loss": 2.8974,
      "step": 2470
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.76485675573349,
      "learning_rate": 0.000599829737442839,
      "loss": 2.9096,
      "step": 2471
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.9115382432937622,
      "learning_rate": 0.000599829599619378,
      "loss": 2.6706,
      "step": 2472
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.8411963582038879,
      "learning_rate": 0.0005998294617401731,
      "loss": 2.7182,
      "step": 2473
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.784993588924408,
      "learning_rate": 0.0005998293238052241,
      "loss": 3.0968,
      "step": 2474
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7429311275482178,
      "learning_rate": 0.0005998291858145314,
      "loss": 2.9684,
      "step": 2475
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7618846297264099,
      "learning_rate": 0.0005998290477680946,
      "loss": 2.9595,
      "step": 2476
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.801938533782959,
      "learning_rate": 0.000599828909665914,
      "loss": 2.765,
      "step": 2477
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7320749163627625,
      "learning_rate": 0.0005998287715079895,
      "loss": 2.7542,
      "step": 2478
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.82017582654953,
      "learning_rate": 0.0005998286332943212,
      "loss": 2.8088,
      "step": 2479
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.6984558701515198,
      "learning_rate": 0.0005998284950249093,
      "loss": 2.8547,
      "step": 2480
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7855092287063599,
      "learning_rate": 0.0005998283566997533,
      "loss": 2.8309,
      "step": 2481
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.9422295093536377,
      "learning_rate": 0.0005998282183188537,
      "loss": 3.0008,
      "step": 2482
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.6921332478523254,
      "learning_rate": 0.0005998280798822103,
      "loss": 2.6654,
      "step": 2483
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7000443935394287,
      "learning_rate": 0.0005998279413898233,
      "loss": 2.6602,
      "step": 2484
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.9521160125732422,
      "learning_rate": 0.0005998278028416926,
      "loss": 2.73,
      "step": 2485
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7351453900337219,
      "learning_rate": 0.0005998276642378182,
      "loss": 2.9437,
      "step": 2486
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.882792055606842,
      "learning_rate": 0.0005998275255782003,
      "loss": 2.9317,
      "step": 2487
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.8557993173599243,
      "learning_rate": 0.0005998273868628387,
      "loss": 2.8613,
      "step": 2488
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7701265215873718,
      "learning_rate": 0.0005998272480917335,
      "loss": 2.843,
      "step": 2489
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.8798103332519531,
      "learning_rate": 0.0005998271092648848,
      "loss": 2.8533,
      "step": 2490
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7267061471939087,
      "learning_rate": 0.0005998269703822926,
      "loss": 2.7794,
      "step": 2491
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7268227934837341,
      "learning_rate": 0.0005998268314439569,
      "loss": 3.1746,
      "step": 2492
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.1462761163711548,
      "learning_rate": 0.0005998266924498777,
      "loss": 2.7565,
      "step": 2493
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.797137975692749,
      "learning_rate": 0.0005998265534000551,
      "loss": 2.6766,
      "step": 2494
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7301762104034424,
      "learning_rate": 0.000599826414294489,
      "loss": 2.7666,
      "step": 2495
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.9763553738594055,
      "learning_rate": 0.0005998262751331796,
      "loss": 3.2243,
      "step": 2496
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.769555389881134,
      "learning_rate": 0.0005998261359161268,
      "loss": 2.6866,
      "step": 2497
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.6903071999549866,
      "learning_rate": 0.0005998259966433307,
      "loss": 2.8906,
      "step": 2498
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.8226003646850586,
      "learning_rate": 0.0005998258573147913,
      "loss": 2.863,
      "step": 2499
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.0303856134414673,
      "learning_rate": 0.0005998257179305085,
      "loss": 2.698,
      "step": 2500
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.8818358778953552,
      "learning_rate": 0.0005998255784904826,
      "loss": 3.123,
      "step": 2501
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.9277511835098267,
      "learning_rate": 0.0005998254389947134,
      "loss": 2.89,
      "step": 2502
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7738892436027527,
      "learning_rate": 0.0005998252994432009,
      "loss": 2.9473,
      "step": 2503
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.761547863483429,
      "learning_rate": 0.0005998251598359453,
      "loss": 2.7949,
      "step": 2504
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.9146711826324463,
      "learning_rate": 0.0005998250201729465,
      "loss": 2.8444,
      "step": 2505
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.9753386378288269,
      "learning_rate": 0.0005998248804542047,
      "loss": 2.9072,
      "step": 2506
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.2484840154647827,
      "learning_rate": 0.0005998247406797196,
      "loss": 2.7982,
      "step": 2507
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.2702466249465942,
      "learning_rate": 0.0005998246008494915,
      "loss": 2.6872,
      "step": 2508
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7815543413162231,
      "learning_rate": 0.0005998244609635204,
      "loss": 2.7899,
      "step": 2509
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.794396698474884,
      "learning_rate": 0.0005998243210218062,
      "loss": 2.8,
      "step": 2510
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.924694299697876,
      "learning_rate": 0.000599824181024349,
      "loss": 2.9732,
      "step": 2511
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7356405258178711,
      "learning_rate": 0.0005998240409711489,
      "loss": 2.7829,
      "step": 2512
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.8032949566841125,
      "learning_rate": 0.0005998239008622058,
      "loss": 2.9426,
      "step": 2513
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.9979055523872375,
      "learning_rate": 0.0005998237606975197,
      "loss": 2.7874,
      "step": 2514
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.705345869064331,
      "learning_rate": 0.0005998236204770909,
      "loss": 2.7982,
      "step": 2515
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.8300780057907104,
      "learning_rate": 0.0005998234802009192,
      "loss": 3.0651,
      "step": 2516
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.8098140358924866,
      "learning_rate": 0.0005998233398690046,
      "loss": 2.89,
      "step": 2517
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7996850609779358,
      "learning_rate": 0.000599823199481347,
      "loss": 2.878,
      "step": 2518
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.9656437039375305,
      "learning_rate": 0.0005998230590379468,
      "loss": 2.8214,
      "step": 2519
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7647343873977661,
      "learning_rate": 0.0005998229185388038,
      "loss": 2.8028,
      "step": 2520
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7871258854866028,
      "learning_rate": 0.000599822777983918,
      "loss": 2.8501,
      "step": 2521
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7593154907226562,
      "learning_rate": 0.0005998226373732896,
      "loss": 2.8614,
      "step": 2522
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.694839596748352,
      "learning_rate": 0.0005998224967069185,
      "loss": 2.7898,
      "step": 2523
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7574983835220337,
      "learning_rate": 0.0005998223559848046,
      "loss": 2.6753,
      "step": 2524
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7279633283615112,
      "learning_rate": 0.0005998222152069483,
      "loss": 2.9128,
      "step": 2525
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.6877351403236389,
      "learning_rate": 0.0005998220743733492,
      "loss": 3.0466,
      "step": 2526
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7943878173828125,
      "learning_rate": 0.0005998219334840075,
      "loss": 2.8144,
      "step": 2527
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.8055921792984009,
      "learning_rate": 0.0005998217925389233,
      "loss": 3.03,
      "step": 2528
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.8259342908859253,
      "learning_rate": 0.0005998216515380966,
      "loss": 2.7668,
      "step": 2529
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7526254057884216,
      "learning_rate": 0.0005998215104815274,
      "loss": 2.9896,
      "step": 2530
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.6936026811599731,
      "learning_rate": 0.0005998213693692157,
      "loss": 2.8551,
      "step": 2531
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.8586190342903137,
      "learning_rate": 0.0005998212282011616,
      "loss": 2.8512,
      "step": 2532
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7664892673492432,
      "learning_rate": 0.0005998210869773649,
      "loss": 2.9409,
      "step": 2533
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.012921929359436,
      "learning_rate": 0.000599820945697826,
      "loss": 2.5601,
      "step": 2534
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7752135396003723,
      "learning_rate": 0.0005998208043625446,
      "loss": 2.7689,
      "step": 2535
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.8374271988868713,
      "learning_rate": 0.0005998206629715209,
      "loss": 2.9424,
      "step": 2536
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.1213085651397705,
      "learning_rate": 0.0005998205215247548,
      "loss": 2.8752,
      "step": 2537
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7103328704833984,
      "learning_rate": 0.0005998203800222466,
      "loss": 2.8843,
      "step": 2538
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7676853537559509,
      "learning_rate": 0.000599820238463996,
      "loss": 2.6595,
      "step": 2539
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.8199318051338196,
      "learning_rate": 0.0005998200968500032,
      "loss": 3.0139,
      "step": 2540
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.9636824727058411,
      "learning_rate": 0.0005998199551802681,
      "loss": 2.5625,
      "step": 2541
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7159861326217651,
      "learning_rate": 0.0005998198134547909,
      "loss": 2.415,
      "step": 2542
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.0072135925292969,
      "learning_rate": 0.0005998196716735715,
      "loss": 2.8958,
      "step": 2543
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.6755673885345459,
      "learning_rate": 0.0005998195298366099,
      "loss": 2.7416,
      "step": 2544
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7443843483924866,
      "learning_rate": 0.0005998193879439063,
      "loss": 2.9711,
      "step": 2545
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.8065887689590454,
      "learning_rate": 0.0005998192459954606,
      "loss": 2.8837,
      "step": 2546
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7390572428703308,
      "learning_rate": 0.0005998191039912728,
      "loss": 3.0102,
      "step": 2547
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7030026912689209,
      "learning_rate": 0.000599818961931343,
      "loss": 3.0961,
      "step": 2548
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7028457522392273,
      "learning_rate": 0.0005998188198156712,
      "loss": 2.7762,
      "step": 2549
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.671823263168335,
      "learning_rate": 0.0005998186776442573,
      "loss": 2.8726,
      "step": 2550
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7720921635627747,
      "learning_rate": 0.0005998185354171016,
      "loss": 2.9186,
      "step": 2551
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.6787992119789124,
      "learning_rate": 0.0005998183931342039,
      "loss": 2.669,
      "step": 2552
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7943384051322937,
      "learning_rate": 0.0005998182507955644,
      "loss": 2.7022,
      "step": 2553
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7784193158149719,
      "learning_rate": 0.0005998181084011829,
      "loss": 2.6775,
      "step": 2554
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7195293307304382,
      "learning_rate": 0.0005998179659510596,
      "loss": 2.645,
      "step": 2555
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7851158380508423,
      "learning_rate": 0.0005998178234451945,
      "loss": 2.9951,
      "step": 2556
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7013583779335022,
      "learning_rate": 0.0005998176808835876,
      "loss": 2.7136,
      "step": 2557
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7285608649253845,
      "learning_rate": 0.000599817538266239,
      "loss": 2.9809,
      "step": 2558
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7133184671401978,
      "learning_rate": 0.0005998173955931485,
      "loss": 2.6472,
      "step": 2559
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7914353609085083,
      "learning_rate": 0.0005998172528643164,
      "loss": 2.9373,
      "step": 2560
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.6958377957344055,
      "learning_rate": 0.0005998171100797426,
      "loss": 2.7056,
      "step": 2561
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.8841256499290466,
      "learning_rate": 0.0005998169672394272,
      "loss": 3.0175,
      "step": 2562
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7264732122421265,
      "learning_rate": 0.00059981682434337,
      "loss": 2.7371,
      "step": 2563
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.6691034436225891,
      "learning_rate": 0.0005998166813915713,
      "loss": 2.8112,
      "step": 2564
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7442305088043213,
      "learning_rate": 0.000599816538384031,
      "loss": 2.791,
      "step": 2565
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7158634662628174,
      "learning_rate": 0.0005998163953207491,
      "loss": 3.0707,
      "step": 2566
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7570067048072815,
      "learning_rate": 0.0005998162522017257,
      "loss": 2.7751,
      "step": 2567
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7225087285041809,
      "learning_rate": 0.0005998161090269609,
      "loss": 2.966,
      "step": 2568
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7250968813896179,
      "learning_rate": 0.0005998159657964546,
      "loss": 2.8805,
      "step": 2569
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.676914632320404,
      "learning_rate": 0.0005998158225102068,
      "loss": 2.674,
      "step": 2570
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7515856623649597,
      "learning_rate": 0.0005998156791682176,
      "loss": 2.942,
      "step": 2571
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.791214644908905,
      "learning_rate": 0.000599815535770487,
      "loss": 2.8324,
      "step": 2572
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.6937572956085205,
      "learning_rate": 0.000599815392317015,
      "loss": 2.8514,
      "step": 2573
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.8064078092575073,
      "learning_rate": 0.0005998152488078016,
      "loss": 2.7123,
      "step": 2574
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.6538204550743103,
      "learning_rate": 0.0005998151052428471,
      "loss": 2.8985,
      "step": 2575
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7820507287979126,
      "learning_rate": 0.0005998149616221511,
      "loss": 3.2655,
      "step": 2576
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7423059344291687,
      "learning_rate": 0.0005998148179457139,
      "loss": 2.8699,
      "step": 2577
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.6710882186889648,
      "learning_rate": 0.0005998146742135354,
      "loss": 2.6281,
      "step": 2578
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.8356372117996216,
      "learning_rate": 0.0005998145304256159,
      "loss": 2.5614,
      "step": 2579
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.6984833478927612,
      "learning_rate": 0.0005998143865819551,
      "loss": 2.7565,
      "step": 2580
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.8613780736923218,
      "learning_rate": 0.0005998142426825531,
      "loss": 2.8473,
      "step": 2581
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7334119081497192,
      "learning_rate": 0.0005998140987274101,
      "loss": 2.7589,
      "step": 2582
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.821083128452301,
      "learning_rate": 0.0005998139547165259,
      "loss": 2.6836,
      "step": 2583
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7984685897827148,
      "learning_rate": 0.0005998138106499005,
      "loss": 2.7985,
      "step": 2584
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7941310405731201,
      "learning_rate": 0.0005998136665275343,
      "loss": 2.9889,
      "step": 2585
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7114253044128418,
      "learning_rate": 0.000599813522349427,
      "loss": 2.6362,
      "step": 2586
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7785787582397461,
      "learning_rate": 0.0005998133781155786,
      "loss": 2.8629,
      "step": 2587
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.9298737049102783,
      "learning_rate": 0.0005998132338259893,
      "loss": 2.8217,
      "step": 2588
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.6825291514396667,
      "learning_rate": 0.0005998130894806591,
      "loss": 3.0365,
      "step": 2589
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.0813308954238892,
      "learning_rate": 0.000599812945079588,
      "loss": 2.8531,
      "step": 2590
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.8429312705993652,
      "learning_rate": 0.000599812800622776,
      "loss": 2.8658,
      "step": 2591
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.925943911075592,
      "learning_rate": 0.0005998126561102231,
      "loss": 3.0319,
      "step": 2592
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.9029713869094849,
      "learning_rate": 0.0005998125115419294,
      "loss": 2.9713,
      "step": 2593
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.8437598943710327,
      "learning_rate": 0.0005998123669178949,
      "loss": 2.8196,
      "step": 2594
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.9147489666938782,
      "learning_rate": 0.0005998122222381195,
      "loss": 2.892,
      "step": 2595
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.204484224319458,
      "learning_rate": 0.0005998120775026035,
      "loss": 2.7469,
      "step": 2596
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7178389430046082,
      "learning_rate": 0.0005998119327113467,
      "loss": 2.9005,
      "step": 2597
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.8294429183006287,
      "learning_rate": 0.0005998117878643493,
      "loss": 3.0109,
      "step": 2598
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7204532623291016,
      "learning_rate": 0.0005998116429616112,
      "loss": 3.0671,
      "step": 2599
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.697908341884613,
      "learning_rate": 0.0005998114980031323,
      "loss": 2.8633,
      "step": 2600
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7171286940574646,
      "learning_rate": 0.000599811352988913,
      "loss": 2.8027,
      "step": 2601
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.793571412563324,
      "learning_rate": 0.0005998112079189529,
      "loss": 3.0115,
      "step": 2602
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7418248057365417,
      "learning_rate": 0.0005998110627932524,
      "loss": 3.0446,
      "step": 2603
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.6698408722877502,
      "learning_rate": 0.0005998109176118113,
      "loss": 2.7487,
      "step": 2604
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7247400283813477,
      "learning_rate": 0.0005998107723746297,
      "loss": 2.7328,
      "step": 2605
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.6916424632072449,
      "learning_rate": 0.0005998106270817076,
      "loss": 2.7693,
      "step": 2606
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7470510601997375,
      "learning_rate": 0.000599810481733045,
      "loss": 2.8555,
      "step": 2607
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7285078763961792,
      "learning_rate": 0.0005998103363286421,
      "loss": 2.7943,
      "step": 2608
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.6838415265083313,
      "learning_rate": 0.0005998101908684987,
      "loss": 2.7895,
      "step": 2609
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7484260201454163,
      "learning_rate": 0.000599810045352615,
      "loss": 2.8311,
      "step": 2610
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.8199912905693054,
      "learning_rate": 0.0005998098997809909,
      "loss": 2.6638,
      "step": 2611
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7727944254875183,
      "learning_rate": 0.0005998097541536265,
      "loss": 2.9283,
      "step": 2612
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7297310829162598,
      "learning_rate": 0.0005998096084705218,
      "loss": 2.7253,
      "step": 2613
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.693716287612915,
      "learning_rate": 0.0005998094627316768,
      "loss": 3.0306,
      "step": 2614
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.8051210641860962,
      "learning_rate": 0.0005998093169370917,
      "loss": 2.8704,
      "step": 2615
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7935771942138672,
      "learning_rate": 0.0005998091710867663,
      "loss": 2.7528,
      "step": 2616
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.6817690134048462,
      "learning_rate": 0.0005998090251807006,
      "loss": 2.782,
      "step": 2617
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.8940475583076477,
      "learning_rate": 0.0005998088792188948,
      "loss": 2.8605,
      "step": 2618
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.738336980342865,
      "learning_rate": 0.000599808733201349,
      "loss": 2.8237,
      "step": 2619
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.6680835485458374,
      "learning_rate": 0.000599808587128063,
      "loss": 2.9818,
      "step": 2620
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7090508937835693,
      "learning_rate": 0.0005998084409990369,
      "loss": 2.6895,
      "step": 2621
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7024458050727844,
      "learning_rate": 0.0005998082948142708,
      "loss": 2.8272,
      "step": 2622
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.8211132884025574,
      "learning_rate": 0.0005998081485737647,
      "loss": 2.826,
      "step": 2623
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.8033308982849121,
      "learning_rate": 0.0005998080022775185,
      "loss": 2.5396,
      "step": 2624
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7339184880256653,
      "learning_rate": 0.0005998078559255324,
      "loss": 3.0837,
      "step": 2625
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.8608270287513733,
      "learning_rate": 0.0005998077095178064,
      "loss": 2.5167,
      "step": 2626
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.8575740456581116,
      "learning_rate": 0.0005998075630543405,
      "loss": 2.5546,
      "step": 2627
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.8507735133171082,
      "learning_rate": 0.0005998074165351347,
      "loss": 2.9447,
      "step": 2628
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.9294947981834412,
      "learning_rate": 0.0005998072699601889,
      "loss": 2.7217,
      "step": 2629
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.8266012668609619,
      "learning_rate": 0.0005998071233295035,
      "loss": 2.77,
      "step": 2630
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.9155175089836121,
      "learning_rate": 0.0005998069766430781,
      "loss": 2.7826,
      "step": 2631
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.8601045608520508,
      "learning_rate": 0.000599806829900913,
      "loss": 2.785,
      "step": 2632
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7012798190116882,
      "learning_rate": 0.0005998066831030081,
      "loss": 2.7194,
      "step": 2633
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.8426799774169922,
      "learning_rate": 0.0005998065362493636,
      "loss": 2.8313,
      "step": 2634
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.843014657497406,
      "learning_rate": 0.0005998063893399793,
      "loss": 2.5466,
      "step": 2635
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7122505307197571,
      "learning_rate": 0.0005998062423748554,
      "loss": 2.8006,
      "step": 2636
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.8143070340156555,
      "learning_rate": 0.0005998060953539918,
      "loss": 2.8538,
      "step": 2637
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.9856811761856079,
      "learning_rate": 0.0005998059482773886,
      "loss": 2.7648,
      "step": 2638
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7210882306098938,
      "learning_rate": 0.0005998058011450458,
      "loss": 2.7075,
      "step": 2639
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.070870280265808,
      "learning_rate": 0.0005998056539569635,
      "loss": 2.709,
      "step": 2640
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.8215241432189941,
      "learning_rate": 0.0005998055067131416,
      "loss": 2.9588,
      "step": 2641
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7510101199150085,
      "learning_rate": 0.0005998053594135802,
      "loss": 2.9512,
      "step": 2642
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.8135839104652405,
      "learning_rate": 0.0005998052120582795,
      "loss": 2.7937,
      "step": 2643
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7121666669845581,
      "learning_rate": 0.0005998050646472392,
      "loss": 2.7485,
      "step": 2644
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7483109831809998,
      "learning_rate": 0.0005998049171804594,
      "loss": 2.8382,
      "step": 2645
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7011560201644897,
      "learning_rate": 0.0005998047696579404,
      "loss": 2.9489,
      "step": 2646
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7419089078903198,
      "learning_rate": 0.0005998046220796819,
      "loss": 2.6824,
      "step": 2647
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7876418232917786,
      "learning_rate": 0.000599804474445684,
      "loss": 2.846,
      "step": 2648
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.8912177681922913,
      "learning_rate": 0.000599804326755947,
      "loss": 2.8477,
      "step": 2649
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7691806554794312,
      "learning_rate": 0.0005998041790104706,
      "loss": 2.645,
      "step": 2650
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.8172603249549866,
      "learning_rate": 0.0005998040312092549,
      "loss": 2.6122,
      "step": 2651
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7368133068084717,
      "learning_rate": 0.0005998038833523,
      "loss": 3.0099,
      "step": 2652
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.9265463352203369,
      "learning_rate": 0.0005998037354396059,
      "loss": 2.7726,
      "step": 2653
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7140633463859558,
      "learning_rate": 0.0005998035874711726,
      "loss": 2.9322,
      "step": 2654
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7744355201721191,
      "learning_rate": 0.0005998034394470003,
      "loss": 2.5628,
      "step": 2655
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7690409421920776,
      "learning_rate": 0.0005998032913670888,
      "loss": 2.81,
      "step": 2656
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.6863447427749634,
      "learning_rate": 0.0005998031432314382,
      "loss": 2.6676,
      "step": 2657
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.9950952529907227,
      "learning_rate": 0.0005998029950400485,
      "loss": 2.4912,
      "step": 2658
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.879135251045227,
      "learning_rate": 0.0005998028467929197,
      "loss": 3.0935,
      "step": 2659
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7755398750305176,
      "learning_rate": 0.000599802698490052,
      "loss": 2.8371,
      "step": 2660
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.8864496350288391,
      "learning_rate": 0.0005998025501314454,
      "loss": 2.7442,
      "step": 2661
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7192245125770569,
      "learning_rate": 0.0005998024017170997,
      "loss": 2.7384,
      "step": 2662
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7673723697662354,
      "learning_rate": 0.000599802253247015,
      "loss": 2.7807,
      "step": 2663
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7586854100227356,
      "learning_rate": 0.0005998021047211915,
      "loss": 2.8592,
      "step": 2664
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7332044839859009,
      "learning_rate": 0.0005998019561396292,
      "loss": 2.6829,
      "step": 2665
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.6824513673782349,
      "learning_rate": 0.0005998018075023281,
      "loss": 2.6618,
      "step": 2666
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.829288125038147,
      "learning_rate": 0.000599801658809288,
      "loss": 2.6685,
      "step": 2667
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7700936198234558,
      "learning_rate": 0.0005998015100605093,
      "loss": 2.8566,
      "step": 2668
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7133655548095703,
      "learning_rate": 0.0005998013612559917,
      "loss": 2.6292,
      "step": 2669
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.6968019008636475,
      "learning_rate": 0.0005998012123957354,
      "loss": 3.0425,
      "step": 2670
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.807086169719696,
      "learning_rate": 0.0005998010634797404,
      "loss": 2.8047,
      "step": 2671
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7324214577674866,
      "learning_rate": 0.0005998009145080068,
      "loss": 2.8385,
      "step": 2672
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.793246865272522,
      "learning_rate": 0.0005998007654805344,
      "loss": 2.8674,
      "step": 2673
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.8439292311668396,
      "learning_rate": 0.0005998006163973235,
      "loss": 3.1472,
      "step": 2674
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.763645589351654,
      "learning_rate": 0.000599800467258374,
      "loss": 2.7222,
      "step": 2675
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7612413763999939,
      "learning_rate": 0.0005998003180636858,
      "loss": 2.7811,
      "step": 2676
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.816241443157196,
      "learning_rate": 0.0005998001688132592,
      "loss": 2.8692,
      "step": 2677
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.8205105662345886,
      "learning_rate": 0.000599800019507094,
      "loss": 2.9311,
      "step": 2678
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.704307496547699,
      "learning_rate": 0.0005997998701451904,
      "loss": 2.7067,
      "step": 2679
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.6915221214294434,
      "learning_rate": 0.0005997997207275483,
      "loss": 2.8654,
      "step": 2680
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.6778483986854553,
      "learning_rate": 0.0005997995712541678,
      "loss": 2.7243,
      "step": 2681
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.8015209436416626,
      "learning_rate": 0.0005997994217250489,
      "loss": 2.8105,
      "step": 2682
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.711263120174408,
      "learning_rate": 0.0005997992721401916,
      "loss": 2.6252,
      "step": 2683
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.8658844828605652,
      "learning_rate": 0.000599799122499596,
      "loss": 2.8113,
      "step": 2684
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.3431543111801147,
      "learning_rate": 0.000599798972803262,
      "loss": 2.9026,
      "step": 2685
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.9395774006843567,
      "learning_rate": 0.0005997988230511898,
      "loss": 2.8155,
      "step": 2686
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7397876381874084,
      "learning_rate": 0.0005997986732433793,
      "loss": 2.7285,
      "step": 2687
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7698303461074829,
      "learning_rate": 0.0005997985233798306,
      "loss": 2.7929,
      "step": 2688
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7513817548751831,
      "learning_rate": 0.0005997983734605437,
      "loss": 2.7306,
      "step": 2689
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7319083213806152,
      "learning_rate": 0.0005997982234855186,
      "loss": 2.7244,
      "step": 2690
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7519435286521912,
      "learning_rate": 0.0005997980734547553,
      "loss": 2.9414,
      "step": 2691
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.6662893295288086,
      "learning_rate": 0.0005997979233682539,
      "loss": 2.7787,
      "step": 2692
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7652427554130554,
      "learning_rate": 0.0005997977732260145,
      "loss": 2.7279,
      "step": 2693
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.731979250907898,
      "learning_rate": 0.0005997976230280369,
      "loss": 2.7798,
      "step": 2694
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7772365212440491,
      "learning_rate": 0.0005997974727743213,
      "loss": 2.9953,
      "step": 2695
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7707005739212036,
      "learning_rate": 0.0005997973224648677,
      "loss": 2.678,
      "step": 2696
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.6945422291755676,
      "learning_rate": 0.0005997971720996762,
      "loss": 2.6581,
      "step": 2697
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.6888855695724487,
      "learning_rate": 0.0005997970216787467,
      "loss": 3.043,
      "step": 2698
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.8083642721176147,
      "learning_rate": 0.0005997968712020792,
      "loss": 2.8927,
      "step": 2699
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7453312277793884,
      "learning_rate": 0.0005997967206696739,
      "loss": 2.9691,
      "step": 2700
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7189197540283203,
      "learning_rate": 0.0005997965700815307,
      "loss": 2.8826,
      "step": 2701
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7483528256416321,
      "learning_rate": 0.0005997964194376496,
      "loss": 3.1179,
      "step": 2702
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7866191864013672,
      "learning_rate": 0.0005997962687380307,
      "loss": 3.0269,
      "step": 2703
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.8325933814048767,
      "learning_rate": 0.0005997961179826741,
      "loss": 2.8742,
      "step": 2704
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.764877438545227,
      "learning_rate": 0.0005997959671715797,
      "loss": 2.715,
      "step": 2705
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.2859296798706055,
      "learning_rate": 0.0005997958163047476,
      "loss": 2.6464,
      "step": 2706
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7390991449356079,
      "learning_rate": 0.0005997956653821777,
      "loss": 2.7237,
      "step": 2707
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7773690819740295,
      "learning_rate": 0.0005997955144038703,
      "loss": 2.9809,
      "step": 2708
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.8928781151771545,
      "learning_rate": 0.0005997953633698251,
      "loss": 2.8374,
      "step": 2709
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7422425150871277,
      "learning_rate": 0.0005997952122800423,
      "loss": 2.7072,
      "step": 2710
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.8691006898880005,
      "learning_rate": 0.0005997950611345219,
      "loss": 2.7435,
      "step": 2711
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.6949936747550964,
      "learning_rate": 0.000599794909933264,
      "loss": 2.9001,
      "step": 2712
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7238205075263977,
      "learning_rate": 0.0005997947586762686,
      "loss": 2.6914,
      "step": 2713
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7895284295082092,
      "learning_rate": 0.0005997946073635356,
      "loss": 2.9294,
      "step": 2714
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.6887364983558655,
      "learning_rate": 0.0005997944559950651,
      "loss": 2.7648,
      "step": 2715
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7494404315948486,
      "learning_rate": 0.0005997943045708573,
      "loss": 2.9884,
      "step": 2716
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.6676808595657349,
      "learning_rate": 0.000599794153090912,
      "loss": 2.9412,
      "step": 2717
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.8495573401451111,
      "learning_rate": 0.0005997940015552292,
      "loss": 2.7661,
      "step": 2718
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7258166670799255,
      "learning_rate": 0.0005997938499638092,
      "loss": 3.0227,
      "step": 2719
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7321751117706299,
      "learning_rate": 0.0005997936983166517,
      "loss": 2.7042,
      "step": 2720
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.753523051738739,
      "learning_rate": 0.0005997935466137571,
      "loss": 2.927,
      "step": 2721
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7093325257301331,
      "learning_rate": 0.0005997933948551251,
      "loss": 2.9312,
      "step": 2722
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7303743958473206,
      "learning_rate": 0.0005997932430407559,
      "loss": 2.8557,
      "step": 2723
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.8139822483062744,
      "learning_rate": 0.0005997930911706493,
      "loss": 2.8524,
      "step": 2724
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.8090891242027283,
      "learning_rate": 0.0005997929392448057,
      "loss": 2.9179,
      "step": 2725
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7787619233131409,
      "learning_rate": 0.0005997927872632249,
      "loss": 2.8448,
      "step": 2726
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7428972721099854,
      "learning_rate": 0.000599792635225907,
      "loss": 2.8553,
      "step": 2727
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.6966132521629333,
      "learning_rate": 0.0005997924831328519,
      "loss": 2.4991,
      "step": 2728
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.804976224899292,
      "learning_rate": 0.0005997923309840597,
      "loss": 2.6621,
      "step": 2729
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7229626178741455,
      "learning_rate": 0.0005997921787795306,
      "loss": 2.6775,
      "step": 2730
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.8704573512077332,
      "learning_rate": 0.0005997920265192642,
      "loss": 2.8249,
      "step": 2731
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7549136281013489,
      "learning_rate": 0.0005997918742032611,
      "loss": 2.7208,
      "step": 2732
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7155278325080872,
      "learning_rate": 0.0005997917218315209,
      "loss": 2.8357,
      "step": 2733
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.8192238807678223,
      "learning_rate": 0.0005997915694040438,
      "loss": 2.6197,
      "step": 2734
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7616353034973145,
      "learning_rate": 0.0005997914169208297,
      "loss": 2.7404,
      "step": 2735
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7620733976364136,
      "learning_rate": 0.0005997912643818788,
      "loss": 2.7917,
      "step": 2736
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.8233075141906738,
      "learning_rate": 0.0005997911117871911,
      "loss": 2.8558,
      "step": 2737
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7469291090965271,
      "learning_rate": 0.0005997909591367664,
      "loss": 2.6594,
      "step": 2738
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7477818131446838,
      "learning_rate": 0.000599790806430605,
      "loss": 2.6257,
      "step": 2739
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7046999335289001,
      "learning_rate": 0.0005997906536687068,
      "loss": 2.6334,
      "step": 2740
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.9025896191596985,
      "learning_rate": 0.0005997905008510719,
      "loss": 2.8545,
      "step": 2741
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.714433491230011,
      "learning_rate": 0.0005997903479777003,
      "loss": 2.5892,
      "step": 2742
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.8834080100059509,
      "learning_rate": 0.000599790195048592,
      "loss": 2.7061,
      "step": 2743
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.8109122514724731,
      "learning_rate": 0.0005997900420637469,
      "loss": 2.8599,
      "step": 2744
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.9522578120231628,
      "learning_rate": 0.0005997898890231654,
      "loss": 2.6268,
      "step": 2745
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.9097983241081238,
      "learning_rate": 0.0005997897359268471,
      "loss": 2.9491,
      "step": 2746
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7552648186683655,
      "learning_rate": 0.0005997895827747924,
      "loss": 3.0456,
      "step": 2747
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.9992645382881165,
      "learning_rate": 0.0005997894295670011,
      "loss": 3.1123,
      "step": 2748
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7972840666770935,
      "learning_rate": 0.0005997892763034733,
      "loss": 2.8975,
      "step": 2749
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.6844323873519897,
      "learning_rate": 0.000599789122984209,
      "loss": 2.7812,
      "step": 2750
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.741947591304779,
      "learning_rate": 0.0005997889696092081,
      "loss": 3.0133,
      "step": 2751
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7747599482536316,
      "learning_rate": 0.0005997888161784709,
      "loss": 2.7815,
      "step": 2752
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.6429771184921265,
      "learning_rate": 0.0005997886626919974,
      "loss": 2.7681,
      "step": 2753
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.8568233251571655,
      "learning_rate": 0.0005997885091497874,
      "loss": 2.6877,
      "step": 2754
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.8108782172203064,
      "learning_rate": 0.0005997883555518411,
      "loss": 2.8133,
      "step": 2755
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7507073879241943,
      "learning_rate": 0.0005997882018981585,
      "loss": 3.0339,
      "step": 2756
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.0547492504119873,
      "learning_rate": 0.0005997880481887395,
      "loss": 2.6928,
      "step": 2757
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.8395648002624512,
      "learning_rate": 0.0005997878944235843,
      "loss": 2.7811,
      "step": 2758
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7434489130973816,
      "learning_rate": 0.000599787740602693,
      "loss": 2.7197,
      "step": 2759
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.8179664015769958,
      "learning_rate": 0.0005997875867260654,
      "loss": 2.7897,
      "step": 2760
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7615586519241333,
      "learning_rate": 0.0005997874327937015,
      "loss": 2.7629,
      "step": 2761
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7657621502876282,
      "learning_rate": 0.0005997872788056017,
      "loss": 3.0623,
      "step": 2762
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.9320579767227173,
      "learning_rate": 0.0005997871247617657,
      "loss": 2.8662,
      "step": 2763
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7350435853004456,
      "learning_rate": 0.0005997869706621936,
      "loss": 2.965,
      "step": 2764
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7999938726425171,
      "learning_rate": 0.0005997868165068854,
      "loss": 3.073,
      "step": 2765
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7170044779777527,
      "learning_rate": 0.0005997866622958412,
      "loss": 2.9161,
      "step": 2766
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.852697491645813,
      "learning_rate": 0.0005997865080290609,
      "loss": 2.8222,
      "step": 2767
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.6693736910820007,
      "learning_rate": 0.0005997863537065448,
      "loss": 2.8987,
      "step": 2768
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.79563969373703,
      "learning_rate": 0.0005997861993282926,
      "loss": 2.9592,
      "step": 2769
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.114333152770996,
      "learning_rate": 0.0005997860448943047,
      "loss": 2.6445,
      "step": 2770
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.9375038146972656,
      "learning_rate": 0.0005997858904045807,
      "loss": 2.6635,
      "step": 2771
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7535377740859985,
      "learning_rate": 0.000599785735859121,
      "loss": 2.7013,
      "step": 2772
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7473438382148743,
      "learning_rate": 0.0005997855812579254,
      "loss": 2.7132,
      "step": 2773
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7300716638565063,
      "learning_rate": 0.000599785426600994,
      "loss": 2.7266,
      "step": 2774
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.0620193481445312,
      "learning_rate": 0.0005997852718883269,
      "loss": 2.7528,
      "step": 2775
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7526024580001831,
      "learning_rate": 0.0005997851171199239,
      "loss": 2.8993,
      "step": 2776
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.1160154342651367,
      "learning_rate": 0.0005997849622957853,
      "loss": 2.7475,
      "step": 2777
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.6986150741577148,
      "learning_rate": 0.0005997848074159111,
      "loss": 2.8274,
      "step": 2778
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.797435998916626,
      "learning_rate": 0.0005997846524803011,
      "loss": 2.8524,
      "step": 2779
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.8772227764129639,
      "learning_rate": 0.0005997844974889555,
      "loss": 2.8686,
      "step": 2780
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.865726113319397,
      "learning_rate": 0.0005997843424418744,
      "loss": 2.6064,
      "step": 2781
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7975723147392273,
      "learning_rate": 0.0005997841873390576,
      "loss": 2.8836,
      "step": 2782
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.8222036361694336,
      "learning_rate": 0.0005997840321805053,
      "loss": 2.9074,
      "step": 2783
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.8784765005111694,
      "learning_rate": 0.0005997838769662175,
      "loss": 2.8506,
      "step": 2784
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7536912560462952,
      "learning_rate": 0.0005997837216961942,
      "loss": 2.7487,
      "step": 2785
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.309242606163025,
      "learning_rate": 0.0005997835663704356,
      "loss": 2.6934,
      "step": 2786
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.8480592966079712,
      "learning_rate": 0.0005997834109889414,
      "loss": 3.0108,
      "step": 2787
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7146170139312744,
      "learning_rate": 0.0005997832555517118,
      "loss": 2.8521,
      "step": 2788
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.042842149734497,
      "learning_rate": 0.000599783100058747,
      "loss": 2.6651,
      "step": 2789
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.6483970880508423,
      "learning_rate": 0.0005997829445100467,
      "loss": 2.9123,
      "step": 2790
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.6439054012298584,
      "learning_rate": 0.0005997827889056112,
      "loss": 2.7573,
      "step": 2791
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.8896760940551758,
      "learning_rate": 0.0005997826332454403,
      "loss": 2.8141,
      "step": 2792
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.70045405626297,
      "learning_rate": 0.0005997824775295342,
      "loss": 2.8352,
      "step": 2793
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7148938775062561,
      "learning_rate": 0.0005997823217578929,
      "loss": 2.9448,
      "step": 2794
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.8313169479370117,
      "learning_rate": 0.0005997821659305164,
      "loss": 2.8243,
      "step": 2795
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7736448049545288,
      "learning_rate": 0.0005997820100474047,
      "loss": 2.6863,
      "step": 2796
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7188523411750793,
      "learning_rate": 0.000599781854108558,
      "loss": 2.7539,
      "step": 2797
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.6716729402542114,
      "learning_rate": 0.0005997816981139761,
      "loss": 2.9357,
      "step": 2798
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.804131031036377,
      "learning_rate": 0.0005997815420636591,
      "loss": 2.8789,
      "step": 2799
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7251685857772827,
      "learning_rate": 0.0005997813859576071,
      "loss": 2.6798,
      "step": 2800
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.8056030869483948,
      "learning_rate": 0.00059978122979582,
      "loss": 2.7943,
      "step": 2801
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7683626413345337,
      "learning_rate": 0.0005997810735782979,
      "loss": 2.6886,
      "step": 2802
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7304604053497314,
      "learning_rate": 0.000599780917305041,
      "loss": 2.9292,
      "step": 2803
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.9024092555046082,
      "learning_rate": 0.0005997807609760491,
      "loss": 2.5445,
      "step": 2804
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7505795955657959,
      "learning_rate": 0.0005997806045913223,
      "loss": 2.6681,
      "step": 2805
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.8200767636299133,
      "learning_rate": 0.0005997804481508606,
      "loss": 2.979,
      "step": 2806
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.9216470122337341,
      "learning_rate": 0.000599780291654664,
      "loss": 2.9995,
      "step": 2807
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7660918235778809,
      "learning_rate": 0.0005997801351027326,
      "loss": 2.8093,
      "step": 2808
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.379623293876648,
      "learning_rate": 0.0005997799784950665,
      "loss": 2.7682,
      "step": 2809
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7583315968513489,
      "learning_rate": 0.0005997798218316656,
      "loss": 2.8322,
      "step": 2810
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7635440826416016,
      "learning_rate": 0.00059977966511253,
      "loss": 2.9144,
      "step": 2811
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.762643039226532,
      "learning_rate": 0.0005997795083376596,
      "loss": 2.9826,
      "step": 2812
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.8870160579681396,
      "learning_rate": 0.0005997793515070546,
      "loss": 2.7154,
      "step": 2813
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7505325675010681,
      "learning_rate": 0.000599779194620715,
      "loss": 2.8669,
      "step": 2814
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.8708193898200989,
      "learning_rate": 0.0005997790376786407,
      "loss": 2.563,
      "step": 2815
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.6989684104919434,
      "learning_rate": 0.0005997788806808319,
      "loss": 2.8976,
      "step": 2816
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.823433518409729,
      "learning_rate": 0.0005997787236272885,
      "loss": 2.7819,
      "step": 2817
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.8036072254180908,
      "learning_rate": 0.0005997785665180106,
      "loss": 2.7122,
      "step": 2818
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7364576458930969,
      "learning_rate": 0.0005997784093529981,
      "loss": 2.5716,
      "step": 2819
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.8436229228973389,
      "learning_rate": 0.0005997782521322513,
      "loss": 2.8986,
      "step": 2820
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.8894460797309875,
      "learning_rate": 0.00059977809485577,
      "loss": 2.7149,
      "step": 2821
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7289884686470032,
      "learning_rate": 0.0005997779375235544,
      "loss": 2.6898,
      "step": 2822
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.9315985441207886,
      "learning_rate": 0.0005997777801356042,
      "loss": 2.5061,
      "step": 2823
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.8481593132019043,
      "learning_rate": 0.0005997776226919199,
      "loss": 2.4675,
      "step": 2824
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.8053861856460571,
      "learning_rate": 0.000599777465192501,
      "loss": 2.7757,
      "step": 2825
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.126994013786316,
      "learning_rate": 0.000599777307637348,
      "loss": 2.775,
      "step": 2826
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7336538434028625,
      "learning_rate": 0.0005997771500264607,
      "loss": 2.9971,
      "step": 2827
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7242480516433716,
      "learning_rate": 0.0005997769923598392,
      "loss": 2.8801,
      "step": 2828
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.8687599897384644,
      "learning_rate": 0.0005997768346374835,
      "loss": 2.7205,
      "step": 2829
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.8770078420639038,
      "learning_rate": 0.0005997766768593935,
      "loss": 2.9541,
      "step": 2830
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.8420789837837219,
      "learning_rate": 0.0005997765190255695,
      "loss": 2.8101,
      "step": 2831
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.016530990600586,
      "learning_rate": 0.0005997763611360113,
      "loss": 2.9031,
      "step": 2832
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7079837918281555,
      "learning_rate": 0.000599776203190719,
      "loss": 2.6465,
      "step": 2833
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.2572073936462402,
      "learning_rate": 0.0005997760451896927,
      "loss": 2.8542,
      "step": 2834
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.0144574642181396,
      "learning_rate": 0.0005997758871329324,
      "loss": 2.8736,
      "step": 2835
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7382087707519531,
      "learning_rate": 0.0005997757290204381,
      "loss": 2.7413,
      "step": 2836
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7561126947402954,
      "learning_rate": 0.0005997755708522099,
      "loss": 2.9622,
      "step": 2837
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7881171107292175,
      "learning_rate": 0.0005997754126282475,
      "loss": 2.6557,
      "step": 2838
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.6987342238426208,
      "learning_rate": 0.0005997752543485514,
      "loss": 2.7494,
      "step": 2839
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.804629385471344,
      "learning_rate": 0.0005997750960131214,
      "loss": 2.8416,
      "step": 2840
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.8081315159797668,
      "learning_rate": 0.0005997749376219576,
      "loss": 2.9564,
      "step": 2841
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7489773035049438,
      "learning_rate": 0.0005997747791750598,
      "loss": 2.8659,
      "step": 2842
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.8331895470619202,
      "learning_rate": 0.0005997746206724284,
      "loss": 2.9289,
      "step": 2843
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7790747880935669,
      "learning_rate": 0.0005997744621140631,
      "loss": 2.8742,
      "step": 2844
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.709983229637146,
      "learning_rate": 0.0005997743034999642,
      "loss": 2.7384,
      "step": 2845
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.8379331827163696,
      "learning_rate": 0.0005997741448301316,
      "loss": 2.8014,
      "step": 2846
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.8247488737106323,
      "learning_rate": 0.0005997739861045651,
      "loss": 2.8896,
      "step": 2847
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7091938257217407,
      "learning_rate": 0.0005997738273232652,
      "loss": 2.8408,
      "step": 2848
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7644179463386536,
      "learning_rate": 0.0005997736684862316,
      "loss": 2.6062,
      "step": 2849
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7611038684844971,
      "learning_rate": 0.0005997735095934644,
      "loss": 2.9506,
      "step": 2850
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.8224937319755554,
      "learning_rate": 0.0005997733506449636,
      "loss": 2.9426,
      "step": 2851
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7585546374320984,
      "learning_rate": 0.0005997731916407295,
      "loss": 2.8227,
      "step": 2852
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7518122792243958,
      "learning_rate": 0.0005997730325807617,
      "loss": 2.6928,
      "step": 2853
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.730835497379303,
      "learning_rate": 0.0005997728734650604,
      "loss": 2.4314,
      "step": 2854
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7472057938575745,
      "learning_rate": 0.0005997727142936258,
      "loss": 2.9855,
      "step": 2855
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.959617555141449,
      "learning_rate": 0.0005997725550664578,
      "loss": 2.9201,
      "step": 2856
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.848347544670105,
      "learning_rate": 0.0005997723957835564,
      "loss": 2.6284,
      "step": 2857
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7768365740776062,
      "learning_rate": 0.0005997722364449215,
      "loss": 3.0654,
      "step": 2858
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.777949333190918,
      "learning_rate": 0.0005997720770505535,
      "loss": 2.9101,
      "step": 2859
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.833117663860321,
      "learning_rate": 0.000599771917600452,
      "loss": 2.6658,
      "step": 2860
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7646105289459229,
      "learning_rate": 0.0005997717580946174,
      "loss": 2.9997,
      "step": 2861
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7480989098548889,
      "learning_rate": 0.0005997715985330495,
      "loss": 2.74,
      "step": 2862
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.8097661137580872,
      "learning_rate": 0.0005997714389157485,
      "loss": 3.0211,
      "step": 2863
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.8420969247817993,
      "learning_rate": 0.0005997712792427142,
      "loss": 3.0486,
      "step": 2864
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7355155944824219,
      "learning_rate": 0.0005997711195139468,
      "loss": 2.6066,
      "step": 2865
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.8832474946975708,
      "learning_rate": 0.0005997709597294463,
      "loss": 2.783,
      "step": 2866
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.9762870073318481,
      "learning_rate": 0.0005997707998892128,
      "loss": 2.7529,
      "step": 2867
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7912784814834595,
      "learning_rate": 0.0005997706399932461,
      "loss": 2.9017,
      "step": 2868
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.99387127161026,
      "learning_rate": 0.0005997704800415465,
      "loss": 2.744,
      "step": 2869
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.3345354795455933,
      "learning_rate": 0.0005997703200341138,
      "loss": 2.8833,
      "step": 2870
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7443711757659912,
      "learning_rate": 0.0005997701599709483,
      "loss": 2.9575,
      "step": 2871
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.198226809501648,
      "learning_rate": 0.0005997699998520497,
      "loss": 2.8806,
      "step": 2872
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.8825127482414246,
      "learning_rate": 0.0005997698396774183,
      "loss": 3.0017,
      "step": 2873
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.6975281834602356,
      "learning_rate": 0.000599769679447054,
      "loss": 2.5258,
      "step": 2874
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.8729137778282166,
      "learning_rate": 0.0005997695191609568,
      "loss": 2.7153,
      "step": 2875
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.8058415055274963,
      "learning_rate": 0.0005997693588191269,
      "loss": 2.7894,
      "step": 2876
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.681928813457489,
      "learning_rate": 0.0005997691984215641,
      "loss": 2.9206,
      "step": 2877
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7746449112892151,
      "learning_rate": 0.0005997690379682687,
      "loss": 2.8604,
      "step": 2878
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.9139981269836426,
      "learning_rate": 0.0005997688774592404,
      "loss": 2.7863,
      "step": 2879
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7785935997962952,
      "learning_rate": 0.0005997687168944795,
      "loss": 2.7659,
      "step": 2880
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.8358689546585083,
      "learning_rate": 0.000599768556273986,
      "loss": 2.7298,
      "step": 2881
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.8695845007896423,
      "learning_rate": 0.0005997683955977598,
      "loss": 2.6032,
      "step": 2882
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7006316184997559,
      "learning_rate": 0.0005997682348658009,
      "loss": 2.7557,
      "step": 2883
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.112969160079956,
      "learning_rate": 0.0005997680740781095,
      "loss": 2.8574,
      "step": 2884
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7908394932746887,
      "learning_rate": 0.0005997679132346856,
      "loss": 2.9495,
      "step": 2885
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7219509482383728,
      "learning_rate": 0.0005997677523355292,
      "loss": 2.8185,
      "step": 2886
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.2903412580490112,
      "learning_rate": 0.0005997675913806404,
      "loss": 2.9679,
      "step": 2887
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.6681073307991028,
      "learning_rate": 0.000599767430370019,
      "loss": 2.8007,
      "step": 2888
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.8855907917022705,
      "learning_rate": 0.0005997672693036652,
      "loss": 2.8727,
      "step": 2889
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.788709282875061,
      "learning_rate": 0.0005997671081815789,
      "loss": 2.8718,
      "step": 2890
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7306891083717346,
      "learning_rate": 0.0005997669470037604,
      "loss": 3.0622,
      "step": 2891
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7225790619850159,
      "learning_rate": 0.0005997667857702095,
      "loss": 2.7248,
      "step": 2892
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7511290907859802,
      "learning_rate": 0.0005997666244809263,
      "loss": 2.7913,
      "step": 2893
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.8423399329185486,
      "learning_rate": 0.0005997664631359109,
      "loss": 2.7232,
      "step": 2894
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.6467767357826233,
      "learning_rate": 0.0005997663017351632,
      "loss": 2.6476,
      "step": 2895
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7853938937187195,
      "learning_rate": 0.0005997661402786833,
      "loss": 2.9254,
      "step": 2896
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.832843005657196,
      "learning_rate": 0.0005997659787664711,
      "loss": 2.7906,
      "step": 2897
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7121613621711731,
      "learning_rate": 0.0005997658171985269,
      "loss": 2.7242,
      "step": 2898
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7887724041938782,
      "learning_rate": 0.0005997656555748504,
      "loss": 2.7225,
      "step": 2899
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.6532392501831055,
      "learning_rate": 0.000599765493895442,
      "loss": 2.7136,
      "step": 2900
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.6847103834152222,
      "learning_rate": 0.0005997653321603014,
      "loss": 2.9093,
      "step": 2901
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.738844096660614,
      "learning_rate": 0.0005997651703694288,
      "loss": 2.7955,
      "step": 2902
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.734303891658783,
      "learning_rate": 0.0005997650085228243,
      "loss": 2.7006,
      "step": 2903
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7151882648468018,
      "learning_rate": 0.0005997648466204877,
      "loss": 2.7883,
      "step": 2904
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7901836037635803,
      "learning_rate": 0.0005997646846624191,
      "loss": 2.8595,
      "step": 2905
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7337841391563416,
      "learning_rate": 0.0005997645226486187,
      "loss": 2.7717,
      "step": 2906
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7333092093467712,
      "learning_rate": 0.0005997643605790865,
      "loss": 2.7301,
      "step": 2907
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7214515805244446,
      "learning_rate": 0.0005997641984538223,
      "loss": 2.7656,
      "step": 2908
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.8477077484130859,
      "learning_rate": 0.0005997640362728263,
      "loss": 2.9814,
      "step": 2909
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7161383032798767,
      "learning_rate": 0.0005997638740360986,
      "loss": 2.7136,
      "step": 2910
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.74419766664505,
      "learning_rate": 0.0005997637117436391,
      "loss": 2.795,
      "step": 2911
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7777158617973328,
      "learning_rate": 0.0005997635493954478,
      "loss": 2.91,
      "step": 2912
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7147798538208008,
      "learning_rate": 0.0005997633869915248,
      "loss": 2.7915,
      "step": 2913
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7776731848716736,
      "learning_rate": 0.0005997632245318702,
      "loss": 2.7015,
      "step": 2914
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7750610113143921,
      "learning_rate": 0.000599763062016484,
      "loss": 3.294,
      "step": 2915
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7533096075057983,
      "learning_rate": 0.0005997628994453661,
      "loss": 2.6293,
      "step": 2916
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7636644840240479,
      "learning_rate": 0.0005997627368185166,
      "loss": 2.8288,
      "step": 2917
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7568842768669128,
      "learning_rate": 0.0005997625741359356,
      "loss": 2.6825,
      "step": 2918
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.679583728313446,
      "learning_rate": 0.000599762411397623,
      "loss": 2.8117,
      "step": 2919
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7014347910881042,
      "learning_rate": 0.000599762248603579,
      "loss": 2.7564,
      "step": 2920
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.6980209350585938,
      "learning_rate": 0.0005997620857538035,
      "loss": 2.8094,
      "step": 2921
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7643210291862488,
      "learning_rate": 0.0005997619228482966,
      "loss": 3.1411,
      "step": 2922
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7291616797447205,
      "learning_rate": 0.0005997617598870582,
      "loss": 2.703,
      "step": 2923
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.8704910278320312,
      "learning_rate": 0.0005997615968700885,
      "loss": 2.7015,
      "step": 2924
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.8312593102455139,
      "learning_rate": 0.0005997614337973875,
      "loss": 2.8361,
      "step": 2925
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7254495024681091,
      "learning_rate": 0.0005997612706689552,
      "loss": 2.8284,
      "step": 2926
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.0827161073684692,
      "learning_rate": 0.0005997611074847916,
      "loss": 2.8215,
      "step": 2927
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7224822044372559,
      "learning_rate": 0.0005997609442448968,
      "loss": 2.9097,
      "step": 2928
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7783997058868408,
      "learning_rate": 0.0005997607809492706,
      "loss": 2.8476,
      "step": 2929
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7977182269096375,
      "learning_rate": 0.0005997606175979133,
      "loss": 2.7653,
      "step": 2930
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7025641202926636,
      "learning_rate": 0.0005997604541908249,
      "loss": 2.9051,
      "step": 2931
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7147460579872131,
      "learning_rate": 0.0005997602907280053,
      "loss": 2.8334,
      "step": 2932
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.8001498579978943,
      "learning_rate": 0.0005997601272094546,
      "loss": 2.6555,
      "step": 2933
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.796575129032135,
      "learning_rate": 0.0005997599636351729,
      "loss": 2.7116,
      "step": 2934
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.6955004930496216,
      "learning_rate": 0.0005997598000051603,
      "loss": 2.7337,
      "step": 2935
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.6469569802284241,
      "learning_rate": 0.0005997596363194165,
      "loss": 2.8223,
      "step": 2936
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.9017661213874817,
      "learning_rate": 0.0005997594725779417,
      "loss": 2.6901,
      "step": 2937
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.6547524929046631,
      "learning_rate": 0.0005997593087807361,
      "loss": 2.7711,
      "step": 2938
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.8353975415229797,
      "learning_rate": 0.0005997591449277994,
      "loss": 3.0426,
      "step": 2939
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7323978543281555,
      "learning_rate": 0.000599758981019132,
      "loss": 2.8298,
      "step": 2940
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.6867600083351135,
      "learning_rate": 0.0005997588170547336,
      "loss": 3.1502,
      "step": 2941
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7839487195014954,
      "learning_rate": 0.0005997586530346045,
      "loss": 2.9631,
      "step": 2942
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7326334118843079,
      "learning_rate": 0.0005997584889587445,
      "loss": 2.5177,
      "step": 2943
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.6810262799263,
      "learning_rate": 0.0005997583248271538,
      "loss": 2.8087,
      "step": 2944
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7134823203086853,
      "learning_rate": 0.0005997581606398324,
      "loss": 3.0778,
      "step": 2945
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.6863507628440857,
      "learning_rate": 0.0005997579963967802,
      "loss": 2.8678,
      "step": 2946
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7127280831336975,
      "learning_rate": 0.0005997578320979974,
      "loss": 2.4148,
      "step": 2947
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.8096237182617188,
      "learning_rate": 0.000599757667743484,
      "loss": 3.034,
      "step": 2948
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.717647910118103,
      "learning_rate": 0.00059975750333324,
      "loss": 2.6611,
      "step": 2949
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.6840240955352783,
      "learning_rate": 0.0005997573388672652,
      "loss": 2.9546,
      "step": 2950
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7405534386634827,
      "learning_rate": 0.0005997571743455601,
      "loss": 2.6777,
      "step": 2951
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7669951319694519,
      "learning_rate": 0.0005997570097681244,
      "loss": 2.8739,
      "step": 2952
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7546408772468567,
      "learning_rate": 0.0005997568451349582,
      "loss": 2.8087,
      "step": 2953
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7826672792434692,
      "learning_rate": 0.0005997566804460616,
      "loss": 2.9747,
      "step": 2954
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.747398316860199,
      "learning_rate": 0.0005997565157014346,
      "loss": 2.8544,
      "step": 2955
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7027176022529602,
      "learning_rate": 0.0005997563509010771,
      "loss": 2.7119,
      "step": 2956
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.6656762361526489,
      "learning_rate": 0.0005997561860449892,
      "loss": 2.6601,
      "step": 2957
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.8083311915397644,
      "learning_rate": 0.0005997560211331711,
      "loss": 2.9023,
      "step": 2958
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7383133769035339,
      "learning_rate": 0.0005997558561656227,
      "loss": 2.7683,
      "step": 2959
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7879669666290283,
      "learning_rate": 0.0005997556911423439,
      "loss": 2.7961,
      "step": 2960
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.9046640992164612,
      "learning_rate": 0.0005997555260633349,
      "loss": 2.8471,
      "step": 2961
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.8371109366416931,
      "learning_rate": 0.0005997553609285957,
      "loss": 2.7202,
      "step": 2962
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.8393042087554932,
      "learning_rate": 0.0005997551957381263,
      "loss": 2.9643,
      "step": 2963
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.751058042049408,
      "learning_rate": 0.0005997550304919269,
      "loss": 2.7773,
      "step": 2964
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.6958792805671692,
      "learning_rate": 0.0005997548651899972,
      "loss": 2.7435,
      "step": 2965
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7204942107200623,
      "learning_rate": 0.0005997546998323375,
      "loss": 2.5774,
      "step": 2966
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7988232970237732,
      "learning_rate": 0.0005997545344189477,
      "loss": 2.6795,
      "step": 2967
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7433374524116516,
      "learning_rate": 0.0005997543689498279,
      "loss": 3.0062,
      "step": 2968
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7145172953605652,
      "learning_rate": 0.0005997542034249781,
      "loss": 2.772,
      "step": 2969
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.8049513697624207,
      "learning_rate": 0.0005997540378443984,
      "loss": 2.8791,
      "step": 2970
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.6659042239189148,
      "learning_rate": 0.0005997538722080887,
      "loss": 2.6111,
      "step": 2971
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7261955738067627,
      "learning_rate": 0.0005997537065160491,
      "loss": 2.9031,
      "step": 2972
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.8940856456756592,
      "learning_rate": 0.0005997535407682796,
      "loss": 2.8187,
      "step": 2973
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.6973537802696228,
      "learning_rate": 0.0005997533749647803,
      "loss": 2.9417,
      "step": 2974
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.1709340810775757,
      "learning_rate": 0.0005997532091055512,
      "loss": 2.6181,
      "step": 2975
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7018151879310608,
      "learning_rate": 0.0005997530431905922,
      "loss": 2.8366,
      "step": 2976
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.8267925977706909,
      "learning_rate": 0.0005997528772199035,
      "loss": 2.9189,
      "step": 2977
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.9229220151901245,
      "learning_rate": 0.0005997527111934852,
      "loss": 2.7775,
      "step": 2978
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7953844666481018,
      "learning_rate": 0.0005997525451113372,
      "loss": 2.8777,
      "step": 2979
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.0234260559082031,
      "learning_rate": 0.0005997523789734594,
      "loss": 2.736,
      "step": 2980
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7743806838989258,
      "learning_rate": 0.0005997522127798522,
      "loss": 2.6901,
      "step": 2981
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.6859956383705139,
      "learning_rate": 0.0005997520465305153,
      "loss": 2.6368,
      "step": 2982
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.6684409976005554,
      "learning_rate": 0.0005997518802254487,
      "loss": 2.7137,
      "step": 2983
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7182648777961731,
      "learning_rate": 0.0005997517138646527,
      "loss": 2.74,
      "step": 2984
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.6549570560455322,
      "learning_rate": 0.0005997515474481271,
      "loss": 2.5708,
      "step": 2985
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7776699662208557,
      "learning_rate": 0.0005997513809758721,
      "loss": 2.6626,
      "step": 2986
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.74493807554245,
      "learning_rate": 0.0005997512144478877,
      "loss": 3.0048,
      "step": 2987
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7256177067756653,
      "learning_rate": 0.0005997510478641738,
      "loss": 2.852,
      "step": 2988
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7982985377311707,
      "learning_rate": 0.0005997508812247306,
      "loss": 2.7528,
      "step": 2989
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.771597146987915,
      "learning_rate": 0.0005997507145295579,
      "loss": 2.7415,
      "step": 2990
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.8012158274650574,
      "learning_rate": 0.0005997505477786561,
      "loss": 2.6319,
      "step": 2991
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7784268856048584,
      "learning_rate": 0.0005997503809720249,
      "loss": 2.8001,
      "step": 2992
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7843162417411804,
      "learning_rate": 0.0005997502141096645,
      "loss": 3.0032,
      "step": 2993
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.6512442827224731,
      "learning_rate": 0.0005997500471915748,
      "loss": 2.4494,
      "step": 2994
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7474708557128906,
      "learning_rate": 0.000599749880217756,
      "loss": 2.7702,
      "step": 2995
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.8550865650177002,
      "learning_rate": 0.000599749713188208,
      "loss": 2.7961,
      "step": 2996
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7275602221488953,
      "learning_rate": 0.0005997495461029308,
      "loss": 2.8761,
      "step": 2997
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.8667158484458923,
      "learning_rate": 0.0005997493789619246,
      "loss": 2.9376,
      "step": 2998
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.765769362449646,
      "learning_rate": 0.0005997492117651893,
      "loss": 2.8148,
      "step": 2999
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7754768133163452,
      "learning_rate": 0.0005997490445127249,
      "loss": 2.9825,
      "step": 3000
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7381401658058167,
      "learning_rate": 0.0005997488772045316,
      "loss": 2.605,
      "step": 3001
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7012333273887634,
      "learning_rate": 0.0005997487098406092,
      "loss": 2.9192,
      "step": 3002
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7251479029655457,
      "learning_rate": 0.0005997485424209581,
      "loss": 2.5309,
      "step": 3003
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.8143923878669739,
      "learning_rate": 0.0005997483749455778,
      "loss": 2.8155,
      "step": 3004
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7788712382316589,
      "learning_rate": 0.0005997482074144687,
      "loss": 2.616,
      "step": 3005
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.9328920245170593,
      "learning_rate": 0.0005997480398276308,
      "loss": 2.7443,
      "step": 3006
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7120596766471863,
      "learning_rate": 0.0005997478721850642,
      "loss": 2.7021,
      "step": 3007
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.080834150314331,
      "learning_rate": 0.0005997477044867686,
      "loss": 2.6432,
      "step": 3008
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.9074439406394958,
      "learning_rate": 0.0005997475367327444,
      "loss": 2.83,
      "step": 3009
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7099738717079163,
      "learning_rate": 0.0005997473689229914,
      "loss": 2.7893,
      "step": 3010
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7154755592346191,
      "learning_rate": 0.0005997472010575096,
      "loss": 2.6156,
      "step": 3011
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7869911193847656,
      "learning_rate": 0.0005997470331362993,
      "loss": 2.6161,
      "step": 3012
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7449909448623657,
      "learning_rate": 0.0005997468651593603,
      "loss": 2.6797,
      "step": 3013
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7981167435646057,
      "learning_rate": 0.0005997466971266927,
      "loss": 2.8217,
      "step": 3014
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7018601894378662,
      "learning_rate": 0.0005997465290382965,
      "loss": 2.5042,
      "step": 3015
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7187404632568359,
      "learning_rate": 0.0005997463608941719,
      "loss": 3.0646,
      "step": 3016
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.8088496327400208,
      "learning_rate": 0.0005997461926943187,
      "loss": 2.6808,
      "step": 3017
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7527032494544983,
      "learning_rate": 0.000599746024438737,
      "loss": 2.6169,
      "step": 3018
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.735998272895813,
      "learning_rate": 0.0005997458561274268,
      "loss": 2.8084,
      "step": 3019
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.8571184277534485,
      "learning_rate": 0.0005997456877603883,
      "loss": 2.7548,
      "step": 3020
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7483448386192322,
      "learning_rate": 0.0005997455193376215,
      "loss": 2.6395,
      "step": 3021
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.6861225366592407,
      "learning_rate": 0.0005997453508591262,
      "loss": 2.9166,
      "step": 3022
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7693709135055542,
      "learning_rate": 0.0005997451823249025,
      "loss": 2.7388,
      "step": 3023
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7317376136779785,
      "learning_rate": 0.0005997450137349506,
      "loss": 2.6462,
      "step": 3024
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.6789331436157227,
      "learning_rate": 0.0005997448450892705,
      "loss": 2.9033,
      "step": 3025
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.6962953805923462,
      "learning_rate": 0.0005997446763878622,
      "loss": 2.7377,
      "step": 3026
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.9716973304748535,
      "learning_rate": 0.0005997445076307257,
      "loss": 2.8519,
      "step": 3027
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7549494504928589,
      "learning_rate": 0.000599744338817861,
      "loss": 2.7361,
      "step": 3028
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7547569274902344,
      "learning_rate": 0.0005997441699492681,
      "loss": 2.7214,
      "step": 3029
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7651475667953491,
      "learning_rate": 0.0005997440010249472,
      "loss": 2.8307,
      "step": 3030
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7338738441467285,
      "learning_rate": 0.0005997438320448982,
      "loss": 2.8383,
      "step": 3031
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7192709445953369,
      "learning_rate": 0.0005997436630091211,
      "loss": 2.7664,
      "step": 3032
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.9019954204559326,
      "learning_rate": 0.0005997434939176161,
      "loss": 2.5566,
      "step": 3033
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7331597208976746,
      "learning_rate": 0.0005997433247703831,
      "loss": 2.9356,
      "step": 3034
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.335499882698059,
      "learning_rate": 0.0005997431555674221,
      "loss": 2.6504,
      "step": 3035
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.731957733631134,
      "learning_rate": 0.0005997429863087333,
      "loss": 2.7142,
      "step": 3036
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.9647489190101624,
      "learning_rate": 0.0005997428169943165,
      "loss": 2.4378,
      "step": 3037
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.2088733911514282,
      "learning_rate": 0.0005997426476241719,
      "loss": 2.912,
      "step": 3038
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7626317739486694,
      "learning_rate": 0.0005997424781982994,
      "loss": 2.9239,
      "step": 3039
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.9694529175758362,
      "learning_rate": 0.0005997423087166992,
      "loss": 2.758,
      "step": 3040
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7698675990104675,
      "learning_rate": 0.0005997421391793714,
      "loss": 2.6403,
      "step": 3041
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7521138787269592,
      "learning_rate": 0.0005997419695863156,
      "loss": 2.8624,
      "step": 3042
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.0229601860046387,
      "learning_rate": 0.0005997417999375322,
      "loss": 2.9124,
      "step": 3043
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7717283964157104,
      "learning_rate": 0.0005997416302330213,
      "loss": 2.827,
      "step": 3044
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7998528480529785,
      "learning_rate": 0.0005997414604727825,
      "loss": 2.9679,
      "step": 3045
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.9129686951637268,
      "learning_rate": 0.0005997412906568163,
      "loss": 2.6254,
      "step": 3046
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7112858295440674,
      "learning_rate": 0.0005997411207851226,
      "loss": 2.6162,
      "step": 3047
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.6621666550636292,
      "learning_rate": 0.0005997409508577013,
      "loss": 2.9028,
      "step": 3048
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.9610365629196167,
      "learning_rate": 0.0005997407808745523,
      "loss": 2.7363,
      "step": 3049
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.6999709010124207,
      "learning_rate": 0.0005997406108356761,
      "loss": 2.6643,
      "step": 3050
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7268728017807007,
      "learning_rate": 0.0005997404407410723,
      "loss": 2.8154,
      "step": 3051
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.0600876808166504,
      "learning_rate": 0.0005997402705907411,
      "loss": 2.5319,
      "step": 3052
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.773597002029419,
      "learning_rate": 0.0005997401003846825,
      "loss": 2.9784,
      "step": 3053
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7367475032806396,
      "learning_rate": 0.0005997399301228966,
      "loss": 2.9696,
      "step": 3054
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.0424847602844238,
      "learning_rate": 0.0005997397598053835,
      "loss": 2.8248,
      "step": 3055
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.8883702158927917,
      "learning_rate": 0.000599739589432143,
      "loss": 2.6284,
      "step": 3056
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.966414213180542,
      "learning_rate": 0.0005997394190031753,
      "loss": 2.7437,
      "step": 3057
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.0200377702713013,
      "learning_rate": 0.0005997392485184804,
      "loss": 2.7201,
      "step": 3058
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.6940847039222717,
      "learning_rate": 0.0005997390779780582,
      "loss": 2.4496,
      "step": 3059
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.0059758424758911,
      "learning_rate": 0.000599738907381909,
      "loss": 2.643,
      "step": 3060
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7733737826347351,
      "learning_rate": 0.0005997387367300327,
      "loss": 2.8622,
      "step": 3061
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7518700361251831,
      "learning_rate": 0.0005997385660224292,
      "loss": 2.9949,
      "step": 3062
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.8187077045440674,
      "learning_rate": 0.0005997383952590986,
      "loss": 2.8049,
      "step": 3063
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.9900141954421997,
      "learning_rate": 0.000599738224440041,
      "loss": 2.7126,
      "step": 3064
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7776948809623718,
      "learning_rate": 0.0005997380535652564,
      "loss": 2.7275,
      "step": 3065
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.8121617436408997,
      "learning_rate": 0.000599737882634745,
      "loss": 2.7555,
      "step": 3066
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7577469944953918,
      "learning_rate": 0.0005997377116485065,
      "loss": 2.7223,
      "step": 3067
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7048601508140564,
      "learning_rate": 0.0005997375406065412,
      "loss": 2.8469,
      "step": 3068
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7540109753608704,
      "learning_rate": 0.0005997373695088489,
      "loss": 2.7513,
      "step": 3069
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.9466202259063721,
      "learning_rate": 0.0005997371983554299,
      "loss": 2.6319,
      "step": 3070
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7684210538864136,
      "learning_rate": 0.0005997370271462841,
      "loss": 2.8453,
      "step": 3071
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7060946822166443,
      "learning_rate": 0.0005997368558814114,
      "loss": 2.8178,
      "step": 3072
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.9322684407234192,
      "learning_rate": 0.0005997366845608121,
      "loss": 2.5194,
      "step": 3073
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7718598246574402,
      "learning_rate": 0.000599736513184486,
      "loss": 2.9529,
      "step": 3074
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7328301072120667,
      "learning_rate": 0.0005997363417524333,
      "loss": 2.9803,
      "step": 3075
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7925010919570923,
      "learning_rate": 0.000599736170264654,
      "loss": 2.5772,
      "step": 3076
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7340567708015442,
      "learning_rate": 0.0005997359987211479,
      "loss": 2.7033,
      "step": 3077
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.1134798526763916,
      "learning_rate": 0.0005997358271219153,
      "loss": 2.6112,
      "step": 3078
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.75868821144104,
      "learning_rate": 0.0005997356554669562,
      "loss": 2.9934,
      "step": 3079
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.0495049953460693,
      "learning_rate": 0.0005997354837562706,
      "loss": 2.8006,
      "step": 3080
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7819251418113708,
      "learning_rate": 0.0005997353119898584,
      "loss": 2.5787,
      "step": 3081
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7724625468254089,
      "learning_rate": 0.0005997351401677199,
      "loss": 2.5238,
      "step": 3082
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7149321436882019,
      "learning_rate": 0.0005997349682898549,
      "loss": 2.5965,
      "step": 3083
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.8860987424850464,
      "learning_rate": 0.0005997347963562635,
      "loss": 2.8475,
      "step": 3084
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7692490816116333,
      "learning_rate": 0.0005997346243669457,
      "loss": 2.5471,
      "step": 3085
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.8333296775817871,
      "learning_rate": 0.0005997344523219016,
      "loss": 2.7996,
      "step": 3086
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.0326958894729614,
      "learning_rate": 0.0005997342802211313,
      "loss": 3.0007,
      "step": 3087
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7275033593177795,
      "learning_rate": 0.0005997341080646346,
      "loss": 2.7238,
      "step": 3088
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.0460745096206665,
      "learning_rate": 0.0005997339358524118,
      "loss": 2.6319,
      "step": 3089
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7949317097663879,
      "learning_rate": 0.0005997337635844628,
      "loss": 2.6468,
      "step": 3090
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7470471262931824,
      "learning_rate": 0.0005997335912607875,
      "loss": 2.7229,
      "step": 3091
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.9879802465438843,
      "learning_rate": 0.0005997334188813861,
      "loss": 2.9385,
      "step": 3092
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7701702117919922,
      "learning_rate": 0.0005997332464462588,
      "loss": 2.7054,
      "step": 3093
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.8376734256744385,
      "learning_rate": 0.0005997330739554052,
      "loss": 2.7975,
      "step": 3094
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.8278533816337585,
      "learning_rate": 0.0005997329014088257,
      "loss": 2.8562,
      "step": 3095
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.9754380583763123,
      "learning_rate": 0.0005997327288065201,
      "loss": 2.5717,
      "step": 3096
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7512340545654297,
      "learning_rate": 0.0005997325561484885,
      "loss": 2.8895,
      "step": 3097
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.0854547023773193,
      "learning_rate": 0.0005997323834347311,
      "loss": 2.8327,
      "step": 3098
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7509421706199646,
      "learning_rate": 0.0005997322106652477,
      "loss": 2.6718,
      "step": 3099
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7870544791221619,
      "learning_rate": 0.0005997320378400385,
      "loss": 2.6779,
      "step": 3100
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7660617232322693,
      "learning_rate": 0.0005997318649591034,
      "loss": 2.8029,
      "step": 3101
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7441657781600952,
      "learning_rate": 0.0005997316920224424,
      "loss": 2.9083,
      "step": 3102
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.6838698387145996,
      "learning_rate": 0.0005997315190300557,
      "loss": 2.9215,
      "step": 3103
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.6458368897438049,
      "learning_rate": 0.0005997313459819434,
      "loss": 2.7551,
      "step": 3104
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7688238024711609,
      "learning_rate": 0.0005997311728781052,
      "loss": 2.5215,
      "step": 3105
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7068668603897095,
      "learning_rate": 0.0005997309997185414,
      "loss": 3.0398,
      "step": 3106
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7291063666343689,
      "learning_rate": 0.0005997308265032518,
      "loss": 2.6047,
      "step": 3107
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7524872422218323,
      "learning_rate": 0.0005997306532322368,
      "loss": 2.446,
      "step": 3108
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.6929060816764832,
      "learning_rate": 0.0005997304799054961,
      "loss": 2.8038,
      "step": 3109
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.8024140000343323,
      "learning_rate": 0.0005997303065230299,
      "loss": 2.5398,
      "step": 3110
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7224769592285156,
      "learning_rate": 0.0005997301330848381,
      "loss": 2.8613,
      "step": 3111
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7105308771133423,
      "learning_rate": 0.0005997299595909208,
      "loss": 2.8247,
      "step": 3112
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.8185892105102539,
      "learning_rate": 0.0005997297860412781,
      "loss": 2.6932,
      "step": 3113
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.6432783007621765,
      "learning_rate": 0.00059972961243591,
      "loss": 2.7725,
      "step": 3114
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.8485085964202881,
      "learning_rate": 0.0005997294387748165,
      "loss": 2.7778,
      "step": 3115
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7976202964782715,
      "learning_rate": 0.0005997292650579976,
      "loss": 2.7116,
      "step": 3116
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.691712498664856,
      "learning_rate": 0.0005997290912854534,
      "loss": 2.9511,
      "step": 3117
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.993497371673584,
      "learning_rate": 0.0005997289174571839,
      "loss": 2.7409,
      "step": 3118
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7518276572227478,
      "learning_rate": 0.000599728743573189,
      "loss": 2.6198,
      "step": 3119
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7206815481185913,
      "learning_rate": 0.0005997285696334691,
      "loss": 2.8914,
      "step": 3120
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7511255145072937,
      "learning_rate": 0.0005997283956380239,
      "loss": 2.7479,
      "step": 3121
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7488914132118225,
      "learning_rate": 0.0005997282215868535,
      "loss": 2.7193,
      "step": 3122
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7989720106124878,
      "learning_rate": 0.000599728047479958,
      "loss": 2.8625,
      "step": 3123
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7312418818473816,
      "learning_rate": 0.0005997278733173374,
      "loss": 2.9827,
      "step": 3124
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7235810160636902,
      "learning_rate": 0.0005997276990989917,
      "loss": 2.7406,
      "step": 3125
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.6498996615409851,
      "learning_rate": 0.000599727524824921,
      "loss": 2.6179,
      "step": 3126
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7012662887573242,
      "learning_rate": 0.0005997273504951253,
      "loss": 2.858,
      "step": 3127
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7212352752685547,
      "learning_rate": 0.0005997271761096045,
      "loss": 2.8152,
      "step": 3128
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7337105870246887,
      "learning_rate": 0.0005997270016683589,
      "loss": 2.7406,
      "step": 3129
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.637610673904419,
      "learning_rate": 0.0005997268271713884,
      "loss": 2.4348,
      "step": 3130
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7462124824523926,
      "learning_rate": 0.000599726652618693,
      "loss": 2.9687,
      "step": 3131
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7466520667076111,
      "learning_rate": 0.0005997264780102727,
      "loss": 2.8805,
      "step": 3132
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.695097804069519,
      "learning_rate": 0.0005997263033461277,
      "loss": 2.5634,
      "step": 3133
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.6590744256973267,
      "learning_rate": 0.0005997261286262578,
      "loss": 2.9139,
      "step": 3134
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7490479350090027,
      "learning_rate": 0.0005997259538506632,
      "loss": 2.6779,
      "step": 3135
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.722741425037384,
      "learning_rate": 0.0005997257790193439,
      "loss": 2.8074,
      "step": 3136
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.9849512577056885,
      "learning_rate": 0.0005997256041322999,
      "loss": 2.8055,
      "step": 3137
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7206329107284546,
      "learning_rate": 0.0005997254291895313,
      "loss": 2.8362,
      "step": 3138
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.0493087768554688,
      "learning_rate": 0.0005997252541910381,
      "loss": 2.567,
      "step": 3139
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7381704449653625,
      "learning_rate": 0.0005997250791368202,
      "loss": 2.9398,
      "step": 3140
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.8268535137176514,
      "learning_rate": 0.0005997249040268779,
      "loss": 2.6084,
      "step": 3141
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.1324081420898438,
      "learning_rate": 0.000599724728861211,
      "loss": 2.7666,
      "step": 3142
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.6749446988105774,
      "learning_rate": 0.0005997245536398196,
      "loss": 2.7835,
      "step": 3143
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.8615117073059082,
      "learning_rate": 0.0005997243783627037,
      "loss": 2.895,
      "step": 3144
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.1580604314804077,
      "learning_rate": 0.0005997242030298635,
      "loss": 2.7939,
      "step": 3145
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7170917391777039,
      "learning_rate": 0.0005997240276412988,
      "loss": 2.964,
      "step": 3146
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.048887848854065,
      "learning_rate": 0.0005997238521970099,
      "loss": 2.7726,
      "step": 3147
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.0461078882217407,
      "learning_rate": 0.0005997236766969965,
      "loss": 2.8589,
      "step": 3148
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7039675712585449,
      "learning_rate": 0.000599723501141259,
      "loss": 2.6999,
      "step": 3149
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.0505027770996094,
      "learning_rate": 0.0005997233255297972,
      "loss": 2.759,
      "step": 3150
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.9340884685516357,
      "learning_rate": 0.000599723149862611,
      "loss": 3.0098,
      "step": 3151
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.8572199940681458,
      "learning_rate": 0.0005997229741397008,
      "loss": 2.6442,
      "step": 3152
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.9982277750968933,
      "learning_rate": 0.0005997227983610663,
      "loss": 2.9443,
      "step": 3153
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7053768038749695,
      "learning_rate": 0.0005997226225267077,
      "loss": 2.5683,
      "step": 3154
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.8384565114974976,
      "learning_rate": 0.000599722446636625,
      "loss": 2.8139,
      "step": 3155
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.8125367760658264,
      "learning_rate": 0.0005997222706908183,
      "loss": 2.8298,
      "step": 3156
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7167412638664246,
      "learning_rate": 0.0005997220946892876,
      "loss": 2.719,
      "step": 3157
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.753718376159668,
      "learning_rate": 0.0005997219186320329,
      "loss": 2.8749,
      "step": 3158
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7236995697021484,
      "learning_rate": 0.0005997217425190542,
      "loss": 2.7303,
      "step": 3159
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.6916564702987671,
      "learning_rate": 0.0005997215663503515,
      "loss": 2.7359,
      "step": 3160
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7033345699310303,
      "learning_rate": 0.0005997213901259251,
      "loss": 2.6868,
      "step": 3161
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7176932692527771,
      "learning_rate": 0.0005997212138457746,
      "loss": 2.7063,
      "step": 3162
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7581945061683655,
      "learning_rate": 0.0005997210375099004,
      "loss": 2.8918,
      "step": 3163
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7304131984710693,
      "learning_rate": 0.0005997208611183024,
      "loss": 2.7459,
      "step": 3164
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.710818886756897,
      "learning_rate": 0.0005997206846709806,
      "loss": 2.8061,
      "step": 3165
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7289183735847473,
      "learning_rate": 0.000599720508167935,
      "loss": 2.8086,
      "step": 3166
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7732434272766113,
      "learning_rate": 0.0005997203316091659,
      "loss": 2.9723,
      "step": 3167
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.140291452407837,
      "learning_rate": 0.0005997201549946729,
      "loss": 2.7692,
      "step": 3168
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7917208075523376,
      "learning_rate": 0.0005997199783244565,
      "loss": 2.7972,
      "step": 3169
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.9499754309654236,
      "learning_rate": 0.0005997198015985164,
      "loss": 2.6997,
      "step": 3170
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.752458393573761,
      "learning_rate": 0.0005997196248168526,
      "loss": 2.5057,
      "step": 3171
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.710110068321228,
      "learning_rate": 0.0005997194479794653,
      "loss": 2.8618,
      "step": 3172
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7349410057067871,
      "learning_rate": 0.0005997192710863547,
      "loss": 3.039,
      "step": 3173
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7859865427017212,
      "learning_rate": 0.0005997190941375204,
      "loss": 2.771,
      "step": 3174
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.669921338558197,
      "learning_rate": 0.0005997189171329628,
      "loss": 2.6641,
      "step": 3175
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7656953930854797,
      "learning_rate": 0.0005997187400726818,
      "loss": 2.8514,
      "step": 3176
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.9133837819099426,
      "learning_rate": 0.0005997185629566773,
      "loss": 2.6928,
      "step": 3177
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.8002128005027771,
      "learning_rate": 0.0005997183857849497,
      "loss": 2.8223,
      "step": 3178
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.6814646124839783,
      "learning_rate": 0.0005997182085574986,
      "loss": 2.8385,
      "step": 3179
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.782913088798523,
      "learning_rate": 0.0005997180312743243,
      "loss": 2.8837,
      "step": 3180
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.8041346073150635,
      "learning_rate": 0.0005997178539354266,
      "loss": 2.9556,
      "step": 3181
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7590048313140869,
      "learning_rate": 0.0005997176765408058,
      "loss": 2.7714,
      "step": 3182
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.0092663764953613,
      "learning_rate": 0.000599717499090462,
      "loss": 2.9965,
      "step": 3183
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.8062834143638611,
      "learning_rate": 0.0005997173215843949,
      "loss": 2.7163,
      "step": 3184
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.8581869006156921,
      "learning_rate": 0.0005997171440226047,
      "loss": 2.7479,
      "step": 3185
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.1597777605056763,
      "learning_rate": 0.0005997169664050914,
      "loss": 2.7889,
      "step": 3186
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7014813423156738,
      "learning_rate": 0.0005997167887318551,
      "loss": 2.4039,
      "step": 3187
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.820532500743866,
      "learning_rate": 0.0005997166110028958,
      "loss": 2.9098,
      "step": 3188
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.8756434321403503,
      "learning_rate": 0.0005997164332182135,
      "loss": 2.8106,
      "step": 3189
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7822791337966919,
      "learning_rate": 0.0005997162553778082,
      "loss": 2.8609,
      "step": 3190
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7004393339157104,
      "learning_rate": 0.0005997160774816801,
      "loss": 2.7442,
      "step": 3191
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.9906315207481384,
      "learning_rate": 0.000599715899529829,
      "loss": 2.5039,
      "step": 3192
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.0155888795852661,
      "learning_rate": 0.0005997157215222552,
      "loss": 2.5291,
      "step": 3193
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.8679779171943665,
      "learning_rate": 0.0005997155434589584,
      "loss": 2.798,
      "step": 3194
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.9041427969932556,
      "learning_rate": 0.0005997153653399391,
      "loss": 3.0406,
      "step": 3195
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.8057238459587097,
      "learning_rate": 0.0005997151871651968,
      "loss": 2.6306,
      "step": 3196
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.751012921333313,
      "learning_rate": 0.0005997150089347319,
      "loss": 2.793,
      "step": 3197
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.9353041648864746,
      "learning_rate": 0.0005997148306485443,
      "loss": 2.714,
      "step": 3198
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7805611491203308,
      "learning_rate": 0.0005997146523066342,
      "loss": 2.6309,
      "step": 3199
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.6671451330184937,
      "learning_rate": 0.0005997144739090012,
      "loss": 2.6734,
      "step": 3200
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.8225909471511841,
      "learning_rate": 0.0005997142954556458,
      "loss": 2.8431,
      "step": 3201
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.8567697405815125,
      "learning_rate": 0.0005997141169465679,
      "loss": 2.5713,
      "step": 3202
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.8205916881561279,
      "learning_rate": 0.0005997139383817673,
      "loss": 2.5271,
      "step": 3203
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.835092306137085,
      "learning_rate": 0.0005997137597612444,
      "loss": 2.739,
      "step": 3204
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.6925826072692871,
      "learning_rate": 0.000599713581084999,
      "loss": 2.4761,
      "step": 3205
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.6857867240905762,
      "learning_rate": 0.0005997134023530312,
      "loss": 2.6309,
      "step": 3206
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.8615903854370117,
      "learning_rate": 0.0005997132235653409,
      "loss": 2.7449,
      "step": 3207
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7102872133255005,
      "learning_rate": 0.0005997130447219284,
      "loss": 2.8253,
      "step": 3208
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.8570089936256409,
      "learning_rate": 0.0005997128658227936,
      "loss": 2.7285,
      "step": 3209
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.9066569805145264,
      "learning_rate": 0.0005997126868679364,
      "loss": 2.8141,
      "step": 3210
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.8974902629852295,
      "learning_rate": 0.000599712507857357,
      "loss": 2.7399,
      "step": 3211
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.8086978793144226,
      "learning_rate": 0.0005997123287910555,
      "loss": 2.8563,
      "step": 3212
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.754443883895874,
      "learning_rate": 0.0005997121496690317,
      "loss": 2.7922,
      "step": 3213
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.8215131163597107,
      "learning_rate": 0.0005997119704912858,
      "loss": 2.8466,
      "step": 3214
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7994630336761475,
      "learning_rate": 0.0005997117912578177,
      "loss": 2.7545,
      "step": 3215
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7245289087295532,
      "learning_rate": 0.0005997116119686276,
      "loss": 2.6966,
      "step": 3216
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.8117051720619202,
      "learning_rate": 0.0005997114326237154,
      "loss": 2.8564,
      "step": 3217
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7359871864318848,
      "learning_rate": 0.0005997112532230813,
      "loss": 2.7817,
      "step": 3218
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.730172872543335,
      "learning_rate": 0.0005997110737667252,
      "loss": 2.9555,
      "step": 3219
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.8210648894309998,
      "learning_rate": 0.000599710894254647,
      "loss": 2.7088,
      "step": 3220
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7766707539558411,
      "learning_rate": 0.0005997107146868471,
      "loss": 2.569,
      "step": 3221
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.6757287979125977,
      "learning_rate": 0.0005997105350633252,
      "loss": 2.8037,
      "step": 3222
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7766783833503723,
      "learning_rate": 0.0005997103553840814,
      "loss": 2.8644,
      "step": 3223
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.8405927419662476,
      "learning_rate": 0.0005997101756491158,
      "loss": 2.7483,
      "step": 3224
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.6442274451255798,
      "learning_rate": 0.0005997099958584285,
      "loss": 2.6338,
      "step": 3225
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7937978506088257,
      "learning_rate": 0.0005997098160120195,
      "loss": 2.9311,
      "step": 3226
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.976836621761322,
      "learning_rate": 0.0005997096361098886,
      "loss": 2.7724,
      "step": 3227
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7001932263374329,
      "learning_rate": 0.0005997094561520362,
      "loss": 2.8059,
      "step": 3228
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.924881637096405,
      "learning_rate": 0.0005997092761384622,
      "loss": 2.8705,
      "step": 3229
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7281851768493652,
      "learning_rate": 0.0005997090960691664,
      "loss": 2.7111,
      "step": 3230
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7466663122177124,
      "learning_rate": 0.0005997089159441491,
      "loss": 2.9349,
      "step": 3231
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.9167428612709045,
      "learning_rate": 0.0005997087357634103,
      "loss": 2.6871,
      "step": 3232
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7199390530586243,
      "learning_rate": 0.00059970855552695,
      "loss": 2.8085,
      "step": 3233
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7727127075195312,
      "learning_rate": 0.0005997083752347681,
      "loss": 2.8084,
      "step": 3234
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7838186621665955,
      "learning_rate": 0.0005997081948868648,
      "loss": 2.7616,
      "step": 3235
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.6208038330078125,
      "learning_rate": 0.0005997080144832401,
      "loss": 2.6797,
      "step": 3236
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7272423505783081,
      "learning_rate": 0.0005997078340238941,
      "loss": 2.7887,
      "step": 3237
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.9444236159324646,
      "learning_rate": 0.0005997076535088267,
      "loss": 2.7735,
      "step": 3238
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.741002082824707,
      "learning_rate": 0.000599707472938038,
      "loss": 2.7524,
      "step": 3239
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.6984192728996277,
      "learning_rate": 0.000599707292311528,
      "loss": 2.7154,
      "step": 3240
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.9909294247627258,
      "learning_rate": 0.0005997071116292967,
      "loss": 2.5875,
      "step": 3241
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.8972401022911072,
      "learning_rate": 0.0005997069308913443,
      "loss": 2.71,
      "step": 3242
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7563709616661072,
      "learning_rate": 0.0005997067500976707,
      "loss": 2.9209,
      "step": 3243
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.6659959554672241,
      "learning_rate": 0.0005997065692482759,
      "loss": 2.6424,
      "step": 3244
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7905386686325073,
      "learning_rate": 0.00059970638834316,
      "loss": 2.6433,
      "step": 3245
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.8219299912452698,
      "learning_rate": 0.0005997062073823231,
      "loss": 2.9174,
      "step": 3246
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.6875882744789124,
      "learning_rate": 0.000599706026365765,
      "loss": 2.8431,
      "step": 3247
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.6590665578842163,
      "learning_rate": 0.000599705845293486,
      "loss": 2.6101,
      "step": 3248
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.6564117074012756,
      "learning_rate": 0.0005997056641654861,
      "loss": 2.4591,
      "step": 3249
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.819888710975647,
      "learning_rate": 0.0005997054829817651,
      "loss": 2.8403,
      "step": 3250
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7530896663665771,
      "learning_rate": 0.0005997053017423234,
      "loss": 2.5301,
      "step": 3251
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7047868967056274,
      "learning_rate": 0.0005997051204471607,
      "loss": 2.9728,
      "step": 3252
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.6887913346290588,
      "learning_rate": 0.0005997049390962771,
      "loss": 2.5807,
      "step": 3253
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.6563579440116882,
      "learning_rate": 0.0005997047576896728,
      "loss": 2.9615,
      "step": 3254
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.830814003944397,
      "learning_rate": 0.0005997045762273477,
      "loss": 2.5727,
      "step": 3255
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7196935415267944,
      "learning_rate": 0.0005997043947093018,
      "loss": 2.6815,
      "step": 3256
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.6945102214813232,
      "learning_rate": 0.0005997042131355352,
      "loss": 2.9926,
      "step": 3257
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.8716636300086975,
      "learning_rate": 0.000599704031506048,
      "loss": 2.7176,
      "step": 3258
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7686424255371094,
      "learning_rate": 0.0005997038498208403,
      "loss": 2.7715,
      "step": 3259
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7428290843963623,
      "learning_rate": 0.0005997036680799117,
      "loss": 2.6219,
      "step": 3260
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.0620545148849487,
      "learning_rate": 0.0005997034862832627,
      "loss": 2.883,
      "step": 3261
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7824640870094299,
      "learning_rate": 0.0005997033044308932,
      "loss": 2.5424,
      "step": 3262
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7989146709442139,
      "learning_rate": 0.0005997031225228031,
      "loss": 2.9023,
      "step": 3263
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.9412013292312622,
      "learning_rate": 0.0005997029405589926,
      "loss": 2.902,
      "step": 3264
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.8274595141410828,
      "learning_rate": 0.0005997027585394615,
      "loss": 2.9815,
      "step": 3265
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7000468969345093,
      "learning_rate": 0.0005997025764642103,
      "loss": 2.8035,
      "step": 3266
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.983852207660675,
      "learning_rate": 0.0005997023943332385,
      "loss": 2.7761,
      "step": 3267
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.822995662689209,
      "learning_rate": 0.0005997022121465463,
      "loss": 2.5024,
      "step": 3268
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.6445481777191162,
      "learning_rate": 0.000599702029904134,
      "loss": 2.654,
      "step": 3269
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.8654457330703735,
      "learning_rate": 0.0005997018476060013,
      "loss": 2.7484,
      "step": 3270
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.9518757462501526,
      "learning_rate": 0.0005997016652521485,
      "loss": 2.971,
      "step": 3271
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7275367379188538,
      "learning_rate": 0.0005997014828425755,
      "loss": 2.8543,
      "step": 3272
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.849179208278656,
      "learning_rate": 0.0005997013003772822,
      "loss": 2.9326,
      "step": 3273
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.1840541362762451,
      "learning_rate": 0.0005997011178562689,
      "loss": 2.5592,
      "step": 3274
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7433173060417175,
      "learning_rate": 0.0005997009352795353,
      "loss": 2.5177,
      "step": 3275
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.9943181276321411,
      "learning_rate": 0.000599700752647082,
      "loss": 2.7555,
      "step": 3276
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.9565038084983826,
      "learning_rate": 0.0005997005699589084,
      "loss": 2.4687,
      "step": 3277
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.670697033405304,
      "learning_rate": 0.0005997003872150148,
      "loss": 2.7324,
      "step": 3278
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.8207202553749084,
      "learning_rate": 0.0005997002044154014,
      "loss": 2.75,
      "step": 3279
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.1483430862426758,
      "learning_rate": 0.000599700021560068,
      "loss": 2.7005,
      "step": 3280
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.9890543222427368,
      "learning_rate": 0.0005996998386490147,
      "loss": 2.8736,
      "step": 3281
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7051663398742676,
      "learning_rate": 0.0005996996556822416,
      "loss": 2.8356,
      "step": 3282
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.272169589996338,
      "learning_rate": 0.0005996994726597486,
      "loss": 2.9054,
      "step": 3283
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.420310378074646,
      "learning_rate": 0.0005996992895815357,
      "loss": 2.8398,
      "step": 3284
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7689629197120667,
      "learning_rate": 0.0005996991064476032,
      "loss": 2.885,
      "step": 3285
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.8278616666793823,
      "learning_rate": 0.000599698923257951,
      "loss": 2.9077,
      "step": 3286
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.9353417158126831,
      "learning_rate": 0.0005996987400125791,
      "loss": 2.7046,
      "step": 3287
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.1472299098968506,
      "learning_rate": 0.0005996985567114874,
      "loss": 2.7367,
      "step": 3288
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7965690493583679,
      "learning_rate": 0.0005996983733546762,
      "loss": 3.0141,
      "step": 3289
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7720872163772583,
      "learning_rate": 0.0005996981899421454,
      "loss": 2.8049,
      "step": 3290
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.8381783366203308,
      "learning_rate": 0.0005996980064738952,
      "loss": 2.9315,
      "step": 3291
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.110802173614502,
      "learning_rate": 0.0005996978229499253,
      "loss": 2.7525,
      "step": 3292
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7448030114173889,
      "learning_rate": 0.000599697639370236,
      "loss": 2.702,
      "step": 3293
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.8254667520523071,
      "learning_rate": 0.000599697455734827,
      "loss": 2.8446,
      "step": 3294
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.9839121699333191,
      "learning_rate": 0.0005996972720436988,
      "loss": 2.8772,
      "step": 3295
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7756014466285706,
      "learning_rate": 0.0005996970882968512,
      "loss": 2.9833,
      "step": 3296
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7212526202201843,
      "learning_rate": 0.0005996969044942843,
      "loss": 2.7671,
      "step": 3297
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.749504029750824,
      "learning_rate": 0.0005996967206359981,
      "loss": 2.765,
      "step": 3298
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.6799428462982178,
      "learning_rate": 0.0005996965367219926,
      "loss": 2.8638,
      "step": 3299
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7072200775146484,
      "learning_rate": 0.0005996963527522679,
      "loss": 2.7739,
      "step": 3300
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7014987468719482,
      "learning_rate": 0.0005996961687268238,
      "loss": 2.7904,
      "step": 3301
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7698090672492981,
      "learning_rate": 0.0005996959846456607,
      "loss": 2.8148,
      "step": 3302
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.911372184753418,
      "learning_rate": 0.0005996958005087784,
      "loss": 2.8517,
      "step": 3303
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7161068916320801,
      "learning_rate": 0.0005996956163161769,
      "loss": 2.5879,
      "step": 3304
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7608746886253357,
      "learning_rate": 0.0005996954320678566,
      "loss": 2.6055,
      "step": 3305
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.8092499375343323,
      "learning_rate": 0.000599695247763817,
      "loss": 2.6438,
      "step": 3306
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.0437308549880981,
      "learning_rate": 0.0005996950634040585,
      "loss": 2.819,
      "step": 3307
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7480100393295288,
      "learning_rate": 0.000599694878988581,
      "loss": 2.6448,
      "step": 3308
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7120299935340881,
      "learning_rate": 0.0005996946945173846,
      "loss": 2.6528,
      "step": 3309
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.8435896039009094,
      "learning_rate": 0.0005996945099904692,
      "loss": 2.8281,
      "step": 3310
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7799150943756104,
      "learning_rate": 0.0005996943254078351,
      "loss": 3.1413,
      "step": 3311
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.733951985836029,
      "learning_rate": 0.0005996941407694821,
      "loss": 2.5915,
      "step": 3312
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7765745520591736,
      "learning_rate": 0.0005996939560754103,
      "loss": 2.9083,
      "step": 3313
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7690094113349915,
      "learning_rate": 0.0005996937713256197,
      "loss": 2.6422,
      "step": 3314
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7929760217666626,
      "learning_rate": 0.0005996935865201105,
      "loss": 3.019,
      "step": 3315
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.6471027135848999,
      "learning_rate": 0.0005996934016588825,
      "loss": 2.6646,
      "step": 3316
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.8137913346290588,
      "learning_rate": 0.0005996932167419357,
      "loss": 2.8864,
      "step": 3317
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7174399495124817,
      "learning_rate": 0.0005996930317692705,
      "loss": 2.9426,
      "step": 3318
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7113115191459656,
      "learning_rate": 0.0005996928467408866,
      "loss": 2.7387,
      "step": 3319
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.6796691417694092,
      "learning_rate": 0.0005996926616567841,
      "loss": 2.9442,
      "step": 3320
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7307446599006653,
      "learning_rate": 0.0005996924765169631,
      "loss": 2.9419,
      "step": 3321
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.6660095453262329,
      "learning_rate": 0.0005996922913214237,
      "loss": 2.744,
      "step": 3322
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.6929724812507629,
      "learning_rate": 0.0005996921060701659,
      "loss": 2.6514,
      "step": 3323
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7667285799980164,
      "learning_rate": 0.0005996919207631895,
      "loss": 2.9662,
      "step": 3324
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7215331792831421,
      "learning_rate": 0.0005996917354004948,
      "loss": 2.8739,
      "step": 3325
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7948225140571594,
      "learning_rate": 0.0005996915499820818,
      "loss": 2.5945,
      "step": 3326
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7759081721305847,
      "learning_rate": 0.0005996913645079504,
      "loss": 2.8507,
      "step": 3327
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7707963585853577,
      "learning_rate": 0.0005996911789781007,
      "loss": 2.7832,
      "step": 3328
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.2082843780517578,
      "learning_rate": 0.0005996909933925328,
      "loss": 2.5197,
      "step": 3329
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7337971329689026,
      "learning_rate": 0.0005996908077512467,
      "loss": 2.6983,
      "step": 3330
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.8612170219421387,
      "learning_rate": 0.0005996906220542425,
      "loss": 2.5906,
      "step": 3331
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.706296443939209,
      "learning_rate": 0.0005996904363015199,
      "loss": 2.6497,
      "step": 3332
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7352116703987122,
      "learning_rate": 0.0005996902504930793,
      "loss": 2.6155,
      "step": 3333
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.8263791799545288,
      "learning_rate": 0.0005996900646289208,
      "loss": 2.9763,
      "step": 3334
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7963769435882568,
      "learning_rate": 0.0005996898787090442,
      "loss": 2.7804,
      "step": 3335
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7568255662918091,
      "learning_rate": 0.0005996896927334494,
      "loss": 2.5591,
      "step": 3336
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7339212894439697,
      "learning_rate": 0.0005996895067021368,
      "loss": 2.7542,
      "step": 3337
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7139725089073181,
      "learning_rate": 0.0005996893206151062,
      "loss": 2.858,
      "step": 3338
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7880306839942932,
      "learning_rate": 0.0005996891344723579,
      "loss": 2.7532,
      "step": 3339
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.8154874444007874,
      "learning_rate": 0.0005996889482738914,
      "loss": 2.6517,
      "step": 3340
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.9041105508804321,
      "learning_rate": 0.0005996887620197072,
      "loss": 2.7775,
      "step": 3341
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.8416385650634766,
      "learning_rate": 0.0005996885757098054,
      "loss": 2.8929,
      "step": 3342
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7667627930641174,
      "learning_rate": 0.0005996883893441855,
      "loss": 2.6379,
      "step": 3343
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7335537075996399,
      "learning_rate": 0.000599688202922848,
      "loss": 2.6768,
      "step": 3344
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7973480820655823,
      "learning_rate": 0.0005996880164457929,
      "loss": 2.9932,
      "step": 3345
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.8260746598243713,
      "learning_rate": 0.00059968782991302,
      "loss": 2.5871,
      "step": 3346
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7221718430519104,
      "learning_rate": 0.0005996876433245297,
      "loss": 2.7292,
      "step": 3347
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.9300965666770935,
      "learning_rate": 0.0005996874566803216,
      "loss": 2.6863,
      "step": 3348
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7026450634002686,
      "learning_rate": 0.0005996872699803961,
      "loss": 2.8177,
      "step": 3349
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.0816584825515747,
      "learning_rate": 0.0005996870832247529,
      "loss": 3.0444,
      "step": 3350
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7373899817466736,
      "learning_rate": 0.0005996868964133923,
      "loss": 2.6445,
      "step": 3351
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7425268888473511,
      "learning_rate": 0.0005996867095463143,
      "loss": 2.6413,
      "step": 3352
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7522526383399963,
      "learning_rate": 0.0005996865226235188,
      "loss": 2.8214,
      "step": 3353
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7880294919013977,
      "learning_rate": 0.0005996863356450059,
      "loss": 2.6259,
      "step": 3354
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.650791585445404,
      "learning_rate": 0.0005996861486107758,
      "loss": 3.0612,
      "step": 3355
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.6913259029388428,
      "learning_rate": 0.0005996859615208283,
      "loss": 2.6934,
      "step": 3356
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7688935995101929,
      "learning_rate": 0.0005996857743751635,
      "loss": 3.0716,
      "step": 3357
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7641525864601135,
      "learning_rate": 0.0005996855871737815,
      "loss": 2.8759,
      "step": 3358
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.704692006111145,
      "learning_rate": 0.0005996853999166822,
      "loss": 2.5233,
      "step": 3359
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.8031063675880432,
      "learning_rate": 0.0005996852126038659,
      "loss": 2.5153,
      "step": 3360
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7178550362586975,
      "learning_rate": 0.0005996850252353324,
      "loss": 2.8224,
      "step": 3361
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7057612538337708,
      "learning_rate": 0.0005996848378110816,
      "loss": 2.7621,
      "step": 3362
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.8784106969833374,
      "learning_rate": 0.0005996846503311139,
      "loss": 2.6622,
      "step": 3363
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7089647054672241,
      "learning_rate": 0.0005996844627954292,
      "loss": 2.8129,
      "step": 3364
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7249414324760437,
      "learning_rate": 0.0005996842752040275,
      "loss": 2.6953,
      "step": 3365
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7084066867828369,
      "learning_rate": 0.0005996840875569087,
      "loss": 2.7746,
      "step": 3366
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7644432187080383,
      "learning_rate": 0.0005996838998540731,
      "loss": 2.6581,
      "step": 3367
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.6897221803665161,
      "learning_rate": 0.0005996837120955205,
      "loss": 2.4064,
      "step": 3368
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7058778405189514,
      "learning_rate": 0.0005996835242812512,
      "loss": 2.5795,
      "step": 3369
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7842355370521545,
      "learning_rate": 0.000599683336411265,
      "loss": 2.7749,
      "step": 3370
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7485859394073486,
      "learning_rate": 0.000599683148485562,
      "loss": 2.8458,
      "step": 3371
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7088431715965271,
      "learning_rate": 0.0005996829605041422,
      "loss": 2.7639,
      "step": 3372
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7073867321014404,
      "learning_rate": 0.0005996827724670057,
      "loss": 3.1348,
      "step": 3373
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7493699789047241,
      "learning_rate": 0.0005996825843741525,
      "loss": 2.8209,
      "step": 3374
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.8172833919525146,
      "learning_rate": 0.0005996823962255828,
      "loss": 2.5853,
      "step": 3375
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7106552124023438,
      "learning_rate": 0.0005996822080212963,
      "loss": 2.6069,
      "step": 3376
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7590303421020508,
      "learning_rate": 0.0005996820197612932,
      "loss": 2.814,
      "step": 3377
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.6957427263259888,
      "learning_rate": 0.0005996818314455738,
      "loss": 2.7736,
      "step": 3378
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7066436409950256,
      "learning_rate": 0.0005996816430741377,
      "loss": 2.6197,
      "step": 3379
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.6935788989067078,
      "learning_rate": 0.0005996814546469852,
      "loss": 2.8263,
      "step": 3380
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.901788592338562,
      "learning_rate": 0.0005996812661641162,
      "loss": 2.9633,
      "step": 3381
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.6755208373069763,
      "learning_rate": 0.0005996810776255308,
      "loss": 2.8868,
      "step": 3382
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7472378611564636,
      "learning_rate": 0.0005996808890312291,
      "loss": 2.6585,
      "step": 3383
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.6950411200523376,
      "learning_rate": 0.000599680700381211,
      "loss": 2.6968,
      "step": 3384
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.6813596487045288,
      "learning_rate": 0.0005996805116754766,
      "loss": 2.4751,
      "step": 3385
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7693156003952026,
      "learning_rate": 0.0005996803229140259,
      "loss": 2.7824,
      "step": 3386
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.6964719891548157,
      "learning_rate": 0.0005996801340968591,
      "loss": 2.8208,
      "step": 3387
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.730888843536377,
      "learning_rate": 0.000599679945223976,
      "loss": 2.6096,
      "step": 3388
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.6614785194396973,
      "learning_rate": 0.0005996797562953768,
      "loss": 2.7666,
      "step": 3389
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7037370800971985,
      "learning_rate": 0.0005996795673110614,
      "loss": 2.8679,
      "step": 3390
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.8321393728256226,
      "learning_rate": 0.00059967937827103,
      "loss": 2.769,
      "step": 3391
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.6738319993019104,
      "learning_rate": 0.0005996791891752825,
      "loss": 2.885,
      "step": 3392
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7224867343902588,
      "learning_rate": 0.000599679000023819,
      "loss": 2.9949,
      "step": 3393
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.8239971399307251,
      "learning_rate": 0.0005996788108166395,
      "loss": 2.6829,
      "step": 3394
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7105649709701538,
      "learning_rate": 0.000599678621553744,
      "loss": 2.9217,
      "step": 3395
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.8046759366989136,
      "learning_rate": 0.0005996784322351328,
      "loss": 2.7505,
      "step": 3396
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.6836703419685364,
      "learning_rate": 0.0005996782428608055,
      "loss": 2.6137,
      "step": 3397
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7364752888679504,
      "learning_rate": 0.0005996780534307625,
      "loss": 2.8171,
      "step": 3398
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.6616615653038025,
      "learning_rate": 0.0005996778639450036,
      "loss": 2.7434,
      "step": 3399
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.6867595314979553,
      "learning_rate": 0.0005996776744035289,
      "loss": 2.8638,
      "step": 3400
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.6687508821487427,
      "learning_rate": 0.0005996774848063385,
      "loss": 2.9048,
      "step": 3401
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.66903156042099,
      "learning_rate": 0.0005996772951534325,
      "loss": 2.8324,
      "step": 3402
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7801522612571716,
      "learning_rate": 0.0005996771054448107,
      "loss": 2.8353,
      "step": 3403
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.6524394750595093,
      "learning_rate": 0.0005996769156804732,
      "loss": 2.8636,
      "step": 3404
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7758791446685791,
      "learning_rate": 0.0005996767258604202,
      "loss": 2.7303,
      "step": 3405
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7573622465133667,
      "learning_rate": 0.0005996765359846518,
      "loss": 2.8957,
      "step": 3406
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.729647696018219,
      "learning_rate": 0.0005996763460531677,
      "loss": 2.6442,
      "step": 3407
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.704620361328125,
      "learning_rate": 0.0005996761560659681,
      "loss": 2.6074,
      "step": 3408
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.6784922480583191,
      "learning_rate": 0.0005996759660230531,
      "loss": 2.7581,
      "step": 3409
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.708784282207489,
      "learning_rate": 0.0005996757759244226,
      "loss": 2.6342,
      "step": 3410
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7133508920669556,
      "learning_rate": 0.0005996755857700769,
      "loss": 2.9683,
      "step": 3411
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7225211262702942,
      "learning_rate": 0.0005996753955600158,
      "loss": 2.9639,
      "step": 3412
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7303190231323242,
      "learning_rate": 0.0005996752052942393,
      "loss": 2.6613,
      "step": 3413
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.6673895120620728,
      "learning_rate": 0.0005996750149727476,
      "loss": 2.716,
      "step": 3414
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7308299541473389,
      "learning_rate": 0.0005996748245955405,
      "loss": 2.8152,
      "step": 3415
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.790764331817627,
      "learning_rate": 0.0005996746341626184,
      "loss": 2.9405,
      "step": 3416
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.6530826687812805,
      "learning_rate": 0.0005996744436739809,
      "loss": 2.906,
      "step": 3417
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.793573260307312,
      "learning_rate": 0.0005996742531296284,
      "loss": 2.6201,
      "step": 3418
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7057533264160156,
      "learning_rate": 0.0005996740625295608,
      "loss": 2.6268,
      "step": 3419
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7041783332824707,
      "learning_rate": 0.0005996738718737781,
      "loss": 2.7308,
      "step": 3420
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7215256094932556,
      "learning_rate": 0.0005996736811622804,
      "loss": 2.7024,
      "step": 3421
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7550361752510071,
      "learning_rate": 0.0005996734903950678,
      "loss": 2.9164,
      "step": 3422
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.727400541305542,
      "learning_rate": 0.0005996732995721401,
      "loss": 2.818,
      "step": 3423
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7382482886314392,
      "learning_rate": 0.0005996731086934976,
      "loss": 2.5776,
      "step": 3424
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.8127594590187073,
      "learning_rate": 0.00059967291775914,
      "loss": 2.4634,
      "step": 3425
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7480801343917847,
      "learning_rate": 0.0005996727267690678,
      "loss": 2.5807,
      "step": 3426
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.8532829284667969,
      "learning_rate": 0.0005996725357232806,
      "loss": 2.66,
      "step": 3427
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.8301892280578613,
      "learning_rate": 0.0005996723446217787,
      "loss": 2.9135,
      "step": 3428
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7359594702720642,
      "learning_rate": 0.000599672153464562,
      "loss": 2.6594,
      "step": 3429
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7560325860977173,
      "learning_rate": 0.0005996719622516307,
      "loss": 2.5575,
      "step": 3430
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.8120593428611755,
      "learning_rate": 0.0005996717709829846,
      "loss": 2.7397,
      "step": 3431
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7307490706443787,
      "learning_rate": 0.0005996715796586239,
      "loss": 2.6567,
      "step": 3432
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.8547279238700867,
      "learning_rate": 0.0005996713882785487,
      "loss": 2.7362,
      "step": 3433
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.8137406706809998,
      "learning_rate": 0.0005996711968427589,
      "loss": 2.9193,
      "step": 3434
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7053623795509338,
      "learning_rate": 0.0005996710053512545,
      "loss": 2.7604,
      "step": 3435
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.665229082107544,
      "learning_rate": 0.0005996708138040355,
      "loss": 2.7823,
      "step": 3436
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7430388331413269,
      "learning_rate": 0.0005996706222011021,
      "loss": 2.6317,
      "step": 3437
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7080830931663513,
      "learning_rate": 0.0005996704305424544,
      "loss": 2.7674,
      "step": 3438
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7388375401496887,
      "learning_rate": 0.0005996702388280922,
      "loss": 2.8692,
      "step": 3439
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.8174409866333008,
      "learning_rate": 0.0005996700470580158,
      "loss": 2.7556,
      "step": 3440
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7728472948074341,
      "learning_rate": 0.0005996698552322249,
      "loss": 2.81,
      "step": 3441
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7240861654281616,
      "learning_rate": 0.0005996696633507198,
      "loss": 2.6981,
      "step": 3442
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.6697321534156799,
      "learning_rate": 0.0005996694714135004,
      "loss": 2.8337,
      "step": 3443
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.8334381580352783,
      "learning_rate": 0.0005996692794205668,
      "loss": 2.6504,
      "step": 3444
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7370799779891968,
      "learning_rate": 0.0005996690873719191,
      "loss": 2.639,
      "step": 3445
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.8583541512489319,
      "learning_rate": 0.0005996688952675572,
      "loss": 2.4792,
      "step": 3446
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7833145260810852,
      "learning_rate": 0.0005996687031074812,
      "loss": 2.8368,
      "step": 3447
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7676903009414673,
      "learning_rate": 0.0005996685108916911,
      "loss": 2.9152,
      "step": 3448
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7253584861755371,
      "learning_rate": 0.0005996683186201871,
      "loss": 2.5486,
      "step": 3449
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.756386399269104,
      "learning_rate": 0.000599668126292969,
      "loss": 2.6712,
      "step": 3450
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.8815136551856995,
      "learning_rate": 0.0005996679339100369,
      "loss": 2.7562,
      "step": 3451
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7889834642410278,
      "learning_rate": 0.0005996677414713909,
      "loss": 2.8965,
      "step": 3452
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.841606616973877,
      "learning_rate": 0.0005996675489770311,
      "loss": 2.9602,
      "step": 3453
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.8714855313301086,
      "learning_rate": 0.0005996673564269573,
      "loss": 2.7552,
      "step": 3454
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7313205599784851,
      "learning_rate": 0.0005996671638211698,
      "loss": 2.7909,
      "step": 3455
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.8845006823539734,
      "learning_rate": 0.0005996669711596684,
      "loss": 2.8308,
      "step": 3456
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.0091100931167603,
      "learning_rate": 0.0005996667784424534,
      "loss": 2.7092,
      "step": 3457
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.7108439207077026,
      "learning_rate": 0.0005996665856695246,
      "loss": 2.7932,
      "step": 3458
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.7964418530464172,
      "learning_rate": 0.0005996663928408821,
      "loss": 2.8611,
      "step": 3459
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.895983099937439,
      "learning_rate": 0.0005996661999565261,
      "loss": 2.7896,
      "step": 3460
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.7321109175682068,
      "learning_rate": 0.0005996660070164564,
      "loss": 2.7777,
      "step": 3461
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.8447698950767517,
      "learning_rate": 0.0005996658140206731,
      "loss": 2.6097,
      "step": 3462
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.8274186849594116,
      "learning_rate": 0.0005996656209691763,
      "loss": 3.1052,
      "step": 3463
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.8030497431755066,
      "learning_rate": 0.000599665427861966,
      "loss": 2.7412,
      "step": 3464
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.7162859439849854,
      "learning_rate": 0.0005996652346990422,
      "loss": 2.9248,
      "step": 3465
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.7666621208190918,
      "learning_rate": 0.000599665041480405,
      "loss": 2.6564,
      "step": 3466
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.7802255153656006,
      "learning_rate": 0.0005996648482060545,
      "loss": 2.9094,
      "step": 3467
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.722649097442627,
      "learning_rate": 0.0005996646548759906,
      "loss": 2.9414,
      "step": 3468
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.8166847229003906,
      "learning_rate": 0.0005996644614902134,
      "loss": 2.8277,
      "step": 3469
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.8723264336585999,
      "learning_rate": 0.0005996642680487228,
      "loss": 2.9199,
      "step": 3470
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.71773761510849,
      "learning_rate": 0.0005996640745515191,
      "loss": 2.6824,
      "step": 3471
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.8908019661903381,
      "learning_rate": 0.0005996638809986021,
      "loss": 2.8807,
      "step": 3472
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.7839584946632385,
      "learning_rate": 0.000599663687389972,
      "loss": 2.6519,
      "step": 3473
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.7076297998428345,
      "learning_rate": 0.0005996634937256287,
      "loss": 2.752,
      "step": 3474
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.7188929915428162,
      "learning_rate": 0.0005996633000055724,
      "loss": 2.7748,
      "step": 3475
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.6704087257385254,
      "learning_rate": 0.000599663106229803,
      "loss": 2.7415,
      "step": 3476
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.8171228170394897,
      "learning_rate": 0.0005996629123983205,
      "loss": 2.6295,
      "step": 3477
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.678619921207428,
      "learning_rate": 0.000599662718511125,
      "loss": 2.8393,
      "step": 3478
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.7374667525291443,
      "learning_rate": 0.0005996625245682166,
      "loss": 2.8171,
      "step": 3479
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.7952606678009033,
      "learning_rate": 0.0005996623305695953,
      "loss": 2.6895,
      "step": 3480
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.7420423626899719,
      "learning_rate": 0.0005996621365152611,
      "loss": 2.7917,
      "step": 3481
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.7698004245758057,
      "learning_rate": 0.000599661942405214,
      "loss": 2.6905,
      "step": 3482
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.6888785362243652,
      "learning_rate": 0.0005996617482394542,
      "loss": 2.9286,
      "step": 3483
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.6989507675170898,
      "learning_rate": 0.0005996615540179815,
      "loss": 2.5798,
      "step": 3484
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.7417728900909424,
      "learning_rate": 0.0005996613597407961,
      "loss": 2.7243,
      "step": 3485
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.7154906988143921,
      "learning_rate": 0.000599661165407898,
      "loss": 2.632,
      "step": 3486
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.7820996642112732,
      "learning_rate": 0.0005996609710192872,
      "loss": 2.5793,
      "step": 3487
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.7383434772491455,
      "learning_rate": 0.0005996607765749637,
      "loss": 2.7592,
      "step": 3488
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.0080546140670776,
      "learning_rate": 0.0005996605820749279,
      "loss": 2.6854,
      "step": 3489
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.7017431855201721,
      "learning_rate": 0.0005996603875191793,
      "loss": 2.4807,
      "step": 3490
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.7530624270439148,
      "learning_rate": 0.0005996601929077181,
      "loss": 2.8073,
      "step": 3491
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.8185667991638184,
      "learning_rate": 0.0005996599982405446,
      "loss": 2.682,
      "step": 3492
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.6591393947601318,
      "learning_rate": 0.0005996598035176587,
      "loss": 2.5899,
      "step": 3493
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.071333408355713,
      "learning_rate": 0.0005996596087390602,
      "loss": 2.7254,
      "step": 3494
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.745639443397522,
      "learning_rate": 0.0005996594139047493,
      "loss": 2.6067,
      "step": 3495
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.862266480922699,
      "learning_rate": 0.0005996592190147264,
      "loss": 2.6239,
      "step": 3496
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.7726739645004272,
      "learning_rate": 0.0005996590240689908,
      "loss": 2.7701,
      "step": 3497
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.643085241317749,
      "learning_rate": 0.0005996588290675431,
      "loss": 2.7844,
      "step": 3498
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.7379307150840759,
      "learning_rate": 0.0005996586340103831,
      "loss": 2.9426,
      "step": 3499
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.8067324161529541,
      "learning_rate": 0.0005996584388975111,
      "loss": 2.7552,
      "step": 3500
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.7503741979598999,
      "learning_rate": 0.0005996582437289268,
      "loss": 2.8277,
      "step": 3501
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.7311037182807922,
      "learning_rate": 0.0005996580485046303,
      "loss": 2.8655,
      "step": 3502
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.7409930229187012,
      "learning_rate": 0.0005996578532246218,
      "loss": 2.8069,
      "step": 3503
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.6931576728820801,
      "learning_rate": 0.0005996576578889012,
      "loss": 2.9846,
      "step": 3504
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.7545197010040283,
      "learning_rate": 0.0005996574624974687,
      "loss": 2.8246,
      "step": 3505
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.751700222492218,
      "learning_rate": 0.0005996572670503241,
      "loss": 2.9154,
      "step": 3506
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.6895864009857178,
      "learning_rate": 0.0005996570715474676,
      "loss": 2.7824,
      "step": 3507
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.8408744931221008,
      "learning_rate": 0.0005996568759888992,
      "loss": 2.7325,
      "step": 3508
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.7122473120689392,
      "learning_rate": 0.0005996566803746189,
      "loss": 2.6401,
      "step": 3509
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.6624438166618347,
      "learning_rate": 0.0005996564847046268,
      "loss": 2.7766,
      "step": 3510
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.7247298955917358,
      "learning_rate": 0.0005996562889789229,
      "loss": 2.9389,
      "step": 3511
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.6557559967041016,
      "learning_rate": 0.0005996560931975072,
      "loss": 2.9302,
      "step": 3512
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.773691713809967,
      "learning_rate": 0.0005996558973603798,
      "loss": 2.806,
      "step": 3513
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.8305358290672302,
      "learning_rate": 0.0005996557014675408,
      "loss": 2.6348,
      "step": 3514
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.7147518396377563,
      "learning_rate": 0.00059965550551899,
      "loss": 2.7342,
      "step": 3515
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.8715436458587646,
      "learning_rate": 0.0005996553095147277,
      "loss": 2.8222,
      "step": 3516
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.7800995707511902,
      "learning_rate": 0.0005996551134547538,
      "loss": 2.3771,
      "step": 3517
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.7557796239852905,
      "learning_rate": 0.0005996549173390682,
      "loss": 2.8904,
      "step": 3518
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.7915821671485901,
      "learning_rate": 0.0005996547211676713,
      "loss": 2.9932,
      "step": 3519
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.7129482626914978,
      "learning_rate": 0.0005996545249405628,
      "loss": 2.9008,
      "step": 3520
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.7094587087631226,
      "learning_rate": 0.000599654328657743,
      "loss": 2.7778,
      "step": 3521
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.9565407037734985,
      "learning_rate": 0.0005996541323192117,
      "loss": 2.8474,
      "step": 3522
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.9202436208724976,
      "learning_rate": 0.000599653935924969,
      "loss": 2.6428,
      "step": 3523
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.6938565969467163,
      "learning_rate": 0.0005996537394750152,
      "loss": 2.6413,
      "step": 3524
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.1952892541885376,
      "learning_rate": 0.0005996535429693498,
      "loss": 2.964,
      "step": 3525
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.7944329977035522,
      "learning_rate": 0.0005996533464079734,
      "loss": 2.9243,
      "step": 3526
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.8735771775245667,
      "learning_rate": 0.0005996531497908857,
      "loss": 2.6236,
      "step": 3527
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.7513441443443298,
      "learning_rate": 0.0005996529531180868,
      "loss": 2.6588,
      "step": 3528
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.711182713508606,
      "learning_rate": 0.0005996527563895767,
      "loss": 2.612,
      "step": 3529
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.7593637108802795,
      "learning_rate": 0.0005996525596053558,
      "loss": 2.7987,
      "step": 3530
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.7656221985816956,
      "learning_rate": 0.0005996523627654235,
      "loss": 2.7996,
      "step": 3531
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.770078182220459,
      "learning_rate": 0.0005996521658697803,
      "loss": 2.6957,
      "step": 3532
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.7665363550186157,
      "learning_rate": 0.000599651968918426,
      "loss": 2.8306,
      "step": 3533
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.7182331085205078,
      "learning_rate": 0.0005996517719113609,
      "loss": 2.8153,
      "step": 3534
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.7116339206695557,
      "learning_rate": 0.0005996515748485848,
      "loss": 2.7502,
      "step": 3535
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.8105201721191406,
      "learning_rate": 0.0005996513777300978,
      "loss": 2.6541,
      "step": 3536
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.693257749080658,
      "learning_rate": 0.0005996511805559,
      "loss": 2.8009,
      "step": 3537
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.8385087251663208,
      "learning_rate": 0.0005996509833259913,
      "loss": 2.6955,
      "step": 3538
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.0544633865356445,
      "learning_rate": 0.0005996507860403719,
      "loss": 2.6496,
      "step": 3539
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.8624138832092285,
      "learning_rate": 0.0005996505886990417,
      "loss": 2.7013,
      "step": 3540
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.9334478974342346,
      "learning_rate": 0.000599650391302001,
      "loss": 2.6904,
      "step": 3541
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.7276901006698608,
      "learning_rate": 0.0005996501938492493,
      "loss": 2.6415,
      "step": 3542
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.8011999726295471,
      "learning_rate": 0.0005996499963407873,
      "loss": 2.8691,
      "step": 3543
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.7950841784477234,
      "learning_rate": 0.0005996497987766145,
      "loss": 2.6592,
      "step": 3544
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.6919822096824646,
      "learning_rate": 0.0005996496011567312,
      "loss": 2.5199,
      "step": 3545
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.743481457233429,
      "learning_rate": 0.0005996494034811374,
      "loss": 2.8801,
      "step": 3546
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.704796552658081,
      "learning_rate": 0.0005996492057498332,
      "loss": 2.855,
      "step": 3547
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.7817076444625854,
      "learning_rate": 0.0005996490079628184,
      "loss": 2.8518,
      "step": 3548
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.6770724654197693,
      "learning_rate": 0.0005996488101200933,
      "loss": 2.7433,
      "step": 3549
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.7841349840164185,
      "learning_rate": 0.0005996486122216579,
      "loss": 2.6254,
      "step": 3550
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.7591677308082581,
      "learning_rate": 0.0005996484142675119,
      "loss": 2.8018,
      "step": 3551
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.6563178300857544,
      "learning_rate": 0.0005996482162576558,
      "loss": 2.6328,
      "step": 3552
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.9161810874938965,
      "learning_rate": 0.0005996480181920895,
      "loss": 2.7222,
      "step": 3553
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.7226253151893616,
      "learning_rate": 0.0005996478200708128,
      "loss": 2.5496,
      "step": 3554
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.761711835861206,
      "learning_rate": 0.0005996476218938261,
      "loss": 3.0558,
      "step": 3555
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.7326072454452515,
      "learning_rate": 0.0005996474236611291,
      "loss": 2.6252,
      "step": 3556
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.8548781275749207,
      "learning_rate": 0.000599647225372722,
      "loss": 2.6963,
      "step": 3557
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.7040917277336121,
      "learning_rate": 0.0005996470270286049,
      "loss": 2.7773,
      "step": 3558
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.7176805138587952,
      "learning_rate": 0.0005996468286287777,
      "loss": 2.714,
      "step": 3559
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.0311869382858276,
      "learning_rate": 0.0005996466301732405,
      "loss": 2.7576,
      "step": 3560
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.7114015817642212,
      "learning_rate": 0.0005996464316619934,
      "loss": 2.591,
      "step": 3561
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.9112000465393066,
      "learning_rate": 0.0005996462330950363,
      "loss": 2.6455,
      "step": 3562
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.9116586446762085,
      "learning_rate": 0.0005996460344723693,
      "loss": 2.6574,
      "step": 3563
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.7209317684173584,
      "learning_rate": 0.0005996458357939925,
      "loss": 2.6748,
      "step": 3564
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.9952433109283447,
      "learning_rate": 0.0005996456370599059,
      "loss": 2.644,
      "step": 3565
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.7315688133239746,
      "learning_rate": 0.0005996454382701094,
      "loss": 2.6918,
      "step": 3566
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.7192685604095459,
      "learning_rate": 0.0005996452394246033,
      "loss": 2.5348,
      "step": 3567
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.7455431818962097,
      "learning_rate": 0.0005996450405233875,
      "loss": 2.5324,
      "step": 3568
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.8104442358016968,
      "learning_rate": 0.0005996448415664619,
      "loss": 2.3963,
      "step": 3569
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.7253889441490173,
      "learning_rate": 0.0005996446425538267,
      "loss": 2.752,
      "step": 3570
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.7188587784767151,
      "learning_rate": 0.0005996444434854819,
      "loss": 2.8843,
      "step": 3571
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.8728600144386292,
      "learning_rate": 0.0005996442443614277,
      "loss": 2.7886,
      "step": 3572
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.6991623640060425,
      "learning_rate": 0.0005996440451816638,
      "loss": 2.8299,
      "step": 3573
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.7279838919639587,
      "learning_rate": 0.0005996438459461906,
      "loss": 2.5819,
      "step": 3574
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.7507889270782471,
      "learning_rate": 0.0005996436466550077,
      "loss": 2.8243,
      "step": 3575
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.6913896799087524,
      "learning_rate": 0.0005996434473081155,
      "loss": 2.6452,
      "step": 3576
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.7216665744781494,
      "learning_rate": 0.000599643247905514,
      "loss": 2.7175,
      "step": 3577
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.6101269125938416,
      "learning_rate": 0.0005996430484472031,
      "loss": 2.7374,
      "step": 3578
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.6749204397201538,
      "learning_rate": 0.0005996428489331829,
      "loss": 2.5138,
      "step": 3579
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.7590287923812866,
      "learning_rate": 0.0005996426493634535,
      "loss": 2.8751,
      "step": 3580
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.6512974500656128,
      "learning_rate": 0.0005996424497380148,
      "loss": 2.7832,
      "step": 3581
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.7537881135940552,
      "learning_rate": 0.000599642250056867,
      "loss": 2.9124,
      "step": 3582
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.6853196620941162,
      "learning_rate": 0.0005996420503200099,
      "loss": 2.3433,
      "step": 3583
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.7405056953430176,
      "learning_rate": 0.0005996418505274438,
      "loss": 2.7993,
      "step": 3584
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.6880627870559692,
      "learning_rate": 0.0005996416506791686,
      "loss": 2.8377,
      "step": 3585
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.8591498136520386,
      "learning_rate": 0.0005996414507751844,
      "loss": 2.5237,
      "step": 3586
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.7348811626434326,
      "learning_rate": 0.0005996412508154911,
      "loss": 2.4337,
      "step": 3587
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.6916965842247009,
      "learning_rate": 0.0005996410508000889,
      "loss": 2.7128,
      "step": 3588
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.7382058501243591,
      "learning_rate": 0.0005996408507289777,
      "loss": 2.8278,
      "step": 3589
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.7075006365776062,
      "learning_rate": 0.0005996406506021578,
      "loss": 2.6349,
      "step": 3590
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.6792402863502502,
      "learning_rate": 0.0005996404504196288,
      "loss": 2.7444,
      "step": 3591
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.9342615604400635,
      "learning_rate": 0.0005996402501813911,
      "loss": 2.5027,
      "step": 3592
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.7025355100631714,
      "learning_rate": 0.0005996400498874447,
      "loss": 2.7053,
      "step": 3593
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.6731183528900146,
      "learning_rate": 0.0005996398495377894,
      "loss": 2.7173,
      "step": 3594
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.7230398058891296,
      "learning_rate": 0.0005996396491324256,
      "loss": 2.983,
      "step": 3595
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.8635936379432678,
      "learning_rate": 0.000599639448671353,
      "loss": 2.7357,
      "step": 3596
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.9360721707344055,
      "learning_rate": 0.0005996392481545718,
      "loss": 2.7847,
      "step": 3597
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.7632573843002319,
      "learning_rate": 0.000599639047582082,
      "loss": 2.7866,
      "step": 3598
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.8140075206756592,
      "learning_rate": 0.0005996388469538835,
      "loss": 2.7321,
      "step": 3599
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.7778942584991455,
      "learning_rate": 0.0005996386462699767,
      "loss": 2.685,
      "step": 3600
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.662981390953064,
      "learning_rate": 0.0005996384455303613,
      "loss": 2.5119,
      "step": 3601
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.8907694816589355,
      "learning_rate": 0.0005996382447350376,
      "loss": 2.612,
      "step": 3602
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.7928252816200256,
      "learning_rate": 0.0005996380438840054,
      "loss": 2.5146,
      "step": 3603
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.7703949213027954,
      "learning_rate": 0.0005996378429772648,
      "loss": 2.6147,
      "step": 3604
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.7270202040672302,
      "learning_rate": 0.0005996376420148159,
      "loss": 2.9855,
      "step": 3605
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.789723813533783,
      "learning_rate": 0.0005996374409966587,
      "loss": 2.5352,
      "step": 3606
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.6653648018836975,
      "learning_rate": 0.0005996372399227931,
      "loss": 2.7668,
      "step": 3607
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.7978569269180298,
      "learning_rate": 0.0005996370387932194,
      "loss": 2.5944,
      "step": 3608
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.7240025401115417,
      "learning_rate": 0.0005996368376079376,
      "loss": 2.562,
      "step": 3609
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.6844765543937683,
      "learning_rate": 0.0005996366363669476,
      "loss": 2.761,
      "step": 3610
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.6929038763046265,
      "learning_rate": 0.0005996364350702495,
      "loss": 2.6345,
      "step": 3611
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.8759015798568726,
      "learning_rate": 0.0005996362337178433,
      "loss": 2.689,
      "step": 3612
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.6495019197463989,
      "learning_rate": 0.0005996360323097291,
      "loss": 2.7116,
      "step": 3613
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.6695153117179871,
      "learning_rate": 0.0005996358308459068,
      "loss": 2.8272,
      "step": 3614
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.8216586112976074,
      "learning_rate": 0.0005996356293263767,
      "loss": 2.7218,
      "step": 3615
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.7663249373435974,
      "learning_rate": 0.0005996354277511386,
      "loss": 2.6912,
      "step": 3616
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.7614197731018066,
      "learning_rate": 0.0005996352261201926,
      "loss": 2.7805,
      "step": 3617
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.7130672931671143,
      "learning_rate": 0.0005996350244335389,
      "loss": 2.6077,
      "step": 3618
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.7582953572273254,
      "learning_rate": 0.0005996348226911773,
      "loss": 2.9614,
      "step": 3619
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.8557789325714111,
      "learning_rate": 0.0005996346208931078,
      "loss": 2.7057,
      "step": 3620
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.8390859961509705,
      "learning_rate": 0.0005996344190393307,
      "loss": 2.7918,
      "step": 3621
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.751508891582489,
      "learning_rate": 0.0005996342171298459,
      "loss": 2.643,
      "step": 3622
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.7133805155754089,
      "learning_rate": 0.0005996340151646534,
      "loss": 2.7756,
      "step": 3623
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.7090493440628052,
      "learning_rate": 0.0005996338131437533,
      "loss": 2.7132,
      "step": 3624
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.9071016311645508,
      "learning_rate": 0.0005996336110671456,
      "loss": 2.6994,
      "step": 3625
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.7164561152458191,
      "learning_rate": 0.0005996334089348305,
      "loss": 2.7264,
      "step": 3626
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.7556436061859131,
      "learning_rate": 0.0005996332067468077,
      "loss": 2.904,
      "step": 3627
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.8680242896080017,
      "learning_rate": 0.0005996330045030775,
      "loss": 2.9115,
      "step": 3628
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.7385838031768799,
      "learning_rate": 0.0005996328022036399,
      "loss": 2.7176,
      "step": 3629
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.8088661432266235,
      "learning_rate": 0.0005996325998484949,
      "loss": 2.9181,
      "step": 3630
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.8892428874969482,
      "learning_rate": 0.0005996323974376425,
      "loss": 2.6403,
      "step": 3631
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.7504913210868835,
      "learning_rate": 0.0005996321949710828,
      "loss": 2.8444,
      "step": 3632
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.6693288683891296,
      "learning_rate": 0.0005996319924488158,
      "loss": 2.8347,
      "step": 3633
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.775115966796875,
      "learning_rate": 0.0005996317898708416,
      "loss": 2.6651,
      "step": 3634
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.6223077178001404,
      "learning_rate": 0.0005996315872371601,
      "loss": 2.615,
      "step": 3635
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.6647838950157166,
      "learning_rate": 0.0005996313845477715,
      "loss": 2.7616,
      "step": 3636
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.8722261786460876,
      "learning_rate": 0.0005996311818026759,
      "loss": 2.8866,
      "step": 3637
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.7702917456626892,
      "learning_rate": 0.0005996309790018731,
      "loss": 2.8397,
      "step": 3638
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.7391378879547119,
      "learning_rate": 0.0005996307761453632,
      "loss": 2.6871,
      "step": 3639
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.6436600685119629,
      "learning_rate": 0.0005996305732331464,
      "loss": 2.4139,
      "step": 3640
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.691157341003418,
      "learning_rate": 0.0005996303702652224,
      "loss": 2.637,
      "step": 3641
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.690938413143158,
      "learning_rate": 0.0005996301672415916,
      "loss": 2.6179,
      "step": 3642
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.7247976064682007,
      "learning_rate": 0.0005996299641622539,
      "loss": 2.8284,
      "step": 3643
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.6865954399108887,
      "learning_rate": 0.0005996297610272093,
      "loss": 2.5738,
      "step": 3644
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.8454104661941528,
      "learning_rate": 0.000599629557836458,
      "loss": 2.5624,
      "step": 3645
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.7379743456840515,
      "learning_rate": 0.0005996293545899998,
      "loss": 2.655,
      "step": 3646
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.7578339576721191,
      "learning_rate": 0.0005996291512878348,
      "loss": 2.6884,
      "step": 3647
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.7370554208755493,
      "learning_rate": 0.0005996289479299631,
      "loss": 2.7703,
      "step": 3648
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.7856513261795044,
      "learning_rate": 0.0005996287445163847,
      "loss": 3.0129,
      "step": 3649
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.6992132067680359,
      "learning_rate": 0.0005996285410470999,
      "loss": 2.7659,
      "step": 3650
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.8609179854393005,
      "learning_rate": 0.0005996283375221082,
      "loss": 2.7942,
      "step": 3651
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.7599971294403076,
      "learning_rate": 0.00059962813394141,
      "loss": 2.7656,
      "step": 3652
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.726553201675415,
      "learning_rate": 0.0005996279303050054,
      "loss": 2.7156,
      "step": 3653
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.8295498490333557,
      "learning_rate": 0.0005996277266128942,
      "loss": 2.6874,
      "step": 3654
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.6869130730628967,
      "learning_rate": 0.0005996275228650766,
      "loss": 2.8242,
      "step": 3655
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.7148303985595703,
      "learning_rate": 0.0005996273190615525,
      "loss": 2.752,
      "step": 3656
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.628729522228241,
      "learning_rate": 0.0005996271152023222,
      "loss": 2.6089,
      "step": 3657
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.8360553979873657,
      "learning_rate": 0.0005996269112873854,
      "loss": 2.3675,
      "step": 3658
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.651212751865387,
      "learning_rate": 0.0005996267073167424,
      "loss": 2.8391,
      "step": 3659
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.7297361493110657,
      "learning_rate": 0.0005996265032903931,
      "loss": 2.795,
      "step": 3660
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.6548327803611755,
      "learning_rate": 0.0005996262992083376,
      "loss": 2.85,
      "step": 3661
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.6843042969703674,
      "learning_rate": 0.0005996260950705759,
      "loss": 2.7766,
      "step": 3662
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.7329358458518982,
      "learning_rate": 0.0005996258908771081,
      "loss": 2.6175,
      "step": 3663
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.7034987211227417,
      "learning_rate": 0.0005996256866279341,
      "loss": 2.6208,
      "step": 3664
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.6730404496192932,
      "learning_rate": 0.000599625482323054,
      "loss": 2.7776,
      "step": 3665
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.6987662315368652,
      "learning_rate": 0.000599625277962468,
      "loss": 2.5356,
      "step": 3666
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.6912817358970642,
      "learning_rate": 0.0005996250735461759,
      "loss": 2.9828,
      "step": 3667
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.7052863240242004,
      "learning_rate": 0.000599624869074178,
      "loss": 2.9568,
      "step": 3668
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.7494381070137024,
      "learning_rate": 0.000599624664546474,
      "loss": 2.7382,
      "step": 3669
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.7878034114837646,
      "learning_rate": 0.0005996244599630642,
      "loss": 2.8148,
      "step": 3670
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.6666281819343567,
      "learning_rate": 0.0005996242553239485,
      "loss": 2.6386,
      "step": 3671
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.6668806672096252,
      "learning_rate": 0.0005996240506291271,
      "loss": 2.7447,
      "step": 3672
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.7983410954475403,
      "learning_rate": 0.0005996238458785998,
      "loss": 2.7541,
      "step": 3673
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.6533145308494568,
      "learning_rate": 0.0005996236410723668,
      "loss": 2.7425,
      "step": 3674
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.6828229427337646,
      "learning_rate": 0.0005996234362104282,
      "loss": 2.7406,
      "step": 3675
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.7606797218322754,
      "learning_rate": 0.0005996232312927838,
      "loss": 2.5738,
      "step": 3676
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.7256486415863037,
      "learning_rate": 0.000599623026319434,
      "loss": 2.8121,
      "step": 3677
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.6700488924980164,
      "learning_rate": 0.0005996228212903785,
      "loss": 2.8503,
      "step": 3678
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.8016119003295898,
      "learning_rate": 0.0005996226162056174,
      "loss": 2.758,
      "step": 3679
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.6302990913391113,
      "learning_rate": 0.0005996224110651508,
      "loss": 2.5675,
      "step": 3680
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.7066866159439087,
      "learning_rate": 0.0005996222058689789,
      "loss": 2.6496,
      "step": 3681
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.6681373715400696,
      "learning_rate": 0.0005996220006171015,
      "loss": 2.5753,
      "step": 3682
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.7258835434913635,
      "learning_rate": 0.0005996217953095186,
      "loss": 2.6989,
      "step": 3683
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.7029567360877991,
      "learning_rate": 0.0005996215899462305,
      "loss": 2.5392,
      "step": 3684
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.7323451042175293,
      "learning_rate": 0.0005996213845272371,
      "loss": 2.8693,
      "step": 3685
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.7913175225257874,
      "learning_rate": 0.0005996211790525384,
      "loss": 2.809,
      "step": 3686
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.7050085663795471,
      "learning_rate": 0.0005996209735221344,
      "loss": 2.617,
      "step": 3687
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.7651745676994324,
      "learning_rate": 0.0005996207679360252,
      "loss": 2.5618,
      "step": 3688
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.7450152039527893,
      "learning_rate": 0.0005996205622942109,
      "loss": 2.6064,
      "step": 3689
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.736870288848877,
      "learning_rate": 0.0005996203565966914,
      "loss": 2.9195,
      "step": 3690
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.656040370464325,
      "learning_rate": 0.0005996201508434669,
      "loss": 2.5385,
      "step": 3691
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.6838341355323792,
      "learning_rate": 0.0005996199450345374,
      "loss": 2.9652,
      "step": 3692
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.660307765007019,
      "learning_rate": 0.0005996197391699028,
      "loss": 2.7675,
      "step": 3693
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.656873881816864,
      "learning_rate": 0.0005996195332495633,
      "loss": 2.5045,
      "step": 3694
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.7205382585525513,
      "learning_rate": 0.0005996193272735189,
      "loss": 2.8402,
      "step": 3695
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.7023299336433411,
      "learning_rate": 0.0005996191212417696,
      "loss": 2.8296,
      "step": 3696
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.7594283223152161,
      "learning_rate": 0.0005996189151543153,
      "loss": 2.594,
      "step": 3697
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.7716628313064575,
      "learning_rate": 0.0005996187090111563,
      "loss": 2.7001,
      "step": 3698
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.765172004699707,
      "learning_rate": 0.0005996185028122926,
      "loss": 2.8545,
      "step": 3699
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.7340953946113586,
      "learning_rate": 0.000599618296557724,
      "loss": 2.6469,
      "step": 3700
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.7107803821563721,
      "learning_rate": 0.0005996180902474508,
      "loss": 2.5375,
      "step": 3701
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.7235217690467834,
      "learning_rate": 0.0005996178838814729,
      "loss": 2.6908,
      "step": 3702
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.6676885485649109,
      "learning_rate": 0.0005996176774597905,
      "loss": 2.9478,
      "step": 3703
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.7574462294578552,
      "learning_rate": 0.0005996174709824033,
      "loss": 2.7421,
      "step": 3704
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.6661620140075684,
      "learning_rate": 0.0005996172644493117,
      "loss": 2.9614,
      "step": 3705
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.7353447079658508,
      "learning_rate": 0.0005996170578605156,
      "loss": 2.68,
      "step": 3706
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.6814978718757629,
      "learning_rate": 0.0005996168512160151,
      "loss": 2.6434,
      "step": 3707
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.6905335187911987,
      "learning_rate": 0.00059961664451581,
      "loss": 2.6764,
      "step": 3708
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.6669712662696838,
      "learning_rate": 0.0005996164377599007,
      "loss": 2.5107,
      "step": 3709
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.7132564783096313,
      "learning_rate": 0.0005996162309482868,
      "loss": 2.602,
      "step": 3710
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.7246723771095276,
      "learning_rate": 0.0005996160240809687,
      "loss": 2.8645,
      "step": 3711
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.6734629273414612,
      "learning_rate": 0.0005996158171579465,
      "loss": 2.7285,
      "step": 3712
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.7032813429832458,
      "learning_rate": 0.0005996156101792199,
      "loss": 2.8452,
      "step": 3713
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.7212178707122803,
      "learning_rate": 0.0005996154031447891,
      "loss": 2.7184,
      "step": 3714
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.6622939705848694,
      "learning_rate": 0.0005996151960546542,
      "loss": 2.6971,
      "step": 3715
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.6543493270874023,
      "learning_rate": 0.0005996149889088151,
      "loss": 2.4517,
      "step": 3716
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.8150997161865234,
      "learning_rate": 0.000599614781707272,
      "loss": 2.9026,
      "step": 3717
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.7011685967445374,
      "learning_rate": 0.0005996145744500248,
      "loss": 3.2052,
      "step": 3718
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.9046491980552673,
      "learning_rate": 0.0005996143671370736,
      "loss": 2.3935,
      "step": 3719
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.6686937212944031,
      "learning_rate": 0.0005996141597684185,
      "loss": 2.7359,
      "step": 3720
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.7937058806419373,
      "learning_rate": 0.0005996139523440594,
      "loss": 2.7213,
      "step": 3721
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.6485825777053833,
      "learning_rate": 0.0005996137448639965,
      "loss": 2.5694,
      "step": 3722
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.7419896125793457,
      "learning_rate": 0.0005996135373282296,
      "loss": 2.5836,
      "step": 3723
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.9343651533126831,
      "learning_rate": 0.0005996133297367589,
      "loss": 2.6648,
      "step": 3724
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.6925092339515686,
      "learning_rate": 0.0005996131220895846,
      "loss": 2.6164,
      "step": 3725
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.6757372617721558,
      "learning_rate": 0.0005996129143867065,
      "loss": 2.7701,
      "step": 3726
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.6857375502586365,
      "learning_rate": 0.0005996127066281245,
      "loss": 2.5738,
      "step": 3727
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.789465069770813,
      "learning_rate": 0.0005996124988138391,
      "loss": 2.4614,
      "step": 3728
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.7149078249931335,
      "learning_rate": 0.00059961229094385,
      "loss": 2.5771,
      "step": 3729
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.7603667378425598,
      "learning_rate": 0.0005996120830181573,
      "loss": 2.7831,
      "step": 3730
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.6565271615982056,
      "learning_rate": 0.0005996118750367609,
      "loss": 2.7262,
      "step": 3731
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.730460524559021,
      "learning_rate": 0.0005996116669996613,
      "loss": 2.6376,
      "step": 3732
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.7210078239440918,
      "learning_rate": 0.000599611458906858,
      "loss": 2.5655,
      "step": 3733
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.6321623921394348,
      "learning_rate": 0.0005996112507583514,
      "loss": 2.7318,
      "step": 3734
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.7205336689949036,
      "learning_rate": 0.0005996110425541414,
      "loss": 2.3806,
      "step": 3735
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.6797001957893372,
      "learning_rate": 0.000599610834294228,
      "loss": 2.5594,
      "step": 3736
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.7904689908027649,
      "learning_rate": 0.0005996106259786114,
      "loss": 2.9582,
      "step": 3737
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.7604937553405762,
      "learning_rate": 0.0005996104176072914,
      "loss": 2.9281,
      "step": 3738
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.7580645084381104,
      "learning_rate": 0.0005996102091802683,
      "loss": 2.8721,
      "step": 3739
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.817680835723877,
      "learning_rate": 0.0005996100006975418,
      "loss": 2.4715,
      "step": 3740
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.6803327202796936,
      "learning_rate": 0.0005996097921591124,
      "loss": 2.6521,
      "step": 3741
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.962755024433136,
      "learning_rate": 0.0005996095835649796,
      "loss": 2.7232,
      "step": 3742
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.6877244114875793,
      "learning_rate": 0.0005996093749151439,
      "loss": 2.585,
      "step": 3743
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.7923077344894409,
      "learning_rate": 0.0005996091662096052,
      "loss": 2.4882,
      "step": 3744
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.7939332127571106,
      "learning_rate": 0.0005996089574483634,
      "loss": 2.6295,
      "step": 3745
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.7719216346740723,
      "learning_rate": 0.0005996087486314187,
      "loss": 2.923,
      "step": 3746
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.7529695630073547,
      "learning_rate": 0.000599608539758771,
      "loss": 2.8259,
      "step": 3747
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.7365591526031494,
      "learning_rate": 0.0005996083308304205,
      "loss": 2.9229,
      "step": 3748
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.7047050595283508,
      "learning_rate": 0.0005996081218463672,
      "loss": 2.5997,
      "step": 3749
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.7201966643333435,
      "learning_rate": 0.000599607912806611,
      "loss": 2.6735,
      "step": 3750
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.7250720858573914,
      "learning_rate": 0.000599607703711152,
      "loss": 2.8834,
      "step": 3751
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.6248542070388794,
      "learning_rate": 0.0005996074945599903,
      "loss": 2.7814,
      "step": 3752
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.7316437363624573,
      "learning_rate": 0.000599607285353126,
      "loss": 2.6579,
      "step": 3753
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.7439720034599304,
      "learning_rate": 0.000599607076090559,
      "loss": 2.5919,
      "step": 3754
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.7075766324996948,
      "learning_rate": 0.0005996068667722894,
      "loss": 2.6557,
      "step": 3755
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.9252955317497253,
      "learning_rate": 0.0005996066573983172,
      "loss": 2.65,
      "step": 3756
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.6472856402397156,
      "learning_rate": 0.0005996064479686425,
      "loss": 2.5825,
      "step": 3757
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.8354807496070862,
      "learning_rate": 0.0005996062384832653,
      "loss": 2.7986,
      "step": 3758
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.8211201429367065,
      "learning_rate": 0.0005996060289421856,
      "loss": 2.6665,
      "step": 3759
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.802625834941864,
      "learning_rate": 0.0005996058193454035,
      "loss": 2.4727,
      "step": 3760
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.8636558651924133,
      "learning_rate": 0.000599605609692919,
      "loss": 2.7355,
      "step": 3761
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.7329061627388,
      "learning_rate": 0.0005996053999847322,
      "loss": 2.6584,
      "step": 3762
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.7966116070747375,
      "learning_rate": 0.0005996051902208431,
      "loss": 2.6429,
      "step": 3763
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.7991119623184204,
      "learning_rate": 0.0005996049804012517,
      "loss": 2.8053,
      "step": 3764
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.7145423889160156,
      "learning_rate": 0.0005996047705259581,
      "loss": 2.7285,
      "step": 3765
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.7841779589653015,
      "learning_rate": 0.0005996045605949624,
      "loss": 2.7195,
      "step": 3766
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.8137551546096802,
      "learning_rate": 0.0005996043506082644,
      "loss": 2.4533,
      "step": 3767
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.9273861646652222,
      "learning_rate": 0.0005996041405658644,
      "loss": 2.5396,
      "step": 3768
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.8340867757797241,
      "learning_rate": 0.0005996039304677624,
      "loss": 2.5692,
      "step": 3769
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.7287812232971191,
      "learning_rate": 0.0005996037203139582,
      "loss": 2.5837,
      "step": 3770
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.7087157964706421,
      "learning_rate": 0.0005996035101044522,
      "loss": 2.5101,
      "step": 3771
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.7207835912704468,
      "learning_rate": 0.0005996032998392441,
      "loss": 2.9591,
      "step": 3772
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.9557005763053894,
      "learning_rate": 0.0005996030895183342,
      "loss": 2.7009,
      "step": 3773
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.7231683731079102,
      "learning_rate": 0.0005996028791417224,
      "loss": 2.7242,
      "step": 3774
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.6974450349807739,
      "learning_rate": 0.0005996026687094087,
      "loss": 2.95,
      "step": 3775
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.8662458062171936,
      "learning_rate": 0.0005996024582213932,
      "loss": 2.7403,
      "step": 3776
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.6696552038192749,
      "learning_rate": 0.0005996022476776761,
      "loss": 2.5121,
      "step": 3777
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.8573384284973145,
      "learning_rate": 0.0005996020370782571,
      "loss": 2.6782,
      "step": 3778
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.6973930597305298,
      "learning_rate": 0.0005996018264231366,
      "loss": 2.8168,
      "step": 3779
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.7930920124053955,
      "learning_rate": 0.0005996016157123143,
      "loss": 2.8963,
      "step": 3780
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.7512115836143494,
      "learning_rate": 0.0005996014049457905,
      "loss": 2.7903,
      "step": 3781
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.6823214292526245,
      "learning_rate": 0.0005996011941235652,
      "loss": 2.8268,
      "step": 3782
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.709335207939148,
      "learning_rate": 0.0005996009832456383,
      "loss": 2.8739,
      "step": 3783
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.9233169555664062,
      "learning_rate": 0.00059960077231201,
      "loss": 2.3716,
      "step": 3784
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.7599151134490967,
      "learning_rate": 0.0005996005613226801,
      "loss": 2.9123,
      "step": 3785
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.9346723556518555,
      "learning_rate": 0.0005996003502776488,
      "loss": 2.7921,
      "step": 3786
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.871523380279541,
      "learning_rate": 0.0005996001391769163,
      "loss": 2.6683,
      "step": 3787
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.7107419371604919,
      "learning_rate": 0.0005995999280204824,
      "loss": 2.6115,
      "step": 3788
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.0507820844650269,
      "learning_rate": 0.0005995997168083473,
      "loss": 2.5535,
      "step": 3789
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.7684757113456726,
      "learning_rate": 0.0005995995055405109,
      "loss": 2.8751,
      "step": 3790
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.7141135334968567,
      "learning_rate": 0.0005995992942169732,
      "loss": 2.861,
      "step": 3791
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.0163037776947021,
      "learning_rate": 0.0005995990828377344,
      "loss": 2.8212,
      "step": 3792
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.7716142535209656,
      "learning_rate": 0.0005995988714027946,
      "loss": 2.8417,
      "step": 3793
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.7115422487258911,
      "learning_rate": 0.0005995986599121536,
      "loss": 2.7073,
      "step": 3794
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.0203064680099487,
      "learning_rate": 0.0005995984483658115,
      "loss": 2.7445,
      "step": 3795
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.6750436425209045,
      "learning_rate": 0.0005995982367637685,
      "loss": 2.764,
      "step": 3796
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.9427126049995422,
      "learning_rate": 0.0005995980251060245,
      "loss": 2.5358,
      "step": 3797
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.009997010231018,
      "learning_rate": 0.0005995978133925795,
      "loss": 2.7982,
      "step": 3798
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.7558155059814453,
      "learning_rate": 0.0005995976016234337,
      "loss": 2.4787,
      "step": 3799
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.8596630096435547,
      "learning_rate": 0.0005995973897985871,
      "loss": 2.713,
      "step": 3800
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.8190658688545227,
      "learning_rate": 0.0005995971779180396,
      "loss": 2.6762,
      "step": 3801
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.842079758644104,
      "learning_rate": 0.0005995969659817913,
      "loss": 2.65,
      "step": 3802
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.649359405040741,
      "learning_rate": 0.0005995967539898422,
      "loss": 2.7421,
      "step": 3803
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.8248977065086365,
      "learning_rate": 0.0005995965419421926,
      "loss": 2.7902,
      "step": 3804
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.6827329993247986,
      "learning_rate": 0.0005995963298388423,
      "loss": 2.7912,
      "step": 3805
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.7969716191291809,
      "learning_rate": 0.0005995961176797914,
      "loss": 2.8497,
      "step": 3806
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.9021100997924805,
      "learning_rate": 0.0005995959054650398,
      "loss": 2.7667,
      "step": 3807
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.7668478488922119,
      "learning_rate": 0.0005995956931945877,
      "loss": 2.7969,
      "step": 3808
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.6493473649024963,
      "learning_rate": 0.0005995954808684352,
      "loss": 2.8573,
      "step": 3809
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.9154091477394104,
      "learning_rate": 0.0005995952684865822,
      "loss": 2.5442,
      "step": 3810
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.6889321804046631,
      "learning_rate": 0.0005995950560490288,
      "loss": 2.815,
      "step": 3811
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.691699743270874,
      "learning_rate": 0.000599594843555775,
      "loss": 2.671,
      "step": 3812
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.8047608137130737,
      "learning_rate": 0.0005995946310068209,
      "loss": 2.6934,
      "step": 3813
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.7021161913871765,
      "learning_rate": 0.0005995944184021665,
      "loss": 2.3917,
      "step": 3814
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.7872621417045593,
      "learning_rate": 0.0005995942057418118,
      "loss": 2.8007,
      "step": 3815
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.7477960586547852,
      "learning_rate": 0.0005995939930257568,
      "loss": 2.5552,
      "step": 3816
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.7991460561752319,
      "learning_rate": 0.0005995937802540018,
      "loss": 2.7624,
      "step": 3817
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.6942827701568604,
      "learning_rate": 0.0005995935674265465,
      "loss": 2.7234,
      "step": 3818
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.7131285667419434,
      "learning_rate": 0.0005995933545433912,
      "loss": 2.6319,
      "step": 3819
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.7119014263153076,
      "learning_rate": 0.0005995931416045358,
      "loss": 2.9212,
      "step": 3820
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.7074767351150513,
      "learning_rate": 0.0005995929286099805,
      "loss": 2.6178,
      "step": 3821
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.7183850407600403,
      "learning_rate": 0.000599592715559725,
      "loss": 2.727,
      "step": 3822
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.7605533003807068,
      "learning_rate": 0.0005995925024537697,
      "loss": 2.8115,
      "step": 3823
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.6960142850875854,
      "learning_rate": 0.0005995922892921145,
      "loss": 2.7413,
      "step": 3824
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.7759429216384888,
      "learning_rate": 0.0005995920760747594,
      "loss": 2.6266,
      "step": 3825
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.7116156220436096,
      "learning_rate": 0.0005995918628017044,
      "loss": 2.6503,
      "step": 3826
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.7690176367759705,
      "learning_rate": 0.0005995916494729497,
      "loss": 2.8398,
      "step": 3827
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.6796813607215881,
      "learning_rate": 0.0005995914360884954,
      "loss": 2.7452,
      "step": 3828
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.6789624094963074,
      "learning_rate": 0.0005995912226483412,
      "loss": 2.7131,
      "step": 3829
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.7262357473373413,
      "learning_rate": 0.0005995910091524874,
      "loss": 2.7076,
      "step": 3830
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.7063474655151367,
      "learning_rate": 0.000599590795600934,
      "loss": 2.5465,
      "step": 3831
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.6887843012809753,
      "learning_rate": 0.000599590581993681,
      "loss": 2.8298,
      "step": 3832
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.6951830983161926,
      "learning_rate": 0.0005995903683307283,
      "loss": 2.6448,
      "step": 3833
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.7111455202102661,
      "learning_rate": 0.0005995901546120764,
      "loss": 2.7089,
      "step": 3834
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.6898537278175354,
      "learning_rate": 0.0005995899408377249,
      "loss": 2.8296,
      "step": 3835
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.74686598777771,
      "learning_rate": 0.0005995897270076739,
      "loss": 2.385,
      "step": 3836
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.7284538745880127,
      "learning_rate": 0.0005995895131219235,
      "loss": 2.7648,
      "step": 3837
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.7760813236236572,
      "learning_rate": 0.0005995892991804739,
      "loss": 2.6688,
      "step": 3838
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.7080743908882141,
      "learning_rate": 0.0005995890851833249,
      "loss": 2.4771,
      "step": 3839
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.6920950412750244,
      "learning_rate": 0.0005995888711304766,
      "loss": 2.5585,
      "step": 3840
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.6811289191246033,
      "learning_rate": 0.0005995886570219291,
      "loss": 2.5906,
      "step": 3841
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.7381570935249329,
      "learning_rate": 0.0005995884428576825,
      "loss": 2.7861,
      "step": 3842
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.6948132514953613,
      "learning_rate": 0.0005995882286377366,
      "loss": 2.6986,
      "step": 3843
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.7532497048377991,
      "learning_rate": 0.0005995880143620918,
      "loss": 2.7402,
      "step": 3844
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.7920396327972412,
      "learning_rate": 0.0005995878000307478,
      "loss": 2.7998,
      "step": 3845
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.6779424548149109,
      "learning_rate": 0.0005995875856437048,
      "loss": 2.6943,
      "step": 3846
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.9302729368209839,
      "learning_rate": 0.0005995873712009627,
      "loss": 2.8063,
      "step": 3847
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.6272329092025757,
      "learning_rate": 0.0005995871567025218,
      "loss": 2.4742,
      "step": 3848
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.043601155281067,
      "learning_rate": 0.0005995869421483819,
      "loss": 2.8121,
      "step": 3849
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.7175539135932922,
      "learning_rate": 0.0005995867275385431,
      "loss": 2.7263,
      "step": 3850
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.8092250227928162,
      "learning_rate": 0.0005995865128730057,
      "loss": 2.5878,
      "step": 3851
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.7377088069915771,
      "learning_rate": 0.0005995862981517693,
      "loss": 2.7135,
      "step": 3852
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.8799555897712708,
      "learning_rate": 0.0005995860833748343,
      "loss": 2.4852,
      "step": 3853
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.7166706323623657,
      "learning_rate": 0.0005995858685422004,
      "loss": 2.5937,
      "step": 3854
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.6651005148887634,
      "learning_rate": 0.000599585653653868,
      "loss": 2.9941,
      "step": 3855
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.8866887092590332,
      "learning_rate": 0.0005995854387098369,
      "loss": 2.8375,
      "step": 3856
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.7587459087371826,
      "learning_rate": 0.0005995852237101072,
      "loss": 2.7073,
      "step": 3857
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.9356045126914978,
      "learning_rate": 0.000599585008654679,
      "loss": 2.6501,
      "step": 3858
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.7166044116020203,
      "learning_rate": 0.0005995847935435522,
      "loss": 2.6851,
      "step": 3859
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.672088623046875,
      "learning_rate": 0.0005995845783767271,
      "loss": 2.6367,
      "step": 3860
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.7280047535896301,
      "learning_rate": 0.0005995843631542034,
      "loss": 2.742,
      "step": 3861
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.7265699505805969,
      "learning_rate": 0.0005995841478759813,
      "loss": 2.7457,
      "step": 3862
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.7898707985877991,
      "learning_rate": 0.000599583932542061,
      "loss": 2.7915,
      "step": 3863
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.0315049886703491,
      "learning_rate": 0.0005995837171524424,
      "loss": 2.7611,
      "step": 3864
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.7214407920837402,
      "learning_rate": 0.0005995835017071254,
      "loss": 2.8538,
      "step": 3865
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.065806269645691,
      "learning_rate": 0.0005995832862061102,
      "loss": 2.6754,
      "step": 3866
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.7166692018508911,
      "learning_rate": 0.0005995830706493969,
      "loss": 2.7833,
      "step": 3867
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.7010578513145447,
      "learning_rate": 0.0005995828550369853,
      "loss": 2.6977,
      "step": 3868
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.8578182458877563,
      "learning_rate": 0.0005995826393688757,
      "loss": 2.838,
      "step": 3869
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.9043206572532654,
      "learning_rate": 0.0005995824236450681,
      "loss": 2.6252,
      "step": 3870
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.8323679566383362,
      "learning_rate": 0.0005995822078655624,
      "loss": 2.6742,
      "step": 3871
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.9971485733985901,
      "learning_rate": 0.0005995819920303586,
      "loss": 2.6519,
      "step": 3872
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.7906187772750854,
      "learning_rate": 0.0005995817761394569,
      "loss": 2.8006,
      "step": 3873
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.790075421333313,
      "learning_rate": 0.0005995815601928575,
      "loss": 2.9274,
      "step": 3874
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.097018837928772,
      "learning_rate": 0.00059958134419056,
      "loss": 2.7599,
      "step": 3875
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.0168840885162354,
      "learning_rate": 0.0005995811281325647,
      "loss": 2.5706,
      "step": 3876
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.7620167136192322,
      "learning_rate": 0.0005995809120188717,
      "loss": 2.9922,
      "step": 3877
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.165700912475586,
      "learning_rate": 0.000599580695849481,
      "loss": 2.7657,
      "step": 3878
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.8048881888389587,
      "learning_rate": 0.0005995804796243925,
      "loss": 2.9622,
      "step": 3879
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.6804617643356323,
      "learning_rate": 0.0005995802633436063,
      "loss": 2.6207,
      "step": 3880
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.787185788154602,
      "learning_rate": 0.0005995800470071225,
      "loss": 2.8118,
      "step": 3881
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.913896918296814,
      "learning_rate": 0.0005995798306149412,
      "loss": 2.6935,
      "step": 3882
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.7400401830673218,
      "learning_rate": 0.0005995796141670623,
      "loss": 2.7015,
      "step": 3883
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.6555640697479248,
      "learning_rate": 0.0005995793976634858,
      "loss": 2.5612,
      "step": 3884
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.9960163831710815,
      "learning_rate": 0.000599579181104212,
      "loss": 2.7993,
      "step": 3885
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.7247079610824585,
      "learning_rate": 0.0005995789644892407,
      "loss": 2.6434,
      "step": 3886
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.7309076189994812,
      "learning_rate": 0.000599578747818572,
      "loss": 2.9008,
      "step": 3887
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.7177507877349854,
      "learning_rate": 0.0005995785310922059,
      "loss": 2.7315,
      "step": 3888
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.7441149950027466,
      "learning_rate": 0.0005995783143101426,
      "loss": 2.9728,
      "step": 3889
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.7281062602996826,
      "learning_rate": 0.000599578097472382,
      "loss": 2.6803,
      "step": 3890
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.6772701740264893,
      "learning_rate": 0.0005995778805789241,
      "loss": 2.8428,
      "step": 3891
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.887294590473175,
      "learning_rate": 0.0005995776636297691,
      "loss": 2.6802,
      "step": 3892
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.6753152012825012,
      "learning_rate": 0.0005995774466249171,
      "loss": 2.5867,
      "step": 3893
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.699776291847229,
      "learning_rate": 0.0005995772295643677,
      "loss": 2.4834,
      "step": 3894
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.6880998015403748,
      "learning_rate": 0.0005995770124481214,
      "loss": 2.5242,
      "step": 3895
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.771550178527832,
      "learning_rate": 0.0005995767952761781,
      "loss": 2.9253,
      "step": 3896
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.725508451461792,
      "learning_rate": 0.0005995765780485378,
      "loss": 2.8486,
      "step": 3897
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.9692317247390747,
      "learning_rate": 0.0005995763607652005,
      "loss": 2.7639,
      "step": 3898
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.6820015907287598,
      "learning_rate": 0.0005995761434261663,
      "loss": 2.6393,
      "step": 3899
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.8346254229545593,
      "learning_rate": 0.0005995759260314353,
      "loss": 2.5361,
      "step": 3900
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.8091678023338318,
      "learning_rate": 0.0005995757085810075,
      "loss": 2.8448,
      "step": 3901
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.7110316157341003,
      "learning_rate": 0.0005995754910748829,
      "loss": 2.9471,
      "step": 3902
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.7927371859550476,
      "learning_rate": 0.0005995752735130615,
      "loss": 2.6845,
      "step": 3903
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.768074095249176,
      "learning_rate": 0.0005995750558955435,
      "loss": 2.7627,
      "step": 3904
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.6605927348136902,
      "learning_rate": 0.0005995748382223287,
      "loss": 2.4782,
      "step": 3905
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.7100399732589722,
      "learning_rate": 0.0005995746204934174,
      "loss": 2.6247,
      "step": 3906
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.6627166867256165,
      "learning_rate": 0.0005995744027088095,
      "loss": 2.5759,
      "step": 3907
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.7759758234024048,
      "learning_rate": 0.000599574184868505,
      "loss": 2.3164,
      "step": 3908
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.6772497296333313,
      "learning_rate": 0.0005995739669725042,
      "loss": 2.6275,
      "step": 3909
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.7779210209846497,
      "learning_rate": 0.0005995737490208067,
      "loss": 2.7556,
      "step": 3910
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.8032159805297852,
      "learning_rate": 0.0005995735310134129,
      "loss": 2.6039,
      "step": 3911
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.7444838881492615,
      "learning_rate": 0.0005995733129503227,
      "loss": 2.6491,
      "step": 3912
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.7446901202201843,
      "learning_rate": 0.0005995730948315361,
      "loss": 3.0437,
      "step": 3913
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.742646336555481,
      "learning_rate": 0.0005995728766570534,
      "loss": 2.724,
      "step": 3914
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.8291356563568115,
      "learning_rate": 0.0005995726584268743,
      "loss": 2.8083,
      "step": 3915
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.6990510821342468,
      "learning_rate": 0.0005995724401409991,
      "loss": 2.7488,
      "step": 3916
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.7615302801132202,
      "learning_rate": 0.0005995722217994276,
      "loss": 2.4975,
      "step": 3917
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.7348071336746216,
      "learning_rate": 0.00059957200340216,
      "loss": 2.8795,
      "step": 3918
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.6787133812904358,
      "learning_rate": 0.0005995717849491964,
      "loss": 2.9326,
      "step": 3919
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.6723086833953857,
      "learning_rate": 0.0005995715664405366,
      "loss": 2.6592,
      "step": 3920
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.664165735244751,
      "learning_rate": 0.0005995713478761809,
      "loss": 2.6198,
      "step": 3921
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.7290799617767334,
      "learning_rate": 0.0005995711292561292,
      "loss": 2.5242,
      "step": 3922
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.7452816963195801,
      "learning_rate": 0.0005995709105803816,
      "loss": 2.6699,
      "step": 3923
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.723201334476471,
      "learning_rate": 0.0005995706918489381,
      "loss": 2.9463,
      "step": 3924
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.7185056805610657,
      "learning_rate": 0.0005995704730617987,
      "loss": 2.468,
      "step": 3925
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.7549328804016113,
      "learning_rate": 0.0005995702542189635,
      "loss": 2.792,
      "step": 3926
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.6883371472358704,
      "learning_rate": 0.0005995700353204327,
      "loss": 2.7339,
      "step": 3927
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.864503026008606,
      "learning_rate": 0.0005995698163662061,
      "loss": 2.4997,
      "step": 3928
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.6905949711799622,
      "learning_rate": 0.0005995695973562837,
      "loss": 2.489,
      "step": 3929
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.7106102705001831,
      "learning_rate": 0.0005995693782906656,
      "loss": 2.8955,
      "step": 3930
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.76903235912323,
      "learning_rate": 0.0005995691591693521,
      "loss": 2.724,
      "step": 3931
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.7676016688346863,
      "learning_rate": 0.0005995689399923429,
      "loss": 2.5119,
      "step": 3932
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.6918005347251892,
      "learning_rate": 0.0005995687207596383,
      "loss": 2.8246,
      "step": 3933
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.6807661056518555,
      "learning_rate": 0.0005995685014712382,
      "loss": 2.6882,
      "step": 3934
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.8555651903152466,
      "learning_rate": 0.0005995682821271426,
      "loss": 2.4766,
      "step": 3935
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.690798282623291,
      "learning_rate": 0.0005995680627273516,
      "loss": 2.9289,
      "step": 3936
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.7919474244117737,
      "learning_rate": 0.0005995678432718653,
      "loss": 2.5527,
      "step": 3937
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.7063793540000916,
      "learning_rate": 0.0005995676237606836,
      "loss": 2.8654,
      "step": 3938
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.7299966812133789,
      "learning_rate": 0.0005995674041938066,
      "loss": 2.5542,
      "step": 3939
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.8097602128982544,
      "learning_rate": 0.0005995671845712344,
      "loss": 2.4769,
      "step": 3940
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.7595064640045166,
      "learning_rate": 0.0005995669648929671,
      "loss": 2.7372,
      "step": 3941
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.6876206398010254,
      "learning_rate": 0.0005995667451590046,
      "loss": 2.8878,
      "step": 3942
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.6433842778205872,
      "learning_rate": 0.000599566525369347,
      "loss": 2.6459,
      "step": 3943
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.6685534119606018,
      "learning_rate": 0.0005995663055239943,
      "loss": 2.6487,
      "step": 3944
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.7611172795295715,
      "learning_rate": 0.0005995660856229466,
      "loss": 2.6476,
      "step": 3945
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.6689639687538147,
      "learning_rate": 0.0005995658656662038,
      "loss": 2.6983,
      "step": 3946
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.7051141858100891,
      "learning_rate": 0.0005995656456537662,
      "loss": 2.7121,
      "step": 3947
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.6623039841651917,
      "learning_rate": 0.0005995654255856336,
      "loss": 2.6827,
      "step": 3948
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.7697027325630188,
      "learning_rate": 0.0005995652054618061,
      "loss": 2.6775,
      "step": 3949
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.6757428646087646,
      "learning_rate": 0.000599564985282284,
      "loss": 2.7594,
      "step": 3950
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.6487859487533569,
      "learning_rate": 0.0005995647650470667,
      "loss": 2.7607,
      "step": 3951
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.7382631301879883,
      "learning_rate": 0.000599564544756155,
      "loss": 2.6946,
      "step": 3952
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.6910587549209595,
      "learning_rate": 0.0005995643244095485,
      "loss": 2.5282,
      "step": 3953
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.7987298369407654,
      "learning_rate": 0.0005995641040072474,
      "loss": 2.8184,
      "step": 3954
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.7442359328269958,
      "learning_rate": 0.0005995638835492516,
      "loss": 2.7451,
      "step": 3955
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.6806663274765015,
      "learning_rate": 0.0005995636630355611,
      "loss": 3.0294,
      "step": 3956
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.8412039279937744,
      "learning_rate": 0.0005995634424661762,
      "loss": 2.7682,
      "step": 3957
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.7250284552574158,
      "learning_rate": 0.0005995632218410968,
      "loss": 2.7796,
      "step": 3958
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.7718241214752197,
      "learning_rate": 0.000599563001160323,
      "loss": 2.7995,
      "step": 3959
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.764895498752594,
      "learning_rate": 0.0005995627804238548,
      "loss": 2.8194,
      "step": 3960
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.6982811093330383,
      "learning_rate": 0.0005995625596316921,
      "loss": 2.5064,
      "step": 3961
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.6766484379768372,
      "learning_rate": 0.0005995623387838351,
      "loss": 2.6425,
      "step": 3962
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.7397156953811646,
      "learning_rate": 0.0005995621178802838,
      "loss": 2.5188,
      "step": 3963
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.7126787304878235,
      "learning_rate": 0.0005995618969210383,
      "loss": 2.6863,
      "step": 3964
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.7047905921936035,
      "learning_rate": 0.0005995616759060985,
      "loss": 2.9207,
      "step": 3965
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.8239327669143677,
      "learning_rate": 0.0005995614548354645,
      "loss": 2.8409,
      "step": 3966
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.7164350152015686,
      "learning_rate": 0.0005995612337091365,
      "loss": 2.7382,
      "step": 3967
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.7556925415992737,
      "learning_rate": 0.0005995610125271144,
      "loss": 2.8478,
      "step": 3968
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.6517524719238281,
      "learning_rate": 0.0005995607912893982,
      "loss": 3.0172,
      "step": 3969
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.7198165655136108,
      "learning_rate": 0.000599560569995988,
      "loss": 2.7773,
      "step": 3970
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.6627523899078369,
      "learning_rate": 0.0005995603486468839,
      "loss": 2.6432,
      "step": 3971
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.705352246761322,
      "learning_rate": 0.0005995601272420857,
      "loss": 2.4969,
      "step": 3972
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.7457765340805054,
      "learning_rate": 0.0005995599057815939,
      "loss": 2.8414,
      "step": 3973
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.6431457996368408,
      "learning_rate": 0.0005995596842654081,
      "loss": 2.5873,
      "step": 3974
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.6821054220199585,
      "learning_rate": 0.0005995594626935284,
      "loss": 2.8415,
      "step": 3975
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.6908231377601624,
      "learning_rate": 0.000599559241065955,
      "loss": 2.7581,
      "step": 3976
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.6701266765594482,
      "learning_rate": 0.000599559019382688,
      "loss": 2.6375,
      "step": 3977
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.6612569689750671,
      "learning_rate": 0.0005995587976437271,
      "loss": 2.7408,
      "step": 3978
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.6906352043151855,
      "learning_rate": 0.0005995585758490728,
      "loss": 2.9949,
      "step": 3979
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.6546461582183838,
      "learning_rate": 0.0005995583539987249,
      "loss": 2.6629,
      "step": 3980
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.6433784365653992,
      "learning_rate": 0.0005995581320926833,
      "loss": 2.7747,
      "step": 3981
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.7590632438659668,
      "learning_rate": 0.0005995579101309483,
      "loss": 2.7471,
      "step": 3982
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.7116715312004089,
      "learning_rate": 0.0005995576881135197,
      "loss": 2.6785,
      "step": 3983
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.7229580879211426,
      "learning_rate": 0.0005995574660403978,
      "loss": 2.7767,
      "step": 3984
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.7573322057723999,
      "learning_rate": 0.0005995572439115825,
      "loss": 2.5532,
      "step": 3985
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.6817473769187927,
      "learning_rate": 0.0005995570217270738,
      "loss": 2.6735,
      "step": 3986
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.7608422040939331,
      "learning_rate": 0.0005995567994868718,
      "loss": 2.6054,
      "step": 3987
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.905318021774292,
      "learning_rate": 0.0005995565771909765,
      "loss": 3.121,
      "step": 3988
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.8270664215087891,
      "learning_rate": 0.0005995563548393881,
      "loss": 2.5455,
      "step": 3989
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.6886564493179321,
      "learning_rate": 0.0005995561324321064,
      "loss": 2.6442,
      "step": 3990
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.6266409754753113,
      "learning_rate": 0.0005995559099691316,
      "loss": 2.7763,
      "step": 3991
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.2681593894958496,
      "learning_rate": 0.0005995556874504637,
      "loss": 2.6654,
      "step": 3992
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.7435198426246643,
      "learning_rate": 0.0005995554648761028,
      "loss": 2.7365,
      "step": 3993
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.6412767767906189,
      "learning_rate": 0.0005995552422460488,
      "loss": 2.6604,
      "step": 3994
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.7414081692695618,
      "learning_rate": 0.0005995550195603018,
      "loss": 2.5808,
      "step": 3995
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.7433432936668396,
      "learning_rate": 0.0005995547968188621,
      "loss": 2.666,
      "step": 3996
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.7307658195495605,
      "learning_rate": 0.0005995545740217293,
      "loss": 2.7715,
      "step": 3997
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.6768355965614319,
      "learning_rate": 0.0005995543511689036,
      "loss": 2.7934,
      "step": 3998
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.6643418073654175,
      "learning_rate": 0.0005995541282603852,
      "loss": 2.5332,
      "step": 3999
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.6870315670967102,
      "learning_rate": 0.0005995539052961741,
      "loss": 2.71,
      "step": 4000
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.7189217805862427,
      "learning_rate": 0.0005995536822762702,
      "loss": 2.7782,
      "step": 4001
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.727331280708313,
      "learning_rate": 0.0005995534592006735,
      "loss": 2.9617,
      "step": 4002
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.6974807381629944,
      "learning_rate": 0.0005995532360693844,
      "loss": 2.9651,
      "step": 4003
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.686678409576416,
      "learning_rate": 0.0005995530128824025,
      "loss": 2.7355,
      "step": 4004
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.8003748059272766,
      "learning_rate": 0.0005995527896397281,
      "loss": 2.6635,
      "step": 4005
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.7129806876182556,
      "learning_rate": 0.0005995525663413612,
      "loss": 2.6451,
      "step": 4006
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.6944684982299805,
      "learning_rate": 0.0005995523429873018,
      "loss": 3.0514,
      "step": 4007
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.8325474262237549,
      "learning_rate": 0.00059955211957755,
      "loss": 2.7931,
      "step": 4008
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.6994345784187317,
      "learning_rate": 0.0005995518961121058,
      "loss": 2.6268,
      "step": 4009
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.7077631950378418,
      "learning_rate": 0.0005995516725909692,
      "loss": 2.8874,
      "step": 4010
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.6291051506996155,
      "learning_rate": 0.0005995514490141404,
      "loss": 2.5549,
      "step": 4011
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.6866896152496338,
      "learning_rate": 0.0005995512253816192,
      "loss": 2.6727,
      "step": 4012
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.808883786201477,
      "learning_rate": 0.0005995510016934058,
      "loss": 2.5926,
      "step": 4013
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.7928246259689331,
      "learning_rate": 0.0005995507779495002,
      "loss": 2.5386,
      "step": 4014
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.784550130367279,
      "learning_rate": 0.0005995505541499025,
      "loss": 2.7194,
      "step": 4015
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.6663126349449158,
      "learning_rate": 0.0005995503302946126,
      "loss": 2.7943,
      "step": 4016
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.705118715763092,
      "learning_rate": 0.0005995501063836308,
      "loss": 2.9399,
      "step": 4017
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.8814824223518372,
      "learning_rate": 0.0005995498824169569,
      "loss": 2.5628,
      "step": 4018
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.7445215582847595,
      "learning_rate": 0.000599549658394591,
      "loss": 2.6844,
      "step": 4019
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.0317500829696655,
      "learning_rate": 0.0005995494343165333,
      "loss": 2.5657,
      "step": 4020
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.7129844427108765,
      "learning_rate": 0.0005995492101827835,
      "loss": 2.7878,
      "step": 4021
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.7055633068084717,
      "learning_rate": 0.000599548985993342,
      "loss": 2.8139,
      "step": 4022
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.146217703819275,
      "learning_rate": 0.0005995487617482087,
      "loss": 2.7453,
      "step": 4023
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.7437198162078857,
      "learning_rate": 0.0005995485374473836,
      "loss": 2.9218,
      "step": 4024
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.7020946741104126,
      "learning_rate": 0.0005995483130908666,
      "loss": 2.7093,
      "step": 4025
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.869351327419281,
      "learning_rate": 0.000599548088678658,
      "loss": 2.5781,
      "step": 4026
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.976130485534668,
      "learning_rate": 0.0005995478642107579,
      "loss": 2.4613,
      "step": 4027
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.7512860298156738,
      "learning_rate": 0.0005995476396871661,
      "loss": 2.7224,
      "step": 4028
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.8086357712745667,
      "learning_rate": 0.0005995474151078828,
      "loss": 2.6417,
      "step": 4029
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.7804334759712219,
      "learning_rate": 0.0005995471904729079,
      "loss": 2.7247,
      "step": 4030
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.7452031373977661,
      "learning_rate": 0.0005995469657822416,
      "loss": 2.6146,
      "step": 4031
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.7728099226951599,
      "learning_rate": 0.0005995467410358838,
      "loss": 2.7131,
      "step": 4032
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.868834912776947,
      "learning_rate": 0.0005995465162338346,
      "loss": 2.8391,
      "step": 4033
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.8520104289054871,
      "learning_rate": 0.0005995462913760941,
      "loss": 2.7222,
      "step": 4034
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.7112880349159241,
      "learning_rate": 0.0005995460664626623,
      "loss": 2.8245,
      "step": 4035
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.8183413147926331,
      "learning_rate": 0.0005995458414935392,
      "loss": 2.937,
      "step": 4036
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.0643569231033325,
      "learning_rate": 0.0005995456164687247,
      "loss": 2.8155,
      "step": 4037
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.703341007232666,
      "learning_rate": 0.0005995453913882193,
      "loss": 2.3724,
      "step": 4038
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.7689452171325684,
      "learning_rate": 0.0005995451662520226,
      "loss": 2.7395,
      "step": 4039
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.8063772916793823,
      "learning_rate": 0.0005995449410601348,
      "loss": 2.6207,
      "step": 4040
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.8982124924659729,
      "learning_rate": 0.0005995447158125561,
      "loss": 2.5049,
      "step": 4041
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.7337118983268738,
      "learning_rate": 0.0005995444905092861,
      "loss": 2.7794,
      "step": 4042
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.822993278503418,
      "learning_rate": 0.0005995442651503253,
      "loss": 2.7267,
      "step": 4043
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.8819081783294678,
      "learning_rate": 0.0005995440397356736,
      "loss": 2.8397,
      "step": 4044
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.7192308902740479,
      "learning_rate": 0.000599543814265331,
      "loss": 2.928,
      "step": 4045
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.7640491724014282,
      "learning_rate": 0.0005995435887392975,
      "loss": 2.6615,
      "step": 4046
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.7505104541778564,
      "learning_rate": 0.0005995433631575733,
      "loss": 2.48,
      "step": 4047
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.7723228931427002,
      "learning_rate": 0.0005995431375201581,
      "loss": 2.7192,
      "step": 4048
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.7094102501869202,
      "learning_rate": 0.0005995429118270524,
      "loss": 2.762,
      "step": 4049
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.7558715343475342,
      "learning_rate": 0.000599542686078256,
      "loss": 2.5855,
      "step": 4050
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.6998162865638733,
      "learning_rate": 0.0005995424602737689,
      "loss": 2.6365,
      "step": 4051
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.6805304288864136,
      "learning_rate": 0.0005995422344135911,
      "loss": 2.467,
      "step": 4052
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.6649808883666992,
      "learning_rate": 0.0005995420084977228,
      "loss": 2.639,
      "step": 4053
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.752515971660614,
      "learning_rate": 0.0005995417825261641,
      "loss": 2.6559,
      "step": 4054
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.6856701970100403,
      "learning_rate": 0.0005995415564989149,
      "loss": 2.3534,
      "step": 4055
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.8393123745918274,
      "learning_rate": 0.0005995413304159752,
      "loss": 2.616,
      "step": 4056
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.7676721215248108,
      "learning_rate": 0.0005995411042773451,
      "loss": 2.6421,
      "step": 4057
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.6600187420845032,
      "learning_rate": 0.0005995408780830247,
      "loss": 2.5456,
      "step": 4058
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.6915035247802734,
      "learning_rate": 0.000599540651833014,
      "loss": 2.5997,
      "step": 4059
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.6854279041290283,
      "learning_rate": 0.000599540425527313,
      "loss": 2.6093,
      "step": 4060
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.893969714641571,
      "learning_rate": 0.0005995401991659218,
      "loss": 2.5539,
      "step": 4061
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.7117682099342346,
      "learning_rate": 0.0005995399727488404,
      "loss": 2.5496,
      "step": 4062
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.7265838980674744,
      "learning_rate": 0.0005995397462760689,
      "loss": 2.801,
      "step": 4063
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.8298216462135315,
      "learning_rate": 0.0005995395197476073,
      "loss": 2.7137,
      "step": 4064
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.6636000871658325,
      "learning_rate": 0.0005995392931634557,
      "loss": 2.5008,
      "step": 4065
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.8301615715026855,
      "learning_rate": 0.000599539066523614,
      "loss": 2.4737,
      "step": 4066
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.7504658699035645,
      "learning_rate": 0.0005995388398280824,
      "loss": 2.9619,
      "step": 4067
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.7760650515556335,
      "learning_rate": 0.0005995386130768609,
      "loss": 2.6473,
      "step": 4068
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.6316482424736023,
      "learning_rate": 0.0005995383862699494,
      "loss": 2.6741,
      "step": 4069
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.8056036829948425,
      "learning_rate": 0.0005995381594073483,
      "loss": 2.7183,
      "step": 4070
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.719580888748169,
      "learning_rate": 0.0005995379324890571,
      "loss": 2.7142,
      "step": 4071
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.774823009967804,
      "learning_rate": 0.0005995377055150763,
      "loss": 2.7076,
      "step": 4072
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.8099996447563171,
      "learning_rate": 0.0005995374784854058,
      "loss": 2.7286,
      "step": 4073
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.66338050365448,
      "learning_rate": 0.0005995372514000456,
      "loss": 2.7861,
      "step": 4074
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.8185650110244751,
      "learning_rate": 0.0005995370242589958,
      "loss": 2.879,
      "step": 4075
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.8647106289863586,
      "learning_rate": 0.0005995367970622563,
      "loss": 2.723,
      "step": 4076
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.7395628690719604,
      "learning_rate": 0.0005995365698098273,
      "loss": 2.8195,
      "step": 4077
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.7847521901130676,
      "learning_rate": 0.0005995363425017089,
      "loss": 2.928,
      "step": 4078
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.6905567646026611,
      "learning_rate": 0.000599536115137901,
      "loss": 2.7534,
      "step": 4079
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.6465857625007629,
      "learning_rate": 0.0005995358877184036,
      "loss": 2.5781,
      "step": 4080
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.7095893621444702,
      "learning_rate": 0.0005995356602432169,
      "loss": 2.71,
      "step": 4081
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.745180606842041,
      "learning_rate": 0.0005995354327123408,
      "loss": 2.5793,
      "step": 4082
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.6602818965911865,
      "learning_rate": 0.0005995352051257754,
      "loss": 2.6419,
      "step": 4083
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.6722280383110046,
      "learning_rate": 0.0005995349774835209,
      "loss": 2.7824,
      "step": 4084
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.7706446051597595,
      "learning_rate": 0.000599534749785577,
      "loss": 2.861,
      "step": 4085
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.6832758784294128,
      "learning_rate": 0.000599534522031944,
      "loss": 2.589,
      "step": 4086
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.643278181552887,
      "learning_rate": 0.0005995342942226219,
      "loss": 2.9498,
      "step": 4087
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.7176345586776733,
      "learning_rate": 0.0005995340663576108,
      "loss": 2.9275,
      "step": 4088
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.6907328367233276,
      "learning_rate": 0.0005995338384369106,
      "loss": 2.4582,
      "step": 4089
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.8072409629821777,
      "learning_rate": 0.0005995336104605214,
      "loss": 2.8959,
      "step": 4090
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.7252004742622375,
      "learning_rate": 0.0005995333824284433,
      "loss": 2.861,
      "step": 4091
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.7766438126564026,
      "learning_rate": 0.0005995331543406762,
      "loss": 2.8091,
      "step": 4092
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.6927722692489624,
      "learning_rate": 0.0005995329261972202,
      "loss": 2.7259,
      "step": 4093
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.6995670199394226,
      "learning_rate": 0.0005995326979980755,
      "loss": 2.8899,
      "step": 4094
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.6786299347877502,
      "learning_rate": 0.000599532469743242,
      "loss": 2.8023,
      "step": 4095
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.7325018644332886,
      "learning_rate": 0.0005995322414327196,
      "loss": 2.5817,
      "step": 4096
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.6869426965713501,
      "learning_rate": 0.0005995320130665087,
      "loss": 2.7928,
      "step": 4097
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.6906819939613342,
      "learning_rate": 0.0005995317846446091,
      "loss": 2.8854,
      "step": 4098
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.7611040472984314,
      "learning_rate": 0.0005995315561670208,
      "loss": 2.8914,
      "step": 4099
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.6969742774963379,
      "learning_rate": 0.000599531327633744,
      "loss": 2.5207,
      "step": 4100
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.7293642163276672,
      "learning_rate": 0.0005995310990447786,
      "loss": 2.5237,
      "step": 4101
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.8450138568878174,
      "learning_rate": 0.0005995308704001248,
      "loss": 2.6525,
      "step": 4102
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.7074275016784668,
      "learning_rate": 0.0005995306416997825,
      "loss": 2.7971,
      "step": 4103
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.6871335506439209,
      "learning_rate": 0.0005995304129437519,
      "loss": 2.9117,
      "step": 4104
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.7443216443061829,
      "learning_rate": 0.0005995301841320329,
      "loss": 2.8949,
      "step": 4105
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.6311076283454895,
      "learning_rate": 0.0005995299552646254,
      "loss": 2.6324,
      "step": 4106
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.6421827077865601,
      "learning_rate": 0.0005995297263415299,
      "loss": 2.6018,
      "step": 4107
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.7245688438415527,
      "learning_rate": 0.000599529497362746,
      "loss": 2.8498,
      "step": 4108
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.6481211185455322,
      "learning_rate": 0.000599529268328274,
      "loss": 2.3628,
      "step": 4109
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.6745325326919556,
      "learning_rate": 0.0005995290392381139,
      "loss": 2.5177,
      "step": 4110
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.6762605905532837,
      "learning_rate": 0.0005995288100922655,
      "loss": 2.5237,
      "step": 4111
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.8148390054702759,
      "learning_rate": 0.0005995285808907292,
      "loss": 2.5899,
      "step": 4112
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.6532413959503174,
      "learning_rate": 0.0005995283516335049,
      "loss": 2.5435,
      "step": 4113
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.7418607473373413,
      "learning_rate": 0.0005995281223205925,
      "loss": 2.9204,
      "step": 4114
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.6876304149627686,
      "learning_rate": 0.0005995278929519923,
      "loss": 2.9015,
      "step": 4115
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.6904596090316772,
      "learning_rate": 0.000599527663527704,
      "loss": 2.5112,
      "step": 4116
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.7353073954582214,
      "learning_rate": 0.0005995274340477281,
      "loss": 2.6662,
      "step": 4117
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.6716001033782959,
      "learning_rate": 0.0005995272045120643,
      "loss": 2.7296,
      "step": 4118
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.6555994749069214,
      "learning_rate": 0.0005995269749207127,
      "loss": 2.7515,
      "step": 4119
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.6734583377838135,
      "learning_rate": 0.0005995267452736734,
      "loss": 2.7838,
      "step": 4120
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.7273914813995361,
      "learning_rate": 0.0005995265155709465,
      "loss": 2.4927,
      "step": 4121
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.6778247356414795,
      "learning_rate": 0.0005995262858125318,
      "loss": 2.5199,
      "step": 4122
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.765192985534668,
      "learning_rate": 0.0005995260559984297,
      "loss": 2.9783,
      "step": 4123
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.7584558725357056,
      "learning_rate": 0.00059952582612864,
      "loss": 2.86,
      "step": 4124
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.6666519641876221,
      "learning_rate": 0.0005995255962031626,
      "loss": 2.6784,
      "step": 4125
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.9252867698669434,
      "learning_rate": 0.000599525366221998,
      "loss": 2.7034,
      "step": 4126
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.8884738683700562,
      "learning_rate": 0.0005995251361851459,
      "loss": 2.6979,
      "step": 4127
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.987975001335144,
      "learning_rate": 0.0005995249060926063,
      "loss": 2.6891,
      "step": 4128
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.7658697366714478,
      "learning_rate": 0.0005995246759443796,
      "loss": 2.7115,
      "step": 4129
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.7420585751533508,
      "learning_rate": 0.0005995244457404654,
      "loss": 2.6695,
      "step": 4130
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.8840348720550537,
      "learning_rate": 0.000599524215480864,
      "loss": 2.9046,
      "step": 4131
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.8577247858047485,
      "learning_rate": 0.0005995239851655754,
      "loss": 2.6065,
      "step": 4132
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.7853448987007141,
      "learning_rate": 0.0005995237547945997,
      "loss": 2.5467,
      "step": 4133
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.8656133413314819,
      "learning_rate": 0.0005995235243679368,
      "loss": 2.6972,
      "step": 4134
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.8515226244926453,
      "learning_rate": 0.0005995232938855869,
      "loss": 2.8893,
      "step": 4135
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.7080196142196655,
      "learning_rate": 0.0005995230633475499,
      "loss": 2.5846,
      "step": 4136
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.7443058490753174,
      "learning_rate": 0.000599522832753826,
      "loss": 2.9193,
      "step": 4137
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.7804380655288696,
      "learning_rate": 0.0005995226021044151,
      "loss": 2.9318,
      "step": 4138
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.6682078838348389,
      "learning_rate": 0.0005995223713993172,
      "loss": 2.8033,
      "step": 4139
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.7806894183158875,
      "learning_rate": 0.0005995221406385325,
      "loss": 2.974,
      "step": 4140
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.7070727944374084,
      "learning_rate": 0.000599521909822061,
      "loss": 2.5449,
      "step": 4141
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.8618009090423584,
      "learning_rate": 0.0005995216789499028,
      "loss": 2.4518,
      "step": 4142
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.7089043259620667,
      "learning_rate": 0.0005995214480220577,
      "loss": 2.6941,
      "step": 4143
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.998263418674469,
      "learning_rate": 0.0005995212170385261,
      "loss": 2.7943,
      "step": 4144
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.7271682024002075,
      "learning_rate": 0.0005995209859993077,
      "loss": 2.7185,
      "step": 4145
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.9059241414070129,
      "learning_rate": 0.0005995207549044027,
      "loss": 2.4572,
      "step": 4146
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.7290046215057373,
      "learning_rate": 0.0005995205237538112,
      "loss": 2.852,
      "step": 4147
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.7701406478881836,
      "learning_rate": 0.0005995202925475332,
      "loss": 2.5894,
      "step": 4148
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.8190118670463562,
      "learning_rate": 0.0005995200612855687,
      "loss": 2.6849,
      "step": 4149
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.7487543225288391,
      "learning_rate": 0.0005995198299679177,
      "loss": 2.7327,
      "step": 4150
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.8216764330863953,
      "learning_rate": 0.0005995195985945804,
      "loss": 2.6801,
      "step": 4151
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.6923763751983643,
      "learning_rate": 0.0005995193671655567,
      "loss": 2.573,
      "step": 4152
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.7120804786682129,
      "learning_rate": 0.0005995191356808467,
      "loss": 2.6978,
      "step": 4153
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.6406145095825195,
      "learning_rate": 0.0005995189041404505,
      "loss": 2.7979,
      "step": 4154
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.8905795812606812,
      "learning_rate": 0.000599518672544368,
      "loss": 2.4721,
      "step": 4155
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.7814649939537048,
      "learning_rate": 0.0005995184408925994,
      "loss": 2.8502,
      "step": 4156
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.0165051221847534,
      "learning_rate": 0.0005995182091851446,
      "loss": 2.6767,
      "step": 4157
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.6234191656112671,
      "learning_rate": 0.0005995179774220037,
      "loss": 2.6406,
      "step": 4158
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.9261308312416077,
      "learning_rate": 0.0005995177456031768,
      "loss": 2.6162,
      "step": 4159
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.8077061772346497,
      "learning_rate": 0.0005995175137286638,
      "loss": 2.676,
      "step": 4160
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.656504213809967,
      "learning_rate": 0.000599517281798465,
      "loss": 2.7238,
      "step": 4161
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.7800907492637634,
      "learning_rate": 0.0005995170498125802,
      "loss": 2.6887,
      "step": 4162
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.8290053606033325,
      "learning_rate": 0.0005995168177710096,
      "loss": 2.7461,
      "step": 4163
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.8705119490623474,
      "learning_rate": 0.0005995165856737531,
      "loss": 2.9382,
      "step": 4164
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.799960732460022,
      "learning_rate": 0.0005995163535208108,
      "loss": 2.714,
      "step": 4165
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.9173892736434937,
      "learning_rate": 0.0005995161213121828,
      "loss": 2.5803,
      "step": 4166
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.7168704867362976,
      "learning_rate": 0.0005995158890478691,
      "loss": 2.8615,
      "step": 4167
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.7170622944831848,
      "learning_rate": 0.0005995156567278697,
      "loss": 2.7569,
      "step": 4168
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.765055775642395,
      "learning_rate": 0.0005995154243521847,
      "loss": 2.7377,
      "step": 4169
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.7223255634307861,
      "learning_rate": 0.0005995151919208141,
      "loss": 2.6004,
      "step": 4170
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.6727193593978882,
      "learning_rate": 0.000599514959433758,
      "loss": 2.5861,
      "step": 4171
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.7540547251701355,
      "learning_rate": 0.0005995147268910164,
      "loss": 2.6827,
      "step": 4172
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.660440981388092,
      "learning_rate": 0.0005995144942925894,
      "loss": 2.5789,
      "step": 4173
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.7706771492958069,
      "learning_rate": 0.0005995142616384769,
      "loss": 2.6235,
      "step": 4174
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.7855486869812012,
      "learning_rate": 0.0005995140289286792,
      "loss": 2.7007,
      "step": 4175
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.6747564077377319,
      "learning_rate": 0.0005995137961631961,
      "loss": 2.7757,
      "step": 4176
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.7674809694290161,
      "learning_rate": 0.0005995135633420277,
      "loss": 2.5638,
      "step": 4177
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.6354509592056274,
      "learning_rate": 0.0005995133304651742,
      "loss": 2.7686,
      "step": 4178
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.9103520512580872,
      "learning_rate": 0.0005995130975326354,
      "loss": 2.7291,
      "step": 4179
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.6854572296142578,
      "learning_rate": 0.0005995128645444115,
      "loss": 2.6654,
      "step": 4180
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.7729571461677551,
      "learning_rate": 0.0005995126315005025,
      "loss": 2.6661,
      "step": 4181
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.853220522403717,
      "learning_rate": 0.0005995123984009085,
      "loss": 2.6407,
      "step": 4182
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.7878636717796326,
      "learning_rate": 0.0005995121652456294,
      "loss": 2.6828,
      "step": 4183
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.7314440011978149,
      "learning_rate": 0.0005995119320346654,
      "loss": 2.7467,
      "step": 4184
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.6965017914772034,
      "learning_rate": 0.0005995116987680165,
      "loss": 2.8147,
      "step": 4185
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.8245173692703247,
      "learning_rate": 0.0005995114654456826,
      "loss": 2.7773,
      "step": 4186
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.6915760636329651,
      "learning_rate": 0.000599511232067664,
      "loss": 2.652,
      "step": 4187
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.717755913734436,
      "learning_rate": 0.0005995109986339605,
      "loss": 2.6752,
      "step": 4188
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.739520788192749,
      "learning_rate": 0.0005995107651445723,
      "loss": 2.589,
      "step": 4189
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.7043514251708984,
      "learning_rate": 0.0005995105315994994,
      "loss": 2.8825,
      "step": 4190
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.8669980764389038,
      "learning_rate": 0.0005995102979987419,
      "loss": 2.5953,
      "step": 4191
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.7492019534111023,
      "learning_rate": 0.0005995100643422996,
      "loss": 2.7521,
      "step": 4192
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.7645072937011719,
      "learning_rate": 0.000599509830630173,
      "loss": 2.6097,
      "step": 4193
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.7058998346328735,
      "learning_rate": 0.0005995095968623616,
      "loss": 2.8194,
      "step": 4194
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.7628397345542908,
      "learning_rate": 0.0005995093630388658,
      "loss": 2.5141,
      "step": 4195
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.6918125152587891,
      "learning_rate": 0.0005995091291596857,
      "loss": 2.6088,
      "step": 4196
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.7687075138092041,
      "learning_rate": 0.000599508895224821,
      "loss": 2.7792,
      "step": 4197
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.6795701384544373,
      "learning_rate": 0.0005995086612342721,
      "loss": 2.8619,
      "step": 4198
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.6348385214805603,
      "learning_rate": 0.0005995084271880387,
      "loss": 2.7495,
      "step": 4199
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.733742356300354,
      "learning_rate": 0.0005995081930861212,
      "loss": 2.8777,
      "step": 4200
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.8102611303329468,
      "learning_rate": 0.0005995079589285195,
      "loss": 2.8949,
      "step": 4201
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.7173080444335938,
      "learning_rate": 0.0005995077247152335,
      "loss": 2.7683,
      "step": 4202
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.8010827302932739,
      "learning_rate": 0.0005995074904462633,
      "loss": 2.6258,
      "step": 4203
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.7043018341064453,
      "learning_rate": 0.0005995072561216092,
      "loss": 2.8208,
      "step": 4204
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.720863401889801,
      "learning_rate": 0.0005995070217412709,
      "loss": 2.8298,
      "step": 4205
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.8562477231025696,
      "learning_rate": 0.0005995067873052486,
      "loss": 2.7244,
      "step": 4206
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.7385909557342529,
      "learning_rate": 0.0005995065528135423,
      "loss": 2.8532,
      "step": 4207
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.7983902096748352,
      "learning_rate": 0.0005995063182661522,
      "loss": 2.7748,
      "step": 4208
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.0094150304794312,
      "learning_rate": 0.0005995060836630782,
      "loss": 2.4439,
      "step": 4209
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.7640581130981445,
      "learning_rate": 0.0005995058490043203,
      "loss": 2.6859,
      "step": 4210
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.7791821956634521,
      "learning_rate": 0.0005995056142898786,
      "loss": 2.8464,
      "step": 4211
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.0009565353393555,
      "learning_rate": 0.0005995053795197533,
      "loss": 2.608,
      "step": 4212
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.7719940543174744,
      "learning_rate": 0.0005995051446939442,
      "loss": 2.7093,
      "step": 4213
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.0961116552352905,
      "learning_rate": 0.0005995049098124513,
      "loss": 2.4625,
      "step": 4214
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.7407582402229309,
      "learning_rate": 0.000599504674875275,
      "loss": 2.6317,
      "step": 4215
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.7105004191398621,
      "learning_rate": 0.0005995044398824149,
      "loss": 2.6535,
      "step": 4216
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.7026314735412598,
      "learning_rate": 0.0005995042048338715,
      "loss": 2.594,
      "step": 4217
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.7146266102790833,
      "learning_rate": 0.0005995039697296445,
      "loss": 2.5856,
      "step": 4218
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.6684063673019409,
      "learning_rate": 0.0005995037345697342,
      "loss": 2.6439,
      "step": 4219
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.6235747337341309,
      "learning_rate": 0.0005995034993541403,
      "loss": 2.4226,
      "step": 4220
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.7091933488845825,
      "learning_rate": 0.0005995032640828631,
      "loss": 2.8312,
      "step": 4221
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.7417935729026794,
      "learning_rate": 0.0005995030287559026,
      "loss": 2.6321,
      "step": 4222
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.6483005285263062,
      "learning_rate": 0.0005995027933732589,
      "loss": 2.4271,
      "step": 4223
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.7514188885688782,
      "learning_rate": 0.0005995025579349319,
      "loss": 2.754,
      "step": 4224
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.6964255571365356,
      "learning_rate": 0.0005995023224409219,
      "loss": 2.6278,
      "step": 4225
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.681320071220398,
      "learning_rate": 0.0005995020868912285,
      "loss": 2.5986,
      "step": 4226
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.6645088791847229,
      "learning_rate": 0.0005995018512858521,
      "loss": 2.9551,
      "step": 4227
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.6541523933410645,
      "learning_rate": 0.0005995016156247928,
      "loss": 2.773,
      "step": 4228
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.8135634064674377,
      "learning_rate": 0.0005995013799080503,
      "loss": 2.8072,
      "step": 4229
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.6118401288986206,
      "learning_rate": 0.0005995011441356249,
      "loss": 2.7858,
      "step": 4230
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.7069330811500549,
      "learning_rate": 0.0005995009083075166,
      "loss": 2.562,
      "step": 4231
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.8129875659942627,
      "learning_rate": 0.0005995006724237254,
      "loss": 2.5047,
      "step": 4232
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.6261624693870544,
      "learning_rate": 0.0005995004364842514,
      "loss": 2.6421,
      "step": 4233
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.6750322580337524,
      "learning_rate": 0.0005995002004890946,
      "loss": 2.8492,
      "step": 4234
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.6566902995109558,
      "learning_rate": 0.0005994999644382551,
      "loss": 2.5849,
      "step": 4235
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.7045894265174866,
      "learning_rate": 0.0005994997283317329,
      "loss": 2.5112,
      "step": 4236
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.7865556478500366,
      "learning_rate": 0.0005994994921695281,
      "loss": 2.5524,
      "step": 4237
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.6689057946205139,
      "learning_rate": 0.0005994992559516405,
      "loss": 2.6487,
      "step": 4238
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.6504020094871521,
      "learning_rate": 0.0005994990196780705,
      "loss": 2.7509,
      "step": 4239
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.7042204141616821,
      "learning_rate": 0.000599498783348818,
      "loss": 2.5718,
      "step": 4240
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.7402991652488708,
      "learning_rate": 0.0005994985469638829,
      "loss": 2.6558,
      "step": 4241
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.7007738351821899,
      "learning_rate": 0.0005994983105232654,
      "loss": 2.5367,
      "step": 4242
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.6481521129608154,
      "learning_rate": 0.0005994980740269655,
      "loss": 2.8105,
      "step": 4243
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.7253128886222839,
      "learning_rate": 0.0005994978374749833,
      "loss": 2.6817,
      "step": 4244
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.6920745968818665,
      "learning_rate": 0.0005994976008673187,
      "loss": 2.7615,
      "step": 4245
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.776671290397644,
      "learning_rate": 0.000599497364203972,
      "loss": 2.8493,
      "step": 4246
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.6650575995445251,
      "learning_rate": 0.000599497127484943,
      "loss": 2.7714,
      "step": 4247
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.842217743396759,
      "learning_rate": 0.0005994968907102318,
      "loss": 2.7861,
      "step": 4248
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.6682327389717102,
      "learning_rate": 0.0005994966538798385,
      "loss": 2.5831,
      "step": 4249
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.6747097373008728,
      "learning_rate": 0.0005994964169937632,
      "loss": 2.6812,
      "step": 4250
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.729140043258667,
      "learning_rate": 0.0005994961800520058,
      "loss": 2.6557,
      "step": 4251
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.7062219381332397,
      "learning_rate": 0.0005994959430545664,
      "loss": 2.5762,
      "step": 4252
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.7477446794509888,
      "learning_rate": 0.000599495706001445,
      "loss": 2.7238,
      "step": 4253
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.8503823280334473,
      "learning_rate": 0.0005994954688926417,
      "loss": 2.4332,
      "step": 4254
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.8002245426177979,
      "learning_rate": 0.0005994952317281566,
      "loss": 2.6952,
      "step": 4255
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.113362431526184,
      "learning_rate": 0.0005994949945079897,
      "loss": 2.5407,
      "step": 4256
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.8763629198074341,
      "learning_rate": 0.000599494757232141,
      "loss": 2.6743,
      "step": 4257
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.917454719543457,
      "learning_rate": 0.0005994945199006106,
      "loss": 2.8061,
      "step": 4258
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.8571025133132935,
      "learning_rate": 0.0005994942825133985,
      "loss": 2.766,
      "step": 4259
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.6976670622825623,
      "learning_rate": 0.0005994940450705047,
      "loss": 2.8097,
      "step": 4260
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.71614670753479,
      "learning_rate": 0.0005994938075719295,
      "loss": 2.721,
      "step": 4261
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.751318097114563,
      "learning_rate": 0.0005994935700176725,
      "loss": 2.7684,
      "step": 4262
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.7111673355102539,
      "learning_rate": 0.0005994933324077341,
      "loss": 2.7627,
      "step": 4263
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.6850671172142029,
      "learning_rate": 0.0005994930947421142,
      "loss": 2.4998,
      "step": 4264
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.7355556488037109,
      "learning_rate": 0.0005994928570208129,
      "loss": 2.8807,
      "step": 4265
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.8513423204421997,
      "learning_rate": 0.0005994926192438301,
      "loss": 2.6075,
      "step": 4266
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.6325476169586182,
      "learning_rate": 0.0005994923814111662,
      "loss": 2.7969,
      "step": 4267
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.8692309260368347,
      "learning_rate": 0.0005994921435228209,
      "loss": 2.6471,
      "step": 4268
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.049916386604309,
      "learning_rate": 0.0005994919055787943,
      "loss": 2.3904,
      "step": 4269
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.6377431750297546,
      "learning_rate": 0.0005994916675790866,
      "loss": 2.6934,
      "step": 4270
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.0803278684616089,
      "learning_rate": 0.0005994914295236977,
      "loss": 2.7335,
      "step": 4271
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.7238118052482605,
      "learning_rate": 0.0005994911914126278,
      "loss": 2.6515,
      "step": 4272
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.6524741053581238,
      "learning_rate": 0.0005994909532458767,
      "loss": 2.7499,
      "step": 4273
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.8543256521224976,
      "learning_rate": 0.0005994907150234447,
      "loss": 2.9047,
      "step": 4274
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.723691999912262,
      "learning_rate": 0.0005994904767453315,
      "loss": 2.7179,
      "step": 4275
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.6929808855056763,
      "learning_rate": 0.0005994902384115375,
      "loss": 2.5823,
      "step": 4276
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.6972787976264954,
      "learning_rate": 0.0005994900000220627,
      "loss": 2.5871,
      "step": 4277
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.8318750858306885,
      "learning_rate": 0.0005994897615769069,
      "loss": 2.6498,
      "step": 4278
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.7429981231689453,
      "learning_rate": 0.0005994895230760703,
      "loss": 2.9242,
      "step": 4279
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.6741469502449036,
      "learning_rate": 0.0005994892845195531,
      "loss": 2.4815,
      "step": 4280
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.8794137835502625,
      "learning_rate": 0.0005994890459073552,
      "loss": 2.7776,
      "step": 4281
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.6352263689041138,
      "learning_rate": 0.0005994888072394766,
      "loss": 2.4673,
      "step": 4282
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.6954870820045471,
      "learning_rate": 0.0005994885685159172,
      "loss": 2.7777,
      "step": 4283
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.8567189574241638,
      "learning_rate": 0.0005994883297366773,
      "loss": 2.4303,
      "step": 4284
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.7060012221336365,
      "learning_rate": 0.000599488090901757,
      "loss": 2.516,
      "step": 4285
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.9537699818611145,
      "learning_rate": 0.0005994878520111561,
      "loss": 2.8164,
      "step": 4286
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.7241714000701904,
      "learning_rate": 0.0005994876130648747,
      "loss": 2.8978,
      "step": 4287
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.7945238351821899,
      "learning_rate": 0.0005994873740629129,
      "loss": 2.568,
      "step": 4288
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.6464090943336487,
      "learning_rate": 0.0005994871350052709,
      "loss": 2.4986,
      "step": 4289
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.9289906024932861,
      "learning_rate": 0.0005994868958919485,
      "loss": 2.7085,
      "step": 4290
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.7949764728546143,
      "learning_rate": 0.0005994866567229458,
      "loss": 2.7499,
      "step": 4291
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.7988872528076172,
      "learning_rate": 0.0005994864174982629,
      "loss": 2.7998,
      "step": 4292
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.8676230907440186,
      "learning_rate": 0.0005994861782178998,
      "loss": 2.6324,
      "step": 4293
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.6453238725662231,
      "learning_rate": 0.0005994859388818567,
      "loss": 2.611,
      "step": 4294
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.7081882357597351,
      "learning_rate": 0.0005994856994901334,
      "loss": 2.4466,
      "step": 4295
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.7410033345222473,
      "learning_rate": 0.0005994854600427301,
      "loss": 2.5341,
      "step": 4296
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.6615664958953857,
      "learning_rate": 0.0005994852205396467,
      "loss": 2.506,
      "step": 4297
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.66382896900177,
      "learning_rate": 0.0005994849809808835,
      "loss": 2.5708,
      "step": 4298
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.8168075680732727,
      "learning_rate": 0.0005994847413664403,
      "loss": 2.4884,
      "step": 4299
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.702949583530426,
      "learning_rate": 0.0005994845016963171,
      "loss": 2.8877,
      "step": 4300
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.7400166392326355,
      "learning_rate": 0.0005994842619705142,
      "loss": 2.7189,
      "step": 4301
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.6741210222244263,
      "learning_rate": 0.0005994840221890316,
      "loss": 2.6267,
      "step": 4302
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.6429733633995056,
      "learning_rate": 0.0005994837823518692,
      "loss": 2.6862,
      "step": 4303
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.7690026164054871,
      "learning_rate": 0.0005994835424590271,
      "loss": 2.6579,
      "step": 4304
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.693341076374054,
      "learning_rate": 0.0005994833025105054,
      "loss": 2.82,
      "step": 4305
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.7691707611083984,
      "learning_rate": 0.000599483062506304,
      "loss": 2.6435,
      "step": 4306
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.7802783846855164,
      "learning_rate": 0.0005994828224464232,
      "loss": 2.6914,
      "step": 4307
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.6879507899284363,
      "learning_rate": 0.0005994825823308627,
      "loss": 2.7081,
      "step": 4308
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.712754487991333,
      "learning_rate": 0.0005994823421596229,
      "loss": 2.7236,
      "step": 4309
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.7330704927444458,
      "learning_rate": 0.0005994821019327036,
      "loss": 2.816,
      "step": 4310
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.6714509129524231,
      "learning_rate": 0.000599481861650105,
      "loss": 2.8332,
      "step": 4311
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.6779072880744934,
      "learning_rate": 0.000599481621311827,
      "loss": 2.7673,
      "step": 4312
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.9973324537277222,
      "learning_rate": 0.0005994813809178697,
      "loss": 2.4397,
      "step": 4313
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.9851924180984497,
      "learning_rate": 0.0005994811404682332,
      "loss": 2.8536,
      "step": 4314
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.726045548915863,
      "learning_rate": 0.0005994808999629176,
      "loss": 2.8578,
      "step": 4315
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.7210597991943359,
      "learning_rate": 0.0005994806594019226,
      "loss": 2.8102,
      "step": 4316
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.7436721920967102,
      "learning_rate": 0.0005994804187852487,
      "loss": 2.9214,
      "step": 4317
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.8471022248268127,
      "learning_rate": 0.0005994801781128956,
      "loss": 2.6506,
      "step": 4318
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.79554283618927,
      "learning_rate": 0.0005994799373848636,
      "loss": 2.7125,
      "step": 4319
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.7324638366699219,
      "learning_rate": 0.0005994796966011525,
      "loss": 2.3948,
      "step": 4320
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.910800576210022,
      "learning_rate": 0.0005994794557617626,
      "loss": 2.6548,
      "step": 4321
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.7446284294128418,
      "learning_rate": 0.0005994792148666936,
      "loss": 2.3884,
      "step": 4322
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.7614309191703796,
      "learning_rate": 0.000599478973915946,
      "loss": 2.5052,
      "step": 4323
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.7920643091201782,
      "learning_rate": 0.0005994787329095196,
      "loss": 2.5601,
      "step": 4324
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.7601531744003296,
      "learning_rate": 0.0005994784918474143,
      "loss": 2.6307,
      "step": 4325
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.7162445783615112,
      "learning_rate": 0.0005994782507296304,
      "loss": 2.6206,
      "step": 4326
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.7846827507019043,
      "learning_rate": 0.0005994780095561679,
      "loss": 2.5645,
      "step": 4327
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.8960736989974976,
      "learning_rate": 0.0005994777683270266,
      "loss": 2.7213,
      "step": 4328
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.6987161636352539,
      "learning_rate": 0.0005994775270422068,
      "loss": 2.8398,
      "step": 4329
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.1052615642547607,
      "learning_rate": 0.0005994772857017086,
      "loss": 2.5269,
      "step": 4330
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.7347179055213928,
      "learning_rate": 0.0005994770443055317,
      "loss": 2.6815,
      "step": 4331
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.7400996088981628,
      "learning_rate": 0.0005994768028536766,
      "loss": 2.7432,
      "step": 4332
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.9536604285240173,
      "learning_rate": 0.0005994765613461429,
      "loss": 2.5076,
      "step": 4333
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.7936434745788574,
      "learning_rate": 0.000599476319782931,
      "loss": 2.7761,
      "step": 4334
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.7000899910926819,
      "learning_rate": 0.0005994760781640407,
      "loss": 2.3341,
      "step": 4335
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.8100899457931519,
      "learning_rate": 0.0005994758364894722,
      "loss": 2.7771,
      "step": 4336
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.7162991762161255,
      "learning_rate": 0.0005994755947592254,
      "loss": 2.5855,
      "step": 4337
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.8710173964500427,
      "learning_rate": 0.0005994753529733006,
      "loss": 2.5826,
      "step": 4338
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.7011458277702332,
      "learning_rate": 0.0005994751111316975,
      "loss": 2.4381,
      "step": 4339
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.6541113257408142,
      "learning_rate": 0.0005994748692344166,
      "loss": 2.7279,
      "step": 4340
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.7374536395072937,
      "learning_rate": 0.0005994746272814574,
      "loss": 2.8108,
      "step": 4341
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.7474557757377625,
      "learning_rate": 0.0005994743852728203,
      "loss": 2.5904,
      "step": 4342
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.7215690612792969,
      "learning_rate": 0.0005994741432085053,
      "loss": 2.6377,
      "step": 4343
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.7657909393310547,
      "learning_rate": 0.0005994739010885123,
      "loss": 2.6675,
      "step": 4344
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.8154903054237366,
      "learning_rate": 0.0005994736589128415,
      "loss": 2.772,
      "step": 4345
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.698173463344574,
      "learning_rate": 0.0005994734166814929,
      "loss": 2.685,
      "step": 4346
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.7368215918540955,
      "learning_rate": 0.0005994731743944665,
      "loss": 2.6986,
      "step": 4347
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.672296941280365,
      "learning_rate": 0.0005994729320517626,
      "loss": 2.9093,
      "step": 4348
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.7424764633178711,
      "learning_rate": 0.0005994726896533807,
      "loss": 2.7177,
      "step": 4349
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.6726154088973999,
      "learning_rate": 0.0005994724471993214,
      "loss": 2.4431,
      "step": 4350
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.826522946357727,
      "learning_rate": 0.0005994722046895844,
      "loss": 2.6349,
      "step": 4351
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.790768027305603,
      "learning_rate": 0.0005994719621241699,
      "loss": 2.6891,
      "step": 4352
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.7110176682472229,
      "learning_rate": 0.000599471719503078,
      "loss": 2.7427,
      "step": 4353
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.7209002375602722,
      "learning_rate": 0.0005994714768263085,
      "loss": 2.8042,
      "step": 4354
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.8498380780220032,
      "learning_rate": 0.0005994712340938617,
      "loss": 2.6234,
      "step": 4355
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.714859127998352,
      "learning_rate": 0.0005994709913057375,
      "loss": 2.4483,
      "step": 4356
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.7761055827140808,
      "learning_rate": 0.000599470748461936,
      "loss": 2.5709,
      "step": 4357
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.6939150094985962,
      "learning_rate": 0.0005994705055624573,
      "loss": 2.5264,
      "step": 4358
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.6936646103858948,
      "learning_rate": 0.0005994702626073013,
      "loss": 2.6252,
      "step": 4359
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.8031648397445679,
      "learning_rate": 0.0005994700195964682,
      "loss": 2.6948,
      "step": 4360
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.6985520124435425,
      "learning_rate": 0.0005994697765299579,
      "loss": 2.5325,
      "step": 4361
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.6966328024864197,
      "learning_rate": 0.0005994695334077706,
      "loss": 2.6989,
      "step": 4362
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.7939538955688477,
      "learning_rate": 0.0005994692902299061,
      "loss": 2.8853,
      "step": 4363
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.7155479192733765,
      "learning_rate": 0.0005994690469963647,
      "loss": 2.7883,
      "step": 4364
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.7517889738082886,
      "learning_rate": 0.0005994688037071463,
      "loss": 2.7241,
      "step": 4365
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.90159010887146,
      "learning_rate": 0.0005994685603622511,
      "loss": 2.7374,
      "step": 4366
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.6723689436912537,
      "learning_rate": 0.0005994683169616789,
      "loss": 2.5718,
      "step": 4367
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.7854658961296082,
      "learning_rate": 0.0005994680735054301,
      "loss": 2.9809,
      "step": 4368
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.6406004428863525,
      "learning_rate": 0.0005994678299935044,
      "loss": 2.8075,
      "step": 4369
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.7221360802650452,
      "learning_rate": 0.000599467586425902,
      "loss": 2.7373,
      "step": 4370
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.6290527582168579,
      "learning_rate": 0.0005994673428026229,
      "loss": 2.7627,
      "step": 4371
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.7325220704078674,
      "learning_rate": 0.0005994670991236672,
      "loss": 2.8077,
      "step": 4372
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.722088634967804,
      "learning_rate": 0.0005994668553890349,
      "loss": 2.6176,
      "step": 4373
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.6795843839645386,
      "learning_rate": 0.0005994666115987261,
      "loss": 2.7029,
      "step": 4374
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.7226772904396057,
      "learning_rate": 0.0005994663677527407,
      "loss": 2.7357,
      "step": 4375
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.6462410092353821,
      "learning_rate": 0.0005994661238510789,
      "loss": 2.7319,
      "step": 4376
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.6589025855064392,
      "learning_rate": 0.0005994658798937407,
      "loss": 2.5518,
      "step": 4377
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.698020339012146,
      "learning_rate": 0.0005994656358807261,
      "loss": 2.7181,
      "step": 4378
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.7631765007972717,
      "learning_rate": 0.0005994653918120353,
      "loss": 2.5154,
      "step": 4379
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.6910272240638733,
      "learning_rate": 0.0005994651476876682,
      "loss": 2.8517,
      "step": 4380
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.6982680559158325,
      "learning_rate": 0.0005994649035076248,
      "loss": 2.5307,
      "step": 4381
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.7143718600273132,
      "learning_rate": 0.0005994646592719053,
      "loss": 2.8033,
      "step": 4382
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.664978563785553,
      "learning_rate": 0.0005994644149805096,
      "loss": 2.7049,
      "step": 4383
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.6830800771713257,
      "learning_rate": 0.0005994641706334378,
      "loss": 2.3732,
      "step": 4384
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.7505698204040527,
      "learning_rate": 0.0005994639262306899,
      "loss": 2.6194,
      "step": 4385
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.6903601884841919,
      "learning_rate": 0.0005994636817722661,
      "loss": 2.7738,
      "step": 4386
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.6591743230819702,
      "learning_rate": 0.0005994634372581664,
      "loss": 2.7302,
      "step": 4387
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.728153645992279,
      "learning_rate": 0.0005994631926883907,
      "loss": 2.6808,
      "step": 4388
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.6827064752578735,
      "learning_rate": 0.0005994629480629393,
      "loss": 2.6753,
      "step": 4389
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.6195400357246399,
      "learning_rate": 0.0005994627033818119,
      "loss": 2.7218,
      "step": 4390
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.6467140913009644,
      "learning_rate": 0.0005994624586450088,
      "loss": 2.8049,
      "step": 4391
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.6662765741348267,
      "learning_rate": 0.00059946221385253,
      "loss": 2.6202,
      "step": 4392
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.7499473690986633,
      "learning_rate": 0.0005994619690043755,
      "loss": 2.5213,
      "step": 4393
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.7332100868225098,
      "learning_rate": 0.0005994617241005453,
      "loss": 2.77,
      "step": 4394
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.6902642846107483,
      "learning_rate": 0.0005994614791410395,
      "loss": 2.3215,
      "step": 4395
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.6887715458869934,
      "learning_rate": 0.0005994612341258583,
      "loss": 2.7293,
      "step": 4396
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.6784630417823792,
      "learning_rate": 0.0005994609890550015,
      "loss": 2.5636,
      "step": 4397
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.7324240207672119,
      "learning_rate": 0.0005994607439284693,
      "loss": 2.6527,
      "step": 4398
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.7601264715194702,
      "learning_rate": 0.0005994604987462616,
      "loss": 2.668,
      "step": 4399
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.7128495573997498,
      "learning_rate": 0.0005994602535083787,
      "loss": 2.5405,
      "step": 4400
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.7523803114891052,
      "learning_rate": 0.0005994600082148204,
      "loss": 2.8829,
      "step": 4401
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.7196244597434998,
      "learning_rate": 0.0005994597628655869,
      "loss": 2.571,
      "step": 4402
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.712841272354126,
      "learning_rate": 0.000599459517460678,
      "loss": 2.8094,
      "step": 4403
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.6682367324829102,
      "learning_rate": 0.0005994592720000941,
      "loss": 2.5908,
      "step": 4404
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.8497452139854431,
      "learning_rate": 0.000599459026483835,
      "loss": 2.6664,
      "step": 4405
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.698122501373291,
      "learning_rate": 0.0005994587809119008,
      "loss": 2.5866,
      "step": 4406
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.9090597033500671,
      "learning_rate": 0.0005994585352842916,
      "loss": 2.7482,
      "step": 4407
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.6796736121177673,
      "learning_rate": 0.0005994582896010073,
      "loss": 2.6679,
      "step": 4408
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.9263333678245544,
      "learning_rate": 0.0005994580438620482,
      "loss": 2.6718,
      "step": 4409
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.6609911918640137,
      "learning_rate": 0.0005994577980674141,
      "loss": 2.7806,
      "step": 4410
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.8719550371170044,
      "learning_rate": 0.0005994575522171052,
      "loss": 2.7349,
      "step": 4411
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.8508161902427673,
      "learning_rate": 0.0005994573063111215,
      "loss": 2.6138,
      "step": 4412
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.7591286301612854,
      "learning_rate": 0.0005994570603494629,
      "loss": 2.4208,
      "step": 4413
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.8290379643440247,
      "learning_rate": 0.0005994568143321297,
      "loss": 2.9033,
      "step": 4414
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.6360684037208557,
      "learning_rate": 0.0005994565682591219,
      "loss": 2.7718,
      "step": 4415
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.7053099870681763,
      "learning_rate": 0.0005994563221304393,
      "loss": 2.6576,
      "step": 4416
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.6680113077163696,
      "learning_rate": 0.0005994560759460822,
      "loss": 2.7566,
      "step": 4417
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.7189313173294067,
      "learning_rate": 0.0005994558297060505,
      "loss": 2.8308,
      "step": 4418
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.7057603001594543,
      "learning_rate": 0.0005994555834103444,
      "loss": 2.3308,
      "step": 4419
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.7805272936820984,
      "learning_rate": 0.0005994553370589639,
      "loss": 2.8134,
      "step": 4420
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.6450151205062866,
      "learning_rate": 0.0005994550906519089,
      "loss": 2.5335,
      "step": 4421
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.6404420137405396,
      "learning_rate": 0.0005994548441891795,
      "loss": 2.4445,
      "step": 4422
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.6676235795021057,
      "learning_rate": 0.000599454597670776,
      "loss": 2.7771,
      "step": 4423
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.7009639143943787,
      "learning_rate": 0.000599454351096698,
      "loss": 2.7302,
      "step": 4424
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.6666666269302368,
      "learning_rate": 0.0005994541044669459,
      "loss": 2.7236,
      "step": 4425
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.7182052731513977,
      "learning_rate": 0.0005994538577815195,
      "loss": 2.6406,
      "step": 4426
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.7004602551460266,
      "learning_rate": 0.0005994536110404191,
      "loss": 2.9975,
      "step": 4427
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.7417539954185486,
      "learning_rate": 0.0005994533642436447,
      "loss": 2.6611,
      "step": 4428
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.7170401215553284,
      "learning_rate": 0.0005994531173911961,
      "loss": 2.6595,
      "step": 4429
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.6927629709243774,
      "learning_rate": 0.0005994528704830735,
      "loss": 2.7017,
      "step": 4430
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.8115329146385193,
      "learning_rate": 0.0005994526235192771,
      "loss": 2.9767,
      "step": 4431
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.78657466173172,
      "learning_rate": 0.0005994523764998068,
      "loss": 2.8146,
      "step": 4432
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.698991060256958,
      "learning_rate": 0.0005994521294246625,
      "loss": 2.6342,
      "step": 4433
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.6591503620147705,
      "learning_rate": 0.0005994518822938446,
      "loss": 2.6517,
      "step": 4434
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.674163281917572,
      "learning_rate": 0.0005994516351073528,
      "loss": 2.6794,
      "step": 4435
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.7000173926353455,
      "learning_rate": 0.0005994513878651873,
      "loss": 2.8393,
      "step": 4436
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.75612473487854,
      "learning_rate": 0.0005994511405673482,
      "loss": 2.7472,
      "step": 4437
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.6583110690116882,
      "learning_rate": 0.0005994508932138353,
      "loss": 2.8649,
      "step": 4438
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.8609763979911804,
      "learning_rate": 0.0005994506458046491,
      "loss": 2.8128,
      "step": 4439
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.720325767993927,
      "learning_rate": 0.0005994503983397892,
      "loss": 2.7989,
      "step": 4440
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.7116880416870117,
      "learning_rate": 0.0005994501508192558,
      "loss": 2.6217,
      "step": 4441
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.6515384912490845,
      "learning_rate": 0.0005994499032430491,
      "loss": 2.5649,
      "step": 4442
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.6439380049705505,
      "learning_rate": 0.0005994496556111689,
      "loss": 2.5611,
      "step": 4443
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.7281669974327087,
      "learning_rate": 0.0005994494079236154,
      "loss": 2.4994,
      "step": 4444
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.684240996837616,
      "learning_rate": 0.0005994491601803886,
      "loss": 2.8141,
      "step": 4445
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.7346693873405457,
      "learning_rate": 0.0005994489123814885,
      "loss": 2.5917,
      "step": 4446
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.6722996830940247,
      "learning_rate": 0.0005994486645269153,
      "loss": 2.7172,
      "step": 4447
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.6671216487884521,
      "learning_rate": 0.0005994484166166688,
      "loss": 2.7148,
      "step": 4448
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.6696630716323853,
      "learning_rate": 0.0005994481686507493,
      "loss": 2.8585,
      "step": 4449
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.6895265579223633,
      "learning_rate": 0.0005994479206291567,
      "loss": 2.707,
      "step": 4450
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.6431670784950256,
      "learning_rate": 0.000599447672551891,
      "loss": 2.6835,
      "step": 4451
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.7180628180503845,
      "learning_rate": 0.0005994474244189525,
      "loss": 2.6976,
      "step": 4452
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.6387816071510315,
      "learning_rate": 0.0005994471762303409,
      "loss": 2.524,
      "step": 4453
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.6874213814735413,
      "learning_rate": 0.0005994469279860565,
      "loss": 2.7003,
      "step": 4454
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.6535957455635071,
      "learning_rate": 0.0005994466796860993,
      "loss": 2.9196,
      "step": 4455
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.6808627843856812,
      "learning_rate": 0.0005994464313304693,
      "loss": 2.4673,
      "step": 4456
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.6284866333007812,
      "learning_rate": 0.0005994461829191664,
      "loss": 2.794,
      "step": 4457
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.6562323570251465,
      "learning_rate": 0.000599445934452191,
      "loss": 2.5826,
      "step": 4458
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.777824342250824,
      "learning_rate": 0.0005994456859295428,
      "loss": 2.5615,
      "step": 4459
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.6878286600112915,
      "learning_rate": 0.0005994454373512221,
      "loss": 2.806,
      "step": 4460
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.9461756348609924,
      "learning_rate": 0.0005994451887172288,
      "loss": 2.6467,
      "step": 4461
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.7807666659355164,
      "learning_rate": 0.0005994449400275629,
      "loss": 2.6296,
      "step": 4462
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.8499508500099182,
      "learning_rate": 0.0005994446912822247,
      "loss": 2.5581,
      "step": 4463
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.8111302852630615,
      "learning_rate": 0.000599444442481214,
      "loss": 3.0196,
      "step": 4464
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.6981711387634277,
      "learning_rate": 0.0005994441936245309,
      "loss": 2.6162,
      "step": 4465
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.2244752645492554,
      "learning_rate": 0.0005994439447121755,
      "loss": 2.6102,
      "step": 4466
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.7791959643363953,
      "learning_rate": 0.0005994436957441478,
      "loss": 2.7942,
      "step": 4467
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.7463213801383972,
      "learning_rate": 0.0005994434467204478,
      "loss": 2.4803,
      "step": 4468
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.8784834742546082,
      "learning_rate": 0.0005994431976410758,
      "loss": 2.5189,
      "step": 4469
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.83497154712677,
      "learning_rate": 0.0005994429485060316,
      "loss": 2.7075,
      "step": 4470
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.7226394414901733,
      "learning_rate": 0.0005994426993153151,
      "loss": 2.7581,
      "step": 4471
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.8866817951202393,
      "learning_rate": 0.0005994424500689268,
      "loss": 2.7824,
      "step": 4472
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.0223572254180908,
      "learning_rate": 0.0005994422007668664,
      "loss": 2.7055,
      "step": 4473
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.6652824282646179,
      "learning_rate": 0.000599441951409134,
      "loss": 2.9947,
      "step": 4474
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.8360607624053955,
      "learning_rate": 0.0005994417019957298,
      "loss": 2.6533,
      "step": 4475
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.9515260457992554,
      "learning_rate": 0.0005994414525266537,
      "loss": 2.5875,
      "step": 4476
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.7346062660217285,
      "learning_rate": 0.0005994412030019057,
      "loss": 2.7383,
      "step": 4477
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.7711383700370789,
      "learning_rate": 0.0005994409534214859,
      "loss": 2.8937,
      "step": 4478
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.9076512455940247,
      "learning_rate": 0.0005994407037853945,
      "loss": 2.7728,
      "step": 4479
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.7654919624328613,
      "learning_rate": 0.0005994404540936314,
      "loss": 2.8141,
      "step": 4480
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.7268379926681519,
      "learning_rate": 0.0005994402043461967,
      "loss": 3.0082,
      "step": 4481
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.9950087070465088,
      "learning_rate": 0.0005994399545430902,
      "loss": 2.5391,
      "step": 4482
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.6817883849143982,
      "learning_rate": 0.0005994397046843123,
      "loss": 2.4621,
      "step": 4483
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.760494589805603,
      "learning_rate": 0.0005994394547698629,
      "loss": 2.856,
      "step": 4484
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.9577007293701172,
      "learning_rate": 0.0005994392047997421,
      "loss": 2.4281,
      "step": 4485
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.7189103364944458,
      "learning_rate": 0.0005994389547739499,
      "loss": 2.7365,
      "step": 4486
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.7684584856033325,
      "learning_rate": 0.0005994387046924863,
      "loss": 2.5859,
      "step": 4487
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.0970896482467651,
      "learning_rate": 0.0005994384545553515,
      "loss": 2.6709,
      "step": 4488
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.7661418914794922,
      "learning_rate": 0.0005994382043625453,
      "loss": 2.709,
      "step": 4489
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.7217205166816711,
      "learning_rate": 0.0005994379541140679,
      "loss": 2.6914,
      "step": 4490
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.8284767866134644,
      "learning_rate": 0.0005994377038099194,
      "loss": 2.6079,
      "step": 4491
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.7088526487350464,
      "learning_rate": 0.0005994374534500997,
      "loss": 2.5074,
      "step": 4492
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.6158615350723267,
      "learning_rate": 0.000599437203034609,
      "loss": 2.5032,
      "step": 4493
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.890636146068573,
      "learning_rate": 0.0005994369525634472,
      "loss": 2.6594,
      "step": 4494
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.7234785556793213,
      "learning_rate": 0.0005994367020366145,
      "loss": 2.429,
      "step": 4495
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.6360534429550171,
      "learning_rate": 0.0005994364514541108,
      "loss": 2.5156,
      "step": 4496
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.9203214049339294,
      "learning_rate": 0.0005994362008159362,
      "loss": 2.8979,
      "step": 4497
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.7732685804367065,
      "learning_rate": 0.0005994359501220909,
      "loss": 2.8219,
      "step": 4498
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.676368236541748,
      "learning_rate": 0.0005994356993725747,
      "loss": 2.5917,
      "step": 4499
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.7022691965103149,
      "learning_rate": 0.0005994354485673877,
      "loss": 2.5589,
      "step": 4500
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.7372612953186035,
      "learning_rate": 0.00059943519770653,
      "loss": 2.5122,
      "step": 4501
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.6894931197166443,
      "learning_rate": 0.0005994349467900017,
      "loss": 2.7443,
      "step": 4502
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.7106151580810547,
      "learning_rate": 0.0005994346958178028,
      "loss": 2.4335,
      "step": 4503
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.0513405799865723,
      "learning_rate": 0.0005994344447899333,
      "loss": 2.736,
      "step": 4504
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.7501143217086792,
      "learning_rate": 0.0005994341937063933,
      "loss": 2.7406,
      "step": 4505
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.92420893907547,
      "learning_rate": 0.0005994339425671829,
      "loss": 2.43,
      "step": 4506
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.7627885937690735,
      "learning_rate": 0.000599433691372302,
      "loss": 2.8119,
      "step": 4507
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.7136853337287903,
      "learning_rate": 0.0005994334401217507,
      "loss": 2.6191,
      "step": 4508
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.6780360341072083,
      "learning_rate": 0.0005994331888155291,
      "loss": 2.5241,
      "step": 4509
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.9073540568351746,
      "learning_rate": 0.0005994329374536372,
      "loss": 2.7807,
      "step": 4510
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.7689542770385742,
      "learning_rate": 0.0005994326860360751,
      "loss": 2.6257,
      "step": 4511
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.6737000346183777,
      "learning_rate": 0.0005994324345628429,
      "loss": 2.5483,
      "step": 4512
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.2027828693389893,
      "learning_rate": 0.0005994321830339403,
      "loss": 2.604,
      "step": 4513
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.7306455373764038,
      "learning_rate": 0.0005994319314493678,
      "loss": 2.7216,
      "step": 4514
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.6654891967773438,
      "learning_rate": 0.0005994316798091253,
      "loss": 2.6721,
      "step": 4515
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.8715664148330688,
      "learning_rate": 0.0005994314281132127,
      "loss": 2.7454,
      "step": 4516
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.0093319416046143,
      "learning_rate": 0.0005994311763616301,
      "loss": 2.701,
      "step": 4517
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.6700738668441772,
      "learning_rate": 0.0005994309245543777,
      "loss": 2.7985,
      "step": 4518
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.9188802242279053,
      "learning_rate": 0.0005994306726914555,
      "loss": 2.8574,
      "step": 4519
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.9774994850158691,
      "learning_rate": 0.0005994304207728632,
      "loss": 2.5919,
      "step": 4520
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.6591722965240479,
      "learning_rate": 0.0005994301687986014,
      "loss": 2.4666,
      "step": 4521
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.8921318650245667,
      "learning_rate": 0.0005994299167686697,
      "loss": 2.4973,
      "step": 4522
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.72304368019104,
      "learning_rate": 0.0005994296646830684,
      "loss": 2.532,
      "step": 4523
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.7597880363464355,
      "learning_rate": 0.0005994294125417976,
      "loss": 2.7958,
      "step": 4524
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.6693265438079834,
      "learning_rate": 0.000599429160344857,
      "loss": 2.5846,
      "step": 4525
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.7648232579231262,
      "learning_rate": 0.0005994289080922471,
      "loss": 2.5135,
      "step": 4526
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.7537727355957031,
      "learning_rate": 0.0005994286557839675,
      "loss": 2.6984,
      "step": 4527
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.767255961894989,
      "learning_rate": 0.0005994284034200186,
      "loss": 2.4281,
      "step": 4528
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.8862345218658447,
      "learning_rate": 0.0005994281510004003,
      "loss": 2.6756,
      "step": 4529
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.7910726070404053,
      "learning_rate": 0.0005994278985251125,
      "loss": 2.7564,
      "step": 4530
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.6289495825767517,
      "learning_rate": 0.0005994276459941556,
      "loss": 2.7877,
      "step": 4531
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.7190102934837341,
      "learning_rate": 0.0005994273934075294,
      "loss": 2.5901,
      "step": 4532
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.7307120561599731,
      "learning_rate": 0.0005994271407652339,
      "loss": 2.8206,
      "step": 4533
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.6963369846343994,
      "learning_rate": 0.0005994268880672693,
      "loss": 2.6739,
      "step": 4534
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.6809975504875183,
      "learning_rate": 0.0005994266353136356,
      "loss": 2.7633,
      "step": 4535
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.699609100818634,
      "learning_rate": 0.0005994263825043328,
      "loss": 2.7724,
      "step": 4536
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.800849437713623,
      "learning_rate": 0.000599426129639361,
      "loss": 2.7,
      "step": 4537
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.7058446407318115,
      "learning_rate": 0.0005994258767187203,
      "loss": 2.5795,
      "step": 4538
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.7058989405632019,
      "learning_rate": 0.0005994256237424106,
      "loss": 2.6127,
      "step": 4539
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.6951714754104614,
      "learning_rate": 0.000599425370710432,
      "loss": 2.4564,
      "step": 4540
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.7728661298751831,
      "learning_rate": 0.0005994251176227847,
      "loss": 2.8211,
      "step": 4541
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.6640047430992126,
      "learning_rate": 0.0005994248644794685,
      "loss": 2.6804,
      "step": 4542
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.7759060263633728,
      "learning_rate": 0.0005994246112804836,
      "loss": 2.7173,
      "step": 4543
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.8433200120925903,
      "learning_rate": 0.00059942435802583,
      "loss": 2.5766,
      "step": 4544
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.7745896577835083,
      "learning_rate": 0.0005994241047155077,
      "loss": 2.6888,
      "step": 4545
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.752890408039093,
      "learning_rate": 0.0005994238513495169,
      "loss": 2.7569,
      "step": 4546
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.7726640105247498,
      "learning_rate": 0.0005994235979278575,
      "loss": 2.5975,
      "step": 4547
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.6486309170722961,
      "learning_rate": 0.0005994233444505297,
      "loss": 2.5984,
      "step": 4548
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.7621941566467285,
      "learning_rate": 0.0005994230909175333,
      "loss": 2.6289,
      "step": 4549
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.6562494039535522,
      "learning_rate": 0.0005994228373288686,
      "loss": 2.8066,
      "step": 4550
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.7198943495750427,
      "learning_rate": 0.0005994225836845356,
      "loss": 2.675,
      "step": 4551
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.7004513144493103,
      "learning_rate": 0.0005994223299845341,
      "loss": 2.6448,
      "step": 4552
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.7338811159133911,
      "learning_rate": 0.0005994220762288645,
      "loss": 2.7411,
      "step": 4553
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.6456345319747925,
      "learning_rate": 0.0005994218224175266,
      "loss": 2.851,
      "step": 4554
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.6860756278038025,
      "learning_rate": 0.0005994215685505205,
      "loss": 2.4768,
      "step": 4555
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.7656115889549255,
      "learning_rate": 0.0005994213146278465,
      "loss": 2.7758,
      "step": 4556
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.7117305994033813,
      "learning_rate": 0.0005994210606495042,
      "loss": 2.6084,
      "step": 4557
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.6604520082473755,
      "learning_rate": 0.0005994208066154938,
      "loss": 2.739,
      "step": 4558
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.6627408266067505,
      "learning_rate": 0.0005994205525258156,
      "loss": 2.6203,
      "step": 4559
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.8011198043823242,
      "learning_rate": 0.0005994202983804696,
      "loss": 2.6743,
      "step": 4560
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.639589250087738,
      "learning_rate": 0.0005994200441794555,
      "loss": 2.6372,
      "step": 4561
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.6937230229377747,
      "learning_rate": 0.0005994197899227736,
      "loss": 2.6495,
      "step": 4562
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.7751061916351318,
      "learning_rate": 0.0005994195356104239,
      "loss": 2.4658,
      "step": 4563
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.6251554489135742,
      "learning_rate": 0.0005994192812424064,
      "loss": 2.6767,
      "step": 4564
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.7837610244750977,
      "learning_rate": 0.0005994190268187212,
      "loss": 2.772,
      "step": 4565
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.6328862905502319,
      "learning_rate": 0.0005994187723393685,
      "loss": 2.671,
      "step": 4566
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.6331732869148254,
      "learning_rate": 0.0005994185178043481,
      "loss": 2.4843,
      "step": 4567
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.6636321544647217,
      "learning_rate": 0.0005994182632136602,
      "loss": 2.552,
      "step": 4568
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.6622540354728699,
      "learning_rate": 0.0005994180085673047,
      "loss": 2.8834,
      "step": 4569
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.7104930877685547,
      "learning_rate": 0.0005994177538652818,
      "loss": 2.5125,
      "step": 4570
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.6836028099060059,
      "learning_rate": 0.0005994174991075915,
      "loss": 2.7944,
      "step": 4571
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.7802362442016602,
      "learning_rate": 0.0005994172442942338,
      "loss": 2.8684,
      "step": 4572
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.6689098477363586,
      "learning_rate": 0.0005994169894252088,
      "loss": 2.7016,
      "step": 4573
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.6592346429824829,
      "learning_rate": 0.0005994167345005166,
      "loss": 2.5244,
      "step": 4574
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.708012044429779,
      "learning_rate": 0.0005994164795201571,
      "loss": 2.3622,
      "step": 4575
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.663794219493866,
      "learning_rate": 0.0005994162244841304,
      "loss": 2.7289,
      "step": 4576
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.689582884311676,
      "learning_rate": 0.0005994159693924366,
      "loss": 2.5876,
      "step": 4577
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.6794767379760742,
      "learning_rate": 0.0005994157142450757,
      "loss": 2.5843,
      "step": 4578
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.6770890355110168,
      "learning_rate": 0.0005994154590420478,
      "loss": 2.6181,
      "step": 4579
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.693732738494873,
      "learning_rate": 0.000599415203783353,
      "loss": 2.6395,
      "step": 4580
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.737111508846283,
      "learning_rate": 0.0005994149484689911,
      "loss": 2.4697,
      "step": 4581
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.7514880299568176,
      "learning_rate": 0.0005994146930989624,
      "loss": 2.7191,
      "step": 4582
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.6814762353897095,
      "learning_rate": 0.0005994144376732668,
      "loss": 2.5871,
      "step": 4583
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.8405857086181641,
      "learning_rate": 0.0005994141821919045,
      "loss": 2.7621,
      "step": 4584
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.7398766279220581,
      "learning_rate": 0.0005994139266548754,
      "loss": 2.7363,
      "step": 4585
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.0660003423690796,
      "learning_rate": 0.0005994136710621795,
      "loss": 2.58,
      "step": 4586
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.813934326171875,
      "learning_rate": 0.0005994134154138171,
      "loss": 2.7069,
      "step": 4587
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.6898792386054993,
      "learning_rate": 0.0005994131597097881,
      "loss": 2.6332,
      "step": 4588
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.102766990661621,
      "learning_rate": 0.0005994129039500923,
      "loss": 2.6957,
      "step": 4589
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.6781343221664429,
      "learning_rate": 0.0005994126481347303,
      "loss": 2.8323,
      "step": 4590
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.7730122208595276,
      "learning_rate": 0.0005994123922637016,
      "loss": 2.6724,
      "step": 4591
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.749010443687439,
      "learning_rate": 0.0005994121363370066,
      "loss": 2.7538,
      "step": 4592
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.6998012661933899,
      "learning_rate": 0.0005994118803546453,
      "loss": 2.6068,
      "step": 4593
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.6544800400733948,
      "learning_rate": 0.0005994116243166176,
      "loss": 2.5313,
      "step": 4594
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.8247462511062622,
      "learning_rate": 0.0005994113682229236,
      "loss": 2.4796,
      "step": 4595
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.751666784286499,
      "learning_rate": 0.0005994111120735633,
      "loss": 2.6789,
      "step": 4596
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.855445384979248,
      "learning_rate": 0.0005994108558685369,
      "loss": 2.6655,
      "step": 4597
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.7468389868736267,
      "learning_rate": 0.0005994105996078444,
      "loss": 2.6082,
      "step": 4598
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.6348046064376831,
      "learning_rate": 0.0005994103432914858,
      "loss": 2.6923,
      "step": 4599
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.8636220693588257,
      "learning_rate": 0.0005994100869194612,
      "loss": 2.6639,
      "step": 4600
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.9116708040237427,
      "learning_rate": 0.0005994098304917706,
      "loss": 2.5732,
      "step": 4601
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.667873740196228,
      "learning_rate": 0.0005994095740084141,
      "loss": 2.8323,
      "step": 4602
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.8525484204292297,
      "learning_rate": 0.0005994093174693916,
      "loss": 2.7,
      "step": 4603
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.7410055994987488,
      "learning_rate": 0.0005994090608747034,
      "loss": 2.8557,
      "step": 4604
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.7708554863929749,
      "learning_rate": 0.0005994088042243492,
      "loss": 2.7097,
      "step": 4605
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.7376061081886292,
      "learning_rate": 0.0005994085475183295,
      "loss": 2.7243,
      "step": 4606
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.8711332082748413,
      "learning_rate": 0.0005994082907566439,
      "loss": 2.4339,
      "step": 4607
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.6809974908828735,
      "learning_rate": 0.0005994080339392926,
      "loss": 2.7721,
      "step": 4608
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.7896355390548706,
      "learning_rate": 0.0005994077770662758,
      "loss": 2.49,
      "step": 4609
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.9400899410247803,
      "learning_rate": 0.0005994075201375935,
      "loss": 2.7345,
      "step": 4610
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.7781103253364563,
      "learning_rate": 0.0005994072631532457,
      "loss": 2.5272,
      "step": 4611
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.6951726078987122,
      "learning_rate": 0.0005994070061132324,
      "loss": 2.4725,
      "step": 4612
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.7740689516067505,
      "learning_rate": 0.0005994067490175535,
      "loss": 2.7183,
      "step": 4613
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.6325488090515137,
      "learning_rate": 0.0005994064918662095,
      "loss": 2.5463,
      "step": 4614
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.6834559440612793,
      "learning_rate": 0.0005994062346592001,
      "loss": 2.7141,
      "step": 4615
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.630506694316864,
      "learning_rate": 0.0005994059773965254,
      "loss": 2.7604,
      "step": 4616
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.6817820072174072,
      "learning_rate": 0.0005994057200781855,
      "loss": 2.556,
      "step": 4617
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.650219202041626,
      "learning_rate": 0.0005994054627041803,
      "loss": 2.5416,
      "step": 4618
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.7006505727767944,
      "learning_rate": 0.0005994052052745101,
      "loss": 2.6186,
      "step": 4619
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.7461041212081909,
      "learning_rate": 0.0005994049477891748,
      "loss": 2.6082,
      "step": 4620
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.719228982925415,
      "learning_rate": 0.0005994046902481745,
      "loss": 2.6513,
      "step": 4621
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.7679947018623352,
      "learning_rate": 0.0005994044326515092,
      "loss": 2.8935,
      "step": 4622
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.673590898513794,
      "learning_rate": 0.0005994041749991789,
      "loss": 2.5794,
      "step": 4623
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.7370790243148804,
      "learning_rate": 0.0005994039172911839,
      "loss": 2.4407,
      "step": 4624
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.678866446018219,
      "learning_rate": 0.0005994036595275238,
      "loss": 2.8582,
      "step": 4625
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.6633836627006531,
      "learning_rate": 0.000599403401708199,
      "loss": 2.5883,
      "step": 4626
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.7504116892814636,
      "learning_rate": 0.0005994031438332095,
      "loss": 2.7032,
      "step": 4627
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.7354314923286438,
      "learning_rate": 0.0005994028859025554,
      "loss": 2.5151,
      "step": 4628
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.8261439800262451,
      "learning_rate": 0.0005994026279162365,
      "loss": 2.475,
      "step": 4629
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.6837735772132874,
      "learning_rate": 0.0005994023698742529,
      "loss": 2.6224,
      "step": 4630
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.7499118447303772,
      "learning_rate": 0.000599402111776605,
      "loss": 2.636,
      "step": 4631
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.6623294949531555,
      "learning_rate": 0.0005994018536232924,
      "loss": 2.7966,
      "step": 4632
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.7159609198570251,
      "learning_rate": 0.0005994015954143156,
      "loss": 2.5663,
      "step": 4633
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.7051804065704346,
      "learning_rate": 0.0005994013371496741,
      "loss": 2.5433,
      "step": 4634
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.7335054278373718,
      "learning_rate": 0.0005994010788293684,
      "loss": 2.7954,
      "step": 4635
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.6512913703918457,
      "learning_rate": 0.0005994008204533984,
      "loss": 2.5276,
      "step": 4636
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.6944605112075806,
      "learning_rate": 0.000599400562021764,
      "loss": 2.7244,
      "step": 4637
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.6536509394645691,
      "learning_rate": 0.0005994003035344655,
      "loss": 2.5406,
      "step": 4638
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.7537279725074768,
      "learning_rate": 0.0005994000449915028,
      "loss": 2.7255,
      "step": 4639
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.7462537288665771,
      "learning_rate": 0.000599399786392876,
      "loss": 2.4638,
      "step": 4640
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.6552457213401794,
      "learning_rate": 0.0005993995277385851,
      "loss": 2.6547,
      "step": 4641
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.678758978843689,
      "learning_rate": 0.0005993992690286301,
      "loss": 2.7958,
      "step": 4642
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.7425903081893921,
      "learning_rate": 0.0005993990102630113,
      "loss": 2.8111,
      "step": 4643
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.7059619426727295,
      "learning_rate": 0.0005993987514417284,
      "loss": 2.6476,
      "step": 4644
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.6900714039802551,
      "learning_rate": 0.0005993984925647817,
      "loss": 2.4947,
      "step": 4645
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.670613706111908,
      "learning_rate": 0.0005993982336321712,
      "loss": 2.8674,
      "step": 4646
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.8671441674232483,
      "learning_rate": 0.0005993979746438969,
      "loss": 2.7557,
      "step": 4647
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.6518563032150269,
      "learning_rate": 0.0005993977155999588,
      "loss": 2.8581,
      "step": 4648
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.8597363233566284,
      "learning_rate": 0.0005993974565003571,
      "loss": 2.6801,
      "step": 4649
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.7045539617538452,
      "learning_rate": 0.0005993971973450918,
      "loss": 2.8297,
      "step": 4650
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.7139927744865417,
      "learning_rate": 0.0005993969381341628,
      "loss": 2.4543,
      "step": 4651
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.662979006767273,
      "learning_rate": 0.0005993966788675703,
      "loss": 2.373,
      "step": 4652
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.6704939007759094,
      "learning_rate": 0.0005993964195453144,
      "loss": 2.5954,
      "step": 4653
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.8205829858779907,
      "learning_rate": 0.0005993961601673948,
      "loss": 2.7074,
      "step": 4654
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.6430983543395996,
      "learning_rate": 0.000599395900733812,
      "loss": 2.6104,
      "step": 4655
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.6331321001052856,
      "learning_rate": 0.0005993956412445659,
      "loss": 2.7692,
      "step": 4656
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.7061254978179932,
      "learning_rate": 0.0005993953816996563,
      "loss": 2.6588,
      "step": 4657
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.7942554354667664,
      "learning_rate": 0.0005993951220990836,
      "loss": 2.8692,
      "step": 4658
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.6859206557273865,
      "learning_rate": 0.0005993948624428477,
      "loss": 2.9918,
      "step": 4659
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.661747932434082,
      "learning_rate": 0.0005993946027309485,
      "loss": 2.5186,
      "step": 4660
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.6682375073432922,
      "learning_rate": 0.0005993943429633864,
      "loss": 2.6204,
      "step": 4661
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.652504563331604,
      "learning_rate": 0.000599394083140161,
      "loss": 2.5997,
      "step": 4662
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.687645673751831,
      "learning_rate": 0.0005993938232612727,
      "loss": 2.6373,
      "step": 4663
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.6861310601234436,
      "learning_rate": 0.0005993935633267215,
      "loss": 2.6293,
      "step": 4664
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.7224230170249939,
      "learning_rate": 0.0005993933033365073,
      "loss": 2.7397,
      "step": 4665
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.7040099501609802,
      "learning_rate": 0.0005993930432906302,
      "loss": 2.776,
      "step": 4666
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.8509358763694763,
      "learning_rate": 0.0005993927831890904,
      "loss": 2.789,
      "step": 4667
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.6688058972358704,
      "learning_rate": 0.0005993925230318878,
      "loss": 2.5277,
      "step": 4668
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.7276569604873657,
      "learning_rate": 0.0005993922628190225,
      "loss": 2.685,
      "step": 4669
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.8855317234992981,
      "learning_rate": 0.0005993920025504945,
      "loss": 2.611,
      "step": 4670
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.8274430632591248,
      "learning_rate": 0.0005993917422263039,
      "loss": 2.3768,
      "step": 4671
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.7539070844650269,
      "learning_rate": 0.0005993914818464507,
      "loss": 2.8065,
      "step": 4672
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.7976836562156677,
      "learning_rate": 0.0005993912214109348,
      "loss": 2.7803,
      "step": 4673
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.6623179316520691,
      "learning_rate": 0.0005993909609197566,
      "loss": 2.6671,
      "step": 4674
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.6525477766990662,
      "learning_rate": 0.000599390700372916,
      "loss": 2.5785,
      "step": 4675
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.8339728116989136,
      "learning_rate": 0.000599390439770413,
      "loss": 2.7366,
      "step": 4676
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.7869872450828552,
      "learning_rate": 0.0005993901791122476,
      "loss": 2.7523,
      "step": 4677
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.6888931393623352,
      "learning_rate": 0.0005993899183984199,
      "loss": 2.5552,
      "step": 4678
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.7714746594429016,
      "learning_rate": 0.00059938965762893,
      "loss": 2.5843,
      "step": 4679
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.7567166686058044,
      "learning_rate": 0.000599389396803778,
      "loss": 2.7394,
      "step": 4680
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.6746329069137573,
      "learning_rate": 0.0005993891359229637,
      "loss": 2.4804,
      "step": 4681
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.6841243505477905,
      "learning_rate": 0.0005993888749864874,
      "loss": 2.6016,
      "step": 4682
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.719213604927063,
      "learning_rate": 0.000599388613994349,
      "loss": 2.3177,
      "step": 4683
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.6716032028198242,
      "learning_rate": 0.0005993883529465487,
      "loss": 2.7826,
      "step": 4684
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.7390695810317993,
      "learning_rate": 0.0005993880918430864,
      "loss": 2.6974,
      "step": 4685
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.6643344759941101,
      "learning_rate": 0.0005993878306839622,
      "loss": 2.5529,
      "step": 4686
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.7348796725273132,
      "learning_rate": 0.0005993875694691762,
      "loss": 2.6229,
      "step": 4687
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.7065025568008423,
      "learning_rate": 0.0005993873081987283,
      "loss": 2.6525,
      "step": 4688
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.7033332586288452,
      "learning_rate": 0.0005993870468726187,
      "loss": 2.5877,
      "step": 4689
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.6897931694984436,
      "learning_rate": 0.0005993867854908473,
      "loss": 2.62,
      "step": 4690
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.729656457901001,
      "learning_rate": 0.0005993865240534144,
      "loss": 2.6094,
      "step": 4691
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.6771336197853088,
      "learning_rate": 0.0005993862625603199,
      "loss": 2.6938,
      "step": 4692
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.6493284702301025,
      "learning_rate": 0.0005993860010115637,
      "loss": 2.5299,
      "step": 4693
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.7414185404777527,
      "learning_rate": 0.0005993857394071461,
      "loss": 2.7649,
      "step": 4694
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.6620787382125854,
      "learning_rate": 0.000599385477747067,
      "loss": 2.7042,
      "step": 4695
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.7053303122520447,
      "learning_rate": 0.0005993852160313265,
      "loss": 2.6565,
      "step": 4696
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.6328616738319397,
      "learning_rate": 0.0005993849542599247,
      "loss": 2.4445,
      "step": 4697
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.7329674363136292,
      "learning_rate": 0.0005993846924328615,
      "loss": 2.476,
      "step": 4698
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.7098202705383301,
      "learning_rate": 0.000599384430550137,
      "loss": 2.4447,
      "step": 4699
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.7337555289268494,
      "learning_rate": 0.0005993841686117514,
      "loss": 2.6013,
      "step": 4700
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.6605578660964966,
      "learning_rate": 0.0005993839066177045,
      "loss": 2.6328,
      "step": 4701
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.7038322687149048,
      "learning_rate": 0.0005993836445679966,
      "loss": 2.6158,
      "step": 4702
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.6395906209945679,
      "learning_rate": 0.0005993833824626276,
      "loss": 2.6794,
      "step": 4703
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.6383308172225952,
      "learning_rate": 0.0005993831203015976,
      "loss": 2.6691,
      "step": 4704
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.7367165088653564,
      "learning_rate": 0.0005993828580849065,
      "loss": 2.6189,
      "step": 4705
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.6737083792686462,
      "learning_rate": 0.0005993825958125545,
      "loss": 2.7479,
      "step": 4706
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.706292986869812,
      "learning_rate": 0.0005993823334845417,
      "loss": 2.8092,
      "step": 4707
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.793861985206604,
      "learning_rate": 0.0005993820711008681,
      "loss": 2.6826,
      "step": 4708
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.7805479168891907,
      "learning_rate": 0.0005993818086615337,
      "loss": 2.7081,
      "step": 4709
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.7850157022476196,
      "learning_rate": 0.0005993815461665386,
      "loss": 2.7152,
      "step": 4710
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.6543735265731812,
      "learning_rate": 0.0005993812836158827,
      "loss": 2.3608,
      "step": 4711
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.7032749652862549,
      "learning_rate": 0.0005993810210095662,
      "loss": 2.7759,
      "step": 4712
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.6750097274780273,
      "learning_rate": 0.0005993807583475892,
      "loss": 2.7569,
      "step": 4713
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.7173452377319336,
      "learning_rate": 0.0005993804956299516,
      "loss": 2.7855,
      "step": 4714
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.6853822469711304,
      "learning_rate": 0.0005993802328566536,
      "loss": 2.5642,
      "step": 4715
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.6493837237358093,
      "learning_rate": 0.000599379970027695,
      "loss": 2.5553,
      "step": 4716
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.6630619168281555,
      "learning_rate": 0.0005993797071430761,
      "loss": 3.0029,
      "step": 4717
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.7174559235572815,
      "learning_rate": 0.000599379444202797,
      "loss": 2.603,
      "step": 4718
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.699142336845398,
      "learning_rate": 0.0005993791812068574,
      "loss": 2.5979,
      "step": 4719
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.6275342702865601,
      "learning_rate": 0.0005993789181552577,
      "loss": 2.7629,
      "step": 4720
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.7212444543838501,
      "learning_rate": 0.0005993786550479977,
      "loss": 2.7941,
      "step": 4721
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.6847842335700989,
      "learning_rate": 0.0005993783918850776,
      "loss": 2.8695,
      "step": 4722
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.7187209725379944,
      "learning_rate": 0.0005993781286664975,
      "loss": 2.7565,
      "step": 4723
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.7734282612800598,
      "learning_rate": 0.0005993778653922572,
      "loss": 3.0085,
      "step": 4724
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.6912293434143066,
      "learning_rate": 0.0005993776020623569,
      "loss": 2.8704,
      "step": 4725
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.6672186255455017,
      "learning_rate": 0.0005993773386767968,
      "loss": 2.6009,
      "step": 4726
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.8154568076133728,
      "learning_rate": 0.0005993770752355767,
      "loss": 2.8594,
      "step": 4727
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.8418693542480469,
      "learning_rate": 0.0005993768117386968,
      "loss": 2.4385,
      "step": 4728
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.6910855174064636,
      "learning_rate": 0.0005993765481861569,
      "loss": 2.6963,
      "step": 4729
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.7319636344909668,
      "learning_rate": 0.0005993762845779575,
      "loss": 2.5633,
      "step": 4730
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.7570474147796631,
      "learning_rate": 0.0005993760209140983,
      "loss": 2.6908,
      "step": 4731
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.7920469045639038,
      "learning_rate": 0.0005993757571945796,
      "loss": 2.5928,
      "step": 4732
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.7141796350479126,
      "learning_rate": 0.000599375493419401,
      "loss": 2.8085,
      "step": 4733
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.7384299635887146,
      "learning_rate": 0.0005993752295885631,
      "loss": 2.6228,
      "step": 4734
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.7083632349967957,
      "learning_rate": 0.0005993749657020656,
      "loss": 2.6507,
      "step": 4735
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.6944400072097778,
      "learning_rate": 0.0005993747017599087,
      "loss": 2.7286,
      "step": 4736
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.7287389636039734,
      "learning_rate": 0.0005993744377620923,
      "loss": 2.7572,
      "step": 4737
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.7091907858848572,
      "learning_rate": 0.0005993741737086166,
      "loss": 2.8606,
      "step": 4738
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.6093535423278809,
      "learning_rate": 0.0005993739095994815,
      "loss": 2.6005,
      "step": 4739
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.6557595133781433,
      "learning_rate": 0.0005993736454346873,
      "loss": 2.6089,
      "step": 4740
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.6573935747146606,
      "learning_rate": 0.0005993733812142337,
      "loss": 2.6601,
      "step": 4741
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.7303245663642883,
      "learning_rate": 0.000599373116938121,
      "loss": 2.6495,
      "step": 4742
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.6852381825447083,
      "learning_rate": 0.0005993728526063493,
      "loss": 2.8379,
      "step": 4743
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.7242599129676819,
      "learning_rate": 0.0005993725882189184,
      "loss": 2.5716,
      "step": 4744
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.7069227695465088,
      "learning_rate": 0.0005993723237758286,
      "loss": 2.7552,
      "step": 4745
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.7350648045539856,
      "learning_rate": 0.0005993720592770797,
      "loss": 2.8024,
      "step": 4746
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.8835002183914185,
      "learning_rate": 0.0005993717947226719,
      "loss": 2.5934,
      "step": 4747
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.6662955284118652,
      "learning_rate": 0.0005993715301126054,
      "loss": 2.7379,
      "step": 4748
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.8928527235984802,
      "learning_rate": 0.0005993712654468799,
      "loss": 2.5961,
      "step": 4749
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.7939461469650269,
      "learning_rate": 0.0005993710007254957,
      "loss": 2.7935,
      "step": 4750
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.6965118050575256,
      "learning_rate": 0.0005993707359484528,
      "loss": 2.7379,
      "step": 4751
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.6764060258865356,
      "learning_rate": 0.0005993704711157511,
      "loss": 2.9656,
      "step": 4752
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.6842449307441711,
      "learning_rate": 0.000599370206227391,
      "loss": 2.5456,
      "step": 4753
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.6170396208763123,
      "learning_rate": 0.0005993699412833722,
      "loss": 2.5764,
      "step": 4754
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.7365778684616089,
      "learning_rate": 0.0005993696762836948,
      "loss": 2.8083,
      "step": 4755
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.7526220083236694,
      "learning_rate": 0.0005993694112283591,
      "loss": 2.5338,
      "step": 4756
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.664665699005127,
      "learning_rate": 0.0005993691461173648,
      "loss": 2.6053,
      "step": 4757
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.7090550065040588,
      "learning_rate": 0.0005993688809507123,
      "loss": 2.7527,
      "step": 4758
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.7542571425437927,
      "learning_rate": 0.0005993686157284014,
      "loss": 2.6805,
      "step": 4759
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.6574395298957825,
      "learning_rate": 0.0005993683504504322,
      "loss": 2.585,
      "step": 4760
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.6783164143562317,
      "learning_rate": 0.0005993680851168049,
      "loss": 2.6524,
      "step": 4761
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.646981418132782,
      "learning_rate": 0.0005993678197275192,
      "loss": 2.551,
      "step": 4762
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.6883442401885986,
      "learning_rate": 0.0005993675542825755,
      "loss": 2.7274,
      "step": 4763
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.6570743918418884,
      "learning_rate": 0.0005993672887819737,
      "loss": 2.5885,
      "step": 4764
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.6615720987319946,
      "learning_rate": 0.0005993670232257138,
      "loss": 2.6673,
      "step": 4765
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.6220018267631531,
      "learning_rate": 0.0005993667576137961,
      "loss": 2.7625,
      "step": 4766
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.6556959748268127,
      "learning_rate": 0.0005993664919462202,
      "loss": 2.6379,
      "step": 4767
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.7022548913955688,
      "learning_rate": 0.0005993662262229867,
      "loss": 2.5633,
      "step": 4768
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.7179526686668396,
      "learning_rate": 0.0005993659604440953,
      "loss": 2.5806,
      "step": 4769
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.6719859838485718,
      "learning_rate": 0.0005993656946095459,
      "loss": 2.6464,
      "step": 4770
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.7849947810173035,
      "learning_rate": 0.000599365428719339,
      "loss": 2.8271,
      "step": 4771
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.6630858778953552,
      "learning_rate": 0.0005993651627734743,
      "loss": 2.6291,
      "step": 4772
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.8916397094726562,
      "learning_rate": 0.0005993648967719521,
      "loss": 2.6644,
      "step": 4773
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.6853311061859131,
      "learning_rate": 0.0005993646307147722,
      "loss": 2.5253,
      "step": 4774
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.897799551486969,
      "learning_rate": 0.0005993643646019347,
      "loss": 2.7898,
      "step": 4775
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.0639278888702393,
      "learning_rate": 0.0005993640984334398,
      "loss": 2.7454,
      "step": 4776
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.7578815221786499,
      "learning_rate": 0.0005993638322092875,
      "loss": 2.4098,
      "step": 4777
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.0695639848709106,
      "learning_rate": 0.0005993635659294779,
      "loss": 2.5061,
      "step": 4778
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.6684733033180237,
      "learning_rate": 0.0005993632995940107,
      "loss": 2.5014,
      "step": 4779
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.7838892936706543,
      "learning_rate": 0.0005993630332028863,
      "loss": 2.9143,
      "step": 4780
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.7611510157585144,
      "learning_rate": 0.0005993627667561048,
      "loss": 2.4845,
      "step": 4781
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.6609251499176025,
      "learning_rate": 0.000599362500253666,
      "loss": 2.6281,
      "step": 4782
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.7445269823074341,
      "learning_rate": 0.0005993622336955702,
      "loss": 2.7695,
      "step": 4783
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.674041748046875,
      "learning_rate": 0.000599361967081817,
      "loss": 2.5402,
      "step": 4784
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.7408985495567322,
      "learning_rate": 0.0005993617004124071,
      "loss": 2.4214,
      "step": 4785
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.6622334122657776,
      "learning_rate": 0.00059936143368734,
      "loss": 2.6492,
      "step": 4786
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.6863186359405518,
      "learning_rate": 0.000599361166906616,
      "loss": 2.79,
      "step": 4787
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.7277592420578003,
      "learning_rate": 0.0005993609000702352,
      "loss": 2.8816,
      "step": 4788
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.6974093914031982,
      "learning_rate": 0.0005993606331781975,
      "loss": 2.7006,
      "step": 4789
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.6450218558311462,
      "learning_rate": 0.0005993603662305029,
      "loss": 2.3906,
      "step": 4790
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.663608729839325,
      "learning_rate": 0.0005993600992271517,
      "loss": 2.5578,
      "step": 4791
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.6927009224891663,
      "learning_rate": 0.0005993598321681437,
      "loss": 2.6987,
      "step": 4792
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.6892904043197632,
      "learning_rate": 0.0005993595650534792,
      "loss": 2.6322,
      "step": 4793
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.6491268873214722,
      "learning_rate": 0.000599359297883158,
      "loss": 2.379,
      "step": 4794
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.6563103795051575,
      "learning_rate": 0.0005993590306571803,
      "loss": 2.6402,
      "step": 4795
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.7380285263061523,
      "learning_rate": 0.000599358763375546,
      "loss": 2.6021,
      "step": 4796
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.6843381524085999,
      "learning_rate": 0.0005993584960382554,
      "loss": 2.6394,
      "step": 4797
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.7407383322715759,
      "learning_rate": 0.0005993582286453083,
      "loss": 2.6905,
      "step": 4798
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.709719181060791,
      "learning_rate": 0.0005993579611967049,
      "loss": 2.6009,
      "step": 4799
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.7639894485473633,
      "learning_rate": 0.0005993576936924452,
      "loss": 2.4881,
      "step": 4800
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.7069433331489563,
      "learning_rate": 0.0005993574261325292,
      "loss": 2.6317,
      "step": 4801
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.7918162941932678,
      "learning_rate": 0.0005993571585169571,
      "loss": 2.5509,
      "step": 4802
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.8498401641845703,
      "learning_rate": 0.0005993568908457288,
      "loss": 2.5491,
      "step": 4803
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.6860686540603638,
      "learning_rate": 0.0005993566231188444,
      "loss": 2.814,
      "step": 4804
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.7099382877349854,
      "learning_rate": 0.000599356355336304,
      "loss": 2.6891,
      "step": 4805
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.9166355133056641,
      "learning_rate": 0.0005993560874981075,
      "loss": 2.66,
      "step": 4806
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.7328106164932251,
      "learning_rate": 0.0005993558196042552,
      "loss": 2.6678,
      "step": 4807
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.6757539510726929,
      "learning_rate": 0.0005993555516547469,
      "loss": 2.7492,
      "step": 4808
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.667317807674408,
      "learning_rate": 0.0005993552836495828,
      "loss": 2.5535,
      "step": 4809
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.6965631246566772,
      "learning_rate": 0.0005993550155887629,
      "loss": 2.8616,
      "step": 4810
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.7844143509864807,
      "learning_rate": 0.0005993547474722872,
      "loss": 2.5135,
      "step": 4811
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.7611298561096191,
      "learning_rate": 0.0005993544793001558,
      "loss": 2.6308,
      "step": 4812
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.6494976878166199,
      "learning_rate": 0.0005993542110723688,
      "loss": 2.6331,
      "step": 4813
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.679042637348175,
      "learning_rate": 0.0005993539427889262,
      "loss": 2.8322,
      "step": 4814
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.6218549013137817,
      "learning_rate": 0.0005993536744498281,
      "loss": 2.6793,
      "step": 4815
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.7323179244995117,
      "learning_rate": 0.0005993534060550744,
      "loss": 2.8785,
      "step": 4816
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.7235984206199646,
      "learning_rate": 0.0005993531376046652,
      "loss": 2.5485,
      "step": 4817
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.7784243822097778,
      "learning_rate": 0.0005993528690986007,
      "loss": 2.7615,
      "step": 4818
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.6627990007400513,
      "learning_rate": 0.0005993526005368809,
      "loss": 2.8051,
      "step": 4819
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.780674159526825,
      "learning_rate": 0.0005993523319195057,
      "loss": 2.5291,
      "step": 4820
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.7323159575462341,
      "learning_rate": 0.0005993520632464753,
      "loss": 2.4146,
      "step": 4821
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.9128153920173645,
      "learning_rate": 0.0005993517945177897,
      "loss": 2.7937,
      "step": 4822
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.7354504466056824,
      "learning_rate": 0.0005993515257334489,
      "loss": 2.9162,
      "step": 4823
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.8677455186843872,
      "learning_rate": 0.000599351256893453,
      "loss": 2.6225,
      "step": 4824
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.6990819573402405,
      "learning_rate": 0.0005993509879978022,
      "loss": 2.7883,
      "step": 4825
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.6496020555496216,
      "learning_rate": 0.0005993507190464963,
      "loss": 2.6712,
      "step": 4826
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.8702654838562012,
      "learning_rate": 0.0005993504500395354,
      "loss": 2.5603,
      "step": 4827
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.7425196766853333,
      "learning_rate": 0.0005993501809769197,
      "loss": 2.7594,
      "step": 4828
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.8328184485435486,
      "learning_rate": 0.000599349911858649,
      "loss": 2.633,
      "step": 4829
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.6701986789703369,
      "learning_rate": 0.0005993496426847237,
      "loss": 2.4838,
      "step": 4830
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.6454482674598694,
      "learning_rate": 0.0005993493734551435,
      "loss": 2.4382,
      "step": 4831
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.6581020355224609,
      "learning_rate": 0.0005993491041699087,
      "loss": 2.4559,
      "step": 4832
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.6969983577728271,
      "learning_rate": 0.0005993488348290192,
      "loss": 2.4185,
      "step": 4833
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.7468435764312744,
      "learning_rate": 0.000599348565432475,
      "loss": 2.9411,
      "step": 4834
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.6823453307151794,
      "learning_rate": 0.0005993482959802764,
      "loss": 2.5521,
      "step": 4835
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.6560533046722412,
      "learning_rate": 0.0005993480264724233,
      "loss": 2.748,
      "step": 4836
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.6625153422355652,
      "learning_rate": 0.0005993477569089156,
      "loss": 2.4627,
      "step": 4837
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.6645216941833496,
      "learning_rate": 0.0005993474872897537,
      "loss": 2.8533,
      "step": 4838
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.7341049909591675,
      "learning_rate": 0.0005993472176149373,
      "loss": 2.5683,
      "step": 4839
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.6873880624771118,
      "learning_rate": 0.0005993469478844667,
      "loss": 2.7383,
      "step": 4840
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.6824484467506409,
      "learning_rate": 0.0005993466780983418,
      "loss": 2.32,
      "step": 4841
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.7210459113121033,
      "learning_rate": 0.0005993464082565627,
      "loss": 2.4489,
      "step": 4842
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.68825364112854,
      "learning_rate": 0.0005993461383591294,
      "loss": 2.7592,
      "step": 4843
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.7249085903167725,
      "learning_rate": 0.0005993458684060421,
      "loss": 2.635,
      "step": 4844
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.768555760383606,
      "learning_rate": 0.0005993455983973007,
      "loss": 2.4084,
      "step": 4845
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.7019956111907959,
      "learning_rate": 0.0005993453283329053,
      "loss": 2.793,
      "step": 4846
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.750049889087677,
      "learning_rate": 0.000599345058212856,
      "loss": 2.4683,
      "step": 4847
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.7229719758033752,
      "learning_rate": 0.0005993447880371528,
      "loss": 2.599,
      "step": 4848
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.6978958249092102,
      "learning_rate": 0.0005993445178057958,
      "loss": 2.5163,
      "step": 4849
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.700337827205658,
      "learning_rate": 0.0005993442475187849,
      "loss": 2.7155,
      "step": 4850
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.7340052127838135,
      "learning_rate": 0.0005993439771761203,
      "loss": 2.5618,
      "step": 4851
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.7750051617622375,
      "learning_rate": 0.0005993437067778019,
      "loss": 2.6433,
      "step": 4852
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.6742136478424072,
      "learning_rate": 0.00059934343632383,
      "loss": 2.8206,
      "step": 4853
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.6853194236755371,
      "learning_rate": 0.0005993431658142045,
      "loss": 2.7064,
      "step": 4854
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.6505461931228638,
      "learning_rate": 0.0005993428952489253,
      "loss": 2.5034,
      "step": 4855
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.7138304710388184,
      "learning_rate": 0.0005993426246279928,
      "loss": 2.7427,
      "step": 4856
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.6963639855384827,
      "learning_rate": 0.0005993423539514067,
      "loss": 2.5473,
      "step": 4857
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.766514778137207,
      "learning_rate": 0.0005993420832191673,
      "loss": 2.7199,
      "step": 4858
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.6899129152297974,
      "learning_rate": 0.0005993418124312746,
      "loss": 2.4809,
      "step": 4859
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.7421445846557617,
      "learning_rate": 0.0005993415415877285,
      "loss": 2.6777,
      "step": 4860
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.726142168045044,
      "learning_rate": 0.0005993412706885292,
      "loss": 2.5996,
      "step": 4861
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.7004103660583496,
      "learning_rate": 0.0005993409997336767,
      "loss": 2.638,
      "step": 4862
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.6790834069252014,
      "learning_rate": 0.0005993407287231711,
      "loss": 2.7463,
      "step": 4863
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.6428506970405579,
      "learning_rate": 0.0005993404576570123,
      "loss": 2.4616,
      "step": 4864
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.7171942591667175,
      "learning_rate": 0.0005993401865352006,
      "loss": 2.758,
      "step": 4865
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.9782264232635498,
      "learning_rate": 0.0005993399153577357,
      "loss": 2.5543,
      "step": 4866
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.6900724768638611,
      "learning_rate": 0.0005993396441246181,
      "loss": 2.5679,
      "step": 4867
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.2313786745071411,
      "learning_rate": 0.0005993393728358474,
      "loss": 2.7234,
      "step": 4868
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.6401429772377014,
      "learning_rate": 0.000599339101491424,
      "loss": 2.6994,
      "step": 4869
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.2878776788711548,
      "learning_rate": 0.0005993388300913478,
      "loss": 2.5473,
      "step": 4870
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.7942840456962585,
      "learning_rate": 0.0005993385586356188,
      "loss": 2.8131,
      "step": 4871
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.0390405654907227,
      "learning_rate": 0.0005993382871242372,
      "loss": 2.569,
      "step": 4872
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.8165076375007629,
      "learning_rate": 0.0005993380155572029,
      "loss": 2.5647,
      "step": 4873
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.764259934425354,
      "learning_rate": 0.000599337743934516,
      "loss": 2.8708,
      "step": 4874
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.7067998051643372,
      "learning_rate": 0.0005993374722561766,
      "loss": 2.3839,
      "step": 4875
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.7391170859336853,
      "learning_rate": 0.0005993372005221847,
      "loss": 2.7636,
      "step": 4876
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.7051681280136108,
      "learning_rate": 0.0005993369287325403,
      "loss": 2.6483,
      "step": 4877
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.8034472465515137,
      "learning_rate": 0.0005993366568872436,
      "loss": 2.7296,
      "step": 4878
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.6914472579956055,
      "learning_rate": 0.0005993363849862946,
      "loss": 2.5101,
      "step": 4879
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.6636930108070374,
      "learning_rate": 0.0005993361130296931,
      "loss": 2.7303,
      "step": 4880
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.6839786767959595,
      "learning_rate": 0.0005993358410174396,
      "loss": 2.4793,
      "step": 4881
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.6946136951446533,
      "learning_rate": 0.0005993355689495339,
      "loss": 2.4762,
      "step": 4882
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.6345370411872864,
      "learning_rate": 0.0005993352968259759,
      "loss": 2.5492,
      "step": 4883
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.6856672763824463,
      "learning_rate": 0.000599335024646766,
      "loss": 2.5111,
      "step": 4884
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.684663712978363,
      "learning_rate": 0.0005993347524119039,
      "loss": 2.817,
      "step": 4885
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.655340313911438,
      "learning_rate": 0.00059933448012139,
      "loss": 2.5934,
      "step": 4886
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.7069035172462463,
      "learning_rate": 0.0005993342077752241,
      "loss": 2.7201,
      "step": 4887
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.6962146759033203,
      "learning_rate": 0.0005993339353734063,
      "loss": 2.5438,
      "step": 4888
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.687262237071991,
      "learning_rate": 0.0005993336629159366,
      "loss": 2.7125,
      "step": 4889
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.7100688815116882,
      "learning_rate": 0.0005993333904028153,
      "loss": 2.6798,
      "step": 4890
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.7643864750862122,
      "learning_rate": 0.0005993331178340421,
      "loss": 2.574,
      "step": 4891
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.641984224319458,
      "learning_rate": 0.0005993328452096173,
      "loss": 2.6501,
      "step": 4892
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.7476207613945007,
      "learning_rate": 0.0005993325725295408,
      "loss": 2.8453,
      "step": 4893
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.7388206720352173,
      "learning_rate": 0.0005993322997938128,
      "loss": 2.6412,
      "step": 4894
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.6837539672851562,
      "learning_rate": 0.0005993320270024333,
      "loss": 2.688,
      "step": 4895
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.7823191285133362,
      "learning_rate": 0.0005993317541554023,
      "loss": 2.8075,
      "step": 4896
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.7463138103485107,
      "learning_rate": 0.0005993314812527199,
      "loss": 2.8761,
      "step": 4897
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.7253130674362183,
      "learning_rate": 0.000599331208294386,
      "loss": 2.6102,
      "step": 4898
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.7289043068885803,
      "learning_rate": 0.000599330935280401,
      "loss": 2.4902,
      "step": 4899
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.6984744071960449,
      "learning_rate": 0.0005993306622107645,
      "loss": 2.5745,
      "step": 4900
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.6556677222251892,
      "learning_rate": 0.0005993303890854769,
      "loss": 2.4493,
      "step": 4901
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.6607082486152649,
      "learning_rate": 0.0005993301159045382,
      "loss": 2.6753,
      "step": 4902
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.7058003544807434,
      "learning_rate": 0.0005993298426679481,
      "loss": 2.9138,
      "step": 4903
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.6379452347755432,
      "learning_rate": 0.0005993295693757072,
      "loss": 2.7741,
      "step": 4904
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.6965972185134888,
      "learning_rate": 0.0005993292960278152,
      "loss": 2.585,
      "step": 4905
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.7252690196037292,
      "learning_rate": 0.0005993290226242723,
      "loss": 2.5759,
      "step": 4906
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.7518689632415771,
      "learning_rate": 0.0005993287491650784,
      "loss": 2.6966,
      "step": 4907
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.6594406366348267,
      "learning_rate": 0.0005993284756502337,
      "loss": 2.6004,
      "step": 4908
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.7625235915184021,
      "learning_rate": 0.0005993282020797381,
      "loss": 2.6047,
      "step": 4909
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.7817944884300232,
      "learning_rate": 0.0005993279284535918,
      "loss": 3.0332,
      "step": 4910
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.6410612463951111,
      "learning_rate": 0.0005993276547717947,
      "loss": 2.5956,
      "step": 4911
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.6632897257804871,
      "learning_rate": 0.0005993273810343471,
      "loss": 2.5868,
      "step": 4912
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.7013951539993286,
      "learning_rate": 0.0005993271072412487,
      "loss": 2.7086,
      "step": 4913
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.7997173070907593,
      "learning_rate": 0.0005993268333924998,
      "loss": 2.5678,
      "step": 4914
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.7156880497932434,
      "learning_rate": 0.0005993265594881006,
      "loss": 2.6617,
      "step": 4915
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.682906448841095,
      "learning_rate": 0.0005993262855280507,
      "loss": 2.7018,
      "step": 4916
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.6972526907920837,
      "learning_rate": 0.0005993260115123505,
      "loss": 2.6412,
      "step": 4917
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.7502843737602234,
      "learning_rate": 0.0005993257374409998,
      "loss": 2.7661,
      "step": 4918
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.6744600534439087,
      "learning_rate": 0.000599325463313999,
      "loss": 2.7012,
      "step": 4919
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.723646879196167,
      "learning_rate": 0.0005993251891313478,
      "loss": 2.5666,
      "step": 4920
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.725290060043335,
      "learning_rate": 0.0005993249148930465,
      "loss": 2.6459,
      "step": 4921
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.6334148049354553,
      "learning_rate": 0.000599324640599095,
      "loss": 2.5722,
      "step": 4922
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.629179835319519,
      "learning_rate": 0.0005993243662494933,
      "loss": 2.4164,
      "step": 4923
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.6792347431182861,
      "learning_rate": 0.0005993240918442418,
      "loss": 2.8122,
      "step": 4924
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.6496804356575012,
      "learning_rate": 0.0005993238173833401,
      "loss": 2.6238,
      "step": 4925
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.6599984765052795,
      "learning_rate": 0.0005993235428667884,
      "loss": 2.6987,
      "step": 4926
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.7164821624755859,
      "learning_rate": 0.000599323268294587,
      "loss": 2.5652,
      "step": 4927
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.7497664093971252,
      "learning_rate": 0.0005993229936667356,
      "loss": 2.495,
      "step": 4928
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.7260527610778809,
      "learning_rate": 0.0005993227189832345,
      "loss": 2.6443,
      "step": 4929
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.8371305465698242,
      "learning_rate": 0.0005993224442440836,
      "loss": 2.6404,
      "step": 4930
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.7623099088668823,
      "learning_rate": 0.0005993221694492831,
      "loss": 2.7041,
      "step": 4931
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.7393266558647156,
      "learning_rate": 0.0005993218945988328,
      "loss": 2.6161,
      "step": 4932
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.6738302707672119,
      "learning_rate": 0.000599321619692733,
      "loss": 2.7643,
      "step": 4933
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.8172332644462585,
      "learning_rate": 0.0005993213447309837,
      "loss": 2.8295,
      "step": 4934
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.6486897468566895,
      "learning_rate": 0.0005993210697135848,
      "loss": 2.7729,
      "step": 4935
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.7224042415618896,
      "learning_rate": 0.0005993207946405365,
      "loss": 2.7303,
      "step": 4936
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.7248297929763794,
      "learning_rate": 0.0005993205195118389,
      "loss": 2.9651,
      "step": 4937
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.722693145275116,
      "learning_rate": 0.0005993202443274919,
      "loss": 2.7388,
      "step": 4938
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.646895706653595,
      "learning_rate": 0.0005993199690874956,
      "loss": 2.4915,
      "step": 4939
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.7466636300086975,
      "learning_rate": 0.0005993196937918501,
      "loss": 2.7196,
      "step": 4940
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.7954763174057007,
      "learning_rate": 0.0005993194184405554,
      "loss": 2.7135,
      "step": 4941
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.7365067601203918,
      "learning_rate": 0.0005993191430336116,
      "loss": 2.5995,
      "step": 4942
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.7492339611053467,
      "learning_rate": 0.0005993188675710187,
      "loss": 2.5111,
      "step": 4943
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.8496761322021484,
      "learning_rate": 0.0005993185920527769,
      "loss": 2.875,
      "step": 4944
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.749193549156189,
      "learning_rate": 0.0005993183164788859,
      "loss": 2.5841,
      "step": 4945
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.8083528876304626,
      "learning_rate": 0.0005993180408493461,
      "loss": 2.4562,
      "step": 4946
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.8853338956832886,
      "learning_rate": 0.0005993177651641573,
      "loss": 2.4366,
      "step": 4947
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.8046033382415771,
      "learning_rate": 0.0005993174894233198,
      "loss": 2.7233,
      "step": 4948
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.9532799124717712,
      "learning_rate": 0.0005993172136268336,
      "loss": 2.6314,
      "step": 4949
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.7044716477394104,
      "learning_rate": 0.0005993169377746985,
      "loss": 2.5286,
      "step": 4950
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.9720735549926758,
      "learning_rate": 0.0005993166618669149,
      "loss": 2.6359,
      "step": 4951
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.8885200023651123,
      "learning_rate": 0.0005993163859034825,
      "loss": 2.5541,
      "step": 4952
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.8755014538764954,
      "learning_rate": 0.0005993161098844017,
      "loss": 2.8532,
      "step": 4953
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.8306611180305481,
      "learning_rate": 0.0005993158338096722,
      "loss": 2.7113,
      "step": 4954
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.6763392686843872,
      "learning_rate": 0.0005993155576792945,
      "loss": 2.4712,
      "step": 4955
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.7489279508590698,
      "learning_rate": 0.0005993152814932682,
      "loss": 2.7963,
      "step": 4956
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.8021039366722107,
      "learning_rate": 0.0005993150052515935,
      "loss": 2.2366,
      "step": 4957
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.6439486742019653,
      "learning_rate": 0.0005993147289542706,
      "loss": 2.582,
      "step": 4958
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.8553255796432495,
      "learning_rate": 0.0005993144526012995,
      "loss": 2.6203,
      "step": 4959
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.7806095480918884,
      "learning_rate": 0.00059931417619268,
      "loss": 2.7542,
      "step": 4960
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.7124270796775818,
      "learning_rate": 0.0005993138997284125,
      "loss": 2.4912,
      "step": 4961
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.7964591979980469,
      "learning_rate": 0.0005993136232084969,
      "loss": 2.6509,
      "step": 4962
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.7035781741142273,
      "learning_rate": 0.0005993133466329332,
      "loss": 2.5095,
      "step": 4963
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.704136073589325,
      "learning_rate": 0.0005993130700017215,
      "loss": 2.7429,
      "step": 4964
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.8069924116134644,
      "learning_rate": 0.0005993127933148618,
      "loss": 2.734,
      "step": 4965
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.6710755825042725,
      "learning_rate": 0.0005993125165723543,
      "loss": 2.621,
      "step": 4966
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.694861888885498,
      "learning_rate": 0.0005993122397741988,
      "loss": 2.4851,
      "step": 4967
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.7510675191879272,
      "learning_rate": 0.0005993119629203956,
      "loss": 2.6789,
      "step": 4968
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.7044188380241394,
      "learning_rate": 0.0005993116860109447,
      "loss": 2.5763,
      "step": 4969
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.6529096961021423,
      "learning_rate": 0.0005993114090458461,
      "loss": 2.7167,
      "step": 4970
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.7639462351799011,
      "learning_rate": 0.0005993111320250998,
      "loss": 2.6898,
      "step": 4971
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.7544530630111694,
      "learning_rate": 0.0005993108549487059,
      "loss": 2.5155,
      "step": 4972
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.7338501811027527,
      "learning_rate": 0.0005993105778166646,
      "loss": 2.6464,
      "step": 4973
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.7006990313529968,
      "learning_rate": 0.0005993103006289756,
      "loss": 2.4898,
      "step": 4974
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.9463292360305786,
      "learning_rate": 0.0005993100233856394,
      "loss": 2.5711,
      "step": 4975
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.6395702958106995,
      "learning_rate": 0.0005993097460866556,
      "loss": 2.7959,
      "step": 4976
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.9318717122077942,
      "learning_rate": 0.0005993094687320246,
      "loss": 2.4435,
      "step": 4977
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.7918896675109863,
      "learning_rate": 0.0005993091913217462,
      "loss": 2.7099,
      "step": 4978
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.7244043350219727,
      "learning_rate": 0.0005993089138558208,
      "loss": 2.4684,
      "step": 4979
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.7474383115768433,
      "learning_rate": 0.000599308636334248,
      "loss": 2.5241,
      "step": 4980
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.6644588112831116,
      "learning_rate": 0.0005993083587570282,
      "loss": 2.9098,
      "step": 4981
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.6453869938850403,
      "learning_rate": 0.0005993080811241613,
      "loss": 2.7542,
      "step": 4982
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.6505493521690369,
      "learning_rate": 0.0005993078034356473,
      "loss": 2.4529,
      "step": 4983
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.7180443406105042,
      "learning_rate": 0.0005993075256914864,
      "loss": 2.6881,
      "step": 4984
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.6620559692382812,
      "learning_rate": 0.0005993072478916786,
      "loss": 2.7808,
      "step": 4985
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.6659979224205017,
      "learning_rate": 0.0005993069700362238,
      "loss": 2.8222,
      "step": 4986
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.6937699317932129,
      "learning_rate": 0.0005993066921251224,
      "loss": 2.5995,
      "step": 4987
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.7188760638237,
      "learning_rate": 0.000599306414158374,
      "loss": 2.6647,
      "step": 4988
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.6804574728012085,
      "learning_rate": 0.0005993061361359791,
      "loss": 2.7286,
      "step": 4989
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.6679001450538635,
      "learning_rate": 0.0005993058580579374,
      "loss": 2.7865,
      "step": 4990
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.7151919603347778,
      "learning_rate": 0.0005993055799242492,
      "loss": 2.5063,
      "step": 4991
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.6492710113525391,
      "learning_rate": 0.0005993053017349144,
      "loss": 2.5033,
      "step": 4992
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.7028322815895081,
      "learning_rate": 0.000599305023489933,
      "loss": 2.6154,
      "step": 4993
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.7047325968742371,
      "learning_rate": 0.0005993047451893053,
      "loss": 2.6369,
      "step": 4994
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.7050532698631287,
      "learning_rate": 0.000599304466833031,
      "loss": 2.4243,
      "step": 4995
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.7002944350242615,
      "learning_rate": 0.0005993041884211106,
      "loss": 2.5144,
      "step": 4996
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.7162440419197083,
      "learning_rate": 0.0005993039099535438,
      "loss": 2.7698,
      "step": 4997
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.6620725989341736,
      "learning_rate": 0.0005993036314303307,
      "loss": 2.6761,
      "step": 4998
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.7905614376068115,
      "learning_rate": 0.0005993033528514714,
      "loss": 2.7367,
      "step": 4999
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.7155806422233582,
      "learning_rate": 0.0005993030742169659,
      "loss": 2.7371,
      "step": 5000
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.7632022500038147,
      "learning_rate": 0.0005993027955268145,
      "loss": 2.72,
      "step": 5001
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.7179052233695984,
      "learning_rate": 0.000599302516781017,
      "loss": 2.6974,
      "step": 5002
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.7207827568054199,
      "learning_rate": 0.0005993022379795735,
      "loss": 2.5892,
      "step": 5003
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.671480655670166,
      "learning_rate": 0.0005993019591224839,
      "loss": 2.6225,
      "step": 5004
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.6544123291969299,
      "learning_rate": 0.0005993016802097486,
      "loss": 2.4644,
      "step": 5005
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.7303997278213501,
      "learning_rate": 0.0005993014012413674,
      "loss": 2.6078,
      "step": 5006
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.7195912003517151,
      "learning_rate": 0.0005993011222173404,
      "loss": 2.6408,
      "step": 5007
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.7044783234596252,
      "learning_rate": 0.0005993008431376677,
      "loss": 2.6072,
      "step": 5008
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.722688615322113,
      "learning_rate": 0.0005993005640023494,
      "loss": 2.706,
      "step": 5009
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.6769305467605591,
      "learning_rate": 0.0005993002848113854,
      "loss": 2.6668,
      "step": 5010
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.6652019619941711,
      "learning_rate": 0.0005993000055647758,
      "loss": 2.6515,
      "step": 5011
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.643875777721405,
      "learning_rate": 0.0005992997262625208,
      "loss": 2.6032,
      "step": 5012
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.6503117084503174,
      "learning_rate": 0.0005992994469046202,
      "loss": 2.7412,
      "step": 5013
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.7985588312149048,
      "learning_rate": 0.0005992991674910743,
      "loss": 2.7894,
      "step": 5014
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.721092939376831,
      "learning_rate": 0.0005992988880218829,
      "loss": 2.6701,
      "step": 5015
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.6262143850326538,
      "learning_rate": 0.0005992986084970462,
      "loss": 2.7973,
      "step": 5016
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.7549450397491455,
      "learning_rate": 0.0005992983289165643,
      "loss": 2.6371,
      "step": 5017
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.689497172832489,
      "learning_rate": 0.0005992980492804372,
      "loss": 2.6472,
      "step": 5018
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.6493018865585327,
      "learning_rate": 0.000599297769588665,
      "loss": 2.5159,
      "step": 5019
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.7017213106155396,
      "learning_rate": 0.0005992974898412476,
      "loss": 2.4414,
      "step": 5020
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.7003387808799744,
      "learning_rate": 0.0005992972100381851,
      "loss": 2.6224,
      "step": 5021
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.6938702464103699,
      "learning_rate": 0.0005992969301794777,
      "loss": 2.4886,
      "step": 5022
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.7215121984481812,
      "learning_rate": 0.0005992966502651253,
      "loss": 2.619,
      "step": 5023
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.7379422783851624,
      "learning_rate": 0.000599296370295128,
      "loss": 2.4438,
      "step": 5024
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.6521074771881104,
      "learning_rate": 0.0005992960902694859,
      "loss": 2.6709,
      "step": 5025
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.7314295172691345,
      "learning_rate": 0.0005992958101881989,
      "loss": 2.622,
      "step": 5026
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.7990078330039978,
      "learning_rate": 0.0005992955300512673,
      "loss": 2.6861,
      "step": 5027
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.7529921531677246,
      "learning_rate": 0.0005992952498586909,
      "loss": 2.4605,
      "step": 5028
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.6789618134498596,
      "learning_rate": 0.0005992949696104699,
      "loss": 2.8237,
      "step": 5029
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.7502813339233398,
      "learning_rate": 0.0005992946893066044,
      "loss": 2.5691,
      "step": 5030
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.6901394724845886,
      "learning_rate": 0.0005992944089470943,
      "loss": 2.6981,
      "step": 5031
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.7163954377174377,
      "learning_rate": 0.0005992941285319398,
      "loss": 2.6579,
      "step": 5032
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.6500762104988098,
      "learning_rate": 0.0005992938480611407,
      "loss": 2.6235,
      "step": 5033
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.6975706219673157,
      "learning_rate": 0.0005992935675346974,
      "loss": 2.8559,
      "step": 5034
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.7056712508201599,
      "learning_rate": 0.0005992932869526097,
      "loss": 2.7292,
      "step": 5035
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.9254961609840393,
      "learning_rate": 0.0005992930063148776,
      "loss": 2.4316,
      "step": 5036
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.7465890645980835,
      "learning_rate": 0.0005992927256215014,
      "loss": 2.8274,
      "step": 5037
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.6831021308898926,
      "learning_rate": 0.000599292444872481,
      "loss": 2.7699,
      "step": 5038
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.6330239176750183,
      "learning_rate": 0.0005992921640678166,
      "loss": 2.5307,
      "step": 5039
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.6739515066146851,
      "learning_rate": 0.0005992918832075081,
      "loss": 2.6193,
      "step": 5040
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.7023676037788391,
      "learning_rate": 0.0005992916022915556,
      "loss": 2.6956,
      "step": 5041
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.7486028075218201,
      "learning_rate": 0.000599291321319959,
      "loss": 2.7266,
      "step": 5042
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.6845025420188904,
      "learning_rate": 0.0005992910402927185,
      "loss": 2.9742,
      "step": 5043
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.8803558349609375,
      "learning_rate": 0.0005992907592098343,
      "loss": 2.7326,
      "step": 5044
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.6976671814918518,
      "learning_rate": 0.0005992904780713062,
      "loss": 2.4228,
      "step": 5045
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.7197511792182922,
      "learning_rate": 0.0005992901968771343,
      "loss": 2.9079,
      "step": 5046
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.7412834167480469,
      "learning_rate": 0.0005992899156273189,
      "loss": 2.7422,
      "step": 5047
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.6775748133659363,
      "learning_rate": 0.0005992896343218596,
      "loss": 2.6002,
      "step": 5048
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.6285853981971741,
      "learning_rate": 0.0005992893529607569,
      "loss": 2.6275,
      "step": 5049
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.6713625192642212,
      "learning_rate": 0.0005992890715440107,
      "loss": 2.7526,
      "step": 5050
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.6649963855743408,
      "learning_rate": 0.0005992887900716209,
      "loss": 2.5829,
      "step": 5051
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.6666872501373291,
      "learning_rate": 0.0005992885085435876,
      "loss": 2.7387,
      "step": 5052
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.6868544816970825,
      "learning_rate": 0.000599288226959911,
      "loss": 2.6425,
      "step": 5053
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.7869729995727539,
      "learning_rate": 0.0005992879453205911,
      "loss": 2.4397,
      "step": 5054
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.704319953918457,
      "learning_rate": 0.0005992876636256279,
      "loss": 2.6658,
      "step": 5055
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.6650059223175049,
      "learning_rate": 0.0005992873818750214,
      "loss": 2.6217,
      "step": 5056
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.7476495504379272,
      "learning_rate": 0.0005992871000687717,
      "loss": 2.5992,
      "step": 5057
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.6719725131988525,
      "learning_rate": 0.0005992868182068791,
      "loss": 2.251,
      "step": 5058
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.6688188314437866,
      "learning_rate": 0.0005992865362893433,
      "loss": 2.4749,
      "step": 5059
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.7590948939323425,
      "learning_rate": 0.0005992862543161644,
      "loss": 2.7235,
      "step": 5060
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.6531848311424255,
      "learning_rate": 0.0005992859722873426,
      "loss": 2.5923,
      "step": 5061
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.7424412965774536,
      "learning_rate": 0.0005992856902028779,
      "loss": 2.531,
      "step": 5062
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.6701064109802246,
      "learning_rate": 0.0005992854080627704,
      "loss": 2.4069,
      "step": 5063
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.7525029182434082,
      "learning_rate": 0.00059928512586702,
      "loss": 2.6532,
      "step": 5064
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.6565815806388855,
      "learning_rate": 0.0005992848436156268,
      "loss": 2.5517,
      "step": 5065
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.6885246634483337,
      "learning_rate": 0.000599284561308591,
      "loss": 2.4445,
      "step": 5066
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.700912356376648,
      "learning_rate": 0.0005992842789459125,
      "loss": 2.6277,
      "step": 5067
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.8319658637046814,
      "learning_rate": 0.0005992839965275916,
      "loss": 2.6269,
      "step": 5068
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.8677083849906921,
      "learning_rate": 0.0005992837140536279,
      "loss": 2.6464,
      "step": 5069
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.6340222358703613,
      "learning_rate": 0.0005992834315240217,
      "loss": 2.7193,
      "step": 5070
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.6656433343887329,
      "learning_rate": 0.0005992831489387732,
      "loss": 2.8463,
      "step": 5071
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.6937810778617859,
      "learning_rate": 0.0005992828662978823,
      "loss": 2.7355,
      "step": 5072
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.7063716650009155,
      "learning_rate": 0.000599282583601349,
      "loss": 2.6983,
      "step": 5073
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.6275193095207214,
      "learning_rate": 0.0005992823008491735,
      "loss": 2.5214,
      "step": 5074
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.7257766723632812,
      "learning_rate": 0.0005992820180413558,
      "loss": 2.8724,
      "step": 5075
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.6591348648071289,
      "learning_rate": 0.0005992817351778958,
      "loss": 2.4877,
      "step": 5076
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.62393718957901,
      "learning_rate": 0.0005992814522587937,
      "loss": 2.6239,
      "step": 5077
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.6359503865242004,
      "learning_rate": 0.0005992811692840496,
      "loss": 2.7798,
      "step": 5078
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.7203819155693054,
      "learning_rate": 0.0005992808862536634,
      "loss": 2.5206,
      "step": 5079
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.7676347494125366,
      "learning_rate": 0.0005992806031676353,
      "loss": 2.5418,
      "step": 5080
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.7842025756835938,
      "learning_rate": 0.0005992803200259652,
      "loss": 2.5464,
      "step": 5081
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.7061057090759277,
      "learning_rate": 0.0005992800368286533,
      "loss": 2.2805,
      "step": 5082
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.7938840985298157,
      "learning_rate": 0.0005992797535756995,
      "loss": 2.7546,
      "step": 5083
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.7000730037689209,
      "learning_rate": 0.0005992794702671041,
      "loss": 2.8437,
      "step": 5084
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.7062889337539673,
      "learning_rate": 0.000599279186902867,
      "loss": 2.8826,
      "step": 5085
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.8407689929008484,
      "learning_rate": 0.0005992789034829881,
      "loss": 2.5469,
      "step": 5086
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.6935869455337524,
      "learning_rate": 0.0005992786200074676,
      "loss": 2.682,
      "step": 5087
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.7982110977172852,
      "learning_rate": 0.0005992783364763058,
      "loss": 2.4884,
      "step": 5088
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.7395660877227783,
      "learning_rate": 0.0005992780528895023,
      "loss": 2.5615,
      "step": 5089
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.6808178424835205,
      "learning_rate": 0.0005992777692470574,
      "loss": 2.4986,
      "step": 5090
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.7387729287147522,
      "learning_rate": 0.0005992774855489711,
      "loss": 2.4832,
      "step": 5091
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.6452436447143555,
      "learning_rate": 0.0005992772017952434,
      "loss": 2.4786,
      "step": 5092
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.6156134009361267,
      "learning_rate": 0.0005992769179858745,
      "loss": 2.6602,
      "step": 5093
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.8988029360771179,
      "learning_rate": 0.0005992766341208645,
      "loss": 2.4289,
      "step": 5094
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.6968806385993958,
      "learning_rate": 0.000599276350200213,
      "loss": 2.8553,
      "step": 5095
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.8189297318458557,
      "learning_rate": 0.0005992760662239207,
      "loss": 2.5092,
      "step": 5096
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.7001622319221497,
      "learning_rate": 0.0005992757821919872,
      "loss": 2.6135,
      "step": 5097
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.8640801310539246,
      "learning_rate": 0.0005992754981044127,
      "loss": 2.7191,
      "step": 5098
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.6876189708709717,
      "learning_rate": 0.0005992752139611972,
      "loss": 2.4935,
      "step": 5099
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.7407279014587402,
      "learning_rate": 0.0005992749297623408,
      "loss": 2.4495,
      "step": 5100
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.7736696600914001,
      "learning_rate": 0.0005992746455078436,
      "loss": 2.5887,
      "step": 5101
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.669057309627533,
      "learning_rate": 0.0005992743611977055,
      "loss": 2.6061,
      "step": 5102
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.6956782937049866,
      "learning_rate": 0.0005992740768319268,
      "loss": 2.8529,
      "step": 5103
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.8002898693084717,
      "learning_rate": 0.0005992737924105072,
      "loss": 2.6687,
      "step": 5104
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.6734086871147156,
      "learning_rate": 0.000599273507933447,
      "loss": 2.5413,
      "step": 5105
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.8333693742752075,
      "learning_rate": 0.0005992732234007462,
      "loss": 2.6014,
      "step": 5106
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.7007728815078735,
      "learning_rate": 0.000599272938812405,
      "loss": 2.6181,
      "step": 5107
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.9100466966629028,
      "learning_rate": 0.0005992726541684233,
      "loss": 2.6331,
      "step": 5108
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.9892756938934326,
      "learning_rate": 0.000599272369468801,
      "loss": 2.6243,
      "step": 5109
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.7760069370269775,
      "learning_rate": 0.0005992720847135385,
      "loss": 2.8842,
      "step": 5110
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.072107195854187,
      "learning_rate": 0.0005992717999026355,
      "loss": 2.592,
      "step": 5111
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.7298984527587891,
      "learning_rate": 0.0005992715150360923,
      "loss": 2.6904,
      "step": 5112
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.8041223287582397,
      "learning_rate": 0.0005992712301139089,
      "loss": 2.5085,
      "step": 5113
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.9644579887390137,
      "learning_rate": 0.0005992709451360853,
      "loss": 2.6415,
      "step": 5114
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.6753812432289124,
      "learning_rate": 0.0005992706601026216,
      "loss": 2.5063,
      "step": 5115
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.7325272560119629,
      "learning_rate": 0.0005992703750135178,
      "loss": 2.5055,
      "step": 5116
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.8192936778068542,
      "learning_rate": 0.0005992700898687739,
      "loss": 2.6616,
      "step": 5117
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.662748396396637,
      "learning_rate": 0.0005992698046683902,
      "loss": 2.7581,
      "step": 5118
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.6451181769371033,
      "learning_rate": 0.0005992695194123666,
      "loss": 2.3933,
      "step": 5119
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.6990336179733276,
      "learning_rate": 0.000599269234100703,
      "loss": 2.7567,
      "step": 5120
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.726179301738739,
      "learning_rate": 0.0005992689487333998,
      "loss": 2.4753,
      "step": 5121
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.688397228717804,
      "learning_rate": 0.0005992686633104568,
      "loss": 2.7546,
      "step": 5122
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.7125899195671082,
      "learning_rate": 0.000599268377831874,
      "loss": 2.6331,
      "step": 5123
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.6669923663139343,
      "learning_rate": 0.0005992680922976516,
      "loss": 2.5704,
      "step": 5124
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.692064642906189,
      "learning_rate": 0.0005992678067077896,
      "loss": 2.8061,
      "step": 5125
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.7465225458145142,
      "learning_rate": 0.000599267521062288,
      "loss": 2.7099,
      "step": 5126
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.696837306022644,
      "learning_rate": 0.0005992672353611471,
      "loss": 2.5612,
      "step": 5127
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.6896172165870667,
      "learning_rate": 0.0005992669496043667,
      "loss": 2.4789,
      "step": 5128
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.7101843953132629,
      "learning_rate": 0.0005992666637919469,
      "loss": 2.597,
      "step": 5129
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.6988750100135803,
      "learning_rate": 0.0005992663779238877,
      "loss": 2.6031,
      "step": 5130
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.6269966959953308,
      "learning_rate": 0.0005992660920001894,
      "loss": 2.7032,
      "step": 5131
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.6789631843566895,
      "learning_rate": 0.0005992658060208518,
      "loss": 2.6294,
      "step": 5132
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.7173779010772705,
      "learning_rate": 0.000599265519985875,
      "loss": 2.7408,
      "step": 5133
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.7625650763511658,
      "learning_rate": 0.000599265233895259,
      "loss": 2.767,
      "step": 5134
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.6804271340370178,
      "learning_rate": 0.0005992649477490042,
      "loss": 2.7268,
      "step": 5135
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.6653990149497986,
      "learning_rate": 0.0005992646615471102,
      "loss": 2.6068,
      "step": 5136
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.6527750492095947,
      "learning_rate": 0.0005992643752895774,
      "loss": 2.6467,
      "step": 5137
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.7473245859146118,
      "learning_rate": 0.0005992640889764056,
      "loss": 2.4135,
      "step": 5138
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.6842423677444458,
      "learning_rate": 0.000599263802607595,
      "loss": 2.5188,
      "step": 5139
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.8077822327613831,
      "learning_rate": 0.0005992635161831457,
      "loss": 2.5917,
      "step": 5140
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.7848588228225708,
      "learning_rate": 0.0005992632297030574,
      "loss": 2.6745,
      "step": 5141
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.8244043588638306,
      "learning_rate": 0.0005992629431673305,
      "loss": 2.5621,
      "step": 5142
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.7061782479286194,
      "learning_rate": 0.0005992626565759651,
      "loss": 2.565,
      "step": 5143
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.8507897853851318,
      "learning_rate": 0.0005992623699289611,
      "loss": 2.7381,
      "step": 5144
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.638895571231842,
      "learning_rate": 0.0005992620832263186,
      "loss": 2.6331,
      "step": 5145
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.7298018932342529,
      "learning_rate": 0.0005992617964680375,
      "loss": 2.6995,
      "step": 5146
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.7992428541183472,
      "learning_rate": 0.000599261509654118,
      "loss": 2.6133,
      "step": 5147
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.7097892165184021,
      "learning_rate": 0.0005992612227845602,
      "loss": 2.6067,
      "step": 5148
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.816429078578949,
      "learning_rate": 0.0005992609358593641,
      "loss": 2.7131,
      "step": 5149
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.6722728610038757,
      "learning_rate": 0.0005992606488785297,
      "loss": 2.6984,
      "step": 5150
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.694780170917511,
      "learning_rate": 0.0005992603618420572,
      "loss": 2.562,
      "step": 5151
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.7136508822441101,
      "learning_rate": 0.0005992600747499464,
      "loss": 2.5309,
      "step": 5152
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.6778815984725952,
      "learning_rate": 0.0005992597876021976,
      "loss": 2.5106,
      "step": 5153
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.6585105061531067,
      "learning_rate": 0.0005992595003988107,
      "loss": 2.4183,
      "step": 5154
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.6617410778999329,
      "learning_rate": 0.0005992592131397859,
      "loss": 2.4828,
      "step": 5155
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.6708059906959534,
      "learning_rate": 0.000599258925825123,
      "loss": 2.6761,
      "step": 5156
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.6828593015670776,
      "learning_rate": 0.0005992586384548224,
      "loss": 2.699,
      "step": 5157
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.7252886891365051,
      "learning_rate": 0.0005992583510288839,
      "loss": 2.9139,
      "step": 5158
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.6579103469848633,
      "learning_rate": 0.0005992580635473076,
      "loss": 2.7567,
      "step": 5159
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.7583925127983093,
      "learning_rate": 0.0005992577760100935,
      "loss": 2.5721,
      "step": 5160
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.7178638577461243,
      "learning_rate": 0.0005992574884172418,
      "loss": 2.8049,
      "step": 5161
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.7479174733161926,
      "learning_rate": 0.0005992572007687526,
      "loss": 2.718,
      "step": 5162
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.6375194191932678,
      "learning_rate": 0.0005992569130646257,
      "loss": 2.5587,
      "step": 5163
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.693152666091919,
      "learning_rate": 0.0005992566253048613,
      "loss": 2.7145,
      "step": 5164
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.6942041516304016,
      "learning_rate": 0.0005992563374894595,
      "loss": 2.8464,
      "step": 5165
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.7274358868598938,
      "learning_rate": 0.0005992560496184202,
      "loss": 2.6886,
      "step": 5166
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.6963492631912231,
      "learning_rate": 0.0005992557616917437,
      "loss": 2.8164,
      "step": 5167
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.712076723575592,
      "learning_rate": 0.0005992554737094298,
      "loss": 2.5271,
      "step": 5168
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.7068694829940796,
      "learning_rate": 0.0005992551856714788,
      "loss": 2.6739,
      "step": 5169
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.7127099633216858,
      "learning_rate": 0.0005992548975778904,
      "loss": 2.8273,
      "step": 5170
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.7359651327133179,
      "learning_rate": 0.0005992546094286649,
      "loss": 2.3536,
      "step": 5171
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.8004870414733887,
      "learning_rate": 0.0005992543212238025,
      "loss": 2.5147,
      "step": 5172
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.6280298829078674,
      "learning_rate": 0.0005992540329633029,
      "loss": 2.7258,
      "step": 5173
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.7419995665550232,
      "learning_rate": 0.0005992537446471663,
      "loss": 2.6023,
      "step": 5174
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.6911303997039795,
      "learning_rate": 0.0005992534562753929,
      "loss": 2.4841,
      "step": 5175
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.6644418835639954,
      "learning_rate": 0.0005992531678479826,
      "loss": 2.6088,
      "step": 5176
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.6810562610626221,
      "learning_rate": 0.0005992528793649356,
      "loss": 2.6094,
      "step": 5177
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.6398979425430298,
      "learning_rate": 0.0005992525908262516,
      "loss": 2.7219,
      "step": 5178
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.6645269393920898,
      "learning_rate": 0.0005992523022319311,
      "loss": 2.702,
      "step": 5179
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.7264502644538879,
      "learning_rate": 0.0005992520135819737,
      "loss": 2.516,
      "step": 5180
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.7380719780921936,
      "learning_rate": 0.0005992517248763799,
      "loss": 2.7809,
      "step": 5181
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.7113589644432068,
      "learning_rate": 0.0005992514361151496,
      "loss": 2.6129,
      "step": 5182
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.6430318355560303,
      "learning_rate": 0.0005992511472982827,
      "loss": 2.4737,
      "step": 5183
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.7322909832000732,
      "learning_rate": 0.0005992508584257794,
      "loss": 2.8967,
      "step": 5184
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.7696714997291565,
      "learning_rate": 0.0005992505694976398,
      "loss": 2.5441,
      "step": 5185
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.7778149247169495,
      "learning_rate": 0.0005992502805138637,
      "loss": 2.6525,
      "step": 5186
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.7707564234733582,
      "learning_rate": 0.0005992499914744514,
      "loss": 2.5356,
      "step": 5187
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.6974090337753296,
      "learning_rate": 0.0005992497023794028,
      "loss": 2.4651,
      "step": 5188
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.64219731092453,
      "learning_rate": 0.0005992494132287182,
      "loss": 2.6362,
      "step": 5189
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.6733871102333069,
      "learning_rate": 0.0005992491240223974,
      "loss": 2.5884,
      "step": 5190
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.7132214903831482,
      "learning_rate": 0.0005992488347604406,
      "loss": 2.9111,
      "step": 5191
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.8136661052703857,
      "learning_rate": 0.0005992485454428477,
      "loss": 2.6017,
      "step": 5192
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.6706809997558594,
      "learning_rate": 0.0005992482560696189,
      "loss": 2.6699,
      "step": 5193
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.6751756072044373,
      "learning_rate": 0.0005992479666407542,
      "loss": 2.7401,
      "step": 5194
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.6376691460609436,
      "learning_rate": 0.0005992476771562536,
      "loss": 2.6675,
      "step": 5195
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.7172503471374512,
      "learning_rate": 0.0005992473876161173,
      "loss": 2.8248,
      "step": 5196
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.7051689624786377,
      "learning_rate": 0.0005992470980203451,
      "loss": 2.6234,
      "step": 5197
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.6312296986579895,
      "learning_rate": 0.0005992468083689373,
      "loss": 2.4587,
      "step": 5198
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.7721794247627258,
      "learning_rate": 0.0005992465186618939,
      "loss": 2.8995,
      "step": 5199
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.631635308265686,
      "learning_rate": 0.000599246228899215,
      "loss": 2.692,
      "step": 5200
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.6871700882911682,
      "learning_rate": 0.0005992459390809005,
      "loss": 2.6129,
      "step": 5201
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.7579436302185059,
      "learning_rate": 0.0005992456492069506,
      "loss": 2.6692,
      "step": 5202
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.6849632263183594,
      "learning_rate": 0.0005992453592773651,
      "loss": 2.7489,
      "step": 5203
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.6607506275177002,
      "learning_rate": 0.0005992450692921444,
      "loss": 2.5577,
      "step": 5204
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.6831445097923279,
      "learning_rate": 0.0005992447792512882,
      "loss": 2.6907,
      "step": 5205
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.6724063754081726,
      "learning_rate": 0.000599244489154797,
      "loss": 2.2901,
      "step": 5206
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.648914098739624,
      "learning_rate": 0.0005992441990026704,
      "loss": 2.6708,
      "step": 5207
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.759148359298706,
      "learning_rate": 0.0005992439087949087,
      "loss": 2.8765,
      "step": 5208
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.783594012260437,
      "learning_rate": 0.0005992436185315121,
      "loss": 2.6132,
      "step": 5209
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.7132992744445801,
      "learning_rate": 0.0005992433282124803,
      "loss": 2.6106,
      "step": 5210
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.6998061537742615,
      "learning_rate": 0.0005992430378378135,
      "loss": 2.9336,
      "step": 5211
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.9113659858703613,
      "learning_rate": 0.0005992427474075118,
      "loss": 2.5015,
      "step": 5212
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.7956206798553467,
      "learning_rate": 0.0005992424569215752,
      "loss": 2.9041,
      "step": 5213
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.7728422284126282,
      "learning_rate": 0.0005992421663800038,
      "loss": 2.6126,
      "step": 5214
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.7062347531318665,
      "learning_rate": 0.0005992418757827977,
      "loss": 2.6903,
      "step": 5215
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.7302107214927673,
      "learning_rate": 0.0005992415851299568,
      "loss": 2.6133,
      "step": 5216
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.8515512943267822,
      "learning_rate": 0.0005992412944214813,
      "loss": 2.7597,
      "step": 5217
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.9512324333190918,
      "learning_rate": 0.0005992410036573711,
      "loss": 2.3757,
      "step": 5218
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.8047437071800232,
      "learning_rate": 0.0005992407128376264,
      "loss": 2.5838,
      "step": 5219
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.8422882556915283,
      "learning_rate": 0.0005992404219622472,
      "loss": 2.8103,
      "step": 5220
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.8131340742111206,
      "learning_rate": 0.0005992401310312335,
      "loss": 2.8122,
      "step": 5221
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.7254877686500549,
      "learning_rate": 0.0005992398400445855,
      "loss": 2.6096,
      "step": 5222
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.1121293306350708,
      "learning_rate": 0.0005992395490023033,
      "loss": 2.7771,
      "step": 5223
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.7756050825119019,
      "learning_rate": 0.0005992392579043865,
      "loss": 2.498,
      "step": 5224
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.6348134279251099,
      "learning_rate": 0.0005992389667508356,
      "loss": 2.4209,
      "step": 5225
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.8512864112854004,
      "learning_rate": 0.0005992386755416506,
      "loss": 2.7635,
      "step": 5226
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.7223175168037415,
      "learning_rate": 0.0005992383842768315,
      "loss": 2.6864,
      "step": 5227
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.8297067284584045,
      "learning_rate": 0.0005992380929563783,
      "loss": 2.5018,
      "step": 5228
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.7761753797531128,
      "learning_rate": 0.000599237801580291,
      "loss": 2.6842,
      "step": 5229
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.7027066946029663,
      "learning_rate": 0.0005992375101485698,
      "loss": 2.6358,
      "step": 5230
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.7302724123001099,
      "learning_rate": 0.0005992372186612148,
      "loss": 2.6703,
      "step": 5231
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.6804288625717163,
      "learning_rate": 0.0005992369271182258,
      "loss": 2.7716,
      "step": 5232
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.677679717540741,
      "learning_rate": 0.000599236635519603,
      "loss": 2.5674,
      "step": 5233
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.6573925018310547,
      "learning_rate": 0.0005992363438653466,
      "loss": 2.6595,
      "step": 5234
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.6395515203475952,
      "learning_rate": 0.0005992360521554564,
      "loss": 2.4508,
      "step": 5235
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.7010831832885742,
      "learning_rate": 0.0005992357603899327,
      "loss": 2.8188,
      "step": 5236
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.6637580394744873,
      "learning_rate": 0.0005992354685687754,
      "loss": 2.6764,
      "step": 5237
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.6300538778305054,
      "learning_rate": 0.0005992351766919845,
      "loss": 2.8932,
      "step": 5238
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.6510240435600281,
      "learning_rate": 0.0005992348847595601,
      "loss": 2.5999,
      "step": 5239
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.6883841753005981,
      "learning_rate": 0.0005992345927715024,
      "loss": 2.5898,
      "step": 5240
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.7152319550514221,
      "learning_rate": 0.0005992343007278113,
      "loss": 2.8262,
      "step": 5241
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.6794715523719788,
      "learning_rate": 0.0005992340086284868,
      "loss": 2.6933,
      "step": 5242
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.6429222822189331,
      "learning_rate": 0.0005992337164735292,
      "loss": 2.5359,
      "step": 5243
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.7697700262069702,
      "learning_rate": 0.0005992334242629383,
      "loss": 2.6193,
      "step": 5244
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.6360096335411072,
      "learning_rate": 0.0005992331319967142,
      "loss": 2.741,
      "step": 5245
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.8913385272026062,
      "learning_rate": 0.0005992328396748572,
      "loss": 2.4383,
      "step": 5246
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.7515619397163391,
      "learning_rate": 0.000599232547297367,
      "loss": 2.7081,
      "step": 5247
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.6655393838882446,
      "learning_rate": 0.0005992322548642439,
      "loss": 2.5489,
      "step": 5248
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.7513810992240906,
      "learning_rate": 0.0005992319623754878,
      "loss": 2.5518,
      "step": 5249
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.6485647559165955,
      "learning_rate": 0.000599231669831099,
      "loss": 2.672,
      "step": 5250
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.8244378566741943,
      "learning_rate": 0.0005992313772310772,
      "loss": 2.5718,
      "step": 5251
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.7208713889122009,
      "learning_rate": 0.0005992310845754228,
      "loss": 2.6388,
      "step": 5252
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.7152236700057983,
      "learning_rate": 0.0005992307918641355,
      "loss": 2.3797,
      "step": 5253
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.7004865407943726,
      "learning_rate": 0.0005992304990972157,
      "loss": 2.5838,
      "step": 5254
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.7552081346511841,
      "learning_rate": 0.0005992302062746632,
      "loss": 2.6456,
      "step": 5255
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.7325856685638428,
      "learning_rate": 0.0005992299133964783,
      "loss": 2.6996,
      "step": 5256
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.6838219165802002,
      "learning_rate": 0.0005992296204626608,
      "loss": 2.5178,
      "step": 5257
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.736932098865509,
      "learning_rate": 0.0005992293274732109,
      "loss": 2.5373,
      "step": 5258
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.6898115277290344,
      "learning_rate": 0.0005992290344281286,
      "loss": 2.7627,
      "step": 5259
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.777316153049469,
      "learning_rate": 0.0005992287413274139,
      "loss": 2.5667,
      "step": 5260
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.739628255367279,
      "learning_rate": 0.000599228448171067,
      "loss": 2.6583,
      "step": 5261
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.6645281910896301,
      "learning_rate": 0.0005992281549590879,
      "loss": 2.7337,
      "step": 5262
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.8265987038612366,
      "learning_rate": 0.0005992278616914767,
      "loss": 2.8256,
      "step": 5263
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.6844860911369324,
      "learning_rate": 0.0005992275683682333,
      "loss": 2.6158,
      "step": 5264
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.7762196660041809,
      "learning_rate": 0.0005992272749893579,
      "loss": 2.7209,
      "step": 5265
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.7394220232963562,
      "learning_rate": 0.0005992269815548504,
      "loss": 2.5597,
      "step": 5266
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.6578311920166016,
      "learning_rate": 0.000599226688064711,
      "loss": 2.5923,
      "step": 5267
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.6850423216819763,
      "learning_rate": 0.0005992263945189398,
      "loss": 2.7197,
      "step": 5268
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.6448923945426941,
      "learning_rate": 0.0005992261009175367,
      "loss": 2.6472,
      "step": 5269
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.7247678637504578,
      "learning_rate": 0.0005992258072605018,
      "loss": 2.5055,
      "step": 5270
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.5740436315536499,
      "learning_rate": 0.0005992255135478352,
      "loss": 2.4145,
      "step": 5271
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.6409401297569275,
      "learning_rate": 0.0005992252197795369,
      "loss": 2.636,
      "step": 5272
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.62470942735672,
      "learning_rate": 0.0005992249259556071,
      "loss": 2.5368,
      "step": 5273
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.6762206554412842,
      "learning_rate": 0.0005992246320760456,
      "loss": 2.4682,
      "step": 5274
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.7073214650154114,
      "learning_rate": 0.0005992243381408525,
      "loss": 2.6111,
      "step": 5275
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.7671424746513367,
      "learning_rate": 0.0005992240441500282,
      "loss": 2.3553,
      "step": 5276
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.7113533616065979,
      "learning_rate": 0.0005992237501035724,
      "loss": 2.7953,
      "step": 5277
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.7564213871955872,
      "learning_rate": 0.0005992234560014852,
      "loss": 2.7358,
      "step": 5278
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.7984827756881714,
      "learning_rate": 0.0005992231618437668,
      "loss": 2.449,
      "step": 5279
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.7143754363059998,
      "learning_rate": 0.0005992228676304171,
      "loss": 2.4174,
      "step": 5280
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.6523286700248718,
      "learning_rate": 0.0005992225733614363,
      "loss": 2.7944,
      "step": 5281
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.6251115202903748,
      "learning_rate": 0.0005992222790368242,
      "loss": 2.6464,
      "step": 5282
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.6363496780395508,
      "learning_rate": 0.0005992219846565812,
      "loss": 2.5393,
      "step": 5283
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.6929298639297485,
      "learning_rate": 0.0005992216902207071,
      "loss": 2.7336,
      "step": 5284
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.6848574876785278,
      "learning_rate": 0.0005992213957292022,
      "loss": 2.8458,
      "step": 5285
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.7948898077011108,
      "learning_rate": 0.0005992211011820661,
      "loss": 2.5461,
      "step": 5286
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.6493974924087524,
      "learning_rate": 0.0005992208065792994,
      "loss": 2.6852,
      "step": 5287
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.7169027328491211,
      "learning_rate": 0.0005992205119209019,
      "loss": 2.6895,
      "step": 5288
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.6991845369338989,
      "learning_rate": 0.0005992202172068735,
      "loss": 2.8897,
      "step": 5289
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.6827204823493958,
      "learning_rate": 0.0005992199224372145,
      "loss": 2.5295,
      "step": 5290
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.7439314723014832,
      "learning_rate": 0.0005992196276119249,
      "loss": 2.5668,
      "step": 5291
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.7594441175460815,
      "learning_rate": 0.0005992193327310047,
      "loss": 2.5959,
      "step": 5292
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.7714183926582336,
      "learning_rate": 0.0005992190377944539,
      "loss": 2.5719,
      "step": 5293
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.6842137575149536,
      "learning_rate": 0.0005992187428022726,
      "loss": 2.6758,
      "step": 5294
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.8543329238891602,
      "learning_rate": 0.000599218447754461,
      "loss": 2.7226,
      "step": 5295
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.6564154028892517,
      "learning_rate": 0.0005992181526510191,
      "loss": 2.5713,
      "step": 5296
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.8995875716209412,
      "learning_rate": 0.0005992178574919468,
      "loss": 2.7314,
      "step": 5297
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.7539907097816467,
      "learning_rate": 0.0005992175622772443,
      "loss": 2.6683,
      "step": 5298
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.6667256355285645,
      "learning_rate": 0.0005992172670069116,
      "loss": 2.63,
      "step": 5299
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.7562835216522217,
      "learning_rate": 0.0005992169716809487,
      "loss": 2.6772,
      "step": 5300
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.730745255947113,
      "learning_rate": 0.0005992166762993557,
      "loss": 2.5082,
      "step": 5301
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.7150346040725708,
      "learning_rate": 0.0005992163808621328,
      "loss": 2.6788,
      "step": 5302
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.8869476914405823,
      "learning_rate": 0.0005992160853692798,
      "loss": 2.542,
      "step": 5303
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.7381536960601807,
      "learning_rate": 0.0005992157898207969,
      "loss": 2.8903,
      "step": 5304
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.0142478942871094,
      "learning_rate": 0.0005992154942166842,
      "loss": 2.6116,
      "step": 5305
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.682136058807373,
      "learning_rate": 0.0005992151985569417,
      "loss": 2.4452,
      "step": 5306
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.7391704320907593,
      "learning_rate": 0.0005992149028415693,
      "loss": 2.6545,
      "step": 5307
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.9628739953041077,
      "learning_rate": 0.0005992146070705674,
      "loss": 2.7384,
      "step": 5308
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.6685848236083984,
      "learning_rate": 0.0005992143112439357,
      "loss": 2.8009,
      "step": 5309
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.7299898862838745,
      "learning_rate": 0.0005992140153616746,
      "loss": 2.529,
      "step": 5310
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.8250930309295654,
      "learning_rate": 0.0005992137194237838,
      "loss": 2.5351,
      "step": 5311
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.6232765913009644,
      "learning_rate": 0.0005992134234302635,
      "loss": 2.6742,
      "step": 5312
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.6723935008049011,
      "learning_rate": 0.0005992131273811139,
      "loss": 2.545,
      "step": 5313
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.8366963863372803,
      "learning_rate": 0.0005992128312763349,
      "loss": 2.6136,
      "step": 5314
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.666304886341095,
      "learning_rate": 0.0005992125351159265,
      "loss": 2.5483,
      "step": 5315
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.7637349963188171,
      "learning_rate": 0.0005992122388998889,
      "loss": 2.2894,
      "step": 5316
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.7564653158187866,
      "learning_rate": 0.0005992119426282221,
      "loss": 2.3913,
      "step": 5317
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.6579099893569946,
      "learning_rate": 0.0005992116463009262,
      "loss": 2.4676,
      "step": 5318
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.8935544490814209,
      "learning_rate": 0.0005992113499180011,
      "loss": 2.6968,
      "step": 5319
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.7297558784484863,
      "learning_rate": 0.0005992110534794471,
      "loss": 2.6712,
      "step": 5320
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.760887622833252,
      "learning_rate": 0.000599210756985264,
      "loss": 2.4733,
      "step": 5321
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.9051790833473206,
      "learning_rate": 0.000599210460435452,
      "loss": 2.5996,
      "step": 5322
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.7253775596618652,
      "learning_rate": 0.0005992101638300112,
      "loss": 2.6303,
      "step": 5323
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.8441263437271118,
      "learning_rate": 0.0005992098671689414,
      "loss": 2.5992,
      "step": 5324
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.6803929805755615,
      "learning_rate": 0.000599209570452243,
      "loss": 2.7144,
      "step": 5325
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.772695004940033,
      "learning_rate": 0.0005992092736799159,
      "loss": 2.6609,
      "step": 5326
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.6606248021125793,
      "learning_rate": 0.00059920897685196,
      "loss": 2.5889,
      "step": 5327
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.7318482398986816,
      "learning_rate": 0.0005992086799683757,
      "loss": 2.8124,
      "step": 5328
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.6531566977500916,
      "learning_rate": 0.0005992083830291627,
      "loss": 2.416,
      "step": 5329
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.6841224431991577,
      "learning_rate": 0.0005992080860343213,
      "loss": 2.6834,
      "step": 5330
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.659809947013855,
      "learning_rate": 0.0005992077889838515,
      "loss": 2.4984,
      "step": 5331
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.7422845363616943,
      "learning_rate": 0.0005992074918777533,
      "loss": 2.637,
      "step": 5332
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.7590669989585876,
      "learning_rate": 0.0005992071947160268,
      "loss": 2.8559,
      "step": 5333
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.6766732931137085,
      "learning_rate": 0.0005992068974986719,
      "loss": 2.9189,
      "step": 5334
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.6909448504447937,
      "learning_rate": 0.0005992066002256888,
      "loss": 2.6489,
      "step": 5335
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.6883195638656616,
      "learning_rate": 0.0005992063028970777,
      "loss": 2.7583,
      "step": 5336
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.6988719701766968,
      "learning_rate": 0.0005992060055128384,
      "loss": 2.9103,
      "step": 5337
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.6715999245643616,
      "learning_rate": 0.000599205708072971,
      "loss": 2.7017,
      "step": 5338
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.6780749559402466,
      "learning_rate": 0.0005992054105774757,
      "loss": 2.8547,
      "step": 5339
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.5956043601036072,
      "learning_rate": 0.0005992051130263525,
      "loss": 2.7391,
      "step": 5340
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.7206576466560364,
      "learning_rate": 0.0005992048154196014,
      "loss": 2.6139,
      "step": 5341
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.6628568172454834,
      "learning_rate": 0.0005992045177572224,
      "loss": 2.3787,
      "step": 5342
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.6702597141265869,
      "learning_rate": 0.0005992042200392157,
      "loss": 2.5543,
      "step": 5343
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.6387813687324524,
      "learning_rate": 0.0005992039222655813,
      "loss": 2.5737,
      "step": 5344
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.6522732973098755,
      "learning_rate": 0.0005992036244363191,
      "loss": 2.4443,
      "step": 5345
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.736820638179779,
      "learning_rate": 0.0005992033265514294,
      "loss": 2.7471,
      "step": 5346
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.6514979600906372,
      "learning_rate": 0.0005992030286109122,
      "loss": 2.8545,
      "step": 5347
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.7819927334785461,
      "learning_rate": 0.0005992027306147675,
      "loss": 2.6405,
      "step": 5348
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.670059084892273,
      "learning_rate": 0.0005992024325629954,
      "loss": 2.5883,
      "step": 5349
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.6320812702178955,
      "learning_rate": 0.0005992021344555958,
      "loss": 2.7597,
      "step": 5350
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.1556237936019897,
      "learning_rate": 0.000599201836292569,
      "loss": 2.85,
      "step": 5351
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.7317711114883423,
      "learning_rate": 0.0005992015380739148,
      "loss": 2.5512,
      "step": 5352
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.8485316038131714,
      "learning_rate": 0.0005992012397996334,
      "loss": 2.7275,
      "step": 5353
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.9862197041511536,
      "learning_rate": 0.0005992009414697249,
      "loss": 2.5052,
      "step": 5354
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.6796484589576721,
      "learning_rate": 0.0005992006430841892,
      "loss": 2.4647,
      "step": 5355
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.9366391897201538,
      "learning_rate": 0.0005992003446430267,
      "loss": 2.787,
      "step": 5356
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.8211142420768738,
      "learning_rate": 0.000599200046146237,
      "loss": 2.8774,
      "step": 5357
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.7059133052825928,
      "learning_rate": 0.0005991997475938204,
      "loss": 2.4569,
      "step": 5358
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.8372510671615601,
      "learning_rate": 0.000599199448985777,
      "loss": 2.4787,
      "step": 5359
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.838367760181427,
      "learning_rate": 0.0005991991503221067,
      "loss": 2.6528,
      "step": 5360
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.7460452318191528,
      "learning_rate": 0.0005991988516028095,
      "loss": 2.3707,
      "step": 5361
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.8045535087585449,
      "learning_rate": 0.0005991985528278858,
      "loss": 2.4524,
      "step": 5362
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.6778832077980042,
      "learning_rate": 0.0005991982539973353,
      "loss": 2.8151,
      "step": 5363
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.7040033936500549,
      "learning_rate": 0.0005991979551111583,
      "loss": 2.4953,
      "step": 5364
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.6739779114723206,
      "learning_rate": 0.0005991976561693547,
      "loss": 2.7719,
      "step": 5365
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.7372159957885742,
      "learning_rate": 0.0005991973571719246,
      "loss": 2.6617,
      "step": 5366
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.7490643858909607,
      "learning_rate": 0.0005991970581188681,
      "loss": 2.8592,
      "step": 5367
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.6967076063156128,
      "learning_rate": 0.0005991967590101853,
      "loss": 2.8067,
      "step": 5368
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.7432982325553894,
      "learning_rate": 0.0005991964598458761,
      "loss": 2.7831,
      "step": 5369
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.6521668434143066,
      "learning_rate": 0.0005991961606259406,
      "loss": 2.6733,
      "step": 5370
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.6584402322769165,
      "learning_rate": 0.0005991958613503788,
      "loss": 2.5181,
      "step": 5371
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.7674710154533386,
      "learning_rate": 0.0005991955620191909,
      "loss": 2.6749,
      "step": 5372
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.6764474511146545,
      "learning_rate": 0.000599195262632377,
      "loss": 2.4525,
      "step": 5373
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.7607212066650391,
      "learning_rate": 0.0005991949631899369,
      "loss": 2.7278,
      "step": 5374
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.6842879056930542,
      "learning_rate": 0.0005991946636918709,
      "loss": 2.5694,
      "step": 5375
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.7033932209014893,
      "learning_rate": 0.000599194364138179,
      "loss": 2.5949,
      "step": 5376
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.6810954809188843,
      "learning_rate": 0.0005991940645288611,
      "loss": 2.5449,
      "step": 5377
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.7296420931816101,
      "learning_rate": 0.0005991937648639176,
      "loss": 2.7243,
      "step": 5378
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.7362990975379944,
      "learning_rate": 0.0005991934651433481,
      "loss": 2.8746,
      "step": 5379
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.7781054973602295,
      "learning_rate": 0.0005991931653671531,
      "loss": 2.6719,
      "step": 5380
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.6832945942878723,
      "learning_rate": 0.0005991928655353322,
      "loss": 2.6135,
      "step": 5381
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.8072234392166138,
      "learning_rate": 0.0005991925656478859,
      "loss": 2.5214,
      "step": 5382
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.7009255290031433,
      "learning_rate": 0.0005991922657048139,
      "loss": 2.8495,
      "step": 5383
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.6604430079460144,
      "learning_rate": 0.0005991919657061164,
      "loss": 2.5316,
      "step": 5384
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.7145859599113464,
      "learning_rate": 0.0005991916656517936,
      "loss": 2.6518,
      "step": 5385
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.6848345398902893,
      "learning_rate": 0.0005991913655418454,
      "loss": 2.5865,
      "step": 5386
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.7172030210494995,
      "learning_rate": 0.0005991910653762719,
      "loss": 2.7475,
      "step": 5387
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.6590501070022583,
      "learning_rate": 0.000599190765155073,
      "loss": 2.5891,
      "step": 5388
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.9162020087242126,
      "learning_rate": 0.000599190464878249,
      "loss": 2.4971,
      "step": 5389
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.8108726739883423,
      "learning_rate": 0.0005991901645457997,
      "loss": 2.558,
      "step": 5390
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.8788207173347473,
      "learning_rate": 0.0005991898641577255,
      "loss": 2.6674,
      "step": 5391
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.0148121118545532,
      "learning_rate": 0.0005991895637140262,
      "loss": 2.4904,
      "step": 5392
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.6395766139030457,
      "learning_rate": 0.0005991892632147019,
      "loss": 2.5708,
      "step": 5393
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.8744062781333923,
      "learning_rate": 0.0005991889626597525,
      "loss": 2.6885,
      "step": 5394
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.102246642112732,
      "learning_rate": 0.0005991886620491785,
      "loss": 2.4621,
      "step": 5395
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.6333746910095215,
      "learning_rate": 0.0005991883613829795,
      "loss": 2.7679,
      "step": 5396
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.9269446730613708,
      "learning_rate": 0.0005991880606611557,
      "loss": 2.7723,
      "step": 5397
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.8867136836051941,
      "learning_rate": 0.0005991877598837073,
      "loss": 2.5029,
      "step": 5398
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.5929705500602722,
      "learning_rate": 0.0005991874590506343,
      "loss": 2.5184,
      "step": 5399
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.7045782804489136,
      "learning_rate": 0.0005991871581619366,
      "loss": 2.8088,
      "step": 5400
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.7906200885772705,
      "learning_rate": 0.0005991868572176144,
      "loss": 2.634,
      "step": 5401
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.7976237535476685,
      "learning_rate": 0.0005991865562176678,
      "loss": 2.4534,
      "step": 5402
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.67075115442276,
      "learning_rate": 0.0005991862551620966,
      "loss": 2.6807,
      "step": 5403
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.8076728582382202,
      "learning_rate": 0.0005991859540509011,
      "loss": 2.6173,
      "step": 5404
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.7215625047683716,
      "learning_rate": 0.0005991856528840812,
      "loss": 2.3857,
      "step": 5405
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.6365950107574463,
      "learning_rate": 0.0005991853516616372,
      "loss": 2.6123,
      "step": 5406
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.8818864226341248,
      "learning_rate": 0.000599185050383569,
      "loss": 2.428,
      "step": 5407
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.674961268901825,
      "learning_rate": 0.0005991847490498765,
      "loss": 2.8315,
      "step": 5408
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.6486936211585999,
      "learning_rate": 0.00059918444766056,
      "loss": 2.6156,
      "step": 5409
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.6690165996551514,
      "learning_rate": 0.0005991841462156194,
      "loss": 2.7692,
      "step": 5410
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.8309535980224609,
      "learning_rate": 0.0005991838447150549,
      "loss": 2.3023,
      "step": 5411
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.8845574259757996,
      "learning_rate": 0.0005991835431588665,
      "loss": 2.4782,
      "step": 5412
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.9657570719718933,
      "learning_rate": 0.0005991832415470542,
      "loss": 2.5827,
      "step": 5413
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.7443044781684875,
      "learning_rate": 0.000599182939879618,
      "loss": 2.6461,
      "step": 5414
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.638194739818573,
      "learning_rate": 0.0005991826381565582,
      "loss": 2.5024,
      "step": 5415
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.6673139929771423,
      "learning_rate": 0.0005991823363778745,
      "loss": 2.6518,
      "step": 5416
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.6939891576766968,
      "learning_rate": 0.0005991820345435673,
      "loss": 2.406,
      "step": 5417
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.6574264764785767,
      "learning_rate": 0.0005991817326536365,
      "loss": 2.4255,
      "step": 5418
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.7388342022895813,
      "learning_rate": 0.0005991814307080821,
      "loss": 2.7581,
      "step": 5419
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.7156449556350708,
      "learning_rate": 0.0005991811287069042,
      "loss": 2.7166,
      "step": 5420
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.6642045378684998,
      "learning_rate": 0.000599180826650103,
      "loss": 2.5998,
      "step": 5421
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.7095226645469666,
      "learning_rate": 0.0005991805245376784,
      "loss": 2.5886,
      "step": 5422
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.774641752243042,
      "learning_rate": 0.0005991802223696305,
      "loss": 2.8982,
      "step": 5423
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.6520995497703552,
      "learning_rate": 0.0005991799201459593,
      "loss": 2.5193,
      "step": 5424
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.6896486282348633,
      "learning_rate": 0.000599179617866665,
      "loss": 2.5777,
      "step": 5425
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.8178496956825256,
      "learning_rate": 0.0005991793155317475,
      "loss": 2.4741,
      "step": 5426
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.6382037401199341,
      "learning_rate": 0.0005991790131412068,
      "loss": 2.6148,
      "step": 5427
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.860903263092041,
      "learning_rate": 0.0005991787106950432,
      "loss": 2.6442,
      "step": 5428
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.6504154801368713,
      "learning_rate": 0.0005991784081932567,
      "loss": 2.6063,
      "step": 5429
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.9033811688423157,
      "learning_rate": 0.0005991781056358472,
      "loss": 2.8358,
      "step": 5430
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.6542550325393677,
      "learning_rate": 0.0005991778030228148,
      "loss": 2.5178,
      "step": 5431
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.6996082067489624,
      "learning_rate": 0.0005991775003541597,
      "loss": 2.561,
      "step": 5432
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.7135111093521118,
      "learning_rate": 0.0005991771976298818,
      "loss": 2.6008,
      "step": 5433
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.7044799327850342,
      "learning_rate": 0.0005991768948499812,
      "loss": 2.5481,
      "step": 5434
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.6126641035079956,
      "learning_rate": 0.000599176592014458,
      "loss": 2.524,
      "step": 5435
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.6652101278305054,
      "learning_rate": 0.0005991762891233121,
      "loss": 2.3798,
      "step": 5436
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.6346352100372314,
      "learning_rate": 0.0005991759861765438,
      "loss": 2.5464,
      "step": 5437
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.6706254482269287,
      "learning_rate": 0.0005991756831741531,
      "loss": 2.8674,
      "step": 5438
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.6625499129295349,
      "learning_rate": 0.0005991753801161399,
      "loss": 2.674,
      "step": 5439
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.9600777626037598,
      "learning_rate": 0.0005991750770025043,
      "loss": 2.5655,
      "step": 5440
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.0388071537017822,
      "learning_rate": 0.0005991747738332465,
      "loss": 2.7448,
      "step": 5441
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.6993106007575989,
      "learning_rate": 0.0005991744706083664,
      "loss": 2.6034,
      "step": 5442
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.667364239692688,
      "learning_rate": 0.0005991741673278642,
      "loss": 2.7023,
      "step": 5443
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.7830653786659241,
      "learning_rate": 0.0005991738639917399,
      "loss": 2.7047,
      "step": 5444
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.7475875616073608,
      "learning_rate": 0.0005991735605999933,
      "loss": 2.5813,
      "step": 5445
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.8113340139389038,
      "learning_rate": 0.0005991732571526249,
      "loss": 2.3921,
      "step": 5446
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.6591033935546875,
      "learning_rate": 0.0005991729536496344,
      "loss": 2.4728,
      "step": 5447
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.6826450824737549,
      "learning_rate": 0.0005991726500910222,
      "loss": 2.6135,
      "step": 5448
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.6358556151390076,
      "learning_rate": 0.000599172346476788,
      "loss": 2.473,
      "step": 5449
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.7467703223228455,
      "learning_rate": 0.0005991720428069321,
      "loss": 2.9325,
      "step": 5450
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.6649808287620544,
      "learning_rate": 0.0005991717390814544,
      "loss": 2.598,
      "step": 5451
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.7031494975090027,
      "learning_rate": 0.000599171435300355,
      "loss": 2.8034,
      "step": 5452
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.7159711122512817,
      "learning_rate": 0.000599171131463634,
      "loss": 2.6156,
      "step": 5453
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.6091896891593933,
      "learning_rate": 0.0005991708275712915,
      "loss": 2.6065,
      "step": 5454
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.6949473023414612,
      "learning_rate": 0.0005991705236233276,
      "loss": 2.5036,
      "step": 5455
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.8708821535110474,
      "learning_rate": 0.0005991702196197421,
      "loss": 2.5238,
      "step": 5456
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.7345240116119385,
      "learning_rate": 0.0005991699155605352,
      "loss": 2.4506,
      "step": 5457
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.6738466024398804,
      "learning_rate": 0.000599169611445707,
      "loss": 2.5906,
      "step": 5458
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.6473127603530884,
      "learning_rate": 0.0005991693072752575,
      "loss": 2.429,
      "step": 5459
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.6802433133125305,
      "learning_rate": 0.0005991690030491869,
      "loss": 2.7591,
      "step": 5460
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.6408730149269104,
      "learning_rate": 0.000599168698767495,
      "loss": 2.6458,
      "step": 5461
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.7421847581863403,
      "learning_rate": 0.0005991683944301821,
      "loss": 2.4972,
      "step": 5462
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.691275954246521,
      "learning_rate": 0.000599168090037248,
      "loss": 2.5013,
      "step": 5463
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.6339531540870667,
      "learning_rate": 0.000599167785588693,
      "loss": 2.528,
      "step": 5464
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.7122384309768677,
      "learning_rate": 0.000599167481084517,
      "loss": 2.2654,
      "step": 5465
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.8318973183631897,
      "learning_rate": 0.0005991671765247201,
      "loss": 2.7734,
      "step": 5466
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.6725801229476929,
      "learning_rate": 0.0005991668719093025,
      "loss": 2.4865,
      "step": 5467
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.6840354204177856,
      "learning_rate": 0.000599166567238264,
      "loss": 2.3798,
      "step": 5468
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.8149449825286865,
      "learning_rate": 0.0005991662625116049,
      "loss": 2.5205,
      "step": 5469
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.6917353272438049,
      "learning_rate": 0.0005991659577293251,
      "loss": 2.7393,
      "step": 5470
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.7227776646614075,
      "learning_rate": 0.0005991656528914248,
      "loss": 2.4693,
      "step": 5471
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.9139142632484436,
      "learning_rate": 0.0005991653479979037,
      "loss": 2.5277,
      "step": 5472
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.7646340131759644,
      "learning_rate": 0.0005991650430487624,
      "loss": 2.4627,
      "step": 5473
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.8439133167266846,
      "learning_rate": 0.0005991647380440005,
      "loss": 2.7996,
      "step": 5474
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.7167437672615051,
      "learning_rate": 0.0005991644329836183,
      "loss": 2.704,
      "step": 5475
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.7054833769798279,
      "learning_rate": 0.0005991641278676158,
      "loss": 2.4217,
      "step": 5476
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.6886852383613586,
      "learning_rate": 0.0005991638226959929,
      "loss": 2.3702,
      "step": 5477
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.8954831957817078,
      "learning_rate": 0.0005991635174687499,
      "loss": 2.5156,
      "step": 5478
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.710559606552124,
      "learning_rate": 0.0005991632121858868,
      "loss": 2.5541,
      "step": 5479
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.8027352094650269,
      "learning_rate": 0.0005991629068474036,
      "loss": 2.5265,
      "step": 5480
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.9217096567153931,
      "learning_rate": 0.0005991626014533002,
      "loss": 2.7133,
      "step": 5481
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.6945326924324036,
      "learning_rate": 0.000599162296003577,
      "loss": 2.7945,
      "step": 5482
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.7302784323692322,
      "learning_rate": 0.0005991619904982339,
      "loss": 2.8018,
      "step": 5483
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.946992039680481,
      "learning_rate": 0.0005991616849372708,
      "loss": 2.4677,
      "step": 5484
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.6786638498306274,
      "learning_rate": 0.000599161379320688,
      "loss": 2.4427,
      "step": 5485
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.7040858864784241,
      "learning_rate": 0.0005991610736484855,
      "loss": 2.4842,
      "step": 5486
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.8763104677200317,
      "learning_rate": 0.0005991607679206632,
      "loss": 2.7089,
      "step": 5487
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.6899072527885437,
      "learning_rate": 0.0005991604621372213,
      "loss": 2.785,
      "step": 5488
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.6641207933425903,
      "learning_rate": 0.0005991601562981598,
      "loss": 2.9452,
      "step": 5489
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.1000921726226807,
      "learning_rate": 0.0005991598504034789,
      "loss": 2.3358,
      "step": 5490
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.7442582845687866,
      "learning_rate": 0.0005991595444531783,
      "loss": 2.7409,
      "step": 5491
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.6769076585769653,
      "learning_rate": 0.0005991592384472585,
      "loss": 2.5409,
      "step": 5492
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.7791173458099365,
      "learning_rate": 0.0005991589323857193,
      "loss": 2.6986,
      "step": 5493
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.7281378507614136,
      "learning_rate": 0.0005991586262685608,
      "loss": 2.6887,
      "step": 5494
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.6668990850448608,
      "learning_rate": 0.000599158320095783,
      "loss": 2.5535,
      "step": 5495
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.7864867448806763,
      "learning_rate": 0.0005991580138673861,
      "loss": 2.6321,
      "step": 5496
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.7863954901695251,
      "learning_rate": 0.00059915770758337,
      "loss": 2.6558,
      "step": 5497
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.7610852122306824,
      "learning_rate": 0.000599157401243735,
      "loss": 2.7323,
      "step": 5498
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.6501052975654602,
      "learning_rate": 0.0005991570948484809,
      "loss": 2.5033,
      "step": 5499
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.856528639793396,
      "learning_rate": 0.0005991567883976077,
      "loss": 2.6072,
      "step": 5500
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.6754344701766968,
      "learning_rate": 0.0005991564818911159,
      "loss": 2.8636,
      "step": 5501
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.8406482934951782,
      "learning_rate": 0.0005991561753290051,
      "loss": 2.6343,
      "step": 5502
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.6300111413002014,
      "learning_rate": 0.0005991558687112754,
      "loss": 2.7803,
      "step": 5503
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.7444325685501099,
      "learning_rate": 0.0005991555620379271,
      "loss": 2.6658,
      "step": 5504
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.7154331207275391,
      "learning_rate": 0.0005991552553089601,
      "loss": 2.6063,
      "step": 5505
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.7888429760932922,
      "learning_rate": 0.0005991549485243746,
      "loss": 2.535,
      "step": 5506
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.6602675914764404,
      "learning_rate": 0.0005991546416841705,
      "loss": 2.6189,
      "step": 5507
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.6807795763015747,
      "learning_rate": 0.0005991543347883478,
      "loss": 2.9202,
      "step": 5508
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.6358578205108643,
      "learning_rate": 0.0005991540278369068,
      "loss": 2.63,
      "step": 5509
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.7082454562187195,
      "learning_rate": 0.0005991537208298474,
      "loss": 2.8858,
      "step": 5510
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.7153276801109314,
      "learning_rate": 0.0005991534137671696,
      "loss": 2.5598,
      "step": 5511
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.6842774152755737,
      "learning_rate": 0.0005991531066488735,
      "loss": 2.4729,
      "step": 5512
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.8133180141448975,
      "learning_rate": 0.0005991527994749593,
      "loss": 2.6387,
      "step": 5513
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.6916728615760803,
      "learning_rate": 0.000599152492245427,
      "loss": 2.6072,
      "step": 5514
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.7129443287849426,
      "learning_rate": 0.0005991521849602764,
      "loss": 2.9017,
      "step": 5515
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.6782224178314209,
      "learning_rate": 0.0005991518776195078,
      "loss": 2.4902,
      "step": 5516
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.6952635049819946,
      "learning_rate": 0.0005991515702231213,
      "loss": 2.6389,
      "step": 5517
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.736625611782074,
      "learning_rate": 0.000599151262771117,
      "loss": 2.7895,
      "step": 5518
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.8221364617347717,
      "learning_rate": 0.0005991509552634947,
      "loss": 2.6634,
      "step": 5519
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.6602575778961182,
      "learning_rate": 0.0005991506477002545,
      "loss": 2.6628,
      "step": 5520
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.824340283870697,
      "learning_rate": 0.0005991503400813966,
      "loss": 2.6286,
      "step": 5521
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.7347072958946228,
      "learning_rate": 0.0005991500324069211,
      "loss": 2.9591,
      "step": 5522
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.8223885297775269,
      "learning_rate": 0.000599149724676828,
      "loss": 2.6963,
      "step": 5523
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.696238100528717,
      "learning_rate": 0.0005991494168911172,
      "loss": 2.5554,
      "step": 5524
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.6564121842384338,
      "learning_rate": 0.0005991491090497889,
      "loss": 2.6507,
      "step": 5525
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.6499052047729492,
      "learning_rate": 0.000599148801152843,
      "loss": 2.601,
      "step": 5526
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.6635452508926392,
      "learning_rate": 0.00059914849320028,
      "loss": 2.6959,
      "step": 5527
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.7104286551475525,
      "learning_rate": 0.0005991481851920994,
      "loss": 2.6685,
      "step": 5528
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.7077323198318481,
      "learning_rate": 0.0005991478771283016,
      "loss": 2.5599,
      "step": 5529
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.7425356507301331,
      "learning_rate": 0.0005991475690088865,
      "loss": 2.2785,
      "step": 5530
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.7515853047370911,
      "learning_rate": 0.0005991472608338544,
      "loss": 2.6174,
      "step": 5531
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.7188951969146729,
      "learning_rate": 0.0005991469526032049,
      "loss": 2.5905,
      "step": 5532
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.6795852184295654,
      "learning_rate": 0.0005991466443169385,
      "loss": 2.5986,
      "step": 5533
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.8769246935844421,
      "learning_rate": 0.000599146335975055,
      "loss": 2.6191,
      "step": 5534
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.6961473822593689,
      "learning_rate": 0.0005991460275775548,
      "loss": 2.4409,
      "step": 5535
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.7708170413970947,
      "learning_rate": 0.0005991457191244374,
      "loss": 2.6008,
      "step": 5536
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.6706063151359558,
      "learning_rate": 0.0005991454106157033,
      "loss": 2.6408,
      "step": 5537
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.6739041805267334,
      "learning_rate": 0.0005991451020513523,
      "loss": 2.5482,
      "step": 5538
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.6367249488830566,
      "learning_rate": 0.0005991447934313847,
      "loss": 2.5194,
      "step": 5539
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.7045905590057373,
      "learning_rate": 0.0005991444847558005,
      "loss": 2.6133,
      "step": 5540
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.7195774912834167,
      "learning_rate": 0.0005991441760245996,
      "loss": 3.0596,
      "step": 5541
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.6819867491722107,
      "learning_rate": 0.0005991438672377821,
      "loss": 2.6605,
      "step": 5542
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.7277317047119141,
      "learning_rate": 0.0005991435583953483,
      "loss": 2.7782,
      "step": 5543
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.6776175498962402,
      "learning_rate": 0.0005991432494972979,
      "loss": 2.7724,
      "step": 5544
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.7918900847434998,
      "learning_rate": 0.0005991429405436311,
      "loss": 2.7032,
      "step": 5545
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.7719610929489136,
      "learning_rate": 0.0005991426315343481,
      "loss": 2.6411,
      "step": 5546
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.7318316102027893,
      "learning_rate": 0.0005991423224694488,
      "loss": 2.6562,
      "step": 5547
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.8605237603187561,
      "learning_rate": 0.0005991420133489334,
      "loss": 2.692,
      "step": 5548
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.6201604604721069,
      "learning_rate": 0.0005991417041728016,
      "loss": 2.7358,
      "step": 5549
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.7913231253623962,
      "learning_rate": 0.0005991413949410539,
      "loss": 2.7904,
      "step": 5550
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.7043113708496094,
      "learning_rate": 0.0005991410856536901,
      "loss": 2.7502,
      "step": 5551
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.7094665169715881,
      "learning_rate": 0.0005991407763107103,
      "loss": 2.4771,
      "step": 5552
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.7239265441894531,
      "learning_rate": 0.0005991404669121146,
      "loss": 2.634,
      "step": 5553
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.9235151410102844,
      "learning_rate": 0.0005991401574579031,
      "loss": 2.6165,
      "step": 5554
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.7507534027099609,
      "learning_rate": 0.0005991398479480757,
      "loss": 2.86,
      "step": 5555
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.9125922918319702,
      "learning_rate": 0.0005991395383826327,
      "loss": 2.4754,
      "step": 5556
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.7154954075813293,
      "learning_rate": 0.000599139228761574,
      "loss": 2.3249,
      "step": 5557
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.7431111931800842,
      "learning_rate": 0.0005991389190848996,
      "loss": 2.8342,
      "step": 5558
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.7681417465209961,
      "learning_rate": 0.0005991386093526095,
      "loss": 2.7555,
      "step": 5559
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.6531360745429993,
      "learning_rate": 0.000599138299564704,
      "loss": 2.6608,
      "step": 5560
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.5946919918060303,
      "learning_rate": 0.0005991379897211831,
      "loss": 2.3216,
      "step": 5561
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.7653853297233582,
      "learning_rate": 0.0005991376798220468,
      "loss": 2.5458,
      "step": 5562
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.6400129199028015,
      "learning_rate": 0.0005991373698672951,
      "loss": 2.7839,
      "step": 5563
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.7170777320861816,
      "learning_rate": 0.0005991370598569281,
      "loss": 2.7106,
      "step": 5564
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.6971319317817688,
      "learning_rate": 0.0005991367497909459,
      "loss": 2.5358,
      "step": 5565
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.6350840330123901,
      "learning_rate": 0.0005991364396693486,
      "loss": 2.4387,
      "step": 5566
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.6557996273040771,
      "learning_rate": 0.0005991361294921361,
      "loss": 2.6386,
      "step": 5567
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.7375866770744324,
      "learning_rate": 0.0005991358192593085,
      "loss": 2.6927,
      "step": 5568
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.778647243976593,
      "learning_rate": 0.000599135508970866,
      "loss": 2.5482,
      "step": 5569
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.6716452240943909,
      "learning_rate": 0.0005991351986268086,
      "loss": 2.4402,
      "step": 5570
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.717484712600708,
      "learning_rate": 0.0005991348882271362,
      "loss": 2.6731,
      "step": 5571
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.6223625540733337,
      "learning_rate": 0.000599134577771849,
      "loss": 2.7615,
      "step": 5572
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.6468020081520081,
      "learning_rate": 0.0005991342672609471,
      "loss": 2.4327,
      "step": 5573
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.6189786195755005,
      "learning_rate": 0.0005991339566944304,
      "loss": 3.0089,
      "step": 5574
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.7241799831390381,
      "learning_rate": 0.0005991336460722992,
      "loss": 2.7234,
      "step": 5575
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.6315216422080994,
      "learning_rate": 0.0005991333353945533,
      "loss": 2.5215,
      "step": 5576
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.6949774622917175,
      "learning_rate": 0.000599133024661193,
      "loss": 2.8485,
      "step": 5577
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.588239848613739,
      "learning_rate": 0.000599132713872218,
      "loss": 2.4467,
      "step": 5578
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.65690678358078,
      "learning_rate": 0.0005991324030276287,
      "loss": 2.5998,
      "step": 5579
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.7224928140640259,
      "learning_rate": 0.0005991320921274251,
      "loss": 2.8183,
      "step": 5580
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.732718288898468,
      "learning_rate": 0.0005991317811716071,
      "loss": 2.6381,
      "step": 5581
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.7030702233314514,
      "learning_rate": 0.0005991314701601749,
      "loss": 2.6263,
      "step": 5582
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.7778303623199463,
      "learning_rate": 0.0005991311590931286,
      "loss": 2.554,
      "step": 5583
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.6776314377784729,
      "learning_rate": 0.000599130847970468,
      "loss": 2.7085,
      "step": 5584
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.6813953518867493,
      "learning_rate": 0.0005991305367921933,
      "loss": 2.6683,
      "step": 5585
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.7428507804870605,
      "learning_rate": 0.0005991302255583048,
      "loss": 2.8208,
      "step": 5586
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.7189213037490845,
      "learning_rate": 0.0005991299142688023,
      "loss": 2.731,
      "step": 5587
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.6443848013877869,
      "learning_rate": 0.0005991296029236857,
      "loss": 2.7109,
      "step": 5588
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.7038126587867737,
      "learning_rate": 0.0005991292915229555,
      "loss": 3.0786,
      "step": 5589
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.7559724450111389,
      "learning_rate": 0.0005991289800666114,
      "loss": 2.5866,
      "step": 5590
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.6857836246490479,
      "learning_rate": 0.0005991286685546536,
      "loss": 2.4367,
      "step": 5591
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.6903072595596313,
      "learning_rate": 0.000599128356987082,
      "loss": 2.6452,
      "step": 5592
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.5925906300544739,
      "learning_rate": 0.000599128045363897,
      "loss": 2.5112,
      "step": 5593
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.6479730010032654,
      "learning_rate": 0.0005991277336850983,
      "loss": 2.7668,
      "step": 5594
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.7205467224121094,
      "learning_rate": 0.0005991274219506862,
      "loss": 2.9059,
      "step": 5595
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.6755150556564331,
      "learning_rate": 0.0005991271101606607,
      "loss": 2.6662,
      "step": 5596
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.6509515643119812,
      "learning_rate": 0.0005991267983150218,
      "loss": 2.4981,
      "step": 5597
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.76788330078125,
      "learning_rate": 0.0005991264864137695,
      "loss": 2.8337,
      "step": 5598
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.6492149233818054,
      "learning_rate": 0.000599126174456904,
      "loss": 2.6622,
      "step": 5599
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.7739354372024536,
      "learning_rate": 0.0005991258624444253,
      "loss": 2.6283,
      "step": 5600
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.6810786724090576,
      "learning_rate": 0.0005991255503763334,
      "loss": 2.9118,
      "step": 5601
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.7011273503303528,
      "learning_rate": 0.0005991252382526284,
      "loss": 2.7165,
      "step": 5602
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.6671568751335144,
      "learning_rate": 0.0005991249260733105,
      "loss": 2.685,
      "step": 5603
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.7591974139213562,
      "learning_rate": 0.0005991246138383796,
      "loss": 2.6868,
      "step": 5604
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.6981967091560364,
      "learning_rate": 0.0005991243015478357,
      "loss": 2.4576,
      "step": 5605
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.6176232099533081,
      "learning_rate": 0.0005991239892016789,
      "loss": 2.9148,
      "step": 5606
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.7561455965042114,
      "learning_rate": 0.0005991236767999094,
      "loss": 2.4488,
      "step": 5607
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.7056254744529724,
      "learning_rate": 0.0005991233643425272,
      "loss": 2.5792,
      "step": 5608
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.6556686162948608,
      "learning_rate": 0.0005991230518295322,
      "loss": 2.6261,
      "step": 5609
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.7479124665260315,
      "learning_rate": 0.0005991227392609247,
      "loss": 2.6039,
      "step": 5610
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.7754007577896118,
      "learning_rate": 0.0005991224266367046,
      "loss": 2.3909,
      "step": 5611
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.6985698342323303,
      "learning_rate": 0.000599122113956872,
      "loss": 2.7263,
      "step": 5612
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.7370452880859375,
      "learning_rate": 0.000599121801221427,
      "loss": 2.7399,
      "step": 5613
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.6674344539642334,
      "learning_rate": 0.0005991214884303695,
      "loss": 2.5332,
      "step": 5614
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.7019598484039307,
      "learning_rate": 0.0005991211755836997,
      "loss": 2.7828,
      "step": 5615
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.6973536014556885,
      "learning_rate": 0.0005991208626814177,
      "loss": 2.506,
      "step": 5616
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.6799753308296204,
      "learning_rate": 0.0005991205497235234,
      "loss": 2.7538,
      "step": 5617
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.6591911315917969,
      "learning_rate": 0.000599120236710017,
      "loss": 2.6484,
      "step": 5618
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.6689466834068298,
      "learning_rate": 0.0005991199236408984,
      "loss": 2.5654,
      "step": 5619
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.7059046030044556,
      "learning_rate": 0.0005991196105161678,
      "loss": 2.3668,
      "step": 5620
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.7909847497940063,
      "learning_rate": 0.0005991192973358253,
      "loss": 2.4986,
      "step": 5621
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.6608689427375793,
      "learning_rate": 0.0005991189840998709,
      "loss": 2.7684,
      "step": 5622
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.8780388236045837,
      "learning_rate": 0.0005991186708083046,
      "loss": 2.7998,
      "step": 5623
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.615189254283905,
      "learning_rate": 0.0005991183574611263,
      "loss": 2.7131,
      "step": 5624
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.6902791857719421,
      "learning_rate": 0.0005991180440583365,
      "loss": 2.6781,
      "step": 5625
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.8100734353065491,
      "learning_rate": 0.0005991177305999349,
      "loss": 2.2369,
      "step": 5626
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.6774414777755737,
      "learning_rate": 0.0005991174170859217,
      "loss": 2.867,
      "step": 5627
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.8768316507339478,
      "learning_rate": 0.0005991171035162969,
      "loss": 2.4996,
      "step": 5628
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.7544248700141907,
      "learning_rate": 0.0005991167898910606,
      "loss": 2.6179,
      "step": 5629
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.8013450503349304,
      "learning_rate": 0.0005991164762102128,
      "loss": 2.6184,
      "step": 5630
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.9763953685760498,
      "learning_rate": 0.0005991161624737536,
      "loss": 2.4511,
      "step": 5631
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.6721451282501221,
      "learning_rate": 0.0005991158486816831,
      "loss": 2.5564,
      "step": 5632
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.8326495885848999,
      "learning_rate": 0.0005991155348340014,
      "loss": 2.7226,
      "step": 5633
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.6945099234580994,
      "learning_rate": 0.0005991152209307083,
      "loss": 2.4465,
      "step": 5634
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.7732552886009216,
      "learning_rate": 0.0005991149069718042,
      "loss": 2.6008,
      "step": 5635
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.6575278639793396,
      "learning_rate": 0.0005991145929572888,
      "loss": 2.5558,
      "step": 5636
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.7247915863990784,
      "learning_rate": 0.0005991142788871626,
      "loss": 2.4695,
      "step": 5637
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.76325923204422,
      "learning_rate": 0.0005991139647614252,
      "loss": 2.5771,
      "step": 5638
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.6582775712013245,
      "learning_rate": 0.000599113650580077,
      "loss": 2.3517,
      "step": 5639
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.7259117960929871,
      "learning_rate": 0.0005991133363431178,
      "loss": 2.746,
      "step": 5640
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.6183555126190186,
      "learning_rate": 0.0005991130220505479,
      "loss": 2.3354,
      "step": 5641
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.8338082432746887,
      "learning_rate": 0.0005991127077023673,
      "loss": 2.654,
      "step": 5642
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.658885657787323,
      "learning_rate": 0.0005991123932985758,
      "loss": 2.6911,
      "step": 5643
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.8110263347625732,
      "learning_rate": 0.0005991120788391737,
      "loss": 2.5885,
      "step": 5644
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.7269749641418457,
      "learning_rate": 0.0005991117643241612,
      "loss": 2.5429,
      "step": 5645
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.674565315246582,
      "learning_rate": 0.000599111449753538,
      "loss": 2.659,
      "step": 5646
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.6435373425483704,
      "learning_rate": 0.0005991111351273044,
      "loss": 2.6577,
      "step": 5647
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.7406240105628967,
      "learning_rate": 0.0005991108204454603,
      "loss": 2.4126,
      "step": 5648
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.641852855682373,
      "learning_rate": 0.000599110505708006,
      "loss": 2.5773,
      "step": 5649
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.7320669293403625,
      "learning_rate": 0.0005991101909149413,
      "loss": 2.8827,
      "step": 5650
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.685021698474884,
      "learning_rate": 0.0005991098760662664,
      "loss": 2.4412,
      "step": 5651
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.7534617185592651,
      "learning_rate": 0.0005991095611619814,
      "loss": 2.5533,
      "step": 5652
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.6381696462631226,
      "learning_rate": 0.0005991092462020862,
      "loss": 2.6563,
      "step": 5653
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.7379310131072998,
      "learning_rate": 0.000599108931186581,
      "loss": 2.6001,
      "step": 5654
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.6492798924446106,
      "learning_rate": 0.0005991086161154658,
      "loss": 2.5712,
      "step": 5655
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.6313210129737854,
      "learning_rate": 0.0005991083009887407,
      "loss": 2.6758,
      "step": 5656
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.6855520009994507,
      "learning_rate": 0.0005991079858064056,
      "loss": 2.3698,
      "step": 5657
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.5840858221054077,
      "learning_rate": 0.0005991076705684608,
      "loss": 2.3229,
      "step": 5658
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.6899279356002808,
      "learning_rate": 0.0005991073552749062,
      "loss": 2.606,
      "step": 5659
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.6751327514648438,
      "learning_rate": 0.0005991070399257417,
      "loss": 2.6466,
      "step": 5660
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.6148250102996826,
      "learning_rate": 0.0005991067245209678,
      "loss": 2.1495,
      "step": 5661
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.7158949375152588,
      "learning_rate": 0.0005991064090605843,
      "loss": 2.606,
      "step": 5662
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.7012340426445007,
      "learning_rate": 0.0005991060935445912,
      "loss": 2.5781,
      "step": 5663
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.858135461807251,
      "learning_rate": 0.0005991057779729886,
      "loss": 2.8889,
      "step": 5664
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.6758764386177063,
      "learning_rate": 0.0005991054623457766,
      "loss": 2.3013,
      "step": 5665
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.667312741279602,
      "learning_rate": 0.0005991051466629553,
      "loss": 2.2564,
      "step": 5666
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.6345599889755249,
      "learning_rate": 0.0005991048309245248,
      "loss": 2.7269,
      "step": 5667
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.671580970287323,
      "learning_rate": 0.0005991045151304849,
      "loss": 2.7916,
      "step": 5668
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.777111291885376,
      "learning_rate": 0.0005991041992808359,
      "loss": 3.063,
      "step": 5669
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.6364699006080627,
      "learning_rate": 0.0005991038833755778,
      "loss": 2.4129,
      "step": 5670
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.6825060844421387,
      "learning_rate": 0.0005991035674147106,
      "loss": 2.6138,
      "step": 5671
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.721496045589447,
      "learning_rate": 0.0005991032513982344,
      "loss": 2.6033,
      "step": 5672
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.6901952624320984,
      "learning_rate": 0.0005991029353261492,
      "loss": 2.4475,
      "step": 5673
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.7206736207008362,
      "learning_rate": 0.0005991026191984552,
      "loss": 2.5062,
      "step": 5674
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.8284646272659302,
      "learning_rate": 0.0005991023030151524,
      "loss": 2.2424,
      "step": 5675
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.8109368085861206,
      "learning_rate": 0.0005991019867762408,
      "loss": 2.5548,
      "step": 5676
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.7837174534797668,
      "learning_rate": 0.0005991016704817205,
      "loss": 2.7217,
      "step": 5677
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.6891933083534241,
      "learning_rate": 0.0005991013541315916,
      "loss": 2.781,
      "step": 5678
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.8175359964370728,
      "learning_rate": 0.000599101037725854,
      "loss": 2.5635,
      "step": 5679
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.6989244818687439,
      "learning_rate": 0.000599100721264508,
      "loss": 2.5756,
      "step": 5680
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.6998776197433472,
      "learning_rate": 0.0005991004047475534,
      "loss": 2.5926,
      "step": 5681
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.8903957605361938,
      "learning_rate": 0.0005991000881749904,
      "loss": 2.4182,
      "step": 5682
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.6570643782615662,
      "learning_rate": 0.0005990997715468191,
      "loss": 2.6736,
      "step": 5683
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.6739201545715332,
      "learning_rate": 0.0005990994548630396,
      "loss": 2.6474,
      "step": 5684
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.7560742497444153,
      "learning_rate": 0.0005990991381236518,
      "loss": 2.5968,
      "step": 5685
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.6142129898071289,
      "learning_rate": 0.0005990988213286558,
      "loss": 2.7384,
      "step": 5686
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.772014319896698,
      "learning_rate": 0.0005990985044780516,
      "loss": 2.5975,
      "step": 5687
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.7013233304023743,
      "learning_rate": 0.0005990981875718394,
      "loss": 2.6925,
      "step": 5688
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.8547606468200684,
      "learning_rate": 0.0005990978706100193,
      "loss": 2.5149,
      "step": 5689
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.6977507472038269,
      "learning_rate": 0.0005990975535925911,
      "loss": 2.519,
      "step": 5690
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.9705190658569336,
      "learning_rate": 0.0005990972365195552,
      "loss": 2.515,
      "step": 5691
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.7049233913421631,
      "learning_rate": 0.0005990969193909113,
      "loss": 2.5065,
      "step": 5692
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.7140249013900757,
      "learning_rate": 0.0005990966022066598,
      "loss": 2.6379,
      "step": 5693
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.7335315942764282,
      "learning_rate": 0.0005990962849668004,
      "loss": 2.3636,
      "step": 5694
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.7055514454841614,
      "learning_rate": 0.0005990959676713335,
      "loss": 2.7167,
      "step": 5695
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.653519332408905,
      "learning_rate": 0.000599095650320259,
      "loss": 2.3437,
      "step": 5696
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.7784969806671143,
      "learning_rate": 0.000599095332913577,
      "loss": 2.3814,
      "step": 5697
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.6982289552688599,
      "learning_rate": 0.0005990950154512875,
      "loss": 2.3336,
      "step": 5698
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.7302451133728027,
      "learning_rate": 0.0005990946979333904,
      "loss": 2.5641,
      "step": 5699
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.736967921257019,
      "learning_rate": 0.0005990943803598863,
      "loss": 2.5222,
      "step": 5700
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.7358262538909912,
      "learning_rate": 0.0005990940627307747,
      "loss": 2.6601,
      "step": 5701
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.8098028302192688,
      "learning_rate": 0.0005990937450460559,
      "loss": 2.2125,
      "step": 5702
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.6395914554595947,
      "learning_rate": 0.0005990934273057298,
      "loss": 2.4341,
      "step": 5703
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.7459745407104492,
      "learning_rate": 0.0005990931095097968,
      "loss": 2.8496,
      "step": 5704
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.7940605282783508,
      "learning_rate": 0.0005990927916582567,
      "loss": 2.5014,
      "step": 5705
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.7146903276443481,
      "learning_rate": 0.0005990924737511095,
      "loss": 2.6913,
      "step": 5706
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.7086785435676575,
      "learning_rate": 0.0005990921557883554,
      "loss": 2.6899,
      "step": 5707
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.8049328327178955,
      "learning_rate": 0.0005990918377699945,
      "loss": 2.413,
      "step": 5708
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.7808650135993958,
      "learning_rate": 0.0005990915196960267,
      "loss": 2.5624,
      "step": 5709
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.2017580270767212,
      "learning_rate": 0.0005990912015664522,
      "loss": 2.6264,
      "step": 5710
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.6735104918479919,
      "learning_rate": 0.000599090883381271,
      "loss": 2.6125,
      "step": 5711
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.7082388997077942,
      "learning_rate": 0.0005990905651404832,
      "loss": 2.5559,
      "step": 5712
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.7096614241600037,
      "learning_rate": 0.0005990902468440886,
      "loss": 2.4646,
      "step": 5713
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.7477995157241821,
      "learning_rate": 0.0005990899284920877,
      "loss": 2.6298,
      "step": 5714
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.7513120770454407,
      "learning_rate": 0.0005990896100844802,
      "loss": 2.6633,
      "step": 5715
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.755954921245575,
      "learning_rate": 0.0005990892916212664,
      "loss": 2.5728,
      "step": 5716
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.7233349680900574,
      "learning_rate": 0.0005990889731024462,
      "loss": 2.6457,
      "step": 5717
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.7838301658630371,
      "learning_rate": 0.0005990886545280198,
      "loss": 2.5082,
      "step": 5718
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.6429664492607117,
      "learning_rate": 0.0005990883358979871,
      "loss": 2.5,
      "step": 5719
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.6419212818145752,
      "learning_rate": 0.0005990880172123482,
      "loss": 2.5209,
      "step": 5720
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.7617743611335754,
      "learning_rate": 0.0005990876984711032,
      "loss": 2.5693,
      "step": 5721
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.7211013436317444,
      "learning_rate": 0.0005990873796742522,
      "loss": 2.9121,
      "step": 5722
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.6981884241104126,
      "learning_rate": 0.0005990870608217952,
      "loss": 2.5568,
      "step": 5723
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.7416951060295105,
      "learning_rate": 0.0005990867419137322,
      "loss": 2.5069,
      "step": 5724
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.6887733340263367,
      "learning_rate": 0.0005990864229500635,
      "loss": 2.5218,
      "step": 5725
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.6982600688934326,
      "learning_rate": 0.0005990861039307888,
      "loss": 2.4469,
      "step": 5726
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.7059308886528015,
      "learning_rate": 0.0005990857848559084,
      "loss": 2.3969,
      "step": 5727
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.6822922229766846,
      "learning_rate": 0.0005990854657254223,
      "loss": 2.7604,
      "step": 5728
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.6427211761474609,
      "learning_rate": 0.0005990851465393307,
      "loss": 2.4148,
      "step": 5729
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.7008519172668457,
      "learning_rate": 0.0005990848272976333,
      "loss": 2.4509,
      "step": 5730
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.7018800377845764,
      "learning_rate": 0.0005990845080003305,
      "loss": 2.4313,
      "step": 5731
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.7137655019760132,
      "learning_rate": 0.0005990841886474223,
      "loss": 2.4866,
      "step": 5732
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.6414592862129211,
      "learning_rate": 0.0005990838692389086,
      "loss": 2.6606,
      "step": 5733
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.6845566630363464,
      "learning_rate": 0.0005990835497747896,
      "loss": 2.4585,
      "step": 5734
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.6855237483978271,
      "learning_rate": 0.0005990832302550654,
      "loss": 2.6353,
      "step": 5735
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.6307904124259949,
      "learning_rate": 0.000599082910679736,
      "loss": 2.4282,
      "step": 5736
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.7418031692504883,
      "learning_rate": 0.0005990825910488013,
      "loss": 2.6324,
      "step": 5737
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.623188316822052,
      "learning_rate": 0.0005990822713622616,
      "loss": 2.3555,
      "step": 5738
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.6534739136695862,
      "learning_rate": 0.0005990819516201168,
      "loss": 2.4222,
      "step": 5739
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.7344323992729187,
      "learning_rate": 0.000599081631822367,
      "loss": 2.6364,
      "step": 5740
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.6841187477111816,
      "learning_rate": 0.0005990813119690124,
      "loss": 2.4314,
      "step": 5741
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.6977101564407349,
      "learning_rate": 0.0005990809920600528,
      "loss": 2.5441,
      "step": 5742
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.8496911525726318,
      "learning_rate": 0.0005990806720954885,
      "loss": 2.5978,
      "step": 5743
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.7065708637237549,
      "learning_rate": 0.0005990803520753193,
      "loss": 2.6115,
      "step": 5744
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.7233691215515137,
      "learning_rate": 0.0005990800319995456,
      "loss": 2.4934,
      "step": 5745
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.7139081954956055,
      "learning_rate": 0.0005990797118681673,
      "loss": 2.6244,
      "step": 5746
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.6016688346862793,
      "learning_rate": 0.0005990793916811843,
      "loss": 2.505,
      "step": 5747
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.6545288562774658,
      "learning_rate": 0.0005990790714385968,
      "loss": 2.5074,
      "step": 5748
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.6981437802314758,
      "learning_rate": 0.0005990787511404049,
      "loss": 2.6828,
      "step": 5749
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.6559470295906067,
      "learning_rate": 0.0005990784307866086,
      "loss": 2.6679,
      "step": 5750
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.656649112701416,
      "learning_rate": 0.000599078110377208,
      "loss": 2.6201,
      "step": 5751
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.7124261260032654,
      "learning_rate": 0.0005990777899122032,
      "loss": 2.8736,
      "step": 5752
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.6598315238952637,
      "learning_rate": 0.000599077469391594,
      "loss": 2.4227,
      "step": 5753
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.6891669034957886,
      "learning_rate": 0.0005990771488153808,
      "loss": 2.7017,
      "step": 5754
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.6534785032272339,
      "learning_rate": 0.0005990768281835634,
      "loss": 2.755,
      "step": 5755
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.6629461646080017,
      "learning_rate": 0.0005990765074961422,
      "loss": 2.3878,
      "step": 5756
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.7130254507064819,
      "learning_rate": 0.0005990761867531168,
      "loss": 2.6839,
      "step": 5757
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.821173369884491,
      "learning_rate": 0.0005990758659544876,
      "loss": 2.6079,
      "step": 5758
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.8754488825798035,
      "learning_rate": 0.0005990755451002545,
      "loss": 2.7383,
      "step": 5759
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.8599565625190735,
      "learning_rate": 0.0005990752241904177,
      "loss": 2.5049,
      "step": 5760
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.7747893929481506,
      "learning_rate": 0.000599074903224977,
      "loss": 2.7567,
      "step": 5761
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.732187032699585,
      "learning_rate": 0.0005990745822039329,
      "loss": 2.8692,
      "step": 5762
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.1053757667541504,
      "learning_rate": 0.0005990742611272848,
      "loss": 2.4121,
      "step": 5763
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.7028875946998596,
      "learning_rate": 0.0005990739399950335,
      "loss": 2.672,
      "step": 5764
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.8794139623641968,
      "learning_rate": 0.0005990736188071786,
      "loss": 2.3712,
      "step": 5765
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.136855959892273,
      "learning_rate": 0.0005990732975637202,
      "loss": 2.7559,
      "step": 5766
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.6303731203079224,
      "learning_rate": 0.0005990729762646585,
      "loss": 2.6101,
      "step": 5767
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.9733497500419617,
      "learning_rate": 0.0005990726549099934,
      "loss": 2.7307,
      "step": 5768
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.9521065354347229,
      "learning_rate": 0.0005990723334997252,
      "loss": 2.6141,
      "step": 5769
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.6651690006256104,
      "learning_rate": 0.0005990720120338536,
      "loss": 2.8076,
      "step": 5770
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.8209234476089478,
      "learning_rate": 0.0005990716905123791,
      "loss": 2.6032,
      "step": 5771
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.8383327126502991,
      "learning_rate": 0.0005990713689353013,
      "loss": 2.7498,
      "step": 5772
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.7101438641548157,
      "learning_rate": 0.0005990710473026208,
      "loss": 2.4488,
      "step": 5773
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.8039918541908264,
      "learning_rate": 0.0005990707256143371,
      "loss": 2.7171,
      "step": 5774
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.7590160965919495,
      "learning_rate": 0.0005990704038704505,
      "loss": 2.4555,
      "step": 5775
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.7627845406532288,
      "learning_rate": 0.0005990700820709612,
      "loss": 2.5213,
      "step": 5776
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.706802248954773,
      "learning_rate": 0.000599069760215869,
      "loss": 2.4684,
      "step": 5777
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.160093069076538,
      "learning_rate": 0.0005990694383051741,
      "loss": 2.6013,
      "step": 5778
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.7229132056236267,
      "learning_rate": 0.0005990691163388766,
      "loss": 2.6829,
      "step": 5779
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.7583427429199219,
      "learning_rate": 0.0005990687943169765,
      "loss": 2.5993,
      "step": 5780
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.7780142426490784,
      "learning_rate": 0.0005990684722394739,
      "loss": 2.6894,
      "step": 5781
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.7482702732086182,
      "learning_rate": 0.0005990681501063688,
      "loss": 2.819,
      "step": 5782
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.653213620185852,
      "learning_rate": 0.0005990678279176612,
      "loss": 2.6188,
      "step": 5783
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.723652720451355,
      "learning_rate": 0.0005990675056733513,
      "loss": 2.6458,
      "step": 5784
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.7758177518844604,
      "learning_rate": 0.0005990671833734392,
      "loss": 2.5704,
      "step": 5785
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.6714454293251038,
      "learning_rate": 0.0005990668610179247,
      "loss": 2.6882,
      "step": 5786
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.7761006951332092,
      "learning_rate": 0.0005990665386068082,
      "loss": 2.5195,
      "step": 5787
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.7510614395141602,
      "learning_rate": 0.0005990662161400894,
      "loss": 2.8605,
      "step": 5788
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.7624061107635498,
      "learning_rate": 0.0005990658936177686,
      "loss": 2.5619,
      "step": 5789
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.7403373718261719,
      "learning_rate": 0.0005990655710398459,
      "loss": 2.4844,
      "step": 5790
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.6546689867973328,
      "learning_rate": 0.0005990652484063212,
      "loss": 2.5613,
      "step": 5791
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.664785623550415,
      "learning_rate": 0.0005990649257171947,
      "loss": 3.0147,
      "step": 5792
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.6795616149902344,
      "learning_rate": 0.0005990646029724663,
      "loss": 2.543,
      "step": 5793
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.6748237013816833,
      "learning_rate": 0.0005990642801721361,
      "loss": 2.3861,
      "step": 5794
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.6796736121177673,
      "learning_rate": 0.0005990639573162043,
      "loss": 2.5664,
      "step": 5795
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.8466955423355103,
      "learning_rate": 0.0005990636344046709,
      "loss": 2.6675,
      "step": 5796
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.6815329790115356,
      "learning_rate": 0.0005990633114375359,
      "loss": 2.5763,
      "step": 5797
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.7077344655990601,
      "learning_rate": 0.0005990629884147993,
      "loss": 2.5715,
      "step": 5798
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.7093895673751831,
      "learning_rate": 0.0005990626653364612,
      "loss": 2.5596,
      "step": 5799
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.6814539432525635,
      "learning_rate": 0.0005990623422025219,
      "loss": 2.6769,
      "step": 5800
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.7424226999282837,
      "learning_rate": 0.0005990620190129811,
      "loss": 2.5883,
      "step": 5801
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.9113209843635559,
      "learning_rate": 0.0005990616957678391,
      "loss": 2.8525,
      "step": 5802
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.6584895253181458,
      "learning_rate": 0.0005990613724670959,
      "loss": 2.5254,
      "step": 5803
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.6792743802070618,
      "learning_rate": 0.0005990610491107515,
      "loss": 2.709,
      "step": 5804
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.8295617699623108,
      "learning_rate": 0.0005990607256988061,
      "loss": 2.6201,
      "step": 5805
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.7152619361877441,
      "learning_rate": 0.0005990604022312595,
      "loss": 2.6386,
      "step": 5806
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.8996104598045349,
      "learning_rate": 0.000599060078708112,
      "loss": 2.336,
      "step": 5807
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.6991797089576721,
      "learning_rate": 0.0005990597551293636,
      "loss": 2.6535,
      "step": 5808
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.0536949634552002,
      "learning_rate": 0.0005990594314950143,
      "loss": 2.6966,
      "step": 5809
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.6455767154693604,
      "learning_rate": 0.0005990591078050642,
      "loss": 2.5003,
      "step": 5810
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.7090990543365479,
      "learning_rate": 0.0005990587840595135,
      "loss": 2.9092,
      "step": 5811
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.8147693872451782,
      "learning_rate": 0.0005990584602583621,
      "loss": 2.7207,
      "step": 5812
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.687101423740387,
      "learning_rate": 0.00059905813640161,
      "loss": 2.7099,
      "step": 5813
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.6728382110595703,
      "learning_rate": 0.0005990578124892574,
      "loss": 2.7005,
      "step": 5814
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.7826700806617737,
      "learning_rate": 0.0005990574885213042,
      "loss": 2.8321,
      "step": 5815
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.6814005970954895,
      "learning_rate": 0.0005990571644977508,
      "loss": 2.5828,
      "step": 5816
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.6906992197036743,
      "learning_rate": 0.0005990568404185968,
      "loss": 2.4812,
      "step": 5817
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.7331451773643494,
      "learning_rate": 0.0005990565162838425,
      "loss": 2.5384,
      "step": 5818
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.6956340074539185,
      "learning_rate": 0.0005990561920934881,
      "loss": 2.7062,
      "step": 5819
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.6916901469230652,
      "learning_rate": 0.0005990558678475335,
      "loss": 2.3093,
      "step": 5820
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.8817009925842285,
      "learning_rate": 0.0005990555435459787,
      "loss": 2.7508,
      "step": 5821
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.6904696226119995,
      "learning_rate": 0.0005990552191888238,
      "loss": 2.6291,
      "step": 5822
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.6331857442855835,
      "learning_rate": 0.0005990548947760689,
      "loss": 2.4387,
      "step": 5823
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.675870418548584,
      "learning_rate": 0.000599054570307714,
      "loss": 2.7656,
      "step": 5824
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.775046169757843,
      "learning_rate": 0.0005990542457837594,
      "loss": 2.5796,
      "step": 5825
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.9003010988235474,
      "learning_rate": 0.0005990539212042048,
      "loss": 2.6458,
      "step": 5826
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.6253405213356018,
      "learning_rate": 0.0005990535965690505,
      "loss": 2.5935,
      "step": 5827
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.8423277735710144,
      "learning_rate": 0.0005990532718782965,
      "loss": 2.2804,
      "step": 5828
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.6715916991233826,
      "learning_rate": 0.0005990529471319427,
      "loss": 2.436,
      "step": 5829
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.6861239075660706,
      "learning_rate": 0.0005990526223299895,
      "loss": 2.6168,
      "step": 5830
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.8793713450431824,
      "learning_rate": 0.0005990522974724366,
      "loss": 2.6883,
      "step": 5831
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.6945399641990662,
      "learning_rate": 0.0005990519725592845,
      "loss": 2.5648,
      "step": 5832
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.6776385307312012,
      "learning_rate": 0.0005990516475905328,
      "loss": 2.6508,
      "step": 5833
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.940932035446167,
      "learning_rate": 0.0005990513225661817,
      "loss": 2.6387,
      "step": 5834
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.7674698829650879,
      "learning_rate": 0.0005990509974862315,
      "loss": 2.6834,
      "step": 5835
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.7360518574714661,
      "learning_rate": 0.0005990506723506819,
      "loss": 2.7513,
      "step": 5836
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.7252227067947388,
      "learning_rate": 0.0005990503471595333,
      "loss": 2.7498,
      "step": 5837
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.885213315486908,
      "learning_rate": 0.0005990500219127854,
      "loss": 2.7189,
      "step": 5838
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.6238514184951782,
      "learning_rate": 0.0005990496966104386,
      "loss": 2.5674,
      "step": 5839
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.8529911637306213,
      "learning_rate": 0.0005990493712524928,
      "loss": 2.6343,
      "step": 5840
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.9246418476104736,
      "learning_rate": 0.0005990490458389481,
      "loss": 2.5474,
      "step": 5841
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.6327809691429138,
      "learning_rate": 0.0005990487203698044,
      "loss": 2.7023,
      "step": 5842
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.893375039100647,
      "learning_rate": 0.0005990483948450621,
      "loss": 2.5408,
      "step": 5843
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.9938632845878601,
      "learning_rate": 0.0005990480692647208,
      "loss": 2.5176,
      "step": 5844
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.6648812294006348,
      "learning_rate": 0.0005990477436287811,
      "loss": 2.5733,
      "step": 5845
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.7739120721817017,
      "learning_rate": 0.0005990474179372426,
      "loss": 2.5589,
      "step": 5846
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.9522550106048584,
      "learning_rate": 0.0005990470921901056,
      "loss": 2.6511,
      "step": 5847
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.6740580201148987,
      "learning_rate": 0.00059904676638737,
      "loss": 2.5635,
      "step": 5848
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.7366714477539062,
      "learning_rate": 0.0005990464405290361,
      "loss": 2.6269,
      "step": 5849
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.8262019753456116,
      "learning_rate": 0.0005990461146151038,
      "loss": 2.5893,
      "step": 5850
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.7623817324638367,
      "learning_rate": 0.0005990457886455731,
      "loss": 2.6471,
      "step": 5851
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.6983287334442139,
      "learning_rate": 0.0005990454626204442,
      "loss": 2.6518,
      "step": 5852
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.7200136184692383,
      "learning_rate": 0.0005990451365397171,
      "loss": 2.6944,
      "step": 5853
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.7023784518241882,
      "learning_rate": 0.0005990448104033919,
      "loss": 2.7497,
      "step": 5854
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.6388518214225769,
      "learning_rate": 0.0005990444842114686,
      "loss": 2.6204,
      "step": 5855
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.6586313247680664,
      "learning_rate": 0.0005990441579639472,
      "loss": 2.8622,
      "step": 5856
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.6935185790061951,
      "learning_rate": 0.000599043831660828,
      "loss": 2.6305,
      "step": 5857
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.6897619962692261,
      "learning_rate": 0.0005990435053021109,
      "loss": 2.4656,
      "step": 5858
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.665859580039978,
      "learning_rate": 0.0005990431788877958,
      "loss": 2.6596,
      "step": 5859
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.6335805654525757,
      "learning_rate": 0.0005990428524178831,
      "loss": 2.645,
      "step": 5860
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.6578793525695801,
      "learning_rate": 0.0005990425258923726,
      "loss": 2.5999,
      "step": 5861
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.6757701635360718,
      "learning_rate": 0.0005990421993112644,
      "loss": 2.6121,
      "step": 5862
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.8145872950553894,
      "learning_rate": 0.0005990418726745587,
      "loss": 2.7424,
      "step": 5863
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.7292375564575195,
      "learning_rate": 0.0005990415459822554,
      "loss": 2.5395,
      "step": 5864
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.7419881224632263,
      "learning_rate": 0.0005990412192343548,
      "loss": 2.7219,
      "step": 5865
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.6824162006378174,
      "learning_rate": 0.0005990408924308565,
      "loss": 2.5798,
      "step": 5866
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.6716104745864868,
      "learning_rate": 0.0005990405655717611,
      "loss": 2.4518,
      "step": 5867
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.9517537951469421,
      "learning_rate": 0.0005990402386570683,
      "loss": 2.3888,
      "step": 5868
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.7248010039329529,
      "learning_rate": 0.0005990399116867784,
      "loss": 2.6761,
      "step": 5869
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.6839510798454285,
      "learning_rate": 0.0005990395846608912,
      "loss": 2.5718,
      "step": 5870
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.7981764674186707,
      "learning_rate": 0.0005990392575794069,
      "loss": 2.865,
      "step": 5871
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.6274316906929016,
      "learning_rate": 0.0005990389304423256,
      "loss": 2.3108,
      "step": 5872
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.8659393191337585,
      "learning_rate": 0.0005990386032496474,
      "loss": 2.5972,
      "step": 5873
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.8898190259933472,
      "learning_rate": 0.0005990382760013721,
      "loss": 2.4894,
      "step": 5874
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.6979283690452576,
      "learning_rate": 0.0005990379486975001,
      "loss": 2.6449,
      "step": 5875
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.6893685460090637,
      "learning_rate": 0.0005990376213380312,
      "loss": 2.5985,
      "step": 5876
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.7195900678634644,
      "learning_rate": 0.0005990372939229656,
      "loss": 2.5889,
      "step": 5877
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.6900420188903809,
      "learning_rate": 0.0005990369664523033,
      "loss": 2.6643,
      "step": 5878
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.6323012709617615,
      "learning_rate": 0.0005990366389260444,
      "loss": 2.7098,
      "step": 5879
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.6630895733833313,
      "learning_rate": 0.000599036311344189,
      "loss": 2.5074,
      "step": 5880
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.7421057224273682,
      "learning_rate": 0.000599035983706737,
      "loss": 2.593,
      "step": 5881
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.7802289128303528,
      "learning_rate": 0.0005990356560136886,
      "loss": 2.745,
      "step": 5882
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.7360767126083374,
      "learning_rate": 0.0005990353282650439,
      "loss": 2.525,
      "step": 5883
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.6744344830513,
      "learning_rate": 0.0005990350004608028,
      "loss": 2.9285,
      "step": 5884
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.6898690462112427,
      "learning_rate": 0.0005990346726009655,
      "loss": 2.632,
      "step": 5885
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.6704944372177124,
      "learning_rate": 0.000599034344685532,
      "loss": 2.6712,
      "step": 5886
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.6844542026519775,
      "learning_rate": 0.0005990340167145023,
      "loss": 2.6146,
      "step": 5887
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.836294949054718,
      "learning_rate": 0.0005990336886878766,
      "loss": 2.4024,
      "step": 5888
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.813580334186554,
      "learning_rate": 0.0005990333606056549,
      "loss": 2.368,
      "step": 5889
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.7115992903709412,
      "learning_rate": 0.0005990330324678372,
      "loss": 2.5393,
      "step": 5890
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.7816085815429688,
      "learning_rate": 0.0005990327042744237,
      "loss": 2.5171,
      "step": 5891
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.6724585890769958,
      "learning_rate": 0.0005990323760254144,
      "loss": 2.488,
      "step": 5892
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.641860842704773,
      "learning_rate": 0.0005990320477208092,
      "loss": 2.4781,
      "step": 5893
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.8195255994796753,
      "learning_rate": 0.0005990317193606084,
      "loss": 2.7616,
      "step": 5894
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.6783080697059631,
      "learning_rate": 0.0005990313909448119,
      "loss": 2.7183,
      "step": 5895
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.6122954487800598,
      "learning_rate": 0.0005990310624734199,
      "loss": 2.5599,
      "step": 5896
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.6660473942756653,
      "learning_rate": 0.0005990307339464322,
      "loss": 2.6032,
      "step": 5897
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.6083094477653503,
      "learning_rate": 0.0005990304053638493,
      "loss": 2.6316,
      "step": 5898
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.6425362229347229,
      "learning_rate": 0.0005990300767256708,
      "loss": 2.3723,
      "step": 5899
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.733905017375946,
      "learning_rate": 0.0005990297480318971,
      "loss": 2.7454,
      "step": 5900
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.6639323830604553,
      "learning_rate": 0.000599029419282528,
      "loss": 2.2128,
      "step": 5901
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.7162458896636963,
      "learning_rate": 0.0005990290904775636,
      "loss": 2.4998,
      "step": 5902
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.7439693212509155,
      "learning_rate": 0.0005990287616170043,
      "loss": 2.8673,
      "step": 5903
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.7409600019454956,
      "learning_rate": 0.0005990284327008498,
      "loss": 2.5399,
      "step": 5904
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.6780275106430054,
      "learning_rate": 0.0005990281037291003,
      "loss": 2.6562,
      "step": 5905
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.81875079870224,
      "learning_rate": 0.0005990277747017558,
      "loss": 2.5046,
      "step": 5906
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.7274858951568604,
      "learning_rate": 0.0005990274456188164,
      "loss": 2.4538,
      "step": 5907
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.6954555511474609,
      "learning_rate": 0.0005990271164802821,
      "loss": 2.7276,
      "step": 5908
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.657224178314209,
      "learning_rate": 0.000599026787286153,
      "loss": 2.5206,
      "step": 5909
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.7035840153694153,
      "learning_rate": 0.0005990264580364293,
      "loss": 3.0145,
      "step": 5910
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.8321822285652161,
      "learning_rate": 0.000599026128731111,
      "loss": 2.4526,
      "step": 5911
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.6744887232780457,
      "learning_rate": 0.0005990257993701978,
      "loss": 2.2423,
      "step": 5912
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.8119180202484131,
      "learning_rate": 0.0005990254699536903,
      "loss": 2.4653,
      "step": 5913
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.9183630347251892,
      "learning_rate": 0.0005990251404815883,
      "loss": 2.6031,
      "step": 5914
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.5901970863342285,
      "learning_rate": 0.0005990248109538918,
      "loss": 2.5772,
      "step": 5915
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.6736716032028198,
      "learning_rate": 0.0005990244813706009,
      "loss": 2.8293,
      "step": 5916
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.7712596654891968,
      "learning_rate": 0.0005990241517317158,
      "loss": 2.676,
      "step": 5917
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.746170699596405,
      "learning_rate": 0.0005990238220372365,
      "loss": 2.4788,
      "step": 5918
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.6655517816543579,
      "learning_rate": 0.0005990234922871629,
      "loss": 2.5093,
      "step": 5919
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.7531667947769165,
      "learning_rate": 0.0005990231624814952,
      "loss": 2.4406,
      "step": 5920
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.8725019097328186,
      "learning_rate": 0.0005990228326202337,
      "loss": 2.7777,
      "step": 5921
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.7018865942955017,
      "learning_rate": 0.0005990225027033779,
      "loss": 2.7456,
      "step": 5922
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.6492190957069397,
      "learning_rate": 0.0005990221727309282,
      "loss": 2.5209,
      "step": 5923
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.7452237606048584,
      "learning_rate": 0.0005990218427028848,
      "loss": 3.0138,
      "step": 5924
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.7533861398696899,
      "learning_rate": 0.0005990215126192475,
      "loss": 2.634,
      "step": 5925
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.8471570014953613,
      "learning_rate": 0.0005990211824800166,
      "loss": 2.6166,
      "step": 5926
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.754650890827179,
      "learning_rate": 0.0005990208522851919,
      "loss": 2.3041,
      "step": 5927
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.9143563508987427,
      "learning_rate": 0.0005990205220347735,
      "loss": 2.5875,
      "step": 5928
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.6358237862586975,
      "learning_rate": 0.0005990201917287617,
      "loss": 2.5813,
      "step": 5929
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.7980577349662781,
      "learning_rate": 0.0005990198613671563,
      "loss": 2.6041,
      "step": 5930
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.7693951725959778,
      "learning_rate": 0.0005990195309499574,
      "loss": 2.48,
      "step": 5931
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.6841574311256409,
      "learning_rate": 0.0005990192004771653,
      "loss": 2.8223,
      "step": 5932
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.6579343676567078,
      "learning_rate": 0.0005990188699487797,
      "loss": 2.7192,
      "step": 5933
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.8432329297065735,
      "learning_rate": 0.000599018539364801,
      "loss": 2.6462,
      "step": 5934
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.6155878305435181,
      "learning_rate": 0.000599018208725229,
      "loss": 2.5856,
      "step": 5935
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.76381915807724,
      "learning_rate": 0.000599017878030064,
      "loss": 2.5584,
      "step": 5936
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.6527290344238281,
      "learning_rate": 0.0005990175472793057,
      "loss": 2.8241,
      "step": 5937
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.6523982286453247,
      "learning_rate": 0.0005990172164729547,
      "loss": 2.7721,
      "step": 5938
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.6204187870025635,
      "learning_rate": 0.0005990168856110106,
      "loss": 2.4936,
      "step": 5939
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.7086950540542603,
      "learning_rate": 0.0005990165546934736,
      "loss": 2.4179,
      "step": 5940
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.6415965557098389,
      "learning_rate": 0.0005990162237203437,
      "loss": 2.5791,
      "step": 5941
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.6742240786552429,
      "learning_rate": 0.0005990158926916212,
      "loss": 2.8344,
      "step": 5942
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.6926474571228027,
      "learning_rate": 0.000599015561607306,
      "loss": 2.5839,
      "step": 5943
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.7552475929260254,
      "learning_rate": 0.000599015230467398,
      "loss": 2.3993,
      "step": 5944
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.6380032300949097,
      "learning_rate": 0.0005990148992718975,
      "loss": 2.551,
      "step": 5945
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.6741670370101929,
      "learning_rate": 0.0005990145680208046,
      "loss": 2.4289,
      "step": 5946
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.6973785161972046,
      "learning_rate": 0.0005990142367141192,
      "loss": 2.4924,
      "step": 5947
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.7851482629776001,
      "learning_rate": 0.0005990139053518413,
      "loss": 2.5376,
      "step": 5948
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.0281391143798828,
      "learning_rate": 0.0005990135739339711,
      "loss": 2.6875,
      "step": 5949
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.8115216493606567,
      "learning_rate": 0.0005990132424605087,
      "loss": 2.4912,
      "step": 5950
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.7142164707183838,
      "learning_rate": 0.0005990129109314541,
      "loss": 2.5351,
      "step": 5951
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.7092147469520569,
      "learning_rate": 0.0005990125793468073,
      "loss": 2.3368,
      "step": 5952
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.9094981551170349,
      "learning_rate": 0.0005990122477065685,
      "loss": 2.8181,
      "step": 5953
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.0293112993240356,
      "learning_rate": 0.0005990119160107375,
      "loss": 2.6131,
      "step": 5954
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.7304027676582336,
      "learning_rate": 0.0005990115842593146,
      "loss": 2.6995,
      "step": 5955
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.3013534545898438,
      "learning_rate": 0.0005990112524522999,
      "loss": 2.4346,
      "step": 5956
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.8141217827796936,
      "learning_rate": 0.0005990109205896934,
      "loss": 2.7094,
      "step": 5957
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.6790919303894043,
      "learning_rate": 0.000599010588671495,
      "loss": 2.6002,
      "step": 5958
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.8320969939231873,
      "learning_rate": 0.000599010256697705,
      "loss": 2.7295,
      "step": 5959
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.6737681031227112,
      "learning_rate": 0.0005990099246683232,
      "loss": 2.7281,
      "step": 5960
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.7023961544036865,
      "learning_rate": 0.0005990095925833499,
      "loss": 2.5736,
      "step": 5961
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.5993263125419617,
      "learning_rate": 0.000599009260442785,
      "loss": 2.4131,
      "step": 5962
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.7223510146141052,
      "learning_rate": 0.0005990089282466287,
      "loss": 2.8477,
      "step": 5963
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.6974598169326782,
      "learning_rate": 0.000599008595994881,
      "loss": 2.6901,
      "step": 5964
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.670365571975708,
      "learning_rate": 0.000599008263687542,
      "loss": 2.4216,
      "step": 5965
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.6568438410758972,
      "learning_rate": 0.0005990079313246116,
      "loss": 2.595,
      "step": 5966
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.708008885383606,
      "learning_rate": 0.0005990075989060901,
      "loss": 2.6368,
      "step": 5967
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.8137847781181335,
      "learning_rate": 0.0005990072664319774,
      "loss": 2.6188,
      "step": 5968
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.6906701326370239,
      "learning_rate": 0.0005990069339022736,
      "loss": 2.7024,
      "step": 5969
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.6903250217437744,
      "learning_rate": 0.0005990066013169789,
      "loss": 2.6833,
      "step": 5970
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.7175257802009583,
      "learning_rate": 0.000599006268676093,
      "loss": 2.6555,
      "step": 5971
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.6658067107200623,
      "learning_rate": 0.0005990059359796163,
      "loss": 2.517,
      "step": 5972
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.6270787119865417,
      "learning_rate": 0.0005990056032275488,
      "loss": 2.5365,
      "step": 5973
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.6139387488365173,
      "learning_rate": 0.0005990052704198905,
      "loss": 2.3779,
      "step": 5974
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.6655412316322327,
      "learning_rate": 0.0005990049375566414,
      "loss": 2.3341,
      "step": 5975
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.663642942905426,
      "learning_rate": 0.0005990046046378016,
      "loss": 2.7677,
      "step": 5976
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.6262170672416687,
      "learning_rate": 0.0005990042716633713,
      "loss": 2.7052,
      "step": 5977
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.6322201490402222,
      "learning_rate": 0.0005990039386333504,
      "loss": 2.4433,
      "step": 5978
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.6536937355995178,
      "learning_rate": 0.0005990036055477391,
      "loss": 2.5114,
      "step": 5979
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.7342046499252319,
      "learning_rate": 0.0005990032724065373,
      "loss": 2.7236,
      "step": 5980
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.838060200214386,
      "learning_rate": 0.0005990029392097452,
      "loss": 2.5056,
      "step": 5981
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.6292357444763184,
      "learning_rate": 0.0005990026059573628,
      "loss": 2.6561,
      "step": 5982
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.6256791949272156,
      "learning_rate": 0.0005990022726493901,
      "loss": 2.7041,
      "step": 5983
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.6385711431503296,
      "learning_rate": 0.0005990019392858273,
      "loss": 2.8095,
      "step": 5984
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.6289070844650269,
      "learning_rate": 0.0005990016058666745,
      "loss": 2.4704,
      "step": 5985
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.6935918927192688,
      "learning_rate": 0.0005990012723919314,
      "loss": 2.4618,
      "step": 5986
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.7466434836387634,
      "learning_rate": 0.0005990009388615986,
      "loss": 2.5531,
      "step": 5987
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.6811363101005554,
      "learning_rate": 0.0005990006052756757,
      "loss": 2.6143,
      "step": 5988
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.7913627624511719,
      "learning_rate": 0.000599000271634163,
      "loss": 2.5114,
      "step": 5989
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.6414808630943298,
      "learning_rate": 0.0005989999379370605,
      "loss": 2.7958,
      "step": 5990
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.6635727882385254,
      "learning_rate": 0.0005989996041843684,
      "loss": 2.426,
      "step": 5991
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.6388680338859558,
      "learning_rate": 0.0005989992703760863,
      "loss": 2.5326,
      "step": 5992
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.753498911857605,
      "learning_rate": 0.0005989989365122149,
      "loss": 2.4822,
      "step": 5993
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.7588245868682861,
      "learning_rate": 0.0005989986025927539,
      "loss": 2.6866,
      "step": 5994
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.6372740864753723,
      "learning_rate": 0.0005989982686177033,
      "loss": 2.6927,
      "step": 5995
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.6385692358016968,
      "learning_rate": 0.0005989979345870634,
      "loss": 2.4932,
      "step": 5996
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.6573485136032104,
      "learning_rate": 0.000598997600500834,
      "loss": 2.5725,
      "step": 5997
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.6285983324050903,
      "learning_rate": 0.0005989972663590154,
      "loss": 2.6126,
      "step": 5998
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.6756321787834167,
      "learning_rate": 0.0005989969321616075,
      "loss": 2.5319,
      "step": 5999
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.607003390789032,
      "learning_rate": 0.0005989965979086105,
      "loss": 2.4386,
      "step": 6000
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.6435961127281189,
      "learning_rate": 0.0005989962636000244,
      "loss": 2.5528,
      "step": 6001
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.6312435269355774,
      "learning_rate": 0.0005989959292358492,
      "loss": 2.6764,
      "step": 6002
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.5998682379722595,
      "learning_rate": 0.0005989955948160851,
      "loss": 2.4248,
      "step": 6003
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.6300725340843201,
      "learning_rate": 0.0005989952603407319,
      "loss": 2.5817,
      "step": 6004
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.6372912526130676,
      "learning_rate": 0.0005989949258097899,
      "loss": 2.4477,
      "step": 6005
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.6631180047988892,
      "learning_rate": 0.0005989945912232592,
      "loss": 2.4272,
      "step": 6006
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.9815291166305542,
      "learning_rate": 0.0005989942565811397,
      "loss": 2.5301,
      "step": 6007
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.6334553360939026,
      "learning_rate": 0.0005989939218834315,
      "loss": 2.6423,
      "step": 6008
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.6548982858657837,
      "learning_rate": 0.0005989935871301347,
      "loss": 2.4153,
      "step": 6009
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.7555802464485168,
      "learning_rate": 0.0005989932523212493,
      "loss": 2.2336,
      "step": 6010
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.6371376514434814,
      "learning_rate": 0.0005989929174567756,
      "loss": 2.3896,
      "step": 6011
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.7400413155555725,
      "learning_rate": 0.0005989925825367133,
      "loss": 2.7589,
      "step": 6012
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.7691681981086731,
      "learning_rate": 0.0005989922475610627,
      "loss": 2.7358,
      "step": 6013
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.6839812994003296,
      "learning_rate": 0.0005989919125298237,
      "loss": 2.7187,
      "step": 6014
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.7126228213310242,
      "learning_rate": 0.0005989915774429965,
      "loss": 2.6679,
      "step": 6015
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.787544846534729,
      "learning_rate": 0.0005989912423005812,
      "loss": 2.7709,
      "step": 6016
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.6562281847000122,
      "learning_rate": 0.0005989909071025777,
      "loss": 2.4837,
      "step": 6017
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.6438561081886292,
      "learning_rate": 0.0005989905718489862,
      "loss": 2.6562,
      "step": 6018
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.7569407224655151,
      "learning_rate": 0.0005989902365398067,
      "loss": 2.7607,
      "step": 6019
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.7044653296470642,
      "learning_rate": 0.0005989899011750393,
      "loss": 2.7059,
      "step": 6020
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.7633028030395508,
      "learning_rate": 0.0005989895657546841,
      "loss": 2.4704,
      "step": 6021
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.6689831018447876,
      "learning_rate": 0.0005989892302787408,
      "loss": 2.8623,
      "step": 6022
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.698901891708374,
      "learning_rate": 0.0005989888947472101,
      "loss": 2.7349,
      "step": 6023
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.6723492741584778,
      "learning_rate": 0.0005989885591600916,
      "loss": 2.8359,
      "step": 6024
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.7696161270141602,
      "learning_rate": 0.0005989882235173854,
      "loss": 2.4952,
      "step": 6025
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.7372270226478577,
      "learning_rate": 0.0005989878878190917,
      "loss": 2.7421,
      "step": 6026
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.742151141166687,
      "learning_rate": 0.0005989875520652104,
      "loss": 2.7098,
      "step": 6027
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.6743579506874084,
      "learning_rate": 0.0005989872162557419,
      "loss": 2.5654,
      "step": 6028
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.6317725777626038,
      "learning_rate": 0.0005989868803906858,
      "loss": 2.5531,
      "step": 6029
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.6814800500869751,
      "learning_rate": 0.0005989865444700425,
      "loss": 2.5224,
      "step": 6030
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.6594027876853943,
      "learning_rate": 0.0005989862084938121,
      "loss": 2.7018,
      "step": 6031
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.6676048040390015,
      "learning_rate": 0.0005989858724619943,
      "loss": 2.5689,
      "step": 6032
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.035919189453125,
      "learning_rate": 0.0005989855363745895,
      "loss": 2.8424,
      "step": 6033
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.8633529543876648,
      "learning_rate": 0.0005989852002315975,
      "loss": 2.6948,
      "step": 6034
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.7073356509208679,
      "learning_rate": 0.0005989848640330186,
      "loss": 2.7635,
      "step": 6035
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.7773066163063049,
      "learning_rate": 0.0005989845277788528,
      "loss": 2.4017,
      "step": 6036
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.6695416569709778,
      "learning_rate": 0.0005989841914691002,
      "loss": 3.0959,
      "step": 6037
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.6403731107711792,
      "learning_rate": 0.0005989838551037607,
      "loss": 2.3942,
      "step": 6038
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.8236017227172852,
      "learning_rate": 0.0005989835186828345,
      "loss": 2.5462,
      "step": 6039
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.6525818109512329,
      "learning_rate": 0.0005989831822063216,
      "loss": 2.6517,
      "step": 6040
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.7089942693710327,
      "learning_rate": 0.000598982845674222,
      "loss": 2.6014,
      "step": 6041
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.7374885678291321,
      "learning_rate": 0.0005989825090865359,
      "loss": 2.5766,
      "step": 6042
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.713871419429779,
      "learning_rate": 0.0005989821724432634,
      "loss": 2.6255,
      "step": 6043
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.6436424851417542,
      "learning_rate": 0.0005989818357444044,
      "loss": 2.7316,
      "step": 6044
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.8660258054733276,
      "learning_rate": 0.000598981498989959,
      "loss": 2.8952,
      "step": 6045
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.7250908613204956,
      "learning_rate": 0.0005989811621799273,
      "loss": 2.5078,
      "step": 6046
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.7120116353034973,
      "learning_rate": 0.0005989808253143094,
      "loss": 2.5764,
      "step": 6047
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.8442214727401733,
      "learning_rate": 0.0005989804883931054,
      "loss": 2.6284,
      "step": 6048
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.8034047484397888,
      "learning_rate": 0.0005989801514163152,
      "loss": 2.5249,
      "step": 6049
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.7131545543670654,
      "learning_rate": 0.0005989798143839389,
      "loss": 2.5295,
      "step": 6050
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.8538998961448669,
      "learning_rate": 0.0005989794772959766,
      "loss": 2.8962,
      "step": 6051
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.6416100263595581,
      "learning_rate": 0.0005989791401524284,
      "loss": 2.6372,
      "step": 6052
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.7840357422828674,
      "learning_rate": 0.0005989788029532945,
      "loss": 2.4729,
      "step": 6053
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.8343158960342407,
      "learning_rate": 0.0005989784656985745,
      "loss": 2.3925,
      "step": 6054
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.8930277824401855,
      "learning_rate": 0.000598978128388269,
      "loss": 2.3638,
      "step": 6055
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.7951227426528931,
      "learning_rate": 0.0005989777910223777,
      "loss": 2.7742,
      "step": 6056
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.9893154501914978,
      "learning_rate": 0.0005989774536009008,
      "loss": 2.5489,
      "step": 6057
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.7223228216171265,
      "learning_rate": 0.0005989771161238383,
      "loss": 2.8239,
      "step": 6058
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.7082763910293579,
      "learning_rate": 0.0005989767785911904,
      "loss": 2.8703,
      "step": 6059
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.868928849697113,
      "learning_rate": 0.0005989764410029571,
      "loss": 2.6236,
      "step": 6060
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.7785822749137878,
      "learning_rate": 0.0005989761033591384,
      "loss": 2.5433,
      "step": 6061
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.6605843305587769,
      "learning_rate": 0.0005989757656597344,
      "loss": 2.7447,
      "step": 6062
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.6737704873085022,
      "learning_rate": 0.0005989754279047451,
      "loss": 2.4978,
      "step": 6063
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.8305814862251282,
      "learning_rate": 0.0005989750900941707,
      "loss": 2.6105,
      "step": 6064
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.6226862072944641,
      "learning_rate": 0.0005989747522280111,
      "loss": 2.4292,
      "step": 6065
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.668029248714447,
      "learning_rate": 0.0005989744143062665,
      "loss": 2.6627,
      "step": 6066
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.780066967010498,
      "learning_rate": 0.0005989740763289369,
      "loss": 2.6373,
      "step": 6067
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.730731189250946,
      "learning_rate": 0.0005989737382960224,
      "loss": 2.5092,
      "step": 6068
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.6538262367248535,
      "learning_rate": 0.000598973400207523,
      "loss": 2.5603,
      "step": 6069
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.752194344997406,
      "learning_rate": 0.0005989730620634388,
      "loss": 2.6647,
      "step": 6070
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.961894690990448,
      "learning_rate": 0.00059897272386377,
      "loss": 2.7606,
      "step": 6071
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.7220955491065979,
      "learning_rate": 0.0005989723856085164,
      "loss": 2.4657,
      "step": 6072
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.6820975542068481,
      "learning_rate": 0.0005989720472976782,
      "loss": 2.8509,
      "step": 6073
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.7768370509147644,
      "learning_rate": 0.0005989717089312554,
      "loss": 2.5524,
      "step": 6074
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.6970723867416382,
      "learning_rate": 0.0005989713705092482,
      "loss": 2.5898,
      "step": 6075
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.6355646252632141,
      "learning_rate": 0.0005989710320316566,
      "loss": 2.675,
      "step": 6076
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.740101158618927,
      "learning_rate": 0.0005989706934984806,
      "loss": 2.6355,
      "step": 6077
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.6544362902641296,
      "learning_rate": 0.0005989703549097203,
      "loss": 2.4,
      "step": 6078
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.6833721399307251,
      "learning_rate": 0.0005989700162653759,
      "loss": 2.6453,
      "step": 6079
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.8959108591079712,
      "learning_rate": 0.0005989696775654471,
      "loss": 2.6505,
      "step": 6080
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.7110438346862793,
      "learning_rate": 0.0005989693388099343,
      "loss": 2.668,
      "step": 6081
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.7153543829917908,
      "learning_rate": 0.0005989689999988375,
      "loss": 2.6752,
      "step": 6082
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.7096334099769592,
      "learning_rate": 0.0005989686611321566,
      "loss": 2.5874,
      "step": 6083
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.7135981321334839,
      "learning_rate": 0.000598968322209892,
      "loss": 2.5026,
      "step": 6084
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.771459698677063,
      "learning_rate": 0.0005989679832320433,
      "loss": 2.5447,
      "step": 6085
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.672787606716156,
      "learning_rate": 0.0005989676441986109,
      "loss": 2.787,
      "step": 6086
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.6441390514373779,
      "learning_rate": 0.0005989673051095948,
      "loss": 2.6748,
      "step": 6087
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.686566174030304,
      "learning_rate": 0.000598966965964995,
      "loss": 2.849,
      "step": 6088
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.7083699703216553,
      "learning_rate": 0.0005989666267648116,
      "loss": 2.5709,
      "step": 6089
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.7068089246749878,
      "learning_rate": 0.0005989662875090447,
      "loss": 2.661,
      "step": 6090
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.6731036901473999,
      "learning_rate": 0.0005989659481976943,
      "loss": 2.8474,
      "step": 6091
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.6665139198303223,
      "learning_rate": 0.0005989656088307605,
      "loss": 2.7072,
      "step": 6092
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.653652548789978,
      "learning_rate": 0.0005989652694082432,
      "loss": 2.6185,
      "step": 6093
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.6544401049613953,
      "learning_rate": 0.0005989649299301429,
      "loss": 2.4778,
      "step": 6094
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.6433331370353699,
      "learning_rate": 0.0005989645903964592,
      "loss": 2.7599,
      "step": 6095
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.607983410358429,
      "learning_rate": 0.0005989642508071924,
      "loss": 2.5026,
      "step": 6096
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.646777331829071,
      "learning_rate": 0.0005989639111623425,
      "loss": 2.3575,
      "step": 6097
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.7055044770240784,
      "learning_rate": 0.0005989635714619095,
      "loss": 2.6839,
      "step": 6098
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.6348552703857422,
      "learning_rate": 0.0005989632317058935,
      "loss": 2.4349,
      "step": 6099
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.6620230674743652,
      "learning_rate": 0.0005989628918942946,
      "loss": 2.8073,
      "step": 6100
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.6938955783843994,
      "learning_rate": 0.000598962552027113,
      "loss": 2.8238,
      "step": 6101
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.7068553566932678,
      "learning_rate": 0.0005989622121043485,
      "loss": 2.4856,
      "step": 6102
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.7014380097389221,
      "learning_rate": 0.0005989618721260014,
      "loss": 2.5021,
      "step": 6103
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.8849391937255859,
      "learning_rate": 0.0005989615320920714,
      "loss": 2.396,
      "step": 6104
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.6529056429862976,
      "learning_rate": 0.000598961192002559,
      "loss": 2.5403,
      "step": 6105
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.6507483124732971,
      "learning_rate": 0.0005989608518574641,
      "loss": 2.6182,
      "step": 6106
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.7830371856689453,
      "learning_rate": 0.0005989605116567867,
      "loss": 2.5858,
      "step": 6107
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.817604124546051,
      "learning_rate": 0.0005989601714005267,
      "loss": 2.6483,
      "step": 6108
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.8808737397193909,
      "learning_rate": 0.0005989598310886846,
      "loss": 2.4868,
      "step": 6109
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.7683161497116089,
      "learning_rate": 0.0005989594907212601,
      "loss": 2.663,
      "step": 6110
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.6681436896324158,
      "learning_rate": 0.0005989591502982535,
      "loss": 2.5568,
      "step": 6111
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.7142639756202698,
      "learning_rate": 0.0005989588098196647,
      "loss": 2.4398,
      "step": 6112
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.6827619075775146,
      "learning_rate": 0.0005989584692854937,
      "loss": 2.3762,
      "step": 6113
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.7310242056846619,
      "learning_rate": 0.0005989581286957408,
      "loss": 2.5674,
      "step": 6114
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.7658408284187317,
      "learning_rate": 0.000598957788050406,
      "loss": 2.6108,
      "step": 6115
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.6361699104309082,
      "learning_rate": 0.0005989574473494893,
      "loss": 2.5505,
      "step": 6116
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.672448992729187,
      "learning_rate": 0.0005989571065929906,
      "loss": 2.7821,
      "step": 6117
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.5850389003753662,
      "learning_rate": 0.0005989567657809101,
      "loss": 2.491,
      "step": 6118
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.6779553890228271,
      "learning_rate": 0.0005989564249132481,
      "loss": 2.6519,
      "step": 6119
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.7387754321098328,
      "learning_rate": 0.0005989560839900044,
      "loss": 2.7871,
      "step": 6120
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.6237959861755371,
      "learning_rate": 0.000598955743011179,
      "loss": 2.4499,
      "step": 6121
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.6771878600120544,
      "learning_rate": 0.0005989554019767723,
      "loss": 2.4879,
      "step": 6122
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.6798510551452637,
      "learning_rate": 0.000598955060886784,
      "loss": 2.5115,
      "step": 6123
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.6991864442825317,
      "learning_rate": 0.0005989547197412143,
      "loss": 2.5536,
      "step": 6124
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.6308783292770386,
      "learning_rate": 0.0005989543785400633,
      "loss": 2.5079,
      "step": 6125
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.6717198491096497,
      "learning_rate": 0.000598954037283331,
      "loss": 2.8799,
      "step": 6126
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.6495696902275085,
      "learning_rate": 0.0005989536959710176,
      "loss": 2.7068,
      "step": 6127
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.7463585138320923,
      "learning_rate": 0.0005989533546031231,
      "loss": 2.4339,
      "step": 6128
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.6378111839294434,
      "learning_rate": 0.0005989530131796474,
      "loss": 2.7298,
      "step": 6129
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.7062200903892517,
      "learning_rate": 0.0005989526717005907,
      "loss": 2.5142,
      "step": 6130
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.6959654688835144,
      "learning_rate": 0.0005989523301659531,
      "loss": 2.6549,
      "step": 6131
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.7086201906204224,
      "learning_rate": 0.0005989519885757346,
      "loss": 2.4027,
      "step": 6132
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.871336042881012,
      "learning_rate": 0.0005989516469299352,
      "loss": 2.4689,
      "step": 6133
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.6414132714271545,
      "learning_rate": 0.0005989513052285552,
      "loss": 2.2494,
      "step": 6134
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.758110523223877,
      "learning_rate": 0.0005989509634715944,
      "loss": 2.7228,
      "step": 6135
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.6614354848861694,
      "learning_rate": 0.000598950621659053,
      "loss": 2.5971,
      "step": 6136
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.7189377546310425,
      "learning_rate": 0.0005989502797909311,
      "loss": 2.4908,
      "step": 6137
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.7023624777793884,
      "learning_rate": 0.0005989499378672286,
      "loss": 2.3464,
      "step": 6138
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.6194258332252502,
      "learning_rate": 0.0005989495958879457,
      "loss": 2.6985,
      "step": 6139
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.6711640357971191,
      "learning_rate": 0.0005989492538530825,
      "loss": 2.5244,
      "step": 6140
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.622454822063446,
      "learning_rate": 0.000598948911762639,
      "loss": 2.6327,
      "step": 6141
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.6300225853919983,
      "learning_rate": 0.0005989485696166152,
      "loss": 2.6609,
      "step": 6142
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.6461419463157654,
      "learning_rate": 0.0005989482274150112,
      "loss": 2.715,
      "step": 6143
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.7687452435493469,
      "learning_rate": 0.000598947885157827,
      "loss": 2.3173,
      "step": 6144
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.6775112152099609,
      "learning_rate": 0.0005989475428450629,
      "loss": 2.656,
      "step": 6145
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.7821913957595825,
      "learning_rate": 0.0005989472004767186,
      "loss": 2.5211,
      "step": 6146
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.8250911235809326,
      "learning_rate": 0.0005989468580527947,
      "loss": 2.6233,
      "step": 6147
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.7077082395553589,
      "learning_rate": 0.0005989465155732907,
      "loss": 2.4971,
      "step": 6148
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.9418608546257019,
      "learning_rate": 0.000598946173038207,
      "loss": 2.667,
      "step": 6149
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.724043607711792,
      "learning_rate": 0.0005989458304475434,
      "loss": 2.6964,
      "step": 6150
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.7921618223190308,
      "learning_rate": 0.0005989454878013003,
      "loss": 2.5285,
      "step": 6151
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.6992756128311157,
      "learning_rate": 0.0005989451450994775,
      "loss": 2.5896,
      "step": 6152
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.6478043794631958,
      "learning_rate": 0.0005989448023420752,
      "loss": 2.4986,
      "step": 6153
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.0058701038360596,
      "learning_rate": 0.0005989444595290934,
      "loss": 2.706,
      "step": 6154
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.8418013453483582,
      "learning_rate": 0.0005989441166605322,
      "loss": 2.5031,
      "step": 6155
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.6648939251899719,
      "learning_rate": 0.0005989437737363916,
      "loss": 2.6342,
      "step": 6156
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.8192908763885498,
      "learning_rate": 0.0005989434307566718,
      "loss": 2.6272,
      "step": 6157
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.9192186594009399,
      "learning_rate": 0.0005989430877213727,
      "loss": 2.575,
      "step": 6158
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.7002468705177307,
      "learning_rate": 0.0005989427446304944,
      "loss": 2.5839,
      "step": 6159
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.8434543013572693,
      "learning_rate": 0.000598942401484037,
      "loss": 2.5275,
      "step": 6160
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.1755765676498413,
      "learning_rate": 0.0005989420582820006,
      "loss": 2.929,
      "step": 6161
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.7565711736679077,
      "learning_rate": 0.0005989417150243853,
      "loss": 2.7117,
      "step": 6162
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.785024106502533,
      "learning_rate": 0.000598941371711191,
      "loss": 2.5444,
      "step": 6163
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.8086509704589844,
      "learning_rate": 0.0005989410283424179,
      "loss": 2.6309,
      "step": 6164
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.7175766825675964,
      "learning_rate": 0.000598940684918066,
      "loss": 2.7216,
      "step": 6165
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.669805645942688,
      "learning_rate": 0.0005989403414381354,
      "loss": 2.6005,
      "step": 6166
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.8454914689064026,
      "learning_rate": 0.0005989399979026261,
      "loss": 2.3473,
      "step": 6167
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.8224222660064697,
      "learning_rate": 0.0005989396543115384,
      "loss": 2.6531,
      "step": 6168
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.6686979532241821,
      "learning_rate": 0.000598939310664872,
      "loss": 2.5738,
      "step": 6169
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.7309778928756714,
      "learning_rate": 0.0005989389669626272,
      "loss": 2.5135,
      "step": 6170
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.0207579135894775,
      "learning_rate": 0.000598938623204804,
      "loss": 2.6323,
      "step": 6171
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.6943195462226868,
      "learning_rate": 0.0005989382793914023,
      "loss": 2.7628,
      "step": 6172
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.7044957280158997,
      "learning_rate": 0.0005989379355224225,
      "loss": 2.5374,
      "step": 6173
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.8043637871742249,
      "learning_rate": 0.0005989375915978645,
      "loss": 2.6046,
      "step": 6174
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.7522092461585999,
      "learning_rate": 0.0005989372476177283,
      "loss": 2.4829,
      "step": 6175
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.644911527633667,
      "learning_rate": 0.000598936903582014,
      "loss": 2.5465,
      "step": 6176
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.7229034304618835,
      "learning_rate": 0.0005989365594907218,
      "loss": 2.526,
      "step": 6177
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.7468616366386414,
      "learning_rate": 0.0005989362153438516,
      "loss": 2.5777,
      "step": 6178
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.654587984085083,
      "learning_rate": 0.0005989358711414035,
      "loss": 2.5255,
      "step": 6179
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.7599443197250366,
      "learning_rate": 0.0005989355268833775,
      "loss": 2.5812,
      "step": 6180
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.6489912271499634,
      "learning_rate": 0.0005989351825697738,
      "loss": 2.5483,
      "step": 6181
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.6986436247825623,
      "learning_rate": 0.0005989348382005925,
      "loss": 2.7889,
      "step": 6182
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.5566356182098389,
      "learning_rate": 0.0005989344937758336,
      "loss": 2.6555,
      "step": 6183
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.6189549565315247,
      "learning_rate": 0.0005989341492954969,
      "loss": 2.5688,
      "step": 6184
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.7733753323554993,
      "learning_rate": 0.0005989338047595828,
      "loss": 2.7185,
      "step": 6185
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.6286993026733398,
      "learning_rate": 0.0005989334601680914,
      "loss": 2.6633,
      "step": 6186
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.5923075079917908,
      "learning_rate": 0.0005989331155210225,
      "loss": 2.6093,
      "step": 6187
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.6121299862861633,
      "learning_rate": 0.0005989327708183763,
      "loss": 2.4667,
      "step": 6188
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.6493319869041443,
      "learning_rate": 0.0005989324260601529,
      "loss": 2.6668,
      "step": 6189
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.6609605550765991,
      "learning_rate": 0.0005989320812463522,
      "loss": 2.6517,
      "step": 6190
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.6939511895179749,
      "learning_rate": 0.0005989317363769745,
      "loss": 2.4363,
      "step": 6191
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.7012224793434143,
      "learning_rate": 0.0005989313914520197,
      "loss": 2.5268,
      "step": 6192
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.7705286145210266,
      "learning_rate": 0.000598931046471488,
      "loss": 2.5919,
      "step": 6193
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.6339167952537537,
      "learning_rate": 0.0005989307014353792,
      "loss": 2.6264,
      "step": 6194
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.6981265544891357,
      "learning_rate": 0.0005989303563436937,
      "loss": 2.6337,
      "step": 6195
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.6270614266395569,
      "learning_rate": 0.0005989300111964314,
      "loss": 2.4432,
      "step": 6196
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.6970755457878113,
      "learning_rate": 0.0005989296659935923,
      "loss": 2.7591,
      "step": 6197
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.7140821218490601,
      "learning_rate": 0.0005989293207351766,
      "loss": 2.6739,
      "step": 6198
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.6885372400283813,
      "learning_rate": 0.0005989289754211842,
      "loss": 2.6774,
      "step": 6199
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.6278656721115112,
      "learning_rate": 0.0005989286300516152,
      "loss": 2.3329,
      "step": 6200
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.7214047908782959,
      "learning_rate": 0.0005989282846264699,
      "loss": 2.5287,
      "step": 6201
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.7195731401443481,
      "learning_rate": 0.0005989279391457481,
      "loss": 2.7415,
      "step": 6202
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.7017256021499634,
      "learning_rate": 0.00059892759360945,
      "loss": 2.6092,
      "step": 6203
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.7265132069587708,
      "learning_rate": 0.0005989272480175756,
      "loss": 2.4973,
      "step": 6204
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.6362007260322571,
      "learning_rate": 0.000598926902370125,
      "loss": 2.6139,
      "step": 6205
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.800751805305481,
      "learning_rate": 0.0005989265566670982,
      "loss": 2.6884,
      "step": 6206
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.7482491731643677,
      "learning_rate": 0.0005989262109084953,
      "loss": 2.6688,
      "step": 6207
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.6364799737930298,
      "learning_rate": 0.0005989258650943164,
      "loss": 2.4543,
      "step": 6208
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.6223861575126648,
      "learning_rate": 0.0005989255192245616,
      "loss": 2.5732,
      "step": 6209
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.7343337535858154,
      "learning_rate": 0.0005989251732992308,
      "loss": 2.7672,
      "step": 6210
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.8010201454162598,
      "learning_rate": 0.0005989248273183242,
      "loss": 2.8593,
      "step": 6211
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.6725952625274658,
      "learning_rate": 0.0005989244812818419,
      "loss": 2.6927,
      "step": 6212
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.6388974785804749,
      "learning_rate": 0.0005989241351897839,
      "loss": 2.3551,
      "step": 6213
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.70618736743927,
      "learning_rate": 0.0005989237890421501,
      "loss": 2.6887,
      "step": 6214
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.6509697437286377,
      "learning_rate": 0.0005989234428389408,
      "loss": 2.5088,
      "step": 6215
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.74836665391922,
      "learning_rate": 0.0005989230965801561,
      "loss": 2.6778,
      "step": 6216
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.6048654317855835,
      "learning_rate": 0.0005989227502657959,
      "loss": 2.6784,
      "step": 6217
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.6870517134666443,
      "learning_rate": 0.0005989224038958602,
      "loss": 2.7622,
      "step": 6218
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.7857804298400879,
      "learning_rate": 0.0005989220574703493,
      "loss": 2.609,
      "step": 6219
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.6894552707672119,
      "learning_rate": 0.0005989217109892631,
      "loss": 2.5393,
      "step": 6220
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.7380504608154297,
      "learning_rate": 0.0005989213644526018,
      "loss": 2.605,
      "step": 6221
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.7615821361541748,
      "learning_rate": 0.0005989210178603653,
      "loss": 2.5723,
      "step": 6222
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.6931268572807312,
      "learning_rate": 0.0005989206712125537,
      "loss": 2.6616,
      "step": 6223
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.7847965359687805,
      "learning_rate": 0.0005989203245091671,
      "loss": 2.4835,
      "step": 6224
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.7809929251670837,
      "learning_rate": 0.0005989199777502056,
      "loss": 2.6679,
      "step": 6225
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.6244871020317078,
      "learning_rate": 0.0005989196309356693,
      "loss": 2.6542,
      "step": 6226
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.6656405925750732,
      "learning_rate": 0.0005989192840655582,
      "loss": 2.7653,
      "step": 6227
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.8069122433662415,
      "learning_rate": 0.0005989189371398723,
      "loss": 2.4131,
      "step": 6228
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.6200644969940186,
      "learning_rate": 0.0005989185901586118,
      "loss": 2.546,
      "step": 6229
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.8449637293815613,
      "learning_rate": 0.0005989182431217766,
      "loss": 2.6164,
      "step": 6230
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.7562331557273865,
      "learning_rate": 0.0005989178960293668,
      "loss": 2.6961,
      "step": 6231
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.6534320712089539,
      "learning_rate": 0.0005989175488813826,
      "loss": 2.6751,
      "step": 6232
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.6660186648368835,
      "learning_rate": 0.0005989172016778241,
      "loss": 2.7988,
      "step": 6233
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.7785606384277344,
      "learning_rate": 0.0005989168544186911,
      "loss": 2.7136,
      "step": 6234
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.6765159368515015,
      "learning_rate": 0.0005989165071039839,
      "loss": 2.4578,
      "step": 6235
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.682619571685791,
      "learning_rate": 0.0005989161597337024,
      "loss": 2.7834,
      "step": 6236
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.6747764348983765,
      "learning_rate": 0.0005989158123078467,
      "loss": 2.502,
      "step": 6237
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.6878972053527832,
      "learning_rate": 0.000598915464826417,
      "loss": 2.4433,
      "step": 6238
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.7879847884178162,
      "learning_rate": 0.0005989151172894133,
      "loss": 2.7517,
      "step": 6239
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.7690902948379517,
      "learning_rate": 0.0005989147696968357,
      "loss": 2.4217,
      "step": 6240
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.7304566502571106,
      "learning_rate": 0.0005989144220486841,
      "loss": 2.7332,
      "step": 6241
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.7989246249198914,
      "learning_rate": 0.0005989140743449587,
      "loss": 2.5587,
      "step": 6242
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.6643127799034119,
      "learning_rate": 0.0005989137265856594,
      "loss": 2.8067,
      "step": 6243
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.6974477171897888,
      "learning_rate": 0.0005989133787707865,
      "loss": 2.6516,
      "step": 6244
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.6583436131477356,
      "learning_rate": 0.0005989130309003399,
      "loss": 2.5119,
      "step": 6245
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.898745596408844,
      "learning_rate": 0.0005989126829743197,
      "loss": 2.512,
      "step": 6246
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.6936960816383362,
      "learning_rate": 0.0005989123349927262,
      "loss": 2.5641,
      "step": 6247
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.6878085136413574,
      "learning_rate": 0.0005989119869555592,
      "loss": 2.9304,
      "step": 6248
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.6742512583732605,
      "learning_rate": 0.0005989116388628187,
      "loss": 2.4872,
      "step": 6249
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.670449435710907,
      "learning_rate": 0.0005989112907145049,
      "loss": 2.6152,
      "step": 6250
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.6454870700836182,
      "learning_rate": 0.0005989109425106179,
      "loss": 2.5536,
      "step": 6251
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.6213502883911133,
      "learning_rate": 0.0005989105942511576,
      "loss": 2.5654,
      "step": 6252
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.5969388484954834,
      "learning_rate": 0.0005989102459361244,
      "loss": 2.4222,
      "step": 6253
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.6618720293045044,
      "learning_rate": 0.0005989098975655179,
      "loss": 2.4413,
      "step": 6254
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.6526410579681396,
      "learning_rate": 0.0005989095491393385,
      "loss": 2.6217,
      "step": 6255
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.5864576697349548,
      "learning_rate": 0.0005989092006575863,
      "loss": 2.6306,
      "step": 6256
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.626047670841217,
      "learning_rate": 0.0005989088521202611,
      "loss": 2.6823,
      "step": 6257
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.6646767258644104,
      "learning_rate": 0.0005989085035273631,
      "loss": 2.5425,
      "step": 6258
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.6313406825065613,
      "learning_rate": 0.0005989081548788924,
      "loss": 2.7063,
      "step": 6259
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.6934822201728821,
      "learning_rate": 0.000598907806174849,
      "loss": 2.7219,
      "step": 6260
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.6338706612586975,
      "learning_rate": 0.0005989074574152331,
      "loss": 2.8204,
      "step": 6261
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.7167535424232483,
      "learning_rate": 0.0005989071086000445,
      "loss": 2.5038,
      "step": 6262
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.6286258101463318,
      "learning_rate": 0.0005989067597292836,
      "loss": 2.9031,
      "step": 6263
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.6542906165122986,
      "learning_rate": 0.0005989064108029502,
      "loss": 2.5891,
      "step": 6264
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.6188015937805176,
      "learning_rate": 0.0005989060618210443,
      "loss": 2.4282,
      "step": 6265
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.6924909949302673,
      "learning_rate": 0.0005989057127835664,
      "loss": 2.516,
      "step": 6266
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.6321437358856201,
      "learning_rate": 0.0005989053636905162,
      "loss": 2.6387,
      "step": 6267
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.6461503505706787,
      "learning_rate": 0.0005989050145418938,
      "loss": 2.5627,
      "step": 6268
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.6965565085411072,
      "learning_rate": 0.0005989046653376993,
      "loss": 2.6754,
      "step": 6269
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.7209272384643555,
      "learning_rate": 0.0005989043160779329,
      "loss": 2.505,
      "step": 6270
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.7170711755752563,
      "learning_rate": 0.0005989039667625944,
      "loss": 2.4818,
      "step": 6271
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.6259094476699829,
      "learning_rate": 0.000598903617391684,
      "loss": 2.6333,
      "step": 6272
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.7038357257843018,
      "learning_rate": 0.0005989032679652018,
      "loss": 2.775,
      "step": 6273
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.6569832563400269,
      "learning_rate": 0.000598902918483148,
      "loss": 2.6702,
      "step": 6274
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.7067884206771851,
      "learning_rate": 0.0005989025689455224,
      "loss": 2.5854,
      "step": 6275
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.7078713774681091,
      "learning_rate": 0.0005989022193523252,
      "loss": 2.5947,
      "step": 6276
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.7525338530540466,
      "learning_rate": 0.0005989018697035563,
      "loss": 2.549,
      "step": 6277
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.7521386742591858,
      "learning_rate": 0.000598901519999216,
      "loss": 2.832,
      "step": 6278
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.7009632587432861,
      "learning_rate": 0.0005989011702393043,
      "loss": 2.5099,
      "step": 6279
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.6398224830627441,
      "learning_rate": 0.0005989008204238212,
      "loss": 2.5891,
      "step": 6280
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.7276092171669006,
      "learning_rate": 0.0005989004705527668,
      "loss": 2.6016,
      "step": 6281
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.6887854933738708,
      "learning_rate": 0.0005989001206261411,
      "loss": 2.6425,
      "step": 6282
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.6825200915336609,
      "learning_rate": 0.0005988997706439443,
      "loss": 2.4696,
      "step": 6283
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.6726831197738647,
      "learning_rate": 0.0005988994206061764,
      "loss": 2.3552,
      "step": 6284
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.6724638342857361,
      "learning_rate": 0.0005988990705128374,
      "loss": 2.8434,
      "step": 6285
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.6979633569717407,
      "learning_rate": 0.0005988987203639274,
      "loss": 2.7866,
      "step": 6286
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.6094707250595093,
      "learning_rate": 0.0005988983701594464,
      "loss": 2.6521,
      "step": 6287
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.6507707238197327,
      "learning_rate": 0.0005988980198993948,
      "loss": 2.7417,
      "step": 6288
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.7473370432853699,
      "learning_rate": 0.0005988976695837723,
      "loss": 2.4629,
      "step": 6289
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.6724266409873962,
      "learning_rate": 0.000598897319212579,
      "loss": 2.5557,
      "step": 6290
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.7448773980140686,
      "learning_rate": 0.0005988969687858153,
      "loss": 2.6218,
      "step": 6291
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.6583198308944702,
      "learning_rate": 0.0005988966183034807,
      "loss": 2.7906,
      "step": 6292
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.6351066827774048,
      "learning_rate": 0.0005988962677655758,
      "loss": 2.5444,
      "step": 6293
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.6708899736404419,
      "learning_rate": 0.0005988959171721002,
      "loss": 2.7505,
      "step": 6294
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.643730640411377,
      "learning_rate": 0.0005988955665230544,
      "loss": 2.4539,
      "step": 6295
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.7345902323722839,
      "learning_rate": 0.0005988952158184382,
      "loss": 2.6325,
      "step": 6296
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.7318258881568909,
      "learning_rate": 0.0005988948650582518,
      "loss": 2.7309,
      "step": 6297
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.6810411214828491,
      "learning_rate": 0.0005988945142424951,
      "loss": 2.5433,
      "step": 6298
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.7177568078041077,
      "learning_rate": 0.0005988941633711684,
      "loss": 2.4323,
      "step": 6299
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.8000138401985168,
      "learning_rate": 0.0005988938124442717,
      "loss": 2.5455,
      "step": 6300
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.6245556473731995,
      "learning_rate": 0.0005988934614618047,
      "loss": 2.4804,
      "step": 6301
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.6788968443870544,
      "learning_rate": 0.000598893110423768,
      "loss": 2.6026,
      "step": 6302
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.686843991279602,
      "learning_rate": 0.0005988927593301613,
      "loss": 2.5882,
      "step": 6303
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.7462832927703857,
      "learning_rate": 0.0005988924081809848,
      "loss": 2.5142,
      "step": 6304
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.6938372850418091,
      "learning_rate": 0.0005988920569762387,
      "loss": 2.5906,
      "step": 6305
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.5983073115348816,
      "learning_rate": 0.0005988917057159228,
      "loss": 2.3669,
      "step": 6306
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.5774730443954468,
      "learning_rate": 0.0005988913544000374,
      "loss": 2.327,
      "step": 6307
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.7796604633331299,
      "learning_rate": 0.0005988910030285823,
      "loss": 2.6971,
      "step": 6308
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.7235021591186523,
      "learning_rate": 0.0005988906516015578,
      "loss": 2.3741,
      "step": 6309
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.7423845529556274,
      "learning_rate": 0.000598890300118964,
      "loss": 2.5247,
      "step": 6310
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.9229227900505066,
      "learning_rate": 0.0005988899485808006,
      "loss": 2.6091,
      "step": 6311
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.6302977800369263,
      "learning_rate": 0.0005988895969870681,
      "loss": 2.4282,
      "step": 6312
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.8717846274375916,
      "learning_rate": 0.0005988892453377664,
      "loss": 2.3174,
      "step": 6313
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.8553140163421631,
      "learning_rate": 0.0005988888936328954,
      "loss": 2.5467,
      "step": 6314
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.6752653121948242,
      "learning_rate": 0.0005988885418724554,
      "loss": 2.4627,
      "step": 6315
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.8661324977874756,
      "learning_rate": 0.0005988881900564464,
      "loss": 2.5765,
      "step": 6316
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.975222110748291,
      "learning_rate": 0.0005988878381848684,
      "loss": 2.5612,
      "step": 6317
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.7433051466941833,
      "learning_rate": 0.0005988874862577216,
      "loss": 2.3004,
      "step": 6318
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.7061551213264465,
      "learning_rate": 0.0005988871342750058,
      "loss": 2.6227,
      "step": 6319
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.0088964700698853,
      "learning_rate": 0.0005988867822367213,
      "loss": 2.3919,
      "step": 6320
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.7374306321144104,
      "learning_rate": 0.0005988864301428682,
      "loss": 2.6417,
      "step": 6321
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.7323024272918701,
      "learning_rate": 0.0005988860779934465,
      "loss": 2.7803,
      "step": 6322
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.8684176206588745,
      "learning_rate": 0.0005988857257884561,
      "loss": 2.5107,
      "step": 6323
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.7113103270530701,
      "learning_rate": 0.0005988853735278973,
      "loss": 2.5775,
      "step": 6324
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.7389875650405884,
      "learning_rate": 0.0005988850212117699,
      "loss": 2.4616,
      "step": 6325
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.693077564239502,
      "learning_rate": 0.0005988846688400743,
      "loss": 2.3375,
      "step": 6326
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.897437572479248,
      "learning_rate": 0.0005988843164128104,
      "loss": 2.4367,
      "step": 6327
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.6988862156867981,
      "learning_rate": 0.0005988839639299782,
      "loss": 2.414,
      "step": 6328
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.7446315288543701,
      "learning_rate": 0.0005988836113915778,
      "loss": 2.3222,
      "step": 6329
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.7248067855834961,
      "learning_rate": 0.0005988832587976095,
      "loss": 2.4872,
      "step": 6330
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.8135746717453003,
      "learning_rate": 0.0005988829061480729,
      "loss": 2.6787,
      "step": 6331
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.869778573513031,
      "learning_rate": 0.0005988825534429684,
      "loss": 2.5117,
      "step": 6332
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.7666957974433899,
      "learning_rate": 0.0005988822006822961,
      "loss": 2.5851,
      "step": 6333
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.6355941891670227,
      "learning_rate": 0.0005988818478660558,
      "loss": 2.7684,
      "step": 6334
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.7483192682266235,
      "learning_rate": 0.0005988814949942478,
      "loss": 2.4713,
      "step": 6335
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.6515748500823975,
      "learning_rate": 0.000598881142066872,
      "loss": 2.8394,
      "step": 6336
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.6681146025657654,
      "learning_rate": 0.0005988807890839287,
      "loss": 2.668,
      "step": 6337
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.6407503485679626,
      "learning_rate": 0.0005988804360454179,
      "loss": 2.4546,
      "step": 6338
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.6715837717056274,
      "learning_rate": 0.0005988800829513394,
      "loss": 2.5905,
      "step": 6339
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.6852036714553833,
      "learning_rate": 0.0005988797298016934,
      "loss": 2.3792,
      "step": 6340
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.660635232925415,
      "learning_rate": 0.0005988793765964802,
      "loss": 2.6727,
      "step": 6341
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.6191858649253845,
      "learning_rate": 0.0005988790233356996,
      "loss": 2.6145,
      "step": 6342
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.6463475823402405,
      "learning_rate": 0.0005988786700193517,
      "loss": 2.652,
      "step": 6343
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.6105049848556519,
      "learning_rate": 0.0005988783166474366,
      "loss": 2.501,
      "step": 6344
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.6020640730857849,
      "learning_rate": 0.0005988779632199545,
      "loss": 2.6282,
      "step": 6345
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.6543192863464355,
      "learning_rate": 0.0005988776097369053,
      "loss": 2.3622,
      "step": 6346
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.69572514295578,
      "learning_rate": 0.0005988772561982891,
      "loss": 2.5544,
      "step": 6347
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.6953793168067932,
      "learning_rate": 0.0005988769026041059,
      "loss": 2.3775,
      "step": 6348
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.6756442189216614,
      "learning_rate": 0.0005988765489543559,
      "loss": 2.464,
      "step": 6349
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.6701520681381226,
      "learning_rate": 0.0005988761952490391,
      "loss": 2.9296,
      "step": 6350
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.7363554835319519,
      "learning_rate": 0.0005988758414881556,
      "loss": 2.4528,
      "step": 6351
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.6345621347427368,
      "learning_rate": 0.0005988754876717054,
      "loss": 2.4382,
      "step": 6352
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.6844720840454102,
      "learning_rate": 0.0005988751337996887,
      "loss": 2.4158,
      "step": 6353
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.9097076654434204,
      "learning_rate": 0.0005988747798721054,
      "loss": 2.8069,
      "step": 6354
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.657168447971344,
      "learning_rate": 0.0005988744258889556,
      "loss": 2.3846,
      "step": 6355
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.7165631055831909,
      "learning_rate": 0.0005988740718502395,
      "loss": 2.6463,
      "step": 6356
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.68097323179245,
      "learning_rate": 0.000598873717755957,
      "loss": 2.5292,
      "step": 6357
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.6539563536643982,
      "learning_rate": 0.0005988733636061082,
      "loss": 2.7959,
      "step": 6358
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.705646812915802,
      "learning_rate": 0.0005988730094006933,
      "loss": 2.5995,
      "step": 6359
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.633435070514679,
      "learning_rate": 0.0005988726551397122,
      "loss": 2.6191,
      "step": 6360
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.7344061732292175,
      "learning_rate": 0.0005988723008231651,
      "loss": 2.6798,
      "step": 6361
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.7432253956794739,
      "learning_rate": 0.0005988719464510519,
      "loss": 2.5494,
      "step": 6362
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.6340922117233276,
      "learning_rate": 0.0005988715920233728,
      "loss": 2.5665,
      "step": 6363
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.6694408655166626,
      "learning_rate": 0.0005988712375401278,
      "loss": 2.5279,
      "step": 6364
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.8059672713279724,
      "learning_rate": 0.0005988708830013171,
      "loss": 2.5787,
      "step": 6365
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.6957392692565918,
      "learning_rate": 0.0005988705284069406,
      "loss": 2.5556,
      "step": 6366
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.7228646278381348,
      "learning_rate": 0.0005988701737569984,
      "loss": 2.6387,
      "step": 6367
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.8082929253578186,
      "learning_rate": 0.0005988698190514906,
      "loss": 2.6735,
      "step": 6368
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.7560129761695862,
      "learning_rate": 0.0005988694642904173,
      "loss": 2.3424,
      "step": 6369
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.6352753639221191,
      "learning_rate": 0.0005988691094737785,
      "loss": 2.6584,
      "step": 6370
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.8022615313529968,
      "learning_rate": 0.0005988687546015743,
      "loss": 2.4339,
      "step": 6371
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.6442965865135193,
      "learning_rate": 0.0005988683996738047,
      "loss": 2.4313,
      "step": 6372
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.6693183779716492,
      "learning_rate": 0.0005988680446904699,
      "loss": 2.5625,
      "step": 6373
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.6736172437667847,
      "learning_rate": 0.0005988676896515698,
      "loss": 2.446,
      "step": 6374
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.6637088656425476,
      "learning_rate": 0.0005988673345571047,
      "loss": 2.7844,
      "step": 6375
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.6446560621261597,
      "learning_rate": 0.0005988669794070743,
      "loss": 2.5957,
      "step": 6376
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.6502614617347717,
      "learning_rate": 0.0005988666242014791,
      "loss": 2.8349,
      "step": 6377
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.661969780921936,
      "learning_rate": 0.0005988662689403189,
      "loss": 2.5443,
      "step": 6378
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.6487600803375244,
      "learning_rate": 0.0005988659136235937,
      "loss": 2.4594,
      "step": 6379
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.6477644443511963,
      "learning_rate": 0.0005988655582513039,
      "loss": 2.5357,
      "step": 6380
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.6638404726982117,
      "learning_rate": 0.0005988652028234492,
      "loss": 2.6609,
      "step": 6381
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.7042891979217529,
      "learning_rate": 0.0005988648473400298,
      "loss": 2.5674,
      "step": 6382
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.6912496089935303,
      "learning_rate": 0.0005988644918010458,
      "loss": 2.5708,
      "step": 6383
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.6494086980819702,
      "learning_rate": 0.0005988641362064972,
      "loss": 2.7822,
      "step": 6384
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.6439980268478394,
      "learning_rate": 0.0005988637805563843,
      "loss": 2.4212,
      "step": 6385
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.6514784693717957,
      "learning_rate": 0.0005988634248507068,
      "loss": 2.6682,
      "step": 6386
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.705540657043457,
      "learning_rate": 0.0005988630690894651,
      "loss": 2.79,
      "step": 6387
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.6469050049781799,
      "learning_rate": 0.0005988627132726591,
      "loss": 2.8157,
      "step": 6388
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.7710680365562439,
      "learning_rate": 0.0005988623574002888,
      "loss": 2.538,
      "step": 6389
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.599474310874939,
      "learning_rate": 0.0005988620014723543,
      "loss": 2.4821,
      "step": 6390
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.6088318228721619,
      "learning_rate": 0.0005988616454888557,
      "loss": 2.4661,
      "step": 6391
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.7731965780258179,
      "learning_rate": 0.0005988612894497932,
      "loss": 2.5499,
      "step": 6392
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.8859328031539917,
      "learning_rate": 0.0005988609333551667,
      "loss": 2.3881,
      "step": 6393
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.6585298180580139,
      "learning_rate": 0.0005988605772049763,
      "loss": 2.56,
      "step": 6394
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.0206924676895142,
      "learning_rate": 0.0005988602209992222,
      "loss": 2.7022,
      "step": 6395
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.764405369758606,
      "learning_rate": 0.0005988598647379042,
      "loss": 2.5568,
      "step": 6396
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.6673586368560791,
      "learning_rate": 0.0005988595084210225,
      "loss": 2.6073,
      "step": 6397
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.9707702398300171,
      "learning_rate": 0.0005988591520485773,
      "loss": 2.7347,
      "step": 6398
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.8019421696662903,
      "learning_rate": 0.0005988587956205684,
      "loss": 2.5482,
      "step": 6399
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.7108573317527771,
      "learning_rate": 0.000598858439136996,
      "loss": 2.3498,
      "step": 6400
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.9382683038711548,
      "learning_rate": 0.0005988580825978603,
      "loss": 2.5599,
      "step": 6401
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.6995742917060852,
      "learning_rate": 0.0005988577260031612,
      "loss": 2.3608,
      "step": 6402
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.6614416241645813,
      "learning_rate": 0.0005988573693528987,
      "loss": 2.3555,
      "step": 6403
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.6857154965400696,
      "learning_rate": 0.0005988570126470731,
      "loss": 2.3937,
      "step": 6404
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.6866230368614197,
      "learning_rate": 0.0005988566558856843,
      "loss": 2.425,
      "step": 6405
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.6271893978118896,
      "learning_rate": 0.0005988562990687324,
      "loss": 2.5263,
      "step": 6406
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.6180968880653381,
      "learning_rate": 0.0005988559421962174,
      "loss": 2.7322,
      "step": 6407
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.7438381314277649,
      "learning_rate": 0.0005988555852681396,
      "loss": 2.4477,
      "step": 6408
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.658460259437561,
      "learning_rate": 0.0005988552282844989,
      "loss": 2.6483,
      "step": 6409
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.6931073069572449,
      "learning_rate": 0.0005988548712452953,
      "loss": 2.7067,
      "step": 6410
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.7233734130859375,
      "learning_rate": 0.0005988545141505289,
      "loss": 2.6833,
      "step": 6411
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.7152987718582153,
      "learning_rate": 0.0005988541570001999,
      "loss": 2.7316,
      "step": 6412
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.7262439131736755,
      "learning_rate": 0.0005988537997943081,
      "loss": 2.6799,
      "step": 6413
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.698441207408905,
      "learning_rate": 0.0005988534425328539,
      "loss": 2.4173,
      "step": 6414
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.6715994477272034,
      "learning_rate": 0.0005988530852158371,
      "loss": 2.5677,
      "step": 6415
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.692310094833374,
      "learning_rate": 0.000598852727843258,
      "loss": 2.5333,
      "step": 6416
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.7569568157196045,
      "learning_rate": 0.0005988523704151164,
      "loss": 2.5272,
      "step": 6417
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.6654061079025269,
      "learning_rate": 0.0005988520129314126,
      "loss": 2.3953,
      "step": 6418
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.6862830519676208,
      "learning_rate": 0.0005988516553921465,
      "loss": 2.6896,
      "step": 6419
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.7184767723083496,
      "learning_rate": 0.0005988512977973182,
      "loss": 2.6303,
      "step": 6420
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.7055305242538452,
      "learning_rate": 0.000598850940146928,
      "loss": 2.5483,
      "step": 6421
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.6599329113960266,
      "learning_rate": 0.0005988505824409756,
      "loss": 2.6627,
      "step": 6422
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.7232977151870728,
      "learning_rate": 0.0005988502246794613,
      "loss": 2.5935,
      "step": 6423
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.7019034028053284,
      "learning_rate": 0.0005988498668623851,
      "loss": 2.5609,
      "step": 6424
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.6924713850021362,
      "learning_rate": 0.000598849508989747,
      "loss": 2.6833,
      "step": 6425
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.7352033853530884,
      "learning_rate": 0.0005988491510615472,
      "loss": 2.6795,
      "step": 6426
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.6519008278846741,
      "learning_rate": 0.0005988487930777858,
      "loss": 2.6109,
      "step": 6427
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.7236891388893127,
      "learning_rate": 0.0005988484350384626,
      "loss": 2.5679,
      "step": 6428
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.6883530020713806,
      "learning_rate": 0.0005988480769435778,
      "loss": 2.5358,
      "step": 6429
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.6680005788803101,
      "learning_rate": 0.0005988477187931316,
      "loss": 2.6717,
      "step": 6430
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.6373096108436584,
      "learning_rate": 0.000598847360587124,
      "loss": 2.5325,
      "step": 6431
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.6808003187179565,
      "learning_rate": 0.0005988470023255549,
      "loss": 2.4904,
      "step": 6432
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.6749096512794495,
      "learning_rate": 0.0005988466440084246,
      "loss": 2.6079,
      "step": 6433
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.6565945744514465,
      "learning_rate": 0.0005988462856357332,
      "loss": 2.4726,
      "step": 6434
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.6418814659118652,
      "learning_rate": 0.0005988459272074804,
      "loss": 2.7935,
      "step": 6435
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.6875126361846924,
      "learning_rate": 0.0005988455687236666,
      "loss": 2.8632,
      "step": 6436
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.6605817675590515,
      "learning_rate": 0.0005988452101842917,
      "loss": 2.5435,
      "step": 6437
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.6375101804733276,
      "learning_rate": 0.000598844851589356,
      "loss": 2.6641,
      "step": 6438
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.6385376453399658,
      "learning_rate": 0.0005988444929388593,
      "loss": 2.7404,
      "step": 6439
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.6532250642776489,
      "learning_rate": 0.0005988441342328017,
      "loss": 2.4733,
      "step": 6440
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.6729637980461121,
      "learning_rate": 0.0005988437754711834,
      "loss": 2.5024,
      "step": 6441
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.6755215525627136,
      "learning_rate": 0.0005988434166540044,
      "loss": 2.7003,
      "step": 6442
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.6376832127571106,
      "learning_rate": 0.0005988430577812649,
      "loss": 2.6044,
      "step": 6443
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.6276475191116333,
      "learning_rate": 0.0005988426988529646,
      "loss": 2.593,
      "step": 6444
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.6227744817733765,
      "learning_rate": 0.0005988423398691039,
      "loss": 2.5303,
      "step": 6445
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.6841514110565186,
      "learning_rate": 0.0005988419808296828,
      "loss": 2.5505,
      "step": 6446
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.6903507709503174,
      "learning_rate": 0.0005988416217347013,
      "loss": 2.5421,
      "step": 6447
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.6601580381393433,
      "learning_rate": 0.0005988412625841594,
      "loss": 2.7258,
      "step": 6448
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.6766355633735657,
      "learning_rate": 0.0005988409033780575,
      "loss": 2.7063,
      "step": 6449
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.7096068859100342,
      "learning_rate": 0.0005988405441163952,
      "loss": 2.4846,
      "step": 6450
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.7289333343505859,
      "learning_rate": 0.0005988401847991729,
      "loss": 2.4581,
      "step": 6451
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.8624585866928101,
      "learning_rate": 0.0005988398254263906,
      "loss": 2.6185,
      "step": 6452
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.6121822595596313,
      "learning_rate": 0.0005988394659980483,
      "loss": 2.5151,
      "step": 6453
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.6196499466896057,
      "learning_rate": 0.0005988391065141461,
      "loss": 2.6563,
      "step": 6454
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.7148838043212891,
      "learning_rate": 0.0005988387469746841,
      "loss": 2.6917,
      "step": 6455
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.6586765646934509,
      "learning_rate": 0.0005988383873796624,
      "loss": 2.4938,
      "step": 6456
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.7203084230422974,
      "learning_rate": 0.0005988380277290809,
      "loss": 2.4859,
      "step": 6457
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.7298551797866821,
      "learning_rate": 0.0005988376680229398,
      "loss": 2.5826,
      "step": 6458
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.6516056060791016,
      "learning_rate": 0.0005988373082612391,
      "loss": 2.7346,
      "step": 6459
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.7060106992721558,
      "learning_rate": 0.000598836948443979,
      "loss": 2.64,
      "step": 6460
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.718166708946228,
      "learning_rate": 0.0005988365885711594,
      "loss": 2.6066,
      "step": 6461
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.633537232875824,
      "learning_rate": 0.0005988362286427804,
      "loss": 2.5012,
      "step": 6462
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.6464781761169434,
      "learning_rate": 0.0005988358686588423,
      "loss": 2.4485,
      "step": 6463
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.6400372982025146,
      "learning_rate": 0.0005988355086193447,
      "loss": 2.548,
      "step": 6464
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.6237861514091492,
      "learning_rate": 0.0005988351485242881,
      "loss": 2.6124,
      "step": 6465
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.6382794380187988,
      "learning_rate": 0.0005988347883736724,
      "loss": 2.4823,
      "step": 6466
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.7594110369682312,
      "learning_rate": 0.0005988344281674976,
      "loss": 2.5023,
      "step": 6467
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.6763437986373901,
      "learning_rate": 0.0005988340679057638,
      "loss": 2.5898,
      "step": 6468
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.7078723311424255,
      "learning_rate": 0.0005988337075884712,
      "loss": 2.8469,
      "step": 6469
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.620124101638794,
      "learning_rate": 0.0005988333472156198,
      "loss": 2.3526,
      "step": 6470
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.9797224998474121,
      "learning_rate": 0.0005988329867872096,
      "loss": 2.4567,
      "step": 6471
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.6692917346954346,
      "learning_rate": 0.0005988326263032406,
      "loss": 2.6959,
      "step": 6472
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.665652334690094,
      "learning_rate": 0.0005988322657637131,
      "loss": 2.493,
      "step": 6473
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.7371056079864502,
      "learning_rate": 0.000598831905168627,
      "loss": 2.3996,
      "step": 6474
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.6342598795890808,
      "learning_rate": 0.0005988315445179824,
      "loss": 2.5482,
      "step": 6475
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.628425657749176,
      "learning_rate": 0.0005988311838117795,
      "loss": 2.4062,
      "step": 6476
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.7142809629440308,
      "learning_rate": 0.0005988308230500181,
      "loss": 2.5778,
      "step": 6477
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.6770504713058472,
      "learning_rate": 0.0005988304622326984,
      "loss": 2.623,
      "step": 6478
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.6497566103935242,
      "learning_rate": 0.0005988301013598206,
      "loss": 2.3572,
      "step": 6479
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.7663500905036926,
      "learning_rate": 0.0005988297404313844,
      "loss": 2.4957,
      "step": 6480
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.6941690444946289,
      "learning_rate": 0.0005988293794473903,
      "loss": 2.6032,
      "step": 6481
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.6358278393745422,
      "learning_rate": 0.0005988290184078382,
      "loss": 2.6196,
      "step": 6482
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.8295659422874451,
      "learning_rate": 0.0005988286573127282,
      "loss": 2.4974,
      "step": 6483
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.6545619368553162,
      "learning_rate": 0.0005988282961620602,
      "loss": 2.5591,
      "step": 6484
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.7152172923088074,
      "learning_rate": 0.0005988279349558343,
      "loss": 2.6143,
      "step": 6485
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.7156105041503906,
      "learning_rate": 0.0005988275736940508,
      "loss": 2.6904,
      "step": 6486
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.6821359395980835,
      "learning_rate": 0.0005988272123767096,
      "loss": 2.7546,
      "step": 6487
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.6920815706253052,
      "learning_rate": 0.0005988268510038107,
      "loss": 2.7346,
      "step": 6488
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.6296843886375427,
      "learning_rate": 0.0005988264895753542,
      "loss": 2.6294,
      "step": 6489
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.6756424307823181,
      "learning_rate": 0.0005988261280913403,
      "loss": 2.6206,
      "step": 6490
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.8854885101318359,
      "learning_rate": 0.000598825766551769,
      "loss": 2.7198,
      "step": 6491
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.6669773459434509,
      "learning_rate": 0.0005988254049566404,
      "loss": 2.6581,
      "step": 6492
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.6966168284416199,
      "learning_rate": 0.0005988250433059544,
      "loss": 2.7255,
      "step": 6493
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.7408192753791809,
      "learning_rate": 0.0005988246815997112,
      "loss": 2.7096,
      "step": 6494
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.6767886877059937,
      "learning_rate": 0.0005988243198379109,
      "loss": 2.6006,
      "step": 6495
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.623013436794281,
      "learning_rate": 0.0005988239580205535,
      "loss": 2.583,
      "step": 6496
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.752544641494751,
      "learning_rate": 0.000598823596147639,
      "loss": 2.441,
      "step": 6497
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.743438720703125,
      "learning_rate": 0.0005988232342191677,
      "loss": 2.6972,
      "step": 6498
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.6033698320388794,
      "learning_rate": 0.0005988228722351394,
      "loss": 2.5407,
      "step": 6499
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.6613069176673889,
      "learning_rate": 0.0005988225101955543,
      "loss": 2.5865,
      "step": 6500
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.6826101541519165,
      "learning_rate": 0.0005988221481004125,
      "loss": 2.6498,
      "step": 6501
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.7665496468544006,
      "learning_rate": 0.0005988217859497141,
      "loss": 2.8585,
      "step": 6502
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.6592845916748047,
      "learning_rate": 0.000598821423743459,
      "loss": 2.5186,
      "step": 6503
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.66612708568573,
      "learning_rate": 0.0005988210614816473,
      "loss": 2.7973,
      "step": 6504
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.7423421740531921,
      "learning_rate": 0.0005988206991642792,
      "loss": 2.3275,
      "step": 6505
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.7207111716270447,
      "learning_rate": 0.0005988203367913547,
      "loss": 2.2995,
      "step": 6506
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.6428468227386475,
      "learning_rate": 0.0005988199743628738,
      "loss": 2.4597,
      "step": 6507
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.7115676999092102,
      "learning_rate": 0.0005988196118788367,
      "loss": 2.6654,
      "step": 6508
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.623005211353302,
      "learning_rate": 0.0005988192493392434,
      "loss": 2.5178,
      "step": 6509
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.6121407747268677,
      "learning_rate": 0.0005988188867440938,
      "loss": 2.6935,
      "step": 6510
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.6702837347984314,
      "learning_rate": 0.0005988185240933882,
      "loss": 2.5223,
      "step": 6511
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.7219038009643555,
      "learning_rate": 0.0005988181613871267,
      "loss": 2.5339,
      "step": 6512
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.6848330497741699,
      "learning_rate": 0.0005988177986253092,
      "loss": 2.5024,
      "step": 6513
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.750412106513977,
      "learning_rate": 0.0005988174358079358,
      "loss": 2.2796,
      "step": 6514
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.665709912776947,
      "learning_rate": 0.0005988170729350067,
      "loss": 2.5258,
      "step": 6515
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.0828310251235962,
      "learning_rate": 0.0005988167100065217,
      "loss": 2.4968,
      "step": 6516
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.6175320148468018,
      "learning_rate": 0.0005988163470224812,
      "loss": 2.3469,
      "step": 6517
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.7818504571914673,
      "learning_rate": 0.0005988159839828851,
      "loss": 2.5425,
      "step": 6518
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.9422858357429504,
      "learning_rate": 0.0005988156208877333,
      "loss": 2.5227,
      "step": 6519
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.6083909869194031,
      "learning_rate": 0.0005988152577370262,
      "loss": 2.6277,
      "step": 6520
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.0152511596679688,
      "learning_rate": 0.0005988148945307636,
      "loss": 2.5165,
      "step": 6521
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.6182723641395569,
      "learning_rate": 0.0005988145312689456,
      "loss": 2.5042,
      "step": 6522
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.7082647085189819,
      "learning_rate": 0.0005988141679515726,
      "loss": 2.5482,
      "step": 6523
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.7002124190330505,
      "learning_rate": 0.0005988138045786442,
      "loss": 2.6245,
      "step": 6524
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.601540744304657,
      "learning_rate": 0.0005988134411501607,
      "loss": 2.4576,
      "step": 6525
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.644865095615387,
      "learning_rate": 0.0005988130776661222,
      "loss": 2.7006,
      "step": 6526
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.7131689786911011,
      "learning_rate": 0.0005988127141265286,
      "loss": 2.5499,
      "step": 6527
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.6972891092300415,
      "learning_rate": 0.0005988123505313802,
      "loss": 2.65,
      "step": 6528
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6687437891960144,
      "learning_rate": 0.0005988119868806769,
      "loss": 2.634,
      "step": 6529
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6306005120277405,
      "learning_rate": 0.0005988116231744188,
      "loss": 2.3605,
      "step": 6530
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.878837525844574,
      "learning_rate": 0.000598811259412606,
      "loss": 2.4058,
      "step": 6531
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.7489515542984009,
      "learning_rate": 0.0005988108955952386,
      "loss": 2.4083,
      "step": 6532
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.7106841802597046,
      "learning_rate": 0.0005988105317223164,
      "loss": 2.7177,
      "step": 6533
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.7296226024627686,
      "learning_rate": 0.0005988101677938399,
      "loss": 2.5121,
      "step": 6534
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6526787281036377,
      "learning_rate": 0.0005988098038098088,
      "loss": 2.4807,
      "step": 6535
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6716052293777466,
      "learning_rate": 0.0005988094397702234,
      "loss": 2.5225,
      "step": 6536
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6549843549728394,
      "learning_rate": 0.0005988090756750838,
      "loss": 2.3338,
      "step": 6537
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6865999102592468,
      "learning_rate": 0.0005988087115243897,
      "loss": 2.6461,
      "step": 6538
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6744703650474548,
      "learning_rate": 0.0005988083473181416,
      "loss": 2.7307,
      "step": 6539
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.7320957183837891,
      "learning_rate": 0.0005988079830563394,
      "loss": 2.6095,
      "step": 6540
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6531369090080261,
      "learning_rate": 0.0005988076187389829,
      "loss": 2.6516,
      "step": 6541
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.5917648673057556,
      "learning_rate": 0.0005988072543660727,
      "loss": 2.5908,
      "step": 6542
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.7412727475166321,
      "learning_rate": 0.0005988068899376085,
      "loss": 2.5283,
      "step": 6543
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.7250657677650452,
      "learning_rate": 0.0005988065254535905,
      "loss": 2.4216,
      "step": 6544
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6145248413085938,
      "learning_rate": 0.0005988061609140187,
      "loss": 2.6319,
      "step": 6545
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6873878240585327,
      "learning_rate": 0.0005988057963188932,
      "loss": 2.639,
      "step": 6546
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6808138489723206,
      "learning_rate": 0.000598805431668214,
      "loss": 2.5001,
      "step": 6547
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.700861930847168,
      "learning_rate": 0.0005988050669619813,
      "loss": 2.6948,
      "step": 6548
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.7141650319099426,
      "learning_rate": 0.000598804702200195,
      "loss": 2.5625,
      "step": 6549
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.7018899321556091,
      "learning_rate": 0.0005988043373828555,
      "loss": 2.7052,
      "step": 6550
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6787887215614319,
      "learning_rate": 0.0005988039725099624,
      "loss": 2.4608,
      "step": 6551
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.7746075391769409,
      "learning_rate": 0.0005988036075815161,
      "loss": 2.5517,
      "step": 6552
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.687537670135498,
      "learning_rate": 0.0005988032425975165,
      "loss": 2.8195,
      "step": 6553
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6399556398391724,
      "learning_rate": 0.0005988028775579637,
      "loss": 2.6457,
      "step": 6554
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.7464374303817749,
      "learning_rate": 0.000598802512462858,
      "loss": 2.5544,
      "step": 6555
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6465396881103516,
      "learning_rate": 0.0005988021473121992,
      "loss": 2.6185,
      "step": 6556
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.7145289778709412,
      "learning_rate": 0.0005988017821059873,
      "loss": 2.4509,
      "step": 6557
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6629137992858887,
      "learning_rate": 0.0005988014168442226,
      "loss": 2.73,
      "step": 6558
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.7060086131095886,
      "learning_rate": 0.0005988010515269051,
      "loss": 2.6214,
      "step": 6559
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6501280665397644,
      "learning_rate": 0.0005988006861540349,
      "loss": 2.4317,
      "step": 6560
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.8868321776390076,
      "learning_rate": 0.0005988003207256119,
      "loss": 2.6477,
      "step": 6561
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6919039487838745,
      "learning_rate": 0.0005987999552416362,
      "loss": 2.4194,
      "step": 6562
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.736912727355957,
      "learning_rate": 0.0005987995897021081,
      "loss": 2.7931,
      "step": 6563
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6890503168106079,
      "learning_rate": 0.0005987992241070275,
      "loss": 2.6047,
      "step": 6564
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6792470216751099,
      "learning_rate": 0.0005987988584563945,
      "loss": 2.5647,
      "step": 6565
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.7222027778625488,
      "learning_rate": 0.0005987984927502091,
      "loss": 2.411,
      "step": 6566
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6847323179244995,
      "learning_rate": 0.0005987981269884715,
      "loss": 2.6434,
      "step": 6567
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6439813375473022,
      "learning_rate": 0.0005987977611711816,
      "loss": 2.6008,
      "step": 6568
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.7798797488212585,
      "learning_rate": 0.0005987973952983396,
      "loss": 2.5842,
      "step": 6569
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.7959632873535156,
      "learning_rate": 0.0005987970293699455,
      "loss": 2.4653,
      "step": 6570
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.7000179290771484,
      "learning_rate": 0.0005987966633859994,
      "loss": 2.6298,
      "step": 6571
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6783825159072876,
      "learning_rate": 0.0005987962973465013,
      "loss": 2.6076,
      "step": 6572
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.7858126163482666,
      "learning_rate": 0.0005987959312514514,
      "loss": 2.644,
      "step": 6573
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.789659321308136,
      "learning_rate": 0.0005987955651008497,
      "loss": 2.576,
      "step": 6574
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.7238166928291321,
      "learning_rate": 0.0005987951988946962,
      "loss": 2.5784,
      "step": 6575
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.9600117206573486,
      "learning_rate": 0.0005987948326329911,
      "loss": 2.6927,
      "step": 6576
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6424219012260437,
      "learning_rate": 0.0005987944663157344,
      "loss": 2.7112,
      "step": 6577
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.689289927482605,
      "learning_rate": 0.0005987940999429262,
      "loss": 2.5011,
      "step": 6578
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.8152804374694824,
      "learning_rate": 0.0005987937335145665,
      "loss": 2.6705,
      "step": 6579
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6590374708175659,
      "learning_rate": 0.0005987933670306554,
      "loss": 2.4356,
      "step": 6580
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6422719955444336,
      "learning_rate": 0.0005987930004911929,
      "loss": 2.5364,
      "step": 6581
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6251105070114136,
      "learning_rate": 0.0005987926338961792,
      "loss": 2.56,
      "step": 6582
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.8060106635093689,
      "learning_rate": 0.0005987922672456143,
      "loss": 2.6844,
      "step": 6583
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6368354558944702,
      "learning_rate": 0.0005987919005394983,
      "loss": 2.5033,
      "step": 6584
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6594902873039246,
      "learning_rate": 0.0005987915337778313,
      "loss": 2.3643,
      "step": 6585
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.7050750255584717,
      "learning_rate": 0.0005987911669606132,
      "loss": 2.764,
      "step": 6586
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.7257915139198303,
      "learning_rate": 0.0005987908000878443,
      "loss": 2.7208,
      "step": 6587
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6587828397750854,
      "learning_rate": 0.0005987904331595244,
      "loss": 2.4036,
      "step": 6588
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.7609341740608215,
      "learning_rate": 0.0005987900661756538,
      "loss": 2.4132,
      "step": 6589
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.79201340675354,
      "learning_rate": 0.0005987896991362325,
      "loss": 2.5142,
      "step": 6590
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6899392008781433,
      "learning_rate": 0.0005987893320412604,
      "loss": 2.8674,
      "step": 6591
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.7889658808708191,
      "learning_rate": 0.000598788964890738,
      "loss": 2.5551,
      "step": 6592
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6945775747299194,
      "learning_rate": 0.0005987885976846649,
      "loss": 2.3163,
      "step": 6593
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.598126232624054,
      "learning_rate": 0.0005987882304230415,
      "loss": 2.5173,
      "step": 6594
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6525728106498718,
      "learning_rate": 0.0005987878631058676,
      "loss": 2.664,
      "step": 6595
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.8089690804481506,
      "learning_rate": 0.0005987874957331435,
      "loss": 2.6756,
      "step": 6596
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6539493799209595,
      "learning_rate": 0.000598787128304869,
      "loss": 2.6443,
      "step": 6597
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6141509413719177,
      "learning_rate": 0.0005987867608210445,
      "loss": 2.5771,
      "step": 6598
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.804589569568634,
      "learning_rate": 0.0005987863932816698,
      "loss": 2.4162,
      "step": 6599
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6801226735115051,
      "learning_rate": 0.0005987860256867451,
      "loss": 2.6733,
      "step": 6600
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.8105794191360474,
      "learning_rate": 0.0005987856580362706,
      "loss": 2.6142,
      "step": 6601
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6108754277229309,
      "learning_rate": 0.000598785290330246,
      "loss": 2.5138,
      "step": 6602
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.658818781375885,
      "learning_rate": 0.0005987849225686715,
      "loss": 2.359,
      "step": 6603
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6490204930305481,
      "learning_rate": 0.0005987845547515474,
      "loss": 2.7283,
      "step": 6604
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6557583808898926,
      "learning_rate": 0.0005987841868788736,
      "loss": 2.5342,
      "step": 6605
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.7618618011474609,
      "learning_rate": 0.0005987838189506502,
      "loss": 2.4254,
      "step": 6606
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6319320797920227,
      "learning_rate": 0.0005987834509668771,
      "loss": 2.368,
      "step": 6607
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.7094379663467407,
      "learning_rate": 0.0005987830829275548,
      "loss": 2.6887,
      "step": 6608
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.740026593208313,
      "learning_rate": 0.0005987827148326829,
      "loss": 2.4932,
      "step": 6609
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6537496447563171,
      "learning_rate": 0.0005987823466822616,
      "loss": 2.3805,
      "step": 6610
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.7057488560676575,
      "learning_rate": 0.000598781978476291,
      "loss": 2.4642,
      "step": 6611
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6754403114318848,
      "learning_rate": 0.0005987816102147713,
      "loss": 2.4854,
      "step": 6612
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6806546449661255,
      "learning_rate": 0.0005987812418977024,
      "loss": 2.527,
      "step": 6613
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.7575433850288391,
      "learning_rate": 0.0005987808735250846,
      "loss": 2.7928,
      "step": 6614
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6358469724655151,
      "learning_rate": 0.0005987805050969176,
      "loss": 2.4213,
      "step": 6615
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.715840220451355,
      "learning_rate": 0.0005987801366132018,
      "loss": 2.6701,
      "step": 6616
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.5846384763717651,
      "learning_rate": 0.0005987797680739369,
      "loss": 2.6154,
      "step": 6617
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6478593945503235,
      "learning_rate": 0.0005987793994791234,
      "loss": 2.6773,
      "step": 6618
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6378324031829834,
      "learning_rate": 0.0005987790308287611,
      "loss": 2.5815,
      "step": 6619
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6901824474334717,
      "learning_rate": 0.0005987786621228503,
      "loss": 2.5474,
      "step": 6620
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6380304098129272,
      "learning_rate": 0.0005987782933613907,
      "loss": 2.4747,
      "step": 6621
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.740236222743988,
      "learning_rate": 0.0005987779245443826,
      "loss": 2.6199,
      "step": 6622
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.7009729146957397,
      "learning_rate": 0.000598777555671826,
      "loss": 2.2435,
      "step": 6623
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6086700558662415,
      "learning_rate": 0.0005987771867437212,
      "loss": 2.4961,
      "step": 6624
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.7808884382247925,
      "learning_rate": 0.000598776817760068,
      "loss": 2.523,
      "step": 6625
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6841534376144409,
      "learning_rate": 0.0005987764487208665,
      "loss": 2.7006,
      "step": 6626
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6840918064117432,
      "learning_rate": 0.0005987760796261169,
      "loss": 2.6495,
      "step": 6627
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6717680096626282,
      "learning_rate": 0.0005987757104758192,
      "loss": 2.4117,
      "step": 6628
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.7704539895057678,
      "learning_rate": 0.0005987753412699733,
      "loss": 2.5905,
      "step": 6629
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.710115373134613,
      "learning_rate": 0.0005987749720085796,
      "loss": 2.6191,
      "step": 6630
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6903898119926453,
      "learning_rate": 0.0005987746026916379,
      "loss": 2.546,
      "step": 6631
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.7146002054214478,
      "learning_rate": 0.0005987742333191483,
      "loss": 2.4065,
      "step": 6632
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.7399858832359314,
      "learning_rate": 0.000598773863891111,
      "loss": 2.6047,
      "step": 6633
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.7413431406021118,
      "learning_rate": 0.000598773494407526,
      "loss": 2.7159,
      "step": 6634
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.7263767719268799,
      "learning_rate": 0.0005987731248683934,
      "loss": 2.5735,
      "step": 6635
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.8358427882194519,
      "learning_rate": 0.0005987727552737132,
      "loss": 2.7708,
      "step": 6636
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6794580817222595,
      "learning_rate": 0.0005987723856234855,
      "loss": 2.3426,
      "step": 6637
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.8396381139755249,
      "learning_rate": 0.0005987720159177104,
      "loss": 2.7817,
      "step": 6638
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6591757535934448,
      "learning_rate": 0.0005987716461563878,
      "loss": 2.6553,
      "step": 6639
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.7933304309844971,
      "learning_rate": 0.0005987712763395181,
      "loss": 2.6423,
      "step": 6640
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6312019228935242,
      "learning_rate": 0.0005987709064671011,
      "loss": 2.3659,
      "step": 6641
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.7090773582458496,
      "learning_rate": 0.000598770536539137,
      "loss": 2.606,
      "step": 6642
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.5940893888473511,
      "learning_rate": 0.0005987701665556258,
      "loss": 2.4894,
      "step": 6643
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.8796002864837646,
      "learning_rate": 0.0005987697965165676,
      "loss": 2.5455,
      "step": 6644
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.704186201095581,
      "learning_rate": 0.0005987694264219624,
      "loss": 2.5911,
      "step": 6645
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6322519183158875,
      "learning_rate": 0.0005987690562718103,
      "loss": 2.2214,
      "step": 6646
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.8305597901344299,
      "learning_rate": 0.0005987686860661115,
      "loss": 2.5607,
      "step": 6647
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.672227144241333,
      "learning_rate": 0.0005987683158048658,
      "loss": 2.6108,
      "step": 6648
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.7116603255271912,
      "learning_rate": 0.0005987679454880736,
      "loss": 2.7026,
      "step": 6649
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.7616502642631531,
      "learning_rate": 0.0005987675751157348,
      "loss": 2.4827,
      "step": 6650
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.8892295956611633,
      "learning_rate": 0.0005987672046878493,
      "loss": 2.6845,
      "step": 6651
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.629255473613739,
      "learning_rate": 0.0005987668342044175,
      "loss": 2.5478,
      "step": 6652
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.9466264843940735,
      "learning_rate": 0.0005987664636654392,
      "loss": 2.6864,
      "step": 6653
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.0450345277786255,
      "learning_rate": 0.0005987660930709147,
      "loss": 2.4554,
      "step": 6654
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.7190718650817871,
      "learning_rate": 0.0005987657224208438,
      "loss": 2.5467,
      "step": 6655
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.8432709574699402,
      "learning_rate": 0.0005987653517152268,
      "loss": 2.4571,
      "step": 6656
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.9764518141746521,
      "learning_rate": 0.0005987649809540636,
      "loss": 2.6328,
      "step": 6657
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.7527310848236084,
      "learning_rate": 0.0005987646101373545,
      "loss": 2.8735,
      "step": 6658
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.8973899483680725,
      "learning_rate": 0.0005987642392650993,
      "loss": 2.711,
      "step": 6659
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.0384193658828735,
      "learning_rate": 0.0005987638683372982,
      "loss": 2.315,
      "step": 6660
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6816759705543518,
      "learning_rate": 0.0005987634973539513,
      "loss": 2.742,
      "step": 6661
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.7105233073234558,
      "learning_rate": 0.0005987631263150586,
      "loss": 2.5121,
      "step": 6662
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.7954583168029785,
      "learning_rate": 0.0005987627552206202,
      "loss": 2.5925,
      "step": 6663
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.8179132342338562,
      "learning_rate": 0.0005987623840706362,
      "loss": 2.6899,
      "step": 6664
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6699665188789368,
      "learning_rate": 0.0005987620128651065,
      "loss": 2.462,
      "step": 6665
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.7028318643569946,
      "learning_rate": 0.0005987616416040314,
      "loss": 2.6692,
      "step": 6666
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.9028293490409851,
      "learning_rate": 0.0005987612702874109,
      "loss": 2.2803,
      "step": 6667
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.681368887424469,
      "learning_rate": 0.000598760898915245,
      "loss": 2.813,
      "step": 6668
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.7319366335868835,
      "learning_rate": 0.0005987605274875338,
      "loss": 2.6352,
      "step": 6669
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.7328131794929504,
      "learning_rate": 0.0005987601560042774,
      "loss": 2.6475,
      "step": 6670
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6641935110092163,
      "learning_rate": 0.0005987597844654759,
      "loss": 2.46,
      "step": 6671
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6980369687080383,
      "learning_rate": 0.0005987594128711292,
      "loss": 2.5722,
      "step": 6672
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6423850655555725,
      "learning_rate": 0.0005987590412212376,
      "loss": 2.6447,
      "step": 6673
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.7631447315216064,
      "learning_rate": 0.0005987586695158009,
      "loss": 2.4983,
      "step": 6674
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6799271702766418,
      "learning_rate": 0.0005987582977548194,
      "loss": 2.5365,
      "step": 6675
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.679436981678009,
      "learning_rate": 0.0005987579259382932,
      "loss": 2.4606,
      "step": 6676
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.7059690356254578,
      "learning_rate": 0.0005987575540662222,
      "loss": 2.5641,
      "step": 6677
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6717028021812439,
      "learning_rate": 0.0005987571821386064,
      "loss": 2.2882,
      "step": 6678
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6738067865371704,
      "learning_rate": 0.0005987568101554461,
      "loss": 2.4181,
      "step": 6679
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6800682544708252,
      "learning_rate": 0.0005987564381167413,
      "loss": 2.7345,
      "step": 6680
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6462404727935791,
      "learning_rate": 0.000598756066022492,
      "loss": 2.409,
      "step": 6681
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.7566583752632141,
      "learning_rate": 0.0005987556938726983,
      "loss": 2.4726,
      "step": 6682
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6488065719604492,
      "learning_rate": 0.0005987553216673604,
      "loss": 2.5782,
      "step": 6683
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6829491257667542,
      "learning_rate": 0.0005987549494064782,
      "loss": 2.671,
      "step": 6684
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6902022361755371,
      "learning_rate": 0.0005987545770900517,
      "loss": 2.6704,
      "step": 6685
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6362689733505249,
      "learning_rate": 0.0005987542047180811,
      "loss": 2.5975,
      "step": 6686
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6470922827720642,
      "learning_rate": 0.0005987538322905665,
      "loss": 2.4895,
      "step": 6687
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6265546083450317,
      "learning_rate": 0.000598753459807508,
      "loss": 2.5318,
      "step": 6688
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6421399712562561,
      "learning_rate": 0.0005987530872689055,
      "loss": 2.6774,
      "step": 6689
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6390974521636963,
      "learning_rate": 0.0005987527146747592,
      "loss": 2.4872,
      "step": 6690
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.645814836025238,
      "learning_rate": 0.000598752342025069,
      "loss": 2.7491,
      "step": 6691
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6106120347976685,
      "learning_rate": 0.0005987519693198352,
      "loss": 2.4066,
      "step": 6692
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6826182007789612,
      "learning_rate": 0.0005987515965590578,
      "loss": 2.6055,
      "step": 6693
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6832844614982605,
      "learning_rate": 0.0005987512237427368,
      "loss": 2.4698,
      "step": 6694
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.7841898798942566,
      "learning_rate": 0.0005987508508708723,
      "loss": 2.5267,
      "step": 6695
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6186678409576416,
      "learning_rate": 0.0005987504779434644,
      "loss": 2.6824,
      "step": 6696
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.7148407101631165,
      "learning_rate": 0.0005987501049605133,
      "loss": 2.5184,
      "step": 6697
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.7254738211631775,
      "learning_rate": 0.0005987497319220186,
      "loss": 2.6595,
      "step": 6698
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.653899073600769,
      "learning_rate": 0.000598749358827981,
      "loss": 2.7113,
      "step": 6699
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6606581211090088,
      "learning_rate": 0.0005987489856784001,
      "loss": 2.6887,
      "step": 6700
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.7100149989128113,
      "learning_rate": 0.0005987486124732761,
      "loss": 2.6309,
      "step": 6701
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6194994449615479,
      "learning_rate": 0.0005987482392126091,
      "loss": 2.6037,
      "step": 6702
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6582846641540527,
      "learning_rate": 0.0005987478658963991,
      "loss": 2.418,
      "step": 6703
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.8315011858940125,
      "learning_rate": 0.0005987474925246465,
      "loss": 2.5083,
      "step": 6704
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6478091478347778,
      "learning_rate": 0.0005987471190973507,
      "loss": 2.3681,
      "step": 6705
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.7763298749923706,
      "learning_rate": 0.0005987467456145126,
      "loss": 2.2302,
      "step": 6706
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.694487988948822,
      "learning_rate": 0.0005987463720761315,
      "loss": 2.3054,
      "step": 6707
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.5939950942993164,
      "learning_rate": 0.000598745998482208,
      "loss": 2.3406,
      "step": 6708
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.7510960102081299,
      "learning_rate": 0.0005987456248327419,
      "loss": 2.4418,
      "step": 6709
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.7711630463600159,
      "learning_rate": 0.0005987452511277334,
      "loss": 2.7531,
      "step": 6710
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.7116421461105347,
      "learning_rate": 0.0005987448773671825,
      "loss": 2.4803,
      "step": 6711
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.685113787651062,
      "learning_rate": 0.0005987445035510892,
      "loss": 2.6103,
      "step": 6712
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6897238492965698,
      "learning_rate": 0.0005987441296794538,
      "loss": 2.4022,
      "step": 6713
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.7034593820571899,
      "learning_rate": 0.0005987437557522762,
      "loss": 2.4178,
      "step": 6714
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6818666458129883,
      "learning_rate": 0.0005987433817695564,
      "loss": 2.6642,
      "step": 6715
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.7134236693382263,
      "learning_rate": 0.0005987430077312947,
      "loss": 2.4857,
      "step": 6716
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6950457096099854,
      "learning_rate": 0.0005987426336374909,
      "loss": 2.6848,
      "step": 6717
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6786654591560364,
      "learning_rate": 0.0005987422594881453,
      "loss": 2.5784,
      "step": 6718
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.7275599837303162,
      "learning_rate": 0.0005987418852832578,
      "loss": 2.6182,
      "step": 6719
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6989513635635376,
      "learning_rate": 0.0005987415110228285,
      "loss": 2.5833,
      "step": 6720
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6104980111122131,
      "learning_rate": 0.0005987411367068577,
      "loss": 2.3779,
      "step": 6721
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.7244486808776855,
      "learning_rate": 0.0005987407623353452,
      "loss": 2.6744,
      "step": 6722
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.7095755934715271,
      "learning_rate": 0.0005987403879082912,
      "loss": 2.5699,
      "step": 6723
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.5903701186180115,
      "learning_rate": 0.0005987400134256956,
      "loss": 2.4612,
      "step": 6724
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6874696612358093,
      "learning_rate": 0.0005987396388875587,
      "loss": 2.5608,
      "step": 6725
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.7003858685493469,
      "learning_rate": 0.0005987392642938803,
      "loss": 2.5651,
      "step": 6726
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6453066468238831,
      "learning_rate": 0.0005987388896446608,
      "loss": 2.5982,
      "step": 6727
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.7532559037208557,
      "learning_rate": 0.0005987385149399,
      "loss": 2.7336,
      "step": 6728
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.7175261378288269,
      "learning_rate": 0.0005987381401795981,
      "loss": 2.7847,
      "step": 6729
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6744384169578552,
      "learning_rate": 0.0005987377653637552,
      "loss": 2.6267,
      "step": 6730
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.8639470934867859,
      "learning_rate": 0.0005987373904923713,
      "loss": 2.4852,
      "step": 6731
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6468212008476257,
      "learning_rate": 0.0005987370155654465,
      "loss": 2.5392,
      "step": 6732
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6498700976371765,
      "learning_rate": 0.0005987366405829808,
      "loss": 2.42,
      "step": 6733
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6668557524681091,
      "learning_rate": 0.0005987362655449742,
      "loss": 2.4853,
      "step": 6734
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6713988184928894,
      "learning_rate": 0.0005987358904514271,
      "loss": 2.8913,
      "step": 6735
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6981301307678223,
      "learning_rate": 0.0005987355153023393,
      "loss": 2.5911,
      "step": 6736
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6764158606529236,
      "learning_rate": 0.0005987351400977109,
      "loss": 2.6845,
      "step": 6737
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.7839989066123962,
      "learning_rate": 0.0005987347648375419,
      "loss": 2.4794,
      "step": 6738
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6370508670806885,
      "learning_rate": 0.0005987343895218326,
      "loss": 2.5808,
      "step": 6739
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.9377682209014893,
      "learning_rate": 0.0005987340141505829,
      "loss": 2.773,
      "step": 6740
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6232581734657288,
      "learning_rate": 0.0005987336387237929,
      "loss": 2.4846,
      "step": 6741
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.7968894243240356,
      "learning_rate": 0.0005987332632414626,
      "loss": 2.7531,
      "step": 6742
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.8520236015319824,
      "learning_rate": 0.0005987328877035923,
      "loss": 2.4439,
      "step": 6743
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.8095607757568359,
      "learning_rate": 0.0005987325121101818,
      "loss": 2.7073,
      "step": 6744
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6504279375076294,
      "learning_rate": 0.0005987321364612314,
      "loss": 2.2426,
      "step": 6745
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6456548571586609,
      "learning_rate": 0.0005987317607567409,
      "loss": 2.6407,
      "step": 6746
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.687545895576477,
      "learning_rate": 0.0005987313849967106,
      "loss": 2.5604,
      "step": 6747
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6381207704544067,
      "learning_rate": 0.0005987310091811406,
      "loss": 2.5173,
      "step": 6748
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6735348105430603,
      "learning_rate": 0.0005987306333100307,
      "loss": 2.6214,
      "step": 6749
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6179091930389404,
      "learning_rate": 0.0005987302573833813,
      "loss": 2.6702,
      "step": 6750
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6582729816436768,
      "learning_rate": 0.0005987298814011922,
      "loss": 2.7149,
      "step": 6751
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6947249174118042,
      "learning_rate": 0.0005987295053634636,
      "loss": 2.483,
      "step": 6752
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6779374480247498,
      "learning_rate": 0.0005987291292701956,
      "loss": 2.6668,
      "step": 6753
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.715731143951416,
      "learning_rate": 0.0005987287531213881,
      "loss": 2.5358,
      "step": 6754
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.8105621933937073,
      "learning_rate": 0.0005987283769170413,
      "loss": 2.506,
      "step": 6755
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6844267249107361,
      "learning_rate": 0.0005987280006571554,
      "loss": 2.6965,
      "step": 6756
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6530445218086243,
      "learning_rate": 0.0005987276243417302,
      "loss": 2.7241,
      "step": 6757
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.734046995639801,
      "learning_rate": 0.000598727247970766,
      "loss": 2.5277,
      "step": 6758
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.7002394199371338,
      "learning_rate": 0.0005987268715442626,
      "loss": 2.5502,
      "step": 6759
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.639766275882721,
      "learning_rate": 0.0005987264950622203,
      "loss": 2.5884,
      "step": 6760
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.7188604474067688,
      "learning_rate": 0.0005987261185246392,
      "loss": 2.6892,
      "step": 6761
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6885307431221008,
      "learning_rate": 0.0005987257419315192,
      "loss": 2.3397,
      "step": 6762
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.7762675881385803,
      "learning_rate": 0.0005987253652828605,
      "loss": 2.5551,
      "step": 6763
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6507958769798279,
      "learning_rate": 0.0005987249885786631,
      "loss": 2.676,
      "step": 6764
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6302632689476013,
      "learning_rate": 0.0005987246118189269,
      "loss": 2.4513,
      "step": 6765
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.7165173888206482,
      "learning_rate": 0.0005987242350036524,
      "loss": 2.6795,
      "step": 6766
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.7468830347061157,
      "learning_rate": 0.0005987238581328393,
      "loss": 2.6541,
      "step": 6767
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.7780815958976746,
      "learning_rate": 0.0005987234812064879,
      "loss": 2.6126,
      "step": 6768
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6954582929611206,
      "learning_rate": 0.000598723104224598,
      "loss": 2.3739,
      "step": 6769
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6822502613067627,
      "learning_rate": 0.00059872272718717,
      "loss": 2.6054,
      "step": 6770
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6794848442077637,
      "learning_rate": 0.0005987223500942037,
      "loss": 2.4467,
      "step": 6771
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6501448750495911,
      "learning_rate": 0.0005987219729456994,
      "loss": 2.6706,
      "step": 6772
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6110792756080627,
      "learning_rate": 0.0005987215957416569,
      "loss": 2.6982,
      "step": 6773
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6232825517654419,
      "learning_rate": 0.0005987212184820765,
      "loss": 2.5713,
      "step": 6774
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.7076258659362793,
      "learning_rate": 0.0005987208411669582,
      "loss": 2.5118,
      "step": 6775
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6994338035583496,
      "learning_rate": 0.000598720463796302,
      "loss": 2.4805,
      "step": 6776
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.65061354637146,
      "learning_rate": 0.0005987200863701081,
      "loss": 2.6753,
      "step": 6777
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.7957923412322998,
      "learning_rate": 0.0005987197088883763,
      "loss": 2.7582,
      "step": 6778
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6215919256210327,
      "learning_rate": 0.0005987193313511071,
      "loss": 2.5082,
      "step": 6779
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.7425833344459534,
      "learning_rate": 0.0005987189537583002,
      "loss": 2.456,
      "step": 6780
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6923490166664124,
      "learning_rate": 0.0005987185761099559,
      "loss": 2.4617,
      "step": 6781
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.665975034236908,
      "learning_rate": 0.0005987181984060742,
      "loss": 2.5473,
      "step": 6782
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6549155712127686,
      "learning_rate": 0.000598717820646655,
      "loss": 2.7041,
      "step": 6783
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.7006388902664185,
      "learning_rate": 0.0005987174428316986,
      "loss": 2.589,
      "step": 6784
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6327369809150696,
      "learning_rate": 0.0005987170649612051,
      "loss": 2.5807,
      "step": 6785
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.725439727306366,
      "learning_rate": 0.0005987166870351743,
      "loss": 2.5678,
      "step": 6786
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.7059317231178284,
      "learning_rate": 0.0005987163090536065,
      "loss": 2.6663,
      "step": 6787
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6183249354362488,
      "learning_rate": 0.0005987159310165017,
      "loss": 2.5941,
      "step": 6788
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.7068209648132324,
      "learning_rate": 0.0005987155529238599,
      "loss": 2.3756,
      "step": 6789
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.704618513584137,
      "learning_rate": 0.0005987151747756813,
      "loss": 2.3454,
      "step": 6790
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.703407347202301,
      "learning_rate": 0.0005987147965719659,
      "loss": 2.5718,
      "step": 6791
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.7368354797363281,
      "learning_rate": 0.0005987144183127138,
      "loss": 2.4706,
      "step": 6792
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6880860924720764,
      "learning_rate": 0.000598714039997925,
      "loss": 2.5947,
      "step": 6793
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6508181691169739,
      "learning_rate": 0.0005987136616275997,
      "loss": 2.5148,
      "step": 6794
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6556184887886047,
      "learning_rate": 0.0005987132832017378,
      "loss": 2.5443,
      "step": 6795
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6813347935676575,
      "learning_rate": 0.0005987129047203395,
      "loss": 2.5548,
      "step": 6796
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.640195906162262,
      "learning_rate": 0.0005987125261834048,
      "loss": 2.6861,
      "step": 6797
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6904531717300415,
      "learning_rate": 0.0005987121475909339,
      "loss": 2.8014,
      "step": 6798
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6498578786849976,
      "learning_rate": 0.0005987117689429267,
      "loss": 2.4267,
      "step": 6799
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6012564897537231,
      "learning_rate": 0.0005987113902393834,
      "loss": 2.6982,
      "step": 6800
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.5991025567054749,
      "learning_rate": 0.0005987110114803039,
      "loss": 2.5277,
      "step": 6801
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.693091869354248,
      "learning_rate": 0.0005987106326656885,
      "loss": 2.6225,
      "step": 6802
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6197702288627625,
      "learning_rate": 0.0005987102537955371,
      "loss": 2.463,
      "step": 6803
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6862390041351318,
      "learning_rate": 0.0005987098748698499,
      "loss": 2.7266,
      "step": 6804
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6195032596588135,
      "learning_rate": 0.0005987094958886268,
      "loss": 2.549,
      "step": 6805
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6247897744178772,
      "learning_rate": 0.0005987091168518679,
      "loss": 2.5474,
      "step": 6806
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.5999574065208435,
      "learning_rate": 0.0005987087377595735,
      "loss": 2.437,
      "step": 6807
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6206039190292358,
      "learning_rate": 0.0005987083586117435,
      "loss": 2.4493,
      "step": 6808
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.7085933685302734,
      "learning_rate": 0.0005987079794083778,
      "loss": 2.4872,
      "step": 6809
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.7117499709129333,
      "learning_rate": 0.0005987076001494769,
      "loss": 2.4687,
      "step": 6810
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6183232665061951,
      "learning_rate": 0.0005987072208350404,
      "loss": 2.3896,
      "step": 6811
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.8212470412254333,
      "learning_rate": 0.0005987068414650687,
      "loss": 2.5089,
      "step": 6812
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6490244269371033,
      "learning_rate": 0.0005987064620395617,
      "loss": 2.5306,
      "step": 6813
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.7623876333236694,
      "learning_rate": 0.0005987060825585197,
      "loss": 2.8172,
      "step": 6814
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.0938326120376587,
      "learning_rate": 0.0005987057030219424,
      "loss": 2.6603,
      "step": 6815
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6783682703971863,
      "learning_rate": 0.0005987053234298301,
      "loss": 2.6426,
      "step": 6816
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.8007262945175171,
      "learning_rate": 0.000598704943782183,
      "loss": 2.6582,
      "step": 6817
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6893534660339355,
      "learning_rate": 0.0005987045640790008,
      "loss": 2.6452,
      "step": 6818
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.8331369757652283,
      "learning_rate": 0.0005987041843202839,
      "loss": 2.8907,
      "step": 6819
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.7739068865776062,
      "learning_rate": 0.0005987038045060321,
      "loss": 2.4048,
      "step": 6820
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6544642448425293,
      "learning_rate": 0.0005987034246362459,
      "loss": 2.5264,
      "step": 6821
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6397749781608582,
      "learning_rate": 0.000598703044710925,
      "loss": 2.5014,
      "step": 6822
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.7389650940895081,
      "learning_rate": 0.0005987026647300694,
      "loss": 2.5367,
      "step": 6823
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.651417076587677,
      "learning_rate": 0.0005987022846936794,
      "loss": 2.448,
      "step": 6824
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6959090232849121,
      "learning_rate": 0.0005987019046017551,
      "loss": 2.4593,
      "step": 6825
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.5994636416435242,
      "learning_rate": 0.0005987015244542965,
      "loss": 2.6206,
      "step": 6826
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6283701062202454,
      "learning_rate": 0.0005987011442513035,
      "loss": 2.9428,
      "step": 6827
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.7115429639816284,
      "learning_rate": 0.0005987007639927765,
      "loss": 2.7071,
      "step": 6828
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6342819333076477,
      "learning_rate": 0.0005987003836787153,
      "loss": 2.5423,
      "step": 6829
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6122657656669617,
      "learning_rate": 0.00059870000330912,
      "loss": 2.5799,
      "step": 6830
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.8901893496513367,
      "learning_rate": 0.0005986996228839908,
      "loss": 2.4091,
      "step": 6831
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6194519400596619,
      "learning_rate": 0.0005986992424033276,
      "loss": 2.669,
      "step": 6832
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6541976928710938,
      "learning_rate": 0.0005986988618671307,
      "loss": 2.6591,
      "step": 6833
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.7010488510131836,
      "learning_rate": 0.0005986984812754,
      "loss": 2.3697,
      "step": 6834
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6607629656791687,
      "learning_rate": 0.0005986981006281356,
      "loss": 2.6934,
      "step": 6835
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6647776365280151,
      "learning_rate": 0.0005986977199253376,
      "loss": 2.8064,
      "step": 6836
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.7826334834098816,
      "learning_rate": 0.000598697339167006,
      "loss": 2.5636,
      "step": 6837
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6648828387260437,
      "learning_rate": 0.000598696958353141,
      "loss": 2.5399,
      "step": 6838
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.8210002183914185,
      "learning_rate": 0.0005986965774837426,
      "loss": 2.505,
      "step": 6839
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6518568396568298,
      "learning_rate": 0.0005986961965588109,
      "loss": 2.4598,
      "step": 6840
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.7481371164321899,
      "learning_rate": 0.0005986958155783459,
      "loss": 2.5687,
      "step": 6841
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6895465850830078,
      "learning_rate": 0.0005986954345423476,
      "loss": 2.5222,
      "step": 6842
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.7073577046394348,
      "learning_rate": 0.0005986950534508164,
      "loss": 2.7884,
      "step": 6843
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6378042101860046,
      "learning_rate": 0.0005986946723037519,
      "loss": 2.5715,
      "step": 6844
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6795982122421265,
      "learning_rate": 0.0005986942911011547,
      "loss": 2.5273,
      "step": 6845
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6347785592079163,
      "learning_rate": 0.0005986939098430243,
      "loss": 2.4509,
      "step": 6846
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6246832609176636,
      "learning_rate": 0.0005986935285293613,
      "loss": 2.6213,
      "step": 6847
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6212682127952576,
      "learning_rate": 0.0005986931471601654,
      "loss": 2.2391,
      "step": 6848
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6805112361907959,
      "learning_rate": 0.0005986927657354368,
      "loss": 2.6785,
      "step": 6849
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.5687932372093201,
      "learning_rate": 0.0005986923842551757,
      "loss": 2.5154,
      "step": 6850
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6157647371292114,
      "learning_rate": 0.000598692002719382,
      "loss": 2.5552,
      "step": 6851
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6572121381759644,
      "learning_rate": 0.0005986916211280557,
      "loss": 2.4439,
      "step": 6852
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6784630417823792,
      "learning_rate": 0.0005986912394811971,
      "loss": 2.6565,
      "step": 6853
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.673930823802948,
      "learning_rate": 0.0005986908577788061,
      "loss": 2.5628,
      "step": 6854
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6625487208366394,
      "learning_rate": 0.0005986904760208829,
      "loss": 2.7001,
      "step": 6855
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.7133857607841492,
      "learning_rate": 0.0005986900942074275,
      "loss": 2.4651,
      "step": 6856
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6234805583953857,
      "learning_rate": 0.0005986897123384399,
      "loss": 2.5721,
      "step": 6857
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6785643696784973,
      "learning_rate": 0.0005986893304139203,
      "loss": 2.5972,
      "step": 6858
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.7264705896377563,
      "learning_rate": 0.0005986889484338688,
      "loss": 2.3272,
      "step": 6859
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.7268667221069336,
      "learning_rate": 0.0005986885663982853,
      "loss": 2.5016,
      "step": 6860
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.7243980169296265,
      "learning_rate": 0.0005986881843071699,
      "loss": 2.4402,
      "step": 6861
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.7486449480056763,
      "learning_rate": 0.0005986878021605228,
      "loss": 2.3886,
      "step": 6862
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.8032441735267639,
      "learning_rate": 0.0005986874199583439,
      "loss": 2.3956,
      "step": 6863
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.7469072341918945,
      "learning_rate": 0.0005986870377006334,
      "loss": 2.4155,
      "step": 6864
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.8120637536048889,
      "learning_rate": 0.0005986866553873915,
      "loss": 2.5253,
      "step": 6865
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6823593378067017,
      "learning_rate": 0.000598686273018618,
      "loss": 2.471,
      "step": 6866
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.8540240526199341,
      "learning_rate": 0.000598685890594313,
      "loss": 2.558,
      "step": 6867
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.7304624319076538,
      "learning_rate": 0.0005986855081144768,
      "loss": 2.4434,
      "step": 6868
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.8071528077125549,
      "learning_rate": 0.0005986851255791092,
      "loss": 2.455,
      "step": 6869
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.7086806297302246,
      "learning_rate": 0.0005986847429882105,
      "loss": 2.6259,
      "step": 6870
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6123499274253845,
      "learning_rate": 0.0005986843603417807,
      "loss": 2.8051,
      "step": 6871
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6621555089950562,
      "learning_rate": 0.0005986839776398197,
      "loss": 2.6212,
      "step": 6872
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.767328143119812,
      "learning_rate": 0.0005986835948823278,
      "loss": 2.7486,
      "step": 6873
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.8303071856498718,
      "learning_rate": 0.0005986832120693049,
      "loss": 2.527,
      "step": 6874
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6710804104804993,
      "learning_rate": 0.0005986828292007513,
      "loss": 2.4107,
      "step": 6875
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.906762957572937,
      "learning_rate": 0.0005986824462766668,
      "loss": 2.7092,
      "step": 6876
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.0063172578811646,
      "learning_rate": 0.0005986820632970515,
      "loss": 2.4205,
      "step": 6877
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.7169694900512695,
      "learning_rate": 0.0005986816802619058,
      "loss": 2.6765,
      "step": 6878
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.2557919025421143,
      "learning_rate": 0.0005986812971712295,
      "loss": 2.5024,
      "step": 6879
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.0694209337234497,
      "learning_rate": 0.0005986809140250225,
      "loss": 2.4521,
      "step": 6880
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6635425090789795,
      "learning_rate": 0.0005986805308232853,
      "loss": 2.5662,
      "step": 6881
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.0237358808517456,
      "learning_rate": 0.0005986801475660177,
      "loss": 2.5508,
      "step": 6882
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.0985705852508545,
      "learning_rate": 0.0005986797642532198,
      "loss": 2.5669,
      "step": 6883
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6752952337265015,
      "learning_rate": 0.0005986793808848917,
      "loss": 2.5021,
      "step": 6884
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.715969979763031,
      "learning_rate": 0.0005986789974610334,
      "loss": 2.5543,
      "step": 6885
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.8159694075584412,
      "learning_rate": 0.0005986786139816451,
      "loss": 2.6474,
      "step": 6886
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.7358167767524719,
      "learning_rate": 0.0005986782304467269,
      "loss": 2.4549,
      "step": 6887
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6924190521240234,
      "learning_rate": 0.0005986778468562786,
      "loss": 2.4976,
      "step": 6888
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.7756954431533813,
      "learning_rate": 0.0005986774632103006,
      "loss": 2.6686,
      "step": 6889
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.760017454624176,
      "learning_rate": 0.0005986770795087927,
      "loss": 2.5348,
      "step": 6890
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6624282598495483,
      "learning_rate": 0.0005986766957517552,
      "loss": 2.6884,
      "step": 6891
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6242402195930481,
      "learning_rate": 0.000598676311939188,
      "loss": 2.8532,
      "step": 6892
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.660413384437561,
      "learning_rate": 0.0005986759280710912,
      "loss": 2.4108,
      "step": 6893
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.624488115310669,
      "learning_rate": 0.000598675544147465,
      "loss": 2.6778,
      "step": 6894
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.666179895401001,
      "learning_rate": 0.0005986751601683093,
      "loss": 2.6554,
      "step": 6895
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6663115620613098,
      "learning_rate": 0.0005986747761336243,
      "loss": 2.622,
      "step": 6896
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6912640333175659,
      "learning_rate": 0.00059867439204341,
      "loss": 2.5575,
      "step": 6897
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6448818445205688,
      "learning_rate": 0.0005986740078976665,
      "loss": 2.6407,
      "step": 6898
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6675575971603394,
      "learning_rate": 0.0005986736236963939,
      "loss": 2.454,
      "step": 6899
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6261221170425415,
      "learning_rate": 0.0005986732394395922,
      "loss": 2.4898,
      "step": 6900
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6980400085449219,
      "learning_rate": 0.0005986728551272614,
      "loss": 2.7126,
      "step": 6901
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.644031822681427,
      "learning_rate": 0.0005986724707594018,
      "loss": 2.5406,
      "step": 6902
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.7329493761062622,
      "learning_rate": 0.0005986720863360133,
      "loss": 2.613,
      "step": 6903
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6985816359519958,
      "learning_rate": 0.000598671701857096,
      "loss": 2.4125,
      "step": 6904
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6177298426628113,
      "learning_rate": 0.0005986713173226501,
      "loss": 2.3806,
      "step": 6905
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6707680821418762,
      "learning_rate": 0.0005986709327326755,
      "loss": 2.7634,
      "step": 6906
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6734897494316101,
      "learning_rate": 0.0005986705480871724,
      "loss": 2.406,
      "step": 6907
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6780497431755066,
      "learning_rate": 0.0005986701633861408,
      "loss": 2.601,
      "step": 6908
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6425063610076904,
      "learning_rate": 0.0005986697786295806,
      "loss": 2.6009,
      "step": 6909
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.8095137476921082,
      "learning_rate": 0.0005986693938174922,
      "loss": 2.6842,
      "step": 6910
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6333062052726746,
      "learning_rate": 0.0005986690089498755,
      "loss": 2.7751,
      "step": 6911
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6258404850959778,
      "learning_rate": 0.0005986686240267307,
      "loss": 2.2263,
      "step": 6912
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6558929085731506,
      "learning_rate": 0.0005986682390480576,
      "loss": 2.481,
      "step": 6913
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6996475458145142,
      "learning_rate": 0.0005986678540138565,
      "loss": 2.5006,
      "step": 6914
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6832035183906555,
      "learning_rate": 0.0005986674689241274,
      "loss": 2.6676,
      "step": 6915
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.686145007610321,
      "learning_rate": 0.0005986670837788703,
      "loss": 2.594,
      "step": 6916
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6911766529083252,
      "learning_rate": 0.0005986666985780855,
      "loss": 2.4425,
      "step": 6917
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.721432626247406,
      "learning_rate": 0.0005986663133217729,
      "loss": 2.6091,
      "step": 6918
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6358629465103149,
      "learning_rate": 0.0005986659280099325,
      "loss": 2.5315,
      "step": 6919
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6798738837242126,
      "learning_rate": 0.0005986655426425646,
      "loss": 2.599,
      "step": 6920
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.683559238910675,
      "learning_rate": 0.000598665157219669,
      "loss": 2.6953,
      "step": 6921
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.7139592170715332,
      "learning_rate": 0.0005986647717412459,
      "loss": 2.649,
      "step": 6922
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6741371154785156,
      "learning_rate": 0.0005986643862072956,
      "loss": 2.6849,
      "step": 6923
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6442165374755859,
      "learning_rate": 0.0005986640006178178,
      "loss": 2.5996,
      "step": 6924
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6021550893783569,
      "learning_rate": 0.0005986636149728126,
      "loss": 2.3141,
      "step": 6925
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6440457701683044,
      "learning_rate": 0.0005986632292722803,
      "loss": 2.5751,
      "step": 6926
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6460087299346924,
      "learning_rate": 0.000598662843516221,
      "loss": 2.4808,
      "step": 6927
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6456383466720581,
      "learning_rate": 0.0005986624577046345,
      "loss": 2.5015,
      "step": 6928
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6979944705963135,
      "learning_rate": 0.0005986620718375211,
      "loss": 2.6286,
      "step": 6929
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.7499094009399414,
      "learning_rate": 0.0005986616859148807,
      "loss": 2.47,
      "step": 6930
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6492599844932556,
      "learning_rate": 0.0005986612999367134,
      "loss": 2.4454,
      "step": 6931
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.7826225161552429,
      "learning_rate": 0.0005986609139030194,
      "loss": 2.4625,
      "step": 6932
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6565482020378113,
      "learning_rate": 0.0005986605278137987,
      "loss": 2.5521,
      "step": 6933
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6572495698928833,
      "learning_rate": 0.0005986601416690515,
      "loss": 2.4956,
      "step": 6934
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6971632242202759,
      "learning_rate": 0.0005986597554687776,
      "loss": 2.3805,
      "step": 6935
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.8749194741249084,
      "learning_rate": 0.0005986593692129772,
      "loss": 2.4189,
      "step": 6936
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6524379253387451,
      "learning_rate": 0.0005986589829016504,
      "loss": 2.4525,
      "step": 6937
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.8889933824539185,
      "learning_rate": 0.0005986585965347972,
      "loss": 2.6962,
      "step": 6938
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.7875267267227173,
      "learning_rate": 0.0005986582101124179,
      "loss": 2.5469,
      "step": 6939
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6197736263275146,
      "learning_rate": 0.0005986578236345123,
      "loss": 2.338,
      "step": 6940
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6683768630027771,
      "learning_rate": 0.0005986574371010806,
      "loss": 2.4039,
      "step": 6941
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.7513442039489746,
      "learning_rate": 0.0005986570505121229,
      "loss": 2.6221,
      "step": 6942
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6458134651184082,
      "learning_rate": 0.0005986566638676391,
      "loss": 2.6376,
      "step": 6943
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.682350218296051,
      "learning_rate": 0.0005986562771676293,
      "loss": 2.387,
      "step": 6944
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6589251160621643,
      "learning_rate": 0.0005986558904120939,
      "loss": 2.4537,
      "step": 6945
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.7078303098678589,
      "learning_rate": 0.0005986555036010325,
      "loss": 2.5952,
      "step": 6946
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.7478620409965515,
      "learning_rate": 0.0005986551167344456,
      "loss": 2.5407,
      "step": 6947
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.8310543894767761,
      "learning_rate": 0.0005986547298123331,
      "loss": 2.2803,
      "step": 6948
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.694227397441864,
      "learning_rate": 0.0005986543428346948,
      "loss": 2.4896,
      "step": 6949
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6618255376815796,
      "learning_rate": 0.0005986539558015311,
      "loss": 2.4604,
      "step": 6950
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.7922073602676392,
      "learning_rate": 0.0005986535687128422,
      "loss": 2.4699,
      "step": 6951
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.9406295418739319,
      "learning_rate": 0.0005986531815686277,
      "loss": 2.5789,
      "step": 6952
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6666076183319092,
      "learning_rate": 0.0005986527943688881,
      "loss": 2.3672,
      "step": 6953
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.8352940082550049,
      "learning_rate": 0.0005986524071136233,
      "loss": 2.5268,
      "step": 6954
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.7385973930358887,
      "learning_rate": 0.0005986520198028334,
      "loss": 2.5781,
      "step": 6955
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6886611580848694,
      "learning_rate": 0.0005986516324365184,
      "loss": 2.6352,
      "step": 6956
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.7907899022102356,
      "learning_rate": 0.0005986512450146784,
      "loss": 2.5845,
      "step": 6957
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.7275945544242859,
      "learning_rate": 0.0005986508575373136,
      "loss": 2.4446,
      "step": 6958
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6998640894889832,
      "learning_rate": 0.0005986504700044238,
      "loss": 2.5083,
      "step": 6959
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.7273959517478943,
      "learning_rate": 0.0005986500824160094,
      "loss": 2.6658,
      "step": 6960
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.7655557990074158,
      "learning_rate": 0.0005986496947720702,
      "loss": 2.5433,
      "step": 6961
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.9299278259277344,
      "learning_rate": 0.0005986493070726064,
      "loss": 2.6256,
      "step": 6962
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.669631838798523,
      "learning_rate": 0.000598648919317618,
      "loss": 2.5425,
      "step": 6963
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.7961668372154236,
      "learning_rate": 0.0005986485315071053,
      "loss": 2.4247,
      "step": 6964
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.630470335483551,
      "learning_rate": 0.0005986481436410681,
      "loss": 2.7828,
      "step": 6965
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.844557523727417,
      "learning_rate": 0.0005986477557195066,
      "loss": 2.7227,
      "step": 6966
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6338574886322021,
      "learning_rate": 0.0005986473677424208,
      "loss": 2.5787,
      "step": 6967
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6777654886245728,
      "learning_rate": 0.0005986469797098109,
      "loss": 2.5575,
      "step": 6968
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.7473551034927368,
      "learning_rate": 0.0005986465916216769,
      "loss": 2.5201,
      "step": 6969
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.839850127696991,
      "learning_rate": 0.0005986462034780187,
      "loss": 2.4475,
      "step": 6970
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6787565350532532,
      "learning_rate": 0.0005986458152788366,
      "loss": 2.7168,
      "step": 6971
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.7842231392860413,
      "learning_rate": 0.0005986454270241308,
      "loss": 2.7515,
      "step": 6972
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.7201688885688782,
      "learning_rate": 0.0005986450387139009,
      "loss": 2.5357,
      "step": 6973
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6143210530281067,
      "learning_rate": 0.0005986446503481474,
      "loss": 2.8101,
      "step": 6974
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6360851526260376,
      "learning_rate": 0.0005986442619268701,
      "loss": 2.5343,
      "step": 6975
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.674041748046875,
      "learning_rate": 0.0005986438734500694,
      "loss": 2.4794,
      "step": 6976
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.941049337387085,
      "learning_rate": 0.000598643484917745,
      "loss": 2.9343,
      "step": 6977
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6514365673065186,
      "learning_rate": 0.0005986430963298971,
      "loss": 2.5773,
      "step": 6978
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.925318717956543,
      "learning_rate": 0.0005986427076865258,
      "loss": 2.5351,
      "step": 6979
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.753154456615448,
      "learning_rate": 0.0005986423189876314,
      "loss": 2.5465,
      "step": 6980
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6836990714073181,
      "learning_rate": 0.0005986419302332136,
      "loss": 2.5318,
      "step": 6981
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.7391754984855652,
      "learning_rate": 0.0005986415414232728,
      "loss": 2.6028,
      "step": 6982
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6396753787994385,
      "learning_rate": 0.0005986411525578087,
      "loss": 2.6399,
      "step": 6983
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.634857714176178,
      "learning_rate": 0.0005986407636368216,
      "loss": 2.5972,
      "step": 6984
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.653419554233551,
      "learning_rate": 0.0005986403746603116,
      "loss": 2.6287,
      "step": 6985
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6626220345497131,
      "learning_rate": 0.0005986399856282786,
      "loss": 2.3845,
      "step": 6986
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6719115972518921,
      "learning_rate": 0.000598639596540723,
      "loss": 2.3047,
      "step": 6987
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6773099303245544,
      "learning_rate": 0.0005986392073976445,
      "loss": 2.5947,
      "step": 6988
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.5985430479049683,
      "learning_rate": 0.0005986388181990434,
      "loss": 2.4147,
      "step": 6989
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6650890111923218,
      "learning_rate": 0.0005986384289449197,
      "loss": 2.4298,
      "step": 6990
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6978152990341187,
      "learning_rate": 0.0005986380396352734,
      "loss": 2.2637,
      "step": 6991
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6708167791366577,
      "learning_rate": 0.0005986376502701048,
      "loss": 2.6074,
      "step": 6992
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6849944591522217,
      "learning_rate": 0.0005986372608494137,
      "loss": 2.5418,
      "step": 6993
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.8402639627456665,
      "learning_rate": 0.0005986368713732003,
      "loss": 2.5526,
      "step": 6994
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6408278942108154,
      "learning_rate": 0.0005986364818414648,
      "loss": 2.7388,
      "step": 6995
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6670267581939697,
      "learning_rate": 0.000598636092254207,
      "loss": 2.348,
      "step": 6996
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6700199842453003,
      "learning_rate": 0.0005986357026114272,
      "loss": 2.6505,
      "step": 6997
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6953592300415039,
      "learning_rate": 0.0005986353129131253,
      "loss": 2.5356,
      "step": 6998
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.680069625377655,
      "learning_rate": 0.0005986349231593015,
      "loss": 2.5688,
      "step": 6999
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.8047581911087036,
      "learning_rate": 0.0005986345333499559,
      "loss": 2.6699,
      "step": 7000
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6116594672203064,
      "learning_rate": 0.0005986341434850884,
      "loss": 2.3881,
      "step": 7001
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.5948764681816101,
      "learning_rate": 0.0005986337535646992,
      "loss": 2.7149,
      "step": 7002
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6554070711135864,
      "learning_rate": 0.0005986333635887884,
      "loss": 2.7345,
      "step": 7003
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6238910555839539,
      "learning_rate": 0.000598632973557356,
      "loss": 2.4516,
      "step": 7004
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6304721832275391,
      "learning_rate": 0.0005986325834704021,
      "loss": 2.7046,
      "step": 7005
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.7105631232261658,
      "learning_rate": 0.0005986321933279267,
      "loss": 2.4246,
      "step": 7006
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6652499437332153,
      "learning_rate": 0.0005986318031299299,
      "loss": 2.6059,
      "step": 7007
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6903533935546875,
      "learning_rate": 0.000598631412876412,
      "loss": 2.5248,
      "step": 7008
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.603339433670044,
      "learning_rate": 0.0005986310225673727,
      "loss": 2.5154,
      "step": 7009
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6215632557868958,
      "learning_rate": 0.0005986306322028124,
      "loss": 2.5543,
      "step": 7010
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.683566153049469,
      "learning_rate": 0.0005986302417827309,
      "loss": 2.5489,
      "step": 7011
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.7620987892150879,
      "learning_rate": 0.0005986298513071285,
      "loss": 2.1882,
      "step": 7012
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6556326746940613,
      "learning_rate": 0.0005986294607760051,
      "loss": 2.5295,
      "step": 7013
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6127854585647583,
      "learning_rate": 0.0005986290701893609,
      "loss": 2.3718,
      "step": 7014
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.7408567667007446,
      "learning_rate": 0.0005986286795471959,
      "loss": 2.6695,
      "step": 7015
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6205833554267883,
      "learning_rate": 0.0005986282888495102,
      "loss": 2.681,
      "step": 7016
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.7414019107818604,
      "learning_rate": 0.0005986278980963038,
      "loss": 2.5986,
      "step": 7017
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.8661572933197021,
      "learning_rate": 0.0005986275072875769,
      "loss": 2.6234,
      "step": 7018
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6512852907180786,
      "learning_rate": 0.0005986271164233295,
      "loss": 2.6726,
      "step": 7019
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.750415027141571,
      "learning_rate": 0.0005986267255035617,
      "loss": 2.4079,
      "step": 7020
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.7547986507415771,
      "learning_rate": 0.0005986263345282735,
      "loss": 2.5458,
      "step": 7021
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.906732439994812,
      "learning_rate": 0.0005986259434974651,
      "loss": 2.6503,
      "step": 7022
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6203466057777405,
      "learning_rate": 0.0005986255524111365,
      "loss": 2.4175,
      "step": 7023
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.770142138004303,
      "learning_rate": 0.0005986251612692876,
      "loss": 2.4379,
      "step": 7024
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.9458652138710022,
      "learning_rate": 0.0005986247700719188,
      "loss": 2.2364,
      "step": 7025
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6428539156913757,
      "learning_rate": 0.00059862437881903,
      "loss": 2.6842,
      "step": 7026
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.162822961807251,
      "learning_rate": 0.0005986239875106214,
      "loss": 2.7144,
      "step": 7027
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.7544260025024414,
      "learning_rate": 0.0005986235961466928,
      "loss": 2.4154,
      "step": 7028
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.7200869917869568,
      "learning_rate": 0.0005986232047272444,
      "loss": 2.5421,
      "step": 7029
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.716909646987915,
      "learning_rate": 0.0005986228132522765,
      "loss": 2.5284,
      "step": 7030
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.872238278388977,
      "learning_rate": 0.0005986224217217889,
      "loss": 2.5973,
      "step": 7031
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6627521514892578,
      "learning_rate": 0.0005986220301357817,
      "loss": 2.6271,
      "step": 7032
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.7159804701805115,
      "learning_rate": 0.0005986216384942551,
      "loss": 2.7627,
      "step": 7033
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.7524796724319458,
      "learning_rate": 0.000598621246797209,
      "loss": 2.5015,
      "step": 7034
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.9504100680351257,
      "learning_rate": 0.0005986208550446437,
      "loss": 2.4925,
      "step": 7035
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6961892247200012,
      "learning_rate": 0.000598620463236559,
      "loss": 2.3365,
      "step": 7036
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.7922385334968567,
      "learning_rate": 0.0005986200713729552,
      "loss": 2.5493,
      "step": 7037
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.0139682292938232,
      "learning_rate": 0.0005986196794538324,
      "loss": 2.487,
      "step": 7038
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.7706255912780762,
      "learning_rate": 0.0005986192874791904,
      "loss": 2.4337,
      "step": 7039
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6439138650894165,
      "learning_rate": 0.0005986188954490296,
      "loss": 2.5178,
      "step": 7040
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.8399945497512817,
      "learning_rate": 0.0005986185033633498,
      "loss": 2.5411,
      "step": 7041
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.7029410004615784,
      "learning_rate": 0.0005986181112221512,
      "loss": 2.6941,
      "step": 7042
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6498504877090454,
      "learning_rate": 0.0005986177190254338,
      "loss": 2.7601,
      "step": 7043
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6715859770774841,
      "learning_rate": 0.0005986173267731978,
      "loss": 2.3733,
      "step": 7044
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.7834171652793884,
      "learning_rate": 0.0005986169344654431,
      "loss": 2.3956,
      "step": 7045
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6463127136230469,
      "learning_rate": 0.0005986165421021701,
      "loss": 2.3653,
      "step": 7046
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.7595943212509155,
      "learning_rate": 0.0005986161496833784,
      "loss": 2.6778,
      "step": 7047
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6691707372665405,
      "learning_rate": 0.0005986157572090685,
      "loss": 2.5783,
      "step": 7048
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6985124349594116,
      "learning_rate": 0.0005986153646792402,
      "loss": 2.6438,
      "step": 7049
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6645815968513489,
      "learning_rate": 0.0005986149720938937,
      "loss": 2.7389,
      "step": 7050
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6334740519523621,
      "learning_rate": 0.000598614579453029,
      "loss": 2.4713,
      "step": 7051
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.8182273507118225,
      "learning_rate": 0.0005986141867566462,
      "loss": 2.3666,
      "step": 7052
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6677923202514648,
      "learning_rate": 0.0005986137940047454,
      "loss": 2.5074,
      "step": 7053
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6552205681800842,
      "learning_rate": 0.0005986134011973268,
      "loss": 2.603,
      "step": 7054
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.7334266901016235,
      "learning_rate": 0.00059861300833439,
      "loss": 2.6533,
      "step": 7055
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.7194545865058899,
      "learning_rate": 0.0005986126154159358,
      "loss": 2.5711,
      "step": 7056
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6444912552833557,
      "learning_rate": 0.0005986122224419636,
      "loss": 2.5304,
      "step": 7057
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6916228532791138,
      "learning_rate": 0.0005986118294124739,
      "loss": 2.3543,
      "step": 7058
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6708291172981262,
      "learning_rate": 0.0005986114363274665,
      "loss": 2.3972,
      "step": 7059
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6258187890052795,
      "learning_rate": 0.0005986110431869417,
      "loss": 2.5449,
      "step": 7060
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6536335349082947,
      "learning_rate": 0.0005986106499908995,
      "loss": 2.2977,
      "step": 7061
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6468694806098938,
      "learning_rate": 0.0005986102567393398,
      "loss": 2.5897,
      "step": 7062
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.7001070976257324,
      "learning_rate": 0.0005986098634322629,
      "loss": 2.5961,
      "step": 7063
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.597008228302002,
      "learning_rate": 0.0005986094700696688,
      "loss": 2.6049,
      "step": 7064
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.720004141330719,
      "learning_rate": 0.0005986090766515576,
      "loss": 2.5432,
      "step": 7065
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6337217688560486,
      "learning_rate": 0.0005986086831779293,
      "loss": 2.4613,
      "step": 7066
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.651444673538208,
      "learning_rate": 0.000598608289648784,
      "loss": 2.5347,
      "step": 7067
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.5891884565353394,
      "learning_rate": 0.0005986078960641217,
      "loss": 2.7713,
      "step": 7068
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6622249484062195,
      "learning_rate": 0.0005986075024239428,
      "loss": 2.7434,
      "step": 7069
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6443158984184265,
      "learning_rate": 0.0005986071087282468,
      "loss": 2.462,
      "step": 7070
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.7068333029747009,
      "learning_rate": 0.0005986067149770343,
      "loss": 2.5136,
      "step": 7071
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.5898246765136719,
      "learning_rate": 0.0005986063211703052,
      "loss": 2.4648,
      "step": 7072
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6632072925567627,
      "learning_rate": 0.0005986059273080594,
      "loss": 2.7389,
      "step": 7073
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6187496781349182,
      "learning_rate": 0.0005986055333902973,
      "loss": 2.8645,
      "step": 7074
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6603623032569885,
      "learning_rate": 0.0005986051394170186,
      "loss": 2.6185,
      "step": 7075
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6816752552986145,
      "learning_rate": 0.0005986047453882238,
      "loss": 2.6394,
      "step": 7076
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6145491600036621,
      "learning_rate": 0.0005986043513039126,
      "loss": 2.3913,
      "step": 7077
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.7091584801673889,
      "learning_rate": 0.0005986039571640851,
      "loss": 2.4183,
      "step": 7078
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.662742555141449,
      "learning_rate": 0.0005986035629687417,
      "loss": 2.5642,
      "step": 7079
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.7219324707984924,
      "learning_rate": 0.0005986031687178822,
      "loss": 2.7185,
      "step": 7080
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6870017647743225,
      "learning_rate": 0.0005986027744115067,
      "loss": 2.5774,
      "step": 7081
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6256429553031921,
      "learning_rate": 0.0005986023800496153,
      "loss": 2.8293,
      "step": 7082
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6580888032913208,
      "learning_rate": 0.000598601985632208,
      "loss": 2.4348,
      "step": 7083
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6413800120353699,
      "learning_rate": 0.0005986015911592852,
      "loss": 2.46,
      "step": 7084
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6322721838951111,
      "learning_rate": 0.0005986011966308465,
      "loss": 2.5981,
      "step": 7085
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.7531217932701111,
      "learning_rate": 0.0005986008020468924,
      "loss": 2.5665,
      "step": 7086
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6620010137557983,
      "learning_rate": 0.0005986004074074226,
      "loss": 2.5205,
      "step": 7087
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6356027722358704,
      "learning_rate": 0.0005986000127124374,
      "loss": 2.7011,
      "step": 7088
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.5954676866531372,
      "learning_rate": 0.0005985996179619368,
      "loss": 2.4474,
      "step": 7089
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6731224656105042,
      "learning_rate": 0.0005985992231559209,
      "loss": 2.7135,
      "step": 7090
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6491554975509644,
      "learning_rate": 0.0005985988282943899,
      "loss": 2.4995,
      "step": 7091
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6895579099655151,
      "learning_rate": 0.0005985984333773436,
      "loss": 2.6225,
      "step": 7092
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6878860592842102,
      "learning_rate": 0.0005985980384047823,
      "loss": 2.5607,
      "step": 7093
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6190419793128967,
      "learning_rate": 0.0005985976433767058,
      "loss": 2.5022,
      "step": 7094
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6374363303184509,
      "learning_rate": 0.0005985972482931146,
      "loss": 2.521,
      "step": 7095
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6441282629966736,
      "learning_rate": 0.0005985968531540083,
      "loss": 2.6556,
      "step": 7096
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6455327272415161,
      "learning_rate": 0.0005985964579593874,
      "loss": 2.5997,
      "step": 7097
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.7094036340713501,
      "learning_rate": 0.0005985960627092517,
      "loss": 2.5106,
      "step": 7098
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.5994331240653992,
      "learning_rate": 0.0005985956674036014,
      "loss": 2.4197,
      "step": 7099
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6628220677375793,
      "learning_rate": 0.0005985952720424365,
      "loss": 2.5808,
      "step": 7100
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6275524497032166,
      "learning_rate": 0.0005985948766257572,
      "loss": 2.5631,
      "step": 7101
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.641494870185852,
      "learning_rate": 0.0005985944811535633,
      "loss": 2.576,
      "step": 7102
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.7486250400543213,
      "learning_rate": 0.0005985940856258552,
      "loss": 2.6636,
      "step": 7103
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6371105909347534,
      "learning_rate": 0.0005985936900426326,
      "loss": 2.5203,
      "step": 7104
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6677248477935791,
      "learning_rate": 0.000598593294403896,
      "loss": 2.5062,
      "step": 7105
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6580598950386047,
      "learning_rate": 0.0005985928987096453,
      "loss": 2.5432,
      "step": 7106
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6388964056968689,
      "learning_rate": 0.0005985925029598805,
      "loss": 2.5123,
      "step": 7107
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6241241693496704,
      "learning_rate": 0.0005985921071546016,
      "loss": 2.7019,
      "step": 7108
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6250942349433899,
      "learning_rate": 0.0005985917112938089,
      "loss": 2.597,
      "step": 7109
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6552076935768127,
      "learning_rate": 0.0005985913153775023,
      "loss": 2.46,
      "step": 7110
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.631458580493927,
      "learning_rate": 0.000598590919405682,
      "loss": 2.6524,
      "step": 7111
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.669356644153595,
      "learning_rate": 0.000598590523378348,
      "loss": 2.612,
      "step": 7112
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6179443001747131,
      "learning_rate": 0.0005985901272955004,
      "loss": 2.4483,
      "step": 7113
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6157717108726501,
      "learning_rate": 0.0005985897311571392,
      "loss": 2.7361,
      "step": 7114
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.5846571326255798,
      "learning_rate": 0.0005985893349632645,
      "loss": 2.5034,
      "step": 7115
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.5801821947097778,
      "learning_rate": 0.0005985889387138765,
      "loss": 2.4306,
      "step": 7116
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6108521819114685,
      "learning_rate": 0.0005985885424089752,
      "loss": 2.5387,
      "step": 7117
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6926488280296326,
      "learning_rate": 0.0005985881460485606,
      "loss": 2.7102,
      "step": 7118
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6419256329536438,
      "learning_rate": 0.0005985877496326328,
      "loss": 2.406,
      "step": 7119
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6792081594467163,
      "learning_rate": 0.000598587353161192,
      "loss": 2.5721,
      "step": 7120
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6744948625564575,
      "learning_rate": 0.0005985869566342381,
      "loss": 2.4871,
      "step": 7121
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6421588063240051,
      "learning_rate": 0.0005985865600517713,
      "loss": 2.4385,
      "step": 7122
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.7138627767562866,
      "learning_rate": 0.0005985861634137915,
      "loss": 2.7087,
      "step": 7123
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6510140299797058,
      "learning_rate": 0.000598585766720299,
      "loss": 2.3255,
      "step": 7124
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6917756795883179,
      "learning_rate": 0.0005985853699712937,
      "loss": 2.4807,
      "step": 7125
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6563383936882019,
      "learning_rate": 0.0005985849731667758,
      "loss": 2.6186,
      "step": 7126
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.7025439143180847,
      "learning_rate": 0.0005985845763067454,
      "loss": 2.4129,
      "step": 7127
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6255115270614624,
      "learning_rate": 0.0005985841793912024,
      "loss": 2.5301,
      "step": 7128
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6791799664497375,
      "learning_rate": 0.000598583782420147,
      "loss": 2.8863,
      "step": 7129
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6432743668556213,
      "learning_rate": 0.0005985833853935791,
      "loss": 2.6659,
      "step": 7130
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.7008416652679443,
      "learning_rate": 0.000598582988311499,
      "loss": 2.104,
      "step": 7131
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6692689061164856,
      "learning_rate": 0.0005985825911739068,
      "loss": 2.4637,
      "step": 7132
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.5950590372085571,
      "learning_rate": 0.0005985821939808023,
      "loss": 2.4955,
      "step": 7133
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6421968340873718,
      "learning_rate": 0.0005985817967321858,
      "loss": 2.5294,
      "step": 7134
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6721810698509216,
      "learning_rate": 0.0005985813994280573,
      "loss": 2.8483,
      "step": 7135
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6022924780845642,
      "learning_rate": 0.0005985810020684169,
      "loss": 2.6274,
      "step": 7136
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.7229729890823364,
      "learning_rate": 0.0005985806046532647,
      "loss": 2.6314,
      "step": 7137
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6516574025154114,
      "learning_rate": 0.0005985802071826005,
      "loss": 2.3846,
      "step": 7138
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6178512573242188,
      "learning_rate": 0.0005985798096564248,
      "loss": 2.5155,
      "step": 7139
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6737584471702576,
      "learning_rate": 0.0005985794120747375,
      "loss": 2.4724,
      "step": 7140
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6668037176132202,
      "learning_rate": 0.0005985790144375386,
      "loss": 2.6433,
      "step": 7141
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6538628339767456,
      "learning_rate": 0.0005985786167448283,
      "loss": 2.494,
      "step": 7142
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6606382131576538,
      "learning_rate": 0.0005985782189966064,
      "loss": 2.6455,
      "step": 7143
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.768252968788147,
      "learning_rate": 0.0005985778211928733,
      "loss": 2.5813,
      "step": 7144
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6790339946746826,
      "learning_rate": 0.000598577423333629,
      "loss": 2.7873,
      "step": 7145
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6577230095863342,
      "learning_rate": 0.0005985770254188735,
      "loss": 2.4622,
      "step": 7146
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.5980417728424072,
      "learning_rate": 0.0005985766274486068,
      "loss": 2.6146,
      "step": 7147
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6779323220252991,
      "learning_rate": 0.0005985762294228291,
      "loss": 2.747,
      "step": 7148
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6787703037261963,
      "learning_rate": 0.0005985758313415405,
      "loss": 2.3837,
      "step": 7149
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.8053194284439087,
      "learning_rate": 0.000598575433204741,
      "loss": 2.6321,
      "step": 7150
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6506823301315308,
      "learning_rate": 0.0005985750350124307,
      "loss": 2.8985,
      "step": 7151
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6585272550582886,
      "learning_rate": 0.0005985746367646096,
      "loss": 2.4125,
      "step": 7152
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6780478954315186,
      "learning_rate": 0.0005985742384612779,
      "loss": 2.3521,
      "step": 7153
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.66163170337677,
      "learning_rate": 0.0005985738401024356,
      "loss": 2.3379,
      "step": 7154
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6300341486930847,
      "learning_rate": 0.0005985734416880828,
      "loss": 2.7034,
      "step": 7155
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.5913305878639221,
      "learning_rate": 0.0005985730432182196,
      "loss": 2.3759,
      "step": 7156
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6321676969528198,
      "learning_rate": 0.000598572644692846,
      "loss": 2.5305,
      "step": 7157
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.657500684261322,
      "learning_rate": 0.0005985722461119621,
      "loss": 2.5228,
      "step": 7158
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6747770309448242,
      "learning_rate": 0.0005985718474755679,
      "loss": 2.6437,
      "step": 7159
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.7526928186416626,
      "learning_rate": 0.0005985714487836637,
      "loss": 2.7065,
      "step": 7160
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.7106351256370544,
      "learning_rate": 0.0005985710500362493,
      "loss": 2.4817,
      "step": 7161
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.7847241759300232,
      "learning_rate": 0.000598570651233325,
      "loss": 2.5,
      "step": 7162
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.7182168960571289,
      "learning_rate": 0.0005985702523748907,
      "loss": 2.5148,
      "step": 7163
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.332865595817566,
      "learning_rate": 0.0005985698534609466,
      "loss": 2.755,
      "step": 7164
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.64719158411026,
      "learning_rate": 0.0005985694544914927,
      "loss": 2.725,
      "step": 7165
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.8287198543548584,
      "learning_rate": 0.0005985690554665291,
      "loss": 2.7105,
      "step": 7166
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.7141351103782654,
      "learning_rate": 0.0005985686563860559,
      "loss": 2.5645,
      "step": 7167
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6258645057678223,
      "learning_rate": 0.0005985682572500733,
      "loss": 2.6924,
      "step": 7168
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6504335403442383,
      "learning_rate": 0.000598567858058581,
      "loss": 2.4583,
      "step": 7169
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.7491286396980286,
      "learning_rate": 0.0005985674588115794,
      "loss": 2.473,
      "step": 7170
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.7711589932441711,
      "learning_rate": 0.0005985670595090685,
      "loss": 2.4157,
      "step": 7171
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6925682425498962,
      "learning_rate": 0.0005985666601510482,
      "loss": 2.7877,
      "step": 7172
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.7248415350914001,
      "learning_rate": 0.000598566260737519,
      "loss": 2.6353,
      "step": 7173
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.685623288154602,
      "learning_rate": 0.0005985658612684804,
      "loss": 2.5037,
      "step": 7174
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.5989860892295837,
      "learning_rate": 0.0005985654617439329,
      "loss": 2.6008,
      "step": 7175
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.8086990714073181,
      "learning_rate": 0.0005985650621638765,
      "loss": 2.4124,
      "step": 7176
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.7365355491638184,
      "learning_rate": 0.0005985646625283112,
      "loss": 2.3833,
      "step": 7177
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6027980446815491,
      "learning_rate": 0.000598564262837237,
      "loss": 2.576,
      "step": 7178
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6652407050132751,
      "learning_rate": 0.0005985638630906542,
      "loss": 2.7875,
      "step": 7179
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6566432118415833,
      "learning_rate": 0.0005985634632885626,
      "loss": 2.6522,
      "step": 7180
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.7189443707466125,
      "learning_rate": 0.0005985630634309625,
      "loss": 2.5731,
      "step": 7181
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6914544105529785,
      "learning_rate": 0.0005985626635178538,
      "loss": 2.6784,
      "step": 7182
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.7478069067001343,
      "learning_rate": 0.0005985622635492368,
      "loss": 2.3015,
      "step": 7183
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.7773191928863525,
      "learning_rate": 0.0005985618635251113,
      "loss": 2.6399,
      "step": 7184
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6602674126625061,
      "learning_rate": 0.0005985614634454776,
      "loss": 2.7757,
      "step": 7185
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6510822176933289,
      "learning_rate": 0.0005985610633103357,
      "loss": 2.7028,
      "step": 7186
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6941406726837158,
      "learning_rate": 0.0005985606631196856,
      "loss": 2.7297,
      "step": 7187
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6510118842124939,
      "learning_rate": 0.0005985602628735275,
      "loss": 2.3612,
      "step": 7188
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.738730251789093,
      "learning_rate": 0.0005985598625718614,
      "loss": 2.4196,
      "step": 7189
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6300956010818481,
      "learning_rate": 0.0005985594622146873,
      "loss": 2.5909,
      "step": 7190
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6716923117637634,
      "learning_rate": 0.0005985590618020055,
      "loss": 2.6608,
      "step": 7191
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.7062913775444031,
      "learning_rate": 0.0005985586613338158,
      "loss": 2.4399,
      "step": 7192
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6866273283958435,
      "learning_rate": 0.0005985582608101185,
      "loss": 2.5344,
      "step": 7193
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.7613223791122437,
      "learning_rate": 0.0005985578602309136,
      "loss": 2.6413,
      "step": 7194
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6672366857528687,
      "learning_rate": 0.0005985574595962012,
      "loss": 2.4821,
      "step": 7195
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6768895387649536,
      "learning_rate": 0.0005985570589059812,
      "loss": 2.7004,
      "step": 7196
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6137174367904663,
      "learning_rate": 0.0005985566581602538,
      "loss": 2.6415,
      "step": 7197
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6689003109931946,
      "learning_rate": 0.0005985562573590192,
      "loss": 2.7381,
      "step": 7198
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6212353110313416,
      "learning_rate": 0.0005985558565022773,
      "loss": 2.2944,
      "step": 7199
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6810935735702515,
      "learning_rate": 0.0005985554555900281,
      "loss": 2.4791,
      "step": 7200
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6885854005813599,
      "learning_rate": 0.000598555054622272,
      "loss": 2.4901,
      "step": 7201
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6980438828468323,
      "learning_rate": 0.0005985546535990088,
      "loss": 2.7159,
      "step": 7202
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6091655492782593,
      "learning_rate": 0.0005985542525202386,
      "loss": 2.4951,
      "step": 7203
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6080271601676941,
      "learning_rate": 0.0005985538513859617,
      "loss": 2.5932,
      "step": 7204
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6251929998397827,
      "learning_rate": 0.0005985534501961779,
      "loss": 2.3433,
      "step": 7205
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6827399730682373,
      "learning_rate": 0.0005985530489508874,
      "loss": 2.4891,
      "step": 7206
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6097628474235535,
      "learning_rate": 0.0005985526476500902,
      "loss": 2.6538,
      "step": 7207
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.678993821144104,
      "learning_rate": 0.0005985522462937865,
      "loss": 2.4191,
      "step": 7208
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6521329283714294,
      "learning_rate": 0.0005985518448819761,
      "loss": 2.5953,
      "step": 7209
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.7137944102287292,
      "learning_rate": 0.0005985514434146596,
      "loss": 2.6345,
      "step": 7210
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6038967370986938,
      "learning_rate": 0.0005985510418918365,
      "loss": 2.3238,
      "step": 7211
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.7759610414505005,
      "learning_rate": 0.0005985506403135072,
      "loss": 2.4109,
      "step": 7212
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6674982905387878,
      "learning_rate": 0.0005985502386796717,
      "loss": 2.3876,
      "step": 7213
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.7147035002708435,
      "learning_rate": 0.0005985498369903301,
      "loss": 2.7281,
      "step": 7214
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.9833343029022217,
      "learning_rate": 0.0005985494352454825,
      "loss": 2.6119,
      "step": 7215
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.7864058613777161,
      "learning_rate": 0.0005985490334451288,
      "loss": 2.6594,
      "step": 7216
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.907568097114563,
      "learning_rate": 0.0005985486315892693,
      "loss": 2.4841,
      "step": 7217
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6947165727615356,
      "learning_rate": 0.0005985482296779039,
      "loss": 2.4172,
      "step": 7218
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6326637268066406,
      "learning_rate": 0.0005985478277110328,
      "loss": 2.6849,
      "step": 7219
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6759470105171204,
      "learning_rate": 0.000598547425688656,
      "loss": 2.6157,
      "step": 7220
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.7073086500167847,
      "learning_rate": 0.0005985470236107737,
      "loss": 2.6434,
      "step": 7221
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6457802653312683,
      "learning_rate": 0.0005985466214773857,
      "loss": 2.5814,
      "step": 7222
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6500868201255798,
      "learning_rate": 0.0005985462192884924,
      "loss": 2.7039,
      "step": 7223
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6546863913536072,
      "learning_rate": 0.0005985458170440935,
      "loss": 2.6336,
      "step": 7224
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.8056735396385193,
      "learning_rate": 0.0005985454147441895,
      "loss": 2.476,
      "step": 7225
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6255847811698914,
      "learning_rate": 0.0005985450123887802,
      "loss": 2.4357,
      "step": 7226
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.7144750356674194,
      "learning_rate": 0.0005985446099778658,
      "loss": 2.5525,
      "step": 7227
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.7403424978256226,
      "learning_rate": 0.0005985442075114463,
      "loss": 2.5288,
      "step": 7228
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6794611215591431,
      "learning_rate": 0.0005985438049895218,
      "loss": 2.3696,
      "step": 7229
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6323621273040771,
      "learning_rate": 0.0005985434024120924,
      "loss": 2.5116,
      "step": 7230
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6280184984207153,
      "learning_rate": 0.000598542999779158,
      "loss": 2.6163,
      "step": 7231
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6592870354652405,
      "learning_rate": 0.0005985425970907189,
      "loss": 2.645,
      "step": 7232
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6416888236999512,
      "learning_rate": 0.0005985421943467751,
      "loss": 2.5305,
      "step": 7233
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6757579445838928,
      "learning_rate": 0.0005985417915473268,
      "loss": 2.4514,
      "step": 7234
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6905078291893005,
      "learning_rate": 0.0005985413886923737,
      "loss": 2.3587,
      "step": 7235
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6473538279533386,
      "learning_rate": 0.0005985409857819162,
      "loss": 2.6506,
      "step": 7236
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.7030317187309265,
      "learning_rate": 0.0005985405828159544,
      "loss": 2.5737,
      "step": 7237
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6955594420433044,
      "learning_rate": 0.0005985401797944882,
      "loss": 2.5849,
      "step": 7238
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6533071994781494,
      "learning_rate": 0.0005985397767175178,
      "loss": 2.4737,
      "step": 7239
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6377283334732056,
      "learning_rate": 0.0005985393735850431,
      "loss": 2.4825,
      "step": 7240
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.7470449209213257,
      "learning_rate": 0.0005985389703970644,
      "loss": 2.6651,
      "step": 7241
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6789727807044983,
      "learning_rate": 0.0005985385671535816,
      "loss": 2.4846,
      "step": 7242
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.7350828647613525,
      "learning_rate": 0.000598538163854595,
      "loss": 2.5002,
      "step": 7243
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.7134870290756226,
      "learning_rate": 0.0005985377605001044,
      "loss": 2.4891,
      "step": 7244
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6321868896484375,
      "learning_rate": 0.00059853735709011,
      "loss": 2.5758,
      "step": 7245
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.7292312383651733,
      "learning_rate": 0.0005985369536246119,
      "loss": 2.4364,
      "step": 7246
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.631573498249054,
      "learning_rate": 0.0005985365501036101,
      "loss": 2.3478,
      "step": 7247
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6139963269233704,
      "learning_rate": 0.0005985361465271048,
      "loss": 2.7636,
      "step": 7248
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6821208596229553,
      "learning_rate": 0.0005985357428950959,
      "loss": 2.5432,
      "step": 7249
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6575299501419067,
      "learning_rate": 0.0005985353392075835,
      "loss": 2.3903,
      "step": 7250
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6814353466033936,
      "learning_rate": 0.000598534935464568,
      "loss": 2.5012,
      "step": 7251
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.639218270778656,
      "learning_rate": 0.000598534531666049,
      "loss": 2.6656,
      "step": 7252
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.7805495262145996,
      "learning_rate": 0.0005985341278120269,
      "loss": 2.8409,
      "step": 7253
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6810937523841858,
      "learning_rate": 0.0005985337239025016,
      "loss": 2.5804,
      "step": 7254
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6855137348175049,
      "learning_rate": 0.0005985333199374733,
      "loss": 2.4733,
      "step": 7255
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.7417010068893433,
      "learning_rate": 0.0005985329159169419,
      "loss": 2.3286,
      "step": 7256
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.647867739200592,
      "learning_rate": 0.0005985325118409078,
      "loss": 2.6394,
      "step": 7257
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6813633441925049,
      "learning_rate": 0.0005985321077093708,
      "loss": 2.491,
      "step": 7258
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6703960299491882,
      "learning_rate": 0.000598531703522331,
      "loss": 2.6465,
      "step": 7259
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.8764626383781433,
      "learning_rate": 0.0005985312992797885,
      "loss": 2.5511,
      "step": 7260
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.7148352861404419,
      "learning_rate": 0.0005985308949817435,
      "loss": 2.6307,
      "step": 7261
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.7938817143440247,
      "learning_rate": 0.0005985304906281958,
      "loss": 2.7983,
      "step": 7262
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.7558965682983398,
      "learning_rate": 0.0005985300862191458,
      "loss": 2.4301,
      "step": 7263
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6751092076301575,
      "learning_rate": 0.0005985296817545933,
      "loss": 2.4662,
      "step": 7264
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.7468081116676331,
      "learning_rate": 0.0005985292772345386,
      "loss": 2.7769,
      "step": 7265
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.8275807499885559,
      "learning_rate": 0.0005985288726589816,
      "loss": 2.7892,
      "step": 7266
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6746152639389038,
      "learning_rate": 0.0005985284680279226,
      "loss": 2.6778,
      "step": 7267
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6273632645606995,
      "learning_rate": 0.0005985280633413614,
      "loss": 2.6299,
      "step": 7268
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.011634111404419,
      "learning_rate": 0.0005985276585992982,
      "loss": 2.7486,
      "step": 7269
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.5972205400466919,
      "learning_rate": 0.0005985272538017332,
      "loss": 2.4631,
      "step": 7270
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.7025418877601624,
      "learning_rate": 0.0005985268489486661,
      "loss": 2.7812,
      "step": 7271
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.66704261302948,
      "learning_rate": 0.0005985264440400974,
      "loss": 2.742,
      "step": 7272
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6757501363754272,
      "learning_rate": 0.000598526039076027,
      "loss": 2.5622,
      "step": 7273
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.610313892364502,
      "learning_rate": 0.0005985256340564548,
      "loss": 2.3969,
      "step": 7274
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.7142403721809387,
      "learning_rate": 0.0005985252289813812,
      "loss": 2.7307,
      "step": 7275
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6623069643974304,
      "learning_rate": 0.0005985248238508061,
      "loss": 2.6999,
      "step": 7276
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6395282745361328,
      "learning_rate": 0.0005985244186647296,
      "loss": 2.5444,
      "step": 7277
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6898958086967468,
      "learning_rate": 0.0005985240134231518,
      "loss": 2.6604,
      "step": 7278
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6715388894081116,
      "learning_rate": 0.0005985236081260727,
      "loss": 2.5304,
      "step": 7279
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6347158551216125,
      "learning_rate": 0.0005985232027734926,
      "loss": 2.5515,
      "step": 7280
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.7323878407478333,
      "learning_rate": 0.0005985227973654112,
      "loss": 2.5956,
      "step": 7281
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6353322863578796,
      "learning_rate": 0.0005985223919018288,
      "loss": 2.4029,
      "step": 7282
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6494109034538269,
      "learning_rate": 0.0005985219863827454,
      "loss": 2.4105,
      "step": 7283
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6484441757202148,
      "learning_rate": 0.0005985215808081613,
      "loss": 2.5627,
      "step": 7284
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6547915935516357,
      "learning_rate": 0.0005985211751780763,
      "loss": 2.6411,
      "step": 7285
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6880411505699158,
      "learning_rate": 0.0005985207694924906,
      "loss": 2.4419,
      "step": 7286
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6289594173431396,
      "learning_rate": 0.0005985203637514043,
      "loss": 2.4841,
      "step": 7287
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6861612796783447,
      "learning_rate": 0.0005985199579548173,
      "loss": 2.5548,
      "step": 7288
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6998611688613892,
      "learning_rate": 0.0005985195521027299,
      "loss": 2.5575,
      "step": 7289
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.7472342252731323,
      "learning_rate": 0.0005985191461951421,
      "loss": 2.5175,
      "step": 7290
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6286420822143555,
      "learning_rate": 0.0005985187402320539,
      "loss": 2.5487,
      "step": 7291
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6956576704978943,
      "learning_rate": 0.0005985183342134654,
      "loss": 2.3995,
      "step": 7292
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6632359027862549,
      "learning_rate": 0.0005985179281393768,
      "loss": 2.3921,
      "step": 7293
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6142668724060059,
      "learning_rate": 0.0005985175220097881,
      "loss": 2.3265,
      "step": 7294
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6532740592956543,
      "learning_rate": 0.0005985171158246992,
      "loss": 2.4319,
      "step": 7295
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6720100045204163,
      "learning_rate": 0.0005985167095841104,
      "loss": 2.3632,
      "step": 7296
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6605995893478394,
      "learning_rate": 0.0005985163032880218,
      "loss": 2.7403,
      "step": 7297
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.670099675655365,
      "learning_rate": 0.0005985158969364333,
      "loss": 2.4033,
      "step": 7298
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6763525009155273,
      "learning_rate": 0.0005985154905293451,
      "loss": 2.3684,
      "step": 7299
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6313521862030029,
      "learning_rate": 0.0005985150840667572,
      "loss": 2.6021,
      "step": 7300
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.7077224254608154,
      "learning_rate": 0.0005985146775486698,
      "loss": 2.487,
      "step": 7301
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6335045099258423,
      "learning_rate": 0.0005985142709750827,
      "loss": 2.5798,
      "step": 7302
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.7267382740974426,
      "learning_rate": 0.0005985138643459963,
      "loss": 2.6072,
      "step": 7303
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.709450364112854,
      "learning_rate": 0.0005985134576614106,
      "loss": 2.8224,
      "step": 7304
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6042000651359558,
      "learning_rate": 0.0005985130509213255,
      "loss": 2.6908,
      "step": 7305
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6621461510658264,
      "learning_rate": 0.0005985126441257412,
      "loss": 2.4951,
      "step": 7306
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6699138283729553,
      "learning_rate": 0.0005985122372746578,
      "loss": 2.6396,
      "step": 7307
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6595937013626099,
      "learning_rate": 0.0005985118303680753,
      "loss": 2.4814,
      "step": 7308
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6307039260864258,
      "learning_rate": 0.0005985114234059939,
      "loss": 2.5685,
      "step": 7309
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.5971055626869202,
      "learning_rate": 0.0005985110163884134,
      "loss": 2.4846,
      "step": 7310
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6511752605438232,
      "learning_rate": 0.0005985106093153342,
      "loss": 2.5132,
      "step": 7311
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6169007420539856,
      "learning_rate": 0.0005985102021867563,
      "loss": 2.5213,
      "step": 7312
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.7473872900009155,
      "learning_rate": 0.0005985097950026797,
      "loss": 2.6562,
      "step": 7313
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.7451711893081665,
      "learning_rate": 0.0005985093877631045,
      "loss": 2.3741,
      "step": 7314
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6398641467094421,
      "learning_rate": 0.0005985089804680306,
      "loss": 2.5115,
      "step": 7315
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6586062908172607,
      "learning_rate": 0.0005985085731174584,
      "loss": 2.3654,
      "step": 7316
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6762623190879822,
      "learning_rate": 0.0005985081657113878,
      "loss": 2.5444,
      "step": 7317
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6744644641876221,
      "learning_rate": 0.0005985077582498189,
      "loss": 2.5646,
      "step": 7318
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6526668071746826,
      "learning_rate": 0.0005985073507327518,
      "loss": 2.6526,
      "step": 7319
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6277316808700562,
      "learning_rate": 0.0005985069431601865,
      "loss": 2.6444,
      "step": 7320
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6453553438186646,
      "learning_rate": 0.000598506535532123,
      "loss": 2.6741,
      "step": 7321
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.621258556842804,
      "learning_rate": 0.0005985061278485618,
      "loss": 2.7023,
      "step": 7322
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6289711594581604,
      "learning_rate": 0.0005985057201095025,
      "loss": 2.5831,
      "step": 7323
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6626883745193481,
      "learning_rate": 0.0005985053123149453,
      "loss": 2.4116,
      "step": 7324
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6929317116737366,
      "learning_rate": 0.0005985049044648905,
      "loss": 2.5701,
      "step": 7325
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6508125066757202,
      "learning_rate": 0.0005985044965593378,
      "loss": 2.616,
      "step": 7326
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6631591320037842,
      "learning_rate": 0.0005985040885982876,
      "loss": 2.4339,
      "step": 7327
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.782131552696228,
      "learning_rate": 0.0005985036805817397,
      "loss": 2.4752,
      "step": 7328
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6728647947311401,
      "learning_rate": 0.0005985032725096945,
      "loss": 2.7083,
      "step": 7329
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.5862302184104919,
      "learning_rate": 0.0005985028643821519,
      "loss": 2.4334,
      "step": 7330
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6795133352279663,
      "learning_rate": 0.0005985024561991119,
      "loss": 2.7213,
      "step": 7331
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.9029554128646851,
      "learning_rate": 0.0005985020479605749,
      "loss": 2.6711,
      "step": 7332
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.71169114112854,
      "learning_rate": 0.0005985016396665404,
      "loss": 2.6517,
      "step": 7333
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.1511296033859253,
      "learning_rate": 0.000598501231317009,
      "loss": 2.3866,
      "step": 7334
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6345675587654114,
      "learning_rate": 0.0005985008229119805,
      "loss": 2.5432,
      "step": 7335
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.7247616052627563,
      "learning_rate": 0.0005985004144514551,
      "loss": 2.5517,
      "step": 7336
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.8833034038543701,
      "learning_rate": 0.0005985000059354329,
      "loss": 2.6864,
      "step": 7337
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6057475805282593,
      "learning_rate": 0.0005984995973639138,
      "loss": 2.2268,
      "step": 7338
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6995180249214172,
      "learning_rate": 0.0005984991887368981,
      "loss": 2.5416,
      "step": 7339
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6314951181411743,
      "learning_rate": 0.0005984987800543857,
      "loss": 2.543,
      "step": 7340
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.7115039229393005,
      "learning_rate": 0.0005984983713163767,
      "loss": 2.6928,
      "step": 7341
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6519601941108704,
      "learning_rate": 0.0005984979625228713,
      "loss": 2.8758,
      "step": 7342
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.5944079756736755,
      "learning_rate": 0.0005984975536738695,
      "loss": 2.7098,
      "step": 7343
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6409475207328796,
      "learning_rate": 0.0005984971447693712,
      "loss": 2.4582,
      "step": 7344
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6545538306236267,
      "learning_rate": 0.0005984967358093769,
      "loss": 2.6784,
      "step": 7345
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.723264753818512,
      "learning_rate": 0.0005984963267938863,
      "loss": 2.4813,
      "step": 7346
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6817567944526672,
      "learning_rate": 0.0005984959177228996,
      "loss": 2.5495,
      "step": 7347
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6687870621681213,
      "learning_rate": 0.0005984955085964168,
      "loss": 2.4326,
      "step": 7348
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.7169615030288696,
      "learning_rate": 0.0005984950994144383,
      "loss": 2.5744,
      "step": 7349
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6403290033340454,
      "learning_rate": 0.0005984946901769636,
      "loss": 2.6489,
      "step": 7350
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.635522723197937,
      "learning_rate": 0.0005984942808839933,
      "loss": 2.5813,
      "step": 7351
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6308916211128235,
      "learning_rate": 0.0005984938715355273,
      "loss": 2.7427,
      "step": 7352
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.7165940999984741,
      "learning_rate": 0.0005984934621315655,
      "loss": 2.4557,
      "step": 7353
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.7238137722015381,
      "learning_rate": 0.0005984930526721082,
      "loss": 2.2975,
      "step": 7354
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6645545959472656,
      "learning_rate": 0.0005984926431571554,
      "loss": 2.3788,
      "step": 7355
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.674834132194519,
      "learning_rate": 0.0005984922335867072,
      "loss": 2.6698,
      "step": 7356
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.649895966053009,
      "learning_rate": 0.0005984918239607638,
      "loss": 2.4561,
      "step": 7357
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.7016710638999939,
      "learning_rate": 0.000598491414279325,
      "loss": 2.2785,
      "step": 7358
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.7429511547088623,
      "learning_rate": 0.000598491004542391,
      "loss": 2.6518,
      "step": 7359
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.7370120882987976,
      "learning_rate": 0.0005984905947499619,
      "loss": 2.4933,
      "step": 7360
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6510459780693054,
      "learning_rate": 0.0005984901849020377,
      "loss": 2.4257,
      "step": 7361
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.8330798149108887,
      "learning_rate": 0.0005984897749986187,
      "loss": 2.3348,
      "step": 7362
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6484537124633789,
      "learning_rate": 0.0005984893650397046,
      "loss": 2.7079,
      "step": 7363
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6584639549255371,
      "learning_rate": 0.0005984889550252959,
      "loss": 2.526,
      "step": 7364
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6520618200302124,
      "learning_rate": 0.0005984885449553923,
      "loss": 2.5183,
      "step": 7365
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6753065586090088,
      "learning_rate": 0.0005984881348299943,
      "loss": 2.4209,
      "step": 7366
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6093519926071167,
      "learning_rate": 0.0005984877246491014,
      "loss": 2.4172,
      "step": 7367
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6670398116111755,
      "learning_rate": 0.0005984873144127143,
      "loss": 2.5257,
      "step": 7368
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.654874861240387,
      "learning_rate": 0.0005984869041208325,
      "loss": 2.6162,
      "step": 7369
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.7049770355224609,
      "learning_rate": 0.0005984864937734565,
      "loss": 2.6153,
      "step": 7370
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6290481090545654,
      "learning_rate": 0.0005984860833705862,
      "loss": 2.5986,
      "step": 7371
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6455714106559753,
      "learning_rate": 0.0005984856729122217,
      "loss": 2.3771,
      "step": 7372
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6625446677207947,
      "learning_rate": 0.0005984852623983631,
      "loss": 2.567,
      "step": 7373
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6427215933799744,
      "learning_rate": 0.0005984848518290105,
      "loss": 2.5114,
      "step": 7374
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.7093666791915894,
      "learning_rate": 0.0005984844412041638,
      "loss": 2.5894,
      "step": 7375
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6989210844039917,
      "learning_rate": 0.0005984840305238233,
      "loss": 2.4465,
      "step": 7376
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.5773652195930481,
      "learning_rate": 0.000598483619787989,
      "loss": 2.6734,
      "step": 7377
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6040113568305969,
      "learning_rate": 0.0005984832089966609,
      "loss": 2.831,
      "step": 7378
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6332191228866577,
      "learning_rate": 0.0005984827981498392,
      "loss": 2.5472,
      "step": 7379
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6922649145126343,
      "learning_rate": 0.0005984823872475239,
      "loss": 2.6361,
      "step": 7380
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6457064151763916,
      "learning_rate": 0.000598481976289715,
      "loss": 2.4343,
      "step": 7381
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.5984686613082886,
      "learning_rate": 0.0005984815652764127,
      "loss": 2.5723,
      "step": 7382
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6575353741645813,
      "learning_rate": 0.000598481154207617,
      "loss": 2.576,
      "step": 7383
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6403270959854126,
      "learning_rate": 0.0005984807430833281,
      "loss": 2.529,
      "step": 7384
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6388635635375977,
      "learning_rate": 0.0005984803319035461,
      "loss": 2.3405,
      "step": 7385
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.7679312229156494,
      "learning_rate": 0.0005984799206682707,
      "loss": 2.4199,
      "step": 7386
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.7037944793701172,
      "learning_rate": 0.0005984795093775026,
      "loss": 2.5043,
      "step": 7387
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6256527900695801,
      "learning_rate": 0.0005984790980312412,
      "loss": 2.5112,
      "step": 7388
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6232386827468872,
      "learning_rate": 0.0005984786866294872,
      "loss": 2.6669,
      "step": 7389
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6761517524719238,
      "learning_rate": 0.0005984782751722402,
      "loss": 2.5463,
      "step": 7390
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.5513626337051392,
      "learning_rate": 0.0005984778636595004,
      "loss": 2.3788,
      "step": 7391
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6347423195838928,
      "learning_rate": 0.000598477452091268,
      "loss": 2.575,
      "step": 7392
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6440300345420837,
      "learning_rate": 0.000598477040467543,
      "loss": 2.6454,
      "step": 7393
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.7096789479255676,
      "learning_rate": 0.0005984766287883255,
      "loss": 2.568,
      "step": 7394
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.7467336058616638,
      "learning_rate": 0.0005984762170536156,
      "loss": 2.5878,
      "step": 7395
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6680461168289185,
      "learning_rate": 0.0005984758052634132,
      "loss": 2.4891,
      "step": 7396
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.7130926251411438,
      "learning_rate": 0.0005984753934177187,
      "loss": 2.5577,
      "step": 7397
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.7925388216972351,
      "learning_rate": 0.0005984749815165319,
      "loss": 2.6833,
      "step": 7398
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6662088632583618,
      "learning_rate": 0.0005984745695598529,
      "loss": 2.3894,
      "step": 7399
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6108753681182861,
      "learning_rate": 0.0005984741575476819,
      "loss": 2.7345,
      "step": 7400
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.7239307761192322,
      "learning_rate": 0.0005984737454800188,
      "loss": 2.6113,
      "step": 7401
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6754918694496155,
      "learning_rate": 0.000598473333356864,
      "loss": 2.7051,
      "step": 7402
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6385837197303772,
      "learning_rate": 0.0005984729211782173,
      "loss": 2.5035,
      "step": 7403
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6149317622184753,
      "learning_rate": 0.0005984725089440788,
      "loss": 2.5931,
      "step": 7404
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6971140503883362,
      "learning_rate": 0.0005984720966544487,
      "loss": 2.482,
      "step": 7405
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6401717066764832,
      "learning_rate": 0.000598471684309327,
      "loss": 2.6118,
      "step": 7406
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6227974891662598,
      "learning_rate": 0.0005984712719087137,
      "loss": 2.5105,
      "step": 7407
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6449311375617981,
      "learning_rate": 0.000598470859452609,
      "loss": 2.5929,
      "step": 7408
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.8520323634147644,
      "learning_rate": 0.000598470446941013,
      "loss": 2.4361,
      "step": 7409
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6183464527130127,
      "learning_rate": 0.0005984700343739256,
      "loss": 2.3922,
      "step": 7410
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6911658644676208,
      "learning_rate": 0.0005984696217513471,
      "loss": 2.4959,
      "step": 7411
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6695586442947388,
      "learning_rate": 0.0005984692090732774,
      "loss": 2.5669,
      "step": 7412
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6424603462219238,
      "learning_rate": 0.0005984687963397166,
      "loss": 2.6314,
      "step": 7413
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6752599477767944,
      "learning_rate": 0.0005984683835506649,
      "loss": 2.7418,
      "step": 7414
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6850757598876953,
      "learning_rate": 0.0005984679707061221,
      "loss": 2.5691,
      "step": 7415
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6615699529647827,
      "learning_rate": 0.0005984675578060887,
      "loss": 2.4833,
      "step": 7416
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.7011368870735168,
      "learning_rate": 0.0005984671448505645,
      "loss": 2.2828,
      "step": 7417
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6977564692497253,
      "learning_rate": 0.0005984667318395495,
      "loss": 2.6438,
      "step": 7418
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6938174366950989,
      "learning_rate": 0.0005984663187730441,
      "loss": 2.7543,
      "step": 7419
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.7467580437660217,
      "learning_rate": 0.000598465905651048,
      "loss": 2.4425,
      "step": 7420
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.639895498752594,
      "learning_rate": 0.0005984654924735616,
      "loss": 2.6347,
      "step": 7421
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.7366781830787659,
      "learning_rate": 0.0005984650792405848,
      "loss": 2.7116,
      "step": 7422
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.609571099281311,
      "learning_rate": 0.0005984646659521176,
      "loss": 2.6107,
      "step": 7423
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6971877813339233,
      "learning_rate": 0.0005984642526081602,
      "loss": 2.6568,
      "step": 7424
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6176165342330933,
      "learning_rate": 0.0005984638392087128,
      "loss": 2.7588,
      "step": 7425
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.7585415244102478,
      "learning_rate": 0.0005984634257537752,
      "loss": 2.6048,
      "step": 7426
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6586027145385742,
      "learning_rate": 0.0005984630122433476,
      "loss": 2.7003,
      "step": 7427
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.689265787601471,
      "learning_rate": 0.0005984625986774301,
      "loss": 2.599,
      "step": 7428
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6257941722869873,
      "learning_rate": 0.0005984621850560229,
      "loss": 2.4788,
      "step": 7429
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.7472779154777527,
      "learning_rate": 0.0005984617713791259,
      "loss": 2.595,
      "step": 7430
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.7045979499816895,
      "learning_rate": 0.0005984613576467392,
      "loss": 2.6178,
      "step": 7431
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6412622928619385,
      "learning_rate": 0.0005984609438588629,
      "loss": 2.6133,
      "step": 7432
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.7075227499008179,
      "learning_rate": 0.0005984605300154971,
      "loss": 2.475,
      "step": 7433
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.7997585535049438,
      "learning_rate": 0.0005984601161166418,
      "loss": 2.5578,
      "step": 7434
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.7793126106262207,
      "learning_rate": 0.0005984597021622971,
      "loss": 2.4067,
      "step": 7435
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.8216275572776794,
      "learning_rate": 0.0005984592881524632,
      "loss": 2.5201,
      "step": 7436
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.8092771768569946,
      "learning_rate": 0.00059845887408714,
      "loss": 2.5433,
      "step": 7437
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.8318068981170654,
      "learning_rate": 0.0005984584599663278,
      "loss": 2.47,
      "step": 7438
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.749832808971405,
      "learning_rate": 0.0005984580457900265,
      "loss": 2.6313,
      "step": 7439
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6810221076011658,
      "learning_rate": 0.0005984576315582361,
      "loss": 2.5822,
      "step": 7440
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6195533871650696,
      "learning_rate": 0.0005984572172709568,
      "loss": 2.5394,
      "step": 7441
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.7004827857017517,
      "learning_rate": 0.0005984568029281887,
      "loss": 2.4679,
      "step": 7442
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6817665100097656,
      "learning_rate": 0.0005984563885299319,
      "loss": 2.5886,
      "step": 7443
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6332522034645081,
      "learning_rate": 0.0005984559740761864,
      "loss": 2.6581,
      "step": 7444
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6628684997558594,
      "learning_rate": 0.0005984555595669523,
      "loss": 2.6548,
      "step": 7445
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6062252521514893,
      "learning_rate": 0.0005984551450022296,
      "loss": 2.694,
      "step": 7446
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6863541007041931,
      "learning_rate": 0.0005984547303820185,
      "loss": 2.5344,
      "step": 7447
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.7365632653236389,
      "learning_rate": 0.000598454315706319,
      "loss": 2.5462,
      "step": 7448
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6884400248527527,
      "learning_rate": 0.0005984539009751314,
      "loss": 2.4257,
      "step": 7449
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6769339442253113,
      "learning_rate": 0.0005984534861884555,
      "loss": 2.4838,
      "step": 7450
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6315547823905945,
      "learning_rate": 0.0005984530713462914,
      "loss": 2.562,
      "step": 7451
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6758385300636292,
      "learning_rate": 0.0005984526564486392,
      "loss": 2.4631,
      "step": 7452
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6355603933334351,
      "learning_rate": 0.000598452241495499,
      "loss": 2.7258,
      "step": 7453
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.7321087121963501,
      "learning_rate": 0.000598451826486871,
      "loss": 2.3837,
      "step": 7454
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.7714791893959045,
      "learning_rate": 0.0005984514114227551,
      "loss": 2.5717,
      "step": 7455
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.7003124952316284,
      "learning_rate": 0.0005984509963031516,
      "loss": 2.6081,
      "step": 7456
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.5933889746665955,
      "learning_rate": 0.0005984505811280602,
      "loss": 2.4568,
      "step": 7457
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.7214502692222595,
      "learning_rate": 0.0005984501658974815,
      "loss": 2.6598,
      "step": 7458
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6151688694953918,
      "learning_rate": 0.000598449750611415,
      "loss": 2.4124,
      "step": 7459
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.7147956490516663,
      "learning_rate": 0.0005984493352698612,
      "loss": 2.6902,
      "step": 7460
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.7427868247032166,
      "learning_rate": 0.00059844891987282,
      "loss": 2.5275,
      "step": 7461
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6604509949684143,
      "learning_rate": 0.0005984485044202915,
      "loss": 2.5317,
      "step": 7462
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.7592334151268005,
      "learning_rate": 0.0005984480889122757,
      "loss": 2.4901,
      "step": 7463
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6341169476509094,
      "learning_rate": 0.0005984476733487729,
      "loss": 2.5424,
      "step": 7464
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.7275309562683105,
      "learning_rate": 0.000598447257729783,
      "loss": 2.5252,
      "step": 7465
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.7441490292549133,
      "learning_rate": 0.0005984468420553061,
      "loss": 2.5569,
      "step": 7466
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6439693570137024,
      "learning_rate": 0.0005984464263253423,
      "loss": 2.5917,
      "step": 7467
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6523005366325378,
      "learning_rate": 0.0005984460105398917,
      "loss": 2.3144,
      "step": 7468
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.766520619392395,
      "learning_rate": 0.0005984455946989543,
      "loss": 2.5629,
      "step": 7469
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.7208495140075684,
      "learning_rate": 0.0005984451788025304,
      "loss": 2.5169,
      "step": 7470
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.7369781136512756,
      "learning_rate": 0.0005984447628506198,
      "loss": 2.3076,
      "step": 7471
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.8907356262207031,
      "learning_rate": 0.0005984443468432226,
      "loss": 2.4599,
      "step": 7472
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6092116832733154,
      "learning_rate": 0.000598443930780339,
      "loss": 2.4126,
      "step": 7473
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.9797945022583008,
      "learning_rate": 0.0005984435146619691,
      "loss": 2.5925,
      "step": 7474
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6806934475898743,
      "learning_rate": 0.0005984430984881129,
      "loss": 2.6088,
      "step": 7475
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6930369138717651,
      "learning_rate": 0.0005984426822587704,
      "loss": 2.5414,
      "step": 7476
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.792151927947998,
      "learning_rate": 0.0005984422659739419,
      "loss": 2.5858,
      "step": 7477
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6916285157203674,
      "learning_rate": 0.0005984418496336274,
      "loss": 2.609,
      "step": 7478
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6401399970054626,
      "learning_rate": 0.0005984414332378268,
      "loss": 2.6095,
      "step": 7479
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.9744380116462708,
      "learning_rate": 0.0005984410167865403,
      "loss": 2.6703,
      "step": 7480
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6492828726768494,
      "learning_rate": 0.0005984406002797681,
      "loss": 2.5795,
      "step": 7481
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6768133640289307,
      "learning_rate": 0.00059844018371751,
      "loss": 2.5725,
      "step": 7482
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.8617453575134277,
      "learning_rate": 0.0005984397670997664,
      "loss": 2.5207,
      "step": 7483
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.7465846538543701,
      "learning_rate": 0.000598439350426537,
      "loss": 2.8702,
      "step": 7484
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.7113721966743469,
      "learning_rate": 0.0005984389336978223,
      "loss": 2.6373,
      "step": 7485
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.7036914229393005,
      "learning_rate": 0.0005984385169136221,
      "loss": 2.6478,
      "step": 7486
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.8390167951583862,
      "learning_rate": 0.0005984381000739365,
      "loss": 2.4609,
      "step": 7487
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.7661296129226685,
      "learning_rate": 0.0005984376831787657,
      "loss": 2.4887,
      "step": 7488
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.7340179681777954,
      "learning_rate": 0.0005984372662281097,
      "loss": 2.4239,
      "step": 7489
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.8289836645126343,
      "learning_rate": 0.0005984368492219684,
      "loss": 2.6446,
      "step": 7490
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.657383918762207,
      "learning_rate": 0.0005984364321603423,
      "loss": 2.712,
      "step": 7491
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.898783802986145,
      "learning_rate": 0.0005984360150432311,
      "loss": 2.4464,
      "step": 7492
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.804744303226471,
      "learning_rate": 0.0005984355978706351,
      "loss": 2.5465,
      "step": 7493
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6739963293075562,
      "learning_rate": 0.0005984351806425542,
      "loss": 2.3425,
      "step": 7494
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.9526901841163635,
      "learning_rate": 0.0005984347633589886,
      "loss": 2.7297,
      "step": 7495
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.084445834159851,
      "learning_rate": 0.0005984343460199383,
      "loss": 2.6356,
      "step": 7496
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6830475926399231,
      "learning_rate": 0.0005984339286254035,
      "loss": 2.6069,
      "step": 7497
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.8144857883453369,
      "learning_rate": 0.0005984335111753841,
      "loss": 2.4565,
      "step": 7498
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.7892549633979797,
      "learning_rate": 0.0005984330936698803,
      "loss": 2.474,
      "step": 7499
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.7011404037475586,
      "learning_rate": 0.0005984326761088923,
      "loss": 2.6099,
      "step": 7500
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6673901677131653,
      "learning_rate": 0.0005984322584924198,
      "loss": 2.4706,
      "step": 7501
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.8209075927734375,
      "learning_rate": 0.0005984318408204633,
      "loss": 2.5509,
      "step": 7502
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.7092770338058472,
      "learning_rate": 0.0005984314230930227,
      "loss": 2.4909,
      "step": 7503
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6459435820579529,
      "learning_rate": 0.0005984310053100979,
      "loss": 2.6262,
      "step": 7504
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.7912046313285828,
      "learning_rate": 0.0005984305874716892,
      "loss": 2.4928,
      "step": 7505
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6865495443344116,
      "learning_rate": 0.0005984301695777966,
      "loss": 2.6196,
      "step": 7506
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6849989295005798,
      "learning_rate": 0.0005984297516284204,
      "loss": 2.7445,
      "step": 7507
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.7072331309318542,
      "learning_rate": 0.0005984293336235602,
      "loss": 2.5823,
      "step": 7508
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.7725932598114014,
      "learning_rate": 0.0005984289155632165,
      "loss": 2.4752,
      "step": 7509
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.676979660987854,
      "learning_rate": 0.0005984284974473893,
      "loss": 2.4195,
      "step": 7510
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6217155456542969,
      "learning_rate": 0.0005984280792760785,
      "loss": 2.5786,
      "step": 7511
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.7927014827728271,
      "learning_rate": 0.0005984276610492843,
      "loss": 2.5625,
      "step": 7512
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.7205677032470703,
      "learning_rate": 0.0005984272427670068,
      "loss": 2.2662,
      "step": 7513
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6770572662353516,
      "learning_rate": 0.000598426824429246,
      "loss": 2.5862,
      "step": 7514
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.7231062650680542,
      "learning_rate": 0.000598426406036002,
      "loss": 2.8601,
      "step": 7515
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.9108113646507263,
      "learning_rate": 0.000598425987587275,
      "loss": 2.3941,
      "step": 7516
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6697226762771606,
      "learning_rate": 0.0005984255690830648,
      "loss": 2.4721,
      "step": 7517
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.7320330739021301,
      "learning_rate": 0.0005984251505233717,
      "loss": 2.7573,
      "step": 7518
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.9869868159294128,
      "learning_rate": 0.0005984247319081959,
      "loss": 2.5869,
      "step": 7519
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6502108573913574,
      "learning_rate": 0.0005984243132375372,
      "loss": 2.7318,
      "step": 7520
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.671619176864624,
      "learning_rate": 0.0005984238945113958,
      "loss": 2.4197,
      "step": 7521
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.8574416637420654,
      "learning_rate": 0.0005984234757297717,
      "loss": 2.6924,
      "step": 7522
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6855572462081909,
      "learning_rate": 0.0005984230568926651,
      "loss": 2.448,
      "step": 7523
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6249758005142212,
      "learning_rate": 0.000598422638000076,
      "loss": 2.3437,
      "step": 7524
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.667758584022522,
      "learning_rate": 0.0005984222190520045,
      "loss": 2.4186,
      "step": 7525
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6867289543151855,
      "learning_rate": 0.0005984218000484508,
      "loss": 2.6795,
      "step": 7526
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6916797757148743,
      "learning_rate": 0.0005984213809894146,
      "loss": 2.6251,
      "step": 7527
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.650076687335968,
      "learning_rate": 0.0005984209618748965,
      "loss": 2.6585,
      "step": 7528
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6818372011184692,
      "learning_rate": 0.0005984205427048961,
      "loss": 2.5281,
      "step": 7529
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6489864587783813,
      "learning_rate": 0.0005984201234794138,
      "loss": 2.5275,
      "step": 7530
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6392775774002075,
      "learning_rate": 0.0005984197041984495,
      "loss": 2.445,
      "step": 7531
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.7807351350784302,
      "learning_rate": 0.0005984192848620034,
      "loss": 2.499,
      "step": 7532
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.7357744574546814,
      "learning_rate": 0.0005984188654700756,
      "loss": 2.5775,
      "step": 7533
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6382204294204712,
      "learning_rate": 0.0005984184460226659,
      "loss": 2.4333,
      "step": 7534
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6982870101928711,
      "learning_rate": 0.0005984180265197747,
      "loss": 2.4394,
      "step": 7535
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6792338490486145,
      "learning_rate": 0.000598417606961402,
      "loss": 2.5473,
      "step": 7536
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6572919487953186,
      "learning_rate": 0.0005984171873475477,
      "loss": 2.5266,
      "step": 7537
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.7334256768226624,
      "learning_rate": 0.0005984167676782122,
      "loss": 2.42,
      "step": 7538
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.7000269293785095,
      "learning_rate": 0.0005984163479533952,
      "loss": 2.3709,
      "step": 7539
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6272670030593872,
      "learning_rate": 0.000598415928173097,
      "loss": 2.6489,
      "step": 7540
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.7032811641693115,
      "learning_rate": 0.0005984155083373177,
      "loss": 2.7072,
      "step": 7541
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6994287371635437,
      "learning_rate": 0.0005984150884460574,
      "loss": 2.5138,
      "step": 7542
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6232630610466003,
      "learning_rate": 0.000598414668499316,
      "loss": 2.4959,
      "step": 7543
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6307439804077148,
      "learning_rate": 0.0005984142484970937,
      "loss": 2.1517,
      "step": 7544
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6445704102516174,
      "learning_rate": 0.0005984138284393904,
      "loss": 2.5054,
      "step": 7545
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6722317337989807,
      "learning_rate": 0.0005984134083262066,
      "loss": 2.4428,
      "step": 7546
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.7033147811889648,
      "learning_rate": 0.000598412988157542,
      "loss": 2.7756,
      "step": 7547
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.641265332698822,
      "learning_rate": 0.0005984125679333968,
      "loss": 2.5805,
      "step": 7548
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.7328931093215942,
      "learning_rate": 0.000598412147653771,
      "loss": 2.5892,
      "step": 7549
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6479615569114685,
      "learning_rate": 0.0005984117273186648,
      "loss": 2.4818,
      "step": 7550
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.63038170337677,
      "learning_rate": 0.0005984113069280783,
      "loss": 2.4287,
      "step": 7551
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.7312529683113098,
      "learning_rate": 0.0005984108864820115,
      "loss": 2.4158,
      "step": 7552
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6255643963813782,
      "learning_rate": 0.0005984104659804644,
      "loss": 2.5814,
      "step": 7553
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.7013826966285706,
      "learning_rate": 0.0005984100454234372,
      "loss": 2.4684,
      "step": 7554
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.5871601104736328,
      "learning_rate": 0.0005984096248109299,
      "loss": 2.5245,
      "step": 7555
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.7164337038993835,
      "learning_rate": 0.0005984092041429426,
      "loss": 2.3064,
      "step": 7556
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6811413764953613,
      "learning_rate": 0.0005984087834194755,
      "loss": 2.6696,
      "step": 7557
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6289144158363342,
      "learning_rate": 0.0005984083626405285,
      "loss": 2.3668,
      "step": 7558
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6433618068695068,
      "learning_rate": 0.0005984079418061018,
      "loss": 2.5477,
      "step": 7559
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6347718834877014,
      "learning_rate": 0.0005984075209161954,
      "loss": 2.2408,
      "step": 7560
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.5969012975692749,
      "learning_rate": 0.0005984070999708094,
      "loss": 2.2668,
      "step": 7561
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6216147541999817,
      "learning_rate": 0.0005984066789699439,
      "loss": 2.6981,
      "step": 7562
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.703754186630249,
      "learning_rate": 0.000598406257913599,
      "loss": 2.545,
      "step": 7563
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6832488775253296,
      "learning_rate": 0.0005984058368017747,
      "loss": 2.5271,
      "step": 7564
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6896129250526428,
      "learning_rate": 0.0005984054156344712,
      "loss": 2.3991,
      "step": 7565
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.596047580242157,
      "learning_rate": 0.0005984049944116884,
      "loss": 2.8532,
      "step": 7566
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6884950995445251,
      "learning_rate": 0.0005984045731334264,
      "loss": 2.6195,
      "step": 7567
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.7650938034057617,
      "learning_rate": 0.0005984041517996855,
      "loss": 2.7215,
      "step": 7568
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6628664135932922,
      "learning_rate": 0.0005984037304104656,
      "loss": 2.4015,
      "step": 7569
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6506144404411316,
      "learning_rate": 0.0005984033089657669,
      "loss": 2.629,
      "step": 7570
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6462854743003845,
      "learning_rate": 0.0005984028874655892,
      "loss": 2.4698,
      "step": 7571
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6858940124511719,
      "learning_rate": 0.0005984024659099329,
      "loss": 2.4831,
      "step": 7572
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6132253408432007,
      "learning_rate": 0.000598402044298798,
      "loss": 2.5705,
      "step": 7573
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.7481646537780762,
      "learning_rate": 0.0005984016226321844,
      "loss": 2.6539,
      "step": 7574
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6472523808479309,
      "learning_rate": 0.0005984012009100923,
      "loss": 2.5757,
      "step": 7575
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6495252251625061,
      "learning_rate": 0.000598400779132522,
      "loss": 2.5317,
      "step": 7576
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6557735204696655,
      "learning_rate": 0.000598400357299473,
      "loss": 2.5473,
      "step": 7577
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.7263521552085876,
      "learning_rate": 0.000598399935410946,
      "loss": 2.6079,
      "step": 7578
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6807079911231995,
      "learning_rate": 0.0005983995134669408,
      "loss": 2.3789,
      "step": 7579
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6445644497871399,
      "learning_rate": 0.0005983990914674574,
      "loss": 2.5812,
      "step": 7580
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6984725594520569,
      "learning_rate": 0.0005983986694124961,
      "loss": 2.5758,
      "step": 7581
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6406078934669495,
      "learning_rate": 0.0005983982473020569,
      "loss": 2.4136,
      "step": 7582
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.7171498537063599,
      "learning_rate": 0.0005983978251361396,
      "loss": 2.6365,
      "step": 7583
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6604505777359009,
      "learning_rate": 0.0005983974029147447,
      "loss": 2.4502,
      "step": 7584
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6232426166534424,
      "learning_rate": 0.0005983969806378721,
      "loss": 2.5894,
      "step": 7585
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.759131908416748,
      "learning_rate": 0.0005983965583055217,
      "loss": 2.5092,
      "step": 7586
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6511456966400146,
      "learning_rate": 0.0005983961359176938,
      "loss": 2.4683,
      "step": 7587
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.5911606550216675,
      "learning_rate": 0.0005983957134743886,
      "loss": 2.3431,
      "step": 7588
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6926339268684387,
      "learning_rate": 0.0005983952909756059,
      "loss": 2.532,
      "step": 7589
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6271889209747314,
      "learning_rate": 0.0005983948684213458,
      "loss": 2.4394,
      "step": 7590
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6241626143455505,
      "learning_rate": 0.0005983944458116085,
      "loss": 2.5131,
      "step": 7591
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.7259817123413086,
      "learning_rate": 0.0005983940231463941,
      "loss": 2.4527,
      "step": 7592
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6408013105392456,
      "learning_rate": 0.0005983936004257026,
      "loss": 2.5517,
      "step": 7593
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6381810307502747,
      "learning_rate": 0.000598393177649534,
      "loss": 2.581,
      "step": 7594
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.7418453693389893,
      "learning_rate": 0.0005983927548178885,
      "loss": 2.3639,
      "step": 7595
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.7731224894523621,
      "learning_rate": 0.0005983923319307663,
      "loss": 2.5037,
      "step": 7596
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6771379709243774,
      "learning_rate": 0.0005983919089881672,
      "loss": 2.797,
      "step": 7597
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6669347286224365,
      "learning_rate": 0.0005983914859900915,
      "loss": 2.7041,
      "step": 7598
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.8248394727706909,
      "learning_rate": 0.0005983910629365391,
      "loss": 2.2548,
      "step": 7599
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.5952841639518738,
      "learning_rate": 0.0005983906398275103,
      "loss": 2.6857,
      "step": 7600
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6725239157676697,
      "learning_rate": 0.0005983902166630048,
      "loss": 2.3362,
      "step": 7601
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.0842856168746948,
      "learning_rate": 0.0005983897934430231,
      "loss": 2.4755,
      "step": 7602
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.636780858039856,
      "learning_rate": 0.0005983893701675652,
      "loss": 2.5529,
      "step": 7603
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.8167639970779419,
      "learning_rate": 0.0005983889468366308,
      "loss": 2.5292,
      "step": 7604
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.9686199426651001,
      "learning_rate": 0.0005983885234502205,
      "loss": 2.6983,
      "step": 7605
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6821402907371521,
      "learning_rate": 0.000598388100008334,
      "loss": 2.3141,
      "step": 7606
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.7108734250068665,
      "learning_rate": 0.0005983876765109717,
      "loss": 2.739,
      "step": 7607
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.956131637096405,
      "learning_rate": 0.0005983872529581334,
      "loss": 2.5651,
      "step": 7608
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.7677019834518433,
      "learning_rate": 0.0005983868293498193,
      "loss": 2.4865,
      "step": 7609
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.675399661064148,
      "learning_rate": 0.0005983864056860294,
      "loss": 2.4959,
      "step": 7610
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.9762360453605652,
      "learning_rate": 0.0005983859819667638,
      "loss": 2.4306,
      "step": 7611
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6897776126861572,
      "learning_rate": 0.0005983855581920227,
      "loss": 2.8966,
      "step": 7612
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.677577018737793,
      "learning_rate": 0.0005983851343618061,
      "loss": 2.5297,
      "step": 7613
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.7225135564804077,
      "learning_rate": 0.000598384710476114,
      "loss": 2.499,
      "step": 7614
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6716689467430115,
      "learning_rate": 0.0005983842865349466,
      "loss": 2.408,
      "step": 7615
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6505428552627563,
      "learning_rate": 0.0005983838625383039,
      "loss": 2.3391,
      "step": 7616
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.5984740853309631,
      "learning_rate": 0.000598383438486186,
      "loss": 2.3383,
      "step": 7617
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6647832989692688,
      "learning_rate": 0.000598383014378593,
      "loss": 2.4684,
      "step": 7618
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.8054327368736267,
      "learning_rate": 0.000598382590215525,
      "loss": 2.4456,
      "step": 7619
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6063205003738403,
      "learning_rate": 0.0005983821659969821,
      "loss": 2.5249,
      "step": 7620
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.7327206134796143,
      "learning_rate": 0.0005983817417229641,
      "loss": 2.5166,
      "step": 7621
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6572653651237488,
      "learning_rate": 0.0005983813173934715,
      "loss": 2.4752,
      "step": 7622
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6440636515617371,
      "learning_rate": 0.0005983808930085041,
      "loss": 2.4788,
      "step": 7623
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.7556066513061523,
      "learning_rate": 0.0005983804685680621,
      "loss": 2.5923,
      "step": 7624
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6611855626106262,
      "learning_rate": 0.0005983800440721455,
      "loss": 2.7247,
      "step": 7625
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6631625890731812,
      "learning_rate": 0.0005983796195207545,
      "loss": 2.5748,
      "step": 7626
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6910608410835266,
      "learning_rate": 0.000598379194913889,
      "loss": 2.6967,
      "step": 7627
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.7917916178703308,
      "learning_rate": 0.0005983787702515492,
      "loss": 2.5301,
      "step": 7628
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6436285376548767,
      "learning_rate": 0.0005983783455337352,
      "loss": 2.4632,
      "step": 7629
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.7026883959770203,
      "learning_rate": 0.000598377920760447,
      "loss": 2.5132,
      "step": 7630
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.7499366402626038,
      "learning_rate": 0.0005983774959316847,
      "loss": 2.4297,
      "step": 7631
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.7030335664749146,
      "learning_rate": 0.0005983770710474484,
      "loss": 2.5017,
      "step": 7632
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6583862900733948,
      "learning_rate": 0.0005983766461077382,
      "loss": 2.4125,
      "step": 7633
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6295758485794067,
      "learning_rate": 0.0005983762211125541,
      "loss": 2.658,
      "step": 7634
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6487185955047607,
      "learning_rate": 0.0005983757960618963,
      "loss": 2.4268,
      "step": 7635
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.653325617313385,
      "learning_rate": 0.0005983753709557648,
      "loss": 2.3771,
      "step": 7636
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.8087547421455383,
      "learning_rate": 0.0005983749457941595,
      "loss": 2.5326,
      "step": 7637
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6656121015548706,
      "learning_rate": 0.0005983745205770808,
      "loss": 2.553,
      "step": 7638
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6400208473205566,
      "learning_rate": 0.0005983740953045287,
      "loss": 2.5418,
      "step": 7639
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.7146795392036438,
      "learning_rate": 0.0005983736699765031,
      "loss": 2.5098,
      "step": 7640
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.7762993574142456,
      "learning_rate": 0.0005983732445930043,
      "loss": 2.7338,
      "step": 7641
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6361713409423828,
      "learning_rate": 0.0005983728191540323,
      "loss": 2.4142,
      "step": 7642
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.7544417977333069,
      "learning_rate": 0.0005983723936595871,
      "loss": 2.7189,
      "step": 7643
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.7214205265045166,
      "learning_rate": 0.0005983719681096689,
      "loss": 2.4484,
      "step": 7644
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.613651692867279,
      "learning_rate": 0.0005983715425042777,
      "loss": 2.5845,
      "step": 7645
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.087677240371704,
      "learning_rate": 0.0005983711168434135,
      "loss": 2.5076,
      "step": 7646
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6712702512741089,
      "learning_rate": 0.0005983706911270766,
      "loss": 2.8805,
      "step": 7647
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.7828832268714905,
      "learning_rate": 0.0005983702653552669,
      "loss": 2.5501,
      "step": 7648
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.7588489055633545,
      "learning_rate": 0.0005983698395279845,
      "loss": 2.5184,
      "step": 7649
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6711752414703369,
      "learning_rate": 0.0005983694136452296,
      "loss": 2.5149,
      "step": 7650
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6914017796516418,
      "learning_rate": 0.0005983689877070022,
      "loss": 2.5465,
      "step": 7651
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6319096088409424,
      "learning_rate": 0.0005983685617133023,
      "loss": 2.4861,
      "step": 7652
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6609293818473816,
      "learning_rate": 0.0005983681356641302,
      "loss": 2.5094,
      "step": 7653
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6553248763084412,
      "learning_rate": 0.0005983677095594856,
      "loss": 2.4307,
      "step": 7654
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6693761348724365,
      "learning_rate": 0.0005983672833993689,
      "loss": 2.3943,
      "step": 7655
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.7027056217193604,
      "learning_rate": 0.0005983668571837801,
      "loss": 2.7252,
      "step": 7656
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6255707740783691,
      "learning_rate": 0.0005983664309127193,
      "loss": 2.5462,
      "step": 7657
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.7329011559486389,
      "learning_rate": 0.0005983660045861865,
      "loss": 2.7221,
      "step": 7658
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.7007840871810913,
      "learning_rate": 0.0005983655782041819,
      "loss": 2.6215,
      "step": 7659
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6780920028686523,
      "learning_rate": 0.0005983651517667055,
      "loss": 2.6034,
      "step": 7660
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.5986554026603699,
      "learning_rate": 0.0005983647252737574,
      "loss": 2.7059,
      "step": 7661
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6987914443016052,
      "learning_rate": 0.0005983642987253375,
      "loss": 2.51,
      "step": 7662
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.7783174514770508,
      "learning_rate": 0.0005983638721214461,
      "loss": 2.7085,
      "step": 7663
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6524061560630798,
      "learning_rate": 0.0005983634454620833,
      "loss": 2.3962,
      "step": 7664
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.7993087768554688,
      "learning_rate": 0.0005983630187472492,
      "loss": 2.5779,
      "step": 7665
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.632021427154541,
      "learning_rate": 0.0005983625919769436,
      "loss": 2.4859,
      "step": 7666
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.7416061758995056,
      "learning_rate": 0.0005983621651511669,
      "loss": 2.6953,
      "step": 7667
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6725522875785828,
      "learning_rate": 0.0005983617382699189,
      "loss": 2.6375,
      "step": 7668
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.7007815837860107,
      "learning_rate": 0.0005983613113331999,
      "loss": 2.667,
      "step": 7669
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.7029858231544495,
      "learning_rate": 0.0005983608843410099,
      "loss": 2.6607,
      "step": 7670
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6686106324195862,
      "learning_rate": 0.0005983604572933491,
      "loss": 2.2712,
      "step": 7671
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6774706840515137,
      "learning_rate": 0.0005983600301902172,
      "loss": 2.4937,
      "step": 7672
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.5986934304237366,
      "learning_rate": 0.0005983596030316147,
      "loss": 2.4858,
      "step": 7673
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6293423175811768,
      "learning_rate": 0.0005983591758175415,
      "loss": 2.5997,
      "step": 7674
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6053018569946289,
      "learning_rate": 0.0005983587485479977,
      "loss": 2.4897,
      "step": 7675
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6042799353599548,
      "learning_rate": 0.0005983583212229834,
      "loss": 2.6804,
      "step": 7676
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.695004940032959,
      "learning_rate": 0.0005983578938424986,
      "loss": 2.6737,
      "step": 7677
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6203343272209167,
      "learning_rate": 0.0005983574664065435,
      "loss": 2.681,
      "step": 7678
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6181293725967407,
      "learning_rate": 0.0005983570389151181,
      "loss": 2.5122,
      "step": 7679
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6055382490158081,
      "learning_rate": 0.0005983566113682224,
      "loss": 2.5881,
      "step": 7680
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6320822238922119,
      "learning_rate": 0.0005983561837658567,
      "loss": 2.5294,
      "step": 7681
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6031661629676819,
      "learning_rate": 0.0005983557561080208,
      "loss": 2.5702,
      "step": 7682
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6850079298019409,
      "learning_rate": 0.000598355328394715,
      "loss": 2.4093,
      "step": 7683
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6701329350471497,
      "learning_rate": 0.0005983549006259393,
      "loss": 2.6675,
      "step": 7684
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6009026169776917,
      "learning_rate": 0.0005983544728016939,
      "loss": 2.4936,
      "step": 7685
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6870402693748474,
      "learning_rate": 0.0005983540449219787,
      "loss": 2.4709,
      "step": 7686
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.7143729329109192,
      "learning_rate": 0.0005983536169867938,
      "loss": 2.5807,
      "step": 7687
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6898208260536194,
      "learning_rate": 0.0005983531889961395,
      "loss": 2.5424,
      "step": 7688
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.690920352935791,
      "learning_rate": 0.0005983527609500156,
      "loss": 2.6143,
      "step": 7689
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.7866610288619995,
      "learning_rate": 0.0005983523328484222,
      "loss": 2.5583,
      "step": 7690
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.7558379769325256,
      "learning_rate": 0.0005983519046913596,
      "loss": 2.7381,
      "step": 7691
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.8136467337608337,
      "learning_rate": 0.0005983514764788276,
      "loss": 2.6225,
      "step": 7692
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6821946501731873,
      "learning_rate": 0.0005983510482108266,
      "loss": 2.3309,
      "step": 7693
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6700845956802368,
      "learning_rate": 0.0005983506198873565,
      "loss": 2.3816,
      "step": 7694
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6579975485801697,
      "learning_rate": 0.0005983501915084173,
      "loss": 2.4831,
      "step": 7695
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.7831387519836426,
      "learning_rate": 0.0005983497630740091,
      "loss": 2.3659,
      "step": 7696
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6770161986351013,
      "learning_rate": 0.0005983493345841322,
      "loss": 2.8182,
      "step": 7697
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6615830659866333,
      "learning_rate": 0.0005983489060387865,
      "loss": 2.5532,
      "step": 7698
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6994954347610474,
      "learning_rate": 0.000598348477437972,
      "loss": 2.4358,
      "step": 7699
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6502992510795593,
      "learning_rate": 0.000598348048781689,
      "loss": 2.8247,
      "step": 7700
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.640091061592102,
      "learning_rate": 0.0005983476200699374,
      "loss": 2.3581,
      "step": 7701
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6652323007583618,
      "learning_rate": 0.0005983471913027174,
      "loss": 2.3763,
      "step": 7702
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6000769138336182,
      "learning_rate": 0.000598346762480029,
      "loss": 2.5353,
      "step": 7703
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6536927223205566,
      "learning_rate": 0.0005983463336018723,
      "loss": 2.6235,
      "step": 7704
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6459961533546448,
      "learning_rate": 0.0005983459046682475,
      "loss": 2.5312,
      "step": 7705
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.7450165748596191,
      "learning_rate": 0.0005983454756791544,
      "loss": 2.6072,
      "step": 7706
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6795887351036072,
      "learning_rate": 0.0005983450466345932,
      "loss": 2.3313,
      "step": 7707
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6550591588020325,
      "learning_rate": 0.0005983446175345642,
      "loss": 2.3814,
      "step": 7708
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.670367419719696,
      "learning_rate": 0.0005983441883790672,
      "loss": 2.545,
      "step": 7709
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6063506603240967,
      "learning_rate": 0.0005983437591681024,
      "loss": 2.5186,
      "step": 7710
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.636911153793335,
      "learning_rate": 0.0005983433299016699,
      "loss": 2.6365,
      "step": 7711
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.7061026096343994,
      "learning_rate": 0.0005983429005797697,
      "loss": 2.4991,
      "step": 7712
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6668699979782104,
      "learning_rate": 0.0005983424712024019,
      "loss": 2.5267,
      "step": 7713
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.685069739818573,
      "learning_rate": 0.0005983420417695667,
      "loss": 2.4081,
      "step": 7714
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6141812801361084,
      "learning_rate": 0.0005983416122812641,
      "loss": 2.4198,
      "step": 7715
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6088850498199463,
      "learning_rate": 0.000598341182737494,
      "loss": 2.5104,
      "step": 7716
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6567775011062622,
      "learning_rate": 0.0005983407531382568,
      "loss": 2.4442,
      "step": 7717
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.7423844933509827,
      "learning_rate": 0.0005983403234835524,
      "loss": 2.5502,
      "step": 7718
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.0233067274093628,
      "learning_rate": 0.0005983398937733808,
      "loss": 2.3011,
      "step": 7719
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6653627753257751,
      "learning_rate": 0.0005983394640077422,
      "loss": 2.4736,
      "step": 7720
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.7140088081359863,
      "learning_rate": 0.0005983390341866367,
      "loss": 2.6445,
      "step": 7721
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6262014508247375,
      "learning_rate": 0.0005983386043100644,
      "loss": 2.5436,
      "step": 7722
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.7228027582168579,
      "learning_rate": 0.0005983381743780252,
      "loss": 2.4288,
      "step": 7723
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6484125256538391,
      "learning_rate": 0.0005983377443905194,
      "loss": 2.6932,
      "step": 7724
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6736935377120972,
      "learning_rate": 0.0005983373143475469,
      "loss": 2.4167,
      "step": 7725
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.7599935531616211,
      "learning_rate": 0.000598336884249108,
      "loss": 2.6338,
      "step": 7726
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6222227215766907,
      "learning_rate": 0.0005983364540952025,
      "loss": 2.4067,
      "step": 7727
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6190105080604553,
      "learning_rate": 0.0005983360238858307,
      "loss": 2.4086,
      "step": 7728
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6288850903511047,
      "learning_rate": 0.0005983355936209926,
      "loss": 2.515,
      "step": 7729
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6000584363937378,
      "learning_rate": 0.0005983351633006882,
      "loss": 2.5509,
      "step": 7730
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6184216141700745,
      "learning_rate": 0.0005983347329249178,
      "loss": 2.4815,
      "step": 7731
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.7347620129585266,
      "learning_rate": 0.0005983343024936812,
      "loss": 2.4651,
      "step": 7732
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6347574591636658,
      "learning_rate": 0.0005983338720069787,
      "loss": 2.7048,
      "step": 7733
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6597171425819397,
      "learning_rate": 0.0005983334414648103,
      "loss": 2.6596,
      "step": 7734
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6652977466583252,
      "learning_rate": 0.000598333010867176,
      "loss": 2.3301,
      "step": 7735
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6549595594406128,
      "learning_rate": 0.000598332580214076,
      "loss": 2.3131,
      "step": 7736
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.7295570373535156,
      "learning_rate": 0.0005983321495055104,
      "loss": 2.3374,
      "step": 7737
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6785460710525513,
      "learning_rate": 0.0005983317187414792,
      "loss": 2.4112,
      "step": 7738
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6081081032752991,
      "learning_rate": 0.0005983312879219825,
      "loss": 2.3894,
      "step": 7739
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.7035863399505615,
      "learning_rate": 0.0005983308570470204,
      "loss": 2.4081,
      "step": 7740
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6939929723739624,
      "learning_rate": 0.0005983304261165929,
      "loss": 2.7658,
      "step": 7741
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6114177107810974,
      "learning_rate": 0.0005983299951307002,
      "loss": 2.4349,
      "step": 7742
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6254716515541077,
      "learning_rate": 0.0005983295640893423,
      "loss": 2.4025,
      "step": 7743
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6532037258148193,
      "learning_rate": 0.0005983291329925193,
      "loss": 2.4639,
      "step": 7744
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6659759879112244,
      "learning_rate": 0.0005983287018402313,
      "loss": 2.6667,
      "step": 7745
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6369481682777405,
      "learning_rate": 0.0005983282706324783,
      "loss": 2.5125,
      "step": 7746
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.762669026851654,
      "learning_rate": 0.0005983278393692606,
      "loss": 2.4561,
      "step": 7747
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.7328315377235413,
      "learning_rate": 0.0005983274080505779,
      "loss": 2.5209,
      "step": 7748
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.7103978991508484,
      "learning_rate": 0.0005983269766764306,
      "loss": 2.694,
      "step": 7749
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.7836076617240906,
      "learning_rate": 0.0005983265452468188,
      "loss": 2.4761,
      "step": 7750
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.5728952884674072,
      "learning_rate": 0.0005983261137617424,
      "loss": 2.3439,
      "step": 7751
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.044461965560913,
      "learning_rate": 0.0005983256822212014,
      "loss": 2.4148,
      "step": 7752
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6981023550033569,
      "learning_rate": 0.0005983252506251962,
      "loss": 2.4594,
      "step": 7753
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6941831111907959,
      "learning_rate": 0.0005983248189737266,
      "loss": 2.4591,
      "step": 7754
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6490741968154907,
      "learning_rate": 0.0005983243872667928,
      "loss": 2.4954,
      "step": 7755
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6508684754371643,
      "learning_rate": 0.000598323955504395,
      "loss": 2.7081,
      "step": 7756
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.669628381729126,
      "learning_rate": 0.0005983235236865329,
      "loss": 2.5911,
      "step": 7757
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.7111071944236755,
      "learning_rate": 0.0005983230918132071,
      "loss": 2.5544,
      "step": 7758
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6250354647636414,
      "learning_rate": 0.0005983226598844172,
      "loss": 2.6123,
      "step": 7759
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6594091653823853,
      "learning_rate": 0.0005983222279001636,
      "loss": 2.5747,
      "step": 7760
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.7442914843559265,
      "learning_rate": 0.0005983217958604462,
      "loss": 2.2333,
      "step": 7761
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6853240132331848,
      "learning_rate": 0.0005983213637652652,
      "loss": 2.6327,
      "step": 7762
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6351997256278992,
      "learning_rate": 0.0005983209316146205,
      "loss": 2.5357,
      "step": 7763
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.7410277724266052,
      "learning_rate": 0.0005983204994085124,
      "loss": 2.3102,
      "step": 7764
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.8831372261047363,
      "learning_rate": 0.000598320067146941,
      "loss": 2.472,
      "step": 7765
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6677225828170776,
      "learning_rate": 0.0005983196348299061,
      "loss": 2.7459,
      "step": 7766
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.7572838664054871,
      "learning_rate": 0.000598319202457408,
      "loss": 2.6287,
      "step": 7767
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.883462131023407,
      "learning_rate": 0.0005983187700294467,
      "loss": 2.4038,
      "step": 7768
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.802269458770752,
      "learning_rate": 0.0005983183375460223,
      "loss": 2.6596,
      "step": 7769
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6608814597129822,
      "learning_rate": 0.000598317905007135,
      "loss": 2.6802,
      "step": 7770
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.9060222506523132,
      "learning_rate": 0.0005983174724127847,
      "loss": 2.5986,
      "step": 7771
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.7443188428878784,
      "learning_rate": 0.0005983170397629716,
      "loss": 2.7619,
      "step": 7772
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6416254043579102,
      "learning_rate": 0.0005983166070576957,
      "loss": 2.6633,
      "step": 7773
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.9159356355667114,
      "learning_rate": 0.0005983161742969571,
      "loss": 2.7274,
      "step": 7774
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.7256761193275452,
      "learning_rate": 0.0005983157414807558,
      "loss": 2.3693,
      "step": 7775
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6151610016822815,
      "learning_rate": 0.0005983153086090921,
      "loss": 2.6285,
      "step": 7776
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.9064498543739319,
      "learning_rate": 0.0005983148756819659,
      "loss": 2.6121,
      "step": 7777
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6714734435081482,
      "learning_rate": 0.0005983144426993773,
      "loss": 2.3334,
      "step": 7778
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.7043139338493347,
      "learning_rate": 0.0005983140096613265,
      "loss": 2.5299,
      "step": 7779
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6945682168006897,
      "learning_rate": 0.0005983135765678134,
      "loss": 2.7005,
      "step": 7780
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.611508846282959,
      "learning_rate": 0.0005983131434188384,
      "loss": 2.5165,
      "step": 7781
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.5938189625740051,
      "learning_rate": 0.0005983127102144011,
      "loss": 2.5909,
      "step": 7782
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6682692170143127,
      "learning_rate": 0.0005983122769545019,
      "loss": 2.5739,
      "step": 7783
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.643032431602478,
      "learning_rate": 0.0005983118436391409,
      "loss": 2.5692,
      "step": 7784
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6280331015586853,
      "learning_rate": 0.000598311410268318,
      "loss": 2.4933,
      "step": 7785
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6389142274856567,
      "learning_rate": 0.0005983109768420335,
      "loss": 2.6545,
      "step": 7786
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.5904707908630371,
      "learning_rate": 0.0005983105433602872,
      "loss": 2.4849,
      "step": 7787
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6356106996536255,
      "learning_rate": 0.0005983101098230795,
      "loss": 2.4892,
      "step": 7788
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6704245805740356,
      "learning_rate": 0.0005983096762304101,
      "loss": 2.5383,
      "step": 7789
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6498720645904541,
      "learning_rate": 0.0005983092425822796,
      "loss": 2.6449,
      "step": 7790
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.659474790096283,
      "learning_rate": 0.0005983088088786875,
      "loss": 2.5385,
      "step": 7791
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6674103140830994,
      "learning_rate": 0.0005983083751196342,
      "loss": 2.5898,
      "step": 7792
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6271387338638306,
      "learning_rate": 0.0005983079413051199,
      "loss": 2.3262,
      "step": 7793
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6873382925987244,
      "learning_rate": 0.0005983075074351444,
      "loss": 2.3848,
      "step": 7794
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.619705080986023,
      "learning_rate": 0.0005983070735097079,
      "loss": 2.4003,
      "step": 7795
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6471266150474548,
      "learning_rate": 0.0005983066395288106,
      "loss": 2.5202,
      "step": 7796
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6575192809104919,
      "learning_rate": 0.0005983062054924524,
      "loss": 2.7453,
      "step": 7797
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6937553286552429,
      "learning_rate": 0.0005983057714006334,
      "loss": 2.654,
      "step": 7798
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6672270894050598,
      "learning_rate": 0.0005983053372533539,
      "loss": 2.565,
      "step": 7799
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.688686192035675,
      "learning_rate": 0.0005983049030506136,
      "loss": 2.5583,
      "step": 7800
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6864151358604431,
      "learning_rate": 0.0005983044687924128,
      "loss": 2.5492,
      "step": 7801
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.7386718988418579,
      "learning_rate": 0.0005983040344787516,
      "loss": 2.5479,
      "step": 7802
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.609946072101593,
      "learning_rate": 0.0005983036001096302,
      "loss": 2.435,
      "step": 7803
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6663846373558044,
      "learning_rate": 0.0005983031656850484,
      "loss": 2.569,
      "step": 7804
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6760094165802002,
      "learning_rate": 0.0005983027312050064,
      "loss": 2.4415,
      "step": 7805
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6042541861534119,
      "learning_rate": 0.0005983022966695043,
      "loss": 2.2117,
      "step": 7806
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6397817134857178,
      "learning_rate": 0.0005983018620785422,
      "loss": 2.6699,
      "step": 7807
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.69804847240448,
      "learning_rate": 0.0005983014274321201,
      "loss": 2.7154,
      "step": 7808
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.631672739982605,
      "learning_rate": 0.0005983009927302381,
      "loss": 2.4976,
      "step": 7809
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6307538747787476,
      "learning_rate": 0.0005983005579728965,
      "loss": 2.4724,
      "step": 7810
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6809224486351013,
      "learning_rate": 0.000598300123160095,
      "loss": 2.6719,
      "step": 7811
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6593120098114014,
      "learning_rate": 0.000598299688291834,
      "loss": 2.6269,
      "step": 7812
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6982907056808472,
      "learning_rate": 0.0005982992533681133,
      "loss": 2.8077,
      "step": 7813
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6401973962783813,
      "learning_rate": 0.0005982988183889333,
      "loss": 2.4144,
      "step": 7814
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6489115357398987,
      "learning_rate": 0.0005982983833542939,
      "loss": 2.4094,
      "step": 7815
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6621511578559875,
      "learning_rate": 0.0005982979482641952,
      "loss": 2.4382,
      "step": 7816
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.7180936336517334,
      "learning_rate": 0.0005982975131186371,
      "loss": 2.5153,
      "step": 7817
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6547003388404846,
      "learning_rate": 0.0005982970779176201,
      "loss": 2.5138,
      "step": 7818
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6472454071044922,
      "learning_rate": 0.0005982966426611439,
      "loss": 2.2473,
      "step": 7819
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6386705636978149,
      "learning_rate": 0.0005982962073492089,
      "loss": 2.4894,
      "step": 7820
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6133588552474976,
      "learning_rate": 0.0005982957719818148,
      "loss": 2.3937,
      "step": 7821
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6570518016815186,
      "learning_rate": 0.000598295336558962,
      "loss": 2.4675,
      "step": 7822
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6616770029067993,
      "learning_rate": 0.0005982949010806504,
      "loss": 2.6363,
      "step": 7823
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6670783758163452,
      "learning_rate": 0.0005982944655468802,
      "loss": 2.4215,
      "step": 7824
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6671282052993774,
      "learning_rate": 0.0005982940299576514,
      "loss": 2.2719,
      "step": 7825
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6828237771987915,
      "learning_rate": 0.0005982935943129641,
      "loss": 2.5934,
      "step": 7826
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6080722808837891,
      "learning_rate": 0.0005982931586128185,
      "loss": 2.5545,
      "step": 7827
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6541597843170166,
      "learning_rate": 0.0005982927228572143,
      "loss": 2.5334,
      "step": 7828
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6492828130722046,
      "learning_rate": 0.0005982922870461521,
      "loss": 2.5262,
      "step": 7829
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6139989495277405,
      "learning_rate": 0.0005982918511796317,
      "loss": 2.6047,
      "step": 7830
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6454970836639404,
      "learning_rate": 0.0005982914152576532,
      "loss": 2.5257,
      "step": 7831
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6228991746902466,
      "learning_rate": 0.0005982909792802166,
      "loss": 2.3296,
      "step": 7832
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6292814016342163,
      "learning_rate": 0.0005982905432473221,
      "loss": 2.4023,
      "step": 7833
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.8308039307594299,
      "learning_rate": 0.0005982901071589699,
      "loss": 2.6355,
      "step": 7834
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6457570791244507,
      "learning_rate": 0.0005982896710151598,
      "loss": 2.4252,
      "step": 7835
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6965077519416809,
      "learning_rate": 0.0005982892348158921,
      "loss": 2.6682,
      "step": 7836
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6994339823722839,
      "learning_rate": 0.0005982887985611667,
      "loss": 2.7842,
      "step": 7837
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.658184289932251,
      "learning_rate": 0.0005982883622509839,
      "loss": 2.3565,
      "step": 7838
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6793228983879089,
      "learning_rate": 0.0005982879258853436,
      "loss": 2.3657,
      "step": 7839
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.5821877121925354,
      "learning_rate": 0.0005982874894642459,
      "loss": 2.5845,
      "step": 7840
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6526559591293335,
      "learning_rate": 0.000598287052987691,
      "loss": 2.5477,
      "step": 7841
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.781440794467926,
      "learning_rate": 0.0005982866164556789,
      "loss": 2.4138,
      "step": 7842
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.7025645971298218,
      "learning_rate": 0.0005982861798682097,
      "loss": 2.6949,
      "step": 7843
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6615837812423706,
      "learning_rate": 0.0005982857432252833,
      "loss": 2.4575,
      "step": 7844
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6320545673370361,
      "learning_rate": 0.0005982853065269002,
      "loss": 2.7751,
      "step": 7845
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6212462782859802,
      "learning_rate": 0.0005982848697730601,
      "loss": 2.6998,
      "step": 7846
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.8304804563522339,
      "learning_rate": 0.0005982844329637633,
      "loss": 2.5529,
      "step": 7847
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.7173497080802917,
      "learning_rate": 0.0005982839960990097,
      "loss": 2.7379,
      "step": 7848
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.87919682264328,
      "learning_rate": 0.0005982835591787994,
      "loss": 2.6318,
      "step": 7849
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6012787818908691,
      "learning_rate": 0.0005982831222031327,
      "loss": 2.5549,
      "step": 7850
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6493604779243469,
      "learning_rate": 0.0005982826851720095,
      "loss": 2.5541,
      "step": 7851
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6782113909721375,
      "learning_rate": 0.0005982822480854299,
      "loss": 2.4613,
      "step": 7852
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.640250563621521,
      "learning_rate": 0.000598281810943394,
      "loss": 2.4612,
      "step": 7853
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6843576431274414,
      "learning_rate": 0.0005982813737459019,
      "loss": 2.64,
      "step": 7854
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.5976601243019104,
      "learning_rate": 0.0005982809364929537,
      "loss": 2.4472,
      "step": 7855
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6746484637260437,
      "learning_rate": 0.0005982804991845494,
      "loss": 2.3736,
      "step": 7856
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6773977875709534,
      "learning_rate": 0.0005982800618206891,
      "loss": 2.7788,
      "step": 7857
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6805411577224731,
      "learning_rate": 0.0005982796244013729,
      "loss": 2.5507,
      "step": 7858
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6165257692337036,
      "learning_rate": 0.0005982791869266009,
      "loss": 2.5038,
      "step": 7859
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.9125310778617859,
      "learning_rate": 0.0005982787493963732,
      "loss": 2.3807,
      "step": 7860
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6449803709983826,
      "learning_rate": 0.0005982783118106898,
      "loss": 2.5213,
      "step": 7861
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6449169516563416,
      "learning_rate": 0.0005982778741695509,
      "loss": 2.478,
      "step": 7862
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.7161549925804138,
      "learning_rate": 0.0005982774364729564,
      "loss": 2.3502,
      "step": 7863
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6648023128509521,
      "learning_rate": 0.0005982769987209067,
      "loss": 2.5488,
      "step": 7864
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6438645720481873,
      "learning_rate": 0.0005982765609134015,
      "loss": 2.5209,
      "step": 7865
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.5686737298965454,
      "learning_rate": 0.0005982761230504412,
      "loss": 2.4916,
      "step": 7866
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6720802187919617,
      "learning_rate": 0.0005982756851320255,
      "loss": 2.4076,
      "step": 7867
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.706577718257904,
      "learning_rate": 0.0005982752471581549,
      "loss": 2.6059,
      "step": 7868
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6679276823997498,
      "learning_rate": 0.0005982748091288292,
      "loss": 2.4923,
      "step": 7869
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6273159980773926,
      "learning_rate": 0.0005982743710440488,
      "loss": 2.7067,
      "step": 7870
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6349222660064697,
      "learning_rate": 0.0005982739329038133,
      "loss": 2.5583,
      "step": 7871
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.932924211025238,
      "learning_rate": 0.0005982734947081232,
      "loss": 2.599,
      "step": 7872
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.823085367679596,
      "learning_rate": 0.0005982730564569784,
      "loss": 2.6018,
      "step": 7873
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.184187889099121,
      "learning_rate": 0.0005982726181503789,
      "loss": 2.362,
      "step": 7874
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6497437953948975,
      "learning_rate": 0.000598272179788325,
      "loss": 2.6084,
      "step": 7875
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.785378634929657,
      "learning_rate": 0.0005982717413708167,
      "loss": 2.4178,
      "step": 7876
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.788683295249939,
      "learning_rate": 0.000598271302897854,
      "loss": 2.4153,
      "step": 7877
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.7509317398071289,
      "learning_rate": 0.0005982708643694371,
      "loss": 2.4953,
      "step": 7878
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.7449393272399902,
      "learning_rate": 0.0005982704257855659,
      "loss": 2.5084,
      "step": 7879
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.7263215780258179,
      "learning_rate": 0.0005982699871462406,
      "loss": 2.7238,
      "step": 7880
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.616098165512085,
      "learning_rate": 0.0005982695484514613,
      "loss": 2.6755,
      "step": 7881
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.7627358436584473,
      "learning_rate": 0.0005982691097012282,
      "loss": 2.4538,
      "step": 7882
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.8448833227157593,
      "learning_rate": 0.000598268670895541,
      "loss": 2.5982,
      "step": 7883
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.5948434472084045,
      "learning_rate": 0.0005982682320344003,
      "loss": 2.5014,
      "step": 7884
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.8628790974617004,
      "learning_rate": 0.0005982677931178058,
      "loss": 2.4777,
      "step": 7885
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.7660025358200073,
      "learning_rate": 0.0005982673541457576,
      "loss": 2.5599,
      "step": 7886
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6314800381660461,
      "learning_rate": 0.0005982669151182559,
      "loss": 2.5617,
      "step": 7887
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6687566041946411,
      "learning_rate": 0.0005982664760353008,
      "loss": 2.5754,
      "step": 7888
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6263685822486877,
      "learning_rate": 0.0005982660368968923,
      "loss": 2.5374,
      "step": 7889
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6425735950469971,
      "learning_rate": 0.0005982655977030305,
      "loss": 2.5106,
      "step": 7890
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.7432887554168701,
      "learning_rate": 0.0005982651584537155,
      "loss": 2.5842,
      "step": 7891
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6616294980049133,
      "learning_rate": 0.0005982647191489474,
      "loss": 2.7178,
      "step": 7892
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6868612766265869,
      "learning_rate": 0.0005982642797887262,
      "loss": 2.5126,
      "step": 7893
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.8324351906776428,
      "learning_rate": 0.0005982638403730521,
      "loss": 2.5058,
      "step": 7894
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6207273602485657,
      "learning_rate": 0.0005982634009019252,
      "loss": 2.6106,
      "step": 7895
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6466776728630066,
      "learning_rate": 0.0005982629613753453,
      "loss": 2.396,
      "step": 7896
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.7499933838844299,
      "learning_rate": 0.0005982625217933129,
      "loss": 2.5266,
      "step": 7897
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.7059639096260071,
      "learning_rate": 0.0005982620821558277,
      "loss": 2.6459,
      "step": 7898
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6897366046905518,
      "learning_rate": 0.00059826164246289,
      "loss": 2.568,
      "step": 7899
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6696028709411621,
      "learning_rate": 0.0005982612027144999,
      "loss": 2.7168,
      "step": 7900
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6099708676338196,
      "learning_rate": 0.0005982607629106574,
      "loss": 2.4435,
      "step": 7901
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6898617744445801,
      "learning_rate": 0.0005982603230513624,
      "loss": 2.365,
      "step": 7902
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.7055236101150513,
      "learning_rate": 0.0005982598831366153,
      "loss": 2.4743,
      "step": 7903
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6795646548271179,
      "learning_rate": 0.0005982594431664162,
      "loss": 2.4166,
      "step": 7904
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6497597098350525,
      "learning_rate": 0.0005982590031407648,
      "loss": 2.8206,
      "step": 7905
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6858435273170471,
      "learning_rate": 0.0005982585630596617,
      "loss": 2.4438,
      "step": 7906
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.7413376569747925,
      "learning_rate": 0.0005982581229231065,
      "loss": 2.8694,
      "step": 7907
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6571233868598938,
      "learning_rate": 0.0005982576827310996,
      "loss": 2.6114,
      "step": 7908
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6269723176956177,
      "learning_rate": 0.0005982572424836408,
      "loss": 2.4544,
      "step": 7909
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6454687714576721,
      "learning_rate": 0.0005982568021807305,
      "loss": 2.5606,
      "step": 7910
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6116252541542053,
      "learning_rate": 0.0005982563618223685,
      "loss": 2.5536,
      "step": 7911
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6593345403671265,
      "learning_rate": 0.0005982559214085552,
      "loss": 2.4119,
      "step": 7912
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6487118005752563,
      "learning_rate": 0.0005982554809392902,
      "loss": 2.5587,
      "step": 7913
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.5957862138748169,
      "learning_rate": 0.0005982550404145742,
      "loss": 2.5239,
      "step": 7914
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6521947979927063,
      "learning_rate": 0.0005982545998344067,
      "loss": 2.3973,
      "step": 7915
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6582186818122864,
      "learning_rate": 0.0005982541591987883,
      "loss": 2.5421,
      "step": 7916
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6691108345985413,
      "learning_rate": 0.0005982537185077186,
      "loss": 2.4035,
      "step": 7917
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6010974049568176,
      "learning_rate": 0.000598253277761198,
      "loss": 2.5682,
      "step": 7918
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6728497743606567,
      "learning_rate": 0.0005982528369592265,
      "loss": 2.6551,
      "step": 7919
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6999430060386658,
      "learning_rate": 0.0005982523961018041,
      "loss": 2.709,
      "step": 7920
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.667329728603363,
      "learning_rate": 0.0005982519551889309,
      "loss": 2.5786,
      "step": 7921
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6215265989303589,
      "learning_rate": 0.0005982515142206072,
      "loss": 2.5504,
      "step": 7922
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.7486138939857483,
      "learning_rate": 0.0005982510731968328,
      "loss": 2.5834,
      "step": 7923
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6442258358001709,
      "learning_rate": 0.000598250632117608,
      "loss": 2.3994,
      "step": 7924
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6432120203971863,
      "learning_rate": 0.0005982501909829326,
      "loss": 2.293,
      "step": 7925
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6613720655441284,
      "learning_rate": 0.0005982497497928071,
      "loss": 2.5887,
      "step": 7926
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.7576360702514648,
      "learning_rate": 0.0005982493085472312,
      "loss": 2.6953,
      "step": 7927
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6178651452064514,
      "learning_rate": 0.0005982488672462051,
      "loss": 2.6069,
      "step": 7928
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.9051840305328369,
      "learning_rate": 0.0005982484258897289,
      "loss": 2.7581,
      "step": 7929
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6816631555557251,
      "learning_rate": 0.0005982479844778028,
      "loss": 2.4511,
      "step": 7930
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6751167178153992,
      "learning_rate": 0.0005982475430104267,
      "loss": 2.3993,
      "step": 7931
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.9051470160484314,
      "learning_rate": 0.0005982471014876007,
      "loss": 2.645,
      "step": 7932
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.7793546915054321,
      "learning_rate": 0.000598246659909325,
      "loss": 2.6341,
      "step": 7933
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6224279999732971,
      "learning_rate": 0.0005982462182755996,
      "loss": 2.4938,
      "step": 7934
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.7784727811813354,
      "learning_rate": 0.0005982457765864246,
      "loss": 2.3917,
      "step": 7935
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.7635826468467712,
      "learning_rate": 0.0005982453348418001,
      "loss": 2.534,
      "step": 7936
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.5822810530662537,
      "learning_rate": 0.000598244893041726,
      "loss": 2.4837,
      "step": 7937
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.7244141697883606,
      "learning_rate": 0.0005982444511862027,
      "loss": 2.6432,
      "step": 7938
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6620782613754272,
      "learning_rate": 0.0005982440092752302,
      "loss": 2.6301,
      "step": 7939
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.7426003813743591,
      "learning_rate": 0.0005982435673088084,
      "loss": 2.4129,
      "step": 7940
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6648527383804321,
      "learning_rate": 0.0005982431252869375,
      "loss": 2.1794,
      "step": 7941
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.7807072401046753,
      "learning_rate": 0.0005982426832096176,
      "loss": 2.5194,
      "step": 7942
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6551426649093628,
      "learning_rate": 0.0005982422410768487,
      "loss": 2.5286,
      "step": 7943
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6432039141654968,
      "learning_rate": 0.000598241798888631,
      "loss": 2.3636,
      "step": 7944
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6164120435714722,
      "learning_rate": 0.0005982413566449645,
      "loss": 2.6154,
      "step": 7945
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6781696677207947,
      "learning_rate": 0.0005982409143458492,
      "loss": 2.5079,
      "step": 7946
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.607656717300415,
      "learning_rate": 0.0005982404719912853,
      "loss": 2.3909,
      "step": 7947
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.615543007850647,
      "learning_rate": 0.0005982400295812731,
      "loss": 2.6031,
      "step": 7948
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.752803385257721,
      "learning_rate": 0.0005982395871158122,
      "loss": 2.5664,
      "step": 7949
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6750388741493225,
      "learning_rate": 0.0005982391445949031,
      "loss": 2.6791,
      "step": 7950
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6274302005767822,
      "learning_rate": 0.0005982387020185457,
      "loss": 2.5157,
      "step": 7951
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6267619132995605,
      "learning_rate": 0.0005982382593867399,
      "loss": 2.6861,
      "step": 7952
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.0833795070648193,
      "learning_rate": 0.0005982378166994862,
      "loss": 2.3729,
      "step": 7953
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6897007822990417,
      "learning_rate": 0.0005982373739567843,
      "loss": 2.5051,
      "step": 7954
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6918200850486755,
      "learning_rate": 0.0005982369311586346,
      "loss": 2.6139,
      "step": 7955
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.7222954034805298,
      "learning_rate": 0.000598236488305037,
      "loss": 2.5321,
      "step": 7956
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.7102094888687134,
      "learning_rate": 0.0005982360453959915,
      "loss": 2.3831,
      "step": 7957
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6550843715667725,
      "learning_rate": 0.0005982356024314983,
      "loss": 2.3307,
      "step": 7958
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.7319834232330322,
      "learning_rate": 0.0005982351594115575,
      "loss": 2.6134,
      "step": 7959
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6411528587341309,
      "learning_rate": 0.0005982347163361691,
      "loss": 2.6104,
      "step": 7960
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6870290040969849,
      "learning_rate": 0.0005982342732053334,
      "loss": 2.5487,
      "step": 7961
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.7023848295211792,
      "learning_rate": 0.0005982338300190502,
      "loss": 2.7291,
      "step": 7962
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.7414310574531555,
      "learning_rate": 0.0005982333867773197,
      "loss": 2.5083,
      "step": 7963
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6436957120895386,
      "learning_rate": 0.0005982329434801421,
      "loss": 2.6965,
      "step": 7964
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6497308015823364,
      "learning_rate": 0.0005982325001275172,
      "loss": 2.268,
      "step": 7965
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.7564697265625,
      "learning_rate": 0.0005982320567194454,
      "loss": 2.6983,
      "step": 7966
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6719186902046204,
      "learning_rate": 0.0005982316132559265,
      "loss": 2.3372,
      "step": 7967
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6267978549003601,
      "learning_rate": 0.0005982311697369607,
      "loss": 2.5469,
      "step": 7968
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6115306615829468,
      "learning_rate": 0.0005982307261625482,
      "loss": 2.5972,
      "step": 7969
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.5973172187805176,
      "learning_rate": 0.0005982302825326889,
      "loss": 2.362,
      "step": 7970
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.7300028800964355,
      "learning_rate": 0.000598229838847383,
      "loss": 2.5335,
      "step": 7971
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6455032825469971,
      "learning_rate": 0.0005982293951066305,
      "loss": 2.2885,
      "step": 7972
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6328946948051453,
      "learning_rate": 0.0005982289513104316,
      "loss": 2.4688,
      "step": 7973
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6143890619277954,
      "learning_rate": 0.0005982285074587863,
      "loss": 2.4994,
      "step": 7974
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.7893555164337158,
      "learning_rate": 0.0005982280635516946,
      "loss": 2.5126,
      "step": 7975
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6083356142044067,
      "learning_rate": 0.0005982276195891568,
      "loss": 2.4749,
      "step": 7976
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6913147568702698,
      "learning_rate": 0.0005982271755711728,
      "loss": 2.4536,
      "step": 7977
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.8155443072319031,
      "learning_rate": 0.0005982267314977428,
      "loss": 2.5212,
      "step": 7978
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.5782451629638672,
      "learning_rate": 0.0005982262873688668,
      "loss": 2.4478,
      "step": 7979
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.9428514242172241,
      "learning_rate": 0.0005982258431845447,
      "loss": 2.7002,
      "step": 7980
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.857058584690094,
      "learning_rate": 0.000598225398944777,
      "loss": 2.3159,
      "step": 7981
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.609947681427002,
      "learning_rate": 0.0005982249546495636,
      "loss": 2.6087,
      "step": 7982
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.715766191482544,
      "learning_rate": 0.0005982245102989043,
      "loss": 2.5091,
      "step": 7983
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.9491913318634033,
      "learning_rate": 0.0005982240658927997,
      "loss": 2.4007,
      "step": 7984
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.661314845085144,
      "learning_rate": 0.0005982236214312496,
      "loss": 2.5962,
      "step": 7985
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.7492116093635559,
      "learning_rate": 0.0005982231769142539,
      "loss": 2.9516,
      "step": 7986
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.7165287137031555,
      "learning_rate": 0.0005982227323418131,
      "loss": 2.5564,
      "step": 7987
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.9837864637374878,
      "learning_rate": 0.0005982222877139269,
      "loss": 2.3315,
      "step": 7988
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6642559170722961,
      "learning_rate": 0.0005982218430305957,
      "loss": 2.6097,
      "step": 7989
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6992695331573486,
      "learning_rate": 0.0005982213982918192,
      "loss": 2.6325,
      "step": 7990
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.9791309833526611,
      "learning_rate": 0.0005982209534975979,
      "loss": 2.4849,
      "step": 7991
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6414692997932434,
      "learning_rate": 0.0005982205086479316,
      "loss": 2.4695,
      "step": 7992
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6979840993881226,
      "learning_rate": 0.0005982200637428206,
      "loss": 2.3962,
      "step": 7993
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.8644602298736572,
      "learning_rate": 0.0005982196187822648,
      "loss": 2.3332,
      "step": 7994
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.7410340309143066,
      "learning_rate": 0.0005982191737662643,
      "loss": 2.5419,
      "step": 7995
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6503161191940308,
      "learning_rate": 0.0005982187286948193,
      "loss": 2.4524,
      "step": 7996
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.8951786160469055,
      "learning_rate": 0.0005982182835679297,
      "loss": 2.6088,
      "step": 7997
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.7192635536193848,
      "learning_rate": 0.0005982178383855958,
      "loss": 2.247,
      "step": 7998
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.5960657000541687,
      "learning_rate": 0.0005982173931478175,
      "loss": 2.6364,
      "step": 7999
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.7092966437339783,
      "learning_rate": 0.0005982169478545951,
      "loss": 2.535,
      "step": 8000
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.7430517673492432,
      "learning_rate": 0.0005982165025059283,
      "loss": 2.6167,
      "step": 8001
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.7724313735961914,
      "learning_rate": 0.0005982160571018177,
      "loss": 2.5202,
      "step": 8002
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.7581549882888794,
      "learning_rate": 0.0005982156116422629,
      "loss": 2.6813,
      "step": 8003
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.7674843072891235,
      "learning_rate": 0.0005982151661272642,
      "loss": 2.6573,
      "step": 8004
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.7076983451843262,
      "learning_rate": 0.0005982147205568218,
      "loss": 2.7379,
      "step": 8005
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.5772421360015869,
      "learning_rate": 0.0005982142749309356,
      "loss": 2.4623,
      "step": 8006
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6322941184043884,
      "learning_rate": 0.0005982138292496056,
      "loss": 2.6149,
      "step": 8007
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.665957510471344,
      "learning_rate": 0.0005982133835128322,
      "loss": 2.7375,
      "step": 8008
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.681603193283081,
      "learning_rate": 0.0005982129377206152,
      "loss": 2.5362,
      "step": 8009
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.5924770832061768,
      "learning_rate": 0.0005982124918729549,
      "loss": 2.6498,
      "step": 8010
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6714165210723877,
      "learning_rate": 0.0005982120459698511,
      "loss": 2.548,
      "step": 8011
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.8342321515083313,
      "learning_rate": 0.0005982116000113042,
      "loss": 2.67,
      "step": 8012
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.5940458178520203,
      "learning_rate": 0.000598211153997314,
      "loss": 2.7119,
      "step": 8013
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.7664582133293152,
      "learning_rate": 0.0005982107079278808,
      "loss": 2.6203,
      "step": 8014
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.590230405330658,
      "learning_rate": 0.0005982102618030046,
      "loss": 2.4986,
      "step": 8015
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6061387658119202,
      "learning_rate": 0.0005982098156226854,
      "loss": 2.5462,
      "step": 8016
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6856030225753784,
      "learning_rate": 0.0005982093693869235,
      "loss": 2.4989,
      "step": 8017
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6486420631408691,
      "learning_rate": 0.0005982089230957188,
      "loss": 2.2754,
      "step": 8018
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.710475504398346,
      "learning_rate": 0.0005982084767490715,
      "loss": 2.362,
      "step": 8019
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6527184844017029,
      "learning_rate": 0.0005982080303469815,
      "loss": 2.4798,
      "step": 8020
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.7224491834640503,
      "learning_rate": 0.000598207583889449,
      "loss": 2.3565,
      "step": 8021
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.7134447693824768,
      "learning_rate": 0.0005982071373764741,
      "loss": 2.5535,
      "step": 8022
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.8974246978759766,
      "learning_rate": 0.0005982066908080569,
      "loss": 2.5234,
      "step": 8023
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.7009579539299011,
      "learning_rate": 0.0005982062441841975,
      "loss": 2.5756,
      "step": 8024
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.7305552363395691,
      "learning_rate": 0.0005982057975048958,
      "loss": 2.4863,
      "step": 8025
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.8226405382156372,
      "learning_rate": 0.0005982053507701521,
      "loss": 2.6786,
      "step": 8026
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6815575957298279,
      "learning_rate": 0.0005982049039799664,
      "loss": 2.5161,
      "step": 8027
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6319547891616821,
      "learning_rate": 0.0005982044571343387,
      "loss": 2.3367,
      "step": 8028
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6430752277374268,
      "learning_rate": 0.0005982040102332692,
      "loss": 2.3279,
      "step": 8029
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6445685029029846,
      "learning_rate": 0.000598203563276758,
      "loss": 2.4589,
      "step": 8030
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6889491677284241,
      "learning_rate": 0.000598203116264805,
      "loss": 2.2518,
      "step": 8031
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.668231189250946,
      "learning_rate": 0.0005982026691974105,
      "loss": 2.531,
      "step": 8032
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6185225248336792,
      "learning_rate": 0.0005982022220745745,
      "loss": 2.5231,
      "step": 8033
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6200512051582336,
      "learning_rate": 0.0005982017748962971,
      "loss": 2.3219,
      "step": 8034
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.7083032727241516,
      "learning_rate": 0.0005982013276625783,
      "loss": 2.4128,
      "step": 8035
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.684013843536377,
      "learning_rate": 0.0005982008803734183,
      "loss": 2.3717,
      "step": 8036
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.7313575148582458,
      "learning_rate": 0.0005982004330288171,
      "loss": 2.4885,
      "step": 8037
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.7142305374145508,
      "learning_rate": 0.0005981999856287748,
      "loss": 2.8397,
      "step": 8038
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6895717978477478,
      "learning_rate": 0.0005981995381732915,
      "loss": 2.3096,
      "step": 8039
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.7095958590507507,
      "learning_rate": 0.0005981990906623673,
      "loss": 2.4798,
      "step": 8040
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.7810613512992859,
      "learning_rate": 0.0005981986430960022,
      "loss": 2.4528,
      "step": 8041
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6172104477882385,
      "learning_rate": 0.0005981981954741964,
      "loss": 2.7518,
      "step": 8042
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.7154507040977478,
      "learning_rate": 0.0005981977477969499,
      "loss": 2.6198,
      "step": 8043
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6952640414237976,
      "learning_rate": 0.0005981973000642628,
      "loss": 2.4692,
      "step": 8044
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6235889792442322,
      "learning_rate": 0.0005981968522761352,
      "loss": 2.5588,
      "step": 8045
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6280491948127747,
      "learning_rate": 0.0005981964044325672,
      "loss": 2.4027,
      "step": 8046
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6165494322776794,
      "learning_rate": 0.0005981959565335589,
      "loss": 2.767,
      "step": 8047
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6540867686271667,
      "learning_rate": 0.0005981955085791103,
      "loss": 2.5361,
      "step": 8048
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6279000639915466,
      "learning_rate": 0.0005981950605692214,
      "loss": 2.4675,
      "step": 8049
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.5966172814369202,
      "learning_rate": 0.0005981946125038927,
      "loss": 2.2585,
      "step": 8050
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6561496257781982,
      "learning_rate": 0.0005981941643831238,
      "loss": 2.7023,
      "step": 8051
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6291009187698364,
      "learning_rate": 0.000598193716206915,
      "loss": 2.4489,
      "step": 8052
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.689629852771759,
      "learning_rate": 0.0005981932679752662,
      "loss": 2.535,
      "step": 8053
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6783906817436218,
      "learning_rate": 0.0005981928196881779,
      "loss": 2.5051,
      "step": 8054
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6467189788818359,
      "learning_rate": 0.0005981923713456498,
      "loss": 2.5507,
      "step": 8055
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6806706190109253,
      "learning_rate": 0.0005981919229476821,
      "loss": 2.6353,
      "step": 8056
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6507692933082581,
      "learning_rate": 0.0005981914744942749,
      "loss": 2.7267,
      "step": 8057
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6562330722808838,
      "learning_rate": 0.0005981910259854284,
      "loss": 2.4172,
      "step": 8058
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6214321255683899,
      "learning_rate": 0.0005981905774211424,
      "loss": 2.4833,
      "step": 8059
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6812872290611267,
      "learning_rate": 0.0005981901288014172,
      "loss": 2.4813,
      "step": 8060
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6275060772895813,
      "learning_rate": 0.0005981896801262529,
      "loss": 2.4567,
      "step": 8061
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.7296592593193054,
      "learning_rate": 0.0005981892313956493,
      "loss": 2.6736,
      "step": 8062
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.712795615196228,
      "learning_rate": 0.0005981887826096069,
      "loss": 2.5779,
      "step": 8063
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6376274228096008,
      "learning_rate": 0.0005981883337681256,
      "loss": 2.6643,
      "step": 8064
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6532543301582336,
      "learning_rate": 0.0005981878848712053,
      "loss": 2.5005,
      "step": 8065
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6554675698280334,
      "learning_rate": 0.0005981874359188464,
      "loss": 2.4072,
      "step": 8066
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6267919540405273,
      "learning_rate": 0.0005981869869110487,
      "loss": 2.5461,
      "step": 8067
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.7150026559829712,
      "learning_rate": 0.0005981865378478124,
      "loss": 2.5184,
      "step": 8068
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.624739408493042,
      "learning_rate": 0.0005981860887291377,
      "loss": 2.5378,
      "step": 8069
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6592607498168945,
      "learning_rate": 0.0005981856395550245,
      "loss": 2.5036,
      "step": 8070
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.7560037970542908,
      "learning_rate": 0.0005981851903254729,
      "loss": 2.4842,
      "step": 8071
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.7789131999015808,
      "learning_rate": 0.0005981847410404832,
      "loss": 2.8957,
      "step": 8072
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6211914420127869,
      "learning_rate": 0.0005981842917000553,
      "loss": 2.6578,
      "step": 8073
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.8353184461593628,
      "learning_rate": 0.0005981838423041892,
      "loss": 2.4057,
      "step": 8074
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6739886403083801,
      "learning_rate": 0.0005981833928528852,
      "loss": 2.6723,
      "step": 8075
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.7124060392379761,
      "learning_rate": 0.0005981829433461432,
      "loss": 2.8591,
      "step": 8076
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6438536643981934,
      "learning_rate": 0.0005981824937839635,
      "loss": 2.3826,
      "step": 8077
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.8754828572273254,
      "learning_rate": 0.0005981820441663459,
      "loss": 2.3893,
      "step": 8078
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.5930502414703369,
      "learning_rate": 0.0005981815944932906,
      "loss": 2.5943,
      "step": 8079
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.7252885699272156,
      "learning_rate": 0.0005981811447647978,
      "loss": 2.4992,
      "step": 8080
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6761842370033264,
      "learning_rate": 0.0005981806949808676,
      "loss": 2.4853,
      "step": 8081
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6545078754425049,
      "learning_rate": 0.0005981802451414998,
      "loss": 2.3605,
      "step": 8082
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6472721099853516,
      "learning_rate": 0.0005981797952466948,
      "loss": 2.7024,
      "step": 8083
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6573949456214905,
      "learning_rate": 0.0005981793452964524,
      "loss": 2.1904,
      "step": 8084
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.8232516050338745,
      "learning_rate": 0.0005981788952907729,
      "loss": 2.5984,
      "step": 8085
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6967775821685791,
      "learning_rate": 0.0005981784452296563,
      "loss": 2.7224,
      "step": 8086
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6659140586853027,
      "learning_rate": 0.0005981779951131028,
      "loss": 2.615,
      "step": 8087
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.64277583360672,
      "learning_rate": 0.0005981775449411123,
      "loss": 2.4837,
      "step": 8088
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.91341233253479,
      "learning_rate": 0.000598177094713685,
      "loss": 2.4362,
      "step": 8089
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6297860741615295,
      "learning_rate": 0.000598176644430821,
      "loss": 2.2791,
      "step": 8090
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6626110672950745,
      "learning_rate": 0.0005981761940925202,
      "loss": 2.6195,
      "step": 8091
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6006499528884888,
      "learning_rate": 0.0005981757436987828,
      "loss": 2.3853,
      "step": 8092
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.7555935978889465,
      "learning_rate": 0.000598175293249609,
      "loss": 2.4864,
      "step": 8093
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6905249953269958,
      "learning_rate": 0.0005981748427449987,
      "loss": 2.6141,
      "step": 8094
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6199466586112976,
      "learning_rate": 0.0005981743921849522,
      "loss": 2.3711,
      "step": 8095
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6725507378578186,
      "learning_rate": 0.0005981739415694693,
      "loss": 2.4993,
      "step": 8096
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.0214884281158447,
      "learning_rate": 0.0005981734908985503,
      "loss": 2.3939,
      "step": 8097
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.7397193312644958,
      "learning_rate": 0.0005981730401721952,
      "loss": 2.4913,
      "step": 8098
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.8228561282157898,
      "learning_rate": 0.0005981725893904042,
      "loss": 2.6582,
      "step": 8099
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.8023183941841125,
      "learning_rate": 0.0005981721385531772,
      "loss": 2.5442,
      "step": 8100
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.7658889293670654,
      "learning_rate": 0.0005981716876605145,
      "loss": 2.5524,
      "step": 8101
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.7107464671134949,
      "learning_rate": 0.0005981712367124159,
      "loss": 2.6042,
      "step": 8102
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.8230931758880615,
      "learning_rate": 0.0005981707857088816,
      "loss": 2.567,
      "step": 8103
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.7680523991584778,
      "learning_rate": 0.0005981703346499118,
      "loss": 2.5528,
      "step": 8104
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.7294806241989136,
      "learning_rate": 0.0005981698835355065,
      "loss": 2.6884,
      "step": 8105
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6298004388809204,
      "learning_rate": 0.0005981694323656659,
      "loss": 2.6222,
      "step": 8106
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.7308469414710999,
      "learning_rate": 0.0005981689811403899,
      "loss": 2.5725,
      "step": 8107
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.7279049158096313,
      "learning_rate": 0.0005981685298596786,
      "loss": 2.4615,
      "step": 8108
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6986856460571289,
      "learning_rate": 0.0005981680785235322,
      "loss": 2.5273,
      "step": 8109
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6521191596984863,
      "learning_rate": 0.0005981676271319507,
      "loss": 2.6677,
      "step": 8110
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.7223019003868103,
      "learning_rate": 0.0005981671756849343,
      "loss": 2.5308,
      "step": 8111
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.7174758315086365,
      "learning_rate": 0.0005981667241824829,
      "loss": 2.3455,
      "step": 8112
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6647269129753113,
      "learning_rate": 0.0005981662726245966,
      "loss": 2.428,
      "step": 8113
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6508035659790039,
      "learning_rate": 0.0005981658210112757,
      "loss": 2.4209,
      "step": 8114
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.7023767232894897,
      "learning_rate": 0.0005981653693425201,
      "loss": 2.7527,
      "step": 8115
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6428881883621216,
      "learning_rate": 0.0005981649176183299,
      "loss": 2.656,
      "step": 8116
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6370943784713745,
      "learning_rate": 0.0005981644658387053,
      "loss": 2.5861,
      "step": 8117
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6225711703300476,
      "learning_rate": 0.0005981640140036461,
      "loss": 2.5006,
      "step": 8118
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.7303615808486938,
      "learning_rate": 0.0005981635621131528,
      "loss": 2.5743,
      "step": 8119
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6757421493530273,
      "learning_rate": 0.0005981631101672251,
      "loss": 2.5474,
      "step": 8120
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6344428658485413,
      "learning_rate": 0.0005981626581658635,
      "loss": 2.7295,
      "step": 8121
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.7217279076576233,
      "learning_rate": 0.0005981622061090676,
      "loss": 2.7845,
      "step": 8122
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6486971378326416,
      "learning_rate": 0.0005981617539968378,
      "loss": 2.3031,
      "step": 8123
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.643983781337738,
      "learning_rate": 0.0005981613018291741,
      "loss": 2.7569,
      "step": 8124
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.724506676197052,
      "learning_rate": 0.0005981608496060765,
      "loss": 2.5738,
      "step": 8125
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.7849035263061523,
      "learning_rate": 0.0005981603973275452,
      "loss": 2.5413,
      "step": 8126
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.5690987706184387,
      "learning_rate": 0.0005981599449935804,
      "loss": 2.2547,
      "step": 8127
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6393381953239441,
      "learning_rate": 0.0005981594926041819,
      "loss": 2.4103,
      "step": 8128
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.7030572295188904,
      "learning_rate": 0.0005981590401593499,
      "loss": 2.359,
      "step": 8129
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6555798649787903,
      "learning_rate": 0.0005981585876590846,
      "loss": 2.7296,
      "step": 8130
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6534884572029114,
      "learning_rate": 0.0005981581351033859,
      "loss": 2.6951,
      "step": 8131
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.7394319176673889,
      "learning_rate": 0.000598157682492254,
      "loss": 2.5692,
      "step": 8132
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6298458576202393,
      "learning_rate": 0.0005981572298256891,
      "loss": 2.4461,
      "step": 8133
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6608792543411255,
      "learning_rate": 0.0005981567771036909,
      "loss": 2.7738,
      "step": 8134
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.7687221765518188,
      "learning_rate": 0.0005981563243262599,
      "loss": 2.3406,
      "step": 8135
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.7319412231445312,
      "learning_rate": 0.000598155871493396,
      "loss": 2.7253,
      "step": 8136
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6815553903579712,
      "learning_rate": 0.0005981554186050992,
      "loss": 2.5185,
      "step": 8137
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.7253161072731018,
      "learning_rate": 0.0005981549656613698,
      "loss": 2.5915,
      "step": 8138
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6483345627784729,
      "learning_rate": 0.0005981545126622078,
      "loss": 2.5864,
      "step": 8139
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.646733820438385,
      "learning_rate": 0.0005981540596076131,
      "loss": 2.5533,
      "step": 8140
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6361313462257385,
      "learning_rate": 0.000598153606497586,
      "loss": 2.6404,
      "step": 8141
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6399514079093933,
      "learning_rate": 0.0005981531533321264,
      "loss": 2.3975,
      "step": 8142
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6441245675086975,
      "learning_rate": 0.0005981527001112347,
      "loss": 2.5588,
      "step": 8143
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6992610096931458,
      "learning_rate": 0.0005981522468349106,
      "loss": 2.4016,
      "step": 8144
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.7083653807640076,
      "learning_rate": 0.0005981517935031545,
      "loss": 2.4362,
      "step": 8145
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.7567200064659119,
      "learning_rate": 0.0005981513401159663,
      "loss": 2.7046,
      "step": 8146
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.7776986956596375,
      "learning_rate": 0.0005981508866733462,
      "loss": 2.7309,
      "step": 8147
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6265500783920288,
      "learning_rate": 0.0005981504331752941,
      "loss": 2.3685,
      "step": 8148
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6689148545265198,
      "learning_rate": 0.0005981499796218103,
      "loss": 2.5125,
      "step": 8149
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.7896822690963745,
      "learning_rate": 0.0005981495260128948,
      "loss": 2.4776,
      "step": 8150
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.7098341584205627,
      "learning_rate": 0.0005981490723485476,
      "loss": 2.4435,
      "step": 8151
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6329162120819092,
      "learning_rate": 0.0005981486186287689,
      "loss": 2.3226,
      "step": 8152
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6732538938522339,
      "learning_rate": 0.0005981481648535587,
      "loss": 2.3991,
      "step": 8153
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6432219743728638,
      "learning_rate": 0.0005981477110229172,
      "loss": 2.6597,
      "step": 8154
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6691877245903015,
      "learning_rate": 0.0005981472571368444,
      "loss": 2.4448,
      "step": 8155
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6540436148643494,
      "learning_rate": 0.0005981468031953405,
      "loss": 2.3075,
      "step": 8156
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6699951887130737,
      "learning_rate": 0.0005981463491984052,
      "loss": 2.412,
      "step": 8157
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6717255711555481,
      "learning_rate": 0.0005981458951460391,
      "loss": 2.5278,
      "step": 8158
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6213685870170593,
      "learning_rate": 0.000598145441038242,
      "loss": 2.4576,
      "step": 8159
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6544904708862305,
      "learning_rate": 0.000598144986875014,
      "loss": 2.5421,
      "step": 8160
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.7237025499343872,
      "learning_rate": 0.0005981445326563552,
      "loss": 2.5559,
      "step": 8161
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.5924318432807922,
      "learning_rate": 0.0005981440783822657,
      "loss": 2.5813,
      "step": 8162
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.605055570602417,
      "learning_rate": 0.0005981436240527457,
      "loss": 2.3713,
      "step": 8163
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.7118439674377441,
      "learning_rate": 0.0005981431696677951,
      "loss": 2.4105,
      "step": 8164
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6344322562217712,
      "learning_rate": 0.000598142715227414,
      "loss": 2.489,
      "step": 8165
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.9426043033599854,
      "learning_rate": 0.0005981422607316026,
      "loss": 2.7289,
      "step": 8166
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.744878351688385,
      "learning_rate": 0.0005981418061803609,
      "loss": 2.7135,
      "step": 8167
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.8137208223342896,
      "learning_rate": 0.0005981413515736892,
      "loss": 2.4869,
      "step": 8168
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.598164439201355,
      "learning_rate": 0.0005981408969115871,
      "loss": 2.5349,
      "step": 8169
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6852863430976868,
      "learning_rate": 0.0005981404421940552,
      "loss": 2.7057,
      "step": 8170
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6899895071983337,
      "learning_rate": 0.0005981399874210932,
      "loss": 2.6649,
      "step": 8171
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6123199462890625,
      "learning_rate": 0.0005981395325927014,
      "loss": 2.4231,
      "step": 8172
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6636830568313599,
      "learning_rate": 0.0005981390777088798,
      "loss": 2.5609,
      "step": 8173
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.7983839511871338,
      "learning_rate": 0.0005981386227696286,
      "loss": 2.5411,
      "step": 8174
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6505733728408813,
      "learning_rate": 0.0005981381677749478,
      "loss": 2.4384,
      "step": 8175
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6798703074455261,
      "learning_rate": 0.0005981377127248374,
      "loss": 2.5107,
      "step": 8176
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.7724603414535522,
      "learning_rate": 0.0005981372576192977,
      "loss": 2.4449,
      "step": 8177
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6137794852256775,
      "learning_rate": 0.0005981368024583286,
      "loss": 2.615,
      "step": 8178
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6025365591049194,
      "learning_rate": 0.0005981363472419302,
      "loss": 2.4491,
      "step": 8179
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6993242502212524,
      "learning_rate": 0.0005981358919701027,
      "loss": 2.2706,
      "step": 8180
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6598519086837769,
      "learning_rate": 0.000598135436642846,
      "loss": 2.3851,
      "step": 8181
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6554807424545288,
      "learning_rate": 0.0005981349812601604,
      "loss": 2.3451,
      "step": 8182
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6031746864318848,
      "learning_rate": 0.0005981345258220458,
      "loss": 2.4438,
      "step": 8183
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.7242757678031921,
      "learning_rate": 0.0005981340703285024,
      "loss": 2.3045,
      "step": 8184
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6592372059822083,
      "learning_rate": 0.0005981336147795303,
      "loss": 2.6823,
      "step": 8185
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.7084770202636719,
      "learning_rate": 0.0005981331591751295,
      "loss": 2.3332,
      "step": 8186
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6864178776741028,
      "learning_rate": 0.0005981327035153,
      "loss": 2.507,
      "step": 8187
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6229888796806335,
      "learning_rate": 0.0005981322478000421,
      "loss": 2.5799,
      "step": 8188
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6331267356872559,
      "learning_rate": 0.0005981317920293558,
      "loss": 2.3703,
      "step": 8189
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6087682247161865,
      "learning_rate": 0.0005981313362032413,
      "loss": 2.4755,
      "step": 8190
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6948145627975464,
      "learning_rate": 0.0005981308803216983,
      "loss": 2.6072,
      "step": 8191
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.7376688718795776,
      "learning_rate": 0.0005981304243847272,
      "loss": 2.337,
      "step": 8192
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.5797277092933655,
      "learning_rate": 0.0005981299683923281,
      "loss": 2.49,
      "step": 8193
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6262540221214294,
      "learning_rate": 0.000598129512344501,
      "loss": 2.6262,
      "step": 8194
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6806369423866272,
      "learning_rate": 0.000598129056241246,
      "loss": 2.4963,
      "step": 8195
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6584946513175964,
      "learning_rate": 0.0005981286000825631,
      "loss": 2.6211,
      "step": 8196
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.7706291675567627,
      "learning_rate": 0.0005981281438684527,
      "loss": 2.4173,
      "step": 8197
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6034191846847534,
      "learning_rate": 0.0005981276875989144,
      "loss": 2.5162,
      "step": 8198
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6085952520370483,
      "learning_rate": 0.0005981272312739486,
      "loss": 2.429,
      "step": 8199
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6930489540100098,
      "learning_rate": 0.0005981267748935554,
      "loss": 2.5741,
      "step": 8200
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.670579195022583,
      "learning_rate": 0.0005981263184577347,
      "loss": 2.4282,
      "step": 8201
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6457816958427429,
      "learning_rate": 0.0005981258619664868,
      "loss": 2.6668,
      "step": 8202
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6187157034873962,
      "learning_rate": 0.0005981254054198117,
      "loss": 2.4857,
      "step": 8203
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6218414902687073,
      "learning_rate": 0.0005981249488177093,
      "loss": 2.634,
      "step": 8204
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6324408054351807,
      "learning_rate": 0.0005981244921601799,
      "loss": 2.4998,
      "step": 8205
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6506546139717102,
      "learning_rate": 0.0005981240354472236,
      "loss": 2.673,
      "step": 8206
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6309453845024109,
      "learning_rate": 0.0005981235786788404,
      "loss": 2.445,
      "step": 8207
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6726518869400024,
      "learning_rate": 0.0005981231218550303,
      "loss": 2.5506,
      "step": 8208
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.7184286713600159,
      "learning_rate": 0.0005981226649757935,
      "loss": 2.6286,
      "step": 8209
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6801681518554688,
      "learning_rate": 0.0005981222080411302,
      "loss": 2.4872,
      "step": 8210
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.631759762763977,
      "learning_rate": 0.0005981217510510402,
      "loss": 2.5293,
      "step": 8211
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6538988947868347,
      "learning_rate": 0.0005981212940055238,
      "loss": 2.4065,
      "step": 8212
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6713083386421204,
      "learning_rate": 0.0005981208369045812,
      "loss": 2.6642,
      "step": 8213
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.5835092067718506,
      "learning_rate": 0.000598120379748212,
      "loss": 2.3937,
      "step": 8214
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6267717480659485,
      "learning_rate": 0.0005981199225364168,
      "loss": 2.4459,
      "step": 8215
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6875749826431274,
      "learning_rate": 0.0005981194652691953,
      "loss": 2.5807,
      "step": 8216
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.5943585634231567,
      "learning_rate": 0.000598119007946548,
      "loss": 2.4181,
      "step": 8217
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.9336447715759277,
      "learning_rate": 0.0005981185505684745,
      "loss": 2.4763,
      "step": 8218
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6896559000015259,
      "learning_rate": 0.0005981180931349753,
      "loss": 2.4793,
      "step": 8219
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6552407145500183,
      "learning_rate": 0.0005981176356460503,
      "loss": 2.2794,
      "step": 8220
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.8792943954467773,
      "learning_rate": 0.0005981171781016996,
      "loss": 2.5469,
      "step": 8221
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.693054735660553,
      "learning_rate": 0.0005981167205019233,
      "loss": 2.4805,
      "step": 8222
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.7030404210090637,
      "learning_rate": 0.0005981162628467215,
      "loss": 2.6111,
      "step": 8223
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.7526409029960632,
      "learning_rate": 0.0005981158051360942,
      "loss": 2.8478,
      "step": 8224
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6446810364723206,
      "learning_rate": 0.0005981153473700416,
      "loss": 2.3868,
      "step": 8225
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.633648693561554,
      "learning_rate": 0.0005981148895485636,
      "loss": 2.5256,
      "step": 8226
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6576952934265137,
      "learning_rate": 0.0005981144316716605,
      "loss": 2.412,
      "step": 8227
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.879296064376831,
      "learning_rate": 0.0005981139737393323,
      "loss": 2.6098,
      "step": 8228
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6074686646461487,
      "learning_rate": 0.000598113515751579,
      "loss": 2.5768,
      "step": 8229
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6688534021377563,
      "learning_rate": 0.0005981130577084009,
      "loss": 2.6595,
      "step": 8230
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6590724587440491,
      "learning_rate": 0.0005981125996097979,
      "loss": 2.6584,
      "step": 8231
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6690737009048462,
      "learning_rate": 0.0005981121414557701,
      "loss": 2.6005,
      "step": 8232
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6915919780731201,
      "learning_rate": 0.0005981116832463177,
      "loss": 2.6845,
      "step": 8233
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.653975248336792,
      "learning_rate": 0.0005981112249814407,
      "loss": 2.4592,
      "step": 8234
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6457062363624573,
      "learning_rate": 0.0005981107666611392,
      "loss": 2.6639,
      "step": 8235
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.7491681575775146,
      "learning_rate": 0.0005981103082854133,
      "loss": 2.4226,
      "step": 8236
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6949813961982727,
      "learning_rate": 0.000598109849854263,
      "loss": 2.5233,
      "step": 8237
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6760812997817993,
      "learning_rate": 0.0005981093913676885,
      "loss": 2.4453,
      "step": 8238
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6165447235107422,
      "learning_rate": 0.0005981089328256898,
      "loss": 2.6755,
      "step": 8239
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6542206406593323,
      "learning_rate": 0.000598108474228267,
      "loss": 2.9295,
      "step": 8240
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.5993267893791199,
      "learning_rate": 0.0005981080155754203,
      "loss": 2.7359,
      "step": 8241
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6010174751281738,
      "learning_rate": 0.0005981075568671496,
      "loss": 2.3169,
      "step": 8242
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6273191571235657,
      "learning_rate": 0.0005981070981034551,
      "loss": 2.5298,
      "step": 8243
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6940675973892212,
      "learning_rate": 0.0005981066392843369,
      "loss": 2.4329,
      "step": 8244
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.7720668911933899,
      "learning_rate": 0.000598106180409795,
      "loss": 2.4044,
      "step": 8245
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.666070282459259,
      "learning_rate": 0.0005981057214798296,
      "loss": 2.511,
      "step": 8246
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6217451691627502,
      "learning_rate": 0.0005981052624944406,
      "loss": 2.5307,
      "step": 8247
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6280093193054199,
      "learning_rate": 0.0005981048034536283,
      "loss": 2.466,
      "step": 8248
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6510432362556458,
      "learning_rate": 0.0005981043443573928,
      "loss": 2.5091,
      "step": 8249
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6667426824569702,
      "learning_rate": 0.0005981038852057338,
      "loss": 2.4155,
      "step": 8250
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6409110426902771,
      "learning_rate": 0.0005981034259986518,
      "loss": 2.5476,
      "step": 8251
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.661636471748352,
      "learning_rate": 0.0005981029667361469,
      "loss": 2.5634,
      "step": 8252
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6786569356918335,
      "learning_rate": 0.0005981025074182189,
      "loss": 2.7265,
      "step": 8253
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6739884614944458,
      "learning_rate": 0.0005981020480448679,
      "loss": 2.5266,
      "step": 8254
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6486541032791138,
      "learning_rate": 0.0005981015886160942,
      "loss": 2.4542,
      "step": 8255
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.7141326665878296,
      "learning_rate": 0.0005981011291318979,
      "loss": 2.6799,
      "step": 8256
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.5887141227722168,
      "learning_rate": 0.0005981006695922788,
      "loss": 2.3918,
      "step": 8257
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.7609298825263977,
      "learning_rate": 0.0005981002099972372,
      "loss": 2.3373,
      "step": 8258
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.5888243913650513,
      "learning_rate": 0.0005980997503467732,
      "loss": 2.4486,
      "step": 8259
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6615488529205322,
      "learning_rate": 0.0005980992906408867,
      "loss": 2.6596,
      "step": 8260
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6665425300598145,
      "learning_rate": 0.0005980988308795781,
      "loss": 2.6187,
      "step": 8261
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.68948894739151,
      "learning_rate": 0.0005980983710628473,
      "loss": 2.3651,
      "step": 8262
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6223335862159729,
      "learning_rate": 0.0005980979111906942,
      "loss": 2.6666,
      "step": 8263
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.7710676193237305,
      "learning_rate": 0.0005980974512631191,
      "loss": 2.5832,
      "step": 8264
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6264872550964355,
      "learning_rate": 0.0005980969912801222,
      "loss": 2.3964,
      "step": 8265
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.7001014351844788,
      "learning_rate": 0.0005980965312417034,
      "loss": 2.5496,
      "step": 8266
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6095307469367981,
      "learning_rate": 0.0005980960711478628,
      "loss": 2.4461,
      "step": 8267
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.8107811212539673,
      "learning_rate": 0.0005980956109986005,
      "loss": 2.3527,
      "step": 8268
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6509283185005188,
      "learning_rate": 0.0005980951507939166,
      "loss": 2.877,
      "step": 8269
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6786615252494812,
      "learning_rate": 0.0005980946905338112,
      "loss": 2.617,
      "step": 8270
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.673875093460083,
      "learning_rate": 0.0005980942302182844,
      "loss": 2.3425,
      "step": 8271
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.7000983357429504,
      "learning_rate": 0.0005980937698473362,
      "loss": 2.7585,
      "step": 8272
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6649901866912842,
      "learning_rate": 0.0005980933094209668,
      "loss": 2.4243,
      "step": 8273
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6408580541610718,
      "learning_rate": 0.0005980928489391762,
      "loss": 2.3675,
      "step": 8274
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6921182870864868,
      "learning_rate": 0.0005980923884019644,
      "loss": 2.583,
      "step": 8275
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6207634210586548,
      "learning_rate": 0.0005980919278093318,
      "loss": 2.2812,
      "step": 8276
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6435607671737671,
      "learning_rate": 0.0005980914671612781,
      "loss": 2.4678,
      "step": 8277
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6721140742301941,
      "learning_rate": 0.0005980910064578037,
      "loss": 2.3851,
      "step": 8278
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6320512294769287,
      "learning_rate": 0.0005980905456989085,
      "loss": 2.7443,
      "step": 8279
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6217962503433228,
      "learning_rate": 0.0005980900848845926,
      "loss": 2.6973,
      "step": 8280
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.7175948619842529,
      "learning_rate": 0.0005980896240148562,
      "loss": 2.4675,
      "step": 8281
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.7522948384284973,
      "learning_rate": 0.0005980891630896993,
      "loss": 2.6736,
      "step": 8282
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.5714806914329529,
      "learning_rate": 0.000598088702109122,
      "loss": 2.4989,
      "step": 8283
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.8005208969116211,
      "learning_rate": 0.0005980882410731243,
      "loss": 2.5611,
      "step": 8284
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.625619113445282,
      "learning_rate": 0.0005980877799817065,
      "loss": 2.3716,
      "step": 8285
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.7073567509651184,
      "learning_rate": 0.0005980873188348684,
      "loss": 2.5596,
      "step": 8286
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6123393177986145,
      "learning_rate": 0.0005980868576326104,
      "loss": 2.9843,
      "step": 8287
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.5981144905090332,
      "learning_rate": 0.0005980863963749324,
      "loss": 2.4123,
      "step": 8288
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6337295770645142,
      "learning_rate": 0.0005980859350618344,
      "loss": 2.496,
      "step": 8289
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6249759793281555,
      "learning_rate": 0.0005980854736933167,
      "loss": 2.642,
      "step": 8290
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6346623301506042,
      "learning_rate": 0.0005980850122693794,
      "loss": 2.8332,
      "step": 8291
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.5881003141403198,
      "learning_rate": 0.0005980845507900221,
      "loss": 2.4186,
      "step": 8292
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6559728384017944,
      "learning_rate": 0.0005980840892552455,
      "loss": 2.6008,
      "step": 8293
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6235474348068237,
      "learning_rate": 0.0005980836276650494,
      "loss": 2.4611,
      "step": 8294
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6539468169212341,
      "learning_rate": 0.000598083166019434,
      "loss": 2.887,
      "step": 8295
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6090924143791199,
      "learning_rate": 0.0005980827043183993,
      "loss": 2.4727,
      "step": 8296
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.5816822052001953,
      "learning_rate": 0.0005980822425619454,
      "loss": 2.6516,
      "step": 8297
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6166996359825134,
      "learning_rate": 0.0005980817807500723,
      "loss": 2.701,
      "step": 8298
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.7538074851036072,
      "learning_rate": 0.0005980813188827802,
      "loss": 2.5261,
      "step": 8299
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6000734567642212,
      "learning_rate": 0.0005980808569600692,
      "loss": 2.5804,
      "step": 8300
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6042718887329102,
      "learning_rate": 0.0005980803949819393,
      "loss": 2.3887,
      "step": 8301
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6354982852935791,
      "learning_rate": 0.0005980799329483906,
      "loss": 2.7602,
      "step": 8302
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6880825161933899,
      "learning_rate": 0.0005980794708594232,
      "loss": 2.5849,
      "step": 8303
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.667349100112915,
      "learning_rate": 0.0005980790087150373,
      "loss": 2.5173,
      "step": 8304
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6972916722297668,
      "learning_rate": 0.0005980785465152329,
      "loss": 2.3437,
      "step": 8305
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6496044397354126,
      "learning_rate": 0.0005980780842600099,
      "loss": 2.5718,
      "step": 8306
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6200510263442993,
      "learning_rate": 0.0005980776219493687,
      "loss": 2.5092,
      "step": 8307
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.799887478351593,
      "learning_rate": 0.0005980771595833093,
      "loss": 2.4062,
      "step": 8308
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6446388363838196,
      "learning_rate": 0.0005980766971618316,
      "loss": 2.5637,
      "step": 8309
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.7144566178321838,
      "learning_rate": 0.0005980762346849358,
      "loss": 2.4555,
      "step": 8310
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.7016186118125916,
      "learning_rate": 0.0005980757721526221,
      "loss": 2.5005,
      "step": 8311
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.716402530670166,
      "learning_rate": 0.0005980753095648904,
      "loss": 2.4412,
      "step": 8312
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.7512695789337158,
      "learning_rate": 0.0005980748469217409,
      "loss": 2.5223,
      "step": 8313
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6644331216812134,
      "learning_rate": 0.0005980743842231737,
      "loss": 2.5418,
      "step": 8314
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6420823931694031,
      "learning_rate": 0.0005980739214691888,
      "loss": 2.3984,
      "step": 8315
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6629011631011963,
      "learning_rate": 0.0005980734586597864,
      "loss": 2.7398,
      "step": 8316
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6639379262924194,
      "learning_rate": 0.0005980729957949665,
      "loss": 2.5917,
      "step": 8317
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6774296164512634,
      "learning_rate": 0.0005980725328747291,
      "loss": 2.318,
      "step": 8318
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.7518160343170166,
      "learning_rate": 0.0005980720698990745,
      "loss": 2.3704,
      "step": 8319
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6502279043197632,
      "learning_rate": 0.0005980716068680026,
      "loss": 2.5167,
      "step": 8320
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6965135931968689,
      "learning_rate": 0.0005980711437815136,
      "loss": 2.4617,
      "step": 8321
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.7118716239929199,
      "learning_rate": 0.0005980706806396075,
      "loss": 2.5378,
      "step": 8322
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6957423090934753,
      "learning_rate": 0.0005980702174422846,
      "loss": 2.4258,
      "step": 8323
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.670049250125885,
      "learning_rate": 0.0005980697541895446,
      "loss": 2.5648,
      "step": 8324
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6453556418418884,
      "learning_rate": 0.0005980692908813879,
      "loss": 2.6003,
      "step": 8325
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.7067574262619019,
      "learning_rate": 0.0005980688275178146,
      "loss": 2.4257,
      "step": 8326
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6471405625343323,
      "learning_rate": 0.0005980683640988245,
      "loss": 2.4915,
      "step": 8327
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6999743580818176,
      "learning_rate": 0.0005980679006244179,
      "loss": 2.7928,
      "step": 8328
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6504315137863159,
      "learning_rate": 0.000598067437094595,
      "loss": 2.4183,
      "step": 8329
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.660767674446106,
      "learning_rate": 0.0005980669735093556,
      "loss": 2.5035,
      "step": 8330
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.703510582447052,
      "learning_rate": 0.0005980665098687,
      "loss": 2.4879,
      "step": 8331
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.684428334236145,
      "learning_rate": 0.0005980660461726281,
      "loss": 2.5931,
      "step": 8332
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6651752591133118,
      "learning_rate": 0.0005980655824211401,
      "loss": 2.3691,
      "step": 8333
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.5840879082679749,
      "learning_rate": 0.0005980651186142362,
      "loss": 2.5144,
      "step": 8334
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6554784774780273,
      "learning_rate": 0.0005980646547519164,
      "loss": 2.6655,
      "step": 8335
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.5995158553123474,
      "learning_rate": 0.0005980641908341805,
      "loss": 2.4063,
      "step": 8336
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.5957391262054443,
      "learning_rate": 0.0005980637268610291,
      "loss": 2.4283,
      "step": 8337
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6942437291145325,
      "learning_rate": 0.0005980632628324619,
      "loss": 2.6425,
      "step": 8338
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6438635587692261,
      "learning_rate": 0.0005980627987484791,
      "loss": 2.5904,
      "step": 8339
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.8704525232315063,
      "learning_rate": 0.0005980623346090809,
      "loss": 2.3236,
      "step": 8340
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6441383957862854,
      "learning_rate": 0.0005980618704142672,
      "loss": 2.4273,
      "step": 8341
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6404320001602173,
      "learning_rate": 0.0005980614061640382,
      "loss": 2.4753,
      "step": 8342
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.7369394898414612,
      "learning_rate": 0.000598060941858394,
      "loss": 2.5262,
      "step": 8343
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6535565853118896,
      "learning_rate": 0.0005980604774973346,
      "loss": 2.6065,
      "step": 8344
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.7210266590118408,
      "learning_rate": 0.0005980600130808602,
      "loss": 2.4717,
      "step": 8345
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6959195733070374,
      "learning_rate": 0.0005980595486089707,
      "loss": 2.2833,
      "step": 8346
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6335094571113586,
      "learning_rate": 0.0005980590840816663,
      "loss": 2.3983,
      "step": 8347
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6245328783988953,
      "learning_rate": 0.0005980586194989472,
      "loss": 2.5329,
      "step": 8348
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.7843618988990784,
      "learning_rate": 0.0005980581548608132,
      "loss": 2.5339,
      "step": 8349
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.653285026550293,
      "learning_rate": 0.0005980576901672647,
      "loss": 2.3574,
      "step": 8350
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6924470067024231,
      "learning_rate": 0.0005980572254183017,
      "loss": 2.5041,
      "step": 8351
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6217312216758728,
      "learning_rate": 0.0005980567606139242,
      "loss": 2.4861,
      "step": 8352
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6571060419082642,
      "learning_rate": 0.0005980562957541323,
      "loss": 2.4832,
      "step": 8353
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6533992886543274,
      "learning_rate": 0.0005980558308389261,
      "loss": 2.5261,
      "step": 8354
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6656874418258667,
      "learning_rate": 0.0005980553658683056,
      "loss": 2.7736,
      "step": 8355
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6631140112876892,
      "learning_rate": 0.0005980549008422712,
      "loss": 2.4905,
      "step": 8356
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6383153796195984,
      "learning_rate": 0.0005980544357608226,
      "loss": 2.7959,
      "step": 8357
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6295650601387024,
      "learning_rate": 0.0005980539706239601,
      "loss": 2.4843,
      "step": 8358
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6392555236816406,
      "learning_rate": 0.0005980535054316837,
      "loss": 2.3078,
      "step": 8359
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.623525083065033,
      "learning_rate": 0.0005980530401839936,
      "loss": 2.3694,
      "step": 8360
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6276407837867737,
      "learning_rate": 0.0005980525748808897,
      "loss": 2.383,
      "step": 8361
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.7261990904808044,
      "learning_rate": 0.0005980521095223724,
      "loss": 2.7497,
      "step": 8362
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.5822699666023254,
      "learning_rate": 0.0005980516441084413,
      "loss": 2.5355,
      "step": 8363
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.611275851726532,
      "learning_rate": 0.0005980511786390971,
      "loss": 2.3393,
      "step": 8364
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6889438033103943,
      "learning_rate": 0.0005980507131143393,
      "loss": 2.5782,
      "step": 8365
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6303746104240417,
      "learning_rate": 0.0005980502475341683,
      "loss": 2.5209,
      "step": 8366
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.627552330493927,
      "learning_rate": 0.0005980497818985842,
      "loss": 2.551,
      "step": 8367
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.7149823307991028,
      "learning_rate": 0.000598049316207587,
      "loss": 2.5417,
      "step": 8368
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6826652884483337,
      "learning_rate": 0.0005980488504611769,
      "loss": 2.8363,
      "step": 8369
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6370360255241394,
      "learning_rate": 0.0005980483846593537,
      "loss": 2.5159,
      "step": 8370
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6883981823921204,
      "learning_rate": 0.0005980479188021178,
      "loss": 2.526,
      "step": 8371
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6890242099761963,
      "learning_rate": 0.0005980474528894692,
      "loss": 2.4753,
      "step": 8372
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6909162998199463,
      "learning_rate": 0.0005980469869214079,
      "loss": 2.445,
      "step": 8373
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.5937796831130981,
      "learning_rate": 0.0005980465208979339,
      "loss": 2.2812,
      "step": 8374
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.692520260810852,
      "learning_rate": 0.0005980460548190476,
      "loss": 2.4568,
      "step": 8375
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6596238613128662,
      "learning_rate": 0.0005980455886847489,
      "loss": 2.7603,
      "step": 8376
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.662251889705658,
      "learning_rate": 0.0005980451224950379,
      "loss": 2.4667,
      "step": 8377
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6004259586334229,
      "learning_rate": 0.0005980446562499146,
      "loss": 2.433,
      "step": 8378
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6662235260009766,
      "learning_rate": 0.0005980441899493792,
      "loss": 2.6169,
      "step": 8379
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6901121735572815,
      "learning_rate": 0.0005980437235934318,
      "loss": 2.4235,
      "step": 8380
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6740500330924988,
      "learning_rate": 0.0005980432571820724,
      "loss": 2.6343,
      "step": 8381
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6940711140632629,
      "learning_rate": 0.0005980427907153012,
      "loss": 2.4521,
      "step": 8382
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.697136640548706,
      "learning_rate": 0.0005980423241931182,
      "loss": 2.6623,
      "step": 8383
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.5933695435523987,
      "learning_rate": 0.0005980418576155235,
      "loss": 2.5408,
      "step": 8384
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6279347538948059,
      "learning_rate": 0.0005980413909825172,
      "loss": 2.3253,
      "step": 8385
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.583621621131897,
      "learning_rate": 0.0005980409242940993,
      "loss": 2.0037,
      "step": 8386
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.663246214389801,
      "learning_rate": 0.0005980404575502701,
      "loss": 2.5673,
      "step": 8387
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6662212014198303,
      "learning_rate": 0.0005980399907510295,
      "loss": 2.4749,
      "step": 8388
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6731066107749939,
      "learning_rate": 0.0005980395238963775,
      "loss": 2.6467,
      "step": 8389
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.7513220310211182,
      "learning_rate": 0.0005980390569863145,
      "loss": 2.3823,
      "step": 8390
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.5989072918891907,
      "learning_rate": 0.0005980385900208403,
      "loss": 2.4321,
      "step": 8391
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.70026695728302,
      "learning_rate": 0.0005980381229999552,
      "loss": 2.3999,
      "step": 8392
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6610466241836548,
      "learning_rate": 0.0005980376559236592,
      "loss": 2.7245,
      "step": 8393
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.5668706893920898,
      "learning_rate": 0.0005980371887919524,
      "loss": 2.4784,
      "step": 8394
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6767945289611816,
      "learning_rate": 0.0005980367216048347,
      "loss": 2.4161,
      "step": 8395
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.609028697013855,
      "learning_rate": 0.0005980362543623064,
      "loss": 2.4333,
      "step": 8396
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6377646923065186,
      "learning_rate": 0.0005980357870643676,
      "loss": 2.5277,
      "step": 8397
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6222871541976929,
      "learning_rate": 0.0005980353197110183,
      "loss": 2.6588,
      "step": 8398
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6321733593940735,
      "learning_rate": 0.0005980348523022586,
      "loss": 2.412,
      "step": 8399
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6685091853141785,
      "learning_rate": 0.0005980343848380886,
      "loss": 2.511,
      "step": 8400
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.5882862210273743,
      "learning_rate": 0.0005980339173185084,
      "loss": 2.5229,
      "step": 8401
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.780131459236145,
      "learning_rate": 0.0005980334497435179,
      "loss": 2.6313,
      "step": 8402
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6148373484611511,
      "learning_rate": 0.0005980329821131176,
      "loss": 2.5628,
      "step": 8403
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.5970473289489746,
      "learning_rate": 0.0005980325144273071,
      "loss": 2.4428,
      "step": 8404
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.5792431831359863,
      "learning_rate": 0.0005980320466860869,
      "loss": 2.2718,
      "step": 8405
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6785708665847778,
      "learning_rate": 0.0005980315788894569,
      "loss": 2.7318,
      "step": 8406
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.714659571647644,
      "learning_rate": 0.0005980311110374173,
      "loss": 2.3289,
      "step": 8407
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.626790463924408,
      "learning_rate": 0.000598030643129968,
      "loss": 2.5141,
      "step": 8408
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6567083597183228,
      "learning_rate": 0.000598030175167109,
      "loss": 2.4816,
      "step": 8409
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6119019985198975,
      "learning_rate": 0.0005980297071488408,
      "loss": 2.4245,
      "step": 8410
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6603200435638428,
      "learning_rate": 0.0005980292390751631,
      "loss": 2.8153,
      "step": 8411
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6119149327278137,
      "learning_rate": 0.0005980287709460763,
      "loss": 2.2036,
      "step": 8412
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6249390840530396,
      "learning_rate": 0.0005980283027615801,
      "loss": 2.5546,
      "step": 8413
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.7633628249168396,
      "learning_rate": 0.0005980278345216751,
      "loss": 2.4193,
      "step": 8414
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6788501143455505,
      "learning_rate": 0.0005980273662263609,
      "loss": 2.7543,
      "step": 8415
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.740169107913971,
      "learning_rate": 0.0005980268978756379,
      "loss": 2.6026,
      "step": 8416
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.693749189376831,
      "learning_rate": 0.000598026429469506,
      "loss": 2.4909,
      "step": 8417
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.7334699630737305,
      "learning_rate": 0.0005980259610079653,
      "loss": 2.6599,
      "step": 8418
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6654489040374756,
      "learning_rate": 0.000598025492491016,
      "loss": 2.6173,
      "step": 8419
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6556733250617981,
      "learning_rate": 0.0005980250239186581,
      "loss": 2.3075,
      "step": 8420
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6264806389808655,
      "learning_rate": 0.0005980245552908919,
      "loss": 2.3224,
      "step": 8421
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6678116321563721,
      "learning_rate": 0.0005980240866077172,
      "loss": 2.3334,
      "step": 8422
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6246213912963867,
      "learning_rate": 0.0005980236178691342,
      "loss": 2.8038,
      "step": 8423
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.7738975882530212,
      "learning_rate": 0.0005980231490751429,
      "loss": 2.3886,
      "step": 8424
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.7361889481544495,
      "learning_rate": 0.0005980226802257435,
      "loss": 2.6492,
      "step": 8425
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.702456533908844,
      "learning_rate": 0.0005980222113209361,
      "loss": 2.3603,
      "step": 8426
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.7328130602836609,
      "learning_rate": 0.0005980217423607208,
      "loss": 2.2662,
      "step": 8427
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6013668775558472,
      "learning_rate": 0.0005980212733450975,
      "loss": 2.5386,
      "step": 8428
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.7631997466087341,
      "learning_rate": 0.0005980208042740666,
      "loss": 2.5894,
      "step": 8429
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.7156897783279419,
      "learning_rate": 0.0005980203351476279,
      "loss": 2.5073,
      "step": 8430
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6527184247970581,
      "learning_rate": 0.0005980198659657816,
      "loss": 2.5158,
      "step": 8431
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6504091620445251,
      "learning_rate": 0.0005980193967285278,
      "loss": 2.5968,
      "step": 8432
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6724507212638855,
      "learning_rate": 0.0005980189274358666,
      "loss": 2.4392,
      "step": 8433
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.0208441019058228,
      "learning_rate": 0.0005980184580877978,
      "loss": 2.4999,
      "step": 8434
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6586653590202332,
      "learning_rate": 0.0005980179886843221,
      "loss": 2.5433,
      "step": 8435
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.2184754610061646,
      "learning_rate": 0.0005980175192254391,
      "loss": 2.5365,
      "step": 8436
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.9410658478736877,
      "learning_rate": 0.000598017049711149,
      "loss": 2.3634,
      "step": 8437
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6967248320579529,
      "learning_rate": 0.0005980165801414518,
      "loss": 2.348,
      "step": 8438
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.2613258361816406,
      "learning_rate": 0.0005980161105163479,
      "loss": 2.593,
      "step": 8439
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.0516566038131714,
      "learning_rate": 0.0005980156408358369,
      "loss": 2.5412,
      "step": 8440
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.5925257802009583,
      "learning_rate": 0.0005980151710999194,
      "loss": 2.5372,
      "step": 8441
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.1779046058654785,
      "learning_rate": 0.0005980147013085952,
      "loss": 2.2861,
      "step": 8442
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.2539594173431396,
      "learning_rate": 0.0005980142314618644,
      "loss": 2.5087,
      "step": 8443
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.7879540920257568,
      "learning_rate": 0.0005980137615597272,
      "loss": 2.311,
      "step": 8444
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.9323221445083618,
      "learning_rate": 0.0005980132916021836,
      "loss": 2.4125,
      "step": 8445
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.8832999467849731,
      "learning_rate": 0.0005980128215892336,
      "loss": 2.8106,
      "step": 8446
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.7504166960716248,
      "learning_rate": 0.0005980123515208775,
      "loss": 2.604,
      "step": 8447
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.605262041091919,
      "learning_rate": 0.0005980118813971153,
      "loss": 2.7844,
      "step": 8448
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6470460295677185,
      "learning_rate": 0.000598011411217947,
      "loss": 2.2723,
      "step": 8449
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.796919047832489,
      "learning_rate": 0.0005980109409833728,
      "loss": 2.4815,
      "step": 8450
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.7180999517440796,
      "learning_rate": 0.0005980104706933926,
      "loss": 2.6291,
      "step": 8451
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6532673239707947,
      "learning_rate": 0.0005980100003480068,
      "loss": 2.5242,
      "step": 8452
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.777812659740448,
      "learning_rate": 0.0005980095299472153,
      "loss": 2.4165,
      "step": 8453
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.682979941368103,
      "learning_rate": 0.0005980090594910181,
      "loss": 2.8508,
      "step": 8454
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.7064735293388367,
      "learning_rate": 0.0005980085889794155,
      "loss": 2.6084,
      "step": 8455
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6248152852058411,
      "learning_rate": 0.0005980081184124074,
      "loss": 2.509,
      "step": 8456
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6693153381347656,
      "learning_rate": 0.000598007647789994,
      "loss": 2.702,
      "step": 8457
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6872237920761108,
      "learning_rate": 0.0005980071771121753,
      "loss": 2.5542,
      "step": 8458
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6388422250747681,
      "learning_rate": 0.0005980067063789515,
      "loss": 2.422,
      "step": 8459
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6447649002075195,
      "learning_rate": 0.0005980062355903226,
      "loss": 2.615,
      "step": 8460
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6844645142555237,
      "learning_rate": 0.0005980057647462887,
      "loss": 2.4731,
      "step": 8461
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.7059194445610046,
      "learning_rate": 0.0005980052938468499,
      "loss": 2.3642,
      "step": 8462
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.131361484527588,
      "learning_rate": 0.0005980048228920065,
      "loss": 2.4768,
      "step": 8463
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6987558007240295,
      "learning_rate": 0.0005980043518817581,
      "loss": 2.5111,
      "step": 8464
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6588321924209595,
      "learning_rate": 0.0005980038808161052,
      "loss": 2.6313,
      "step": 8465
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6155559420585632,
      "learning_rate": 0.0005980034096950477,
      "loss": 2.4818,
      "step": 8466
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6786640286445618,
      "learning_rate": 0.0005980029385185857,
      "loss": 2.5016,
      "step": 8467
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6541230082511902,
      "learning_rate": 0.0005980024672867194,
      "loss": 2.3581,
      "step": 8468
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.5884223580360413,
      "learning_rate": 0.0005980019959994488,
      "loss": 2.4689,
      "step": 8469
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6231350898742676,
      "learning_rate": 0.0005980015246567741,
      "loss": 2.5448,
      "step": 8470
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6330718398094177,
      "learning_rate": 0.0005980010532586951,
      "loss": 2.5656,
      "step": 8471
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6916640996932983,
      "learning_rate": 0.0005980005818052122,
      "loss": 2.6157,
      "step": 8472
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.676498532295227,
      "learning_rate": 0.0005980001102963254,
      "loss": 2.506,
      "step": 8473
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.5910623669624329,
      "learning_rate": 0.0005979996387320346,
      "loss": 2.6264,
      "step": 8474
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6366076469421387,
      "learning_rate": 0.0005979991671123402,
      "loss": 2.6145,
      "step": 8475
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6525055170059204,
      "learning_rate": 0.0005979986954372421,
      "loss": 2.5687,
      "step": 8476
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6305607557296753,
      "learning_rate": 0.0005979982237067404,
      "loss": 2.6198,
      "step": 8477
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.659572184085846,
      "learning_rate": 0.0005979977519208351,
      "loss": 2.5907,
      "step": 8478
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6678158640861511,
      "learning_rate": 0.0005979972800795265,
      "loss": 2.4384,
      "step": 8479
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6994065642356873,
      "learning_rate": 0.0005979968081828146,
      "loss": 2.4401,
      "step": 8480
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6250452995300293,
      "learning_rate": 0.0005979963362306995,
      "loss": 2.4092,
      "step": 8481
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6523944139480591,
      "learning_rate": 0.0005979958642231811,
      "loss": 2.7759,
      "step": 8482
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6237634420394897,
      "learning_rate": 0.0005979953921602597,
      "loss": 2.5815,
      "step": 8483
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6368109583854675,
      "learning_rate": 0.0005979949200419355,
      "loss": 2.4989,
      "step": 8484
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6080477833747864,
      "learning_rate": 0.0005979944478682083,
      "loss": 2.4399,
      "step": 8485
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6432697772979736,
      "learning_rate": 0.0005979939756390782,
      "loss": 2.6233,
      "step": 8486
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.7784900665283203,
      "learning_rate": 0.0005979935033545456,
      "loss": 2.2637,
      "step": 8487
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.5910780429840088,
      "learning_rate": 0.0005979930310146102,
      "loss": 2.5508,
      "step": 8488
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6705781817436218,
      "learning_rate": 0.0005979925586192722,
      "loss": 2.4387,
      "step": 8489
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6787704229354858,
      "learning_rate": 0.000597992086168532,
      "loss": 2.526,
      "step": 8490
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.7251600027084351,
      "learning_rate": 0.0005979916136623893,
      "loss": 2.2654,
      "step": 8491
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6801232695579529,
      "learning_rate": 0.0005979911411008444,
      "loss": 2.6024,
      "step": 8492
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6756705641746521,
      "learning_rate": 0.0005979906684838973,
      "loss": 2.4603,
      "step": 8493
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6426135897636414,
      "learning_rate": 0.0005979901958115481,
      "loss": 2.7845,
      "step": 8494
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6159802079200745,
      "learning_rate": 0.0005979897230837969,
      "loss": 2.4288,
      "step": 8495
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6719342470169067,
      "learning_rate": 0.0005979892503006439,
      "loss": 2.564,
      "step": 8496
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6697418093681335,
      "learning_rate": 0.0005979887774620889,
      "loss": 2.6755,
      "step": 8497
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.606090784072876,
      "learning_rate": 0.0005979883045681323,
      "loss": 2.6372,
      "step": 8498
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6525300145149231,
      "learning_rate": 0.000597987831618774,
      "loss": 2.6499,
      "step": 8499
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6907907724380493,
      "learning_rate": 0.000597987358614014,
      "loss": 2.4119,
      "step": 8500
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6218242049217224,
      "learning_rate": 0.0005979868855538527,
      "loss": 2.3418,
      "step": 8501
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6322482824325562,
      "learning_rate": 0.0005979864124382899,
      "loss": 2.3559,
      "step": 8502
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6272202134132385,
      "learning_rate": 0.0005979859392673259,
      "loss": 2.7624,
      "step": 8503
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6585597395896912,
      "learning_rate": 0.0005979854660409606,
      "loss": 2.2722,
      "step": 8504
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6414037346839905,
      "learning_rate": 0.0005979849927591943,
      "loss": 2.3999,
      "step": 8505
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6902700066566467,
      "learning_rate": 0.0005979845194220267,
      "loss": 2.4973,
      "step": 8506
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6702406406402588,
      "learning_rate": 0.0005979840460294584,
      "loss": 2.4597,
      "step": 8507
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.60816490650177,
      "learning_rate": 0.0005979835725814892,
      "loss": 2.3561,
      "step": 8508
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6599938869476318,
      "learning_rate": 0.0005979830990781193,
      "loss": 2.342,
      "step": 8509
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6824219226837158,
      "learning_rate": 0.0005979826255193486,
      "loss": 2.2162,
      "step": 8510
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.636642336845398,
      "learning_rate": 0.0005979821519051773,
      "loss": 2.4746,
      "step": 8511
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6602890491485596,
      "learning_rate": 0.0005979816782356054,
      "loss": 2.2962,
      "step": 8512
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.600551187992096,
      "learning_rate": 0.0005979812045106333,
      "loss": 2.2897,
      "step": 8513
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.8478353023529053,
      "learning_rate": 0.0005979807307302607,
      "loss": 2.6554,
      "step": 8514
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.5823778510093689,
      "learning_rate": 0.000597980256894488,
      "loss": 2.6637,
      "step": 8515
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6070581674575806,
      "learning_rate": 0.0005979797830033151,
      "loss": 2.4582,
      "step": 8516
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6249002814292908,
      "learning_rate": 0.0005979793090567421,
      "loss": 2.6398,
      "step": 8517
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.8895130157470703,
      "learning_rate": 0.0005979788350547691,
      "loss": 2.6632,
      "step": 8518
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.5641293525695801,
      "learning_rate": 0.0005979783609973962,
      "loss": 2.4381,
      "step": 8519
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.8143356442451477,
      "learning_rate": 0.0005979778868846237,
      "loss": 2.6057,
      "step": 8520
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6725379824638367,
      "learning_rate": 0.0005979774127164513,
      "loss": 2.5633,
      "step": 8521
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6144375801086426,
      "learning_rate": 0.0005979769384928792,
      "loss": 2.7788,
      "step": 8522
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.7496508359909058,
      "learning_rate": 0.0005979764642139077,
      "loss": 2.5856,
      "step": 8523
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6526830792427063,
      "learning_rate": 0.0005979759898795368,
      "loss": 2.4851,
      "step": 8524
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6489012241363525,
      "learning_rate": 0.0005979755154897664,
      "loss": 2.4718,
      "step": 8525
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.8634769916534424,
      "learning_rate": 0.0005979750410445967,
      "loss": 2.5049,
      "step": 8526
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6760299205780029,
      "learning_rate": 0.000597974566544028,
      "loss": 2.4711,
      "step": 8527
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6311708688735962,
      "learning_rate": 0.00059797409198806,
      "loss": 2.6227,
      "step": 8528
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6893621683120728,
      "learning_rate": 0.0005979736173766931,
      "loss": 2.54,
      "step": 8529
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6642248034477234,
      "learning_rate": 0.0005979731427099272,
      "loss": 2.3641,
      "step": 8530
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6674564480781555,
      "learning_rate": 0.0005979726679877627,
      "loss": 2.5435,
      "step": 8531
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.7293508052825928,
      "learning_rate": 0.0005979721932101991,
      "loss": 2.4447,
      "step": 8532
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.7242094278335571,
      "learning_rate": 0.0005979717183772371,
      "loss": 2.6046,
      "step": 8533
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.665008008480072,
      "learning_rate": 0.0005979712434888765,
      "loss": 2.6982,
      "step": 8534
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6628060936927795,
      "learning_rate": 0.0005979707685451174,
      "loss": 2.6809,
      "step": 8535
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6642266511917114,
      "learning_rate": 0.0005979702935459599,
      "loss": 2.7123,
      "step": 8536
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6808599829673767,
      "learning_rate": 0.0005979698184914041,
      "loss": 2.6192,
      "step": 8537
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6420002579689026,
      "learning_rate": 0.0005979693433814501,
      "loss": 2.4581,
      "step": 8538
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6361576318740845,
      "learning_rate": 0.0005979688682160979,
      "loss": 2.5859,
      "step": 8539
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6894296407699585,
      "learning_rate": 0.0005979683929953478,
      "loss": 2.4126,
      "step": 8540
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6532701849937439,
      "learning_rate": 0.0005979679177191996,
      "loss": 2.6065,
      "step": 8541
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.7170461416244507,
      "learning_rate": 0.0005979674423876537,
      "loss": 2.6935,
      "step": 8542
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6941001415252686,
      "learning_rate": 0.0005979669670007099,
      "loss": 2.3492,
      "step": 8543
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6125115156173706,
      "learning_rate": 0.0005979664915583685,
      "loss": 2.5695,
      "step": 8544
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.5972003936767578,
      "learning_rate": 0.0005979660160606294,
      "loss": 2.4602,
      "step": 8545
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.654306948184967,
      "learning_rate": 0.0005979655405074928,
      "loss": 2.4124,
      "step": 8546
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6758958101272583,
      "learning_rate": 0.0005979650648989589,
      "loss": 2.4752,
      "step": 8547
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.63906329870224,
      "learning_rate": 0.0005979645892350276,
      "loss": 2.3924,
      "step": 8548
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.7276355028152466,
      "learning_rate": 0.000597964113515699,
      "loss": 2.4505,
      "step": 8549
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.641037106513977,
      "learning_rate": 0.0005979636377409734,
      "loss": 2.3265,
      "step": 8550
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.7154620289802551,
      "learning_rate": 0.0005979631619108507,
      "loss": 2.5621,
      "step": 8551
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6839312314987183,
      "learning_rate": 0.0005979626860253309,
      "loss": 2.536,
      "step": 8552
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.7932697534561157,
      "learning_rate": 0.0005979622100844142,
      "loss": 2.6404,
      "step": 8553
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.607150137424469,
      "learning_rate": 0.0005979617340881009,
      "loss": 2.5167,
      "step": 8554
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.726447582244873,
      "learning_rate": 0.0005979612580363907,
      "loss": 2.5203,
      "step": 8555
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.7443660497665405,
      "learning_rate": 0.0005979607819292839,
      "loss": 2.4909,
      "step": 8556
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6814913749694824,
      "learning_rate": 0.0005979603057667806,
      "loss": 2.5473,
      "step": 8557
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.695855975151062,
      "learning_rate": 0.000597959829548881,
      "loss": 2.4017,
      "step": 8558
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.7381858825683594,
      "learning_rate": 0.0005979593532755848,
      "loss": 2.6556,
      "step": 8559
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6651283502578735,
      "learning_rate": 0.0005979588769468925,
      "loss": 2.6884,
      "step": 8560
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.7407071590423584,
      "learning_rate": 0.0005979584005628041,
      "loss": 2.3479,
      "step": 8561
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6830853819847107,
      "learning_rate": 0.0005979579241233194,
      "loss": 2.4701,
      "step": 8562
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6341378688812256,
      "learning_rate": 0.0005979574476284387,
      "loss": 2.2461,
      "step": 8563
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.767728328704834,
      "learning_rate": 0.0005979569710781622,
      "loss": 2.5002,
      "step": 8564
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.632210910320282,
      "learning_rate": 0.0005979564944724897,
      "loss": 2.4487,
      "step": 8565
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6361459493637085,
      "learning_rate": 0.0005979560178114217,
      "loss": 2.3692,
      "step": 8566
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.608773410320282,
      "learning_rate": 0.000597955541094958,
      "loss": 2.5129,
      "step": 8567
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6762409806251526,
      "learning_rate": 0.0005979550643230987,
      "loss": 2.5892,
      "step": 8568
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6295363306999207,
      "learning_rate": 0.0005979545874958438,
      "loss": 2.4049,
      "step": 8569
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6554909348487854,
      "learning_rate": 0.0005979541106131937,
      "loss": 2.8213,
      "step": 8570
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6200321912765503,
      "learning_rate": 0.0005979536336751482,
      "loss": 2.301,
      "step": 8571
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.7468177676200867,
      "learning_rate": 0.0005979531566817076,
      "loss": 2.3148,
      "step": 8572
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.7060205936431885,
      "learning_rate": 0.0005979526796328718,
      "loss": 2.928,
      "step": 8573
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6616849303245544,
      "learning_rate": 0.000597952202528641,
      "loss": 2.6261,
      "step": 8574
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6402880549430847,
      "learning_rate": 0.0005979517253690153,
      "loss": 2.4322,
      "step": 8575
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.615038275718689,
      "learning_rate": 0.0005979512481539946,
      "loss": 2.5998,
      "step": 8576
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.624523937702179,
      "learning_rate": 0.0005979507708835793,
      "loss": 2.3445,
      "step": 8577
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6595109701156616,
      "learning_rate": 0.0005979502935577692,
      "loss": 2.5196,
      "step": 8578
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6027969121932983,
      "learning_rate": 0.0005979498161765647,
      "loss": 2.405,
      "step": 8579
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.7606577277183533,
      "learning_rate": 0.0005979493387399655,
      "loss": 2.4948,
      "step": 8580
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6664678454399109,
      "learning_rate": 0.0005979488612479721,
      "loss": 2.5297,
      "step": 8581
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6308221220970154,
      "learning_rate": 0.0005979483837005842,
      "loss": 2.4482,
      "step": 8582
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6805912256240845,
      "learning_rate": 0.0005979479060978022,
      "loss": 2.5093,
      "step": 8583
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6662761569023132,
      "learning_rate": 0.0005979474284396259,
      "loss": 2.5081,
      "step": 8584
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6793003678321838,
      "learning_rate": 0.0005979469507260557,
      "loss": 2.4409,
      "step": 8585
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6493514776229858,
      "learning_rate": 0.0005979464729570915,
      "loss": 2.5293,
      "step": 8586
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6227704882621765,
      "learning_rate": 0.0005979459951327334,
      "loss": 2.4437,
      "step": 8587
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.7985476851463318,
      "learning_rate": 0.0005979455172529816,
      "loss": 2.3402,
      "step": 8588
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6992195248603821,
      "learning_rate": 0.0005979450393178359,
      "loss": 2.7952,
      "step": 8589
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.8294516801834106,
      "learning_rate": 0.0005979445613272967,
      "loss": 2.4532,
      "step": 8590
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.9336721897125244,
      "learning_rate": 0.0005979440832813641,
      "loss": 2.4815,
      "step": 8591
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6653026342391968,
      "learning_rate": 0.0005979436051800379,
      "loss": 2.5062,
      "step": 8592
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.7697787880897522,
      "learning_rate": 0.0005979431270233185,
      "loss": 2.2263,
      "step": 8593
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.7802372574806213,
      "learning_rate": 0.0005979426488112058,
      "loss": 2.4523,
      "step": 8594
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6172847151756287,
      "learning_rate": 0.0005979421705436999,
      "loss": 2.3897,
      "step": 8595
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.7795912027359009,
      "learning_rate": 0.0005979416922208009,
      "loss": 2.2564,
      "step": 8596
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.8760755658149719,
      "learning_rate": 0.000597941213842509,
      "loss": 2.5299,
      "step": 8597
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.7497047185897827,
      "learning_rate": 0.0005979407354088242,
      "loss": 2.4698,
      "step": 8598
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.8591505885124207,
      "learning_rate": 0.0005979402569197466,
      "loss": 2.211,
      "step": 8599
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.9255728721618652,
      "learning_rate": 0.0005979397783752761,
      "loss": 2.6307,
      "step": 8600
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.669773519039154,
      "learning_rate": 0.0005979392997754132,
      "loss": 2.6144,
      "step": 8601
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.7578276991844177,
      "learning_rate": 0.0005979388211201577,
      "loss": 2.4759,
      "step": 8602
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.7170863151550293,
      "learning_rate": 0.0005979383424095097,
      "loss": 2.7504,
      "step": 8603
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6923009157180786,
      "learning_rate": 0.0005979378636434693,
      "loss": 2.2653,
      "step": 8604
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.867741584777832,
      "learning_rate": 0.0005979373848220367,
      "loss": 2.5983,
      "step": 8605
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6992696523666382,
      "learning_rate": 0.0005979369059452119,
      "loss": 2.4817,
      "step": 8606
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.9467296004295349,
      "learning_rate": 0.0005979364270129951,
      "loss": 2.595,
      "step": 8607
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6494059562683105,
      "learning_rate": 0.0005979359480253862,
      "loss": 2.4329,
      "step": 8608
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6788463592529297,
      "learning_rate": 0.0005979354689823853,
      "loss": 2.3204,
      "step": 8609
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6861779689788818,
      "learning_rate": 0.0005979349898839928,
      "loss": 2.5088,
      "step": 8610
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6234354972839355,
      "learning_rate": 0.0005979345107302084,
      "loss": 2.7088,
      "step": 8611
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.7354544401168823,
      "learning_rate": 0.0005979340315210324,
      "loss": 2.3966,
      "step": 8612
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.7475653886795044,
      "learning_rate": 0.0005979335522564648,
      "loss": 2.5681,
      "step": 8613
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6339936852455139,
      "learning_rate": 0.0005979330729365057,
      "loss": 2.3128,
      "step": 8614
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6943144798278809,
      "learning_rate": 0.0005979325935611553,
      "loss": 2.2739,
      "step": 8615
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6554667949676514,
      "learning_rate": 0.0005979321141304135,
      "loss": 2.4767,
      "step": 8616
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6939142346382141,
      "learning_rate": 0.0005979316346442806,
      "loss": 2.3763,
      "step": 8617
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6440284252166748,
      "learning_rate": 0.0005979311551027565,
      "loss": 2.6155,
      "step": 8618
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.8274462223052979,
      "learning_rate": 0.0005979306755058415,
      "loss": 2.7524,
      "step": 8619
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.7663728594779968,
      "learning_rate": 0.0005979301958535354,
      "loss": 2.3436,
      "step": 8620
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.7523809671401978,
      "learning_rate": 0.0005979297161458386,
      "loss": 2.3353,
      "step": 8621
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.659089982509613,
      "learning_rate": 0.0005979292363827509,
      "loss": 2.5727,
      "step": 8622
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6413401365280151,
      "learning_rate": 0.0005979287565642727,
      "loss": 2.358,
      "step": 8623
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6257014274597168,
      "learning_rate": 0.0005979282766904037,
      "loss": 2.4656,
      "step": 8624
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6758393049240112,
      "learning_rate": 0.0005979277967611445,
      "loss": 2.3295,
      "step": 8625
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6125257015228271,
      "learning_rate": 0.0005979273167764947,
      "loss": 2.4194,
      "step": 8626
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.5683146119117737,
      "learning_rate": 0.0005979268367364546,
      "loss": 2.448,
      "step": 8627
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6959031224250793,
      "learning_rate": 0.0005979263566410244,
      "loss": 2.5727,
      "step": 8628
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6347568035125732,
      "learning_rate": 0.0005979258764902039,
      "loss": 2.4118,
      "step": 8629
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6132092475891113,
      "learning_rate": 0.0005979253962839934,
      "loss": 2.4887,
      "step": 8630
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.7147738933563232,
      "learning_rate": 0.0005979249160223929,
      "loss": 2.2387,
      "step": 8631
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6589148044586182,
      "learning_rate": 0.0005979244357054026,
      "loss": 2.4281,
      "step": 8632
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.7190052270889282,
      "learning_rate": 0.0005979239553330226,
      "loss": 2.6796,
      "step": 8633
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6286527514457703,
      "learning_rate": 0.0005979234749052528,
      "loss": 2.5673,
      "step": 8634
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.782584547996521,
      "learning_rate": 0.0005979229944220936,
      "loss": 2.7074,
      "step": 8635
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6323524713516235,
      "learning_rate": 0.0005979225138835446,
      "loss": 2.5896,
      "step": 8636
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6135575771331787,
      "learning_rate": 0.0005979220332896063,
      "loss": 2.472,
      "step": 8637
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.7181174159049988,
      "learning_rate": 0.0005979215526402788,
      "loss": 2.5698,
      "step": 8638
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.9215032458305359,
      "learning_rate": 0.000597921071935562,
      "loss": 2.508,
      "step": 8639
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6004571318626404,
      "learning_rate": 0.0005979205911754559,
      "loss": 2.4745,
      "step": 8640
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.70345139503479,
      "learning_rate": 0.0005979201103599608,
      "loss": 2.3565,
      "step": 8641
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6625435948371887,
      "learning_rate": 0.0005979196294890768,
      "loss": 2.7195,
      "step": 8642
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.7129363417625427,
      "learning_rate": 0.0005979191485628039,
      "loss": 2.5474,
      "step": 8643
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.7612552046775818,
      "learning_rate": 0.0005979186675811421,
      "loss": 2.5597,
      "step": 8644
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.7214411497116089,
      "learning_rate": 0.0005979181865440918,
      "loss": 2.5563,
      "step": 8645
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6878314018249512,
      "learning_rate": 0.0005979177054516527,
      "loss": 2.5072,
      "step": 8646
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.7453218102455139,
      "learning_rate": 0.0005979172243038252,
      "loss": 2.3843,
      "step": 8647
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.7038173079490662,
      "learning_rate": 0.0005979167431006092,
      "loss": 2.5955,
      "step": 8648
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.775863528251648,
      "learning_rate": 0.0005979162618420049,
      "loss": 2.317,
      "step": 8649
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.8210248947143555,
      "learning_rate": 0.0005979157805280122,
      "loss": 2.4906,
      "step": 8650
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6184552907943726,
      "learning_rate": 0.0005979152991586314,
      "loss": 2.8726,
      "step": 8651
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6474803686141968,
      "learning_rate": 0.0005979148177338627,
      "loss": 2.4224,
      "step": 8652
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6781133413314819,
      "learning_rate": 0.0005979143362537058,
      "loss": 2.7253,
      "step": 8653
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.7795550227165222,
      "learning_rate": 0.0005979138547181611,
      "loss": 2.5015,
      "step": 8654
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6464529633522034,
      "learning_rate": 0.0005979133731272286,
      "loss": 2.3749,
      "step": 8655
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6639126539230347,
      "learning_rate": 0.0005979128914809083,
      "loss": 2.3463,
      "step": 8656
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.7972289323806763,
      "learning_rate": 0.0005979124097792004,
      "loss": 2.3576,
      "step": 8657
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6367520093917847,
      "learning_rate": 0.000597911928022105,
      "loss": 2.6593,
      "step": 8658
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6498738527297974,
      "learning_rate": 0.0005979114462096222,
      "loss": 2.4566,
      "step": 8659
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.7679353356361389,
      "learning_rate": 0.0005979109643417519,
      "loss": 2.6504,
      "step": 8660
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6968933939933777,
      "learning_rate": 0.0005979104824184944,
      "loss": 2.8282,
      "step": 8661
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6076825857162476,
      "learning_rate": 0.0005979100004398497,
      "loss": 2.5612,
      "step": 8662
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6922442317008972,
      "learning_rate": 0.0005979095184058179,
      "loss": 2.3236,
      "step": 8663
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.7748520374298096,
      "learning_rate": 0.0005979090363163991,
      "loss": 2.5506,
      "step": 8664
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6437479853630066,
      "learning_rate": 0.0005979085541715933,
      "loss": 2.5813,
      "step": 8665
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6855214834213257,
      "learning_rate": 0.0005979080719714009,
      "loss": 2.6041,
      "step": 8666
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6602147817611694,
      "learning_rate": 0.0005979075897158216,
      "loss": 2.4507,
      "step": 8667
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6628488898277283,
      "learning_rate": 0.0005979071074048558,
      "loss": 2.456,
      "step": 8668
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6445878744125366,
      "learning_rate": 0.0005979066250385033,
      "loss": 2.5158,
      "step": 8669
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6404288411140442,
      "learning_rate": 0.0005979061426167644,
      "loss": 2.6156,
      "step": 8670
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.7326463460922241,
      "learning_rate": 0.000597905660139639,
      "loss": 2.4557,
      "step": 8671
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6704663634300232,
      "learning_rate": 0.0005979051776071275,
      "loss": 2.5268,
      "step": 8672
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.66791170835495,
      "learning_rate": 0.0005979046950192297,
      "loss": 2.613,
      "step": 8673
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6134548783302307,
      "learning_rate": 0.0005979042123759458,
      "loss": 2.4356,
      "step": 8674
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6398153901100159,
      "learning_rate": 0.0005979037296772758,
      "loss": 2.5015,
      "step": 8675
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.5917001366615295,
      "learning_rate": 0.00059790324692322,
      "loss": 2.4727,
      "step": 8676
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.8673519492149353,
      "learning_rate": 0.0005979027641137784,
      "loss": 2.6386,
      "step": 8677
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.680990993976593,
      "learning_rate": 0.000597902281248951,
      "loss": 2.5222,
      "step": 8678
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6420652866363525,
      "learning_rate": 0.0005979017983287379,
      "loss": 2.4971,
      "step": 8679
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.678249716758728,
      "learning_rate": 0.0005979013153531393,
      "loss": 2.4854,
      "step": 8680
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.7666689157485962,
      "learning_rate": 0.0005979008323221552,
      "loss": 2.5157,
      "step": 8681
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6304771900177002,
      "learning_rate": 0.0005979003492357856,
      "loss": 2.6798,
      "step": 8682
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.7407318353652954,
      "learning_rate": 0.0005978998660940308,
      "loss": 2.5314,
      "step": 8683
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6387050747871399,
      "learning_rate": 0.0005978993828968909,
      "loss": 2.6353,
      "step": 8684
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.7750783562660217,
      "learning_rate": 0.0005978988996443657,
      "loss": 2.6766,
      "step": 8685
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6975477337837219,
      "learning_rate": 0.0005978984163364556,
      "loss": 2.5442,
      "step": 8686
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6329005360603333,
      "learning_rate": 0.0005978979329731604,
      "loss": 2.5985,
      "step": 8687
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.7552101612091064,
      "learning_rate": 0.0005978974495544806,
      "loss": 2.725,
      "step": 8688
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.869186282157898,
      "learning_rate": 0.0005978969660804159,
      "loss": 2.5248,
      "step": 8689
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6280601024627686,
      "learning_rate": 0.0005978964825509664,
      "loss": 2.5347,
      "step": 8690
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.8614486455917358,
      "learning_rate": 0.0005978959989661325,
      "loss": 2.6301,
      "step": 8691
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6238195300102234,
      "learning_rate": 0.0005978955153259141,
      "loss": 2.3291,
      "step": 8692
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.646362841129303,
      "learning_rate": 0.0005978950316303112,
      "loss": 2.604,
      "step": 8693
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6069594025611877,
      "learning_rate": 0.000597894547879324,
      "loss": 2.3757,
      "step": 8694
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.7144143581390381,
      "learning_rate": 0.0005978940640729526,
      "loss": 2.3402,
      "step": 8695
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6185102462768555,
      "learning_rate": 0.0005978935802111972,
      "loss": 2.4718,
      "step": 8696
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.5995498299598694,
      "learning_rate": 0.0005978930962940576,
      "loss": 2.6312,
      "step": 8697
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.7407141923904419,
      "learning_rate": 0.0005978926123215341,
      "loss": 2.6153,
      "step": 8698
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6858460903167725,
      "learning_rate": 0.0005978921282936267,
      "loss": 2.7448,
      "step": 8699
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.7305383682250977,
      "learning_rate": 0.0005978916442103357,
      "loss": 2.6204,
      "step": 8700
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6910662651062012,
      "learning_rate": 0.0005978911600716608,
      "loss": 2.4769,
      "step": 8701
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.5997154712677002,
      "learning_rate": 0.0005978906758776025,
      "loss": 2.5942,
      "step": 8702
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6087592244148254,
      "learning_rate": 0.0005978901916281606,
      "loss": 2.5392,
      "step": 8703
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6352441906929016,
      "learning_rate": 0.0005978897073233353,
      "loss": 2.5764,
      "step": 8704
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6428564190864563,
      "learning_rate": 0.0005978892229631267,
      "loss": 2.5024,
      "step": 8705
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6004164218902588,
      "learning_rate": 0.0005978887385475348,
      "loss": 2.628,
      "step": 8706
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6374674439430237,
      "learning_rate": 0.0005978882540765599,
      "loss": 2.6619,
      "step": 8707
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6527577042579651,
      "learning_rate": 0.0005978877695502017,
      "loss": 2.5262,
      "step": 8708
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6552874445915222,
      "learning_rate": 0.0005978872849684608,
      "loss": 2.5837,
      "step": 8709
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.7628086805343628,
      "learning_rate": 0.000597886800331337,
      "loss": 2.7545,
      "step": 8710
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.600767970085144,
      "learning_rate": 0.0005978863156388303,
      "loss": 2.4458,
      "step": 8711
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6241584420204163,
      "learning_rate": 0.0005978858308909409,
      "loss": 2.7696,
      "step": 8712
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6034501791000366,
      "learning_rate": 0.000597885346087669,
      "loss": 2.5253,
      "step": 8713
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.7316293120384216,
      "learning_rate": 0.0005978848612290146,
      "loss": 2.4973,
      "step": 8714
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6037749648094177,
      "learning_rate": 0.0005978843763149778,
      "loss": 2.4408,
      "step": 8715
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6237314939498901,
      "learning_rate": 0.0005978838913455586,
      "loss": 2.6094,
      "step": 8716
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.7362722754478455,
      "learning_rate": 0.0005978834063207571,
      "loss": 2.3649,
      "step": 8717
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6197445392608643,
      "learning_rate": 0.0005978829212405735,
      "loss": 2.7719,
      "step": 8718
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.1875450611114502,
      "learning_rate": 0.0005978824361050079,
      "loss": 2.3328,
      "step": 8719
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.7218047380447388,
      "learning_rate": 0.0005978819509140604,
      "loss": 2.8482,
      "step": 8720
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.7130162715911865,
      "learning_rate": 0.0005978814656677308,
      "loss": 2.7653,
      "step": 8721
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.8644399046897888,
      "learning_rate": 0.0005978809803660196,
      "loss": 2.4073,
      "step": 8722
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6476062536239624,
      "learning_rate": 0.0005978804950089266,
      "loss": 2.6722,
      "step": 8723
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6936860680580139,
      "learning_rate": 0.0005978800095964521,
      "loss": 2.4434,
      "step": 8724
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.7416517734527588,
      "learning_rate": 0.0005978795241285959,
      "loss": 2.2007,
      "step": 8725
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.613042414188385,
      "learning_rate": 0.0005978790386053584,
      "loss": 2.5841,
      "step": 8726
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6479109525680542,
      "learning_rate": 0.0005978785530267396,
      "loss": 2.6233,
      "step": 8727
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6439782977104187,
      "learning_rate": 0.0005978780673927395,
      "loss": 2.4636,
      "step": 8728
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6567809581756592,
      "learning_rate": 0.0005978775817033581,
      "loss": 2.332,
      "step": 8729
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6652424335479736,
      "learning_rate": 0.0005978770959585959,
      "loss": 2.4656,
      "step": 8730
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.585483193397522,
      "learning_rate": 0.0005978766101584526,
      "loss": 2.7082,
      "step": 8731
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6371074914932251,
      "learning_rate": 0.0005978761243029284,
      "loss": 2.2838,
      "step": 8732
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6675920486450195,
      "learning_rate": 0.0005978756383920235,
      "loss": 2.534,
      "step": 8733
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6264447569847107,
      "learning_rate": 0.0005978751524257378,
      "loss": 2.6491,
      "step": 8734
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.7017096281051636,
      "learning_rate": 0.0005978746664040715,
      "loss": 2.3808,
      "step": 8735
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.630547821521759,
      "learning_rate": 0.0005978741803270247,
      "loss": 2.5131,
      "step": 8736
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6460360884666443,
      "learning_rate": 0.0005978736941945974,
      "loss": 2.6561,
      "step": 8737
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6647377014160156,
      "learning_rate": 0.0005978732080067898,
      "loss": 2.3873,
      "step": 8738
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.7252362966537476,
      "learning_rate": 0.0005978727217636021,
      "loss": 2.5361,
      "step": 8739
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6196104288101196,
      "learning_rate": 0.000597872235465034,
      "loss": 2.438,
      "step": 8740
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.7168704867362976,
      "learning_rate": 0.0005978717491110859,
      "loss": 2.6046,
      "step": 8741
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.626266360282898,
      "learning_rate": 0.0005978712627017578,
      "loss": 2.797,
      "step": 8742
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.70225989818573,
      "learning_rate": 0.0005978707762370499,
      "loss": 2.5432,
      "step": 8743
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6817074418067932,
      "learning_rate": 0.0005978702897169622,
      "loss": 2.6767,
      "step": 8744
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6968148350715637,
      "learning_rate": 0.0005978698031414947,
      "loss": 2.3209,
      "step": 8745
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6392139196395874,
      "learning_rate": 0.0005978693165106477,
      "loss": 2.3834,
      "step": 8746
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.702943742275238,
      "learning_rate": 0.0005978688298244211,
      "loss": 2.3993,
      "step": 8747
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6114577054977417,
      "learning_rate": 0.0005978683430828151,
      "loss": 2.4811,
      "step": 8748
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6849681735038757,
      "learning_rate": 0.0005978678562858297,
      "loss": 2.7966,
      "step": 8749
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6714168787002563,
      "learning_rate": 0.000597867369433465,
      "loss": 2.3851,
      "step": 8750
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6568624973297119,
      "learning_rate": 0.0005978668825257212,
      "loss": 2.5188,
      "step": 8751
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.5972064733505249,
      "learning_rate": 0.0005978663955625983,
      "loss": 2.5477,
      "step": 8752
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6157382130622864,
      "learning_rate": 0.0005978659085440965,
      "loss": 2.5757,
      "step": 8753
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6628873348236084,
      "learning_rate": 0.0005978654214702157,
      "loss": 2.3465,
      "step": 8754
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6111583709716797,
      "learning_rate": 0.0005978649343409561,
      "loss": 2.436,
      "step": 8755
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.5992520451545715,
      "learning_rate": 0.0005978644471563178,
      "loss": 2.5035,
      "step": 8756
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.5924449563026428,
      "learning_rate": 0.0005978639599163009,
      "loss": 2.3636,
      "step": 8757
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.7251550555229187,
      "learning_rate": 0.0005978634726209054,
      "loss": 2.505,
      "step": 8758
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.5746108889579773,
      "learning_rate": 0.0005978629852701315,
      "loss": 2.4528,
      "step": 8759
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6363506317138672,
      "learning_rate": 0.0005978624978639793,
      "loss": 2.3858,
      "step": 8760
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6133317351341248,
      "learning_rate": 0.0005978620104024489,
      "loss": 2.4258,
      "step": 8761
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6294135451316833,
      "learning_rate": 0.0005978615228855401,
      "loss": 2.3926,
      "step": 8762
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6479763388633728,
      "learning_rate": 0.0005978610353132534,
      "loss": 2.4073,
      "step": 8763
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6761638522148132,
      "learning_rate": 0.0005978605476855887,
      "loss": 2.6815,
      "step": 8764
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6497777104377747,
      "learning_rate": 0.0005978600600025461,
      "loss": 2.6022,
      "step": 8765
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6786485910415649,
      "learning_rate": 0.0005978595722641256,
      "loss": 2.2618,
      "step": 8766
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6118470430374146,
      "learning_rate": 0.0005978590844703274,
      "loss": 2.4213,
      "step": 8767
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.5779686570167542,
      "learning_rate": 0.0005978585966211516,
      "loss": 2.4838,
      "step": 8768
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6334221363067627,
      "learning_rate": 0.0005978581087165983,
      "loss": 2.6696,
      "step": 8769
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.603775680065155,
      "learning_rate": 0.0005978576207566676,
      "loss": 2.5836,
      "step": 8770
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.5689224600791931,
      "learning_rate": 0.0005978571327413594,
      "loss": 2.2907,
      "step": 8771
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6905726194381714,
      "learning_rate": 0.0005978566446706741,
      "loss": 2.5138,
      "step": 8772
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6143700480461121,
      "learning_rate": 0.0005978561565446115,
      "loss": 2.5089,
      "step": 8773
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6122174263000488,
      "learning_rate": 0.0005978556683631719,
      "loss": 2.7212,
      "step": 8774
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6063877940177917,
      "learning_rate": 0.0005978551801263551,
      "loss": 2.4163,
      "step": 8775
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.633496105670929,
      "learning_rate": 0.0005978546918341617,
      "loss": 2.3903,
      "step": 8776
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6109859943389893,
      "learning_rate": 0.0005978542034865913,
      "loss": 2.4984,
      "step": 8777
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.5815884470939636,
      "learning_rate": 0.0005978537150836443,
      "loss": 2.4772,
      "step": 8778
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6178130507469177,
      "learning_rate": 0.0005978532266253207,
      "loss": 2.5469,
      "step": 8779
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6988568902015686,
      "learning_rate": 0.0005978527381116204,
      "loss": 2.7462,
      "step": 8780
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.702244222164154,
      "learning_rate": 0.0005978522495425437,
      "loss": 2.4732,
      "step": 8781
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.5903748869895935,
      "learning_rate": 0.0005978517609180906,
      "loss": 2.484,
      "step": 8782
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6397982835769653,
      "learning_rate": 0.0005978512722382614,
      "loss": 2.6993,
      "step": 8783
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6158543229103088,
      "learning_rate": 0.0005978507835030559,
      "loss": 2.6961,
      "step": 8784
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.5861629247665405,
      "learning_rate": 0.0005978502947124744,
      "loss": 2.3308,
      "step": 8785
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6032495498657227,
      "learning_rate": 0.0005978498058665169,
      "loss": 2.6604,
      "step": 8786
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.5696730017662048,
      "learning_rate": 0.0005978493169651834,
      "loss": 2.574,
      "step": 8787
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6506989002227783,
      "learning_rate": 0.0005978488280084742,
      "loss": 2.537,
      "step": 8788
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6794750094413757,
      "learning_rate": 0.0005978483389963892,
      "loss": 2.4832,
      "step": 8789
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.573340892791748,
      "learning_rate": 0.0005978478499289285,
      "loss": 2.4202,
      "step": 8790
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.5789421796798706,
      "learning_rate": 0.0005978473608060925,
      "loss": 2.3068,
      "step": 8791
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6532979607582092,
      "learning_rate": 0.0005978468716278808,
      "loss": 2.6238,
      "step": 8792
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6498589515686035,
      "learning_rate": 0.0005978463823942939,
      "loss": 2.6753,
      "step": 8793
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6078606247901917,
      "learning_rate": 0.0005978458931053317,
      "loss": 2.3335,
      "step": 8794
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6105705499649048,
      "learning_rate": 0.0005978454037609943,
      "loss": 2.4084,
      "step": 8795
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6952765583992004,
      "learning_rate": 0.0005978449143612818,
      "loss": 2.569,
      "step": 8796
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6305511593818665,
      "learning_rate": 0.0005978444249061945,
      "loss": 2.5205,
      "step": 8797
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.601637065410614,
      "learning_rate": 0.000597843935395732,
      "loss": 2.4287,
      "step": 8798
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6408176422119141,
      "learning_rate": 0.0005978434458298948,
      "loss": 2.4129,
      "step": 8799
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.686555027961731,
      "learning_rate": 0.000597842956208683,
      "loss": 2.3355,
      "step": 8800
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.5795655250549316,
      "learning_rate": 0.0005978424665320965,
      "loss": 2.2877,
      "step": 8801
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6622928380966187,
      "learning_rate": 0.0005978419768001355,
      "loss": 2.4352,
      "step": 8802
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6662110090255737,
      "learning_rate": 0.0005978414870128,
      "loss": 2.5634,
      "step": 8803
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.619411826133728,
      "learning_rate": 0.0005978409971700902,
      "loss": 2.4668,
      "step": 8804
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.7360923886299133,
      "learning_rate": 0.000597840507272006,
      "loss": 2.4842,
      "step": 8805
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6807368993759155,
      "learning_rate": 0.0005978400173185477,
      "loss": 2.3446,
      "step": 8806
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.7243147492408752,
      "learning_rate": 0.0005978395273097154,
      "loss": 2.1858,
      "step": 8807
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.7657108902931213,
      "learning_rate": 0.0005978390372455089,
      "loss": 2.505,
      "step": 8808
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6300418972969055,
      "learning_rate": 0.0005978385471259287,
      "loss": 2.4116,
      "step": 8809
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.64210444688797,
      "learning_rate": 0.0005978380569509746,
      "loss": 2.4112,
      "step": 8810
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6545637845993042,
      "learning_rate": 0.0005978375667206469,
      "loss": 2.4802,
      "step": 8811
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6826974153518677,
      "learning_rate": 0.0005978370764349454,
      "loss": 2.6048,
      "step": 8812
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.8177730441093445,
      "learning_rate": 0.0005978365860938705,
      "loss": 2.4808,
      "step": 8813
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.8332353830337524,
      "learning_rate": 0.0005978360956974221,
      "loss": 2.5526,
      "step": 8814
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.7084010243415833,
      "learning_rate": 0.0005978356052456005,
      "loss": 2.4623,
      "step": 8815
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.7106593251228333,
      "learning_rate": 0.0005978351147384054,
      "loss": 2.7241,
      "step": 8816
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.5628960132598877,
      "learning_rate": 0.0005978346241758373,
      "loss": 2.693,
      "step": 8817
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.7912003993988037,
      "learning_rate": 0.0005978341335578961,
      "loss": 2.3843,
      "step": 8818
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.7122154235839844,
      "learning_rate": 0.0005978336428845818,
      "loss": 2.454,
      "step": 8819
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.676338255405426,
      "learning_rate": 0.0005978331521558947,
      "loss": 2.4013,
      "step": 8820
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6064475774765015,
      "learning_rate": 0.0005978326613718348,
      "loss": 2.3071,
      "step": 8821
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6205782890319824,
      "learning_rate": 0.0005978321705324021,
      "loss": 2.7213,
      "step": 8822
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6238483786582947,
      "learning_rate": 0.0005978316796375968,
      "loss": 2.7397,
      "step": 8823
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6270769238471985,
      "learning_rate": 0.000597831188687419,
      "loss": 2.5177,
      "step": 8824
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6629714965820312,
      "learning_rate": 0.0005978306976818688,
      "loss": 2.5565,
      "step": 8825
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.5730498433113098,
      "learning_rate": 0.0005978302066209462,
      "loss": 2.5247,
      "step": 8826
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6246973872184753,
      "learning_rate": 0.0005978297155046513,
      "loss": 2.4065,
      "step": 8827
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6128515005111694,
      "learning_rate": 0.0005978292243329843,
      "loss": 2.5673,
      "step": 8828
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6713677644729614,
      "learning_rate": 0.0005978287331059452,
      "loss": 2.6424,
      "step": 8829
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6743282675743103,
      "learning_rate": 0.0005978282418235341,
      "loss": 2.6342,
      "step": 8830
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6429764628410339,
      "learning_rate": 0.0005978277504857512,
      "loss": 2.3901,
      "step": 8831
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.5935376882553101,
      "learning_rate": 0.0005978272590925964,
      "loss": 2.4668,
      "step": 8832
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6817456483840942,
      "learning_rate": 0.0005978267676440699,
      "loss": 2.48,
      "step": 8833
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6674640774726868,
      "learning_rate": 0.0005978262761401719,
      "loss": 2.501,
      "step": 8834
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6347115635871887,
      "learning_rate": 0.0005978257845809022,
      "loss": 2.5528,
      "step": 8835
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.635552704334259,
      "learning_rate": 0.0005978252929662612,
      "loss": 2.3924,
      "step": 8836
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.634221076965332,
      "learning_rate": 0.0005978248012962488,
      "loss": 2.4107,
      "step": 8837
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6180589199066162,
      "learning_rate": 0.0005978243095708652,
      "loss": 2.4423,
      "step": 8838
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6032606363296509,
      "learning_rate": 0.0005978238177901104,
      "loss": 2.422,
      "step": 8839
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6001256704330444,
      "learning_rate": 0.0005978233259539844,
      "loss": 2.2644,
      "step": 8840
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6788462996482849,
      "learning_rate": 0.0005978228340624875,
      "loss": 2.6514,
      "step": 8841
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6117393970489502,
      "learning_rate": 0.0005978223421156198,
      "loss": 2.7518,
      "step": 8842
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6397362351417542,
      "learning_rate": 0.0005978218501133813,
      "loss": 2.3302,
      "step": 8843
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.627070426940918,
      "learning_rate": 0.000597821358055772,
      "loss": 2.4396,
      "step": 8844
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6796997785568237,
      "learning_rate": 0.0005978208659427922,
      "loss": 2.6502,
      "step": 8845
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6509037613868713,
      "learning_rate": 0.0005978203737744418,
      "loss": 2.5484,
      "step": 8846
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6435530185699463,
      "learning_rate": 0.000597819881550721,
      "loss": 2.5461,
      "step": 8847
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.5633792877197266,
      "learning_rate": 0.0005978193892716299,
      "loss": 2.3858,
      "step": 8848
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6822198033332825,
      "learning_rate": 0.0005978188969371685,
      "loss": 2.7031,
      "step": 8849
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6743674874305725,
      "learning_rate": 0.0005978184045473369,
      "loss": 2.5562,
      "step": 8850
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6315644979476929,
      "learning_rate": 0.0005978179121021353,
      "loss": 2.3498,
      "step": 8851
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6075302958488464,
      "learning_rate": 0.0005978174196015637,
      "loss": 2.611,
      "step": 8852
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6572428345680237,
      "learning_rate": 0.0005978169270456223,
      "loss": 2.4354,
      "step": 8853
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.5857031345367432,
      "learning_rate": 0.000597816434434311,
      "loss": 2.444,
      "step": 8854
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6887085437774658,
      "learning_rate": 0.00059781594176763,
      "loss": 2.743,
      "step": 8855
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6475780606269836,
      "learning_rate": 0.0005978154490455795,
      "loss": 2.3528,
      "step": 8856
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6493271589279175,
      "learning_rate": 0.0005978149562681595,
      "loss": 2.4288,
      "step": 8857
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.5831653475761414,
      "learning_rate": 0.00059781446343537,
      "loss": 2.5497,
      "step": 8858
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6340052485466003,
      "learning_rate": 0.0005978139705472112,
      "loss": 2.373,
      "step": 8859
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.609021008014679,
      "learning_rate": 0.0005978134776036832,
      "loss": 2.556,
      "step": 8860
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6557884216308594,
      "learning_rate": 0.0005978129846047859,
      "loss": 2.4917,
      "step": 8861
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6335002183914185,
      "learning_rate": 0.0005978124915505197,
      "loss": 2.5155,
      "step": 8862
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6874011158943176,
      "learning_rate": 0.0005978119984408844,
      "loss": 2.3096,
      "step": 8863
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.55534428358078,
      "learning_rate": 0.0005978115052758803,
      "loss": 2.2905,
      "step": 8864
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.681315541267395,
      "learning_rate": 0.0005978110120555075,
      "loss": 2.4911,
      "step": 8865
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6153788566589355,
      "learning_rate": 0.0005978105187797659,
      "loss": 2.7191,
      "step": 8866
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6750826239585876,
      "learning_rate": 0.0005978100254486557,
      "loss": 2.6093,
      "step": 8867
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6555083394050598,
      "learning_rate": 0.0005978095320621771,
      "loss": 2.505,
      "step": 8868
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6145590543746948,
      "learning_rate": 0.00059780903862033,
      "loss": 2.626,
      "step": 8869
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.7201048731803894,
      "learning_rate": 0.0005978085451231146,
      "loss": 2.504,
      "step": 8870
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6647611260414124,
      "learning_rate": 0.000597808051570531,
      "loss": 2.459,
      "step": 8871
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.616117000579834,
      "learning_rate": 0.0005978075579625791,
      "loss": 2.4894,
      "step": 8872
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.672499418258667,
      "learning_rate": 0.0005978070642992594,
      "loss": 2.5778,
      "step": 8873
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.5740575194358826,
      "learning_rate": 0.0005978065705805716,
      "loss": 2.5888,
      "step": 8874
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6358161568641663,
      "learning_rate": 0.0005978060768065159,
      "loss": 2.3422,
      "step": 8875
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.7460781931877136,
      "learning_rate": 0.0005978055829770926,
      "loss": 2.5854,
      "step": 8876
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6460410952568054,
      "learning_rate": 0.0005978050890923015,
      "loss": 2.5675,
      "step": 8877
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.7558256387710571,
      "learning_rate": 0.0005978045951521428,
      "loss": 2.4994,
      "step": 8878
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6337019801139832,
      "learning_rate": 0.0005978041011566167,
      "loss": 2.5871,
      "step": 8879
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6081916093826294,
      "learning_rate": 0.0005978036071057231,
      "loss": 2.5706,
      "step": 8880
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.7033549547195435,
      "learning_rate": 0.0005978031129994622,
      "loss": 2.3453,
      "step": 8881
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.7633167505264282,
      "learning_rate": 0.0005978026188378341,
      "loss": 2.4869,
      "step": 8882
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.5744752287864685,
      "learning_rate": 0.0005978021246208387,
      "loss": 2.4578,
      "step": 8883
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.8506788015365601,
      "learning_rate": 0.0005978016303484766,
      "loss": 2.3274,
      "step": 8884
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.708968997001648,
      "learning_rate": 0.0005978011360207473,
      "loss": 2.353,
      "step": 8885
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.7288742065429688,
      "learning_rate": 0.0005978006416376512,
      "loss": 2.3344,
      "step": 8886
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6906796097755432,
      "learning_rate": 0.0005978001471991885,
      "loss": 2.5632,
      "step": 8887
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6430925726890564,
      "learning_rate": 0.0005977996527053589,
      "loss": 2.3891,
      "step": 8888
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6666097640991211,
      "learning_rate": 0.0005977991581561628,
      "loss": 2.6472,
      "step": 8889
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6895706653594971,
      "learning_rate": 0.0005977986635516004,
      "loss": 2.4353,
      "step": 8890
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6971353888511658,
      "learning_rate": 0.0005977981688916713,
      "loss": 2.5226,
      "step": 8891
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6113727688789368,
      "learning_rate": 0.0005977976741763761,
      "loss": 2.3842,
      "step": 8892
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6901161074638367,
      "learning_rate": 0.0005977971794057146,
      "loss": 2.4893,
      "step": 8893
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6308662295341492,
      "learning_rate": 0.0005977966845796871,
      "loss": 2.3361,
      "step": 8894
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6074249148368835,
      "learning_rate": 0.0005977961896982935,
      "loss": 2.6002,
      "step": 8895
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6394092440605164,
      "learning_rate": 0.0005977956947615339,
      "loss": 2.4583,
      "step": 8896
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.689909040927887,
      "learning_rate": 0.0005977951997694086,
      "loss": 2.3054,
      "step": 8897
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6247351765632629,
      "learning_rate": 0.0005977947047219174,
      "loss": 2.5423,
      "step": 8898
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6389184594154358,
      "learning_rate": 0.0005977942096190606,
      "loss": 2.4317,
      "step": 8899
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6741263270378113,
      "learning_rate": 0.0005977937144608383,
      "loss": 2.3351,
      "step": 8900
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6704778075218201,
      "learning_rate": 0.0005977932192472505,
      "loss": 2.6062,
      "step": 8901
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.740999698638916,
      "learning_rate": 0.0005977927239782973,
      "loss": 2.4783,
      "step": 8902
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6844959259033203,
      "learning_rate": 0.0005977922286539789,
      "loss": 2.5319,
      "step": 8903
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.7860072255134583,
      "learning_rate": 0.0005977917332742951,
      "loss": 2.5037,
      "step": 8904
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6055996417999268,
      "learning_rate": 0.0005977912378392463,
      "loss": 2.3158,
      "step": 8905
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.9797891974449158,
      "learning_rate": 0.0005977907423488325,
      "loss": 2.4652,
      "step": 8906
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6307656764984131,
      "learning_rate": 0.0005977902468030538,
      "loss": 2.6778,
      "step": 8907
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.7514820098876953,
      "learning_rate": 0.0005977897512019103,
      "loss": 2.5848,
      "step": 8908
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.0891531705856323,
      "learning_rate": 0.0005977892555454019,
      "loss": 2.5455,
      "step": 8909
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6806963086128235,
      "learning_rate": 0.000597788759833529,
      "loss": 2.6159,
      "step": 8910
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.7630891799926758,
      "learning_rate": 0.0005977882640662915,
      "loss": 2.4161,
      "step": 8911
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.7959657311439514,
      "learning_rate": 0.0005977877682436896,
      "loss": 2.4034,
      "step": 8912
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6858930587768555,
      "learning_rate": 0.0005977872723657233,
      "loss": 2.6302,
      "step": 8913
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.7282989621162415,
      "learning_rate": 0.0005977867764323928,
      "loss": 2.3848,
      "step": 8914
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.8207541108131409,
      "learning_rate": 0.000597786280443698,
      "loss": 2.6409,
      "step": 8915
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6258091330528259,
      "learning_rate": 0.0005977857843996391,
      "loss": 2.5224,
      "step": 8916
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.702744722366333,
      "learning_rate": 0.0005977852883002161,
      "loss": 2.3821,
      "step": 8917
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6677694916725159,
      "learning_rate": 0.0005977847921454295,
      "loss": 2.4012,
      "step": 8918
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6345475316047668,
      "learning_rate": 0.0005977842959352789,
      "loss": 2.4523,
      "step": 8919
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6591619849205017,
      "learning_rate": 0.0005977837996697646,
      "loss": 2.771,
      "step": 8920
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6128720641136169,
      "learning_rate": 0.0005977833033488867,
      "loss": 2.683,
      "step": 8921
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6599496603012085,
      "learning_rate": 0.0005977828069726452,
      "loss": 2.6335,
      "step": 8922
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.7651750445365906,
      "learning_rate": 0.0005977823105410403,
      "loss": 2.463,
      "step": 8923
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6577313542366028,
      "learning_rate": 0.000597781814054072,
      "loss": 2.6672,
      "step": 8924
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6341843605041504,
      "learning_rate": 0.0005977813175117405,
      "loss": 2.6242,
      "step": 8925
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.5850799679756165,
      "learning_rate": 0.0005977808209140458,
      "loss": 2.4791,
      "step": 8926
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6509455442428589,
      "learning_rate": 0.0005977803242609881,
      "loss": 2.8146,
      "step": 8927
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.5953621864318848,
      "learning_rate": 0.0005977798275525674,
      "loss": 2.5646,
      "step": 8928
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6158341765403748,
      "learning_rate": 0.0005977793307887836,
      "loss": 2.6956,
      "step": 8929
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6325731873512268,
      "learning_rate": 0.0005977788339696371,
      "loss": 2.3158,
      "step": 8930
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.5924303531646729,
      "learning_rate": 0.000597778337095128,
      "loss": 2.3056,
      "step": 8931
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.7429385781288147,
      "learning_rate": 0.0005977778401652563,
      "loss": 2.7844,
      "step": 8932
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6176870465278625,
      "learning_rate": 0.000597777343180022,
      "loss": 2.5371,
      "step": 8933
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6601493954658508,
      "learning_rate": 0.0005977768461394253,
      "loss": 2.5582,
      "step": 8934
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6412186026573181,
      "learning_rate": 0.0005977763490434663,
      "loss": 2.3894,
      "step": 8935
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6486330628395081,
      "learning_rate": 0.000597775851892145,
      "loss": 2.7522,
      "step": 8936
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.5975974202156067,
      "learning_rate": 0.0005977753546854616,
      "loss": 2.409,
      "step": 8937
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.5787684917449951,
      "learning_rate": 0.000597774857423416,
      "loss": 2.6128,
      "step": 8938
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.584010124206543,
      "learning_rate": 0.0005977743601060085,
      "loss": 2.4655,
      "step": 8939
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.589307427406311,
      "learning_rate": 0.0005977738627332392,
      "loss": 2.1743,
      "step": 8940
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6186389327049255,
      "learning_rate": 0.000597773365305108,
      "loss": 2.3173,
      "step": 8941
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6796738505363464,
      "learning_rate": 0.0005977728678216152,
      "loss": 2.4291,
      "step": 8942
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6729338765144348,
      "learning_rate": 0.0005977723702827608,
      "loss": 2.4608,
      "step": 8943
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6950299143791199,
      "learning_rate": 0.0005977718726885449,
      "loss": 2.4341,
      "step": 8944
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.7317744493484497,
      "learning_rate": 0.0005977713750389676,
      "loss": 2.4932,
      "step": 8945
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6299111247062683,
      "learning_rate": 0.000597770877334029,
      "loss": 2.3309,
      "step": 8946
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.602156400680542,
      "learning_rate": 0.0005977703795737291,
      "loss": 2.4311,
      "step": 8947
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6092016696929932,
      "learning_rate": 0.0005977698817580681,
      "loss": 2.5726,
      "step": 8948
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6724998354911804,
      "learning_rate": 0.000597769383887046,
      "loss": 2.2222,
      "step": 8949
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6285260319709778,
      "learning_rate": 0.0005977688859606631,
      "loss": 2.3467,
      "step": 8950
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6299049258232117,
      "learning_rate": 0.0005977683879789192,
      "loss": 2.4963,
      "step": 8951
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.7394514679908752,
      "learning_rate": 0.0005977678899418146,
      "loss": 2.7164,
      "step": 8952
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.8927268981933594,
      "learning_rate": 0.0005977673918493492,
      "loss": 2.5461,
      "step": 8953
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6153150796890259,
      "learning_rate": 0.0005977668937015234,
      "loss": 2.5313,
      "step": 8954
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6668916344642639,
      "learning_rate": 0.000597766395498337,
      "loss": 2.3386,
      "step": 8955
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.669075608253479,
      "learning_rate": 0.0005977658972397903,
      "loss": 2.5479,
      "step": 8956
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6653465032577515,
      "learning_rate": 0.0005977653989258832,
      "loss": 2.5358,
      "step": 8957
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6403233408927917,
      "learning_rate": 0.000597764900556616,
      "loss": 2.3119,
      "step": 8958
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.7690974473953247,
      "learning_rate": 0.0005977644021319886,
      "loss": 2.603,
      "step": 8959
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6361300349235535,
      "learning_rate": 0.0005977639036520012,
      "loss": 2.6398,
      "step": 8960
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6770948767662048,
      "learning_rate": 0.0005977634051166538,
      "loss": 2.5712,
      "step": 8961
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6449307203292847,
      "learning_rate": 0.0005977629065259467,
      "loss": 2.2642,
      "step": 8962
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.633651077747345,
      "learning_rate": 0.0005977624078798798,
      "loss": 2.2689,
      "step": 8963
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6245059370994568,
      "learning_rate": 0.0005977619091784532,
      "loss": 2.4873,
      "step": 8964
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6950671076774597,
      "learning_rate": 0.0005977614104216671,
      "loss": 2.34,
      "step": 8965
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.7121109366416931,
      "learning_rate": 0.0005977609116095215,
      "loss": 2.4281,
      "step": 8966
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6541294455528259,
      "learning_rate": 0.0005977604127420165,
      "loss": 2.7368,
      "step": 8967
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.706524133682251,
      "learning_rate": 0.0005977599138191523,
      "loss": 2.4005,
      "step": 8968
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.632689893245697,
      "learning_rate": 0.0005977594148409289,
      "loss": 2.5097,
      "step": 8969
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.5978854894638062,
      "learning_rate": 0.0005977589158073463,
      "loss": 2.4293,
      "step": 8970
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.5822513103485107,
      "learning_rate": 0.0005977584167184048,
      "loss": 2.2198,
      "step": 8971
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.5970476269721985,
      "learning_rate": 0.0005977579175741043,
      "loss": 2.3955,
      "step": 8972
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6918531656265259,
      "learning_rate": 0.000597757418374445,
      "loss": 2.5525,
      "step": 8973
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6858021020889282,
      "learning_rate": 0.0005977569191194271,
      "loss": 2.573,
      "step": 8974
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6336687803268433,
      "learning_rate": 0.0005977564198090504,
      "loss": 2.5064,
      "step": 8975
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.5925370454788208,
      "learning_rate": 0.0005977559204433153,
      "loss": 2.4952,
      "step": 8976
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6193891167640686,
      "learning_rate": 0.0005977554210222217,
      "loss": 2.7278,
      "step": 8977
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.7419455647468567,
      "learning_rate": 0.0005977549215457697,
      "loss": 2.5439,
      "step": 8978
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6662939786911011,
      "learning_rate": 0.0005977544220139595,
      "loss": 2.6584,
      "step": 8979
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.7061561346054077,
      "learning_rate": 0.0005977539224267911,
      "loss": 2.5584,
      "step": 8980
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.8194267749786377,
      "learning_rate": 0.0005977534227842647,
      "loss": 2.5402,
      "step": 8981
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6690295338630676,
      "learning_rate": 0.0005977529230863802,
      "loss": 2.6472,
      "step": 8982
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.9696255326271057,
      "learning_rate": 0.0005977524233331379,
      "loss": 2.1266,
      "step": 8983
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.8717625141143799,
      "learning_rate": 0.0005977519235245378,
      "loss": 2.3958,
      "step": 8984
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.7124236226081848,
      "learning_rate": 0.0005977514236605799,
      "loss": 2.2659,
      "step": 8985
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.7918037176132202,
      "learning_rate": 0.0005977509237412644,
      "loss": 2.3149,
      "step": 8986
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.7326030731201172,
      "learning_rate": 0.0005977504237665916,
      "loss": 2.5208,
      "step": 8987
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.641870379447937,
      "learning_rate": 0.000597749923736561,
      "loss": 2.3645,
      "step": 8988
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.7501170039176941,
      "learning_rate": 0.0005977494236511733,
      "loss": 2.5903,
      "step": 8989
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.7375509142875671,
      "learning_rate": 0.0005977489235104284,
      "loss": 2.3966,
      "step": 8990
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6686384677886963,
      "learning_rate": 0.0005977484233143264,
      "loss": 2.5698,
      "step": 8991
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6863053441047668,
      "learning_rate": 0.0005977479230628672,
      "loss": 2.6053,
      "step": 8992
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.9020069241523743,
      "learning_rate": 0.0005977474227560511,
      "loss": 2.6207,
      "step": 8993
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6708282828330994,
      "learning_rate": 0.0005977469223938781,
      "loss": 2.2475,
      "step": 8994
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6778668761253357,
      "learning_rate": 0.0005977464219763484,
      "loss": 2.6189,
      "step": 8995
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.7599838972091675,
      "learning_rate": 0.000597745921503462,
      "loss": 2.3211,
      "step": 8996
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6495219469070435,
      "learning_rate": 0.0005977454209752189,
      "loss": 2.3459,
      "step": 8997
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6810801029205322,
      "learning_rate": 0.0005977449203916195,
      "loss": 2.3573,
      "step": 8998
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.7329819798469543,
      "learning_rate": 0.0005977444197526635,
      "loss": 2.5627,
      "step": 8999
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.5781172513961792,
      "learning_rate": 0.0005977439190583513,
      "loss": 2.4902,
      "step": 9000
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6600070595741272,
      "learning_rate": 0.0005977434183086828,
      "loss": 2.7586,
      "step": 9001
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6135002374649048,
      "learning_rate": 0.0005977429175036583,
      "loss": 2.4131,
      "step": 9002
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.7837481498718262,
      "learning_rate": 0.0005977424166432777,
      "loss": 2.6812,
      "step": 9003
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6363039612770081,
      "learning_rate": 0.0005977419157275412,
      "loss": 2.7548,
      "step": 9004
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6307020783424377,
      "learning_rate": 0.0005977414147564489,
      "loss": 2.3407,
      "step": 9005
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.632817268371582,
      "learning_rate": 0.0005977409137300008,
      "loss": 2.2118,
      "step": 9006
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6384451985359192,
      "learning_rate": 0.0005977404126481969,
      "loss": 2.5349,
      "step": 9007
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6546534299850464,
      "learning_rate": 0.0005977399115110377,
      "loss": 2.4288,
      "step": 9008
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6993544101715088,
      "learning_rate": 0.0005977394103185228,
      "loss": 2.5801,
      "step": 9009
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6643832325935364,
      "learning_rate": 0.0005977389090706526,
      "loss": 2.5686,
      "step": 9010
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6447634696960449,
      "learning_rate": 0.0005977384077674272,
      "loss": 2.5731,
      "step": 9011
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.7578436732292175,
      "learning_rate": 0.0005977379064088465,
      "loss": 2.3158,
      "step": 9012
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6539149284362793,
      "learning_rate": 0.0005977374049949107,
      "loss": 2.612,
      "step": 9013
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.643598735332489,
      "learning_rate": 0.0005977369035256201,
      "loss": 2.4939,
      "step": 9014
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6550506949424744,
      "learning_rate": 0.0005977364020009744,
      "loss": 2.2418,
      "step": 9015
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6578044295310974,
      "learning_rate": 0.0005977359004209738,
      "loss": 2.8523,
      "step": 9016
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6245808601379395,
      "learning_rate": 0.0005977353987856187,
      "loss": 2.6159,
      "step": 9017
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6578909754753113,
      "learning_rate": 0.0005977348970949088,
      "loss": 2.5717,
      "step": 9018
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6990849375724792,
      "learning_rate": 0.0005977343953488445,
      "loss": 2.643,
      "step": 9019
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6493768692016602,
      "learning_rate": 0.0005977338935474256,
      "loss": 2.4339,
      "step": 9020
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6090483665466309,
      "learning_rate": 0.0005977333916906525,
      "loss": 2.4852,
      "step": 9021
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6585456728935242,
      "learning_rate": 0.0005977328897785251,
      "loss": 2.5794,
      "step": 9022
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6920861601829529,
      "learning_rate": 0.0005977323878110434,
      "loss": 2.6398,
      "step": 9023
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6620957255363464,
      "learning_rate": 0.0005977318857882078,
      "loss": 2.4842,
      "step": 9024
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.5949217081069946,
      "learning_rate": 0.0005977313837100182,
      "loss": 2.4281,
      "step": 9025
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.7355388402938843,
      "learning_rate": 0.0005977308815764746,
      "loss": 2.3945,
      "step": 9026
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.7159362435340881,
      "learning_rate": 0.0005977303793875773,
      "loss": 2.5126,
      "step": 9027
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6228941082954407,
      "learning_rate": 0.0005977298771433264,
      "loss": 2.5237,
      "step": 9028
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6714333295822144,
      "learning_rate": 0.0005977293748437217,
      "loss": 2.4255,
      "step": 9029
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.7530218362808228,
      "learning_rate": 0.0005977288724887636,
      "loss": 2.4731,
      "step": 9030
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6016474962234497,
      "learning_rate": 0.000597728370078452,
      "loss": 2.4103,
      "step": 9031
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.7185631394386292,
      "learning_rate": 0.0005977278676127872,
      "loss": 2.4044,
      "step": 9032
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6204631328582764,
      "learning_rate": 0.000597727365091769,
      "loss": 2.614,
      "step": 9033
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.7097406983375549,
      "learning_rate": 0.0005977268625153978,
      "loss": 2.6096,
      "step": 9034
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.68733149766922,
      "learning_rate": 0.0005977263598836735,
      "loss": 2.6392,
      "step": 9035
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6916921138763428,
      "learning_rate": 0.0005977258571965964,
      "loss": 2.767,
      "step": 9036
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6404944658279419,
      "learning_rate": 0.0005977253544541662,
      "loss": 2.5221,
      "step": 9037
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6911348700523376,
      "learning_rate": 0.0005977248516563833,
      "loss": 2.3678,
      "step": 9038
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.7202990651130676,
      "learning_rate": 0.0005977243488032479,
      "loss": 2.493,
      "step": 9039
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6620497107505798,
      "learning_rate": 0.0005977238458947598,
      "loss": 2.6629,
      "step": 9040
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.744201123714447,
      "learning_rate": 0.0005977233429309192,
      "loss": 2.4412,
      "step": 9041
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.7529190182685852,
      "learning_rate": 0.0005977228399117263,
      "loss": 2.3568,
      "step": 9042
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.7050167322158813,
      "learning_rate": 0.0005977223368371809,
      "loss": 2.5419,
      "step": 9043
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6647268533706665,
      "learning_rate": 0.0005977218337072835,
      "loss": 2.1886,
      "step": 9044
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6920205950737,
      "learning_rate": 0.0005977213305220338,
      "loss": 2.5025,
      "step": 9045
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6453115940093994,
      "learning_rate": 0.0005977208272814323,
      "loss": 2.6308,
      "step": 9046
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6738089919090271,
      "learning_rate": 0.0005977203239854788,
      "loss": 2.5807,
      "step": 9047
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.7340417504310608,
      "learning_rate": 0.0005977198206341735,
      "loss": 2.5153,
      "step": 9048
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6462744474411011,
      "learning_rate": 0.0005977193172275164,
      "loss": 2.6548,
      "step": 9049
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6708645224571228,
      "learning_rate": 0.0005977188137655077,
      "loss": 2.3726,
      "step": 9050
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.7357429265975952,
      "learning_rate": 0.0005977183102481474,
      "loss": 2.4296,
      "step": 9051
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6713259220123291,
      "learning_rate": 0.0005977178066754356,
      "loss": 2.4804,
      "step": 9052
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.8384856581687927,
      "learning_rate": 0.0005977173030473726,
      "loss": 2.3381,
      "step": 9053
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6223863363265991,
      "learning_rate": 0.0005977167993639583,
      "loss": 2.5174,
      "step": 9054
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6888610124588013,
      "learning_rate": 0.0005977162956251928,
      "loss": 2.6214,
      "step": 9055
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.9019646644592285,
      "learning_rate": 0.0005977157918310762,
      "loss": 2.5357,
      "step": 9056
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.608445942401886,
      "learning_rate": 0.0005977152879816086,
      "loss": 2.3764,
      "step": 9057
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.8105952739715576,
      "learning_rate": 0.0005977147840767901,
      "loss": 2.5587,
      "step": 9058
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.7907708287239075,
      "learning_rate": 0.0005977142801166208,
      "loss": 2.7537,
      "step": 9059
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.5865733623504639,
      "learning_rate": 0.0005977137761011009,
      "loss": 2.4655,
      "step": 9060
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.8154267072677612,
      "learning_rate": 0.0005977132720302302,
      "loss": 2.3847,
      "step": 9061
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.7122148871421814,
      "learning_rate": 0.0005977127679040092,
      "loss": 2.5167,
      "step": 9062
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6925473213195801,
      "learning_rate": 0.0005977122637224377,
      "loss": 2.3244,
      "step": 9063
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6216008067131042,
      "learning_rate": 0.0005977117594855157,
      "loss": 2.3669,
      "step": 9064
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.7204646468162537,
      "learning_rate": 0.0005977112551932436,
      "loss": 2.6869,
      "step": 9065
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6090957522392273,
      "learning_rate": 0.0005977107508456215,
      "loss": 2.3643,
      "step": 9066
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.624254047870636,
      "learning_rate": 0.0005977102464426492,
      "loss": 2.4608,
      "step": 9067
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.8628873229026794,
      "learning_rate": 0.0005977097419843269,
      "loss": 2.6602,
      "step": 9068
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6695460081100464,
      "learning_rate": 0.0005977092374706548,
      "loss": 2.7279,
      "step": 9069
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.8926200866699219,
      "learning_rate": 0.000597708732901633,
      "loss": 2.5968,
      "step": 9070
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.643154501914978,
      "learning_rate": 0.0005977082282772613,
      "loss": 2.473,
      "step": 9071
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6624605059623718,
      "learning_rate": 0.0005977077235975402,
      "loss": 2.5187,
      "step": 9072
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.7428131699562073,
      "learning_rate": 0.0005977072188624697,
      "loss": 2.5534,
      "step": 9073
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.7155025005340576,
      "learning_rate": 0.0005977067140720497,
      "loss": 2.6934,
      "step": 9074
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6285994052886963,
      "learning_rate": 0.0005977062092262804,
      "loss": 2.649,
      "step": 9075
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.7973483204841614,
      "learning_rate": 0.0005977057043251619,
      "loss": 2.6436,
      "step": 9076
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6960684657096863,
      "learning_rate": 0.0005977051993686943,
      "loss": 2.3357,
      "step": 9077
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.8100269436836243,
      "learning_rate": 0.0005977046943568777,
      "loss": 2.3023,
      "step": 9078
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6927614808082581,
      "learning_rate": 0.0005977041892897122,
      "loss": 2.4174,
      "step": 9079
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6934711337089539,
      "learning_rate": 0.0005977036841671978,
      "loss": 2.6396,
      "step": 9080
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6070442795753479,
      "learning_rate": 0.0005977031789893346,
      "loss": 2.7526,
      "step": 9081
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.7466270327568054,
      "learning_rate": 0.0005977026737561228,
      "loss": 2.1413,
      "step": 9082
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6765740513801575,
      "learning_rate": 0.0005977021684675626,
      "loss": 2.3688,
      "step": 9083
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.723379909992218,
      "learning_rate": 0.0005977016631236538,
      "loss": 2.217,
      "step": 9084
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6958324909210205,
      "learning_rate": 0.0005977011577243967,
      "loss": 2.5362,
      "step": 9085
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.654574990272522,
      "learning_rate": 0.0005977006522697913,
      "loss": 2.3653,
      "step": 9086
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6074076294898987,
      "learning_rate": 0.0005977001467598378,
      "loss": 2.5659,
      "step": 9087
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6513310074806213,
      "learning_rate": 0.0005976996411945362,
      "loss": 2.517,
      "step": 9088
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6405103802680969,
      "learning_rate": 0.0005976991355738866,
      "loss": 2.6714,
      "step": 9089
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6518906950950623,
      "learning_rate": 0.0005976986298978891,
      "loss": 2.5751,
      "step": 9090
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6504568457603455,
      "learning_rate": 0.0005976981241665437,
      "loss": 2.6172,
      "step": 9091
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6478431820869446,
      "learning_rate": 0.0005976976183798508,
      "loss": 2.4666,
      "step": 9092
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6663208603858948,
      "learning_rate": 0.0005976971125378101,
      "loss": 2.8299,
      "step": 9093
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6460630297660828,
      "learning_rate": 0.0005976966066404221,
      "loss": 2.5236,
      "step": 9094
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6168675422668457,
      "learning_rate": 0.0005976961006876866,
      "loss": 2.3086,
      "step": 9095
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.7575297951698303,
      "learning_rate": 0.0005976955946796037,
      "loss": 2.3089,
      "step": 9096
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6402947306632996,
      "learning_rate": 0.0005976950886161736,
      "loss": 2.3976,
      "step": 9097
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.7353620529174805,
      "learning_rate": 0.0005976945824973965,
      "loss": 2.3315,
      "step": 9098
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6554626226425171,
      "learning_rate": 0.0005976940763232722,
      "loss": 2.5441,
      "step": 9099
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.7709004878997803,
      "learning_rate": 0.000597693570093801,
      "loss": 2.4827,
      "step": 9100
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6313326358795166,
      "learning_rate": 0.0005976930638089829,
      "loss": 2.6558,
      "step": 9101
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6772848963737488,
      "learning_rate": 0.0005976925574688181,
      "loss": 2.2977,
      "step": 9102
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6054846048355103,
      "learning_rate": 0.0005976920510733066,
      "loss": 2.5374,
      "step": 9103
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6563945412635803,
      "learning_rate": 0.0005976915446224485,
      "loss": 2.5043,
      "step": 9104
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6390348076820374,
      "learning_rate": 0.000597691038116244,
      "loss": 2.8181,
      "step": 9105
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6807928085327148,
      "learning_rate": 0.0005976905315546931,
      "loss": 2.5853,
      "step": 9106
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6725947260856628,
      "learning_rate": 0.0005976900249377959,
      "loss": 2.435,
      "step": 9107
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.5833192467689514,
      "learning_rate": 0.0005976895182655524,
      "loss": 2.5503,
      "step": 9108
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.586933434009552,
      "learning_rate": 0.000597689011537963,
      "loss": 2.4987,
      "step": 9109
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6207230687141418,
      "learning_rate": 0.0005976885047550275,
      "loss": 2.5437,
      "step": 9110
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6112599968910217,
      "learning_rate": 0.000597687997916746,
      "loss": 2.5077,
      "step": 9111
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.5820552110671997,
      "learning_rate": 0.0005976874910231188,
      "loss": 2.3095,
      "step": 9112
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.72879958152771,
      "learning_rate": 0.0005976869840741459,
      "loss": 2.416,
      "step": 9113
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6646095514297485,
      "learning_rate": 0.0005976864770698273,
      "loss": 2.5252,
      "step": 9114
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6864641308784485,
      "learning_rate": 0.000597685970010163,
      "loss": 2.355,
      "step": 9115
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6799298524856567,
      "learning_rate": 0.0005976854628951536,
      "loss": 2.7548,
      "step": 9116
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6445333361625671,
      "learning_rate": 0.0005976849557247986,
      "loss": 2.6471,
      "step": 9117
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.7214179635047913,
      "learning_rate": 0.0005976844484990984,
      "loss": 2.6036,
      "step": 9118
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6824414134025574,
      "learning_rate": 0.0005976839412180532,
      "loss": 2.4587,
      "step": 9119
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6337586045265198,
      "learning_rate": 0.0005976834338816628,
      "loss": 2.5363,
      "step": 9120
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.5938040018081665,
      "learning_rate": 0.0005976829264899274,
      "loss": 2.6989,
      "step": 9121
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.7537724375724792,
      "learning_rate": 0.0005976824190428472,
      "loss": 2.1958,
      "step": 9122
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6548065543174744,
      "learning_rate": 0.0005976819115404221,
      "loss": 2.5689,
      "step": 9123
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6470126509666443,
      "learning_rate": 0.0005976814039826524,
      "loss": 2.4106,
      "step": 9124
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6664608120918274,
      "learning_rate": 0.0005976808963695381,
      "loss": 2.5548,
      "step": 9125
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.7723535299301147,
      "learning_rate": 0.0005976803887010793,
      "loss": 2.5598,
      "step": 9126
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6482446193695068,
      "learning_rate": 0.0005976798809772761,
      "loss": 2.6866,
      "step": 9127
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6907306909561157,
      "learning_rate": 0.0005976793731981286,
      "loss": 2.523,
      "step": 9128
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.673203706741333,
      "learning_rate": 0.0005976788653636369,
      "loss": 2.7696,
      "step": 9129
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.640871524810791,
      "learning_rate": 0.000597678357473801,
      "loss": 2.4908,
      "step": 9130
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6466402411460876,
      "learning_rate": 0.0005976778495286212,
      "loss": 2.5765,
      "step": 9131
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6364684700965881,
      "learning_rate": 0.0005976773415280974,
      "loss": 2.4681,
      "step": 9132
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6045383214950562,
      "learning_rate": 0.0005976768334722298,
      "loss": 2.4226,
      "step": 9133
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6767545938491821,
      "learning_rate": 0.0005976763253610184,
      "loss": 2.4895,
      "step": 9134
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6345009803771973,
      "learning_rate": 0.0005976758171944633,
      "loss": 2.5299,
      "step": 9135
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6524790525436401,
      "learning_rate": 0.0005976753089725647,
      "loss": 2.4743,
      "step": 9136
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6884654760360718,
      "learning_rate": 0.0005976748006953227,
      "loss": 2.3109,
      "step": 9137
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.605446994304657,
      "learning_rate": 0.0005976742923627374,
      "loss": 2.41,
      "step": 9138
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.696772038936615,
      "learning_rate": 0.0005976737839748087,
      "loss": 2.5615,
      "step": 9139
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6105514168739319,
      "learning_rate": 0.0005976732755315368,
      "loss": 2.2913,
      "step": 9140
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6868053078651428,
      "learning_rate": 0.0005976727670329218,
      "loss": 2.3702,
      "step": 9141
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.652385413646698,
      "learning_rate": 0.000597672258478964,
      "loss": 2.3493,
      "step": 9142
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.629520833492279,
      "learning_rate": 0.0005976717498696631,
      "loss": 2.2133,
      "step": 9143
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6689314246177673,
      "learning_rate": 0.0005976712412050195,
      "loss": 2.4265,
      "step": 9144
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6617286205291748,
      "learning_rate": 0.0005976707324850333,
      "loss": 2.6483,
      "step": 9145
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6448078751564026,
      "learning_rate": 0.0005976702237097043,
      "loss": 2.4906,
      "step": 9146
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6349143385887146,
      "learning_rate": 0.0005976697148790329,
      "loss": 2.7243,
      "step": 9147
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6498738527297974,
      "learning_rate": 0.0005976692059930191,
      "loss": 2.5936,
      "step": 9148
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6506467461585999,
      "learning_rate": 0.0005976686970516629,
      "loss": 2.4315,
      "step": 9149
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6013219952583313,
      "learning_rate": 0.0005976681880549645,
      "loss": 2.5697,
      "step": 9150
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6556915044784546,
      "learning_rate": 0.000597667679002924,
      "loss": 2.6365,
      "step": 9151
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.617097795009613,
      "learning_rate": 0.0005976671698955414,
      "loss": 2.1777,
      "step": 9152
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6304683685302734,
      "learning_rate": 0.0005976666607328168,
      "loss": 2.3775,
      "step": 9153
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6301159858703613,
      "learning_rate": 0.0005976661515147505,
      "loss": 2.4892,
      "step": 9154
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6274382472038269,
      "learning_rate": 0.0005976656422413423,
      "loss": 2.5904,
      "step": 9155
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6662918329238892,
      "learning_rate": 0.0005976651329125925,
      "loss": 2.7369,
      "step": 9156
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.7374709248542786,
      "learning_rate": 0.0005976646235285012,
      "loss": 2.4496,
      "step": 9157
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.79621422290802,
      "learning_rate": 0.0005976641140890684,
      "loss": 2.2514,
      "step": 9158
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6915442943572998,
      "learning_rate": 0.0005976636045942941,
      "loss": 2.5478,
      "step": 9159
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.8580842018127441,
      "learning_rate": 0.0005976630950441786,
      "loss": 2.2005,
      "step": 9160
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.771332323551178,
      "learning_rate": 0.000597662585438722,
      "loss": 2.7558,
      "step": 9161
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6870948076248169,
      "learning_rate": 0.0005976620757779242,
      "loss": 2.4346,
      "step": 9162
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.28816819190979,
      "learning_rate": 0.0005976615660617854,
      "loss": 2.4825,
      "step": 9163
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6706182360649109,
      "learning_rate": 0.0005976610562903057,
      "loss": 2.4617,
      "step": 9164
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6299802660942078,
      "learning_rate": 0.0005976605464634852,
      "loss": 2.8098,
      "step": 9165
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.7138885855674744,
      "learning_rate": 0.0005976600365813241,
      "loss": 2.4908,
      "step": 9166
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.850715696811676,
      "learning_rate": 0.0005976595266438223,
      "loss": 2.5756,
      "step": 9167
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.7248154878616333,
      "learning_rate": 0.0005976590166509798,
      "loss": 2.4353,
      "step": 9168
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.691409170627594,
      "learning_rate": 0.0005976585066027971,
      "loss": 2.6109,
      "step": 9169
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.8066449165344238,
      "learning_rate": 0.000597657996499274,
      "loss": 2.5142,
      "step": 9170
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.7486743927001953,
      "learning_rate": 0.0005976574863404107,
      "loss": 2.3023,
      "step": 9171
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6867202520370483,
      "learning_rate": 0.0005976569761262071,
      "loss": 2.6372,
      "step": 9172
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.7701345682144165,
      "learning_rate": 0.0005976564658566636,
      "loss": 2.4945,
      "step": 9173
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.8879254460334778,
      "learning_rate": 0.0005976559555317801,
      "loss": 2.7675,
      "step": 9174
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.637226402759552,
      "learning_rate": 0.0005976554451515567,
      "loss": 2.6238,
      "step": 9175
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6683356165885925,
      "learning_rate": 0.0005976549347159936,
      "loss": 2.5055,
      "step": 9176
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.7106610536575317,
      "learning_rate": 0.0005976544242250908,
      "loss": 2.5167,
      "step": 9177
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.7387312054634094,
      "learning_rate": 0.0005976539136788485,
      "loss": 2.5408,
      "step": 9178
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6808874607086182,
      "learning_rate": 0.0005976534030772665,
      "loss": 2.4406,
      "step": 9179
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6355615854263306,
      "learning_rate": 0.0005976528924203454,
      "loss": 2.6056,
      "step": 9180
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.701094925403595,
      "learning_rate": 0.0005976523817080848,
      "loss": 2.4688,
      "step": 9181
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6910794973373413,
      "learning_rate": 0.000597651870940485,
      "loss": 2.4761,
      "step": 9182
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6399173736572266,
      "learning_rate": 0.0005976513601175462,
      "loss": 2.2804,
      "step": 9183
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.7283966541290283,
      "learning_rate": 0.0005976508492392683,
      "loss": 2.495,
      "step": 9184
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.8626879453659058,
      "learning_rate": 0.0005976503383056516,
      "loss": 2.3118,
      "step": 9185
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.596400797367096,
      "learning_rate": 0.0005976498273166961,
      "loss": 2.1555,
      "step": 9186
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6159172654151917,
      "learning_rate": 0.0005976493162724017,
      "loss": 2.4973,
      "step": 9187
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.757919192314148,
      "learning_rate": 0.0005976488051727688,
      "loss": 2.5961,
      "step": 9188
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6641363501548767,
      "learning_rate": 0.0005976482940177974,
      "loss": 2.5549,
      "step": 9189
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6422681212425232,
      "learning_rate": 0.0005976477828074875,
      "loss": 2.6083,
      "step": 9190
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6572309136390686,
      "learning_rate": 0.0005976472715418392,
      "loss": 2.5277,
      "step": 9191
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.9003673195838928,
      "learning_rate": 0.0005976467602208528,
      "loss": 2.4615,
      "step": 9192
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6419776082038879,
      "learning_rate": 0.0005976462488445283,
      "loss": 2.6099,
      "step": 9193
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.7417586445808411,
      "learning_rate": 0.0005976457374128655,
      "loss": 2.5235,
      "step": 9194
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.739295244216919,
      "learning_rate": 0.0005976452259258649,
      "loss": 2.2693,
      "step": 9195
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.639298141002655,
      "learning_rate": 0.0005976447143835264,
      "loss": 2.4478,
      "step": 9196
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.9132916331291199,
      "learning_rate": 0.0005976442027858501,
      "loss": 2.3366,
      "step": 9197
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.7292041182518005,
      "learning_rate": 0.0005976436911328361,
      "loss": 2.3322,
      "step": 9198
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.7441701292991638,
      "learning_rate": 0.0005976431794244846,
      "loss": 2.622,
      "step": 9199
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6222226619720459,
      "learning_rate": 0.0005976426676607956,
      "loss": 2.2417,
      "step": 9200
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.8206707239151001,
      "learning_rate": 0.0005976421558417692,
      "loss": 2.4776,
      "step": 9201
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.7691220641136169,
      "learning_rate": 0.0005976416439674054,
      "loss": 2.7079,
      "step": 9202
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6438469290733337,
      "learning_rate": 0.0005976411320377045,
      "loss": 2.3124,
      "step": 9203
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.74967360496521,
      "learning_rate": 0.0005976406200526666,
      "loss": 2.6949,
      "step": 9204
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.7308462262153625,
      "learning_rate": 0.0005976401080122916,
      "loss": 2.5795,
      "step": 9205
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6009274125099182,
      "learning_rate": 0.0005976395959165796,
      "loss": 2.4741,
      "step": 9206
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6571493744850159,
      "learning_rate": 0.0005976390837655308,
      "loss": 2.4576,
      "step": 9207
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6815400719642639,
      "learning_rate": 0.0005976385715591454,
      "loss": 2.6174,
      "step": 9208
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6589906811714172,
      "learning_rate": 0.0005976380592974233,
      "loss": 2.4419,
      "step": 9209
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6035147905349731,
      "learning_rate": 0.0005976375469803647,
      "loss": 2.4489,
      "step": 9210
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6188952922821045,
      "learning_rate": 0.0005976370346079696,
      "loss": 2.51,
      "step": 9211
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6877496838569641,
      "learning_rate": 0.0005976365221802382,
      "loss": 2.3725,
      "step": 9212
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.701566219329834,
      "learning_rate": 0.0005976360096971705,
      "loss": 2.6761,
      "step": 9213
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6524765491485596,
      "learning_rate": 0.0005976354971587667,
      "loss": 2.439,
      "step": 9214
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6933678984642029,
      "learning_rate": 0.0005976349845650268,
      "loss": 2.6567,
      "step": 9215
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6498165130615234,
      "learning_rate": 0.000597634471915951,
      "loss": 2.6316,
      "step": 9216
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.7285436987876892,
      "learning_rate": 0.0005976339592115393,
      "loss": 2.4667,
      "step": 9217
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6544997692108154,
      "learning_rate": 0.0005976334464517919,
      "loss": 2.4604,
      "step": 9218
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6240621209144592,
      "learning_rate": 0.0005976329336367086,
      "loss": 2.5782,
      "step": 9219
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6373094916343689,
      "learning_rate": 0.0005976324207662899,
      "loss": 2.6632,
      "step": 9220
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6150234937667847,
      "learning_rate": 0.0005976319078405357,
      "loss": 2.4588,
      "step": 9221
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6312364935874939,
      "learning_rate": 0.0005976313948594461,
      "loss": 2.4584,
      "step": 9222
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6067540049552917,
      "learning_rate": 0.0005976308818230211,
      "loss": 2.5536,
      "step": 9223
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.7679184675216675,
      "learning_rate": 0.0005976303687312611,
      "loss": 2.4142,
      "step": 9224
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6925670504570007,
      "learning_rate": 0.0005976298555841658,
      "loss": 2.3841,
      "step": 9225
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6136474609375,
      "learning_rate": 0.0005976293423817356,
      "loss": 2.6298,
      "step": 9226
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.7236453294754028,
      "learning_rate": 0.0005976288291239705,
      "loss": 2.6637,
      "step": 9227
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.7159112095832825,
      "learning_rate": 0.0005976283158108706,
      "loss": 2.4248,
      "step": 9228
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6517639756202698,
      "learning_rate": 0.000597627802442436,
      "loss": 2.7083,
      "step": 9229
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6789841055870056,
      "learning_rate": 0.0005976272890186666,
      "loss": 2.4793,
      "step": 9230
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.7549801468849182,
      "learning_rate": 0.0005976267755395628,
      "loss": 2.5779,
      "step": 9231
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6545198559761047,
      "learning_rate": 0.0005976262620051245,
      "loss": 2.6321,
      "step": 9232
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.660414457321167,
      "learning_rate": 0.0005976257484153519,
      "loss": 2.3811,
      "step": 9233
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.7328870892524719,
      "learning_rate": 0.0005976252347702451,
      "loss": 2.5566,
      "step": 9234
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.7113665342330933,
      "learning_rate": 0.0005976247210698041,
      "loss": 2.4643,
      "step": 9235
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.7286781668663025,
      "learning_rate": 0.0005976242073140291,
      "loss": 2.561,
      "step": 9236
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.8169940114021301,
      "learning_rate": 0.00059762369350292,
      "loss": 2.3776,
      "step": 9237
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.7985203266143799,
      "learning_rate": 0.0005976231796364772,
      "loss": 2.5662,
      "step": 9238
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6443610191345215,
      "learning_rate": 0.0005976226657147005,
      "loss": 2.4722,
      "step": 9239
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.7098340392112732,
      "learning_rate": 0.0005976221517375902,
      "loss": 2.4931,
      "step": 9240
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.7879757881164551,
      "learning_rate": 0.0005976216377051464,
      "loss": 2.3558,
      "step": 9241
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6581434607505798,
      "learning_rate": 0.000597621123617369,
      "loss": 2.4425,
      "step": 9242
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.7204377055168152,
      "learning_rate": 0.0005976206094742582,
      "loss": 2.4281,
      "step": 9243
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.7987725138664246,
      "learning_rate": 0.0005976200952758141,
      "loss": 2.4748,
      "step": 9244
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.7033754587173462,
      "learning_rate": 0.000597619581022037,
      "loss": 2.4127,
      "step": 9245
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.7788264155387878,
      "learning_rate": 0.0005976190667129266,
      "loss": 2.2283,
      "step": 9246
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6521145701408386,
      "learning_rate": 0.0005976185523484833,
      "loss": 2.6391,
      "step": 9247
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6267082095146179,
      "learning_rate": 0.000597618037928707,
      "loss": 2.4061,
      "step": 9248
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.8033210039138794,
      "learning_rate": 0.000597617523453598,
      "loss": 2.6493,
      "step": 9249
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.5625874400138855,
      "learning_rate": 0.0005976170089231562,
      "loss": 2.3567,
      "step": 9250
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6857611536979675,
      "learning_rate": 0.0005976164943373818,
      "loss": 2.6309,
      "step": 9251
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6612685322761536,
      "learning_rate": 0.000597615979696275,
      "loss": 2.6081,
      "step": 9252
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.5824531316757202,
      "learning_rate": 0.0005976154649998356,
      "loss": 2.3907,
      "step": 9253
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6948447227478027,
      "learning_rate": 0.0005976149502480638,
      "loss": 2.5732,
      "step": 9254
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.5896996259689331,
      "learning_rate": 0.00059761443544096,
      "loss": 2.4278,
      "step": 9255
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6171615719795227,
      "learning_rate": 0.000597613920578524,
      "loss": 2.3749,
      "step": 9256
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.5960289835929871,
      "learning_rate": 0.0005976134056607558,
      "loss": 2.583,
      "step": 9257
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.5945549607276917,
      "learning_rate": 0.0005976128906876558,
      "loss": 2.4788,
      "step": 9258
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.5984545350074768,
      "learning_rate": 0.0005976123756592239,
      "loss": 2.4864,
      "step": 9259
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6163726449012756,
      "learning_rate": 0.0005976118605754602,
      "loss": 2.6234,
      "step": 9260
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6027385592460632,
      "learning_rate": 0.0005976113454363649,
      "loss": 2.3969,
      "step": 9261
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.5960747003555298,
      "learning_rate": 0.000597610830241938,
      "loss": 2.6737,
      "step": 9262
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6561993956565857,
      "learning_rate": 0.0005976103149921796,
      "loss": 2.6744,
      "step": 9263
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6550663113594055,
      "learning_rate": 0.0005976097996870899,
      "loss": 2.6314,
      "step": 9264
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6684076189994812,
      "learning_rate": 0.0005976092843266689,
      "loss": 2.6749,
      "step": 9265
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6099563837051392,
      "learning_rate": 0.0005976087689109166,
      "loss": 2.6545,
      "step": 9266
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6450193524360657,
      "learning_rate": 0.0005976082534398334,
      "loss": 2.5289,
      "step": 9267
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.7040346264839172,
      "learning_rate": 0.0005976077379134192,
      "loss": 2.6013,
      "step": 9268
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.5954053401947021,
      "learning_rate": 0.000597607222331674,
      "loss": 2.6097,
      "step": 9269
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6867691278457642,
      "learning_rate": 0.000597606706694598,
      "loss": 2.4121,
      "step": 9270
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.7331819534301758,
      "learning_rate": 0.0005976061910021914,
      "loss": 2.5839,
      "step": 9271
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6042304039001465,
      "learning_rate": 0.0005976056752544541,
      "loss": 2.5002,
      "step": 9272
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6467564702033997,
      "learning_rate": 0.0005976051594513862,
      "loss": 2.5091,
      "step": 9273
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6132211089134216,
      "learning_rate": 0.000597604643592988,
      "loss": 2.4276,
      "step": 9274
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.7143293619155884,
      "learning_rate": 0.0005976041276792595,
      "loss": 2.3672,
      "step": 9275
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6486926078796387,
      "learning_rate": 0.0005976036117102008,
      "loss": 2.4066,
      "step": 9276
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6310781836509705,
      "learning_rate": 0.0005976030956858117,
      "loss": 2.6601,
      "step": 9277
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6322148442268372,
      "learning_rate": 0.0005976025796060929,
      "loss": 2.5645,
      "step": 9278
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.7015872001647949,
      "learning_rate": 0.000597602063471044,
      "loss": 2.7296,
      "step": 9279
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6589545011520386,
      "learning_rate": 0.0005976015472806652,
      "loss": 2.8129,
      "step": 9280
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6614254117012024,
      "learning_rate": 0.0005976010310349568,
      "loss": 2.5387,
      "step": 9281
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.7364760041236877,
      "learning_rate": 0.0005976005147339188,
      "loss": 2.3934,
      "step": 9282
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6304479837417603,
      "learning_rate": 0.0005975999983775511,
      "loss": 2.6133,
      "step": 9283
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6566867232322693,
      "learning_rate": 0.0005975994819658539,
      "loss": 2.4609,
      "step": 9284
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6713210940361023,
      "learning_rate": 0.0005975989654988274,
      "loss": 2.3663,
      "step": 9285
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.7097674608230591,
      "learning_rate": 0.0005975984489764719,
      "loss": 2.3401,
      "step": 9286
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6150882840156555,
      "learning_rate": 0.0005975979323987869,
      "loss": 2.244,
      "step": 9287
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6368316411972046,
      "learning_rate": 0.0005975974157657729,
      "loss": 2.4942,
      "step": 9288
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.7161740064620972,
      "learning_rate": 0.0005975968990774298,
      "loss": 2.7014,
      "step": 9289
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6732980608940125,
      "learning_rate": 0.0005975963823337581,
      "loss": 2.332,
      "step": 9290
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.65184086561203,
      "learning_rate": 0.0005975958655347575,
      "loss": 2.256,
      "step": 9291
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.5978068113327026,
      "learning_rate": 0.0005975953486804282,
      "loss": 2.322,
      "step": 9292
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.7641469836235046,
      "learning_rate": 0.0005975948317707703,
      "loss": 2.3425,
      "step": 9293
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6204509139060974,
      "learning_rate": 0.0005975943148057839,
      "loss": 2.6749,
      "step": 9294
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.7209970951080322,
      "learning_rate": 0.0005975937977854691,
      "loss": 2.517,
      "step": 9295
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.7073755264282227,
      "learning_rate": 0.0005975932807098259,
      "loss": 2.2851,
      "step": 9296
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6442673206329346,
      "learning_rate": 0.0005975927635788546,
      "loss": 2.6932,
      "step": 9297
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6934264898300171,
      "learning_rate": 0.0005975922463925552,
      "loss": 2.4383,
      "step": 9298
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6407778263092041,
      "learning_rate": 0.0005975917291509278,
      "loss": 2.2844,
      "step": 9299
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.7056527733802795,
      "learning_rate": 0.0005975912118539724,
      "loss": 2.4692,
      "step": 9300
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6464372873306274,
      "learning_rate": 0.0005975906945016893,
      "loss": 2.5483,
      "step": 9301
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6516836881637573,
      "learning_rate": 0.0005975901770940783,
      "loss": 2.6121,
      "step": 9302
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6164288520812988,
      "learning_rate": 0.0005975896596311397,
      "loss": 2.5103,
      "step": 9303
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6264887452125549,
      "learning_rate": 0.0005975891421128737,
      "loss": 2.4928,
      "step": 9304
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6800885796546936,
      "learning_rate": 0.0005975886245392801,
      "loss": 2.4774,
      "step": 9305
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.72189861536026,
      "learning_rate": 0.0005975881069103593,
      "loss": 2.6141,
      "step": 9306
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.7255359292030334,
      "learning_rate": 0.0005975875892261112,
      "loss": 2.3284,
      "step": 9307
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.7098193168640137,
      "learning_rate": 0.0005975870714865359,
      "loss": 2.444,
      "step": 9308
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6103895306587219,
      "learning_rate": 0.0005975865536916336,
      "loss": 2.6984,
      "step": 9309
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6181144118309021,
      "learning_rate": 0.0005975860358414044,
      "loss": 2.3106,
      "step": 9310
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.7277249097824097,
      "learning_rate": 0.0005975855179358482,
      "loss": 2.4647,
      "step": 9311
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6556454300880432,
      "learning_rate": 0.0005975849999749653,
      "loss": 2.5713,
      "step": 9312
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.7655468583106995,
      "learning_rate": 0.0005975844819587556,
      "loss": 2.4644,
      "step": 9313
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.598681628704071,
      "learning_rate": 0.0005975839638872197,
      "loss": 2.6234,
      "step": 9314
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6907820105552673,
      "learning_rate": 0.000597583445760357,
      "loss": 2.6596,
      "step": 9315
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.7956867218017578,
      "learning_rate": 0.000597582927578168,
      "loss": 2.6029,
      "step": 9316
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.8574706315994263,
      "learning_rate": 0.0005975824093406527,
      "loss": 2.6244,
      "step": 9317
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.7640001773834229,
      "learning_rate": 0.0005975818910478112,
      "loss": 2.4935,
      "step": 9318
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.7375756502151489,
      "learning_rate": 0.0005975813726996436,
      "loss": 2.5837,
      "step": 9319
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.7069522738456726,
      "learning_rate": 0.0005975808542961501,
      "loss": 2.5726,
      "step": 9320
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.7057202458381653,
      "learning_rate": 0.0005975803358373307,
      "loss": 2.7704,
      "step": 9321
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.889983594417572,
      "learning_rate": 0.0005975798173231854,
      "loss": 2.4401,
      "step": 9322
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6883452534675598,
      "learning_rate": 0.0005975792987537145,
      "loss": 2.3209,
      "step": 9323
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6428436040878296,
      "learning_rate": 0.0005975787801289179,
      "loss": 2.5192,
      "step": 9324
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.8337064981460571,
      "learning_rate": 0.0005975782614487959,
      "loss": 2.3526,
      "step": 9325
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6651191115379333,
      "learning_rate": 0.0005975777427133484,
      "loss": 2.4958,
      "step": 9326
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6832571029663086,
      "learning_rate": 0.0005975772239225754,
      "loss": 2.4055,
      "step": 9327
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.5969125628471375,
      "learning_rate": 0.0005975767050764775,
      "loss": 2.3594,
      "step": 9328
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.642754852771759,
      "learning_rate": 0.0005975761861750543,
      "loss": 2.6426,
      "step": 9329
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6526721715927124,
      "learning_rate": 0.0005975756672183061,
      "loss": 2.5091,
      "step": 9330
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6814718842506409,
      "learning_rate": 0.0005975751482062329,
      "loss": 2.621,
      "step": 9331
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6228416562080383,
      "learning_rate": 0.000597574629138835,
      "loss": 2.2451,
      "step": 9332
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6254288554191589,
      "learning_rate": 0.0005975741100161122,
      "loss": 2.4561,
      "step": 9333
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6544535756111145,
      "learning_rate": 0.0005975735908380649,
      "loss": 2.5383,
      "step": 9334
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.611839234828949,
      "learning_rate": 0.000597573071604693,
      "loss": 2.5116,
      "step": 9335
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.624771773815155,
      "learning_rate": 0.0005975725523159967,
      "loss": 2.4146,
      "step": 9336
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.5953608751296997,
      "learning_rate": 0.0005975720329719759,
      "loss": 2.395,
      "step": 9337
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6720494627952576,
      "learning_rate": 0.000597571513572631,
      "loss": 2.482,
      "step": 9338
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6798122525215149,
      "learning_rate": 0.000597570994117962,
      "loss": 2.5519,
      "step": 9339
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6270776391029358,
      "learning_rate": 0.0005975704746079688,
      "loss": 2.4532,
      "step": 9340
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6325424313545227,
      "learning_rate": 0.0005975699550426516,
      "loss": 2.4827,
      "step": 9341
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.7093266844749451,
      "learning_rate": 0.0005975694354220106,
      "loss": 2.3766,
      "step": 9342
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6375361680984497,
      "learning_rate": 0.0005975689157460458,
      "loss": 2.5709,
      "step": 9343
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.655449628829956,
      "learning_rate": 0.0005975683960147573,
      "loss": 2.6196,
      "step": 9344
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.7520560622215271,
      "learning_rate": 0.0005975678762281452,
      "loss": 2.5173,
      "step": 9345
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6345621347427368,
      "learning_rate": 0.0005975673563862096,
      "loss": 2.3858,
      "step": 9346
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.5757044553756714,
      "learning_rate": 0.0005975668364889508,
      "loss": 2.437,
      "step": 9347
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6155162453651428,
      "learning_rate": 0.0005975663165363684,
      "loss": 2.3809,
      "step": 9348
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.7946271896362305,
      "learning_rate": 0.0005975657965284631,
      "loss": 2.4409,
      "step": 9349
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.7329941987991333,
      "learning_rate": 0.0005975652764652346,
      "loss": 2.463,
      "step": 9350
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.9860031604766846,
      "learning_rate": 0.000597564756346683,
      "loss": 2.4596,
      "step": 9351
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6060137152671814,
      "learning_rate": 0.0005975642361728086,
      "loss": 2.547,
      "step": 9352
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.7208117246627808,
      "learning_rate": 0.0005975637159436113,
      "loss": 2.4502,
      "step": 9353
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.7121568918228149,
      "learning_rate": 0.0005975631956590914,
      "loss": 2.4014,
      "step": 9354
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.7947120070457458,
      "learning_rate": 0.0005975626753192488,
      "loss": 2.7845,
      "step": 9355
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.7773042321205139,
      "learning_rate": 0.0005975621549240838,
      "loss": 2.347,
      "step": 9356
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6488456726074219,
      "learning_rate": 0.0005975616344735963,
      "loss": 2.5595,
      "step": 9357
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6087527275085449,
      "learning_rate": 0.0005975611139677864,
      "loss": 2.4549,
      "step": 9358
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6363966464996338,
      "learning_rate": 0.0005975605934066543,
      "loss": 2.7897,
      "step": 9359
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6303510665893555,
      "learning_rate": 0.0005975600727902002,
      "loss": 2.7199,
      "step": 9360
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6404280066490173,
      "learning_rate": 0.000597559552118424,
      "loss": 2.2889,
      "step": 9361
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6082245707511902,
      "learning_rate": 0.0005975590313913257,
      "loss": 2.331,
      "step": 9362
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6272193789482117,
      "learning_rate": 0.0005975585106089057,
      "loss": 2.3003,
      "step": 9363
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6872128844261169,
      "learning_rate": 0.000597557989771164,
      "loss": 2.596,
      "step": 9364
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6651809811592102,
      "learning_rate": 0.0005975574688781007,
      "loss": 2.2512,
      "step": 9365
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.7100322246551514,
      "learning_rate": 0.0005975569479297157,
      "loss": 2.4751,
      "step": 9366
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6601682305335999,
      "learning_rate": 0.0005975564269260093,
      "loss": 2.7821,
      "step": 9367
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.651451051235199,
      "learning_rate": 0.0005975559058669815,
      "loss": 2.4034,
      "step": 9368
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.5832798480987549,
      "learning_rate": 0.0005975553847526326,
      "loss": 2.4919,
      "step": 9369
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6492547988891602,
      "learning_rate": 0.0005975548635829623,
      "loss": 2.5892,
      "step": 9370
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.7515953779220581,
      "learning_rate": 0.0005975543423579711,
      "loss": 2.7188,
      "step": 9371
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6517366170883179,
      "learning_rate": 0.0005975538210776589,
      "loss": 2.462,
      "step": 9372
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.603129506111145,
      "learning_rate": 0.0005975532997420258,
      "loss": 2.4235,
      "step": 9373
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6430273056030273,
      "learning_rate": 0.000597552778351072,
      "loss": 2.9069,
      "step": 9374
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6546298861503601,
      "learning_rate": 0.0005975522569047975,
      "loss": 2.4978,
      "step": 9375
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.5827358961105347,
      "learning_rate": 0.0005975517354032024,
      "loss": 2.2415,
      "step": 9376
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.5681771039962769,
      "learning_rate": 0.0005975512138462867,
      "loss": 2.5483,
      "step": 9377
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6258590221405029,
      "learning_rate": 0.0005975506922340508,
      "loss": 2.6653,
      "step": 9378
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6170920133590698,
      "learning_rate": 0.0005975501705664944,
      "loss": 2.5223,
      "step": 9379
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.647199809551239,
      "learning_rate": 0.000597549648843618,
      "loss": 2.6914,
      "step": 9380
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6525857448577881,
      "learning_rate": 0.0005975491270654214,
      "loss": 2.6048,
      "step": 9381
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.5891449451446533,
      "learning_rate": 0.0005975486052319049,
      "loss": 2.5934,
      "step": 9382
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6427908539772034,
      "learning_rate": 0.0005975480833430683,
      "loss": 2.2974,
      "step": 9383
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6429479122161865,
      "learning_rate": 0.0005975475613989121,
      "loss": 2.4206,
      "step": 9384
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6013923287391663,
      "learning_rate": 0.0005975470393994361,
      "loss": 2.6833,
      "step": 9385
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6173408627510071,
      "learning_rate": 0.0005975465173446406,
      "loss": 2.7708,
      "step": 9386
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.5974838137626648,
      "learning_rate": 0.0005975459952345254,
      "loss": 2.5186,
      "step": 9387
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.7457132339477539,
      "learning_rate": 0.0005975454730690908,
      "loss": 2.6236,
      "step": 9388
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6592904925346375,
      "learning_rate": 0.000597544950848337,
      "loss": 2.4197,
      "step": 9389
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6958774328231812,
      "learning_rate": 0.0005975444285722639,
      "loss": 2.2695,
      "step": 9390
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.5980205535888672,
      "learning_rate": 0.0005975439062408718,
      "loss": 2.3635,
      "step": 9391
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.7075120806694031,
      "learning_rate": 0.0005975433838541606,
      "loss": 2.4811,
      "step": 9392
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6064386367797852,
      "learning_rate": 0.0005975428614121304,
      "loss": 2.4884,
      "step": 9393
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.7765331268310547,
      "learning_rate": 0.0005975423389147814,
      "loss": 2.6145,
      "step": 9394
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.5776457190513611,
      "learning_rate": 0.0005975418163621137,
      "loss": 2.3448,
      "step": 9395
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6722944378852844,
      "learning_rate": 0.0005975412937541273,
      "loss": 2.5251,
      "step": 9396
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6637967824935913,
      "learning_rate": 0.0005975407710908223,
      "loss": 2.402,
      "step": 9397
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6740685701370239,
      "learning_rate": 0.000597540248372199,
      "loss": 2.413,
      "step": 9398
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6398542523384094,
      "learning_rate": 0.0005975397255982573,
      "loss": 2.4594,
      "step": 9399
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6787790656089783,
      "learning_rate": 0.0005975392027689974,
      "loss": 2.488,
      "step": 9400
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.5871837735176086,
      "learning_rate": 0.0005975386798844193,
      "loss": 2.5516,
      "step": 9401
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6237117648124695,
      "learning_rate": 0.000597538156944523,
      "loss": 2.487,
      "step": 9402
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.5863282084465027,
      "learning_rate": 0.0005975376339493089,
      "loss": 2.502,
      "step": 9403
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6039919853210449,
      "learning_rate": 0.000597537110898777,
      "loss": 2.2044,
      "step": 9404
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6236836910247803,
      "learning_rate": 0.0005975365877929271,
      "loss": 2.3595,
      "step": 9405
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6621925234794617,
      "learning_rate": 0.0005975360646317598,
      "loss": 2.5859,
      "step": 9406
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.687091052532196,
      "learning_rate": 0.0005975355414152748,
      "loss": 2.1225,
      "step": 9407
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6183642148971558,
      "learning_rate": 0.0005975350181434724,
      "loss": 2.4253,
      "step": 9408
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6379200220108032,
      "learning_rate": 0.0005975344948163524,
      "loss": 2.4648,
      "step": 9409
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6907662153244019,
      "learning_rate": 0.0005975339714339153,
      "loss": 2.4213,
      "step": 9410
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.5773701667785645,
      "learning_rate": 0.000597533447996161,
      "loss": 2.2836,
      "step": 9411
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6232163906097412,
      "learning_rate": 0.0005975329245030897,
      "loss": 2.4022,
      "step": 9412
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6819777488708496,
      "learning_rate": 0.0005975324009547013,
      "loss": 2.4403,
      "step": 9413
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.7427830100059509,
      "learning_rate": 0.000597531877350996,
      "loss": 2.3071,
      "step": 9414
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6237528324127197,
      "learning_rate": 0.0005975313536919739,
      "loss": 2.5532,
      "step": 9415
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.8410951495170593,
      "learning_rate": 0.0005975308299776352,
      "loss": 2.4141,
      "step": 9416
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6036801338195801,
      "learning_rate": 0.0005975303062079797,
      "loss": 2.2532,
      "step": 9417
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6641757488250732,
      "learning_rate": 0.000597529782383008,
      "loss": 2.4558,
      "step": 9418
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6475343704223633,
      "learning_rate": 0.0005975292585027197,
      "loss": 2.4263,
      "step": 9419
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.617002546787262,
      "learning_rate": 0.0005975287345671151,
      "loss": 2.5085,
      "step": 9420
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6134065985679626,
      "learning_rate": 0.0005975282105761943,
      "loss": 2.3949,
      "step": 9421
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6052200794219971,
      "learning_rate": 0.0005975276865299576,
      "loss": 2.5037,
      "step": 9422
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.7097326517105103,
      "learning_rate": 0.0005975271624284046,
      "loss": 2.6082,
      "step": 9423
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.622654378414154,
      "learning_rate": 0.0005975266382715358,
      "loss": 2.7182,
      "step": 9424
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6681594848632812,
      "learning_rate": 0.0005975261140593512,
      "loss": 2.729,
      "step": 9425
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.626351535320282,
      "learning_rate": 0.0005975255897918508,
      "loss": 2.5091,
      "step": 9426
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6166532039642334,
      "learning_rate": 0.0005975250654690348,
      "loss": 2.5582,
      "step": 9427
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.5653041005134583,
      "learning_rate": 0.0005975245410909034,
      "loss": 2.4118,
      "step": 9428
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6232761740684509,
      "learning_rate": 0.0005975240166574564,
      "loss": 2.7078,
      "step": 9429
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.5927965044975281,
      "learning_rate": 0.0005975234921686941,
      "loss": 2.4441,
      "step": 9430
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.5955779552459717,
      "learning_rate": 0.0005975229676246167,
      "loss": 2.4651,
      "step": 9431
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6506328582763672,
      "learning_rate": 0.0005975224430252241,
      "loss": 2.5191,
      "step": 9432
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6202992796897888,
      "learning_rate": 0.0005975219183705164,
      "loss": 2.5572,
      "step": 9433
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6188558340072632,
      "learning_rate": 0.0005975213936604938,
      "loss": 2.5961,
      "step": 9434
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.7548429369926453,
      "learning_rate": 0.0005975208688951563,
      "loss": 2.6413,
      "step": 9435
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.7019578218460083,
      "learning_rate": 0.0005975203440745042,
      "loss": 2.5802,
      "step": 9436
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6040416359901428,
      "learning_rate": 0.0005975198191985374,
      "loss": 2.1869,
      "step": 9437
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.7490707039833069,
      "learning_rate": 0.000597519294267256,
      "loss": 2.4059,
      "step": 9438
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6328274011611938,
      "learning_rate": 0.00059751876928066,
      "loss": 2.2976,
      "step": 9439
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6109668612480164,
      "learning_rate": 0.0005975182442387499,
      "loss": 2.3557,
      "step": 9440
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6451379060745239,
      "learning_rate": 0.0005975177191415255,
      "loss": 2.7631,
      "step": 9441
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6214423179626465,
      "learning_rate": 0.0005975171939889868,
      "loss": 2.5537,
      "step": 9442
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.7066730260848999,
      "learning_rate": 0.0005975166687811342,
      "loss": 2.3926,
      "step": 9443
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.7520071268081665,
      "learning_rate": 0.0005975161435179675,
      "loss": 2.2361,
      "step": 9444
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6393216252326965,
      "learning_rate": 0.000597515618199487,
      "loss": 2.5077,
      "step": 9445
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.5673801302909851,
      "learning_rate": 0.0005975150928256926,
      "loss": 2.556,
      "step": 9446
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.7045040726661682,
      "learning_rate": 0.0005975145673965847,
      "loss": 2.4122,
      "step": 9447
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6825363039970398,
      "learning_rate": 0.0005975140419121632,
      "loss": 2.6602,
      "step": 9448
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6176456809043884,
      "learning_rate": 0.0005975135163724281,
      "loss": 2.4335,
      "step": 9449
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6463978886604309,
      "learning_rate": 0.0005975129907773799,
      "loss": 2.3132,
      "step": 9450
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6931820511817932,
      "learning_rate": 0.0005975124651270181,
      "loss": 2.4134,
      "step": 9451
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6252059936523438,
      "learning_rate": 0.0005975119394213432,
      "loss": 2.5073,
      "step": 9452
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6284531950950623,
      "learning_rate": 0.0005975114136603553,
      "loss": 2.5471,
      "step": 9453
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6659746170043945,
      "learning_rate": 0.0005975108878440545,
      "loss": 2.4303,
      "step": 9454
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6314805150032043,
      "learning_rate": 0.0005975103619724406,
      "loss": 2.4285,
      "step": 9455
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6116612553596497,
      "learning_rate": 0.000597509836045514,
      "loss": 2.4948,
      "step": 9456
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6251944303512573,
      "learning_rate": 0.0005975093100632747,
      "loss": 2.3854,
      "step": 9457
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6760063767433167,
      "learning_rate": 0.0005975087840257229,
      "loss": 2.386,
      "step": 9458
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6504049301147461,
      "learning_rate": 0.0005975082579328585,
      "loss": 2.4328,
      "step": 9459
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.7618524432182312,
      "learning_rate": 0.0005975077317846816,
      "loss": 2.4398,
      "step": 9460
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6176484227180481,
      "learning_rate": 0.0005975072055811926,
      "loss": 2.3441,
      "step": 9461
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.7060741186141968,
      "learning_rate": 0.0005975066793223912,
      "loss": 2.5893,
      "step": 9462
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6361497640609741,
      "learning_rate": 0.000597506153008278,
      "loss": 2.3016,
      "step": 9463
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6227588653564453,
      "learning_rate": 0.0005975056266388524,
      "loss": 2.519,
      "step": 9464
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.985935389995575,
      "learning_rate": 0.0005975051002141151,
      "loss": 2.5766,
      "step": 9465
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.7870603203773499,
      "learning_rate": 0.000597504573734066,
      "loss": 2.603,
      "step": 9466
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.7087574601173401,
      "learning_rate": 0.000597504047198705,
      "loss": 2.3673,
      "step": 9467
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6088423728942871,
      "learning_rate": 0.0005975035206080327,
      "loss": 2.6714,
      "step": 9468
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.8887844681739807,
      "learning_rate": 0.0005975029939620486,
      "loss": 2.5826,
      "step": 9469
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6801655292510986,
      "learning_rate": 0.0005975024672607532,
      "loss": 2.4197,
      "step": 9470
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.7804757356643677,
      "learning_rate": 0.0005975019405041464,
      "loss": 2.5197,
      "step": 9471
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.5904048681259155,
      "learning_rate": 0.0005975014136922284,
      "loss": 2.4448,
      "step": 9472
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6483222246170044,
      "learning_rate": 0.0005975008868249994,
      "loss": 2.3971,
      "step": 9473
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6644185781478882,
      "learning_rate": 0.0005975003599024592,
      "loss": 2.7368,
      "step": 9474
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6661714315414429,
      "learning_rate": 0.0005974998329246081,
      "loss": 2.3815,
      "step": 9475
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.5948825478553772,
      "learning_rate": 0.0005974993058914462,
      "loss": 2.2116,
      "step": 9476
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6333258152008057,
      "learning_rate": 0.0005974987788029736,
      "loss": 2.5995,
      "step": 9477
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6174795627593994,
      "learning_rate": 0.0005974982516591904,
      "loss": 2.6954,
      "step": 9478
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6533438563346863,
      "learning_rate": 0.0005974977244600965,
      "loss": 2.5141,
      "step": 9479
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6560048460960388,
      "learning_rate": 0.0005974971972056923,
      "loss": 2.5888,
      "step": 9480
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6551209688186646,
      "learning_rate": 0.0005974966698959777,
      "loss": 2.5145,
      "step": 9481
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6951030492782593,
      "learning_rate": 0.0005974961425309528,
      "loss": 2.4856,
      "step": 9482
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6240506172180176,
      "learning_rate": 0.0005974956151106178,
      "loss": 2.5483,
      "step": 9483
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.678217887878418,
      "learning_rate": 0.0005974950876349728,
      "loss": 2.5451,
      "step": 9484
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6198241114616394,
      "learning_rate": 0.0005974945601040177,
      "loss": 2.5356,
      "step": 9485
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.5822380185127258,
      "learning_rate": 0.000597494032517753,
      "loss": 2.7006,
      "step": 9486
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.7236335277557373,
      "learning_rate": 0.0005974935048761783,
      "loss": 2.577,
      "step": 9487
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.5925907492637634,
      "learning_rate": 0.000597492977179294,
      "loss": 2.4969,
      "step": 9488
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.5992047786712646,
      "learning_rate": 0.0005974924494271003,
      "loss": 2.4837,
      "step": 9489
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6850569248199463,
      "learning_rate": 0.000597491921619597,
      "loss": 2.5836,
      "step": 9490
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6276118159294128,
      "learning_rate": 0.0005974913937567842,
      "loss": 2.4525,
      "step": 9491
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6103301644325256,
      "learning_rate": 0.0005974908658386624,
      "loss": 2.3941,
      "step": 9492
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6642000675201416,
      "learning_rate": 0.0005974903378652312,
      "loss": 2.4446,
      "step": 9493
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.8682554960250854,
      "learning_rate": 0.000597489809836491,
      "loss": 2.3463,
      "step": 9494
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.621099054813385,
      "learning_rate": 0.000597489281752442,
      "loss": 2.3231,
      "step": 9495
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.7289367318153381,
      "learning_rate": 0.0005974887536130839,
      "loss": 2.7273,
      "step": 9496
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.9061070084571838,
      "learning_rate": 0.0005974882254184172,
      "loss": 2.3734,
      "step": 9497
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6342456936836243,
      "learning_rate": 0.0005974876971684417,
      "loss": 2.485,
      "step": 9498
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6500625014305115,
      "learning_rate": 0.0005974871688631577,
      "loss": 2.3219,
      "step": 9499
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.0322866439819336,
      "learning_rate": 0.0005974866405025652,
      "loss": 2.5595,
      "step": 9500
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.5935968160629272,
      "learning_rate": 0.0005974861120866643,
      "loss": 2.4926,
      "step": 9501
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.7442952990531921,
      "learning_rate": 0.0005974855836154551,
      "loss": 2.4428,
      "step": 9502
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.7122770547866821,
      "learning_rate": 0.0005974850550889378,
      "loss": 2.6633,
      "step": 9503
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6572673320770264,
      "learning_rate": 0.0005974845265071123,
      "loss": 2.5847,
      "step": 9504
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6241582036018372,
      "learning_rate": 0.0005974839978699789,
      "loss": 2.5912,
      "step": 9505
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.683864176273346,
      "learning_rate": 0.0005974834691775375,
      "loss": 2.6543,
      "step": 9506
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.885795533657074,
      "learning_rate": 0.0005974829404297884,
      "loss": 2.6108,
      "step": 9507
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.7294597029685974,
      "learning_rate": 0.0005974824116267316,
      "loss": 2.5182,
      "step": 9508
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.7602143287658691,
      "learning_rate": 0.0005974818827683672,
      "loss": 2.6374,
      "step": 9509
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.7913572192192078,
      "learning_rate": 0.0005974813538546953,
      "loss": 2.4156,
      "step": 9510
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6445379853248596,
      "learning_rate": 0.000597480824885716,
      "loss": 2.7659,
      "step": 9511
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6880363821983337,
      "learning_rate": 0.0005974802958614293,
      "loss": 2.38,
      "step": 9512
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.654551088809967,
      "learning_rate": 0.0005974797667818356,
      "loss": 2.4197,
      "step": 9513
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6869388818740845,
      "learning_rate": 0.0005974792376469346,
      "loss": 2.309,
      "step": 9514
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.5900585055351257,
      "learning_rate": 0.0005974787084567267,
      "loss": 2.5278,
      "step": 9515
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.7031912803649902,
      "learning_rate": 0.0005974781792112119,
      "loss": 2.415,
      "step": 9516
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6679295301437378,
      "learning_rate": 0.0005974776499103903,
      "loss": 2.3692,
      "step": 9517
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6457729339599609,
      "learning_rate": 0.0005974771205542619,
      "loss": 2.3871,
      "step": 9518
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.7903349995613098,
      "learning_rate": 0.0005974765911428269,
      "loss": 2.5762,
      "step": 9519
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6427486538887024,
      "learning_rate": 0.0005974760616760855,
      "loss": 2.4944,
      "step": 9520
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6422420740127563,
      "learning_rate": 0.0005974755321540376,
      "loss": 2.6893,
      "step": 9521
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6585607528686523,
      "learning_rate": 0.0005974750025766833,
      "loss": 2.2568,
      "step": 9522
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6908891201019287,
      "learning_rate": 0.0005974744729440229,
      "loss": 2.6842,
      "step": 9523
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.5567489266395569,
      "learning_rate": 0.0005974739432560563,
      "loss": 2.3946,
      "step": 9524
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.5832399129867554,
      "learning_rate": 0.0005974734135127838,
      "loss": 2.4147,
      "step": 9525
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.570545494556427,
      "learning_rate": 0.0005974728837142054,
      "loss": 2.6705,
      "step": 9526
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.5782137513160706,
      "learning_rate": 0.000597472353860321,
      "loss": 2.5917,
      "step": 9527
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6649336218833923,
      "learning_rate": 0.000597471823951131,
      "loss": 2.5039,
      "step": 9528
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.7038012742996216,
      "learning_rate": 0.0005974712939866353,
      "loss": 2.8907,
      "step": 9529
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6390303373336792,
      "learning_rate": 0.0005974707639668341,
      "loss": 2.5892,
      "step": 9530
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6805537343025208,
      "learning_rate": 0.0005974702338917275,
      "loss": 2.5796,
      "step": 9531
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.7555768489837646,
      "learning_rate": 0.0005974697037613155,
      "loss": 2.5193,
      "step": 9532
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6756000518798828,
      "learning_rate": 0.0005974691735755983,
      "loss": 2.3546,
      "step": 9533
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6655682325363159,
      "learning_rate": 0.000597468643334576,
      "loss": 2.4528,
      "step": 9534
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.5693031549453735,
      "learning_rate": 0.0005974681130382487,
      "loss": 2.3763,
      "step": 9535
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6163374781608582,
      "learning_rate": 0.0005974675826866164,
      "loss": 2.3218,
      "step": 9536
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6015415191650391,
      "learning_rate": 0.0005974670522796792,
      "loss": 2.5738,
      "step": 9537
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.7075997591018677,
      "learning_rate": 0.0005974665218174374,
      "loss": 2.4132,
      "step": 9538
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6535462141036987,
      "learning_rate": 0.0005974659912998908,
      "loss": 2.4333,
      "step": 9539
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6319313049316406,
      "learning_rate": 0.0005974654607270398,
      "loss": 2.4531,
      "step": 9540
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.8043545484542847,
      "learning_rate": 0.0005974649300988843,
      "loss": 2.5217,
      "step": 9541
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.5815404057502747,
      "learning_rate": 0.0005974643994154245,
      "loss": 2.4446,
      "step": 9542
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6571712493896484,
      "learning_rate": 0.0005974638686766603,
      "loss": 2.4988,
      "step": 9543
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.656921923160553,
      "learning_rate": 0.0005974633378825921,
      "loss": 2.4658,
      "step": 9544
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.664361834526062,
      "learning_rate": 0.0005974628070332199,
      "loss": 2.5146,
      "step": 9545
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6458097100257874,
      "learning_rate": 0.0005974622761285435,
      "loss": 2.671,
      "step": 9546
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.7514081597328186,
      "learning_rate": 0.0005974617451685634,
      "loss": 2.7291,
      "step": 9547
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6416335701942444,
      "learning_rate": 0.0005974612141532796,
      "loss": 2.5171,
      "step": 9548
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6142485737800598,
      "learning_rate": 0.0005974606830826922,
      "loss": 2.5622,
      "step": 9549
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6011993885040283,
      "learning_rate": 0.000597460151956801,
      "loss": 2.6035,
      "step": 9550
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6250151991844177,
      "learning_rate": 0.0005974596207756064,
      "loss": 2.4695,
      "step": 9551
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.608584463596344,
      "learning_rate": 0.0005974590895391086,
      "loss": 2.5991,
      "step": 9552
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6474897861480713,
      "learning_rate": 0.0005974585582473076,
      "loss": 2.4191,
      "step": 9553
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.5796136856079102,
      "learning_rate": 0.0005974580269002032,
      "loss": 2.4088,
      "step": 9554
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.5974692702293396,
      "learning_rate": 0.0005974574954977958,
      "loss": 2.4413,
      "step": 9555
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6535781621932983,
      "learning_rate": 0.0005974569640400855,
      "loss": 2.7565,
      "step": 9556
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6104631423950195,
      "learning_rate": 0.0005974564325270723,
      "loss": 2.5524,
      "step": 9557
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.5856164693832397,
      "learning_rate": 0.0005974559009587564,
      "loss": 2.5148,
      "step": 9558
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6576020121574402,
      "learning_rate": 0.0005974553693351378,
      "loss": 2.7272,
      "step": 9559
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.666203498840332,
      "learning_rate": 0.0005974548376562166,
      "loss": 2.527,
      "step": 9560
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.7567726373672485,
      "learning_rate": 0.0005974543059219929,
      "loss": 2.5155,
      "step": 9561
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6899966597557068,
      "learning_rate": 0.0005974537741324669,
      "loss": 2.4777,
      "step": 9562
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6641089916229248,
      "learning_rate": 0.0005974532422876385,
      "loss": 2.4091,
      "step": 9563
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6436910033226013,
      "learning_rate": 0.0005974527103875081,
      "loss": 2.3553,
      "step": 9564
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6850947737693787,
      "learning_rate": 0.0005974521784320756,
      "loss": 2.3284,
      "step": 9565
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6175239682197571,
      "learning_rate": 0.000597451646421341,
      "loss": 2.3027,
      "step": 9566
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.613810658454895,
      "learning_rate": 0.0005974511143553046,
      "loss": 2.4668,
      "step": 9567
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6370190978050232,
      "learning_rate": 0.0005974505822339664,
      "loss": 2.4631,
      "step": 9568
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6381818056106567,
      "learning_rate": 0.0005974500500573266,
      "loss": 2.3748,
      "step": 9569
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.5977756381034851,
      "learning_rate": 0.0005974495178253851,
      "loss": 2.556,
      "step": 9570
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6537011861801147,
      "learning_rate": 0.0005974489855381423,
      "loss": 2.4609,
      "step": 9571
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.5909146070480347,
      "learning_rate": 0.000597448453195598,
      "loss": 2.4692,
      "step": 9572
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.5854516625404358,
      "learning_rate": 0.0005974479207977524,
      "loss": 2.2698,
      "step": 9573
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6899049878120422,
      "learning_rate": 0.0005974473883446056,
      "loss": 2.5658,
      "step": 9574
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6322171688079834,
      "learning_rate": 0.0005974468558361578,
      "loss": 2.4076,
      "step": 9575
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6171317100524902,
      "learning_rate": 0.000597446323272409,
      "loss": 2.5068,
      "step": 9576
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.680110514163971,
      "learning_rate": 0.0005974457906533592,
      "loss": 2.5692,
      "step": 9577
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6415993571281433,
      "learning_rate": 0.0005974452579790088,
      "loss": 2.4159,
      "step": 9578
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6203016042709351,
      "learning_rate": 0.0005974447252493576,
      "loss": 2.2692,
      "step": 9579
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6343503594398499,
      "learning_rate": 0.0005974441924644058,
      "loss": 2.4602,
      "step": 9580
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6954516768455505,
      "learning_rate": 0.0005974436596241535,
      "loss": 2.5273,
      "step": 9581
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.7424353957176208,
      "learning_rate": 0.0005974431267286008,
      "loss": 2.5092,
      "step": 9582
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6262906789779663,
      "learning_rate": 0.0005974425937777479,
      "loss": 2.563,
      "step": 9583
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.8874003887176514,
      "learning_rate": 0.0005974420607715947,
      "loss": 2.6808,
      "step": 9584
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.7262342572212219,
      "learning_rate": 0.0005974415277101415,
      "loss": 2.3214,
      "step": 9585
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6305210590362549,
      "learning_rate": 0.0005974409945933883,
      "loss": 2.5285,
      "step": 9586
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.7514241933822632,
      "learning_rate": 0.0005974404614213351,
      "loss": 2.7035,
      "step": 9587
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.7732177376747131,
      "learning_rate": 0.0005974399281939822,
      "loss": 2.4351,
      "step": 9588
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6389257907867432,
      "learning_rate": 0.0005974393949113296,
      "loss": 2.3292,
      "step": 9589
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6925956606864929,
      "learning_rate": 0.0005974388615733774,
      "loss": 2.38,
      "step": 9590
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.7247251272201538,
      "learning_rate": 0.0005974383281801257,
      "loss": 2.6231,
      "step": 9591
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6453830003738403,
      "learning_rate": 0.0005974377947315745,
      "loss": 2.7069,
      "step": 9592
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6897680163383484,
      "learning_rate": 0.0005974372612277242,
      "loss": 2.4544,
      "step": 9593
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6160435080528259,
      "learning_rate": 0.0005974367276685745,
      "loss": 2.7175,
      "step": 9594
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6686676740646362,
      "learning_rate": 0.0005974361940541258,
      "loss": 2.5213,
      "step": 9595
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.624761700630188,
      "learning_rate": 0.000597435660384378,
      "loss": 2.6789,
      "step": 9596
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6075937151908875,
      "learning_rate": 0.0005974351266593315,
      "loss": 2.5839,
      "step": 9597
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6862640380859375,
      "learning_rate": 0.0005974345928789859,
      "loss": 2.4118,
      "step": 9598
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.5992501974105835,
      "learning_rate": 0.0005974340590433419,
      "loss": 2.3848,
      "step": 9599
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6806066632270813,
      "learning_rate": 0.0005974335251523991,
      "loss": 2.3868,
      "step": 9600
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6431940793991089,
      "learning_rate": 0.0005974329912061579,
      "loss": 2.5992,
      "step": 9601
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6585692167282104,
      "learning_rate": 0.0005974324572046182,
      "loss": 2.6544,
      "step": 9602
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6227931976318359,
      "learning_rate": 0.0005974319231477803,
      "loss": 2.5019,
      "step": 9603
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.691974401473999,
      "learning_rate": 0.000597431389035644,
      "loss": 2.4777,
      "step": 9604
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6049957871437073,
      "learning_rate": 0.0005974308548682097,
      "loss": 2.4847,
      "step": 9605
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6427229642868042,
      "learning_rate": 0.0005974303206454774,
      "loss": 2.3995,
      "step": 9606
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.7234681248664856,
      "learning_rate": 0.0005974297863674473,
      "loss": 2.366,
      "step": 9607
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6380309462547302,
      "learning_rate": 0.0005974292520341192,
      "loss": 2.422,
      "step": 9608
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6196337342262268,
      "learning_rate": 0.0005974287176454935,
      "loss": 2.4879,
      "step": 9609
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6700419783592224,
      "learning_rate": 0.0005974281832015701,
      "loss": 2.3925,
      "step": 9610
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6532106399536133,
      "learning_rate": 0.0005974276487023493,
      "loss": 2.4957,
      "step": 9611
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.669128954410553,
      "learning_rate": 0.000597427114147831,
      "loss": 2.522,
      "step": 9612
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6729409098625183,
      "learning_rate": 0.0005974265795380154,
      "loss": 2.5443,
      "step": 9613
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6801761984825134,
      "learning_rate": 0.0005974260448729026,
      "loss": 2.3778,
      "step": 9614
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6611616611480713,
      "learning_rate": 0.0005974255101524927,
      "loss": 2.667,
      "step": 9615
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.71599280834198,
      "learning_rate": 0.0005974249753767857,
      "loss": 2.2998,
      "step": 9616
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6229802966117859,
      "learning_rate": 0.0005974244405457819,
      "loss": 2.3551,
      "step": 9617
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6104423999786377,
      "learning_rate": 0.0005974239056594812,
      "loss": 2.4664,
      "step": 9618
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.639239490032196,
      "learning_rate": 0.0005974233707178838,
      "loss": 2.6741,
      "step": 9619
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6125588417053223,
      "learning_rate": 0.0005974228357209898,
      "loss": 2.4829,
      "step": 9620
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6304116249084473,
      "learning_rate": 0.0005974223006687993,
      "loss": 2.5193,
      "step": 9621
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6806975603103638,
      "learning_rate": 0.0005974217655613124,
      "loss": 2.5392,
      "step": 9622
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6135203242301941,
      "learning_rate": 0.0005974212303985291,
      "loss": 2.381,
      "step": 9623
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6559920310974121,
      "learning_rate": 0.0005974206951804495,
      "loss": 2.5313,
      "step": 9624
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.644106924533844,
      "learning_rate": 0.000597420159907074,
      "loss": 2.5402,
      "step": 9625
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6412453055381775,
      "learning_rate": 0.0005974196245784023,
      "loss": 2.5598,
      "step": 9626
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.7373601794242859,
      "learning_rate": 0.0005974190891944348,
      "loss": 2.5949,
      "step": 9627
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6187333464622498,
      "learning_rate": 0.0005974185537551714,
      "loss": 2.3858,
      "step": 9628
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6377822160720825,
      "learning_rate": 0.0005974180182606123,
      "loss": 2.8431,
      "step": 9629
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6891158819198608,
      "learning_rate": 0.0005974174827107576,
      "loss": 2.6161,
      "step": 9630
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.5977186560630798,
      "learning_rate": 0.0005974169471056073,
      "loss": 2.3481,
      "step": 9631
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6729185581207275,
      "learning_rate": 0.0005974164114451616,
      "loss": 2.2528,
      "step": 9632
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6906160712242126,
      "learning_rate": 0.0005974158757294206,
      "loss": 2.6111,
      "step": 9633
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6592735648155212,
      "learning_rate": 0.0005974153399583845,
      "loss": 2.5171,
      "step": 9634
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.5830219984054565,
      "learning_rate": 0.000597414804132053,
      "loss": 2.296,
      "step": 9635
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6677455902099609,
      "learning_rate": 0.0005974142682504268,
      "loss": 2.4847,
      "step": 9636
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6794869899749756,
      "learning_rate": 0.0005974137323135054,
      "loss": 2.7102,
      "step": 9637
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.5613369941711426,
      "learning_rate": 0.0005974131963212893,
      "loss": 2.6314,
      "step": 9638
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.9646713137626648,
      "learning_rate": 0.0005974126602737783,
      "loss": 2.767,
      "step": 9639
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6122754812240601,
      "learning_rate": 0.0005974121241709728,
      "loss": 2.4969,
      "step": 9640
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6214684247970581,
      "learning_rate": 0.0005974115880128729,
      "loss": 2.4003,
      "step": 9641
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.8203109502792358,
      "learning_rate": 0.0005974110517994785,
      "loss": 2.3319,
      "step": 9642
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.8847433924674988,
      "learning_rate": 0.0005974105155307896,
      "loss": 2.6886,
      "step": 9643
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.7057415843009949,
      "learning_rate": 0.0005974099792068067,
      "loss": 2.4255,
      "step": 9644
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.7709195017814636,
      "learning_rate": 0.0005974094428275296,
      "loss": 2.3495,
      "step": 9645
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6560505628585815,
      "learning_rate": 0.0005974089063929584,
      "loss": 2.3611,
      "step": 9646
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6942526698112488,
      "learning_rate": 0.0005974083699030933,
      "loss": 2.4167,
      "step": 9647
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6233760714530945,
      "learning_rate": 0.0005974078333579344,
      "loss": 2.3946,
      "step": 9648
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6480439305305481,
      "learning_rate": 0.0005974072967574818,
      "loss": 2.2369,
      "step": 9649
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.5760138034820557,
      "learning_rate": 0.0005974067601017354,
      "loss": 2.3647,
      "step": 9650
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.685636579990387,
      "learning_rate": 0.0005974062233906957,
      "loss": 2.4882,
      "step": 9651
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6781402826309204,
      "learning_rate": 0.0005974056866243625,
      "loss": 2.5719,
      "step": 9652
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.5681188106536865,
      "learning_rate": 0.0005974051498027358,
      "loss": 2.7303,
      "step": 9653
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6005931496620178,
      "learning_rate": 0.0005974046129258161,
      "loss": 2.3341,
      "step": 9654
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6662949919700623,
      "learning_rate": 0.0005974040759936032,
      "loss": 2.4382,
      "step": 9655
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6963602304458618,
      "learning_rate": 0.0005974035390060972,
      "loss": 2.4032,
      "step": 9656
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6553775072097778,
      "learning_rate": 0.0005974030019632983,
      "loss": 2.5721,
      "step": 9657
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6541913747787476,
      "learning_rate": 0.0005974024648652065,
      "loss": 2.4176,
      "step": 9658
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6213387846946716,
      "learning_rate": 0.000597401927711822,
      "loss": 2.5419,
      "step": 9659
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6532111763954163,
      "learning_rate": 0.000597401390503145,
      "loss": 2.4095,
      "step": 9660
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6820009350776672,
      "learning_rate": 0.0005974008532391753,
      "loss": 2.5358,
      "step": 9661
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6644058227539062,
      "learning_rate": 0.0005974003159199133,
      "loss": 2.4291,
      "step": 9662
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6037951707839966,
      "learning_rate": 0.0005973997785453589,
      "loss": 2.5214,
      "step": 9663
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6772308945655823,
      "learning_rate": 0.0005973992411155123,
      "loss": 2.627,
      "step": 9664
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6315735578536987,
      "learning_rate": 0.0005973987036303735,
      "loss": 2.4494,
      "step": 9665
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6679795980453491,
      "learning_rate": 0.0005973981660899427,
      "loss": 2.2764,
      "step": 9666
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6120812296867371,
      "learning_rate": 0.0005973976284942199,
      "loss": 2.7261,
      "step": 9667
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6755666732788086,
      "learning_rate": 0.0005973970908432054,
      "loss": 2.7546,
      "step": 9668
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6370528340339661,
      "learning_rate": 0.0005973965531368992,
      "loss": 2.2872,
      "step": 9669
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6313104033470154,
      "learning_rate": 0.0005973960153753012,
      "loss": 2.1675,
      "step": 9670
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.7052389979362488,
      "learning_rate": 0.0005973954775584118,
      "loss": 2.5707,
      "step": 9671
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.64186030626297,
      "learning_rate": 0.0005973949396862309,
      "loss": 2.4377,
      "step": 9672
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.637966513633728,
      "learning_rate": 0.0005973944017587586,
      "loss": 2.5459,
      "step": 9673
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6851649284362793,
      "learning_rate": 0.0005973938637759952,
      "loss": 2.5718,
      "step": 9674
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.65233314037323,
      "learning_rate": 0.0005973933257379406,
      "loss": 2.4383,
      "step": 9675
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.9882882833480835,
      "learning_rate": 0.000597392787644595,
      "loss": 2.3659,
      "step": 9676
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.5985305309295654,
      "learning_rate": 0.0005973922494959585,
      "loss": 2.2336,
      "step": 9677
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.7484848499298096,
      "learning_rate": 0.0005973917112920311,
      "loss": 2.3034,
      "step": 9678
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.8409428596496582,
      "learning_rate": 0.000597391173032813,
      "loss": 2.2987,
      "step": 9679
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6756424307823181,
      "learning_rate": 0.0005973906347183042,
      "loss": 2.6362,
      "step": 9680
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6838330626487732,
      "learning_rate": 0.000597390096348505,
      "loss": 2.5561,
      "step": 9681
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.7516229748725891,
      "learning_rate": 0.0005973895579234152,
      "loss": 2.4997,
      "step": 9682
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.586576521396637,
      "learning_rate": 0.0005973890194430351,
      "loss": 2.6289,
      "step": 9683
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.7172752022743225,
      "learning_rate": 0.0005973884809073648,
      "loss": 2.4564,
      "step": 9684
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6680992841720581,
      "learning_rate": 0.0005973879423164044,
      "loss": 2.4147,
      "step": 9685
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.636073648929596,
      "learning_rate": 0.000597387403670154,
      "loss": 2.5383,
      "step": 9686
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.635175347328186,
      "learning_rate": 0.0005973868649686137,
      "loss": 2.4113,
      "step": 9687
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6644116640090942,
      "learning_rate": 0.0005973863262117834,
      "loss": 2.5418,
      "step": 9688
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.640261173248291,
      "learning_rate": 0.0005973857873996635,
      "loss": 2.3613,
      "step": 9689
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6371264457702637,
      "learning_rate": 0.0005973852485322539,
      "loss": 2.6078,
      "step": 9690
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6609454154968262,
      "learning_rate": 0.0005973847096095548,
      "loss": 2.5621,
      "step": 9691
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.7156602740287781,
      "learning_rate": 0.0005973841706315662,
      "loss": 2.433,
      "step": 9692
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6094685792922974,
      "learning_rate": 0.0005973836315982884,
      "loss": 2.5615,
      "step": 9693
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6317328810691833,
      "learning_rate": 0.0005973830925097213,
      "loss": 2.5769,
      "step": 9694
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.7284839153289795,
      "learning_rate": 0.0005973825533658651,
      "loss": 2.3766,
      "step": 9695
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.5807275772094727,
      "learning_rate": 0.0005973820141667198,
      "loss": 2.5706,
      "step": 9696
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.7432609796524048,
      "learning_rate": 0.0005973814749122856,
      "loss": 2.695,
      "step": 9697
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6486495733261108,
      "learning_rate": 0.0005973809356025625,
      "loss": 2.4566,
      "step": 9698
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6054480671882629,
      "learning_rate": 0.0005973803962375509,
      "loss": 2.4398,
      "step": 9699
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.669575572013855,
      "learning_rate": 0.0005973798568172505,
      "loss": 2.5685,
      "step": 9700
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6483359336853027,
      "learning_rate": 0.0005973793173416616,
      "loss": 2.6077,
      "step": 9701
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6085394024848938,
      "learning_rate": 0.0005973787778107843,
      "loss": 2.603,
      "step": 9702
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6495712399482727,
      "learning_rate": 0.0005973782382246186,
      "loss": 2.586,
      "step": 9703
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.7317155003547668,
      "learning_rate": 0.0005973776985831647,
      "loss": 2.5265,
      "step": 9704
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6781693696975708,
      "learning_rate": 0.0005973771588864226,
      "loss": 2.6045,
      "step": 9705
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6565478444099426,
      "learning_rate": 0.0005973766191343927,
      "loss": 2.7012,
      "step": 9706
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6813040971755981,
      "learning_rate": 0.0005973760793270747,
      "loss": 2.4721,
      "step": 9707
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.5873168110847473,
      "learning_rate": 0.0005973755394644689,
      "loss": 2.5482,
      "step": 9708
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6432517170906067,
      "learning_rate": 0.0005973749995465753,
      "loss": 2.586,
      "step": 9709
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6417774558067322,
      "learning_rate": 0.0005973744595733942,
      "loss": 2.3779,
      "step": 9710
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.648996889591217,
      "learning_rate": 0.0005973739195449254,
      "loss": 2.4862,
      "step": 9711
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.7508783340454102,
      "learning_rate": 0.0005973733794611694,
      "loss": 2.3966,
      "step": 9712
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.5930466651916504,
      "learning_rate": 0.000597372839322126,
      "loss": 2.5595,
      "step": 9713
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6082624197006226,
      "learning_rate": 0.0005973722991277953,
      "loss": 2.3762,
      "step": 9714
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6923101544380188,
      "learning_rate": 0.0005973717588781775,
      "loss": 2.4993,
      "step": 9715
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6658212542533875,
      "learning_rate": 0.0005973712185732727,
      "loss": 2.3362,
      "step": 9716
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6843787431716919,
      "learning_rate": 0.000597370678213081,
      "loss": 2.3803,
      "step": 9717
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6956909894943237,
      "learning_rate": 0.0005973701377976025,
      "loss": 2.763,
      "step": 9718
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6105166077613831,
      "learning_rate": 0.0005973695973268372,
      "loss": 2.3734,
      "step": 9719
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.5558908581733704,
      "learning_rate": 0.0005973690568007853,
      "loss": 2.2897,
      "step": 9720
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6715424656867981,
      "learning_rate": 0.000597368516219447,
      "loss": 2.3403,
      "step": 9721
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6870969533920288,
      "learning_rate": 0.0005973679755828221,
      "loss": 2.5931,
      "step": 9722
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.5561820864677429,
      "learning_rate": 0.000597367434890911,
      "loss": 2.5571,
      "step": 9723
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6516152620315552,
      "learning_rate": 0.0005973668941437136,
      "loss": 2.3977,
      "step": 9724
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.8206365704536438,
      "learning_rate": 0.0005973663533412302,
      "loss": 2.3678,
      "step": 9725
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.692156970500946,
      "learning_rate": 0.0005973658124834607,
      "loss": 2.4658,
      "step": 9726
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.5935378074645996,
      "learning_rate": 0.0005973652715704053,
      "loss": 2.5201,
      "step": 9727
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6474798321723938,
      "learning_rate": 0.000597364730602064,
      "loss": 2.5726,
      "step": 9728
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6253500580787659,
      "learning_rate": 0.000597364189578437,
      "loss": 2.3981,
      "step": 9729
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6519289612770081,
      "learning_rate": 0.0005973636484995244,
      "loss": 2.5597,
      "step": 9730
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.7185409069061279,
      "learning_rate": 0.0005973631073653263,
      "loss": 2.2908,
      "step": 9731
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6578043103218079,
      "learning_rate": 0.0005973625661758429,
      "loss": 2.4684,
      "step": 9732
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.701417863368988,
      "learning_rate": 0.0005973620249310741,
      "loss": 2.6146,
      "step": 9733
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6446415185928345,
      "learning_rate": 0.00059736148363102,
      "loss": 2.3736,
      "step": 9734
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6570353507995605,
      "learning_rate": 0.0005973609422756809,
      "loss": 2.5945,
      "step": 9735
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.647384524345398,
      "learning_rate": 0.0005973604008650568,
      "loss": 2.3765,
      "step": 9736
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6918826699256897,
      "learning_rate": 0.0005973598593991477,
      "loss": 2.621,
      "step": 9737
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.7107141017913818,
      "learning_rate": 0.0005973593178779538,
      "loss": 2.4053,
      "step": 9738
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6648068428039551,
      "learning_rate": 0.0005973587763014752,
      "loss": 2.5219,
      "step": 9739
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6624457240104675,
      "learning_rate": 0.000597358234669712,
      "loss": 2.3974,
      "step": 9740
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6582625508308411,
      "learning_rate": 0.0005973576929826643,
      "loss": 2.5631,
      "step": 9741
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6196325421333313,
      "learning_rate": 0.0005973571512403322,
      "loss": 2.4199,
      "step": 9742
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6268435716629028,
      "learning_rate": 0.0005973566094427158,
      "loss": 2.4439,
      "step": 9743
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.7362654805183411,
      "learning_rate": 0.0005973560675898152,
      "loss": 2.4576,
      "step": 9744
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6638598442077637,
      "learning_rate": 0.0005973555256816304,
      "loss": 2.4938,
      "step": 9745
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6748959422111511,
      "learning_rate": 0.0005973549837181616,
      "loss": 2.5913,
      "step": 9746
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6651603579521179,
      "learning_rate": 0.000597354441699409,
      "loss": 2.647,
      "step": 9747
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6678398847579956,
      "learning_rate": 0.0005973538996253726,
      "loss": 2.4618,
      "step": 9748
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.63625168800354,
      "learning_rate": 0.0005973533574960524,
      "loss": 2.2853,
      "step": 9749
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.795071005821228,
      "learning_rate": 0.0005973528153114487,
      "loss": 2.2992,
      "step": 9750
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6740697026252747,
      "learning_rate": 0.0005973522730715615,
      "loss": 2.6004,
      "step": 9751
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.0434564352035522,
      "learning_rate": 0.0005973517307763908,
      "loss": 2.3866,
      "step": 9752
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6343405246734619,
      "learning_rate": 0.0005973511884259367,
      "loss": 1.9899,
      "step": 9753
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6625386476516724,
      "learning_rate": 0.0005973506460201996,
      "loss": 2.5283,
      "step": 9754
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.7225762605667114,
      "learning_rate": 0.0005973501035591794,
      "loss": 2.5535,
      "step": 9755
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.5880325436592102,
      "learning_rate": 0.0005973495610428762,
      "loss": 2.274,
      "step": 9756
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.7650577425956726,
      "learning_rate": 0.00059734901847129,
      "loss": 2.5096,
      "step": 9757
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6153398156166077,
      "learning_rate": 0.0005973484758444211,
      "loss": 2.5821,
      "step": 9758
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6088938117027283,
      "learning_rate": 0.0005973479331622695,
      "loss": 2.2513,
      "step": 9759
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6517243385314941,
      "learning_rate": 0.0005973473904248352,
      "loss": 2.2306,
      "step": 9760
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6037260293960571,
      "learning_rate": 0.0005973468476321185,
      "loss": 2.218,
      "step": 9761
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6252620816230774,
      "learning_rate": 0.0005973463047841194,
      "loss": 2.5298,
      "step": 9762
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6272823810577393,
      "learning_rate": 0.000597345761880838,
      "loss": 2.4723,
      "step": 9763
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.7001492381095886,
      "learning_rate": 0.0005973452189222744,
      "loss": 2.2192,
      "step": 9764
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6121810674667358,
      "learning_rate": 0.0005973446759084287,
      "loss": 2.3402,
      "step": 9765
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6536738276481628,
      "learning_rate": 0.0005973441328393011,
      "loss": 2.6194,
      "step": 9766
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6161366701126099,
      "learning_rate": 0.0005973435897148915,
      "loss": 2.4294,
      "step": 9767
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6419764161109924,
      "learning_rate": 0.0005973430465352001,
      "loss": 2.5302,
      "step": 9768
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.7832861542701721,
      "learning_rate": 0.0005973425033002271,
      "loss": 2.4725,
      "step": 9769
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6767429113388062,
      "learning_rate": 0.0005973419600099725,
      "loss": 2.6287,
      "step": 9770
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.5735018253326416,
      "learning_rate": 0.0005973414166644364,
      "loss": 2.4133,
      "step": 9771
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6321657299995422,
      "learning_rate": 0.0005973408732636189,
      "loss": 2.6448,
      "step": 9772
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6245401501655579,
      "learning_rate": 0.0005973403298075202,
      "loss": 2.5504,
      "step": 9773
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.5910741090774536,
      "learning_rate": 0.0005973397862961402,
      "loss": 2.5105,
      "step": 9774
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6080684661865234,
      "learning_rate": 0.0005973392427294792,
      "loss": 2.5642,
      "step": 9775
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.630255401134491,
      "learning_rate": 0.0005973386991075373,
      "loss": 2.4193,
      "step": 9776
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6115902066230774,
      "learning_rate": 0.0005973381554303143,
      "loss": 2.5805,
      "step": 9777
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6523119211196899,
      "learning_rate": 0.0005973376116978107,
      "loss": 2.2848,
      "step": 9778
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.5774565935134888,
      "learning_rate": 0.0005973370679100263,
      "loss": 2.4438,
      "step": 9779
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.601318895816803,
      "learning_rate": 0.0005973365240669616,
      "loss": 2.5572,
      "step": 9780
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6269409656524658,
      "learning_rate": 0.000597335980168616,
      "loss": 2.5984,
      "step": 9781
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6419196724891663,
      "learning_rate": 0.0005973354362149903,
      "loss": 2.3961,
      "step": 9782
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.5992161631584167,
      "learning_rate": 0.0005973348922060843,
      "loss": 2.6401,
      "step": 9783
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6084559559822083,
      "learning_rate": 0.0005973343481418982,
      "loss": 2.437,
      "step": 9784
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.59451824426651,
      "learning_rate": 0.0005973338040224319,
      "loss": 2.4706,
      "step": 9785
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6210563778877258,
      "learning_rate": 0.0005973332598476857,
      "loss": 2.4205,
      "step": 9786
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6376014351844788,
      "learning_rate": 0.0005973327156176596,
      "loss": 2.7176,
      "step": 9787
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.7205485701560974,
      "learning_rate": 0.0005973321713323538,
      "loss": 2.393,
      "step": 9788
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6619419455528259,
      "learning_rate": 0.0005973316269917683,
      "loss": 2.5505,
      "step": 9789
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.9631059169769287,
      "learning_rate": 0.0005973310825959032,
      "loss": 2.6426,
      "step": 9790
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.8385759592056274,
      "learning_rate": 0.0005973305381447587,
      "loss": 2.7806,
      "step": 9791
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.7539694905281067,
      "learning_rate": 0.0005973299936383349,
      "loss": 2.4258,
      "step": 9792
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.1204090118408203,
      "learning_rate": 0.0005973294490766316,
      "loss": 2.3149,
      "step": 9793
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.5901235938072205,
      "learning_rate": 0.0005973289044596494,
      "loss": 2.6318,
      "step": 9794
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.8472350239753723,
      "learning_rate": 0.000597328359787388,
      "loss": 2.2958,
      "step": 9795
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.838686466217041,
      "learning_rate": 0.0005973278150598477,
      "loss": 2.506,
      "step": 9796
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6460367441177368,
      "learning_rate": 0.0005973272702770285,
      "loss": 2.4698,
      "step": 9797
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.9086384773254395,
      "learning_rate": 0.0005973267254389305,
      "loss": 2.3033,
      "step": 9798
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6762154698371887,
      "learning_rate": 0.000597326180545554,
      "loss": 2.558,
      "step": 9799
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6143321394920349,
      "learning_rate": 0.0005973256355968989,
      "loss": 2.4188,
      "step": 9800
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6119604110717773,
      "learning_rate": 0.0005973250905929653,
      "loss": 2.4671,
      "step": 9801
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.8577539324760437,
      "learning_rate": 0.0005973245455337534,
      "loss": 2.6163,
      "step": 9802
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6497827172279358,
      "learning_rate": 0.0005973240004192632,
      "loss": 2.5289,
      "step": 9803
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.813210129737854,
      "learning_rate": 0.0005973234552494949,
      "loss": 2.3022,
      "step": 9804
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6894188523292542,
      "learning_rate": 0.0005973229100244485,
      "loss": 2.7422,
      "step": 9805
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.5665503144264221,
      "learning_rate": 0.0005973223647441243,
      "loss": 2.3212,
      "step": 9806
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6033083200454712,
      "learning_rate": 0.0005973218194085222,
      "loss": 2.4996,
      "step": 9807
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6421417593955994,
      "learning_rate": 0.0005973212740176423,
      "loss": 2.5734,
      "step": 9808
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.7160526514053345,
      "learning_rate": 0.0005973207285714848,
      "loss": 2.3753,
      "step": 9809
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6469672322273254,
      "learning_rate": 0.0005973201830700497,
      "loss": 2.4818,
      "step": 9810
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6243100166320801,
      "learning_rate": 0.0005973196375133373,
      "loss": 2.6588,
      "step": 9811
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6398212313652039,
      "learning_rate": 0.0005973190919013474,
      "loss": 2.4717,
      "step": 9812
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6422368288040161,
      "learning_rate": 0.0005973185462340804,
      "loss": 2.5834,
      "step": 9813
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6158798336982727,
      "learning_rate": 0.0005973180005115362,
      "loss": 2.4875,
      "step": 9814
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.5909608006477356,
      "learning_rate": 0.0005973174547337151,
      "loss": 2.4621,
      "step": 9815
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.5866550803184509,
      "learning_rate": 0.000597316908900617,
      "loss": 2.2406,
      "step": 9816
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.5412720441818237,
      "learning_rate": 0.000597316363012242,
      "loss": 2.4661,
      "step": 9817
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6112219095230103,
      "learning_rate": 0.0005973158170685904,
      "loss": 2.4311,
      "step": 9818
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6499000191688538,
      "learning_rate": 0.000597315271069662,
      "loss": 2.5609,
      "step": 9819
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6931999921798706,
      "learning_rate": 0.0005973147250154572,
      "loss": 2.6948,
      "step": 9820
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6290821433067322,
      "learning_rate": 0.000597314178905976,
      "loss": 2.629,
      "step": 9821
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6885320544242859,
      "learning_rate": 0.0005973136327412184,
      "loss": 2.2796,
      "step": 9822
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6628959774971008,
      "learning_rate": 0.0005973130865211846,
      "loss": 2.2651,
      "step": 9823
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6223399043083191,
      "learning_rate": 0.0005973125402458747,
      "loss": 2.6259,
      "step": 9824
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6498981714248657,
      "learning_rate": 0.0005973119939152888,
      "loss": 2.6365,
      "step": 9825
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.5608633160591125,
      "learning_rate": 0.000597311447529427,
      "loss": 2.3693,
      "step": 9826
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6987928748130798,
      "learning_rate": 0.0005973109010882894,
      "loss": 2.6224,
      "step": 9827
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6736286878585815,
      "learning_rate": 0.000597310354591876,
      "loss": 2.6394,
      "step": 9828
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6569401025772095,
      "learning_rate": 0.000597309808040187,
      "loss": 2.7084,
      "step": 9829
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6069628000259399,
      "learning_rate": 0.0005973092614332227,
      "loss": 2.3377,
      "step": 9830
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6720625162124634,
      "learning_rate": 0.0005973087147709828,
      "loss": 2.45,
      "step": 9831
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6534051895141602,
      "learning_rate": 0.0005973081680534677,
      "loss": 2.2774,
      "step": 9832
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6525243520736694,
      "learning_rate": 0.0005973076212806773,
      "loss": 2.3432,
      "step": 9833
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.7435030937194824,
      "learning_rate": 0.0005973070744526119,
      "loss": 2.5308,
      "step": 9834
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6252011060714722,
      "learning_rate": 0.0005973065275692714,
      "loss": 2.5217,
      "step": 9835
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6255733370780945,
      "learning_rate": 0.000597305980630656,
      "loss": 2.6482,
      "step": 9836
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6329558491706848,
      "learning_rate": 0.0005973054336367659,
      "loss": 2.4307,
      "step": 9837
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6029762029647827,
      "learning_rate": 0.0005973048865876011,
      "loss": 2.348,
      "step": 9838
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6202753186225891,
      "learning_rate": 0.0005973043394831617,
      "loss": 2.5942,
      "step": 9839
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6427822709083557,
      "learning_rate": 0.0005973037923234478,
      "loss": 2.3619,
      "step": 9840
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.56390780210495,
      "learning_rate": 0.0005973032451084594,
      "loss": 2.576,
      "step": 9841
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.597891092300415,
      "learning_rate": 0.0005973026978381969,
      "loss": 2.4992,
      "step": 9842
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.5976977348327637,
      "learning_rate": 0.0005973021505126601,
      "loss": 2.3269,
      "step": 9843
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6454935073852539,
      "learning_rate": 0.0005973016031318492,
      "loss": 2.5822,
      "step": 9844
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6269341707229614,
      "learning_rate": 0.0005973010556957645,
      "loss": 2.599,
      "step": 9845
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6871995329856873,
      "learning_rate": 0.0005973005082044057,
      "loss": 2.4011,
      "step": 9846
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6323846578598022,
      "learning_rate": 0.0005972999606577733,
      "loss": 2.5,
      "step": 9847
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6680314540863037,
      "learning_rate": 0.0005972994130558671,
      "loss": 2.4714,
      "step": 9848
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6522159576416016,
      "learning_rate": 0.0005972988653986873,
      "loss": 2.478,
      "step": 9849
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.5989124178886414,
      "learning_rate": 0.0005972983176862342,
      "loss": 2.5046,
      "step": 9850
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6730567216873169,
      "learning_rate": 0.0005972977699185075,
      "loss": 2.3973,
      "step": 9851
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.643714427947998,
      "learning_rate": 0.0005972972220955078,
      "loss": 2.6063,
      "step": 9852
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6755796670913696,
      "learning_rate": 0.0005972966742172348,
      "loss": 2.3901,
      "step": 9853
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6530597805976868,
      "learning_rate": 0.0005972961262836887,
      "loss": 2.3365,
      "step": 9854
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6284698247909546,
      "learning_rate": 0.0005972955782948696,
      "loss": 2.5244,
      "step": 9855
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6404896378517151,
      "learning_rate": 0.0005972950302507779,
      "loss": 2.4257,
      "step": 9856
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6485183238983154,
      "learning_rate": 0.0005972944821514131,
      "loss": 2.3677,
      "step": 9857
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6727850437164307,
      "learning_rate": 0.0005972939339967759,
      "loss": 2.3309,
      "step": 9858
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.659366250038147,
      "learning_rate": 0.000597293385786866,
      "loss": 2.5605,
      "step": 9859
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.700453519821167,
      "learning_rate": 0.0005972928375216837,
      "loss": 2.6287,
      "step": 9860
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6746672987937927,
      "learning_rate": 0.0005972922892012291,
      "loss": 2.5549,
      "step": 9861
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.752103865146637,
      "learning_rate": 0.0005972917408255021,
      "loss": 2.3142,
      "step": 9862
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6248006224632263,
      "learning_rate": 0.000597291192394503,
      "loss": 2.3894,
      "step": 9863
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6651391983032227,
      "learning_rate": 0.000597290643908232,
      "loss": 2.546,
      "step": 9864
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.630745530128479,
      "learning_rate": 0.0005972900953666888,
      "loss": 2.6365,
      "step": 9865
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6768046021461487,
      "learning_rate": 0.000597289546769874,
      "loss": 2.5909,
      "step": 9866
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6317228078842163,
      "learning_rate": 0.0005972889981177874,
      "loss": 2.5406,
      "step": 9867
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6395677924156189,
      "learning_rate": 0.0005972884494104292,
      "loss": 2.399,
      "step": 9868
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6696763038635254,
      "learning_rate": 0.0005972879006477993,
      "loss": 2.3456,
      "step": 9869
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6188483238220215,
      "learning_rate": 0.0005972873518298982,
      "loss": 2.4769,
      "step": 9870
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.7530800104141235,
      "learning_rate": 0.0005972868029567256,
      "loss": 2.3884,
      "step": 9871
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6836698055267334,
      "learning_rate": 0.0005972862540282817,
      "loss": 2.4716,
      "step": 9872
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.7024833559989929,
      "learning_rate": 0.0005972857050445669,
      "loss": 2.4446,
      "step": 9873
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6172358393669128,
      "learning_rate": 0.000597285156005581,
      "loss": 2.3181,
      "step": 9874
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.7320370674133301,
      "learning_rate": 0.0005972846069113242,
      "loss": 2.6133,
      "step": 9875
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6433345675468445,
      "learning_rate": 0.0005972840577617964,
      "loss": 2.4725,
      "step": 9876
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6948400139808655,
      "learning_rate": 0.000597283508556998,
      "loss": 2.4905,
      "step": 9877
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6845726370811462,
      "learning_rate": 0.000597282959296929,
      "loss": 2.4673,
      "step": 9878
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6046758890151978,
      "learning_rate": 0.0005972824099815895,
      "loss": 2.5315,
      "step": 9879
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6517435908317566,
      "learning_rate": 0.0005972818606109796,
      "loss": 2.3579,
      "step": 9880
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.5902876257896423,
      "learning_rate": 0.0005972813111850994,
      "loss": 2.5763,
      "step": 9881
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.673804521560669,
      "learning_rate": 0.0005972807617039489,
      "loss": 2.5002,
      "step": 9882
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6278915405273438,
      "learning_rate": 0.0005972802121675283,
      "loss": 2.2807,
      "step": 9883
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6879754662513733,
      "learning_rate": 0.0005972796625758378,
      "loss": 2.5311,
      "step": 9884
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.8235624432563782,
      "learning_rate": 0.0005972791129288773,
      "loss": 2.6083,
      "step": 9885
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6340550780296326,
      "learning_rate": 0.000597278563226647,
      "loss": 2.3422,
      "step": 9886
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6906378269195557,
      "learning_rate": 0.000597278013469147,
      "loss": 2.551,
      "step": 9887
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.5688841342926025,
      "learning_rate": 0.0005972774636563775,
      "loss": 2.5829,
      "step": 9888
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6478791832923889,
      "learning_rate": 0.0005972769137883385,
      "loss": 2.6266,
      "step": 9889
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6763994097709656,
      "learning_rate": 0.0005972763638650301,
      "loss": 2.5755,
      "step": 9890
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6732326149940491,
      "learning_rate": 0.0005972758138864523,
      "loss": 2.3974,
      "step": 9891
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6475211381912231,
      "learning_rate": 0.0005972752638526054,
      "loss": 2.3293,
      "step": 9892
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6679795384407043,
      "learning_rate": 0.0005972747137634895,
      "loss": 2.6434,
      "step": 9893
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.5824352502822876,
      "learning_rate": 0.0005972741636191044,
      "loss": 2.3277,
      "step": 9894
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6431677341461182,
      "learning_rate": 0.0005972736134194506,
      "loss": 2.5984,
      "step": 9895
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6282922625541687,
      "learning_rate": 0.000597273063164528,
      "loss": 2.5101,
      "step": 9896
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6399768590927124,
      "learning_rate": 0.0005972725128543367,
      "loss": 2.4178,
      "step": 9897
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6427382826805115,
      "learning_rate": 0.0005972719624888768,
      "loss": 2.5108,
      "step": 9898
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6752212643623352,
      "learning_rate": 0.0005972714120681485,
      "loss": 2.4652,
      "step": 9899
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6538428068161011,
      "learning_rate": 0.0005972708615921518,
      "loss": 2.6213,
      "step": 9900
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.5702461004257202,
      "learning_rate": 0.0005972703110608868,
      "loss": 2.5694,
      "step": 9901
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6251476407051086,
      "learning_rate": 0.0005972697604743536,
      "loss": 2.8531,
      "step": 9902
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.7381705045700073,
      "learning_rate": 0.0005972692098325525,
      "loss": 2.389,
      "step": 9903
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6672613024711609,
      "learning_rate": 0.0005972686591354834,
      "loss": 2.314,
      "step": 9904
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.693509578704834,
      "learning_rate": 0.0005972681083831464,
      "loss": 2.5246,
      "step": 9905
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6173236966133118,
      "learning_rate": 0.0005972675575755417,
      "loss": 2.6208,
      "step": 9906
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6071292161941528,
      "learning_rate": 0.0005972670067126692,
      "loss": 2.5495,
      "step": 9907
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6295875906944275,
      "learning_rate": 0.0005972664557945292,
      "loss": 2.3901,
      "step": 9908
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6413577198982239,
      "learning_rate": 0.0005972659048211218,
      "loss": 2.4933,
      "step": 9909
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6276688575744629,
      "learning_rate": 0.0005972653537924471,
      "loss": 2.5647,
      "step": 9910
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6691890358924866,
      "learning_rate": 0.0005972648027085052,
      "loss": 2.284,
      "step": 9911
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6275571584701538,
      "learning_rate": 0.000597264251569296,
      "loss": 2.4101,
      "step": 9912
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.7399587631225586,
      "learning_rate": 0.0005972637003748199,
      "loss": 2.4623,
      "step": 9913
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6315930485725403,
      "learning_rate": 0.0005972631491250768,
      "loss": 2.513,
      "step": 9914
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.5758311152458191,
      "learning_rate": 0.0005972625978200669,
      "loss": 2.5352,
      "step": 9915
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.8512519001960754,
      "learning_rate": 0.0005972620464597901,
      "loss": 2.25,
      "step": 9916
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6119183301925659,
      "learning_rate": 0.0005972614950442469,
      "loss": 2.4659,
      "step": 9917
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.7708292603492737,
      "learning_rate": 0.0005972609435734371,
      "loss": 2.2718,
      "step": 9918
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.669509768486023,
      "learning_rate": 0.0005972603920473609,
      "loss": 2.7424,
      "step": 9919
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.653984785079956,
      "learning_rate": 0.0005972598404660183,
      "loss": 2.8351,
      "step": 9920
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6454519629478455,
      "learning_rate": 0.0005972592888294095,
      "loss": 2.4512,
      "step": 9921
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6236579418182373,
      "learning_rate": 0.0005972587371375347,
      "loss": 2.4441,
      "step": 9922
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6387051939964294,
      "learning_rate": 0.0005972581853903938,
      "loss": 2.4729,
      "step": 9923
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6860491633415222,
      "learning_rate": 0.0005972576335879871,
      "loss": 2.3992,
      "step": 9924
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.7140533328056335,
      "learning_rate": 0.0005972570817303145,
      "loss": 2.3772,
      "step": 9925
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.642238974571228,
      "learning_rate": 0.0005972565298173762,
      "loss": 2.6792,
      "step": 9926
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.9828016757965088,
      "learning_rate": 0.0005972559778491723,
      "loss": 2.5191,
      "step": 9927
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6696702837944031,
      "learning_rate": 0.0005972554258257029,
      "loss": 2.3149,
      "step": 9928
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.9235528707504272,
      "learning_rate": 0.0005972548737469681,
      "loss": 2.492,
      "step": 9929
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6680126786231995,
      "learning_rate": 0.0005972543216129681,
      "loss": 2.3993,
      "step": 9930
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.7310847640037537,
      "learning_rate": 0.0005972537694237029,
      "loss": 2.3831,
      "step": 9931
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6448658108711243,
      "learning_rate": 0.0005972532171791725,
      "loss": 2.6456,
      "step": 9932
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6001796722412109,
      "learning_rate": 0.0005972526648793771,
      "loss": 2.4582,
      "step": 9933
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6407669186592102,
      "learning_rate": 0.0005972521125243168,
      "loss": 2.3141,
      "step": 9934
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.708377480506897,
      "learning_rate": 0.0005972515601139919,
      "loss": 2.5492,
      "step": 9935
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6831198334693909,
      "learning_rate": 0.0005972510076484022,
      "loss": 2.3811,
      "step": 9936
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6952113509178162,
      "learning_rate": 0.000597250455127548,
      "loss": 2.3652,
      "step": 9937
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6424943804740906,
      "learning_rate": 0.0005972499025514292,
      "loss": 2.3121,
      "step": 9938
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6140084862709045,
      "learning_rate": 0.0005972493499200462,
      "loss": 2.3358,
      "step": 9939
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.8208557367324829,
      "learning_rate": 0.0005972487972333987,
      "loss": 2.5202,
      "step": 9940
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.7647756934165955,
      "learning_rate": 0.0005972482444914873,
      "loss": 2.5344,
      "step": 9941
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6495298743247986,
      "learning_rate": 0.0005972476916943117,
      "loss": 2.3481,
      "step": 9942
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.7204620242118835,
      "learning_rate": 0.0005972471388418722,
      "loss": 2.4918,
      "step": 9943
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.7286180257797241,
      "learning_rate": 0.0005972465859341688,
      "loss": 2.3523,
      "step": 9944
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6601896286010742,
      "learning_rate": 0.0005972460329712017,
      "loss": 2.5167,
      "step": 9945
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.723276674747467,
      "learning_rate": 0.0005972454799529709,
      "loss": 2.357,
      "step": 9946
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.7225772738456726,
      "learning_rate": 0.0005972449268794765,
      "loss": 2.5392,
      "step": 9947
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.7250691056251526,
      "learning_rate": 0.0005972443737507188,
      "loss": 2.5846,
      "step": 9948
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.7353677153587341,
      "learning_rate": 0.0005972438205666976,
      "loss": 2.6427,
      "step": 9949
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6901823878288269,
      "learning_rate": 0.0005972432673274133,
      "loss": 2.3722,
      "step": 9950
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6219435930252075,
      "learning_rate": 0.0005972427140328658,
      "loss": 2.5659,
      "step": 9951
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6487213373184204,
      "learning_rate": 0.0005972421606830553,
      "loss": 2.5464,
      "step": 9952
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.701423168182373,
      "learning_rate": 0.0005972416072779819,
      "loss": 2.5346,
      "step": 9953
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.7062534093856812,
      "learning_rate": 0.0005972410538176456,
      "loss": 2.6345,
      "step": 9954
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6916684508323669,
      "learning_rate": 0.0005972405003020466,
      "loss": 2.6052,
      "step": 9955
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.7284784317016602,
      "learning_rate": 0.000597239946731185,
      "loss": 2.4517,
      "step": 9956
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6278971433639526,
      "learning_rate": 0.0005972393931050609,
      "loss": 2.8517,
      "step": 9957
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6409364938735962,
      "learning_rate": 0.0005972388394236744,
      "loss": 2.7172,
      "step": 9958
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6294249296188354,
      "learning_rate": 0.0005972382856870255,
      "loss": 2.4339,
      "step": 9959
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6842355728149414,
      "learning_rate": 0.0005972377318951144,
      "loss": 2.4349,
      "step": 9960
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6080299019813538,
      "learning_rate": 0.0005972371780479413,
      "loss": 2.6752,
      "step": 9961
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.756405234336853,
      "learning_rate": 0.0005972366241455061,
      "loss": 2.5998,
      "step": 9962
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6557538509368896,
      "learning_rate": 0.000597236070187809,
      "loss": 2.4828,
      "step": 9963
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6695542335510254,
      "learning_rate": 0.0005972355161748501,
      "loss": 2.5084,
      "step": 9964
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6208447813987732,
      "learning_rate": 0.0005972349621066297,
      "loss": 2.561,
      "step": 9965
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.5938739776611328,
      "learning_rate": 0.0005972344079831475,
      "loss": 2.3454,
      "step": 9966
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6142030954360962,
      "learning_rate": 0.0005972338538044038,
      "loss": 2.5319,
      "step": 9967
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.637821614742279,
      "learning_rate": 0.0005972332995703988,
      "loss": 2.4494,
      "step": 9968
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6130824685096741,
      "learning_rate": 0.0005972327452811324,
      "loss": 2.4481,
      "step": 9969
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6384583711624146,
      "learning_rate": 0.000597232190936605,
      "loss": 2.4926,
      "step": 9970
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6209374070167542,
      "learning_rate": 0.0005972316365368165,
      "loss": 2.5045,
      "step": 9971
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6063224673271179,
      "learning_rate": 0.0005972310820817669,
      "loss": 2.4451,
      "step": 9972
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6163194179534912,
      "learning_rate": 0.0005972305275714566,
      "loss": 2.4909,
      "step": 9973
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6379382610321045,
      "learning_rate": 0.0005972299730058854,
      "loss": 2.5031,
      "step": 9974
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6252753138542175,
      "learning_rate": 0.0005972294183850536,
      "loss": 2.482,
      "step": 9975
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6373335719108582,
      "learning_rate": 0.0005972288637089612,
      "loss": 2.4824,
      "step": 9976
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6313934326171875,
      "learning_rate": 0.0005972283089776084,
      "loss": 2.6669,
      "step": 9977
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.7749989032745361,
      "learning_rate": 0.0005972277541909951,
      "loss": 2.3361,
      "step": 9978
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6588315367698669,
      "learning_rate": 0.0005972271993491217,
      "loss": 2.6077,
      "step": 9979
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.7291847467422485,
      "learning_rate": 0.0005972266444519882,
      "loss": 2.4966,
      "step": 9980
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.742984414100647,
      "learning_rate": 0.0005972260894995945,
      "loss": 2.6555,
      "step": 9981
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6320390701293945,
      "learning_rate": 0.0005972255344919409,
      "loss": 2.6647,
      "step": 9982
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.7116559743881226,
      "learning_rate": 0.0005972249794290275,
      "loss": 2.5172,
      "step": 9983
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6742904782295227,
      "learning_rate": 0.0005972244243108543,
      "loss": 2.5056,
      "step": 9984
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.7100692391395569,
      "learning_rate": 0.0005972238691374216,
      "loss": 2.7078,
      "step": 9985
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6748220920562744,
      "learning_rate": 0.0005972233139087293,
      "loss": 2.4592,
      "step": 9986
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6740191578865051,
      "learning_rate": 0.0005972227586247775,
      "loss": 2.4667,
      "step": 9987
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6483249068260193,
      "learning_rate": 0.0005972222032855666,
      "loss": 2.6098,
      "step": 9988
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6613853573799133,
      "learning_rate": 0.0005972216478910962,
      "loss": 2.5592,
      "step": 9989
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6108744144439697,
      "learning_rate": 0.0005972210924413668,
      "loss": 2.5525,
      "step": 9990
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6346829533576965,
      "learning_rate": 0.0005972205369363784,
      "loss": 2.5564,
      "step": 9991
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.5989104509353638,
      "learning_rate": 0.0005972199813761311,
      "loss": 2.49,
      "step": 9992
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6672708988189697,
      "learning_rate": 0.000597219425760625,
      "loss": 2.4108,
      "step": 9993
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6218686699867249,
      "learning_rate": 0.0005972188700898602,
      "loss": 2.3009,
      "step": 9994
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6311861276626587,
      "learning_rate": 0.0005972183143638368,
      "loss": 2.4824,
      "step": 9995
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6186041235923767,
      "learning_rate": 0.000597217758582555,
      "loss": 2.3326,
      "step": 9996
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6318625211715698,
      "learning_rate": 0.0005972172027460146,
      "loss": 2.5165,
      "step": 9997
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6382810473442078,
      "learning_rate": 0.000597216646854216,
      "loss": 2.4617,
      "step": 9998
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.634028971195221,
      "learning_rate": 0.0005972160909071593,
      "loss": 2.4106,
      "step": 9999
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6292397379875183,
      "learning_rate": 0.0005972155349048445,
      "loss": 2.4164,
      "step": 10000
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6298540234565735,
      "learning_rate": 0.0005972149788472716,
      "loss": 2.5027,
      "step": 10001
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.5808999538421631,
      "learning_rate": 0.0005972144227344409,
      "loss": 2.6202,
      "step": 10002
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6642223596572876,
      "learning_rate": 0.0005972138665663525,
      "loss": 2.447,
      "step": 10003
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.5987881422042847,
      "learning_rate": 0.0005972133103430063,
      "loss": 2.2798,
      "step": 10004
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.5994360446929932,
      "learning_rate": 0.0005972127540644026,
      "loss": 2.5109,
      "step": 10005
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6545674800872803,
      "learning_rate": 0.0005972121977305414,
      "loss": 2.3646,
      "step": 10006
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6779726147651672,
      "learning_rate": 0.000597211641341423,
      "loss": 2.5697,
      "step": 10007
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.583670437335968,
      "learning_rate": 0.0005972110848970471,
      "loss": 2.5224,
      "step": 10008
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.7869941592216492,
      "learning_rate": 0.0005972105283974142,
      "loss": 2.4877,
      "step": 10009
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6031255125999451,
      "learning_rate": 0.0005972099718425242,
      "loss": 2.3769,
      "step": 10010
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.7157207727432251,
      "learning_rate": 0.0005972094152323773,
      "loss": 2.5274,
      "step": 10011
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6403871178627014,
      "learning_rate": 0.0005972088585669736,
      "loss": 2.4293,
      "step": 10012
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.5741674304008484,
      "learning_rate": 0.000597208301846313,
      "loss": 2.6158,
      "step": 10013
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.7176091074943542,
      "learning_rate": 0.0005972077450703959,
      "loss": 2.3238,
      "step": 10014
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6396023035049438,
      "learning_rate": 0.0005972071882392223,
      "loss": 2.1921,
      "step": 10015
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6784250140190125,
      "learning_rate": 0.0005972066313527922,
      "loss": 2.3698,
      "step": 10016
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6498278379440308,
      "learning_rate": 0.0005972060744111057,
      "loss": 2.6123,
      "step": 10017
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.589643657207489,
      "learning_rate": 0.0005972055174141631,
      "loss": 2.4994,
      "step": 10018
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6767696142196655,
      "learning_rate": 0.0005972049603619644,
      "loss": 2.5313,
      "step": 10019
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6809276342391968,
      "learning_rate": 0.0005972044032545095,
      "loss": 2.6485,
      "step": 10020
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.5822134017944336,
      "learning_rate": 0.0005972038460917989,
      "loss": 2.6036,
      "step": 10021
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6222313642501831,
      "learning_rate": 0.0005972032888738324,
      "loss": 2.4621,
      "step": 10022
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6077473759651184,
      "learning_rate": 0.0005972027316006102,
      "loss": 2.3138,
      "step": 10023
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.5377402901649475,
      "learning_rate": 0.0005972021742721324,
      "loss": 2.4226,
      "step": 10024
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6316465139389038,
      "learning_rate": 0.0005972016168883992,
      "loss": 2.5899,
      "step": 10025
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.5646108984947205,
      "learning_rate": 0.0005972010594494106,
      "loss": 2.3469,
      "step": 10026
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.662731409072876,
      "learning_rate": 0.0005972005019551667,
      "loss": 2.5461,
      "step": 10027
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6671806573867798,
      "learning_rate": 0.0005971999444056675,
      "loss": 2.668,
      "step": 10028
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.7498214840888977,
      "learning_rate": 0.0005971993868009133,
      "loss": 2.5153,
      "step": 10029
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6854942440986633,
      "learning_rate": 0.000597198829140904,
      "loss": 2.3768,
      "step": 10030
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.5914482474327087,
      "learning_rate": 0.0005971982714256401,
      "loss": 2.5408,
      "step": 10031
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.8949824571609497,
      "learning_rate": 0.0005971977136551213,
      "loss": 2.664,
      "step": 10032
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.579264760017395,
      "learning_rate": 0.0005971971558293476,
      "loss": 2.2531,
      "step": 10033
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6284900307655334,
      "learning_rate": 0.0005971965979483196,
      "loss": 2.6894,
      "step": 10034
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6786414384841919,
      "learning_rate": 0.0005971960400120371,
      "loss": 2.5875,
      "step": 10035
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.635779619216919,
      "learning_rate": 0.0005971954820205002,
      "loss": 2.5478,
      "step": 10036
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.604235053062439,
      "learning_rate": 0.0005971949239737092,
      "loss": 2.2783,
      "step": 10037
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6290302276611328,
      "learning_rate": 0.0005971943658716638,
      "loss": 2.5433,
      "step": 10038
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6793655157089233,
      "learning_rate": 0.0005971938077143646,
      "loss": 2.3532,
      "step": 10039
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.5649346709251404,
      "learning_rate": 0.0005971932495018113,
      "loss": 2.3053,
      "step": 10040
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6284574270248413,
      "learning_rate": 0.0005971926912340043,
      "loss": 2.2883,
      "step": 10041
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.631000280380249,
      "learning_rate": 0.0005971921329109435,
      "loss": 2.4976,
      "step": 10042
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6271595358848572,
      "learning_rate": 0.000597191574532629,
      "loss": 2.4537,
      "step": 10043
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.7349210977554321,
      "learning_rate": 0.0005971910160990611,
      "loss": 2.3572,
      "step": 10044
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.5879079699516296,
      "learning_rate": 0.0005971904576102397,
      "loss": 2.471,
      "step": 10045
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.664365828037262,
      "learning_rate": 0.0005971898990661651,
      "loss": 2.4187,
      "step": 10046
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6204824447631836,
      "learning_rate": 0.000597189340466837,
      "loss": 2.4306,
      "step": 10047
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6495524048805237,
      "learning_rate": 0.000597188781812256,
      "loss": 2.4132,
      "step": 10048
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.8536451458930969,
      "learning_rate": 0.0005971882231024221,
      "loss": 2.4337,
      "step": 10049
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.7039238214492798,
      "learning_rate": 0.0005971876643373351,
      "loss": 2.275,
      "step": 10050
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.754486083984375,
      "learning_rate": 0.0005971871055169954,
      "loss": 2.5781,
      "step": 10051
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.663038432598114,
      "learning_rate": 0.000597186546641403,
      "loss": 2.456,
      "step": 10052
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.7098893523216248,
      "learning_rate": 0.0005971859877105581,
      "loss": 2.53,
      "step": 10053
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.7535934448242188,
      "learning_rate": 0.0005971854287244605,
      "loss": 2.6604,
      "step": 10054
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6780107617378235,
      "learning_rate": 0.0005971848696831106,
      "loss": 2.5341,
      "step": 10055
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6228988766670227,
      "learning_rate": 0.0005971843105865085,
      "loss": 2.6478,
      "step": 10056
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.865201473236084,
      "learning_rate": 0.0005971837514346541,
      "loss": 2.1988,
      "step": 10057
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6104435920715332,
      "learning_rate": 0.0005971831922275477,
      "loss": 2.4346,
      "step": 10058
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.7024231553077698,
      "learning_rate": 0.0005971826329651895,
      "loss": 2.494,
      "step": 10059
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.63327956199646,
      "learning_rate": 0.0005971820736475793,
      "loss": 2.6693,
      "step": 10060
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6808226108551025,
      "learning_rate": 0.0005971815142747172,
      "loss": 2.3542,
      "step": 10061
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6187642216682434,
      "learning_rate": 0.0005971809548466037,
      "loss": 2.2341,
      "step": 10062
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.5883192420005798,
      "learning_rate": 0.0005971803953632385,
      "loss": 2.4714,
      "step": 10063
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6970887184143066,
      "learning_rate": 0.000597179835824622,
      "loss": 2.5279,
      "step": 10064
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6765530705451965,
      "learning_rate": 0.0005971792762307539,
      "loss": 2.5896,
      "step": 10065
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.655146598815918,
      "learning_rate": 0.0005971787165816347,
      "loss": 2.4545,
      "step": 10066
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6039657592773438,
      "learning_rate": 0.0005971781568772644,
      "loss": 2.391,
      "step": 10067
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6998122334480286,
      "learning_rate": 0.000597177597117643,
      "loss": 2.5347,
      "step": 10068
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6676484942436218,
      "learning_rate": 0.0005971770373027707,
      "loss": 2.6634,
      "step": 10069
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6353152394294739,
      "learning_rate": 0.0005971764774326476,
      "loss": 2.5414,
      "step": 10070
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.8457022905349731,
      "learning_rate": 0.0005971759175072739,
      "loss": 2.6159,
      "step": 10071
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.8490616083145142,
      "learning_rate": 0.0005971753575266494,
      "loss": 2.3707,
      "step": 10072
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6485710740089417,
      "learning_rate": 0.0005971747974907744,
      "loss": 2.4466,
      "step": 10073
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6917357444763184,
      "learning_rate": 0.0005971742373996491,
      "loss": 2.5663,
      "step": 10074
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.9136384129524231,
      "learning_rate": 0.0005971736772532734,
      "loss": 2.3555,
      "step": 10075
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6644141674041748,
      "learning_rate": 0.0005971731170516477,
      "loss": 2.2235,
      "step": 10076
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.622124433517456,
      "learning_rate": 0.0005971725567947717,
      "loss": 2.4388,
      "step": 10077
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.8265409469604492,
      "learning_rate": 0.0005971719964826458,
      "loss": 2.3959,
      "step": 10078
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.7525562047958374,
      "learning_rate": 0.00059717143611527,
      "loss": 2.5432,
      "step": 10079
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6544646620750427,
      "learning_rate": 0.0005971708756926443,
      "loss": 2.5917,
      "step": 10080
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6933104395866394,
      "learning_rate": 0.0005971703152147692,
      "loss": 2.3761,
      "step": 10081
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.8131663203239441,
      "learning_rate": 0.0005971697546816443,
      "loss": 2.5258,
      "step": 10082
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.715053379535675,
      "learning_rate": 0.0005971691940932699,
      "loss": 2.3482,
      "step": 10083
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6266768574714661,
      "learning_rate": 0.0005971686334496463,
      "loss": 2.6692,
      "step": 10084
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.8863802552223206,
      "learning_rate": 0.0005971680727507734,
      "loss": 2.3442,
      "step": 10085
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6160505414009094,
      "learning_rate": 0.0005971675119966514,
      "loss": 2.5591,
      "step": 10086
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.5848363041877747,
      "learning_rate": 0.0005971669511872802,
      "loss": 2.541,
      "step": 10087
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6920766830444336,
      "learning_rate": 0.0005971663903226601,
      "loss": 2.4805,
      "step": 10088
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6416361927986145,
      "learning_rate": 0.0005971658294027913,
      "loss": 2.4427,
      "step": 10089
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.8023493885993958,
      "learning_rate": 0.0005971652684276736,
      "loss": 2.4832,
      "step": 10090
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6148412227630615,
      "learning_rate": 0.0005971647073973073,
      "loss": 2.3562,
      "step": 10091
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6727737784385681,
      "learning_rate": 0.0005971641463116925,
      "loss": 2.6373,
      "step": 10092
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.7144609093666077,
      "learning_rate": 0.0005971635851708293,
      "loss": 2.6273,
      "step": 10093
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.5788330435752869,
      "learning_rate": 0.0005971630239747177,
      "loss": 2.3637,
      "step": 10094
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.7163341045379639,
      "learning_rate": 0.0005971624627233579,
      "loss": 2.4399,
      "step": 10095
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6946975588798523,
      "learning_rate": 0.0005971619014167501,
      "loss": 2.1854,
      "step": 10096
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.65666264295578,
      "learning_rate": 0.0005971613400548942,
      "loss": 2.4946,
      "step": 10097
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6659119129180908,
      "learning_rate": 0.0005971607786377903,
      "loss": 2.5268,
      "step": 10098
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6291908025741577,
      "learning_rate": 0.0005971602171654389,
      "loss": 2.553,
      "step": 10099
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.7760518193244934,
      "learning_rate": 0.0005971596556378395,
      "loss": 2.2891,
      "step": 10100
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6756052374839783,
      "learning_rate": 0.0005971590940549926,
      "loss": 2.4429,
      "step": 10101
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6352217197418213,
      "learning_rate": 0.0005971585324168983,
      "loss": 2.6731,
      "step": 10102
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.7441098093986511,
      "learning_rate": 0.0005971579707235566,
      "loss": 2.5264,
      "step": 10103
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6358941793441772,
      "learning_rate": 0.0005971574089749676,
      "loss": 2.457,
      "step": 10104
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6013880372047424,
      "learning_rate": 0.0005971568471711314,
      "loss": 2.2136,
      "step": 10105
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6793437600135803,
      "learning_rate": 0.0005971562853120481,
      "loss": 2.3981,
      "step": 10106
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6114166378974915,
      "learning_rate": 0.0005971557233977179,
      "loss": 2.318,
      "step": 10107
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6328495144844055,
      "learning_rate": 0.0005971551614281409,
      "loss": 2.4516,
      "step": 10108
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6453105807304382,
      "learning_rate": 0.000597154599403317,
      "loss": 2.4607,
      "step": 10109
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6327193975448608,
      "learning_rate": 0.0005971540373232465,
      "loss": 2.3893,
      "step": 10110
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6094194054603577,
      "learning_rate": 0.0005971534751879295,
      "loss": 2.4847,
      "step": 10111
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6468018889427185,
      "learning_rate": 0.0005971529129973661,
      "loss": 2.6599,
      "step": 10112
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6784260272979736,
      "learning_rate": 0.0005971523507515563,
      "loss": 2.3485,
      "step": 10113
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.5972556471824646,
      "learning_rate": 0.0005971517884505002,
      "loss": 2.5065,
      "step": 10114
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.5889819860458374,
      "learning_rate": 0.0005971512260941981,
      "loss": 2.4289,
      "step": 10115
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.7480442523956299,
      "learning_rate": 0.0005971506636826499,
      "loss": 2.6917,
      "step": 10116
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.761374831199646,
      "learning_rate": 0.0005971501012158559,
      "loss": 2.458,
      "step": 10117
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.7133693695068359,
      "learning_rate": 0.0005971495386938158,
      "loss": 2.5133,
      "step": 10118
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.8857423663139343,
      "learning_rate": 0.0005971489761165303,
      "loss": 2.5118,
      "step": 10119
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6766601800918579,
      "learning_rate": 0.0005971484134839991,
      "loss": 2.6946,
      "step": 10120
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.8025092482566833,
      "learning_rate": 0.0005971478507962223,
      "loss": 2.4065,
      "step": 10121
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.7188072800636292,
      "learning_rate": 0.0005971472880532002,
      "loss": 2.6349,
      "step": 10122
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6239507794380188,
      "learning_rate": 0.0005971467252549328,
      "loss": 2.3667,
      "step": 10123
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.5810233354568481,
      "learning_rate": 0.0005971461624014202,
      "loss": 2.5673,
      "step": 10124
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6548093557357788,
      "learning_rate": 0.0005971455994926626,
      "loss": 2.6273,
      "step": 10125
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6788854002952576,
      "learning_rate": 0.00059714503652866,
      "loss": 2.4275,
      "step": 10126
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.5797258019447327,
      "learning_rate": 0.0005971444735094124,
      "loss": 2.2543,
      "step": 10127
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.7490060329437256,
      "learning_rate": 0.0005971439104349201,
      "loss": 2.512,
      "step": 10128
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6819316744804382,
      "learning_rate": 0.0005971433473051832,
      "loss": 2.5113,
      "step": 10129
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6505802869796753,
      "learning_rate": 0.0005971427841202018,
      "loss": 2.3797,
      "step": 10130
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6344186663627625,
      "learning_rate": 0.0005971422208799758,
      "loss": 2.4053,
      "step": 10131
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.9100697636604309,
      "learning_rate": 0.0005971416575845054,
      "loss": 2.6058,
      "step": 10132
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.7103368639945984,
      "learning_rate": 0.0005971410942337909,
      "loss": 2.4332,
      "step": 10133
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.9970961213111877,
      "learning_rate": 0.0005971405308278322,
      "loss": 2.319,
      "step": 10134
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.64005047082901,
      "learning_rate": 0.0005971399673666295,
      "loss": 2.5766,
      "step": 10135
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.7840222120285034,
      "learning_rate": 0.0005971394038501829,
      "loss": 2.4261,
      "step": 10136
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.7849117517471313,
      "learning_rate": 0.0005971388402784924,
      "loss": 2.3883,
      "step": 10137
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6233367919921875,
      "learning_rate": 0.0005971382766515582,
      "loss": 2.2707,
      "step": 10138
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.7423081398010254,
      "learning_rate": 0.0005971377129693804,
      "loss": 2.6723,
      "step": 10139
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6892527341842651,
      "learning_rate": 0.0005971371492319591,
      "loss": 2.4814,
      "step": 10140
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6168640851974487,
      "learning_rate": 0.0005971365854392943,
      "loss": 2.5471,
      "step": 10141
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.7326837182044983,
      "learning_rate": 0.0005971360215913864,
      "loss": 2.5173,
      "step": 10142
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.7129960060119629,
      "learning_rate": 0.0005971354576882351,
      "loss": 2.3967,
      "step": 10143
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6409669518470764,
      "learning_rate": 0.0005971348937298409,
      "loss": 2.565,
      "step": 10144
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6795448660850525,
      "learning_rate": 0.0005971343297162036,
      "loss": 2.7269,
      "step": 10145
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6335002183914185,
      "learning_rate": 0.0005971337656473233,
      "loss": 2.3571,
      "step": 10146
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.688497006893158,
      "learning_rate": 0.0005971332015232005,
      "loss": 2.598,
      "step": 10147
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.7046929597854614,
      "learning_rate": 0.0005971326373438348,
      "loss": 2.2282,
      "step": 10148
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.5807036757469177,
      "learning_rate": 0.0005971320731092266,
      "loss": 2.3433,
      "step": 10149
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.654331624507904,
      "learning_rate": 0.000597131508819376,
      "loss": 2.4393,
      "step": 10150
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6101180911064148,
      "learning_rate": 0.0005971309444742829,
      "loss": 2.4854,
      "step": 10151
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6294077634811401,
      "learning_rate": 0.0005971303800739478,
      "loss": 2.4962,
      "step": 10152
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.5818518400192261,
      "learning_rate": 0.0005971298156183703,
      "loss": 2.2767,
      "step": 10153
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6301205158233643,
      "learning_rate": 0.0005971292511075509,
      "loss": 2.5193,
      "step": 10154
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6323294639587402,
      "learning_rate": 0.0005971286865414896,
      "loss": 2.4914,
      "step": 10155
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.5727779865264893,
      "learning_rate": 0.0005971281219201863,
      "loss": 2.5388,
      "step": 10156
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6244205236434937,
      "learning_rate": 0.0005971275572436415,
      "loss": 2.7338,
      "step": 10157
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6059735417366028,
      "learning_rate": 0.0005971269925118548,
      "loss": 2.5618,
      "step": 10158
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6327356100082397,
      "learning_rate": 0.0005971264277248268,
      "loss": 2.6993,
      "step": 10159
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.5575670003890991,
      "learning_rate": 0.0005971258628825573,
      "loss": 2.596,
      "step": 10160
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.7054449915885925,
      "learning_rate": 0.0005971252979850466,
      "loss": 2.3219,
      "step": 10161
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6602901816368103,
      "learning_rate": 0.0005971247330322945,
      "loss": 2.5591,
      "step": 10162
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6296201944351196,
      "learning_rate": 0.0005971241680243014,
      "loss": 2.317,
      "step": 10163
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.7290756702423096,
      "learning_rate": 0.0005971236029610673,
      "loss": 2.4455,
      "step": 10164
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.641910970211029,
      "learning_rate": 0.0005971230378425924,
      "loss": 2.3593,
      "step": 10165
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6067615151405334,
      "learning_rate": 0.0005971224726688765,
      "loss": 2.2069,
      "step": 10166
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6472363471984863,
      "learning_rate": 0.0005971219074399201,
      "loss": 2.55,
      "step": 10167
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.7224085330963135,
      "learning_rate": 0.0005971213421557232,
      "loss": 2.6318,
      "step": 10168
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.631848156452179,
      "learning_rate": 0.0005971207768162857,
      "loss": 2.5802,
      "step": 10169
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.5602810382843018,
      "learning_rate": 0.000597120211421608,
      "loss": 2.5132,
      "step": 10170
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.7134763598442078,
      "learning_rate": 0.0005971196459716899,
      "loss": 2.508,
      "step": 10171
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6376556754112244,
      "learning_rate": 0.0005971190804665316,
      "loss": 2.4562,
      "step": 10172
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6554585695266724,
      "learning_rate": 0.0005971185149061334,
      "loss": 2.5122,
      "step": 10173
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6897973418235779,
      "learning_rate": 0.0005971179492904952,
      "loss": 2.3302,
      "step": 10174
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.9172040820121765,
      "learning_rate": 0.0005971173836196171,
      "loss": 2.5426,
      "step": 10175
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.5972715020179749,
      "learning_rate": 0.0005971168178934993,
      "loss": 2.3088,
      "step": 10176
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.7238768339157104,
      "learning_rate": 0.0005971162521121419,
      "loss": 2.5791,
      "step": 10177
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6017717719078064,
      "learning_rate": 0.0005971156862755449,
      "loss": 2.3786,
      "step": 10178
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.5353584885597229,
      "learning_rate": 0.0005971151203837086,
      "loss": 2.2751,
      "step": 10179
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.7369135022163391,
      "learning_rate": 0.0005971145544366329,
      "loss": 2.6405,
      "step": 10180
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6173352003097534,
      "learning_rate": 0.000597113988434318,
      "loss": 2.5957,
      "step": 10181
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6642436981201172,
      "learning_rate": 0.0005971134223767642,
      "loss": 2.2784,
      "step": 10182
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6493363380432129,
      "learning_rate": 0.0005971128562639712,
      "loss": 2.4528,
      "step": 10183
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6499666571617126,
      "learning_rate": 0.0005971122900959394,
      "loss": 2.5221,
      "step": 10184
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.7828931212425232,
      "learning_rate": 0.0005971117238726687,
      "loss": 2.3371,
      "step": 10185
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6671351790428162,
      "learning_rate": 0.0005971111575941595,
      "loss": 2.4096,
      "step": 10186
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6186683177947998,
      "learning_rate": 0.0005971105912604116,
      "loss": 2.4115,
      "step": 10187
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.5940440893173218,
      "learning_rate": 0.0005971100248714252,
      "loss": 2.3796,
      "step": 10188
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6952428221702576,
      "learning_rate": 0.0005971094584272006,
      "loss": 2.4477,
      "step": 10189
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6174731254577637,
      "learning_rate": 0.0005971088919277376,
      "loss": 2.4832,
      "step": 10190
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.7356269359588623,
      "learning_rate": 0.0005971083253730365,
      "loss": 2.6367,
      "step": 10191
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6565449833869934,
      "learning_rate": 0.0005971077587630973,
      "loss": 2.52,
      "step": 10192
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6645349264144897,
      "learning_rate": 0.0005971071920979203,
      "loss": 2.3711,
      "step": 10193
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6232324838638306,
      "learning_rate": 0.0005971066253775053,
      "loss": 2.9105,
      "step": 10194
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6124207973480225,
      "learning_rate": 0.0005971060586018527,
      "loss": 2.5471,
      "step": 10195
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6064692139625549,
      "learning_rate": 0.0005971054917709624,
      "loss": 2.2661,
      "step": 10196
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6046184301376343,
      "learning_rate": 0.0005971049248848346,
      "loss": 2.4611,
      "step": 10197
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.5584084987640381,
      "learning_rate": 0.0005971043579434695,
      "loss": 2.5277,
      "step": 10198
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.5927500128746033,
      "learning_rate": 0.0005971037909468668,
      "loss": 2.4432,
      "step": 10199
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.7894229292869568,
      "learning_rate": 0.0005971032238950272,
      "loss": 2.5931,
      "step": 10200
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6209644079208374,
      "learning_rate": 0.0005971026567879503,
      "loss": 2.5297,
      "step": 10201
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.7390706539154053,
      "learning_rate": 0.0005971020896256365,
      "loss": 2.5603,
      "step": 10202
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6931877136230469,
      "learning_rate": 0.0005971015224080858,
      "loss": 2.7133,
      "step": 10203
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6637492775917053,
      "learning_rate": 0.0005971009551352984,
      "loss": 2.5629,
      "step": 10204
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6417529582977295,
      "learning_rate": 0.0005971003878072742,
      "loss": 2.6358,
      "step": 10205
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.7177567481994629,
      "learning_rate": 0.0005970998204240134,
      "loss": 2.5323,
      "step": 10206
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.7369118928909302,
      "learning_rate": 0.0005970992529855162,
      "loss": 2.6195,
      "step": 10207
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6223832368850708,
      "learning_rate": 0.0005970986854917827,
      "loss": 2.6276,
      "step": 10208
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.9018802642822266,
      "learning_rate": 0.0005970981179428129,
      "loss": 2.7046,
      "step": 10209
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6562245488166809,
      "learning_rate": 0.0005970975503386069,
      "loss": 2.4251,
      "step": 10210
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6463072299957275,
      "learning_rate": 0.0005970969826791649,
      "loss": 2.5435,
      "step": 10211
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6856294870376587,
      "learning_rate": 0.0005970964149644869,
      "loss": 2.4262,
      "step": 10212
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6419367790222168,
      "learning_rate": 0.0005970958471945732,
      "loss": 2.6159,
      "step": 10213
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.7969492673873901,
      "learning_rate": 0.0005970952793694237,
      "loss": 2.6217,
      "step": 10214
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6284271478652954,
      "learning_rate": 0.0005970947114890385,
      "loss": 2.4588,
      "step": 10215
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.7919228076934814,
      "learning_rate": 0.0005970941435534179,
      "loss": 2.508,
      "step": 10216
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.7113685607910156,
      "learning_rate": 0.000597093575562562,
      "loss": 2.4443,
      "step": 10217
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.650775671005249,
      "learning_rate": 0.0005970930075164706,
      "loss": 2.4891,
      "step": 10218
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.8814980983734131,
      "learning_rate": 0.000597092439415144,
      "loss": 2.6742,
      "step": 10219
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6147571206092834,
      "learning_rate": 0.0005970918712585824,
      "loss": 2.6203,
      "step": 10220
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6722485423088074,
      "learning_rate": 0.0005970913030467857,
      "loss": 2.4949,
      "step": 10221
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.7558441758155823,
      "learning_rate": 0.0005970907347797543,
      "loss": 2.5302,
      "step": 10222
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6932177543640137,
      "learning_rate": 0.000597090166457488,
      "loss": 2.6039,
      "step": 10223
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.5812422633171082,
      "learning_rate": 0.0005970895980799871,
      "loss": 2.4291,
      "step": 10224
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.5984801054000854,
      "learning_rate": 0.0005970890296472515,
      "loss": 2.3507,
      "step": 10225
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6041004061698914,
      "learning_rate": 0.0005970884611592815,
      "loss": 2.5083,
      "step": 10226
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6366809010505676,
      "learning_rate": 0.0005970878926160772,
      "loss": 2.483,
      "step": 10227
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6693763136863708,
      "learning_rate": 0.0005970873240176386,
      "loss": 2.8137,
      "step": 10228
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6122682690620422,
      "learning_rate": 0.000597086755363966,
      "loss": 2.5258,
      "step": 10229
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.709674596786499,
      "learning_rate": 0.0005970861866550592,
      "loss": 2.6614,
      "step": 10230
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6832488775253296,
      "learning_rate": 0.0005970856178909186,
      "loss": 2.3724,
      "step": 10231
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.7071173787117004,
      "learning_rate": 0.000597085049071544,
      "loss": 2.3942,
      "step": 10232
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6302321553230286,
      "learning_rate": 0.0005970844801969358,
      "loss": 2.5365,
      "step": 10233
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6459634900093079,
      "learning_rate": 0.0005970839112670939,
      "loss": 2.6247,
      "step": 10234
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6197597980499268,
      "learning_rate": 0.0005970833422820186,
      "loss": 2.4495,
      "step": 10235
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6743104457855225,
      "learning_rate": 0.0005970827732417099,
      "loss": 2.6247,
      "step": 10236
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.5833186507225037,
      "learning_rate": 0.0005970822041461678,
      "loss": 2.1535,
      "step": 10237
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6489455699920654,
      "learning_rate": 0.0005970816349953927,
      "loss": 2.5789,
      "step": 10238
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6412034034729004,
      "learning_rate": 0.0005970810657893843,
      "loss": 2.2734,
      "step": 10239
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6550008058547974,
      "learning_rate": 0.000597080496528143,
      "loss": 2.1089,
      "step": 10240
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6099883317947388,
      "learning_rate": 0.0005970799272116689,
      "loss": 2.3433,
      "step": 10241
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.8132157921791077,
      "learning_rate": 0.0005970793578399619,
      "loss": 2.5384,
      "step": 10242
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.720879316329956,
      "learning_rate": 0.0005970787884130223,
      "loss": 2.4689,
      "step": 10243
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6454402208328247,
      "learning_rate": 0.0005970782189308502,
      "loss": 2.3409,
      "step": 10244
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.9674580097198486,
      "learning_rate": 0.0005970776493934457,
      "loss": 2.3022,
      "step": 10245
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.7057787775993347,
      "learning_rate": 0.0005970770798008087,
      "loss": 2.4802,
      "step": 10246
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6656293869018555,
      "learning_rate": 0.0005970765101529395,
      "loss": 2.8255,
      "step": 10247
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6558194160461426,
      "learning_rate": 0.0005970759404498382,
      "loss": 2.5093,
      "step": 10248
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.7458356618881226,
      "learning_rate": 0.0005970753706915049,
      "loss": 2.7705,
      "step": 10249
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6238061189651489,
      "learning_rate": 0.0005970748008779397,
      "loss": 2.3805,
      "step": 10250
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6365326046943665,
      "learning_rate": 0.0005970742310091425,
      "loss": 2.7762,
      "step": 10251
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.8331921100616455,
      "learning_rate": 0.0005970736610851138,
      "loss": 2.4726,
      "step": 10252
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6611765623092651,
      "learning_rate": 0.0005970730911058535,
      "loss": 2.795,
      "step": 10253
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.5999999642372131,
      "learning_rate": 0.0005970725210713617,
      "loss": 2.4512,
      "step": 10254
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.7403091788291931,
      "learning_rate": 0.0005970719509816383,
      "loss": 2.3877,
      "step": 10255
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6946302056312561,
      "learning_rate": 0.0005970713808366838,
      "loss": 2.5154,
      "step": 10256
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6288330554962158,
      "learning_rate": 0.0005970708106364981,
      "loss": 2.424,
      "step": 10257
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.7173245549201965,
      "learning_rate": 0.0005970702403810814,
      "loss": 2.3652,
      "step": 10258
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6306650042533875,
      "learning_rate": 0.0005970696700704336,
      "loss": 2.3413,
      "step": 10259
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6208289861679077,
      "learning_rate": 0.0005970690997045549,
      "loss": 2.3633,
      "step": 10260
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6324766278266907,
      "learning_rate": 0.0005970685292834456,
      "loss": 2.4106,
      "step": 10261
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.7532114386558533,
      "learning_rate": 0.0005970679588071056,
      "loss": 2.649,
      "step": 10262
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.7053565979003906,
      "learning_rate": 0.000597067388275535,
      "loss": 2.501,
      "step": 10263
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6523607969284058,
      "learning_rate": 0.0005970668176887342,
      "loss": 2.3188,
      "step": 10264
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6782897114753723,
      "learning_rate": 0.0005970662470467027,
      "loss": 2.547,
      "step": 10265
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.8131237626075745,
      "learning_rate": 0.0005970656763494411,
      "loss": 2.4822,
      "step": 10266
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6874427199363708,
      "learning_rate": 0.0005970651055969495,
      "loss": 2.4616,
      "step": 10267
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6127464771270752,
      "learning_rate": 0.0005970645347892279,
      "loss": 2.4892,
      "step": 10268
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.7143076658248901,
      "learning_rate": 0.0005970639639262763,
      "loss": 2.4257,
      "step": 10269
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.656143307685852,
      "learning_rate": 0.0005970633930080949,
      "loss": 2.6316,
      "step": 10270
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6968182325363159,
      "learning_rate": 0.0005970628220346839,
      "loss": 2.6319,
      "step": 10271
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6540284752845764,
      "learning_rate": 0.0005970622510060432,
      "loss": 2.6532,
      "step": 10272
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.7133622169494629,
      "learning_rate": 0.0005970616799221731,
      "loss": 2.4862,
      "step": 10273
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.7608844041824341,
      "learning_rate": 0.0005970611087830737,
      "loss": 2.1663,
      "step": 10274
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6602775454521179,
      "learning_rate": 0.0005970605375887448,
      "loss": 2.2774,
      "step": 10275
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6023558378219604,
      "learning_rate": 0.0005970599663391869,
      "loss": 2.6221,
      "step": 10276
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.8052432537078857,
      "learning_rate": 0.0005970593950343999,
      "loss": 2.3245,
      "step": 10277
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.7006747722625732,
      "learning_rate": 0.000597058823674384,
      "loss": 2.7564,
      "step": 10278
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6181423664093018,
      "learning_rate": 0.0005970582522591393,
      "loss": 2.295,
      "step": 10279
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6205700039863586,
      "learning_rate": 0.0005970576807886657,
      "loss": 2.344,
      "step": 10280
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.820746660232544,
      "learning_rate": 0.0005970571092629636,
      "loss": 2.2997,
      "step": 10281
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.742455780506134,
      "learning_rate": 0.0005970565376820329,
      "loss": 2.1259,
      "step": 10282
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6148857474327087,
      "learning_rate": 0.0005970559660458738,
      "loss": 2.659,
      "step": 10283
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.8827542066574097,
      "learning_rate": 0.0005970553943544864,
      "loss": 2.3231,
      "step": 10284
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6340488791465759,
      "learning_rate": 0.0005970548226078709,
      "loss": 2.5882,
      "step": 10285
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6392003297805786,
      "learning_rate": 0.0005970542508060271,
      "loss": 2.4855,
      "step": 10286
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.774078905582428,
      "learning_rate": 0.0005970536789489554,
      "loss": 2.5697,
      "step": 10287
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.7577046751976013,
      "learning_rate": 0.0005970531070366559,
      "loss": 2.3766,
      "step": 10288
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6931131482124329,
      "learning_rate": 0.0005970525350691285,
      "loss": 2.3683,
      "step": 10289
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.8683518171310425,
      "learning_rate": 0.0005970519630463735,
      "loss": 2.514,
      "step": 10290
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.7051246762275696,
      "learning_rate": 0.0005970513909683908,
      "loss": 2.2526,
      "step": 10291
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6303568482398987,
      "learning_rate": 0.0005970508188351808,
      "loss": 2.6485,
      "step": 10292
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6212684512138367,
      "learning_rate": 0.0005970502466467434,
      "loss": 2.4563,
      "step": 10293
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6390775442123413,
      "learning_rate": 0.0005970496744030787,
      "loss": 2.4577,
      "step": 10294
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6731254458427429,
      "learning_rate": 0.0005970491021041869,
      "loss": 2.824,
      "step": 10295
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.863357424736023,
      "learning_rate": 0.0005970485297500681,
      "loss": 2.3697,
      "step": 10296
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6027673482894897,
      "learning_rate": 0.0005970479573407224,
      "loss": 2.5668,
      "step": 10297
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.7696351408958435,
      "learning_rate": 0.0005970473848761497,
      "loss": 2.2032,
      "step": 10298
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6621620059013367,
      "learning_rate": 0.0005970468123563504,
      "loss": 2.5544,
      "step": 10299
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6112158894538879,
      "learning_rate": 0.0005970462397813245,
      "loss": 2.4553,
      "step": 10300
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6385949850082397,
      "learning_rate": 0.0005970456671510721,
      "loss": 2.4506,
      "step": 10301
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6750091910362244,
      "learning_rate": 0.0005970450944655933,
      "loss": 2.4576,
      "step": 10302
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.5840277671813965,
      "learning_rate": 0.0005970445217248881,
      "loss": 2.6112,
      "step": 10303
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.7047901153564453,
      "learning_rate": 0.0005970439489289569,
      "loss": 2.5431,
      "step": 10304
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6011863946914673,
      "learning_rate": 0.0005970433760777996,
      "loss": 2.2821,
      "step": 10305
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6399878859519958,
      "learning_rate": 0.0005970428031714162,
      "loss": 2.2377,
      "step": 10306
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.7329175472259521,
      "learning_rate": 0.0005970422302098069,
      "loss": 2.3176,
      "step": 10307
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6084122657775879,
      "learning_rate": 0.000597041657192972,
      "loss": 2.4946,
      "step": 10308
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6899876594543457,
      "learning_rate": 0.0005970410841209115,
      "loss": 2.4819,
      "step": 10309
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6523743867874146,
      "learning_rate": 0.0005970405109936253,
      "loss": 2.2757,
      "step": 10310
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6389380693435669,
      "learning_rate": 0.0005970399378111137,
      "loss": 2.6439,
      "step": 10311
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.5957863926887512,
      "learning_rate": 0.0005970393645733766,
      "loss": 2.3868,
      "step": 10312
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.5596105456352234,
      "learning_rate": 0.0005970387912804146,
      "loss": 2.5691,
      "step": 10313
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6769130229949951,
      "learning_rate": 0.0005970382179322273,
      "loss": 2.3108,
      "step": 10314
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6146937012672424,
      "learning_rate": 0.000597037644528815,
      "loss": 2.4585,
      "step": 10315
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6118184328079224,
      "learning_rate": 0.0005970370710701779,
      "loss": 2.4459,
      "step": 10316
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.5952894687652588,
      "learning_rate": 0.0005970364975563158,
      "loss": 2.4427,
      "step": 10317
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.627499520778656,
      "learning_rate": 0.0005970359239872292,
      "loss": 2.4268,
      "step": 10318
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.57649827003479,
      "learning_rate": 0.0005970353503629179,
      "loss": 2.5617,
      "step": 10319
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6565335988998413,
      "learning_rate": 0.0005970347766833822,
      "loss": 2.6574,
      "step": 10320
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6621333956718445,
      "learning_rate": 0.0005970342029486221,
      "loss": 2.3721,
      "step": 10321
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6373052000999451,
      "learning_rate": 0.0005970336291586377,
      "loss": 2.4046,
      "step": 10322
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6372042298316956,
      "learning_rate": 0.0005970330553134291,
      "loss": 2.4655,
      "step": 10323
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.5832250714302063,
      "learning_rate": 0.0005970324814129967,
      "loss": 2.414,
      "step": 10324
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6749229431152344,
      "learning_rate": 0.00059703190745734,
      "loss": 2.4822,
      "step": 10325
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.5815304517745972,
      "learning_rate": 0.0005970313334464596,
      "loss": 2.418,
      "step": 10326
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6310195922851562,
      "learning_rate": 0.0005970307593803556,
      "loss": 2.408,
      "step": 10327
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.5638130307197571,
      "learning_rate": 0.0005970301852590278,
      "loss": 2.6085,
      "step": 10328
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6291355490684509,
      "learning_rate": 0.0005970296110824765,
      "loss": 2.7252,
      "step": 10329
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6338585615158081,
      "learning_rate": 0.0005970290368507019,
      "loss": 2.3317,
      "step": 10330
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.62285315990448,
      "learning_rate": 0.0005970284625637039,
      "loss": 2.5881,
      "step": 10331
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.0884976387023926,
      "learning_rate": 0.0005970278882214828,
      "loss": 2.4968,
      "step": 10332
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6766538619995117,
      "learning_rate": 0.0005970273138240386,
      "loss": 2.7428,
      "step": 10333
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6554348468780518,
      "learning_rate": 0.0005970267393713713,
      "loss": 2.5865,
      "step": 10334
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6082080602645874,
      "learning_rate": 0.0005970261648634812,
      "loss": 2.5277,
      "step": 10335
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6288308501243591,
      "learning_rate": 0.0005970255903003683,
      "loss": 2.2553,
      "step": 10336
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6680644154548645,
      "learning_rate": 0.0005970250156820328,
      "loss": 2.6453,
      "step": 10337
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.7491670846939087,
      "learning_rate": 0.0005970244410084747,
      "loss": 2.4766,
      "step": 10338
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.5984532833099365,
      "learning_rate": 0.0005970238662796942,
      "loss": 2.4284,
      "step": 10339
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6214187741279602,
      "learning_rate": 0.0005970232914956912,
      "loss": 2.4038,
      "step": 10340
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6026079058647156,
      "learning_rate": 0.0005970227166564661,
      "loss": 2.5033,
      "step": 10341
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6701863408088684,
      "learning_rate": 0.0005970221417620188,
      "loss": 2.532,
      "step": 10342
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6607012152671814,
      "learning_rate": 0.0005970215668123494,
      "loss": 2.5705,
      "step": 10343
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6013587713241577,
      "learning_rate": 0.0005970209918074583,
      "loss": 2.5431,
      "step": 10344
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6713538765907288,
      "learning_rate": 0.0005970204167473453,
      "loss": 2.2779,
      "step": 10345
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6400020122528076,
      "learning_rate": 0.0005970198416320106,
      "loss": 2.5673,
      "step": 10346
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.7557499408721924,
      "learning_rate": 0.0005970192664614542,
      "loss": 2.3651,
      "step": 10347
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6242865920066833,
      "learning_rate": 0.0005970186912356764,
      "loss": 2.3972,
      "step": 10348
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6885284185409546,
      "learning_rate": 0.0005970181159546773,
      "loss": 2.3217,
      "step": 10349
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.8587239384651184,
      "learning_rate": 0.0005970175406184568,
      "loss": 2.5549,
      "step": 10350
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6700085401535034,
      "learning_rate": 0.0005970169652270152,
      "loss": 2.3127,
      "step": 10351
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6210494041442871,
      "learning_rate": 0.0005970163897803525,
      "loss": 2.4984,
      "step": 10352
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.5691633224487305,
      "learning_rate": 0.0005970158142784689,
      "loss": 2.5235,
      "step": 10353
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6706447005271912,
      "learning_rate": 0.0005970152387213644,
      "loss": 2.2862,
      "step": 10354
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.826450526714325,
      "learning_rate": 0.0005970146631090392,
      "loss": 2.5458,
      "step": 10355
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6935165524482727,
      "learning_rate": 0.0005970140874414934,
      "loss": 2.5326,
      "step": 10356
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6774663925170898,
      "learning_rate": 0.0005970135117187269,
      "loss": 2.3688,
      "step": 10357
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6451311707496643,
      "learning_rate": 0.0005970129359407401,
      "loss": 2.4052,
      "step": 10358
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.653525710105896,
      "learning_rate": 0.000597012360107533,
      "loss": 2.5813,
      "step": 10359
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6592279076576233,
      "learning_rate": 0.0005970117842191057,
      "loss": 2.5524,
      "step": 10360
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6379331350326538,
      "learning_rate": 0.0005970112082754583,
      "loss": 2.3425,
      "step": 10361
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.7074746489524841,
      "learning_rate": 0.0005970106322765909,
      "loss": 2.5336,
      "step": 10362
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6172309517860413,
      "learning_rate": 0.0005970100562225037,
      "loss": 2.4514,
      "step": 10363
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6245402693748474,
      "learning_rate": 0.0005970094801131966,
      "loss": 2.2302,
      "step": 10364
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6783676147460938,
      "learning_rate": 0.0005970089039486698,
      "loss": 2.6117,
      "step": 10365
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6356350183486938,
      "learning_rate": 0.0005970083277289235,
      "loss": 2.6052,
      "step": 10366
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6295722126960754,
      "learning_rate": 0.0005970077514539577,
      "loss": 2.4181,
      "step": 10367
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6720356345176697,
      "learning_rate": 0.0005970071751237728,
      "loss": 2.4666,
      "step": 10368
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.703113317489624,
      "learning_rate": 0.0005970065987383684,
      "loss": 2.5414,
      "step": 10369
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.7373449206352234,
      "learning_rate": 0.0005970060222977449,
      "loss": 2.3192,
      "step": 10370
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.643535852432251,
      "learning_rate": 0.0005970054458019024,
      "loss": 2.6729,
      "step": 10371
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.8439120054244995,
      "learning_rate": 0.000597004869250841,
      "loss": 2.3554,
      "step": 10372
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6222894787788391,
      "learning_rate": 0.0005970042926445608,
      "loss": 2.4904,
      "step": 10373
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.7418373227119446,
      "learning_rate": 0.0005970037159830618,
      "loss": 2.2098,
      "step": 10374
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6656388640403748,
      "learning_rate": 0.0005970031392663444,
      "loss": 2.4602,
      "step": 10375
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6016170382499695,
      "learning_rate": 0.0005970025624944083,
      "loss": 2.4173,
      "step": 10376
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.583328127861023,
      "learning_rate": 0.0005970019856672539,
      "loss": 2.3584,
      "step": 10377
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6511187553405762,
      "learning_rate": 0.0005970014087848813,
      "loss": 2.5901,
      "step": 10378
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6358575820922852,
      "learning_rate": 0.0005970008318472904,
      "loss": 2.5273,
      "step": 10379
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.9514508843421936,
      "learning_rate": 0.0005970002548544815,
      "loss": 2.3586,
      "step": 10380
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.7118512988090515,
      "learning_rate": 0.0005969996778064546,
      "loss": 2.5335,
      "step": 10381
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.7694718837738037,
      "learning_rate": 0.00059699910070321,
      "loss": 2.459,
      "step": 10382
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.594500720500946,
      "learning_rate": 0.0005969985235447474,
      "loss": 2.5821,
      "step": 10383
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.7491704225540161,
      "learning_rate": 0.0005969979463310674,
      "loss": 2.7656,
      "step": 10384
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.5684564113616943,
      "learning_rate": 0.0005969973690621699,
      "loss": 2.6263,
      "step": 10385
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6574798226356506,
      "learning_rate": 0.0005969967917380548,
      "loss": 2.1757,
      "step": 10386
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6617535352706909,
      "learning_rate": 0.0005969962143587225,
      "loss": 2.3546,
      "step": 10387
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.578450620174408,
      "learning_rate": 0.000596995636924173,
      "loss": 2.5832,
      "step": 10388
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6785446405410767,
      "learning_rate": 0.0005969950594344063,
      "loss": 2.3738,
      "step": 10389
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6657451391220093,
      "learning_rate": 0.0005969944818894228,
      "loss": 2.8522,
      "step": 10390
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.7006286978721619,
      "learning_rate": 0.0005969939042892222,
      "loss": 2.4731,
      "step": 10391
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.7971580028533936,
      "learning_rate": 0.0005969933266338049,
      "loss": 2.4489,
      "step": 10392
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.7520081996917725,
      "learning_rate": 0.0005969927489231711,
      "loss": 2.4849,
      "step": 10393
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6265853047370911,
      "learning_rate": 0.0005969921711573206,
      "loss": 2.3937,
      "step": 10394
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.7233743071556091,
      "learning_rate": 0.0005969915933362536,
      "loss": 2.5232,
      "step": 10395
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.711217999458313,
      "learning_rate": 0.0005969910154599703,
      "loss": 2.4715,
      "step": 10396
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6339811682701111,
      "learning_rate": 0.0005969904375284709,
      "loss": 2.6089,
      "step": 10397
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6471011638641357,
      "learning_rate": 0.0005969898595417552,
      "loss": 2.6188,
      "step": 10398
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.7474302649497986,
      "learning_rate": 0.0005969892814998235,
      "loss": 2.4423,
      "step": 10399
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6496654152870178,
      "learning_rate": 0.0005969887034026759,
      "loss": 2.5338,
      "step": 10400
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6352464556694031,
      "learning_rate": 0.0005969881252503125,
      "loss": 2.7151,
      "step": 10401
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.7035480737686157,
      "learning_rate": 0.0005969875470427335,
      "loss": 2.5794,
      "step": 10402
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.7318065166473389,
      "learning_rate": 0.0005969869687799387,
      "loss": 2.4203,
      "step": 10403
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.5705307126045227,
      "learning_rate": 0.0005969863904619286,
      "loss": 2.7387,
      "step": 10404
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6092009544372559,
      "learning_rate": 0.000596985812088703,
      "loss": 2.5328,
      "step": 10405
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6640744209289551,
      "learning_rate": 0.0005969852336602622,
      "loss": 2.4263,
      "step": 10406
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.7682480216026306,
      "learning_rate": 0.0005969846551766063,
      "loss": 2.208,
      "step": 10407
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6447153687477112,
      "learning_rate": 0.0005969840766377353,
      "loss": 2.6392,
      "step": 10408
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6887147426605225,
      "learning_rate": 0.0005969834980436493,
      "loss": 2.7928,
      "step": 10409
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.7516818046569824,
      "learning_rate": 0.0005969829193943484,
      "loss": 2.5155,
      "step": 10410
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6530569791793823,
      "learning_rate": 0.0005969823406898329,
      "loss": 2.4353,
      "step": 10411
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6584988832473755,
      "learning_rate": 0.0005969817619301028,
      "loss": 2.3874,
      "step": 10412
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6966056823730469,
      "learning_rate": 0.0005969811831151581,
      "loss": 2.4424,
      "step": 10413
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6535009741783142,
      "learning_rate": 0.0005969806042449991,
      "loss": 2.7356,
      "step": 10414
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6612852215766907,
      "learning_rate": 0.0005969800253196257,
      "loss": 2.6123,
      "step": 10415
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6241016387939453,
      "learning_rate": 0.0005969794463390382,
      "loss": 2.5116,
      "step": 10416
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6512463688850403,
      "learning_rate": 0.0005969788673032365,
      "loss": 2.602,
      "step": 10417
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6314390897750854,
      "learning_rate": 0.0005969782882122208,
      "loss": 2.5193,
      "step": 10418
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6396954655647278,
      "learning_rate": 0.0005969777090659912,
      "loss": 2.5042,
      "step": 10419
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6442422866821289,
      "learning_rate": 0.000596977129864548,
      "loss": 2.4482,
      "step": 10420
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.699729323387146,
      "learning_rate": 0.0005969765506078911,
      "loss": 2.4932,
      "step": 10421
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6128555536270142,
      "learning_rate": 0.0005969759712960206,
      "loss": 2.219,
      "step": 10422
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.7826679944992065,
      "learning_rate": 0.0005969753919289367,
      "loss": 2.4268,
      "step": 10423
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.7611706256866455,
      "learning_rate": 0.0005969748125066395,
      "loss": 2.6264,
      "step": 10424
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6810503005981445,
      "learning_rate": 0.000596974233029129,
      "loss": 2.4583,
      "step": 10425
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.7247908115386963,
      "learning_rate": 0.0005969736534964055,
      "loss": 2.3533,
      "step": 10426
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6721895933151245,
      "learning_rate": 0.0005969730739084689,
      "loss": 2.5637,
      "step": 10427
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6111849546432495,
      "learning_rate": 0.0005969724942653194,
      "loss": 2.3409,
      "step": 10428
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6434260606765747,
      "learning_rate": 0.0005969719145669571,
      "loss": 2.4419,
      "step": 10429
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6582525372505188,
      "learning_rate": 0.0005969713348133822,
      "loss": 2.4662,
      "step": 10430
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6152166724205017,
      "learning_rate": 0.0005969707550045946,
      "loss": 2.6542,
      "step": 10431
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.609237790107727,
      "learning_rate": 0.0005969701751405946,
      "loss": 2.4574,
      "step": 10432
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.5848544239997864,
      "learning_rate": 0.0005969695952213823,
      "loss": 2.3874,
      "step": 10433
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6344856023788452,
      "learning_rate": 0.0005969690152469576,
      "loss": 2.4859,
      "step": 10434
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.7018263936042786,
      "learning_rate": 0.0005969684352173208,
      "loss": 2.4288,
      "step": 10435
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6524510979652405,
      "learning_rate": 0.000596967855132472,
      "loss": 2.4793,
      "step": 10436
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.5933322906494141,
      "learning_rate": 0.0005969672749924113,
      "loss": 2.4641,
      "step": 10437
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6888641715049744,
      "learning_rate": 0.0005969666947971386,
      "loss": 2.5321,
      "step": 10438
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6802991032600403,
      "learning_rate": 0.0005969661145466545,
      "loss": 2.504,
      "step": 10439
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.9270418286323547,
      "learning_rate": 0.0005969655342409586,
      "loss": 2.4581,
      "step": 10440
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.5517604351043701,
      "learning_rate": 0.0005969649538800512,
      "loss": 2.2887,
      "step": 10441
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.631296694278717,
      "learning_rate": 0.0005969643734639324,
      "loss": 2.2283,
      "step": 10442
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6138027310371399,
      "learning_rate": 0.0005969637929926023,
      "loss": 2.6708,
      "step": 10443
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6073071360588074,
      "learning_rate": 0.0005969632124660612,
      "loss": 2.2931,
      "step": 10444
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6488249897956848,
      "learning_rate": 0.0005969626318843088,
      "loss": 2.424,
      "step": 10445
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.018965721130371,
      "learning_rate": 0.0005969620512473456,
      "loss": 2.4711,
      "step": 10446
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6062617897987366,
      "learning_rate": 0.0005969614705551716,
      "loss": 2.6125,
      "step": 10447
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.7031390070915222,
      "learning_rate": 0.0005969608898077867,
      "loss": 2.4068,
      "step": 10448
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6442315578460693,
      "learning_rate": 0.0005969603090051912,
      "loss": 2.3887,
      "step": 10449
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6581152081489563,
      "learning_rate": 0.0005969597281473851,
      "loss": 2.6459,
      "step": 10450
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.5854976773262024,
      "learning_rate": 0.0005969591472343687,
      "loss": 2.21,
      "step": 10451
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.8195285797119141,
      "learning_rate": 0.000596958566266142,
      "loss": 2.5547,
      "step": 10452
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.9444708824157715,
      "learning_rate": 0.000596957985242705,
      "loss": 2.6149,
      "step": 10453
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6209285259246826,
      "learning_rate": 0.000596957404164058,
      "loss": 2.4552,
      "step": 10454
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6383792161941528,
      "learning_rate": 0.000596956823030201,
      "loss": 2.3122,
      "step": 10455
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.7423154711723328,
      "learning_rate": 0.0005969562418411341,
      "loss": 2.5264,
      "step": 10456
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6558618545532227,
      "learning_rate": 0.0005969556605968574,
      "loss": 2.283,
      "step": 10457
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6356598138809204,
      "learning_rate": 0.000596955079297371,
      "loss": 2.4064,
      "step": 10458
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.7272088527679443,
      "learning_rate": 0.0005969544979426752,
      "loss": 2.3607,
      "step": 10459
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6627175807952881,
      "learning_rate": 0.0005969539165327698,
      "loss": 2.5797,
      "step": 10460
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6273086071014404,
      "learning_rate": 0.0005969533350676552,
      "loss": 2.5258,
      "step": 10461
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.5914068818092346,
      "learning_rate": 0.0005969527535473313,
      "loss": 2.4008,
      "step": 10462
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.7266722917556763,
      "learning_rate": 0.0005969521719717982,
      "loss": 2.5987,
      "step": 10463
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6737789511680603,
      "learning_rate": 0.0005969515903410562,
      "loss": 2.6739,
      "step": 10464
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.5857255458831787,
      "learning_rate": 0.0005969510086551052,
      "loss": 2.5802,
      "step": 10465
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.5941060185432434,
      "learning_rate": 0.0005969504269139454,
      "loss": 2.6052,
      "step": 10466
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6116632223129272,
      "learning_rate": 0.0005969498451175771,
      "loss": 2.3419,
      "step": 10467
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.5931688547134399,
      "learning_rate": 0.000596949263266,
      "loss": 2.0972,
      "step": 10468
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6493574380874634,
      "learning_rate": 0.0005969486813592146,
      "loss": 2.5629,
      "step": 10469
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6125971078872681,
      "learning_rate": 0.0005969480993972206,
      "loss": 2.3784,
      "step": 10470
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6570553779602051,
      "learning_rate": 0.0005969475173800186,
      "loss": 2.6396,
      "step": 10471
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.5828420519828796,
      "learning_rate": 0.0005969469353076084,
      "loss": 2.4536,
      "step": 10472
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6387428641319275,
      "learning_rate": 0.00059694635317999,
      "loss": 2.433,
      "step": 10473
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.676886260509491,
      "learning_rate": 0.0005969457709971638,
      "loss": 2.4396,
      "step": 10474
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6195522546768188,
      "learning_rate": 0.0005969451887591297,
      "loss": 2.2221,
      "step": 10475
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6122720241546631,
      "learning_rate": 0.000596944606465888,
      "loss": 2.5462,
      "step": 10476
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6017999053001404,
      "learning_rate": 0.0005969440241174386,
      "loss": 2.3328,
      "step": 10477
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6671714782714844,
      "learning_rate": 0.0005969434417137817,
      "loss": 2.7325,
      "step": 10478
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.9085389375686646,
      "learning_rate": 0.0005969428592549174,
      "loss": 2.5441,
      "step": 10479
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6979418992996216,
      "learning_rate": 0.0005969422767408459,
      "loss": 2.6312,
      "step": 10480
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.0150096416473389,
      "learning_rate": 0.0005969416941715672,
      "loss": 2.3529,
      "step": 10481
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6689999103546143,
      "learning_rate": 0.0005969411115470813,
      "loss": 2.6618,
      "step": 10482
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6868634819984436,
      "learning_rate": 0.0005969405288673886,
      "loss": 2.5205,
      "step": 10483
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6301511526107788,
      "learning_rate": 0.000596939946132489,
      "loss": 2.3795,
      "step": 10484
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.680791974067688,
      "learning_rate": 0.0005969393633423825,
      "loss": 2.5102,
      "step": 10485
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6368735432624817,
      "learning_rate": 0.0005969387804970696,
      "loss": 2.5799,
      "step": 10486
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6109771728515625,
      "learning_rate": 0.0005969381975965502,
      "loss": 2.413,
      "step": 10487
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.7468564510345459,
      "learning_rate": 0.0005969376146408242,
      "loss": 2.3046,
      "step": 10488
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6457937359809875,
      "learning_rate": 0.000596937031629892,
      "loss": 2.3293,
      "step": 10489
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.5900033116340637,
      "learning_rate": 0.0005969364485637536,
      "loss": 2.4674,
      "step": 10490
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6976571083068848,
      "learning_rate": 0.0005969358654424091,
      "loss": 2.6378,
      "step": 10491
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6396822929382324,
      "learning_rate": 0.0005969352822658585,
      "loss": 2.5445,
      "step": 10492
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.7317028641700745,
      "learning_rate": 0.0005969346990341022,
      "loss": 2.5023,
      "step": 10493
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.5639148950576782,
      "learning_rate": 0.0005969341157471401,
      "loss": 2.4651,
      "step": 10494
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6155460476875305,
      "learning_rate": 0.0005969335324049723,
      "loss": 2.465,
      "step": 10495
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.5937711000442505,
      "learning_rate": 0.000596932949007599,
      "loss": 2.4011,
      "step": 10496
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6475529670715332,
      "learning_rate": 0.0005969323655550202,
      "loss": 2.6454,
      "step": 10497
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6117889881134033,
      "learning_rate": 0.000596931782047236,
      "loss": 2.5065,
      "step": 10498
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.592656135559082,
      "learning_rate": 0.0005969311984842467,
      "loss": 2.5523,
      "step": 10499
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.7275214195251465,
      "learning_rate": 0.0005969306148660522,
      "loss": 2.5218,
      "step": 10500
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6653031706809998,
      "learning_rate": 0.0005969300311926528,
      "loss": 2.7541,
      "step": 10501
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6658241748809814,
      "learning_rate": 0.0005969294474640484,
      "loss": 2.4498,
      "step": 10502
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6715738773345947,
      "learning_rate": 0.0005969288636802392,
      "loss": 2.2903,
      "step": 10503
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6704468131065369,
      "learning_rate": 0.0005969282798412254,
      "loss": 2.5069,
      "step": 10504
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.5588786602020264,
      "learning_rate": 0.000596927695947007,
      "loss": 2.3347,
      "step": 10505
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6310806274414062,
      "learning_rate": 0.0005969271119975842,
      "loss": 2.4072,
      "step": 10506
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.7072657346725464,
      "learning_rate": 0.000596926527992957,
      "loss": 2.3674,
      "step": 10507
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6704182624816895,
      "learning_rate": 0.0005969259439331255,
      "loss": 2.595,
      "step": 10508
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6137667298316956,
      "learning_rate": 0.0005969253598180899,
      "loss": 2.6899,
      "step": 10509
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.653976321220398,
      "learning_rate": 0.0005969247756478503,
      "loss": 2.5602,
      "step": 10510
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6991986632347107,
      "learning_rate": 0.0005969241914224067,
      "loss": 2.5608,
      "step": 10511
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.7528054118156433,
      "learning_rate": 0.0005969236071417594,
      "loss": 2.524,
      "step": 10512
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.5991412997245789,
      "learning_rate": 0.0005969230228059083,
      "loss": 2.3294,
      "step": 10513
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6487817168235779,
      "learning_rate": 0.0005969224384148536,
      "loss": 2.4646,
      "step": 10514
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.5750755071640015,
      "learning_rate": 0.0005969218539685954,
      "loss": 2.5046,
      "step": 10515
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6170506477355957,
      "learning_rate": 0.000596921269467134,
      "loss": 2.4257,
      "step": 10516
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6051267385482788,
      "learning_rate": 0.0005969206849104692,
      "loss": 2.4385,
      "step": 10517
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.749482274055481,
      "learning_rate": 0.0005969201002986012,
      "loss": 2.3044,
      "step": 10518
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6523364186286926,
      "learning_rate": 0.0005969195156315301,
      "loss": 2.4197,
      "step": 10519
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6549036502838135,
      "learning_rate": 0.0005969189309092562,
      "loss": 2.4305,
      "step": 10520
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.9401421546936035,
      "learning_rate": 0.0005969183461317794,
      "loss": 2.2683,
      "step": 10521
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.5954610705375671,
      "learning_rate": 0.0005969177612990999,
      "loss": 2.4587,
      "step": 10522
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.9892322421073914,
      "learning_rate": 0.0005969171764112178,
      "loss": 2.408,
      "step": 10523
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.7224103212356567,
      "learning_rate": 0.0005969165914681331,
      "loss": 2.4896,
      "step": 10524
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.7651209831237793,
      "learning_rate": 0.0005969160064698461,
      "loss": 2.3976,
      "step": 10525
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.854110598564148,
      "learning_rate": 0.0005969154214163568,
      "loss": 2.6968,
      "step": 10526
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.66254061460495,
      "learning_rate": 0.0005969148363076653,
      "loss": 2.3366,
      "step": 10527
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6910557746887207,
      "learning_rate": 0.0005969142511437717,
      "loss": 2.4698,
      "step": 10528
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.7432602047920227,
      "learning_rate": 0.0005969136659246762,
      "loss": 2.3839,
      "step": 10529
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6375722289085388,
      "learning_rate": 0.0005969130806503788,
      "loss": 2.3359,
      "step": 10530
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6922611594200134,
      "learning_rate": 0.0005969124953208796,
      "loss": 2.4472,
      "step": 10531
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6152721047401428,
      "learning_rate": 0.0005969119099361789,
      "loss": 2.2459,
      "step": 10532
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.7126865983009338,
      "learning_rate": 0.0005969113244962765,
      "loss": 2.355,
      "step": 10533
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6475855708122253,
      "learning_rate": 0.0005969107390011728,
      "loss": 2.5287,
      "step": 10534
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6724991798400879,
      "learning_rate": 0.0005969101534508678,
      "loss": 2.4874,
      "step": 10535
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6475015878677368,
      "learning_rate": 0.0005969095678453615,
      "loss": 2.4402,
      "step": 10536
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.7127569317817688,
      "learning_rate": 0.0005969089821846542,
      "loss": 2.5082,
      "step": 10537
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6508040428161621,
      "learning_rate": 0.0005969083964687459,
      "loss": 2.4211,
      "step": 10538
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6861485242843628,
      "learning_rate": 0.0005969078106976367,
      "loss": 2.4358,
      "step": 10539
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6338791251182556,
      "learning_rate": 0.0005969072248713267,
      "loss": 2.5924,
      "step": 10540
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6776090860366821,
      "learning_rate": 0.000596906638989816,
      "loss": 2.4591,
      "step": 10541
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.5923290848731995,
      "learning_rate": 0.0005969060530531049,
      "loss": 2.5396,
      "step": 10542
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6455259323120117,
      "learning_rate": 0.0005969054670611933,
      "loss": 2.4807,
      "step": 10543
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.5936241745948792,
      "learning_rate": 0.0005969048810140813,
      "loss": 2.3979,
      "step": 10544
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.5967992544174194,
      "learning_rate": 0.0005969042949117691,
      "loss": 2.4609,
      "step": 10545
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6243000030517578,
      "learning_rate": 0.000596903708754257,
      "loss": 2.538,
      "step": 10546
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6547600030899048,
      "learning_rate": 0.0005969031225415447,
      "loss": 2.3918,
      "step": 10547
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.5466272830963135,
      "learning_rate": 0.0005969025362736324,
      "loss": 2.4512,
      "step": 10548
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6391960382461548,
      "learning_rate": 0.0005969019499505205,
      "loss": 2.3233,
      "step": 10549
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.5965597629547119,
      "learning_rate": 0.0005969013635722088,
      "loss": 2.4542,
      "step": 10550
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6439383625984192,
      "learning_rate": 0.0005969007771386977,
      "loss": 2.5047,
      "step": 10551
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6171985864639282,
      "learning_rate": 0.000596900190649987,
      "loss": 2.4273,
      "step": 10552
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.762826681137085,
      "learning_rate": 0.000596899604106077,
      "loss": 2.2568,
      "step": 10553
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6098453998565674,
      "learning_rate": 0.0005968990175069678,
      "loss": 2.5763,
      "step": 10554
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.7641507983207703,
      "learning_rate": 0.0005968984308526593,
      "loss": 2.6877,
      "step": 10555
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6860194802284241,
      "learning_rate": 0.000596897844143152,
      "loss": 2.4867,
      "step": 10556
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6304017305374146,
      "learning_rate": 0.0005968972573784456,
      "loss": 2.5665,
      "step": 10557
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.5839477181434631,
      "learning_rate": 0.0005968966705585405,
      "loss": 2.3474,
      "step": 10558
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.7002192139625549,
      "learning_rate": 0.0005968960836834367,
      "loss": 2.5302,
      "step": 10559
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6174377202987671,
      "learning_rate": 0.0005968954967531342,
      "loss": 2.5714,
      "step": 10560
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.7146278619766235,
      "learning_rate": 0.0005968949097676335,
      "loss": 2.5979,
      "step": 10561
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6234228014945984,
      "learning_rate": 0.0005968943227269342,
      "loss": 2.5372,
      "step": 10562
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.5887318849563599,
      "learning_rate": 0.0005968937356310366,
      "loss": 2.4808,
      "step": 10563
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6115169525146484,
      "learning_rate": 0.0005968931484799409,
      "loss": 2.0577,
      "step": 10564
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6241650581359863,
      "learning_rate": 0.0005968925612736473,
      "loss": 2.0563,
      "step": 10565
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.5914401412010193,
      "learning_rate": 0.0005968919740121557,
      "loss": 2.291,
      "step": 10566
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6516197323799133,
      "learning_rate": 0.0005968913866954661,
      "loss": 2.5994,
      "step": 10567
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.5916681885719299,
      "learning_rate": 0.000596890799323579,
      "loss": 2.5926,
      "step": 10568
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.692994236946106,
      "learning_rate": 0.0005968902118964943,
      "loss": 2.599,
      "step": 10569
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.7072107791900635,
      "learning_rate": 0.000596889624414212,
      "loss": 2.4892,
      "step": 10570
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6963334083557129,
      "learning_rate": 0.0005968890368767323,
      "loss": 2.5447,
      "step": 10571
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.709551990032196,
      "learning_rate": 0.0005968884492840553,
      "loss": 2.5894,
      "step": 10572
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.7089252471923828,
      "learning_rate": 0.0005968878616361813,
      "loss": 2.4732,
      "step": 10573
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.5714463591575623,
      "learning_rate": 0.0005968872739331102,
      "loss": 2.2457,
      "step": 10574
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6349987983703613,
      "learning_rate": 0.0005968866861748421,
      "loss": 2.2354,
      "step": 10575
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6748655438423157,
      "learning_rate": 0.0005968860983613771,
      "loss": 2.5904,
      "step": 10576
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6360769271850586,
      "learning_rate": 0.0005968855104927154,
      "loss": 2.5143,
      "step": 10577
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6240923404693604,
      "learning_rate": 0.0005968849225688571,
      "loss": 2.6062,
      "step": 10578
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6434323191642761,
      "learning_rate": 0.0005968843345898023,
      "loss": 2.4915,
      "step": 10579
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6946004629135132,
      "learning_rate": 0.0005968837465555512,
      "loss": 2.474,
      "step": 10580
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.5777431130409241,
      "learning_rate": 0.0005968831584661036,
      "loss": 2.4614,
      "step": 10581
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.5796769857406616,
      "learning_rate": 0.00059688257032146,
      "loss": 2.4495,
      "step": 10582
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6410006284713745,
      "learning_rate": 0.0005968819821216202,
      "loss": 2.2865,
      "step": 10583
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6587525606155396,
      "learning_rate": 0.0005968813938665845,
      "loss": 2.4468,
      "step": 10584
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.693072497844696,
      "learning_rate": 0.0005968808055563529,
      "loss": 2.4159,
      "step": 10585
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6837619543075562,
      "learning_rate": 0.0005968802171909256,
      "loss": 2.2276,
      "step": 10586
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6030138731002808,
      "learning_rate": 0.0005968796287703027,
      "loss": 2.5785,
      "step": 10587
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.7660945057868958,
      "learning_rate": 0.0005968790402944841,
      "loss": 2.4569,
      "step": 10588
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6339420676231384,
      "learning_rate": 0.0005968784517634703,
      "loss": 2.2213,
      "step": 10589
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.677056074142456,
      "learning_rate": 0.000596877863177261,
      "loss": 2.5226,
      "step": 10590
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.5920601487159729,
      "learning_rate": 0.0005968772745358566,
      "loss": 2.2358,
      "step": 10591
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6752526164054871,
      "learning_rate": 0.0005968766858392571,
      "loss": 2.6161,
      "step": 10592
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6467880606651306,
      "learning_rate": 0.0005968760970874626,
      "loss": 2.5169,
      "step": 10593
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.7968893647193909,
      "learning_rate": 0.0005968755082804733,
      "loss": 2.5892,
      "step": 10594
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6151551008224487,
      "learning_rate": 0.0005968749194182892,
      "loss": 2.4271,
      "step": 10595
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6424019932746887,
      "learning_rate": 0.0005968743305009104,
      "loss": 2.4227,
      "step": 10596
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6848896145820618,
      "learning_rate": 0.0005968737415283371,
      "loss": 2.2128,
      "step": 10597
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6086883544921875,
      "learning_rate": 0.0005968731525005694,
      "loss": 2.4581,
      "step": 10598
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.5995286107063293,
      "learning_rate": 0.0005968725634176073,
      "loss": 2.3435,
      "step": 10599
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.5757849216461182,
      "learning_rate": 0.0005968719742794511,
      "loss": 2.3963,
      "step": 10600
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.9009119868278503,
      "learning_rate": 0.0005968713850861007,
      "loss": 2.3802,
      "step": 10601
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6546069383621216,
      "learning_rate": 0.0005968707958375563,
      "loss": 2.5873,
      "step": 10602
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6217939853668213,
      "learning_rate": 0.000596870206533818,
      "loss": 2.4784,
      "step": 10603
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.668341338634491,
      "learning_rate": 0.000596869617174886,
      "loss": 2.7424,
      "step": 10604
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6328529119491577,
      "learning_rate": 0.0005968690277607604,
      "loss": 2.3723,
      "step": 10605
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6363928914070129,
      "learning_rate": 0.0005968684382914411,
      "loss": 2.6569,
      "step": 10606
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.5972472429275513,
      "learning_rate": 0.0005968678487669284,
      "loss": 2.514,
      "step": 10607
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.5855308771133423,
      "learning_rate": 0.0005968672591872223,
      "loss": 2.2924,
      "step": 10608
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.5792936682701111,
      "learning_rate": 0.0005968666695523231,
      "loss": 2.4086,
      "step": 10609
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6488394141197205,
      "learning_rate": 0.0005968660798622308,
      "loss": 2.3874,
      "step": 10610
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.5821366310119629,
      "learning_rate": 0.0005968654901169453,
      "loss": 2.4912,
      "step": 10611
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6046046018600464,
      "learning_rate": 0.000596864900316467,
      "loss": 2.3511,
      "step": 10612
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6432217359542847,
      "learning_rate": 0.000596864310460796,
      "loss": 2.5782,
      "step": 10613
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6510798335075378,
      "learning_rate": 0.0005968637205499322,
      "loss": 2.4186,
      "step": 10614
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6954120993614197,
      "learning_rate": 0.0005968631305838758,
      "loss": 2.502,
      "step": 10615
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.7582851648330688,
      "learning_rate": 0.0005968625405626271,
      "loss": 2.4993,
      "step": 10616
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.697576105594635,
      "learning_rate": 0.000596861950486186,
      "loss": 2.3968,
      "step": 10617
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.611240565776825,
      "learning_rate": 0.0005968613603545526,
      "loss": 2.4138,
      "step": 10618
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.7191531658172607,
      "learning_rate": 0.0005968607701677271,
      "loss": 2.5512,
      "step": 10619
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.648887038230896,
      "learning_rate": 0.0005968601799257096,
      "loss": 2.3218,
      "step": 10620
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6958590745925903,
      "learning_rate": 0.0005968595896285001,
      "loss": 2.6395,
      "step": 10621
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6519407033920288,
      "learning_rate": 0.000596858999276099,
      "loss": 2.4725,
      "step": 10622
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.7349162101745605,
      "learning_rate": 0.000596858408868506,
      "loss": 2.3841,
      "step": 10623
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.604694664478302,
      "learning_rate": 0.0005968578184057215,
      "loss": 2.2212,
      "step": 10624
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6911583542823792,
      "learning_rate": 0.0005968572278877455,
      "loss": 2.5113,
      "step": 10625
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6384603381156921,
      "learning_rate": 0.0005968566373145782,
      "loss": 2.6779,
      "step": 10626
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.5585354566574097,
      "learning_rate": 0.0005968560466862196,
      "loss": 2.2875,
      "step": 10627
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6549355983734131,
      "learning_rate": 0.0005968554560026697,
      "loss": 2.4194,
      "step": 10628
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6391273140907288,
      "learning_rate": 0.000596854865263929,
      "loss": 2.3278,
      "step": 10629
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.5985932350158691,
      "learning_rate": 0.0005968542744699974,
      "loss": 2.4353,
      "step": 10630
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6389263272285461,
      "learning_rate": 0.0005968536836208748,
      "loss": 2.3083,
      "step": 10631
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.7314048409461975,
      "learning_rate": 0.0005968530927165616,
      "loss": 2.4137,
      "step": 10632
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6194809675216675,
      "learning_rate": 0.0005968525017570578,
      "loss": 2.3276,
      "step": 10633
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6410596966743469,
      "learning_rate": 0.0005968519107423635,
      "loss": 2.3584,
      "step": 10634
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6853398084640503,
      "learning_rate": 0.0005968513196724788,
      "loss": 2.562,
      "step": 10635
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6153501868247986,
      "learning_rate": 0.0005968507285474039,
      "loss": 2.3619,
      "step": 10636
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6301124095916748,
      "learning_rate": 0.0005968501373671388,
      "loss": 2.5079,
      "step": 10637
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6661325097084045,
      "learning_rate": 0.0005968495461316837,
      "loss": 2.3262,
      "step": 10638
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6423086524009705,
      "learning_rate": 0.0005968489548410387,
      "loss": 2.3433,
      "step": 10639
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.5459222197532654,
      "learning_rate": 0.0005968483634952037,
      "loss": 2.5167,
      "step": 10640
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.7012109160423279,
      "learning_rate": 0.0005968477720941792,
      "loss": 2.5749,
      "step": 10641
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6041885614395142,
      "learning_rate": 0.0005968471806379649,
      "loss": 2.5263,
      "step": 10642
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6679813265800476,
      "learning_rate": 0.0005968465891265613,
      "loss": 2.4395,
      "step": 10643
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6676004528999329,
      "learning_rate": 0.0005968459975599682,
      "loss": 2.611,
      "step": 10644
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.7149245142936707,
      "learning_rate": 0.0005968454059381858,
      "loss": 2.4896,
      "step": 10645
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.5995820164680481,
      "learning_rate": 0.0005968448142612144,
      "loss": 2.3108,
      "step": 10646
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6694799065589905,
      "learning_rate": 0.0005968442225290538,
      "loss": 2.32,
      "step": 10647
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6301116347312927,
      "learning_rate": 0.0005968436307417042,
      "loss": 2.4436,
      "step": 10648
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6449050307273865,
      "learning_rate": 0.0005968430388991659,
      "loss": 2.2754,
      "step": 10649
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.5948941707611084,
      "learning_rate": 0.0005968424470014389,
      "loss": 2.2823,
      "step": 10650
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.5813538432121277,
      "learning_rate": 0.0005968418550485231,
      "loss": 2.4133,
      "step": 10651
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6013718843460083,
      "learning_rate": 0.000596841263040419,
      "loss": 2.3323,
      "step": 10652
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.728193461894989,
      "learning_rate": 0.0005968406709771262,
      "loss": 2.4026,
      "step": 10653
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6355567574501038,
      "learning_rate": 0.0005968400788586455,
      "loss": 2.5667,
      "step": 10654
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6555256247520447,
      "learning_rate": 0.0005968394866849764,
      "loss": 2.46,
      "step": 10655
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6637013554573059,
      "learning_rate": 0.0005968388944561193,
      "loss": 2.4383,
      "step": 10656
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6127904057502747,
      "learning_rate": 0.0005968383021720743,
      "loss": 2.4038,
      "step": 10657
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.7092586159706116,
      "learning_rate": 0.0005968377098328413,
      "loss": 2.6229,
      "step": 10658
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6375876069068909,
      "learning_rate": 0.0005968371174384208,
      "loss": 2.4949,
      "step": 10659
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.631645143032074,
      "learning_rate": 0.0005968365249888124,
      "loss": 2.4967,
      "step": 10660
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.5918810367584229,
      "learning_rate": 0.0005968359324840166,
      "loss": 2.4448,
      "step": 10661
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6766376495361328,
      "learning_rate": 0.0005968353399240335,
      "loss": 2.5967,
      "step": 10662
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.5909596681594849,
      "learning_rate": 0.000596834747308863,
      "loss": 2.3771,
      "step": 10663
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6465012431144714,
      "learning_rate": 0.0005968341546385053,
      "loss": 2.3887,
      "step": 10664
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.626442015171051,
      "learning_rate": 0.0005968335619129606,
      "loss": 2.4972,
      "step": 10665
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6218616366386414,
      "learning_rate": 0.0005968329691322289,
      "loss": 2.4215,
      "step": 10666
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.5970725417137146,
      "learning_rate": 0.0005968323762963103,
      "loss": 2.4454,
      "step": 10667
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.579925000667572,
      "learning_rate": 0.0005968317834052051,
      "loss": 2.6292,
      "step": 10668
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6523076295852661,
      "learning_rate": 0.0005968311904589131,
      "loss": 2.3113,
      "step": 10669
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6271548271179199,
      "learning_rate": 0.0005968305974574347,
      "loss": 2.5315,
      "step": 10670
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6369943618774414,
      "learning_rate": 0.0005968300044007698,
      "loss": 2.2298,
      "step": 10671
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6262112259864807,
      "learning_rate": 0.0005968294112889187,
      "loss": 2.6216,
      "step": 10672
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.5996101498603821,
      "learning_rate": 0.0005968288181218814,
      "loss": 2.5704,
      "step": 10673
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6360034942626953,
      "learning_rate": 0.0005968282248996579,
      "loss": 2.2854,
      "step": 10674
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.656386137008667,
      "learning_rate": 0.0005968276316222486,
      "loss": 2.4971,
      "step": 10675
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6219696998596191,
      "learning_rate": 0.0005968270382896533,
      "loss": 2.4357,
      "step": 10676
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.586223840713501,
      "learning_rate": 0.0005968264449018723,
      "loss": 2.7462,
      "step": 10677
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.575313150882721,
      "learning_rate": 0.0005968258514589057,
      "loss": 2.3304,
      "step": 10678
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6120709776878357,
      "learning_rate": 0.0005968252579607535,
      "loss": 2.5579,
      "step": 10679
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.716036856174469,
      "learning_rate": 0.0005968246644074159,
      "loss": 2.5725,
      "step": 10680
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.5617828965187073,
      "learning_rate": 0.000596824070798893,
      "loss": 2.4509,
      "step": 10681
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.7059904932975769,
      "learning_rate": 0.0005968234771351849,
      "loss": 2.5296,
      "step": 10682
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.7932099103927612,
      "learning_rate": 0.0005968228834162917,
      "loss": 2.3976,
      "step": 10683
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6288747191429138,
      "learning_rate": 0.0005968222896422136,
      "loss": 2.6699,
      "step": 10684
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.8856366872787476,
      "learning_rate": 0.0005968216958129505,
      "loss": 2.3153,
      "step": 10685
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.7256510257720947,
      "learning_rate": 0.0005968211019285027,
      "loss": 2.6867,
      "step": 10686
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6344443559646606,
      "learning_rate": 0.0005968205079888703,
      "loss": 2.4856,
      "step": 10687
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.9382501244544983,
      "learning_rate": 0.0005968199139940534,
      "loss": 2.4203,
      "step": 10688
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6916602253913879,
      "learning_rate": 0.000596819319944052,
      "loss": 2.6197,
      "step": 10689
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6807950735092163,
      "learning_rate": 0.0005968187258388663,
      "loss": 2.4216,
      "step": 10690
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.823748767375946,
      "learning_rate": 0.0005968181316784964,
      "loss": 2.7298,
      "step": 10691
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.7098150253295898,
      "learning_rate": 0.0005968175374629425,
      "loss": 2.5254,
      "step": 10692
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6408316493034363,
      "learning_rate": 0.0005968169431922044,
      "loss": 2.5859,
      "step": 10693
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6622753739356995,
      "learning_rate": 0.0005968163488662826,
      "loss": 2.3575,
      "step": 10694
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.7236363291740417,
      "learning_rate": 0.0005968157544851769,
      "loss": 2.4131,
      "step": 10695
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6879100203514099,
      "learning_rate": 0.0005968151600488877,
      "loss": 2.3914,
      "step": 10696
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6147075295448303,
      "learning_rate": 0.0005968145655574149,
      "loss": 2.421,
      "step": 10697
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6439223885536194,
      "learning_rate": 0.0005968139710107586,
      "loss": 2.5967,
      "step": 10698
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.5744770765304565,
      "learning_rate": 0.0005968133764089191,
      "loss": 2.2472,
      "step": 10699
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6623076796531677,
      "learning_rate": 0.0005968127817518962,
      "loss": 2.3054,
      "step": 10700
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6555650234222412,
      "learning_rate": 0.0005968121870396904,
      "loss": 2.6533,
      "step": 10701
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6122620701789856,
      "learning_rate": 0.0005968115922723016,
      "loss": 2.4254,
      "step": 10702
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.672458291053772,
      "learning_rate": 0.0005968109974497298,
      "loss": 2.2939,
      "step": 10703
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6898948550224304,
      "learning_rate": 0.0005968104025719753,
      "loss": 2.4946,
      "step": 10704
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6105888485908508,
      "learning_rate": 0.0005968098076390381,
      "loss": 2.5928,
      "step": 10705
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6988084316253662,
      "learning_rate": 0.0005968092126509185,
      "loss": 2.4714,
      "step": 10706
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.8585172295570374,
      "learning_rate": 0.0005968086176076163,
      "loss": 2.4798,
      "step": 10707
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.5847667455673218,
      "learning_rate": 0.0005968080225091318,
      "loss": 2.3763,
      "step": 10708
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.761907696723938,
      "learning_rate": 0.0005968074273554652,
      "loss": 2.4229,
      "step": 10709
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.8187811374664307,
      "learning_rate": 0.0005968068321466164,
      "loss": 2.4535,
      "step": 10710
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.582031786441803,
      "learning_rate": 0.0005968062368825857,
      "loss": 2.5,
      "step": 10711
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.0687572956085205,
      "learning_rate": 0.000596805641563373,
      "loss": 2.3683,
      "step": 10712
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.7501674890518188,
      "learning_rate": 0.0005968050461889786,
      "loss": 2.4995,
      "step": 10713
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.5737470984458923,
      "learning_rate": 0.0005968044507594025,
      "loss": 2.68,
      "step": 10714
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.7015398144721985,
      "learning_rate": 0.0005968038552746449,
      "loss": 2.5535,
      "step": 10715
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6704570651054382,
      "learning_rate": 0.0005968032597347058,
      "loss": 2.4221,
      "step": 10716
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.7206635475158691,
      "learning_rate": 0.0005968026641395854,
      "loss": 2.4939,
      "step": 10717
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6219021677970886,
      "learning_rate": 0.0005968020684892838,
      "loss": 2.4709,
      "step": 10718
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.7171791195869446,
      "learning_rate": 0.0005968014727838009,
      "loss": 2.7375,
      "step": 10719
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.676345705986023,
      "learning_rate": 0.0005968008770231373,
      "loss": 2.6003,
      "step": 10720
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6181333065032959,
      "learning_rate": 0.0005968002812072927,
      "loss": 2.4667,
      "step": 10721
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.718398928642273,
      "learning_rate": 0.0005967996853362672,
      "loss": 2.208,
      "step": 10722
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.7001915574073792,
      "learning_rate": 0.0005967990894100612,
      "loss": 2.6868,
      "step": 10723
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6447550058364868,
      "learning_rate": 0.0005967984934286746,
      "loss": 2.4757,
      "step": 10724
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6389889717102051,
      "learning_rate": 0.0005967978973921076,
      "loss": 2.5983,
      "step": 10725
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6492058038711548,
      "learning_rate": 0.0005967973013003601,
      "loss": 2.4546,
      "step": 10726
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6895818114280701,
      "learning_rate": 0.0005967967051534326,
      "loss": 2.3317,
      "step": 10727
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.619138240814209,
      "learning_rate": 0.0005967961089513249,
      "loss": 2.4768,
      "step": 10728
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6449183821678162,
      "learning_rate": 0.0005967955126940372,
      "loss": 2.3471,
      "step": 10729
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6341464519500732,
      "learning_rate": 0.0005967949163815696,
      "loss": 2.5182,
      "step": 10730
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.612480640411377,
      "learning_rate": 0.0005967943200139221,
      "loss": 2.4428,
      "step": 10731
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6082081198692322,
      "learning_rate": 0.0005967937235910951,
      "loss": 2.3435,
      "step": 10732
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.682842493057251,
      "learning_rate": 0.0005967931271130886,
      "loss": 2.3324,
      "step": 10733
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6375752091407776,
      "learning_rate": 0.0005967925305799026,
      "loss": 2.3036,
      "step": 10734
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6667459011077881,
      "learning_rate": 0.0005967919339915372,
      "loss": 2.4569,
      "step": 10735
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6031714677810669,
      "learning_rate": 0.0005967913373479925,
      "loss": 2.5884,
      "step": 10736
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.636239767074585,
      "learning_rate": 0.0005967907406492688,
      "loss": 2.5513,
      "step": 10737
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.602034866809845,
      "learning_rate": 0.0005967901438953661,
      "loss": 2.494,
      "step": 10738
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6438502669334412,
      "learning_rate": 0.0005967895470862845,
      "loss": 2.6796,
      "step": 10739
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.5924251675605774,
      "learning_rate": 0.0005967889502220241,
      "loss": 2.2095,
      "step": 10740
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6216034889221191,
      "learning_rate": 0.0005967883533025851,
      "loss": 2.3375,
      "step": 10741
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6064806580543518,
      "learning_rate": 0.0005967877563279675,
      "loss": 2.6336,
      "step": 10742
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6695525646209717,
      "learning_rate": 0.0005967871592981714,
      "loss": 2.329,
      "step": 10743
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6328254342079163,
      "learning_rate": 0.000596786562213197,
      "loss": 2.3681,
      "step": 10744
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6485521793365479,
      "learning_rate": 0.0005967859650730444,
      "loss": 2.4644,
      "step": 10745
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6817097663879395,
      "learning_rate": 0.0005967853678777137,
      "loss": 2.5052,
      "step": 10746
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6078527569770813,
      "learning_rate": 0.000596784770627205,
      "loss": 2.3575,
      "step": 10747
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.7807750105857849,
      "learning_rate": 0.0005967841733215184,
      "loss": 2.4243,
      "step": 10748
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.7228149175643921,
      "learning_rate": 0.000596783575960654,
      "loss": 2.5362,
      "step": 10749
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6267308592796326,
      "learning_rate": 0.0005967829785446119,
      "loss": 2.4588,
      "step": 10750
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6525087952613831,
      "learning_rate": 0.0005967823810733923,
      "loss": 2.361,
      "step": 10751
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.825239896774292,
      "learning_rate": 0.0005967817835469952,
      "loss": 2.4868,
      "step": 10752
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.7022112011909485,
      "learning_rate": 0.0005967811859654208,
      "loss": 2.1814,
      "step": 10753
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.9572242498397827,
      "learning_rate": 0.0005967805883286692,
      "loss": 2.3007,
      "step": 10754
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.7876980900764465,
      "learning_rate": 0.0005967799906367404,
      "loss": 2.5089,
      "step": 10755
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.7851356863975525,
      "learning_rate": 0.0005967793928896346,
      "loss": 2.4337,
      "step": 10756
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.8781489729881287,
      "learning_rate": 0.000596778795087352,
      "loss": 2.647,
      "step": 10757
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.7816498875617981,
      "learning_rate": 0.0005967781972298925,
      "loss": 2.4168,
      "step": 10758
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6997456550598145,
      "learning_rate": 0.0005967775993172565,
      "loss": 2.5612,
      "step": 10759
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.9634829759597778,
      "learning_rate": 0.0005967770013494439,
      "loss": 2.1455,
      "step": 10760
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6471894979476929,
      "learning_rate": 0.0005967764033264547,
      "loss": 2.5023,
      "step": 10761
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6309053301811218,
      "learning_rate": 0.0005967758052482894,
      "loss": 2.3818,
      "step": 10762
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.7301556468009949,
      "learning_rate": 0.0005967752071149477,
      "loss": 2.8535,
      "step": 10763
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.709204912185669,
      "learning_rate": 0.0005967746089264299,
      "loss": 2.5208,
      "step": 10764
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6077545285224915,
      "learning_rate": 0.0005967740106827362,
      "loss": 2.5607,
      "step": 10765
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6413904428482056,
      "learning_rate": 0.0005967734123838665,
      "loss": 2.4761,
      "step": 10766
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6732115745544434,
      "learning_rate": 0.0005967728140298211,
      "loss": 2.7197,
      "step": 10767
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6323347091674805,
      "learning_rate": 0.0005967722156205999,
      "loss": 2.5322,
      "step": 10768
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.7101665735244751,
      "learning_rate": 0.0005967716171562033,
      "loss": 2.6215,
      "step": 10769
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6107850074768066,
      "learning_rate": 0.0005967710186366311,
      "loss": 2.2869,
      "step": 10770
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6253819465637207,
      "learning_rate": 0.0005967704200618838,
      "loss": 2.5202,
      "step": 10771
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6598767638206482,
      "learning_rate": 0.0005967698214319611,
      "loss": 2.3529,
      "step": 10772
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6207315921783447,
      "learning_rate": 0.0005967692227468633,
      "loss": 2.6101,
      "step": 10773
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6363651752471924,
      "learning_rate": 0.0005967686240065904,
      "loss": 2.5634,
      "step": 10774
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.7669740319252014,
      "learning_rate": 0.0005967680252111429,
      "loss": 2.4975,
      "step": 10775
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.5906442999839783,
      "learning_rate": 0.0005967674263605205,
      "loss": 2.3032,
      "step": 10776
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6784011125564575,
      "learning_rate": 0.0005967668274547233,
      "loss": 2.4731,
      "step": 10777
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.7113088369369507,
      "learning_rate": 0.0005967662284937516,
      "loss": 2.56,
      "step": 10778
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6947278380393982,
      "learning_rate": 0.0005967656294776055,
      "loss": 2.776,
      "step": 10779
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.63336181640625,
      "learning_rate": 0.000596765030406285,
      "loss": 2.7109,
      "step": 10780
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6437405943870544,
      "learning_rate": 0.0005967644312797904,
      "loss": 2.5005,
      "step": 10781
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.587122917175293,
      "learning_rate": 0.0005967638320981217,
      "loss": 2.5241,
      "step": 10782
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6119290590286255,
      "learning_rate": 0.0005967632328612788,
      "loss": 2.5601,
      "step": 10783
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6338399648666382,
      "learning_rate": 0.0005967626335692622,
      "loss": 2.3785,
      "step": 10784
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6672061681747437,
      "learning_rate": 0.0005967620342220717,
      "loss": 2.6602,
      "step": 10785
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6174966096878052,
      "learning_rate": 0.0005967614348197077,
      "loss": 2.4927,
      "step": 10786
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.0211533308029175,
      "learning_rate": 0.0005967608353621701,
      "loss": 2.5587,
      "step": 10787
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.5982392430305481,
      "learning_rate": 0.0005967602358494589,
      "loss": 2.3528,
      "step": 10788
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.5793414115905762,
      "learning_rate": 0.0005967596362815745,
      "loss": 2.3839,
      "step": 10789
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6294105052947998,
      "learning_rate": 0.0005967590366585168,
      "loss": 2.4557,
      "step": 10790
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6186838150024414,
      "learning_rate": 0.0005967584369802861,
      "loss": 2.5504,
      "step": 10791
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.631067156791687,
      "learning_rate": 0.0005967578372468823,
      "loss": 2.3606,
      "step": 10792
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6098647713661194,
      "learning_rate": 0.0005967572374583057,
      "loss": 2.4151,
      "step": 10793
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.5671263337135315,
      "learning_rate": 0.0005967566376145564,
      "loss": 2.528,
      "step": 10794
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6391357779502869,
      "learning_rate": 0.0005967560377156343,
      "loss": 2.5532,
      "step": 10795
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6455801725387573,
      "learning_rate": 0.0005967554377615397,
      "loss": 2.3208,
      "step": 10796
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6216136813163757,
      "learning_rate": 0.0005967548377522726,
      "loss": 2.3863,
      "step": 10797
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6011995673179626,
      "learning_rate": 0.0005967542376878333,
      "loss": 2.2613,
      "step": 10798
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.5862423181533813,
      "learning_rate": 0.0005967536375682216,
      "loss": 2.2758,
      "step": 10799
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.589040219783783,
      "learning_rate": 0.000596753037393438,
      "loss": 2.4973,
      "step": 10800
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.688669741153717,
      "learning_rate": 0.0005967524371634824,
      "loss": 2.6199,
      "step": 10801
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.5671298503875732,
      "learning_rate": 0.0005967518368783548,
      "loss": 2.4318,
      "step": 10802
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6037269234657288,
      "learning_rate": 0.0005967512365380554,
      "loss": 2.4609,
      "step": 10803
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6201900839805603,
      "learning_rate": 0.0005967506361425844,
      "loss": 2.6254,
      "step": 10804
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6725558638572693,
      "learning_rate": 0.0005967500356919419,
      "loss": 2.2965,
      "step": 10805
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6253112554550171,
      "learning_rate": 0.000596749435186128,
      "loss": 2.4259,
      "step": 10806
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6706428527832031,
      "learning_rate": 0.0005967488346251427,
      "loss": 2.3261,
      "step": 10807
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.679145097732544,
      "learning_rate": 0.0005967482340089862,
      "loss": 2.2481,
      "step": 10808
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6632742285728455,
      "learning_rate": 0.0005967476333376586,
      "loss": 2.4545,
      "step": 10809
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.7012258768081665,
      "learning_rate": 0.00059674703261116,
      "loss": 2.4432,
      "step": 10810
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.7236858606338501,
      "learning_rate": 0.0005967464318294906,
      "loss": 2.5497,
      "step": 10811
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6581076383590698,
      "learning_rate": 0.0005967458309926503,
      "loss": 2.3566,
      "step": 10812
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.5975296497344971,
      "learning_rate": 0.0005967452301006395,
      "loss": 2.3684,
      "step": 10813
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6722646951675415,
      "learning_rate": 0.000596744629153458,
      "loss": 2.4953,
      "step": 10814
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.7308429479598999,
      "learning_rate": 0.0005967440281511062,
      "loss": 2.5171,
      "step": 10815
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6794073581695557,
      "learning_rate": 0.0005967434270935841,
      "loss": 2.4487,
      "step": 10816
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6749916076660156,
      "learning_rate": 0.0005967428259808917,
      "loss": 2.5205,
      "step": 10817
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6424546241760254,
      "learning_rate": 0.0005967422248130293,
      "loss": 2.7347,
      "step": 10818
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6274073719978333,
      "learning_rate": 0.0005967416235899969,
      "loss": 2.462,
      "step": 10819
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.5999743342399597,
      "learning_rate": 0.0005967410223117946,
      "loss": 2.3934,
      "step": 10820
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.7227845191955566,
      "learning_rate": 0.0005967404209784225,
      "loss": 2.4343,
      "step": 10821
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6199055910110474,
      "learning_rate": 0.0005967398195898808,
      "loss": 2.4772,
      "step": 10822
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.552807629108429,
      "learning_rate": 0.0005967392181461695,
      "loss": 2.633,
      "step": 10823
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6382250785827637,
      "learning_rate": 0.0005967386166472889,
      "loss": 2.3731,
      "step": 10824
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6136714816093445,
      "learning_rate": 0.0005967380150932389,
      "loss": 2.4132,
      "step": 10825
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.5545854568481445,
      "learning_rate": 0.0005967374134840198,
      "loss": 2.5227,
      "step": 10826
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6162423491477966,
      "learning_rate": 0.0005967368118196315,
      "loss": 2.4735,
      "step": 10827
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6626287698745728,
      "learning_rate": 0.0005967362101000743,
      "loss": 2.4182,
      "step": 10828
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6185793280601501,
      "learning_rate": 0.0005967356083253482,
      "loss": 2.4825,
      "step": 10829
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.5859081149101257,
      "learning_rate": 0.0005967350064954534,
      "loss": 2.5001,
      "step": 10830
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6140537261962891,
      "learning_rate": 0.0005967344046103898,
      "loss": 2.2276,
      "step": 10831
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6277854442596436,
      "learning_rate": 0.0005967338026701579,
      "loss": 2.4653,
      "step": 10832
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6062596440315247,
      "learning_rate": 0.0005967332006747575,
      "loss": 2.5509,
      "step": 10833
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.5648977756500244,
      "learning_rate": 0.0005967325986241887,
      "loss": 2.5966,
      "step": 10834
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6431127190589905,
      "learning_rate": 0.0005967319965184518,
      "loss": 2.2968,
      "step": 10835
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.7079285383224487,
      "learning_rate": 0.0005967313943575468,
      "loss": 2.3434,
      "step": 10836
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.5811459422111511,
      "learning_rate": 0.0005967307921414739,
      "loss": 2.533,
      "step": 10837
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6221058368682861,
      "learning_rate": 0.000596730189870233,
      "loss": 2.4761,
      "step": 10838
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.5820402503013611,
      "learning_rate": 0.0005967295875438245,
      "loss": 2.4409,
      "step": 10839
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6164530515670776,
      "learning_rate": 0.0005967289851622483,
      "loss": 2.629,
      "step": 10840
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.5735510587692261,
      "learning_rate": 0.0005967283827255045,
      "loss": 2.2955,
      "step": 10841
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.7028712630271912,
      "learning_rate": 0.0005967277802335935,
      "loss": 2.3427,
      "step": 10842
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6587905287742615,
      "learning_rate": 0.0005967271776865149,
      "loss": 2.6585,
      "step": 10843
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6038923263549805,
      "learning_rate": 0.0005967265750842694,
      "loss": 2.5274,
      "step": 10844
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6430425643920898,
      "learning_rate": 0.0005967259724268567,
      "loss": 2.5881,
      "step": 10845
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.5987979173660278,
      "learning_rate": 0.0005967253697142771,
      "loss": 2.3329,
      "step": 10846
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6552396416664124,
      "learning_rate": 0.0005967247669465307,
      "loss": 2.2436,
      "step": 10847
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6278548836708069,
      "learning_rate": 0.0005967241641236174,
      "loss": 2.4116,
      "step": 10848
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.5610304474830627,
      "learning_rate": 0.0005967235612455375,
      "loss": 2.4353,
      "step": 10849
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.5869784355163574,
      "learning_rate": 0.0005967229583122912,
      "loss": 2.2503,
      "step": 10850
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.5961142182350159,
      "learning_rate": 0.0005967223553238786,
      "loss": 2.4787,
      "step": 10851
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6353887319564819,
      "learning_rate": 0.0005967217522802994,
      "loss": 2.4743,
      "step": 10852
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.636417031288147,
      "learning_rate": 0.0005967211491815542,
      "loss": 2.5895,
      "step": 10853
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6343681812286377,
      "learning_rate": 0.0005967205460276429,
      "loss": 2.4764,
      "step": 10854
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6515078544616699,
      "learning_rate": 0.0005967199428185657,
      "loss": 2.4961,
      "step": 10855
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.830721378326416,
      "learning_rate": 0.0005967193395543227,
      "loss": 2.4791,
      "step": 10856
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.7287517189979553,
      "learning_rate": 0.0005967187362349138,
      "loss": 2.5108,
      "step": 10857
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6671527028083801,
      "learning_rate": 0.0005967181328603395,
      "loss": 2.5896,
      "step": 10858
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6696211099624634,
      "learning_rate": 0.0005967175294305996,
      "loss": 2.5544,
      "step": 10859
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6180935502052307,
      "learning_rate": 0.0005967169259456943,
      "loss": 2.4157,
      "step": 10860
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6011789441108704,
      "learning_rate": 0.0005967163224056236,
      "loss": 2.3831,
      "step": 10861
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6769741177558899,
      "learning_rate": 0.000596715718810388,
      "loss": 2.4632,
      "step": 10862
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6340346932411194,
      "learning_rate": 0.0005967151151599872,
      "loss": 2.5101,
      "step": 10863
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6315608620643616,
      "learning_rate": 0.0005967145114544215,
      "loss": 2.3643,
      "step": 10864
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6743994355201721,
      "learning_rate": 0.0005967139076936908,
      "loss": 2.3123,
      "step": 10865
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6204217672348022,
      "learning_rate": 0.0005967133038777956,
      "loss": 2.3121,
      "step": 10866
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.679336428642273,
      "learning_rate": 0.0005967127000067357,
      "loss": 2.6908,
      "step": 10867
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.7389911413192749,
      "learning_rate": 0.0005967120960805114,
      "loss": 2.2076,
      "step": 10868
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6403916478157043,
      "learning_rate": 0.0005967114920991226,
      "loss": 2.6099,
      "step": 10869
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.807701051235199,
      "learning_rate": 0.0005967108880625696,
      "loss": 2.5686,
      "step": 10870
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6603949666023254,
      "learning_rate": 0.0005967102839708524,
      "loss": 2.5647,
      "step": 10871
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.8060441017150879,
      "learning_rate": 0.0005967096798239713,
      "loss": 2.6201,
      "step": 10872
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6430435180664062,
      "learning_rate": 0.000596709075621926,
      "loss": 2.5474,
      "step": 10873
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6367698907852173,
      "learning_rate": 0.000596708471364717,
      "loss": 2.3999,
      "step": 10874
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6230854988098145,
      "learning_rate": 0.0005967078670523445,
      "loss": 2.4931,
      "step": 10875
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.5628519654273987,
      "learning_rate": 0.0005967072626848082,
      "loss": 2.4036,
      "step": 10876
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6302041411399841,
      "learning_rate": 0.0005967066582621085,
      "loss": 2.2622,
      "step": 10877
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6067613363265991,
      "learning_rate": 0.0005967060537842453,
      "loss": 2.5268,
      "step": 10878
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6064058542251587,
      "learning_rate": 0.000596705449251219,
      "loss": 2.3458,
      "step": 10879
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.5769565105438232,
      "learning_rate": 0.0005967048446630295,
      "loss": 2.4543,
      "step": 10880
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.7178485989570618,
      "learning_rate": 0.0005967042400196769,
      "loss": 2.3942,
      "step": 10881
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6412632465362549,
      "learning_rate": 0.0005967036353211614,
      "loss": 2.4682,
      "step": 10882
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.797990083694458,
      "learning_rate": 0.0005967030305674832,
      "loss": 2.5203,
      "step": 10883
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6268919110298157,
      "learning_rate": 0.0005967024257586422,
      "loss": 2.47,
      "step": 10884
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.7650843858718872,
      "learning_rate": 0.0005967018208946387,
      "loss": 2.5323,
      "step": 10885
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6801800727844238,
      "learning_rate": 0.0005967012159754726,
      "loss": 2.3599,
      "step": 10886
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.63343745470047,
      "learning_rate": 0.0005967006110011442,
      "loss": 2.43,
      "step": 10887
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.7369970679283142,
      "learning_rate": 0.0005967000059716536,
      "loss": 2.447,
      "step": 10888
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.7076720595359802,
      "learning_rate": 0.0005966994008870009,
      "loss": 2.4581,
      "step": 10889
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.7337560057640076,
      "learning_rate": 0.0005966987957471861,
      "loss": 2.4277,
      "step": 10890
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.889514684677124,
      "learning_rate": 0.0005966981905522095,
      "loss": 2.5475,
      "step": 10891
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.7054124474525452,
      "learning_rate": 0.000596697585302071,
      "loss": 2.4134,
      "step": 10892
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6317917108535767,
      "learning_rate": 0.0005966969799967708,
      "loss": 2.6674,
      "step": 10893
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.8108214735984802,
      "learning_rate": 0.0005966963746363092,
      "loss": 2.4719,
      "step": 10894
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.7849704027175903,
      "learning_rate": 0.000596695769220686,
      "loss": 2.4729,
      "step": 10895
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6442909836769104,
      "learning_rate": 0.0005966951637499015,
      "loss": 2.6622,
      "step": 10896
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6835851669311523,
      "learning_rate": 0.0005966945582239557,
      "loss": 2.4508,
      "step": 10897
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.7802690863609314,
      "learning_rate": 0.0005966939526428489,
      "loss": 2.485,
      "step": 10898
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.8189159631729126,
      "learning_rate": 0.0005966933470065809,
      "loss": 2.544,
      "step": 10899
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.5985653400421143,
      "learning_rate": 0.0005966927413151522,
      "loss": 2.5493,
      "step": 10900
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6323816180229187,
      "learning_rate": 0.0005966921355685627,
      "loss": 2.464,
      "step": 10901
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.0480659008026123,
      "learning_rate": 0.0005966915297668124,
      "loss": 2.5427,
      "step": 10902
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6473625898361206,
      "learning_rate": 0.0005966909239099016,
      "loss": 2.6324,
      "step": 10903
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6935660243034363,
      "learning_rate": 0.0005966903179978304,
      "loss": 2.563,
      "step": 10904
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.9788943529129028,
      "learning_rate": 0.0005966897120305988,
      "loss": 2.4052,
      "step": 10905
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.8124904036521912,
      "learning_rate": 0.0005966891060082072,
      "loss": 2.6782,
      "step": 10906
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6074514389038086,
      "learning_rate": 0.0005966884999306552,
      "loss": 2.2933,
      "step": 10907
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.7769947648048401,
      "learning_rate": 0.0005966878937979434,
      "loss": 2.4345,
      "step": 10908
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.877626359462738,
      "learning_rate": 0.0005966872876100717,
      "loss": 2.4011,
      "step": 10909
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.0015027523040771,
      "learning_rate": 0.0005966866813670402,
      "loss": 2.2297,
      "step": 10910
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6972882151603699,
      "learning_rate": 0.000596686075068849,
      "loss": 2.3723,
      "step": 10911
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.7642276287078857,
      "learning_rate": 0.0005966854687154982,
      "loss": 2.3135,
      "step": 10912
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.9174495935440063,
      "learning_rate": 0.0005966848623069882,
      "loss": 2.4807,
      "step": 10913
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.7108278870582581,
      "learning_rate": 0.0005966842558433186,
      "loss": 2.5801,
      "step": 10914
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6020301580429077,
      "learning_rate": 0.00059668364932449,
      "loss": 2.5388,
      "step": 10915
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.9249383211135864,
      "learning_rate": 0.0005966830427505022,
      "loss": 2.3216,
      "step": 10916
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6407445669174194,
      "learning_rate": 0.0005966824361213556,
      "loss": 2.6236,
      "step": 10917
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.5604188442230225,
      "learning_rate": 0.0005966818294370499,
      "loss": 2.3992,
      "step": 10918
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.5984171032905579,
      "learning_rate": 0.0005966812226975856,
      "loss": 2.6704,
      "step": 10919
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6439564228057861,
      "learning_rate": 0.0005966806159029627,
      "loss": 2.4922,
      "step": 10920
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6726202368736267,
      "learning_rate": 0.0005966800090531811,
      "loss": 2.3949,
      "step": 10921
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6251885890960693,
      "learning_rate": 0.0005966794021482412,
      "loss": 2.4265,
      "step": 10922
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6504806280136108,
      "learning_rate": 0.000596678795188143,
      "loss": 2.3427,
      "step": 10923
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6211374402046204,
      "learning_rate": 0.0005966781881728866,
      "loss": 2.4762,
      "step": 10924
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6652228832244873,
      "learning_rate": 0.000596677581102472,
      "loss": 2.5229,
      "step": 10925
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6504983305931091,
      "learning_rate": 0.0005966769739768997,
      "loss": 2.3186,
      "step": 10926
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6537336111068726,
      "learning_rate": 0.0005966763667961694,
      "loss": 2.4889,
      "step": 10927
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6286065578460693,
      "learning_rate": 0.0005966757595602813,
      "loss": 2.4191,
      "step": 10928
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6089401245117188,
      "learning_rate": 0.0005966751522692356,
      "loss": 2.4705,
      "step": 10929
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6311361193656921,
      "learning_rate": 0.0005966745449230324,
      "loss": 2.3206,
      "step": 10930
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.677901566028595,
      "learning_rate": 0.0005966739375216719,
      "loss": 2.5335,
      "step": 10931
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6108775734901428,
      "learning_rate": 0.000596673330065154,
      "loss": 2.6304,
      "step": 10932
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.5859431624412537,
      "learning_rate": 0.0005966727225534791,
      "loss": 2.2601,
      "step": 10933
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.5883254408836365,
      "learning_rate": 0.000596672114986647,
      "loss": 2.3891,
      "step": 10934
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6685048937797546,
      "learning_rate": 0.0005966715073646579,
      "loss": 2.5219,
      "step": 10935
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6896971464157104,
      "learning_rate": 0.0005966708996875121,
      "loss": 2.4793,
      "step": 10936
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6584489345550537,
      "learning_rate": 0.0005966702919552095,
      "loss": 2.4138,
      "step": 10937
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6369951367378235,
      "learning_rate": 0.0005966696841677503,
      "loss": 2.5149,
      "step": 10938
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6233928203582764,
      "learning_rate": 0.0005966690763251346,
      "loss": 2.2958,
      "step": 10939
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6171474456787109,
      "learning_rate": 0.0005966684684273625,
      "loss": 2.403,
      "step": 10940
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6008889079093933,
      "learning_rate": 0.0005966678604744342,
      "loss": 2.4036,
      "step": 10941
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.635013222694397,
      "learning_rate": 0.0005966672524663496,
      "loss": 2.1933,
      "step": 10942
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.582780659198761,
      "learning_rate": 0.0005966666444031091,
      "loss": 2.4203,
      "step": 10943
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6342353224754333,
      "learning_rate": 0.0005966660362847127,
      "loss": 2.3169,
      "step": 10944
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.7547576427459717,
      "learning_rate": 0.0005966654281111604,
      "loss": 2.6159,
      "step": 10945
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.649514377117157,
      "learning_rate": 0.0005966648198824524,
      "loss": 2.2881,
      "step": 10946
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6075389981269836,
      "learning_rate": 0.0005966642115985888,
      "loss": 2.4654,
      "step": 10947
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.587086021900177,
      "learning_rate": 0.0005966636032595698,
      "loss": 2.2537,
      "step": 10948
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.7521639466285706,
      "learning_rate": 0.0005966629948653954,
      "loss": 2.2817,
      "step": 10949
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6079087257385254,
      "learning_rate": 0.0005966623864160658,
      "loss": 2.4903,
      "step": 10950
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.82796710729599,
      "learning_rate": 0.0005966617779115809,
      "loss": 2.6879,
      "step": 10951
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6960907578468323,
      "learning_rate": 0.0005966611693519411,
      "loss": 2.451,
      "step": 10952
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.5899884104728699,
      "learning_rate": 0.0005966605607371463,
      "loss": 2.4083,
      "step": 10953
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.8136417865753174,
      "learning_rate": 0.0005966599520671968,
      "loss": 2.3908,
      "step": 10954
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.685962438583374,
      "learning_rate": 0.0005966593433420926,
      "loss": 2.4333,
      "step": 10955
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6115421652793884,
      "learning_rate": 0.0005966587345618338,
      "loss": 2.4824,
      "step": 10956
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6435356140136719,
      "learning_rate": 0.0005966581257264205,
      "loss": 2.419,
      "step": 10957
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6507173776626587,
      "learning_rate": 0.000596657516835853,
      "loss": 2.6874,
      "step": 10958
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.5862829089164734,
      "learning_rate": 0.0005966569078901311,
      "loss": 2.5458,
      "step": 10959
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.5888246893882751,
      "learning_rate": 0.0005966562988892551,
      "loss": 2.4366,
      "step": 10960
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6854004859924316,
      "learning_rate": 0.0005966556898332252,
      "loss": 2.3319,
      "step": 10961
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.5865437984466553,
      "learning_rate": 0.0005966550807220413,
      "loss": 2.6078,
      "step": 10962
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6669836640357971,
      "learning_rate": 0.0005966544715557037,
      "loss": 2.8256,
      "step": 10963
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.5959824919700623,
      "learning_rate": 0.0005966538623342124,
      "loss": 2.5135,
      "step": 10964
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.624382495880127,
      "learning_rate": 0.0005966532530575676,
      "loss": 2.4777,
      "step": 10965
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6265516877174377,
      "learning_rate": 0.0005966526437257693,
      "loss": 2.3143,
      "step": 10966
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6377465724945068,
      "learning_rate": 0.0005966520343388177,
      "loss": 2.3538,
      "step": 10967
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.7054502964019775,
      "learning_rate": 0.0005966514248967129,
      "loss": 2.5741,
      "step": 10968
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.5941289663314819,
      "learning_rate": 0.000596650815399455,
      "loss": 2.1309,
      "step": 10969
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.694978654384613,
      "learning_rate": 0.0005966502058470441,
      "loss": 2.5406,
      "step": 10970
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6232943534851074,
      "learning_rate": 0.0005966495962394803,
      "loss": 2.397,
      "step": 10971
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6902921199798584,
      "learning_rate": 0.0005966489865767637,
      "loss": 2.6344,
      "step": 10972
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.7360005378723145,
      "learning_rate": 0.0005966483768588946,
      "loss": 2.6818,
      "step": 10973
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.7220354080200195,
      "learning_rate": 0.0005966477670858729,
      "loss": 2.5068,
      "step": 10974
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6661038398742676,
      "learning_rate": 0.0005966471572576987,
      "loss": 2.3905,
      "step": 10975
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.8202484250068665,
      "learning_rate": 0.0005966465473743722,
      "loss": 2.5977,
      "step": 10976
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6101971864700317,
      "learning_rate": 0.0005966459374358936,
      "loss": 2.4616,
      "step": 10977
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.573498010635376,
      "learning_rate": 0.0005966453274422629,
      "loss": 2.2461,
      "step": 10978
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6966973543167114,
      "learning_rate": 0.0005966447173934801,
      "loss": 2.4037,
      "step": 10979
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6786918640136719,
      "learning_rate": 0.0005966441072895456,
      "loss": 2.6064,
      "step": 10980
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6196733117103577,
      "learning_rate": 0.0005966434971304594,
      "loss": 2.53,
      "step": 10981
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6288759112358093,
      "learning_rate": 0.0005966428869162214,
      "loss": 2.4918,
      "step": 10982
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6627954244613647,
      "learning_rate": 0.000596642276646832,
      "loss": 2.4723,
      "step": 10983
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.5317959189414978,
      "learning_rate": 0.0005966416663222913,
      "loss": 2.2202,
      "step": 10984
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.7040538787841797,
      "learning_rate": 0.0005966410559425992,
      "loss": 2.1799,
      "step": 10985
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6027804017066956,
      "learning_rate": 0.000596640445507756,
      "loss": 2.4202,
      "step": 10986
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6750354766845703,
      "learning_rate": 0.0005966398350177616,
      "loss": 2.5517,
      "step": 10987
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.582576334476471,
      "learning_rate": 0.0005966392244726163,
      "loss": 2.5123,
      "step": 10988
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6514717936515808,
      "learning_rate": 0.0005966386138723202,
      "loss": 2.5402,
      "step": 10989
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6378254890441895,
      "learning_rate": 0.0005966380032168733,
      "loss": 2.3375,
      "step": 10990
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.625555157661438,
      "learning_rate": 0.0005966373925062759,
      "loss": 2.4336,
      "step": 10991
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6001371741294861,
      "learning_rate": 0.0005966367817405281,
      "loss": 2.3268,
      "step": 10992
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6442163586616516,
      "learning_rate": 0.0005966361709196298,
      "loss": 2.5704,
      "step": 10993
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6454977989196777,
      "learning_rate": 0.0005966355600435812,
      "loss": 2.6536,
      "step": 10994
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6250717043876648,
      "learning_rate": 0.0005966349491123824,
      "loss": 2.4333,
      "step": 10995
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6746440529823303,
      "learning_rate": 0.0005966343381260338,
      "loss": 2.4191,
      "step": 10996
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.739101767539978,
      "learning_rate": 0.000596633727084535,
      "loss": 2.5082,
      "step": 10997
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.5771949887275696,
      "learning_rate": 0.0005966331159878866,
      "loss": 2.4305,
      "step": 10998
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.5996419787406921,
      "learning_rate": 0.0005966325048360884,
      "loss": 2.3843,
      "step": 10999
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.802688479423523,
      "learning_rate": 0.0005966318936291407,
      "loss": 2.5501,
      "step": 11000
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.5980005264282227,
      "learning_rate": 0.0005966312823670435,
      "loss": 2.761,
      "step": 11001
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.5953645706176758,
      "learning_rate": 0.0005966306710497968,
      "loss": 2.5113,
      "step": 11002
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.7015076279640198,
      "learning_rate": 0.000596630059677401,
      "loss": 2.5645,
      "step": 11003
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.7478089332580566,
      "learning_rate": 0.0005966294482498561,
      "loss": 2.4086,
      "step": 11004
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6250908970832825,
      "learning_rate": 0.0005966288367671622,
      "loss": 2.5953,
      "step": 11005
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6853102445602417,
      "learning_rate": 0.0005966282252293193,
      "loss": 2.38,
      "step": 11006
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.9536029100418091,
      "learning_rate": 0.0005966276136363276,
      "loss": 2.4827,
      "step": 11007
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6321784853935242,
      "learning_rate": 0.0005966270019881874,
      "loss": 2.4763,
      "step": 11008
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6139656901359558,
      "learning_rate": 0.0005966263902848984,
      "loss": 2.5889,
      "step": 11009
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.8041810393333435,
      "learning_rate": 0.0005966257785264611,
      "loss": 2.4856,
      "step": 11010
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.7903731465339661,
      "learning_rate": 0.0005966251667128754,
      "loss": 2.4037,
      "step": 11011
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6543002724647522,
      "learning_rate": 0.0005966245548441416,
      "loss": 2.693,
      "step": 11012
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6361592411994934,
      "learning_rate": 0.0005966239429202596,
      "loss": 2.3278,
      "step": 11013
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.8157806396484375,
      "learning_rate": 0.0005966233309412296,
      "loss": 2.5424,
      "step": 11014
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.7236446738243103,
      "learning_rate": 0.0005966227189070518,
      "loss": 2.5981,
      "step": 11015
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.5732444524765015,
      "learning_rate": 0.0005966221068177262,
      "loss": 2.3556,
      "step": 11016
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.752033531665802,
      "learning_rate": 0.000596621494673253,
      "loss": 2.3605,
      "step": 11017
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.717888593673706,
      "learning_rate": 0.0005966208824736321,
      "loss": 2.3769,
      "step": 11018
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6057502627372742,
      "learning_rate": 0.000596620270218864,
      "loss": 2.6569,
      "step": 11019
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.5932708382606506,
      "learning_rate": 0.0005966196579089484,
      "loss": 2.4019,
      "step": 11020
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.8401192426681519,
      "learning_rate": 0.0005966190455438857,
      "loss": 2.5115,
      "step": 11021
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.7810575366020203,
      "learning_rate": 0.0005966184331236759,
      "loss": 2.4601,
      "step": 11022
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.5694762468338013,
      "learning_rate": 0.0005966178206483192,
      "loss": 2.242,
      "step": 11023
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6775099635124207,
      "learning_rate": 0.0005966172081178155,
      "loss": 2.3083,
      "step": 11024
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6033853888511658,
      "learning_rate": 0.0005966165955321653,
      "loss": 2.3676,
      "step": 11025
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6469272971153259,
      "learning_rate": 0.0005966159828913682,
      "loss": 2.5158,
      "step": 11026
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6931873559951782,
      "learning_rate": 0.0005966153701954248,
      "loss": 2.4715,
      "step": 11027
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6580382585525513,
      "learning_rate": 0.0005966147574443349,
      "loss": 2.5342,
      "step": 11028
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.5844393968582153,
      "learning_rate": 0.0005966141446380987,
      "loss": 2.3144,
      "step": 11029
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6384482979774475,
      "learning_rate": 0.0005966135317767164,
      "loss": 2.3289,
      "step": 11030
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.614240288734436,
      "learning_rate": 0.000596612918860188,
      "loss": 2.3013,
      "step": 11031
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6605213284492493,
      "learning_rate": 0.0005966123058885137,
      "loss": 2.414,
      "step": 11032
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6482704281806946,
      "learning_rate": 0.0005966116928616935,
      "loss": 2.7296,
      "step": 11033
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6307321786880493,
      "learning_rate": 0.0005966110797797276,
      "loss": 2.4277,
      "step": 11034
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.597907543182373,
      "learning_rate": 0.0005966104666426162,
      "loss": 2.5519,
      "step": 11035
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6287845373153687,
      "learning_rate": 0.0005966098534503592,
      "loss": 2.494,
      "step": 11036
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.5983588695526123,
      "learning_rate": 0.0005966092402029569,
      "loss": 2.6249,
      "step": 11037
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6453049778938293,
      "learning_rate": 0.0005966086269004094,
      "loss": 2.4711,
      "step": 11038
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6011337041854858,
      "learning_rate": 0.0005966080135427166,
      "loss": 2.4886,
      "step": 11039
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6182146072387695,
      "learning_rate": 0.000596607400129879,
      "loss": 2.5421,
      "step": 11040
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6168204545974731,
      "learning_rate": 0.0005966067866618963,
      "loss": 2.3689,
      "step": 11041
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6289331912994385,
      "learning_rate": 0.0005966061731387688,
      "loss": 2.5164,
      "step": 11042
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6466227173805237,
      "learning_rate": 0.0005966055595604966,
      "loss": 2.5289,
      "step": 11043
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6142280697822571,
      "learning_rate": 0.00059660494592708,
      "loss": 2.5946,
      "step": 11044
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.605426549911499,
      "learning_rate": 0.0005966043322385187,
      "loss": 2.5529,
      "step": 11045
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.5982348322868347,
      "learning_rate": 0.0005966037184948131,
      "loss": 2.3949,
      "step": 11046
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.599589467048645,
      "learning_rate": 0.0005966031046959634,
      "loss": 2.4719,
      "step": 11047
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6174471974372864,
      "learning_rate": 0.0005966024908419695,
      "loss": 2.3197,
      "step": 11048
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.7412441372871399,
      "learning_rate": 0.0005966018769328318,
      "loss": 2.359,
      "step": 11049
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6456069946289062,
      "learning_rate": 0.0005966012629685499,
      "loss": 2.4965,
      "step": 11050
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6101399064064026,
      "learning_rate": 0.0005966006489491244,
      "loss": 2.5619,
      "step": 11051
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6723490357398987,
      "learning_rate": 0.0005966000348745553,
      "loss": 2.423,
      "step": 11052
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.571120023727417,
      "learning_rate": 0.0005965994207448425,
      "loss": 2.7357,
      "step": 11053
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6767477989196777,
      "learning_rate": 0.0005965988065599863,
      "loss": 2.4594,
      "step": 11054
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6797743439674377,
      "learning_rate": 0.0005965981923199868,
      "loss": 2.3876,
      "step": 11055
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.604901134967804,
      "learning_rate": 0.0005965975780248442,
      "loss": 2.4641,
      "step": 11056
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.7439126968383789,
      "learning_rate": 0.0005965969636745584,
      "loss": 2.3638,
      "step": 11057
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.5721115469932556,
      "learning_rate": 0.0005965963492691297,
      "loss": 2.4236,
      "step": 11058
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6358751058578491,
      "learning_rate": 0.000596595734808558,
      "loss": 2.5166,
      "step": 11059
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.646169900894165,
      "learning_rate": 0.0005965951202928437,
      "loss": 2.7665,
      "step": 11060
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.7014162540435791,
      "learning_rate": 0.0005965945057219867,
      "loss": 2.4266,
      "step": 11061
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6201635599136353,
      "learning_rate": 0.0005965938910959871,
      "loss": 2.2186,
      "step": 11062
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6454347968101501,
      "learning_rate": 0.0005965932764148453,
      "loss": 2.3728,
      "step": 11063
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6029136180877686,
      "learning_rate": 0.0005965926616785611,
      "loss": 2.7187,
      "step": 11064
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.5998747944831848,
      "learning_rate": 0.0005965920468871348,
      "loss": 2.4833,
      "step": 11065
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.7164855599403381,
      "learning_rate": 0.0005965914320405662,
      "loss": 2.5981,
      "step": 11066
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6864527463912964,
      "learning_rate": 0.0005965908171388559,
      "loss": 2.5492,
      "step": 11067
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6592985391616821,
      "learning_rate": 0.0005965902021820037,
      "loss": 2.4841,
      "step": 11068
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6158699989318848,
      "learning_rate": 0.0005965895871700098,
      "loss": 2.3632,
      "step": 11069
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.7545741200447083,
      "learning_rate": 0.0005965889721028742,
      "loss": 2.3423,
      "step": 11070
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6216909289360046,
      "learning_rate": 0.0005965883569805972,
      "loss": 2.6118,
      "step": 11071
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.5666261911392212,
      "learning_rate": 0.0005965877418031789,
      "loss": 2.3777,
      "step": 11072
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6521636247634888,
      "learning_rate": 0.0005965871265706192,
      "loss": 2.3592,
      "step": 11073
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6968585848808289,
      "learning_rate": 0.0005965865112829183,
      "loss": 2.4413,
      "step": 11074
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6650688648223877,
      "learning_rate": 0.0005965858959400766,
      "loss": 2.553,
      "step": 11075
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.733379065990448,
      "learning_rate": 0.0005965852805420937,
      "loss": 2.5375,
      "step": 11076
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6817901730537415,
      "learning_rate": 0.0005965846650889703,
      "loss": 2.3508,
      "step": 11077
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6092878580093384,
      "learning_rate": 0.000596584049580706,
      "loss": 2.4693,
      "step": 11078
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.7160388827323914,
      "learning_rate": 0.0005965834340173013,
      "loss": 2.586,
      "step": 11079
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6733819842338562,
      "learning_rate": 0.000596582818398756,
      "loss": 2.3422,
      "step": 11080
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.7988100051879883,
      "learning_rate": 0.0005965822027250704,
      "loss": 2.395,
      "step": 11081
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.7596462965011597,
      "learning_rate": 0.0005965815869962446,
      "loss": 2.5655,
      "step": 11082
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.8964523077011108,
      "learning_rate": 0.0005965809712122786,
      "loss": 2.7767,
      "step": 11083
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.645675539970398,
      "learning_rate": 0.0005965803553731726,
      "loss": 2.0524,
      "step": 11084
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6226573586463928,
      "learning_rate": 0.0005965797394789268,
      "loss": 2.2441,
      "step": 11085
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.766981303691864,
      "learning_rate": 0.0005965791235295412,
      "loss": 2.4843,
      "step": 11086
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6549212336540222,
      "learning_rate": 0.0005965785075250159,
      "loss": 2.5404,
      "step": 11087
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.5657612681388855,
      "learning_rate": 0.000596577891465351,
      "loss": 2.2734,
      "step": 11088
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6047368049621582,
      "learning_rate": 0.0005965772753505468,
      "loss": 2.164,
      "step": 11089
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.778380811214447,
      "learning_rate": 0.0005965766591806034,
      "loss": 2.45,
      "step": 11090
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.5799656510353088,
      "learning_rate": 0.0005965760429555205,
      "loss": 2.2893,
      "step": 11091
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6304563879966736,
      "learning_rate": 0.0005965754266752987,
      "loss": 2.4234,
      "step": 11092
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6597208380699158,
      "learning_rate": 0.0005965748103399378,
      "loss": 2.5278,
      "step": 11093
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6051527857780457,
      "learning_rate": 0.0005965741939494382,
      "loss": 2.6551,
      "step": 11094
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.791696310043335,
      "learning_rate": 0.0005965735775037999,
      "loss": 2.5093,
      "step": 11095
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6897534728050232,
      "learning_rate": 0.0005965729610030228,
      "loss": 2.6674,
      "step": 11096
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.7263342142105103,
      "learning_rate": 0.0005965723444471072,
      "loss": 2.4864,
      "step": 11097
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6419054269790649,
      "learning_rate": 0.0005965717278360533,
      "loss": 2.4262,
      "step": 11098
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6421425938606262,
      "learning_rate": 0.000596571111169861,
      "loss": 2.5262,
      "step": 11099
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.7409632802009583,
      "learning_rate": 0.0005965704944485307,
      "loss": 2.485,
      "step": 11100
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6195642352104187,
      "learning_rate": 0.0005965698776720622,
      "loss": 2.5587,
      "step": 11101
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6201468110084534,
      "learning_rate": 0.0005965692608404559,
      "loss": 2.3725,
      "step": 11102
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.5626359581947327,
      "learning_rate": 0.0005965686439537117,
      "loss": 2.487,
      "step": 11103
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6514777541160583,
      "learning_rate": 0.0005965680270118296,
      "loss": 2.3478,
      "step": 11104
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.5935809016227722,
      "learning_rate": 0.0005965674100148102,
      "loss": 2.3994,
      "step": 11105
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6650243997573853,
      "learning_rate": 0.0005965667929626532,
      "loss": 2.5297,
      "step": 11106
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.694521963596344,
      "learning_rate": 0.0005965661758553589,
      "loss": 2.3735,
      "step": 11107
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6399654746055603,
      "learning_rate": 0.0005965655586929273,
      "loss": 2.4691,
      "step": 11108
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.5906069278717041,
      "learning_rate": 0.0005965649414753586,
      "loss": 2.6399,
      "step": 11109
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6866975426673889,
      "learning_rate": 0.0005965643242026528,
      "loss": 2.503,
      "step": 11110
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.7456153631210327,
      "learning_rate": 0.0005965637068748101,
      "loss": 2.3268,
      "step": 11111
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6935821771621704,
      "learning_rate": 0.0005965630894918307,
      "loss": 2.4094,
      "step": 11112
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6705050468444824,
      "learning_rate": 0.0005965624720537145,
      "loss": 2.2894,
      "step": 11113
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6743288636207581,
      "learning_rate": 0.0005965618545604618,
      "loss": 2.2122,
      "step": 11114
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6066238880157471,
      "learning_rate": 0.0005965612370120727,
      "loss": 2.3093,
      "step": 11115
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6855326294898987,
      "learning_rate": 0.0005965606194085472,
      "loss": 2.5447,
      "step": 11116
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6380661129951477,
      "learning_rate": 0.0005965600017498855,
      "loss": 2.7364,
      "step": 11117
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6359298825263977,
      "learning_rate": 0.0005965593840360877,
      "loss": 2.4617,
      "step": 11118
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.5855011940002441,
      "learning_rate": 0.0005965587662671539,
      "loss": 2.6589,
      "step": 11119
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6285533308982849,
      "learning_rate": 0.0005965581484430843,
      "loss": 2.3754,
      "step": 11120
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6739700436592102,
      "learning_rate": 0.0005965575305638789,
      "loss": 2.3702,
      "step": 11121
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.619804859161377,
      "learning_rate": 0.0005965569126295378,
      "loss": 2.5887,
      "step": 11122
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6106443405151367,
      "learning_rate": 0.0005965562946400612,
      "loss": 2.453,
      "step": 11123
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6739941835403442,
      "learning_rate": 0.0005965556765954492,
      "loss": 2.7091,
      "step": 11124
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.61541748046875,
      "learning_rate": 0.0005965550584957019,
      "loss": 2.6553,
      "step": 11125
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6922603249549866,
      "learning_rate": 0.0005965544403408195,
      "loss": 2.5788,
      "step": 11126
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6019058227539062,
      "learning_rate": 0.0005965538221308018,
      "loss": 2.3515,
      "step": 11127
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.7214269042015076,
      "learning_rate": 0.0005965532038656494,
      "loss": 2.361,
      "step": 11128
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6412522792816162,
      "learning_rate": 0.0005965525855453621,
      "loss": 2.4724,
      "step": 11129
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6543062925338745,
      "learning_rate": 0.00059655196716994,
      "loss": 2.5472,
      "step": 11130
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6592717170715332,
      "learning_rate": 0.0005965513487393834,
      "loss": 2.2691,
      "step": 11131
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.7228254675865173,
      "learning_rate": 0.0005965507302536922,
      "loss": 2.3252,
      "step": 11132
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.7456545233726501,
      "learning_rate": 0.0005965501117128667,
      "loss": 2.4129,
      "step": 11133
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6325293183326721,
      "learning_rate": 0.0005965494931169068,
      "loss": 2.4461,
      "step": 11134
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.7850411534309387,
      "learning_rate": 0.0005965488744658128,
      "loss": 2.5198,
      "step": 11135
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.7206006646156311,
      "learning_rate": 0.0005965482557595849,
      "loss": 2.5965,
      "step": 11136
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.838465690612793,
      "learning_rate": 0.000596547636998223,
      "loss": 2.4742,
      "step": 11137
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6390239000320435,
      "learning_rate": 0.0005965470181817273,
      "loss": 2.458,
      "step": 11138
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.680127739906311,
      "learning_rate": 0.0005965463993100979,
      "loss": 2.5833,
      "step": 11139
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.7413803935050964,
      "learning_rate": 0.0005965457803833349,
      "loss": 2.5939,
      "step": 11140
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.8770302534103394,
      "learning_rate": 0.0005965451614014384,
      "loss": 2.2384,
      "step": 11141
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.7101647257804871,
      "learning_rate": 0.0005965445423644087,
      "loss": 2.5374,
      "step": 11142
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.7507122159004211,
      "learning_rate": 0.0005965439232722457,
      "loss": 2.6946,
      "step": 11143
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6736733913421631,
      "learning_rate": 0.0005965433041249496,
      "loss": 2.4565,
      "step": 11144
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6040743589401245,
      "learning_rate": 0.0005965426849225204,
      "loss": 2.5466,
      "step": 11145
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6760327219963074,
      "learning_rate": 0.0005965420656649585,
      "loss": 2.5146,
      "step": 11146
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.676615297794342,
      "learning_rate": 0.0005965414463522637,
      "loss": 2.6594,
      "step": 11147
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6572498083114624,
      "learning_rate": 0.0005965408269844362,
      "loss": 2.5714,
      "step": 11148
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.622380256652832,
      "learning_rate": 0.0005965402075614764,
      "loss": 2.4475,
      "step": 11149
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.5995519161224365,
      "learning_rate": 0.000596539588083384,
      "loss": 2.4487,
      "step": 11150
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6140119433403015,
      "learning_rate": 0.0005965389685501592,
      "loss": 2.3825,
      "step": 11151
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6170735359191895,
      "learning_rate": 0.0005965383489618025,
      "loss": 2.6227,
      "step": 11152
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.5907945036888123,
      "learning_rate": 0.0005965377293183135,
      "loss": 2.5115,
      "step": 11153
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.5830386281013489,
      "learning_rate": 0.0005965371096196925,
      "loss": 2.5337,
      "step": 11154
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6333519220352173,
      "learning_rate": 0.0005965364898659397,
      "loss": 2.5672,
      "step": 11155
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.5723732113838196,
      "learning_rate": 0.0005965358700570552,
      "loss": 2.4953,
      "step": 11156
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.713661789894104,
      "learning_rate": 0.0005965352501930392,
      "loss": 2.4008,
      "step": 11157
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6428641676902771,
      "learning_rate": 0.0005965346302738916,
      "loss": 2.3637,
      "step": 11158
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6050410270690918,
      "learning_rate": 0.0005965340102996126,
      "loss": 2.3512,
      "step": 11159
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6412469148635864,
      "learning_rate": 0.0005965333902702023,
      "loss": 2.5083,
      "step": 11160
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6330875754356384,
      "learning_rate": 0.0005965327701856609,
      "loss": 2.3083,
      "step": 11161
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.5982580780982971,
      "learning_rate": 0.0005965321500459884,
      "loss": 2.4631,
      "step": 11162
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.702294111251831,
      "learning_rate": 0.000596531529851185,
      "loss": 2.548,
      "step": 11163
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.5673443675041199,
      "learning_rate": 0.0005965309096012508,
      "loss": 2.3278,
      "step": 11164
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6151697039604187,
      "learning_rate": 0.000596530289296186,
      "loss": 2.2495,
      "step": 11165
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6764315962791443,
      "learning_rate": 0.0005965296689359905,
      "loss": 2.3936,
      "step": 11166
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6316456198692322,
      "learning_rate": 0.0005965290485206646,
      "loss": 2.4538,
      "step": 11167
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.7131786942481995,
      "learning_rate": 0.0005965284280502084,
      "loss": 2.5894,
      "step": 11168
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6038476228713989,
      "learning_rate": 0.0005965278075246218,
      "loss": 2.6564,
      "step": 11169
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.66449373960495,
      "learning_rate": 0.0005965271869439053,
      "loss": 2.5582,
      "step": 11170
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.8798317909240723,
      "learning_rate": 0.0005965265663080585,
      "loss": 2.5556,
      "step": 11171
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6008279919624329,
      "learning_rate": 0.000596525945617082,
      "loss": 2.5279,
      "step": 11172
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6281502842903137,
      "learning_rate": 0.0005965253248709758,
      "loss": 2.3853,
      "step": 11173
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6818780303001404,
      "learning_rate": 0.0005965247040697399,
      "loss": 2.3007,
      "step": 11174
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6768479943275452,
      "learning_rate": 0.0005965240832133744,
      "loss": 2.3411,
      "step": 11175
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.5626506209373474,
      "learning_rate": 0.0005965234623018796,
      "loss": 2.4705,
      "step": 11176
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6768739223480225,
      "learning_rate": 0.0005965228413352553,
      "loss": 2.7608,
      "step": 11177
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6081969141960144,
      "learning_rate": 0.000596522220313502,
      "loss": 2.4878,
      "step": 11178
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6281827688217163,
      "learning_rate": 0.0005965215992366194,
      "loss": 2.6967,
      "step": 11179
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6096413135528564,
      "learning_rate": 0.000596520978104608,
      "loss": 2.3807,
      "step": 11180
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6908275485038757,
      "learning_rate": 0.0005965203569174676,
      "loss": 2.4258,
      "step": 11181
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6091203093528748,
      "learning_rate": 0.0005965197356751987,
      "loss": 2.4779,
      "step": 11182
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6656270623207092,
      "learning_rate": 0.0005965191143778011,
      "loss": 2.3193,
      "step": 11183
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6631394028663635,
      "learning_rate": 0.0005965184930252749,
      "loss": 2.2454,
      "step": 11184
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6343845725059509,
      "learning_rate": 0.0005965178716176204,
      "loss": 2.5406,
      "step": 11185
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.5764690041542053,
      "learning_rate": 0.0005965172501548376,
      "loss": 2.3908,
      "step": 11186
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.5834968090057373,
      "learning_rate": 0.0005965166286369266,
      "loss": 2.5587,
      "step": 11187
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6362544298171997,
      "learning_rate": 0.0005965160070638876,
      "loss": 2.4852,
      "step": 11188
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6860610842704773,
      "learning_rate": 0.0005965153854357208,
      "loss": 2.4282,
      "step": 11189
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6500458121299744,
      "learning_rate": 0.000596514763752426,
      "loss": 2.2793,
      "step": 11190
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6925369501113892,
      "learning_rate": 0.0005965141420140037,
      "loss": 2.3197,
      "step": 11191
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6444427967071533,
      "learning_rate": 0.0005965135202204536,
      "loss": 2.4249,
      "step": 11192
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.8103059530258179,
      "learning_rate": 0.0005965128983717762,
      "loss": 2.497,
      "step": 11193
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.5911943316459656,
      "learning_rate": 0.0005965122764679715,
      "loss": 2.4323,
      "step": 11194
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6674668788909912,
      "learning_rate": 0.0005965116545090394,
      "loss": 2.3929,
      "step": 11195
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.7116475701332092,
      "learning_rate": 0.0005965110324949803,
      "loss": 2.7452,
      "step": 11196
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.7320455312728882,
      "learning_rate": 0.0005965104104257941,
      "loss": 2.2723,
      "step": 11197
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6371845602989197,
      "learning_rate": 0.0005965097883014811,
      "loss": 2.3732,
      "step": 11198
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.5881354212760925,
      "learning_rate": 0.0005965091661220414,
      "loss": 2.6408,
      "step": 11199
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6838396787643433,
      "learning_rate": 0.0005965085438874749,
      "loss": 2.596,
      "step": 11200
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.604475200176239,
      "learning_rate": 0.000596507921597782,
      "loss": 2.593,
      "step": 11201
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6452457308769226,
      "learning_rate": 0.0005965072992529626,
      "loss": 2.4178,
      "step": 11202
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6279132962226868,
      "learning_rate": 0.0005965066768530169,
      "loss": 2.4528,
      "step": 11203
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.5830195546150208,
      "learning_rate": 0.000596506054397945,
      "loss": 2.5991,
      "step": 11204
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6666715145111084,
      "learning_rate": 0.0005965054318877471,
      "loss": 2.4198,
      "step": 11205
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.562131941318512,
      "learning_rate": 0.0005965048093224232,
      "loss": 2.4197,
      "step": 11206
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6074761748313904,
      "learning_rate": 0.0005965041867019734,
      "loss": 2.5167,
      "step": 11207
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6914757490158081,
      "learning_rate": 0.0005965035640263979,
      "loss": 2.6217,
      "step": 11208
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6157527565956116,
      "learning_rate": 0.0005965029412956969,
      "loss": 2.393,
      "step": 11209
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.585180401802063,
      "learning_rate": 0.0005965023185098703,
      "loss": 2.1508,
      "step": 11210
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6951733827590942,
      "learning_rate": 0.0005965016956689185,
      "loss": 2.5703,
      "step": 11211
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6638120412826538,
      "learning_rate": 0.0005965010727728412,
      "loss": 2.4811,
      "step": 11212
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6580566167831421,
      "learning_rate": 0.0005965004498216388,
      "loss": 2.5119,
      "step": 11213
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6372485756874084,
      "learning_rate": 0.0005964998268153114,
      "loss": 2.5411,
      "step": 11214
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6940471529960632,
      "learning_rate": 0.0005964992037538591,
      "loss": 2.6346,
      "step": 11215
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6220663189888,
      "learning_rate": 0.000596498580637282,
      "loss": 2.4246,
      "step": 11216
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6476545929908752,
      "learning_rate": 0.0005964979574655802,
      "loss": 2.5481,
      "step": 11217
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6200573444366455,
      "learning_rate": 0.0005964973342387538,
      "loss": 2.3012,
      "step": 11218
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6137760877609253,
      "learning_rate": 0.000596496710956803,
      "loss": 2.5668,
      "step": 11219
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.581587016582489,
      "learning_rate": 0.000596496087619728,
      "loss": 2.5137,
      "step": 11220
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6268777251243591,
      "learning_rate": 0.0005964954642275285,
      "loss": 2.5706,
      "step": 11221
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.5770921111106873,
      "learning_rate": 0.000596494840780205,
      "loss": 2.4906,
      "step": 11222
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.9418737888336182,
      "learning_rate": 0.0005964942172777577,
      "loss": 2.5982,
      "step": 11223
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6960903406143188,
      "learning_rate": 0.0005964935937201863,
      "loss": 2.52,
      "step": 11224
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6099253296852112,
      "learning_rate": 0.0005964929701074912,
      "loss": 2.4337,
      "step": 11225
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6246395111083984,
      "learning_rate": 0.0005964923464396725,
      "loss": 2.3617,
      "step": 11226
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6619659662246704,
      "learning_rate": 0.0005964917227167303,
      "loss": 2.4929,
      "step": 11227
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6345218420028687,
      "learning_rate": 0.0005964910989386647,
      "loss": 2.1681,
      "step": 11228
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6081985235214233,
      "learning_rate": 0.0005964904751054757,
      "loss": 2.5852,
      "step": 11229
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.5907387137413025,
      "learning_rate": 0.0005964898512171636,
      "loss": 2.3149,
      "step": 11230
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6037260293960571,
      "learning_rate": 0.0005964892272737283,
      "loss": 2.7042,
      "step": 11231
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.7727641463279724,
      "learning_rate": 0.0005964886032751702,
      "loss": 2.4417,
      "step": 11232
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.7124708890914917,
      "learning_rate": 0.0005964879792214893,
      "loss": 2.4801,
      "step": 11233
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.558525800704956,
      "learning_rate": 0.0005964873551126856,
      "loss": 2.4287,
      "step": 11234
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6238637566566467,
      "learning_rate": 0.0005964867309487593,
      "loss": 2.3216,
      "step": 11235
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.766105592250824,
      "learning_rate": 0.0005964861067297106,
      "loss": 2.3638,
      "step": 11236
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6454051733016968,
      "learning_rate": 0.0005964854824555395,
      "loss": 2.4186,
      "step": 11237
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.7322096228599548,
      "learning_rate": 0.0005964848581262462,
      "loss": 2.4735,
      "step": 11238
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6548643708229065,
      "learning_rate": 0.0005964842337418307,
      "loss": 2.5511,
      "step": 11239
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.640036940574646,
      "learning_rate": 0.0005964836093022933,
      "loss": 2.5811,
      "step": 11240
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.5805623531341553,
      "learning_rate": 0.0005964829848076338,
      "loss": 2.5176,
      "step": 11241
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.8292458057403564,
      "learning_rate": 0.0005964823602578526,
      "loss": 2.4892,
      "step": 11242
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6213497519493103,
      "learning_rate": 0.0005964817356529498,
      "loss": 2.6083,
      "step": 11243
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.8668631315231323,
      "learning_rate": 0.0005964811109929254,
      "loss": 2.4072,
      "step": 11244
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6119476556777954,
      "learning_rate": 0.0005964804862777796,
      "loss": 2.33,
      "step": 11245
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6210264563560486,
      "learning_rate": 0.0005964798615075124,
      "loss": 2.2044,
      "step": 11246
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6845678687095642,
      "learning_rate": 0.000596479236682124,
      "loss": 2.515,
      "step": 11247
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.663675844669342,
      "learning_rate": 0.0005964786118016146,
      "loss": 2.4895,
      "step": 11248
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.596501350402832,
      "learning_rate": 0.0005964779868659842,
      "loss": 2.4703,
      "step": 11249
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.7242147326469421,
      "learning_rate": 0.0005964773618752329,
      "loss": 2.5247,
      "step": 11250
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.607174277305603,
      "learning_rate": 0.000596476736829361,
      "loss": 2.4934,
      "step": 11251
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.5961015820503235,
      "learning_rate": 0.0005964761117283683,
      "loss": 2.3619,
      "step": 11252
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6139715909957886,
      "learning_rate": 0.0005964754865722552,
      "loss": 2.253,
      "step": 11253
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6320545077323914,
      "learning_rate": 0.0005964748613610216,
      "loss": 2.2053,
      "step": 11254
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.620817244052887,
      "learning_rate": 0.0005964742360946678,
      "loss": 2.3454,
      "step": 11255
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6593802571296692,
      "learning_rate": 0.0005964736107731939,
      "loss": 2.4072,
      "step": 11256
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6751028299331665,
      "learning_rate": 0.0005964729853966,
      "loss": 2.42,
      "step": 11257
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6813238859176636,
      "learning_rate": 0.0005964723599648859,
      "loss": 2.4819,
      "step": 11258
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6904914975166321,
      "learning_rate": 0.0005964717344780523,
      "loss": 2.4267,
      "step": 11259
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6611477732658386,
      "learning_rate": 0.0005964711089360989,
      "loss": 2.4054,
      "step": 11260
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6321795582771301,
      "learning_rate": 0.0005964704833390259,
      "loss": 2.6661,
      "step": 11261
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.5965412259101868,
      "learning_rate": 0.0005964698576868334,
      "loss": 2.3958,
      "step": 11262
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6234661340713501,
      "learning_rate": 0.0005964692319795216,
      "loss": 2.6295,
      "step": 11263
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6323285102844238,
      "learning_rate": 0.0005964686062170906,
      "loss": 2.4221,
      "step": 11264
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.7034164071083069,
      "learning_rate": 0.0005964679803995405,
      "loss": 2.4935,
      "step": 11265
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6494519710540771,
      "learning_rate": 0.0005964673545268713,
      "loss": 2.4806,
      "step": 11266
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.7667366862297058,
      "learning_rate": 0.0005964667285990833,
      "loss": 2.3447,
      "step": 11267
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6864420771598816,
      "learning_rate": 0.0005964661026161765,
      "loss": 2.4521,
      "step": 11268
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.5797712206840515,
      "learning_rate": 0.0005964654765781512,
      "loss": 2.3438,
      "step": 11269
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.614166259765625,
      "learning_rate": 0.0005964648504850072,
      "loss": 2.3285,
      "step": 11270
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6295118927955627,
      "learning_rate": 0.0005964642243367449,
      "loss": 2.6279,
      "step": 11271
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6271460652351379,
      "learning_rate": 0.0005964635981333642,
      "loss": 2.4228,
      "step": 11272
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.7097454071044922,
      "learning_rate": 0.0005964629718748654,
      "loss": 2.434,
      "step": 11273
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.7076077461242676,
      "learning_rate": 0.0005964623455612485,
      "loss": 2.4875,
      "step": 11274
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6438940167427063,
      "learning_rate": 0.0005964617191925137,
      "loss": 2.2855,
      "step": 11275
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6145570874214172,
      "learning_rate": 0.000596461092768661,
      "loss": 2.4951,
      "step": 11276
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.5642693638801575,
      "learning_rate": 0.0005964604662896907,
      "loss": 2.2839,
      "step": 11277
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.5689609050750732,
      "learning_rate": 0.0005964598397556028,
      "loss": 2.2523,
      "step": 11278
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6094170808792114,
      "learning_rate": 0.0005964592131663972,
      "loss": 2.5451,
      "step": 11279
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6865517497062683,
      "learning_rate": 0.0005964585865220744,
      "loss": 2.2971,
      "step": 11280
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.5949928760528564,
      "learning_rate": 0.0005964579598226343,
      "loss": 2.7481,
      "step": 11281
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.596343457698822,
      "learning_rate": 0.0005964573330680771,
      "loss": 2.3298,
      "step": 11282
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.7856374382972717,
      "learning_rate": 0.000596456706258403,
      "loss": 2.5635,
      "step": 11283
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6206802129745483,
      "learning_rate": 0.0005964560793936119,
      "loss": 2.8571,
      "step": 11284
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.7162651419639587,
      "learning_rate": 0.000596455452473704,
      "loss": 2.6189,
      "step": 11285
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6905816793441772,
      "learning_rate": 0.0005964548254986795,
      "loss": 2.6565,
      "step": 11286
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.5825034379959106,
      "learning_rate": 0.0005964541984685383,
      "loss": 2.642,
      "step": 11287
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.5953493714332581,
      "learning_rate": 0.0005964535713832808,
      "loss": 2.3779,
      "step": 11288
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6309961676597595,
      "learning_rate": 0.000596452944242907,
      "loss": 2.3074,
      "step": 11289
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.7221006751060486,
      "learning_rate": 0.0005964523170474169,
      "loss": 2.547,
      "step": 11290
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6488330960273743,
      "learning_rate": 0.0005964516897968107,
      "loss": 2.495,
      "step": 11291
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.5584665536880493,
      "learning_rate": 0.0005964510624910887,
      "loss": 2.2749,
      "step": 11292
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.5916877388954163,
      "learning_rate": 0.0005964504351302507,
      "loss": 2.5041,
      "step": 11293
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.603744387626648,
      "learning_rate": 0.000596449807714297,
      "loss": 2.5874,
      "step": 11294
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6371937990188599,
      "learning_rate": 0.0005964491802432278,
      "loss": 2.6271,
      "step": 11295
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6407679915428162,
      "learning_rate": 0.000596448552717043,
      "loss": 2.5321,
      "step": 11296
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.5898533463478088,
      "learning_rate": 0.0005964479251357428,
      "loss": 2.3512,
      "step": 11297
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.5498112440109253,
      "learning_rate": 0.0005964472974993274,
      "loss": 2.3148,
      "step": 11298
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6587537527084351,
      "learning_rate": 0.0005964466698077968,
      "loss": 2.4179,
      "step": 11299
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.5933809876441956,
      "learning_rate": 0.0005964460420611513,
      "loss": 2.5876,
      "step": 11300
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.655215859413147,
      "learning_rate": 0.0005964454142593906,
      "loss": 2.4528,
      "step": 11301
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6146606802940369,
      "learning_rate": 0.0005964447864025154,
      "loss": 2.4575,
      "step": 11302
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.5939921736717224,
      "learning_rate": 0.0005964441584905253,
      "loss": 2.552,
      "step": 11303
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6675825119018555,
      "learning_rate": 0.0005964435305234208,
      "loss": 2.4084,
      "step": 11304
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.5770561695098877,
      "learning_rate": 0.0005964429025012017,
      "loss": 2.439,
      "step": 11305
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.645277738571167,
      "learning_rate": 0.0005964422744238684,
      "loss": 2.5727,
      "step": 11306
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.5841680765151978,
      "learning_rate": 0.0005964416462914208,
      "loss": 2.287,
      "step": 11307
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6181126236915588,
      "learning_rate": 0.0005964410181038592,
      "loss": 2.4984,
      "step": 11308
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6211051940917969,
      "learning_rate": 0.0005964403898611835,
      "loss": 2.3964,
      "step": 11309
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.5328419208526611,
      "learning_rate": 0.000596439761563394,
      "loss": 2.3995,
      "step": 11310
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6251624226570129,
      "learning_rate": 0.0005964391332104907,
      "loss": 2.3096,
      "step": 11311
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.628395676612854,
      "learning_rate": 0.0005964385048024739,
      "loss": 2.3449,
      "step": 11312
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6547375917434692,
      "learning_rate": 0.0005964378763393434,
      "loss": 2.4102,
      "step": 11313
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6796374917030334,
      "learning_rate": 0.0005964372478210997,
      "loss": 2.8085,
      "step": 11314
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6525807976722717,
      "learning_rate": 0.0005964366192477426,
      "loss": 2.2984,
      "step": 11315
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6251565217971802,
      "learning_rate": 0.0005964359906192724,
      "loss": 2.3438,
      "step": 11316
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6802334785461426,
      "learning_rate": 0.0005964353619356891,
      "loss": 2.5974,
      "step": 11317
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6130837798118591,
      "learning_rate": 0.0005964347331969929,
      "loss": 2.5091,
      "step": 11318
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6118658781051636,
      "learning_rate": 0.0005964341044031839,
      "loss": 2.2518,
      "step": 11319
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.5687533617019653,
      "learning_rate": 0.0005964334755542621,
      "loss": 2.1802,
      "step": 11320
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6144570112228394,
      "learning_rate": 0.0005964328466502278,
      "loss": 2.3432,
      "step": 11321
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6179496049880981,
      "learning_rate": 0.000596432217691081,
      "loss": 2.3781,
      "step": 11322
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.5748546719551086,
      "learning_rate": 0.0005964315886768219,
      "loss": 2.2836,
      "step": 11323
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.616188645362854,
      "learning_rate": 0.0005964309596074506,
      "loss": 2.4154,
      "step": 11324
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6376246213912964,
      "learning_rate": 0.0005964303304829672,
      "loss": 2.3582,
      "step": 11325
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6311289072036743,
      "learning_rate": 0.0005964297013033716,
      "loss": 2.3209,
      "step": 11326
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6400183439254761,
      "learning_rate": 0.0005964290720686644,
      "loss": 2.2547,
      "step": 11327
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6414493918418884,
      "learning_rate": 0.0005964284427788454,
      "loss": 2.4793,
      "step": 11328
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6212357878684998,
      "learning_rate": 0.0005964278134339146,
      "loss": 2.6271,
      "step": 11329
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6077311635017395,
      "learning_rate": 0.0005964271840338723,
      "loss": 2.4169,
      "step": 11330
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6229468584060669,
      "learning_rate": 0.0005964265545787186,
      "loss": 2.4691,
      "step": 11331
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6529892086982727,
      "learning_rate": 0.0005964259250684537,
      "loss": 2.4653,
      "step": 11332
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.5980494618415833,
      "learning_rate": 0.0005964252955030776,
      "loss": 2.4411,
      "step": 11333
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6180815696716309,
      "learning_rate": 0.0005964246658825904,
      "loss": 2.2953,
      "step": 11334
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6151129603385925,
      "learning_rate": 0.0005964240362069923,
      "loss": 2.4114,
      "step": 11335
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.711336612701416,
      "learning_rate": 0.0005964234064762833,
      "loss": 2.763,
      "step": 11336
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6374185085296631,
      "learning_rate": 0.0005964227766904637,
      "loss": 2.3613,
      "step": 11337
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6162078380584717,
      "learning_rate": 0.0005964221468495334,
      "loss": 2.3688,
      "step": 11338
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.600654661655426,
      "learning_rate": 0.0005964215169534927,
      "loss": 2.4637,
      "step": 11339
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6778227090835571,
      "learning_rate": 0.0005964208870023417,
      "loss": 2.6495,
      "step": 11340
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.610787570476532,
      "learning_rate": 0.0005964202569960804,
      "loss": 2.6149,
      "step": 11341
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6428425908088684,
      "learning_rate": 0.0005964196269347088,
      "loss": 2.5095,
      "step": 11342
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.7277795076370239,
      "learning_rate": 0.0005964189968182275,
      "loss": 2.6369,
      "step": 11343
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.582344114780426,
      "learning_rate": 0.0005964183666466362,
      "loss": 2.5329,
      "step": 11344
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.5799597501754761,
      "learning_rate": 0.000596417736419935,
      "loss": 2.6577,
      "step": 11345
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6682284474372864,
      "learning_rate": 0.0005964171061381242,
      "loss": 2.2173,
      "step": 11346
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.5911563038825989,
      "learning_rate": 0.000596416475801204,
      "loss": 2.3697,
      "step": 11347
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6610494256019592,
      "learning_rate": 0.0005964158454091743,
      "loss": 2.4925,
      "step": 11348
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.7000911831855774,
      "learning_rate": 0.0005964152149620352,
      "loss": 2.3912,
      "step": 11349
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6219296455383301,
      "learning_rate": 0.0005964145844597872,
      "loss": 2.5188,
      "step": 11350
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6298628449440002,
      "learning_rate": 0.0005964139539024299,
      "loss": 2.3872,
      "step": 11351
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.5825904607772827,
      "learning_rate": 0.0005964133232899637,
      "loss": 2.4595,
      "step": 11352
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.5746140480041504,
      "learning_rate": 0.0005964126926223886,
      "loss": 2.5108,
      "step": 11353
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6638596653938293,
      "learning_rate": 0.0005964120618997049,
      "loss": 2.7312,
      "step": 11354
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6529353857040405,
      "learning_rate": 0.0005964114311219126,
      "loss": 2.4572,
      "step": 11355
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6624923944473267,
      "learning_rate": 0.0005964108002890117,
      "loss": 2.6428,
      "step": 11356
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.593839168548584,
      "learning_rate": 0.0005964101694010026,
      "loss": 2.5141,
      "step": 11357
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.559850811958313,
      "learning_rate": 0.000596409538457885,
      "loss": 2.628,
      "step": 11358
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.5975345373153687,
      "learning_rate": 0.0005964089074596596,
      "loss": 2.353,
      "step": 11359
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6416581869125366,
      "learning_rate": 0.000596408276406326,
      "loss": 2.5752,
      "step": 11360
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6878833770751953,
      "learning_rate": 0.0005964076452978845,
      "loss": 2.2519,
      "step": 11361
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.5756858587265015,
      "learning_rate": 0.0005964070141343353,
      "loss": 2.3253,
      "step": 11362
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6258602142333984,
      "learning_rate": 0.0005964063829156784,
      "loss": 2.3416,
      "step": 11363
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6924970149993896,
      "learning_rate": 0.0005964057516419139,
      "loss": 2.2855,
      "step": 11364
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6204981207847595,
      "learning_rate": 0.000596405120313042,
      "loss": 2.3969,
      "step": 11365
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.8193890452384949,
      "learning_rate": 0.000596404488929063,
      "loss": 2.4213,
      "step": 11366
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6828495264053345,
      "learning_rate": 0.0005964038574899766,
      "loss": 2.6119,
      "step": 11367
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.5942646265029907,
      "learning_rate": 0.0005964032259957831,
      "loss": 2.463,
      "step": 11368
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6643864512443542,
      "learning_rate": 0.0005964025944464827,
      "loss": 2.5189,
      "step": 11369
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.704903781414032,
      "learning_rate": 0.0005964019628420755,
      "loss": 2.5071,
      "step": 11370
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.624309778213501,
      "learning_rate": 0.0005964013311825616,
      "loss": 2.2094,
      "step": 11371
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.7089808583259583,
      "learning_rate": 0.000596400699467941,
      "loss": 2.4752,
      "step": 11372
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.7422556281089783,
      "learning_rate": 0.000596400067698214,
      "loss": 2.5552,
      "step": 11373
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.7017947435379028,
      "learning_rate": 0.0005963994358733805,
      "loss": 2.4833,
      "step": 11374
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6865647435188293,
      "learning_rate": 0.000596398803993441,
      "loss": 2.5535,
      "step": 11375
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.69640052318573,
      "learning_rate": 0.0005963981720583951,
      "loss": 2.2653,
      "step": 11376
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.5926687121391296,
      "learning_rate": 0.0005963975400682433,
      "loss": 2.3433,
      "step": 11377
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6518266201019287,
      "learning_rate": 0.0005963969080229857,
      "loss": 2.3854,
      "step": 11378
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6038743257522583,
      "learning_rate": 0.0005963962759226221,
      "loss": 2.3985,
      "step": 11379
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6771509647369385,
      "learning_rate": 0.000596395643767153,
      "loss": 2.4258,
      "step": 11380
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6009404063224792,
      "learning_rate": 0.0005963950115565784,
      "loss": 2.384,
      "step": 11381
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6729490160942078,
      "learning_rate": 0.0005963943792908982,
      "loss": 2.6175,
      "step": 11382
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.7391873002052307,
      "learning_rate": 0.0005963937469701128,
      "loss": 2.5556,
      "step": 11383
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6663256287574768,
      "learning_rate": 0.0005963931145942221,
      "loss": 2.4649,
      "step": 11384
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.7416300773620605,
      "learning_rate": 0.0005963924821632264,
      "loss": 2.4186,
      "step": 11385
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6998288035392761,
      "learning_rate": 0.0005963918496771259,
      "loss": 2.5639,
      "step": 11386
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6819164156913757,
      "learning_rate": 0.0005963912171359203,
      "loss": 2.5243,
      "step": 11387
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6056577563285828,
      "learning_rate": 0.0005963905845396101,
      "loss": 2.582,
      "step": 11388
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.631645679473877,
      "learning_rate": 0.0005963899518881953,
      "loss": 2.3319,
      "step": 11389
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.615473747253418,
      "learning_rate": 0.000596389319181676,
      "loss": 2.4674,
      "step": 11390
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6071648001670837,
      "learning_rate": 0.0005963886864200524,
      "loss": 2.5405,
      "step": 11391
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6206197738647461,
      "learning_rate": 0.0005963880536033245,
      "loss": 2.375,
      "step": 11392
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.5988948345184326,
      "learning_rate": 0.0005963874207314924,
      "loss": 2.3923,
      "step": 11393
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.626454770565033,
      "learning_rate": 0.0005963867878045563,
      "loss": 2.408,
      "step": 11394
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6850239038467407,
      "learning_rate": 0.0005963861548225163,
      "loss": 2.4478,
      "step": 11395
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.5976617932319641,
      "learning_rate": 0.0005963855217853727,
      "loss": 2.4334,
      "step": 11396
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6838744282722473,
      "learning_rate": 0.0005963848886931252,
      "loss": 2.5239,
      "step": 11397
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6217890977859497,
      "learning_rate": 0.0005963842555457743,
      "loss": 2.4578,
      "step": 11398
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.670000433921814,
      "learning_rate": 0.0005963836223433198,
      "loss": 2.3746,
      "step": 11399
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6221661567687988,
      "learning_rate": 0.0005963829890857621,
      "loss": 2.3904,
      "step": 11400
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6703569293022156,
      "learning_rate": 0.0005963823557731012,
      "loss": 2.4806,
      "step": 11401
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.5676008462905884,
      "learning_rate": 0.0005963817224053373,
      "loss": 2.3771,
      "step": 11402
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6854238510131836,
      "learning_rate": 0.0005963810889824704,
      "loss": 2.318,
      "step": 11403
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6005125045776367,
      "learning_rate": 0.0005963804555045006,
      "loss": 2.4733,
      "step": 11404
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6358500123023987,
      "learning_rate": 0.0005963798219714282,
      "loss": 2.6072,
      "step": 11405
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6298362612724304,
      "learning_rate": 0.0005963791883832531,
      "loss": 2.4126,
      "step": 11406
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.618520200252533,
      "learning_rate": 0.0005963785547399755,
      "loss": 2.0329,
      "step": 11407
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6349058151245117,
      "learning_rate": 0.0005963779210415956,
      "loss": 2.2908,
      "step": 11408
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6337188482284546,
      "learning_rate": 0.0005963772872881134,
      "loss": 2.4381,
      "step": 11409
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.7015753984451294,
      "learning_rate": 0.000596376653479529,
      "loss": 2.4436,
      "step": 11410
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6551092267036438,
      "learning_rate": 0.0005963760196158427,
      "loss": 2.5741,
      "step": 11411
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.7487976551055908,
      "learning_rate": 0.0005963753856970545,
      "loss": 2.3994,
      "step": 11412
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.5971583127975464,
      "learning_rate": 0.0005963747517231644,
      "loss": 2.5243,
      "step": 11413
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6698170304298401,
      "learning_rate": 0.0005963741176941728,
      "loss": 2.746,
      "step": 11414
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6688721776008606,
      "learning_rate": 0.0005963734836100796,
      "loss": 2.5462,
      "step": 11415
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.585936427116394,
      "learning_rate": 0.0005963728494708849,
      "loss": 2.7622,
      "step": 11416
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6530554294586182,
      "learning_rate": 0.0005963722152765889,
      "loss": 2.2536,
      "step": 11417
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6315364241600037,
      "learning_rate": 0.0005963715810271918,
      "loss": 2.5056,
      "step": 11418
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6443581581115723,
      "learning_rate": 0.0005963709467226936,
      "loss": 2.5334,
      "step": 11419
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6957296133041382,
      "learning_rate": 0.0005963703123630945,
      "loss": 2.4677,
      "step": 11420
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6087192296981812,
      "learning_rate": 0.0005963696779483945,
      "loss": 2.2719,
      "step": 11421
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6490957736968994,
      "learning_rate": 0.0005963690434785937,
      "loss": 2.3518,
      "step": 11422
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6517124772071838,
      "learning_rate": 0.0005963684089536924,
      "loss": 2.7258,
      "step": 11423
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.7941171526908875,
      "learning_rate": 0.0005963677743736906,
      "loss": 2.41,
      "step": 11424
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6105358600616455,
      "learning_rate": 0.0005963671397385884,
      "loss": 2.668,
      "step": 11425
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.7349241971969604,
      "learning_rate": 0.000596366505048386,
      "loss": 2.487,
      "step": 11426
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.7426580786705017,
      "learning_rate": 0.0005963658703030836,
      "loss": 2.454,
      "step": 11427
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6425443291664124,
      "learning_rate": 0.000596365235502681,
      "loss": 2.624,
      "step": 11428
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6190304756164551,
      "learning_rate": 0.0005963646006471785,
      "loss": 2.5385,
      "step": 11429
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6284753680229187,
      "learning_rate": 0.0005963639657365764,
      "loss": 2.3973,
      "step": 11430
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.612332284450531,
      "learning_rate": 0.0005963633307708745,
      "loss": 2.4617,
      "step": 11431
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.5819888710975647,
      "learning_rate": 0.0005963626957500731,
      "loss": 2.4217,
      "step": 11432
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.5828876495361328,
      "learning_rate": 0.0005963620606741723,
      "loss": 2.3639,
      "step": 11433
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.680185854434967,
      "learning_rate": 0.0005963614255431721,
      "loss": 2.3838,
      "step": 11434
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6522648930549622,
      "learning_rate": 0.0005963607903570729,
      "loss": 2.3526,
      "step": 11435
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.586113691329956,
      "learning_rate": 0.0005963601551158745,
      "loss": 2.4226,
      "step": 11436
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.638150155544281,
      "learning_rate": 0.0005963595198195772,
      "loss": 2.5598,
      "step": 11437
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.7749245762825012,
      "learning_rate": 0.000596358884468181,
      "loss": 2.4545,
      "step": 11438
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6254743337631226,
      "learning_rate": 0.0005963582490616862,
      "loss": 2.5768,
      "step": 11439
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.670477569103241,
      "learning_rate": 0.0005963576136000928,
      "loss": 2.3934,
      "step": 11440
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6728320717811584,
      "learning_rate": 0.0005963569780834008,
      "loss": 2.3347,
      "step": 11441
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6058281660079956,
      "learning_rate": 0.0005963563425116106,
      "loss": 2.321,
      "step": 11442
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.5583984851837158,
      "learning_rate": 0.000596355706884722,
      "loss": 2.2778,
      "step": 11443
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.7316851615905762,
      "learning_rate": 0.0005963550712027353,
      "loss": 2.5144,
      "step": 11444
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.7193658947944641,
      "learning_rate": 0.0005963544354656507,
      "loss": 2.3694,
      "step": 11445
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.575853168964386,
      "learning_rate": 0.0005963537996734682,
      "loss": 2.3885,
      "step": 11446
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.7356239557266235,
      "learning_rate": 0.000596353163826188,
      "loss": 2.3759,
      "step": 11447
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.9029324650764465,
      "learning_rate": 0.00059635252792381,
      "loss": 2.3906,
      "step": 11448
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6915371417999268,
      "learning_rate": 0.0005963518919663345,
      "loss": 2.4446,
      "step": 11449
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.8609820604324341,
      "learning_rate": 0.0005963512559537617,
      "loss": 2.3999,
      "step": 11450
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6810221076011658,
      "learning_rate": 0.0005963506198860915,
      "loss": 2.1387,
      "step": 11451
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6410793662071228,
      "learning_rate": 0.0005963499837633242,
      "loss": 2.2584,
      "step": 11452
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6580737829208374,
      "learning_rate": 0.0005963493475854598,
      "loss": 2.4402,
      "step": 11453
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6375979781150818,
      "learning_rate": 0.0005963487113524984,
      "loss": 2.2985,
      "step": 11454
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6434521079063416,
      "learning_rate": 0.0005963480750644402,
      "loss": 2.5759,
      "step": 11455
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6990933418273926,
      "learning_rate": 0.0005963474387212853,
      "loss": 2.4338,
      "step": 11456
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.8757732510566711,
      "learning_rate": 0.000596346802323034,
      "loss": 2.6789,
      "step": 11457
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6486431956291199,
      "learning_rate": 0.000596346165869686,
      "loss": 2.3788,
      "step": 11458
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6542367339134216,
      "learning_rate": 0.0005963455293612418,
      "loss": 2.6822,
      "step": 11459
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6328597664833069,
      "learning_rate": 0.0005963448927977014,
      "loss": 2.4782,
      "step": 11460
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.7106459736824036,
      "learning_rate": 0.0005963442561790648,
      "loss": 2.2994,
      "step": 11461
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.7196095585823059,
      "learning_rate": 0.0005963436195053322,
      "loss": 2.599,
      "step": 11462
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.634955644607544,
      "learning_rate": 0.0005963429827765037,
      "loss": 2.4177,
      "step": 11463
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.8528315424919128,
      "learning_rate": 0.0005963423459925795,
      "loss": 2.5573,
      "step": 11464
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6472347974777222,
      "learning_rate": 0.0005963417091535597,
      "loss": 2.4179,
      "step": 11465
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6656234264373779,
      "learning_rate": 0.0005963410722594442,
      "loss": 2.3725,
      "step": 11466
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.7129955291748047,
      "learning_rate": 0.0005963404353102335,
      "loss": 2.5107,
      "step": 11467
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6831302046775818,
      "learning_rate": 0.0005963397983059274,
      "loss": 2.4396,
      "step": 11468
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6666471362113953,
      "learning_rate": 0.0005963391612465263,
      "loss": 2.4792,
      "step": 11469
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.592124879360199,
      "learning_rate": 0.00059633852413203,
      "loss": 2.3688,
      "step": 11470
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6439687013626099,
      "learning_rate": 0.0005963378869624389,
      "loss": 2.4511,
      "step": 11471
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.644188404083252,
      "learning_rate": 0.0005963372497377529,
      "loss": 2.6409,
      "step": 11472
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6285408139228821,
      "learning_rate": 0.0005963366124579722,
      "loss": 2.6096,
      "step": 11473
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.571072518825531,
      "learning_rate": 0.000596335975123097,
      "loss": 2.3715,
      "step": 11474
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6260322332382202,
      "learning_rate": 0.0005963353377331273,
      "loss": 2.5096,
      "step": 11475
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6517922878265381,
      "learning_rate": 0.0005963347002880632,
      "loss": 2.4952,
      "step": 11476
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6211538314819336,
      "learning_rate": 0.0005963340627879051,
      "loss": 2.5525,
      "step": 11477
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6075696349143982,
      "learning_rate": 0.0005963334252326527,
      "loss": 2.4777,
      "step": 11478
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6692193746566772,
      "learning_rate": 0.0005963327876223063,
      "loss": 2.7414,
      "step": 11479
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6704444289207458,
      "learning_rate": 0.0005963321499568662,
      "loss": 2.596,
      "step": 11480
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6239479184150696,
      "learning_rate": 0.0005963315122363323,
      "loss": 2.4487,
      "step": 11481
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.606115996837616,
      "learning_rate": 0.0005963308744607048,
      "loss": 2.5188,
      "step": 11482
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.643994152545929,
      "learning_rate": 0.0005963302366299838,
      "loss": 2.7312,
      "step": 11483
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.7207504510879517,
      "learning_rate": 0.0005963295987441693,
      "loss": 2.5796,
      "step": 11484
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6576446294784546,
      "learning_rate": 0.0005963289608032616,
      "loss": 2.5164,
      "step": 11485
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6167716979980469,
      "learning_rate": 0.0005963283228072608,
      "loss": 2.6012,
      "step": 11486
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6874476075172424,
      "learning_rate": 0.0005963276847561669,
      "loss": 2.6819,
      "step": 11487
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.7213411331176758,
      "learning_rate": 0.0005963270466499801,
      "loss": 2.5935,
      "step": 11488
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.614688515663147,
      "learning_rate": 0.0005963264084887005,
      "loss": 2.1574,
      "step": 11489
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.8275210857391357,
      "learning_rate": 0.0005963257702723283,
      "loss": 2.4766,
      "step": 11490
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6423904299736023,
      "learning_rate": 0.0005963251320008636,
      "loss": 2.4416,
      "step": 11491
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6924368143081665,
      "learning_rate": 0.0005963244936743064,
      "loss": 2.4892,
      "step": 11492
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6154115200042725,
      "learning_rate": 0.0005963238552926569,
      "loss": 2.4039,
      "step": 11493
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6176578402519226,
      "learning_rate": 0.0005963232168559151,
      "loss": 2.4965,
      "step": 11494
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6220815777778625,
      "learning_rate": 0.0005963225783640813,
      "loss": 2.2225,
      "step": 11495
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6198908090591431,
      "learning_rate": 0.0005963219398171554,
      "loss": 2.4173,
      "step": 11496
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6451581120491028,
      "learning_rate": 0.0005963213012151378,
      "loss": 2.6277,
      "step": 11497
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6909437775611877,
      "learning_rate": 0.0005963206625580285,
      "loss": 2.5544,
      "step": 11498
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.5586854815483093,
      "learning_rate": 0.0005963200238458276,
      "loss": 2.3202,
      "step": 11499
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6551730036735535,
      "learning_rate": 0.0005963193850785351,
      "loss": 2.5368,
      "step": 11500
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.8055827021598816,
      "learning_rate": 0.0005963187462561514,
      "loss": 2.5336,
      "step": 11501
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.5981952548027039,
      "learning_rate": 0.0005963181073786762,
      "loss": 2.5189,
      "step": 11502
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6585246920585632,
      "learning_rate": 0.0005963174684461101,
      "loss": 2.6064,
      "step": 11503
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6186164617538452,
      "learning_rate": 0.0005963168294584529,
      "loss": 2.5112,
      "step": 11504
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6714003086090088,
      "learning_rate": 0.0005963161904157047,
      "loss": 2.5014,
      "step": 11505
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6046439409255981,
      "learning_rate": 0.000596315551317866,
      "loss": 2.4414,
      "step": 11506
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6055604219436646,
      "learning_rate": 0.0005963149121649364,
      "loss": 2.5038,
      "step": 11507
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6514601707458496,
      "learning_rate": 0.0005963142729569163,
      "loss": 2.5386,
      "step": 11508
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.7180113196372986,
      "learning_rate": 0.0005963136336938059,
      "loss": 2.7368,
      "step": 11509
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.7272889614105225,
      "learning_rate": 0.0005963129943756051,
      "loss": 2.2329,
      "step": 11510
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6103624105453491,
      "learning_rate": 0.0005963123550023142,
      "loss": 2.4025,
      "step": 11511
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.5978131890296936,
      "learning_rate": 0.0005963117155739332,
      "loss": 2.6002,
      "step": 11512
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.7318323850631714,
      "learning_rate": 0.0005963110760904622,
      "loss": 2.66,
      "step": 11513
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6442745327949524,
      "learning_rate": 0.0005963104365519015,
      "loss": 2.5853,
      "step": 11514
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6459358334541321,
      "learning_rate": 0.000596309796958251,
      "loss": 2.4112,
      "step": 11515
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6764480471611023,
      "learning_rate": 0.000596309157309511,
      "loss": 2.2977,
      "step": 11516
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6097608208656311,
      "learning_rate": 0.0005963085176056814,
      "loss": 2.5483,
      "step": 11517
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.5956771969795227,
      "learning_rate": 0.0005963078778467626,
      "loss": 2.3254,
      "step": 11518
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.652824342250824,
      "learning_rate": 0.0005963072380327544,
      "loss": 2.4938,
      "step": 11519
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.670874297618866,
      "learning_rate": 0.0005963065981636573,
      "loss": 2.3378,
      "step": 11520
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6950606107711792,
      "learning_rate": 0.000596305958239471,
      "loss": 2.4565,
      "step": 11521
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6586180329322815,
      "learning_rate": 0.0005963053182601961,
      "loss": 2.4334,
      "step": 11522
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.5861239433288574,
      "learning_rate": 0.0005963046782258322,
      "loss": 2.2051,
      "step": 11523
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6100037693977356,
      "learning_rate": 0.0005963040381363798,
      "loss": 2.4614,
      "step": 11524
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6320622563362122,
      "learning_rate": 0.0005963033979918388,
      "loss": 2.2575,
      "step": 11525
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.7056825160980225,
      "learning_rate": 0.0005963027577922095,
      "loss": 2.628,
      "step": 11526
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6681949496269226,
      "learning_rate": 0.0005963021175374918,
      "loss": 2.2907,
      "step": 11527
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.600038468837738,
      "learning_rate": 0.000596301477227686,
      "loss": 2.3838,
      "step": 11528
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6215845942497253,
      "learning_rate": 0.0005963008368627922,
      "loss": 2.4202,
      "step": 11529
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6260676980018616,
      "learning_rate": 0.0005963001964428104,
      "loss": 2.4368,
      "step": 11530
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6454449892044067,
      "learning_rate": 0.0005962995559677408,
      "loss": 2.5664,
      "step": 11531
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.5658130049705505,
      "learning_rate": 0.0005962989154375837,
      "loss": 2.4092,
      "step": 11532
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6143683195114136,
      "learning_rate": 0.0005962982748523389,
      "loss": 2.4802,
      "step": 11533
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6770437359809875,
      "learning_rate": 0.0005962976342120067,
      "loss": 2.4247,
      "step": 11534
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6498855352401733,
      "learning_rate": 0.0005962969935165871,
      "loss": 2.593,
      "step": 11535
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.597453773021698,
      "learning_rate": 0.0005962963527660803,
      "loss": 2.3813,
      "step": 11536
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.5573212504386902,
      "learning_rate": 0.0005962957119604865,
      "loss": 2.1803,
      "step": 11537
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.5933131575584412,
      "learning_rate": 0.0005962950710998057,
      "loss": 2.4157,
      "step": 11538
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6101275086402893,
      "learning_rate": 0.0005962944301840381,
      "loss": 2.5327,
      "step": 11539
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6789463758468628,
      "learning_rate": 0.0005962937892131836,
      "loss": 2.5833,
      "step": 11540
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6057182550430298,
      "learning_rate": 0.0005962931481872427,
      "loss": 2.6107,
      "step": 11541
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6413453221321106,
      "learning_rate": 0.0005962925071062152,
      "loss": 2.5792,
      "step": 11542
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.7005075216293335,
      "learning_rate": 0.0005962918659701013,
      "loss": 2.5811,
      "step": 11543
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6128180623054504,
      "learning_rate": 0.0005962912247789012,
      "loss": 2.4074,
      "step": 11544
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.5970283150672913,
      "learning_rate": 0.000596290583532615,
      "loss": 2.3136,
      "step": 11545
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.5904926657676697,
      "learning_rate": 0.0005962899422312427,
      "loss": 2.4041,
      "step": 11546
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.580416738986969,
      "learning_rate": 0.0005962893008747846,
      "loss": 2.3105,
      "step": 11547
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6736274361610413,
      "learning_rate": 0.0005962886594632407,
      "loss": 2.4938,
      "step": 11548
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6686013340950012,
      "learning_rate": 0.0005962880179966112,
      "loss": 2.4409,
      "step": 11549
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.5834082961082458,
      "learning_rate": 0.0005962873764748961,
      "loss": 2.2535,
      "step": 11550
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.8013529181480408,
      "learning_rate": 0.0005962867348980955,
      "loss": 2.7425,
      "step": 11551
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6319385170936584,
      "learning_rate": 0.0005962860932662097,
      "loss": 2.3749,
      "step": 11552
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6116765737533569,
      "learning_rate": 0.0005962854515792387,
      "loss": 2.4511,
      "step": 11553
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.7022843360900879,
      "learning_rate": 0.0005962848098371827,
      "loss": 2.38,
      "step": 11554
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.5965883731842041,
      "learning_rate": 0.0005962841680400417,
      "loss": 2.5194,
      "step": 11555
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6185353398323059,
      "learning_rate": 0.0005962835261878158,
      "loss": 2.3458,
      "step": 11556
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.591101348400116,
      "learning_rate": 0.0005962828842805053,
      "loss": 2.272,
      "step": 11557
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6328172087669373,
      "learning_rate": 0.0005962822423181102,
      "loss": 2.7028,
      "step": 11558
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.5989089608192444,
      "learning_rate": 0.0005962816003006306,
      "loss": 2.2256,
      "step": 11559
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6902540922164917,
      "learning_rate": 0.0005962809582280668,
      "loss": 2.7546,
      "step": 11560
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6589159965515137,
      "learning_rate": 0.0005962803161004188,
      "loss": 2.4346,
      "step": 11561
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.592179536819458,
      "learning_rate": 0.0005962796739176864,
      "loss": 2.4236,
      "step": 11562
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.624729573726654,
      "learning_rate": 0.0005962790316798703,
      "loss": 2.4574,
      "step": 11563
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.7936517000198364,
      "learning_rate": 0.0005962783893869703,
      "loss": 2.6107,
      "step": 11564
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6013132333755493,
      "learning_rate": 0.0005962777470389863,
      "loss": 2.5386,
      "step": 11565
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.7839146852493286,
      "learning_rate": 0.0005962771046359189,
      "loss": 2.5577,
      "step": 11566
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.614525318145752,
      "learning_rate": 0.000596276462177768,
      "loss": 2.727,
      "step": 11567
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6214706301689148,
      "learning_rate": 0.0005962758196645337,
      "loss": 2.4515,
      "step": 11568
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.7161359786987305,
      "learning_rate": 0.000596275177096216,
      "loss": 2.3815,
      "step": 11569
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6077137589454651,
      "learning_rate": 0.0005962745344728153,
      "loss": 2.4857,
      "step": 11570
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6147674322128296,
      "learning_rate": 0.0005962738917943314,
      "loss": 2.1752,
      "step": 11571
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6261560916900635,
      "learning_rate": 0.0005962732490607649,
      "loss": 2.3495,
      "step": 11572
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6734616756439209,
      "learning_rate": 0.0005962726062721153,
      "loss": 2.3971,
      "step": 11573
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.570334792137146,
      "learning_rate": 0.0005962719634283832,
      "loss": 2.404,
      "step": 11574
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6473824977874756,
      "learning_rate": 0.0005962713205295685,
      "loss": 2.637,
      "step": 11575
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.5952850580215454,
      "learning_rate": 0.0005962706775756714,
      "loss": 2.3658,
      "step": 11576
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.7136637568473816,
      "learning_rate": 0.0005962700345666919,
      "loss": 2.4929,
      "step": 11577
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6540914177894592,
      "learning_rate": 0.0005962693915026303,
      "loss": 2.538,
      "step": 11578
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6983373165130615,
      "learning_rate": 0.0005962687483834866,
      "loss": 2.4056,
      "step": 11579
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6745260953903198,
      "learning_rate": 0.000596268105209261,
      "loss": 2.5491,
      "step": 11580
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.7027052640914917,
      "learning_rate": 0.0005962674619799535,
      "loss": 2.5125,
      "step": 11581
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6842101216316223,
      "learning_rate": 0.0005962668186955643,
      "loss": 2.4323,
      "step": 11582
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6004122495651245,
      "learning_rate": 0.0005962661753560935,
      "loss": 2.5705,
      "step": 11583
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6942507028579712,
      "learning_rate": 0.0005962655319615412,
      "loss": 2.6232,
      "step": 11584
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6060343980789185,
      "learning_rate": 0.0005962648885119076,
      "loss": 2.4734,
      "step": 11585
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6149864792823792,
      "learning_rate": 0.0005962642450071927,
      "loss": 2.6586,
      "step": 11586
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.7953563332557678,
      "learning_rate": 0.0005962636014473967,
      "loss": 2.4892,
      "step": 11587
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.5897312164306641,
      "learning_rate": 0.0005962629578325198,
      "loss": 2.4646,
      "step": 11588
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6044555902481079,
      "learning_rate": 0.0005962623141625619,
      "loss": 2.2283,
      "step": 11589
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6443065404891968,
      "learning_rate": 0.0005962616704375234,
      "loss": 2.4164,
      "step": 11590
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6157188415527344,
      "learning_rate": 0.0005962610266574041,
      "loss": 2.3214,
      "step": 11591
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6219939589500427,
      "learning_rate": 0.0005962603828222043,
      "loss": 2.4193,
      "step": 11592
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6029102206230164,
      "learning_rate": 0.0005962597389319241,
      "loss": 2.3294,
      "step": 11593
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6344435811042786,
      "learning_rate": 0.0005962590949865637,
      "loss": 2.4209,
      "step": 11594
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.676399827003479,
      "learning_rate": 0.000596258450986123,
      "loss": 2.6378,
      "step": 11595
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.589907169342041,
      "learning_rate": 0.0005962578069306025,
      "loss": 2.508,
      "step": 11596
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.5743394494056702,
      "learning_rate": 0.0005962571628200019,
      "loss": 2.4656,
      "step": 11597
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.5827499628067017,
      "learning_rate": 0.0005962565186543215,
      "loss": 2.4187,
      "step": 11598
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.5897182822227478,
      "learning_rate": 0.0005962558744335615,
      "loss": 2.4645,
      "step": 11599
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.5416726469993591,
      "learning_rate": 0.0005962552301577219,
      "loss": 2.2657,
      "step": 11600
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6822784543037415,
      "learning_rate": 0.0005962545858268029,
      "loss": 2.3439,
      "step": 11601
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.562549889087677,
      "learning_rate": 0.0005962539414408046,
      "loss": 2.3134,
      "step": 11602
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6658889055252075,
      "learning_rate": 0.000596253296999727,
      "loss": 2.3776,
      "step": 11603
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.5996624231338501,
      "learning_rate": 0.0005962526525035704,
      "loss": 2.375,
      "step": 11604
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6174895763397217,
      "learning_rate": 0.0005962520079523348,
      "loss": 2.5693,
      "step": 11605
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.7689122557640076,
      "learning_rate": 0.0005962513633460204,
      "loss": 2.4047,
      "step": 11606
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.7236721515655518,
      "learning_rate": 0.0005962507186846272,
      "loss": 2.5377,
      "step": 11607
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.663418710231781,
      "learning_rate": 0.0005962500739681555,
      "loss": 2.4544,
      "step": 11608
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6416810750961304,
      "learning_rate": 0.0005962494291966052,
      "loss": 2.4065,
      "step": 11609
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6542936563491821,
      "learning_rate": 0.0005962487843699767,
      "loss": 2.3267,
      "step": 11610
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6181926727294922,
      "learning_rate": 0.0005962481394882699,
      "loss": 2.5961,
      "step": 11611
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6439699530601501,
      "learning_rate": 0.0005962474945514849,
      "loss": 2.2424,
      "step": 11612
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6267533302307129,
      "learning_rate": 0.0005962468495596219,
      "loss": 2.5171,
      "step": 11613
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.5623725056648254,
      "learning_rate": 0.000596246204512681,
      "loss": 2.4322,
      "step": 11614
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.7830151915550232,
      "learning_rate": 0.0005962455594106625,
      "loss": 2.3218,
      "step": 11615
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.693305492401123,
      "learning_rate": 0.0005962449142535662,
      "loss": 2.6795,
      "step": 11616
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.5912370681762695,
      "learning_rate": 0.0005962442690413925,
      "loss": 2.4496,
      "step": 11617
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6723439693450928,
      "learning_rate": 0.0005962436237741413,
      "loss": 2.5404,
      "step": 11618
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.5788262486457825,
      "learning_rate": 0.0005962429784518129,
      "loss": 2.4006,
      "step": 11619
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6779223680496216,
      "learning_rate": 0.0005962423330744072,
      "loss": 2.3523,
      "step": 11620
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6517859101295471,
      "learning_rate": 0.0005962416876419246,
      "loss": 2.5289,
      "step": 11621
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.640683650970459,
      "learning_rate": 0.000596241042154365,
      "loss": 2.3534,
      "step": 11622
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6197714805603027,
      "learning_rate": 0.0005962403966117286,
      "loss": 2.4845,
      "step": 11623
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6027525067329407,
      "learning_rate": 0.0005962397510140156,
      "loss": 2.6705,
      "step": 11624
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6079016327857971,
      "learning_rate": 0.000596239105361226,
      "loss": 2.4248,
      "step": 11625
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6111804842948914,
      "learning_rate": 0.0005962384596533599,
      "loss": 2.3426,
      "step": 11626
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6150492429733276,
      "learning_rate": 0.0005962378138904175,
      "loss": 2.3764,
      "step": 11627
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.5917300581932068,
      "learning_rate": 0.000596237168072399,
      "loss": 2.3384,
      "step": 11628
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6674749851226807,
      "learning_rate": 0.0005962365221993043,
      "loss": 2.6396,
      "step": 11629
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6404228210449219,
      "learning_rate": 0.0005962358762711338,
      "loss": 2.6136,
      "step": 11630
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.5735630393028259,
      "learning_rate": 0.0005962352302878873,
      "loss": 2.3954,
      "step": 11631
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.5631161332130432,
      "learning_rate": 0.000596234584249565,
      "loss": 2.5342,
      "step": 11632
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6780403852462769,
      "learning_rate": 0.0005962339381561672,
      "loss": 2.5637,
      "step": 11633
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.7379934787750244,
      "learning_rate": 0.0005962332920076939,
      "loss": 2.3536,
      "step": 11634
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.5992153286933899,
      "learning_rate": 0.0005962326458041454,
      "loss": 2.4309,
      "step": 11635
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6732681393623352,
      "learning_rate": 0.0005962319995455215,
      "loss": 2.2835,
      "step": 11636
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6370517611503601,
      "learning_rate": 0.0005962313532318225,
      "loss": 2.4958,
      "step": 11637
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6377142071723938,
      "learning_rate": 0.0005962307068630486,
      "loss": 2.3593,
      "step": 11638
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6216119527816772,
      "learning_rate": 0.0005962300604391997,
      "loss": 2.3958,
      "step": 11639
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6116071343421936,
      "learning_rate": 0.000596229413960276,
      "loss": 2.1995,
      "step": 11640
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.8263964056968689,
      "learning_rate": 0.0005962287674262778,
      "loss": 2.5903,
      "step": 11641
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.5836607813835144,
      "learning_rate": 0.0005962281208372049,
      "loss": 2.2069,
      "step": 11642
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6906614899635315,
      "learning_rate": 0.0005962274741930577,
      "loss": 2.4535,
      "step": 11643
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6801156997680664,
      "learning_rate": 0.0005962268274938363,
      "loss": 2.5638,
      "step": 11644
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6168856024742126,
      "learning_rate": 0.0005962261807395407,
      "loss": 2.2328,
      "step": 11645
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.5999285578727722,
      "learning_rate": 0.000596225533930171,
      "loss": 2.4959,
      "step": 11646
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6826470494270325,
      "learning_rate": 0.0005962248870657274,
      "loss": 2.5153,
      "step": 11647
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6028268337249756,
      "learning_rate": 0.00059622424014621,
      "loss": 2.5323,
      "step": 11648
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6212331056594849,
      "learning_rate": 0.000596223593171619,
      "loss": 2.4742,
      "step": 11649
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6189820766448975,
      "learning_rate": 0.0005962229461419544,
      "loss": 2.33,
      "step": 11650
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6455333232879639,
      "learning_rate": 0.0005962222990572163,
      "loss": 2.4693,
      "step": 11651
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.575815737247467,
      "learning_rate": 0.000596221651917405,
      "loss": 2.5207,
      "step": 11652
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6149454712867737,
      "learning_rate": 0.0005962210047225203,
      "loss": 2.2193,
      "step": 11653
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6776230931282043,
      "learning_rate": 0.0005962203574725627,
      "loss": 2.2665,
      "step": 11654
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6084030866622925,
      "learning_rate": 0.000596219710167532,
      "loss": 2.5485,
      "step": 11655
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6559980511665344,
      "learning_rate": 0.0005962190628074287,
      "loss": 2.5966,
      "step": 11656
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6121323108673096,
      "learning_rate": 0.0005962184153922525,
      "loss": 2.507,
      "step": 11657
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6054604053497314,
      "learning_rate": 0.0005962177679220038,
      "loss": 2.2349,
      "step": 11658
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.589972734451294,
      "learning_rate": 0.0005962171203966825,
      "loss": 2.5973,
      "step": 11659
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6196160912513733,
      "learning_rate": 0.0005962164728162889,
      "loss": 2.522,
      "step": 11660
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6466572284698486,
      "learning_rate": 0.0005962158251808232,
      "loss": 2.5546,
      "step": 11661
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.7170338034629822,
      "learning_rate": 0.0005962151774902851,
      "loss": 2.4501,
      "step": 11662
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6473994851112366,
      "learning_rate": 0.0005962145297446753,
      "loss": 2.366,
      "step": 11663
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6516795754432678,
      "learning_rate": 0.0005962138819439935,
      "loss": 2.6565,
      "step": 11664
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.7099058032035828,
      "learning_rate": 0.00059621323408824,
      "loss": 2.4442,
      "step": 11665
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6111395359039307,
      "learning_rate": 0.0005962125861774148,
      "loss": 2.4184,
      "step": 11666
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.643537163734436,
      "learning_rate": 0.0005962119382115182,
      "loss": 2.7055,
      "step": 11667
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6489663124084473,
      "learning_rate": 0.00059621129019055,
      "loss": 2.5236,
      "step": 11668
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.7545811533927917,
      "learning_rate": 0.0005962106421145107,
      "loss": 2.4663,
      "step": 11669
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6375373601913452,
      "learning_rate": 0.0005962099939834003,
      "loss": 2.2606,
      "step": 11670
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.7810790538787842,
      "learning_rate": 0.0005962093457972187,
      "loss": 2.4008,
      "step": 11671
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6240819096565247,
      "learning_rate": 0.0005962086975559662,
      "loss": 2.4497,
      "step": 11672
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6920105218887329,
      "learning_rate": 0.000596208049259643,
      "loss": 2.4818,
      "step": 11673
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.7061219811439514,
      "learning_rate": 0.0005962074009082492,
      "loss": 2.5147,
      "step": 11674
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.8149372339248657,
      "learning_rate": 0.0005962067525017848,
      "loss": 2.426,
      "step": 11675
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.70351642370224,
      "learning_rate": 0.0005962061040402499,
      "loss": 2.3894,
      "step": 11676
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.572637140750885,
      "learning_rate": 0.0005962054555236447,
      "loss": 2.4467,
      "step": 11677
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.771974503993988,
      "learning_rate": 0.0005962048069519694,
      "loss": 2.4936,
      "step": 11678
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.5911643505096436,
      "learning_rate": 0.0005962041583252238,
      "loss": 2.4579,
      "step": 11679
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.8696768283843994,
      "learning_rate": 0.0005962035096434085,
      "loss": 2.6319,
      "step": 11680
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.7316741943359375,
      "learning_rate": 0.0005962028609065234,
      "loss": 2.3973,
      "step": 11681
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.5781893134117126,
      "learning_rate": 0.0005962022121145684,
      "loss": 2.1568,
      "step": 11682
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6223843693733215,
      "learning_rate": 0.000596201563267544,
      "loss": 2.7174,
      "step": 11683
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6022017598152161,
      "learning_rate": 0.0005962009143654501,
      "loss": 2.2111,
      "step": 11684
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.5875415205955505,
      "learning_rate": 0.0005962002654082868,
      "loss": 2.3813,
      "step": 11685
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6235182285308838,
      "learning_rate": 0.0005961996163960543,
      "loss": 2.2994,
      "step": 11686
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6148957014083862,
      "learning_rate": 0.0005961989673287526,
      "loss": 2.3648,
      "step": 11687
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6976120471954346,
      "learning_rate": 0.0005961983182063822,
      "loss": 2.5274,
      "step": 11688
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6321697235107422,
      "learning_rate": 0.0005961976690289426,
      "loss": 2.6283,
      "step": 11689
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.561809241771698,
      "learning_rate": 0.0005961970197964345,
      "loss": 2.4758,
      "step": 11690
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6193451881408691,
      "learning_rate": 0.0005961963705088577,
      "loss": 2.4036,
      "step": 11691
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6526128053665161,
      "learning_rate": 0.0005961957211662125,
      "loss": 2.3368,
      "step": 11692
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.5753313899040222,
      "learning_rate": 0.0005961950717684988,
      "loss": 2.4398,
      "step": 11693
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6501908302307129,
      "learning_rate": 0.0005961944223157169,
      "loss": 2.6613,
      "step": 11694
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.5889447331428528,
      "learning_rate": 0.0005961937728078668,
      "loss": 2.3718,
      "step": 11695
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6321893334388733,
      "learning_rate": 0.0005961931232449487,
      "loss": 2.4163,
      "step": 11696
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6167053580284119,
      "learning_rate": 0.0005961924736269628,
      "loss": 2.2508,
      "step": 11697
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.576299250125885,
      "learning_rate": 0.000596191823953909,
      "loss": 2.3049,
      "step": 11698
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6695291996002197,
      "learning_rate": 0.0005961911742257877,
      "loss": 2.3953,
      "step": 11699
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.5966570377349854,
      "learning_rate": 0.0005961905244425987,
      "loss": 2.739,
      "step": 11700
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6276807188987732,
      "learning_rate": 0.0005961898746043424,
      "loss": 2.3831,
      "step": 11701
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.8519243001937866,
      "learning_rate": 0.0005961892247110188,
      "loss": 2.5573,
      "step": 11702
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6086157560348511,
      "learning_rate": 0.000596188574762628,
      "loss": 2.5599,
      "step": 11703
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.7357355952262878,
      "learning_rate": 0.0005961879247591701,
      "loss": 2.4992,
      "step": 11704
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.1493792533874512,
      "learning_rate": 0.0005961872747006455,
      "loss": 2.2918,
      "step": 11705
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6362924575805664,
      "learning_rate": 0.0005961866245870539,
      "loss": 2.4499,
      "step": 11706
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.9133944511413574,
      "learning_rate": 0.0005961859744183957,
      "loss": 2.4739,
      "step": 11707
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.8323266506195068,
      "learning_rate": 0.0005961853241946708,
      "loss": 2.4841,
      "step": 11708
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6282296180725098,
      "learning_rate": 0.0005961846739158796,
      "loss": 2.4422,
      "step": 11709
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.8393511176109314,
      "learning_rate": 0.0005961840235820221,
      "loss": 2.3918,
      "step": 11710
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6614386439323425,
      "learning_rate": 0.0005961833731930983,
      "loss": 2.6647,
      "step": 11711
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.634917140007019,
      "learning_rate": 0.0005961827227491084,
      "loss": 2.3466,
      "step": 11712
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6187928318977356,
      "learning_rate": 0.0005961820722500526,
      "loss": 2.5011,
      "step": 11713
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6090554594993591,
      "learning_rate": 0.000596181421695931,
      "loss": 2.3513,
      "step": 11714
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.5982932448387146,
      "learning_rate": 0.0005961807710867436,
      "loss": 2.2366,
      "step": 11715
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6081823706626892,
      "learning_rate": 0.0005961801204224906,
      "loss": 2.6336,
      "step": 11716
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6525804996490479,
      "learning_rate": 0.0005961794697031723,
      "loss": 2.4535,
      "step": 11717
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.7810242176055908,
      "learning_rate": 0.0005961788189287884,
      "loss": 2.6627,
      "step": 11718
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.635140597820282,
      "learning_rate": 0.0005961781680993395,
      "loss": 2.3592,
      "step": 11719
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6639537215232849,
      "learning_rate": 0.0005961775172148254,
      "loss": 2.4301,
      "step": 11720
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.7820079326629639,
      "learning_rate": 0.0005961768662752461,
      "loss": 2.3973,
      "step": 11721
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6916341781616211,
      "learning_rate": 0.0005961762152806022,
      "loss": 2.271,
      "step": 11722
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.706810474395752,
      "learning_rate": 0.0005961755642308935,
      "loss": 2.2297,
      "step": 11723
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.8588224053382874,
      "learning_rate": 0.0005961749131261201,
      "loss": 2.2441,
      "step": 11724
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6752805709838867,
      "learning_rate": 0.0005961742619662822,
      "loss": 2.3037,
      "step": 11725
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6514682769775391,
      "learning_rate": 0.0005961736107513799,
      "loss": 2.3673,
      "step": 11726
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.77907794713974,
      "learning_rate": 0.0005961729594814134,
      "loss": 2.596,
      "step": 11727
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.7348697781562805,
      "learning_rate": 0.0005961723081563827,
      "loss": 2.2563,
      "step": 11728
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6253810524940491,
      "learning_rate": 0.000596171656776288,
      "loss": 2.6168,
      "step": 11729
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6150892972946167,
      "learning_rate": 0.0005961710053411295,
      "loss": 2.7779,
      "step": 11730
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.786751389503479,
      "learning_rate": 0.000596170353850907,
      "loss": 2.5207,
      "step": 11731
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6695559620857239,
      "learning_rate": 0.000596169702305621,
      "loss": 2.4781,
      "step": 11732
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6220799684524536,
      "learning_rate": 0.0005961690507052715,
      "loss": 2.3207,
      "step": 11733
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6130435466766357,
      "learning_rate": 0.0005961683990498585,
      "loss": 2.5001,
      "step": 11734
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.8501202464103699,
      "learning_rate": 0.0005961677473393821,
      "loss": 2.1939,
      "step": 11735
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.7173826098442078,
      "learning_rate": 0.0005961670955738428,
      "loss": 2.7803,
      "step": 11736
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6852366328239441,
      "learning_rate": 0.0005961664437532402,
      "loss": 2.1095,
      "step": 11737
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.8301659226417542,
      "learning_rate": 0.0005961657918775748,
      "loss": 2.5397,
      "step": 11738
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6598018407821655,
      "learning_rate": 0.0005961651399468465,
      "loss": 2.4784,
      "step": 11739
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6400232315063477,
      "learning_rate": 0.0005961644879610555,
      "loss": 2.4957,
      "step": 11740
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6673219203948975,
      "learning_rate": 0.000596163835920202,
      "loss": 2.2672,
      "step": 11741
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.7073506712913513,
      "learning_rate": 0.000596163183824286,
      "loss": 2.499,
      "step": 11742
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6026692390441895,
      "learning_rate": 0.0005961625316733077,
      "loss": 2.2844,
      "step": 11743
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6791858673095703,
      "learning_rate": 0.0005961618794672672,
      "loss": 2.4561,
      "step": 11744
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6995101571083069,
      "learning_rate": 0.0005961612272061647,
      "loss": 2.4541,
      "step": 11745
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6300156116485596,
      "learning_rate": 0.0005961605748900001,
      "loss": 2.191,
      "step": 11746
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.5451065897941589,
      "learning_rate": 0.0005961599225187737,
      "loss": 2.465,
      "step": 11747
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.711790144443512,
      "learning_rate": 0.0005961592700924856,
      "loss": 2.4383,
      "step": 11748
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6802167892456055,
      "learning_rate": 0.0005961586176111359,
      "loss": 2.3097,
      "step": 11749
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.5731715559959412,
      "learning_rate": 0.0005961579650747247,
      "loss": 2.3781,
      "step": 11750
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6078522801399231,
      "learning_rate": 0.0005961573124832522,
      "loss": 2.4313,
      "step": 11751
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.5978575944900513,
      "learning_rate": 0.0005961566598367184,
      "loss": 2.2526,
      "step": 11752
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6370769739151001,
      "learning_rate": 0.0005961560071351235,
      "loss": 2.4405,
      "step": 11753
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6695687174797058,
      "learning_rate": 0.0005961553543784676,
      "loss": 2.4522,
      "step": 11754
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6120510101318359,
      "learning_rate": 0.0005961547015667509,
      "loss": 2.5926,
      "step": 11755
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6849433183670044,
      "learning_rate": 0.0005961540486999733,
      "loss": 2.2579,
      "step": 11756
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6108267903327942,
      "learning_rate": 0.0005961533957781353,
      "loss": 2.4777,
      "step": 11757
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.7482524514198303,
      "learning_rate": 0.0005961527428012366,
      "loss": 2.4528,
      "step": 11758
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6602693200111389,
      "learning_rate": 0.0005961520897692775,
      "loss": 2.538,
      "step": 11759
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6287938952445984,
      "learning_rate": 0.0005961514366822582,
      "loss": 2.3689,
      "step": 11760
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.58537757396698,
      "learning_rate": 0.0005961507835401789,
      "loss": 2.5094,
      "step": 11761
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6115125417709351,
      "learning_rate": 0.0005961501303430394,
      "loss": 2.5871,
      "step": 11762
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6045248508453369,
      "learning_rate": 0.0005961494770908401,
      "loss": 2.5417,
      "step": 11763
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.5540022850036621,
      "learning_rate": 0.000596148823783581,
      "loss": 2.331,
      "step": 11764
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6482625007629395,
      "learning_rate": 0.0005961481704212622,
      "loss": 2.5601,
      "step": 11765
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6124506592750549,
      "learning_rate": 0.0005961475170038838,
      "loss": 2.5073,
      "step": 11766
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.5952260494232178,
      "learning_rate": 0.0005961468635314461,
      "loss": 2.4145,
      "step": 11767
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6059144735336304,
      "learning_rate": 0.000596146210003949,
      "loss": 2.2157,
      "step": 11768
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6250486373901367,
      "learning_rate": 0.0005961455564213929,
      "loss": 2.3953,
      "step": 11769
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.5725736618041992,
      "learning_rate": 0.0005961449027837776,
      "loss": 2.3453,
      "step": 11770
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6719762682914734,
      "learning_rate": 0.0005961442490911035,
      "loss": 2.5738,
      "step": 11771
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.5963003039360046,
      "learning_rate": 0.0005961435953433705,
      "loss": 2.2462,
      "step": 11772
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6170646548271179,
      "learning_rate": 0.0005961429415405788,
      "loss": 2.4998,
      "step": 11773
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6619748473167419,
      "learning_rate": 0.0005961422876827286,
      "loss": 2.4859,
      "step": 11774
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6399399042129517,
      "learning_rate": 0.00059614163376982,
      "loss": 2.3722,
      "step": 11775
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.5882710814476013,
      "learning_rate": 0.0005961409798018529,
      "loss": 2.3993,
      "step": 11776
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6387542486190796,
      "learning_rate": 0.0005961403257788277,
      "loss": 2.4289,
      "step": 11777
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6554719805717468,
      "learning_rate": 0.0005961396717007445,
      "loss": 2.3783,
      "step": 11778
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6036887764930725,
      "learning_rate": 0.0005961390175676034,
      "loss": 2.276,
      "step": 11779
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6255596280097961,
      "learning_rate": 0.0005961383633794042,
      "loss": 2.4915,
      "step": 11780
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.7690397500991821,
      "learning_rate": 0.0005961377091361475,
      "loss": 2.4282,
      "step": 11781
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6462006568908691,
      "learning_rate": 0.0005961370548378331,
      "loss": 2.3963,
      "step": 11782
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6113114953041077,
      "learning_rate": 0.0005961364004844614,
      "loss": 2.4669,
      "step": 11783
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6026763916015625,
      "learning_rate": 0.0005961357460760321,
      "loss": 2.2413,
      "step": 11784
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6746910214424133,
      "learning_rate": 0.0005961350916125458,
      "loss": 2.4438,
      "step": 11785
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.7460949420928955,
      "learning_rate": 0.0005961344370940023,
      "loss": 2.3368,
      "step": 11786
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6862072944641113,
      "learning_rate": 0.0005961337825204018,
      "loss": 2.6093,
      "step": 11787
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6475768685340881,
      "learning_rate": 0.0005961331278917445,
      "loss": 2.4962,
      "step": 11788
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.7564918398857117,
      "learning_rate": 0.0005961324732080305,
      "loss": 2.3723,
      "step": 11789
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6194743514060974,
      "learning_rate": 0.0005961318184692598,
      "loss": 2.4587,
      "step": 11790
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.752168595790863,
      "learning_rate": 0.0005961311636754325,
      "loss": 2.7868,
      "step": 11791
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.7053651809692383,
      "learning_rate": 0.000596130508826549,
      "loss": 2.5107,
      "step": 11792
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.642936646938324,
      "learning_rate": 0.0005961298539226092,
      "loss": 2.63,
      "step": 11793
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.5730147957801819,
      "learning_rate": 0.0005961291989636132,
      "loss": 2.3239,
      "step": 11794
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6331064701080322,
      "learning_rate": 0.0005961285439495613,
      "loss": 2.4873,
      "step": 11795
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6411354541778564,
      "learning_rate": 0.0005961278888804534,
      "loss": 2.5193,
      "step": 11796
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6114848256111145,
      "learning_rate": 0.0005961272337562898,
      "loss": 2.647,
      "step": 11797
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.7432637810707092,
      "learning_rate": 0.0005961265785770706,
      "loss": 2.4842,
      "step": 11798
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6316579580307007,
      "learning_rate": 0.0005961259233427959,
      "loss": 2.5989,
      "step": 11799
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.8446943759918213,
      "learning_rate": 0.0005961252680534657,
      "loss": 2.3252,
      "step": 11800
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.7591243386268616,
      "learning_rate": 0.0005961246127090802,
      "loss": 2.7612,
      "step": 11801
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6082806587219238,
      "learning_rate": 0.0005961239573096396,
      "loss": 2.7705,
      "step": 11802
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.7530112862586975,
      "learning_rate": 0.000596123301855144,
      "loss": 2.4301,
      "step": 11803
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.8001552224159241,
      "learning_rate": 0.0005961226463455934,
      "loss": 2.3713,
      "step": 11804
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.7256406545639038,
      "learning_rate": 0.0005961219907809882,
      "loss": 2.5568,
      "step": 11805
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.0975943803787231,
      "learning_rate": 0.0005961213351613282,
      "loss": 2.3328,
      "step": 11806
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6408949494361877,
      "learning_rate": 0.0005961206794866136,
      "loss": 2.5306,
      "step": 11807
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.7669581174850464,
      "learning_rate": 0.0005961200237568446,
      "loss": 2.3939,
      "step": 11808
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6414011716842651,
      "learning_rate": 0.0005961193679720213,
      "loss": 2.4827,
      "step": 11809
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6305474638938904,
      "learning_rate": 0.0005961187121321439,
      "loss": 2.5101,
      "step": 11810
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.7006016373634338,
      "learning_rate": 0.0005961180562372123,
      "loss": 2.4176,
      "step": 11811
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6722041964530945,
      "learning_rate": 0.0005961174002872269,
      "loss": 2.3863,
      "step": 11812
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6350228786468506,
      "learning_rate": 0.0005961167442821877,
      "loss": 2.5493,
      "step": 11813
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6882438659667969,
      "learning_rate": 0.0005961160882220948,
      "loss": 2.5311,
      "step": 11814
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6889041662216187,
      "learning_rate": 0.0005961154321069482,
      "loss": 2.5525,
      "step": 11815
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.5867998600006104,
      "learning_rate": 0.0005961147759367482,
      "loss": 2.2808,
      "step": 11816
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6969982385635376,
      "learning_rate": 0.000596114119711495,
      "loss": 2.3857,
      "step": 11817
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6619728207588196,
      "learning_rate": 0.0005961134634311884,
      "loss": 2.4079,
      "step": 11818
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.739494800567627,
      "learning_rate": 0.0005961128070958289,
      "loss": 2.553,
      "step": 11819
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.661586582660675,
      "learning_rate": 0.0005961121507054163,
      "loss": 2.5593,
      "step": 11820
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6822705864906311,
      "learning_rate": 0.000596111494259951,
      "loss": 2.597,
      "step": 11821
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6219005584716797,
      "learning_rate": 0.0005961108377594329,
      "loss": 2.2268,
      "step": 11822
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.7417631149291992,
      "learning_rate": 0.0005961101812038622,
      "loss": 2.5608,
      "step": 11823
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6436653733253479,
      "learning_rate": 0.000596109524593239,
      "loss": 2.3133,
      "step": 11824
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.5850890278816223,
      "learning_rate": 0.0005961088679275634,
      "loss": 2.5198,
      "step": 11825
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6211220622062683,
      "learning_rate": 0.0005961082112068358,
      "loss": 2.3243,
      "step": 11826
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.8636797070503235,
      "learning_rate": 0.0005961075544310559,
      "loss": 2.6428,
      "step": 11827
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.58903568983078,
      "learning_rate": 0.0005961068976002241,
      "loss": 2.2663,
      "step": 11828
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6204880475997925,
      "learning_rate": 0.0005961062407143404,
      "loss": 2.3083,
      "step": 11829
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6348477005958557,
      "learning_rate": 0.0005961055837734049,
      "loss": 2.4211,
      "step": 11830
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6186686158180237,
      "learning_rate": 0.0005961049267774179,
      "loss": 2.4881,
      "step": 11831
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6595578789710999,
      "learning_rate": 0.0005961042697263793,
      "loss": 2.5232,
      "step": 11832
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6347153782844543,
      "learning_rate": 0.0005961036126202894,
      "loss": 2.7367,
      "step": 11833
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6903033256530762,
      "learning_rate": 0.0005961029554591483,
      "loss": 2.3783,
      "step": 11834
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6298293471336365,
      "learning_rate": 0.0005961022982429559,
      "loss": 2.4605,
      "step": 11835
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6583850979804993,
      "learning_rate": 0.0005961016409717126,
      "loss": 2.5053,
      "step": 11836
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6532574892044067,
      "learning_rate": 0.0005961009836454184,
      "loss": 2.4805,
      "step": 11837
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6233096718788147,
      "learning_rate": 0.0005961003262640735,
      "loss": 2.5393,
      "step": 11838
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6723080277442932,
      "learning_rate": 0.0005960996688276779,
      "loss": 2.5678,
      "step": 11839
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6184588074684143,
      "learning_rate": 0.0005960990113362318,
      "loss": 2.3286,
      "step": 11840
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6602113842964172,
      "learning_rate": 0.0005960983537897353,
      "loss": 2.6245,
      "step": 11841
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6701827049255371,
      "learning_rate": 0.0005960976961881886,
      "loss": 2.4432,
      "step": 11842
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6093883514404297,
      "learning_rate": 0.0005960970385315916,
      "loss": 2.6661,
      "step": 11843
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.7004104852676392,
      "learning_rate": 0.0005960963808199446,
      "loss": 2.5389,
      "step": 11844
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6177834868431091,
      "learning_rate": 0.0005960957230532477,
      "loss": 2.3585,
      "step": 11845
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6243053078651428,
      "learning_rate": 0.0005960950652315011,
      "loss": 2.4932,
      "step": 11846
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.7456081509590149,
      "learning_rate": 0.0005960944073547048,
      "loss": 2.4268,
      "step": 11847
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.5824837684631348,
      "learning_rate": 0.000596093749422859,
      "loss": 2.3014,
      "step": 11848
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6732528209686279,
      "learning_rate": 0.0005960930914359636,
      "loss": 2.2974,
      "step": 11849
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6005662679672241,
      "learning_rate": 0.0005960924333940191,
      "loss": 2.3722,
      "step": 11850
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6567248702049255,
      "learning_rate": 0.0005960917752970255,
      "loss": 2.6246,
      "step": 11851
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6314740180969238,
      "learning_rate": 0.0005960911171449826,
      "loss": 2.5373,
      "step": 11852
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6623558402061462,
      "learning_rate": 0.0005960904589378909,
      "loss": 2.4488,
      "step": 11853
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.5676780343055725,
      "learning_rate": 0.0005960898006757504,
      "loss": 2.5384,
      "step": 11854
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6108441352844238,
      "learning_rate": 0.0005960891423585612,
      "loss": 2.2079,
      "step": 11855
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.5955575704574585,
      "learning_rate": 0.0005960884839863235,
      "loss": 2.2943,
      "step": 11856
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.5758733153343201,
      "learning_rate": 0.0005960878255590372,
      "loss": 2.1971,
      "step": 11857
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6111462116241455,
      "learning_rate": 0.0005960871670767027,
      "loss": 2.3904,
      "step": 11858
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6773044466972351,
      "learning_rate": 0.00059608650853932,
      "loss": 2.3678,
      "step": 11859
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.5902470946311951,
      "learning_rate": 0.0005960858499468892,
      "loss": 2.1848,
      "step": 11860
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6681932806968689,
      "learning_rate": 0.0005960851912994104,
      "loss": 2.2909,
      "step": 11861
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6360501050949097,
      "learning_rate": 0.0005960845325968839,
      "loss": 2.6629,
      "step": 11862
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6945191025733948,
      "learning_rate": 0.0005960838738393097,
      "loss": 2.3452,
      "step": 11863
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6615024209022522,
      "learning_rate": 0.0005960832150266878,
      "loss": 2.5834,
      "step": 11864
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6861979961395264,
      "learning_rate": 0.0005960825561590185,
      "loss": 2.309,
      "step": 11865
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6588037014007568,
      "learning_rate": 0.0005960818972363019,
      "loss": 2.6249,
      "step": 11866
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.9050395488739014,
      "learning_rate": 0.000596081238258538,
      "loss": 2.4051,
      "step": 11867
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6028799414634705,
      "learning_rate": 0.0005960805792257271,
      "loss": 2.2277,
      "step": 11868
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.5930482149124146,
      "learning_rate": 0.0005960799201378692,
      "loss": 2.545,
      "step": 11869
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.706283688545227,
      "learning_rate": 0.0005960792609949644,
      "loss": 2.4746,
      "step": 11870
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6803207397460938,
      "learning_rate": 0.000596078601797013,
      "loss": 2.5088,
      "step": 11871
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6267485618591309,
      "learning_rate": 0.0005960779425440148,
      "loss": 2.3805,
      "step": 11872
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.622014045715332,
      "learning_rate": 0.0005960772832359704,
      "loss": 2.2819,
      "step": 11873
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.8124858140945435,
      "learning_rate": 0.0005960766238728794,
      "loss": 2.4615,
      "step": 11874
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6860105991363525,
      "learning_rate": 0.0005960759644547423,
      "loss": 2.612,
      "step": 11875
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.5904139876365662,
      "learning_rate": 0.000596075304981559,
      "loss": 2.5472,
      "step": 11876
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.7176240086555481,
      "learning_rate": 0.0005960746454533298,
      "loss": 2.5525,
      "step": 11877
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.5888896584510803,
      "learning_rate": 0.0005960739858700546,
      "loss": 2.3129,
      "step": 11878
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.7048057317733765,
      "learning_rate": 0.0005960733262317338,
      "loss": 2.3175,
      "step": 11879
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6711780428886414,
      "learning_rate": 0.0005960726665383674,
      "loss": 2.5578,
      "step": 11880
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.7065590023994446,
      "learning_rate": 0.0005960720067899553,
      "loss": 2.4209,
      "step": 11881
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.8371025323867798,
      "learning_rate": 0.000596071346986498,
      "loss": 2.2212,
      "step": 11882
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6038088798522949,
      "learning_rate": 0.0005960706871279953,
      "loss": 2.5274,
      "step": 11883
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6298666000366211,
      "learning_rate": 0.0005960700272144477,
      "loss": 2.4476,
      "step": 11884
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6751270294189453,
      "learning_rate": 0.000596069367245855,
      "loss": 2.3496,
      "step": 11885
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6364180445671082,
      "learning_rate": 0.0005960687072222174,
      "loss": 2.4409,
      "step": 11886
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.7573513984680176,
      "learning_rate": 0.000596068047143535,
      "loss": 2.5026,
      "step": 11887
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.607159435749054,
      "learning_rate": 0.0005960673870098079,
      "loss": 2.6041,
      "step": 11888
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6067250370979309,
      "learning_rate": 0.0005960667268210364,
      "loss": 2.2968,
      "step": 11889
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.7704424262046814,
      "learning_rate": 0.0005960660665772204,
      "loss": 2.443,
      "step": 11890
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.7249883413314819,
      "learning_rate": 0.0005960654062783604,
      "loss": 2.3456,
      "step": 11891
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.7214980721473694,
      "learning_rate": 0.000596064745924456,
      "loss": 2.4171,
      "step": 11892
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6743715405464172,
      "learning_rate": 0.0005960640855155075,
      "loss": 2.475,
      "step": 11893
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6639725565910339,
      "learning_rate": 0.0005960634250515152,
      "loss": 2.3787,
      "step": 11894
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.642245352268219,
      "learning_rate": 0.0005960627645324792,
      "loss": 2.4673,
      "step": 11895
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.8102311491966248,
      "learning_rate": 0.0005960621039583995,
      "loss": 2.5604,
      "step": 11896
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.5915446877479553,
      "learning_rate": 0.0005960614433292762,
      "loss": 2.4206,
      "step": 11897
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6189004778862,
      "learning_rate": 0.0005960607826451097,
      "loss": 2.2535,
      "step": 11898
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.743337869644165,
      "learning_rate": 0.0005960601219058996,
      "loss": 2.5117,
      "step": 11899
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.7212623953819275,
      "learning_rate": 0.0005960594611116466,
      "loss": 2.4555,
      "step": 11900
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.5859333872795105,
      "learning_rate": 0.0005960588002623504,
      "loss": 2.4729,
      "step": 11901
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6992912888526917,
      "learning_rate": 0.0005960581393580114,
      "loss": 2.6638,
      "step": 11902
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6978188753128052,
      "learning_rate": 0.0005960574783986295,
      "loss": 2.6088,
      "step": 11903
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.5809839367866516,
      "learning_rate": 0.0005960568173842049,
      "loss": 2.5,
      "step": 11904
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6139849424362183,
      "learning_rate": 0.0005960561563147378,
      "loss": 2.3693,
      "step": 11905
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6454300284385681,
      "learning_rate": 0.0005960554951902284,
      "loss": 2.4758,
      "step": 11906
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6978464126586914,
      "learning_rate": 0.0005960548340106764,
      "loss": 2.4532,
      "step": 11907
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6280727386474609,
      "learning_rate": 0.0005960541727760826,
      "loss": 2.6148,
      "step": 11908
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.620793879032135,
      "learning_rate": 0.0005960535114864464,
      "loss": 2.3064,
      "step": 11909
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.7666211724281311,
      "learning_rate": 0.0005960528501417684,
      "loss": 2.535,
      "step": 11910
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.5820550322532654,
      "learning_rate": 0.0005960521887420486,
      "loss": 2.234,
      "step": 11911
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6039104461669922,
      "learning_rate": 0.0005960515272872871,
      "loss": 2.236,
      "step": 11912
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.7518450021743774,
      "learning_rate": 0.000596050865777484,
      "loss": 2.5421,
      "step": 11913
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6981897950172424,
      "learning_rate": 0.0005960502042126395,
      "loss": 2.5142,
      "step": 11914
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6729994416236877,
      "learning_rate": 0.0005960495425927537,
      "loss": 2.5837,
      "step": 11915
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6273000836372375,
      "learning_rate": 0.0005960488809178267,
      "loss": 2.2516,
      "step": 11916
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6147109866142273,
      "learning_rate": 0.0005960482191878586,
      "loss": 2.4942,
      "step": 11917
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6713438630104065,
      "learning_rate": 0.0005960475574028495,
      "loss": 2.4094,
      "step": 11918
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.5935841202735901,
      "learning_rate": 0.0005960468955627995,
      "loss": 2.5672,
      "step": 11919
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6781858205795288,
      "learning_rate": 0.000596046233667709,
      "loss": 2.6742,
      "step": 11920
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.7743045091629028,
      "learning_rate": 0.0005960455717175778,
      "loss": 2.2221,
      "step": 11921
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6443683505058289,
      "learning_rate": 0.0005960449097124062,
      "loss": 2.365,
      "step": 11922
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.662365734577179,
      "learning_rate": 0.0005960442476521942,
      "loss": 2.5682,
      "step": 11923
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.5990147590637207,
      "learning_rate": 0.000596043585536942,
      "loss": 2.2457,
      "step": 11924
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.610175371170044,
      "learning_rate": 0.0005960429233666497,
      "loss": 2.5551,
      "step": 11925
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.5777228474617004,
      "learning_rate": 0.0005960422611413175,
      "loss": 2.5188,
      "step": 11926
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.5826134085655212,
      "learning_rate": 0.0005960415988609454,
      "loss": 2.4605,
      "step": 11927
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.5914050936698914,
      "learning_rate": 0.0005960409365255335,
      "loss": 2.5543,
      "step": 11928
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.5994159579277039,
      "learning_rate": 0.0005960402741350821,
      "loss": 2.6291,
      "step": 11929
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6309251189231873,
      "learning_rate": 0.0005960396116895913,
      "loss": 2.358,
      "step": 11930
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6006885170936584,
      "learning_rate": 0.000596038949189061,
      "loss": 2.3378,
      "step": 11931
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.5925979018211365,
      "learning_rate": 0.0005960382866334915,
      "loss": 2.4927,
      "step": 11932
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.579635739326477,
      "learning_rate": 0.000596037624022883,
      "loss": 2.389,
      "step": 11933
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.5929218530654907,
      "learning_rate": 0.0005960369613572353,
      "loss": 2.4717,
      "step": 11934
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.668941855430603,
      "learning_rate": 0.000596036298636549,
      "loss": 2.4758,
      "step": 11935
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.5964135527610779,
      "learning_rate": 0.0005960356358608238,
      "loss": 2.5695,
      "step": 11936
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6275980472564697,
      "learning_rate": 0.0005960349730300601,
      "loss": 2.3263,
      "step": 11937
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.5888087153434753,
      "learning_rate": 0.0005960343101442578,
      "loss": 2.4767,
      "step": 11938
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6049890518188477,
      "learning_rate": 0.0005960336472034172,
      "loss": 2.3238,
      "step": 11939
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6256352663040161,
      "learning_rate": 0.0005960329842075383,
      "loss": 2.5878,
      "step": 11940
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6959105730056763,
      "learning_rate": 0.0005960323211566214,
      "loss": 2.4323,
      "step": 11941
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6217500567436218,
      "learning_rate": 0.0005960316580506663,
      "loss": 2.5536,
      "step": 11942
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6424882411956787,
      "learning_rate": 0.0005960309948896735,
      "loss": 2.6787,
      "step": 11943
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6615626811981201,
      "learning_rate": 0.0005960303316736429,
      "loss": 2.3827,
      "step": 11944
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.5906899571418762,
      "learning_rate": 0.0005960296684025747,
      "loss": 2.4942,
      "step": 11945
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6101887226104736,
      "learning_rate": 0.0005960290050764688,
      "loss": 2.4637,
      "step": 11946
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6783953309059143,
      "learning_rate": 0.0005960283416953258,
      "loss": 2.597,
      "step": 11947
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6310749650001526,
      "learning_rate": 0.0005960276782591454,
      "loss": 2.2937,
      "step": 11948
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.634704053401947,
      "learning_rate": 0.0005960270147679279,
      "loss": 2.4149,
      "step": 11949
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6069240570068359,
      "learning_rate": 0.0005960263512216734,
      "loss": 2.5175,
      "step": 11950
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6389649510383606,
      "learning_rate": 0.000596025687620382,
      "loss": 2.3993,
      "step": 11951
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6361013054847717,
      "learning_rate": 0.0005960250239640539,
      "loss": 2.6056,
      "step": 11952
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6058818101882935,
      "learning_rate": 0.000596024360252689,
      "loss": 2.3694,
      "step": 11953
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.5877243280410767,
      "learning_rate": 0.0005960236964862877,
      "loss": 2.4091,
      "step": 11954
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.7119395732879639,
      "learning_rate": 0.00059602303266485,
      "loss": 2.5668,
      "step": 11955
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.5712053775787354,
      "learning_rate": 0.000596022368788376,
      "loss": 2.49,
      "step": 11956
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6981078386306763,
      "learning_rate": 0.0005960217048568659,
      "loss": 2.6068,
      "step": 11957
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6452879905700684,
      "learning_rate": 0.0005960210408703197,
      "loss": 2.6723,
      "step": 11958
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6016330122947693,
      "learning_rate": 0.0005960203768287376,
      "loss": 2.053,
      "step": 11959
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.5902830362319946,
      "learning_rate": 0.0005960197127321198,
      "loss": 2.4553,
      "step": 11960
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6469347476959229,
      "learning_rate": 0.0005960190485804663,
      "loss": 2.5369,
      "step": 11961
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6557285189628601,
      "learning_rate": 0.0005960183843737773,
      "loss": 2.3748,
      "step": 11962
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.599545419216156,
      "learning_rate": 0.0005960177201120529,
      "loss": 2.3783,
      "step": 11963
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6033300161361694,
      "learning_rate": 0.0005960170557952932,
      "loss": 2.6236,
      "step": 11964
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6400107145309448,
      "learning_rate": 0.0005960163914234984,
      "loss": 2.1103,
      "step": 11965
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6748949289321899,
      "learning_rate": 0.0005960157269966685,
      "loss": 2.6581,
      "step": 11966
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.5345866084098816,
      "learning_rate": 0.0005960150625148036,
      "loss": 2.4506,
      "step": 11967
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6512179374694824,
      "learning_rate": 0.0005960143979779041,
      "loss": 2.387,
      "step": 11968
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6136133074760437,
      "learning_rate": 0.0005960137333859698,
      "loss": 2.3053,
      "step": 11969
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6611376404762268,
      "learning_rate": 0.000596013068739001,
      "loss": 2.5674,
      "step": 11970
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6902095079421997,
      "learning_rate": 0.0005960124040369978,
      "loss": 2.2504,
      "step": 11971
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6373786330223083,
      "learning_rate": 0.0005960117392799603,
      "loss": 2.4351,
      "step": 11972
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.610459566116333,
      "learning_rate": 0.0005960110744678886,
      "loss": 2.1251,
      "step": 11973
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.5918134450912476,
      "learning_rate": 0.000596010409600783,
      "loss": 2.4435,
      "step": 11974
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.7949053049087524,
      "learning_rate": 0.0005960097446786433,
      "loss": 2.3722,
      "step": 11975
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6811559796333313,
      "learning_rate": 0.0005960090797014698,
      "loss": 2.3311,
      "step": 11976
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6422229409217834,
      "learning_rate": 0.0005960084146692627,
      "loss": 2.5042,
      "step": 11977
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.7524557709693909,
      "learning_rate": 0.0005960077495820221,
      "loss": 2.5141,
      "step": 11978
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.620318591594696,
      "learning_rate": 0.0005960070844397479,
      "loss": 2.5278,
      "step": 11979
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.7001469135284424,
      "learning_rate": 0.0005960064192424405,
      "loss": 2.2789,
      "step": 11980
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.764343798160553,
      "learning_rate": 0.0005960057539901,
      "loss": 2.3093,
      "step": 11981
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6618667244911194,
      "learning_rate": 0.0005960050886827262,
      "loss": 2.2729,
      "step": 11982
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.924699604511261,
      "learning_rate": 0.0005960044233203198,
      "loss": 2.5632,
      "step": 11983
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.7482582926750183,
      "learning_rate": 0.0005960037579028803,
      "loss": 2.6697,
      "step": 11984
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6863317489624023,
      "learning_rate": 0.0005960030924304083,
      "loss": 2.3284,
      "step": 11985
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6941779255867004,
      "learning_rate": 0.0005960024269029036,
      "loss": 2.3327,
      "step": 11986
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.658230185508728,
      "learning_rate": 0.0005960017613203665,
      "loss": 2.4039,
      "step": 11987
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.65619957447052,
      "learning_rate": 0.0005960010956827972,
      "loss": 2.1035,
      "step": 11988
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.8114888668060303,
      "learning_rate": 0.0005960004299901955,
      "loss": 2.4252,
      "step": 11989
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6773231029510498,
      "learning_rate": 0.0005959997642425619,
      "loss": 2.4737,
      "step": 11990
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.5845602750778198,
      "learning_rate": 0.0005959990984398964,
      "loss": 2.3121,
      "step": 11991
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6532509326934814,
      "learning_rate": 0.0005959984325821989,
      "loss": 2.3441,
      "step": 11992
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.715991199016571,
      "learning_rate": 0.0005959977666694698,
      "loss": 2.4023,
      "step": 11993
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6497959494590759,
      "learning_rate": 0.0005959971007017091,
      "loss": 2.3403,
      "step": 11994
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6148560047149658,
      "learning_rate": 0.000595996434678917,
      "loss": 2.3326,
      "step": 11995
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6385363340377808,
      "learning_rate": 0.0005959957686010936,
      "loss": 2.418,
      "step": 11996
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.649816632270813,
      "learning_rate": 0.000595995102468239,
      "loss": 2.491,
      "step": 11997
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6752068400382996,
      "learning_rate": 0.0005959944362803532,
      "loss": 2.626,
      "step": 11998
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6317740082740784,
      "learning_rate": 0.0005959937700374364,
      "loss": 2.546,
      "step": 11999
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.8529885411262512,
      "learning_rate": 0.000595993103739489,
      "loss": 2.6032,
      "step": 12000
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.5902942419052124,
      "learning_rate": 0.0005959924373865108,
      "loss": 2.4256,
      "step": 12001
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.7000183463096619,
      "learning_rate": 0.000595991770978502,
      "loss": 2.4388,
      "step": 12002
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.7452231645584106,
      "learning_rate": 0.0005959911045154627,
      "loss": 2.5726,
      "step": 12003
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.5530989170074463,
      "learning_rate": 0.0005959904379973931,
      "loss": 2.2896,
      "step": 12004
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.5970464944839478,
      "learning_rate": 0.0005959897714242933,
      "loss": 2.5852,
      "step": 12005
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.7171601057052612,
      "learning_rate": 0.0005959891047961634,
      "loss": 2.5294,
      "step": 12006
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.7040982842445374,
      "learning_rate": 0.0005959884381130035,
      "loss": 2.5551,
      "step": 12007
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.8164087533950806,
      "learning_rate": 0.0005959877713748139,
      "loss": 2.6059,
      "step": 12008
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6265565752983093,
      "learning_rate": 0.0005959871045815944,
      "loss": 2.2907,
      "step": 12009
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6012459993362427,
      "learning_rate": 0.0005959864377333455,
      "loss": 2.5219,
      "step": 12010
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6723520755767822,
      "learning_rate": 0.000595985770830067,
      "loss": 2.4504,
      "step": 12011
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6561788320541382,
      "learning_rate": 0.0005959851038717593,
      "loss": 2.4975,
      "step": 12012
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.5872759222984314,
      "learning_rate": 0.0005959844368584223,
      "loss": 2.6018,
      "step": 12013
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6180245280265808,
      "learning_rate": 0.0005959837697900562,
      "loss": 2.5155,
      "step": 12014
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6336045265197754,
      "learning_rate": 0.000595983102666661,
      "loss": 2.4315,
      "step": 12015
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6185802817344666,
      "learning_rate": 0.0005959824354882372,
      "loss": 2.4628,
      "step": 12016
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6127337217330933,
      "learning_rate": 0.0005959817682547845,
      "loss": 2.178,
      "step": 12017
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6058192253112793,
      "learning_rate": 0.0005959811009663033,
      "loss": 2.4938,
      "step": 12018
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6039780974388123,
      "learning_rate": 0.0005959804336227936,
      "loss": 2.5059,
      "step": 12019
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.5863058567047119,
      "learning_rate": 0.0005959797662242555,
      "loss": 2.5013,
      "step": 12020
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.7860399484634399,
      "learning_rate": 0.0005959790987706892,
      "loss": 2.3489,
      "step": 12021
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.619172990322113,
      "learning_rate": 0.0005959784312620948,
      "loss": 2.4415,
      "step": 12022
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6644783020019531,
      "learning_rate": 0.0005959777636984724,
      "loss": 2.3135,
      "step": 12023
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.679373562335968,
      "learning_rate": 0.0005959770960798222,
      "loss": 2.5294,
      "step": 12024
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.793609082698822,
      "learning_rate": 0.0005959764284061442,
      "loss": 2.6098,
      "step": 12025
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.5953798294067383,
      "learning_rate": 0.0005959757606774387,
      "loss": 2.3218,
      "step": 12026
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.69632488489151,
      "learning_rate": 0.0005959750928937056,
      "loss": 2.4575,
      "step": 12027
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.7293112874031067,
      "learning_rate": 0.0005959744250549452,
      "loss": 2.1856,
      "step": 12028
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6136937737464905,
      "learning_rate": 0.0005959737571611576,
      "loss": 2.2227,
      "step": 12029
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6111305952072144,
      "learning_rate": 0.0005959730892123429,
      "loss": 2.7581,
      "step": 12030
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.5870492458343506,
      "learning_rate": 0.0005959724212085012,
      "loss": 2.3385,
      "step": 12031
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.5747693181037903,
      "learning_rate": 0.0005959717531496325,
      "loss": 2.3645,
      "step": 12032
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.616540789604187,
      "learning_rate": 0.0005959710850357373,
      "loss": 2.5579,
      "step": 12033
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.5508792400360107,
      "learning_rate": 0.0005959704168668153,
      "loss": 2.058,
      "step": 12034
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.5853741765022278,
      "learning_rate": 0.000595969748642867,
      "loss": 2.4386,
      "step": 12035
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6456872224807739,
      "learning_rate": 0.0005959690803638921,
      "loss": 2.1596,
      "step": 12036
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6534593105316162,
      "learning_rate": 0.0005959684120298911,
      "loss": 2.4383,
      "step": 12037
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6925458908081055,
      "learning_rate": 0.000595967743640864,
      "loss": 2.5634,
      "step": 12038
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6334072351455688,
      "learning_rate": 0.0005959670751968108,
      "loss": 2.6658,
      "step": 12039
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.7314723134040833,
      "learning_rate": 0.0005959664066977319,
      "loss": 2.4422,
      "step": 12040
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6878793835639954,
      "learning_rate": 0.0005959657381436272,
      "loss": 2.5252,
      "step": 12041
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6742544770240784,
      "learning_rate": 0.0005959650695344968,
      "loss": 2.3828,
      "step": 12042
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.8502618074417114,
      "learning_rate": 0.000595964400870341,
      "loss": 2.3254,
      "step": 12043
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6546152234077454,
      "learning_rate": 0.0005959637321511597,
      "loss": 2.3969,
      "step": 12044
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6865438222885132,
      "learning_rate": 0.0005959630633769533,
      "loss": 2.6483,
      "step": 12045
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.7035430073738098,
      "learning_rate": 0.0005959623945477217,
      "loss": 2.3389,
      "step": 12046
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6096007227897644,
      "learning_rate": 0.0005959617256634652,
      "loss": 2.1489,
      "step": 12047
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6527408361434937,
      "learning_rate": 0.0005959610567241837,
      "loss": 2.2719,
      "step": 12048
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.5985191464424133,
      "learning_rate": 0.0005959603877298775,
      "loss": 2.4338,
      "step": 12049
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6177800893783569,
      "learning_rate": 0.0005959597186805467,
      "loss": 2.2815,
      "step": 12050
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6456542015075684,
      "learning_rate": 0.0005959590495761914,
      "loss": 2.3275,
      "step": 12051
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6384612917900085,
      "learning_rate": 0.0005959583804168117,
      "loss": 2.417,
      "step": 12052
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6261629462242126,
      "learning_rate": 0.0005959577112024078,
      "loss": 2.4711,
      "step": 12053
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.5920979976654053,
      "learning_rate": 0.0005959570419329797,
      "loss": 2.182,
      "step": 12054
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6888094544410706,
      "learning_rate": 0.0005959563726085276,
      "loss": 2.4641,
      "step": 12055
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.5912419557571411,
      "learning_rate": 0.0005959557032290517,
      "loss": 2.4122,
      "step": 12056
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6993262767791748,
      "learning_rate": 0.000595955033794552,
      "loss": 2.6886,
      "step": 12057
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6222591400146484,
      "learning_rate": 0.0005959543643050287,
      "loss": 2.409,
      "step": 12058
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.7749051451683044,
      "learning_rate": 0.0005959536947604819,
      "loss": 2.5995,
      "step": 12059
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.5841348171234131,
      "learning_rate": 0.0005959530251609117,
      "loss": 2.6536,
      "step": 12060
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6110056042671204,
      "learning_rate": 0.0005959523555063181,
      "loss": 2.4554,
      "step": 12061
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6196413636207581,
      "learning_rate": 0.0005959516857967016,
      "loss": 2.2934,
      "step": 12062
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6640018820762634,
      "learning_rate": 0.0005959510160320621,
      "loss": 2.5128,
      "step": 12063
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.677498996257782,
      "learning_rate": 0.0005959503462123996,
      "loss": 2.4311,
      "step": 12064
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6056123375892639,
      "learning_rate": 0.0005959496763377144,
      "loss": 2.4995,
      "step": 12065
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6375855207443237,
      "learning_rate": 0.0005959490064080065,
      "loss": 2.4546,
      "step": 12066
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6456995606422424,
      "learning_rate": 0.0005959483364232762,
      "loss": 2.4151,
      "step": 12067
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.580078125,
      "learning_rate": 0.0005959476663835234,
      "loss": 2.4074,
      "step": 12068
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.5966961979866028,
      "learning_rate": 0.0005959469962887484,
      "loss": 2.5416,
      "step": 12069
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6194609999656677,
      "learning_rate": 0.0005959463261389513,
      "loss": 2.2584,
      "step": 12070
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6169577240943909,
      "learning_rate": 0.0005959456559341321,
      "loss": 2.3543,
      "step": 12071
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6190685033798218,
      "learning_rate": 0.0005959449856742912,
      "loss": 2.4554,
      "step": 12072
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6416359543800354,
      "learning_rate": 0.0005959443153594283,
      "loss": 2.4601,
      "step": 12073
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.633491039276123,
      "learning_rate": 0.0005959436449895439,
      "loss": 2.3861,
      "step": 12074
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.7637479305267334,
      "learning_rate": 0.000595942974564638,
      "loss": 2.407,
      "step": 12075
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.5901463627815247,
      "learning_rate": 0.0005959423040847107,
      "loss": 2.1135,
      "step": 12076
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6324986815452576,
      "learning_rate": 0.0005959416335497621,
      "loss": 2.4883,
      "step": 12077
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.7284421324729919,
      "learning_rate": 0.0005959409629597924,
      "loss": 2.6176,
      "step": 12078
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.7801578640937805,
      "learning_rate": 0.0005959402923148017,
      "loss": 2.4545,
      "step": 12079
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6748027801513672,
      "learning_rate": 0.00059593962161479,
      "loss": 2.4832,
      "step": 12080
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.7921614646911621,
      "learning_rate": 0.0005959389508597578,
      "loss": 2.2486,
      "step": 12081
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6883832216262817,
      "learning_rate": 0.0005959382800497046,
      "loss": 2.3734,
      "step": 12082
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.716610848903656,
      "learning_rate": 0.0005959376091846311,
      "loss": 2.7145,
      "step": 12083
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6722906231880188,
      "learning_rate": 0.0005959369382645373,
      "loss": 2.4087,
      "step": 12084
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.8480040431022644,
      "learning_rate": 0.000595936267289423,
      "loss": 2.551,
      "step": 12085
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6367802619934082,
      "learning_rate": 0.0005959355962592888,
      "loss": 2.3951,
      "step": 12086
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.7679480910301208,
      "learning_rate": 0.0005959349251741345,
      "loss": 2.4389,
      "step": 12087
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.734123706817627,
      "learning_rate": 0.0005959342540339602,
      "loss": 2.4745,
      "step": 12088
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.5935754179954529,
      "learning_rate": 0.0005959335828387663,
      "loss": 2.357,
      "step": 12089
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6713042855262756,
      "learning_rate": 0.0005959329115885527,
      "loss": 2.4258,
      "step": 12090
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.7254728674888611,
      "learning_rate": 0.0005959322402833195,
      "loss": 2.3808,
      "step": 12091
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.5725638270378113,
      "learning_rate": 0.000595931568923067,
      "loss": 2.4232,
      "step": 12092
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6036897897720337,
      "learning_rate": 0.0005959308975077952,
      "loss": 2.3136,
      "step": 12093
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.791930615901947,
      "learning_rate": 0.0005959302260375042,
      "loss": 2.7388,
      "step": 12094
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.5779044032096863,
      "learning_rate": 0.0005959295545121943,
      "loss": 2.352,
      "step": 12095
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6205859184265137,
      "learning_rate": 0.0005959288829318656,
      "loss": 2.3016,
      "step": 12096
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.7118037343025208,
      "learning_rate": 0.000595928211296518,
      "loss": 2.4973,
      "step": 12097
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.0727801322937012,
      "learning_rate": 0.0005959275396061516,
      "loss": 2.4893,
      "step": 12098
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.5389978885650635,
      "learning_rate": 0.0005959268678607669,
      "loss": 2.3751,
      "step": 12099
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.7021308541297913,
      "learning_rate": 0.0005959261960603638,
      "loss": 2.207,
      "step": 12100
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6943048238754272,
      "learning_rate": 0.0005959255242049424,
      "loss": 2.3306,
      "step": 12101
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.7772398591041565,
      "learning_rate": 0.0005959248522945028,
      "loss": 2.2605,
      "step": 12102
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6110318303108215,
      "learning_rate": 0.0005959241803290453,
      "loss": 2.3744,
      "step": 12103
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.9310815930366516,
      "learning_rate": 0.0005959235083085698,
      "loss": 2.3968,
      "step": 12104
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.0468688011169434,
      "learning_rate": 0.0005959228362330766,
      "loss": 2.5591,
      "step": 12105
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6966803669929504,
      "learning_rate": 0.0005959221641025658,
      "loss": 2.4033,
      "step": 12106
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.74898362159729,
      "learning_rate": 0.0005959214919170374,
      "loss": 2.5399,
      "step": 12107
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.7486448287963867,
      "learning_rate": 0.0005959208196764916,
      "loss": 2.3757,
      "step": 12108
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6203789114952087,
      "learning_rate": 0.0005959201473809285,
      "loss": 2.3442,
      "step": 12109
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6608865261077881,
      "learning_rate": 0.0005959194750303484,
      "loss": 2.2998,
      "step": 12110
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.7203398942947388,
      "learning_rate": 0.0005959188026247511,
      "loss": 2.3575,
      "step": 12111
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.673646867275238,
      "learning_rate": 0.000595918130164137,
      "loss": 2.4205,
      "step": 12112
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.61042720079422,
      "learning_rate": 0.0005959174576485063,
      "loss": 2.573,
      "step": 12113
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6356189250946045,
      "learning_rate": 0.0005959167850778587,
      "loss": 2.2277,
      "step": 12114
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6526492834091187,
      "learning_rate": 0.0005959161124521947,
      "loss": 2.3732,
      "step": 12115
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6201659440994263,
      "learning_rate": 0.0005959154397715144,
      "loss": 2.3858,
      "step": 12116
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6177546381950378,
      "learning_rate": 0.0005959147670358176,
      "loss": 2.1802,
      "step": 12117
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6109762191772461,
      "learning_rate": 0.0005959140942451049,
      "loss": 2.4316,
      "step": 12118
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.7190537452697754,
      "learning_rate": 0.000595913421399376,
      "loss": 2.2483,
      "step": 12119
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.7124150395393372,
      "learning_rate": 0.0005959127484986313,
      "loss": 2.3096,
      "step": 12120
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.7289009690284729,
      "learning_rate": 0.0005959120755428709,
      "loss": 2.3299,
      "step": 12121
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6285607814788818,
      "learning_rate": 0.0005959114025320946,
      "loss": 2.1912,
      "step": 12122
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.7367085814476013,
      "learning_rate": 0.0005959107294663031,
      "loss": 2.3417,
      "step": 12123
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6832256317138672,
      "learning_rate": 0.000595910056345496,
      "loss": 2.4048,
      "step": 12124
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6142098307609558,
      "learning_rate": 0.0005959093831696738,
      "loss": 2.5514,
      "step": 12125
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.7372921705245972,
      "learning_rate": 0.0005959087099388363,
      "loss": 2.6039,
      "step": 12126
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.733905017375946,
      "learning_rate": 0.000595908036652984,
      "loss": 2.496,
      "step": 12127
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6028193831443787,
      "learning_rate": 0.0005959073633121166,
      "loss": 2.3551,
      "step": 12128
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.8042628169059753,
      "learning_rate": 0.0005959066899162345,
      "loss": 2.2467,
      "step": 12129
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.0438098907470703,
      "learning_rate": 0.0005959060164653377,
      "loss": 2.4275,
      "step": 12130
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6451034545898438,
      "learning_rate": 0.0005959053429594265,
      "loss": 2.61,
      "step": 12131
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.2254871129989624,
      "learning_rate": 0.0005959046693985008,
      "loss": 2.3349,
      "step": 12132
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.0383577346801758,
      "learning_rate": 0.0005959039957825609,
      "loss": 2.2757,
      "step": 12133
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6801068782806396,
      "learning_rate": 0.0005959033221116069,
      "loss": 2.1505,
      "step": 12134
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6567107439041138,
      "learning_rate": 0.0005959026483856388,
      "loss": 2.5575,
      "step": 12135
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.8944458365440369,
      "learning_rate": 0.0005959019746046569,
      "loss": 2.6313,
      "step": 12136
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.7218782305717468,
      "learning_rate": 0.0005959013007686611,
      "loss": 2.6785,
      "step": 12137
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.666021466255188,
      "learning_rate": 0.0005959006268776518,
      "loss": 2.669,
      "step": 12138
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6468877792358398,
      "learning_rate": 0.0005958999529316289,
      "loss": 2.53,
      "step": 12139
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6179128289222717,
      "learning_rate": 0.0005958992789305926,
      "loss": 2.494,
      "step": 12140
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6343403458595276,
      "learning_rate": 0.0005958986048745432,
      "loss": 2.4998,
      "step": 12141
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6718188524246216,
      "learning_rate": 0.0005958979307634806,
      "loss": 2.4216,
      "step": 12142
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.8116128444671631,
      "learning_rate": 0.0005958972565974049,
      "loss": 2.4134,
      "step": 12143
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6593231558799744,
      "learning_rate": 0.0005958965823763164,
      "loss": 2.4487,
      "step": 12144
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.7331723570823669,
      "learning_rate": 0.000595895908100215,
      "loss": 2.3111,
      "step": 12145
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6130020618438721,
      "learning_rate": 0.0005958952337691011,
      "loss": 2.2996,
      "step": 12146
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.5976493954658508,
      "learning_rate": 0.0005958945593829746,
      "loss": 2.4473,
      "step": 12147
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.154972791671753,
      "learning_rate": 0.0005958938849418357,
      "loss": 2.5291,
      "step": 12148
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6539224982261658,
      "learning_rate": 0.0005958932104456847,
      "loss": 2.3655,
      "step": 12149
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.7171534895896912,
      "learning_rate": 0.0005958925358945215,
      "loss": 2.2503,
      "step": 12150
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6468304991722107,
      "learning_rate": 0.0005958918612883463,
      "loss": 2.4217,
      "step": 12151
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6617279648780823,
      "learning_rate": 0.0005958911866271591,
      "loss": 2.5703,
      "step": 12152
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6048163771629333,
      "learning_rate": 0.0005958905119109603,
      "loss": 2.5457,
      "step": 12153
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6158647537231445,
      "learning_rate": 0.0005958898371397498,
      "loss": 2.5149,
      "step": 12154
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.5919598937034607,
      "learning_rate": 0.0005958891623135278,
      "loss": 2.5851,
      "step": 12155
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6475518941879272,
      "learning_rate": 0.0005958884874322944,
      "loss": 2.1811,
      "step": 12156
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6685137152671814,
      "learning_rate": 0.0005958878124960497,
      "loss": 2.7137,
      "step": 12157
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.58171147108078,
      "learning_rate": 0.000595887137504794,
      "loss": 2.4253,
      "step": 12158
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6030445098876953,
      "learning_rate": 0.0005958864624585272,
      "loss": 2.415,
      "step": 12159
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.5831418633460999,
      "learning_rate": 0.0005958857873572496,
      "loss": 2.3998,
      "step": 12160
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.5904157757759094,
      "learning_rate": 0.0005958851122009612,
      "loss": 2.5304,
      "step": 12161
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6564982533454895,
      "learning_rate": 0.0005958844369896621,
      "loss": 2.6485,
      "step": 12162
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6147428750991821,
      "learning_rate": 0.0005958837617233525,
      "loss": 2.3517,
      "step": 12163
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.7278943061828613,
      "learning_rate": 0.0005958830864020327,
      "loss": 2.3719,
      "step": 12164
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.5694571137428284,
      "learning_rate": 0.0005958824110257026,
      "loss": 2.3119,
      "step": 12165
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6018064618110657,
      "learning_rate": 0.0005958817355943623,
      "loss": 2.3265,
      "step": 12166
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.5512104630470276,
      "learning_rate": 0.0005958810601080119,
      "loss": 2.4896,
      "step": 12167
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.5645288228988647,
      "learning_rate": 0.0005958803845666518,
      "loss": 2.3747,
      "step": 12168
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6177809238433838,
      "learning_rate": 0.0005958797089702818,
      "loss": 2.5428,
      "step": 12169
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.5989894866943359,
      "learning_rate": 0.0005958790333189023,
      "loss": 2.6604,
      "step": 12170
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.5566153526306152,
      "learning_rate": 0.0005958783576125134,
      "loss": 2.4117,
      "step": 12171
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.5953989028930664,
      "learning_rate": 0.0005958776818511149,
      "loss": 2.3954,
      "step": 12172
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6569821834564209,
      "learning_rate": 0.0005958770060347072,
      "loss": 2.4056,
      "step": 12173
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6624526381492615,
      "learning_rate": 0.0005958763301632904,
      "loss": 2.2943,
      "step": 12174
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6595308184623718,
      "learning_rate": 0.0005958756542368646,
      "loss": 2.5372,
      "step": 12175
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.5671760439872742,
      "learning_rate": 0.00059587497825543,
      "loss": 2.4132,
      "step": 12176
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.639407217502594,
      "learning_rate": 0.0005958743022189866,
      "loss": 2.3643,
      "step": 12177
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.649322509765625,
      "learning_rate": 0.0005958736261275347,
      "loss": 2.6748,
      "step": 12178
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.5956193208694458,
      "learning_rate": 0.0005958729499810742,
      "loss": 2.7005,
      "step": 12179
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6067251563072205,
      "learning_rate": 0.0005958722737796053,
      "loss": 2.5156,
      "step": 12180
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.7276591062545776,
      "learning_rate": 0.0005958715975231283,
      "loss": 2.4058,
      "step": 12181
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6510406732559204,
      "learning_rate": 0.0005958709212116431,
      "loss": 2.306,
      "step": 12182
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.7336041927337646,
      "learning_rate": 0.0005958702448451499,
      "loss": 2.492,
      "step": 12183
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.7584144473075867,
      "learning_rate": 0.0005958695684236487,
      "loss": 2.4634,
      "step": 12184
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.5505803227424622,
      "learning_rate": 0.0005958688919471399,
      "loss": 2.5577,
      "step": 12185
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.730521023273468,
      "learning_rate": 0.0005958682154156235,
      "loss": 2.5801,
      "step": 12186
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6886211633682251,
      "learning_rate": 0.0005958675388290996,
      "loss": 2.4928,
      "step": 12187
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6282191872596741,
      "learning_rate": 0.0005958668621875684,
      "loss": 2.3824,
      "step": 12188
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6996548175811768,
      "learning_rate": 0.0005958661854910299,
      "loss": 2.4298,
      "step": 12189
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6289024353027344,
      "learning_rate": 0.0005958655087394844,
      "loss": 2.5232,
      "step": 12190
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6155381798744202,
      "learning_rate": 0.0005958648319329317,
      "loss": 2.4186,
      "step": 12191
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6284354329109192,
      "learning_rate": 0.0005958641550713723,
      "loss": 2.4993,
      "step": 12192
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.5922391414642334,
      "learning_rate": 0.0005958634781548061,
      "loss": 2.4438,
      "step": 12193
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6284885406494141,
      "learning_rate": 0.0005958628011832333,
      "loss": 2.4821,
      "step": 12194
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6282495856285095,
      "learning_rate": 0.000595862124156654,
      "loss": 2.4705,
      "step": 12195
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6357755661010742,
      "learning_rate": 0.0005958614470750684,
      "loss": 2.7053,
      "step": 12196
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.589059591293335,
      "learning_rate": 0.0005958607699384766,
      "loss": 2.4598,
      "step": 12197
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6312928199768066,
      "learning_rate": 0.0005958600927468787,
      "loss": 2.496,
      "step": 12198
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.5795687437057495,
      "learning_rate": 0.0005958594155002747,
      "loss": 2.4499,
      "step": 12199
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.5844430923461914,
      "learning_rate": 0.000595858738198665,
      "loss": 2.6432,
      "step": 12200
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6402779817581177,
      "learning_rate": 0.0005958580608420495,
      "loss": 2.4909,
      "step": 12201
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6407081484794617,
      "learning_rate": 0.0005958573834304283,
      "loss": 2.3686,
      "step": 12202
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.5832959413528442,
      "learning_rate": 0.0005958567059638018,
      "loss": 2.2968,
      "step": 12203
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6085513234138489,
      "learning_rate": 0.0005958560284421699,
      "loss": 2.4201,
      "step": 12204
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.590552806854248,
      "learning_rate": 0.0005958553508655328,
      "loss": 2.2549,
      "step": 12205
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6031258702278137,
      "learning_rate": 0.0005958546732338906,
      "loss": 2.3346,
      "step": 12206
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6990598440170288,
      "learning_rate": 0.0005958539955472434,
      "loss": 2.3962,
      "step": 12207
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6816282272338867,
      "learning_rate": 0.0005958533178055913,
      "loss": 2.3227,
      "step": 12208
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6112331748008728,
      "learning_rate": 0.0005958526400089346,
      "loss": 2.5535,
      "step": 12209
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6596196293830872,
      "learning_rate": 0.0005958519621572732,
      "loss": 2.2384,
      "step": 12210
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.645377516746521,
      "learning_rate": 0.0005958512842506074,
      "loss": 2.5986,
      "step": 12211
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6531816124916077,
      "learning_rate": 0.0005958506062889372,
      "loss": 2.5016,
      "step": 12212
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.5897014141082764,
      "learning_rate": 0.000595849928272263,
      "loss": 2.2539,
      "step": 12213
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.7982143759727478,
      "learning_rate": 0.0005958492502005845,
      "loss": 2.4682,
      "step": 12214
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6592479348182678,
      "learning_rate": 0.0005958485720739021,
      "loss": 2.3506,
      "step": 12215
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.5737791657447815,
      "learning_rate": 0.0005958478938922158,
      "loss": 2.0916,
      "step": 12216
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6666169166564941,
      "learning_rate": 0.0005958472156555259,
      "loss": 2.4875,
      "step": 12217
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6486328840255737,
      "learning_rate": 0.0005958465373638324,
      "loss": 2.5473,
      "step": 12218
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6731278300285339,
      "learning_rate": 0.0005958458590171353,
      "loss": 2.2954,
      "step": 12219
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.7452244162559509,
      "learning_rate": 0.0005958451806154351,
      "loss": 2.5827,
      "step": 12220
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.5587005615234375,
      "learning_rate": 0.0005958445021587316,
      "loss": 2.2289,
      "step": 12221
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6163980960845947,
      "learning_rate": 0.0005958438236470249,
      "loss": 2.4111,
      "step": 12222
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.5937985777854919,
      "learning_rate": 0.0005958431450803154,
      "loss": 2.3781,
      "step": 12223
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6629444360733032,
      "learning_rate": 0.000595842466458603,
      "loss": 2.5039,
      "step": 12224
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.7163636684417725,
      "learning_rate": 0.000595841787781888,
      "loss": 2.5047,
      "step": 12225
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6087387800216675,
      "learning_rate": 0.0005958411090501703,
      "loss": 2.5188,
      "step": 12226
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6775175333023071,
      "learning_rate": 0.0005958404302634502,
      "loss": 2.5078,
      "step": 12227
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6339508295059204,
      "learning_rate": 0.0005958397514217278,
      "loss": 2.3643,
      "step": 12228
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.5877841711044312,
      "learning_rate": 0.0005958390725250032,
      "loss": 2.4218,
      "step": 12229
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6107394099235535,
      "learning_rate": 0.0005958383935732765,
      "loss": 2.4958,
      "step": 12230
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.7217422127723694,
      "learning_rate": 0.000595837714566548,
      "loss": 2.2731,
      "step": 12231
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.616097629070282,
      "learning_rate": 0.0005958370355048175,
      "loss": 2.3839,
      "step": 12232
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.5979983806610107,
      "learning_rate": 0.0005958363563880854,
      "loss": 2.3947,
      "step": 12233
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6205058097839355,
      "learning_rate": 0.0005958356772163516,
      "loss": 2.3822,
      "step": 12234
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6321347951889038,
      "learning_rate": 0.0005958349979896166,
      "loss": 2.5068,
      "step": 12235
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.556568443775177,
      "learning_rate": 0.0005958343187078801,
      "loss": 2.3096,
      "step": 12236
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6611916422843933,
      "learning_rate": 0.0005958336393711425,
      "loss": 2.2342,
      "step": 12237
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.7322202920913696,
      "learning_rate": 0.0005958329599794038,
      "loss": 2.4147,
      "step": 12238
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6081271767616272,
      "learning_rate": 0.0005958322805326643,
      "loss": 2.3965,
      "step": 12239
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6865025758743286,
      "learning_rate": 0.0005958316010309238,
      "loss": 2.542,
      "step": 12240
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6342522501945496,
      "learning_rate": 0.0005958309214741828,
      "loss": 2.4272,
      "step": 12241
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6460334658622742,
      "learning_rate": 0.0005958302418624411,
      "loss": 2.5053,
      "step": 12242
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.7230326533317566,
      "learning_rate": 0.000595829562195699,
      "loss": 2.3801,
      "step": 12243
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6088948249816895,
      "learning_rate": 0.0005958288824739566,
      "loss": 2.4158,
      "step": 12244
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6126999855041504,
      "learning_rate": 0.0005958282026972141,
      "loss": 2.3883,
      "step": 12245
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6524773240089417,
      "learning_rate": 0.0005958275228654715,
      "loss": 2.444,
      "step": 12246
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6434701681137085,
      "learning_rate": 0.000595826842978729,
      "loss": 2.2475,
      "step": 12247
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6244715452194214,
      "learning_rate": 0.0005958261630369867,
      "loss": 2.6056,
      "step": 12248
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.685020923614502,
      "learning_rate": 0.0005958254830402448,
      "loss": 2.4751,
      "step": 12249
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6783138513565063,
      "learning_rate": 0.0005958248029885033,
      "loss": 2.4463,
      "step": 12250
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6047438979148865,
      "learning_rate": 0.0005958241228817623,
      "loss": 2.3139,
      "step": 12251
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6510053873062134,
      "learning_rate": 0.0005958234427200221,
      "loss": 2.3255,
      "step": 12252
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6234787106513977,
      "learning_rate": 0.0005958227625032827,
      "loss": 2.2894,
      "step": 12253
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6341988444328308,
      "learning_rate": 0.0005958220822315444,
      "loss": 2.4786,
      "step": 12254
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.5843548774719238,
      "learning_rate": 0.000595821401904807,
      "loss": 2.3182,
      "step": 12255
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.8051236271858215,
      "learning_rate": 0.0005958207215230709,
      "loss": 2.5657,
      "step": 12256
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6190229654312134,
      "learning_rate": 0.0005958200410863361,
      "loss": 2.4863,
      "step": 12257
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.8407018780708313,
      "learning_rate": 0.0005958193605946028,
      "loss": 2.58,
      "step": 12258
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6668974757194519,
      "learning_rate": 0.0005958186800478712,
      "loss": 2.5825,
      "step": 12259
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6394312977790833,
      "learning_rate": 0.0005958179994461412,
      "loss": 2.5701,
      "step": 12260
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6933131814002991,
      "learning_rate": 0.0005958173187894132,
      "loss": 2.5496,
      "step": 12261
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.5808754563331604,
      "learning_rate": 0.0005958166380776869,
      "loss": 2.4736,
      "step": 12262
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6642571091651917,
      "learning_rate": 0.0005958159573109629,
      "loss": 2.4466,
      "step": 12263
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.61529141664505,
      "learning_rate": 0.0005958152764892411,
      "loss": 2.5291,
      "step": 12264
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.8805144429206848,
      "learning_rate": 0.0005958145956125217,
      "loss": 2.4305,
      "step": 12265
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.5906698703765869,
      "learning_rate": 0.0005958139146808047,
      "loss": 2.3339,
      "step": 12266
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.7080219388008118,
      "learning_rate": 0.0005958132336940903,
      "loss": 2.3516,
      "step": 12267
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6289828419685364,
      "learning_rate": 0.0005958125526523786,
      "loss": 2.2208,
      "step": 12268
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.5684627294540405,
      "learning_rate": 0.0005958118715556699,
      "loss": 2.4392,
      "step": 12269
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6195226907730103,
      "learning_rate": 0.0005958111904039641,
      "loss": 2.415,
      "step": 12270
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.8502873182296753,
      "learning_rate": 0.0005958105091972615,
      "loss": 2.3712,
      "step": 12271
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6960150003433228,
      "learning_rate": 0.000595809827935562,
      "loss": 2.3703,
      "step": 12272
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6614586114883423,
      "learning_rate": 0.000595809146618866,
      "loss": 2.3315,
      "step": 12273
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6787287592887878,
      "learning_rate": 0.0005958084652471735,
      "loss": 2.3796,
      "step": 12274
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.5546265244483948,
      "learning_rate": 0.0005958077838204846,
      "loss": 2.2756,
      "step": 12275
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6551249027252197,
      "learning_rate": 0.0005958071023387994,
      "loss": 2.6462,
      "step": 12276
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6256275177001953,
      "learning_rate": 0.000595806420802118,
      "loss": 2.4012,
      "step": 12277
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.59605872631073,
      "learning_rate": 0.0005958057392104407,
      "loss": 2.1163,
      "step": 12278
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6082004904747009,
      "learning_rate": 0.0005958050575637676,
      "loss": 2.4569,
      "step": 12279
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.5647637844085693,
      "learning_rate": 0.0005958043758620987,
      "loss": 2.4023,
      "step": 12280
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6121503114700317,
      "learning_rate": 0.000595803694105434,
      "loss": 2.3329,
      "step": 12281
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6534958481788635,
      "learning_rate": 0.0005958030122937742,
      "loss": 2.1654,
      "step": 12282
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.8075503706932068,
      "learning_rate": 0.0005958023304271187,
      "loss": 2.5226,
      "step": 12283
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6292729377746582,
      "learning_rate": 0.0005958016485054682,
      "loss": 2.2939,
      "step": 12284
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6723456978797913,
      "learning_rate": 0.0005958009665288224,
      "loss": 2.5709,
      "step": 12285
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.856702446937561,
      "learning_rate": 0.0005958002844971817,
      "loss": 2.3971,
      "step": 12286
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6939182877540588,
      "learning_rate": 0.0005957996024105461,
      "loss": 2.5274,
      "step": 12287
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.9126936793327332,
      "learning_rate": 0.0005957989202689158,
      "loss": 2.4396,
      "step": 12288
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6632241010665894,
      "learning_rate": 0.0005957982380722909,
      "loss": 2.5749,
      "step": 12289
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6240130662918091,
      "learning_rate": 0.0005957975558206715,
      "loss": 2.3593,
      "step": 12290
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.8156013488769531,
      "learning_rate": 0.0005957968735140577,
      "loss": 2.5326,
      "step": 12291
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.5866040587425232,
      "learning_rate": 0.0005957961911524498,
      "loss": 2.372,
      "step": 12292
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6019151210784912,
      "learning_rate": 0.0005957955087358478,
      "loss": 2.4203,
      "step": 12293
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6311354041099548,
      "learning_rate": 0.0005957948262642518,
      "loss": 2.3282,
      "step": 12294
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6824167370796204,
      "learning_rate": 0.0005957941437376619,
      "loss": 2.4961,
      "step": 12295
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.7209900617599487,
      "learning_rate": 0.0005957934611560783,
      "loss": 2.3681,
      "step": 12296
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.5836352705955505,
      "learning_rate": 0.0005957927785195011,
      "loss": 2.5118,
      "step": 12297
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.699221670627594,
      "learning_rate": 0.0005957920958279305,
      "loss": 2.3878,
      "step": 12298
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.7897758483886719,
      "learning_rate": 0.0005957914130813666,
      "loss": 2.446,
      "step": 12299
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6176161170005798,
      "learning_rate": 0.0005957907302798094,
      "loss": 2.3723,
      "step": 12300
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.7856740355491638,
      "learning_rate": 0.0005957900474232591,
      "loss": 2.2795,
      "step": 12301
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6370342373847961,
      "learning_rate": 0.000595789364511716,
      "loss": 2.4745,
      "step": 12302
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6329460144042969,
      "learning_rate": 0.00059578868154518,
      "loss": 2.3059,
      "step": 12303
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.7160779237747192,
      "learning_rate": 0.0005957879985236512,
      "loss": 2.405,
      "step": 12304
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6523833870887756,
      "learning_rate": 0.00059578731544713,
      "loss": 2.449,
      "step": 12305
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.63715660572052,
      "learning_rate": 0.0005957866323156162,
      "loss": 2.6007,
      "step": 12306
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.5943587422370911,
      "learning_rate": 0.0005957859491291101,
      "loss": 2.2344,
      "step": 12307
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6114950180053711,
      "learning_rate": 0.0005957852658876119,
      "loss": 2.4183,
      "step": 12308
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.698673665523529,
      "learning_rate": 0.0005957845825911215,
      "loss": 2.5658,
      "step": 12309
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.642361581325531,
      "learning_rate": 0.0005957838992396394,
      "loss": 2.4251,
      "step": 12310
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.5494600534439087,
      "learning_rate": 0.0005957832158331653,
      "loss": 2.459,
      "step": 12311
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6848791241645813,
      "learning_rate": 0.0005957825323716995,
      "loss": 2.2751,
      "step": 12312
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.744550883769989,
      "learning_rate": 0.0005957818488552422,
      "loss": 2.466,
      "step": 12313
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6163972616195679,
      "learning_rate": 0.0005957811652837935,
      "loss": 2.5261,
      "step": 12314
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.5832148194313049,
      "learning_rate": 0.0005957804816573534,
      "loss": 2.4416,
      "step": 12315
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.7245216965675354,
      "learning_rate": 0.0005957797979759222,
      "loss": 2.4149,
      "step": 12316
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6633694171905518,
      "learning_rate": 0.0005957791142395,
      "loss": 2.5672,
      "step": 12317
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6712243556976318,
      "learning_rate": 0.0005957784304480867,
      "loss": 2.3178,
      "step": 12318
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6819769740104675,
      "learning_rate": 0.0005957777466016827,
      "loss": 2.5231,
      "step": 12319
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.746830940246582,
      "learning_rate": 0.0005957770627002881,
      "loss": 2.3118,
      "step": 12320
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.7149978876113892,
      "learning_rate": 0.000595776378743903,
      "loss": 2.6299,
      "step": 12321
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6194916367530823,
      "learning_rate": 0.0005957756947325274,
      "loss": 2.5221,
      "step": 12322
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6554877758026123,
      "learning_rate": 0.0005957750106661615,
      "loss": 2.5084,
      "step": 12323
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.7073716521263123,
      "learning_rate": 0.0005957743265448054,
      "loss": 2.4458,
      "step": 12324
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.7553874254226685,
      "learning_rate": 0.0005957736423684595,
      "loss": 2.5128,
      "step": 12325
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.663147509098053,
      "learning_rate": 0.0005957729581371235,
      "loss": 2.4189,
      "step": 12326
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6785917282104492,
      "learning_rate": 0.0005957722738507978,
      "loss": 2.4793,
      "step": 12327
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6640896797180176,
      "learning_rate": 0.0005957715895094823,
      "loss": 2.5704,
      "step": 12328
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6508941054344177,
      "learning_rate": 0.0005957709051131775,
      "loss": 2.6005,
      "step": 12329
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6782536506652832,
      "learning_rate": 0.0005957702206618831,
      "loss": 2.2894,
      "step": 12330
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.5953578352928162,
      "learning_rate": 0.0005957695361555995,
      "loss": 2.3601,
      "step": 12331
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.7000299692153931,
      "learning_rate": 0.0005957688515943268,
      "loss": 2.4295,
      "step": 12332
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.61517333984375,
      "learning_rate": 0.0005957681669780652,
      "loss": 2.4807,
      "step": 12333
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6245983242988586,
      "learning_rate": 0.0005957674823068145,
      "loss": 2.4728,
      "step": 12334
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6166977286338806,
      "learning_rate": 0.0005957667975805751,
      "loss": 2.7226,
      "step": 12335
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6358797550201416,
      "learning_rate": 0.0005957661127993471,
      "loss": 2.3529,
      "step": 12336
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6933853626251221,
      "learning_rate": 0.0005957654279631306,
      "loss": 2.4476,
      "step": 12337
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6725096106529236,
      "learning_rate": 0.0005957647430719258,
      "loss": 2.6414,
      "step": 12338
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.5735477209091187,
      "learning_rate": 0.0005957640581257326,
      "loss": 2.3816,
      "step": 12339
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6858884692192078,
      "learning_rate": 0.0005957633731245514,
      "loss": 2.5915,
      "step": 12340
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6833366751670837,
      "learning_rate": 0.0005957626880683821,
      "loss": 2.4991,
      "step": 12341
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6092836856842041,
      "learning_rate": 0.0005957620029572251,
      "loss": 2.3614,
      "step": 12342
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6614179611206055,
      "learning_rate": 0.0005957613177910801,
      "loss": 2.3781,
      "step": 12343
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6638736724853516,
      "learning_rate": 0.0005957606325699477,
      "loss": 2.4172,
      "step": 12344
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6109820008277893,
      "learning_rate": 0.0005957599472938278,
      "loss": 2.3829,
      "step": 12345
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6827453970909119,
      "learning_rate": 0.0005957592619627205,
      "loss": 2.3403,
      "step": 12346
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6905596852302551,
      "learning_rate": 0.0005957585765766259,
      "loss": 2.3855,
      "step": 12347
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.582362174987793,
      "learning_rate": 0.0005957578911355443,
      "loss": 2.2051,
      "step": 12348
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6995424032211304,
      "learning_rate": 0.0005957572056394758,
      "loss": 2.383,
      "step": 12349
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6072821617126465,
      "learning_rate": 0.0005957565200884202,
      "loss": 2.6043,
      "step": 12350
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6331631541252136,
      "learning_rate": 0.0005957558344823781,
      "loss": 2.4236,
      "step": 12351
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6724332571029663,
      "learning_rate": 0.0005957551488213493,
      "loss": 2.5063,
      "step": 12352
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6227366924285889,
      "learning_rate": 0.000595754463105334,
      "loss": 2.4975,
      "step": 12353
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6076623201370239,
      "learning_rate": 0.0005957537773343325,
      "loss": 2.4281,
      "step": 12354
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.624475359916687,
      "learning_rate": 0.0005957530915083447,
      "loss": 2.5788,
      "step": 12355
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.7029486298561096,
      "learning_rate": 0.0005957524056273708,
      "loss": 2.5786,
      "step": 12356
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6625258922576904,
      "learning_rate": 0.000595751719691411,
      "loss": 2.2038,
      "step": 12357
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6316978931427002,
      "learning_rate": 0.0005957510337004654,
      "loss": 2.4778,
      "step": 12358
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6582883596420288,
      "learning_rate": 0.000595750347654534,
      "loss": 2.4647,
      "step": 12359
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.5863237977027893,
      "learning_rate": 0.0005957496615536171,
      "loss": 2.2661,
      "step": 12360
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.5848878622055054,
      "learning_rate": 0.0005957489753977147,
      "loss": 2.4101,
      "step": 12361
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.5793195366859436,
      "learning_rate": 0.000595748289186827,
      "loss": 2.3843,
      "step": 12362
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6294224262237549,
      "learning_rate": 0.0005957476029209541,
      "loss": 2.5034,
      "step": 12363
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6502212882041931,
      "learning_rate": 0.0005957469166000963,
      "loss": 2.2623,
      "step": 12364
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6276658773422241,
      "learning_rate": 0.0005957462302242535,
      "loss": 2.6339,
      "step": 12365
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6454776525497437,
      "learning_rate": 0.0005957455437934257,
      "loss": 2.5237,
      "step": 12366
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.623837411403656,
      "learning_rate": 0.0005957448573076133,
      "loss": 2.561,
      "step": 12367
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.5764253735542297,
      "learning_rate": 0.0005957441707668164,
      "loss": 2.4564,
      "step": 12368
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.695762038230896,
      "learning_rate": 0.0005957434841710352,
      "loss": 2.566,
      "step": 12369
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6626880168914795,
      "learning_rate": 0.0005957427975202696,
      "loss": 2.3495,
      "step": 12370
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6113166213035583,
      "learning_rate": 0.0005957421108145197,
      "loss": 2.3865,
      "step": 12371
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.7927023768424988,
      "learning_rate": 0.0005957414240537859,
      "loss": 2.2948,
      "step": 12372
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.7378587126731873,
      "learning_rate": 0.0005957407372380681,
      "loss": 2.3541,
      "step": 12373
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.7407943606376648,
      "learning_rate": 0.0005957400503673666,
      "loss": 2.5535,
      "step": 12374
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.7767994403839111,
      "learning_rate": 0.0005957393634416815,
      "loss": 2.582,
      "step": 12375
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.7476010918617249,
      "learning_rate": 0.0005957386764610127,
      "loss": 2.3194,
      "step": 12376
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.5958960652351379,
      "learning_rate": 0.0005957379894253606,
      "loss": 2.4027,
      "step": 12377
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6776840090751648,
      "learning_rate": 0.0005957373023347252,
      "loss": 2.5681,
      "step": 12378
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6963869333267212,
      "learning_rate": 0.0005957366151891066,
      "loss": 2.2728,
      "step": 12379
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6149833798408508,
      "learning_rate": 0.0005957359279885051,
      "loss": 2.5489,
      "step": 12380
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.588263213634491,
      "learning_rate": 0.0005957352407329207,
      "loss": 2.3534,
      "step": 12381
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6185320019721985,
      "learning_rate": 0.0005957345534223534,
      "loss": 2.3737,
      "step": 12382
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.7619665861129761,
      "learning_rate": 0.0005957338660568036,
      "loss": 2.4456,
      "step": 12383
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.5892338156700134,
      "learning_rate": 0.0005957331786362712,
      "loss": 2.3572,
      "step": 12384
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6232090592384338,
      "learning_rate": 0.0005957324911607565,
      "loss": 2.4088,
      "step": 12385
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6151285767555237,
      "learning_rate": 0.0005957318036302594,
      "loss": 2.2205,
      "step": 12386
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.7705705761909485,
      "learning_rate": 0.0005957311160447803,
      "loss": 2.3554,
      "step": 12387
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.613609254360199,
      "learning_rate": 0.0005957304284043192,
      "loss": 2.2843,
      "step": 12388
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6265242099761963,
      "learning_rate": 0.0005957297407088762,
      "loss": 2.4507,
      "step": 12389
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.7276263236999512,
      "learning_rate": 0.0005957290529584515,
      "loss": 2.315,
      "step": 12390
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6102842092514038,
      "learning_rate": 0.0005957283651530452,
      "loss": 2.4236,
      "step": 12391
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6737645864486694,
      "learning_rate": 0.0005957276772926574,
      "loss": 2.5396,
      "step": 12392
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6212086081504822,
      "learning_rate": 0.0005957269893772882,
      "loss": 2.4731,
      "step": 12393
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6641502380371094,
      "learning_rate": 0.0005957263014069377,
      "loss": 2.3865,
      "step": 12394
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6119101643562317,
      "learning_rate": 0.0005957256133816062,
      "loss": 2.5083,
      "step": 12395
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.5563268661499023,
      "learning_rate": 0.0005957249253012938,
      "loss": 2.5057,
      "step": 12396
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6705081462860107,
      "learning_rate": 0.0005957242371660003,
      "loss": 2.3541,
      "step": 12397
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6295661330223083,
      "learning_rate": 0.0005957235489757264,
      "loss": 2.4005,
      "step": 12398
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6289983987808228,
      "learning_rate": 0.0005957228607304717,
      "loss": 2.5674,
      "step": 12399
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6349841952323914,
      "learning_rate": 0.0005957221724302366,
      "loss": 2.5248,
      "step": 12400
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.7030736804008484,
      "learning_rate": 0.0005957214840750212,
      "loss": 2.3588,
      "step": 12401
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.5879294872283936,
      "learning_rate": 0.0005957207956648255,
      "loss": 2.5148,
      "step": 12402
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.710746705532074,
      "learning_rate": 0.0005957201071996498,
      "loss": 2.3883,
      "step": 12403
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.7255792617797852,
      "learning_rate": 0.0005957194186794942,
      "loss": 2.6239,
      "step": 12404
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.5902495384216309,
      "learning_rate": 0.0005957187301043586,
      "loss": 2.206,
      "step": 12405
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6337420344352722,
      "learning_rate": 0.0005957180414742435,
      "loss": 2.5198,
      "step": 12406
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6376080513000488,
      "learning_rate": 0.0005957173527891486,
      "loss": 2.3602,
      "step": 12407
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6975991725921631,
      "learning_rate": 0.0005957166640490746,
      "loss": 2.595,
      "step": 12408
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6369723677635193,
      "learning_rate": 0.000595715975254021,
      "loss": 2.4092,
      "step": 12409
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6752029061317444,
      "learning_rate": 0.0005957152864039883,
      "loss": 2.512,
      "step": 12410
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.7653048634529114,
      "learning_rate": 0.0005957145974989765,
      "loss": 2.5725,
      "step": 12411
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6864395141601562,
      "learning_rate": 0.0005957139085389859,
      "loss": 2.6253,
      "step": 12412
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6313852071762085,
      "learning_rate": 0.0005957132195240164,
      "loss": 2.4488,
      "step": 12413
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6275789737701416,
      "learning_rate": 0.0005957125304540683,
      "loss": 2.5937,
      "step": 12414
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.5904840230941772,
      "learning_rate": 0.0005957118413291416,
      "loss": 2.3674,
      "step": 12415
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.5730214715003967,
      "learning_rate": 0.0005957111521492365,
      "loss": 2.2384,
      "step": 12416
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.5947672724723816,
      "learning_rate": 0.0005957104629143532,
      "loss": 2.3223,
      "step": 12417
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6584564447402954,
      "learning_rate": 0.0005957097736244917,
      "loss": 2.3619,
      "step": 12418
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.5792748928070068,
      "learning_rate": 0.0005957090842796521,
      "loss": 2.4789,
      "step": 12419
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6219573020935059,
      "learning_rate": 0.0005957083948798347,
      "loss": 2.562,
      "step": 12420
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6376219391822815,
      "learning_rate": 0.0005957077054250394,
      "loss": 2.3858,
      "step": 12421
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.5693609118461609,
      "learning_rate": 0.0005957070159152665,
      "loss": 2.5579,
      "step": 12422
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.633263885974884,
      "learning_rate": 0.0005957063263505162,
      "loss": 2.4878,
      "step": 12423
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6033735871315002,
      "learning_rate": 0.0005957056367307883,
      "loss": 2.4135,
      "step": 12424
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6007349491119385,
      "learning_rate": 0.0005957049470560834,
      "loss": 2.4076,
      "step": 12425
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6898699402809143,
      "learning_rate": 0.0005957042573264011,
      "loss": 2.4951,
      "step": 12426
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.5738131403923035,
      "learning_rate": 0.000595703567541742,
      "loss": 2.1997,
      "step": 12427
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6205295324325562,
      "learning_rate": 0.0005957028777021059,
      "loss": 2.4836,
      "step": 12428
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.586568295955658,
      "learning_rate": 0.0005957021878074931,
      "loss": 2.4062,
      "step": 12429
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6357492804527283,
      "learning_rate": 0.0005957014978579037,
      "loss": 2.5369,
      "step": 12430
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6354519128799438,
      "learning_rate": 0.0005957008078533378,
      "loss": 2.2986,
      "step": 12431
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6453379988670349,
      "learning_rate": 0.0005957001177937956,
      "loss": 2.4556,
      "step": 12432
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.7035958766937256,
      "learning_rate": 0.0005956994276792771,
      "loss": 2.5679,
      "step": 12433
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6204046607017517,
      "learning_rate": 0.0005956987375097825,
      "loss": 2.3706,
      "step": 12434
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6153363585472107,
      "learning_rate": 0.0005956980472853119,
      "loss": 2.6749,
      "step": 12435
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6200594902038574,
      "learning_rate": 0.0005956973570058654,
      "loss": 2.416,
      "step": 12436
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6031878590583801,
      "learning_rate": 0.0005956966666714433,
      "loss": 2.5362,
      "step": 12437
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6135590076446533,
      "learning_rate": 0.0005956959762820455,
      "loss": 2.4022,
      "step": 12438
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.7214146256446838,
      "learning_rate": 0.0005956952858376724,
      "loss": 2.3996,
      "step": 12439
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6148656010627747,
      "learning_rate": 0.0005956945953383238,
      "loss": 2.3849,
      "step": 12440
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6263214349746704,
      "learning_rate": 0.000595693904784,
      "loss": 2.534,
      "step": 12441
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6274461150169373,
      "learning_rate": 0.0005956932141747012,
      "loss": 2.1046,
      "step": 12442
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6566115617752075,
      "learning_rate": 0.0005956925235104275,
      "loss": 2.2668,
      "step": 12443
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.5656628012657166,
      "learning_rate": 0.0005956918327911789,
      "loss": 2.4167,
      "step": 12444
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6386486887931824,
      "learning_rate": 0.0005956911420169556,
      "loss": 2.348,
      "step": 12445
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6003882884979248,
      "learning_rate": 0.0005956904511877577,
      "loss": 2.331,
      "step": 12446
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6665410399436951,
      "learning_rate": 0.0005956897603035853,
      "loss": 2.387,
      "step": 12447
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6009839773178101,
      "learning_rate": 0.0005956890693644388,
      "loss": 2.5562,
      "step": 12448
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6147626042366028,
      "learning_rate": 0.000595688378370318,
      "loss": 2.2911,
      "step": 12449
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.5850514769554138,
      "learning_rate": 0.0005956876873212232,
      "loss": 2.4186,
      "step": 12450
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.5979822874069214,
      "learning_rate": 0.0005956869962171545,
      "loss": 2.3196,
      "step": 12451
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6096721291542053,
      "learning_rate": 0.0005956863050581119,
      "loss": 2.381,
      "step": 12452
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.5996695160865784,
      "learning_rate": 0.0005956856138440957,
      "loss": 2.6268,
      "step": 12453
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6891120076179504,
      "learning_rate": 0.0005956849225751059,
      "loss": 2.4344,
      "step": 12454
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.5931925773620605,
      "learning_rate": 0.0005956842312511428,
      "loss": 2.3701,
      "step": 12455
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6252967715263367,
      "learning_rate": 0.0005956835398722064,
      "loss": 2.6398,
      "step": 12456
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.5899118185043335,
      "learning_rate": 0.0005956828484382968,
      "loss": 2.2889,
      "step": 12457
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6328521966934204,
      "learning_rate": 0.0005956821569494141,
      "loss": 2.5154,
      "step": 12458
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6135086417198181,
      "learning_rate": 0.0005956814654055586,
      "loss": 2.3015,
      "step": 12459
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6656346917152405,
      "learning_rate": 0.0005956807738067303,
      "loss": 2.6121,
      "step": 12460
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6988592743873596,
      "learning_rate": 0.0005956800821529294,
      "loss": 2.4921,
      "step": 12461
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.5883285999298096,
      "learning_rate": 0.000595679390444156,
      "loss": 2.4197,
      "step": 12462
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6224285364151001,
      "learning_rate": 0.0005956786986804102,
      "loss": 2.4187,
      "step": 12463
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6439995169639587,
      "learning_rate": 0.0005956780068616922,
      "loss": 2.2736,
      "step": 12464
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.5765979290008545,
      "learning_rate": 0.0005956773149880019,
      "loss": 2.6381,
      "step": 12465
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.5463854074478149,
      "learning_rate": 0.0005956766230593398,
      "loss": 2.5233,
      "step": 12466
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6491203904151917,
      "learning_rate": 0.0005956759310757057,
      "loss": 2.681,
      "step": 12467
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6648562550544739,
      "learning_rate": 0.0005956752390371,
      "loss": 2.5382,
      "step": 12468
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6175904870033264,
      "learning_rate": 0.0005956745469435226,
      "loss": 2.4261,
      "step": 12469
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.5679407119750977,
      "learning_rate": 0.0005956738547949738,
      "loss": 2.1561,
      "step": 12470
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.7346347570419312,
      "learning_rate": 0.0005956731625914534,
      "loss": 2.4467,
      "step": 12471
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6420857906341553,
      "learning_rate": 0.000595672470332962,
      "loss": 2.4578,
      "step": 12472
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6765074729919434,
      "learning_rate": 0.0005956717780194996,
      "loss": 2.571,
      "step": 12473
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6151600480079651,
      "learning_rate": 0.0005956710856510661,
      "loss": 2.5096,
      "step": 12474
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6199254989624023,
      "learning_rate": 0.0005956703932276618,
      "loss": 2.4689,
      "step": 12475
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.7556071281433105,
      "learning_rate": 0.0005956697007492867,
      "loss": 2.4216,
      "step": 12476
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6718012094497681,
      "learning_rate": 0.0005956690082159411,
      "loss": 2.4006,
      "step": 12477
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.7085601687431335,
      "learning_rate": 0.0005956683156276251,
      "loss": 2.5165,
      "step": 12478
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.5894942283630371,
      "learning_rate": 0.0005956676229843386,
      "loss": 2.3414,
      "step": 12479
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.5893047451972961,
      "learning_rate": 0.0005956669302860821,
      "loss": 2.4501,
      "step": 12480
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6754174828529358,
      "learning_rate": 0.0005956662375328555,
      "loss": 2.4838,
      "step": 12481
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6057627201080322,
      "learning_rate": 0.0005956655447246588,
      "loss": 2.4521,
      "step": 12482
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6129095554351807,
      "learning_rate": 0.0005956648518614925,
      "loss": 2.464,
      "step": 12483
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6013817191123962,
      "learning_rate": 0.0005956641589433564,
      "loss": 2.4734,
      "step": 12484
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6483771204948425,
      "learning_rate": 0.0005956634659702508,
      "loss": 2.3805,
      "step": 12485
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.7190366387367249,
      "learning_rate": 0.0005956627729421758,
      "loss": 2.3723,
      "step": 12486
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6422158479690552,
      "learning_rate": 0.0005956620798591315,
      "loss": 2.4683,
      "step": 12487
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6905373930931091,
      "learning_rate": 0.000595661386721118,
      "loss": 2.4482,
      "step": 12488
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6567405462265015,
      "learning_rate": 0.0005956606935281354,
      "loss": 2.342,
      "step": 12489
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6448929309844971,
      "learning_rate": 0.000595660000280184,
      "loss": 2.5752,
      "step": 12490
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6392616033554077,
      "learning_rate": 0.0005956593069772639,
      "loss": 2.5954,
      "step": 12491
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.5741593837738037,
      "learning_rate": 0.000595658613619375,
      "loss": 2.396,
      "step": 12492
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.5720511078834534,
      "learning_rate": 0.0005956579202065176,
      "loss": 2.2597,
      "step": 12493
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.8016802072525024,
      "learning_rate": 0.0005956572267386919,
      "loss": 2.3987,
      "step": 12494
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6177756786346436,
      "learning_rate": 0.0005956565332158979,
      "loss": 2.4027,
      "step": 12495
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6818811893463135,
      "learning_rate": 0.0005956558396381358,
      "loss": 2.5183,
      "step": 12496
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6565409302711487,
      "learning_rate": 0.0005956551460054056,
      "loss": 2.2935,
      "step": 12497
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6273826360702515,
      "learning_rate": 0.0005956544523177076,
      "loss": 2.3447,
      "step": 12498
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.7502606511116028,
      "learning_rate": 0.000595653758575042,
      "loss": 2.3916,
      "step": 12499
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.5990976095199585,
      "learning_rate": 0.0005956530647774085,
      "loss": 2.3017,
      "step": 12500
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.8212986588478088,
      "learning_rate": 0.0005956523709248077,
      "loss": 2.4652,
      "step": 12501
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6223284602165222,
      "learning_rate": 0.0005956516770172395,
      "loss": 2.2495,
      "step": 12502
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6262094974517822,
      "learning_rate": 0.000595650983054704,
      "loss": 2.3601,
      "step": 12503
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6294679045677185,
      "learning_rate": 0.0005956502890372015,
      "loss": 2.373,
      "step": 12504
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.672285258769989,
      "learning_rate": 0.0005956495949647321,
      "loss": 2.6193,
      "step": 12505
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.9059709906578064,
      "learning_rate": 0.0005956489008372957,
      "loss": 2.3006,
      "step": 12506
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6310124397277832,
      "learning_rate": 0.0005956482066548926,
      "loss": 2.3682,
      "step": 12507
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6745541095733643,
      "learning_rate": 0.0005956475124175231,
      "loss": 2.396,
      "step": 12508
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.7314947843551636,
      "learning_rate": 0.0005956468181251869,
      "loss": 2.409,
      "step": 12509
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6434938311576843,
      "learning_rate": 0.0005956461237778846,
      "loss": 2.336,
      "step": 12510
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6053330302238464,
      "learning_rate": 0.0005956454293756159,
      "loss": 2.516,
      "step": 12511
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6500532031059265,
      "learning_rate": 0.0005956447349183812,
      "loss": 2.3571,
      "step": 12512
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.7207421660423279,
      "learning_rate": 0.0005956440404061806,
      "loss": 2.4182,
      "step": 12513
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6561006903648376,
      "learning_rate": 0.0005956433458390142,
      "loss": 2.5968,
      "step": 12514
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6518824696540833,
      "learning_rate": 0.000595642651216882,
      "loss": 2.3223,
      "step": 12515
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.8246752619743347,
      "learning_rate": 0.0005956419565397845,
      "loss": 2.6438,
      "step": 12516
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6785721182823181,
      "learning_rate": 0.0005956412618077214,
      "loss": 2.4595,
      "step": 12517
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6568939089775085,
      "learning_rate": 0.0005956405670206929,
      "loss": 2.3217,
      "step": 12518
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.8237878084182739,
      "learning_rate": 0.0005956398721786995,
      "loss": 2.5272,
      "step": 12519
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6777797937393188,
      "learning_rate": 0.000595639177281741,
      "loss": 2.3536,
      "step": 12520
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.5863814353942871,
      "learning_rate": 0.0005956384823298175,
      "loss": 2.2381,
      "step": 12521
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.7337208390235901,
      "learning_rate": 0.0005956377873229293,
      "loss": 2.6137,
      "step": 12522
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6706114411354065,
      "learning_rate": 0.0005956370922610764,
      "loss": 2.3811,
      "step": 12523
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6578709483146667,
      "learning_rate": 0.000595636397144259,
      "loss": 2.3659,
      "step": 12524
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6196327209472656,
      "learning_rate": 0.0005956357019724772,
      "loss": 2.3143,
      "step": 12525
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6325011253356934,
      "learning_rate": 0.0005956350067457312,
      "loss": 2.4551,
      "step": 12526
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6186221837997437,
      "learning_rate": 0.000595634311464021,
      "loss": 2.4658,
      "step": 12527
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.5950503349304199,
      "learning_rate": 0.0005956336161273469,
      "loss": 2.3278,
      "step": 12528
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6539725065231323,
      "learning_rate": 0.0005956329207357088,
      "loss": 2.5606,
      "step": 12529
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6261513829231262,
      "learning_rate": 0.000595632225289107,
      "loss": 2.4651,
      "step": 12530
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6193493604660034,
      "learning_rate": 0.0005956315297875416,
      "loss": 2.3737,
      "step": 12531
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.5437427759170532,
      "learning_rate": 0.0005956308342310128,
      "loss": 2.4038,
      "step": 12532
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.5882784724235535,
      "learning_rate": 0.0005956301386195205,
      "loss": 2.3504,
      "step": 12533
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.62400221824646,
      "learning_rate": 0.0005956294429530651,
      "loss": 2.2684,
      "step": 12534
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.7006198167800903,
      "learning_rate": 0.0005956287472316465,
      "loss": 2.3466,
      "step": 12535
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6632863283157349,
      "learning_rate": 0.000595628051455265,
      "loss": 2.4145,
      "step": 12536
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6143240928649902,
      "learning_rate": 0.0005956273556239208,
      "loss": 2.2565,
      "step": 12537
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.5327351689338684,
      "learning_rate": 0.0005956266597376136,
      "loss": 2.4268,
      "step": 12538
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.635378360748291,
      "learning_rate": 0.000595625963796344,
      "loss": 2.3542,
      "step": 12539
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6361414790153503,
      "learning_rate": 0.000595625267800112,
      "loss": 2.2699,
      "step": 12540
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.5966184735298157,
      "learning_rate": 0.0005956245717489176,
      "loss": 2.4406,
      "step": 12541
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6266454458236694,
      "learning_rate": 0.000595623875642761,
      "loss": 2.4479,
      "step": 12542
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6083360314369202,
      "learning_rate": 0.0005956231794816424,
      "loss": 2.2882,
      "step": 12543
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6767668724060059,
      "learning_rate": 0.0005956224832655618,
      "loss": 2.2829,
      "step": 12544
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.5979330539703369,
      "learning_rate": 0.0005956217869945194,
      "loss": 2.3864,
      "step": 12545
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6146278381347656,
      "learning_rate": 0.0005956210906685154,
      "loss": 2.5783,
      "step": 12546
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6961871385574341,
      "learning_rate": 0.0005956203942875498,
      "loss": 2.507,
      "step": 12547
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6322394013404846,
      "learning_rate": 0.0005956196978516229,
      "loss": 2.2114,
      "step": 12548
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6629475355148315,
      "learning_rate": 0.0005956190013607346,
      "loss": 2.3677,
      "step": 12549
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.7103893756866455,
      "learning_rate": 0.0005956183048148853,
      "loss": 2.4558,
      "step": 12550
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6281524896621704,
      "learning_rate": 0.0005956176082140747,
      "loss": 2.3559,
      "step": 12551
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6465052366256714,
      "learning_rate": 0.0005956169115583033,
      "loss": 2.4337,
      "step": 12552
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.7375622987747192,
      "learning_rate": 0.0005956162148475714,
      "loss": 2.3449,
      "step": 12553
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.5908907055854797,
      "learning_rate": 0.0005956155180818786,
      "loss": 2.3311,
      "step": 12554
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.571319580078125,
      "learning_rate": 0.0005956148212612253,
      "loss": 2.4039,
      "step": 12555
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.5964644551277161,
      "learning_rate": 0.0005956141243856117,
      "loss": 2.419,
      "step": 12556
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6093254089355469,
      "learning_rate": 0.0005956134274550379,
      "loss": 2.522,
      "step": 12557
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.578244149684906,
      "learning_rate": 0.0005956127304695039,
      "loss": 2.4995,
      "step": 12558
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6757527589797974,
      "learning_rate": 0.0005956120334290099,
      "loss": 2.5774,
      "step": 12559
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6987674832344055,
      "learning_rate": 0.0005956113363335561,
      "loss": 2.4471,
      "step": 12560
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.5940378308296204,
      "learning_rate": 0.0005956106391831425,
      "loss": 2.5813,
      "step": 12561
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.5993611812591553,
      "learning_rate": 0.0005956099419777694,
      "loss": 2.4118,
      "step": 12562
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6107091307640076,
      "learning_rate": 0.0005956092447174368,
      "loss": 2.4837,
      "step": 12563
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.672252893447876,
      "learning_rate": 0.0005956085474021448,
      "loss": 2.322,
      "step": 12564
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6086410880088806,
      "learning_rate": 0.0005956078500318937,
      "loss": 2.2508,
      "step": 12565
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6469944715499878,
      "learning_rate": 0.0005956071526066834,
      "loss": 2.6773,
      "step": 12566
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.7724903225898743,
      "learning_rate": 0.0005956064551265143,
      "loss": 2.5604,
      "step": 12567
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6548627614974976,
      "learning_rate": 0.0005956057575913862,
      "loss": 2.5728,
      "step": 12568
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6976568698883057,
      "learning_rate": 0.0005956050600012995,
      "loss": 2.4173,
      "step": 12569
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6356382966041565,
      "learning_rate": 0.0005956043623562543,
      "loss": 2.4186,
      "step": 12570
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.5920447707176208,
      "learning_rate": 0.0005956036646562506,
      "loss": 2.4045,
      "step": 12571
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.5999201536178589,
      "learning_rate": 0.0005956029669012885,
      "loss": 2.5406,
      "step": 12572
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6819778084754944,
      "learning_rate": 0.0005956022690913684,
      "loss": 2.3955,
      "step": 12573
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6064930558204651,
      "learning_rate": 0.0005956015712264902,
      "loss": 2.3193,
      "step": 12574
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6655203104019165,
      "learning_rate": 0.000595600873306654,
      "loss": 2.3058,
      "step": 12575
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.5961378216743469,
      "learning_rate": 0.0005956001753318601,
      "loss": 2.6027,
      "step": 12576
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6783189177513123,
      "learning_rate": 0.0005955994773021087,
      "loss": 2.3621,
      "step": 12577
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6609869599342346,
      "learning_rate": 0.0005955987792173995,
      "loss": 2.3172,
      "step": 12578
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.5700416564941406,
      "learning_rate": 0.0005955980810777331,
      "loss": 2.5279,
      "step": 12579
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6466618776321411,
      "learning_rate": 0.0005955973828831093,
      "loss": 2.4651,
      "step": 12580
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.7378541827201843,
      "learning_rate": 0.0005955966846335285,
      "loss": 2.5488,
      "step": 12581
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.5884055495262146,
      "learning_rate": 0.0005955959863289906,
      "loss": 2.5586,
      "step": 12582
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6830253005027771,
      "learning_rate": 0.0005955952879694959,
      "loss": 2.5706,
      "step": 12583
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6495095491409302,
      "learning_rate": 0.0005955945895550444,
      "loss": 2.3657,
      "step": 12584
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6321707963943481,
      "learning_rate": 0.0005955938910856362,
      "loss": 2.5542,
      "step": 12585
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.5982640385627747,
      "learning_rate": 0.0005955931925612717,
      "loss": 2.6268,
      "step": 12586
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6514655351638794,
      "learning_rate": 0.0005955924939819508,
      "loss": 2.4527,
      "step": 12587
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6086635589599609,
      "learning_rate": 0.0005955917953476736,
      "loss": 2.2774,
      "step": 12588
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.5930711030960083,
      "learning_rate": 0.0005955910966584403,
      "loss": 2.4236,
      "step": 12589
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6011815071105957,
      "learning_rate": 0.0005955903979142511,
      "loss": 2.1811,
      "step": 12590
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6030049324035645,
      "learning_rate": 0.0005955896991151061,
      "loss": 2.4661,
      "step": 12591
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6853002309799194,
      "learning_rate": 0.0005955890002610053,
      "loss": 2.5008,
      "step": 12592
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6270806193351746,
      "learning_rate": 0.0005955883013519489,
      "loss": 2.5079,
      "step": 12593
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6237412095069885,
      "learning_rate": 0.0005955876023879372,
      "loss": 2.1704,
      "step": 12594
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6491472125053406,
      "learning_rate": 0.0005955869033689701,
      "loss": 2.6444,
      "step": 12595
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6862017512321472,
      "learning_rate": 0.0005955862042950477,
      "loss": 2.3293,
      "step": 12596
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.57359379529953,
      "learning_rate": 0.0005955855051661704,
      "loss": 2.3177,
      "step": 12597
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.5948317050933838,
      "learning_rate": 0.0005955848059823382,
      "loss": 2.3796,
      "step": 12598
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6229901313781738,
      "learning_rate": 0.0005955841067435511,
      "loss": 2.5006,
      "step": 12599
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6729129552841187,
      "learning_rate": 0.0005955834074498095,
      "loss": 2.473,
      "step": 12600
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6033657789230347,
      "learning_rate": 0.0005955827081011131,
      "loss": 2.4564,
      "step": 12601
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6502174139022827,
      "learning_rate": 0.0005955820086974626,
      "loss": 2.3226,
      "step": 12602
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.5878258943557739,
      "learning_rate": 0.0005955813092388576,
      "loss": 2.1039,
      "step": 12603
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6976875066757202,
      "learning_rate": 0.0005955806097252985,
      "loss": 2.3546,
      "step": 12604
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6881551146507263,
      "learning_rate": 0.0005955799101567854,
      "loss": 2.615,
      "step": 12605
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6096228361129761,
      "learning_rate": 0.0005955792105333184,
      "loss": 2.4078,
      "step": 12606
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6482760310173035,
      "learning_rate": 0.0005955785108548976,
      "loss": 2.5647,
      "step": 12607
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6307240128517151,
      "learning_rate": 0.0005955778111215233,
      "loss": 2.384,
      "step": 12608
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.7221387624740601,
      "learning_rate": 0.0005955771113331955,
      "loss": 2.5959,
      "step": 12609
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6778092980384827,
      "learning_rate": 0.0005955764114899142,
      "loss": 2.3887,
      "step": 12610
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6750928163528442,
      "learning_rate": 0.0005955757115916797,
      "loss": 2.3162,
      "step": 12611
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.5619370341300964,
      "learning_rate": 0.0005955750116384921,
      "loss": 2.3363,
      "step": 12612
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.5887474417686462,
      "learning_rate": 0.0005955743116303517,
      "loss": 2.3735,
      "step": 12613
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6418864727020264,
      "learning_rate": 0.0005955736115672583,
      "loss": 2.3633,
      "step": 12614
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.5798753499984741,
      "learning_rate": 0.0005955729114492123,
      "loss": 2.2562,
      "step": 12615
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6025374531745911,
      "learning_rate": 0.0005955722112762136,
      "loss": 2.6191,
      "step": 12616
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.5840688943862915,
      "learning_rate": 0.0005955715110482624,
      "loss": 2.2997,
      "step": 12617
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.5863187313079834,
      "learning_rate": 0.000595570810765359,
      "loss": 2.3258,
      "step": 12618
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6187755465507507,
      "learning_rate": 0.0005955701104275033,
      "loss": 2.3683,
      "step": 12619
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.589793860912323,
      "learning_rate": 0.0005955694100346957,
      "loss": 2.5534,
      "step": 12620
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.5929055213928223,
      "learning_rate": 0.000595568709586936,
      "loss": 2.4522,
      "step": 12621
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6928941011428833,
      "learning_rate": 0.0005955680090842246,
      "loss": 2.6083,
      "step": 12622
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6905994415283203,
      "learning_rate": 0.0005955673085265615,
      "loss": 2.4899,
      "step": 12623
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6089033484458923,
      "learning_rate": 0.0005955666079139467,
      "loss": 2.283,
      "step": 12624
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6755490899085999,
      "learning_rate": 0.0005955659072463807,
      "loss": 2.4614,
      "step": 12625
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.5853464603424072,
      "learning_rate": 0.0005955652065238634,
      "loss": 2.2994,
      "step": 12626
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.5743324160575867,
      "learning_rate": 0.0005955645057463948,
      "loss": 2.3398,
      "step": 12627
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.5781077146530151,
      "learning_rate": 0.0005955638049139753,
      "loss": 2.3826,
      "step": 12628
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6369286179542542,
      "learning_rate": 0.0005955631040266048,
      "loss": 2.4035,
      "step": 12629
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.5872021317481995,
      "learning_rate": 0.0005955624030842837,
      "loss": 2.4344,
      "step": 12630
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6236475110054016,
      "learning_rate": 0.0005955617020870119,
      "loss": 2.673,
      "step": 12631
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.5934102535247803,
      "learning_rate": 0.0005955610010347896,
      "loss": 2.3,
      "step": 12632
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.5887667536735535,
      "learning_rate": 0.0005955602999276169,
      "loss": 2.4781,
      "step": 12633
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.59507155418396,
      "learning_rate": 0.0005955595987654941,
      "loss": 2.587,
      "step": 12634
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6184981465339661,
      "learning_rate": 0.000595558897548421,
      "loss": 2.3722,
      "step": 12635
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6311295628547668,
      "learning_rate": 0.0005955581962763981,
      "loss": 2.369,
      "step": 12636
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6338129043579102,
      "learning_rate": 0.0005955574949494252,
      "loss": 2.4247,
      "step": 12637
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6222009658813477,
      "learning_rate": 0.0005955567935675026,
      "loss": 2.6729,
      "step": 12638
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6328542232513428,
      "learning_rate": 0.0005955560921306305,
      "loss": 2.3169,
      "step": 12639
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6373052597045898,
      "learning_rate": 0.000595555390638809,
      "loss": 2.238,
      "step": 12640
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.5727847814559937,
      "learning_rate": 0.000595554689092038,
      "loss": 2.3926,
      "step": 12641
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.5677606463432312,
      "learning_rate": 0.0005955539874903178,
      "loss": 2.495,
      "step": 12642
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.5970603823661804,
      "learning_rate": 0.0005955532858336486,
      "loss": 2.3907,
      "step": 12643
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6146588921546936,
      "learning_rate": 0.0005955525841220304,
      "loss": 2.4679,
      "step": 12644
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6377555131912231,
      "learning_rate": 0.0005955518823554636,
      "loss": 2.4204,
      "step": 12645
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6364931464195251,
      "learning_rate": 0.000595551180533948,
      "loss": 2.2612,
      "step": 12646
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.7082656025886536,
      "learning_rate": 0.0005955504786574838,
      "loss": 2.5241,
      "step": 12647
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.5711513161659241,
      "learning_rate": 0.0005955497767260713,
      "loss": 2.2983,
      "step": 12648
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6272248029708862,
      "learning_rate": 0.0005955490747397104,
      "loss": 2.3354,
      "step": 12649
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6276767253875732,
      "learning_rate": 0.0005955483726984014,
      "loss": 2.5377,
      "step": 12650
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.7294474244117737,
      "learning_rate": 0.0005955476706021444,
      "loss": 2.4516,
      "step": 12651
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.672818124294281,
      "learning_rate": 0.0005955469684509395,
      "loss": 2.4315,
      "step": 12652
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.8150985240936279,
      "learning_rate": 0.0005955462662447868,
      "loss": 2.4119,
      "step": 12653
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.5878908038139343,
      "learning_rate": 0.0005955455639836865,
      "loss": 2.4536,
      "step": 12654
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.7862368226051331,
      "learning_rate": 0.0005955448616676388,
      "loss": 2.0476,
      "step": 12655
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6025721430778503,
      "learning_rate": 0.0005955441592966435,
      "loss": 2.5733,
      "step": 12656
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6140789985656738,
      "learning_rate": 0.0005955434568707012,
      "loss": 2.24,
      "step": 12657
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.671553909778595,
      "learning_rate": 0.0005955427543898116,
      "loss": 2.2712,
      "step": 12658
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.5999133586883545,
      "learning_rate": 0.0005955420518539751,
      "loss": 2.3284,
      "step": 12659
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6486288905143738,
      "learning_rate": 0.0005955413492631918,
      "loss": 2.4948,
      "step": 12660
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6455615758895874,
      "learning_rate": 0.0005955406466174617,
      "loss": 2.3416,
      "step": 12661
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.5847514271736145,
      "learning_rate": 0.0005955399439167852,
      "loss": 2.3634,
      "step": 12662
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6259042620658875,
      "learning_rate": 0.000595539241161162,
      "loss": 2.3994,
      "step": 12663
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6556034088134766,
      "learning_rate": 0.0005955385383505927,
      "loss": 2.2856,
      "step": 12664
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6406075954437256,
      "learning_rate": 0.000595537835485077,
      "loss": 2.4746,
      "step": 12665
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6471419334411621,
      "learning_rate": 0.0005955371325646154,
      "loss": 2.469,
      "step": 12666
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6507464051246643,
      "learning_rate": 0.0005955364295892078,
      "loss": 2.2271,
      "step": 12667
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.631148636341095,
      "learning_rate": 0.0005955357265588543,
      "loss": 2.6018,
      "step": 12668
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6377298831939697,
      "learning_rate": 0.0005955350234735554,
      "loss": 2.5036,
      "step": 12669
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6446254253387451,
      "learning_rate": 0.0005955343203333107,
      "loss": 2.7651,
      "step": 12670
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.619792640209198,
      "learning_rate": 0.0005955336171381206,
      "loss": 2.5443,
      "step": 12671
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.5695185661315918,
      "learning_rate": 0.0005955329138879854,
      "loss": 2.1472,
      "step": 12672
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.7614805698394775,
      "learning_rate": 0.0005955322105829049,
      "loss": 2.5324,
      "step": 12673
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6134738922119141,
      "learning_rate": 0.0005955315072228794,
      "loss": 2.4273,
      "step": 12674
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6141229867935181,
      "learning_rate": 0.0005955308038079092,
      "loss": 2.2365,
      "step": 12675
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.7501198649406433,
      "learning_rate": 0.000595530100337994,
      "loss": 2.5077,
      "step": 12676
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6207150220870972,
      "learning_rate": 0.0005955293968131343,
      "loss": 2.1881,
      "step": 12677
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6147658824920654,
      "learning_rate": 0.00059552869323333,
      "loss": 2.5291,
      "step": 12678
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6024813652038574,
      "learning_rate": 0.0005955279895985815,
      "loss": 2.4,
      "step": 12679
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.7819308042526245,
      "learning_rate": 0.0005955272859088887,
      "loss": 2.2201,
      "step": 12680
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6522967219352722,
      "learning_rate": 0.0005955265821642518,
      "loss": 2.5192,
      "step": 12681
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6014478802680969,
      "learning_rate": 0.0005955258783646709,
      "loss": 2.3643,
      "step": 12682
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.7097195386886597,
      "learning_rate": 0.0005955251745101461,
      "loss": 2.4454,
      "step": 12683
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6981124877929688,
      "learning_rate": 0.0005955244706006777,
      "loss": 2.5005,
      "step": 12684
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6165120005607605,
      "learning_rate": 0.0005955237666362657,
      "loss": 2.4422,
      "step": 12685
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6959424614906311,
      "learning_rate": 0.0005955230626169103,
      "loss": 2.241,
      "step": 12686
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.7813331484794617,
      "learning_rate": 0.0005955223585426115,
      "loss": 2.501,
      "step": 12687
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.8049210906028748,
      "learning_rate": 0.0005955216544133695,
      "loss": 2.4662,
      "step": 12688
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.7526063919067383,
      "learning_rate": 0.0005955209502291845,
      "loss": 2.4851,
      "step": 12689
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.9022781252861023,
      "learning_rate": 0.0005955202459900565,
      "loss": 2.4101,
      "step": 12690
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.5879133343696594,
      "learning_rate": 0.0005955195416959858,
      "loss": 2.4996,
      "step": 12691
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.7361893057823181,
      "learning_rate": 0.0005955188373469724,
      "loss": 2.3971,
      "step": 12692
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6316893696784973,
      "learning_rate": 0.0005955181329430165,
      "loss": 2.4236,
      "step": 12693
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6649508476257324,
      "learning_rate": 0.0005955174284841182,
      "loss": 2.332,
      "step": 12694
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6589847207069397,
      "learning_rate": 0.0005955167239702776,
      "loss": 2.6304,
      "step": 12695
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.7304743528366089,
      "learning_rate": 0.000595516019401495,
      "loss": 2.5532,
      "step": 12696
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6626830697059631,
      "learning_rate": 0.0005955153147777702,
      "loss": 2.1788,
      "step": 12697
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.5729318857192993,
      "learning_rate": 0.0005955146100991036,
      "loss": 2.552,
      "step": 12698
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.5983331203460693,
      "learning_rate": 0.0005955139053654952,
      "loss": 2.5231,
      "step": 12699
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6584053635597229,
      "learning_rate": 0.0005955132005769453,
      "loss": 2.441,
      "step": 12700
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.5577085614204407,
      "learning_rate": 0.0005955124957334538,
      "loss": 2.5339,
      "step": 12701
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6520765423774719,
      "learning_rate": 0.000595511790835021,
      "loss": 2.4189,
      "step": 12702
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6133004426956177,
      "learning_rate": 0.000595511085881647,
      "loss": 2.3803,
      "step": 12703
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6464288234710693,
      "learning_rate": 0.000595510380873332,
      "loss": 2.1314,
      "step": 12704
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.7553089261054993,
      "learning_rate": 0.0005955096758100759,
      "loss": 2.4419,
      "step": 12705
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6871857643127441,
      "learning_rate": 0.000595508970691879,
      "loss": 2.2882,
      "step": 12706
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6288037896156311,
      "learning_rate": 0.0005955082655187413,
      "loss": 2.5132,
      "step": 12707
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.684392511844635,
      "learning_rate": 0.0005955075602906633,
      "loss": 2.2336,
      "step": 12708
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6850599646568298,
      "learning_rate": 0.0005955068550076446,
      "loss": 2.4964,
      "step": 12709
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6912832260131836,
      "learning_rate": 0.0005955061496696858,
      "loss": 2.2323,
      "step": 12710
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6521572470664978,
      "learning_rate": 0.0005955054442767867,
      "loss": 2.6611,
      "step": 12711
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.7541236877441406,
      "learning_rate": 0.0005955047388289475,
      "loss": 2.2635,
      "step": 12712
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.8258076906204224,
      "learning_rate": 0.0005955040333261686,
      "loss": 2.2432,
      "step": 12713
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6364219784736633,
      "learning_rate": 0.0005955033277684498,
      "loss": 2.4288,
      "step": 12714
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.7418485879898071,
      "learning_rate": 0.0005955026221557913,
      "loss": 2.4943,
      "step": 12715
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.8512399196624756,
      "learning_rate": 0.0005955019164881934,
      "loss": 2.5637,
      "step": 12716
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6626085638999939,
      "learning_rate": 0.000595501210765656,
      "loss": 2.4461,
      "step": 12717
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6942647695541382,
      "learning_rate": 0.0005955005049881795,
      "loss": 2.2977,
      "step": 12718
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.8094398975372314,
      "learning_rate": 0.0005954997991557638,
      "loss": 2.6923,
      "step": 12719
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.7964133024215698,
      "learning_rate": 0.0005954990932684091,
      "loss": 2.8697,
      "step": 12720
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.5732296109199524,
      "learning_rate": 0.0005954983873261156,
      "loss": 2.2122,
      "step": 12721
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.8690279722213745,
      "learning_rate": 0.0005954976813288832,
      "loss": 2.3512,
      "step": 12722
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.7711700797080994,
      "learning_rate": 0.0005954969752767123,
      "loss": 2.3128,
      "step": 12723
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6937940716743469,
      "learning_rate": 0.000595496269169603,
      "loss": 2.6169,
      "step": 12724
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.7227501273155212,
      "learning_rate": 0.0005954955630075554,
      "loss": 2.5211,
      "step": 12725
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.5904147028923035,
      "learning_rate": 0.0005954948567905694,
      "loss": 2.6395,
      "step": 12726
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6045660972595215,
      "learning_rate": 0.0005954941505186455,
      "loss": 2.3858,
      "step": 12727
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.5800291895866394,
      "learning_rate": 0.0005954934441917836,
      "loss": 2.1704,
      "step": 12728
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6065509915351868,
      "learning_rate": 0.0005954927378099839,
      "loss": 2.4722,
      "step": 12729
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.5903403162956238,
      "learning_rate": 0.0005954920313732465,
      "loss": 2.3726,
      "step": 12730
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6147186756134033,
      "learning_rate": 0.0005954913248815716,
      "loss": 2.2851,
      "step": 12731
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6570031642913818,
      "learning_rate": 0.0005954906183349592,
      "loss": 2.4425,
      "step": 12732
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6125010251998901,
      "learning_rate": 0.0005954899117334096,
      "loss": 2.5736,
      "step": 12733
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6333361268043518,
      "learning_rate": 0.0005954892050769229,
      "loss": 2.5536,
      "step": 12734
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.5480200052261353,
      "learning_rate": 0.0005954884983654991,
      "loss": 2.437,
      "step": 12735
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.7074161171913147,
      "learning_rate": 0.0005954877915991384,
      "loss": 2.5355,
      "step": 12736
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.5632653832435608,
      "learning_rate": 0.0005954870847778409,
      "loss": 2.2929,
      "step": 12737
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6079148650169373,
      "learning_rate": 0.0005954863779016068,
      "loss": 2.5579,
      "step": 12738
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.582133412361145,
      "learning_rate": 0.0005954856709704362,
      "loss": 2.3654,
      "step": 12739
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6585184335708618,
      "learning_rate": 0.0005954849639843293,
      "loss": 2.5846,
      "step": 12740
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6280367374420166,
      "learning_rate": 0.0005954842569432862,
      "loss": 2.3746,
      "step": 12741
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6426148414611816,
      "learning_rate": 0.0005954835498473068,
      "loss": 2.446,
      "step": 12742
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.60460364818573,
      "learning_rate": 0.0005954828426963917,
      "loss": 2.2342,
      "step": 12743
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6381863951683044,
      "learning_rate": 0.0005954821354905406,
      "loss": 2.5101,
      "step": 12744
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6547441482543945,
      "learning_rate": 0.0005954814282297538,
      "loss": 2.4478,
      "step": 12745
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.56455397605896,
      "learning_rate": 0.0005954807209140315,
      "loss": 2.4786,
      "step": 12746
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.582141101360321,
      "learning_rate": 0.0005954800135433736,
      "loss": 2.4197,
      "step": 12747
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6012505888938904,
      "learning_rate": 0.0005954793061177805,
      "loss": 2.4779,
      "step": 12748
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.553422749042511,
      "learning_rate": 0.0005954785986372522,
      "loss": 2.3765,
      "step": 12749
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6282353401184082,
      "learning_rate": 0.0005954778911017889,
      "loss": 2.4425,
      "step": 12750
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.583138644695282,
      "learning_rate": 0.0005954771835113906,
      "loss": 2.4579,
      "step": 12751
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6220523715019226,
      "learning_rate": 0.0005954764758660576,
      "loss": 2.2718,
      "step": 12752
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6204253435134888,
      "learning_rate": 0.0005954757681657898,
      "loss": 2.5676,
      "step": 12753
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6174138784408569,
      "learning_rate": 0.0005954750604105876,
      "loss": 2.7684,
      "step": 12754
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6183657646179199,
      "learning_rate": 0.0005954743526004509,
      "loss": 2.4792,
      "step": 12755
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.5714380145072937,
      "learning_rate": 0.0005954736447353801,
      "loss": 2.5173,
      "step": 12756
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.5988909602165222,
      "learning_rate": 0.0005954729368153752,
      "loss": 2.6019,
      "step": 12757
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6178581118583679,
      "learning_rate": 0.0005954722288404362,
      "loss": 2.3565,
      "step": 12758
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6797946691513062,
      "learning_rate": 0.0005954715208105633,
      "loss": 2.5311,
      "step": 12759
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6595849990844727,
      "learning_rate": 0.0005954708127257567,
      "loss": 2.4257,
      "step": 12760
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.5981230139732361,
      "learning_rate": 0.0005954701045860165,
      "loss": 2.5347,
      "step": 12761
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6058387160301208,
      "learning_rate": 0.0005954693963913428,
      "loss": 2.5602,
      "step": 12762
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6625917553901672,
      "learning_rate": 0.0005954686881417358,
      "loss": 2.466,
      "step": 12763
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.643890917301178,
      "learning_rate": 0.0005954679798371956,
      "loss": 2.3371,
      "step": 12764
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6940909624099731,
      "learning_rate": 0.0005954672714777223,
      "loss": 2.3175,
      "step": 12765
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6186918020248413,
      "learning_rate": 0.000595466563063316,
      "loss": 2.3608,
      "step": 12766
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6326895952224731,
      "learning_rate": 0.000595465854593977,
      "loss": 2.2859,
      "step": 12767
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.738594651222229,
      "learning_rate": 0.0005954651460697053,
      "loss": 2.4056,
      "step": 12768
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.640361487865448,
      "learning_rate": 0.0005954644374905009,
      "loss": 2.64,
      "step": 12769
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.656421422958374,
      "learning_rate": 0.0005954637288563642,
      "loss": 2.2635,
      "step": 12770
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6045520901679993,
      "learning_rate": 0.0005954630201672954,
      "loss": 2.4521,
      "step": 12771
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6197421550750732,
      "learning_rate": 0.0005954623114232942,
      "loss": 2.3967,
      "step": 12772
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.5813832879066467,
      "learning_rate": 0.000595461602624361,
      "loss": 2.5974,
      "step": 12773
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.5809935331344604,
      "learning_rate": 0.0005954608937704961,
      "loss": 2.4628,
      "step": 12774
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6582489609718323,
      "learning_rate": 0.0005954601848616992,
      "loss": 2.3534,
      "step": 12775
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6584590077400208,
      "learning_rate": 0.0005954594758979708,
      "loss": 2.4884,
      "step": 12776
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.7826147675514221,
      "learning_rate": 0.0005954587668793109,
      "loss": 2.4476,
      "step": 12777
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6304947733879089,
      "learning_rate": 0.0005954580578057196,
      "loss": 2.4975,
      "step": 12778
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.8038548231124878,
      "learning_rate": 0.0005954573486771971,
      "loss": 2.3429,
      "step": 12779
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6666539311408997,
      "learning_rate": 0.0005954566394937435,
      "loss": 2.3982,
      "step": 12780
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6388494372367859,
      "learning_rate": 0.0005954559302553589,
      "loss": 2.3973,
      "step": 12781
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6139105558395386,
      "learning_rate": 0.0005954552209620435,
      "loss": 2.3141,
      "step": 12782
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6264169812202454,
      "learning_rate": 0.0005954545116137975,
      "loss": 2.2899,
      "step": 12783
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6854623556137085,
      "learning_rate": 0.0005954538022106208,
      "loss": 2.4784,
      "step": 12784
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.5846042633056641,
      "learning_rate": 0.0005954530927525137,
      "loss": 2.2798,
      "step": 12785
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.7065688967704773,
      "learning_rate": 0.0005954523832394762,
      "loss": 2.5388,
      "step": 12786
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6181793808937073,
      "learning_rate": 0.0005954516736715086,
      "loss": 2.3266,
      "step": 12787
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6202096939086914,
      "learning_rate": 0.000595450964048611,
      "loss": 2.4573,
      "step": 12788
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6057283878326416,
      "learning_rate": 0.0005954502543707834,
      "loss": 2.5953,
      "step": 12789
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.602448046207428,
      "learning_rate": 0.0005954495446380262,
      "loss": 2.4396,
      "step": 12790
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6864514946937561,
      "learning_rate": 0.0005954488348503391,
      "loss": 2.7725,
      "step": 12791
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.624116063117981,
      "learning_rate": 0.0005954481250077226,
      "loss": 2.4011,
      "step": 12792
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.8344675302505493,
      "learning_rate": 0.0005954474151101767,
      "loss": 2.565,
      "step": 12793
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6263577938079834,
      "learning_rate": 0.0005954467051577017,
      "loss": 2.2999,
      "step": 12794
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6124509572982788,
      "learning_rate": 0.0005954459951502973,
      "loss": 2.4614,
      "step": 12795
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6716457605361938,
      "learning_rate": 0.0005954452850879641,
      "loss": 2.3155,
      "step": 12796
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6008827686309814,
      "learning_rate": 0.0005954445749707021,
      "loss": 2.4015,
      "step": 12797
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.5920440554618835,
      "learning_rate": 0.0005954438647985112,
      "loss": 2.4558,
      "step": 12798
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6440578699111938,
      "learning_rate": 0.0005954431545713918,
      "loss": 2.0685,
      "step": 12799
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6381173133850098,
      "learning_rate": 0.0005954424442893439,
      "loss": 2.561,
      "step": 12800
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6168218851089478,
      "learning_rate": 0.0005954417339523678,
      "loss": 2.4436,
      "step": 12801
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6420425176620483,
      "learning_rate": 0.0005954410235604634,
      "loss": 2.471,
      "step": 12802
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6465246677398682,
      "learning_rate": 0.000595440313113631,
      "loss": 2.4157,
      "step": 12803
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6053411960601807,
      "learning_rate": 0.0005954396026118705,
      "loss": 2.3295,
      "step": 12804
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6255506873130798,
      "learning_rate": 0.0005954388920551823,
      "loss": 2.4423,
      "step": 12805
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6361347436904907,
      "learning_rate": 0.0005954381814435664,
      "loss": 2.4219,
      "step": 12806
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.5897390842437744,
      "learning_rate": 0.0005954374707770231,
      "loss": 2.4081,
      "step": 12807
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6402094960212708,
      "learning_rate": 0.0005954367600555523,
      "loss": 2.5093,
      "step": 12808
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.62252277135849,
      "learning_rate": 0.0005954360492791543,
      "loss": 2.5205,
      "step": 12809
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6299590468406677,
      "learning_rate": 0.0005954353384478291,
      "loss": 2.6822,
      "step": 12810
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6345799565315247,
      "learning_rate": 0.0005954346275615769,
      "loss": 2.5391,
      "step": 12811
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6195477247238159,
      "learning_rate": 0.0005954339166203978,
      "loss": 2.3698,
      "step": 12812
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6773567199707031,
      "learning_rate": 0.000595433205624292,
      "loss": 2.5142,
      "step": 12813
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6070290207862854,
      "learning_rate": 0.0005954324945732595,
      "loss": 2.1785,
      "step": 12814
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6552549004554749,
      "learning_rate": 0.0005954317834673006,
      "loss": 2.3658,
      "step": 12815
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.911129355430603,
      "learning_rate": 0.0005954310723064154,
      "loss": 2.4683,
      "step": 12816
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.643456220626831,
      "learning_rate": 0.0005954303610906039,
      "loss": 2.4961,
      "step": 12817
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.5715998411178589,
      "learning_rate": 0.0005954296498198664,
      "loss": 2.2673,
      "step": 12818
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6137990355491638,
      "learning_rate": 0.0005954289384942029,
      "loss": 2.4545,
      "step": 12819
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.7721591591835022,
      "learning_rate": 0.0005954282271136136,
      "loss": 2.3756,
      "step": 12820
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6125487089157104,
      "learning_rate": 0.0005954275156780986,
      "loss": 2.56,
      "step": 12821
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6285493969917297,
      "learning_rate": 0.000595426804187658,
      "loss": 2.3839,
      "step": 12822
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.5687277317047119,
      "learning_rate": 0.0005954260926422921,
      "loss": 2.2999,
      "step": 12823
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6046808362007141,
      "learning_rate": 0.0005954253810420008,
      "loss": 2.2578,
      "step": 12824
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.536838173866272,
      "learning_rate": 0.0005954246693867844,
      "loss": 2.2559,
      "step": 12825
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.5970261096954346,
      "learning_rate": 0.000595423957676643,
      "loss": 2.235,
      "step": 12826
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6255545020103455,
      "learning_rate": 0.0005954232459115767,
      "loss": 2.5129,
      "step": 12827
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6747703552246094,
      "learning_rate": 0.0005954225340915856,
      "loss": 2.3659,
      "step": 12828
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6132392287254333,
      "learning_rate": 0.0005954218222166699,
      "loss": 2.3822,
      "step": 12829
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6065260171890259,
      "learning_rate": 0.0005954211102868297,
      "loss": 2.5565,
      "step": 12830
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.665471613407135,
      "learning_rate": 0.0005954203983020651,
      "loss": 2.3384,
      "step": 12831
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6320359706878662,
      "learning_rate": 0.0005954196862623763,
      "loss": 2.416,
      "step": 12832
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.5869975090026855,
      "learning_rate": 0.0005954189741677634,
      "loss": 2.2915,
      "step": 12833
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.5939508080482483,
      "learning_rate": 0.0005954182620182266,
      "loss": 2.1324,
      "step": 12834
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.5912168622016907,
      "learning_rate": 0.0005954175498137659,
      "loss": 2.1719,
      "step": 12835
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.5395904779434204,
      "learning_rate": 0.0005954168375543815,
      "loss": 2.3793,
      "step": 12836
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6937162280082703,
      "learning_rate": 0.0005954161252400736,
      "loss": 2.3886,
      "step": 12837
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6706662178039551,
      "learning_rate": 0.0005954154128708422,
      "loss": 2.6538,
      "step": 12838
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6545775532722473,
      "learning_rate": 0.0005954147004466876,
      "loss": 2.4688,
      "step": 12839
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.612347424030304,
      "learning_rate": 0.0005954139879676098,
      "loss": 2.2933,
      "step": 12840
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6038225293159485,
      "learning_rate": 0.0005954132754336089,
      "loss": 2.5525,
      "step": 12841
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6096563935279846,
      "learning_rate": 0.000595412562844685,
      "loss": 2.46,
      "step": 12842
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6466810703277588,
      "learning_rate": 0.0005954118502008386,
      "loss": 2.5484,
      "step": 12843
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.8488520383834839,
      "learning_rate": 0.0005954111375020693,
      "loss": 2.4797,
      "step": 12844
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.5898380875587463,
      "learning_rate": 0.0005954104247483775,
      "loss": 2.4614,
      "step": 12845
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.62513667345047,
      "learning_rate": 0.0005954097119397635,
      "loss": 2.4855,
      "step": 12846
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6059803366661072,
      "learning_rate": 0.0005954089990762272,
      "loss": 2.2117,
      "step": 12847
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.669861376285553,
      "learning_rate": 0.0005954082861577688,
      "loss": 2.5752,
      "step": 12848
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6090192198753357,
      "learning_rate": 0.0005954075731843884,
      "loss": 2.4063,
      "step": 12849
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6050126552581787,
      "learning_rate": 0.000595406860156086,
      "loss": 2.2162,
      "step": 12850
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6661521196365356,
      "learning_rate": 0.0005954061470728621,
      "loss": 2.6662,
      "step": 12851
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.5947993397712708,
      "learning_rate": 0.0005954054339347166,
      "loss": 2.4683,
      "step": 12852
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6267414689064026,
      "learning_rate": 0.0005954047207416495,
      "loss": 2.5899,
      "step": 12853
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6463542580604553,
      "learning_rate": 0.0005954040074936612,
      "loss": 2.4482,
      "step": 12854
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6400993466377258,
      "learning_rate": 0.0005954032941907518,
      "loss": 2.4153,
      "step": 12855
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6202576756477356,
      "learning_rate": 0.0005954025808329213,
      "loss": 2.2914,
      "step": 12856
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6257836222648621,
      "learning_rate": 0.0005954018674201698,
      "loss": 2.5595,
      "step": 12857
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6315696239471436,
      "learning_rate": 0.0005954011539524976,
      "loss": 2.4647,
      "step": 12858
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.5802702903747559,
      "learning_rate": 0.0005954004404299047,
      "loss": 2.4482,
      "step": 12859
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6834787726402283,
      "learning_rate": 0.0005953997268523912,
      "loss": 2.3444,
      "step": 12860
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.5705727934837341,
      "learning_rate": 0.0005953990132199575,
      "loss": 2.3739,
      "step": 12861
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.5984106063842773,
      "learning_rate": 0.0005953982995326034,
      "loss": 2.1788,
      "step": 12862
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6423070430755615,
      "learning_rate": 0.0005953975857903293,
      "loss": 2.3015,
      "step": 12863
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.5898205041885376,
      "learning_rate": 0.0005953968719931352,
      "loss": 2.6002,
      "step": 12864
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.5796158313751221,
      "learning_rate": 0.0005953961581410212,
      "loss": 2.5763,
      "step": 12865
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.608872652053833,
      "learning_rate": 0.0005953954442339875,
      "loss": 2.2568,
      "step": 12866
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6090657711029053,
      "learning_rate": 0.0005953947302720341,
      "loss": 2.269,
      "step": 12867
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6139117479324341,
      "learning_rate": 0.0005953940162551613,
      "loss": 2.571,
      "step": 12868
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6468753814697266,
      "learning_rate": 0.0005953933021833693,
      "loss": 2.617,
      "step": 12869
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6310786008834839,
      "learning_rate": 0.0005953925880566581,
      "loss": 2.5928,
      "step": 12870
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6351000070571899,
      "learning_rate": 0.0005953918738750277,
      "loss": 2.424,
      "step": 12871
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6461789608001709,
      "learning_rate": 0.0005953911596384785,
      "loss": 2.1458,
      "step": 12872
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6096922755241394,
      "learning_rate": 0.0005953904453470104,
      "loss": 2.5913,
      "step": 12873
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6759588122367859,
      "learning_rate": 0.0005953897310006236,
      "loss": 2.5468,
      "step": 12874
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6462697982788086,
      "learning_rate": 0.0005953890165993184,
      "loss": 2.4627,
      "step": 12875
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6936183571815491,
      "learning_rate": 0.0005953883021430947,
      "loss": 2.0679,
      "step": 12876
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.5969958901405334,
      "learning_rate": 0.0005953875876319528,
      "loss": 2.3178,
      "step": 12877
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.7747374176979065,
      "learning_rate": 0.0005953868730658929,
      "loss": 2.3365,
      "step": 12878
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6242204308509827,
      "learning_rate": 0.0005953861584449148,
      "loss": 2.1895,
      "step": 12879
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6337366104125977,
      "learning_rate": 0.0005953854437690189,
      "loss": 2.4894,
      "step": 12880
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.5973371863365173,
      "learning_rate": 0.0005953847290382053,
      "loss": 2.6061,
      "step": 12881
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.8171886205673218,
      "learning_rate": 0.0005953840142524741,
      "loss": 2.4765,
      "step": 12882
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6054530143737793,
      "learning_rate": 0.0005953832994118254,
      "loss": 2.3456,
      "step": 12883
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6212767362594604,
      "learning_rate": 0.0005953825845162594,
      "loss": 2.3914,
      "step": 12884
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.7090849876403809,
      "learning_rate": 0.0005953818695657762,
      "loss": 2.5598,
      "step": 12885
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.7246648073196411,
      "learning_rate": 0.0005953811545603759,
      "loss": 2.4563,
      "step": 12886
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6057901382446289,
      "learning_rate": 0.0005953804395000587,
      "loss": 2.4995,
      "step": 12887
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.7740185260772705,
      "learning_rate": 0.0005953797243848247,
      "loss": 2.406,
      "step": 12888
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.7599644064903259,
      "learning_rate": 0.000595379009214674,
      "loss": 2.2969,
      "step": 12889
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6620689630508423,
      "learning_rate": 0.0005953782939896069,
      "loss": 2.403,
      "step": 12890
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.7104587554931641,
      "learning_rate": 0.0005953775787096232,
      "loss": 2.3393,
      "step": 12891
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.7731215953826904,
      "learning_rate": 0.0005953768633747234,
      "loss": 2.4867,
      "step": 12892
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.678838849067688,
      "learning_rate": 0.0005953761479849073,
      "loss": 2.3376,
      "step": 12893
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6381223797798157,
      "learning_rate": 0.0005953754325401753,
      "loss": 2.3224,
      "step": 12894
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.7571846842765808,
      "learning_rate": 0.0005953747170405274,
      "loss": 2.3246,
      "step": 12895
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.7391674518585205,
      "learning_rate": 0.0005953740014859639,
      "loss": 2.4425,
      "step": 12896
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6184645295143127,
      "learning_rate": 0.0005953732858764847,
      "loss": 2.3982,
      "step": 12897
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.8417806625366211,
      "learning_rate": 0.0005953725702120899,
      "loss": 2.292,
      "step": 12898
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.7429128289222717,
      "learning_rate": 0.0005953718544927798,
      "loss": 2.384,
      "step": 12899
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.5877998471260071,
      "learning_rate": 0.0005953711387185547,
      "loss": 2.2098,
      "step": 12900
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.7262651920318604,
      "learning_rate": 0.0005953704228894144,
      "loss": 2.6206,
      "step": 12901
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.7040029764175415,
      "learning_rate": 0.0005953697070053592,
      "loss": 2.4902,
      "step": 12902
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6510828137397766,
      "learning_rate": 0.0005953689910663891,
      "loss": 2.4889,
      "step": 12903
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6256200671195984,
      "learning_rate": 0.0005953682750725045,
      "loss": 2.5116,
      "step": 12904
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.7392482161521912,
      "learning_rate": 0.0005953675590237051,
      "loss": 2.4979,
      "step": 12905
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6937255263328552,
      "learning_rate": 0.0005953668429199915,
      "loss": 2.1997,
      "step": 12906
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6571969985961914,
      "learning_rate": 0.0005953661267613636,
      "loss": 2.5245,
      "step": 12907
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6904395222663879,
      "learning_rate": 0.0005953654105478216,
      "loss": 2.402,
      "step": 12908
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.7494750022888184,
      "learning_rate": 0.0005953646942793655,
      "loss": 2.4417,
      "step": 12909
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6519047021865845,
      "learning_rate": 0.0005953639779559956,
      "loss": 2.5349,
      "step": 12910
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.7173169851303101,
      "learning_rate": 0.0005953632615777119,
      "loss": 2.2678,
      "step": 12911
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6034578680992126,
      "learning_rate": 0.0005953625451445146,
      "loss": 2.4709,
      "step": 12912
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6452828645706177,
      "learning_rate": 0.0005953618286564038,
      "loss": 2.6155,
      "step": 12913
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6047242283821106,
      "learning_rate": 0.0005953611121133798,
      "loss": 2.1614,
      "step": 12914
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6898658275604248,
      "learning_rate": 0.0005953603955154424,
      "loss": 2.6269,
      "step": 12915
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6544516682624817,
      "learning_rate": 0.0005953596788625921,
      "loss": 2.5562,
      "step": 12916
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.5910308957099915,
      "learning_rate": 0.0005953589621548288,
      "loss": 2.3774,
      "step": 12917
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.7587538957595825,
      "learning_rate": 0.0005953582453921526,
      "loss": 2.5258,
      "step": 12918
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.604732096195221,
      "learning_rate": 0.0005953575285745639,
      "loss": 2.4895,
      "step": 12919
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.5440731048583984,
      "learning_rate": 0.0005953568117020626,
      "loss": 2.3493,
      "step": 12920
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.7279612421989441,
      "learning_rate": 0.0005953560947746489,
      "loss": 2.381,
      "step": 12921
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6172910928726196,
      "learning_rate": 0.0005953553777923228,
      "loss": 2.4334,
      "step": 12922
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.5654440522193909,
      "learning_rate": 0.0005953546607550846,
      "loss": 2.2095,
      "step": 12923
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6263501644134521,
      "learning_rate": 0.0005953539436629344,
      "loss": 2.6195,
      "step": 12924
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6021814942359924,
      "learning_rate": 0.0005953532265158724,
      "loss": 2.5704,
      "step": 12925
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6647794246673584,
      "learning_rate": 0.0005953525093138986,
      "loss": 2.7555,
      "step": 12926
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.5539483428001404,
      "learning_rate": 0.0005953517920570133,
      "loss": 2.6174,
      "step": 12927
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.621846616268158,
      "learning_rate": 0.0005953510747452165,
      "loss": 2.5614,
      "step": 12928
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6334342956542969,
      "learning_rate": 0.0005953503573785082,
      "loss": 2.2567,
      "step": 12929
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6198709607124329,
      "learning_rate": 0.0005953496399568888,
      "loss": 2.2596,
      "step": 12930
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.5994240045547485,
      "learning_rate": 0.0005953489224803584,
      "loss": 2.3127,
      "step": 12931
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.626776933670044,
      "learning_rate": 0.000595348204948917,
      "loss": 2.2176,
      "step": 12932
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6554755568504333,
      "learning_rate": 0.0005953474873625648,
      "loss": 2.2371,
      "step": 12933
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.653424859046936,
      "learning_rate": 0.0005953467697213018,
      "loss": 2.5315,
      "step": 12934
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6116036176681519,
      "learning_rate": 0.0005953460520251284,
      "loss": 2.5041,
      "step": 12935
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6431429982185364,
      "learning_rate": 0.0005953453342740446,
      "loss": 2.3579,
      "step": 12936
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.5976411700248718,
      "learning_rate": 0.0005953446164680505,
      "loss": 2.581,
      "step": 12937
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.7107464075088501,
      "learning_rate": 0.0005953438986071463,
      "loss": 2.2459,
      "step": 12938
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6452625393867493,
      "learning_rate": 0.0005953431806913319,
      "loss": 2.2341,
      "step": 12939
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.5896743535995483,
      "learning_rate": 0.0005953424627206079,
      "loss": 2.4529,
      "step": 12940
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6821721792221069,
      "learning_rate": 0.000595341744694974,
      "loss": 2.5575,
      "step": 12941
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6421210169792175,
      "learning_rate": 0.0005953410266144305,
      "loss": 2.6488,
      "step": 12942
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6778173446655273,
      "learning_rate": 0.0005953403084789776,
      "loss": 2.6211,
      "step": 12943
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6639911532402039,
      "learning_rate": 0.0005953395902886153,
      "loss": 2.4807,
      "step": 12944
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6269101500511169,
      "learning_rate": 0.0005953388720433438,
      "loss": 2.5482,
      "step": 12945
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.5538162589073181,
      "learning_rate": 0.0005953381537431632,
      "loss": 2.2042,
      "step": 12946
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.5695092678070068,
      "learning_rate": 0.0005953374353880737,
      "loss": 2.3976,
      "step": 12947
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6226438879966736,
      "learning_rate": 0.0005953367169780755,
      "loss": 2.4806,
      "step": 12948
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.5858644247055054,
      "learning_rate": 0.0005953359985131685,
      "loss": 2.3851,
      "step": 12949
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.5979216694831848,
      "learning_rate": 0.000595335279993353,
      "loss": 2.4168,
      "step": 12950
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6480662822723389,
      "learning_rate": 0.000595334561418629,
      "loss": 2.6005,
      "step": 12951
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.5750571489334106,
      "learning_rate": 0.0005953338427889969,
      "loss": 2.6773,
      "step": 12952
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6353687047958374,
      "learning_rate": 0.0005953331241044566,
      "loss": 2.3539,
      "step": 12953
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.5806393623352051,
      "learning_rate": 0.0005953324053650082,
      "loss": 2.6242,
      "step": 12954
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.5878538489341736,
      "learning_rate": 0.000595331686570652,
      "loss": 2.4775,
      "step": 12955
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6243459582328796,
      "learning_rate": 0.0005953309677213882,
      "loss": 2.4636,
      "step": 12956
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6529830694198608,
      "learning_rate": 0.0005953302488172166,
      "loss": 2.447,
      "step": 12957
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.5680794715881348,
      "learning_rate": 0.0005953295298581376,
      "loss": 2.4678,
      "step": 12958
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6371210813522339,
      "learning_rate": 0.0005953288108441513,
      "loss": 2.3625,
      "step": 12959
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.5987048745155334,
      "learning_rate": 0.0005953280917752578,
      "loss": 2.5135,
      "step": 12960
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6386908888816833,
      "learning_rate": 0.0005953273726514572,
      "loss": 2.3026,
      "step": 12961
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6115283370018005,
      "learning_rate": 0.0005953266534727497,
      "loss": 2.2093,
      "step": 12962
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6343084573745728,
      "learning_rate": 0.0005953259342391353,
      "loss": 2.2072,
      "step": 12963
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.5844615697860718,
      "learning_rate": 0.0005953252149506143,
      "loss": 2.243,
      "step": 12964
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.7118635773658752,
      "learning_rate": 0.0005953244956071868,
      "loss": 2.432,
      "step": 12965
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6023437976837158,
      "learning_rate": 0.0005953237762088529,
      "loss": 2.5558,
      "step": 12966
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6481278538703918,
      "learning_rate": 0.0005953230567556127,
      "loss": 2.4762,
      "step": 12967
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6858156323432922,
      "learning_rate": 0.0005953223372474664,
      "loss": 2.4266,
      "step": 12968
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6323655247688293,
      "learning_rate": 0.000595321617684414,
      "loss": 2.3079,
      "step": 12969
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.5952260494232178,
      "learning_rate": 0.0005953208980664559,
      "loss": 2.633,
      "step": 12970
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6379806995391846,
      "learning_rate": 0.000595320178393592,
      "loss": 2.4391,
      "step": 12971
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.5979498624801636,
      "learning_rate": 0.0005953194586658224,
      "loss": 2.1916,
      "step": 12972
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.5934239029884338,
      "learning_rate": 0.0005953187388831475,
      "loss": 2.3839,
      "step": 12973
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6494731903076172,
      "learning_rate": 0.0005953180190455673,
      "loss": 2.1944,
      "step": 12974
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.5398604273796082,
      "learning_rate": 0.0005953172991530818,
      "loss": 2.4544,
      "step": 12975
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.7002862691879272,
      "learning_rate": 0.0005953165792056911,
      "loss": 2.5006,
      "step": 12976
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.592038094997406,
      "learning_rate": 0.0005953158592033956,
      "loss": 2.4058,
      "step": 12977
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.610802412033081,
      "learning_rate": 0.0005953151391461954,
      "loss": 2.9591,
      "step": 12978
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6259754300117493,
      "learning_rate": 0.0005953144190340905,
      "loss": 2.5882,
      "step": 12979
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6053544878959656,
      "learning_rate": 0.000595313698867081,
      "loss": 2.3575,
      "step": 12980
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.616722047328949,
      "learning_rate": 0.0005953129786451672,
      "loss": 2.2969,
      "step": 12981
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6101661920547485,
      "learning_rate": 0.0005953122583683491,
      "loss": 2.5007,
      "step": 12982
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6297109127044678,
      "learning_rate": 0.0005953115380366268,
      "loss": 2.4067,
      "step": 12983
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6711931228637695,
      "learning_rate": 0.0005953108176500006,
      "loss": 2.5892,
      "step": 12984
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6067755818367004,
      "learning_rate": 0.0005953100972084706,
      "loss": 2.5281,
      "step": 12985
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.5857492089271545,
      "learning_rate": 0.0005953093767120368,
      "loss": 2.3655,
      "step": 12986
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6553483605384827,
      "learning_rate": 0.0005953086561606995,
      "loss": 2.3878,
      "step": 12987
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6099570989608765,
      "learning_rate": 0.0005953079355544587,
      "loss": 2.5608,
      "step": 12988
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6079362034797668,
      "learning_rate": 0.0005953072148933144,
      "loss": 2.4525,
      "step": 12989
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6433495283126831,
      "learning_rate": 0.0005953064941772671,
      "loss": 2.5876,
      "step": 12990
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.5688823461532593,
      "learning_rate": 0.0005953057734063167,
      "loss": 2.3799,
      "step": 12991
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.623978316783905,
      "learning_rate": 0.0005953050525804634,
      "loss": 2.4668,
      "step": 12992
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.8473644852638245,
      "learning_rate": 0.0005953043316997074,
      "loss": 2.4499,
      "step": 12993
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6323683261871338,
      "learning_rate": 0.0005953036107640486,
      "loss": 2.4655,
      "step": 12994
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.9548338055610657,
      "learning_rate": 0.0005953028897734873,
      "loss": 2.1876,
      "step": 12995
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6019313335418701,
      "learning_rate": 0.0005953021687280237,
      "loss": 2.3533,
      "step": 12996
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6484758853912354,
      "learning_rate": 0.0005953014476276578,
      "loss": 2.5427,
      "step": 12997
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6942462921142578,
      "learning_rate": 0.0005953007264723897,
      "loss": 2.48,
      "step": 12998
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.5899963974952698,
      "learning_rate": 0.0005953000052622199,
      "loss": 2.3264,
      "step": 12999
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.7700066566467285,
      "learning_rate": 0.000595299283997148,
      "loss": 2.5358,
      "step": 13000
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.5947355031967163,
      "learning_rate": 0.0005952985626771745,
      "loss": 2.0333,
      "step": 13001
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6505340933799744,
      "learning_rate": 0.0005952978413022994,
      "loss": 2.3454,
      "step": 13002
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6481537222862244,
      "learning_rate": 0.0005952971198725228,
      "loss": 2.4482,
      "step": 13003
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6468991041183472,
      "learning_rate": 0.000595296398387845,
      "loss": 2.3501,
      "step": 13004
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6231803297996521,
      "learning_rate": 0.0005952956768482659,
      "loss": 2.2854,
      "step": 13005
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6284063458442688,
      "learning_rate": 0.0005952949552537858,
      "loss": 2.3577,
      "step": 13006
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.5967590808868408,
      "learning_rate": 0.0005952942336044049,
      "loss": 2.646,
      "step": 13007
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.7248889803886414,
      "learning_rate": 0.0005952935119001231,
      "loss": 2.3542,
      "step": 13008
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.5750230550765991,
      "learning_rate": 0.0005952927901409407,
      "loss": 2.3709,
      "step": 13009
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.5868270993232727,
      "learning_rate": 0.0005952920683268578,
      "loss": 2.6907,
      "step": 13010
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6885842084884644,
      "learning_rate": 0.0005952913464578745,
      "loss": 2.3538,
      "step": 13011
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6232209801673889,
      "learning_rate": 0.0005952906245339911,
      "loss": 2.4173,
      "step": 13012
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6117938160896301,
      "learning_rate": 0.0005952899025552075,
      "loss": 2.3351,
      "step": 13013
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6115007996559143,
      "learning_rate": 0.0005952891805215239,
      "loss": 2.5145,
      "step": 13014
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.5958430171012878,
      "learning_rate": 0.0005952884584329405,
      "loss": 2.2394,
      "step": 13015
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.7137146592140198,
      "learning_rate": 0.0005952877362894574,
      "loss": 2.2677,
      "step": 13016
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.5800946950912476,
      "learning_rate": 0.0005952870140910747,
      "loss": 2.2257,
      "step": 13017
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.577070415019989,
      "learning_rate": 0.0005952862918377926,
      "loss": 2.3065,
      "step": 13018
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6699832677841187,
      "learning_rate": 0.0005952855695296113,
      "loss": 2.4471,
      "step": 13019
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.600247859954834,
      "learning_rate": 0.0005952848471665307,
      "loss": 2.474,
      "step": 13020
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.7636064887046814,
      "learning_rate": 0.0005952841247485511,
      "loss": 2.3197,
      "step": 13021
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.7176245450973511,
      "learning_rate": 0.0005952834022756726,
      "loss": 2.4199,
      "step": 13022
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6861674189567566,
      "learning_rate": 0.0005952826797478954,
      "loss": 2.373,
      "step": 13023
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6417062878608704,
      "learning_rate": 0.0005952819571652196,
      "loss": 2.4057,
      "step": 13024
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6507015228271484,
      "learning_rate": 0.0005952812345276453,
      "loss": 2.6515,
      "step": 13025
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6813477277755737,
      "learning_rate": 0.0005952805118351726,
      "loss": 2.4299,
      "step": 13026
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6577736735343933,
      "learning_rate": 0.0005952797890878016,
      "loss": 2.6275,
      "step": 13027
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6423627138137817,
      "learning_rate": 0.0005952790662855326,
      "loss": 2.1114,
      "step": 13028
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.5855582356452942,
      "learning_rate": 0.0005952783434283657,
      "loss": 2.5434,
      "step": 13029
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6695955991744995,
      "learning_rate": 0.0005952776205163008,
      "loss": 2.4657,
      "step": 13030
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.5960384011268616,
      "learning_rate": 0.0005952768975493384,
      "loss": 2.2922,
      "step": 13031
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6662245392799377,
      "learning_rate": 0.0005952761745274783,
      "loss": 2.4663,
      "step": 13032
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.7238306999206543,
      "learning_rate": 0.000595275451450721,
      "loss": 2.4366,
      "step": 13033
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6059329509735107,
      "learning_rate": 0.0005952747283190663,
      "loss": 2.4246,
      "step": 13034
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6452153325080872,
      "learning_rate": 0.0005952740051325144,
      "loss": 2.7202,
      "step": 13035
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6355555057525635,
      "learning_rate": 0.0005952732818910656,
      "loss": 2.4864,
      "step": 13036
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.5696261525154114,
      "learning_rate": 0.0005952725585947198,
      "loss": 2.2589,
      "step": 13037
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6424294114112854,
      "learning_rate": 0.0005952718352434773,
      "loss": 2.4388,
      "step": 13038
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.5777132511138916,
      "learning_rate": 0.0005952711118373382,
      "loss": 2.452,
      "step": 13039
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.5567835569381714,
      "learning_rate": 0.0005952703883763028,
      "loss": 2.3061,
      "step": 13040
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6332861185073853,
      "learning_rate": 0.0005952696648603708,
      "loss": 2.2509,
      "step": 13041
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6573231220245361,
      "learning_rate": 0.0005952689412895427,
      "loss": 2.3971,
      "step": 13042
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.581784725189209,
      "learning_rate": 0.0005952682176638185,
      "loss": 2.3347,
      "step": 13043
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.5741614103317261,
      "learning_rate": 0.0005952674939831984,
      "loss": 2.3236,
      "step": 13044
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6092082262039185,
      "learning_rate": 0.0005952667702476825,
      "loss": 2.3167,
      "step": 13045
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6071892976760864,
      "learning_rate": 0.000595266046457271,
      "loss": 2.3791,
      "step": 13046
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6504823565483093,
      "learning_rate": 0.0005952653226119638,
      "loss": 2.6055,
      "step": 13047
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.5643245577812195,
      "learning_rate": 0.0005952645987117613,
      "loss": 2.2748,
      "step": 13048
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6003895998001099,
      "learning_rate": 0.0005952638747566635,
      "loss": 2.4014,
      "step": 13049
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6089752912521362,
      "learning_rate": 0.0005952631507466705,
      "loss": 2.3446,
      "step": 13050
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6531996130943298,
      "learning_rate": 0.0005952624266817827,
      "loss": 2.6366,
      "step": 13051
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6086733341217041,
      "learning_rate": 0.000595261702562,
      "loss": 2.4587,
      "step": 13052
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6395584344863892,
      "learning_rate": 0.0005952609783873225,
      "loss": 2.2107,
      "step": 13053
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6269132494926453,
      "learning_rate": 0.0005952602541577504,
      "loss": 2.4827,
      "step": 13054
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.5760467648506165,
      "learning_rate": 0.000595259529873284,
      "loss": 2.3688,
      "step": 13055
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6048380136489868,
      "learning_rate": 0.0005952588055339231,
      "loss": 2.4681,
      "step": 13056
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6636499762535095,
      "learning_rate": 0.0005952580811396681,
      "loss": 2.7536,
      "step": 13057
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.5813751220703125,
      "learning_rate": 0.000595257356690519,
      "loss": 2.5668,
      "step": 13058
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6246032118797302,
      "learning_rate": 0.000595256632186476,
      "loss": 2.1161,
      "step": 13059
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6085720658302307,
      "learning_rate": 0.0005952559076275393,
      "loss": 2.5152,
      "step": 13060
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6069469451904297,
      "learning_rate": 0.0005952551830137088,
      "loss": 2.6715,
      "step": 13061
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6285097002983093,
      "learning_rate": 0.0005952544583449849,
      "loss": 2.3641,
      "step": 13062
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.577065646648407,
      "learning_rate": 0.0005952537336213677,
      "loss": 2.3115,
      "step": 13063
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.5920494198799133,
      "learning_rate": 0.000595253008842857,
      "loss": 2.3695,
      "step": 13064
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.7296168804168701,
      "learning_rate": 0.0005952522840094534,
      "loss": 2.4188,
      "step": 13065
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6338276267051697,
      "learning_rate": 0.0005952515591211568,
      "loss": 2.3725,
      "step": 13066
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6132553219795227,
      "learning_rate": 0.0005952508341779673,
      "loss": 2.4054,
      "step": 13067
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.5930180549621582,
      "learning_rate": 0.0005952501091798852,
      "loss": 2.4521,
      "step": 13068
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6343861222267151,
      "learning_rate": 0.0005952493841269105,
      "loss": 2.3504,
      "step": 13069
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.7024634480476379,
      "learning_rate": 0.0005952486590190432,
      "loss": 2.3645,
      "step": 13070
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6220941543579102,
      "learning_rate": 0.0005952479338562838,
      "loss": 2.3932,
      "step": 13071
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.7526718974113464,
      "learning_rate": 0.0005952472086386321,
      "loss": 2.7959,
      "step": 13072
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6018333435058594,
      "learning_rate": 0.0005952464833660885,
      "loss": 2.3119,
      "step": 13073
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.5933240652084351,
      "learning_rate": 0.0005952457580386529,
      "loss": 2.4206,
      "step": 13074
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6581436991691589,
      "learning_rate": 0.0005952450326563257,
      "loss": 2.4649,
      "step": 13075
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6472573280334473,
      "learning_rate": 0.0005952443072191067,
      "loss": 2.2168,
      "step": 13076
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.5601869821548462,
      "learning_rate": 0.0005952435817269962,
      "loss": 2.4025,
      "step": 13077
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6532735824584961,
      "learning_rate": 0.0005952428561799945,
      "loss": 2.4305,
      "step": 13078
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.5827743411064148,
      "learning_rate": 0.0005952421305781014,
      "loss": 2.3674,
      "step": 13079
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.8479958772659302,
      "learning_rate": 0.0005952414049213173,
      "loss": 2.4555,
      "step": 13080
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6151198744773865,
      "learning_rate": 0.0005952406792096422,
      "loss": 2.4255,
      "step": 13081
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6858637928962708,
      "learning_rate": 0.0005952399534430765,
      "loss": 2.2556,
      "step": 13082
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.659364640712738,
      "learning_rate": 0.0005952392276216199,
      "loss": 2.4922,
      "step": 13083
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.7790805697441101,
      "learning_rate": 0.0005952385017452728,
      "loss": 2.4868,
      "step": 13084
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6600651144981384,
      "learning_rate": 0.0005952377758140353,
      "loss": 2.3587,
      "step": 13085
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6930665373802185,
      "learning_rate": 0.0005952370498279075,
      "loss": 2.2498,
      "step": 13086
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6946220993995667,
      "learning_rate": 0.0005952363237868896,
      "loss": 2.3826,
      "step": 13087
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6114068627357483,
      "learning_rate": 0.0005952355976909817,
      "loss": 2.4435,
      "step": 13088
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6212752461433411,
      "learning_rate": 0.0005952348715401839,
      "loss": 2.375,
      "step": 13089
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6771752238273621,
      "learning_rate": 0.0005952341453344964,
      "loss": 2.254,
      "step": 13090
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.5625040531158447,
      "learning_rate": 0.0005952334190739193,
      "loss": 2.5424,
      "step": 13091
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6331538558006287,
      "learning_rate": 0.0005952326927584527,
      "loss": 2.4369,
      "step": 13092
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.5959020853042603,
      "learning_rate": 0.0005952319663880967,
      "loss": 2.3698,
      "step": 13093
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6941200494766235,
      "learning_rate": 0.0005952312399628516,
      "loss": 2.4362,
      "step": 13094
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6148723363876343,
      "learning_rate": 0.0005952305134827174,
      "loss": 2.2728,
      "step": 13095
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6639029383659363,
      "learning_rate": 0.0005952297869476944,
      "loss": 2.2824,
      "step": 13096
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6231313943862915,
      "learning_rate": 0.0005952290603577825,
      "loss": 2.3294,
      "step": 13097
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6673132181167603,
      "learning_rate": 0.0005952283337129819,
      "loss": 2.3929,
      "step": 13098
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6065090298652649,
      "learning_rate": 0.0005952276070132929,
      "loss": 2.3028,
      "step": 13099
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.5832709074020386,
      "learning_rate": 0.0005952268802587153,
      "loss": 2.1381,
      "step": 13100
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.580468475818634,
      "learning_rate": 0.0005952261534492496,
      "loss": 2.468,
      "step": 13101
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.7110089063644409,
      "learning_rate": 0.0005952254265848959,
      "loss": 2.7355,
      "step": 13102
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.7053078413009644,
      "learning_rate": 0.0005952246996656541,
      "loss": 2.6691,
      "step": 13103
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6233256459236145,
      "learning_rate": 0.0005952239726915245,
      "loss": 2.4251,
      "step": 13104
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6309945583343506,
      "learning_rate": 0.0005952232456625071,
      "loss": 2.5393,
      "step": 13105
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.5851072669029236,
      "learning_rate": 0.0005952225185786021,
      "loss": 2.4064,
      "step": 13106
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.7142336368560791,
      "learning_rate": 0.0005952217914398098,
      "loss": 2.3309,
      "step": 13107
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6890339255332947,
      "learning_rate": 0.0005952210642461302,
      "loss": 2.4056,
      "step": 13108
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.7222290635108948,
      "learning_rate": 0.0005952203369975634,
      "loss": 2.6617,
      "step": 13109
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.608453631401062,
      "learning_rate": 0.0005952196096941095,
      "loss": 2.4448,
      "step": 13110
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.7059262990951538,
      "learning_rate": 0.0005952188823357687,
      "loss": 2.2448,
      "step": 13111
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.5877324342727661,
      "learning_rate": 0.0005952181549225412,
      "loss": 2.5215,
      "step": 13112
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6191747188568115,
      "learning_rate": 0.000595217427454427,
      "loss": 2.3763,
      "step": 13113
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6304024457931519,
      "learning_rate": 0.0005952166999314264,
      "loss": 2.4403,
      "step": 13114
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6569305062294006,
      "learning_rate": 0.0005952159723535393,
      "loss": 2.5178,
      "step": 13115
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.5996124148368835,
      "learning_rate": 0.0005952152447207661,
      "loss": 2.1798,
      "step": 13116
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.7471257448196411,
      "learning_rate": 0.0005952145170331068,
      "loss": 2.402,
      "step": 13117
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.595196008682251,
      "learning_rate": 0.0005952137892905615,
      "loss": 2.2918,
      "step": 13118
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6209195256233215,
      "learning_rate": 0.0005952130614931305,
      "loss": 2.4269,
      "step": 13119
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.5324110388755798,
      "learning_rate": 0.0005952123336408137,
      "loss": 2.2765,
      "step": 13120
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6826107501983643,
      "learning_rate": 0.0005952116057336114,
      "loss": 2.5757,
      "step": 13121
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.7809308171272278,
      "learning_rate": 0.0005952108777715236,
      "loss": 2.5526,
      "step": 13122
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.7305382490158081,
      "learning_rate": 0.0005952101497545507,
      "loss": 2.3495,
      "step": 13123
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.7409481406211853,
      "learning_rate": 0.0005952094216826925,
      "loss": 2.3112,
      "step": 13124
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6011318564414978,
      "learning_rate": 0.0005952086935559494,
      "loss": 2.3283,
      "step": 13125
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.859360933303833,
      "learning_rate": 0.0005952079653743214,
      "loss": 2.1561,
      "step": 13126
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.7330498099327087,
      "learning_rate": 0.0005952072371378087,
      "loss": 2.4341,
      "step": 13127
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.5851783752441406,
      "learning_rate": 0.0005952065088464113,
      "loss": 2.5809,
      "step": 13128
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.791347086429596,
      "learning_rate": 0.0005952057805001295,
      "loss": 2.6709,
      "step": 13129
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6467728018760681,
      "learning_rate": 0.0005952050520989634,
      "loss": 2.315,
      "step": 13130
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6298443675041199,
      "learning_rate": 0.0005952043236429131,
      "loss": 2.4959,
      "step": 13131
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.7289495468139648,
      "learning_rate": 0.0005952035951319787,
      "loss": 2.4298,
      "step": 13132
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.7946979999542236,
      "learning_rate": 0.0005952028665661604,
      "loss": 2.4324,
      "step": 13133
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6029999256134033,
      "learning_rate": 0.0005952021379454583,
      "loss": 2.2287,
      "step": 13134
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6454488039016724,
      "learning_rate": 0.0005952014092698726,
      "loss": 2.3881,
      "step": 13135
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6979177594184875,
      "learning_rate": 0.0005952006805394033,
      "loss": 2.393,
      "step": 13136
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6648120284080505,
      "learning_rate": 0.0005951999517540507,
      "loss": 2.5061,
      "step": 13137
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.617267906665802,
      "learning_rate": 0.0005951992229138148,
      "loss": 2.5971,
      "step": 13138
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6690038442611694,
      "learning_rate": 0.0005951984940186959,
      "loss": 2.4511,
      "step": 13139
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6590774059295654,
      "learning_rate": 0.0005951977650686939,
      "loss": 2.5599,
      "step": 13140
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.5886281728744507,
      "learning_rate": 0.0005951970360638092,
      "loss": 2.3841,
      "step": 13141
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.668887734413147,
      "learning_rate": 0.0005951963070040417,
      "loss": 2.5499,
      "step": 13142
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.7824971675872803,
      "learning_rate": 0.0005951955778893916,
      "loss": 2.2837,
      "step": 13143
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.634904146194458,
      "learning_rate": 0.0005951948487198591,
      "loss": 2.4627,
      "step": 13144
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.583478569984436,
      "learning_rate": 0.0005951941194954444,
      "loss": 2.533,
      "step": 13145
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.8375807404518127,
      "learning_rate": 0.0005951933902161474,
      "loss": 2.6232,
      "step": 13146
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6243330240249634,
      "learning_rate": 0.0005951926608819685,
      "loss": 2.3473,
      "step": 13147
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.7900183200836182,
      "learning_rate": 0.0005951919314929077,
      "loss": 2.3013,
      "step": 13148
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6779983639717102,
      "learning_rate": 0.000595191202048965,
      "loss": 2.5839,
      "step": 13149
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6153779029846191,
      "learning_rate": 0.0005951904725501408,
      "loss": 2.2954,
      "step": 13150
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6112457513809204,
      "learning_rate": 0.0005951897429964351,
      "loss": 2.2238,
      "step": 13151
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6335875391960144,
      "learning_rate": 0.000595189013387848,
      "loss": 2.6426,
      "step": 13152
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.7329973578453064,
      "learning_rate": 0.0005951882837243799,
      "loss": 2.4965,
      "step": 13153
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6298798322677612,
      "learning_rate": 0.0005951875540060305,
      "loss": 2.4161,
      "step": 13154
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6565174460411072,
      "learning_rate": 0.0005951868242328002,
      "loss": 2.3757,
      "step": 13155
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6608954071998596,
      "learning_rate": 0.0005951860944046892,
      "loss": 2.4442,
      "step": 13156
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.5474289059638977,
      "learning_rate": 0.0005951853645216975,
      "loss": 2.1411,
      "step": 13157
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6483538746833801,
      "learning_rate": 0.0005951846345838252,
      "loss": 2.3191,
      "step": 13158
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.721996009349823,
      "learning_rate": 0.0005951839045910725,
      "loss": 2.372,
      "step": 13159
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.5980218052864075,
      "learning_rate": 0.0005951831745434396,
      "loss": 2.1909,
      "step": 13160
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.7109087705612183,
      "learning_rate": 0.0005951824444409265,
      "loss": 2.575,
      "step": 13161
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6112803220748901,
      "learning_rate": 0.0005951817142835335,
      "loss": 2.3708,
      "step": 13162
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.8155049085617065,
      "learning_rate": 0.0005951809840712607,
      "loss": 2.4988,
      "step": 13163
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6967744827270508,
      "learning_rate": 0.0005951802538041081,
      "loss": 2.3532,
      "step": 13164
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6655816435813904,
      "learning_rate": 0.0005951795234820759,
      "loss": 2.272,
      "step": 13165
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.7102547883987427,
      "learning_rate": 0.0005951787931051643,
      "loss": 2.4329,
      "step": 13166
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6471986770629883,
      "learning_rate": 0.0005951780626733732,
      "loss": 2.3715,
      "step": 13167
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6813372373580933,
      "learning_rate": 0.0005951773321867032,
      "loss": 2.4998,
      "step": 13168
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6366866827011108,
      "learning_rate": 0.000595176601645154,
      "loss": 2.3291,
      "step": 13169
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6469152569770813,
      "learning_rate": 0.000595175871048726,
      "loss": 2.2439,
      "step": 13170
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6041995882987976,
      "learning_rate": 0.0005951751403974193,
      "loss": 2.1586,
      "step": 13171
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6330136060714722,
      "learning_rate": 0.0005951744096912337,
      "loss": 2.326,
      "step": 13172
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.7021203637123108,
      "learning_rate": 0.0005951736789301697,
      "loss": 2.2903,
      "step": 13173
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.7049496173858643,
      "learning_rate": 0.0005951729481142275,
      "loss": 2.4292,
      "step": 13174
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.654213011264801,
      "learning_rate": 0.000595172217243407,
      "loss": 2.2701,
      "step": 13175
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.612244188785553,
      "learning_rate": 0.0005951714863177084,
      "loss": 2.4632,
      "step": 13176
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.5834502577781677,
      "learning_rate": 0.0005951707553371318,
      "loss": 2.2784,
      "step": 13177
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6432413458824158,
      "learning_rate": 0.0005951700243016773,
      "loss": 2.5824,
      "step": 13178
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6471774578094482,
      "learning_rate": 0.0005951692932113454,
      "loss": 2.1726,
      "step": 13179
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6829515099525452,
      "learning_rate": 0.0005951685620661358,
      "loss": 2.2425,
      "step": 13180
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.646004319190979,
      "learning_rate": 0.0005951678308660488,
      "loss": 2.5131,
      "step": 13181
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.5479313731193542,
      "learning_rate": 0.0005951670996110844,
      "loss": 2.527,
      "step": 13182
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6602110862731934,
      "learning_rate": 0.0005951663683012431,
      "loss": 2.3636,
      "step": 13183
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.7029314041137695,
      "learning_rate": 0.0005951656369365246,
      "loss": 2.6043,
      "step": 13184
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6538217663764954,
      "learning_rate": 0.0005951649055169294,
      "loss": 2.5703,
      "step": 13185
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6649460196495056,
      "learning_rate": 0.0005951641740424573,
      "loss": 2.4741,
      "step": 13186
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.8873382210731506,
      "learning_rate": 0.0005951634425131087,
      "loss": 2.737,
      "step": 13187
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6775749921798706,
      "learning_rate": 0.0005951627109288837,
      "loss": 2.2638,
      "step": 13188
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6319996118545532,
      "learning_rate": 0.0005951619792897823,
      "loss": 2.6127,
      "step": 13189
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6114501953125,
      "learning_rate": 0.0005951612475958047,
      "loss": 2.4633,
      "step": 13190
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6027089953422546,
      "learning_rate": 0.0005951605158469512,
      "loss": 2.4892,
      "step": 13191
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6762559413909912,
      "learning_rate": 0.0005951597840432217,
      "loss": 2.505,
      "step": 13192
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.5863875150680542,
      "learning_rate": 0.0005951590521846164,
      "loss": 2.2629,
      "step": 13193
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.676261842250824,
      "learning_rate": 0.0005951583202711354,
      "loss": 2.467,
      "step": 13194
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.5749344825744629,
      "learning_rate": 0.0005951575883027789,
      "loss": 2.4862,
      "step": 13195
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6864995360374451,
      "learning_rate": 0.0005951568562795472,
      "loss": 2.4711,
      "step": 13196
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6357167363166809,
      "learning_rate": 0.0005951561242014402,
      "loss": 2.304,
      "step": 13197
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6119638085365295,
      "learning_rate": 0.000595155392068458,
      "loss": 2.6737,
      "step": 13198
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.5929674506187439,
      "learning_rate": 0.000595154659880601,
      "loss": 2.3976,
      "step": 13199
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6114567518234253,
      "learning_rate": 0.000595153927637869,
      "loss": 2.6392,
      "step": 13200
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6090983152389526,
      "learning_rate": 0.0005951531953402624,
      "loss": 2.5018,
      "step": 13201
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6192864775657654,
      "learning_rate": 0.0005951524629877813,
      "loss": 2.4248,
      "step": 13202
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6278165578842163,
      "learning_rate": 0.0005951517305804257,
      "loss": 2.4979,
      "step": 13203
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6848317980766296,
      "learning_rate": 0.0005951509981181959,
      "loss": 2.5631,
      "step": 13204
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6619279384613037,
      "learning_rate": 0.0005951502656010919,
      "loss": 2.4202,
      "step": 13205
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6094606518745422,
      "learning_rate": 0.0005951495330291139,
      "loss": 2.4388,
      "step": 13206
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6113583445549011,
      "learning_rate": 0.000595148800402262,
      "loss": 2.5315,
      "step": 13207
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6042605638504028,
      "learning_rate": 0.0005951480677205365,
      "loss": 2.7548,
      "step": 13208
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.7807658910751343,
      "learning_rate": 0.0005951473349839374,
      "loss": 2.366,
      "step": 13209
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.5860719680786133,
      "learning_rate": 0.0005951466021924647,
      "loss": 2.4706,
      "step": 13210
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6130884885787964,
      "learning_rate": 0.0005951458693461187,
      "loss": 2.1578,
      "step": 13211
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6658837795257568,
      "learning_rate": 0.0005951451364448995,
      "loss": 2.5216,
      "step": 13212
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6104994416236877,
      "learning_rate": 0.0005951444034888074,
      "loss": 2.4408,
      "step": 13213
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6492927670478821,
      "learning_rate": 0.0005951436704778422,
      "loss": 2.3721,
      "step": 13214
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6001760959625244,
      "learning_rate": 0.0005951429374120043,
      "loss": 2.6356,
      "step": 13215
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.5968586206436157,
      "learning_rate": 0.0005951422042912937,
      "loss": 2.2276,
      "step": 13216
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6094486117362976,
      "learning_rate": 0.0005951414711157107,
      "loss": 2.2807,
      "step": 13217
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6226825714111328,
      "learning_rate": 0.0005951407378852553,
      "loss": 2.5042,
      "step": 13218
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.5954722762107849,
      "learning_rate": 0.0005951400045999276,
      "loss": 2.2888,
      "step": 13219
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.5915767550468445,
      "learning_rate": 0.0005951392712597279,
      "loss": 2.2526,
      "step": 13220
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.5850605964660645,
      "learning_rate": 0.0005951385378646561,
      "loss": 2.4009,
      "step": 13221
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.618204653263092,
      "learning_rate": 0.0005951378044147126,
      "loss": 2.3322,
      "step": 13222
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6171244978904724,
      "learning_rate": 0.0005951370709098973,
      "loss": 2.2823,
      "step": 13223
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6559624671936035,
      "learning_rate": 0.0005951363373502106,
      "loss": 2.3834,
      "step": 13224
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6292652487754822,
      "learning_rate": 0.0005951356037356523,
      "loss": 2.6339,
      "step": 13225
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.7031228542327881,
      "learning_rate": 0.0005951348700662228,
      "loss": 2.4928,
      "step": 13226
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6437432169914246,
      "learning_rate": 0.0005951341363419222,
      "loss": 2.5878,
      "step": 13227
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6236011385917664,
      "learning_rate": 0.0005951334025627506,
      "loss": 2.487,
      "step": 13228
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.5951499938964844,
      "learning_rate": 0.000595132668728708,
      "loss": 2.3598,
      "step": 13229
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6241762638092041,
      "learning_rate": 0.0005951319348397948,
      "loss": 2.595,
      "step": 13230
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.65592360496521,
      "learning_rate": 0.000595131200896011,
      "loss": 2.4754,
      "step": 13231
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6106663346290588,
      "learning_rate": 0.0005951304668973566,
      "loss": 2.5794,
      "step": 13232
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6102449893951416,
      "learning_rate": 0.0005951297328438321,
      "loss": 2.3928,
      "step": 13233
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6216627955436707,
      "learning_rate": 0.0005951289987354373,
      "loss": 2.3992,
      "step": 13234
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.8302240371704102,
      "learning_rate": 0.0005951282645721724,
      "loss": 2.4973,
      "step": 13235
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.536693274974823,
      "learning_rate": 0.0005951275303540376,
      "loss": 2.4838,
      "step": 13236
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6374915242195129,
      "learning_rate": 0.0005951267960810331,
      "loss": 2.3089,
      "step": 13237
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.568035900592804,
      "learning_rate": 0.0005951260617531589,
      "loss": 2.1741,
      "step": 13238
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.5999070405960083,
      "learning_rate": 0.0005951253273704151,
      "loss": 2.3982,
      "step": 13239
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.5696361064910889,
      "learning_rate": 0.0005951245929328021,
      "loss": 2.2163,
      "step": 13240
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.747014045715332,
      "learning_rate": 0.0005951238584403197,
      "loss": 2.6528,
      "step": 13241
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6261518597602844,
      "learning_rate": 0.0005951231238929684,
      "loss": 2.591,
      "step": 13242
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.7723913192749023,
      "learning_rate": 0.000595122389290748,
      "loss": 2.6559,
      "step": 13243
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6041038632392883,
      "learning_rate": 0.0005951216546336588,
      "loss": 2.4198,
      "step": 13244
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.5864822864532471,
      "learning_rate": 0.0005951209199217009,
      "loss": 2.4479,
      "step": 13245
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6107797026634216,
      "learning_rate": 0.0005951201851548745,
      "loss": 2.4938,
      "step": 13246
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6165268421173096,
      "learning_rate": 0.0005951194503331797,
      "loss": 2.4577,
      "step": 13247
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6194262504577637,
      "learning_rate": 0.0005951187154566165,
      "loss": 2.4575,
      "step": 13248
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.603821337223053,
      "learning_rate": 0.0005951179805251853,
      "loss": 2.4041,
      "step": 13249
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6724569797515869,
      "learning_rate": 0.0005951172455388861,
      "loss": 2.4078,
      "step": 13250
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.5520743727684021,
      "learning_rate": 0.000595116510497719,
      "loss": 2.5007,
      "step": 13251
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6020578742027283,
      "learning_rate": 0.0005951157754016841,
      "loss": 2.4509,
      "step": 13252
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.7200084328651428,
      "learning_rate": 0.0005951150402507817,
      "loss": 2.3678,
      "step": 13253
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.8056513667106628,
      "learning_rate": 0.0005951143050450117,
      "loss": 2.2469,
      "step": 13254
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6208098530769348,
      "learning_rate": 0.0005951135697843745,
      "loss": 2.4217,
      "step": 13255
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.7671307325363159,
      "learning_rate": 0.00059511283446887,
      "loss": 2.5754,
      "step": 13256
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6318127512931824,
      "learning_rate": 0.0005951120990984987,
      "loss": 2.5155,
      "step": 13257
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.645461916923523,
      "learning_rate": 0.0005951113636732603,
      "loss": 2.3741,
      "step": 13258
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6904052495956421,
      "learning_rate": 0.0005951106281931551,
      "loss": 2.6965,
      "step": 13259
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.7829707264900208,
      "learning_rate": 0.0005951098926581834,
      "loss": 2.3277,
      "step": 13260
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6277196407318115,
      "learning_rate": 0.0005951091570683451,
      "loss": 2.6474,
      "step": 13261
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.654521644115448,
      "learning_rate": 0.0005951084214236405,
      "loss": 2.4863,
      "step": 13262
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6360495686531067,
      "learning_rate": 0.0005951076857240696,
      "loss": 2.3709,
      "step": 13263
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6096669435501099,
      "learning_rate": 0.0005951069499696327,
      "loss": 2.4141,
      "step": 13264
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.5684890151023865,
      "learning_rate": 0.0005951062141603297,
      "loss": 2.4534,
      "step": 13265
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6550427675247192,
      "learning_rate": 0.0005951054782961611,
      "loss": 2.3737,
      "step": 13266
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6744121313095093,
      "learning_rate": 0.0005951047423771266,
      "loss": 2.529,
      "step": 13267
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.5848099589347839,
      "learning_rate": 0.0005951040064032267,
      "loss": 2.3073,
      "step": 13268
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.59791499376297,
      "learning_rate": 0.0005951032703744614,
      "loss": 2.2274,
      "step": 13269
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.627168595790863,
      "learning_rate": 0.0005951025342908308,
      "loss": 2.5034,
      "step": 13270
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.5624632835388184,
      "learning_rate": 0.000595101798152335,
      "loss": 2.3271,
      "step": 13271
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6003212332725525,
      "learning_rate": 0.0005951010619589742,
      "loss": 2.5244,
      "step": 13272
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6324310898780823,
      "learning_rate": 0.0005951003257107486,
      "loss": 2.3766,
      "step": 13273
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6866126656532288,
      "learning_rate": 0.0005950995894076584,
      "loss": 2.5095,
      "step": 13274
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6670867204666138,
      "learning_rate": 0.0005950988530497034,
      "loss": 2.4844,
      "step": 13275
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6277041435241699,
      "learning_rate": 0.0005950981166368841,
      "loss": 2.5672,
      "step": 13276
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6406087875366211,
      "learning_rate": 0.0005950973801692004,
      "loss": 2.4101,
      "step": 13277
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6038140058517456,
      "learning_rate": 0.0005950966436466526,
      "loss": 2.3977,
      "step": 13278
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6146206855773926,
      "learning_rate": 0.0005950959070692407,
      "loss": 2.2602,
      "step": 13279
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6226595044136047,
      "learning_rate": 0.0005950951704369649,
      "loss": 2.3316,
      "step": 13280
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.7440433502197266,
      "learning_rate": 0.0005950944337498254,
      "loss": 2.3947,
      "step": 13281
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6865355968475342,
      "learning_rate": 0.0005950936970078222,
      "loss": 2.5071,
      "step": 13282
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.7172225117683411,
      "learning_rate": 0.0005950929602109555,
      "loss": 2.3871,
      "step": 13283
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.805504560470581,
      "learning_rate": 0.0005950922233592255,
      "loss": 2.4067,
      "step": 13284
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6550947427749634,
      "learning_rate": 0.0005950914864526324,
      "loss": 2.3324,
      "step": 13285
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6021618247032166,
      "learning_rate": 0.0005950907494911761,
      "loss": 2.4049,
      "step": 13286
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.8813460469245911,
      "learning_rate": 0.0005950900124748568,
      "loss": 2.3244,
      "step": 13287
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6356754302978516,
      "learning_rate": 0.0005950892754036748,
      "loss": 2.4209,
      "step": 13288
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6029495000839233,
      "learning_rate": 0.0005950885382776301,
      "loss": 2.4671,
      "step": 13289
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.7142360806465149,
      "learning_rate": 0.0005950878010967229,
      "loss": 2.334,
      "step": 13290
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6613218188285828,
      "learning_rate": 0.0005950870638609533,
      "loss": 2.4644,
      "step": 13291
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.596907377243042,
      "learning_rate": 0.0005950863265703213,
      "loss": 2.2231,
      "step": 13292
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.823872983455658,
      "learning_rate": 0.0005950855892248274,
      "loss": 2.3122,
      "step": 13293
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.7004374265670776,
      "learning_rate": 0.0005950848518244714,
      "loss": 2.3475,
      "step": 13294
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.673490583896637,
      "learning_rate": 0.0005950841143692536,
      "loss": 2.2395,
      "step": 13295
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6156083345413208,
      "learning_rate": 0.000595083376859174,
      "loss": 2.2282,
      "step": 13296
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.7413361072540283,
      "learning_rate": 0.000595082639294233,
      "loss": 2.6859,
      "step": 13297
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.8850548267364502,
      "learning_rate": 0.0005950819016744304,
      "loss": 2.3108,
      "step": 13298
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.5986711978912354,
      "learning_rate": 0.0005950811639997666,
      "loss": 2.5759,
      "step": 13299
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6348157525062561,
      "learning_rate": 0.0005950804262702415,
      "loss": 2.4786,
      "step": 13300
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.962800920009613,
      "learning_rate": 0.0005950796884858557,
      "loss": 2.7443,
      "step": 13301
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6851189732551575,
      "learning_rate": 0.0005950789506466087,
      "loss": 2.5058,
      "step": 13302
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6256011724472046,
      "learning_rate": 0.000595078212752501,
      "loss": 2.4559,
      "step": 13303
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.7969416379928589,
      "learning_rate": 0.0005950774748035328,
      "loss": 2.5132,
      "step": 13304
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.649050772190094,
      "learning_rate": 0.0005950767367997041,
      "loss": 2.5484,
      "step": 13305
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6048488616943359,
      "learning_rate": 0.000595075998741015,
      "loss": 2.2222,
      "step": 13306
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6852624416351318,
      "learning_rate": 0.0005950752606274658,
      "loss": 2.5505,
      "step": 13307
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.7228743433952332,
      "learning_rate": 0.0005950745224590565,
      "loss": 2.314,
      "step": 13308
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6185970306396484,
      "learning_rate": 0.0005950737842357872,
      "loss": 2.2488,
      "step": 13309
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6239425539970398,
      "learning_rate": 0.0005950730459576581,
      "loss": 2.6456,
      "step": 13310
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6968902945518494,
      "learning_rate": 0.0005950723076246695,
      "loss": 2.1494,
      "step": 13311
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6738144159317017,
      "learning_rate": 0.0005950715692368213,
      "loss": 2.5241,
      "step": 13312
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6285556554794312,
      "learning_rate": 0.0005950708307941138,
      "loss": 2.4468,
      "step": 13313
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.658064603805542,
      "learning_rate": 0.000595070092296547,
      "loss": 2.6023,
      "step": 13314
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6863554120063782,
      "learning_rate": 0.0005950693537441211,
      "loss": 2.2097,
      "step": 13315
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.7460922598838806,
      "learning_rate": 0.0005950686151368362,
      "loss": 2.4771,
      "step": 13316
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.5999784469604492,
      "learning_rate": 0.0005950678764746925,
      "loss": 2.4473,
      "step": 13317
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.8335185050964355,
      "learning_rate": 0.0005950671377576901,
      "loss": 2.4199,
      "step": 13318
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.0015054941177368,
      "learning_rate": 0.0005950663989858292,
      "loss": 2.2088,
      "step": 13319
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.7322089076042175,
      "learning_rate": 0.0005950656601591099,
      "loss": 2.6002,
      "step": 13320
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.0165293216705322,
      "learning_rate": 0.0005950649212775322,
      "loss": 2.3393,
      "step": 13321
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.8246908187866211,
      "learning_rate": 0.0005950641823410965,
      "loss": 2.3694,
      "step": 13322
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.5531664490699768,
      "learning_rate": 0.0005950634433498027,
      "loss": 2.173,
      "step": 13323
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.9381155371665955,
      "learning_rate": 0.0005950627043036511,
      "loss": 2.5575,
      "step": 13324
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6515582799911499,
      "learning_rate": 0.0005950619652026418,
      "loss": 2.4764,
      "step": 13325
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6075226664543152,
      "learning_rate": 0.0005950612260467749,
      "loss": 2.3135,
      "step": 13326
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6155640482902527,
      "learning_rate": 0.0005950604868360505,
      "loss": 2.56,
      "step": 13327
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6226473450660706,
      "learning_rate": 0.0005950597475704688,
      "loss": 2.4174,
      "step": 13328
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6394111514091492,
      "learning_rate": 0.0005950590082500299,
      "loss": 2.4279,
      "step": 13329
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.7366637587547302,
      "learning_rate": 0.000595058268874734,
      "loss": 2.2728,
      "step": 13330
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6441391110420227,
      "learning_rate": 0.0005950575294445811,
      "loss": 2.4639,
      "step": 13331
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6828996539115906,
      "learning_rate": 0.0005950567899595716,
      "loss": 2.4631,
      "step": 13332
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.8621404767036438,
      "learning_rate": 0.0005950560504197054,
      "loss": 2.3841,
      "step": 13333
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6230081915855408,
      "learning_rate": 0.0005950553108249827,
      "loss": 2.3916,
      "step": 13334
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.5864787101745605,
      "learning_rate": 0.0005950545711754035,
      "loss": 2.6349,
      "step": 13335
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6145949959754944,
      "learning_rate": 0.0005950538314709684,
      "loss": 2.4536,
      "step": 13336
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6737204790115356,
      "learning_rate": 0.000595053091711677,
      "loss": 2.3799,
      "step": 13337
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6645854711532593,
      "learning_rate": 0.0005950523518975297,
      "loss": 2.5181,
      "step": 13338
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.569649875164032,
      "learning_rate": 0.0005950516120285266,
      "loss": 2.2722,
      "step": 13339
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.5640931725502014,
      "learning_rate": 0.0005950508721046678,
      "loss": 2.4287,
      "step": 13340
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.5768028497695923,
      "learning_rate": 0.0005950501321259536,
      "loss": 2.5703,
      "step": 13341
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6484702825546265,
      "learning_rate": 0.0005950493920923839,
      "loss": 2.6616,
      "step": 13342
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6835558414459229,
      "learning_rate": 0.000595048652003959,
      "loss": 2.5086,
      "step": 13343
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6313767433166504,
      "learning_rate": 0.000595047911860679,
      "loss": 2.4362,
      "step": 13344
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6728746891021729,
      "learning_rate": 0.000595047171662544,
      "loss": 2.5947,
      "step": 13345
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.600971519947052,
      "learning_rate": 0.0005950464314095541,
      "loss": 2.4782,
      "step": 13346
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.5769680142402649,
      "learning_rate": 0.0005950456911017096,
      "loss": 2.5325,
      "step": 13347
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.5635454058647156,
      "learning_rate": 0.0005950449507390105,
      "loss": 2.4354,
      "step": 13348
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.5509067177772522,
      "learning_rate": 0.000595044210321457,
      "loss": 2.4311,
      "step": 13349
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.5746660828590393,
      "learning_rate": 0.0005950434698490492,
      "loss": 2.5548,
      "step": 13350
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6121848821640015,
      "learning_rate": 0.0005950427293217872,
      "loss": 2.4789,
      "step": 13351
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6335697174072266,
      "learning_rate": 0.0005950419887396711,
      "loss": 2.3977,
      "step": 13352
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.585659921169281,
      "learning_rate": 0.0005950412481027013,
      "loss": 2.5624,
      "step": 13353
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.586992621421814,
      "learning_rate": 0.0005950405074108778,
      "loss": 2.3774,
      "step": 13354
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6829656958580017,
      "learning_rate": 0.0005950397666642006,
      "loss": 2.4117,
      "step": 13355
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.5814487934112549,
      "learning_rate": 0.0005950390258626699,
      "loss": 2.3533,
      "step": 13356
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.5905731916427612,
      "learning_rate": 0.0005950382850062858,
      "loss": 2.4373,
      "step": 13357
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6125299334526062,
      "learning_rate": 0.0005950375440950487,
      "loss": 2.4775,
      "step": 13358
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6631931066513062,
      "learning_rate": 0.0005950368031289585,
      "loss": 2.6881,
      "step": 13359
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.5834437608718872,
      "learning_rate": 0.0005950360621080153,
      "loss": 2.3829,
      "step": 13360
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.64898681640625,
      "learning_rate": 0.0005950353210322195,
      "loss": 2.3242,
      "step": 13361
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.651252031326294,
      "learning_rate": 0.0005950345799015709,
      "loss": 2.5169,
      "step": 13362
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6324951648712158,
      "learning_rate": 0.0005950338387160699,
      "loss": 2.317,
      "step": 13363
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6295956373214722,
      "learning_rate": 0.0005950330974757164,
      "loss": 2.4089,
      "step": 13364
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.5993140935897827,
      "learning_rate": 0.0005950323561805109,
      "loss": 2.5385,
      "step": 13365
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6125678420066833,
      "learning_rate": 0.0005950316148304531,
      "loss": 2.439,
      "step": 13366
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6396788358688354,
      "learning_rate": 0.0005950308734255436,
      "loss": 2.4574,
      "step": 13367
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6686750054359436,
      "learning_rate": 0.000595030131965782,
      "loss": 2.3541,
      "step": 13368
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6000226736068726,
      "learning_rate": 0.000595029390451169,
      "loss": 2.2745,
      "step": 13369
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6400159597396851,
      "learning_rate": 0.0005950286488817043,
      "loss": 2.4964,
      "step": 13370
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6136487722396851,
      "learning_rate": 0.0005950279072573883,
      "loss": 2.5312,
      "step": 13371
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.719169020652771,
      "learning_rate": 0.000595027165578221,
      "loss": 2.6179,
      "step": 13372
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6212184429168701,
      "learning_rate": 0.0005950264238442026,
      "loss": 2.5056,
      "step": 13373
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6978772878646851,
      "learning_rate": 0.0005950256820553331,
      "loss": 2.3382,
      "step": 13374
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6594289541244507,
      "learning_rate": 0.0005950249402116129,
      "loss": 2.4322,
      "step": 13375
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6785624623298645,
      "learning_rate": 0.0005950241983130419,
      "loss": 2.4098,
      "step": 13376
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6850975155830383,
      "learning_rate": 0.0005950234563596203,
      "loss": 2.2237,
      "step": 13377
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.7177351713180542,
      "learning_rate": 0.0005950227143513484,
      "loss": 2.5856,
      "step": 13378
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.5843554139137268,
      "learning_rate": 0.0005950219722882262,
      "loss": 2.6784,
      "step": 13379
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.574856698513031,
      "learning_rate": 0.0005950212301702538,
      "loss": 2.4198,
      "step": 13380
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6649998426437378,
      "learning_rate": 0.0005950204879974313,
      "loss": 2.4571,
      "step": 13381
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.597853422164917,
      "learning_rate": 0.000595019745769759,
      "loss": 2.4522,
      "step": 13382
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6198598742485046,
      "learning_rate": 0.000595019003487237,
      "loss": 2.2808,
      "step": 13383
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.8332056999206543,
      "learning_rate": 0.0005950182611498653,
      "loss": 2.4178,
      "step": 13384
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6019541621208191,
      "learning_rate": 0.0005950175187576442,
      "loss": 2.5779,
      "step": 13385
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.677941083908081,
      "learning_rate": 0.0005950167763105738,
      "loss": 2.5352,
      "step": 13386
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6135096549987793,
      "learning_rate": 0.0005950160338086542,
      "loss": 2.437,
      "step": 13387
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6000427007675171,
      "learning_rate": 0.0005950152912518855,
      "loss": 2.3729,
      "step": 13388
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.5843909978866577,
      "learning_rate": 0.0005950145486402679,
      "loss": 2.4969,
      "step": 13389
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6333547830581665,
      "learning_rate": 0.0005950138059738016,
      "loss": 2.349,
      "step": 13390
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6576404571533203,
      "learning_rate": 0.0005950130632524866,
      "loss": 2.4053,
      "step": 13391
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6141228675842285,
      "learning_rate": 0.0005950123204763231,
      "loss": 2.4134,
      "step": 13392
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6433093547821045,
      "learning_rate": 0.0005950115776453112,
      "loss": 2.5668,
      "step": 13393
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6346269845962524,
      "learning_rate": 0.0005950108347594512,
      "loss": 2.3309,
      "step": 13394
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.7161707282066345,
      "learning_rate": 0.0005950100918187431,
      "loss": 2.5907,
      "step": 13395
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.5977925658226013,
      "learning_rate": 0.000595009348823187,
      "loss": 2.6218,
      "step": 13396
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6362645030021667,
      "learning_rate": 0.0005950086057727831,
      "loss": 2.5259,
      "step": 13397
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.9513427019119263,
      "learning_rate": 0.0005950078626675315,
      "loss": 2.2849,
      "step": 13398
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6040568351745605,
      "learning_rate": 0.0005950071195074325,
      "loss": 2.4092,
      "step": 13399
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6329551935195923,
      "learning_rate": 0.000595006376292486,
      "loss": 2.505,
      "step": 13400
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6530963182449341,
      "learning_rate": 0.0005950056330226922,
      "loss": 2.3649,
      "step": 13401
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6541812419891357,
      "learning_rate": 0.0005950048896980513,
      "loss": 2.3286,
      "step": 13402
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6079944372177124,
      "learning_rate": 0.0005950041463185635,
      "loss": 2.524,
      "step": 13403
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.7955199480056763,
      "learning_rate": 0.0005950034028842289,
      "loss": 2.346,
      "step": 13404
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.5877126455307007,
      "learning_rate": 0.0005950026593950474,
      "loss": 2.5136,
      "step": 13405
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.7411888837814331,
      "learning_rate": 0.0005950019158510195,
      "loss": 2.3382,
      "step": 13406
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6039113998413086,
      "learning_rate": 0.0005950011722521452,
      "loss": 2.2919,
      "step": 13407
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6741193532943726,
      "learning_rate": 0.0005950004285984245,
      "loss": 2.3828,
      "step": 13408
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6108478903770447,
      "learning_rate": 0.0005949996848898577,
      "loss": 2.2641,
      "step": 13409
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6849870085716248,
      "learning_rate": 0.0005949989411264449,
      "loss": 2.3404,
      "step": 13410
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.5761305689811707,
      "learning_rate": 0.0005949981973081863,
      "loss": 2.2107,
      "step": 13411
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.5872482657432556,
      "learning_rate": 0.0005949974534350818,
      "loss": 2.4065,
      "step": 13412
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6472939848899841,
      "learning_rate": 0.0005949967095071317,
      "loss": 2.4285,
      "step": 13413
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6520707011222839,
      "learning_rate": 0.0005949959655243363,
      "loss": 2.4941,
      "step": 13414
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6469614505767822,
      "learning_rate": 0.0005949952214866956,
      "loss": 2.4589,
      "step": 13415
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.59214186668396,
      "learning_rate": 0.0005949944773942096,
      "loss": 2.1965,
      "step": 13416
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6167407035827637,
      "learning_rate": 0.0005949937332468786,
      "loss": 2.4245,
      "step": 13417
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.7602287530899048,
      "learning_rate": 0.0005949929890447027,
      "loss": 2.4588,
      "step": 13418
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6023978590965271,
      "learning_rate": 0.000594992244787682,
      "loss": 2.2707,
      "step": 13419
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.8360719680786133,
      "learning_rate": 0.0005949915004758167,
      "loss": 2.5055,
      "step": 13420
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6608201265335083,
      "learning_rate": 0.000594990756109107,
      "loss": 2.4502,
      "step": 13421
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6463047862052917,
      "learning_rate": 0.0005949900116875528,
      "loss": 2.4508,
      "step": 13422
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6648276448249817,
      "learning_rate": 0.0005949892672111544,
      "loss": 2.3585,
      "step": 13423
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.7260428667068481,
      "learning_rate": 0.000594988522679912,
      "loss": 2.3402,
      "step": 13424
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6238386034965515,
      "learning_rate": 0.0005949877780938256,
      "loss": 2.1758,
      "step": 13425
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.5898251533508301,
      "learning_rate": 0.0005949870334528956,
      "loss": 2.2769,
      "step": 13426
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6231193542480469,
      "learning_rate": 0.0005949862887571217,
      "loss": 2.6731,
      "step": 13427
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6186731457710266,
      "learning_rate": 0.0005949855440065044,
      "loss": 2.4976,
      "step": 13428
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.5696854591369629,
      "learning_rate": 0.0005949847992010436,
      "loss": 2.297,
      "step": 13429
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6375076770782471,
      "learning_rate": 0.0005949840543407397,
      "loss": 2.4575,
      "step": 13430
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6072131395339966,
      "learning_rate": 0.0005949833094255926,
      "loss": 2.4232,
      "step": 13431
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6004274487495422,
      "learning_rate": 0.0005949825644556027,
      "loss": 2.5889,
      "step": 13432
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.5981467366218567,
      "learning_rate": 0.0005949818194307697,
      "loss": 2.6829,
      "step": 13433
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.5491353869438171,
      "learning_rate": 0.0005949810743510941,
      "loss": 2.4001,
      "step": 13434
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.5466251373291016,
      "learning_rate": 0.0005949803292165761,
      "loss": 2.3369,
      "step": 13435
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.7250614762306213,
      "learning_rate": 0.0005949795840272156,
      "loss": 2.2163,
      "step": 13436
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6817313432693481,
      "learning_rate": 0.0005949788387830129,
      "loss": 2.4507,
      "step": 13437
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.5843056440353394,
      "learning_rate": 0.000594978093483968,
      "loss": 2.5057,
      "step": 13438
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6645482778549194,
      "learning_rate": 0.0005949773481300811,
      "loss": 2.5485,
      "step": 13439
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.5921181440353394,
      "learning_rate": 0.0005949766027213522,
      "loss": 2.3957,
      "step": 13440
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6845781207084656,
      "learning_rate": 0.0005949758572577818,
      "loss": 2.2848,
      "step": 13441
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.594075620174408,
      "learning_rate": 0.0005949751117393697,
      "loss": 2.5339,
      "step": 13442
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.5931477546691895,
      "learning_rate": 0.0005949743661661163,
      "loss": 2.2869,
      "step": 13443
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6791473031044006,
      "learning_rate": 0.0005949736205380214,
      "loss": 2.4233,
      "step": 13444
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6420292258262634,
      "learning_rate": 0.0005949728748550854,
      "loss": 2.5179,
      "step": 13445
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6172261834144592,
      "learning_rate": 0.0005949721291173085,
      "loss": 2.5286,
      "step": 13446
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6074711084365845,
      "learning_rate": 0.0005949713833246906,
      "loss": 2.2709,
      "step": 13447
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6638199090957642,
      "learning_rate": 0.0005949706374772319,
      "loss": 2.3321,
      "step": 13448
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.615117609500885,
      "learning_rate": 0.0005949698915749327,
      "loss": 2.5299,
      "step": 13449
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6710269451141357,
      "learning_rate": 0.0005949691456177931,
      "loss": 2.5383,
      "step": 13450
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6821721792221069,
      "learning_rate": 0.000594968399605813,
      "loss": 2.4642,
      "step": 13451
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.642467200756073,
      "learning_rate": 0.0005949676535389927,
      "loss": 2.3268,
      "step": 13452
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6321666240692139,
      "learning_rate": 0.0005949669074173325,
      "loss": 2.5142,
      "step": 13453
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.7447185516357422,
      "learning_rate": 0.0005949661612408323,
      "loss": 2.6341,
      "step": 13454
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6374556422233582,
      "learning_rate": 0.0005949654150094922,
      "loss": 2.3024,
      "step": 13455
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6078581809997559,
      "learning_rate": 0.0005949646687233126,
      "loss": 2.3161,
      "step": 13456
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.674902081489563,
      "learning_rate": 0.0005949639223822935,
      "loss": 2.0624,
      "step": 13457
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6217489838600159,
      "learning_rate": 0.0005949631759864351,
      "loss": 2.3517,
      "step": 13458
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6970566511154175,
      "learning_rate": 0.0005949624295357374,
      "loss": 2.4767,
      "step": 13459
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6398435235023499,
      "learning_rate": 0.0005949616830302005,
      "loss": 2.4317,
      "step": 13460
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.7260926961898804,
      "learning_rate": 0.0005949609364698248,
      "loss": 2.3706,
      "step": 13461
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6437602639198303,
      "learning_rate": 0.0005949601898546103,
      "loss": 2.5884,
      "step": 13462
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.662187397480011,
      "learning_rate": 0.0005949594431845571,
      "loss": 2.4468,
      "step": 13463
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6457569003105164,
      "learning_rate": 0.0005949586964596654,
      "loss": 2.2816,
      "step": 13464
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.5963628888130188,
      "learning_rate": 0.0005949579496799352,
      "loss": 2.422,
      "step": 13465
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.60024493932724,
      "learning_rate": 0.0005949572028453668,
      "loss": 2.5948,
      "step": 13466
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6646436452865601,
      "learning_rate": 0.0005949564559559604,
      "loss": 2.5159,
      "step": 13467
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.5981054902076721,
      "learning_rate": 0.000594955709011716,
      "loss": 2.3583,
      "step": 13468
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6088199019432068,
      "learning_rate": 0.0005949549620126336,
      "loss": 2.636,
      "step": 13469
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.603909969329834,
      "learning_rate": 0.0005949542149587137,
      "loss": 2.2531,
      "step": 13470
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.5658599734306335,
      "learning_rate": 0.000594953467849956,
      "loss": 2.1666,
      "step": 13471
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6433791518211365,
      "learning_rate": 0.0005949527206863611,
      "loss": 2.4821,
      "step": 13472
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6456917524337769,
      "learning_rate": 0.0005949519734679288,
      "loss": 2.4219,
      "step": 13473
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6790578961372375,
      "learning_rate": 0.0005949512261946595,
      "loss": 2.5452,
      "step": 13474
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6635841131210327,
      "learning_rate": 0.0005949504788665531,
      "loss": 2.2883,
      "step": 13475
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.5842373371124268,
      "learning_rate": 0.0005949497314836099,
      "loss": 2.4267,
      "step": 13476
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.7081807255744934,
      "learning_rate": 0.0005949489840458299,
      "loss": 2.5503,
      "step": 13477
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6694175004959106,
      "learning_rate": 0.0005949482365532133,
      "loss": 2.3945,
      "step": 13478
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6441482901573181,
      "learning_rate": 0.0005949474890057603,
      "loss": 2.5906,
      "step": 13479
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6994189620018005,
      "learning_rate": 0.000594946741403471,
      "loss": 2.447,
      "step": 13480
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.5814400911331177,
      "learning_rate": 0.0005949459937463456,
      "loss": 2.5712,
      "step": 13481
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.5919193625450134,
      "learning_rate": 0.0005949452460343841,
      "loss": 2.2415,
      "step": 13482
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6671947836875916,
      "learning_rate": 0.0005949444982675867,
      "loss": 2.3773,
      "step": 13483
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6158980131149292,
      "learning_rate": 0.0005949437504459537,
      "loss": 2.4081,
      "step": 13484
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.5777930617332458,
      "learning_rate": 0.0005949430025694849,
      "loss": 2.297,
      "step": 13485
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.5725317597389221,
      "learning_rate": 0.0005949422546381807,
      "loss": 2.2206,
      "step": 13486
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.5930712819099426,
      "learning_rate": 0.000594941506652041,
      "loss": 2.4806,
      "step": 13487
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6713021993637085,
      "learning_rate": 0.0005949407586110663,
      "loss": 2.5874,
      "step": 13488
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.640902042388916,
      "learning_rate": 0.0005949400105152566,
      "loss": 2.3837,
      "step": 13489
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6448187828063965,
      "learning_rate": 0.0005949392623646119,
      "loss": 2.6451,
      "step": 13490
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6027529835700989,
      "learning_rate": 0.0005949385141591324,
      "loss": 2.4193,
      "step": 13491
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.5609195828437805,
      "learning_rate": 0.0005949377658988182,
      "loss": 2.473,
      "step": 13492
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.5690372586250305,
      "learning_rate": 0.0005949370175836696,
      "loss": 2.5254,
      "step": 13493
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6200988292694092,
      "learning_rate": 0.0005949362692136866,
      "loss": 2.5963,
      "step": 13494
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.7935812473297119,
      "learning_rate": 0.0005949355207888695,
      "loss": 2.1908,
      "step": 13495
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6674346923828125,
      "learning_rate": 0.0005949347723092181,
      "loss": 2.584,
      "step": 13496
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6678338646888733,
      "learning_rate": 0.0005949340237747329,
      "loss": 2.394,
      "step": 13497
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6459780931472778,
      "learning_rate": 0.0005949332751854139,
      "loss": 2.2592,
      "step": 13498
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.7896745204925537,
      "learning_rate": 0.0005949325265412612,
      "loss": 2.2121,
      "step": 13499
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.5545274019241333,
      "learning_rate": 0.000594931777842275,
      "loss": 2.4141,
      "step": 13500
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.60394287109375,
      "learning_rate": 0.0005949310290884554,
      "loss": 2.3303,
      "step": 13501
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6889804601669312,
      "learning_rate": 0.0005949302802798026,
      "loss": 2.2818,
      "step": 13502
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6139864921569824,
      "learning_rate": 0.0005949295314163166,
      "loss": 2.4199,
      "step": 13503
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.64058518409729,
      "learning_rate": 0.0005949287824979977,
      "loss": 2.3036,
      "step": 13504
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6011619567871094,
      "learning_rate": 0.000594928033524846,
      "loss": 2.4042,
      "step": 13505
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6671206951141357,
      "learning_rate": 0.0005949272844968616,
      "loss": 2.3124,
      "step": 13506
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6335325241088867,
      "learning_rate": 0.0005949265354140446,
      "loss": 2.4591,
      "step": 13507
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6463833451271057,
      "learning_rate": 0.0005949257862763952,
      "loss": 2.2284,
      "step": 13508
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.7807073593139648,
      "learning_rate": 0.0005949250370839136,
      "loss": 2.2557,
      "step": 13509
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.5929926037788391,
      "learning_rate": 0.0005949242878365996,
      "loss": 2.4045,
      "step": 13510
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6393749713897705,
      "learning_rate": 0.0005949235385344539,
      "loss": 2.3876,
      "step": 13511
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.5873545408248901,
      "learning_rate": 0.0005949227891774763,
      "loss": 2.5658,
      "step": 13512
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.5886614322662354,
      "learning_rate": 0.0005949220397656669,
      "loss": 2.5324,
      "step": 13513
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.599445641040802,
      "learning_rate": 0.000594921290299026,
      "loss": 2.1588,
      "step": 13514
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6060319542884827,
      "learning_rate": 0.0005949205407775537,
      "loss": 2.5017,
      "step": 13515
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.625054121017456,
      "learning_rate": 0.00059491979120125,
      "loss": 2.4217,
      "step": 13516
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.616733968257904,
      "learning_rate": 0.0005949190415701151,
      "loss": 2.4209,
      "step": 13517
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6032246351242065,
      "learning_rate": 0.0005949182918841494,
      "loss": 2.3616,
      "step": 13518
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.5693563222885132,
      "learning_rate": 0.0005949175421433526,
      "loss": 2.4779,
      "step": 13519
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.57206130027771,
      "learning_rate": 0.0005949167923477252,
      "loss": 2.2007,
      "step": 13520
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.5875398516654968,
      "learning_rate": 0.0005949160424972671,
      "loss": 2.3492,
      "step": 13521
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6381019353866577,
      "learning_rate": 0.0005949152925919785,
      "loss": 2.5399,
      "step": 13522
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6222661137580872,
      "learning_rate": 0.0005949145426318598,
      "loss": 2.4656,
      "step": 13523
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6170911192893982,
      "learning_rate": 0.0005949137926169109,
      "loss": 2.7116,
      "step": 13524
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.5728052258491516,
      "learning_rate": 0.0005949130425471317,
      "loss": 2.6226,
      "step": 13525
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6528790593147278,
      "learning_rate": 0.0005949122924225228,
      "loss": 2.4446,
      "step": 13526
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6731751561164856,
      "learning_rate": 0.000594911542243084,
      "loss": 2.4042,
      "step": 13527
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6216723918914795,
      "learning_rate": 0.0005949107920088158,
      "loss": 2.6972,
      "step": 13528
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6420906782150269,
      "learning_rate": 0.000594910041719718,
      "loss": 2.5231,
      "step": 13529
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6405669450759888,
      "learning_rate": 0.0005949092913757908,
      "loss": 2.6149,
      "step": 13530
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.5883237719535828,
      "learning_rate": 0.0005949085409770344,
      "loss": 2.5032,
      "step": 13531
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6021261215209961,
      "learning_rate": 0.000594907790523449,
      "loss": 2.5597,
      "step": 13532
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.7447994351387024,
      "learning_rate": 0.0005949070400150347,
      "loss": 2.5247,
      "step": 13533
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6254193186759949,
      "learning_rate": 0.0005949062894517915,
      "loss": 2.3541,
      "step": 13534
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6106089353561401,
      "learning_rate": 0.0005949055388337198,
      "loss": 2.4999,
      "step": 13535
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6937354803085327,
      "learning_rate": 0.0005949047881608194,
      "loss": 2.385,
      "step": 13536
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6433785557746887,
      "learning_rate": 0.0005949040374330907,
      "loss": 2.4575,
      "step": 13537
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6160752177238464,
      "learning_rate": 0.0005949032866505339,
      "loss": 2.4074,
      "step": 13538
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.577738881111145,
      "learning_rate": 0.0005949025358131489,
      "loss": 2.5195,
      "step": 13539
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.5531056523323059,
      "learning_rate": 0.0005949017849209361,
      "loss": 2.2458,
      "step": 13540
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6317073106765747,
      "learning_rate": 0.0005949010339738953,
      "loss": 2.0201,
      "step": 13541
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.5881479382514954,
      "learning_rate": 0.0005949002829720269,
      "loss": 2.3248,
      "step": 13542
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.5850074291229248,
      "learning_rate": 0.000594899531915331,
      "loss": 2.427,
      "step": 13543
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6073331832885742,
      "learning_rate": 0.0005948987808038076,
      "loss": 2.4277,
      "step": 13544
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.5650675296783447,
      "learning_rate": 0.0005948980296374572,
      "loss": 2.1782,
      "step": 13545
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.5779293179512024,
      "learning_rate": 0.0005948972784162794,
      "loss": 2.196,
      "step": 13546
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6738240718841553,
      "learning_rate": 0.0005948965271402747,
      "loss": 2.5798,
      "step": 13547
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.5871108770370483,
      "learning_rate": 0.0005948957758094433,
      "loss": 2.3477,
      "step": 13548
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6028518676757812,
      "learning_rate": 0.0005948950244237851,
      "loss": 2.6539,
      "step": 13549
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.5829794406890869,
      "learning_rate": 0.0005948942729833005,
      "loss": 2.1386,
      "step": 13550
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.7158008217811584,
      "learning_rate": 0.0005948935214879893,
      "loss": 2.7093,
      "step": 13551
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6437811851501465,
      "learning_rate": 0.0005948927699378518,
      "loss": 2.3674,
      "step": 13552
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.5720471143722534,
      "learning_rate": 0.0005948920183328883,
      "loss": 2.4055,
      "step": 13553
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6193353533744812,
      "learning_rate": 0.0005948912666730988,
      "loss": 2.4863,
      "step": 13554
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.5616745352745056,
      "learning_rate": 0.0005948905149584833,
      "loss": 2.5439,
      "step": 13555
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6381232142448425,
      "learning_rate": 0.0005948897631890423,
      "loss": 2.4676,
      "step": 13556
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6044015884399414,
      "learning_rate": 0.0005948890113647756,
      "loss": 2.5111,
      "step": 13557
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.5889226794242859,
      "learning_rate": 0.0005948882594856835,
      "loss": 2.5938,
      "step": 13558
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.5909936428070068,
      "learning_rate": 0.0005948875075517661,
      "loss": 2.51,
      "step": 13559
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6016305685043335,
      "learning_rate": 0.0005948867555630236,
      "loss": 2.6735,
      "step": 13560
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6062238812446594,
      "learning_rate": 0.000594886003519456,
      "loss": 2.5021,
      "step": 13561
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.5648387670516968,
      "learning_rate": 0.0005948852514210636,
      "loss": 2.4571,
      "step": 13562
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.5795348882675171,
      "learning_rate": 0.0005948844992678464,
      "loss": 2.4761,
      "step": 13563
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.5489165782928467,
      "learning_rate": 0.0005948837470598047,
      "loss": 2.5363,
      "step": 13564
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.5651484727859497,
      "learning_rate": 0.0005948829947969384,
      "loss": 2.32,
      "step": 13565
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6245830655097961,
      "learning_rate": 0.0005948822424792479,
      "loss": 2.376,
      "step": 13566
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6243774890899658,
      "learning_rate": 0.0005948814901067332,
      "loss": 2.5449,
      "step": 13567
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.5515695810317993,
      "learning_rate": 0.0005948807376793945,
      "loss": 2.3526,
      "step": 13568
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6273607015609741,
      "learning_rate": 0.0005948799851972317,
      "loss": 2.4213,
      "step": 13569
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.5912098288536072,
      "learning_rate": 0.0005948792326602453,
      "loss": 2.5424,
      "step": 13570
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.572331428527832,
      "learning_rate": 0.0005948784800684354,
      "loss": 2.3281,
      "step": 13571
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6190966963768005,
      "learning_rate": 0.0005948777274218018,
      "loss": 2.4235,
      "step": 13572
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.5873916745185852,
      "learning_rate": 0.0005948769747203451,
      "loss": 2.6477,
      "step": 13573
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.5671141147613525,
      "learning_rate": 0.0005948762219640649,
      "loss": 2.5235,
      "step": 13574
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.5477463006973267,
      "learning_rate": 0.0005948754691529618,
      "loss": 2.2515,
      "step": 13575
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6324975490570068,
      "learning_rate": 0.0005948747162870359,
      "loss": 2.3967,
      "step": 13576
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6147582530975342,
      "learning_rate": 0.0005948739633662871,
      "loss": 2.5057,
      "step": 13577
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6372228264808655,
      "learning_rate": 0.0005948732103907157,
      "loss": 2.4049,
      "step": 13578
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.678427517414093,
      "learning_rate": 0.0005948724573603218,
      "loss": 2.3618,
      "step": 13579
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6300115585327148,
      "learning_rate": 0.0005948717042751056,
      "loss": 2.4327,
      "step": 13580
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.5927145481109619,
      "learning_rate": 0.0005948709511350671,
      "loss": 2.4181,
      "step": 13581
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6557661294937134,
      "learning_rate": 0.0005948701979402065,
      "loss": 2.4293,
      "step": 13582
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.5472704172134399,
      "learning_rate": 0.000594869444690524,
      "loss": 2.3415,
      "step": 13583
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6541814208030701,
      "learning_rate": 0.0005948686913860197,
      "loss": 2.2412,
      "step": 13584
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6412220001220703,
      "learning_rate": 0.0005948679380266938,
      "loss": 2.6873,
      "step": 13585
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6218368411064148,
      "learning_rate": 0.0005948671846125464,
      "loss": 2.7065,
      "step": 13586
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.621612548828125,
      "learning_rate": 0.0005948664311435775,
      "loss": 2.6706,
      "step": 13587
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.5704042315483093,
      "learning_rate": 0.0005948656776197875,
      "loss": 2.4915,
      "step": 13588
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.591793954372406,
      "learning_rate": 0.0005948649240411763,
      "loss": 2.4772,
      "step": 13589
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6180429458618164,
      "learning_rate": 0.0005948641704077442,
      "loss": 2.3707,
      "step": 13590
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.5968827605247498,
      "learning_rate": 0.0005948634167194913,
      "loss": 2.4014,
      "step": 13591
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6298989653587341,
      "learning_rate": 0.0005948626629764176,
      "loss": 2.5637,
      "step": 13592
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6484348177909851,
      "learning_rate": 0.0005948619091785235,
      "loss": 2.3619,
      "step": 13593
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6570892333984375,
      "learning_rate": 0.000594861155325809,
      "loss": 2.6313,
      "step": 13594
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6202755570411682,
      "learning_rate": 0.0005948604014182742,
      "loss": 2.3959,
      "step": 13595
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6521520614624023,
      "learning_rate": 0.0005948596474559193,
      "loss": 2.5025,
      "step": 13596
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.5599690675735474,
      "learning_rate": 0.0005948588934387444,
      "loss": 2.3991,
      "step": 13597
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6160020232200623,
      "learning_rate": 0.0005948581393667497,
      "loss": 2.2425,
      "step": 13598
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6295548677444458,
      "learning_rate": 0.0005948573852399353,
      "loss": 2.2635,
      "step": 13599
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6537402868270874,
      "learning_rate": 0.0005948566310583013,
      "loss": 2.267,
      "step": 13600
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6236485838890076,
      "learning_rate": 0.0005948558768218479,
      "loss": 2.3233,
      "step": 13601
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.7174886465072632,
      "learning_rate": 0.0005948551225305753,
      "loss": 2.51,
      "step": 13602
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.5932166576385498,
      "learning_rate": 0.0005948543681844835,
      "loss": 2.5737,
      "step": 13603
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6631160974502563,
      "learning_rate": 0.0005948536137835727,
      "loss": 2.5984,
      "step": 13604
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.8477990627288818,
      "learning_rate": 0.000594852859327843,
      "loss": 2.2562,
      "step": 13605
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6427867412567139,
      "learning_rate": 0.0005948521048172948,
      "loss": 2.3306,
      "step": 13606
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.8687239289283752,
      "learning_rate": 0.0005948513502519278,
      "loss": 2.7786,
      "step": 13607
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6255694627761841,
      "learning_rate": 0.0005948505956317426,
      "loss": 2.4311,
      "step": 13608
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6239529848098755,
      "learning_rate": 0.0005948498409567389,
      "loss": 2.2146,
      "step": 13609
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.747490406036377,
      "learning_rate": 0.0005948490862269171,
      "loss": 2.4527,
      "step": 13610
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.5896645188331604,
      "learning_rate": 0.0005948483314422773,
      "loss": 2.5117,
      "step": 13611
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6707630157470703,
      "learning_rate": 0.0005948475766028197,
      "loss": 2.522,
      "step": 13612
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6479586362838745,
      "learning_rate": 0.0005948468217085442,
      "loss": 2.5114,
      "step": 13613
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.706071674823761,
      "learning_rate": 0.0005948460667594512,
      "loss": 2.5726,
      "step": 13614
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6079217195510864,
      "learning_rate": 0.0005948453117555407,
      "loss": 2.5431,
      "step": 13615
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6699856519699097,
      "learning_rate": 0.0005948445566968131,
      "loss": 2.3734,
      "step": 13616
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.5650193095207214,
      "learning_rate": 0.0005948438015832681,
      "loss": 2.4407,
      "step": 13617
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.5542470216751099,
      "learning_rate": 0.0005948430464149062,
      "loss": 2.2909,
      "step": 13618
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.5945084691047668,
      "learning_rate": 0.0005948422911917274,
      "loss": 2.6306,
      "step": 13619
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.610683262348175,
      "learning_rate": 0.0005948415359137318,
      "loss": 2.0442,
      "step": 13620
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6451095938682556,
      "learning_rate": 0.0005948407805809196,
      "loss": 2.3428,
      "step": 13621
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.5665882229804993,
      "learning_rate": 0.0005948400251932909,
      "loss": 2.4248,
      "step": 13622
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6185846328735352,
      "learning_rate": 0.0005948392697508459,
      "loss": 2.5318,
      "step": 13623
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6739230155944824,
      "learning_rate": 0.0005948385142535848,
      "loss": 2.27,
      "step": 13624
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.5648642778396606,
      "learning_rate": 0.0005948377587015076,
      "loss": 2.3895,
      "step": 13625
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6475327014923096,
      "learning_rate": 0.0005948370030946145,
      "loss": 2.4417,
      "step": 13626
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.8089457154273987,
      "learning_rate": 0.0005948362474329056,
      "loss": 2.3822,
      "step": 13627
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.602087676525116,
      "learning_rate": 0.000594835491716381,
      "loss": 2.5538,
      "step": 13628
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6146280169487,
      "learning_rate": 0.0005948347359450412,
      "loss": 2.5027,
      "step": 13629
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.5625877976417542,
      "learning_rate": 0.0005948339801188857,
      "loss": 2.4473,
      "step": 13630
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.628440797328949,
      "learning_rate": 0.0005948332242379153,
      "loss": 2.3763,
      "step": 13631
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.578820526599884,
      "learning_rate": 0.0005948324683021297,
      "loss": 2.4141,
      "step": 13632
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6117576956748962,
      "learning_rate": 0.0005948317123115292,
      "loss": 2.5597,
      "step": 13633
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6918303966522217,
      "learning_rate": 0.0005948309562661138,
      "loss": 2.6707,
      "step": 13634
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6369006633758545,
      "learning_rate": 0.0005948302001658839,
      "loss": 2.4858,
      "step": 13635
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6238405108451843,
      "learning_rate": 0.0005948294440108394,
      "loss": 2.6417,
      "step": 13636
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.572736382484436,
      "learning_rate": 0.0005948286878009806,
      "loss": 2.3363,
      "step": 13637
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6614992618560791,
      "learning_rate": 0.0005948279315363075,
      "loss": 2.6043,
      "step": 13638
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6159353852272034,
      "learning_rate": 0.0005948271752168206,
      "loss": 2.6659,
      "step": 13639
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.5887539386749268,
      "learning_rate": 0.0005948264188425195,
      "loss": 2.3359,
      "step": 13640
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6110301613807678,
      "learning_rate": 0.0005948256624134047,
      "loss": 2.3745,
      "step": 13641
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6210810542106628,
      "learning_rate": 0.0005948249059294763,
      "loss": 2.0054,
      "step": 13642
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.607437252998352,
      "learning_rate": 0.0005948241493907342,
      "loss": 2.4743,
      "step": 13643
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6259310841560364,
      "learning_rate": 0.0005948233927971789,
      "loss": 2.5316,
      "step": 13644
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6329923272132874,
      "learning_rate": 0.0005948226361488103,
      "loss": 2.7324,
      "step": 13645
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6227499842643738,
      "learning_rate": 0.0005948218794456286,
      "loss": 2.3251,
      "step": 13646
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6542433500289917,
      "learning_rate": 0.0005948211226876339,
      "loss": 2.4413,
      "step": 13647
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6439056396484375,
      "learning_rate": 0.0005948203658748264,
      "loss": 2.5075,
      "step": 13648
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6060140132904053,
      "learning_rate": 0.0005948196090072063,
      "loss": 2.6189,
      "step": 13649
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.5639419555664062,
      "learning_rate": 0.0005948188520847736,
      "loss": 2.1861,
      "step": 13650
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6263946294784546,
      "learning_rate": 0.0005948180951075286,
      "loss": 2.6376,
      "step": 13651
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6943204402923584,
      "learning_rate": 0.0005948173380754714,
      "loss": 2.6322,
      "step": 13652
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6167874336242676,
      "learning_rate": 0.0005948165809886018,
      "loss": 2.6103,
      "step": 13653
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.5984241962432861,
      "learning_rate": 0.0005948158238469205,
      "loss": 2.4089,
      "step": 13654
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6831508278846741,
      "learning_rate": 0.0005948150666504273,
      "loss": 2.6231,
      "step": 13655
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6179364323616028,
      "learning_rate": 0.0005948143093991224,
      "loss": 2.4605,
      "step": 13656
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.5895823836326599,
      "learning_rate": 0.0005948135520930059,
      "loss": 2.3267,
      "step": 13657
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6246792674064636,
      "learning_rate": 0.0005948127947320781,
      "loss": 2.6256,
      "step": 13658
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.629882276058197,
      "learning_rate": 0.000594812037316339,
      "loss": 2.6114,
      "step": 13659
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6570702195167542,
      "learning_rate": 0.0005948112798457888,
      "loss": 2.4626,
      "step": 13660
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.5945261716842651,
      "learning_rate": 0.0005948105223204277,
      "loss": 2.3461,
      "step": 13661
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6035777926445007,
      "learning_rate": 0.0005948097647402556,
      "loss": 2.5086,
      "step": 13662
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6097206473350525,
      "learning_rate": 0.0005948090071052728,
      "loss": 2.336,
      "step": 13663
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.5844933986663818,
      "learning_rate": 0.0005948082494154795,
      "loss": 2.2749,
      "step": 13664
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6811867952346802,
      "learning_rate": 0.0005948074916708758,
      "loss": 2.471,
      "step": 13665
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6013518571853638,
      "learning_rate": 0.0005948067338714618,
      "loss": 2.1696,
      "step": 13666
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6132528781890869,
      "learning_rate": 0.0005948059760172376,
      "loss": 2.6042,
      "step": 13667
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6142184734344482,
      "learning_rate": 0.0005948052181082035,
      "loss": 2.3187,
      "step": 13668
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6285222172737122,
      "learning_rate": 0.0005948044601443595,
      "loss": 2.555,
      "step": 13669
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6065221428871155,
      "learning_rate": 0.0005948037021257058,
      "loss": 2.3943,
      "step": 13670
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6648041009902954,
      "learning_rate": 0.0005948029440522425,
      "loss": 2.4565,
      "step": 13671
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6531583070755005,
      "learning_rate": 0.0005948021859239698,
      "loss": 2.344,
      "step": 13672
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.5936684012413025,
      "learning_rate": 0.0005948014277408878,
      "loss": 2.2565,
      "step": 13673
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6282951235771179,
      "learning_rate": 0.0005948006695029966,
      "loss": 2.2601,
      "step": 13674
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6168058514595032,
      "learning_rate": 0.0005947999112102964,
      "loss": 2.6234,
      "step": 13675
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6018487811088562,
      "learning_rate": 0.0005947991528627874,
      "loss": 2.4203,
      "step": 13676
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6145367622375488,
      "learning_rate": 0.0005947983944604696,
      "loss": 2.1899,
      "step": 13677
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6450887322425842,
      "learning_rate": 0.0005947976360033433,
      "loss": 2.6345,
      "step": 13678
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6059520840644836,
      "learning_rate": 0.0005947968774914086,
      "loss": 2.3517,
      "step": 13679
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.5606125593185425,
      "learning_rate": 0.0005947961189246654,
      "loss": 2.5335,
      "step": 13680
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.7000762820243835,
      "learning_rate": 0.0005947953603031141,
      "loss": 2.4786,
      "step": 13681
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.5861596465110779,
      "learning_rate": 0.0005947946016267549,
      "loss": 2.5172,
      "step": 13682
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6301013827323914,
      "learning_rate": 0.0005947938428955878,
      "loss": 2.4675,
      "step": 13683
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6111066937446594,
      "learning_rate": 0.0005947930841096129,
      "loss": 2.2704,
      "step": 13684
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.5898814797401428,
      "learning_rate": 0.0005947923252688303,
      "loss": 2.2649,
      "step": 13685
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6334382891654968,
      "learning_rate": 0.0005947915663732404,
      "loss": 2.4653,
      "step": 13686
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6231450438499451,
      "learning_rate": 0.0005947908074228431,
      "loss": 2.515,
      "step": 13687
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.7611697316169739,
      "learning_rate": 0.0005947900484176386,
      "loss": 2.2469,
      "step": 13688
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.5931432247161865,
      "learning_rate": 0.0005947892893576271,
      "loss": 2.3821,
      "step": 13689
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.5722842216491699,
      "learning_rate": 0.0005947885302428088,
      "loss": 2.5959,
      "step": 13690
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6275922656059265,
      "learning_rate": 0.0005947877710731837,
      "loss": 2.7059,
      "step": 13691
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.590542197227478,
      "learning_rate": 0.0005947870118487518,
      "loss": 2.367,
      "step": 13692
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6725398898124695,
      "learning_rate": 0.0005947862525695137,
      "loss": 2.2259,
      "step": 13693
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6178542375564575,
      "learning_rate": 0.0005947854932354692,
      "loss": 2.6186,
      "step": 13694
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6406182646751404,
      "learning_rate": 0.0005947847338466185,
      "loss": 2.4242,
      "step": 13695
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6061611175537109,
      "learning_rate": 0.0005947839744029616,
      "loss": 2.2282,
      "step": 13696
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6087133884429932,
      "learning_rate": 0.000594783214904499,
      "loss": 2.2896,
      "step": 13697
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6073716282844543,
      "learning_rate": 0.0005947824553512305,
      "loss": 2.4944,
      "step": 13698
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6303741931915283,
      "learning_rate": 0.0005947816957431564,
      "loss": 2.2648,
      "step": 13699
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6063796877861023,
      "learning_rate": 0.0005947809360802769,
      "loss": 2.4023,
      "step": 13700
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6365950107574463,
      "learning_rate": 0.000594780176362592,
      "loss": 2.235,
      "step": 13701
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.8884082436561584,
      "learning_rate": 0.000594779416590102,
      "loss": 2.2071,
      "step": 13702
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.7235609889030457,
      "learning_rate": 0.0005947786567628068,
      "loss": 2.3841,
      "step": 13703
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6372131705284119,
      "learning_rate": 0.0005947778968807067,
      "loss": 2.4462,
      "step": 13704
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.002804160118103,
      "learning_rate": 0.0005947771369438019,
      "loss": 2.3823,
      "step": 13705
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.579816997051239,
      "learning_rate": 0.0005947763769520925,
      "loss": 2.3469,
      "step": 13706
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.864262580871582,
      "learning_rate": 0.0005947756169055785,
      "loss": 2.4441,
      "step": 13707
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.7752406597137451,
      "learning_rate": 0.0005947748568042602,
      "loss": 2.4025,
      "step": 13708
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6488090753555298,
      "learning_rate": 0.0005947740966481376,
      "loss": 2.3324,
      "step": 13709
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6709998250007629,
      "learning_rate": 0.0005947733364372109,
      "loss": 2.4446,
      "step": 13710
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.7714928388595581,
      "learning_rate": 0.0005947725761714805,
      "loss": 2.4237,
      "step": 13711
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.713470995426178,
      "learning_rate": 0.0005947718158509462,
      "loss": 2.2513,
      "step": 13712
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6646007895469666,
      "learning_rate": 0.0005947710554756083,
      "loss": 2.3843,
      "step": 13713
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.8186127543449402,
      "learning_rate": 0.0005947702950454668,
      "loss": 2.5347,
      "step": 13714
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6573935747146606,
      "learning_rate": 0.0005947695345605219,
      "loss": 2.171,
      "step": 13715
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.647100031375885,
      "learning_rate": 0.0005947687740207739,
      "loss": 2.673,
      "step": 13716
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6464183330535889,
      "learning_rate": 0.0005947680134262228,
      "loss": 2.4412,
      "step": 13717
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.8590682148933411,
      "learning_rate": 0.0005947672527768687,
      "loss": 2.344,
      "step": 13718
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6955447793006897,
      "learning_rate": 0.0005947664920727118,
      "loss": 2.3965,
      "step": 13719
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6035520434379578,
      "learning_rate": 0.0005947657313137524,
      "loss": 2.5027,
      "step": 13720
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.8529040813446045,
      "learning_rate": 0.0005947649704999904,
      "loss": 2.4065,
      "step": 13721
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6936883330345154,
      "learning_rate": 0.000594764209631426,
      "loss": 2.2423,
      "step": 13722
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.5883587002754211,
      "learning_rate": 0.0005947634487080593,
      "loss": 2.3306,
      "step": 13723
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6569019556045532,
      "learning_rate": 0.0005947626877298906,
      "loss": 2.3625,
      "step": 13724
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6216627955436707,
      "learning_rate": 0.0005947619266969199,
      "loss": 2.2988,
      "step": 13725
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.583987295627594,
      "learning_rate": 0.0005947611656091475,
      "loss": 2.3996,
      "step": 13726
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6608036160469055,
      "learning_rate": 0.0005947604044665732,
      "loss": 2.3122,
      "step": 13727
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6241626739501953,
      "learning_rate": 0.0005947596432691977,
      "loss": 2.4978,
      "step": 13728
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.7302932739257812,
      "learning_rate": 0.0005947588820170206,
      "loss": 2.2283,
      "step": 13729
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6141861081123352,
      "learning_rate": 0.0005947581207100424,
      "loss": 2.3508,
      "step": 13730
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6197513341903687,
      "learning_rate": 0.000594757359348263,
      "loss": 2.2468,
      "step": 13731
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6459219455718994,
      "learning_rate": 0.0005947565979316827,
      "loss": 2.4395,
      "step": 13732
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6088085770606995,
      "learning_rate": 0.0005947558364603015,
      "loss": 2.3137,
      "step": 13733
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.5875057578086853,
      "learning_rate": 0.0005947550749341197,
      "loss": 2.5368,
      "step": 13734
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.5991402268409729,
      "learning_rate": 0.0005947543133531373,
      "loss": 2.34,
      "step": 13735
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.5893569588661194,
      "learning_rate": 0.0005947535517173546,
      "loss": 2.539,
      "step": 13736
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.637520968914032,
      "learning_rate": 0.0005947527900267716,
      "loss": 2.4044,
      "step": 13737
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.8173031806945801,
      "learning_rate": 0.0005947520282813886,
      "loss": 2.3437,
      "step": 13738
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.5971081256866455,
      "learning_rate": 0.0005947512664812054,
      "loss": 2.1502,
      "step": 13739
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6569421887397766,
      "learning_rate": 0.0005947505046262224,
      "loss": 2.3496,
      "step": 13740
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.686116635799408,
      "learning_rate": 0.0005947497427164399,
      "loss": 2.3425,
      "step": 13741
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.5999675989151001,
      "learning_rate": 0.0005947489807518577,
      "loss": 2.2307,
      "step": 13742
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6676475405693054,
      "learning_rate": 0.0005947482187324763,
      "loss": 2.4335,
      "step": 13743
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.7469459772109985,
      "learning_rate": 0.0005947474566582955,
      "loss": 2.2845,
      "step": 13744
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6360676884651184,
      "learning_rate": 0.0005947466945293156,
      "loss": 2.5823,
      "step": 13745
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6535835862159729,
      "learning_rate": 0.0005947459323455368,
      "loss": 2.4545,
      "step": 13746
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6416302919387817,
      "learning_rate": 0.000594745170106959,
      "loss": 2.5542,
      "step": 13747
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.7805540561676025,
      "learning_rate": 0.0005947444078135826,
      "loss": 2.4368,
      "step": 13748
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.5951452851295471,
      "learning_rate": 0.0005947436454654077,
      "loss": 2.3012,
      "step": 13749
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.620048463344574,
      "learning_rate": 0.0005947428830624344,
      "loss": 2.4725,
      "step": 13750
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.9795018434524536,
      "learning_rate": 0.0005947421206046628,
      "loss": 2.5047,
      "step": 13751
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6349906921386719,
      "learning_rate": 0.0005947413580920931,
      "loss": 2.2554,
      "step": 13752
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.7271134257316589,
      "learning_rate": 0.0005947405955247254,
      "loss": 2.2405,
      "step": 13753
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.8338669538497925,
      "learning_rate": 0.0005947398329025598,
      "loss": 2.2585,
      "step": 13754
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.5745161771774292,
      "learning_rate": 0.0005947390702255966,
      "loss": 2.3514,
      "step": 13755
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6819459199905396,
      "learning_rate": 0.0005947383074938357,
      "loss": 2.2985,
      "step": 13756
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6642230153083801,
      "learning_rate": 0.0005947375447072776,
      "loss": 2.3929,
      "step": 13757
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.595893144607544,
      "learning_rate": 0.000594736781865922,
      "loss": 2.3579,
      "step": 13758
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.5806500315666199,
      "learning_rate": 0.0005947360189697693,
      "loss": 2.1432,
      "step": 13759
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.698031485080719,
      "learning_rate": 0.0005947352560188197,
      "loss": 2.4434,
      "step": 13760
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6855114698410034,
      "learning_rate": 0.0005947344930130733,
      "loss": 2.331,
      "step": 13761
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.5988770723342896,
      "learning_rate": 0.0005947337299525301,
      "loss": 2.4272,
      "step": 13762
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.7570013999938965,
      "learning_rate": 0.0005947329668371903,
      "loss": 2.4045,
      "step": 13763
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.71330326795578,
      "learning_rate": 0.0005947322036670542,
      "loss": 2.3941,
      "step": 13764
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6517919898033142,
      "learning_rate": 0.0005947314404421218,
      "loss": 2.3755,
      "step": 13765
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.5845262408256531,
      "learning_rate": 0.0005947306771623931,
      "loss": 2.2562,
      "step": 13766
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6844894289970398,
      "learning_rate": 0.0005947299138278686,
      "loss": 2.4516,
      "step": 13767
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6472015380859375,
      "learning_rate": 0.0005947291504385482,
      "loss": 2.2533,
      "step": 13768
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6911405920982361,
      "learning_rate": 0.000594728386994432,
      "loss": 2.3714,
      "step": 13769
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6098929643630981,
      "learning_rate": 0.0005947276234955203,
      "loss": 2.6657,
      "step": 13770
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.617789089679718,
      "learning_rate": 0.0005947268599418131,
      "loss": 2.5005,
      "step": 13771
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6194166541099548,
      "learning_rate": 0.0005947260963333108,
      "loss": 2.6506,
      "step": 13772
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.622667670249939,
      "learning_rate": 0.0005947253326700131,
      "loss": 2.6194,
      "step": 13773
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6069373488426208,
      "learning_rate": 0.0005947245689519205,
      "loss": 2.4688,
      "step": 13774
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6057300567626953,
      "learning_rate": 0.0005947238051790332,
      "loss": 2.3085,
      "step": 13775
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6332162618637085,
      "learning_rate": 0.0005947230413513511,
      "loss": 2.218,
      "step": 13776
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.680240273475647,
      "learning_rate": 0.0005947222774688743,
      "loss": 2.3125,
      "step": 13777
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6683917045593262,
      "learning_rate": 0.0005947215135316031,
      "loss": 2.4698,
      "step": 13778
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6657328605651855,
      "learning_rate": 0.0005947207495395376,
      "loss": 2.3893,
      "step": 13779
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.5534421801567078,
      "learning_rate": 0.0005947199854926781,
      "loss": 2.5545,
      "step": 13780
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.5812377333641052,
      "learning_rate": 0.0005947192213910244,
      "loss": 2.3135,
      "step": 13781
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6469442844390869,
      "learning_rate": 0.000594718457234577,
      "loss": 2.4734,
      "step": 13782
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6331062316894531,
      "learning_rate": 0.0005947176930233358,
      "loss": 2.5353,
      "step": 13783
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.5784369111061096,
      "learning_rate": 0.0005947169287573009,
      "loss": 2.3744,
      "step": 13784
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.5894170999526978,
      "learning_rate": 0.0005947161644364727,
      "loss": 2.4834,
      "step": 13785
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6503947973251343,
      "learning_rate": 0.0005947154000608512,
      "loss": 2.3479,
      "step": 13786
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6216854453086853,
      "learning_rate": 0.0005947146356304365,
      "loss": 2.4221,
      "step": 13787
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6745907068252563,
      "learning_rate": 0.0005947138711452287,
      "loss": 2.702,
      "step": 13788
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6395936012268066,
      "learning_rate": 0.0005947131066052282,
      "loss": 2.3658,
      "step": 13789
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.5672436952590942,
      "learning_rate": 0.0005947123420104349,
      "loss": 2.3522,
      "step": 13790
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6061877012252808,
      "learning_rate": 0.0005947115773608491,
      "loss": 2.3862,
      "step": 13791
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6090368628501892,
      "learning_rate": 0.0005947108126564707,
      "loss": 2.5565,
      "step": 13792
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.5570923089981079,
      "learning_rate": 0.0005947100478973001,
      "loss": 2.4848,
      "step": 13793
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.5975067615509033,
      "learning_rate": 0.0005947092830833373,
      "loss": 2.1897,
      "step": 13794
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6610047817230225,
      "learning_rate": 0.0005947085182145825,
      "loss": 2.526,
      "step": 13795
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6555641293525696,
      "learning_rate": 0.0005947077532910358,
      "loss": 2.3917,
      "step": 13796
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6095333099365234,
      "learning_rate": 0.0005947069883126973,
      "loss": 2.2764,
      "step": 13797
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6678768396377563,
      "learning_rate": 0.0005947062232795672,
      "loss": 2.589,
      "step": 13798
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.595265805721283,
      "learning_rate": 0.0005947054581916458,
      "loss": 2.3096,
      "step": 13799
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6196656227111816,
      "learning_rate": 0.0005947046930489331,
      "loss": 2.605,
      "step": 13800
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6147734522819519,
      "learning_rate": 0.0005947039278514291,
      "loss": 2.3329,
      "step": 13801
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.7566469311714172,
      "learning_rate": 0.0005947031625991342,
      "loss": 2.4109,
      "step": 13802
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6356944441795349,
      "learning_rate": 0.0005947023972920482,
      "loss": 2.6864,
      "step": 13803
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.7836588621139526,
      "learning_rate": 0.0005947016319301717,
      "loss": 2.2546,
      "step": 13804
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6316655874252319,
      "learning_rate": 0.0005947008665135046,
      "loss": 2.536,
      "step": 13805
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6278917193412781,
      "learning_rate": 0.0005947001010420469,
      "loss": 2.1222,
      "step": 13806
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6575261354446411,
      "learning_rate": 0.0005946993355157989,
      "loss": 2.6738,
      "step": 13807
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6259804964065552,
      "learning_rate": 0.0005946985699347608,
      "loss": 2.4224,
      "step": 13808
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.7134105563163757,
      "learning_rate": 0.0005946978042989326,
      "loss": 2.5627,
      "step": 13809
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6371834874153137,
      "learning_rate": 0.0005946970386083146,
      "loss": 2.2785,
      "step": 13810
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.7290182113647461,
      "learning_rate": 0.0005946962728629069,
      "loss": 2.4806,
      "step": 13811
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6195067763328552,
      "learning_rate": 0.0005946955070627095,
      "loss": 2.3789,
      "step": 13812
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6038464307785034,
      "learning_rate": 0.0005946947412077226,
      "loss": 2.5293,
      "step": 13813
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.7809808850288391,
      "learning_rate": 0.0005946939752979464,
      "loss": 2.1775,
      "step": 13814
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.5553752183914185,
      "learning_rate": 0.0005946932093333812,
      "loss": 2.2844,
      "step": 13815
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6498607397079468,
      "learning_rate": 0.0005946924433140269,
      "loss": 2.6034,
      "step": 13816
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6301259398460388,
      "learning_rate": 0.0005946916772398836,
      "loss": 2.6044,
      "step": 13817
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6798137426376343,
      "learning_rate": 0.0005946909111109517,
      "loss": 2.5178,
      "step": 13818
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.70461106300354,
      "learning_rate": 0.0005946901449272311,
      "loss": 2.462,
      "step": 13819
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6557601094245911,
      "learning_rate": 0.000594689378688722,
      "loss": 2.3232,
      "step": 13820
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.968036413192749,
      "learning_rate": 0.0005946886123954247,
      "loss": 2.3942,
      "step": 13821
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.5803546905517578,
      "learning_rate": 0.0005946878460473391,
      "loss": 2.5032,
      "step": 13822
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6485296487808228,
      "learning_rate": 0.0005946870796444655,
      "loss": 2.3596,
      "step": 13823
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.66599041223526,
      "learning_rate": 0.000594686313186804,
      "loss": 2.4094,
      "step": 13824
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6381587982177734,
      "learning_rate": 0.0005946855466743548,
      "loss": 2.6772,
      "step": 13825
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.567030668258667,
      "learning_rate": 0.0005946847801071181,
      "loss": 2.4192,
      "step": 13826
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6452181339263916,
      "learning_rate": 0.0005946840134850937,
      "loss": 2.428,
      "step": 13827
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6916592717170715,
      "learning_rate": 0.0005946832468082822,
      "loss": 2.3835,
      "step": 13828
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6056642532348633,
      "learning_rate": 0.0005946824800766834,
      "loss": 2.7402,
      "step": 13829
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6368411183357239,
      "learning_rate": 0.0005946817132902976,
      "loss": 2.4107,
      "step": 13830
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6410033702850342,
      "learning_rate": 0.0005946809464491249,
      "loss": 2.4856,
      "step": 13831
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.558530867099762,
      "learning_rate": 0.0005946801795531654,
      "loss": 2.5215,
      "step": 13832
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6103267669677734,
      "learning_rate": 0.0005946794126024193,
      "loss": 2.4759,
      "step": 13833
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6482540369033813,
      "learning_rate": 0.0005946786455968868,
      "loss": 2.4692,
      "step": 13834
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6633281707763672,
      "learning_rate": 0.000594677878536568,
      "loss": 2.355,
      "step": 13835
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.698496401309967,
      "learning_rate": 0.0005946771114214629,
      "loss": 2.462,
      "step": 13836
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.5831876993179321,
      "learning_rate": 0.0005946763442515718,
      "loss": 2.3665,
      "step": 13837
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6670299768447876,
      "learning_rate": 0.0005946755770268949,
      "loss": 2.0994,
      "step": 13838
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6423171162605286,
      "learning_rate": 0.0005946748097474321,
      "loss": 2.3152,
      "step": 13839
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6006947755813599,
      "learning_rate": 0.0005946740424131839,
      "loss": 2.3935,
      "step": 13840
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6123360395431519,
      "learning_rate": 0.00059467327502415,
      "loss": 2.6871,
      "step": 13841
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6405594348907471,
      "learning_rate": 0.0005946725075803309,
      "loss": 2.3904,
      "step": 13842
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.5980573296546936,
      "learning_rate": 0.0005946717400817267,
      "loss": 2.5377,
      "step": 13843
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6558144092559814,
      "learning_rate": 0.0005946709725283372,
      "loss": 2.4013,
      "step": 13844
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.5955427289009094,
      "learning_rate": 0.000594670204920163,
      "loss": 2.6749,
      "step": 13845
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.5592686533927917,
      "learning_rate": 0.0005946694372572041,
      "loss": 2.4449,
      "step": 13846
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.5455209612846375,
      "learning_rate": 0.0005946686695394605,
      "loss": 2.1444,
      "step": 13847
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6541905403137207,
      "learning_rate": 0.0005946679017669324,
      "loss": 2.3073,
      "step": 13848
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6016272902488708,
      "learning_rate": 0.0005946671339396199,
      "loss": 2.4739,
      "step": 13849
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.866870641708374,
      "learning_rate": 0.0005946663660575234,
      "loss": 2.6109,
      "step": 13850
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.5974417924880981,
      "learning_rate": 0.0005946655981206428,
      "loss": 2.332,
      "step": 13851
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.5980930924415588,
      "learning_rate": 0.0005946648301289783,
      "loss": 2.3031,
      "step": 13852
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6133742928504944,
      "learning_rate": 0.00059466406208253,
      "loss": 2.3279,
      "step": 13853
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.5659340620040894,
      "learning_rate": 0.0005946632939812981,
      "loss": 2.3877,
      "step": 13854
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.5787131190299988,
      "learning_rate": 0.0005946625258252828,
      "loss": 2.3922,
      "step": 13855
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6989911794662476,
      "learning_rate": 0.0005946617576144842,
      "loss": 2.4057,
      "step": 13856
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6124586462974548,
      "learning_rate": 0.0005946609893489023,
      "loss": 2.4502,
      "step": 13857
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6476709246635437,
      "learning_rate": 0.0005946602210285374,
      "loss": 2.3263,
      "step": 13858
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.5765248537063599,
      "learning_rate": 0.0005946594526533896,
      "loss": 2.4325,
      "step": 13859
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6088507175445557,
      "learning_rate": 0.0005946586842234591,
      "loss": 2.5359,
      "step": 13860
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.5988410711288452,
      "learning_rate": 0.0005946579157387459,
      "loss": 2.3093,
      "step": 13861
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6404033899307251,
      "learning_rate": 0.0005946571471992502,
      "loss": 2.3654,
      "step": 13862
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6638894081115723,
      "learning_rate": 0.0005946563786049723,
      "loss": 2.5048,
      "step": 13863
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6259859204292297,
      "learning_rate": 0.0005946556099559122,
      "loss": 2.3944,
      "step": 13864
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6806290745735168,
      "learning_rate": 0.00059465484125207,
      "loss": 2.4809,
      "step": 13865
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.5913764834403992,
      "learning_rate": 0.0005946540724934459,
      "loss": 2.2333,
      "step": 13866
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6514862775802612,
      "learning_rate": 0.0005946533036800401,
      "loss": 2.4432,
      "step": 13867
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6175577640533447,
      "learning_rate": 0.0005946525348118527,
      "loss": 2.6251,
      "step": 13868
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.5713266134262085,
      "learning_rate": 0.0005946517658888837,
      "loss": 2.389,
      "step": 13869
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.5339353680610657,
      "learning_rate": 0.0005946509969111336,
      "loss": 2.1555,
      "step": 13870
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6606882810592651,
      "learning_rate": 0.0005946502278786021,
      "loss": 2.1598,
      "step": 13871
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6091047525405884,
      "learning_rate": 0.0005946494587912897,
      "loss": 2.4996,
      "step": 13872
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6618897914886475,
      "learning_rate": 0.0005946486896491964,
      "loss": 2.3495,
      "step": 13873
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.645963728427887,
      "learning_rate": 0.0005946479204523222,
      "loss": 2.4556,
      "step": 13874
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6099081635475159,
      "learning_rate": 0.0005946471512006675,
      "loss": 2.3894,
      "step": 13875
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6610605716705322,
      "learning_rate": 0.0005946463818942324,
      "loss": 2.2886,
      "step": 13876
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.5895524024963379,
      "learning_rate": 0.0005946456125330168,
      "loss": 2.3182,
      "step": 13877
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.5916256308555603,
      "learning_rate": 0.0005946448431170212,
      "loss": 2.4859,
      "step": 13878
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.5867120623588562,
      "learning_rate": 0.0005946440736462454,
      "loss": 2.5975,
      "step": 13879
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6075891852378845,
      "learning_rate": 0.0005946433041206897,
      "loss": 2.3526,
      "step": 13880
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6317872405052185,
      "learning_rate": 0.0005946425345403544,
      "loss": 2.768,
      "step": 13881
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.5765465497970581,
      "learning_rate": 0.0005946417649052394,
      "loss": 2.5423,
      "step": 13882
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6965852379798889,
      "learning_rate": 0.000594640995215345,
      "loss": 2.1372,
      "step": 13883
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6240060329437256,
      "learning_rate": 0.0005946402254706712,
      "loss": 2.3864,
      "step": 13884
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.749926745891571,
      "learning_rate": 0.0005946394556712183,
      "loss": 2.2652,
      "step": 13885
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6668254137039185,
      "learning_rate": 0.0005946386858169863,
      "loss": 2.528,
      "step": 13886
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.5512781739234924,
      "learning_rate": 0.0005946379159079754,
      "loss": 2.342,
      "step": 13887
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6211279630661011,
      "learning_rate": 0.0005946371459441857,
      "loss": 2.2756,
      "step": 13888
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6150392293930054,
      "learning_rate": 0.0005946363759256175,
      "loss": 2.2731,
      "step": 13889
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6482754349708557,
      "learning_rate": 0.0005946356058522708,
      "loss": 2.3763,
      "step": 13890
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.5710055232048035,
      "learning_rate": 0.0005946348357241458,
      "loss": 2.4539,
      "step": 13891
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6006652116775513,
      "learning_rate": 0.0005946340655412427,
      "loss": 2.3353,
      "step": 13892
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6340860724449158,
      "learning_rate": 0.0005946332953035615,
      "loss": 2.4284,
      "step": 13893
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.5861768126487732,
      "learning_rate": 0.0005946325250111023,
      "loss": 2.5535,
      "step": 13894
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.5966032147407532,
      "learning_rate": 0.0005946317546638654,
      "loss": 2.4098,
      "step": 13895
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6392712593078613,
      "learning_rate": 0.0005946309842618509,
      "loss": 2.4045,
      "step": 13896
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.7021943926811218,
      "learning_rate": 0.000594630213805059,
      "loss": 2.5011,
      "step": 13897
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6586278676986694,
      "learning_rate": 0.0005946294432934898,
      "loss": 2.4774,
      "step": 13898
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.7813237309455872,
      "learning_rate": 0.0005946286727271433,
      "loss": 2.4716,
      "step": 13899
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6744385361671448,
      "learning_rate": 0.0005946279021060199,
      "loss": 2.3463,
      "step": 13900
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6877837181091309,
      "learning_rate": 0.0005946271314301195,
      "loss": 2.5002,
      "step": 13901
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6666974425315857,
      "learning_rate": 0.0005946263606994423,
      "loss": 2.5017,
      "step": 13902
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.7332651615142822,
      "learning_rate": 0.0005946255899139887,
      "loss": 2.4226,
      "step": 13903
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.5809179544448853,
      "learning_rate": 0.0005946248190737586,
      "loss": 2.5545,
      "step": 13904
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.7363604307174683,
      "learning_rate": 0.0005946240481787521,
      "loss": 2.2302,
      "step": 13905
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.68861323595047,
      "learning_rate": 0.0005946232772289695,
      "loss": 2.5981,
      "step": 13906
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6341160535812378,
      "learning_rate": 0.0005946225062244109,
      "loss": 2.4039,
      "step": 13907
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.5901350975036621,
      "learning_rate": 0.0005946217351650762,
      "loss": 2.4473,
      "step": 13908
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6184715032577515,
      "learning_rate": 0.0005946209640509658,
      "loss": 2.6762,
      "step": 13909
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6710211634635925,
      "learning_rate": 0.00059462019288208,
      "loss": 2.3302,
      "step": 13910
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6211569905281067,
      "learning_rate": 0.0005946194216584187,
      "loss": 2.2133,
      "step": 13911
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6440826654434204,
      "learning_rate": 0.000594618650379982,
      "loss": 2.3211,
      "step": 13912
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.5955305695533752,
      "learning_rate": 0.0005946178790467701,
      "loss": 2.3961,
      "step": 13913
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6220107674598694,
      "learning_rate": 0.0005946171076587832,
      "loss": 2.4161,
      "step": 13914
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6618260145187378,
      "learning_rate": 0.0005946163362160216,
      "loss": 2.3712,
      "step": 13915
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.5747061967849731,
      "learning_rate": 0.0005946155647184851,
      "loss": 2.6256,
      "step": 13916
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6145161986351013,
      "learning_rate": 0.000594614793166174,
      "loss": 2.2418,
      "step": 13917
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.7157288193702698,
      "learning_rate": 0.0005946140215590885,
      "loss": 2.2174,
      "step": 13918
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6141568422317505,
      "learning_rate": 0.0005946132498972286,
      "loss": 2.4697,
      "step": 13919
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.7553056478500366,
      "learning_rate": 0.0005946124781805946,
      "loss": 2.3176,
      "step": 13920
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6483667492866516,
      "learning_rate": 0.0005946117064091867,
      "loss": 2.623,
      "step": 13921
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6451340317726135,
      "learning_rate": 0.0005946109345830047,
      "loss": 2.3803,
      "step": 13922
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.710851788520813,
      "learning_rate": 0.0005946101627020491,
      "loss": 2.4369,
      "step": 13923
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.5961554050445557,
      "learning_rate": 0.0005946093907663199,
      "loss": 2.436,
      "step": 13924
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6415092349052429,
      "learning_rate": 0.0005946086187758172,
      "loss": 2.423,
      "step": 13925
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6500274538993835,
      "learning_rate": 0.0005946078467305412,
      "loss": 2.3469,
      "step": 13926
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6249402761459351,
      "learning_rate": 0.0005946070746304921,
      "loss": 2.393,
      "step": 13927
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.5977994203567505,
      "learning_rate": 0.00059460630247567,
      "loss": 2.6056,
      "step": 13928
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6074367165565491,
      "learning_rate": 0.000594605530266075,
      "loss": 2.5194,
      "step": 13929
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6113567352294922,
      "learning_rate": 0.0005946047580017072,
      "loss": 2.3434,
      "step": 13930
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6120709180831909,
      "learning_rate": 0.0005946039856825669,
      "loss": 2.4682,
      "step": 13931
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6664702892303467,
      "learning_rate": 0.0005946032133086541,
      "loss": 2.2269,
      "step": 13932
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6616501808166504,
      "learning_rate": 0.000594602440879969,
      "loss": 2.632,
      "step": 13933
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.664647102355957,
      "learning_rate": 0.0005946016683965118,
      "loss": 2.5058,
      "step": 13934
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.5804508924484253,
      "learning_rate": 0.0005946008958582824,
      "loss": 2.4165,
      "step": 13935
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6432382464408875,
      "learning_rate": 0.0005946001232652815,
      "loss": 2.3427,
      "step": 13936
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6233971118927002,
      "learning_rate": 0.0005945993506175086,
      "loss": 2.5389,
      "step": 13937
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6143187880516052,
      "learning_rate": 0.0005945985779149641,
      "loss": 2.4956,
      "step": 13938
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6512585878372192,
      "learning_rate": 0.0005945978051576483,
      "loss": 2.406,
      "step": 13939
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6374672055244446,
      "learning_rate": 0.0005945970323455611,
      "loss": 2.4496,
      "step": 13940
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.5786138772964478,
      "learning_rate": 0.0005945962594787029,
      "loss": 2.4549,
      "step": 13941
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.59531170129776,
      "learning_rate": 0.0005945954865570736,
      "loss": 2.2941,
      "step": 13942
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6745465397834778,
      "learning_rate": 0.0005945947135806734,
      "loss": 2.545,
      "step": 13943
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.5604616403579712,
      "learning_rate": 0.0005945939405495025,
      "loss": 2.2367,
      "step": 13944
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6082635521888733,
      "learning_rate": 0.0005945931674635611,
      "loss": 2.3312,
      "step": 13945
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6474577188491821,
      "learning_rate": 0.0005945923943228492,
      "loss": 2.4956,
      "step": 13946
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6130780577659607,
      "learning_rate": 0.0005945916211273671,
      "loss": 2.2272,
      "step": 13947
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6464109420776367,
      "learning_rate": 0.0005945908478771147,
      "loss": 2.2748,
      "step": 13948
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.560920774936676,
      "learning_rate": 0.0005945900745720924,
      "loss": 2.2285,
      "step": 13949
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6180071234703064,
      "learning_rate": 0.0005945893012123002,
      "loss": 2.3516,
      "step": 13950
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.650277316570282,
      "learning_rate": 0.0005945885277977384,
      "loss": 2.4491,
      "step": 13951
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.5775216221809387,
      "learning_rate": 0.000594587754328407,
      "loss": 2.4943,
      "step": 13952
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6175071597099304,
      "learning_rate": 0.0005945869808043061,
      "loss": 2.2557,
      "step": 13953
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.5962519645690918,
      "learning_rate": 0.000594586207225436,
      "loss": 2.3496,
      "step": 13954
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6770408153533936,
      "learning_rate": 0.0005945854335917966,
      "loss": 2.1686,
      "step": 13955
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6098384261131287,
      "learning_rate": 0.0005945846599033883,
      "loss": 2.3854,
      "step": 13956
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.7187370657920837,
      "learning_rate": 0.0005945838861602113,
      "loss": 2.351,
      "step": 13957
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.8044708967208862,
      "learning_rate": 0.0005945831123622654,
      "loss": 2.572,
      "step": 13958
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.5872346758842468,
      "learning_rate": 0.000594582338509551,
      "loss": 2.4651,
      "step": 13959
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.8671247959136963,
      "learning_rate": 0.0005945815646020682,
      "loss": 2.5378,
      "step": 13960
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6452133655548096,
      "learning_rate": 0.0005945807906398171,
      "loss": 2.1155,
      "step": 13961
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.7943941950798035,
      "learning_rate": 0.0005945800166227978,
      "loss": 2.3101,
      "step": 13962
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.7551522850990295,
      "learning_rate": 0.0005945792425510107,
      "loss": 2.5105,
      "step": 13963
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.625163197517395,
      "learning_rate": 0.0005945784684244556,
      "loss": 2.3738,
      "step": 13964
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.7184911370277405,
      "learning_rate": 0.0005945776942431328,
      "loss": 2.2243,
      "step": 13965
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.7344866991043091,
      "learning_rate": 0.0005945769200070425,
      "loss": 2.4303,
      "step": 13966
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6258655190467834,
      "learning_rate": 0.0005945761457161849,
      "loss": 2.26,
      "step": 13967
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.7019901871681213,
      "learning_rate": 0.0005945753713705599,
      "loss": 2.3235,
      "step": 13968
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6027786135673523,
      "learning_rate": 0.0005945745969701678,
      "loss": 2.7018,
      "step": 13969
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.649158239364624,
      "learning_rate": 0.0005945738225150087,
      "loss": 2.6661,
      "step": 13970
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.7085934281349182,
      "learning_rate": 0.0005945730480050828,
      "loss": 2.319,
      "step": 13971
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6471622586250305,
      "learning_rate": 0.0005945722734403901,
      "loss": 2.3977,
      "step": 13972
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6116256713867188,
      "learning_rate": 0.0005945714988209311,
      "loss": 2.4186,
      "step": 13973
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.7378679513931274,
      "learning_rate": 0.0005945707241467055,
      "loss": 2.6479,
      "step": 13974
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.7040027976036072,
      "learning_rate": 0.0005945699494177137,
      "loss": 2.5997,
      "step": 13975
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.5873791575431824,
      "learning_rate": 0.0005945691746339558,
      "loss": 2.5157,
      "step": 13976
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6483922600746155,
      "learning_rate": 0.0005945683997954319,
      "loss": 2.4745,
      "step": 13977
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.650779128074646,
      "learning_rate": 0.0005945676249021421,
      "loss": 2.3972,
      "step": 13978
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6141162514686584,
      "learning_rate": 0.0005945668499540868,
      "loss": 2.4854,
      "step": 13979
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6363713145256042,
      "learning_rate": 0.0005945660749512658,
      "loss": 2.417,
      "step": 13980
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.7691216468811035,
      "learning_rate": 0.0005945652998936796,
      "loss": 2.5318,
      "step": 13981
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6090009212493896,
      "learning_rate": 0.0005945645247813279,
      "loss": 2.3091,
      "step": 13982
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6222363114356995,
      "learning_rate": 0.0005945637496142113,
      "loss": 2.2212,
      "step": 13983
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6796690225601196,
      "learning_rate": 0.0005945629743923296,
      "loss": 2.4536,
      "step": 13984
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.7904744148254395,
      "learning_rate": 0.0005945621991156831,
      "loss": 2.3691,
      "step": 13985
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.564169704914093,
      "learning_rate": 0.0005945614237842721,
      "loss": 2.5805,
      "step": 13986
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.7151418328285217,
      "learning_rate": 0.0005945606483980963,
      "loss": 2.5875,
      "step": 13987
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6793763041496277,
      "learning_rate": 0.0005945598729571563,
      "loss": 2.4242,
      "step": 13988
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6230502128601074,
      "learning_rate": 0.000594559097461452,
      "loss": 2.373,
      "step": 13989
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.8080693483352661,
      "learning_rate": 0.0005945583219109836,
      "loss": 2.4562,
      "step": 13990
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6204670071601868,
      "learning_rate": 0.0005945575463057513,
      "loss": 2.1912,
      "step": 13991
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6302924752235413,
      "learning_rate": 0.0005945567706457551,
      "loss": 2.2325,
      "step": 13992
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6397067308425903,
      "learning_rate": 0.0005945559949309953,
      "loss": 2.4488,
      "step": 13993
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6805353760719299,
      "learning_rate": 0.000594555219161472,
      "loss": 2.3339,
      "step": 13994
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6702293753623962,
      "learning_rate": 0.0005945544433371854,
      "loss": 2.5886,
      "step": 13995
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.5848265290260315,
      "learning_rate": 0.0005945536674581354,
      "loss": 2.4789,
      "step": 13996
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6135608553886414,
      "learning_rate": 0.0005945528915243223,
      "loss": 2.3252,
      "step": 13997
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.5922843813896179,
      "learning_rate": 0.0005945521155357464,
      "loss": 2.4909,
      "step": 13998
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6886267066001892,
      "learning_rate": 0.0005945513394924076,
      "loss": 2.4924,
      "step": 13999
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6712905168533325,
      "learning_rate": 0.0005945505633943063,
      "loss": 2.4634,
      "step": 14000
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.692379355430603,
      "learning_rate": 0.0005945497872414423,
      "loss": 2.2864,
      "step": 14001
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6089169979095459,
      "learning_rate": 0.0005945490110338159,
      "loss": 2.2891,
      "step": 14002
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6525683403015137,
      "learning_rate": 0.0005945482347714274,
      "loss": 2.3816,
      "step": 14003
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.663157045841217,
      "learning_rate": 0.0005945474584542769,
      "loss": 2.3941,
      "step": 14004
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6125277876853943,
      "learning_rate": 0.0005945466820823644,
      "loss": 2.326,
      "step": 14005
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.639917254447937,
      "learning_rate": 0.00059454590565569,
      "loss": 2.4941,
      "step": 14006
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6163619756698608,
      "learning_rate": 0.0005945451291742541,
      "loss": 2.5135,
      "step": 14007
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6420195698738098,
      "learning_rate": 0.0005945443526380567,
      "loss": 2.3322,
      "step": 14008
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6135909557342529,
      "learning_rate": 0.0005945435760470978,
      "loss": 2.3481,
      "step": 14009
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6273384094238281,
      "learning_rate": 0.0005945427994013778,
      "loss": 2.3798,
      "step": 14010
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.7166039943695068,
      "learning_rate": 0.0005945420227008967,
      "loss": 2.4456,
      "step": 14011
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6694671511650085,
      "learning_rate": 0.0005945412459456547,
      "loss": 2.2002,
      "step": 14012
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6182705760002136,
      "learning_rate": 0.0005945404691356519,
      "loss": 2.2797,
      "step": 14013
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6969395875930786,
      "learning_rate": 0.0005945396922708884,
      "loss": 2.5178,
      "step": 14014
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6837332248687744,
      "learning_rate": 0.0005945389153513646,
      "loss": 2.3432,
      "step": 14015
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.59559565782547,
      "learning_rate": 0.0005945381383770803,
      "loss": 2.6418,
      "step": 14016
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6303397417068481,
      "learning_rate": 0.0005945373613480359,
      "loss": 2.4468,
      "step": 14017
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6245408654212952,
      "learning_rate": 0.0005945365842642314,
      "loss": 2.3466,
      "step": 14018
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6707871556282043,
      "learning_rate": 0.000594535807125667,
      "loss": 2.4219,
      "step": 14019
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.7457634210586548,
      "learning_rate": 0.0005945350299323427,
      "loss": 2.6186,
      "step": 14020
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.5988614559173584,
      "learning_rate": 0.000594534252684259,
      "loss": 2.5402,
      "step": 14021
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6099734902381897,
      "learning_rate": 0.0005945334753814157,
      "loss": 2.4627,
      "step": 14022
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.5931496620178223,
      "learning_rate": 0.0005945326980238131,
      "loss": 2.3783,
      "step": 14023
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.681315541267395,
      "learning_rate": 0.0005945319206114513,
      "loss": 2.4281,
      "step": 14024
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.5753620266914368,
      "learning_rate": 0.0005945311431443304,
      "loss": 2.1375,
      "step": 14025
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6441396474838257,
      "learning_rate": 0.0005945303656224506,
      "loss": 2.2906,
      "step": 14026
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.5744447708129883,
      "learning_rate": 0.0005945295880458121,
      "loss": 2.4499,
      "step": 14027
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6571962237358093,
      "learning_rate": 0.000594528810414415,
      "loss": 2.4439,
      "step": 14028
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6081585884094238,
      "learning_rate": 0.0005945280327282595,
      "loss": 2.2727,
      "step": 14029
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.5787054896354675,
      "learning_rate": 0.0005945272549873456,
      "loss": 2.2683,
      "step": 14030
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.5823553204536438,
      "learning_rate": 0.0005945264771916735,
      "loss": 2.2182,
      "step": 14031
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.7052894830703735,
      "learning_rate": 0.0005945256993412434,
      "loss": 2.4306,
      "step": 14032
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6311241984367371,
      "learning_rate": 0.0005945249214360554,
      "loss": 2.5448,
      "step": 14033
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6328207850456238,
      "learning_rate": 0.0005945241434761096,
      "loss": 2.4026,
      "step": 14034
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.644869863986969,
      "learning_rate": 0.0005945233654614063,
      "loss": 2.5293,
      "step": 14035
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6318793296813965,
      "learning_rate": 0.0005945225873919455,
      "loss": 2.5749,
      "step": 14036
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6176579594612122,
      "learning_rate": 0.0005945218092677274,
      "loss": 2.3899,
      "step": 14037
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.7644437551498413,
      "learning_rate": 0.0005945210310887521,
      "loss": 2.4442,
      "step": 14038
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6035483479499817,
      "learning_rate": 0.0005945202528550199,
      "loss": 2.4358,
      "step": 14039
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.5959306955337524,
      "learning_rate": 0.0005945194745665308,
      "loss": 2.4548,
      "step": 14040
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6324282288551331,
      "learning_rate": 0.0005945186962232849,
      "loss": 2.7253,
      "step": 14041
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.7574411034584045,
      "learning_rate": 0.0005945179178252824,
      "loss": 2.3103,
      "step": 14042
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6452775597572327,
      "learning_rate": 0.0005945171393725234,
      "loss": 2.368,
      "step": 14043
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.7294325828552246,
      "learning_rate": 0.0005945163608650082,
      "loss": 2.4747,
      "step": 14044
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.7456701397895813,
      "learning_rate": 0.000594515582302737,
      "loss": 2.2661,
      "step": 14045
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6734309196472168,
      "learning_rate": 0.0005945148036857095,
      "loss": 2.5208,
      "step": 14046
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.7408337593078613,
      "learning_rate": 0.0005945140250139263,
      "loss": 2.5826,
      "step": 14047
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.7658661603927612,
      "learning_rate": 0.0005945132462873874,
      "loss": 2.4709,
      "step": 14048
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.5722621083259583,
      "learning_rate": 0.0005945124675060929,
      "loss": 2.3917,
      "step": 14049
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6747748255729675,
      "learning_rate": 0.0005945116886700429,
      "loss": 2.2809,
      "step": 14050
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6061657071113586,
      "learning_rate": 0.0005945109097792376,
      "loss": 2.3248,
      "step": 14051
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6794477701187134,
      "learning_rate": 0.0005945101308336774,
      "loss": 2.2712,
      "step": 14052
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.711236298084259,
      "learning_rate": 0.000594509351833362,
      "loss": 2.4932,
      "step": 14053
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.7397414445877075,
      "learning_rate": 0.0005945085727782918,
      "loss": 2.5746,
      "step": 14054
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6850473880767822,
      "learning_rate": 0.0005945077936684669,
      "loss": 2.5361,
      "step": 14055
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.572394609451294,
      "learning_rate": 0.0005945070145038874,
      "loss": 2.0129,
      "step": 14056
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6079181432723999,
      "learning_rate": 0.0005945062352845535,
      "loss": 2.4857,
      "step": 14057
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.5804174542427063,
      "learning_rate": 0.0005945054560104654,
      "loss": 2.5137,
      "step": 14058
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6479015946388245,
      "learning_rate": 0.0005945046766816232,
      "loss": 2.4945,
      "step": 14059
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.7264747023582458,
      "learning_rate": 0.0005945038972980269,
      "loss": 2.6184,
      "step": 14060
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.684536337852478,
      "learning_rate": 0.0005945031178596769,
      "loss": 2.5406,
      "step": 14061
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6780819296836853,
      "learning_rate": 0.000594502338366573,
      "loss": 2.4878,
      "step": 14062
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6239978671073914,
      "learning_rate": 0.0005945015588187157,
      "loss": 2.3621,
      "step": 14063
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6446120738983154,
      "learning_rate": 0.000594500779216105,
      "loss": 2.4923,
      "step": 14064
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6134488582611084,
      "learning_rate": 0.0005944999995587411,
      "loss": 2.4822,
      "step": 14065
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6594745516777039,
      "learning_rate": 0.0005944992198466239,
      "loss": 2.3097,
      "step": 14066
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.627303421497345,
      "learning_rate": 0.0005944984400797539,
      "loss": 2.279,
      "step": 14067
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6141054630279541,
      "learning_rate": 0.0005944976602581311,
      "loss": 2.3536,
      "step": 14068
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.5691677331924438,
      "learning_rate": 0.0005944968803817556,
      "loss": 2.5023,
      "step": 14069
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6114333868026733,
      "learning_rate": 0.0005944961004506275,
      "loss": 2.4351,
      "step": 14070
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6382431387901306,
      "learning_rate": 0.0005944953204647471,
      "loss": 2.3298,
      "step": 14071
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.5925363302230835,
      "learning_rate": 0.0005944945404241144,
      "loss": 2.4345,
      "step": 14072
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6049261689186096,
      "learning_rate": 0.0005944937603287296,
      "loss": 2.5208,
      "step": 14073
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6179981231689453,
      "learning_rate": 0.0005944929801785929,
      "loss": 2.615,
      "step": 14074
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6273331642150879,
      "learning_rate": 0.0005944921999737043,
      "loss": 2.4596,
      "step": 14075
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6281945109367371,
      "learning_rate": 0.0005944914197140642,
      "loss": 2.4131,
      "step": 14076
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6143537163734436,
      "learning_rate": 0.0005944906393996725,
      "loss": 2.2875,
      "step": 14077
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6145966053009033,
      "learning_rate": 0.0005944898590305295,
      "loss": 2.3941,
      "step": 14078
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6593654751777649,
      "learning_rate": 0.0005944890786066351,
      "loss": 2.4426,
      "step": 14079
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.5734072923660278,
      "learning_rate": 0.0005944882981279897,
      "loss": 2.3897,
      "step": 14080
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.5834250450134277,
      "learning_rate": 0.0005944875175945935,
      "loss": 2.4672,
      "step": 14081
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.7135958075523376,
      "learning_rate": 0.0005944867370064463,
      "loss": 2.3755,
      "step": 14082
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6186698079109192,
      "learning_rate": 0.0005944859563635486,
      "loss": 2.301,
      "step": 14083
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6211986541748047,
      "learning_rate": 0.0005944851756659004,
      "loss": 2.5494,
      "step": 14084
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6644830107688904,
      "learning_rate": 0.0005944843949135018,
      "loss": 2.6383,
      "step": 14085
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.7578418850898743,
      "learning_rate": 0.000594483614106353,
      "loss": 2.389,
      "step": 14086
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.1079902648925781,
      "learning_rate": 0.0005944828332444541,
      "loss": 2.353,
      "step": 14087
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.5991947054862976,
      "learning_rate": 0.0005944820523278053,
      "loss": 2.5111,
      "step": 14088
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6966270208358765,
      "learning_rate": 0.0005944812713564069,
      "loss": 2.6105,
      "step": 14089
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.5653650164604187,
      "learning_rate": 0.0005944804903302586,
      "loss": 2.1236,
      "step": 14090
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.5987493991851807,
      "learning_rate": 0.000594479709249361,
      "loss": 2.3017,
      "step": 14091
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6035085916519165,
      "learning_rate": 0.000594478928113714,
      "loss": 2.3662,
      "step": 14092
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6668792963027954,
      "learning_rate": 0.0005944781469233178,
      "loss": 2.1343,
      "step": 14093
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.5691353678703308,
      "learning_rate": 0.0005944773656781726,
      "loss": 2.3645,
      "step": 14094
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6775779724121094,
      "learning_rate": 0.0005944765843782784,
      "loss": 2.6189,
      "step": 14095
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.5794703960418701,
      "learning_rate": 0.0005944758030236356,
      "loss": 2.5461,
      "step": 14096
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6760302186012268,
      "learning_rate": 0.0005944750216142439,
      "loss": 2.2442,
      "step": 14097
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6476757526397705,
      "learning_rate": 0.000594474240150104,
      "loss": 2.6203,
      "step": 14098
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.5839778184890747,
      "learning_rate": 0.0005944734586312158,
      "loss": 2.4656,
      "step": 14099
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6301483511924744,
      "learning_rate": 0.0005944726770575793,
      "loss": 2.6005,
      "step": 14100
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6548895835876465,
      "learning_rate": 0.0005944718954291948,
      "loss": 2.4363,
      "step": 14101
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.5971547961235046,
      "learning_rate": 0.0005944711137460624,
      "loss": 2.5147,
      "step": 14102
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6800278425216675,
      "learning_rate": 0.0005944703320081824,
      "loss": 2.3374,
      "step": 14103
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6885166764259338,
      "learning_rate": 0.0005944695502155547,
      "loss": 2.6097,
      "step": 14104
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6913202404975891,
      "learning_rate": 0.0005944687683681794,
      "loss": 2.3125,
      "step": 14105
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.625540554523468,
      "learning_rate": 0.0005944679864660571,
      "loss": 2.2404,
      "step": 14106
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6505255103111267,
      "learning_rate": 0.0005944672045091874,
      "loss": 2.5825,
      "step": 14107
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6040834188461304,
      "learning_rate": 0.0005944664224975708,
      "loss": 2.3234,
      "step": 14108
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6259313225746155,
      "learning_rate": 0.0005944656404312073,
      "loss": 2.6141,
      "step": 14109
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.5818769931793213,
      "learning_rate": 0.0005944648583100971,
      "loss": 2.5936,
      "step": 14110
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.5763487219810486,
      "learning_rate": 0.0005944640761342402,
      "loss": 2.666,
      "step": 14111
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.5745701193809509,
      "learning_rate": 0.000594463293903637,
      "loss": 2.4269,
      "step": 14112
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.589826762676239,
      "learning_rate": 0.0005944625116182876,
      "loss": 2.3628,
      "step": 14113
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.5715101361274719,
      "learning_rate": 0.0005944617292781919,
      "loss": 2.4967,
      "step": 14114
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.636907696723938,
      "learning_rate": 0.0005944609468833503,
      "loss": 2.3134,
      "step": 14115
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6336680054664612,
      "learning_rate": 0.0005944601644337627,
      "loss": 2.3874,
      "step": 14116
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.5870243906974792,
      "learning_rate": 0.0005944593819294297,
      "loss": 2.4523,
      "step": 14117
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.5982416272163391,
      "learning_rate": 0.0005944585993703508,
      "loss": 2.2736,
      "step": 14118
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6712840795516968,
      "learning_rate": 0.0005944578167565267,
      "loss": 2.4896,
      "step": 14119
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6344760060310364,
      "learning_rate": 0.0005944570340879572,
      "loss": 2.3755,
      "step": 14120
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.5895719528198242,
      "learning_rate": 0.0005944562513646428,
      "loss": 2.3216,
      "step": 14121
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6375544667243958,
      "learning_rate": 0.0005944554685865833,
      "loss": 2.3153,
      "step": 14122
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6194722056388855,
      "learning_rate": 0.0005944546857537789,
      "loss": 2.2957,
      "step": 14123
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6437780261039734,
      "learning_rate": 0.0005944539028662299,
      "loss": 2.4573,
      "step": 14124
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.656864583492279,
      "learning_rate": 0.0005944531199239363,
      "loss": 2.4063,
      "step": 14125
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6480685472488403,
      "learning_rate": 0.0005944523369268984,
      "loss": 2.238,
      "step": 14126
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.7120686769485474,
      "learning_rate": 0.0005944515538751161,
      "loss": 2.3365,
      "step": 14127
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6076106429100037,
      "learning_rate": 0.0005944507707685897,
      "loss": 2.3529,
      "step": 14128
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6505674719810486,
      "learning_rate": 0.0005944499876073194,
      "loss": 2.3287,
      "step": 14129
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.646989107131958,
      "learning_rate": 0.0005944492043913055,
      "loss": 2.6578,
      "step": 14130
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6120565533638,
      "learning_rate": 0.0005944484211205477,
      "loss": 2.1648,
      "step": 14131
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6195019483566284,
      "learning_rate": 0.0005944476377950465,
      "loss": 2.6058,
      "step": 14132
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6509648561477661,
      "learning_rate": 0.0005944468544148018,
      "loss": 2.5228,
      "step": 14133
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6248160600662231,
      "learning_rate": 0.0005944460709798139,
      "loss": 2.502,
      "step": 14134
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.638766348361969,
      "learning_rate": 0.000594445287490083,
      "loss": 2.3228,
      "step": 14135
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6384394764900208,
      "learning_rate": 0.0005944445039456091,
      "loss": 2.4468,
      "step": 14136
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.5554043054580688,
      "learning_rate": 0.0005944437203463924,
      "loss": 2.3826,
      "step": 14137
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.563809335231781,
      "learning_rate": 0.0005944429366924331,
      "loss": 2.5262,
      "step": 14138
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6431106328964233,
      "learning_rate": 0.0005944421529837312,
      "loss": 2.4358,
      "step": 14139
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.699900209903717,
      "learning_rate": 0.0005944413692202871,
      "loss": 2.2842,
      "step": 14140
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6170838475227356,
      "learning_rate": 0.0005944405854021007,
      "loss": 2.5904,
      "step": 14141
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6601865291595459,
      "learning_rate": 0.0005944398015291722,
      "loss": 2.5541,
      "step": 14142
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6068218946456909,
      "learning_rate": 0.0005944390176015018,
      "loss": 2.4165,
      "step": 14143
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6098437309265137,
      "learning_rate": 0.0005944382336190897,
      "loss": 2.1906,
      "step": 14144
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6048662662506104,
      "learning_rate": 0.000594437449581936,
      "loss": 2.4776,
      "step": 14145
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6590048670768738,
      "learning_rate": 0.0005944366654900406,
      "loss": 2.4285,
      "step": 14146
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6214314103126526,
      "learning_rate": 0.0005944358813434041,
      "loss": 2.5185,
      "step": 14147
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.8197100758552551,
      "learning_rate": 0.0005944350971420263,
      "loss": 2.4839,
      "step": 14148
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.658409595489502,
      "learning_rate": 0.0005944343128859075,
      "loss": 2.3255,
      "step": 14149
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.5970758199691772,
      "learning_rate": 0.0005944335285750479,
      "loss": 2.3368,
      "step": 14150
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6274685859680176,
      "learning_rate": 0.0005944327442094474,
      "loss": 2.2164,
      "step": 14151
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6962413191795349,
      "learning_rate": 0.0005944319597891063,
      "loss": 2.1796,
      "step": 14152
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.611425518989563,
      "learning_rate": 0.0005944311753140248,
      "loss": 2.355,
      "step": 14153
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.5862756371498108,
      "learning_rate": 0.0005944303907842029,
      "loss": 2.7303,
      "step": 14154
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.5771706104278564,
      "learning_rate": 0.000594429606199641,
      "loss": 2.4248,
      "step": 14155
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6261806488037109,
      "learning_rate": 0.0005944288215603389,
      "loss": 2.4824,
      "step": 14156
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6355408430099487,
      "learning_rate": 0.000594428036866297,
      "loss": 2.5468,
      "step": 14157
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6117433309555054,
      "learning_rate": 0.0005944272521175154,
      "loss": 2.371,
      "step": 14158
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.5990714430809021,
      "learning_rate": 0.0005944264673139941,
      "loss": 2.6207,
      "step": 14159
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6761159896850586,
      "learning_rate": 0.0005944256824557335,
      "loss": 2.272,
      "step": 14160
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6207965016365051,
      "learning_rate": 0.0005944248975427336,
      "loss": 2.4833,
      "step": 14161
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6429013013839722,
      "learning_rate": 0.0005944241125749944,
      "loss": 2.6468,
      "step": 14162
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.5808757543563843,
      "learning_rate": 0.0005944233275525163,
      "loss": 2.4657,
      "step": 14163
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.5917670130729675,
      "learning_rate": 0.0005944225424752994,
      "loss": 2.5145,
      "step": 14164
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6137210726737976,
      "learning_rate": 0.0005944217573433438,
      "loss": 2.5379,
      "step": 14165
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.616672158241272,
      "learning_rate": 0.0005944209721566496,
      "loss": 2.2797,
      "step": 14166
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6054398417472839,
      "learning_rate": 0.0005944201869152169,
      "loss": 2.5904,
      "step": 14167
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6018138527870178,
      "learning_rate": 0.0005944194016190461,
      "loss": 2.6095,
      "step": 14168
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6057834029197693,
      "learning_rate": 0.000594418616268137,
      "loss": 2.2727,
      "step": 14169
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.5614697933197021,
      "learning_rate": 0.00059441783086249,
      "loss": 2.4439,
      "step": 14170
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.5919164419174194,
      "learning_rate": 0.0005944170454021051,
      "loss": 2.4779,
      "step": 14171
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6061266660690308,
      "learning_rate": 0.0005944162598869827,
      "loss": 2.3728,
      "step": 14172
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6346363425254822,
      "learning_rate": 0.0005944154743171226,
      "loss": 2.6802,
      "step": 14173
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6568766236305237,
      "learning_rate": 0.0005944146886925251,
      "loss": 2.3751,
      "step": 14174
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6023133397102356,
      "learning_rate": 0.0005944139030131904,
      "loss": 2.3775,
      "step": 14175
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6400668025016785,
      "learning_rate": 0.0005944131172791186,
      "loss": 2.3608,
      "step": 14176
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.5978460311889648,
      "learning_rate": 0.0005944123314903098,
      "loss": 2.4022,
      "step": 14177
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.7110781073570251,
      "learning_rate": 0.0005944115456467643,
      "loss": 2.3941,
      "step": 14178
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.5777397155761719,
      "learning_rate": 0.000594410759748482,
      "loss": 2.4729,
      "step": 14179
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.582852840423584,
      "learning_rate": 0.0005944099737954633,
      "loss": 2.5413,
      "step": 14180
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6528147459030151,
      "learning_rate": 0.0005944091877877082,
      "loss": 2.4377,
      "step": 14181
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.5859209299087524,
      "learning_rate": 0.0005944084017252167,
      "loss": 2.3295,
      "step": 14182
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.7265684008598328,
      "learning_rate": 0.0005944076156079893,
      "loss": 2.3507,
      "step": 14183
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.621486485004425,
      "learning_rate": 0.000594406829436026,
      "loss": 2.3155,
      "step": 14184
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.5115605592727661,
      "learning_rate": 0.0005944060432093269,
      "loss": 2.3312,
      "step": 14185
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6179931163787842,
      "learning_rate": 0.000594405256927892,
      "loss": 2.4012,
      "step": 14186
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6793397068977356,
      "learning_rate": 0.0005944044705917218,
      "loss": 2.5293,
      "step": 14187
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.565040111541748,
      "learning_rate": 0.0005944036842008162,
      "loss": 2.4404,
      "step": 14188
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6090429425239563,
      "learning_rate": 0.0005944028977551753,
      "loss": 2.6031,
      "step": 14189
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.5869806408882141,
      "learning_rate": 0.0005944021112547994,
      "loss": 2.3693,
      "step": 14190
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.626804769039154,
      "learning_rate": 0.0005944013246996886,
      "loss": 2.3234,
      "step": 14191
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6077307462692261,
      "learning_rate": 0.000594400538089843,
      "loss": 2.337,
      "step": 14192
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.5900352597236633,
      "learning_rate": 0.0005943997514252628,
      "loss": 2.4499,
      "step": 14193
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6443743705749512,
      "learning_rate": 0.0005943989647059483,
      "loss": 2.4583,
      "step": 14194
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6282657980918884,
      "learning_rate": 0.0005943981779318992,
      "loss": 2.447,
      "step": 14195
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6250261068344116,
      "learning_rate": 0.000594397391103116,
      "loss": 2.6491,
      "step": 14196
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.7645854353904724,
      "learning_rate": 0.0005943966042195989,
      "loss": 2.2885,
      "step": 14197
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6164730787277222,
      "learning_rate": 0.0005943958172813478,
      "loss": 2.4079,
      "step": 14198
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6217644214630127,
      "learning_rate": 0.000594395030288363,
      "loss": 2.5486,
      "step": 14199
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.647982120513916,
      "learning_rate": 0.0005943942432406446,
      "loss": 2.4816,
      "step": 14200
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6183875799179077,
      "learning_rate": 0.0005943934561381927,
      "loss": 2.2905,
      "step": 14201
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.643254280090332,
      "learning_rate": 0.0005943926689810075,
      "loss": 2.2567,
      "step": 14202
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6300073266029358,
      "learning_rate": 0.0005943918817690893,
      "loss": 2.2951,
      "step": 14203
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6347557902336121,
      "learning_rate": 0.0005943910945024379,
      "loss": 2.6637,
      "step": 14204
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.5603340864181519,
      "learning_rate": 0.0005943903071810538,
      "loss": 2.4232,
      "step": 14205
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.5598201751708984,
      "learning_rate": 0.0005943895198049369,
      "loss": 2.1787,
      "step": 14206
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.5835832953453064,
      "learning_rate": 0.0005943887323740874,
      "loss": 2.2854,
      "step": 14207
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6726608276367188,
      "learning_rate": 0.0005943879448885055,
      "loss": 2.274,
      "step": 14208
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6462631821632385,
      "learning_rate": 0.0005943871573481914,
      "loss": 2.3503,
      "step": 14209
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.5612764954566956,
      "learning_rate": 0.000594386369753145,
      "loss": 2.411,
      "step": 14210
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6197121143341064,
      "learning_rate": 0.0005943855821033668,
      "loss": 2.1736,
      "step": 14211
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6118569374084473,
      "learning_rate": 0.0005943847943988566,
      "loss": 2.3118,
      "step": 14212
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6074196696281433,
      "learning_rate": 0.0005943840066396148,
      "loss": 2.5074,
      "step": 14213
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6526360511779785,
      "learning_rate": 0.0005943832188256415,
      "loss": 2.4323,
      "step": 14214
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.7786007523536682,
      "learning_rate": 0.0005943824309569368,
      "loss": 2.5073,
      "step": 14215
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.7011995911598206,
      "learning_rate": 0.0005943816430335007,
      "loss": 2.4217,
      "step": 14216
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6912050843238831,
      "learning_rate": 0.0005943808550553337,
      "loss": 2.38,
      "step": 14217
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.5891363024711609,
      "learning_rate": 0.0005943800670224356,
      "loss": 2.3188,
      "step": 14218
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.7113338708877563,
      "learning_rate": 0.0005943792789348067,
      "loss": 2.4494,
      "step": 14219
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6518826484680176,
      "learning_rate": 0.0005943784907924473,
      "loss": 2.2897,
      "step": 14220
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.5986726880073547,
      "learning_rate": 0.0005943777025953571,
      "loss": 2.4252,
      "step": 14221
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6253679394721985,
      "learning_rate": 0.0005943769143435367,
      "loss": 2.346,
      "step": 14222
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6298472285270691,
      "learning_rate": 0.000594376126036986,
      "loss": 2.3797,
      "step": 14223
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.5705521702766418,
      "learning_rate": 0.0005943753376757054,
      "loss": 2.0798,
      "step": 14224
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6595010161399841,
      "learning_rate": 0.0005943745492596946,
      "loss": 2.6516,
      "step": 14225
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6305853128433228,
      "learning_rate": 0.0005943737607889542,
      "loss": 2.1506,
      "step": 14226
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.8512017726898193,
      "learning_rate": 0.0005943729722634842,
      "loss": 2.3192,
      "step": 14227
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6366754770278931,
      "learning_rate": 0.0005943721836832845,
      "loss": 2.4217,
      "step": 14228
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.7786884903907776,
      "learning_rate": 0.0005943713950483557,
      "loss": 2.4521,
      "step": 14229
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.8492106795310974,
      "learning_rate": 0.0005943706063586975,
      "loss": 2.3156,
      "step": 14230
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.7767971754074097,
      "learning_rate": 0.0005943698176143103,
      "loss": 2.2182,
      "step": 14231
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.8593736886978149,
      "learning_rate": 0.0005943690288151943,
      "loss": 2.4035,
      "step": 14232
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.615565299987793,
      "learning_rate": 0.0005943682399613495,
      "loss": 2.2309,
      "step": 14233
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.7128530740737915,
      "learning_rate": 0.000594367451052776,
      "loss": 2.4257,
      "step": 14234
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.9032661318778992,
      "learning_rate": 0.0005943666620894741,
      "loss": 2.1914,
      "step": 14235
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6282352209091187,
      "learning_rate": 0.0005943658730714438,
      "loss": 2.3641,
      "step": 14236
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6356613636016846,
      "learning_rate": 0.0005943650839986855,
      "loss": 2.4278,
      "step": 14237
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.7359389662742615,
      "learning_rate": 0.0005943642948711991,
      "loss": 2.3394,
      "step": 14238
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6885504722595215,
      "learning_rate": 0.0005943635056889847,
      "loss": 2.3728,
      "step": 14239
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6408309936523438,
      "learning_rate": 0.0005943627164520426,
      "loss": 2.4955,
      "step": 14240
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.8173355460166931,
      "learning_rate": 0.0005943619271603731,
      "loss": 2.4733,
      "step": 14241
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.7146652340888977,
      "learning_rate": 0.000594361137813976,
      "loss": 2.4075,
      "step": 14242
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6546050906181335,
      "learning_rate": 0.0005943603484128517,
      "loss": 2.2725,
      "step": 14243
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6478679180145264,
      "learning_rate": 0.0005943595589570001,
      "loss": 2.4422,
      "step": 14244
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.5764312744140625,
      "learning_rate": 0.0005943587694464216,
      "loss": 2.4724,
      "step": 14245
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.5408453345298767,
      "learning_rate": 0.0005943579798811162,
      "loss": 2.3124,
      "step": 14246
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6551764607429504,
      "learning_rate": 0.0005943571902610842,
      "loss": 2.3376,
      "step": 14247
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6176984906196594,
      "learning_rate": 0.0005943564005863256,
      "loss": 2.3497,
      "step": 14248
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6765878796577454,
      "learning_rate": 0.0005943556108568406,
      "loss": 2.4546,
      "step": 14249
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6204997301101685,
      "learning_rate": 0.0005943548210726292,
      "loss": 2.5229,
      "step": 14250
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.5767655968666077,
      "learning_rate": 0.0005943540312336919,
      "loss": 2.3487,
      "step": 14251
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.617072343826294,
      "learning_rate": 0.0005943532413400285,
      "loss": 2.2572,
      "step": 14252
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6079607009887695,
      "learning_rate": 0.0005943524513916393,
      "loss": 2.4661,
      "step": 14253
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6919021010398865,
      "learning_rate": 0.0005943516613885244,
      "loss": 2.2866,
      "step": 14254
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.5874783396720886,
      "learning_rate": 0.0005943508713306839,
      "loss": 2.3862,
      "step": 14255
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.64698725938797,
      "learning_rate": 0.0005943500812181182,
      "loss": 2.4246,
      "step": 14256
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.627199113368988,
      "learning_rate": 0.0005943492910508271,
      "loss": 2.2551,
      "step": 14257
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.5952585935592651,
      "learning_rate": 0.000594348500828811,
      "loss": 2.0355,
      "step": 14258
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.583289623260498,
      "learning_rate": 0.0005943477105520699,
      "loss": 2.3869,
      "step": 14259
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6130359172821045,
      "learning_rate": 0.0005943469202206041,
      "loss": 2.5638,
      "step": 14260
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6015077233314514,
      "learning_rate": 0.0005943461298344136,
      "loss": 2.4257,
      "step": 14261
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6105605363845825,
      "learning_rate": 0.0005943453393934986,
      "loss": 2.4439,
      "step": 14262
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.5833396911621094,
      "learning_rate": 0.0005943445488978593,
      "loss": 2.5695,
      "step": 14263
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6512017250061035,
      "learning_rate": 0.0005943437583474957,
      "loss": 2.489,
      "step": 14264
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.7168059945106506,
      "learning_rate": 0.000594342967742408,
      "loss": 2.6881,
      "step": 14265
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6613391637802124,
      "learning_rate": 0.0005943421770825964,
      "loss": 2.2857,
      "step": 14266
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.7061359882354736,
      "learning_rate": 0.0005943413863680611,
      "loss": 2.2674,
      "step": 14267
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6146553158760071,
      "learning_rate": 0.0005943405955988022,
      "loss": 2.4551,
      "step": 14268
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6114497780799866,
      "learning_rate": 0.0005943398047748197,
      "loss": 2.5399,
      "step": 14269
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6599135994911194,
      "learning_rate": 0.000594339013896114,
      "loss": 2.5756,
      "step": 14270
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6262415647506714,
      "learning_rate": 0.0005943382229626851,
      "loss": 2.4293,
      "step": 14271
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.7790961861610413,
      "learning_rate": 0.0005943374319745332,
      "loss": 2.1442,
      "step": 14272
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.5868929624557495,
      "learning_rate": 0.0005943366409316583,
      "loss": 2.3975,
      "step": 14273
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.5835980176925659,
      "learning_rate": 0.0005943358498340607,
      "loss": 2.4031,
      "step": 14274
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6168310642242432,
      "learning_rate": 0.0005943350586817407,
      "loss": 2.6575,
      "step": 14275
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6120644807815552,
      "learning_rate": 0.0005943342674746981,
      "loss": 2.5094,
      "step": 14276
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6358915567398071,
      "learning_rate": 0.0005943334762129332,
      "loss": 2.4613,
      "step": 14277
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.7173973917961121,
      "learning_rate": 0.000594332684896446,
      "loss": 2.2516,
      "step": 14278
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.7180003523826599,
      "learning_rate": 0.000594331893525237,
      "loss": 2.4629,
      "step": 14279
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6414711475372314,
      "learning_rate": 0.0005943311020993061,
      "loss": 2.3864,
      "step": 14280
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.82486492395401,
      "learning_rate": 0.0005943303106186535,
      "loss": 2.5519,
      "step": 14281
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6047924757003784,
      "learning_rate": 0.0005943295190832794,
      "loss": 2.5291,
      "step": 14282
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.678004264831543,
      "learning_rate": 0.0005943287274931838,
      "loss": 2.294,
      "step": 14283
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.7045817971229553,
      "learning_rate": 0.0005943279358483669,
      "loss": 2.4819,
      "step": 14284
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6079593300819397,
      "learning_rate": 0.000594327144148829,
      "loss": 2.252,
      "step": 14285
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.7156205773353577,
      "learning_rate": 0.00059432635239457,
      "loss": 2.4328,
      "step": 14286
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6160838007926941,
      "learning_rate": 0.0005943255605855902,
      "loss": 2.53,
      "step": 14287
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6577841639518738,
      "learning_rate": 0.0005943247687218898,
      "loss": 2.3058,
      "step": 14288
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.542777955532074,
      "learning_rate": 0.0005943239768034688,
      "loss": 2.3143,
      "step": 14289
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6051889061927795,
      "learning_rate": 0.0005943231848303275,
      "loss": 2.3771,
      "step": 14290
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6605547070503235,
      "learning_rate": 0.0005943223928024659,
      "loss": 2.372,
      "step": 14291
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6331718564033508,
      "learning_rate": 0.0005943216007198842,
      "loss": 2.3209,
      "step": 14292
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6435246467590332,
      "learning_rate": 0.0005943208085825826,
      "loss": 2.3232,
      "step": 14293
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6744422316551208,
      "learning_rate": 0.0005943200163905612,
      "loss": 2.4533,
      "step": 14294
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6417614221572876,
      "learning_rate": 0.0005943192241438201,
      "loss": 2.3887,
      "step": 14295
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6762133240699768,
      "learning_rate": 0.0005943184318423595,
      "loss": 2.3995,
      "step": 14296
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6106231808662415,
      "learning_rate": 0.0005943176394861796,
      "loss": 2.393,
      "step": 14297
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.5937562584877014,
      "learning_rate": 0.0005943168470752805,
      "loss": 2.2367,
      "step": 14298
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.5944352149963379,
      "learning_rate": 0.0005943160546096623,
      "loss": 2.3849,
      "step": 14299
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.5835317969322205,
      "learning_rate": 0.0005943152620893252,
      "loss": 2.4131,
      "step": 14300
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6016920208930969,
      "learning_rate": 0.0005943144695142693,
      "loss": 2.2444,
      "step": 14301
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.5786558985710144,
      "learning_rate": 0.0005943136768844948,
      "loss": 2.345,
      "step": 14302
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6743737459182739,
      "learning_rate": 0.000594312884200002,
      "loss": 2.5674,
      "step": 14303
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.5960249900817871,
      "learning_rate": 0.0005943120914607907,
      "loss": 2.5408,
      "step": 14304
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.5884495377540588,
      "learning_rate": 0.0005943112986668612,
      "loss": 2.3016,
      "step": 14305
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6483281254768372,
      "learning_rate": 0.0005943105058182138,
      "loss": 2.3208,
      "step": 14306
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6647107601165771,
      "learning_rate": 0.0005943097129148485,
      "loss": 2.3894,
      "step": 14307
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6780515909194946,
      "learning_rate": 0.0005943089199567654,
      "loss": 2.3083,
      "step": 14308
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.597476065158844,
      "learning_rate": 0.0005943081269439648,
      "loss": 2.2908,
      "step": 14309
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6185491681098938,
      "learning_rate": 0.0005943073338764468,
      "loss": 2.362,
      "step": 14310
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6384192705154419,
      "learning_rate": 0.0005943065407542113,
      "loss": 2.4409,
      "step": 14311
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6336387395858765,
      "learning_rate": 0.0005943057475772588,
      "loss": 2.5208,
      "step": 14312
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6267604827880859,
      "learning_rate": 0.0005943049543455892,
      "loss": 2.3185,
      "step": 14313
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.62395840883255,
      "learning_rate": 0.000594304161059203,
      "loss": 2.5469,
      "step": 14314
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6580280065536499,
      "learning_rate": 0.0005943033677180999,
      "loss": 2.296,
      "step": 14315
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6247718930244446,
      "learning_rate": 0.0005943025743222803,
      "loss": 2.3379,
      "step": 14316
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6123887300491333,
      "learning_rate": 0.0005943017808717442,
      "loss": 2.4081,
      "step": 14317
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6453898549079895,
      "learning_rate": 0.000594300987366492,
      "loss": 2.3313,
      "step": 14318
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.5870235562324524,
      "learning_rate": 0.0005943001938065236,
      "loss": 2.2681,
      "step": 14319
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6588127613067627,
      "learning_rate": 0.0005942994001918393,
      "loss": 2.305,
      "step": 14320
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.5638928413391113,
      "learning_rate": 0.0005942986065224391,
      "loss": 2.2302,
      "step": 14321
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6295063495635986,
      "learning_rate": 0.0005942978127983234,
      "loss": 2.536,
      "step": 14322
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6303079128265381,
      "learning_rate": 0.000594297019019492,
      "loss": 2.592,
      "step": 14323
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.7228965163230896,
      "learning_rate": 0.0005942962251859452,
      "loss": 2.5267,
      "step": 14324
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.5539856553077698,
      "learning_rate": 0.0005942954312976833,
      "loss": 2.4592,
      "step": 14325
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.9524465799331665,
      "learning_rate": 0.0005942946373547063,
      "loss": 2.3875,
      "step": 14326
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6407214999198914,
      "learning_rate": 0.0005942938433570144,
      "loss": 2.469,
      "step": 14327
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.59380704164505,
      "learning_rate": 0.0005942930493046077,
      "loss": 2.6604,
      "step": 14328
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.623735249042511,
      "learning_rate": 0.0005942922551974864,
      "loss": 2.339,
      "step": 14329
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.607055127620697,
      "learning_rate": 0.0005942914610356505,
      "loss": 2.2972,
      "step": 14330
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6822301745414734,
      "learning_rate": 0.0005942906668191003,
      "loss": 2.2429,
      "step": 14331
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6094381213188171,
      "learning_rate": 0.000594289872547836,
      "loss": 2.5156,
      "step": 14332
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.7761529684066772,
      "learning_rate": 0.0005942890782218576,
      "loss": 2.3892,
      "step": 14333
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.74270099401474,
      "learning_rate": 0.0005942882838411653,
      "loss": 2.4636,
      "step": 14334
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6484827995300293,
      "learning_rate": 0.0005942874894057592,
      "loss": 2.5217,
      "step": 14335
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6092790365219116,
      "learning_rate": 0.0005942866949156395,
      "loss": 2.3769,
      "step": 14336
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6463902592658997,
      "learning_rate": 0.0005942859003708065,
      "loss": 2.3889,
      "step": 14337
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.7448166608810425,
      "learning_rate": 0.0005942851057712601,
      "loss": 2.3565,
      "step": 14338
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6417331695556641,
      "learning_rate": 0.0005942843111170005,
      "loss": 2.4395,
      "step": 14339
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.9257818460464478,
      "learning_rate": 0.000594283516408028,
      "loss": 2.4795,
      "step": 14340
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6291006803512573,
      "learning_rate": 0.0005942827216443425,
      "loss": 2.4673,
      "step": 14341
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6823915243148804,
      "learning_rate": 0.0005942819268259443,
      "loss": 2.6883,
      "step": 14342
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.7276707887649536,
      "learning_rate": 0.0005942811319528336,
      "loss": 2.4068,
      "step": 14343
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.661648154258728,
      "learning_rate": 0.0005942803370250106,
      "loss": 2.1427,
      "step": 14344
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.5837375521659851,
      "learning_rate": 0.0005942795420424752,
      "loss": 2.5565,
      "step": 14345
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.7315502762794495,
      "learning_rate": 0.0005942787470052277,
      "loss": 2.2533,
      "step": 14346
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.665486752986908,
      "learning_rate": 0.0005942779519132682,
      "loss": 2.4182,
      "step": 14347
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6108877062797546,
      "learning_rate": 0.0005942771567665968,
      "loss": 2.3935,
      "step": 14348
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.610267162322998,
      "learning_rate": 0.0005942763615652139,
      "loss": 2.4058,
      "step": 14349
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6219996809959412,
      "learning_rate": 0.0005942755663091193,
      "loss": 2.3524,
      "step": 14350
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6748225688934326,
      "learning_rate": 0.0005942747709983135,
      "loss": 2.4696,
      "step": 14351
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6156757473945618,
      "learning_rate": 0.0005942739756327963,
      "loss": 2.187,
      "step": 14352
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6204050183296204,
      "learning_rate": 0.000594273180212568,
      "loss": 2.446,
      "step": 14353
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.7882580161094666,
      "learning_rate": 0.000594272384737629,
      "loss": 2.1759,
      "step": 14354
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.5891259908676147,
      "learning_rate": 0.0005942715892079789,
      "loss": 2.3953,
      "step": 14355
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.5694341063499451,
      "learning_rate": 0.0005942707936236184,
      "loss": 2.3739,
      "step": 14356
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.7276756763458252,
      "learning_rate": 0.0005942699979845472,
      "loss": 2.5792,
      "step": 14357
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.5813198685646057,
      "learning_rate": 0.0005942692022907659,
      "loss": 2.4785,
      "step": 14358
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.5738033652305603,
      "learning_rate": 0.0005942684065422742,
      "loss": 2.1806,
      "step": 14359
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6174483895301819,
      "learning_rate": 0.0005942676107390725,
      "loss": 2.372,
      "step": 14360
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.623982310295105,
      "learning_rate": 0.0005942668148811609,
      "loss": 2.294,
      "step": 14361
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.5953948497772217,
      "learning_rate": 0.0005942660189685395,
      "loss": 2.4746,
      "step": 14362
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.645965576171875,
      "learning_rate": 0.0005942652230012085,
      "loss": 2.3669,
      "step": 14363
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6137030124664307,
      "learning_rate": 0.0005942644269791681,
      "loss": 2.5783,
      "step": 14364
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.5773633718490601,
      "learning_rate": 0.0005942636309024183,
      "loss": 2.3202,
      "step": 14365
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.5949872136116028,
      "learning_rate": 0.0005942628347709595,
      "loss": 2.276,
      "step": 14366
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6457946300506592,
      "learning_rate": 0.0005942620385847915,
      "loss": 2.4256,
      "step": 14367
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.634786069393158,
      "learning_rate": 0.0005942612423439149,
      "loss": 2.5457,
      "step": 14368
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6359910368919373,
      "learning_rate": 0.0005942604460483293,
      "loss": 2.3779,
      "step": 14369
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.8718155026435852,
      "learning_rate": 0.0005942596496980353,
      "loss": 2.4529,
      "step": 14370
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6348642110824585,
      "learning_rate": 0.0005942588532930328,
      "loss": 2.6307,
      "step": 14371
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6110776662826538,
      "learning_rate": 0.000594258056833322,
      "loss": 2.3146,
      "step": 14372
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6993510127067566,
      "learning_rate": 0.0005942572603189031,
      "loss": 2.3449,
      "step": 14373
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6580825448036194,
      "learning_rate": 0.0005942564637497762,
      "loss": 2.3584,
      "step": 14374
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6636309623718262,
      "learning_rate": 0.0005942556671259415,
      "loss": 2.4885,
      "step": 14375
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6826887130737305,
      "learning_rate": 0.0005942548704473991,
      "loss": 2.134,
      "step": 14376
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6313955187797546,
      "learning_rate": 0.0005942540737141492,
      "loss": 2.4066,
      "step": 14377
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6835268139839172,
      "learning_rate": 0.0005942532769261918,
      "loss": 2.5298,
      "step": 14378
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6476001143455505,
      "learning_rate": 0.0005942524800835274,
      "loss": 2.1911,
      "step": 14379
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.627210259437561,
      "learning_rate": 0.0005942516831861557,
      "loss": 2.2409,
      "step": 14380
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6616988778114319,
      "learning_rate": 0.0005942508862340771,
      "loss": 2.4363,
      "step": 14381
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.5973248481750488,
      "learning_rate": 0.0005942500892272917,
      "loss": 2.2809,
      "step": 14382
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.7525434494018555,
      "learning_rate": 0.0005942492921657996,
      "loss": 2.3844,
      "step": 14383
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6616978049278259,
      "learning_rate": 0.0005942484950496012,
      "loss": 2.5863,
      "step": 14384
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6385126709938049,
      "learning_rate": 0.0005942476978786964,
      "loss": 2.3182,
      "step": 14385
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.720391035079956,
      "learning_rate": 0.0005942469006530853,
      "loss": 2.4669,
      "step": 14386
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.7564514875411987,
      "learning_rate": 0.0005942461033727681,
      "loss": 2.4328,
      "step": 14387
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6413608193397522,
      "learning_rate": 0.0005942453060377451,
      "loss": 2.24,
      "step": 14388
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.5832787156105042,
      "learning_rate": 0.0005942445086480164,
      "loss": 2.434,
      "step": 14389
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.8024449944496155,
      "learning_rate": 0.0005942437112035819,
      "loss": 2.4631,
      "step": 14390
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6291840672492981,
      "learning_rate": 0.000594242913704442,
      "loss": 2.4683,
      "step": 14391
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6078444123268127,
      "learning_rate": 0.0005942421161505969,
      "loss": 2.4168,
      "step": 14392
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6329485774040222,
      "learning_rate": 0.0005942413185420465,
      "loss": 2.6045,
      "step": 14393
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.7830799221992493,
      "learning_rate": 0.0005942405208787912,
      "loss": 2.3262,
      "step": 14394
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.5768173933029175,
      "learning_rate": 0.000594239723160831,
      "loss": 2.3652,
      "step": 14395
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.7101264595985413,
      "learning_rate": 0.000594238925388166,
      "loss": 2.5007,
      "step": 14396
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.7591583132743835,
      "learning_rate": 0.0005942381275607965,
      "loss": 2.3418,
      "step": 14397
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.5672739148139954,
      "learning_rate": 0.0005942373296787226,
      "loss": 2.0915,
      "step": 14398
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.5847833156585693,
      "learning_rate": 0.0005942365317419445,
      "loss": 2.2811,
      "step": 14399
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6620118618011475,
      "learning_rate": 0.0005942357337504621,
      "loss": 2.5812,
      "step": 14400
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.7372521162033081,
      "learning_rate": 0.0005942349357042758,
      "loss": 2.527,
      "step": 14401
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6636370420455933,
      "learning_rate": 0.0005942341376033857,
      "loss": 2.5705,
      "step": 14402
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6511253118515015,
      "learning_rate": 0.0005942333394477919,
      "loss": 2.4812,
      "step": 14403
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.7765006422996521,
      "learning_rate": 0.0005942325412374946,
      "loss": 2.4658,
      "step": 14404
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6411676406860352,
      "learning_rate": 0.0005942317429724938,
      "loss": 2.5181,
      "step": 14405
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.5547661781311035,
      "learning_rate": 0.0005942309446527899,
      "loss": 2.3237,
      "step": 14406
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6762474179267883,
      "learning_rate": 0.0005942301462783828,
      "loss": 2.2156,
      "step": 14407
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.637029767036438,
      "learning_rate": 0.0005942293478492729,
      "loss": 2.2217,
      "step": 14408
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.5987662076950073,
      "learning_rate": 0.0005942285493654601,
      "loss": 2.5375,
      "step": 14409
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.5820299983024597,
      "learning_rate": 0.0005942277508269446,
      "loss": 2.2866,
      "step": 14410
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6291992664337158,
      "learning_rate": 0.0005942269522337268,
      "loss": 2.2676,
      "step": 14411
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.71342533826828,
      "learning_rate": 0.0005942261535858064,
      "loss": 2.4682,
      "step": 14412
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6050439476966858,
      "learning_rate": 0.000594225354883184,
      "loss": 2.5221,
      "step": 14413
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6199425458908081,
      "learning_rate": 0.0005942245561258595,
      "loss": 2.2862,
      "step": 14414
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6735776662826538,
      "learning_rate": 0.0005942237573138331,
      "loss": 2.2928,
      "step": 14415
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6755149364471436,
      "learning_rate": 0.0005942229584471049,
      "loss": 2.3928,
      "step": 14416
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.5833616852760315,
      "learning_rate": 0.0005942221595256753,
      "loss": 2.1845,
      "step": 14417
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.8652138113975525,
      "learning_rate": 0.000594221360549544,
      "loss": 2.4623,
      "step": 14418
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6902893781661987,
      "learning_rate": 0.0005942205615187115,
      "loss": 2.4338,
      "step": 14419
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6025230884552002,
      "learning_rate": 0.0005942197624331778,
      "loss": 2.4865,
      "step": 14420
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.66583251953125,
      "learning_rate": 0.0005942189632929432,
      "loss": 2.5792,
      "step": 14421
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.7233748435974121,
      "learning_rate": 0.0005942181640980076,
      "loss": 2.356,
      "step": 14422
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6995699405670166,
      "learning_rate": 0.0005942173648483714,
      "loss": 2.4574,
      "step": 14423
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.5960542559623718,
      "learning_rate": 0.0005942165655440346,
      "loss": 2.2047,
      "step": 14424
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.7238632440567017,
      "learning_rate": 0.0005942157661849974,
      "loss": 2.5063,
      "step": 14425
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.7230268120765686,
      "learning_rate": 0.00059421496677126,
      "loss": 2.301,
      "step": 14426
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6192593574523926,
      "learning_rate": 0.0005942141673028223,
      "loss": 2.4156,
      "step": 14427
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6777769327163696,
      "learning_rate": 0.0005942133677796848,
      "loss": 2.2363,
      "step": 14428
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6959869861602783,
      "learning_rate": 0.0005942125682018473,
      "loss": 2.3042,
      "step": 14429
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6368836760520935,
      "learning_rate": 0.0005942117685693103,
      "loss": 2.4172,
      "step": 14430
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6190692186355591,
      "learning_rate": 0.0005942109688820737,
      "loss": 2.2751,
      "step": 14431
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6567954421043396,
      "learning_rate": 0.0005942101691401378,
      "loss": 2.2296,
      "step": 14432
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6979400515556335,
      "learning_rate": 0.0005942093693435027,
      "loss": 2.3324,
      "step": 14433
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.585290789604187,
      "learning_rate": 0.0005942085694921684,
      "loss": 2.3221,
      "step": 14434
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.7012324929237366,
      "learning_rate": 0.0005942077695861352,
      "loss": 2.3893,
      "step": 14435
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.7067323923110962,
      "learning_rate": 0.0005942069696254033,
      "loss": 2.1713,
      "step": 14436
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.5677809119224548,
      "learning_rate": 0.0005942061696099727,
      "loss": 2.5418,
      "step": 14437
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6118848919868469,
      "learning_rate": 0.0005942053695398436,
      "loss": 2.4184,
      "step": 14438
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.5922635197639465,
      "learning_rate": 0.0005942045694150163,
      "loss": 2.432,
      "step": 14439
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.594743013381958,
      "learning_rate": 0.0005942037692354907,
      "loss": 2.4851,
      "step": 14440
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.5933988094329834,
      "learning_rate": 0.000594202969001267,
      "loss": 2.4574,
      "step": 14441
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6454781293869019,
      "learning_rate": 0.0005942021687123456,
      "loss": 2.3483,
      "step": 14442
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6369906067848206,
      "learning_rate": 0.0005942013683687264,
      "loss": 2.6092,
      "step": 14443
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.5548858046531677,
      "learning_rate": 0.0005942005679704096,
      "loss": 2.3771,
      "step": 14444
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6366153359413147,
      "learning_rate": 0.0005941997675173952,
      "loss": 2.5649,
      "step": 14445
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.7091972231864929,
      "learning_rate": 0.0005941989670096837,
      "loss": 2.4166,
      "step": 14446
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6476405262947083,
      "learning_rate": 0.0005941981664472751,
      "loss": 2.4686,
      "step": 14447
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6551616191864014,
      "learning_rate": 0.0005941973658301694,
      "loss": 2.4052,
      "step": 14448
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.5991888642311096,
      "learning_rate": 0.0005941965651583668,
      "loss": 2.4167,
      "step": 14449
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6857704520225525,
      "learning_rate": 0.0005941957644318676,
      "loss": 2.5446,
      "step": 14450
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6216049194335938,
      "learning_rate": 0.0005941949636506718,
      "loss": 2.5637,
      "step": 14451
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6007784008979797,
      "learning_rate": 0.0005941941628147797,
      "loss": 2.2227,
      "step": 14452
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6694532036781311,
      "learning_rate": 0.0005941933619241913,
      "loss": 2.4481,
      "step": 14453
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6153402924537659,
      "learning_rate": 0.0005941925609789068,
      "loss": 2.1584,
      "step": 14454
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6268171072006226,
      "learning_rate": 0.0005941917599789264,
      "loss": 2.5248,
      "step": 14455
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6533272862434387,
      "learning_rate": 0.00059419095892425,
      "loss": 2.2773,
      "step": 14456
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.7374516725540161,
      "learning_rate": 0.0005941901578148781,
      "loss": 2.6008,
      "step": 14457
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.5880725979804993,
      "learning_rate": 0.0005941893566508107,
      "loss": 2.4151,
      "step": 14458
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6461004614830017,
      "learning_rate": 0.000594188555432048,
      "loss": 2.3213,
      "step": 14459
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6494332551956177,
      "learning_rate": 0.00059418775415859,
      "loss": 2.3487,
      "step": 14460
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.599694550037384,
      "learning_rate": 0.000594186952830437,
      "loss": 2.602,
      "step": 14461
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6652019023895264,
      "learning_rate": 0.000594186151447589,
      "loss": 2.4088,
      "step": 14462
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6521045565605164,
      "learning_rate": 0.0005941853500100462,
      "loss": 2.4216,
      "step": 14463
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6176599264144897,
      "learning_rate": 0.000594184548517809,
      "loss": 2.5392,
      "step": 14464
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.7200865149497986,
      "learning_rate": 0.0005941837469708772,
      "loss": 2.2263,
      "step": 14465
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6127220392227173,
      "learning_rate": 0.000594182945369251,
      "loss": 2.6026,
      "step": 14466
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.5837346315383911,
      "learning_rate": 0.0005941821437129308,
      "loss": 2.5124,
      "step": 14467
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6353915333747864,
      "learning_rate": 0.0005941813420019166,
      "loss": 2.2881,
      "step": 14468
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.691828727722168,
      "learning_rate": 0.0005941805402362084,
      "loss": 2.4632,
      "step": 14469
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.5704010725021362,
      "learning_rate": 0.0005941797384158066,
      "loss": 2.4565,
      "step": 14470
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.5934005379676819,
      "learning_rate": 0.0005941789365407112,
      "loss": 2.3818,
      "step": 14471
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.547653079032898,
      "learning_rate": 0.0005941781346109223,
      "loss": 2.4604,
      "step": 14472
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.5622397661209106,
      "learning_rate": 0.0005941773326264403,
      "loss": 2.3973,
      "step": 14473
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.641091525554657,
      "learning_rate": 0.0005941765305872651,
      "loss": 2.7601,
      "step": 14474
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.5796235203742981,
      "learning_rate": 0.0005941757284933969,
      "loss": 2.3374,
      "step": 14475
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.5717597007751465,
      "learning_rate": 0.000594174926344836,
      "loss": 2.501,
      "step": 14476
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6022550463676453,
      "learning_rate": 0.0005941741241415823,
      "loss": 2.2301,
      "step": 14477
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6569278836250305,
      "learning_rate": 0.000594173321883636,
      "loss": 2.4838,
      "step": 14478
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6003401875495911,
      "learning_rate": 0.0005941725195709975,
      "loss": 2.3867,
      "step": 14479
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6118708848953247,
      "learning_rate": 0.0005941717172036667,
      "loss": 2.5126,
      "step": 14480
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6058114171028137,
      "learning_rate": 0.0005941709147816438,
      "loss": 2.5682,
      "step": 14481
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.5564715266227722,
      "learning_rate": 0.0005941701123049289,
      "loss": 2.5463,
      "step": 14482
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.5882148742675781,
      "learning_rate": 0.0005941693097735224,
      "loss": 2.4923,
      "step": 14483
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6127632856369019,
      "learning_rate": 0.0005941685071874242,
      "loss": 2.4997,
      "step": 14484
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6532765030860901,
      "learning_rate": 0.0005941677045466345,
      "loss": 2.4896,
      "step": 14485
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.673811137676239,
      "learning_rate": 0.0005941669018511536,
      "loss": 2.5255,
      "step": 14486
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.5813994407653809,
      "learning_rate": 0.0005941660991009814,
      "loss": 2.3064,
      "step": 14487
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.5848090052604675,
      "learning_rate": 0.0005941652962961181,
      "loss": 2.4185,
      "step": 14488
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.5993654727935791,
      "learning_rate": 0.000594164493436564,
      "loss": 2.4877,
      "step": 14489
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.7021685838699341,
      "learning_rate": 0.0005941636905223191,
      "loss": 2.4191,
      "step": 14490
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.5479027628898621,
      "learning_rate": 0.0005941628875533838,
      "loss": 2.4524,
      "step": 14491
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6157630681991577,
      "learning_rate": 0.0005941620845297579,
      "loss": 2.4916,
      "step": 14492
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.5998648405075073,
      "learning_rate": 0.0005941612814514419,
      "loss": 2.5266,
      "step": 14493
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6209566593170166,
      "learning_rate": 0.0005941604783184355,
      "loss": 2.3357,
      "step": 14494
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.5966188311576843,
      "learning_rate": 0.0005941596751307392,
      "loss": 2.626,
      "step": 14495
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6254185438156128,
      "learning_rate": 0.0005941588718883532,
      "loss": 2.3647,
      "step": 14496
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.5847976803779602,
      "learning_rate": 0.0005941580685912774,
      "loss": 2.2718,
      "step": 14497
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6044082045555115,
      "learning_rate": 0.0005941572652395121,
      "loss": 2.623,
      "step": 14498
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.7715416550636292,
      "learning_rate": 0.0005941564618330573,
      "loss": 2.3161,
      "step": 14499
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6152493953704834,
      "learning_rate": 0.0005941556583719134,
      "loss": 2.374,
      "step": 14500
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6636034846305847,
      "learning_rate": 0.0005941548548560804,
      "loss": 2.2054,
      "step": 14501
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.5728440880775452,
      "learning_rate": 0.0005941540512855584,
      "loss": 2.3376,
      "step": 14502
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6611511707305908,
      "learning_rate": 0.0005941532476603477,
      "loss": 2.1915,
      "step": 14503
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6367290019989014,
      "learning_rate": 0.0005941524439804482,
      "loss": 2.5483,
      "step": 14504
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6878201365470886,
      "learning_rate": 0.0005941516402458604,
      "loss": 2.4483,
      "step": 14505
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6999656558036804,
      "learning_rate": 0.0005941508364565841,
      "loss": 2.5168,
      "step": 14506
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6534832119941711,
      "learning_rate": 0.0005941500326126196,
      "loss": 2.2935,
      "step": 14507
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6051353216171265,
      "learning_rate": 0.0005941492287139672,
      "loss": 2.3761,
      "step": 14508
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6483721137046814,
      "learning_rate": 0.0005941484247606268,
      "loss": 2.4882,
      "step": 14509
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6072786450386047,
      "learning_rate": 0.0005941476207525988,
      "loss": 2.5698,
      "step": 14510
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.5999689102172852,
      "learning_rate": 0.000594146816689883,
      "loss": 2.1509,
      "step": 14511
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6206774711608887,
      "learning_rate": 0.0005941460125724798,
      "loss": 2.512,
      "step": 14512
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6570848822593689,
      "learning_rate": 0.0005941452084003894,
      "loss": 2.3796,
      "step": 14513
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.5724762082099915,
      "learning_rate": 0.0005941444041736119,
      "loss": 2.5063,
      "step": 14514
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6704308390617371,
      "learning_rate": 0.0005941435998921472,
      "loss": 2.4603,
      "step": 14515
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.595567524433136,
      "learning_rate": 0.0005941427955559958,
      "loss": 2.3575,
      "step": 14516
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6317728757858276,
      "learning_rate": 0.0005941419911651576,
      "loss": 2.1886,
      "step": 14517
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.5655925869941711,
      "learning_rate": 0.000594141186719633,
      "loss": 2.4255,
      "step": 14518
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6576170325279236,
      "learning_rate": 0.0005941403822194219,
      "loss": 2.5042,
      "step": 14519
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6053290367126465,
      "learning_rate": 0.0005941395776645245,
      "loss": 2.2106,
      "step": 14520
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.662957489490509,
      "learning_rate": 0.0005941387730549412,
      "loss": 2.4476,
      "step": 14521
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6285120844841003,
      "learning_rate": 0.0005941379683906717,
      "loss": 2.479,
      "step": 14522
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6368845701217651,
      "learning_rate": 0.0005941371636717165,
      "loss": 2.3484,
      "step": 14523
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6926906704902649,
      "learning_rate": 0.0005941363588980757,
      "loss": 2.2497,
      "step": 14524
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.7714981436729431,
      "learning_rate": 0.0005941355540697493,
      "loss": 2.4048,
      "step": 14525
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6185338497161865,
      "learning_rate": 0.0005941347491867377,
      "loss": 2.4503,
      "step": 14526
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.7063753604888916,
      "learning_rate": 0.0005941339442490408,
      "loss": 2.0891,
      "step": 14527
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.672173798084259,
      "learning_rate": 0.0005941331392566588,
      "loss": 2.4937,
      "step": 14528
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6480982899665833,
      "learning_rate": 0.000594132334209592,
      "loss": 2.4294,
      "step": 14529
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6508743762969971,
      "learning_rate": 0.0005941315291078404,
      "loss": 2.3814,
      "step": 14530
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.5891779065132141,
      "learning_rate": 0.0005941307239514041,
      "loss": 2.4849,
      "step": 14531
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6509063839912415,
      "learning_rate": 0.0005941299187402835,
      "loss": 2.1347,
      "step": 14532
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6805288791656494,
      "learning_rate": 0.0005941291134744785,
      "loss": 2.4088,
      "step": 14533
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.645300567150116,
      "learning_rate": 0.0005941283081539894,
      "loss": 2.4999,
      "step": 14534
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6327648162841797,
      "learning_rate": 0.0005941275027788163,
      "loss": 2.3503,
      "step": 14535
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.7382708191871643,
      "learning_rate": 0.0005941266973489593,
      "loss": 2.4833,
      "step": 14536
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.618987500667572,
      "learning_rate": 0.0005941258918644187,
      "loss": 2.4682,
      "step": 14537
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.677371621131897,
      "learning_rate": 0.0005941250863251944,
      "loss": 2.5301,
      "step": 14538
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.7717645168304443,
      "learning_rate": 0.0005941242807312868,
      "loss": 2.3775,
      "step": 14539
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6325538158416748,
      "learning_rate": 0.0005941234750826958,
      "loss": 2.5931,
      "step": 14540
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.7270453572273254,
      "learning_rate": 0.0005941226693794219,
      "loss": 2.4898,
      "step": 14541
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6855454444885254,
      "learning_rate": 0.0005941218636214649,
      "loss": 2.4267,
      "step": 14542
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.5907104015350342,
      "learning_rate": 0.0005941210578088251,
      "loss": 2.3911,
      "step": 14543
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6706135272979736,
      "learning_rate": 0.0005941202519415028,
      "loss": 2.4131,
      "step": 14544
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6907113194465637,
      "learning_rate": 0.0005941194460194978,
      "loss": 2.5678,
      "step": 14545
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6149402260780334,
      "learning_rate": 0.0005941186400428106,
      "loss": 2.4064,
      "step": 14546
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.5981208086013794,
      "learning_rate": 0.000594117834011441,
      "loss": 2.4578,
      "step": 14547
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.8377780914306641,
      "learning_rate": 0.0005941170279253894,
      "loss": 2.238,
      "step": 14548
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6288866400718689,
      "learning_rate": 0.000594116221784656,
      "loss": 2.4974,
      "step": 14549
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6994075775146484,
      "learning_rate": 0.0005941154155892407,
      "loss": 2.4234,
      "step": 14550
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6265841126441956,
      "learning_rate": 0.000594114609339144,
      "loss": 2.1764,
      "step": 14551
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6083808541297913,
      "learning_rate": 0.0005941138030343657,
      "loss": 2.4784,
      "step": 14552
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6274319887161255,
      "learning_rate": 0.0005941129966749061,
      "loss": 2.381,
      "step": 14553
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6752704381942749,
      "learning_rate": 0.0005941121902607653,
      "loss": 2.4564,
      "step": 14554
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6227609515190125,
      "learning_rate": 0.0005941113837919436,
      "loss": 2.5051,
      "step": 14555
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6854211091995239,
      "learning_rate": 0.000594110577268441,
      "loss": 2.3309,
      "step": 14556
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.7033138275146484,
      "learning_rate": 0.0005941097706902576,
      "loss": 2.4431,
      "step": 14557
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.5847679376602173,
      "learning_rate": 0.0005941089640573938,
      "loss": 2.387,
      "step": 14558
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.5813064575195312,
      "learning_rate": 0.0005941081573698494,
      "loss": 2.6248,
      "step": 14559
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.7236112952232361,
      "learning_rate": 0.0005941073506276249,
      "loss": 2.5498,
      "step": 14560
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.597448468208313,
      "learning_rate": 0.0005941065438307203,
      "loss": 2.3106,
      "step": 14561
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6809996366500854,
      "learning_rate": 0.0005941057369791356,
      "loss": 2.5171,
      "step": 14562
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6269632577896118,
      "learning_rate": 0.0005941049300728712,
      "loss": 2.6751,
      "step": 14563
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6222587823867798,
      "learning_rate": 0.0005941041231119271,
      "loss": 2.1947,
      "step": 14564
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.5997345447540283,
      "learning_rate": 0.0005941033160963036,
      "loss": 2.4263,
      "step": 14565
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6271112561225891,
      "learning_rate": 0.0005941025090260006,
      "loss": 2.3952,
      "step": 14566
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.570121169090271,
      "learning_rate": 0.0005941017019010184,
      "loss": 2.4028,
      "step": 14567
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.5695847272872925,
      "learning_rate": 0.0005941008947213573,
      "loss": 2.6146,
      "step": 14568
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.5977528691291809,
      "learning_rate": 0.0005941000874870172,
      "loss": 2.4054,
      "step": 14569
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.5769743323326111,
      "learning_rate": 0.0005940992801979982,
      "loss": 2.4827,
      "step": 14570
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6094610095024109,
      "learning_rate": 0.0005940984728543008,
      "loss": 2.1945,
      "step": 14571
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.7818821668624878,
      "learning_rate": 0.0005940976654559248,
      "loss": 2.1707,
      "step": 14572
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6209227442741394,
      "learning_rate": 0.0005940968580028706,
      "loss": 2.3996,
      "step": 14573
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.8632463216781616,
      "learning_rate": 0.0005940960504951381,
      "loss": 2.6871,
      "step": 14574
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6393624544143677,
      "learning_rate": 0.0005940952429327278,
      "loss": 2.4795,
      "step": 14575
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.7206871509552002,
      "learning_rate": 0.0005940944353156395,
      "loss": 2.3706,
      "step": 14576
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6123329997062683,
      "learning_rate": 0.0005940936276438735,
      "loss": 2.2654,
      "step": 14577
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6110594868659973,
      "learning_rate": 0.00059409281991743,
      "loss": 2.649,
      "step": 14578
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6055610179901123,
      "learning_rate": 0.000594092012136309,
      "loss": 2.393,
      "step": 14579
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6368427872657776,
      "learning_rate": 0.0005940912043005108,
      "loss": 2.5089,
      "step": 14580
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.5853919386863708,
      "learning_rate": 0.0005940903964100356,
      "loss": 2.2867,
      "step": 14581
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6021246314048767,
      "learning_rate": 0.0005940895884648833,
      "loss": 2.3681,
      "step": 14582
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.755959689617157,
      "learning_rate": 0.0005940887804650541,
      "loss": 2.2417,
      "step": 14583
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6699970960617065,
      "learning_rate": 0.0005940879724105484,
      "loss": 2.5134,
      "step": 14584
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6564936637878418,
      "learning_rate": 0.0005940871643013662,
      "loss": 2.345,
      "step": 14585
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.7325570583343506,
      "learning_rate": 0.0005940863561375077,
      "loss": 2.3359,
      "step": 14586
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6199967861175537,
      "learning_rate": 0.0005940855479189729,
      "loss": 2.538,
      "step": 14587
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.8179835081100464,
      "learning_rate": 0.0005940847396457621,
      "loss": 2.4665,
      "step": 14588
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6315475106239319,
      "learning_rate": 0.0005940839313178753,
      "loss": 2.3883,
      "step": 14589
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6621096134185791,
      "learning_rate": 0.0005940831229353128,
      "loss": 2.5983,
      "step": 14590
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6121019124984741,
      "learning_rate": 0.0005940823144980747,
      "loss": 2.4873,
      "step": 14591
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.5494087934494019,
      "learning_rate": 0.0005940815060061611,
      "loss": 2.3903,
      "step": 14592
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6724576354026794,
      "learning_rate": 0.0005940806974595723,
      "loss": 2.5903,
      "step": 14593
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.5465282201766968,
      "learning_rate": 0.0005940798888583082,
      "loss": 2.5032,
      "step": 14594
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6101499795913696,
      "learning_rate": 0.0005940790802023692,
      "loss": 2.1995,
      "step": 14595
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.7004754543304443,
      "learning_rate": 0.0005940782714917553,
      "loss": 2.3717,
      "step": 14596
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6375650763511658,
      "learning_rate": 0.0005940774627264668,
      "loss": 2.4719,
      "step": 14597
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.5718677043914795,
      "learning_rate": 0.0005940766539065036,
      "loss": 2.31,
      "step": 14598
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6243003010749817,
      "learning_rate": 0.0005940758450318661,
      "loss": 2.4625,
      "step": 14599
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6366034746170044,
      "learning_rate": 0.0005940750361025543,
      "loss": 2.4492,
      "step": 14600
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6119937896728516,
      "learning_rate": 0.0005940742271185685,
      "loss": 2.1881,
      "step": 14601
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.5863659381866455,
      "learning_rate": 0.0005940734180799087,
      "loss": 2.1246,
      "step": 14602
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6427812576293945,
      "learning_rate": 0.0005940726089865751,
      "loss": 2.2063,
      "step": 14603
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6847882866859436,
      "learning_rate": 0.0005940717998385678,
      "loss": 2.409,
      "step": 14604
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.5639604330062866,
      "learning_rate": 0.0005940709906358869,
      "loss": 2.5998,
      "step": 14605
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.8321141600608826,
      "learning_rate": 0.0005940701813785329,
      "loss": 2.5228,
      "step": 14606
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6056519746780396,
      "learning_rate": 0.0005940693720665056,
      "loss": 2.5692,
      "step": 14607
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6489289999008179,
      "learning_rate": 0.0005940685626998052,
      "loss": 2.6175,
      "step": 14608
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.7830774188041687,
      "learning_rate": 0.0005940677532784319,
      "loss": 2.2688,
      "step": 14609
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6581966876983643,
      "learning_rate": 0.0005940669438023859,
      "loss": 2.3902,
      "step": 14610
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6290775537490845,
      "learning_rate": 0.0005940661342716673,
      "loss": 2.5284,
      "step": 14611
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6545864939689636,
      "learning_rate": 0.0005940653246862762,
      "loss": 2.5374,
      "step": 14612
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6694729924201965,
      "learning_rate": 0.000594064515046213,
      "loss": 2.4572,
      "step": 14613
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6256482005119324,
      "learning_rate": 0.0005940637053514775,
      "loss": 2.5986,
      "step": 14614
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6837547421455383,
      "learning_rate": 0.00059406289560207,
      "loss": 2.3745,
      "step": 14615
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6364279985427856,
      "learning_rate": 0.0005940620857979907,
      "loss": 2.3725,
      "step": 14616
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.7288228869438171,
      "learning_rate": 0.0005940612759392397,
      "loss": 2.2723,
      "step": 14617
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.589434802532196,
      "learning_rate": 0.0005940604660258171,
      "loss": 2.3713,
      "step": 14618
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6283012628555298,
      "learning_rate": 0.0005940596560577231,
      "loss": 2.5033,
      "step": 14619
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6285251975059509,
      "learning_rate": 0.0005940588460349579,
      "loss": 2.3738,
      "step": 14620
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.7479548454284668,
      "learning_rate": 0.0005940580359575216,
      "loss": 2.4674,
      "step": 14621
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6289752721786499,
      "learning_rate": 0.0005940572258254144,
      "loss": 2.6313,
      "step": 14622
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.7438903450965881,
      "learning_rate": 0.0005940564156386364,
      "loss": 2.5455,
      "step": 14623
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6905999779701233,
      "learning_rate": 0.0005940556053971878,
      "loss": 2.3851,
      "step": 14624
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.632790207862854,
      "learning_rate": 0.0005940547951010686,
      "loss": 2.5466,
      "step": 14625
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6627785563468933,
      "learning_rate": 0.0005940539847502791,
      "loss": 2.1624,
      "step": 14626
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.643252968788147,
      "learning_rate": 0.0005940531743448195,
      "loss": 2.3233,
      "step": 14627
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.613422155380249,
      "learning_rate": 0.0005940523638846898,
      "loss": 2.3746,
      "step": 14628
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.5953102707862854,
      "learning_rate": 0.0005940515533698902,
      "loss": 2.4426,
      "step": 14629
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6295931339263916,
      "learning_rate": 0.0005940507428004209,
      "loss": 2.5918,
      "step": 14630
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.608651876449585,
      "learning_rate": 0.000594049932176282,
      "loss": 2.2577,
      "step": 14631
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.5795879364013672,
      "learning_rate": 0.0005940491214974737,
      "loss": 2.3087,
      "step": 14632
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6076993942260742,
      "learning_rate": 0.000594048310763996,
      "loss": 2.4052,
      "step": 14633
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.5700856447219849,
      "learning_rate": 0.0005940474999758493,
      "loss": 2.3194,
      "step": 14634
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6149216294288635,
      "learning_rate": 0.0005940466891330335,
      "loss": 2.3484,
      "step": 14635
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6441201567649841,
      "learning_rate": 0.000594045878235549,
      "loss": 2.2781,
      "step": 14636
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6658328771591187,
      "learning_rate": 0.0005940450672833957,
      "loss": 2.3324,
      "step": 14637
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6162993907928467,
      "learning_rate": 0.000594044256276574,
      "loss": 2.505,
      "step": 14638
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.5907325744628906,
      "learning_rate": 0.0005940434452150838,
      "loss": 2.2947,
      "step": 14639
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6399633288383484,
      "learning_rate": 0.0005940426340989254,
      "loss": 2.4572,
      "step": 14640
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.7228971719741821,
      "learning_rate": 0.000594041822928099,
      "loss": 2.381,
      "step": 14641
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6210504174232483,
      "learning_rate": 0.0005940410117026047,
      "loss": 2.3229,
      "step": 14642
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.5881081819534302,
      "learning_rate": 0.0005940402004224425,
      "loss": 2.4378,
      "step": 14643
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6603935956954956,
      "learning_rate": 0.0005940393890876127,
      "loss": 2.4542,
      "step": 14644
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6146784424781799,
      "learning_rate": 0.0005940385776981155,
      "loss": 2.3882,
      "step": 14645
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6616199612617493,
      "learning_rate": 0.000594037766253951,
      "loss": 2.469,
      "step": 14646
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.7431735396385193,
      "learning_rate": 0.0005940369547551192,
      "loss": 1.9878,
      "step": 14647
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.739206850528717,
      "learning_rate": 0.0005940361432016205,
      "loss": 2.5733,
      "step": 14648
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.5980345606803894,
      "learning_rate": 0.0005940353315934549,
      "loss": 2.4986,
      "step": 14649
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6971349120140076,
      "learning_rate": 0.0005940345199306224,
      "loss": 2.305,
      "step": 14650
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6549588441848755,
      "learning_rate": 0.0005940337082131236,
      "loss": 2.2863,
      "step": 14651
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6564720273017883,
      "learning_rate": 0.0005940328964409583,
      "loss": 2.4878,
      "step": 14652
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.7159494161605835,
      "learning_rate": 0.0005940320846141267,
      "loss": 2.2805,
      "step": 14653
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.681792676448822,
      "learning_rate": 0.000594031272732629,
      "loss": 2.5492,
      "step": 14654
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.5752888321876526,
      "learning_rate": 0.0005940304607964653,
      "loss": 2.1615,
      "step": 14655
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6483904123306274,
      "learning_rate": 0.0005940296488056358,
      "loss": 2.1364,
      "step": 14656
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.7200914621353149,
      "learning_rate": 0.0005940288367601406,
      "loss": 2.5193,
      "step": 14657
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.632434606552124,
      "learning_rate": 0.00059402802465998,
      "loss": 2.3595,
      "step": 14658
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6626625657081604,
      "learning_rate": 0.000594027212505154,
      "loss": 2.3725,
      "step": 14659
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6637300252914429,
      "learning_rate": 0.0005940264002956627,
      "loss": 2.3877,
      "step": 14660
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6343436241149902,
      "learning_rate": 0.0005940255880315065,
      "loss": 2.2044,
      "step": 14661
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6463467478752136,
      "learning_rate": 0.0005940247757126852,
      "loss": 2.5438,
      "step": 14662
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.8240390419960022,
      "learning_rate": 0.0005940239633391993,
      "loss": 2.4593,
      "step": 14663
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6292071342468262,
      "learning_rate": 0.0005940231509110488,
      "loss": 2.2287,
      "step": 14664
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.724410355091095,
      "learning_rate": 0.0005940223384282338,
      "loss": 2.065,
      "step": 14665
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.7479274272918701,
      "learning_rate": 0.0005940215258907545,
      "loss": 2.4597,
      "step": 14666
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6064271926879883,
      "learning_rate": 0.0005940207132986109,
      "loss": 2.4142,
      "step": 14667
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6776224374771118,
      "learning_rate": 0.0005940199006518035,
      "loss": 2.3713,
      "step": 14668
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.8424626588821411,
      "learning_rate": 0.0005940190879503322,
      "loss": 2.4625,
      "step": 14669
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.5935909748077393,
      "learning_rate": 0.0005940182751941973,
      "loss": 2.546,
      "step": 14670
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.7768937945365906,
      "learning_rate": 0.0005940174623833987,
      "loss": 2.4882,
      "step": 14671
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.7308084964752197,
      "learning_rate": 0.0005940166495179367,
      "loss": 2.5133,
      "step": 14672
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6391400098800659,
      "learning_rate": 0.0005940158365978115,
      "loss": 2.4859,
      "step": 14673
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6357037425041199,
      "learning_rate": 0.0005940150236230232,
      "loss": 2.4406,
      "step": 14674
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.630564272403717,
      "learning_rate": 0.0005940142105935719,
      "loss": 2.3942,
      "step": 14675
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6006367206573486,
      "learning_rate": 0.0005940133975094579,
      "loss": 2.1632,
      "step": 14676
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6376349329948425,
      "learning_rate": 0.0005940125843706813,
      "loss": 2.6765,
      "step": 14677
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6786635518074036,
      "learning_rate": 0.000594011771177242,
      "loss": 2.2753,
      "step": 14678
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.7320868968963623,
      "learning_rate": 0.0005940109579291406,
      "loss": 2.4665,
      "step": 14679
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6591870188713074,
      "learning_rate": 0.0005940101446263769,
      "loss": 2.5111,
      "step": 14680
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6250616312026978,
      "learning_rate": 0.0005940093312689512,
      "loss": 2.2821,
      "step": 14681
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6199101209640503,
      "learning_rate": 0.0005940085178568637,
      "loss": 2.5494,
      "step": 14682
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6327706575393677,
      "learning_rate": 0.0005940077043901142,
      "loss": 2.2781,
      "step": 14683
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6666492223739624,
      "learning_rate": 0.0005940068908687034,
      "loss": 2.369,
      "step": 14684
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6228147745132446,
      "learning_rate": 0.000594006077292631,
      "loss": 2.4621,
      "step": 14685
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6170146465301514,
      "learning_rate": 0.0005940052636618973,
      "loss": 2.1786,
      "step": 14686
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6253049373626709,
      "learning_rate": 0.0005940044499765026,
      "loss": 2.4668,
      "step": 14687
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6459276080131531,
      "learning_rate": 0.0005940036362364468,
      "loss": 2.4968,
      "step": 14688
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6651090979576111,
      "learning_rate": 0.0005940028224417303,
      "loss": 2.6174,
      "step": 14689
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6150634288787842,
      "learning_rate": 0.000594002008592353,
      "loss": 2.4217,
      "step": 14690
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6341713666915894,
      "learning_rate": 0.0005940011946883153,
      "loss": 2.595,
      "step": 14691
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.676213264465332,
      "learning_rate": 0.0005940003807296172,
      "loss": 2.3611,
      "step": 14692
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6045046448707581,
      "learning_rate": 0.0005939995667162589,
      "loss": 2.3043,
      "step": 14693
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6476709842681885,
      "learning_rate": 0.0005939987526482405,
      "loss": 2.2735,
      "step": 14694
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6783095598220825,
      "learning_rate": 0.0005939979385255621,
      "loss": 2.3104,
      "step": 14695
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6071242094039917,
      "learning_rate": 0.0005939971243482241,
      "loss": 2.5489,
      "step": 14696
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6498851180076599,
      "learning_rate": 0.0005939963101162263,
      "loss": 2.3888,
      "step": 14697
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6277197599411011,
      "learning_rate": 0.0005939954958295691,
      "loss": 2.2631,
      "step": 14698
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6315537691116333,
      "learning_rate": 0.0005939946814882526,
      "loss": 2.5297,
      "step": 14699
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.635552167892456,
      "learning_rate": 0.0005939938670922769,
      "loss": 2.3609,
      "step": 14700
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6840671896934509,
      "learning_rate": 0.0005939930526416423,
      "loss": 2.2455,
      "step": 14701
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6255480647087097,
      "learning_rate": 0.0005939922381363487,
      "loss": 2.3309,
      "step": 14702
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6809256076812744,
      "learning_rate": 0.0005939914235763965,
      "loss": 2.343,
      "step": 14703
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.7140795588493347,
      "learning_rate": 0.0005939906089617857,
      "loss": 2.514,
      "step": 14704
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6012350916862488,
      "learning_rate": 0.0005939897942925166,
      "loss": 2.2777,
      "step": 14705
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6914215683937073,
      "learning_rate": 0.0005939889795685892,
      "loss": 2.278,
      "step": 14706
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.8429942727088928,
      "learning_rate": 0.0005939881647900035,
      "loss": 2.4181,
      "step": 14707
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6307505369186401,
      "learning_rate": 0.0005939873499567601,
      "loss": 2.4031,
      "step": 14708
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.7496418356895447,
      "learning_rate": 0.0005939865350688588,
      "loss": 2.3227,
      "step": 14709
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.760083794593811,
      "learning_rate": 0.0005939857201263,
      "loss": 2.3705,
      "step": 14710
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.5941267013549805,
      "learning_rate": 0.0005939849051290834,
      "loss": 2.2112,
      "step": 14711
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.9390059113502502,
      "learning_rate": 0.0005939840900772097,
      "loss": 2.2879,
      "step": 14712
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.820456862449646,
      "learning_rate": 0.0005939832749706787,
      "loss": 2.1569,
      "step": 14713
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6445344090461731,
      "learning_rate": 0.0005939824598094907,
      "loss": 2.1547,
      "step": 14714
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.881158709526062,
      "learning_rate": 0.0005939816445936458,
      "loss": 2.2659,
      "step": 14715
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6621359586715698,
      "learning_rate": 0.0005939808293231441,
      "loss": 2.4027,
      "step": 14716
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6627939939498901,
      "learning_rate": 0.000593980013997986,
      "loss": 2.4168,
      "step": 14717
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6436210870742798,
      "learning_rate": 0.0005939791986181713,
      "loss": 2.2969,
      "step": 14718
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6575116515159607,
      "learning_rate": 0.0005939783831837004,
      "loss": 2.4736,
      "step": 14719
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.7539700269699097,
      "learning_rate": 0.0005939775676945733,
      "loss": 2.4804,
      "step": 14720
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.5738502740859985,
      "learning_rate": 0.0005939767521507902,
      "loss": 2.2727,
      "step": 14721
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.8046050667762756,
      "learning_rate": 0.0005939759365523515,
      "loss": 2.4464,
      "step": 14722
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6461926698684692,
      "learning_rate": 0.0005939751208992568,
      "loss": 2.4552,
      "step": 14723
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6846016049385071,
      "learning_rate": 0.0005939743051915068,
      "loss": 2.3686,
      "step": 14724
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6834732294082642,
      "learning_rate": 0.0005939734894291013,
      "loss": 2.4794,
      "step": 14725
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6950286030769348,
      "learning_rate": 0.0005939726736120407,
      "loss": 2.531,
      "step": 14726
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6064551472663879,
      "learning_rate": 0.0005939718577403249,
      "loss": 2.3986,
      "step": 14727
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6286966800689697,
      "learning_rate": 0.0005939710418139543,
      "loss": 2.673,
      "step": 14728
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.646413266658783,
      "learning_rate": 0.0005939702258329289,
      "loss": 2.5346,
      "step": 14729
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6830237507820129,
      "learning_rate": 0.0005939694097972488,
      "loss": 2.39,
      "step": 14730
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.5517739057540894,
      "learning_rate": 0.0005939685937069144,
      "loss": 2.473,
      "step": 14731
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6186028122901917,
      "learning_rate": 0.0005939677775619256,
      "loss": 2.2602,
      "step": 14732
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6519533395767212,
      "learning_rate": 0.0005939669613622826,
      "loss": 2.4611,
      "step": 14733
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6240105628967285,
      "learning_rate": 0.0005939661451079857,
      "loss": 2.3447,
      "step": 14734
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.5828201174736023,
      "learning_rate": 0.0005939653287990348,
      "loss": 2.5272,
      "step": 14735
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.5859115719795227,
      "learning_rate": 0.0005939645124354303,
      "loss": 2.2122,
      "step": 14736
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6609982252120972,
      "learning_rate": 0.0005939636960171722,
      "loss": 2.3344,
      "step": 14737
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6097577214241028,
      "learning_rate": 0.0005939628795442607,
      "loss": 2.3215,
      "step": 14738
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.5871477723121643,
      "learning_rate": 0.000593962063016696,
      "loss": 2.3313,
      "step": 14739
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.5976101160049438,
      "learning_rate": 0.0005939612464344781,
      "loss": 2.3774,
      "step": 14740
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.5796903967857361,
      "learning_rate": 0.0005939604297976075,
      "loss": 2.371,
      "step": 14741
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.5831222534179688,
      "learning_rate": 0.0005939596131060838,
      "loss": 2.6208,
      "step": 14742
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6130056977272034,
      "learning_rate": 0.0005939587963599077,
      "loss": 2.4874,
      "step": 14743
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.5898910164833069,
      "learning_rate": 0.000593957979559079,
      "loss": 2.3004,
      "step": 14744
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6272373199462891,
      "learning_rate": 0.0005939571627035979,
      "loss": 2.503,
      "step": 14745
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6405439376831055,
      "learning_rate": 0.0005939563457934647,
      "loss": 2.2506,
      "step": 14746
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.5642221570014954,
      "learning_rate": 0.0005939555288286794,
      "loss": 2.3469,
      "step": 14747
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6060115098953247,
      "learning_rate": 0.0005939547118092424,
      "loss": 2.596,
      "step": 14748
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6557959318161011,
      "learning_rate": 0.0005939538947351535,
      "loss": 2.3725,
      "step": 14749
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.5582727193832397,
      "learning_rate": 0.0005939530776064131,
      "loss": 2.3882,
      "step": 14750
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.565902829170227,
      "learning_rate": 0.0005939522604230212,
      "loss": 2.3412,
      "step": 14751
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6459419131278992,
      "learning_rate": 0.0005939514431849781,
      "loss": 2.6751,
      "step": 14752
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6084916591644287,
      "learning_rate": 0.0005939506258922837,
      "loss": 2.2889,
      "step": 14753
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.5974093675613403,
      "learning_rate": 0.0005939498085449386,
      "loss": 2.389,
      "step": 14754
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6699308156967163,
      "learning_rate": 0.0005939489911429425,
      "loss": 2.3425,
      "step": 14755
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.5770442485809326,
      "learning_rate": 0.0005939481736862958,
      "loss": 2.3806,
      "step": 14756
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6132913827896118,
      "learning_rate": 0.0005939473561749985,
      "loss": 2.2847,
      "step": 14757
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.630952775478363,
      "learning_rate": 0.000593946538609051,
      "loss": 2.3847,
      "step": 14758
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.724530816078186,
      "learning_rate": 0.0005939457209884532,
      "loss": 2.3833,
      "step": 14759
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.7780616283416748,
      "learning_rate": 0.0005939449033132053,
      "loss": 2.5268,
      "step": 14760
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6266986727714539,
      "learning_rate": 0.0005939440855833076,
      "loss": 2.2965,
      "step": 14761
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.699347198009491,
      "learning_rate": 0.00059394326779876,
      "loss": 2.3734,
      "step": 14762
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6105374693870544,
      "learning_rate": 0.0005939424499595629,
      "loss": 2.2183,
      "step": 14763
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.7179051041603088,
      "learning_rate": 0.0005939416320657163,
      "loss": 2.2731,
      "step": 14764
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6562581062316895,
      "learning_rate": 0.0005939408141172205,
      "loss": 2.4962,
      "step": 14765
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.5934320688247681,
      "learning_rate": 0.0005939399961140755,
      "loss": 2.4129,
      "step": 14766
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.7140524983406067,
      "learning_rate": 0.0005939391780562814,
      "loss": 2.6524,
      "step": 14767
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6723238229751587,
      "learning_rate": 0.0005939383599438386,
      "loss": 2.5388,
      "step": 14768
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6538092494010925,
      "learning_rate": 0.0005939375417767471,
      "loss": 2.5515,
      "step": 14769
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.714515209197998,
      "learning_rate": 0.000593936723555007,
      "loss": 2.6052,
      "step": 14770
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6928940415382385,
      "learning_rate": 0.0005939359052786185,
      "loss": 2.4652,
      "step": 14771
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6397455334663391,
      "learning_rate": 0.0005939350869475818,
      "loss": 2.4127,
      "step": 14772
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6196566224098206,
      "learning_rate": 0.0005939342685618971,
      "loss": 2.3001,
      "step": 14773
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6310972571372986,
      "learning_rate": 0.0005939334501215643,
      "loss": 2.2608,
      "step": 14774
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6092454791069031,
      "learning_rate": 0.0005939326316265839,
      "loss": 2.3794,
      "step": 14775
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.5979146957397461,
      "learning_rate": 0.0005939318130769558,
      "loss": 2.5334,
      "step": 14776
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.5685299038887024,
      "learning_rate": 0.0005939309944726802,
      "loss": 2.3185,
      "step": 14777
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6651948094367981,
      "learning_rate": 0.0005939301758137573,
      "loss": 2.3074,
      "step": 14778
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.7010842561721802,
      "learning_rate": 0.0005939293571001872,
      "loss": 2.4855,
      "step": 14779
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.5749409794807434,
      "learning_rate": 0.0005939285383319701,
      "loss": 2.4446,
      "step": 14780
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.7352791428565979,
      "learning_rate": 0.0005939277195091061,
      "loss": 2.3428,
      "step": 14781
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6626996397972107,
      "learning_rate": 0.0005939269006315956,
      "loss": 2.3691,
      "step": 14782
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6562911868095398,
      "learning_rate": 0.0005939260816994383,
      "loss": 2.4986,
      "step": 14783
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.5647907257080078,
      "learning_rate": 0.0005939252627126347,
      "loss": 2.6128,
      "step": 14784
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6699612140655518,
      "learning_rate": 0.0005939244436711849,
      "loss": 2.4374,
      "step": 14785
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6596697568893433,
      "learning_rate": 0.0005939236245750888,
      "loss": 2.5014,
      "step": 14786
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.5720492005348206,
      "learning_rate": 0.0005939228054243469,
      "loss": 2.303,
      "step": 14787
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6719653606414795,
      "learning_rate": 0.0005939219862189591,
      "loss": 2.6007,
      "step": 14788
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6590623259544373,
      "learning_rate": 0.0005939211669589258,
      "loss": 2.5834,
      "step": 14789
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.603379487991333,
      "learning_rate": 0.0005939203476442469,
      "loss": 2.5188,
      "step": 14790
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6245797872543335,
      "learning_rate": 0.0005939195282749227,
      "loss": 2.364,
      "step": 14791
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.7209185361862183,
      "learning_rate": 0.0005939187088509533,
      "loss": 2.3007,
      "step": 14792
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.7052667737007141,
      "learning_rate": 0.0005939178893723389,
      "loss": 2.6893,
      "step": 14793
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6079875826835632,
      "learning_rate": 0.0005939170698390795,
      "loss": 2.2373,
      "step": 14794
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6442368030548096,
      "learning_rate": 0.0005939162502511755,
      "loss": 2.6181,
      "step": 14795
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6552770733833313,
      "learning_rate": 0.0005939154306086269,
      "loss": 2.2349,
      "step": 14796
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.583330512046814,
      "learning_rate": 0.0005939146109114338,
      "loss": 2.6723,
      "step": 14797
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6472862958908081,
      "learning_rate": 0.0005939137911595966,
      "loss": 2.3855,
      "step": 14798
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6221295595169067,
      "learning_rate": 0.0005939129713531151,
      "loss": 2.4411,
      "step": 14799
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.8232280611991882,
      "learning_rate": 0.0005939121514919897,
      "loss": 2.4478,
      "step": 14800
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.591680109500885,
      "learning_rate": 0.0005939113315762204,
      "loss": 2.4367,
      "step": 14801
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6491394639015198,
      "learning_rate": 0.0005939105116058074,
      "loss": 2.2206,
      "step": 14802
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6425482034683228,
      "learning_rate": 0.0005939096915807511,
      "loss": 2.2341,
      "step": 14803
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.659721851348877,
      "learning_rate": 0.0005939088715010513,
      "loss": 2.5559,
      "step": 14804
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6621274352073669,
      "learning_rate": 0.0005939080513667084,
      "loss": 2.4496,
      "step": 14805
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.7493748068809509,
      "learning_rate": 0.0005939072311777223,
      "loss": 2.4199,
      "step": 14806
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6244761943817139,
      "learning_rate": 0.0005939064109340933,
      "loss": 2.4456,
      "step": 14807
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6240447759628296,
      "learning_rate": 0.0005939055906358217,
      "loss": 2.2694,
      "step": 14808
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.5811648964881897,
      "learning_rate": 0.0005939047702829074,
      "loss": 2.3545,
      "step": 14809
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6613402962684631,
      "learning_rate": 0.0005939039498753506,
      "loss": 2.5934,
      "step": 14810
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.5902686715126038,
      "learning_rate": 0.0005939031294131516,
      "loss": 2.4899,
      "step": 14811
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.7182528972625732,
      "learning_rate": 0.0005939023088963105,
      "loss": 2.4841,
      "step": 14812
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6443418264389038,
      "learning_rate": 0.0005939014883248274,
      "loss": 2.2257,
      "step": 14813
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6694983243942261,
      "learning_rate": 0.0005939006676987024,
      "loss": 2.4153,
      "step": 14814
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6021093130111694,
      "learning_rate": 0.0005938998470179357,
      "loss": 2.7633,
      "step": 14815
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6263813972473145,
      "learning_rate": 0.0005938990262825275,
      "loss": 2.5148,
      "step": 14816
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.633469820022583,
      "learning_rate": 0.0005938982054924779,
      "loss": 2.3601,
      "step": 14817
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6349534392356873,
      "learning_rate": 0.0005938973846477872,
      "loss": 2.496,
      "step": 14818
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.627654492855072,
      "learning_rate": 0.0005938965637484553,
      "loss": 2.4138,
      "step": 14819
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6736600399017334,
      "learning_rate": 0.0005938957427944825,
      "loss": 2.4967,
      "step": 14820
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6381001472473145,
      "learning_rate": 0.0005938949217858689,
      "loss": 2.4691,
      "step": 14821
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.5915248394012451,
      "learning_rate": 0.0005938941007226148,
      "loss": 2.2758,
      "step": 14822
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.633764386177063,
      "learning_rate": 0.0005938932796047202,
      "loss": 2.4986,
      "step": 14823
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6709913015365601,
      "learning_rate": 0.0005938924584321852,
      "loss": 2.318,
      "step": 14824
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6363086700439453,
      "learning_rate": 0.0005938916372050101,
      "loss": 2.3087,
      "step": 14825
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6461431384086609,
      "learning_rate": 0.000593890815923195,
      "loss": 2.2858,
      "step": 14826
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6800873875617981,
      "learning_rate": 0.00059388999458674,
      "loss": 2.5386,
      "step": 14827
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.5818771719932556,
      "learning_rate": 0.0005938891731956454,
      "loss": 2.5118,
      "step": 14828
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6157791614532471,
      "learning_rate": 0.0005938883517499112,
      "loss": 2.4447,
      "step": 14829
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.7221017479896545,
      "learning_rate": 0.0005938875302495377,
      "loss": 2.5344,
      "step": 14830
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6895783543586731,
      "learning_rate": 0.0005938867086945249,
      "loss": 2.5891,
      "step": 14831
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.7002099752426147,
      "learning_rate": 0.000593885887084873,
      "loss": 2.5512,
      "step": 14832
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6127439141273499,
      "learning_rate": 0.0005938850654205821,
      "loss": 2.4916,
      "step": 14833
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.7504713535308838,
      "learning_rate": 0.0005938842437016525,
      "loss": 2.5306,
      "step": 14834
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.710518479347229,
      "learning_rate": 0.0005938834219280843,
      "loss": 2.5683,
      "step": 14835
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6663077473640442,
      "learning_rate": 0.0005938826000998775,
      "loss": 2.5964,
      "step": 14836
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6411309838294983,
      "learning_rate": 0.0005938817782170325,
      "loss": 2.469,
      "step": 14837
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.7833052277565002,
      "learning_rate": 0.0005938809562795492,
      "loss": 2.5308,
      "step": 14838
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6879239678382874,
      "learning_rate": 0.000593880134287428,
      "loss": 2.4496,
      "step": 14839
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6530067324638367,
      "learning_rate": 0.0005938793122406689,
      "loss": 2.546,
      "step": 14840
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6860150098800659,
      "learning_rate": 0.0005938784901392721,
      "loss": 2.3669,
      "step": 14841
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6374291181564331,
      "learning_rate": 0.0005938776679832378,
      "loss": 2.5688,
      "step": 14842
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.7252781391143799,
      "learning_rate": 0.0005938768457725659,
      "loss": 2.6298,
      "step": 14843
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6922014951705933,
      "learning_rate": 0.000593876023507257,
      "loss": 2.6933,
      "step": 14844
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6037281155586243,
      "learning_rate": 0.0005938752011873108,
      "loss": 2.3063,
      "step": 14845
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.8577268123626709,
      "learning_rate": 0.0005938743788127278,
      "loss": 2.422,
      "step": 14846
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.7837038636207581,
      "learning_rate": 0.0005938735563835079,
      "loss": 2.2978,
      "step": 14847
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6625120043754578,
      "learning_rate": 0.0005938727338996514,
      "loss": 2.3724,
      "step": 14848
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.7888272404670715,
      "learning_rate": 0.0005938719113611585,
      "loss": 2.3617,
      "step": 14849
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.7231996655464172,
      "learning_rate": 0.0005938710887680291,
      "loss": 2.4906,
      "step": 14850
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.582700252532959,
      "learning_rate": 0.0005938702661202636,
      "loss": 2.4706,
      "step": 14851
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.8212234973907471,
      "learning_rate": 0.000593869443417862,
      "loss": 2.2741,
      "step": 14852
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.163232445716858,
      "learning_rate": 0.0005938686206608246,
      "loss": 2.4325,
      "step": 14853
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.7919206023216248,
      "learning_rate": 0.0005938677978491514,
      "loss": 2.5222,
      "step": 14854
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6466214060783386,
      "learning_rate": 0.0005938669749828427,
      "loss": 2.4879,
      "step": 14855
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.7788830399513245,
      "learning_rate": 0.0005938661520618985,
      "loss": 2.4148,
      "step": 14856
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.7350104451179504,
      "learning_rate": 0.0005938653290863192,
      "loss": 2.5763,
      "step": 14857
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6537249684333801,
      "learning_rate": 0.0005938645060561046,
      "loss": 2.426,
      "step": 14858
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6117645502090454,
      "learning_rate": 0.0005938636829712552,
      "loss": 2.353,
      "step": 14859
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.8525485992431641,
      "learning_rate": 0.0005938628598317709,
      "loss": 2.4439,
      "step": 14860
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6213257312774658,
      "learning_rate": 0.0005938620366376519,
      "loss": 2.5366,
      "step": 14861
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.5679046511650085,
      "learning_rate": 0.0005938612133888986,
      "loss": 2.2133,
      "step": 14862
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.7078580260276794,
      "learning_rate": 0.0005938603900855108,
      "loss": 2.3345,
      "step": 14863
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6229312419891357,
      "learning_rate": 0.0005938595667274888,
      "loss": 2.4553,
      "step": 14864
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6218521595001221,
      "learning_rate": 0.0005938587433148328,
      "loss": 2.548,
      "step": 14865
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.578885555267334,
      "learning_rate": 0.0005938579198475429,
      "loss": 2.2979,
      "step": 14866
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.5637385845184326,
      "learning_rate": 0.0005938570963256193,
      "loss": 2.3066,
      "step": 14867
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.7054590582847595,
      "learning_rate": 0.0005938562727490621,
      "loss": 2.684,
      "step": 14868
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6144655346870422,
      "learning_rate": 0.0005938554491178714,
      "loss": 2.4778,
      "step": 14869
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.5972294807434082,
      "learning_rate": 0.0005938546254320475,
      "loss": 2.4737,
      "step": 14870
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6239006519317627,
      "learning_rate": 0.0005938538016915905,
      "loss": 2.3503,
      "step": 14871
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.5817297101020813,
      "learning_rate": 0.0005938529778965005,
      "loss": 2.6635,
      "step": 14872
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.580741286277771,
      "learning_rate": 0.0005938521540467777,
      "loss": 2.3079,
      "step": 14873
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.60663241147995,
      "learning_rate": 0.0005938513301424221,
      "loss": 2.4056,
      "step": 14874
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.5197696685791016,
      "learning_rate": 0.0005938505061834341,
      "loss": 2.0477,
      "step": 14875
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.5443888902664185,
      "learning_rate": 0.0005938496821698138,
      "loss": 2.2839,
      "step": 14876
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.5698967576026917,
      "learning_rate": 0.0005938488581015613,
      "loss": 2.4261,
      "step": 14877
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6097471714019775,
      "learning_rate": 0.0005938480339786767,
      "loss": 2.4336,
      "step": 14878
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6585192680358887,
      "learning_rate": 0.0005938472098011602,
      "loss": 2.5882,
      "step": 14879
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6606988310813904,
      "learning_rate": 0.0005938463855690119,
      "loss": 2.4608,
      "step": 14880
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6393681168556213,
      "learning_rate": 0.0005938455612822321,
      "loss": 2.3083,
      "step": 14881
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.5723340511322021,
      "learning_rate": 0.0005938447369408208,
      "loss": 2.432,
      "step": 14882
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6279258728027344,
      "learning_rate": 0.0005938439125447783,
      "loss": 2.5126,
      "step": 14883
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6668124794960022,
      "learning_rate": 0.0005938430880941046,
      "loss": 2.4689,
      "step": 14884
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.630622148513794,
      "learning_rate": 0.0005938422635888,
      "loss": 2.5177,
      "step": 14885
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.7056212425231934,
      "learning_rate": 0.0005938414390288644,
      "loss": 2.5313,
      "step": 14886
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.65461266040802,
      "learning_rate": 0.0005938406144142983,
      "loss": 2.5439,
      "step": 14887
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6109277009963989,
      "learning_rate": 0.0005938397897451016,
      "loss": 2.4709,
      "step": 14888
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.9689602255821228,
      "learning_rate": 0.0005938389650212745,
      "loss": 2.3184,
      "step": 14889
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6189351081848145,
      "learning_rate": 0.0005938381402428173,
      "loss": 2.4136,
      "step": 14890
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6621211171150208,
      "learning_rate": 0.0005938373154097301,
      "loss": 2.3953,
      "step": 14891
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6297734379768372,
      "learning_rate": 0.0005938364905220129,
      "loss": 2.3089,
      "step": 14892
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6553855538368225,
      "learning_rate": 0.000593835665579666,
      "loss": 2.6421,
      "step": 14893
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6347640752792358,
      "learning_rate": 0.0005938348405826894,
      "loss": 2.3767,
      "step": 14894
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6365699172019958,
      "learning_rate": 0.0005938340155310834,
      "loss": 2.6217,
      "step": 14895
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6759190559387207,
      "learning_rate": 0.0005938331904248481,
      "loss": 2.4083,
      "step": 14896
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6570369601249695,
      "learning_rate": 0.0005938323652639837,
      "loss": 2.2964,
      "step": 14897
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.5751867294311523,
      "learning_rate": 0.0005938315400484903,
      "loss": 2.2389,
      "step": 14898
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.7076283693313599,
      "learning_rate": 0.0005938307147783681,
      "loss": 2.2037,
      "step": 14899
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6916014552116394,
      "learning_rate": 0.0005938298894536171,
      "loss": 2.129,
      "step": 14900
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.5801648497581482,
      "learning_rate": 0.0005938290640742377,
      "loss": 2.4948,
      "step": 14901
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6283568739891052,
      "learning_rate": 0.0005938282386402299,
      "loss": 2.3137,
      "step": 14902
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.65510094165802,
      "learning_rate": 0.0005938274131515939,
      "loss": 2.5416,
      "step": 14903
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6184253096580505,
      "learning_rate": 0.0005938265876083298,
      "loss": 2.4116,
      "step": 14904
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.619053065776825,
      "learning_rate": 0.0005938257620104378,
      "loss": 2.0957,
      "step": 14905
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6208084225654602,
      "learning_rate": 0.0005938249363579181,
      "loss": 2.3334,
      "step": 14906
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.7097657322883606,
      "learning_rate": 0.0005938241106507707,
      "loss": 2.4618,
      "step": 14907
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.5507319569587708,
      "learning_rate": 0.000593823284888996,
      "loss": 2.3305,
      "step": 14908
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.5957084894180298,
      "learning_rate": 0.0005938224590725939,
      "loss": 2.3673,
      "step": 14909
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6096281409263611,
      "learning_rate": 0.0005938216332015647,
      "loss": 2.3804,
      "step": 14910
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6603080630302429,
      "learning_rate": 0.0005938208072759085,
      "loss": 2.6008,
      "step": 14911
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6715001463890076,
      "learning_rate": 0.0005938199812956254,
      "loss": 2.3831,
      "step": 14912
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.668500542640686,
      "learning_rate": 0.0005938191552607156,
      "loss": 2.5142,
      "step": 14913
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.7423521280288696,
      "learning_rate": 0.0005938183291711793,
      "loss": 2.5868,
      "step": 14914
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6111606955528259,
      "learning_rate": 0.0005938175030270167,
      "loss": 2.4296,
      "step": 14915
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6939785480499268,
      "learning_rate": 0.0005938166768282278,
      "loss": 2.5831,
      "step": 14916
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.5779374241828918,
      "learning_rate": 0.0005938158505748129,
      "loss": 2.4691,
      "step": 14917
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6336827874183655,
      "learning_rate": 0.0005938150242667721,
      "loss": 2.5291,
      "step": 14918
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.5857847929000854,
      "learning_rate": 0.0005938141979041056,
      "loss": 2.2222,
      "step": 14919
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.718698263168335,
      "learning_rate": 0.0005938133714868133,
      "loss": 2.4624,
      "step": 14920
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6276825070381165,
      "learning_rate": 0.0005938125450148956,
      "loss": 2.4197,
      "step": 14921
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6054090857505798,
      "learning_rate": 0.0005938117184883528,
      "loss": 2.4689,
      "step": 14922
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.601415753364563,
      "learning_rate": 0.0005938108919071847,
      "loss": 2.481,
      "step": 14923
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.538153350353241,
      "learning_rate": 0.0005938100652713915,
      "loss": 2.3959,
      "step": 14924
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6037492156028748,
      "learning_rate": 0.0005938092385809737,
      "loss": 2.3825,
      "step": 14925
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6108547449111938,
      "learning_rate": 0.000593808411835931,
      "loss": 2.2177,
      "step": 14926
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6021835803985596,
      "learning_rate": 0.0005938075850362638,
      "loss": 2.377,
      "step": 14927
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.5706120729446411,
      "learning_rate": 0.0005938067581819724,
      "loss": 2.5197,
      "step": 14928
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6738558411598206,
      "learning_rate": 0.0005938059312730567,
      "loss": 2.222,
      "step": 14929
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.7358272671699524,
      "learning_rate": 0.0005938051043095168,
      "loss": 2.7319,
      "step": 14930
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6157673001289368,
      "learning_rate": 0.0005938042772913532,
      "loss": 2.4233,
      "step": 14931
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6170373558998108,
      "learning_rate": 0.0005938034502185656,
      "loss": 2.462,
      "step": 14932
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6561896800994873,
      "learning_rate": 0.0005938026230911546,
      "loss": 2.3412,
      "step": 14933
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6534988284111023,
      "learning_rate": 0.00059380179590912,
      "loss": 2.6456,
      "step": 14934
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6866115927696228,
      "learning_rate": 0.0005938009686724622,
      "loss": 2.3288,
      "step": 14935
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6603937149047852,
      "learning_rate": 0.0005938001413811812,
      "loss": 2.4263,
      "step": 14936
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6046598553657532,
      "learning_rate": 0.0005937993140352772,
      "loss": 2.4645,
      "step": 14937
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.8095317482948303,
      "learning_rate": 0.0005937984866347505,
      "loss": 2.438,
      "step": 14938
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6540483832359314,
      "learning_rate": 0.000593797659179601,
      "loss": 2.2966,
      "step": 14939
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6447094082832336,
      "learning_rate": 0.0005937968316698289,
      "loss": 2.4726,
      "step": 14940
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6480410695075989,
      "learning_rate": 0.0005937960041054345,
      "loss": 2.5826,
      "step": 14941
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6234128475189209,
      "learning_rate": 0.0005937951764864179,
      "loss": 2.5359,
      "step": 14942
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6276340484619141,
      "learning_rate": 0.0005937943488127791,
      "loss": 2.5376,
      "step": 14943
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6011742949485779,
      "learning_rate": 0.0005937935210845185,
      "loss": 2.4559,
      "step": 14944
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6458849906921387,
      "learning_rate": 0.0005937926933016361,
      "loss": 2.5718,
      "step": 14945
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6029564738273621,
      "learning_rate": 0.0005937918654641321,
      "loss": 2.2414,
      "step": 14946
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6211027503013611,
      "learning_rate": 0.0005937910375720068,
      "loss": 2.4669,
      "step": 14947
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6240233778953552,
      "learning_rate": 0.0005937902096252599,
      "loss": 2.287,
      "step": 14948
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.7137060761451721,
      "learning_rate": 0.0005937893816238922,
      "loss": 2.6631,
      "step": 14949
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.5933602452278137,
      "learning_rate": 0.0005937885535679033,
      "loss": 2.1405,
      "step": 14950
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6456506848335266,
      "learning_rate": 0.0005937877254572935,
      "loss": 2.4369,
      "step": 14951
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.7020355463027954,
      "learning_rate": 0.0005937868972920631,
      "loss": 2.2756,
      "step": 14952
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6153427958488464,
      "learning_rate": 0.0005937860690722123,
      "loss": 2.3373,
      "step": 14953
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6491724252700806,
      "learning_rate": 0.0005937852407977409,
      "loss": 2.6473,
      "step": 14954
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6628751158714294,
      "learning_rate": 0.0005937844124686494,
      "loss": 2.3387,
      "step": 14955
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.5892234444618225,
      "learning_rate": 0.0005937835840849377,
      "loss": 2.2599,
      "step": 14956
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6205121278762817,
      "learning_rate": 0.0005937827556466062,
      "loss": 2.5043,
      "step": 14957
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.5889759063720703,
      "learning_rate": 0.000593781927153655,
      "loss": 2.6185,
      "step": 14958
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6084558963775635,
      "learning_rate": 0.0005937810986060842,
      "loss": 2.3006,
      "step": 14959
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.5730893611907959,
      "learning_rate": 0.0005937802700038938,
      "loss": 2.331,
      "step": 14960
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.5903852581977844,
      "learning_rate": 0.0005937794413470842,
      "loss": 2.3721,
      "step": 14961
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6159250140190125,
      "learning_rate": 0.0005937786126356554,
      "loss": 2.5902,
      "step": 14962
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.5624069571495056,
      "learning_rate": 0.0005937777838696077,
      "loss": 2.2897,
      "step": 14963
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6405599117279053,
      "learning_rate": 0.000593776955048941,
      "loss": 2.3693,
      "step": 14964
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6492586731910706,
      "learning_rate": 0.0005937761261736559,
      "loss": 2.4969,
      "step": 14965
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.5799474716186523,
      "learning_rate": 0.000593775297243752,
      "loss": 2.3962,
      "step": 14966
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6490231156349182,
      "learning_rate": 0.0005937744682592298,
      "loss": 2.7076,
      "step": 14967
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6242507100105286,
      "learning_rate": 0.0005937736392200894,
      "loss": 2.4526,
      "step": 14968
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.5770947933197021,
      "learning_rate": 0.000593772810126331,
      "loss": 2.3954,
      "step": 14969
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.614538311958313,
      "learning_rate": 0.0005937719809779547,
      "loss": 2.3683,
      "step": 14970
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6655508279800415,
      "learning_rate": 0.0005937711517749605,
      "loss": 2.2875,
      "step": 14971
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.5566482543945312,
      "learning_rate": 0.0005937703225173488,
      "loss": 2.2659,
      "step": 14972
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6074018478393555,
      "learning_rate": 0.0005937694932051196,
      "loss": 2.3301,
      "step": 14973
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6319126486778259,
      "learning_rate": 0.0005937686638382731,
      "loss": 2.7489,
      "step": 14974
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.5921199917793274,
      "learning_rate": 0.0005937678344168096,
      "loss": 2.5279,
      "step": 14975
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.8007797598838806,
      "learning_rate": 0.000593767004940729,
      "loss": 2.4258,
      "step": 14976
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6152784824371338,
      "learning_rate": 0.0005937661754100315,
      "loss": 2.4839,
      "step": 14977
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6194912195205688,
      "learning_rate": 0.0005937653458247174,
      "loss": 2.3249,
      "step": 14978
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.573998212814331,
      "learning_rate": 0.0005937645161847867,
      "loss": 2.3364,
      "step": 14979
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.5882576704025269,
      "learning_rate": 0.0005937636864902398,
      "loss": 2.3435,
      "step": 14980
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.5613572597503662,
      "learning_rate": 0.0005937628567410766,
      "loss": 2.3219,
      "step": 14981
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.5604222416877747,
      "learning_rate": 0.0005937620269372973,
      "loss": 2.3972,
      "step": 14982
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.5949171781539917,
      "learning_rate": 0.0005937611970789021,
      "loss": 2.3327,
      "step": 14983
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.646475076675415,
      "learning_rate": 0.0005937603671658913,
      "loss": 2.294,
      "step": 14984
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.69701087474823,
      "learning_rate": 0.0005937595371982647,
      "loss": 2.3606,
      "step": 14985
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6385924220085144,
      "learning_rate": 0.0005937587071760227,
      "loss": 2.4265,
      "step": 14986
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6094233393669128,
      "learning_rate": 0.0005937578770991655,
      "loss": 2.2746,
      "step": 14987
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.5711051821708679,
      "learning_rate": 0.0005937570469676931,
      "loss": 2.4759,
      "step": 14988
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.5826224088668823,
      "learning_rate": 0.0005937562167816057,
      "loss": 2.3921,
      "step": 14989
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6032643914222717,
      "learning_rate": 0.0005937553865409035,
      "loss": 2.432,
      "step": 14990
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6575214862823486,
      "learning_rate": 0.0005937545562455866,
      "loss": 2.692,
      "step": 14991
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.5951986312866211,
      "learning_rate": 0.0005937537258956552,
      "loss": 2.5292,
      "step": 14992
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6339316964149475,
      "learning_rate": 0.0005937528954911096,
      "loss": 2.6515,
      "step": 14993
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6059452891349792,
      "learning_rate": 0.0005937520650319496,
      "loss": 2.4309,
      "step": 14994
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.5779791474342346,
      "learning_rate": 0.0005937512345181756,
      "loss": 2.2058,
      "step": 14995
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.5908662676811218,
      "learning_rate": 0.0005937504039497877,
      "loss": 2.558,
      "step": 14996
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6007867455482483,
      "learning_rate": 0.000593749573326786,
      "loss": 2.4739,
      "step": 14997
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.8565379977226257,
      "learning_rate": 0.0005937487426491707,
      "loss": 2.4534,
      "step": 14998
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6994374394416809,
      "learning_rate": 0.0005937479119169421,
      "loss": 2.4905,
      "step": 14999
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6399199366569519,
      "learning_rate": 0.0005937470811301001,
      "loss": 2.6198,
      "step": 15000
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.5990753173828125,
      "learning_rate": 0.000593746250288645,
      "loss": 2.5059,
      "step": 15001
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6203057765960693,
      "learning_rate": 0.0005937454193925769,
      "loss": 2.3432,
      "step": 15002
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6753350496292114,
      "learning_rate": 0.0005937445884418961,
      "loss": 2.6361,
      "step": 15003
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.5932822227478027,
      "learning_rate": 0.0005937437574366025,
      "loss": 2.2736,
      "step": 15004
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.5998404622077942,
      "learning_rate": 0.0005937429263766965,
      "loss": 2.4458,
      "step": 15005
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6377029418945312,
      "learning_rate": 0.0005937420952621781,
      "loss": 2.264,
      "step": 15006
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6393153071403503,
      "learning_rate": 0.0005937412640930475,
      "loss": 2.6007,
      "step": 15007
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.5509071350097656,
      "learning_rate": 0.0005937404328693048,
      "loss": 2.3269,
      "step": 15008
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6088960766792297,
      "learning_rate": 0.0005937396015909503,
      "loss": 2.5722,
      "step": 15009
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.594689667224884,
      "learning_rate": 0.000593738770257984,
      "loss": 2.4158,
      "step": 15010
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6029193997383118,
      "learning_rate": 0.0005937379388704061,
      "loss": 2.351,
      "step": 15011
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6196299195289612,
      "learning_rate": 0.0005937371074282168,
      "loss": 2.6085,
      "step": 15012
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6879898309707642,
      "learning_rate": 0.0005937362759314163,
      "loss": 2.4792,
      "step": 15013
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.5759212374687195,
      "learning_rate": 0.0005937354443800045,
      "loss": 2.4203,
      "step": 15014
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6052161455154419,
      "learning_rate": 0.0005937346127739819,
      "loss": 2.4908,
      "step": 15015
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.678449273109436,
      "learning_rate": 0.0005937337811133484,
      "loss": 2.4925,
      "step": 15016
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6459224224090576,
      "learning_rate": 0.0005937329493981043,
      "loss": 2.6519,
      "step": 15017
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.688946545124054,
      "learning_rate": 0.0005937321176282497,
      "loss": 2.4565,
      "step": 15018
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.7077546119689941,
      "learning_rate": 0.0005937312858037848,
      "loss": 2.5053,
      "step": 15019
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.5933636426925659,
      "learning_rate": 0.0005937304539247096,
      "loss": 2.273,
      "step": 15020
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6141822934150696,
      "learning_rate": 0.0005937296219910244,
      "loss": 2.3251,
      "step": 15021
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.5753827691078186,
      "learning_rate": 0.0005937287900027294,
      "loss": 2.4565,
      "step": 15022
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.5948293805122375,
      "learning_rate": 0.0005937279579598246,
      "loss": 2.4343,
      "step": 15023
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.5927544832229614,
      "learning_rate": 0.0005937271258623102,
      "loss": 2.5305,
      "step": 15024
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.8280673027038574,
      "learning_rate": 0.0005937262937101864,
      "loss": 2.5789,
      "step": 15025
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6274802684783936,
      "learning_rate": 0.0005937254615034534,
      "loss": 2.441,
      "step": 15026
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.636815071105957,
      "learning_rate": 0.0005937246292421112,
      "loss": 2.3129,
      "step": 15027
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.598932683467865,
      "learning_rate": 0.0005937237969261601,
      "loss": 2.4795,
      "step": 15028
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.5847542881965637,
      "learning_rate": 0.0005937229645556002,
      "loss": 2.2987,
      "step": 15029
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6231252551078796,
      "learning_rate": 0.0005937221321304317,
      "loss": 2.5453,
      "step": 15030
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.5975620150566101,
      "learning_rate": 0.0005937212996506546,
      "loss": 2.27,
      "step": 15031
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.65432208776474,
      "learning_rate": 0.0005937204671162693,
      "loss": 2.3477,
      "step": 15032
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6419850587844849,
      "learning_rate": 0.0005937196345272757,
      "loss": 2.7748,
      "step": 15033
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6481443643569946,
      "learning_rate": 0.0005937188018836742,
      "loss": 2.4086,
      "step": 15034
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6149788498878479,
      "learning_rate": 0.0005937179691854647,
      "loss": 2.29,
      "step": 15035
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.634113073348999,
      "learning_rate": 0.0005937171364326476,
      "loss": 2.411,
      "step": 15036
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.5620366930961609,
      "learning_rate": 0.0005937163036252229,
      "loss": 2.3322,
      "step": 15037
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.8370744585990906,
      "learning_rate": 0.0005937154707631908,
      "loss": 2.3665,
      "step": 15038
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6206813454627991,
      "learning_rate": 0.0005937146378465514,
      "loss": 2.5521,
      "step": 15039
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.5611216425895691,
      "learning_rate": 0.0005937138048753049,
      "loss": 2.3896,
      "step": 15040
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.676886796951294,
      "learning_rate": 0.0005937129718494515,
      "loss": 2.5089,
      "step": 15041
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6828594207763672,
      "learning_rate": 0.0005937121387689912,
      "loss": 2.3871,
      "step": 15042
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.5992676615715027,
      "learning_rate": 0.0005937113056339244,
      "loss": 2.7616,
      "step": 15043
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6297211647033691,
      "learning_rate": 0.0005937104724442513,
      "loss": 2.2486,
      "step": 15044
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6836045980453491,
      "learning_rate": 0.0005937096391999715,
      "loss": 2.5459,
      "step": 15045
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6419191956520081,
      "learning_rate": 0.0005937088059010857,
      "loss": 2.2928,
      "step": 15046
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.5875447392463684,
      "learning_rate": 0.000593707972547594,
      "loss": 2.3556,
      "step": 15047
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.5570858716964722,
      "learning_rate": 0.0005937071391394963,
      "loss": 2.098,
      "step": 15048
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.5867494940757751,
      "learning_rate": 0.000593706305676793,
      "loss": 2.5578,
      "step": 15049
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.630623996257782,
      "learning_rate": 0.0005937054721594841,
      "loss": 2.4635,
      "step": 15050
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6350367665290833,
      "learning_rate": 0.0005937046385875698,
      "loss": 2.351,
      "step": 15051
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6793156266212463,
      "learning_rate": 0.0005937038049610502,
      "loss": 2.5897,
      "step": 15052
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6429599523544312,
      "learning_rate": 0.0005937029712799256,
      "loss": 2.3725,
      "step": 15053
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6141203045845032,
      "learning_rate": 0.0005937021375441961,
      "loss": 2.3253,
      "step": 15054
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6474381685256958,
      "learning_rate": 0.0005937013037538618,
      "loss": 2.473,
      "step": 15055
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.5596509575843811,
      "learning_rate": 0.0005937004699089228,
      "loss": 2.3937,
      "step": 15056
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.599345326423645,
      "learning_rate": 0.0005936996360093794,
      "loss": 2.6035,
      "step": 15057
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6065781116485596,
      "learning_rate": 0.0005936988020552318,
      "loss": 2.2764,
      "step": 15058
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.5822856426239014,
      "learning_rate": 0.0005936979680464799,
      "loss": 2.4981,
      "step": 15059
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.5866644382476807,
      "learning_rate": 0.0005936971339831241,
      "loss": 2.4086,
      "step": 15060
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6440689563751221,
      "learning_rate": 0.0005936962998651644,
      "loss": 2.4186,
      "step": 15061
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6075630784034729,
      "learning_rate": 0.000593695465692601,
      "loss": 2.4531,
      "step": 15062
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6293396353721619,
      "learning_rate": 0.0005936946314654342,
      "loss": 2.3756,
      "step": 15063
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.5996359586715698,
      "learning_rate": 0.0005936937971836638,
      "loss": 2.3773,
      "step": 15064
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.5839152932167053,
      "learning_rate": 0.0005936929628472903,
      "loss": 2.4128,
      "step": 15065
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.636211097240448,
      "learning_rate": 0.0005936921284563138,
      "loss": 2.3414,
      "step": 15066
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6145835518836975,
      "learning_rate": 0.0005936912940107343,
      "loss": 2.4415,
      "step": 15067
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.5716711282730103,
      "learning_rate": 0.000593690459510552,
      "loss": 2.4341,
      "step": 15068
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6737074851989746,
      "learning_rate": 0.0005936896249557673,
      "loss": 2.4775,
      "step": 15069
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6601740717887878,
      "learning_rate": 0.00059368879034638,
      "loss": 2.4067,
      "step": 15070
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6635695099830627,
      "learning_rate": 0.0005936879556823904,
      "loss": 2.2846,
      "step": 15071
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6152868270874023,
      "learning_rate": 0.0005936871209637987,
      "loss": 2.1941,
      "step": 15072
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.8155438899993896,
      "learning_rate": 0.000593686286190605,
      "loss": 2.4801,
      "step": 15073
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.665730893611908,
      "learning_rate": 0.0005936854513628096,
      "loss": 2.2518,
      "step": 15074
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.5925864577293396,
      "learning_rate": 0.0005936846164804123,
      "loss": 2.4081,
      "step": 15075
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.7252152562141418,
      "learning_rate": 0.0005936837815434138,
      "loss": 2.3664,
      "step": 15076
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.7276555299758911,
      "learning_rate": 0.0005936829465518137,
      "loss": 2.561,
      "step": 15077
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.600678563117981,
      "learning_rate": 0.0005936821115056124,
      "loss": 2.2408,
      "step": 15078
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6994792819023132,
      "learning_rate": 0.00059368127640481,
      "loss": 2.3883,
      "step": 15079
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.926769495010376,
      "learning_rate": 0.0005936804412494069,
      "loss": 2.223,
      "step": 15080
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.650928795337677,
      "learning_rate": 0.0005936796060394031,
      "loss": 2.4673,
      "step": 15081
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.7030432820320129,
      "learning_rate": 0.0005936787707747985,
      "loss": 2.533,
      "step": 15082
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.9552426338195801,
      "learning_rate": 0.0005936779354555935,
      "loss": 2.1813,
      "step": 15083
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6562157273292542,
      "learning_rate": 0.0005936771000817884,
      "loss": 2.7291,
      "step": 15084
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6766862869262695,
      "learning_rate": 0.000593676264653383,
      "loss": 2.3502,
      "step": 15085
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.7623746991157532,
      "learning_rate": 0.0005936754291703777,
      "loss": 2.363,
      "step": 15086
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.7299222350120544,
      "learning_rate": 0.0005936745936327726,
      "loss": 2.4,
      "step": 15087
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.5832730531692505,
      "learning_rate": 0.0005936737580405679,
      "loss": 2.4248,
      "step": 15088
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.7030543088912964,
      "learning_rate": 0.0005936729223937636,
      "loss": 2.453,
      "step": 15089
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.9272881150245667,
      "learning_rate": 0.0005936720866923601,
      "loss": 2.4701,
      "step": 15090
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6265828013420105,
      "learning_rate": 0.0005936712509363573,
      "loss": 2.5341,
      "step": 15091
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6805484890937805,
      "learning_rate": 0.0005936704151257555,
      "loss": 2.4691,
      "step": 15092
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.8152106404304504,
      "learning_rate": 0.0005936695792605549,
      "loss": 2.6223,
      "step": 15093
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6554708480834961,
      "learning_rate": 0.0005936687433407555,
      "loss": 2.4235,
      "step": 15094
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6112444400787354,
      "learning_rate": 0.0005936679073663576,
      "loss": 2.3576,
      "step": 15095
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.7584661245346069,
      "learning_rate": 0.0005936670713373611,
      "loss": 2.2581,
      "step": 15096
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.8408022522926331,
      "learning_rate": 0.0005936662352537666,
      "loss": 2.3428,
      "step": 15097
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.5803773403167725,
      "learning_rate": 0.0005936653991155739,
      "loss": 2.5076,
      "step": 15098
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.7911646366119385,
      "learning_rate": 0.0005936645629227832,
      "loss": 2.4815,
      "step": 15099
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.8254378437995911,
      "learning_rate": 0.0005936637266753948,
      "loss": 2.5703,
      "step": 15100
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.5913897752761841,
      "learning_rate": 0.0005936628903734087,
      "loss": 2.4785,
      "step": 15101
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6654983758926392,
      "learning_rate": 0.0005936620540168252,
      "loss": 2.2503,
      "step": 15102
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6424477696418762,
      "learning_rate": 0.0005936612176056443,
      "loss": 2.3107,
      "step": 15103
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.592124342918396,
      "learning_rate": 0.0005936603811398663,
      "loss": 2.5727,
      "step": 15104
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6072965264320374,
      "learning_rate": 0.0005936595446194914,
      "loss": 2.4316,
      "step": 15105
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6308732032775879,
      "learning_rate": 0.0005936587080445194,
      "loss": 2.4283,
      "step": 15106
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.7380020022392273,
      "learning_rate": 0.0005936578714149509,
      "loss": 2.365,
      "step": 15107
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.608380913734436,
      "learning_rate": 0.0005936570347307858,
      "loss": 2.6217,
      "step": 15108
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.613078773021698,
      "learning_rate": 0.0005936561979920243,
      "loss": 2.3273,
      "step": 15109
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6941304802894592,
      "learning_rate": 0.0005936553611986665,
      "loss": 2.481,
      "step": 15110
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6229105591773987,
      "learning_rate": 0.0005936545243507127,
      "loss": 2.4565,
      "step": 15111
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6205213069915771,
      "learning_rate": 0.000593653687448163,
      "loss": 2.4733,
      "step": 15112
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6220144629478455,
      "learning_rate": 0.0005936528504910175,
      "loss": 2.4615,
      "step": 15113
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.7126523852348328,
      "learning_rate": 0.0005936520134792765,
      "loss": 2.5039,
      "step": 15114
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6450642347335815,
      "learning_rate": 0.0005936511764129399,
      "loss": 2.6655,
      "step": 15115
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6011869311332703,
      "learning_rate": 0.000593650339292008,
      "loss": 2.7075,
      "step": 15116
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6532961130142212,
      "learning_rate": 0.000593649502116481,
      "loss": 2.449,
      "step": 15117
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6937238574028015,
      "learning_rate": 0.0005936486648863591,
      "loss": 2.2547,
      "step": 15118
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.7186643481254578,
      "learning_rate": 0.0005936478276016423,
      "loss": 2.4546,
      "step": 15119
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6771155595779419,
      "learning_rate": 0.0005936469902623309,
      "loss": 2.2243,
      "step": 15120
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.7329831123352051,
      "learning_rate": 0.0005936461528684248,
      "loss": 2.3505,
      "step": 15121
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.5662640929222107,
      "learning_rate": 0.0005936453154199245,
      "loss": 2.3109,
      "step": 15122
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.5800113677978516,
      "learning_rate": 0.0005936444779168299,
      "loss": 2.3134,
      "step": 15123
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.7182475328445435,
      "learning_rate": 0.0005936436403591413,
      "loss": 2.5089,
      "step": 15124
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6168128252029419,
      "learning_rate": 0.0005936428027468587,
      "loss": 2.2661,
      "step": 15125
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6255788207054138,
      "learning_rate": 0.0005936419650799824,
      "loss": 2.2934,
      "step": 15126
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6058762669563293,
      "learning_rate": 0.0005936411273585126,
      "loss": 2.4147,
      "step": 15127
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6046686172485352,
      "learning_rate": 0.0005936402895824494,
      "loss": 2.3086,
      "step": 15128
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.5398826599121094,
      "learning_rate": 0.0005936394517517928,
      "loss": 2.3091,
      "step": 15129
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.5583674907684326,
      "learning_rate": 0.0005936386138665432,
      "loss": 2.2775,
      "step": 15130
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6757526993751526,
      "learning_rate": 0.0005936377759267005,
      "loss": 2.2608,
      "step": 15131
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6296241283416748,
      "learning_rate": 0.0005936369379322651,
      "loss": 2.3461,
      "step": 15132
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.5772387385368347,
      "learning_rate": 0.0005936360998832371,
      "loss": 2.372,
      "step": 15133
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.7310851812362671,
      "learning_rate": 0.0005936352617796166,
      "loss": 2.4705,
      "step": 15134
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6327391862869263,
      "learning_rate": 0.0005936344236214036,
      "loss": 2.7055,
      "step": 15135
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.587726354598999,
      "learning_rate": 0.0005936335854085986,
      "loss": 2.0119,
      "step": 15136
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.5725233554840088,
      "learning_rate": 0.0005936327471412014,
      "loss": 2.327,
      "step": 15137
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.5851740837097168,
      "learning_rate": 0.0005936319088192124,
      "loss": 2.4158,
      "step": 15138
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.5730997323989868,
      "learning_rate": 0.0005936310704426318,
      "loss": 2.2719,
      "step": 15139
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.5578925609588623,
      "learning_rate": 0.0005936302320114596,
      "loss": 2.358,
      "step": 15140
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6529173254966736,
      "learning_rate": 0.0005936293935256958,
      "loss": 2.375,
      "step": 15141
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.5571177005767822,
      "learning_rate": 0.0005936285549853409,
      "loss": 2.3139,
      "step": 15142
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6031104922294617,
      "learning_rate": 0.0005936277163903949,
      "loss": 2.1563,
      "step": 15143
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.612139880657196,
      "learning_rate": 0.0005936268777408581,
      "loss": 2.5733,
      "step": 15144
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6466453075408936,
      "learning_rate": 0.0005936260390367303,
      "loss": 2.4286,
      "step": 15145
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.5592882037162781,
      "learning_rate": 0.000593625200278012,
      "loss": 2.5718,
      "step": 15146
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6092969179153442,
      "learning_rate": 0.0005936243614647032,
      "loss": 2.2151,
      "step": 15147
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.5852983593940735,
      "learning_rate": 0.0005936235225968041,
      "loss": 2.4346,
      "step": 15148
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6263770461082458,
      "learning_rate": 0.0005936226836743148,
      "loss": 2.475,
      "step": 15149
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6176319718360901,
      "learning_rate": 0.0005936218446972356,
      "loss": 2.3754,
      "step": 15150
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.7314199209213257,
      "learning_rate": 0.0005936210056655664,
      "loss": 2.4029,
      "step": 15151
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6166898608207703,
      "learning_rate": 0.0005936201665793076,
      "loss": 2.298,
      "step": 15152
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.5908403992652893,
      "learning_rate": 0.0005936193274384594,
      "loss": 2.4829,
      "step": 15153
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6255510449409485,
      "learning_rate": 0.0005936184882430216,
      "loss": 2.3208,
      "step": 15154
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.7107358574867249,
      "learning_rate": 0.0005936176489929946,
      "loss": 2.6129,
      "step": 15155
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.5872095227241516,
      "learning_rate": 0.0005936168096883787,
      "loss": 2.398,
      "step": 15156
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.7711706161499023,
      "learning_rate": 0.0005936159703291738,
      "loss": 2.0368,
      "step": 15157
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.5705687403678894,
      "learning_rate": 0.0005936151309153801,
      "loss": 2.2543,
      "step": 15158
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6002491116523743,
      "learning_rate": 0.000593614291446998,
      "loss": 2.4116,
      "step": 15159
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6092900633811951,
      "learning_rate": 0.0005936134519240272,
      "loss": 2.1952,
      "step": 15160
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6036258935928345,
      "learning_rate": 0.0005936126123464682,
      "loss": 2.4516,
      "step": 15161
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6325948238372803,
      "learning_rate": 0.0005936117727143212,
      "loss": 1.9857,
      "step": 15162
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.7067246437072754,
      "learning_rate": 0.0005936109330275861,
      "loss": 2.3368,
      "step": 15163
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.5719369053840637,
      "learning_rate": 0.0005936100932862632,
      "loss": 2.4259,
      "step": 15164
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.1497621536254883,
      "learning_rate": 0.0005936092534903528,
      "loss": 2.1938,
      "step": 15165
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6132294535636902,
      "learning_rate": 0.0005936084136398547,
      "loss": 2.4583,
      "step": 15166
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6202477216720581,
      "learning_rate": 0.0005936075737347693,
      "loss": 2.4116,
      "step": 15167
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.631055474281311,
      "learning_rate": 0.0005936067337750967,
      "loss": 2.4181,
      "step": 15168
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.7180120944976807,
      "learning_rate": 0.0005936058937608371,
      "loss": 2.4376,
      "step": 15169
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.5442050695419312,
      "learning_rate": 0.0005936050536919906,
      "loss": 2.0362,
      "step": 15170
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6022278666496277,
      "learning_rate": 0.0005936042135685573,
      "loss": 2.1875,
      "step": 15171
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.5685933232307434,
      "learning_rate": 0.0005936033733905376,
      "loss": 2.4751,
      "step": 15172
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6456491947174072,
      "learning_rate": 0.0005936025331579315,
      "loss": 2.6686,
      "step": 15173
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6168811321258545,
      "learning_rate": 0.000593601692870739,
      "loss": 2.603,
      "step": 15174
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6079816222190857,
      "learning_rate": 0.0005936008525289605,
      "loss": 2.5238,
      "step": 15175
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.5967622995376587,
      "learning_rate": 0.0005936000121325959,
      "loss": 2.4758,
      "step": 15176
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.5961902737617493,
      "learning_rate": 0.0005935991716816458,
      "loss": 2.3864,
      "step": 15177
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.5749253034591675,
      "learning_rate": 0.0005935983311761098,
      "loss": 2.4029,
      "step": 15178
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6428055763244629,
      "learning_rate": 0.0005935974906159886,
      "loss": 2.4311,
      "step": 15179
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.692939817905426,
      "learning_rate": 0.0005935966500012818,
      "loss": 2.3155,
      "step": 15180
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.5451411008834839,
      "learning_rate": 0.00059359580933199,
      "loss": 2.3342,
      "step": 15181
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6178564429283142,
      "learning_rate": 0.0005935949686081132,
      "loss": 2.1226,
      "step": 15182
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6162519454956055,
      "learning_rate": 0.0005935941278296515,
      "loss": 2.502,
      "step": 15183
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.5722342133522034,
      "learning_rate": 0.0005935932869966051,
      "loss": 2.4512,
      "step": 15184
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6145076155662537,
      "learning_rate": 0.0005935924461089742,
      "loss": 2.2975,
      "step": 15185
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.601473867893219,
      "learning_rate": 0.0005935916051667589,
      "loss": 2.3443,
      "step": 15186
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6296748518943787,
      "learning_rate": 0.0005935907641699595,
      "loss": 2.6644,
      "step": 15187
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6023319959640503,
      "learning_rate": 0.0005935899231185759,
      "loss": 2.3048,
      "step": 15188
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.5685650110244751,
      "learning_rate": 0.0005935890820126083,
      "loss": 2.2721,
      "step": 15189
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6079946160316467,
      "learning_rate": 0.0005935882408520572,
      "loss": 2.4258,
      "step": 15190
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.5810761451721191,
      "learning_rate": 0.0005935873996369223,
      "loss": 2.4407,
      "step": 15191
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6075600385665894,
      "learning_rate": 0.0005935865583672041,
      "loss": 2.3557,
      "step": 15192
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6572250723838806,
      "learning_rate": 0.0005935857170429026,
      "loss": 2.3805,
      "step": 15193
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6110928654670715,
      "learning_rate": 0.0005935848756640178,
      "loss": 2.2416,
      "step": 15194
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6239150762557983,
      "learning_rate": 0.0005935840342305502,
      "loss": 2.5596,
      "step": 15195
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6217668056488037,
      "learning_rate": 0.0005935831927424997,
      "loss": 2.4012,
      "step": 15196
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.5685409307479858,
      "learning_rate": 0.0005935823511998667,
      "loss": 2.2921,
      "step": 15197
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.60346919298172,
      "learning_rate": 0.000593581509602651,
      "loss": 2.4758,
      "step": 15198
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.5936671495437622,
      "learning_rate": 0.0005935806679508531,
      "loss": 2.5647,
      "step": 15199
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.7258913516998291,
      "learning_rate": 0.0005935798262444729,
      "loss": 2.2762,
      "step": 15200
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.5819043517112732,
      "learning_rate": 0.0005935789844835107,
      "loss": 2.4842,
      "step": 15201
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6761916279792786,
      "learning_rate": 0.0005935781426679666,
      "loss": 2.4549,
      "step": 15202
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.8473602533340454,
      "learning_rate": 0.0005935773007978408,
      "loss": 2.505,
      "step": 15203
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6125186085700989,
      "learning_rate": 0.0005935764588731336,
      "loss": 2.3386,
      "step": 15204
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.7346956729888916,
      "learning_rate": 0.0005935756168938448,
      "loss": 2.2023,
      "step": 15205
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.7575731873512268,
      "learning_rate": 0.0005935747748599748,
      "loss": 2.4503,
      "step": 15206
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6816880106925964,
      "learning_rate": 0.0005935739327715237,
      "loss": 2.292,
      "step": 15207
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.7348017692565918,
      "learning_rate": 0.0005935730906284917,
      "loss": 2.5525,
      "step": 15208
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.842266321182251,
      "learning_rate": 0.0005935722484308788,
      "loss": 2.3828,
      "step": 15209
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.5945667624473572,
      "learning_rate": 0.0005935714061786855,
      "loss": 2.4352,
      "step": 15210
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.7872348427772522,
      "learning_rate": 0.0005935705638719116,
      "loss": 2.3209,
      "step": 15211
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.9007721543312073,
      "learning_rate": 0.0005935697215105573,
      "loss": 2.6659,
      "step": 15212
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6146591901779175,
      "learning_rate": 0.000593568879094623,
      "loss": 2.2602,
      "step": 15213
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.7248380184173584,
      "learning_rate": 0.0005935680366241087,
      "loss": 2.4382,
      "step": 15214
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.2604434490203857,
      "learning_rate": 0.0005935671940990145,
      "loss": 2.4925,
      "step": 15215
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.5838544964790344,
      "learning_rate": 0.0005935663515193406,
      "loss": 2.2458,
      "step": 15216
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6269239187240601,
      "learning_rate": 0.0005935655088850873,
      "loss": 2.3044,
      "step": 15217
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.5874547958374023,
      "learning_rate": 0.0005935646661962544,
      "loss": 2.2928,
      "step": 15218
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6018447875976562,
      "learning_rate": 0.0005935638234528426,
      "loss": 2.4328,
      "step": 15219
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.549118161201477,
      "learning_rate": 0.0005935629806548514,
      "loss": 2.5234,
      "step": 15220
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.5849975943565369,
      "learning_rate": 0.0005935621378022816,
      "loss": 2.3198,
      "step": 15221
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6975026726722717,
      "learning_rate": 0.0005935612948951329,
      "loss": 2.4692,
      "step": 15222
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6136564612388611,
      "learning_rate": 0.0005935604519334057,
      "loss": 2.3003,
      "step": 15223
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.5727823376655579,
      "learning_rate": 0.0005935596089170999,
      "loss": 2.5597,
      "step": 15224
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.5812771320343018,
      "learning_rate": 0.000593558765846216,
      "loss": 2.4453,
      "step": 15225
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.5485468506813049,
      "learning_rate": 0.000593557922720754,
      "loss": 2.4352,
      "step": 15226
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6165515184402466,
      "learning_rate": 0.0005935570795407138,
      "loss": 2.5676,
      "step": 15227
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.5667301416397095,
      "learning_rate": 0.000593556236306096,
      "loss": 2.6122,
      "step": 15228
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.7147234082221985,
      "learning_rate": 0.0005935553930169006,
      "loss": 2.1892,
      "step": 15229
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6031506657600403,
      "learning_rate": 0.0005935545496731275,
      "loss": 2.4559,
      "step": 15230
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6416369080543518,
      "learning_rate": 0.0005935537062747772,
      "loss": 2.6141,
      "step": 15231
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6013379096984863,
      "learning_rate": 0.0005935528628218498,
      "loss": 2.2689,
      "step": 15232
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.607873797416687,
      "learning_rate": 0.0005935520193143453,
      "loss": 2.3621,
      "step": 15233
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6315118074417114,
      "learning_rate": 0.0005935511757522639,
      "loss": 2.2866,
      "step": 15234
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.7006651163101196,
      "learning_rate": 0.0005935503321356058,
      "loss": 2.4649,
      "step": 15235
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6299013495445251,
      "learning_rate": 0.0005935494884643712,
      "loss": 2.2401,
      "step": 15236
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6517322659492493,
      "learning_rate": 0.0005935486447385602,
      "loss": 2.2413,
      "step": 15237
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6205085515975952,
      "learning_rate": 0.000593547800958173,
      "loss": 2.3545,
      "step": 15238
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.576093852519989,
      "learning_rate": 0.0005935469571232097,
      "loss": 2.4298,
      "step": 15239
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.5985701680183411,
      "learning_rate": 0.0005935461132336703,
      "loss": 2.2306,
      "step": 15240
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6493507027626038,
      "learning_rate": 0.0005935452692895554,
      "loss": 2.3328,
      "step": 15241
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6532111167907715,
      "learning_rate": 0.0005935444252908648,
      "loss": 2.5325,
      "step": 15242
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6675894856452942,
      "learning_rate": 0.0005935435812375987,
      "loss": 2.4595,
      "step": 15243
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6587591171264648,
      "learning_rate": 0.0005935427371297574,
      "loss": 2.4758,
      "step": 15244
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6500464081764221,
      "learning_rate": 0.0005935418929673409,
      "loss": 2.4826,
      "step": 15245
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6281213164329529,
      "learning_rate": 0.0005935410487503494,
      "loss": 2.2776,
      "step": 15246
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.68175208568573,
      "learning_rate": 0.0005935402044787831,
      "loss": 2.4081,
      "step": 15247
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.621636688709259,
      "learning_rate": 0.0005935393601526422,
      "loss": 2.5797,
      "step": 15248
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.5912261605262756,
      "learning_rate": 0.0005935385157719267,
      "loss": 2.4239,
      "step": 15249
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.5831478834152222,
      "learning_rate": 0.0005935376713366369,
      "loss": 2.5325,
      "step": 15250
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6131815910339355,
      "learning_rate": 0.0005935368268467729,
      "loss": 2.5364,
      "step": 15251
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6435122489929199,
      "learning_rate": 0.0005935359823023348,
      "loss": 2.0774,
      "step": 15252
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6668978333473206,
      "learning_rate": 0.0005935351377033229,
      "loss": 2.4149,
      "step": 15253
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.5722616314888,
      "learning_rate": 0.0005935342930497372,
      "loss": 2.4523,
      "step": 15254
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.5758702158927917,
      "learning_rate": 0.0005935334483415781,
      "loss": 2.3446,
      "step": 15255
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.5914912223815918,
      "learning_rate": 0.0005935326035788455,
      "loss": 2.2438,
      "step": 15256
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6100442409515381,
      "learning_rate": 0.0005935317587615395,
      "loss": 2.3268,
      "step": 15257
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6201992034912109,
      "learning_rate": 0.0005935309138896607,
      "loss": 2.2906,
      "step": 15258
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6016458868980408,
      "learning_rate": 0.0005935300689632087,
      "loss": 2.4352,
      "step": 15259
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6592903733253479,
      "learning_rate": 0.000593529223982184,
      "loss": 2.5455,
      "step": 15260
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6568964719772339,
      "learning_rate": 0.0005935283789465867,
      "loss": 2.4428,
      "step": 15261
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.7588028907775879,
      "learning_rate": 0.0005935275338564169,
      "loss": 2.5424,
      "step": 15262
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6457856893539429,
      "learning_rate": 0.0005935266887116747,
      "loss": 2.6876,
      "step": 15263
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6577922701835632,
      "learning_rate": 0.0005935258435123606,
      "loss": 2.3866,
      "step": 15264
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.5363904237747192,
      "learning_rate": 0.0005935249982584744,
      "loss": 2.3144,
      "step": 15265
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.649785578250885,
      "learning_rate": 0.0005935241529500163,
      "loss": 2.3588,
      "step": 15266
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.5850895047187805,
      "learning_rate": 0.0005935233075869866,
      "loss": 2.4584,
      "step": 15267
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6207150816917419,
      "learning_rate": 0.0005935224621693853,
      "loss": 2.5563,
      "step": 15268
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6687473654747009,
      "learning_rate": 0.0005935216166972127,
      "loss": 2.7065,
      "step": 15269
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.5980244278907776,
      "learning_rate": 0.0005935207711704688,
      "loss": 2.2954,
      "step": 15270
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6530675888061523,
      "learning_rate": 0.0005935199255891539,
      "loss": 2.7156,
      "step": 15271
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6089553833007812,
      "learning_rate": 0.0005935190799532681,
      "loss": 2.4854,
      "step": 15272
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.5809222459793091,
      "learning_rate": 0.0005935182342628115,
      "loss": 2.4997,
      "step": 15273
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6501213908195496,
      "learning_rate": 0.0005935173885177845,
      "loss": 2.3759,
      "step": 15274
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6021178364753723,
      "learning_rate": 0.0005935165427181869,
      "loss": 2.4787,
      "step": 15275
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6307219862937927,
      "learning_rate": 0.0005935156968640192,
      "loss": 2.2218,
      "step": 15276
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6421147584915161,
      "learning_rate": 0.0005935148509552811,
      "loss": 2.6799,
      "step": 15277
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.605219841003418,
      "learning_rate": 0.0005935140049919734,
      "loss": 2.2951,
      "step": 15278
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.5937692523002625,
      "learning_rate": 0.0005935131589740957,
      "loss": 2.4176,
      "step": 15279
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.5918068885803223,
      "learning_rate": 0.0005935123129016483,
      "loss": 2.4837,
      "step": 15280
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.5797212719917297,
      "learning_rate": 0.0005935114667746316,
      "loss": 2.1216,
      "step": 15281
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6369115114212036,
      "learning_rate": 0.0005935106205930455,
      "loss": 2.3984,
      "step": 15282
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6028324961662292,
      "learning_rate": 0.0005935097743568902,
      "loss": 2.6555,
      "step": 15283
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6015257835388184,
      "learning_rate": 0.000593508928066166,
      "loss": 2.4583,
      "step": 15284
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.5667284727096558,
      "learning_rate": 0.0005935080817208728,
      "loss": 2.5763,
      "step": 15285
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.633701503276825,
      "learning_rate": 0.000593507235321011,
      "loss": 2.6436,
      "step": 15286
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.614382266998291,
      "learning_rate": 0.0005935063888665807,
      "loss": 2.3973,
      "step": 15287
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.538118839263916,
      "learning_rate": 0.0005935055423575819,
      "loss": 2.641,
      "step": 15288
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6167306303977966,
      "learning_rate": 0.000593504695794015,
      "loss": 2.4375,
      "step": 15289
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6320206522941589,
      "learning_rate": 0.00059350384917588,
      "loss": 2.459,
      "step": 15290
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.5697451829910278,
      "learning_rate": 0.0005935030025031771,
      "loss": 2.5955,
      "step": 15291
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.5783919095993042,
      "learning_rate": 0.0005935021557759065,
      "loss": 2.1938,
      "step": 15292
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6383456587791443,
      "learning_rate": 0.0005935013089940683,
      "loss": 2.5809,
      "step": 15293
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.5664982795715332,
      "learning_rate": 0.0005935004621576625,
      "loss": 2.3622,
      "step": 15294
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6384258270263672,
      "learning_rate": 0.0005934996152666895,
      "loss": 2.4024,
      "step": 15295
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.5679977536201477,
      "learning_rate": 0.0005934987683211495,
      "loss": 2.3247,
      "step": 15296
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6085945963859558,
      "learning_rate": 0.0005934979213210424,
      "loss": 2.525,
      "step": 15297
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6717448234558105,
      "learning_rate": 0.0005934970742663685,
      "loss": 2.4017,
      "step": 15298
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.5870187282562256,
      "learning_rate": 0.0005934962271571281,
      "loss": 2.3806,
      "step": 15299
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.7318888902664185,
      "learning_rate": 0.0005934953799933211,
      "loss": 2.503,
      "step": 15300
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6237927675247192,
      "learning_rate": 0.0005934945327749477,
      "loss": 2.3106,
      "step": 15301
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6340026259422302,
      "learning_rate": 0.0005934936855020082,
      "loss": 2.3049,
      "step": 15302
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6294072866439819,
      "learning_rate": 0.0005934928381745027,
      "loss": 2.2475,
      "step": 15303
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6144286394119263,
      "learning_rate": 0.0005934919907924314,
      "loss": 2.7447,
      "step": 15304
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6166226863861084,
      "learning_rate": 0.0005934911433557943,
      "loss": 2.6219,
      "step": 15305
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6088989973068237,
      "learning_rate": 0.0005934902958645917,
      "loss": 2.3542,
      "step": 15306
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.5826570391654968,
      "learning_rate": 0.0005934894483188237,
      "loss": 2.167,
      "step": 15307
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.5767415761947632,
      "learning_rate": 0.0005934886007184904,
      "loss": 2.2983,
      "step": 15308
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6135898232460022,
      "learning_rate": 0.0005934877530635922,
      "loss": 2.4519,
      "step": 15309
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6401816606521606,
      "learning_rate": 0.0005934869053541291,
      "loss": 2.3399,
      "step": 15310
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6476436853408813,
      "learning_rate": 0.000593486057590101,
      "loss": 2.4881,
      "step": 15311
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6809312701225281,
      "learning_rate": 0.0005934852097715085,
      "loss": 2.4075,
      "step": 15312
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6230345964431763,
      "learning_rate": 0.0005934843618983515,
      "loss": 2.4018,
      "step": 15313
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6175747513771057,
      "learning_rate": 0.0005934835139706302,
      "loss": 2.3941,
      "step": 15314
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6396057605743408,
      "learning_rate": 0.0005934826659883448,
      "loss": 2.3305,
      "step": 15315
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6349465250968933,
      "learning_rate": 0.0005934818179514954,
      "loss": 2.3804,
      "step": 15316
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6850021481513977,
      "learning_rate": 0.0005934809698600823,
      "loss": 2.5603,
      "step": 15317
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.621196985244751,
      "learning_rate": 0.0005934801217141053,
      "loss": 2.4001,
      "step": 15318
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6222901940345764,
      "learning_rate": 0.000593479273513565,
      "loss": 2.2649,
      "step": 15319
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6271321773529053,
      "learning_rate": 0.0005934784252584614,
      "loss": 2.3886,
      "step": 15320
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.588230311870575,
      "learning_rate": 0.0005934775769487945,
      "loss": 2.6625,
      "step": 15321
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6753333210945129,
      "learning_rate": 0.0005934767285845646,
      "loss": 2.3249,
      "step": 15322
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6203591227531433,
      "learning_rate": 0.0005934758801657719,
      "loss": 2.4299,
      "step": 15323
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6933040022850037,
      "learning_rate": 0.0005934750316924165,
      "loss": 2.2322,
      "step": 15324
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.640418529510498,
      "learning_rate": 0.0005934741831644984,
      "loss": 2.3208,
      "step": 15325
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6799264550209045,
      "learning_rate": 0.0005934733345820181,
      "loss": 2.3018,
      "step": 15326
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.5502544641494751,
      "learning_rate": 0.0005934724859449755,
      "loss": 2.405,
      "step": 15327
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.5816115736961365,
      "learning_rate": 0.0005934716372533708,
      "loss": 2.3336,
      "step": 15328
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6515821814537048,
      "learning_rate": 0.0005934707885072042,
      "loss": 2.1121,
      "step": 15329
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6029374599456787,
      "learning_rate": 0.0005934699397064759,
      "loss": 2.54,
      "step": 15330
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.5581336617469788,
      "learning_rate": 0.0005934690908511858,
      "loss": 2.4426,
      "step": 15331
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6410341262817383,
      "learning_rate": 0.0005934682419413345,
      "loss": 2.4391,
      "step": 15332
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6693092584609985,
      "learning_rate": 0.0005934673929769218,
      "loss": 2.363,
      "step": 15333
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6412435173988342,
      "learning_rate": 0.000593466543957948,
      "loss": 2.3445,
      "step": 15334
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.7962426543235779,
      "learning_rate": 0.0005934656948844132,
      "loss": 2.2272,
      "step": 15335
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6324183344841003,
      "learning_rate": 0.0005934648457563176,
      "loss": 2.4486,
      "step": 15336
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.5902553200721741,
      "learning_rate": 0.0005934639965736614,
      "loss": 2.3064,
      "step": 15337
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6790423393249512,
      "learning_rate": 0.0005934631473364445,
      "loss": 2.2206,
      "step": 15338
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6507657766342163,
      "learning_rate": 0.0005934622980446675,
      "loss": 2.3821,
      "step": 15339
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6606670022010803,
      "learning_rate": 0.0005934614486983302,
      "loss": 2.4857,
      "step": 15340
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6294838786125183,
      "learning_rate": 0.0005934605992974328,
      "loss": 2.2539,
      "step": 15341
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.7426590919494629,
      "learning_rate": 0.0005934597498419757,
      "loss": 2.5032,
      "step": 15342
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.5966631770133972,
      "learning_rate": 0.0005934589003319589,
      "loss": 2.4905,
      "step": 15343
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6528815031051636,
      "learning_rate": 0.0005934580507673824,
      "loss": 2.4166,
      "step": 15344
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6050369739532471,
      "learning_rate": 0.0005934572011482466,
      "loss": 2.3126,
      "step": 15345
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6190921068191528,
      "learning_rate": 0.0005934563514745515,
      "loss": 2.3584,
      "step": 15346
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6567434668540955,
      "learning_rate": 0.0005934555017462975,
      "loss": 2.2452,
      "step": 15347
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.7602118253707886,
      "learning_rate": 0.0005934546519634844,
      "loss": 2.409,
      "step": 15348
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6167252063751221,
      "learning_rate": 0.0005934538021261127,
      "loss": 2.6278,
      "step": 15349
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6126117706298828,
      "learning_rate": 0.0005934529522341823,
      "loss": 2.4161,
      "step": 15350
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.586471438407898,
      "learning_rate": 0.0005934521022876935,
      "loss": 2.6066,
      "step": 15351
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.7005413770675659,
      "learning_rate": 0.0005934512522866463,
      "loss": 2.3333,
      "step": 15352
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6568750143051147,
      "learning_rate": 0.000593450402231041,
      "loss": 2.5568,
      "step": 15353
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6410946249961853,
      "learning_rate": 0.0005934495521208778,
      "loss": 2.4564,
      "step": 15354
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.5908768177032471,
      "learning_rate": 0.0005934487019561568,
      "loss": 2.4847,
      "step": 15355
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.609472930431366,
      "learning_rate": 0.0005934478517368782,
      "loss": 2.6408,
      "step": 15356
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.5959329009056091,
      "learning_rate": 0.0005934470014630419,
      "loss": 2.4731,
      "step": 15357
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.5635244846343994,
      "learning_rate": 0.0005934461511346485,
      "loss": 2.2445,
      "step": 15358
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6022559404373169,
      "learning_rate": 0.0005934453007516977,
      "loss": 2.2625,
      "step": 15359
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6149040460586548,
      "learning_rate": 0.00059344445031419,
      "loss": 2.3498,
      "step": 15360
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6104057431221008,
      "learning_rate": 0.0005934435998221255,
      "loss": 2.1545,
      "step": 15361
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.5910236835479736,
      "learning_rate": 0.0005934427492755042,
      "loss": 2.4114,
      "step": 15362
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.5993151664733887,
      "learning_rate": 0.0005934418986743263,
      "loss": 2.2517,
      "step": 15363
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.5971941947937012,
      "learning_rate": 0.0005934410480185921,
      "loss": 2.4402,
      "step": 15364
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6277773380279541,
      "learning_rate": 0.0005934401973083017,
      "loss": 2.2263,
      "step": 15365
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.5850499272346497,
      "learning_rate": 0.0005934393465434551,
      "loss": 2.2866,
      "step": 15366
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.606015145778656,
      "learning_rate": 0.0005934384957240527,
      "loss": 2.2205,
      "step": 15367
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.5592811703681946,
      "learning_rate": 0.0005934376448500945,
      "loss": 2.3953,
      "step": 15368
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.64114910364151,
      "learning_rate": 0.0005934367939215807,
      "loss": 2.2734,
      "step": 15369
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.555397629737854,
      "learning_rate": 0.0005934359429385115,
      "loss": 2.5293,
      "step": 15370
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.5827817320823669,
      "learning_rate": 0.000593435091900887,
      "loss": 2.3612,
      "step": 15371
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.713438868522644,
      "learning_rate": 0.0005934342408087074,
      "loss": 2.3527,
      "step": 15372
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6398415565490723,
      "learning_rate": 0.0005934333896619728,
      "loss": 2.3411,
      "step": 15373
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.585882842540741,
      "learning_rate": 0.0005934325384606833,
      "loss": 2.31,
      "step": 15374
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.652239203453064,
      "learning_rate": 0.0005934316872048394,
      "loss": 2.5263,
      "step": 15375
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6796143054962158,
      "learning_rate": 0.0005934308358944408,
      "loss": 2.1109,
      "step": 15376
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6098994612693787,
      "learning_rate": 0.000593429984529488,
      "loss": 2.5681,
      "step": 15377
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.5907787084579468,
      "learning_rate": 0.000593429133109981,
      "loss": 2.5748,
      "step": 15378
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6465650796890259,
      "learning_rate": 0.0005934282816359199,
      "loss": 2.5945,
      "step": 15379
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.593849778175354,
      "learning_rate": 0.0005934274301073049,
      "loss": 2.5618,
      "step": 15380
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.5608627796173096,
      "learning_rate": 0.0005934265785241363,
      "loss": 2.1727,
      "step": 15381
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.5820915102958679,
      "learning_rate": 0.0005934257268864142,
      "loss": 2.5442,
      "step": 15382
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.599639356136322,
      "learning_rate": 0.0005934248751941387,
      "loss": 2.3385,
      "step": 15383
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6555098295211792,
      "learning_rate": 0.0005934240234473099,
      "loss": 2.3104,
      "step": 15384
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.5922607183456421,
      "learning_rate": 0.0005934231716459282,
      "loss": 2.4283,
      "step": 15385
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.640411913394928,
      "learning_rate": 0.0005934223197899934,
      "loss": 2.6407,
      "step": 15386
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6342067122459412,
      "learning_rate": 0.000593421467879506,
      "loss": 2.3659,
      "step": 15387
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6616266965866089,
      "learning_rate": 0.0005934206159144658,
      "loss": 2.4473,
      "step": 15388
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6113191843032837,
      "learning_rate": 0.0005934197638948733,
      "loss": 2.3086,
      "step": 15389
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6713930368423462,
      "learning_rate": 0.0005934189118207287,
      "loss": 2.4656,
      "step": 15390
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.5961610674858093,
      "learning_rate": 0.0005934180596920317,
      "loss": 2.4709,
      "step": 15391
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6954433917999268,
      "learning_rate": 0.0005934172075087829,
      "loss": 2.452,
      "step": 15392
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.5689604878425598,
      "learning_rate": 0.0005934163552709822,
      "loss": 2.2324,
      "step": 15393
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.5908241868019104,
      "learning_rate": 0.00059341550297863,
      "loss": 2.4028,
      "step": 15394
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.5955140590667725,
      "learning_rate": 0.0005934146506317262,
      "loss": 2.3862,
      "step": 15395
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6698179244995117,
      "learning_rate": 0.0005934137982302712,
      "loss": 2.2003,
      "step": 15396
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6594130992889404,
      "learning_rate": 0.000593412945774265,
      "loss": 2.3482,
      "step": 15397
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6164024472236633,
      "learning_rate": 0.0005934120932637076,
      "loss": 2.3826,
      "step": 15398
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6040437817573547,
      "learning_rate": 0.0005934112406985996,
      "loss": 2.3369,
      "step": 15399
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6937112212181091,
      "learning_rate": 0.0005934103880789408,
      "loss": 2.3375,
      "step": 15400
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.7237913012504578,
      "learning_rate": 0.0005934095354047314,
      "loss": 2.4444,
      "step": 15401
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6214109659194946,
      "learning_rate": 0.0005934086826759718,
      "loss": 2.3949,
      "step": 15402
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6498422026634216,
      "learning_rate": 0.0005934078298926619,
      "loss": 2.4226,
      "step": 15403
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6137441992759705,
      "learning_rate": 0.000593406977054802,
      "loss": 2.377,
      "step": 15404
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6765185594558716,
      "learning_rate": 0.0005934061241623922,
      "loss": 2.2122,
      "step": 15405
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6957813501358032,
      "learning_rate": 0.0005934052712154326,
      "loss": 2.3183,
      "step": 15406
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.5803024172782898,
      "learning_rate": 0.0005934044182139234,
      "loss": 2.3741,
      "step": 15407
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6500717401504517,
      "learning_rate": 0.0005934035651578648,
      "loss": 2.3865,
      "step": 15408
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.5890194773674011,
      "learning_rate": 0.000593402712047257,
      "loss": 2.5356,
      "step": 15409
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.5512077212333679,
      "learning_rate": 0.0005934018588821,
      "loss": 2.5617,
      "step": 15410
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6133751273155212,
      "learning_rate": 0.0005934010056623942,
      "loss": 2.6278,
      "step": 15411
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6453882455825806,
      "learning_rate": 0.0005934001523881394,
      "loss": 2.3114,
      "step": 15412
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6679003238677979,
      "learning_rate": 0.0005933992990593362,
      "loss": 2.5421,
      "step": 15413
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.5583257079124451,
      "learning_rate": 0.0005933984456759845,
      "loss": 2.4824,
      "step": 15414
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6100863814353943,
      "learning_rate": 0.0005933975922380844,
      "loss": 2.3298,
      "step": 15415
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.5952521562576294,
      "learning_rate": 0.0005933967387456361,
      "loss": 2.4968,
      "step": 15416
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.5204778909683228,
      "learning_rate": 0.00059339588519864,
      "loss": 2.363,
      "step": 15417
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6441508531570435,
      "learning_rate": 0.0005933950315970959,
      "loss": 2.0648,
      "step": 15418
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.5723034143447876,
      "learning_rate": 0.0005933941779410042,
      "loss": 2.2918,
      "step": 15419
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6311332583427429,
      "learning_rate": 0.0005933933242303649,
      "loss": 2.3259,
      "step": 15420
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6067522764205933,
      "learning_rate": 0.0005933924704651783,
      "loss": 2.2548,
      "step": 15421
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6166906952857971,
      "learning_rate": 0.0005933916166454445,
      "loss": 2.394,
      "step": 15422
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6240358948707581,
      "learning_rate": 0.0005933907627711637,
      "loss": 2.2568,
      "step": 15423
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.7557345628738403,
      "learning_rate": 0.000593389908842336,
      "loss": 2.4674,
      "step": 15424
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.642252504825592,
      "learning_rate": 0.0005933890548589615,
      "loss": 2.5344,
      "step": 15425
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.7854491472244263,
      "learning_rate": 0.0005933882008210405,
      "loss": 2.3235,
      "step": 15426
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6482799649238586,
      "learning_rate": 0.0005933873467285731,
      "loss": 2.3752,
      "step": 15427
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6244904398918152,
      "learning_rate": 0.0005933864925815594,
      "loss": 2.3325,
      "step": 15428
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6776254177093506,
      "learning_rate": 0.0005933856383799997,
      "loss": 2.4171,
      "step": 15429
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6643375754356384,
      "learning_rate": 0.000593384784123894,
      "loss": 2.4278,
      "step": 15430
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.573250412940979,
      "learning_rate": 0.0005933839298132425,
      "loss": 2.6802,
      "step": 15431
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6310175657272339,
      "learning_rate": 0.0005933830754480456,
      "loss": 2.5368,
      "step": 15432
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.7865091562271118,
      "learning_rate": 0.000593382221028303,
      "loss": 2.1313,
      "step": 15433
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.564699649810791,
      "learning_rate": 0.0005933813665540153,
      "loss": 2.4086,
      "step": 15434
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.7076418399810791,
      "learning_rate": 0.0005933805120251824,
      "loss": 2.403,
      "step": 15435
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.7006324529647827,
      "learning_rate": 0.0005933796574418044,
      "loss": 2.3609,
      "step": 15436
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6273160576820374,
      "learning_rate": 0.0005933788028038817,
      "loss": 2.4185,
      "step": 15437
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.5722918510437012,
      "learning_rate": 0.0005933779481114143,
      "loss": 2.2694,
      "step": 15438
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.629588782787323,
      "learning_rate": 0.0005933770933644024,
      "loss": 2.3284,
      "step": 15439
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.8610945343971252,
      "learning_rate": 0.0005933762385628463,
      "loss": 2.5924,
      "step": 15440
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.7317659258842468,
      "learning_rate": 0.0005933753837067458,
      "loss": 2.3754,
      "step": 15441
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.7992863059043884,
      "learning_rate": 0.0005933745287961014,
      "loss": 2.3751,
      "step": 15442
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6510744094848633,
      "learning_rate": 0.0005933736738309132,
      "loss": 2.4121,
      "step": 15443
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.686255931854248,
      "learning_rate": 0.0005933728188111812,
      "loss": 2.4581,
      "step": 15444
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6670354604721069,
      "learning_rate": 0.0005933719637369057,
      "loss": 2.5766,
      "step": 15445
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.7322503924369812,
      "learning_rate": 0.0005933711086080867,
      "loss": 2.3571,
      "step": 15446
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.672508716583252,
      "learning_rate": 0.0005933702534247246,
      "loss": 2.3869,
      "step": 15447
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.7996103763580322,
      "learning_rate": 0.0005933693981868194,
      "loss": 2.2045,
      "step": 15448
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.5769756436347961,
      "learning_rate": 0.0005933685428943712,
      "loss": 2.4555,
      "step": 15449
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6228584051132202,
      "learning_rate": 0.0005933676875473804,
      "loss": 2.5787,
      "step": 15450
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.7231734991073608,
      "learning_rate": 0.0005933668321458469,
      "loss": 2.155,
      "step": 15451
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6644999980926514,
      "learning_rate": 0.000593365976689771,
      "loss": 2.2296,
      "step": 15452
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.5850051045417786,
      "learning_rate": 0.0005933651211791529,
      "loss": 2.3929,
      "step": 15453
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6611945033073425,
      "learning_rate": 0.0005933642656139926,
      "loss": 2.4128,
      "step": 15454
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6563112735748291,
      "learning_rate": 0.0005933634099942904,
      "loss": 2.3598,
      "step": 15455
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.61151123046875,
      "learning_rate": 0.0005933625543200464,
      "loss": 2.5417,
      "step": 15456
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.582541823387146,
      "learning_rate": 0.0005933616985912607,
      "loss": 2.2825,
      "step": 15457
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.5995410084724426,
      "learning_rate": 0.0005933608428079336,
      "loss": 2.3866,
      "step": 15458
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6708210110664368,
      "learning_rate": 0.000593359986970065,
      "loss": 2.3446,
      "step": 15459
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.604215681552887,
      "learning_rate": 0.0005933591310776555,
      "loss": 2.1989,
      "step": 15460
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6122661828994751,
      "learning_rate": 0.0005933582751307049,
      "loss": 2.5199,
      "step": 15461
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6836912035942078,
      "learning_rate": 0.0005933574191292134,
      "loss": 2.3404,
      "step": 15462
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6337041258811951,
      "learning_rate": 0.0005933565630731813,
      "loss": 2.4586,
      "step": 15463
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6533780097961426,
      "learning_rate": 0.0005933557069626085,
      "loss": 2.4852,
      "step": 15464
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6567960381507874,
      "learning_rate": 0.0005933548507974955,
      "loss": 2.5165,
      "step": 15465
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.5900282263755798,
      "learning_rate": 0.0005933539945778423,
      "loss": 2.3446,
      "step": 15466
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.5964122414588928,
      "learning_rate": 0.000593353138303649,
      "loss": 2.2909,
      "step": 15467
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6367158889770508,
      "learning_rate": 0.0005933522819749159,
      "loss": 2.2292,
      "step": 15468
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.5684940218925476,
      "learning_rate": 0.000593351425591643,
      "loss": 2.3911,
      "step": 15469
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.5869200229644775,
      "learning_rate": 0.0005933505691538306,
      "loss": 2.4617,
      "step": 15470
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6326208710670471,
      "learning_rate": 0.0005933497126614787,
      "loss": 2.3472,
      "step": 15471
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.7087582945823669,
      "learning_rate": 0.0005933488561145876,
      "loss": 2.5513,
      "step": 15472
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.7364493608474731,
      "learning_rate": 0.0005933479995131574,
      "loss": 2.336,
      "step": 15473
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6518871784210205,
      "learning_rate": 0.0005933471428571882,
      "loss": 2.3923,
      "step": 15474
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.8435178995132446,
      "learning_rate": 0.0005933462861466804,
      "loss": 2.2338,
      "step": 15475
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6879441738128662,
      "learning_rate": 0.0005933454293816339,
      "loss": 2.3019,
      "step": 15476
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.9572340250015259,
      "learning_rate": 0.0005933445725620489,
      "loss": 2.4347,
      "step": 15477
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.830507755279541,
      "learning_rate": 0.0005933437156879257,
      "loss": 2.526,
      "step": 15478
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.5964970588684082,
      "learning_rate": 0.0005933428587592642,
      "loss": 2.5177,
      "step": 15479
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6534627079963684,
      "learning_rate": 0.0005933420017760648,
      "loss": 2.4915,
      "step": 15480
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6746882200241089,
      "learning_rate": 0.0005933411447383276,
      "loss": 2.2229,
      "step": 15481
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.810184895992279,
      "learning_rate": 0.0005933402876460528,
      "loss": 2.2196,
      "step": 15482
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6470386385917664,
      "learning_rate": 0.0005933394304992404,
      "loss": 2.2255,
      "step": 15483
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.5946974754333496,
      "learning_rate": 0.0005933385732978907,
      "loss": 2.5001,
      "step": 15484
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.651599109172821,
      "learning_rate": 0.000593337716042004,
      "loss": 2.5538,
      "step": 15485
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6196733713150024,
      "learning_rate": 0.0005933368587315801,
      "loss": 2.3379,
      "step": 15486
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6089733242988586,
      "learning_rate": 0.0005933360013666194,
      "loss": 2.6256,
      "step": 15487
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6671513319015503,
      "learning_rate": 0.0005933351439471219,
      "loss": 2.6167,
      "step": 15488
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6057042479515076,
      "learning_rate": 0.0005933342864730879,
      "loss": 2.3525,
      "step": 15489
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6292568445205688,
      "learning_rate": 0.0005933334289445175,
      "loss": 2.4852,
      "step": 15490
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.647100031375885,
      "learning_rate": 0.000593332571361411,
      "loss": 2.6689,
      "step": 15491
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6170984506607056,
      "learning_rate": 0.0005933317137237683,
      "loss": 2.3778,
      "step": 15492
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6144692897796631,
      "learning_rate": 0.0005933308560315898,
      "loss": 2.4305,
      "step": 15493
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6827316880226135,
      "learning_rate": 0.0005933299982848755,
      "loss": 2.3503,
      "step": 15494
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.5978901386260986,
      "learning_rate": 0.0005933291404836256,
      "loss": 2.5675,
      "step": 15495
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6727644205093384,
      "learning_rate": 0.0005933282826278404,
      "loss": 2.3417,
      "step": 15496
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6158502697944641,
      "learning_rate": 0.0005933274247175198,
      "loss": 2.2641,
      "step": 15497
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6119911074638367,
      "learning_rate": 0.0005933265667526641,
      "loss": 2.2672,
      "step": 15498
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.643110454082489,
      "learning_rate": 0.0005933257087332735,
      "loss": 2.4874,
      "step": 15499
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.5760570764541626,
      "learning_rate": 0.0005933248506593482,
      "loss": 2.4207,
      "step": 15500
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6264762878417969,
      "learning_rate": 0.0005933239925308881,
      "loss": 2.2576,
      "step": 15501
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6207436919212341,
      "learning_rate": 0.0005933231343478937,
      "loss": 2.5486,
      "step": 15502
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6352595090866089,
      "learning_rate": 0.0005933222761103649,
      "loss": 2.473,
      "step": 15503
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6004443168640137,
      "learning_rate": 0.0005933214178183018,
      "loss": 2.5407,
      "step": 15504
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6080203056335449,
      "learning_rate": 0.000593320559471705,
      "loss": 2.4101,
      "step": 15505
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.5715046525001526,
      "learning_rate": 0.0005933197010705742,
      "loss": 2.1348,
      "step": 15506
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6155967712402344,
      "learning_rate": 0.0005933188426149098,
      "loss": 2.1163,
      "step": 15507
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6169881820678711,
      "learning_rate": 0.0005933179841047119,
      "loss": 2.5333,
      "step": 15508
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6001137495040894,
      "learning_rate": 0.0005933171255399805,
      "loss": 2.3601,
      "step": 15509
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.5947117805480957,
      "learning_rate": 0.0005933162669207161,
      "loss": 2.4339,
      "step": 15510
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.618012011051178,
      "learning_rate": 0.0005933154082469186,
      "loss": 2.5901,
      "step": 15511
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6164198517799377,
      "learning_rate": 0.0005933145495185882,
      "loss": 2.2848,
      "step": 15512
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6410367488861084,
      "learning_rate": 0.0005933136907357252,
      "loss": 2.5198,
      "step": 15513
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.5991383194923401,
      "learning_rate": 0.0005933128318983294,
      "loss": 2.4679,
      "step": 15514
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6152974367141724,
      "learning_rate": 0.0005933119730064014,
      "loss": 2.2812,
      "step": 15515
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.5978394746780396,
      "learning_rate": 0.0005933111140599411,
      "loss": 2.4614,
      "step": 15516
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6020076274871826,
      "learning_rate": 0.0005933102550589488,
      "loss": 2.274,
      "step": 15517
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.5989572405815125,
      "learning_rate": 0.0005933093960034245,
      "loss": 2.479,
      "step": 15518
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.5770909786224365,
      "learning_rate": 0.0005933085368933685,
      "loss": 2.1005,
      "step": 15519
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.598387598991394,
      "learning_rate": 0.0005933076777287809,
      "loss": 2.5529,
      "step": 15520
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.7014548778533936,
      "learning_rate": 0.0005933068185096619,
      "loss": 2.5632,
      "step": 15521
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6371676325798035,
      "learning_rate": 0.0005933059592360116,
      "loss": 2.4492,
      "step": 15522
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6423801183700562,
      "learning_rate": 0.0005933050999078302,
      "loss": 2.5178,
      "step": 15523
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.5661649107933044,
      "learning_rate": 0.0005933042405251177,
      "loss": 2.2103,
      "step": 15524
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6155081391334534,
      "learning_rate": 0.0005933033810878746,
      "loss": 2.4982,
      "step": 15525
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.670443058013916,
      "learning_rate": 0.0005933025215961007,
      "loss": 2.5308,
      "step": 15526
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.633937656879425,
      "learning_rate": 0.0005933016620497965,
      "loss": 2.4032,
      "step": 15527
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6077983975410461,
      "learning_rate": 0.0005933008024489619,
      "loss": 2.5366,
      "step": 15528
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.7151567935943604,
      "learning_rate": 0.0005932999427935971,
      "loss": 2.5644,
      "step": 15529
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.598869264125824,
      "learning_rate": 0.0005932990830837024,
      "loss": 2.3234,
      "step": 15530
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6071131229400635,
      "learning_rate": 0.0005932982233192778,
      "loss": 2.4281,
      "step": 15531
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6248470544815063,
      "learning_rate": 0.0005932973635003236,
      "loss": 2.5756,
      "step": 15532
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6130199432373047,
      "learning_rate": 0.0005932965036268398,
      "loss": 2.2642,
      "step": 15533
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6570276021957397,
      "learning_rate": 0.0005932956436988266,
      "loss": 2.3182,
      "step": 15534
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6382980942726135,
      "learning_rate": 0.0005932947837162842,
      "loss": 2.3443,
      "step": 15535
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.5922840237617493,
      "learning_rate": 0.0005932939236792129,
      "loss": 2.4363,
      "step": 15536
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.633852481842041,
      "learning_rate": 0.0005932930635876127,
      "loss": 2.4015,
      "step": 15537
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6256272792816162,
      "learning_rate": 0.0005932922034414837,
      "loss": 2.4951,
      "step": 15538
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6065280437469482,
      "learning_rate": 0.0005932913432408262,
      "loss": 2.4757,
      "step": 15539
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6079772710800171,
      "learning_rate": 0.0005932904829856402,
      "loss": 2.433,
      "step": 15540
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.7398645877838135,
      "learning_rate": 0.0005932896226759261,
      "loss": 2.633,
      "step": 15541
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.650313138961792,
      "learning_rate": 0.0005932887623116838,
      "loss": 2.5228,
      "step": 15542
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6337786912918091,
      "learning_rate": 0.0005932879018929137,
      "loss": 2.3183,
      "step": 15543
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6051360964775085,
      "learning_rate": 0.0005932870414196159,
      "loss": 2.3318,
      "step": 15544
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.696509063243866,
      "learning_rate": 0.0005932861808917903,
      "loss": 2.4862,
      "step": 15545
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6281442642211914,
      "learning_rate": 0.0005932853203094374,
      "loss": 2.2232,
      "step": 15546
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6099459528923035,
      "learning_rate": 0.0005932844596725571,
      "loss": 2.1154,
      "step": 15547
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.5999733805656433,
      "learning_rate": 0.0005932835989811498,
      "loss": 2.2337,
      "step": 15548
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6468223929405212,
      "learning_rate": 0.0005932827382352156,
      "loss": 2.3488,
      "step": 15549
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.7265027761459351,
      "learning_rate": 0.0005932818774347545,
      "loss": 2.4997,
      "step": 15550
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.7410313487052917,
      "learning_rate": 0.0005932810165797668,
      "loss": 2.259,
      "step": 15551
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6005243062973022,
      "learning_rate": 0.0005932801556702526,
      "loss": 2.425,
      "step": 15552
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.5916047692298889,
      "learning_rate": 0.000593279294706212,
      "loss": 2.5607,
      "step": 15553
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6083918213844299,
      "learning_rate": 0.0005932784336876453,
      "loss": 2.357,
      "step": 15554
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6897696256637573,
      "learning_rate": 0.0005932775726145527,
      "loss": 2.5816,
      "step": 15555
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6118239164352417,
      "learning_rate": 0.0005932767114869342,
      "loss": 2.2692,
      "step": 15556
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6012611985206604,
      "learning_rate": 0.00059327585030479,
      "loss": 2.4159,
      "step": 15557
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6359925866127014,
      "learning_rate": 0.0005932749890681202,
      "loss": 2.4774,
      "step": 15558
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6258564591407776,
      "learning_rate": 0.0005932741277769252,
      "loss": 2.5272,
      "step": 15559
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.7452746033668518,
      "learning_rate": 0.000593273266431205,
      "loss": 2.2988,
      "step": 15560
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6308916211128235,
      "learning_rate": 0.0005932724050309597,
      "loss": 2.2986,
      "step": 15561
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.679581880569458,
      "learning_rate": 0.0005932715435761895,
      "loss": 2.3833,
      "step": 15562
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.7070268392562866,
      "learning_rate": 0.0005932706820668946,
      "loss": 2.4485,
      "step": 15563
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.7945404052734375,
      "learning_rate": 0.0005932698205030752,
      "loss": 2.451,
      "step": 15564
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6472659111022949,
      "learning_rate": 0.0005932689588847313,
      "loss": 2.4084,
      "step": 15565
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6066572070121765,
      "learning_rate": 0.0005932680972118632,
      "loss": 2.5744,
      "step": 15566
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6571353077888489,
      "learning_rate": 0.0005932672354844711,
      "loss": 2.4475,
      "step": 15567
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6560191512107849,
      "learning_rate": 0.0005932663737025551,
      "loss": 2.154,
      "step": 15568
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6453914046287537,
      "learning_rate": 0.0005932655118661152,
      "loss": 2.515,
      "step": 15569
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.7273544669151306,
      "learning_rate": 0.0005932646499751519,
      "loss": 2.4086,
      "step": 15570
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.575896143913269,
      "learning_rate": 0.000593263788029665,
      "loss": 2.2441,
      "step": 15571
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.8194968104362488,
      "learning_rate": 0.0005932629260296549,
      "loss": 2.4156,
      "step": 15572
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6383013725280762,
      "learning_rate": 0.0005932620639751216,
      "loss": 2.2307,
      "step": 15573
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6230703592300415,
      "learning_rate": 0.0005932612018660654,
      "loss": 2.2924,
      "step": 15574
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.7423701286315918,
      "learning_rate": 0.0005932603397024864,
      "loss": 2.4012,
      "step": 15575
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6442760825157166,
      "learning_rate": 0.0005932594774843849,
      "loss": 2.3806,
      "step": 15576
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.594996452331543,
      "learning_rate": 0.0005932586152117607,
      "loss": 2.3484,
      "step": 15577
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6214941143989563,
      "learning_rate": 0.0005932577528846144,
      "loss": 2.4578,
      "step": 15578
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6931633353233337,
      "learning_rate": 0.0005932568905029458,
      "loss": 2.6156,
      "step": 15579
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6695025563240051,
      "learning_rate": 0.0005932560280667553,
      "loss": 2.3816,
      "step": 15580
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.5922461152076721,
      "learning_rate": 0.0005932551655760429,
      "loss": 2.1996,
      "step": 15581
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6624966859817505,
      "learning_rate": 0.0005932543030308089,
      "loss": 2.334,
      "step": 15582
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.0320007801055908,
      "learning_rate": 0.0005932534404310534,
      "loss": 2.3678,
      "step": 15583
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6136851906776428,
      "learning_rate": 0.0005932525777767764,
      "loss": 2.3511,
      "step": 15584
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.7316869497299194,
      "learning_rate": 0.0005932517150679783,
      "loss": 2.5522,
      "step": 15585
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.806077241897583,
      "learning_rate": 0.0005932508523046592,
      "loss": 2.6432,
      "step": 15586
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.5654841065406799,
      "learning_rate": 0.0005932499894868192,
      "loss": 2.2783,
      "step": 15587
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.5894576907157898,
      "learning_rate": 0.0005932491266144585,
      "loss": 2.4345,
      "step": 15588
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.0433870553970337,
      "learning_rate": 0.0005932482636875773,
      "loss": 2.3556,
      "step": 15589
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.604775607585907,
      "learning_rate": 0.0005932474007061757,
      "loss": 2.3358,
      "step": 15590
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6055439114570618,
      "learning_rate": 0.0005932465376702538,
      "loss": 2.6227,
      "step": 15591
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.8768641948699951,
      "learning_rate": 0.0005932456745798118,
      "loss": 2.2395,
      "step": 15592
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6794201731681824,
      "learning_rate": 0.00059324481143485,
      "loss": 2.5805,
      "step": 15593
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6025211811065674,
      "learning_rate": 0.0005932439482353684,
      "loss": 2.4506,
      "step": 15594
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.5841242671012878,
      "learning_rate": 0.0005932430849813672,
      "loss": 2.341,
      "step": 15595
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6584285497665405,
      "learning_rate": 0.0005932422216728466,
      "loss": 2.6059,
      "step": 15596
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.8338549733161926,
      "learning_rate": 0.0005932413583098068,
      "loss": 2.2586,
      "step": 15597
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6182674765586853,
      "learning_rate": 0.0005932404948922478,
      "loss": 2.4573,
      "step": 15598
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.5977731347084045,
      "learning_rate": 0.0005932396314201699,
      "loss": 2.4315,
      "step": 15599
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6282696723937988,
      "learning_rate": 0.0005932387678935732,
      "loss": 2.2202,
      "step": 15600
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6830780506134033,
      "learning_rate": 0.000593237904312458,
      "loss": 2.4784,
      "step": 15601
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6807683706283569,
      "learning_rate": 0.0005932370406768241,
      "loss": 2.4411,
      "step": 15602
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6041979789733887,
      "learning_rate": 0.0005932361769866721,
      "loss": 2.4283,
      "step": 15603
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.7104384303092957,
      "learning_rate": 0.0005932353132420019,
      "loss": 2.3829,
      "step": 15604
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6888723373413086,
      "learning_rate": 0.0005932344494428136,
      "loss": 2.319,
      "step": 15605
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6659220457077026,
      "learning_rate": 0.0005932335855891077,
      "loss": 2.526,
      "step": 15606
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6338881850242615,
      "learning_rate": 0.000593232721680884,
      "loss": 2.3601,
      "step": 15607
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6851962208747864,
      "learning_rate": 0.0005932318577181428,
      "loss": 2.214,
      "step": 15608
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.62008136510849,
      "learning_rate": 0.0005932309937008843,
      "loss": 2.3491,
      "step": 15609
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6229756474494934,
      "learning_rate": 0.0005932301296291086,
      "loss": 2.3209,
      "step": 15610
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.597485363483429,
      "learning_rate": 0.0005932292655028159,
      "loss": 2.0891,
      "step": 15611
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.5669679641723633,
      "learning_rate": 0.0005932284013220063,
      "loss": 2.4576,
      "step": 15612
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6551102995872498,
      "learning_rate": 0.0005932275370866801,
      "loss": 2.223,
      "step": 15613
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.5943433046340942,
      "learning_rate": 0.0005932266727968374,
      "loss": 2.5535,
      "step": 15614
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6355820298194885,
      "learning_rate": 0.0005932258084524782,
      "loss": 2.5736,
      "step": 15615
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.765107274055481,
      "learning_rate": 0.0005932249440536028,
      "loss": 2.5341,
      "step": 15616
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6025041937828064,
      "learning_rate": 0.0005932240796002115,
      "loss": 2.433,
      "step": 15617
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6390200853347778,
      "learning_rate": 0.000593223215092304,
      "loss": 2.4504,
      "step": 15618
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6131909489631653,
      "learning_rate": 0.0005932223505298811,
      "loss": 2.4999,
      "step": 15619
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6340221762657166,
      "learning_rate": 0.0005932214859129425,
      "loss": 2.4322,
      "step": 15620
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6619765758514404,
      "learning_rate": 0.0005932206212414884,
      "loss": 2.3978,
      "step": 15621
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6383512020111084,
      "learning_rate": 0.0005932197565155191,
      "loss": 2.2209,
      "step": 15622
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.8693178296089172,
      "learning_rate": 0.0005932188917350347,
      "loss": 2.304,
      "step": 15623
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.5974388718605042,
      "learning_rate": 0.0005932180269000354,
      "loss": 2.2268,
      "step": 15624
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.7259423732757568,
      "learning_rate": 0.0005932171620105214,
      "loss": 2.4404,
      "step": 15625
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6890689134597778,
      "learning_rate": 0.0005932162970664927,
      "loss": 2.0754,
      "step": 15626
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.675319254398346,
      "learning_rate": 0.0005932154320679497,
      "loss": 2.528,
      "step": 15627
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.596950352191925,
      "learning_rate": 0.0005932145670148922,
      "loss": 2.5344,
      "step": 15628
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.9764941334724426,
      "learning_rate": 0.0005932137019073208,
      "loss": 2.2891,
      "step": 15629
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.7430503368377686,
      "learning_rate": 0.0005932128367452352,
      "loss": 2.4134,
      "step": 15630
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.5541655421257019,
      "learning_rate": 0.000593211971528636,
      "loss": 2.3024,
      "step": 15631
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.7469785809516907,
      "learning_rate": 0.0005932111062575231,
      "loss": 2.336,
      "step": 15632
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6329395771026611,
      "learning_rate": 0.0005932102409318967,
      "loss": 2.4778,
      "step": 15633
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6097753643989563,
      "learning_rate": 0.0005932093755517569,
      "loss": 2.5011,
      "step": 15634
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.5823140144348145,
      "learning_rate": 0.0005932085101171041,
      "loss": 2.3672,
      "step": 15635
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.5963686108589172,
      "learning_rate": 0.0005932076446279383,
      "loss": 2.2892,
      "step": 15636
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.5882214307785034,
      "learning_rate": 0.0005932067790842595,
      "loss": 2.2553,
      "step": 15637
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.5995051860809326,
      "learning_rate": 0.0005932059134860682,
      "loss": 2.308,
      "step": 15638
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6300459504127502,
      "learning_rate": 0.0005932050478333643,
      "loss": 2.5353,
      "step": 15639
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6292588114738464,
      "learning_rate": 0.0005932041821261481,
      "loss": 2.8062,
      "step": 15640
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6328408122062683,
      "learning_rate": 0.0005932033163644196,
      "loss": 2.3557,
      "step": 15641
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.5969027280807495,
      "learning_rate": 0.0005932024505481791,
      "loss": 2.4059,
      "step": 15642
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.612369954586029,
      "learning_rate": 0.0005932015846774268,
      "loss": 2.4218,
      "step": 15643
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6474032402038574,
      "learning_rate": 0.0005932007187521627,
      "loss": 2.3628,
      "step": 15644
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6310977935791016,
      "learning_rate": 0.0005931998527723871,
      "loss": 2.3059,
      "step": 15645
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.5922847390174866,
      "learning_rate": 0.0005931989867381002,
      "loss": 2.3825,
      "step": 15646
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.5941964387893677,
      "learning_rate": 0.0005931981206493019,
      "loss": 2.4468,
      "step": 15647
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6950934529304504,
      "learning_rate": 0.0005931972545059927,
      "loss": 2.3983,
      "step": 15648
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.7107961177825928,
      "learning_rate": 0.0005931963883081726,
      "loss": 2.3552,
      "step": 15649
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6185213923454285,
      "learning_rate": 0.0005931955220558416,
      "loss": 2.4675,
      "step": 15650
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.7131078243255615,
      "learning_rate": 0.0005931946557490001,
      "loss": 2.345,
      "step": 15651
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6562556624412537,
      "learning_rate": 0.0005931937893876481,
      "loss": 2.4475,
      "step": 15652
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6334577798843384,
      "learning_rate": 0.000593192922971786,
      "loss": 2.3485,
      "step": 15653
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6828067302703857,
      "learning_rate": 0.0005931920565014138,
      "loss": 2.6929,
      "step": 15654
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6271493434906006,
      "learning_rate": 0.0005931911899765315,
      "loss": 2.595,
      "step": 15655
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.5935032963752747,
      "learning_rate": 0.0005931903233971395,
      "loss": 2.265,
      "step": 15656
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6271631717681885,
      "learning_rate": 0.0005931894567632378,
      "loss": 2.5522,
      "step": 15657
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.5775883197784424,
      "learning_rate": 0.0005931885900748268,
      "loss": 2.288,
      "step": 15658
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6523883938789368,
      "learning_rate": 0.0005931877233319064,
      "loss": 2.4172,
      "step": 15659
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6362996101379395,
      "learning_rate": 0.000593186856534477,
      "loss": 2.3286,
      "step": 15660
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6393117308616638,
      "learning_rate": 0.0005931859896825385,
      "loss": 2.3084,
      "step": 15661
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6707589626312256,
      "learning_rate": 0.0005931851227760912,
      "loss": 2.3234,
      "step": 15662
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6237701177597046,
      "learning_rate": 0.0005931842558151353,
      "loss": 2.3428,
      "step": 15663
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.646425187587738,
      "learning_rate": 0.0005931833887996709,
      "loss": 2.5978,
      "step": 15664
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6175057291984558,
      "learning_rate": 0.0005931825217296981,
      "loss": 2.5717,
      "step": 15665
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.5998168587684631,
      "learning_rate": 0.0005931816546052173,
      "loss": 2.428,
      "step": 15666
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6202929615974426,
      "learning_rate": 0.0005931807874262284,
      "loss": 2.424,
      "step": 15667
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6607178449630737,
      "learning_rate": 0.0005931799201927317,
      "loss": 2.6713,
      "step": 15668
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6711148023605347,
      "learning_rate": 0.0005931790529047271,
      "loss": 2.3185,
      "step": 15669
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6262480020523071,
      "learning_rate": 0.0005931781855622152,
      "loss": 2.5763,
      "step": 15670
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6398406028747559,
      "learning_rate": 0.000593177318165196,
      "loss": 2.3116,
      "step": 15671
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6137746572494507,
      "learning_rate": 0.0005931764507136695,
      "loss": 2.212,
      "step": 15672
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.7131582498550415,
      "learning_rate": 0.0005931755832076359,
      "loss": 2.2745,
      "step": 15673
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6300376653671265,
      "learning_rate": 0.0005931747156470954,
      "loss": 2.4206,
      "step": 15674
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6520740389823914,
      "learning_rate": 0.0005931738480320483,
      "loss": 2.2584,
      "step": 15675
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6614176630973816,
      "learning_rate": 0.0005931729803624946,
      "loss": 2.3295,
      "step": 15676
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6566151976585388,
      "learning_rate": 0.0005931721126384346,
      "loss": 2.5346,
      "step": 15677
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6078662276268005,
      "learning_rate": 0.0005931712448598683,
      "loss": 2.5886,
      "step": 15678
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6667183637619019,
      "learning_rate": 0.0005931703770267959,
      "loss": 2.2739,
      "step": 15679
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6222203969955444,
      "learning_rate": 0.0005931695091392177,
      "loss": 2.5718,
      "step": 15680
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6068462133407593,
      "learning_rate": 0.0005931686411971336,
      "loss": 2.4701,
      "step": 15681
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.571740984916687,
      "learning_rate": 0.000593167773200544,
      "loss": 2.4038,
      "step": 15682
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.5837828516960144,
      "learning_rate": 0.000593166905149449,
      "loss": 2.3858,
      "step": 15683
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6028950214385986,
      "learning_rate": 0.0005931660370438487,
      "loss": 2.4012,
      "step": 15684
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.5843843817710876,
      "learning_rate": 0.0005931651688837434,
      "loss": 2.3525,
      "step": 15685
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6348235011100769,
      "learning_rate": 0.000593164300669133,
      "loss": 2.2045,
      "step": 15686
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.7771673798561096,
      "learning_rate": 0.0005931634324000179,
      "loss": 2.3882,
      "step": 15687
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6022575497627258,
      "learning_rate": 0.0005931625640763982,
      "loss": 2.2128,
      "step": 15688
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.5916458368301392,
      "learning_rate": 0.000593161695698274,
      "loss": 2.6733,
      "step": 15689
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.696018397808075,
      "learning_rate": 0.0005931608272656454,
      "loss": 2.3222,
      "step": 15690
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.7298977375030518,
      "learning_rate": 0.0005931599587785129,
      "loss": 2.3379,
      "step": 15691
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6096284985542297,
      "learning_rate": 0.0005931590902368762,
      "loss": 2.4246,
      "step": 15692
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6460460424423218,
      "learning_rate": 0.0005931582216407359,
      "loss": 2.5744,
      "step": 15693
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6305779218673706,
      "learning_rate": 0.0005931573529900919,
      "loss": 2.5986,
      "step": 15694
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6129165291786194,
      "learning_rate": 0.0005931564842849445,
      "loss": 2.2301,
      "step": 15695
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.5898638367652893,
      "learning_rate": 0.0005931556155252935,
      "loss": 2.2054,
      "step": 15696
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.5854387283325195,
      "learning_rate": 0.0005931547467111396,
      "loss": 2.6221,
      "step": 15697
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6228315830230713,
      "learning_rate": 0.0005931538778424825,
      "loss": 2.4568,
      "step": 15698
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6274449229240417,
      "learning_rate": 0.0005931530089193227,
      "loss": 2.4694,
      "step": 15699
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.5988364815711975,
      "learning_rate": 0.0005931521399416602,
      "loss": 2.379,
      "step": 15700
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6774535775184631,
      "learning_rate": 0.0005931512709094951,
      "loss": 2.4836,
      "step": 15701
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6972647309303284,
      "learning_rate": 0.0005931504018228277,
      "loss": 2.3814,
      "step": 15702
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6571512222290039,
      "learning_rate": 0.0005931495326816582,
      "loss": 2.468,
      "step": 15703
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.7698896527290344,
      "learning_rate": 0.0005931486634859865,
      "loss": 2.328,
      "step": 15704
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6348475813865662,
      "learning_rate": 0.0005931477942358131,
      "loss": 2.3895,
      "step": 15705
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6675208210945129,
      "learning_rate": 0.0005931469249311378,
      "loss": 2.1144,
      "step": 15706
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.7852674722671509,
      "learning_rate": 0.0005931460555719611,
      "loss": 2.4165,
      "step": 15707
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.7391642928123474,
      "learning_rate": 0.000593145186158283,
      "loss": 2.6386,
      "step": 15708
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6427516341209412,
      "learning_rate": 0.0005931443166901036,
      "loss": 2.3791,
      "step": 15709
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6933728456497192,
      "learning_rate": 0.0005931434471674232,
      "loss": 2.4043,
      "step": 15710
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6061711311340332,
      "learning_rate": 0.0005931425775902419,
      "loss": 2.38,
      "step": 15711
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6034574508666992,
      "learning_rate": 0.0005931417079585598,
      "loss": 2.2963,
      "step": 15712
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6278432011604309,
      "learning_rate": 0.0005931408382723772,
      "loss": 2.4393,
      "step": 15713
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.7692143321037292,
      "learning_rate": 0.0005931399685316941,
      "loss": 2.2126,
      "step": 15714
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6561068296432495,
      "learning_rate": 0.0005931390987365108,
      "loss": 2.2411,
      "step": 15715
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6212752461433411,
      "learning_rate": 0.0005931382288868275,
      "loss": 2.1306,
      "step": 15716
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.5965747237205505,
      "learning_rate": 0.0005931373589826442,
      "loss": 2.3095,
      "step": 15717
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.5795819163322449,
      "learning_rate": 0.000593136489023961,
      "loss": 2.3003,
      "step": 15718
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6015210747718811,
      "learning_rate": 0.0005931356190107784,
      "loss": 2.526,
      "step": 15719
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6844760775566101,
      "learning_rate": 0.0005931347489430962,
      "loss": 2.5154,
      "step": 15720
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6398378014564514,
      "learning_rate": 0.0005931338788209148,
      "loss": 2.4695,
      "step": 15721
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6316945552825928,
      "learning_rate": 0.0005931330086442342,
      "loss": 2.5356,
      "step": 15722
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6171243786811829,
      "learning_rate": 0.0005931321384130547,
      "loss": 2.288,
      "step": 15723
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6430086493492126,
      "learning_rate": 0.0005931312681273764,
      "loss": 2.3904,
      "step": 15724
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.7052066922187805,
      "learning_rate": 0.0005931303977871995,
      "loss": 2.5397,
      "step": 15725
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.5860933661460876,
      "learning_rate": 0.000593129527392524,
      "loss": 2.2897,
      "step": 15726
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6054971218109131,
      "learning_rate": 0.0005931286569433504,
      "loss": 2.6689,
      "step": 15727
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6297438144683838,
      "learning_rate": 0.0005931277864396785,
      "loss": 2.4969,
      "step": 15728
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6578280329704285,
      "learning_rate": 0.0005931269158815086,
      "loss": 2.4689,
      "step": 15729
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6125984787940979,
      "learning_rate": 0.0005931260452688409,
      "loss": 2.5251,
      "step": 15730
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6442188024520874,
      "learning_rate": 0.0005931251746016756,
      "loss": 2.1806,
      "step": 15731
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6683287024497986,
      "learning_rate": 0.0005931243038800127,
      "loss": 2.6505,
      "step": 15732
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.5263848900794983,
      "learning_rate": 0.0005931234331038525,
      "loss": 2.1692,
      "step": 15733
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.7106137871742249,
      "learning_rate": 0.0005931225622731951,
      "loss": 2.5291,
      "step": 15734
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6485304832458496,
      "learning_rate": 0.0005931216913880407,
      "loss": 2.4001,
      "step": 15735
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6308802366256714,
      "learning_rate": 0.0005931208204483895,
      "loss": 2.3428,
      "step": 15736
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.546014666557312,
      "learning_rate": 0.0005931199494542415,
      "loss": 2.1476,
      "step": 15737
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.5763817429542542,
      "learning_rate": 0.000593119078405597,
      "loss": 2.4715,
      "step": 15738
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6094769239425659,
      "learning_rate": 0.0005931182073024561,
      "loss": 2.4137,
      "step": 15739
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6002017259597778,
      "learning_rate": 0.0005931173361448191,
      "loss": 2.5707,
      "step": 15740
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.5881676077842712,
      "learning_rate": 0.0005931164649326861,
      "loss": 2.3787,
      "step": 15741
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6780153512954712,
      "learning_rate": 0.000593115593666057,
      "loss": 2.3736,
      "step": 15742
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.5773376822471619,
      "learning_rate": 0.0005931147223449323,
      "loss": 2.3976,
      "step": 15743
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.673612654209137,
      "learning_rate": 0.0005931138509693121,
      "loss": 2.2966,
      "step": 15744
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.603610634803772,
      "learning_rate": 0.0005931129795391964,
      "loss": 2.4526,
      "step": 15745
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.595724880695343,
      "learning_rate": 0.0005931121080545855,
      "loss": 2.1615,
      "step": 15746
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6471928358078003,
      "learning_rate": 0.0005931112365154795,
      "loss": 2.4171,
      "step": 15747
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.5592705011367798,
      "learning_rate": 0.0005931103649218786,
      "loss": 2.3227,
      "step": 15748
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.5633092522621155,
      "learning_rate": 0.000593109493273783,
      "loss": 2.3022,
      "step": 15749
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.639743983745575,
      "learning_rate": 0.0005931086215711927,
      "loss": 2.6167,
      "step": 15750
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.5929054021835327,
      "learning_rate": 0.0005931077498141081,
      "loss": 2.301,
      "step": 15751
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.5716732144355774,
      "learning_rate": 0.0005931068780025291,
      "loss": 2.4269,
      "step": 15752
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6501918435096741,
      "learning_rate": 0.0005931060061364562,
      "loss": 2.1451,
      "step": 15753
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6912462711334229,
      "learning_rate": 0.0005931051342158893,
      "loss": 2.3837,
      "step": 15754
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.5715140700340271,
      "learning_rate": 0.0005931042622408285,
      "loss": 2.2916,
      "step": 15755
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6037752032279968,
      "learning_rate": 0.0005931033902112742,
      "loss": 2.4232,
      "step": 15756
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6401923894882202,
      "learning_rate": 0.0005931025181272264,
      "loss": 2.3253,
      "step": 15757
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6390208601951599,
      "learning_rate": 0.0005931016459886852,
      "loss": 2.2831,
      "step": 15758
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6316415667533875,
      "learning_rate": 0.0005931007737956511,
      "loss": 2.2733,
      "step": 15759
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6016120314598083,
      "learning_rate": 0.0005930999015481238,
      "loss": 2.2352,
      "step": 15760
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6141035556793213,
      "learning_rate": 0.0005930990292461038,
      "loss": 2.4579,
      "step": 15761
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6432951092720032,
      "learning_rate": 0.0005930981568895912,
      "loss": 2.4679,
      "step": 15762
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6382824182510376,
      "learning_rate": 0.0005930972844785861,
      "loss": 2.5084,
      "step": 15763
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6225870251655579,
      "learning_rate": 0.0005930964120130888,
      "loss": 2.0992,
      "step": 15764
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.8237408399581909,
      "learning_rate": 0.0005930955394930991,
      "loss": 2.239,
      "step": 15765
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6243847012519836,
      "learning_rate": 0.0005930946669186175,
      "loss": 2.3837,
      "step": 15766
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.5806120038032532,
      "learning_rate": 0.0005930937942896441,
      "loss": 2.2073,
      "step": 15767
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.5589138269424438,
      "learning_rate": 0.000593092921606179,
      "loss": 2.2126,
      "step": 15768
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.683448314666748,
      "learning_rate": 0.0005930920488682224,
      "loss": 2.3937,
      "step": 15769
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.5829939842224121,
      "learning_rate": 0.0005930911760757745,
      "loss": 2.3738,
      "step": 15770
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.725503921508789,
      "learning_rate": 0.0005930903032288354,
      "loss": 2.3368,
      "step": 15771
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6833738684654236,
      "learning_rate": 0.0005930894303274053,
      "loss": 2.4841,
      "step": 15772
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.5650634169578552,
      "learning_rate": 0.0005930885573714842,
      "loss": 2.308,
      "step": 15773
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6031234264373779,
      "learning_rate": 0.0005930876843610725,
      "loss": 2.2062,
      "step": 15774
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6234272122383118,
      "learning_rate": 0.0005930868112961704,
      "loss": 2.56,
      "step": 15775
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6735011339187622,
      "learning_rate": 0.0005930859381767778,
      "loss": 2.4178,
      "step": 15776
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.5830525159835815,
      "learning_rate": 0.000593085065002895,
      "loss": 2.4453,
      "step": 15777
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6709761619567871,
      "learning_rate": 0.0005930841917745222,
      "loss": 2.3533,
      "step": 15778
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.5994971394538879,
      "learning_rate": 0.0005930833184916594,
      "loss": 2.425,
      "step": 15779
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.5778807401657104,
      "learning_rate": 0.000593082445154307,
      "loss": 2.438,
      "step": 15780
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6000523567199707,
      "learning_rate": 0.0005930815717624649,
      "loss": 2.5098,
      "step": 15781
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.5993235111236572,
      "learning_rate": 0.0005930806983161336,
      "loss": 2.6906,
      "step": 15782
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6348836421966553,
      "learning_rate": 0.0005930798248153129,
      "loss": 2.2545,
      "step": 15783
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6118364334106445,
      "learning_rate": 0.0005930789512600032,
      "loss": 2.5951,
      "step": 15784
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6070964336395264,
      "learning_rate": 0.0005930780776502045,
      "loss": 2.3602,
      "step": 15785
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.5601100921630859,
      "learning_rate": 0.0005930772039859173,
      "loss": 2.4039,
      "step": 15786
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6055660247802734,
      "learning_rate": 0.0005930763302671412,
      "loss": 2.4658,
      "step": 15787
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.5932363867759705,
      "learning_rate": 0.0005930754564938769,
      "loss": 2.3593,
      "step": 15788
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.694950520992279,
      "learning_rate": 0.0005930745826661243,
      "loss": 2.5861,
      "step": 15789
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6215215921401978,
      "learning_rate": 0.0005930737087838835,
      "loss": 2.3911,
      "step": 15790
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6470693945884705,
      "learning_rate": 0.0005930728348471549,
      "loss": 2.4491,
      "step": 15791
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6307413578033447,
      "learning_rate": 0.0005930719608559384,
      "loss": 2.261,
      "step": 15792
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6910635232925415,
      "learning_rate": 0.0005930710868102344,
      "loss": 2.4235,
      "step": 15793
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.5727124810218811,
      "learning_rate": 0.0005930702127100428,
      "loss": 2.2412,
      "step": 15794
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.5809880495071411,
      "learning_rate": 0.0005930693385553641,
      "loss": 2.4201,
      "step": 15795
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.5679453015327454,
      "learning_rate": 0.0005930684643461982,
      "loss": 2.6438,
      "step": 15796
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6992397904396057,
      "learning_rate": 0.0005930675900825453,
      "loss": 2.4852,
      "step": 15797
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.8017711639404297,
      "learning_rate": 0.0005930667157644057,
      "loss": 2.4778,
      "step": 15798
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6125373244285583,
      "learning_rate": 0.0005930658413917794,
      "loss": 2.4892,
      "step": 15799
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.7149658799171448,
      "learning_rate": 0.0005930649669646667,
      "loss": 2.4656,
      "step": 15800
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6413015127182007,
      "learning_rate": 0.0005930640924830676,
      "loss": 2.5155,
      "step": 15801
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6230284571647644,
      "learning_rate": 0.0005930632179469823,
      "loss": 2.4818,
      "step": 15802
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6103947758674622,
      "learning_rate": 0.0005930623433564111,
      "loss": 2.3712,
      "step": 15803
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.7109236121177673,
      "learning_rate": 0.0005930614687113541,
      "loss": 2.4702,
      "step": 15804
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6130158305168152,
      "learning_rate": 0.0005930605940118114,
      "loss": 2.1784,
      "step": 15805
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6170234084129333,
      "learning_rate": 0.0005930597192577833,
      "loss": 2.4519,
      "step": 15806
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6342974901199341,
      "learning_rate": 0.0005930588444492698,
      "loss": 2.3939,
      "step": 15807
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.7426080703735352,
      "learning_rate": 0.000593057969586271,
      "loss": 2.5147,
      "step": 15808
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6277146339416504,
      "learning_rate": 0.0005930570946687874,
      "loss": 2.5049,
      "step": 15809
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6541557908058167,
      "learning_rate": 0.0005930562196968188,
      "loss": 2.4143,
      "step": 15810
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.8088351488113403,
      "learning_rate": 0.0005930553446703656,
      "loss": 2.1611,
      "step": 15811
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6285464763641357,
      "learning_rate": 0.0005930544695894279,
      "loss": 2.2345,
      "step": 15812
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.5598056316375732,
      "learning_rate": 0.0005930535944540057,
      "loss": 2.2318,
      "step": 15813
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.7400639057159424,
      "learning_rate": 0.0005930527192640995,
      "loss": 2.4163,
      "step": 15814
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.5738533139228821,
      "learning_rate": 0.0005930518440197091,
      "loss": 2.3983,
      "step": 15815
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6227317452430725,
      "learning_rate": 0.0005930509687208349,
      "loss": 2.527,
      "step": 15816
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.5643194913864136,
      "learning_rate": 0.0005930500933674771,
      "loss": 2.3109,
      "step": 15817
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6345847249031067,
      "learning_rate": 0.0005930492179596356,
      "loss": 2.426,
      "step": 15818
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.5695292949676514,
      "learning_rate": 0.0005930483424973108,
      "loss": 2.3525,
      "step": 15819
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6187460422515869,
      "learning_rate": 0.0005930474669805027,
      "loss": 2.2699,
      "step": 15820
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6427887082099915,
      "learning_rate": 0.0005930465914092116,
      "loss": 2.4005,
      "step": 15821
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6233567595481873,
      "learning_rate": 0.0005930457157834376,
      "loss": 2.5251,
      "step": 15822
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6347270011901855,
      "learning_rate": 0.0005930448401031809,
      "loss": 2.3288,
      "step": 15823
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.7062190771102905,
      "learning_rate": 0.0005930439643684416,
      "loss": 2.4192,
      "step": 15824
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6155635118484497,
      "learning_rate": 0.0005930430885792199,
      "loss": 2.5162,
      "step": 15825
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6895586848258972,
      "learning_rate": 0.000593042212735516,
      "loss": 2.3677,
      "step": 15826
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6065614223480225,
      "learning_rate": 0.0005930413368373299,
      "loss": 2.52,
      "step": 15827
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6523678302764893,
      "learning_rate": 0.0005930404608846619,
      "loss": 2.4547,
      "step": 15828
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.628017008304596,
      "learning_rate": 0.0005930395848775123,
      "loss": 2.5074,
      "step": 15829
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6802464723587036,
      "learning_rate": 0.0005930387088158809,
      "loss": 2.351,
      "step": 15830
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.5935710072517395,
      "learning_rate": 0.0005930378326997683,
      "loss": 2.4753,
      "step": 15831
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6557178497314453,
      "learning_rate": 0.0005930369565291742,
      "loss": 2.4291,
      "step": 15832
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.5683077573776245,
      "learning_rate": 0.0005930360803040992,
      "loss": 2.359,
      "step": 15833
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6783230304718018,
      "learning_rate": 0.0005930352040245432,
      "loss": 2.3205,
      "step": 15834
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.5708480477333069,
      "learning_rate": 0.0005930343276905064,
      "loss": 2.2591,
      "step": 15835
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6230959892272949,
      "learning_rate": 0.000593033451301989,
      "loss": 2.2404,
      "step": 15836
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6202940344810486,
      "learning_rate": 0.0005930325748589912,
      "loss": 2.572,
      "step": 15837
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.7077124118804932,
      "learning_rate": 0.000593031698361513,
      "loss": 2.3943,
      "step": 15838
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6534340977668762,
      "learning_rate": 0.0005930308218095548,
      "loss": 2.4385,
      "step": 15839
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6149000525474548,
      "learning_rate": 0.0005930299452031165,
      "loss": 2.3199,
      "step": 15840
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.579343318939209,
      "learning_rate": 0.0005930290685421985,
      "loss": 2.4436,
      "step": 15841
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6262816190719604,
      "learning_rate": 0.0005930281918268009,
      "loss": 2.6051,
      "step": 15842
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6210670471191406,
      "learning_rate": 0.0005930273150569237,
      "loss": 2.6085,
      "step": 15843
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.5644201636314392,
      "learning_rate": 0.0005930264382325673,
      "loss": 2.1614,
      "step": 15844
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6357430219650269,
      "learning_rate": 0.0005930255613537317,
      "loss": 2.5607,
      "step": 15845
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6076481342315674,
      "learning_rate": 0.0005930246844204172,
      "loss": 2.467,
      "step": 15846
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6365967392921448,
      "learning_rate": 0.0005930238074326237,
      "loss": 2.3439,
      "step": 15847
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6511483788490295,
      "learning_rate": 0.0005930229303903518,
      "loss": 2.359,
      "step": 15848
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6254995465278625,
      "learning_rate": 0.0005930220532936012,
      "loss": 2.5303,
      "step": 15849
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6723172664642334,
      "learning_rate": 0.0005930211761423724,
      "loss": 2.5025,
      "step": 15850
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6320679187774658,
      "learning_rate": 0.0005930202989366652,
      "loss": 2.6386,
      "step": 15851
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.5754071474075317,
      "learning_rate": 0.0005930194216764802,
      "loss": 2.1694,
      "step": 15852
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6082718372344971,
      "learning_rate": 0.0005930185443618173,
      "loss": 2.3385,
      "step": 15853
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.5901169180870056,
      "learning_rate": 0.0005930176669926767,
      "loss": 2.3672,
      "step": 15854
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.5940488576889038,
      "learning_rate": 0.0005930167895690587,
      "loss": 2.3486,
      "step": 15855
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6595292687416077,
      "learning_rate": 0.0005930159120909633,
      "loss": 2.267,
      "step": 15856
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6022744178771973,
      "learning_rate": 0.0005930150345583907,
      "loss": 2.2541,
      "step": 15857
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.5873709917068481,
      "learning_rate": 0.000593014156971341,
      "loss": 2.3469,
      "step": 15858
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.7674506902694702,
      "learning_rate": 0.0005930132793298145,
      "loss": 2.4004,
      "step": 15859
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.5665304660797119,
      "learning_rate": 0.0005930124016338112,
      "loss": 2.2957,
      "step": 15860
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6547380685806274,
      "learning_rate": 0.0005930115238833315,
      "loss": 2.5601,
      "step": 15861
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.5854324698448181,
      "learning_rate": 0.0005930106460783754,
      "loss": 2.1398,
      "step": 15862
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.5862603783607483,
      "learning_rate": 0.0005930097682189431,
      "loss": 2.3471,
      "step": 15863
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.605767548084259,
      "learning_rate": 0.0005930088903050348,
      "loss": 2.2849,
      "step": 15864
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.666812539100647,
      "learning_rate": 0.0005930080123366505,
      "loss": 2.4311,
      "step": 15865
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.5296332836151123,
      "learning_rate": 0.0005930071343137904,
      "loss": 2.2097,
      "step": 15866
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6366888880729675,
      "learning_rate": 0.000593006256236455,
      "loss": 2.4757,
      "step": 15867
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6613064408302307,
      "learning_rate": 0.000593005378104644,
      "loss": 2.4583,
      "step": 15868
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6451460719108582,
      "learning_rate": 0.0005930044999183579,
      "loss": 2.2127,
      "step": 15869
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6639456152915955,
      "learning_rate": 0.0005930036216775966,
      "loss": 2.5047,
      "step": 15870
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6013420820236206,
      "learning_rate": 0.0005930027433823606,
      "loss": 2.3875,
      "step": 15871
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6688448786735535,
      "learning_rate": 0.0005930018650326496,
      "loss": 2.5349,
      "step": 15872
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6348939538002014,
      "learning_rate": 0.0005930009866284641,
      "loss": 2.5439,
      "step": 15873
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6284000873565674,
      "learning_rate": 0.0005930001081698043,
      "loss": 2.1973,
      "step": 15874
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6155555844306946,
      "learning_rate": 0.0005929992296566702,
      "loss": 2.2763,
      "step": 15875
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6439703106880188,
      "learning_rate": 0.000592998351089062,
      "loss": 2.256,
      "step": 15876
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6342697143554688,
      "learning_rate": 0.0005929974724669799,
      "loss": 2.5741,
      "step": 15877
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.608021080493927,
      "learning_rate": 0.0005929965937904239,
      "loss": 2.3527,
      "step": 15878
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.5808513760566711,
      "learning_rate": 0.0005929957150593945,
      "loss": 2.416,
      "step": 15879
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6738088130950928,
      "learning_rate": 0.0005929948362738915,
      "loss": 2.4351,
      "step": 15880
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6002655029296875,
      "learning_rate": 0.0005929939574339153,
      "loss": 2.4981,
      "step": 15881
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.7599519491195679,
      "learning_rate": 0.000592993078539466,
      "loss": 2.4667,
      "step": 15882
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6525521278381348,
      "learning_rate": 0.0005929921995905438,
      "loss": 2.2456,
      "step": 15883
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6362680196762085,
      "learning_rate": 0.0005929913205871488,
      "loss": 2.3268,
      "step": 15884
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6365355253219604,
      "learning_rate": 0.0005929904415292811,
      "loss": 2.1873,
      "step": 15885
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.7355194091796875,
      "learning_rate": 0.0005929895624169411,
      "loss": 2.5325,
      "step": 15886
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.7088150978088379,
      "learning_rate": 0.0005929886832501287,
      "loss": 2.2467,
      "step": 15887
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6481630802154541,
      "learning_rate": 0.0005929878040288442,
      "loss": 2.4527,
      "step": 15888
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.7686746716499329,
      "learning_rate": 0.0005929869247530877,
      "loss": 2.3299,
      "step": 15889
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6185203790664673,
      "learning_rate": 0.0005929860454228595,
      "loss": 2.2435,
      "step": 15890
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.5957791209220886,
      "learning_rate": 0.0005929851660381595,
      "loss": 2.1674,
      "step": 15891
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.5663236975669861,
      "learning_rate": 0.0005929842865989881,
      "loss": 2.4223,
      "step": 15892
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.528843104839325,
      "learning_rate": 0.0005929834071053454,
      "loss": 2.5073,
      "step": 15893
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.630693793296814,
      "learning_rate": 0.0005929825275572316,
      "loss": 2.4724,
      "step": 15894
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6474079489707947,
      "learning_rate": 0.0005929816479546468,
      "loss": 2.263,
      "step": 15895
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.5700711011886597,
      "learning_rate": 0.000592980768297591,
      "loss": 2.2736,
      "step": 15896
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6264325976371765,
      "learning_rate": 0.0005929798885860647,
      "loss": 2.3149,
      "step": 15897
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6325185894966125,
      "learning_rate": 0.000592979008820068,
      "loss": 2.3496,
      "step": 15898
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.5957326889038086,
      "learning_rate": 0.0005929781289996009,
      "loss": 2.2365,
      "step": 15899
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6071534752845764,
      "learning_rate": 0.0005929772491246636,
      "loss": 2.3713,
      "step": 15900
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.679707407951355,
      "learning_rate": 0.0005929763691952564,
      "loss": 2.3944,
      "step": 15901
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.5833297967910767,
      "learning_rate": 0.0005929754892113792,
      "loss": 2.3704,
      "step": 15902
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6243922710418701,
      "learning_rate": 0.0005929746091730325,
      "loss": 2.4864,
      "step": 15903
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.7251608371734619,
      "learning_rate": 0.0005929737290802161,
      "loss": 2.6021,
      "step": 15904
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6086403131484985,
      "learning_rate": 0.0005929728489329305,
      "loss": 2.313,
      "step": 15905
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6135138869285583,
      "learning_rate": 0.0005929719687311757,
      "loss": 2.263,
      "step": 15906
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6377099752426147,
      "learning_rate": 0.0005929710884749517,
      "loss": 2.4302,
      "step": 15907
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.7304145097732544,
      "learning_rate": 0.0005929702081642591,
      "loss": 2.1542,
      "step": 15908
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6588062644004822,
      "learning_rate": 0.0005929693277990977,
      "loss": 2.5405,
      "step": 15909
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6730713248252869,
      "learning_rate": 0.0005929684473794678,
      "loss": 2.3222,
      "step": 15910
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6241852641105652,
      "learning_rate": 0.0005929675669053695,
      "loss": 2.5857,
      "step": 15911
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6701911687850952,
      "learning_rate": 0.0005929666863768029,
      "loss": 2.2328,
      "step": 15912
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.7216091752052307,
      "learning_rate": 0.0005929658057937685,
      "loss": 2.1869,
      "step": 15913
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6174240708351135,
      "learning_rate": 0.0005929649251562661,
      "loss": 2.3562,
      "step": 15914
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.5869938135147095,
      "learning_rate": 0.0005929640444642959,
      "loss": 2.2032,
      "step": 15915
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6411793828010559,
      "learning_rate": 0.0005929631637178582,
      "loss": 2.4379,
      "step": 15916
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6975463628768921,
      "learning_rate": 0.0005929622829169533,
      "loss": 2.2616,
      "step": 15917
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6434167623519897,
      "learning_rate": 0.000592961402061581,
      "loss": 2.523,
      "step": 15918
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6887151002883911,
      "learning_rate": 0.0005929605211517417,
      "loss": 2.3097,
      "step": 15919
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6827523112297058,
      "learning_rate": 0.0005929596401874355,
      "loss": 2.8555,
      "step": 15920
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6460933685302734,
      "learning_rate": 0.0005929587591686625,
      "loss": 2.2144,
      "step": 15921
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.5829272866249084,
      "learning_rate": 0.0005929578780954231,
      "loss": 2.557,
      "step": 15922
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6345947980880737,
      "learning_rate": 0.0005929569969677171,
      "loss": 2.2618,
      "step": 15923
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.734929084777832,
      "learning_rate": 0.0005929561157855451,
      "loss": 2.3637,
      "step": 15924
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6289005875587463,
      "learning_rate": 0.0005929552345489068,
      "loss": 2.2233,
      "step": 15925
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6094548106193542,
      "learning_rate": 0.0005929543532578027,
      "loss": 2.2057,
      "step": 15926
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.7117123603820801,
      "learning_rate": 0.0005929534719122328,
      "loss": 2.411,
      "step": 15927
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6518746614456177,
      "learning_rate": 0.0005929525905121974,
      "loss": 2.7333,
      "step": 15928
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.5849213600158691,
      "learning_rate": 0.0005929517090576965,
      "loss": 2.2175,
      "step": 15929
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.7332351207733154,
      "learning_rate": 0.0005929508275487305,
      "loss": 2.3166,
      "step": 15930
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.5936474204063416,
      "learning_rate": 0.0005929499459852993,
      "loss": 2.394,
      "step": 15931
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.657282292842865,
      "learning_rate": 0.0005929490643674032,
      "loss": 2.2239,
      "step": 15932
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.673461377620697,
      "learning_rate": 0.0005929481826950423,
      "loss": 2.5013,
      "step": 15933
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6049990653991699,
      "learning_rate": 0.0005929473009682167,
      "loss": 2.3237,
      "step": 15934
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6098470687866211,
      "learning_rate": 0.0005929464191869268,
      "loss": 2.5187,
      "step": 15935
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.5732332468032837,
      "learning_rate": 0.0005929455373511727,
      "loss": 2.1674,
      "step": 15936
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.5837128162384033,
      "learning_rate": 0.0005929446554609544,
      "loss": 2.1778,
      "step": 15937
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6486043334007263,
      "learning_rate": 0.0005929437735162722,
      "loss": 2.2584,
      "step": 15938
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6169524192810059,
      "learning_rate": 0.0005929428915171262,
      "loss": 2.4232,
      "step": 15939
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.7661792039871216,
      "learning_rate": 0.0005929420094635166,
      "loss": 2.3369,
      "step": 15940
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6630600094795227,
      "learning_rate": 0.0005929411273554435,
      "loss": 2.1892,
      "step": 15941
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.5689401626586914,
      "learning_rate": 0.0005929402451929071,
      "loss": 2.3664,
      "step": 15942
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6193954944610596,
      "learning_rate": 0.0005929393629759077,
      "loss": 2.3867,
      "step": 15943
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.7686270475387573,
      "learning_rate": 0.0005929384807044452,
      "loss": 2.4526,
      "step": 15944
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6021899580955505,
      "learning_rate": 0.00059293759837852,
      "loss": 2.2564,
      "step": 15945
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.8038155436515808,
      "learning_rate": 0.0005929367159981322,
      "loss": 2.2939,
      "step": 15946
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.749556303024292,
      "learning_rate": 0.0005929358335632819,
      "loss": 2.4545,
      "step": 15947
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6330647468566895,
      "learning_rate": 0.0005929349510739693,
      "loss": 2.6106,
      "step": 15948
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.8767969608306885,
      "learning_rate": 0.0005929340685301945,
      "loss": 2.2894,
      "step": 15949
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.7776541709899902,
      "learning_rate": 0.0005929331859319578,
      "loss": 2.5519,
      "step": 15950
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6682950258255005,
      "learning_rate": 0.0005929323032792592,
      "loss": 2.4478,
      "step": 15951
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6503379940986633,
      "learning_rate": 0.0005929314205720991,
      "loss": 2.4162,
      "step": 15952
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.7882962226867676,
      "learning_rate": 0.0005929305378104774,
      "loss": 2.3991,
      "step": 15953
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.761021077632904,
      "learning_rate": 0.0005929296549943944,
      "loss": 2.4697,
      "step": 15954
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.679538905620575,
      "learning_rate": 0.0005929287721238503,
      "loss": 2.3868,
      "step": 15955
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.9673811793327332,
      "learning_rate": 0.0005929278891988452,
      "loss": 2.4469,
      "step": 15956
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6044257879257202,
      "learning_rate": 0.0005929270062193792,
      "loss": 2.337,
      "step": 15957
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6676796078681946,
      "learning_rate": 0.0005929261231854526,
      "loss": 2.4138,
      "step": 15958
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.682869017124176,
      "learning_rate": 0.0005929252400970655,
      "loss": 2.3313,
      "step": 15959
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.662988543510437,
      "learning_rate": 0.0005929243569542181,
      "loss": 2.5343,
      "step": 15960
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.7201736569404602,
      "learning_rate": 0.0005929234737569105,
      "loss": 2.2697,
      "step": 15961
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6878864765167236,
      "learning_rate": 0.0005929225905051429,
      "loss": 2.2647,
      "step": 15962
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.648881733417511,
      "learning_rate": 0.0005929217071989155,
      "loss": 2.3777,
      "step": 15963
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6288160681724548,
      "learning_rate": 0.0005929208238382284,
      "loss": 2.2626,
      "step": 15964
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6257340908050537,
      "learning_rate": 0.0005929199404230818,
      "loss": 2.2413,
      "step": 15965
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6302317976951599,
      "learning_rate": 0.0005929190569534757,
      "loss": 2.6265,
      "step": 15966
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.5956476926803589,
      "learning_rate": 0.0005929181734294106,
      "loss": 2.5,
      "step": 15967
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.693071186542511,
      "learning_rate": 0.0005929172898508864,
      "loss": 2.4489,
      "step": 15968
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6179369688034058,
      "learning_rate": 0.0005929164062179034,
      "loss": 2.3988,
      "step": 15969
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6159621477127075,
      "learning_rate": 0.0005929155225304618,
      "loss": 2.309,
      "step": 15970
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6257442831993103,
      "learning_rate": 0.0005929146387885615,
      "loss": 2.5516,
      "step": 15971
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6286070942878723,
      "learning_rate": 0.0005929137549922029,
      "loss": 2.2903,
      "step": 15972
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6602117419242859,
      "learning_rate": 0.0005929128711413862,
      "loss": 2.3612,
      "step": 15973
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6351726651191711,
      "learning_rate": 0.0005929119872361112,
      "loss": 2.2072,
      "step": 15974
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6828826665878296,
      "learning_rate": 0.0005929111032763786,
      "loss": 2.1807,
      "step": 15975
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.8231227397918701,
      "learning_rate": 0.0005929102192621881,
      "loss": 2.5361,
      "step": 15976
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6567966341972351,
      "learning_rate": 0.0005929093351935402,
      "loss": 2.6158,
      "step": 15977
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.7151581645011902,
      "learning_rate": 0.000592908451070435,
      "loss": 2.4125,
      "step": 15978
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6824446320533752,
      "learning_rate": 0.0005929075668928724,
      "loss": 2.3355,
      "step": 15979
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.7443550229072571,
      "learning_rate": 0.0005929066826608529,
      "loss": 2.2338,
      "step": 15980
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6793239712715149,
      "learning_rate": 0.0005929057983743764,
      "loss": 2.5059,
      "step": 15981
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6900585293769836,
      "learning_rate": 0.0005929049140334432,
      "loss": 2.5043,
      "step": 15982
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.8167033195495605,
      "learning_rate": 0.0005929040296380535,
      "loss": 2.2426,
      "step": 15983
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.5645244717597961,
      "learning_rate": 0.0005929031451882074,
      "loss": 2.4739,
      "step": 15984
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.818155825138092,
      "learning_rate": 0.0005929022606839051,
      "loss": 2.4224,
      "step": 15985
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.8047183752059937,
      "learning_rate": 0.0005929013761251467,
      "loss": 2.4308,
      "step": 15986
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6434840559959412,
      "learning_rate": 0.0005929004915119324,
      "loss": 2.4682,
      "step": 15987
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.7612195014953613,
      "learning_rate": 0.0005928996068442624,
      "loss": 2.4,
      "step": 15988
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.8849800825119019,
      "learning_rate": 0.0005928987221221368,
      "loss": 2.3417,
      "step": 15989
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6306418180465698,
      "learning_rate": 0.0005928978373455558,
      "loss": 2.4989,
      "step": 15990
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6212899684906006,
      "learning_rate": 0.0005928969525145196,
      "loss": 2.5478,
      "step": 15991
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.7488164901733398,
      "learning_rate": 0.0005928960676290283,
      "loss": 2.382,
      "step": 15992
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.7545222640037537,
      "learning_rate": 0.0005928951826890821,
      "loss": 2.5638,
      "step": 15993
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.5862548351287842,
      "learning_rate": 0.0005928942976946812,
      "loss": 2.328,
      "step": 15994
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.5850315093994141,
      "learning_rate": 0.0005928934126458256,
      "loss": 2.1861,
      "step": 15995
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6327567100524902,
      "learning_rate": 0.0005928925275425158,
      "loss": 2.1614,
      "step": 15996
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6777877807617188,
      "learning_rate": 0.0005928916423847515,
      "loss": 2.4447,
      "step": 15997
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.5884994864463806,
      "learning_rate": 0.0005928907571725333,
      "loss": 2.2841,
      "step": 15998
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6952178478240967,
      "learning_rate": 0.0005928898719058611,
      "loss": 2.4441,
      "step": 15999
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.7615206837654114,
      "learning_rate": 0.0005928889865847352,
      "loss": 2.3979,
      "step": 16000
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6763661503791809,
      "learning_rate": 0.0005928881012091556,
      "loss": 2.5372,
      "step": 16001
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.5782885551452637,
      "learning_rate": 0.0005928872157791227,
      "loss": 2.4109,
      "step": 16002
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6302943825721741,
      "learning_rate": 0.0005928863302946364,
      "loss": 2.2819,
      "step": 16003
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6181270480155945,
      "learning_rate": 0.0005928854447556971,
      "loss": 2.4394,
      "step": 16004
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6361867189407349,
      "learning_rate": 0.0005928845591623049,
      "loss": 2.5188,
      "step": 16005
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6575269103050232,
      "learning_rate": 0.0005928836735144599,
      "loss": 2.2785,
      "step": 16006
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6543749570846558,
      "learning_rate": 0.0005928827878121623,
      "loss": 2.4267,
      "step": 16007
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6007812023162842,
      "learning_rate": 0.0005928819020554122,
      "loss": 2.1914,
      "step": 16008
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6619647145271301,
      "learning_rate": 0.0005928810162442098,
      "loss": 2.4281,
      "step": 16009
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6461902260780334,
      "learning_rate": 0.0005928801303785553,
      "loss": 2.6063,
      "step": 16010
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.5759245157241821,
      "learning_rate": 0.000592879244458449,
      "loss": 2.3443,
      "step": 16011
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6895238757133484,
      "learning_rate": 0.0005928783584838907,
      "loss": 2.127,
      "step": 16012
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6317258477210999,
      "learning_rate": 0.000592877472454881,
      "loss": 2.1967,
      "step": 16013
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6230169534683228,
      "learning_rate": 0.0005928765863714198,
      "loss": 2.6558,
      "step": 16014
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6054377555847168,
      "learning_rate": 0.0005928757002335072,
      "loss": 2.3484,
      "step": 16015
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.7021152973175049,
      "learning_rate": 0.0005928748140411436,
      "loss": 2.3414,
      "step": 16016
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6863214373588562,
      "learning_rate": 0.0005928739277943289,
      "loss": 2.4531,
      "step": 16017
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6103709936141968,
      "learning_rate": 0.0005928730414930636,
      "loss": 2.3797,
      "step": 16018
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6789670586585999,
      "learning_rate": 0.0005928721551373476,
      "loss": 2.4168,
      "step": 16019
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6361008286476135,
      "learning_rate": 0.0005928712687271812,
      "loss": 2.3226,
      "step": 16020
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6806416511535645,
      "learning_rate": 0.0005928703822625644,
      "loss": 2.1591,
      "step": 16021
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.5327994227409363,
      "learning_rate": 0.0005928694957434975,
      "loss": 2.3803,
      "step": 16022
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.5710878968238831,
      "learning_rate": 0.0005928686091699806,
      "loss": 2.7376,
      "step": 16023
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6329009532928467,
      "learning_rate": 0.0005928677225420139,
      "loss": 2.3759,
      "step": 16024
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6731716394424438,
      "learning_rate": 0.0005928668358595975,
      "loss": 2.7187,
      "step": 16025
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6295234560966492,
      "learning_rate": 0.0005928659491227317,
      "loss": 2.3739,
      "step": 16026
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.694810152053833,
      "learning_rate": 0.0005928650623314166,
      "loss": 2.6688,
      "step": 16027
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6833564639091492,
      "learning_rate": 0.0005928641754856524,
      "loss": 2.3291,
      "step": 16028
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6328989863395691,
      "learning_rate": 0.0005928632885854392,
      "loss": 2.3334,
      "step": 16029
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6253068447113037,
      "learning_rate": 0.0005928624016307771,
      "loss": 2.3092,
      "step": 16030
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.657458484172821,
      "learning_rate": 0.0005928615146216664,
      "loss": 2.6171,
      "step": 16031
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.8240768909454346,
      "learning_rate": 0.0005928606275581073,
      "loss": 2.3694,
      "step": 16032
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.5900765061378479,
      "learning_rate": 0.0005928597404400998,
      "loss": 2.307,
      "step": 16033
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.7131653428077698,
      "learning_rate": 0.0005928588532676441,
      "loss": 2.3518,
      "step": 16034
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.7582563161849976,
      "learning_rate": 0.0005928579660407405,
      "loss": 2.3649,
      "step": 16035
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6579803824424744,
      "learning_rate": 0.0005928570787593891,
      "loss": 2.5161,
      "step": 16036
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6253241896629333,
      "learning_rate": 0.00059285619142359,
      "loss": 2.3438,
      "step": 16037
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6692624688148499,
      "learning_rate": 0.0005928553040333433,
      "loss": 2.4829,
      "step": 16038
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.7047153115272522,
      "learning_rate": 0.0005928544165886494,
      "loss": 2.4209,
      "step": 16039
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6646760106086731,
      "learning_rate": 0.0005928535290895083,
      "loss": 2.5127,
      "step": 16040
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.725220799446106,
      "learning_rate": 0.0005928526415359202,
      "loss": 2.6699,
      "step": 16041
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6378692984580994,
      "learning_rate": 0.0005928517539278852,
      "loss": 2.1927,
      "step": 16042
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6212318539619446,
      "learning_rate": 0.0005928508662654037,
      "loss": 2.2376,
      "step": 16043
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6644997000694275,
      "learning_rate": 0.0005928499785484755,
      "loss": 2.2274,
      "step": 16044
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6343592405319214,
      "learning_rate": 0.0005928490907771011,
      "loss": 2.352,
      "step": 16045
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6412286162376404,
      "learning_rate": 0.0005928482029512804,
      "loss": 2.4152,
      "step": 16046
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6948909759521484,
      "learning_rate": 0.0005928473150710138,
      "loss": 2.325,
      "step": 16047
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6590009927749634,
      "learning_rate": 0.0005928464271363013,
      "loss": 2.503,
      "step": 16048
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.7152059674263,
      "learning_rate": 0.0005928455391471432,
      "loss": 2.4686,
      "step": 16049
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.8162912726402283,
      "learning_rate": 0.0005928446511035396,
      "loss": 2.548,
      "step": 16050
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.7741297483444214,
      "learning_rate": 0.0005928437630054905,
      "loss": 2.3965,
      "step": 16051
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.5870251655578613,
      "learning_rate": 0.0005928428748529964,
      "loss": 2.4127,
      "step": 16052
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6986740231513977,
      "learning_rate": 0.0005928419866460572,
      "loss": 2.346,
      "step": 16053
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.7383201718330383,
      "learning_rate": 0.000592841098384673,
      "loss": 2.4782,
      "step": 16054
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.847273588180542,
      "learning_rate": 0.0005928402100688442,
      "loss": 2.2549,
      "step": 16055
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6662265658378601,
      "learning_rate": 0.0005928393216985711,
      "loss": 2.1694,
      "step": 16056
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.7616650462150574,
      "learning_rate": 0.0005928384332738534,
      "loss": 2.2913,
      "step": 16057
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6634114384651184,
      "learning_rate": 0.0005928375447946916,
      "loss": 2.2617,
      "step": 16058
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6413760185241699,
      "learning_rate": 0.0005928366562610856,
      "loss": 2.572,
      "step": 16059
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.8519395589828491,
      "learning_rate": 0.0005928357676730359,
      "loss": 2.377,
      "step": 16060
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.7455410361289978,
      "learning_rate": 0.0005928348790305426,
      "loss": 2.4522,
      "step": 16061
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.5929798483848572,
      "learning_rate": 0.0005928339903336056,
      "loss": 2.7337,
      "step": 16062
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6012343168258667,
      "learning_rate": 0.0005928331015822253,
      "loss": 2.3372,
      "step": 16063
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.681783139705658,
      "learning_rate": 0.0005928322127764018,
      "loss": 2.3092,
      "step": 16064
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6877365112304688,
      "learning_rate": 0.0005928313239161353,
      "loss": 2.2153,
      "step": 16065
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.616907000541687,
      "learning_rate": 0.0005928304350014259,
      "loss": 2.0841,
      "step": 16066
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6141152381896973,
      "learning_rate": 0.0005928295460322737,
      "loss": 2.2791,
      "step": 16067
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6251314878463745,
      "learning_rate": 0.000592828657008679,
      "loss": 2.2744,
      "step": 16068
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6246762871742249,
      "learning_rate": 0.000592827767930642,
      "loss": 2.5469,
      "step": 16069
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.644037127494812,
      "learning_rate": 0.0005928268787981628,
      "loss": 2.5682,
      "step": 16070
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.5937702655792236,
      "learning_rate": 0.0005928259896112414,
      "loss": 2.368,
      "step": 16071
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6531049609184265,
      "learning_rate": 0.0005928251003698783,
      "loss": 2.3881,
      "step": 16072
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6321228742599487,
      "learning_rate": 0.0005928242110740734,
      "loss": 2.5306,
      "step": 16073
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.5838385224342346,
      "learning_rate": 0.0005928233217238269,
      "loss": 2.508,
      "step": 16074
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6529474854469299,
      "learning_rate": 0.0005928224323191392,
      "loss": 2.5925,
      "step": 16075
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6558376550674438,
      "learning_rate": 0.00059282154286001,
      "loss": 2.3685,
      "step": 16076
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.5376758575439453,
      "learning_rate": 0.00059282065334644,
      "loss": 2.2035,
      "step": 16077
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6005656719207764,
      "learning_rate": 0.0005928197637784289,
      "loss": 2.3181,
      "step": 16078
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.5725539922714233,
      "learning_rate": 0.0005928188741559773,
      "loss": 2.2772,
      "step": 16079
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.5994009971618652,
      "learning_rate": 0.000592817984479085,
      "loss": 2.5163,
      "step": 16080
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6375702023506165,
      "learning_rate": 0.0005928170947477523,
      "loss": 2.6224,
      "step": 16081
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6136853098869324,
      "learning_rate": 0.0005928162049619795,
      "loss": 2.5562,
      "step": 16082
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6175168752670288,
      "learning_rate": 0.0005928153151217666,
      "loss": 2.3933,
      "step": 16083
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.5666313171386719,
      "learning_rate": 0.0005928144252271137,
      "loss": 2.2588,
      "step": 16084
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6031266450881958,
      "learning_rate": 0.0005928135352780212,
      "loss": 2.2008,
      "step": 16085
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.5706034302711487,
      "learning_rate": 0.0005928126452744891,
      "loss": 2.2192,
      "step": 16086
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6033743023872375,
      "learning_rate": 0.0005928117552165175,
      "loss": 2.5171,
      "step": 16087
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6299945116043091,
      "learning_rate": 0.0005928108651041068,
      "loss": 2.3456,
      "step": 16088
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.5871455073356628,
      "learning_rate": 0.000592809974937257,
      "loss": 2.5,
      "step": 16089
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.723589301109314,
      "learning_rate": 0.0005928090847159682,
      "loss": 2.1913,
      "step": 16090
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.5901269316673279,
      "learning_rate": 0.0005928081944402408,
      "loss": 2.2536,
      "step": 16091
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.5627843737602234,
      "learning_rate": 0.0005928073041100748,
      "loss": 2.2521,
      "step": 16092
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6318639516830444,
      "learning_rate": 0.0005928064137254704,
      "loss": 2.4492,
      "step": 16093
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.5556340217590332,
      "learning_rate": 0.0005928055232864277,
      "loss": 2.3205,
      "step": 16094
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6601184010505676,
      "learning_rate": 0.0005928046327929469,
      "loss": 2.3271,
      "step": 16095
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6428089141845703,
      "learning_rate": 0.0005928037422450284,
      "loss": 2.5755,
      "step": 16096
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.7062668204307556,
      "learning_rate": 0.000592802851642672,
      "loss": 2.3877,
      "step": 16097
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6196726560592651,
      "learning_rate": 0.0005928019609858779,
      "loss": 2.3584,
      "step": 16098
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.5886944532394409,
      "learning_rate": 0.0005928010702746466,
      "loss": 2.4607,
      "step": 16099
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6394200325012207,
      "learning_rate": 0.0005928001795089779,
      "loss": 2.5927,
      "step": 16100
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.5968820452690125,
      "learning_rate": 0.0005927992886888723,
      "loss": 2.3654,
      "step": 16101
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6272449493408203,
      "learning_rate": 0.0005927983978143296,
      "loss": 2.1665,
      "step": 16102
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6317843198776245,
      "learning_rate": 0.0005927975068853503,
      "loss": 2.3678,
      "step": 16103
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.5819616913795471,
      "learning_rate": 0.0005927966159019343,
      "loss": 2.2276,
      "step": 16104
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6858924031257629,
      "learning_rate": 0.0005927957248640819,
      "loss": 2.4828,
      "step": 16105
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.5860427021980286,
      "learning_rate": 0.0005927948337717933,
      "loss": 2.4245,
      "step": 16106
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6297350525856018,
      "learning_rate": 0.0005927939426250686,
      "loss": 2.3208,
      "step": 16107
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6127680540084839,
      "learning_rate": 0.000592793051423908,
      "loss": 2.32,
      "step": 16108
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6848432421684265,
      "learning_rate": 0.0005927921601683115,
      "loss": 2.5011,
      "step": 16109
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.5873380899429321,
      "learning_rate": 0.0005927912688582796,
      "loss": 2.4208,
      "step": 16110
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.5866541862487793,
      "learning_rate": 0.0005927903774938122,
      "loss": 2.4054,
      "step": 16111
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.561659038066864,
      "learning_rate": 0.0005927894860749095,
      "loss": 2.2712,
      "step": 16112
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6467036604881287,
      "learning_rate": 0.0005927885946015717,
      "loss": 2.7257,
      "step": 16113
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.5747194886207581,
      "learning_rate": 0.000592787703073799,
      "loss": 2.4871,
      "step": 16114
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.5648949146270752,
      "learning_rate": 0.0005927868114915917,
      "loss": 2.2887,
      "step": 16115
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.5776881575584412,
      "learning_rate": 0.0005927859198549496,
      "loss": 2.4198,
      "step": 16116
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6442054510116577,
      "learning_rate": 0.0005927850281638731,
      "loss": 2.5414,
      "step": 16117
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6085842847824097,
      "learning_rate": 0.0005927841364183625,
      "loss": 2.6046,
      "step": 16118
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6190785765647888,
      "learning_rate": 0.0005927832446184176,
      "loss": 2.4547,
      "step": 16119
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.5651711821556091,
      "learning_rate": 0.0005927823527640388,
      "loss": 2.4369,
      "step": 16120
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6336466073989868,
      "learning_rate": 0.0005927814608552263,
      "loss": 2.5792,
      "step": 16121
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6822290420532227,
      "learning_rate": 0.0005927805688919802,
      "loss": 2.4429,
      "step": 16122
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.5566661357879639,
      "learning_rate": 0.0005927796768743006,
      "loss": 2.4914,
      "step": 16123
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6644113659858704,
      "learning_rate": 0.0005927787848021877,
      "loss": 2.4759,
      "step": 16124
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6248963475227356,
      "learning_rate": 0.0005927778926756419,
      "loss": 2.34,
      "step": 16125
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6578441858291626,
      "learning_rate": 0.0005927770004946631,
      "loss": 2.3688,
      "step": 16126
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.538250744342804,
      "learning_rate": 0.0005927761082592513,
      "loss": 2.2604,
      "step": 16127
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6610088348388672,
      "learning_rate": 0.0005927752159694071,
      "loss": 2.4471,
      "step": 16128
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.640021800994873,
      "learning_rate": 0.0005927743236251305,
      "loss": 2.5273,
      "step": 16129
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6242701411247253,
      "learning_rate": 0.0005927734312264214,
      "loss": 2.1317,
      "step": 16130
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.8391276001930237,
      "learning_rate": 0.0005927725387732804,
      "loss": 2.3131,
      "step": 16131
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6456416249275208,
      "learning_rate": 0.0005927716462657074,
      "loss": 2.3411,
      "step": 16132
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6266180276870728,
      "learning_rate": 0.0005927707537037027,
      "loss": 2.3443,
      "step": 16133
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6439316272735596,
      "learning_rate": 0.0005927698610872662,
      "loss": 2.5392,
      "step": 16134
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.7155918478965759,
      "learning_rate": 0.0005927689684163984,
      "loss": 2.7296,
      "step": 16135
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6236096620559692,
      "learning_rate": 0.0005927680756910992,
      "loss": 2.2693,
      "step": 16136
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6100817322731018,
      "learning_rate": 0.0005927671829113691,
      "loss": 2.3605,
      "step": 16137
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6984397768974304,
      "learning_rate": 0.0005927662900772079,
      "loss": 2.4678,
      "step": 16138
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.7461010217666626,
      "learning_rate": 0.000592765397188616,
      "loss": 2.562,
      "step": 16139
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6271092295646667,
      "learning_rate": 0.0005927645042455934,
      "loss": 2.4351,
      "step": 16140
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6810237765312195,
      "learning_rate": 0.0005927636112481404,
      "loss": 2.2919,
      "step": 16141
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6122378706932068,
      "learning_rate": 0.0005927627181962571,
      "loss": 2.61,
      "step": 16142
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.5680477023124695,
      "learning_rate": 0.0005927618250899437,
      "loss": 2.2736,
      "step": 16143
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6232805252075195,
      "learning_rate": 0.0005927609319292004,
      "loss": 2.4878,
      "step": 16144
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.7744967341423035,
      "learning_rate": 0.0005927600387140272,
      "loss": 2.6007,
      "step": 16145
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.8331509232521057,
      "learning_rate": 0.0005927591454444244,
      "loss": 2.2518,
      "step": 16146
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.5868799090385437,
      "learning_rate": 0.0005927582521203923,
      "loss": 2.692,
      "step": 16147
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.561053454875946,
      "learning_rate": 0.0005927573587419308,
      "loss": 2.414,
      "step": 16148
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6790440678596497,
      "learning_rate": 0.0005927564653090402,
      "loss": 2.4263,
      "step": 16149
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.632043719291687,
      "learning_rate": 0.0005927555718217205,
      "loss": 2.6349,
      "step": 16150
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6089423894882202,
      "learning_rate": 0.0005927546782799723,
      "loss": 2.6282,
      "step": 16151
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.563825786113739,
      "learning_rate": 0.0005927537846837953,
      "loss": 2.5095,
      "step": 16152
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.5817410945892334,
      "learning_rate": 0.0005927528910331898,
      "loss": 2.4993,
      "step": 16153
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6083444952964783,
      "learning_rate": 0.0005927519973281561,
      "loss": 2.5382,
      "step": 16154
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.7351128458976746,
      "learning_rate": 0.0005927511035686943,
      "loss": 2.3182,
      "step": 16155
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6594401597976685,
      "learning_rate": 0.0005927502097548045,
      "loss": 2.3194,
      "step": 16156
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.8511615991592407,
      "learning_rate": 0.0005927493158864868,
      "loss": 2.3866,
      "step": 16157
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6901085376739502,
      "learning_rate": 0.0005927484219637417,
      "loss": 2.52,
      "step": 16158
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.647323727607727,
      "learning_rate": 0.000592747527986569,
      "loss": 2.3815,
      "step": 16159
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.7015216946601868,
      "learning_rate": 0.000592746633954969,
      "loss": 2.3935,
      "step": 16160
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6489140391349792,
      "learning_rate": 0.0005927457398689419,
      "loss": 2.4222,
      "step": 16161
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6134150624275208,
      "learning_rate": 0.0005927448457284879,
      "loss": 2.4006,
      "step": 16162
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6263222098350525,
      "learning_rate": 0.0005927439515336071,
      "loss": 2.4848,
      "step": 16163
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.7844492793083191,
      "learning_rate": 0.0005927430572842997,
      "loss": 2.4222,
      "step": 16164
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6089938282966614,
      "learning_rate": 0.0005927421629805656,
      "loss": 2.1752,
      "step": 16165
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6421418190002441,
      "learning_rate": 0.0005927412686224055,
      "loss": 2.2458,
      "step": 16166
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6918745040893555,
      "learning_rate": 0.0005927403742098192,
      "loss": 2.3047,
      "step": 16167
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.7092336416244507,
      "learning_rate": 0.0005927394797428068,
      "loss": 2.4196,
      "step": 16168
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6637004613876343,
      "learning_rate": 0.0005927385852213687,
      "loss": 2.3102,
      "step": 16169
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.8512107133865356,
      "learning_rate": 0.000592737690645505,
      "loss": 2.5925,
      "step": 16170
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.7137554287910461,
      "learning_rate": 0.0005927367960152158,
      "loss": 2.4755,
      "step": 16171
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6937549114227295,
      "learning_rate": 0.0005927359013305014,
      "loss": 2.2747,
      "step": 16172
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.7970254421234131,
      "learning_rate": 0.0005927350065913617,
      "loss": 2.3342,
      "step": 16173
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.7164129614830017,
      "learning_rate": 0.0005927341117977972,
      "loss": 2.3147,
      "step": 16174
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.5626000761985779,
      "learning_rate": 0.0005927332169498077,
      "loss": 2.5613,
      "step": 16175
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6918309330940247,
      "learning_rate": 0.0005927323220473937,
      "loss": 2.6508,
      "step": 16176
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6985851526260376,
      "learning_rate": 0.0005927314270905552,
      "loss": 2.3407,
      "step": 16177
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.608290433883667,
      "learning_rate": 0.0005927305320792925,
      "loss": 2.4254,
      "step": 16178
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.575569748878479,
      "learning_rate": 0.0005927296370136056,
      "loss": 2.4642,
      "step": 16179
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6758687496185303,
      "learning_rate": 0.0005927287418934947,
      "loss": 2.2972,
      "step": 16180
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.5833088755607605,
      "learning_rate": 0.0005927278467189601,
      "loss": 2.3532,
      "step": 16181
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6105799674987793,
      "learning_rate": 0.0005927269514900017,
      "loss": 2.3257,
      "step": 16182
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.5823343396186829,
      "learning_rate": 0.00059272605620662,
      "loss": 2.405,
      "step": 16183
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.695812463760376,
      "learning_rate": 0.0005927251608688149,
      "loss": 2.2791,
      "step": 16184
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.5547595620155334,
      "learning_rate": 0.0005927242654765867,
      "loss": 2.4848,
      "step": 16185
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.7542863488197327,
      "learning_rate": 0.0005927233700299356,
      "loss": 2.4372,
      "step": 16186
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6059430241584778,
      "learning_rate": 0.0005927224745288617,
      "loss": 2.3129,
      "step": 16187
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.5668745040893555,
      "learning_rate": 0.000592721578973365,
      "loss": 2.2448,
      "step": 16188
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.5716577172279358,
      "learning_rate": 0.0005927206833634459,
      "loss": 2.3031,
      "step": 16189
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.5997764468193054,
      "learning_rate": 0.0005927197876991045,
      "loss": 2.3511,
      "step": 16190
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6433826684951782,
      "learning_rate": 0.000592718891980341,
      "loss": 2.5012,
      "step": 16191
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6251102089881897,
      "learning_rate": 0.0005927179962071556,
      "loss": 2.4767,
      "step": 16192
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6410203576087952,
      "learning_rate": 0.0005927171003795483,
      "loss": 2.3441,
      "step": 16193
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6592394709587097,
      "learning_rate": 0.0005927162044975194,
      "loss": 2.4367,
      "step": 16194
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.5902048349380493,
      "learning_rate": 0.000592715308561069,
      "loss": 2.3456,
      "step": 16195
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.5896995067596436,
      "learning_rate": 0.0005927144125701973,
      "loss": 2.4611,
      "step": 16196
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6009373068809509,
      "learning_rate": 0.0005927135165249046,
      "loss": 2.1894,
      "step": 16197
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.5886870622634888,
      "learning_rate": 0.0005927126204251908,
      "loss": 2.2468,
      "step": 16198
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6017231941223145,
      "learning_rate": 0.0005927117242710562,
      "loss": 2.2188,
      "step": 16199
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.5978803634643555,
      "learning_rate": 0.000592710828062501,
      "loss": 2.4425,
      "step": 16200
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6618629097938538,
      "learning_rate": 0.0005927099317995253,
      "loss": 2.3774,
      "step": 16201
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.5530021786689758,
      "learning_rate": 0.0005927090354821294,
      "loss": 2.311,
      "step": 16202
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6051048636436462,
      "learning_rate": 0.0005927081391103132,
      "loss": 2.4056,
      "step": 16203
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.5889014601707458,
      "learning_rate": 0.0005927072426840771,
      "loss": 2.4785,
      "step": 16204
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6268645524978638,
      "learning_rate": 0.0005927063462034212,
      "loss": 2.3511,
      "step": 16205
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6186964511871338,
      "learning_rate": 0.0005927054496683458,
      "loss": 2.097,
      "step": 16206
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6487101912498474,
      "learning_rate": 0.0005927045530788507,
      "loss": 2.438,
      "step": 16207
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6503812074661255,
      "learning_rate": 0.0005927036564349364,
      "loss": 2.2121,
      "step": 16208
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.5911717414855957,
      "learning_rate": 0.000592702759736603,
      "loss": 2.3827,
      "step": 16209
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.5435114502906799,
      "learning_rate": 0.0005927018629838506,
      "loss": 2.4519,
      "step": 16210
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.5795292258262634,
      "learning_rate": 0.0005927009661766794,
      "loss": 2.4283,
      "step": 16211
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.5984902381896973,
      "learning_rate": 0.0005927000693150896,
      "loss": 2.5776,
      "step": 16212
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.5779632329940796,
      "learning_rate": 0.0005926991723990813,
      "loss": 2.1063,
      "step": 16213
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6209304332733154,
      "learning_rate": 0.0005926982754286547,
      "loss": 2.4045,
      "step": 16214
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6271655559539795,
      "learning_rate": 0.0005926973784038099,
      "loss": 2.4141,
      "step": 16215
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6239400506019592,
      "learning_rate": 0.0005926964813245473,
      "loss": 2.6048,
      "step": 16216
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.5935445427894592,
      "learning_rate": 0.0005926955841908668,
      "loss": 2.5048,
      "step": 16217
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.639589250087738,
      "learning_rate": 0.0005926946870027686,
      "loss": 2.3013,
      "step": 16218
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6429893374443054,
      "learning_rate": 0.000592693789760253,
      "loss": 2.3884,
      "step": 16219
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6046529412269592,
      "learning_rate": 0.0005926928924633201,
      "loss": 2.446,
      "step": 16220
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.613444447517395,
      "learning_rate": 0.00059269199511197,
      "loss": 2.234,
      "step": 16221
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.5762456655502319,
      "learning_rate": 0.000592691097706203,
      "loss": 2.3203,
      "step": 16222
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.5640515685081482,
      "learning_rate": 0.0005926902002460193,
      "loss": 2.2782,
      "step": 16223
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6309368014335632,
      "learning_rate": 0.0005926893027314188,
      "loss": 2.2337,
      "step": 16224
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6156130433082581,
      "learning_rate": 0.0005926884051624019,
      "loss": 2.2848,
      "step": 16225
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6039855480194092,
      "learning_rate": 0.0005926875075389687,
      "loss": 2.256,
      "step": 16226
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6229126453399658,
      "learning_rate": 0.0005926866098611193,
      "loss": 2.4077,
      "step": 16227
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6257598996162415,
      "learning_rate": 0.0005926857121288541,
      "loss": 2.4347,
      "step": 16228
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6614488363265991,
      "learning_rate": 0.000592684814342173,
      "loss": 2.4119,
      "step": 16229
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6922187209129333,
      "learning_rate": 0.0005926839165010762,
      "loss": 2.2316,
      "step": 16230
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6543877124786377,
      "learning_rate": 0.000592683018605564,
      "loss": 2.3749,
      "step": 16231
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6102409362792969,
      "learning_rate": 0.0005926821206556365,
      "loss": 2.306,
      "step": 16232
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6700664162635803,
      "learning_rate": 0.0005926812226512938,
      "loss": 2.4001,
      "step": 16233
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6362878084182739,
      "learning_rate": 0.0005926803245925363,
      "loss": 2.5465,
      "step": 16234
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6047970652580261,
      "learning_rate": 0.0005926794264793639,
      "loss": 2.648,
      "step": 16235
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6459817290306091,
      "learning_rate": 0.0005926785283117769,
      "loss": 2.534,
      "step": 16236
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.595318078994751,
      "learning_rate": 0.0005926776300897754,
      "loss": 2.5237,
      "step": 16237
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6344490647315979,
      "learning_rate": 0.0005926767318133597,
      "loss": 2.4479,
      "step": 16238
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6063999533653259,
      "learning_rate": 0.0005926758334825297,
      "loss": 2.3968,
      "step": 16239
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6608055830001831,
      "learning_rate": 0.0005926749350972858,
      "loss": 2.2362,
      "step": 16240
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6024240851402283,
      "learning_rate": 0.0005926740366576282,
      "loss": 2.4432,
      "step": 16241
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6141383051872253,
      "learning_rate": 0.0005926731381635569,
      "loss": 2.3338,
      "step": 16242
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6713558435440063,
      "learning_rate": 0.0005926722396150722,
      "loss": 2.3788,
      "step": 16243
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.5955862998962402,
      "learning_rate": 0.0005926713410121741,
      "loss": 2.4842,
      "step": 16244
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6154022216796875,
      "learning_rate": 0.0005926704423548629,
      "loss": 2.5509,
      "step": 16245
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6438125967979431,
      "learning_rate": 0.0005926695436431387,
      "loss": 2.3066,
      "step": 16246
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6039091944694519,
      "learning_rate": 0.0005926686448770019,
      "loss": 2.3864,
      "step": 16247
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6114537119865417,
      "learning_rate": 0.0005926677460564524,
      "loss": 2.4061,
      "step": 16248
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6407271027565002,
      "learning_rate": 0.0005926668471814903,
      "loss": 2.5427,
      "step": 16249
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6125162243843079,
      "learning_rate": 0.000592665948252116,
      "loss": 2.2987,
      "step": 16250
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.5654097199440002,
      "learning_rate": 0.0005926650492683296,
      "loss": 2.577,
      "step": 16251
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6242508888244629,
      "learning_rate": 0.0005926641502301312,
      "loss": 2.4741,
      "step": 16252
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6579198241233826,
      "learning_rate": 0.000592663251137521,
      "loss": 2.1419,
      "step": 16253
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6380335688591003,
      "learning_rate": 0.0005926623519904992,
      "loss": 2.4676,
      "step": 16254
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6877772212028503,
      "learning_rate": 0.000592661452789066,
      "loss": 2.2926,
      "step": 16255
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.670367956161499,
      "learning_rate": 0.0005926605535332214,
      "loss": 2.3997,
      "step": 16256
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6306771636009216,
      "learning_rate": 0.0005926596542229658,
      "loss": 2.4643,
      "step": 16257
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.7066945433616638,
      "learning_rate": 0.0005926587548582992,
      "loss": 2.518,
      "step": 16258
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.8183748722076416,
      "learning_rate": 0.0005926578554392218,
      "loss": 2.5146,
      "step": 16259
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6145703196525574,
      "learning_rate": 0.0005926569559657337,
      "loss": 2.4976,
      "step": 16260
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6423473358154297,
      "learning_rate": 0.0005926560564378353,
      "loss": 2.3269,
      "step": 16261
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.7177550196647644,
      "learning_rate": 0.0005926551568555265,
      "loss": 2.6081,
      "step": 16262
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6088459491729736,
      "learning_rate": 0.0005926542572188076,
      "loss": 2.3009,
      "step": 16263
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.5350183844566345,
      "learning_rate": 0.0005926533575276787,
      "loss": 2.2739,
      "step": 16264
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6613390445709229,
      "learning_rate": 0.0005926524577821401,
      "loss": 2.3841,
      "step": 16265
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6962381601333618,
      "learning_rate": 0.000592651557982192,
      "loss": 2.4372,
      "step": 16266
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.5851055383682251,
      "learning_rate": 0.0005926506581278343,
      "loss": 2.3778,
      "step": 16267
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.7295133471488953,
      "learning_rate": 0.0005926497582190673,
      "loss": 2.3992,
      "step": 16268
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.5934313535690308,
      "learning_rate": 0.0005926488582558911,
      "loss": 2.4449,
      "step": 16269
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.5925232768058777,
      "learning_rate": 0.0005926479582383062,
      "loss": 2.3761,
      "step": 16270
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.5851154327392578,
      "learning_rate": 0.0005926470581663124,
      "loss": 2.3811,
      "step": 16271
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6899107694625854,
      "learning_rate": 0.00059264615803991,
      "loss": 2.3535,
      "step": 16272
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6474761366844177,
      "learning_rate": 0.0005926452578590991,
      "loss": 2.3797,
      "step": 16273
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6159665584564209,
      "learning_rate": 0.00059264435762388,
      "loss": 2.3581,
      "step": 16274
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6221139430999756,
      "learning_rate": 0.0005926434573342527,
      "loss": 2.3826,
      "step": 16275
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.5780224204063416,
      "learning_rate": 0.0005926425569902175,
      "loss": 2.2179,
      "step": 16276
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6694228053092957,
      "learning_rate": 0.0005926416565917746,
      "loss": 2.3183,
      "step": 16277
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6476938724517822,
      "learning_rate": 0.000592640756138924,
      "loss": 2.4123,
      "step": 16278
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6183919906616211,
      "learning_rate": 0.000592639855631666,
      "loss": 2.2371,
      "step": 16279
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6156599521636963,
      "learning_rate": 0.0005926389550700008,
      "loss": 2.4106,
      "step": 16280
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6409339308738708,
      "learning_rate": 0.0005926380544539284,
      "loss": 2.3224,
      "step": 16281
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.64841628074646,
      "learning_rate": 0.000592637153783449,
      "loss": 2.5086,
      "step": 16282
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6275238990783691,
      "learning_rate": 0.0005926362530585629,
      "loss": 2.2902,
      "step": 16283
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.5750252604484558,
      "learning_rate": 0.0005926353522792702,
      "loss": 2.5902,
      "step": 16284
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6200417280197144,
      "learning_rate": 0.000592634451445571,
      "loss": 2.0623,
      "step": 16285
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6675677299499512,
      "learning_rate": 0.0005926335505574657,
      "loss": 2.5962,
      "step": 16286
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6343095302581787,
      "learning_rate": 0.0005926326496149541,
      "loss": 2.4609,
      "step": 16287
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6204013228416443,
      "learning_rate": 0.0005926317486180366,
      "loss": 2.3337,
      "step": 16288
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.5832129120826721,
      "learning_rate": 0.0005926308475667136,
      "loss": 2.5008,
      "step": 16289
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6452401280403137,
      "learning_rate": 0.0005926299464609847,
      "loss": 2.6568,
      "step": 16290
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.5940415263175964,
      "learning_rate": 0.0005926290453008504,
      "loss": 2.3598,
      "step": 16291
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.5816642045974731,
      "learning_rate": 0.000592628144086311,
      "loss": 2.2757,
      "step": 16292
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6532630324363708,
      "learning_rate": 0.0005926272428173663,
      "loss": 2.4408,
      "step": 16293
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6067154407501221,
      "learning_rate": 0.0005926263414940168,
      "loss": 2.4377,
      "step": 16294
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6472457051277161,
      "learning_rate": 0.0005926254401162625,
      "loss": 2.3226,
      "step": 16295
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6537804007530212,
      "learning_rate": 0.0005926245386841037,
      "loss": 2.314,
      "step": 16296
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.5782991647720337,
      "learning_rate": 0.0005926236371975403,
      "loss": 2.3742,
      "step": 16297
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.7013685703277588,
      "learning_rate": 0.0005926227356565729,
      "loss": 2.5265,
      "step": 16298
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.647529125213623,
      "learning_rate": 0.0005926218340612012,
      "loss": 2.4979,
      "step": 16299
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.5949440002441406,
      "learning_rate": 0.0005926209324114256,
      "loss": 2.3273,
      "step": 16300
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6312925219535828,
      "learning_rate": 0.0005926200307072463,
      "loss": 2.3775,
      "step": 16301
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.631104052066803,
      "learning_rate": 0.0005926191289486633,
      "loss": 2.4127,
      "step": 16302
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.7026100158691406,
      "learning_rate": 0.000592618227135677,
      "loss": 2.3619,
      "step": 16303
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6126165390014648,
      "learning_rate": 0.0005926173252682875,
      "loss": 2.4692,
      "step": 16304
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.5695275068283081,
      "learning_rate": 0.0005926164233464947,
      "loss": 2.3106,
      "step": 16305
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.598651647567749,
      "learning_rate": 0.0005926155213702992,
      "loss": 2.2414,
      "step": 16306
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.7155926823616028,
      "learning_rate": 0.0005926146193397008,
      "loss": 2.2518,
      "step": 16307
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.5773846507072449,
      "learning_rate": 0.0005926137172546998,
      "loss": 2.3172,
      "step": 16308
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6094586849212646,
      "learning_rate": 0.0005926128151152965,
      "loss": 2.3306,
      "step": 16309
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.590110182762146,
      "learning_rate": 0.000592611912921491,
      "loss": 2.3795,
      "step": 16310
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.5889648795127869,
      "learning_rate": 0.0005926110106732833,
      "loss": 2.33,
      "step": 16311
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.674518883228302,
      "learning_rate": 0.0005926101083706737,
      "loss": 2.2416,
      "step": 16312
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6149779558181763,
      "learning_rate": 0.0005926092060136624,
      "loss": 2.6067,
      "step": 16313
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.5598203539848328,
      "learning_rate": 0.0005926083036022496,
      "loss": 2.4093,
      "step": 16314
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.5968289375305176,
      "learning_rate": 0.0005926074011364353,
      "loss": 2.372,
      "step": 16315
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6229177117347717,
      "learning_rate": 0.0005926064986162197,
      "loss": 2.4332,
      "step": 16316
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6333149671554565,
      "learning_rate": 0.0005926055960416032,
      "loss": 2.3073,
      "step": 16317
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.649197518825531,
      "learning_rate": 0.0005926046934125856,
      "loss": 2.6721,
      "step": 16318
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6766908168792725,
      "learning_rate": 0.0005926037907291673,
      "loss": 2.4495,
      "step": 16319
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6176275014877319,
      "learning_rate": 0.0005926028879913485,
      "loss": 2.5033,
      "step": 16320
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6056455373764038,
      "learning_rate": 0.0005926019851991293,
      "loss": 2.0406,
      "step": 16321
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.5988695621490479,
      "learning_rate": 0.0005926010823525099,
      "loss": 2.3072,
      "step": 16322
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6703783273696899,
      "learning_rate": 0.0005926001794514904,
      "loss": 2.1858,
      "step": 16323
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6680948734283447,
      "learning_rate": 0.000592599276496071,
      "loss": 2.4349,
      "step": 16324
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6932280659675598,
      "learning_rate": 0.0005925983734862519,
      "loss": 2.5263,
      "step": 16325
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6635175943374634,
      "learning_rate": 0.0005925974704220332,
      "loss": 2.4416,
      "step": 16326
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.5895622372627258,
      "learning_rate": 0.0005925965673034151,
      "loss": 2.3074,
      "step": 16327
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.5979746580123901,
      "learning_rate": 0.0005925956641303978,
      "loss": 2.689,
      "step": 16328
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6714795827865601,
      "learning_rate": 0.0005925947609029815,
      "loss": 2.6252,
      "step": 16329
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6223028302192688,
      "learning_rate": 0.0005925938576211661,
      "loss": 2.3922,
      "step": 16330
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6260518431663513,
      "learning_rate": 0.0005925929542849521,
      "loss": 2.3363,
      "step": 16331
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6026412844657898,
      "learning_rate": 0.0005925920508943396,
      "loss": 2.2073,
      "step": 16332
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.5606334209442139,
      "learning_rate": 0.0005925911474493287,
      "loss": 2.3296,
      "step": 16333
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.5926303863525391,
      "learning_rate": 0.0005925902439499197,
      "loss": 2.5345,
      "step": 16334
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6842694878578186,
      "learning_rate": 0.0005925893403961124,
      "loss": 2.5731,
      "step": 16335
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.5964046716690063,
      "learning_rate": 0.0005925884367879073,
      "loss": 2.3496,
      "step": 16336
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6950357556343079,
      "learning_rate": 0.0005925875331253045,
      "loss": 2.2497,
      "step": 16337
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6374891996383667,
      "learning_rate": 0.0005925866294083042,
      "loss": 2.2693,
      "step": 16338
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6158180236816406,
      "learning_rate": 0.0005925857256369065,
      "loss": 2.6087,
      "step": 16339
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.7362374067306519,
      "learning_rate": 0.0005925848218111116,
      "loss": 2.4665,
      "step": 16340
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.737089991569519,
      "learning_rate": 0.0005925839179309197,
      "loss": 2.579,
      "step": 16341
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.5861368179321289,
      "learning_rate": 0.0005925830139963308,
      "loss": 2.4647,
      "step": 16342
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.746868371963501,
      "learning_rate": 0.0005925821100073454,
      "loss": 2.5562,
      "step": 16343
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.7560282945632935,
      "learning_rate": 0.0005925812059639632,
      "loss": 2.3033,
      "step": 16344
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6628333330154419,
      "learning_rate": 0.0005925803018661848,
      "loss": 2.316,
      "step": 16345
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6269702315330505,
      "learning_rate": 0.0005925793977140101,
      "loss": 2.3419,
      "step": 16346
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6672741770744324,
      "learning_rate": 0.0005925784935074395,
      "loss": 2.5099,
      "step": 16347
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6631382703781128,
      "learning_rate": 0.000592577589246473,
      "loss": 2.3743,
      "step": 16348
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6333550810813904,
      "learning_rate": 0.0005925766849311107,
      "loss": 2.2732,
      "step": 16349
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6332792639732361,
      "learning_rate": 0.000592575780561353,
      "loss": 2.2221,
      "step": 16350
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6940414905548096,
      "learning_rate": 0.0005925748761371998,
      "loss": 2.6294,
      "step": 16351
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.7333903312683105,
      "learning_rate": 0.0005925739716586515,
      "loss": 2.2975,
      "step": 16352
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6449540853500366,
      "learning_rate": 0.0005925730671257082,
      "loss": 2.6405,
      "step": 16353
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6440787315368652,
      "learning_rate": 0.00059257216253837,
      "loss": 2.3276,
      "step": 16354
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6097280979156494,
      "learning_rate": 0.0005925712578966371,
      "loss": 2.2397,
      "step": 16355
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6825025677680969,
      "learning_rate": 0.0005925703532005097,
      "loss": 2.2865,
      "step": 16356
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.5950647592544556,
      "learning_rate": 0.000592569448449988,
      "loss": 2.5876,
      "step": 16357
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6453235745429993,
      "learning_rate": 0.000592568543645072,
      "loss": 2.353,
      "step": 16358
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6134476661682129,
      "learning_rate": 0.000592567638785762,
      "loss": 2.173,
      "step": 16359
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.5902847051620483,
      "learning_rate": 0.0005925667338720583,
      "loss": 2.267,
      "step": 16360
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.5788105726242065,
      "learning_rate": 0.0005925658289039609,
      "loss": 2.2143,
      "step": 16361
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6395125985145569,
      "learning_rate": 0.0005925649238814698,
      "loss": 2.2135,
      "step": 16362
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.5705952644348145,
      "learning_rate": 0.0005925640188045855,
      "loss": 2.6137,
      "step": 16363
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6339254379272461,
      "learning_rate": 0.0005925631136733079,
      "loss": 2.4242,
      "step": 16364
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6195516586303711,
      "learning_rate": 0.0005925622084876375,
      "loss": 2.3538,
      "step": 16365
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6323096752166748,
      "learning_rate": 0.0005925613032475741,
      "loss": 2.0788,
      "step": 16366
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.8531841039657593,
      "learning_rate": 0.0005925603979531181,
      "loss": 2.0786,
      "step": 16367
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6402493119239807,
      "learning_rate": 0.0005925594926042696,
      "loss": 2.5721,
      "step": 16368
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6142857074737549,
      "learning_rate": 0.0005925585872010288,
      "loss": 2.4811,
      "step": 16369
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.5999394655227661,
      "learning_rate": 0.0005925576817433957,
      "loss": 2.4148,
      "step": 16370
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.5825631618499756,
      "learning_rate": 0.0005925567762313707,
      "loss": 2.3813,
      "step": 16371
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6607882380485535,
      "learning_rate": 0.0005925558706649539,
      "loss": 2.5927,
      "step": 16372
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6177992224693298,
      "learning_rate": 0.0005925549650441454,
      "loss": 2.1677,
      "step": 16373
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6470689177513123,
      "learning_rate": 0.0005925540593689454,
      "loss": 2.3421,
      "step": 16374
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.7003798484802246,
      "learning_rate": 0.0005925531536393541,
      "loss": 2.5759,
      "step": 16375
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6077347993850708,
      "learning_rate": 0.0005925522478553716,
      "loss": 2.4159,
      "step": 16376
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.572030782699585,
      "learning_rate": 0.0005925513420169982,
      "loss": 2.3453,
      "step": 16377
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6266578435897827,
      "learning_rate": 0.000592550436124234,
      "loss": 2.5011,
      "step": 16378
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6531533598899841,
      "learning_rate": 0.000592549530177079,
      "loss": 2.2112,
      "step": 16379
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6205577850341797,
      "learning_rate": 0.0005925486241755337,
      "loss": 2.6515,
      "step": 16380
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6693223118782043,
      "learning_rate": 0.000592547718119598,
      "loss": 2.3682,
      "step": 16381
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.7404983043670654,
      "learning_rate": 0.0005925468120092721,
      "loss": 2.6093,
      "step": 16382
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6685624718666077,
      "learning_rate": 0.0005925459058445563,
      "loss": 2.2491,
      "step": 16383
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6139832735061646,
      "learning_rate": 0.0005925449996254506,
      "loss": 2.1845,
      "step": 16384
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6272143125534058,
      "learning_rate": 0.0005925440933519555,
      "loss": 2.4513,
      "step": 16385
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.633847713470459,
      "learning_rate": 0.0005925431870240707,
      "loss": 2.2915,
      "step": 16386
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.7503389716148376,
      "learning_rate": 0.0005925422806417966,
      "loss": 2.4468,
      "step": 16387
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.594277024269104,
      "learning_rate": 0.0005925413742051335,
      "loss": 2.3828,
      "step": 16388
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6215713620185852,
      "learning_rate": 0.0005925404677140813,
      "loss": 2.3084,
      "step": 16389
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6458708047866821,
      "learning_rate": 0.0005925395611686404,
      "loss": 2.5795,
      "step": 16390
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.7331948280334473,
      "learning_rate": 0.0005925386545688109,
      "loss": 2.4703,
      "step": 16391
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.600272536277771,
      "learning_rate": 0.0005925377479145929,
      "loss": 2.5409,
      "step": 16392
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.8276382684707642,
      "learning_rate": 0.0005925368412059867,
      "loss": 2.203,
      "step": 16393
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.614184558391571,
      "learning_rate": 0.0005925359344429922,
      "loss": 1.9769,
      "step": 16394
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6361179947853088,
      "learning_rate": 0.0005925350276256099,
      "loss": 2.5976,
      "step": 16395
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.7957091927528381,
      "learning_rate": 0.0005925341207538397,
      "loss": 2.5879,
      "step": 16396
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.7095101475715637,
      "learning_rate": 0.0005925332138276819,
      "loss": 2.41,
      "step": 16397
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6519250869750977,
      "learning_rate": 0.0005925323068471367,
      "loss": 2.4733,
      "step": 16398
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6229072213172913,
      "learning_rate": 0.0005925313998122042,
      "loss": 2.1595,
      "step": 16399
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6806896328926086,
      "learning_rate": 0.0005925304927228846,
      "loss": 2.3175,
      "step": 16400
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6557904481887817,
      "learning_rate": 0.0005925295855791781,
      "loss": 2.5775,
      "step": 16401
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6069420576095581,
      "learning_rate": 0.0005925286783810847,
      "loss": 2.5455,
      "step": 16402
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.7376970052719116,
      "learning_rate": 0.0005925277711286049,
      "loss": 2.6669,
      "step": 16403
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.5949994921684265,
      "learning_rate": 0.0005925268638217384,
      "loss": 2.4192,
      "step": 16404
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6901650428771973,
      "learning_rate": 0.0005925259564604859,
      "loss": 2.3043,
      "step": 16405
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6282534003257751,
      "learning_rate": 0.0005925250490448472,
      "loss": 2.4119,
      "step": 16406
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6361590623855591,
      "learning_rate": 0.0005925241415748225,
      "loss": 2.6229,
      "step": 16407
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6014980673789978,
      "learning_rate": 0.0005925232340504122,
      "loss": 2.6743,
      "step": 16408
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6404474377632141,
      "learning_rate": 0.0005925223264716161,
      "loss": 2.5036,
      "step": 16409
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.5871490836143494,
      "learning_rate": 0.0005925214188384347,
      "loss": 2.2009,
      "step": 16410
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6246122121810913,
      "learning_rate": 0.000592520511150868,
      "loss": 2.4883,
      "step": 16411
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6340320110321045,
      "learning_rate": 0.0005925196034089163,
      "loss": 2.1765,
      "step": 16412
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6855765581130981,
      "learning_rate": 0.0005925186956125796,
      "loss": 2.4289,
      "step": 16413
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.5817277431488037,
      "learning_rate": 0.0005925177877618582,
      "loss": 2.2623,
      "step": 16414
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6151990294456482,
      "learning_rate": 0.0005925168798567522,
      "loss": 2.4805,
      "step": 16415
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.5705024003982544,
      "learning_rate": 0.0005925159718972618,
      "loss": 2.5103,
      "step": 16416
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6576249599456787,
      "learning_rate": 0.0005925150638833871,
      "loss": 2.2996,
      "step": 16417
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.5795753598213196,
      "learning_rate": 0.0005925141558151284,
      "loss": 2.3776,
      "step": 16418
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6476909518241882,
      "learning_rate": 0.0005925132476924858,
      "loss": 2.4366,
      "step": 16419
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6377889513969421,
      "learning_rate": 0.0005925123395154594,
      "loss": 2.4204,
      "step": 16420
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6186497211456299,
      "learning_rate": 0.0005925114312840495,
      "loss": 2.5099,
      "step": 16421
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.5778704285621643,
      "learning_rate": 0.0005925105229982562,
      "loss": 2.0807,
      "step": 16422
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6690744161605835,
      "learning_rate": 0.0005925096146580797,
      "loss": 2.3608,
      "step": 16423
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.5979333519935608,
      "learning_rate": 0.00059250870626352,
      "loss": 2.3983,
      "step": 16424
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6925561428070068,
      "learning_rate": 0.0005925077978145776,
      "loss": 2.6751,
      "step": 16425
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.5855411291122437,
      "learning_rate": 0.0005925068893112524,
      "loss": 2.4632,
      "step": 16426
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.5725460648536682,
      "learning_rate": 0.0005925059807535446,
      "loss": 2.3517,
      "step": 16427
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.5996792912483215,
      "learning_rate": 0.0005925050721414544,
      "loss": 2.2156,
      "step": 16428
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.548522412776947,
      "learning_rate": 0.0005925041634749821,
      "loss": 2.1803,
      "step": 16429
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.5657812356948853,
      "learning_rate": 0.0005925032547541277,
      "loss": 2.3122,
      "step": 16430
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.7041831016540527,
      "learning_rate": 0.0005925023459788914,
      "loss": 2.2233,
      "step": 16431
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6632214784622192,
      "learning_rate": 0.0005925014371492733,
      "loss": 2.3865,
      "step": 16432
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6418792009353638,
      "learning_rate": 0.0005925005282652738,
      "loss": 2.2513,
      "step": 16433
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6307882070541382,
      "learning_rate": 0.000592499619326893,
      "loss": 2.2891,
      "step": 16434
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6306554079055786,
      "learning_rate": 0.0005924987103341308,
      "loss": 2.2178,
      "step": 16435
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.645735502243042,
      "learning_rate": 0.0005924978012869877,
      "loss": 2.2529,
      "step": 16436
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6470157504081726,
      "learning_rate": 0.0005924968921854637,
      "loss": 2.4154,
      "step": 16437
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.5939277410507202,
      "learning_rate": 0.0005924959830295591,
      "loss": 2.4247,
      "step": 16438
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6572065353393555,
      "learning_rate": 0.0005924950738192738,
      "loss": 2.4533,
      "step": 16439
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6223199367523193,
      "learning_rate": 0.0005924941645546083,
      "loss": 2.6051,
      "step": 16440
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6150791645050049,
      "learning_rate": 0.0005924932552355624,
      "loss": 2.6272,
      "step": 16441
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6199024319648743,
      "learning_rate": 0.0005924923458621367,
      "loss": 2.4021,
      "step": 16442
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6527654528617859,
      "learning_rate": 0.0005924914364343311,
      "loss": 2.4213,
      "step": 16443
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.5778728723526001,
      "learning_rate": 0.0005924905269521458,
      "loss": 2.4831,
      "step": 16444
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6174225211143494,
      "learning_rate": 0.000592489617415581,
      "loss": 2.174,
      "step": 16445
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.7045391201972961,
      "learning_rate": 0.0005924887078246367,
      "loss": 2.5012,
      "step": 16446
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.5974153876304626,
      "learning_rate": 0.0005924877981793135,
      "loss": 2.33,
      "step": 16447
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6623116135597229,
      "learning_rate": 0.0005924868884796112,
      "loss": 2.4119,
      "step": 16448
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6226667761802673,
      "learning_rate": 0.00059248597872553,
      "loss": 2.5597,
      "step": 16449
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6281902194023132,
      "learning_rate": 0.0005924850689170702,
      "loss": 2.602,
      "step": 16450
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.5872663259506226,
      "learning_rate": 0.0005924841590542318,
      "loss": 2.4173,
      "step": 16451
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6476262211799622,
      "learning_rate": 0.0005924832491370151,
      "loss": 2.3853,
      "step": 16452
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6022763252258301,
      "learning_rate": 0.0005924823391654204,
      "loss": 2.281,
      "step": 16453
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.5582574009895325,
      "learning_rate": 0.0005924814291394476,
      "loss": 2.3344,
      "step": 16454
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.7162729501724243,
      "learning_rate": 0.000592480519059097,
      "loss": 2.4718,
      "step": 16455
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6306344270706177,
      "learning_rate": 0.0005924796089243687,
      "loss": 2.5685,
      "step": 16456
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.5568484663963318,
      "learning_rate": 0.0005924786987352629,
      "loss": 2.3222,
      "step": 16457
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.5691396594047546,
      "learning_rate": 0.00059247778849178,
      "loss": 2.2645,
      "step": 16458
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6456562876701355,
      "learning_rate": 0.0005924768781939197,
      "loss": 2.3159,
      "step": 16459
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.5605900287628174,
      "learning_rate": 0.0005924759678416826,
      "loss": 2.4617,
      "step": 16460
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6017793416976929,
      "learning_rate": 0.0005924750574350686,
      "loss": 2.5278,
      "step": 16461
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6184453368186951,
      "learning_rate": 0.0005924741469740779,
      "loss": 2.1784,
      "step": 16462
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.7314894795417786,
      "learning_rate": 0.000592473236458711,
      "loss": 2.3862,
      "step": 16463
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.5838156342506409,
      "learning_rate": 0.0005924723258889676,
      "loss": 2.3262,
      "step": 16464
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6605850458145142,
      "learning_rate": 0.000592471415264848,
      "loss": 2.2403,
      "step": 16465
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.5662124752998352,
      "learning_rate": 0.0005924705045863526,
      "loss": 2.5132,
      "step": 16466
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.5384504795074463,
      "learning_rate": 0.0005924695938534813,
      "loss": 2.2868,
      "step": 16467
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.5824654698371887,
      "learning_rate": 0.0005924686830662345,
      "loss": 2.2611,
      "step": 16468
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.5973441004753113,
      "learning_rate": 0.0005924677722246122,
      "loss": 2.6284,
      "step": 16469
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6069483160972595,
      "learning_rate": 0.0005924668613286147,
      "loss": 2.3602,
      "step": 16470
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.793651282787323,
      "learning_rate": 0.000592465950378242,
      "loss": 2.3434,
      "step": 16471
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.7534142136573792,
      "learning_rate": 0.0005924650393734943,
      "loss": 2.3389,
      "step": 16472
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.5712526440620422,
      "learning_rate": 0.0005924641283143719,
      "loss": 2.2725,
      "step": 16473
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.7237074375152588,
      "learning_rate": 0.0005924632172008749,
      "loss": 2.3952,
      "step": 16474
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.5759514570236206,
      "learning_rate": 0.0005924623060330035,
      "loss": 2.4417,
      "step": 16475
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6257093548774719,
      "learning_rate": 0.0005924613948107578,
      "loss": 2.4606,
      "step": 16476
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.5905453562736511,
      "learning_rate": 0.000592460483534138,
      "loss": 2.3598,
      "step": 16477
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6581283807754517,
      "learning_rate": 0.0005924595722031443,
      "loss": 2.3146,
      "step": 16478
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.5783911943435669,
      "learning_rate": 0.0005924586608177767,
      "loss": 2.4613,
      "step": 16479
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6062476634979248,
      "learning_rate": 0.0005924577493780358,
      "loss": 2.4948,
      "step": 16480
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.5798715353012085,
      "learning_rate": 0.0005924568378839213,
      "loss": 2.1575,
      "step": 16481
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.5479938387870789,
      "learning_rate": 0.0005924559263354335,
      "loss": 2.429,
      "step": 16482
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6044781804084778,
      "learning_rate": 0.0005924550147325727,
      "loss": 2.4207,
      "step": 16483
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6246796250343323,
      "learning_rate": 0.000592454103075339,
      "loss": 2.2903,
      "step": 16484
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6371452212333679,
      "learning_rate": 0.0005924531913637325,
      "loss": 2.2925,
      "step": 16485
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6095631122589111,
      "learning_rate": 0.0005924522795977535,
      "loss": 2.31,
      "step": 16486
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6933519840240479,
      "learning_rate": 0.0005924513677774021,
      "loss": 2.3071,
      "step": 16487
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.5448145270347595,
      "learning_rate": 0.0005924504559026785,
      "loss": 2.3948,
      "step": 16488
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6129376292228699,
      "learning_rate": 0.0005924495439735828,
      "loss": 2.2629,
      "step": 16489
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6634050607681274,
      "learning_rate": 0.0005924486319901152,
      "loss": 2.4164,
      "step": 16490
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6134651303291321,
      "learning_rate": 0.0005924477199522758,
      "loss": 2.3417,
      "step": 16491
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6165927648544312,
      "learning_rate": 0.000592446807860065,
      "loss": 2.3678,
      "step": 16492
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6433903574943542,
      "learning_rate": 0.0005924458957134827,
      "loss": 2.4823,
      "step": 16493
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6025428175926208,
      "learning_rate": 0.0005924449835125293,
      "loss": 2.463,
      "step": 16494
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6092498302459717,
      "learning_rate": 0.0005924440712572047,
      "loss": 2.3291,
      "step": 16495
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6367486715316772,
      "learning_rate": 0.0005924431589475093,
      "loss": 2.3743,
      "step": 16496
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.617818295955658,
      "learning_rate": 0.0005924422465834432,
      "loss": 2.3566,
      "step": 16497
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.7016054391860962,
      "learning_rate": 0.0005924413341650066,
      "loss": 2.3554,
      "step": 16498
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6764572858810425,
      "learning_rate": 0.0005924404216921997,
      "loss": 2.5424,
      "step": 16499
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6191476583480835,
      "learning_rate": 0.0005924395091650225,
      "loss": 2.3726,
      "step": 16500
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.7409144043922424,
      "learning_rate": 0.0005924385965834754,
      "loss": 2.2733,
      "step": 16501
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6057373881340027,
      "learning_rate": 0.0005924376839475583,
      "loss": 2.4639,
      "step": 16502
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6201624870300293,
      "learning_rate": 0.0005924367712572715,
      "loss": 2.3329,
      "step": 16503
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6425915956497192,
      "learning_rate": 0.0005924358585126153,
      "loss": 2.426,
      "step": 16504
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6560635566711426,
      "learning_rate": 0.0005924349457135897,
      "loss": 1.9723,
      "step": 16505
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6149619817733765,
      "learning_rate": 0.0005924340328601949,
      "loss": 2.5075,
      "step": 16506
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6954256892204285,
      "learning_rate": 0.0005924331199524311,
      "loss": 2.2669,
      "step": 16507
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6735919713973999,
      "learning_rate": 0.0005924322069902984,
      "loss": 2.509,
      "step": 16508
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.5772172212600708,
      "learning_rate": 0.0005924312939737972,
      "loss": 2.0206,
      "step": 16509
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.657065212726593,
      "learning_rate": 0.0005924303809029274,
      "loss": 2.4306,
      "step": 16510
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.7913072109222412,
      "learning_rate": 0.0005924294677776893,
      "loss": 2.4334,
      "step": 16511
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.5877152681350708,
      "learning_rate": 0.0005924285545980829,
      "loss": 2.7828,
      "step": 16512
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6451622843742371,
      "learning_rate": 0.0005924276413641087,
      "loss": 2.5161,
      "step": 16513
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6791207194328308,
      "learning_rate": 0.0005924267280757666,
      "loss": 2.4077,
      "step": 16514
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.5828705430030823,
      "learning_rate": 0.0005924258147330569,
      "loss": 2.3333,
      "step": 16515
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6047652363777161,
      "learning_rate": 0.0005924249013359797,
      "loss": 2.3083,
      "step": 16516
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6054340600967407,
      "learning_rate": 0.0005924239878845352,
      "loss": 2.2198,
      "step": 16517
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6466259956359863,
      "learning_rate": 0.0005924230743787235,
      "loss": 2.4178,
      "step": 16518
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6528105735778809,
      "learning_rate": 0.0005924221608185449,
      "loss": 2.356,
      "step": 16519
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6712236404418945,
      "learning_rate": 0.0005924212472039994,
      "loss": 2.427,
      "step": 16520
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.5973581671714783,
      "learning_rate": 0.0005924203335350874,
      "loss": 2.4559,
      "step": 16521
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.5967690944671631,
      "learning_rate": 0.0005924194198118089,
      "loss": 2.379,
      "step": 16522
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6361263990402222,
      "learning_rate": 0.000592418506034164,
      "loss": 2.5328,
      "step": 16523
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.5814335942268372,
      "learning_rate": 0.0005924175922021531,
      "loss": 2.3932,
      "step": 16524
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6224238276481628,
      "learning_rate": 0.0005924166783157763,
      "loss": 2.4187,
      "step": 16525
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6170127987861633,
      "learning_rate": 0.0005924157643750336,
      "loss": 2.3321,
      "step": 16526
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6343439221382141,
      "learning_rate": 0.0005924148503799253,
      "loss": 2.4251,
      "step": 16527
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6040249466896057,
      "learning_rate": 0.0005924139363304516,
      "loss": 2.2597,
      "step": 16528
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6367881894111633,
      "learning_rate": 0.0005924130222266127,
      "loss": 2.4148,
      "step": 16529
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6054040789604187,
      "learning_rate": 0.0005924121080684086,
      "loss": 2.2303,
      "step": 16530
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6119741201400757,
      "learning_rate": 0.0005924111938558395,
      "loss": 2.5621,
      "step": 16531
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6592182517051697,
      "learning_rate": 0.0005924102795889058,
      "loss": 2.5677,
      "step": 16532
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.7549651861190796,
      "learning_rate": 0.0005924093652676074,
      "loss": 2.4849,
      "step": 16533
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6509197354316711,
      "learning_rate": 0.0005924084508919447,
      "loss": 2.3868,
      "step": 16534
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.5972083210945129,
      "learning_rate": 0.0005924075364619176,
      "loss": 2.54,
      "step": 16535
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.7842536568641663,
      "learning_rate": 0.0005924066219775265,
      "loss": 2.3109,
      "step": 16536
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6584132313728333,
      "learning_rate": 0.0005924057074387715,
      "loss": 2.2717,
      "step": 16537
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.602098822593689,
      "learning_rate": 0.0005924047928456528,
      "loss": 2.3498,
      "step": 16538
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.564732551574707,
      "learning_rate": 0.0005924038781981704,
      "loss": 2.6328,
      "step": 16539
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.7506321668624878,
      "learning_rate": 0.0005924029634963246,
      "loss": 2.2985,
      "step": 16540
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6863647699356079,
      "learning_rate": 0.0005924020487401157,
      "loss": 2.3975,
      "step": 16541
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.5909731984138489,
      "learning_rate": 0.0005924011339295437,
      "loss": 2.519,
      "step": 16542
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.5706395506858826,
      "learning_rate": 0.0005924002190646087,
      "loss": 2.4253,
      "step": 16543
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.7378079295158386,
      "learning_rate": 0.000592399304145311,
      "loss": 2.4203,
      "step": 16544
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6689403653144836,
      "learning_rate": 0.0005923983891716509,
      "loss": 2.2412,
      "step": 16545
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.5860252976417542,
      "learning_rate": 0.0005923974741436284,
      "loss": 2.2484,
      "step": 16546
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6671127080917358,
      "learning_rate": 0.0005923965590612435,
      "loss": 2.4402,
      "step": 16547
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.7342131733894348,
      "learning_rate": 0.0005923956439244967,
      "loss": 2.277,
      "step": 16548
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.5537638664245605,
      "learning_rate": 0.0005923947287333881,
      "loss": 2.3787,
      "step": 16549
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6546609997749329,
      "learning_rate": 0.0005923938134879176,
      "loss": 2.3411,
      "step": 16550
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6787679195404053,
      "learning_rate": 0.0005923928981880857,
      "loss": 2.4036,
      "step": 16551
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.653863251209259,
      "learning_rate": 0.0005923919828338925,
      "loss": 2.2582,
      "step": 16552
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6443300843238831,
      "learning_rate": 0.000592391067425338,
      "loss": 2.5387,
      "step": 16553
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6444995999336243,
      "learning_rate": 0.0005923901519624226,
      "loss": 2.3424,
      "step": 16554
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6137462258338928,
      "learning_rate": 0.0005923892364451461,
      "loss": 2.2911,
      "step": 16555
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.626589298248291,
      "learning_rate": 0.0005923883208735092,
      "loss": 2.4778,
      "step": 16556
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.5786643028259277,
      "learning_rate": 0.0005923874052475118,
      "loss": 2.4943,
      "step": 16557
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6080809831619263,
      "learning_rate": 0.0005923864895671538,
      "loss": 2.3731,
      "step": 16558
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.5599908232688904,
      "learning_rate": 0.0005923855738324357,
      "loss": 2.3006,
      "step": 16559
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.5899995565414429,
      "learning_rate": 0.0005923846580433578,
      "loss": 2.1142,
      "step": 16560
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6074804067611694,
      "learning_rate": 0.0005923837421999199,
      "loss": 2.3359,
      "step": 16561
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.5903860926628113,
      "learning_rate": 0.0005923828263021224,
      "loss": 2.235,
      "step": 16562
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.5565071702003479,
      "learning_rate": 0.0005923819103499654,
      "loss": 2.2868,
      "step": 16563
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6530274152755737,
      "learning_rate": 0.0005923809943434491,
      "loss": 2.6119,
      "step": 16564
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6436467170715332,
      "learning_rate": 0.0005923800782825737,
      "loss": 2.6016,
      "step": 16565
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6365898251533508,
      "learning_rate": 0.0005923791621673393,
      "loss": 2.4501,
      "step": 16566
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.5908880233764648,
      "learning_rate": 0.000592378245997746,
      "loss": 2.4272,
      "step": 16567
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.5761273503303528,
      "learning_rate": 0.0005923773297737942,
      "loss": 2.1427,
      "step": 16568
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6019858121871948,
      "learning_rate": 0.0005923764134954838,
      "loss": 2.3682,
      "step": 16569
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6490857005119324,
      "learning_rate": 0.0005923754971628152,
      "loss": 2.4544,
      "step": 16570
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6033302545547485,
      "learning_rate": 0.0005923745807757885,
      "loss": 2.3619,
      "step": 16571
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6452798247337341,
      "learning_rate": 0.0005923736643344037,
      "loss": 2.2974,
      "step": 16572
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.5970985293388367,
      "learning_rate": 0.0005923727478386612,
      "loss": 2.2222,
      "step": 16573
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6386809945106506,
      "learning_rate": 0.0005923718312885612,
      "loss": 2.5497,
      "step": 16574
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.5838280320167542,
      "learning_rate": 0.0005923709146841036,
      "loss": 2.6694,
      "step": 16575
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.7113288640975952,
      "learning_rate": 0.0005923699980252887,
      "loss": 2.3935,
      "step": 16576
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6297157406806946,
      "learning_rate": 0.0005923690813121168,
      "loss": 2.1971,
      "step": 16577
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6542760133743286,
      "learning_rate": 0.000592368164544588,
      "loss": 2.3577,
      "step": 16578
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.5663485527038574,
      "learning_rate": 0.0005923672477227024,
      "loss": 2.4802,
      "step": 16579
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.7143349647521973,
      "learning_rate": 0.0005923663308464602,
      "loss": 2.3358,
      "step": 16580
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.5760495662689209,
      "learning_rate": 0.0005923654139158616,
      "loss": 2.2344,
      "step": 16581
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.5599726438522339,
      "learning_rate": 0.0005923644969309068,
      "loss": 2.2376,
      "step": 16582
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.5494681596755981,
      "learning_rate": 0.0005923635798915958,
      "loss": 2.4078,
      "step": 16583
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6497138738632202,
      "learning_rate": 0.000592362662797929,
      "loss": 2.4841,
      "step": 16584
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.628394603729248,
      "learning_rate": 0.0005923617456499063,
      "loss": 2.3291,
      "step": 16585
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.670143187046051,
      "learning_rate": 0.0005923608284475281,
      "loss": 2.2716,
      "step": 16586
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6328485012054443,
      "learning_rate": 0.0005923599111907946,
      "loss": 2.4508,
      "step": 16587
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6074950098991394,
      "learning_rate": 0.0005923589938797058,
      "loss": 2.3028,
      "step": 16588
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6082724332809448,
      "learning_rate": 0.000592358076514262,
      "loss": 2.1698,
      "step": 16589
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.7810431718826294,
      "learning_rate": 0.0005923571590944634,
      "loss": 2.4449,
      "step": 16590
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.7640938758850098,
      "learning_rate": 0.0005923562416203099,
      "loss": 2.5697,
      "step": 16591
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6590625643730164,
      "learning_rate": 0.0005923553240918019,
      "loss": 2.2731,
      "step": 16592
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.7546003460884094,
      "learning_rate": 0.0005923544065089395,
      "loss": 2.3119,
      "step": 16593
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.7393882274627686,
      "learning_rate": 0.0005923534888717229,
      "loss": 2.3275,
      "step": 16594
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.5914192795753479,
      "learning_rate": 0.0005923525711801523,
      "loss": 2.5808,
      "step": 16595
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.0399478673934937,
      "learning_rate": 0.0005923516534342278,
      "loss": 2.5286,
      "step": 16596
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6048893928527832,
      "learning_rate": 0.0005923507356339498,
      "loss": 2.4893,
      "step": 16597
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6223859190940857,
      "learning_rate": 0.000592349817779318,
      "loss": 2.4626,
      "step": 16598
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6010947227478027,
      "learning_rate": 0.0005923488998703331,
      "loss": 2.41,
      "step": 16599
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6158782243728638,
      "learning_rate": 0.0005923479819069949,
      "loss": 2.2463,
      "step": 16600
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.5680322647094727,
      "learning_rate": 0.0005923470638893037,
      "loss": 2.3352,
      "step": 16601
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6653839945793152,
      "learning_rate": 0.0005923461458172596,
      "loss": 2.1223,
      "step": 16602
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6083952784538269,
      "learning_rate": 0.0005923452276908629,
      "loss": 2.3656,
      "step": 16603
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6844989657402039,
      "learning_rate": 0.0005923443095101137,
      "loss": 2.1963,
      "step": 16604
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.5983195304870605,
      "learning_rate": 0.0005923433912750122,
      "loss": 2.445,
      "step": 16605
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.607855498790741,
      "learning_rate": 0.0005923424729855585,
      "loss": 2.2863,
      "step": 16606
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.7028598785400391,
      "learning_rate": 0.0005923415546417529,
      "loss": 2.2154,
      "step": 16607
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6457166075706482,
      "learning_rate": 0.0005923406362435954,
      "loss": 2.452,
      "step": 16608
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.5946012735366821,
      "learning_rate": 0.0005923397177910863,
      "loss": 2.1289,
      "step": 16609
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6038007140159607,
      "learning_rate": 0.0005923387992842257,
      "loss": 2.3787,
      "step": 16610
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.5651270151138306,
      "learning_rate": 0.0005923378807230138,
      "loss": 2.5229,
      "step": 16611
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.587209165096283,
      "learning_rate": 0.0005923369621074507,
      "loss": 2.4908,
      "step": 16612
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.9380958676338196,
      "learning_rate": 0.0005923360434375367,
      "loss": 2.4678,
      "step": 16613
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6589874625205994,
      "learning_rate": 0.000592335124713272,
      "loss": 2.4191,
      "step": 16614
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.5915796160697937,
      "learning_rate": 0.0005923342059346566,
      "loss": 2.5315,
      "step": 16615
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6464729905128479,
      "learning_rate": 0.0005923332871016907,
      "loss": 2.4918,
      "step": 16616
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6694993376731873,
      "learning_rate": 0.0005923323682143747,
      "loss": 2.4527,
      "step": 16617
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.5927040576934814,
      "learning_rate": 0.0005923314492727084,
      "loss": 2.4068,
      "step": 16618
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6546197533607483,
      "learning_rate": 0.0005923305302766922,
      "loss": 2.2407,
      "step": 16619
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.5984021425247192,
      "learning_rate": 0.0005923296112263263,
      "loss": 2.4463,
      "step": 16620
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.5932420492172241,
      "learning_rate": 0.0005923286921216108,
      "loss": 2.4444,
      "step": 16621
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.7131137847900391,
      "learning_rate": 0.0005923277729625459,
      "loss": 2.4399,
      "step": 16622
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.7329021096229553,
      "learning_rate": 0.0005923268537491317,
      "loss": 2.4046,
      "step": 16623
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.5960516333580017,
      "learning_rate": 0.0005923259344813685,
      "loss": 2.4066,
      "step": 16624
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6941406726837158,
      "learning_rate": 0.0005923250151592563,
      "loss": 2.4965,
      "step": 16625
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.5977239012718201,
      "learning_rate": 0.0005923240957827953,
      "loss": 2.6868,
      "step": 16626
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6228362321853638,
      "learning_rate": 0.0005923231763519859,
      "loss": 2.3043,
      "step": 16627
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.5968128442764282,
      "learning_rate": 0.000592322256866828,
      "loss": 2.3053,
      "step": 16628
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6334611177444458,
      "learning_rate": 0.0005923213373273219,
      "loss": 2.294,
      "step": 16629
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.7136015295982361,
      "learning_rate": 0.0005923204177334677,
      "loss": 2.2307,
      "step": 16630
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6213379502296448,
      "learning_rate": 0.0005923194980852656,
      "loss": 2.5041,
      "step": 16631
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6548280715942383,
      "learning_rate": 0.000592318578382716,
      "loss": 2.1593,
      "step": 16632
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6501486897468567,
      "learning_rate": 0.0005923176586258187,
      "loss": 2.3524,
      "step": 16633
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.690288245677948,
      "learning_rate": 0.000592316738814574,
      "loss": 2.3038,
      "step": 16634
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6573688387870789,
      "learning_rate": 0.0005923158189489821,
      "loss": 2.3837,
      "step": 16635
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6328122019767761,
      "learning_rate": 0.0005923148990290432,
      "loss": 2.415,
      "step": 16636
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.5987637042999268,
      "learning_rate": 0.0005923139790547575,
      "loss": 2.3404,
      "step": 16637
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.591884970664978,
      "learning_rate": 0.000592313059026125,
      "loss": 2.3873,
      "step": 16638
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6127307415008545,
      "learning_rate": 0.0005923121389431462,
      "loss": 2.392,
      "step": 16639
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6398486495018005,
      "learning_rate": 0.0005923112188058208,
      "loss": 2.6356,
      "step": 16640
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6093161702156067,
      "learning_rate": 0.0005923102986141493,
      "loss": 2.1517,
      "step": 16641
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.5703513622283936,
      "learning_rate": 0.0005923093783681318,
      "loss": 2.5359,
      "step": 16642
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.5991871953010559,
      "learning_rate": 0.0005923084580677686,
      "loss": 2.4052,
      "step": 16643
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6876322031021118,
      "learning_rate": 0.0005923075377130596,
      "loss": 2.3839,
      "step": 16644
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6042889952659607,
      "learning_rate": 0.0005923066173040051,
      "loss": 2.1313,
      "step": 16645
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6262187361717224,
      "learning_rate": 0.0005923056968406054,
      "loss": 2.4693,
      "step": 16646
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6514617800712585,
      "learning_rate": 0.0005923047763228604,
      "loss": 2.4246,
      "step": 16647
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6629899740219116,
      "learning_rate": 0.0005923038557507706,
      "loss": 2.34,
      "step": 16648
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6032540202140808,
      "learning_rate": 0.0005923029351243359,
      "loss": 2.3827,
      "step": 16649
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.7017053365707397,
      "learning_rate": 0.0005923020144435565,
      "loss": 2.3684,
      "step": 16650
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.5747143030166626,
      "learning_rate": 0.0005923010937084328,
      "loss": 2.4413,
      "step": 16651
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.568787157535553,
      "learning_rate": 0.0005923001729189647,
      "loss": 2.4894,
      "step": 16652
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6159371137619019,
      "learning_rate": 0.0005922992520751524,
      "loss": 2.6795,
      "step": 16653
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.5829298496246338,
      "learning_rate": 0.0005922983311769963,
      "loss": 2.4902,
      "step": 16654
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.607458233833313,
      "learning_rate": 0.0005922974102244964,
      "loss": 2.282,
      "step": 16655
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.601915717124939,
      "learning_rate": 0.0005922964892176528,
      "loss": 2.1652,
      "step": 16656
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.5747219920158386,
      "learning_rate": 0.0005922955681564658,
      "loss": 2.3164,
      "step": 16657
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.5505082011222839,
      "learning_rate": 0.0005922946470409356,
      "loss": 2.1482,
      "step": 16658
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6364285349845886,
      "learning_rate": 0.0005922937258710623,
      "loss": 2.3861,
      "step": 16659
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6215318441390991,
      "learning_rate": 0.000592292804646846,
      "loss": 2.3455,
      "step": 16660
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6023939847946167,
      "learning_rate": 0.0005922918833682869,
      "loss": 2.254,
      "step": 16661
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6093040704727173,
      "learning_rate": 0.0005922909620353855,
      "loss": 2.4057,
      "step": 16662
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.5761168599128723,
      "learning_rate": 0.0005922900406481414,
      "loss": 2.231,
      "step": 16663
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.5550161600112915,
      "learning_rate": 0.0005922891192065552,
      "loss": 2.3141,
      "step": 16664
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6332278847694397,
      "learning_rate": 0.000592288197710627,
      "loss": 2.4007,
      "step": 16665
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.619217574596405,
      "learning_rate": 0.0005922872761603568,
      "loss": 2.2482,
      "step": 16666
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6212513446807861,
      "learning_rate": 0.0005922863545557449,
      "loss": 2.4288,
      "step": 16667
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6221107840538025,
      "learning_rate": 0.0005922854328967915,
      "loss": 2.4694,
      "step": 16668
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.7146843671798706,
      "learning_rate": 0.0005922845111834966,
      "loss": 2.1441,
      "step": 16669
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.5514606833457947,
      "learning_rate": 0.0005922835894158605,
      "loss": 2.262,
      "step": 16670
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6197280883789062,
      "learning_rate": 0.0005922826675938836,
      "loss": 2.4976,
      "step": 16671
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.5800872445106506,
      "learning_rate": 0.0005922817457175656,
      "loss": 2.5227,
      "step": 16672
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6192359924316406,
      "learning_rate": 0.000592280823786907,
      "loss": 2.5052,
      "step": 16673
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.573458194732666,
      "learning_rate": 0.0005922799018019078,
      "loss": 2.4567,
      "step": 16674
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6337711215019226,
      "learning_rate": 0.0005922789797625683,
      "loss": 2.409,
      "step": 16675
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6379788517951965,
      "learning_rate": 0.0005922780576688886,
      "loss": 2.329,
      "step": 16676
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6069305539131165,
      "learning_rate": 0.0005922771355208689,
      "loss": 2.4681,
      "step": 16677
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.5946251749992371,
      "learning_rate": 0.0005922762133185093,
      "loss": 2.3388,
      "step": 16678
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.587871253490448,
      "learning_rate": 0.0005922752910618101,
      "loss": 2.232,
      "step": 16679
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.5888282656669617,
      "learning_rate": 0.0005922743687507715,
      "loss": 2.4359,
      "step": 16680
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6978843808174133,
      "learning_rate": 0.0005922734463853934,
      "loss": 2.2243,
      "step": 16681
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.5789117217063904,
      "learning_rate": 0.0005922725239656762,
      "loss": 2.3037,
      "step": 16682
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6271824836730957,
      "learning_rate": 0.0005922716014916202,
      "loss": 2.5092,
      "step": 16683
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6575348377227783,
      "learning_rate": 0.0005922706789632252,
      "loss": 2.4679,
      "step": 16684
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6188545823097229,
      "learning_rate": 0.0005922697563804916,
      "loss": 2.5696,
      "step": 16685
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6297759413719177,
      "learning_rate": 0.0005922688337434195,
      "loss": 2.4366,
      "step": 16686
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.7689386010169983,
      "learning_rate": 0.0005922679110520092,
      "loss": 2.2957,
      "step": 16687
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6440496444702148,
      "learning_rate": 0.0005922669883062608,
      "loss": 2.5035,
      "step": 16688
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.5980592370033264,
      "learning_rate": 0.0005922660655061744,
      "loss": 2.4132,
      "step": 16689
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.7719046473503113,
      "learning_rate": 0.0005922651426517501,
      "loss": 2.377,
      "step": 16690
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.5553433299064636,
      "learning_rate": 0.0005922642197429884,
      "loss": 2.5475,
      "step": 16691
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.657930314540863,
      "learning_rate": 0.0005922632967798891,
      "loss": 2.3098,
      "step": 16692
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6578158140182495,
      "learning_rate": 0.0005922623737624526,
      "loss": 2.4147,
      "step": 16693
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6080171465873718,
      "learning_rate": 0.0005922614506906791,
      "loss": 2.3428,
      "step": 16694
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.612684965133667,
      "learning_rate": 0.0005922605275645685,
      "loss": 2.4287,
      "step": 16695
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6614896655082703,
      "learning_rate": 0.0005922596043841212,
      "loss": 2.3268,
      "step": 16696
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6509006023406982,
      "learning_rate": 0.0005922586811493374,
      "loss": 2.4893,
      "step": 16697
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6458833813667297,
      "learning_rate": 0.0005922577578602172,
      "loss": 2.3865,
      "step": 16698
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.5968186259269714,
      "learning_rate": 0.0005922568345167607,
      "loss": 2.4113,
      "step": 16699
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6572378873825073,
      "learning_rate": 0.0005922559111189681,
      "loss": 2.2962,
      "step": 16700
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6576716899871826,
      "learning_rate": 0.0005922549876668398,
      "loss": 2.5545,
      "step": 16701
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.5922136902809143,
      "learning_rate": 0.0005922540641603755,
      "loss": 2.1971,
      "step": 16702
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6670195460319519,
      "learning_rate": 0.0005922531405995759,
      "loss": 2.6082,
      "step": 16703
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6788457036018372,
      "learning_rate": 0.0005922522169844408,
      "loss": 2.3537,
      "step": 16704
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6245529055595398,
      "learning_rate": 0.0005922512933149705,
      "loss": 2.3212,
      "step": 16705
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6296159029006958,
      "learning_rate": 0.0005922503695911652,
      "loss": 2.4104,
      "step": 16706
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.5844163298606873,
      "learning_rate": 0.000592249445813025,
      "loss": 2.3923,
      "step": 16707
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6530099511146545,
      "learning_rate": 0.0005922485219805502,
      "loss": 2.3935,
      "step": 16708
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6209015846252441,
      "learning_rate": 0.0005922475980937408,
      "loss": 2.2195,
      "step": 16709
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6129791140556335,
      "learning_rate": 0.0005922466741525971,
      "loss": 2.1679,
      "step": 16710
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6701849102973938,
      "learning_rate": 0.0005922457501571192,
      "loss": 2.2915,
      "step": 16711
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6669245958328247,
      "learning_rate": 0.0005922448261073073,
      "loss": 2.4718,
      "step": 16712
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.7414994239807129,
      "learning_rate": 0.0005922439020031615,
      "loss": 2.1432,
      "step": 16713
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.5967690348625183,
      "learning_rate": 0.0005922429778446822,
      "loss": 2.366,
      "step": 16714
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.5834382772445679,
      "learning_rate": 0.0005922420536318693,
      "loss": 2.4531,
      "step": 16715
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6770193576812744,
      "learning_rate": 0.0005922411293647231,
      "loss": 2.2583,
      "step": 16716
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.5801775455474854,
      "learning_rate": 0.0005922402050432438,
      "loss": 2.4513,
      "step": 16717
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6203391551971436,
      "learning_rate": 0.0005922392806674315,
      "loss": 2.3104,
      "step": 16718
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.5470962524414062,
      "learning_rate": 0.0005922383562372864,
      "loss": 2.4174,
      "step": 16719
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.5844630002975464,
      "learning_rate": 0.0005922374317528087,
      "loss": 2.5768,
      "step": 16720
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6367896795272827,
      "learning_rate": 0.0005922365072139986,
      "loss": 2.427,
      "step": 16721
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6023921966552734,
      "learning_rate": 0.0005922355826208561,
      "loss": 2.5956,
      "step": 16722
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6820335984230042,
      "learning_rate": 0.0005922346579733815,
      "loss": 2.6053,
      "step": 16723
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.5876402258872986,
      "learning_rate": 0.000592233733271575,
      "loss": 2.5235,
      "step": 16724
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.5637238025665283,
      "learning_rate": 0.0005922328085154367,
      "loss": 2.4282,
      "step": 16725
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6942053437232971,
      "learning_rate": 0.0005922318837049668,
      "loss": 2.4378,
      "step": 16726
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.5912957191467285,
      "learning_rate": 0.0005922309588401656,
      "loss": 2.3868,
      "step": 16727
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.61959308385849,
      "learning_rate": 0.000592230033921033,
      "loss": 2.374,
      "step": 16728
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6279906034469604,
      "learning_rate": 0.0005922291089475693,
      "loss": 2.3029,
      "step": 16729
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6016985177993774,
      "learning_rate": 0.0005922281839197748,
      "loss": 2.3852,
      "step": 16730
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6338446736335754,
      "learning_rate": 0.0005922272588376495,
      "loss": 2.3385,
      "step": 16731
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.5396109223365784,
      "learning_rate": 0.0005922263337011937,
      "loss": 2.4251,
      "step": 16732
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6240090727806091,
      "learning_rate": 0.0005922254085104075,
      "loss": 2.5666,
      "step": 16733
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.5700209736824036,
      "learning_rate": 0.000592224483265291,
      "loss": 2.4279,
      "step": 16734
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6388768553733826,
      "learning_rate": 0.0005922235579658445,
      "loss": 2.1386,
      "step": 16735
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.615753173828125,
      "learning_rate": 0.0005922226326120682,
      "loss": 2.3733,
      "step": 16736
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6599332094192505,
      "learning_rate": 0.000592221707203962,
      "loss": 2.4622,
      "step": 16737
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.5906931161880493,
      "learning_rate": 0.0005922207817415264,
      "loss": 2.4934,
      "step": 16738
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6141283512115479,
      "learning_rate": 0.0005922198562247615,
      "loss": 2.6385,
      "step": 16739
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.582272469997406,
      "learning_rate": 0.0005922189306536673,
      "loss": 2.3067,
      "step": 16740
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6134709119796753,
      "learning_rate": 0.0005922180050282442,
      "loss": 2.3172,
      "step": 16741
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6332697868347168,
      "learning_rate": 0.0005922170793484922,
      "loss": 2.4105,
      "step": 16742
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6430773138999939,
      "learning_rate": 0.0005922161536144116,
      "loss": 2.4582,
      "step": 16743
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.654964804649353,
      "learning_rate": 0.0005922152278260023,
      "loss": 2.333,
      "step": 16744
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6779365539550781,
      "learning_rate": 0.0005922143019832649,
      "loss": 2.4141,
      "step": 16745
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6914663910865784,
      "learning_rate": 0.0005922133760861993,
      "loss": 2.3746,
      "step": 16746
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6176292896270752,
      "learning_rate": 0.0005922124501348056,
      "loss": 2.41,
      "step": 16747
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6272285580635071,
      "learning_rate": 0.000592211524129084,
      "loss": 2.463,
      "step": 16748
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6603943109512329,
      "learning_rate": 0.0005922105980690351,
      "loss": 2.486,
      "step": 16749
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6125796437263489,
      "learning_rate": 0.0005922096719546585,
      "loss": 2.627,
      "step": 16750
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6212265491485596,
      "learning_rate": 0.0005922087457859546,
      "loss": 2.2663,
      "step": 16751
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.0124417543411255,
      "learning_rate": 0.0005922078195629236,
      "loss": 2.381,
      "step": 16752
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6164771914482117,
      "learning_rate": 0.0005922068932855657,
      "loss": 2.3407,
      "step": 16753
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6230460405349731,
      "learning_rate": 0.0005922059669538811,
      "loss": 2.3933,
      "step": 16754
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.63276606798172,
      "learning_rate": 0.0005922050405678696,
      "loss": 2.2483,
      "step": 16755
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.621663510799408,
      "learning_rate": 0.0005922041141275319,
      "loss": 2.5224,
      "step": 16756
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.596230149269104,
      "learning_rate": 0.0005922031876328679,
      "loss": 2.4251,
      "step": 16757
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6466110348701477,
      "learning_rate": 0.0005922022610838778,
      "loss": 2.5907,
      "step": 16758
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6229799389839172,
      "learning_rate": 0.0005922013344805618,
      "loss": 2.5382,
      "step": 16759
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.5662292242050171,
      "learning_rate": 0.0005922004078229202,
      "loss": 2.2072,
      "step": 16760
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6162566542625427,
      "learning_rate": 0.0005921994811109528,
      "loss": 2.3352,
      "step": 16761
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.5688733458518982,
      "learning_rate": 0.0005921985543446601,
      "loss": 2.389,
      "step": 16762
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.614853024482727,
      "learning_rate": 0.0005921976275240421,
      "loss": 2.5879,
      "step": 16763
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6849094033241272,
      "learning_rate": 0.0005921967006490991,
      "loss": 2.2522,
      "step": 16764
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.5994472503662109,
      "learning_rate": 0.0005921957737198313,
      "loss": 2.2676,
      "step": 16765
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.7350615859031677,
      "learning_rate": 0.0005921948467362387,
      "loss": 2.4018,
      "step": 16766
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6418910622596741,
      "learning_rate": 0.0005921939196983215,
      "loss": 2.1829,
      "step": 16767
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6250776052474976,
      "learning_rate": 0.00059219299260608,
      "loss": 2.4248,
      "step": 16768
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6111946105957031,
      "learning_rate": 0.0005921920654595143,
      "loss": 2.3616,
      "step": 16769
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6338273286819458,
      "learning_rate": 0.0005921911382586247,
      "loss": 2.5379,
      "step": 16770
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6594429612159729,
      "learning_rate": 0.0005921902110034111,
      "loss": 2.4793,
      "step": 16771
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.658795952796936,
      "learning_rate": 0.0005921892836938737,
      "loss": 2.2454,
      "step": 16772
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6051355600357056,
      "learning_rate": 0.0005921883563300131,
      "loss": 2.4446,
      "step": 16773
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.5805842876434326,
      "learning_rate": 0.000592187428911829,
      "loss": 2.3962,
      "step": 16774
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6580848693847656,
      "learning_rate": 0.0005921865014393219,
      "loss": 2.5211,
      "step": 16775
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6178141832351685,
      "learning_rate": 0.0005921855739124916,
      "loss": 2.6299,
      "step": 16776
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.593873143196106,
      "learning_rate": 0.0005921846463313386,
      "loss": 2.3957,
      "step": 16777
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6768808364868164,
      "learning_rate": 0.000592183718695863,
      "loss": 2.2735,
      "step": 16778
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6890826225280762,
      "learning_rate": 0.0005921827910060648,
      "loss": 2.3974,
      "step": 16779
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6380640268325806,
      "learning_rate": 0.0005921818632619444,
      "loss": 2.3603,
      "step": 16780
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.627427875995636,
      "learning_rate": 0.0005921809354635018,
      "loss": 2.5536,
      "step": 16781
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.641999363899231,
      "learning_rate": 0.0005921800076107372,
      "loss": 2.5482,
      "step": 16782
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.5819442272186279,
      "learning_rate": 0.000592179079703651,
      "loss": 2.471,
      "step": 16783
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.7208496332168579,
      "learning_rate": 0.0005921781517422431,
      "loss": 2.1713,
      "step": 16784
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.5838353633880615,
      "learning_rate": 0.0005921772237265138,
      "loss": 2.2183,
      "step": 16785
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6691707372665405,
      "learning_rate": 0.0005921762956564632,
      "loss": 2.1733,
      "step": 16786
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6389482021331787,
      "learning_rate": 0.0005921753675320916,
      "loss": 2.4125,
      "step": 16787
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6571905016899109,
      "learning_rate": 0.0005921744393533989,
      "loss": 2.3344,
      "step": 16788
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6279585361480713,
      "learning_rate": 0.0005921735111203856,
      "loss": 2.2939,
      "step": 16789
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6038618683815002,
      "learning_rate": 0.0005921725828330518,
      "loss": 2.2791,
      "step": 16790
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6124432682991028,
      "learning_rate": 0.0005921716544913974,
      "loss": 2.4056,
      "step": 16791
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.5722084641456604,
      "learning_rate": 0.0005921707260954229,
      "loss": 2.4302,
      "step": 16792
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.5691798329353333,
      "learning_rate": 0.0005921697976451283,
      "loss": 2.1413,
      "step": 16793
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.5092031359672546,
      "learning_rate": 0.0005921688691405138,
      "loss": 2.1088,
      "step": 16794
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.5513834357261658,
      "learning_rate": 0.0005921679405815796,
      "loss": 2.3175,
      "step": 16795
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.601439893245697,
      "learning_rate": 0.000592167011968326,
      "loss": 2.3523,
      "step": 16796
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6051845550537109,
      "learning_rate": 0.0005921660833007529,
      "loss": 2.4543,
      "step": 16797
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.5981064438819885,
      "learning_rate": 0.0005921651545788606,
      "loss": 2.5585,
      "step": 16798
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6419835090637207,
      "learning_rate": 0.0005921642258026493,
      "loss": 2.3632,
      "step": 16799
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6939189434051514,
      "learning_rate": 0.0005921632969721192,
      "loss": 2.4391,
      "step": 16800
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6342283487319946,
      "learning_rate": 0.0005921623680872704,
      "loss": 2.3764,
      "step": 16801
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.7139143347740173,
      "learning_rate": 0.0005921614391481032,
      "loss": 2.4035,
      "step": 16802
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6076734066009521,
      "learning_rate": 0.0005921605101546175,
      "loss": 2.3277,
      "step": 16803
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6013746857643127,
      "learning_rate": 0.0005921595811068137,
      "loss": 2.3904,
      "step": 16804
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6616379618644714,
      "learning_rate": 0.000592158652004692,
      "loss": 2.4483,
      "step": 16805
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6127489805221558,
      "learning_rate": 0.0005921577228482524,
      "loss": 2.3988,
      "step": 16806
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6676231026649475,
      "learning_rate": 0.0005921567936374953,
      "loss": 2.5961,
      "step": 16807
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.7165262699127197,
      "learning_rate": 0.0005921558643724206,
      "loss": 2.5131,
      "step": 16808
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6316068172454834,
      "learning_rate": 0.0005921549350530286,
      "loss": 2.3173,
      "step": 16809
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6665476560592651,
      "learning_rate": 0.0005921540056793195,
      "loss": 2.3822,
      "step": 16810
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6183268427848816,
      "learning_rate": 0.0005921530762512936,
      "loss": 2.3293,
      "step": 16811
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.7215952277183533,
      "learning_rate": 0.0005921521467689508,
      "loss": 2.5453,
      "step": 16812
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.5853464603424072,
      "learning_rate": 0.0005921512172322915,
      "loss": 2.3579,
      "step": 16813
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.5733778476715088,
      "learning_rate": 0.0005921502876413157,
      "loss": 2.2683,
      "step": 16814
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.7014251947402954,
      "learning_rate": 0.0005921493579960236,
      "loss": 2.5577,
      "step": 16815
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.7297843098640442,
      "learning_rate": 0.0005921484282964154,
      "loss": 2.4282,
      "step": 16816
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6191238760948181,
      "learning_rate": 0.0005921474985424914,
      "loss": 2.4925,
      "step": 16817
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6578313708305359,
      "learning_rate": 0.0005921465687342516,
      "loss": 2.4809,
      "step": 16818
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6921558380126953,
      "learning_rate": 0.0005921456388716963,
      "loss": 2.3404,
      "step": 16819
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6531760096549988,
      "learning_rate": 0.0005921447089548255,
      "loss": 2.368,
      "step": 16820
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6503062844276428,
      "learning_rate": 0.0005921437789836395,
      "loss": 2.5368,
      "step": 16821
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.7442650198936462,
      "learning_rate": 0.0005921428489581385,
      "loss": 2.382,
      "step": 16822
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6530988216400146,
      "learning_rate": 0.0005921419188783226,
      "loss": 2.5806,
      "step": 16823
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.9435631632804871,
      "learning_rate": 0.0005921409887441921,
      "loss": 2.5802,
      "step": 16824
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6341094374656677,
      "learning_rate": 0.0005921400585557471,
      "loss": 2.369,
      "step": 16825
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.8006277680397034,
      "learning_rate": 0.0005921391283129876,
      "loss": 2.4355,
      "step": 16826
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.794765830039978,
      "learning_rate": 0.000592138198015914,
      "loss": 2.6391,
      "step": 16827
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.7389241456985474,
      "learning_rate": 0.0005921372676645264,
      "loss": 2.5082,
      "step": 16828
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.711999237537384,
      "learning_rate": 0.0005921363372588249,
      "loss": 2.6281,
      "step": 16829
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.5860468149185181,
      "learning_rate": 0.0005921354067988098,
      "loss": 2.3689,
      "step": 16830
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.7031198143959045,
      "learning_rate": 0.0005921344762844812,
      "loss": 2.1474,
      "step": 16831
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6374766230583191,
      "learning_rate": 0.0005921335457158393,
      "loss": 2.4225,
      "step": 16832
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.5761542916297913,
      "learning_rate": 0.0005921326150928842,
      "loss": 2.4232,
      "step": 16833
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.7928118705749512,
      "learning_rate": 0.0005921316844156162,
      "loss": 2.7954,
      "step": 16834
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6740519404411316,
      "learning_rate": 0.0005921307536840354,
      "loss": 2.4197,
      "step": 16835
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6360670924186707,
      "learning_rate": 0.0005921298228981419,
      "loss": 2.4128,
      "step": 16836
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.9860410690307617,
      "learning_rate": 0.000592128892057936,
      "loss": 2.1009,
      "step": 16837
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6645480394363403,
      "learning_rate": 0.000592127961163418,
      "loss": 2.4279,
      "step": 16838
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6458685398101807,
      "learning_rate": 0.0005921270302145877,
      "loss": 2.4466,
      "step": 16839
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.7567803859710693,
      "learning_rate": 0.0005921260992114455,
      "loss": 2.4309,
      "step": 16840
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.7574859261512756,
      "learning_rate": 0.0005921251681539915,
      "loss": 2.3499,
      "step": 16841
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.5809177756309509,
      "learning_rate": 0.000592124237042226,
      "loss": 2.0266,
      "step": 16842
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.7661013007164001,
      "learning_rate": 0.000592123305876149,
      "loss": 2.3799,
      "step": 16843
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.8178643584251404,
      "learning_rate": 0.0005921223746557609,
      "loss": 2.347,
      "step": 16844
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.573820173740387,
      "learning_rate": 0.0005921214433810616,
      "loss": 2.2888,
      "step": 16845
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6076598763465881,
      "learning_rate": 0.0005921205120520515,
      "loss": 2.4694,
      "step": 16846
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.7770372629165649,
      "learning_rate": 0.0005921195806687307,
      "loss": 2.4411,
      "step": 16847
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6447718143463135,
      "learning_rate": 0.0005921186492310992,
      "loss": 2.3604,
      "step": 16848
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6110460758209229,
      "learning_rate": 0.0005921177177391575,
      "loss": 2.311,
      "step": 16849
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6610086560249329,
      "learning_rate": 0.0005921167861929055,
      "loss": 2.387,
      "step": 16850
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.7400078773498535,
      "learning_rate": 0.0005921158545923436,
      "loss": 2.3388,
      "step": 16851
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.5981714129447937,
      "learning_rate": 0.0005921149229374717,
      "loss": 2.3001,
      "step": 16852
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6149550080299377,
      "learning_rate": 0.0005921139912282901,
      "loss": 2.1156,
      "step": 16853
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6786439418792725,
      "learning_rate": 0.0005921130594647991,
      "loss": 2.3629,
      "step": 16854
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.5423035621643066,
      "learning_rate": 0.0005921121276469988,
      "loss": 2.3525,
      "step": 16855
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.5933107733726501,
      "learning_rate": 0.0005921111957748892,
      "loss": 2.4019,
      "step": 16856
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6798427700996399,
      "learning_rate": 0.0005921102638484707,
      "loss": 2.3284,
      "step": 16857
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6123900413513184,
      "learning_rate": 0.0005921093318677435,
      "loss": 2.4264,
      "step": 16858
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6843467354774475,
      "learning_rate": 0.0005921083998327076,
      "loss": 2.3081,
      "step": 16859
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6471331715583801,
      "learning_rate": 0.0005921074677433631,
      "loss": 2.4966,
      "step": 16860
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6581140756607056,
      "learning_rate": 0.0005921065355997103,
      "loss": 2.1174,
      "step": 16861
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6315082907676697,
      "learning_rate": 0.0005921056034017496,
      "loss": 2.3826,
      "step": 16862
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.5826219320297241,
      "learning_rate": 0.0005921046711494808,
      "loss": 2.2749,
      "step": 16863
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6354193687438965,
      "learning_rate": 0.0005921037388429042,
      "loss": 2.6659,
      "step": 16864
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.5879863500595093,
      "learning_rate": 0.0005921028064820201,
      "loss": 2.4536,
      "step": 16865
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6192161440849304,
      "learning_rate": 0.0005921018740668285,
      "loss": 2.3181,
      "step": 16866
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.8225277662277222,
      "learning_rate": 0.0005921009415973298,
      "loss": 2.4808,
      "step": 16867
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.8000513315200806,
      "learning_rate": 0.0005921000090735239,
      "loss": 2.2681,
      "step": 16868
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6627589464187622,
      "learning_rate": 0.000592099076495411,
      "loss": 2.4423,
      "step": 16869
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6766940951347351,
      "learning_rate": 0.0005920981438629915,
      "loss": 2.358,
      "step": 16870
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6030606031417847,
      "learning_rate": 0.0005920972111762654,
      "loss": 2.1547,
      "step": 16871
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6848204135894775,
      "learning_rate": 0.0005920962784352328,
      "loss": 2.4953,
      "step": 16872
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.7738335728645325,
      "learning_rate": 0.0005920953456398941,
      "loss": 2.458,
      "step": 16873
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6421241760253906,
      "learning_rate": 0.0005920944127902494,
      "loss": 2.5339,
      "step": 16874
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6146195530891418,
      "learning_rate": 0.0005920934798862989,
      "loss": 2.3602,
      "step": 16875
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6741869449615479,
      "learning_rate": 0.0005920925469280426,
      "loss": 2.5521,
      "step": 16876
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.7347344756126404,
      "learning_rate": 0.0005920916139154808,
      "loss": 2.3417,
      "step": 16877
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.5859801173210144,
      "learning_rate": 0.0005920906808486137,
      "loss": 2.2796,
      "step": 16878
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6536536812782288,
      "learning_rate": 0.0005920897477274413,
      "loss": 2.4883,
      "step": 16879
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.5979702472686768,
      "learning_rate": 0.0005920888145519639,
      "loss": 2.3209,
      "step": 16880
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.8145719766616821,
      "learning_rate": 0.0005920878813221818,
      "loss": 2.4617,
      "step": 16881
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.5796117186546326,
      "learning_rate": 0.000592086948038095,
      "loss": 2.3901,
      "step": 16882
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.7513440847396851,
      "learning_rate": 0.0005920860146997038,
      "loss": 2.3815,
      "step": 16883
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6305709481239319,
      "learning_rate": 0.0005920850813070082,
      "loss": 2.3392,
      "step": 16884
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.697161853313446,
      "learning_rate": 0.0005920841478600085,
      "loss": 2.268,
      "step": 16885
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.5853132605552673,
      "learning_rate": 0.0005920832143587048,
      "loss": 2.3498,
      "step": 16886
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.73017418384552,
      "learning_rate": 0.0005920822808030973,
      "loss": 2.5042,
      "step": 16887
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6009006500244141,
      "learning_rate": 0.0005920813471931864,
      "loss": 2.2512,
      "step": 16888
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.5708277821540833,
      "learning_rate": 0.0005920804135289718,
      "loss": 2.0959,
      "step": 16889
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6867687106132507,
      "learning_rate": 0.0005920794798104541,
      "loss": 2.4676,
      "step": 16890
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6162070631980896,
      "learning_rate": 0.0005920785460376333,
      "loss": 2.2301,
      "step": 16891
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6413276195526123,
      "learning_rate": 0.0005920776122105095,
      "loss": 2.2953,
      "step": 16892
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6487038135528564,
      "learning_rate": 0.000592076678329083,
      "loss": 2.4195,
      "step": 16893
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.8004935383796692,
      "learning_rate": 0.0005920757443933539,
      "loss": 2.4398,
      "step": 16894
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.5611138343811035,
      "learning_rate": 0.0005920748104033224,
      "loss": 2.5035,
      "step": 16895
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6417557597160339,
      "learning_rate": 0.0005920738763589888,
      "loss": 2.6021,
      "step": 16896
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6512656807899475,
      "learning_rate": 0.0005920729422603531,
      "loss": 2.3658,
      "step": 16897
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6414502859115601,
      "learning_rate": 0.0005920720081074155,
      "loss": 2.4196,
      "step": 16898
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6125813722610474,
      "learning_rate": 0.0005920710739001761,
      "loss": 2.3431,
      "step": 16899
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6115381121635437,
      "learning_rate": 0.0005920701396386353,
      "loss": 2.4463,
      "step": 16900
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.5911998152732849,
      "learning_rate": 0.0005920692053227931,
      "loss": 2.2395,
      "step": 16901
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.568709671497345,
      "learning_rate": 0.0005920682709526498,
      "loss": 2.1599,
      "step": 16902
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.5742197036743164,
      "learning_rate": 0.0005920673365282054,
      "loss": 2.2271,
      "step": 16903
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6651715636253357,
      "learning_rate": 0.0005920664020494602,
      "loss": 2.1351,
      "step": 16904
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6015282869338989,
      "learning_rate": 0.0005920654675164144,
      "loss": 2.4373,
      "step": 16905
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.5682627558708191,
      "learning_rate": 0.0005920645329290681,
      "loss": 2.3818,
      "step": 16906
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.5620541572570801,
      "learning_rate": 0.0005920635982874213,
      "loss": 2.4544,
      "step": 16907
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6775439381599426,
      "learning_rate": 0.0005920626635914746,
      "loss": 2.3541,
      "step": 16908
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6719586849212646,
      "learning_rate": 0.0005920617288412279,
      "loss": 2.296,
      "step": 16909
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.617590606212616,
      "learning_rate": 0.0005920607940366814,
      "loss": 2.5456,
      "step": 16910
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6516596078872681,
      "learning_rate": 0.0005920598591778351,
      "loss": 2.3004,
      "step": 16911
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.8003160357475281,
      "learning_rate": 0.0005920589242646896,
      "loss": 2.4654,
      "step": 16912
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6504318118095398,
      "learning_rate": 0.0005920579892972448,
      "loss": 2.3126,
      "step": 16913
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.8834227919578552,
      "learning_rate": 0.0005920570542755008,
      "loss": 2.4353,
      "step": 16914
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6080533862113953,
      "learning_rate": 0.0005920561191994579,
      "loss": 2.6654,
      "step": 16915
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6393657326698303,
      "learning_rate": 0.0005920551840691164,
      "loss": 2.4602,
      "step": 16916
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6343574523925781,
      "learning_rate": 0.0005920542488844761,
      "loss": 2.3343,
      "step": 16917
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6361282467842102,
      "learning_rate": 0.0005920533136455375,
      "loss": 2.4292,
      "step": 16918
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6543464064598083,
      "learning_rate": 0.0005920523783523008,
      "loss": 2.2602,
      "step": 16919
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6323987245559692,
      "learning_rate": 0.0005920514430047659,
      "loss": 2.2131,
      "step": 16920
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.600773274898529,
      "learning_rate": 0.0005920505076029332,
      "loss": 2.4529,
      "step": 16921
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6256496906280518,
      "learning_rate": 0.0005920495721468028,
      "loss": 2.3994,
      "step": 16922
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6023755669593811,
      "learning_rate": 0.0005920486366363748,
      "loss": 2.2248,
      "step": 16923
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.7004652619361877,
      "learning_rate": 0.0005920477010716495,
      "loss": 2.5249,
      "step": 16924
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.63374263048172,
      "learning_rate": 0.000592046765452627,
      "loss": 2.3755,
      "step": 16925
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6949065923690796,
      "learning_rate": 0.0005920458297793075,
      "loss": 2.3202,
      "step": 16926
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.598501443862915,
      "learning_rate": 0.0005920448940516911,
      "loss": 2.2651,
      "step": 16927
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.7415949702262878,
      "learning_rate": 0.0005920439582697781,
      "loss": 2.3495,
      "step": 16928
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6309731006622314,
      "learning_rate": 0.0005920430224335687,
      "loss": 2.5647,
      "step": 16929
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6628048419952393,
      "learning_rate": 0.0005920420865430628,
      "loss": 2.4148,
      "step": 16930
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.7497132420539856,
      "learning_rate": 0.0005920411505982609,
      "loss": 2.3219,
      "step": 16931
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.7350860834121704,
      "learning_rate": 0.000592040214599163,
      "loss": 2.4236,
      "step": 16932
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6201733350753784,
      "learning_rate": 0.0005920392785457694,
      "loss": 2.4505,
      "step": 16933
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6512597799301147,
      "learning_rate": 0.0005920383424380801,
      "loss": 2.3347,
      "step": 16934
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.7114850878715515,
      "learning_rate": 0.0005920374062760954,
      "loss": 2.3114,
      "step": 16935
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6285358667373657,
      "learning_rate": 0.0005920364700598153,
      "loss": 2.19,
      "step": 16936
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6177322864532471,
      "learning_rate": 0.0005920355337892402,
      "loss": 2.3243,
      "step": 16937
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.5892811417579651,
      "learning_rate": 0.0005920345974643703,
      "loss": 2.4769,
      "step": 16938
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.5980849862098694,
      "learning_rate": 0.0005920336610852055,
      "loss": 2.1319,
      "step": 16939
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.5972169637680054,
      "learning_rate": 0.0005920327246517462,
      "loss": 2.2789,
      "step": 16940
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.5914101600646973,
      "learning_rate": 0.0005920317881639924,
      "loss": 2.2497,
      "step": 16941
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.5646548867225647,
      "learning_rate": 0.0005920308516219445,
      "loss": 2.5421,
      "step": 16942
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6650839447975159,
      "learning_rate": 0.0005920299150256025,
      "loss": 2.3995,
      "step": 16943
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.677707314491272,
      "learning_rate": 0.0005920289783749666,
      "loss": 2.4671,
      "step": 16944
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.5831491947174072,
      "learning_rate": 0.0005920280416700371,
      "loss": 2.0582,
      "step": 16945
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6291358470916748,
      "learning_rate": 0.0005920271049108139,
      "loss": 2.5181,
      "step": 16946
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6736151576042175,
      "learning_rate": 0.0005920261680972975,
      "loss": 2.2798,
      "step": 16947
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.5549107193946838,
      "learning_rate": 0.000592025231229488,
      "loss": 2.5221,
      "step": 16948
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6349652409553528,
      "learning_rate": 0.0005920242943073853,
      "loss": 2.2798,
      "step": 16949
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6249046921730042,
      "learning_rate": 0.0005920233573309899,
      "loss": 2.3084,
      "step": 16950
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.5897096991539001,
      "learning_rate": 0.0005920224203003018,
      "loss": 2.3284,
      "step": 16951
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.5851407051086426,
      "learning_rate": 0.0005920214832153211,
      "loss": 2.276,
      "step": 16952
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6279082894325256,
      "learning_rate": 0.0005920205460760483,
      "loss": 2.4244,
      "step": 16953
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.5851338505744934,
      "learning_rate": 0.0005920196088824832,
      "loss": 2.5208,
      "step": 16954
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.608979344367981,
      "learning_rate": 0.0005920186716346263,
      "loss": 2.3776,
      "step": 16955
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6451758742332458,
      "learning_rate": 0.0005920177343324776,
      "loss": 2.3223,
      "step": 16956
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6810699701309204,
      "learning_rate": 0.0005920167969760371,
      "loss": 2.4185,
      "step": 16957
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.5857864022254944,
      "learning_rate": 0.0005920158595653053,
      "loss": 2.5636,
      "step": 16958
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6177387833595276,
      "learning_rate": 0.0005920149221002823,
      "loss": 2.3579,
      "step": 16959
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.5814101696014404,
      "learning_rate": 0.0005920139845809681,
      "loss": 2.3706,
      "step": 16960
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6328529715538025,
      "learning_rate": 0.000592013047007363,
      "loss": 2.6569,
      "step": 16961
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.5803563594818115,
      "learning_rate": 0.0005920121093794672,
      "loss": 2.2402,
      "step": 16962
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.5602127313613892,
      "learning_rate": 0.0005920111716972809,
      "loss": 2.395,
      "step": 16963
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.5950925350189209,
      "learning_rate": 0.0005920102339608042,
      "loss": 2.1635,
      "step": 16964
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6725519895553589,
      "learning_rate": 0.0005920092961700372,
      "loss": 2.5636,
      "step": 16965
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.5997202396392822,
      "learning_rate": 0.0005920083583249801,
      "loss": 2.366,
      "step": 16966
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.5944783687591553,
      "learning_rate": 0.0005920074204256333,
      "loss": 2.4178,
      "step": 16967
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.561565101146698,
      "learning_rate": 0.0005920064824719967,
      "loss": 2.217,
      "step": 16968
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.5991957783699036,
      "learning_rate": 0.0005920055444640707,
      "loss": 2.562,
      "step": 16969
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6880638003349304,
      "learning_rate": 0.0005920046064018552,
      "loss": 2.2822,
      "step": 16970
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.598856508731842,
      "learning_rate": 0.0005920036682853506,
      "loss": 2.4994,
      "step": 16971
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.705556333065033,
      "learning_rate": 0.000592002730114557,
      "loss": 2.286,
      "step": 16972
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6336967945098877,
      "learning_rate": 0.0005920017918894747,
      "loss": 2.4401,
      "step": 16973
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.632142961025238,
      "learning_rate": 0.0005920008536101036,
      "loss": 2.1638,
      "step": 16974
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.5991120934486389,
      "learning_rate": 0.0005919999152764441,
      "loss": 2.5462,
      "step": 16975
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6103670597076416,
      "learning_rate": 0.0005919989768884962,
      "loss": 2.4727,
      "step": 16976
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6275961995124817,
      "learning_rate": 0.0005919980384462603,
      "loss": 2.2551,
      "step": 16977
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6196670532226562,
      "learning_rate": 0.0005919970999497363,
      "loss": 2.4316,
      "step": 16978
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6374498605728149,
      "learning_rate": 0.0005919961613989247,
      "loss": 2.2874,
      "step": 16979
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6787136793136597,
      "learning_rate": 0.0005919952227938255,
      "loss": 2.409,
      "step": 16980
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6507186889648438,
      "learning_rate": 0.0005919942841344387,
      "loss": 2.3619,
      "step": 16981
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6102304458618164,
      "learning_rate": 0.0005919933454207649,
      "loss": 2.2218,
      "step": 16982
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6674664616584778,
      "learning_rate": 0.0005919924066528038,
      "loss": 2.223,
      "step": 16983
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6919228434562683,
      "learning_rate": 0.0005919914678305559,
      "loss": 2.5347,
      "step": 16984
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6151365041732788,
      "learning_rate": 0.0005919905289540213,
      "loss": 2.2956,
      "step": 16985
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6297808885574341,
      "learning_rate": 0.0005919895900232001,
      "loss": 2.4355,
      "step": 16986
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.662463366985321,
      "learning_rate": 0.0005919886510380926,
      "loss": 2.1975,
      "step": 16987
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6197097301483154,
      "learning_rate": 0.0005919877119986987,
      "loss": 2.3883,
      "step": 16988
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.5943291187286377,
      "learning_rate": 0.000591986772905019,
      "loss": 2.4716,
      "step": 16989
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6046866178512573,
      "learning_rate": 0.0005919858337570534,
      "loss": 2.3682,
      "step": 16990
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6739745736122131,
      "learning_rate": 0.000591984894554802,
      "loss": 2.3255,
      "step": 16991
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.7085188627243042,
      "learning_rate": 0.0005919839552982652,
      "loss": 2.4716,
      "step": 16992
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6994019746780396,
      "learning_rate": 0.0005919830159874431,
      "loss": 2.2431,
      "step": 16993
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6051629781723022,
      "learning_rate": 0.0005919820766223358,
      "loss": 2.2106,
      "step": 16994
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6326409578323364,
      "learning_rate": 0.0005919811372029434,
      "loss": 2.6235,
      "step": 16995
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.7140017151832581,
      "learning_rate": 0.0005919801977292663,
      "loss": 2.3477,
      "step": 16996
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.639773428440094,
      "learning_rate": 0.0005919792582013046,
      "loss": 2.2647,
      "step": 16997
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.5722242593765259,
      "learning_rate": 0.0005919783186190585,
      "loss": 2.5046,
      "step": 16998
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6270626187324524,
      "learning_rate": 0.000591977378982528,
      "loss": 2.4937,
      "step": 16999
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.604337751865387,
      "learning_rate": 0.0005919764392917134,
      "loss": 2.3333,
      "step": 17000
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6666783094406128,
      "learning_rate": 0.0005919754995466149,
      "loss": 2.6567,
      "step": 17001
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6164002418518066,
      "learning_rate": 0.0005919745597472327,
      "loss": 2.4812,
      "step": 17002
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.5851845145225525,
      "learning_rate": 0.0005919736198935669,
      "loss": 2.5904,
      "step": 17003
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.5899354219436646,
      "learning_rate": 0.0005919726799856176,
      "loss": 2.2978,
      "step": 17004
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.5873079299926758,
      "learning_rate": 0.0005919717400233852,
      "loss": 2.4895,
      "step": 17005
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6454975605010986,
      "learning_rate": 0.0005919708000068696,
      "loss": 2.1832,
      "step": 17006
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6312726140022278,
      "learning_rate": 0.0005919698599360712,
      "loss": 2.2753,
      "step": 17007
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6334764957427979,
      "learning_rate": 0.0005919689198109901,
      "loss": 2.426,
      "step": 17008
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6820006370544434,
      "learning_rate": 0.0005919679796316265,
      "loss": 2.3304,
      "step": 17009
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6119639277458191,
      "learning_rate": 0.0005919670393979803,
      "loss": 2.3771,
      "step": 17010
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.607045590877533,
      "learning_rate": 0.0005919660991100522,
      "loss": 2.2958,
      "step": 17011
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6501018404960632,
      "learning_rate": 0.000591965158767842,
      "loss": 2.4067,
      "step": 17012
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6830505132675171,
      "learning_rate": 0.0005919642183713499,
      "loss": 2.4,
      "step": 17013
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.7042180299758911,
      "learning_rate": 0.0005919632779205762,
      "loss": 2.2108,
      "step": 17014
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6431519389152527,
      "learning_rate": 0.000591962337415521,
      "loss": 2.6268,
      "step": 17015
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6400499939918518,
      "learning_rate": 0.0005919613968561845,
      "loss": 2.4093,
      "step": 17016
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.5833885669708252,
      "learning_rate": 0.0005919604562425669,
      "loss": 2.2993,
      "step": 17017
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6308485269546509,
      "learning_rate": 0.0005919595155746683,
      "loss": 2.3635,
      "step": 17018
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6264289021492004,
      "learning_rate": 0.0005919585748524889,
      "loss": 2.2527,
      "step": 17019
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6249431371688843,
      "learning_rate": 0.0005919576340760289,
      "loss": 2.3559,
      "step": 17020
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.5559868216514587,
      "learning_rate": 0.0005919566932452885,
      "loss": 2.3785,
      "step": 17021
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.59688800573349,
      "learning_rate": 0.0005919557523602679,
      "loss": 2.3404,
      "step": 17022
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6598301529884338,
      "learning_rate": 0.0005919548114209671,
      "loss": 2.2835,
      "step": 17023
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.5738358497619629,
      "learning_rate": 0.0005919538704273864,
      "loss": 2.3291,
      "step": 17024
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6508033275604248,
      "learning_rate": 0.000591952929379526,
      "loss": 2.5255,
      "step": 17025
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6267104744911194,
      "learning_rate": 0.000591951988277386,
      "loss": 2.385,
      "step": 17026
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6069943904876709,
      "learning_rate": 0.0005919510471209666,
      "loss": 2.2707,
      "step": 17027
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6327119469642639,
      "learning_rate": 0.000591950105910268,
      "loss": 2.2071,
      "step": 17028
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.5632972121238708,
      "learning_rate": 0.0005919491646452904,
      "loss": 2.4356,
      "step": 17029
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.7012183666229248,
      "learning_rate": 0.000591948223326034,
      "loss": 2.3343,
      "step": 17030
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.5890501737594604,
      "learning_rate": 0.0005919472819524988,
      "loss": 2.5397,
      "step": 17031
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6823121905326843,
      "learning_rate": 0.0005919463405246852,
      "loss": 2.3286,
      "step": 17032
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6561938524246216,
      "learning_rate": 0.0005919453990425932,
      "loss": 2.4791,
      "step": 17033
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.5998415350914001,
      "learning_rate": 0.000591944457506223,
      "loss": 2.119,
      "step": 17034
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6367914080619812,
      "learning_rate": 0.0005919435159155748,
      "loss": 2.6203,
      "step": 17035
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6386182308197021,
      "learning_rate": 0.0005919425742706489,
      "loss": 2.3359,
      "step": 17036
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6163610219955444,
      "learning_rate": 0.0005919416325714452,
      "loss": 2.3378,
      "step": 17037
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.7071796655654907,
      "learning_rate": 0.0005919406908179642,
      "loss": 2.4064,
      "step": 17038
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6077617406845093,
      "learning_rate": 0.0005919397490102058,
      "loss": 2.1136,
      "step": 17039
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6044275164604187,
      "learning_rate": 0.0005919388071481704,
      "loss": 2.7382,
      "step": 17040
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6432967185974121,
      "learning_rate": 0.000591937865231858,
      "loss": 2.4517,
      "step": 17041
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6409494280815125,
      "learning_rate": 0.0005919369232612689,
      "loss": 2.4425,
      "step": 17042
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6533574461936951,
      "learning_rate": 0.0005919359812364033,
      "loss": 2.5346,
      "step": 17043
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6122610569000244,
      "learning_rate": 0.000591935039157261,
      "loss": 2.4325,
      "step": 17044
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.5708215832710266,
      "learning_rate": 0.0005919340970238426,
      "loss": 2.5119,
      "step": 17045
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.585267186164856,
      "learning_rate": 0.0005919331548361482,
      "loss": 2.1723,
      "step": 17046
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.5605276823043823,
      "learning_rate": 0.0005919322125941779,
      "loss": 2.2091,
      "step": 17047
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6026204228401184,
      "learning_rate": 0.0005919312702979319,
      "loss": 2.4267,
      "step": 17048
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.5831796526908875,
      "learning_rate": 0.0005919303279474103,
      "loss": 2.4283,
      "step": 17049
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.7190932631492615,
      "learning_rate": 0.0005919293855426134,
      "loss": 2.648,
      "step": 17050
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6188379526138306,
      "learning_rate": 0.0005919284430835412,
      "loss": 2.4508,
      "step": 17051
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.5621160864830017,
      "learning_rate": 0.0005919275005701941,
      "loss": 2.2738,
      "step": 17052
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6312599182128906,
      "learning_rate": 0.0005919265580025722,
      "loss": 2.4284,
      "step": 17053
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6157751083374023,
      "learning_rate": 0.0005919256153806757,
      "loss": 2.1308,
      "step": 17054
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6292615532875061,
      "learning_rate": 0.0005919246727045046,
      "loss": 2.4378,
      "step": 17055
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6250542402267456,
      "learning_rate": 0.0005919237299740591,
      "loss": 2.2358,
      "step": 17056
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6104691028594971,
      "learning_rate": 0.0005919227871893396,
      "loss": 2.2404,
      "step": 17057
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.5796547532081604,
      "learning_rate": 0.0005919218443503462,
      "loss": 2.4534,
      "step": 17058
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.5860272645950317,
      "learning_rate": 0.0005919209014570789,
      "loss": 2.4728,
      "step": 17059
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6085783243179321,
      "learning_rate": 0.000591919958509538,
      "loss": 2.6068,
      "step": 17060
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6252356171607971,
      "learning_rate": 0.0005919190155077238,
      "loss": 2.39,
      "step": 17061
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.63954097032547,
      "learning_rate": 0.0005919180724516361,
      "loss": 2.2491,
      "step": 17062
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.7050032019615173,
      "learning_rate": 0.0005919171293412756,
      "loss": 2.3443,
      "step": 17063
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6217424869537354,
      "learning_rate": 0.000591916186176642,
      "loss": 2.2255,
      "step": 17064
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6007429957389832,
      "learning_rate": 0.0005919152429577358,
      "loss": 2.3634,
      "step": 17065
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6659173965454102,
      "learning_rate": 0.0005919142996845569,
      "loss": 2.1699,
      "step": 17066
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.5854385495185852,
      "learning_rate": 0.0005919133563571057,
      "loss": 2.552,
      "step": 17067
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6092057228088379,
      "learning_rate": 0.0005919124129753823,
      "loss": 2.1173,
      "step": 17068
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6133769750595093,
      "learning_rate": 0.000591911469539387,
      "loss": 2.3426,
      "step": 17069
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.5929353833198547,
      "learning_rate": 0.0005919105260491196,
      "loss": 2.3977,
      "step": 17070
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6082125902175903,
      "learning_rate": 0.0005919095825045806,
      "loss": 2.3927,
      "step": 17071
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6036715507507324,
      "learning_rate": 0.0005919086389057702,
      "loss": 2.2671,
      "step": 17072
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.584346354007721,
      "learning_rate": 0.0005919076952526885,
      "loss": 2.2699,
      "step": 17073
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.7065370678901672,
      "learning_rate": 0.0005919067515453355,
      "loss": 2.4116,
      "step": 17074
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6294009685516357,
      "learning_rate": 0.0005919058077837115,
      "loss": 2.3013,
      "step": 17075
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6011586785316467,
      "learning_rate": 0.0005919048639678169,
      "loss": 2.3045,
      "step": 17076
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.5969843864440918,
      "learning_rate": 0.0005919039200976516,
      "loss": 2.5078,
      "step": 17077
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.634639322757721,
      "learning_rate": 0.0005919029761732158,
      "loss": 2.1629,
      "step": 17078
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6203798651695251,
      "learning_rate": 0.0005919020321945097,
      "loss": 2.5382,
      "step": 17079
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6412186026573181,
      "learning_rate": 0.0005919010881615336,
      "loss": 2.6185,
      "step": 17080
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6090942621231079,
      "learning_rate": 0.0005919001440742875,
      "loss": 2.3574,
      "step": 17081
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6975072622299194,
      "learning_rate": 0.0005918991999327716,
      "loss": 2.3003,
      "step": 17082
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6025858521461487,
      "learning_rate": 0.0005918982557369863,
      "loss": 2.2508,
      "step": 17083
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6476991772651672,
      "learning_rate": 0.0005918973114869315,
      "loss": 2.199,
      "step": 17084
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.7617314457893372,
      "learning_rate": 0.0005918963671826074,
      "loss": 2.1405,
      "step": 17085
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6394479870796204,
      "learning_rate": 0.0005918954228240144,
      "loss": 2.23,
      "step": 17086
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6786594986915588,
      "learning_rate": 0.0005918944784111524,
      "loss": 2.4326,
      "step": 17087
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.7936109304428101,
      "learning_rate": 0.0005918935339440218,
      "loss": 2.5265,
      "step": 17088
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6302357316017151,
      "learning_rate": 0.0005918925894226227,
      "loss": 2.4309,
      "step": 17089
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.5934892296791077,
      "learning_rate": 0.0005918916448469551,
      "loss": 2.3885,
      "step": 17090
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6320841312408447,
      "learning_rate": 0.0005918907002170195,
      "loss": 2.4393,
      "step": 17091
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.7983628511428833,
      "learning_rate": 0.0005918897555328158,
      "loss": 2.4544,
      "step": 17092
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6197075843811035,
      "learning_rate": 0.0005918888107943445,
      "loss": 2.1549,
      "step": 17093
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6207609176635742,
      "learning_rate": 0.0005918878660016053,
      "loss": 2.3177,
      "step": 17094
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.7242636680603027,
      "learning_rate": 0.0005918869211545988,
      "loss": 2.4661,
      "step": 17095
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6456491351127625,
      "learning_rate": 0.0005918859762533249,
      "loss": 2.3136,
      "step": 17096
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.632986843585968,
      "learning_rate": 0.0005918850312977839,
      "loss": 2.2754,
      "step": 17097
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.8364865183830261,
      "learning_rate": 0.0005918840862879761,
      "loss": 2.2262,
      "step": 17098
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6369173526763916,
      "learning_rate": 0.0005918831412239015,
      "loss": 2.1954,
      "step": 17099
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.5584569573402405,
      "learning_rate": 0.0005918821961055601,
      "loss": 2.4423,
      "step": 17100
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6135119199752808,
      "learning_rate": 0.0005918812509329525,
      "loss": 2.3903,
      "step": 17101
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6429154276847839,
      "learning_rate": 0.0005918803057060786,
      "loss": 2.4423,
      "step": 17102
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.579273521900177,
      "learning_rate": 0.0005918793604249387,
      "loss": 2.3342,
      "step": 17103
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.604569137096405,
      "learning_rate": 0.000591878415089533,
      "loss": 2.2918,
      "step": 17104
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6009767055511475,
      "learning_rate": 0.0005918774696998613,
      "loss": 2.3305,
      "step": 17105
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.7001545429229736,
      "learning_rate": 0.0005918765242559243,
      "loss": 2.3751,
      "step": 17106
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.5964400172233582,
      "learning_rate": 0.0005918755787577219,
      "loss": 2.315,
      "step": 17107
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6063077449798584,
      "learning_rate": 0.0005918746332052542,
      "loss": 2.3429,
      "step": 17108
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.8204475045204163,
      "learning_rate": 0.0005918736875985217,
      "loss": 2.5327,
      "step": 17109
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.7116391658782959,
      "learning_rate": 0.0005918727419375242,
      "loss": 2.5351,
      "step": 17110
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.8049463033676147,
      "learning_rate": 0.0005918717962222621,
      "loss": 2.2268,
      "step": 17111
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6052977442741394,
      "learning_rate": 0.0005918708504527356,
      "loss": 2.1704,
      "step": 17112
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.5564152598381042,
      "learning_rate": 0.0005918699046289446,
      "loss": 2.327,
      "step": 17113
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.7026044130325317,
      "learning_rate": 0.0005918689587508897,
      "loss": 2.2665,
      "step": 17114
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6027398109436035,
      "learning_rate": 0.0005918680128185708,
      "loss": 2.3588,
      "step": 17115
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6581486463546753,
      "learning_rate": 0.000591867066831988,
      "loss": 2.6045,
      "step": 17116
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.5529767870903015,
      "learning_rate": 0.0005918661207911418,
      "loss": 2.1671,
      "step": 17117
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6408519148826599,
      "learning_rate": 0.000591865174696032,
      "loss": 2.1161,
      "step": 17118
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6087643504142761,
      "learning_rate": 0.000591864228546659,
      "loss": 2.3942,
      "step": 17119
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6762489080429077,
      "learning_rate": 0.0005918632823430231,
      "loss": 2.3119,
      "step": 17120
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6400740742683411,
      "learning_rate": 0.0005918623360851241,
      "loss": 2.5562,
      "step": 17121
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6888191103935242,
      "learning_rate": 0.0005918613897729624,
      "loss": 2.5247,
      "step": 17122
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.7737705111503601,
      "learning_rate": 0.0005918604434065382,
      "loss": 2.5882,
      "step": 17123
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.5899479985237122,
      "learning_rate": 0.0005918594969858517,
      "loss": 2.4749,
      "step": 17124
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.8551108241081238,
      "learning_rate": 0.0005918585505109029,
      "loss": 2.2557,
      "step": 17125
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.7188910841941833,
      "learning_rate": 0.0005918576039816921,
      "loss": 2.3639,
      "step": 17126
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6142218112945557,
      "learning_rate": 0.0005918566573982195,
      "loss": 2.2478,
      "step": 17127
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.8037009835243225,
      "learning_rate": 0.0005918557107604853,
      "loss": 2.3235,
      "step": 17128
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.7111163139343262,
      "learning_rate": 0.0005918547640684895,
      "loss": 2.5528,
      "step": 17129
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6031774878501892,
      "learning_rate": 0.0005918538173222324,
      "loss": 2.351,
      "step": 17130
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.7489068508148193,
      "learning_rate": 0.0005918528705217143,
      "loss": 2.3687,
      "step": 17131
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6638113260269165,
      "learning_rate": 0.0005918519236669352,
      "loss": 2.3081,
      "step": 17132
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6955138444900513,
      "learning_rate": 0.0005918509767578953,
      "loss": 2.5066,
      "step": 17133
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6493297219276428,
      "learning_rate": 0.0005918500297945946,
      "loss": 2.337,
      "step": 17134
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.5910294651985168,
      "learning_rate": 0.0005918490827770337,
      "loss": 2.2939,
      "step": 17135
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6473994851112366,
      "learning_rate": 0.0005918481357052125,
      "loss": 2.4994,
      "step": 17136
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6292394399642944,
      "learning_rate": 0.0005918471885791312,
      "loss": 2.3097,
      "step": 17137
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6203160285949707,
      "learning_rate": 0.00059184624139879,
      "loss": 2.4342,
      "step": 17138
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.7280811071395874,
      "learning_rate": 0.0005918452941641891,
      "loss": 2.1522,
      "step": 17139
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6897701025009155,
      "learning_rate": 0.0005918443468753285,
      "loss": 2.4412,
      "step": 17140
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6945146918296814,
      "learning_rate": 0.0005918433995322087,
      "loss": 2.4795,
      "step": 17141
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.5646635293960571,
      "learning_rate": 0.0005918424521348297,
      "loss": 2.4557,
      "step": 17142
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.7762191295623779,
      "learning_rate": 0.0005918415046831916,
      "loss": 2.4069,
      "step": 17143
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.5908223986625671,
      "learning_rate": 0.0005918405571772947,
      "loss": 2.3101,
      "step": 17144
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6299812197685242,
      "learning_rate": 0.000591839609617139,
      "loss": 2.5273,
      "step": 17145
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6660536527633667,
      "learning_rate": 0.0005918386620027251,
      "loss": 2.3719,
      "step": 17146
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.602179229259491,
      "learning_rate": 0.0005918377143340526,
      "loss": 2.538,
      "step": 17147
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6381971836090088,
      "learning_rate": 0.0005918367666111221,
      "loss": 2.4192,
      "step": 17148
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6331343650817871,
      "learning_rate": 0.0005918358188339336,
      "loss": 2.4864,
      "step": 17149
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.557914674282074,
      "learning_rate": 0.0005918348710024873,
      "loss": 2.1632,
      "step": 17150
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6648911237716675,
      "learning_rate": 0.0005918339231167834,
      "loss": 2.3728,
      "step": 17151
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6621159911155701,
      "learning_rate": 0.0005918329751768221,
      "loss": 2.2847,
      "step": 17152
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6604700088500977,
      "learning_rate": 0.0005918320271826035,
      "loss": 2.5117,
      "step": 17153
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.5739292502403259,
      "learning_rate": 0.0005918310791341279,
      "loss": 2.4758,
      "step": 17154
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6206468939781189,
      "learning_rate": 0.0005918301310313953,
      "loss": 2.4533,
      "step": 17155
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.599830150604248,
      "learning_rate": 0.000591829182874406,
      "loss": 2.4191,
      "step": 17156
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6449548602104187,
      "learning_rate": 0.0005918282346631602,
      "loss": 2.3235,
      "step": 17157
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6631696820259094,
      "learning_rate": 0.0005918272863976579,
      "loss": 2.3495,
      "step": 17158
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.5507327914237976,
      "learning_rate": 0.0005918263380778995,
      "loss": 2.0716,
      "step": 17159
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.7445036768913269,
      "learning_rate": 0.000591825389703885,
      "loss": 2.5903,
      "step": 17160
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6371435523033142,
      "learning_rate": 0.0005918244412756147,
      "loss": 2.6561,
      "step": 17161
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6443544030189514,
      "learning_rate": 0.0005918234927930887,
      "loss": 2.4629,
      "step": 17162
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.7511642575263977,
      "learning_rate": 0.0005918225442563073,
      "loss": 2.729,
      "step": 17163
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.5931403636932373,
      "learning_rate": 0.0005918215956652704,
      "loss": 2.3117,
      "step": 17164
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6053200364112854,
      "learning_rate": 0.0005918206470199786,
      "loss": 2.3016,
      "step": 17165
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.7170502543449402,
      "learning_rate": 0.0005918196983204316,
      "loss": 2.5461,
      "step": 17166
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.701392650604248,
      "learning_rate": 0.0005918187495666298,
      "loss": 2.3834,
      "step": 17167
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6083493828773499,
      "learning_rate": 0.0005918178007585736,
      "loss": 2.2163,
      "step": 17168
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6115067601203918,
      "learning_rate": 0.0005918168518962628,
      "loss": 2.4142,
      "step": 17169
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6679039597511292,
      "learning_rate": 0.0005918159029796978,
      "loss": 2.4885,
      "step": 17170
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.5962598323822021,
      "learning_rate": 0.0005918149540088787,
      "loss": 2.3415,
      "step": 17171
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.5858402848243713,
      "learning_rate": 0.0005918140049838058,
      "loss": 2.4755,
      "step": 17172
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6232991218566895,
      "learning_rate": 0.000591813055904479,
      "loss": 2.2793,
      "step": 17173
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6182102560997009,
      "learning_rate": 0.0005918121067708987,
      "loss": 2.416,
      "step": 17174
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6750932931900024,
      "learning_rate": 0.000591811157583065,
      "loss": 2.3722,
      "step": 17175
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6942341923713684,
      "learning_rate": 0.0005918102083409782,
      "loss": 2.4721,
      "step": 17176
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6588820815086365,
      "learning_rate": 0.0005918092590446383,
      "loss": 2.3875,
      "step": 17177
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6147258877754211,
      "learning_rate": 0.0005918083096940456,
      "loss": 2.3471,
      "step": 17178
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6206568479537964,
      "learning_rate": 0.0005918073602892,
      "loss": 2.3726,
      "step": 17179
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6070470213890076,
      "learning_rate": 0.0005918064108301022,
      "loss": 2.412,
      "step": 17180
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6441602110862732,
      "learning_rate": 0.000591805461316752,
      "loss": 2.3709,
      "step": 17181
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.5684126019477844,
      "learning_rate": 0.0005918045117491495,
      "loss": 2.2648,
      "step": 17182
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6788024306297302,
      "learning_rate": 0.0005918035621272951,
      "loss": 2.6424,
      "step": 17183
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.5807851552963257,
      "learning_rate": 0.000591802612451189,
      "loss": 2.081,
      "step": 17184
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6269130706787109,
      "learning_rate": 0.0005918016627208313,
      "loss": 2.622,
      "step": 17185
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6212854981422424,
      "learning_rate": 0.0005918007129362221,
      "loss": 2.4328,
      "step": 17186
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.607341468334198,
      "learning_rate": 0.0005917997630973616,
      "loss": 2.2478,
      "step": 17187
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.5821375846862793,
      "learning_rate": 0.0005917988132042501,
      "loss": 2.5042,
      "step": 17188
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.574845016002655,
      "learning_rate": 0.0005917978632568877,
      "loss": 2.492,
      "step": 17189
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6057098507881165,
      "learning_rate": 0.0005917969132552745,
      "loss": 2.4269,
      "step": 17190
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.7245128154754639,
      "learning_rate": 0.0005917959631994107,
      "loss": 2.5385,
      "step": 17191
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.5652766823768616,
      "learning_rate": 0.0005917950130892967,
      "loss": 2.6122,
      "step": 17192
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.5888753533363342,
      "learning_rate": 0.0005917940629249324,
      "loss": 2.128,
      "step": 17193
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6337082982063293,
      "learning_rate": 0.000591793112706318,
      "loss": 2.4656,
      "step": 17194
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.645287811756134,
      "learning_rate": 0.0005917921624334539,
      "loss": 2.2048,
      "step": 17195
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6675395369529724,
      "learning_rate": 0.00059179121210634,
      "loss": 2.3238,
      "step": 17196
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.648503839969635,
      "learning_rate": 0.0005917902617249767,
      "loss": 2.6803,
      "step": 17197
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6043696403503418,
      "learning_rate": 0.000591789311289364,
      "loss": 2.3077,
      "step": 17198
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.5924739837646484,
      "learning_rate": 0.0005917883607995021,
      "loss": 2.3571,
      "step": 17199
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6109692454338074,
      "learning_rate": 0.0005917874102553914,
      "loss": 2.2974,
      "step": 17200
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6131673455238342,
      "learning_rate": 0.0005917864596570319,
      "loss": 2.4698,
      "step": 17201
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6881905198097229,
      "learning_rate": 0.0005917855090044237,
      "loss": 2.5108,
      "step": 17202
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.589746356010437,
      "learning_rate": 0.0005917845582975672,
      "loss": 2.395,
      "step": 17203
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6282652616500854,
      "learning_rate": 0.0005917836075364622,
      "loss": 2.5748,
      "step": 17204
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6192938685417175,
      "learning_rate": 0.0005917826567211092,
      "loss": 2.2094,
      "step": 17205
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6809926629066467,
      "learning_rate": 0.0005917817058515085,
      "loss": 2.6021,
      "step": 17206
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.5797560811042786,
      "learning_rate": 0.0005917807549276599,
      "loss": 2.3617,
      "step": 17207
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6671321392059326,
      "learning_rate": 0.0005917798039495638,
      "loss": 2.4811,
      "step": 17208
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.565405011177063,
      "learning_rate": 0.0005917788529172203,
      "loss": 2.4559,
      "step": 17209
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.5795716643333435,
      "learning_rate": 0.0005917779018306296,
      "loss": 2.5964,
      "step": 17210
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6041696667671204,
      "learning_rate": 0.0005917769506897919,
      "loss": 2.3438,
      "step": 17211
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.617648184299469,
      "learning_rate": 0.0005917759994947074,
      "loss": 2.3782,
      "step": 17212
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6880923509597778,
      "learning_rate": 0.0005917750482453761,
      "loss": 2.4679,
      "step": 17213
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6765525937080383,
      "learning_rate": 0.0005917740969417984,
      "loss": 2.3249,
      "step": 17214
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6487724184989929,
      "learning_rate": 0.0005917731455839744,
      "loss": 2.2945,
      "step": 17215
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6959637403488159,
      "learning_rate": 0.0005917721941719043,
      "loss": 2.6917,
      "step": 17216
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.5972937345504761,
      "learning_rate": 0.0005917712427055882,
      "loss": 2.2973,
      "step": 17217
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.7375737428665161,
      "learning_rate": 0.0005917702911850263,
      "loss": 2.3389,
      "step": 17218
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6379202604293823,
      "learning_rate": 0.0005917693396102188,
      "loss": 2.4205,
      "step": 17219
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.5846556425094604,
      "learning_rate": 0.0005917683879811659,
      "loss": 2.662,
      "step": 17220
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6631814241409302,
      "learning_rate": 0.0005917674362978677,
      "loss": 2.5337,
      "step": 17221
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.5931336879730225,
      "learning_rate": 0.0005917664845603245,
      "loss": 2.4106,
      "step": 17222
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6411663889884949,
      "learning_rate": 0.0005917655327685363,
      "loss": 2.4918,
      "step": 17223
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6664015054702759,
      "learning_rate": 0.0005917645809225034,
      "loss": 2.377,
      "step": 17224
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.5938827395439148,
      "learning_rate": 0.0005917636290222259,
      "loss": 2.5113,
      "step": 17225
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.8146361112594604,
      "learning_rate": 0.0005917626770677042,
      "loss": 2.2562,
      "step": 17226
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6945332288742065,
      "learning_rate": 0.0005917617250589382,
      "loss": 2.2594,
      "step": 17227
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.7081158757209778,
      "learning_rate": 0.0005917607729959282,
      "loss": 2.3273,
      "step": 17228
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.7428263425827026,
      "learning_rate": 0.0005917598208786745,
      "loss": 2.1925,
      "step": 17229
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.58844393491745,
      "learning_rate": 0.000591758868707177,
      "loss": 2.534,
      "step": 17230
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.5651671290397644,
      "learning_rate": 0.000591757916481436,
      "loss": 2.2678,
      "step": 17231
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6354098916053772,
      "learning_rate": 0.0005917569642014518,
      "loss": 2.5947,
      "step": 17232
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6618444323539734,
      "learning_rate": 0.0005917560118672245,
      "loss": 2.2736,
      "step": 17233
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6804773211479187,
      "learning_rate": 0.0005917550594787541,
      "loss": 2.2552,
      "step": 17234
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.5564666986465454,
      "learning_rate": 0.000591754107036041,
      "loss": 2.2824,
      "step": 17235
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.5875210762023926,
      "learning_rate": 0.0005917531545390853,
      "loss": 2.2364,
      "step": 17236
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.65299391746521,
      "learning_rate": 0.0005917522019878871,
      "loss": 2.3602,
      "step": 17237
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6169107556343079,
      "learning_rate": 0.0005917512493824468,
      "loss": 2.4157,
      "step": 17238
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.5811507105827332,
      "learning_rate": 0.0005917502967227644,
      "loss": 2.4916,
      "step": 17239
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.64308100938797,
      "learning_rate": 0.0005917493440088401,
      "loss": 2.2753,
      "step": 17240
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.7068622708320618,
      "learning_rate": 0.000591748391240674,
      "loss": 2.4683,
      "step": 17241
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.5813337564468384,
      "learning_rate": 0.0005917474384182664,
      "loss": 2.2362,
      "step": 17242
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6087464690208435,
      "learning_rate": 0.0005917464855416175,
      "loss": 2.1914,
      "step": 17243
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.5917766690254211,
      "learning_rate": 0.0005917455326107275,
      "loss": 2.2844,
      "step": 17244
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6816511750221252,
      "learning_rate": 0.0005917445796255964,
      "loss": 2.4369,
      "step": 17245
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.5931470394134521,
      "learning_rate": 0.0005917436265862245,
      "loss": 2.4963,
      "step": 17246
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.7022151947021484,
      "learning_rate": 0.0005917426734926119,
      "loss": 2.3941,
      "step": 17247
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6011471152305603,
      "learning_rate": 0.0005917417203447588,
      "loss": 2.4082,
      "step": 17248
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6474723219871521,
      "learning_rate": 0.0005917407671426655,
      "loss": 2.5402,
      "step": 17249
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6803001165390015,
      "learning_rate": 0.0005917398138863321,
      "loss": 2.3719,
      "step": 17250
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6119241118431091,
      "learning_rate": 0.0005917388605757587,
      "loss": 2.3851,
      "step": 17251
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.5802586674690247,
      "learning_rate": 0.0005917379072109455,
      "loss": 2.2925,
      "step": 17252
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6664456725120544,
      "learning_rate": 0.0005917369537918928,
      "loss": 2.1684,
      "step": 17253
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6098693013191223,
      "learning_rate": 0.0005917360003186007,
      "loss": 2.3288,
      "step": 17254
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.7261695861816406,
      "learning_rate": 0.0005917350467910693,
      "loss": 2.4332,
      "step": 17255
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6500346064567566,
      "learning_rate": 0.0005917340932092989,
      "loss": 2.3076,
      "step": 17256
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.7204270362854004,
      "learning_rate": 0.0005917331395732896,
      "loss": 2.3804,
      "step": 17257
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6493664383888245,
      "learning_rate": 0.0005917321858830417,
      "loss": 2.1891,
      "step": 17258
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.623220682144165,
      "learning_rate": 0.0005917312321385551,
      "loss": 2.6781,
      "step": 17259
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6300441026687622,
      "learning_rate": 0.0005917302783398303,
      "loss": 2.2177,
      "step": 17260
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.7151634097099304,
      "learning_rate": 0.0005917293244868672,
      "loss": 2.6285,
      "step": 17261
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.5837169289588928,
      "learning_rate": 0.0005917283705796663,
      "loss": 2.3496,
      "step": 17262
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.8353265523910522,
      "learning_rate": 0.0005917274166182274,
      "loss": 2.4442,
      "step": 17263
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.9648017287254333,
      "learning_rate": 0.000591726462602551,
      "loss": 2.2229,
      "step": 17264
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.65180903673172,
      "learning_rate": 0.000591725508532637,
      "loss": 2.4979,
      "step": 17265
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.5831831693649292,
      "learning_rate": 0.0005917245544084859,
      "loss": 2.6787,
      "step": 17266
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.7171081900596619,
      "learning_rate": 0.0005917236002300976,
      "loss": 2.3193,
      "step": 17267
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.5878472924232483,
      "learning_rate": 0.0005917226459974723,
      "loss": 2.3248,
      "step": 17268
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.601553201675415,
      "learning_rate": 0.0005917216917106105,
      "loss": 2.336,
      "step": 17269
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.7010558843612671,
      "learning_rate": 0.0005917207373695119,
      "loss": 2.2152,
      "step": 17270
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.5941214561462402,
      "learning_rate": 0.000591719782974177,
      "loss": 2.3329,
      "step": 17271
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.5985793471336365,
      "learning_rate": 0.0005917188285246059,
      "loss": 2.5786,
      "step": 17272
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.5624070763587952,
      "learning_rate": 0.0005917178740207987,
      "loss": 2.2801,
      "step": 17273
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6181527376174927,
      "learning_rate": 0.0005917169194627557,
      "loss": 2.2375,
      "step": 17274
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.633797287940979,
      "learning_rate": 0.000591715964850477,
      "loss": 2.6503,
      "step": 17275
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.5881206393241882,
      "learning_rate": 0.0005917150101839628,
      "loss": 2.3898,
      "step": 17276
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.5903227925300598,
      "learning_rate": 0.0005917140554632133,
      "loss": 2.6142,
      "step": 17277
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.617560863494873,
      "learning_rate": 0.0005917131006882285,
      "loss": 2.181,
      "step": 17278
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.5822889804840088,
      "learning_rate": 0.0005917121458590089,
      "loss": 2.4333,
      "step": 17279
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6524328589439392,
      "learning_rate": 0.0005917111909755545,
      "loss": 2.5304,
      "step": 17280
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6124866008758545,
      "learning_rate": 0.0005917102360378653,
      "loss": 2.3122,
      "step": 17281
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.7379790544509888,
      "learning_rate": 0.0005917092810459419,
      "loss": 2.3668,
      "step": 17282
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.620003879070282,
      "learning_rate": 0.0005917083259997841,
      "loss": 2.302,
      "step": 17283
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6981155276298523,
      "learning_rate": 0.0005917073708993922,
      "loss": 2.3389,
      "step": 17284
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.5810133814811707,
      "learning_rate": 0.0005917064157447664,
      "loss": 2.2321,
      "step": 17285
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6777700781822205,
      "learning_rate": 0.0005917054605359069,
      "loss": 2.3381,
      "step": 17286
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6909803152084351,
      "learning_rate": 0.0005917045052728139,
      "loss": 2.2734,
      "step": 17287
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.5900611877441406,
      "learning_rate": 0.0005917035499554875,
      "loss": 2.0704,
      "step": 17288
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6078745722770691,
      "learning_rate": 0.0005917025945839278,
      "loss": 2.3059,
      "step": 17289
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6600989103317261,
      "learning_rate": 0.0005917016391581352,
      "loss": 2.336,
      "step": 17290
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6410177946090698,
      "learning_rate": 0.0005917006836781097,
      "loss": 2.3572,
      "step": 17291
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6098657846450806,
      "learning_rate": 0.0005916997281438515,
      "loss": 2.5099,
      "step": 17292
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6491407752037048,
      "learning_rate": 0.0005916987725553609,
      "loss": 2.2836,
      "step": 17293
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6490058898925781,
      "learning_rate": 0.000591697816912638,
      "loss": 2.6655,
      "step": 17294
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.5991870760917664,
      "learning_rate": 0.0005916968612156829,
      "loss": 2.3139,
      "step": 17295
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.5523091554641724,
      "learning_rate": 0.000591695905464496,
      "loss": 2.4051,
      "step": 17296
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6318620443344116,
      "learning_rate": 0.0005916949496590771,
      "loss": 2.335,
      "step": 17297
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.5948887467384338,
      "learning_rate": 0.0005916939937994268,
      "loss": 2.3765,
      "step": 17298
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.621389627456665,
      "learning_rate": 0.000591693037885545,
      "loss": 2.3794,
      "step": 17299
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.5946601033210754,
      "learning_rate": 0.0005916920819174319,
      "loss": 2.4901,
      "step": 17300
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.7013084292411804,
      "learning_rate": 0.0005916911258950878,
      "loss": 2.6455,
      "step": 17301
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.5771340727806091,
      "learning_rate": 0.0005916901698185127,
      "loss": 2.5914,
      "step": 17302
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.5665885210037231,
      "learning_rate": 0.0005916892136877071,
      "loss": 2.3363,
      "step": 17303
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.5905203223228455,
      "learning_rate": 0.0005916882575026708,
      "loss": 2.5082,
      "step": 17304
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6058043241500854,
      "learning_rate": 0.0005916873012634042,
      "loss": 2.7364,
      "step": 17305
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6108012795448303,
      "learning_rate": 0.0005916863449699075,
      "loss": 2.4619,
      "step": 17306
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6019692420959473,
      "learning_rate": 0.0005916853886221808,
      "loss": 2.3083,
      "step": 17307
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6162144541740417,
      "learning_rate": 0.0005916844322202241,
      "loss": 2.4821,
      "step": 17308
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6184638142585754,
      "learning_rate": 0.0005916834757640379,
      "loss": 2.3846,
      "step": 17309
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6435683369636536,
      "learning_rate": 0.0005916825192536221,
      "loss": 2.3686,
      "step": 17310
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6479054093360901,
      "learning_rate": 0.0005916815626889773,
      "loss": 2.5851,
      "step": 17311
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6605281829833984,
      "learning_rate": 0.0005916806060701032,
      "loss": 2.6468,
      "step": 17312
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.600754976272583,
      "learning_rate": 0.000591679649397,
      "loss": 2.3363,
      "step": 17313
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6574917435646057,
      "learning_rate": 0.0005916786926696684,
      "loss": 2.2858,
      "step": 17314
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.5881226062774658,
      "learning_rate": 0.000591677735888108,
      "loss": 2.3503,
      "step": 17315
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6498476266860962,
      "learning_rate": 0.0005916767790523193,
      "loss": 2.183,
      "step": 17316
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.574288547039032,
      "learning_rate": 0.0005916758221623023,
      "loss": 2.5089,
      "step": 17317
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6116025447845459,
      "learning_rate": 0.0005916748652180574,
      "loss": 2.3302,
      "step": 17318
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.637673020362854,
      "learning_rate": 0.0005916739082195845,
      "loss": 2.3354,
      "step": 17319
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6680563688278198,
      "learning_rate": 0.000591672951166884,
      "loss": 2.4381,
      "step": 17320
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6319157481193542,
      "learning_rate": 0.0005916719940599559,
      "loss": 2.2871,
      "step": 17321
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6326442956924438,
      "learning_rate": 0.0005916710368988004,
      "loss": 2.4791,
      "step": 17322
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6488668322563171,
      "learning_rate": 0.0005916700796834179,
      "loss": 2.3331,
      "step": 17323
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6546176075935364,
      "learning_rate": 0.0005916691224138084,
      "loss": 2.5695,
      "step": 17324
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.5889515280723572,
      "learning_rate": 0.0005916681650899721,
      "loss": 2.4305,
      "step": 17325
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6065460443496704,
      "learning_rate": 0.0005916672077119092,
      "loss": 2.3023,
      "step": 17326
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6606476902961731,
      "learning_rate": 0.0005916662502796198,
      "loss": 2.2933,
      "step": 17327
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.5711015462875366,
      "learning_rate": 0.0005916652927931042,
      "loss": 2.3804,
      "step": 17328
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6398429274559021,
      "learning_rate": 0.0005916643352523625,
      "loss": 2.489,
      "step": 17329
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6508769989013672,
      "learning_rate": 0.0005916633776573949,
      "loss": 2.3622,
      "step": 17330
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.5955080389976501,
      "learning_rate": 0.0005916624200082015,
      "loss": 2.6615,
      "step": 17331
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.629164457321167,
      "learning_rate": 0.0005916614623047826,
      "loss": 2.4938,
      "step": 17332
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6105042099952698,
      "learning_rate": 0.0005916605045471384,
      "loss": 2.3989,
      "step": 17333
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.5918390154838562,
      "learning_rate": 0.000591659546735269,
      "loss": 2.6865,
      "step": 17334
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6025588512420654,
      "learning_rate": 0.0005916585888691745,
      "loss": 2.4443,
      "step": 17335
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6062098741531372,
      "learning_rate": 0.0005916576309488551,
      "loss": 2.3821,
      "step": 17336
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.647447943687439,
      "learning_rate": 0.0005916566729743111,
      "loss": 2.1006,
      "step": 17337
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6069098114967346,
      "learning_rate": 0.0005916557149455427,
      "loss": 2.264,
      "step": 17338
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6341550946235657,
      "learning_rate": 0.00059165475686255,
      "loss": 2.4969,
      "step": 17339
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6084566116333008,
      "learning_rate": 0.0005916537987253331,
      "loss": 2.2305,
      "step": 17340
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6042293310165405,
      "learning_rate": 0.0005916528405338922,
      "loss": 2.2623,
      "step": 17341
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6512556076049805,
      "learning_rate": 0.0005916518822882277,
      "loss": 2.3333,
      "step": 17342
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.5978987812995911,
      "learning_rate": 0.0005916509239883395,
      "loss": 2.2147,
      "step": 17343
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6194216012954712,
      "learning_rate": 0.0005916499656342279,
      "loss": 2.3755,
      "step": 17344
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6171963214874268,
      "learning_rate": 0.000591649007225893,
      "loss": 2.3455,
      "step": 17345
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6286950707435608,
      "learning_rate": 0.0005916480487633352,
      "loss": 2.1986,
      "step": 17346
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6145004034042358,
      "learning_rate": 0.0005916470902465544,
      "loss": 2.4853,
      "step": 17347
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.588493287563324,
      "learning_rate": 0.0005916461316755509,
      "loss": 2.5397,
      "step": 17348
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6198959350585938,
      "learning_rate": 0.000591645173050325,
      "loss": 2.4361,
      "step": 17349
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6213573217391968,
      "learning_rate": 0.0005916442143708766,
      "loss": 2.3213,
      "step": 17350
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6379910707473755,
      "learning_rate": 0.0005916432556372062,
      "loss": 2.412,
      "step": 17351
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6469449400901794,
      "learning_rate": 0.0005916422968493137,
      "loss": 2.4749,
      "step": 17352
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.5968279838562012,
      "learning_rate": 0.0005916413380071993,
      "loss": 2.3222,
      "step": 17353
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.5726639628410339,
      "learning_rate": 0.0005916403791108634,
      "loss": 2.3935,
      "step": 17354
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.5663841366767883,
      "learning_rate": 0.0005916394201603061,
      "loss": 2.2087,
      "step": 17355
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6465103626251221,
      "learning_rate": 0.0005916384611555275,
      "loss": 2.3506,
      "step": 17356
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.5875303149223328,
      "learning_rate": 0.0005916375020965278,
      "loss": 2.508,
      "step": 17357
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6101332902908325,
      "learning_rate": 0.0005916365429833071,
      "loss": 2.3406,
      "step": 17358
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.619441568851471,
      "learning_rate": 0.0005916355838158657,
      "loss": 2.2497,
      "step": 17359
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6016408801078796,
      "learning_rate": 0.0005916346245942037,
      "loss": 2.5515,
      "step": 17360
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6357102990150452,
      "learning_rate": 0.0005916336653183214,
      "loss": 2.6016,
      "step": 17361
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.636495053768158,
      "learning_rate": 0.0005916327059882189,
      "loss": 2.4214,
      "step": 17362
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6073524355888367,
      "learning_rate": 0.0005916317466038963,
      "loss": 2.5214,
      "step": 17363
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6682648062705994,
      "learning_rate": 0.000591630787165354,
      "loss": 2.4624,
      "step": 17364
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6308607459068298,
      "learning_rate": 0.0005916298276725919,
      "loss": 2.3828,
      "step": 17365
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6559003591537476,
      "learning_rate": 0.0005916288681256103,
      "loss": 2.4058,
      "step": 17366
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6201838850975037,
      "learning_rate": 0.0005916279085244094,
      "loss": 2.5072,
      "step": 17367
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.5878891348838806,
      "learning_rate": 0.0005916269488689894,
      "loss": 2.3477,
      "step": 17368
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.5420571565628052,
      "learning_rate": 0.0005916259891593504,
      "loss": 2.3054,
      "step": 17369
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.7073853015899658,
      "learning_rate": 0.0005916250293954927,
      "loss": 2.649,
      "step": 17370
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6047576069831848,
      "learning_rate": 0.0005916240695774163,
      "loss": 2.4707,
      "step": 17371
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.7262892127037048,
      "learning_rate": 0.0005916231097051216,
      "loss": 2.4187,
      "step": 17372
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.5846555233001709,
      "learning_rate": 0.0005916221497786086,
      "loss": 2.4227,
      "step": 17373
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6045824289321899,
      "learning_rate": 0.0005916211897978776,
      "loss": 2.3977,
      "step": 17374
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6320884227752686,
      "learning_rate": 0.0005916202297629286,
      "loss": 2.5656,
      "step": 17375
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.5986950993537903,
      "learning_rate": 0.0005916192696737619,
      "loss": 2.5379,
      "step": 17376
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.5767408013343811,
      "learning_rate": 0.0005916183095303778,
      "loss": 2.2436,
      "step": 17377
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6003484129905701,
      "learning_rate": 0.0005916173493327763,
      "loss": 2.3854,
      "step": 17378
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.5685359239578247,
      "learning_rate": 0.0005916163890809576,
      "loss": 2.3188,
      "step": 17379
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6203973293304443,
      "learning_rate": 0.0005916154287749219,
      "loss": 2.4481,
      "step": 17380
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.562890887260437,
      "learning_rate": 0.0005916144684146693,
      "loss": 2.3357,
      "step": 17381
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.5618078112602234,
      "learning_rate": 0.0005916135080002002,
      "loss": 2.3327,
      "step": 17382
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6209571361541748,
      "learning_rate": 0.0005916125475315146,
      "loss": 2.5462,
      "step": 17383
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6231935620307922,
      "learning_rate": 0.0005916115870086127,
      "loss": 2.3871,
      "step": 17384
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.5716196894645691,
      "learning_rate": 0.0005916106264314948,
      "loss": 2.2839,
      "step": 17385
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.7098633646965027,
      "learning_rate": 0.000591609665800161,
      "loss": 2.4228,
      "step": 17386
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6557557582855225,
      "learning_rate": 0.0005916087051146113,
      "loss": 2.3145,
      "step": 17387
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6610888838768005,
      "learning_rate": 0.0005916077443748461,
      "loss": 2.2042,
      "step": 17388
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.5953640341758728,
      "learning_rate": 0.0005916067835808655,
      "loss": 2.4037,
      "step": 17389
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6139073371887207,
      "learning_rate": 0.0005916058227326698,
      "loss": 2.4394,
      "step": 17390
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6281252503395081,
      "learning_rate": 0.000591604861830259,
      "loss": 2.2324,
      "step": 17391
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.651983380317688,
      "learning_rate": 0.0005916039008736333,
      "loss": 2.4198,
      "step": 17392
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.5953751802444458,
      "learning_rate": 0.000591602939862793,
      "loss": 2.3986,
      "step": 17393
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.5876410007476807,
      "learning_rate": 0.0005916019787977381,
      "loss": 2.1805,
      "step": 17394
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6256920099258423,
      "learning_rate": 0.000591601017678469,
      "loss": 2.2487,
      "step": 17395
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6158996820449829,
      "learning_rate": 0.0005916000565049859,
      "loss": 2.3301,
      "step": 17396
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6206069588661194,
      "learning_rate": 0.0005915990952772885,
      "loss": 2.3883,
      "step": 17397
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6234563589096069,
      "learning_rate": 0.0005915981339953775,
      "loss": 2.4805,
      "step": 17398
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.5925639271736145,
      "learning_rate": 0.0005915971726592529,
      "loss": 2.2458,
      "step": 17399
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6159135699272156,
      "learning_rate": 0.0005915962112689149,
      "loss": 2.4744,
      "step": 17400
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6023901700973511,
      "learning_rate": 0.0005915952498243638,
      "loss": 2.5006,
      "step": 17401
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.5424709916114807,
      "learning_rate": 0.0005915942883255995,
      "loss": 2.2977,
      "step": 17402
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.5643942356109619,
      "learning_rate": 0.0005915933267726222,
      "loss": 2.311,
      "step": 17403
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6018670797348022,
      "learning_rate": 0.0005915923651654323,
      "loss": 2.3644,
      "step": 17404
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6137202978134155,
      "learning_rate": 0.0005915914035040299,
      "loss": 2.6689,
      "step": 17405
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6250285506248474,
      "learning_rate": 0.0005915904417884152,
      "loss": 2.5662,
      "step": 17406
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6176413297653198,
      "learning_rate": 0.0005915894800185882,
      "loss": 2.3375,
      "step": 17407
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.5595508813858032,
      "learning_rate": 0.0005915885181945492,
      "loss": 2.4093,
      "step": 17408
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6423996686935425,
      "learning_rate": 0.0005915875563162986,
      "loss": 2.3736,
      "step": 17409
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.5741569995880127,
      "learning_rate": 0.0005915865943838362,
      "loss": 2.4355,
      "step": 17410
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.5732694268226624,
      "learning_rate": 0.0005915856323971624,
      "loss": 2.2163,
      "step": 17411
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6073222756385803,
      "learning_rate": 0.0005915846703562772,
      "loss": 2.4997,
      "step": 17412
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6867368221282959,
      "learning_rate": 0.0005915837082611812,
      "loss": 2.2825,
      "step": 17413
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.5832141041755676,
      "learning_rate": 0.000591582746111874,
      "loss": 2.3572,
      "step": 17414
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.58866286277771,
      "learning_rate": 0.0005915817839083562,
      "loss": 2.445,
      "step": 17415
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6506401896476746,
      "learning_rate": 0.0005915808216506277,
      "loss": 2.6055,
      "step": 17416
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6438153386116028,
      "learning_rate": 0.000591579859338689,
      "loss": 2.4423,
      "step": 17417
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.5678859949111938,
      "learning_rate": 0.00059157889697254,
      "loss": 2.4739,
      "step": 17418
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6288229823112488,
      "learning_rate": 0.000591577934552181,
      "loss": 2.3012,
      "step": 17419
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6462923288345337,
      "learning_rate": 0.0005915769720776122,
      "loss": 2.2126,
      "step": 17420
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.5765368342399597,
      "learning_rate": 0.0005915760095488337,
      "loss": 2.6358,
      "step": 17421
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6242533326148987,
      "learning_rate": 0.0005915750469658458,
      "loss": 2.4574,
      "step": 17422
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.5910988450050354,
      "learning_rate": 0.0005915740843286485,
      "loss": 2.2894,
      "step": 17423
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6026703715324402,
      "learning_rate": 0.000591573121637242,
      "loss": 2.3299,
      "step": 17424
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.7209562063217163,
      "learning_rate": 0.0005915721588916267,
      "loss": 2.2582,
      "step": 17425
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.5699760913848877,
      "learning_rate": 0.0005915711960918026,
      "loss": 2.5378,
      "step": 17426
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.5835921168327332,
      "learning_rate": 0.0005915702332377698,
      "loss": 2.5256,
      "step": 17427
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.5912199020385742,
      "learning_rate": 0.0005915692703295287,
      "loss": 2.361,
      "step": 17428
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6126841902732849,
      "learning_rate": 0.0005915683073670794,
      "loss": 2.2274,
      "step": 17429
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6217760443687439,
      "learning_rate": 0.000591567344350422,
      "loss": 2.703,
      "step": 17430
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.7346127033233643,
      "learning_rate": 0.0005915663812795567,
      "loss": 2.2148,
      "step": 17431
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6395018100738525,
      "learning_rate": 0.0005915654181544836,
      "loss": 2.3391,
      "step": 17432
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6399160027503967,
      "learning_rate": 0.0005915644549752032,
      "loss": 2.3376,
      "step": 17433
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.8768515586853027,
      "learning_rate": 0.0005915634917417154,
      "loss": 2.5082,
      "step": 17434
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.8604424595832825,
      "learning_rate": 0.0005915625284540204,
      "loss": 2.4613,
      "step": 17435
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.9135926365852356,
      "learning_rate": 0.0005915615651121183,
      "loss": 2.4707,
      "step": 17436
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6892595887184143,
      "learning_rate": 0.0005915606017160097,
      "loss": 2.3749,
      "step": 17437
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.8468711376190186,
      "learning_rate": 0.0005915596382656943,
      "loss": 2.4615,
      "step": 17438
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6544553637504578,
      "learning_rate": 0.0005915586747611725,
      "loss": 2.4083,
      "step": 17439
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.8111764192581177,
      "learning_rate": 0.0005915577112024444,
      "loss": 2.0498,
      "step": 17440
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.7362029552459717,
      "learning_rate": 0.0005915567475895102,
      "loss": 2.365,
      "step": 17441
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6474376320838928,
      "learning_rate": 0.0005915557839223702,
      "loss": 2.1604,
      "step": 17442
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6555480360984802,
      "learning_rate": 0.0005915548202010244,
      "loss": 2.4595,
      "step": 17443
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6472779512405396,
      "learning_rate": 0.000591553856425473,
      "loss": 2.4153,
      "step": 17444
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.660868763923645,
      "learning_rate": 0.0005915528925957163,
      "loss": 2.5894,
      "step": 17445
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.578528106212616,
      "learning_rate": 0.0005915519287117545,
      "loss": 2.1229,
      "step": 17446
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.5793160796165466,
      "learning_rate": 0.0005915509647735876,
      "loss": 2.3048,
      "step": 17447
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.749686062335968,
      "learning_rate": 0.0005915500007812158,
      "loss": 2.3585,
      "step": 17448
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6724002957344055,
      "learning_rate": 0.0005915490367346394,
      "loss": 2.2346,
      "step": 17449
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6655895709991455,
      "learning_rate": 0.0005915480726338587,
      "loss": 2.434,
      "step": 17450
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.7584390044212341,
      "learning_rate": 0.0005915471084788734,
      "loss": 2.446,
      "step": 17451
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.609959065914154,
      "learning_rate": 0.0005915461442696843,
      "loss": 2.3283,
      "step": 17452
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6013283729553223,
      "learning_rate": 0.0005915451800062911,
      "loss": 2.4434,
      "step": 17453
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6833123564720154,
      "learning_rate": 0.0005915442156886942,
      "loss": 2.3741,
      "step": 17454
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6866865158081055,
      "learning_rate": 0.0005915432513168937,
      "loss": 2.4278,
      "step": 17455
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.60523521900177,
      "learning_rate": 0.0005915422868908898,
      "loss": 2.3083,
      "step": 17456
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.656269371509552,
      "learning_rate": 0.0005915413224106827,
      "loss": 2.3292,
      "step": 17457
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.5964441895484924,
      "learning_rate": 0.0005915403578762724,
      "loss": 2.4465,
      "step": 17458
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.7407733201980591,
      "learning_rate": 0.0005915393932876595,
      "loss": 2.4116,
      "step": 17459
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.557702898979187,
      "learning_rate": 0.0005915384286448439,
      "loss": 2.3196,
      "step": 17460
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.5856769680976868,
      "learning_rate": 0.0005915374639478256,
      "loss": 2.4103,
      "step": 17461
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.7062947154045105,
      "learning_rate": 0.0005915364991966052,
      "loss": 2.2433,
      "step": 17462
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.5872236490249634,
      "learning_rate": 0.0005915355343911826,
      "loss": 2.3014,
      "step": 17463
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6255722045898438,
      "learning_rate": 0.000591534569531558,
      "loss": 2.3636,
      "step": 17464
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.7167164087295532,
      "learning_rate": 0.0005915336046177316,
      "loss": 2.2879,
      "step": 17465
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.2273448705673218,
      "learning_rate": 0.0005915326396497036,
      "loss": 2.4305,
      "step": 17466
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6498959064483643,
      "learning_rate": 0.0005915316746274743,
      "loss": 2.3201,
      "step": 17467
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6476115584373474,
      "learning_rate": 0.0005915307095510436,
      "loss": 2.3316,
      "step": 17468
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.7044664621353149,
      "learning_rate": 0.000591529744420412,
      "loss": 2.4703,
      "step": 17469
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.5924627780914307,
      "learning_rate": 0.0005915287792355794,
      "loss": 2.2615,
      "step": 17470
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.5918035507202148,
      "learning_rate": 0.0005915278139965461,
      "loss": 2.286,
      "step": 17471
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6837812066078186,
      "learning_rate": 0.0005915268487033123,
      "loss": 2.4801,
      "step": 17472
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6200200915336609,
      "learning_rate": 0.0005915258833558781,
      "loss": 2.5296,
      "step": 17473
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6838415265083313,
      "learning_rate": 0.0005915249179542438,
      "loss": 2.4006,
      "step": 17474
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.5932849049568176,
      "learning_rate": 0.0005915239524984096,
      "loss": 2.3389,
      "step": 17475
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6553565263748169,
      "learning_rate": 0.0005915229869883755,
      "loss": 2.2041,
      "step": 17476
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6308945417404175,
      "learning_rate": 0.0005915220214241418,
      "loss": 2.4003,
      "step": 17477
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.5812474489212036,
      "learning_rate": 0.0005915210558057086,
      "loss": 2.5222,
      "step": 17478
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.630018949508667,
      "learning_rate": 0.0005915200901330761,
      "loss": 2.5554,
      "step": 17479
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.7984712719917297,
      "learning_rate": 0.0005915191244062446,
      "loss": 2.4804,
      "step": 17480
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6273117661476135,
      "learning_rate": 0.0005915181586252142,
      "loss": 2.6949,
      "step": 17481
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.5759595036506653,
      "learning_rate": 0.000591517192789985,
      "loss": 2.545,
      "step": 17482
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.665589451789856,
      "learning_rate": 0.0005915162269005573,
      "loss": 2.371,
      "step": 17483
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6214540004730225,
      "learning_rate": 0.0005915152609569312,
      "loss": 2.3042,
      "step": 17484
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6129361987113953,
      "learning_rate": 0.000591514294959107,
      "loss": 2.6667,
      "step": 17485
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.5842055082321167,
      "learning_rate": 0.0005915133289070846,
      "loss": 2.0928,
      "step": 17486
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6888399720191956,
      "learning_rate": 0.0005915123628008645,
      "loss": 2.2555,
      "step": 17487
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6354576349258423,
      "learning_rate": 0.0005915113966404468,
      "loss": 2.4441,
      "step": 17488
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.5794022679328918,
      "learning_rate": 0.0005915104304258315,
      "loss": 2.3085,
      "step": 17489
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6103563904762268,
      "learning_rate": 0.0005915094641570191,
      "loss": 2.5055,
      "step": 17490
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.8517723679542542,
      "learning_rate": 0.0005915084978340094,
      "loss": 2.209,
      "step": 17491
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6010738015174866,
      "learning_rate": 0.0005915075314568028,
      "loss": 2.2818,
      "step": 17492
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.5860599875450134,
      "learning_rate": 0.0005915065650253995,
      "loss": 2.3053,
      "step": 17493
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.7855478525161743,
      "learning_rate": 0.0005915055985397996,
      "loss": 2.4603,
      "step": 17494
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6100360751152039,
      "learning_rate": 0.0005915046320000033,
      "loss": 2.3565,
      "step": 17495
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6236811876296997,
      "learning_rate": 0.0005915036654060108,
      "loss": 2.3285,
      "step": 17496
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.7090363502502441,
      "learning_rate": 0.0005915026987578223,
      "loss": 2.5284,
      "step": 17497
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.684572696685791,
      "learning_rate": 0.0005915017320554378,
      "loss": 2.3623,
      "step": 17498
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6533570289611816,
      "learning_rate": 0.0005915007652988578,
      "loss": 2.4723,
      "step": 17499
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6585021018981934,
      "learning_rate": 0.0005914997984880823,
      "loss": 2.4316,
      "step": 17500
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6180310845375061,
      "learning_rate": 0.0005914988316231114,
      "loss": 2.404,
      "step": 17501
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6319675445556641,
      "learning_rate": 0.0005914978647039453,
      "loss": 2.2831,
      "step": 17502
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6735346913337708,
      "learning_rate": 0.0005914968977305844,
      "loss": 2.2884,
      "step": 17503
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6764410734176636,
      "learning_rate": 0.0005914959307030287,
      "loss": 2.4629,
      "step": 17504
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.5205037593841553,
      "learning_rate": 0.0005914949636212783,
      "loss": 2.2954,
      "step": 17505
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6470337510108948,
      "learning_rate": 0.0005914939964853336,
      "loss": 2.5462,
      "step": 17506
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.601978063583374,
      "learning_rate": 0.0005914930292951945,
      "loss": 2.3713,
      "step": 17507
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6183549165725708,
      "learning_rate": 0.0005914920620508615,
      "loss": 2.4164,
      "step": 17508
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6084624528884888,
      "learning_rate": 0.0005914910947523346,
      "loss": 2.4146,
      "step": 17509
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.631023108959198,
      "learning_rate": 0.0005914901273996139,
      "loss": 2.2509,
      "step": 17510
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.5823765993118286,
      "learning_rate": 0.0005914891599926998,
      "loss": 2.3406,
      "step": 17511
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.632682740688324,
      "learning_rate": 0.0005914881925315924,
      "loss": 2.3138,
      "step": 17512
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.5926030874252319,
      "learning_rate": 0.0005914872250162918,
      "loss": 2.351,
      "step": 17513
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6230476498603821,
      "learning_rate": 0.0005914862574467982,
      "loss": 2.3042,
      "step": 17514
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.5674728155136108,
      "learning_rate": 0.0005914852898231118,
      "loss": 2.3336,
      "step": 17515
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6273548603057861,
      "learning_rate": 0.0005914843221452328,
      "loss": 2.5799,
      "step": 17516
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6260058879852295,
      "learning_rate": 0.0005914833544131612,
      "loss": 2.6064,
      "step": 17517
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.5610143542289734,
      "learning_rate": 0.0005914823866268975,
      "loss": 2.2294,
      "step": 17518
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6319355964660645,
      "learning_rate": 0.0005914814187864418,
      "loss": 2.4241,
      "step": 17519
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6278893947601318,
      "learning_rate": 0.000591480450891794,
      "loss": 2.3704,
      "step": 17520
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.5642006397247314,
      "learning_rate": 0.0005914794829429546,
      "loss": 2.1635,
      "step": 17521
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6332024335861206,
      "learning_rate": 0.0005914785149399237,
      "loss": 2.4256,
      "step": 17522
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6469594836235046,
      "learning_rate": 0.0005914775468827013,
      "loss": 2.3657,
      "step": 17523
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.572076678276062,
      "learning_rate": 0.0005914765787712879,
      "loss": 2.4608,
      "step": 17524
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6211495995521545,
      "learning_rate": 0.0005914756106056834,
      "loss": 2.401,
      "step": 17525
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6075646877288818,
      "learning_rate": 0.0005914746423858881,
      "loss": 2.5432,
      "step": 17526
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.5927340388298035,
      "learning_rate": 0.0005914736741119022,
      "loss": 2.2723,
      "step": 17527
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.5757551193237305,
      "learning_rate": 0.0005914727057837257,
      "loss": 2.2301,
      "step": 17528
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6488621830940247,
      "learning_rate": 0.000591471737401359,
      "loss": 2.563,
      "step": 17529
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6433739066123962,
      "learning_rate": 0.0005914707689648023,
      "loss": 2.1877,
      "step": 17530
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.5600062012672424,
      "learning_rate": 0.0005914698004740556,
      "loss": 2.0721,
      "step": 17531
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.5839982032775879,
      "learning_rate": 0.0005914688319291192,
      "loss": 2.6692,
      "step": 17532
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6187384128570557,
      "learning_rate": 0.0005914678633299931,
      "loss": 2.4498,
      "step": 17533
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6299294233322144,
      "learning_rate": 0.0005914668946766778,
      "loss": 2.4573,
      "step": 17534
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6066994071006775,
      "learning_rate": 0.0005914659259691733,
      "loss": 2.4388,
      "step": 17535
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6141282320022583,
      "learning_rate": 0.0005914649572074796,
      "loss": 2.2297,
      "step": 17536
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6000007390975952,
      "learning_rate": 0.0005914639883915972,
      "loss": 2.3692,
      "step": 17537
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6421633362770081,
      "learning_rate": 0.0005914630195215262,
      "loss": 2.2984,
      "step": 17538
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6043990254402161,
      "learning_rate": 0.0005914620505972666,
      "loss": 2.3435,
      "step": 17539
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.7937813997268677,
      "learning_rate": 0.0005914610816188187,
      "loss": 2.4417,
      "step": 17540
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6024314761161804,
      "learning_rate": 0.0005914601125861828,
      "loss": 2.2688,
      "step": 17541
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6661416888237,
      "learning_rate": 0.0005914591434993589,
      "loss": 2.2848,
      "step": 17542
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.5730730891227722,
      "learning_rate": 0.0005914581743583472,
      "loss": 2.3421,
      "step": 17543
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6488872766494751,
      "learning_rate": 0.000591457205163148,
      "loss": 2.3072,
      "step": 17544
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.655113697052002,
      "learning_rate": 0.0005914562359137614,
      "loss": 2.4023,
      "step": 17545
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.663567841053009,
      "learning_rate": 0.0005914552666101875,
      "loss": 2.3938,
      "step": 17546
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6417375802993774,
      "learning_rate": 0.0005914542972524266,
      "loss": 2.4802,
      "step": 17547
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6086624264717102,
      "learning_rate": 0.000591453327840479,
      "loss": 2.3432,
      "step": 17548
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.5834698677062988,
      "learning_rate": 0.0005914523583743445,
      "loss": 2.3468,
      "step": 17549
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6462255716323853,
      "learning_rate": 0.0005914513888540235,
      "loss": 2.3397,
      "step": 17550
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6488681435585022,
      "learning_rate": 0.0005914504192795164,
      "loss": 2.4443,
      "step": 17551
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6749423146247864,
      "learning_rate": 0.0005914494496508229,
      "loss": 2.3508,
      "step": 17552
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6068698167800903,
      "learning_rate": 0.0005914484799679437,
      "loss": 2.4351,
      "step": 17553
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6192547678947449,
      "learning_rate": 0.0005914475102308786,
      "loss": 2.3609,
      "step": 17554
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6011633276939392,
      "learning_rate": 0.0005914465404396279,
      "loss": 2.5531,
      "step": 17555
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6302428245544434,
      "learning_rate": 0.0005914455705941918,
      "loss": 2.1582,
      "step": 17556
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.5993103384971619,
      "learning_rate": 0.0005914446006945705,
      "loss": 2.3554,
      "step": 17557
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6231852173805237,
      "learning_rate": 0.0005914436307407642,
      "loss": 2.7044,
      "step": 17558
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.8243353962898254,
      "learning_rate": 0.0005914426607327728,
      "loss": 2.5364,
      "step": 17559
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.7075626850128174,
      "learning_rate": 0.0005914416906705969,
      "loss": 2.104,
      "step": 17560
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6759673357009888,
      "learning_rate": 0.0005914407205542364,
      "loss": 2.5864,
      "step": 17561
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.5752148628234863,
      "learning_rate": 0.0005914397503836916,
      "loss": 2.5098,
      "step": 17562
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6148203015327454,
      "learning_rate": 0.0005914387801589627,
      "loss": 2.3471,
      "step": 17563
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.7251635193824768,
      "learning_rate": 0.0005914378098800498,
      "loss": 2.2651,
      "step": 17564
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.5920475125312805,
      "learning_rate": 0.0005914368395469531,
      "loss": 2.3232,
      "step": 17565
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6784692406654358,
      "learning_rate": 0.0005914358691596728,
      "loss": 2.4981,
      "step": 17566
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.5946099758148193,
      "learning_rate": 0.0005914348987182091,
      "loss": 2.3084,
      "step": 17567
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6120158433914185,
      "learning_rate": 0.000591433928222562,
      "loss": 2.3621,
      "step": 17568
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.5936734080314636,
      "learning_rate": 0.000591432957672732,
      "loss": 2.4388,
      "step": 17569
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6130625605583191,
      "learning_rate": 0.0005914319870687191,
      "loss": 2.4281,
      "step": 17570
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6147868633270264,
      "learning_rate": 0.0005914310164105234,
      "loss": 2.6591,
      "step": 17571
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.7674708962440491,
      "learning_rate": 0.0005914300456981452,
      "loss": 2.3602,
      "step": 17572
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6240364909172058,
      "learning_rate": 0.0005914290749315847,
      "loss": 2.0876,
      "step": 17573
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6442149877548218,
      "learning_rate": 0.000591428104110842,
      "loss": 2.2258,
      "step": 17574
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.7291965484619141,
      "learning_rate": 0.0005914271332359174,
      "loss": 2.2621,
      "step": 17575
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.7432341575622559,
      "learning_rate": 0.0005914261623068109,
      "loss": 2.1941,
      "step": 17576
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6446530818939209,
      "learning_rate": 0.0005914251913235228,
      "loss": 2.4039,
      "step": 17577
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.5962463617324829,
      "learning_rate": 0.0005914242202860533,
      "loss": 2.499,
      "step": 17578
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.7307490110397339,
      "learning_rate": 0.0005914232491944024,
      "loss": 2.5868,
      "step": 17579
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6973565816879272,
      "learning_rate": 0.0005914222780485706,
      "loss": 2.536,
      "step": 17580
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.653733491897583,
      "learning_rate": 0.0005914213068485577,
      "loss": 2.5754,
      "step": 17581
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.5713644027709961,
      "learning_rate": 0.0005914203355943642,
      "loss": 2.6817,
      "step": 17582
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6781857013702393,
      "learning_rate": 0.0005914193642859901,
      "loss": 2.4085,
      "step": 17583
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.661730945110321,
      "learning_rate": 0.0005914183929234358,
      "loss": 2.4098,
      "step": 17584
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.5881383419036865,
      "learning_rate": 0.0005914174215067011,
      "loss": 2.176,
      "step": 17585
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.695358395576477,
      "learning_rate": 0.0005914164500357865,
      "loss": 2.2782,
      "step": 17586
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6085706353187561,
      "learning_rate": 0.0005914154785106921,
      "loss": 2.3333,
      "step": 17587
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6553109288215637,
      "learning_rate": 0.0005914145069314179,
      "loss": 2.7415,
      "step": 17588
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.7158437371253967,
      "learning_rate": 0.0005914135352979645,
      "loss": 2.448,
      "step": 17589
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6894208788871765,
      "learning_rate": 0.0005914125636103316,
      "loss": 2.4143,
      "step": 17590
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.5789822936058044,
      "learning_rate": 0.0005914115918685198,
      "loss": 2.1807,
      "step": 17591
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.5956504344940186,
      "learning_rate": 0.0005914106200725289,
      "loss": 2.5811,
      "step": 17592
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6149575710296631,
      "learning_rate": 0.0005914096482223595,
      "loss": 2.4305,
      "step": 17593
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6515751481056213,
      "learning_rate": 0.0005914086763180113,
      "loss": 2.3044,
      "step": 17594
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.599754810333252,
      "learning_rate": 0.0005914077043594849,
      "loss": 2.5231,
      "step": 17595
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.5896777510643005,
      "learning_rate": 0.0005914067323467803,
      "loss": 2.5482,
      "step": 17596
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6345060467720032,
      "learning_rate": 0.0005914057602798976,
      "loss": 2.2714,
      "step": 17597
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6264289021492004,
      "learning_rate": 0.0005914047881588369,
      "loss": 2.3158,
      "step": 17598
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.647271454334259,
      "learning_rate": 0.0005914038159835988,
      "loss": 2.1962,
      "step": 17599
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6044416427612305,
      "learning_rate": 0.0005914028437541833,
      "loss": 2.4464,
      "step": 17600
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.601868212223053,
      "learning_rate": 0.0005914018714705903,
      "loss": 2.6562,
      "step": 17601
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.5603986978530884,
      "learning_rate": 0.0005914008991328203,
      "loss": 2.2467,
      "step": 17602
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6445021629333496,
      "learning_rate": 0.0005913999267408733,
      "loss": 2.2677,
      "step": 17603
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6203965544700623,
      "learning_rate": 0.0005913989542947496,
      "loss": 2.3648,
      "step": 17604
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6736214756965637,
      "learning_rate": 0.0005913979817944495,
      "loss": 2.371,
      "step": 17605
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6311670541763306,
      "learning_rate": 0.0005913970092399727,
      "loss": 2.3542,
      "step": 17606
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6977776885032654,
      "learning_rate": 0.0005913960366313198,
      "loss": 2.6535,
      "step": 17607
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6257930397987366,
      "learning_rate": 0.000591395063968491,
      "loss": 2.4753,
      "step": 17608
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6732868552207947,
      "learning_rate": 0.0005913940912514863,
      "loss": 2.35,
      "step": 17609
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.5926724076271057,
      "learning_rate": 0.0005913931184803059,
      "loss": 2.2927,
      "step": 17610
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6124989986419678,
      "learning_rate": 0.0005913921456549501,
      "loss": 2.6596,
      "step": 17611
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6379804015159607,
      "learning_rate": 0.0005913911727754188,
      "loss": 2.4335,
      "step": 17612
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.5995156168937683,
      "learning_rate": 0.0005913901998417125,
      "loss": 2.2004,
      "step": 17613
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6842386722564697,
      "learning_rate": 0.0005913892268538313,
      "loss": 2.3763,
      "step": 17614
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.63824862241745,
      "learning_rate": 0.0005913882538117754,
      "loss": 2.2509,
      "step": 17615
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6494688987731934,
      "learning_rate": 0.0005913872807155449,
      "loss": 2.4364,
      "step": 17616
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.5367636680603027,
      "learning_rate": 0.0005913863075651399,
      "loss": 2.3649,
      "step": 17617
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6044687032699585,
      "learning_rate": 0.0005913853343605607,
      "loss": 2.3695,
      "step": 17618
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.8209645748138428,
      "learning_rate": 0.0005913843611018074,
      "loss": 2.4145,
      "step": 17619
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6726528406143188,
      "learning_rate": 0.0005913833877888803,
      "loss": 2.5156,
      "step": 17620
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.5968509316444397,
      "learning_rate": 0.0005913824144217797,
      "loss": 2.2369,
      "step": 17621
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6072056889533997,
      "learning_rate": 0.0005913814410005054,
      "loss": 2.4461,
      "step": 17622
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.612822949886322,
      "learning_rate": 0.0005913804675250579,
      "loss": 2.5512,
      "step": 17623
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6316772103309631,
      "learning_rate": 0.0005913794939954373,
      "loss": 2.4494,
      "step": 17624
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.578414261341095,
      "learning_rate": 0.0005913785204116436,
      "loss": 2.3799,
      "step": 17625
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6492790579795837,
      "learning_rate": 0.0005913775467736773,
      "loss": 2.4319,
      "step": 17626
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6104375720024109,
      "learning_rate": 0.0005913765730815383,
      "loss": 2.1492,
      "step": 17627
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6695579886436462,
      "learning_rate": 0.000591375599335227,
      "loss": 2.6024,
      "step": 17628
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6078680753707886,
      "learning_rate": 0.0005913746255347433,
      "loss": 2.5815,
      "step": 17629
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.666257381439209,
      "learning_rate": 0.0005913736516800877,
      "loss": 2.7439,
      "step": 17630
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6774369478225708,
      "learning_rate": 0.0005913726777712602,
      "loss": 2.4353,
      "step": 17631
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.5997393131256104,
      "learning_rate": 0.0005913717038082611,
      "loss": 2.365,
      "step": 17632
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.5820026993751526,
      "learning_rate": 0.0005913707297910904,
      "loss": 2.3633,
      "step": 17633
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6026504635810852,
      "learning_rate": 0.0005913697557197483,
      "loss": 2.3121,
      "step": 17634
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6036786437034607,
      "learning_rate": 0.0005913687815942352,
      "loss": 2.2857,
      "step": 17635
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6214343905448914,
      "learning_rate": 0.0005913678074145511,
      "loss": 2.3677,
      "step": 17636
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6488041877746582,
      "learning_rate": 0.0005913668331806963,
      "loss": 2.226,
      "step": 17637
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6215164661407471,
      "learning_rate": 0.0005913658588926708,
      "loss": 2.2781,
      "step": 17638
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6532192230224609,
      "learning_rate": 0.0005913648845504749,
      "loss": 2.4633,
      "step": 17639
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6699473857879639,
      "learning_rate": 0.0005913639101541089,
      "loss": 2.2359,
      "step": 17640
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.8984994292259216,
      "learning_rate": 0.0005913629357035727,
      "loss": 2.5805,
      "step": 17641
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6581630110740662,
      "learning_rate": 0.0005913619611988666,
      "loss": 2.2029,
      "step": 17642
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6427106261253357,
      "learning_rate": 0.0005913609866399909,
      "loss": 2.3406,
      "step": 17643
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6187332272529602,
      "learning_rate": 0.0005913600120269457,
      "loss": 2.4145,
      "step": 17644
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6959636211395264,
      "learning_rate": 0.0005913590373597312,
      "loss": 2.3896,
      "step": 17645
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.609411895275116,
      "learning_rate": 0.0005913580626383475,
      "loss": 2.3005,
      "step": 17646
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6503731608390808,
      "learning_rate": 0.0005913570878627949,
      "loss": 2.2946,
      "step": 17647
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6261921525001526,
      "learning_rate": 0.0005913561130330734,
      "loss": 2.3908,
      "step": 17648
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6008002161979675,
      "learning_rate": 0.0005913551381491834,
      "loss": 2.4221,
      "step": 17649
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.5790508985519409,
      "learning_rate": 0.0005913541632111248,
      "loss": 2.5142,
      "step": 17650
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6192867755889893,
      "learning_rate": 0.0005913531882188982,
      "loss": 2.402,
      "step": 17651
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6348335146903992,
      "learning_rate": 0.0005913522131725035,
      "loss": 2.3404,
      "step": 17652
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.5725609660148621,
      "learning_rate": 0.0005913512380719408,
      "loss": 2.3751,
      "step": 17653
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6620392203330994,
      "learning_rate": 0.0005913502629172104,
      "loss": 2.4821,
      "step": 17654
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6400270462036133,
      "learning_rate": 0.0005913492877083127,
      "loss": 2.3811,
      "step": 17655
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.5799642205238342,
      "learning_rate": 0.0005913483124452474,
      "loss": 2.2405,
      "step": 17656
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6218028664588928,
      "learning_rate": 0.0005913473371280152,
      "loss": 2.2739,
      "step": 17657
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6852091550827026,
      "learning_rate": 0.0005913463617566158,
      "loss": 2.5888,
      "step": 17658
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6114898324012756,
      "learning_rate": 0.0005913453863310498,
      "loss": 2.5064,
      "step": 17659
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6191304326057434,
      "learning_rate": 0.0005913444108513171,
      "loss": 2.1623,
      "step": 17660
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6157969236373901,
      "learning_rate": 0.000591343435317418,
      "loss": 2.084,
      "step": 17661
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.592495322227478,
      "learning_rate": 0.0005913424597293526,
      "loss": 2.4025,
      "step": 17662
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6274166107177734,
      "learning_rate": 0.0005913414840871212,
      "loss": 2.2365,
      "step": 17663
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.5859852433204651,
      "learning_rate": 0.0005913405083907239,
      "loss": 2.4427,
      "step": 17664
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.5791913866996765,
      "learning_rate": 0.0005913395326401609,
      "loss": 2.3373,
      "step": 17665
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6703787446022034,
      "learning_rate": 0.0005913385568354324,
      "loss": 2.5036,
      "step": 17666
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6214412450790405,
      "learning_rate": 0.0005913375809765385,
      "loss": 2.4394,
      "step": 17667
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6061923503875732,
      "learning_rate": 0.0005913366050634795,
      "loss": 2.1827,
      "step": 17668
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6304381489753723,
      "learning_rate": 0.0005913356290962556,
      "loss": 2.5407,
      "step": 17669
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6635023355484009,
      "learning_rate": 0.0005913346530748668,
      "loss": 2.2399,
      "step": 17670
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6518003940582275,
      "learning_rate": 0.0005913336769993134,
      "loss": 2.4756,
      "step": 17671
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6430043578147888,
      "learning_rate": 0.0005913327008695955,
      "loss": 2.5301,
      "step": 17672
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6431523561477661,
      "learning_rate": 0.0005913317246857135,
      "loss": 2.3837,
      "step": 17673
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.5934010744094849,
      "learning_rate": 0.0005913307484476674,
      "loss": 2.303,
      "step": 17674
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.663704514503479,
      "learning_rate": 0.0005913297721554573,
      "loss": 2.5649,
      "step": 17675
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.7034761309623718,
      "learning_rate": 0.0005913287958090836,
      "loss": 2.3873,
      "step": 17676
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6467084884643555,
      "learning_rate": 0.0005913278194085463,
      "loss": 2.3985,
      "step": 17677
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6096858978271484,
      "learning_rate": 0.0005913268429538457,
      "loss": 2.3906,
      "step": 17678
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.0995980501174927,
      "learning_rate": 0.000591325866444982,
      "loss": 2.2626,
      "step": 17679
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.7705328464508057,
      "learning_rate": 0.0005913248898819553,
      "loss": 2.2463,
      "step": 17680
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6448100209236145,
      "learning_rate": 0.0005913239132647657,
      "loss": 2.2667,
      "step": 17681
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.8835278749465942,
      "learning_rate": 0.0005913229365934136,
      "loss": 2.4316,
      "step": 17682
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6467980742454529,
      "learning_rate": 0.000591321959867899,
      "loss": 2.5614,
      "step": 17683
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.5551620125770569,
      "learning_rate": 0.0005913209830882222,
      "loss": 2.301,
      "step": 17684
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.685693085193634,
      "learning_rate": 0.0005913200062543833,
      "loss": 2.5451,
      "step": 17685
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.8711903095245361,
      "learning_rate": 0.0005913190293663825,
      "loss": 2.4354,
      "step": 17686
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.648961067199707,
      "learning_rate": 0.0005913180524242201,
      "loss": 2.4196,
      "step": 17687
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6051490902900696,
      "learning_rate": 0.000591317075427896,
      "loss": 2.5037,
      "step": 17688
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.7283118963241577,
      "learning_rate": 0.0005913160983774106,
      "loss": 2.4734,
      "step": 17689
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6273114681243896,
      "learning_rate": 0.0005913151212727642,
      "loss": 2.2967,
      "step": 17690
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6247866153717041,
      "learning_rate": 0.0005913141441139567,
      "loss": 2.3324,
      "step": 17691
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.7416406273841858,
      "learning_rate": 0.0005913131669009885,
      "loss": 2.3555,
      "step": 17692
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6319222450256348,
      "learning_rate": 0.0005913121896338595,
      "loss": 2.4882,
      "step": 17693
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6125388741493225,
      "learning_rate": 0.0005913112123125702,
      "loss": 2.4982,
      "step": 17694
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6273959875106812,
      "learning_rate": 0.0005913102349371205,
      "loss": 2.3621,
      "step": 17695
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6443325281143188,
      "learning_rate": 0.0005913092575075108,
      "loss": 2.414,
      "step": 17696
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6297797560691833,
      "learning_rate": 0.0005913082800237413,
      "loss": 2.3537,
      "step": 17697
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.5879702568054199,
      "learning_rate": 0.000591307302485812,
      "loss": 2.4573,
      "step": 17698
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6212419867515564,
      "learning_rate": 0.0005913063248937232,
      "loss": 2.3443,
      "step": 17699
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6978178024291992,
      "learning_rate": 0.000591305347247475,
      "loss": 2.5251,
      "step": 17700
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6135774850845337,
      "learning_rate": 0.0005913043695470677,
      "loss": 2.3455,
      "step": 17701
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6262553930282593,
      "learning_rate": 0.0005913033917925015,
      "loss": 2.4634,
      "step": 17702
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6246405839920044,
      "learning_rate": 0.0005913024139837764,
      "loss": 2.3914,
      "step": 17703
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.5754877328872681,
      "learning_rate": 0.0005913014361208927,
      "loss": 2.4658,
      "step": 17704
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6083065867424011,
      "learning_rate": 0.0005913004582038505,
      "loss": 2.4759,
      "step": 17705
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.5973026156425476,
      "learning_rate": 0.0005912994802326502,
      "loss": 2.2689,
      "step": 17706
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.5480842590332031,
      "learning_rate": 0.0005912985022072917,
      "loss": 2.1016,
      "step": 17707
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.5809306502342224,
      "learning_rate": 0.0005912975241277754,
      "loss": 2.3286,
      "step": 17708
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.5801264047622681,
      "learning_rate": 0.0005912965459941013,
      "loss": 2.6113,
      "step": 17709
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.5558379292488098,
      "learning_rate": 0.0005912955678062697,
      "loss": 2.3651,
      "step": 17710
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6586349010467529,
      "learning_rate": 0.0005912945895642808,
      "loss": 2.2663,
      "step": 17711
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.661741316318512,
      "learning_rate": 0.0005912936112681347,
      "loss": 2.4823,
      "step": 17712
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.7562384009361267,
      "learning_rate": 0.0005912926329178316,
      "loss": 2.3571,
      "step": 17713
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6283273100852966,
      "learning_rate": 0.0005912916545133718,
      "loss": 2.4319,
      "step": 17714
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.627748966217041,
      "learning_rate": 0.0005912906760547553,
      "loss": 2.2849,
      "step": 17715
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6679596900939941,
      "learning_rate": 0.0005912896975419823,
      "loss": 2.2505,
      "step": 17716
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6956193447113037,
      "learning_rate": 0.0005912887189750531,
      "loss": 2.3898,
      "step": 17717
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6831051111221313,
      "learning_rate": 0.0005912877403539679,
      "loss": 2.4962,
      "step": 17718
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6142656803131104,
      "learning_rate": 0.0005912867616787268,
      "loss": 2.3237,
      "step": 17719
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.7682567834854126,
      "learning_rate": 0.0005912857829493299,
      "loss": 2.4541,
      "step": 17720
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6045519113540649,
      "learning_rate": 0.0005912848041657775,
      "loss": 2.2743,
      "step": 17721
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6303843259811401,
      "learning_rate": 0.0005912838253280698,
      "loss": 2.2637,
      "step": 17722
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6463274359703064,
      "learning_rate": 0.0005912828464362069,
      "loss": 2.2448,
      "step": 17723
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6165431141853333,
      "learning_rate": 0.0005912818674901891,
      "loss": 2.3425,
      "step": 17724
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6088429689407349,
      "learning_rate": 0.0005912808884900164,
      "loss": 2.343,
      "step": 17725
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6238028407096863,
      "learning_rate": 0.0005912799094356891,
      "loss": 2.3447,
      "step": 17726
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6010552048683167,
      "learning_rate": 0.0005912789303272074,
      "loss": 2.1863,
      "step": 17727
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6200903654098511,
      "learning_rate": 0.0005912779511645714,
      "loss": 2.5893,
      "step": 17728
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.5318120121955872,
      "learning_rate": 0.0005912769719477814,
      "loss": 2.2896,
      "step": 17729
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.5761203765869141,
      "learning_rate": 0.0005912759926768376,
      "loss": 2.2824,
      "step": 17730
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6548786759376526,
      "learning_rate": 0.0005912750133517399,
      "loss": 2.5149,
      "step": 17731
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.588616132736206,
      "learning_rate": 0.0005912740339724888,
      "loss": 2.3244,
      "step": 17732
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.5818692445755005,
      "learning_rate": 0.0005912730545390842,
      "loss": 2.1077,
      "step": 17733
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6057912707328796,
      "learning_rate": 0.0005912720750515266,
      "loss": 2.25,
      "step": 17734
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6068164706230164,
      "learning_rate": 0.000591271095509816,
      "loss": 2.3304,
      "step": 17735
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6166127324104309,
      "learning_rate": 0.0005912701159139527,
      "loss": 2.3053,
      "step": 17736
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.5592964291572571,
      "learning_rate": 0.0005912691362639366,
      "loss": 2.1939,
      "step": 17737
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6641290187835693,
      "learning_rate": 0.0005912681565597681,
      "loss": 2.581,
      "step": 17738
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6133838295936584,
      "learning_rate": 0.0005912671768014474,
      "loss": 2.2845,
      "step": 17739
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.679857611656189,
      "learning_rate": 0.0005912661969889746,
      "loss": 2.3759,
      "step": 17740
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6577156186103821,
      "learning_rate": 0.0005912652171223499,
      "loss": 2.1889,
      "step": 17741
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.5828139185905457,
      "learning_rate": 0.0005912642372015736,
      "loss": 2.3231,
      "step": 17742
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6146567463874817,
      "learning_rate": 0.0005912632572266458,
      "loss": 2.2923,
      "step": 17743
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6619553565979004,
      "learning_rate": 0.0005912622771975665,
      "loss": 2.3385,
      "step": 17744
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.5732108950614929,
      "learning_rate": 0.0005912612971143362,
      "loss": 2.7363,
      "step": 17745
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6052930951118469,
      "learning_rate": 0.0005912603169769548,
      "loss": 2.4742,
      "step": 17746
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.7088416218757629,
      "learning_rate": 0.0005912593367854227,
      "loss": 2.3483,
      "step": 17747
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.7184180617332458,
      "learning_rate": 0.00059125835653974,
      "loss": 2.2894,
      "step": 17748
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.5984380841255188,
      "learning_rate": 0.0005912573762399068,
      "loss": 2.4926,
      "step": 17749
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.7291138768196106,
      "learning_rate": 0.0005912563958859233,
      "loss": 2.5995,
      "step": 17750
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6905052661895752,
      "learning_rate": 0.0005912554154777898,
      "loss": 2.2287,
      "step": 17751
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6513817310333252,
      "learning_rate": 0.0005912544350155065,
      "loss": 2.4022,
      "step": 17752
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6061190366744995,
      "learning_rate": 0.0005912534544990735,
      "loss": 2.2643,
      "step": 17753
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.7284359931945801,
      "learning_rate": 0.0005912524739284909,
      "loss": 2.5547,
      "step": 17754
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.7430345416069031,
      "learning_rate": 0.000591251493303759,
      "loss": 2.2984,
      "step": 17755
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6161226034164429,
      "learning_rate": 0.000591250512624878,
      "loss": 2.3755,
      "step": 17756
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.5869682431221008,
      "learning_rate": 0.0005912495318918479,
      "loss": 2.2246,
      "step": 17757
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.751991331577301,
      "learning_rate": 0.0005912485511046691,
      "loss": 2.3519,
      "step": 17758
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.7314650416374207,
      "learning_rate": 0.0005912475702633417,
      "loss": 2.2507,
      "step": 17759
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.612493634223938,
      "learning_rate": 0.0005912465893678658,
      "loss": 2.3255,
      "step": 17760
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.7126142978668213,
      "learning_rate": 0.0005912456084182418,
      "loss": 2.3963,
      "step": 17761
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6533578038215637,
      "learning_rate": 0.0005912446274144698,
      "loss": 2.291,
      "step": 17762
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6081981658935547,
      "learning_rate": 0.0005912436463565496,
      "loss": 2.6178,
      "step": 17763
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6837480664253235,
      "learning_rate": 0.0005912426652444821,
      "loss": 2.2553,
      "step": 17764
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.7290772199630737,
      "learning_rate": 0.0005912416840782669,
      "loss": 2.2099,
      "step": 17765
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.5890096426010132,
      "learning_rate": 0.0005912407028579043,
      "loss": 2.1725,
      "step": 17766
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.5792267322540283,
      "learning_rate": 0.0005912397215833946,
      "loss": 2.2593,
      "step": 17767
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.5756087899208069,
      "learning_rate": 0.000591238740254738,
      "loss": 2.3259,
      "step": 17768
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6171374917030334,
      "learning_rate": 0.0005912377588719346,
      "loss": 2.2799,
      "step": 17769
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6125362515449524,
      "learning_rate": 0.0005912367774349845,
      "loss": 2.297,
      "step": 17770
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.7420734763145447,
      "learning_rate": 0.0005912357959438881,
      "loss": 2.1943,
      "step": 17771
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.593573808670044,
      "learning_rate": 0.0005912348143986456,
      "loss": 2.4555,
      "step": 17772
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.5994219779968262,
      "learning_rate": 0.0005912338327992568,
      "loss": 2.5225,
      "step": 17773
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6308886408805847,
      "learning_rate": 0.0005912328511457223,
      "loss": 2.443,
      "step": 17774
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.619705319404602,
      "learning_rate": 0.000591231869438042,
      "loss": 2.3591,
      "step": 17775
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6056683659553528,
      "learning_rate": 0.0005912308876762162,
      "loss": 2.5717,
      "step": 17776
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.587554931640625,
      "learning_rate": 0.0005912299058602451,
      "loss": 2.1987,
      "step": 17777
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6153813004493713,
      "learning_rate": 0.0005912289239901289,
      "loss": 2.3744,
      "step": 17778
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6425881385803223,
      "learning_rate": 0.0005912279420658677,
      "loss": 2.3372,
      "step": 17779
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.5793973803520203,
      "learning_rate": 0.0005912269600874618,
      "loss": 2.2319,
      "step": 17780
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.590889573097229,
      "learning_rate": 0.0005912259780549113,
      "loss": 2.5422,
      "step": 17781
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6830265522003174,
      "learning_rate": 0.0005912249959682163,
      "loss": 2.3176,
      "step": 17782
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.653226912021637,
      "learning_rate": 0.0005912240138273772,
      "loss": 2.3407,
      "step": 17783
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6035827994346619,
      "learning_rate": 0.000591223031632394,
      "loss": 2.2427,
      "step": 17784
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6188779473304749,
      "learning_rate": 0.000591222049383267,
      "loss": 2.7073,
      "step": 17785
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.5895932912826538,
      "learning_rate": 0.0005912210670799962,
      "loss": 2.3131,
      "step": 17786
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.7906530499458313,
      "learning_rate": 0.000591220084722582,
      "loss": 2.2694,
      "step": 17787
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6379029750823975,
      "learning_rate": 0.0005912191023110244,
      "loss": 2.3229,
      "step": 17788
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6279810667037964,
      "learning_rate": 0.0005912181198453238,
      "loss": 2.4372,
      "step": 17789
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.8300719261169434,
      "learning_rate": 0.0005912171373254802,
      "loss": 2.3713,
      "step": 17790
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.5885128378868103,
      "learning_rate": 0.0005912161547514937,
      "loss": 2.5218,
      "step": 17791
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.7237515449523926,
      "learning_rate": 0.0005912151721233648,
      "loss": 2.4281,
      "step": 17792
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6032415628433228,
      "learning_rate": 0.0005912141894410934,
      "loss": 2.2816,
      "step": 17793
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.7014862298965454,
      "learning_rate": 0.0005912132067046799,
      "loss": 2.3894,
      "step": 17794
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6016561985015869,
      "learning_rate": 0.0005912122239141243,
      "loss": 2.3242,
      "step": 17795
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6461367607116699,
      "learning_rate": 0.0005912112410694268,
      "loss": 2.4363,
      "step": 17796
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.740627646446228,
      "learning_rate": 0.0005912102581705877,
      "loss": 2.2692,
      "step": 17797
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.7938846945762634,
      "learning_rate": 0.000591209275217607,
      "loss": 2.2156,
      "step": 17798
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.7134242057800293,
      "learning_rate": 0.0005912082922104852,
      "loss": 2.2954,
      "step": 17799
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.7210978269577026,
      "learning_rate": 0.0005912073091492221,
      "loss": 2.2122,
      "step": 17800
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.9493202567100525,
      "learning_rate": 0.0005912063260338183,
      "loss": 2.5156,
      "step": 17801
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6442264914512634,
      "learning_rate": 0.0005912053428642736,
      "loss": 2.5569,
      "step": 17802
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6299258470535278,
      "learning_rate": 0.0005912043596405883,
      "loss": 2.3107,
      "step": 17803
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.8370144963264465,
      "learning_rate": 0.0005912033763627626,
      "loss": 2.2306,
      "step": 17804
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.9809398651123047,
      "learning_rate": 0.0005912023930307967,
      "loss": 2.5656,
      "step": 17805
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.5665794014930725,
      "learning_rate": 0.0005912014096446907,
      "loss": 2.3957,
      "step": 17806
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6552757620811462,
      "learning_rate": 0.0005912004262044451,
      "loss": 2.2098,
      "step": 17807
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.8640333414077759,
      "learning_rate": 0.0005911994427100596,
      "loss": 2.2628,
      "step": 17808
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.5999638438224792,
      "learning_rate": 0.0005911984591615348,
      "loss": 2.5988,
      "step": 17809
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6461067795753479,
      "learning_rate": 0.0005911974755588706,
      "loss": 2.767,
      "step": 17810
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.7488369941711426,
      "learning_rate": 0.0005911964919020674,
      "loss": 2.3994,
      "step": 17811
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6143149137496948,
      "learning_rate": 0.0005911955081911251,
      "loss": 2.5638,
      "step": 17812
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6560420989990234,
      "learning_rate": 0.0005911945244260442,
      "loss": 2.442,
      "step": 17813
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6041389107704163,
      "learning_rate": 0.0005911935406068248,
      "loss": 2.2964,
      "step": 17814
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6761433482170105,
      "learning_rate": 0.0005911925567334669,
      "loss": 2.2506,
      "step": 17815
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6183937191963196,
      "learning_rate": 0.0005911915728059709,
      "loss": 2.3438,
      "step": 17816
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6225142478942871,
      "learning_rate": 0.0005911905888243368,
      "loss": 2.3326,
      "step": 17817
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6115429997444153,
      "learning_rate": 0.0005911896047885649,
      "loss": 2.4417,
      "step": 17818
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6921790242195129,
      "learning_rate": 0.0005911886206986553,
      "loss": 2.4209,
      "step": 17819
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6604570746421814,
      "learning_rate": 0.0005911876365546084,
      "loss": 2.3371,
      "step": 17820
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6395418047904968,
      "learning_rate": 0.0005911866523564241,
      "loss": 2.4941,
      "step": 17821
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.0564498901367188,
      "learning_rate": 0.0005911856681041027,
      "loss": 2.453,
      "step": 17822
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6674873232841492,
      "learning_rate": 0.0005911846837976444,
      "loss": 2.3292,
      "step": 17823
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6611958146095276,
      "learning_rate": 0.0005911836994370494,
      "loss": 2.4745,
      "step": 17824
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.7210325002670288,
      "learning_rate": 0.0005911827150223179,
      "loss": 2.426,
      "step": 17825
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6187616586685181,
      "learning_rate": 0.00059118173055345,
      "loss": 2.3282,
      "step": 17826
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6706961393356323,
      "learning_rate": 0.0005911807460304458,
      "loss": 2.0158,
      "step": 17827
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6143769025802612,
      "learning_rate": 0.0005911797614533057,
      "loss": 2.2482,
      "step": 17828
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.7009199261665344,
      "learning_rate": 0.0005911787768220298,
      "loss": 2.571,
      "step": 17829
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6566109657287598,
      "learning_rate": 0.0005911777921366183,
      "loss": 2.4506,
      "step": 17830
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6632504463195801,
      "learning_rate": 0.0005911768073970713,
      "loss": 2.4843,
      "step": 17831
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6195100545883179,
      "learning_rate": 0.000591175822603389,
      "loss": 2.3392,
      "step": 17832
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.668890655040741,
      "learning_rate": 0.0005911748377555717,
      "loss": 2.1623,
      "step": 17833
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.7036476135253906,
      "learning_rate": 0.0005911738528536194,
      "loss": 2.2,
      "step": 17834
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.5997345447540283,
      "learning_rate": 0.0005911728678975324,
      "loss": 2.4404,
      "step": 17835
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6149289011955261,
      "learning_rate": 0.000591171882887311,
      "loss": 2.5452,
      "step": 17836
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.5749427676200867,
      "learning_rate": 0.0005911708978229552,
      "loss": 2.4894,
      "step": 17837
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6959632039070129,
      "learning_rate": 0.0005911699127044651,
      "loss": 2.2408,
      "step": 17838
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6682748198509216,
      "learning_rate": 0.0005911689275318411,
      "loss": 2.5032,
      "step": 17839
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.59041827917099,
      "learning_rate": 0.0005911679423050834,
      "loss": 2.5546,
      "step": 17840
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6461611986160278,
      "learning_rate": 0.000591166957024192,
      "loss": 2.1461,
      "step": 17841
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6130921840667725,
      "learning_rate": 0.0005911659716891671,
      "loss": 2.3624,
      "step": 17842
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6333284974098206,
      "learning_rate": 0.000591164986300009,
      "loss": 2.2723,
      "step": 17843
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.5939963459968567,
      "learning_rate": 0.0005911640008567179,
      "loss": 2.3907,
      "step": 17844
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.5901204943656921,
      "learning_rate": 0.0005911630153592938,
      "loss": 2.4598,
      "step": 17845
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.5657439231872559,
      "learning_rate": 0.0005911620298077371,
      "loss": 2.5724,
      "step": 17846
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.5844074487686157,
      "learning_rate": 0.0005911610442020479,
      "loss": 2.2825,
      "step": 17847
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6437111496925354,
      "learning_rate": 0.0005911600585422264,
      "loss": 2.378,
      "step": 17848
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6567716598510742,
      "learning_rate": 0.0005911590728282726,
      "loss": 2.3372,
      "step": 17849
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6028341054916382,
      "learning_rate": 0.0005911580870601869,
      "loss": 2.1567,
      "step": 17850
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6412850618362427,
      "learning_rate": 0.0005911571012379694,
      "loss": 2.3899,
      "step": 17851
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6581328511238098,
      "learning_rate": 0.0005911561153616203,
      "loss": 2.2628,
      "step": 17852
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6050542593002319,
      "learning_rate": 0.0005911551294311397,
      "loss": 2.323,
      "step": 17853
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.569459080696106,
      "learning_rate": 0.0005911541434465281,
      "loss": 2.1432,
      "step": 17854
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.649002194404602,
      "learning_rate": 0.0005911531574077852,
      "loss": 2.348,
      "step": 17855
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.5748676657676697,
      "learning_rate": 0.0005911521713149115,
      "loss": 2.4389,
      "step": 17856
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.5971648097038269,
      "learning_rate": 0.0005911511851679072,
      "loss": 2.3606,
      "step": 17857
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.704221785068512,
      "learning_rate": 0.0005911501989667723,
      "loss": 2.4564,
      "step": 17858
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.5958907604217529,
      "learning_rate": 0.0005911492127115072,
      "loss": 2.4451,
      "step": 17859
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.5738261938095093,
      "learning_rate": 0.0005911482264021119,
      "loss": 2.6062,
      "step": 17860
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6560497879981995,
      "learning_rate": 0.0005911472400385866,
      "loss": 2.3681,
      "step": 17861
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6039591431617737,
      "learning_rate": 0.0005911462536209316,
      "loss": 2.4372,
      "step": 17862
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6126328110694885,
      "learning_rate": 0.000591145267149147,
      "loss": 2.3131,
      "step": 17863
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.5985059142112732,
      "learning_rate": 0.000591144280623233,
      "loss": 2.4126,
      "step": 17864
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.7492591738700867,
      "learning_rate": 0.0005911432940431898,
      "loss": 2.421,
      "step": 17865
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6504553556442261,
      "learning_rate": 0.0005911423074090175,
      "loss": 2.3154,
      "step": 17866
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.5699146389961243,
      "learning_rate": 0.0005911413207207164,
      "loss": 2.4873,
      "step": 17867
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.5929052829742432,
      "learning_rate": 0.0005911403339782866,
      "loss": 2.3745,
      "step": 17868
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.5848796367645264,
      "learning_rate": 0.0005911393471817284,
      "loss": 2.5147,
      "step": 17869
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6372538208961487,
      "learning_rate": 0.0005911383603310418,
      "loss": 2.5889,
      "step": 17870
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.649115264415741,
      "learning_rate": 0.0005911373734262272,
      "loss": 2.1869,
      "step": 17871
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6811822056770325,
      "learning_rate": 0.0005911363864672845,
      "loss": 2.3008,
      "step": 17872
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.5307469964027405,
      "learning_rate": 0.0005911353994542142,
      "loss": 2.3592,
      "step": 17873
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6054986715316772,
      "learning_rate": 0.0005911344123870162,
      "loss": 2.5116,
      "step": 17874
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6159423589706421,
      "learning_rate": 0.0005911334252656909,
      "loss": 2.5861,
      "step": 17875
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6727316975593567,
      "learning_rate": 0.0005911324380902384,
      "loss": 2.4202,
      "step": 17876
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.5885460376739502,
      "learning_rate": 0.0005911314508606589,
      "loss": 2.1625,
      "step": 17877
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6094031929969788,
      "learning_rate": 0.0005911304635769526,
      "loss": 2.2977,
      "step": 17878
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6161205172538757,
      "learning_rate": 0.0005911294762391196,
      "loss": 2.1477,
      "step": 17879
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6049507856369019,
      "learning_rate": 0.00059112848884716,
      "loss": 2.2797,
      "step": 17880
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.602144181728363,
      "learning_rate": 0.0005911275014010743,
      "loss": 2.1535,
      "step": 17881
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6621884107589722,
      "learning_rate": 0.0005911265139008624,
      "loss": 2.3463,
      "step": 17882
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6470181941986084,
      "learning_rate": 0.0005911255263465247,
      "loss": 2.2711,
      "step": 17883
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6243168115615845,
      "learning_rate": 0.0005911245387380613,
      "loss": 2.5211,
      "step": 17884
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6642001271247864,
      "learning_rate": 0.0005911235510754722,
      "loss": 2.4228,
      "step": 17885
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.5992788076400757,
      "learning_rate": 0.0005911225633587578,
      "loss": 2.2777,
      "step": 17886
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6405757069587708,
      "learning_rate": 0.0005911215755879182,
      "loss": 2.2519,
      "step": 17887
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.7395951151847839,
      "learning_rate": 0.0005911205877629536,
      "loss": 2.5655,
      "step": 17888
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.5691306591033936,
      "learning_rate": 0.0005911195998838641,
      "loss": 2.1215,
      "step": 17889
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.7157325148582458,
      "learning_rate": 0.0005911186119506502,
      "loss": 2.2193,
      "step": 17890
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6684571504592896,
      "learning_rate": 0.0005911176239633116,
      "loss": 2.2685,
      "step": 17891
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6680248379707336,
      "learning_rate": 0.000591116635921849,
      "loss": 2.5231,
      "step": 17892
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.703326404094696,
      "learning_rate": 0.0005911156478262622,
      "loss": 2.3612,
      "step": 17893
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6517661213874817,
      "learning_rate": 0.0005911146596765515,
      "loss": 2.4817,
      "step": 17894
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.541787326335907,
      "learning_rate": 0.000591113671472717,
      "loss": 2.5213,
      "step": 17895
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6101418137550354,
      "learning_rate": 0.0005911126832147592,
      "loss": 2.3093,
      "step": 17896
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.779268741607666,
      "learning_rate": 0.0005911116949026779,
      "loss": 2.2427,
      "step": 17897
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6181201934814453,
      "learning_rate": 0.0005911107065364734,
      "loss": 2.6637,
      "step": 17898
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6513669490814209,
      "learning_rate": 0.0005911097181161459,
      "loss": 2.4395,
      "step": 17899
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6722812056541443,
      "learning_rate": 0.0005911087296416958,
      "loss": 2.5222,
      "step": 17900
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6909682154655457,
      "learning_rate": 0.0005911077411131229,
      "loss": 2.3134,
      "step": 17901
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6989681720733643,
      "learning_rate": 0.0005911067525304277,
      "loss": 2.3247,
      "step": 17902
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6460075974464417,
      "learning_rate": 0.0005911057638936102,
      "loss": 2.2944,
      "step": 17903
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.5603254437446594,
      "learning_rate": 0.0005911047752026707,
      "loss": 2.1947,
      "step": 17904
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.5865738391876221,
      "learning_rate": 0.0005911037864576094,
      "loss": 2.4258,
      "step": 17905
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6337721347808838,
      "learning_rate": 0.0005911027976584262,
      "loss": 2.2594,
      "step": 17906
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.5601739883422852,
      "learning_rate": 0.0005911018088051215,
      "loss": 2.2628,
      "step": 17907
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.5736044049263,
      "learning_rate": 0.0005911008198976956,
      "loss": 2.1682,
      "step": 17908
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6135255098342896,
      "learning_rate": 0.0005910998309361485,
      "loss": 2.4429,
      "step": 17909
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.5894288420677185,
      "learning_rate": 0.0005910988419204806,
      "loss": 2.4382,
      "step": 17910
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.5971383452415466,
      "learning_rate": 0.0005910978528506918,
      "loss": 2.4957,
      "step": 17911
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6813445091247559,
      "learning_rate": 0.0005910968637267824,
      "loss": 2.3511,
      "step": 17912
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6332822442054749,
      "learning_rate": 0.0005910958745487527,
      "loss": 2.4207,
      "step": 17913
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.653590738773346,
      "learning_rate": 0.0005910948853166027,
      "loss": 2.4574,
      "step": 17914
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6344131231307983,
      "learning_rate": 0.0005910938960303326,
      "loss": 2.3879,
      "step": 17915
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.5841922760009766,
      "learning_rate": 0.0005910929066899427,
      "loss": 2.2549,
      "step": 17916
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.607796311378479,
      "learning_rate": 0.0005910919172954332,
      "loss": 2.3219,
      "step": 17917
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6831410527229309,
      "learning_rate": 0.0005910909278468042,
      "loss": 2.3947,
      "step": 17918
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6023773550987244,
      "learning_rate": 0.0005910899383440558,
      "loss": 2.4325,
      "step": 17919
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6354015469551086,
      "learning_rate": 0.0005910889487871884,
      "loss": 2.3357,
      "step": 17920
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6580221652984619,
      "learning_rate": 0.000591087959176202,
      "loss": 2.6886,
      "step": 17921
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6772993803024292,
      "learning_rate": 0.0005910869695110969,
      "loss": 2.3053,
      "step": 17922
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6178650259971619,
      "learning_rate": 0.0005910859797918733,
      "loss": 2.4529,
      "step": 17923
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6440972685813904,
      "learning_rate": 0.0005910849900185312,
      "loss": 2.3571,
      "step": 17924
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.7334065437316895,
      "learning_rate": 0.0005910840001910709,
      "loss": 2.3929,
      "step": 17925
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.5606977939605713,
      "learning_rate": 0.0005910830103094927,
      "loss": 2.3251,
      "step": 17926
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6577188968658447,
      "learning_rate": 0.0005910820203737966,
      "loss": 2.3111,
      "step": 17927
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6052924394607544,
      "learning_rate": 0.0005910810303839829,
      "loss": 2.4652,
      "step": 17928
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.5844426155090332,
      "learning_rate": 0.0005910800403400517,
      "loss": 2.2539,
      "step": 17929
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.5706654191017151,
      "learning_rate": 0.0005910790502420033,
      "loss": 2.3944,
      "step": 17930
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.5851973295211792,
      "learning_rate": 0.0005910780600898377,
      "loss": 2.3513,
      "step": 17931
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.7014071941375732,
      "learning_rate": 0.0005910770698835552,
      "loss": 2.3706,
      "step": 17932
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6222972273826599,
      "learning_rate": 0.000591076079623156,
      "loss": 2.2386,
      "step": 17933
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6063003540039062,
      "learning_rate": 0.0005910750893086403,
      "loss": 2.4572,
      "step": 17934
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6657074689865112,
      "learning_rate": 0.0005910740989400082,
      "loss": 2.4262,
      "step": 17935
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.5348498225212097,
      "learning_rate": 0.0005910731085172601,
      "loss": 2.248,
      "step": 17936
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6513791084289551,
      "learning_rate": 0.0005910721180403958,
      "loss": 2.298,
      "step": 17937
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6795707941055298,
      "learning_rate": 0.0005910711275094158,
      "loss": 2.461,
      "step": 17938
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6258490085601807,
      "learning_rate": 0.0005910701369243202,
      "loss": 2.1195,
      "step": 17939
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6636934876441956,
      "learning_rate": 0.0005910691462851091,
      "loss": 2.202,
      "step": 17940
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6751032471656799,
      "learning_rate": 0.0005910681555917827,
      "loss": 2.3142,
      "step": 17941
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.5735987424850464,
      "learning_rate": 0.0005910671648443413,
      "loss": 2.3799,
      "step": 17942
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.7602788209915161,
      "learning_rate": 0.0005910661740427851,
      "loss": 2.4943,
      "step": 17943
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6466507315635681,
      "learning_rate": 0.0005910651831871141,
      "loss": 2.3989,
      "step": 17944
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6933383345603943,
      "learning_rate": 0.0005910641922773286,
      "loss": 2.5347,
      "step": 17945
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6270187497138977,
      "learning_rate": 0.0005910632013134289,
      "loss": 2.4049,
      "step": 17946
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.5752281546592712,
      "learning_rate": 0.0005910622102954149,
      "loss": 2.4021,
      "step": 17947
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6033563613891602,
      "learning_rate": 0.000591061219223287,
      "loss": 2.3491,
      "step": 17948
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6063661575317383,
      "learning_rate": 0.0005910602280970453,
      "loss": 2.4243,
      "step": 17949
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6863558888435364,
      "learning_rate": 0.00059105923691669,
      "loss": 2.4064,
      "step": 17950
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.595049262046814,
      "learning_rate": 0.0005910582456822213,
      "loss": 2.2637,
      "step": 17951
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6453537940979004,
      "learning_rate": 0.0005910572543936395,
      "loss": 2.3971,
      "step": 17952
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6484963297843933,
      "learning_rate": 0.0005910562630509445,
      "loss": 2.4514,
      "step": 17953
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.5660973787307739,
      "learning_rate": 0.0005910552716541367,
      "loss": 1.9905,
      "step": 17954
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.5950981974601746,
      "learning_rate": 0.0005910542802032164,
      "loss": 2.1714,
      "step": 17955
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6450952291488647,
      "learning_rate": 0.0005910532886981833,
      "loss": 2.4426,
      "step": 17956
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.5604104399681091,
      "learning_rate": 0.0005910522971390381,
      "loss": 2.4302,
      "step": 17957
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.5790059566497803,
      "learning_rate": 0.0005910513055257808,
      "loss": 2.1575,
      "step": 17958
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.5828315019607544,
      "learning_rate": 0.0005910503138584115,
      "loss": 2.5559,
      "step": 17959
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6056738495826721,
      "learning_rate": 0.0005910493221369304,
      "loss": 2.2515,
      "step": 17960
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6468691825866699,
      "learning_rate": 0.0005910483303613378,
      "loss": 2.3945,
      "step": 17961
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6301375031471252,
      "learning_rate": 0.0005910473385316339,
      "loss": 2.4497,
      "step": 17962
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.5890749096870422,
      "learning_rate": 0.0005910463466478187,
      "loss": 2.442,
      "step": 17963
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6449437737464905,
      "learning_rate": 0.0005910453547098924,
      "loss": 2.4557,
      "step": 17964
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6026979088783264,
      "learning_rate": 0.0005910443627178554,
      "loss": 2.1181,
      "step": 17965
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6192470788955688,
      "learning_rate": 0.0005910433706717077,
      "loss": 2.5151,
      "step": 17966
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.7035093903541565,
      "learning_rate": 0.0005910423785714496,
      "loss": 2.3212,
      "step": 17967
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.5826371908187866,
      "learning_rate": 0.0005910413864170813,
      "loss": 2.4226,
      "step": 17968
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.5781599283218384,
      "learning_rate": 0.0005910403942086029,
      "loss": 2.4605,
      "step": 17969
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6179317831993103,
      "learning_rate": 0.0005910394019460145,
      "loss": 2.3762,
      "step": 17970
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6771470308303833,
      "learning_rate": 0.0005910384096293164,
      "loss": 2.4047,
      "step": 17971
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6068975925445557,
      "learning_rate": 0.0005910374172585088,
      "loss": 2.3363,
      "step": 17972
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6228682994842529,
      "learning_rate": 0.0005910364248335918,
      "loss": 2.7106,
      "step": 17973
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.5918586850166321,
      "learning_rate": 0.0005910354323545656,
      "loss": 2.3143,
      "step": 17974
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.630889356136322,
      "learning_rate": 0.0005910344398214306,
      "loss": 2.4787,
      "step": 17975
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.5469934344291687,
      "learning_rate": 0.0005910334472341866,
      "loss": 2.3993,
      "step": 17976
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.5910034775733948,
      "learning_rate": 0.0005910324545928341,
      "loss": 2.3507,
      "step": 17977
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6289666891098022,
      "learning_rate": 0.0005910314618973731,
      "loss": 2.2363,
      "step": 17978
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.644167423248291,
      "learning_rate": 0.0005910304691478039,
      "loss": 2.3492,
      "step": 17979
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.645679235458374,
      "learning_rate": 0.0005910294763441267,
      "loss": 2.5204,
      "step": 17980
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6484468579292297,
      "learning_rate": 0.0005910284834863415,
      "loss": 2.1293,
      "step": 17981
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.714271068572998,
      "learning_rate": 0.0005910274905744487,
      "loss": 2.3715,
      "step": 17982
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6842129826545715,
      "learning_rate": 0.0005910264976084484,
      "loss": 2.3881,
      "step": 17983
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.7275351881980896,
      "learning_rate": 0.0005910255045883407,
      "loss": 2.1513,
      "step": 17984
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6170468926429749,
      "learning_rate": 0.000591024511514126,
      "loss": 2.4784,
      "step": 17985
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6570940017700195,
      "learning_rate": 0.0005910235183858043,
      "loss": 2.444,
      "step": 17986
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.7248684167861938,
      "learning_rate": 0.0005910225252033758,
      "loss": 2.3925,
      "step": 17987
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.8475263118743896,
      "learning_rate": 0.0005910215319668408,
      "loss": 2.2366,
      "step": 17988
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6248179078102112,
      "learning_rate": 0.0005910205386761993,
      "loss": 2.3589,
      "step": 17989
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6445391774177551,
      "learning_rate": 0.0005910195453314516,
      "loss": 2.3023,
      "step": 17990
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.8264738917350769,
      "learning_rate": 0.0005910185519325978,
      "loss": 2.549,
      "step": 17991
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6128003597259521,
      "learning_rate": 0.0005910175584796383,
      "loss": 2.2539,
      "step": 17992
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.5922605395317078,
      "learning_rate": 0.0005910165649725732,
      "loss": 2.4423,
      "step": 17993
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6757247447967529,
      "learning_rate": 0.0005910155714114024,
      "loss": 2.5502,
      "step": 17994
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.5775583386421204,
      "learning_rate": 0.0005910145777961265,
      "loss": 2.3865,
      "step": 17995
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6272025108337402,
      "learning_rate": 0.0005910135841267454,
      "loss": 2.3241,
      "step": 17996
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.7245728969573975,
      "learning_rate": 0.0005910125904032594,
      "loss": 2.6271,
      "step": 17997
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6471807956695557,
      "learning_rate": 0.0005910115966256687,
      "loss": 2.1211,
      "step": 17998
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6348088383674622,
      "learning_rate": 0.0005910106027939735,
      "loss": 2.2962,
      "step": 17999
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6144682765007019,
      "learning_rate": 0.0005910096089081739,
      "loss": 2.4223,
      "step": 18000
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6244847178459167,
      "learning_rate": 0.00059100861496827,
      "loss": 2.2241,
      "step": 18001
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6700040698051453,
      "learning_rate": 0.0005910076209742622,
      "loss": 2.7132,
      "step": 18002
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6135087609291077,
      "learning_rate": 0.0005910066269261505,
      "loss": 2.2614,
      "step": 18003
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6303872466087341,
      "learning_rate": 0.0005910056328239353,
      "loss": 2.2353,
      "step": 18004
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6421384811401367,
      "learning_rate": 0.0005910046386676166,
      "loss": 2.2929,
      "step": 18005
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.5998193621635437,
      "learning_rate": 0.0005910036444571947,
      "loss": 2.2067,
      "step": 18006
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6649309396743774,
      "learning_rate": 0.0005910026501926696,
      "loss": 2.2918,
      "step": 18007
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6844375133514404,
      "learning_rate": 0.0005910016558740418,
      "loss": 2.4233,
      "step": 18008
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6996004581451416,
      "learning_rate": 0.0005910006615013112,
      "loss": 2.6969,
      "step": 18009
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6681561470031738,
      "learning_rate": 0.000590999667074478,
      "loss": 2.6833,
      "step": 18010
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6561343669891357,
      "learning_rate": 0.0005909986725935426,
      "loss": 2.5696,
      "step": 18011
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6383271813392639,
      "learning_rate": 0.0005909976780585051,
      "loss": 2.428,
      "step": 18012
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6807460188865662,
      "learning_rate": 0.0005909966834693655,
      "loss": 2.1377,
      "step": 18013
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6074671149253845,
      "learning_rate": 0.0005909956888261242,
      "loss": 2.2612,
      "step": 18014
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6510061621665955,
      "learning_rate": 0.0005909946941287812,
      "loss": 2.6269,
      "step": 18015
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6202317476272583,
      "learning_rate": 0.0005909936993773369,
      "loss": 2.5513,
      "step": 18016
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6350908875465393,
      "learning_rate": 0.0005909927045717914,
      "loss": 2.3832,
      "step": 18017
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.5604076981544495,
      "learning_rate": 0.0005909917097121448,
      "loss": 2.2045,
      "step": 18018
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6749951243400574,
      "learning_rate": 0.0005909907147983973,
      "loss": 2.4924,
      "step": 18019
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6615902185440063,
      "learning_rate": 0.0005909897198305492,
      "loss": 2.617,
      "step": 18020
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6367053985595703,
      "learning_rate": 0.0005909887248086007,
      "loss": 2.4428,
      "step": 18021
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.5877547264099121,
      "learning_rate": 0.0005909877297325518,
      "loss": 2.5969,
      "step": 18022
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.5945631265640259,
      "learning_rate": 0.0005909867346024027,
      "loss": 2.4613,
      "step": 18023
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6490222811698914,
      "learning_rate": 0.0005909857394181538,
      "loss": 2.4358,
      "step": 18024
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6609274744987488,
      "learning_rate": 0.0005909847441798051,
      "loss": 2.6757,
      "step": 18025
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.639406144618988,
      "learning_rate": 0.000590983748887357,
      "loss": 2.2971,
      "step": 18026
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.7081433534622192,
      "learning_rate": 0.0005909827535408094,
      "loss": 2.4239,
      "step": 18027
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6547099947929382,
      "learning_rate": 0.0005909817581401626,
      "loss": 2.4559,
      "step": 18028
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6508765816688538,
      "learning_rate": 0.0005909807626854168,
      "loss": 2.4057,
      "step": 18029
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.7018435597419739,
      "learning_rate": 0.0005909797671765722,
      "loss": 2.3517,
      "step": 18030
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6760221719741821,
      "learning_rate": 0.000590978771613629,
      "loss": 2.647,
      "step": 18031
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.62154620885849,
      "learning_rate": 0.0005909777759965873,
      "loss": 2.3874,
      "step": 18032
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6093444228172302,
      "learning_rate": 0.0005909767803254474,
      "loss": 2.3087,
      "step": 18033
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.5763214230537415,
      "learning_rate": 0.0005909757846002093,
      "loss": 2.379,
      "step": 18034
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6107524037361145,
      "learning_rate": 0.0005909747888208735,
      "loss": 2.5878,
      "step": 18035
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.5818366408348083,
      "learning_rate": 0.0005909737929874399,
      "loss": 2.3596,
      "step": 18036
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6307654976844788,
      "learning_rate": 0.0005909727970999087,
      "loss": 2.5651,
      "step": 18037
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.637433648109436,
      "learning_rate": 0.0005909718011582803,
      "loss": 2.4971,
      "step": 18038
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.5844615697860718,
      "learning_rate": 0.0005909708051625547,
      "loss": 2.3863,
      "step": 18039
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6110050082206726,
      "learning_rate": 0.0005909698091127321,
      "loss": 2.4096,
      "step": 18040
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6163103580474854,
      "learning_rate": 0.0005909688130088127,
      "loss": 2.5607,
      "step": 18041
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6409177780151367,
      "learning_rate": 0.0005909678168507968,
      "loss": 2.3118,
      "step": 18042
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6090025901794434,
      "learning_rate": 0.0005909668206386843,
      "loss": 2.435,
      "step": 18043
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6279924511909485,
      "learning_rate": 0.0005909658243724758,
      "loss": 2.51,
      "step": 18044
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6227037310600281,
      "learning_rate": 0.0005909648280521712,
      "loss": 2.4118,
      "step": 18045
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6055462956428528,
      "learning_rate": 0.0005909638316777707,
      "loss": 2.1307,
      "step": 18046
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6712727546691895,
      "learning_rate": 0.0005909628352492746,
      "loss": 2.1771,
      "step": 18047
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6858943700790405,
      "learning_rate": 0.0005909618387666829,
      "loss": 2.3581,
      "step": 18048
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.5770111680030823,
      "learning_rate": 0.000590960842229996,
      "loss": 2.3802,
      "step": 18049
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.678728461265564,
      "learning_rate": 0.000590959845639214,
      "loss": 2.5129,
      "step": 18050
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6359549164772034,
      "learning_rate": 0.000590958848994337,
      "loss": 2.5987,
      "step": 18051
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.593359112739563,
      "learning_rate": 0.0005909578522953653,
      "loss": 2.4508,
      "step": 18052
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.5626408457756042,
      "learning_rate": 0.000590956855542299,
      "loss": 2.4343,
      "step": 18053
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6353928446769714,
      "learning_rate": 0.0005909558587351383,
      "loss": 2.294,
      "step": 18054
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6971363425254822,
      "learning_rate": 0.0005909548618738835,
      "loss": 2.2927,
      "step": 18055
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6148403286933899,
      "learning_rate": 0.0005909538649585347,
      "loss": 2.4121,
      "step": 18056
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6452682614326477,
      "learning_rate": 0.000590952867989092,
      "loss": 2.6193,
      "step": 18057
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6182791590690613,
      "learning_rate": 0.0005909518709655558,
      "loss": 2.3867,
      "step": 18058
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.5501250624656677,
      "learning_rate": 0.0005909508738879261,
      "loss": 2.2908,
      "step": 18059
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.612560510635376,
      "learning_rate": 0.000590949876756203,
      "loss": 2.2997,
      "step": 18060
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.5877555012702942,
      "learning_rate": 0.000590948879570387,
      "loss": 2.395,
      "step": 18061
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.7142874598503113,
      "learning_rate": 0.0005909478823304781,
      "loss": 2.4987,
      "step": 18062
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6300989985466003,
      "learning_rate": 0.0005909468850364765,
      "loss": 2.3549,
      "step": 18063
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.599199116230011,
      "learning_rate": 0.0005909458876883823,
      "loss": 2.5466,
      "step": 18064
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6113558411598206,
      "learning_rate": 0.0005909448902861957,
      "loss": 2.3255,
      "step": 18065
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.5846195220947266,
      "learning_rate": 0.0005909438928299171,
      "loss": 2.3728,
      "step": 18066
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6520463228225708,
      "learning_rate": 0.0005909428953195465,
      "loss": 2.4933,
      "step": 18067
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6459307670593262,
      "learning_rate": 0.0005909418977550841,
      "loss": 2.4249,
      "step": 18068
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.5842866897583008,
      "learning_rate": 0.0005909409001365302,
      "loss": 2.4544,
      "step": 18069
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.5562739372253418,
      "learning_rate": 0.0005909399024638848,
      "loss": 2.3556,
      "step": 18070
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6366298198699951,
      "learning_rate": 0.0005909389047371482,
      "loss": 2.4556,
      "step": 18071
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.621160626411438,
      "learning_rate": 0.0005909379069563205,
      "loss": 2.4835,
      "step": 18072
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6588096022605896,
      "learning_rate": 0.0005909369091214021,
      "loss": 2.5726,
      "step": 18073
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6450326442718506,
      "learning_rate": 0.0005909359112323929,
      "loss": 2.2162,
      "step": 18074
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.5678673386573792,
      "learning_rate": 0.0005909349132892932,
      "loss": 2.2202,
      "step": 18075
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.5836635828018188,
      "learning_rate": 0.0005909339152921034,
      "loss": 2.4554,
      "step": 18076
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.5784475803375244,
      "learning_rate": 0.0005909329172408234,
      "loss": 2.1842,
      "step": 18077
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6374452710151672,
      "learning_rate": 0.0005909319191354535,
      "loss": 2.5743,
      "step": 18078
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.5910422801971436,
      "learning_rate": 0.0005909309209759937,
      "loss": 2.169,
      "step": 18079
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.5649372935295105,
      "learning_rate": 0.0005909299227624445,
      "loss": 2.364,
      "step": 18080
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6536467671394348,
      "learning_rate": 0.000590928924494806,
      "loss": 2.3589,
      "step": 18081
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.539480447769165,
      "learning_rate": 0.0005909279261730781,
      "loss": 2.3294,
      "step": 18082
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.7066720128059387,
      "learning_rate": 0.0005909269277972614,
      "loss": 2.4851,
      "step": 18083
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.5868262052536011,
      "learning_rate": 0.0005909259293673558,
      "loss": 2.4388,
      "step": 18084
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6298353672027588,
      "learning_rate": 0.0005909249308833617,
      "loss": 2.317,
      "step": 18085
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.7754485011100769,
      "learning_rate": 0.000590923932345279,
      "loss": 2.2869,
      "step": 18086
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6489719748497009,
      "learning_rate": 0.0005909229337531082,
      "loss": 2.4237,
      "step": 18087
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6172739267349243,
      "learning_rate": 0.0005909219351068492,
      "loss": 2.4153,
      "step": 18088
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.8924967646598816,
      "learning_rate": 0.0005909209364065023,
      "loss": 2.311,
      "step": 18089
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6648573875427246,
      "learning_rate": 0.0005909199376520678,
      "loss": 2.3646,
      "step": 18090
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6098268628120422,
      "learning_rate": 0.0005909189388435458,
      "loss": 2.2936,
      "step": 18091
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.5481922626495361,
      "learning_rate": 0.0005909179399809365,
      "loss": 2.1681,
      "step": 18092
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.7051233053207397,
      "learning_rate": 0.00059091694106424,
      "loss": 2.4131,
      "step": 18093
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6251344084739685,
      "learning_rate": 0.0005909159420934565,
      "loss": 2.4033,
      "step": 18094
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6700793504714966,
      "learning_rate": 0.0005909149430685863,
      "loss": 2.3239,
      "step": 18095
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.7975068092346191,
      "learning_rate": 0.0005909139439896296,
      "loss": 2.0428,
      "step": 18096
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.587914764881134,
      "learning_rate": 0.0005909129448565864,
      "loss": 2.4669,
      "step": 18097
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.5609033703804016,
      "learning_rate": 0.000590911945669457,
      "loss": 2.3627,
      "step": 18098
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.7515871524810791,
      "learning_rate": 0.0005909109464282416,
      "loss": 2.4246,
      "step": 18099
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6213637590408325,
      "learning_rate": 0.0005909099471329404,
      "loss": 2.3421,
      "step": 18100
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6613563895225525,
      "learning_rate": 0.0005909089477835535,
      "loss": 2.3052,
      "step": 18101
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.5753303170204163,
      "learning_rate": 0.0005909079483800812,
      "loss": 2.5458,
      "step": 18102
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6008350849151611,
      "learning_rate": 0.0005909069489225234,
      "loss": 2.4095,
      "step": 18103
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.5981447100639343,
      "learning_rate": 0.0005909059494108807,
      "loss": 2.1947,
      "step": 18104
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6490551233291626,
      "learning_rate": 0.0005909049498451531,
      "loss": 2.262,
      "step": 18105
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.5999326705932617,
      "learning_rate": 0.0005909039502253407,
      "loss": 2.2454,
      "step": 18106
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6639235019683838,
      "learning_rate": 0.0005909029505514439,
      "loss": 2.4721,
      "step": 18107
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.5798352956771851,
      "learning_rate": 0.0005909019508234626,
      "loss": 2.2755,
      "step": 18108
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.5474237203598022,
      "learning_rate": 0.0005909009510413972,
      "loss": 2.1391,
      "step": 18109
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6956933736801147,
      "learning_rate": 0.0005908999512052478,
      "loss": 2.416,
      "step": 18110
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.594027578830719,
      "learning_rate": 0.0005908989513150147,
      "loss": 2.2443,
      "step": 18111
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.656579852104187,
      "learning_rate": 0.0005908979513706979,
      "loss": 2.3458,
      "step": 18112
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6237122416496277,
      "learning_rate": 0.0005908969513722977,
      "loss": 2.3808,
      "step": 18113
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.5591349005699158,
      "learning_rate": 0.0005908959513198143,
      "loss": 2.1986,
      "step": 18114
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6100477576255798,
      "learning_rate": 0.0005908949512132477,
      "loss": 2.3733,
      "step": 18115
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.5901742577552795,
      "learning_rate": 0.0005908939510525984,
      "loss": 2.2004,
      "step": 18116
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6139922142028809,
      "learning_rate": 0.0005908929508378664,
      "loss": 2.0812,
      "step": 18117
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6571904420852661,
      "learning_rate": 0.0005908919505690519,
      "loss": 2.4429,
      "step": 18118
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.7182285785675049,
      "learning_rate": 0.000590890950246155,
      "loss": 2.5881,
      "step": 18119
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6610496044158936,
      "learning_rate": 0.0005908899498691761,
      "loss": 2.3851,
      "step": 18120
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6971520781517029,
      "learning_rate": 0.0005908889494381153,
      "loss": 2.2595,
      "step": 18121
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6714118123054504,
      "learning_rate": 0.0005908879489529726,
      "loss": 2.3376,
      "step": 18122
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6988039612770081,
      "learning_rate": 0.0005908869484137485,
      "loss": 2.6353,
      "step": 18123
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6604608297348022,
      "learning_rate": 0.0005908859478204429,
      "loss": 2.3885,
      "step": 18124
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6492508053779602,
      "learning_rate": 0.0005908849471730562,
      "loss": 2.3665,
      "step": 18125
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6320772767066956,
      "learning_rate": 0.0005908839464715884,
      "loss": 2.587,
      "step": 18126
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6809914112091064,
      "learning_rate": 0.0005908829457160399,
      "loss": 2.359,
      "step": 18127
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6379947066307068,
      "learning_rate": 0.0005908819449064106,
      "loss": 2.561,
      "step": 18128
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6982579231262207,
      "learning_rate": 0.000590880944042701,
      "loss": 2.3561,
      "step": 18129
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6886407732963562,
      "learning_rate": 0.0005908799431249112,
      "loss": 2.4532,
      "step": 18130
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.5901674628257751,
      "learning_rate": 0.0005908789421530412,
      "loss": 2.4356,
      "step": 18131
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.7824816703796387,
      "learning_rate": 0.0005908779411270914,
      "loss": 2.4563,
      "step": 18132
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.5800958871841431,
      "learning_rate": 0.0005908769400470618,
      "loss": 2.3737,
      "step": 18133
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6615361571311951,
      "learning_rate": 0.0005908759389129527,
      "loss": 2.4914,
      "step": 18134
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6676782965660095,
      "learning_rate": 0.0005908749377247644,
      "loss": 2.2783,
      "step": 18135
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6671178936958313,
      "learning_rate": 0.0005908739364824968,
      "loss": 2.2597,
      "step": 18136
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6438852548599243,
      "learning_rate": 0.0005908729351861503,
      "loss": 2.4751,
      "step": 18137
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.596739649772644,
      "learning_rate": 0.0005908719338357251,
      "loss": 2.2642,
      "step": 18138
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6753901243209839,
      "learning_rate": 0.0005908709324312212,
      "loss": 2.2343,
      "step": 18139
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6350045800209045,
      "learning_rate": 0.0005908699309726389,
      "loss": 2.4016,
      "step": 18140
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6056579947471619,
      "learning_rate": 0.0005908689294599785,
      "loss": 2.3048,
      "step": 18141
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6817612648010254,
      "learning_rate": 0.00059086792789324,
      "loss": 2.3591,
      "step": 18142
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6518334150314331,
      "learning_rate": 0.0005908669262724238,
      "loss": 2.3539,
      "step": 18143
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.5993120074272156,
      "learning_rate": 0.0005908659245975298,
      "loss": 2.2273,
      "step": 18144
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6542579531669617,
      "learning_rate": 0.0005908649228685582,
      "loss": 2.422,
      "step": 18145
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.707690417766571,
      "learning_rate": 0.0005908639210855097,
      "loss": 2.3033,
      "step": 18146
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6171280741691589,
      "learning_rate": 0.0005908629192483838,
      "loss": 2.3511,
      "step": 18147
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6637652516365051,
      "learning_rate": 0.0005908619173571811,
      "loss": 2.3291,
      "step": 18148
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6500171422958374,
      "learning_rate": 0.0005908609154119017,
      "loss": 2.4062,
      "step": 18149
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.7449961304664612,
      "learning_rate": 0.0005908599134125458,
      "loss": 2.4164,
      "step": 18150
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.7125546932220459,
      "learning_rate": 0.0005908589113591134,
      "loss": 2.3229,
      "step": 18151
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6422698497772217,
      "learning_rate": 0.000590857909251605,
      "loss": 2.4668,
      "step": 18152
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6343643665313721,
      "learning_rate": 0.0005908569070900205,
      "loss": 2.1598,
      "step": 18153
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.5577104091644287,
      "learning_rate": 0.0005908559048743603,
      "loss": 2.3277,
      "step": 18154
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6150062680244446,
      "learning_rate": 0.0005908549026046243,
      "loss": 2.1896,
      "step": 18155
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.5887977480888367,
      "learning_rate": 0.000590853900280813,
      "loss": 2.3828,
      "step": 18156
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6354933381080627,
      "learning_rate": 0.0005908528979029266,
      "loss": 2.2154,
      "step": 18157
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6307415962219238,
      "learning_rate": 0.000590851895470965,
      "loss": 2.7379,
      "step": 18158
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6199992895126343,
      "learning_rate": 0.0005908508929849287,
      "loss": 2.5314,
      "step": 18159
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6559861898422241,
      "learning_rate": 0.0005908498904448177,
      "loss": 2.1846,
      "step": 18160
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.5846691131591797,
      "learning_rate": 0.0005908488878506321,
      "loss": 2.4326,
      "step": 18161
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.5875659584999084,
      "learning_rate": 0.0005908478852023723,
      "loss": 2.254,
      "step": 18162
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6964630484580994,
      "learning_rate": 0.0005908468825000383,
      "loss": 2.4844,
      "step": 18163
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.7973552942276001,
      "learning_rate": 0.0005908458797436304,
      "loss": 2.5855,
      "step": 18164
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.5989704728126526,
      "learning_rate": 0.0005908448769331488,
      "loss": 2.4561,
      "step": 18165
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.5806506276130676,
      "learning_rate": 0.0005908438740685935,
      "loss": 2.3946,
      "step": 18166
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.5965906977653503,
      "learning_rate": 0.000590842871149965,
      "loss": 2.3296,
      "step": 18167
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6224905848503113,
      "learning_rate": 0.0005908418681772633,
      "loss": 2.2073,
      "step": 18168
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6001049280166626,
      "learning_rate": 0.0005908408651504886,
      "loss": 2.6031,
      "step": 18169
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.5975974798202515,
      "learning_rate": 0.000590839862069641,
      "loss": 2.2915,
      "step": 18170
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.5877649188041687,
      "learning_rate": 0.000590838858934721,
      "loss": 2.5541,
      "step": 18171
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6099656224250793,
      "learning_rate": 0.0005908378557457284,
      "loss": 2.564,
      "step": 18172
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6412608027458191,
      "learning_rate": 0.0005908368525026635,
      "loss": 2.5901,
      "step": 18173
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6042008399963379,
      "learning_rate": 0.0005908358492055266,
      "loss": 2.3323,
      "step": 18174
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6326178312301636,
      "learning_rate": 0.0005908348458543179,
      "loss": 2.419,
      "step": 18175
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.746001660823822,
      "learning_rate": 0.0005908338424490374,
      "loss": 2.4316,
      "step": 18176
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.575802206993103,
      "learning_rate": 0.0005908328389896855,
      "loss": 2.184,
      "step": 18177
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6179981231689453,
      "learning_rate": 0.0005908318354762622,
      "loss": 2.2001,
      "step": 18178
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.622845470905304,
      "learning_rate": 0.0005908308319087679,
      "loss": 2.3343,
      "step": 18179
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6119473576545715,
      "learning_rate": 0.0005908298282872025,
      "loss": 2.4263,
      "step": 18180
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.700046956539154,
      "learning_rate": 0.0005908288246115664,
      "loss": 2.6863,
      "step": 18181
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6070247888565063,
      "learning_rate": 0.0005908278208818597,
      "loss": 2.2301,
      "step": 18182
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6221839189529419,
      "learning_rate": 0.0005908268170980827,
      "loss": 2.507,
      "step": 18183
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6118003129959106,
      "learning_rate": 0.0005908258132602355,
      "loss": 2.4077,
      "step": 18184
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6649597883224487,
      "learning_rate": 0.0005908248093683182,
      "loss": 2.4017,
      "step": 18185
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6212201118469238,
      "learning_rate": 0.0005908238054223311,
      "loss": 2.6728,
      "step": 18186
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6375420689582825,
      "learning_rate": 0.0005908228014222744,
      "loss": 2.4636,
      "step": 18187
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6144694089889526,
      "learning_rate": 0.0005908217973681483,
      "loss": 2.3836,
      "step": 18188
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.5900467038154602,
      "learning_rate": 0.0005908207932599529,
      "loss": 2.272,
      "step": 18189
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.5885268449783325,
      "learning_rate": 0.0005908197890976884,
      "loss": 2.25,
      "step": 18190
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6145026683807373,
      "learning_rate": 0.000590818784881355,
      "loss": 2.4724,
      "step": 18191
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.5401143431663513,
      "learning_rate": 0.0005908177806109529,
      "loss": 2.3816,
      "step": 18192
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.655872106552124,
      "learning_rate": 0.0005908167762864823,
      "loss": 2.38,
      "step": 18193
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.5799531936645508,
      "learning_rate": 0.0005908157719079434,
      "loss": 2.4697,
      "step": 18194
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.5869227647781372,
      "learning_rate": 0.0005908147674753364,
      "loss": 2.2748,
      "step": 18195
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.5887656211853027,
      "learning_rate": 0.0005908137629886613,
      "loss": 2.472,
      "step": 18196
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6853396892547607,
      "learning_rate": 0.0005908127584479185,
      "loss": 2.3344,
      "step": 18197
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.7236054539680481,
      "learning_rate": 0.0005908117538531082,
      "loss": 2.2949,
      "step": 18198
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.645115077495575,
      "learning_rate": 0.0005908107492042304,
      "loss": 2.5443,
      "step": 18199
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6420386433601379,
      "learning_rate": 0.0005908097445012855,
      "loss": 2.2631,
      "step": 18200
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.8181069493293762,
      "learning_rate": 0.0005908087397442735,
      "loss": 2.3005,
      "step": 18201
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6050268411636353,
      "learning_rate": 0.0005908077349331948,
      "loss": 2.4018,
      "step": 18202
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.5873995423316956,
      "learning_rate": 0.0005908067300680492,
      "loss": 2.4478,
      "step": 18203
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6304742097854614,
      "learning_rate": 0.0005908057251488373,
      "loss": 2.036,
      "step": 18204
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6610552668571472,
      "learning_rate": 0.0005908047201755591,
      "loss": 2.3659,
      "step": 18205
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6028740406036377,
      "learning_rate": 0.0005908037151482148,
      "loss": 2.2967,
      "step": 18206
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.5685185194015503,
      "learning_rate": 0.0005908027100668046,
      "loss": 2.2203,
      "step": 18207
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6307514905929565,
      "learning_rate": 0.0005908017049313287,
      "loss": 2.5646,
      "step": 18208
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.616671621799469,
      "learning_rate": 0.0005908006997417872,
      "loss": 2.3152,
      "step": 18209
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6061789989471436,
      "learning_rate": 0.0005907996944981806,
      "loss": 2.61,
      "step": 18210
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.5982519388198853,
      "learning_rate": 0.0005907986892005085,
      "loss": 2.44,
      "step": 18211
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6592146754264832,
      "learning_rate": 0.0005907976838487718,
      "loss": 2.3866,
      "step": 18212
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6340520977973938,
      "learning_rate": 0.0005907966784429701,
      "loss": 2.4539,
      "step": 18213
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6323225498199463,
      "learning_rate": 0.0005907956729831038,
      "loss": 2.2161,
      "step": 18214
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.5784289836883545,
      "learning_rate": 0.0005907946674691731,
      "loss": 2.2717,
      "step": 18215
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6265243291854858,
      "learning_rate": 0.0005907936619011782,
      "loss": 2.5112,
      "step": 18216
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.594056248664856,
      "learning_rate": 0.0005907926562791193,
      "loss": 2.4756,
      "step": 18217
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.5461655855178833,
      "learning_rate": 0.0005907916506029966,
      "loss": 2.6123,
      "step": 18218
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6199274659156799,
      "learning_rate": 0.0005907906448728102,
      "loss": 2.3077,
      "step": 18219
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6365892887115479,
      "learning_rate": 0.0005907896390885603,
      "loss": 2.3531,
      "step": 18220
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6153301000595093,
      "learning_rate": 0.0005907886332502471,
      "loss": 2.599,
      "step": 18221
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.619325578212738,
      "learning_rate": 0.0005907876273578709,
      "loss": 2.3973,
      "step": 18222
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6866475939750671,
      "learning_rate": 0.0005907866214114317,
      "loss": 2.4478,
      "step": 18223
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6127750277519226,
      "learning_rate": 0.0005907856154109297,
      "loss": 2.2147,
      "step": 18224
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6519774794578552,
      "learning_rate": 0.0005907846093563654,
      "loss": 2.4013,
      "step": 18225
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6207131743431091,
      "learning_rate": 0.0005907836032477387,
      "loss": 2.3858,
      "step": 18226
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6421499848365784,
      "learning_rate": 0.0005907825970850497,
      "loss": 2.2638,
      "step": 18227
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6052955389022827,
      "learning_rate": 0.0005907815908682987,
      "loss": 2.4879,
      "step": 18228
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6174860000610352,
      "learning_rate": 0.000590780584597486,
      "loss": 2.2558,
      "step": 18229
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6110562682151794,
      "learning_rate": 0.0005907795782726118,
      "loss": 2.2448,
      "step": 18230
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6197536587715149,
      "learning_rate": 0.000590778571893676,
      "loss": 2.3489,
      "step": 18231
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.749001145362854,
      "learning_rate": 0.000590777565460679,
      "loss": 2.3905,
      "step": 18232
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6021873950958252,
      "learning_rate": 0.0005907765589736211,
      "loss": 2.1694,
      "step": 18233
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.7263410687446594,
      "learning_rate": 0.0005907755524325023,
      "loss": 2.47,
      "step": 18234
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6267933249473572,
      "learning_rate": 0.0005907745458373228,
      "loss": 2.5973,
      "step": 18235
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6348791718482971,
      "learning_rate": 0.0005907735391880828,
      "loss": 2.2509,
      "step": 18236
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6147758364677429,
      "learning_rate": 0.0005907725324847827,
      "loss": 2.515,
      "step": 18237
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.7405598759651184,
      "learning_rate": 0.0005907715257274222,
      "loss": 2.396,
      "step": 18238
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.8051338195800781,
      "learning_rate": 0.000590770518916002,
      "loss": 2.3328,
      "step": 18239
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6854144334793091,
      "learning_rate": 0.000590769512050522,
      "loss": 2.2946,
      "step": 18240
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.7073598504066467,
      "learning_rate": 0.0005907685051309825,
      "loss": 2.4953,
      "step": 18241
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6890037655830383,
      "learning_rate": 0.0005907674981573836,
      "loss": 2.4147,
      "step": 18242
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.9824185371398926,
      "learning_rate": 0.0005907664911297255,
      "loss": 2.2158,
      "step": 18243
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.614467203617096,
      "learning_rate": 0.0005907654840480084,
      "loss": 2.6689,
      "step": 18244
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.9819903373718262,
      "learning_rate": 0.0005907644769122326,
      "loss": 2.1357,
      "step": 18245
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.9021379351615906,
      "learning_rate": 0.0005907634697223982,
      "loss": 2.359,
      "step": 18246
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.5532124638557434,
      "learning_rate": 0.0005907624624785054,
      "loss": 2.4614,
      "step": 18247
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.9204107522964478,
      "learning_rate": 0.0005907614551805542,
      "loss": 2.4341,
      "step": 18248
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.0376724004745483,
      "learning_rate": 0.000590760447828545,
      "loss": 2.2258,
      "step": 18249
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6317911148071289,
      "learning_rate": 0.0005907594404224782,
      "loss": 2.8041,
      "step": 18250
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.7442574501037598,
      "learning_rate": 0.0005907584329623535,
      "loss": 2.1777,
      "step": 18251
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.089421272277832,
      "learning_rate": 0.0005907574254481714,
      "loss": 2.1852,
      "step": 18252
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.688174843788147,
      "learning_rate": 0.0005907564178799319,
      "loss": 2.2844,
      "step": 18253
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.7258638143539429,
      "learning_rate": 0.0005907554102576354,
      "loss": 2.2658,
      "step": 18254
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.1790963411331177,
      "learning_rate": 0.0005907544025812819,
      "loss": 2.5219,
      "step": 18255
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.8324573636054993,
      "learning_rate": 0.0005907533948508717,
      "loss": 2.381,
      "step": 18256
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6807228326797485,
      "learning_rate": 0.000590752387066405,
      "loss": 2.2964,
      "step": 18257
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.7442469596862793,
      "learning_rate": 0.000590751379227882,
      "loss": 2.3308,
      "step": 18258
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.8617738485336304,
      "learning_rate": 0.0005907503713353027,
      "loss": 2.5361,
      "step": 18259
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.5917783975601196,
      "learning_rate": 0.0005907493633886675,
      "loss": 2.2754,
      "step": 18260
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6236482262611389,
      "learning_rate": 0.0005907483553879764,
      "loss": 2.4133,
      "step": 18261
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6616067290306091,
      "learning_rate": 0.0005907473473332298,
      "loss": 2.3599,
      "step": 18262
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.7744837403297424,
      "learning_rate": 0.0005907463392244277,
      "loss": 2.5087,
      "step": 18263
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.696747362613678,
      "learning_rate": 0.0005907453310615705,
      "loss": 2.4534,
      "step": 18264
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.5737082362174988,
      "learning_rate": 0.000590744322844658,
      "loss": 2.4359,
      "step": 18265
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.5959712862968445,
      "learning_rate": 0.0005907433145736909,
      "loss": 2.4845,
      "step": 18266
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6576908230781555,
      "learning_rate": 0.000590742306248669,
      "loss": 2.4743,
      "step": 18267
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6547756791114807,
      "learning_rate": 0.0005907412978695926,
      "loss": 2.2613,
      "step": 18268
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.5867646336555481,
      "learning_rate": 0.0005907402894364621,
      "loss": 2.4831,
      "step": 18269
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6485133171081543,
      "learning_rate": 0.0005907392809492773,
      "loss": 2.3992,
      "step": 18270
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.8572827577590942,
      "learning_rate": 0.0005907382724080386,
      "loss": 2.2747,
      "step": 18271
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6594774127006531,
      "learning_rate": 0.0005907372638127462,
      "loss": 2.5004,
      "step": 18272
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6527321338653564,
      "learning_rate": 0.0005907362551634002,
      "loss": 2.4368,
      "step": 18273
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.7159895896911621,
      "learning_rate": 0.000590735246460001,
      "loss": 2.3667,
      "step": 18274
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6316291689872742,
      "learning_rate": 0.0005907342377025485,
      "loss": 2.4063,
      "step": 18275
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.7228878736495972,
      "learning_rate": 0.000590733228891043,
      "loss": 2.236,
      "step": 18276
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6879189014434814,
      "learning_rate": 0.0005907322200254848,
      "loss": 2.4003,
      "step": 18277
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6099728941917419,
      "learning_rate": 0.0005907312111058738,
      "loss": 2.3851,
      "step": 18278
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.5906291604042053,
      "learning_rate": 0.0005907302021322105,
      "loss": 2.4487,
      "step": 18279
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6649441123008728,
      "learning_rate": 0.000590729193104495,
      "loss": 2.3188,
      "step": 18280
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.5854303240776062,
      "learning_rate": 0.0005907281840227275,
      "loss": 2.4333,
      "step": 18281
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.5686608552932739,
      "learning_rate": 0.000590727174886908,
      "loss": 2.5783,
      "step": 18282
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.59517902135849,
      "learning_rate": 0.000590726165697037,
      "loss": 2.4327,
      "step": 18283
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6445969343185425,
      "learning_rate": 0.0005907251564531144,
      "loss": 2.6976,
      "step": 18284
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6235507726669312,
      "learning_rate": 0.0005907241471551405,
      "loss": 2.0491,
      "step": 18285
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.5753697752952576,
      "learning_rate": 0.0005907231378031155,
      "loss": 2.6659,
      "step": 18286
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.5922691822052002,
      "learning_rate": 0.0005907221283970397,
      "loss": 2.7263,
      "step": 18287
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6202332377433777,
      "learning_rate": 0.000590721118936913,
      "loss": 2.3082,
      "step": 18288
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.5671362280845642,
      "learning_rate": 0.0005907201094227358,
      "loss": 2.6384,
      "step": 18289
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6259698271751404,
      "learning_rate": 0.0005907190998545083,
      "loss": 2.4452,
      "step": 18290
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6242589950561523,
      "learning_rate": 0.0005907180902322306,
      "loss": 2.1698,
      "step": 18291
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6223614811897278,
      "learning_rate": 0.0005907170805559029,
      "loss": 2.1751,
      "step": 18292
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.5735807418823242,
      "learning_rate": 0.0005907160708255255,
      "loss": 2.4995,
      "step": 18293
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.5850667953491211,
      "learning_rate": 0.0005907150610410984,
      "loss": 2.2115,
      "step": 18294
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6443589925765991,
      "learning_rate": 0.0005907140512026219,
      "loss": 2.3414,
      "step": 18295
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.5896043181419373,
      "learning_rate": 0.0005907130413100962,
      "loss": 2.5465,
      "step": 18296
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.5878885984420776,
      "learning_rate": 0.0005907120313635214,
      "loss": 2.3895,
      "step": 18297
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.651003897190094,
      "learning_rate": 0.0005907110213628978,
      "loss": 2.6754,
      "step": 18298
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6150368452072144,
      "learning_rate": 0.0005907100113082256,
      "loss": 2.3022,
      "step": 18299
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6355466246604919,
      "learning_rate": 0.0005907090011995048,
      "loss": 2.4631,
      "step": 18300
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.5922060608863831,
      "learning_rate": 0.0005907079910367357,
      "loss": 2.6124,
      "step": 18301
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6373087763786316,
      "learning_rate": 0.0005907069808199185,
      "loss": 2.0943,
      "step": 18302
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.7133458256721497,
      "learning_rate": 0.0005907059705490535,
      "loss": 2.362,
      "step": 18303
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6011818051338196,
      "learning_rate": 0.0005907049602241407,
      "loss": 2.6727,
      "step": 18304
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6328787207603455,
      "learning_rate": 0.0005907039498451804,
      "loss": 2.7425,
      "step": 18305
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6467156410217285,
      "learning_rate": 0.0005907029394121727,
      "loss": 2.2807,
      "step": 18306
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6173548102378845,
      "learning_rate": 0.0005907019289251178,
      "loss": 2.5083,
      "step": 18307
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6068389415740967,
      "learning_rate": 0.000590700918384016,
      "loss": 2.6822,
      "step": 18308
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6012086868286133,
      "learning_rate": 0.0005906999077888674,
      "loss": 2.2808,
      "step": 18309
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6205344200134277,
      "learning_rate": 0.0005906988971396722,
      "loss": 2.4073,
      "step": 18310
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6207512617111206,
      "learning_rate": 0.0005906978864364305,
      "loss": 2.5568,
      "step": 18311
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6515049338340759,
      "learning_rate": 0.0005906968756791426,
      "loss": 2.3647,
      "step": 18312
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.5706259608268738,
      "learning_rate": 0.0005906958648678088,
      "loss": 2.4114,
      "step": 18313
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.5865777730941772,
      "learning_rate": 0.000590694854002429,
      "loss": 2.3851,
      "step": 18314
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6173310875892639,
      "learning_rate": 0.0005906938430830037,
      "loss": 2.3311,
      "step": 18315
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6299098134040833,
      "learning_rate": 0.0005906928321095328,
      "loss": 2.4841,
      "step": 18316
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6381822228431702,
      "learning_rate": 0.0005906918210820166,
      "loss": 2.4517,
      "step": 18317
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.5445230603218079,
      "learning_rate": 0.0005906908100004553,
      "loss": 2.3014,
      "step": 18318
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.623163640499115,
      "learning_rate": 0.0005906897988648492,
      "loss": 2.0632,
      "step": 18319
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.5998711585998535,
      "learning_rate": 0.0005906887876751983,
      "loss": 2.4535,
      "step": 18320
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.5825781226158142,
      "learning_rate": 0.0005906877764315029,
      "loss": 2.3323,
      "step": 18321
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.7037911415100098,
      "learning_rate": 0.0005906867651337632,
      "loss": 2.4107,
      "step": 18322
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6418112516403198,
      "learning_rate": 0.0005906857537819792,
      "loss": 2.3158,
      "step": 18323
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6426312923431396,
      "learning_rate": 0.0005906847423761514,
      "loss": 2.4737,
      "step": 18324
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6098247766494751,
      "learning_rate": 0.0005906837309162797,
      "loss": 2.3851,
      "step": 18325
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.5671477913856506,
      "learning_rate": 0.0005906827194023644,
      "loss": 2.3627,
      "step": 18326
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.5886126160621643,
      "learning_rate": 0.0005906817078344058,
      "loss": 2.3167,
      "step": 18327
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6330596208572388,
      "learning_rate": 0.0005906806962124039,
      "loss": 2.4082,
      "step": 18328
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.5835723280906677,
      "learning_rate": 0.000590679684536359,
      "loss": 2.6084,
      "step": 18329
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6295208930969238,
      "learning_rate": 0.0005906786728062713,
      "loss": 2.5511,
      "step": 18330
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6263414025306702,
      "learning_rate": 0.0005906776610221407,
      "loss": 2.5518,
      "step": 18331
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.609452486038208,
      "learning_rate": 0.0005906766491839679,
      "loss": 2.4468,
      "step": 18332
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6269476413726807,
      "learning_rate": 0.0005906756372917528,
      "loss": 2.2576,
      "step": 18333
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.572881281375885,
      "learning_rate": 0.0005906746253454956,
      "loss": 2.2283,
      "step": 18334
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.5761004090309143,
      "learning_rate": 0.0005906736133451965,
      "loss": 2.2716,
      "step": 18335
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6274913549423218,
      "learning_rate": 0.0005906726012908556,
      "loss": 2.2044,
      "step": 18336
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6413082480430603,
      "learning_rate": 0.0005906715891824732,
      "loss": 2.4087,
      "step": 18337
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.5914546847343445,
      "learning_rate": 0.0005906705770200495,
      "loss": 2.5017,
      "step": 18338
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.5836436152458191,
      "learning_rate": 0.0005906695648035847,
      "loss": 2.3385,
      "step": 18339
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6856353282928467,
      "learning_rate": 0.0005906685525330789,
      "loss": 2.2715,
      "step": 18340
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6328321695327759,
      "learning_rate": 0.0005906675402085322,
      "loss": 2.4403,
      "step": 18341
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.649882435798645,
      "learning_rate": 0.0005906665278299451,
      "loss": 2.2836,
      "step": 18342
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.7707738876342773,
      "learning_rate": 0.0005906655153973176,
      "loss": 2.4095,
      "step": 18343
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.5866604447364807,
      "learning_rate": 0.0005906645029106498,
      "loss": 2.3078,
      "step": 18344
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.7105346322059631,
      "learning_rate": 0.0005906634903699421,
      "loss": 2.4251,
      "step": 18345
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6523656249046326,
      "learning_rate": 0.0005906624777751945,
      "loss": 2.5845,
      "step": 18346
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.5965498685836792,
      "learning_rate": 0.0005906614651264071,
      "loss": 2.5596,
      "step": 18347
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.5632983446121216,
      "learning_rate": 0.0005906604524235805,
      "loss": 2.4115,
      "step": 18348
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6457123160362244,
      "learning_rate": 0.0005906594396667145,
      "loss": 2.2641,
      "step": 18349
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6888351440429688,
      "learning_rate": 0.0005906584268558095,
      "loss": 2.4515,
      "step": 18350
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.605218231678009,
      "learning_rate": 0.0005906574139908655,
      "loss": 2.3134,
      "step": 18351
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6149601340293884,
      "learning_rate": 0.0005906564010718829,
      "loss": 2.4495,
      "step": 18352
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6147202849388123,
      "learning_rate": 0.0005906553880988618,
      "loss": 2.3263,
      "step": 18353
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6286763548851013,
      "learning_rate": 0.0005906543750718023,
      "loss": 2.5793,
      "step": 18354
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6150030493736267,
      "learning_rate": 0.0005906533619907048,
      "loss": 2.263,
      "step": 18355
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6135005950927734,
      "learning_rate": 0.0005906523488555692,
      "loss": 2.3275,
      "step": 18356
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.7351021766662598,
      "learning_rate": 0.0005906513356663959,
      "loss": 2.2477,
      "step": 18357
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6162844896316528,
      "learning_rate": 0.000590650322423185,
      "loss": 2.4213,
      "step": 18358
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6512635946273804,
      "learning_rate": 0.0005906493091259367,
      "loss": 2.0945,
      "step": 18359
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6320863366127014,
      "learning_rate": 0.0005906482957746512,
      "loss": 2.5142,
      "step": 18360
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.5964892506599426,
      "learning_rate": 0.0005906472823693288,
      "loss": 2.4205,
      "step": 18361
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6251099109649658,
      "learning_rate": 0.0005906462689099696,
      "loss": 2.3272,
      "step": 18362
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6359462141990662,
      "learning_rate": 0.0005906452553965737,
      "loss": 2.4982,
      "step": 18363
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6102954745292664,
      "learning_rate": 0.0005906442418291413,
      "loss": 2.3656,
      "step": 18364
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6204177737236023,
      "learning_rate": 0.0005906432282076727,
      "loss": 2.3377,
      "step": 18365
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.632632315158844,
      "learning_rate": 0.0005906422145321681,
      "loss": 2.5369,
      "step": 18366
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.5485479235649109,
      "learning_rate": 0.0005906412008026275,
      "loss": 2.314,
      "step": 18367
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6279571652412415,
      "learning_rate": 0.0005906401870190513,
      "loss": 2.2658,
      "step": 18368
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.545129120349884,
      "learning_rate": 0.0005906391731814397,
      "loss": 2.1527,
      "step": 18369
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.5658183693885803,
      "learning_rate": 0.0005906381592897926,
      "loss": 2.2787,
      "step": 18370
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6360193490982056,
      "learning_rate": 0.0005906371453441105,
      "loss": 2.4104,
      "step": 18371
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6048110127449036,
      "learning_rate": 0.0005906361313443933,
      "loss": 2.4484,
      "step": 18372
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6008849740028381,
      "learning_rate": 0.0005906351172906417,
      "loss": 2.362,
      "step": 18373
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6278961300849915,
      "learning_rate": 0.0005906341031828552,
      "loss": 2.3212,
      "step": 18374
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6327820420265198,
      "learning_rate": 0.0005906330890210346,
      "loss": 2.3027,
      "step": 18375
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.622890830039978,
      "learning_rate": 0.0005906320748051796,
      "loss": 2.242,
      "step": 18376
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.600710391998291,
      "learning_rate": 0.0005906310605352907,
      "loss": 2.2052,
      "step": 18377
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6928284168243408,
      "learning_rate": 0.000590630046211368,
      "loss": 2.6497,
      "step": 18378
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6320167779922485,
      "learning_rate": 0.0005906290318334117,
      "loss": 2.5626,
      "step": 18379
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.7405136227607727,
      "learning_rate": 0.000590628017401422,
      "loss": 2.3381,
      "step": 18380
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6308146715164185,
      "learning_rate": 0.000590627002915399,
      "loss": 2.5069,
      "step": 18381
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.5989260673522949,
      "learning_rate": 0.0005906259883753429,
      "loss": 2.4809,
      "step": 18382
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.5931291580200195,
      "learning_rate": 0.0005906249737812541,
      "loss": 2.2337,
      "step": 18383
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.5727179050445557,
      "learning_rate": 0.0005906239591331326,
      "loss": 2.3661,
      "step": 18384
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.5756070613861084,
      "learning_rate": 0.0005906229444309786,
      "loss": 2.4483,
      "step": 18385
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6030264496803284,
      "learning_rate": 0.0005906219296747923,
      "loss": 2.4341,
      "step": 18386
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6116260290145874,
      "learning_rate": 0.0005906209148645739,
      "loss": 2.4593,
      "step": 18387
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.5681875944137573,
      "learning_rate": 0.0005906199000003235,
      "loss": 2.3168,
      "step": 18388
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6011179685592651,
      "learning_rate": 0.0005906188850820414,
      "loss": 2.3743,
      "step": 18389
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6508798599243164,
      "learning_rate": 0.0005906178701097279,
      "loss": 2.3933,
      "step": 18390
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.5550638437271118,
      "learning_rate": 0.0005906168550833829,
      "loss": 2.2732,
      "step": 18391
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.5550836324691772,
      "learning_rate": 0.0005906158400030069,
      "loss": 2.3733,
      "step": 18392
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.5964147448539734,
      "learning_rate": 0.0005906148248685998,
      "loss": 2.1577,
      "step": 18393
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.5495170950889587,
      "learning_rate": 0.000590613809680162,
      "loss": 2.4433,
      "step": 18394
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.5808961391448975,
      "learning_rate": 0.0005906127944376935,
      "loss": 2.2176,
      "step": 18395
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6313912868499756,
      "learning_rate": 0.0005906117791411946,
      "loss": 2.708,
      "step": 18396
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.62636798620224,
      "learning_rate": 0.0005906107637906657,
      "loss": 2.5014,
      "step": 18397
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6576387882232666,
      "learning_rate": 0.0005906097483861065,
      "loss": 2.3788,
      "step": 18398
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.5726398825645447,
      "learning_rate": 0.0005906087329275176,
      "loss": 2.377,
      "step": 18399
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.5768764615058899,
      "learning_rate": 0.000590607717414899,
      "loss": 2.2567,
      "step": 18400
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.565189778804779,
      "learning_rate": 0.000590606701848251,
      "loss": 2.3159,
      "step": 18401
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.5949511528015137,
      "learning_rate": 0.0005906056862275737,
      "loss": 2.2022,
      "step": 18402
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.5872638821601868,
      "learning_rate": 0.0005906046705528674,
      "loss": 2.3527,
      "step": 18403
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6100642681121826,
      "learning_rate": 0.0005906036548241321,
      "loss": 2.4254,
      "step": 18404
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6187747120857239,
      "learning_rate": 0.000590602639041368,
      "loss": 2.416,
      "step": 18405
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.5985169410705566,
      "learning_rate": 0.0005906016232045756,
      "loss": 2.3234,
      "step": 18406
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6052901744842529,
      "learning_rate": 0.0005906006073137547,
      "loss": 2.2881,
      "step": 18407
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6906227469444275,
      "learning_rate": 0.0005905995913689058,
      "loss": 2.4287,
      "step": 18408
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6456729769706726,
      "learning_rate": 0.0005905985753700288,
      "loss": 2.3991,
      "step": 18409
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.5583072900772095,
      "learning_rate": 0.0005905975593171241,
      "loss": 2.3142,
      "step": 18410
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.659927487373352,
      "learning_rate": 0.0005905965432101919,
      "loss": 2.3499,
      "step": 18411
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.7447700500488281,
      "learning_rate": 0.0005905955270492322,
      "loss": 2.3651,
      "step": 18412
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6321161389350891,
      "learning_rate": 0.0005905945108342454,
      "loss": 2.2369,
      "step": 18413
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6914260983467102,
      "learning_rate": 0.0005905934945652316,
      "loss": 2.5153,
      "step": 18414
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.7247455716133118,
      "learning_rate": 0.0005905924782421909,
      "loss": 2.4605,
      "step": 18415
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6090975403785706,
      "learning_rate": 0.0005905914618651236,
      "loss": 2.3247,
      "step": 18416
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6938773989677429,
      "learning_rate": 0.0005905904454340299,
      "loss": 2.3254,
      "step": 18417
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6559879779815674,
      "learning_rate": 0.00059058942894891,
      "loss": 2.4026,
      "step": 18418
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.5647534132003784,
      "learning_rate": 0.000590588412409764,
      "loss": 2.1286,
      "step": 18419
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6975098252296448,
      "learning_rate": 0.000590587395816592,
      "loss": 2.3406,
      "step": 18420
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6224891543388367,
      "learning_rate": 0.0005905863791693945,
      "loss": 2.2817,
      "step": 18421
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6502634286880493,
      "learning_rate": 0.0005905853624681714,
      "loss": 2.2422,
      "step": 18422
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6323431134223938,
      "learning_rate": 0.000590584345712923,
      "loss": 2.2144,
      "step": 18423
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.7236103415489197,
      "learning_rate": 0.0005905833289036495,
      "loss": 2.2792,
      "step": 18424
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.7171328067779541,
      "learning_rate": 0.0005905823120403511,
      "loss": 2.5112,
      "step": 18425
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.7518364787101746,
      "learning_rate": 0.000590581295123028,
      "loss": 2.4513,
      "step": 18426
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6319157481193542,
      "learning_rate": 0.0005905802781516802,
      "loss": 2.4577,
      "step": 18427
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6041073799133301,
      "learning_rate": 0.0005905792611263082,
      "loss": 2.3877,
      "step": 18428
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6674061417579651,
      "learning_rate": 0.000590578244046912,
      "loss": 2.5316,
      "step": 18429
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.7139891386032104,
      "learning_rate": 0.0005905772269134918,
      "loss": 2.5262,
      "step": 18430
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6375575065612793,
      "learning_rate": 0.0005905762097260477,
      "loss": 2.4931,
      "step": 18431
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6756638288497925,
      "learning_rate": 0.0005905751924845801,
      "loss": 2.2653,
      "step": 18432
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6802226901054382,
      "learning_rate": 0.0005905741751890892,
      "loss": 2.4212,
      "step": 18433
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.602983832359314,
      "learning_rate": 0.0005905731578395749,
      "loss": 2.4015,
      "step": 18434
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.5690773725509644,
      "learning_rate": 0.0005905721404360376,
      "loss": 2.4414,
      "step": 18435
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.594424307346344,
      "learning_rate": 0.0005905711229784776,
      "loss": 2.3463,
      "step": 18436
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.7375971674919128,
      "learning_rate": 0.0005905701054668948,
      "loss": 2.4467,
      "step": 18437
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6314366459846497,
      "learning_rate": 0.0005905690879012895,
      "loss": 2.48,
      "step": 18438
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.5952087044715881,
      "learning_rate": 0.000590568070281662,
      "loss": 2.4627,
      "step": 18439
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6735010147094727,
      "learning_rate": 0.0005905670526080124,
      "loss": 2.3861,
      "step": 18440
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6005065441131592,
      "learning_rate": 0.0005905660348803409,
      "loss": 2.3569,
      "step": 18441
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.589693546295166,
      "learning_rate": 0.0005905650170986476,
      "loss": 2.2928,
      "step": 18442
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6315454840660095,
      "learning_rate": 0.0005905639992629329,
      "loss": 2.3838,
      "step": 18443
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6116651892662048,
      "learning_rate": 0.0005905629813731969,
      "loss": 2.3671,
      "step": 18444
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.5930579900741577,
      "learning_rate": 0.0005905619634294396,
      "loss": 2.3506,
      "step": 18445
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.7398588061332703,
      "learning_rate": 0.0005905609454316614,
      "loss": 2.2953,
      "step": 18446
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6461532115936279,
      "learning_rate": 0.0005905599273798625,
      "loss": 2.2749,
      "step": 18447
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.5451327562332153,
      "learning_rate": 0.000590558909274043,
      "loss": 2.056,
      "step": 18448
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6712853312492371,
      "learning_rate": 0.0005905578911142032,
      "loss": 2.5375,
      "step": 18449
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6197553873062134,
      "learning_rate": 0.0005905568729003429,
      "loss": 2.683,
      "step": 18450
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.676578938961029,
      "learning_rate": 0.0005905558546324629,
      "loss": 2.5777,
      "step": 18451
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.629698634147644,
      "learning_rate": 0.000590554836310563,
      "loss": 2.5322,
      "step": 18452
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.8310420513153076,
      "learning_rate": 0.0005905538179346435,
      "loss": 2.2001,
      "step": 18453
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6628568172454834,
      "learning_rate": 0.0005905527995047045,
      "loss": 2.5352,
      "step": 18454
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.571111261844635,
      "learning_rate": 0.0005905517810207462,
      "loss": 2.33,
      "step": 18455
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6066177487373352,
      "learning_rate": 0.0005905507624827689,
      "loss": 2.3744,
      "step": 18456
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6853155493736267,
      "learning_rate": 0.0005905497438907727,
      "loss": 2.1585,
      "step": 18457
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6206864714622498,
      "learning_rate": 0.0005905487252447578,
      "loss": 2.1267,
      "step": 18458
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.667186975479126,
      "learning_rate": 0.0005905477065447246,
      "loss": 2.6363,
      "step": 18459
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.7705121636390686,
      "learning_rate": 0.0005905466877906729,
      "loss": 2.3455,
      "step": 18460
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.7281651496887207,
      "learning_rate": 0.0005905456689826031,
      "loss": 2.5089,
      "step": 18461
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.5893397927284241,
      "learning_rate": 0.0005905446501205154,
      "loss": 2.6119,
      "step": 18462
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6197016835212708,
      "learning_rate": 0.00059054363120441,
      "loss": 2.4932,
      "step": 18463
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.578227698802948,
      "learning_rate": 0.000590542612234287,
      "loss": 2.4163,
      "step": 18464
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6300244331359863,
      "learning_rate": 0.0005905415932101467,
      "loss": 2.5175,
      "step": 18465
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.5734255313873291,
      "learning_rate": 0.0005905405741319892,
      "loss": 2.3153,
      "step": 18466
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6281532645225525,
      "learning_rate": 0.0005905395549998147,
      "loss": 2.2983,
      "step": 18467
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6160625219345093,
      "learning_rate": 0.0005905385358136235,
      "loss": 2.3973,
      "step": 18468
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6582415699958801,
      "learning_rate": 0.0005905375165734157,
      "loss": 2.4735,
      "step": 18469
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6532080769538879,
      "learning_rate": 0.0005905364972791914,
      "loss": 2.5475,
      "step": 18470
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.9501135945320129,
      "learning_rate": 0.0005905354779309509,
      "loss": 2.4866,
      "step": 18471
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6397354602813721,
      "learning_rate": 0.0005905344585286944,
      "loss": 2.3484,
      "step": 18472
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.7973611950874329,
      "learning_rate": 0.000590533439072422,
      "loss": 2.4403,
      "step": 18473
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.8084685802459717,
      "learning_rate": 0.000590532419562134,
      "loss": 2.3611,
      "step": 18474
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6250868439674377,
      "learning_rate": 0.0005905313999978306,
      "loss": 2.3691,
      "step": 18475
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.652863085269928,
      "learning_rate": 0.0005905303803795119,
      "loss": 2.4851,
      "step": 18476
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.8245227336883545,
      "learning_rate": 0.000590529360707178,
      "loss": 2.2152,
      "step": 18477
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.5719079375267029,
      "learning_rate": 0.0005905283409808293,
      "loss": 2.3068,
      "step": 18478
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6577907204627991,
      "learning_rate": 0.0005905273212004659,
      "loss": 2.6203,
      "step": 18479
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6790502667427063,
      "learning_rate": 0.000590526301366088,
      "loss": 2.3208,
      "step": 18480
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.667360246181488,
      "learning_rate": 0.0005905252814776958,
      "loss": 2.522,
      "step": 18481
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.5959815979003906,
      "learning_rate": 0.0005905242615352893,
      "loss": 2.2673,
      "step": 18482
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.7350106239318848,
      "learning_rate": 0.000590523241538869,
      "loss": 2.3169,
      "step": 18483
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6625178456306458,
      "learning_rate": 0.0005905222214884349,
      "loss": 2.3829,
      "step": 18484
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6130388379096985,
      "learning_rate": 0.0005905212013839873,
      "loss": 2.1629,
      "step": 18485
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6502560377120972,
      "learning_rate": 0.0005905201812255262,
      "loss": 2.489,
      "step": 18486
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6085477471351624,
      "learning_rate": 0.000590519161013052,
      "loss": 2.3762,
      "step": 18487
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.5908966064453125,
      "learning_rate": 0.0005905181407465648,
      "loss": 2.1685,
      "step": 18488
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6516970992088318,
      "learning_rate": 0.0005905171204260649,
      "loss": 2.2585,
      "step": 18489
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.7409035563468933,
      "learning_rate": 0.0005905161000515522,
      "loss": 2.2551,
      "step": 18490
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.7527430653572083,
      "learning_rate": 0.0005905150796230271,
      "loss": 2.5059,
      "step": 18491
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6667411923408508,
      "learning_rate": 0.0005905140591404898,
      "loss": 2.3045,
      "step": 18492
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6150256991386414,
      "learning_rate": 0.0005905130386039405,
      "loss": 2.2263,
      "step": 18493
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.5929767489433289,
      "learning_rate": 0.0005905120180133794,
      "loss": 2.3382,
      "step": 18494
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6178829669952393,
      "learning_rate": 0.0005905109973688065,
      "loss": 2.318,
      "step": 18495
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6480668187141418,
      "learning_rate": 0.0005905099766702221,
      "loss": 2.362,
      "step": 18496
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.669444739818573,
      "learning_rate": 0.0005905089559176265,
      "loss": 2.3696,
      "step": 18497
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6435242891311646,
      "learning_rate": 0.0005905079351110197,
      "loss": 2.7452,
      "step": 18498
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.5774031281471252,
      "learning_rate": 0.0005905069142504021,
      "loss": 2.4623,
      "step": 18499
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6783932447433472,
      "learning_rate": 0.0005905058933357738,
      "loss": 2.2251,
      "step": 18500
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6413115859031677,
      "learning_rate": 0.0005905048723671349,
      "loss": 2.4472,
      "step": 18501
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.7672565579414368,
      "learning_rate": 0.0005905038513444856,
      "loss": 2.2183,
      "step": 18502
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.7254680395126343,
      "learning_rate": 0.0005905028302678263,
      "loss": 2.386,
      "step": 18503
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6311713457107544,
      "learning_rate": 0.0005905018091371569,
      "loss": 2.4447,
      "step": 18504
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6011752486228943,
      "learning_rate": 0.0005905007879524779,
      "loss": 2.2038,
      "step": 18505
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6938033699989319,
      "learning_rate": 0.0005904997667137892,
      "loss": 2.6385,
      "step": 18506
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.5766940116882324,
      "learning_rate": 0.0005904987454210911,
      "loss": 2.2941,
      "step": 18507
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6658801436424255,
      "learning_rate": 0.0005904977240743838,
      "loss": 2.107,
      "step": 18508
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.7014679312705994,
      "learning_rate": 0.0005904967026736675,
      "loss": 2.4163,
      "step": 18509
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.664659857749939,
      "learning_rate": 0.0005904956812189425,
      "loss": 2.3117,
      "step": 18510
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6262656450271606,
      "learning_rate": 0.0005904946597102088,
      "loss": 2.3745,
      "step": 18511
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6707485914230347,
      "learning_rate": 0.0005904936381474666,
      "loss": 2.4527,
      "step": 18512
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.7043043971061707,
      "learning_rate": 0.0005904926165307161,
      "loss": 2.2971,
      "step": 18513
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.5403081774711609,
      "learning_rate": 0.0005904915948599576,
      "loss": 1.954,
      "step": 18514
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.8179050087928772,
      "learning_rate": 0.0005904905731351913,
      "loss": 2.2091,
      "step": 18515
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6737557649612427,
      "learning_rate": 0.0005904895513564173,
      "loss": 2.3912,
      "step": 18516
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.5962871313095093,
      "learning_rate": 0.0005904885295236357,
      "loss": 2.5076,
      "step": 18517
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.5710803866386414,
      "learning_rate": 0.0005904875076368468,
      "loss": 2.2829,
      "step": 18518
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6684005856513977,
      "learning_rate": 0.000590486485696051,
      "loss": 2.2027,
      "step": 18519
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6189022064208984,
      "learning_rate": 0.000590485463701248,
      "loss": 2.4998,
      "step": 18520
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6577281951904297,
      "learning_rate": 0.0005904844416524384,
      "loss": 2.5038,
      "step": 18521
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6084014773368835,
      "learning_rate": 0.0005904834195496223,
      "loss": 2.3627,
      "step": 18522
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.7136300802230835,
      "learning_rate": 0.0005904823973927998,
      "loss": 2.254,
      "step": 18523
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.5859276056289673,
      "learning_rate": 0.0005904813751819711,
      "loss": 2.1506,
      "step": 18524
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6836026906967163,
      "learning_rate": 0.0005904803529171366,
      "loss": 2.4152,
      "step": 18525
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6619290113449097,
      "learning_rate": 0.0005904793305982961,
      "loss": 2.1299,
      "step": 18526
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6034098863601685,
      "learning_rate": 0.0005904783082254501,
      "loss": 2.3405,
      "step": 18527
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6152466535568237,
      "learning_rate": 0.0005904772857985988,
      "loss": 2.6654,
      "step": 18528
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6995987296104431,
      "learning_rate": 0.0005904762633177422,
      "loss": 2.3659,
      "step": 18529
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6935610175132751,
      "learning_rate": 0.0005904752407828806,
      "loss": 2.3747,
      "step": 18530
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6817725300788879,
      "learning_rate": 0.0005904742181940141,
      "loss": 2.4199,
      "step": 18531
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.7881429195404053,
      "learning_rate": 0.000590473195551143,
      "loss": 2.4357,
      "step": 18532
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6397304534912109,
      "learning_rate": 0.0005904721728542674,
      "loss": 2.3054,
      "step": 18533
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6015338897705078,
      "learning_rate": 0.0005904711501033877,
      "loss": 2.1515,
      "step": 18534
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.7668033242225647,
      "learning_rate": 0.0005904701272985038,
      "loss": 2.2534,
      "step": 18535
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.5877746343612671,
      "learning_rate": 0.0005904691044396161,
      "loss": 2.4589,
      "step": 18536
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6549368500709534,
      "learning_rate": 0.0005904680815267245,
      "loss": 2.2738,
      "step": 18537
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6223468780517578,
      "learning_rate": 0.0005904670585598296,
      "loss": 2.5185,
      "step": 18538
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.7348851561546326,
      "learning_rate": 0.0005904660355389314,
      "loss": 2.2586,
      "step": 18539
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.7014195322990417,
      "learning_rate": 0.00059046501246403,
      "loss": 2.4858,
      "step": 18540
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.5951696634292603,
      "learning_rate": 0.0005904639893351258,
      "loss": 2.352,
      "step": 18541
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6423092484474182,
      "learning_rate": 0.0005904629661522187,
      "loss": 2.7657,
      "step": 18542
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6760572791099548,
      "learning_rate": 0.0005904619429153091,
      "loss": 2.2766,
      "step": 18543
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.5976812839508057,
      "learning_rate": 0.0005904609196243971,
      "loss": 2.4305,
      "step": 18544
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.614054262638092,
      "learning_rate": 0.0005904598962794831,
      "loss": 2.4001,
      "step": 18545
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6045619249343872,
      "learning_rate": 0.000590458872880567,
      "loss": 2.1768,
      "step": 18546
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6886780858039856,
      "learning_rate": 0.0005904578494276491,
      "loss": 2.2998,
      "step": 18547
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.5878896117210388,
      "learning_rate": 0.0005904568259207295,
      "loss": 2.2323,
      "step": 18548
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.5832096338272095,
      "learning_rate": 0.0005904558023598087,
      "loss": 2.3809,
      "step": 18549
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6853975653648376,
      "learning_rate": 0.0005904547787448866,
      "loss": 2.427,
      "step": 18550
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6035110354423523,
      "learning_rate": 0.0005904537550759635,
      "loss": 2.241,
      "step": 18551
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.8354074358940125,
      "learning_rate": 0.0005904527313530395,
      "loss": 2.0214,
      "step": 18552
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6056275963783264,
      "learning_rate": 0.0005904517075761149,
      "loss": 2.4835,
      "step": 18553
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6033548712730408,
      "learning_rate": 0.0005904506837451899,
      "loss": 2.4193,
      "step": 18554
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6309451460838318,
      "learning_rate": 0.0005904496598602646,
      "loss": 2.4575,
      "step": 18555
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.7075836658477783,
      "learning_rate": 0.0005904486359213392,
      "loss": 2.4669,
      "step": 18556
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.5701587200164795,
      "learning_rate": 0.0005904476119284139,
      "loss": 2.3378,
      "step": 18557
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.8017728924751282,
      "learning_rate": 0.0005904465878814889,
      "loss": 2.3697,
      "step": 18558
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6125187277793884,
      "learning_rate": 0.0005904455637805644,
      "loss": 2.3443,
      "step": 18559
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6010349988937378,
      "learning_rate": 0.0005904445396256406,
      "loss": 2.5541,
      "step": 18560
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.7294386029243469,
      "learning_rate": 0.0005904435154167176,
      "loss": 2.3932,
      "step": 18561
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6683494448661804,
      "learning_rate": 0.0005904424911537959,
      "loss": 2.2083,
      "step": 18562
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.577991247177124,
      "learning_rate": 0.0005904414668368752,
      "loss": 2.1369,
      "step": 18563
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6626772880554199,
      "learning_rate": 0.000590440442465956,
      "loss": 2.4078,
      "step": 18564
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6231114268302917,
      "learning_rate": 0.0005904394180410386,
      "loss": 2.3553,
      "step": 18565
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6023157238960266,
      "learning_rate": 0.000590438393562123,
      "loss": 2.2243,
      "step": 18566
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6373980045318604,
      "learning_rate": 0.0005904373690292093,
      "loss": 2.1268,
      "step": 18567
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.597263753414154,
      "learning_rate": 0.0005904363444422979,
      "loss": 2.3348,
      "step": 18568
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.632400631904602,
      "learning_rate": 0.0005904353198013888,
      "loss": 2.388,
      "step": 18569
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.5948858857154846,
      "learning_rate": 0.0005904342951064824,
      "loss": 2.3666,
      "step": 18570
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.767049252986908,
      "learning_rate": 0.0005904332703575788,
      "loss": 2.415,
      "step": 18571
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.623028039932251,
      "learning_rate": 0.000590432245554678,
      "loss": 2.3908,
      "step": 18572
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6366173624992371,
      "learning_rate": 0.0005904312206977805,
      "loss": 2.3152,
      "step": 18573
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.5815147161483765,
      "learning_rate": 0.0005904301957868863,
      "loss": 2.1587,
      "step": 18574
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.7044437527656555,
      "learning_rate": 0.0005904291708219957,
      "loss": 2.3482,
      "step": 18575
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6345039010047913,
      "learning_rate": 0.0005904281458031088,
      "loss": 2.1157,
      "step": 18576
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.664819598197937,
      "learning_rate": 0.0005904271207302258,
      "loss": 2.5561,
      "step": 18577
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.5921841859817505,
      "learning_rate": 0.000590426095603347,
      "loss": 2.3192,
      "step": 18578
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.627721905708313,
      "learning_rate": 0.0005904250704224723,
      "loss": 2.3747,
      "step": 18579
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6726378798484802,
      "learning_rate": 0.0005904240451876023,
      "loss": 2.4241,
      "step": 18580
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6530556082725525,
      "learning_rate": 0.000590423019898737,
      "loss": 2.4935,
      "step": 18581
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6015145778656006,
      "learning_rate": 0.0005904219945558765,
      "loss": 2.393,
      "step": 18582
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6680452823638916,
      "learning_rate": 0.0005904209691590211,
      "loss": 2.4234,
      "step": 18583
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.5910633206367493,
      "learning_rate": 0.000590419943708171,
      "loss": 2.2415,
      "step": 18584
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6082665920257568,
      "learning_rate": 0.0005904189182033262,
      "loss": 2.2194,
      "step": 18585
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6505070328712463,
      "learning_rate": 0.0005904178926444872,
      "loss": 2.3935,
      "step": 18586
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6454569697380066,
      "learning_rate": 0.0005904168670316539,
      "loss": 2.132,
      "step": 18587
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.674260675907135,
      "learning_rate": 0.0005904158413648267,
      "loss": 2.2832,
      "step": 18588
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6362196207046509,
      "learning_rate": 0.0005904148156440057,
      "loss": 2.42,
      "step": 18589
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6294332146644592,
      "learning_rate": 0.0005904137898691911,
      "loss": 2.4775,
      "step": 18590
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.619059681892395,
      "learning_rate": 0.0005904127640403832,
      "loss": 2.4618,
      "step": 18591
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.7630256414413452,
      "learning_rate": 0.0005904117381575818,
      "loss": 2.4253,
      "step": 18592
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6245628595352173,
      "learning_rate": 0.0005904107122207876,
      "loss": 2.3958,
      "step": 18593
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6026871204376221,
      "learning_rate": 0.0005904096862300005,
      "loss": 2.3889,
      "step": 18594
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.7066110968589783,
      "learning_rate": 0.0005904086601852209,
      "loss": 2.4802,
      "step": 18595
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.660683274269104,
      "learning_rate": 0.0005904076340864486,
      "loss": 2.4918,
      "step": 18596
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.626309871673584,
      "learning_rate": 0.0005904066079336843,
      "loss": 2.4066,
      "step": 18597
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.664219081401825,
      "learning_rate": 0.0005904055817269278,
      "loss": 2.2693,
      "step": 18598
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.7263409495353699,
      "learning_rate": 0.0005904045554661794,
      "loss": 2.4445,
      "step": 18599
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6262907981872559,
      "learning_rate": 0.0005904035291514393,
      "loss": 2.4302,
      "step": 18600
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.805585503578186,
      "learning_rate": 0.0005904025027827078,
      "loss": 2.2682,
      "step": 18601
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.7309877872467041,
      "learning_rate": 0.0005904014763599848,
      "loss": 2.6116,
      "step": 18602
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.5700385570526123,
      "learning_rate": 0.0005904004498832709,
      "loss": 2.5086,
      "step": 18603
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.8001025319099426,
      "learning_rate": 0.000590399423352566,
      "loss": 2.5552,
      "step": 18604
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.713589072227478,
      "learning_rate": 0.0005903983967678703,
      "loss": 2.3679,
      "step": 18605
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.594984233379364,
      "learning_rate": 0.0005903973701291841,
      "loss": 2.5968,
      "step": 18606
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.5889256596565247,
      "learning_rate": 0.0005903963434365077,
      "loss": 2.4062,
      "step": 18607
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.8029318451881409,
      "learning_rate": 0.000590395316689841,
      "loss": 2.1958,
      "step": 18608
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.5661454200744629,
      "learning_rate": 0.0005903942898891843,
      "loss": 2.4801,
      "step": 18609
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.5529112815856934,
      "learning_rate": 0.0005903932630345378,
      "loss": 2.2553,
      "step": 18610
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6587868928909302,
      "learning_rate": 0.0005903922361259018,
      "loss": 2.5261,
      "step": 18611
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6489633321762085,
      "learning_rate": 0.0005903912091632763,
      "loss": 2.5252,
      "step": 18612
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.5667805671691895,
      "learning_rate": 0.0005903901821466617,
      "loss": 2.3664,
      "step": 18613
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6265280842781067,
      "learning_rate": 0.000590389155076058,
      "loss": 2.5409,
      "step": 18614
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6747999787330627,
      "learning_rate": 0.0005903881279514655,
      "loss": 2.4296,
      "step": 18615
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6328262686729431,
      "learning_rate": 0.0005903871007728843,
      "loss": 2.4411,
      "step": 18616
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6510121822357178,
      "learning_rate": 0.0005903860735403147,
      "loss": 2.4649,
      "step": 18617
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.7101624608039856,
      "learning_rate": 0.0005903850462537569,
      "loss": 2.4728,
      "step": 18618
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6454978585243225,
      "learning_rate": 0.0005903840189132111,
      "loss": 2.6526,
      "step": 18619
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6607129573822021,
      "learning_rate": 0.0005903829915186774,
      "loss": 2.6033,
      "step": 18620
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.5772209763526917,
      "learning_rate": 0.0005903819640701559,
      "loss": 2.4347,
      "step": 18621
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.5813984870910645,
      "learning_rate": 0.0005903809365676469,
      "loss": 2.1936,
      "step": 18622
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6829419136047363,
      "learning_rate": 0.0005903799090111508,
      "loss": 2.2483,
      "step": 18623
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.7665891051292419,
      "learning_rate": 0.0005903788814006674,
      "loss": 2.5365,
      "step": 18624
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6015447974205017,
      "learning_rate": 0.0005903778537361971,
      "loss": 2.5743,
      "step": 18625
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6077275276184082,
      "learning_rate": 0.0005903768260177403,
      "loss": 2.4906,
      "step": 18626
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.7630374431610107,
      "learning_rate": 0.0005903757982452967,
      "loss": 2.4516,
      "step": 18627
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.5746740698814392,
      "learning_rate": 0.0005903747704188669,
      "loss": 2.1733,
      "step": 18628
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.637811541557312,
      "learning_rate": 0.0005903737425384509,
      "loss": 2.2675,
      "step": 18629
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6117470264434814,
      "learning_rate": 0.0005903727146040489,
      "loss": 2.5277,
      "step": 18630
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.646888792514801,
      "learning_rate": 0.0005903716866156613,
      "loss": 2.2903,
      "step": 18631
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.5724194645881653,
      "learning_rate": 0.0005903706585732879,
      "loss": 2.5566,
      "step": 18632
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6208615303039551,
      "learning_rate": 0.0005903696304769292,
      "loss": 2.5512,
      "step": 18633
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6575355529785156,
      "learning_rate": 0.0005903686023265854,
      "loss": 2.3553,
      "step": 18634
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.5979747772216797,
      "learning_rate": 0.0005903675741222564,
      "loss": 2.2603,
      "step": 18635
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.5973946452140808,
      "learning_rate": 0.0005903665458639427,
      "loss": 2.5456,
      "step": 18636
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6294847130775452,
      "learning_rate": 0.0005903655175516445,
      "loss": 2.1891,
      "step": 18637
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.7194143533706665,
      "learning_rate": 0.0005903644891853616,
      "loss": 2.5749,
      "step": 18638
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6337725520133972,
      "learning_rate": 0.0005903634607650947,
      "loss": 2.3186,
      "step": 18639
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6054397225379944,
      "learning_rate": 0.0005903624322908436,
      "loss": 2.4289,
      "step": 18640
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.7493932843208313,
      "learning_rate": 0.0005903614037626088,
      "loss": 2.4164,
      "step": 18641
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6244384050369263,
      "learning_rate": 0.0005903603751803902,
      "loss": 2.3655,
      "step": 18642
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6111972332000732,
      "learning_rate": 0.0005903593465441881,
      "loss": 2.4799,
      "step": 18643
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.8337114453315735,
      "learning_rate": 0.0005903583178540028,
      "loss": 2.3348,
      "step": 18644
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.755578339099884,
      "learning_rate": 0.0005903572891098344,
      "loss": 2.251,
      "step": 18645
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.5669571757316589,
      "learning_rate": 0.000590356260311683,
      "loss": 2.4603,
      "step": 18646
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.7465293407440186,
      "learning_rate": 0.000590355231459549,
      "loss": 2.2446,
      "step": 18647
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6162881851196289,
      "learning_rate": 0.0005903542025534324,
      "loss": 2.4736,
      "step": 18648
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.586478054523468,
      "learning_rate": 0.0005903531735933336,
      "loss": 2.4478,
      "step": 18649
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6716135144233704,
      "learning_rate": 0.0005903521445792526,
      "loss": 2.3164,
      "step": 18650
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6557338833808899,
      "learning_rate": 0.0005903511155111897,
      "loss": 2.2566,
      "step": 18651
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6311870217323303,
      "learning_rate": 0.000590350086389145,
      "loss": 2.3901,
      "step": 18652
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.624641478061676,
      "learning_rate": 0.0005903490572131187,
      "loss": 2.3321,
      "step": 18653
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6682277917861938,
      "learning_rate": 0.0005903480279831111,
      "loss": 2.5123,
      "step": 18654
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6240561008453369,
      "learning_rate": 0.0005903469986991224,
      "loss": 2.3411,
      "step": 18655
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6121060252189636,
      "learning_rate": 0.0005903459693611525,
      "loss": 2.4767,
      "step": 18656
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.5794023275375366,
      "learning_rate": 0.0005903449399692019,
      "loss": 2.2114,
      "step": 18657
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6373031735420227,
      "learning_rate": 0.0005903439105232706,
      "loss": 2.1333,
      "step": 18658
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6168357729911804,
      "learning_rate": 0.0005903428810233591,
      "loss": 2.4065,
      "step": 18659
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.5788647532463074,
      "learning_rate": 0.0005903418514694672,
      "loss": 2.4083,
      "step": 18660
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6429316401481628,
      "learning_rate": 0.0005903408218615953,
      "loss": 2.2468,
      "step": 18661
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.755528450012207,
      "learning_rate": 0.0005903397921997436,
      "loss": 2.3469,
      "step": 18662
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6741929054260254,
      "learning_rate": 0.0005903387624839123,
      "loss": 2.2884,
      "step": 18663
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6616148948669434,
      "learning_rate": 0.0005903377327141014,
      "loss": 2.3814,
      "step": 18664
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6116428375244141,
      "learning_rate": 0.0005903367028903114,
      "loss": 2.273,
      "step": 18665
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.5963684320449829,
      "learning_rate": 0.0005903356730125422,
      "loss": 2.3512,
      "step": 18666
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6625138521194458,
      "learning_rate": 0.0005903346430807942,
      "loss": 2.3873,
      "step": 18667
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.5764678120613098,
      "learning_rate": 0.0005903336130950674,
      "loss": 2.2607,
      "step": 18668
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.5643506646156311,
      "learning_rate": 0.0005903325830553622,
      "loss": 2.4113,
      "step": 18669
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6480903029441833,
      "learning_rate": 0.0005903315529616788,
      "loss": 2.7211,
      "step": 18670
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.5923982858657837,
      "learning_rate": 0.000590330522814017,
      "loss": 2.3421,
      "step": 18671
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6058311462402344,
      "learning_rate": 0.0005903294926123774,
      "loss": 2.3819,
      "step": 18672
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.7023485898971558,
      "learning_rate": 0.0005903284623567602,
      "loss": 2.4424,
      "step": 18673
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6184532642364502,
      "learning_rate": 0.0005903274320471653,
      "loss": 2.2513,
      "step": 18674
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6696659326553345,
      "learning_rate": 0.0005903264016835931,
      "loss": 2.2489,
      "step": 18675
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6395169496536255,
      "learning_rate": 0.0005903253712660438,
      "loss": 2.3311,
      "step": 18676
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6320831179618835,
      "learning_rate": 0.0005903243407945174,
      "loss": 2.6129,
      "step": 18677
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6040910482406616,
      "learning_rate": 0.0005903233102690143,
      "loss": 2.592,
      "step": 18678
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6836581230163574,
      "learning_rate": 0.0005903222796895347,
      "loss": 2.2751,
      "step": 18679
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6178381443023682,
      "learning_rate": 0.0005903212490560786,
      "loss": 2.5417,
      "step": 18680
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.5841951370239258,
      "learning_rate": 0.0005903202183686463,
      "loss": 2.4799,
      "step": 18681
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6093471050262451,
      "learning_rate": 0.000590319187627238,
      "loss": 2.2859,
      "step": 18682
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6233540773391724,
      "learning_rate": 0.0005903181568318539,
      "loss": 2.198,
      "step": 18683
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.612794816493988,
      "learning_rate": 0.0005903171259824942,
      "loss": 2.3402,
      "step": 18684
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6330246925354004,
      "learning_rate": 0.000590316095079159,
      "loss": 2.6007,
      "step": 18685
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.8218087553977966,
      "learning_rate": 0.0005903150641218486,
      "loss": 2.3144,
      "step": 18686
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6556017994880676,
      "learning_rate": 0.0005903140331105631,
      "loss": 2.4187,
      "step": 18687
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6107372045516968,
      "learning_rate": 0.0005903130020453029,
      "loss": 2.2666,
      "step": 18688
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6967557668685913,
      "learning_rate": 0.0005903119709260679,
      "loss": 2.3294,
      "step": 18689
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6615535616874695,
      "learning_rate": 0.0005903109397528585,
      "loss": 2.484,
      "step": 18690
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6032288074493408,
      "learning_rate": 0.0005903099085256747,
      "loss": 2.3034,
      "step": 18691
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6032179594039917,
      "learning_rate": 0.0005903088772445169,
      "loss": 2.2966,
      "step": 18692
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6230179071426392,
      "learning_rate": 0.0005903078459093852,
      "loss": 2.334,
      "step": 18693
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6469153165817261,
      "learning_rate": 0.0005903068145202796,
      "loss": 2.2617,
      "step": 18694
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6586501002311707,
      "learning_rate": 0.0005903057830772007,
      "loss": 2.3784,
      "step": 18695
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6534134745597839,
      "learning_rate": 0.0005903047515801485,
      "loss": 2.525,
      "step": 18696
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6400465965270996,
      "learning_rate": 0.000590303720029123,
      "loss": 2.2685,
      "step": 18697
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6949769854545593,
      "learning_rate": 0.0005903026884241247,
      "loss": 2.3479,
      "step": 18698
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.5753885507583618,
      "learning_rate": 0.0005903016567651535,
      "loss": 2.5707,
      "step": 18699
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6108922362327576,
      "learning_rate": 0.00059030062505221,
      "loss": 2.3759,
      "step": 18700
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6231948733329773,
      "learning_rate": 0.0005902995932852939,
      "loss": 2.2416,
      "step": 18701
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6789126992225647,
      "learning_rate": 0.0005902985614644057,
      "loss": 2.3492,
      "step": 18702
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.5951964855194092,
      "learning_rate": 0.0005902975295895454,
      "loss": 2.4007,
      "step": 18703
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6009921431541443,
      "learning_rate": 0.0005902964976607134,
      "loss": 2.4359,
      "step": 18704
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6599736213684082,
      "learning_rate": 0.00059029546567791,
      "loss": 2.381,
      "step": 18705
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6264281868934631,
      "learning_rate": 0.0005902944336411349,
      "loss": 2.382,
      "step": 18706
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.5941585898399353,
      "learning_rate": 0.0005902934015503887,
      "loss": 2.3444,
      "step": 18707
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.5936942100524902,
      "learning_rate": 0.0005902923694056716,
      "loss": 2.1791,
      "step": 18708
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6399331092834473,
      "learning_rate": 0.0005902913372069835,
      "loss": 2.4638,
      "step": 18709
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6232512593269348,
      "learning_rate": 0.0005902903049543249,
      "loss": 2.2153,
      "step": 18710
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.558919370174408,
      "learning_rate": 0.0005902892726476957,
      "loss": 2.2525,
      "step": 18711
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.598900318145752,
      "learning_rate": 0.0005902882402870962,
      "loss": 2.2847,
      "step": 18712
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6176506876945496,
      "learning_rate": 0.0005902872078725268,
      "loss": 2.2173,
      "step": 18713
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6023719906806946,
      "learning_rate": 0.0005902861754039874,
      "loss": 2.3061,
      "step": 18714
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.5679218173027039,
      "learning_rate": 0.0005902851428814785,
      "loss": 2.3509,
      "step": 18715
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6290790438652039,
      "learning_rate": 0.000590284110305,
      "loss": 2.2884,
      "step": 18716
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6075629591941833,
      "learning_rate": 0.0005902830776745523,
      "loss": 2.4479,
      "step": 18717
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6103559136390686,
      "learning_rate": 0.0005902820449901353,
      "loss": 2.6923,
      "step": 18718
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6176425814628601,
      "learning_rate": 0.0005902810122517495,
      "loss": 2.3738,
      "step": 18719
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.5942279696464539,
      "learning_rate": 0.000590279979459395,
      "loss": 2.3416,
      "step": 18720
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.5776798129081726,
      "learning_rate": 0.0005902789466130719,
      "loss": 2.4264,
      "step": 18721
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.5801795721054077,
      "learning_rate": 0.0005902779137127806,
      "loss": 2.3064,
      "step": 18722
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.5980773568153381,
      "learning_rate": 0.000590276880758521,
      "loss": 2.2353,
      "step": 18723
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.5767364501953125,
      "learning_rate": 0.0005902758477502935,
      "loss": 2.2755,
      "step": 18724
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6029600501060486,
      "learning_rate": 0.0005902748146880982,
      "loss": 2.1868,
      "step": 18725
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6119039058685303,
      "learning_rate": 0.0005902737815719355,
      "loss": 2.4861,
      "step": 18726
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6016523838043213,
      "learning_rate": 0.0005902727484018052,
      "loss": 2.4049,
      "step": 18727
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.5479305982589722,
      "learning_rate": 0.0005902717151777079,
      "loss": 2.4689,
      "step": 18728
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6077131032943726,
      "learning_rate": 0.0005902706818996436,
      "loss": 2.3058,
      "step": 18729
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6133869290351868,
      "learning_rate": 0.0005902696485676124,
      "loss": 2.4715,
      "step": 18730
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6252127289772034,
      "learning_rate": 0.0005902686151816145,
      "loss": 2.4408,
      "step": 18731
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6116371750831604,
      "learning_rate": 0.0005902675817416504,
      "loss": 2.3075,
      "step": 18732
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.748119056224823,
      "learning_rate": 0.00059026654824772,
      "loss": 2.3669,
      "step": 18733
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.5935390591621399,
      "learning_rate": 0.0005902655146998235,
      "loss": 2.5136,
      "step": 18734
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6430345177650452,
      "learning_rate": 0.0005902644810979613,
      "loss": 2.1429,
      "step": 18735
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.7008723616600037,
      "learning_rate": 0.0005902634474421334,
      "loss": 2.2958,
      "step": 18736
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.5681701898574829,
      "learning_rate": 0.00059026241373234,
      "loss": 2.1619,
      "step": 18737
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6125060319900513,
      "learning_rate": 0.0005902613799685814,
      "loss": 2.2299,
      "step": 18738
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6627113819122314,
      "learning_rate": 0.0005902603461508577,
      "loss": 2.43,
      "step": 18739
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.5756826400756836,
      "learning_rate": 0.0005902593122791691,
      "loss": 2.2502,
      "step": 18740
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6084570288658142,
      "learning_rate": 0.0005902582783535159,
      "loss": 2.346,
      "step": 18741
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6137893199920654,
      "learning_rate": 0.0005902572443738981,
      "loss": 2.2985,
      "step": 18742
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.593889057636261,
      "learning_rate": 0.0005902562103403161,
      "loss": 2.4347,
      "step": 18743
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6404366493225098,
      "learning_rate": 0.0005902551762527698,
      "loss": 2.3244,
      "step": 18744
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.5921782851219177,
      "learning_rate": 0.0005902541421112597,
      "loss": 2.362,
      "step": 18745
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.5919020175933838,
      "learning_rate": 0.0005902531079157859,
      "loss": 2.5707,
      "step": 18746
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.5881884694099426,
      "learning_rate": 0.0005902520736663486,
      "loss": 2.2166,
      "step": 18747
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6398798227310181,
      "learning_rate": 0.000590251039362948,
      "loss": 2.4727,
      "step": 18748
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6009671092033386,
      "learning_rate": 0.0005902500050055841,
      "loss": 2.4536,
      "step": 18749
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.5872182250022888,
      "learning_rate": 0.0005902489705942573,
      "loss": 2.3815,
      "step": 18750
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6384440660476685,
      "learning_rate": 0.0005902479361289678,
      "loss": 2.6505,
      "step": 18751
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6280556917190552,
      "learning_rate": 0.0005902469016097157,
      "loss": 2.4518,
      "step": 18752
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.652438759803772,
      "learning_rate": 0.0005902458670365011,
      "loss": 2.371,
      "step": 18753
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6210436820983887,
      "learning_rate": 0.0005902448324093245,
      "loss": 2.4363,
      "step": 18754
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6293069124221802,
      "learning_rate": 0.0005902437977281858,
      "loss": 2.428,
      "step": 18755
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6210616230964661,
      "learning_rate": 0.0005902427629930853,
      "loss": 2.3089,
      "step": 18756
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6905114054679871,
      "learning_rate": 0.0005902417282040233,
      "loss": 2.566,
      "step": 18757
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6221113204956055,
      "learning_rate": 0.0005902406933609998,
      "loss": 2.468,
      "step": 18758
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6167081594467163,
      "learning_rate": 0.000590239658464015,
      "loss": 2.3517,
      "step": 18759
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6113839745521545,
      "learning_rate": 0.0005902386235130693,
      "loss": 2.1181,
      "step": 18760
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.5893325209617615,
      "learning_rate": 0.0005902375885081626,
      "loss": 2.3866,
      "step": 18761
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6179377436637878,
      "learning_rate": 0.0005902365534492955,
      "loss": 2.3294,
      "step": 18762
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6451270580291748,
      "learning_rate": 0.0005902355183364678,
      "loss": 2.3961,
      "step": 18763
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6845940947532654,
      "learning_rate": 0.0005902344831696797,
      "loss": 2.5559,
      "step": 18764
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6323817372322083,
      "learning_rate": 0.0005902334479489318,
      "loss": 2.4569,
      "step": 18765
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.5287951231002808,
      "learning_rate": 0.0005902324126742239,
      "loss": 2.2423,
      "step": 18766
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6396123766899109,
      "learning_rate": 0.0005902313773455563,
      "loss": 2.3058,
      "step": 18767
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6352444291114807,
      "learning_rate": 0.0005902303419629292,
      "loss": 2.4313,
      "step": 18768
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.635901927947998,
      "learning_rate": 0.0005902293065263429,
      "loss": 2.2282,
      "step": 18769
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.5758444666862488,
      "learning_rate": 0.0005902282710357974,
      "loss": 2.1875,
      "step": 18770
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6187465190887451,
      "learning_rate": 0.000590227235491293,
      "loss": 2.5304,
      "step": 18771
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6598175168037415,
      "learning_rate": 0.0005902261998928298,
      "loss": 2.2274,
      "step": 18772
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6731622219085693,
      "learning_rate": 0.0005902251642404082,
      "loss": 2.4083,
      "step": 18773
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.7551484704017639,
      "learning_rate": 0.0005902241285340282,
      "loss": 2.6493,
      "step": 18774
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6599305868148804,
      "learning_rate": 0.0005902230927736901,
      "loss": 2.4071,
      "step": 18775
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.7165205478668213,
      "learning_rate": 0.0005902220569593941,
      "loss": 2.2653,
      "step": 18776
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.57529217004776,
      "learning_rate": 0.0005902210210911402,
      "loss": 2.3035,
      "step": 18777
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6732574105262756,
      "learning_rate": 0.0005902199851689288,
      "loss": 2.4038,
      "step": 18778
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6873428225517273,
      "learning_rate": 0.00059021894919276,
      "loss": 2.2778,
      "step": 18779
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.5945762395858765,
      "learning_rate": 0.000590217913162634,
      "loss": 2.0532,
      "step": 18780
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6200391054153442,
      "learning_rate": 0.0005902168770785511,
      "loss": 2.6262,
      "step": 18781
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6775554418563843,
      "learning_rate": 0.0005902158409405114,
      "loss": 2.4491,
      "step": 18782
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6505280137062073,
      "learning_rate": 0.000590214804748515,
      "loss": 2.2209,
      "step": 18783
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.5708056092262268,
      "learning_rate": 0.0005902137685025623,
      "loss": 2.2722,
      "step": 18784
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.5930577516555786,
      "learning_rate": 0.0005902127322026533,
      "loss": 2.5177,
      "step": 18785
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.729087233543396,
      "learning_rate": 0.0005902116958487883,
      "loss": 2.3972,
      "step": 18786
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.8114675283432007,
      "learning_rate": 0.0005902106594409674,
      "loss": 2.3744,
      "step": 18787
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6344621181488037,
      "learning_rate": 0.0005902096229791911,
      "loss": 2.4251,
      "step": 18788
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.682141125202179,
      "learning_rate": 0.0005902085864634591,
      "loss": 2.4732,
      "step": 18789
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.8892878890037537,
      "learning_rate": 0.0005902075498937719,
      "loss": 2.46,
      "step": 18790
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.639056921005249,
      "learning_rate": 0.0005902065132701298,
      "loss": 2.3745,
      "step": 18791
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.7424671649932861,
      "learning_rate": 0.0005902054765925326,
      "loss": 2.5725,
      "step": 18792
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.8660399317741394,
      "learning_rate": 0.0005902044398609809,
      "loss": 2.1455,
      "step": 18793
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6808276176452637,
      "learning_rate": 0.0005902034030754746,
      "loss": 2.5593,
      "step": 18794
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.661064624786377,
      "learning_rate": 0.000590202366236014,
      "loss": 2.3761,
      "step": 18795
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6426891684532166,
      "learning_rate": 0.0005902013293425994,
      "loss": 2.3503,
      "step": 18796
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.627450168132782,
      "learning_rate": 0.0005902002923952308,
      "loss": 2.4097,
      "step": 18797
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.5949242115020752,
      "learning_rate": 0.0005901992553939085,
      "loss": 2.39,
      "step": 18798
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.7094790935516357,
      "learning_rate": 0.0005901982183386328,
      "loss": 2.4813,
      "step": 18799
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6228352785110474,
      "learning_rate": 0.0005901971812294037,
      "loss": 2.4843,
      "step": 18800
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.7479145526885986,
      "learning_rate": 0.0005901961440662214,
      "loss": 2.2526,
      "step": 18801
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6277294158935547,
      "learning_rate": 0.0005901951068490862,
      "loss": 2.232,
      "step": 18802
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6479378938674927,
      "learning_rate": 0.0005901940695779983,
      "loss": 2.5307,
      "step": 18803
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.8438276052474976,
      "learning_rate": 0.0005901930322529578,
      "loss": 2.3988,
      "step": 18804
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.5932357311248779,
      "learning_rate": 0.000590191994873965,
      "loss": 2.4086,
      "step": 18805
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.5928816199302673,
      "learning_rate": 0.00059019095744102,
      "loss": 2.1267,
      "step": 18806
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.7033454775810242,
      "learning_rate": 0.0005901899199541229,
      "loss": 2.2766,
      "step": 18807
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6107547283172607,
      "learning_rate": 0.0005901888824132741,
      "loss": 2.3085,
      "step": 18808
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.619665265083313,
      "learning_rate": 0.0005901878448184737,
      "loss": 2.1566,
      "step": 18809
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.0248912572860718,
      "learning_rate": 0.000590186807169722,
      "loss": 2.2633,
      "step": 18810
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6228656768798828,
      "learning_rate": 0.000590185769467019,
      "loss": 2.4862,
      "step": 18811
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6354923844337463,
      "learning_rate": 0.000590184731710365,
      "loss": 2.3393,
      "step": 18812
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6182696223258972,
      "learning_rate": 0.0005901836938997601,
      "loss": 2.2855,
      "step": 18813
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.596555769443512,
      "learning_rate": 0.0005901826560352048,
      "loss": 2.3327,
      "step": 18814
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.5551061630249023,
      "learning_rate": 0.0005901816181166988,
      "loss": 2.2348,
      "step": 18815
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6214603781700134,
      "learning_rate": 0.0005901805801442428,
      "loss": 2.4476,
      "step": 18816
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6213133335113525,
      "learning_rate": 0.0005901795421178367,
      "loss": 2.4456,
      "step": 18817
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6537750959396362,
      "learning_rate": 0.0005901785040374806,
      "loss": 2.4603,
      "step": 18818
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6532503366470337,
      "learning_rate": 0.0005901774659031749,
      "loss": 2.4206,
      "step": 18819
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6387225985527039,
      "learning_rate": 0.0005901764277149198,
      "loss": 2.1814,
      "step": 18820
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6542513966560364,
      "learning_rate": 0.0005901753894727154,
      "loss": 2.5574,
      "step": 18821
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.5944386124610901,
      "learning_rate": 0.0005901743511765618,
      "loss": 2.2348,
      "step": 18822
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.625281810760498,
      "learning_rate": 0.0005901733128264595,
      "loss": 2.2957,
      "step": 18823
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6027037501335144,
      "learning_rate": 0.0005901722744224084,
      "loss": 2.3262,
      "step": 18824
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.5914575457572937,
      "learning_rate": 0.0005901712359644088,
      "loss": 2.7147,
      "step": 18825
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6350815892219543,
      "learning_rate": 0.000590170197452461,
      "loss": 2.1616,
      "step": 18826
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.5753756165504456,
      "learning_rate": 0.0005901691588865649,
      "loss": 2.3517,
      "step": 18827
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.5462006330490112,
      "learning_rate": 0.000590168120266721,
      "loss": 2.1911,
      "step": 18828
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.5849543809890747,
      "learning_rate": 0.0005901670815929294,
      "loss": 2.2267,
      "step": 18829
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.5922539830207825,
      "learning_rate": 0.0005901660428651901,
      "loss": 2.404,
      "step": 18830
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.7112520337104797,
      "learning_rate": 0.0005901650040835035,
      "loss": 1.996,
      "step": 18831
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.5881611108779907,
      "learning_rate": 0.0005901639652478699,
      "loss": 2.2328,
      "step": 18832
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6332485675811768,
      "learning_rate": 0.0005901629263582892,
      "loss": 2.4593,
      "step": 18833
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6001299023628235,
      "learning_rate": 0.0005901618874147618,
      "loss": 2.3453,
      "step": 18834
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6064125299453735,
      "learning_rate": 0.0005901608484172878,
      "loss": 2.4961,
      "step": 18835
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.625855565071106,
      "learning_rate": 0.0005901598093658673,
      "loss": 2.4856,
      "step": 18836
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6350751519203186,
      "learning_rate": 0.0005901587702605008,
      "loss": 2.5545,
      "step": 18837
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.4353216886520386,
      "learning_rate": 0.0005901577311011883,
      "loss": 2.2053,
      "step": 18838
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.5582419037818909,
      "learning_rate": 0.00059015669188793,
      "loss": 2.4466,
      "step": 18839
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6680046916007996,
      "learning_rate": 0.0005901556526207259,
      "loss": 2.3292,
      "step": 18840
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.5805051922798157,
      "learning_rate": 0.0005901546132995766,
      "loss": 2.3886,
      "step": 18841
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6815705299377441,
      "learning_rate": 0.000590153573924482,
      "loss": 2.1286,
      "step": 18842
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6435818672180176,
      "learning_rate": 0.0005901525344954423,
      "loss": 2.227,
      "step": 18843
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6770961880683899,
      "learning_rate": 0.0005901514950124579,
      "loss": 2.3374,
      "step": 18844
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6402919888496399,
      "learning_rate": 0.0005901504554755288,
      "loss": 2.2925,
      "step": 18845
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6427971720695496,
      "learning_rate": 0.0005901494158846553,
      "loss": 2.5344,
      "step": 18846
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6910325884819031,
      "learning_rate": 0.0005901483762398374,
      "loss": 2.3537,
      "step": 18847
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6180546879768372,
      "learning_rate": 0.0005901473365410757,
      "loss": 2.4005,
      "step": 18848
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.5982248187065125,
      "learning_rate": 0.0005901462967883699,
      "loss": 2.1338,
      "step": 18849
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.8212851285934448,
      "learning_rate": 0.0005901452569817206,
      "loss": 2.1227,
      "step": 18850
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.684307873249054,
      "learning_rate": 0.0005901442171211276,
      "loss": 2.3227,
      "step": 18851
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6845945119857788,
      "learning_rate": 0.0005901431772065914,
      "loss": 2.2297,
      "step": 18852
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6436927914619446,
      "learning_rate": 0.0005901421372381121,
      "loss": 2.2063,
      "step": 18853
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6415500640869141,
      "learning_rate": 0.0005901410972156901,
      "loss": 2.4993,
      "step": 18854
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6428177356719971,
      "learning_rate": 0.0005901400571393252,
      "loss": 2.2055,
      "step": 18855
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6633723974227905,
      "learning_rate": 0.0005901390170090178,
      "loss": 2.2652,
      "step": 18856
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6178169846534729,
      "learning_rate": 0.0005901379768247681,
      "loss": 2.4567,
      "step": 18857
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6080141663551331,
      "learning_rate": 0.0005901369365865763,
      "loss": 2.4677,
      "step": 18858
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.734727144241333,
      "learning_rate": 0.0005901358962944425,
      "loss": 2.1964,
      "step": 18859
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6488398909568787,
      "learning_rate": 0.0005901348559483671,
      "loss": 2.3014,
      "step": 18860
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6416381597518921,
      "learning_rate": 0.0005901338155483499,
      "loss": 2.3214,
      "step": 18861
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6507130265235901,
      "learning_rate": 0.0005901327750943916,
      "loss": 2.2318,
      "step": 18862
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.5988208651542664,
      "learning_rate": 0.000590131734586492,
      "loss": 2.3592,
      "step": 18863
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.5720009207725525,
      "learning_rate": 0.0005901306940246516,
      "loss": 2.4124,
      "step": 18864
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.633693277835846,
      "learning_rate": 0.0005901296534088702,
      "loss": 2.4608,
      "step": 18865
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.649160623550415,
      "learning_rate": 0.0005901286127391484,
      "loss": 2.2802,
      "step": 18866
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.5707831382751465,
      "learning_rate": 0.000590127572015486,
      "loss": 2.2621,
      "step": 18867
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6263837218284607,
      "learning_rate": 0.0005901265312378837,
      "loss": 2.2673,
      "step": 18868
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6165306568145752,
      "learning_rate": 0.0005901254904063412,
      "loss": 2.3841,
      "step": 18869
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6740937232971191,
      "learning_rate": 0.0005901244495208589,
      "loss": 2.5009,
      "step": 18870
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6075131893157959,
      "learning_rate": 0.000590123408581437,
      "loss": 2.26,
      "step": 18871
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.657950758934021,
      "learning_rate": 0.0005901223675880759,
      "loss": 2.3167,
      "step": 18872
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6270225048065186,
      "learning_rate": 0.0005901213265407753,
      "loss": 2.5425,
      "step": 18873
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6368448138237,
      "learning_rate": 0.0005901202854395357,
      "loss": 2.3462,
      "step": 18874
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.567882239818573,
      "learning_rate": 0.0005901192442843573,
      "loss": 2.1514,
      "step": 18875
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6090810298919678,
      "learning_rate": 0.0005901182030752403,
      "loss": 2.201,
      "step": 18876
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.5753006935119629,
      "learning_rate": 0.0005901171618121849,
      "loss": 2.5599,
      "step": 18877
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6187876462936401,
      "learning_rate": 0.0005901161204951911,
      "loss": 2.456,
      "step": 18878
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6663223505020142,
      "learning_rate": 0.0005901150791242593,
      "loss": 2.4544,
      "step": 18879
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6712798476219177,
      "learning_rate": 0.0005901140376993897,
      "loss": 2.1516,
      "step": 18880
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.5759031176567078,
      "learning_rate": 0.0005901129962205823,
      "loss": 2.4066,
      "step": 18881
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6369487047195435,
      "learning_rate": 0.0005901119546878375,
      "loss": 2.4027,
      "step": 18882
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.5746141672134399,
      "learning_rate": 0.0005901109131011554,
      "loss": 2.2103,
      "step": 18883
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.5996376872062683,
      "learning_rate": 0.0005901098714605362,
      "loss": 2.4328,
      "step": 18884
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.5743860006332397,
      "learning_rate": 0.0005901088297659801,
      "loss": 2.3014,
      "step": 18885
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6196907758712769,
      "learning_rate": 0.0005901077880174873,
      "loss": 2.5456,
      "step": 18886
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.5821970105171204,
      "learning_rate": 0.000590106746215058,
      "loss": 2.4873,
      "step": 18887
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6423284411430359,
      "learning_rate": 0.0005901057043586923,
      "loss": 2.3747,
      "step": 18888
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6309012174606323,
      "learning_rate": 0.0005901046624483906,
      "loss": 2.2504,
      "step": 18889
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6251806616783142,
      "learning_rate": 0.0005901036204841528,
      "loss": 2.5501,
      "step": 18890
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6029586791992188,
      "learning_rate": 0.0005901025784659793,
      "loss": 2.2776,
      "step": 18891
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.8387243151664734,
      "learning_rate": 0.0005901015363938704,
      "loss": 2.389,
      "step": 18892
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.576022207736969,
      "learning_rate": 0.0005901004942678261,
      "loss": 2.3964,
      "step": 18893
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.638212263584137,
      "learning_rate": 0.0005900994520878466,
      "loss": 2.5036,
      "step": 18894
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.5825662612915039,
      "learning_rate": 0.0005900984098539321,
      "loss": 2.2954,
      "step": 18895
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6610274314880371,
      "learning_rate": 0.0005900973675660829,
      "loss": 2.5032,
      "step": 18896
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6402155756950378,
      "learning_rate": 0.000590096325224299,
      "loss": 2.3589,
      "step": 18897
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.5676382184028625,
      "learning_rate": 0.0005900952828285809,
      "loss": 2.3253,
      "step": 18898
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6055352091789246,
      "learning_rate": 0.0005900942403789285,
      "loss": 2.2934,
      "step": 18899
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6896239519119263,
      "learning_rate": 0.0005900931978753421,
      "loss": 2.2538,
      "step": 18900
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6853751540184021,
      "learning_rate": 0.0005900921553178219,
      "loss": 2.4466,
      "step": 18901
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.7560638189315796,
      "learning_rate": 0.0005900911127063682,
      "loss": 2.1874,
      "step": 18902
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.7166817784309387,
      "learning_rate": 0.000590090070040981,
      "loss": 2.4384,
      "step": 18903
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.5861034989356995,
      "learning_rate": 0.0005900890273216605,
      "loss": 2.2501,
      "step": 18904
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6257449984550476,
      "learning_rate": 0.0005900879845484071,
      "loss": 2.4067,
      "step": 18905
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6425216794013977,
      "learning_rate": 0.0005900869417212207,
      "loss": 2.3152,
      "step": 18906
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6261116862297058,
      "learning_rate": 0.0005900858988401018,
      "loss": 2.236,
      "step": 18907
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.5333425998687744,
      "learning_rate": 0.0005900848559050506,
      "loss": 2.5526,
      "step": 18908
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6618121266365051,
      "learning_rate": 0.0005900838129160669,
      "loss": 2.3368,
      "step": 18909
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6814447045326233,
      "learning_rate": 0.0005900827698731513,
      "loss": 2.2714,
      "step": 18910
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.5899161696434021,
      "learning_rate": 0.0005900817267763036,
      "loss": 2.2218,
      "step": 18911
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.607655942440033,
      "learning_rate": 0.0005900806836255244,
      "loss": 2.3161,
      "step": 18912
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6793384552001953,
      "learning_rate": 0.0005900796404208138,
      "loss": 2.4378,
      "step": 18913
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.623109757900238,
      "learning_rate": 0.0005900785971621719,
      "loss": 2.4374,
      "step": 18914
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.641573965549469,
      "learning_rate": 0.0005900775538495988,
      "loss": 2.3882,
      "step": 18915
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.5696762800216675,
      "learning_rate": 0.0005900765104830948,
      "loss": 2.5828,
      "step": 18916
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6095738410949707,
      "learning_rate": 0.0005900754670626603,
      "loss": 2.5549,
      "step": 18917
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.633955180644989,
      "learning_rate": 0.0005900744235882951,
      "loss": 2.3413,
      "step": 18918
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6062475442886353,
      "learning_rate": 0.0005900733800599996,
      "loss": 2.3265,
      "step": 18919
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6354842782020569,
      "learning_rate": 0.0005900723364777741,
      "loss": 2.5249,
      "step": 18920
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6888893842697144,
      "learning_rate": 0.0005900712928416186,
      "loss": 2.2728,
      "step": 18921
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6277602910995483,
      "learning_rate": 0.0005900702491515333,
      "loss": 2.6416,
      "step": 18922
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6707834005355835,
      "learning_rate": 0.0005900692054075186,
      "loss": 2.5656,
      "step": 18923
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6042405962944031,
      "learning_rate": 0.0005900681616095746,
      "loss": 2.3519,
      "step": 18924
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6384168267250061,
      "learning_rate": 0.0005900671177577013,
      "loss": 2.4616,
      "step": 18925
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.655475914478302,
      "learning_rate": 0.0005900660738518991,
      "loss": 2.4568,
      "step": 18926
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.5740047097206116,
      "learning_rate": 0.0005900650298921681,
      "loss": 2.472,
      "step": 18927
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6141355037689209,
      "learning_rate": 0.0005900639858785087,
      "loss": 2.3707,
      "step": 18928
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.5953922867774963,
      "learning_rate": 0.0005900629418109207,
      "loss": 2.1852,
      "step": 18929
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.5884532332420349,
      "learning_rate": 0.0005900618976894047,
      "loss": 2.2981,
      "step": 18930
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.5771907567977905,
      "learning_rate": 0.0005900608535139606,
      "loss": 2.3708,
      "step": 18931
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.5708381533622742,
      "learning_rate": 0.0005900598092845888,
      "loss": 2.4701,
      "step": 18932
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.598308265209198,
      "learning_rate": 0.0005900587650012894,
      "loss": 2.3804,
      "step": 18933
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6322420835494995,
      "learning_rate": 0.0005900577206640626,
      "loss": 2.5583,
      "step": 18934
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6393530964851379,
      "learning_rate": 0.0005900566762729085,
      "loss": 2.4347,
      "step": 18935
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.5803239941596985,
      "learning_rate": 0.0005900556318278275,
      "loss": 2.3592,
      "step": 18936
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6241784691810608,
      "learning_rate": 0.0005900545873288196,
      "loss": 2.3016,
      "step": 18937
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.5740824937820435,
      "learning_rate": 0.0005900535427758851,
      "loss": 2.3846,
      "step": 18938
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6052505373954773,
      "learning_rate": 0.0005900524981690242,
      "loss": 2.2482,
      "step": 18939
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.632594883441925,
      "learning_rate": 0.000590051453508237,
      "loss": 2.3402,
      "step": 18940
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.63545823097229,
      "learning_rate": 0.0005900504087935238,
      "loss": 2.502,
      "step": 18941
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.590461790561676,
      "learning_rate": 0.0005900493640248848,
      "loss": 2.3693,
      "step": 18942
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6082057952880859,
      "learning_rate": 0.0005900483192023201,
      "loss": 2.5377,
      "step": 18943
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6207879781723022,
      "learning_rate": 0.00059004727432583,
      "loss": 2.5344,
      "step": 18944
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6287047863006592,
      "learning_rate": 0.0005900462293954146,
      "loss": 2.4942,
      "step": 18945
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.631843090057373,
      "learning_rate": 0.0005900451844110741,
      "loss": 2.2171,
      "step": 18946
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6815001964569092,
      "learning_rate": 0.0005900441393728087,
      "loss": 2.0733,
      "step": 18947
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.5906294584274292,
      "learning_rate": 0.0005900430942806187,
      "loss": 2.5339,
      "step": 18948
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.7158602476119995,
      "learning_rate": 0.0005900420491345042,
      "loss": 2.5437,
      "step": 18949
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6002002954483032,
      "learning_rate": 0.0005900410039344654,
      "loss": 2.6491,
      "step": 18950
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.7842938303947449,
      "learning_rate": 0.0005900399586805025,
      "loss": 2.4587,
      "step": 18951
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.5789796113967896,
      "learning_rate": 0.0005900389133726157,
      "loss": 2.2981,
      "step": 18952
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6377108693122864,
      "learning_rate": 0.0005900378680108052,
      "loss": 2.3508,
      "step": 18953
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6620407700538635,
      "learning_rate": 0.0005900368225950711,
      "loss": 2.3073,
      "step": 18954
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6071254014968872,
      "learning_rate": 0.0005900357771254138,
      "loss": 2.295,
      "step": 18955
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.7267528176307678,
      "learning_rate": 0.0005900347316018333,
      "loss": 2.5182,
      "step": 18956
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.5799949765205383,
      "learning_rate": 0.0005900336860243299,
      "loss": 2.2296,
      "step": 18957
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.7055584788322449,
      "learning_rate": 0.0005900326403929037,
      "loss": 2.3501,
      "step": 18958
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6725634932518005,
      "learning_rate": 0.000590031594707555,
      "loss": 2.414,
      "step": 18959
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.5506527423858643,
      "learning_rate": 0.000590030548968284,
      "loss": 2.2763,
      "step": 18960
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.7205870151519775,
      "learning_rate": 0.0005900295031750908,
      "loss": 2.4038,
      "step": 18961
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.7374616861343384,
      "learning_rate": 0.0005900284573279757,
      "loss": 2.4489,
      "step": 18962
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.5933261513710022,
      "learning_rate": 0.0005900274114269387,
      "loss": 2.4969,
      "step": 18963
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.719859778881073,
      "learning_rate": 0.0005900263654719803,
      "loss": 2.4634,
      "step": 18964
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6899721026420593,
      "learning_rate": 0.0005900253194631004,
      "loss": 2.2611,
      "step": 18965
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.5728422403335571,
      "learning_rate": 0.0005900242734002993,
      "loss": 2.4555,
      "step": 18966
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6804852485656738,
      "learning_rate": 0.0005900232272835773,
      "loss": 2.4517,
      "step": 18967
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6568480730056763,
      "learning_rate": 0.0005900221811129345,
      "loss": 2.4079,
      "step": 18968
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.7552019953727722,
      "learning_rate": 0.000590021134888371,
      "loss": 2.2646,
      "step": 18969
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.568699061870575,
      "learning_rate": 0.0005900200886098872,
      "loss": 2.373,
      "step": 18970
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.9637249112129211,
      "learning_rate": 0.0005900190422774833,
      "loss": 2.202,
      "step": 18971
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6614163517951965,
      "learning_rate": 0.0005900179958911591,
      "loss": 2.2839,
      "step": 18972
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.638777494430542,
      "learning_rate": 0.0005900169494509152,
      "loss": 2.4584,
      "step": 18973
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6086466312408447,
      "learning_rate": 0.0005900159029567517,
      "loss": 2.2913,
      "step": 18974
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6766548156738281,
      "learning_rate": 0.0005900148564086688,
      "loss": 2.1776,
      "step": 18975
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.5960213541984558,
      "learning_rate": 0.0005900138098066666,
      "loss": 2.3491,
      "step": 18976
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.603029191493988,
      "learning_rate": 0.0005900127631507455,
      "loss": 2.4481,
      "step": 18977
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.7092415690422058,
      "learning_rate": 0.0005900117164409055,
      "loss": 2.4204,
      "step": 18978
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.5843710899353027,
      "learning_rate": 0.0005900106696771467,
      "loss": 2.3171,
      "step": 18979
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6090735793113708,
      "learning_rate": 0.0005900096228594695,
      "loss": 2.5118,
      "step": 18980
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.7493252158164978,
      "learning_rate": 0.0005900085759878741,
      "loss": 2.4481,
      "step": 18981
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6403782963752747,
      "learning_rate": 0.0005900075290623606,
      "loss": 2.6524,
      "step": 18982
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6600384712219238,
      "learning_rate": 0.0005900064820829293,
      "loss": 2.3777,
      "step": 18983
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6214821934700012,
      "learning_rate": 0.0005900054350495802,
      "loss": 2.3115,
      "step": 18984
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6968660354614258,
      "learning_rate": 0.0005900043879623137,
      "loss": 2.3704,
      "step": 18985
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6281618475914001,
      "learning_rate": 0.0005900033408211299,
      "loss": 2.2894,
      "step": 18986
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6566017270088196,
      "learning_rate": 0.0005900022936260289,
      "loss": 2.4412,
      "step": 18987
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6480258703231812,
      "learning_rate": 0.0005900012463770111,
      "loss": 2.3789,
      "step": 18988
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6119734048843384,
      "learning_rate": 0.0005900001990740766,
      "loss": 2.5375,
      "step": 18989
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6152169108390808,
      "learning_rate": 0.0005899991517172256,
      "loss": 2.0756,
      "step": 18990
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6529128551483154,
      "learning_rate": 0.0005899981043064582,
      "loss": 2.6296,
      "step": 18991
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.5811052918434143,
      "learning_rate": 0.0005899970568417748,
      "loss": 2.5154,
      "step": 18992
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6091244220733643,
      "learning_rate": 0.0005899960093231753,
      "loss": 2.4906,
      "step": 18993
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.5800988078117371,
      "learning_rate": 0.0005899949617506603,
      "loss": 2.3987,
      "step": 18994
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6644338369369507,
      "learning_rate": 0.0005899939141242296,
      "loss": 2.3105,
      "step": 18995
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.664438784122467,
      "learning_rate": 0.0005899928664438836,
      "loss": 2.2606,
      "step": 18996
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.5880032777786255,
      "learning_rate": 0.0005899918187096223,
      "loss": 2.4284,
      "step": 18997
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6563414335250854,
      "learning_rate": 0.0005899907709214462,
      "loss": 2.5695,
      "step": 18998
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6753334403038025,
      "learning_rate": 0.0005899897230793554,
      "loss": 2.3458,
      "step": 18999
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.5984131097793579,
      "learning_rate": 0.0005899886751833499,
      "loss": 2.203,
      "step": 19000
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6393370032310486,
      "learning_rate": 0.0005899876272334302,
      "loss": 2.2898,
      "step": 19001
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.7020742893218994,
      "learning_rate": 0.0005899865792295962,
      "loss": 2.3571,
      "step": 19002
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6027427315711975,
      "learning_rate": 0.0005899855311718482,
      "loss": 2.3132,
      "step": 19003
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6026108860969543,
      "learning_rate": 0.0005899844830601865,
      "loss": 2.303,
      "step": 19004
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6144483685493469,
      "learning_rate": 0.0005899834348946111,
      "loss": 2.2842,
      "step": 19005
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6826854944229126,
      "learning_rate": 0.0005899823866751224,
      "loss": 2.5759,
      "step": 19006
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6063036918640137,
      "learning_rate": 0.0005899813384017205,
      "loss": 2.4238,
      "step": 19007
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6343251466751099,
      "learning_rate": 0.0005899802900744055,
      "loss": 2.369,
      "step": 19008
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6746459007263184,
      "learning_rate": 0.0005899792416931779,
      "loss": 2.329,
      "step": 19009
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6266413331031799,
      "learning_rate": 0.0005899781932580375,
      "loss": 2.103,
      "step": 19010
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.5901556611061096,
      "learning_rate": 0.0005899771447689847,
      "loss": 2.53,
      "step": 19011
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.9008921980857849,
      "learning_rate": 0.0005899760962260196,
      "loss": 2.5198,
      "step": 19012
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.5929195284843445,
      "learning_rate": 0.0005899750476291427,
      "loss": 2.1773,
      "step": 19013
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6115415096282959,
      "learning_rate": 0.0005899739989783539,
      "loss": 2.2086,
      "step": 19014
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6429992318153381,
      "learning_rate": 0.0005899729502736533,
      "loss": 2.2885,
      "step": 19015
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.565546989440918,
      "learning_rate": 0.0005899719015150413,
      "loss": 2.3394,
      "step": 19016
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6255471110343933,
      "learning_rate": 0.0005899708527025181,
      "loss": 2.2481,
      "step": 19017
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6519847512245178,
      "learning_rate": 0.0005899698038360838,
      "loss": 2.4153,
      "step": 19018
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6243886351585388,
      "learning_rate": 0.0005899687549157387,
      "loss": 2.2339,
      "step": 19019
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6104297041893005,
      "learning_rate": 0.0005899677059414831,
      "loss": 2.2744,
      "step": 19020
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.5972732901573181,
      "learning_rate": 0.0005899666569133167,
      "loss": 2.4596,
      "step": 19021
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6371461153030396,
      "learning_rate": 0.0005899656078312402,
      "loss": 2.3345,
      "step": 19022
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6290237903594971,
      "learning_rate": 0.0005899645586952536,
      "loss": 2.2408,
      "step": 19023
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6777938008308411,
      "learning_rate": 0.0005899635095053572,
      "loss": 2.3331,
      "step": 19024
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.8425570130348206,
      "learning_rate": 0.000589962460261551,
      "loss": 2.3086,
      "step": 19025
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6749793887138367,
      "learning_rate": 0.0005899614109638353,
      "loss": 2.3339,
      "step": 19026
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.585016667842865,
      "learning_rate": 0.0005899603616122103,
      "loss": 2.3332,
      "step": 19027
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6315187811851501,
      "learning_rate": 0.0005899593122066762,
      "loss": 2.3721,
      "step": 19028
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6125847101211548,
      "learning_rate": 0.0005899582627472333,
      "loss": 2.2727,
      "step": 19029
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6193244457244873,
      "learning_rate": 0.0005899572132338816,
      "loss": 2.4121,
      "step": 19030
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.7682043313980103,
      "learning_rate": 0.0005899561636666213,
      "loss": 2.2955,
      "step": 19031
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6037091612815857,
      "learning_rate": 0.0005899551140454527,
      "loss": 2.1231,
      "step": 19032
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.5884033441543579,
      "learning_rate": 0.0005899540643703761,
      "loss": 2.4781,
      "step": 19033
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6117624640464783,
      "learning_rate": 0.0005899530146413916,
      "loss": 2.3272,
      "step": 19034
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6317802667617798,
      "learning_rate": 0.0005899519648584992,
      "loss": 2.3781,
      "step": 19035
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6478883028030396,
      "learning_rate": 0.0005899509150216993,
      "loss": 2.4844,
      "step": 19036
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.5897660851478577,
      "learning_rate": 0.000589949865130992,
      "loss": 2.4653,
      "step": 19037
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.5822965502738953,
      "learning_rate": 0.0005899488151863777,
      "loss": 2.2789,
      "step": 19038
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6059053540229797,
      "learning_rate": 0.0005899477651878563,
      "loss": 2.2171,
      "step": 19039
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6115894913673401,
      "learning_rate": 0.0005899467151354281,
      "loss": 2.5181,
      "step": 19040
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.5522167086601257,
      "learning_rate": 0.0005899456650290935,
      "loss": 2.4548,
      "step": 19041
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.630868136882782,
      "learning_rate": 0.0005899446148688524,
      "loss": 2.5356,
      "step": 19042
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6037421226501465,
      "learning_rate": 0.0005899435646547052,
      "loss": 2.3337,
      "step": 19043
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6014372110366821,
      "learning_rate": 0.000589942514386652,
      "loss": 2.3739,
      "step": 19044
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.605154812335968,
      "learning_rate": 0.0005899414640646929,
      "loss": 2.3792,
      "step": 19045
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.5983595848083496,
      "learning_rate": 0.0005899404136888284,
      "loss": 2.3544,
      "step": 19046
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.5890485048294067,
      "learning_rate": 0.0005899393632590584,
      "loss": 2.3935,
      "step": 19047
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.7054840922355652,
      "learning_rate": 0.0005899383127753832,
      "loss": 2.5079,
      "step": 19048
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6141728758811951,
      "learning_rate": 0.000589937262237803,
      "loss": 2.7038,
      "step": 19049
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6250135898590088,
      "learning_rate": 0.000589936211646318,
      "loss": 2.5063,
      "step": 19050
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6043050289154053,
      "learning_rate": 0.0005899351610009284,
      "loss": 2.6694,
      "step": 19051
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6042745113372803,
      "learning_rate": 0.0005899341103016343,
      "loss": 2.2266,
      "step": 19052
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.5940436124801636,
      "learning_rate": 0.0005899330595484361,
      "loss": 2.2793,
      "step": 19053
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.595437228679657,
      "learning_rate": 0.0005899320087413339,
      "loss": 2.233,
      "step": 19054
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6667682528495789,
      "learning_rate": 0.0005899309578803277,
      "loss": 2.3123,
      "step": 19055
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6237764358520508,
      "learning_rate": 0.0005899299069654179,
      "loss": 2.3028,
      "step": 19056
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.563632071018219,
      "learning_rate": 0.0005899288559966047,
      "loss": 2.2473,
      "step": 19057
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6374741196632385,
      "learning_rate": 0.0005899278049738883,
      "loss": 2.6813,
      "step": 19058
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.5791323781013489,
      "learning_rate": 0.0005899267538972688,
      "loss": 2.2865,
      "step": 19059
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6437785029411316,
      "learning_rate": 0.0005899257027667464,
      "loss": 2.5536,
      "step": 19060
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.582248330116272,
      "learning_rate": 0.0005899246515823214,
      "loss": 2.3337,
      "step": 19061
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6036053895950317,
      "learning_rate": 0.0005899236003439939,
      "loss": 2.3188,
      "step": 19062
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.628320038318634,
      "learning_rate": 0.0005899225490517642,
      "loss": 2.5017,
      "step": 19063
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.635765790939331,
      "learning_rate": 0.0005899214977056323,
      "loss": 2.1759,
      "step": 19064
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6889346241950989,
      "learning_rate": 0.0005899204463055986,
      "loss": 2.5099,
      "step": 19065
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6518712043762207,
      "learning_rate": 0.0005899193948516632,
      "loss": 2.4038,
      "step": 19066
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.5687475204467773,
      "learning_rate": 0.0005899183433438264,
      "loss": 2.3018,
      "step": 19067
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.5612667202949524,
      "learning_rate": 0.0005899172917820882,
      "loss": 2.3479,
      "step": 19068
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6418914198875427,
      "learning_rate": 0.0005899162401664489,
      "loss": 2.6871,
      "step": 19069
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.7045721411705017,
      "learning_rate": 0.0005899151884969087,
      "loss": 2.5918,
      "step": 19070
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.7385132908821106,
      "learning_rate": 0.0005899141367734679,
      "loss": 2.3323,
      "step": 19071
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.5926464796066284,
      "learning_rate": 0.0005899130849961265,
      "loss": 2.2369,
      "step": 19072
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.7009502053260803,
      "learning_rate": 0.0005899120331648847,
      "loss": 2.4704,
      "step": 19073
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6865634322166443,
      "learning_rate": 0.000589910981279743,
      "loss": 2.5695,
      "step": 19074
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6162546873092651,
      "learning_rate": 0.0005899099293407013,
      "loss": 2.5102,
      "step": 19075
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6184990406036377,
      "learning_rate": 0.0005899088773477598,
      "loss": 2.382,
      "step": 19076
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6063005328178406,
      "learning_rate": 0.0005899078253009187,
      "loss": 2.4855,
      "step": 19077
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6372746825218201,
      "learning_rate": 0.0005899067732001784,
      "loss": 2.4479,
      "step": 19078
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.745800256729126,
      "learning_rate": 0.0005899057210455389,
      "loss": 2.3096,
      "step": 19079
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6388473510742188,
      "learning_rate": 0.0005899046688370005,
      "loss": 2.1739,
      "step": 19080
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6567081809043884,
      "learning_rate": 0.0005899036165745632,
      "loss": 2.4471,
      "step": 19081
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6495442390441895,
      "learning_rate": 0.0005899025642582276,
      "loss": 2.5203,
      "step": 19082
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6566307544708252,
      "learning_rate": 0.0005899015118879935,
      "loss": 2.6624,
      "step": 19083
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6406759023666382,
      "learning_rate": 0.0005899004594638612,
      "loss": 2.2155,
      "step": 19084
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.5770545601844788,
      "learning_rate": 0.0005898994069858311,
      "loss": 2.4119,
      "step": 19085
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.5867779850959778,
      "learning_rate": 0.000589898354453903,
      "loss": 2.2578,
      "step": 19086
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.5925582647323608,
      "learning_rate": 0.0005898973018680774,
      "loss": 2.5647,
      "step": 19087
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6497646570205688,
      "learning_rate": 0.0005898962492283544,
      "loss": 2.4778,
      "step": 19088
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6447924971580505,
      "learning_rate": 0.0005898951965347343,
      "loss": 2.4584,
      "step": 19089
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.5932339429855347,
      "learning_rate": 0.0005898941437872171,
      "loss": 2.3966,
      "step": 19090
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.567162275314331,
      "learning_rate": 0.0005898930909858033,
      "loss": 2.4012,
      "step": 19091
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6075626015663147,
      "learning_rate": 0.0005898920381304928,
      "loss": 2.4323,
      "step": 19092
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6179080009460449,
      "learning_rate": 0.0005898909852212858,
      "loss": 2.2582,
      "step": 19093
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6760329008102417,
      "learning_rate": 0.0005898899322581827,
      "loss": 2.1465,
      "step": 19094
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6530023813247681,
      "learning_rate": 0.0005898888792411836,
      "loss": 2.3696,
      "step": 19095
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6178691983222961,
      "learning_rate": 0.0005898878261702886,
      "loss": 2.536,
      "step": 19096
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6450145244598389,
      "learning_rate": 0.000589886773045498,
      "loss": 2.4808,
      "step": 19097
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6321431398391724,
      "learning_rate": 0.0005898857198668121,
      "loss": 2.0656,
      "step": 19098
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.5770352482795715,
      "learning_rate": 0.0005898846666342308,
      "loss": 2.2968,
      "step": 19099
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6341553330421448,
      "learning_rate": 0.0005898836133477546,
      "loss": 2.5666,
      "step": 19100
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.5652109384536743,
      "learning_rate": 0.0005898825600073834,
      "loss": 2.4842,
      "step": 19101
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6127262115478516,
      "learning_rate": 0.0005898815066131178,
      "loss": 2.3586,
      "step": 19102
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.618666410446167,
      "learning_rate": 0.0005898804531649576,
      "loss": 2.3037,
      "step": 19103
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.5597882270812988,
      "learning_rate": 0.0005898793996629031,
      "loss": 2.1427,
      "step": 19104
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6342803835868835,
      "learning_rate": 0.0005898783461069547,
      "loss": 2.4081,
      "step": 19105
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.5751782655715942,
      "learning_rate": 0.0005898772924971123,
      "loss": 2.3537,
      "step": 19106
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.5526115894317627,
      "learning_rate": 0.0005898762388333763,
      "loss": 2.0917,
      "step": 19107
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6566314101219177,
      "learning_rate": 0.0005898751851157469,
      "loss": 2.4639,
      "step": 19108
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6142187714576721,
      "learning_rate": 0.0005898741313442241,
      "loss": 2.5188,
      "step": 19109
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.5924042463302612,
      "learning_rate": 0.0005898730775188083,
      "loss": 2.3225,
      "step": 19110
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6114335656166077,
      "learning_rate": 0.0005898720236394996,
      "loss": 2.4285,
      "step": 19111
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6323907375335693,
      "learning_rate": 0.0005898709697062983,
      "loss": 2.3741,
      "step": 19112
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.5710875988006592,
      "learning_rate": 0.0005898699157192045,
      "loss": 2.4263,
      "step": 19113
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6084110140800476,
      "learning_rate": 0.0005898688616782183,
      "loss": 2.5006,
      "step": 19114
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.5876439213752747,
      "learning_rate": 0.0005898678075833401,
      "loss": 2.1435,
      "step": 19115
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6286327242851257,
      "learning_rate": 0.0005898667534345699,
      "loss": 2.2673,
      "step": 19116
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6025181412696838,
      "learning_rate": 0.0005898656992319081,
      "loss": 2.3487,
      "step": 19117
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6653029322624207,
      "learning_rate": 0.0005898646449753549,
      "loss": 2.3709,
      "step": 19118
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6408463716506958,
      "learning_rate": 0.0005898635906649103,
      "loss": 2.1741,
      "step": 19119
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6343407034873962,
      "learning_rate": 0.0005898625363005745,
      "loss": 2.3386,
      "step": 19120
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.7138534784317017,
      "learning_rate": 0.0005898614818823477,
      "loss": 2.4451,
      "step": 19121
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.621575117111206,
      "learning_rate": 0.0005898604274102304,
      "loss": 2.3877,
      "step": 19122
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.5885766744613647,
      "learning_rate": 0.0005898593728842225,
      "loss": 2.3925,
      "step": 19123
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6205556392669678,
      "learning_rate": 0.0005898583183043243,
      "loss": 2.439,
      "step": 19124
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6303769946098328,
      "learning_rate": 0.0005898572636705359,
      "loss": 2.4833,
      "step": 19125
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6244428753852844,
      "learning_rate": 0.0005898562089828575,
      "loss": 2.4775,
      "step": 19126
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.5217202305793762,
      "learning_rate": 0.0005898551542412895,
      "loss": 2.1527,
      "step": 19127
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6061410307884216,
      "learning_rate": 0.0005898540994458319,
      "loss": 2.4331,
      "step": 19128
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.5573717355728149,
      "learning_rate": 0.000589853044596485,
      "loss": 2.1645,
      "step": 19129
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.5954563021659851,
      "learning_rate": 0.0005898519896932489,
      "loss": 2.2166,
      "step": 19130
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.593446671962738,
      "learning_rate": 0.0005898509347361237,
      "loss": 2.3167,
      "step": 19131
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.5920711755752563,
      "learning_rate": 0.0005898498797251101,
      "loss": 2.3821,
      "step": 19132
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.5660685300827026,
      "learning_rate": 0.0005898488246602077,
      "loss": 2.3958,
      "step": 19133
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.5944464206695557,
      "learning_rate": 0.0005898477695414169,
      "loss": 2.4715,
      "step": 19134
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6181437373161316,
      "learning_rate": 0.000589846714368738,
      "loss": 2.2379,
      "step": 19135
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6929592490196228,
      "learning_rate": 0.000589845659142171,
      "loss": 2.5484,
      "step": 19136
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.618497908115387,
      "learning_rate": 0.0005898446038617163,
      "loss": 2.2002,
      "step": 19137
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.5680475831031799,
      "learning_rate": 0.0005898435485273742,
      "loss": 2.3421,
      "step": 19138
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.5866919159889221,
      "learning_rate": 0.0005898424931391445,
      "loss": 2.2613,
      "step": 19139
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.5949503779411316,
      "learning_rate": 0.0005898414376970277,
      "loss": 2.2755,
      "step": 19140
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.5966396927833557,
      "learning_rate": 0.0005898403822010239,
      "loss": 2.172,
      "step": 19141
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6118288040161133,
      "learning_rate": 0.0005898393266511332,
      "loss": 2.3917,
      "step": 19142
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.5951048731803894,
      "learning_rate": 0.000589838271047356,
      "loss": 2.163,
      "step": 19143
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6674728989601135,
      "learning_rate": 0.0005898372153896924,
      "loss": 2.4815,
      "step": 19144
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6020375490188599,
      "learning_rate": 0.0005898361596781425,
      "loss": 2.4256,
      "step": 19145
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.5814228057861328,
      "learning_rate": 0.0005898351039127067,
      "loss": 2.4988,
      "step": 19146
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.5384716987609863,
      "learning_rate": 0.000589834048093385,
      "loss": 2.2512,
      "step": 19147
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.5861217975616455,
      "learning_rate": 0.0005898329922201778,
      "loss": 2.3629,
      "step": 19148
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6763927936553955,
      "learning_rate": 0.000589831936293085,
      "loss": 2.5491,
      "step": 19149
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.631380021572113,
      "learning_rate": 0.000589830880312107,
      "loss": 2.3471,
      "step": 19150
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6958214640617371,
      "learning_rate": 0.0005898298242772441,
      "loss": 2.1936,
      "step": 19151
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.7219322323799133,
      "learning_rate": 0.0005898287681884962,
      "loss": 2.3172,
      "step": 19152
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6007047295570374,
      "learning_rate": 0.0005898277120458638,
      "loss": 2.3,
      "step": 19153
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.7026960253715515,
      "learning_rate": 0.0005898266558493468,
      "loss": 2.3203,
      "step": 19154
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6534067392349243,
      "learning_rate": 0.0005898255995989458,
      "loss": 2.3679,
      "step": 19155
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6350573301315308,
      "learning_rate": 0.0005898245432946605,
      "loss": 2.4111,
      "step": 19156
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.5720799565315247,
      "learning_rate": 0.0005898234869364915,
      "loss": 2.3016,
      "step": 19157
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6121637225151062,
      "learning_rate": 0.0005898224305244387,
      "loss": 2.2826,
      "step": 19158
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.5925812721252441,
      "learning_rate": 0.0005898213740585026,
      "loss": 2.4979,
      "step": 19159
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.5999295711517334,
      "learning_rate": 0.0005898203175386832,
      "loss": 2.4426,
      "step": 19160
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6296688914299011,
      "learning_rate": 0.0005898192609649806,
      "loss": 2.522,
      "step": 19161
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6669101715087891,
      "learning_rate": 0.0005898182043373954,
      "loss": 2.3222,
      "step": 19162
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6229444742202759,
      "learning_rate": 0.0005898171476559274,
      "loss": 2.19,
      "step": 19163
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6781307458877563,
      "learning_rate": 0.0005898160909205768,
      "loss": 2.2793,
      "step": 19164
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6387618780136108,
      "learning_rate": 0.000589815034131344,
      "loss": 2.3723,
      "step": 19165
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6498586535453796,
      "learning_rate": 0.0005898139772882291,
      "loss": 2.3225,
      "step": 19166
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6012629270553589,
      "learning_rate": 0.0005898129203912324,
      "loss": 2.4356,
      "step": 19167
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6589272022247314,
      "learning_rate": 0.0005898118634403539,
      "loss": 2.4558,
      "step": 19168
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.591787576675415,
      "learning_rate": 0.0005898108064355939,
      "loss": 2.3727,
      "step": 19169
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6841111183166504,
      "learning_rate": 0.0005898097493769527,
      "loss": 2.3808,
      "step": 19170
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.7099546194076538,
      "learning_rate": 0.0005898086922644303,
      "loss": 2.3211,
      "step": 19171
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6410824060440063,
      "learning_rate": 0.000589807635098027,
      "loss": 2.4892,
      "step": 19172
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.5768436193466187,
      "learning_rate": 0.0005898065778777431,
      "loss": 2.5884,
      "step": 19173
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6785401701927185,
      "learning_rate": 0.0005898055206035786,
      "loss": 2.4882,
      "step": 19174
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6071150898933411,
      "learning_rate": 0.0005898044632755337,
      "loss": 2.315,
      "step": 19175
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6051432490348816,
      "learning_rate": 0.0005898034058936088,
      "loss": 2.4241,
      "step": 19176
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6394636034965515,
      "learning_rate": 0.000589802348457804,
      "loss": 2.2331,
      "step": 19177
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.5618134140968323,
      "learning_rate": 0.0005898012909681193,
      "loss": 2.2539,
      "step": 19178
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.5932955145835876,
      "learning_rate": 0.0005898002334245553,
      "loss": 2.1085,
      "step": 19179
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6207334995269775,
      "learning_rate": 0.0005897991758271118,
      "loss": 2.4025,
      "step": 19180
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.608198344707489,
      "learning_rate": 0.0005897981181757892,
      "loss": 2.3093,
      "step": 19181
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6027187705039978,
      "learning_rate": 0.0005897970604705877,
      "loss": 2.327,
      "step": 19182
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6353341937065125,
      "learning_rate": 0.0005897960027115074,
      "loss": 2.4688,
      "step": 19183
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6626016497612,
      "learning_rate": 0.0005897949448985485,
      "loss": 2.4178,
      "step": 19184
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.729548990726471,
      "learning_rate": 0.0005897938870317115,
      "loss": 2.2435,
      "step": 19185
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6658602952957153,
      "learning_rate": 0.0005897928291109961,
      "loss": 2.3304,
      "step": 19186
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6513554453849792,
      "learning_rate": 0.0005897917711364028,
      "loss": 2.3372,
      "step": 19187
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6743602752685547,
      "learning_rate": 0.0005897907131079318,
      "loss": 2.4234,
      "step": 19188
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6742560267448425,
      "learning_rate": 0.0005897896550255832,
      "loss": 2.2257,
      "step": 19189
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.5929920077323914,
      "learning_rate": 0.0005897885968893572,
      "loss": 2.2744,
      "step": 19190
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6512583494186401,
      "learning_rate": 0.0005897875386992541,
      "loss": 2.2902,
      "step": 19191
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.694681704044342,
      "learning_rate": 0.000589786480455274,
      "loss": 2.4753,
      "step": 19192
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.606350839138031,
      "learning_rate": 0.0005897854221574171,
      "loss": 2.3102,
      "step": 19193
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6412538886070251,
      "learning_rate": 0.0005897843638056836,
      "loss": 2.3529,
      "step": 19194
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6832489371299744,
      "learning_rate": 0.0005897833054000738,
      "loss": 2.4343,
      "step": 19195
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6096020936965942,
      "learning_rate": 0.0005897822469405877,
      "loss": 2.4174,
      "step": 19196
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6283811926841736,
      "learning_rate": 0.0005897811884272257,
      "loss": 2.4189,
      "step": 19197
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.603579044342041,
      "learning_rate": 0.0005897801298599878,
      "loss": 2.2897,
      "step": 19198
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6276399493217468,
      "learning_rate": 0.0005897790712388745,
      "loss": 2.3465,
      "step": 19199
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6136121153831482,
      "learning_rate": 0.0005897780125638857,
      "loss": 2.3452,
      "step": 19200
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6196726560592651,
      "learning_rate": 0.0005897769538350216,
      "loss": 2.4646,
      "step": 19201
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6727147698402405,
      "learning_rate": 0.0005897758950522826,
      "loss": 2.181,
      "step": 19202
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.656438410282135,
      "learning_rate": 0.0005897748362156688,
      "loss": 2.5358,
      "step": 19203
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6072973608970642,
      "learning_rate": 0.0005897737773251803,
      "loss": 2.455,
      "step": 19204
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.7006335854530334,
      "learning_rate": 0.0005897727183808174,
      "loss": 2.3887,
      "step": 19205
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6990605592727661,
      "learning_rate": 0.0005897716593825803,
      "loss": 2.3655,
      "step": 19206
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6601833701133728,
      "learning_rate": 0.0005897706003304692,
      "loss": 2.2432,
      "step": 19207
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.7853944897651672,
      "learning_rate": 0.0005897695412244844,
      "loss": 2.4249,
      "step": 19208
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.5934057235717773,
      "learning_rate": 0.0005897684820646258,
      "loss": 2.3017,
      "step": 19209
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6599906086921692,
      "learning_rate": 0.0005897674228508938,
      "loss": 2.2672,
      "step": 19210
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.7331460118293762,
      "learning_rate": 0.0005897663635832886,
      "loss": 2.5411,
      "step": 19211
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6051892042160034,
      "learning_rate": 0.0005897653042618103,
      "loss": 2.3304,
      "step": 19212
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6053197383880615,
      "learning_rate": 0.0005897642448864593,
      "loss": 2.4764,
      "step": 19213
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.5936607122421265,
      "learning_rate": 0.0005897631854572355,
      "loss": 2.2639,
      "step": 19214
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6261047720909119,
      "learning_rate": 0.0005897621259741393,
      "loss": 2.2737,
      "step": 19215
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.5964658260345459,
      "learning_rate": 0.0005897610664371708,
      "loss": 2.3515,
      "step": 19216
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6843112111091614,
      "learning_rate": 0.0005897600068463304,
      "loss": 2.5135,
      "step": 19217
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6648052930831909,
      "learning_rate": 0.000589758947201618,
      "loss": 2.6112,
      "step": 19218
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6603195071220398,
      "learning_rate": 0.0005897578875030341,
      "loss": 2.427,
      "step": 19219
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6183756589889526,
      "learning_rate": 0.0005897568277505786,
      "loss": 2.5012,
      "step": 19220
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.5706055164337158,
      "learning_rate": 0.0005897557679442519,
      "loss": 2.3275,
      "step": 19221
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.5855624079704285,
      "learning_rate": 0.0005897547080840541,
      "loss": 2.2301,
      "step": 19222
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6804029941558838,
      "learning_rate": 0.0005897536481699855,
      "loss": 2.4655,
      "step": 19223
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6375485062599182,
      "learning_rate": 0.0005897525882020461,
      "loss": 2.2908,
      "step": 19224
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6723778247833252,
      "learning_rate": 0.0005897515281802363,
      "loss": 2.3083,
      "step": 19225
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.60182785987854,
      "learning_rate": 0.0005897504681045563,
      "loss": 2.5282,
      "step": 19226
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6012338995933533,
      "learning_rate": 0.0005897494079750061,
      "loss": 2.2765,
      "step": 19227
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.9767650961875916,
      "learning_rate": 0.000589748347791586,
      "loss": 2.2274,
      "step": 19228
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.589593768119812,
      "learning_rate": 0.0005897472875542963,
      "loss": 2.2577,
      "step": 19229
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.7976326942443848,
      "learning_rate": 0.0005897462272631371,
      "loss": 2.3193,
      "step": 19230
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6631305813789368,
      "learning_rate": 0.0005897451669181085,
      "loss": 2.3272,
      "step": 19231
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6101282835006714,
      "learning_rate": 0.0005897441065192109,
      "loss": 2.3416,
      "step": 19232
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6856511235237122,
      "learning_rate": 0.0005897430460664444,
      "loss": 2.2429,
      "step": 19233
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6458245515823364,
      "learning_rate": 0.0005897419855598092,
      "loss": 2.4766,
      "step": 19234
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6500857472419739,
      "learning_rate": 0.0005897409249993054,
      "loss": 2.4614,
      "step": 19235
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6436301469802856,
      "learning_rate": 0.0005897398643849334,
      "loss": 2.562,
      "step": 19236
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6006619334220886,
      "learning_rate": 0.0005897388037166933,
      "loss": 2.446,
      "step": 19237
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6092613339424133,
      "learning_rate": 0.0005897377429945851,
      "loss": 2.4891,
      "step": 19238
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.5966171026229858,
      "learning_rate": 0.0005897366822186095,
      "loss": 2.5081,
      "step": 19239
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6570160388946533,
      "learning_rate": 0.0005897356213887662,
      "loss": 2.6325,
      "step": 19240
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6158875823020935,
      "learning_rate": 0.0005897345605050555,
      "loss": 2.352,
      "step": 19241
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.5797860026359558,
      "learning_rate": 0.0005897334995674778,
      "loss": 2.1352,
      "step": 19242
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.5855969786643982,
      "learning_rate": 0.0005897324385760332,
      "loss": 2.4073,
      "step": 19243
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6352090835571289,
      "learning_rate": 0.0005897313775307218,
      "loss": 2.1812,
      "step": 19244
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.5965516567230225,
      "learning_rate": 0.0005897303164315438,
      "loss": 2.6098,
      "step": 19245
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.7012587189674377,
      "learning_rate": 0.0005897292552784996,
      "loss": 2.5713,
      "step": 19246
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.5793684720993042,
      "learning_rate": 0.0005897281940715893,
      "loss": 2.3046,
      "step": 19247
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6144462823867798,
      "learning_rate": 0.0005897271328108129,
      "loss": 2.5929,
      "step": 19248
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6612248420715332,
      "learning_rate": 0.0005897260714961708,
      "loss": 2.2982,
      "step": 19249
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6520583629608154,
      "learning_rate": 0.0005897250101276632,
      "loss": 2.3978,
      "step": 19250
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6233805418014526,
      "learning_rate": 0.0005897239487052902,
      "loss": 2.3175,
      "step": 19251
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6817672252655029,
      "learning_rate": 0.0005897228872290522,
      "loss": 2.4609,
      "step": 19252
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6289673447608948,
      "learning_rate": 0.0005897218256989491,
      "loss": 2.3911,
      "step": 19253
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6770325899124146,
      "learning_rate": 0.0005897207641149812,
      "loss": 2.2214,
      "step": 19254
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.5779017210006714,
      "learning_rate": 0.0005897197024771488,
      "loss": 2.5811,
      "step": 19255
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.7393454909324646,
      "learning_rate": 0.0005897186407854521,
      "loss": 2.2515,
      "step": 19256
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.619210422039032,
      "learning_rate": 0.0005897175790398911,
      "loss": 2.4989,
      "step": 19257
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.5896542072296143,
      "learning_rate": 0.0005897165172404663,
      "loss": 2.3695,
      "step": 19258
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6242687702178955,
      "learning_rate": 0.0005897154553871776,
      "loss": 2.2375,
      "step": 19259
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.5992498993873596,
      "learning_rate": 0.0005897143934800253,
      "loss": 2.205,
      "step": 19260
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6488088965415955,
      "learning_rate": 0.0005897133315190098,
      "loss": 2.3115,
      "step": 19261
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.7350878119468689,
      "learning_rate": 0.000589712269504131,
      "loss": 2.4092,
      "step": 19262
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6306765079498291,
      "learning_rate": 0.0005897112074353892,
      "loss": 2.288,
      "step": 19263
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.5988415479660034,
      "learning_rate": 0.0005897101453127847,
      "loss": 2.2681,
      "step": 19264
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6698165535926819,
      "learning_rate": 0.0005897090831363176,
      "loss": 2.4362,
      "step": 19265
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.5934426188468933,
      "learning_rate": 0.0005897080209059881,
      "loss": 2.3512,
      "step": 19266
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6614199876785278,
      "learning_rate": 0.0005897069586217964,
      "loss": 2.286,
      "step": 19267
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6159797310829163,
      "learning_rate": 0.0005897058962837427,
      "loss": 2.4964,
      "step": 19268
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.7167978882789612,
      "learning_rate": 0.0005897048338918273,
      "loss": 2.5366,
      "step": 19269
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.7002747058868408,
      "learning_rate": 0.0005897037714460502,
      "loss": 2.3238,
      "step": 19270
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6498704552650452,
      "learning_rate": 0.0005897027089464116,
      "loss": 2.3002,
      "step": 19271
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6101701855659485,
      "learning_rate": 0.0005897016463929119,
      "loss": 2.2629,
      "step": 19272
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.5995712280273438,
      "learning_rate": 0.0005897005837855512,
      "loss": 2.3429,
      "step": 19273
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.7042867541313171,
      "learning_rate": 0.0005896995211243297,
      "loss": 2.264,
      "step": 19274
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6083147525787354,
      "learning_rate": 0.0005896984584092477,
      "loss": 2.4756,
      "step": 19275
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6032108068466187,
      "learning_rate": 0.0005896973956403052,
      "loss": 2.1345,
      "step": 19276
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6089229583740234,
      "learning_rate": 0.0005896963328175024,
      "loss": 2.3847,
      "step": 19277
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6587749719619751,
      "learning_rate": 0.0005896952699408396,
      "loss": 2.2695,
      "step": 19278
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.5988097786903381,
      "learning_rate": 0.0005896942070103172,
      "loss": 2.2634,
      "step": 19279
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6515429615974426,
      "learning_rate": 0.0005896931440259349,
      "loss": 2.5562,
      "step": 19280
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6452140808105469,
      "learning_rate": 0.0005896920809876933,
      "loss": 2.3016,
      "step": 19281
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6068831086158752,
      "learning_rate": 0.0005896910178955924,
      "loss": 2.47,
      "step": 19282
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.639672040939331,
      "learning_rate": 0.0005896899547496327,
      "loss": 2.163,
      "step": 19283
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.5791974067687988,
      "learning_rate": 0.0005896888915498139,
      "loss": 2.2503,
      "step": 19284
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.7790243029594421,
      "learning_rate": 0.0005896878282961365,
      "loss": 2.3211,
      "step": 19285
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6920846700668335,
      "learning_rate": 0.0005896867649886008,
      "loss": 2.502,
      "step": 19286
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6494740843772888,
      "learning_rate": 0.0005896857016272068,
      "loss": 2.385,
      "step": 19287
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6813788414001465,
      "learning_rate": 0.0005896846382119547,
      "loss": 2.5044,
      "step": 19288
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6414472460746765,
      "learning_rate": 0.0005896835747428448,
      "loss": 2.467,
      "step": 19289
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6523342728614807,
      "learning_rate": 0.0005896825112198773,
      "loss": 2.4145,
      "step": 19290
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.536854088306427,
      "learning_rate": 0.0005896814476430524,
      "loss": 2.5289,
      "step": 19291
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.5986621379852295,
      "learning_rate": 0.0005896803840123702,
      "loss": 2.5014,
      "step": 19292
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.7432829737663269,
      "learning_rate": 0.0005896793203278308,
      "loss": 2.2446,
      "step": 19293
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.612760603427887,
      "learning_rate": 0.0005896782565894348,
      "loss": 2.3651,
      "step": 19294
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6223379969596863,
      "learning_rate": 0.0005896771927971819,
      "loss": 2.5281,
      "step": 19295
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6896042823791504,
      "learning_rate": 0.0005896761289510726,
      "loss": 2.3045,
      "step": 19296
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6806758642196655,
      "learning_rate": 0.0005896750650511072,
      "loss": 2.4454,
      "step": 19297
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6423717141151428,
      "learning_rate": 0.0005896740010972856,
      "loss": 2.6612,
      "step": 19298
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6040069460868835,
      "learning_rate": 0.0005896729370896081,
      "loss": 2.3626,
      "step": 19299
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6063817739486694,
      "learning_rate": 0.000589671873028075,
      "loss": 2.3692,
      "step": 19300
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6034315228462219,
      "learning_rate": 0.0005896708089126864,
      "loss": 2.2598,
      "step": 19301
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6000741124153137,
      "learning_rate": 0.0005896697447434426,
      "loss": 2.203,
      "step": 19302
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6231207847595215,
      "learning_rate": 0.0005896686805203436,
      "loss": 2.1848,
      "step": 19303
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6662471890449524,
      "learning_rate": 0.00058966761624339,
      "loss": 2.4687,
      "step": 19304
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6151602268218994,
      "learning_rate": 0.0005896665519125815,
      "loss": 2.2084,
      "step": 19305
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6719136834144592,
      "learning_rate": 0.0005896654875279185,
      "loss": 2.2254,
      "step": 19306
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.5608089566230774,
      "learning_rate": 0.0005896644230894013,
      "loss": 2.1045,
      "step": 19307
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6248505711555481,
      "learning_rate": 0.0005896633585970299,
      "loss": 2.4686,
      "step": 19308
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.5741749405860901,
      "learning_rate": 0.0005896622940508048,
      "loss": 2.4226,
      "step": 19309
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.5784003734588623,
      "learning_rate": 0.0005896612294507259,
      "loss": 2.6,
      "step": 19310
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6178515553474426,
      "learning_rate": 0.0005896601647967934,
      "loss": 2.4688,
      "step": 19311
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6194531321525574,
      "learning_rate": 0.0005896591000890078,
      "loss": 2.4729,
      "step": 19312
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6659808158874512,
      "learning_rate": 0.000589658035327369,
      "loss": 2.2769,
      "step": 19313
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.7363026142120361,
      "learning_rate": 0.0005896569705118774,
      "loss": 2.103,
      "step": 19314
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6171594858169556,
      "learning_rate": 0.000589655905642533,
      "loss": 2.4013,
      "step": 19315
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6012776494026184,
      "learning_rate": 0.0005896548407193362,
      "loss": 2.1806,
      "step": 19316
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6281763911247253,
      "learning_rate": 0.0005896537757422869,
      "loss": 2.358,
      "step": 19317
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6688773036003113,
      "learning_rate": 0.0005896527107113858,
      "loss": 2.3005,
      "step": 19318
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.589812695980072,
      "learning_rate": 0.0005896516456266325,
      "loss": 2.287,
      "step": 19319
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.5945449471473694,
      "learning_rate": 0.0005896505804880276,
      "loss": 2.2574,
      "step": 19320
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.7393333911895752,
      "learning_rate": 0.0005896495152955712,
      "loss": 2.3659,
      "step": 19321
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6522085070610046,
      "learning_rate": 0.0005896484500492635,
      "loss": 2.2705,
      "step": 19322
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6447814702987671,
      "learning_rate": 0.0005896473847491046,
      "loss": 2.4654,
      "step": 19323
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6105117201805115,
      "learning_rate": 0.000589646319395095,
      "loss": 2.2067,
      "step": 19324
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.5836685299873352,
      "learning_rate": 0.0005896452539872345,
      "loss": 2.3056,
      "step": 19325
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.5800789594650269,
      "learning_rate": 0.0005896441885255235,
      "loss": 2.5405,
      "step": 19326
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.5886763334274292,
      "learning_rate": 0.0005896431230099622,
      "loss": 2.2187,
      "step": 19327
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6343831419944763,
      "learning_rate": 0.0005896420574405507,
      "loss": 2.3306,
      "step": 19328
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.5642277598381042,
      "learning_rate": 0.0005896409918172894,
      "loss": 2.3443,
      "step": 19329
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6634723544120789,
      "learning_rate": 0.0005896399261401783,
      "loss": 2.4879,
      "step": 19330
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6173926591873169,
      "learning_rate": 0.0005896388604092178,
      "loss": 2.452,
      "step": 19331
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.551008403301239,
      "learning_rate": 0.0005896377946244078,
      "loss": 2.3422,
      "step": 19332
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6186172366142273,
      "learning_rate": 0.0005896367287857487,
      "loss": 2.352,
      "step": 19333
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6693629622459412,
      "learning_rate": 0.0005896356628932407,
      "loss": 2.366,
      "step": 19334
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6033982634544373,
      "learning_rate": 0.000589634596946884,
      "loss": 2.386,
      "step": 19335
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6243082880973816,
      "learning_rate": 0.0005896335309466788,
      "loss": 2.4011,
      "step": 19336
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.573554277420044,
      "learning_rate": 0.000589632464892625,
      "loss": 2.3268,
      "step": 19337
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.5982137322425842,
      "learning_rate": 0.0005896313987847233,
      "loss": 2.3697,
      "step": 19338
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.570701003074646,
      "learning_rate": 0.0005896303326229736,
      "loss": 2.4035,
      "step": 19339
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.5884910225868225,
      "learning_rate": 0.0005896292664073761,
      "loss": 2.3693,
      "step": 19340
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6625190377235413,
      "learning_rate": 0.0005896282001379311,
      "loss": 2.3152,
      "step": 19341
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.5947977900505066,
      "learning_rate": 0.0005896271338146387,
      "loss": 2.1408,
      "step": 19342
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.630098819732666,
      "learning_rate": 0.0005896260674374993,
      "loss": 2.4476,
      "step": 19343
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.5890368223190308,
      "learning_rate": 0.0005896250010065128,
      "loss": 2.4202,
      "step": 19344
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6319440007209778,
      "learning_rate": 0.0005896239345216795,
      "loss": 2.4619,
      "step": 19345
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6382072567939758,
      "learning_rate": 0.0005896228679829998,
      "loss": 2.2133,
      "step": 19346
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.629975438117981,
      "learning_rate": 0.0005896218013904738,
      "loss": 2.5065,
      "step": 19347
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6270806193351746,
      "learning_rate": 0.0005896207347441014,
      "loss": 2.3933,
      "step": 19348
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.5947123169898987,
      "learning_rate": 0.0005896196680438831,
      "loss": 2.3079,
      "step": 19349
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6271857023239136,
      "learning_rate": 0.0005896186012898191,
      "loss": 2.3723,
      "step": 19350
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6444771885871887,
      "learning_rate": 0.0005896175344819095,
      "loss": 2.4841,
      "step": 19351
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.612513542175293,
      "learning_rate": 0.0005896164676201546,
      "loss": 2.4089,
      "step": 19352
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.650687575340271,
      "learning_rate": 0.0005896154007045544,
      "loss": 2.3744,
      "step": 19353
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6774559617042542,
      "learning_rate": 0.0005896143337351094,
      "loss": 2.2193,
      "step": 19354
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.5962126851081848,
      "learning_rate": 0.0005896132667118195,
      "loss": 2.3777,
      "step": 19355
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6016948223114014,
      "learning_rate": 0.0005896121996346852,
      "loss": 2.2435,
      "step": 19356
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6637272834777832,
      "learning_rate": 0.0005896111325037063,
      "loss": 2.5148,
      "step": 19357
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6104598045349121,
      "learning_rate": 0.0005896100653188833,
      "loss": 2.1605,
      "step": 19358
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6087485551834106,
      "learning_rate": 0.0005896089980802164,
      "loss": 2.1998,
      "step": 19359
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6324177384376526,
      "learning_rate": 0.0005896079307877056,
      "loss": 2.4264,
      "step": 19360
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6810277700424194,
      "learning_rate": 0.0005896068634413513,
      "loss": 2.3777,
      "step": 19361
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6286217570304871,
      "learning_rate": 0.0005896057960411537,
      "loss": 2.4028,
      "step": 19362
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.7003162503242493,
      "learning_rate": 0.0005896047285871128,
      "loss": 2.6476,
      "step": 19363
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.9374281167984009,
      "learning_rate": 0.0005896036610792289,
      "loss": 2.2795,
      "step": 19364
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.5974892973899841,
      "learning_rate": 0.0005896025935175022,
      "loss": 2.5996,
      "step": 19365
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6832311749458313,
      "learning_rate": 0.0005896015259019329,
      "loss": 2.1864,
      "step": 19366
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.764741063117981,
      "learning_rate": 0.0005896004582325214,
      "loss": 2.4975,
      "step": 19367
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6190151572227478,
      "learning_rate": 0.0005895993905092675,
      "loss": 2.3648,
      "step": 19368
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6735753417015076,
      "learning_rate": 0.0005895983227321717,
      "loss": 2.3948,
      "step": 19369
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.651604413986206,
      "learning_rate": 0.0005895972549012341,
      "loss": 2.4441,
      "step": 19370
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6194783449172974,
      "learning_rate": 0.0005895961870164549,
      "loss": 2.4787,
      "step": 19371
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6156930327415466,
      "learning_rate": 0.0005895951190778342,
      "loss": 2.3309,
      "step": 19372
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.5832324028015137,
      "learning_rate": 0.0005895940510853724,
      "loss": 2.5511,
      "step": 19373
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.5768131017684937,
      "learning_rate": 0.0005895929830390696,
      "loss": 2.2572,
      "step": 19374
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6280835270881653,
      "learning_rate": 0.000589591914938926,
      "loss": 2.2161,
      "step": 19375
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6086428761482239,
      "learning_rate": 0.0005895908467849418,
      "loss": 2.3887,
      "step": 19376
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.5979940891265869,
      "learning_rate": 0.0005895897785771172,
      "loss": 2.356,
      "step": 19377
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6491015553474426,
      "learning_rate": 0.0005895887103154524,
      "loss": 2.5136,
      "step": 19378
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6187483668327332,
      "learning_rate": 0.0005895876419999476,
      "loss": 2.4564,
      "step": 19379
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6379246115684509,
      "learning_rate": 0.0005895865736306029,
      "loss": 2.3254,
      "step": 19380
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6986438632011414,
      "learning_rate": 0.0005895855052074187,
      "loss": 2.2293,
      "step": 19381
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6241894960403442,
      "learning_rate": 0.000589584436730395,
      "loss": 2.3578,
      "step": 19382
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6938402056694031,
      "learning_rate": 0.0005895833681995321,
      "loss": 2.3967,
      "step": 19383
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.7254769206047058,
      "learning_rate": 0.0005895822996148302,
      "loss": 2.351,
      "step": 19384
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6216787099838257,
      "learning_rate": 0.0005895812309762895,
      "loss": 2.2747,
      "step": 19385
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.652758002281189,
      "learning_rate": 0.0005895801622839102,
      "loss": 2.3464,
      "step": 19386
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6809754371643066,
      "learning_rate": 0.0005895790935376924,
      "loss": 2.5968,
      "step": 19387
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6641231775283813,
      "learning_rate": 0.0005895780247376365,
      "loss": 2.3376,
      "step": 19388
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.5959640145301819,
      "learning_rate": 0.0005895769558837425,
      "loss": 2.185,
      "step": 19389
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.5642205476760864,
      "learning_rate": 0.0005895758869760106,
      "loss": 2.1303,
      "step": 19390
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6507117748260498,
      "learning_rate": 0.0005895748180144413,
      "loss": 2.2007,
      "step": 19391
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.5962026119232178,
      "learning_rate": 0.0005895737489990344,
      "loss": 2.4717,
      "step": 19392
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6012230515480042,
      "learning_rate": 0.0005895726799297903,
      "loss": 2.3596,
      "step": 19393
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6053544282913208,
      "learning_rate": 0.0005895716108067091,
      "loss": 2.1873,
      "step": 19394
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6931871175765991,
      "learning_rate": 0.0005895705416297912,
      "loss": 2.4677,
      "step": 19395
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6176376938819885,
      "learning_rate": 0.0005895694723990366,
      "loss": 2.3351,
      "step": 19396
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.5849825739860535,
      "learning_rate": 0.0005895684031144456,
      "loss": 2.0994,
      "step": 19397
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6045962572097778,
      "learning_rate": 0.0005895673337760185,
      "loss": 2.2912,
      "step": 19398
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.687145471572876,
      "learning_rate": 0.0005895662643837553,
      "loss": 2.2874,
      "step": 19399
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6044153571128845,
      "learning_rate": 0.0005895651949376561,
      "loss": 2.3221,
      "step": 19400
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6420724987983704,
      "learning_rate": 0.0005895641254377213,
      "loss": 2.5035,
      "step": 19401
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.682379424571991,
      "learning_rate": 0.0005895630558839512,
      "loss": 2.3213,
      "step": 19402
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6119786500930786,
      "learning_rate": 0.0005895619862763458,
      "loss": 2.5154,
      "step": 19403
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6241840720176697,
      "learning_rate": 0.0005895609166149053,
      "loss": 2.6476,
      "step": 19404
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.674146831035614,
      "learning_rate": 0.0005895598468996301,
      "loss": 2.3638,
      "step": 19405
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.5911528468132019,
      "learning_rate": 0.0005895587771305201,
      "loss": 2.4751,
      "step": 19406
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6757546663284302,
      "learning_rate": 0.0005895577073075757,
      "loss": 2.6771,
      "step": 19407
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6124923229217529,
      "learning_rate": 0.000589556637430797,
      "loss": 2.315,
      "step": 19408
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.622643768787384,
      "learning_rate": 0.0005895555675001844,
      "loss": 2.5958,
      "step": 19409
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.5746405720710754,
      "learning_rate": 0.000589554497515738,
      "loss": 2.5801,
      "step": 19410
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6428678631782532,
      "learning_rate": 0.0005895534274774578,
      "loss": 2.5117,
      "step": 19411
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6437849402427673,
      "learning_rate": 0.0005895523573853442,
      "loss": 2.4474,
      "step": 19412
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6036965250968933,
      "learning_rate": 0.0005895512872393973,
      "loss": 2.3021,
      "step": 19413
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.5683984160423279,
      "learning_rate": 0.0005895502170396175,
      "loss": 2.3434,
      "step": 19414
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.7316920757293701,
      "learning_rate": 0.0005895491467860047,
      "loss": 2.4318,
      "step": 19415
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6301095485687256,
      "learning_rate": 0.0005895480764785593,
      "loss": 2.5991,
      "step": 19416
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.5945765376091003,
      "learning_rate": 0.0005895470061172814,
      "loss": 2.331,
      "step": 19417
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6558617353439331,
      "learning_rate": 0.0005895459357021714,
      "loss": 2.6274,
      "step": 19418
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.5999367833137512,
      "learning_rate": 0.0005895448652332291,
      "loss": 2.1191,
      "step": 19419
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6188918352127075,
      "learning_rate": 0.0005895437947104551,
      "loss": 2.3452,
      "step": 19420
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6219976544380188,
      "learning_rate": 0.0005895427241338494,
      "loss": 2.2587,
      "step": 19421
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6117211580276489,
      "learning_rate": 0.0005895416535034124,
      "loss": 2.4062,
      "step": 19422
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6513567566871643,
      "learning_rate": 0.0005895405828191439,
      "loss": 2.3787,
      "step": 19423
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6365996599197388,
      "learning_rate": 0.0005895395120810445,
      "loss": 2.3197,
      "step": 19424
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6753911375999451,
      "learning_rate": 0.0005895384412891142,
      "loss": 2.5574,
      "step": 19425
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6244925856590271,
      "learning_rate": 0.0005895373704433532,
      "loss": 2.6797,
      "step": 19426
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6534832119941711,
      "learning_rate": 0.0005895362995437617,
      "loss": 2.2301,
      "step": 19427
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6510921120643616,
      "learning_rate": 0.0005895352285903401,
      "loss": 2.2037,
      "step": 19428
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.5654094815254211,
      "learning_rate": 0.0005895341575830884,
      "loss": 2.3762,
      "step": 19429
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.8183324933052063,
      "learning_rate": 0.0005895330865220066,
      "loss": 2.3519,
      "step": 19430
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.5914217829704285,
      "learning_rate": 0.0005895320154070953,
      "loss": 2.512,
      "step": 19431
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.5987626314163208,
      "learning_rate": 0.0005895309442383545,
      "loss": 2.0786,
      "step": 19432
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6320852637290955,
      "learning_rate": 0.0005895298730157845,
      "loss": 2.2806,
      "step": 19433
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6407947540283203,
      "learning_rate": 0.0005895288017393855,
      "loss": 2.489,
      "step": 19434
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6898014545440674,
      "learning_rate": 0.0005895277304091575,
      "loss": 2.2902,
      "step": 19435
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.65439772605896,
      "learning_rate": 0.0005895266590251008,
      "loss": 2.4127,
      "step": 19436
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6133215427398682,
      "learning_rate": 0.0005895255875872158,
      "loss": 2.5783,
      "step": 19437
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.5532944798469543,
      "learning_rate": 0.0005895245160955023,
      "loss": 2.3706,
      "step": 19438
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6849278807640076,
      "learning_rate": 0.0005895234445499609,
      "loss": 2.342,
      "step": 19439
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6536678075790405,
      "learning_rate": 0.0005895223729505915,
      "loss": 2.4174,
      "step": 19440
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6380829215049744,
      "learning_rate": 0.0005895213012973946,
      "loss": 2.2181,
      "step": 19441
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6158738136291504,
      "learning_rate": 0.00058952022959037,
      "loss": 2.4985,
      "step": 19442
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6220908761024475,
      "learning_rate": 0.0005895191578295183,
      "loss": 2.541,
      "step": 19443
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6024753451347351,
      "learning_rate": 0.0005895180860148394,
      "loss": 2.379,
      "step": 19444
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.5709633231163025,
      "learning_rate": 0.0005895170141463337,
      "loss": 2.4521,
      "step": 19445
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6866888403892517,
      "learning_rate": 0.0005895159422240014,
      "loss": 2.3228,
      "step": 19446
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6958616375923157,
      "learning_rate": 0.0005895148702478425,
      "loss": 2.3323,
      "step": 19447
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6081501245498657,
      "learning_rate": 0.0005895137982178573,
      "loss": 2.5447,
      "step": 19448
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6380619406700134,
      "learning_rate": 0.0005895127261340461,
      "loss": 2.3176,
      "step": 19449
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.5766690373420715,
      "learning_rate": 0.000589511653996409,
      "loss": 2.4516,
      "step": 19450
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.650328516960144,
      "learning_rate": 0.0005895105818049461,
      "loss": 2.2782,
      "step": 19451
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.5852457284927368,
      "learning_rate": 0.0005895095095596579,
      "loss": 2.1885,
      "step": 19452
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.790009617805481,
      "learning_rate": 0.0005895084372605444,
      "loss": 2.4026,
      "step": 19453
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6922093033790588,
      "learning_rate": 0.0005895073649076057,
      "loss": 2.4872,
      "step": 19454
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.7268657684326172,
      "learning_rate": 0.0005895062925008422,
      "loss": 2.2288,
      "step": 19455
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.5755982398986816,
      "learning_rate": 0.0005895052200402541,
      "loss": 2.2492,
      "step": 19456
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6504544615745544,
      "learning_rate": 0.0005895041475258414,
      "loss": 2.3287,
      "step": 19457
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.9547520875930786,
      "learning_rate": 0.0005895030749576043,
      "loss": 2.2622,
      "step": 19458
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.7083131670951843,
      "learning_rate": 0.0005895020023355433,
      "loss": 2.4765,
      "step": 19459
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.663052499294281,
      "learning_rate": 0.0005895009296596584,
      "loss": 2.3713,
      "step": 19460
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.7321622371673584,
      "learning_rate": 0.0005894998569299497,
      "loss": 2.3055,
      "step": 19461
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6698023676872253,
      "learning_rate": 0.0005894987841464175,
      "loss": 2.5196,
      "step": 19462
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.7536687254905701,
      "learning_rate": 0.0005894977113090621,
      "loss": 2.2815,
      "step": 19463
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.61474609375,
      "learning_rate": 0.0005894966384178837,
      "loss": 2.456,
      "step": 19464
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.7643588185310364,
      "learning_rate": 0.0005894955654728823,
      "loss": 2.1872,
      "step": 19465
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6273922920227051,
      "learning_rate": 0.0005894944924740581,
      "loss": 2.247,
      "step": 19466
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6013280153274536,
      "learning_rate": 0.0005894934194214116,
      "loss": 2.2182,
      "step": 19467
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6532031893730164,
      "learning_rate": 0.0005894923463149427,
      "loss": 2.3929,
      "step": 19468
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6366875171661377,
      "learning_rate": 0.0005894912731546517,
      "loss": 2.3954,
      "step": 19469
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6991337537765503,
      "learning_rate": 0.0005894901999405389,
      "loss": 2.2282,
      "step": 19470
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.7106743454933167,
      "learning_rate": 0.0005894891266726044,
      "loss": 2.5081,
      "step": 19471
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6381155252456665,
      "learning_rate": 0.0005894880533508484,
      "loss": 2.3359,
      "step": 19472
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.5633847117424011,
      "learning_rate": 0.000589486979975271,
      "loss": 2.6026,
      "step": 19473
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6442494988441467,
      "learning_rate": 0.0005894859065458726,
      "loss": 2.4,
      "step": 19474
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6986718773841858,
      "learning_rate": 0.0005894848330626533,
      "loss": 2.2662,
      "step": 19475
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6833512187004089,
      "learning_rate": 0.0005894837595256132,
      "loss": 2.6077,
      "step": 19476
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6149198412895203,
      "learning_rate": 0.0005894826859347527,
      "loss": 2.7624,
      "step": 19477
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6180112361907959,
      "learning_rate": 0.0005894816122900719,
      "loss": 2.1561,
      "step": 19478
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6191917657852173,
      "learning_rate": 0.000589480538591571,
      "loss": 2.1764,
      "step": 19479
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6253513097763062,
      "learning_rate": 0.0005894794648392502,
      "loss": 2.3835,
      "step": 19480
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6011136174201965,
      "learning_rate": 0.0005894783910331097,
      "loss": 2.2907,
      "step": 19481
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6541195511817932,
      "learning_rate": 0.0005894773171731498,
      "loss": 2.4185,
      "step": 19482
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6038358807563782,
      "learning_rate": 0.0005894762432593705,
      "loss": 2.3184,
      "step": 19483
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.5787863731384277,
      "learning_rate": 0.0005894751692917721,
      "loss": 2.4483,
      "step": 19484
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6213653683662415,
      "learning_rate": 0.0005894740952703549,
      "loss": 2.2661,
      "step": 19485
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6281708478927612,
      "learning_rate": 0.0005894730211951188,
      "loss": 2.246,
      "step": 19486
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6156278848648071,
      "learning_rate": 0.0005894719470660644,
      "loss": 2.3834,
      "step": 19487
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6135799884796143,
      "learning_rate": 0.0005894708728831915,
      "loss": 2.4326,
      "step": 19488
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6237561702728271,
      "learning_rate": 0.0005894697986465006,
      "loss": 2.4193,
      "step": 19489
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.5672804713249207,
      "learning_rate": 0.0005894687243559919,
      "loss": 2.205,
      "step": 19490
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.5897289514541626,
      "learning_rate": 0.0005894676500116654,
      "loss": 2.2307,
      "step": 19491
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6224902868270874,
      "learning_rate": 0.0005894665756135214,
      "loss": 2.2498,
      "step": 19492
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6400893330574036,
      "learning_rate": 0.0005894655011615601,
      "loss": 2.4179,
      "step": 19493
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6101834774017334,
      "learning_rate": 0.0005894644266557817,
      "loss": 2.149,
      "step": 19494
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6606155633926392,
      "learning_rate": 0.0005894633520961864,
      "loss": 2.5624,
      "step": 19495
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.5787675976753235,
      "learning_rate": 0.0005894622774827745,
      "loss": 2.2153,
      "step": 19496
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6233423948287964,
      "learning_rate": 0.000589461202815546,
      "loss": 2.3791,
      "step": 19497
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.645235538482666,
      "learning_rate": 0.0005894601280945011,
      "loss": 2.2749,
      "step": 19498
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6603893637657166,
      "learning_rate": 0.0005894590533196403,
      "loss": 2.1543,
      "step": 19499
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.5614765882492065,
      "learning_rate": 0.0005894579784909634,
      "loss": 2.4389,
      "step": 19500
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6027105450630188,
      "learning_rate": 0.000589456903608471,
      "loss": 2.1266,
      "step": 19501
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6292558312416077,
      "learning_rate": 0.000589455828672163,
      "loss": 2.2932,
      "step": 19502
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6291292309761047,
      "learning_rate": 0.0005894547536820397,
      "loss": 2.3593,
      "step": 19503
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6012874245643616,
      "learning_rate": 0.0005894536786381013,
      "loss": 2.5454,
      "step": 19504
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6269654631614685,
      "learning_rate": 0.000589452603540348,
      "loss": 2.412,
      "step": 19505
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6750720739364624,
      "learning_rate": 0.0005894515283887801,
      "loss": 2.3459,
      "step": 19506
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6066955327987671,
      "learning_rate": 0.0005894504531833975,
      "loss": 2.6278,
      "step": 19507
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6024454236030579,
      "learning_rate": 0.0005894493779242008,
      "loss": 2.2314,
      "step": 19508
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.7882658243179321,
      "learning_rate": 0.0005894483026111898,
      "loss": 2.4435,
      "step": 19509
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6041732430458069,
      "learning_rate": 0.0005894472272443649,
      "loss": 2.2847,
      "step": 19510
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6497464776039124,
      "learning_rate": 0.0005894461518237264,
      "loss": 2.4095,
      "step": 19511
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6179651618003845,
      "learning_rate": 0.0005894450763492744,
      "loss": 2.4793,
      "step": 19512
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.5676968097686768,
      "learning_rate": 0.0005894440008210091,
      "loss": 2.3131,
      "step": 19513
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.650900661945343,
      "learning_rate": 0.0005894429252389306,
      "loss": 2.4143,
      "step": 19514
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6168516874313354,
      "learning_rate": 0.0005894418496030393,
      "loss": 2.7707,
      "step": 19515
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6260925531387329,
      "learning_rate": 0.0005894407739133352,
      "loss": 2.44,
      "step": 19516
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.7095556855201721,
      "learning_rate": 0.0005894396981698187,
      "loss": 2.5038,
      "step": 19517
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6431817412376404,
      "learning_rate": 0.0005894386223724899,
      "loss": 2.1551,
      "step": 19518
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.5694644451141357,
      "learning_rate": 0.000589437546521349,
      "loss": 2.4656,
      "step": 19519
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.5883898735046387,
      "learning_rate": 0.000589436470616396,
      "loss": 2.1993,
      "step": 19520
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.5890559554100037,
      "learning_rate": 0.0005894353946576315,
      "loss": 2.5622,
      "step": 19521
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6004788279533386,
      "learning_rate": 0.0005894343186450554,
      "loss": 2.389,
      "step": 19522
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.5636028051376343,
      "learning_rate": 0.0005894332425786681,
      "loss": 2.3722,
      "step": 19523
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6004706025123596,
      "learning_rate": 0.0005894321664584697,
      "loss": 2.4081,
      "step": 19524
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.5870421528816223,
      "learning_rate": 0.0005894310902844603,
      "loss": 2.5236,
      "step": 19525
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.5834537744522095,
      "learning_rate": 0.0005894300140566402,
      "loss": 2.4337,
      "step": 19526
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6891178488731384,
      "learning_rate": 0.0005894289377750096,
      "loss": 2.3073,
      "step": 19527
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6094239354133606,
      "learning_rate": 0.0005894278614395687,
      "loss": 2.4492,
      "step": 19528
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6012842655181885,
      "learning_rate": 0.0005894267850503178,
      "loss": 2.4554,
      "step": 19529
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6155920028686523,
      "learning_rate": 0.0005894257086072569,
      "loss": 2.1431,
      "step": 19530
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6387118697166443,
      "learning_rate": 0.0005894246321103862,
      "loss": 2.2422,
      "step": 19531
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.5972066521644592,
      "learning_rate": 0.0005894235555597061,
      "loss": 2.3474,
      "step": 19532
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.5900198817253113,
      "learning_rate": 0.0005894224789552168,
      "loss": 2.2374,
      "step": 19533
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6960070729255676,
      "learning_rate": 0.0005894214022969183,
      "loss": 2.2892,
      "step": 19534
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6553557515144348,
      "learning_rate": 0.0005894203255848109,
      "loss": 2.4837,
      "step": 19535
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.5822532773017883,
      "learning_rate": 0.0005894192488188948,
      "loss": 2.4068,
      "step": 19536
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.655568540096283,
      "learning_rate": 0.0005894181719991702,
      "loss": 2.3587,
      "step": 19537
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6404099464416504,
      "learning_rate": 0.0005894170951256374,
      "loss": 2.3805,
      "step": 19538
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6029796004295349,
      "learning_rate": 0.0005894160181982963,
      "loss": 2.3227,
      "step": 19539
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.5749870538711548,
      "learning_rate": 0.0005894149412171474,
      "loss": 2.3526,
      "step": 19540
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6041407585144043,
      "learning_rate": 0.0005894138641821908,
      "loss": 2.3541,
      "step": 19541
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.609830379486084,
      "learning_rate": 0.0005894127870934267,
      "loss": 2.3273,
      "step": 19542
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.5790125727653503,
      "learning_rate": 0.0005894117099508553,
      "loss": 2.5511,
      "step": 19543
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6523988246917725,
      "learning_rate": 0.0005894106327544768,
      "loss": 2.3165,
      "step": 19544
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.5943390130996704,
      "learning_rate": 0.0005894095555042914,
      "loss": 2.5238,
      "step": 19545
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6161895990371704,
      "learning_rate": 0.0005894084782002994,
      "loss": 2.1715,
      "step": 19546
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6001499891281128,
      "learning_rate": 0.0005894074008425007,
      "loss": 2.2837,
      "step": 19547
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6040279269218445,
      "learning_rate": 0.0005894063234308958,
      "loss": 2.3529,
      "step": 19548
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6582818627357483,
      "learning_rate": 0.0005894052459654848,
      "loss": 2.1328,
      "step": 19549
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.664598822593689,
      "learning_rate": 0.0005894041684462679,
      "loss": 2.4298,
      "step": 19550
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.590999186038971,
      "learning_rate": 0.0005894030908732453,
      "loss": 2.5133,
      "step": 19551
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6418374180793762,
      "learning_rate": 0.0005894020132464172,
      "loss": 2.2665,
      "step": 19552
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6839677095413208,
      "learning_rate": 0.0005894009355657838,
      "loss": 2.314,
      "step": 19553
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.7056020498275757,
      "learning_rate": 0.0005893998578313453,
      "loss": 2.5056,
      "step": 19554
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.7040131688117981,
      "learning_rate": 0.0005893987800431019,
      "loss": 2.3489,
      "step": 19555
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6116200685501099,
      "learning_rate": 0.0005893977022010539,
      "loss": 2.2794,
      "step": 19556
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.5732945203781128,
      "learning_rate": 0.0005893966243052013,
      "loss": 2.0271,
      "step": 19557
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6169410943984985,
      "learning_rate": 0.0005893955463555444,
      "loss": 2.5151,
      "step": 19558
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.609822690486908,
      "learning_rate": 0.0005893944683520835,
      "loss": 2.205,
      "step": 19559
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.8240966200828552,
      "learning_rate": 0.0005893933902948186,
      "loss": 2.3468,
      "step": 19560
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6750565767288208,
      "learning_rate": 0.0005893923121837501,
      "loss": 2.4194,
      "step": 19561
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6924691796302795,
      "learning_rate": 0.000589391234018878,
      "loss": 2.2394,
      "step": 19562
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6415669322013855,
      "learning_rate": 0.0005893901558002027,
      "loss": 2.1825,
      "step": 19563
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6265659332275391,
      "learning_rate": 0.0005893890775277242,
      "loss": 2.1958,
      "step": 19564
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.5842944979667664,
      "learning_rate": 0.0005893879992014429,
      "loss": 2.4332,
      "step": 19565
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.5898535251617432,
      "learning_rate": 0.0005893869208213588,
      "loss": 2.4354,
      "step": 19566
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.5703423023223877,
      "learning_rate": 0.0005893858423874724,
      "loss": 2.1537,
      "step": 19567
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6842023134231567,
      "learning_rate": 0.0005893847638997836,
      "loss": 2.2238,
      "step": 19568
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.5929647088050842,
      "learning_rate": 0.0005893836853582928,
      "loss": 2.2872,
      "step": 19569
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.5786975622177124,
      "learning_rate": 0.0005893826067629999,
      "loss": 2.4053,
      "step": 19570
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6155381798744202,
      "learning_rate": 0.0005893815281139055,
      "loss": 2.2548,
      "step": 19571
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6471995115280151,
      "learning_rate": 0.0005893804494110095,
      "loss": 2.4698,
      "step": 19572
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.5837440490722656,
      "learning_rate": 0.0005893793706543123,
      "loss": 2.3142,
      "step": 19573
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.5970867872238159,
      "learning_rate": 0.000589378291843814,
      "loss": 2.3758,
      "step": 19574
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.5542500019073486,
      "learning_rate": 0.0005893772129795148,
      "loss": 2.3992,
      "step": 19575
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6275284886360168,
      "learning_rate": 0.0005893761340614148,
      "loss": 2.5579,
      "step": 19576
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.5985214114189148,
      "learning_rate": 0.0005893750550895144,
      "loss": 2.1175,
      "step": 19577
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6189035773277283,
      "learning_rate": 0.0005893739760638137,
      "loss": 2.5229,
      "step": 19578
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.5825063586235046,
      "learning_rate": 0.000589372896984313,
      "loss": 2.4039,
      "step": 19579
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.5987986922264099,
      "learning_rate": 0.0005893718178510123,
      "loss": 2.464,
      "step": 19580
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.645435094833374,
      "learning_rate": 0.000589370738663912,
      "loss": 2.4152,
      "step": 19581
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6014233231544495,
      "learning_rate": 0.0005893696594230121,
      "loss": 2.2636,
      "step": 19582
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6216374039649963,
      "learning_rate": 0.0005893685801283131,
      "loss": 2.2263,
      "step": 19583
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6544443964958191,
      "learning_rate": 0.0005893675007798149,
      "loss": 2.43,
      "step": 19584
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.7011714577674866,
      "learning_rate": 0.0005893664213775179,
      "loss": 2.3473,
      "step": 19585
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.5677512288093567,
      "learning_rate": 0.0005893653419214222,
      "loss": 2.3026,
      "step": 19586
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.5970284938812256,
      "learning_rate": 0.0005893642624115279,
      "loss": 2.4312,
      "step": 19587
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.5926663279533386,
      "learning_rate": 0.0005893631828478354,
      "loss": 2.1548,
      "step": 19588
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6266356706619263,
      "learning_rate": 0.0005893621032303449,
      "loss": 2.2689,
      "step": 19589
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6510646343231201,
      "learning_rate": 0.0005893610235590563,
      "loss": 2.3836,
      "step": 19590
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6280748844146729,
      "learning_rate": 0.0005893599438339703,
      "loss": 2.2865,
      "step": 19591
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6514776349067688,
      "learning_rate": 0.0005893588640550867,
      "loss": 2.6416,
      "step": 19592
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6849085688591003,
      "learning_rate": 0.0005893577842224059,
      "loss": 2.4548,
      "step": 19593
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.5822395086288452,
      "learning_rate": 0.0005893567043359279,
      "loss": 2.3468,
      "step": 19594
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6655734181404114,
      "learning_rate": 0.0005893556243956531,
      "loss": 2.5251,
      "step": 19595
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.7174286246299744,
      "learning_rate": 0.0005893545444015816,
      "loss": 2.3823,
      "step": 19596
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6432532668113708,
      "learning_rate": 0.0005893534643537136,
      "loss": 2.3084,
      "step": 19597
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.7290204167366028,
      "learning_rate": 0.0005893523842520494,
      "loss": 2.3192,
      "step": 19598
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6253748536109924,
      "learning_rate": 0.0005893513040965891,
      "loss": 2.6206,
      "step": 19599
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.7611189484596252,
      "learning_rate": 0.0005893502238873328,
      "loss": 2.3111,
      "step": 19600
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.689795196056366,
      "learning_rate": 0.000589349143624281,
      "loss": 2.3463,
      "step": 19601
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6837233304977417,
      "learning_rate": 0.0005893480633074337,
      "loss": 2.1817,
      "step": 19602
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.7164292931556702,
      "learning_rate": 0.0005893469829367911,
      "loss": 2.3431,
      "step": 19603
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6599650382995605,
      "learning_rate": 0.0005893459025123535,
      "loss": 2.6095,
      "step": 19604
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.7203201055526733,
      "learning_rate": 0.000589344822034121,
      "loss": 2.4208,
      "step": 19605
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.7191029191017151,
      "learning_rate": 0.0005893437415020938,
      "loss": 2.2744,
      "step": 19606
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6513303518295288,
      "learning_rate": 0.0005893426609162722,
      "loss": 2.3048,
      "step": 19607
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6337825059890747,
      "learning_rate": 0.0005893415802766562,
      "loss": 2.4011,
      "step": 19608
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6111157536506653,
      "learning_rate": 0.0005893404995832463,
      "loss": 2.4492,
      "step": 19609
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6981368064880371,
      "learning_rate": 0.0005893394188360425,
      "loss": 2.5541,
      "step": 19610
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6035091280937195,
      "learning_rate": 0.000589338338035045,
      "loss": 2.4152,
      "step": 19611
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.5843917727470398,
      "learning_rate": 0.0005893372571802541,
      "loss": 2.1522,
      "step": 19612
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6065986156463623,
      "learning_rate": 0.0005893361762716699,
      "loss": 2.2254,
      "step": 19613
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6277502179145813,
      "learning_rate": 0.0005893350953092927,
      "loss": 2.5087,
      "step": 19614
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.5932431221008301,
      "learning_rate": 0.0005893340142931225,
      "loss": 2.5931,
      "step": 19615
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.7182852029800415,
      "learning_rate": 0.0005893329332231597,
      "loss": 2.431,
      "step": 19616
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6267365217208862,
      "learning_rate": 0.0005893318520994045,
      "loss": 2.249,
      "step": 19617
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.7831720113754272,
      "learning_rate": 0.0005893307709218571,
      "loss": 2.3066,
      "step": 19618
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6777582168579102,
      "learning_rate": 0.0005893296896905175,
      "loss": 2.1289,
      "step": 19619
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6386719346046448,
      "learning_rate": 0.0005893286084053861,
      "loss": 2.2792,
      "step": 19620
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6442005038261414,
      "learning_rate": 0.0005893275270664631,
      "loss": 2.3219,
      "step": 19621
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.7517246603965759,
      "learning_rate": 0.0005893264456737487,
      "loss": 2.2666,
      "step": 19622
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.604836642742157,
      "learning_rate": 0.0005893253642272429,
      "loss": 2.5973,
      "step": 19623
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6045557856559753,
      "learning_rate": 0.0005893242827269462,
      "loss": 2.4289,
      "step": 19624
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.734756350517273,
      "learning_rate": 0.0005893232011728585,
      "loss": 2.5037,
      "step": 19625
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.672805666923523,
      "learning_rate": 0.0005893221195649803,
      "loss": 2.5386,
      "step": 19626
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6223301887512207,
      "learning_rate": 0.0005893210379033116,
      "loss": 2.4909,
      "step": 19627
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6442101001739502,
      "learning_rate": 0.0005893199561878527,
      "loss": 2.4285,
      "step": 19628
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6622931361198425,
      "learning_rate": 0.0005893188744186037,
      "loss": 2.4614,
      "step": 19629
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6235668063163757,
      "learning_rate": 0.0005893177925955648,
      "loss": 2.4199,
      "step": 19630
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.5603939890861511,
      "learning_rate": 0.0005893167107187364,
      "loss": 2.2755,
      "step": 19631
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6043646931648254,
      "learning_rate": 0.0005893156287881184,
      "loss": 2.5259,
      "step": 19632
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.5436203479766846,
      "learning_rate": 0.0005893145468037113,
      "loss": 2.1334,
      "step": 19633
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6775426864624023,
      "learning_rate": 0.0005893134647655152,
      "loss": 2.4154,
      "step": 19634
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.612682044506073,
      "learning_rate": 0.0005893123826735302,
      "loss": 2.4896,
      "step": 19635
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6445179581642151,
      "learning_rate": 0.0005893113005277566,
      "loss": 2.3531,
      "step": 19636
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6080504655838013,
      "learning_rate": 0.0005893102183281944,
      "loss": 2.4162,
      "step": 19637
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.5805773735046387,
      "learning_rate": 0.0005893091360748441,
      "loss": 2.3378,
      "step": 19638
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6413072943687439,
      "learning_rate": 0.0005893080537677058,
      "loss": 2.6045,
      "step": 19639
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.7368849515914917,
      "learning_rate": 0.0005893069714067796,
      "loss": 2.4274,
      "step": 19640
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.640072762966156,
      "learning_rate": 0.0005893058889920659,
      "loss": 2.3232,
      "step": 19641
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.625276505947113,
      "learning_rate": 0.0005893048065235646,
      "loss": 2.3602,
      "step": 19642
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6131141185760498,
      "learning_rate": 0.0005893037240012763,
      "loss": 2.4091,
      "step": 19643
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.7070086002349854,
      "learning_rate": 0.0005893026414252008,
      "loss": 2.2934,
      "step": 19644
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6750478744506836,
      "learning_rate": 0.0005893015587953384,
      "loss": 2.1774,
      "step": 19645
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.7498328685760498,
      "learning_rate": 0.0005893004761116895,
      "loss": 2.5608,
      "step": 19646
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.5993680953979492,
      "learning_rate": 0.0005892993933742542,
      "loss": 2.1885,
      "step": 19647
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6578724980354309,
      "learning_rate": 0.0005892983105830326,
      "loss": 2.3891,
      "step": 19648
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.7494819164276123,
      "learning_rate": 0.0005892972277380252,
      "loss": 2.4712,
      "step": 19649
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.5784692764282227,
      "learning_rate": 0.0005892961448392318,
      "loss": 2.305,
      "step": 19650
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.639693558216095,
      "learning_rate": 0.0005892950618866528,
      "loss": 2.493,
      "step": 19651
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.5930973291397095,
      "learning_rate": 0.0005892939788802883,
      "loss": 2.4243,
      "step": 19652
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.7000563144683838,
      "learning_rate": 0.0005892928958201387,
      "loss": 2.4764,
      "step": 19653
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.611822247505188,
      "learning_rate": 0.000589291812706204,
      "loss": 2.323,
      "step": 19654
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6981938481330872,
      "learning_rate": 0.0005892907295384847,
      "loss": 2.5253,
      "step": 19655
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.768267035484314,
      "learning_rate": 0.0005892896463169806,
      "loss": 2.4231,
      "step": 19656
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.5974270105361938,
      "learning_rate": 0.0005892885630416921,
      "loss": 2.3175,
      "step": 19657
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.709869384765625,
      "learning_rate": 0.0005892874797126195,
      "loss": 2.3451,
      "step": 19658
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6947294473648071,
      "learning_rate": 0.0005892863963297628,
      "loss": 2.2949,
      "step": 19659
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6074613928794861,
      "learning_rate": 0.0005892853128931225,
      "loss": 2.1924,
      "step": 19660
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6462013721466064,
      "learning_rate": 0.0005892842294026984,
      "loss": 2.2051,
      "step": 19661
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6760273575782776,
      "learning_rate": 0.0005892831458584909,
      "loss": 2.2519,
      "step": 19662
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.7241520881652832,
      "learning_rate": 0.0005892820622605002,
      "loss": 2.3221,
      "step": 19663
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.5928599238395691,
      "learning_rate": 0.0005892809786087267,
      "loss": 2.457,
      "step": 19664
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6449540257453918,
      "learning_rate": 0.0005892798949031702,
      "loss": 2.4884,
      "step": 19665
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6358058452606201,
      "learning_rate": 0.0005892788111438311,
      "loss": 2.2413,
      "step": 19666
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.72129225730896,
      "learning_rate": 0.0005892777273307097,
      "loss": 2.3039,
      "step": 19667
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6290231943130493,
      "learning_rate": 0.0005892766434638061,
      "loss": 2.4277,
      "step": 19668
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.7028083205223083,
      "learning_rate": 0.0005892755595431205,
      "loss": 2.3975,
      "step": 19669
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6679048538208008,
      "learning_rate": 0.0005892744755686531,
      "loss": 2.3922,
      "step": 19670
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.5864623785018921,
      "learning_rate": 0.0005892733915404041,
      "loss": 2.5151,
      "step": 19671
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6873441934585571,
      "learning_rate": 0.0005892723074583737,
      "loss": 2.3218,
      "step": 19672
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.5965785980224609,
      "learning_rate": 0.0005892712233225622,
      "loss": 2.2726,
      "step": 19673
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.5844816565513611,
      "learning_rate": 0.0005892701391329697,
      "loss": 2.1215,
      "step": 19674
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6114857792854309,
      "learning_rate": 0.0005892690548895964,
      "loss": 2.2589,
      "step": 19675
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6462028622627258,
      "learning_rate": 0.0005892679705924426,
      "loss": 2.2744,
      "step": 19676
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6555254459381104,
      "learning_rate": 0.0005892668862415082,
      "loss": 2.4779,
      "step": 19677
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6396928429603577,
      "learning_rate": 0.0005892658018367937,
      "loss": 2.3612,
      "step": 19678
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6575271487236023,
      "learning_rate": 0.0005892647173782994,
      "loss": 2.2314,
      "step": 19679
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.637943685054779,
      "learning_rate": 0.0005892636328660252,
      "loss": 2.4459,
      "step": 19680
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6653540134429932,
      "learning_rate": 0.0005892625482999715,
      "loss": 2.4902,
      "step": 19681
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6354467868804932,
      "learning_rate": 0.0005892614636801383,
      "loss": 2.219,
      "step": 19682
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.5941219925880432,
      "learning_rate": 0.0005892603790065261,
      "loss": 2.3142,
      "step": 19683
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6623323559761047,
      "learning_rate": 0.0005892592942791349,
      "loss": 2.3551,
      "step": 19684
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6755517721176147,
      "learning_rate": 0.0005892582094979647,
      "loss": 2.3899,
      "step": 19685
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.682248055934906,
      "learning_rate": 0.0005892571246630162,
      "loss": 2.5432,
      "step": 19686
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6171594858169556,
      "learning_rate": 0.0005892560397742894,
      "loss": 2.3236,
      "step": 19687
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6494122743606567,
      "learning_rate": 0.0005892549548317842,
      "loss": 2.4146,
      "step": 19688
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.7924973368644714,
      "learning_rate": 0.0005892538698355011,
      "loss": 2.4895,
      "step": 19689
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.5792537927627563,
      "learning_rate": 0.0005892527847854404,
      "loss": 2.4139,
      "step": 19690
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6018516421318054,
      "learning_rate": 0.000589251699681602,
      "loss": 2.2107,
      "step": 19691
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6112335920333862,
      "learning_rate": 0.0005892506145239863,
      "loss": 2.3354,
      "step": 19692
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6413803696632385,
      "learning_rate": 0.0005892495293125935,
      "loss": 2.4004,
      "step": 19693
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6575980186462402,
      "learning_rate": 0.0005892484440474237,
      "loss": 2.407,
      "step": 19694
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6424975395202637,
      "learning_rate": 0.0005892473587284771,
      "loss": 2.374,
      "step": 19695
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6488622426986694,
      "learning_rate": 0.000589246273355754,
      "loss": 2.3511,
      "step": 19696
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6491824984550476,
      "learning_rate": 0.0005892451879292546,
      "loss": 2.2807,
      "step": 19697
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6446017622947693,
      "learning_rate": 0.000589244102448979,
      "loss": 2.368,
      "step": 19698
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.651176929473877,
      "learning_rate": 0.0005892430169149275,
      "loss": 2.4736,
      "step": 19699
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.5915326476097107,
      "learning_rate": 0.0005892419313271003,
      "loss": 2.1585,
      "step": 19700
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6181569695472717,
      "learning_rate": 0.0005892408456854975,
      "loss": 2.4584,
      "step": 19701
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.5658906102180481,
      "learning_rate": 0.0005892397599901193,
      "loss": 2.2442,
      "step": 19702
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6480717062950134,
      "learning_rate": 0.0005892386742409662,
      "loss": 2.2358,
      "step": 19703
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.5691711902618408,
      "learning_rate": 0.0005892375884380379,
      "loss": 2.2,
      "step": 19704
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6431910395622253,
      "learning_rate": 0.0005892365025813351,
      "loss": 2.5163,
      "step": 19705
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.7067506313323975,
      "learning_rate": 0.0005892354166708576,
      "loss": 2.3011,
      "step": 19706
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.5873537063598633,
      "learning_rate": 0.0005892343307066059,
      "loss": 2.4127,
      "step": 19707
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6640119552612305,
      "learning_rate": 0.0005892332446885799,
      "loss": 2.4055,
      "step": 19708
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6624225974082947,
      "learning_rate": 0.0005892321586167802,
      "loss": 2.6987,
      "step": 19709
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6457952857017517,
      "learning_rate": 0.0005892310724912067,
      "loss": 2.5059,
      "step": 19710
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.5743449330329895,
      "learning_rate": 0.0005892299863118597,
      "loss": 2.2232,
      "step": 19711
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6225104331970215,
      "learning_rate": 0.0005892289000787393,
      "loss": 2.614,
      "step": 19712
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6656765937805176,
      "learning_rate": 0.0005892278137918459,
      "loss": 1.9708,
      "step": 19713
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6267098188400269,
      "learning_rate": 0.0005892267274511796,
      "loss": 2.1481,
      "step": 19714
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6439216732978821,
      "learning_rate": 0.0005892256410567404,
      "loss": 2.3025,
      "step": 19715
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6641427874565125,
      "learning_rate": 0.0005892245546085289,
      "loss": 2.4988,
      "step": 19716
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.654067873954773,
      "learning_rate": 0.000589223468106545,
      "loss": 2.4507,
      "step": 19717
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6811230778694153,
      "learning_rate": 0.0005892223815507889,
      "loss": 2.4023,
      "step": 19718
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.5925551652908325,
      "learning_rate": 0.0005892212949412611,
      "loss": 2.4152,
      "step": 19719
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6212186813354492,
      "learning_rate": 0.0005892202082779615,
      "loss": 2.4662,
      "step": 19720
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.5927039980888367,
      "learning_rate": 0.0005892191215608905,
      "loss": 2.2446,
      "step": 19721
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6103044748306274,
      "learning_rate": 0.0005892180347900481,
      "loss": 2.3587,
      "step": 19722
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6665217280387878,
      "learning_rate": 0.0005892169479654346,
      "loss": 2.4617,
      "step": 19723
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.5921168923377991,
      "learning_rate": 0.0005892158610870502,
      "loss": 2.2606,
      "step": 19724
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.5909814238548279,
      "learning_rate": 0.0005892147741548952,
      "loss": 2.3869,
      "step": 19725
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6061194539070129,
      "learning_rate": 0.0005892136871689696,
      "loss": 2.2087,
      "step": 19726
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6733676791191101,
      "learning_rate": 0.0005892126001292738,
      "loss": 2.3699,
      "step": 19727
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6029180288314819,
      "learning_rate": 0.000589211513035808,
      "loss": 2.3842,
      "step": 19728
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.5713291168212891,
      "learning_rate": 0.0005892104258885722,
      "loss": 2.1739,
      "step": 19729
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6084640622138977,
      "learning_rate": 0.0005892093386875668,
      "loss": 2.6296,
      "step": 19730
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6250729560852051,
      "learning_rate": 0.0005892082514327918,
      "loss": 2.346,
      "step": 19731
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.5970594882965088,
      "learning_rate": 0.0005892071641242476,
      "loss": 2.3738,
      "step": 19732
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.5829141139984131,
      "learning_rate": 0.0005892060767619344,
      "loss": 2.3921,
      "step": 19733
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6308659911155701,
      "learning_rate": 0.0005892049893458522,
      "loss": 2.732,
      "step": 19734
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6223192811012268,
      "learning_rate": 0.0005892039018760013,
      "loss": 2.4718,
      "step": 19735
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.7235245108604431,
      "learning_rate": 0.0005892028143523821,
      "loss": 2.4818,
      "step": 19736
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.5981234312057495,
      "learning_rate": 0.0005892017267749945,
      "loss": 2.1802,
      "step": 19737
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6189389824867249,
      "learning_rate": 0.000589200639143839,
      "loss": 2.2262,
      "step": 19738
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6748290061950684,
      "learning_rate": 0.0005891995514589155,
      "loss": 2.5581,
      "step": 19739
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.5984917879104614,
      "learning_rate": 0.0005891984637202244,
      "loss": 2.2896,
      "step": 19740
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.5753982663154602,
      "learning_rate": 0.0005891973759277659,
      "loss": 2.2152,
      "step": 19741
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.5879873633384705,
      "learning_rate": 0.00058919628808154,
      "loss": 2.2715,
      "step": 19742
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6939426064491272,
      "learning_rate": 0.0005891952001815472,
      "loss": 2.3659,
      "step": 19743
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.5707606077194214,
      "learning_rate": 0.0005891941122277875,
      "loss": 2.3037,
      "step": 19744
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6802578568458557,
      "learning_rate": 0.0005891930242202611,
      "loss": 2.5095,
      "step": 19745
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6840245723724365,
      "learning_rate": 0.0005891919361589684,
      "loss": 2.3693,
      "step": 19746
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.5761076211929321,
      "learning_rate": 0.0005891908480439093,
      "loss": 2.3528,
      "step": 19747
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.5838109254837036,
      "learning_rate": 0.0005891897598750842,
      "loss": 2.3081,
      "step": 19748
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6267464756965637,
      "learning_rate": 0.0005891886716524933,
      "loss": 2.3847,
      "step": 19749
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6025059819221497,
      "learning_rate": 0.0005891875833761368,
      "loss": 2.4398,
      "step": 19750
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.5801295042037964,
      "learning_rate": 0.0005891864950460148,
      "loss": 2.3032,
      "step": 19751
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.5667622685432434,
      "learning_rate": 0.0005891854066621276,
      "loss": 2.4025,
      "step": 19752
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.5933359861373901,
      "learning_rate": 0.0005891843182244754,
      "loss": 2.3912,
      "step": 19753
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.5913364887237549,
      "learning_rate": 0.0005891832297330584,
      "loss": 2.1678,
      "step": 19754
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.5989915728569031,
      "learning_rate": 0.0005891821411878767,
      "loss": 2.2094,
      "step": 19755
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6584053635597229,
      "learning_rate": 0.0005891810525889306,
      "loss": 2.4996,
      "step": 19756
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6416795253753662,
      "learning_rate": 0.0005891799639362202,
      "loss": 2.4375,
      "step": 19757
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.668877899646759,
      "learning_rate": 0.000589178875229746,
      "loss": 2.3604,
      "step": 19758
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6206642389297485,
      "learning_rate": 0.0005891777864695079,
      "loss": 2.1951,
      "step": 19759
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.5579840540885925,
      "learning_rate": 0.0005891766976555061,
      "loss": 2.4692,
      "step": 19760
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6348091959953308,
      "learning_rate": 0.000589175608787741,
      "loss": 2.3547,
      "step": 19761
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6612170934677124,
      "learning_rate": 0.0005891745198662126,
      "loss": 2.44,
      "step": 19762
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.5879413485527039,
      "learning_rate": 0.0005891734308909213,
      "loss": 2.3541,
      "step": 19763
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.5955310463905334,
      "learning_rate": 0.000589172341861867,
      "loss": 2.5645,
      "step": 19764
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6079116463661194,
      "learning_rate": 0.0005891712527790504,
      "loss": 2.436,
      "step": 19765
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.633600652217865,
      "learning_rate": 0.0005891701636424712,
      "loss": 2.3339,
      "step": 19766
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.576314389705658,
      "learning_rate": 0.0005891690744521299,
      "loss": 2.2607,
      "step": 19767
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.7909441590309143,
      "learning_rate": 0.0005891679852080265,
      "loss": 2.5023,
      "step": 19768
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.7607720494270325,
      "learning_rate": 0.0005891668959101614,
      "loss": 2.2947,
      "step": 19769
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.5922412872314453,
      "learning_rate": 0.0005891658065585347,
      "loss": 2.5096,
      "step": 19770
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6842932105064392,
      "learning_rate": 0.0005891647171531466,
      "loss": 2.2767,
      "step": 19771
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.7426810264587402,
      "learning_rate": 0.0005891636276939974,
      "loss": 2.274,
      "step": 19772
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6658949851989746,
      "learning_rate": 0.0005891625381810871,
      "loss": 2.3425,
      "step": 19773
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.7001386880874634,
      "learning_rate": 0.0005891614486144161,
      "loss": 2.5301,
      "step": 19774
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.070845127105713,
      "learning_rate": 0.0005891603589939846,
      "loss": 2.2352,
      "step": 19775
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6637163758277893,
      "learning_rate": 0.0005891592693197926,
      "loss": 2.2607,
      "step": 19776
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.860466718673706,
      "learning_rate": 0.0005891581795918405,
      "loss": 2.3032,
      "step": 19777
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.7627271413803101,
      "learning_rate": 0.0005891570898101285,
      "loss": 2.0665,
      "step": 19778
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6369737982749939,
      "learning_rate": 0.0005891559999746566,
      "loss": 2.367,
      "step": 19779
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6885549426078796,
      "learning_rate": 0.0005891549100854251,
      "loss": 2.373,
      "step": 19780
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.746816873550415,
      "learning_rate": 0.0005891538201424344,
      "loss": 2.3463,
      "step": 19781
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6299141049385071,
      "learning_rate": 0.0005891527301456845,
      "loss": 2.3015,
      "step": 19782
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6318045258522034,
      "learning_rate": 0.0005891516400951756,
      "loss": 2.042,
      "step": 19783
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6585956811904907,
      "learning_rate": 0.000589150549990908,
      "loss": 2.3672,
      "step": 19784
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.7166453003883362,
      "learning_rate": 0.0005891494598328819,
      "loss": 2.2716,
      "step": 19785
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6432164907455444,
      "learning_rate": 0.0005891483696210974,
      "loss": 2.6497,
      "step": 19786
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6266275644302368,
      "learning_rate": 0.0005891472793555548,
      "loss": 2.2743,
      "step": 19787
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6679242849349976,
      "learning_rate": 0.0005891461890362542,
      "loss": 2.3169,
      "step": 19788
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6061934232711792,
      "learning_rate": 0.0005891450986631959,
      "loss": 2.3423,
      "step": 19789
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6871477365493774,
      "learning_rate": 0.0005891440082363801,
      "loss": 2.4216,
      "step": 19790
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6466752290725708,
      "learning_rate": 0.0005891429177558069,
      "loss": 2.372,
      "step": 19791
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6780895590782166,
      "learning_rate": 0.0005891418272214766,
      "loss": 2.48,
      "step": 19792
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6191987991333008,
      "learning_rate": 0.0005891407366333894,
      "loss": 2.7049,
      "step": 19793
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6054609417915344,
      "learning_rate": 0.0005891396459915454,
      "loss": 2.3535,
      "step": 19794
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6910228729248047,
      "learning_rate": 0.000589138555295945,
      "loss": 2.3346,
      "step": 19795
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.5956436991691589,
      "learning_rate": 0.0005891374645465881,
      "loss": 2.4323,
      "step": 19796
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.7507289052009583,
      "learning_rate": 0.0005891363737434753,
      "loss": 2.3913,
      "step": 19797
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6473859548568726,
      "learning_rate": 0.0005891352828866066,
      "loss": 2.3738,
      "step": 19798
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6023919582366943,
      "learning_rate": 0.000589134191975982,
      "loss": 2.4547,
      "step": 19799
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6274340152740479,
      "learning_rate": 0.000589133101011602,
      "loss": 2.326,
      "step": 19800
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6170200705528259,
      "learning_rate": 0.0005891320099934668,
      "loss": 2.216,
      "step": 19801
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.5813193321228027,
      "learning_rate": 0.0005891309189215763,
      "loss": 2.357,
      "step": 19802
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6245505213737488,
      "learning_rate": 0.0005891298277959311,
      "loss": 2.2946,
      "step": 19803
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.5843949913978577,
      "learning_rate": 0.000589128736616531,
      "loss": 2.3749,
      "step": 19804
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.5786961913108826,
      "learning_rate": 0.0005891276453833766,
      "loss": 2.4205,
      "step": 19805
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.5791752338409424,
      "learning_rate": 0.0005891265540964677,
      "loss": 2.3051,
      "step": 19806
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6052854657173157,
      "learning_rate": 0.0005891254627558048,
      "loss": 2.245,
      "step": 19807
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.5754321217536926,
      "learning_rate": 0.0005891243713613882,
      "loss": 2.4796,
      "step": 19808
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6303190588951111,
      "learning_rate": 0.0005891232799132178,
      "loss": 2.4568,
      "step": 19809
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.5893928408622742,
      "learning_rate": 0.0005891221884112938,
      "loss": 2.2937,
      "step": 19810
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6447031497955322,
      "learning_rate": 0.0005891210968556167,
      "loss": 2.1302,
      "step": 19811
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6233313679695129,
      "learning_rate": 0.0005891200052461865,
      "loss": 2.3179,
      "step": 19812
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6368551850318909,
      "learning_rate": 0.0005891189135830033,
      "loss": 2.4487,
      "step": 19813
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.5791592001914978,
      "learning_rate": 0.0005891178218660676,
      "loss": 2.2662,
      "step": 19814
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.632991373538971,
      "learning_rate": 0.0005891167300953794,
      "loss": 2.3462,
      "step": 19815
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.5861008763313293,
      "learning_rate": 0.0005891156382709389,
      "loss": 2.3844,
      "step": 19816
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.5685020089149475,
      "learning_rate": 0.0005891145463927463,
      "loss": 2.3512,
      "step": 19817
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6395618319511414,
      "learning_rate": 0.0005891134544608019,
      "loss": 2.3726,
      "step": 19818
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.631292998790741,
      "learning_rate": 0.000589112362475106,
      "loss": 2.402,
      "step": 19819
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.611012876033783,
      "learning_rate": 0.0005891112704356584,
      "loss": 2.3647,
      "step": 19820
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.666877269744873,
      "learning_rate": 0.0005891101783424597,
      "loss": 2.2413,
      "step": 19821
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6703782081604004,
      "learning_rate": 0.0005891090861955099,
      "loss": 2.3613,
      "step": 19822
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.667605996131897,
      "learning_rate": 0.0005891079939948094,
      "loss": 2.4358,
      "step": 19823
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.7245851159095764,
      "learning_rate": 0.0005891069017403581,
      "loss": 2.2389,
      "step": 19824
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.62251216173172,
      "learning_rate": 0.0005891058094321565,
      "loss": 2.2982,
      "step": 19825
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6295362114906311,
      "learning_rate": 0.0005891047170702046,
      "loss": 2.6135,
      "step": 19826
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6428106427192688,
      "learning_rate": 0.0005891036246545027,
      "loss": 2.4086,
      "step": 19827
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.626507580280304,
      "learning_rate": 0.0005891025321850509,
      "loss": 2.3435,
      "step": 19828
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6417906880378723,
      "learning_rate": 0.0005891014396618495,
      "loss": 2.5227,
      "step": 19829
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6066588759422302,
      "learning_rate": 0.0005891003470848989,
      "loss": 2.3547,
      "step": 19830
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.5876665115356445,
      "learning_rate": 0.0005890992544541989,
      "loss": 2.3928,
      "step": 19831
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.568880558013916,
      "learning_rate": 0.0005890981617697499,
      "loss": 2.2301,
      "step": 19832
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.5796816349029541,
      "learning_rate": 0.0005890970690315521,
      "loss": 2.4559,
      "step": 19833
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6303099393844604,
      "learning_rate": 0.0005890959762396056,
      "loss": 2.4111,
      "step": 19834
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.5645760297775269,
      "learning_rate": 0.0005890948833939108,
      "loss": 2.2072,
      "step": 19835
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.603410542011261,
      "learning_rate": 0.0005890937904944679,
      "loss": 2.3353,
      "step": 19836
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.5735573768615723,
      "learning_rate": 0.000589092697541277,
      "loss": 2.3142,
      "step": 19837
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.5875784158706665,
      "learning_rate": 0.0005890916045343381,
      "loss": 2.3835,
      "step": 19838
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6114844083786011,
      "learning_rate": 0.0005890905114736517,
      "loss": 2.4437,
      "step": 19839
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6028801798820496,
      "learning_rate": 0.000589089418359218,
      "loss": 2.2505,
      "step": 19840
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6000057458877563,
      "learning_rate": 0.0005890883251910371,
      "loss": 2.3293,
      "step": 19841
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.5695698857307434,
      "learning_rate": 0.0005890872319691091,
      "loss": 2.3415,
      "step": 19842
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6583850979804993,
      "learning_rate": 0.0005890861386934345,
      "loss": 2.4615,
      "step": 19843
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6568188667297363,
      "learning_rate": 0.0005890850453640134,
      "loss": 2.2267,
      "step": 19844
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6163493394851685,
      "learning_rate": 0.0005890839519808456,
      "loss": 2.4433,
      "step": 19845
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.5959498882293701,
      "learning_rate": 0.0005890828585439319,
      "loss": 2.3546,
      "step": 19846
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.5914461016654968,
      "learning_rate": 0.0005890817650532721,
      "loss": 2.2631,
      "step": 19847
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6207512021064758,
      "learning_rate": 0.0005890806715088667,
      "loss": 2.4219,
      "step": 19848
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6534784436225891,
      "learning_rate": 0.0005890795779107156,
      "loss": 2.3953,
      "step": 19849
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6561049818992615,
      "learning_rate": 0.0005890784842588193,
      "loss": 2.6097,
      "step": 19850
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6715332865715027,
      "learning_rate": 0.0005890773905531777,
      "loss": 2.386,
      "step": 19851
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.577782154083252,
      "learning_rate": 0.0005890762967937912,
      "loss": 2.3123,
      "step": 19852
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.5819810032844543,
      "learning_rate": 0.00058907520298066,
      "loss": 2.4708,
      "step": 19853
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6427773833274841,
      "learning_rate": 0.0005890741091137842,
      "loss": 2.4991,
      "step": 19854
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6045777201652527,
      "learning_rate": 0.0005890730151931641,
      "loss": 2.3426,
      "step": 19855
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.63009113073349,
      "learning_rate": 0.0005890719212188,
      "loss": 2.5161,
      "step": 19856
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.5906200408935547,
      "learning_rate": 0.0005890708271906917,
      "loss": 2.1059,
      "step": 19857
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6350219249725342,
      "learning_rate": 0.0005890697331088399,
      "loss": 2.3036,
      "step": 19858
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.5853891372680664,
      "learning_rate": 0.0005890686389732444,
      "loss": 2.277,
      "step": 19859
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6503416895866394,
      "learning_rate": 0.0005890675447839057,
      "loss": 2.5698,
      "step": 19860
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6653226017951965,
      "learning_rate": 0.000589066450540824,
      "loss": 2.4351,
      "step": 19861
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.615462601184845,
      "learning_rate": 0.0005890653562439992,
      "loss": 2.0593,
      "step": 19862
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6645674705505371,
      "learning_rate": 0.0005890642618934317,
      "loss": 2.2892,
      "step": 19863
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6895731091499329,
      "learning_rate": 0.0005890631674891218,
      "loss": 2.3628,
      "step": 19864
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6039180755615234,
      "learning_rate": 0.0005890620730310696,
      "loss": 2.2345,
      "step": 19865
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6581975817680359,
      "learning_rate": 0.0005890609785192752,
      "loss": 2.3581,
      "step": 19866
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6703171730041504,
      "learning_rate": 0.0005890598839537389,
      "loss": 2.3616,
      "step": 19867
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6214330196380615,
      "learning_rate": 0.000589058789334461,
      "loss": 2.3455,
      "step": 19868
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6256765127182007,
      "learning_rate": 0.0005890576946614417,
      "loss": 2.3868,
      "step": 19869
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.7112361788749695,
      "learning_rate": 0.000589056599934681,
      "loss": 2.4544,
      "step": 19870
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6861073970794678,
      "learning_rate": 0.0005890555051541792,
      "loss": 2.3537,
      "step": 19871
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6408506631851196,
      "learning_rate": 0.0005890544103199367,
      "loss": 2.2901,
      "step": 19872
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6023632287979126,
      "learning_rate": 0.0005890533154319533,
      "loss": 2.4826,
      "step": 19873
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6087659597396851,
      "learning_rate": 0.0005890522204902294,
      "loss": 2.1788,
      "step": 19874
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6404205560684204,
      "learning_rate": 0.0005890511254947655,
      "loss": 2.305,
      "step": 19875
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.625445544719696,
      "learning_rate": 0.0005890500304455613,
      "loss": 2.1162,
      "step": 19876
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.5985286831855774,
      "learning_rate": 0.0005890489353426174,
      "loss": 2.3683,
      "step": 19877
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6667014360427856,
      "learning_rate": 0.0005890478401859337,
      "loss": 2.3906,
      "step": 19878
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6299923658370972,
      "learning_rate": 0.0005890467449755107,
      "loss": 2.3616,
      "step": 19879
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6227346658706665,
      "learning_rate": 0.0005890456497113484,
      "loss": 2.4736,
      "step": 19880
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6829869747161865,
      "learning_rate": 0.000589044554393447,
      "loss": 2.4058,
      "step": 19881
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.5609011054039001,
      "learning_rate": 0.0005890434590218068,
      "loss": 2.1914,
      "step": 19882
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6391488909721375,
      "learning_rate": 0.000589042363596428,
      "loss": 2.4067,
      "step": 19883
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6660096049308777,
      "learning_rate": 0.0005890412681173107,
      "loss": 2.4544,
      "step": 19884
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6111307740211487,
      "learning_rate": 0.0005890401725844553,
      "loss": 2.4889,
      "step": 19885
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.5662770867347717,
      "learning_rate": 0.0005890390769978617,
      "loss": 2.3457,
      "step": 19886
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6075985431671143,
      "learning_rate": 0.0005890379813575304,
      "loss": 2.4869,
      "step": 19887
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.641505241394043,
      "learning_rate": 0.0005890368856634614,
      "loss": 2.285,
      "step": 19888
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.957766056060791,
      "learning_rate": 0.000589035789915655,
      "loss": 2.4422,
      "step": 19889
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.5722981095314026,
      "learning_rate": 0.0005890346941141116,
      "loss": 2.2986,
      "step": 19890
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.7608084678649902,
      "learning_rate": 0.000589033598258831,
      "loss": 2.4755,
      "step": 19891
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.7357677221298218,
      "learning_rate": 0.0005890325023498136,
      "loss": 2.5485,
      "step": 19892
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.5759683847427368,
      "learning_rate": 0.0005890314063870596,
      "loss": 2.1709,
      "step": 19893
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.5952354073524475,
      "learning_rate": 0.0005890303103705692,
      "loss": 2.3991,
      "step": 19894
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6331331729888916,
      "learning_rate": 0.0005890292143003427,
      "loss": 2.2504,
      "step": 19895
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6705763339996338,
      "learning_rate": 0.0005890281181763802,
      "loss": 2.3119,
      "step": 19896
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.593580961227417,
      "learning_rate": 0.0005890270219986818,
      "loss": 2.3067,
      "step": 19897
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6587485074996948,
      "learning_rate": 0.0005890259257672479,
      "loss": 2.3892,
      "step": 19898
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.7050133347511292,
      "learning_rate": 0.0005890248294820786,
      "loss": 2.2798,
      "step": 19899
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6204468011856079,
      "learning_rate": 0.0005890237331431741,
      "loss": 2.5905,
      "step": 19900
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.5867926478385925,
      "learning_rate": 0.0005890226367505347,
      "loss": 2.3586,
      "step": 19901
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6475635170936584,
      "learning_rate": 0.0005890215403041606,
      "loss": 2.4541,
      "step": 19902
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.7546780705451965,
      "learning_rate": 0.0005890204438040518,
      "loss": 2.2461,
      "step": 19903
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6555845141410828,
      "learning_rate": 0.0005890193472502087,
      "loss": 2.4619,
      "step": 19904
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.800309956073761,
      "learning_rate": 0.0005890182506426314,
      "loss": 2.3374,
      "step": 19905
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.862639844417572,
      "learning_rate": 0.0005890171539813203,
      "loss": 2.157,
      "step": 19906
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6053760647773743,
      "learning_rate": 0.0005890160572662752,
      "loss": 2.4443,
      "step": 19907
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.0483721494674683,
      "learning_rate": 0.0005890149604974968,
      "loss": 2.3855,
      "step": 19908
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.692871630191803,
      "learning_rate": 0.0005890138636749849,
      "loss": 2.2185,
      "step": 19909
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.7147870659828186,
      "learning_rate": 0.00058901276679874,
      "loss": 2.3963,
      "step": 19910
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.8436168432235718,
      "learning_rate": 0.0005890116698687621,
      "loss": 2.3601,
      "step": 19911
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6148374080657959,
      "learning_rate": 0.0005890105728850515,
      "loss": 2.1259,
      "step": 19912
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6457136273384094,
      "learning_rate": 0.0005890094758476084,
      "loss": 2.2219,
      "step": 19913
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6607803702354431,
      "learning_rate": 0.0005890083787564329,
      "loss": 2.2097,
      "step": 19914
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.705422043800354,
      "learning_rate": 0.0005890072816115253,
      "loss": 2.3951,
      "step": 19915
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6817917823791504,
      "learning_rate": 0.0005890061844128859,
      "loss": 2.4336,
      "step": 19916
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.616616427898407,
      "learning_rate": 0.0005890050871605147,
      "loss": 2.3147,
      "step": 19917
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6939421892166138,
      "learning_rate": 0.000589003989854412,
      "loss": 2.1533,
      "step": 19918
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.7244040966033936,
      "learning_rate": 0.0005890028924945781,
      "loss": 2.4885,
      "step": 19919
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6301900148391724,
      "learning_rate": 0.000589001795081013,
      "loss": 2.458,
      "step": 19920
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.7423888444900513,
      "learning_rate": 0.000589000697613717,
      "loss": 2.4854,
      "step": 19921
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6812490820884705,
      "learning_rate": 0.0005889996000926904,
      "loss": 2.532,
      "step": 19922
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.5862059593200684,
      "learning_rate": 0.0005889985025179332,
      "loss": 2.5033,
      "step": 19923
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.5905715227127075,
      "learning_rate": 0.0005889974048894458,
      "loss": 2.354,
      "step": 19924
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6954615712165833,
      "learning_rate": 0.0005889963072072283,
      "loss": 2.2985,
      "step": 19925
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.7333570122718811,
      "learning_rate": 0.0005889952094712811,
      "loss": 2.5096,
      "step": 19926
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6056604981422424,
      "learning_rate": 0.000588994111681604,
      "loss": 2.4213,
      "step": 19927
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6151033639907837,
      "learning_rate": 0.0005889930138381975,
      "loss": 2.4592,
      "step": 19928
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6562976241111755,
      "learning_rate": 0.0005889919159410618,
      "loss": 2.1597,
      "step": 19929
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6785181760787964,
      "learning_rate": 0.0005889908179901971,
      "loss": 2.4646,
      "step": 19930
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6232467889785767,
      "learning_rate": 0.0005889897199856034,
      "loss": 2.283,
      "step": 19931
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6615303158760071,
      "learning_rate": 0.0005889886219272812,
      "loss": 2.1525,
      "step": 19932
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.7235254645347595,
      "learning_rate": 0.0005889875238152304,
      "loss": 2.3549,
      "step": 19933
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.5865868330001831,
      "learning_rate": 0.0005889864256494515,
      "loss": 2.3834,
      "step": 19934
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6701450943946838,
      "learning_rate": 0.0005889853274299445,
      "loss": 2.6249,
      "step": 19935
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.7294657230377197,
      "learning_rate": 0.0005889842291567097,
      "loss": 2.4355,
      "step": 19936
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.560814380645752,
      "learning_rate": 0.0005889831308297473,
      "loss": 2.3819,
      "step": 19937
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6917672753334045,
      "learning_rate": 0.0005889820324490574,
      "loss": 2.5171,
      "step": 19938
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.7956876754760742,
      "learning_rate": 0.0005889809340146404,
      "loss": 2.0389,
      "step": 19939
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6360887289047241,
      "learning_rate": 0.0005889798355264963,
      "loss": 2.2558,
      "step": 19940
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6496652364730835,
      "learning_rate": 0.0005889787369846254,
      "loss": 2.1793,
      "step": 19941
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.595389187335968,
      "learning_rate": 0.000588977638389028,
      "loss": 2.3543,
      "step": 19942
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6331989765167236,
      "learning_rate": 0.0005889765397397041,
      "loss": 2.1867,
      "step": 19943
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6040075421333313,
      "learning_rate": 0.000588975441036654,
      "loss": 2.3017,
      "step": 19944
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6486386656761169,
      "learning_rate": 0.000588974342279878,
      "loss": 2.5228,
      "step": 19945
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.5650258660316467,
      "learning_rate": 0.0005889732434693761,
      "loss": 2.2824,
      "step": 19946
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.596293032169342,
      "learning_rate": 0.0005889721446051487,
      "loss": 2.3999,
      "step": 19947
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6482611894607544,
      "learning_rate": 0.0005889710456871959,
      "loss": 2.124,
      "step": 19948
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6674001216888428,
      "learning_rate": 0.0005889699467155179,
      "loss": 2.1214,
      "step": 19949
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6086909770965576,
      "learning_rate": 0.0005889688476901149,
      "loss": 2.4864,
      "step": 19950
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.7400578260421753,
      "learning_rate": 0.0005889677486109872,
      "loss": 2.4213,
      "step": 19951
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.5702235698699951,
      "learning_rate": 0.000588966649478135,
      "loss": 2.334,
      "step": 19952
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.5869911313056946,
      "learning_rate": 0.0005889655502915583,
      "loss": 2.4348,
      "step": 19953
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.621097981929779,
      "learning_rate": 0.0005889644510512575,
      "loss": 2.254,
      "step": 19954
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.5772818922996521,
      "learning_rate": 0.0005889633517572329,
      "loss": 2.2602,
      "step": 19955
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6780115365982056,
      "learning_rate": 0.0005889622524094843,
      "loss": 2.2289,
      "step": 19956
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.5895830392837524,
      "learning_rate": 0.0005889611530080123,
      "loss": 2.4082,
      "step": 19957
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.7882907390594482,
      "learning_rate": 0.000588960053552817,
      "loss": 2.5372,
      "step": 19958
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6121701002120972,
      "learning_rate": 0.0005889589540438987,
      "loss": 2.3833,
      "step": 19959
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.7089914083480835,
      "learning_rate": 0.0005889578544812572,
      "loss": 2.265,
      "step": 19960
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6248605847358704,
      "learning_rate": 0.0005889567548648931,
      "loss": 2.0928,
      "step": 19961
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6516909599304199,
      "learning_rate": 0.0005889556551948065,
      "loss": 2.0978,
      "step": 19962
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.5682076215744019,
      "learning_rate": 0.0005889545554709976,
      "loss": 2.491,
      "step": 19963
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6124280095100403,
      "learning_rate": 0.0005889534556934666,
      "loss": 2.127,
      "step": 19964
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6111162900924683,
      "learning_rate": 0.0005889523558622136,
      "loss": 2.484,
      "step": 19965
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6464693546295166,
      "learning_rate": 0.000588951255977239,
      "loss": 2.6933,
      "step": 19966
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6324582099914551,
      "learning_rate": 0.0005889501560385428,
      "loss": 2.4181,
      "step": 19967
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6638493537902832,
      "learning_rate": 0.0005889490560461254,
      "loss": 2.3658,
      "step": 19968
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6382937431335449,
      "learning_rate": 0.0005889479559999869,
      "loss": 2.3901,
      "step": 19969
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.5849992632865906,
      "learning_rate": 0.0005889468559001276,
      "loss": 2.3497,
      "step": 19970
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6435551047325134,
      "learning_rate": 0.0005889457557465475,
      "loss": 2.1575,
      "step": 19971
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6683706641197205,
      "learning_rate": 0.000588944655539247,
      "loss": 2.225,
      "step": 19972
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6079264283180237,
      "learning_rate": 0.0005889435552782262,
      "loss": 2.3219,
      "step": 19973
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.5658044219017029,
      "learning_rate": 0.0005889424549634854,
      "loss": 2.2593,
      "step": 19974
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6086711883544922,
      "learning_rate": 0.0005889413545950246,
      "loss": 2.6034,
      "step": 19975
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.5832304954528809,
      "learning_rate": 0.0005889402541728444,
      "loss": 2.3363,
      "step": 19976
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6071563959121704,
      "learning_rate": 0.0005889391536969445,
      "loss": 2.3315,
      "step": 19977
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6164506077766418,
      "learning_rate": 0.0005889380531673255,
      "loss": 2.351,
      "step": 19978
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6013817191123962,
      "learning_rate": 0.0005889369525839874,
      "loss": 2.448,
      "step": 19979
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6355580687522888,
      "learning_rate": 0.0005889358519469304,
      "loss": 2.6463,
      "step": 19980
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.5723319053649902,
      "learning_rate": 0.0005889347512561549,
      "loss": 2.2592,
      "step": 19981
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6235222220420837,
      "learning_rate": 0.000588933650511661,
      "loss": 2.1363,
      "step": 19982
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.7996794581413269,
      "learning_rate": 0.0005889325497134487,
      "loss": 2.5878,
      "step": 19983
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.675984263420105,
      "learning_rate": 0.0005889314488615185,
      "loss": 2.3965,
      "step": 19984
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6797811985015869,
      "learning_rate": 0.0005889303479558705,
      "loss": 2.7029,
      "step": 19985
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6590502858161926,
      "learning_rate": 0.0005889292469965049,
      "loss": 2.503,
      "step": 19986
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.7412282824516296,
      "learning_rate": 0.0005889281459834218,
      "loss": 2.6752,
      "step": 19987
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6402228474617004,
      "learning_rate": 0.0005889270449166217,
      "loss": 2.2843,
      "step": 19988
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6873735189437866,
      "learning_rate": 0.0005889259437961044,
      "loss": 2.4141,
      "step": 19989
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.8420676589012146,
      "learning_rate": 0.0005889248426218704,
      "loss": 2.4553,
      "step": 19990
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6616306304931641,
      "learning_rate": 0.0005889237413939199,
      "loss": 2.3968,
      "step": 19991
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6655629873275757,
      "learning_rate": 0.0005889226401122529,
      "loss": 2.3095,
      "step": 19992
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.8466308116912842,
      "learning_rate": 0.0005889215387768698,
      "loss": 2.3602,
      "step": 19993
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6615862846374512,
      "learning_rate": 0.0005889204373877707,
      "loss": 2.4342,
      "step": 19994
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6209032535552979,
      "learning_rate": 0.0005889193359449558,
      "loss": 2.552,
      "step": 19995
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6950071454048157,
      "learning_rate": 0.0005889182344484253,
      "loss": 2.2588,
      "step": 19996
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6839767098426819,
      "learning_rate": 0.0005889171328981797,
      "loss": 2.6496,
      "step": 19997
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6756007075309753,
      "learning_rate": 0.0005889160312942186,
      "loss": 2.1914,
      "step": 19998
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6660605072975159,
      "learning_rate": 0.0005889149296365428,
      "loss": 2.3768,
      "step": 19999
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.7129884362220764,
      "learning_rate": 0.0005889138279251522,
      "loss": 2.2999,
      "step": 20000
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6560854315757751,
      "learning_rate": 0.000588912726160047,
      "loss": 2.2214,
      "step": 20001
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6183344125747681,
      "learning_rate": 0.0005889116243412276,
      "loss": 2.4675,
      "step": 20002
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6035577654838562,
      "learning_rate": 0.000588910522468694,
      "loss": 2.1316,
      "step": 20003
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6130273342132568,
      "learning_rate": 0.0005889094205424466,
      "loss": 2.1867,
      "step": 20004
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6322999000549316,
      "learning_rate": 0.0005889083185624852,
      "loss": 2.4165,
      "step": 20005
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.5921987295150757,
      "learning_rate": 0.0005889072165288105,
      "loss": 2.3861,
      "step": 20006
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.7848691940307617,
      "learning_rate": 0.0005889061144414225,
      "loss": 2.1795,
      "step": 20007
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.5686087608337402,
      "learning_rate": 0.0005889050123003213,
      "loss": 2.37,
      "step": 20008
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6950764060020447,
      "learning_rate": 0.0005889039101055072,
      "loss": 2.4812,
      "step": 20009
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6104880571365356,
      "learning_rate": 0.0005889028078569805,
      "loss": 2.3916,
      "step": 20010
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.642136812210083,
      "learning_rate": 0.0005889017055547412,
      "loss": 2.3351,
      "step": 20011
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6180530190467834,
      "learning_rate": 0.0005889006031987898,
      "loss": 2.4215,
      "step": 20012
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6758033633232117,
      "learning_rate": 0.0005888995007891262,
      "loss": 2.44,
      "step": 20013
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.7822884917259216,
      "learning_rate": 0.0005888983983257508,
      "loss": 2.3991,
      "step": 20014
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.5975266695022583,
      "learning_rate": 0.0005888972958086637,
      "loss": 2.3562,
      "step": 20015
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6642191410064697,
      "learning_rate": 0.0005888961932378652,
      "loss": 2.4204,
      "step": 20016
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.7783423066139221,
      "learning_rate": 0.0005888950906133553,
      "loss": 2.2883,
      "step": 20017
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6303291916847229,
      "learning_rate": 0.0005888939879351345,
      "loss": 2.5077,
      "step": 20018
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6004005074501038,
      "learning_rate": 0.0005888928852032028,
      "loss": 2.4453,
      "step": 20019
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6352382302284241,
      "learning_rate": 0.0005888917824175603,
      "loss": 2.2996,
      "step": 20020
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6576938629150391,
      "learning_rate": 0.0005888906795782075,
      "loss": 2.4858,
      "step": 20021
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6847400069236755,
      "learning_rate": 0.0005888895766851445,
      "loss": 2.3989,
      "step": 20022
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.5836697220802307,
      "learning_rate": 0.0005888884737383715,
      "loss": 2.5132,
      "step": 20023
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6566895842552185,
      "learning_rate": 0.0005888873707378886,
      "loss": 2.308,
      "step": 20024
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6791690587997437,
      "learning_rate": 0.0005888862676836961,
      "loss": 2.3364,
      "step": 20025
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.5815874338150024,
      "learning_rate": 0.0005888851645757942,
      "loss": 2.2951,
      "step": 20026
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.633928120136261,
      "learning_rate": 0.0005888840614141831,
      "loss": 2.1532,
      "step": 20027
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.815597653388977,
      "learning_rate": 0.000588882958198863,
      "loss": 2.2717,
      "step": 20028
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.5837910175323486,
      "learning_rate": 0.0005888818549298341,
      "loss": 2.1532,
      "step": 20029
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6942067742347717,
      "learning_rate": 0.0005888807516070966,
      "loss": 2.3458,
      "step": 20030
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6326562166213989,
      "learning_rate": 0.0005888796482306507,
      "loss": 2.4579,
      "step": 20031
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.7131418585777283,
      "learning_rate": 0.0005888785448004966,
      "loss": 1.9673,
      "step": 20032
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.615929365158081,
      "learning_rate": 0.0005888774413166346,
      "loss": 2.3919,
      "step": 20033
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6283026337623596,
      "learning_rate": 0.0005888763377790648,
      "loss": 2.6099,
      "step": 20034
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6783023476600647,
      "learning_rate": 0.0005888752341877875,
      "loss": 2.3316,
      "step": 20035
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.570243775844574,
      "learning_rate": 0.0005888741305428028,
      "loss": 2.3753,
      "step": 20036
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.5633139610290527,
      "learning_rate": 0.0005888730268441109,
      "loss": 2.2931,
      "step": 20037
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.9245963096618652,
      "learning_rate": 0.000588871923091712,
      "loss": 2.4707,
      "step": 20038
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.696956217288971,
      "learning_rate": 0.0005888708192856065,
      "loss": 2.4818,
      "step": 20039
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6733085513114929,
      "learning_rate": 0.0005888697154257943,
      "loss": 2.2834,
      "step": 20040
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.7260119915008545,
      "learning_rate": 0.000588868611512276,
      "loss": 2.1461,
      "step": 20041
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6252374649047852,
      "learning_rate": 0.0005888675075450514,
      "loss": 2.3438,
      "step": 20042
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.585715115070343,
      "learning_rate": 0.000588866403524121,
      "loss": 2.3444,
      "step": 20043
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.7269707322120667,
      "learning_rate": 0.0005888652994494847,
      "loss": 2.2889,
      "step": 20044
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6814842820167542,
      "learning_rate": 0.000588864195321143,
      "loss": 2.3437,
      "step": 20045
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.5826891660690308,
      "learning_rate": 0.000588863091139096,
      "loss": 2.3616,
      "step": 20046
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6088009476661682,
      "learning_rate": 0.0005888619869033439,
      "loss": 2.3062,
      "step": 20047
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.8398049473762512,
      "learning_rate": 0.0005888608826138868,
      "loss": 2.3237,
      "step": 20048
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.5674474835395813,
      "learning_rate": 0.0005888597782707252,
      "loss": 2.3496,
      "step": 20049
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6511173844337463,
      "learning_rate": 0.000588858673873859,
      "loss": 2.4932,
      "step": 20050
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.7443585991859436,
      "learning_rate": 0.0005888575694232885,
      "loss": 2.52,
      "step": 20051
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.7152652144432068,
      "learning_rate": 0.0005888564649190139,
      "loss": 2.4133,
      "step": 20052
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.620975911617279,
      "learning_rate": 0.0005888553603610355,
      "loss": 2.4575,
      "step": 20053
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6895487308502197,
      "learning_rate": 0.0005888542557493533,
      "loss": 2.3904,
      "step": 20054
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.7120354771614075,
      "learning_rate": 0.0005888531510839678,
      "loss": 2.4735,
      "step": 20055
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6204076409339905,
      "learning_rate": 0.000588852046364879,
      "loss": 2.3891,
      "step": 20056
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6621137857437134,
      "learning_rate": 0.0005888509415920872,
      "loss": 2.2227,
      "step": 20057
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.678709864616394,
      "learning_rate": 0.0005888498367655925,
      "loss": 2.2197,
      "step": 20058
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.7042774558067322,
      "learning_rate": 0.0005888487318853952,
      "loss": 2.6225,
      "step": 20059
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6195453405380249,
      "learning_rate": 0.0005888476269514954,
      "loss": 2.5277,
      "step": 20060
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.633938193321228,
      "learning_rate": 0.0005888465219638934,
      "loss": 2.3062,
      "step": 20061
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.7679898142814636,
      "learning_rate": 0.0005888454169225894,
      "loss": 2.4979,
      "step": 20062
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6683292984962463,
      "learning_rate": 0.0005888443118275837,
      "loss": 2.3799,
      "step": 20063
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6217067241668701,
      "learning_rate": 0.0005888432066788762,
      "loss": 2.3033,
      "step": 20064
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.5803296566009521,
      "learning_rate": 0.0005888421014764673,
      "loss": 2.2538,
      "step": 20065
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6484450101852417,
      "learning_rate": 0.0005888409962203573,
      "loss": 2.46,
      "step": 20066
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.744011640548706,
      "learning_rate": 0.0005888398909105463,
      "loss": 2.2658,
      "step": 20067
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.5989468693733215,
      "learning_rate": 0.0005888387855470345,
      "loss": 2.2975,
      "step": 20068
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.7785720229148865,
      "learning_rate": 0.0005888376801298221,
      "loss": 2.3938,
      "step": 20069
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.7923465967178345,
      "learning_rate": 0.0005888365746589094,
      "loss": 2.5093,
      "step": 20070
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6006879806518555,
      "learning_rate": 0.0005888354691342964,
      "loss": 2.3855,
      "step": 20071
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.7487168312072754,
      "learning_rate": 0.0005888343635559835,
      "loss": 2.3183,
      "step": 20072
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.617677628993988,
      "learning_rate": 0.0005888332579239708,
      "loss": 2.2886,
      "step": 20073
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6623137593269348,
      "learning_rate": 0.0005888321522382586,
      "loss": 2.1944,
      "step": 20074
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.5565118193626404,
      "learning_rate": 0.0005888310464988469,
      "loss": 2.3244,
      "step": 20075
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6646037697792053,
      "learning_rate": 0.0005888299407057363,
      "loss": 2.1762,
      "step": 20076
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6393905282020569,
      "learning_rate": 0.0005888288348589266,
      "loss": 2.5861,
      "step": 20077
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6181954741477966,
      "learning_rate": 0.0005888277289584181,
      "loss": 2.2425,
      "step": 20078
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.5942330360412598,
      "learning_rate": 0.0005888266230042113,
      "loss": 2.4023,
      "step": 20079
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.7022625207901001,
      "learning_rate": 0.0005888255169963061,
      "loss": 2.366,
      "step": 20080
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6622912883758545,
      "learning_rate": 0.0005888244109347027,
      "loss": 2.2589,
      "step": 20081
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.5533978939056396,
      "learning_rate": 0.0005888233048194014,
      "loss": 2.4891,
      "step": 20082
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6955373883247375,
      "learning_rate": 0.0005888221986504026,
      "loss": 2.3439,
      "step": 20083
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6411138772964478,
      "learning_rate": 0.000588821092427706,
      "loss": 2.3158,
      "step": 20084
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.7479783296585083,
      "learning_rate": 0.0005888199861513123,
      "loss": 2.406,
      "step": 20085
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.7051032781600952,
      "learning_rate": 0.0005888188798212215,
      "loss": 2.2737,
      "step": 20086
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6553272604942322,
      "learning_rate": 0.0005888177734374336,
      "loss": 2.3602,
      "step": 20087
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6938564777374268,
      "learning_rate": 0.0005888166669999493,
      "loss": 2.4528,
      "step": 20088
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.7322353720664978,
      "learning_rate": 0.0005888155605087684,
      "loss": 2.6583,
      "step": 20089
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6136193871498108,
      "learning_rate": 0.0005888144539638913,
      "loss": 2.3236,
      "step": 20090
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.5838804841041565,
      "learning_rate": 0.000588813347365318,
      "loss": 2.4673,
      "step": 20091
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.699453592300415,
      "learning_rate": 0.0005888122407130491,
      "loss": 2.2036,
      "step": 20092
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6843780279159546,
      "learning_rate": 0.0005888111340070842,
      "loss": 2.2318,
      "step": 20093
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6148914694786072,
      "learning_rate": 0.0005888100272474242,
      "loss": 2.4904,
      "step": 20094
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6690001487731934,
      "learning_rate": 0.0005888089204340688,
      "loss": 2.4035,
      "step": 20095
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6591535210609436,
      "learning_rate": 0.0005888078135670184,
      "loss": 2.3652,
      "step": 20096
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6119675636291504,
      "learning_rate": 0.0005888067066462732,
      "loss": 2.4131,
      "step": 20097
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.5950195789337158,
      "learning_rate": 0.0005888055996718333,
      "loss": 2.1928,
      "step": 20098
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6414496898651123,
      "learning_rate": 0.000588804492643699,
      "loss": 2.6967,
      "step": 20099
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.7002755403518677,
      "learning_rate": 0.0005888033855618706,
      "loss": 2.6746,
      "step": 20100
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.5853836536407471,
      "learning_rate": 0.0005888022784263481,
      "loss": 2.6534,
      "step": 20101
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.629673957824707,
      "learning_rate": 0.0005888011712371319,
      "loss": 2.3806,
      "step": 20102
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6455566883087158,
      "learning_rate": 0.000588800063994222,
      "loss": 2.2143,
      "step": 20103
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.7399397492408752,
      "learning_rate": 0.0005887989566976189,
      "loss": 2.3245,
      "step": 20104
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6433131694793701,
      "learning_rate": 0.0005887978493473224,
      "loss": 2.2472,
      "step": 20105
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6121687889099121,
      "learning_rate": 0.000588796741943333,
      "loss": 2.3076,
      "step": 20106
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.7083669304847717,
      "learning_rate": 0.0005887956344856508,
      "loss": 2.1455,
      "step": 20107
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6097461581230164,
      "learning_rate": 0.0005887945269742762,
      "loss": 2.4327,
      "step": 20108
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6475360989570618,
      "learning_rate": 0.0005887934194092091,
      "loss": 2.4135,
      "step": 20109
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6369929909706116,
      "learning_rate": 0.0005887923117904499,
      "loss": 2.1798,
      "step": 20110
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.7058691382408142,
      "learning_rate": 0.0005887912041179987,
      "loss": 2.5901,
      "step": 20111
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6666561365127563,
      "learning_rate": 0.0005887900963918558,
      "loss": 2.2356,
      "step": 20112
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6208491325378418,
      "learning_rate": 0.0005887889886120214,
      "loss": 2.2775,
      "step": 20113
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6692763566970825,
      "learning_rate": 0.0005887878807784956,
      "loss": 2.3354,
      "step": 20114
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.7119508981704712,
      "learning_rate": 0.0005887867728912788,
      "loss": 2.4063,
      "step": 20115
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.5997087955474854,
      "learning_rate": 0.000588785664950371,
      "loss": 2.3385,
      "step": 20116
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6298611760139465,
      "learning_rate": 0.0005887845569557725,
      "loss": 2.4495,
      "step": 20117
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6788062453269958,
      "learning_rate": 0.0005887834489074835,
      "loss": 2.5379,
      "step": 20118
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6719114184379578,
      "learning_rate": 0.0005887823408055042,
      "loss": 2.458,
      "step": 20119
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.5981507301330566,
      "learning_rate": 0.0005887812326498349,
      "loss": 2.2058,
      "step": 20120
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6077229380607605,
      "learning_rate": 0.0005887801244404756,
      "loss": 2.3641,
      "step": 20121
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6542785167694092,
      "learning_rate": 0.0005887790161774267,
      "loss": 2.2515,
      "step": 20122
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.8421675562858582,
      "learning_rate": 0.0005887779078606883,
      "loss": 2.2757,
      "step": 20123
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6443635821342468,
      "learning_rate": 0.0005887767994902606,
      "loss": 2.261,
      "step": 20124
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6315926909446716,
      "learning_rate": 0.0005887756910661439,
      "loss": 2.5693,
      "step": 20125
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6843317151069641,
      "learning_rate": 0.0005887745825883384,
      "loss": 2.3208,
      "step": 20126
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6916472315788269,
      "learning_rate": 0.0005887734740568441,
      "loss": 2.3606,
      "step": 20127
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6509028673171997,
      "learning_rate": 0.0005887723654716615,
      "loss": 2.5553,
      "step": 20128
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6428075432777405,
      "learning_rate": 0.0005887712568327906,
      "loss": 2.4063,
      "step": 20129
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.7004150748252869,
      "learning_rate": 0.0005887701481402317,
      "loss": 2.4735,
      "step": 20130
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6765354871749878,
      "learning_rate": 0.0005887690393939849,
      "loss": 2.3589,
      "step": 20131
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6704404354095459,
      "learning_rate": 0.0005887679305940506,
      "loss": 2.373,
      "step": 20132
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.695415735244751,
      "learning_rate": 0.0005887668217404289,
      "loss": 2.3449,
      "step": 20133
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6481074690818787,
      "learning_rate": 0.0005887657128331199,
      "loss": 2.3406,
      "step": 20134
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6379085779190063,
      "learning_rate": 0.0005887646038721239,
      "loss": 2.334,
      "step": 20135
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6039347052574158,
      "learning_rate": 0.0005887634948574411,
      "loss": 2.3268,
      "step": 20136
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6032497882843018,
      "learning_rate": 0.0005887623857890719,
      "loss": 2.1783,
      "step": 20137
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.5482355356216431,
      "learning_rate": 0.000588761276667016,
      "loss": 2.189,
      "step": 20138
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.610471785068512,
      "learning_rate": 0.0005887601674912742,
      "loss": 2.5647,
      "step": 20139
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.641883373260498,
      "learning_rate": 0.0005887590582618464,
      "loss": 2.2697,
      "step": 20140
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.5920238494873047,
      "learning_rate": 0.0005887579489787328,
      "loss": 2.3513,
      "step": 20141
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6095806956291199,
      "learning_rate": 0.0005887568396419336,
      "loss": 2.539,
      "step": 20142
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.5633605122566223,
      "learning_rate": 0.0005887557302514491,
      "loss": 2.2713,
      "step": 20143
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6079790592193604,
      "learning_rate": 0.0005887546208072794,
      "loss": 2.3309,
      "step": 20144
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.5891544222831726,
      "learning_rate": 0.0005887535113094248,
      "loss": 2.2556,
      "step": 20145
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6413944959640503,
      "learning_rate": 0.0005887524017578855,
      "loss": 2.4347,
      "step": 20146
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.576674222946167,
      "learning_rate": 0.0005887512921526616,
      "loss": 2.3329,
      "step": 20147
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.661961555480957,
      "learning_rate": 0.0005887501824937535,
      "loss": 2.3041,
      "step": 20148
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6001736521720886,
      "learning_rate": 0.0005887490727811613,
      "loss": 2.316,
      "step": 20149
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6176961660385132,
      "learning_rate": 0.0005887479630148851,
      "loss": 2.4424,
      "step": 20150
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6566501259803772,
      "learning_rate": 0.0005887468531949252,
      "loss": 2.4547,
      "step": 20151
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6858675479888916,
      "learning_rate": 0.0005887457433212819,
      "loss": 2.4723,
      "step": 20152
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6121724247932434,
      "learning_rate": 0.0005887446333939552,
      "loss": 2.5042,
      "step": 20153
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6460424661636353,
      "learning_rate": 0.0005887435234129456,
      "loss": 2.3056,
      "step": 20154
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.7146596908569336,
      "learning_rate": 0.0005887424133782529,
      "loss": 2.3256,
      "step": 20155
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.7789240479469299,
      "learning_rate": 0.0005887413032898776,
      "loss": 2.2454,
      "step": 20156
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6948626041412354,
      "learning_rate": 0.00058874019314782,
      "loss": 2.5078,
      "step": 20157
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6652244329452515,
      "learning_rate": 0.00058873908295208,
      "loss": 2.5722,
      "step": 20158
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6526780724525452,
      "learning_rate": 0.000588737972702658,
      "loss": 2.2332,
      "step": 20159
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.5940232276916504,
      "learning_rate": 0.0005887368623995541,
      "loss": 2.3131,
      "step": 20160
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6272929906845093,
      "learning_rate": 0.0005887357520427687,
      "loss": 2.1446,
      "step": 20161
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.7058641314506531,
      "learning_rate": 0.0005887346416323018,
      "loss": 2.3894,
      "step": 20162
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6224063634872437,
      "learning_rate": 0.0005887335311681537,
      "loss": 2.4372,
      "step": 20163
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6530404090881348,
      "learning_rate": 0.0005887324206503245,
      "loss": 2.3625,
      "step": 20164
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.5951699614524841,
      "learning_rate": 0.0005887313100788145,
      "loss": 2.4713,
      "step": 20165
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6989395022392273,
      "learning_rate": 0.000588730199453624,
      "loss": 2.2751,
      "step": 20166
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.7299672365188599,
      "learning_rate": 0.000588729088774753,
      "loss": 2.4892,
      "step": 20167
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.5904514193534851,
      "learning_rate": 0.0005887279780422018,
      "loss": 2.4079,
      "step": 20168
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.657319188117981,
      "learning_rate": 0.0005887268672559707,
      "loss": 2.1365,
      "step": 20169
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6054277420043945,
      "learning_rate": 0.0005887257564160598,
      "loss": 2.1267,
      "step": 20170
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6340782642364502,
      "learning_rate": 0.0005887246455224694,
      "loss": 2.5228,
      "step": 20171
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6230682134628296,
      "learning_rate": 0.0005887235345751996,
      "loss": 2.4637,
      "step": 20172
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.637399435043335,
      "learning_rate": 0.0005887224235742506,
      "loss": 2.3812,
      "step": 20173
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6366235613822937,
      "learning_rate": 0.0005887213125196228,
      "loss": 2.4282,
      "step": 20174
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6098031401634216,
      "learning_rate": 0.000588720201411316,
      "loss": 2.5725,
      "step": 20175
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.5943068861961365,
      "learning_rate": 0.0005887190902493308,
      "loss": 2.2821,
      "step": 20176
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6101806163787842,
      "learning_rate": 0.0005887179790336673,
      "loss": 2.4531,
      "step": 20177
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6793304085731506,
      "learning_rate": 0.0005887168677643257,
      "loss": 2.1386,
      "step": 20178
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.5763719081878662,
      "learning_rate": 0.000588715756441306,
      "loss": 2.3573,
      "step": 20179
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.630081295967102,
      "learning_rate": 0.0005887146450646088,
      "loss": 2.2561,
      "step": 20180
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.5868640542030334,
      "learning_rate": 0.0005887135336342341,
      "loss": 2.2585,
      "step": 20181
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.5997222065925598,
      "learning_rate": 0.0005887124221501819,
      "loss": 2.3798,
      "step": 20182
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.621483325958252,
      "learning_rate": 0.0005887113106124528,
      "loss": 2.5571,
      "step": 20183
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6134450435638428,
      "learning_rate": 0.0005887101990210467,
      "loss": 2.3554,
      "step": 20184
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6897093653678894,
      "learning_rate": 0.000588709087375964,
      "loss": 2.3505,
      "step": 20185
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6019213199615479,
      "learning_rate": 0.0005887079756772048,
      "loss": 2.3193,
      "step": 20186
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.5992847084999084,
      "learning_rate": 0.0005887068639247693,
      "loss": 2.2511,
      "step": 20187
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.5672641396522522,
      "learning_rate": 0.0005887057521186577,
      "loss": 2.2888,
      "step": 20188
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6299306154251099,
      "learning_rate": 0.0005887046402588705,
      "loss": 2.2367,
      "step": 20189
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6879687905311584,
      "learning_rate": 0.0005887035283454075,
      "loss": 2.266,
      "step": 20190
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6229931116104126,
      "learning_rate": 0.0005887024163782691,
      "loss": 2.5006,
      "step": 20191
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6170170307159424,
      "learning_rate": 0.0005887013043574553,
      "loss": 2.4868,
      "step": 20192
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6387842893600464,
      "learning_rate": 0.0005887001922829667,
      "loss": 2.2701,
      "step": 20193
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.5605010986328125,
      "learning_rate": 0.0005886990801548032,
      "loss": 2.4137,
      "step": 20194
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.5989656448364258,
      "learning_rate": 0.0005886979679729651,
      "loss": 2.4274,
      "step": 20195
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.582274317741394,
      "learning_rate": 0.0005886968557374526,
      "loss": 2.2473,
      "step": 20196
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6260966658592224,
      "learning_rate": 0.0005886957434482658,
      "loss": 2.3118,
      "step": 20197
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.618867039680481,
      "learning_rate": 0.0005886946311054051,
      "loss": 2.3135,
      "step": 20198
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6752640008926392,
      "learning_rate": 0.0005886935187088706,
      "loss": 2.3459,
      "step": 20199
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6276578903198242,
      "learning_rate": 0.0005886924062586626,
      "loss": 2.2341,
      "step": 20200
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6088931560516357,
      "learning_rate": 0.0005886912937547812,
      "loss": 2.2439,
      "step": 20201
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6286481022834778,
      "learning_rate": 0.0005886901811972266,
      "loss": 2.3719,
      "step": 20202
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6689923405647278,
      "learning_rate": 0.0005886890685859991,
      "loss": 2.5409,
      "step": 20203
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6077237129211426,
      "learning_rate": 0.0005886879559210987,
      "loss": 2.4899,
      "step": 20204
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6646762490272522,
      "learning_rate": 0.000588686843202526,
      "loss": 2.3407,
      "step": 20205
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.654174268245697,
      "learning_rate": 0.0005886857304302809,
      "loss": 2.4921,
      "step": 20206
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6436002254486084,
      "learning_rate": 0.0005886846176043635,
      "loss": 2.346,
      "step": 20207
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6540073752403259,
      "learning_rate": 0.0005886835047247742,
      "loss": 2.5731,
      "step": 20208
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6244359612464905,
      "learning_rate": 0.0005886823917915134,
      "loss": 2.3019,
      "step": 20209
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.608641505241394,
      "learning_rate": 0.000588681278804581,
      "loss": 2.4407,
      "step": 20210
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.7056922912597656,
      "learning_rate": 0.0005886801657639773,
      "loss": 2.22,
      "step": 20211
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.5672862529754639,
      "learning_rate": 0.0005886790526697024,
      "loss": 2.3881,
      "step": 20212
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.5881142020225525,
      "learning_rate": 0.0005886779395217567,
      "loss": 2.2018,
      "step": 20213
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.7083337306976318,
      "learning_rate": 0.0005886768263201404,
      "loss": 2.4144,
      "step": 20214
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.634213387966156,
      "learning_rate": 0.0005886757130648536,
      "loss": 2.3395,
      "step": 20215
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6200717687606812,
      "learning_rate": 0.0005886745997558964,
      "loss": 2.5252,
      "step": 20216
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.5843782424926758,
      "learning_rate": 0.0005886734863932693,
      "loss": 2.3577,
      "step": 20217
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6237550377845764,
      "learning_rate": 0.0005886723729769722,
      "loss": 2.387,
      "step": 20218
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6748533844947815,
      "learning_rate": 0.0005886712595070055,
      "loss": 2.5431,
      "step": 20219
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.5868416428565979,
      "learning_rate": 0.0005886701459833695,
      "loss": 2.3017,
      "step": 20220
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.5987997055053711,
      "learning_rate": 0.000588669032406064,
      "loss": 2.196,
      "step": 20221
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6109647750854492,
      "learning_rate": 0.0005886679187750898,
      "loss": 2.367,
      "step": 20222
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.5955805778503418,
      "learning_rate": 0.0005886668050904466,
      "loss": 2.5277,
      "step": 20223
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.62064129114151,
      "learning_rate": 0.0005886656913521348,
      "loss": 2.3734,
      "step": 20224
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.5817649364471436,
      "learning_rate": 0.0005886645775601547,
      "loss": 2.3861,
      "step": 20225
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.8059753775596619,
      "learning_rate": 0.0005886634637145063,
      "loss": 2.4307,
      "step": 20226
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6119059324264526,
      "learning_rate": 0.0005886623498151899,
      "loss": 2.2448,
      "step": 20227
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6486070156097412,
      "learning_rate": 0.0005886612358622057,
      "loss": 2.3662,
      "step": 20228
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6663832664489746,
      "learning_rate": 0.000588660121855554,
      "loss": 2.2267,
      "step": 20229
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.7048481702804565,
      "learning_rate": 0.000588659007795235,
      "loss": 2.1549,
      "step": 20230
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.5692009329795837,
      "learning_rate": 0.0005886578936812487,
      "loss": 2.303,
      "step": 20231
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.7748960256576538,
      "learning_rate": 0.0005886567795135956,
      "loss": 2.6344,
      "step": 20232
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6020353436470032,
      "learning_rate": 0.0005886556652922756,
      "loss": 2.3857,
      "step": 20233
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6297488808631897,
      "learning_rate": 0.0005886545510172892,
      "loss": 2.4044,
      "step": 20234
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6230536103248596,
      "learning_rate": 0.0005886534366886364,
      "loss": 2.2207,
      "step": 20235
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6825688481330872,
      "learning_rate": 0.0005886523223063174,
      "loss": 2.3768,
      "step": 20236
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.631397008895874,
      "learning_rate": 0.0005886512078703326,
      "loss": 2.2173,
      "step": 20237
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.5882651805877686,
      "learning_rate": 0.000588650093380682,
      "loss": 2.298,
      "step": 20238
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6589441299438477,
      "learning_rate": 0.000588648978837366,
      "loss": 2.5708,
      "step": 20239
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.664917528629303,
      "learning_rate": 0.0005886478642403847,
      "loss": 2.4812,
      "step": 20240
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6236724853515625,
      "learning_rate": 0.0005886467495897383,
      "loss": 2.2478,
      "step": 20241
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.7013722062110901,
      "learning_rate": 0.0005886456348854269,
      "loss": 2.3028,
      "step": 20242
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6249365210533142,
      "learning_rate": 0.000588644520127451,
      "loss": 2.5969,
      "step": 20243
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6135771870613098,
      "learning_rate": 0.0005886434053158106,
      "loss": 2.4286,
      "step": 20244
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6499995589256287,
      "learning_rate": 0.0005886422904505059,
      "loss": 2.3002,
      "step": 20245
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6065585613250732,
      "learning_rate": 0.0005886411755315371,
      "loss": 2.4836,
      "step": 20246
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6481863856315613,
      "learning_rate": 0.0005886400605589045,
      "loss": 2.3527,
      "step": 20247
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6620792746543884,
      "learning_rate": 0.0005886389455326083,
      "loss": 2.4444,
      "step": 20248
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6668228507041931,
      "learning_rate": 0.0005886378304526486,
      "loss": 2.433,
      "step": 20249
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6156335473060608,
      "learning_rate": 0.0005886367153190258,
      "loss": 2.4527,
      "step": 20250
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6163827776908875,
      "learning_rate": 0.0005886356001317399,
      "loss": 2.2972,
      "step": 20251
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.65618497133255,
      "learning_rate": 0.0005886344848907912,
      "loss": 2.25,
      "step": 20252
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6593404412269592,
      "learning_rate": 0.0005886333695961799,
      "loss": 2.2732,
      "step": 20253
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6167004704475403,
      "learning_rate": 0.0005886322542479062,
      "loss": 2.153,
      "step": 20254
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6199048161506653,
      "learning_rate": 0.0005886311388459704,
      "loss": 2.4775,
      "step": 20255
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.5691092014312744,
      "learning_rate": 0.0005886300233903725,
      "loss": 2.2848,
      "step": 20256
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.639950692653656,
      "learning_rate": 0.0005886289078811128,
      "loss": 2.199,
      "step": 20257
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6468165516853333,
      "learning_rate": 0.0005886277923181916,
      "loss": 2.4144,
      "step": 20258
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6137051582336426,
      "learning_rate": 0.0005886266767016092,
      "loss": 2.4492,
      "step": 20259
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.8574895262718201,
      "learning_rate": 0.0005886255610313654,
      "loss": 2.4034,
      "step": 20260
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6795781254768372,
      "learning_rate": 0.0005886244453074608,
      "loss": 2.6348,
      "step": 20261
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6186901330947876,
      "learning_rate": 0.0005886233295298955,
      "loss": 2.1791,
      "step": 20262
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.0718414783477783,
      "learning_rate": 0.0005886222136986696,
      "loss": 2.3607,
      "step": 20263
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.5380247235298157,
      "learning_rate": 0.0005886210978137834,
      "loss": 2.3079,
      "step": 20264
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6151934862136841,
      "learning_rate": 0.0005886199818752371,
      "loss": 2.3751,
      "step": 20265
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.5654801726341248,
      "learning_rate": 0.0005886188658830309,
      "loss": 2.2251,
      "step": 20266
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.5953516364097595,
      "learning_rate": 0.0005886177498371651,
      "loss": 2.5384,
      "step": 20267
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6367629766464233,
      "learning_rate": 0.0005886166337376395,
      "loss": 2.5622,
      "step": 20268
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.61281818151474,
      "learning_rate": 0.0005886155175844549,
      "loss": 2.4525,
      "step": 20269
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.5910235047340393,
      "learning_rate": 0.0005886144013776112,
      "loss": 2.4214,
      "step": 20270
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6068271994590759,
      "learning_rate": 0.0005886132851171085,
      "loss": 2.4903,
      "step": 20271
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6707121729850769,
      "learning_rate": 0.0005886121688029473,
      "loss": 2.2562,
      "step": 20272
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6375088095664978,
      "learning_rate": 0.0005886110524351275,
      "loss": 2.2848,
      "step": 20273
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.5746967196464539,
      "learning_rate": 0.0005886099360136495,
      "loss": 2.2102,
      "step": 20274
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.7955329418182373,
      "learning_rate": 0.0005886088195385137,
      "loss": 2.4617,
      "step": 20275
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6358968019485474,
      "learning_rate": 0.0005886077030097198,
      "loss": 2.2427,
      "step": 20276
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.5738916993141174,
      "learning_rate": 0.0005886065864272683,
      "loss": 2.1563,
      "step": 20277
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6406446099281311,
      "learning_rate": 0.0005886054697911595,
      "loss": 2.4015,
      "step": 20278
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.7456488609313965,
      "learning_rate": 0.0005886043531013935,
      "loss": 2.4035,
      "step": 20279
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.5845844745635986,
      "learning_rate": 0.0005886032363579703,
      "loss": 2.3097,
      "step": 20280
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.7825804352760315,
      "learning_rate": 0.0005886021195608905,
      "loss": 2.3235,
      "step": 20281
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6450791358947754,
      "learning_rate": 0.0005886010027101542,
      "loss": 2.3979,
      "step": 20282
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6247047185897827,
      "learning_rate": 0.0005885998858057613,
      "loss": 2.259,
      "step": 20283
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.7143011689186096,
      "learning_rate": 0.0005885987688477122,
      "loss": 2.2152,
      "step": 20284
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6143670082092285,
      "learning_rate": 0.0005885976518360073,
      "loss": 2.3297,
      "step": 20285
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6766425371170044,
      "learning_rate": 0.0005885965347706466,
      "loss": 2.5479,
      "step": 20286
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.7052308320999146,
      "learning_rate": 0.0005885954176516304,
      "loss": 2.4579,
      "step": 20287
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.612359881401062,
      "learning_rate": 0.0005885943004789589,
      "loss": 2.2062,
      "step": 20288
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.7281279563903809,
      "learning_rate": 0.0005885931832526322,
      "loss": 2.3209,
      "step": 20289
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6118858456611633,
      "learning_rate": 0.0005885920659726506,
      "loss": 2.2817,
      "step": 20290
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6384198665618896,
      "learning_rate": 0.0005885909486390143,
      "loss": 2.2479,
      "step": 20291
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6069425344467163,
      "learning_rate": 0.0005885898312517234,
      "loss": 2.4873,
      "step": 20292
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.5814080238342285,
      "learning_rate": 0.0005885887138107783,
      "loss": 2.2175,
      "step": 20293
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6773502230644226,
      "learning_rate": 0.0005885875963161792,
      "loss": 2.3327,
      "step": 20294
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.582236647605896,
      "learning_rate": 0.000588586478767926,
      "loss": 2.3217,
      "step": 20295
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6164108514785767,
      "learning_rate": 0.0005885853611660193,
      "loss": 2.3141,
      "step": 20296
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.5779958367347717,
      "learning_rate": 0.000588584243510459,
      "loss": 2.4796,
      "step": 20297
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.5943319797515869,
      "learning_rate": 0.0005885831258012456,
      "loss": 2.3393,
      "step": 20298
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.5665205717086792,
      "learning_rate": 0.000588582008038379,
      "loss": 2.185,
      "step": 20299
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6310475468635559,
      "learning_rate": 0.0005885808902218596,
      "loss": 2.3353,
      "step": 20300
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.5704888105392456,
      "learning_rate": 0.0005885797723516876,
      "loss": 2.3859,
      "step": 20301
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6686408519744873,
      "learning_rate": 0.0005885786544278631,
      "loss": 2.5431,
      "step": 20302
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6643503904342651,
      "learning_rate": 0.0005885775364503865,
      "loss": 2.2775,
      "step": 20303
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6399805545806885,
      "learning_rate": 0.0005885764184192577,
      "loss": 2.5032,
      "step": 20304
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6133096218109131,
      "learning_rate": 0.0005885753003344773,
      "loss": 2.3075,
      "step": 20305
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6831080317497253,
      "learning_rate": 0.0005885741821960453,
      "loss": 2.4035,
      "step": 20306
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.5793873071670532,
      "learning_rate": 0.0005885730640039617,
      "loss": 2.4225,
      "step": 20307
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6172734498977661,
      "learning_rate": 0.0005885719457582271,
      "loss": 2.3122,
      "step": 20308
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6433480381965637,
      "learning_rate": 0.0005885708274588415,
      "loss": 2.2793,
      "step": 20309
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6161266565322876,
      "learning_rate": 0.000588569709105805,
      "loss": 2.205,
      "step": 20310
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6210894584655762,
      "learning_rate": 0.0005885685906991182,
      "loss": 2.3352,
      "step": 20311
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6507752537727356,
      "learning_rate": 0.000588567472238781,
      "loss": 2.1357,
      "step": 20312
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6574810147285461,
      "learning_rate": 0.0005885663537247935,
      "loss": 2.2638,
      "step": 20313
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.5910732746124268,
      "learning_rate": 0.0005885652351571561,
      "loss": 2.3532,
      "step": 20314
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.627364456653595,
      "learning_rate": 0.000588564116535869,
      "loss": 2.259,
      "step": 20315
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6196815967559814,
      "learning_rate": 0.0005885629978609324,
      "loss": 2.445,
      "step": 20316
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6504454016685486,
      "learning_rate": 0.0005885618791323465,
      "loss": 2.5157,
      "step": 20317
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6339383721351624,
      "learning_rate": 0.0005885607603501116,
      "loss": 2.4196,
      "step": 20318
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6801348328590393,
      "learning_rate": 0.0005885596415142276,
      "loss": 2.252,
      "step": 20319
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6350319981575012,
      "learning_rate": 0.000588558522624695,
      "loss": 2.1391,
      "step": 20320
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6733400225639343,
      "learning_rate": 0.000588557403681514,
      "loss": 2.5101,
      "step": 20321
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.5964173078536987,
      "learning_rate": 0.0005885562846846846,
      "loss": 2.4309,
      "step": 20322
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6663246154785156,
      "learning_rate": 0.0005885551656342072,
      "loss": 2.2369,
      "step": 20323
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.8147689700126648,
      "learning_rate": 0.000588554046530082,
      "loss": 2.3399,
      "step": 20324
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.613780677318573,
      "learning_rate": 0.0005885529273723092,
      "loss": 2.3159,
      "step": 20325
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6608061194419861,
      "learning_rate": 0.0005885518081608888,
      "loss": 2.2946,
      "step": 20326
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6545200943946838,
      "learning_rate": 0.0005885506888958212,
      "loss": 2.4088,
      "step": 20327
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.5745505690574646,
      "learning_rate": 0.0005885495695771067,
      "loss": 2.2763,
      "step": 20328
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.5715209245681763,
      "learning_rate": 0.0005885484502047453,
      "loss": 2.2551,
      "step": 20329
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6679174900054932,
      "learning_rate": 0.0005885473307787372,
      "loss": 2.2919,
      "step": 20330
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6103388071060181,
      "learning_rate": 0.000588546211299083,
      "loss": 2.1398,
      "step": 20331
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6828221082687378,
      "learning_rate": 0.0005885450917657824,
      "loss": 2.2016,
      "step": 20332
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6207649111747742,
      "learning_rate": 0.0005885439721788358,
      "loss": 2.261,
      "step": 20333
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6713354587554932,
      "learning_rate": 0.0005885428525382435,
      "loss": 2.2074,
      "step": 20334
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6009374856948853,
      "learning_rate": 0.0005885417328440057,
      "loss": 2.2174,
      "step": 20335
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6961429119110107,
      "learning_rate": 0.0005885406130961224,
      "loss": 2.2975,
      "step": 20336
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6542671322822571,
      "learning_rate": 0.000588539493294594,
      "loss": 2.3799,
      "step": 20337
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6078916192054749,
      "learning_rate": 0.0005885383734394208,
      "loss": 2.3397,
      "step": 20338
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.7145928740501404,
      "learning_rate": 0.0005885372535306026,
      "loss": 2.3013,
      "step": 20339
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.7128191590309143,
      "learning_rate": 0.00058853613356814,
      "loss": 2.4256,
      "step": 20340
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6213778257369995,
      "learning_rate": 0.0005885350135520332,
      "loss": 2.4282,
      "step": 20341
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.7384962439537048,
      "learning_rate": 0.0005885338934822821,
      "loss": 2.2144,
      "step": 20342
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6549769639968872,
      "learning_rate": 0.0005885327733588872,
      "loss": 2.4463,
      "step": 20343
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6198668479919434,
      "learning_rate": 0.0005885316531818487,
      "loss": 2.4599,
      "step": 20344
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6122260689735413,
      "learning_rate": 0.0005885305329511666,
      "loss": 2.3064,
      "step": 20345
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6826464533805847,
      "learning_rate": 0.0005885294126668413,
      "loss": 2.1412,
      "step": 20346
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.5862244367599487,
      "learning_rate": 0.0005885282923288729,
      "loss": 2.2779,
      "step": 20347
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.5903187990188599,
      "learning_rate": 0.0005885271719372617,
      "loss": 2.3529,
      "step": 20348
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6574226021766663,
      "learning_rate": 0.0005885260514920077,
      "loss": 2.3803,
      "step": 20349
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6075858473777771,
      "learning_rate": 0.0005885249309931115,
      "loss": 2.4191,
      "step": 20350
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6188310384750366,
      "learning_rate": 0.0005885238104405728,
      "loss": 2.2286,
      "step": 20351
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.5946212410926819,
      "learning_rate": 0.0005885226898343923,
      "loss": 2.3304,
      "step": 20352
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6085175275802612,
      "learning_rate": 0.0005885215691745699,
      "loss": 2.5517,
      "step": 20353
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6752578616142273,
      "learning_rate": 0.0005885204484611059,
      "loss": 2.3874,
      "step": 20354
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.5759969353675842,
      "learning_rate": 0.0005885193276940004,
      "loss": 2.3637,
      "step": 20355
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6440129280090332,
      "learning_rate": 0.0005885182068732539,
      "loss": 2.0658,
      "step": 20356
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6130216121673584,
      "learning_rate": 0.0005885170859988663,
      "loss": 2.4685,
      "step": 20357
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.5601508021354675,
      "learning_rate": 0.0005885159650708379,
      "loss": 2.3742,
      "step": 20358
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.5816053748130798,
      "learning_rate": 0.000588514844089169,
      "loss": 2.3654,
      "step": 20359
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.5846443772315979,
      "learning_rate": 0.0005885137230538597,
      "loss": 2.4057,
      "step": 20360
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6258159875869751,
      "learning_rate": 0.0005885126019649102,
      "loss": 2.4237,
      "step": 20361
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6152052283287048,
      "learning_rate": 0.0005885114808223209,
      "loss": 2.727,
      "step": 20362
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6028663516044617,
      "learning_rate": 0.0005885103596260918,
      "loss": 2.3523,
      "step": 20363
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.594824492931366,
      "learning_rate": 0.0005885092383762232,
      "loss": 2.3269,
      "step": 20364
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.5748500823974609,
      "learning_rate": 0.0005885081170727153,
      "loss": 2.1395,
      "step": 20365
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.5720879435539246,
      "learning_rate": 0.0005885069957155683,
      "loss": 2.3157,
      "step": 20366
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.5749507546424866,
      "learning_rate": 0.0005885058743047823,
      "loss": 2.2861,
      "step": 20367
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6195714473724365,
      "learning_rate": 0.0005885047528403577,
      "loss": 2.5488,
      "step": 20368
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.578795850276947,
      "learning_rate": 0.0005885036313222946,
      "loss": 2.3222,
      "step": 20369
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6088318228721619,
      "learning_rate": 0.0005885025097505933,
      "loss": 2.4546,
      "step": 20370
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.5544840693473816,
      "learning_rate": 0.0005885013881252538,
      "loss": 2.5839,
      "step": 20371
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6641742587089539,
      "learning_rate": 0.0005885002664462765,
      "loss": 2.1807,
      "step": 20372
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6902639865875244,
      "learning_rate": 0.0005884991447136616,
      "loss": 2.408,
      "step": 20373
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6513479351997375,
      "learning_rate": 0.0005884980229274093,
      "loss": 2.3401,
      "step": 20374
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.5776448845863342,
      "learning_rate": 0.0005884969010875197,
      "loss": 2.3002,
      "step": 20375
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6215602159500122,
      "learning_rate": 0.000588495779193993,
      "loss": 2.4405,
      "step": 20376
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6550956964492798,
      "learning_rate": 0.0005884946572468296,
      "loss": 2.6039,
      "step": 20377
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.611765444278717,
      "learning_rate": 0.0005884935352460296,
      "loss": 2.1721,
      "step": 20378
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.5999203324317932,
      "learning_rate": 0.0005884924131915932,
      "loss": 2.4135,
      "step": 20379
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6316702365875244,
      "learning_rate": 0.0005884912910835206,
      "loss": 2.4326,
      "step": 20380
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6182013154029846,
      "learning_rate": 0.000588490168921812,
      "loss": 2.27,
      "step": 20381
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6297332048416138,
      "learning_rate": 0.0005884890467064677,
      "loss": 2.3247,
      "step": 20382
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6125136017799377,
      "learning_rate": 0.0005884879244374878,
      "loss": 2.3631,
      "step": 20383
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.7158964276313782,
      "learning_rate": 0.0005884868021148724,
      "loss": 2.5543,
      "step": 20384
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6836689114570618,
      "learning_rate": 0.0005884856797386221,
      "loss": 2.3397,
      "step": 20385
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.633227527141571,
      "learning_rate": 0.0005884845573087368,
      "loss": 2.4459,
      "step": 20386
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6553515195846558,
      "learning_rate": 0.0005884834348252166,
      "loss": 2.2868,
      "step": 20387
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6201923489570618,
      "learning_rate": 0.0005884823122880621,
      "loss": 2.1062,
      "step": 20388
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6213303208351135,
      "learning_rate": 0.0005884811896972733,
      "loss": 2.2698,
      "step": 20389
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6817079186439514,
      "learning_rate": 0.0005884800670528502,
      "loss": 2.4274,
      "step": 20390
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6332652568817139,
      "learning_rate": 0.0005884789443547934,
      "loss": 2.3924,
      "step": 20391
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.5760279297828674,
      "learning_rate": 0.0005884778216031028,
      "loss": 2.5068,
      "step": 20392
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6595928072929382,
      "learning_rate": 0.0005884766987977789,
      "loss": 2.1692,
      "step": 20393
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.60826575756073,
      "learning_rate": 0.0005884755759388216,
      "loss": 2.4588,
      "step": 20394
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.5683049559593201,
      "learning_rate": 0.0005884744530262312,
      "loss": 2.3097,
      "step": 20395
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6102636456489563,
      "learning_rate": 0.0005884733300600079,
      "loss": 2.5915,
      "step": 20396
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.8499834537506104,
      "learning_rate": 0.0005884722070401522,
      "loss": 2.3427,
      "step": 20397
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6835817098617554,
      "learning_rate": 0.0005884710839666639,
      "loss": 2.2253,
      "step": 20398
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.8520731925964355,
      "learning_rate": 0.0005884699608395434,
      "loss": 2.3829,
      "step": 20399
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6311661601066589,
      "learning_rate": 0.0005884688376587909,
      "loss": 2.4453,
      "step": 20400
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6654883027076721,
      "learning_rate": 0.0005884677144244066,
      "loss": 2.7631,
      "step": 20401
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.7386816740036011,
      "learning_rate": 0.0005884665911363908,
      "loss": 2.1985,
      "step": 20402
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6074990034103394,
      "learning_rate": 0.0005884654677947435,
      "loss": 2.534,
      "step": 20403
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.7507616877555847,
      "learning_rate": 0.000588464344399465,
      "loss": 2.1303,
      "step": 20404
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.5578970313072205,
      "learning_rate": 0.0005884632209505556,
      "loss": 2.4065,
      "step": 20405
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.7811433672904968,
      "learning_rate": 0.0005884620974480155,
      "loss": 2.1325,
      "step": 20406
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6921496987342834,
      "learning_rate": 0.0005884609738918448,
      "loss": 2.408,
      "step": 20407
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6165916323661804,
      "learning_rate": 0.0005884598502820437,
      "loss": 2.2833,
      "step": 20408
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6586714386940002,
      "learning_rate": 0.0005884587266186125,
      "loss": 2.4265,
      "step": 20409
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6652794480323792,
      "learning_rate": 0.0005884576029015514,
      "loss": 2.2069,
      "step": 20410
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6177204847335815,
      "learning_rate": 0.0005884564791308605,
      "loss": 2.5866,
      "step": 20411
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6044119596481323,
      "learning_rate": 0.0005884553553065403,
      "loss": 2.2625,
      "step": 20412
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6098878979682922,
      "learning_rate": 0.0005884542314285906,
      "loss": 2.5044,
      "step": 20413
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.5779340267181396,
      "learning_rate": 0.0005884531074970119,
      "loss": 2.0837,
      "step": 20414
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.7209612131118774,
      "learning_rate": 0.0005884519835118044,
      "loss": 2.3142,
      "step": 20415
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.7998355627059937,
      "learning_rate": 0.0005884508594729681,
      "loss": 2.3642,
      "step": 20416
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.5901052355766296,
      "learning_rate": 0.0005884497353805033,
      "loss": 2.4946,
      "step": 20417
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6878056526184082,
      "learning_rate": 0.0005884486112344104,
      "loss": 2.4995,
      "step": 20418
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.7384702563285828,
      "learning_rate": 0.0005884474870346894,
      "loss": 2.3838,
      "step": 20419
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6273283958435059,
      "learning_rate": 0.0005884463627813405,
      "loss": 1.9971,
      "step": 20420
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6604796648025513,
      "learning_rate": 0.000588445238474364,
      "loss": 2.4786,
      "step": 20421
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6637629866600037,
      "learning_rate": 0.0005884441141137602,
      "loss": 2.2132,
      "step": 20422
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6651859879493713,
      "learning_rate": 0.0005884429896995291,
      "loss": 2.2675,
      "step": 20423
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6271117329597473,
      "learning_rate": 0.000588441865231671,
      "loss": 2.5029,
      "step": 20424
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6305059194564819,
      "learning_rate": 0.0005884407407101862,
      "loss": 2.3383,
      "step": 20425
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6377590298652649,
      "learning_rate": 0.0005884396161350747,
      "loss": 2.4374,
      "step": 20426
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.5816718935966492,
      "learning_rate": 0.0005884384915063369,
      "loss": 2.3263,
      "step": 20427
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6139891743659973,
      "learning_rate": 0.0005884373668239729,
      "loss": 2.1681,
      "step": 20428
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6084825396537781,
      "learning_rate": 0.000588436242087983,
      "loss": 2.2495,
      "step": 20429
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6370331645011902,
      "learning_rate": 0.0005884351172983673,
      "loss": 2.4249,
      "step": 20430
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6161773204803467,
      "learning_rate": 0.0005884339924551261,
      "loss": 2.2954,
      "step": 20431
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6541309952735901,
      "learning_rate": 0.0005884328675582596,
      "loss": 2.3059,
      "step": 20432
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6134081482887268,
      "learning_rate": 0.0005884317426077679,
      "loss": 2.3213,
      "step": 20433
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6533977389335632,
      "learning_rate": 0.0005884306176036514,
      "loss": 2.3967,
      "step": 20434
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6321741938591003,
      "learning_rate": 0.0005884294925459101,
      "loss": 2.4081,
      "step": 20435
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6165477633476257,
      "learning_rate": 0.0005884283674345444,
      "loss": 2.2515,
      "step": 20436
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6602423191070557,
      "learning_rate": 0.0005884272422695543,
      "loss": 2.4624,
      "step": 20437
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6577099561691284,
      "learning_rate": 0.0005884261170509403,
      "loss": 2.3023,
      "step": 20438
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.5828048586845398,
      "learning_rate": 0.0005884249917787024,
      "loss": 2.3556,
      "step": 20439
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6361441016197205,
      "learning_rate": 0.0005884238664528408,
      "loss": 2.3352,
      "step": 20440
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6804518103599548,
      "learning_rate": 0.0005884227410733558,
      "loss": 2.4126,
      "step": 20441
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6182104349136353,
      "learning_rate": 0.0005884216156402474,
      "loss": 2.5181,
      "step": 20442
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6398042440414429,
      "learning_rate": 0.0005884204901535162,
      "loss": 2.2243,
      "step": 20443
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.5895371437072754,
      "learning_rate": 0.0005884193646131621,
      "loss": 2.3222,
      "step": 20444
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.5870128273963928,
      "learning_rate": 0.0005884182390191854,
      "loss": 2.3988,
      "step": 20445
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6154887676239014,
      "learning_rate": 0.0005884171133715864,
      "loss": 2.0729,
      "step": 20446
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6148134469985962,
      "learning_rate": 0.000588415987670365,
      "loss": 2.2966,
      "step": 20447
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6965844631195068,
      "learning_rate": 0.0005884148619155217,
      "loss": 2.4291,
      "step": 20448
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6085178256034851,
      "learning_rate": 0.0005884137361070568,
      "loss": 2.248,
      "step": 20449
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.654370903968811,
      "learning_rate": 0.0005884126102449701,
      "loss": 2.3751,
      "step": 20450
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.5745508670806885,
      "learning_rate": 0.0005884114843292623,
      "loss": 2.1326,
      "step": 20451
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6134896278381348,
      "learning_rate": 0.0005884103583599331,
      "loss": 2.4518,
      "step": 20452
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6344102621078491,
      "learning_rate": 0.0005884092323369831,
      "loss": 2.6103,
      "step": 20453
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6625958681106567,
      "learning_rate": 0.0005884081062604124,
      "loss": 2.3012,
      "step": 20454
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.652216911315918,
      "learning_rate": 0.0005884069801302212,
      "loss": 2.3289,
      "step": 20455
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6286156177520752,
      "learning_rate": 0.0005884058539464096,
      "loss": 2.4885,
      "step": 20456
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6526229977607727,
      "learning_rate": 0.0005884047277089781,
      "loss": 2.3495,
      "step": 20457
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6164197325706482,
      "learning_rate": 0.0005884036014179266,
      "loss": 2.4851,
      "step": 20458
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6200069189071655,
      "learning_rate": 0.0005884024750732553,
      "loss": 2.21,
      "step": 20459
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6282731294631958,
      "learning_rate": 0.0005884013486749647,
      "loss": 2.3535,
      "step": 20460
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6214876770973206,
      "learning_rate": 0.0005884002222230548,
      "loss": 2.5418,
      "step": 20461
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6208205819129944,
      "learning_rate": 0.0005883990957175259,
      "loss": 2.281,
      "step": 20462
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.5747798085212708,
      "learning_rate": 0.0005883979691583781,
      "loss": 2.409,
      "step": 20463
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.5641356706619263,
      "learning_rate": 0.0005883968425456117,
      "loss": 2.4504,
      "step": 20464
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.5791234374046326,
      "learning_rate": 0.0005883957158792268,
      "loss": 2.4129,
      "step": 20465
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6913052201271057,
      "learning_rate": 0.0005883945891592238,
      "loss": 2.4427,
      "step": 20466
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6847884654998779,
      "learning_rate": 0.0005883934623856027,
      "loss": 2.2362,
      "step": 20467
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.5912488698959351,
      "learning_rate": 0.000588392335558364,
      "loss": 2.4511,
      "step": 20468
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.694722056388855,
      "learning_rate": 0.0005883912086775075,
      "loss": 2.201,
      "step": 20469
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6289738416671753,
      "learning_rate": 0.0005883900817430337,
      "loss": 2.3472,
      "step": 20470
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6901600360870361,
      "learning_rate": 0.0005883889547549429,
      "loss": 2.3911,
      "step": 20471
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.7328317165374756,
      "learning_rate": 0.000588387827713235,
      "loss": 2.3714,
      "step": 20472
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.7325197458267212,
      "learning_rate": 0.0005883867006179104,
      "loss": 2.3226,
      "step": 20473
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6033924221992493,
      "learning_rate": 0.0005883855734689692,
      "loss": 2.3324,
      "step": 20474
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6439830660820007,
      "learning_rate": 0.0005883844462664117,
      "loss": 2.3262,
      "step": 20475
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.7012233734130859,
      "learning_rate": 0.0005883833190102381,
      "loss": 2.3926,
      "step": 20476
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.573108971118927,
      "learning_rate": 0.0005883821917004486,
      "loss": 2.3917,
      "step": 20477
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.855830729007721,
      "learning_rate": 0.0005883810643370434,
      "loss": 2.4926,
      "step": 20478
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6670471429824829,
      "learning_rate": 0.0005883799369200228,
      "loss": 2.4057,
      "step": 20479
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.620435357093811,
      "learning_rate": 0.0005883788094493868,
      "loss": 2.4137,
      "step": 20480
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.7290065288543701,
      "learning_rate": 0.0005883776819251359,
      "loss": 2.3681,
      "step": 20481
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.5961008667945862,
      "learning_rate": 0.00058837655434727,
      "loss": 2.4689,
      "step": 20482
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.7565209865570068,
      "learning_rate": 0.0005883754267157895,
      "loss": 2.3638,
      "step": 20483
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6163576245307922,
      "learning_rate": 0.0005883742990306946,
      "loss": 2.4613,
      "step": 20484
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6289559602737427,
      "learning_rate": 0.0005883731712919854,
      "loss": 2.3348,
      "step": 20485
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6471918821334839,
      "learning_rate": 0.0005883720434996623,
      "loss": 2.4426,
      "step": 20486
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6449438333511353,
      "learning_rate": 0.0005883709156537252,
      "loss": 2.3832,
      "step": 20487
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6747992038726807,
      "learning_rate": 0.0005883697877541747,
      "loss": 2.4126,
      "step": 20488
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.7276688814163208,
      "learning_rate": 0.0005883686598010107,
      "loss": 2.5879,
      "step": 20489
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.586736261844635,
      "learning_rate": 0.0005883675317942336,
      "loss": 2.349,
      "step": 20490
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.0675750970840454,
      "learning_rate": 0.0005883664037338433,
      "loss": 2.4141,
      "step": 20491
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6338815689086914,
      "learning_rate": 0.0005883652756198405,
      "loss": 2.0376,
      "step": 20492
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6498741507530212,
      "learning_rate": 0.0005883641474522252,
      "loss": 2.3744,
      "step": 20493
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.5965084433555603,
      "learning_rate": 0.0005883630192309974,
      "loss": 2.3212,
      "step": 20494
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6757400631904602,
      "learning_rate": 0.0005883618909561574,
      "loss": 2.492,
      "step": 20495
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.627644419670105,
      "learning_rate": 0.0005883607626277057,
      "loss": 2.4484,
      "step": 20496
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6147551536560059,
      "learning_rate": 0.0005883596342456421,
      "loss": 2.3253,
      "step": 20497
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6945663094520569,
      "learning_rate": 0.0005883585058099671,
      "loss": 2.5964,
      "step": 20498
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.635113000869751,
      "learning_rate": 0.0005883573773206808,
      "loss": 2.3942,
      "step": 20499
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.5614943504333496,
      "learning_rate": 0.0005883562487777835,
      "loss": 2.3663,
      "step": 20500
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6273114681243896,
      "learning_rate": 0.0005883551201812752,
      "loss": 2.6265,
      "step": 20501
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6072075366973877,
      "learning_rate": 0.0005883539915311562,
      "loss": 2.3051,
      "step": 20502
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6484718322753906,
      "learning_rate": 0.0005883528628274269,
      "loss": 2.4093,
      "step": 20503
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6165904998779297,
      "learning_rate": 0.0005883517340700874,
      "loss": 2.4058,
      "step": 20504
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6230954527854919,
      "learning_rate": 0.0005883506052591377,
      "loss": 2.3772,
      "step": 20505
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.7380741238594055,
      "learning_rate": 0.0005883494763945782,
      "loss": 2.4004,
      "step": 20506
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6488640308380127,
      "learning_rate": 0.0005883483474764092,
      "loss": 2.4283,
      "step": 20507
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.640379786491394,
      "learning_rate": 0.0005883472185046308,
      "loss": 2.2707,
      "step": 20508
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6069270372390747,
      "learning_rate": 0.000588346089479243,
      "loss": 2.6268,
      "step": 20509
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6908114552497864,
      "learning_rate": 0.0005883449604002464,
      "loss": 2.3131,
      "step": 20510
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.5769740343093872,
      "learning_rate": 0.0005883438312676409,
      "loss": 2.3028,
      "step": 20511
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.5994858741760254,
      "learning_rate": 0.000588342702081427,
      "loss": 2.2679,
      "step": 20512
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.619086503982544,
      "learning_rate": 0.0005883415728416047,
      "loss": 2.3875,
      "step": 20513
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.5877184271812439,
      "learning_rate": 0.0005883404435481742,
      "loss": 2.3028,
      "step": 20514
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6362447142601013,
      "learning_rate": 0.0005883393142011358,
      "loss": 2.4467,
      "step": 20515
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.7176388502120972,
      "learning_rate": 0.0005883381848004897,
      "loss": 2.2619,
      "step": 20516
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6146478056907654,
      "learning_rate": 0.0005883370553462361,
      "loss": 2.172,
      "step": 20517
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6022102236747742,
      "learning_rate": 0.0005883359258383752,
      "loss": 2.4094,
      "step": 20518
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6052389740943909,
      "learning_rate": 0.0005883347962769072,
      "loss": 2.2108,
      "step": 20519
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.7211845517158508,
      "learning_rate": 0.0005883336666618323,
      "loss": 2.2072,
      "step": 20520
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.635257363319397,
      "learning_rate": 0.0005883325369931507,
      "loss": 2.3864,
      "step": 20521
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6404235363006592,
      "learning_rate": 0.0005883314072708626,
      "loss": 2.3823,
      "step": 20522
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.7163237929344177,
      "learning_rate": 0.0005883302774949684,
      "loss": 2.3856,
      "step": 20523
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.7027817368507385,
      "learning_rate": 0.0005883291476654681,
      "loss": 2.3202,
      "step": 20524
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6597670912742615,
      "learning_rate": 0.0005883280177823619,
      "loss": 2.4163,
      "step": 20525
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.689234733581543,
      "learning_rate": 0.0005883268878456501,
      "loss": 2.3365,
      "step": 20526
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.7092199325561523,
      "learning_rate": 0.000588325757855333,
      "loss": 2.3529,
      "step": 20527
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6396147608757019,
      "learning_rate": 0.0005883246278114105,
      "loss": 2.6085,
      "step": 20528
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6664196848869324,
      "learning_rate": 0.0005883234977138831,
      "loss": 2.4981,
      "step": 20529
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6096158027648926,
      "learning_rate": 0.0005883223675627509,
      "loss": 2.5525,
      "step": 20530
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6749395728111267,
      "learning_rate": 0.0005883212373580143,
      "loss": 2.4841,
      "step": 20531
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6056616306304932,
      "learning_rate": 0.0005883201070996731,
      "loss": 2.4678,
      "step": 20532
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6315767765045166,
      "learning_rate": 0.000588318976787728,
      "loss": 2.3794,
      "step": 20533
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6400429606437683,
      "learning_rate": 0.0005883178464221788,
      "loss": 2.2908,
      "step": 20534
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.590836226940155,
      "learning_rate": 0.0005883167160030258,
      "loss": 2.4655,
      "step": 20535
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6184918284416199,
      "learning_rate": 0.0005883155855302693,
      "loss": 2.4242,
      "step": 20536
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.5421755313873291,
      "learning_rate": 0.0005883144550039096,
      "loss": 2.156,
      "step": 20537
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.703272819519043,
      "learning_rate": 0.0005883133244239467,
      "loss": 2.5315,
      "step": 20538
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.5933457612991333,
      "learning_rate": 0.0005883121937903811,
      "loss": 2.4313,
      "step": 20539
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6275108456611633,
      "learning_rate": 0.0005883110631032126,
      "loss": 2.5816,
      "step": 20540
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6937834620475769,
      "learning_rate": 0.0005883099323624418,
      "loss": 2.7253,
      "step": 20541
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6086611747741699,
      "learning_rate": 0.0005883088015680687,
      "loss": 2.3456,
      "step": 20542
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6341451406478882,
      "learning_rate": 0.0005883076707200934,
      "loss": 2.6082,
      "step": 20543
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6221868395805359,
      "learning_rate": 0.0005883065398185164,
      "loss": 2.1532,
      "step": 20544
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6791969537734985,
      "learning_rate": 0.0005883054088633378,
      "loss": 2.3187,
      "step": 20545
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6172279119491577,
      "learning_rate": 0.0005883042778545577,
      "loss": 2.3954,
      "step": 20546
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6085194945335388,
      "learning_rate": 0.0005883031467921764,
      "loss": 2.3515,
      "step": 20547
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6076472401618958,
      "learning_rate": 0.0005883020156761941,
      "loss": 2.5659,
      "step": 20548
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6461965441703796,
      "learning_rate": 0.000588300884506611,
      "loss": 2.3434,
      "step": 20549
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6449726819992065,
      "learning_rate": 0.0005882997532834274,
      "loss": 2.3222,
      "step": 20550
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.646609902381897,
      "learning_rate": 0.0005882986220066435,
      "loss": 2.5246,
      "step": 20551
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.7626405954360962,
      "learning_rate": 0.0005882974906762593,
      "loss": 2.3559,
      "step": 20552
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6896569728851318,
      "learning_rate": 0.0005882963592922752,
      "loss": 2.5142,
      "step": 20553
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6386638283729553,
      "learning_rate": 0.0005882952278546914,
      "loss": 2.1946,
      "step": 20554
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.706242024898529,
      "learning_rate": 0.0005882940963635079,
      "loss": 2.4019,
      "step": 20555
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.613070011138916,
      "learning_rate": 0.0005882929648187253,
      "loss": 1.963,
      "step": 20556
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.5619387030601501,
      "learning_rate": 0.0005882918332203435,
      "loss": 2.1427,
      "step": 20557
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.5596954822540283,
      "learning_rate": 0.0005882907015683629,
      "loss": 2.2628,
      "step": 20558
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6196125745773315,
      "learning_rate": 0.0005882895698627835,
      "loss": 2.4265,
      "step": 20559
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6754372119903564,
      "learning_rate": 0.0005882884381036057,
      "loss": 2.5666,
      "step": 20560
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.5754923820495605,
      "learning_rate": 0.0005882873062908296,
      "loss": 2.1142,
      "step": 20561
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6383069157600403,
      "learning_rate": 0.0005882861744244554,
      "loss": 2.2691,
      "step": 20562
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.7063806653022766,
      "learning_rate": 0.0005882850425044834,
      "loss": 2.1102,
      "step": 20563
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.780191957950592,
      "learning_rate": 0.0005882839105309138,
      "loss": 2.2488,
      "step": 20564
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.5945978760719299,
      "learning_rate": 0.0005882827785037467,
      "loss": 2.2938,
      "step": 20565
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.7226786613464355,
      "learning_rate": 0.0005882816464229825,
      "loss": 2.4197,
      "step": 20566
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6944779753684998,
      "learning_rate": 0.0005882805142886212,
      "loss": 2.4386,
      "step": 20567
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6575864553451538,
      "learning_rate": 0.0005882793821006631,
      "loss": 2.3408,
      "step": 20568
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.9642167687416077,
      "learning_rate": 0.0005882782498591085,
      "loss": 2.0862,
      "step": 20569
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.9047536253929138,
      "learning_rate": 0.0005882771175639575,
      "loss": 2.3371,
      "step": 20570
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.8601881265640259,
      "learning_rate": 0.0005882759852152104,
      "loss": 2.3944,
      "step": 20571
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.7477226853370667,
      "learning_rate": 0.0005882748528128673,
      "loss": 2.4018,
      "step": 20572
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.9082930684089661,
      "learning_rate": 0.0005882737203569285,
      "loss": 2.3061,
      "step": 20573
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.5702605843544006,
      "learning_rate": 0.0005882725878473941,
      "loss": 2.4979,
      "step": 20574
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.7395544052124023,
      "learning_rate": 0.0005882714552842644,
      "loss": 2.3021,
      "step": 20575
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.0480965375900269,
      "learning_rate": 0.0005882703226675396,
      "loss": 2.5657,
      "step": 20576
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6575866341590881,
      "learning_rate": 0.00058826918999722,
      "loss": 2.4782,
      "step": 20577
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6809982061386108,
      "learning_rate": 0.0005882680572733056,
      "loss": 2.154,
      "step": 20578
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.7786884307861328,
      "learning_rate": 0.0005882669244957968,
      "loss": 2.5457,
      "step": 20579
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.7450222373008728,
      "learning_rate": 0.0005882657916646938,
      "loss": 2.3107,
      "step": 20580
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6233415603637695,
      "learning_rate": 0.0005882646587799966,
      "loss": 2.2791,
      "step": 20581
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.7527179718017578,
      "learning_rate": 0.0005882635258417056,
      "loss": 2.3201,
      "step": 20582
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.9282814860343933,
      "learning_rate": 0.000588262392849821,
      "loss": 2.3792,
      "step": 20583
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6706283688545227,
      "learning_rate": 0.0005882612598043429,
      "loss": 2.3537,
      "step": 20584
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.7289859056472778,
      "learning_rate": 0.0005882601267052717,
      "loss": 2.4395,
      "step": 20585
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.908869743347168,
      "learning_rate": 0.0005882589935526074,
      "loss": 2.4253,
      "step": 20586
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.592323362827301,
      "learning_rate": 0.0005882578603463504,
      "loss": 2.3237,
      "step": 20587
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6191153526306152,
      "learning_rate": 0.0005882567270865007,
      "loss": 2.3888,
      "step": 20588
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.7352402806282043,
      "learning_rate": 0.0005882555937730588,
      "loss": 2.3238,
      "step": 20589
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.7861933708190918,
      "learning_rate": 0.0005882544604060247,
      "loss": 2.3358,
      "step": 20590
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6185815930366516,
      "learning_rate": 0.0005882533269853987,
      "loss": 2.3546,
      "step": 20591
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.588422954082489,
      "learning_rate": 0.0005882521935111808,
      "loss": 2.1641,
      "step": 20592
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.905633807182312,
      "learning_rate": 0.0005882510599833715,
      "loss": 2.1316,
      "step": 20593
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6335996985435486,
      "learning_rate": 0.0005882499264019708,
      "loss": 2.2894,
      "step": 20594
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6528635025024414,
      "learning_rate": 0.000588248792766979,
      "loss": 2.3637,
      "step": 20595
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.8352476358413696,
      "learning_rate": 0.0005882476590783963,
      "loss": 2.5521,
      "step": 20596
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6211544275283813,
      "learning_rate": 0.000588246525336223,
      "loss": 2.2531,
      "step": 20597
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.610597550868988,
      "learning_rate": 0.0005882453915404592,
      "loss": 2.1436,
      "step": 20598
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6711626648902893,
      "learning_rate": 0.0005882442576911052,
      "loss": 2.3981,
      "step": 20599
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6466303467750549,
      "learning_rate": 0.0005882431237881612,
      "loss": 2.284,
      "step": 20600
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6032947301864624,
      "learning_rate": 0.0005882419898316271,
      "loss": 2.2303,
      "step": 20601
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6511006355285645,
      "learning_rate": 0.0005882408558215036,
      "loss": 2.2598,
      "step": 20602
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6498374938964844,
      "learning_rate": 0.0005882397217577905,
      "loss": 2.5164,
      "step": 20603
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6817505955696106,
      "learning_rate": 0.0005882385876404884,
      "loss": 2.5479,
      "step": 20604
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6744042634963989,
      "learning_rate": 0.0005882374534695972,
      "loss": 2.4357,
      "step": 20605
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6420680284500122,
      "learning_rate": 0.0005882363192451172,
      "loss": 2.1567,
      "step": 20606
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.7641158103942871,
      "learning_rate": 0.0005882351849670487,
      "loss": 2.2904,
      "step": 20607
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.5984863042831421,
      "learning_rate": 0.0005882340506353918,
      "loss": 2.542,
      "step": 20608
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6399049162864685,
      "learning_rate": 0.0005882329162501468,
      "loss": 2.4821,
      "step": 20609
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.7388916611671448,
      "learning_rate": 0.0005882317818113138,
      "loss": 2.4667,
      "step": 20610
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6391964554786682,
      "learning_rate": 0.000588230647318893,
      "loss": 2.3655,
      "step": 20611
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6646338701248169,
      "learning_rate": 0.0005882295127728848,
      "loss": 2.452,
      "step": 20612
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.7275276780128479,
      "learning_rate": 0.0005882283781732892,
      "loss": 2.3055,
      "step": 20613
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6034241914749146,
      "learning_rate": 0.0005882272435201066,
      "loss": 2.1548,
      "step": 20614
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.599281370639801,
      "learning_rate": 0.0005882261088133371,
      "loss": 2.6765,
      "step": 20615
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6169976592063904,
      "learning_rate": 0.0005882249740529809,
      "loss": 2.6307,
      "step": 20616
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.7039421200752258,
      "learning_rate": 0.0005882238392390382,
      "loss": 2.4102,
      "step": 20617
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6261467337608337,
      "learning_rate": 0.0005882227043715093,
      "loss": 2.5906,
      "step": 20618
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6134860515594482,
      "learning_rate": 0.0005882215694503943,
      "loss": 2.3103,
      "step": 20619
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6167423725128174,
      "learning_rate": 0.0005882204344756935,
      "loss": 2.0779,
      "step": 20620
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6663835048675537,
      "learning_rate": 0.0005882192994474072,
      "loss": 2.4735,
      "step": 20621
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6021235585212708,
      "learning_rate": 0.0005882181643655352,
      "loss": 2.4953,
      "step": 20622
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.656000554561615,
      "learning_rate": 0.0005882170292300782,
      "loss": 2.1896,
      "step": 20623
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.7222245931625366,
      "learning_rate": 0.0005882158940410363,
      "loss": 2.3565,
      "step": 20624
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.658327043056488,
      "learning_rate": 0.0005882147587984094,
      "loss": 2.6248,
      "step": 20625
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.587208092212677,
      "learning_rate": 0.0005882136235021981,
      "loss": 2.3956,
      "step": 20626
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.5932483077049255,
      "learning_rate": 0.0005882124881524025,
      "loss": 2.5687,
      "step": 20627
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6475301384925842,
      "learning_rate": 0.0005882113527490225,
      "loss": 2.5499,
      "step": 20628
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6342837810516357,
      "learning_rate": 0.0005882102172920588,
      "loss": 2.1636,
      "step": 20629
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.5911768078804016,
      "learning_rate": 0.0005882090817815114,
      "loss": 2.3285,
      "step": 20630
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6114684343338013,
      "learning_rate": 0.0005882079462173804,
      "loss": 2.254,
      "step": 20631
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6217537522315979,
      "learning_rate": 0.000588206810599666,
      "loss": 2.4899,
      "step": 20632
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.5846412181854248,
      "learning_rate": 0.0005882056749283687,
      "loss": 2.4321,
      "step": 20633
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6267091035842896,
      "learning_rate": 0.0005882045392034884,
      "loss": 2.1973,
      "step": 20634
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6245474219322205,
      "learning_rate": 0.0005882034034250254,
      "loss": 2.3581,
      "step": 20635
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6114814877510071,
      "learning_rate": 0.0005882022675929801,
      "loss": 2.2062,
      "step": 20636
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6620431542396545,
      "learning_rate": 0.0005882011317073526,
      "loss": 2.2242,
      "step": 20637
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.5695250630378723,
      "learning_rate": 0.0005881999957681428,
      "loss": 2.2128,
      "step": 20638
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.642511248588562,
      "learning_rate": 0.0005881988597753513,
      "loss": 2.3255,
      "step": 20639
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6438935995101929,
      "learning_rate": 0.0005881977237289783,
      "loss": 2.4524,
      "step": 20640
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6446663737297058,
      "learning_rate": 0.0005881965876290238,
      "loss": 2.25,
      "step": 20641
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.5821854472160339,
      "learning_rate": 0.0005881954514754882,
      "loss": 2.3686,
      "step": 20642
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6859320998191833,
      "learning_rate": 0.0005881943152683715,
      "loss": 2.3952,
      "step": 20643
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6213756799697876,
      "learning_rate": 0.0005881931790076741,
      "loss": 2.482,
      "step": 20644
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.5740644931793213,
      "learning_rate": 0.0005881920426933962,
      "loss": 2.1908,
      "step": 20645
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6153537034988403,
      "learning_rate": 0.0005881909063255379,
      "loss": 2.1549,
      "step": 20646
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6771345734596252,
      "learning_rate": 0.0005881897699040995,
      "loss": 2.23,
      "step": 20647
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6342718601226807,
      "learning_rate": 0.0005881886334290811,
      "loss": 2.226,
      "step": 20648
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6603760719299316,
      "learning_rate": 0.0005881874969004831,
      "loss": 2.4067,
      "step": 20649
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6519920229911804,
      "learning_rate": 0.0005881863603183057,
      "loss": 2.3451,
      "step": 20650
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6527299880981445,
      "learning_rate": 0.0005881852236825488,
      "loss": 2.246,
      "step": 20651
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6449800133705139,
      "learning_rate": 0.0005881840869932128,
      "loss": 2.3611,
      "step": 20652
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6421283483505249,
      "learning_rate": 0.0005881829502502981,
      "loss": 2.2407,
      "step": 20653
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6488471627235413,
      "learning_rate": 0.0005881818134538048,
      "loss": 2.4933,
      "step": 20654
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6484007835388184,
      "learning_rate": 0.0005881806766037329,
      "loss": 2.3805,
      "step": 20655
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6466914415359497,
      "learning_rate": 0.0005881795397000828,
      "loss": 2.5245,
      "step": 20656
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.646518886089325,
      "learning_rate": 0.0005881784027428547,
      "loss": 2.3909,
      "step": 20657
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6373323798179626,
      "learning_rate": 0.0005881772657320488,
      "loss": 2.0234,
      "step": 20658
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6094235181808472,
      "learning_rate": 0.0005881761286676653,
      "loss": 2.2553,
      "step": 20659
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6288781762123108,
      "learning_rate": 0.0005881749915497044,
      "loss": 2.1408,
      "step": 20660
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.633542001247406,
      "learning_rate": 0.0005881738543781664,
      "loss": 2.3925,
      "step": 20661
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.7035269737243652,
      "learning_rate": 0.0005881727171530514,
      "loss": 2.4213,
      "step": 20662
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6028204560279846,
      "learning_rate": 0.0005881715798743598,
      "loss": 2.311,
      "step": 20663
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.606468141078949,
      "learning_rate": 0.0005881704425420914,
      "loss": 2.3838,
      "step": 20664
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.7758027911186218,
      "learning_rate": 0.0005881693051562468,
      "loss": 2.6547,
      "step": 20665
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6199645400047302,
      "learning_rate": 0.000588168167716826,
      "loss": 2.4013,
      "step": 20666
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6211555004119873,
      "learning_rate": 0.0005881670302238294,
      "loss": 2.3312,
      "step": 20667
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6732335090637207,
      "learning_rate": 0.0005881658926772572,
      "loss": 2.4775,
      "step": 20668
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6599908471107483,
      "learning_rate": 0.0005881647550771093,
      "loss": 2.2869,
      "step": 20669
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.9754161238670349,
      "learning_rate": 0.0005881636174233863,
      "loss": 2.5129,
      "step": 20670
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6218610405921936,
      "learning_rate": 0.0005881624797160881,
      "loss": 2.2893,
      "step": 20671
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.5771602392196655,
      "learning_rate": 0.0005881613419552153,
      "loss": 2.2569,
      "step": 20672
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.596564531326294,
      "learning_rate": 0.0005881602041407676,
      "loss": 2.5222,
      "step": 20673
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6108877062797546,
      "learning_rate": 0.0005881590662727456,
      "loss": 2.2735,
      "step": 20674
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.5894399881362915,
      "learning_rate": 0.0005881579283511493,
      "loss": 2.3358,
      "step": 20675
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6529217958450317,
      "learning_rate": 0.0005881567903759791,
      "loss": 2.4866,
      "step": 20676
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.5595535039901733,
      "learning_rate": 0.0005881556523472351,
      "loss": 2.282,
      "step": 20677
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.60710209608078,
      "learning_rate": 0.0005881545142649176,
      "loss": 2.1805,
      "step": 20678
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.7294418215751648,
      "learning_rate": 0.0005881533761290266,
      "loss": 2.3737,
      "step": 20679
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6542319655418396,
      "learning_rate": 0.0005881522379395625,
      "loss": 2.2643,
      "step": 20680
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.7266443967819214,
      "learning_rate": 0.0005881510996965254,
      "loss": 2.4777,
      "step": 20681
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.779940128326416,
      "learning_rate": 0.0005881499613999157,
      "loss": 2.3735,
      "step": 20682
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6152299642562866,
      "learning_rate": 0.0005881488230497334,
      "loss": 2.3012,
      "step": 20683
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6000643968582153,
      "learning_rate": 0.0005881476846459788,
      "loss": 2.4146,
      "step": 20684
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.7347477078437805,
      "learning_rate": 0.0005881465461886521,
      "loss": 2.4519,
      "step": 20685
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.7156407833099365,
      "learning_rate": 0.0005881454076777535,
      "loss": 2.3057,
      "step": 20686
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.5853729844093323,
      "learning_rate": 0.0005881442691132834,
      "loss": 2.2694,
      "step": 20687
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.7456978559494019,
      "learning_rate": 0.0005881431304952416,
      "loss": 2.4624,
      "step": 20688
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6613048315048218,
      "learning_rate": 0.0005881419918236286,
      "loss": 2.2379,
      "step": 20689
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6946187615394592,
      "learning_rate": 0.0005881408530984447,
      "loss": 2.511,
      "step": 20690
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6833044290542603,
      "learning_rate": 0.0005881397143196899,
      "loss": 2.4736,
      "step": 20691
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6714262366294861,
      "learning_rate": 0.0005881385754873645,
      "loss": 2.5237,
      "step": 20692
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.5792884230613708,
      "learning_rate": 0.0005881374366014686,
      "loss": 2.3241,
      "step": 20693
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6538657546043396,
      "learning_rate": 0.0005881362976620026,
      "loss": 2.2039,
      "step": 20694
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6507801413536072,
      "learning_rate": 0.0005881351586689667,
      "loss": 2.3198,
      "step": 20695
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6549820303916931,
      "learning_rate": 0.000588134019622361,
      "loss": 2.4099,
      "step": 20696
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6775983572006226,
      "learning_rate": 0.0005881328805221855,
      "loss": 2.3834,
      "step": 20697
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.7688465118408203,
      "learning_rate": 0.000588131741368441,
      "loss": 2.2718,
      "step": 20698
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.5659500956535339,
      "learning_rate": 0.0005881306021611272,
      "loss": 2.2337,
      "step": 20699
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.7019038796424866,
      "learning_rate": 0.0005881294629002446,
      "loss": 2.3175,
      "step": 20700
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6026976108551025,
      "learning_rate": 0.0005881283235857931,
      "loss": 2.1951,
      "step": 20701
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6786700487136841,
      "learning_rate": 0.0005881271842177732,
      "loss": 2.1694,
      "step": 20702
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6074581146240234,
      "learning_rate": 0.0005881260447961851,
      "loss": 2.4516,
      "step": 20703
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6498615145683289,
      "learning_rate": 0.0005881249053210288,
      "loss": 2.4704,
      "step": 20704
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.660966694355011,
      "learning_rate": 0.0005881237657923048,
      "loss": 2.3262,
      "step": 20705
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.7209450602531433,
      "learning_rate": 0.0005881226262100129,
      "loss": 2.4968,
      "step": 20706
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6276998519897461,
      "learning_rate": 0.0005881214865741538,
      "loss": 2.5336,
      "step": 20707
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6146437525749207,
      "learning_rate": 0.0005881203468847273,
      "loss": 2.5309,
      "step": 20708
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.5907920002937317,
      "learning_rate": 0.0005881192071417339,
      "loss": 2.3205,
      "step": 20709
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6321459412574768,
      "learning_rate": 0.0005881180673451737,
      "loss": 2.4956,
      "step": 20710
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6150679588317871,
      "learning_rate": 0.000588116927495047,
      "loss": 2.0657,
      "step": 20711
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6090306043624878,
      "learning_rate": 0.0005881157875913537,
      "loss": 2.4483,
      "step": 20712
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6101357340812683,
      "learning_rate": 0.0005881146476340944,
      "loss": 2.582,
      "step": 20713
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.5759391784667969,
      "learning_rate": 0.0005881135076232692,
      "loss": 2.2654,
      "step": 20714
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6440243721008301,
      "learning_rate": 0.0005881123675588781,
      "loss": 2.2707,
      "step": 20715
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6279211640357971,
      "learning_rate": 0.0005881112274409216,
      "loss": 2.6074,
      "step": 20716
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.5899998545646667,
      "learning_rate": 0.0005881100872693997,
      "loss": 2.4637,
      "step": 20717
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.589024543762207,
      "learning_rate": 0.0005881089470443127,
      "loss": 2.1954,
      "step": 20718
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.5806246995925903,
      "learning_rate": 0.0005881078067656608,
      "loss": 2.2888,
      "step": 20719
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6424360275268555,
      "learning_rate": 0.0005881066664334443,
      "loss": 2.3581,
      "step": 20720
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6005542874336243,
      "learning_rate": 0.0005881055260476632,
      "loss": 2.3181,
      "step": 20721
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6004992723464966,
      "learning_rate": 0.000588104385608318,
      "loss": 2.5671,
      "step": 20722
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6299502849578857,
      "learning_rate": 0.0005881032451154086,
      "loss": 2.2482,
      "step": 20723
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.5841407775878906,
      "learning_rate": 0.0005881021045689355,
      "loss": 2.4004,
      "step": 20724
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6260671615600586,
      "learning_rate": 0.0005881009639688988,
      "loss": 2.3986,
      "step": 20725
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6608031988143921,
      "learning_rate": 0.0005880998233152986,
      "loss": 2.471,
      "step": 20726
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.5762455463409424,
      "learning_rate": 0.0005880986826081352,
      "loss": 2.4614,
      "step": 20727
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.61978679895401,
      "learning_rate": 0.0005880975418474088,
      "loss": 2.2908,
      "step": 20728
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.5546919703483582,
      "learning_rate": 0.0005880964010331197,
      "loss": 2.2893,
      "step": 20729
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.5793250799179077,
      "learning_rate": 0.000588095260165268,
      "loss": 2.3933,
      "step": 20730
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6251701712608337,
      "learning_rate": 0.000588094119243854,
      "loss": 2.5571,
      "step": 20731
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.641653299331665,
      "learning_rate": 0.0005880929782688778,
      "loss": 2.3585,
      "step": 20732
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.5741167068481445,
      "learning_rate": 0.0005880918372403398,
      "loss": 2.3179,
      "step": 20733
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6010855436325073,
      "learning_rate": 0.0005880906961582399,
      "loss": 2.3533,
      "step": 20734
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6101718544960022,
      "learning_rate": 0.0005880895550225787,
      "loss": 2.2885,
      "step": 20735
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6368575692176819,
      "learning_rate": 0.000588088413833356,
      "loss": 2.4361,
      "step": 20736
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6458845138549805,
      "learning_rate": 0.0005880872725905723,
      "loss": 2.3571,
      "step": 20737
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.661719024181366,
      "learning_rate": 0.0005880861312942279,
      "loss": 2.5363,
      "step": 20738
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.5562655329704285,
      "learning_rate": 0.0005880849899443227,
      "loss": 2.7636,
      "step": 20739
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6285609602928162,
      "learning_rate": 0.000588083848540857,
      "loss": 2.3081,
      "step": 20740
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.5631808042526245,
      "learning_rate": 0.0005880827070838312,
      "loss": 2.1839,
      "step": 20741
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6011231541633606,
      "learning_rate": 0.0005880815655732454,
      "loss": 2.4796,
      "step": 20742
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6311399936676025,
      "learning_rate": 0.0005880804240090998,
      "loss": 2.4935,
      "step": 20743
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6999459862709045,
      "learning_rate": 0.0005880792823913945,
      "loss": 2.4244,
      "step": 20744
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.5841995477676392,
      "learning_rate": 0.0005880781407201299,
      "loss": 2.1815,
      "step": 20745
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.796486496925354,
      "learning_rate": 0.0005880769989953062,
      "loss": 2.4282,
      "step": 20746
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6134026050567627,
      "learning_rate": 0.0005880758572169234,
      "loss": 2.5967,
      "step": 20747
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6244166493415833,
      "learning_rate": 0.0005880747153849819,
      "loss": 2.4338,
      "step": 20748
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.7003054618835449,
      "learning_rate": 0.0005880735734994819,
      "loss": 2.345,
      "step": 20749
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.7224261164665222,
      "learning_rate": 0.0005880724315604235,
      "loss": 2.2732,
      "step": 20750
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6270129084587097,
      "learning_rate": 0.0005880712895678071,
      "loss": 2.1728,
      "step": 20751
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6131342649459839,
      "learning_rate": 0.0005880701475216328,
      "loss": 2.2449,
      "step": 20752
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6662365198135376,
      "learning_rate": 0.0005880690054219009,
      "loss": 2.1754,
      "step": 20753
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.7020848393440247,
      "learning_rate": 0.0005880678632686114,
      "loss": 2.6007,
      "step": 20754
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.67452073097229,
      "learning_rate": 0.0005880667210617647,
      "loss": 2.404,
      "step": 20755
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.8255343437194824,
      "learning_rate": 0.0005880655788013609,
      "loss": 2.4162,
      "step": 20756
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.677895188331604,
      "learning_rate": 0.0005880644364874004,
      "loss": 2.3195,
      "step": 20757
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6206761598587036,
      "learning_rate": 0.0005880632941198832,
      "loss": 2.4763,
      "step": 20758
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.7216781973838806,
      "learning_rate": 0.0005880621516988096,
      "loss": 2.191,
      "step": 20759
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6513577699661255,
      "learning_rate": 0.0005880610092241798,
      "loss": 2.3058,
      "step": 20760
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6162956357002258,
      "learning_rate": 0.000588059866695994,
      "loss": 2.5906,
      "step": 20761
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.7076271772384644,
      "learning_rate": 0.0005880587241142525,
      "loss": 2.4315,
      "step": 20762
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6520934104919434,
      "learning_rate": 0.0005880575814789554,
      "loss": 2.4743,
      "step": 20763
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6120932698249817,
      "learning_rate": 0.000588056438790103,
      "loss": 2.4295,
      "step": 20764
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6245542764663696,
      "learning_rate": 0.0005880552960476955,
      "loss": 2.2648,
      "step": 20765
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6714977622032166,
      "learning_rate": 0.0005880541532517329,
      "loss": 2.1349,
      "step": 20766
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.5810667872428894,
      "learning_rate": 0.0005880530104022158,
      "loss": 2.3743,
      "step": 20767
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6435149312019348,
      "learning_rate": 0.0005880518674991442,
      "loss": 2.483,
      "step": 20768
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6911484599113464,
      "learning_rate": 0.0005880507245425181,
      "loss": 2.4997,
      "step": 20769
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.5963571667671204,
      "learning_rate": 0.0005880495815323381,
      "loss": 2.2136,
      "step": 20770
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6234270930290222,
      "learning_rate": 0.0005880484384686042,
      "loss": 2.3957,
      "step": 20771
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.5595616698265076,
      "learning_rate": 0.0005880472953513167,
      "loss": 2.4117,
      "step": 20772
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.5608252286911011,
      "learning_rate": 0.0005880461521804757,
      "loss": 2.2347,
      "step": 20773
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.5543109774589539,
      "learning_rate": 0.0005880450089560817,
      "loss": 2.3046,
      "step": 20774
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.8780747056007385,
      "learning_rate": 0.0005880438656781344,
      "loss": 2.372,
      "step": 20775
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.5827773809432983,
      "learning_rate": 0.0005880427223466345,
      "loss": 2.6133,
      "step": 20776
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6533207297325134,
      "learning_rate": 0.000588041578961582,
      "loss": 2.3692,
      "step": 20777
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6615782976150513,
      "learning_rate": 0.0005880404355229771,
      "loss": 2.2261,
      "step": 20778
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6128902435302734,
      "learning_rate": 0.0005880392920308199,
      "loss": 2.4212,
      "step": 20779
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6118434071540833,
      "learning_rate": 0.000588038148485111,
      "loss": 2.3229,
      "step": 20780
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6034579277038574,
      "learning_rate": 0.0005880370048858503,
      "loss": 2.3718,
      "step": 20781
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.5785830616950989,
      "learning_rate": 0.000588035861233038,
      "loss": 2.276,
      "step": 20782
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.5966362953186035,
      "learning_rate": 0.0005880347175266746,
      "loss": 2.4287,
      "step": 20783
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6777548789978027,
      "learning_rate": 0.0005880335737667599,
      "loss": 2.2392,
      "step": 20784
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6280476450920105,
      "learning_rate": 0.0005880324299532943,
      "loss": 2.4136,
      "step": 20785
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.5789987444877625,
      "learning_rate": 0.0005880312860862782,
      "loss": 2.2555,
      "step": 20786
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.7252737879753113,
      "learning_rate": 0.0005880301421657115,
      "loss": 2.4952,
      "step": 20787
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.7106830477714539,
      "learning_rate": 0.0005880289981915946,
      "loss": 2.3717,
      "step": 20788
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.5495457649230957,
      "learning_rate": 0.0005880278541639277,
      "loss": 2.39,
      "step": 20789
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6064981818199158,
      "learning_rate": 0.000588026710082711,
      "loss": 2.3244,
      "step": 20790
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6037034392356873,
      "learning_rate": 0.0005880255659479447,
      "loss": 2.3139,
      "step": 20791
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6156810522079468,
      "learning_rate": 0.0005880244217596289,
      "loss": 2.5202,
      "step": 20792
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.5807627439498901,
      "learning_rate": 0.0005880232775177641,
      "loss": 2.1784,
      "step": 20793
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.543879508972168,
      "learning_rate": 0.0005880221332223503,
      "loss": 2.3159,
      "step": 20794
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.5746620893478394,
      "learning_rate": 0.0005880209888733877,
      "loss": 2.3557,
      "step": 20795
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.5772425532341003,
      "learning_rate": 0.0005880198444708765,
      "loss": 2.5234,
      "step": 20796
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6338167786598206,
      "learning_rate": 0.000588018700014817,
      "loss": 2.3364,
      "step": 20797
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.7228369116783142,
      "learning_rate": 0.0005880175555052095,
      "loss": 2.4932,
      "step": 20798
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6244817972183228,
      "learning_rate": 0.000588016410942054,
      "loss": 2.2168,
      "step": 20799
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6319260597229004,
      "learning_rate": 0.0005880152663253508,
      "loss": 2.6202,
      "step": 20800
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6781609058380127,
      "learning_rate": 0.0005880141216551001,
      "loss": 2.2436,
      "step": 20801
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6878251433372498,
      "learning_rate": 0.0005880129769313022,
      "loss": 2.3159,
      "step": 20802
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6351966261863708,
      "learning_rate": 0.0005880118321539572,
      "loss": 2.383,
      "step": 20803
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.5895098447799683,
      "learning_rate": 0.0005880106873230655,
      "loss": 2.3588,
      "step": 20804
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.8596716523170471,
      "learning_rate": 0.0005880095424386271,
      "loss": 2.2194,
      "step": 20805
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.76495760679245,
      "learning_rate": 0.0005880083975006422,
      "loss": 2.3891,
      "step": 20806
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.738097608089447,
      "learning_rate": 0.0005880072525091112,
      "loss": 2.1263,
      "step": 20807
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.592069685459137,
      "learning_rate": 0.0005880061074640342,
      "loss": 2.2764,
      "step": 20808
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.633257269859314,
      "learning_rate": 0.0005880049623654114,
      "loss": 2.5745,
      "step": 20809
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6653617024421692,
      "learning_rate": 0.0005880038172132431,
      "loss": 2.4452,
      "step": 20810
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.7877769470214844,
      "learning_rate": 0.0005880026720075294,
      "loss": 2.2239,
      "step": 20811
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6666091084480286,
      "learning_rate": 0.0005880015267482705,
      "loss": 2.4154,
      "step": 20812
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.623075544834137,
      "learning_rate": 0.0005880003814354669,
      "loss": 2.4317,
      "step": 20813
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.7454056143760681,
      "learning_rate": 0.0005879992360691184,
      "loss": 2.584,
      "step": 20814
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6658127307891846,
      "learning_rate": 0.0005879980906492254,
      "loss": 2.1,
      "step": 20815
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6466654539108276,
      "learning_rate": 0.0005879969451757882,
      "loss": 2.6037,
      "step": 20816
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6412954330444336,
      "learning_rate": 0.0005879957996488069,
      "loss": 2.2298,
      "step": 20817
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.7663253545761108,
      "learning_rate": 0.0005879946540682817,
      "loss": 2.263,
      "step": 20818
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.5698791146278381,
      "learning_rate": 0.000587993508434213,
      "loss": 2.363,
      "step": 20819
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6451497077941895,
      "learning_rate": 0.0005879923627466007,
      "loss": 2.2348,
      "step": 20820
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.7329026460647583,
      "learning_rate": 0.0005879912170054454,
      "loss": 2.4757,
      "step": 20821
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6054043769836426,
      "learning_rate": 0.0005879900712107469,
      "loss": 2.3887,
      "step": 20822
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.57401442527771,
      "learning_rate": 0.0005879889253625057,
      "loss": 2.457,
      "step": 20823
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.5918635129928589,
      "learning_rate": 0.0005879877794607219,
      "loss": 2.3885,
      "step": 20824
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6914970874786377,
      "learning_rate": 0.0005879866335053957,
      "loss": 2.5584,
      "step": 20825
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6562433242797852,
      "learning_rate": 0.0005879854874965274,
      "loss": 2.3633,
      "step": 20826
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6208232045173645,
      "learning_rate": 0.000587984341434117,
      "loss": 2.3692,
      "step": 20827
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6065628528594971,
      "learning_rate": 0.0005879831953181651,
      "loss": 2.489,
      "step": 20828
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.5986159443855286,
      "learning_rate": 0.0005879820491486716,
      "loss": 2.5731,
      "step": 20829
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.645117998123169,
      "learning_rate": 0.0005879809029256369,
      "loss": 2.5309,
      "step": 20830
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.593101978302002,
      "learning_rate": 0.000587979756649061,
      "loss": 2.3645,
      "step": 20831
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.605384886264801,
      "learning_rate": 0.0005879786103189442,
      "loss": 2.2597,
      "step": 20832
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6173176765441895,
      "learning_rate": 0.0005879774639352867,
      "loss": 2.5367,
      "step": 20833
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.5846701264381409,
      "learning_rate": 0.000587976317498089,
      "loss": 2.4105,
      "step": 20834
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.5822181105613708,
      "learning_rate": 0.0005879751710073509,
      "loss": 2.32,
      "step": 20835
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6308282017707825,
      "learning_rate": 0.0005879740244630726,
      "loss": 2.4804,
      "step": 20836
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.614740252494812,
      "learning_rate": 0.0005879728778652547,
      "loss": 2.4206,
      "step": 20837
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.5841541886329651,
      "learning_rate": 0.0005879717312138972,
      "loss": 2.2977,
      "step": 20838
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.602453351020813,
      "learning_rate": 0.0005879705845090004,
      "loss": 2.2806,
      "step": 20839
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6683524250984192,
      "learning_rate": 0.0005879694377505643,
      "loss": 2.5307,
      "step": 20840
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6188340187072754,
      "learning_rate": 0.0005879682909385893,
      "loss": 2.6232,
      "step": 20841
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6432375907897949,
      "learning_rate": 0.0005879671440730755,
      "loss": 2.4229,
      "step": 20842
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.592795729637146,
      "learning_rate": 0.0005879659971540232,
      "loss": 2.2453,
      "step": 20843
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.558485209941864,
      "learning_rate": 0.0005879648501814326,
      "loss": 2.3399,
      "step": 20844
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6123465299606323,
      "learning_rate": 0.000587963703155304,
      "loss": 2.2907,
      "step": 20845
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.601965606212616,
      "learning_rate": 0.0005879625560756374,
      "loss": 2.3503,
      "step": 20846
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.5923047065734863,
      "learning_rate": 0.0005879614089424331,
      "loss": 2.4546,
      "step": 20847
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.5818508267402649,
      "learning_rate": 0.0005879602617556913,
      "loss": 2.3599,
      "step": 20848
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.5837268829345703,
      "learning_rate": 0.0005879591145154123,
      "loss": 2.6256,
      "step": 20849
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.5609029531478882,
      "learning_rate": 0.0005879579672215964,
      "loss": 2.4834,
      "step": 20850
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6069080829620361,
      "learning_rate": 0.0005879568198742435,
      "loss": 2.2899,
      "step": 20851
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.5844355821609497,
      "learning_rate": 0.0005879556724733542,
      "loss": 2.5323,
      "step": 20852
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6764238476753235,
      "learning_rate": 0.0005879545250189282,
      "loss": 2.2642,
      "step": 20853
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.7913020849227905,
      "learning_rate": 0.0005879533775109663,
      "loss": 2.4283,
      "step": 20854
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6722264289855957,
      "learning_rate": 0.0005879522299494682,
      "loss": 2.5059,
      "step": 20855
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6717031002044678,
      "learning_rate": 0.0005879510823344345,
      "loss": 2.3227,
      "step": 20856
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.7275176048278809,
      "learning_rate": 0.0005879499346658653,
      "loss": 2.4037,
      "step": 20857
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6146785616874695,
      "learning_rate": 0.0005879487869437605,
      "loss": 2.5933,
      "step": 20858
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.674206554889679,
      "learning_rate": 0.0005879476391681208,
      "loss": 2.4397,
      "step": 20859
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.621269166469574,
      "learning_rate": 0.0005879464913389462,
      "loss": 2.4923,
      "step": 20860
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.709345281124115,
      "learning_rate": 0.0005879453434562368,
      "loss": 2.2414,
      "step": 20861
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.5832513570785522,
      "learning_rate": 0.0005879441955199931,
      "loss": 2.328,
      "step": 20862
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6317736506462097,
      "learning_rate": 0.0005879430475302151,
      "loss": 2.3694,
      "step": 20863
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6358219981193542,
      "learning_rate": 0.000587941899486903,
      "loss": 2.1993,
      "step": 20864
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.5947153568267822,
      "learning_rate": 0.0005879407513900571,
      "loss": 2.2643,
      "step": 20865
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.5800060629844666,
      "learning_rate": 0.0005879396032396775,
      "loss": 2.4095,
      "step": 20866
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.7356171607971191,
      "learning_rate": 0.0005879384550357645,
      "loss": 2.3037,
      "step": 20867
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6705437302589417,
      "learning_rate": 0.0005879373067783183,
      "loss": 2.5954,
      "step": 20868
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6768478751182556,
      "learning_rate": 0.0005879361584673391,
      "loss": 2.3609,
      "step": 20869
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6185570955276489,
      "learning_rate": 0.0005879350101028273,
      "loss": 2.3933,
      "step": 20870
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.7055125832557678,
      "learning_rate": 0.0005879338616847829,
      "loss": 2.3571,
      "step": 20871
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.5929871201515198,
      "learning_rate": 0.000587932713213206,
      "loss": 2.361,
      "step": 20872
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6386027336120605,
      "learning_rate": 0.0005879315646880972,
      "loss": 2.4377,
      "step": 20873
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6148137450218201,
      "learning_rate": 0.0005879304161094562,
      "loss": 2.3735,
      "step": 20874
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6405426859855652,
      "learning_rate": 0.0005879292674772838,
      "loss": 2.4671,
      "step": 20875
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.5799481272697449,
      "learning_rate": 0.0005879281187915798,
      "loss": 2.4585,
      "step": 20876
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6038681268692017,
      "learning_rate": 0.0005879269700523444,
      "loss": 2.1595,
      "step": 20877
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.5483949780464172,
      "learning_rate": 0.000587925821259578,
      "loss": 2.2164,
      "step": 20878
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.652060329914093,
      "learning_rate": 0.0005879246724132808,
      "loss": 2.5181,
      "step": 20879
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.7169815301895142,
      "learning_rate": 0.000587923523513453,
      "loss": 2.4658,
      "step": 20880
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.59844571352005,
      "learning_rate": 0.0005879223745600948,
      "loss": 2.2607,
      "step": 20881
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6200881004333496,
      "learning_rate": 0.0005879212255532064,
      "loss": 2.4994,
      "step": 20882
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6429047584533691,
      "learning_rate": 0.0005879200764927879,
      "loss": 2.3556,
      "step": 20883
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6052663922309875,
      "learning_rate": 0.0005879189273788396,
      "loss": 2.1977,
      "step": 20884
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.8837676644325256,
      "learning_rate": 0.0005879177782113619,
      "loss": 2.2999,
      "step": 20885
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6675915718078613,
      "learning_rate": 0.0005879166289903548,
      "loss": 2.3205,
      "step": 20886
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6388642191886902,
      "learning_rate": 0.0005879154797158184,
      "loss": 2.3556,
      "step": 20887
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6054638624191284,
      "learning_rate": 0.0005879143303877532,
      "loss": 2.382,
      "step": 20888
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.5851168632507324,
      "learning_rate": 0.0005879131810061594,
      "loss": 2.3221,
      "step": 20889
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.7249556183815002,
      "learning_rate": 0.000587912031571037,
      "loss": 2.1326,
      "step": 20890
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.5941275954246521,
      "learning_rate": 0.0005879108820823863,
      "loss": 2.5015,
      "step": 20891
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.5835919976234436,
      "learning_rate": 0.0005879097325402076,
      "loss": 2.1606,
      "step": 20892
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6360006332397461,
      "learning_rate": 0.000587908582944501,
      "loss": 2.4234,
      "step": 20893
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.592711329460144,
      "learning_rate": 0.0005879074332952668,
      "loss": 2.2931,
      "step": 20894
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6077413558959961,
      "learning_rate": 0.0005879062835925051,
      "loss": 2.1961,
      "step": 20895
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.5632214546203613,
      "learning_rate": 0.0005879051338362162,
      "loss": 2.4942,
      "step": 20896
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6195211410522461,
      "learning_rate": 0.0005879039840264003,
      "loss": 2.5668,
      "step": 20897
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.567612886428833,
      "learning_rate": 0.0005879028341630577,
      "loss": 2.212,
      "step": 20898
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6392291188240051,
      "learning_rate": 0.0005879016842461883,
      "loss": 2.3889,
      "step": 20899
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.5824880599975586,
      "learning_rate": 0.0005879005342757928,
      "loss": 2.076,
      "step": 20900
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.5776225328445435,
      "learning_rate": 0.0005878993842518711,
      "loss": 2.3093,
      "step": 20901
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6093795895576477,
      "learning_rate": 0.0005878982341744233,
      "loss": 2.4816,
      "step": 20902
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.5973010659217834,
      "learning_rate": 0.00058789708404345,
      "loss": 2.317,
      "step": 20903
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6344763040542603,
      "learning_rate": 0.0005878959338589511,
      "loss": 2.4578,
      "step": 20904
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6125070452690125,
      "learning_rate": 0.0005878947836209269,
      "loss": 2.3097,
      "step": 20905
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.5708536505699158,
      "learning_rate": 0.0005878936333293777,
      "loss": 2.3209,
      "step": 20906
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.747444748878479,
      "learning_rate": 0.0005878924829843035,
      "loss": 2.356,
      "step": 20907
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6351221799850464,
      "learning_rate": 0.0005878913325857049,
      "loss": 2.3144,
      "step": 20908
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6002723574638367,
      "learning_rate": 0.0005878901821335815,
      "loss": 2.2842,
      "step": 20909
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6456857323646545,
      "learning_rate": 0.0005878890316279342,
      "loss": 2.1918,
      "step": 20910
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.5765355825424194,
      "learning_rate": 0.0005878878810687627,
      "loss": 2.2603,
      "step": 20911
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.603082001209259,
      "learning_rate": 0.0005878867304560676,
      "loss": 2.5323,
      "step": 20912
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.5921998620033264,
      "learning_rate": 0.0005878855797898487,
      "loss": 2.5154,
      "step": 20913
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.5501524209976196,
      "learning_rate": 0.0005878844290701066,
      "loss": 2.3713,
      "step": 20914
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6723116636276245,
      "learning_rate": 0.0005878832782968414,
      "loss": 2.6165,
      "step": 20915
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6100448966026306,
      "learning_rate": 0.0005878821274700531,
      "loss": 2.4827,
      "step": 20916
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.7302963137626648,
      "learning_rate": 0.0005878809765897422,
      "loss": 2.3882,
      "step": 20917
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.688064455986023,
      "learning_rate": 0.0005878798256559087,
      "loss": 2.1739,
      "step": 20918
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.5801066160202026,
      "learning_rate": 0.0005878786746685529,
      "loss": 2.3756,
      "step": 20919
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6779982447624207,
      "learning_rate": 0.0005878775236276752,
      "loss": 2.4398,
      "step": 20920
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.5747855305671692,
      "learning_rate": 0.0005878763725332754,
      "loss": 2.3847,
      "step": 20921
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6261058449745178,
      "learning_rate": 0.000587875221385354,
      "loss": 2.4011,
      "step": 20922
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6252062916755676,
      "learning_rate": 0.0005878740701839113,
      "loss": 2.3219,
      "step": 20923
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6681230664253235,
      "learning_rate": 0.0005878729189289473,
      "loss": 2.4279,
      "step": 20924
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6170828938484192,
      "learning_rate": 0.0005878717676204622,
      "loss": 2.4494,
      "step": 20925
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6459372043609619,
      "learning_rate": 0.0005878706162584564,
      "loss": 2.4716,
      "step": 20926
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.7540595531463623,
      "learning_rate": 0.0005878694648429301,
      "loss": 2.2115,
      "step": 20927
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6210243105888367,
      "learning_rate": 0.0005878683133738832,
      "loss": 2.2588,
      "step": 20928
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6286531686782837,
      "learning_rate": 0.0005878671618513164,
      "loss": 2.3702,
      "step": 20929
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.8011969327926636,
      "learning_rate": 0.0005878660102752295,
      "loss": 2.3144,
      "step": 20930
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6902930736541748,
      "learning_rate": 0.0005878648586456228,
      "loss": 2.3784,
      "step": 20931
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6019160151481628,
      "learning_rate": 0.0005878637069624968,
      "loss": 2.3441,
      "step": 20932
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6711896657943726,
      "learning_rate": 0.0005878625552258514,
      "loss": 2.3702,
      "step": 20933
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.9432099461555481,
      "learning_rate": 0.0005878614034356868,
      "loss": 2.3659,
      "step": 20934
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6070415377616882,
      "learning_rate": 0.0005878602515920036,
      "loss": 2.3246,
      "step": 20935
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.657250702381134,
      "learning_rate": 0.0005878590996948016,
      "loss": 2.1654,
      "step": 20936
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.70283442735672,
      "learning_rate": 0.000587857947744081,
      "loss": 2.5937,
      "step": 20937
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.7651161551475525,
      "learning_rate": 0.0005878567957398424,
      "loss": 2.2162,
      "step": 20938
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6671104431152344,
      "learning_rate": 0.0005878556436820858,
      "loss": 2.2387,
      "step": 20939
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6915179491043091,
      "learning_rate": 0.0005878544915708113,
      "loss": 2.5371,
      "step": 20940
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.7486032247543335,
      "learning_rate": 0.0005878533394060192,
      "loss": 2.1547,
      "step": 20941
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.5960940718650818,
      "learning_rate": 0.0005878521871877098,
      "loss": 2.1987,
      "step": 20942
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6841585636138916,
      "learning_rate": 0.0005878510349158831,
      "loss": 2.4239,
      "step": 20943
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6025869846343994,
      "learning_rate": 0.0005878498825905397,
      "loss": 2.2748,
      "step": 20944
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.649874210357666,
      "learning_rate": 0.0005878487302116794,
      "loss": 2.4625,
      "step": 20945
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6321378350257874,
      "learning_rate": 0.0005878475777793027,
      "loss": 2.2865,
      "step": 20946
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6692774891853333,
      "learning_rate": 0.0005878464252934096,
      "loss": 2.4275,
      "step": 20947
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.598539412021637,
      "learning_rate": 0.0005878452727540004,
      "loss": 2.5237,
      "step": 20948
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6059318780899048,
      "learning_rate": 0.0005878441201610754,
      "loss": 2.2519,
      "step": 20949
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6635137796401978,
      "learning_rate": 0.0005878429675146348,
      "loss": 2.5051,
      "step": 20950
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6018337607383728,
      "learning_rate": 0.0005878418148146786,
      "loss": 2.4493,
      "step": 20951
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6209614276885986,
      "learning_rate": 0.0005878406620612072,
      "loss": 2.4446,
      "step": 20952
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6126751899719238,
      "learning_rate": 0.0005878395092542209,
      "loss": 2.3265,
      "step": 20953
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.608252227306366,
      "learning_rate": 0.0005878383563937198,
      "loss": 2.4387,
      "step": 20954
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6324303150177002,
      "learning_rate": 0.000587837203479704,
      "loss": 2.3427,
      "step": 20955
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.5984248518943787,
      "learning_rate": 0.000587836050512174,
      "loss": 1.9913,
      "step": 20956
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6223453879356384,
      "learning_rate": 0.0005878348974911297,
      "loss": 2.3486,
      "step": 20957
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6871870160102844,
      "learning_rate": 0.0005878337444165715,
      "loss": 2.4402,
      "step": 20958
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6105547547340393,
      "learning_rate": 0.0005878325912884995,
      "loss": 2.4038,
      "step": 20959
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.5725296139717102,
      "learning_rate": 0.0005878314381069141,
      "loss": 2.571,
      "step": 20960
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6520906686782837,
      "learning_rate": 0.0005878302848718154,
      "loss": 2.1511,
      "step": 20961
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.5662213563919067,
      "learning_rate": 0.0005878291315832035,
      "loss": 2.3388,
      "step": 20962
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.7351252436637878,
      "learning_rate": 0.0005878279782410788,
      "loss": 2.3964,
      "step": 20963
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6760061383247375,
      "learning_rate": 0.0005878268248454416,
      "loss": 2.2426,
      "step": 20964
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.7095102667808533,
      "learning_rate": 0.0005878256713962918,
      "loss": 2.1927,
      "step": 20965
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.682157039642334,
      "learning_rate": 0.0005878245178936297,
      "loss": 2.2979,
      "step": 20966
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6322926878929138,
      "learning_rate": 0.0005878233643374556,
      "loss": 2.6562,
      "step": 20967
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6665017604827881,
      "learning_rate": 0.0005878222107277699,
      "loss": 2.2202,
      "step": 20968
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6077487468719482,
      "learning_rate": 0.0005878210570645724,
      "loss": 2.5154,
      "step": 20969
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6461212038993835,
      "learning_rate": 0.0005878199033478637,
      "loss": 2.3507,
      "step": 20970
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6537450551986694,
      "learning_rate": 0.0005878187495776438,
      "loss": 2.465,
      "step": 20971
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6119550466537476,
      "learning_rate": 0.0005878175957539129,
      "loss": 2.3986,
      "step": 20972
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6332510113716125,
      "learning_rate": 0.0005878164418766713,
      "loss": 2.3698,
      "step": 20973
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6462979316711426,
      "learning_rate": 0.0005878152879459193,
      "loss": 2.3083,
      "step": 20974
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6500452756881714,
      "learning_rate": 0.0005878141339616569,
      "loss": 2.2355,
      "step": 20975
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6471279859542847,
      "learning_rate": 0.0005878129799238844,
      "loss": 2.4129,
      "step": 20976
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6786884069442749,
      "learning_rate": 0.0005878118258326021,
      "loss": 2.3111,
      "step": 20977
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.5957291126251221,
      "learning_rate": 0.0005878106716878101,
      "loss": 2.5238,
      "step": 20978
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.8202466368675232,
      "learning_rate": 0.0005878095174895086,
      "loss": 2.2868,
      "step": 20979
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6884375810623169,
      "learning_rate": 0.000587808363237698,
      "loss": 2.2505,
      "step": 20980
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.624981164932251,
      "learning_rate": 0.0005878072089323783,
      "loss": 2.2656,
      "step": 20981
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.9020838141441345,
      "learning_rate": 0.00058780605457355,
      "loss": 2.4486,
      "step": 20982
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.7293612360954285,
      "learning_rate": 0.0005878049001612129,
      "loss": 2.4333,
      "step": 20983
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6313952207565308,
      "learning_rate": 0.0005878037456953676,
      "loss": 2.4719,
      "step": 20984
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.7306304574012756,
      "learning_rate": 0.000587802591176014,
      "loss": 2.1889,
      "step": 20985
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.777519166469574,
      "learning_rate": 0.0005878014366031525,
      "loss": 2.4607,
      "step": 20986
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6424841284751892,
      "learning_rate": 0.0005878002819767834,
      "loss": 2.7848,
      "step": 20987
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6978187561035156,
      "learning_rate": 0.0005877991272969066,
      "loss": 2.5296,
      "step": 20988
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.8638870120048523,
      "learning_rate": 0.0005877979725635227,
      "loss": 2.4179,
      "step": 20989
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6367807388305664,
      "learning_rate": 0.0005877968177766316,
      "loss": 2.2646,
      "step": 20990
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6688685417175293,
      "learning_rate": 0.0005877956629362337,
      "loss": 2.4193,
      "step": 20991
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6375488042831421,
      "learning_rate": 0.000587794508042329,
      "loss": 2.2505,
      "step": 20992
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6102805137634277,
      "learning_rate": 0.0005877933530949182,
      "loss": 2.5439,
      "step": 20993
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6979647874832153,
      "learning_rate": 0.0005877921980940008,
      "loss": 2.2941,
      "step": 20994
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6446409821510315,
      "learning_rate": 0.0005877910430395777,
      "loss": 2.2257,
      "step": 20995
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.8206890225410461,
      "learning_rate": 0.0005877898879316486,
      "loss": 2.4796,
      "step": 20996
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6735294461250305,
      "learning_rate": 0.0005877887327702141,
      "loss": 2.2937,
      "step": 20997
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6765100955963135,
      "learning_rate": 0.0005877875775552741,
      "loss": 2.2972,
      "step": 20998
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6763978600502014,
      "learning_rate": 0.0005877864222868291,
      "loss": 2.2315,
      "step": 20999
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.8089715838432312,
      "learning_rate": 0.0005877852669648791,
      "loss": 2.2375,
      "step": 21000
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6474184989929199,
      "learning_rate": 0.0005877841115894244,
      "loss": 2.3507,
      "step": 21001
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.7310771942138672,
      "learning_rate": 0.0005877829561604652,
      "loss": 2.1335,
      "step": 21002
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.8149392008781433,
      "learning_rate": 0.0005877818006780015,
      "loss": 2.5975,
      "step": 21003
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.5804483890533447,
      "learning_rate": 0.0005877806451420341,
      "loss": 2.5402,
      "step": 21004
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6929343938827515,
      "learning_rate": 0.0005877794895525626,
      "loss": 2.6752,
      "step": 21005
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.675888180732727,
      "learning_rate": 0.0005877783339095875,
      "loss": 2.4124,
      "step": 21006
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.7596396207809448,
      "learning_rate": 0.000587777178213109,
      "loss": 2.5186,
      "step": 21007
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.7251130938529968,
      "learning_rate": 0.0005877760224631273,
      "loss": 2.2968,
      "step": 21008
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.712347149848938,
      "learning_rate": 0.0005877748666596426,
      "loss": 2.1522,
      "step": 21009
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6272944808006287,
      "learning_rate": 0.0005877737108026552,
      "loss": 2.284,
      "step": 21010
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.7044889330863953,
      "learning_rate": 0.0005877725548921652,
      "loss": 2.4929,
      "step": 21011
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.616421639919281,
      "learning_rate": 0.0005877713989281728,
      "loss": 2.2044,
      "step": 21012
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6494197845458984,
      "learning_rate": 0.0005877702429106783,
      "loss": 2.3424,
      "step": 21013
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6256356835365295,
      "learning_rate": 0.0005877690868396819,
      "loss": 2.4071,
      "step": 21014
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6447513103485107,
      "learning_rate": 0.0005877679307151836,
      "loss": 2.116,
      "step": 21015
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6005074381828308,
      "learning_rate": 0.000587766774537184,
      "loss": 2.3787,
      "step": 21016
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6062427163124084,
      "learning_rate": 0.0005877656183056831,
      "loss": 2.346,
      "step": 21017
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6606791019439697,
      "learning_rate": 0.0005877644620206812,
      "loss": 2.2434,
      "step": 21018
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.7209485769271851,
      "learning_rate": 0.0005877633056821784,
      "loss": 2.5067,
      "step": 21019
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6313932538032532,
      "learning_rate": 0.0005877621492901749,
      "loss": 2.5114,
      "step": 21020
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.7322601675987244,
      "learning_rate": 0.000587760992844671,
      "loss": 2.3353,
      "step": 21021
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6337326765060425,
      "learning_rate": 0.000587759836345667,
      "loss": 2.5883,
      "step": 21022
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6233537197113037,
      "learning_rate": 0.0005877586797931629,
      "loss": 2.302,
      "step": 21023
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6241306662559509,
      "learning_rate": 0.000587757523187159,
      "loss": 2.3164,
      "step": 21024
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.626347005367279,
      "learning_rate": 0.0005877563665276556,
      "loss": 2.3809,
      "step": 21025
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6315220594406128,
      "learning_rate": 0.0005877552098146529,
      "loss": 2.3024,
      "step": 21026
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6387298703193665,
      "learning_rate": 0.0005877540530481512,
      "loss": 2.537,
      "step": 21027
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6477011442184448,
      "learning_rate": 0.0005877528962281504,
      "loss": 2.3759,
      "step": 21028
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.5942174196243286,
      "learning_rate": 0.0005877517393546509,
      "loss": 2.4206,
      "step": 21029
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6526198387145996,
      "learning_rate": 0.0005877505824276529,
      "loss": 2.5825,
      "step": 21030
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6075803637504578,
      "learning_rate": 0.0005877494254471567,
      "loss": 2.3872,
      "step": 21031
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.5574163198471069,
      "learning_rate": 0.0005877482684131625,
      "loss": 2.4183,
      "step": 21032
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6188697814941406,
      "learning_rate": 0.0005877471113256705,
      "loss": 2.2917,
      "step": 21033
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6227777600288391,
      "learning_rate": 0.0005877459541846807,
      "loss": 2.3806,
      "step": 21034
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6856203675270081,
      "learning_rate": 0.0005877447969901936,
      "loss": 2.2902,
      "step": 21035
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.5992771983146667,
      "learning_rate": 0.0005877436397422093,
      "loss": 2.4465,
      "step": 21036
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.632936418056488,
      "learning_rate": 0.000587742482440728,
      "loss": 2.1943,
      "step": 21037
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.5856232047080994,
      "learning_rate": 0.00058774132508575,
      "loss": 2.4124,
      "step": 21038
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.5836695432662964,
      "learning_rate": 0.0005877401676772755,
      "loss": 2.4219,
      "step": 21039
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6271048784255981,
      "learning_rate": 0.0005877390102153046,
      "loss": 2.4314,
      "step": 21040
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6562842130661011,
      "learning_rate": 0.0005877378526998376,
      "loss": 2.3143,
      "step": 21041
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6690356731414795,
      "learning_rate": 0.0005877366951308747,
      "loss": 2.4304,
      "step": 21042
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6625240445137024,
      "learning_rate": 0.0005877355375084161,
      "loss": 2.2515,
      "step": 21043
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6347424387931824,
      "learning_rate": 0.000587734379832462,
      "loss": 2.2616,
      "step": 21044
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.633907675743103,
      "learning_rate": 0.0005877332221030127,
      "loss": 2.3236,
      "step": 21045
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.585347592830658,
      "learning_rate": 0.0005877320643200684,
      "loss": 2.3835,
      "step": 21046
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6560395956039429,
      "learning_rate": 0.0005877309064836291,
      "loss": 2.4133,
      "step": 21047
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6345646381378174,
      "learning_rate": 0.0005877297485936953,
      "loss": 2.3789,
      "step": 21048
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6498736143112183,
      "learning_rate": 0.0005877285906502672,
      "loss": 2.3781,
      "step": 21049
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6033329367637634,
      "learning_rate": 0.0005877274326533448,
      "loss": 2.3816,
      "step": 21050
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6119557619094849,
      "learning_rate": 0.0005877262746029285,
      "loss": 2.3736,
      "step": 21051
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6099070310592651,
      "learning_rate": 0.0005877251164990184,
      "loss": 2.2851,
      "step": 21052
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6466641426086426,
      "learning_rate": 0.0005877239583416149,
      "loss": 2.4836,
      "step": 21053
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6357307434082031,
      "learning_rate": 0.000587722800130718,
      "loss": 2.1843,
      "step": 21054
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6065051555633545,
      "learning_rate": 0.000587721641866328,
      "loss": 2.066,
      "step": 21055
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6875705122947693,
      "learning_rate": 0.0005877204835484451,
      "loss": 2.3926,
      "step": 21056
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.8187911510467529,
      "learning_rate": 0.0005877193251770695,
      "loss": 2.3293,
      "step": 21057
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6475741863250732,
      "learning_rate": 0.0005877181667522016,
      "loss": 2.37,
      "step": 21058
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.5511094927787781,
      "learning_rate": 0.0005877170082738413,
      "loss": 2.2803,
      "step": 21059
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.7738000154495239,
      "learning_rate": 0.0005877158497419891,
      "loss": 2.4585,
      "step": 21060
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6511989235877991,
      "learning_rate": 0.0005877146911566449,
      "loss": 2.463,
      "step": 21061
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6004251837730408,
      "learning_rate": 0.0005877135325178093,
      "loss": 2.2765,
      "step": 21062
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.62924724817276,
      "learning_rate": 0.0005877123738254822,
      "loss": 2.3181,
      "step": 21063
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.614717423915863,
      "learning_rate": 0.000587711215079664,
      "loss": 2.266,
      "step": 21064
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.5689480304718018,
      "learning_rate": 0.0005877100562803549,
      "loss": 2.1118,
      "step": 21065
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6174150109291077,
      "learning_rate": 0.000587708897427555,
      "loss": 2.3274,
      "step": 21066
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6266536116600037,
      "learning_rate": 0.0005877077385212646,
      "loss": 2.2941,
      "step": 21067
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6239758729934692,
      "learning_rate": 0.0005877065795614839,
      "loss": 2.2117,
      "step": 21068
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.736261785030365,
      "learning_rate": 0.0005877054205482131,
      "loss": 2.568,
      "step": 21069
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6347513794898987,
      "learning_rate": 0.0005877042614814525,
      "loss": 2.5517,
      "step": 21070
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.5787646174430847,
      "learning_rate": 0.0005877031023612022,
      "loss": 2.3745,
      "step": 21071
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6270764470100403,
      "learning_rate": 0.0005877019431874624,
      "loss": 2.2696,
      "step": 21072
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6704222559928894,
      "learning_rate": 0.0005877007839602335,
      "loss": 2.4074,
      "step": 21073
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.5919007062911987,
      "learning_rate": 0.0005876996246795155,
      "loss": 2.3328,
      "step": 21074
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.62571120262146,
      "learning_rate": 0.0005876984653453087,
      "loss": 2.4046,
      "step": 21075
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.725395679473877,
      "learning_rate": 0.0005876973059576134,
      "loss": 2.2434,
      "step": 21076
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6605812907218933,
      "learning_rate": 0.0005876961465164297,
      "loss": 2.2592,
      "step": 21077
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6358982920646667,
      "learning_rate": 0.0005876949870217579,
      "loss": 2.2459,
      "step": 21078
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6171255707740784,
      "learning_rate": 0.0005876938274735981,
      "loss": 2.4327,
      "step": 21079
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.5917350649833679,
      "learning_rate": 0.0005876926678719506,
      "loss": 2.4665,
      "step": 21080
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.5635292530059814,
      "learning_rate": 0.0005876915082168156,
      "loss": 2.2686,
      "step": 21081
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.62172532081604,
      "learning_rate": 0.0005876903485081934,
      "loss": 2.3261,
      "step": 21082
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6547101140022278,
      "learning_rate": 0.000587689188746084,
      "loss": 2.1083,
      "step": 21083
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.5650318264961243,
      "learning_rate": 0.0005876880289304879,
      "loss": 2.451,
      "step": 21084
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6503364443778992,
      "learning_rate": 0.000587686869061405,
      "loss": 2.6021,
      "step": 21085
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6512081027030945,
      "learning_rate": 0.0005876857091388358,
      "loss": 2.3756,
      "step": 21086
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6311806440353394,
      "learning_rate": 0.0005876845491627803,
      "loss": 2.1726,
      "step": 21087
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6626993417739868,
      "learning_rate": 0.000587683389133239,
      "loss": 2.3467,
      "step": 21088
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.5673943161964417,
      "learning_rate": 0.0005876822290502117,
      "loss": 2.1348,
      "step": 21089
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.5980492830276489,
      "learning_rate": 0.000587681068913699,
      "loss": 2.2205,
      "step": 21090
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.5813028216362,
      "learning_rate": 0.0005876799087237009,
      "loss": 2.2851,
      "step": 21091
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.599861204624176,
      "learning_rate": 0.0005876787484802177,
      "loss": 2.404,
      "step": 21092
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.5878483057022095,
      "learning_rate": 0.0005876775881832495,
      "loss": 2.0801,
      "step": 21093
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6202650666236877,
      "learning_rate": 0.0005876764278327967,
      "loss": 2.5146,
      "step": 21094
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6350546479225159,
      "learning_rate": 0.0005876752674288594,
      "loss": 2.2812,
      "step": 21095
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.628555417060852,
      "learning_rate": 0.0005876741069714379,
      "loss": 2.5042,
      "step": 21096
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6082779765129089,
      "learning_rate": 0.0005876729464605323,
      "loss": 2.3053,
      "step": 21097
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.5772237777709961,
      "learning_rate": 0.0005876717858961428,
      "loss": 2.3907,
      "step": 21098
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.5872969031333923,
      "learning_rate": 0.0005876706252782697,
      "loss": 2.1736,
      "step": 21099
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6491249203681946,
      "learning_rate": 0.0005876694646069132,
      "loss": 2.3141,
      "step": 21100
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.5955985188484192,
      "learning_rate": 0.0005876683038820737,
      "loss": 2.4327,
      "step": 21101
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6324961185455322,
      "learning_rate": 0.000587667143103751,
      "loss": 2.4455,
      "step": 21102
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.5989647507667542,
      "learning_rate": 0.0005876659822719457,
      "loss": 2.2581,
      "step": 21103
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6428355574607849,
      "learning_rate": 0.0005876648213866578,
      "loss": 2.1688,
      "step": 21104
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6538336873054504,
      "learning_rate": 0.0005876636604478876,
      "loss": 2.2234,
      "step": 21105
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6078993678092957,
      "learning_rate": 0.0005876624994556352,
      "loss": 2.0926,
      "step": 21106
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.7159355282783508,
      "learning_rate": 0.000587661338409901,
      "loss": 2.3808,
      "step": 21107
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6179028749465942,
      "learning_rate": 0.0005876601773106852,
      "loss": 2.2814,
      "step": 21108
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.60878586769104,
      "learning_rate": 0.0005876590161579878,
      "loss": 2.2699,
      "step": 21109
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.7925174832344055,
      "learning_rate": 0.0005876578549518092,
      "loss": 2.3074,
      "step": 21110
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6320801973342896,
      "learning_rate": 0.0005876566936921496,
      "loss": 2.4306,
      "step": 21111
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6475780606269836,
      "learning_rate": 0.0005876555323790092,
      "loss": 2.3347,
      "step": 21112
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.5714463591575623,
      "learning_rate": 0.0005876543710123882,
      "loss": 2.2364,
      "step": 21113
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.7827404737472534,
      "learning_rate": 0.0005876532095922868,
      "loss": 2.3019,
      "step": 21114
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.617110013961792,
      "learning_rate": 0.0005876520481187052,
      "loss": 2.2675,
      "step": 21115
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.5961424708366394,
      "learning_rate": 0.0005876508865916437,
      "loss": 2.3743,
      "step": 21116
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.5663119554519653,
      "learning_rate": 0.0005876497250111026,
      "loss": 2.1412,
      "step": 21117
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6095224618911743,
      "learning_rate": 0.0005876485633770818,
      "loss": 2.2926,
      "step": 21118
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6402555108070374,
      "learning_rate": 0.0005876474016895817,
      "loss": 2.3057,
      "step": 21119
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6287232637405396,
      "learning_rate": 0.0005876462399486026,
      "loss": 2.532,
      "step": 21120
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.7021483778953552,
      "learning_rate": 0.0005876450781541446,
      "loss": 2.2228,
      "step": 21121
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.7040830850601196,
      "learning_rate": 0.000587643916306208,
      "loss": 2.2473,
      "step": 21122
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.633632481098175,
      "learning_rate": 0.000587642754404793,
      "loss": 2.2898,
      "step": 21123
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.5930676460266113,
      "learning_rate": 0.0005876415924498997,
      "loss": 2.4422,
      "step": 21124
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6772800087928772,
      "learning_rate": 0.0005876404304415284,
      "loss": 2.4246,
      "step": 21125
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.5506609678268433,
      "learning_rate": 0.0005876392683796793,
      "loss": 1.8083,
      "step": 21126
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.5990240573883057,
      "learning_rate": 0.0005876381062643527,
      "loss": 2.3226,
      "step": 21127
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6561887860298157,
      "learning_rate": 0.0005876369440955487,
      "loss": 2.2026,
      "step": 21128
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.593951940536499,
      "learning_rate": 0.0005876357818732675,
      "loss": 2.2054,
      "step": 21129
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6100112199783325,
      "learning_rate": 0.0005876346195975095,
      "loss": 2.4015,
      "step": 21130
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6443670392036438,
      "learning_rate": 0.0005876334572682747,
      "loss": 2.5266,
      "step": 21131
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6091685891151428,
      "learning_rate": 0.0005876322948855635,
      "loss": 2.4882,
      "step": 21132
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6497427225112915,
      "learning_rate": 0.000587631132449376,
      "loss": 2.1508,
      "step": 21133
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6924371123313904,
      "learning_rate": 0.0005876299699597124,
      "loss": 2.479,
      "step": 21134
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6081730127334595,
      "learning_rate": 0.0005876288074165729,
      "loss": 2.4707,
      "step": 21135
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.5326008200645447,
      "learning_rate": 0.0005876276448199579,
      "loss": 2.2085,
      "step": 21136
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6102328300476074,
      "learning_rate": 0.0005876264821698676,
      "loss": 2.4542,
      "step": 21137
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.5876702666282654,
      "learning_rate": 0.0005876253194663019,
      "loss": 2.338,
      "step": 21138
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6035782694816589,
      "learning_rate": 0.0005876241567092612,
      "loss": 2.1467,
      "step": 21139
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6546614766120911,
      "learning_rate": 0.0005876229938987458,
      "loss": 2.3293,
      "step": 21140
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.599855899810791,
      "learning_rate": 0.000587621831034756,
      "loss": 2.3828,
      "step": 21141
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.7819929122924805,
      "learning_rate": 0.0005876206681172916,
      "loss": 2.2441,
      "step": 21142
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.5944084525108337,
      "learning_rate": 0.0005876195051463533,
      "loss": 2.3785,
      "step": 21143
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6020196676254272,
      "learning_rate": 0.0005876183421219411,
      "loss": 2.4054,
      "step": 21144
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.5711384415626526,
      "learning_rate": 0.0005876171790440551,
      "loss": 2.3775,
      "step": 21145
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.614635169506073,
      "learning_rate": 0.0005876160159126956,
      "loss": 2.3997,
      "step": 21146
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6216229796409607,
      "learning_rate": 0.000587614852727863,
      "loss": 2.4634,
      "step": 21147
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6199110746383667,
      "learning_rate": 0.0005876136894895574,
      "loss": 2.6756,
      "step": 21148
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6381173729896545,
      "learning_rate": 0.0005876125261977789,
      "loss": 2.3338,
      "step": 21149
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.5814995169639587,
      "learning_rate": 0.0005876113628525278,
      "loss": 2.4629,
      "step": 21150
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.8282850980758667,
      "learning_rate": 0.0005876101994538043,
      "loss": 2.2438,
      "step": 21151
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.703434407711029,
      "learning_rate": 0.0005876090360016087,
      "loss": 2.2727,
      "step": 21152
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.679564356803894,
      "learning_rate": 0.0005876078724959411,
      "loss": 2.426,
      "step": 21153
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6289288997650146,
      "learning_rate": 0.0005876067089368018,
      "loss": 2.3197,
      "step": 21154
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.623363733291626,
      "learning_rate": 0.0005876055453241909,
      "loss": 2.4114,
      "step": 21155
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.587807834148407,
      "learning_rate": 0.0005876043816581088,
      "loss": 2.2432,
      "step": 21156
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6537626385688782,
      "learning_rate": 0.0005876032179385556,
      "loss": 2.3626,
      "step": 21157
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6452221870422363,
      "learning_rate": 0.0005876020541655315,
      "loss": 2.3285,
      "step": 21158
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.5913355946540833,
      "learning_rate": 0.0005876008903390367,
      "loss": 2.1417,
      "step": 21159
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6002015471458435,
      "learning_rate": 0.0005875997264590716,
      "loss": 2.2777,
      "step": 21160
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6562157273292542,
      "learning_rate": 0.0005875985625256361,
      "loss": 2.035,
      "step": 21161
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6081899404525757,
      "learning_rate": 0.0005875973985387308,
      "loss": 2.2167,
      "step": 21162
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6119465827941895,
      "learning_rate": 0.0005875962344983556,
      "loss": 2.4066,
      "step": 21163
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6624761819839478,
      "learning_rate": 0.0005875950704045109,
      "loss": 2.2994,
      "step": 21164
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.5989330410957336,
      "learning_rate": 0.0005875939062571969,
      "loss": 2.6228,
      "step": 21165
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.7197827696800232,
      "learning_rate": 0.0005875927420564136,
      "loss": 2.3191,
      "step": 21166
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6137412786483765,
      "learning_rate": 0.0005875915778021614,
      "loss": 2.4147,
      "step": 21167
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6637687683105469,
      "learning_rate": 0.0005875904134944406,
      "loss": 2.3464,
      "step": 21168
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6668174862861633,
      "learning_rate": 0.0005875892491332513,
      "loss": 2.3354,
      "step": 21169
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6116291284561157,
      "learning_rate": 0.0005875880847185937,
      "loss": 2.3229,
      "step": 21170
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6283291578292847,
      "learning_rate": 0.000587586920250468,
      "loss": 2.357,
      "step": 21171
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6767416596412659,
      "learning_rate": 0.0005875857557288744,
      "loss": 2.2738,
      "step": 21172
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6300908923149109,
      "learning_rate": 0.0005875845911538133,
      "loss": 2.4632,
      "step": 21173
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.7059652805328369,
      "learning_rate": 0.0005875834265252848,
      "loss": 2.4641,
      "step": 21174
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6633944511413574,
      "learning_rate": 0.000587582261843289,
      "loss": 2.428,
      "step": 21175
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6536203622817993,
      "learning_rate": 0.0005875810971078264,
      "loss": 2.4304,
      "step": 21176
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6132804155349731,
      "learning_rate": 0.000587579932318897,
      "loss": 2.2839,
      "step": 21177
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6108801364898682,
      "learning_rate": 0.000587578767476501,
      "loss": 2.2823,
      "step": 21178
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.5962348580360413,
      "learning_rate": 0.0005875776025806386,
      "loss": 2.423,
      "step": 21179
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.5898259282112122,
      "learning_rate": 0.0005875764376313103,
      "loss": 2.2593,
      "step": 21180
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6063387989997864,
      "learning_rate": 0.000587575272628516,
      "loss": 2.5471,
      "step": 21181
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6590233445167542,
      "learning_rate": 0.000587574107572256,
      "loss": 2.5267,
      "step": 21182
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6389722228050232,
      "learning_rate": 0.0005875729424625306,
      "loss": 2.2501,
      "step": 21183
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6145783066749573,
      "learning_rate": 0.00058757177729934,
      "loss": 2.3601,
      "step": 21184
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.7288950085639954,
      "learning_rate": 0.0005875706120826843,
      "loss": 2.4241,
      "step": 21185
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6975966691970825,
      "learning_rate": 0.0005875694468125638,
      "loss": 2.3371,
      "step": 21186
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.642970860004425,
      "learning_rate": 0.0005875682814889787,
      "loss": 2.3374,
      "step": 21187
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.7822847962379456,
      "learning_rate": 0.0005875671161119293,
      "loss": 2.2288,
      "step": 21188
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.5959429144859314,
      "learning_rate": 0.0005875659506814157,
      "loss": 2.3234,
      "step": 21189
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6334298253059387,
      "learning_rate": 0.0005875647851974382,
      "loss": 2.4126,
      "step": 21190
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6101486086845398,
      "learning_rate": 0.000587563619659997,
      "loss": 2.3929,
      "step": 21191
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6132434010505676,
      "learning_rate": 0.0005875624540690921,
      "loss": 2.3915,
      "step": 21192
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.66131991147995,
      "learning_rate": 0.0005875612884247241,
      "loss": 2.2415,
      "step": 21193
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.5916810035705566,
      "learning_rate": 0.000587560122726893,
      "loss": 2.2409,
      "step": 21194
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6150932312011719,
      "learning_rate": 0.000587558956975599,
      "loss": 2.1455,
      "step": 21195
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6787622570991516,
      "learning_rate": 0.0005875577911708424,
      "loss": 2.44,
      "step": 21196
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.584666907787323,
      "learning_rate": 0.0005875566253126233,
      "loss": 2.3046,
      "step": 21197
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6277652382850647,
      "learning_rate": 0.000587555459400942,
      "loss": 2.2875,
      "step": 21198
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.7471083402633667,
      "learning_rate": 0.0005875542934357989,
      "loss": 2.5633,
      "step": 21199
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6371135115623474,
      "learning_rate": 0.0005875531274171939,
      "loss": 2.3424,
      "step": 21200
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6290123462677002,
      "learning_rate": 0.0005875519613451273,
      "loss": 2.3965,
      "step": 21201
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6823283433914185,
      "learning_rate": 0.0005875507952195994,
      "loss": 2.1878,
      "step": 21202
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.7121450901031494,
      "learning_rate": 0.0005875496290406104,
      "loss": 2.489,
      "step": 21203
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.5935010313987732,
      "learning_rate": 0.0005875484628081605,
      "loss": 2.5025,
      "step": 21204
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6447870135307312,
      "learning_rate": 0.0005875472965222499,
      "loss": 2.3463,
      "step": 21205
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6578724980354309,
      "learning_rate": 0.0005875461301828788,
      "loss": 2.285,
      "step": 21206
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6571719646453857,
      "learning_rate": 0.0005875449637900474,
      "loss": 2.1148,
      "step": 21207
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6206710338592529,
      "learning_rate": 0.0005875437973437561,
      "loss": 2.3089,
      "step": 21208
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.64995938539505,
      "learning_rate": 0.0005875426308440049,
      "loss": 2.3137,
      "step": 21209
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6581335067749023,
      "learning_rate": 0.0005875414642907941,
      "loss": 2.4544,
      "step": 21210
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6027240753173828,
      "learning_rate": 0.000587540297684124,
      "loss": 2.2638,
      "step": 21211
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6786210536956787,
      "learning_rate": 0.0005875391310239946,
      "loss": 2.5695,
      "step": 21212
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6021767854690552,
      "learning_rate": 0.0005875379643104063,
      "loss": 2.4186,
      "step": 21213
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.5982218384742737,
      "learning_rate": 0.0005875367975433593,
      "loss": 2.3637,
      "step": 21214
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6311527490615845,
      "learning_rate": 0.0005875356307228538,
      "loss": 2.1608,
      "step": 21215
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6578230857849121,
      "learning_rate": 0.0005875344638488899,
      "loss": 2.5548,
      "step": 21216
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.7069328427314758,
      "learning_rate": 0.0005875332969214679,
      "loss": 2.3963,
      "step": 21217
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6108324527740479,
      "learning_rate": 0.0005875321299405881,
      "loss": 2.2914,
      "step": 21218
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6633799076080322,
      "learning_rate": 0.0005875309629062507,
      "loss": 2.3564,
      "step": 21219
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.8043368458747864,
      "learning_rate": 0.0005875297958184558,
      "loss": 2.2885,
      "step": 21220
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6204060316085815,
      "learning_rate": 0.0005875286286772036,
      "loss": 2.2155,
      "step": 21221
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6540666222572327,
      "learning_rate": 0.0005875274614824946,
      "loss": 2.142,
      "step": 21222
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6143677234649658,
      "learning_rate": 0.0005875262942343286,
      "loss": 2.3502,
      "step": 21223
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.7725840210914612,
      "learning_rate": 0.0005875251269327062,
      "loss": 2.599,
      "step": 21224
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.5822833180427551,
      "learning_rate": 0.0005875239595776275,
      "loss": 2.284,
      "step": 21225
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6543164849281311,
      "learning_rate": 0.0005875227921690925,
      "loss": 2.4082,
      "step": 21226
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.7396418452262878,
      "learning_rate": 0.0005875216247071016,
      "loss": 2.3931,
      "step": 21227
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.5660434365272522,
      "learning_rate": 0.0005875204571916551,
      "loss": 2.216,
      "step": 21228
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6342810988426208,
      "learning_rate": 0.0005875192896227531,
      "loss": 2.1342,
      "step": 21229
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6168568134307861,
      "learning_rate": 0.0005875181220003959,
      "loss": 2.2528,
      "step": 21230
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6459985375404358,
      "learning_rate": 0.0005875169543245835,
      "loss": 2.3566,
      "step": 21231
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6277165412902832,
      "learning_rate": 0.0005875157865953163,
      "loss": 2.325,
      "step": 21232
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6038707494735718,
      "learning_rate": 0.0005875146188125946,
      "loss": 2.1708,
      "step": 21233
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6434051394462585,
      "learning_rate": 0.0005875134509764184,
      "loss": 2.5432,
      "step": 21234
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6798257231712341,
      "learning_rate": 0.0005875122830867881,
      "loss": 2.1788,
      "step": 21235
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6609283685684204,
      "learning_rate": 0.0005875111151437037,
      "loss": 2.7271,
      "step": 21236
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6003280282020569,
      "learning_rate": 0.0005875099471471657,
      "loss": 2.4108,
      "step": 21237
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6162393689155579,
      "learning_rate": 0.0005875087790971741,
      "loss": 2.5369,
      "step": 21238
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6198207139968872,
      "learning_rate": 0.0005875076109937293,
      "loss": 2.3405,
      "step": 21239
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6100420951843262,
      "learning_rate": 0.0005875064428368312,
      "loss": 2.2724,
      "step": 21240
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.7011069655418396,
      "learning_rate": 0.0005875052746264804,
      "loss": 2.1149,
      "step": 21241
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6189225912094116,
      "learning_rate": 0.0005875041063626768,
      "loss": 2.2533,
      "step": 21242
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6698734164237976,
      "learning_rate": 0.0005875029380454209,
      "loss": 2.5509,
      "step": 21243
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6058200001716614,
      "learning_rate": 0.0005875017696747128,
      "loss": 2.3258,
      "step": 21244
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6003767848014832,
      "learning_rate": 0.0005875006012505525,
      "loss": 2.2888,
      "step": 21245
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.8591356873512268,
      "learning_rate": 0.0005874994327729406,
      "loss": 2.4363,
      "step": 21246
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.635265588760376,
      "learning_rate": 0.0005874982642418771,
      "loss": 2.2965,
      "step": 21247
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6146875023841858,
      "learning_rate": 0.0005874970956573621,
      "loss": 2.4035,
      "step": 21248
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6544952988624573,
      "learning_rate": 0.0005874959270193961,
      "loss": 2.2709,
      "step": 21249
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6488986611366272,
      "learning_rate": 0.0005874947583279791,
      "loss": 2.3814,
      "step": 21250
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6232735514640808,
      "learning_rate": 0.0005874935895831115,
      "loss": 2.2075,
      "step": 21251
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6774372458457947,
      "learning_rate": 0.0005874924207847933,
      "loss": 2.4278,
      "step": 21252
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.7007415890693665,
      "learning_rate": 0.000587491251933025,
      "loss": 2.5399,
      "step": 21253
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6644342541694641,
      "learning_rate": 0.0005874900830278065,
      "loss": 2.3311,
      "step": 21254
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.5779931545257568,
      "learning_rate": 0.0005874889140691382,
      "loss": 2.6016,
      "step": 21255
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.584678053855896,
      "learning_rate": 0.0005874877450570203,
      "loss": 2.2696,
      "step": 21256
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6138927340507507,
      "learning_rate": 0.000587486575991453,
      "loss": 2.5935,
      "step": 21257
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6259078979492188,
      "learning_rate": 0.0005874854068724365,
      "loss": 2.4393,
      "step": 21258
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6739187240600586,
      "learning_rate": 0.000587484237699971,
      "loss": 2.394,
      "step": 21259
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.5944294333457947,
      "learning_rate": 0.0005874830684740568,
      "loss": 2.3311,
      "step": 21260
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.5634474754333496,
      "learning_rate": 0.0005874818991946941,
      "loss": 2.393,
      "step": 21261
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.5800895094871521,
      "learning_rate": 0.0005874807298618831,
      "loss": 2.3455,
      "step": 21262
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.5734137296676636,
      "learning_rate": 0.0005874795604756239,
      "loss": 2.1221,
      "step": 21263
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6929906010627747,
      "learning_rate": 0.000587478391035917,
      "loss": 2.1458,
      "step": 21264
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.5827476382255554,
      "learning_rate": 0.0005874772215427622,
      "loss": 2.4432,
      "step": 21265
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.631813645362854,
      "learning_rate": 0.0005874760519961602,
      "loss": 2.5082,
      "step": 21266
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.5898799896240234,
      "learning_rate": 0.0005874748823961108,
      "loss": 2.404,
      "step": 21267
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6787263751029968,
      "learning_rate": 0.0005874737127426145,
      "loss": 2.4704,
      "step": 21268
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.5873774290084839,
      "learning_rate": 0.0005874725430356714,
      "loss": 2.3281,
      "step": 21269
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6289363503456116,
      "learning_rate": 0.0005874713732752817,
      "loss": 2.4303,
      "step": 21270
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6575683951377869,
      "learning_rate": 0.0005874702034614456,
      "loss": 1.9391,
      "step": 21271
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.607986330986023,
      "learning_rate": 0.0005874690335941633,
      "loss": 2.2473,
      "step": 21272
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.7162503004074097,
      "learning_rate": 0.0005874678636734353,
      "loss": 2.3038,
      "step": 21273
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.5853670239448547,
      "learning_rate": 0.0005874666936992615,
      "loss": 2.3593,
      "step": 21274
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.5908024311065674,
      "learning_rate": 0.0005874655236716422,
      "loss": 2.3235,
      "step": 21275
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6569535732269287,
      "learning_rate": 0.0005874643535905776,
      "loss": 2.5749,
      "step": 21276
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6599689722061157,
      "learning_rate": 0.0005874631834560679,
      "loss": 2.7423,
      "step": 21277
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6174258589744568,
      "learning_rate": 0.0005874620132681136,
      "loss": 2.4851,
      "step": 21278
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6773346662521362,
      "learning_rate": 0.0005874608430267145,
      "loss": 2.1879,
      "step": 21279
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6105896234512329,
      "learning_rate": 0.000587459672731871,
      "loss": 2.1864,
      "step": 21280
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6043210625648499,
      "learning_rate": 0.0005874585023835834,
      "loss": 2.2393,
      "step": 21281
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6544267535209656,
      "learning_rate": 0.0005874573319818517,
      "loss": 2.192,
      "step": 21282
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6039109826087952,
      "learning_rate": 0.0005874561615266765,
      "loss": 2.2737,
      "step": 21283
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.5981627702713013,
      "learning_rate": 0.0005874549910180576,
      "loss": 2.5654,
      "step": 21284
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6180279850959778,
      "learning_rate": 0.0005874538204559954,
      "loss": 2.4216,
      "step": 21285
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.664864182472229,
      "learning_rate": 0.00058745264984049,
      "loss": 2.2038,
      "step": 21286
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.7013056874275208,
      "learning_rate": 0.0005874514791715419,
      "loss": 2.4756,
      "step": 21287
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.2257636785507202,
      "learning_rate": 0.0005874503084491511,
      "loss": 2.5589,
      "step": 21288
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.7077783942222595,
      "learning_rate": 0.0005874491376733178,
      "loss": 2.3753,
      "step": 21289
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6086981296539307,
      "learning_rate": 0.0005874479668440424,
      "loss": 2.4012,
      "step": 21290
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.5767700672149658,
      "learning_rate": 0.0005874467959613249,
      "loss": 2.1402,
      "step": 21291
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.8079380393028259,
      "learning_rate": 0.0005874456250251657,
      "loss": 2.405,
      "step": 21292
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6827789545059204,
      "learning_rate": 0.0005874444540355649,
      "loss": 2.1617,
      "step": 21293
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6133137345314026,
      "learning_rate": 0.0005874432829925227,
      "loss": 2.3856,
      "step": 21294
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6475778222084045,
      "learning_rate": 0.0005874421118960394,
      "loss": 2.5411,
      "step": 21295
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6606371402740479,
      "learning_rate": 0.0005874409407461152,
      "loss": 2.2753,
      "step": 21296
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6701843738555908,
      "learning_rate": 0.0005874397695427502,
      "loss": 2.4567,
      "step": 21297
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.5662627220153809,
      "learning_rate": 0.0005874385982859449,
      "loss": 2.3451,
      "step": 21298
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.5945048332214355,
      "learning_rate": 0.0005874374269756992,
      "loss": 2.2545,
      "step": 21299
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6414592266082764,
      "learning_rate": 0.0005874362556120135,
      "loss": 2.2096,
      "step": 21300
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6209652423858643,
      "learning_rate": 0.0005874350841948879,
      "loss": 2.4388,
      "step": 21301
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6288567781448364,
      "learning_rate": 0.0005874339127243228,
      "loss": 2.2214,
      "step": 21302
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.5880723595619202,
      "learning_rate": 0.0005874327412003184,
      "loss": 2.3905,
      "step": 21303
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6385050415992737,
      "learning_rate": 0.0005874315696228747,
      "loss": 2.4819,
      "step": 21304
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6396379470825195,
      "learning_rate": 0.000587430397991992,
      "loss": 2.3602,
      "step": 21305
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.5951237678527832,
      "learning_rate": 0.0005874292263076706,
      "loss": 2.3946,
      "step": 21306
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6536090970039368,
      "learning_rate": 0.0005874280545699107,
      "loss": 2.4501,
      "step": 21307
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6067218780517578,
      "learning_rate": 0.0005874268827787125,
      "loss": 2.7205,
      "step": 21308
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6101948022842407,
      "learning_rate": 0.0005874257109340762,
      "loss": 2.365,
      "step": 21309
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6861656308174133,
      "learning_rate": 0.000587424539036002,
      "loss": 2.4376,
      "step": 21310
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.639784038066864,
      "learning_rate": 0.0005874233670844903,
      "loss": 2.082,
      "step": 21311
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.5924512147903442,
      "learning_rate": 0.000587422195079541,
      "loss": 1.9605,
      "step": 21312
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.635915994644165,
      "learning_rate": 0.0005874210230211546,
      "loss": 2.443,
      "step": 21313
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6070748567581177,
      "learning_rate": 0.0005874198509093311,
      "loss": 2.4243,
      "step": 21314
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6163251996040344,
      "learning_rate": 0.000587418678744071,
      "loss": 2.4616,
      "step": 21315
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6112297773361206,
      "learning_rate": 0.0005874175065253742,
      "loss": 2.1894,
      "step": 21316
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6048494577407837,
      "learning_rate": 0.0005874163342532411,
      "loss": 2.3687,
      "step": 21317
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.617136538028717,
      "learning_rate": 0.0005874151619276719,
      "loss": 2.3686,
      "step": 21318
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6244665384292603,
      "learning_rate": 0.0005874139895486667,
      "loss": 2.2721,
      "step": 21319
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6124871373176575,
      "learning_rate": 0.0005874128171162259,
      "loss": 2.2522,
      "step": 21320
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.617568850517273,
      "learning_rate": 0.0005874116446303496,
      "loss": 2.1753,
      "step": 21321
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6430547833442688,
      "learning_rate": 0.000587410472091038,
      "loss": 2.6116,
      "step": 21322
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6410645842552185,
      "learning_rate": 0.0005874092994982915,
      "loss": 2.1916,
      "step": 21323
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6353652477264404,
      "learning_rate": 0.0005874081268521101,
      "loss": 2.3809,
      "step": 21324
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6015602350234985,
      "learning_rate": 0.0005874069541524942,
      "loss": 2.4559,
      "step": 21325
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.8084766268730164,
      "learning_rate": 0.0005874057813994437,
      "loss": 2.1952,
      "step": 21326
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6874851584434509,
      "learning_rate": 0.0005874046085929593,
      "loss": 2.4427,
      "step": 21327
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6769680380821228,
      "learning_rate": 0.0005874034357330407,
      "loss": 2.4889,
      "step": 21328
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6603610515594482,
      "learning_rate": 0.0005874022628196885,
      "loss": 2.3226,
      "step": 21329
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.5699657797813416,
      "learning_rate": 0.0005874010898529029,
      "loss": 2.0747,
      "step": 21330
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.5535205602645874,
      "learning_rate": 0.0005873999168326839,
      "loss": 2.2397,
      "step": 21331
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6614715456962585,
      "learning_rate": 0.0005873987437590318,
      "loss": 2.3519,
      "step": 21332
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.7092761993408203,
      "learning_rate": 0.000587397570631947,
      "loss": 2.1945,
      "step": 21333
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.5736756324768066,
      "learning_rate": 0.0005873963974514295,
      "loss": 2.197,
      "step": 21334
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6647040843963623,
      "learning_rate": 0.0005873952242174795,
      "loss": 2.1936,
      "step": 21335
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6630143523216248,
      "learning_rate": 0.0005873940509300974,
      "loss": 2.4461,
      "step": 21336
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.599216639995575,
      "learning_rate": 0.0005873928775892832,
      "loss": 2.2025,
      "step": 21337
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.608364999294281,
      "learning_rate": 0.0005873917041950374,
      "loss": 2.4697,
      "step": 21338
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.5938664078712463,
      "learning_rate": 0.0005873905307473599,
      "loss": 2.5277,
      "step": 21339
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6315932273864746,
      "learning_rate": 0.0005873893572462511,
      "loss": 2.3554,
      "step": 21340
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6068268418312073,
      "learning_rate": 0.0005873881836917113,
      "loss": 2.0457,
      "step": 21341
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6046348214149475,
      "learning_rate": 0.0005873870100837406,
      "loss": 2.2535,
      "step": 21342
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6557561755180359,
      "learning_rate": 0.0005873858364223392,
      "loss": 2.2042,
      "step": 21343
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.7588624954223633,
      "learning_rate": 0.0005873846627075073,
      "loss": 2.2764,
      "step": 21344
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6311317682266235,
      "learning_rate": 0.0005873834889392452,
      "loss": 2.4808,
      "step": 21345
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6250712871551514,
      "learning_rate": 0.000587382315117553,
      "loss": 2.5354,
      "step": 21346
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6443333625793457,
      "learning_rate": 0.000587381141242431,
      "loss": 2.2575,
      "step": 21347
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6124696135520935,
      "learning_rate": 0.0005873799673138797,
      "loss": 2.2765,
      "step": 21348
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6232799291610718,
      "learning_rate": 0.0005873787933318988,
      "loss": 2.2353,
      "step": 21349
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6667383313179016,
      "learning_rate": 0.0005873776192964887,
      "loss": 2.272,
      "step": 21350
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6903325319290161,
      "learning_rate": 0.0005873764452076498,
      "loss": 2.1008,
      "step": 21351
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6289148330688477,
      "learning_rate": 0.0005873752710653823,
      "loss": 2.3428,
      "step": 21352
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.7706542611122131,
      "learning_rate": 0.0005873740968696861,
      "loss": 2.3684,
      "step": 21353
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6516255736351013,
      "learning_rate": 0.0005873729226205617,
      "loss": 2.361,
      "step": 21354
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6707762479782104,
      "learning_rate": 0.0005873717483180093,
      "loss": 2.4208,
      "step": 21355
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.624203622341156,
      "learning_rate": 0.0005873705739620291,
      "loss": 2.2122,
      "step": 21356
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.7397857904434204,
      "learning_rate": 0.0005873693995526212,
      "loss": 2.302,
      "step": 21357
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.5799313187599182,
      "learning_rate": 0.0005873682250897859,
      "loss": 1.9002,
      "step": 21358
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6993218660354614,
      "learning_rate": 0.0005873670505735234,
      "loss": 2.2784,
      "step": 21359
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.8969380259513855,
      "learning_rate": 0.0005873658760038341,
      "loss": 2.433,
      "step": 21360
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6070532202720642,
      "learning_rate": 0.0005873647013807179,
      "loss": 2.385,
      "step": 21361
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.9068513512611389,
      "learning_rate": 0.0005873635267041752,
      "loss": 2.3792,
      "step": 21362
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.7378866076469421,
      "learning_rate": 0.0005873623519742062,
      "loss": 2.6332,
      "step": 21363
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6257916688919067,
      "learning_rate": 0.0005873611771908111,
      "loss": 2.1955,
      "step": 21364
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.683000385761261,
      "learning_rate": 0.0005873600023539903,
      "loss": 2.4711,
      "step": 21365
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6602396368980408,
      "learning_rate": 0.0005873588274637436,
      "loss": 2.3956,
      "step": 21366
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.7865193486213684,
      "learning_rate": 0.0005873576525200716,
      "loss": 2.2002,
      "step": 21367
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.7122863531112671,
      "learning_rate": 0.0005873564775229743,
      "loss": 2.4545,
      "step": 21368
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.983879566192627,
      "learning_rate": 0.0005873553024724522,
      "loss": 2.3756,
      "step": 21369
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6142573356628418,
      "learning_rate": 0.0005873541273685052,
      "loss": 2.329,
      "step": 21370
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.5997785329818726,
      "learning_rate": 0.0005873529522111336,
      "loss": 2.4021,
      "step": 21371
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6970247626304626,
      "learning_rate": 0.0005873517770003377,
      "loss": 2.5048,
      "step": 21372
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.7170687317848206,
      "learning_rate": 0.0005873506017361178,
      "loss": 2.3623,
      "step": 21373
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.666259229183197,
      "learning_rate": 0.0005873494264184738,
      "loss": 2.6321,
      "step": 21374
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.650285542011261,
      "learning_rate": 0.0005873482510474063,
      "loss": 2.3651,
      "step": 21375
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.7126390933990479,
      "learning_rate": 0.0005873470756229152,
      "loss": 2.4135,
      "step": 21376
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6511712074279785,
      "learning_rate": 0.0005873459001450008,
      "loss": 2.2287,
      "step": 21377
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6556635499000549,
      "learning_rate": 0.0005873447246136635,
      "loss": 2.4318,
      "step": 21378
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.7074099779129028,
      "learning_rate": 0.0005873435490289034,
      "loss": 2.3521,
      "step": 21379
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.0254851579666138,
      "learning_rate": 0.0005873423733907207,
      "loss": 2.3255,
      "step": 21380
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6504067182540894,
      "learning_rate": 0.0005873411976991157,
      "loss": 2.5352,
      "step": 21381
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6996719241142273,
      "learning_rate": 0.0005873400219540883,
      "loss": 2.3509,
      "step": 21382
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6889462471008301,
      "learning_rate": 0.0005873388461556391,
      "loss": 2.4593,
      "step": 21383
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6977318525314331,
      "learning_rate": 0.0005873376703037682,
      "loss": 2.3121,
      "step": 21384
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6165215969085693,
      "learning_rate": 0.0005873364943984759,
      "loss": 2.3132,
      "step": 21385
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.7680442333221436,
      "learning_rate": 0.0005873353184397621,
      "loss": 2.2568,
      "step": 21386
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6334434151649475,
      "learning_rate": 0.0005873341424276274,
      "loss": 2.4367,
      "step": 21387
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6308262348175049,
      "learning_rate": 0.0005873329663620719,
      "loss": 2.5909,
      "step": 21388
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6352701187133789,
      "learning_rate": 0.0005873317902430958,
      "loss": 2.3433,
      "step": 21389
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6848891973495483,
      "learning_rate": 0.0005873306140706991,
      "loss": 2.41,
      "step": 21390
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6410877108573914,
      "learning_rate": 0.0005873294378448822,
      "loss": 2.259,
      "step": 21391
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.5804505944252014,
      "learning_rate": 0.0005873282615656456,
      "loss": 2.259,
      "step": 21392
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6144751310348511,
      "learning_rate": 0.0005873270852329891,
      "loss": 2.2913,
      "step": 21393
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.7204684019088745,
      "learning_rate": 0.000587325908846913,
      "loss": 2.1429,
      "step": 21394
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6375719904899597,
      "learning_rate": 0.0005873247324074177,
      "loss": 2.4627,
      "step": 21395
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.5735428333282471,
      "learning_rate": 0.0005873235559145033,
      "loss": 2.1915,
      "step": 21396
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.777070939540863,
      "learning_rate": 0.00058732237936817,
      "loss": 2.3657,
      "step": 21397
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6319011449813843,
      "learning_rate": 0.000587321202768418,
      "loss": 2.5264,
      "step": 21398
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6024858951568604,
      "learning_rate": 0.0005873200261152476,
      "loss": 2.4886,
      "step": 21399
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.7280905842781067,
      "learning_rate": 0.000587318849408659,
      "loss": 2.4937,
      "step": 21400
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.7770252227783203,
      "learning_rate": 0.0005873176726486523,
      "loss": 2.0822,
      "step": 21401
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6857087016105652,
      "learning_rate": 0.000587316495835228,
      "loss": 2.4722,
      "step": 21402
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.7458872199058533,
      "learning_rate": 0.000587315318968386,
      "loss": 2.3186,
      "step": 21403
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.5996968150138855,
      "learning_rate": 0.0005873141420481268,
      "loss": 2.3267,
      "step": 21404
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.5761843919754028,
      "learning_rate": 0.0005873129650744504,
      "loss": 2.3886,
      "step": 21405
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6735930442810059,
      "learning_rate": 0.000587311788047357,
      "loss": 2.5112,
      "step": 21406
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.7126277089118958,
      "learning_rate": 0.0005873106109668471,
      "loss": 2.3837,
      "step": 21407
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.5915688872337341,
      "learning_rate": 0.0005873094338329206,
      "loss": 2.3905,
      "step": 21408
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.655836820602417,
      "learning_rate": 0.0005873082566455779,
      "loss": 2.1786,
      "step": 21409
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6785680055618286,
      "learning_rate": 0.0005873070794048192,
      "loss": 2.3537,
      "step": 21410
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.598298966884613,
      "learning_rate": 0.0005873059021106446,
      "loss": 2.3787,
      "step": 21411
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6798186898231506,
      "learning_rate": 0.0005873047247630544,
      "loss": 2.4382,
      "step": 21412
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6064943075180054,
      "learning_rate": 0.0005873035473620489,
      "loss": 2.4811,
      "step": 21413
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6581460237503052,
      "learning_rate": 0.0005873023699076283,
      "loss": 2.2705,
      "step": 21414
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.8132840394973755,
      "learning_rate": 0.0005873011923997926,
      "loss": 2.2463,
      "step": 21415
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6869968771934509,
      "learning_rate": 0.0005873000148385424,
      "loss": 2.136,
      "step": 21416
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6189373731613159,
      "learning_rate": 0.0005872988372238776,
      "loss": 2.3266,
      "step": 21417
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6078449487686157,
      "learning_rate": 0.0005872976595557985,
      "loss": 2.3648,
      "step": 21418
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6236634254455566,
      "learning_rate": 0.0005872964818343054,
      "loss": 2.3552,
      "step": 21419
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.7640326619148254,
      "learning_rate": 0.0005872953040593984,
      "loss": 2.3063,
      "step": 21420
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6264318823814392,
      "learning_rate": 0.0005872941262310778,
      "loss": 2.4876,
      "step": 21421
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6008245944976807,
      "learning_rate": 0.0005872929483493438,
      "loss": 2.2545,
      "step": 21422
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6224733591079712,
      "learning_rate": 0.0005872917704141966,
      "loss": 2.3081,
      "step": 21423
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6852133870124817,
      "learning_rate": 0.0005872905924256364,
      "loss": 2.4433,
      "step": 21424
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.5621989369392395,
      "learning_rate": 0.0005872894143836635,
      "loss": 2.1686,
      "step": 21425
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.5741779804229736,
      "learning_rate": 0.000587288236288278,
      "loss": 2.3088,
      "step": 21426
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.5991063714027405,
      "learning_rate": 0.0005872870581394803,
      "loss": 2.315,
      "step": 21427
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.667508065700531,
      "learning_rate": 0.0005872858799372706,
      "loss": 2.1797,
      "step": 21428
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.5754227638244629,
      "learning_rate": 0.0005872847016816489,
      "loss": 2.2883,
      "step": 21429
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.5745160579681396,
      "learning_rate": 0.0005872835233726155,
      "loss": 2.4107,
      "step": 21430
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.7538194060325623,
      "learning_rate": 0.0005872823450101709,
      "loss": 2.4793,
      "step": 21431
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.7405744791030884,
      "learning_rate": 0.0005872811665943149,
      "loss": 2.4547,
      "step": 21432
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6025179028511047,
      "learning_rate": 0.0005872799881250479,
      "loss": 2.4953,
      "step": 21433
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.7139994502067566,
      "learning_rate": 0.0005872788096023702,
      "loss": 2.1723,
      "step": 21434
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.7052307724952698,
      "learning_rate": 0.0005872776310262819,
      "loss": 2.3082,
      "step": 21435
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6387046575546265,
      "learning_rate": 0.0005872764523967832,
      "loss": 2.243,
      "step": 21436
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6950941681861877,
      "learning_rate": 0.0005872752737138744,
      "loss": 2.1714,
      "step": 21437
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.7412753701210022,
      "learning_rate": 0.0005872740949775559,
      "loss": 2.3838,
      "step": 21438
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.7273872494697571,
      "learning_rate": 0.0005872729161878275,
      "loss": 2.4898,
      "step": 21439
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.5969626903533936,
      "learning_rate": 0.0005872717373446897,
      "loss": 2.242,
      "step": 21440
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6459525227546692,
      "learning_rate": 0.0005872705584481429,
      "loss": 2.4178,
      "step": 21441
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.7532056570053101,
      "learning_rate": 0.0005872693794981869,
      "loss": 2.5508,
      "step": 21442
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.7693567276000977,
      "learning_rate": 0.000587268200494822,
      "loss": 2.3839,
      "step": 21443
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.683840811252594,
      "learning_rate": 0.0005872670214380486,
      "loss": 2.4563,
      "step": 21444
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.7407904267311096,
      "learning_rate": 0.0005872658423278669,
      "loss": 2.2232,
      "step": 21445
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.7562825083732605,
      "learning_rate": 0.0005872646631642769,
      "loss": 2.5023,
      "step": 21446
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.7147536277770996,
      "learning_rate": 0.000587263483947279,
      "loss": 2.3816,
      "step": 21447
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6525028944015503,
      "learning_rate": 0.0005872623046768735,
      "loss": 2.2176,
      "step": 21448
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.8488350510597229,
      "learning_rate": 0.0005872611253530605,
      "loss": 2.3013,
      "step": 21449
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.7149558663368225,
      "learning_rate": 0.0005872599459758402,
      "loss": 2.3659,
      "step": 21450
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6044089794158936,
      "learning_rate": 0.0005872587665452128,
      "loss": 2.4072,
      "step": 21451
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.812335193157196,
      "learning_rate": 0.0005872575870611787,
      "loss": 2.2897,
      "step": 21452
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.607536792755127,
      "learning_rate": 0.0005872564075237379,
      "loss": 2.2934,
      "step": 21453
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.5959620475769043,
      "learning_rate": 0.0005872552279328908,
      "loss": 2.4054,
      "step": 21454
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6068947315216064,
      "learning_rate": 0.0005872540482886375,
      "loss": 2.5567,
      "step": 21455
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.697727620601654,
      "learning_rate": 0.0005872528685909781,
      "loss": 2.2488,
      "step": 21456
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6503592729568481,
      "learning_rate": 0.0005872516888399132,
      "loss": 2.2063,
      "step": 21457
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6545949578285217,
      "learning_rate": 0.0005872505090354427,
      "loss": 2.2783,
      "step": 21458
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6545396447181702,
      "learning_rate": 0.0005872493291775668,
      "loss": 2.2548,
      "step": 21459
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.1559104919433594,
      "learning_rate": 0.0005872481492662859,
      "loss": 2.4683,
      "step": 21460
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6408923864364624,
      "learning_rate": 0.0005872469693016003,
      "loss": 2.4849,
      "step": 21461
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.647364616394043,
      "learning_rate": 0.0005872457892835098,
      "loss": 2.3297,
      "step": 21462
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.5988074541091919,
      "learning_rate": 0.000587244609212015,
      "loss": 2.2125,
      "step": 21463
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.7173283100128174,
      "learning_rate": 0.0005872434290871161,
      "loss": 2.4778,
      "step": 21464
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6182786822319031,
      "learning_rate": 0.0005872422489088131,
      "loss": 2.1145,
      "step": 21465
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6151959300041199,
      "learning_rate": 0.0005872410686771064,
      "loss": 2.1572,
      "step": 21466
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6518675684928894,
      "learning_rate": 0.0005872398883919961,
      "loss": 2.3703,
      "step": 21467
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6397592425346375,
      "learning_rate": 0.0005872387080534826,
      "loss": 2.4178,
      "step": 21468
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.625516951084137,
      "learning_rate": 0.0005872375276615659,
      "loss": 2.3859,
      "step": 21469
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.5955031514167786,
      "learning_rate": 0.0005872363472162464,
      "loss": 2.4048,
      "step": 21470
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.5703365802764893,
      "learning_rate": 0.0005872351667175242,
      "loss": 2.4066,
      "step": 21471
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6027001738548279,
      "learning_rate": 0.0005872339861653995,
      "loss": 2.4391,
      "step": 21472
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.559123694896698,
      "learning_rate": 0.0005872328055598726,
      "loss": 2.1983,
      "step": 21473
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.60848468542099,
      "learning_rate": 0.0005872316249009438,
      "loss": 2.3911,
      "step": 21474
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6635375022888184,
      "learning_rate": 0.0005872304441886132,
      "loss": 2.4471,
      "step": 21475
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6012917757034302,
      "learning_rate": 0.0005872292634228809,
      "loss": 2.4581,
      "step": 21476
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6210851669311523,
      "learning_rate": 0.0005872280826037474,
      "loss": 2.4688,
      "step": 21477
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6967567205429077,
      "learning_rate": 0.0005872269017312127,
      "loss": 2.5748,
      "step": 21478
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.5907238125801086,
      "learning_rate": 0.0005872257208052771,
      "loss": 2.3666,
      "step": 21479
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6213423609733582,
      "learning_rate": 0.0005872245398259409,
      "loss": 2.1307,
      "step": 21480
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.7216904759407043,
      "learning_rate": 0.0005872233587932042,
      "loss": 2.4109,
      "step": 21481
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.5749866962432861,
      "learning_rate": 0.0005872221777070673,
      "loss": 2.3297,
      "step": 21482
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.5948203802108765,
      "learning_rate": 0.0005872209965675303,
      "loss": 2.3912,
      "step": 21483
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6303343176841736,
      "learning_rate": 0.0005872198153745934,
      "loss": 2.4046,
      "step": 21484
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6207962036132812,
      "learning_rate": 0.0005872186341282571,
      "loss": 2.5355,
      "step": 21485
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.668601930141449,
      "learning_rate": 0.0005872174528285213,
      "loss": 2.3619,
      "step": 21486
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6319615244865417,
      "learning_rate": 0.0005872162714753864,
      "loss": 2.3108,
      "step": 21487
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6738702058792114,
      "learning_rate": 0.0005872150900688526,
      "loss": 2.6194,
      "step": 21488
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.5584213137626648,
      "learning_rate": 0.00058721390860892,
      "loss": 2.2934,
      "step": 21489
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6478174924850464,
      "learning_rate": 0.0005872127270955891,
      "loss": 2.1953,
      "step": 21490
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.5972542762756348,
      "learning_rate": 0.0005872115455288597,
      "loss": 2.3459,
      "step": 21491
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6060391664505005,
      "learning_rate": 0.0005872103639087325,
      "loss": 2.3662,
      "step": 21492
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.597572386264801,
      "learning_rate": 0.0005872091822352074,
      "loss": 2.1717,
      "step": 21493
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.5811884999275208,
      "learning_rate": 0.0005872080005082846,
      "loss": 2.1096,
      "step": 21494
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.5942128896713257,
      "learning_rate": 0.0005872068187279644,
      "loss": 2.4229,
      "step": 21495
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6183183193206787,
      "learning_rate": 0.0005872056368942471,
      "loss": 2.1966,
      "step": 21496
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6215829253196716,
      "learning_rate": 0.0005872044550071328,
      "loss": 2.2529,
      "step": 21497
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.8394485116004944,
      "learning_rate": 0.0005872032730666218,
      "loss": 2.3702,
      "step": 21498
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.5944265127182007,
      "learning_rate": 0.0005872020910727144,
      "loss": 2.3638,
      "step": 21499
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6908746957778931,
      "learning_rate": 0.0005872009090254106,
      "loss": 2.2422,
      "step": 21500
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.5960770845413208,
      "learning_rate": 0.0005871997269247107,
      "loss": 1.9949,
      "step": 21501
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.7490776777267456,
      "learning_rate": 0.0005871985447706149,
      "loss": 2.2588,
      "step": 21502
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6218324303627014,
      "learning_rate": 0.0005871973625631236,
      "loss": 2.2696,
      "step": 21503
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6186602711677551,
      "learning_rate": 0.0005871961803022369,
      "loss": 2.7986,
      "step": 21504
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6053215861320496,
      "learning_rate": 0.000587194997987955,
      "loss": 2.1506,
      "step": 21505
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.5876561999320984,
      "learning_rate": 0.000587193815620278,
      "loss": 2.1468,
      "step": 21506
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.615365743637085,
      "learning_rate": 0.0005871926331992063,
      "loss": 2.5631,
      "step": 21507
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6244293451309204,
      "learning_rate": 0.0005871914507247401,
      "loss": 2.284,
      "step": 21508
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6299985647201538,
      "learning_rate": 0.0005871902681968795,
      "loss": 2.6414,
      "step": 21509
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6819565892219543,
      "learning_rate": 0.000587189085615625,
      "loss": 2.6593,
      "step": 21510
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6444952487945557,
      "learning_rate": 0.0005871879029809764,
      "loss": 2.6176,
      "step": 21511
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6147811412811279,
      "learning_rate": 0.0005871867202929344,
      "loss": 2.2694,
      "step": 21512
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6353753805160522,
      "learning_rate": 0.0005871855375514988,
      "loss": 2.3653,
      "step": 21513
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6191346645355225,
      "learning_rate": 0.0005871843547566699,
      "loss": 2.3247,
      "step": 21514
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6066220998764038,
      "learning_rate": 0.0005871831719084482,
      "loss": 2.2772,
      "step": 21515
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6216627955436707,
      "learning_rate": 0.0005871819890068337,
      "loss": 2.2533,
      "step": 21516
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6669723391532898,
      "learning_rate": 0.0005871808060518266,
      "loss": 2.4112,
      "step": 21517
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6484671831130981,
      "learning_rate": 0.0005871796230434271,
      "loss": 2.277,
      "step": 21518
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6063224673271179,
      "learning_rate": 0.0005871784399816356,
      "loss": 2.4026,
      "step": 21519
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6141124367713928,
      "learning_rate": 0.0005871772568664522,
      "loss": 2.2733,
      "step": 21520
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.677689790725708,
      "learning_rate": 0.000587176073697877,
      "loss": 2.6425,
      "step": 21521
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.5890234708786011,
      "learning_rate": 0.0005871748904759104,
      "loss": 2.3722,
      "step": 21522
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6127776503562927,
      "learning_rate": 0.0005871737072005528,
      "loss": 2.3462,
      "step": 21523
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6480463147163391,
      "learning_rate": 0.000587172523871804,
      "loss": 2.1865,
      "step": 21524
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6437179446220398,
      "learning_rate": 0.0005871713404896643,
      "loss": 2.4175,
      "step": 21525
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6240917444229126,
      "learning_rate": 0.0005871701570541342,
      "loss": 2.1661,
      "step": 21526
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.5999454259872437,
      "learning_rate": 0.0005871689735652136,
      "loss": 2.6776,
      "step": 21527
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.5535484552383423,
      "learning_rate": 0.0005871677900229029,
      "loss": 2.3697,
      "step": 21528
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6892412304878235,
      "learning_rate": 0.0005871666064272024,
      "loss": 2.6203,
      "step": 21529
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.5813614726066589,
      "learning_rate": 0.0005871654227781121,
      "loss": 2.2995,
      "step": 21530
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6194660067558289,
      "learning_rate": 0.0005871642390756324,
      "loss": 2.3381,
      "step": 21531
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6179253458976746,
      "learning_rate": 0.0005871630553197633,
      "loss": 2.4889,
      "step": 21532
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6274253726005554,
      "learning_rate": 0.0005871618715105054,
      "loss": 2.4292,
      "step": 21533
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6014004349708557,
      "learning_rate": 0.0005871606876478586,
      "loss": 2.2872,
      "step": 21534
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6335980296134949,
      "learning_rate": 0.000587159503731823,
      "loss": 2.216,
      "step": 21535
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6329416632652283,
      "learning_rate": 0.0005871583197623992,
      "loss": 2.4321,
      "step": 21536
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6293573975563049,
      "learning_rate": 0.0005871571357395874,
      "loss": 2.1965,
      "step": 21537
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6235040426254272,
      "learning_rate": 0.0005871559516633874,
      "loss": 2.2302,
      "step": 21538
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6661280393600464,
      "learning_rate": 0.0005871547675337999,
      "loss": 2.259,
      "step": 21539
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6471177935600281,
      "learning_rate": 0.0005871535833508248,
      "loss": 2.5679,
      "step": 21540
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.599737286567688,
      "learning_rate": 0.0005871523991144624,
      "loss": 2.4792,
      "step": 21541
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.5940648317337036,
      "learning_rate": 0.000587151214824713,
      "loss": 2.4424,
      "step": 21542
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6464853286743164,
      "learning_rate": 0.0005871500304815767,
      "loss": 2.3674,
      "step": 21543
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6064411997795105,
      "learning_rate": 0.0005871488460850538,
      "loss": 2.3165,
      "step": 21544
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6261571645736694,
      "learning_rate": 0.0005871476616351447,
      "loss": 2.2207,
      "step": 21545
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.5498424172401428,
      "learning_rate": 0.0005871464771318492,
      "loss": 2.3594,
      "step": 21546
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6235253214836121,
      "learning_rate": 0.0005871452925751679,
      "loss": 2.1971,
      "step": 21547
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.5857054591178894,
      "learning_rate": 0.0005871441079651008,
      "loss": 2.2553,
      "step": 21548
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.5787224769592285,
      "learning_rate": 0.0005871429233016482,
      "loss": 2.2715,
      "step": 21549
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.5981268286705017,
      "learning_rate": 0.0005871417385848103,
      "loss": 2.1235,
      "step": 21550
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6236032843589783,
      "learning_rate": 0.0005871405538145873,
      "loss": 2.2683,
      "step": 21551
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6700395345687866,
      "learning_rate": 0.0005871393689909795,
      "loss": 2.2983,
      "step": 21552
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.8353337645530701,
      "learning_rate": 0.000587138184113987,
      "loss": 2.2884,
      "step": 21553
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6143769025802612,
      "learning_rate": 0.0005871369991836102,
      "loss": 2.2785,
      "step": 21554
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6079739332199097,
      "learning_rate": 0.0005871358141998492,
      "loss": 2.5413,
      "step": 21555
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6327811479568481,
      "learning_rate": 0.0005871346291627041,
      "loss": 2.4287,
      "step": 21556
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.58861243724823,
      "learning_rate": 0.0005871334440721754,
      "loss": 2.6177,
      "step": 21557
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6006156802177429,
      "learning_rate": 0.000587132258928263,
      "loss": 2.2898,
      "step": 21558
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6189601421356201,
      "learning_rate": 0.0005871310737309675,
      "loss": 2.4489,
      "step": 21559
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6615749001502991,
      "learning_rate": 0.0005871298884802888,
      "loss": 2.4339,
      "step": 21560
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6238582134246826,
      "learning_rate": 0.0005871287031762273,
      "loss": 2.5226,
      "step": 21561
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6401941180229187,
      "learning_rate": 0.0005871275178187831,
      "loss": 2.4319,
      "step": 21562
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.5870355367660522,
      "learning_rate": 0.0005871263324079564,
      "loss": 2.3274,
      "step": 21563
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.5942126512527466,
      "learning_rate": 0.0005871251469437476,
      "loss": 2.3273,
      "step": 21564
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6061538457870483,
      "learning_rate": 0.0005871239614261568,
      "loss": 2.5972,
      "step": 21565
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6446285247802734,
      "learning_rate": 0.0005871227758551843,
      "loss": 2.4592,
      "step": 21566
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6030258536338806,
      "learning_rate": 0.0005871215902308302,
      "loss": 2.3062,
      "step": 21567
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.5605159401893616,
      "learning_rate": 0.0005871204045530947,
      "loss": 2.529,
      "step": 21568
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.7489686608314514,
      "learning_rate": 0.0005871192188219782,
      "loss": 2.4099,
      "step": 21569
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6017446517944336,
      "learning_rate": 0.0005871180330374808,
      "loss": 2.2862,
      "step": 21570
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.679660439491272,
      "learning_rate": 0.0005871168471996027,
      "loss": 2.3527,
      "step": 21571
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6255801916122437,
      "learning_rate": 0.0005871156613083443,
      "loss": 2.3463,
      "step": 21572
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.7293089628219604,
      "learning_rate": 0.0005871144753637054,
      "loss": 2.303,
      "step": 21573
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.5836759805679321,
      "learning_rate": 0.0005871132893656867,
      "loss": 2.1558,
      "step": 21574
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.7644594311714172,
      "learning_rate": 0.0005871121033142882,
      "loss": 2.2119,
      "step": 21575
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6694994568824768,
      "learning_rate": 0.00058711091720951,
      "loss": 2.5841,
      "step": 21576
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.7227995991706848,
      "learning_rate": 0.0005871097310513526,
      "loss": 2.3867,
      "step": 21577
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6376033425331116,
      "learning_rate": 0.0005871085448398161,
      "loss": 2.374,
      "step": 21578
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.678928792476654,
      "learning_rate": 0.0005871073585749007,
      "loss": 2.2006,
      "step": 21579
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.543902575969696,
      "learning_rate": 0.0005871061722566065,
      "loss": 2.5321,
      "step": 21580
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.5734500288963318,
      "learning_rate": 0.000587104985884934,
      "loss": 2.3639,
      "step": 21581
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.7278679013252258,
      "learning_rate": 0.0005871037994598831,
      "loss": 2.3181,
      "step": 21582
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6760936379432678,
      "learning_rate": 0.0005871026129814543,
      "loss": 2.3887,
      "step": 21583
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.788834273815155,
      "learning_rate": 0.0005871014264496477,
      "loss": 2.5589,
      "step": 21584
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6212846040725708,
      "learning_rate": 0.0005871002398644635,
      "loss": 2.3167,
      "step": 21585
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.602059543132782,
      "learning_rate": 0.000587099053225902,
      "loss": 2.2605,
      "step": 21586
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6634105443954468,
      "learning_rate": 0.0005870978665339633,
      "loss": 2.3912,
      "step": 21587
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.5981552600860596,
      "learning_rate": 0.0005870966797886476,
      "loss": 2.2357,
      "step": 21588
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6761292815208435,
      "learning_rate": 0.0005870954929899554,
      "loss": 2.4931,
      "step": 21589
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.651968777179718,
      "learning_rate": 0.0005870943061378867,
      "loss": 2.5692,
      "step": 21590
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.5945040583610535,
      "learning_rate": 0.0005870931192324415,
      "loss": 2.2118,
      "step": 21591
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6880916953086853,
      "learning_rate": 0.0005870919322736206,
      "loss": 2.4198,
      "step": 21592
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6500527262687683,
      "learning_rate": 0.0005870907452614237,
      "loss": 2.2157,
      "step": 21593
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.7931695580482483,
      "learning_rate": 0.0005870895581958513,
      "loss": 2.4737,
      "step": 21594
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.64787358045578,
      "learning_rate": 0.0005870883710769034,
      "loss": 2.3368,
      "step": 21595
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.7445289492607117,
      "learning_rate": 0.0005870871839045805,
      "loss": 2.4953,
      "step": 21596
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.5965578556060791,
      "learning_rate": 0.0005870859966788825,
      "loss": 2.3818,
      "step": 21597
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6533336639404297,
      "learning_rate": 0.0005870848093998101,
      "loss": 2.4009,
      "step": 21598
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6566709280014038,
      "learning_rate": 0.0005870836220673629,
      "loss": 2.2769,
      "step": 21599
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6515154242515564,
      "learning_rate": 0.0005870824346815416,
      "loss": 2.2842,
      "step": 21600
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6717603206634521,
      "learning_rate": 0.0005870812472423462,
      "loss": 2.5268,
      "step": 21601
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6564782857894897,
      "learning_rate": 0.000587080059749777,
      "loss": 2.4429,
      "step": 21602
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6367768049240112,
      "learning_rate": 0.0005870788722038342,
      "loss": 2.2655,
      "step": 21603
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.5749916434288025,
      "learning_rate": 0.000587077684604518,
      "loss": 2.2695,
      "step": 21604
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.696087658405304,
      "learning_rate": 0.0005870764969518286,
      "loss": 2.5795,
      "step": 21605
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.5848249197006226,
      "learning_rate": 0.0005870753092457663,
      "loss": 2.552,
      "step": 21606
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.5741173028945923,
      "learning_rate": 0.0005870741214863313,
      "loss": 2.4627,
      "step": 21607
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.5847718715667725,
      "learning_rate": 0.0005870729336735238,
      "loss": 2.3294,
      "step": 21608
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6198562979698181,
      "learning_rate": 0.000587071745807344,
      "loss": 2.6355,
      "step": 21609
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.5837005376815796,
      "learning_rate": 0.0005870705578877923,
      "loss": 2.2811,
      "step": 21610
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.5800806283950806,
      "learning_rate": 0.0005870693699148686,
      "loss": 2.2277,
      "step": 21611
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.628750205039978,
      "learning_rate": 0.0005870681818885733,
      "loss": 2.4741,
      "step": 21612
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6252565979957581,
      "learning_rate": 0.0005870669938089066,
      "loss": 2.2492,
      "step": 21613
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.5860912799835205,
      "learning_rate": 0.0005870658056758687,
      "loss": 2.4068,
      "step": 21614
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6829506158828735,
      "learning_rate": 0.0005870646174894599,
      "loss": 2.3498,
      "step": 21615
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6393158435821533,
      "learning_rate": 0.0005870634292496804,
      "loss": 2.4316,
      "step": 21616
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.721457839012146,
      "learning_rate": 0.0005870622409565303,
      "loss": 2.271,
      "step": 21617
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.5949485898017883,
      "learning_rate": 0.0005870610526100099,
      "loss": 2.3598,
      "step": 21618
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6194322109222412,
      "learning_rate": 0.0005870598642101195,
      "loss": 2.3209,
      "step": 21619
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6408851146697998,
      "learning_rate": 0.0005870586757568593,
      "loss": 2.3891,
      "step": 21620
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6599549055099487,
      "learning_rate": 0.0005870574872502294,
      "loss": 2.543,
      "step": 21621
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.635057270526886,
      "learning_rate": 0.0005870562986902301,
      "loss": 2.3664,
      "step": 21622
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6262311935424805,
      "learning_rate": 0.0005870551100768616,
      "loss": 2.1758,
      "step": 21623
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6959165930747986,
      "learning_rate": 0.0005870539214101242,
      "loss": 2.4617,
      "step": 21624
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.5852522850036621,
      "learning_rate": 0.0005870527326900179,
      "loss": 2.3003,
      "step": 21625
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6621957421302795,
      "learning_rate": 0.0005870515439165433,
      "loss": 2.1676,
      "step": 21626
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.7000706195831299,
      "learning_rate": 0.0005870503550897003,
      "loss": 2.1828,
      "step": 21627
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6554822325706482,
      "learning_rate": 0.0005870491662094892,
      "loss": 2.377,
      "step": 21628
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6244986057281494,
      "learning_rate": 0.0005870479772759103,
      "loss": 2.3291,
      "step": 21629
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.7989857196807861,
      "learning_rate": 0.0005870467882889638,
      "loss": 2.0864,
      "step": 21630
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6274752616882324,
      "learning_rate": 0.0005870455992486497,
      "loss": 2.4755,
      "step": 21631
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6659833788871765,
      "learning_rate": 0.0005870444101549685,
      "loss": 2.4493,
      "step": 21632
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6794289946556091,
      "learning_rate": 0.0005870432210079204,
      "loss": 2.3321,
      "step": 21633
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6665361523628235,
      "learning_rate": 0.0005870420318075055,
      "loss": 2.4825,
      "step": 21634
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.8283224701881409,
      "learning_rate": 0.000587040842553724,
      "loss": 2.1888,
      "step": 21635
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.5718423128128052,
      "learning_rate": 0.0005870396532465762,
      "loss": 2.351,
      "step": 21636
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.7148411870002747,
      "learning_rate": 0.0005870384638860624,
      "loss": 2.1031,
      "step": 21637
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6783385276794434,
      "learning_rate": 0.0005870372744721827,
      "loss": 2.2804,
      "step": 21638
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6321309804916382,
      "learning_rate": 0.0005870360850049373,
      "loss": 2.4553,
      "step": 21639
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6804018616676331,
      "learning_rate": 0.0005870348954843266,
      "loss": 2.5774,
      "step": 21640
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6841340065002441,
      "learning_rate": 0.0005870337059103505,
      "loss": 2.3122,
      "step": 21641
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.671660304069519,
      "learning_rate": 0.0005870325162830095,
      "loss": 2.2926,
      "step": 21642
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.5953428149223328,
      "learning_rate": 0.0005870313266023038,
      "loss": 2.3158,
      "step": 21643
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6309192776679993,
      "learning_rate": 0.0005870301368682335,
      "loss": 2.3865,
      "step": 21644
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6329704523086548,
      "learning_rate": 0.0005870289470807988,
      "loss": 2.4958,
      "step": 21645
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6262144446372986,
      "learning_rate": 0.0005870277572400001,
      "loss": 2.4381,
      "step": 21646
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6352405548095703,
      "learning_rate": 0.0005870265673458376,
      "loss": 2.1331,
      "step": 21647
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6380241513252258,
      "learning_rate": 0.0005870253773983113,
      "loss": 2.2586,
      "step": 21648
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.5861122608184814,
      "learning_rate": 0.0005870241873974217,
      "loss": 2.3472,
      "step": 21649
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6958680748939514,
      "learning_rate": 0.0005870229973431687,
      "loss": 2.1923,
      "step": 21650
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6328980326652527,
      "learning_rate": 0.0005870218072355529,
      "loss": 2.4551,
      "step": 21651
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6006565093994141,
      "learning_rate": 0.0005870206170745743,
      "loss": 2.3097,
      "step": 21652
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.5736081004142761,
      "learning_rate": 0.0005870194268602331,
      "loss": 2.204,
      "step": 21653
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6338348984718323,
      "learning_rate": 0.0005870182365925294,
      "loss": 2.2597,
      "step": 21654
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.662996232509613,
      "learning_rate": 0.0005870170462714639,
      "loss": 2.4764,
      "step": 21655
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6781187057495117,
      "learning_rate": 0.0005870158558970363,
      "loss": 2.2958,
      "step": 21656
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.7064972519874573,
      "learning_rate": 0.0005870146654692472,
      "loss": 2.5477,
      "step": 21657
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.8192757368087769,
      "learning_rate": 0.0005870134749880965,
      "loss": 2.5284,
      "step": 21658
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6575867533683777,
      "learning_rate": 0.0005870122844535846,
      "loss": 2.2771,
      "step": 21659
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6775894165039062,
      "learning_rate": 0.0005870110938657118,
      "loss": 2.3711,
      "step": 21660
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.206972599029541,
      "learning_rate": 0.0005870099032244782,
      "loss": 2.5474,
      "step": 21661
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.5975313782691956,
      "learning_rate": 0.0005870087125298839,
      "loss": 2.1949,
      "step": 21662
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.7663557529449463,
      "learning_rate": 0.0005870075217819294,
      "loss": 2.2704,
      "step": 21663
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6704506874084473,
      "learning_rate": 0.0005870063309806147,
      "loss": 2.2706,
      "step": 21664
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6554314494132996,
      "learning_rate": 0.0005870051401259402,
      "loss": 2.3875,
      "step": 21665
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6058093309402466,
      "learning_rate": 0.000587003949217906,
      "loss": 2.0626,
      "step": 21666
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.7933790683746338,
      "learning_rate": 0.0005870027582565123,
      "loss": 2.6049,
      "step": 21667
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.5912429094314575,
      "learning_rate": 0.0005870015672417594,
      "loss": 2.3742,
      "step": 21668
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.836020827293396,
      "learning_rate": 0.0005870003761736475,
      "loss": 2.4769,
      "step": 21669
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6580154895782471,
      "learning_rate": 0.0005869991850521768,
      "loss": 2.3111,
      "step": 21670
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6474782228469849,
      "learning_rate": 0.0005869979938773475,
      "loss": 2.4428,
      "step": 21671
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.5879762172698975,
      "learning_rate": 0.0005869968026491598,
      "loss": 2.0002,
      "step": 21672
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.580775797367096,
      "learning_rate": 0.0005869956113676142,
      "loss": 2.3182,
      "step": 21673
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6047011017799377,
      "learning_rate": 0.0005869944200327105,
      "loss": 2.2902,
      "step": 21674
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.5934274792671204,
      "learning_rate": 0.0005869932286444491,
      "loss": 2.3767,
      "step": 21675
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6051144599914551,
      "learning_rate": 0.0005869920372028304,
      "loss": 2.4383,
      "step": 21676
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6406912803649902,
      "learning_rate": 0.0005869908457078543,
      "loss": 2.3364,
      "step": 21677
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.7124046087265015,
      "learning_rate": 0.0005869896541595214,
      "loss": 2.2413,
      "step": 21678
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6287687420845032,
      "learning_rate": 0.0005869884625578314,
      "loss": 2.4566,
      "step": 21679
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.5612950921058655,
      "learning_rate": 0.0005869872709027851,
      "loss": 2.4454,
      "step": 21680
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6491769552230835,
      "learning_rate": 0.0005869860791943823,
      "loss": 2.2622,
      "step": 21681
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.5720451474189758,
      "learning_rate": 0.0005869848874326235,
      "loss": 2.2934,
      "step": 21682
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6329240798950195,
      "learning_rate": 0.0005869836956175087,
      "loss": 2.3094,
      "step": 21683
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.7146260142326355,
      "learning_rate": 0.0005869825037490382,
      "loss": 2.1626,
      "step": 21684
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6256871223449707,
      "learning_rate": 0.0005869813118272123,
      "loss": 2.3284,
      "step": 21685
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6905988454818726,
      "learning_rate": 0.0005869801198520312,
      "loss": 2.3736,
      "step": 21686
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6146938800811768,
      "learning_rate": 0.0005869789278234949,
      "loss": 2.4835,
      "step": 21687
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.9173072576522827,
      "learning_rate": 0.000586977735741604,
      "loss": 2.2638,
      "step": 21688
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6899033188819885,
      "learning_rate": 0.0005869765436063584,
      "loss": 2.399,
      "step": 21689
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6841602921485901,
      "learning_rate": 0.0005869753514177584,
      "loss": 2.5602,
      "step": 21690
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.5922034978866577,
      "learning_rate": 0.0005869741591758044,
      "loss": 2.3411,
      "step": 21691
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.5761955380439758,
      "learning_rate": 0.0005869729668804963,
      "loss": 2.2239,
      "step": 21692
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.9571153521537781,
      "learning_rate": 0.0005869717745318347,
      "loss": 2.427,
      "step": 21693
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6306942701339722,
      "learning_rate": 0.0005869705821298195,
      "loss": 2.4439,
      "step": 21694
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6468955278396606,
      "learning_rate": 0.0005869693896744511,
      "loss": 2.3094,
      "step": 21695
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.8950621485710144,
      "learning_rate": 0.0005869681971657297,
      "loss": 2.3503,
      "step": 21696
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6443832516670227,
      "learning_rate": 0.0005869670046036555,
      "loss": 2.7378,
      "step": 21697
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.5968058705329895,
      "learning_rate": 0.0005869658119882288,
      "loss": 2.5053,
      "step": 21698
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6841174364089966,
      "learning_rate": 0.0005869646193194495,
      "loss": 2.1357,
      "step": 21699
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6708331108093262,
      "learning_rate": 0.0005869634265973183,
      "loss": 2.1407,
      "step": 21700
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6350027322769165,
      "learning_rate": 0.0005869622338218351,
      "loss": 2.3721,
      "step": 21701
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6830715537071228,
      "learning_rate": 0.0005869610409930001,
      "loss": 2.3808,
      "step": 21702
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.7953559756278992,
      "learning_rate": 0.0005869598481108137,
      "loss": 2.4497,
      "step": 21703
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6865178346633911,
      "learning_rate": 0.0005869586551752761,
      "loss": 2.4195,
      "step": 21704
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6564729809761047,
      "learning_rate": 0.0005869574621863873,
      "loss": 2.3241,
      "step": 21705
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6088521480560303,
      "learning_rate": 0.0005869562691441479,
      "loss": 2.3451,
      "step": 21706
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6132965087890625,
      "learning_rate": 0.0005869550760485578,
      "loss": 2.4478,
      "step": 21707
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.5666503310203552,
      "learning_rate": 0.0005869538828996173,
      "loss": 2.2208,
      "step": 21708
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.638599157333374,
      "learning_rate": 0.0005869526896973268,
      "loss": 2.3112,
      "step": 21709
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6465344429016113,
      "learning_rate": 0.0005869514964416863,
      "loss": 2.2946,
      "step": 21710
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6715567111968994,
      "learning_rate": 0.0005869503031326961,
      "loss": 2.3401,
      "step": 21711
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.584018349647522,
      "learning_rate": 0.0005869491097703563,
      "loss": 2.4666,
      "step": 21712
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6449626088142395,
      "learning_rate": 0.0005869479163546675,
      "loss": 2.5373,
      "step": 21713
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6687923669815063,
      "learning_rate": 0.0005869467228856295,
      "loss": 2.4812,
      "step": 21714
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.70304274559021,
      "learning_rate": 0.0005869455293632427,
      "loss": 2.4669,
      "step": 21715
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.5974293351173401,
      "learning_rate": 0.0005869443357875073,
      "loss": 2.3744,
      "step": 21716
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6304726004600525,
      "learning_rate": 0.0005869431421584235,
      "loss": 2.2813,
      "step": 21717
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6632594466209412,
      "learning_rate": 0.0005869419484759917,
      "loss": 2.0476,
      "step": 21718
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6103869676589966,
      "learning_rate": 0.0005869407547402118,
      "loss": 2.2213,
      "step": 21719
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6377456188201904,
      "learning_rate": 0.0005869395609510843,
      "loss": 2.6109,
      "step": 21720
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.7537604570388794,
      "learning_rate": 0.0005869383671086093,
      "loss": 2.2836,
      "step": 21721
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.5781575441360474,
      "learning_rate": 0.000586937173212787,
      "loss": 2.3607,
      "step": 21722
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.667000412940979,
      "learning_rate": 0.0005869359792636178,
      "loss": 2.4385,
      "step": 21723
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6293243169784546,
      "learning_rate": 0.0005869347852611016,
      "loss": 2.2312,
      "step": 21724
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6157903075218201,
      "learning_rate": 0.0005869335912052389,
      "loss": 2.2354,
      "step": 21725
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.5713865756988525,
      "learning_rate": 0.0005869323970960299,
      "loss": 2.4992,
      "step": 21726
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6479446291923523,
      "learning_rate": 0.0005869312029334747,
      "loss": 2.4713,
      "step": 21727
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.590215802192688,
      "learning_rate": 0.0005869300087175735,
      "loss": 2.1002,
      "step": 21728
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6542434096336365,
      "learning_rate": 0.0005869288144483266,
      "loss": 2.1035,
      "step": 21729
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6161807775497437,
      "learning_rate": 0.0005869276201257343,
      "loss": 2.3054,
      "step": 21730
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6717596054077148,
      "learning_rate": 0.0005869264257497967,
      "loss": 2.2871,
      "step": 21731
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6543985605239868,
      "learning_rate": 0.000586925231320514,
      "loss": 2.6219,
      "step": 21732
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6113221049308777,
      "learning_rate": 0.0005869240368378866,
      "loss": 2.4432,
      "step": 21733
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6173381805419922,
      "learning_rate": 0.0005869228423019146,
      "loss": 2.167,
      "step": 21734
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6471589207649231,
      "learning_rate": 0.0005869216477125982,
      "loss": 2.2739,
      "step": 21735
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6165313720703125,
      "learning_rate": 0.0005869204530699376,
      "loss": 2.1694,
      "step": 21736
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.634183943271637,
      "learning_rate": 0.0005869192583739331,
      "loss": 2.3426,
      "step": 21737
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.5792558789253235,
      "learning_rate": 0.0005869180636245848,
      "loss": 2.4075,
      "step": 21738
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6943092346191406,
      "learning_rate": 0.0005869168688218932,
      "loss": 2.6909,
      "step": 21739
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6010228991508484,
      "learning_rate": 0.0005869156739658582,
      "loss": 2.3841,
      "step": 21740
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.5919987559318542,
      "learning_rate": 0.0005869144790564803,
      "loss": 2.1744,
      "step": 21741
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6424044966697693,
      "learning_rate": 0.0005869132840937595,
      "loss": 2.3151,
      "step": 21742
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6504115462303162,
      "learning_rate": 0.000586912089077696,
      "loss": 2.4013,
      "step": 21743
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6133332252502441,
      "learning_rate": 0.0005869108940082902,
      "loss": 2.4306,
      "step": 21744
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6347965598106384,
      "learning_rate": 0.0005869096988855423,
      "loss": 2.3615,
      "step": 21745
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6296674013137817,
      "learning_rate": 0.0005869085037094524,
      "loss": 2.2606,
      "step": 21746
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.5980709791183472,
      "learning_rate": 0.0005869073084800208,
      "loss": 2.3875,
      "step": 21747
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.5912790894508362,
      "learning_rate": 0.0005869061131972478,
      "loss": 2.3475,
      "step": 21748
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.7393098473548889,
      "learning_rate": 0.0005869049178611335,
      "loss": 2.5176,
      "step": 21749
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6200145483016968,
      "learning_rate": 0.0005869037224716781,
      "loss": 2.3149,
      "step": 21750
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6254377961158752,
      "learning_rate": 0.0005869025270288821,
      "loss": 2.4409,
      "step": 21751
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6148521304130554,
      "learning_rate": 0.0005869013315327452,
      "loss": 2.3105,
      "step": 21752
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6980910897254944,
      "learning_rate": 0.0005869001359832681,
      "loss": 2.3316,
      "step": 21753
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.5799604654312134,
      "learning_rate": 0.0005868989403804508,
      "loss": 2.1979,
      "step": 21754
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6276283264160156,
      "learning_rate": 0.0005868977447242936,
      "loss": 2.549,
      "step": 21755
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.7478834986686707,
      "learning_rate": 0.0005868965490147967,
      "loss": 2.5324,
      "step": 21756
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.599517822265625,
      "learning_rate": 0.0005868953532519603,
      "loss": 2.3363,
      "step": 21757
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.7434790134429932,
      "learning_rate": 0.0005868941574357846,
      "loss": 2.3142,
      "step": 21758
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6604510545730591,
      "learning_rate": 0.0005868929615662698,
      "loss": 2.2886,
      "step": 21759
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.5876660943031311,
      "learning_rate": 0.0005868917656434164,
      "loss": 2.2613,
      "step": 21760
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.587833046913147,
      "learning_rate": 0.0005868905696672242,
      "loss": 2.3079,
      "step": 21761
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6620759963989258,
      "learning_rate": 0.0005868893736376938,
      "loss": 2.3978,
      "step": 21762
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.5845174789428711,
      "learning_rate": 0.000586888177554825,
      "loss": 2.3142,
      "step": 21763
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6213585734367371,
      "learning_rate": 0.0005868869814186186,
      "loss": 2.4189,
      "step": 21764
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6975885033607483,
      "learning_rate": 0.0005868857852290742,
      "loss": 2.0897,
      "step": 21765
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6957066059112549,
      "learning_rate": 0.0005868845889861925,
      "loss": 2.384,
      "step": 21766
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6275348663330078,
      "learning_rate": 0.0005868833926899735,
      "loss": 2.4062,
      "step": 21767
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.609971821308136,
      "learning_rate": 0.0005868821963404175,
      "loss": 2.345,
      "step": 21768
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.5968821048736572,
      "learning_rate": 0.0005868809999375246,
      "loss": 2.2222,
      "step": 21769
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6088358163833618,
      "learning_rate": 0.0005868798034812951,
      "loss": 2.4602,
      "step": 21770
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.64582759141922,
      "learning_rate": 0.0005868786069717292,
      "loss": 2.4796,
      "step": 21771
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.5743518471717834,
      "learning_rate": 0.0005868774104088272,
      "loss": 2.2679,
      "step": 21772
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6156346797943115,
      "learning_rate": 0.0005868762137925894,
      "loss": 2.4433,
      "step": 21773
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6275951862335205,
      "learning_rate": 0.0005868750171230157,
      "loss": 2.3463,
      "step": 21774
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.5924016237258911,
      "learning_rate": 0.0005868738204001067,
      "loss": 2.4353,
      "step": 21775
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.5948498845100403,
      "learning_rate": 0.0005868726236238623,
      "loss": 2.4275,
      "step": 21776
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6592317819595337,
      "learning_rate": 0.0005868714267942828,
      "loss": 2.3844,
      "step": 21777
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6951635479927063,
      "learning_rate": 0.0005868702299113688,
      "loss": 2.1641,
      "step": 21778
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6034936904907227,
      "learning_rate": 0.0005868690329751199,
      "loss": 2.5714,
      "step": 21779
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6149101853370667,
      "learning_rate": 0.0005868678359855368,
      "loss": 2.3486,
      "step": 21780
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.5901272296905518,
      "learning_rate": 0.0005868666389426194,
      "loss": 2.3865,
      "step": 21781
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6288324594497681,
      "learning_rate": 0.0005868654418463682,
      "loss": 2.5012,
      "step": 21782
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6017377972602844,
      "learning_rate": 0.0005868642446967832,
      "loss": 2.3531,
      "step": 21783
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.5773383975028992,
      "learning_rate": 0.0005868630474938648,
      "loss": 2.5659,
      "step": 21784
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6165453195571899,
      "learning_rate": 0.0005868618502376132,
      "loss": 2.2193,
      "step": 21785
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6388640999794006,
      "learning_rate": 0.0005868606529280285,
      "loss": 2.4086,
      "step": 21786
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6241331100463867,
      "learning_rate": 0.000586859455565111,
      "loss": 2.1248,
      "step": 21787
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.5437979102134705,
      "learning_rate": 0.0005868582581488609,
      "loss": 2.1015,
      "step": 21788
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.7697943449020386,
      "learning_rate": 0.0005868570606792784,
      "loss": 2.2308,
      "step": 21789
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6409794092178345,
      "learning_rate": 0.0005868558631563638,
      "loss": 2.402,
      "step": 21790
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6487287282943726,
      "learning_rate": 0.0005868546655801173,
      "loss": 2.2004,
      "step": 21791
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6585187315940857,
      "learning_rate": 0.000586853467950539,
      "loss": 2.577,
      "step": 21792
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6701366901397705,
      "learning_rate": 0.0005868522702676294,
      "loss": 2.5694,
      "step": 21793
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6115591526031494,
      "learning_rate": 0.0005868510725313884,
      "loss": 2.3255,
      "step": 21794
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6420215368270874,
      "learning_rate": 0.0005868498747418165,
      "loss": 2.333,
      "step": 21795
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.720621645450592,
      "learning_rate": 0.0005868486768989137,
      "loss": 2.4426,
      "step": 21796
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6121610403060913,
      "learning_rate": 0.0005868474790026803,
      "loss": 2.3086,
      "step": 21797
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6870060563087463,
      "learning_rate": 0.0005868462810531166,
      "loss": 2.2672,
      "step": 21798
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6928660869598389,
      "learning_rate": 0.0005868450830502227,
      "loss": 2.5556,
      "step": 21799
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6038962602615356,
      "learning_rate": 0.000586843884993999,
      "loss": 2.3074,
      "step": 21800
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6212624311447144,
      "learning_rate": 0.0005868426868844454,
      "loss": 2.3697,
      "step": 21801
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6494671106338501,
      "learning_rate": 0.0005868414887215625,
      "loss": 2.3793,
      "step": 21802
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6926173567771912,
      "learning_rate": 0.0005868402905053504,
      "loss": 2.4531,
      "step": 21803
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.5569009780883789,
      "learning_rate": 0.0005868390922358091,
      "loss": 2.3652,
      "step": 21804
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.8632453680038452,
      "learning_rate": 0.000586837893912939,
      "loss": 2.2666,
      "step": 21805
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6383241415023804,
      "learning_rate": 0.0005868366955367405,
      "loss": 2.4168,
      "step": 21806
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.5899526476860046,
      "learning_rate": 0.0005868354971072135,
      "loss": 2.2095,
      "step": 21807
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.9338343739509583,
      "learning_rate": 0.0005868342986243585,
      "loss": 2.3296,
      "step": 21808
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.646411657333374,
      "learning_rate": 0.0005868331000881755,
      "loss": 2.4311,
      "step": 21809
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.8207621574401855,
      "learning_rate": 0.0005868319014986648,
      "loss": 2.414,
      "step": 21810
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6216822862625122,
      "learning_rate": 0.0005868307028558266,
      "loss": 2.4679,
      "step": 21811
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.676369309425354,
      "learning_rate": 0.0005868295041596613,
      "loss": 2.3896,
      "step": 21812
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6402596831321716,
      "learning_rate": 0.0005868283054101689,
      "loss": 2.3412,
      "step": 21813
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6172335743904114,
      "learning_rate": 0.0005868271066073496,
      "loss": 2.4955,
      "step": 21814
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6894996166229248,
      "learning_rate": 0.0005868259077512039,
      "loss": 2.3996,
      "step": 21815
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.5723135471343994,
      "learning_rate": 0.0005868247088417316,
      "loss": 2.308,
      "step": 21816
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.5663129687309265,
      "learning_rate": 0.0005868235098789334,
      "loss": 2.2699,
      "step": 21817
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.5541039705276489,
      "learning_rate": 0.0005868223108628093,
      "loss": 2.5173,
      "step": 21818
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.641554594039917,
      "learning_rate": 0.0005868211117933595,
      "loss": 2.1014,
      "step": 21819
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6254475712776184,
      "learning_rate": 0.0005868199126705841,
      "loss": 2.2965,
      "step": 21820
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6373755931854248,
      "learning_rate": 0.0005868187134944836,
      "loss": 2.0253,
      "step": 21821
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.603810727596283,
      "learning_rate": 0.0005868175142650581,
      "loss": 2.4231,
      "step": 21822
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6214921474456787,
      "learning_rate": 0.0005868163149823077,
      "loss": 2.4784,
      "step": 21823
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6260692477226257,
      "learning_rate": 0.0005868151156462328,
      "loss": 2.4456,
      "step": 21824
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.655831515789032,
      "learning_rate": 0.0005868139162568336,
      "loss": 2.4771,
      "step": 21825
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6189211010932922,
      "learning_rate": 0.0005868127168141103,
      "loss": 2.6129,
      "step": 21826
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6444577574729919,
      "learning_rate": 0.000586811517318063,
      "loss": 2.3368,
      "step": 21827
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6199029088020325,
      "learning_rate": 0.000586810317768692,
      "loss": 2.2906,
      "step": 21828
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6590985655784607,
      "learning_rate": 0.0005868091181659975,
      "loss": 2.338,
      "step": 21829
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6053292751312256,
      "learning_rate": 0.0005868079185099799,
      "loss": 2.5104,
      "step": 21830
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.9960523843765259,
      "learning_rate": 0.0005868067188006393,
      "loss": 2.3481,
      "step": 21831
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.5927563309669495,
      "learning_rate": 0.0005868055190379759,
      "loss": 2.3878,
      "step": 21832
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6916100978851318,
      "learning_rate": 0.0005868043192219899,
      "loss": 2.2195,
      "step": 21833
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6932411789894104,
      "learning_rate": 0.0005868031193526816,
      "loss": 2.6506,
      "step": 21834
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.646578311920166,
      "learning_rate": 0.0005868019194300512,
      "loss": 2.291,
      "step": 21835
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.666847825050354,
      "learning_rate": 0.0005868007194540988,
      "loss": 2.3962,
      "step": 21836
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.7562270164489746,
      "learning_rate": 0.0005867995194248248,
      "loss": 2.3481,
      "step": 21837
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.8171002864837646,
      "learning_rate": 0.0005867983193422293,
      "loss": 2.1702,
      "step": 21838
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6414061188697815,
      "learning_rate": 0.0005867971192063127,
      "loss": 2.4379,
      "step": 21839
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.7363359332084656,
      "learning_rate": 0.0005867959190170751,
      "loss": 2.4833,
      "step": 21840
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.8058580756187439,
      "learning_rate": 0.0005867947187745168,
      "loss": 2.39,
      "step": 21841
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.7873343229293823,
      "learning_rate": 0.0005867935184786377,
      "loss": 2.4593,
      "step": 21842
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.607782781124115,
      "learning_rate": 0.0005867923181294384,
      "loss": 2.1782,
      "step": 21843
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6927644610404968,
      "learning_rate": 0.0005867911177269191,
      "loss": 2.2946,
      "step": 21844
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6694027781486511,
      "learning_rate": 0.0005867899172710797,
      "loss": 2.443,
      "step": 21845
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6038270592689514,
      "learning_rate": 0.0005867887167619207,
      "loss": 2.2422,
      "step": 21846
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6190814971923828,
      "learning_rate": 0.0005867875161994424,
      "loss": 2.4759,
      "step": 21847
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6683584451675415,
      "learning_rate": 0.0005867863155836448,
      "loss": 2.2849,
      "step": 21848
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6639167666435242,
      "learning_rate": 0.0005867851149145282,
      "loss": 2.4297,
      "step": 21849
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6570119261741638,
      "learning_rate": 0.0005867839141920928,
      "loss": 2.3107,
      "step": 21850
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.7724210619926453,
      "learning_rate": 0.0005867827134163389,
      "loss": 2.4261,
      "step": 21851
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.7492312788963318,
      "learning_rate": 0.0005867815125872667,
      "loss": 2.7064,
      "step": 21852
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.5734153985977173,
      "learning_rate": 0.0005867803117048764,
      "loss": 2.2606,
      "step": 21853
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6241087317466736,
      "learning_rate": 0.0005867791107691682,
      "loss": 2.504,
      "step": 21854
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.5962574481964111,
      "learning_rate": 0.0005867779097801423,
      "loss": 2.4464,
      "step": 21855
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6639083027839661,
      "learning_rate": 0.0005867767087377991,
      "loss": 2.2857,
      "step": 21856
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.759820282459259,
      "learning_rate": 0.0005867755076421386,
      "loss": 2.4041,
      "step": 21857
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.7053171396255493,
      "learning_rate": 0.0005867743064931611,
      "loss": 2.2241,
      "step": 21858
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6722258925437927,
      "learning_rate": 0.000586773105290867,
      "loss": 2.2364,
      "step": 21859
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6381587982177734,
      "learning_rate": 0.0005867719040352563,
      "loss": 2.5461,
      "step": 21860
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6405900716781616,
      "learning_rate": 0.0005867707027263291,
      "loss": 2.4194,
      "step": 21861
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.5877699851989746,
      "learning_rate": 0.000586769501364086,
      "loss": 2.1908,
      "step": 21862
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.7304110527038574,
      "learning_rate": 0.000586768299948527,
      "loss": 2.3579,
      "step": 21863
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6456515192985535,
      "learning_rate": 0.0005867670984796525,
      "loss": 2.5233,
      "step": 21864
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.5878811478614807,
      "learning_rate": 0.0005867658969574623,
      "loss": 2.3318,
      "step": 21865
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.7142014503479004,
      "learning_rate": 0.000586764695381957,
      "loss": 2.3643,
      "step": 21866
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6947014927864075,
      "learning_rate": 0.0005867634937531369,
      "loss": 2.4586,
      "step": 21867
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.5666529536247253,
      "learning_rate": 0.0005867622920710018,
      "loss": 2.409,
      "step": 21868
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6861087679862976,
      "learning_rate": 0.0005867610903355524,
      "loss": 2.4898,
      "step": 21869
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6721151471138,
      "learning_rate": 0.0005867598885467886,
      "loss": 2.4294,
      "step": 21870
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.63176029920578,
      "learning_rate": 0.0005867586867047107,
      "loss": 2.5833,
      "step": 21871
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.7409936785697937,
      "learning_rate": 0.000586757484809319,
      "loss": 2.5196,
      "step": 21872
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.5675601959228516,
      "learning_rate": 0.0005867562828606136,
      "loss": 2.3899,
      "step": 21873
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.5870588421821594,
      "learning_rate": 0.0005867550808585949,
      "loss": 2.3199,
      "step": 21874
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6228135824203491,
      "learning_rate": 0.000586753878803263,
      "loss": 2.2132,
      "step": 21875
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.5787447094917297,
      "learning_rate": 0.000586752676694618,
      "loss": 2.2674,
      "step": 21876
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.5992916226387024,
      "learning_rate": 0.0005867514745326605,
      "loss": 2.3124,
      "step": 21877
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.5746532082557678,
      "learning_rate": 0.0005867502723173903,
      "loss": 2.2997,
      "step": 21878
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6684446334838867,
      "learning_rate": 0.0005867490700488079,
      "loss": 2.3808,
      "step": 21879
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.5909091830253601,
      "learning_rate": 0.0005867478677269134,
      "loss": 2.4119,
      "step": 21880
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6964153051376343,
      "learning_rate": 0.0005867466653517071,
      "loss": 2.1701,
      "step": 21881
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.5913282632827759,
      "learning_rate": 0.0005867454629231891,
      "loss": 2.366,
      "step": 21882
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.5928865671157837,
      "learning_rate": 0.0005867442604413597,
      "loss": 2.18,
      "step": 21883
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.7270786762237549,
      "learning_rate": 0.0005867430579062192,
      "loss": 2.4394,
      "step": 21884
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6312475204467773,
      "learning_rate": 0.0005867418553177678,
      "loss": 2.4788,
      "step": 21885
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6406133770942688,
      "learning_rate": 0.0005867406526760056,
      "loss": 2.2871,
      "step": 21886
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.5896158814430237,
      "learning_rate": 0.0005867394499809328,
      "loss": 2.4691,
      "step": 21887
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6557082533836365,
      "learning_rate": 0.0005867382472325499,
      "loss": 2.1679,
      "step": 21888
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.656021773815155,
      "learning_rate": 0.0005867370444308568,
      "loss": 2.3286,
      "step": 21889
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.5992753505706787,
      "learning_rate": 0.0005867358415758539,
      "loss": 2.4092,
      "step": 21890
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6428514122962952,
      "learning_rate": 0.0005867346386675416,
      "loss": 2.1834,
      "step": 21891
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.5960305333137512,
      "learning_rate": 0.0005867334357059197,
      "loss": 2.3717,
      "step": 21892
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6488248109817505,
      "learning_rate": 0.0005867322326909887,
      "loss": 2.4293,
      "step": 21893
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.5788470506668091,
      "learning_rate": 0.0005867310296227488,
      "loss": 2.3085,
      "step": 21894
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6585472226142883,
      "learning_rate": 0.0005867298265012002,
      "loss": 2.2351,
      "step": 21895
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.5877448916435242,
      "learning_rate": 0.000586728623326343,
      "loss": 2.4,
      "step": 21896
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6906668543815613,
      "learning_rate": 0.0005867274200981776,
      "loss": 2.2787,
      "step": 21897
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6039592623710632,
      "learning_rate": 0.0005867262168167042,
      "loss": 2.45,
      "step": 21898
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6950685977935791,
      "learning_rate": 0.0005867250134819229,
      "loss": 2.6032,
      "step": 21899
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6788811087608337,
      "learning_rate": 0.0005867238100938341,
      "loss": 2.522,
      "step": 21900
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6424638032913208,
      "learning_rate": 0.0005867226066524378,
      "loss": 2.4687,
      "step": 21901
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.5987175107002258,
      "learning_rate": 0.0005867214031577345,
      "loss": 2.4491,
      "step": 21902
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6891586780548096,
      "learning_rate": 0.0005867201996097243,
      "loss": 2.3292,
      "step": 21903
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6647480130195618,
      "learning_rate": 0.0005867189960084073,
      "loss": 2.3567,
      "step": 21904
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.5944836139678955,
      "learning_rate": 0.0005867177923537839,
      "loss": 2.279,
      "step": 21905
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6191976070404053,
      "learning_rate": 0.0005867165886458541,
      "loss": 2.4799,
      "step": 21906
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6339287757873535,
      "learning_rate": 0.0005867153848846184,
      "loss": 2.285,
      "step": 21907
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.5914263725280762,
      "learning_rate": 0.0005867141810700769,
      "loss": 2.3584,
      "step": 21908
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6620516180992126,
      "learning_rate": 0.0005867129772022298,
      "loss": 2.3385,
      "step": 21909
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6188250780105591,
      "learning_rate": 0.0005867117732810774,
      "loss": 2.1314,
      "step": 21910
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6226703524589539,
      "learning_rate": 0.0005867105693066197,
      "loss": 2.0813,
      "step": 21911
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.599121630191803,
      "learning_rate": 0.0005867093652788572,
      "loss": 2.4292,
      "step": 21912
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.7125908732414246,
      "learning_rate": 0.0005867081611977901,
      "loss": 2.3701,
      "step": 21913
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6652950048446655,
      "learning_rate": 0.0005867069570634184,
      "loss": 2.5103,
      "step": 21914
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6185433268547058,
      "learning_rate": 0.0005867057528757425,
      "loss": 2.373,
      "step": 21915
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.577752947807312,
      "learning_rate": 0.0005867045486347627,
      "loss": 2.3049,
      "step": 21916
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6265890002250671,
      "learning_rate": 0.000586703344340479,
      "loss": 2.2084,
      "step": 21917
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6046698689460754,
      "learning_rate": 0.0005867021399928918,
      "loss": 2.22,
      "step": 21918
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6117857694625854,
      "learning_rate": 0.0005867009355920011,
      "loss": 2.6067,
      "step": 21919
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6320905089378357,
      "learning_rate": 0.0005866997311378075,
      "loss": 2.2993,
      "step": 21920
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.5890364646911621,
      "learning_rate": 0.000586698526630311,
      "loss": 2.299,
      "step": 21921
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.7007710933685303,
      "learning_rate": 0.0005866973220695116,
      "loss": 2.2106,
      "step": 21922
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.5323536396026611,
      "learning_rate": 0.00058669611745541,
      "loss": 2.2142,
      "step": 21923
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.679630696773529,
      "learning_rate": 0.000586694912788006,
      "loss": 2.4156,
      "step": 21924
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6706875562667847,
      "learning_rate": 0.0005866937080673002,
      "loss": 2.572,
      "step": 21925
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.7015853524208069,
      "learning_rate": 0.0005866925032932924,
      "loss": 2.3088,
      "step": 21926
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6840390563011169,
      "learning_rate": 0.0005866912984659832,
      "loss": 2.4552,
      "step": 21927
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.5900819897651672,
      "learning_rate": 0.0005866900935853727,
      "loss": 2.1562,
      "step": 21928
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6586308479309082,
      "learning_rate": 0.000586688888651461,
      "loss": 2.5603,
      "step": 21929
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.5947305560112,
      "learning_rate": 0.0005866876836642485,
      "loss": 2.3618,
      "step": 21930
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6140957474708557,
      "learning_rate": 0.0005866864786237353,
      "loss": 2.4483,
      "step": 21931
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6640512347221375,
      "learning_rate": 0.0005866852735299217,
      "loss": 2.2149,
      "step": 21932
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6766620874404907,
      "learning_rate": 0.000586684068382808,
      "loss": 2.4543,
      "step": 21933
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6010302901268005,
      "learning_rate": 0.0005866828631823941,
      "loss": 2.4885,
      "step": 21934
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6081100702285767,
      "learning_rate": 0.0005866816579286805,
      "loss": 2.352,
      "step": 21935
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6756176948547363,
      "learning_rate": 0.0005866804526216675,
      "loss": 2.4951,
      "step": 21936
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.5979520678520203,
      "learning_rate": 0.000586679247261355,
      "loss": 2.4028,
      "step": 21937
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6635509729385376,
      "learning_rate": 0.0005866780418477436,
      "loss": 2.1897,
      "step": 21938
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6063398718833923,
      "learning_rate": 0.0005866768363808332,
      "loss": 2.4799,
      "step": 21939
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6632903814315796,
      "learning_rate": 0.0005866756308606243,
      "loss": 2.1627,
      "step": 21940
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6242433190345764,
      "learning_rate": 0.0005866744252871169,
      "loss": 2.3383,
      "step": 21941
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6129829287528992,
      "learning_rate": 0.0005866732196603113,
      "loss": 2.2485,
      "step": 21942
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.5846977829933167,
      "learning_rate": 0.0005866720139802078,
      "loss": 2.1499,
      "step": 21943
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6452532410621643,
      "learning_rate": 0.0005866708082468066,
      "loss": 2.3186,
      "step": 21944
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6538214683532715,
      "learning_rate": 0.0005866696024601078,
      "loss": 2.1324,
      "step": 21945
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.58966064453125,
      "learning_rate": 0.0005866683966201117,
      "loss": 2.2261,
      "step": 21946
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6335719227790833,
      "learning_rate": 0.0005866671907268186,
      "loss": 2.3774,
      "step": 21947
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6027308702468872,
      "learning_rate": 0.0005866659847802285,
      "loss": 2.1226,
      "step": 21948
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6272103786468506,
      "learning_rate": 0.000586664778780342,
      "loss": 2.2626,
      "step": 21949
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6188896298408508,
      "learning_rate": 0.000586663572727159,
      "loss": 2.4349,
      "step": 21950
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.5641252398490906,
      "learning_rate": 0.0005866623666206798,
      "loss": 2.3872,
      "step": 21951
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6389639973640442,
      "learning_rate": 0.0005866611604609047,
      "loss": 2.3953,
      "step": 21952
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.5716841220855713,
      "learning_rate": 0.0005866599542478339,
      "loss": 2.4847,
      "step": 21953
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6260290741920471,
      "learning_rate": 0.0005866587479814676,
      "loss": 2.3609,
      "step": 21954
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6479421257972717,
      "learning_rate": 0.000586657541661806,
      "loss": 1.9586,
      "step": 21955
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.5423350930213928,
      "learning_rate": 0.0005866563352888493,
      "loss": 2.5398,
      "step": 21956
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6677786111831665,
      "learning_rate": 0.0005866551288625978,
      "loss": 2.2898,
      "step": 21957
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6131508946418762,
      "learning_rate": 0.0005866539223830517,
      "loss": 2.3202,
      "step": 21958
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.587702214717865,
      "learning_rate": 0.0005866527158502113,
      "loss": 2.3895,
      "step": 21959
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6031713485717773,
      "learning_rate": 0.0005866515092640766,
      "loss": 2.322,
      "step": 21960
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6714830994606018,
      "learning_rate": 0.0005866503026246482,
      "loss": 2.4074,
      "step": 21961
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6056488752365112,
      "learning_rate": 0.000586649095931926,
      "loss": 2.3996,
      "step": 21962
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.7530678510665894,
      "learning_rate": 0.0005866478891859103,
      "loss": 2.4659,
      "step": 21963
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6352806687355042,
      "learning_rate": 0.0005866466823866012,
      "loss": 2.5198,
      "step": 21964
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6913530230522156,
      "learning_rate": 0.0005866454755339993,
      "loss": 2.0365,
      "step": 21965
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6917020082473755,
      "learning_rate": 0.0005866442686281046,
      "loss": 2.4664,
      "step": 21966
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6567592024803162,
      "learning_rate": 0.0005866430616689171,
      "loss": 2.558,
      "step": 21967
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6978681087493896,
      "learning_rate": 0.0005866418546564375,
      "loss": 2.4528,
      "step": 21968
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6402080655097961,
      "learning_rate": 0.0005866406475906656,
      "loss": 2.4613,
      "step": 21969
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.61851567029953,
      "learning_rate": 0.0005866394404716018,
      "loss": 2.2958,
      "step": 21970
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6522489786148071,
      "learning_rate": 0.0005866382332992464,
      "loss": 2.3814,
      "step": 21971
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6336705088615417,
      "learning_rate": 0.0005866370260735994,
      "loss": 2.1337,
      "step": 21972
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6109737753868103,
      "learning_rate": 0.0005866358187946614,
      "loss": 2.4049,
      "step": 21973
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.5799899697303772,
      "learning_rate": 0.0005866346114624322,
      "loss": 2.3673,
      "step": 21974
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.8251888751983643,
      "learning_rate": 0.0005866334040769123,
      "loss": 2.4503,
      "step": 21975
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6146422028541565,
      "learning_rate": 0.0005866321966381017,
      "loss": 2.3479,
      "step": 21976
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6704283356666565,
      "learning_rate": 0.0005866309891460009,
      "loss": 2.2935,
      "step": 21977
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6867198944091797,
      "learning_rate": 0.0005866297816006099,
      "loss": 2.4456,
      "step": 21978
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.8637245893478394,
      "learning_rate": 0.0005866285740019292,
      "loss": 2.2708,
      "step": 21979
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6784806847572327,
      "learning_rate": 0.0005866273663499587,
      "loss": 2.3729,
      "step": 21980
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.808580756187439,
      "learning_rate": 0.0005866261586446988,
      "loss": 2.3813,
      "step": 21981
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.73880535364151,
      "learning_rate": 0.0005866249508861496,
      "loss": 2.7526,
      "step": 21982
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6104072332382202,
      "learning_rate": 0.0005866237430743115,
      "loss": 2.2378,
      "step": 21983
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6908884644508362,
      "learning_rate": 0.0005866225352091846,
      "loss": 2.3214,
      "step": 21984
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6904217600822449,
      "learning_rate": 0.0005866213272907691,
      "loss": 2.2912,
      "step": 21985
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6316170692443848,
      "learning_rate": 0.0005866201193190654,
      "loss": 2.5261,
      "step": 21986
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6921656727790833,
      "learning_rate": 0.0005866189112940735,
      "loss": 2.4812,
      "step": 21987
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6715486645698547,
      "learning_rate": 0.0005866177032157938,
      "loss": 2.1839,
      "step": 21988
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6069977879524231,
      "learning_rate": 0.0005866164950842265,
      "loss": 2.3183,
      "step": 21989
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6160703897476196,
      "learning_rate": 0.0005866152868993717,
      "loss": 2.319,
      "step": 21990
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.5939936637878418,
      "learning_rate": 0.0005866140786612297,
      "loss": 2.3512,
      "step": 21991
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6534346342086792,
      "learning_rate": 0.0005866128703698008,
      "loss": 2.48,
      "step": 21992
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.0687272548675537,
      "learning_rate": 0.0005866116620250852,
      "loss": 2.3889,
      "step": 21993
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.5798852443695068,
      "learning_rate": 0.0005866104536270829,
      "loss": 2.5641,
      "step": 21994
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6546042561531067,
      "learning_rate": 0.0005866092451757945,
      "loss": 2.2963,
      "step": 21995
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6365757584571838,
      "learning_rate": 0.00058660803667122,
      "loss": 2.3177,
      "step": 21996
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6783560514450073,
      "learning_rate": 0.0005866068281133595,
      "loss": 2.2992,
      "step": 21997
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6102325320243835,
      "learning_rate": 0.0005866056195022136,
      "loss": 2.6935,
      "step": 21998
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6410807371139526,
      "learning_rate": 0.0005866044108377822,
      "loss": 2.137,
      "step": 21999
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.7183986306190491,
      "learning_rate": 0.0005866032021200656,
      "loss": 2.2799,
      "step": 22000
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.617426872253418,
      "learning_rate": 0.000586601993349064,
      "loss": 2.2743,
      "step": 22001
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.7526170611381531,
      "learning_rate": 0.0005866007845247778,
      "loss": 2.2725,
      "step": 22002
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6247989535331726,
      "learning_rate": 0.000586599575647207,
      "loss": 2.313,
      "step": 22003
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.7915392518043518,
      "learning_rate": 0.0005865983667163521,
      "loss": 2.2549,
      "step": 22004
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6724074482917786,
      "learning_rate": 0.0005865971577322129,
      "loss": 2.3281,
      "step": 22005
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.706139326095581,
      "learning_rate": 0.0005865959486947901,
      "loss": 2.4481,
      "step": 22006
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.7229531407356262,
      "learning_rate": 0.0005865947396040837,
      "loss": 2.2962,
      "step": 22007
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.7363561987876892,
      "learning_rate": 0.0005865935304600938,
      "loss": 2.2405,
      "step": 22008
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.7403717041015625,
      "learning_rate": 0.0005865923212628207,
      "loss": 2.2473,
      "step": 22009
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6332787275314331,
      "learning_rate": 0.0005865911120122649,
      "loss": 2.418,
      "step": 22010
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.586155116558075,
      "learning_rate": 0.0005865899027084263,
      "loss": 2.5109,
      "step": 22011
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.582997739315033,
      "learning_rate": 0.0005865886933513052,
      "loss": 2.4016,
      "step": 22012
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6200946569442749,
      "learning_rate": 0.0005865874839409019,
      "loss": 2.1468,
      "step": 22013
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6902689337730408,
      "learning_rate": 0.0005865862744772165,
      "loss": 2.2218,
      "step": 22014
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.599770188331604,
      "learning_rate": 0.0005865850649602493,
      "loss": 2.2255,
      "step": 22015
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6433429718017578,
      "learning_rate": 0.0005865838553900007,
      "loss": 2.1452,
      "step": 22016
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6539743542671204,
      "learning_rate": 0.0005865826457664706,
      "loss": 2.2083,
      "step": 22017
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.655921995639801,
      "learning_rate": 0.0005865814360896594,
      "loss": 2.3018,
      "step": 22018
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6268081068992615,
      "learning_rate": 0.0005865802263595672,
      "loss": 2.206,
      "step": 22019
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.5944958329200745,
      "learning_rate": 0.0005865790165761945,
      "loss": 2.2819,
      "step": 22020
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6868929862976074,
      "learning_rate": 0.0005865778067395413,
      "loss": 2.4237,
      "step": 22021
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.5929145812988281,
      "learning_rate": 0.0005865765968496078,
      "loss": 2.3016,
      "step": 22022
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6650186777114868,
      "learning_rate": 0.0005865753869063944,
      "loss": 2.3972,
      "step": 22023
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.8080282211303711,
      "learning_rate": 0.0005865741769099012,
      "loss": 2.218,
      "step": 22024
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6243386268615723,
      "learning_rate": 0.0005865729668601284,
      "loss": 2.0485,
      "step": 22025
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.814490556716919,
      "learning_rate": 0.0005865717567570763,
      "loss": 2.3167,
      "step": 22026
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6845512390136719,
      "learning_rate": 0.000586570546600745,
      "loss": 2.2949,
      "step": 22027
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.7243284583091736,
      "learning_rate": 0.000586569336391135,
      "loss": 2.303,
      "step": 22028
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6895156502723694,
      "learning_rate": 0.0005865681261282463,
      "loss": 2.3267,
      "step": 22029
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.709820032119751,
      "learning_rate": 0.0005865669158120792,
      "loss": 2.3707,
      "step": 22030
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.7531311511993408,
      "learning_rate": 0.0005865657054426339,
      "loss": 2.3642,
      "step": 22031
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.5984776616096497,
      "learning_rate": 0.0005865644950199105,
      "loss": 2.2886,
      "step": 22032
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6397123336791992,
      "learning_rate": 0.0005865632845439095,
      "loss": 2.3021,
      "step": 22033
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.596984326839447,
      "learning_rate": 0.0005865620740146309,
      "loss": 2.5266,
      "step": 22034
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6533368229866028,
      "learning_rate": 0.000586560863432075,
      "loss": 2.3481,
      "step": 22035
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.5740128755569458,
      "learning_rate": 0.000586559652796242,
      "loss": 2.3575,
      "step": 22036
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6189271211624146,
      "learning_rate": 0.0005865584421071323,
      "loss": 2.5447,
      "step": 22037
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6680055856704712,
      "learning_rate": 0.0005865572313647459,
      "loss": 2.2089,
      "step": 22038
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6049551367759705,
      "learning_rate": 0.000586556020569083,
      "loss": 2.2615,
      "step": 22039
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6592182517051697,
      "learning_rate": 0.0005865548097201441,
      "loss": 2.3763,
      "step": 22040
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.5584405660629272,
      "learning_rate": 0.000586553598817929,
      "loss": 2.5072,
      "step": 22041
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6446685791015625,
      "learning_rate": 0.0005865523878624384,
      "loss": 2.3857,
      "step": 22042
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.5811766982078552,
      "learning_rate": 0.0005865511768536723,
      "loss": 2.3401,
      "step": 22043
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6059805154800415,
      "learning_rate": 0.0005865499657916308,
      "loss": 2.5968,
      "step": 22044
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6076182126998901,
      "learning_rate": 0.0005865487546763143,
      "loss": 2.6404,
      "step": 22045
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6373373866081238,
      "learning_rate": 0.000586547543507723,
      "loss": 2.314,
      "step": 22046
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6216115355491638,
      "learning_rate": 0.0005865463322858571,
      "loss": 2.4272,
      "step": 22047
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6096190810203552,
      "learning_rate": 0.0005865451210107168,
      "loss": 2.4046,
      "step": 22048
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6532699465751648,
      "learning_rate": 0.0005865439096823024,
      "loss": 2.1776,
      "step": 22049
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6628851294517517,
      "learning_rate": 0.000586542698300614,
      "loss": 2.3206,
      "step": 22050
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6643299460411072,
      "learning_rate": 0.000586541486865652,
      "loss": 2.2678,
      "step": 22051
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6645070910453796,
      "learning_rate": 0.0005865402753774165,
      "loss": 2.2504,
      "step": 22052
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.8763129711151123,
      "learning_rate": 0.0005865390638359078,
      "loss": 2.6508,
      "step": 22053
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6344942450523376,
      "learning_rate": 0.0005865378522411259,
      "loss": 2.4409,
      "step": 22054
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6688067317008972,
      "learning_rate": 0.0005865366405930713,
      "loss": 2.4185,
      "step": 22055
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6174025535583496,
      "learning_rate": 0.0005865354288917442,
      "loss": 2.2152,
      "step": 22056
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6197236180305481,
      "learning_rate": 0.0005865342171371447,
      "loss": 2.091,
      "step": 22057
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.7178435921669006,
      "learning_rate": 0.000586533005329273,
      "loss": 2.2691,
      "step": 22058
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6486182808876038,
      "learning_rate": 0.0005865317934681295,
      "loss": 2.206,
      "step": 22059
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.7316755056381226,
      "learning_rate": 0.0005865305815537144,
      "loss": 2.4976,
      "step": 22060
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6074167490005493,
      "learning_rate": 0.0005865293695860277,
      "loss": 2.6268,
      "step": 22061
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.9170013666152954,
      "learning_rate": 0.0005865281575650699,
      "loss": 2.6185,
      "step": 22062
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6413376927375793,
      "learning_rate": 0.0005865269454908411,
      "loss": 2.2645,
      "step": 22063
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.7654488682746887,
      "learning_rate": 0.0005865257333633415,
      "loss": 2.1649,
      "step": 22064
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6384376287460327,
      "learning_rate": 0.0005865245211825713,
      "loss": 2.3298,
      "step": 22065
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6120893955230713,
      "learning_rate": 0.0005865233089485309,
      "loss": 2.4189,
      "step": 22066
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.5973191261291504,
      "learning_rate": 0.0005865220966612204,
      "loss": 2.4597,
      "step": 22067
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.679226279258728,
      "learning_rate": 0.0005865208843206399,
      "loss": 2.3594,
      "step": 22068
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.7107590436935425,
      "learning_rate": 0.0005865196719267898,
      "loss": 2.2404,
      "step": 22069
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.5642511248588562,
      "learning_rate": 0.0005865184594796704,
      "loss": 2.2224,
      "step": 22070
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.7157425284385681,
      "learning_rate": 0.0005865172469792817,
      "loss": 2.2334,
      "step": 22071
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6482813954353333,
      "learning_rate": 0.0005865160344256241,
      "loss": 2.453,
      "step": 22072
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6186397075653076,
      "learning_rate": 0.0005865148218186977,
      "loss": 2.2474,
      "step": 22073
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6336362361907959,
      "learning_rate": 0.0005865136091585028,
      "loss": 2.4315,
      "step": 22074
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.7258223295211792,
      "learning_rate": 0.0005865123964450396,
      "loss": 2.3363,
      "step": 22075
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.87944096326828,
      "learning_rate": 0.0005865111836783084,
      "loss": 2.0371,
      "step": 22076
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.7217679023742676,
      "learning_rate": 0.0005865099708583094,
      "loss": 2.4005,
      "step": 22077
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.7124618887901306,
      "learning_rate": 0.0005865087579850426,
      "loss": 2.4798,
      "step": 22078
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6438264846801758,
      "learning_rate": 0.0005865075450585086,
      "loss": 2.3008,
      "step": 22079
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6082393527030945,
      "learning_rate": 0.0005865063320787073,
      "loss": 2.3183,
      "step": 22080
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.5640751719474792,
      "learning_rate": 0.0005865051190456392,
      "loss": 2.5326,
      "step": 22081
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6712155938148499,
      "learning_rate": 0.0005865039059593043,
      "loss": 2.4151,
      "step": 22082
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6694757342338562,
      "learning_rate": 0.0005865026928197029,
      "loss": 2.3815,
      "step": 22083
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.5799570679664612,
      "learning_rate": 0.0005865014796268353,
      "loss": 2.3245,
      "step": 22084
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.627565860748291,
      "learning_rate": 0.0005865002663807017,
      "loss": 2.3233,
      "step": 22085
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6323362588882446,
      "learning_rate": 0.0005864990530813022,
      "loss": 2.3182,
      "step": 22086
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.629801332950592,
      "learning_rate": 0.0005864978397286372,
      "loss": 2.1934,
      "step": 22087
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6217613220214844,
      "learning_rate": 0.0005864966263227067,
      "loss": 2.2426,
      "step": 22088
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.5620179176330566,
      "learning_rate": 0.0005864954128635112,
      "loss": 2.3343,
      "step": 22089
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6904541850090027,
      "learning_rate": 0.0005864941993510508,
      "loss": 2.3454,
      "step": 22090
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6918346285820007,
      "learning_rate": 0.0005864929857853256,
      "loss": 2.247,
      "step": 22091
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6119025945663452,
      "learning_rate": 0.0005864917721663362,
      "loss": 2.4114,
      "step": 22092
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.5942320823669434,
      "learning_rate": 0.0005864905584940823,
      "loss": 2.4962,
      "step": 22093
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.8709026575088501,
      "learning_rate": 0.0005864893447685645,
      "loss": 2.1475,
      "step": 22094
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6941127777099609,
      "learning_rate": 0.000586488130989783,
      "loss": 2.2645,
      "step": 22095
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6869032382965088,
      "learning_rate": 0.0005864869171577379,
      "loss": 2.2804,
      "step": 22096
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.7428769469261169,
      "learning_rate": 0.0005864857032724293,
      "loss": 2.5718,
      "step": 22097
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6503232717514038,
      "learning_rate": 0.0005864844893338578,
      "loss": 2.2601,
      "step": 22098
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6284470558166504,
      "learning_rate": 0.0005864832753420233,
      "loss": 2.4347,
      "step": 22099
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6972184181213379,
      "learning_rate": 0.0005864820612969263,
      "loss": 2.4338,
      "step": 22100
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.7489049434661865,
      "learning_rate": 0.0005864808471985668,
      "loss": 2.2108,
      "step": 22101
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.5631019473075867,
      "learning_rate": 0.000586479633046945,
      "loss": 2.3086,
      "step": 22102
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6327385306358337,
      "learning_rate": 0.0005864784188420614,
      "loss": 2.5548,
      "step": 22103
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6168259382247925,
      "learning_rate": 0.0005864772045839161,
      "loss": 2.5731,
      "step": 22104
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6747860908508301,
      "learning_rate": 0.0005864759902725091,
      "loss": 2.0882,
      "step": 22105
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6111578941345215,
      "learning_rate": 0.0005864747759078408,
      "loss": 2.5313,
      "step": 22106
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6251558661460876,
      "learning_rate": 0.0005864735614899115,
      "loss": 2.4682,
      "step": 22107
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.8584100604057312,
      "learning_rate": 0.0005864723470187214,
      "loss": 2.3835,
      "step": 22108
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6319474577903748,
      "learning_rate": 0.0005864711324942706,
      "loss": 2.2078,
      "step": 22109
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.5965981483459473,
      "learning_rate": 0.0005864699179165595,
      "loss": 2.3609,
      "step": 22110
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6555898785591125,
      "learning_rate": 0.0005864687032855882,
      "loss": 2.2358,
      "step": 22111
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.8022245764732361,
      "learning_rate": 0.000586467488601357,
      "loss": 2.3556,
      "step": 22112
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6496621966362,
      "learning_rate": 0.000586466273863866,
      "loss": 2.7579,
      "step": 22113
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6479076147079468,
      "learning_rate": 0.0005864650590731155,
      "loss": 2.332,
      "step": 22114
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6573375463485718,
      "learning_rate": 0.0005864638442291059,
      "loss": 2.5721,
      "step": 22115
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6481090188026428,
      "learning_rate": 0.000586462629331837,
      "loss": 2.3698,
      "step": 22116
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.5880855321884155,
      "learning_rate": 0.0005864614143813095,
      "loss": 2.4002,
      "step": 22117
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6084222197532654,
      "learning_rate": 0.0005864601993775234,
      "loss": 2.4285,
      "step": 22118
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.664182722568512,
      "learning_rate": 0.0005864589843204788,
      "loss": 2.5062,
      "step": 22119
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6847338676452637,
      "learning_rate": 0.0005864577692101762,
      "loss": 2.4163,
      "step": 22120
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6472575068473816,
      "learning_rate": 0.0005864565540466156,
      "loss": 2.5058,
      "step": 22121
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.611926794052124,
      "learning_rate": 0.0005864553388297975,
      "loss": 2.4258,
      "step": 22122
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.5724714994430542,
      "learning_rate": 0.0005864541235597217,
      "loss": 2.266,
      "step": 22123
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6152317523956299,
      "learning_rate": 0.0005864529082363888,
      "loss": 2.3074,
      "step": 22124
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6190387606620789,
      "learning_rate": 0.0005864516928597989,
      "loss": 2.382,
      "step": 22125
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6897021532058716,
      "learning_rate": 0.0005864504774299523,
      "loss": 2.4763,
      "step": 22126
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6792134642601013,
      "learning_rate": 0.000586449261946849,
      "loss": 2.1667,
      "step": 22127
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.5964236259460449,
      "learning_rate": 0.0005864480464104894,
      "loss": 2.2435,
      "step": 22128
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.5931342244148254,
      "learning_rate": 0.0005864468308208738,
      "loss": 2.2671,
      "step": 22129
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6488432288169861,
      "learning_rate": 0.0005864456151780023,
      "loss": 2.4797,
      "step": 22130
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.7108564376831055,
      "learning_rate": 0.0005864443994818751,
      "loss": 2.3138,
      "step": 22131
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6543397903442383,
      "learning_rate": 0.0005864431837324925,
      "loss": 2.4877,
      "step": 22132
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.7094582319259644,
      "learning_rate": 0.0005864419679298547,
      "loss": 2.4897,
      "step": 22133
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6048556566238403,
      "learning_rate": 0.0005864407520739618,
      "loss": 2.4542,
      "step": 22134
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.5846845507621765,
      "learning_rate": 0.0005864395361648144,
      "loss": 2.3358,
      "step": 22135
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.5983158946037292,
      "learning_rate": 0.0005864383202024124,
      "loss": 2.3463,
      "step": 22136
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.5961453318595886,
      "learning_rate": 0.000586437104186756,
      "loss": 2.2419,
      "step": 22137
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6171978712081909,
      "learning_rate": 0.0005864358881178457,
      "loss": 2.4593,
      "step": 22138
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6607473492622375,
      "learning_rate": 0.0005864346719956815,
      "loss": 2.6916,
      "step": 22139
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6008761525154114,
      "learning_rate": 0.0005864334558202636,
      "loss": 2.3763,
      "step": 22140
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6487520933151245,
      "learning_rate": 0.0005864322395915923,
      "loss": 2.488,
      "step": 22141
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6180583238601685,
      "learning_rate": 0.000586431023309668,
      "loss": 2.0077,
      "step": 22142
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.5902778506278992,
      "learning_rate": 0.0005864298069744908,
      "loss": 2.3651,
      "step": 22143
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.5770993828773499,
      "learning_rate": 0.0005864285905860607,
      "loss": 2.2498,
      "step": 22144
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.5899266600608826,
      "learning_rate": 0.0005864273741443782,
      "loss": 2.1574,
      "step": 22145
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6921943426132202,
      "learning_rate": 0.0005864261576494433,
      "loss": 2.5116,
      "step": 22146
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6167530417442322,
      "learning_rate": 0.0005864249411012567,
      "loss": 2.5241,
      "step": 22147
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6321101784706116,
      "learning_rate": 0.000586423724499818,
      "loss": 2.4096,
      "step": 22148
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.623619794845581,
      "learning_rate": 0.0005864225078451279,
      "loss": 2.3941,
      "step": 22149
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6914361715316772,
      "learning_rate": 0.0005864212911371864,
      "loss": 2.3216,
      "step": 22150
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6276997327804565,
      "learning_rate": 0.0005864200743759938,
      "loss": 2.3599,
      "step": 22151
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.5846766829490662,
      "learning_rate": 0.0005864188575615501,
      "loss": 2.5065,
      "step": 22152
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.632027268409729,
      "learning_rate": 0.0005864176406938559,
      "loss": 2.2378,
      "step": 22153
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6876674890518188,
      "learning_rate": 0.0005864164237729113,
      "loss": 2.4575,
      "step": 22154
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6407703757286072,
      "learning_rate": 0.0005864152067987164,
      "loss": 2.2867,
      "step": 22155
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.5937644839286804,
      "learning_rate": 0.0005864139897712715,
      "loss": 2.2845,
      "step": 22156
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6132723689079285,
      "learning_rate": 0.0005864127726905769,
      "loss": 2.5344,
      "step": 22157
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.5689875483512878,
      "learning_rate": 0.0005864115555566327,
      "loss": 2.3302,
      "step": 22158
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6104567050933838,
      "learning_rate": 0.0005864103383694392,
      "loss": 2.3136,
      "step": 22159
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6266356110572815,
      "learning_rate": 0.0005864091211289966,
      "loss": 2.5182,
      "step": 22160
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6801270246505737,
      "learning_rate": 0.0005864079038353052,
      "loss": 2.3646,
      "step": 22161
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6310006976127625,
      "learning_rate": 0.0005864066864883651,
      "loss": 2.3056,
      "step": 22162
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.5696709156036377,
      "learning_rate": 0.0005864054690881766,
      "loss": 2.3804,
      "step": 22163
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6153964996337891,
      "learning_rate": 0.00058640425163474,
      "loss": 2.5408,
      "step": 22164
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.602310061454773,
      "learning_rate": 0.0005864030341280553,
      "loss": 2.3264,
      "step": 22165
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.623336911201477,
      "learning_rate": 0.000586401816568123,
      "loss": 2.2661,
      "step": 22166
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6501767635345459,
      "learning_rate": 0.0005864005989549431,
      "loss": 2.4576,
      "step": 22167
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6212320923805237,
      "learning_rate": 0.000586399381288516,
      "loss": 2.245,
      "step": 22168
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6486644148826599,
      "learning_rate": 0.0005863981635688418,
      "loss": 2.3579,
      "step": 22169
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.654637336730957,
      "learning_rate": 0.0005863969457959207,
      "loss": 2.2507,
      "step": 22170
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.615244448184967,
      "learning_rate": 0.0005863957279697531,
      "loss": 2.3959,
      "step": 22171
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.5957436561584473,
      "learning_rate": 0.0005863945100903391,
      "loss": 2.3903,
      "step": 22172
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6797547936439514,
      "learning_rate": 0.000586393292157679,
      "loss": 2.3385,
      "step": 22173
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.5955123901367188,
      "learning_rate": 0.0005863920741717729,
      "loss": 2.4713,
      "step": 22174
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6571951508522034,
      "learning_rate": 0.0005863908561326212,
      "loss": 2.5088,
      "step": 22175
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6356348395347595,
      "learning_rate": 0.0005863896380402239,
      "loss": 2.3983,
      "step": 22176
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.5848836302757263,
      "learning_rate": 0.0005863884198945815,
      "loss": 2.0822,
      "step": 22177
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6769452095031738,
      "learning_rate": 0.0005863872016956941,
      "loss": 2.4351,
      "step": 22178
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6030338406562805,
      "learning_rate": 0.0005863859834435618,
      "loss": 2.2531,
      "step": 22179
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6078559160232544,
      "learning_rate": 0.0005863847651381849,
      "loss": 2.3379,
      "step": 22180
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6842253804206848,
      "learning_rate": 0.0005863835467795638,
      "loss": 2.431,
      "step": 22181
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6119009256362915,
      "learning_rate": 0.0005863823283676986,
      "loss": 2.3244,
      "step": 22182
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6694185733795166,
      "learning_rate": 0.0005863811099025894,
      "loss": 2.5665,
      "step": 22183
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6392660737037659,
      "learning_rate": 0.0005863798913842367,
      "loss": 2.1915,
      "step": 22184
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6980384588241577,
      "learning_rate": 0.0005863786728126405,
      "loss": 2.3453,
      "step": 22185
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6164600849151611,
      "learning_rate": 0.000586377454187801,
      "loss": 2.5475,
      "step": 22186
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.579829216003418,
      "learning_rate": 0.0005863762355097186,
      "loss": 2.1879,
      "step": 22187
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.5615344047546387,
      "learning_rate": 0.0005863750167783935,
      "loss": 2.2761,
      "step": 22188
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.5875815153121948,
      "learning_rate": 0.0005863737979938258,
      "loss": 2.2515,
      "step": 22189
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6352855563163757,
      "learning_rate": 0.0005863725791560158,
      "loss": 2.3679,
      "step": 22190
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.5991148352622986,
      "learning_rate": 0.0005863713602649638,
      "loss": 2.499,
      "step": 22191
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6255800724029541,
      "learning_rate": 0.00058637014132067,
      "loss": 2.3257,
      "step": 22192
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.5969159603118896,
      "learning_rate": 0.0005863689223231344,
      "loss": 2.3628,
      "step": 22193
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.5745090246200562,
      "learning_rate": 0.0005863677032723576,
      "loss": 2.3191,
      "step": 22194
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6085887551307678,
      "learning_rate": 0.0005863664841683396,
      "loss": 2.2958,
      "step": 22195
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6223925948143005,
      "learning_rate": 0.0005863652650110806,
      "loss": 2.4005,
      "step": 22196
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6558571457862854,
      "learning_rate": 0.0005863640458005809,
      "loss": 2.5095,
      "step": 22197
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.5478860139846802,
      "learning_rate": 0.0005863628265368408,
      "loss": 2.2496,
      "step": 22198
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6959918141365051,
      "learning_rate": 0.0005863616072198604,
      "loss": 2.2949,
      "step": 22199
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6492296457290649,
      "learning_rate": 0.00058636038784964,
      "loss": 2.3413,
      "step": 22200
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6203964948654175,
      "learning_rate": 0.0005863591684261796,
      "loss": 2.2547,
      "step": 22201
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6032686233520508,
      "learning_rate": 0.0005863579489494799,
      "loss": 2.2323,
      "step": 22202
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6622941493988037,
      "learning_rate": 0.0005863567294195407,
      "loss": 2.2039,
      "step": 22203
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.5824435353279114,
      "learning_rate": 0.0005863555098363625,
      "loss": 2.325,
      "step": 22204
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6636731624603271,
      "learning_rate": 0.0005863542901999452,
      "loss": 2.3227,
      "step": 22205
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6540135741233826,
      "learning_rate": 0.0005863530705102894,
      "loss": 2.2697,
      "step": 22206
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6449575424194336,
      "learning_rate": 0.0005863518507673952,
      "loss": 2.3459,
      "step": 22207
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6083142757415771,
      "learning_rate": 0.0005863506309712627,
      "loss": 2.3243,
      "step": 22208
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.7187841534614563,
      "learning_rate": 0.0005863494111218921,
      "loss": 2.2985,
      "step": 22209
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6185686588287354,
      "learning_rate": 0.0005863481912192839,
      "loss": 2.2603,
      "step": 22210
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6698963642120361,
      "learning_rate": 0.0005863469712634381,
      "loss": 2.33,
      "step": 22211
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6308958530426025,
      "learning_rate": 0.000586345751254355,
      "loss": 2.3293,
      "step": 22212
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6238777041435242,
      "learning_rate": 0.0005863445311920347,
      "loss": 2.3854,
      "step": 22213
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.5662353038787842,
      "learning_rate": 0.0005863433110764777,
      "loss": 2.1051,
      "step": 22214
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6589636206626892,
      "learning_rate": 0.0005863420909076842,
      "loss": 2.4791,
      "step": 22215
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.5835100412368774,
      "learning_rate": 0.0005863408706856541,
      "loss": 2.3427,
      "step": 22216
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6066339015960693,
      "learning_rate": 0.0005863396504103878,
      "loss": 2.3293,
      "step": 22217
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6866388320922852,
      "learning_rate": 0.0005863384300818856,
      "loss": 2.3772,
      "step": 22218
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.5926928520202637,
      "learning_rate": 0.0005863372097001478,
      "loss": 2.5004,
      "step": 22219
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.5672058463096619,
      "learning_rate": 0.0005863359892651744,
      "loss": 2.1527,
      "step": 22220
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6203522682189941,
      "learning_rate": 0.0005863347687769658,
      "loss": 2.1757,
      "step": 22221
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.5889682173728943,
      "learning_rate": 0.0005863335482355221,
      "loss": 2.5062,
      "step": 22222
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.8112678527832031,
      "learning_rate": 0.0005863323276408436,
      "loss": 2.2537,
      "step": 22223
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6384578943252563,
      "learning_rate": 0.0005863311069929304,
      "loss": 2.3541,
      "step": 22224
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6415631175041199,
      "learning_rate": 0.000586329886291783,
      "loss": 2.227,
      "step": 22225
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.8024395108222961,
      "learning_rate": 0.0005863286655374014,
      "loss": 2.1119,
      "step": 22226
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6350398063659668,
      "learning_rate": 0.000586327444729786,
      "loss": 2.6034,
      "step": 22227
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.7433761954307556,
      "learning_rate": 0.0005863262238689369,
      "loss": 2.1405,
      "step": 22228
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.7718763947486877,
      "learning_rate": 0.0005863250029548542,
      "loss": 2.241,
      "step": 22229
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.8208304047584534,
      "learning_rate": 0.0005863237819875384,
      "loss": 2.2509,
      "step": 22230
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.0980396270751953,
      "learning_rate": 0.0005863225609669896,
      "loss": 2.2678,
      "step": 22231
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.7082228660583496,
      "learning_rate": 0.000586321339893208,
      "loss": 2.2207,
      "step": 22232
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6269805431365967,
      "learning_rate": 0.0005863201187661939,
      "loss": 2.1299,
      "step": 22233
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.9706161022186279,
      "learning_rate": 0.0005863188975859474,
      "loss": 2.4913,
      "step": 22234
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6825557351112366,
      "learning_rate": 0.0005863176763524689,
      "loss": 2.4525,
      "step": 22235
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.597576379776001,
      "learning_rate": 0.0005863164550657585,
      "loss": 2.3044,
      "step": 22236
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.918541431427002,
      "learning_rate": 0.0005863152337258165,
      "loss": 2.4834,
      "step": 22237
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6968344449996948,
      "learning_rate": 0.000586314012332643,
      "loss": 2.2271,
      "step": 22238
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.5486655831336975,
      "learning_rate": 0.0005863127908862384,
      "loss": 2.2729,
      "step": 22239
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.712575376033783,
      "learning_rate": 0.0005863115693866028,
      "loss": 2.408,
      "step": 22240
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.8161788582801819,
      "learning_rate": 0.0005863103478337366,
      "loss": 2.5859,
      "step": 22241
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.5336196422576904,
      "learning_rate": 0.0005863091262276397,
      "loss": 2.1274,
      "step": 22242
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6094899773597717,
      "learning_rate": 0.0005863079045683127,
      "loss": 2.3264,
      "step": 22243
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.8790073990821838,
      "learning_rate": 0.0005863066828557555,
      "loss": 2.3052,
      "step": 22244
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6142553091049194,
      "learning_rate": 0.0005863054610899686,
      "loss": 2.1449,
      "step": 22245
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6346596479415894,
      "learning_rate": 0.0005863042392709522,
      "loss": 2.3538,
      "step": 22246
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.7694063782691956,
      "learning_rate": 0.0005863030173987063,
      "loss": 2.1506,
      "step": 22247
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.5712594389915466,
      "learning_rate": 0.0005863017954732312,
      "loss": 2.2781,
      "step": 22248
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6243980526924133,
      "learning_rate": 0.0005863005734945274,
      "loss": 2.3119,
      "step": 22249
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.7909961342811584,
      "learning_rate": 0.0005862993514625948,
      "loss": 2.4963,
      "step": 22250
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6229500770568848,
      "learning_rate": 0.0005862981293774336,
      "loss": 2.4112,
      "step": 22251
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6685925722122192,
      "learning_rate": 0.0005862969072390443,
      "loss": 2.3187,
      "step": 22252
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6097951531410217,
      "learning_rate": 0.0005862956850474271,
      "loss": 2.1289,
      "step": 22253
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.7507815957069397,
      "learning_rate": 0.0005862944628025821,
      "loss": 2.2808,
      "step": 22254
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.7311871647834778,
      "learning_rate": 0.0005862932405045094,
      "loss": 2.4121,
      "step": 22255
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6245905160903931,
      "learning_rate": 0.0005862920181532095,
      "loss": 2.505,
      "step": 22256
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.7647819519042969,
      "learning_rate": 0.0005862907957486826,
      "loss": 2.3343,
      "step": 22257
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6248877048492432,
      "learning_rate": 0.0005862895732909288,
      "loss": 2.2962,
      "step": 22258
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6218901872634888,
      "learning_rate": 0.0005862883507799482,
      "loss": 2.1927,
      "step": 22259
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6352114677429199,
      "learning_rate": 0.0005862871282157413,
      "loss": 2.3765,
      "step": 22260
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6632822155952454,
      "learning_rate": 0.0005862859055983082,
      "loss": 2.1838,
      "step": 22261
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6207400560379028,
      "learning_rate": 0.0005862846829276491,
      "loss": 2.4018,
      "step": 22262
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6155230402946472,
      "learning_rate": 0.0005862834602037644,
      "loss": 2.5495,
      "step": 22263
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6634040474891663,
      "learning_rate": 0.0005862822374266541,
      "loss": 2.3768,
      "step": 22264
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6927944421768188,
      "learning_rate": 0.0005862810145963185,
      "loss": 2.4178,
      "step": 22265
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.7797200083732605,
      "learning_rate": 0.0005862797917127581,
      "loss": 2.4461,
      "step": 22266
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.7054208517074585,
      "learning_rate": 0.0005862785687759726,
      "loss": 2.182,
      "step": 22267
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6245810985565186,
      "learning_rate": 0.0005862773457859625,
      "loss": 2.4092,
      "step": 22268
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6010891795158386,
      "learning_rate": 0.0005862761227427282,
      "loss": 2.5216,
      "step": 22269
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.5575035214424133,
      "learning_rate": 0.0005862748996462697,
      "loss": 2.4962,
      "step": 22270
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.5980758666992188,
      "learning_rate": 0.0005862736764965874,
      "loss": 2.3306,
      "step": 22271
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6236733198165894,
      "learning_rate": 0.0005862724532936812,
      "loss": 2.3799,
      "step": 22272
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6606125235557556,
      "learning_rate": 0.0005862712300375517,
      "loss": 2.4235,
      "step": 22273
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6237185001373291,
      "learning_rate": 0.0005862700067281989,
      "loss": 2.57,
      "step": 22274
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6142688989639282,
      "learning_rate": 0.0005862687833656233,
      "loss": 2.6195,
      "step": 22275
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.567267894744873,
      "learning_rate": 0.0005862675599498247,
      "loss": 2.4027,
      "step": 22276
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6122021079063416,
      "learning_rate": 0.0005862663364808036,
      "loss": 2.1736,
      "step": 22277
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6101768016815186,
      "learning_rate": 0.0005862651129585602,
      "loss": 2.3245,
      "step": 22278
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.650804877281189,
      "learning_rate": 0.0005862638893830947,
      "loss": 2.255,
      "step": 22279
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.5881638526916504,
      "learning_rate": 0.0005862626657544074,
      "loss": 2.4675,
      "step": 22280
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6274135112762451,
      "learning_rate": 0.0005862614420724984,
      "loss": 2.4838,
      "step": 22281
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6867254376411438,
      "learning_rate": 0.0005862602183373681,
      "loss": 2.492,
      "step": 22282
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6378814578056335,
      "learning_rate": 0.0005862589945490165,
      "loss": 2.4313,
      "step": 22283
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6934277415275574,
      "learning_rate": 0.000586257770707444,
      "loss": 2.307,
      "step": 22284
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6669415831565857,
      "learning_rate": 0.0005862565468126508,
      "loss": 2.2416,
      "step": 22285
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.5795162320137024,
      "learning_rate": 0.000586255322864637,
      "loss": 2.656,
      "step": 22286
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.7491625547409058,
      "learning_rate": 0.0005862540988634031,
      "loss": 2.4153,
      "step": 22287
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6250505447387695,
      "learning_rate": 0.000586252874808949,
      "loss": 2.2719,
      "step": 22288
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.604394793510437,
      "learning_rate": 0.0005862516507012752,
      "loss": 2.3182,
      "step": 22289
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6872200965881348,
      "learning_rate": 0.0005862504265403818,
      "loss": 2.591,
      "step": 22290
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6817818284034729,
      "learning_rate": 0.000586249202326269,
      "loss": 2.4348,
      "step": 22291
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.5626243948936462,
      "learning_rate": 0.000586247978058937,
      "loss": 2.2677,
      "step": 22292
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.5974481105804443,
      "learning_rate": 0.0005862467537383862,
      "loss": 2.2159,
      "step": 22293
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6409104466438293,
      "learning_rate": 0.0005862455293646167,
      "loss": 2.3228,
      "step": 22294
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6044275164604187,
      "learning_rate": 0.0005862443049376287,
      "loss": 2.1322,
      "step": 22295
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6247235536575317,
      "learning_rate": 0.0005862430804574226,
      "loss": 2.4692,
      "step": 22296
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6460534930229187,
      "learning_rate": 0.0005862418559239985,
      "loss": 2.2015,
      "step": 22297
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.5606371760368347,
      "learning_rate": 0.0005862406313373566,
      "loss": 2.2889,
      "step": 22298
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6894519925117493,
      "learning_rate": 0.0005862394066974971,
      "loss": 2.2472,
      "step": 22299
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6105765700340271,
      "learning_rate": 0.0005862381820044203,
      "loss": 2.4585,
      "step": 22300
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.636512815952301,
      "learning_rate": 0.0005862369572581265,
      "loss": 2.3008,
      "step": 22301
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6581477522850037,
      "learning_rate": 0.0005862357324586158,
      "loss": 2.4115,
      "step": 22302
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6837928891181946,
      "learning_rate": 0.0005862345076058885,
      "loss": 2.4226,
      "step": 22303
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.7088595032691956,
      "learning_rate": 0.0005862332826999447,
      "loss": 2.1907,
      "step": 22304
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.692220151424408,
      "learning_rate": 0.0005862320577407849,
      "loss": 2.3954,
      "step": 22305
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6981484889984131,
      "learning_rate": 0.000586230832728409,
      "loss": 2.3456,
      "step": 22306
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6296624541282654,
      "learning_rate": 0.0005862296076628175,
      "loss": 2.543,
      "step": 22307
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.5955786108970642,
      "learning_rate": 0.0005862283825440104,
      "loss": 2.4181,
      "step": 22308
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6089245676994324,
      "learning_rate": 0.0005862271573719882,
      "loss": 2.4474,
      "step": 22309
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6905113458633423,
      "learning_rate": 0.0005862259321467509,
      "loss": 2.4015,
      "step": 22310
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.7070464491844177,
      "learning_rate": 0.0005862247068682987,
      "loss": 2.2297,
      "step": 22311
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6629619598388672,
      "learning_rate": 0.0005862234815366319,
      "loss": 2.1342,
      "step": 22312
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.636671781539917,
      "learning_rate": 0.0005862222561517509,
      "loss": 2.1969,
      "step": 22313
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.62665194272995,
      "learning_rate": 0.0005862210307136557,
      "loss": 2.3362,
      "step": 22314
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.8371526002883911,
      "learning_rate": 0.0005862198052223466,
      "loss": 2.3511,
      "step": 22315
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6597988605499268,
      "learning_rate": 0.000586218579677824,
      "loss": 2.3993,
      "step": 22316
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6992006301879883,
      "learning_rate": 0.0005862173540800877,
      "loss": 2.4325,
      "step": 22317
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6701912879943848,
      "learning_rate": 0.0005862161284291384,
      "loss": 2.161,
      "step": 22318
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6125249266624451,
      "learning_rate": 0.0005862149027249761,
      "loss": 2.3837,
      "step": 22319
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6045969128608704,
      "learning_rate": 0.0005862136769676009,
      "loss": 2.3177,
      "step": 22320
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6491397023200989,
      "learning_rate": 0.0005862124511570134,
      "loss": 2.5259,
      "step": 22321
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6003787517547607,
      "learning_rate": 0.0005862112252932133,
      "loss": 2.1828,
      "step": 22322
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.5696667432785034,
      "learning_rate": 0.0005862099993762013,
      "loss": 2.2634,
      "step": 22323
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6250574588775635,
      "learning_rate": 0.0005862087734059776,
      "loss": 2.1919,
      "step": 22324
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6614711880683899,
      "learning_rate": 0.000586207547382542,
      "loss": 2.3484,
      "step": 22325
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.5999390482902527,
      "learning_rate": 0.0005862063213058952,
      "loss": 2.357,
      "step": 22326
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6281364560127258,
      "learning_rate": 0.0005862050951760372,
      "loss": 2.2395,
      "step": 22327
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6216967105865479,
      "learning_rate": 0.0005862038689929683,
      "loss": 2.2337,
      "step": 22328
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6419721245765686,
      "learning_rate": 0.0005862026427566887,
      "loss": 2.2634,
      "step": 22329
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.651543378829956,
      "learning_rate": 0.0005862014164671986,
      "loss": 2.4204,
      "step": 22330
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6465452909469604,
      "learning_rate": 0.0005862001901244981,
      "loss": 2.3677,
      "step": 22331
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6176648736000061,
      "learning_rate": 0.0005861989637285878,
      "loss": 2.2495,
      "step": 22332
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.5751775503158569,
      "learning_rate": 0.0005861977372794676,
      "loss": 2.3047,
      "step": 22333
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.5725627541542053,
      "learning_rate": 0.0005861965107771379,
      "loss": 2.28,
      "step": 22334
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.5929720997810364,
      "learning_rate": 0.0005861952842215989,
      "loss": 2.1792,
      "step": 22335
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.5353869795799255,
      "learning_rate": 0.0005861940576128507,
      "loss": 2.2204,
      "step": 22336
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.9513634443283081,
      "learning_rate": 0.0005861928309508936,
      "loss": 2.3722,
      "step": 22337
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6619614958763123,
      "learning_rate": 0.000586191604235728,
      "loss": 2.3951,
      "step": 22338
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6962717175483704,
      "learning_rate": 0.0005861903774673538,
      "loss": 2.4816,
      "step": 22339
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6931736469268799,
      "learning_rate": 0.0005861891506457716,
      "loss": 2.4165,
      "step": 22340
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6147047281265259,
      "learning_rate": 0.0005861879237709813,
      "loss": 2.2821,
      "step": 22341
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6315364241600037,
      "learning_rate": 0.0005861866968429833,
      "loss": 2.4206,
      "step": 22342
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6481972336769104,
      "learning_rate": 0.0005861854698617778,
      "loss": 2.4114,
      "step": 22343
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6608195304870605,
      "learning_rate": 0.000586184242827365,
      "loss": 2.567,
      "step": 22344
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6343982219696045,
      "learning_rate": 0.0005861830157397451,
      "loss": 2.2654,
      "step": 22345
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6170063018798828,
      "learning_rate": 0.0005861817885989184,
      "loss": 2.3703,
      "step": 22346
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6026688814163208,
      "learning_rate": 0.0005861805614048852,
      "loss": 2.3396,
      "step": 22347
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6379516124725342,
      "learning_rate": 0.0005861793341576456,
      "loss": 2.239,
      "step": 22348
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.5821338295936584,
      "learning_rate": 0.0005861781068571998,
      "loss": 2.2476,
      "step": 22349
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6182034015655518,
      "learning_rate": 0.0005861768795035481,
      "loss": 2.3469,
      "step": 22350
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6306012868881226,
      "learning_rate": 0.0005861756520966906,
      "loss": 2.2319,
      "step": 22351
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.5575228333473206,
      "learning_rate": 0.0005861744246366278,
      "loss": 2.4827,
      "step": 22352
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6140748858451843,
      "learning_rate": 0.0005861731971233598,
      "loss": 2.3472,
      "step": 22353
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.7031579613685608,
      "learning_rate": 0.0005861719695568867,
      "loss": 2.2866,
      "step": 22354
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.7257130742073059,
      "learning_rate": 0.0005861707419372089,
      "loss": 2.2415,
      "step": 22355
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6209709644317627,
      "learning_rate": 0.0005861695142643265,
      "loss": 2.2035,
      "step": 22356
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6381226181983948,
      "learning_rate": 0.0005861682865382398,
      "loss": 2.2671,
      "step": 22357
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.7699988484382629,
      "learning_rate": 0.0005861670587589491,
      "loss": 2.1285,
      "step": 22358
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.5616135001182556,
      "learning_rate": 0.0005861658309264545,
      "loss": 2.3199,
      "step": 22359
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6219530701637268,
      "learning_rate": 0.0005861646030407561,
      "loss": 2.2399,
      "step": 22360
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6060453057289124,
      "learning_rate": 0.0005861633751018545,
      "loss": 2.3113,
      "step": 22361
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6915456056594849,
      "learning_rate": 0.0005861621471097497,
      "loss": 2.5699,
      "step": 22362
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6374472975730896,
      "learning_rate": 0.0005861609190644419,
      "loss": 2.1512,
      "step": 22363
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.5827531218528748,
      "learning_rate": 0.0005861596909659314,
      "loss": 2.3306,
      "step": 22364
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6259410977363586,
      "learning_rate": 0.0005861584628142184,
      "loss": 2.2605,
      "step": 22365
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.7123234272003174,
      "learning_rate": 0.0005861572346093031,
      "loss": 2.6208,
      "step": 22366
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6305265426635742,
      "learning_rate": 0.0005861560063511859,
      "loss": 2.4774,
      "step": 22367
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6419248580932617,
      "learning_rate": 0.0005861547780398669,
      "loss": 2.4044,
      "step": 22368
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6902803778648376,
      "learning_rate": 0.0005861535496753461,
      "loss": 2.3542,
      "step": 22369
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6238833069801331,
      "learning_rate": 0.0005861523212576242,
      "loss": 2.5014,
      "step": 22370
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.682742178440094,
      "learning_rate": 0.0005861510927867011,
      "loss": 2.4107,
      "step": 22371
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6604490876197815,
      "learning_rate": 0.0005861498642625771,
      "loss": 2.3601,
      "step": 22372
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.7637783885002136,
      "learning_rate": 0.0005861486356852524,
      "loss": 2.1095,
      "step": 22373
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.7332249283790588,
      "learning_rate": 0.0005861474070547274,
      "loss": 2.1673,
      "step": 22374
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6703830361366272,
      "learning_rate": 0.000586146178371002,
      "loss": 2.4926,
      "step": 22375
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.8008843660354614,
      "learning_rate": 0.0005861449496340767,
      "loss": 2.3259,
      "step": 22376
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.7096485495567322,
      "learning_rate": 0.0005861437208439516,
      "loss": 2.4358,
      "step": 22377
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.5888844728469849,
      "learning_rate": 0.0005861424920006271,
      "loss": 2.0985,
      "step": 22378
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.7025505900382996,
      "learning_rate": 0.0005861412631041033,
      "loss": 2.3576,
      "step": 22379
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6994945406913757,
      "learning_rate": 0.0005861400341543804,
      "loss": 2.3363,
      "step": 22380
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.7587405443191528,
      "learning_rate": 0.0005861388051514586,
      "loss": 2.5094,
      "step": 22381
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6294442415237427,
      "learning_rate": 0.0005861375760953382,
      "loss": 2.3945,
      "step": 22382
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6355040073394775,
      "learning_rate": 0.0005861363469860196,
      "loss": 2.2373,
      "step": 22383
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6073889136314392,
      "learning_rate": 0.0005861351178235027,
      "loss": 2.2847,
      "step": 22384
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6330854892730713,
      "learning_rate": 0.0005861338886077878,
      "loss": 2.3059,
      "step": 22385
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6326122879981995,
      "learning_rate": 0.0005861326593388754,
      "loss": 2.145,
      "step": 22386
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.635232150554657,
      "learning_rate": 0.0005861314300167655,
      "loss": 2.2579,
      "step": 22387
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6979195475578308,
      "learning_rate": 0.0005861302006414583,
      "loss": 2.4026,
      "step": 22388
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6237610578536987,
      "learning_rate": 0.0005861289712129541,
      "loss": 2.3222,
      "step": 22389
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6862503886222839,
      "learning_rate": 0.0005861277417312531,
      "loss": 2.1926,
      "step": 22390
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6681166887283325,
      "learning_rate": 0.0005861265121963556,
      "loss": 2.3342,
      "step": 22391
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.7134770154953003,
      "learning_rate": 0.0005861252826082618,
      "loss": 2.4581,
      "step": 22392
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6848170161247253,
      "learning_rate": 0.0005861240529669719,
      "loss": 2.3871,
      "step": 22393
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6961140036582947,
      "learning_rate": 0.0005861228232724861,
      "loss": 2.3572,
      "step": 22394
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6208831667900085,
      "learning_rate": 0.0005861215935248048,
      "loss": 2.3397,
      "step": 22395
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6170759201049805,
      "learning_rate": 0.0005861203637239279,
      "loss": 2.3697,
      "step": 22396
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6666913032531738,
      "learning_rate": 0.0005861191338698559,
      "loss": 2.5044,
      "step": 22397
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6875422596931458,
      "learning_rate": 0.000586117903962589,
      "loss": 2.3213,
      "step": 22398
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6728271842002869,
      "learning_rate": 0.0005861166740021272,
      "loss": 2.2662,
      "step": 22399
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.5665462017059326,
      "learning_rate": 0.0005861154439884712,
      "loss": 2.4012,
      "step": 22400
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6630334854125977,
      "learning_rate": 0.0005861142139216208,
      "loss": 2.5131,
      "step": 22401
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.5805209875106812,
      "learning_rate": 0.0005861129838015763,
      "loss": 2.4539,
      "step": 22402
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6014371514320374,
      "learning_rate": 0.0005861117536283381,
      "loss": 2.1943,
      "step": 22403
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6088457107543945,
      "learning_rate": 0.0005861105234019063,
      "loss": 2.5375,
      "step": 22404
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.599600076675415,
      "learning_rate": 0.000586109293122281,
      "loss": 2.3438,
      "step": 22405
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.7014380693435669,
      "learning_rate": 0.0005861080627894628,
      "loss": 2.2236,
      "step": 22406
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.7815187573432922,
      "learning_rate": 0.0005861068324034516,
      "loss": 2.3486,
      "step": 22407
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6214717030525208,
      "learning_rate": 0.0005861056019642477,
      "loss": 2.2625,
      "step": 22408
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6233176589012146,
      "learning_rate": 0.0005861043714718515,
      "loss": 2.1655,
      "step": 22409
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6972914338111877,
      "learning_rate": 0.000586103140926263,
      "loss": 2.311,
      "step": 22410
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.591878354549408,
      "learning_rate": 0.0005861019103274826,
      "loss": 2.2974,
      "step": 22411
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.5836681127548218,
      "learning_rate": 0.0005861006796755103,
      "loss": 2.4288,
      "step": 22412
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6277576088905334,
      "learning_rate": 0.0005860994489703467,
      "loss": 2.4837,
      "step": 22413
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6609704494476318,
      "learning_rate": 0.0005860982182119917,
      "loss": 2.4151,
      "step": 22414
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6968185901641846,
      "learning_rate": 0.0005860969874004456,
      "loss": 2.3851,
      "step": 22415
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.5862535238265991,
      "learning_rate": 0.0005860957565357088,
      "loss": 2.3883,
      "step": 22416
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.5639713406562805,
      "learning_rate": 0.0005860945256177812,
      "loss": 2.3031,
      "step": 22417
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.7737834453582764,
      "learning_rate": 0.0005860932946466633,
      "loss": 2.268,
      "step": 22418
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.625646710395813,
      "learning_rate": 0.0005860920636223553,
      "loss": 2.5791,
      "step": 22419
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.7314865589141846,
      "learning_rate": 0.0005860908325448574,
      "loss": 2.4343,
      "step": 22420
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6863022446632385,
      "learning_rate": 0.0005860896014141697,
      "loss": 2.3867,
      "step": 22421
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6787909865379333,
      "learning_rate": 0.0005860883702302927,
      "loss": 2.4493,
      "step": 22422
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6112185716629028,
      "learning_rate": 0.0005860871389932264,
      "loss": 2.3892,
      "step": 22423
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.5900600552558899,
      "learning_rate": 0.0005860859077029711,
      "loss": 2.5055,
      "step": 22424
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.7169315814971924,
      "learning_rate": 0.000586084676359527,
      "loss": 2.4057,
      "step": 22425
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6513309478759766,
      "learning_rate": 0.0005860834449628945,
      "loss": 2.2369,
      "step": 22426
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6522055268287659,
      "learning_rate": 0.0005860822135130735,
      "loss": 2.307,
      "step": 22427
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6702945828437805,
      "learning_rate": 0.0005860809820100645,
      "loss": 2.2972,
      "step": 22428
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6431701183319092,
      "learning_rate": 0.0005860797504538677,
      "loss": 2.4394,
      "step": 22429
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6619659662246704,
      "learning_rate": 0.0005860785188444831,
      "loss": 2.4653,
      "step": 22430
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6553467512130737,
      "learning_rate": 0.0005860772871819113,
      "loss": 2.179,
      "step": 22431
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6037057042121887,
      "learning_rate": 0.0005860760554661522,
      "loss": 2.3532,
      "step": 22432
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.635504961013794,
      "learning_rate": 0.0005860748236972063,
      "loss": 2.4807,
      "step": 22433
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.7101634740829468,
      "learning_rate": 0.0005860735918750736,
      "loss": 2.3057,
      "step": 22434
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.8892752528190613,
      "learning_rate": 0.0005860723599997544,
      "loss": 2.4234,
      "step": 22435
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.59212726354599,
      "learning_rate": 0.0005860711280712489,
      "loss": 2.3941,
      "step": 22436
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6927657723426819,
      "learning_rate": 0.0005860698960895574,
      "loss": 2.0631,
      "step": 22437
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6145894527435303,
      "learning_rate": 0.0005860686640546802,
      "loss": 2.171,
      "step": 22438
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6746314764022827,
      "learning_rate": 0.0005860674319666174,
      "loss": 2.4475,
      "step": 22439
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6739412546157837,
      "learning_rate": 0.0005860661998253692,
      "loss": 2.3396,
      "step": 22440
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.8566661477088928,
      "learning_rate": 0.0005860649676309361,
      "loss": 2.4323,
      "step": 22441
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.5963780283927917,
      "learning_rate": 0.0005860637353833179,
      "loss": 2.5713,
      "step": 22442
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.5851109027862549,
      "learning_rate": 0.0005860625030825151,
      "loss": 2.2589,
      "step": 22443
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6920463442802429,
      "learning_rate": 0.0005860612707285278,
      "loss": 2.0958,
      "step": 22444
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6737070679664612,
      "learning_rate": 0.0005860600383213565,
      "loss": 2.274,
      "step": 22445
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6604719161987305,
      "learning_rate": 0.0005860588058610012,
      "loss": 2.2453,
      "step": 22446
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.5942577719688416,
      "learning_rate": 0.000586057573347462,
      "loss": 2.2314,
      "step": 22447
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.7320784330368042,
      "learning_rate": 0.0005860563407807395,
      "loss": 2.3994,
      "step": 22448
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6154409050941467,
      "learning_rate": 0.0005860551081608336,
      "loss": 2.3715,
      "step": 22449
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6762418150901794,
      "learning_rate": 0.0005860538754877447,
      "loss": 2.3507,
      "step": 22450
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6956599950790405,
      "learning_rate": 0.000586052642761473,
      "loss": 2.3858,
      "step": 22451
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6518788933753967,
      "learning_rate": 0.0005860514099820187,
      "loss": 2.5018,
      "step": 22452
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6179426312446594,
      "learning_rate": 0.000586050177149382,
      "loss": 2.2683,
      "step": 22453
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6672044396400452,
      "learning_rate": 0.0005860489442635633,
      "loss": 2.4854,
      "step": 22454
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.5666614770889282,
      "learning_rate": 0.0005860477113245625,
      "loss": 2.3133,
      "step": 22455
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.5786226391792297,
      "learning_rate": 0.0005860464783323802,
      "loss": 2.4947,
      "step": 22456
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6287715435028076,
      "learning_rate": 0.0005860452452870164,
      "loss": 2.3089,
      "step": 22457
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.5647507905960083,
      "learning_rate": 0.0005860440121884715,
      "loss": 2.477,
      "step": 22458
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6688392758369446,
      "learning_rate": 0.0005860427790367454,
      "loss": 2.4802,
      "step": 22459
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6283255815505981,
      "learning_rate": 0.0005860415458318386,
      "loss": 2.5539,
      "step": 22460
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6073377728462219,
      "learning_rate": 0.0005860403125737514,
      "loss": 2.3513,
      "step": 22461
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.7481886744499207,
      "learning_rate": 0.000586039079262484,
      "loss": 2.5884,
      "step": 22462
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6594886183738708,
      "learning_rate": 0.0005860378458980363,
      "loss": 2.4008,
      "step": 22463
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.5610778331756592,
      "learning_rate": 0.0005860366124804088,
      "loss": 2.3109,
      "step": 22464
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6256945133209229,
      "learning_rate": 0.0005860353790096017,
      "loss": 2.3258,
      "step": 22465
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6638215780258179,
      "learning_rate": 0.0005860341454856152,
      "loss": 2.4006,
      "step": 22466
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6847900748252869,
      "learning_rate": 0.0005860329119084497,
      "loss": 2.3709,
      "step": 22467
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6063206195831299,
      "learning_rate": 0.0005860316782781052,
      "loss": 2.4497,
      "step": 22468
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6005525588989258,
      "learning_rate": 0.000586030444594582,
      "loss": 2.3325,
      "step": 22469
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6363013386726379,
      "learning_rate": 0.0005860292108578804,
      "loss": 2.4681,
      "step": 22470
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6486800909042358,
      "learning_rate": 0.0005860279770680005,
      "loss": 2.1832,
      "step": 22471
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6174710392951965,
      "learning_rate": 0.0005860267432249425,
      "loss": 2.5606,
      "step": 22472
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6346749067306519,
      "learning_rate": 0.000586025509328707,
      "loss": 2.1721,
      "step": 22473
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6465297937393188,
      "learning_rate": 0.0005860242753792938,
      "loss": 2.1733,
      "step": 22474
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6241899132728577,
      "learning_rate": 0.0005860230413767034,
      "loss": 2.3456,
      "step": 22475
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6630524396896362,
      "learning_rate": 0.0005860218073209357,
      "loss": 2.3874,
      "step": 22476
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.607835054397583,
      "learning_rate": 0.0005860205732119912,
      "loss": 2.4171,
      "step": 22477
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.5692940950393677,
      "learning_rate": 0.0005860193390498702,
      "loss": 2.2541,
      "step": 22478
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6571222543716431,
      "learning_rate": 0.0005860181048345728,
      "loss": 2.3423,
      "step": 22479
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6807143092155457,
      "learning_rate": 0.0005860168705660991,
      "loss": 2.5304,
      "step": 22480
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.5990533828735352,
      "learning_rate": 0.0005860156362444495,
      "loss": 2.5023,
      "step": 22481
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.5913675427436829,
      "learning_rate": 0.0005860144018696243,
      "loss": 2.4174,
      "step": 22482
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.604308545589447,
      "learning_rate": 0.0005860131674416236,
      "loss": 2.475,
      "step": 22483
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6261659860610962,
      "learning_rate": 0.0005860119329604476,
      "loss": 2.3825,
      "step": 22484
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6421130895614624,
      "learning_rate": 0.0005860106984260966,
      "loss": 2.2932,
      "step": 22485
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.5696882605552673,
      "learning_rate": 0.0005860094638385708,
      "loss": 2.5087,
      "step": 22486
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6681681275367737,
      "learning_rate": 0.0005860082291978705,
      "loss": 2.1436,
      "step": 22487
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6036677956581116,
      "learning_rate": 0.0005860069945039958,
      "loss": 2.4676,
      "step": 22488
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.629131555557251,
      "learning_rate": 0.000586005759756947,
      "loss": 2.5288,
      "step": 22489
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6597038507461548,
      "learning_rate": 0.0005860045249567244,
      "loss": 2.4991,
      "step": 22490
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.616931676864624,
      "learning_rate": 0.0005860032901033281,
      "loss": 2.317,
      "step": 22491
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6019511222839355,
      "learning_rate": 0.0005860020551967584,
      "loss": 2.422,
      "step": 22492
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.5703595280647278,
      "learning_rate": 0.0005860008202370155,
      "loss": 1.8749,
      "step": 22493
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6628609299659729,
      "learning_rate": 0.0005859995852240997,
      "loss": 2.5295,
      "step": 22494
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.631217896938324,
      "learning_rate": 0.0005859983501580111,
      "loss": 2.246,
      "step": 22495
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6076434254646301,
      "learning_rate": 0.0005859971150387501,
      "loss": 2.3245,
      "step": 22496
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6901917457580566,
      "learning_rate": 0.0005859958798663167,
      "loss": 2.493,
      "step": 22497
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.5740445852279663,
      "learning_rate": 0.0005859946446407114,
      "loss": 2.3795,
      "step": 22498
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6915382146835327,
      "learning_rate": 0.0005859934093619344,
      "loss": 2.1957,
      "step": 22499
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6340103149414062,
      "learning_rate": 0.0005859921740299855,
      "loss": 2.0952,
      "step": 22500
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6415131688117981,
      "learning_rate": 0.0005859909386448655,
      "loss": 2.7455,
      "step": 22501
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6388224959373474,
      "learning_rate": 0.0005859897032065744,
      "loss": 2.5468,
      "step": 22502
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6179394125938416,
      "learning_rate": 0.0005859884677151124,
      "loss": 2.5333,
      "step": 22503
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.7191564440727234,
      "learning_rate": 0.0005859872321704796,
      "loss": 2.4355,
      "step": 22504
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.5927925109863281,
      "learning_rate": 0.0005859859965726764,
      "loss": 2.3585,
      "step": 22505
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6309416890144348,
      "learning_rate": 0.0005859847609217031,
      "loss": 2.2569,
      "step": 22506
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6392673850059509,
      "learning_rate": 0.0005859835252175599,
      "loss": 2.4061,
      "step": 22507
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6331329941749573,
      "learning_rate": 0.0005859822894602469,
      "loss": 2.4143,
      "step": 22508
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6545720100402832,
      "learning_rate": 0.0005859810536497643,
      "loss": 2.518,
      "step": 22509
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.647222101688385,
      "learning_rate": 0.0005859798177861124,
      "loss": 2.45,
      "step": 22510
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.5679820775985718,
      "learning_rate": 0.0005859785818692916,
      "loss": 2.3538,
      "step": 22511
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6208441853523254,
      "learning_rate": 0.0005859773458993019,
      "loss": 2.2164,
      "step": 22512
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6897504925727844,
      "learning_rate": 0.0005859761098761437,
      "loss": 2.2553,
      "step": 22513
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6013594269752502,
      "learning_rate": 0.0005859748737998171,
      "loss": 2.2732,
      "step": 22514
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.5674411058425903,
      "learning_rate": 0.0005859736376703223,
      "loss": 2.2385,
      "step": 22515
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6268728375434875,
      "learning_rate": 0.0005859724014876597,
      "loss": 2.2533,
      "step": 22516
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6007399559020996,
      "learning_rate": 0.0005859711652518294,
      "loss": 2.4673,
      "step": 22517
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6631883978843689,
      "learning_rate": 0.0005859699289628317,
      "loss": 2.326,
      "step": 22518
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6964666247367859,
      "learning_rate": 0.0005859686926206666,
      "loss": 2.4483,
      "step": 22519
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6163209080696106,
      "learning_rate": 0.0005859674562253347,
      "loss": 2.5038,
      "step": 22520
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.5466971397399902,
      "learning_rate": 0.0005859662197768361,
      "loss": 2.3337,
      "step": 22521
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6278645396232605,
      "learning_rate": 0.0005859649832751708,
      "loss": 2.2939,
      "step": 22522
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6736800670623779,
      "learning_rate": 0.0005859637467203393,
      "loss": 2.5088,
      "step": 22523
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6226540207862854,
      "learning_rate": 0.0005859625101123418,
      "loss": 2.2423,
      "step": 22524
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6417853236198425,
      "learning_rate": 0.0005859612734511784,
      "loss": 2.3021,
      "step": 22525
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6108308434486389,
      "learning_rate": 0.0005859600367368494,
      "loss": 2.2326,
      "step": 22526
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.7361042499542236,
      "learning_rate": 0.0005859587999693551,
      "loss": 2.447,
      "step": 22527
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.617111325263977,
      "learning_rate": 0.0005859575631486957,
      "loss": 2.4004,
      "step": 22528
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6394984126091003,
      "learning_rate": 0.0005859563262748713,
      "loss": 2.3749,
      "step": 22529
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.7554535865783691,
      "learning_rate": 0.0005859550893478822,
      "loss": 2.3395,
      "step": 22530
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6028558015823364,
      "learning_rate": 0.0005859538523677287,
      "loss": 2.4737,
      "step": 22531
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6174036264419556,
      "learning_rate": 0.0005859526153344109,
      "loss": 2.1154,
      "step": 22532
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6597700715065002,
      "learning_rate": 0.0005859513782479292,
      "loss": 2.2125,
      "step": 22533
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.5764264464378357,
      "learning_rate": 0.0005859501411082837,
      "loss": 2.2872,
      "step": 22534
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.7039048671722412,
      "learning_rate": 0.0005859489039154747,
      "loss": 2.3731,
      "step": 22535
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6830554008483887,
      "learning_rate": 0.0005859476666695023,
      "loss": 2.5965,
      "step": 22536
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.5836930871009827,
      "learning_rate": 0.000585946429370367,
      "loss": 2.2844,
      "step": 22537
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.584553062915802,
      "learning_rate": 0.0005859451920180688,
      "loss": 2.4794,
      "step": 22538
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6374651789665222,
      "learning_rate": 0.0005859439546126079,
      "loss": 2.2938,
      "step": 22539
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6053069233894348,
      "learning_rate": 0.0005859427171539847,
      "loss": 2.3457,
      "step": 22540
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.5974894165992737,
      "learning_rate": 0.0005859414796421993,
      "loss": 2.1801,
      "step": 22541
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6245163679122925,
      "learning_rate": 0.0005859402420772522,
      "loss": 2.2575,
      "step": 22542
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6498836874961853,
      "learning_rate": 0.0005859390044591432,
      "loss": 2.1593,
      "step": 22543
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6646912097930908,
      "learning_rate": 0.0005859377667878727,
      "loss": 2.3765,
      "step": 22544
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.5872918367385864,
      "learning_rate": 0.0005859365290634411,
      "loss": 2.2433,
      "step": 22545
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6683664917945862,
      "learning_rate": 0.0005859352912858484,
      "loss": 2.1379,
      "step": 22546
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6209481954574585,
      "learning_rate": 0.000585934053455095,
      "loss": 2.3356,
      "step": 22547
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6431496143341064,
      "learning_rate": 0.0005859328155711811,
      "loss": 2.3594,
      "step": 22548
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6347280740737915,
      "learning_rate": 0.0005859315776341068,
      "loss": 2.4038,
      "step": 22549
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.5989338755607605,
      "learning_rate": 0.0005859303396438725,
      "loss": 2.2768,
      "step": 22550
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6696068644523621,
      "learning_rate": 0.0005859291016004783,
      "loss": 2.363,
      "step": 22551
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.7238800525665283,
      "learning_rate": 0.0005859278635039245,
      "loss": 2.4128,
      "step": 22552
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6590606570243835,
      "learning_rate": 0.0005859266253542113,
      "loss": 2.4068,
      "step": 22553
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.8217854499816895,
      "learning_rate": 0.0005859253871513389,
      "loss": 2.243,
      "step": 22554
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.617234468460083,
      "learning_rate": 0.0005859241488953076,
      "loss": 2.3581,
      "step": 22555
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6024115085601807,
      "learning_rate": 0.0005859229105861176,
      "loss": 2.3311,
      "step": 22556
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.632649302482605,
      "learning_rate": 0.0005859216722237693,
      "loss": 2.3903,
      "step": 22557
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.647014856338501,
      "learning_rate": 0.0005859204338082625,
      "loss": 2.5336,
      "step": 22558
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6420093178749084,
      "learning_rate": 0.0005859191953395979,
      "loss": 2.4036,
      "step": 22559
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6328885555267334,
      "learning_rate": 0.0005859179568177754,
      "loss": 2.4806,
      "step": 22560
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6193171143531799,
      "learning_rate": 0.0005859167182427953,
      "loss": 2.2448,
      "step": 22561
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6707769632339478,
      "learning_rate": 0.0005859154796146581,
      "loss": 2.3352,
      "step": 22562
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6746311187744141,
      "learning_rate": 0.0005859142409333636,
      "loss": 2.402,
      "step": 22563
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6825746893882751,
      "learning_rate": 0.0005859130021989124,
      "loss": 2.4337,
      "step": 22564
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6057994365692139,
      "learning_rate": 0.0005859117634113046,
      "loss": 2.3387,
      "step": 22565
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.7018051743507385,
      "learning_rate": 0.0005859105245705403,
      "loss": 2.4331,
      "step": 22566
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.602654218673706,
      "learning_rate": 0.0005859092856766197,
      "loss": 2.3541,
      "step": 22567
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6225320100784302,
      "learning_rate": 0.0005859080467295434,
      "loss": 2.3302,
      "step": 22568
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6486538648605347,
      "learning_rate": 0.0005859068077293114,
      "loss": 2.4441,
      "step": 22569
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6416037082672119,
      "learning_rate": 0.0005859055686759238,
      "loss": 2.3793,
      "step": 22570
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6840828657150269,
      "learning_rate": 0.0005859043295693811,
      "loss": 2.2618,
      "step": 22571
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6150646209716797,
      "learning_rate": 0.0005859030904096832,
      "loss": 2.311,
      "step": 22572
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6369417905807495,
      "learning_rate": 0.0005859018511968307,
      "loss": 2.7545,
      "step": 22573
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6452338695526123,
      "learning_rate": 0.0005859006119308235,
      "loss": 2.2855,
      "step": 22574
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6231570839881897,
      "learning_rate": 0.0005858993726116621,
      "loss": 2.4137,
      "step": 22575
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6819791793823242,
      "learning_rate": 0.0005858981332393465,
      "loss": 2.5072,
      "step": 22576
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6482526063919067,
      "learning_rate": 0.0005858968938138771,
      "loss": 2.2273,
      "step": 22577
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.5921302437782288,
      "learning_rate": 0.0005858956543352541,
      "loss": 2.3005,
      "step": 22578
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.5202111601829529,
      "learning_rate": 0.0005858944148034776,
      "loss": 2.4509,
      "step": 22579
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6455020308494568,
      "learning_rate": 0.0005858931752185481,
      "loss": 2.3495,
      "step": 22580
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.5957404375076294,
      "learning_rate": 0.0005858919355804655,
      "loss": 2.3628,
      "step": 22581
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6066769361495972,
      "learning_rate": 0.0005858906958892302,
      "loss": 2.2938,
      "step": 22582
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.7813178300857544,
      "learning_rate": 0.0005858894561448425,
      "loss": 2.4836,
      "step": 22583
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6258257031440735,
      "learning_rate": 0.0005858882163473026,
      "loss": 2.3001,
      "step": 22584
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6341906785964966,
      "learning_rate": 0.0005858869764966105,
      "loss": 2.3283,
      "step": 22585
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.73854660987854,
      "learning_rate": 0.0005858857365927669,
      "loss": 2.2954,
      "step": 22586
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6148121953010559,
      "learning_rate": 0.0005858844966357715,
      "loss": 2.2442,
      "step": 22587
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.5718419551849365,
      "learning_rate": 0.0005858832566256248,
      "loss": 2.4884,
      "step": 22588
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.7049494981765747,
      "learning_rate": 0.0005858820165623271,
      "loss": 2.5801,
      "step": 22589
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.659210205078125,
      "learning_rate": 0.0005858807764458785,
      "loss": 2.5105,
      "step": 22590
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.5725172162055969,
      "learning_rate": 0.0005858795362762793,
      "loss": 2.2751,
      "step": 22591
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.7237269282341003,
      "learning_rate": 0.0005858782960535296,
      "loss": 2.5372,
      "step": 22592
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.7220215201377869,
      "learning_rate": 0.0005858770557776297,
      "loss": 2.4917,
      "step": 22593
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6486254930496216,
      "learning_rate": 0.00058587581544858,
      "loss": 2.3444,
      "step": 22594
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6966944336891174,
      "learning_rate": 0.0005858745750663806,
      "loss": 2.3979,
      "step": 22595
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.767067015171051,
      "learning_rate": 0.0005858733346310316,
      "loss": 2.292,
      "step": 22596
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.9476358890533447,
      "learning_rate": 0.0005858720941425335,
      "loss": 2.3191,
      "step": 22597
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6254554390907288,
      "learning_rate": 0.0005858708536008862,
      "loss": 2.4092,
      "step": 22598
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.8048425316810608,
      "learning_rate": 0.0005858696130060903,
      "loss": 2.2016,
      "step": 22599
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.8589920401573181,
      "learning_rate": 0.0005858683723581456,
      "loss": 2.3619,
      "step": 22600
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6949777007102966,
      "learning_rate": 0.0005858671316570528,
      "loss": 2.3977,
      "step": 22601
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.7531182765960693,
      "learning_rate": 0.0005858658909028118,
      "loss": 2.446,
      "step": 22602
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6260786056518555,
      "learning_rate": 0.000585864650095423,
      "loss": 2.4918,
      "step": 22603
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6369323134422302,
      "learning_rate": 0.0005858634092348865,
      "loss": 2.1175,
      "step": 22604
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.8166059255599976,
      "learning_rate": 0.0005858621683212026,
      "loss": 2.3547,
      "step": 22605
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.7658480405807495,
      "learning_rate": 0.0005858609273543716,
      "loss": 2.3583,
      "step": 22606
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.7550672292709351,
      "learning_rate": 0.0005858596863343936,
      "loss": 2.439,
      "step": 22607
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6703885197639465,
      "learning_rate": 0.0005858584452612688,
      "loss": 2.3911,
      "step": 22608
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6115309596061707,
      "learning_rate": 0.0005858572041349976,
      "loss": 2.1695,
      "step": 22609
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.664355456829071,
      "learning_rate": 0.0005858559629555802,
      "loss": 2.5604,
      "step": 22610
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.7132686376571655,
      "learning_rate": 0.0005858547217230167,
      "loss": 2.4668,
      "step": 22611
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6476016640663147,
      "learning_rate": 0.0005858534804373075,
      "loss": 2.3361,
      "step": 22612
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.646109402179718,
      "learning_rate": 0.0005858522390984527,
      "loss": 2.4231,
      "step": 22613
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.5893889665603638,
      "learning_rate": 0.0005858509977064525,
      "loss": 2.2816,
      "step": 22614
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.628585934638977,
      "learning_rate": 0.0005858497562613073,
      "loss": 2.6417,
      "step": 22615
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.5936053991317749,
      "learning_rate": 0.0005858485147630173,
      "loss": 2.4782,
      "step": 22616
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6581953763961792,
      "learning_rate": 0.0005858472732115825,
      "loss": 2.4198,
      "step": 22617
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6949490904808044,
      "learning_rate": 0.0005858460316070034,
      "loss": 2.5952,
      "step": 22618
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6387307047843933,
      "learning_rate": 0.00058584478994928,
      "loss": 2.3255,
      "step": 22619
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6333564519882202,
      "learning_rate": 0.0005858435482384128,
      "loss": 2.2625,
      "step": 22620
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.727822482585907,
      "learning_rate": 0.0005858423064744018,
      "loss": 2.2442,
      "step": 22621
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6240638494491577,
      "learning_rate": 0.0005858410646572474,
      "loss": 2.1895,
      "step": 22622
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.634380578994751,
      "learning_rate": 0.0005858398227869497,
      "loss": 2.289,
      "step": 22623
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.5966334342956543,
      "learning_rate": 0.000585838580863509,
      "loss": 2.4838,
      "step": 22624
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.5618183016777039,
      "learning_rate": 0.0005858373388869254,
      "loss": 2.5276,
      "step": 22625
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.5983877182006836,
      "learning_rate": 0.0005858360968571994,
      "loss": 2.2086,
      "step": 22626
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6505873203277588,
      "learning_rate": 0.0005858348547743311,
      "loss": 2.3682,
      "step": 22627
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6302871108055115,
      "learning_rate": 0.0005858336126383206,
      "loss": 2.7064,
      "step": 22628
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6494474411010742,
      "learning_rate": 0.0005858323704491683,
      "loss": 2.156,
      "step": 22629
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6154435276985168,
      "learning_rate": 0.0005858311282068742,
      "loss": 2.4641,
      "step": 22630
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6487009525299072,
      "learning_rate": 0.0005858298859114389,
      "loss": 2.3499,
      "step": 22631
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.5849985480308533,
      "learning_rate": 0.0005858286435628623,
      "loss": 2.251,
      "step": 22632
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6019129753112793,
      "learning_rate": 0.0005858274011611448,
      "loss": 2.4694,
      "step": 22633
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.5842230916023254,
      "learning_rate": 0.0005858261587062866,
      "loss": 2.2911,
      "step": 22634
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6305392980575562,
      "learning_rate": 0.000585824916198288,
      "loss": 2.4685,
      "step": 22635
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6052907109260559,
      "learning_rate": 0.0005858236736371491,
      "loss": 2.5701,
      "step": 22636
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6443467736244202,
      "learning_rate": 0.0005858224310228701,
      "loss": 2.4058,
      "step": 22637
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6952332258224487,
      "learning_rate": 0.0005858211883554514,
      "loss": 2.0974,
      "step": 22638
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6110389828681946,
      "learning_rate": 0.0005858199456348931,
      "loss": 2.3823,
      "step": 22639
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6714463233947754,
      "learning_rate": 0.0005858187028611956,
      "loss": 2.1969,
      "step": 22640
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6599361300468445,
      "learning_rate": 0.0005858174600343588,
      "loss": 2.4104,
      "step": 22641
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.5828186273574829,
      "learning_rate": 0.0005858162171543833,
      "loss": 2.25,
      "step": 22642
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.7271581888198853,
      "learning_rate": 0.000585814974221269,
      "loss": 2.2714,
      "step": 22643
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.5934022665023804,
      "learning_rate": 0.0005858137312350165,
      "loss": 2.368,
      "step": 22644
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.636389970779419,
      "learning_rate": 0.0005858124881956258,
      "loss": 2.6577,
      "step": 22645
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.5996392369270325,
      "learning_rate": 0.000585811245103097,
      "loss": 2.3132,
      "step": 22646
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.7619482278823853,
      "learning_rate": 0.0005858100019574306,
      "loss": 2.4718,
      "step": 22647
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6488437652587891,
      "learning_rate": 0.0005858087587586267,
      "loss": 2.4281,
      "step": 22648
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.5505159497261047,
      "learning_rate": 0.0005858075155066857,
      "loss": 2.233,
      "step": 22649
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.614483654499054,
      "learning_rate": 0.0005858062722016075,
      "loss": 2.5328,
      "step": 22650
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6095069050788879,
      "learning_rate": 0.0005858050288433927,
      "loss": 2.2517,
      "step": 22651
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6610845327377319,
      "learning_rate": 0.0005858037854320412,
      "loss": 2.3067,
      "step": 22652
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6761252284049988,
      "learning_rate": 0.0005858025419675535,
      "loss": 2.2631,
      "step": 22653
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.593908965587616,
      "learning_rate": 0.0005858012984499295,
      "loss": 2.128,
      "step": 22654
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.636069118976593,
      "learning_rate": 0.0005858000548791698,
      "loss": 2.4221,
      "step": 22655
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6361876130104065,
      "learning_rate": 0.0005857988112552744,
      "loss": 2.4526,
      "step": 22656
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.623366117477417,
      "learning_rate": 0.0005857975675782437,
      "loss": 2.3161,
      "step": 22657
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6311463713645935,
      "learning_rate": 0.0005857963238480779,
      "loss": 2.2089,
      "step": 22658
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.5758768320083618,
      "learning_rate": 0.000585795080064777,
      "loss": 2.2854,
      "step": 22659
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.5898515582084656,
      "learning_rate": 0.0005857938362283415,
      "loss": 2.3542,
      "step": 22660
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6082277894020081,
      "learning_rate": 0.0005857925923387715,
      "loss": 2.5157,
      "step": 22661
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6354969143867493,
      "learning_rate": 0.0005857913483960672,
      "loss": 2.2612,
      "step": 22662
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6196473836898804,
      "learning_rate": 0.000585790104400229,
      "loss": 2.2351,
      "step": 22663
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.5814499855041504,
      "learning_rate": 0.000585788860351257,
      "loss": 2.2082,
      "step": 22664
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6078512072563171,
      "learning_rate": 0.0005857876162491514,
      "loss": 2.0951,
      "step": 22665
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6595908403396606,
      "learning_rate": 0.0005857863720939126,
      "loss": 2.5282,
      "step": 22666
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.5922245979309082,
      "learning_rate": 0.0005857851278855406,
      "loss": 2.2178,
      "step": 22667
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.5776203870773315,
      "learning_rate": 0.0005857838836240358,
      "loss": 2.1564,
      "step": 22668
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.5909841656684875,
      "learning_rate": 0.0005857826393093985,
      "loss": 2.3023,
      "step": 22669
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6278020143508911,
      "learning_rate": 0.0005857813949416285,
      "loss": 2.2815,
      "step": 22670
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6145736575126648,
      "learning_rate": 0.0005857801505207266,
      "loss": 2.5037,
      "step": 22671
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6145807504653931,
      "learning_rate": 0.0005857789060466927,
      "loss": 2.2365,
      "step": 22672
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6638277173042297,
      "learning_rate": 0.0005857776615195271,
      "loss": 2.1235,
      "step": 22673
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.7663031816482544,
      "learning_rate": 0.0005857764169392301,
      "loss": 2.4564,
      "step": 22674
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6369129419326782,
      "learning_rate": 0.0005857751723058018,
      "loss": 2.3875,
      "step": 22675
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6555057168006897,
      "learning_rate": 0.0005857739276192425,
      "loss": 2.4092,
      "step": 22676
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6936769485473633,
      "learning_rate": 0.0005857726828795525,
      "loss": 2.4108,
      "step": 22677
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.7230923771858215,
      "learning_rate": 0.0005857714380867319,
      "loss": 2.136,
      "step": 22678
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6694352030754089,
      "learning_rate": 0.000585770193240781,
      "loss": 2.1582,
      "step": 22679
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.9196509122848511,
      "learning_rate": 0.0005857689483417001,
      "loss": 2.3115,
      "step": 22680
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6541484594345093,
      "learning_rate": 0.0005857677033894893,
      "loss": 2.3837,
      "step": 22681
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6074689030647278,
      "learning_rate": 0.0005857664583841488,
      "loss": 2.4864,
      "step": 22682
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.5832299590110779,
      "learning_rate": 0.0005857652133256791,
      "loss": 2.4373,
      "step": 22683
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.8297412991523743,
      "learning_rate": 0.0005857639682140801,
      "loss": 2.3721,
      "step": 22684
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.5642563104629517,
      "learning_rate": 0.0005857627230493524,
      "loss": 2.0798,
      "step": 22685
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6125807166099548,
      "learning_rate": 0.0005857614778314958,
      "loss": 2.3554,
      "step": 22686
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.9726195335388184,
      "learning_rate": 0.0005857602325605107,
      "loss": 2.335,
      "step": 22687
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.7058549523353577,
      "learning_rate": 0.0005857589872363976,
      "loss": 2.2612,
      "step": 22688
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.617338240146637,
      "learning_rate": 0.0005857577418591564,
      "loss": 2.2989,
      "step": 22689
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6084780097007751,
      "learning_rate": 0.0005857564964287874,
      "loss": 2.4266,
      "step": 22690
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6859648823738098,
      "learning_rate": 0.0005857552509452909,
      "loss": 2.3123,
      "step": 22691
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.626952052116394,
      "learning_rate": 0.0005857540054086671,
      "loss": 2.4867,
      "step": 22692
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.5855010151863098,
      "learning_rate": 0.0005857527598189163,
      "loss": 2.5875,
      "step": 22693
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6637688875198364,
      "learning_rate": 0.0005857515141760386,
      "loss": 2.4692,
      "step": 22694
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.7250495553016663,
      "learning_rate": 0.0005857502684800343,
      "loss": 2.3693,
      "step": 22695
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6584062576293945,
      "learning_rate": 0.0005857490227309036,
      "loss": 2.3012,
      "step": 22696
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6878920197486877,
      "learning_rate": 0.0005857477769286468,
      "loss": 2.5092,
      "step": 22697
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.5972249507904053,
      "learning_rate": 0.0005857465310732642,
      "loss": 2.4248,
      "step": 22698
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.5852405428886414,
      "learning_rate": 0.0005857452851647558,
      "loss": 2.1342,
      "step": 22699
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.5938159227371216,
      "learning_rate": 0.000585744039203122,
      "loss": 2.253,
      "step": 22700
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6736249923706055,
      "learning_rate": 0.0005857427931883631,
      "loss": 2.451,
      "step": 22701
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6458155512809753,
      "learning_rate": 0.000585741547120479,
      "loss": 2.2339,
      "step": 22702
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6511887907981873,
      "learning_rate": 0.0005857403009994703,
      "loss": 2.3246,
      "step": 22703
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.7136543989181519,
      "learning_rate": 0.000585739054825337,
      "loss": 2.2933,
      "step": 22704
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6120644807815552,
      "learning_rate": 0.0005857378085980795,
      "loss": 2.2979,
      "step": 22705
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6425040364265442,
      "learning_rate": 0.0005857365623176979,
      "loss": 2.366,
      "step": 22706
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6845939755439758,
      "learning_rate": 0.0005857353159841925,
      "loss": 2.4957,
      "step": 22707
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.5487704873085022,
      "learning_rate": 0.0005857340695975635,
      "loss": 2.2417,
      "step": 22708
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.698509156703949,
      "learning_rate": 0.0005857328231578112,
      "loss": 2.3361,
      "step": 22709
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6653898358345032,
      "learning_rate": 0.0005857315766649357,
      "loss": 2.3259,
      "step": 22710
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.5911577343940735,
      "learning_rate": 0.0005857303301189372,
      "loss": 2.3953,
      "step": 22711
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6731353402137756,
      "learning_rate": 0.0005857290835198163,
      "loss": 2.3515,
      "step": 22712
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6663743257522583,
      "learning_rate": 0.0005857278368675727,
      "loss": 2.0103,
      "step": 22713
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6228776574134827,
      "learning_rate": 0.000585726590162207,
      "loss": 2.3349,
      "step": 22714
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6308668851852417,
      "learning_rate": 0.0005857253434037195,
      "loss": 2.4553,
      "step": 22715
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.690593957901001,
      "learning_rate": 0.0005857240965921101,
      "loss": 2.3734,
      "step": 22716
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.718970000743866,
      "learning_rate": 0.0005857228497273792,
      "loss": 2.3424,
      "step": 22717
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6425133347511292,
      "learning_rate": 0.000585721602809527,
      "loss": 2.5781,
      "step": 22718
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6320648789405823,
      "learning_rate": 0.0005857203558385539,
      "loss": 2.4643,
      "step": 22719
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6837247014045715,
      "learning_rate": 0.0005857191088144599,
      "loss": 2.2508,
      "step": 22720
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6138430833816528,
      "learning_rate": 0.0005857178617372453,
      "loss": 2.2943,
      "step": 22721
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6389514803886414,
      "learning_rate": 0.0005857166146069103,
      "loss": 2.1262,
      "step": 22722
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6133124232292175,
      "learning_rate": 0.0005857153674234554,
      "loss": 2.2505,
      "step": 22723
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.644952654838562,
      "learning_rate": 0.0005857141201868805,
      "loss": 2.5537,
      "step": 22724
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6122103333473206,
      "learning_rate": 0.0005857128728971859,
      "loss": 2.2062,
      "step": 22725
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6945317983627319,
      "learning_rate": 0.0005857116255543719,
      "loss": 2.4489,
      "step": 22726
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.5936198830604553,
      "learning_rate": 0.0005857103781584387,
      "loss": 1.9931,
      "step": 22727
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6998224854469299,
      "learning_rate": 0.0005857091307093866,
      "loss": 2.5001,
      "step": 22728
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6930689215660095,
      "learning_rate": 0.0005857078832072159,
      "loss": 2.4192,
      "step": 22729
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6116741299629211,
      "learning_rate": 0.0005857066356519266,
      "loss": 2.3465,
      "step": 22730
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.5808780193328857,
      "learning_rate": 0.0005857053880435189,
      "loss": 2.1867,
      "step": 22731
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6066437363624573,
      "learning_rate": 0.0005857041403819933,
      "loss": 2.4058,
      "step": 22732
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6042923331260681,
      "learning_rate": 0.00058570289266735,
      "loss": 2.3479,
      "step": 22733
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6047117114067078,
      "learning_rate": 0.000585701644899589,
      "loss": 2.323,
      "step": 22734
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6426445245742798,
      "learning_rate": 0.0005857003970787107,
      "loss": 2.364,
      "step": 22735
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.608081579208374,
      "learning_rate": 0.0005856991492047153,
      "loss": 2.22,
      "step": 22736
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6002166271209717,
      "learning_rate": 0.0005856979012776032,
      "loss": 2.2491,
      "step": 22737
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6291121244430542,
      "learning_rate": 0.0005856966532973742,
      "loss": 2.3237,
      "step": 22738
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6576854586601257,
      "learning_rate": 0.0005856954052640289,
      "loss": 2.2128,
      "step": 22739
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6324437856674194,
      "learning_rate": 0.0005856941571775675,
      "loss": 2.3883,
      "step": 22740
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6757965683937073,
      "learning_rate": 0.0005856929090379901,
      "loss": 2.3347,
      "step": 22741
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6114482879638672,
      "learning_rate": 0.000585691660845297,
      "loss": 2.4203,
      "step": 22742
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6404784917831421,
      "learning_rate": 0.0005856904125994885,
      "loss": 2.2133,
      "step": 22743
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.593222975730896,
      "learning_rate": 0.0005856891643005647,
      "loss": 2.3294,
      "step": 22744
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.603050172328949,
      "learning_rate": 0.0005856879159485259,
      "loss": 2.3501,
      "step": 22745
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6751529574394226,
      "learning_rate": 0.0005856866675433722,
      "loss": 2.283,
      "step": 22746
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.7013863325119019,
      "learning_rate": 0.0005856854190851041,
      "loss": 2.2745,
      "step": 22747
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6160498857498169,
      "learning_rate": 0.0005856841705737218,
      "loss": 2.2254,
      "step": 22748
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.7275826334953308,
      "learning_rate": 0.0005856829220092252,
      "loss": 2.3704,
      "step": 22749
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.67619389295578,
      "learning_rate": 0.0005856816733916148,
      "loss": 2.4689,
      "step": 22750
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6215026378631592,
      "learning_rate": 0.0005856804247208909,
      "loss": 2.3152,
      "step": 22751
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.739068329334259,
      "learning_rate": 0.0005856791759970535,
      "loss": 2.5227,
      "step": 22752
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6848208904266357,
      "learning_rate": 0.000585677927220103,
      "loss": 2.1992,
      "step": 22753
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6054918766021729,
      "learning_rate": 0.0005856766783900395,
      "loss": 2.4505,
      "step": 22754
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.5765348076820374,
      "learning_rate": 0.0005856754295068635,
      "loss": 2.2194,
      "step": 22755
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6357936859130859,
      "learning_rate": 0.0005856741805705749,
      "loss": 2.2571,
      "step": 22756
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.5940415263175964,
      "learning_rate": 0.000585672931581174,
      "loss": 2.4471,
      "step": 22757
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.5775214433670044,
      "learning_rate": 0.0005856716825386613,
      "loss": 2.1797,
      "step": 22758
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6606203317642212,
      "learning_rate": 0.0005856704334430366,
      "loss": 2.2044,
      "step": 22759
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6629649996757507,
      "learning_rate": 0.0005856691842943006,
      "loss": 2.3016,
      "step": 22760
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.5894731283187866,
      "learning_rate": 0.0005856679350924532,
      "loss": 2.1903,
      "step": 22761
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6889567375183105,
      "learning_rate": 0.0005856666858374949,
      "loss": 2.5178,
      "step": 22762
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6126210689544678,
      "learning_rate": 0.0005856654365294255,
      "loss": 2.3412,
      "step": 22763
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6484808921813965,
      "learning_rate": 0.0005856641871682457,
      "loss": 2.2934,
      "step": 22764
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.7011069655418396,
      "learning_rate": 0.0005856629377539555,
      "loss": 2.5138,
      "step": 22765
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6464223265647888,
      "learning_rate": 0.0005856616882865552,
      "loss": 2.4267,
      "step": 22766
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.5887444615364075,
      "learning_rate": 0.0005856604387660449,
      "loss": 2.5695,
      "step": 22767
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6448846459388733,
      "learning_rate": 0.000585659189192425,
      "loss": 2.1909,
      "step": 22768
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6206989288330078,
      "learning_rate": 0.0005856579395656956,
      "loss": 2.4305,
      "step": 22769
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.8518843054771423,
      "learning_rate": 0.000585656689885857,
      "loss": 2.3197,
      "step": 22770
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6319823265075684,
      "learning_rate": 0.0005856554401529095,
      "loss": 2.0993,
      "step": 22771
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6731138229370117,
      "learning_rate": 0.0005856541903668533,
      "loss": 2.0985,
      "step": 22772
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.7521506547927856,
      "learning_rate": 0.0005856529405276884,
      "loss": 2.2836,
      "step": 22773
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.7123811841011047,
      "learning_rate": 0.0005856516906354154,
      "loss": 2.3171,
      "step": 22774
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6594160199165344,
      "learning_rate": 0.0005856504406900343,
      "loss": 2.2573,
      "step": 22775
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.9200899004936218,
      "learning_rate": 0.0005856491906915454,
      "loss": 2.4633,
      "step": 22776
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6519778370857239,
      "learning_rate": 0.000585647940639949,
      "loss": 2.2931,
      "step": 22777
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6712919473648071,
      "learning_rate": 0.0005856466905352451,
      "loss": 2.4018,
      "step": 22778
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.687099814414978,
      "learning_rate": 0.0005856454403774343,
      "loss": 2.6053,
      "step": 22779
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6761068105697632,
      "learning_rate": 0.0005856441901665164,
      "loss": 2.252,
      "step": 22780
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6925272941589355,
      "learning_rate": 0.0005856429399024919,
      "loss": 2.1988,
      "step": 22781
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6531601548194885,
      "learning_rate": 0.0005856416895853611,
      "loss": 2.638,
      "step": 22782
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6780882477760315,
      "learning_rate": 0.0005856404392151239,
      "loss": 2.4867,
      "step": 22783
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6805292367935181,
      "learning_rate": 0.0005856391887917809,
      "loss": 2.3391,
      "step": 22784
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.5815324187278748,
      "learning_rate": 0.0005856379383153323,
      "loss": 2.2094,
      "step": 22785
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.7225025296211243,
      "learning_rate": 0.000585636687785778,
      "loss": 2.1789,
      "step": 22786
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6694965362548828,
      "learning_rate": 0.0005856354372031186,
      "loss": 2.1938,
      "step": 22787
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6045805215835571,
      "learning_rate": 0.0005856341865673541,
      "loss": 2.388,
      "step": 22788
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6213579177856445,
      "learning_rate": 0.0005856329358784848,
      "loss": 2.1312,
      "step": 22789
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.7338509559631348,
      "learning_rate": 0.000585631685136511,
      "loss": 2.3676,
      "step": 22790
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6116598844528198,
      "learning_rate": 0.0005856304343414328,
      "loss": 2.5262,
      "step": 22791
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.5833805799484253,
      "learning_rate": 0.0005856291834932506,
      "loss": 2.1648,
      "step": 22792
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.5803082585334778,
      "learning_rate": 0.0005856279325919646,
      "loss": 2.4013,
      "step": 22793
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.7166169285774231,
      "learning_rate": 0.0005856266816375748,
      "loss": 2.3602,
      "step": 22794
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.5948235392570496,
      "learning_rate": 0.0005856254306300817,
      "loss": 2.5097,
      "step": 22795
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.5763047933578491,
      "learning_rate": 0.0005856241795694855,
      "loss": 2.313,
      "step": 22796
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6471887826919556,
      "learning_rate": 0.0005856229284557863,
      "loss": 2.5418,
      "step": 22797
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6623044610023499,
      "learning_rate": 0.0005856216772889844,
      "loss": 2.2681,
      "step": 22798
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6373446583747864,
      "learning_rate": 0.0005856204260690801,
      "loss": 2.2122,
      "step": 22799
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6483263373374939,
      "learning_rate": 0.0005856191747960735,
      "loss": 2.5656,
      "step": 22800
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.661651074886322,
      "learning_rate": 0.0005856179234699649,
      "loss": 2.3348,
      "step": 22801
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6241543889045715,
      "learning_rate": 0.0005856166720907545,
      "loss": 2.5269,
      "step": 22802
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6729534864425659,
      "learning_rate": 0.0005856154206584427,
      "loss": 2.2413,
      "step": 22803
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.5898981094360352,
      "learning_rate": 0.0005856141691730295,
      "loss": 2.4266,
      "step": 22804
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.8405860066413879,
      "learning_rate": 0.0005856129176345154,
      "loss": 2.343,
      "step": 22805
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6375419497489929,
      "learning_rate": 0.0005856116660429003,
      "loss": 2.3764,
      "step": 22806
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6667471528053284,
      "learning_rate": 0.0005856104143981846,
      "loss": 2.1262,
      "step": 22807
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.8913766145706177,
      "learning_rate": 0.0005856091627003685,
      "loss": 2.0789,
      "step": 22808
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.7720472812652588,
      "learning_rate": 0.0005856079109494524,
      "loss": 2.597,
      "step": 22809
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.7046300172805786,
      "learning_rate": 0.0005856066591454364,
      "loss": 2.2956,
      "step": 22810
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6220241785049438,
      "learning_rate": 0.0005856054072883206,
      "loss": 2.3443,
      "step": 22811
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6708984971046448,
      "learning_rate": 0.0005856041553781054,
      "loss": 2.1925,
      "step": 22812
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6713219881057739,
      "learning_rate": 0.0005856029034147911,
      "loss": 2.3342,
      "step": 22813
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6771467924118042,
      "learning_rate": 0.0005856016513983777,
      "loss": 2.2159,
      "step": 22814
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6109189987182617,
      "learning_rate": 0.0005856003993288657,
      "loss": 2.3501,
      "step": 22815
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.653929591178894,
      "learning_rate": 0.000585599147206255,
      "loss": 2.4513,
      "step": 22816
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.7358899116516113,
      "learning_rate": 0.0005855978950305461,
      "loss": 2.2635,
      "step": 22817
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6044290065765381,
      "learning_rate": 0.0005855966428017392,
      "loss": 2.2543,
      "step": 22818
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6893331408500671,
      "learning_rate": 0.0005855953905198345,
      "loss": 2.3469,
      "step": 22819
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.7194303274154663,
      "learning_rate": 0.0005855941381848322,
      "loss": 2.2774,
      "step": 22820
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6807002425193787,
      "learning_rate": 0.0005855928857967326,
      "loss": 2.3151,
      "step": 22821
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.7307453751564026,
      "learning_rate": 0.0005855916333555359,
      "loss": 2.277,
      "step": 22822
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.7521135210990906,
      "learning_rate": 0.0005855903808612423,
      "loss": 2.4277,
      "step": 22823
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6309458017349243,
      "learning_rate": 0.000585589128313852,
      "loss": 2.4486,
      "step": 22824
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6404184699058533,
      "learning_rate": 0.0005855878757133654,
      "loss": 2.3375,
      "step": 22825
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.5894700288772583,
      "learning_rate": 0.0005855866230597825,
      "loss": 2.4769,
      "step": 22826
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.8343960046768188,
      "learning_rate": 0.0005855853703531038,
      "loss": 2.2888,
      "step": 22827
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6340977549552917,
      "learning_rate": 0.0005855841175933293,
      "loss": 2.1499,
      "step": 22828
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6022235751152039,
      "learning_rate": 0.0005855828647804593,
      "loss": 2.4001,
      "step": 22829
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.7946213483810425,
      "learning_rate": 0.0005855816119144941,
      "loss": 2.2053,
      "step": 22830
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.7625089287757874,
      "learning_rate": 0.0005855803589954339,
      "loss": 2.53,
      "step": 22831
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.585574746131897,
      "learning_rate": 0.0005855791060232789,
      "loss": 2.1961,
      "step": 22832
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6401146054267883,
      "learning_rate": 0.0005855778529980293,
      "loss": 2.5334,
      "step": 22833
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.8047348260879517,
      "learning_rate": 0.0005855765999196854,
      "loss": 2.2956,
      "step": 22834
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.7092089056968689,
      "learning_rate": 0.0005855753467882475,
      "loss": 2.257,
      "step": 22835
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.5667331218719482,
      "learning_rate": 0.0005855740936037157,
      "loss": 2.4472,
      "step": 22836
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6510978937149048,
      "learning_rate": 0.0005855728403660902,
      "loss": 2.1061,
      "step": 22837
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6585808396339417,
      "learning_rate": 0.0005855715870753715,
      "loss": 2.413,
      "step": 22838
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.7246180772781372,
      "learning_rate": 0.0005855703337315595,
      "loss": 2.1368,
      "step": 22839
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6385607719421387,
      "learning_rate": 0.0005855690803346547,
      "loss": 2.5034,
      "step": 22840
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6239228844642639,
      "learning_rate": 0.0005855678268846572,
      "loss": 2.4105,
      "step": 22841
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6986321210861206,
      "learning_rate": 0.0005855665733815671,
      "loss": 2.1918,
      "step": 22842
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6663874983787537,
      "learning_rate": 0.000585565319825385,
      "loss": 2.378,
      "step": 22843
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.641184389591217,
      "learning_rate": 0.0005855640662161107,
      "loss": 2.2943,
      "step": 22844
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6396353244781494,
      "learning_rate": 0.0005855628125537447,
      "loss": 2.2155,
      "step": 22845
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6499223113059998,
      "learning_rate": 0.0005855615588382873,
      "loss": 2.1893,
      "step": 22846
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6405664682388306,
      "learning_rate": 0.0005855603050697386,
      "loss": 2.4394,
      "step": 22847
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6941046714782715,
      "learning_rate": 0.0005855590512480988,
      "loss": 2.276,
      "step": 22848
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6445363163948059,
      "learning_rate": 0.0005855577973733682,
      "loss": 2.3442,
      "step": 22849
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.747208297252655,
      "learning_rate": 0.000585556543445547,
      "loss": 2.429,
      "step": 22850
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6471468210220337,
      "learning_rate": 0.0005855552894646355,
      "loss": 2.3822,
      "step": 22851
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6291349530220032,
      "learning_rate": 0.0005855540354306338,
      "loss": 2.5615,
      "step": 22852
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6995058655738831,
      "learning_rate": 0.0005855527813435423,
      "loss": 2.4017,
      "step": 22853
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6091071367263794,
      "learning_rate": 0.0005855515272033611,
      "loss": 2.3688,
      "step": 22854
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.7168489098548889,
      "learning_rate": 0.0005855502730100904,
      "loss": 2.3762,
      "step": 22855
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6402578949928284,
      "learning_rate": 0.0005855490187637307,
      "loss": 2.3103,
      "step": 22856
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.7536543011665344,
      "learning_rate": 0.0005855477644642819,
      "loss": 2.0372,
      "step": 22857
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.8008838295936584,
      "learning_rate": 0.0005855465101117445,
      "loss": 2.3364,
      "step": 22858
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6654618978500366,
      "learning_rate": 0.0005855452557061184,
      "loss": 2.2933,
      "step": 22859
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6805926561355591,
      "learning_rate": 0.0005855440012474042,
      "loss": 2.5059,
      "step": 22860
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.622554361820221,
      "learning_rate": 0.0005855427467356019,
      "loss": 2.2862,
      "step": 22861
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.5630868673324585,
      "learning_rate": 0.0005855414921707119,
      "loss": 2.3889,
      "step": 22862
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6836686730384827,
      "learning_rate": 0.0005855402375527344,
      "loss": 2.5132,
      "step": 22863
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6046960353851318,
      "learning_rate": 0.0005855389828816694,
      "loss": 2.4366,
      "step": 22864
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.5873668193817139,
      "learning_rate": 0.0005855377281575175,
      "loss": 2.2921,
      "step": 22865
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6398365497589111,
      "learning_rate": 0.0005855364733802787,
      "loss": 2.1551,
      "step": 22866
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6510084867477417,
      "learning_rate": 0.0005855352185499532,
      "loss": 2.3506,
      "step": 22867
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6098311543464661,
      "learning_rate": 0.0005855339636665414,
      "loss": 2.2715,
      "step": 22868
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6637264490127563,
      "learning_rate": 0.0005855327087300433,
      "loss": 2.6337,
      "step": 22869
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6444315910339355,
      "learning_rate": 0.0005855314537404594,
      "loss": 2.5749,
      "step": 22870
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6174612045288086,
      "learning_rate": 0.0005855301986977898,
      "loss": 2.443,
      "step": 22871
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6521400213241577,
      "learning_rate": 0.0005855289436020348,
      "loss": 2.4138,
      "step": 22872
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.56406170129776,
      "learning_rate": 0.0005855276884531945,
      "loss": 2.0774,
      "step": 22873
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.5871613621711731,
      "learning_rate": 0.0005855264332512692,
      "loss": 2.3954,
      "step": 22874
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6247346997261047,
      "learning_rate": 0.0005855251779962592,
      "loss": 2.2479,
      "step": 22875
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.5894052386283875,
      "learning_rate": 0.0005855239226881646,
      "loss": 2.227,
      "step": 22876
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.5749943256378174,
      "learning_rate": 0.0005855226673269858,
      "loss": 2.6371,
      "step": 22877
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6382158398628235,
      "learning_rate": 0.0005855214119127229,
      "loss": 2.3403,
      "step": 22878
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6033471822738647,
      "learning_rate": 0.0005855201564453761,
      "loss": 2.3007,
      "step": 22879
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6268160343170166,
      "learning_rate": 0.0005855189009249459,
      "loss": 2.5038,
      "step": 22880
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6382197141647339,
      "learning_rate": 0.0005855176453514323,
      "loss": 2.3532,
      "step": 22881
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.7949873805046082,
      "learning_rate": 0.0005855163897248355,
      "loss": 2.3629,
      "step": 22882
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6688769459724426,
      "learning_rate": 0.0005855151340451558,
      "loss": 2.593,
      "step": 22883
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.669630229473114,
      "learning_rate": 0.0005855138783123935,
      "loss": 2.4279,
      "step": 22884
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6448851823806763,
      "learning_rate": 0.0005855126225265488,
      "loss": 2.2131,
      "step": 22885
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6686293482780457,
      "learning_rate": 0.0005855113666876219,
      "loss": 2.3402,
      "step": 22886
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.569729208946228,
      "learning_rate": 0.000585510110795613,
      "loss": 2.2846,
      "step": 22887
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.5710384249687195,
      "learning_rate": 0.0005855088548505225,
      "loss": 2.248,
      "step": 22888
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6226708889007568,
      "learning_rate": 0.0005855075988523504,
      "loss": 2.2924,
      "step": 22889
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6392664313316345,
      "learning_rate": 0.0005855063428010971,
      "loss": 2.5328,
      "step": 22890
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.7151001691818237,
      "learning_rate": 0.0005855050866967628,
      "loss": 2.3057,
      "step": 22891
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.5974056720733643,
      "learning_rate": 0.0005855038305393477,
      "loss": 2.2715,
      "step": 22892
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6405742168426514,
      "learning_rate": 0.000585502574328852,
      "loss": 2.4482,
      "step": 22893
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.5841674208641052,
      "learning_rate": 0.0005855013180652759,
      "loss": 2.5107,
      "step": 22894
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.5943900346755981,
      "learning_rate": 0.0005855000617486199,
      "loss": 2.2434,
      "step": 22895
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6625385284423828,
      "learning_rate": 0.000585498805378884,
      "loss": 2.5022,
      "step": 22896
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6219693422317505,
      "learning_rate": 0.0005854975489560684,
      "loss": 2.4164,
      "step": 22897
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.5914860367774963,
      "learning_rate": 0.0005854962924801735,
      "loss": 2.2401,
      "step": 22898
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.660271942615509,
      "learning_rate": 0.0005854950359511994,
      "loss": 2.4657,
      "step": 22899
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6347837448120117,
      "learning_rate": 0.0005854937793691464,
      "loss": 2.5161,
      "step": 22900
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.59058678150177,
      "learning_rate": 0.0005854925227340148,
      "loss": 2.3587,
      "step": 22901
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.5948511958122253,
      "learning_rate": 0.0005854912660458046,
      "loss": 2.3398,
      "step": 22902
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6868765950202942,
      "learning_rate": 0.0005854900093045163,
      "loss": 2.4369,
      "step": 22903
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6853068470954895,
      "learning_rate": 0.0005854887525101499,
      "loss": 2.2418,
      "step": 22904
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6448009014129639,
      "learning_rate": 0.0005854874956627059,
      "loss": 2.3573,
      "step": 22905
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6428016424179077,
      "learning_rate": 0.0005854862387621843,
      "loss": 2.4722,
      "step": 22906
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6585713028907776,
      "learning_rate": 0.0005854849818085854,
      "loss": 2.3173,
      "step": 22907
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.7136391997337341,
      "learning_rate": 0.0005854837248019095,
      "loss": 2.2569,
      "step": 22908
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.651798665523529,
      "learning_rate": 0.0005854824677421569,
      "loss": 2.4335,
      "step": 22909
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.8418160080909729,
      "learning_rate": 0.0005854812106293276,
      "loss": 2.3992,
      "step": 22910
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.5972301363945007,
      "learning_rate": 0.0005854799534634218,
      "loss": 2.4183,
      "step": 22911
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6490210294723511,
      "learning_rate": 0.00058547869624444,
      "loss": 2.2532,
      "step": 22912
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6582841873168945,
      "learning_rate": 0.0005854774389723825,
      "loss": 2.219,
      "step": 22913
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6990581750869751,
      "learning_rate": 0.0005854761816472491,
      "loss": 2.2214,
      "step": 22914
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6524949669837952,
      "learning_rate": 0.0005854749242690404,
      "loss": 2.2042,
      "step": 22915
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6956087350845337,
      "learning_rate": 0.0005854736668377565,
      "loss": 2.1364,
      "step": 22916
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.7105384469032288,
      "learning_rate": 0.0005854724093533977,
      "loss": 2.5176,
      "step": 22917
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6595457792282104,
      "learning_rate": 0.0005854711518159643,
      "loss": 2.5795,
      "step": 22918
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6350311636924744,
      "learning_rate": 0.0005854698942254562,
      "loss": 2.3945,
      "step": 22919
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6401134133338928,
      "learning_rate": 0.000585468636581874,
      "loss": 2.1871,
      "step": 22920
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6003631353378296,
      "learning_rate": 0.0005854673788852177,
      "loss": 2.3678,
      "step": 22921
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6483247876167297,
      "learning_rate": 0.0005854661211354877,
      "loss": 2.1114,
      "step": 22922
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.605034351348877,
      "learning_rate": 0.0005854648633326842,
      "loss": 2.4606,
      "step": 22923
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.7105035185813904,
      "learning_rate": 0.0005854636054768073,
      "loss": 2.3213,
      "step": 22924
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.661045253276825,
      "learning_rate": 0.0005854623475678573,
      "loss": 2.3521,
      "step": 22925
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6428069472312927,
      "learning_rate": 0.0005854610896058346,
      "loss": 2.5238,
      "step": 22926
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6235846281051636,
      "learning_rate": 0.0005854598315907392,
      "loss": 2.2833,
      "step": 22927
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.5593435168266296,
      "learning_rate": 0.0005854585735225714,
      "loss": 2.1996,
      "step": 22928
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6104913353919983,
      "learning_rate": 0.0005854573154013315,
      "loss": 2.4825,
      "step": 22929
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.5742684006690979,
      "learning_rate": 0.0005854560572270197,
      "loss": 2.3093,
      "step": 22930
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6600301265716553,
      "learning_rate": 0.0005854547989996362,
      "loss": 2.4681,
      "step": 22931
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.7226242423057556,
      "learning_rate": 0.0005854535407191813,
      "loss": 2.2778,
      "step": 22932
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.5867406129837036,
      "learning_rate": 0.0005854522823856552,
      "loss": 2.3687,
      "step": 22933
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6828544735908508,
      "learning_rate": 0.0005854510239990581,
      "loss": 2.4933,
      "step": 22934
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6027602553367615,
      "learning_rate": 0.0005854497655593903,
      "loss": 2.2792,
      "step": 22935
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.657875120639801,
      "learning_rate": 0.000585448507066652,
      "loss": 2.1841,
      "step": 22936
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6595340967178345,
      "learning_rate": 0.0005854472485208433,
      "loss": 2.6141,
      "step": 22937
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6148855090141296,
      "learning_rate": 0.0005854459899219648,
      "loss": 2.468,
      "step": 22938
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6024457812309265,
      "learning_rate": 0.0005854447312700164,
      "loss": 2.5008,
      "step": 22939
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6472818851470947,
      "learning_rate": 0.0005854434725649985,
      "loss": 2.1464,
      "step": 22940
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6237838864326477,
      "learning_rate": 0.0005854422138069111,
      "loss": 2.376,
      "step": 22941
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6897937655448914,
      "learning_rate": 0.0005854409549957547,
      "loss": 2.4931,
      "step": 22942
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.634951651096344,
      "learning_rate": 0.0005854396961315295,
      "loss": 2.3215,
      "step": 22943
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6253550052642822,
      "learning_rate": 0.0005854384372142356,
      "loss": 2.3759,
      "step": 22944
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.7030522227287292,
      "learning_rate": 0.0005854371782438733,
      "loss": 2.4988,
      "step": 22945
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.5885396599769592,
      "learning_rate": 0.0005854359192204429,
      "loss": 2.2575,
      "step": 22946
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6064016819000244,
      "learning_rate": 0.0005854346601439445,
      "loss": 2.2641,
      "step": 22947
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6579204201698303,
      "learning_rate": 0.0005854334010143784,
      "loss": 2.1595,
      "step": 22948
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.7276551723480225,
      "learning_rate": 0.0005854321418317449,
      "loss": 2.2224,
      "step": 22949
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6391649842262268,
      "learning_rate": 0.0005854308825960442,
      "loss": 2.3934,
      "step": 22950
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.7345120906829834,
      "learning_rate": 0.0005854296233072765,
      "loss": 2.3255,
      "step": 22951
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6826996803283691,
      "learning_rate": 0.000585428363965442,
      "loss": 2.2754,
      "step": 22952
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.5895020961761475,
      "learning_rate": 0.000585427104570541,
      "loss": 2.3559,
      "step": 22953
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6467206478118896,
      "learning_rate": 0.0005854258451225737,
      "loss": 2.3096,
      "step": 22954
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6892069578170776,
      "learning_rate": 0.0005854245856215404,
      "loss": 2.4484,
      "step": 22955
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6699364185333252,
      "learning_rate": 0.0005854233260674412,
      "loss": 2.3301,
      "step": 22956
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6574783325195312,
      "learning_rate": 0.0005854220664602765,
      "loss": 2.1946,
      "step": 22957
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6760389804840088,
      "learning_rate": 0.0005854208068000464,
      "loss": 2.487,
      "step": 22958
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.7540900707244873,
      "learning_rate": 0.0005854195470867512,
      "loss": 2.4456,
      "step": 22959
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.7128425240516663,
      "learning_rate": 0.0005854182873203911,
      "loss": 2.2672,
      "step": 22960
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6375112533569336,
      "learning_rate": 0.0005854170275009664,
      "loss": 2.439,
      "step": 22961
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6544685363769531,
      "learning_rate": 0.0005854157676284772,
      "loss": 2.2326,
      "step": 22962
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6823172569274902,
      "learning_rate": 0.0005854145077029239,
      "loss": 2.3484,
      "step": 22963
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.790119469165802,
      "learning_rate": 0.0005854132477243066,
      "loss": 2.6686,
      "step": 22964
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6810285449028015,
      "learning_rate": 0.0005854119876926256,
      "loss": 2.4557,
      "step": 22965
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.8305625915527344,
      "learning_rate": 0.0005854107276078811,
      "loss": 2.4023,
      "step": 22966
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6595916152000427,
      "learning_rate": 0.0005854094674700735,
      "loss": 2.6196,
      "step": 22967
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6669105291366577,
      "learning_rate": 0.0005854082072792027,
      "loss": 2.3581,
      "step": 22968
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.76530522108078,
      "learning_rate": 0.0005854069470352692,
      "loss": 2.2862,
      "step": 22969
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6256659030914307,
      "learning_rate": 0.0005854056867382733,
      "loss": 2.1424,
      "step": 22970
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.7255831360816956,
      "learning_rate": 0.000585404426388215,
      "loss": 2.5171,
      "step": 22971
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6172863245010376,
      "learning_rate": 0.0005854031659850946,
      "loss": 2.2367,
      "step": 22972
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.7058067321777344,
      "learning_rate": 0.0005854019055289123,
      "loss": 2.1528,
      "step": 22973
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6511839032173157,
      "learning_rate": 0.0005854006450196685,
      "loss": 2.3613,
      "step": 22974
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6763205528259277,
      "learning_rate": 0.0005853993844573633,
      "loss": 2.1812,
      "step": 22975
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6744667887687683,
      "learning_rate": 0.000585398123841997,
      "loss": 2.1259,
      "step": 22976
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6642630100250244,
      "learning_rate": 0.0005853968631735698,
      "loss": 2.374,
      "step": 22977
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.5748379230499268,
      "learning_rate": 0.0005853956024520818,
      "loss": 2.2724,
      "step": 22978
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6284176707267761,
      "learning_rate": 0.0005853943416775335,
      "loss": 2.4684,
      "step": 22979
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6095319390296936,
      "learning_rate": 0.000585393080849925,
      "loss": 2.2075,
      "step": 22980
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6504237055778503,
      "learning_rate": 0.0005853918199692565,
      "loss": 2.2161,
      "step": 22981
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6386438608169556,
      "learning_rate": 0.0005853905590355282,
      "loss": 2.2805,
      "step": 22982
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6325228214263916,
      "learning_rate": 0.0005853892980487406,
      "loss": 2.3751,
      "step": 22983
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6301680207252502,
      "learning_rate": 0.0005853880370088936,
      "loss": 2.5639,
      "step": 22984
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6088030934333801,
      "learning_rate": 0.0005853867759159876,
      "loss": 2.3289,
      "step": 22985
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6225309371948242,
      "learning_rate": 0.0005853855147700228,
      "loss": 2.3316,
      "step": 22986
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.5894404053688049,
      "learning_rate": 0.0005853842535709996,
      "loss": 2.2068,
      "step": 22987
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6146872639656067,
      "learning_rate": 0.000585382992318918,
      "loss": 2.6511,
      "step": 22988
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.675731360912323,
      "learning_rate": 0.0005853817310137782,
      "loss": 2.2595,
      "step": 22989
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.5910192728042603,
      "learning_rate": 0.0005853804696555806,
      "loss": 2.3006,
      "step": 22990
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6926197409629822,
      "learning_rate": 0.0005853792082443255,
      "loss": 2.4543,
      "step": 22991
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6714377403259277,
      "learning_rate": 0.0005853779467800129,
      "loss": 2.4195,
      "step": 22992
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6118062734603882,
      "learning_rate": 0.000585376685262643,
      "loss": 2.2713,
      "step": 22993
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6670289635658264,
      "learning_rate": 0.0005853754236922165,
      "loss": 2.3004,
      "step": 22994
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.5805383324623108,
      "learning_rate": 0.0005853741620687332,
      "loss": 2.4419,
      "step": 22995
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.5995901823043823,
      "learning_rate": 0.0005853729003921933,
      "loss": 2.2129,
      "step": 22996
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6438453793525696,
      "learning_rate": 0.0005853716386625974,
      "loss": 2.3905,
      "step": 22997
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6634480953216553,
      "learning_rate": 0.0005853703768799455,
      "loss": 2.4719,
      "step": 22998
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6322416663169861,
      "learning_rate": 0.0005853691150442378,
      "loss": 2.2914,
      "step": 22999
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6160828471183777,
      "learning_rate": 0.0005853678531554747,
      "loss": 2.3977,
      "step": 23000
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6698728799819946,
      "learning_rate": 0.0005853665912136562,
      "loss": 2.2236,
      "step": 23001
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.5798138976097107,
      "learning_rate": 0.0005853653292187827,
      "loss": 2.346,
      "step": 23002
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6483283042907715,
      "learning_rate": 0.0005853640671708544,
      "loss": 2.1196,
      "step": 23003
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.635842502117157,
      "learning_rate": 0.0005853628050698715,
      "loss": 2.5499,
      "step": 23004
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6531243324279785,
      "learning_rate": 0.0005853615429158343,
      "loss": 2.4303,
      "step": 23005
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.5896225571632385,
      "learning_rate": 0.000585360280708743,
      "loss": 2.3702,
      "step": 23006
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6319963932037354,
      "learning_rate": 0.0005853590184485978,
      "loss": 2.3769,
      "step": 23007
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6958357691764832,
      "learning_rate": 0.0005853577561353991,
      "loss": 2.4366,
      "step": 23008
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6174589991569519,
      "learning_rate": 0.000585356493769147,
      "loss": 2.1926,
      "step": 23009
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6336919069290161,
      "learning_rate": 0.0005853552313498416,
      "loss": 2.3629,
      "step": 23010
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6435483694076538,
      "learning_rate": 0.0005853539688774834,
      "loss": 2.2371,
      "step": 23011
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.5975717306137085,
      "learning_rate": 0.0005853527063520723,
      "loss": 2.2044,
      "step": 23012
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6447381377220154,
      "learning_rate": 0.0005853514437736089,
      "loss": 2.4595,
      "step": 23013
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6246134638786316,
      "learning_rate": 0.0005853501811420934,
      "loss": 2.459,
      "step": 23014
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.7673783898353577,
      "learning_rate": 0.0005853489184575258,
      "loss": 2.4044,
      "step": 23015
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.5619645714759827,
      "learning_rate": 0.0005853476557199064,
      "loss": 2.2032,
      "step": 23016
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6178178191184998,
      "learning_rate": 0.0005853463929292355,
      "loss": 2.1948,
      "step": 23017
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6456907987594604,
      "learning_rate": 0.0005853451300855134,
      "loss": 2.1923,
      "step": 23018
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6436534523963928,
      "learning_rate": 0.0005853438671887401,
      "loss": 2.1935,
      "step": 23019
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6943212151527405,
      "learning_rate": 0.0005853426042389162,
      "loss": 2.1163,
      "step": 23020
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6618263125419617,
      "learning_rate": 0.0005853413412360416,
      "loss": 2.3255,
      "step": 23021
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6803529262542725,
      "learning_rate": 0.0005853400781801166,
      "loss": 2.3974,
      "step": 23022
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.5980973243713379,
      "learning_rate": 0.0005853388150711416,
      "loss": 2.4998,
      "step": 23023
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6737265586853027,
      "learning_rate": 0.0005853375519091167,
      "loss": 2.2348,
      "step": 23024
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.660483181476593,
      "learning_rate": 0.0005853362886940422,
      "loss": 2.4149,
      "step": 23025
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.7438337206840515,
      "learning_rate": 0.0005853350254259182,
      "loss": 2.3951,
      "step": 23026
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6104307174682617,
      "learning_rate": 0.0005853337621047451,
      "loss": 2.3832,
      "step": 23027
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.7451360821723938,
      "learning_rate": 0.0005853324987305231,
      "loss": 2.3463,
      "step": 23028
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6427340507507324,
      "learning_rate": 0.0005853312353032523,
      "loss": 2.6085,
      "step": 23029
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6091526746749878,
      "learning_rate": 0.0005853299718229331,
      "loss": 2.4088,
      "step": 23030
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.7319037318229675,
      "learning_rate": 0.0005853287082895656,
      "loss": 2.2358,
      "step": 23031
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6087214946746826,
      "learning_rate": 0.0005853274447031501,
      "loss": 2.3745,
      "step": 23032
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6447780132293701,
      "learning_rate": 0.0005853261810636871,
      "loss": 2.2374,
      "step": 23033
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6227220296859741,
      "learning_rate": 0.0005853249173711763,
      "loss": 2.1749,
      "step": 23034
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6038976311683655,
      "learning_rate": 0.0005853236536256184,
      "loss": 2.421,
      "step": 23035
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6491398811340332,
      "learning_rate": 0.0005853223898270133,
      "loss": 2.3407,
      "step": 23036
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6691486239433289,
      "learning_rate": 0.0005853211259753614,
      "loss": 2.4419,
      "step": 23037
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6258876919746399,
      "learning_rate": 0.000585319862070663,
      "loss": 2.2679,
      "step": 23038
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.5961533188819885,
      "learning_rate": 0.0005853185981129182,
      "loss": 2.2446,
      "step": 23039
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6080077290534973,
      "learning_rate": 0.0005853173341021273,
      "loss": 2.2082,
      "step": 23040
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6001861691474915,
      "learning_rate": 0.0005853160700382905,
      "loss": 2.197,
      "step": 23041
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6212047934532166,
      "learning_rate": 0.000585314805921408,
      "loss": 2.2982,
      "step": 23042
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6403180956840515,
      "learning_rate": 0.0005853135417514802,
      "loss": 2.2131,
      "step": 23043
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6417934894561768,
      "learning_rate": 0.0005853122775285072,
      "loss": 2.3719,
      "step": 23044
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.5876855850219727,
      "learning_rate": 0.0005853110132524892,
      "loss": 2.2816,
      "step": 23045
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.5864166617393494,
      "learning_rate": 0.0005853097489234265,
      "loss": 2.2178,
      "step": 23046
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.658955991268158,
      "learning_rate": 0.0005853084845413194,
      "loss": 2.4213,
      "step": 23047
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6562253832817078,
      "learning_rate": 0.0005853072201061681,
      "loss": 2.432,
      "step": 23048
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.5652223229408264,
      "learning_rate": 0.0005853059556179727,
      "loss": 2.1943,
      "step": 23049
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6512643098831177,
      "learning_rate": 0.0005853046910767335,
      "loss": 2.3948,
      "step": 23050
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6678904891014099,
      "learning_rate": 0.0005853034264824509,
      "loss": 2.2911,
      "step": 23051
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6737334132194519,
      "learning_rate": 0.0005853021618351251,
      "loss": 2.3472,
      "step": 23052
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.5412242412567139,
      "learning_rate": 0.0005853008971347561,
      "loss": 2.378,
      "step": 23053
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6097880005836487,
      "learning_rate": 0.0005852996323813442,
      "loss": 1.9906,
      "step": 23054
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6486107707023621,
      "learning_rate": 0.0005852983675748898,
      "loss": 2.3994,
      "step": 23055
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.7403272986412048,
      "learning_rate": 0.0005852971027153931,
      "loss": 2.506,
      "step": 23056
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6248809695243835,
      "learning_rate": 0.0005852958378028541,
      "loss": 2.2241,
      "step": 23057
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.7126094102859497,
      "learning_rate": 0.0005852945728372734,
      "loss": 2.3536,
      "step": 23058
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6836820244789124,
      "learning_rate": 0.0005852933078186511,
      "loss": 2.3531,
      "step": 23059
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.846674382686615,
      "learning_rate": 0.0005852920427469872,
      "loss": 2.377,
      "step": 23060
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6536315679550171,
      "learning_rate": 0.0005852907776222822,
      "loss": 2.4373,
      "step": 23061
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.8426722884178162,
      "learning_rate": 0.0005852895124445363,
      "loss": 2.2691,
      "step": 23062
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.7160772681236267,
      "learning_rate": 0.0005852882472137496,
      "loss": 2.2133,
      "step": 23063
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6726346611976624,
      "learning_rate": 0.0005852869819299225,
      "loss": 2.323,
      "step": 23064
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6311158537864685,
      "learning_rate": 0.0005852857165930551,
      "loss": 2.2338,
      "step": 23065
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6948297619819641,
      "learning_rate": 0.0005852844512031479,
      "loss": 2.3268,
      "step": 23066
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.8178634643554688,
      "learning_rate": 0.0005852831857602008,
      "loss": 2.2834,
      "step": 23067
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6470155119895935,
      "learning_rate": 0.0005852819202642141,
      "loss": 2.346,
      "step": 23068
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6749481558799744,
      "learning_rate": 0.0005852806547151882,
      "loss": 2.2417,
      "step": 23069
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.7447367310523987,
      "learning_rate": 0.0005852793891131232,
      "loss": 2.4255,
      "step": 23070
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.7333180904388428,
      "learning_rate": 0.0005852781234580194,
      "loss": 2.5384,
      "step": 23071
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6187593936920166,
      "learning_rate": 0.000585276857749877,
      "loss": 2.2071,
      "step": 23072
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6265416145324707,
      "learning_rate": 0.0005852755919886962,
      "loss": 2.3213,
      "step": 23073
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6790817975997925,
      "learning_rate": 0.0005852743261744774,
      "loss": 2.1878,
      "step": 23074
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6768001317977905,
      "learning_rate": 0.0005852730603072206,
      "loss": 2.2974,
      "step": 23075
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.60670006275177,
      "learning_rate": 0.0005852717943869262,
      "loss": 2.1738,
      "step": 23076
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.637452244758606,
      "learning_rate": 0.0005852705284135944,
      "loss": 2.1846,
      "step": 23077
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.7782952189445496,
      "learning_rate": 0.0005852692623872255,
      "loss": 2.4339,
      "step": 23078
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6009678244590759,
      "learning_rate": 0.0005852679963078194,
      "loss": 2.5494,
      "step": 23079
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6870774626731873,
      "learning_rate": 0.0005852667301753768,
      "loss": 2.3347,
      "step": 23080
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6925985217094421,
      "learning_rate": 0.0005852654639898978,
      "loss": 2.489,
      "step": 23081
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.7531340718269348,
      "learning_rate": 0.0005852641977513825,
      "loss": 2.3941,
      "step": 23082
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6330543756484985,
      "learning_rate": 0.0005852629314598312,
      "loss": 2.3977,
      "step": 23083
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6492253541946411,
      "learning_rate": 0.000585261665115244,
      "loss": 2.2605,
      "step": 23084
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6325399279594421,
      "learning_rate": 0.0005852603987176214,
      "loss": 2.4737,
      "step": 23085
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6314283609390259,
      "learning_rate": 0.0005852591322669635,
      "loss": 2.4809,
      "step": 23086
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6540313363075256,
      "learning_rate": 0.0005852578657632705,
      "loss": 2.5148,
      "step": 23087
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6333711743354797,
      "learning_rate": 0.0005852565992065427,
      "loss": 2.2762,
      "step": 23088
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.7502557039260864,
      "learning_rate": 0.0005852553325967803,
      "loss": 2.2528,
      "step": 23089
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6214126944541931,
      "learning_rate": 0.0005852540659339835,
      "loss": 2.2013,
      "step": 23090
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6981953382492065,
      "learning_rate": 0.0005852527992181527,
      "loss": 2.3347,
      "step": 23091
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6471696496009827,
      "learning_rate": 0.000585251532449288,
      "loss": 2.4689,
      "step": 23092
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.5922691226005554,
      "learning_rate": 0.0005852502656273896,
      "loss": 2.2887,
      "step": 23093
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.5690873861312866,
      "learning_rate": 0.0005852489987524578,
      "loss": 2.4321,
      "step": 23094
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.666249692440033,
      "learning_rate": 0.0005852477318244928,
      "loss": 2.423,
      "step": 23095
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.7118594646453857,
      "learning_rate": 0.000585246464843495,
      "loss": 2.4254,
      "step": 23096
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6701691150665283,
      "learning_rate": 0.0005852451978094643,
      "loss": 2.2909,
      "step": 23097
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6411784291267395,
      "learning_rate": 0.0005852439307224013,
      "loss": 2.3656,
      "step": 23098
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.5502027273178101,
      "learning_rate": 0.000585242663582306,
      "loss": 1.9286,
      "step": 23099
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.7697005271911621,
      "learning_rate": 0.0005852413963891787,
      "loss": 2.2879,
      "step": 23100
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6087802052497864,
      "learning_rate": 0.0005852401291430196,
      "loss": 2.3157,
      "step": 23101
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6550844311714172,
      "learning_rate": 0.000585238861843829,
      "loss": 2.2067,
      "step": 23102
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.680557131767273,
      "learning_rate": 0.0005852375944916073,
      "loss": 2.4367,
      "step": 23103
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6447175145149231,
      "learning_rate": 0.0005852363270863542,
      "loss": 2.0112,
      "step": 23104
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6921098828315735,
      "learning_rate": 0.0005852350596280705,
      "loss": 2.4222,
      "step": 23105
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6995445489883423,
      "learning_rate": 0.0005852337921167563,
      "loss": 2.2565,
      "step": 23106
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.7487590909004211,
      "learning_rate": 0.0005852325245524116,
      "loss": 2.3223,
      "step": 23107
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.5983477830886841,
      "learning_rate": 0.0005852312569350368,
      "loss": 2.3062,
      "step": 23108
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.9003401398658752,
      "learning_rate": 0.0005852299892646321,
      "loss": 2.1853,
      "step": 23109
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6794763803482056,
      "learning_rate": 0.0005852287215411978,
      "loss": 2.2028,
      "step": 23110
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6799808144569397,
      "learning_rate": 0.0005852274537647341,
      "loss": 2.42,
      "step": 23111
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.701113760471344,
      "learning_rate": 0.0005852261859352413,
      "loss": 2.3812,
      "step": 23112
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6238594651222229,
      "learning_rate": 0.0005852249180527194,
      "loss": 2.3586,
      "step": 23113
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.5634920001029968,
      "learning_rate": 0.000585223650117169,
      "loss": 2.3399,
      "step": 23114
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6062717437744141,
      "learning_rate": 0.0005852223821285901,
      "loss": 2.3755,
      "step": 23115
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6382765173912048,
      "learning_rate": 0.0005852211140869829,
      "loss": 2.4219,
      "step": 23116
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6272822618484497,
      "learning_rate": 0.0005852198459923478,
      "loss": 2.2844,
      "step": 23117
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.7268388271331787,
      "learning_rate": 0.0005852185778446848,
      "loss": 2.3826,
      "step": 23118
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.694311797618866,
      "learning_rate": 0.0005852173096439944,
      "loss": 2.2215,
      "step": 23119
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6998764872550964,
      "learning_rate": 0.0005852160413902766,
      "loss": 2.4453,
      "step": 23120
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6377657651901245,
      "learning_rate": 0.0005852147730835319,
      "loss": 2.3678,
      "step": 23121
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6194554567337036,
      "learning_rate": 0.0005852135047237603,
      "loss": 2.4509,
      "step": 23122
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6788811087608337,
      "learning_rate": 0.0005852122363109622,
      "loss": 2.7706,
      "step": 23123
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.754828155040741,
      "learning_rate": 0.0005852109678451377,
      "loss": 2.4907,
      "step": 23124
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.642334520816803,
      "learning_rate": 0.000585209699326287,
      "loss": 2.4587,
      "step": 23125
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.635959804058075,
      "learning_rate": 0.0005852084307544106,
      "loss": 2.5484,
      "step": 23126
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6505984663963318,
      "learning_rate": 0.0005852071621295085,
      "loss": 2.4869,
      "step": 23127
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.5784425139427185,
      "learning_rate": 0.000585205893451581,
      "loss": 2.1464,
      "step": 23128
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.5875011086463928,
      "learning_rate": 0.0005852046247206284,
      "loss": 2.2583,
      "step": 23129
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6565141677856445,
      "learning_rate": 0.0005852033559366509,
      "loss": 2.177,
      "step": 23130
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.7913489937782288,
      "learning_rate": 0.0005852020870996486,
      "loss": 2.4081,
      "step": 23131
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6072804927825928,
      "learning_rate": 0.0005852008182096219,
      "loss": 2.2006,
      "step": 23132
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6718692779541016,
      "learning_rate": 0.000585199549266571,
      "loss": 2.3669,
      "step": 23133
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.7277945876121521,
      "learning_rate": 0.0005851982802704961,
      "loss": 2.1886,
      "step": 23134
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6354756355285645,
      "learning_rate": 0.0005851970112213975,
      "loss": 2.472,
      "step": 23135
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6981301307678223,
      "learning_rate": 0.0005851957421192753,
      "loss": 2.3965,
      "step": 23136
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.7422189116477966,
      "learning_rate": 0.0005851944729641299,
      "loss": 2.0875,
      "step": 23137
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.5835999846458435,
      "learning_rate": 0.0005851932037559615,
      "loss": 2.3099,
      "step": 23138
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.631249189376831,
      "learning_rate": 0.0005851919344947702,
      "loss": 2.183,
      "step": 23139
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6202483773231506,
      "learning_rate": 0.0005851906651805563,
      "loss": 2.513,
      "step": 23140
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6486998796463013,
      "learning_rate": 0.0005851893958133201,
      "loss": 2.6459,
      "step": 23141
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6212607622146606,
      "learning_rate": 0.0005851881263930619,
      "loss": 2.5836,
      "step": 23142
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6319687962532043,
      "learning_rate": 0.0005851868569197818,
      "loss": 2.4318,
      "step": 23143
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.5355720520019531,
      "learning_rate": 0.0005851855873934801,
      "loss": 2.3081,
      "step": 23144
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6201876997947693,
      "learning_rate": 0.000585184317814157,
      "loss": 2.597,
      "step": 23145
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.568093478679657,
      "learning_rate": 0.0005851830481818128,
      "loss": 2.3605,
      "step": 23146
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6275843977928162,
      "learning_rate": 0.0005851817784964476,
      "loss": 2.2721,
      "step": 23147
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.5937836766242981,
      "learning_rate": 0.0005851805087580618,
      "loss": 2.3124,
      "step": 23148
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6863164305686951,
      "learning_rate": 0.0005851792389666555,
      "loss": 2.2082,
      "step": 23149
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6378519535064697,
      "learning_rate": 0.0005851779691222291,
      "loss": 2.1211,
      "step": 23150
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.5824531316757202,
      "learning_rate": 0.0005851766992247825,
      "loss": 2.4149,
      "step": 23151
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6378312706947327,
      "learning_rate": 0.0005851754292743165,
      "loss": 2.2335,
      "step": 23152
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.640465259552002,
      "learning_rate": 0.0005851741592708307,
      "loss": 2.2533,
      "step": 23153
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6113501787185669,
      "learning_rate": 0.0005851728892143259,
      "loss": 2.3363,
      "step": 23154
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6664566397666931,
      "learning_rate": 0.0005851716191048019,
      "loss": 2.5345,
      "step": 23155
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6095212697982788,
      "learning_rate": 0.0005851703489422592,
      "loss": 2.3669,
      "step": 23156
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6130800843238831,
      "learning_rate": 0.0005851690787266978,
      "loss": 2.1718,
      "step": 23157
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.7297441959381104,
      "learning_rate": 0.0005851678084581182,
      "loss": 2.3984,
      "step": 23158
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6622141599655151,
      "learning_rate": 0.0005851665381365205,
      "loss": 2.2221,
      "step": 23159
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6046634316444397,
      "learning_rate": 0.0005851652677619049,
      "loss": 2.4371,
      "step": 23160
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6424777507781982,
      "learning_rate": 0.0005851639973342718,
      "loss": 2.5999,
      "step": 23161
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.5663066506385803,
      "learning_rate": 0.0005851627268536213,
      "loss": 2.3292,
      "step": 23162
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.5827298164367676,
      "learning_rate": 0.0005851614563199537,
      "loss": 2.4127,
      "step": 23163
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.5808956623077393,
      "learning_rate": 0.0005851601857332692,
      "loss": 2.1406,
      "step": 23164
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.819340705871582,
      "learning_rate": 0.000585158915093568,
      "loss": 2.3905,
      "step": 23165
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6471235156059265,
      "learning_rate": 0.0005851576444008504,
      "loss": 2.246,
      "step": 23166
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6286099553108215,
      "learning_rate": 0.0005851563736551166,
      "loss": 2.2858,
      "step": 23167
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6057042479515076,
      "learning_rate": 0.0005851551028563667,
      "loss": 2.3006,
      "step": 23168
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.5985127687454224,
      "learning_rate": 0.0005851538320046013,
      "loss": 2.2542,
      "step": 23169
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.5809201598167419,
      "learning_rate": 0.0005851525610998203,
      "loss": 2.1972,
      "step": 23170
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6306589245796204,
      "learning_rate": 0.000585151290142024,
      "loss": 2.2962,
      "step": 23171
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.5979126691818237,
      "learning_rate": 0.0005851500191312129,
      "loss": 2.3662,
      "step": 23172
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.7163227200508118,
      "learning_rate": 0.0005851487480673869,
      "loss": 2.2352,
      "step": 23173
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6192169785499573,
      "learning_rate": 0.0005851474769505464,
      "loss": 2.4682,
      "step": 23174
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.5530100464820862,
      "learning_rate": 0.0005851462057806916,
      "loss": 2.1304,
      "step": 23175
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.5901675224304199,
      "learning_rate": 0.0005851449345578226,
      "loss": 2.1024,
      "step": 23176
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6168569326400757,
      "learning_rate": 0.00058514366328194,
      "loss": 2.3542,
      "step": 23177
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.5725098848342896,
      "learning_rate": 0.0005851423919530437,
      "loss": 2.3041,
      "step": 23178
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.5524611473083496,
      "learning_rate": 0.000585141120571134,
      "loss": 2.1987,
      "step": 23179
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6286930441856384,
      "learning_rate": 0.0005851398491362113,
      "loss": 2.4531,
      "step": 23180
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6248045563697815,
      "learning_rate": 0.0005851385776482756,
      "loss": 2.2031,
      "step": 23181
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.5777131915092468,
      "learning_rate": 0.0005851373061073273,
      "loss": 2.2674,
      "step": 23182
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6033762097358704,
      "learning_rate": 0.0005851360345133667,
      "loss": 2.359,
      "step": 23183
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.5647855401039124,
      "learning_rate": 0.0005851347628663937,
      "loss": 2.3319,
      "step": 23184
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6568484306335449,
      "learning_rate": 0.0005851334911664089,
      "loss": 2.3234,
      "step": 23185
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6379219889640808,
      "learning_rate": 0.0005851322194134124,
      "loss": 2.345,
      "step": 23186
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.7033310532569885,
      "learning_rate": 0.0005851309476074044,
      "loss": 2.6101,
      "step": 23187
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6595873832702637,
      "learning_rate": 0.0005851296757483853,
      "loss": 2.5277,
      "step": 23188
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.736832320690155,
      "learning_rate": 0.0005851284038363551,
      "loss": 2.2427,
      "step": 23189
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6596762537956238,
      "learning_rate": 0.0005851271318713142,
      "loss": 2.1939,
      "step": 23190
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.687856912612915,
      "learning_rate": 0.0005851258598532627,
      "loss": 2.3288,
      "step": 23191
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.7640979886054993,
      "learning_rate": 0.000585124587782201,
      "loss": 2.2373,
      "step": 23192
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6147533059120178,
      "learning_rate": 0.0005851233156581292,
      "loss": 2.3805,
      "step": 23193
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.5991683006286621,
      "learning_rate": 0.0005851220434810477,
      "loss": 2.4055,
      "step": 23194
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.7310516238212585,
      "learning_rate": 0.0005851207712509566,
      "loss": 2.3438,
      "step": 23195
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.642780601978302,
      "learning_rate": 0.0005851194989678561,
      "loss": 2.1898,
      "step": 23196
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6418935656547546,
      "learning_rate": 0.0005851182266317464,
      "loss": 2.1768,
      "step": 23197
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6789405345916748,
      "learning_rate": 0.0005851169542426281,
      "loss": 2.3848,
      "step": 23198
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.7648030519485474,
      "learning_rate": 0.0005851156818005009,
      "loss": 2.5409,
      "step": 23199
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6205959320068359,
      "learning_rate": 0.0005851144093053655,
      "loss": 2.3548,
      "step": 23200
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6187490820884705,
      "learning_rate": 0.0005851131367572219,
      "loss": 2.5517,
      "step": 23201
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.7021358609199524,
      "learning_rate": 0.0005851118641560704,
      "loss": 2.384,
      "step": 23202
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6509243845939636,
      "learning_rate": 0.0005851105915019113,
      "loss": 2.3257,
      "step": 23203
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.7419472336769104,
      "learning_rate": 0.0005851093187947446,
      "loss": 2.2924,
      "step": 23204
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6171609163284302,
      "learning_rate": 0.0005851080460345708,
      "loss": 2.2686,
      "step": 23205
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6383768320083618,
      "learning_rate": 0.00058510677322139,
      "loss": 2.3075,
      "step": 23206
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6949058175086975,
      "learning_rate": 0.0005851055003552024,
      "loss": 2.3412,
      "step": 23207
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6365262866020203,
      "learning_rate": 0.0005851042274360084,
      "loss": 2.1505,
      "step": 23208
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.5989910364151001,
      "learning_rate": 0.0005851029544638081,
      "loss": 2.4191,
      "step": 23209
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6787495613098145,
      "learning_rate": 0.0005851016814386018,
      "loss": 2.4023,
      "step": 23210
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.7326937317848206,
      "learning_rate": 0.0005851004083603897,
      "loss": 2.5086,
      "step": 23211
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6995792984962463,
      "learning_rate": 0.0005850991352291721,
      "loss": 2.4157,
      "step": 23212
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.5984048843383789,
      "learning_rate": 0.0005850978620449491,
      "loss": 2.4226,
      "step": 23213
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6123197078704834,
      "learning_rate": 0.0005850965888077211,
      "loss": 2.0639,
      "step": 23214
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6175928115844727,
      "learning_rate": 0.0005850953155174882,
      "loss": 2.4931,
      "step": 23215
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6480545401573181,
      "learning_rate": 0.0005850940421742507,
      "loss": 2.3661,
      "step": 23216
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.7478861212730408,
      "learning_rate": 0.000585092768778009,
      "loss": 2.4371,
      "step": 23217
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6193781495094299,
      "learning_rate": 0.000585091495328763,
      "loss": 2.4333,
      "step": 23218
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6403079032897949,
      "learning_rate": 0.0005850902218265131,
      "loss": 2.5149,
      "step": 23219
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.5765488743782043,
      "learning_rate": 0.0005850889482712596,
      "loss": 2.194,
      "step": 23220
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.5898004174232483,
      "learning_rate": 0.0005850876746630027,
      "loss": 2.3342,
      "step": 23221
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.7720164060592651,
      "learning_rate": 0.0005850864010017426,
      "loss": 2.2954,
      "step": 23222
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6355376839637756,
      "learning_rate": 0.0005850851272874795,
      "loss": 2.2968,
      "step": 23223
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6104452610015869,
      "learning_rate": 0.0005850838535202138,
      "loss": 2.4272,
      "step": 23224
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6242018938064575,
      "learning_rate": 0.0005850825796999456,
      "loss": 2.1698,
      "step": 23225
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6298930048942566,
      "learning_rate": 0.0005850813058266752,
      "loss": 2.4058,
      "step": 23226
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6738190650939941,
      "learning_rate": 0.0005850800319004028,
      "loss": 2.4848,
      "step": 23227
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6470218300819397,
      "learning_rate": 0.0005850787579211285,
      "loss": 2.2042,
      "step": 23228
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6306672096252441,
      "learning_rate": 0.0005850774838888529,
      "loss": 2.3087,
      "step": 23229
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.632876455783844,
      "learning_rate": 0.0005850762098035758,
      "loss": 2.2626,
      "step": 23230
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.5891966819763184,
      "learning_rate": 0.0005850749356652977,
      "loss": 2.3944,
      "step": 23231
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6316004395484924,
      "learning_rate": 0.0005850736614740189,
      "loss": 2.3412,
      "step": 23232
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6203036308288574,
      "learning_rate": 0.0005850723872297394,
      "loss": 2.2808,
      "step": 23233
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6074853539466858,
      "learning_rate": 0.0005850711129324597,
      "loss": 2.4981,
      "step": 23234
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.5658197402954102,
      "learning_rate": 0.0005850698385821797,
      "loss": 2.2643,
      "step": 23235
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.658659815788269,
      "learning_rate": 0.0005850685641789,
      "loss": 2.4092,
      "step": 23236
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.7143689393997192,
      "learning_rate": 0.0005850672897226206,
      "loss": 2.4174,
      "step": 23237
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6168334484100342,
      "learning_rate": 0.0005850660152133418,
      "loss": 1.9336,
      "step": 23238
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.5938180088996887,
      "learning_rate": 0.000585064740651064,
      "loss": 2.1297,
      "step": 23239
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6777728199958801,
      "learning_rate": 0.0005850634660357871,
      "loss": 2.545,
      "step": 23240
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6349185705184937,
      "learning_rate": 0.0005850621913675116,
      "loss": 2.4463,
      "step": 23241
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6271650195121765,
      "learning_rate": 0.0005850609166462376,
      "loss": 2.5134,
      "step": 23242
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6355887055397034,
      "learning_rate": 0.0005850596418719655,
      "loss": 2.3778,
      "step": 23243
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6293443441390991,
      "learning_rate": 0.0005850583670446954,
      "loss": 2.5722,
      "step": 23244
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.5952062010765076,
      "learning_rate": 0.0005850570921644275,
      "loss": 2.1725,
      "step": 23245
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6191828846931458,
      "learning_rate": 0.0005850558172311622,
      "loss": 2.277,
      "step": 23246
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6241409182548523,
      "learning_rate": 0.0005850545422448996,
      "loss": 2.4475,
      "step": 23247
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.5627567172050476,
      "learning_rate": 0.0005850532672056399,
      "loss": 2.267,
      "step": 23248
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.7037215828895569,
      "learning_rate": 0.0005850519921133835,
      "loss": 2.3412,
      "step": 23249
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6618985533714294,
      "learning_rate": 0.0005850507169681305,
      "loss": 2.5931,
      "step": 23250
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6015440225601196,
      "learning_rate": 0.0005850494417698812,
      "loss": 2.6899,
      "step": 23251
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6258458495140076,
      "learning_rate": 0.0005850481665186359,
      "loss": 2.5074,
      "step": 23252
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6872637271881104,
      "learning_rate": 0.0005850468912143947,
      "loss": 2.3528,
      "step": 23253
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6000239253044128,
      "learning_rate": 0.000585045615857158,
      "loss": 2.1773,
      "step": 23254
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6269522905349731,
      "learning_rate": 0.0005850443404469258,
      "loss": 2.3611,
      "step": 23255
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6198199391365051,
      "learning_rate": 0.0005850430649836986,
      "loss": 2.1073,
      "step": 23256
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.5753613710403442,
      "learning_rate": 0.0005850417894674764,
      "loss": 2.1588,
      "step": 23257
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6024442315101624,
      "learning_rate": 0.0005850405138982597,
      "loss": 2.34,
      "step": 23258
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6470254063606262,
      "learning_rate": 0.0005850392382760484,
      "loss": 2.4178,
      "step": 23259
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6051399111747742,
      "learning_rate": 0.0005850379626008431,
      "loss": 2.2164,
      "step": 23260
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6073514223098755,
      "learning_rate": 0.0005850366868726437,
      "loss": 2.1766,
      "step": 23261
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6477763056755066,
      "learning_rate": 0.0005850354110914507,
      "loss": 2.3468,
      "step": 23262
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.5920262932777405,
      "learning_rate": 0.0005850341352572642,
      "loss": 2.3907,
      "step": 23263
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6545705795288086,
      "learning_rate": 0.0005850328593700845,
      "loss": 2.348,
      "step": 23264
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.5742001533508301,
      "learning_rate": 0.0005850315834299118,
      "loss": 2.2755,
      "step": 23265
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.615856409072876,
      "learning_rate": 0.0005850303074367464,
      "loss": 2.2419,
      "step": 23266
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6739621758460999,
      "learning_rate": 0.0005850290313905883,
      "loss": 2.107,
      "step": 23267
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.5928996801376343,
      "learning_rate": 0.0005850277552914381,
      "loss": 2.2938,
      "step": 23268
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.5837178826332092,
      "learning_rate": 0.0005850264791392958,
      "loss": 2.2773,
      "step": 23269
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.5925608277320862,
      "learning_rate": 0.0005850252029341617,
      "loss": 2.4146,
      "step": 23270
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.5978277325630188,
      "learning_rate": 0.0005850239266760361,
      "loss": 2.2383,
      "step": 23271
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6178216338157654,
      "learning_rate": 0.0005850226503649192,
      "loss": 2.3247,
      "step": 23272
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6218653321266174,
      "learning_rate": 0.000585021374000811,
      "loss": 2.3471,
      "step": 23273
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.7188295125961304,
      "learning_rate": 0.0005850200975837121,
      "loss": 2.4001,
      "step": 23274
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6277140974998474,
      "learning_rate": 0.0005850188211136226,
      "loss": 2.5295,
      "step": 23275
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6241655349731445,
      "learning_rate": 0.0005850175445905427,
      "loss": 2.3428,
      "step": 23276
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6241991519927979,
      "learning_rate": 0.0005850162680144725,
      "loss": 2.4756,
      "step": 23277
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6568514704704285,
      "learning_rate": 0.0005850149913854127,
      "loss": 2.1972,
      "step": 23278
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6309500932693481,
      "learning_rate": 0.000585013714703363,
      "loss": 2.3432,
      "step": 23279
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.5937926769256592,
      "learning_rate": 0.0005850124379683239,
      "loss": 2.2624,
      "step": 23280
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6917098760604858,
      "learning_rate": 0.0005850111611802956,
      "loss": 2.2665,
      "step": 23281
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6223908066749573,
      "learning_rate": 0.0005850098843392785,
      "loss": 2.1118,
      "step": 23282
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6409511566162109,
      "learning_rate": 0.0005850086074452725,
      "loss": 2.433,
      "step": 23283
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6476671695709229,
      "learning_rate": 0.0005850073304982782,
      "loss": 2.407,
      "step": 23284
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6778042912483215,
      "learning_rate": 0.0005850060534982954,
      "loss": 2.4349,
      "step": 23285
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.620406985282898,
      "learning_rate": 0.0005850047764453249,
      "loss": 2.3017,
      "step": 23286
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.7366907000541687,
      "learning_rate": 0.0005850034993393664,
      "loss": 2.1669,
      "step": 23287
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6714667081832886,
      "learning_rate": 0.0005850022221804205,
      "loss": 2.2976,
      "step": 23288
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6783148646354675,
      "learning_rate": 0.0005850009449684872,
      "loss": 2.4719,
      "step": 23289
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6169376373291016,
      "learning_rate": 0.0005849996677035669,
      "loss": 2.514,
      "step": 23290
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6105660796165466,
      "learning_rate": 0.0005849983903856598,
      "loss": 2.4822,
      "step": 23291
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6811347603797913,
      "learning_rate": 0.000584997113014766,
      "loss": 2.282,
      "step": 23292
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6355990767478943,
      "learning_rate": 0.000584995835590886,
      "loss": 2.3097,
      "step": 23293
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.7013354897499084,
      "learning_rate": 0.0005849945581140199,
      "loss": 2.2377,
      "step": 23294
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6554171442985535,
      "learning_rate": 0.0005849932805841679,
      "loss": 2.2944,
      "step": 23295
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6249479055404663,
      "learning_rate": 0.0005849920030013302,
      "loss": 2.2072,
      "step": 23296
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.5964764952659607,
      "learning_rate": 0.0005849907253655072,
      "loss": 2.2367,
      "step": 23297
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6689789891242981,
      "learning_rate": 0.0005849894476766989,
      "loss": 2.4462,
      "step": 23298
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.669039249420166,
      "learning_rate": 0.0005849881699349058,
      "loss": 2.1053,
      "step": 23299
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6176581382751465,
      "learning_rate": 0.0005849868921401281,
      "loss": 2.4226,
      "step": 23300
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.640884280204773,
      "learning_rate": 0.0005849856142923658,
      "loss": 2.111,
      "step": 23301
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6939715147018433,
      "learning_rate": 0.0005849843363916194,
      "loss": 2.3627,
      "step": 23302
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.576229453086853,
      "learning_rate": 0.0005849830584378889,
      "loss": 2.3269,
      "step": 23303
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6993489861488342,
      "learning_rate": 0.0005849817804311748,
      "loss": 2.428,
      "step": 23304
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.5962690711021423,
      "learning_rate": 0.0005849805023714772,
      "loss": 2.3788,
      "step": 23305
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6187693476676941,
      "learning_rate": 0.0005849792242587963,
      "loss": 2.3753,
      "step": 23306
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6095587015151978,
      "learning_rate": 0.0005849779460931324,
      "loss": 2.3605,
      "step": 23307
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.7162838578224182,
      "learning_rate": 0.0005849766678744857,
      "loss": 2.3268,
      "step": 23308
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.612311601638794,
      "learning_rate": 0.0005849753896028564,
      "loss": 2.2098,
      "step": 23309
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.5896335244178772,
      "learning_rate": 0.000584974111278245,
      "loss": 2.0481,
      "step": 23310
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6312089562416077,
      "learning_rate": 0.0005849728329006513,
      "loss": 2.3769,
      "step": 23311
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6411843299865723,
      "learning_rate": 0.000584971554470076,
      "loss": 2.4432,
      "step": 23312
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6600753664970398,
      "learning_rate": 0.0005849702759865189,
      "loss": 2.2169,
      "step": 23313
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6835828423500061,
      "learning_rate": 0.0005849689974499805,
      "loss": 2.3461,
      "step": 23314
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.7108334898948669,
      "learning_rate": 0.0005849677188604611,
      "loss": 2.3816,
      "step": 23315
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6066386699676514,
      "learning_rate": 0.0005849664402179606,
      "loss": 2.2316,
      "step": 23316
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.5885305404663086,
      "learning_rate": 0.0005849651615224796,
      "loss": 2.1593,
      "step": 23317
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.561784565448761,
      "learning_rate": 0.0005849638827740184,
      "loss": 2.4037,
      "step": 23318
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6265929937362671,
      "learning_rate": 0.0005849626039725768,
      "loss": 2.1779,
      "step": 23319
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6457975506782532,
      "learning_rate": 0.0005849613251181552,
      "loss": 2.2612,
      "step": 23320
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.5833485722541809,
      "learning_rate": 0.000584960046210754,
      "loss": 2.2366,
      "step": 23321
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6707159876823425,
      "learning_rate": 0.0005849587672503734,
      "loss": 2.4608,
      "step": 23322
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6401216983795166,
      "learning_rate": 0.0005849574882370135,
      "loss": 2.4737,
      "step": 23323
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6643351912498474,
      "learning_rate": 0.0005849562091706748,
      "loss": 2.4176,
      "step": 23324
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6816383600234985,
      "learning_rate": 0.000584954930051357,
      "loss": 2.3523,
      "step": 23325
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.5849781036376953,
      "learning_rate": 0.0005849536508790611,
      "loss": 2.3272,
      "step": 23326
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6475915908813477,
      "learning_rate": 0.0005849523716537868,
      "loss": 2.3734,
      "step": 23327
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6974050998687744,
      "learning_rate": 0.0005849510923755344,
      "loss": 2.4442,
      "step": 23328
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6138068437576294,
      "learning_rate": 0.0005849498130443042,
      "loss": 2.4925,
      "step": 23329
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6298613548278809,
      "learning_rate": 0.0005849485336600965,
      "loss": 2.2674,
      "step": 23330
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.5954158902168274,
      "learning_rate": 0.0005849472542229115,
      "loss": 2.4047,
      "step": 23331
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6298312544822693,
      "learning_rate": 0.0005849459747327495,
      "loss": 2.3352,
      "step": 23332
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6342455148696899,
      "learning_rate": 0.0005849446951896106,
      "loss": 2.2695,
      "step": 23333
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6438506841659546,
      "learning_rate": 0.000584943415593495,
      "loss": 2.1097,
      "step": 23334
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6552618741989136,
      "learning_rate": 0.0005849421359444032,
      "loss": 2.1782,
      "step": 23335
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6410648226737976,
      "learning_rate": 0.0005849408562423352,
      "loss": 2.4131,
      "step": 23336
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6713624596595764,
      "learning_rate": 0.0005849395764872913,
      "loss": 2.4492,
      "step": 23337
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.5583301782608032,
      "learning_rate": 0.0005849382966792718,
      "loss": 2.2628,
      "step": 23338
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.5851129293441772,
      "learning_rate": 0.0005849370168182769,
      "loss": 2.3878,
      "step": 23339
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6146193742752075,
      "learning_rate": 0.0005849357369043067,
      "loss": 2.5335,
      "step": 23340
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.7663589715957642,
      "learning_rate": 0.0005849344569373617,
      "loss": 2.2752,
      "step": 23341
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6111132502555847,
      "learning_rate": 0.000584933176917442,
      "loss": 2.2772,
      "step": 23342
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.5655980706214905,
      "learning_rate": 0.0005849318968445478,
      "loss": 2.1388,
      "step": 23343
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6183517575263977,
      "learning_rate": 0.0005849306167186794,
      "loss": 2.1883,
      "step": 23344
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6514652967453003,
      "learning_rate": 0.000584929336539837,
      "loss": 2.4881,
      "step": 23345
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6282050609588623,
      "learning_rate": 0.0005849280563080209,
      "loss": 2.3841,
      "step": 23346
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6232648491859436,
      "learning_rate": 0.0005849267760232313,
      "loss": 2.3553,
      "step": 23347
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6703262329101562,
      "learning_rate": 0.0005849254956854684,
      "loss": 2.2971,
      "step": 23348
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.5911521911621094,
      "learning_rate": 0.0005849242152947325,
      "loss": 2.445,
      "step": 23349
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6792001128196716,
      "learning_rate": 0.0005849229348510237,
      "loss": 2.1582,
      "step": 23350
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.5894098877906799,
      "learning_rate": 0.0005849216543543425,
      "loss": 2.281,
      "step": 23351
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6128988862037659,
      "learning_rate": 0.000584920373804689,
      "loss": 2.2894,
      "step": 23352
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.7176359295845032,
      "learning_rate": 0.0005849190932020633,
      "loss": 2.4562,
      "step": 23353
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6144500374794006,
      "learning_rate": 0.0005849178125464658,
      "loss": 2.302,
      "step": 23354
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6148587465286255,
      "learning_rate": 0.0005849165318378966,
      "loss": 2.2914,
      "step": 23355
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.7238908410072327,
      "learning_rate": 0.0005849152510763562,
      "loss": 2.4107,
      "step": 23356
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6709657907485962,
      "learning_rate": 0.0005849139702618447,
      "loss": 2.3671,
      "step": 23357
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.7392452955245972,
      "learning_rate": 0.0005849126893943623,
      "loss": 2.2829,
      "step": 23358
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.7151591181755066,
      "learning_rate": 0.0005849114084739091,
      "loss": 2.3899,
      "step": 23359
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6079453825950623,
      "learning_rate": 0.0005849101275004855,
      "loss": 2.2447,
      "step": 23360
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.8373441100120544,
      "learning_rate": 0.0005849088464740919,
      "loss": 2.2862,
      "step": 23361
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.5615222454071045,
      "learning_rate": 0.0005849075653947283,
      "loss": 2.417,
      "step": 23362
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6370579600334167,
      "learning_rate": 0.000584906284262395,
      "loss": 2.6236,
      "step": 23363
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.5971802473068237,
      "learning_rate": 0.0005849050030770922,
      "loss": 2.3223,
      "step": 23364
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6613466143608093,
      "learning_rate": 0.0005849037218388202,
      "loss": 2.2949,
      "step": 23365
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6278714537620544,
      "learning_rate": 0.0005849024405475792,
      "loss": 2.2598,
      "step": 23366
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6686866879463196,
      "learning_rate": 0.0005849011592033695,
      "loss": 2.2131,
      "step": 23367
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.8122358322143555,
      "learning_rate": 0.0005848998778061914,
      "loss": 2.4126,
      "step": 23368
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6660515666007996,
      "learning_rate": 0.0005848985963560448,
      "loss": 2.4366,
      "step": 23369
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6601291298866272,
      "learning_rate": 0.0005848973148529304,
      "loss": 2.7676,
      "step": 23370
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6565638780593872,
      "learning_rate": 0.000584896033296848,
      "loss": 2.2271,
      "step": 23371
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6966099739074707,
      "learning_rate": 0.0005848947516877982,
      "loss": 2.3238,
      "step": 23372
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6773216128349304,
      "learning_rate": 0.0005848934700257811,
      "loss": 2.4081,
      "step": 23373
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.59584641456604,
      "learning_rate": 0.0005848921883107968,
      "loss": 2.2681,
      "step": 23374
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6440677046775818,
      "learning_rate": 0.0005848909065428457,
      "loss": 2.3678,
      "step": 23375
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.723645031452179,
      "learning_rate": 0.0005848896247219282,
      "loss": 2.4588,
      "step": 23376
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.7948747873306274,
      "learning_rate": 0.000584888342848044,
      "loss": 2.522,
      "step": 23377
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.8028474450111389,
      "learning_rate": 0.0005848870609211938,
      "loss": 2.292,
      "step": 23378
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.7043032646179199,
      "learning_rate": 0.0005848857789413779,
      "loss": 2.1661,
      "step": 23379
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6083268523216248,
      "learning_rate": 0.0005848844969085962,
      "loss": 2.3235,
      "step": 23380
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.8094050288200378,
      "learning_rate": 0.0005848832148228492,
      "loss": 2.2538,
      "step": 23381
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.8956036567687988,
      "learning_rate": 0.0005848819326841368,
      "loss": 2.507,
      "step": 23382
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6214161515235901,
      "learning_rate": 0.0005848806504924598,
      "loss": 2.0886,
      "step": 23383
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.8128135800361633,
      "learning_rate": 0.0005848793682478178,
      "loss": 2.5198,
      "step": 23384
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.0495942831039429,
      "learning_rate": 0.0005848780859502116,
      "loss": 2.2372,
      "step": 23385
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6328269243240356,
      "learning_rate": 0.0005848768035996411,
      "loss": 2.263,
      "step": 23386
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.5967017412185669,
      "learning_rate": 0.0005848755211961066,
      "loss": 2.3315,
      "step": 23387
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.8478493094444275,
      "learning_rate": 0.0005848742387396084,
      "loss": 2.3295,
      "step": 23388
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.7565146088600159,
      "learning_rate": 0.0005848729562301466,
      "loss": 2.2394,
      "step": 23389
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6311084628105164,
      "learning_rate": 0.0005848716736677216,
      "loss": 2.4281,
      "step": 23390
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.5701872110366821,
      "learning_rate": 0.0005848703910523336,
      "loss": 2.1497,
      "step": 23391
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6616604328155518,
      "learning_rate": 0.0005848691083839828,
      "loss": 2.7431,
      "step": 23392
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.64290851354599,
      "learning_rate": 0.0005848678256626694,
      "loss": 2.0988,
      "step": 23393
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6573613286018372,
      "learning_rate": 0.0005848665428883937,
      "loss": 2.2395,
      "step": 23394
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.5916516184806824,
      "learning_rate": 0.000584865260061156,
      "loss": 2.4493,
      "step": 23395
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6635118722915649,
      "learning_rate": 0.0005848639771809564,
      "loss": 2.5906,
      "step": 23396
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6369277238845825,
      "learning_rate": 0.0005848626942477953,
      "loss": 2.1987,
      "step": 23397
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6188475489616394,
      "learning_rate": 0.0005848614112616728,
      "loss": 2.1962,
      "step": 23398
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6274181604385376,
      "learning_rate": 0.0005848601282225892,
      "loss": 2.5279,
      "step": 23399
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.663394033908844,
      "learning_rate": 0.0005848588451305447,
      "loss": 2.3516,
      "step": 23400
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6393255591392517,
      "learning_rate": 0.0005848575619855397,
      "loss": 2.3403,
      "step": 23401
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6868915557861328,
      "learning_rate": 0.0005848562787875741,
      "loss": 2.1101,
      "step": 23402
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6267874240875244,
      "learning_rate": 0.0005848549955366484,
      "loss": 2.4621,
      "step": 23403
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6112529635429382,
      "learning_rate": 0.0005848537122327629,
      "loss": 2.2211,
      "step": 23404
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6669716238975525,
      "learning_rate": 0.0005848524288759176,
      "loss": 2.2512,
      "step": 23405
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6294412612915039,
      "learning_rate": 0.0005848511454661128,
      "loss": 2.2365,
      "step": 23406
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6244426965713501,
      "learning_rate": 0.0005848498620033489,
      "loss": 2.5464,
      "step": 23407
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6320454478263855,
      "learning_rate": 0.0005848485784876261,
      "loss": 2.3489,
      "step": 23408
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6574872732162476,
      "learning_rate": 0.0005848472949189444,
      "loss": 2.2215,
      "step": 23409
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6338102221488953,
      "learning_rate": 0.0005848460112973043,
      "loss": 2.5068,
      "step": 23410
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6663219332695007,
      "learning_rate": 0.000584844727622706,
      "loss": 2.5709,
      "step": 23411
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6764634251594543,
      "learning_rate": 0.0005848434438951496,
      "loss": 2.1999,
      "step": 23412
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6022545695304871,
      "learning_rate": 0.0005848421601146354,
      "loss": 2.2743,
      "step": 23413
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6064913272857666,
      "learning_rate": 0.0005848408762811638,
      "loss": 2.4461,
      "step": 23414
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6391739845275879,
      "learning_rate": 0.0005848395923947348,
      "loss": 2.4481,
      "step": 23415
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.5969458818435669,
      "learning_rate": 0.0005848383084553488,
      "loss": 2.4104,
      "step": 23416
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6670927405357361,
      "learning_rate": 0.000584837024463006,
      "loss": 2.3854,
      "step": 23417
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6101980805397034,
      "learning_rate": 0.0005848357404177064,
      "loss": 2.3229,
      "step": 23418
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.7011899352073669,
      "learning_rate": 0.0005848344563194508,
      "loss": 2.0893,
      "step": 23419
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6293044686317444,
      "learning_rate": 0.0005848331721682389,
      "loss": 2.504,
      "step": 23420
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.5983445048332214,
      "learning_rate": 0.0005848318879640712,
      "loss": 2.1259,
      "step": 23421
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6878484487533569,
      "learning_rate": 0.0005848306037069478,
      "loss": 2.4744,
      "step": 23422
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6869450211524963,
      "learning_rate": 0.000584829319396869,
      "loss": 2.3325,
      "step": 23423
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6268351078033447,
      "learning_rate": 0.0005848280350338351,
      "loss": 2.2578,
      "step": 23424
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6929752826690674,
      "learning_rate": 0.0005848267506178463,
      "loss": 2.2014,
      "step": 23425
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.659666895866394,
      "learning_rate": 0.0005848254661489029,
      "loss": 2.2866,
      "step": 23426
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6558037996292114,
      "learning_rate": 0.000584824181627005,
      "loss": 2.282,
      "step": 23427
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.7231647372245789,
      "learning_rate": 0.0005848228970521529,
      "loss": 2.2215,
      "step": 23428
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6130163073539734,
      "learning_rate": 0.0005848216124243468,
      "loss": 2.5985,
      "step": 23429
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.5720332264900208,
      "learning_rate": 0.000584820327743587,
      "loss": 2.3028,
      "step": 23430
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6315734386444092,
      "learning_rate": 0.0005848190430098739,
      "loss": 2.0567,
      "step": 23431
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.8423391580581665,
      "learning_rate": 0.0005848177582232074,
      "loss": 2.3954,
      "step": 23432
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.5925186276435852,
      "learning_rate": 0.0005848164733835879,
      "loss": 2.0928,
      "step": 23433
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.7516863346099854,
      "learning_rate": 0.0005848151884910157,
      "loss": 2.356,
      "step": 23434
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6179414391517639,
      "learning_rate": 0.0005848139035454909,
      "loss": 2.1519,
      "step": 23435
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.705367386341095,
      "learning_rate": 0.0005848126185470139,
      "loss": 2.425,
      "step": 23436
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6990695595741272,
      "learning_rate": 0.0005848113334955848,
      "loss": 2.1541,
      "step": 23437
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6825123429298401,
      "learning_rate": 0.000584810048391204,
      "loss": 2.4948,
      "step": 23438
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.8207573890686035,
      "learning_rate": 0.0005848087632338715,
      "loss": 2.4172,
      "step": 23439
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6756792664527893,
      "learning_rate": 0.0005848074780235877,
      "loss": 2.2111,
      "step": 23440
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6243700385093689,
      "learning_rate": 0.0005848061927603529,
      "loss": 2.1265,
      "step": 23441
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.699504017829895,
      "learning_rate": 0.0005848049074441672,
      "loss": 2.1832,
      "step": 23442
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.7515805959701538,
      "learning_rate": 0.0005848036220750309,
      "loss": 2.161,
      "step": 23443
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.5969823598861694,
      "learning_rate": 0.0005848023366529442,
      "loss": 2.3849,
      "step": 23444
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6748111248016357,
      "learning_rate": 0.0005848010511779074,
      "loss": 2.3548,
      "step": 23445
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.7280203700065613,
      "learning_rate": 0.0005847997656499207,
      "loss": 2.5044,
      "step": 23446
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.7421847581863403,
      "learning_rate": 0.0005847984800689843,
      "loss": 2.4624,
      "step": 23447
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.7007922530174255,
      "learning_rate": 0.0005847971944350986,
      "loss": 2.105,
      "step": 23448
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.716566264629364,
      "learning_rate": 0.0005847959087482637,
      "loss": 2.3378,
      "step": 23449
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.7332574725151062,
      "learning_rate": 0.0005847946230084799,
      "loss": 2.2838,
      "step": 23450
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.7192285060882568,
      "learning_rate": 0.0005847933372157472,
      "loss": 2.2838,
      "step": 23451
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6764348149299622,
      "learning_rate": 0.0005847920513700663,
      "loss": 2.3797,
      "step": 23452
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.5875955820083618,
      "learning_rate": 0.000584790765471437,
      "loss": 2.2723,
      "step": 23453
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.5883382558822632,
      "learning_rate": 0.0005847894795198597,
      "loss": 2.3854,
      "step": 23454
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6077105402946472,
      "learning_rate": 0.0005847881935153348,
      "loss": 2.1434,
      "step": 23455
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6572098135948181,
      "learning_rate": 0.0005847869074578623,
      "loss": 2.2658,
      "step": 23456
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6783831715583801,
      "learning_rate": 0.0005847856213474427,
      "loss": 2.3126,
      "step": 23457
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6364059448242188,
      "learning_rate": 0.0005847843351840759,
      "loss": 2.224,
      "step": 23458
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6293011903762817,
      "learning_rate": 0.0005847830489677623,
      "loss": 2.3832,
      "step": 23459
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.5877147316932678,
      "learning_rate": 0.0005847817626985023,
      "loss": 2.2477,
      "step": 23460
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6817450523376465,
      "learning_rate": 0.0005847804763762959,
      "loss": 2.3798,
      "step": 23461
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.599624514579773,
      "learning_rate": 0.0005847791900011434,
      "loss": 2.23,
      "step": 23462
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6108149886131287,
      "learning_rate": 0.0005847779035730452,
      "loss": 2.4339,
      "step": 23463
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.7063953280448914,
      "learning_rate": 0.0005847766170920013,
      "loss": 2.3078,
      "step": 23464
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6610434651374817,
      "learning_rate": 0.000584775330558012,
      "loss": 2.2147,
      "step": 23465
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.7470234632492065,
      "learning_rate": 0.0005847740439710778,
      "loss": 2.2449,
      "step": 23466
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6089985966682434,
      "learning_rate": 0.0005847727573311985,
      "loss": 2.4022,
      "step": 23467
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.7422601580619812,
      "learning_rate": 0.0005847714706383747,
      "loss": 2.2657,
      "step": 23468
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6622098088264465,
      "learning_rate": 0.0005847701838926065,
      "loss": 2.1864,
      "step": 23469
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6026357412338257,
      "learning_rate": 0.0005847688970938942,
      "loss": 2.4503,
      "step": 23470
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6046375632286072,
      "learning_rate": 0.0005847676102422379,
      "loss": 2.6413,
      "step": 23471
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.9062739014625549,
      "learning_rate": 0.0005847663233376379,
      "loss": 2.37,
      "step": 23472
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6445111632347107,
      "learning_rate": 0.0005847650363800946,
      "loss": 1.9733,
      "step": 23473
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.7299977540969849,
      "learning_rate": 0.0005847637493696079,
      "loss": 2.5198,
      "step": 23474
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.750259518623352,
      "learning_rate": 0.0005847624623061784,
      "loss": 2.2609,
      "step": 23475
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.7278708815574646,
      "learning_rate": 0.0005847611751898062,
      "loss": 2.4007,
      "step": 23476
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.693601667881012,
      "learning_rate": 0.0005847598880204915,
      "loss": 2.3242,
      "step": 23477
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6755395531654358,
      "learning_rate": 0.0005847586007982344,
      "loss": 2.3877,
      "step": 23478
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.7298420667648315,
      "learning_rate": 0.0005847573135230354,
      "loss": 2.4445,
      "step": 23479
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.7036729454994202,
      "learning_rate": 0.0005847560261948947,
      "loss": 2.1542,
      "step": 23480
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6060724854469299,
      "learning_rate": 0.0005847547388138125,
      "loss": 2.3311,
      "step": 23481
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.7337963581085205,
      "learning_rate": 0.0005847534513797889,
      "loss": 2.3817,
      "step": 23482
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.718593180179596,
      "learning_rate": 0.0005847521638928243,
      "loss": 2.3236,
      "step": 23483
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6315837502479553,
      "learning_rate": 0.000584750876352919,
      "loss": 2.207,
      "step": 23484
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.5825697183609009,
      "learning_rate": 0.000584749588760073,
      "loss": 2.1555,
      "step": 23485
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6648110747337341,
      "learning_rate": 0.0005847483011142868,
      "loss": 2.2999,
      "step": 23486
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.715173602104187,
      "learning_rate": 0.0005847470134155605,
      "loss": 2.1328,
      "step": 23487
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.7044844031333923,
      "learning_rate": 0.0005847457256638942,
      "loss": 2.2989,
      "step": 23488
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6368972659111023,
      "learning_rate": 0.0005847444378592884,
      "loss": 2.3281,
      "step": 23489
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.5966011881828308,
      "learning_rate": 0.0005847431500017433,
      "loss": 2.342,
      "step": 23490
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.5910224914550781,
      "learning_rate": 0.0005847418620912589,
      "loss": 2.2949,
      "step": 23491
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6970888376235962,
      "learning_rate": 0.0005847405741278358,
      "loss": 2.3367,
      "step": 23492
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6669664978981018,
      "learning_rate": 0.000584739286111474,
      "loss": 2.1723,
      "step": 23493
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6564658880233765,
      "learning_rate": 0.0005847379980421737,
      "loss": 2.4424,
      "step": 23494
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6532771587371826,
      "learning_rate": 0.0005847367099199354,
      "loss": 2.466,
      "step": 23495
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6629942059516907,
      "learning_rate": 0.000584735421744759,
      "loss": 2.2288,
      "step": 23496
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6355014443397522,
      "learning_rate": 0.000584734133516645,
      "loss": 2.1062,
      "step": 23497
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.5729548931121826,
      "learning_rate": 0.0005847328452355935,
      "loss": 2.4072,
      "step": 23498
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6229448914527893,
      "learning_rate": 0.0005847315569016048,
      "loss": 2.3015,
      "step": 23499
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6213172674179077,
      "learning_rate": 0.0005847302685146791,
      "loss": 2.4447,
      "step": 23500
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.5942946076393127,
      "learning_rate": 0.0005847289800748168,
      "loss": 2.082,
      "step": 23501
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.5817359089851379,
      "learning_rate": 0.0005847276915820178,
      "loss": 2.2752,
      "step": 23502
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6009392142295837,
      "learning_rate": 0.0005847264030362828,
      "loss": 2.3408,
      "step": 23503
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6837539076805115,
      "learning_rate": 0.0005847251144376115,
      "loss": 2.3496,
      "step": 23504
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.681849479675293,
      "learning_rate": 0.0005847238257860047,
      "loss": 2.3876,
      "step": 23505
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6533017158508301,
      "learning_rate": 0.0005847225370814623,
      "loss": 2.3325,
      "step": 23506
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6845089793205261,
      "learning_rate": 0.0005847212483239845,
      "loss": 2.4253,
      "step": 23507
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.7279295921325684,
      "learning_rate": 0.0005847199595135717,
      "loss": 2.472,
      "step": 23508
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.7497146129608154,
      "learning_rate": 0.0005847186706502241,
      "loss": 2.4535,
      "step": 23509
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.5948310494422913,
      "learning_rate": 0.000584717381733942,
      "loss": 2.2808,
      "step": 23510
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6671271324157715,
      "learning_rate": 0.0005847160927647254,
      "loss": 2.1794,
      "step": 23511
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6487931609153748,
      "learning_rate": 0.0005847148037425747,
      "loss": 2.5629,
      "step": 23512
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6822675466537476,
      "learning_rate": 0.0005847135146674903,
      "loss": 2.2059,
      "step": 23513
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.673319399356842,
      "learning_rate": 0.0005847122255394721,
      "loss": 2.2686,
      "step": 23514
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.5986342430114746,
      "learning_rate": 0.0005847109363585207,
      "loss": 2.3599,
      "step": 23515
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.5859841108322144,
      "learning_rate": 0.0005847096471246361,
      "loss": 2.3581,
      "step": 23516
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.5722505450248718,
      "learning_rate": 0.0005847083578378186,
      "loss": 2.5418,
      "step": 23517
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6294001936912537,
      "learning_rate": 0.0005847070684980684,
      "loss": 2.4855,
      "step": 23518
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6586548089981079,
      "learning_rate": 0.0005847057791053859,
      "loss": 2.3346,
      "step": 23519
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6670079231262207,
      "learning_rate": 0.0005847044896597711,
      "loss": 2.2566,
      "step": 23520
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.7125144004821777,
      "learning_rate": 0.0005847032001612244,
      "loss": 2.2617,
      "step": 23521
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6663742661476135,
      "learning_rate": 0.000584701910609746,
      "loss": 2.5296,
      "step": 23522
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6763395667076111,
      "learning_rate": 0.0005847006210053361,
      "loss": 2.1829,
      "step": 23523
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6347346305847168,
      "learning_rate": 0.0005846993313479951,
      "loss": 2.3008,
      "step": 23524
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.690757691860199,
      "learning_rate": 0.000584698041637723,
      "loss": 2.6126,
      "step": 23525
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6589935421943665,
      "learning_rate": 0.0005846967518745202,
      "loss": 2.4081,
      "step": 23526
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6122561693191528,
      "learning_rate": 0.0005846954620583869,
      "loss": 2.4506,
      "step": 23527
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6338807344436646,
      "learning_rate": 0.0005846941721893233,
      "loss": 2.2287,
      "step": 23528
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6235470771789551,
      "learning_rate": 0.0005846928822673297,
      "loss": 2.4227,
      "step": 23529
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.5704609155654907,
      "learning_rate": 0.0005846915922924063,
      "loss": 2.351,
      "step": 23530
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6974005103111267,
      "learning_rate": 0.0005846903022645535,
      "loss": 2.3864,
      "step": 23531
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.5440914034843445,
      "learning_rate": 0.0005846890121837712,
      "loss": 2.3396,
      "step": 23532
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6028609871864319,
      "learning_rate": 0.00058468772205006,
      "loss": 2.3506,
      "step": 23533
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.5887333154678345,
      "learning_rate": 0.0005846864318634199,
      "loss": 2.5593,
      "step": 23534
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6493231058120728,
      "learning_rate": 0.0005846851416238513,
      "loss": 2.1922,
      "step": 23535
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6074606776237488,
      "learning_rate": 0.0005846838513313543,
      "loss": 2.4237,
      "step": 23536
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.5899000763893127,
      "learning_rate": 0.0005846825609859291,
      "loss": 2.3239,
      "step": 23537
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6563500761985779,
      "learning_rate": 0.0005846812705875761,
      "loss": 2.3529,
      "step": 23538
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6026456952095032,
      "learning_rate": 0.0005846799801362955,
      "loss": 2.1063,
      "step": 23539
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.5971706509590149,
      "learning_rate": 0.0005846786896320876,
      "loss": 2.165,
      "step": 23540
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6094571948051453,
      "learning_rate": 0.0005846773990749525,
      "loss": 2.4755,
      "step": 23541
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.5822012424468994,
      "learning_rate": 0.0005846761084648904,
      "loss": 2.3559,
      "step": 23542
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.652001142501831,
      "learning_rate": 0.0005846748178019018,
      "loss": 2.4699,
      "step": 23543
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6254344582557678,
      "learning_rate": 0.0005846735270859867,
      "loss": 2.5134,
      "step": 23544
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6362888813018799,
      "learning_rate": 0.0005846722363171454,
      "loss": 2.2924,
      "step": 23545
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6860836148262024,
      "learning_rate": 0.0005846709454953781,
      "loss": 2.4861,
      "step": 23546
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.5936338901519775,
      "learning_rate": 0.0005846696546206853,
      "loss": 2.3199,
      "step": 23547
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.612147331237793,
      "learning_rate": 0.0005846683636930669,
      "loss": 2.24,
      "step": 23548
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6300418972969055,
      "learning_rate": 0.0005846670727125232,
      "loss": 2.2489,
      "step": 23549
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.5895463228225708,
      "learning_rate": 0.0005846657816790547,
      "loss": 2.2149,
      "step": 23550
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6028521656990051,
      "learning_rate": 0.0005846644905926613,
      "loss": 2.3342,
      "step": 23551
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.5741589069366455,
      "learning_rate": 0.0005846631994533434,
      "loss": 2.1138,
      "step": 23552
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6224207878112793,
      "learning_rate": 0.0005846619082611013,
      "loss": 2.1556,
      "step": 23553
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6317179799079895,
      "learning_rate": 0.0005846606170159352,
      "loss": 2.4267,
      "step": 23554
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.5784379243850708,
      "learning_rate": 0.0005846593257178452,
      "loss": 2.0516,
      "step": 23555
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6253319382667542,
      "learning_rate": 0.0005846580343668318,
      "loss": 2.2256,
      "step": 23556
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6687678098678589,
      "learning_rate": 0.0005846567429628951,
      "loss": 2.3158,
      "step": 23557
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.7257391214370728,
      "learning_rate": 0.0005846554515060353,
      "loss": 2.5099,
      "step": 23558
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6208750605583191,
      "learning_rate": 0.0005846541599962526,
      "loss": 2.2104,
      "step": 23559
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.626586377620697,
      "learning_rate": 0.0005846528684335474,
      "loss": 2.3283,
      "step": 23560
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.7548537254333496,
      "learning_rate": 0.0005846515768179198,
      "loss": 2.5709,
      "step": 23561
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6296597719192505,
      "learning_rate": 0.0005846502851493702,
      "loss": 2.3856,
      "step": 23562
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.626747727394104,
      "learning_rate": 0.0005846489934278985,
      "loss": 2.2296,
      "step": 23563
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6227150559425354,
      "learning_rate": 0.0005846477016535055,
      "loss": 2.2376,
      "step": 23564
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6399815082550049,
      "learning_rate": 0.000584646409826191,
      "loss": 2.5073,
      "step": 23565
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.5986676216125488,
      "learning_rate": 0.0005846451179459553,
      "loss": 2.2725,
      "step": 23566
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6209380626678467,
      "learning_rate": 0.0005846438260127988,
      "loss": 2.3397,
      "step": 23567
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6508453488349915,
      "learning_rate": 0.0005846425340267215,
      "loss": 2.3651,
      "step": 23568
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6145966649055481,
      "learning_rate": 0.0005846412419877238,
      "loss": 2.4212,
      "step": 23569
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.7627849578857422,
      "learning_rate": 0.0005846399498958061,
      "loss": 2.3406,
      "step": 23570
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.661637544631958,
      "learning_rate": 0.0005846386577509682,
      "loss": 2.3778,
      "step": 23571
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6530712246894836,
      "learning_rate": 0.0005846373655532108,
      "loss": 2.5016,
      "step": 23572
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6800968050956726,
      "learning_rate": 0.0005846360733025339,
      "loss": 2.2881,
      "step": 23573
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.7048726081848145,
      "learning_rate": 0.0005846347809989376,
      "loss": 2.2323,
      "step": 23574
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6955576539039612,
      "learning_rate": 0.0005846334886424226,
      "loss": 2.3603,
      "step": 23575
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6386945843696594,
      "learning_rate": 0.0005846321962329887,
      "loss": 2.4252,
      "step": 23576
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6589611768722534,
      "learning_rate": 0.0005846309037706364,
      "loss": 2.5154,
      "step": 23577
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6068547964096069,
      "learning_rate": 0.0005846296112553658,
      "loss": 2.4155,
      "step": 23578
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6556171774864197,
      "learning_rate": 0.0005846283186871771,
      "loss": 2.4366,
      "step": 23579
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6185547113418579,
      "learning_rate": 0.0005846270260660707,
      "loss": 2.1473,
      "step": 23580
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6197511553764343,
      "learning_rate": 0.0005846257333920468,
      "loss": 2.1865,
      "step": 23581
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6538901925086975,
      "learning_rate": 0.0005846244406651056,
      "loss": 2.1636,
      "step": 23582
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.7042112350463867,
      "learning_rate": 0.0005846231478852472,
      "loss": 2.3015,
      "step": 23583
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6314213275909424,
      "learning_rate": 0.0005846218550524721,
      "loss": 2.3726,
      "step": 23584
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6809605956077576,
      "learning_rate": 0.0005846205621667804,
      "loss": 2.1658,
      "step": 23585
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6274712085723877,
      "learning_rate": 0.0005846192692281723,
      "loss": 2.601,
      "step": 23586
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.637728750705719,
      "learning_rate": 0.0005846179762366482,
      "loss": 2.4868,
      "step": 23587
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6158468723297119,
      "learning_rate": 0.0005846166831922083,
      "loss": 2.4002,
      "step": 23588
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.693297266960144,
      "learning_rate": 0.0005846153900948527,
      "loss": 2.2912,
      "step": 23589
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6280844807624817,
      "learning_rate": 0.0005846140969445817,
      "loss": 2.2889,
      "step": 23590
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6243276596069336,
      "learning_rate": 0.0005846128037413956,
      "loss": 2.4225,
      "step": 23591
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.636562168598175,
      "learning_rate": 0.0005846115104852946,
      "loss": 2.5238,
      "step": 23592
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6731379628181458,
      "learning_rate": 0.000584610217176279,
      "loss": 2.2506,
      "step": 23593
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.5910216569900513,
      "learning_rate": 0.0005846089238143489,
      "loss": 2.3434,
      "step": 23594
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6412050724029541,
      "learning_rate": 0.0005846076303995047,
      "loss": 2.4799,
      "step": 23595
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.731857419013977,
      "learning_rate": 0.0005846063369317465,
      "loss": 2.0898,
      "step": 23596
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.7249108552932739,
      "learning_rate": 0.0005846050434110747,
      "loss": 2.3574,
      "step": 23597
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.706792414188385,
      "learning_rate": 0.0005846037498374894,
      "loss": 2.4574,
      "step": 23598
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.9957842826843262,
      "learning_rate": 0.0005846024562109909,
      "loss": 2.425,
      "step": 23599
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.7519511580467224,
      "learning_rate": 0.0005846011625315794,
      "loss": 2.4205,
      "step": 23600
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6517036557197571,
      "learning_rate": 0.0005845998687992552,
      "loss": 2.2713,
      "step": 23601
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.7367932200431824,
      "learning_rate": 0.0005845985750140185,
      "loss": 2.1141,
      "step": 23602
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.7191349864006042,
      "learning_rate": 0.0005845972811758695,
      "loss": 2.3593,
      "step": 23603
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.7038015723228455,
      "learning_rate": 0.0005845959872848086,
      "loss": 2.6338,
      "step": 23604
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6990468502044678,
      "learning_rate": 0.0005845946933408359,
      "loss": 2.194,
      "step": 23605
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.7501401901245117,
      "learning_rate": 0.0005845933993439517,
      "loss": 2.4507,
      "step": 23606
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6224359273910522,
      "learning_rate": 0.0005845921052941561,
      "loss": 2.2786,
      "step": 23607
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.640254020690918,
      "learning_rate": 0.0005845908111914496,
      "loss": 2.6872,
      "step": 23608
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.7191490530967712,
      "learning_rate": 0.0005845895170358322,
      "loss": 2.4255,
      "step": 23609
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6431466937065125,
      "learning_rate": 0.0005845882228273043,
      "loss": 2.1652,
      "step": 23610
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6403827667236328,
      "learning_rate": 0.000584586928565866,
      "loss": 2.2113,
      "step": 23611
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6957761645317078,
      "learning_rate": 0.0005845856342515176,
      "loss": 2.4955,
      "step": 23612
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.7466236352920532,
      "learning_rate": 0.0005845843398842595,
      "loss": 2.4702,
      "step": 23613
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6271187663078308,
      "learning_rate": 0.0005845830454640916,
      "loss": 2.4111,
      "step": 23614
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.5957887172698975,
      "learning_rate": 0.0005845817509910145,
      "loss": 2.4451,
      "step": 23615
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6678964495658875,
      "learning_rate": 0.0005845804564650283,
      "loss": 2.2054,
      "step": 23616
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6774624586105347,
      "learning_rate": 0.0005845791618861332,
      "loss": 2.2048,
      "step": 23617
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.5868739485740662,
      "learning_rate": 0.0005845778672543294,
      "loss": 2.266,
      "step": 23618
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6362584829330444,
      "learning_rate": 0.0005845765725696172,
      "loss": 2.373,
      "step": 23619
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6488379836082458,
      "learning_rate": 0.0005845752778319969,
      "loss": 2.2555,
      "step": 23620
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.829795241355896,
      "learning_rate": 0.0005845739830414687,
      "loss": 2.4415,
      "step": 23621
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.623379647731781,
      "learning_rate": 0.0005845726881980327,
      "loss": 2.2933,
      "step": 23622
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.5988222360610962,
      "learning_rate": 0.0005845713933016894,
      "loss": 2.4063,
      "step": 23623
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6144587397575378,
      "learning_rate": 0.0005845700983524388,
      "loss": 2.2391,
      "step": 23624
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.8549879789352417,
      "learning_rate": 0.0005845688033502813,
      "loss": 2.2012,
      "step": 23625
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6662706732749939,
      "learning_rate": 0.000584567508295217,
      "loss": 2.1527,
      "step": 23626
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.7267071008682251,
      "learning_rate": 0.0005845662131872463,
      "loss": 2.4703,
      "step": 23627
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6879901885986328,
      "learning_rate": 0.0005845649180263694,
      "loss": 2.2803,
      "step": 23628
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6454087495803833,
      "learning_rate": 0.0005845636228125866,
      "loss": 2.4063,
      "step": 23629
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.5834619998931885,
      "learning_rate": 0.0005845623275458979,
      "loss": 2.1304,
      "step": 23630
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.670443058013916,
      "learning_rate": 0.0005845610322263037,
      "loss": 2.5766,
      "step": 23631
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6383382081985474,
      "learning_rate": 0.0005845597368538043,
      "loss": 2.4921,
      "step": 23632
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.7418515086174011,
      "learning_rate": 0.0005845584414283999,
      "loss": 2.513,
      "step": 23633
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.5738123059272766,
      "learning_rate": 0.0005845571459500905,
      "loss": 2.4749,
      "step": 23634
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.7744625210762024,
      "learning_rate": 0.0005845558504188768,
      "loss": 2.2663,
      "step": 23635
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.7438793182373047,
      "learning_rate": 0.0005845545548347587,
      "loss": 2.359,
      "step": 23636
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.7180402278900146,
      "learning_rate": 0.0005845532591977365,
      "loss": 2.2848,
      "step": 23637
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.638037383556366,
      "learning_rate": 0.0005845519635078106,
      "loss": 2.2505,
      "step": 23638
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6631921529769897,
      "learning_rate": 0.000584550667764981,
      "loss": 2.4089,
      "step": 23639
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6777923107147217,
      "learning_rate": 0.0005845493719692481,
      "loss": 2.1701,
      "step": 23640
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.561791718006134,
      "learning_rate": 0.0005845480761206121,
      "loss": 2.2484,
      "step": 23641
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6303778290748596,
      "learning_rate": 0.0005845467802190733,
      "loss": 2.5243,
      "step": 23642
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6702463626861572,
      "learning_rate": 0.0005845454842646318,
      "loss": 2.3616,
      "step": 23643
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6192637085914612,
      "learning_rate": 0.000584544188257288,
      "loss": 2.2978,
      "step": 23644
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6077053546905518,
      "learning_rate": 0.0005845428921970419,
      "loss": 2.1792,
      "step": 23645
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6205878257751465,
      "learning_rate": 0.000584541596083894,
      "loss": 2.4432,
      "step": 23646
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.8979598879814148,
      "learning_rate": 0.0005845402999178445,
      "loss": 2.3872,
      "step": 23647
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.7566477656364441,
      "learning_rate": 0.0005845390036988935,
      "loss": 2.3052,
      "step": 23648
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6501035094261169,
      "learning_rate": 0.0005845377074270414,
      "loss": 2.7037,
      "step": 23649
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6681540608406067,
      "learning_rate": 0.0005845364111022884,
      "loss": 1.9958,
      "step": 23650
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6979055404663086,
      "learning_rate": 0.0005845351147246346,
      "loss": 2.3778,
      "step": 23651
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.624426007270813,
      "learning_rate": 0.0005845338182940804,
      "loss": 2.1607,
      "step": 23652
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.706968367099762,
      "learning_rate": 0.0005845325218106261,
      "loss": 2.745,
      "step": 23653
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6844991445541382,
      "learning_rate": 0.0005845312252742717,
      "loss": 2.4508,
      "step": 23654
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.7144170999526978,
      "learning_rate": 0.0005845299286850175,
      "loss": 2.3388,
      "step": 23655
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.5992053747177124,
      "learning_rate": 0.000584528632042864,
      "loss": 2.3099,
      "step": 23656
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.7301350831985474,
      "learning_rate": 0.0005845273353478112,
      "loss": 2.2321,
      "step": 23657
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6565427780151367,
      "learning_rate": 0.0005845260385998595,
      "loss": 2.4216,
      "step": 23658
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.629728376865387,
      "learning_rate": 0.0005845247417990087,
      "loss": 2.3178,
      "step": 23659
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6309941411018372,
      "learning_rate": 0.0005845234449452597,
      "loss": 2.2024,
      "step": 23660
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.8136575222015381,
      "learning_rate": 0.0005845221480386123,
      "loss": 2.2756,
      "step": 23661
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6008157730102539,
      "learning_rate": 0.0005845208510790669,
      "loss": 2.4553,
      "step": 23662
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.60688316822052,
      "learning_rate": 0.0005845195540666237,
      "loss": 2.3482,
      "step": 23663
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.9251866340637207,
      "learning_rate": 0.0005845182570012828,
      "loss": 2.6601,
      "step": 23664
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6209244728088379,
      "learning_rate": 0.0005845169598830448,
      "loss": 2.4095,
      "step": 23665
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6414819955825806,
      "learning_rate": 0.0005845156627119096,
      "loss": 2.3721,
      "step": 23666
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.777193009853363,
      "learning_rate": 0.0005845143654878775,
      "loss": 2.4041,
      "step": 23667
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.7881138324737549,
      "learning_rate": 0.000584513068210949,
      "loss": 2.2997,
      "step": 23668
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6455926895141602,
      "learning_rate": 0.0005845117708811241,
      "loss": 2.4034,
      "step": 23669
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.7964146137237549,
      "learning_rate": 0.000584510473498403,
      "loss": 2.2998,
      "step": 23670
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.744634211063385,
      "learning_rate": 0.0005845091760627862,
      "loss": 2.4997,
      "step": 23671
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.5909519791603088,
      "learning_rate": 0.0005845078785742736,
      "loss": 2.3407,
      "step": 23672
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.5907936692237854,
      "learning_rate": 0.0005845065810328656,
      "loss": 2.3302,
      "step": 23673
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6245936751365662,
      "learning_rate": 0.0005845052834385626,
      "loss": 2.3625,
      "step": 23674
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.811214029788971,
      "learning_rate": 0.0005845039857913647,
      "loss": 2.1482,
      "step": 23675
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.5984456539154053,
      "learning_rate": 0.000584502688091272,
      "loss": 2.3063,
      "step": 23676
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6341551542282104,
      "learning_rate": 0.000584501390338285,
      "loss": 2.173,
      "step": 23677
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.723173201084137,
      "learning_rate": 0.0005845000925324038,
      "loss": 2.2382,
      "step": 23678
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6050119996070862,
      "learning_rate": 0.0005844987946736286,
      "loss": 2.3702,
      "step": 23679
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6198926568031311,
      "learning_rate": 0.0005844974967619597,
      "loss": 2.3364,
      "step": 23680
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6709310412406921,
      "learning_rate": 0.0005844961987973974,
      "loss": 2.1901,
      "step": 23681
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6260024905204773,
      "learning_rate": 0.0005844949007799418,
      "loss": 2.438,
      "step": 23682
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.5842427611351013,
      "learning_rate": 0.0005844936027095933,
      "loss": 2.5475,
      "step": 23683
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6396581530570984,
      "learning_rate": 0.0005844923045863521,
      "loss": 2.2626,
      "step": 23684
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6040405631065369,
      "learning_rate": 0.0005844910064102183,
      "loss": 2.327,
      "step": 23685
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.636131763458252,
      "learning_rate": 0.0005844897081811924,
      "loss": 2.5563,
      "step": 23686
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6078965067863464,
      "learning_rate": 0.0005844884098992743,
      "loss": 2.2902,
      "step": 23687
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.5996271967887878,
      "learning_rate": 0.0005844871115644647,
      "loss": 2.3571,
      "step": 23688
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6240525841712952,
      "learning_rate": 0.0005844858131767634,
      "loss": 2.4551,
      "step": 23689
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6672347187995911,
      "learning_rate": 0.0005844845147361708,
      "loss": 2.3349,
      "step": 23690
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6399441361427307,
      "learning_rate": 0.0005844832162426873,
      "loss": 2.377,
      "step": 23691
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6357889175415039,
      "learning_rate": 0.0005844819176963128,
      "loss": 2.2388,
      "step": 23692
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6287137866020203,
      "learning_rate": 0.0005844806190970478,
      "loss": 2.403,
      "step": 23693
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.7006676197052002,
      "learning_rate": 0.0005844793204448926,
      "loss": 2.458,
      "step": 23694
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.5895296335220337,
      "learning_rate": 0.0005844780217398473,
      "loss": 2.3423,
      "step": 23695
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.5790294408798218,
      "learning_rate": 0.0005844767229819121,
      "loss": 2.2313,
      "step": 23696
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6033892035484314,
      "learning_rate": 0.0005844754241710872,
      "loss": 2.3418,
      "step": 23697
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.554438591003418,
      "learning_rate": 0.0005844741253073732,
      "loss": 2.2407,
      "step": 23698
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6718344688415527,
      "learning_rate": 0.00058447282639077,
      "loss": 2.6202,
      "step": 23699
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.5758548378944397,
      "learning_rate": 0.0005844715274212778,
      "loss": 2.4905,
      "step": 23700
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.637682318687439,
      "learning_rate": 0.0005844702283988971,
      "loss": 2.2797,
      "step": 23701
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6288418173789978,
      "learning_rate": 0.000584468929323628,
      "loss": 2.2554,
      "step": 23702
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6758129000663757,
      "learning_rate": 0.0005844676301954708,
      "loss": 2.6618,
      "step": 23703
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6170172095298767,
      "learning_rate": 0.0005844663310144256,
      "loss": 2.4124,
      "step": 23704
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6212338209152222,
      "learning_rate": 0.0005844650317804928,
      "loss": 2.1847,
      "step": 23705
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.622678279876709,
      "learning_rate": 0.0005844637324936725,
      "loss": 2.482,
      "step": 23706
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6611369252204895,
      "learning_rate": 0.0005844624331539651,
      "loss": 2.2412,
      "step": 23707
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6037846207618713,
      "learning_rate": 0.0005844611337613707,
      "loss": 2.2231,
      "step": 23708
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6600832939147949,
      "learning_rate": 0.0005844598343158897,
      "loss": 2.4121,
      "step": 23709
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6566809415817261,
      "learning_rate": 0.0005844585348175221,
      "loss": 2.459,
      "step": 23710
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.5987794399261475,
      "learning_rate": 0.0005844572352662685,
      "loss": 2.2021,
      "step": 23711
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.666969358921051,
      "learning_rate": 0.0005844559356621287,
      "loss": 2.2413,
      "step": 23712
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.5720076560974121,
      "learning_rate": 0.0005844546360051034,
      "loss": 2.4043,
      "step": 23713
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.5860699415206909,
      "learning_rate": 0.0005844533362951925,
      "loss": 1.9775,
      "step": 23714
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6462884545326233,
      "learning_rate": 0.0005844520365323963,
      "loss": 2.3963,
      "step": 23715
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6086066365242004,
      "learning_rate": 0.0005844507367167152,
      "loss": 2.3081,
      "step": 23716
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6516969799995422,
      "learning_rate": 0.0005844494368481492,
      "loss": 2.5024,
      "step": 23717
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6068065166473389,
      "learning_rate": 0.0005844481369266989,
      "loss": 2.3696,
      "step": 23718
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.5959874987602234,
      "learning_rate": 0.0005844468369523642,
      "loss": 2.2389,
      "step": 23719
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.5682145357131958,
      "learning_rate": 0.0005844455369251454,
      "loss": 2.1622,
      "step": 23720
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.5776035189628601,
      "learning_rate": 0.0005844442368450429,
      "loss": 2.2013,
      "step": 23721
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.5914220809936523,
      "learning_rate": 0.0005844429367120568,
      "loss": 2.283,
      "step": 23722
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6729758977890015,
      "learning_rate": 0.0005844416365261874,
      "loss": 2.4209,
      "step": 23723
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6235074996948242,
      "learning_rate": 0.000584440336287435,
      "loss": 2.2989,
      "step": 23724
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.5910454988479614,
      "learning_rate": 0.0005844390359957996,
      "loss": 2.4653,
      "step": 23725
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6110098361968994,
      "learning_rate": 0.0005844377356512817,
      "loss": 2.2727,
      "step": 23726
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6128109693527222,
      "learning_rate": 0.0005844364352538815,
      "loss": 2.4663,
      "step": 23727
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.5970115661621094,
      "learning_rate": 0.0005844351348035992,
      "loss": 2.3219,
      "step": 23728
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6298834085464478,
      "learning_rate": 0.000584433834300435,
      "loss": 2.3006,
      "step": 23729
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6239727735519409,
      "learning_rate": 0.0005844325337443893,
      "loss": 2.24,
      "step": 23730
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.5964252948760986,
      "learning_rate": 0.0005844312331354621,
      "loss": 2.2951,
      "step": 23731
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6275169849395752,
      "learning_rate": 0.0005844299324736537,
      "loss": 2.5452,
      "step": 23732
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6206313371658325,
      "learning_rate": 0.0005844286317589645,
      "loss": 2.2859,
      "step": 23733
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6750363707542419,
      "learning_rate": 0.0005844273309913946,
      "loss": 2.3115,
      "step": 23734
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.565080463886261,
      "learning_rate": 0.0005844260301709443,
      "loss": 2.3059,
      "step": 23735
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6240730881690979,
      "learning_rate": 0.0005844247292976139,
      "loss": 2.1796,
      "step": 23736
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6832392811775208,
      "learning_rate": 0.0005844234283714035,
      "loss": 2.236,
      "step": 23737
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6023816466331482,
      "learning_rate": 0.0005844221273923134,
      "loss": 2.526,
      "step": 23738
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6654704213142395,
      "learning_rate": 0.000584420826360344,
      "loss": 2.3124,
      "step": 23739
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.5796099901199341,
      "learning_rate": 0.0005844195252754953,
      "loss": 2.1603,
      "step": 23740
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6352638602256775,
      "learning_rate": 0.0005844182241377676,
      "loss": 2.4637,
      "step": 23741
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6897040009498596,
      "learning_rate": 0.0005844169229471612,
      "loss": 2.2174,
      "step": 23742
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6894580721855164,
      "learning_rate": 0.0005844156217036764,
      "loss": 2.5118,
      "step": 23743
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6474833488464355,
      "learning_rate": 0.0005844143204073133,
      "loss": 2.2959,
      "step": 23744
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6087042093276978,
      "learning_rate": 0.0005844130190580722,
      "loss": 2.5468,
      "step": 23745
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6311746835708618,
      "learning_rate": 0.0005844117176559534,
      "loss": 2.4978,
      "step": 23746
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.7757232785224915,
      "learning_rate": 0.0005844104162009571,
      "loss": 2.4391,
      "step": 23747
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.647718608379364,
      "learning_rate": 0.0005844091146930835,
      "loss": 2.3291,
      "step": 23748
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6553388833999634,
      "learning_rate": 0.000584407813132333,
      "loss": 2.4217,
      "step": 23749
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6056042313575745,
      "learning_rate": 0.0005844065115187055,
      "loss": 2.2424,
      "step": 23750
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.611535370349884,
      "learning_rate": 0.0005844052098522016,
      "loss": 2.3737,
      "step": 23751
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.5977026224136353,
      "learning_rate": 0.0005844039081328213,
      "loss": 2.2802,
      "step": 23752
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6600380539894104,
      "learning_rate": 0.0005844026063605651,
      "loss": 2.6259,
      "step": 23753
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6708570122718811,
      "learning_rate": 0.0005844013045354329,
      "loss": 2.2668,
      "step": 23754
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.7329061627388,
      "learning_rate": 0.0005844000026574253,
      "loss": 2.2951,
      "step": 23755
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6972470879554749,
      "learning_rate": 0.0005843987007265423,
      "loss": 2.1338,
      "step": 23756
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6326231360435486,
      "learning_rate": 0.0005843973987427842,
      "loss": 2.3827,
      "step": 23757
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6497889757156372,
      "learning_rate": 0.0005843960967061513,
      "loss": 2.5982,
      "step": 23758
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.5958397388458252,
      "learning_rate": 0.0005843947946166437,
      "loss": 2.2611,
      "step": 23759
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6456906199455261,
      "learning_rate": 0.0005843934924742618,
      "loss": 2.327,
      "step": 23760
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.7032658457756042,
      "learning_rate": 0.0005843921902790058,
      "loss": 2.6133,
      "step": 23761
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6861735582351685,
      "learning_rate": 0.0005843908880308759,
      "loss": 2.1734,
      "step": 23762
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6420365571975708,
      "learning_rate": 0.0005843895857298724,
      "loss": 2.2555,
      "step": 23763
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6183538436889648,
      "learning_rate": 0.0005843882833759954,
      "loss": 2.2679,
      "step": 23764
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.777068018913269,
      "learning_rate": 0.0005843869809692454,
      "loss": 2.1233,
      "step": 23765
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.5423346757888794,
      "learning_rate": 0.0005843856785096225,
      "loss": 2.2666,
      "step": 23766
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6275453567504883,
      "learning_rate": 0.0005843843759971268,
      "loss": 2.2565,
      "step": 23767
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.7497653961181641,
      "learning_rate": 0.0005843830734317588,
      "loss": 2.45,
      "step": 23768
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.816498875617981,
      "learning_rate": 0.0005843817708135185,
      "loss": 2.2022,
      "step": 23769
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.5767515301704407,
      "learning_rate": 0.0005843804681424063,
      "loss": 2.2046,
      "step": 23770
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.0419349670410156,
      "learning_rate": 0.0005843791654184224,
      "loss": 2.1933,
      "step": 23771
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.5762893557548523,
      "learning_rate": 0.000584377862641567,
      "loss": 2.1912,
      "step": 23772
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.657373309135437,
      "learning_rate": 0.0005843765598118405,
      "loss": 2.5299,
      "step": 23773
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6683188676834106,
      "learning_rate": 0.000584375256929243,
      "loss": 2.3965,
      "step": 23774
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6536915898323059,
      "learning_rate": 0.0005843739539937747,
      "loss": 2.2873,
      "step": 23775
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.635516881942749,
      "learning_rate": 0.000584372651005436,
      "loss": 2.6727,
      "step": 23776
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6099022030830383,
      "learning_rate": 0.0005843713479642271,
      "loss": 2.4746,
      "step": 23777
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6576849222183228,
      "learning_rate": 0.000584370044870148,
      "loss": 2.2095,
      "step": 23778
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6961989998817444,
      "learning_rate": 0.0005843687417231993,
      "loss": 2.2627,
      "step": 23779
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6391093730926514,
      "learning_rate": 0.000584367438523381,
      "loss": 2.2082,
      "step": 23780
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.5840246677398682,
      "learning_rate": 0.0005843661352706935,
      "loss": 2.3592,
      "step": 23781
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6521970629692078,
      "learning_rate": 0.0005843648319651371,
      "loss": 2.1743,
      "step": 23782
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6420706510543823,
      "learning_rate": 0.0005843635286067115,
      "loss": 2.0693,
      "step": 23783
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6442790031433105,
      "learning_rate": 0.0005843622251954177,
      "loss": 2.284,
      "step": 23784
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.606934130191803,
      "learning_rate": 0.0005843609217312556,
      "loss": 2.2404,
      "step": 23785
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6350623965263367,
      "learning_rate": 0.0005843596182142253,
      "loss": 2.3812,
      "step": 23786
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6610500812530518,
      "learning_rate": 0.0005843583146443272,
      "loss": 2.3577,
      "step": 23787
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.5859366059303284,
      "learning_rate": 0.0005843570110215615,
      "loss": 2.47,
      "step": 23788
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6369591951370239,
      "learning_rate": 0.0005843557073459285,
      "loss": 2.2226,
      "step": 23789
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6421722769737244,
      "learning_rate": 0.0005843544036174284,
      "loss": 2.2979,
      "step": 23790
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6727985143661499,
      "learning_rate": 0.0005843530998360614,
      "loss": 2.3289,
      "step": 23791
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.642460823059082,
      "learning_rate": 0.0005843517960018279,
      "loss": 2.257,
      "step": 23792
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6335367560386658,
      "learning_rate": 0.000584350492114728,
      "loss": 2.2263,
      "step": 23793
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.5797938108444214,
      "learning_rate": 0.0005843491881747619,
      "loss": 2.3046,
      "step": 23794
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6142883896827698,
      "learning_rate": 0.00058434788418193,
      "loss": 2.1195,
      "step": 23795
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.9267433881759644,
      "learning_rate": 0.0005843465801362324,
      "loss": 2.3098,
      "step": 23796
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.5725167393684387,
      "learning_rate": 0.0005843452760376694,
      "loss": 2.3174,
      "step": 23797
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6909403800964355,
      "learning_rate": 0.0005843439718862413,
      "loss": 2.2952,
      "step": 23798
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6964330077171326,
      "learning_rate": 0.0005843426676819482,
      "loss": 2.0539,
      "step": 23799
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.7467255592346191,
      "learning_rate": 0.0005843413634247906,
      "loss": 2.454,
      "step": 23800
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6765312552452087,
      "learning_rate": 0.0005843400591147685,
      "loss": 2.152,
      "step": 23801
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6536023616790771,
      "learning_rate": 0.0005843387547518821,
      "loss": 2.2887,
      "step": 23802
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.7504843473434448,
      "learning_rate": 0.0005843374503361319,
      "loss": 2.4795,
      "step": 23803
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6095620393753052,
      "learning_rate": 0.0005843361458675179,
      "loss": 2.2693,
      "step": 23804
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6675182580947876,
      "learning_rate": 0.0005843348413460405,
      "loss": 2.0735,
      "step": 23805
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6015819907188416,
      "learning_rate": 0.0005843335367716999,
      "loss": 2.3424,
      "step": 23806
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.5807301998138428,
      "learning_rate": 0.0005843322321444963,
      "loss": 2.2791,
      "step": 23807
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.7138820290565491,
      "learning_rate": 0.0005843309274644299,
      "loss": 2.3616,
      "step": 23808
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6269199848175049,
      "learning_rate": 0.0005843296227315012,
      "loss": 2.3156,
      "step": 23809
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6970131397247314,
      "learning_rate": 0.0005843283179457101,
      "loss": 2.2917,
      "step": 23810
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.5476449131965637,
      "learning_rate": 0.000584327013107057,
      "loss": 2.2653,
      "step": 23811
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6866700053215027,
      "learning_rate": 0.0005843257082155422,
      "loss": 2.431,
      "step": 23812
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.5856034159660339,
      "learning_rate": 0.0005843244032711659,
      "loss": 2.2732,
      "step": 23813
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6784339547157288,
      "learning_rate": 0.0005843230982739283,
      "loss": 2.0045,
      "step": 23814
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6232609748840332,
      "learning_rate": 0.0005843217932238297,
      "loss": 2.4115,
      "step": 23815
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.606468677520752,
      "learning_rate": 0.0005843204881208702,
      "loss": 2.1493,
      "step": 23816
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.63200443983078,
      "learning_rate": 0.0005843191829650503,
      "loss": 2.0767,
      "step": 23817
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.650641918182373,
      "learning_rate": 0.00058431787775637,
      "loss": 2.2609,
      "step": 23818
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6278043389320374,
      "learning_rate": 0.0005843165724948296,
      "loss": 2.3047,
      "step": 23819
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6378223299980164,
      "learning_rate": 0.0005843152671804294,
      "loss": 2.3471,
      "step": 23820
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6382823586463928,
      "learning_rate": 0.0005843139618131697,
      "loss": 2.431,
      "step": 23821
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6181250810623169,
      "learning_rate": 0.0005843126563930507,
      "loss": 2.2199,
      "step": 23822
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.5890845656394958,
      "learning_rate": 0.0005843113509200725,
      "loss": 2.264,
      "step": 23823
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6453791856765747,
      "learning_rate": 0.0005843100453942356,
      "loss": 2.3671,
      "step": 23824
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.5687557458877563,
      "learning_rate": 0.00058430873981554,
      "loss": 2.4391,
      "step": 23825
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6212948560714722,
      "learning_rate": 0.0005843074341839859,
      "loss": 2.1982,
      "step": 23826
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6615602970123291,
      "learning_rate": 0.0005843061284995739,
      "loss": 2.5482,
      "step": 23827
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6489495038986206,
      "learning_rate": 0.000584304822762304,
      "loss": 2.4145,
      "step": 23828
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6336973905563354,
      "learning_rate": 0.0005843035169721763,
      "loss": 2.4345,
      "step": 23829
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6414491534233093,
      "learning_rate": 0.0005843022111291914,
      "loss": 2.2576,
      "step": 23830
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.7067669630050659,
      "learning_rate": 0.0005843009052333493,
      "loss": 2.3313,
      "step": 23831
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6279929280281067,
      "learning_rate": 0.0005842995992846503,
      "loss": 2.3881,
      "step": 23832
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.5948300361633301,
      "learning_rate": 0.0005842982932830946,
      "loss": 2.0938,
      "step": 23833
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6091703772544861,
      "learning_rate": 0.0005842969872286824,
      "loss": 2.0658,
      "step": 23834
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.7089720964431763,
      "learning_rate": 0.0005842956811214141,
      "loss": 2.3152,
      "step": 23835
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.694469153881073,
      "learning_rate": 0.00058429437496129,
      "loss": 2.4225,
      "step": 23836
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.631333589553833,
      "learning_rate": 0.00058429306874831,
      "loss": 2.4845,
      "step": 23837
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6377507448196411,
      "learning_rate": 0.0005842917624824746,
      "loss": 2.5813,
      "step": 23838
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.7216606736183167,
      "learning_rate": 0.000584290456163784,
      "loss": 2.1386,
      "step": 23839
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6030609607696533,
      "learning_rate": 0.0005842891497922385,
      "loss": 2.2575,
      "step": 23840
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.688396155834198,
      "learning_rate": 0.0005842878433678382,
      "loss": 2.3747,
      "step": 23841
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6445686221122742,
      "learning_rate": 0.0005842865368905835,
      "loss": 2.3021,
      "step": 23842
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6777306795120239,
      "learning_rate": 0.0005842852303604745,
      "loss": 2.5424,
      "step": 23843
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6546247601509094,
      "learning_rate": 0.0005842839237775116,
      "loss": 2.3027,
      "step": 23844
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6186208724975586,
      "learning_rate": 0.0005842826171416949,
      "loss": 2.4015,
      "step": 23845
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6462066769599915,
      "learning_rate": 0.0005842813104530247,
      "loss": 2.3164,
      "step": 23846
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.7884687781333923,
      "learning_rate": 0.0005842800037115012,
      "loss": 2.2416,
      "step": 23847
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6372030377388,
      "learning_rate": 0.0005842786969171247,
      "loss": 2.3345,
      "step": 23848
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6862592101097107,
      "learning_rate": 0.0005842773900698956,
      "loss": 2.3895,
      "step": 23849
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6753905415534973,
      "learning_rate": 0.0005842760831698138,
      "loss": 2.4348,
      "step": 23850
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6734483242034912,
      "learning_rate": 0.0005842747762168796,
      "loss": 2.3304,
      "step": 23851
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6249371767044067,
      "learning_rate": 0.0005842734692110934,
      "loss": 2.4008,
      "step": 23852
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6252853274345398,
      "learning_rate": 0.0005842721621524556,
      "loss": 2.2338,
      "step": 23853
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.5972471237182617,
      "learning_rate": 0.000584270855040966,
      "loss": 2.3407,
      "step": 23854
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6640647053718567,
      "learning_rate": 0.0005842695478766253,
      "loss": 2.2526,
      "step": 23855
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6344742178916931,
      "learning_rate": 0.0005842682406594333,
      "loss": 2.5146,
      "step": 23856
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6154377460479736,
      "learning_rate": 0.0005842669333893907,
      "loss": 2.1028,
      "step": 23857
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6392545104026794,
      "learning_rate": 0.0005842656260664974,
      "loss": 2.5589,
      "step": 23858
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6758818626403809,
      "learning_rate": 0.0005842643186907537,
      "loss": 2.5117,
      "step": 23859
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.5849689245223999,
      "learning_rate": 0.00058426301126216,
      "loss": 2.2358,
      "step": 23860
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6671217679977417,
      "learning_rate": 0.0005842617037807163,
      "loss": 2.3297,
      "step": 23861
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6513389348983765,
      "learning_rate": 0.0005842603962464231,
      "loss": 2.4639,
      "step": 23862
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.5635071396827698,
      "learning_rate": 0.0005842590886592805,
      "loss": 2.3279,
      "step": 23863
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.7553592324256897,
      "learning_rate": 0.0005842577810192888,
      "loss": 2.3785,
      "step": 23864
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6329942941665649,
      "learning_rate": 0.0005842564733264481,
      "loss": 2.3253,
      "step": 23865
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6833222508430481,
      "learning_rate": 0.0005842551655807588,
      "loss": 2.4764,
      "step": 23866
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.585487425327301,
      "learning_rate": 0.0005842538577822212,
      "loss": 2.3189,
      "step": 23867
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6309952735900879,
      "learning_rate": 0.0005842525499308354,
      "loss": 2.3068,
      "step": 23868
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6213576793670654,
      "learning_rate": 0.0005842512420266016,
      "loss": 2.2282,
      "step": 23869
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.7182978987693787,
      "learning_rate": 0.0005842499340695202,
      "loss": 2.1992,
      "step": 23870
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6766787767410278,
      "learning_rate": 0.0005842486260595914,
      "loss": 2.4348,
      "step": 23871
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6344073414802551,
      "learning_rate": 0.0005842473179968153,
      "loss": 2.3523,
      "step": 23872
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.7144971489906311,
      "learning_rate": 0.0005842460098811924,
      "loss": 2.1956,
      "step": 23873
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6134117841720581,
      "learning_rate": 0.0005842447017127227,
      "loss": 2.323,
      "step": 23874
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6620182394981384,
      "learning_rate": 0.0005842433934914066,
      "loss": 2.5232,
      "step": 23875
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.613215982913971,
      "learning_rate": 0.0005842420852172442,
      "loss": 2.1235,
      "step": 23876
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.646758496761322,
      "learning_rate": 0.0005842407768902359,
      "loss": 2.4989,
      "step": 23877
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.607745885848999,
      "learning_rate": 0.0005842394685103819,
      "loss": 2.3987,
      "step": 23878
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6464806795120239,
      "learning_rate": 0.0005842381600776824,
      "loss": 2.1305,
      "step": 23879
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6134693026542664,
      "learning_rate": 0.0005842368515921376,
      "loss": 2.2666,
      "step": 23880
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.5457233190536499,
      "learning_rate": 0.0005842355430537479,
      "loss": 2.1797,
      "step": 23881
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.7957303524017334,
      "learning_rate": 0.0005842342344625134,
      "loss": 2.4179,
      "step": 23882
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6124467253684998,
      "learning_rate": 0.0005842329258184344,
      "loss": 2.3452,
      "step": 23883
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6033533811569214,
      "learning_rate": 0.0005842316171215112,
      "loss": 2.3061,
      "step": 23884
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.643077552318573,
      "learning_rate": 0.0005842303083717438,
      "loss": 2.32,
      "step": 23885
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.584959864616394,
      "learning_rate": 0.0005842289995691327,
      "loss": 2.3334,
      "step": 23886
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.576038122177124,
      "learning_rate": 0.0005842276907136781,
      "loss": 2.2847,
      "step": 23887
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6440293788909912,
      "learning_rate": 0.0005842263818053801,
      "loss": 2.3825,
      "step": 23888
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.7692602872848511,
      "learning_rate": 0.0005842250728442393,
      "loss": 2.3607,
      "step": 23889
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.606035590171814,
      "learning_rate": 0.0005842237638302554,
      "loss": 2.2473,
      "step": 23890
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6420968174934387,
      "learning_rate": 0.0005842224547634292,
      "loss": 2.3615,
      "step": 23891
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6605753898620605,
      "learning_rate": 0.0005842211456437605,
      "loss": 2.415,
      "step": 23892
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.672961413860321,
      "learning_rate": 0.0005842198364712498,
      "loss": 2.3713,
      "step": 23893
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6023067831993103,
      "learning_rate": 0.0005842185272458972,
      "loss": 2.3209,
      "step": 23894
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6917572021484375,
      "learning_rate": 0.0005842172179677031,
      "loss": 2.5435,
      "step": 23895
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6358690857887268,
      "learning_rate": 0.0005842159086366676,
      "loss": 2.4603,
      "step": 23896
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6416420340538025,
      "learning_rate": 0.0005842145992527909,
      "loss": 2.4667,
      "step": 23897
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6291124224662781,
      "learning_rate": 0.0005842132898160735,
      "loss": 2.555,
      "step": 23898
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6734112501144409,
      "learning_rate": 0.0005842119803265154,
      "loss": 2.3616,
      "step": 23899
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6125218272209167,
      "learning_rate": 0.000584210670784117,
      "loss": 2.4071,
      "step": 23900
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6123447418212891,
      "learning_rate": 0.0005842093611888784,
      "loss": 2.2755,
      "step": 23901
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6295526623725891,
      "learning_rate": 0.0005842080515408001,
      "loss": 2.1217,
      "step": 23902
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6700730323791504,
      "learning_rate": 0.0005842067418398819,
      "loss": 2.3891,
      "step": 23903
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.5975480675697327,
      "learning_rate": 0.0005842054320861244,
      "loss": 2.0836,
      "step": 23904
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6456741690635681,
      "learning_rate": 0.0005842041222795277,
      "loss": 2.391,
      "step": 23905
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6021678447723389,
      "learning_rate": 0.0005842028124200923,
      "loss": 2.3049,
      "step": 23906
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6034472584724426,
      "learning_rate": 0.000584201502507818,
      "loss": 2.3432,
      "step": 23907
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6460736989974976,
      "learning_rate": 0.0005842001925427053,
      "loss": 2.1979,
      "step": 23908
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.5911082029342651,
      "learning_rate": 0.0005841988825247545,
      "loss": 2.2686,
      "step": 23909
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.5945557951927185,
      "learning_rate": 0.0005841975724539657,
      "loss": 2.1752,
      "step": 23910
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6138144135475159,
      "learning_rate": 0.0005841962623303392,
      "loss": 2.3133,
      "step": 23911
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.8375283479690552,
      "learning_rate": 0.0005841949521538753,
      "loss": 2.4202,
      "step": 23912
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6298021078109741,
      "learning_rate": 0.0005841936419245743,
      "loss": 2.229,
      "step": 23913
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6498690247535706,
      "learning_rate": 0.0005841923316424361,
      "loss": 2.4305,
      "step": 23914
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.5877526998519897,
      "learning_rate": 0.0005841910213074614,
      "loss": 2.274,
      "step": 23915
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6770603060722351,
      "learning_rate": 0.0005841897109196501,
      "loss": 2.4077,
      "step": 23916
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6422775387763977,
      "learning_rate": 0.0005841884004790025,
      "loss": 2.2775,
      "step": 23917
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.776076078414917,
      "learning_rate": 0.000584187089985519,
      "loss": 2.6331,
      "step": 23918
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.641638994216919,
      "learning_rate": 0.0005841857794391998,
      "loss": 2.1608,
      "step": 23919
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6796273589134216,
      "learning_rate": 0.0005841844688400451,
      "loss": 2.2654,
      "step": 23920
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.7790470123291016,
      "learning_rate": 0.000584183158188055,
      "loss": 2.2214,
      "step": 23921
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.7417762279510498,
      "learning_rate": 0.0005841818474832301,
      "loss": 2.2039,
      "step": 23922
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.8080212473869324,
      "learning_rate": 0.0005841805367255704,
      "loss": 2.6323,
      "step": 23923
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6458842158317566,
      "learning_rate": 0.000584179225915076,
      "loss": 2.2126,
      "step": 23924
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.7148489952087402,
      "learning_rate": 0.0005841779150517473,
      "loss": 2.2699,
      "step": 23925
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.7633419036865234,
      "learning_rate": 0.0005841766041355848,
      "loss": 2.444,
      "step": 23926
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6710695624351501,
      "learning_rate": 0.0005841752931665884,
      "loss": 2.1806,
      "step": 23927
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.73647540807724,
      "learning_rate": 0.0005841739821447583,
      "loss": 2.4734,
      "step": 23928
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.7233278751373291,
      "learning_rate": 0.0005841726710700951,
      "loss": 2.4228,
      "step": 23929
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.5897470712661743,
      "learning_rate": 0.0005841713599425988,
      "loss": 2.2856,
      "step": 23930
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6817732453346252,
      "learning_rate": 0.0005841700487622696,
      "loss": 2.2721,
      "step": 23931
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6307050585746765,
      "learning_rate": 0.0005841687375291079,
      "loss": 2.3233,
      "step": 23932
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6354132890701294,
      "learning_rate": 0.0005841674262431137,
      "loss": 2.3874,
      "step": 23933
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.674892246723175,
      "learning_rate": 0.0005841661149042876,
      "loss": 2.2037,
      "step": 23934
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.7208989858627319,
      "learning_rate": 0.0005841648035126297,
      "loss": 2.2111,
      "step": 23935
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6649957299232483,
      "learning_rate": 0.00058416349206814,
      "loss": 2.0453,
      "step": 23936
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6005469560623169,
      "learning_rate": 0.0005841621805708191,
      "loss": 2.4211,
      "step": 23937
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6987032294273376,
      "learning_rate": 0.000584160869020667,
      "loss": 2.3497,
      "step": 23938
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6715323328971863,
      "learning_rate": 0.000584159557417684,
      "loss": 2.5533,
      "step": 23939
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.7272917032241821,
      "learning_rate": 0.0005841582457618704,
      "loss": 2.1168,
      "step": 23940
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6272487044334412,
      "learning_rate": 0.0005841569340532265,
      "loss": 2.1695,
      "step": 23941
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6678072810173035,
      "learning_rate": 0.0005841556222917524,
      "loss": 2.5618,
      "step": 23942
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6967519521713257,
      "learning_rate": 0.0005841543104774484,
      "loss": 2.6277,
      "step": 23943
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6676385998725891,
      "learning_rate": 0.0005841529986103148,
      "loss": 2.3525,
      "step": 23944
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.681503415107727,
      "learning_rate": 0.0005841516866903517,
      "loss": 2.2747,
      "step": 23945
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.5714132785797119,
      "learning_rate": 0.0005841503747175596,
      "loss": 2.2933,
      "step": 23946
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.760287880897522,
      "learning_rate": 0.0005841490626919384,
      "loss": 2.3579,
      "step": 23947
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.7655467391014099,
      "learning_rate": 0.0005841477506134886,
      "loss": 2.3441,
      "step": 23948
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.7126284241676331,
      "learning_rate": 0.0005841464384822104,
      "loss": 2.297,
      "step": 23949
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.8695242404937744,
      "learning_rate": 0.0005841451262981039,
      "loss": 2.3254,
      "step": 23950
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.7165164947509766,
      "learning_rate": 0.0005841438140611696,
      "loss": 2.28,
      "step": 23951
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.651081919670105,
      "learning_rate": 0.0005841425017714076,
      "loss": 2.5491,
      "step": 23952
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.8227536678314209,
      "learning_rate": 0.000584141189428818,
      "loss": 2.2577,
      "step": 23953
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.7196912169456482,
      "learning_rate": 0.0005841398770334013,
      "loss": 2.2265,
      "step": 23954
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.621489942073822,
      "learning_rate": 0.0005841385645851576,
      "loss": 2.1562,
      "step": 23955
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.7245913743972778,
      "learning_rate": 0.0005841372520840872,
      "loss": 2.1767,
      "step": 23956
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.7990608215332031,
      "learning_rate": 0.0005841359395301904,
      "loss": 2.1045,
      "step": 23957
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.673628568649292,
      "learning_rate": 0.0005841346269234672,
      "loss": 2.34,
      "step": 23958
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.5971578359603882,
      "learning_rate": 0.0005841333142639181,
      "loss": 2.2921,
      "step": 23959
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.7055385112762451,
      "learning_rate": 0.0005841320015515432,
      "loss": 2.3471,
      "step": 23960
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6829471588134766,
      "learning_rate": 0.0005841306887863428,
      "loss": 2.1678,
      "step": 23961
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.7103369832038879,
      "learning_rate": 0.0005841293759683172,
      "loss": 2.322,
      "step": 23962
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6282728314399719,
      "learning_rate": 0.0005841280630974667,
      "loss": 2.3255,
      "step": 23963
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.7818235754966736,
      "learning_rate": 0.0005841267501737912,
      "loss": 2.297,
      "step": 23964
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.676861584186554,
      "learning_rate": 0.0005841254371972912,
      "loss": 2.3134,
      "step": 23965
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6765103936195374,
      "learning_rate": 0.000584124124167967,
      "loss": 2.5014,
      "step": 23966
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.8736774921417236,
      "learning_rate": 0.0005841228110858188,
      "loss": 2.1354,
      "step": 23967
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6861364245414734,
      "learning_rate": 0.0005841214979508467,
      "loss": 2.2038,
      "step": 23968
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6914151906967163,
      "learning_rate": 0.0005841201847630511,
      "loss": 2.3281,
      "step": 23969
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6568108797073364,
      "learning_rate": 0.0005841188715224321,
      "loss": 2.4087,
      "step": 23970
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.5981976389884949,
      "learning_rate": 0.0005841175582289903,
      "loss": 2.1978,
      "step": 23971
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6992143988609314,
      "learning_rate": 0.0005841162448827255,
      "loss": 2.3794,
      "step": 23972
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6075024008750916,
      "learning_rate": 0.0005841149314836381,
      "loss": 2.541,
      "step": 23973
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.7833825349807739,
      "learning_rate": 0.0005841136180317284,
      "loss": 2.2155,
      "step": 23974
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6647272109985352,
      "learning_rate": 0.0005841123045269966,
      "loss": 2.2791,
      "step": 23975
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6446850299835205,
      "learning_rate": 0.0005841109909694431,
      "loss": 2.4604,
      "step": 23976
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.7072615027427673,
      "learning_rate": 0.0005841096773590678,
      "loss": 2.466,
      "step": 23977
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6916069388389587,
      "learning_rate": 0.0005841083636958713,
      "loss": 2.5557,
      "step": 23978
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6359573006629944,
      "learning_rate": 0.0005841070499798536,
      "loss": 2.5455,
      "step": 23979
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6139342784881592,
      "learning_rate": 0.0005841057362110151,
      "loss": 2.3866,
      "step": 23980
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6651986241340637,
      "learning_rate": 0.0005841044223893559,
      "loss": 2.2832,
      "step": 23981
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6997746825218201,
      "learning_rate": 0.0005841031085148764,
      "loss": 2.1984,
      "step": 23982
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6742774844169617,
      "learning_rate": 0.0005841017945875768,
      "loss": 2.2578,
      "step": 23983
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.7158710360527039,
      "learning_rate": 0.0005841004806074573,
      "loss": 2.4066,
      "step": 23984
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6366904377937317,
      "learning_rate": 0.000584099166574518,
      "loss": 2.4726,
      "step": 23985
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6666886806488037,
      "learning_rate": 0.0005840978524887595,
      "loss": 2.5275,
      "step": 23986
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6410212516784668,
      "learning_rate": 0.0005840965383501817,
      "loss": 2.6251,
      "step": 23987
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.7049877047538757,
      "learning_rate": 0.000584095224158785,
      "loss": 2.345,
      "step": 23988
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6604739427566528,
      "learning_rate": 0.0005840939099145697,
      "loss": 2.294,
      "step": 23989
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.5723395347595215,
      "learning_rate": 0.0005840925956175361,
      "loss": 2.4281,
      "step": 23990
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6368697881698608,
      "learning_rate": 0.0005840912812676841,
      "loss": 2.2628,
      "step": 23991
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.7681713700294495,
      "learning_rate": 0.0005840899668650144,
      "loss": 2.2657,
      "step": 23992
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.5892363786697388,
      "learning_rate": 0.0005840886524095269,
      "loss": 2.4049,
      "step": 23993
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6205773949623108,
      "learning_rate": 0.0005840873379012218,
      "loss": 2.2815,
      "step": 23994
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6139190793037415,
      "learning_rate": 0.0005840860233400996,
      "loss": 2.4326,
      "step": 23995
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.7927055358886719,
      "learning_rate": 0.0005840847087261606,
      "loss": 2.3684,
      "step": 23996
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.5879732370376587,
      "learning_rate": 0.0005840833940594047,
      "loss": 2.2049,
      "step": 23997
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6038284301757812,
      "learning_rate": 0.0005840820793398324,
      "loss": 2.1533,
      "step": 23998
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6437138915061951,
      "learning_rate": 0.0005840807645674438,
      "loss": 2.0874,
      "step": 23999
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.8123903870582581,
      "learning_rate": 0.0005840794497422394,
      "loss": 2.4884,
      "step": 24000
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6334413886070251,
      "learning_rate": 0.0005840781348642192,
      "loss": 2.4431,
      "step": 24001
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.7382674217224121,
      "learning_rate": 0.0005840768199333834,
      "loss": 2.3207,
      "step": 24002
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6825761795043945,
      "learning_rate": 0.0005840755049497325,
      "loss": 2.299,
      "step": 24003
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6985417604446411,
      "learning_rate": 0.0005840741899132666,
      "loss": 2.4174,
      "step": 24004
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6029908061027527,
      "learning_rate": 0.0005840728748239859,
      "loss": 2.2836,
      "step": 24005
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.683586835861206,
      "learning_rate": 0.0005840715596818906,
      "loss": 2.3613,
      "step": 24006
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6877187490463257,
      "learning_rate": 0.0005840702444869811,
      "loss": 2.3124,
      "step": 24007
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.7002319097518921,
      "learning_rate": 0.0005840689292392577,
      "loss": 2.2841,
      "step": 24008
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6690076589584351,
      "learning_rate": 0.0005840676139387205,
      "loss": 2.338,
      "step": 24009
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6280272603034973,
      "learning_rate": 0.0005840662985853696,
      "loss": 2.4518,
      "step": 24010
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.7176491618156433,
      "learning_rate": 0.0005840649831792055,
      "loss": 2.2036,
      "step": 24011
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6824204921722412,
      "learning_rate": 0.0005840636677202283,
      "loss": 2.457,
      "step": 24012
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6005271673202515,
      "learning_rate": 0.0005840623522084384,
      "loss": 2.1224,
      "step": 24013
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.648272693157196,
      "learning_rate": 0.000584061036643836,
      "loss": 2.2507,
      "step": 24014
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.7100560665130615,
      "learning_rate": 0.000584059721026421,
      "loss": 2.2662,
      "step": 24015
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6123352646827698,
      "learning_rate": 0.0005840584053561943,
      "loss": 2.5005,
      "step": 24016
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6358662247657776,
      "learning_rate": 0.0005840570896331555,
      "loss": 2.375,
      "step": 24017
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6350505352020264,
      "learning_rate": 0.0005840557738573053,
      "loss": 2.2521,
      "step": 24018
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6330297589302063,
      "learning_rate": 0.0005840544580286436,
      "loss": 2.2664,
      "step": 24019
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.640893816947937,
      "learning_rate": 0.0005840531421471711,
      "loss": 2.1202,
      "step": 24020
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.7119927406311035,
      "learning_rate": 0.0005840518262128875,
      "loss": 2.5212,
      "step": 24021
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6874606013298035,
      "learning_rate": 0.0005840505102257934,
      "loss": 2.3188,
      "step": 24022
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.5604796409606934,
      "learning_rate": 0.0005840491941858889,
      "loss": 2.2043,
      "step": 24023
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6243599653244019,
      "learning_rate": 0.0005840478780931743,
      "loss": 2.1798,
      "step": 24024
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6800087094306946,
      "learning_rate": 0.0005840465619476499,
      "loss": 2.291,
      "step": 24025
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.5912792086601257,
      "learning_rate": 0.0005840452457493158,
      "loss": 2.2336,
      "step": 24026
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6453113555908203,
      "learning_rate": 0.0005840439294981725,
      "loss": 2.5644,
      "step": 24027
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6284849047660828,
      "learning_rate": 0.0005840426131942199,
      "loss": 2.2551,
      "step": 24028
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.677088737487793,
      "learning_rate": 0.0005840412968374585,
      "loss": 2.4174,
      "step": 24029
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.7300508618354797,
      "learning_rate": 0.0005840399804278884,
      "loss": 2.6222,
      "step": 24030
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6669169664382935,
      "learning_rate": 0.0005840386639655099,
      "loss": 2.3944,
      "step": 24031
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6171168088912964,
      "learning_rate": 0.0005840373474503233,
      "loss": 2.389,
      "step": 24032
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6226876974105835,
      "learning_rate": 0.0005840360308823288,
      "loss": 2.2854,
      "step": 24033
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.734009325504303,
      "learning_rate": 0.0005840347142615265,
      "loss": 2.3264,
      "step": 24034
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6492920517921448,
      "learning_rate": 0.000584033397587917,
      "loss": 2.385,
      "step": 24035
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6313895583152771,
      "learning_rate": 0.0005840320808615002,
      "loss": 2.186,
      "step": 24036
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.7165001034736633,
      "learning_rate": 0.0005840307640822765,
      "loss": 2.3222,
      "step": 24037
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6230586171150208,
      "learning_rate": 0.0005840294472502462,
      "loss": 2.2785,
      "step": 24038
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6499520540237427,
      "learning_rate": 0.0005840281303654093,
      "loss": 2.4427,
      "step": 24039
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.674933135509491,
      "learning_rate": 0.0005840268134277664,
      "loss": 2.3879,
      "step": 24040
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.7319930195808411,
      "learning_rate": 0.0005840254964373174,
      "loss": 2.4553,
      "step": 24041
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6197807192802429,
      "learning_rate": 0.0005840241793940627,
      "loss": 2.4744,
      "step": 24042
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6499165296554565,
      "learning_rate": 0.0005840228622980025,
      "loss": 2.5735,
      "step": 24043
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6658365726470947,
      "learning_rate": 0.0005840215451491373,
      "loss": 2.3976,
      "step": 24044
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6329634785652161,
      "learning_rate": 0.000584020227947467,
      "loss": 2.3779,
      "step": 24045
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.646802544593811,
      "learning_rate": 0.000584018910692992,
      "loss": 2.2862,
      "step": 24046
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6612443923950195,
      "learning_rate": 0.0005840175933857126,
      "loss": 2.1654,
      "step": 24047
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.7413287162780762,
      "learning_rate": 0.0005840162760256288,
      "loss": 2.4032,
      "step": 24048
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6845149993896484,
      "learning_rate": 0.0005840149586127412,
      "loss": 2.4094,
      "step": 24049
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6915122270584106,
      "learning_rate": 0.0005840136411470496,
      "loss": 2.2402,
      "step": 24050
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.667008638381958,
      "learning_rate": 0.0005840123236285547,
      "loss": 2.1623,
      "step": 24051
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6857837438583374,
      "learning_rate": 0.0005840110060572566,
      "loss": 2.4132,
      "step": 24052
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.613887369632721,
      "learning_rate": 0.0005840096884331553,
      "loss": 2.2562,
      "step": 24053
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6502558588981628,
      "learning_rate": 0.0005840083707562514,
      "loss": 2.1546,
      "step": 24054
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6245357990264893,
      "learning_rate": 0.000584007053026545,
      "loss": 2.3542,
      "step": 24055
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.7028293609619141,
      "learning_rate": 0.0005840057352440362,
      "loss": 2.411,
      "step": 24056
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6463459730148315,
      "learning_rate": 0.0005840044174087254,
      "loss": 2.5349,
      "step": 24057
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6207841038703918,
      "learning_rate": 0.0005840030995206128,
      "loss": 2.5404,
      "step": 24058
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6113770604133606,
      "learning_rate": 0.0005840017815796988,
      "loss": 2.5764,
      "step": 24059
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.5914439558982849,
      "learning_rate": 0.0005840004635859833,
      "loss": 2.3308,
      "step": 24060
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.611599326133728,
      "learning_rate": 0.000583999145539467,
      "loss": 2.1712,
      "step": 24061
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.5706149935722351,
      "learning_rate": 0.0005839978274401497,
      "loss": 2.3649,
      "step": 24062
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6280057430267334,
      "learning_rate": 0.0005839965092880319,
      "loss": 2.1817,
      "step": 24063
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6492063999176025,
      "learning_rate": 0.0005839951910831138,
      "loss": 2.5623,
      "step": 24064
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6063084602355957,
      "learning_rate": 0.0005839938728253956,
      "loss": 1.9743,
      "step": 24065
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.5525090098381042,
      "learning_rate": 0.0005839925545148776,
      "loss": 2.4802,
      "step": 24066
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.5708015561103821,
      "learning_rate": 0.0005839912361515601,
      "loss": 2.4261,
      "step": 24067
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.664106547832489,
      "learning_rate": 0.0005839899177354431,
      "loss": 2.5113,
      "step": 24068
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6140214800834656,
      "learning_rate": 0.0005839885992665272,
      "loss": 2.2747,
      "step": 24069
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.5968796014785767,
      "learning_rate": 0.0005839872807448124,
      "loss": 2.2112,
      "step": 24070
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6413686871528625,
      "learning_rate": 0.0005839859621702989,
      "loss": 2.4013,
      "step": 24071
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6097222566604614,
      "learning_rate": 0.0005839846435429871,
      "loss": 2.4145,
      "step": 24072
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.5764525532722473,
      "learning_rate": 0.0005839833248628773,
      "loss": 2.2298,
      "step": 24073
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.5998101830482483,
      "learning_rate": 0.0005839820061299695,
      "loss": 2.3587,
      "step": 24074
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.5818518400192261,
      "learning_rate": 0.0005839806873442642,
      "loss": 2.1851,
      "step": 24075
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.703863799571991,
      "learning_rate": 0.0005839793685057614,
      "loss": 2.494,
      "step": 24076
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6096221208572388,
      "learning_rate": 0.0005839780496144616,
      "loss": 2.1147,
      "step": 24077
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.5894100069999695,
      "learning_rate": 0.0005839767306703649,
      "loss": 2.1507,
      "step": 24078
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6874092817306519,
      "learning_rate": 0.0005839754116734714,
      "loss": 2.3131,
      "step": 24079
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6232205033302307,
      "learning_rate": 0.0005839740926237816,
      "loss": 2.474,
      "step": 24080
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.7277628183364868,
      "learning_rate": 0.0005839727735212957,
      "loss": 2.3654,
      "step": 24081
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6245390176773071,
      "learning_rate": 0.000583971454366014,
      "loss": 2.4213,
      "step": 24082
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.600496768951416,
      "learning_rate": 0.0005839701351579364,
      "loss": 2.4041,
      "step": 24083
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6185629367828369,
      "learning_rate": 0.0005839688158970637,
      "loss": 2.4475,
      "step": 24084
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6341404914855957,
      "learning_rate": 0.0005839674965833956,
      "loss": 2.3358,
      "step": 24085
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6821431517601013,
      "learning_rate": 0.0005839661772169326,
      "loss": 2.3229,
      "step": 24086
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.625043511390686,
      "learning_rate": 0.000583964857797675,
      "loss": 2.3944,
      "step": 24087
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6398951411247253,
      "learning_rate": 0.000583963538325623,
      "loss": 2.3643,
      "step": 24088
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6444312930107117,
      "learning_rate": 0.0005839622188007768,
      "loss": 2.3326,
      "step": 24089
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6766110062599182,
      "learning_rate": 0.0005839608992231366,
      "loss": 2.5303,
      "step": 24090
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.5828059911727905,
      "learning_rate": 0.0005839595795927027,
      "loss": 2.3048,
      "step": 24091
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.5815752744674683,
      "learning_rate": 0.0005839582599094754,
      "loss": 2.2661,
      "step": 24092
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6417850852012634,
      "learning_rate": 0.0005839569401734549,
      "loss": 2.4773,
      "step": 24093
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.5992148518562317,
      "learning_rate": 0.0005839556203846415,
      "loss": 2.351,
      "step": 24094
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6361143589019775,
      "learning_rate": 0.0005839543005430354,
      "loss": 2.2079,
      "step": 24095
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6254546642303467,
      "learning_rate": 0.0005839529806486367,
      "loss": 2.3359,
      "step": 24096
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.7353657484054565,
      "learning_rate": 0.0005839516607014459,
      "loss": 2.4877,
      "step": 24097
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.665008008480072,
      "learning_rate": 0.000583950340701463,
      "loss": 2.34,
      "step": 24098
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6558102965354919,
      "learning_rate": 0.0005839490206486884,
      "loss": 2.108,
      "step": 24099
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6252822279930115,
      "learning_rate": 0.0005839477005431225,
      "loss": 2.2596,
      "step": 24100
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6318053603172302,
      "learning_rate": 0.0005839463803847651,
      "loss": 2.3863,
      "step": 24101
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6692707538604736,
      "learning_rate": 0.000583945060173617,
      "loss": 2.344,
      "step": 24102
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6270720958709717,
      "learning_rate": 0.0005839437399096779,
      "loss": 2.4076,
      "step": 24103
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6651374697685242,
      "learning_rate": 0.0005839424195929483,
      "loss": 2.1723,
      "step": 24104
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6352524757385254,
      "learning_rate": 0.0005839410992234286,
      "loss": 2.3363,
      "step": 24105
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6666225790977478,
      "learning_rate": 0.0005839397788011188,
      "loss": 2.3387,
      "step": 24106
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6015021204948425,
      "learning_rate": 0.0005839384583260193,
      "loss": 2.1987,
      "step": 24107
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6518068313598633,
      "learning_rate": 0.0005839371377981302,
      "loss": 2.2028,
      "step": 24108
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6021158695220947,
      "learning_rate": 0.0005839358172174519,
      "loss": 2.5687,
      "step": 24109
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6376941204071045,
      "learning_rate": 0.0005839344965839846,
      "loss": 2.1275,
      "step": 24110
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6177117228507996,
      "learning_rate": 0.0005839331758977284,
      "loss": 2.2402,
      "step": 24111
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6320683360099792,
      "learning_rate": 0.0005839318551586837,
      "loss": 2.2864,
      "step": 24112
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.7656095027923584,
      "learning_rate": 0.0005839305343668507,
      "loss": 2.4221,
      "step": 24113
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.7010321617126465,
      "learning_rate": 0.0005839292135222297,
      "loss": 2.3112,
      "step": 24114
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6404857039451599,
      "learning_rate": 0.0005839278926248209,
      "loss": 2.4132,
      "step": 24115
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.5644763112068176,
      "learning_rate": 0.0005839265716746244,
      "loss": 2.0359,
      "step": 24116
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6099374294281006,
      "learning_rate": 0.0005839252506716408,
      "loss": 2.5017,
      "step": 24117
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6163784265518188,
      "learning_rate": 0.00058392392961587,
      "loss": 2.1925,
      "step": 24118
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6473762392997742,
      "learning_rate": 0.0005839226085073124,
      "loss": 2.4144,
      "step": 24119
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.634739100933075,
      "learning_rate": 0.0005839212873459684,
      "loss": 2.3662,
      "step": 24120
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6598541140556335,
      "learning_rate": 0.0005839199661318379,
      "loss": 2.4624,
      "step": 24121
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.5948499441146851,
      "learning_rate": 0.0005839186448649213,
      "loss": 2.2818,
      "step": 24122
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6729117631912231,
      "learning_rate": 0.0005839173235452189,
      "loss": 2.5232,
      "step": 24123
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6941917538642883,
      "learning_rate": 0.0005839160021727309,
      "loss": 2.4409,
      "step": 24124
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6756534576416016,
      "learning_rate": 0.0005839146807474577,
      "loss": 2.3066,
      "step": 24125
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6116777658462524,
      "learning_rate": 0.0005839133592693992,
      "loss": 2.2323,
      "step": 24126
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6377179622650146,
      "learning_rate": 0.000583912037738556,
      "loss": 2.2435,
      "step": 24127
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.8016166687011719,
      "learning_rate": 0.0005839107161549281,
      "loss": 2.4908,
      "step": 24128
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6111739277839661,
      "learning_rate": 0.0005839093945185158,
      "loss": 2.3086,
      "step": 24129
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.5946198105812073,
      "learning_rate": 0.0005839080728293195,
      "loss": 2.6405,
      "step": 24130
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6650694608688354,
      "learning_rate": 0.0005839067510873393,
      "loss": 1.904,
      "step": 24131
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6603953242301941,
      "learning_rate": 0.0005839054292925754,
      "loss": 2.3754,
      "step": 24132
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.9018610715866089,
      "learning_rate": 0.0005839041074450282,
      "loss": 2.3939,
      "step": 24133
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6520094275474548,
      "learning_rate": 0.0005839027855446978,
      "loss": 2.5455,
      "step": 24134
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6367055773735046,
      "learning_rate": 0.0005839014635915846,
      "loss": 2.194,
      "step": 24135
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.7693650722503662,
      "learning_rate": 0.0005839001415856887,
      "loss": 2.2415,
      "step": 24136
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.8962950706481934,
      "learning_rate": 0.0005838988195270104,
      "loss": 2.3404,
      "step": 24137
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.656048059463501,
      "learning_rate": 0.0005838974974155499,
      "loss": 2.3127,
      "step": 24138
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.7306256294250488,
      "learning_rate": 0.0005838961752513076,
      "loss": 2.5395,
      "step": 24139
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.9392040371894836,
      "learning_rate": 0.0005838948530342837,
      "loss": 2.3925,
      "step": 24140
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6334867477416992,
      "learning_rate": 0.0005838935307644782,
      "loss": 2.3432,
      "step": 24141
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.7006402611732483,
      "learning_rate": 0.0005838922084418917,
      "loss": 2.4632,
      "step": 24142
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.9717079997062683,
      "learning_rate": 0.0005838908860665242,
      "loss": 2.4205,
      "step": 24143
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.7484990954399109,
      "learning_rate": 0.0005838895636383761,
      "loss": 2.307,
      "step": 24144
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6709509491920471,
      "learning_rate": 0.0005838882411574474,
      "loss": 2.0714,
      "step": 24145
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.915147066116333,
      "learning_rate": 0.0005838869186237386,
      "loss": 2.1371,
      "step": 24146
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.805106520652771,
      "learning_rate": 0.00058388559603725,
      "loss": 2.3973,
      "step": 24147
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.8049954175949097,
      "learning_rate": 0.0005838842733979816,
      "loss": 2.1817,
      "step": 24148
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6170265078544617,
      "learning_rate": 0.0005838829507059338,
      "loss": 2.57,
      "step": 24149
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.0935413837432861,
      "learning_rate": 0.0005838816279611066,
      "loss": 2.4704,
      "step": 24150
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6730209589004517,
      "learning_rate": 0.0005838803051635006,
      "loss": 2.1398,
      "step": 24151
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.7115151882171631,
      "learning_rate": 0.0005838789823131159,
      "loss": 2.4367,
      "step": 24152
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.7175822854042053,
      "learning_rate": 0.0005838776594099528,
      "loss": 2.4239,
      "step": 24153
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.692241370677948,
      "learning_rate": 0.0005838763364540113,
      "loss": 2.2274,
      "step": 24154
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.7541492581367493,
      "learning_rate": 0.0005838750134452921,
      "loss": 2.4898,
      "step": 24155
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6499146819114685,
      "learning_rate": 0.0005838736903837949,
      "loss": 2.3203,
      "step": 24156
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6166585683822632,
      "learning_rate": 0.0005838723672695204,
      "loss": 2.4085,
      "step": 24157
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.7281532883644104,
      "learning_rate": 0.0005838710441024686,
      "loss": 2.5068,
      "step": 24158
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6376699209213257,
      "learning_rate": 0.0005838697208826398,
      "loss": 2.3286,
      "step": 24159
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6195692420005798,
      "learning_rate": 0.0005838683976100343,
      "loss": 2.4292,
      "step": 24160
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.549062967300415,
      "learning_rate": 0.0005838670742846523,
      "loss": 2.2935,
      "step": 24161
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.5798742771148682,
      "learning_rate": 0.000583865750906494,
      "loss": 2.247,
      "step": 24162
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6235165596008301,
      "learning_rate": 0.0005838644274755597,
      "loss": 2.1349,
      "step": 24163
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6543322205543518,
      "learning_rate": 0.0005838631039918497,
      "loss": 2.5468,
      "step": 24164
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6201812028884888,
      "learning_rate": 0.0005838617804553642,
      "loss": 2.2054,
      "step": 24165
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.5759468674659729,
      "learning_rate": 0.0005838604568661034,
      "loss": 2.229,
      "step": 24166
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.5694968104362488,
      "learning_rate": 0.0005838591332240677,
      "loss": 2.1366,
      "step": 24167
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6987345814704895,
      "learning_rate": 0.0005838578095292571,
      "loss": 2.4374,
      "step": 24168
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.671451985836029,
      "learning_rate": 0.0005838564857816719,
      "loss": 2.3264,
      "step": 24169
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6112551093101501,
      "learning_rate": 0.0005838551619813126,
      "loss": 2.3397,
      "step": 24170
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6457622051239014,
      "learning_rate": 0.0005838538381281792,
      "loss": 2.3972,
      "step": 24171
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6714969277381897,
      "learning_rate": 0.000583852514222272,
      "loss": 2.5217,
      "step": 24172
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.639176607131958,
      "learning_rate": 0.0005838511902635913,
      "loss": 2.3773,
      "step": 24173
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.644372284412384,
      "learning_rate": 0.0005838498662521372,
      "loss": 2.5972,
      "step": 24174
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6238730549812317,
      "learning_rate": 0.0005838485421879101,
      "loss": 2.041,
      "step": 24175
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6724796891212463,
      "learning_rate": 0.0005838472180709103,
      "loss": 2.3363,
      "step": 24176
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.5887683629989624,
      "learning_rate": 0.0005838458939011378,
      "loss": 2.29,
      "step": 24177
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6195729970932007,
      "learning_rate": 0.000583844569678593,
      "loss": 2.4223,
      "step": 24178
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6601974368095398,
      "learning_rate": 0.0005838432454032763,
      "loss": 2.384,
      "step": 24179
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6181919574737549,
      "learning_rate": 0.0005838419210751876,
      "loss": 2.5319,
      "step": 24180
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.5997921228408813,
      "learning_rate": 0.0005838405966943274,
      "loss": 2.2903,
      "step": 24181
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6477270722389221,
      "learning_rate": 0.0005838392722606958,
      "loss": 2.3048,
      "step": 24182
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6334503293037415,
      "learning_rate": 0.0005838379477742933,
      "loss": 2.3933,
      "step": 24183
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6495792269706726,
      "learning_rate": 0.0005838366232351199,
      "loss": 2.3857,
      "step": 24184
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6205026507377625,
      "learning_rate": 0.0005838352986431758,
      "loss": 2.5029,
      "step": 24185
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6095507144927979,
      "learning_rate": 0.0005838339739984614,
      "loss": 2.0563,
      "step": 24186
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6381425857543945,
      "learning_rate": 0.000583832649300977,
      "loss": 2.3953,
      "step": 24187
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.5847908854484558,
      "learning_rate": 0.0005838313245507227,
      "loss": 2.2791,
      "step": 24188
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6420064568519592,
      "learning_rate": 0.0005838299997476988,
      "loss": 2.5208,
      "step": 24189
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6789341568946838,
      "learning_rate": 0.0005838286748919056,
      "loss": 2.4363,
      "step": 24190
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6872968673706055,
      "learning_rate": 0.0005838273499833432,
      "loss": 2.5666,
      "step": 24191
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.5654913187026978,
      "learning_rate": 0.000583826025022012,
      "loss": 2.16,
      "step": 24192
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6438478827476501,
      "learning_rate": 0.000583824700007912,
      "loss": 2.3099,
      "step": 24193
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6228921413421631,
      "learning_rate": 0.0005838233749410438,
      "loss": 2.2637,
      "step": 24194
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6085297465324402,
      "learning_rate": 0.0005838220498214075,
      "loss": 2.268,
      "step": 24195
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6344279050827026,
      "learning_rate": 0.0005838207246490033,
      "loss": 2.2552,
      "step": 24196
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.705483078956604,
      "learning_rate": 0.0005838193994238314,
      "loss": 2.322,
      "step": 24197
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6614872813224792,
      "learning_rate": 0.0005838180741458922,
      "loss": 2.4982,
      "step": 24198
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6785485148429871,
      "learning_rate": 0.0005838167488151859,
      "loss": 2.2552,
      "step": 24199
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6695261001586914,
      "learning_rate": 0.0005838154234317126,
      "loss": 2.3838,
      "step": 24200
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6235082745552063,
      "learning_rate": 0.0005838140979954727,
      "loss": 2.2891,
      "step": 24201
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6396263241767883,
      "learning_rate": 0.0005838127725064665,
      "loss": 2.2078,
      "step": 24202
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.577233612537384,
      "learning_rate": 0.000583811446964694,
      "loss": 2.3212,
      "step": 24203
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6429069638252258,
      "learning_rate": 0.0005838101213701556,
      "loss": 2.5387,
      "step": 24204
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.660772979259491,
      "learning_rate": 0.0005838087957228517,
      "loss": 2.2323,
      "step": 24205
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6847135424613953,
      "learning_rate": 0.0005838074700227822,
      "loss": 2.3649,
      "step": 24206
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6446236968040466,
      "learning_rate": 0.0005838061442699476,
      "loss": 2.3418,
      "step": 24207
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6788621544837952,
      "learning_rate": 0.0005838048184643481,
      "loss": 2.3938,
      "step": 24208
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6247130632400513,
      "learning_rate": 0.0005838034926059839,
      "loss": 2.1623,
      "step": 24209
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.646306574344635,
      "learning_rate": 0.0005838021666948553,
      "loss": 2.59,
      "step": 24210
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.5998944044113159,
      "learning_rate": 0.0005838008407309625,
      "loss": 2.3227,
      "step": 24211
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.604511559009552,
      "learning_rate": 0.0005837995147143058,
      "loss": 2.0989,
      "step": 24212
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.5991301536560059,
      "learning_rate": 0.0005837981886448854,
      "loss": 2.5929,
      "step": 24213
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6045476198196411,
      "learning_rate": 0.0005837968625227015,
      "loss": 2.1779,
      "step": 24214
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6140270233154297,
      "learning_rate": 0.0005837955363477544,
      "loss": 2.3989,
      "step": 24215
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6389384865760803,
      "learning_rate": 0.0005837942101200444,
      "loss": 2.4377,
      "step": 24216
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6121140122413635,
      "learning_rate": 0.0005837928838395717,
      "loss": 2.5316,
      "step": 24217
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.5916282534599304,
      "learning_rate": 0.0005837915575063366,
      "loss": 2.2166,
      "step": 24218
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.62519770860672,
      "learning_rate": 0.0005837902311203393,
      "loss": 2.2712,
      "step": 24219
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.5787196159362793,
      "learning_rate": 0.00058378890468158,
      "loss": 2.2627,
      "step": 24220
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6506495475769043,
      "learning_rate": 0.000583787578190059,
      "loss": 2.2413,
      "step": 24221
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.572736382484436,
      "learning_rate": 0.0005837862516457764,
      "loss": 2.3467,
      "step": 24222
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.5682388544082642,
      "learning_rate": 0.0005837849250487327,
      "loss": 2.246,
      "step": 24223
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6336462497711182,
      "learning_rate": 0.000583783598398928,
      "loss": 2.4301,
      "step": 24224
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6542946100234985,
      "learning_rate": 0.0005837822716963626,
      "loss": 2.3582,
      "step": 24225
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6701659560203552,
      "learning_rate": 0.0005837809449410367,
      "loss": 2.2945,
      "step": 24226
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6183397173881531,
      "learning_rate": 0.0005837796181329505,
      "loss": 2.2656,
      "step": 24227
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6444987058639526,
      "learning_rate": 0.0005837782912721044,
      "loss": 2.2279,
      "step": 24228
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6334666013717651,
      "learning_rate": 0.0005837769643584985,
      "loss": 2.3793,
      "step": 24229
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6431203484535217,
      "learning_rate": 0.0005837756373921331,
      "loss": 2.2401,
      "step": 24230
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6071122288703918,
      "learning_rate": 0.0005837743103730086,
      "loss": 2.4011,
      "step": 24231
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6408738493919373,
      "learning_rate": 0.000583772983301125,
      "loss": 2.3643,
      "step": 24232
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.5882686972618103,
      "learning_rate": 0.0005837716561764826,
      "loss": 2.4493,
      "step": 24233
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6222311854362488,
      "learning_rate": 0.0005837703289990817,
      "loss": 2.2823,
      "step": 24234
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6611345410346985,
      "learning_rate": 0.0005837690017689226,
      "loss": 2.4867,
      "step": 24235
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6160606145858765,
      "learning_rate": 0.0005837676744860055,
      "loss": 2.2183,
      "step": 24236
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6095678806304932,
      "learning_rate": 0.0005837663471503305,
      "loss": 2.2018,
      "step": 24237
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.604823648929596,
      "learning_rate": 0.000583765019761898,
      "loss": 2.4523,
      "step": 24238
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6680428385734558,
      "learning_rate": 0.0005837636923207084,
      "loss": 2.281,
      "step": 24239
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.565534234046936,
      "learning_rate": 0.0005837623648267616,
      "loss": 2.286,
      "step": 24240
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6157010197639465,
      "learning_rate": 0.0005837610372800581,
      "loss": 2.2548,
      "step": 24241
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6489158868789673,
      "learning_rate": 0.0005837597096805981,
      "loss": 2.4808,
      "step": 24242
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6142208576202393,
      "learning_rate": 0.0005837583820283817,
      "loss": 2.1552,
      "step": 24243
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.5770255923271179,
      "learning_rate": 0.0005837570543234093,
      "loss": 2.2202,
      "step": 24244
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.673530638217926,
      "learning_rate": 0.0005837557265656811,
      "loss": 2.2018,
      "step": 24245
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6625780463218689,
      "learning_rate": 0.0005837543987551975,
      "loss": 2.488,
      "step": 24246
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6718377470970154,
      "learning_rate": 0.0005837530708919585,
      "loss": 2.5207,
      "step": 24247
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.5870451927185059,
      "learning_rate": 0.0005837517429759645,
      "loss": 2.3213,
      "step": 24248
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6273689866065979,
      "learning_rate": 0.0005837504150072156,
      "loss": 2.4987,
      "step": 24249
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.697260320186615,
      "learning_rate": 0.0005837490869857122,
      "loss": 2.3869,
      "step": 24250
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6953146457672119,
      "learning_rate": 0.0005837477589114544,
      "loss": 2.1787,
      "step": 24251
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6677613258361816,
      "learning_rate": 0.0005837464307844425,
      "loss": 2.3116,
      "step": 24252
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.5920993089675903,
      "learning_rate": 0.0005837451026046769,
      "loss": 2.3883,
      "step": 24253
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6415932774543762,
      "learning_rate": 0.0005837437743721577,
      "loss": 2.4637,
      "step": 24254
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6775728464126587,
      "learning_rate": 0.0005837424460868854,
      "loss": 2.6012,
      "step": 24255
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.5868790745735168,
      "learning_rate": 0.0005837411177488596,
      "loss": 2.2104,
      "step": 24256
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6477880477905273,
      "learning_rate": 0.0005837397893580813,
      "loss": 2.3579,
      "step": 24257
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6822900772094727,
      "learning_rate": 0.0005837384609145502,
      "loss": 2.3462,
      "step": 24258
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6330322027206421,
      "learning_rate": 0.0005837371324182669,
      "loss": 2.0561,
      "step": 24259
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6048722267150879,
      "learning_rate": 0.0005837358038692315,
      "loss": 2.101,
      "step": 24260
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6285939812660217,
      "learning_rate": 0.0005837344752674441,
      "loss": 2.3567,
      "step": 24261
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.5679210424423218,
      "learning_rate": 0.0005837331466129054,
      "loss": 2.2741,
      "step": 24262
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6361525654792786,
      "learning_rate": 0.0005837318179056151,
      "loss": 2.5736,
      "step": 24263
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.7745077013969421,
      "learning_rate": 0.0005837304891455738,
      "loss": 2.1535,
      "step": 24264
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6438892483711243,
      "learning_rate": 0.0005837291603327816,
      "loss": 2.3533,
      "step": 24265
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6215096116065979,
      "learning_rate": 0.0005837278314672388,
      "loss": 2.4267,
      "step": 24266
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6052409410476685,
      "learning_rate": 0.0005837265025489457,
      "loss": 2.2428,
      "step": 24267
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6090896129608154,
      "learning_rate": 0.0005837251735779025,
      "loss": 2.3085,
      "step": 24268
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6059294939041138,
      "learning_rate": 0.0005837238445541093,
      "loss": 2.2955,
      "step": 24269
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.5883268713951111,
      "learning_rate": 0.0005837225154775666,
      "loss": 2.2832,
      "step": 24270
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6901974678039551,
      "learning_rate": 0.0005837211863482745,
      "loss": 2.5624,
      "step": 24271
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6313353776931763,
      "learning_rate": 0.0005837198571662333,
      "loss": 2.4209,
      "step": 24272
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6299301385879517,
      "learning_rate": 0.0005837185279314432,
      "loss": 2.2433,
      "step": 24273
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6852782368659973,
      "learning_rate": 0.0005837171986439045,
      "loss": 2.4193,
      "step": 24274
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6166853308677673,
      "learning_rate": 0.0005837158693036173,
      "loss": 2.1795,
      "step": 24275
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6337915658950806,
      "learning_rate": 0.0005837145399105821,
      "loss": 2.3124,
      "step": 24276
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.5571111440658569,
      "learning_rate": 0.000583713210464799,
      "loss": 2.1055,
      "step": 24277
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6104053854942322,
      "learning_rate": 0.0005837118809662681,
      "loss": 2.3972,
      "step": 24278
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6980929970741272,
      "learning_rate": 0.00058371055141499,
      "loss": 2.3194,
      "step": 24279
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6520057320594788,
      "learning_rate": 0.0005837092218109647,
      "loss": 2.2177,
      "step": 24280
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6826642155647278,
      "learning_rate": 0.0005837078921541925,
      "loss": 2.4385,
      "step": 24281
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6190551519393921,
      "learning_rate": 0.0005837065624446735,
      "loss": 2.2449,
      "step": 24282
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6743213534355164,
      "learning_rate": 0.0005837052326824083,
      "loss": 2.1977,
      "step": 24283
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.5916023850440979,
      "learning_rate": 0.0005837039028673969,
      "loss": 2.3086,
      "step": 24284
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6227591037750244,
      "learning_rate": 0.0005837025729996395,
      "loss": 2.2733,
      "step": 24285
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6175629496574402,
      "learning_rate": 0.0005837012430791365,
      "loss": 2.4065,
      "step": 24286
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6559593677520752,
      "learning_rate": 0.0005836999131058881,
      "loss": 2.313,
      "step": 24287
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6378793120384216,
      "learning_rate": 0.0005836985830798945,
      "loss": 2.401,
      "step": 24288
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6977820992469788,
      "learning_rate": 0.0005836972530011559,
      "loss": 2.4324,
      "step": 24289
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6055286526679993,
      "learning_rate": 0.0005836959228696728,
      "loss": 2.1809,
      "step": 24290
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.637347936630249,
      "learning_rate": 0.000583694592685445,
      "loss": 2.2342,
      "step": 24291
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6170509457588196,
      "learning_rate": 0.0005836932624484734,
      "loss": 2.3017,
      "step": 24292
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.7108365297317505,
      "learning_rate": 0.0005836919321587576,
      "loss": 2.2426,
      "step": 24293
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6450469493865967,
      "learning_rate": 0.0005836906018162982,
      "loss": 2.5879,
      "step": 24294
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6594661474227905,
      "learning_rate": 0.0005836892714210953,
      "loss": 2.3458,
      "step": 24295
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.5897042155265808,
      "learning_rate": 0.0005836879409731493,
      "loss": 2.3797,
      "step": 24296
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6639398336410522,
      "learning_rate": 0.0005836866104724603,
      "loss": 2.6294,
      "step": 24297
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.618378221988678,
      "learning_rate": 0.0005836852799190286,
      "loss": 2.4919,
      "step": 24298
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6413694024085999,
      "learning_rate": 0.0005836839493128544,
      "loss": 2.1831,
      "step": 24299
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.5969457030296326,
      "learning_rate": 0.0005836826186539381,
      "loss": 2.2292,
      "step": 24300
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.687886118888855,
      "learning_rate": 0.0005836812879422799,
      "loss": 2.0352,
      "step": 24301
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6512519121170044,
      "learning_rate": 0.0005836799571778799,
      "loss": 2.6114,
      "step": 24302
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6513792276382446,
      "learning_rate": 0.0005836786263607384,
      "loss": 2.5213,
      "step": 24303
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6293739676475525,
      "learning_rate": 0.0005836772954908557,
      "loss": 2.4508,
      "step": 24304
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.7211278676986694,
      "learning_rate": 0.0005836759645682321,
      "loss": 2.147,
      "step": 24305
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.687088668346405,
      "learning_rate": 0.0005836746335928678,
      "loss": 2.4083,
      "step": 24306
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.594334065914154,
      "learning_rate": 0.0005836733025647629,
      "loss": 2.3253,
      "step": 24307
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6299303770065308,
      "learning_rate": 0.0005836719714839178,
      "loss": 2.19,
      "step": 24308
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6174908876419067,
      "learning_rate": 0.0005836706403503329,
      "loss": 2.245,
      "step": 24309
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6661426424980164,
      "learning_rate": 0.0005836693091640081,
      "loss": 2.2756,
      "step": 24310
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.5993409752845764,
      "learning_rate": 0.0005836679779249439,
      "loss": 2.3075,
      "step": 24311
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.5513662695884705,
      "learning_rate": 0.0005836666466331404,
      "loss": 2.2164,
      "step": 24312
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6403467655181885,
      "learning_rate": 0.000583665315288598,
      "loss": 2.1546,
      "step": 24313
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.628929615020752,
      "learning_rate": 0.0005836639838913168,
      "loss": 2.1765,
      "step": 24314
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.5964713096618652,
      "learning_rate": 0.0005836626524412972,
      "loss": 2.3704,
      "step": 24315
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6415290832519531,
      "learning_rate": 0.0005836613209385393,
      "loss": 2.1865,
      "step": 24316
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6793166995048523,
      "learning_rate": 0.0005836599893830434,
      "loss": 2.4302,
      "step": 24317
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6212486624717712,
      "learning_rate": 0.0005836586577748097,
      "loss": 2.2745,
      "step": 24318
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6615061163902283,
      "learning_rate": 0.0005836573261138386,
      "loss": 2.3441,
      "step": 24319
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6016382575035095,
      "learning_rate": 0.0005836559944001303,
      "loss": 2.3101,
      "step": 24320
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.7573739290237427,
      "learning_rate": 0.0005836546626336849,
      "loss": 2.3072,
      "step": 24321
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6944015026092529,
      "learning_rate": 0.0005836533308145026,
      "loss": 2.231,
      "step": 24322
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6664083003997803,
      "learning_rate": 0.000583651998942584,
      "loss": 2.2546,
      "step": 24323
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.5999226570129395,
      "learning_rate": 0.0005836506670179291,
      "loss": 2.155,
      "step": 24324
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.632408857345581,
      "learning_rate": 0.0005836493350405382,
      "loss": 2.3302,
      "step": 24325
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6315432190895081,
      "learning_rate": 0.0005836480030104115,
      "loss": 2.5621,
      "step": 24326
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.5631446838378906,
      "learning_rate": 0.0005836466709275491,
      "loss": 2.26,
      "step": 24327
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6270554065704346,
      "learning_rate": 0.0005836453387919518,
      "loss": 2.3478,
      "step": 24328
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6549839973449707,
      "learning_rate": 0.0005836440066036192,
      "loss": 2.5374,
      "step": 24329
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.5990826487541199,
      "learning_rate": 0.0005836426743625519,
      "loss": 2.2085,
      "step": 24330
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6070649027824402,
      "learning_rate": 0.00058364134206875,
      "loss": 2.261,
      "step": 24331
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6856454014778137,
      "learning_rate": 0.000583640009722214,
      "loss": 2.3186,
      "step": 24332
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.7216841578483582,
      "learning_rate": 0.0005836386773229439,
      "loss": 2.2577,
      "step": 24333
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6416621208190918,
      "learning_rate": 0.00058363734487094,
      "loss": 2.3942,
      "step": 24334
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6089155077934265,
      "learning_rate": 0.0005836360123662025,
      "loss": 2.2744,
      "step": 24335
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6259613633155823,
      "learning_rate": 0.0005836346798087317,
      "loss": 2.2259,
      "step": 24336
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6538369059562683,
      "learning_rate": 0.000583633347198528,
      "loss": 2.5398,
      "step": 24337
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.7739297747612,
      "learning_rate": 0.0005836320145355914,
      "loss": 2.4615,
      "step": 24338
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.61020827293396,
      "learning_rate": 0.0005836306818199224,
      "loss": 2.3724,
      "step": 24339
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6318249106407166,
      "learning_rate": 0.0005836293490515209,
      "loss": 2.1665,
      "step": 24340
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6839147806167603,
      "learning_rate": 0.0005836280162303874,
      "loss": 2.3777,
      "step": 24341
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.633589506149292,
      "learning_rate": 0.0005836266833565222,
      "loss": 2.1755,
      "step": 24342
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.5916553735733032,
      "learning_rate": 0.0005836253504299254,
      "loss": 2.3914,
      "step": 24343
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.7058851718902588,
      "learning_rate": 0.0005836240174505972,
      "loss": 2.3062,
      "step": 24344
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6708608865737915,
      "learning_rate": 0.000583622684418538,
      "loss": 2.3849,
      "step": 24345
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.7023098468780518,
      "learning_rate": 0.000583621351333748,
      "loss": 2.4132,
      "step": 24346
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6685280799865723,
      "learning_rate": 0.0005836200181962275,
      "loss": 2.1897,
      "step": 24347
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.8995210528373718,
      "learning_rate": 0.0005836186850059766,
      "loss": 2.5847,
      "step": 24348
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.666225790977478,
      "learning_rate": 0.0005836173517629957,
      "loss": 2.6462,
      "step": 24349
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6854593753814697,
      "learning_rate": 0.0005836160184672849,
      "loss": 2.3743,
      "step": 24350
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.8826635479927063,
      "learning_rate": 0.0005836146851188446,
      "loss": 2.2848,
      "step": 24351
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6348151564598083,
      "learning_rate": 0.0005836133517176749,
      "loss": 2.3171,
      "step": 24352
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6562252044677734,
      "learning_rate": 0.0005836120182637762,
      "loss": 2.3313,
      "step": 24353
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.7229725122451782,
      "learning_rate": 0.0005836106847571486,
      "loss": 2.398,
      "step": 24354
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6830334663391113,
      "learning_rate": 0.0005836093511977924,
      "loss": 2.2799,
      "step": 24355
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6332049369812012,
      "learning_rate": 0.0005836080175857081,
      "loss": 2.2387,
      "step": 24356
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6257025599479675,
      "learning_rate": 0.0005836066839208955,
      "loss": 2.2933,
      "step": 24357
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.921592652797699,
      "learning_rate": 0.0005836053502033551,
      "loss": 2.2129,
      "step": 24358
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6522373557090759,
      "learning_rate": 0.000583604016433087,
      "loss": 2.4515,
      "step": 24359
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6394425630569458,
      "learning_rate": 0.0005836026826100917,
      "loss": 2.237,
      "step": 24360
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6228806376457214,
      "learning_rate": 0.0005836013487343694,
      "loss": 2.3471,
      "step": 24361
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.8929717540740967,
      "learning_rate": 0.0005836000148059201,
      "loss": 2.2006,
      "step": 24362
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.5995067358016968,
      "learning_rate": 0.0005835986808247442,
      "loss": 2.2638,
      "step": 24363
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.7271974086761475,
      "learning_rate": 0.0005835973467908421,
      "loss": 2.3749,
      "step": 24364
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.7691770195960999,
      "learning_rate": 0.0005835960127042137,
      "loss": 2.3548,
      "step": 24365
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.632544755935669,
      "learning_rate": 0.0005835946785648596,
      "loss": 2.4027,
      "step": 24366
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.7258260846138,
      "learning_rate": 0.0005835933443727798,
      "loss": 2.4656,
      "step": 24367
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6483359932899475,
      "learning_rate": 0.0005835920101279747,
      "loss": 2.3631,
      "step": 24368
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.7224524617195129,
      "learning_rate": 0.0005835906758304445,
      "loss": 2.3656,
      "step": 24369
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6774787902832031,
      "learning_rate": 0.0005835893414801894,
      "loss": 2.3703,
      "step": 24370
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6881345510482788,
      "learning_rate": 0.0005835880070772097,
      "loss": 2.3556,
      "step": 24371
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.8760215640068054,
      "learning_rate": 0.0005835866726215057,
      "loss": 2.4354,
      "step": 24372
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6332786679267883,
      "learning_rate": 0.0005835853381130774,
      "loss": 2.3989,
      "step": 24373
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6557115316390991,
      "learning_rate": 0.0005835840035519255,
      "loss": 2.4891,
      "step": 24374
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6950598955154419,
      "learning_rate": 0.0005835826689380498,
      "loss": 2.5585,
      "step": 24375
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6982450485229492,
      "learning_rate": 0.0005835813342714507,
      "loss": 2.4285,
      "step": 24376
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6028584241867065,
      "learning_rate": 0.0005835799995521287,
      "loss": 2.192,
      "step": 24377
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.7061815857887268,
      "learning_rate": 0.0005835786647800836,
      "loss": 2.2107,
      "step": 24378
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6174936890602112,
      "learning_rate": 0.0005835773299553158,
      "loss": 2.1682,
      "step": 24379
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.7213394045829773,
      "learning_rate": 0.0005835759950778258,
      "loss": 2.3375,
      "step": 24380
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6132155656814575,
      "learning_rate": 0.0005835746601476138,
      "loss": 2.392,
      "step": 24381
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.7216672301292419,
      "learning_rate": 0.0005835733251646797,
      "loss": 2.3523,
      "step": 24382
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6697679162025452,
      "learning_rate": 0.000583571990129024,
      "loss": 2.5383,
      "step": 24383
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6323404312133789,
      "learning_rate": 0.000583570655040647,
      "loss": 2.2272,
      "step": 24384
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6854894757270813,
      "learning_rate": 0.0005835693198995487,
      "loss": 2.3962,
      "step": 24385
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6502081155776978,
      "learning_rate": 0.0005835679847057296,
      "loss": 2.2928,
      "step": 24386
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6367278695106506,
      "learning_rate": 0.00058356664945919,
      "loss": 2.3418,
      "step": 24387
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6426449418067932,
      "learning_rate": 0.0005835653141599298,
      "loss": 2.5203,
      "step": 24388
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6946721076965332,
      "learning_rate": 0.0005835639788079496,
      "loss": 2.3771,
      "step": 24389
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.697333037853241,
      "learning_rate": 0.0005835626434032493,
      "loss": 2.2691,
      "step": 24390
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.7127047181129456,
      "learning_rate": 0.0005835613079458296,
      "loss": 2.4224,
      "step": 24391
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6522172093391418,
      "learning_rate": 0.0005835599724356903,
      "loss": 2.2854,
      "step": 24392
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6068823933601379,
      "learning_rate": 0.000583558636872832,
      "loss": 2.2628,
      "step": 24393
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6187616586685181,
      "learning_rate": 0.0005835573012572547,
      "loss": 2.3485,
      "step": 24394
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.635007917881012,
      "learning_rate": 0.0005835559655889588,
      "loss": 2.3667,
      "step": 24395
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.5931081175804138,
      "learning_rate": 0.0005835546298679445,
      "loss": 2.1983,
      "step": 24396
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.575262188911438,
      "learning_rate": 0.0005835532940942119,
      "loss": 2.2136,
      "step": 24397
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.7171892523765564,
      "learning_rate": 0.0005835519582677616,
      "loss": 2.3646,
      "step": 24398
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6188459396362305,
      "learning_rate": 0.0005835506223885935,
      "loss": 2.3416,
      "step": 24399
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6373037099838257,
      "learning_rate": 0.000583549286456708,
      "loss": 2.2234,
      "step": 24400
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6813783645629883,
      "learning_rate": 0.0005835479504721053,
      "loss": 2.2374,
      "step": 24401
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6725174188613892,
      "learning_rate": 0.0005835466144347858,
      "loss": 2.2425,
      "step": 24402
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6901754140853882,
      "learning_rate": 0.0005835452783447496,
      "loss": 2.3909,
      "step": 24403
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6110143065452576,
      "learning_rate": 0.0005835439422019969,
      "loss": 2.418,
      "step": 24404
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.656880259513855,
      "learning_rate": 0.0005835426060065281,
      "loss": 2.398,
      "step": 24405
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.606255292892456,
      "learning_rate": 0.0005835412697583434,
      "loss": 2.3347,
      "step": 24406
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6826657652854919,
      "learning_rate": 0.000583539933457443,
      "loss": 2.5407,
      "step": 24407
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.5801865458488464,
      "learning_rate": 0.0005835385971038272,
      "loss": 2.2344,
      "step": 24408
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.5993055105209351,
      "learning_rate": 0.0005835372606974961,
      "loss": 2.2597,
      "step": 24409
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6988961696624756,
      "learning_rate": 0.0005835359242384501,
      "loss": 2.4476,
      "step": 24410
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6896149516105652,
      "learning_rate": 0.0005835345877266896,
      "loss": 2.385,
      "step": 24411
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.7252143025398254,
      "learning_rate": 0.0005835332511622145,
      "loss": 2.4793,
      "step": 24412
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6259768009185791,
      "learning_rate": 0.0005835319145450253,
      "loss": 2.3149,
      "step": 24413
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6195605993270874,
      "learning_rate": 0.0005835305778751221,
      "loss": 2.0479,
      "step": 24414
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6329002380371094,
      "learning_rate": 0.0005835292411525052,
      "loss": 2.1364,
      "step": 24415
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6192231774330139,
      "learning_rate": 0.000583527904377175,
      "loss": 2.3909,
      "step": 24416
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6632577180862427,
      "learning_rate": 0.0005835265675491315,
      "loss": 2.25,
      "step": 24417
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6248432397842407,
      "learning_rate": 0.000583525230668375,
      "loss": 2.437,
      "step": 24418
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6469014883041382,
      "learning_rate": 0.0005835238937349058,
      "loss": 2.4819,
      "step": 24419
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6115742921829224,
      "learning_rate": 0.0005835225567487243,
      "loss": 2.223,
      "step": 24420
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.5979347825050354,
      "learning_rate": 0.0005835212197098307,
      "loss": 2.3265,
      "step": 24421
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.5958656668663025,
      "learning_rate": 0.0005835198826182249,
      "loss": 2.266,
      "step": 24422
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6047744750976562,
      "learning_rate": 0.0005835185454739076,
      "loss": 2.2341,
      "step": 24423
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.7170195579528809,
      "learning_rate": 0.0005835172082768786,
      "loss": 2.3643,
      "step": 24424
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6648361682891846,
      "learning_rate": 0.0005835158710271386,
      "loss": 2.5024,
      "step": 24425
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6290650963783264,
      "learning_rate": 0.0005835145337246875,
      "loss": 2.3489,
      "step": 24426
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6250032186508179,
      "learning_rate": 0.0005835131963695258,
      "loss": 2.4768,
      "step": 24427
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6690031886100769,
      "learning_rate": 0.0005835118589616536,
      "loss": 2.4859,
      "step": 24428
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.5967712998390198,
      "learning_rate": 0.0005835105215010714,
      "loss": 2.189,
      "step": 24429
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6161264181137085,
      "learning_rate": 0.000583509183987779,
      "loss": 2.5175,
      "step": 24430
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6079670190811157,
      "learning_rate": 0.000583507846421777,
      "loss": 2.3821,
      "step": 24431
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6212205290794373,
      "learning_rate": 0.0005835065088030655,
      "loss": 2.3907,
      "step": 24432
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6519315838813782,
      "learning_rate": 0.0005835051711316448,
      "loss": 2.2955,
      "step": 24433
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6268728375434875,
      "learning_rate": 0.0005835038334075151,
      "loss": 2.4824,
      "step": 24434
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6328234672546387,
      "learning_rate": 0.0005835024956306766,
      "loss": 2.2791,
      "step": 24435
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.628210186958313,
      "learning_rate": 0.0005835011578011299,
      "loss": 2.1702,
      "step": 24436
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6600956320762634,
      "learning_rate": 0.0005834998199188748,
      "loss": 2.5308,
      "step": 24437
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.5878975987434387,
      "learning_rate": 0.0005834984819839118,
      "loss": 2.3252,
      "step": 24438
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.693385660648346,
      "learning_rate": 0.0005834971439962409,
      "loss": 2.2436,
      "step": 24439
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6583733558654785,
      "learning_rate": 0.0005834958059558628,
      "loss": 2.3336,
      "step": 24440
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6856179237365723,
      "learning_rate": 0.0005834944678627773,
      "loss": 2.3425,
      "step": 24441
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.8157281279563904,
      "learning_rate": 0.0005834931297169849,
      "loss": 2.3522,
      "step": 24442
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6409434080123901,
      "learning_rate": 0.0005834917915184857,
      "loss": 2.3584,
      "step": 24443
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.5981801748275757,
      "learning_rate": 0.00058349045326728,
      "loss": 2.1639,
      "step": 24444
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.7172802686691284,
      "learning_rate": 0.0005834891149633682,
      "loss": 2.2847,
      "step": 24445
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.5986862182617188,
      "learning_rate": 0.0005834877766067504,
      "loss": 2.3622,
      "step": 24446
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6088824272155762,
      "learning_rate": 0.0005834864381974268,
      "loss": 2.4269,
      "step": 24447
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.5858023166656494,
      "learning_rate": 0.0005834850997353979,
      "loss": 2.4935,
      "step": 24448
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6657828688621521,
      "learning_rate": 0.0005834837612206636,
      "loss": 2.7348,
      "step": 24449
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.5689647197723389,
      "learning_rate": 0.0005834824226532243,
      "loss": 2.1717,
      "step": 24450
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6770444512367249,
      "learning_rate": 0.0005834810840330804,
      "loss": 2.444,
      "step": 24451
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.641962468624115,
      "learning_rate": 0.0005834797453602319,
      "loss": 2.4135,
      "step": 24452
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.690118134021759,
      "learning_rate": 0.0005834784066346792,
      "loss": 2.3413,
      "step": 24453
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.608228862285614,
      "learning_rate": 0.0005834770678564225,
      "loss": 2.6044,
      "step": 24454
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6129176616668701,
      "learning_rate": 0.000583475729025462,
      "loss": 2.3291,
      "step": 24455
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.5924570560455322,
      "learning_rate": 0.0005834743901417982,
      "loss": 2.2497,
      "step": 24456
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.5989783406257629,
      "learning_rate": 0.000583473051205431,
      "loss": 2.3785,
      "step": 24457
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6315779089927673,
      "learning_rate": 0.0005834717122163609,
      "loss": 2.2623,
      "step": 24458
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6342922449111938,
      "learning_rate": 0.000583470373174588,
      "loss": 2.3491,
      "step": 24459
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.638866662979126,
      "learning_rate": 0.0005834690340801125,
      "loss": 2.2458,
      "step": 24460
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6781018376350403,
      "learning_rate": 0.0005834676949329349,
      "loss": 2.6654,
      "step": 24461
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.5646901726722717,
      "learning_rate": 0.0005834663557330554,
      "loss": 2.1473,
      "step": 24462
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.576310396194458,
      "learning_rate": 0.000583465016480474,
      "loss": 2.299,
      "step": 24463
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6302894353866577,
      "learning_rate": 0.000583463677175191,
      "loss": 2.243,
      "step": 24464
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6141575574874878,
      "learning_rate": 0.000583462337817207,
      "loss": 2.2197,
      "step": 24465
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6640001535415649,
      "learning_rate": 0.0005834609984065219,
      "loss": 2.1461,
      "step": 24466
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6193934082984924,
      "learning_rate": 0.000583459658943136,
      "loss": 2.296,
      "step": 24467
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6042346358299255,
      "learning_rate": 0.0005834583194270496,
      "loss": 2.3773,
      "step": 24468
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6324546933174133,
      "learning_rate": 0.000583456979858263,
      "loss": 2.4916,
      "step": 24469
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.5708721876144409,
      "learning_rate": 0.0005834556402367764,
      "loss": 2.1831,
      "step": 24470
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.5984455943107605,
      "learning_rate": 0.00058345430056259,
      "loss": 2.5627,
      "step": 24471
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6634904742240906,
      "learning_rate": 0.0005834529608357042,
      "loss": 2.5787,
      "step": 24472
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6236060261726379,
      "learning_rate": 0.0005834516210561191,
      "loss": 2.368,
      "step": 24473
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.5878305435180664,
      "learning_rate": 0.0005834502812238348,
      "loss": 2.3867,
      "step": 24474
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.604567289352417,
      "learning_rate": 0.000583448941338852,
      "loss": 2.351,
      "step": 24475
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.5995954871177673,
      "learning_rate": 0.0005834476014011706,
      "loss": 2.0921,
      "step": 24476
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6298183798789978,
      "learning_rate": 0.0005834462614107909,
      "loss": 2.2524,
      "step": 24477
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.7217579483985901,
      "learning_rate": 0.0005834449213677132,
      "loss": 2.2728,
      "step": 24478
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.7032366394996643,
      "learning_rate": 0.0005834435812719378,
      "loss": 2.319,
      "step": 24479
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6010800004005432,
      "learning_rate": 0.0005834422411234648,
      "loss": 2.3616,
      "step": 24480
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6116513013839722,
      "learning_rate": 0.0005834409009222946,
      "loss": 2.5484,
      "step": 24481
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6186842322349548,
      "learning_rate": 0.0005834395606684273,
      "loss": 2.2473,
      "step": 24482
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.5981619954109192,
      "learning_rate": 0.0005834382203618633,
      "loss": 2.2444,
      "step": 24483
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6122602224349976,
      "learning_rate": 0.0005834368800026029,
      "loss": 2.2421,
      "step": 24484
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.7362003922462463,
      "learning_rate": 0.000583435539590646,
      "loss": 2.331,
      "step": 24485
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6370182633399963,
      "learning_rate": 0.0005834341991259933,
      "loss": 2.1208,
      "step": 24486
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6432681083679199,
      "learning_rate": 0.0005834328586086447,
      "loss": 2.5891,
      "step": 24487
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6066710948944092,
      "learning_rate": 0.0005834315180386007,
      "loss": 2.5462,
      "step": 24488
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6309509873390198,
      "learning_rate": 0.0005834301774158613,
      "loss": 2.4078,
      "step": 24489
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6499704122543335,
      "learning_rate": 0.0005834288367404269,
      "loss": 2.2473,
      "step": 24490
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.738693356513977,
      "learning_rate": 0.0005834274960122978,
      "loss": 2.4571,
      "step": 24491
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6587144732475281,
      "learning_rate": 0.0005834261552314741,
      "loss": 2.2982,
      "step": 24492
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6163500547409058,
      "learning_rate": 0.0005834248143979562,
      "loss": 2.3962,
      "step": 24493
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.8491998314857483,
      "learning_rate": 0.0005834234735117443,
      "loss": 2.4837,
      "step": 24494
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.624753475189209,
      "learning_rate": 0.0005834221325728384,
      "loss": 2.6169,
      "step": 24495
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.5959471464157104,
      "learning_rate": 0.0005834207915812392,
      "loss": 2.1181,
      "step": 24496
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6433407068252563,
      "learning_rate": 0.0005834194505369467,
      "loss": 2.3291,
      "step": 24497
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.5872529149055481,
      "learning_rate": 0.0005834181094399612,
      "loss": 2.5855,
      "step": 24498
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6267510652542114,
      "learning_rate": 0.0005834167682902829,
      "loss": 2.0704,
      "step": 24499
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.7365046739578247,
      "learning_rate": 0.000583415427087912,
      "loss": 2.3614,
      "step": 24500
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.5948323011398315,
      "learning_rate": 0.0005834140858328488,
      "loss": 2.4693,
      "step": 24501
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6397449970245361,
      "learning_rate": 0.0005834127445250937,
      "loss": 2.2847,
      "step": 24502
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6186478137969971,
      "learning_rate": 0.0005834114031646467,
      "loss": 2.2776,
      "step": 24503
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6506509780883789,
      "learning_rate": 0.0005834100617515082,
      "loss": 2.6196,
      "step": 24504
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6314049959182739,
      "learning_rate": 0.0005834087202856785,
      "loss": 2.3532,
      "step": 24505
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.5775078535079956,
      "learning_rate": 0.0005834073787671577,
      "loss": 2.3503,
      "step": 24506
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6150712370872498,
      "learning_rate": 0.0005834060371959462,
      "loss": 2.3518,
      "step": 24507
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6417924165725708,
      "learning_rate": 0.000583404695572044,
      "loss": 2.2907,
      "step": 24508
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.629204511642456,
      "learning_rate": 0.0005834033538954517,
      "loss": 2.2953,
      "step": 24509
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6218809485435486,
      "learning_rate": 0.0005834020121661692,
      "loss": 2.1804,
      "step": 24510
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6071044206619263,
      "learning_rate": 0.000583400670384197,
      "loss": 2.3859,
      "step": 24511
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6313351988792419,
      "learning_rate": 0.0005833993285495353,
      "loss": 2.2116,
      "step": 24512
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6332101225852966,
      "learning_rate": 0.0005833979866621843,
      "loss": 2.2371,
      "step": 24513
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6998740434646606,
      "learning_rate": 0.0005833966447221442,
      "loss": 2.2848,
      "step": 24514
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.5852080583572388,
      "learning_rate": 0.0005833953027294154,
      "loss": 2.1448,
      "step": 24515
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6597111225128174,
      "learning_rate": 0.000583393960683998,
      "loss": 2.5249,
      "step": 24516
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6489187479019165,
      "learning_rate": 0.0005833926185858923,
      "loss": 2.3639,
      "step": 24517
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6093646287918091,
      "learning_rate": 0.0005833912764350987,
      "loss": 2.1592,
      "step": 24518
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.5600168704986572,
      "learning_rate": 0.0005833899342316172,
      "loss": 2.3029,
      "step": 24519
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6395792365074158,
      "learning_rate": 0.0005833885919754482,
      "loss": 2.5001,
      "step": 24520
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6068308353424072,
      "learning_rate": 0.0005833872496665919,
      "loss": 2.3997,
      "step": 24521
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6487604975700378,
      "learning_rate": 0.0005833859073050485,
      "loss": 2.3331,
      "step": 24522
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.65092933177948,
      "learning_rate": 0.0005833845648908183,
      "loss": 2.2584,
      "step": 24523
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.5882639288902283,
      "learning_rate": 0.0005833832224239016,
      "loss": 2.2354,
      "step": 24524
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6091277599334717,
      "learning_rate": 0.0005833818799042986,
      "loss": 2.3909,
      "step": 24525
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6170525550842285,
      "learning_rate": 0.0005833805373320097,
      "loss": 2.2524,
      "step": 24526
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.5873812437057495,
      "learning_rate": 0.0005833791947070349,
      "loss": 2.2269,
      "step": 24527
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6285306215286255,
      "learning_rate": 0.0005833778520293745,
      "loss": 2.3185,
      "step": 24528
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.632112443447113,
      "learning_rate": 0.0005833765092990289,
      "loss": 2.3081,
      "step": 24529
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.7402811646461487,
      "learning_rate": 0.0005833751665159982,
      "loss": 2.4505,
      "step": 24530
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6679448485374451,
      "learning_rate": 0.0005833738236802827,
      "loss": 2.3912,
      "step": 24531
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.5987241268157959,
      "learning_rate": 0.0005833724807918827,
      "loss": 2.4054,
      "step": 24532
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.7270421981811523,
      "learning_rate": 0.0005833711378507983,
      "loss": 2.1639,
      "step": 24533
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.570270299911499,
      "learning_rate": 0.00058336979485703,
      "loss": 2.319,
      "step": 24534
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6505542993545532,
      "learning_rate": 0.0005833684518105779,
      "loss": 2.3126,
      "step": 24535
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.5934382677078247,
      "learning_rate": 0.0005833671087114422,
      "loss": 2.126,
      "step": 24536
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6264271140098572,
      "learning_rate": 0.0005833657655596233,
      "loss": 2.2718,
      "step": 24537
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.828126847743988,
      "learning_rate": 0.0005833644223551212,
      "loss": 2.155,
      "step": 24538
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.7534767389297485,
      "learning_rate": 0.0005833630790979364,
      "loss": 2.4043,
      "step": 24539
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.5854179263114929,
      "learning_rate": 0.0005833617357880689,
      "loss": 2.1919,
      "step": 24540
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6366217136383057,
      "learning_rate": 0.0005833603924255194,
      "loss": 2.3878,
      "step": 24541
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6011453866958618,
      "learning_rate": 0.0005833590490102876,
      "loss": 2.2473,
      "step": 24542
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.8184024691581726,
      "learning_rate": 0.0005833577055423741,
      "loss": 2.5063,
      "step": 24543
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.7369702458381653,
      "learning_rate": 0.0005833563620217791,
      "loss": 2.3421,
      "step": 24544
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6261276602745056,
      "learning_rate": 0.0005833550184485027,
      "loss": 2.3371,
      "step": 24545
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.799213707447052,
      "learning_rate": 0.0005833536748225453,
      "loss": 2.4094,
      "step": 24546
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6861506104469299,
      "learning_rate": 0.0005833523311439071,
      "loss": 2.4586,
      "step": 24547
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.5862553119659424,
      "learning_rate": 0.0005833509874125884,
      "loss": 2.4995,
      "step": 24548
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.7030026912689209,
      "learning_rate": 0.0005833496436285894,
      "loss": 2.3165,
      "step": 24549
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6657087206840515,
      "learning_rate": 0.0005833482997919103,
      "loss": 2.5176,
      "step": 24550
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6393424868583679,
      "learning_rate": 0.0005833469559025513,
      "loss": 2.2795,
      "step": 24551
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6072680354118347,
      "learning_rate": 0.000583345611960513,
      "loss": 2.3366,
      "step": 24552
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.634714663028717,
      "learning_rate": 0.0005833442679657952,
      "loss": 2.5649,
      "step": 24553
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6254163384437561,
      "learning_rate": 0.0005833429239183984,
      "loss": 2.3877,
      "step": 24554
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6130639314651489,
      "learning_rate": 0.0005833415798183228,
      "loss": 2.205,
      "step": 24555
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6204171180725098,
      "learning_rate": 0.0005833402356655687,
      "loss": 2.2757,
      "step": 24556
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.5942158102989197,
      "learning_rate": 0.0005833388914601363,
      "loss": 2.3961,
      "step": 24557
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6271855235099792,
      "learning_rate": 0.0005833375472020258,
      "loss": 2.3909,
      "step": 24558
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6380589604377747,
      "learning_rate": 0.0005833362028912376,
      "loss": 2.1173,
      "step": 24559
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.7186734080314636,
      "learning_rate": 0.0005833348585277718,
      "loss": 2.3812,
      "step": 24560
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.607731282711029,
      "learning_rate": 0.0005833335141116287,
      "loss": 2.4368,
      "step": 24561
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.7092514038085938,
      "learning_rate": 0.0005833321696428085,
      "loss": 2.1504,
      "step": 24562
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.7137359976768494,
      "learning_rate": 0.0005833308251213115,
      "loss": 2.3395,
      "step": 24563
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.7084527015686035,
      "learning_rate": 0.000583329480547138,
      "loss": 2.2428,
      "step": 24564
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6815042495727539,
      "learning_rate": 0.0005833281359202882,
      "loss": 2.1909,
      "step": 24565
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.5801648497581482,
      "learning_rate": 0.0005833267912407624,
      "loss": 2.2763,
      "step": 24566
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6755126118659973,
      "learning_rate": 0.0005833254465085607,
      "loss": 2.2253,
      "step": 24567
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6812751293182373,
      "learning_rate": 0.0005833241017236835,
      "loss": 2.3968,
      "step": 24568
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6351103186607361,
      "learning_rate": 0.0005833227568861311,
      "loss": 2.4945,
      "step": 24569
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6745394468307495,
      "learning_rate": 0.0005833214119959035,
      "loss": 2.3666,
      "step": 24570
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.7296375036239624,
      "learning_rate": 0.0005833200670530011,
      "loss": 2.2965,
      "step": 24571
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6696237921714783,
      "learning_rate": 0.0005833187220574241,
      "loss": 2.3946,
      "step": 24572
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.693769633769989,
      "learning_rate": 0.000583317377009173,
      "loss": 2.2865,
      "step": 24573
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6587412357330322,
      "learning_rate": 0.0005833160319082477,
      "loss": 2.3707,
      "step": 24574
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.9322420954704285,
      "learning_rate": 0.0005833146867546487,
      "loss": 2.3254,
      "step": 24575
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6027212738990784,
      "learning_rate": 0.0005833133415483761,
      "loss": 2.2425,
      "step": 24576
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.7213086485862732,
      "learning_rate": 0.0005833119962894302,
      "loss": 2.254,
      "step": 24577
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.5988545417785645,
      "learning_rate": 0.0005833106509778112,
      "loss": 2.2568,
      "step": 24578
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6093863248825073,
      "learning_rate": 0.0005833093056135196,
      "loss": 2.1071,
      "step": 24579
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.7083190083503723,
      "learning_rate": 0.0005833079601965553,
      "loss": 2.3505,
      "step": 24580
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6143321990966797,
      "learning_rate": 0.0005833066147269187,
      "loss": 2.3172,
      "step": 24581
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6551318764686584,
      "learning_rate": 0.00058330526920461,
      "loss": 2.2984,
      "step": 24582
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.672719419002533,
      "learning_rate": 0.0005833039236296297,
      "loss": 2.5218,
      "step": 24583
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6089533567428589,
      "learning_rate": 0.0005833025780019777,
      "loss": 2.4127,
      "step": 24584
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.623782753944397,
      "learning_rate": 0.0005833012323216544,
      "loss": 2.55,
      "step": 24585
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6294255256652832,
      "learning_rate": 0.00058329988658866,
      "loss": 2.13,
      "step": 24586
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6891812086105347,
      "learning_rate": 0.0005832985408029949,
      "loss": 2.4929,
      "step": 24587
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6758654117584229,
      "learning_rate": 0.0005832971949646592,
      "loss": 2.3801,
      "step": 24588
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6484326720237732,
      "learning_rate": 0.0005832958490736532,
      "loss": 2.5151,
      "step": 24589
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6344383358955383,
      "learning_rate": 0.0005832945031299773,
      "loss": 2.3388,
      "step": 24590
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6051319241523743,
      "learning_rate": 0.0005832931571336315,
      "loss": 2.3369,
      "step": 24591
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.5961199402809143,
      "learning_rate": 0.0005832918110846161,
      "loss": 2.4592,
      "step": 24592
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.7578656673431396,
      "learning_rate": 0.0005832904649829315,
      "loss": 2.3909,
      "step": 24593
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.5920156240463257,
      "learning_rate": 0.0005832891188285778,
      "loss": 2.1454,
      "step": 24594
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.5943770408630371,
      "learning_rate": 0.0005832877726215554,
      "loss": 2.2938,
      "step": 24595
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.5893829464912415,
      "learning_rate": 0.0005832864263618644,
      "loss": 2.2977,
      "step": 24596
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6525117754936218,
      "learning_rate": 0.000583285080049505,
      "loss": 2.295,
      "step": 24597
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.7271701097488403,
      "learning_rate": 0.0005832837336844776,
      "loss": 2.2405,
      "step": 24598
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.5765397548675537,
      "learning_rate": 0.0005832823872667825,
      "loss": 2.335,
      "step": 24599
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6462503671646118,
      "learning_rate": 0.0005832810407964198,
      "loss": 2.1752,
      "step": 24600
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.590660810470581,
      "learning_rate": 0.0005832796942733898,
      "loss": 2.2921,
      "step": 24601
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.576470136642456,
      "learning_rate": 0.0005832783476976928,
      "loss": 2.4053,
      "step": 24602
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.5812440514564514,
      "learning_rate": 0.000583277001069329,
      "loss": 2.4203,
      "step": 24603
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6580289006233215,
      "learning_rate": 0.0005832756543882986,
      "loss": 2.3966,
      "step": 24604
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6307480931282043,
      "learning_rate": 0.0005832743076546019,
      "loss": 2.5597,
      "step": 24605
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.0810669660568237,
      "learning_rate": 0.0005832729608682393,
      "loss": 2.2711,
      "step": 24606
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6397084593772888,
      "learning_rate": 0.0005832716140292107,
      "loss": 2.2787,
      "step": 24607
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.671356201171875,
      "learning_rate": 0.0005832702671375167,
      "loss": 2.5374,
      "step": 24608
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6321620941162109,
      "learning_rate": 0.0005832689201931574,
      "loss": 2.3396,
      "step": 24609
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6379536986351013,
      "learning_rate": 0.0005832675731961331,
      "loss": 2.4868,
      "step": 24610
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6369987726211548,
      "learning_rate": 0.0005832662261464438,
      "loss": 2.3114,
      "step": 24611
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6409140825271606,
      "learning_rate": 0.0005832648790440901,
      "loss": 2.4345,
      "step": 24612
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.643061637878418,
      "learning_rate": 0.0005832635318890722,
      "loss": 2.3311,
      "step": 24613
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6216879487037659,
      "learning_rate": 0.0005832621846813902,
      "loss": 2.2865,
      "step": 24614
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6951949596405029,
      "learning_rate": 0.0005832608374210445,
      "loss": 2.3241,
      "step": 24615
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.5916338562965393,
      "learning_rate": 0.000583259490108035,
      "loss": 2.2113,
      "step": 24616
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6968157887458801,
      "learning_rate": 0.0005832581427423624,
      "loss": 2.3762,
      "step": 24617
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.8016669154167175,
      "learning_rate": 0.0005832567953240268,
      "loss": 2.306,
      "step": 24618
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6185688972473145,
      "learning_rate": 0.0005832554478530283,
      "loss": 2.3719,
      "step": 24619
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.5889557003974915,
      "learning_rate": 0.0005832541003293674,
      "loss": 2.317,
      "step": 24620
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.7114004492759705,
      "learning_rate": 0.0005832527527530441,
      "loss": 2.4471,
      "step": 24621
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.7732604742050171,
      "learning_rate": 0.0005832514051240589,
      "loss": 2.4884,
      "step": 24622
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6353569030761719,
      "learning_rate": 0.0005832500574424117,
      "loss": 2.5171,
      "step": 24623
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.576736569404602,
      "learning_rate": 0.000583248709708103,
      "loss": 2.2634,
      "step": 24624
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.7078192234039307,
      "learning_rate": 0.0005832473619211332,
      "loss": 2.4811,
      "step": 24625
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6313559412956238,
      "learning_rate": 0.0005832460140815023,
      "loss": 2.444,
      "step": 24626
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6275390386581421,
      "learning_rate": 0.0005832446661892106,
      "loss": 2.3725,
      "step": 24627
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6521428227424622,
      "learning_rate": 0.0005832433182442583,
      "loss": 2.2489,
      "step": 24628
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.7206679582595825,
      "learning_rate": 0.0005832419702466458,
      "loss": 2.3656,
      "step": 24629
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6512297987937927,
      "learning_rate": 0.0005832406221963733,
      "loss": 2.1785,
      "step": 24630
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6881589889526367,
      "learning_rate": 0.0005832392740934409,
      "loss": 2.4454,
      "step": 24631
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.760198175907135,
      "learning_rate": 0.0005832379259378491,
      "loss": 2.2067,
      "step": 24632
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.685509443283081,
      "learning_rate": 0.000583236577729598,
      "loss": 2.5612,
      "step": 24633
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6318178176879883,
      "learning_rate": 0.0005832352294686878,
      "loss": 2.2193,
      "step": 24634
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6401739716529846,
      "learning_rate": 0.0005832338811551189,
      "loss": 2.2712,
      "step": 24635
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.701479434967041,
      "learning_rate": 0.0005832325327888914,
      "loss": 2.3194,
      "step": 24636
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.7494970560073853,
      "learning_rate": 0.0005832311843700057,
      "loss": 2.4694,
      "step": 24637
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6094453930854797,
      "learning_rate": 0.000583229835898462,
      "loss": 2.1844,
      "step": 24638
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6316802501678467,
      "learning_rate": 0.0005832284873742604,
      "loss": 2.4054,
      "step": 24639
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6433680057525635,
      "learning_rate": 0.0005832271387974014,
      "loss": 2.6847,
      "step": 24640
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.5769058465957642,
      "learning_rate": 0.000583225790167885,
      "loss": 2.2145,
      "step": 24641
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.612175464630127,
      "learning_rate": 0.0005832244414857117,
      "loss": 2.2998,
      "step": 24642
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6083342432975769,
      "learning_rate": 0.0005832230927508816,
      "loss": 2.2813,
      "step": 24643
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6755123734474182,
      "learning_rate": 0.000583221743963395,
      "loss": 2.3109,
      "step": 24644
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6478018760681152,
      "learning_rate": 0.0005832203951232521,
      "loss": 2.5172,
      "step": 24645
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.614576518535614,
      "learning_rate": 0.0005832190462304532,
      "loss": 2.3931,
      "step": 24646
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6929343342781067,
      "learning_rate": 0.0005832176972849985,
      "loss": 2.4191,
      "step": 24647
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6624758243560791,
      "learning_rate": 0.0005832163482868883,
      "loss": 2.5681,
      "step": 24648
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6415777206420898,
      "learning_rate": 0.0005832149992361229,
      "loss": 2.2924,
      "step": 24649
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6361322999000549,
      "learning_rate": 0.0005832136501327024,
      "loss": 2.2552,
      "step": 24650
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6659133434295654,
      "learning_rate": 0.0005832123009766271,
      "loss": 2.5284,
      "step": 24651
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.5834241509437561,
      "learning_rate": 0.0005832109517678975,
      "loss": 2.4883,
      "step": 24652
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6466190218925476,
      "learning_rate": 0.0005832096025065133,
      "loss": 2.4876,
      "step": 24653
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6664950251579285,
      "learning_rate": 0.0005832082531924753,
      "loss": 2.4949,
      "step": 24654
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.687491238117218,
      "learning_rate": 0.0005832069038257835,
      "loss": 2.2442,
      "step": 24655
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.5849964618682861,
      "learning_rate": 0.0005832055544064382,
      "loss": 2.1265,
      "step": 24656
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6176571249961853,
      "learning_rate": 0.0005832042049344397,
      "loss": 2.4518,
      "step": 24657
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6169373393058777,
      "learning_rate": 0.0005832028554097881,
      "loss": 2.2124,
      "step": 24658
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6294845938682556,
      "learning_rate": 0.0005832015058324839,
      "loss": 2.3487,
      "step": 24659
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6937146782875061,
      "learning_rate": 0.000583200156202527,
      "loss": 2.3037,
      "step": 24660
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.5785266757011414,
      "learning_rate": 0.000583198806519918,
      "loss": 2.3029,
      "step": 24661
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.5652889609336853,
      "learning_rate": 0.0005831974567846568,
      "loss": 2.271,
      "step": 24662
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.5988055467605591,
      "learning_rate": 0.0005831961069967441,
      "loss": 2.2284,
      "step": 24663
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6208221316337585,
      "learning_rate": 0.0005831947571561796,
      "loss": 2.4448,
      "step": 24664
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6668733954429626,
      "learning_rate": 0.000583193407262964,
      "loss": 2.5087,
      "step": 24665
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6100484728813171,
      "learning_rate": 0.0005831920573170975,
      "loss": 2.1324,
      "step": 24666
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6175578236579895,
      "learning_rate": 0.0005831907073185801,
      "loss": 2.4058,
      "step": 24667
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6547341346740723,
      "learning_rate": 0.0005831893572674123,
      "loss": 2.2439,
      "step": 24668
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6428421139717102,
      "learning_rate": 0.0005831880071635943,
      "loss": 2.4018,
      "step": 24669
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.5722661018371582,
      "learning_rate": 0.0005831866570071262,
      "loss": 1.9868,
      "step": 24670
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6325848698616028,
      "learning_rate": 0.0005831853067980083,
      "loss": 2.4078,
      "step": 24671
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.7177642583847046,
      "learning_rate": 0.000583183956536241,
      "loss": 2.1067,
      "step": 24672
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6178116202354431,
      "learning_rate": 0.0005831826062218244,
      "loss": 2.2995,
      "step": 24673
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.7547614574432373,
      "learning_rate": 0.0005831812558547589,
      "loss": 2.4057,
      "step": 24674
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.7206624150276184,
      "learning_rate": 0.0005831799054350446,
      "loss": 2.1648,
      "step": 24675
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6594089865684509,
      "learning_rate": 0.0005831785549626818,
      "loss": 2.2711,
      "step": 24676
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.5841926336288452,
      "learning_rate": 0.0005831772044376707,
      "loss": 2.296,
      "step": 24677
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6803175806999207,
      "learning_rate": 0.0005831758538600118,
      "loss": 2.2471,
      "step": 24678
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6445342302322388,
      "learning_rate": 0.000583174503229705,
      "loss": 2.1747,
      "step": 24679
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6619896292686462,
      "learning_rate": 0.0005831731525467508,
      "loss": 2.2364,
      "step": 24680
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6626577377319336,
      "learning_rate": 0.0005831718018111493,
      "loss": 2.1847,
      "step": 24681
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6615908741950989,
      "learning_rate": 0.0005831704510229009,
      "loss": 2.1914,
      "step": 24682
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.7068854570388794,
      "learning_rate": 0.0005831691001820057,
      "loss": 2.2399,
      "step": 24683
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6802363991737366,
      "learning_rate": 0.0005831677492884641,
      "loss": 2.439,
      "step": 24684
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6001304388046265,
      "learning_rate": 0.000583166398342276,
      "loss": 2.3319,
      "step": 24685
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6258274912834167,
      "learning_rate": 0.0005831650473434422,
      "loss": 2.0597,
      "step": 24686
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.5877242088317871,
      "learning_rate": 0.0005831636962919626,
      "loss": 2.5089,
      "step": 24687
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6768296360969543,
      "learning_rate": 0.0005831623451878376,
      "loss": 2.365,
      "step": 24688
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6129754185676575,
      "learning_rate": 0.0005831609940310673,
      "loss": 2.2537,
      "step": 24689
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6412891745567322,
      "learning_rate": 0.000583159642821652,
      "loss": 2.251,
      "step": 24690
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6585320234298706,
      "learning_rate": 0.000583158291559592,
      "loss": 2.127,
      "step": 24691
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.626569390296936,
      "learning_rate": 0.0005831569402448875,
      "loss": 2.3668,
      "step": 24692
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6419271230697632,
      "learning_rate": 0.0005831555888775389,
      "loss": 2.2763,
      "step": 24693
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6134958267211914,
      "learning_rate": 0.0005831542374575461,
      "loss": 2.3065,
      "step": 24694
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6186870336532593,
      "learning_rate": 0.0005831528859849099,
      "loss": 2.2214,
      "step": 24695
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6435635089874268,
      "learning_rate": 0.0005831515344596299,
      "loss": 2.4948,
      "step": 24696
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6574677228927612,
      "learning_rate": 0.0005831501828817069,
      "loss": 2.2397,
      "step": 24697
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6114076972007751,
      "learning_rate": 0.0005831488312511409,
      "loss": 2.3116,
      "step": 24698
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6441512107849121,
      "learning_rate": 0.0005831474795679321,
      "loss": 2.4371,
      "step": 24699
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6090693473815918,
      "learning_rate": 0.000583146127832081,
      "loss": 2.4661,
      "step": 24700
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6211798787117004,
      "learning_rate": 0.0005831447760435875,
      "loss": 2.2878,
      "step": 24701
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6051018834114075,
      "learning_rate": 0.0005831434242024522,
      "loss": 2.482,
      "step": 24702
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6628425121307373,
      "learning_rate": 0.0005831420723086751,
      "loss": 2.1061,
      "step": 24703
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.7135918736457825,
      "learning_rate": 0.0005831407203622566,
      "loss": 2.3356,
      "step": 24704
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6456117033958435,
      "learning_rate": 0.0005831393683631968,
      "loss": 2.2472,
      "step": 24705
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6576611995697021,
      "learning_rate": 0.0005831380163114962,
      "loss": 2.6091,
      "step": 24706
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.5916333794593811,
      "learning_rate": 0.0005831366642071548,
      "loss": 2.4955,
      "step": 24707
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6169940233230591,
      "learning_rate": 0.000583135312050173,
      "loss": 2.2939,
      "step": 24708
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6389339566230774,
      "learning_rate": 0.0005831339598405509,
      "loss": 2.2843,
      "step": 24709
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6298994421958923,
      "learning_rate": 0.0005831326075782889,
      "loss": 2.4671,
      "step": 24710
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.7082569599151611,
      "learning_rate": 0.0005831312552633872,
      "loss": 2.3629,
      "step": 24711
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6493144035339355,
      "learning_rate": 0.0005831299028958461,
      "loss": 2.2104,
      "step": 24712
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6670494079589844,
      "learning_rate": 0.0005831285504756658,
      "loss": 2.3687,
      "step": 24713
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.7481920123100281,
      "learning_rate": 0.0005831271980028466,
      "loss": 2.3139,
      "step": 24714
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.642811119556427,
      "learning_rate": 0.0005831258454773886,
      "loss": 2.3861,
      "step": 24715
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.63620924949646,
      "learning_rate": 0.0005831244928992922,
      "loss": 2.5518,
      "step": 24716
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6124970316886902,
      "learning_rate": 0.0005831231402685577,
      "loss": 2.1597,
      "step": 24717
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.5906147956848145,
      "learning_rate": 0.0005831217875851851,
      "loss": 2.3329,
      "step": 24718
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.5853129625320435,
      "learning_rate": 0.0005831204348491749,
      "loss": 2.3283,
      "step": 24719
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.66982102394104,
      "learning_rate": 0.0005831190820605272,
      "loss": 2.4838,
      "step": 24720
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6767013669013977,
      "learning_rate": 0.0005831177292192425,
      "loss": 2.2641,
      "step": 24721
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6913638710975647,
      "learning_rate": 0.0005831163763253207,
      "loss": 2.573,
      "step": 24722
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.5810213088989258,
      "learning_rate": 0.0005831150233787623,
      "loss": 2.3552,
      "step": 24723
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.7194638848304749,
      "learning_rate": 0.0005831136703795675,
      "loss": 2.1867,
      "step": 24724
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6605715155601501,
      "learning_rate": 0.0005831123173277364,
      "loss": 2.3795,
      "step": 24725
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.5885291695594788,
      "learning_rate": 0.0005831109642232696,
      "loss": 2.2861,
      "step": 24726
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.667996883392334,
      "learning_rate": 0.0005831096110661669,
      "loss": 2.1483,
      "step": 24727
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.686334490776062,
      "learning_rate": 0.0005831082578564289,
      "loss": 2.4502,
      "step": 24728
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.5721434354782104,
      "learning_rate": 0.0005831069045940556,
      "loss": 2.193,
      "step": 24729
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.613470733165741,
      "learning_rate": 0.0005831055512790475,
      "loss": 2.4031,
      "step": 24730
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.5652454495429993,
      "learning_rate": 0.0005831041979114047,
      "loss": 2.1319,
      "step": 24731
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.8716368675231934,
      "learning_rate": 0.0005831028444911275,
      "loss": 2.3181,
      "step": 24732
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6428086161613464,
      "learning_rate": 0.0005831014910182161,
      "loss": 1.8926,
      "step": 24733
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.7291222214698792,
      "learning_rate": 0.0005831001374926708,
      "loss": 2.1992,
      "step": 24734
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.5853254795074463,
      "learning_rate": 0.0005830987839144919,
      "loss": 2.1578,
      "step": 24735
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.611411452293396,
      "learning_rate": 0.0005830974302836795,
      "loss": 2.4945,
      "step": 24736
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6211637258529663,
      "learning_rate": 0.000583096076600234,
      "loss": 2.285,
      "step": 24737
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.7557386159896851,
      "learning_rate": 0.0005830947228641556,
      "loss": 2.4154,
      "step": 24738
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.715032696723938,
      "learning_rate": 0.0005830933690754445,
      "loss": 2.2597,
      "step": 24739
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6655172109603882,
      "learning_rate": 0.0005830920152341011,
      "loss": 2.1203,
      "step": 24740
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.580604076385498,
      "learning_rate": 0.0005830906613401254,
      "loss": 2.2672,
      "step": 24741
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6696619987487793,
      "learning_rate": 0.0005830893073935179,
      "loss": 2.2868,
      "step": 24742
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.5711128115653992,
      "learning_rate": 0.0005830879533942788,
      "loss": 2.2153,
      "step": 24743
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6129617691040039,
      "learning_rate": 0.0005830865993424082,
      "loss": 2.4965,
      "step": 24744
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6230131387710571,
      "learning_rate": 0.0005830852452379065,
      "loss": 2.31,
      "step": 24745
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6752668023109436,
      "learning_rate": 0.0005830838910807739,
      "loss": 2.3745,
      "step": 24746
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.622504711151123,
      "learning_rate": 0.0005830825368710107,
      "loss": 2.3551,
      "step": 24747
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6304698586463928,
      "learning_rate": 0.0005830811826086171,
      "loss": 2.3112,
      "step": 24748
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.5966682434082031,
      "learning_rate": 0.0005830798282935933,
      "loss": 2.4658,
      "step": 24749
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6680156588554382,
      "learning_rate": 0.0005830784739259397,
      "loss": 2.4764,
      "step": 24750
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.836614191532135,
      "learning_rate": 0.0005830771195056566,
      "loss": 2.2314,
      "step": 24751
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6589751243591309,
      "learning_rate": 0.0005830757650327438,
      "loss": 2.489,
      "step": 24752
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.5630137920379639,
      "learning_rate": 0.0005830744105072021,
      "loss": 2.4405,
      "step": 24753
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6997992992401123,
      "learning_rate": 0.0005830730559290314,
      "loss": 2.2161,
      "step": 24754
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6713548898696899,
      "learning_rate": 0.0005830717012982322,
      "loss": 2.3007,
      "step": 24755
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.5546353459358215,
      "learning_rate": 0.0005830703466148044,
      "loss": 2.1199,
      "step": 24756
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6446219086647034,
      "learning_rate": 0.0005830689918787487,
      "loss": 2.0019,
      "step": 24757
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6866292953491211,
      "learning_rate": 0.0005830676370900651,
      "loss": 2.537,
      "step": 24758
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6852195858955383,
      "learning_rate": 0.0005830662822487538,
      "loss": 2.1632,
      "step": 24759
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.663395345211029,
      "learning_rate": 0.0005830649273548152,
      "loss": 2.1693,
      "step": 24760
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6367887258529663,
      "learning_rate": 0.0005830635724082495,
      "loss": 2.3916,
      "step": 24761
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.606493353843689,
      "learning_rate": 0.0005830622174090569,
      "loss": 2.1566,
      "step": 24762
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6333288550376892,
      "learning_rate": 0.0005830608623572377,
      "loss": 2.0723,
      "step": 24763
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6330822706222534,
      "learning_rate": 0.0005830595072527921,
      "loss": 2.3746,
      "step": 24764
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6563894748687744,
      "learning_rate": 0.0005830581520957205,
      "loss": 2.3552,
      "step": 24765
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6169402003288269,
      "learning_rate": 0.000583056796886023,
      "loss": 2.275,
      "step": 24766
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6030893325805664,
      "learning_rate": 0.0005830554416236998,
      "loss": 2.1959,
      "step": 24767
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6777637600898743,
      "learning_rate": 0.0005830540863087514,
      "loss": 2.4158,
      "step": 24768
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6932345032691956,
      "learning_rate": 0.0005830527309411778,
      "loss": 2.529,
      "step": 24769
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6338347792625427,
      "learning_rate": 0.0005830513755209794,
      "loss": 2.3019,
      "step": 24770
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6443374156951904,
      "learning_rate": 0.0005830500200481564,
      "loss": 2.3689,
      "step": 24771
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.7172282934188843,
      "learning_rate": 0.0005830486645227091,
      "loss": 2.4417,
      "step": 24772
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.611775279045105,
      "learning_rate": 0.0005830473089446376,
      "loss": 2.2708,
      "step": 24773
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.809788167476654,
      "learning_rate": 0.0005830459533139424,
      "loss": 2.2343,
      "step": 24774
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6977962255477905,
      "learning_rate": 0.0005830445976306235,
      "loss": 2.4534,
      "step": 24775
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.5843303203582764,
      "learning_rate": 0.0005830432418946814,
      "loss": 2.1962,
      "step": 24776
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.7247715592384338,
      "learning_rate": 0.0005830418861061161,
      "loss": 2.468,
      "step": 24777
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.7488082051277161,
      "learning_rate": 0.0005830405302649279,
      "loss": 2.3139,
      "step": 24778
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.5859563946723938,
      "learning_rate": 0.0005830391743711172,
      "loss": 2.1139,
      "step": 24779
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.5660607814788818,
      "learning_rate": 0.0005830378184246843,
      "loss": 2.4522,
      "step": 24780
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.7809011936187744,
      "learning_rate": 0.0005830364624256291,
      "loss": 2.3295,
      "step": 24781
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.620276927947998,
      "learning_rate": 0.0005830351063739523,
      "loss": 2.1437,
      "step": 24782
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.5752159953117371,
      "learning_rate": 0.0005830337502696537,
      "loss": 2.237,
      "step": 24783
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.7222132086753845,
      "learning_rate": 0.0005830323941127339,
      "loss": 2.2906,
      "step": 24784
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.646571695804596,
      "learning_rate": 0.0005830310379031932,
      "loss": 2.4036,
      "step": 24785
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6151593327522278,
      "learning_rate": 0.0005830296816410315,
      "loss": 2.1471,
      "step": 24786
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6089077591896057,
      "learning_rate": 0.0005830283253262493,
      "loss": 2.2486,
      "step": 24787
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.7286438345909119,
      "learning_rate": 0.0005830269689588467,
      "loss": 2.5201,
      "step": 24788
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6097305417060852,
      "learning_rate": 0.0005830256125388242,
      "loss": 2.3506,
      "step": 24789
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6269832253456116,
      "learning_rate": 0.0005830242560661817,
      "loss": 2.4373,
      "step": 24790
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6657689213752747,
      "learning_rate": 0.0005830228995409198,
      "loss": 2.3002,
      "step": 24791
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.7152084708213806,
      "learning_rate": 0.0005830215429630385,
      "loss": 2.3583,
      "step": 24792
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6190317273139954,
      "learning_rate": 0.0005830201863325382,
      "loss": 2.3074,
      "step": 24793
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6642652153968811,
      "learning_rate": 0.0005830188296494191,
      "loss": 2.3214,
      "step": 24794
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6595533490180969,
      "learning_rate": 0.0005830174729136814,
      "loss": 2.1987,
      "step": 24795
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6459239721298218,
      "learning_rate": 0.0005830161161253255,
      "loss": 2.2003,
      "step": 24796
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6493139863014221,
      "learning_rate": 0.0005830147592843516,
      "loss": 2.2264,
      "step": 24797
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.5680103302001953,
      "learning_rate": 0.0005830134023907599,
      "loss": 2.5293,
      "step": 24798
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6701355576515198,
      "learning_rate": 0.0005830120454445506,
      "loss": 2.374,
      "step": 24799
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6301862001419067,
      "learning_rate": 0.0005830106884457241,
      "loss": 2.1651,
      "step": 24800
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6154470443725586,
      "learning_rate": 0.0005830093313942804,
      "loss": 2.2122,
      "step": 24801
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6234455704689026,
      "learning_rate": 0.0005830079742902201,
      "loss": 2.4551,
      "step": 24802
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.591637909412384,
      "learning_rate": 0.0005830066171335432,
      "loss": 2.3262,
      "step": 24803
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.5822123885154724,
      "learning_rate": 0.00058300525992425,
      "loss": 2.3903,
      "step": 24804
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6491561532020569,
      "learning_rate": 0.0005830039026623408,
      "loss": 2.2571,
      "step": 24805
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.5972142815589905,
      "learning_rate": 0.000583002545347816,
      "loss": 2.3963,
      "step": 24806
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6799896955490112,
      "learning_rate": 0.0005830011879806755,
      "loss": 2.1546,
      "step": 24807
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6579515337944031,
      "learning_rate": 0.0005829998305609197,
      "loss": 2.2974,
      "step": 24808
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6029432415962219,
      "learning_rate": 0.000582998473088549,
      "loss": 2.2618,
      "step": 24809
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6495121121406555,
      "learning_rate": 0.0005829971155635634,
      "loss": 2.5262,
      "step": 24810
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6968076825141907,
      "learning_rate": 0.0005829957579859635,
      "loss": 2.1847,
      "step": 24811
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.5675108432769775,
      "learning_rate": 0.0005829944003557493,
      "loss": 2.3098,
      "step": 24812
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6281658411026001,
      "learning_rate": 0.0005829930426729209,
      "loss": 2.1809,
      "step": 24813
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6111581325531006,
      "learning_rate": 0.000582991684937479,
      "loss": 2.2579,
      "step": 24814
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.733457624912262,
      "learning_rate": 0.0005829903271494235,
      "loss": 2.2347,
      "step": 24815
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.5964659452438354,
      "learning_rate": 0.0005829889693087548,
      "loss": 2.5456,
      "step": 24816
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.7169690132141113,
      "learning_rate": 0.0005829876114154729,
      "loss": 2.2699,
      "step": 24817
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6271659731864929,
      "learning_rate": 0.0005829862534695785,
      "loss": 2.4476,
      "step": 24818
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6471816301345825,
      "learning_rate": 0.0005829848954710716,
      "loss": 2.3884,
      "step": 24819
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.5631025433540344,
      "learning_rate": 0.0005829835374199523,
      "loss": 2.2165,
      "step": 24820
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6930261850357056,
      "learning_rate": 0.0005829821793162212,
      "loss": 2.1169,
      "step": 24821
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6381708979606628,
      "learning_rate": 0.0005829808211598782,
      "loss": 2.3781,
      "step": 24822
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6653083562850952,
      "learning_rate": 0.0005829794629509239,
      "loss": 2.3193,
      "step": 24823
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6401097774505615,
      "learning_rate": 0.0005829781046893583,
      "loss": 2.2497,
      "step": 24824
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6711621880531311,
      "learning_rate": 0.0005829767463751817,
      "loss": 2.2243,
      "step": 24825
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6207196116447449,
      "learning_rate": 0.0005829753880083944,
      "loss": 2.398,
      "step": 24826
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.9465746879577637,
      "learning_rate": 0.0005829740295889966,
      "loss": 2.2015,
      "step": 24827
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6314958333969116,
      "learning_rate": 0.0005829726711169887,
      "loss": 2.47,
      "step": 24828
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.7554060220718384,
      "learning_rate": 0.0005829713125923707,
      "loss": 2.3328,
      "step": 24829
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6167944669723511,
      "learning_rate": 0.0005829699540151431,
      "loss": 2.2453,
      "step": 24830
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6603744029998779,
      "learning_rate": 0.000582968595385306,
      "loss": 2.1774,
      "step": 24831
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.75117427110672,
      "learning_rate": 0.0005829672367028597,
      "loss": 2.6271,
      "step": 24832
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6431267857551575,
      "learning_rate": 0.0005829658779678044,
      "loss": 2.2943,
      "step": 24833
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6182037591934204,
      "learning_rate": 0.0005829645191801404,
      "loss": 2.1414,
      "step": 24834
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6430876851081848,
      "learning_rate": 0.000582963160339868,
      "loss": 2.4293,
      "step": 24835
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.5803250670433044,
      "learning_rate": 0.0005829618014469874,
      "loss": 2.281,
      "step": 24836
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.5749213099479675,
      "learning_rate": 0.000582960442501499,
      "loss": 2.3308,
      "step": 24837
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.5825473070144653,
      "learning_rate": 0.0005829590835034027,
      "loss": 2.475,
      "step": 24838
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.5950208902359009,
      "learning_rate": 0.000582957724452699,
      "loss": 2.2699,
      "step": 24839
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6156138777732849,
      "learning_rate": 0.0005829563653493883,
      "loss": 2.2306,
      "step": 24840
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6912708878517151,
      "learning_rate": 0.0005829550061934705,
      "loss": 2.1697,
      "step": 24841
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6520462036132812,
      "learning_rate": 0.000582953646984946,
      "loss": 2.3572,
      "step": 24842
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6168006658554077,
      "learning_rate": 0.0005829522877238151,
      "loss": 2.1332,
      "step": 24843
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.7130336165428162,
      "learning_rate": 0.0005829509284100781,
      "loss": 2.3832,
      "step": 24844
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6810777187347412,
      "learning_rate": 0.0005829495690437351,
      "loss": 2.3505,
      "step": 24845
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.5570406913757324,
      "learning_rate": 0.0005829482096247865,
      "loss": 2.1628,
      "step": 24846
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6592262983322144,
      "learning_rate": 0.0005829468501532325,
      "loss": 1.995,
      "step": 24847
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.5901775360107422,
      "learning_rate": 0.0005829454906290733,
      "loss": 2.3452,
      "step": 24848
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6220871806144714,
      "learning_rate": 0.0005829441310523091,
      "loss": 2.4139,
      "step": 24849
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6202936172485352,
      "learning_rate": 0.0005829427714229404,
      "loss": 2.5161,
      "step": 24850
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.604266881942749,
      "learning_rate": 0.0005829414117409673,
      "loss": 2.3868,
      "step": 24851
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6027892231941223,
      "learning_rate": 0.00058294005200639,
      "loss": 2.389,
      "step": 24852
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.5852933526039124,
      "learning_rate": 0.0005829386922192087,
      "loss": 2.1043,
      "step": 24853
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6321554780006409,
      "learning_rate": 0.000582937332379424,
      "loss": 2.4168,
      "step": 24854
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6600567102432251,
      "learning_rate": 0.0005829359724870356,
      "loss": 2.4111,
      "step": 24855
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6336112022399902,
      "learning_rate": 0.0005829346125420442,
      "loss": 2.3259,
      "step": 24856
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6563584804534912,
      "learning_rate": 0.00058293325254445,
      "loss": 2.5035,
      "step": 24857
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.5780400633811951,
      "learning_rate": 0.0005829318924942531,
      "loss": 2.1571,
      "step": 24858
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6304931640625,
      "learning_rate": 0.0005829305323914538,
      "loss": 2.4221,
      "step": 24859
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6849469542503357,
      "learning_rate": 0.0005829291722360524,
      "loss": 2.1932,
      "step": 24860
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.7051482200622559,
      "learning_rate": 0.0005829278120280491,
      "loss": 2.2556,
      "step": 24861
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6814330220222473,
      "learning_rate": 0.0005829264517674443,
      "loss": 2.3784,
      "step": 24862
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.607977032661438,
      "learning_rate": 0.0005829250914542379,
      "loss": 2.1491,
      "step": 24863
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6638239026069641,
      "learning_rate": 0.0005829237310884306,
      "loss": 2.2301,
      "step": 24864
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6284521818161011,
      "learning_rate": 0.0005829223706700223,
      "loss": 2.3098,
      "step": 24865
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6518121957778931,
      "learning_rate": 0.0005829210101990134,
      "loss": 2.5267,
      "step": 24866
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6797918677330017,
      "learning_rate": 0.0005829196496754042,
      "loss": 2.3648,
      "step": 24867
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.7708946466445923,
      "learning_rate": 0.000582918289099195,
      "loss": 1.8823,
      "step": 24868
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.7089229822158813,
      "learning_rate": 0.0005829169284703858,
      "loss": 2.4478,
      "step": 24869
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.7387076020240784,
      "learning_rate": 0.000582915567788977,
      "loss": 2.5108,
      "step": 24870
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6498951315879822,
      "learning_rate": 0.000582914207054969,
      "loss": 2.3411,
      "step": 24871
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6602941155433655,
      "learning_rate": 0.0005829128462683617,
      "loss": 2.4267,
      "step": 24872
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.763703465461731,
      "learning_rate": 0.0005829114854291557,
      "loss": 2.1572,
      "step": 24873
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6800512075424194,
      "learning_rate": 0.0005829101245373511,
      "loss": 2.1814,
      "step": 24874
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.7188758850097656,
      "learning_rate": 0.0005829087635929481,
      "loss": 2.0235,
      "step": 24875
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6858710646629333,
      "learning_rate": 0.000582907402595947,
      "loss": 2.4841,
      "step": 24876
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.7067106366157532,
      "learning_rate": 0.0005829060415463482,
      "loss": 2.6558,
      "step": 24877
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.7007890939712524,
      "learning_rate": 0.0005829046804441518,
      "loss": 2.5435,
      "step": 24878
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6475352644920349,
      "learning_rate": 0.000582903319289358,
      "loss": 2.2679,
      "step": 24879
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.725688099861145,
      "learning_rate": 0.0005829019580819672,
      "loss": 2.2256,
      "step": 24880
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.7110030651092529,
      "learning_rate": 0.0005829005968219796,
      "loss": 2.4935,
      "step": 24881
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6285479068756104,
      "learning_rate": 0.0005828992355093954,
      "loss": 2.3879,
      "step": 24882
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6489153504371643,
      "learning_rate": 0.0005828978741442148,
      "loss": 2.3386,
      "step": 24883
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6812493801116943,
      "learning_rate": 0.0005828965127264383,
      "loss": 2.3226,
      "step": 24884
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.5824198126792908,
      "learning_rate": 0.000582895151256066,
      "loss": 2.2994,
      "step": 24885
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6005682349205017,
      "learning_rate": 0.000582893789733098,
      "loss": 2.2782,
      "step": 24886
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6641315221786499,
      "learning_rate": 0.0005828924281575348,
      "loss": 2.2968,
      "step": 24887
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.7205772399902344,
      "learning_rate": 0.0005828910665293766,
      "loss": 2.2997,
      "step": 24888
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6058046221733093,
      "learning_rate": 0.0005828897048486235,
      "loss": 2.3299,
      "step": 24889
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6305578351020813,
      "learning_rate": 0.0005828883431152759,
      "loss": 2.2758,
      "step": 24890
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.7209980487823486,
      "learning_rate": 0.0005828869813293341,
      "loss": 2.4127,
      "step": 24891
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.7693559527397156,
      "learning_rate": 0.0005828856194907983,
      "loss": 2.3532,
      "step": 24892
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.7513302564620972,
      "learning_rate": 0.0005828842575996686,
      "loss": 2.624,
      "step": 24893
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6677743196487427,
      "learning_rate": 0.0005828828956559454,
      "loss": 2.1973,
      "step": 24894
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.7155247926712036,
      "learning_rate": 0.0005828815336596289,
      "loss": 2.5396,
      "step": 24895
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.7771711349487305,
      "learning_rate": 0.0005828801716107195,
      "loss": 2.4337,
      "step": 24896
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.8582422137260437,
      "learning_rate": 0.0005828788095092173,
      "loss": 2.4552,
      "step": 24897
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.5921650528907776,
      "learning_rate": 0.0005828774473551225,
      "loss": 2.3635,
      "step": 24898
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.7072550654411316,
      "learning_rate": 0.0005828760851484355,
      "loss": 2.3652,
      "step": 24899
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6605277061462402,
      "learning_rate": 0.0005828747228891565,
      "loss": 2.2572,
      "step": 24900
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6274515986442566,
      "learning_rate": 0.0005828733605772857,
      "loss": 2.2751,
      "step": 24901
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6717150211334229,
      "learning_rate": 0.0005828719982128235,
      "loss": 2.4032,
      "step": 24902
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.7024921774864197,
      "learning_rate": 0.00058287063579577,
      "loss": 2.5691,
      "step": 24903
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.7742976546287537,
      "learning_rate": 0.0005828692733261255,
      "loss": 2.6061,
      "step": 24904
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6382071375846863,
      "learning_rate": 0.0005828679108038901,
      "loss": 2.1786,
      "step": 24905
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6700871586799622,
      "learning_rate": 0.0005828665482290645,
      "loss": 2.3092,
      "step": 24906
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.609863817691803,
      "learning_rate": 0.0005828651856016485,
      "loss": 2.4309,
      "step": 24907
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6746228337287903,
      "learning_rate": 0.0005828638229216426,
      "loss": 2.1388,
      "step": 24908
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6179614663124084,
      "learning_rate": 0.000582862460189047,
      "loss": 2.011,
      "step": 24909
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.654349684715271,
      "learning_rate": 0.0005828610974038618,
      "loss": 2.2468,
      "step": 24910
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.5652233958244324,
      "learning_rate": 0.0005828597345660876,
      "loss": 2.5902,
      "step": 24911
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.5975959897041321,
      "learning_rate": 0.0005828583716757243,
      "loss": 2.2365,
      "step": 24912
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6108420491218567,
      "learning_rate": 0.0005828570087327722,
      "loss": 2.4533,
      "step": 24913
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.5966638922691345,
      "learning_rate": 0.0005828556457372318,
      "loss": 2.3814,
      "step": 24914
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6449152231216431,
      "learning_rate": 0.0005828542826891031,
      "loss": 2.2403,
      "step": 24915
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6186298727989197,
      "learning_rate": 0.0005828529195883865,
      "loss": 2.3034,
      "step": 24916
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6062206625938416,
      "learning_rate": 0.0005828515564350822,
      "loss": 2.3552,
      "step": 24917
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6110153794288635,
      "learning_rate": 0.0005828501932291905,
      "loss": 2.3507,
      "step": 24918
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6017696857452393,
      "learning_rate": 0.0005828488299707115,
      "loss": 2.3803,
      "step": 24919
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6179897785186768,
      "learning_rate": 0.0005828474666596457,
      "loss": 2.3754,
      "step": 24920
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6755077242851257,
      "learning_rate": 0.000582846103295993,
      "loss": 2.415,
      "step": 24921
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6788111329078674,
      "learning_rate": 0.000582844739879754,
      "loss": 2.3874,
      "step": 24922
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.7225425243377686,
      "learning_rate": 0.0005828433764109288,
      "loss": 2.2776,
      "step": 24923
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6459522843360901,
      "learning_rate": 0.0005828420128895177,
      "loss": 2.434,
      "step": 24924
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.785232663154602,
      "learning_rate": 0.0005828406493155209,
      "loss": 2.3062,
      "step": 24925
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6753544211387634,
      "learning_rate": 0.0005828392856889387,
      "loss": 2.2293,
      "step": 24926
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6850900650024414,
      "learning_rate": 0.0005828379220097714,
      "loss": 2.3567,
      "step": 24927
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6391360759735107,
      "learning_rate": 0.0005828365582780192,
      "loss": 2.4986,
      "step": 24928
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.7026270627975464,
      "learning_rate": 0.0005828351944936823,
      "loss": 2.2804,
      "step": 24929
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.5620194673538208,
      "learning_rate": 0.0005828338306567609,
      "loss": 2.2596,
      "step": 24930
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6179647445678711,
      "learning_rate": 0.0005828324667672554,
      "loss": 2.5125,
      "step": 24931
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6880326867103577,
      "learning_rate": 0.0005828311028251662,
      "loss": 2.704,
      "step": 24932
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6210208535194397,
      "learning_rate": 0.0005828297388304931,
      "loss": 2.1634,
      "step": 24933
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6089330315589905,
      "learning_rate": 0.0005828283747832368,
      "loss": 2.0773,
      "step": 24934
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.7006660103797913,
      "learning_rate": 0.0005828270106833973,
      "loss": 2.182,
      "step": 24935
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.7296327352523804,
      "learning_rate": 0.0005828256465309748,
      "loss": 2.4566,
      "step": 24936
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6469297409057617,
      "learning_rate": 0.0005828242823259698,
      "loss": 2.5186,
      "step": 24937
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6765342354774475,
      "learning_rate": 0.0005828229180683824,
      "loss": 2.1951,
      "step": 24938
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.675267219543457,
      "learning_rate": 0.0005828215537582129,
      "loss": 2.1783,
      "step": 24939
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.5695211887359619,
      "learning_rate": 0.0005828201893954615,
      "loss": 2.3205,
      "step": 24940
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6768084764480591,
      "learning_rate": 0.0005828188249801286,
      "loss": 2.3155,
      "step": 24941
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6351487636566162,
      "learning_rate": 0.0005828174605122142,
      "loss": 2.1567,
      "step": 24942
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6494408845901489,
      "learning_rate": 0.0005828160959917187,
      "loss": 2.2685,
      "step": 24943
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6503959894180298,
      "learning_rate": 0.0005828147314186424,
      "loss": 2.4403,
      "step": 24944
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6869871616363525,
      "learning_rate": 0.0005828133667929855,
      "loss": 2.3074,
      "step": 24945
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.7483373284339905,
      "learning_rate": 0.0005828120021147483,
      "loss": 2.3787,
      "step": 24946
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6595770716667175,
      "learning_rate": 0.0005828106373839309,
      "loss": 2.4759,
      "step": 24947
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6119999289512634,
      "learning_rate": 0.0005828092726005338,
      "loss": 2.3192,
      "step": 24948
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.7416343688964844,
      "learning_rate": 0.0005828079077645571,
      "loss": 2.4201,
      "step": 24949
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.5838721394538879,
      "learning_rate": 0.0005828065428760009,
      "loss": 2.0965,
      "step": 24950
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.5913177132606506,
      "learning_rate": 0.0005828051779348658,
      "loss": 2.3914,
      "step": 24951
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.962137758731842,
      "learning_rate": 0.0005828038129411518,
      "loss": 2.2874,
      "step": 24952
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6623040437698364,
      "learning_rate": 0.0005828024478948592,
      "loss": 2.3467,
      "step": 24953
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.7147537469863892,
      "learning_rate": 0.0005828010827959884,
      "loss": 2.334,
      "step": 24954
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.8652344942092896,
      "learning_rate": 0.0005827997176445396,
      "loss": 2.1874,
      "step": 24955
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6361243724822998,
      "learning_rate": 0.0005827983524405127,
      "loss": 2.2211,
      "step": 24956
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6015188097953796,
      "learning_rate": 0.0005827969871839084,
      "loss": 2.2746,
      "step": 24957
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.7830474376678467,
      "learning_rate": 0.0005827956218747268,
      "loss": 2.43,
      "step": 24958
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.0538837909698486,
      "learning_rate": 0.0005827942565129683,
      "loss": 2.159,
      "step": 24959
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.642808198928833,
      "learning_rate": 0.0005827928910986329,
      "loss": 2.3275,
      "step": 24960
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.9524761438369751,
      "learning_rate": 0.0005827915256317209,
      "loss": 2.2916,
      "step": 24961
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.8255537748336792,
      "learning_rate": 0.0005827901601122327,
      "loss": 2.3974,
      "step": 24962
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6200330853462219,
      "learning_rate": 0.0005827887945401685,
      "loss": 2.394,
      "step": 24963
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.7406836152076721,
      "learning_rate": 0.0005827874289155285,
      "loss": 2.1446,
      "step": 24964
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.9825923442840576,
      "learning_rate": 0.0005827860632383129,
      "loss": 2.1587,
      "step": 24965
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.7782084345817566,
      "learning_rate": 0.000582784697508522,
      "loss": 2.1591,
      "step": 24966
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6867427825927734,
      "learning_rate": 0.0005827833317261562,
      "loss": 2.1466,
      "step": 24967
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.8933209180831909,
      "learning_rate": 0.0005827819658912156,
      "loss": 2.1384,
      "step": 24968
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.7919518947601318,
      "learning_rate": 0.0005827806000037005,
      "loss": 2.3554,
      "step": 24969
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6057455539703369,
      "learning_rate": 0.0005827792340636111,
      "loss": 2.3147,
      "step": 24970
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.7222571969032288,
      "learning_rate": 0.0005827778680709478,
      "loss": 2.3486,
      "step": 24971
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.7119116187095642,
      "learning_rate": 0.0005827765020257107,
      "loss": 2.018,
      "step": 24972
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6815095543861389,
      "learning_rate": 0.0005827751359279001,
      "loss": 2.3056,
      "step": 24973
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6187523007392883,
      "learning_rate": 0.0005827737697775163,
      "loss": 2.236,
      "step": 24974
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.69278883934021,
      "learning_rate": 0.0005827724035745596,
      "loss": 2.5309,
      "step": 24975
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.7094152569770813,
      "learning_rate": 0.0005827710373190301,
      "loss": 2.3113,
      "step": 24976
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6249139308929443,
      "learning_rate": 0.000582769671010928,
      "loss": 2.2944,
      "step": 24977
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6149362325668335,
      "learning_rate": 0.0005827683046502538,
      "loss": 2.387,
      "step": 24978
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6049175262451172,
      "learning_rate": 0.0005827669382370076,
      "loss": 2.3197,
      "step": 24979
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.7069419622421265,
      "learning_rate": 0.0005827655717711897,
      "loss": 2.535,
      "step": 24980
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6670204997062683,
      "learning_rate": 0.0005827642052528004,
      "loss": 2.303,
      "step": 24981
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.71947181224823,
      "learning_rate": 0.0005827628386818398,
      "loss": 2.2271,
      "step": 24982
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6336167454719543,
      "learning_rate": 0.0005827614720583083,
      "loss": 2.2257,
      "step": 24983
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6298348903656006,
      "learning_rate": 0.0005827601053822061,
      "loss": 2.2327,
      "step": 24984
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6217668652534485,
      "learning_rate": 0.0005827587386535335,
      "loss": 2.5362,
      "step": 24985
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6382434964179993,
      "learning_rate": 0.0005827573718722907,
      "loss": 2.3949,
      "step": 24986
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6861317753791809,
      "learning_rate": 0.0005827560050384779,
      "loss": 2.4301,
      "step": 24987
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6188148260116577,
      "learning_rate": 0.0005827546381520954,
      "loss": 2.5752,
      "step": 24988
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6560617089271545,
      "learning_rate": 0.0005827532712131436,
      "loss": 2.4962,
      "step": 24989
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6351675391197205,
      "learning_rate": 0.0005827519042216225,
      "loss": 2.4441,
      "step": 24990
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.7392823100090027,
      "learning_rate": 0.0005827505371775326,
      "loss": 2.2878,
      "step": 24991
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6525423526763916,
      "learning_rate": 0.000582749170080874,
      "loss": 2.2443,
      "step": 24992
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6765691637992859,
      "learning_rate": 0.000582747802931647,
      "loss": 2.4407,
      "step": 24993
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6737653613090515,
      "learning_rate": 0.0005827464357298518,
      "loss": 2.3714,
      "step": 24994
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6581434011459351,
      "learning_rate": 0.0005827450684754887,
      "loss": 2.3511,
      "step": 24995
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6427704691886902,
      "learning_rate": 0.000582743701168558,
      "loss": 2.0094,
      "step": 24996
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6970934867858887,
      "learning_rate": 0.0005827423338090598,
      "loss": 2.3622,
      "step": 24997
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.7339738607406616,
      "learning_rate": 0.0005827409663969946,
      "loss": 2.4167,
      "step": 24998
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.64363032579422,
      "learning_rate": 0.0005827395989323625,
      "loss": 2.3235,
      "step": 24999
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.5716578960418701,
      "learning_rate": 0.0005827382314151637,
      "loss": 2.4423,
      "step": 25000
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.5910738706588745,
      "learning_rate": 0.0005827368638453987,
      "loss": 2.2538,
      "step": 25001
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6494468450546265,
      "learning_rate": 0.0005827354962230674,
      "loss": 2.0935,
      "step": 25002
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.13494873046875,
      "learning_rate": 0.0005827341285481702,
      "loss": 2.4356,
      "step": 25003
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.5824286341667175,
      "learning_rate": 0.0005827327608207075,
      "loss": 2.2597,
      "step": 25004
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6265184879302979,
      "learning_rate": 0.0005827313930406794,
      "loss": 2.2262,
      "step": 25005
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6221470236778259,
      "learning_rate": 0.0005827300252080862,
      "loss": 2.4667,
      "step": 25006
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6897850632667542,
      "learning_rate": 0.0005827286573229281,
      "loss": 2.6108,
      "step": 25007
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.5934675931930542,
      "learning_rate": 0.0005827272893852056,
      "loss": 2.4601,
      "step": 25008
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.575516939163208,
      "learning_rate": 0.0005827259213949186,
      "loss": 2.2703,
      "step": 25009
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6274740695953369,
      "learning_rate": 0.0005827245533520675,
      "loss": 2.1534,
      "step": 25010
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.7265876531600952,
      "learning_rate": 0.0005827231852566526,
      "loss": 2.3305,
      "step": 25011
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.5884896516799927,
      "learning_rate": 0.0005827218171086741,
      "loss": 2.2578,
      "step": 25012
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.7116584181785583,
      "learning_rate": 0.0005827204489081324,
      "loss": 2.1645,
      "step": 25013
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.5909199118614197,
      "learning_rate": 0.0005827190806550275,
      "loss": 2.3061,
      "step": 25014
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.7468436360359192,
      "learning_rate": 0.0005827177123493599,
      "loss": 2.2574,
      "step": 25015
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6577247977256775,
      "learning_rate": 0.0005827163439911296,
      "loss": 2.1654,
      "step": 25016
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6728075742721558,
      "learning_rate": 0.000582714975580337,
      "loss": 2.239,
      "step": 25017
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6668593287467957,
      "learning_rate": 0.0005827136071169824,
      "loss": 2.3212,
      "step": 25018
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.688242495059967,
      "learning_rate": 0.0005827122386010661,
      "loss": 2.0998,
      "step": 25019
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6092954277992249,
      "learning_rate": 0.0005827108700325881,
      "loss": 2.38,
      "step": 25020
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.5914990901947021,
      "learning_rate": 0.0005827095014115489,
      "loss": 2.2136,
      "step": 25021
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6918637156486511,
      "learning_rate": 0.0005827081327379487,
      "loss": 2.3822,
      "step": 25022
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6219040155410767,
      "learning_rate": 0.0005827067640117876,
      "loss": 2.5388,
      "step": 25023
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6653903126716614,
      "learning_rate": 0.0005827053952330662,
      "loss": 2.4587,
      "step": 25024
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6190176606178284,
      "learning_rate": 0.0005827040264017843,
      "loss": 2.3063,
      "step": 25025
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.5890154242515564,
      "learning_rate": 0.0005827026575179425,
      "loss": 2.1525,
      "step": 25026
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.608376145362854,
      "learning_rate": 0.0005827012885815409,
      "loss": 2.2536,
      "step": 25027
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6313280463218689,
      "learning_rate": 0.0005826999195925798,
      "loss": 2.3183,
      "step": 25028
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.5977763533592224,
      "learning_rate": 0.0005826985505510594,
      "loss": 2.1283,
      "step": 25029
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.7578724026679993,
      "learning_rate": 0.00058269718145698,
      "loss": 2.4628,
      "step": 25030
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6577624678611755,
      "learning_rate": 0.000582695812310342,
      "loss": 2.3571,
      "step": 25031
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6847257614135742,
      "learning_rate": 0.0005826944431111454,
      "loss": 2.5479,
      "step": 25032
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6486998200416565,
      "learning_rate": 0.0005826930738593905,
      "loss": 2.4544,
      "step": 25033
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.7572203278541565,
      "learning_rate": 0.0005826917045550778,
      "loss": 2.482,
      "step": 25034
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6185376048088074,
      "learning_rate": 0.0005826903351982073,
      "loss": 2.6136,
      "step": 25035
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6796565055847168,
      "learning_rate": 0.0005826889657887793,
      "loss": 2.4463,
      "step": 25036
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.7013842463493347,
      "learning_rate": 0.0005826875963267941,
      "loss": 2.0575,
      "step": 25037
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6600250601768494,
      "learning_rate": 0.0005826862268122519,
      "loss": 2.5149,
      "step": 25038
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6387295126914978,
      "learning_rate": 0.0005826848572451531,
      "loss": 2.0967,
      "step": 25039
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6490746736526489,
      "learning_rate": 0.0005826834876254978,
      "loss": 2.3334,
      "step": 25040
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.7346084117889404,
      "learning_rate": 0.0005826821179532862,
      "loss": 2.3957,
      "step": 25041
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6431447267532349,
      "learning_rate": 0.0005826807482285188,
      "loss": 2.4591,
      "step": 25042
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.576877772808075,
      "learning_rate": 0.0005826793784511957,
      "loss": 2.3816,
      "step": 25043
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.668215274810791,
      "learning_rate": 0.0005826780086213171,
      "loss": 2.5619,
      "step": 25044
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.637320876121521,
      "learning_rate": 0.0005826766387388834,
      "loss": 2.2217,
      "step": 25045
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6136496663093567,
      "learning_rate": 0.0005826752688038947,
      "loss": 2.2889,
      "step": 25046
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6519791483879089,
      "learning_rate": 0.0005826738988163513,
      "loss": 2.3782,
      "step": 25047
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.7172661423683167,
      "learning_rate": 0.0005826725287762535,
      "loss": 2.4624,
      "step": 25048
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.5857719779014587,
      "learning_rate": 0.0005826711586836016,
      "loss": 2.1953,
      "step": 25049
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.629780113697052,
      "learning_rate": 0.0005826697885383958,
      "loss": 2.3378,
      "step": 25050
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6322205662727356,
      "learning_rate": 0.0005826684183406364,
      "loss": 2.5174,
      "step": 25051
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6067525744438171,
      "learning_rate": 0.0005826670480903235,
      "loss": 2.2497,
      "step": 25052
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.621862530708313,
      "learning_rate": 0.0005826656777874576,
      "loss": 2.0013,
      "step": 25053
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.5753516554832458,
      "learning_rate": 0.0005826643074320387,
      "loss": 2.2517,
      "step": 25054
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6531106233596802,
      "learning_rate": 0.0005826629370240671,
      "loss": 2.4465,
      "step": 25055
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.587192952632904,
      "learning_rate": 0.0005826615665635431,
      "loss": 2.5998,
      "step": 25056
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6213844418525696,
      "learning_rate": 0.0005826601960504672,
      "loss": 2.1166,
      "step": 25057
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6186853051185608,
      "learning_rate": 0.0005826588254848392,
      "loss": 2.3885,
      "step": 25058
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6690840125083923,
      "learning_rate": 0.0005826574548666597,
      "loss": 2.3328,
      "step": 25059
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.5578515529632568,
      "learning_rate": 0.000582656084195929,
      "loss": 2.3718,
      "step": 25060
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.568477988243103,
      "learning_rate": 0.000582654713472647,
      "loss": 2.3057,
      "step": 25061
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6082996726036072,
      "learning_rate": 0.0005826533426968142,
      "loss": 2.2183,
      "step": 25062
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6527823805809021,
      "learning_rate": 0.0005826519718684308,
      "loss": 2.3294,
      "step": 25063
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.5668207406997681,
      "learning_rate": 0.0005826506009874971,
      "loss": 1.9747,
      "step": 25064
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6212407350540161,
      "learning_rate": 0.0005826492300540132,
      "loss": 2.4187,
      "step": 25065
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6999427676200867,
      "learning_rate": 0.0005826478590679796,
      "loss": 2.3351,
      "step": 25066
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6369931101799011,
      "learning_rate": 0.0005826464880293964,
      "loss": 2.5463,
      "step": 25067
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6946811676025391,
      "learning_rate": 0.0005826451169382639,
      "loss": 2.2956,
      "step": 25068
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6335475444793701,
      "learning_rate": 0.0005826437457945823,
      "loss": 2.1494,
      "step": 25069
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6769834160804749,
      "learning_rate": 0.000582642374598352,
      "loss": 2.3795,
      "step": 25070
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.5820180773735046,
      "learning_rate": 0.0005826410033495731,
      "loss": 2.4221,
      "step": 25071
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6701614260673523,
      "learning_rate": 0.0005826396320482457,
      "loss": 2.3622,
      "step": 25072
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6599496006965637,
      "learning_rate": 0.0005826382606943705,
      "loss": 2.3566,
      "step": 25073
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6083779335021973,
      "learning_rate": 0.0005826368892879474,
      "loss": 2.195,
      "step": 25074
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6126435399055481,
      "learning_rate": 0.0005826355178289767,
      "loss": 2.3998,
      "step": 25075
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6687057614326477,
      "learning_rate": 0.0005826341463174589,
      "loss": 2.5291,
      "step": 25076
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6900810599327087,
      "learning_rate": 0.000582632774753394,
      "loss": 2.0913,
      "step": 25077
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.66558837890625,
      "learning_rate": 0.0005826314031367823,
      "loss": 2.3166,
      "step": 25078
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6296546459197998,
      "learning_rate": 0.0005826300314676241,
      "loss": 2.4531,
      "step": 25079
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.7850432991981506,
      "learning_rate": 0.0005826286597459196,
      "loss": 2.3417,
      "step": 25080
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.5953898429870605,
      "learning_rate": 0.0005826272879716692,
      "loss": 2.3482,
      "step": 25081
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.7869109511375427,
      "learning_rate": 0.0005826259161448729,
      "loss": 2.4855,
      "step": 25082
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.7314872741699219,
      "learning_rate": 0.0005826245442655313,
      "loss": 2.263,
      "step": 25083
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.7061389684677124,
      "learning_rate": 0.0005826231723336443,
      "loss": 2.4022,
      "step": 25084
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6319327354431152,
      "learning_rate": 0.0005826218003492124,
      "loss": 2.4482,
      "step": 25085
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6673714518547058,
      "learning_rate": 0.0005826204283122358,
      "loss": 2.1615,
      "step": 25086
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6768727898597717,
      "learning_rate": 0.0005826190562227147,
      "loss": 2.1327,
      "step": 25087
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.7042261362075806,
      "learning_rate": 0.0005826176840806493,
      "loss": 2.4896,
      "step": 25088
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.5845189094543457,
      "learning_rate": 0.00058261631188604,
      "loss": 2.3078,
      "step": 25089
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.612294614315033,
      "learning_rate": 0.0005826149396388871,
      "loss": 2.1481,
      "step": 25090
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.611247718334198,
      "learning_rate": 0.0005826135673391907,
      "loss": 2.1129,
      "step": 25091
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.683908998966217,
      "learning_rate": 0.0005826121949869509,
      "loss": 2.3208,
      "step": 25092
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6152042150497437,
      "learning_rate": 0.0005826108225821683,
      "loss": 2.1856,
      "step": 25093
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6579762697219849,
      "learning_rate": 0.0005826094501248431,
      "loss": 2.1794,
      "step": 25094
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6189337968826294,
      "learning_rate": 0.0005826080776149754,
      "loss": 2.3232,
      "step": 25095
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.5881708860397339,
      "learning_rate": 0.0005826067050525654,
      "loss": 2.2238,
      "step": 25096
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6156266927719116,
      "learning_rate": 0.0005826053324376137,
      "loss": 2.2358,
      "step": 25097
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6320299506187439,
      "learning_rate": 0.0005826039597701202,
      "loss": 2.3585,
      "step": 25098
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.5924741625785828,
      "learning_rate": 0.0005826025870500854,
      "loss": 2.2891,
      "step": 25099
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6318599581718445,
      "learning_rate": 0.0005826012142775092,
      "loss": 2.5039,
      "step": 25100
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6095775365829468,
      "learning_rate": 0.0005825998414523924,
      "loss": 2.2131,
      "step": 25101
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6510090231895447,
      "learning_rate": 0.0005825984685747347,
      "loss": 2.5422,
      "step": 25102
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6346328854560852,
      "learning_rate": 0.0005825970956445368,
      "loss": 2.3063,
      "step": 25103
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.7285169959068298,
      "learning_rate": 0.0005825957226617987,
      "loss": 2.4154,
      "step": 25104
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6561399102210999,
      "learning_rate": 0.0005825943496265207,
      "loss": 2.202,
      "step": 25105
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.5963577032089233,
      "learning_rate": 0.000582592976538703,
      "loss": 2.3144,
      "step": 25106
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6867563724517822,
      "learning_rate": 0.0005825916033983461,
      "loss": 2.4109,
      "step": 25107
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.7924195528030396,
      "learning_rate": 0.00058259023020545,
      "loss": 2.4715,
      "step": 25108
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.585354208946228,
      "learning_rate": 0.000582588856960015,
      "loss": 2.3271,
      "step": 25109
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.7875884771347046,
      "learning_rate": 0.0005825874836620415,
      "loss": 2.3915,
      "step": 25110
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6564565300941467,
      "learning_rate": 0.0005825861103115295,
      "loss": 2.2422,
      "step": 25111
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.5958511233329773,
      "learning_rate": 0.0005825847369084796,
      "loss": 2.2407,
      "step": 25112
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6269663572311401,
      "learning_rate": 0.0005825833634528918,
      "loss": 2.3585,
      "step": 25113
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6185362935066223,
      "learning_rate": 0.0005825819899447663,
      "loss": 2.1643,
      "step": 25114
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6662676334381104,
      "learning_rate": 0.0005825806163841037,
      "loss": 2.5958,
      "step": 25115
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6700215339660645,
      "learning_rate": 0.0005825792427709038,
      "loss": 2.4695,
      "step": 25116
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6370853781700134,
      "learning_rate": 0.0005825778691051672,
      "loss": 2.3675,
      "step": 25117
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6083910465240479,
      "learning_rate": 0.0005825764953868941,
      "loss": 2.3025,
      "step": 25118
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6384016275405884,
      "learning_rate": 0.0005825751216160845,
      "loss": 2.5403,
      "step": 25119
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.5950831770896912,
      "learning_rate": 0.0005825737477927389,
      "loss": 2.3991,
      "step": 25120
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6486724615097046,
      "learning_rate": 0.0005825723739168576,
      "loss": 2.2423,
      "step": 25121
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6079416275024414,
      "learning_rate": 0.0005825709999884408,
      "loss": 2.3948,
      "step": 25122
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6259569525718689,
      "learning_rate": 0.0005825696260074886,
      "loss": 2.2146,
      "step": 25123
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6080685257911682,
      "learning_rate": 0.0005825682519740014,
      "loss": 2.4367,
      "step": 25124
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6090734004974365,
      "learning_rate": 0.0005825668778879795,
      "loss": 1.9746,
      "step": 25125
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6561335325241089,
      "learning_rate": 0.000582565503749423,
      "loss": 2.5059,
      "step": 25126
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.650570809841156,
      "learning_rate": 0.0005825641295583323,
      "loss": 2.5065,
      "step": 25127
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6088061928749084,
      "learning_rate": 0.0005825627553147075,
      "loss": 2.2911,
      "step": 25128
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.5913824439048767,
      "learning_rate": 0.0005825613810185491,
      "loss": 2.3943,
      "step": 25129
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6409056186676025,
      "learning_rate": 0.000582560006669857,
      "loss": 2.5259,
      "step": 25130
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.5792309641838074,
      "learning_rate": 0.0005825586322686318,
      "loss": 2.1074,
      "step": 25131
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6572988629341125,
      "learning_rate": 0.0005825572578148737,
      "loss": 2.2283,
      "step": 25132
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6398695707321167,
      "learning_rate": 0.0005825558833085828,
      "loss": 2.4159,
      "step": 25133
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.5819849967956543,
      "learning_rate": 0.0005825545087497593,
      "loss": 2.3762,
      "step": 25134
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6282418966293335,
      "learning_rate": 0.0005825531341384037,
      "loss": 2.2766,
      "step": 25135
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6882180571556091,
      "learning_rate": 0.0005825517594745162,
      "loss": 2.428,
      "step": 25136
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6624480485916138,
      "learning_rate": 0.0005825503847580968,
      "loss": 2.2508,
      "step": 25137
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.660211443901062,
      "learning_rate": 0.0005825490099891461,
      "loss": 2.2913,
      "step": 25138
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6319132447242737,
      "learning_rate": 0.0005825476351676642,
      "loss": 2.4726,
      "step": 25139
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6065455675125122,
      "learning_rate": 0.0005825462602936513,
      "loss": 2.4859,
      "step": 25140
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.7429108023643494,
      "learning_rate": 0.0005825448853671078,
      "loss": 2.3096,
      "step": 25141
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.5887006521224976,
      "learning_rate": 0.0005825435103880337,
      "loss": 2.4696,
      "step": 25142
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6418167948722839,
      "learning_rate": 0.0005825421353564296,
      "loss": 2.3493,
      "step": 25143
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6385989189147949,
      "learning_rate": 0.0005825407602722955,
      "loss": 2.3251,
      "step": 25144
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.5898087024688721,
      "learning_rate": 0.0005825393851356318,
      "loss": 2.1597,
      "step": 25145
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6529752016067505,
      "learning_rate": 0.0005825380099464387,
      "loss": 2.3258,
      "step": 25146
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6965832114219666,
      "learning_rate": 0.0005825366347047163,
      "loss": 2.472,
      "step": 25147
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6420895457267761,
      "learning_rate": 0.0005825352594104652,
      "loss": 2.2712,
      "step": 25148
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.7459296584129333,
      "learning_rate": 0.0005825338840636854,
      "loss": 2.2373,
      "step": 25149
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.7293382287025452,
      "learning_rate": 0.0005825325086643772,
      "loss": 2.2884,
      "step": 25150
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.9834908843040466,
      "learning_rate": 0.0005825311332125409,
      "loss": 2.352,
      "step": 25151
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6762557625770569,
      "learning_rate": 0.0005825297577081766,
      "loss": 2.2716,
      "step": 25152
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.7134751081466675,
      "learning_rate": 0.0005825283821512849,
      "loss": 2.4339,
      "step": 25153
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.7462331652641296,
      "learning_rate": 0.0005825270065418657,
      "loss": 2.3935,
      "step": 25154
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.7495304346084595,
      "learning_rate": 0.0005825256308799195,
      "loss": 2.2709,
      "step": 25155
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6397241353988647,
      "learning_rate": 0.0005825242551654462,
      "loss": 2.1815,
      "step": 25156
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.7775985598564148,
      "learning_rate": 0.0005825228793984467,
      "loss": 2.3267,
      "step": 25157
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6863636374473572,
      "learning_rate": 0.0005825215035789205,
      "loss": 2.2345,
      "step": 25158
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.8563390970230103,
      "learning_rate": 0.0005825201277068684,
      "loss": 2.3776,
      "step": 25159
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6875479221343994,
      "learning_rate": 0.0005825187517822905,
      "loss": 2.4023,
      "step": 25160
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.9792277812957764,
      "learning_rate": 0.0005825173758051869,
      "loss": 2.325,
      "step": 25161
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.8056467771530151,
      "learning_rate": 0.0005825159997755582,
      "loss": 2.1615,
      "step": 25162
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6515629887580872,
      "learning_rate": 0.0005825146236934042,
      "loss": 2.4003,
      "step": 25163
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.8993028998374939,
      "learning_rate": 0.0005825132475587255,
      "loss": 2.2992,
      "step": 25164
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6985281109809875,
      "learning_rate": 0.0005825118713715222,
      "loss": 2.0557,
      "step": 25165
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6391345262527466,
      "learning_rate": 0.0005825104951317947,
      "loss": 2.5191,
      "step": 25166
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6239008903503418,
      "learning_rate": 0.0005825091188395432,
      "loss": 2.4603,
      "step": 25167
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.7496886849403381,
      "learning_rate": 0.0005825077424947678,
      "loss": 2.2976,
      "step": 25168
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.7140108942985535,
      "learning_rate": 0.000582506366097469,
      "loss": 2.4492,
      "step": 25169
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.7020326256752014,
      "learning_rate": 0.0005825049896476468,
      "loss": 2.3969,
      "step": 25170
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6240708827972412,
      "learning_rate": 0.0005825036131453017,
      "loss": 2.2622,
      "step": 25171
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6747208833694458,
      "learning_rate": 0.0005825022365904338,
      "loss": 2.2489,
      "step": 25172
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6982706189155579,
      "learning_rate": 0.0005825008599830433,
      "loss": 2.4006,
      "step": 25173
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6217158436775208,
      "learning_rate": 0.0005824994833231308,
      "loss": 2.3528,
      "step": 25174
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.563390851020813,
      "learning_rate": 0.0005824981066106962,
      "loss": 2.222,
      "step": 25175
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.69297194480896,
      "learning_rate": 0.0005824967298457398,
      "loss": 2.4855,
      "step": 25176
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6255553960800171,
      "learning_rate": 0.0005824953530282619,
      "loss": 2.116,
      "step": 25177
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.618230938911438,
      "learning_rate": 0.000582493976158263,
      "loss": 2.1119,
      "step": 25178
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6225988864898682,
      "learning_rate": 0.0005824925992357429,
      "loss": 2.1879,
      "step": 25179
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.7345491051673889,
      "learning_rate": 0.0005824912222607022,
      "loss": 2.3954,
      "step": 25180
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6013068556785583,
      "learning_rate": 0.000582489845233141,
      "loss": 2.2595,
      "step": 25181
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.8541828393936157,
      "learning_rate": 0.0005824884681530597,
      "loss": 2.431,
      "step": 25182
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6844406127929688,
      "learning_rate": 0.0005824870910204583,
      "loss": 2.5445,
      "step": 25183
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6546546220779419,
      "learning_rate": 0.0005824857138353374,
      "loss": 2.2536,
      "step": 25184
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6696547269821167,
      "learning_rate": 0.000582484336597697,
      "loss": 2.5168,
      "step": 25185
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.7538850903511047,
      "learning_rate": 0.0005824829593075374,
      "loss": 2.4833,
      "step": 25186
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6303116679191589,
      "learning_rate": 0.0005824815819648588,
      "loss": 2.1642,
      "step": 25187
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6922826170921326,
      "learning_rate": 0.0005824802045696618,
      "loss": 2.4204,
      "step": 25188
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6261059045791626,
      "learning_rate": 0.0005824788271219461,
      "loss": 2.2949,
      "step": 25189
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.702314019203186,
      "learning_rate": 0.0005824774496217126,
      "loss": 2.4148,
      "step": 25190
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.5998620390892029,
      "learning_rate": 0.0005824760720689609,
      "loss": 2.4983,
      "step": 25191
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6548731327056885,
      "learning_rate": 0.0005824746944636917,
      "loss": 2.3285,
      "step": 25192
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6919648051261902,
      "learning_rate": 0.0005824733168059051,
      "loss": 2.3522,
      "step": 25193
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6227542757987976,
      "learning_rate": 0.0005824719390956014,
      "loss": 2.3049,
      "step": 25194
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6599626541137695,
      "learning_rate": 0.0005824705613327807,
      "loss": 2.3463,
      "step": 25195
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.7694052457809448,
      "learning_rate": 0.0005824691835174436,
      "loss": 2.5112,
      "step": 25196
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.652654230594635,
      "learning_rate": 0.00058246780564959,
      "loss": 2.2728,
      "step": 25197
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.544575572013855,
      "learning_rate": 0.0005824664277292203,
      "loss": 2.0907,
      "step": 25198
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.8726235628128052,
      "learning_rate": 0.0005824650497563348,
      "loss": 2.4253,
      "step": 25199
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6666215062141418,
      "learning_rate": 0.0005824636717309338,
      "loss": 2.1258,
      "step": 25200
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.5960062742233276,
      "learning_rate": 0.0005824622936530174,
      "loss": 2.2453,
      "step": 25201
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6779007315635681,
      "learning_rate": 0.0005824609155225859,
      "loss": 2.4203,
      "step": 25202
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6587713956832886,
      "learning_rate": 0.0005824595373396396,
      "loss": 2.2463,
      "step": 25203
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6573001742362976,
      "learning_rate": 0.0005824581591041787,
      "loss": 2.5128,
      "step": 25204
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6544927358627319,
      "learning_rate": 0.0005824567808162035,
      "loss": 2.6057,
      "step": 25205
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6221027374267578,
      "learning_rate": 0.0005824554024757144,
      "loss": 2.5969,
      "step": 25206
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6120738983154297,
      "learning_rate": 0.0005824540240827113,
      "loss": 2.2687,
      "step": 25207
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.624608039855957,
      "learning_rate": 0.0005824526456371949,
      "loss": 2.3859,
      "step": 25208
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.771477997303009,
      "learning_rate": 0.0005824512671391651,
      "loss": 2.4532,
      "step": 25209
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6295265555381775,
      "learning_rate": 0.0005824498885886222,
      "loss": 2.1918,
      "step": 25210
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.8053866028785706,
      "learning_rate": 0.0005824485099855666,
      "loss": 2.3001,
      "step": 25211
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6340405344963074,
      "learning_rate": 0.0005824471313299986,
      "loss": 2.2337,
      "step": 25212
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6273051500320435,
      "learning_rate": 0.0005824457526219181,
      "loss": 2.1492,
      "step": 25213
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6149004697799683,
      "learning_rate": 0.0005824443738613258,
      "loss": 2.4409,
      "step": 25214
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6446276903152466,
      "learning_rate": 0.0005824429950482217,
      "loss": 2.3832,
      "step": 25215
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6155521273612976,
      "learning_rate": 0.0005824416161826062,
      "loss": 2.3816,
      "step": 25216
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6113007068634033,
      "learning_rate": 0.0005824402372644794,
      "loss": 2.2103,
      "step": 25217
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6760789752006531,
      "learning_rate": 0.0005824388582938415,
      "loss": 2.2476,
      "step": 25218
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.636688768863678,
      "learning_rate": 0.0005824374792706931,
      "loss": 2.2917,
      "step": 25219
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6719199419021606,
      "learning_rate": 0.0005824361001950341,
      "loss": 2.4541,
      "step": 25220
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.5923079252243042,
      "learning_rate": 0.000582434721066865,
      "loss": 2.0381,
      "step": 25221
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.5808064341545105,
      "learning_rate": 0.0005824333418861859,
      "loss": 2.2978,
      "step": 25222
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6829557418823242,
      "learning_rate": 0.000582431962652997,
      "loss": 2.3108,
      "step": 25223
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6608734726905823,
      "learning_rate": 0.0005824305833672988,
      "loss": 2.333,
      "step": 25224
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6521112322807312,
      "learning_rate": 0.0005824292040290914,
      "loss": 2.2788,
      "step": 25225
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6707218885421753,
      "learning_rate": 0.0005824278246383751,
      "loss": 2.2462,
      "step": 25226
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6331016421318054,
      "learning_rate": 0.0005824264451951501,
      "loss": 2.3303,
      "step": 25227
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6156286001205444,
      "learning_rate": 0.0005824250656994167,
      "loss": 2.3261,
      "step": 25228
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.8094543814659119,
      "learning_rate": 0.0005824236861511751,
      "loss": 2.3604,
      "step": 25229
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6099384427070618,
      "learning_rate": 0.0005824223065504258,
      "loss": 2.3668,
      "step": 25230
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.612372100353241,
      "learning_rate": 0.0005824209268971687,
      "loss": 2.2568,
      "step": 25231
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6243273019790649,
      "learning_rate": 0.0005824195471914042,
      "loss": 2.2531,
      "step": 25232
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6351688504219055,
      "learning_rate": 0.0005824181674331326,
      "loss": 2.1851,
      "step": 25233
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6689608097076416,
      "learning_rate": 0.0005824167876223542,
      "loss": 2.5271,
      "step": 25234
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6499654054641724,
      "learning_rate": 0.0005824154077590691,
      "loss": 2.3454,
      "step": 25235
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6732881665229797,
      "learning_rate": 0.0005824140278432777,
      "loss": 2.3473,
      "step": 25236
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6245930194854736,
      "learning_rate": 0.0005824126478749802,
      "loss": 2.2561,
      "step": 25237
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6349846720695496,
      "learning_rate": 0.0005824112678541768,
      "loss": 2.3409,
      "step": 25238
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.5622334480285645,
      "learning_rate": 0.0005824098877808678,
      "loss": 2.367,
      "step": 25239
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6046833992004395,
      "learning_rate": 0.0005824085076550536,
      "loss": 2.203,
      "step": 25240
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.66131591796875,
      "learning_rate": 0.0005824071274767342,
      "loss": 2.3337,
      "step": 25241
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6161395311355591,
      "learning_rate": 0.0005824057472459101,
      "loss": 2.107,
      "step": 25242
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6973391175270081,
      "learning_rate": 0.0005824043669625813,
      "loss": 2.3655,
      "step": 25243
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6373895406723022,
      "learning_rate": 0.0005824029866267484,
      "loss": 2.3141,
      "step": 25244
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6927968859672546,
      "learning_rate": 0.0005824016062384112,
      "loss": 2.2944,
      "step": 25245
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6303735375404358,
      "learning_rate": 0.0005824002257975704,
      "loss": 2.2931,
      "step": 25246
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.584437370300293,
      "learning_rate": 0.0005823988453042261,
      "loss": 2.2614,
      "step": 25247
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6108043193817139,
      "learning_rate": 0.0005823974647583783,
      "loss": 2.3024,
      "step": 25248
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6352493762969971,
      "learning_rate": 0.0005823960841600277,
      "loss": 2.3943,
      "step": 25249
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.632147490978241,
      "learning_rate": 0.0005823947035091743,
      "loss": 2.3938,
      "step": 25250
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6669321656227112,
      "learning_rate": 0.0005823933228058183,
      "loss": 2.4034,
      "step": 25251
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6267411112785339,
      "learning_rate": 0.00058239194204996,
      "loss": 2.3736,
      "step": 25252
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6381189823150635,
      "learning_rate": 0.0005823905612415999,
      "loss": 2.3328,
      "step": 25253
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6621772050857544,
      "learning_rate": 0.0005823891803807379,
      "loss": 2.4625,
      "step": 25254
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.7988758683204651,
      "learning_rate": 0.0005823877994673745,
      "loss": 2.4232,
      "step": 25255
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6943254470825195,
      "learning_rate": 0.0005823864185015099,
      "loss": 2.2868,
      "step": 25256
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6589471101760864,
      "learning_rate": 0.0005823850374831442,
      "loss": 2.1734,
      "step": 25257
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.7224122285842896,
      "learning_rate": 0.000582383656412278,
      "loss": 2.1724,
      "step": 25258
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.8174529671669006,
      "learning_rate": 0.0005823822752889111,
      "loss": 2.3134,
      "step": 25259
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.607352614402771,
      "learning_rate": 0.0005823808941130442,
      "loss": 2.0294,
      "step": 25260
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.5992041826248169,
      "learning_rate": 0.0005823795128846774,
      "loss": 2.2893,
      "step": 25261
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6341364979743958,
      "learning_rate": 0.0005823781316038108,
      "loss": 2.4789,
      "step": 25262
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.7166840434074402,
      "learning_rate": 0.0005823767502704447,
      "loss": 2.4136,
      "step": 25263
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6597305536270142,
      "learning_rate": 0.0005823753688845796,
      "loss": 2.3256,
      "step": 25264
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6607782244682312,
      "learning_rate": 0.0005823739874462155,
      "loss": 2.2729,
      "step": 25265
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6293292045593262,
      "learning_rate": 0.0005823726059553528,
      "loss": 2.155,
      "step": 25266
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.7143867015838623,
      "learning_rate": 0.0005823712244119916,
      "loss": 2.2691,
      "step": 25267
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6031273007392883,
      "learning_rate": 0.0005823698428161323,
      "loss": 2.458,
      "step": 25268
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.5991767048835754,
      "learning_rate": 0.0005823684611677753,
      "loss": 2.5465,
      "step": 25269
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.7961958646774292,
      "learning_rate": 0.0005823670794669204,
      "loss": 2.4702,
      "step": 25270
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6135521531105042,
      "learning_rate": 0.0005823656977135683,
      "loss": 2.2277,
      "step": 25271
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6396005153656006,
      "learning_rate": 0.000582364315907719,
      "loss": 2.3014,
      "step": 25272
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6288231015205383,
      "learning_rate": 0.0005823629340493728,
      "loss": 2.3812,
      "step": 25273
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6256171464920044,
      "learning_rate": 0.0005823615521385301,
      "loss": 2.4115,
      "step": 25274
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6840924024581909,
      "learning_rate": 0.0005823601701751911,
      "loss": 2.6368,
      "step": 25275
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.664997398853302,
      "learning_rate": 0.000582358788159356,
      "loss": 2.345,
      "step": 25276
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6117044687271118,
      "learning_rate": 0.000582357406091025,
      "loss": 2.4484,
      "step": 25277
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6001471877098083,
      "learning_rate": 0.0005823560239701984,
      "loss": 2.4749,
      "step": 25278
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.5955349802970886,
      "learning_rate": 0.0005823546417968765,
      "loss": 2.2932,
      "step": 25279
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.7213928699493408,
      "learning_rate": 0.0005823532595710596,
      "loss": 2.3901,
      "step": 25280
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6229369640350342,
      "learning_rate": 0.0005823518772927478,
      "loss": 2.3873,
      "step": 25281
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6679911017417908,
      "learning_rate": 0.0005823504949619416,
      "loss": 2.4932,
      "step": 25282
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.7543113231658936,
      "learning_rate": 0.000582349112578641,
      "loss": 2.0937,
      "step": 25283
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.5912857055664062,
      "learning_rate": 0.0005823477301428463,
      "loss": 2.2434,
      "step": 25284
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6530205607414246,
      "learning_rate": 0.0005823463476545581,
      "loss": 2.5387,
      "step": 25285
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6591458916664124,
      "learning_rate": 0.0005823449651137762,
      "loss": 2.4452,
      "step": 25286
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.8526555895805359,
      "learning_rate": 0.000582343582520501,
      "loss": 2.2617,
      "step": 25287
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6119113564491272,
      "learning_rate": 0.000582342199874733,
      "loss": 2.338,
      "step": 25288
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.8525633811950684,
      "learning_rate": 0.0005823408171764721,
      "loss": 2.4692,
      "step": 25289
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6892998218536377,
      "learning_rate": 0.0005823394344257187,
      "loss": 2.2017,
      "step": 25290
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6398141980171204,
      "learning_rate": 0.0005823380516224731,
      "loss": 2.3556,
      "step": 25291
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6946249008178711,
      "learning_rate": 0.0005823366687667355,
      "loss": 2.3511,
      "step": 25292
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.9047562479972839,
      "learning_rate": 0.0005823352858585062,
      "loss": 2.415,
      "step": 25293
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6214971542358398,
      "learning_rate": 0.0005823339028977855,
      "loss": 2.3835,
      "step": 25294
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6386387944221497,
      "learning_rate": 0.0005823325198845735,
      "loss": 2.3111,
      "step": 25295
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.7392086386680603,
      "learning_rate": 0.0005823311368188706,
      "loss": 2.2409,
      "step": 25296
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6756726503372192,
      "learning_rate": 0.000582329753700677,
      "loss": 2.4133,
      "step": 25297
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6053128242492676,
      "learning_rate": 0.000582328370529993,
      "loss": 2.3514,
      "step": 25298
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6300166249275208,
      "learning_rate": 0.0005823269873068189,
      "loss": 2.1573,
      "step": 25299
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.7304731607437134,
      "learning_rate": 0.0005823256040311548,
      "loss": 2.3125,
      "step": 25300
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.7137007713317871,
      "learning_rate": 0.000582324220703001,
      "loss": 2.3676,
      "step": 25301
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6529538631439209,
      "learning_rate": 0.0005823228373223579,
      "loss": 2.2581,
      "step": 25302
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6741687059402466,
      "learning_rate": 0.0005823214538892255,
      "loss": 2.241,
      "step": 25303
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.7580272555351257,
      "learning_rate": 0.0005823200704036043,
      "loss": 2.4313,
      "step": 25304
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6820244789123535,
      "learning_rate": 0.0005823186868654945,
      "loss": 2.4278,
      "step": 25305
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6412776708602905,
      "learning_rate": 0.0005823173032748962,
      "loss": 2.3643,
      "step": 25306
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6239842772483826,
      "learning_rate": 0.00058231591963181,
      "loss": 2.4748,
      "step": 25307
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.8069377541542053,
      "learning_rate": 0.0005823145359362358,
      "loss": 2.2648,
      "step": 25308
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.626611590385437,
      "learning_rate": 0.0005823131521881741,
      "loss": 2.4436,
      "step": 25309
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6350750923156738,
      "learning_rate": 0.0005823117683876249,
      "loss": 2.268,
      "step": 25310
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6296165585517883,
      "learning_rate": 0.0005823103845345887,
      "loss": 2.2176,
      "step": 25311
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.603093147277832,
      "learning_rate": 0.0005823090006290657,
      "loss": 2.3995,
      "step": 25312
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6644212007522583,
      "learning_rate": 0.000582307616671056,
      "loss": 2.3453,
      "step": 25313
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6268483996391296,
      "learning_rate": 0.0005823062326605601,
      "loss": 2.2564,
      "step": 25314
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6486343145370483,
      "learning_rate": 0.0005823048485975781,
      "loss": 2.2704,
      "step": 25315
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6861788630485535,
      "learning_rate": 0.0005823034644821104,
      "loss": 2.2714,
      "step": 25316
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.771150529384613,
      "learning_rate": 0.0005823020803141571,
      "loss": 2.3174,
      "step": 25317
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.5745494365692139,
      "learning_rate": 0.0005823006960937185,
      "loss": 2.5295,
      "step": 25318
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6249558925628662,
      "learning_rate": 0.0005822993118207948,
      "loss": 2.201,
      "step": 25319
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6480101346969604,
      "learning_rate": 0.0005822979274953863,
      "loss": 2.3625,
      "step": 25320
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6481195092201233,
      "learning_rate": 0.0005822965431174935,
      "loss": 2.4123,
      "step": 25321
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.616908609867096,
      "learning_rate": 0.0005822951586871162,
      "loss": 2.3979,
      "step": 25322
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6471664309501648,
      "learning_rate": 0.0005822937742042551,
      "loss": 2.275,
      "step": 25323
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6229555606842041,
      "learning_rate": 0.0005822923896689102,
      "loss": 2.3951,
      "step": 25324
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.7199602127075195,
      "learning_rate": 0.0005822910050810818,
      "loss": 2.2342,
      "step": 25325
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6893252730369568,
      "learning_rate": 0.0005822896204407702,
      "loss": 2.5104,
      "step": 25326
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6709836721420288,
      "learning_rate": 0.0005822882357479755,
      "loss": 2.3223,
      "step": 25327
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6297069191932678,
      "learning_rate": 0.0005822868510026982,
      "loss": 2.3002,
      "step": 25328
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6340252161026001,
      "learning_rate": 0.0005822854662049385,
      "loss": 2.3183,
      "step": 25329
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6472325921058655,
      "learning_rate": 0.0005822840813546965,
      "loss": 2.0562,
      "step": 25330
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6418537497520447,
      "learning_rate": 0.0005822826964519726,
      "loss": 2.3422,
      "step": 25331
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.5900433659553528,
      "learning_rate": 0.0005822813114967671,
      "loss": 2.3614,
      "step": 25332
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.7074799537658691,
      "learning_rate": 0.00058227992648908,
      "loss": 2.2743,
      "step": 25333
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6404757499694824,
      "learning_rate": 0.0005822785414289119,
      "loss": 2.3481,
      "step": 25334
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6851840615272522,
      "learning_rate": 0.0005822771563162629,
      "loss": 2.2694,
      "step": 25335
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6136655807495117,
      "learning_rate": 0.0005822757711511331,
      "loss": 2.2964,
      "step": 25336
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6080067753791809,
      "learning_rate": 0.0005822743859335229,
      "loss": 2.2079,
      "step": 25337
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6456789374351501,
      "learning_rate": 0.0005822730006634326,
      "loss": 2.2967,
      "step": 25338
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6026651859283447,
      "learning_rate": 0.0005822716153408626,
      "loss": 2.372,
      "step": 25339
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.7090078592300415,
      "learning_rate": 0.0005822702299658127,
      "loss": 2.1191,
      "step": 25340
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.5992605686187744,
      "learning_rate": 0.0005822688445382836,
      "loss": 2.417,
      "step": 25341
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6732155084609985,
      "learning_rate": 0.0005822674590582754,
      "loss": 2.348,
      "step": 25342
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6347072124481201,
      "learning_rate": 0.0005822660735257883,
      "loss": 2.1776,
      "step": 25343
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6008303761482239,
      "learning_rate": 0.0005822646879408227,
      "loss": 2.1283,
      "step": 25344
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6595344543457031,
      "learning_rate": 0.0005822633023033787,
      "loss": 2.4311,
      "step": 25345
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.7279118299484253,
      "learning_rate": 0.0005822619166134565,
      "loss": 2.2762,
      "step": 25346
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6616912484169006,
      "learning_rate": 0.0005822605308710567,
      "loss": 2.4614,
      "step": 25347
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6489744782447815,
      "learning_rate": 0.0005822591450761793,
      "loss": 2.5209,
      "step": 25348
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.660903811454773,
      "learning_rate": 0.0005822577592288244,
      "loss": 2.3228,
      "step": 25349
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6312288641929626,
      "learning_rate": 0.0005822563733289926,
      "loss": 2.2887,
      "step": 25350
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.5662280917167664,
      "learning_rate": 0.000582254987376684,
      "loss": 2.2579,
      "step": 25351
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6525977253913879,
      "learning_rate": 0.0005822536013718989,
      "loss": 2.1216,
      "step": 25352
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6905461549758911,
      "learning_rate": 0.0005822522153146374,
      "loss": 2.3429,
      "step": 25353
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6389736533164978,
      "learning_rate": 0.0005822508292049,
      "loss": 2.3079,
      "step": 25354
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6588981747627258,
      "learning_rate": 0.0005822494430426869,
      "loss": 2.586,
      "step": 25355
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6708357930183411,
      "learning_rate": 0.0005822480568279982,
      "loss": 2.2371,
      "step": 25356
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6129299998283386,
      "learning_rate": 0.0005822466705608343,
      "loss": 2.3644,
      "step": 25357
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6034036874771118,
      "learning_rate": 0.0005822452842411954,
      "loss": 2.3162,
      "step": 25358
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6927869319915771,
      "learning_rate": 0.0005822438978690818,
      "loss": 2.4748,
      "step": 25359
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6398517489433289,
      "learning_rate": 0.0005822425114444937,
      "loss": 2.3482,
      "step": 25360
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.5791757106781006,
      "learning_rate": 0.0005822411249674314,
      "loss": 2.3953,
      "step": 25361
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.614013671875,
      "learning_rate": 0.0005822397384378952,
      "loss": 2.3988,
      "step": 25362
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.7007092833518982,
      "learning_rate": 0.0005822383518558851,
      "loss": 2.5064,
      "step": 25363
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6474355459213257,
      "learning_rate": 0.0005822369652214018,
      "loss": 2.5337,
      "step": 25364
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.590957760810852,
      "learning_rate": 0.0005822355785344451,
      "loss": 2.2803,
      "step": 25365
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6467776298522949,
      "learning_rate": 0.0005822341917950157,
      "loss": 2.4844,
      "step": 25366
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6532199382781982,
      "learning_rate": 0.0005822328050031136,
      "loss": 2.3897,
      "step": 25367
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6262290477752686,
      "learning_rate": 0.000582231418158739,
      "loss": 2.2757,
      "step": 25368
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.5701581835746765,
      "learning_rate": 0.0005822300312618923,
      "loss": 2.4066,
      "step": 25369
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6145760416984558,
      "learning_rate": 0.0005822286443125736,
      "loss": 2.4644,
      "step": 25370
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6221158504486084,
      "learning_rate": 0.0005822272573107834,
      "loss": 2.2419,
      "step": 25371
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6344056725502014,
      "learning_rate": 0.0005822258702565218,
      "loss": 2.2595,
      "step": 25372
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6300283670425415,
      "learning_rate": 0.0005822244831497891,
      "loss": 2.5374,
      "step": 25373
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6487753391265869,
      "learning_rate": 0.0005822230959905856,
      "loss": 2.3117,
      "step": 25374
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6115468740463257,
      "learning_rate": 0.0005822217087789114,
      "loss": 2.3181,
      "step": 25375
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6253764629364014,
      "learning_rate": 0.0005822203215147669,
      "loss": 2.384,
      "step": 25376
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6181463599205017,
      "learning_rate": 0.0005822189341981522,
      "loss": 2.4566,
      "step": 25377
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6580487489700317,
      "learning_rate": 0.0005822175468290678,
      "loss": 2.5228,
      "step": 25378
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6560379266738892,
      "learning_rate": 0.0005822161594075138,
      "loss": 2.3854,
      "step": 25379
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6092599630355835,
      "learning_rate": 0.0005822147719334905,
      "loss": 2.4459,
      "step": 25380
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.5797522664070129,
      "learning_rate": 0.0005822133844069981,
      "loss": 2.2351,
      "step": 25381
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6146755814552307,
      "learning_rate": 0.0005822119968280369,
      "loss": 1.9655,
      "step": 25382
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6328467130661011,
      "learning_rate": 0.0005822106091966072,
      "loss": 2.335,
      "step": 25383
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6333959698677063,
      "learning_rate": 0.0005822092215127092,
      "loss": 2.499,
      "step": 25384
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.5993067026138306,
      "learning_rate": 0.0005822078337763433,
      "loss": 2.3169,
      "step": 25385
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.5979703664779663,
      "learning_rate": 0.0005822064459875094,
      "loss": 2.4406,
      "step": 25386
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.7483640909194946,
      "learning_rate": 0.0005822050581462081,
      "loss": 2.46,
      "step": 25387
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6587545871734619,
      "learning_rate": 0.0005822036702524396,
      "loss": 2.1635,
      "step": 25388
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.7735824584960938,
      "learning_rate": 0.0005822022823062041,
      "loss": 2.2891,
      "step": 25389
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.5596444010734558,
      "learning_rate": 0.0005822008943075019,
      "loss": 2.1192,
      "step": 25390
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6375821232795715,
      "learning_rate": 0.0005821995062563331,
      "loss": 2.3367,
      "step": 25391
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6005914211273193,
      "learning_rate": 0.0005821981181526983,
      "loss": 2.3627,
      "step": 25392
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6992703676223755,
      "learning_rate": 0.0005821967299965973,
      "loss": 2.4765,
      "step": 25393
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6172500252723694,
      "learning_rate": 0.0005821953417880308,
      "loss": 2.3311,
      "step": 25394
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.5876532793045044,
      "learning_rate": 0.0005821939535269987,
      "loss": 2.1776,
      "step": 25395
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6500244736671448,
      "learning_rate": 0.0005821925652135016,
      "loss": 2.3979,
      "step": 25396
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.7010473012924194,
      "learning_rate": 0.0005821911768475393,
      "loss": 2.5323,
      "step": 25397
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6795535087585449,
      "learning_rate": 0.0005821897884291127,
      "loss": 2.295,
      "step": 25398
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6230847835540771,
      "learning_rate": 0.0005821883999582215,
      "loss": 2.3044,
      "step": 25399
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.830053448677063,
      "learning_rate": 0.000582187011434866,
      "loss": 2.3707,
      "step": 25400
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.7413055896759033,
      "learning_rate": 0.0005821856228590468,
      "loss": 2.331,
      "step": 25401
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6743507981300354,
      "learning_rate": 0.0005821842342307639,
      "loss": 2.3886,
      "step": 25402
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.7434960603713989,
      "learning_rate": 0.0005821828455500177,
      "loss": 2.4432,
      "step": 25403
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.7646614909172058,
      "learning_rate": 0.0005821814568168083,
      "loss": 2.2981,
      "step": 25404
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6249368190765381,
      "learning_rate": 0.000582180068031136,
      "loss": 2.2034,
      "step": 25405
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.7327839732170105,
      "learning_rate": 0.0005821786791930012,
      "loss": 2.3035,
      "step": 25406
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.7469154596328735,
      "learning_rate": 0.000582177290302404,
      "loss": 2.2494,
      "step": 25407
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6479851007461548,
      "learning_rate": 0.0005821759013593448,
      "loss": 2.5181,
      "step": 25408
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6412784457206726,
      "learning_rate": 0.0005821745123638236,
      "loss": 2.3775,
      "step": 25409
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.7476626634597778,
      "learning_rate": 0.000582173123315841,
      "loss": 2.558,
      "step": 25410
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6774997711181641,
      "learning_rate": 0.000582171734215397,
      "loss": 2.3442,
      "step": 25411
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.785960853099823,
      "learning_rate": 0.000582170345062492,
      "loss": 2.3538,
      "step": 25412
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6532281637191772,
      "learning_rate": 0.0005821689558571262,
      "loss": 2.3158,
      "step": 25413
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.9473452568054199,
      "learning_rate": 0.0005821675665992999,
      "loss": 2.3065,
      "step": 25414
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.640529215335846,
      "learning_rate": 0.0005821661772890132,
      "loss": 2.1744,
      "step": 25415
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.606204092502594,
      "learning_rate": 0.0005821647879262666,
      "loss": 2.3401,
      "step": 25416
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.629144012928009,
      "learning_rate": 0.0005821633985110601,
      "loss": 2.189,
      "step": 25417
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6516357064247131,
      "learning_rate": 0.0005821620090433942,
      "loss": 2.2497,
      "step": 25418
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.713056206703186,
      "learning_rate": 0.000582160619523269,
      "loss": 2.3068,
      "step": 25419
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6731119751930237,
      "learning_rate": 0.000582159229950685,
      "loss": 2.3746,
      "step": 25420
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.7987262606620789,
      "learning_rate": 0.000582157840325642,
      "loss": 2.3188,
      "step": 25421
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6598108410835266,
      "learning_rate": 0.0005821564506481407,
      "loss": 2.286,
      "step": 25422
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6272817254066467,
      "learning_rate": 0.0005821550609181812,
      "loss": 2.2746,
      "step": 25423
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6288565993309021,
      "learning_rate": 0.0005821536711357637,
      "loss": 2.3287,
      "step": 25424
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6239264011383057,
      "learning_rate": 0.0005821522813008885,
      "loss": 2.26,
      "step": 25425
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6503636240959167,
      "learning_rate": 0.0005821508914135558,
      "loss": 2.2877,
      "step": 25426
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.588262140750885,
      "learning_rate": 0.0005821495014737661,
      "loss": 2.3081,
      "step": 25427
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6212794184684753,
      "learning_rate": 0.0005821481114815193,
      "loss": 2.4377,
      "step": 25428
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.621067225933075,
      "learning_rate": 0.000582146721436816,
      "loss": 2.297,
      "step": 25429
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6027593016624451,
      "learning_rate": 0.0005821453313396562,
      "loss": 2.104,
      "step": 25430
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6626054048538208,
      "learning_rate": 0.0005821439411900402,
      "loss": 2.2321,
      "step": 25431
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6241121292114258,
      "learning_rate": 0.0005821425509879684,
      "loss": 2.3649,
      "step": 25432
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.7974205017089844,
      "learning_rate": 0.000582141160733441,
      "loss": 2.4076,
      "step": 25433
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.7461347579956055,
      "learning_rate": 0.0005821397704264582,
      "loss": 2.0779,
      "step": 25434
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6409624218940735,
      "learning_rate": 0.0005821383800670203,
      "loss": 2.3208,
      "step": 25435
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.7412526607513428,
      "learning_rate": 0.0005821369896551275,
      "loss": 2.5926,
      "step": 25436
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6028646230697632,
      "learning_rate": 0.0005821355991907801,
      "loss": 2.3314,
      "step": 25437
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6465893983840942,
      "learning_rate": 0.0005821342086739785,
      "loss": 2.3431,
      "step": 25438
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6226933598518372,
      "learning_rate": 0.0005821328181047227,
      "loss": 2.1793,
      "step": 25439
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6456241011619568,
      "learning_rate": 0.0005821314274830131,
      "loss": 2.1939,
      "step": 25440
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.7125493288040161,
      "learning_rate": 0.00058213003680885,
      "loss": 2.3108,
      "step": 25441
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6001982688903809,
      "learning_rate": 0.0005821286460822335,
      "loss": 2.2196,
      "step": 25442
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6853473782539368,
      "learning_rate": 0.000582127255303164,
      "loss": 2.28,
      "step": 25443
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.7317748069763184,
      "learning_rate": 0.0005821258644716418,
      "loss": 2.2648,
      "step": 25444
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6889416575431824,
      "learning_rate": 0.0005821244735876671,
      "loss": 2.3014,
      "step": 25445
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6680094003677368,
      "learning_rate": 0.00058212308265124,
      "loss": 2.1618,
      "step": 25446
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6660802364349365,
      "learning_rate": 0.000582121691662361,
      "loss": 2.2479,
      "step": 25447
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.7307655215263367,
      "learning_rate": 0.0005821203006210301,
      "loss": 2.3736,
      "step": 25448
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6296443343162537,
      "learning_rate": 0.0005821189095272479,
      "loss": 2.4577,
      "step": 25449
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.5961489677429199,
      "learning_rate": 0.0005821175183810143,
      "loss": 2.3005,
      "step": 25450
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.7632931470870972,
      "learning_rate": 0.0005821161271823298,
      "loss": 2.2136,
      "step": 25451
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6763167381286621,
      "learning_rate": 0.0005821147359311946,
      "loss": 2.307,
      "step": 25452
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6372260451316833,
      "learning_rate": 0.0005821133446276089,
      "loss": 2.5095,
      "step": 25453
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.600927472114563,
      "learning_rate": 0.000582111953271573,
      "loss": 2.3728,
      "step": 25454
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6318042278289795,
      "learning_rate": 0.0005821105618630872,
      "loss": 2.3489,
      "step": 25455
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.680992841720581,
      "learning_rate": 0.0005821091704021517,
      "loss": 2.3331,
      "step": 25456
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6012474298477173,
      "learning_rate": 0.0005821077788887669,
      "loss": 2.3861,
      "step": 25457
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.5685482621192932,
      "learning_rate": 0.0005821063873229327,
      "loss": 2.0083,
      "step": 25458
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.7284665107727051,
      "learning_rate": 0.0005821049957046499,
      "loss": 2.4703,
      "step": 25459
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6353099942207336,
      "learning_rate": 0.0005821036040339182,
      "loss": 2.439,
      "step": 25460
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.64866703748703,
      "learning_rate": 0.0005821022123107382,
      "loss": 2.563,
      "step": 25461
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.591379702091217,
      "learning_rate": 0.0005821008205351101,
      "loss": 2.3786,
      "step": 25462
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6477964520454407,
      "learning_rate": 0.0005820994287070342,
      "loss": 2.4613,
      "step": 25463
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6843164563179016,
      "learning_rate": 0.0005820980368265106,
      "loss": 2.4516,
      "step": 25464
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6958979964256287,
      "learning_rate": 0.0005820966448935395,
      "loss": 2.6181,
      "step": 25465
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6486101746559143,
      "learning_rate": 0.0005820952529081216,
      "loss": 2.3868,
      "step": 25466
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6000362634658813,
      "learning_rate": 0.0005820938608702567,
      "loss": 2.3477,
      "step": 25467
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.5924211740493774,
      "learning_rate": 0.0005820924687799453,
      "loss": 2.2319,
      "step": 25468
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6030301451683044,
      "learning_rate": 0.0005820910766371876,
      "loss": 2.4946,
      "step": 25469
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6554051637649536,
      "learning_rate": 0.0005820896844419837,
      "loss": 2.3818,
      "step": 25470
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.622749924659729,
      "learning_rate": 0.0005820882921943341,
      "loss": 2.2062,
      "step": 25471
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.8427638411521912,
      "learning_rate": 0.000582086899894239,
      "loss": 2.1776,
      "step": 25472
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6438499093055725,
      "learning_rate": 0.0005820855075416985,
      "loss": 2.146,
      "step": 25473
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6612287759780884,
      "learning_rate": 0.0005820841151367131,
      "loss": 2.3484,
      "step": 25474
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6140042543411255,
      "learning_rate": 0.0005820827226792829,
      "loss": 2.1869,
      "step": 25475
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.5999203324317932,
      "learning_rate": 0.0005820813301694081,
      "loss": 2.4816,
      "step": 25476
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.630989134311676,
      "learning_rate": 0.0005820799376070892,
      "loss": 2.3797,
      "step": 25477
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6648005843162537,
      "learning_rate": 0.0005820785449923262,
      "loss": 2.2499,
      "step": 25478
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6416863799095154,
      "learning_rate": 0.0005820771523251196,
      "loss": 2.2717,
      "step": 25479
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6011932492256165,
      "learning_rate": 0.0005820757596054695,
      "loss": 2.3068,
      "step": 25480
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6161022186279297,
      "learning_rate": 0.0005820743668333761,
      "loss": 2.2945,
      "step": 25481
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.5476816892623901,
      "learning_rate": 0.0005820729740088398,
      "loss": 2.1958,
      "step": 25482
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.5759716629981995,
      "learning_rate": 0.0005820715811318608,
      "loss": 2.1575,
      "step": 25483
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.5681136250495911,
      "learning_rate": 0.0005820701882024394,
      "loss": 2.1815,
      "step": 25484
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.656544029712677,
      "learning_rate": 0.0005820687952205758,
      "loss": 2.2353,
      "step": 25485
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6847208738327026,
      "learning_rate": 0.0005820674021862703,
      "loss": 2.0934,
      "step": 25486
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6392495632171631,
      "learning_rate": 0.0005820660090995231,
      "loss": 2.4654,
      "step": 25487
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.5970272421836853,
      "learning_rate": 0.0005820646159603346,
      "loss": 2.4775,
      "step": 25488
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6766499876976013,
      "learning_rate": 0.0005820632227687049,
      "loss": 2.3919,
      "step": 25489
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6427197456359863,
      "learning_rate": 0.0005820618295246343,
      "loss": 2.1923,
      "step": 25490
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6069050431251526,
      "learning_rate": 0.0005820604362281232,
      "loss": 2.356,
      "step": 25491
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6093325614929199,
      "learning_rate": 0.0005820590428791716,
      "loss": 2.3224,
      "step": 25492
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6459929347038269,
      "learning_rate": 0.00058205764947778,
      "loss": 2.5875,
      "step": 25493
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6090033650398254,
      "learning_rate": 0.0005820562560239485,
      "loss": 2.1605,
      "step": 25494
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6656369566917419,
      "learning_rate": 0.0005820548625176774,
      "loss": 2.1875,
      "step": 25495
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.655102550983429,
      "learning_rate": 0.0005820534689589671,
      "loss": 2.1465,
      "step": 25496
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6398311257362366,
      "learning_rate": 0.0005820520753478176,
      "loss": 2.1027,
      "step": 25497
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6352992057800293,
      "learning_rate": 0.0005820506816842293,
      "loss": 2.1205,
      "step": 25498
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.5975019931793213,
      "learning_rate": 0.0005820492879682025,
      "loss": 2.2762,
      "step": 25499
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6457773447036743,
      "learning_rate": 0.0005820478941997374,
      "loss": 2.2693,
      "step": 25500
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.7033353447914124,
      "learning_rate": 0.0005820465003788343,
      "loss": 2.5878,
      "step": 25501
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.5884512662887573,
      "learning_rate": 0.0005820451065054934,
      "loss": 2.0571,
      "step": 25502
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.639886736869812,
      "learning_rate": 0.000582043712579715,
      "loss": 2.334,
      "step": 25503
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.624118983745575,
      "learning_rate": 0.0005820423186014994,
      "loss": 2.3216,
      "step": 25504
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6947282552719116,
      "learning_rate": 0.0005820409245708468,
      "loss": 2.6449,
      "step": 25505
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6754695177078247,
      "learning_rate": 0.0005820395304877574,
      "loss": 2.2868,
      "step": 25506
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.7102111577987671,
      "learning_rate": 0.0005820381363522316,
      "loss": 2.0586,
      "step": 25507
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6511492729187012,
      "learning_rate": 0.0005820367421642696,
      "loss": 2.3283,
      "step": 25508
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6135963797569275,
      "learning_rate": 0.0005820353479238715,
      "loss": 2.3969,
      "step": 25509
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6853123307228088,
      "learning_rate": 0.0005820339536310378,
      "loss": 2.3762,
      "step": 25510
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.5931273102760315,
      "learning_rate": 0.0005820325592857686,
      "loss": 2.1546,
      "step": 25511
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.7231075763702393,
      "learning_rate": 0.0005820311648880644,
      "loss": 2.3122,
      "step": 25512
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.7780959010124207,
      "learning_rate": 0.0005820297704379251,
      "loss": 2.5336,
      "step": 25513
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6175257563591003,
      "learning_rate": 0.0005820283759353513,
      "loss": 2.5141,
      "step": 25514
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6797354221343994,
      "learning_rate": 0.000582026981380343,
      "loss": 2.1647,
      "step": 25515
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.8536408543586731,
      "learning_rate": 0.0005820255867729005,
      "loss": 2.2761,
      "step": 25516
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.7231519818305969,
      "learning_rate": 0.0005820241921130242,
      "loss": 2.3172,
      "step": 25517
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6664501428604126,
      "learning_rate": 0.0005820227974007142,
      "loss": 2.6233,
      "step": 25518
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.702695906162262,
      "learning_rate": 0.000582021402635971,
      "loss": 2.2431,
      "step": 25519
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6572725176811218,
      "learning_rate": 0.0005820200078187946,
      "loss": 2.359,
      "step": 25520
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6635738611221313,
      "learning_rate": 0.0005820186129491853,
      "loss": 2.4119,
      "step": 25521
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.7278946042060852,
      "learning_rate": 0.0005820172180271434,
      "loss": 2.3568,
      "step": 25522
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6764307618141174,
      "learning_rate": 0.0005820158230526694,
      "loss": 2.3852,
      "step": 25523
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.7200837731361389,
      "learning_rate": 0.0005820144280257631,
      "loss": 2.6499,
      "step": 25524
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6436821222305298,
      "learning_rate": 0.000582013032946425,
      "loss": 2.2367,
      "step": 25525
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6425123810768127,
      "learning_rate": 0.0005820116378146556,
      "loss": 2.3706,
      "step": 25526
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6308023929595947,
      "learning_rate": 0.0005820102426304547,
      "loss": 2.3885,
      "step": 25527
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6345821022987366,
      "learning_rate": 0.0005820088473938228,
      "loss": 2.3217,
      "step": 25528
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.5934509038925171,
      "learning_rate": 0.0005820074521047601,
      "loss": 2.3268,
      "step": 25529
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.610670804977417,
      "learning_rate": 0.000582006056763267,
      "loss": 2.2192,
      "step": 25530
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.656496524810791,
      "learning_rate": 0.0005820046613693436,
      "loss": 2.1904,
      "step": 25531
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6397536993026733,
      "learning_rate": 0.0005820032659229902,
      "loss": 2.298,
      "step": 25532
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6287707090377808,
      "learning_rate": 0.0005820018704242072,
      "loss": 2.3496,
      "step": 25533
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6296114921569824,
      "learning_rate": 0.0005820004748729945,
      "loss": 2.4196,
      "step": 25534
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6883772015571594,
      "learning_rate": 0.0005819990792693527,
      "loss": 2.4468,
      "step": 25535
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.7021329402923584,
      "learning_rate": 0.000581997683613282,
      "loss": 1.9577,
      "step": 25536
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6251311898231506,
      "learning_rate": 0.0005819962879047825,
      "loss": 2.4732,
      "step": 25537
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6862941384315491,
      "learning_rate": 0.0005819948921438546,
      "loss": 2.1998,
      "step": 25538
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.627602219581604,
      "learning_rate": 0.0005819934963304986,
      "loss": 2.2398,
      "step": 25539
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6428422927856445,
      "learning_rate": 0.0005819921004647146,
      "loss": 2.4564,
      "step": 25540
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.8325660824775696,
      "learning_rate": 0.000581990704546503,
      "loss": 2.1353,
      "step": 25541
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6076174974441528,
      "learning_rate": 0.000581989308575864,
      "loss": 2.0987,
      "step": 25542
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6415696740150452,
      "learning_rate": 0.0005819879125527978,
      "loss": 2.4824,
      "step": 25543
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6287246346473694,
      "learning_rate": 0.0005819865164773048,
      "loss": 2.4166,
      "step": 25544
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.633449137210846,
      "learning_rate": 0.0005819851203493851,
      "loss": 2.3568,
      "step": 25545
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6624097228050232,
      "learning_rate": 0.0005819837241690391,
      "loss": 2.3036,
      "step": 25546
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6369844675064087,
      "learning_rate": 0.000581982327936267,
      "loss": 2.6398,
      "step": 25547
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6135121583938599,
      "learning_rate": 0.0005819809316510691,
      "loss": 2.2604,
      "step": 25548
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6427420377731323,
      "learning_rate": 0.0005819795353134455,
      "loss": 2.4194,
      "step": 25549
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6891784071922302,
      "learning_rate": 0.0005819781389233967,
      "loss": 2.3509,
      "step": 25550
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6542055606842041,
      "learning_rate": 0.0005819767424809228,
      "loss": 2.0257,
      "step": 25551
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.7210078239440918,
      "learning_rate": 0.0005819753459860241,
      "loss": 2.3891,
      "step": 25552
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6818491220474243,
      "learning_rate": 0.0005819739494387009,
      "loss": 2.5041,
      "step": 25553
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.593896746635437,
      "learning_rate": 0.0005819725528389534,
      "loss": 2.1951,
      "step": 25554
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.7394858598709106,
      "learning_rate": 0.0005819711561867819,
      "loss": 2.3076,
      "step": 25555
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.5676949620246887,
      "learning_rate": 0.0005819697594821866,
      "loss": 2.3376,
      "step": 25556
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.5972990393638611,
      "learning_rate": 0.0005819683627251678,
      "loss": 2.3938,
      "step": 25557
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6390663981437683,
      "learning_rate": 0.0005819669659157257,
      "loss": 2.3782,
      "step": 25558
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.9243740439414978,
      "learning_rate": 0.0005819655690538607,
      "loss": 2.6023,
      "step": 25559
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.5910114049911499,
      "learning_rate": 0.0005819641721395729,
      "loss": 2.2987,
      "step": 25560
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6203234195709229,
      "learning_rate": 0.0005819627751728627,
      "loss": 2.4145,
      "step": 25561
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6210066676139832,
      "learning_rate": 0.0005819613781537304,
      "loss": 2.3848,
      "step": 25562
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6558350920677185,
      "learning_rate": 0.0005819599810821761,
      "loss": 2.4148,
      "step": 25563
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.5967084169387817,
      "learning_rate": 0.0005819585839582,
      "loss": 2.4764,
      "step": 25564
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6562933921813965,
      "learning_rate": 0.0005819571867818025,
      "loss": 2.4737,
      "step": 25565
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.7045770883560181,
      "learning_rate": 0.000581955789552984,
      "loss": 2.308,
      "step": 25566
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6116120219230652,
      "learning_rate": 0.0005819543922717444,
      "loss": 2.0832,
      "step": 25567
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6308761239051819,
      "learning_rate": 0.0005819529949380842,
      "loss": 2.3076,
      "step": 25568
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.8249804973602295,
      "learning_rate": 0.0005819515975520037,
      "loss": 2.305,
      "step": 25569
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6280595660209656,
      "learning_rate": 0.0005819502001135029,
      "loss": 2.4582,
      "step": 25570
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.7007794380187988,
      "learning_rate": 0.0005819488026225825,
      "loss": 2.4609,
      "step": 25571
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6311670541763306,
      "learning_rate": 0.0005819474050792424,
      "loss": 2.4544,
      "step": 25572
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6816428303718567,
      "learning_rate": 0.0005819460074834829,
      "loss": 2.3453,
      "step": 25573
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.602358877658844,
      "learning_rate": 0.0005819446098353043,
      "loss": 2.4245,
      "step": 25574
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6219470500946045,
      "learning_rate": 0.0005819432121347068,
      "loss": 2.4425,
      "step": 25575
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6728177070617676,
      "learning_rate": 0.000581941814381691,
      "loss": 2.2662,
      "step": 25576
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.662499189376831,
      "learning_rate": 0.0005819404165762566,
      "loss": 2.2772,
      "step": 25577
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.5809969305992126,
      "learning_rate": 0.0005819390187184044,
      "loss": 2.2478,
      "step": 25578
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6199853420257568,
      "learning_rate": 0.0005819376208081343,
      "loss": 2.2702,
      "step": 25579
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.604337751865387,
      "learning_rate": 0.0005819362228454467,
      "loss": 2.532,
      "step": 25580
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.7491326332092285,
      "learning_rate": 0.0005819348248303418,
      "loss": 2.5374,
      "step": 25581
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.623716413974762,
      "learning_rate": 0.0005819334267628198,
      "loss": 2.3967,
      "step": 25582
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.5915409326553345,
      "learning_rate": 0.0005819320286428812,
      "loss": 2.1856,
      "step": 25583
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.7093499898910522,
      "learning_rate": 0.0005819306304705261,
      "loss": 2.4147,
      "step": 25584
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.7113229036331177,
      "learning_rate": 0.0005819292322457548,
      "loss": 2.425,
      "step": 25585
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6140915155410767,
      "learning_rate": 0.0005819278339685674,
      "loss": 2.2608,
      "step": 25586
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6713376045227051,
      "learning_rate": 0.0005819264356389643,
      "loss": 2.3166,
      "step": 25587
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6314402222633362,
      "learning_rate": 0.0005819250372569459,
      "loss": 2.4005,
      "step": 25588
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.61454176902771,
      "learning_rate": 0.0005819236388225123,
      "loss": 2.5244,
      "step": 25589
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6043261289596558,
      "learning_rate": 0.0005819222403356637,
      "loss": 2.3192,
      "step": 25590
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6351571679115295,
      "learning_rate": 0.0005819208417964004,
      "loss": 2.3741,
      "step": 25591
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6092910170555115,
      "learning_rate": 0.0005819194432047227,
      "loss": 2.4065,
      "step": 25592
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6529115438461304,
      "learning_rate": 0.000581918044560631,
      "loss": 2.1281,
      "step": 25593
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6584324240684509,
      "learning_rate": 0.0005819166458641253,
      "loss": 2.469,
      "step": 25594
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.616988480091095,
      "learning_rate": 0.0005819152471152059,
      "loss": 2.2182,
      "step": 25595
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6387444734573364,
      "learning_rate": 0.0005819138483138733,
      "loss": 2.5355,
      "step": 25596
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.5933451056480408,
      "learning_rate": 0.0005819124494601275,
      "loss": 2.3595,
      "step": 25597
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.638767659664154,
      "learning_rate": 0.0005819110505539688,
      "loss": 2.3015,
      "step": 25598
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.7575997114181519,
      "learning_rate": 0.0005819096515953977,
      "loss": 2.3216,
      "step": 25599
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.674950122833252,
      "learning_rate": 0.0005819082525844141,
      "loss": 2.321,
      "step": 25600
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.5582696199417114,
      "learning_rate": 0.0005819068535210185,
      "loss": 2.1284,
      "step": 25601
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.5760388374328613,
      "learning_rate": 0.0005819054544052111,
      "loss": 2.1895,
      "step": 25602
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.5878983736038208,
      "learning_rate": 0.0005819040552369921,
      "loss": 2.3192,
      "step": 25603
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6253543496131897,
      "learning_rate": 0.0005819026560163618,
      "loss": 2.3506,
      "step": 25604
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.7464446425437927,
      "learning_rate": 0.0005819012567433206,
      "loss": 2.2054,
      "step": 25605
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6095081567764282,
      "learning_rate": 0.0005818998574178686,
      "loss": 2.4946,
      "step": 25606
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.631335973739624,
      "learning_rate": 0.0005818984580400061,
      "loss": 2.3028,
      "step": 25607
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.7165789604187012,
      "learning_rate": 0.0005818970586097332,
      "loss": 2.319,
      "step": 25608
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6457968950271606,
      "learning_rate": 0.0005818956591270505,
      "loss": 2.4055,
      "step": 25609
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6629583835601807,
      "learning_rate": 0.000581894259591958,
      "loss": 2.0954,
      "step": 25610
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6772628426551819,
      "learning_rate": 0.0005818928600044561,
      "loss": 2.3739,
      "step": 25611
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6615632772445679,
      "learning_rate": 0.000581891460364545,
      "loss": 2.7336,
      "step": 25612
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6294687986373901,
      "learning_rate": 0.0005818900606722248,
      "loss": 2.3033,
      "step": 25613
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6119058728218079,
      "learning_rate": 0.000581888660927496,
      "loss": 2.1403,
      "step": 25614
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6400124430656433,
      "learning_rate": 0.0005818872611303588,
      "loss": 2.3515,
      "step": 25615
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6257029175758362,
      "learning_rate": 0.0005818858612808135,
      "loss": 2.3057,
      "step": 25616
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.581874668598175,
      "learning_rate": 0.0005818844613788602,
      "loss": 2.4041,
      "step": 25617
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6283895373344421,
      "learning_rate": 0.0005818830614244992,
      "loss": 2.3882,
      "step": 25618
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.5685592293739319,
      "learning_rate": 0.0005818816614177309,
      "loss": 2.1188,
      "step": 25619
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.692145586013794,
      "learning_rate": 0.0005818802613585556,
      "loss": 2.3911,
      "step": 25620
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6259089708328247,
      "learning_rate": 0.0005818788612469732,
      "loss": 2.2425,
      "step": 25621
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6530827879905701,
      "learning_rate": 0.0005818774610829844,
      "loss": 2.3505,
      "step": 25622
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.7170001864433289,
      "learning_rate": 0.0005818760608665891,
      "loss": 2.4,
      "step": 25623
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6298331022262573,
      "learning_rate": 0.0005818746605977879,
      "loss": 2.2502,
      "step": 25624
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6357465982437134,
      "learning_rate": 0.0005818732602765807,
      "loss": 2.4709,
      "step": 25625
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6225118637084961,
      "learning_rate": 0.000581871859902968,
      "loss": 2.2245,
      "step": 25626
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6275559663772583,
      "learning_rate": 0.0005818704594769499,
      "loss": 2.4914,
      "step": 25627
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.5765104293823242,
      "learning_rate": 0.000581869058998527,
      "loss": 2.1939,
      "step": 25628
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6400202512741089,
      "learning_rate": 0.0005818676584676991,
      "loss": 2.45,
      "step": 25629
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6229012608528137,
      "learning_rate": 0.0005818662578844668,
      "loss": 2.4372,
      "step": 25630
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6672868132591248,
      "learning_rate": 0.0005818648572488303,
      "loss": 2.2582,
      "step": 25631
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6111781001091003,
      "learning_rate": 0.0005818634565607897,
      "loss": 2.2299,
      "step": 25632
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6334740519523621,
      "learning_rate": 0.0005818620558203454,
      "loss": 2.3562,
      "step": 25633
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.7017740607261658,
      "learning_rate": 0.0005818606550274976,
      "loss": 2.3546,
      "step": 25634
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6258436441421509,
      "learning_rate": 0.0005818592541822465,
      "loss": 2.3095,
      "step": 25635
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6630619764328003,
      "learning_rate": 0.0005818578532845927,
      "loss": 2.2144,
      "step": 25636
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.774188756942749,
      "learning_rate": 0.000581856452334536,
      "loss": 2.0444,
      "step": 25637
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.5726372003555298,
      "learning_rate": 0.0005818550513320769,
      "loss": 2.4065,
      "step": 25638
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6285721063613892,
      "learning_rate": 0.0005818536502772156,
      "loss": 2.4473,
      "step": 25639
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6541749835014343,
      "learning_rate": 0.0005818522491699525,
      "loss": 2.2829,
      "step": 25640
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6899814009666443,
      "learning_rate": 0.0005818508480102876,
      "loss": 2.4219,
      "step": 25641
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6322447657585144,
      "learning_rate": 0.0005818494467982213,
      "loss": 2.3998,
      "step": 25642
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6622315645217896,
      "learning_rate": 0.000581848045533754,
      "loss": 2.5559,
      "step": 25643
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6271074414253235,
      "learning_rate": 0.0005818466442168858,
      "loss": 2.54,
      "step": 25644
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.7195440530776978,
      "learning_rate": 0.0005818452428476169,
      "loss": 2.4696,
      "step": 25645
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.5831514596939087,
      "learning_rate": 0.0005818438414259477,
      "loss": 2.1868,
      "step": 25646
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6186720728874207,
      "learning_rate": 0.0005818424399518784,
      "loss": 2.4546,
      "step": 25647
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6686955094337463,
      "learning_rate": 0.0005818410384254094,
      "loss": 2.4161,
      "step": 25648
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6030098795890808,
      "learning_rate": 0.0005818396368465407,
      "loss": 2.3408,
      "step": 25649
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6787851452827454,
      "learning_rate": 0.0005818382352152727,
      "loss": 2.3545,
      "step": 25650
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6054686307907104,
      "learning_rate": 0.0005818368335316056,
      "loss": 2.4513,
      "step": 25651
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6431744694709778,
      "learning_rate": 0.0005818354317955398,
      "loss": 2.1261,
      "step": 25652
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6115590929985046,
      "learning_rate": 0.0005818340300070754,
      "loss": 2.3086,
      "step": 25653
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.628067672252655,
      "learning_rate": 0.0005818326281662128,
      "loss": 2.164,
      "step": 25654
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6467095017433167,
      "learning_rate": 0.0005818312262729521,
      "loss": 2.1363,
      "step": 25655
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.5737326741218567,
      "learning_rate": 0.0005818298243272938,
      "loss": 2.4926,
      "step": 25656
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6120058298110962,
      "learning_rate": 0.0005818284223292379,
      "loss": 2.532,
      "step": 25657
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6084572076797485,
      "learning_rate": 0.0005818270202787849,
      "loss": 2.3774,
      "step": 25658
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.7588807940483093,
      "learning_rate": 0.0005818256181759348,
      "loss": 2.3791,
      "step": 25659
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6218753457069397,
      "learning_rate": 0.0005818242160206881,
      "loss": 2.386,
      "step": 25660
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6156992316246033,
      "learning_rate": 0.0005818228138130448,
      "loss": 2.3778,
      "step": 25661
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.5658289194107056,
      "learning_rate": 0.0005818214115530055,
      "loss": 2.2675,
      "step": 25662
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6822137832641602,
      "learning_rate": 0.0005818200092405703,
      "loss": 2.4187,
      "step": 25663
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6203960180282593,
      "learning_rate": 0.0005818186068757392,
      "loss": 2.2534,
      "step": 25664
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6260963082313538,
      "learning_rate": 0.0005818172044585129,
      "loss": 2.352,
      "step": 25665
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.5667542219161987,
      "learning_rate": 0.0005818158019888914,
      "loss": 2.0939,
      "step": 25666
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6897059679031372,
      "learning_rate": 0.0005818143994668751,
      "loss": 2.4467,
      "step": 25667
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.654595136642456,
      "learning_rate": 0.0005818129968924641,
      "loss": 2.3554,
      "step": 25668
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6284070014953613,
      "learning_rate": 0.0005818115942656587,
      "loss": 2.2025,
      "step": 25669
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6433919072151184,
      "learning_rate": 0.0005818101915864594,
      "loss": 2.434,
      "step": 25670
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6412467360496521,
      "learning_rate": 0.0005818087888548662,
      "loss": 2.5612,
      "step": 25671
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6252185702323914,
      "learning_rate": 0.0005818073860708793,
      "loss": 2.3423,
      "step": 25672
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6322634816169739,
      "learning_rate": 0.000581805983234499,
      "loss": 2.3128,
      "step": 25673
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6204479336738586,
      "learning_rate": 0.0005818045803457259,
      "loss": 2.4175,
      "step": 25674
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6780108213424683,
      "learning_rate": 0.0005818031774045599,
      "loss": 2.3602,
      "step": 25675
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.5922275185585022,
      "learning_rate": 0.0005818017744110013,
      "loss": 2.1166,
      "step": 25676
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6322140693664551,
      "learning_rate": 0.0005818003713650505,
      "loss": 2.2932,
      "step": 25677
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6074820756912231,
      "learning_rate": 0.0005817989682667078,
      "loss": 2.2406,
      "step": 25678
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6444985270500183,
      "learning_rate": 0.0005817975651159732,
      "loss": 2.1611,
      "step": 25679
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6301335692405701,
      "learning_rate": 0.0005817961619128472,
      "loss": 2.2222,
      "step": 25680
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6154231429100037,
      "learning_rate": 0.0005817947586573299,
      "loss": 2.4541,
      "step": 25681
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.619768500328064,
      "learning_rate": 0.0005817933553494217,
      "loss": 2.3717,
      "step": 25682
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6255348324775696,
      "learning_rate": 0.0005817919519891227,
      "loss": 2.2139,
      "step": 25683
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.7049162983894348,
      "learning_rate": 0.0005817905485764334,
      "loss": 2.4905,
      "step": 25684
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6165250539779663,
      "learning_rate": 0.0005817891451113539,
      "loss": 2.3577,
      "step": 25685
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6105530858039856,
      "learning_rate": 0.0005817877415938843,
      "loss": 2.3637,
      "step": 25686
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.573756992816925,
      "learning_rate": 0.0005817863380240253,
      "loss": 2.2785,
      "step": 25687
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.5897049307823181,
      "learning_rate": 0.0005817849344017768,
      "loss": 2.1215,
      "step": 25688
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.5798435211181641,
      "learning_rate": 0.000581783530727139,
      "loss": 2.518,
      "step": 25689
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6706337332725525,
      "learning_rate": 0.0005817821270001124,
      "loss": 2.3179,
      "step": 25690
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6654784679412842,
      "learning_rate": 0.0005817807232206973,
      "loss": 2.2481,
      "step": 25691
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.693448543548584,
      "learning_rate": 0.0005817793193888938,
      "loss": 2.3101,
      "step": 25692
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6898648142814636,
      "learning_rate": 0.0005817779155047022,
      "loss": 2.1566,
      "step": 25693
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6101966500282288,
      "learning_rate": 0.0005817765115681227,
      "loss": 2.4345,
      "step": 25694
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6828134655952454,
      "learning_rate": 0.0005817751075791557,
      "loss": 2.3153,
      "step": 25695
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.7266025543212891,
      "learning_rate": 0.0005817737035378014,
      "loss": 2.3531,
      "step": 25696
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.675417959690094,
      "learning_rate": 0.00058177229944406,
      "loss": 2.2079,
      "step": 25697
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6590842008590698,
      "learning_rate": 0.0005817708952979318,
      "loss": 2.3055,
      "step": 25698
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6390736103057861,
      "learning_rate": 0.0005817694910994172,
      "loss": 2.3949,
      "step": 25699
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.8031089901924133,
      "learning_rate": 0.0005817680868485161,
      "loss": 2.2772,
      "step": 25700
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.7114183306694031,
      "learning_rate": 0.0005817666825452292,
      "loss": 2.2541,
      "step": 25701
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6523195505142212,
      "learning_rate": 0.0005817652781895565,
      "loss": 2.5235,
      "step": 25702
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.012500524520874,
      "learning_rate": 0.0005817638737814982,
      "loss": 2.2974,
      "step": 25703
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6171818971633911,
      "learning_rate": 0.0005817624693210548,
      "loss": 2.3795,
      "step": 25704
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.919136106967926,
      "learning_rate": 0.0005817610648082265,
      "loss": 2.0928,
      "step": 25705
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6093403697013855,
      "learning_rate": 0.0005817596602430133,
      "loss": 2.4705,
      "step": 25706
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6112070083618164,
      "learning_rate": 0.0005817582556254157,
      "loss": 2.1005,
      "step": 25707
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6315016150474548,
      "learning_rate": 0.000581756850955434,
      "loss": 2.0377,
      "step": 25708
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6239300966262817,
      "learning_rate": 0.0005817554462330684,
      "loss": 2.4219,
      "step": 25709
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6773992776870728,
      "learning_rate": 0.0005817540414583192,
      "loss": 2.5667,
      "step": 25710
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6596900224685669,
      "learning_rate": 0.0005817526366311865,
      "loss": 2.4461,
      "step": 25711
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6803475022315979,
      "learning_rate": 0.0005817512317516707,
      "loss": 2.2761,
      "step": 25712
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6397674679756165,
      "learning_rate": 0.000581749826819772,
      "loss": 2.3001,
      "step": 25713
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6276796460151672,
      "learning_rate": 0.0005817484218354907,
      "loss": 2.3737,
      "step": 25714
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6529070138931274,
      "learning_rate": 0.0005817470167988271,
      "loss": 2.3325,
      "step": 25715
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6940520405769348,
      "learning_rate": 0.0005817456117097814,
      "loss": 2.4005,
      "step": 25716
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6416887044906616,
      "learning_rate": 0.0005817442065683537,
      "loss": 2.2669,
      "step": 25717
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6297704577445984,
      "learning_rate": 0.0005817428013745447,
      "loss": 2.2808,
      "step": 25718
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6327967047691345,
      "learning_rate": 0.0005817413961283542,
      "loss": 2.2202,
      "step": 25719
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.9808080196380615,
      "learning_rate": 0.0005817399908297828,
      "loss": 2.2641,
      "step": 25720
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.7583242654800415,
      "learning_rate": 0.0005817385854788305,
      "loss": 2.345,
      "step": 25721
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.8436206579208374,
      "learning_rate": 0.0005817371800754977,
      "loss": 2.376,
      "step": 25722
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6405591368675232,
      "learning_rate": 0.0005817357746197847,
      "loss": 2.3538,
      "step": 25723
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6555359363555908,
      "learning_rate": 0.0005817343691116917,
      "loss": 2.4882,
      "step": 25724
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.7072018384933472,
      "learning_rate": 0.0005817329635512189,
      "loss": 2.3797,
      "step": 25725
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.7793118953704834,
      "learning_rate": 0.0005817315579383666,
      "loss": 2.4811,
      "step": 25726
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6148805022239685,
      "learning_rate": 0.0005817301522731351,
      "loss": 2.4627,
      "step": 25727
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6645533442497253,
      "learning_rate": 0.0005817287465555248,
      "loss": 2.4539,
      "step": 25728
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.8093058466911316,
      "learning_rate": 0.0005817273407855356,
      "loss": 2.2855,
      "step": 25729
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6531903147697449,
      "learning_rate": 0.000581725934963168,
      "loss": 2.3226,
      "step": 25730
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.58432936668396,
      "learning_rate": 0.0005817245290884224,
      "loss": 2.3224,
      "step": 25731
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6516015529632568,
      "learning_rate": 0.0005817231231612988,
      "loss": 2.5402,
      "step": 25732
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6655108332633972,
      "learning_rate": 0.0005817217171817976,
      "loss": 2.2068,
      "step": 25733
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6718326807022095,
      "learning_rate": 0.0005817203111499189,
      "loss": 2.3976,
      "step": 25734
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6211421489715576,
      "learning_rate": 0.000581718905065663,
      "loss": 2.3684,
      "step": 25735
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.7021694183349609,
      "learning_rate": 0.0005817174989290304,
      "loss": 2.389,
      "step": 25736
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6949802041053772,
      "learning_rate": 0.0005817160927400212,
      "loss": 2.3159,
      "step": 25737
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.8108488321304321,
      "learning_rate": 0.0005817146864986356,
      "loss": 2.5047,
      "step": 25738
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.5833590030670166,
      "learning_rate": 0.0005817132802048739,
      "loss": 2.0336,
      "step": 25739
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.7049698829650879,
      "learning_rate": 0.0005817118738587365,
      "loss": 2.3048,
      "step": 25740
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6635711193084717,
      "learning_rate": 0.0005817104674602235,
      "loss": 2.3082,
      "step": 25741
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.5943832397460938,
      "learning_rate": 0.0005817090610093351,
      "loss": 2.5845,
      "step": 25742
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6810966730117798,
      "learning_rate": 0.0005817076545060717,
      "loss": 2.2968,
      "step": 25743
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.7408403158187866,
      "learning_rate": 0.0005817062479504337,
      "loss": 2.5444,
      "step": 25744
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6862673759460449,
      "learning_rate": 0.000581704841342421,
      "loss": 2.3845,
      "step": 25745
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6171705722808838,
      "learning_rate": 0.0005817034346820341,
      "loss": 2.4193,
      "step": 25746
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6136375665664673,
      "learning_rate": 0.0005817020279692733,
      "loss": 2.3938,
      "step": 25747
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6273588538169861,
      "learning_rate": 0.0005817006212041387,
      "loss": 2.4458,
      "step": 25748
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6433042287826538,
      "learning_rate": 0.0005816992143866306,
      "loss": 2.4784,
      "step": 25749
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.583193838596344,
      "learning_rate": 0.0005816978075167494,
      "loss": 2.4122,
      "step": 25750
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.5761305689811707,
      "learning_rate": 0.0005816964005944952,
      "loss": 2.0682,
      "step": 25751
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.5773890018463135,
      "learning_rate": 0.0005816949936198684,
      "loss": 2.4763,
      "step": 25752
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6265507340431213,
      "learning_rate": 0.000581693586592869,
      "loss": 2.5957,
      "step": 25753
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.5912128686904907,
      "learning_rate": 0.0005816921795134976,
      "loss": 2.3737,
      "step": 25754
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6486682295799255,
      "learning_rate": 0.0005816907723817542,
      "loss": 2.2908,
      "step": 25755
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6671944856643677,
      "learning_rate": 0.0005816893651976392,
      "loss": 2.2332,
      "step": 25756
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6019869446754456,
      "learning_rate": 0.0005816879579611529,
      "loss": 2.3663,
      "step": 25757
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.5944968461990356,
      "learning_rate": 0.0005816865506722955,
      "loss": 2.4032,
      "step": 25758
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.8510375022888184,
      "learning_rate": 0.0005816851433310671,
      "loss": 2.225,
      "step": 25759
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6930539011955261,
      "learning_rate": 0.0005816837359374682,
      "loss": 2.328,
      "step": 25760
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6469429135322571,
      "learning_rate": 0.000581682328491499,
      "loss": 2.2822,
      "step": 25761
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.7201108336448669,
      "learning_rate": 0.0005816809209931599,
      "loss": 2.3489,
      "step": 25762
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.649834156036377,
      "learning_rate": 0.0005816795134424508,
      "loss": 2.3552,
      "step": 25763
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.7529852390289307,
      "learning_rate": 0.0005816781058393722,
      "loss": 2.4777,
      "step": 25764
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.7826261520385742,
      "learning_rate": 0.0005816766981839242,
      "loss": 2.1142,
      "step": 25765
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6071142554283142,
      "learning_rate": 0.0005816752904761074,
      "loss": 2.4635,
      "step": 25766
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6533108353614807,
      "learning_rate": 0.0005816738827159217,
      "loss": 2.4709,
      "step": 25767
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.7465366721153259,
      "learning_rate": 0.0005816724749033676,
      "loss": 2.4441,
      "step": 25768
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.855769157409668,
      "learning_rate": 0.0005816710670384452,
      "loss": 2.0981,
      "step": 25769
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6976012587547302,
      "learning_rate": 0.0005816696591211548,
      "loss": 2.3878,
      "step": 25770
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.9006213545799255,
      "learning_rate": 0.0005816682511514969,
      "loss": 2.4434,
      "step": 25771
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6698273420333862,
      "learning_rate": 0.0005816668431294713,
      "loss": 2.3713,
      "step": 25772
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6189523935317993,
      "learning_rate": 0.0005816654350550786,
      "loss": 2.2618,
      "step": 25773
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6587539315223694,
      "learning_rate": 0.000581664026928319,
      "loss": 2.3166,
      "step": 25774
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6070554852485657,
      "learning_rate": 0.0005816626187491928,
      "loss": 2.3171,
      "step": 25775
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6977628469467163,
      "learning_rate": 0.0005816612105177001,
      "loss": 2.3094,
      "step": 25776
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6270889043807983,
      "learning_rate": 0.0005816598022338413,
      "loss": 2.1612,
      "step": 25777
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6515527367591858,
      "learning_rate": 0.0005816583938976167,
      "loss": 2.0979,
      "step": 25778
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.623147189617157,
      "learning_rate": 0.0005816569855090264,
      "loss": 2.337,
      "step": 25779
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6369523406028748,
      "learning_rate": 0.0005816555770680708,
      "loss": 2.3106,
      "step": 25780
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.544188380241394,
      "learning_rate": 0.0005816541685747501,
      "loss": 2.117,
      "step": 25781
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6164036393165588,
      "learning_rate": 0.0005816527600290645,
      "loss": 2.3335,
      "step": 25782
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6070275902748108,
      "learning_rate": 0.0005816513514310144,
      "loss": 2.2214,
      "step": 25783
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6591363549232483,
      "learning_rate": 0.0005816499427806001,
      "loss": 2.3758,
      "step": 25784
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6490110754966736,
      "learning_rate": 0.0005816485340778216,
      "loss": 2.2637,
      "step": 25785
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.642349362373352,
      "learning_rate": 0.0005816471253226793,
      "loss": 2.3999,
      "step": 25786
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6307351589202881,
      "learning_rate": 0.0005816457165151736,
      "loss": 2.2164,
      "step": 25787
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6290777921676636,
      "learning_rate": 0.0005816443076553045,
      "loss": 2.4804,
      "step": 25788
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6787443161010742,
      "learning_rate": 0.0005816428987430726,
      "loss": 2.4829,
      "step": 25789
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6065176129341125,
      "learning_rate": 0.0005816414897784779,
      "loss": 2.0834,
      "step": 25790
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6244292855262756,
      "learning_rate": 0.0005816400807615206,
      "loss": 2.4037,
      "step": 25791
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6904205679893494,
      "learning_rate": 0.0005816386716922013,
      "loss": 2.3203,
      "step": 25792
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.7733414173126221,
      "learning_rate": 0.00058163726257052,
      "loss": 2.24,
      "step": 25793
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.774393618106842,
      "learning_rate": 0.0005816358533964769,
      "loss": 2.3774,
      "step": 25794
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6511837840080261,
      "learning_rate": 0.0005816344441700724,
      "loss": 2.5067,
      "step": 25795
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6800753474235535,
      "learning_rate": 0.0005816330348913069,
      "loss": 2.2658,
      "step": 25796
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6240050792694092,
      "learning_rate": 0.0005816316255601803,
      "loss": 2.3117,
      "step": 25797
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.655759334564209,
      "learning_rate": 0.0005816302161766932,
      "loss": 2.2907,
      "step": 25798
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.7312699556350708,
      "learning_rate": 0.0005816288067408456,
      "loss": 2.3838,
      "step": 25799
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6677759289741516,
      "learning_rate": 0.000581627397252638,
      "loss": 2.324,
      "step": 25800
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6466881632804871,
      "learning_rate": 0.0005816259877120706,
      "loss": 2.2756,
      "step": 25801
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.7062942385673523,
      "learning_rate": 0.0005816245781191435,
      "loss": 2.1254,
      "step": 25802
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.7336272597312927,
      "learning_rate": 0.000581623168473857,
      "loss": 2.4858,
      "step": 25803
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.7031810879707336,
      "learning_rate": 0.0005816217587762116,
      "loss": 2.3433,
      "step": 25804
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.8856362104415894,
      "learning_rate": 0.0005816203490262074,
      "loss": 2.3248,
      "step": 25805
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6434950828552246,
      "learning_rate": 0.0005816189392238445,
      "loss": 2.4842,
      "step": 25806
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6651949286460876,
      "learning_rate": 0.0005816175293691235,
      "loss": 2.194,
      "step": 25807
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.7722551822662354,
      "learning_rate": 0.0005816161194620444,
      "loss": 2.3953,
      "step": 25808
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.5948531031608582,
      "learning_rate": 0.0005816147095026075,
      "loss": 2.4573,
      "step": 25809
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6547122597694397,
      "learning_rate": 0.0005816132994908133,
      "loss": 2.4302,
      "step": 25810
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.7208897471427917,
      "learning_rate": 0.0005816118894266617,
      "loss": 2.2746,
      "step": 25811
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6154723167419434,
      "learning_rate": 0.0005816104793101532,
      "loss": 2.1593,
      "step": 25812
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6716099977493286,
      "learning_rate": 0.0005816090691412879,
      "loss": 2.0824,
      "step": 25813
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6543041467666626,
      "learning_rate": 0.0005816076589200662,
      "loss": 2.4416,
      "step": 25814
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6016440987586975,
      "learning_rate": 0.0005816062486464884,
      "loss": 2.4541,
      "step": 25815
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6576808094978333,
      "learning_rate": 0.0005816048383205547,
      "loss": 2.3688,
      "step": 25816
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6764301657676697,
      "learning_rate": 0.0005816034279422653,
      "loss": 2.5141,
      "step": 25817
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6275957822799683,
      "learning_rate": 0.0005816020175116204,
      "loss": 2.2066,
      "step": 25818
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.603840172290802,
      "learning_rate": 0.0005816006070286204,
      "loss": 2.2735,
      "step": 25819
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6222963929176331,
      "learning_rate": 0.0005815991964932656,
      "loss": 2.2093,
      "step": 25820
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6408982872962952,
      "learning_rate": 0.0005815977859055562,
      "loss": 2.3048,
      "step": 25821
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6865091919898987,
      "learning_rate": 0.0005815963752654925,
      "loss": 2.2326,
      "step": 25822
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.5802394151687622,
      "learning_rate": 0.0005815949645730746,
      "loss": 2.2807,
      "step": 25823
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6366052627563477,
      "learning_rate": 0.0005815935538283029,
      "loss": 2.4231,
      "step": 25824
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6337811946868896,
      "learning_rate": 0.0005815921430311777,
      "loss": 2.2773,
      "step": 25825
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6634505391120911,
      "learning_rate": 0.0005815907321816992,
      "loss": 2.3137,
      "step": 25826
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6566521525382996,
      "learning_rate": 0.0005815893212798677,
      "loss": 2.4301,
      "step": 25827
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6562199592590332,
      "learning_rate": 0.0005815879103256834,
      "loss": 2.5848,
      "step": 25828
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.5708783864974976,
      "learning_rate": 0.0005815864993191465,
      "loss": 2.3109,
      "step": 25829
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.684066116809845,
      "learning_rate": 0.0005815850882602574,
      "loss": 2.3941,
      "step": 25830
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6790978312492371,
      "learning_rate": 0.0005815836771490164,
      "loss": 2.4321,
      "step": 25831
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6274964213371277,
      "learning_rate": 0.0005815822659854237,
      "loss": 2.2862,
      "step": 25832
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6184062361717224,
      "learning_rate": 0.0005815808547694794,
      "loss": 2.1968,
      "step": 25833
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6931771636009216,
      "learning_rate": 0.000581579443501184,
      "loss": 2.2335,
      "step": 25834
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.5974286198616028,
      "learning_rate": 0.0005815780321805376,
      "loss": 2.6052,
      "step": 25835
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6267291903495789,
      "learning_rate": 0.0005815766208075406,
      "loss": 2.4448,
      "step": 25836
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6075490117073059,
      "learning_rate": 0.0005815752093821931,
      "loss": 2.3495,
      "step": 25837
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.5922878980636597,
      "learning_rate": 0.0005815737979044955,
      "loss": 2.3993,
      "step": 25838
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.5908045768737793,
      "learning_rate": 0.000581572386374448,
      "loss": 2.3204,
      "step": 25839
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6360955238342285,
      "learning_rate": 0.0005815709747920509,
      "loss": 2.1135,
      "step": 25840
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6075769662857056,
      "learning_rate": 0.0005815695631573044,
      "loss": 2.3502,
      "step": 25841
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6534874439239502,
      "learning_rate": 0.000581568151470209,
      "loss": 2.4403,
      "step": 25842
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6765883564949036,
      "learning_rate": 0.0005815667397307645,
      "loss": 2.1168,
      "step": 25843
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6585378646850586,
      "learning_rate": 0.0005815653279389716,
      "loss": 2.3086,
      "step": 25844
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6385040879249573,
      "learning_rate": 0.0005815639160948303,
      "loss": 2.3412,
      "step": 25845
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6356061100959778,
      "learning_rate": 0.000581562504198341,
      "loss": 2.4067,
      "step": 25846
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6309154033660889,
      "learning_rate": 0.0005815610922495039,
      "loss": 2.3629,
      "step": 25847
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6758080124855042,
      "learning_rate": 0.0005815596802483192,
      "loss": 2.2195,
      "step": 25848
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.7078237533569336,
      "learning_rate": 0.0005815582681947874,
      "loss": 2.2775,
      "step": 25849
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6801012754440308,
      "learning_rate": 0.0005815568560889085,
      "loss": 2.4244,
      "step": 25850
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.614851713180542,
      "learning_rate": 0.0005815554439306829,
      "loss": 2.0619,
      "step": 25851
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6545867323875427,
      "learning_rate": 0.0005815540317201108,
      "loss": 2.474,
      "step": 25852
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6575943827629089,
      "learning_rate": 0.0005815526194571925,
      "loss": 2.1715,
      "step": 25853
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6391040086746216,
      "learning_rate": 0.0005815512071419283,
      "loss": 2.3129,
      "step": 25854
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6143973469734192,
      "learning_rate": 0.0005815497947743183,
      "loss": 2.3585,
      "step": 25855
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.642231822013855,
      "learning_rate": 0.000581548382354363,
      "loss": 2.3633,
      "step": 25856
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6389625668525696,
      "learning_rate": 0.0005815469698820625,
      "loss": 2.2312,
      "step": 25857
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6327040791511536,
      "learning_rate": 0.0005815455573574171,
      "loss": 2.2585,
      "step": 25858
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6463385224342346,
      "learning_rate": 0.000581544144780427,
      "loss": 2.3192,
      "step": 25859
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6038196086883545,
      "learning_rate": 0.0005815427321510927,
      "loss": 2.3592,
      "step": 25860
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6307191252708435,
      "learning_rate": 0.0005815413194694141,
      "loss": 2.2753,
      "step": 25861
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6331202983856201,
      "learning_rate": 0.0005815399067353918,
      "loss": 2.1424,
      "step": 25862
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6603968739509583,
      "learning_rate": 0.0005815384939490258,
      "loss": 2.3565,
      "step": 25863
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6605440974235535,
      "learning_rate": 0.0005815370811103165,
      "loss": 2.1101,
      "step": 25864
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6861604452133179,
      "learning_rate": 0.0005815356682192643,
      "loss": 2.3401,
      "step": 25865
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6347041130065918,
      "learning_rate": 0.0005815342552758691,
      "loss": 2.3112,
      "step": 25866
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.7091829180717468,
      "learning_rate": 0.0005815328422801314,
      "loss": 2.5792,
      "step": 25867
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6049337983131409,
      "learning_rate": 0.0005815314292320515,
      "loss": 2.4282,
      "step": 25868
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.5875838398933411,
      "learning_rate": 0.0005815300161316296,
      "loss": 2.2513,
      "step": 25869
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.7447033524513245,
      "learning_rate": 0.0005815286029788659,
      "loss": 2.3684,
      "step": 25870
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.7315929532051086,
      "learning_rate": 0.0005815271897737608,
      "loss": 2.3417,
      "step": 25871
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.7100211977958679,
      "learning_rate": 0.0005815257765163145,
      "loss": 2.4054,
      "step": 25872
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.7595391273498535,
      "learning_rate": 0.0005815243632065272,
      "loss": 2.401,
      "step": 25873
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6875366568565369,
      "learning_rate": 0.0005815229498443991,
      "loss": 2.1744,
      "step": 25874
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6586808562278748,
      "learning_rate": 0.0005815215364299306,
      "loss": 2.4023,
      "step": 25875
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6258586049079895,
      "learning_rate": 0.000581520122963122,
      "loss": 2.2033,
      "step": 25876
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6638349890708923,
      "learning_rate": 0.0005815187094439735,
      "loss": 2.3625,
      "step": 25877
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6577914953231812,
      "learning_rate": 0.0005815172958724853,
      "loss": 2.6255,
      "step": 25878
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6821237802505493,
      "learning_rate": 0.0005815158822486577,
      "loss": 2.4396,
      "step": 25879
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.7973218560218811,
      "learning_rate": 0.0005815144685724911,
      "loss": 2.1863,
      "step": 25880
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6455915570259094,
      "learning_rate": 0.0005815130548439856,
      "loss": 2.3314,
      "step": 25881
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6007376909255981,
      "learning_rate": 0.0005815116410631414,
      "loss": 2.2241,
      "step": 25882
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.0832167863845825,
      "learning_rate": 0.0005815102272299591,
      "loss": 2.3433,
      "step": 25883
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6507272720336914,
      "learning_rate": 0.0005815088133444385,
      "loss": 2.3558,
      "step": 25884
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.7602343559265137,
      "learning_rate": 0.0005815073994065803,
      "loss": 2.4086,
      "step": 25885
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.8354634642601013,
      "learning_rate": 0.0005815059854163844,
      "loss": 2.4475,
      "step": 25886
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6126537322998047,
      "learning_rate": 0.0005815045713738513,
      "loss": 2.457,
      "step": 25887
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6707897186279297,
      "learning_rate": 0.0005815031572789812,
      "loss": 2.2363,
      "step": 25888
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6388869285583496,
      "learning_rate": 0.0005815017431317742,
      "loss": 2.2353,
      "step": 25889
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6359435319900513,
      "learning_rate": 0.0005815003289322309,
      "loss": 2.258,
      "step": 25890
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6718207597732544,
      "learning_rate": 0.0005814989146803513,
      "loss": 2.378,
      "step": 25891
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6347609758377075,
      "learning_rate": 0.0005814975003761358,
      "loss": 2.38,
      "step": 25892
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.7135626077651978,
      "learning_rate": 0.0005814960860195845,
      "loss": 2.3085,
      "step": 25893
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6919323205947876,
      "learning_rate": 0.0005814946716106978,
      "loss": 2.4174,
      "step": 25894
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.636537492275238,
      "learning_rate": 0.000581493257149476,
      "loss": 2.353,
      "step": 25895
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6605460047721863,
      "learning_rate": 0.0005814918426359192,
      "loss": 1.9795,
      "step": 25896
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.613588273525238,
      "learning_rate": 0.0005814904280700277,
      "loss": 2.2288,
      "step": 25897
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6441142559051514,
      "learning_rate": 0.0005814890134518019,
      "loss": 2.4613,
      "step": 25898
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.570885956287384,
      "learning_rate": 0.0005814875987812419,
      "loss": 2.3598,
      "step": 25899
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6400547027587891,
      "learning_rate": 0.0005814861840583482,
      "loss": 2.0895,
      "step": 25900
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.7022759318351746,
      "learning_rate": 0.0005814847692831208,
      "loss": 2.2488,
      "step": 25901
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6009405255317688,
      "learning_rate": 0.0005814833544555601,
      "loss": 2.0365,
      "step": 25902
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6243774890899658,
      "learning_rate": 0.0005814819395756663,
      "loss": 2.2753,
      "step": 25903
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6617527008056641,
      "learning_rate": 0.0005814805246434398,
      "loss": 2.2581,
      "step": 25904
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.7843572497367859,
      "learning_rate": 0.0005814791096588806,
      "loss": 2.3734,
      "step": 25905
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.5925623178482056,
      "learning_rate": 0.0005814776946219892,
      "loss": 2.1297,
      "step": 25906
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6297915577888489,
      "learning_rate": 0.0005814762795327658,
      "loss": 2.404,
      "step": 25907
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.7055248618125916,
      "learning_rate": 0.0005814748643912105,
      "loss": 2.4055,
      "step": 25908
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.8567669987678528,
      "learning_rate": 0.000581473449197324,
      "loss": 2.2629,
      "step": 25909
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6316715478897095,
      "learning_rate": 0.000581472033951106,
      "loss": 2.326,
      "step": 25910
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6853234767913818,
      "learning_rate": 0.0005814706186525572,
      "loss": 2.1116,
      "step": 25911
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.8416805863380432,
      "learning_rate": 0.0005814692033016776,
      "loss": 2.2815,
      "step": 25912
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6893496513366699,
      "learning_rate": 0.0005814677878984676,
      "loss": 2.3616,
      "step": 25913
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.7609348297119141,
      "learning_rate": 0.0005814663724429275,
      "loss": 2.4192,
      "step": 25914
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.9388720393180847,
      "learning_rate": 0.0005814649569350573,
      "loss": 2.4267,
      "step": 25915
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6936537027359009,
      "learning_rate": 0.0005814635413748576,
      "loss": 2.0444,
      "step": 25916
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.8392847776412964,
      "learning_rate": 0.0005814621257623284,
      "loss": 2.2842,
      "step": 25917
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.7288355231285095,
      "learning_rate": 0.0005814607100974702,
      "loss": 2.2562,
      "step": 25918
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6328635215759277,
      "learning_rate": 0.0005814592943802832,
      "loss": 2.3081,
      "step": 25919
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.7308474183082581,
      "learning_rate": 0.0005814578786107674,
      "loss": 2.405,
      "step": 25920
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.7136160135269165,
      "learning_rate": 0.0005814564627889234,
      "loss": 2.4673,
      "step": 25921
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6120850443840027,
      "learning_rate": 0.0005814550469147512,
      "loss": 2.4251,
      "step": 25922
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6545038819313049,
      "learning_rate": 0.0005814536309882513,
      "loss": 2.4428,
      "step": 25923
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6372079253196716,
      "learning_rate": 0.000581452215009424,
      "loss": 2.3064,
      "step": 25924
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6405465006828308,
      "learning_rate": 0.0005814507989782692,
      "loss": 2.1669,
      "step": 25925
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6280663013458252,
      "learning_rate": 0.0005814493828947874,
      "loss": 2.1675,
      "step": 25926
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6199722290039062,
      "learning_rate": 0.0005814479667589789,
      "loss": 2.3947,
      "step": 25927
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6655603647232056,
      "learning_rate": 0.0005814465505708439,
      "loss": 2.4493,
      "step": 25928
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6506255865097046,
      "learning_rate": 0.0005814451343303827,
      "loss": 2.1974,
      "step": 25929
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.62458336353302,
      "learning_rate": 0.0005814437180375955,
      "loss": 2.2854,
      "step": 25930
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.602264404296875,
      "learning_rate": 0.0005814423016924828,
      "loss": 2.4026,
      "step": 25931
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.570132315158844,
      "learning_rate": 0.0005814408852950443,
      "loss": 2.5101,
      "step": 25932
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6321052312850952,
      "learning_rate": 0.0005814394688452809,
      "loss": 2.4694,
      "step": 25933
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.5909283757209778,
      "learning_rate": 0.0005814380523431925,
      "loss": 1.965,
      "step": 25934
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6616671085357666,
      "learning_rate": 0.0005814366357887794,
      "loss": 2.171,
      "step": 25935
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.580815315246582,
      "learning_rate": 0.000581435219182042,
      "loss": 2.2586,
      "step": 25936
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6326513290405273,
      "learning_rate": 0.0005814338025229804,
      "loss": 2.4144,
      "step": 25937
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6507244110107422,
      "learning_rate": 0.000581432385811595,
      "loss": 2.3138,
      "step": 25938
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.620175838470459,
      "learning_rate": 0.0005814309690478861,
      "loss": 2.3044,
      "step": 25939
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6182546019554138,
      "learning_rate": 0.0005814295522318537,
      "loss": 2.4489,
      "step": 25940
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6007350087165833,
      "learning_rate": 0.0005814281353634983,
      "loss": 2.3134,
      "step": 25941
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.5828850865364075,
      "learning_rate": 0.0005814267184428201,
      "loss": 2.1763,
      "step": 25942
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6342741847038269,
      "learning_rate": 0.0005814253014698194,
      "loss": 2.3829,
      "step": 25943
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6169868111610413,
      "learning_rate": 0.0005814238844444963,
      "loss": 2.3606,
      "step": 25944
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.7223935127258301,
      "learning_rate": 0.0005814224673668513,
      "loss": 2.3224,
      "step": 25945
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6540842652320862,
      "learning_rate": 0.0005814210502368846,
      "loss": 2.2868,
      "step": 25946
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6409876942634583,
      "learning_rate": 0.0005814196330545964,
      "loss": 2.2202,
      "step": 25947
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6138046383857727,
      "learning_rate": 0.0005814182158199869,
      "loss": 2.4608,
      "step": 25948
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.606504499912262,
      "learning_rate": 0.0005814167985330565,
      "loss": 2.2353,
      "step": 25949
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6012985110282898,
      "learning_rate": 0.0005814153811938053,
      "loss": 2.446,
      "step": 25950
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6801371574401855,
      "learning_rate": 0.0005814139638022337,
      "loss": 2.2896,
      "step": 25951
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6570085287094116,
      "learning_rate": 0.0005814125463583419,
      "loss": 2.2726,
      "step": 25952
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.7460758686065674,
      "learning_rate": 0.0005814111288621304,
      "loss": 2.1997,
      "step": 25953
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6310920119285583,
      "learning_rate": 0.0005814097113135991,
      "loss": 2.4034,
      "step": 25954
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.5875423550605774,
      "learning_rate": 0.0005814082937127484,
      "loss": 2.1794,
      "step": 25955
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.5899314284324646,
      "learning_rate": 0.0005814068760595786,
      "loss": 2.2541,
      "step": 25956
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.8044130206108093,
      "learning_rate": 0.00058140545835409,
      "loss": 2.2051,
      "step": 25957
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6070972681045532,
      "learning_rate": 0.0005814040405962828,
      "loss": 1.9293,
      "step": 25958
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6229360103607178,
      "learning_rate": 0.0005814026227861573,
      "loss": 2.3817,
      "step": 25959
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.7524041533470154,
      "learning_rate": 0.0005814012049237137,
      "loss": 2.191,
      "step": 25960
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6530249714851379,
      "learning_rate": 0.0005813997870089524,
      "loss": 2.0798,
      "step": 25961
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6415453553199768,
      "learning_rate": 0.0005813983690418735,
      "loss": 2.3842,
      "step": 25962
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6331802010536194,
      "learning_rate": 0.0005813969510224773,
      "loss": 2.4565,
      "step": 25963
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.7029156684875488,
      "learning_rate": 0.000581395532950764,
      "loss": 2.3711,
      "step": 25964
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6760379076004028,
      "learning_rate": 0.0005813941148267341,
      "loss": 1.981,
      "step": 25965
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6496424674987793,
      "learning_rate": 0.0005813926966503877,
      "loss": 2.3766,
      "step": 25966
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.718299388885498,
      "learning_rate": 0.0005813912784217251,
      "loss": 2.578,
      "step": 25967
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6821991801261902,
      "learning_rate": 0.0005813898601407465,
      "loss": 2.356,
      "step": 25968
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.623275101184845,
      "learning_rate": 0.0005813884418074522,
      "loss": 2.1781,
      "step": 25969
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6259307265281677,
      "learning_rate": 0.0005813870234218425,
      "loss": 2.3871,
      "step": 25970
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6395869851112366,
      "learning_rate": 0.0005813856049839177,
      "loss": 2.3747,
      "step": 25971
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.845309853553772,
      "learning_rate": 0.000581384186493678,
      "loss": 2.1555,
      "step": 25972
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.7623927593231201,
      "learning_rate": 0.0005813827679511237,
      "loss": 2.2276,
      "step": 25973
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6820357441902161,
      "learning_rate": 0.0005813813493562548,
      "loss": 2.397,
      "step": 25974
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.7455716133117676,
      "learning_rate": 0.000581379930709072,
      "loss": 2.3217,
      "step": 25975
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.7559753656387329,
      "learning_rate": 0.0005813785120095754,
      "loss": 2.237,
      "step": 25976
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.725515604019165,
      "learning_rate": 0.0005813770932577651,
      "loss": 2.3682,
      "step": 25977
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6369705200195312,
      "learning_rate": 0.0005813756744536415,
      "loss": 2.0384,
      "step": 25978
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.7475936412811279,
      "learning_rate": 0.0005813742555972049,
      "loss": 2.4171,
      "step": 25979
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.8082791566848755,
      "learning_rate": 0.0005813728366884554,
      "loss": 2.1412,
      "step": 25980
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.7332524657249451,
      "learning_rate": 0.0005813714177273936,
      "loss": 2.3096,
      "step": 25981
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6425700187683105,
      "learning_rate": 0.0005813699987140193,
      "loss": 2.3149,
      "step": 25982
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6671159267425537,
      "learning_rate": 0.0005813685796483332,
      "loss": 2.4187,
      "step": 25983
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6789140105247498,
      "learning_rate": 0.0005813671605303353,
      "loss": 2.3905,
      "step": 25984
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6714151501655579,
      "learning_rate": 0.0005813657413600259,
      "loss": 2.268,
      "step": 25985
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.7074413299560547,
      "learning_rate": 0.0005813643221374053,
      "loss": 2.5488,
      "step": 25986
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6502803564071655,
      "learning_rate": 0.0005813629028624738,
      "loss": 2.3161,
      "step": 25987
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6418343782424927,
      "learning_rate": 0.0005813614835352316,
      "loss": 2.422,
      "step": 25988
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.7784654498100281,
      "learning_rate": 0.0005813600641556791,
      "loss": 2.3508,
      "step": 25989
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6816797852516174,
      "learning_rate": 0.0005813586447238162,
      "loss": 2.2497,
      "step": 25990
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6402162313461304,
      "learning_rate": 0.0005813572252396436,
      "loss": 2.1607,
      "step": 25991
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6776546835899353,
      "learning_rate": 0.0005813558057031614,
      "loss": 2.2651,
      "step": 25992
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6287025213241577,
      "learning_rate": 0.0005813543861143698,
      "loss": 2.1674,
      "step": 25993
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.720553994178772,
      "learning_rate": 0.0005813529664732691,
      "loss": 2.5395,
      "step": 25994
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6650837659835815,
      "learning_rate": 0.0005813515467798596,
      "loss": 2.5727,
      "step": 25995
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6613273620605469,
      "learning_rate": 0.0005813501270341415,
      "loss": 2.3473,
      "step": 25996
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6180377006530762,
      "learning_rate": 0.000581348707236115,
      "loss": 2.4286,
      "step": 25997
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.7577950954437256,
      "learning_rate": 0.0005813472873857806,
      "loss": 2.2943,
      "step": 25998
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6174047589302063,
      "learning_rate": 0.0005813458674831384,
      "loss": 2.4117,
      "step": 25999
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6286974549293518,
      "learning_rate": 0.0005813444475281887,
      "loss": 2.1194,
      "step": 26000
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6368849873542786,
      "learning_rate": 0.0005813430275209317,
      "loss": 2.3912,
      "step": 26001
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.5663120746612549,
      "learning_rate": 0.0005813416074613679,
      "loss": 2.3117,
      "step": 26002
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6178837418556213,
      "learning_rate": 0.0005813401873494972,
      "loss": 2.1974,
      "step": 26003
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6802358031272888,
      "learning_rate": 0.0005813387671853201,
      "loss": 2.2576,
      "step": 26004
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6342674493789673,
      "learning_rate": 0.0005813373469688368,
      "loss": 2.1874,
      "step": 26005
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6970568895339966,
      "learning_rate": 0.0005813359267000476,
      "loss": 2.1863,
      "step": 26006
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.7773095369338989,
      "learning_rate": 0.0005813345063789527,
      "loss": 2.5472,
      "step": 26007
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.672831118106842,
      "learning_rate": 0.0005813330860055525,
      "loss": 2.3361,
      "step": 26008
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6049827337265015,
      "learning_rate": 0.000581331665579847,
      "loss": 2.5386,
      "step": 26009
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6260182857513428,
      "learning_rate": 0.0005813302451018368,
      "loss": 2.3863,
      "step": 26010
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.5995084047317505,
      "learning_rate": 0.0005813288245715219,
      "loss": 2.1923,
      "step": 26011
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6920987367630005,
      "learning_rate": 0.0005813274039889027,
      "loss": 2.2112,
      "step": 26012
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6242803335189819,
      "learning_rate": 0.0005813259833539795,
      "loss": 2.3659,
      "step": 26013
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6678446531295776,
      "learning_rate": 0.0005813245626667524,
      "loss": 2.2312,
      "step": 26014
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6214403510093689,
      "learning_rate": 0.0005813231419272217,
      "loss": 2.2638,
      "step": 26015
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6614525318145752,
      "learning_rate": 0.0005813217211353878,
      "loss": 2.3866,
      "step": 26016
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.7100443243980408,
      "learning_rate": 0.0005813203002912509,
      "loss": 2.3838,
      "step": 26017
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6095941066741943,
      "learning_rate": 0.0005813188793948112,
      "loss": 2.1756,
      "step": 26018
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.9603554606437683,
      "learning_rate": 0.000581317458446069,
      "loss": 2.246,
      "step": 26019
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6566357612609863,
      "learning_rate": 0.0005813160374450246,
      "loss": 2.4111,
      "step": 26020
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6927056908607483,
      "learning_rate": 0.0005813146163916782,
      "loss": 2.3439,
      "step": 26021
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6354675889015198,
      "learning_rate": 0.0005813131952860303,
      "loss": 2.3511,
      "step": 26022
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6478479504585266,
      "learning_rate": 0.0005813117741280807,
      "loss": 2.2154,
      "step": 26023
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6542181372642517,
      "learning_rate": 0.0005813103529178301,
      "loss": 2.3302,
      "step": 26024
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.57330322265625,
      "learning_rate": 0.0005813089316552784,
      "loss": 2.3114,
      "step": 26025
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6310253143310547,
      "learning_rate": 0.0005813075103404263,
      "loss": 2.1836,
      "step": 26026
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6409161686897278,
      "learning_rate": 0.0005813060889732737,
      "loss": 2.2949,
      "step": 26027
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.5867726802825928,
      "learning_rate": 0.000581304667553821,
      "loss": 2.1918,
      "step": 26028
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6508913636207581,
      "learning_rate": 0.0005813032460820684,
      "loss": 2.3291,
      "step": 26029
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6133301258087158,
      "learning_rate": 0.0005813018245580163,
      "loss": 2.5072,
      "step": 26030
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6536859273910522,
      "learning_rate": 0.0005813004029816648,
      "loss": 2.4881,
      "step": 26031
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6436787843704224,
      "learning_rate": 0.0005812989813530143,
      "loss": 2.5711,
      "step": 26032
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6335482001304626,
      "learning_rate": 0.0005812975596720651,
      "loss": 2.424,
      "step": 26033
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6516212224960327,
      "learning_rate": 0.0005812961379388172,
      "loss": 2.2942,
      "step": 26034
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.7410359978675842,
      "learning_rate": 0.0005812947161532712,
      "loss": 2.2888,
      "step": 26035
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6463249921798706,
      "learning_rate": 0.000581293294315427,
      "loss": 2.2215,
      "step": 26036
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6967027187347412,
      "learning_rate": 0.0005812918724252853,
      "loss": 2.4107,
      "step": 26037
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6860085725784302,
      "learning_rate": 0.0005812904504828459,
      "loss": 2.2764,
      "step": 26038
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6294246912002563,
      "learning_rate": 0.0005812890284881095,
      "loss": 2.1326,
      "step": 26039
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.5955684185028076,
      "learning_rate": 0.0005812876064410761,
      "loss": 2.5139,
      "step": 26040
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6217383146286011,
      "learning_rate": 0.000581286184341746,
      "loss": 2.139,
      "step": 26041
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.60264652967453,
      "learning_rate": 0.0005812847621901194,
      "loss": 2.2653,
      "step": 26042
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.8388547301292419,
      "learning_rate": 0.0005812833399861968,
      "loss": 2.3976,
      "step": 26043
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.5953347086906433,
      "learning_rate": 0.0005812819177299783,
      "loss": 2.0759,
      "step": 26044
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.7844394445419312,
      "learning_rate": 0.000581280495421464,
      "loss": 2.4475,
      "step": 26045
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6627346277236938,
      "learning_rate": 0.0005812790730606546,
      "loss": 2.2575,
      "step": 26046
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6233546733856201,
      "learning_rate": 0.00058127765064755,
      "loss": 2.2867,
      "step": 26047
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.7159403562545776,
      "learning_rate": 0.0005812762281821505,
      "loss": 2.2883,
      "step": 26048
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.7178946137428284,
      "learning_rate": 0.0005812748056644565,
      "loss": 2.4076,
      "step": 26049
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.589131772518158,
      "learning_rate": 0.0005812733830944682,
      "loss": 2.3618,
      "step": 26050
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6416588425636292,
      "learning_rate": 0.0005812719604721859,
      "loss": 2.3148,
      "step": 26051
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6783321499824524,
      "learning_rate": 0.0005812705377976099,
      "loss": 2.447,
      "step": 26052
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.616206705570221,
      "learning_rate": 0.0005812691150707402,
      "loss": 2.4047,
      "step": 26053
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.5989311933517456,
      "learning_rate": 0.0005812676922915775,
      "loss": 2.1296,
      "step": 26054
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6165961027145386,
      "learning_rate": 0.0005812662694601216,
      "loss": 2.2489,
      "step": 26055
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6172420978546143,
      "learning_rate": 0.0005812648465763731,
      "loss": 2.2661,
      "step": 26056
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.645472526550293,
      "learning_rate": 0.0005812634236403322,
      "loss": 2.2674,
      "step": 26057
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.7632404565811157,
      "learning_rate": 0.000581262000651999,
      "loss": 2.375,
      "step": 26058
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.670993447303772,
      "learning_rate": 0.000581260577611374,
      "loss": 2.2602,
      "step": 26059
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.5834769606590271,
      "learning_rate": 0.0005812591545184573,
      "loss": 2.1984,
      "step": 26060
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6314943432807922,
      "learning_rate": 0.0005812577313732491,
      "loss": 2.1555,
      "step": 26061
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6215154528617859,
      "learning_rate": 0.0005812563081757499,
      "loss": 2.4208,
      "step": 26062
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6142405271530151,
      "learning_rate": 0.00058125488492596,
      "loss": 2.393,
      "step": 26063
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6476033329963684,
      "learning_rate": 0.0005812534616238793,
      "loss": 2.3295,
      "step": 26064
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6646994948387146,
      "learning_rate": 0.0005812520382695083,
      "loss": 2.5054,
      "step": 26065
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6013259887695312,
      "learning_rate": 0.0005812506148628472,
      "loss": 2.2704,
      "step": 26066
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6217949390411377,
      "learning_rate": 0.0005812491914038964,
      "loss": 2.3083,
      "step": 26067
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6243450045585632,
      "learning_rate": 0.000581247767892656,
      "loss": 2.3592,
      "step": 26068
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6408895254135132,
      "learning_rate": 0.0005812463443291264,
      "loss": 2.1376,
      "step": 26069
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.606511652469635,
      "learning_rate": 0.0005812449207133078,
      "loss": 2.0903,
      "step": 26070
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6925829648971558,
      "learning_rate": 0.0005812434970452004,
      "loss": 2.3549,
      "step": 26071
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.5954943895339966,
      "learning_rate": 0.0005812420733248045,
      "loss": 2.2705,
      "step": 26072
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6709488034248352,
      "learning_rate": 0.0005812406495521204,
      "loss": 2.284,
      "step": 26073
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6437393426895142,
      "learning_rate": 0.0005812392257271484,
      "loss": 2.2247,
      "step": 26074
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.636142909526825,
      "learning_rate": 0.0005812378018498888,
      "loss": 2.5169,
      "step": 26075
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.707514226436615,
      "learning_rate": 0.0005812363779203416,
      "loss": 2.2945,
      "step": 26076
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6475133895874023,
      "learning_rate": 0.0005812349539385073,
      "loss": 2.2187,
      "step": 26077
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6659235954284668,
      "learning_rate": 0.0005812335299043862,
      "loss": 2.5727,
      "step": 26078
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6073641777038574,
      "learning_rate": 0.0005812321058179784,
      "loss": 2.5051,
      "step": 26079
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6859219670295715,
      "learning_rate": 0.0005812306816792843,
      "loss": 2.0799,
      "step": 26080
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6236618757247925,
      "learning_rate": 0.0005812292574883041,
      "loss": 2.347,
      "step": 26081
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6402702331542969,
      "learning_rate": 0.000581227833245038,
      "loss": 2.4238,
      "step": 26082
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6287938952445984,
      "learning_rate": 0.0005812264089494864,
      "loss": 2.4126,
      "step": 26083
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6232797503471375,
      "learning_rate": 0.0005812249846016494,
      "loss": 2.3011,
      "step": 26084
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6240367293357849,
      "learning_rate": 0.0005812235602015275,
      "loss": 2.2499,
      "step": 26085
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6059389114379883,
      "learning_rate": 0.0005812221357491208,
      "loss": 2.4591,
      "step": 26086
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.582384467124939,
      "learning_rate": 0.0005812207112444296,
      "loss": 2.2462,
      "step": 26087
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6521455645561218,
      "learning_rate": 0.0005812192866874542,
      "loss": 2.2683,
      "step": 26088
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.5895342826843262,
      "learning_rate": 0.0005812178620781947,
      "loss": 2.5115,
      "step": 26089
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6339779496192932,
      "learning_rate": 0.0005812164374166516,
      "loss": 2.2835,
      "step": 26090
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.59125155210495,
      "learning_rate": 0.000581215012702825,
      "loss": 2.4983,
      "step": 26091
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6937459707260132,
      "learning_rate": 0.0005812135879367152,
      "loss": 2.2153,
      "step": 26092
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.7446779608726501,
      "learning_rate": 0.0005812121631183226,
      "loss": 2.4392,
      "step": 26093
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.5914767980575562,
      "learning_rate": 0.0005812107382476472,
      "loss": 2.3826,
      "step": 26094
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6151250600814819,
      "learning_rate": 0.0005812093133246895,
      "loss": 2.2248,
      "step": 26095
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.634964108467102,
      "learning_rate": 0.0005812078883494497,
      "loss": 2.5728,
      "step": 26096
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.7012479305267334,
      "learning_rate": 0.0005812064633219279,
      "loss": 2.5662,
      "step": 26097
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.5561320781707764,
      "learning_rate": 0.0005812050382421247,
      "loss": 1.9852,
      "step": 26098
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6793424487113953,
      "learning_rate": 0.0005812036131100401,
      "loss": 2.2943,
      "step": 26099
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6031100153923035,
      "learning_rate": 0.0005812021879256744,
      "loss": 2.2451,
      "step": 26100
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6312190294265747,
      "learning_rate": 0.000581200762689028,
      "loss": 2.4318,
      "step": 26101
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.7654415965080261,
      "learning_rate": 0.0005811993374001009,
      "loss": 2.4513,
      "step": 26102
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.5884124636650085,
      "learning_rate": 0.0005811979120588938,
      "loss": 2.1089,
      "step": 26103
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.7449952960014343,
      "learning_rate": 0.0005811964866654065,
      "loss": 2.2282,
      "step": 26104
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6176977753639221,
      "learning_rate": 0.0005811950612196394,
      "loss": 2.3298,
      "step": 26105
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.609769344329834,
      "learning_rate": 0.0005811936357215931,
      "loss": 2.5299,
      "step": 26106
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6274175643920898,
      "learning_rate": 0.0005811922101712673,
      "loss": 2.4367,
      "step": 26107
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6621924638748169,
      "learning_rate": 0.0005811907845686628,
      "loss": 2.2602,
      "step": 26108
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6964804530143738,
      "learning_rate": 0.0005811893589137794,
      "loss": 2.6332,
      "step": 26109
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.647646427154541,
      "learning_rate": 0.0005811879332066178,
      "loss": 2.4131,
      "step": 26110
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6715236902236938,
      "learning_rate": 0.000581186507447178,
      "loss": 2.3333,
      "step": 26111
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6539617776870728,
      "learning_rate": 0.0005811850816354603,
      "loss": 2.502,
      "step": 26112
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.598252534866333,
      "learning_rate": 0.000581183655771465,
      "loss": 2.3515,
      "step": 26113
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.7102690935134888,
      "learning_rate": 0.0005811822298551922,
      "loss": 2.2308,
      "step": 26114
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6700822710990906,
      "learning_rate": 0.0005811808038866425,
      "loss": 2.3942,
      "step": 26115
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.5771726369857788,
      "learning_rate": 0.000581179377865816,
      "loss": 2.3443,
      "step": 26116
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.5796581506729126,
      "learning_rate": 0.0005811779517927128,
      "loss": 2.3544,
      "step": 26117
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6474080681800842,
      "learning_rate": 0.0005811765256673334,
      "loss": 2.4138,
      "step": 26118
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6093466281890869,
      "learning_rate": 0.000581175099489678,
      "loss": 2.3866,
      "step": 26119
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.5880640149116516,
      "learning_rate": 0.0005811736732597467,
      "loss": 2.2064,
      "step": 26120
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6437089443206787,
      "learning_rate": 0.00058117224697754,
      "loss": 2.2666,
      "step": 26121
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6027914881706238,
      "learning_rate": 0.0005811708206430582,
      "loss": 2.273,
      "step": 26122
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6494637131690979,
      "learning_rate": 0.0005811693942563013,
      "loss": 2.2188,
      "step": 26123
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6542783379554749,
      "learning_rate": 0.0005811679678172696,
      "loss": 2.3776,
      "step": 26124
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6145679950714111,
      "learning_rate": 0.0005811665413259637,
      "loss": 2.4867,
      "step": 26125
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.620587170124054,
      "learning_rate": 0.0005811651147823835,
      "loss": 2.1387,
      "step": 26126
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6687279343605042,
      "learning_rate": 0.0005811636881865295,
      "loss": 2.2176,
      "step": 26127
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6824581027030945,
      "learning_rate": 0.0005811622615384016,
      "loss": 2.2164,
      "step": 26128
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6567847728729248,
      "learning_rate": 0.0005811608348380006,
      "loss": 2.2636,
      "step": 26129
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6120212078094482,
      "learning_rate": 0.0005811594080853263,
      "loss": 2.4652,
      "step": 26130
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6135448217391968,
      "learning_rate": 0.0005811579812803793,
      "loss": 2.2395,
      "step": 26131
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6792681217193604,
      "learning_rate": 0.0005811565544231597,
      "loss": 2.1691,
      "step": 26132
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6552236080169678,
      "learning_rate": 0.0005811551275136678,
      "loss": 2.2065,
      "step": 26133
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.5867249965667725,
      "learning_rate": 0.0005811537005519038,
      "loss": 2.3988,
      "step": 26134
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6390319466590881,
      "learning_rate": 0.0005811522735378681,
      "loss": 2.2156,
      "step": 26135
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.733841061592102,
      "learning_rate": 0.0005811508464715608,
      "loss": 2.3198,
      "step": 26136
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6317387819290161,
      "learning_rate": 0.0005811494193529823,
      "loss": 2.1966,
      "step": 26137
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.7450711727142334,
      "learning_rate": 0.0005811479921821327,
      "loss": 2.4392,
      "step": 26138
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6694661974906921,
      "learning_rate": 0.0005811465649590126,
      "loss": 2.1783,
      "step": 26139
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6618458032608032,
      "learning_rate": 0.0005811451376836219,
      "loss": 2.3778,
      "step": 26140
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6313132047653198,
      "learning_rate": 0.000581143710355961,
      "loss": 2.2968,
      "step": 26141
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6494787931442261,
      "learning_rate": 0.0005811422829760303,
      "loss": 2.2897,
      "step": 26142
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6919484734535217,
      "learning_rate": 0.0005811408555438299,
      "loss": 2.1756,
      "step": 26143
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.5845808386802673,
      "learning_rate": 0.00058113942805936,
      "loss": 2.2651,
      "step": 26144
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6169514060020447,
      "learning_rate": 0.0005811380005226209,
      "loss": 2.3305,
      "step": 26145
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.64002525806427,
      "learning_rate": 0.0005811365729336132,
      "loss": 2.3223,
      "step": 26146
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6541207432746887,
      "learning_rate": 0.0005811351452923367,
      "loss": 2.0981,
      "step": 26147
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.650835394859314,
      "learning_rate": 0.000581133717598792,
      "loss": 2.2143,
      "step": 26148
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6193585991859436,
      "learning_rate": 0.0005811322898529792,
      "loss": 2.1469,
      "step": 26149
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6858620047569275,
      "learning_rate": 0.0005811308620548985,
      "loss": 2.2667,
      "step": 26150
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.8177154064178467,
      "learning_rate": 0.0005811294342045504,
      "loss": 2.1851,
      "step": 26151
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6536146402359009,
      "learning_rate": 0.000581128006301935,
      "loss": 2.3668,
      "step": 26152
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.7384441494941711,
      "learning_rate": 0.0005811265783470526,
      "loss": 2.3174,
      "step": 26153
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.8460744023323059,
      "learning_rate": 0.0005811251503399034,
      "loss": 2.2867,
      "step": 26154
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6648468971252441,
      "learning_rate": 0.0005811237222804878,
      "loss": 2.2622,
      "step": 26155
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.640080988407135,
      "learning_rate": 0.000581122294168806,
      "loss": 2.4082,
      "step": 26156
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6731557250022888,
      "learning_rate": 0.0005811208660048582,
      "loss": 2.5071,
      "step": 26157
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6435702443122864,
      "learning_rate": 0.0005811194377886447,
      "loss": 2.4691,
      "step": 26158
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.7391510605812073,
      "learning_rate": 0.0005811180095201658,
      "loss": 2.6183,
      "step": 26159
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6652306318283081,
      "learning_rate": 0.0005811165811994218,
      "loss": 2.3867,
      "step": 26160
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.7351309061050415,
      "learning_rate": 0.000581115152826413,
      "loss": 2.1938,
      "step": 26161
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.7342413663864136,
      "learning_rate": 0.0005811137244011394,
      "loss": 2.3509,
      "step": 26162
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6318502426147461,
      "learning_rate": 0.0005811122959236017,
      "loss": 2.1343,
      "step": 26163
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6328320503234863,
      "learning_rate": 0.0005811108673937997,
      "loss": 2.2157,
      "step": 26164
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.7845767140388489,
      "learning_rate": 0.0005811094388117339,
      "loss": 2.4266,
      "step": 26165
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.7149472832679749,
      "learning_rate": 0.0005811080101774046,
      "loss": 2.5485,
      "step": 26166
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.5784597396850586,
      "learning_rate": 0.0005811065814908121,
      "loss": 2.0685,
      "step": 26167
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.633479118347168,
      "learning_rate": 0.0005811051527519564,
      "loss": 2.1659,
      "step": 26168
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6293123364448547,
      "learning_rate": 0.0005811037239608381,
      "loss": 2.3682,
      "step": 26169
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6870065331459045,
      "learning_rate": 0.0005811022951174573,
      "loss": 2.1481,
      "step": 26170
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.7040804624557495,
      "learning_rate": 0.0005811008662218142,
      "loss": 2.3391,
      "step": 26171
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6744619011878967,
      "learning_rate": 0.0005810994372739092,
      "loss": 2.2691,
      "step": 26172
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6186566352844238,
      "learning_rate": 0.0005810980082737424,
      "loss": 2.2666,
      "step": 26173
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.7227048277854919,
      "learning_rate": 0.0005810965792213143,
      "loss": 2.3256,
      "step": 26174
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6911154389381409,
      "learning_rate": 0.000581095150116625,
      "loss": 2.6028,
      "step": 26175
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.7204148769378662,
      "learning_rate": 0.0005810937209596749,
      "loss": 2.2862,
      "step": 26176
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.8779371976852417,
      "learning_rate": 0.000581092291750464,
      "loss": 2.4088,
      "step": 26177
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6294168829917908,
      "learning_rate": 0.0005810908624889928,
      "loss": 2.3196,
      "step": 26178
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6519826650619507,
      "learning_rate": 0.0005810894331752615,
      "loss": 2.3432,
      "step": 26179
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6770326495170593,
      "learning_rate": 0.0005810880038092705,
      "loss": 2.4281,
      "step": 26180
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.7200237512588501,
      "learning_rate": 0.0005810865743910197,
      "loss": 2.4776,
      "step": 26181
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.5797921419143677,
      "learning_rate": 0.0005810851449205097,
      "loss": 2.0292,
      "step": 26182
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.717570424079895,
      "learning_rate": 0.0005810837153977408,
      "loss": 2.3393,
      "step": 26183
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6956343054771423,
      "learning_rate": 0.000581082285822713,
      "loss": 2.3484,
      "step": 26184
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6106548309326172,
      "learning_rate": 0.0005810808561954266,
      "loss": 2.285,
      "step": 26185
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6582479476928711,
      "learning_rate": 0.0005810794265158822,
      "loss": 1.9166,
      "step": 26186
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6355887651443481,
      "learning_rate": 0.0005810779967840796,
      "loss": 2.4249,
      "step": 26187
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.5832211375236511,
      "learning_rate": 0.0005810765670000195,
      "loss": 2.3526,
      "step": 26188
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6233799457550049,
      "learning_rate": 0.0005810751371637017,
      "loss": 2.3281,
      "step": 26189
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6324160695075989,
      "learning_rate": 0.000581073707275127,
      "loss": 2.3213,
      "step": 26190
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.5949576497077942,
      "learning_rate": 0.0005810722773342953,
      "loss": 2.2464,
      "step": 26191
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.5967501997947693,
      "learning_rate": 0.0005810708473412069,
      "loss": 2.5702,
      "step": 26192
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6194813847541809,
      "learning_rate": 0.0005810694172958621,
      "loss": 2.3965,
      "step": 26193
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6312713027000427,
      "learning_rate": 0.0005810679871982613,
      "loss": 2.2626,
      "step": 26194
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.5938237309455872,
      "learning_rate": 0.0005810665570484046,
      "loss": 2.1271,
      "step": 26195
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6859480738639832,
      "learning_rate": 0.0005810651268462923,
      "loss": 2.2722,
      "step": 26196
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.684780478477478,
      "learning_rate": 0.0005810636965919247,
      "loss": 2.2808,
      "step": 26197
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.5980008244514465,
      "learning_rate": 0.0005810622662853021,
      "loss": 2.3795,
      "step": 26198
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6022534966468811,
      "learning_rate": 0.0005810608359264246,
      "loss": 2.1942,
      "step": 26199
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6754307746887207,
      "learning_rate": 0.0005810594055152927,
      "loss": 2.4381,
      "step": 26200
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.700508713722229,
      "learning_rate": 0.0005810579750519066,
      "loss": 2.4341,
      "step": 26201
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6153152585029602,
      "learning_rate": 0.0005810565445362664,
      "loss": 2.4064,
      "step": 26202
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6603583097457886,
      "learning_rate": 0.0005810551139683725,
      "loss": 2.5468,
      "step": 26203
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.658284604549408,
      "learning_rate": 0.0005810536833482252,
      "loss": 2.1288,
      "step": 26204
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6092149615287781,
      "learning_rate": 0.0005810522526758247,
      "loss": 2.1669,
      "step": 26205
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6030750870704651,
      "learning_rate": 0.0005810508219511712,
      "loss": 2.2394,
      "step": 26206
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.7156838774681091,
      "learning_rate": 0.0005810493911742651,
      "loss": 2.3509,
      "step": 26207
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.660097599029541,
      "learning_rate": 0.0005810479603451067,
      "loss": 2.2293,
      "step": 26208
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.661855161190033,
      "learning_rate": 0.0005810465294636961,
      "loss": 2.3063,
      "step": 26209
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.7582082748413086,
      "learning_rate": 0.0005810450985300337,
      "loss": 2.4004,
      "step": 26210
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.7040150165557861,
      "learning_rate": 0.0005810436675441195,
      "loss": 2.3929,
      "step": 26211
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6083287000656128,
      "learning_rate": 0.0005810422365059542,
      "loss": 2.3086,
      "step": 26212
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6730604767799377,
      "learning_rate": 0.0005810408054155378,
      "loss": 2.1591,
      "step": 26213
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6986847519874573,
      "learning_rate": 0.0005810393742728706,
      "loss": 2.3472,
      "step": 26214
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6182792782783508,
      "learning_rate": 0.0005810379430779528,
      "loss": 2.4084,
      "step": 26215
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6597681045532227,
      "learning_rate": 0.0005810365118307848,
      "loss": 2.2458,
      "step": 26216
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.7796357274055481,
      "learning_rate": 0.0005810350805313669,
      "loss": 2.5615,
      "step": 26217
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.5986844897270203,
      "learning_rate": 0.0005810336491796992,
      "loss": 2.5965,
      "step": 26218
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6062542200088501,
      "learning_rate": 0.000581032217775782,
      "loss": 2.2446,
      "step": 26219
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.5959845781326294,
      "learning_rate": 0.0005810307863196156,
      "loss": 2.3409,
      "step": 26220
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.585502028465271,
      "learning_rate": 0.0005810293548112004,
      "loss": 2.3793,
      "step": 26221
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6108485460281372,
      "learning_rate": 0.0005810279232505364,
      "loss": 2.3576,
      "step": 26222
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.599281370639801,
      "learning_rate": 0.0005810264916376241,
      "loss": 2.5166,
      "step": 26223
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6418737769126892,
      "learning_rate": 0.0005810250599724636,
      "loss": 2.2764,
      "step": 26224
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.5727266669273376,
      "learning_rate": 0.0005810236282550552,
      "loss": 2.2808,
      "step": 26225
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6658342480659485,
      "learning_rate": 0.0005810221964853993,
      "loss": 2.6251,
      "step": 26226
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.59506756067276,
      "learning_rate": 0.0005810207646634961,
      "loss": 2.2987,
      "step": 26227
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6225281357765198,
      "learning_rate": 0.0005810193327893457,
      "loss": 2.3424,
      "step": 26228
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.639994740486145,
      "learning_rate": 0.0005810179008629486,
      "loss": 2.4124,
      "step": 26229
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6183754801750183,
      "learning_rate": 0.0005810164688843048,
      "loss": 2.2165,
      "step": 26230
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6495296955108643,
      "learning_rate": 0.0005810150368534149,
      "loss": 2.1402,
      "step": 26231
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.5853462219238281,
      "learning_rate": 0.000581013604770279,
      "loss": 2.2877,
      "step": 26232
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.5818517804145813,
      "learning_rate": 0.0005810121726348973,
      "loss": 2.3384,
      "step": 26233
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.5820422768592834,
      "learning_rate": 0.0005810107404472701,
      "loss": 2.37,
      "step": 26234
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6219704747200012,
      "learning_rate": 0.0005810093082073978,
      "loss": 2.4111,
      "step": 26235
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6152334809303284,
      "learning_rate": 0.0005810078759152805,
      "loss": 2.3966,
      "step": 26236
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6501820087432861,
      "learning_rate": 0.0005810064435709186,
      "loss": 2.323,
      "step": 26237
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6988420486450195,
      "learning_rate": 0.0005810050111743122,
      "loss": 2.4893,
      "step": 26238
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6936340928077698,
      "learning_rate": 0.0005810035787254618,
      "loss": 2.3257,
      "step": 26239
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6200029850006104,
      "learning_rate": 0.0005810021462243674,
      "loss": 2.2944,
      "step": 26240
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6411730647087097,
      "learning_rate": 0.0005810007136710295,
      "loss": 2.3738,
      "step": 26241
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.651957094669342,
      "learning_rate": 0.0005809992810654483,
      "loss": 2.2152,
      "step": 26242
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6723676919937134,
      "learning_rate": 0.0005809978484076239,
      "loss": 2.5537,
      "step": 26243
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6265376806259155,
      "learning_rate": 0.0005809964156975566,
      "loss": 2.2519,
      "step": 26244
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6348481774330139,
      "learning_rate": 0.000580994982935247,
      "loss": 2.2278,
      "step": 26245
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.590969443321228,
      "learning_rate": 0.000580993550120695,
      "loss": 2.2376,
      "step": 26246
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6488975286483765,
      "learning_rate": 0.000580992117253901,
      "loss": 2.5593,
      "step": 26247
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6545653343200684,
      "learning_rate": 0.0005809906843348653,
      "loss": 2.4902,
      "step": 26248
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.5905066728591919,
      "learning_rate": 0.0005809892513635882,
      "loss": 2.1531,
      "step": 26249
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6341028809547424,
      "learning_rate": 0.0005809878183400697,
      "loss": 2.392,
      "step": 26250
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6214573383331299,
      "learning_rate": 0.0005809863852643104,
      "loss": 2.3648,
      "step": 26251
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6122562289237976,
      "learning_rate": 0.0005809849521363104,
      "loss": 2.3681,
      "step": 26252
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6002786755561829,
      "learning_rate": 0.0005809835189560699,
      "loss": 2.2349,
      "step": 26253
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6724344491958618,
      "learning_rate": 0.0005809820857235894,
      "loss": 2.4004,
      "step": 26254
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.5973078608512878,
      "learning_rate": 0.000580980652438869,
      "loss": 2.4906,
      "step": 26255
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.585478663444519,
      "learning_rate": 0.000580979219101909,
      "loss": 2.2888,
      "step": 26256
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6087353229522705,
      "learning_rate": 0.0005809777857127096,
      "loss": 2.2881,
      "step": 26257
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6034859418869019,
      "learning_rate": 0.0005809763522712712,
      "loss": 2.2991,
      "step": 26258
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6454847455024719,
      "learning_rate": 0.0005809749187775939,
      "loss": 2.3748,
      "step": 26259
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.5886864066123962,
      "learning_rate": 0.000580973485231678,
      "loss": 2.4654,
      "step": 26260
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6262059807777405,
      "learning_rate": 0.000580972051633524,
      "loss": 2.3183,
      "step": 26261
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.5912092924118042,
      "learning_rate": 0.0005809706179831319,
      "loss": 2.1564,
      "step": 26262
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.7399795055389404,
      "learning_rate": 0.0005809691842805021,
      "loss": 2.0968,
      "step": 26263
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.671282172203064,
      "learning_rate": 0.0005809677505256349,
      "loss": 2.2085,
      "step": 26264
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.7260920405387878,
      "learning_rate": 0.0005809663167185303,
      "loss": 2.5587,
      "step": 26265
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6198509335517883,
      "learning_rate": 0.0005809648828591889,
      "loss": 2.5405,
      "step": 26266
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.614371120929718,
      "learning_rate": 0.0005809634489476107,
      "loss": 2.3465,
      "step": 26267
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.699920117855072,
      "learning_rate": 0.0005809620149837962,
      "loss": 2.4213,
      "step": 26268
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6463153958320618,
      "learning_rate": 0.0005809605809677455,
      "loss": 2.6376,
      "step": 26269
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6597638726234436,
      "learning_rate": 0.000580959146899459,
      "loss": 2.4659,
      "step": 26270
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6496771574020386,
      "learning_rate": 0.0005809577127789368,
      "loss": 2.2682,
      "step": 26271
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6575420498847961,
      "learning_rate": 0.0005809562786061793,
      "loss": 2.26,
      "step": 26272
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.615333080291748,
      "learning_rate": 0.0005809548443811867,
      "loss": 2.107,
      "step": 26273
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6718018651008606,
      "learning_rate": 0.0005809534101039593,
      "loss": 2.412,
      "step": 26274
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6478577852249146,
      "learning_rate": 0.0005809519757744974,
      "loss": 2.3151,
      "step": 26275
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6617730855941772,
      "learning_rate": 0.0005809505413928012,
      "loss": 2.1455,
      "step": 26276
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6300541162490845,
      "learning_rate": 0.0005809491069588709,
      "loss": 2.2777,
      "step": 26277
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6541717648506165,
      "learning_rate": 0.000580947672472707,
      "loss": 1.9973,
      "step": 26278
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6455358862876892,
      "learning_rate": 0.0005809462379343096,
      "loss": 2.6445,
      "step": 26279
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.615455150604248,
      "learning_rate": 0.000580944803343679,
      "loss": 2.4068,
      "step": 26280
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6838878393173218,
      "learning_rate": 0.0005809433687008154,
      "loss": 2.355,
      "step": 26281
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6307502388954163,
      "learning_rate": 0.0005809419340057191,
      "loss": 2.3908,
      "step": 26282
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6022658944129944,
      "learning_rate": 0.0005809404992583904,
      "loss": 2.1843,
      "step": 26283
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.5898898243904114,
      "learning_rate": 0.0005809390644588297,
      "loss": 2.4024,
      "step": 26284
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.5994088649749756,
      "learning_rate": 0.000580937629607037,
      "loss": 2.3192,
      "step": 26285
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6217674016952515,
      "learning_rate": 0.0005809361947030127,
      "loss": 2.2348,
      "step": 26286
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.7107299566268921,
      "learning_rate": 0.000580934759746757,
      "loss": 2.2728,
      "step": 26287
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6619473099708557,
      "learning_rate": 0.0005809333247382703,
      "loss": 2.3146,
      "step": 26288
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.5711475014686584,
      "learning_rate": 0.0005809318896775528,
      "loss": 1.9543,
      "step": 26289
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6668288111686707,
      "learning_rate": 0.0005809304545646047,
      "loss": 2.3359,
      "step": 26290
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.707017183303833,
      "learning_rate": 0.0005809290193994264,
      "loss": 2.1616,
      "step": 26291
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.7179140448570251,
      "learning_rate": 0.000580927584182018,
      "loss": 2.4317,
      "step": 26292
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.68348228931427,
      "learning_rate": 0.00058092614891238,
      "loss": 2.227,
      "step": 26293
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.7263129949569702,
      "learning_rate": 0.0005809247135905124,
      "loss": 2.6424,
      "step": 26294
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6254876852035522,
      "learning_rate": 0.0005809232782164157,
      "loss": 2.5081,
      "step": 26295
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.5870373249053955,
      "learning_rate": 0.00058092184279009,
      "loss": 2.2174,
      "step": 26296
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6223241686820984,
      "learning_rate": 0.0005809204073115356,
      "loss": 2.3663,
      "step": 26297
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6471733450889587,
      "learning_rate": 0.0005809189717807528,
      "loss": 2.4207,
      "step": 26298
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6704460382461548,
      "learning_rate": 0.0005809175361977418,
      "loss": 2.3333,
      "step": 26299
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6077295541763306,
      "learning_rate": 0.0005809161005625031,
      "loss": 2.6375,
      "step": 26300
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6963446140289307,
      "learning_rate": 0.0005809146648750366,
      "loss": 2.4916,
      "step": 26301
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.7118518948554993,
      "learning_rate": 0.0005809132291353429,
      "loss": 2.4035,
      "step": 26302
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.5826747417449951,
      "learning_rate": 0.0005809117933434221,
      "loss": 2.2711,
      "step": 26303
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.7585581541061401,
      "learning_rate": 0.0005809103574992745,
      "loss": 2.1879,
      "step": 26304
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.615853488445282,
      "learning_rate": 0.0005809089216029003,
      "loss": 2.2165,
      "step": 26305
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6139353513717651,
      "learning_rate": 0.0005809074856543,
      "loss": 2.5809,
      "step": 26306
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.599206268787384,
      "learning_rate": 0.0005809060496534735,
      "loss": 2.1944,
      "step": 26307
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6224763989448547,
      "learning_rate": 0.0005809046136004214,
      "loss": 2.3561,
      "step": 26308
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6341978907585144,
      "learning_rate": 0.0005809031774951437,
      "loss": 2.4124,
      "step": 26309
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6218451261520386,
      "learning_rate": 0.0005809017413376409,
      "loss": 2.2129,
      "step": 26310
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6728217601776123,
      "learning_rate": 0.0005809003051279131,
      "loss": 2.573,
      "step": 26311
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6401208639144897,
      "learning_rate": 0.0005808988688659607,
      "loss": 2.4304,
      "step": 26312
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6343356966972351,
      "learning_rate": 0.0005808974325517838,
      "loss": 2.5194,
      "step": 26313
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6999520659446716,
      "learning_rate": 0.0005808959961853828,
      "loss": 2.4058,
      "step": 26314
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.60533207654953,
      "learning_rate": 0.000580894559766758,
      "loss": 2.4828,
      "step": 26315
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6464540958404541,
      "learning_rate": 0.0005808931232959095,
      "loss": 2.2549,
      "step": 26316
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.7748495936393738,
      "learning_rate": 0.0005808916867728378,
      "loss": 2.0619,
      "step": 26317
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6577528119087219,
      "learning_rate": 0.0005808902501975428,
      "loss": 2.4003,
      "step": 26318
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.7507805228233337,
      "learning_rate": 0.0005808888135700251,
      "loss": 2.3771,
      "step": 26319
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6506578326225281,
      "learning_rate": 0.000580887376890285,
      "loss": 2.564,
      "step": 26320
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6197508573532104,
      "learning_rate": 0.0005808859401583226,
      "loss": 2.4048,
      "step": 26321
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6478790044784546,
      "learning_rate": 0.0005808845033741381,
      "loss": 2.225,
      "step": 26322
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6510237455368042,
      "learning_rate": 0.000580883066537732,
      "loss": 2.4392,
      "step": 26323
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.7381997108459473,
      "learning_rate": 0.0005808816296491043,
      "loss": 2.5023,
      "step": 26324
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.5968660116195679,
      "learning_rate": 0.0005808801927082555,
      "loss": 2.4694,
      "step": 26325
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6168646812438965,
      "learning_rate": 0.0005808787557151857,
      "loss": 2.3874,
      "step": 26326
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6098443269729614,
      "learning_rate": 0.0005808773186698953,
      "loss": 2.4131,
      "step": 26327
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6241546869277954,
      "learning_rate": 0.0005808758815723844,
      "loss": 2.2117,
      "step": 26328
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6519285440444946,
      "learning_rate": 0.0005808744444226535,
      "loss": 2.1747,
      "step": 26329
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.7150251865386963,
      "learning_rate": 0.0005808730072207026,
      "loss": 2.4958,
      "step": 26330
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6164901256561279,
      "learning_rate": 0.0005808715699665323,
      "loss": 2.2827,
      "step": 26331
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6151144504547119,
      "learning_rate": 0.0005808701326601425,
      "loss": 2.2028,
      "step": 26332
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6582016944885254,
      "learning_rate": 0.0005808686953015337,
      "loss": 2.3928,
      "step": 26333
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6019651293754578,
      "learning_rate": 0.0005808672578907061,
      "loss": 2.3258,
      "step": 26334
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6353206038475037,
      "learning_rate": 0.00058086582042766,
      "loss": 2.2632,
      "step": 26335
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6980026960372925,
      "learning_rate": 0.0005808643829123956,
      "loss": 2.4194,
      "step": 26336
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6256100535392761,
      "learning_rate": 0.0005808629453449133,
      "loss": 2.2647,
      "step": 26337
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6899055242538452,
      "learning_rate": 0.0005808615077252131,
      "loss": 2.4566,
      "step": 26338
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6421663761138916,
      "learning_rate": 0.0005808600700532956,
      "loss": 2.2849,
      "step": 26339
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6039871573448181,
      "learning_rate": 0.000580858632329161,
      "loss": 2.1899,
      "step": 26340
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6416906118392944,
      "learning_rate": 0.0005808571945528092,
      "loss": 2.1236,
      "step": 26341
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6662567853927612,
      "learning_rate": 0.0005808557567242409,
      "loss": 2.161,
      "step": 26342
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6922445893287659,
      "learning_rate": 0.0005808543188434562,
      "loss": 2.342,
      "step": 26343
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6245649456977844,
      "learning_rate": 0.0005808528809104553,
      "loss": 2.2307,
      "step": 26344
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6273607015609741,
      "learning_rate": 0.0005808514429252388,
      "loss": 2.257,
      "step": 26345
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.632803738117218,
      "learning_rate": 0.0005808500048878066,
      "loss": 2.4553,
      "step": 26346
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6424241662025452,
      "learning_rate": 0.0005808485667981589,
      "loss": 2.3815,
      "step": 26347
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.7203538417816162,
      "learning_rate": 0.0005808471286562963,
      "loss": 2.4043,
      "step": 26348
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6421167850494385,
      "learning_rate": 0.000580845690462219,
      "loss": 2.4342,
      "step": 26349
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6476545929908752,
      "learning_rate": 0.000580844252215927,
      "loss": 2.4846,
      "step": 26350
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6639658212661743,
      "learning_rate": 0.0005808428139174209,
      "loss": 2.4793,
      "step": 26351
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.7359931468963623,
      "learning_rate": 0.0005808413755667007,
      "loss": 2.4539,
      "step": 26352
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6356775760650635,
      "learning_rate": 0.0005808399371637668,
      "loss": 2.4555,
      "step": 26353
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.5847716331481934,
      "learning_rate": 0.0005808384987086195,
      "loss": 2.2735,
      "step": 26354
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6823842525482178,
      "learning_rate": 0.0005808370602012591,
      "loss": 2.3681,
      "step": 26355
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6140037775039673,
      "learning_rate": 0.0005808356216416856,
      "loss": 2.4235,
      "step": 26356
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.7107511758804321,
      "learning_rate": 0.0005808341830298996,
      "loss": 2.472,
      "step": 26357
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6030884981155396,
      "learning_rate": 0.0005808327443659012,
      "loss": 2.4299,
      "step": 26358
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6553738117218018,
      "learning_rate": 0.0005808313056496906,
      "loss": 2.3639,
      "step": 26359
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6546801924705505,
      "learning_rate": 0.0005808298668812683,
      "loss": 2.389,
      "step": 26360
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6238353252410889,
      "learning_rate": 0.0005808284280606344,
      "loss": 2.3624,
      "step": 26361
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6626064777374268,
      "learning_rate": 0.0005808269891877892,
      "loss": 2.2358,
      "step": 26362
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6217078566551208,
      "learning_rate": 0.0005808255502627329,
      "loss": 2.3379,
      "step": 26363
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.657107949256897,
      "learning_rate": 0.0005808241112854658,
      "loss": 2.2457,
      "step": 26364
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6564656496047974,
      "learning_rate": 0.0005808226722559883,
      "loss": 2.336,
      "step": 26365
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6559865474700928,
      "learning_rate": 0.0005808212331743004,
      "loss": 2.2338,
      "step": 26366
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6530756950378418,
      "learning_rate": 0.0005808197940404028,
      "loss": 2.3951,
      "step": 26367
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6335193514823914,
      "learning_rate": 0.0005808183548542953,
      "loss": 2.2644,
      "step": 26368
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6012577414512634,
      "learning_rate": 0.0005808169156159784,
      "loss": 2.2912,
      "step": 26369
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6024282574653625,
      "learning_rate": 0.0005808154763254523,
      "loss": 2.1543,
      "step": 26370
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6214210987091064,
      "learning_rate": 0.0005808140369827174,
      "loss": 2.3773,
      "step": 26371
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.7228367924690247,
      "learning_rate": 0.0005808125975877739,
      "loss": 2.3414,
      "step": 26372
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6433529257774353,
      "learning_rate": 0.0005808111581406219,
      "loss": 2.3977,
      "step": 26373
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.7524605989456177,
      "learning_rate": 0.0005808097186412618,
      "loss": 2.2298,
      "step": 26374
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6852588653564453,
      "learning_rate": 0.000580808279089694,
      "loss": 2.3078,
      "step": 26375
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.7636950016021729,
      "learning_rate": 0.0005808068394859185,
      "loss": 2.2973,
      "step": 26376
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6262348294258118,
      "learning_rate": 0.0005808053998299358,
      "loss": 2.1795,
      "step": 26377
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6315234303474426,
      "learning_rate": 0.0005808039601217461,
      "loss": 2.2716,
      "step": 26378
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.7053335309028625,
      "learning_rate": 0.0005808025203613496,
      "loss": 2.4018,
      "step": 26379
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6105000972747803,
      "learning_rate": 0.0005808010805487465,
      "loss": 2.252,
      "step": 26380
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.681464672088623,
      "learning_rate": 0.0005807996406839373,
      "loss": 2.1385,
      "step": 26381
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6669883728027344,
      "learning_rate": 0.0005807982007669222,
      "loss": 2.2275,
      "step": 26382
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6358395218849182,
      "learning_rate": 0.0005807967607977014,
      "loss": 2.329,
      "step": 26383
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.5785837173461914,
      "learning_rate": 0.0005807953207762751,
      "loss": 2.263,
      "step": 26384
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6340932846069336,
      "learning_rate": 0.0005807938807026436,
      "loss": 2.2851,
      "step": 26385
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6563870906829834,
      "learning_rate": 0.0005807924405768074,
      "loss": 2.301,
      "step": 26386
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6170825958251953,
      "learning_rate": 0.0005807910003987665,
      "loss": 2.405,
      "step": 26387
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6002277731895447,
      "learning_rate": 0.0005807895601685211,
      "loss": 2.3762,
      "step": 26388
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6323695778846741,
      "learning_rate": 0.0005807881198860718,
      "loss": 2.1762,
      "step": 26389
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.7493754029273987,
      "learning_rate": 0.0005807866795514187,
      "loss": 2.5448,
      "step": 26390
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.5714055299758911,
      "learning_rate": 0.000580785239164562,
      "loss": 2.1243,
      "step": 26391
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6454352140426636,
      "learning_rate": 0.0005807837987255019,
      "loss": 2.3979,
      "step": 26392
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.9360889792442322,
      "learning_rate": 0.000580782358234239,
      "loss": 2.0678,
      "step": 26393
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6283671855926514,
      "learning_rate": 0.0005807809176907733,
      "loss": 2.4738,
      "step": 26394
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.7152742743492126,
      "learning_rate": 0.0005807794770951051,
      "loss": 2.2065,
      "step": 26395
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.7978829145431519,
      "learning_rate": 0.0005807780364472347,
      "loss": 2.4024,
      "step": 26396
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6682263612747192,
      "learning_rate": 0.0005807765957471623,
      "loss": 2.2373,
      "step": 26397
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6399916410446167,
      "learning_rate": 0.0005807751549948884,
      "loss": 2.1808,
      "step": 26398
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.8283401727676392,
      "learning_rate": 0.000580773714190413,
      "loss": 1.8444,
      "step": 26399
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6178953647613525,
      "learning_rate": 0.0005807722733337364,
      "loss": 2.1758,
      "step": 26400
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6224835515022278,
      "learning_rate": 0.000580770832424859,
      "loss": 2.2748,
      "step": 26401
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.702846109867096,
      "learning_rate": 0.000580769391463781,
      "loss": 2.31,
      "step": 26402
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.669105052947998,
      "learning_rate": 0.0005807679504505026,
      "loss": 2.4176,
      "step": 26403
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.5934601426124573,
      "learning_rate": 0.0005807665093850243,
      "loss": 2.4839,
      "step": 26404
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6344919204711914,
      "learning_rate": 0.000580765068267346,
      "loss": 2.3931,
      "step": 26405
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6333638429641724,
      "learning_rate": 0.0005807636270974683,
      "loss": 2.4098,
      "step": 26406
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6124464273452759,
      "learning_rate": 0.0005807621858753913,
      "loss": 2.2961,
      "step": 26407
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.644253134727478,
      "learning_rate": 0.0005807607446011154,
      "loss": 2.3849,
      "step": 26408
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6400184035301208,
      "learning_rate": 0.0005807593032746407,
      "loss": 2.3801,
      "step": 26409
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6004652380943298,
      "learning_rate": 0.0005807578618959676,
      "loss": 2.1639,
      "step": 26410
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6422223448753357,
      "learning_rate": 0.0005807564204650962,
      "loss": 2.4621,
      "step": 26411
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6734799742698669,
      "learning_rate": 0.0005807549789820269,
      "loss": 2.5253,
      "step": 26412
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.5619546175003052,
      "learning_rate": 0.0005807535374467601,
      "loss": 2.1854,
      "step": 26413
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6975290179252625,
      "learning_rate": 0.0005807520958592958,
      "loss": 2.4864,
      "step": 26414
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6297325491905212,
      "learning_rate": 0.0005807506542196344,
      "loss": 2.2944,
      "step": 26415
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.5665217041969299,
      "learning_rate": 0.0005807492125277762,
      "loss": 2.1092,
      "step": 26416
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6260098814964294,
      "learning_rate": 0.0005807477707837214,
      "loss": 2.4457,
      "step": 26417
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6216356754302979,
      "learning_rate": 0.0005807463289874702,
      "loss": 2.4045,
      "step": 26418
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6012833714485168,
      "learning_rate": 0.0005807448871390231,
      "loss": 2.2753,
      "step": 26419
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.7116245031356812,
      "learning_rate": 0.0005807434452383801,
      "loss": 2.1167,
      "step": 26420
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6235012412071228,
      "learning_rate": 0.0005807420032855416,
      "loss": 2.4938,
      "step": 26421
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.7323325872421265,
      "learning_rate": 0.0005807405612805079,
      "loss": 2.2932,
      "step": 26422
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.5887838006019592,
      "learning_rate": 0.0005807391192232793,
      "loss": 2.2205,
      "step": 26423
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6238933801651001,
      "learning_rate": 0.000580737677113856,
      "loss": 2.1607,
      "step": 26424
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6334389448165894,
      "learning_rate": 0.0005807362349522381,
      "loss": 2.2106,
      "step": 26425
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6492305994033813,
      "learning_rate": 0.0005807347927384262,
      "loss": 2.365,
      "step": 26426
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.593282163143158,
      "learning_rate": 0.0005807333504724203,
      "loss": 2.1835,
      "step": 26427
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.5544028878211975,
      "learning_rate": 0.0005807319081542208,
      "loss": 2.1458,
      "step": 26428
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.7144896984100342,
      "learning_rate": 0.000580730465783828,
      "loss": 2.4541,
      "step": 26429
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6463798880577087,
      "learning_rate": 0.000580729023361242,
      "loss": 2.2925,
      "step": 26430
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6310654878616333,
      "learning_rate": 0.0005807275808864633,
      "loss": 2.1765,
      "step": 26431
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.7088571786880493,
      "learning_rate": 0.000580726138359492,
      "loss": 2.3605,
      "step": 26432
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.5891562700271606,
      "learning_rate": 0.0005807246957803283,
      "loss": 2.3323,
      "step": 26433
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.590650200843811,
      "learning_rate": 0.0005807232531489727,
      "loss": 2.2721,
      "step": 26434
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6838497519493103,
      "learning_rate": 0.0005807218104654254,
      "loss": 2.2463,
      "step": 26435
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6191612482070923,
      "learning_rate": 0.0005807203677296865,
      "loss": 1.9679,
      "step": 26436
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6007161736488342,
      "learning_rate": 0.0005807189249417565,
      "loss": 2.3196,
      "step": 26437
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6578047275543213,
      "learning_rate": 0.0005807174821016353,
      "loss": 2.4446,
      "step": 26438
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.8062171339988708,
      "learning_rate": 0.0005807160392093237,
      "loss": 2.4524,
      "step": 26439
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6339791417121887,
      "learning_rate": 0.0005807145962648216,
      "loss": 2.2962,
      "step": 26440
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6611512303352356,
      "learning_rate": 0.0005807131532681293,
      "loss": 2.5238,
      "step": 26441
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6412623524665833,
      "learning_rate": 0.0005807117102192472,
      "loss": 2.0849,
      "step": 26442
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6096084713935852,
      "learning_rate": 0.0005807102671181755,
      "loss": 2.2887,
      "step": 26443
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6014310121536255,
      "learning_rate": 0.0005807088239649144,
      "loss": 2.4857,
      "step": 26444
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6377179622650146,
      "learning_rate": 0.0005807073807594642,
      "loss": 2.273,
      "step": 26445
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6612652540206909,
      "learning_rate": 0.0005807059375018252,
      "loss": 2.3827,
      "step": 26446
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6356362104415894,
      "learning_rate": 0.0005807044941919978,
      "loss": 2.3005,
      "step": 26447
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.7324222922325134,
      "learning_rate": 0.0005807030508299821,
      "loss": 2.2389,
      "step": 26448
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.7665376663208008,
      "learning_rate": 0.0005807016074157782,
      "loss": 2.355,
      "step": 26449
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6670366525650024,
      "learning_rate": 0.0005807001639493868,
      "loss": 2.6538,
      "step": 26450
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.7693137526512146,
      "learning_rate": 0.0005806987204308078,
      "loss": 2.078,
      "step": 26451
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6683257818222046,
      "learning_rate": 0.0005806972768600418,
      "loss": 2.3128,
      "step": 26452
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.690942645072937,
      "learning_rate": 0.0005806958332370888,
      "loss": 2.272,
      "step": 26453
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6819694638252258,
      "learning_rate": 0.0005806943895619491,
      "loss": 2.3011,
      "step": 26454
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.8123951554298401,
      "learning_rate": 0.000580692945834623,
      "loss": 2.3359,
      "step": 26455
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.5858708620071411,
      "learning_rate": 0.0005806915020551109,
      "loss": 2.2681,
      "step": 26456
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6572409868240356,
      "learning_rate": 0.0005806900582234128,
      "loss": 2.2843,
      "step": 26457
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.9042646884918213,
      "learning_rate": 0.0005806886143395291,
      "loss": 2.1007,
      "step": 26458
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6276865005493164,
      "learning_rate": 0.0005806871704034602,
      "loss": 2.5135,
      "step": 26459
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.8993521928787231,
      "learning_rate": 0.0005806857264152063,
      "loss": 2.3905,
      "step": 26460
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.8777164220809937,
      "learning_rate": 0.0005806842823747675,
      "loss": 2.247,
      "step": 26461
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6048029661178589,
      "learning_rate": 0.0005806828382821442,
      "loss": 2.4442,
      "step": 26462
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.7624896168708801,
      "learning_rate": 0.0005806813941373367,
      "loss": 2.2319,
      "step": 26463
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.7287706136703491,
      "learning_rate": 0.0005806799499403452,
      "loss": 2.2474,
      "step": 26464
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6517140865325928,
      "learning_rate": 0.0005806785056911701,
      "loss": 2.471,
      "step": 26465
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.7629269957542419,
      "learning_rate": 0.0005806770613898115,
      "loss": 2.3115,
      "step": 26466
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.7496461272239685,
      "learning_rate": 0.0005806756170362697,
      "loss": 2.3175,
      "step": 26467
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6474292874336243,
      "learning_rate": 0.000580674172630545,
      "loss": 2.3836,
      "step": 26468
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6023052930831909,
      "learning_rate": 0.0005806727281726377,
      "loss": 2.419,
      "step": 26469
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6524572372436523,
      "learning_rate": 0.0005806712836625481,
      "loss": 2.286,
      "step": 26470
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6862829923629761,
      "learning_rate": 0.0005806698391002764,
      "loss": 2.0678,
      "step": 26471
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6719675660133362,
      "learning_rate": 0.0005806683944858228,
      "loss": 2.2249,
      "step": 26472
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6425113081932068,
      "learning_rate": 0.0005806669498191877,
      "loss": 2.0673,
      "step": 26473
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.5901926755905151,
      "learning_rate": 0.0005806655051003713,
      "loss": 2.1805,
      "step": 26474
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6266688704490662,
      "learning_rate": 0.0005806640603293738,
      "loss": 2.2528,
      "step": 26475
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6548197269439697,
      "learning_rate": 0.0005806626155061957,
      "loss": 2.1635,
      "step": 26476
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.660407543182373,
      "learning_rate": 0.0005806611706308372,
      "loss": 2.3253,
      "step": 26477
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.5857254266738892,
      "learning_rate": 0.0005806597257032983,
      "loss": 2.1704,
      "step": 26478
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6927403807640076,
      "learning_rate": 0.0005806582807235795,
      "loss": 2.6193,
      "step": 26479
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.9851593375205994,
      "learning_rate": 0.000580656835691681,
      "loss": 2.4087,
      "step": 26480
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6840090155601501,
      "learning_rate": 0.0005806553906076031,
      "loss": 2.3358,
      "step": 26481
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6648579239845276,
      "learning_rate": 0.0005806539454713462,
      "loss": 2.3883,
      "step": 26482
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6066513657569885,
      "learning_rate": 0.0005806525002829103,
      "loss": 2.2722,
      "step": 26483
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6340793967247009,
      "learning_rate": 0.0005806510550422959,
      "loss": 2.2507,
      "step": 26484
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6021133661270142,
      "learning_rate": 0.0005806496097495031,
      "loss": 2.2568,
      "step": 26485
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6615431308746338,
      "learning_rate": 0.0005806481644045322,
      "loss": 2.3831,
      "step": 26486
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.613429844379425,
      "learning_rate": 0.0005806467190073837,
      "loss": 2.4786,
      "step": 26487
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6320881843566895,
      "learning_rate": 0.0005806452735580575,
      "loss": 2.1689,
      "step": 26488
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6595976948738098,
      "learning_rate": 0.000580643828056554,
      "loss": 2.3115,
      "step": 26489
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.612339437007904,
      "learning_rate": 0.0005806423825028736,
      "loss": 2.1561,
      "step": 26490
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.7055020332336426,
      "learning_rate": 0.0005806409368970166,
      "loss": 2.3543,
      "step": 26491
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6492602229118347,
      "learning_rate": 0.0005806394912389828,
      "loss": 2.401,
      "step": 26492
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6333531737327576,
      "learning_rate": 0.0005806380455287732,
      "loss": 2.3214,
      "step": 26493
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.643509566783905,
      "learning_rate": 0.0005806365997663875,
      "loss": 2.1324,
      "step": 26494
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6675848960876465,
      "learning_rate": 0.0005806351539518262,
      "loss": 2.3258,
      "step": 26495
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.63984614610672,
      "learning_rate": 0.0005806337080850896,
      "loss": 2.2748,
      "step": 26496
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6082830429077148,
      "learning_rate": 0.0005806322621661779,
      "loss": 2.3255,
      "step": 26497
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.7079832553863525,
      "learning_rate": 0.0005806308161950913,
      "loss": 2.4101,
      "step": 26498
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6394177675247192,
      "learning_rate": 0.00058062937017183,
      "loss": 2.1829,
      "step": 26499
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.5863664746284485,
      "learning_rate": 0.0005806279240963946,
      "loss": 2.5085,
      "step": 26500
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6291571259498596,
      "learning_rate": 0.0005806264779687851,
      "loss": 2.3927,
      "step": 26501
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.5966008305549622,
      "learning_rate": 0.0005806250317890019,
      "loss": 2.3484,
      "step": 26502
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.7176985144615173,
      "learning_rate": 0.0005806235855570451,
      "loss": 2.3903,
      "step": 26503
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6459612846374512,
      "learning_rate": 0.0005806221392729151,
      "loss": 2.6128,
      "step": 26504
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6605072617530823,
      "learning_rate": 0.0005806206929366122,
      "loss": 2.562,
      "step": 26505
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.5963929891586304,
      "learning_rate": 0.0005806192465481366,
      "loss": 2.2638,
      "step": 26506
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.5968031883239746,
      "learning_rate": 0.0005806178001074886,
      "loss": 2.2697,
      "step": 26507
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6459422707557678,
      "learning_rate": 0.0005806163536146684,
      "loss": 2.2601,
      "step": 26508
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6734426021575928,
      "learning_rate": 0.0005806149070696763,
      "loss": 2.2456,
      "step": 26509
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6571781635284424,
      "learning_rate": 0.0005806134604725127,
      "loss": 2.2323,
      "step": 26510
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.9494508504867554,
      "learning_rate": 0.0005806120138231777,
      "loss": 2.3489,
      "step": 26511
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.7736528515815735,
      "learning_rate": 0.0005806105671216716,
      "loss": 2.2309,
      "step": 26512
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.612199604511261,
      "learning_rate": 0.0005806091203679948,
      "loss": 2.3115,
      "step": 26513
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6311692595481873,
      "learning_rate": 0.0005806076735621474,
      "loss": 2.2633,
      "step": 26514
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.7319746613502502,
      "learning_rate": 0.0005806062267041298,
      "loss": 2.3553,
      "step": 26515
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6188629269599915,
      "learning_rate": 0.0005806047797939421,
      "loss": 2.2132,
      "step": 26516
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6730851531028748,
      "learning_rate": 0.0005806033328315846,
      "loss": 2.251,
      "step": 26517
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6263257265090942,
      "learning_rate": 0.0005806018858170579,
      "loss": 2.2459,
      "step": 26518
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6514079570770264,
      "learning_rate": 0.0005806004387503619,
      "loss": 2.4336,
      "step": 26519
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6311110258102417,
      "learning_rate": 0.0005805989916314968,
      "loss": 2.0361,
      "step": 26520
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.5981844663619995,
      "learning_rate": 0.0005805975444604633,
      "loss": 2.2412,
      "step": 26521
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6915512681007385,
      "learning_rate": 0.0005805960972372614,
      "loss": 2.3801,
      "step": 26522
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.621082603931427,
      "learning_rate": 0.0005805946499618913,
      "loss": 2.3439,
      "step": 26523
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.5931176543235779,
      "learning_rate": 0.0005805932026343533,
      "loss": 2.3032,
      "step": 26524
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6004355549812317,
      "learning_rate": 0.0005805917552546478,
      "loss": 2.2749,
      "step": 26525
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6949558258056641,
      "learning_rate": 0.000580590307822775,
      "loss": 2.3872,
      "step": 26526
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6415287256240845,
      "learning_rate": 0.0005805888603387352,
      "loss": 2.2072,
      "step": 26527
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6351245045661926,
      "learning_rate": 0.0005805874128025285,
      "loss": 2.3668,
      "step": 26528
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6713667511940002,
      "learning_rate": 0.0005805859652141556,
      "loss": 2.4171,
      "step": 26529
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6695696115493774,
      "learning_rate": 0.0005805845175736162,
      "loss": 2.4978,
      "step": 26530
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.5826697945594788,
      "learning_rate": 0.0005805830698809109,
      "loss": 2.3417,
      "step": 26531
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6199624538421631,
      "learning_rate": 0.0005805816221360399,
      "loss": 2.3193,
      "step": 26532
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.7379801869392395,
      "learning_rate": 0.0005805801743390035,
      "loss": 2.3743,
      "step": 26533
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6248971819877625,
      "learning_rate": 0.000580578726489802,
      "loss": 2.1571,
      "step": 26534
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6292338967323303,
      "learning_rate": 0.0005805772785884356,
      "loss": 2.1369,
      "step": 26535
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.7325088381767273,
      "learning_rate": 0.0005805758306349046,
      "loss": 2.2744,
      "step": 26536
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6380749344825745,
      "learning_rate": 0.0005805743826292092,
      "loss": 2.2972,
      "step": 26537
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6474719643592834,
      "learning_rate": 0.0005805729345713497,
      "loss": 2.4063,
      "step": 26538
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.7700640559196472,
      "learning_rate": 0.0005805714864613264,
      "loss": 2.4434,
      "step": 26539
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.59430992603302,
      "learning_rate": 0.0005805700382991397,
      "loss": 2.1963,
      "step": 26540
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.652503490447998,
      "learning_rate": 0.0005805685900847896,
      "loss": 2.5178,
      "step": 26541
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6548728346824646,
      "learning_rate": 0.0005805671418182766,
      "loss": 2.2537,
      "step": 26542
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6814064383506775,
      "learning_rate": 0.0005805656934996007,
      "loss": 2.4542,
      "step": 26543
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6679558753967285,
      "learning_rate": 0.0005805642451287624,
      "loss": 2.1654,
      "step": 26544
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.5601281523704529,
      "learning_rate": 0.0005805627967057621,
      "loss": 2.2943,
      "step": 26545
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6329568028450012,
      "learning_rate": 0.0005805613482305997,
      "loss": 2.3092,
      "step": 26546
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6587190628051758,
      "learning_rate": 0.0005805598997032757,
      "loss": 2.4553,
      "step": 26547
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6180723309516907,
      "learning_rate": 0.0005805584511237902,
      "loss": 2.3579,
      "step": 26548
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6278331279754639,
      "learning_rate": 0.0005805570024921439,
      "loss": 2.2925,
      "step": 26549
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.5663402080535889,
      "learning_rate": 0.0005805555538083365,
      "loss": 2.4066,
      "step": 26550
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6152099370956421,
      "learning_rate": 0.0005805541050723686,
      "loss": 2.1973,
      "step": 26551
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6417143940925598,
      "learning_rate": 0.0005805526562842403,
      "loss": 2.4336,
      "step": 26552
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6757468581199646,
      "learning_rate": 0.000580551207443952,
      "loss": 2.2997,
      "step": 26553
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.5970455408096313,
      "learning_rate": 0.0005805497585515041,
      "loss": 2.3083,
      "step": 26554
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6717846393585205,
      "learning_rate": 0.0005805483096068965,
      "loss": 2.3028,
      "step": 26555
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6303973197937012,
      "learning_rate": 0.0005805468606101299,
      "loss": 2.1308,
      "step": 26556
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.585127592086792,
      "learning_rate": 0.0005805454115612042,
      "loss": 2.1966,
      "step": 26557
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6135671734809875,
      "learning_rate": 0.0005805439624601197,
      "loss": 2.3672,
      "step": 26558
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.8794673085212708,
      "learning_rate": 0.000580542513306877,
      "loss": 2.4537,
      "step": 26559
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6702775359153748,
      "learning_rate": 0.0005805410641014759,
      "loss": 2.4526,
      "step": 26560
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.7001023292541504,
      "learning_rate": 0.0005805396148439172,
      "loss": 2.5703,
      "step": 26561
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6231720447540283,
      "learning_rate": 0.0005805381655342007,
      "loss": 2.2052,
      "step": 26562
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6259322166442871,
      "learning_rate": 0.0005805367161723269,
      "loss": 2.2163,
      "step": 26563
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6338571906089783,
      "learning_rate": 0.0005805352667582961,
      "loss": 2.099,
      "step": 26564
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6824115514755249,
      "learning_rate": 0.0005805338172921084,
      "loss": 2.2702,
      "step": 26565
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6274542212486267,
      "learning_rate": 0.0005805323677737643,
      "loss": 2.2384,
      "step": 26566
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.676769495010376,
      "learning_rate": 0.0005805309182032639,
      "loss": 2.5868,
      "step": 26567
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.826265275478363,
      "learning_rate": 0.0005805294685806074,
      "loss": 2.1669,
      "step": 26568
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.643924355506897,
      "learning_rate": 0.0005805280189057952,
      "loss": 2.2079,
      "step": 26569
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.586574137210846,
      "learning_rate": 0.0005805265691788275,
      "loss": 2.5235,
      "step": 26570
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6600983142852783,
      "learning_rate": 0.0005805251193997048,
      "loss": 2.028,
      "step": 26571
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6435387134552002,
      "learning_rate": 0.0005805236695684271,
      "loss": 2.3174,
      "step": 26572
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6408746242523193,
      "learning_rate": 0.0005805222196849947,
      "loss": 2.2037,
      "step": 26573
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.5682553052902222,
      "learning_rate": 0.0005805207697494079,
      "loss": 2.4442,
      "step": 26574
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6933727264404297,
      "learning_rate": 0.0005805193197616671,
      "loss": 2.3716,
      "step": 26575
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.642093300819397,
      "learning_rate": 0.0005805178697217724,
      "loss": 2.4421,
      "step": 26576
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.7377443909645081,
      "learning_rate": 0.0005805164196297242,
      "loss": 2.1813,
      "step": 26577
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6978496313095093,
      "learning_rate": 0.0005805149694855225,
      "loss": 2.3861,
      "step": 26578
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6748852133750916,
      "learning_rate": 0.000580513519289168,
      "loss": 2.2064,
      "step": 26579
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6030442714691162,
      "learning_rate": 0.0005805120690406607,
      "loss": 2.3044,
      "step": 26580
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6811394691467285,
      "learning_rate": 0.0005805106187400009,
      "loss": 2.2922,
      "step": 26581
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.625617265701294,
      "learning_rate": 0.0005805091683871888,
      "loss": 2.2786,
      "step": 26582
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.646362841129303,
      "learning_rate": 0.0005805077179822249,
      "loss": 2.0444,
      "step": 26583
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.7541010975837708,
      "learning_rate": 0.0005805062675251091,
      "loss": 2.3832,
      "step": 26584
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6558346152305603,
      "learning_rate": 0.0005805048170158422,
      "loss": 2.352,
      "step": 26585
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6480472087860107,
      "learning_rate": 0.0005805033664544239,
      "loss": 2.4297,
      "step": 26586
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6537106037139893,
      "learning_rate": 0.0005805019158408548,
      "loss": 2.3767,
      "step": 26587
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6752639412879944,
      "learning_rate": 0.000580500465175135,
      "loss": 2.0953,
      "step": 26588
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6111146211624146,
      "learning_rate": 0.0005804990144572651,
      "loss": 2.377,
      "step": 26589
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6513031721115112,
      "learning_rate": 0.0005804975636872449,
      "loss": 1.9974,
      "step": 26590
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6659631729125977,
      "learning_rate": 0.000580496112865075,
      "loss": 2.3416,
      "step": 26591
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6422410607337952,
      "learning_rate": 0.0005804946619907557,
      "loss": 2.3514,
      "step": 26592
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6196425557136536,
      "learning_rate": 0.000580493211064287,
      "loss": 2.5259,
      "step": 26593
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6659544706344604,
      "learning_rate": 0.0005804917600856694,
      "loss": 2.3945,
      "step": 26594
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.8373249769210815,
      "learning_rate": 0.000580490309054903,
      "loss": 2.3437,
      "step": 26595
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6449874043464661,
      "learning_rate": 0.0005804888579719883,
      "loss": 2.2339,
      "step": 26596
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6453577876091003,
      "learning_rate": 0.0005804874068369253,
      "loss": 2.1441,
      "step": 26597
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6918759942054749,
      "learning_rate": 0.0005804859556497145,
      "loss": 2.3495,
      "step": 26598
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.622624933719635,
      "learning_rate": 0.000580484504410356,
      "loss": 2.1744,
      "step": 26599
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.7253162264823914,
      "learning_rate": 0.0005804830531188502,
      "loss": 2.2301,
      "step": 26600
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.7645441889762878,
      "learning_rate": 0.0005804816017751972,
      "loss": 2.2074,
      "step": 26601
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6938669681549072,
      "learning_rate": 0.0005804801503793973,
      "loss": 2.4085,
      "step": 26602
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.8875617980957031,
      "learning_rate": 0.000580478698931451,
      "loss": 2.2709,
      "step": 26603
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6715825200080872,
      "learning_rate": 0.0005804772474313583,
      "loss": 2.3319,
      "step": 26604
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6240010857582092,
      "learning_rate": 0.0005804757958791198,
      "loss": 2.1495,
      "step": 26605
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.7784007787704468,
      "learning_rate": 0.0005804743442747353,
      "loss": 2.2217,
      "step": 26606
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6262449622154236,
      "learning_rate": 0.0005804728926182054,
      "loss": 2.3253,
      "step": 26607
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.0601434707641602,
      "learning_rate": 0.0005804714409095304,
      "loss": 2.3577,
      "step": 26608
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6539442539215088,
      "learning_rate": 0.0005804699891487104,
      "loss": 2.1846,
      "step": 26609
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6397750377655029,
      "learning_rate": 0.0005804685373357458,
      "loss": 2.1162,
      "step": 26610
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6199907064437866,
      "learning_rate": 0.0005804670854706367,
      "loss": 2.1466,
      "step": 26611
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.5972431898117065,
      "learning_rate": 0.0005804656335533835,
      "loss": 2.2249,
      "step": 26612
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6259147524833679,
      "learning_rate": 0.0005804641815839865,
      "loss": 2.3319,
      "step": 26613
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.753255307674408,
      "learning_rate": 0.0005804627295624458,
      "loss": 2.5184,
      "step": 26614
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6462975144386292,
      "learning_rate": 0.0005804612774887619,
      "loss": 2.3221,
      "step": 26615
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6409288644790649,
      "learning_rate": 0.0005804598253629349,
      "loss": 2.4289,
      "step": 26616
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6857232451438904,
      "learning_rate": 0.0005804583731849652,
      "loss": 2.2981,
      "step": 26617
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.7071089744567871,
      "learning_rate": 0.0005804569209548529,
      "loss": 2.2975,
      "step": 26618
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.5818012952804565,
      "learning_rate": 0.0005804554686725984,
      "loss": 2.3187,
      "step": 26619
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6941638588905334,
      "learning_rate": 0.000580454016338202,
      "loss": 2.2759,
      "step": 26620
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6662298440933228,
      "learning_rate": 0.0005804525639516638,
      "loss": 2.3713,
      "step": 26621
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.7134122252464294,
      "learning_rate": 0.0005804511115129841,
      "loss": 2.3227,
      "step": 26622
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6815289258956909,
      "learning_rate": 0.0005804496590221634,
      "loss": 2.2501,
      "step": 26623
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6462666392326355,
      "learning_rate": 0.0005804482064792018,
      "loss": 2.4626,
      "step": 26624
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.8478708267211914,
      "learning_rate": 0.0005804467538840995,
      "loss": 2.2407,
      "step": 26625
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.7615612149238586,
      "learning_rate": 0.0005804453012368569,
      "loss": 2.6237,
      "step": 26626
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.5886037349700928,
      "learning_rate": 0.0005804438485374743,
      "loss": 2.2986,
      "step": 26627
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.7025833129882812,
      "learning_rate": 0.0005804423957859518,
      "loss": 2.342,
      "step": 26628
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6637660264968872,
      "learning_rate": 0.0005804409429822898,
      "loss": 2.4214,
      "step": 26629
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.644207239151001,
      "learning_rate": 0.0005804394901264885,
      "loss": 2.2538,
      "step": 26630
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6437115669250488,
      "learning_rate": 0.0005804380372185482,
      "loss": 2.147,
      "step": 26631
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.9490353465080261,
      "learning_rate": 0.0005804365842584693,
      "loss": 2.388,
      "step": 26632
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.7087036967277527,
      "learning_rate": 0.0005804351312462518,
      "loss": 2.3041,
      "step": 26633
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.7153847217559814,
      "learning_rate": 0.0005804336781818962,
      "loss": 2.4487,
      "step": 26634
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.8476704955101013,
      "learning_rate": 0.0005804322250654026,
      "loss": 2.1187,
      "step": 26635
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.7663556933403015,
      "learning_rate": 0.0005804307718967715,
      "loss": 2.1393,
      "step": 26636
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.666782796382904,
      "learning_rate": 0.0005804293186760029,
      "loss": 2.3021,
      "step": 26637
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.7121871709823608,
      "learning_rate": 0.0005804278654030972,
      "loss": 2.3568,
      "step": 26638
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6681528687477112,
      "learning_rate": 0.0005804264120780547,
      "loss": 2.722,
      "step": 26639
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.7002843618392944,
      "learning_rate": 0.0005804249587008756,
      "loss": 2.4422,
      "step": 26640
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6882327198982239,
      "learning_rate": 0.0005804235052715603,
      "loss": 2.2086,
      "step": 26641
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6706753969192505,
      "learning_rate": 0.0005804220517901088,
      "loss": 2.1412,
      "step": 26642
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6248556971549988,
      "learning_rate": 0.0005804205982565218,
      "loss": 2.1687,
      "step": 26643
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.7382559776306152,
      "learning_rate": 0.0005804191446707992,
      "loss": 2.4804,
      "step": 26644
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6583164930343628,
      "learning_rate": 0.0005804176910329413,
      "loss": 2.3128,
      "step": 26645
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6187151670455933,
      "learning_rate": 0.0005804162373429486,
      "loss": 2.3144,
      "step": 26646
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.8004915118217468,
      "learning_rate": 0.0005804147836008211,
      "loss": 2.1379,
      "step": 26647
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6067808270454407,
      "learning_rate": 0.0005804133298065592,
      "loss": 2.3185,
      "step": 26648
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6863384246826172,
      "learning_rate": 0.0005804118759601632,
      "loss": 2.4345,
      "step": 26649
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.7816964387893677,
      "learning_rate": 0.0005804104220616334,
      "loss": 2.4115,
      "step": 26650
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6914058923721313,
      "learning_rate": 0.00058040896811097,
      "loss": 2.3195,
      "step": 26651
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6123889684677124,
      "learning_rate": 0.0005804075141081733,
      "loss": 2.4911,
      "step": 26652
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6329941153526306,
      "learning_rate": 0.0005804060600532434,
      "loss": 2.2627,
      "step": 26653
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6462773084640503,
      "learning_rate": 0.0005804046059461809,
      "loss": 2.5053,
      "step": 26654
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.7216110825538635,
      "learning_rate": 0.0005804031517869858,
      "loss": 2.1761,
      "step": 26655
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6411988735198975,
      "learning_rate": 0.0005804016975756585,
      "loss": 2.3193,
      "step": 26656
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.5836421251296997,
      "learning_rate": 0.0005804002433121992,
      "loss": 2.3,
      "step": 26657
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6217087507247925,
      "learning_rate": 0.0005803987889966082,
      "loss": 2.5462,
      "step": 26658
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.7326694130897522,
      "learning_rate": 0.0005803973346288858,
      "loss": 2.4186,
      "step": 26659
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.657335638999939,
      "learning_rate": 0.0005803958802090322,
      "loss": 2.3454,
      "step": 26660
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.7327207922935486,
      "learning_rate": 0.0005803944257370476,
      "loss": 2.1823,
      "step": 26661
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6068090796470642,
      "learning_rate": 0.0005803929712129327,
      "loss": 2.4366,
      "step": 26662
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6064268350601196,
      "learning_rate": 0.0005803915166366872,
      "loss": 2.2394,
      "step": 26663
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6822835206985474,
      "learning_rate": 0.0005803900620083118,
      "loss": 2.1858,
      "step": 26664
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6574994921684265,
      "learning_rate": 0.0005803886073278064,
      "loss": 2.2672,
      "step": 26665
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6566165685653687,
      "learning_rate": 0.0005803871525951715,
      "loss": 2.2217,
      "step": 26666
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6709312796592712,
      "learning_rate": 0.0005803856978104074,
      "loss": 2.3613,
      "step": 26667
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6308204531669617,
      "learning_rate": 0.0005803842429735143,
      "loss": 2.2507,
      "step": 26668
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6365288496017456,
      "learning_rate": 0.0005803827880844926,
      "loss": 2.31,
      "step": 26669
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6118119359016418,
      "learning_rate": 0.0005803813331433423,
      "loss": 2.3515,
      "step": 26670
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6125001311302185,
      "learning_rate": 0.0005803798781500638,
      "loss": 2.2279,
      "step": 26671
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6529813408851624,
      "learning_rate": 0.0005803784231046574,
      "loss": 2.3156,
      "step": 26672
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6682192087173462,
      "learning_rate": 0.0005803769680071234,
      "loss": 2.4664,
      "step": 26673
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6495604515075684,
      "learning_rate": 0.0005803755128574621,
      "loss": 2.4496,
      "step": 26674
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.639477014541626,
      "learning_rate": 0.0005803740576556735,
      "loss": 2.4513,
      "step": 26675
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6856020092964172,
      "learning_rate": 0.0005803726024017581,
      "loss": 2.4949,
      "step": 26676
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6186405420303345,
      "learning_rate": 0.0005803711470957163,
      "loss": 2.4363,
      "step": 26677
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.5845228433609009,
      "learning_rate": 0.0005803696917375482,
      "loss": 2.2535,
      "step": 26678
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6277633905410767,
      "learning_rate": 0.000580368236327254,
      "loss": 2.245,
      "step": 26679
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.5731624960899353,
      "learning_rate": 0.0005803667808648341,
      "loss": 2.5442,
      "step": 26680
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6517151594161987,
      "learning_rate": 0.0005803653253502887,
      "loss": 2.3989,
      "step": 26681
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6105311512947083,
      "learning_rate": 0.000580363869783618,
      "loss": 2.4794,
      "step": 26682
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6152835488319397,
      "learning_rate": 0.0005803624141648225,
      "loss": 2.3318,
      "step": 26683
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6591892242431641,
      "learning_rate": 0.0005803609584939023,
      "loss": 2.4908,
      "step": 26684
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6038860082626343,
      "learning_rate": 0.0005803595027708577,
      "loss": 2.4525,
      "step": 26685
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.5932938456535339,
      "learning_rate": 0.000580358046995689,
      "loss": 2.3499,
      "step": 26686
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6258280277252197,
      "learning_rate": 0.0005803565911683964,
      "loss": 2.2889,
      "step": 26687
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6606150269508362,
      "learning_rate": 0.0005803551352889803,
      "loss": 2.5467,
      "step": 26688
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6346251368522644,
      "learning_rate": 0.0005803536793574408,
      "loss": 2.3607,
      "step": 26689
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6414647698402405,
      "learning_rate": 0.0005803522233737781,
      "loss": 2.6076,
      "step": 26690
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6056024432182312,
      "learning_rate": 0.0005803507673379929,
      "loss": 2.2103,
      "step": 26691
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6678501963615417,
      "learning_rate": 0.0005803493112500851,
      "loss": 2.4922,
      "step": 26692
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6062985062599182,
      "learning_rate": 0.0005803478551100551,
      "loss": 2.4517,
      "step": 26693
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6348349452018738,
      "learning_rate": 0.0005803463989179031,
      "loss": 2.4314,
      "step": 26694
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6333866119384766,
      "learning_rate": 0.0005803449426736295,
      "loss": 2.4063,
      "step": 26695
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6336199045181274,
      "learning_rate": 0.0005803434863772344,
      "loss": 2.2966,
      "step": 26696
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.5928572416305542,
      "learning_rate": 0.0005803420300287182,
      "loss": 2.3644,
      "step": 26697
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.672849714756012,
      "learning_rate": 0.0005803405736280811,
      "loss": 2.3026,
      "step": 26698
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6309106349945068,
      "learning_rate": 0.0005803391171753234,
      "loss": 2.3945,
      "step": 26699
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6201989650726318,
      "learning_rate": 0.0005803376606704453,
      "loss": 2.5677,
      "step": 26700
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.63034588098526,
      "learning_rate": 0.0005803362041134473,
      "loss": 2.1602,
      "step": 26701
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6157155632972717,
      "learning_rate": 0.0005803347475043293,
      "loss": 2.3032,
      "step": 26702
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6559233069419861,
      "learning_rate": 0.000580333290843092,
      "loss": 2.4103,
      "step": 26703
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6250582337379456,
      "learning_rate": 0.0005803318341297352,
      "loss": 2.317,
      "step": 26704
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6511046290397644,
      "learning_rate": 0.0005803303773642596,
      "loss": 2.2067,
      "step": 26705
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.7755621671676636,
      "learning_rate": 0.0005803289205466652,
      "loss": 2.3848,
      "step": 26706
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6504067182540894,
      "learning_rate": 0.0005803274636769525,
      "loss": 2.3416,
      "step": 26707
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6191875338554382,
      "learning_rate": 0.0005803260067551215,
      "loss": 2.3566,
      "step": 26708
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6232089400291443,
      "learning_rate": 0.0005803245497811726,
      "loss": 2.4812,
      "step": 26709
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6573009490966797,
      "learning_rate": 0.0005803230927551061,
      "loss": 2.3014,
      "step": 26710
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6648749709129333,
      "learning_rate": 0.0005803216356769223,
      "loss": 2.4438,
      "step": 26711
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6523008346557617,
      "learning_rate": 0.0005803201785466214,
      "loss": 2.461,
      "step": 26712
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6717752814292908,
      "learning_rate": 0.0005803187213642036,
      "loss": 2.3251,
      "step": 26713
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.7137899398803711,
      "learning_rate": 0.0005803172641296694,
      "loss": 2.3868,
      "step": 26714
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6799297332763672,
      "learning_rate": 0.0005803158068430189,
      "loss": 2.3586,
      "step": 26715
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.7447639107704163,
      "learning_rate": 0.0005803143495042523,
      "loss": 2.1567,
      "step": 26716
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.674012303352356,
      "learning_rate": 0.0005803128921133701,
      "loss": 2.2671,
      "step": 26717
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.627146303653717,
      "learning_rate": 0.0005803114346703723,
      "loss": 2.4026,
      "step": 26718
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.7112419605255127,
      "learning_rate": 0.0005803099771752594,
      "loss": 2.3941,
      "step": 26719
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.660072386264801,
      "learning_rate": 0.0005803085196280315,
      "loss": 2.1673,
      "step": 26720
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.7387009263038635,
      "learning_rate": 0.000580307062028689,
      "loss": 2.1956,
      "step": 26721
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.650478720664978,
      "learning_rate": 0.0005803056043772322,
      "loss": 2.5773,
      "step": 26722
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6487130522727966,
      "learning_rate": 0.0005803041466736612,
      "loss": 2.3392,
      "step": 26723
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6046154499053955,
      "learning_rate": 0.0005803026889179765,
      "loss": 2.1626,
      "step": 26724
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6547984480857849,
      "learning_rate": 0.0005803012311101782,
      "loss": 2.4317,
      "step": 26725
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.8078051209449768,
      "learning_rate": 0.0005802997732502665,
      "loss": 2.4856,
      "step": 26726
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6541172862052917,
      "learning_rate": 0.0005802983153382419,
      "loss": 2.536,
      "step": 26727
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6383011341094971,
      "learning_rate": 0.0005802968573741044,
      "loss": 2.2247,
      "step": 26728
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.7026932239532471,
      "learning_rate": 0.0005802953993578545,
      "loss": 2.552,
      "step": 26729
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6564397811889648,
      "learning_rate": 0.0005802939412894924,
      "loss": 2.1384,
      "step": 26730
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6559231877326965,
      "learning_rate": 0.0005802924831690185,
      "loss": 2.2337,
      "step": 26731
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6684702634811401,
      "learning_rate": 0.0005802910249964327,
      "loss": 2.6208,
      "step": 26732
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6632776260375977,
      "learning_rate": 0.0005802895667717357,
      "loss": 2.3234,
      "step": 26733
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.589102566242218,
      "learning_rate": 0.0005802881084949274,
      "loss": 2.192,
      "step": 26734
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.678520143032074,
      "learning_rate": 0.0005802866501660084,
      "loss": 2.2446,
      "step": 26735
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6323050856590271,
      "learning_rate": 0.0005802851917849788,
      "loss": 2.4323,
      "step": 26736
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6938719749450684,
      "learning_rate": 0.0005802837333518389,
      "loss": 2.3802,
      "step": 26737
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6527679562568665,
      "learning_rate": 0.0005802822748665888,
      "loss": 2.2006,
      "step": 26738
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6987885236740112,
      "learning_rate": 0.0005802808163292291,
      "loss": 2.289,
      "step": 26739
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6592986583709717,
      "learning_rate": 0.00058027935773976,
      "loss": 2.2586,
      "step": 26740
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6346469521522522,
      "learning_rate": 0.0005802778990981815,
      "loss": 2.3402,
      "step": 26741
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6179246306419373,
      "learning_rate": 0.000580276440404494,
      "loss": 2.1045,
      "step": 26742
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6467647552490234,
      "learning_rate": 0.0005802749816586979,
      "loss": 2.2746,
      "step": 26743
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6369129419326782,
      "learning_rate": 0.0005802735228607934,
      "loss": 2.1755,
      "step": 26744
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.5500664710998535,
      "learning_rate": 0.0005802720640107808,
      "loss": 2.0251,
      "step": 26745
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6170493960380554,
      "learning_rate": 0.0005802706051086603,
      "loss": 2.3905,
      "step": 26746
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6477544903755188,
      "learning_rate": 0.0005802691461544322,
      "loss": 2.1599,
      "step": 26747
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6389087438583374,
      "learning_rate": 0.0005802676871480968,
      "loss": 2.3948,
      "step": 26748
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.622958242893219,
      "learning_rate": 0.0005802662280896542,
      "loss": 2.2423,
      "step": 26749
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6197144389152527,
      "learning_rate": 0.000580264768979105,
      "loss": 2.096,
      "step": 26750
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6660962700843811,
      "learning_rate": 0.0005802633098164492,
      "loss": 2.2308,
      "step": 26751
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.7097039222717285,
      "learning_rate": 0.0005802618506016872,
      "loss": 2.1466,
      "step": 26752
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.620594322681427,
      "learning_rate": 0.0005802603913348192,
      "loss": 2.2255,
      "step": 26753
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.5977410078048706,
      "learning_rate": 0.0005802589320158455,
      "loss": 2.4207,
      "step": 26754
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6174601912498474,
      "learning_rate": 0.0005802574726447665,
      "loss": 2.3927,
      "step": 26755
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6611079573631287,
      "learning_rate": 0.0005802560132215822,
      "loss": 2.2857,
      "step": 26756
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6226574182510376,
      "learning_rate": 0.0005802545537462931,
      "loss": 2.1094,
      "step": 26757
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.721122145652771,
      "learning_rate": 0.0005802530942188993,
      "loss": 2.4124,
      "step": 26758
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.7443239688873291,
      "learning_rate": 0.0005802516346394012,
      "loss": 2.3171,
      "step": 26759
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.7200210094451904,
      "learning_rate": 0.0005802501750077991,
      "loss": 2.3666,
      "step": 26760
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6423527002334595,
      "learning_rate": 0.000580248715324093,
      "loss": 2.4241,
      "step": 26761
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.8448760509490967,
      "learning_rate": 0.0005802472555882836,
      "loss": 2.3031,
      "step": 26762
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.687493622303009,
      "learning_rate": 0.0005802457958003709,
      "loss": 2.3294,
      "step": 26763
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.597002387046814,
      "learning_rate": 0.0005802443359603552,
      "loss": 2.209,
      "step": 26764
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6400431990623474,
      "learning_rate": 0.0005802428760682366,
      "loss": 2.4067,
      "step": 26765
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.656999409198761,
      "learning_rate": 0.0005802414161240158,
      "loss": 2.3657,
      "step": 26766
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6151689887046814,
      "learning_rate": 0.0005802399561276927,
      "loss": 2.106,
      "step": 26767
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6529373526573181,
      "learning_rate": 0.0005802384960792678,
      "loss": 2.3267,
      "step": 26768
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.7121928930282593,
      "learning_rate": 0.0005802370359787412,
      "loss": 2.6276,
      "step": 26769
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.792104184627533,
      "learning_rate": 0.0005802355758261133,
      "loss": 2.2406,
      "step": 26770
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6702459454536438,
      "learning_rate": 0.0005802341156213843,
      "loss": 2.4926,
      "step": 26771
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6827144622802734,
      "learning_rate": 0.0005802326553645545,
      "loss": 2.2236,
      "step": 26772
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6293845176696777,
      "learning_rate": 0.0005802311950556242,
      "loss": 2.2275,
      "step": 26773
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6682848930358887,
      "learning_rate": 0.0005802297346945935,
      "loss": 2.4891,
      "step": 26774
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.64980548620224,
      "learning_rate": 0.0005802282742814629,
      "loss": 2.268,
      "step": 26775
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6431592106819153,
      "learning_rate": 0.0005802268138162325,
      "loss": 2.1534,
      "step": 26776
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.7937051653862,
      "learning_rate": 0.0005802253532989027,
      "loss": 2.3832,
      "step": 26777
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6078837513923645,
      "learning_rate": 0.0005802238927294737,
      "loss": 2.1899,
      "step": 26778
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.5854292511940002,
      "learning_rate": 0.0005802224321079458,
      "loss": 2.1748,
      "step": 26779
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.8431334495544434,
      "learning_rate": 0.0005802209714343193,
      "loss": 2.2921,
      "step": 26780
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.8723734617233276,
      "learning_rate": 0.0005802195107085943,
      "loss": 2.4314,
      "step": 26781
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6074331998825073,
      "learning_rate": 0.0005802180499307714,
      "loss": 2.2077,
      "step": 26782
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.7805114388465881,
      "learning_rate": 0.0005802165891008505,
      "loss": 2.226,
      "step": 26783
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.565605103969574,
      "learning_rate": 0.0005802151282188321,
      "loss": 2.1282,
      "step": 26784
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.666793167591095,
      "learning_rate": 0.0005802136672847164,
      "loss": 2.2445,
      "step": 26785
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.5722004771232605,
      "learning_rate": 0.0005802122062985038,
      "loss": 2.2811,
      "step": 26786
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.7671300172805786,
      "learning_rate": 0.0005802107452601943,
      "loss": 2.4785,
      "step": 26787
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.5684707164764404,
      "learning_rate": 0.0005802092841697884,
      "loss": 2.1893,
      "step": 26788
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6541914343833923,
      "learning_rate": 0.0005802078230272864,
      "loss": 2.897,
      "step": 26789
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6265064477920532,
      "learning_rate": 0.0005802063618326884,
      "loss": 2.2675,
      "step": 26790
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.5406583547592163,
      "learning_rate": 0.0005802049005859947,
      "loss": 2.1665,
      "step": 26791
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6506218910217285,
      "learning_rate": 0.0005802034392872057,
      "loss": 2.4828,
      "step": 26792
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.651647686958313,
      "learning_rate": 0.0005802019779363216,
      "loss": 2.4317,
      "step": 26793
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6999605894088745,
      "learning_rate": 0.0005802005165333426,
      "loss": 2.4562,
      "step": 26794
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6628484725952148,
      "learning_rate": 0.000580199055078269,
      "loss": 2.0782,
      "step": 26795
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6014090180397034,
      "learning_rate": 0.0005801975935711012,
      "loss": 2.3038,
      "step": 26796
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6639772057533264,
      "learning_rate": 0.0005801961320118393,
      "loss": 2.4071,
      "step": 26797
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.5634359121322632,
      "learning_rate": 0.0005801946704004837,
      "loss": 2.1799,
      "step": 26798
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6515985727310181,
      "learning_rate": 0.0005801932087370346,
      "loss": 2.2966,
      "step": 26799
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6727861166000366,
      "learning_rate": 0.0005801917470214924,
      "loss": 2.6219,
      "step": 26800
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.626730740070343,
      "learning_rate": 0.0005801902852538572,
      "loss": 2.3395,
      "step": 26801
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.644166886806488,
      "learning_rate": 0.0005801888234341292,
      "loss": 2.1121,
      "step": 26802
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6401656866073608,
      "learning_rate": 0.000580187361562309,
      "loss": 2.4829,
      "step": 26803
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6655569672584534,
      "learning_rate": 0.0005801858996383964,
      "loss": 2.1093,
      "step": 26804
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6738066077232361,
      "learning_rate": 0.0005801844376623922,
      "loss": 2.2968,
      "step": 26805
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6199941039085388,
      "learning_rate": 0.0005801829756342964,
      "loss": 2.4266,
      "step": 26806
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6101313233375549,
      "learning_rate": 0.0005801815135541092,
      "loss": 2.4082,
      "step": 26807
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6200380921363831,
      "learning_rate": 0.0005801800514218311,
      "loss": 2.5299,
      "step": 26808
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6214584112167358,
      "learning_rate": 0.0005801785892374622,
      "loss": 2.2523,
      "step": 26809
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.9230895042419434,
      "learning_rate": 0.0005801771270010028,
      "loss": 2.2969,
      "step": 26810
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.5967028737068176,
      "learning_rate": 0.0005801756647124532,
      "loss": 2.5129,
      "step": 26811
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.0785232782363892,
      "learning_rate": 0.0005801742023718136,
      "loss": 2.1839,
      "step": 26812
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6528357267379761,
      "learning_rate": 0.0005801727399790843,
      "loss": 2.138,
      "step": 26813
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.702958345413208,
      "learning_rate": 0.0005801712775342656,
      "loss": 2.421,
      "step": 26814
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.7146260738372803,
      "learning_rate": 0.000580169815037358,
      "loss": 2.368,
      "step": 26815
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.7105469703674316,
      "learning_rate": 0.0005801683524883612,
      "loss": 2.1907,
      "step": 26816
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6194331049919128,
      "learning_rate": 0.0005801668898872761,
      "loss": 2.3723,
      "step": 26817
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.7027361989021301,
      "learning_rate": 0.0005801654272341025,
      "loss": 2.6231,
      "step": 26818
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.5935777425765991,
      "learning_rate": 0.000580163964528841,
      "loss": 2.5288,
      "step": 26819
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6264243125915527,
      "learning_rate": 0.0005801625017714916,
      "loss": 2.1377,
      "step": 26820
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6571211218833923,
      "learning_rate": 0.0005801610389620548,
      "loss": 2.5371,
      "step": 26821
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6148551106452942,
      "learning_rate": 0.0005801595761005307,
      "loss": 2.4219,
      "step": 26822
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6476455330848694,
      "learning_rate": 0.0005801581131869197,
      "loss": 2.4426,
      "step": 26823
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6100637912750244,
      "learning_rate": 0.000580156650221222,
      "loss": 2.5543,
      "step": 26824
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.5924605131149292,
      "learning_rate": 0.0005801551872034378,
      "loss": 2.3291,
      "step": 26825
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.5854295492172241,
      "learning_rate": 0.0005801537241335676,
      "loss": 2.3496,
      "step": 26826
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6137083172798157,
      "learning_rate": 0.0005801522610116114,
      "loss": 2.2219,
      "step": 26827
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.5916731357574463,
      "learning_rate": 0.0005801507978375699,
      "loss": 2.5248,
      "step": 26828
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6386590003967285,
      "learning_rate": 0.0005801493346114428,
      "loss": 2.1979,
      "step": 26829
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6154048442840576,
      "learning_rate": 0.0005801478713332307,
      "loss": 2.3484,
      "step": 26830
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.5926229357719421,
      "learning_rate": 0.0005801464080029338,
      "loss": 2.2748,
      "step": 26831
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6433006525039673,
      "learning_rate": 0.0005801449446205525,
      "loss": 2.4119,
      "step": 26832
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6159801483154297,
      "learning_rate": 0.0005801434811860868,
      "loss": 2.2278,
      "step": 26833
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.5990481376647949,
      "learning_rate": 0.0005801420176995373,
      "loss": 2.3866,
      "step": 26834
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.5916231870651245,
      "learning_rate": 0.000580140554160904,
      "loss": 2.6487,
      "step": 26835
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.627813994884491,
      "learning_rate": 0.0005801390905701873,
      "loss": 2.1761,
      "step": 26836
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6130495667457581,
      "learning_rate": 0.0005801376269273876,
      "loss": 2.1391,
      "step": 26837
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6923011541366577,
      "learning_rate": 0.0005801361632325049,
      "loss": 2.5246,
      "step": 26838
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6613107323646545,
      "learning_rate": 0.0005801346994855396,
      "loss": 2.3536,
      "step": 26839
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6160924434661865,
      "learning_rate": 0.0005801332356864921,
      "loss": 2.3117,
      "step": 26840
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.7431180477142334,
      "learning_rate": 0.0005801317718353623,
      "loss": 2.4236,
      "step": 26841
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.7143204212188721,
      "learning_rate": 0.0005801303079321509,
      "loss": 2.3283,
      "step": 26842
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.717749297618866,
      "learning_rate": 0.000580128843976858,
      "loss": 2.4546,
      "step": 26843
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.7742488980293274,
      "learning_rate": 0.0005801273799694837,
      "loss": 2.2096,
      "step": 26844
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.7551416158676147,
      "learning_rate": 0.0005801259159100285,
      "loss": 2.4334,
      "step": 26845
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.5613329410552979,
      "learning_rate": 0.0005801244517984926,
      "loss": 2.2436,
      "step": 26846
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6309730410575867,
      "learning_rate": 0.0005801229876348764,
      "loss": 2.2447,
      "step": 26847
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6636571884155273,
      "learning_rate": 0.0005801215234191799,
      "loss": 2.3629,
      "step": 26848
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6273710131645203,
      "learning_rate": 0.0005801200591514037,
      "loss": 2.5564,
      "step": 26849
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6478692889213562,
      "learning_rate": 0.0005801185948315476,
      "loss": 2.191,
      "step": 26850
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.9026604890823364,
      "learning_rate": 0.0005801171304596124,
      "loss": 2.4959,
      "step": 26851
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6682243347167969,
      "learning_rate": 0.0005801156660355981,
      "loss": 2.299,
      "step": 26852
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.654138445854187,
      "learning_rate": 0.0005801142015595049,
      "loss": 2.4983,
      "step": 26853
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6133281588554382,
      "learning_rate": 0.0005801127370313333,
      "loss": 2.3281,
      "step": 26854
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.7062997817993164,
      "learning_rate": 0.0005801112724510835,
      "loss": 1.9994,
      "step": 26855
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.7247060537338257,
      "learning_rate": 0.0005801098078187557,
      "loss": 2.2131,
      "step": 26856
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.727531373500824,
      "learning_rate": 0.0005801083431343501,
      "loss": 2.3298,
      "step": 26857
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.659798800945282,
      "learning_rate": 0.0005801068783978671,
      "loss": 2.3126,
      "step": 26858
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6189098954200745,
      "learning_rate": 0.000580105413609307,
      "loss": 2.0289,
      "step": 26859
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6660752892494202,
      "learning_rate": 0.00058010394876867,
      "loss": 2.4024,
      "step": 26860
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6578737497329712,
      "learning_rate": 0.0005801024838759562,
      "loss": 2.448,
      "step": 26861
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.570239245891571,
      "learning_rate": 0.0005801010189311663,
      "loss": 2.3524,
      "step": 26862
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.7240305542945862,
      "learning_rate": 0.0005800995539343003,
      "loss": 2.4338,
      "step": 26863
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6141300201416016,
      "learning_rate": 0.0005800980888853585,
      "loss": 2.1451,
      "step": 26864
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6582244038581848,
      "learning_rate": 0.0005800966237843411,
      "loss": 2.326,
      "step": 26865
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6768122315406799,
      "learning_rate": 0.0005800951586312485,
      "loss": 2.4339,
      "step": 26866
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6142621040344238,
      "learning_rate": 0.0005800936934260809,
      "loss": 2.5134,
      "step": 26867
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6688772439956665,
      "learning_rate": 0.0005800922281688387,
      "loss": 2.329,
      "step": 26868
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.652884840965271,
      "learning_rate": 0.0005800907628595219,
      "loss": 2.3948,
      "step": 26869
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6221352219581604,
      "learning_rate": 0.000580089297498131,
      "loss": 2.2275,
      "step": 26870
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6039418578147888,
      "learning_rate": 0.0005800878320846663,
      "loss": 2.3703,
      "step": 26871
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.7500925064086914,
      "learning_rate": 0.0005800863666191279,
      "loss": 2.1872,
      "step": 26872
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6509643197059631,
      "learning_rate": 0.0005800849011015161,
      "loss": 2.4638,
      "step": 26873
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6014090776443481,
      "learning_rate": 0.0005800834355318314,
      "loss": 2.3529,
      "step": 26874
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6512938141822815,
      "learning_rate": 0.0005800819699100738,
      "loss": 2.453,
      "step": 26875
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.8159506916999817,
      "learning_rate": 0.0005800805042362436,
      "loss": 2.195,
      "step": 26876
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6673114895820618,
      "learning_rate": 0.0005800790385103413,
      "loss": 2.2096,
      "step": 26877
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.748541533946991,
      "learning_rate": 0.000580077572732367,
      "loss": 2.4976,
      "step": 26878
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6288380026817322,
      "learning_rate": 0.0005800761069023209,
      "loss": 2.201,
      "step": 26879
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6483588814735413,
      "learning_rate": 0.0005800746410202035,
      "loss": 2.4495,
      "step": 26880
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.5961581468582153,
      "learning_rate": 0.0005800731750860148,
      "loss": 2.4178,
      "step": 26881
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6477116942405701,
      "learning_rate": 0.0005800717090997553,
      "loss": 2.0633,
      "step": 26882
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6910370588302612,
      "learning_rate": 0.0005800702430614251,
      "loss": 2.0681,
      "step": 26883
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6069654226303101,
      "learning_rate": 0.0005800687769710245,
      "loss": 2.3636,
      "step": 26884
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6211944818496704,
      "learning_rate": 0.0005800673108285541,
      "loss": 2.1884,
      "step": 26885
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6941606402397156,
      "learning_rate": 0.0005800658446340136,
      "loss": 2.488,
      "step": 26886
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.7063273191452026,
      "learning_rate": 0.0005800643783874037,
      "loss": 2.1788,
      "step": 26887
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6738278865814209,
      "learning_rate": 0.0005800629120887245,
      "loss": 2.4597,
      "step": 26888
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6487999558448792,
      "learning_rate": 0.0005800614457379765,
      "loss": 2.3582,
      "step": 26889
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6815968155860901,
      "learning_rate": 0.0005800599793351595,
      "loss": 2.4961,
      "step": 26890
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6249327063560486,
      "learning_rate": 0.0005800585128802742,
      "loss": 2.0279,
      "step": 26891
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6449186205863953,
      "learning_rate": 0.0005800570463733208,
      "loss": 2.1937,
      "step": 26892
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.5980067253112793,
      "learning_rate": 0.0005800555798142994,
      "loss": 2.564,
      "step": 26893
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.7943875789642334,
      "learning_rate": 0.0005800541132032105,
      "loss": 2.442,
      "step": 26894
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6754232048988342,
      "learning_rate": 0.0005800526465400542,
      "loss": 2.1307,
      "step": 26895
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.679839551448822,
      "learning_rate": 0.0005800511798248309,
      "loss": 2.3554,
      "step": 26896
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.7418435215950012,
      "learning_rate": 0.0005800497130575407,
      "loss": 2.3091,
      "step": 26897
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6800909638404846,
      "learning_rate": 0.0005800482462381839,
      "loss": 2.5141,
      "step": 26898
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.7416391968727112,
      "learning_rate": 0.0005800467793667609,
      "loss": 2.4698,
      "step": 26899
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.8405999541282654,
      "learning_rate": 0.0005800453124432719,
      "loss": 2.4557,
      "step": 26900
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.71392422914505,
      "learning_rate": 0.0005800438454677174,
      "loss": 2.2956,
      "step": 26901
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6886312961578369,
      "learning_rate": 0.0005800423784400973,
      "loss": 2.5844,
      "step": 26902
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.7973476648330688,
      "learning_rate": 0.0005800409113604119,
      "loss": 2.1964,
      "step": 26903
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6750009059906006,
      "learning_rate": 0.0005800394442286619,
      "loss": 2.4852,
      "step": 26904
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6311566829681396,
      "learning_rate": 0.0005800379770448471,
      "loss": 2.4731,
      "step": 26905
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.9514658451080322,
      "learning_rate": 0.000580036509808968,
      "loss": 2.3705,
      "step": 26906
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6563698649406433,
      "learning_rate": 0.0005800350425210248,
      "loss": 2.367,
      "step": 26907
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6870530843734741,
      "learning_rate": 0.0005800335751810178,
      "loss": 2.3678,
      "step": 26908
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6117283701896667,
      "learning_rate": 0.0005800321077889472,
      "loss": 2.311,
      "step": 26909
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6193177103996277,
      "learning_rate": 0.0005800306403448136,
      "loss": 2.412,
      "step": 26910
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6495373845100403,
      "learning_rate": 0.0005800291728486167,
      "loss": 2.2663,
      "step": 26911
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6526059508323669,
      "learning_rate": 0.0005800277053003573,
      "loss": 2.0874,
      "step": 26912
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6905961036682129,
      "learning_rate": 0.0005800262377000354,
      "loss": 2.4183,
      "step": 26913
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6943082809448242,
      "learning_rate": 0.0005800247700476512,
      "loss": 1.9461,
      "step": 26914
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6902804970741272,
      "learning_rate": 0.0005800233023432054,
      "loss": 2.2836,
      "step": 26915
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6194450259208679,
      "learning_rate": 0.0005800218345866978,
      "loss": 2.2688,
      "step": 26916
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.7203478813171387,
      "learning_rate": 0.0005800203667781289,
      "loss": 2.2232,
      "step": 26917
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.7021436095237732,
      "learning_rate": 0.0005800188989174989,
      "loss": 2.3931,
      "step": 26918
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6266292929649353,
      "learning_rate": 0.0005800174310048081,
      "loss": 2.3751,
      "step": 26919
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.7321296334266663,
      "learning_rate": 0.0005800159630400568,
      "loss": 2.3699,
      "step": 26920
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.7149727940559387,
      "learning_rate": 0.0005800144950232452,
      "loss": 2.5418,
      "step": 26921
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.7065470814704895,
      "learning_rate": 0.0005800130269543736,
      "loss": 2.4963,
      "step": 26922
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.7454004883766174,
      "learning_rate": 0.0005800115588334424,
      "loss": 2.402,
      "step": 26923
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.7202045917510986,
      "learning_rate": 0.0005800100906604517,
      "loss": 2.5142,
      "step": 26924
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.7173725366592407,
      "learning_rate": 0.0005800086224354019,
      "loss": 2.2653,
      "step": 26925
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6350707411766052,
      "learning_rate": 0.0005800071541582931,
      "loss": 2.3118,
      "step": 26926
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.5810220837593079,
      "learning_rate": 0.0005800056858291258,
      "loss": 2.4448,
      "step": 26927
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.7459020018577576,
      "learning_rate": 0.0005800042174479001,
      "loss": 2.2016,
      "step": 26928
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.653769314289093,
      "learning_rate": 0.0005800027490146163,
      "loss": 2.3582,
      "step": 26929
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6662362813949585,
      "learning_rate": 0.0005800012805292747,
      "loss": 2.0342,
      "step": 26930
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6826421618461609,
      "learning_rate": 0.0005799998119918756,
      "loss": 2.257,
      "step": 26931
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6211981177330017,
      "learning_rate": 0.0005799983434024193,
      "loss": 2.1404,
      "step": 26932
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6946744322776794,
      "learning_rate": 0.000579996874760906,
      "loss": 2.2586,
      "step": 26933
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.7080943584442139,
      "learning_rate": 0.000579995406067336,
      "loss": 2.0882,
      "step": 26934
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.7990931272506714,
      "learning_rate": 0.0005799939373217095,
      "loss": 2.2616,
      "step": 26935
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.7474820017814636,
      "learning_rate": 0.000579992468524027,
      "loss": 2.1302,
      "step": 26936
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.7471941113471985,
      "learning_rate": 0.0005799909996742885,
      "loss": 2.3795,
      "step": 26937
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6378636360168457,
      "learning_rate": 0.0005799895307724944,
      "loss": 2.2925,
      "step": 26938
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.7667298316955566,
      "learning_rate": 0.000579988061818645,
      "loss": 2.135,
      "step": 26939
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6696726679801941,
      "learning_rate": 0.0005799865928127404,
      "loss": 2.4015,
      "step": 26940
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6209779381752014,
      "learning_rate": 0.0005799851237547811,
      "loss": 2.1921,
      "step": 26941
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6353320479393005,
      "learning_rate": 0.0005799836546447673,
      "loss": 2.3598,
      "step": 26942
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6530837416648865,
      "learning_rate": 0.0005799821854826992,
      "loss": 2.4614,
      "step": 26943
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6708739399909973,
      "learning_rate": 0.0005799807162685771,
      "loss": 2.385,
      "step": 26944
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.5662838816642761,
      "learning_rate": 0.0005799792470024014,
      "loss": 2.2954,
      "step": 26945
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6754138469696045,
      "learning_rate": 0.0005799777776841722,
      "loss": 2.4588,
      "step": 26946
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.7568691372871399,
      "learning_rate": 0.0005799763083138899,
      "loss": 2.2542,
      "step": 26947
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6154218912124634,
      "learning_rate": 0.0005799748388915548,
      "loss": 2.3101,
      "step": 26948
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6147028803825378,
      "learning_rate": 0.0005799733694171669,
      "loss": 2.2635,
      "step": 26949
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.653545081615448,
      "learning_rate": 0.0005799718998907268,
      "loss": 2.3093,
      "step": 26950
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6362476348876953,
      "learning_rate": 0.0005799704303122346,
      "loss": 2.2871,
      "step": 26951
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.5952627658843994,
      "learning_rate": 0.0005799689606816905,
      "loss": 2.2932,
      "step": 26952
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6539112329483032,
      "learning_rate": 0.000579967490999095,
      "loss": 2.2664,
      "step": 26953
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6191897988319397,
      "learning_rate": 0.0005799660212644482,
      "loss": 2.3226,
      "step": 26954
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.694908857345581,
      "learning_rate": 0.0005799645514777506,
      "loss": 2.4047,
      "step": 26955
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.7026556134223938,
      "learning_rate": 0.0005799630816390022,
      "loss": 2.2463,
      "step": 26956
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6303671002388,
      "learning_rate": 0.0005799616117482033,
      "loss": 2.5021,
      "step": 26957
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6602061986923218,
      "learning_rate": 0.0005799601418053544,
      "loss": 2.1737,
      "step": 26958
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6306611895561218,
      "learning_rate": 0.0005799586718104556,
      "loss": 2.3734,
      "step": 26959
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6205613613128662,
      "learning_rate": 0.000579957201763507,
      "loss": 2.3955,
      "step": 26960
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.7048942446708679,
      "learning_rate": 0.0005799557316645092,
      "loss": 2.3003,
      "step": 26961
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6856035590171814,
      "learning_rate": 0.0005799542615134624,
      "loss": 2.0922,
      "step": 26962
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6452313661575317,
      "learning_rate": 0.0005799527913103668,
      "loss": 2.1933,
      "step": 26963
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.64079350233078,
      "learning_rate": 0.0005799513210552227,
      "loss": 2.4043,
      "step": 26964
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.8862414956092834,
      "learning_rate": 0.0005799498507480303,
      "loss": 2.2429,
      "step": 26965
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6464040279388428,
      "learning_rate": 0.00057994838038879,
      "loss": 2.2464,
      "step": 26966
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6571405529975891,
      "learning_rate": 0.0005799469099775019,
      "loss": 2.255,
      "step": 26967
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.7287579774856567,
      "learning_rate": 0.0005799454395141665,
      "loss": 2.4071,
      "step": 26968
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6392391920089722,
      "learning_rate": 0.000579943968998784,
      "loss": 2.0916,
      "step": 26969
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6267512440681458,
      "learning_rate": 0.0005799424984313545,
      "loss": 2.3902,
      "step": 26970
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6626600623130798,
      "learning_rate": 0.0005799410278118784,
      "loss": 2.5024,
      "step": 26971
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.7806633710861206,
      "learning_rate": 0.0005799395571403561,
      "loss": 2.347,
      "step": 26972
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.7164724469184875,
      "learning_rate": 0.0005799380864167877,
      "loss": 2.4251,
      "step": 26973
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.7136653065681458,
      "learning_rate": 0.0005799366156411735,
      "loss": 2.3031,
      "step": 26974
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.7878042459487915,
      "learning_rate": 0.0005799351448135139,
      "loss": 2.4254,
      "step": 26975
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6760825514793396,
      "learning_rate": 0.0005799336739338089,
      "loss": 2.3051,
      "step": 26976
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6136844754219055,
      "learning_rate": 0.000579932203002059,
      "loss": 2.2824,
      "step": 26977
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6710241436958313,
      "learning_rate": 0.0005799307320182645,
      "loss": 2.2495,
      "step": 26978
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.7338106036186218,
      "learning_rate": 0.0005799292609824255,
      "loss": 2.4883,
      "step": 26979
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.663947343826294,
      "learning_rate": 0.0005799277898945424,
      "loss": 2.3757,
      "step": 26980
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6373106241226196,
      "learning_rate": 0.0005799263187546155,
      "loss": 2.5185,
      "step": 26981
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6645002961158752,
      "learning_rate": 0.000579924847562645,
      "loss": 2.5024,
      "step": 26982
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6512279510498047,
      "learning_rate": 0.0005799233763186311,
      "loss": 2.3001,
      "step": 26983
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.7001825571060181,
      "learning_rate": 0.0005799219050225743,
      "loss": 2.2629,
      "step": 26984
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6135714650154114,
      "learning_rate": 0.0005799204336744745,
      "loss": 2.1792,
      "step": 26985
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.5780208706855774,
      "learning_rate": 0.0005799189622743324,
      "loss": 2.4063,
      "step": 26986
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.5907486081123352,
      "learning_rate": 0.0005799174908221481,
      "loss": 2.3633,
      "step": 26987
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6500486731529236,
      "learning_rate": 0.0005799160193179218,
      "loss": 2.323,
      "step": 26988
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.5898603200912476,
      "learning_rate": 0.0005799145477616538,
      "loss": 2.1213,
      "step": 26989
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.674243152141571,
      "learning_rate": 0.0005799130761533444,
      "loss": 2.2584,
      "step": 26990
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6236706972122192,
      "learning_rate": 0.0005799116044929939,
      "loss": 2.4924,
      "step": 26991
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6778797507286072,
      "learning_rate": 0.0005799101327806026,
      "loss": 2.2882,
      "step": 26992
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.5969259738922119,
      "learning_rate": 0.0005799086610161706,
      "loss": 2.2365,
      "step": 26993
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6253164410591125,
      "learning_rate": 0.0005799071891996984,
      "loss": 2.0929,
      "step": 26994
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6228911876678467,
      "learning_rate": 0.0005799057173311861,
      "loss": 2.4346,
      "step": 26995
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.5956201553344727,
      "learning_rate": 0.0005799042454106341,
      "loss": 2.123,
      "step": 26996
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6250074505805969,
      "learning_rate": 0.0005799027734380425,
      "loss": 2.5241,
      "step": 26997
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.5979002118110657,
      "learning_rate": 0.0005799013014134119,
      "loss": 2.1759,
      "step": 26998
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.5938346982002258,
      "learning_rate": 0.0005798998293367422,
      "loss": 2.2613,
      "step": 26999
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6693408489227295,
      "learning_rate": 0.0005798983572080339,
      "loss": 2.6704,
      "step": 27000
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6287779808044434,
      "learning_rate": 0.0005798968850272872,
      "loss": 2.4075,
      "step": 27001
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6723875999450684,
      "learning_rate": 0.0005798954127945023,
      "loss": 2.3329,
      "step": 27002
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6535650491714478,
      "learning_rate": 0.0005798939405096797,
      "loss": 2.2961,
      "step": 27003
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6057156920433044,
      "learning_rate": 0.0005798924681728195,
      "loss": 2.0309,
      "step": 27004
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.7046760320663452,
      "learning_rate": 0.0005798909957839219,
      "loss": 2.0903,
      "step": 27005
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6207306981086731,
      "learning_rate": 0.0005798895233429874,
      "loss": 2.3592,
      "step": 27006
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6702345013618469,
      "learning_rate": 0.0005798880508500161,
      "loss": 2.4114,
      "step": 27007
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.7398874759674072,
      "learning_rate": 0.0005798865783050084,
      "loss": 2.283,
      "step": 27008
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6066060662269592,
      "learning_rate": 0.0005798851057079644,
      "loss": 2.2811,
      "step": 27009
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.5921414494514465,
      "learning_rate": 0.0005798836330588846,
      "loss": 2.2074,
      "step": 27010
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.7008029222488403,
      "learning_rate": 0.0005798821603577692,
      "loss": 2.3279,
      "step": 27011
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6269589066505432,
      "learning_rate": 0.0005798806876046182,
      "loss": 2.3189,
      "step": 27012
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6438650488853455,
      "learning_rate": 0.0005798792147994321,
      "loss": 2.3489,
      "step": 27013
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6378695964813232,
      "learning_rate": 0.0005798777419422112,
      "loss": 2.3968,
      "step": 27014
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6545892953872681,
      "learning_rate": 0.000579876269032956,
      "loss": 2.3215,
      "step": 27015
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.8484004139900208,
      "learning_rate": 0.0005798747960716662,
      "loss": 2.4675,
      "step": 27016
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.8656587600708008,
      "learning_rate": 0.0005798733230583426,
      "loss": 2.3128,
      "step": 27017
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6511625051498413,
      "learning_rate": 0.0005798718499929852,
      "loss": 2.2973,
      "step": 27018
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6641950011253357,
      "learning_rate": 0.0005798703768755943,
      "loss": 2.3353,
      "step": 27019
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.7330968379974365,
      "learning_rate": 0.0005798689037061702,
      "loss": 2.304,
      "step": 27020
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6737914085388184,
      "learning_rate": 0.0005798674304847133,
      "loss": 2.3881,
      "step": 27021
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.7591213583946228,
      "learning_rate": 0.0005798659572112236,
      "loss": 2.4046,
      "step": 27022
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6326737403869629,
      "learning_rate": 0.0005798644838857017,
      "loss": 2.2519,
      "step": 27023
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6586238741874695,
      "learning_rate": 0.0005798630105081476,
      "loss": 2.5099,
      "step": 27024
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6708562970161438,
      "learning_rate": 0.0005798615370785617,
      "loss": 1.9909,
      "step": 27025
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6454222798347473,
      "learning_rate": 0.0005798600635969443,
      "loss": 2.626,
      "step": 27026
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6919674277305603,
      "learning_rate": 0.0005798585900632956,
      "loss": 2.4841,
      "step": 27027
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6453232765197754,
      "learning_rate": 0.000579857116477616,
      "loss": 2.2345,
      "step": 27028
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6994619369506836,
      "learning_rate": 0.0005798556428399055,
      "loss": 2.2874,
      "step": 27029
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6266087293624878,
      "learning_rate": 0.0005798541691501647,
      "loss": 2.2126,
      "step": 27030
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.5843597054481506,
      "learning_rate": 0.0005798526954083937,
      "loss": 2.302,
      "step": 27031
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.9070834517478943,
      "learning_rate": 0.0005798512216145927,
      "loss": 2.3908,
      "step": 27032
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.5970584154129028,
      "learning_rate": 0.0005798497477687622,
      "loss": 2.2367,
      "step": 27033
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.5978009700775146,
      "learning_rate": 0.0005798482738709023,
      "loss": 2.376,
      "step": 27034
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.9223513007164001,
      "learning_rate": 0.0005798467999210133,
      "loss": 2.2302,
      "step": 27035
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6554684042930603,
      "learning_rate": 0.0005798453259190955,
      "loss": 2.4115,
      "step": 27036
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.5961574912071228,
      "learning_rate": 0.0005798438518651492,
      "loss": 2.532,
      "step": 27037
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.8017905354499817,
      "learning_rate": 0.0005798423777591746,
      "loss": 2.3381,
      "step": 27038
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6704971194267273,
      "learning_rate": 0.0005798409036011722,
      "loss": 2.2907,
      "step": 27039
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6131525635719299,
      "learning_rate": 0.0005798394293911419,
      "loss": 2.2511,
      "step": 27040
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6661485433578491,
      "learning_rate": 0.0005798379551290842,
      "loss": 2.2657,
      "step": 27041
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.9801290035247803,
      "learning_rate": 0.0005798364808149993,
      "loss": 2.292,
      "step": 27042
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6391452550888062,
      "learning_rate": 0.0005798350064488877,
      "loss": 2.5216,
      "step": 27043
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.8307188749313354,
      "learning_rate": 0.0005798335320307493,
      "loss": 2.2148,
      "step": 27044
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6787518262863159,
      "learning_rate": 0.0005798320575605846,
      "loss": 2.2076,
      "step": 27045
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.7690195441246033,
      "learning_rate": 0.0005798305830383939,
      "loss": 2.3841,
      "step": 27046
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6286706328392029,
      "learning_rate": 0.0005798291084641773,
      "loss": 2.4389,
      "step": 27047
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.7713147401809692,
      "learning_rate": 0.0005798276338379353,
      "loss": 2.0943,
      "step": 27048
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.7228469252586365,
      "learning_rate": 0.0005798261591596681,
      "loss": 2.091,
      "step": 27049
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.7145320773124695,
      "learning_rate": 0.0005798246844293757,
      "loss": 2.334,
      "step": 27050
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6825727224349976,
      "learning_rate": 0.0005798232096470589,
      "loss": 2.3226,
      "step": 27051
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6721373796463013,
      "learning_rate": 0.0005798217348127175,
      "loss": 2.4373,
      "step": 27052
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.7251735925674438,
      "learning_rate": 0.0005798202599263521,
      "loss": 2.43,
      "step": 27053
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6542491316795349,
      "learning_rate": 0.0005798187849879626,
      "loss": 2.2726,
      "step": 27054
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6406742334365845,
      "learning_rate": 0.0005798173099975498,
      "loss": 2.1431,
      "step": 27055
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.728434145450592,
      "learning_rate": 0.0005798158349551134,
      "loss": 2.3138,
      "step": 27056
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6548416018486023,
      "learning_rate": 0.0005798143598606542,
      "loss": 2.3562,
      "step": 27057
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6465424299240112,
      "learning_rate": 0.000579812884714172,
      "loss": 2.2769,
      "step": 27058
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.5817879438400269,
      "learning_rate": 0.0005798114095156675,
      "loss": 2.2799,
      "step": 27059
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6365734338760376,
      "learning_rate": 0.0005798099342651407,
      "loss": 2.3415,
      "step": 27060
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6227615475654602,
      "learning_rate": 0.0005798084589625919,
      "loss": 2.1299,
      "step": 27061
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.598824679851532,
      "learning_rate": 0.0005798069836080215,
      "loss": 2.4687,
      "step": 27062
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6283334493637085,
      "learning_rate": 0.0005798055082014297,
      "loss": 2.5897,
      "step": 27063
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.598270058631897,
      "learning_rate": 0.0005798040327428168,
      "loss": 2.3462,
      "step": 27064
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6269626021385193,
      "learning_rate": 0.000579802557232183,
      "loss": 2.2291,
      "step": 27065
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6038095951080322,
      "learning_rate": 0.0005798010816695286,
      "loss": 2.1984,
      "step": 27066
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6203604340553284,
      "learning_rate": 0.0005797996060548539,
      "loss": 2.3469,
      "step": 27067
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6031585335731506,
      "learning_rate": 0.0005797981303881591,
      "loss": 2.4638,
      "step": 27068
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6268320679664612,
      "learning_rate": 0.0005797966546694447,
      "loss": 2.2028,
      "step": 27069
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6593037843704224,
      "learning_rate": 0.0005797951788987108,
      "loss": 2.0911,
      "step": 27070
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6581695079803467,
      "learning_rate": 0.0005797937030759576,
      "loss": 2.2814,
      "step": 27071
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6883793473243713,
      "learning_rate": 0.0005797922272011855,
      "loss": 2.179,
      "step": 27072
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.663000762462616,
      "learning_rate": 0.0005797907512743947,
      "loss": 2.1246,
      "step": 27073
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.7655686140060425,
      "learning_rate": 0.0005797892752955856,
      "loss": 2.3877,
      "step": 27074
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.5917879343032837,
      "learning_rate": 0.0005797877992647583,
      "loss": 2.1493,
      "step": 27075
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.680940568447113,
      "learning_rate": 0.0005797863231819133,
      "loss": 2.2888,
      "step": 27076
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6557303667068481,
      "learning_rate": 0.0005797848470470507,
      "loss": 2.3818,
      "step": 27077
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6738454699516296,
      "learning_rate": 0.0005797833708601708,
      "loss": 2.605,
      "step": 27078
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6061534285545349,
      "learning_rate": 0.0005797818946212739,
      "loss": 2.5212,
      "step": 27079
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6166258454322815,
      "learning_rate": 0.0005797804183303602,
      "loss": 2.3434,
      "step": 27080
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.635510265827179,
      "learning_rate": 0.0005797789419874301,
      "loss": 2.1626,
      "step": 27081
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6115184426307678,
      "learning_rate": 0.0005797774655924838,
      "loss": 2.3491,
      "step": 27082
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6458042860031128,
      "learning_rate": 0.0005797759891455216,
      "loss": 2.3991,
      "step": 27083
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6436767578125,
      "learning_rate": 0.0005797745126465437,
      "loss": 2.2377,
      "step": 27084
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6481337547302246,
      "learning_rate": 0.0005797730360955506,
      "loss": 2.4505,
      "step": 27085
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6258864998817444,
      "learning_rate": 0.0005797715594925422,
      "loss": 2.1072,
      "step": 27086
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.7764955163002014,
      "learning_rate": 0.0005797700828375191,
      "loss": 2.3579,
      "step": 27087
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6030167937278748,
      "learning_rate": 0.0005797686061304815,
      "loss": 2.4286,
      "step": 27088
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.5816143155097961,
      "learning_rate": 0.0005797671293714295,
      "loss": 2.3024,
      "step": 27089
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6575132012367249,
      "learning_rate": 0.0005797656525603636,
      "loss": 2.4444,
      "step": 27090
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6117950677871704,
      "learning_rate": 0.000579764175697284,
      "loss": 2.1787,
      "step": 27091
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6396900415420532,
      "learning_rate": 0.0005797626987821909,
      "loss": 2.2235,
      "step": 27092
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6543675065040588,
      "learning_rate": 0.0005797612218150847,
      "loss": 2.0413,
      "step": 27093
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6289651989936829,
      "learning_rate": 0.0005797597447959656,
      "loss": 2.3298,
      "step": 27094
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6059786677360535,
      "learning_rate": 0.0005797582677248338,
      "loss": 2.4436,
      "step": 27095
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.7918374538421631,
      "learning_rate": 0.0005797567906016898,
      "loss": 2.1898,
      "step": 27096
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.5969332456588745,
      "learning_rate": 0.0005797553134265336,
      "loss": 2.0981,
      "step": 27097
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6339575052261353,
      "learning_rate": 0.0005797538361993656,
      "loss": 2.209,
      "step": 27098
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6763978004455566,
      "learning_rate": 0.0005797523589201861,
      "loss": 2.3935,
      "step": 27099
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.7347848415374756,
      "learning_rate": 0.0005797508815889955,
      "loss": 2.4154,
      "step": 27100
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6862822771072388,
      "learning_rate": 0.0005797494042057938,
      "loss": 2.5981,
      "step": 27101
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6295124292373657,
      "learning_rate": 0.0005797479267705814,
      "loss": 2.2622,
      "step": 27102
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6359638571739197,
      "learning_rate": 0.0005797464492833586,
      "loss": 2.428,
      "step": 27103
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.617123007774353,
      "learning_rate": 0.0005797449717441257,
      "loss": 2.4032,
      "step": 27104
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.5834434032440186,
      "learning_rate": 0.0005797434941528829,
      "loss": 2.1567,
      "step": 27105
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6127311587333679,
      "learning_rate": 0.0005797420165096304,
      "loss": 2.6045,
      "step": 27106
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.5831398367881775,
      "learning_rate": 0.0005797405388143687,
      "loss": 2.2834,
      "step": 27107
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6398712992668152,
      "learning_rate": 0.000579739061067098,
      "loss": 2.3455,
      "step": 27108
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.7118551135063171,
      "learning_rate": 0.0005797375832678184,
      "loss": 2.2455,
      "step": 27109
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6444565057754517,
      "learning_rate": 0.0005797361054165303,
      "loss": 2.3967,
      "step": 27110
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.54902583360672,
      "learning_rate": 0.000579734627513234,
      "loss": 2.2667,
      "step": 27111
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6094775795936584,
      "learning_rate": 0.0005797331495579299,
      "loss": 2.3111,
      "step": 27112
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6422340869903564,
      "learning_rate": 0.000579731671550618,
      "loss": 2.3902,
      "step": 27113
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6085690259933472,
      "learning_rate": 0.0005797301934912987,
      "loss": 2.2194,
      "step": 27114
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6062719225883484,
      "learning_rate": 0.0005797287153799723,
      "loss": 2.159,
      "step": 27115
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.711772620677948,
      "learning_rate": 0.000579727237216639,
      "loss": 2.279,
      "step": 27116
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6440443396568298,
      "learning_rate": 0.0005797257590012992,
      "loss": 2.1712,
      "step": 27117
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.659247100353241,
      "learning_rate": 0.000579724280733953,
      "loss": 2.5353,
      "step": 27118
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6688565611839294,
      "learning_rate": 0.0005797228024146009,
      "loss": 2.4504,
      "step": 27119
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.7665070295333862,
      "learning_rate": 0.0005797213240432429,
      "loss": 2.4762,
      "step": 27120
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6362990736961365,
      "learning_rate": 0.0005797198456198796,
      "loss": 2.4494,
      "step": 27121
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.7206308245658875,
      "learning_rate": 0.000579718367144511,
      "loss": 2.3232,
      "step": 27122
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6278427243232727,
      "learning_rate": 0.0005797168886171375,
      "loss": 2.2443,
      "step": 27123
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6270710229873657,
      "learning_rate": 0.0005797154100377593,
      "loss": 2.3762,
      "step": 27124
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6276366710662842,
      "learning_rate": 0.0005797139314063769,
      "loss": 2.4084,
      "step": 27125
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6303402781486511,
      "learning_rate": 0.0005797124527229902,
      "loss": 2.3419,
      "step": 27126
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.608009934425354,
      "learning_rate": 0.0005797109739875998,
      "loss": 2.4258,
      "step": 27127
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.628363847732544,
      "learning_rate": 0.0005797094952002058,
      "loss": 2.3284,
      "step": 27128
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6258141994476318,
      "learning_rate": 0.0005797080163608085,
      "loss": 2.2038,
      "step": 27129
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6109122037887573,
      "learning_rate": 0.0005797065374694082,
      "loss": 2.375,
      "step": 27130
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6268970966339111,
      "learning_rate": 0.0005797050585260052,
      "loss": 2.3846,
      "step": 27131
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.627183735370636,
      "learning_rate": 0.0005797035795305997,
      "loss": 2.2392,
      "step": 27132
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6292206645011902,
      "learning_rate": 0.000579702100483192,
      "loss": 2.0232,
      "step": 27133
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.7038524746894836,
      "learning_rate": 0.0005797006213837826,
      "loss": 2.3809,
      "step": 27134
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6193267107009888,
      "learning_rate": 0.0005796991422323714,
      "loss": 2.2985,
      "step": 27135
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6167404651641846,
      "learning_rate": 0.000579697663028959,
      "loss": 2.2823,
      "step": 27136
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6710385680198669,
      "learning_rate": 0.0005796961837735453,
      "loss": 2.2596,
      "step": 27137
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6426745653152466,
      "learning_rate": 0.000579694704466131,
      "loss": 2.1469,
      "step": 27138
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6089933514595032,
      "learning_rate": 0.0005796932251067161,
      "loss": 2.4008,
      "step": 27139
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6043813824653625,
      "learning_rate": 0.0005796917456953009,
      "loss": 2.3555,
      "step": 27140
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6584099531173706,
      "learning_rate": 0.0005796902662318857,
      "loss": 2.4542,
      "step": 27141
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6242638826370239,
      "learning_rate": 0.0005796887867164709,
      "loss": 2.3656,
      "step": 27142
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.7407003045082092,
      "learning_rate": 0.0005796873071490567,
      "loss": 2.3219,
      "step": 27143
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6428921222686768,
      "learning_rate": 0.0005796858275296433,
      "loss": 2.2473,
      "step": 27144
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6544104218482971,
      "learning_rate": 0.000579684347858231,
      "loss": 2.1849,
      "step": 27145
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6725636720657349,
      "learning_rate": 0.0005796828681348201,
      "loss": 2.4502,
      "step": 27146
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.5858444571495056,
      "learning_rate": 0.000579681388359411,
      "loss": 2.1969,
      "step": 27147
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6946588754653931,
      "learning_rate": 0.0005796799085320037,
      "loss": 2.4774,
      "step": 27148
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.7327607870101929,
      "learning_rate": 0.0005796784286525987,
      "loss": 2.4098,
      "step": 27149
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.663752019405365,
      "learning_rate": 0.0005796769487211961,
      "loss": 2.4138,
      "step": 27150
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6472877860069275,
      "learning_rate": 0.0005796754687377963,
      "loss": 2.3287,
      "step": 27151
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.7262444496154785,
      "learning_rate": 0.0005796739887023997,
      "loss": 2.2817,
      "step": 27152
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6119592785835266,
      "learning_rate": 0.0005796725086150063,
      "loss": 2.4233,
      "step": 27153
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.8542931079864502,
      "learning_rate": 0.0005796710284756165,
      "loss": 2.4427,
      "step": 27154
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6601734757423401,
      "learning_rate": 0.0005796695482842306,
      "loss": 2.3839,
      "step": 27155
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6858461499214172,
      "learning_rate": 0.0005796680680408489,
      "loss": 2.275,
      "step": 27156
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6479769945144653,
      "learning_rate": 0.0005796665877454716,
      "loss": 2.5215,
      "step": 27157
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6219457983970642,
      "learning_rate": 0.0005796651073980989,
      "loss": 2.429,
      "step": 27158
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6379019021987915,
      "learning_rate": 0.0005796636269987313,
      "loss": 2.3909,
      "step": 27159
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.5951487421989441,
      "learning_rate": 0.000579662146547369,
      "loss": 2.5907,
      "step": 27160
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6420665979385376,
      "learning_rate": 0.0005796606660440121,
      "loss": 2.1853,
      "step": 27161
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.621105968952179,
      "learning_rate": 0.000579659185488661,
      "loss": 2.1381,
      "step": 27162
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6609212160110474,
      "learning_rate": 0.000579657704881316,
      "loss": 1.9772,
      "step": 27163
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6690093278884888,
      "learning_rate": 0.0005796562242219774,
      "loss": 2.4272,
      "step": 27164
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6960071325302124,
      "learning_rate": 0.0005796547435106455,
      "loss": 2.3394,
      "step": 27165
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6335780024528503,
      "learning_rate": 0.0005796532627473203,
      "loss": 2.2776,
      "step": 27166
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6330140233039856,
      "learning_rate": 0.0005796517819320025,
      "loss": 2.1888,
      "step": 27167
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.7924250960350037,
      "learning_rate": 0.000579650301064692,
      "loss": 2.3347,
      "step": 27168
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6378979086875916,
      "learning_rate": 0.0005796488201453892,
      "loss": 2.4562,
      "step": 27169
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.771767258644104,
      "learning_rate": 0.0005796473391740945,
      "loss": 2.157,
      "step": 27170
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6687121987342834,
      "learning_rate": 0.0005796458581508081,
      "loss": 2.2583,
      "step": 27171
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6667789816856384,
      "learning_rate": 0.0005796443770755302,
      "loss": 2.3011,
      "step": 27172
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.7261523008346558,
      "learning_rate": 0.0005796428959482612,
      "loss": 2.2281,
      "step": 27173
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6227521896362305,
      "learning_rate": 0.0005796414147690013,
      "loss": 2.1642,
      "step": 27174
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.5955706238746643,
      "learning_rate": 0.0005796399335377506,
      "loss": 2.3637,
      "step": 27175
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6729121804237366,
      "learning_rate": 0.0005796384522545097,
      "loss": 2.4421,
      "step": 27176
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.7027557492256165,
      "learning_rate": 0.0005796369709192788,
      "loss": 2.3671,
      "step": 27177
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6627798676490784,
      "learning_rate": 0.000579635489532058,
      "loss": 2.2154,
      "step": 27178
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6075406074523926,
      "learning_rate": 0.0005796340080928478,
      "loss": 2.1798,
      "step": 27179
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6380348801612854,
      "learning_rate": 0.0005796325266016483,
      "loss": 2.071,
      "step": 27180
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6135486364364624,
      "learning_rate": 0.0005796310450584597,
      "loss": 2.4927,
      "step": 27181
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6222361326217651,
      "learning_rate": 0.0005796295634632826,
      "loss": 2.2065,
      "step": 27182
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6772174835205078,
      "learning_rate": 0.000579628081816117,
      "loss": 2.4023,
      "step": 27183
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6605626344680786,
      "learning_rate": 0.0005796266001169633,
      "loss": 2.1339,
      "step": 27184
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6499691605567932,
      "learning_rate": 0.0005796251183658217,
      "loss": 2.1615,
      "step": 27185
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6661390066146851,
      "learning_rate": 0.0005796236365626925,
      "loss": 2.3097,
      "step": 27186
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6743067502975464,
      "learning_rate": 0.000579622154707576,
      "loss": 2.2373,
      "step": 27187
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6299710273742676,
      "learning_rate": 0.0005796206728004724,
      "loss": 2.3054,
      "step": 27188
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.7154704332351685,
      "learning_rate": 0.0005796191908413823,
      "loss": 2.2739,
      "step": 27189
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6509379148483276,
      "learning_rate": 0.0005796177088303055,
      "loss": 2.4156,
      "step": 27190
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6195226311683655,
      "learning_rate": 0.0005796162267672425,
      "loss": 2.2819,
      "step": 27191
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.591049313545227,
      "learning_rate": 0.0005796147446521935,
      "loss": 2.3625,
      "step": 27192
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6307036280632019,
      "learning_rate": 0.0005796132624851589,
      "loss": 2.3175,
      "step": 27193
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.674284040927887,
      "learning_rate": 0.0005796117802661391,
      "loss": 2.401,
      "step": 27194
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.629260778427124,
      "learning_rate": 0.0005796102979951339,
      "loss": 2.21,
      "step": 27195
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6990639567375183,
      "learning_rate": 0.0005796088156721441,
      "loss": 2.3587,
      "step": 27196
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6846395134925842,
      "learning_rate": 0.0005796073332971696,
      "loss": 2.1139,
      "step": 27197
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.7038827538490295,
      "learning_rate": 0.0005796058508702108,
      "loss": 2.3493,
      "step": 27198
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6204450130462646,
      "learning_rate": 0.0005796043683912681,
      "loss": 2.3605,
      "step": 27199
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6112281084060669,
      "learning_rate": 0.0005796028858603416,
      "loss": 2.3135,
      "step": 27200
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6273175477981567,
      "learning_rate": 0.0005796014032774316,
      "loss": 2.4134,
      "step": 27201
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6680084466934204,
      "learning_rate": 0.0005795999206425386,
      "loss": 2.4163,
      "step": 27202
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.5835078954696655,
      "learning_rate": 0.0005795984379556626,
      "loss": 2.3822,
      "step": 27203
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6155243515968323,
      "learning_rate": 0.0005795969552168039,
      "loss": 2.4083,
      "step": 27204
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.634220540523529,
      "learning_rate": 0.000579595472425963,
      "loss": 2.4175,
      "step": 27205
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.598709762096405,
      "learning_rate": 0.0005795939895831398,
      "loss": 2.281,
      "step": 27206
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.5874260663986206,
      "learning_rate": 0.000579592506688335,
      "loss": 2.2937,
      "step": 27207
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.631467878818512,
      "learning_rate": 0.0005795910237415486,
      "loss": 2.4718,
      "step": 27208
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6417511105537415,
      "learning_rate": 0.000579589540742781,
      "loss": 2.5004,
      "step": 27209
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.729871392250061,
      "learning_rate": 0.0005795880576920324,
      "loss": 2.6556,
      "step": 27210
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6927446126937866,
      "learning_rate": 0.000579586574589303,
      "loss": 2.33,
      "step": 27211
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.5751376152038574,
      "learning_rate": 0.0005795850914345934,
      "loss": 2.2827,
      "step": 27212
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6695142984390259,
      "learning_rate": 0.0005795836082279034,
      "loss": 2.1288,
      "step": 27213
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6410514116287231,
      "learning_rate": 0.0005795821249692337,
      "loss": 2.3765,
      "step": 27214
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.5710915327072144,
      "learning_rate": 0.0005795806416585843,
      "loss": 2.2968,
      "step": 27215
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6582640409469604,
      "learning_rate": 0.0005795791582959558,
      "loss": 2.4141,
      "step": 27216
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6971249580383301,
      "learning_rate": 0.000579577674881348,
      "loss": 2.4034,
      "step": 27217
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.5913313031196594,
      "learning_rate": 0.0005795761914147616,
      "loss": 2.2485,
      "step": 27218
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6202163696289062,
      "learning_rate": 0.0005795747078961965,
      "loss": 2.514,
      "step": 27219
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6772646903991699,
      "learning_rate": 0.0005795732243256534,
      "loss": 2.256,
      "step": 27220
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6527252197265625,
      "learning_rate": 0.0005795717407031323,
      "loss": 2.2942,
      "step": 27221
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6635414958000183,
      "learning_rate": 0.0005795702570286335,
      "loss": 2.3991,
      "step": 27222
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6469273567199707,
      "learning_rate": 0.0005795687733021573,
      "loss": 2.3031,
      "step": 27223
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.7133957743644714,
      "learning_rate": 0.0005795672895237041,
      "loss": 2.5338,
      "step": 27224
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6627176403999329,
      "learning_rate": 0.0005795658056932739,
      "loss": 2.3183,
      "step": 27225
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6419790387153625,
      "learning_rate": 0.0005795643218108672,
      "loss": 2.2482,
      "step": 27226
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.7908522486686707,
      "learning_rate": 0.0005795628378764843,
      "loss": 2.6723,
      "step": 27227
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6387324333190918,
      "learning_rate": 0.0005795613538901252,
      "loss": 2.3497,
      "step": 27228
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6676527261734009,
      "learning_rate": 0.0005795598698517906,
      "loss": 2.2375,
      "step": 27229
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.728369951248169,
      "learning_rate": 0.0005795583857614805,
      "loss": 2.2104,
      "step": 27230
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6261076927185059,
      "learning_rate": 0.000579556901619195,
      "loss": 2.3187,
      "step": 27231
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6027548909187317,
      "learning_rate": 0.0005795554174249348,
      "loss": 2.418,
      "step": 27232
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6037859916687012,
      "learning_rate": 0.0005795539331786999,
      "loss": 2.2208,
      "step": 27233
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.7262136340141296,
      "learning_rate": 0.0005795524488804906,
      "loss": 2.4754,
      "step": 27234
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6186352968215942,
      "learning_rate": 0.0005795509645303074,
      "loss": 2.5156,
      "step": 27235
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6692588329315186,
      "learning_rate": 0.0005795494801281503,
      "loss": 2.2789,
      "step": 27236
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6130773425102234,
      "learning_rate": 0.0005795479956740196,
      "loss": 2.4816,
      "step": 27237
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6218245625495911,
      "learning_rate": 0.0005795465111679157,
      "loss": 2.1732,
      "step": 27238
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6902756690979004,
      "learning_rate": 0.0005795450266098389,
      "loss": 2.2851,
      "step": 27239
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.670657753944397,
      "learning_rate": 0.0005795435419997894,
      "loss": 2.3025,
      "step": 27240
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6462356448173523,
      "learning_rate": 0.0005795420573377674,
      "loss": 2.3912,
      "step": 27241
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.7123387455940247,
      "learning_rate": 0.0005795405726237733,
      "loss": 2.3871,
      "step": 27242
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6219972968101501,
      "learning_rate": 0.0005795390878578072,
      "loss": 2.388,
      "step": 27243
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6571629047393799,
      "learning_rate": 0.0005795376030398696,
      "loss": 2.2496,
      "step": 27244
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6521808505058289,
      "learning_rate": 0.0005795361181699608,
      "loss": 2.2108,
      "step": 27245
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.7161310911178589,
      "learning_rate": 0.0005795346332480809,
      "loss": 2.4275,
      "step": 27246
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6735745072364807,
      "learning_rate": 0.0005795331482742302,
      "loss": 2.436,
      "step": 27247
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.8371105194091797,
      "learning_rate": 0.000579531663248409,
      "loss": 2.4373,
      "step": 27248
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6234657764434814,
      "learning_rate": 0.0005795301781706176,
      "loss": 2.2329,
      "step": 27249
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.7424442768096924,
      "learning_rate": 0.0005795286930408562,
      "loss": 2.4421,
      "step": 27250
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.7109754681587219,
      "learning_rate": 0.0005795272078591251,
      "loss": 2.4053,
      "step": 27251
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.7301962375640869,
      "learning_rate": 0.0005795257226254248,
      "loss": 2.3307,
      "step": 27252
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.7715457081794739,
      "learning_rate": 0.0005795242373397553,
      "loss": 2.1891,
      "step": 27253
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6395998597145081,
      "learning_rate": 0.0005795227520021169,
      "loss": 2.3283,
      "step": 27254
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.8088030815124512,
      "learning_rate": 0.0005795212666125101,
      "loss": 2.2516,
      "step": 27255
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.7655753493309021,
      "learning_rate": 0.0005795197811709349,
      "loss": 2.4095,
      "step": 27256
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6492000818252563,
      "learning_rate": 0.0005795182956773916,
      "loss": 2.0519,
      "step": 27257
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6924058198928833,
      "learning_rate": 0.0005795168101318808,
      "loss": 2.2932,
      "step": 27258
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6308773159980774,
      "learning_rate": 0.0005795153245344025,
      "loss": 1.984,
      "step": 27259
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6239627003669739,
      "learning_rate": 0.000579513838884957,
      "loss": 2.2143,
      "step": 27260
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6154724359512329,
      "learning_rate": 0.0005795123531835446,
      "loss": 2.2389,
      "step": 27261
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6022526025772095,
      "learning_rate": 0.0005795108674301656,
      "loss": 2.3394,
      "step": 27262
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6572718620300293,
      "learning_rate": 0.0005795093816248201,
      "loss": 2.3621,
      "step": 27263
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6634010672569275,
      "learning_rate": 0.0005795078957675087,
      "loss": 2.3404,
      "step": 27264
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6700526475906372,
      "learning_rate": 0.0005795064098582314,
      "loss": 2.6353,
      "step": 27265
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.5908690094947815,
      "learning_rate": 0.0005795049238969886,
      "loss": 2.5392,
      "step": 27266
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6414212584495544,
      "learning_rate": 0.0005795034378837805,
      "loss": 2.198,
      "step": 27267
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6570290923118591,
      "learning_rate": 0.0005795019518186076,
      "loss": 2.2414,
      "step": 27268
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6173003911972046,
      "learning_rate": 0.0005795004657014699,
      "loss": 2.1097,
      "step": 27269
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.675977885723114,
      "learning_rate": 0.0005794989795323679,
      "loss": 2.2816,
      "step": 27270
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.673658013343811,
      "learning_rate": 0.0005794974933113016,
      "loss": 2.4187,
      "step": 27271
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6451579332351685,
      "learning_rate": 0.0005794960070382716,
      "loss": 2.2184,
      "step": 27272
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6769396662712097,
      "learning_rate": 0.0005794945207132779,
      "loss": 2.3227,
      "step": 27273
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.701572835445404,
      "learning_rate": 0.0005794930343363209,
      "loss": 2.2692,
      "step": 27274
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.7678862810134888,
      "learning_rate": 0.0005794915479074009,
      "loss": 2.4279,
      "step": 27275
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6075178384780884,
      "learning_rate": 0.0005794900614265182,
      "loss": 2.3139,
      "step": 27276
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.7042283415794373,
      "learning_rate": 0.0005794885748936729,
      "loss": 2.4141,
      "step": 27277
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.660041332244873,
      "learning_rate": 0.0005794870883088656,
      "loss": 2.4658,
      "step": 27278
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.7266327738761902,
      "learning_rate": 0.0005794856016720961,
      "loss": 2.4173,
      "step": 27279
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6239683032035828,
      "learning_rate": 0.0005794841149833652,
      "loss": 2.367,
      "step": 27280
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6311590671539307,
      "learning_rate": 0.0005794826282426728,
      "loss": 2.2109,
      "step": 27281
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6094640493392944,
      "learning_rate": 0.0005794811414500192,
      "loss": 2.4693,
      "step": 27282
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6365195512771606,
      "learning_rate": 0.000579479654605405,
      "loss": 2.2447,
      "step": 27283
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.5979081392288208,
      "learning_rate": 0.0005794781677088301,
      "loss": 2.1769,
      "step": 27284
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.664257287979126,
      "learning_rate": 0.000579476680760295,
      "loss": 2.4166,
      "step": 27285
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6914960145950317,
      "learning_rate": 0.0005794751937598,
      "loss": 2.3975,
      "step": 27286
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6771700382232666,
      "learning_rate": 0.0005794737067073452,
      "loss": 2.3677,
      "step": 27287
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6472541093826294,
      "learning_rate": 0.0005794722196029309,
      "loss": 2.4503,
      "step": 27288
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6923065185546875,
      "learning_rate": 0.0005794707324465575,
      "loss": 2.7374,
      "step": 27289
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6246362924575806,
      "learning_rate": 0.0005794692452382252,
      "loss": 2.3784,
      "step": 27290
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6040433049201965,
      "learning_rate": 0.0005794677579779343,
      "loss": 2.382,
      "step": 27291
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6805804967880249,
      "learning_rate": 0.000579466270665685,
      "loss": 2.0454,
      "step": 27292
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6454419493675232,
      "learning_rate": 0.0005794647833014777,
      "loss": 2.365,
      "step": 27293
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6479642987251282,
      "learning_rate": 0.0005794632958853127,
      "loss": 2.4305,
      "step": 27294
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6252209544181824,
      "learning_rate": 0.00057946180841719,
      "loss": 2.4756,
      "step": 27295
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.5172578692436218,
      "learning_rate": 0.0005794603208971103,
      "loss": 2.0909,
      "step": 27296
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6869930028915405,
      "learning_rate": 0.0005794588333250736,
      "loss": 2.5126,
      "step": 27297
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6344983577728271,
      "learning_rate": 0.0005794573457010802,
      "loss": 2.3452,
      "step": 27298
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6579793691635132,
      "learning_rate": 0.0005794558580251303,
      "loss": 2.3446,
      "step": 27299
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.64863121509552,
      "learning_rate": 0.0005794543702972244,
      "loss": 2.3874,
      "step": 27300
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.5996944308280945,
      "learning_rate": 0.0005794528825173626,
      "loss": 2.4602,
      "step": 27301
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6212192177772522,
      "learning_rate": 0.0005794513946855453,
      "loss": 2.3383,
      "step": 27302
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6019037961959839,
      "learning_rate": 0.0005794499068017726,
      "loss": 2.5249,
      "step": 27303
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.5933734774589539,
      "learning_rate": 0.000579448418866045,
      "loss": 2.2473,
      "step": 27304
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6143445372581482,
      "learning_rate": 0.0005794469308783626,
      "loss": 2.2939,
      "step": 27305
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6297267079353333,
      "learning_rate": 0.0005794454428387258,
      "loss": 2.3009,
      "step": 27306
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6526179313659668,
      "learning_rate": 0.0005794439547471348,
      "loss": 2.3586,
      "step": 27307
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6401427984237671,
      "learning_rate": 0.0005794424666035899,
      "loss": 2.2958,
      "step": 27308
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6548711061477661,
      "learning_rate": 0.0005794409784080913,
      "loss": 2.4512,
      "step": 27309
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.641355574131012,
      "learning_rate": 0.0005794394901606394,
      "loss": 2.3196,
      "step": 27310
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6643667817115784,
      "learning_rate": 0.0005794380018612344,
      "loss": 2.1286,
      "step": 27311
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.5941593050956726,
      "learning_rate": 0.0005794365135098766,
      "loss": 2.372,
      "step": 27312
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6318778395652771,
      "learning_rate": 0.0005794350251065663,
      "loss": 2.3026,
      "step": 27313
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6544882655143738,
      "learning_rate": 0.0005794335366513038,
      "loss": 2.4707,
      "step": 27314
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6123448610305786,
      "learning_rate": 0.0005794320481440893,
      "loss": 2.3237,
      "step": 27315
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.7000053524971008,
      "learning_rate": 0.000579430559584923,
      "loss": 2.3987,
      "step": 27316
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.653595507144928,
      "learning_rate": 0.0005794290709738054,
      "loss": 2.28,
      "step": 27317
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6893277168273926,
      "learning_rate": 0.0005794275823107366,
      "loss": 2.2728,
      "step": 27318
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.591360330581665,
      "learning_rate": 0.000579426093595717,
      "loss": 2.4102,
      "step": 27319
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6517922282218933,
      "learning_rate": 0.0005794246048287468,
      "loss": 2.2767,
      "step": 27320
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6777656078338623,
      "learning_rate": 0.0005794231160098262,
      "loss": 2.3519,
      "step": 27321
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6238040328025818,
      "learning_rate": 0.0005794216271389557,
      "loss": 2.2372,
      "step": 27322
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6125339269638062,
      "learning_rate": 0.0005794201382161353,
      "loss": 2.524,
      "step": 27323
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.5876415371894836,
      "learning_rate": 0.0005794186492413656,
      "loss": 2.2207,
      "step": 27324
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6151682138442993,
      "learning_rate": 0.0005794171602146465,
      "loss": 2.1548,
      "step": 27325
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6498314738273621,
      "learning_rate": 0.0005794156711359787,
      "loss": 2.262,
      "step": 27326
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6146642565727234,
      "learning_rate": 0.0005794141820053621,
      "loss": 2.2511,
      "step": 27327
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.612890899181366,
      "learning_rate": 0.0005794126928227971,
      "loss": 2.5674,
      "step": 27328
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6258560419082642,
      "learning_rate": 0.0005794112035882841,
      "loss": 2.3443,
      "step": 27329
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.5899704694747925,
      "learning_rate": 0.0005794097143018233,
      "loss": 2.4554,
      "step": 27330
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6217465400695801,
      "learning_rate": 0.0005794082249634149,
      "loss": 2.2798,
      "step": 27331
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6450929641723633,
      "learning_rate": 0.0005794067355730593,
      "loss": 2.2059,
      "step": 27332
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6647964715957642,
      "learning_rate": 0.0005794052461307566,
      "loss": 2.2909,
      "step": 27333
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.5929965972900391,
      "learning_rate": 0.0005794037566365073,
      "loss": 2.4361,
      "step": 27334
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.5878216028213501,
      "learning_rate": 0.0005794022670903115,
      "loss": 2.2344,
      "step": 27335
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.5881213545799255,
      "learning_rate": 0.0005794007774921695,
      "loss": 2.184,
      "step": 27336
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6475085616111755,
      "learning_rate": 0.0005793992878420817,
      "loss": 2.3325,
      "step": 27337
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6331186890602112,
      "learning_rate": 0.0005793977981400484,
      "loss": 2.195,
      "step": 27338
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6003205180168152,
      "learning_rate": 0.0005793963083860696,
      "loss": 2.4364,
      "step": 27339
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.644684374332428,
      "learning_rate": 0.0005793948185801458,
      "loss": 2.4974,
      "step": 27340
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.5996277332305908,
      "learning_rate": 0.0005793933287222774,
      "loss": 2.254,
      "step": 27341
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.5975160598754883,
      "learning_rate": 0.0005793918388124642,
      "loss": 2.2274,
      "step": 27342
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.7080497145652771,
      "learning_rate": 0.0005793903488507069,
      "loss": 2.4072,
      "step": 27343
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6270186305046082,
      "learning_rate": 0.0005793888588370058,
      "loss": 2.3178,
      "step": 27344
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.5541966557502747,
      "learning_rate": 0.0005793873687713609,
      "loss": 2.1114,
      "step": 27345
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6481489539146423,
      "learning_rate": 0.0005793858786537726,
      "loss": 2.356,
      "step": 27346
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.625220537185669,
      "learning_rate": 0.0005793843884842413,
      "loss": 2.1964,
      "step": 27347
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6599239110946655,
      "learning_rate": 0.000579382898262767,
      "loss": 2.4283,
      "step": 27348
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.5837617516517639,
      "learning_rate": 0.0005793814079893502,
      "loss": 2.2176,
      "step": 27349
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.7611891031265259,
      "learning_rate": 0.0005793799176639913,
      "loss": 2.3412,
      "step": 27350
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6322768926620483,
      "learning_rate": 0.0005793784272866903,
      "loss": 2.1487,
      "step": 27351
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6185305714607239,
      "learning_rate": 0.0005793769368574476,
      "loss": 2.3335,
      "step": 27352
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.806800901889801,
      "learning_rate": 0.0005793754463762633,
      "loss": 2.3683,
      "step": 27353
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6799511313438416,
      "learning_rate": 0.0005793739558431379,
      "loss": 2.152,
      "step": 27354
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6307477355003357,
      "learning_rate": 0.0005793724652580717,
      "loss": 2.0827,
      "step": 27355
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.7302381992340088,
      "learning_rate": 0.0005793709746210648,
      "loss": 2.1739,
      "step": 27356
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.894269585609436,
      "learning_rate": 0.0005793694839321176,
      "loss": 2.3137,
      "step": 27357
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6087737679481506,
      "learning_rate": 0.0005793679931912303,
      "loss": 2.409,
      "step": 27358
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6766271591186523,
      "learning_rate": 0.0005793665023984033,
      "loss": 2.1572,
      "step": 27359
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6784675121307373,
      "learning_rate": 0.0005793650115536367,
      "loss": 2.3559,
      "step": 27360
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6786543130874634,
      "learning_rate": 0.0005793635206569309,
      "loss": 2.1085,
      "step": 27361
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6173014044761658,
      "learning_rate": 0.0005793620297082862,
      "loss": 2.3625,
      "step": 27362
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6981422305107117,
      "learning_rate": 0.0005793605387077027,
      "loss": 2.4339,
      "step": 27363
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6911876797676086,
      "learning_rate": 0.0005793590476551809,
      "loss": 2.0666,
      "step": 27364
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.639214038848877,
      "learning_rate": 0.000579357556550721,
      "loss": 2.4241,
      "step": 27365
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6068812608718872,
      "learning_rate": 0.0005793560653943233,
      "loss": 2.3136,
      "step": 27366
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.0303928852081299,
      "learning_rate": 0.0005793545741859878,
      "loss": 2.2116,
      "step": 27367
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6639132499694824,
      "learning_rate": 0.0005793530829257152,
      "loss": 2.3544,
      "step": 27368
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6829387545585632,
      "learning_rate": 0.0005793515916135057,
      "loss": 2.2613,
      "step": 27369
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.0954222679138184,
      "learning_rate": 0.0005793501002493592,
      "loss": 2.0128,
      "step": 27370
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.5774264931678772,
      "learning_rate": 0.0005793486088332764,
      "loss": 2.1865,
      "step": 27371
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.7562294006347656,
      "learning_rate": 0.0005793471173652574,
      "loss": 2.2141,
      "step": 27372
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.907879114151001,
      "learning_rate": 0.0005793456258453025,
      "loss": 2.1496,
      "step": 27373
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6215454936027527,
      "learning_rate": 0.000579344134273412,
      "loss": 2.2255,
      "step": 27374
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6230800747871399,
      "learning_rate": 0.000579342642649586,
      "loss": 2.4417,
      "step": 27375
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.8267137408256531,
      "learning_rate": 0.0005793411509738251,
      "loss": 2.2272,
      "step": 27376
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6697019338607788,
      "learning_rate": 0.0005793396592461294,
      "loss": 2.3414,
      "step": 27377
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6389238238334656,
      "learning_rate": 0.000579338167466499,
      "loss": 2.4101,
      "step": 27378
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6721838116645813,
      "learning_rate": 0.0005793366756349346,
      "loss": 2.105,
      "step": 27379
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.7659922242164612,
      "learning_rate": 0.0005793351837514361,
      "loss": 2.2227,
      "step": 27380
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6382683515548706,
      "learning_rate": 0.0005793336918160039,
      "loss": 2.3706,
      "step": 27381
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.7680817246437073,
      "learning_rate": 0.0005793321998286383,
      "loss": 2.4292,
      "step": 27382
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6016711592674255,
      "learning_rate": 0.0005793307077893397,
      "loss": 2.3497,
      "step": 27383
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6017640233039856,
      "learning_rate": 0.0005793292156981082,
      "loss": 2.1379,
      "step": 27384
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6532878875732422,
      "learning_rate": 0.0005793277235549441,
      "loss": 2.2778,
      "step": 27385
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.586926281452179,
      "learning_rate": 0.0005793262313598476,
      "loss": 2.4856,
      "step": 27386
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6970362067222595,
      "learning_rate": 0.0005793247391128192,
      "loss": 2.2558,
      "step": 27387
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.7882223725318909,
      "learning_rate": 0.000579323246813859,
      "loss": 2.3821,
      "step": 27388
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6377875804901123,
      "learning_rate": 0.0005793217544629674,
      "loss": 2.3344,
      "step": 27389
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.5699254870414734,
      "learning_rate": 0.0005793202620601447,
      "loss": 2.3674,
      "step": 27390
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6278435587882996,
      "learning_rate": 0.0005793187696053908,
      "loss": 2.221,
      "step": 27391
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6447164416313171,
      "learning_rate": 0.0005793172770987065,
      "loss": 2.1583,
      "step": 27392
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6262601613998413,
      "learning_rate": 0.0005793157845400918,
      "loss": 2.0245,
      "step": 27393
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6204991340637207,
      "learning_rate": 0.000579314291929547,
      "loss": 2.3414,
      "step": 27394
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6326727271080017,
      "learning_rate": 0.0005793127992670725,
      "loss": 2.2302,
      "step": 27395
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6532077193260193,
      "learning_rate": 0.0005793113065526683,
      "loss": 2.3618,
      "step": 27396
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6709433197975159,
      "learning_rate": 0.0005793098137863349,
      "loss": 2.1273,
      "step": 27397
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6686950922012329,
      "learning_rate": 0.0005793083209680727,
      "loss": 2.3168,
      "step": 27398
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6335768103599548,
      "learning_rate": 0.0005793068280978816,
      "loss": 2.3099,
      "step": 27399
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.8038686513900757,
      "learning_rate": 0.0005793053351757622,
      "loss": 2.3774,
      "step": 27400
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6242024302482605,
      "learning_rate": 0.0005793038422017146,
      "loss": 2.3326,
      "step": 27401
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6386765837669373,
      "learning_rate": 0.0005793023491757393,
      "loss": 2.2233,
      "step": 27402
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6621774435043335,
      "learning_rate": 0.0005793008560978362,
      "loss": 2.3142,
      "step": 27403
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.613655686378479,
      "learning_rate": 0.0005792993629680059,
      "loss": 2.233,
      "step": 27404
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6086425185203552,
      "learning_rate": 0.0005792978697862486,
      "loss": 2.43,
      "step": 27405
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6850014925003052,
      "learning_rate": 0.0005792963765525645,
      "loss": 2.4283,
      "step": 27406
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6434251666069031,
      "learning_rate": 0.000579294883266954,
      "loss": 2.4054,
      "step": 27407
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.7113127708435059,
      "learning_rate": 0.0005792933899294172,
      "loss": 2.4285,
      "step": 27408
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6711985468864441,
      "learning_rate": 0.0005792918965399547,
      "loss": 2.5433,
      "step": 27409
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6702861189842224,
      "learning_rate": 0.0005792904030985665,
      "loss": 2.3271,
      "step": 27410
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6761139035224915,
      "learning_rate": 0.0005792889096052527,
      "loss": 2.4689,
      "step": 27411
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6172769665718079,
      "learning_rate": 0.0005792874160600141,
      "loss": 2.286,
      "step": 27412
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6361299157142639,
      "learning_rate": 0.0005792859224628505,
      "loss": 2.2711,
      "step": 27413
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6300633549690247,
      "learning_rate": 0.0005792844288137625,
      "loss": 2.4911,
      "step": 27414
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6742219924926758,
      "learning_rate": 0.0005792829351127502,
      "loss": 2.2716,
      "step": 27415
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.5993929505348206,
      "learning_rate": 0.0005792814413598139,
      "loss": 2.093,
      "step": 27416
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6813771724700928,
      "learning_rate": 0.000579279947554954,
      "loss": 2.1453,
      "step": 27417
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6313143968582153,
      "learning_rate": 0.0005792784536981706,
      "loss": 2.483,
      "step": 27418
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6360476613044739,
      "learning_rate": 0.0005792769597894641,
      "loss": 2.3619,
      "step": 27419
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.642966091632843,
      "learning_rate": 0.0005792754658288348,
      "loss": 2.473,
      "step": 27420
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6978209018707275,
      "learning_rate": 0.0005792739718162829,
      "loss": 2.5592,
      "step": 27421
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6859345436096191,
      "learning_rate": 0.0005792724777518085,
      "loss": 2.5569,
      "step": 27422
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6070441007614136,
      "learning_rate": 0.0005792709836354124,
      "loss": 2.1044,
      "step": 27423
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6163944005966187,
      "learning_rate": 0.0005792694894670943,
      "loss": 2.441,
      "step": 27424
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.7198411226272583,
      "learning_rate": 0.0005792679952468548,
      "loss": 2.5309,
      "step": 27425
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6402919888496399,
      "learning_rate": 0.0005792665009746942,
      "loss": 2.3024,
      "step": 27426
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6629758477210999,
      "learning_rate": 0.0005792650066506126,
      "loss": 2.1644,
      "step": 27427
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6867009997367859,
      "learning_rate": 0.0005792635122746103,
      "loss": 2.2641,
      "step": 27428
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6261088252067566,
      "learning_rate": 0.0005792620178466878,
      "loss": 2.6724,
      "step": 27429
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6765959858894348,
      "learning_rate": 0.0005792605233668451,
      "loss": 2.444,
      "step": 27430
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6756187677383423,
      "learning_rate": 0.0005792590288350827,
      "loss": 2.3433,
      "step": 27431
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6472021341323853,
      "learning_rate": 0.0005792575342514006,
      "loss": 2.2872,
      "step": 27432
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.693987250328064,
      "learning_rate": 0.0005792560396157994,
      "loss": 2.4273,
      "step": 27433
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6396238207817078,
      "learning_rate": 0.0005792545449282792,
      "loss": 2.2748,
      "step": 27434
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6753979921340942,
      "learning_rate": 0.0005792530501888403,
      "loss": 2.3071,
      "step": 27435
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6952250003814697,
      "learning_rate": 0.000579251555397483,
      "loss": 2.1393,
      "step": 27436
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.7128084897994995,
      "learning_rate": 0.0005792500605542076,
      "loss": 2.3022,
      "step": 27437
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6658620238304138,
      "learning_rate": 0.0005792485656590143,
      "loss": 2.2006,
      "step": 27438
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6865856647491455,
      "learning_rate": 0.0005792470707119033,
      "loss": 2.2328,
      "step": 27439
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6176879405975342,
      "learning_rate": 0.0005792455757128752,
      "loss": 2.453,
      "step": 27440
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.5904578566551208,
      "learning_rate": 0.00057924408066193,
      "loss": 2.3137,
      "step": 27441
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6421430110931396,
      "learning_rate": 0.000579242585559068,
      "loss": 2.3164,
      "step": 27442
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6888300776481628,
      "learning_rate": 0.0005792410904042896,
      "loss": 2.2476,
      "step": 27443
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.853138267993927,
      "learning_rate": 0.0005792395951975949,
      "loss": 1.9929,
      "step": 27444
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6681101322174072,
      "learning_rate": 0.0005792380999389844,
      "loss": 2.3535,
      "step": 27445
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6975663304328918,
      "learning_rate": 0.0005792366046284583,
      "loss": 2.1874,
      "step": 27446
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6437574028968811,
      "learning_rate": 0.0005792351092660168,
      "loss": 2.251,
      "step": 27447
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6299781203269958,
      "learning_rate": 0.0005792336138516601,
      "loss": 2.3469,
      "step": 27448
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6829551458358765,
      "learning_rate": 0.0005792321183853888,
      "loss": 2.2599,
      "step": 27449
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6740782260894775,
      "learning_rate": 0.0005792306228672028,
      "loss": 2.2136,
      "step": 27450
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6048629879951477,
      "learning_rate": 0.0005792291272971027,
      "loss": 2.3289,
      "step": 27451
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6773272752761841,
      "learning_rate": 0.0005792276316750886,
      "loss": 2.3044,
      "step": 27452
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6696779727935791,
      "learning_rate": 0.0005792261360011607,
      "loss": 2.2603,
      "step": 27453
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6617149114608765,
      "learning_rate": 0.0005792246402753196,
      "loss": 2.3116,
      "step": 27454
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6719552874565125,
      "learning_rate": 0.0005792231444975652,
      "loss": 2.0332,
      "step": 27455
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6715546250343323,
      "learning_rate": 0.000579221648667898,
      "loss": 2.3215,
      "step": 27456
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.7140885591506958,
      "learning_rate": 0.0005792201527863182,
      "loss": 2.3066,
      "step": 27457
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6262444853782654,
      "learning_rate": 0.0005792186568528261,
      "loss": 2.4055,
      "step": 27458
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.7314510345458984,
      "learning_rate": 0.000579217160867422,
      "loss": 2.5002,
      "step": 27459
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.7749486565589905,
      "learning_rate": 0.0005792156648301062,
      "loss": 2.2703,
      "step": 27460
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6317622065544128,
      "learning_rate": 0.0005792141687408789,
      "loss": 2.3097,
      "step": 27461
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.5596126914024353,
      "learning_rate": 0.0005792126725997404,
      "loss": 2.2974,
      "step": 27462
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6458618640899658,
      "learning_rate": 0.0005792111764066911,
      "loss": 2.258,
      "step": 27463
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.676953136920929,
      "learning_rate": 0.000579209680161731,
      "loss": 2.4247,
      "step": 27464
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6454353332519531,
      "learning_rate": 0.0005792081838648606,
      "loss": 2.2891,
      "step": 27465
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6474463939666748,
      "learning_rate": 0.0005792066875160802,
      "loss": 2.1712,
      "step": 27466
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6053210496902466,
      "learning_rate": 0.00057920519111539,
      "loss": 2.292,
      "step": 27467
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.709104597568512,
      "learning_rate": 0.0005792036946627902,
      "loss": 2.3401,
      "step": 27468
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6183823943138123,
      "learning_rate": 0.0005792021981582812,
      "loss": 2.2049,
      "step": 27469
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6496178507804871,
      "learning_rate": 0.0005792007016018633,
      "loss": 2.442,
      "step": 27470
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6937959790229797,
      "learning_rate": 0.0005791992049935367,
      "loss": 2.0997,
      "step": 27471
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.7178338766098022,
      "learning_rate": 0.0005791977083333017,
      "loss": 2.297,
      "step": 27472
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.7765084505081177,
      "learning_rate": 0.0005791962116211585,
      "loss": 2.3128,
      "step": 27473
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.7133308053016663,
      "learning_rate": 0.0005791947148571074,
      "loss": 2.2017,
      "step": 27474
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.9209576845169067,
      "learning_rate": 0.0005791932180411489,
      "loss": 2.4332,
      "step": 27475
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6103322505950928,
      "learning_rate": 0.000579191721173283,
      "loss": 2.3818,
      "step": 27476
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.7942706346511841,
      "learning_rate": 0.0005791902242535102,
      "loss": 2.1546,
      "step": 27477
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.77601557970047,
      "learning_rate": 0.0005791887272818306,
      "loss": 2.2827,
      "step": 27478
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.7146576046943665,
      "learning_rate": 0.0005791872302582445,
      "loss": 2.3869,
      "step": 27479
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.7812102437019348,
      "learning_rate": 0.0005791857331827522,
      "loss": 2.3445,
      "step": 27480
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6537201404571533,
      "learning_rate": 0.0005791842360553542,
      "loss": 2.2213,
      "step": 27481
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.5666338205337524,
      "learning_rate": 0.0005791827388760504,
      "loss": 2.3378,
      "step": 27482
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6890388131141663,
      "learning_rate": 0.0005791812416448413,
      "loss": 2.1171,
      "step": 27483
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6623342037200928,
      "learning_rate": 0.0005791797443617272,
      "loss": 2.2987,
      "step": 27484
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.7399567365646362,
      "learning_rate": 0.0005791782470267083,
      "loss": 2.3951,
      "step": 27485
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6869868040084839,
      "learning_rate": 0.0005791767496397847,
      "loss": 2.3843,
      "step": 27486
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6717659831047058,
      "learning_rate": 0.0005791752522009571,
      "loss": 2.2492,
      "step": 27487
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6873297691345215,
      "learning_rate": 0.0005791737547102255,
      "loss": 2.3617,
      "step": 27488
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6101332902908325,
      "learning_rate": 0.0005791722571675903,
      "loss": 2.3469,
      "step": 27489
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.7097792625427246,
      "learning_rate": 0.0005791707595730515,
      "loss": 2.2507,
      "step": 27490
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6540156602859497,
      "learning_rate": 0.0005791692619266098,
      "loss": 2.4197,
      "step": 27491
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6604256629943848,
      "learning_rate": 0.0005791677642282651,
      "loss": 2.3787,
      "step": 27492
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6287246942520142,
      "learning_rate": 0.0005791662664780179,
      "loss": 2.2881,
      "step": 27493
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6195764541625977,
      "learning_rate": 0.0005791647686758685,
      "loss": 2.5567,
      "step": 27494
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6158549785614014,
      "learning_rate": 0.000579163270821817,
      "loss": 2.2689,
      "step": 27495
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.655959963798523,
      "learning_rate": 0.0005791617729158637,
      "loss": 2.3631,
      "step": 27496
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6714302897453308,
      "learning_rate": 0.0005791602749580092,
      "loss": 2.2177,
      "step": 27497
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6865556240081787,
      "learning_rate": 0.0005791587769482533,
      "loss": 2.4784,
      "step": 27498
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.8030230402946472,
      "learning_rate": 0.0005791572788865967,
      "loss": 2.1029,
      "step": 27499
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6689445376396179,
      "learning_rate": 0.0005791557807730394,
      "loss": 2.2748,
      "step": 27500
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6812207698822021,
      "learning_rate": 0.0005791542826075818,
      "loss": 2.3098,
      "step": 27501
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6051505208015442,
      "learning_rate": 0.000579152784390224,
      "loss": 2.3382,
      "step": 27502
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6418469548225403,
      "learning_rate": 0.0005791512861209666,
      "loss": 2.2082,
      "step": 27503
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.7410216927528381,
      "learning_rate": 0.0005791497877998097,
      "loss": 2.312,
      "step": 27504
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6398460268974304,
      "learning_rate": 0.0005791482894267535,
      "loss": 2.1522,
      "step": 27505
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.7311972379684448,
      "learning_rate": 0.0005791467910017985,
      "loss": 2.2304,
      "step": 27506
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6760526299476624,
      "learning_rate": 0.0005791452925249447,
      "loss": 2.1195,
      "step": 27507
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6551665663719177,
      "learning_rate": 0.0005791437939961926,
      "loss": 2.1561,
      "step": 27508
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6177725791931152,
      "learning_rate": 0.0005791422954155423,
      "loss": 2.3246,
      "step": 27509
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6446384191513062,
      "learning_rate": 0.0005791407967829944,
      "loss": 2.5428,
      "step": 27510
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6522995233535767,
      "learning_rate": 0.0005791392980985488,
      "loss": 2.3431,
      "step": 27511
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6350517868995667,
      "learning_rate": 0.0005791377993622058,
      "loss": 2.4416,
      "step": 27512
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6106560826301575,
      "learning_rate": 0.0005791363005739661,
      "loss": 2.1283,
      "step": 27513
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6976733207702637,
      "learning_rate": 0.0005791348017338294,
      "loss": 2.284,
      "step": 27514
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6608991026878357,
      "learning_rate": 0.0005791333028417965,
      "loss": 2.0674,
      "step": 27515
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6177860498428345,
      "learning_rate": 0.0005791318038978674,
      "loss": 2.3165,
      "step": 27516
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6224508285522461,
      "learning_rate": 0.0005791303049020424,
      "loss": 2.0471,
      "step": 27517
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6073564887046814,
      "learning_rate": 0.0005791288058543218,
      "loss": 2.504,
      "step": 27518
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.7014425992965698,
      "learning_rate": 0.0005791273067547059,
      "loss": 2.5119,
      "step": 27519
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6966930031776428,
      "learning_rate": 0.0005791258076031949,
      "loss": 2.0974,
      "step": 27520
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6678856015205383,
      "learning_rate": 0.0005791243083997893,
      "loss": 2.4464,
      "step": 27521
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.7543789148330688,
      "learning_rate": 0.0005791228091444891,
      "loss": 2.2837,
      "step": 27522
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6794644594192505,
      "learning_rate": 0.0005791213098372946,
      "loss": 2.0749,
      "step": 27523
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6549012660980225,
      "learning_rate": 0.0005791198104782064,
      "loss": 2.2657,
      "step": 27524
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.7194657921791077,
      "learning_rate": 0.0005791183110672245,
      "loss": 2.2675,
      "step": 27525
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.832349419593811,
      "learning_rate": 0.0005791168116043491,
      "loss": 2.3017,
      "step": 27526
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6952117681503296,
      "learning_rate": 0.0005791153120895808,
      "loss": 2.2301,
      "step": 27527
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6566836833953857,
      "learning_rate": 0.0005791138125229195,
      "loss": 2.2094,
      "step": 27528
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6495917439460754,
      "learning_rate": 0.0005791123129043658,
      "loss": 2.2412,
      "step": 27529
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.7909334897994995,
      "learning_rate": 0.00057911081323392,
      "loss": 2.4249,
      "step": 27530
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.7054287791252136,
      "learning_rate": 0.0005791093135115819,
      "loss": 2.3058,
      "step": 27531
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.671945333480835,
      "learning_rate": 0.0005791078137373524,
      "loss": 2.1093,
      "step": 27532
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6822897791862488,
      "learning_rate": 0.0005791063139112313,
      "loss": 2.4781,
      "step": 27533
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6119848489761353,
      "learning_rate": 0.0005791048140332192,
      "loss": 2.246,
      "step": 27534
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6545458436012268,
      "learning_rate": 0.0005791033141033161,
      "loss": 2.4343,
      "step": 27535
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.7204993367195129,
      "learning_rate": 0.0005791018141215226,
      "loss": 2.5166,
      "step": 27536
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.7009117603302002,
      "learning_rate": 0.0005791003140878387,
      "loss": 2.1217,
      "step": 27537
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6451205015182495,
      "learning_rate": 0.0005790988140022648,
      "loss": 2.3552,
      "step": 27538
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6918128728866577,
      "learning_rate": 0.0005790973138648011,
      "loss": 2.3367,
      "step": 27539
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6506822109222412,
      "learning_rate": 0.000579095813675448,
      "loss": 2.2067,
      "step": 27540
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.9132933020591736,
      "learning_rate": 0.0005790943134342057,
      "loss": 2.1148,
      "step": 27541
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6684697866439819,
      "learning_rate": 0.0005790928131410746,
      "loss": 2.4203,
      "step": 27542
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.7567638158798218,
      "learning_rate": 0.0005790913127960547,
      "loss": 2.4594,
      "step": 27543
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6558840870857239,
      "learning_rate": 0.0005790898123991466,
      "loss": 2.3419,
      "step": 27544
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6095597147941589,
      "learning_rate": 0.0005790883119503504,
      "loss": 2.3125,
      "step": 27545
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.8451119065284729,
      "learning_rate": 0.0005790868114496664,
      "loss": 2.3849,
      "step": 27546
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.7574730515480042,
      "learning_rate": 0.0005790853108970949,
      "loss": 2.3135,
      "step": 27547
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6699912548065186,
      "learning_rate": 0.0005790838102926361,
      "loss": 2.5262,
      "step": 27548
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6464727520942688,
      "learning_rate": 0.0005790823096362904,
      "loss": 2.4304,
      "step": 27549
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.7686857581138611,
      "learning_rate": 0.0005790808089280582,
      "loss": 1.981,
      "step": 27550
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.5969953536987305,
      "learning_rate": 0.0005790793081679394,
      "loss": 2.5178,
      "step": 27551
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6935573816299438,
      "learning_rate": 0.0005790778073559345,
      "loss": 2.477,
      "step": 27552
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.7026118040084839,
      "learning_rate": 0.0005790763064920439,
      "loss": 2.3595,
      "step": 27553
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6663509011268616,
      "learning_rate": 0.0005790748055762676,
      "loss": 2.2463,
      "step": 27554
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6566027402877808,
      "learning_rate": 0.0005790733046086061,
      "loss": 2.4572,
      "step": 27555
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6504522562026978,
      "learning_rate": 0.0005790718035890597,
      "loss": 2.2381,
      "step": 27556
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6225071549415588,
      "learning_rate": 0.0005790703025176283,
      "loss": 2.4706,
      "step": 27557
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6765667200088501,
      "learning_rate": 0.0005790688013943126,
      "loss": 2.2427,
      "step": 27558
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.608565092086792,
      "learning_rate": 0.0005790673002191129,
      "loss": 2.3414,
      "step": 27559
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6905441284179688,
      "learning_rate": 0.0005790657989920291,
      "loss": 2.2502,
      "step": 27560
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6295905113220215,
      "learning_rate": 0.0005790642977130619,
      "loss": 2.2157,
      "step": 27561
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6614428758621216,
      "learning_rate": 0.0005790627963822112,
      "loss": 2.3112,
      "step": 27562
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6843243837356567,
      "learning_rate": 0.0005790612949994776,
      "loss": 2.3966,
      "step": 27563
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6128536462783813,
      "learning_rate": 0.0005790597935648611,
      "loss": 2.4153,
      "step": 27564
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6074572801589966,
      "learning_rate": 0.0005790582920783623,
      "loss": 2.5088,
      "step": 27565
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.5985679626464844,
      "learning_rate": 0.0005790567905399811,
      "loss": 2.271,
      "step": 27566
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6461614370346069,
      "learning_rate": 0.0005790552889497181,
      "loss": 2.2301,
      "step": 27567
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6534115076065063,
      "learning_rate": 0.0005790537873075735,
      "loss": 2.3994,
      "step": 27568
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6496558785438538,
      "learning_rate": 0.0005790522856135474,
      "loss": 2.6922,
      "step": 27569
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.7264584898948669,
      "learning_rate": 0.0005790507838676403,
      "loss": 2.4211,
      "step": 27570
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6123146414756775,
      "learning_rate": 0.0005790492820698523,
      "loss": 2.3467,
      "step": 27571
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6662023663520813,
      "learning_rate": 0.0005790477802201839,
      "loss": 2.3353,
      "step": 27572
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.9286986589431763,
      "learning_rate": 0.0005790462783186351,
      "loss": 2.2964,
      "step": 27573
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6394127607345581,
      "learning_rate": 0.0005790447763652064,
      "loss": 2.5568,
      "step": 27574
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.7809778451919556,
      "learning_rate": 0.000579043274359898,
      "loss": 2.3068,
      "step": 27575
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6382836699485779,
      "learning_rate": 0.0005790417723027103,
      "loss": 2.5261,
      "step": 27576
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6366741061210632,
      "learning_rate": 0.0005790402701936434,
      "loss": 2.1761,
      "step": 27577
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.5954352617263794,
      "learning_rate": 0.0005790387680326976,
      "loss": 2.3577,
      "step": 27578
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.7224125862121582,
      "learning_rate": 0.0005790372658198732,
      "loss": 2.5791,
      "step": 27579
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6172147989273071,
      "learning_rate": 0.0005790357635551705,
      "loss": 2.3401,
      "step": 27580
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.608363926410675,
      "learning_rate": 0.0005790342612385899,
      "loss": 2.4722,
      "step": 27581
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6582928895950317,
      "learning_rate": 0.0005790327588701315,
      "loss": 2.4127,
      "step": 27582
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6946738958358765,
      "learning_rate": 0.0005790312564497956,
      "loss": 2.5453,
      "step": 27583
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6118597388267517,
      "learning_rate": 0.0005790297539775825,
      "loss": 2.3037,
      "step": 27584
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6167632341384888,
      "learning_rate": 0.0005790282514534926,
      "loss": 2.2553,
      "step": 27585
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.686187744140625,
      "learning_rate": 0.0005790267488775261,
      "loss": 2.2934,
      "step": 27586
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6257979273796082,
      "learning_rate": 0.0005790252462496832,
      "loss": 2.3161,
      "step": 27587
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6212390065193176,
      "learning_rate": 0.0005790237435699643,
      "loss": 2.2191,
      "step": 27588
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.611992597579956,
      "learning_rate": 0.0005790222408383695,
      "loss": 2.3703,
      "step": 27589
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.623579204082489,
      "learning_rate": 0.0005790207380548993,
      "loss": 2.3758,
      "step": 27590
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6971458792686462,
      "learning_rate": 0.0005790192352195539,
      "loss": 2.3763,
      "step": 27591
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.5978402495384216,
      "learning_rate": 0.0005790177323323334,
      "loss": 2.2494,
      "step": 27592
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6220914125442505,
      "learning_rate": 0.0005790162293932384,
      "loss": 2.3287,
      "step": 27593
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6466558575630188,
      "learning_rate": 0.0005790147264022689,
      "loss": 2.3243,
      "step": 27594
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6135136485099792,
      "learning_rate": 0.0005790132233594253,
      "loss": 2.513,
      "step": 27595
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6196011304855347,
      "learning_rate": 0.000579011720264708,
      "loss": 2.2883,
      "step": 27596
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6736399531364441,
      "learning_rate": 0.000579010217118117,
      "loss": 2.5519,
      "step": 27597
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.5638031363487244,
      "learning_rate": 0.0005790087139196529,
      "loss": 2.4949,
      "step": 27598
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.5874485373497009,
      "learning_rate": 0.0005790072106693156,
      "loss": 2.3353,
      "step": 27599
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6422496438026428,
      "learning_rate": 0.0005790057073671058,
      "loss": 2.2631,
      "step": 27600
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.66683030128479,
      "learning_rate": 0.0005790042040130234,
      "loss": 2.4017,
      "step": 27601
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6549389362335205,
      "learning_rate": 0.000579002700607069,
      "loss": 2.3561,
      "step": 27602
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6829841136932373,
      "learning_rate": 0.0005790011971492427,
      "loss": 2.2847,
      "step": 27603
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6984069347381592,
      "learning_rate": 0.0005789996936395447,
      "loss": 2.4307,
      "step": 27604
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.5486485362052917,
      "learning_rate": 0.0005789981900779756,
      "loss": 2.3704,
      "step": 27605
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.711694598197937,
      "learning_rate": 0.0005789966864645353,
      "loss": 2.3779,
      "step": 27606
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.8419893383979797,
      "learning_rate": 0.0005789951827992242,
      "loss": 2.4465,
      "step": 27607
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6476280093193054,
      "learning_rate": 0.0005789936790820429,
      "loss": 2.3659,
      "step": 27608
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6071414351463318,
      "learning_rate": 0.0005789921753129911,
      "loss": 2.2872,
      "step": 27609
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.7264912128448486,
      "learning_rate": 0.0005789906714920696,
      "loss": 2.4914,
      "step": 27610
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6252796053886414,
      "learning_rate": 0.0005789891676192783,
      "loss": 2.2025,
      "step": 27611
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6940146088600159,
      "learning_rate": 0.0005789876636946177,
      "loss": 2.3873,
      "step": 27612
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.042226791381836,
      "learning_rate": 0.0005789861597180881,
      "loss": 2.3966,
      "step": 27613
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.7367271184921265,
      "learning_rate": 0.0005789846556896897,
      "loss": 2.3679,
      "step": 27614
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6478911638259888,
      "learning_rate": 0.0005789831516094228,
      "loss": 2.3144,
      "step": 27615
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.5838518738746643,
      "learning_rate": 0.0005789816474772875,
      "loss": 2.0427,
      "step": 27616
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6127484440803528,
      "learning_rate": 0.0005789801432932845,
      "loss": 2.1921,
      "step": 27617
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.8455987572669983,
      "learning_rate": 0.0005789786390574136,
      "loss": 2.2032,
      "step": 27618
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6860530376434326,
      "learning_rate": 0.0005789771347696754,
      "loss": 2.1902,
      "step": 27619
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6794017553329468,
      "learning_rate": 0.0005789756304300701,
      "loss": 2.1985,
      "step": 27620
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6304509043693542,
      "learning_rate": 0.000578974126038598,
      "loss": 2.235,
      "step": 27621
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6416863203048706,
      "learning_rate": 0.0005789726215952592,
      "loss": 2.339,
      "step": 27622
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6222143173217773,
      "learning_rate": 0.0005789711171000542,
      "loss": 2.4653,
      "step": 27623
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.771309494972229,
      "learning_rate": 0.0005789696125529832,
      "loss": 2.2291,
      "step": 27624
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6754313111305237,
      "learning_rate": 0.0005789681079540465,
      "loss": 2.2385,
      "step": 27625
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6173055768013,
      "learning_rate": 0.0005789666033032444,
      "loss": 2.2094,
      "step": 27626
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6206759214401245,
      "learning_rate": 0.0005789650986005771,
      "loss": 2.1771,
      "step": 27627
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6538341045379639,
      "learning_rate": 0.0005789635938460449,
      "loss": 2.1668,
      "step": 27628
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.7096328735351562,
      "learning_rate": 0.0005789620890396482,
      "loss": 2.2297,
      "step": 27629
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6495996117591858,
      "learning_rate": 0.0005789605841813871,
      "loss": 2.4062,
      "step": 27630
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6464183330535889,
      "learning_rate": 0.000578959079271262,
      "loss": 2.2751,
      "step": 27631
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6323439478874207,
      "learning_rate": 0.000578957574309273,
      "loss": 2.2215,
      "step": 27632
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6516760587692261,
      "learning_rate": 0.0005789560692954207,
      "loss": 2.4876,
      "step": 27633
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6199496984481812,
      "learning_rate": 0.0005789545642297051,
      "loss": 2.3673,
      "step": 27634
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.7566804885864258,
      "learning_rate": 0.0005789530591121266,
      "loss": 2.1238,
      "step": 27635
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6633416414260864,
      "learning_rate": 0.0005789515539426855,
      "loss": 2.2627,
      "step": 27636
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6635300517082214,
      "learning_rate": 0.000578950048721382,
      "loss": 2.3835,
      "step": 27637
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.636366069316864,
      "learning_rate": 0.0005789485434482165,
      "loss": 2.3603,
      "step": 27638
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6480618119239807,
      "learning_rate": 0.0005789470381231891,
      "loss": 2.3504,
      "step": 27639
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.676074743270874,
      "learning_rate": 0.0005789455327463002,
      "loss": 2.1753,
      "step": 27640
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6867640018463135,
      "learning_rate": 0.0005789440273175501,
      "loss": 2.2366,
      "step": 27641
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6797086596488953,
      "learning_rate": 0.000578942521836939,
      "loss": 2.3177,
      "step": 27642
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6563263535499573,
      "learning_rate": 0.0005789410163044672,
      "loss": 2.4234,
      "step": 27643
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6176376938819885,
      "learning_rate": 0.0005789395107201351,
      "loss": 2.3908,
      "step": 27644
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6261677145957947,
      "learning_rate": 0.0005789380050839428,
      "loss": 2.348,
      "step": 27645
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6385241746902466,
      "learning_rate": 0.0005789364993958907,
      "loss": 2.2934,
      "step": 27646
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.7139105200767517,
      "learning_rate": 0.000578934993655979,
      "loss": 2.3034,
      "step": 27647
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.601395845413208,
      "learning_rate": 0.000578933487864208,
      "loss": 2.2461,
      "step": 27648
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.5862590074539185,
      "learning_rate": 0.000578931982020578,
      "loss": 2.3054,
      "step": 27649
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.0063145160675049,
      "learning_rate": 0.0005789304761250894,
      "loss": 2.4433,
      "step": 27650
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6204516291618347,
      "learning_rate": 0.0005789289701777422,
      "loss": 2.2645,
      "step": 27651
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.7112347483634949,
      "learning_rate": 0.000578927464178537,
      "loss": 2.3637,
      "step": 27652
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6661891937255859,
      "learning_rate": 0.0005789259581274738,
      "loss": 2.4256,
      "step": 27653
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6568484306335449,
      "learning_rate": 0.0005789244520245531,
      "loss": 2.4801,
      "step": 27654
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6697264909744263,
      "learning_rate": 0.0005789229458697749,
      "loss": 2.4646,
      "step": 27655
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6741417050361633,
      "learning_rate": 0.0005789214396631398,
      "loss": 2.3538,
      "step": 27656
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6415122747421265,
      "learning_rate": 0.0005789199334046478,
      "loss": 2.1811,
      "step": 27657
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6541082262992859,
      "learning_rate": 0.0005789184270942995,
      "loss": 2.2163,
      "step": 27658
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.7930756211280823,
      "learning_rate": 0.0005789169207320949,
      "loss": 2.1141,
      "step": 27659
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6242858171463013,
      "learning_rate": 0.0005789154143180345,
      "loss": 2.3592,
      "step": 27660
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6438081860542297,
      "learning_rate": 0.0005789139078521183,
      "loss": 2.1959,
      "step": 27661
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.687335729598999,
      "learning_rate": 0.0005789124013343467,
      "loss": 2.3215,
      "step": 27662
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6037210822105408,
      "learning_rate": 0.0005789108947647203,
      "loss": 2.2804,
      "step": 27663
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6690006852149963,
      "learning_rate": 0.0005789093881432389,
      "loss": 2.3718,
      "step": 27664
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.63409823179245,
      "learning_rate": 0.000578907881469903,
      "loss": 2.4188,
      "step": 27665
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.5958955883979797,
      "learning_rate": 0.0005789063747447128,
      "loss": 2.279,
      "step": 27666
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.7227504849433899,
      "learning_rate": 0.0005789048679676688,
      "loss": 2.3474,
      "step": 27667
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.7367081046104431,
      "learning_rate": 0.000578903361138771,
      "loss": 2.403,
      "step": 27668
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.7320654392242432,
      "learning_rate": 0.0005789018542580198,
      "loss": 2.4769,
      "step": 27669
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.627609133720398,
      "learning_rate": 0.0005789003473254155,
      "loss": 2.411,
      "step": 27670
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6477734446525574,
      "learning_rate": 0.0005788988403409584,
      "loss": 2.1278,
      "step": 27671
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6202255487442017,
      "learning_rate": 0.0005788973333046486,
      "loss": 2.2676,
      "step": 27672
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6363105177879333,
      "learning_rate": 0.0005788958262164867,
      "loss": 2.2761,
      "step": 27673
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6934155225753784,
      "learning_rate": 0.0005788943190764728,
      "loss": 2.2872,
      "step": 27674
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6500599980354309,
      "learning_rate": 0.0005788928118846071,
      "loss": 2.1729,
      "step": 27675
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6813166737556458,
      "learning_rate": 0.0005788913046408899,
      "loss": 2.4653,
      "step": 27676
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6970953345298767,
      "learning_rate": 0.0005788897973453217,
      "loss": 2.2919,
      "step": 27677
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6492809057235718,
      "learning_rate": 0.0005788882899979025,
      "loss": 2.3127,
      "step": 27678
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.651455283164978,
      "learning_rate": 0.0005788867825986327,
      "loss": 2.07,
      "step": 27679
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6505257487297058,
      "learning_rate": 0.0005788852751475126,
      "loss": 2.5019,
      "step": 27680
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6376057267189026,
      "learning_rate": 0.0005788837676445425,
      "loss": 2.2126,
      "step": 27681
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6346815228462219,
      "learning_rate": 0.0005788822600897226,
      "loss": 2.3889,
      "step": 27682
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6261664628982544,
      "learning_rate": 0.0005788807524830533,
      "loss": 2.3986,
      "step": 27683
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.56742262840271,
      "learning_rate": 0.0005788792448245348,
      "loss": 2.2083,
      "step": 27684
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.5959411263465881,
      "learning_rate": 0.0005788777371141673,
      "loss": 2.3829,
      "step": 27685
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6187498569488525,
      "learning_rate": 0.0005788762293519513,
      "loss": 2.4201,
      "step": 27686
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6936042904853821,
      "learning_rate": 0.0005788747215378868,
      "loss": 2.4458,
      "step": 27687
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6437271237373352,
      "learning_rate": 0.0005788732136719743,
      "loss": 2.1468,
      "step": 27688
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6785808205604553,
      "learning_rate": 0.000578871705754214,
      "loss": 2.4174,
      "step": 27689
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6425001621246338,
      "learning_rate": 0.0005788701977846061,
      "loss": 2.245,
      "step": 27690
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.640215277671814,
      "learning_rate": 0.0005788686897631511,
      "loss": 2.3331,
      "step": 27691
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.670341968536377,
      "learning_rate": 0.000578867181689849,
      "loss": 2.1412,
      "step": 27692
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6254570484161377,
      "learning_rate": 0.0005788656735647003,
      "loss": 2.066,
      "step": 27693
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6139822006225586,
      "learning_rate": 0.0005788641653877052,
      "loss": 2.4958,
      "step": 27694
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.7292101979255676,
      "learning_rate": 0.000578862657158864,
      "loss": 2.4232,
      "step": 27695
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6274720430374146,
      "learning_rate": 0.000578861148878177,
      "loss": 2.1183,
      "step": 27696
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.7571378946304321,
      "learning_rate": 0.0005788596405456444,
      "loss": 2.5401,
      "step": 27697
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.5966646671295166,
      "learning_rate": 0.0005788581321612665,
      "loss": 2.3333,
      "step": 27698
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.7214334607124329,
      "learning_rate": 0.0005788566237250437,
      "loss": 2.3408,
      "step": 27699
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6769042611122131,
      "learning_rate": 0.0005788551152369762,
      "loss": 2.4336,
      "step": 27700
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.7045136094093323,
      "learning_rate": 0.0005788536066970642,
      "loss": 2.3203,
      "step": 27701
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6023701429367065,
      "learning_rate": 0.000578852098105308,
      "loss": 2.6974,
      "step": 27702
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.5679107308387756,
      "learning_rate": 0.0005788505894617081,
      "loss": 2.1934,
      "step": 27703
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6410475969314575,
      "learning_rate": 0.0005788490807662644,
      "loss": 2.3389,
      "step": 27704
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.7232301235198975,
      "learning_rate": 0.0005788475720189776,
      "loss": 2.2903,
      "step": 27705
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6359960436820984,
      "learning_rate": 0.0005788460632198476,
      "loss": 2.4242,
      "step": 27706
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6886021494865417,
      "learning_rate": 0.0005788445543688749,
      "loss": 2.3235,
      "step": 27707
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.7032310366630554,
      "learning_rate": 0.0005788430454660597,
      "loss": 2.1747,
      "step": 27708
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6832205653190613,
      "learning_rate": 0.0005788415365114024,
      "loss": 2.2523,
      "step": 27709
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6192331314086914,
      "learning_rate": 0.0005788400275049033,
      "loss": 2.4097,
      "step": 27710
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.1756112575531006,
      "learning_rate": 0.0005788385184465623,
      "loss": 2.1577,
      "step": 27711
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6525822877883911,
      "learning_rate": 0.0005788370093363802,
      "loss": 2.4002,
      "step": 27712
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.849657416343689,
      "learning_rate": 0.0005788355001743569,
      "loss": 2.1826,
      "step": 27713
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6821568012237549,
      "learning_rate": 0.0005788339909604929,
      "loss": 2.2165,
      "step": 27714
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6604001522064209,
      "learning_rate": 0.0005788324816947884,
      "loss": 2.2735,
      "step": 27715
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.7438984513282776,
      "learning_rate": 0.0005788309723772436,
      "loss": 2.2297,
      "step": 27716
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6388592720031738,
      "learning_rate": 0.0005788294630078588,
      "loss": 2.2129,
      "step": 27717
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6750214695930481,
      "learning_rate": 0.0005788279535866345,
      "loss": 2.2193,
      "step": 27718
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6299809217453003,
      "learning_rate": 0.0005788264441135707,
      "loss": 2.361,
      "step": 27719
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6079923510551453,
      "learning_rate": 0.0005788249345886679,
      "loss": 2.3886,
      "step": 27720
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.610438883304596,
      "learning_rate": 0.0005788234250119263,
      "loss": 2.3316,
      "step": 27721
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6661330461502075,
      "learning_rate": 0.000578821915383346,
      "loss": 2.2443,
      "step": 27722
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6241629719734192,
      "learning_rate": 0.0005788204057029275,
      "loss": 2.4123,
      "step": 27723
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.709295928478241,
      "learning_rate": 0.0005788188959706712,
      "loss": 2.3265,
      "step": 27724
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6190609335899353,
      "learning_rate": 0.000578817386186577,
      "loss": 2.2721,
      "step": 27725
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6298283934593201,
      "learning_rate": 0.0005788158763506455,
      "loss": 2.1156,
      "step": 27726
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.709671676158905,
      "learning_rate": 0.0005788143664628767,
      "loss": 2.2872,
      "step": 27727
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6483332514762878,
      "learning_rate": 0.0005788128565232712,
      "loss": 2.1567,
      "step": 27728
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.7611895799636841,
      "learning_rate": 0.0005788113465318291,
      "loss": 2.3747,
      "step": 27729
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.5955640077590942,
      "learning_rate": 0.0005788098364885506,
      "loss": 2.1239,
      "step": 27730
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.7106309533119202,
      "learning_rate": 0.0005788083263934363,
      "loss": 2.285,
      "step": 27731
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.587161123752594,
      "learning_rate": 0.0005788068162464862,
      "loss": 2.3725,
      "step": 27732
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6367245316505432,
      "learning_rate": 0.0005788053060477006,
      "loss": 2.4135,
      "step": 27733
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6725303530693054,
      "learning_rate": 0.0005788037957970797,
      "loss": 2.2782,
      "step": 27734
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.593665361404419,
      "learning_rate": 0.0005788022854946241,
      "loss": 2.2959,
      "step": 27735
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6928353905677795,
      "learning_rate": 0.0005788007751403338,
      "loss": 2.2648,
      "step": 27736
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6450080871582031,
      "learning_rate": 0.0005787992647342092,
      "loss": 2.3813,
      "step": 27737
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6099846363067627,
      "learning_rate": 0.0005787977542762507,
      "loss": 2.3799,
      "step": 27738
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6701552271842957,
      "learning_rate": 0.0005787962437664582,
      "loss": 2.1494,
      "step": 27739
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6225707530975342,
      "learning_rate": 0.0005787947332048323,
      "loss": 2.3188,
      "step": 27740
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6404434442520142,
      "learning_rate": 0.0005787932225913733,
      "loss": 2.1768,
      "step": 27741
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6715446710586548,
      "learning_rate": 0.0005787917119260813,
      "loss": 2.0572,
      "step": 27742
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6709569692611694,
      "learning_rate": 0.0005787902012089565,
      "loss": 2.4229,
      "step": 27743
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6564173102378845,
      "learning_rate": 0.0005787886904399996,
      "loss": 2.2575,
      "step": 27744
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6520128846168518,
      "learning_rate": 0.0005787871796192106,
      "loss": 2.3496,
      "step": 27745
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6990010142326355,
      "learning_rate": 0.0005787856687465896,
      "loss": 2.3612,
      "step": 27746
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6267711520195007,
      "learning_rate": 0.0005787841578221371,
      "loss": 2.3927,
      "step": 27747
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.7035689949989319,
      "learning_rate": 0.0005787826468458535,
      "loss": 2.113,
      "step": 27748
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.7710591554641724,
      "learning_rate": 0.000578781135817739,
      "loss": 2.4066,
      "step": 27749
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6296369433403015,
      "learning_rate": 0.0005787796247377937,
      "loss": 2.5058,
      "step": 27750
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6547799706459045,
      "learning_rate": 0.0005787781136060179,
      "loss": 2.457,
      "step": 27751
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6583263278007507,
      "learning_rate": 0.0005787766024224122,
      "loss": 2.3334,
      "step": 27752
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.679703414440155,
      "learning_rate": 0.0005787750911869766,
      "loss": 2.2897,
      "step": 27753
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6712707281112671,
      "learning_rate": 0.0005787735798997115,
      "loss": 2.3976,
      "step": 27754
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6914992928504944,
      "learning_rate": 0.0005787720685606169,
      "loss": 2.4714,
      "step": 27755
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.660778820514679,
      "learning_rate": 0.0005787705571696936,
      "loss": 2.1607,
      "step": 27756
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.619184672832489,
      "learning_rate": 0.0005787690457269414,
      "loss": 2.6146,
      "step": 27757
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.608734667301178,
      "learning_rate": 0.0005787675342323608,
      "loss": 2.3557,
      "step": 27758
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6949827671051025,
      "learning_rate": 0.0005787660226859522,
      "loss": 2.3544,
      "step": 27759
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.8268179297447205,
      "learning_rate": 0.0005787645110877155,
      "loss": 2.3846,
      "step": 27760
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6936354637145996,
      "learning_rate": 0.0005787629994376514,
      "loss": 2.307,
      "step": 27761
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.7045008540153503,
      "learning_rate": 0.0005787614877357598,
      "loss": 2.1291,
      "step": 27762
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6438559889793396,
      "learning_rate": 0.0005787599759820414,
      "loss": 2.3313,
      "step": 27763
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.7007122039794922,
      "learning_rate": 0.0005787584641764961,
      "loss": 2.275,
      "step": 27764
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6886013746261597,
      "learning_rate": 0.0005787569523191244,
      "loss": 2.4452,
      "step": 27765
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.748677670955658,
      "learning_rate": 0.0005787554404099265,
      "loss": 2.3446,
      "step": 27766
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.7424733638763428,
      "learning_rate": 0.0005787539284489028,
      "loss": 2.1591,
      "step": 27767
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6171644926071167,
      "learning_rate": 0.0005787524164360534,
      "loss": 2.1494,
      "step": 27768
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6387138962745667,
      "learning_rate": 0.0005787509043713787,
      "loss": 2.4342,
      "step": 27769
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.692945659160614,
      "learning_rate": 0.0005787493922548789,
      "loss": 2.3962,
      "step": 27770
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.7721610069274902,
      "learning_rate": 0.0005787478800865543,
      "loss": 2.1898,
      "step": 27771
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6357513070106506,
      "learning_rate": 0.0005787463678664053,
      "loss": 2.2209,
      "step": 27772
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.5847139358520508,
      "learning_rate": 0.0005787448555944321,
      "loss": 2.4154,
      "step": 27773
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.841850996017456,
      "learning_rate": 0.0005787433432706348,
      "loss": 2.3497,
      "step": 27774
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.7193838357925415,
      "learning_rate": 0.0005787418308950141,
      "loss": 2.2506,
      "step": 27775
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.7364742755889893,
      "learning_rate": 0.0005787403184675698,
      "loss": 2.4922,
      "step": 27776
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6840186715126038,
      "learning_rate": 0.0005787388059883025,
      "loss": 2.4124,
      "step": 27777
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.817005455493927,
      "learning_rate": 0.0005787372934572124,
      "loss": 2.1563,
      "step": 27778
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6553179025650024,
      "learning_rate": 0.0005787357808742997,
      "loss": 2.3721,
      "step": 27779
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6799712181091309,
      "learning_rate": 0.0005787342682395649,
      "loss": 2.5384,
      "step": 27780
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.7529270648956299,
      "learning_rate": 0.0005787327555530081,
      "loss": 2.2495,
      "step": 27781
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.7177132368087769,
      "learning_rate": 0.0005787312428146296,
      "loss": 2.2501,
      "step": 27782
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.7084015011787415,
      "learning_rate": 0.0005787297300244296,
      "loss": 2.3535,
      "step": 27783
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6814277768135071,
      "learning_rate": 0.0005787282171824086,
      "loss": 2.4259,
      "step": 27784
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.7660267353057861,
      "learning_rate": 0.0005787267042885667,
      "loss": 2.2012,
      "step": 27785
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6218283772468567,
      "learning_rate": 0.0005787251913429043,
      "loss": 2.2281,
      "step": 27786
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6298564076423645,
      "learning_rate": 0.0005787236783454216,
      "loss": 2.3969,
      "step": 27787
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6325021386146545,
      "learning_rate": 0.000578722165296119,
      "loss": 2.3024,
      "step": 27788
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.7691733241081238,
      "learning_rate": 0.0005787206521949966,
      "loss": 2.458,
      "step": 27789
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6046663522720337,
      "learning_rate": 0.0005787191390420547,
      "loss": 2.3029,
      "step": 27790
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6335516571998596,
      "learning_rate": 0.0005787176258372937,
      "loss": 2.3646,
      "step": 27791
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.5706393718719482,
      "learning_rate": 0.0005787161125807139,
      "loss": 2.5571,
      "step": 27792
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6396344304084778,
      "learning_rate": 0.0005787145992723155,
      "loss": 2.3702,
      "step": 27793
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.7068721652030945,
      "learning_rate": 0.0005787130859120987,
      "loss": 2.3136,
      "step": 27794
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.7054308652877808,
      "learning_rate": 0.0005787115725000639,
      "loss": 2.0927,
      "step": 27795
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6247085332870483,
      "learning_rate": 0.0005787100590362114,
      "loss": 2.3278,
      "step": 27796
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6579838991165161,
      "learning_rate": 0.0005787085455205414,
      "loss": 2.281,
      "step": 27797
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6909993290901184,
      "learning_rate": 0.0005787070319530544,
      "loss": 2.3279,
      "step": 27798
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6216599941253662,
      "learning_rate": 0.0005787055183337503,
      "loss": 2.1188,
      "step": 27799
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.5823714137077332,
      "learning_rate": 0.0005787040046626296,
      "loss": 2.2697,
      "step": 27800
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.5849681496620178,
      "learning_rate": 0.0005787024909396927,
      "loss": 2.06,
      "step": 27801
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6922834515571594,
      "learning_rate": 0.0005787009771649396,
      "loss": 2.3549,
      "step": 27802
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6903700232505798,
      "learning_rate": 0.0005786994633383707,
      "loss": 2.4991,
      "step": 27803
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.615894079208374,
      "learning_rate": 0.0005786979494599864,
      "loss": 2.3015,
      "step": 27804
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6758268475532532,
      "learning_rate": 0.0005786964355297869,
      "loss": 2.4909,
      "step": 27805
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.719318687915802,
      "learning_rate": 0.0005786949215477725,
      "loss": 2.1708,
      "step": 27806
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.8210867047309875,
      "learning_rate": 0.0005786934075139434,
      "loss": 2.347,
      "step": 27807
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6366984844207764,
      "learning_rate": 0.0005786918934282999,
      "loss": 2.4434,
      "step": 27808
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.660481333732605,
      "learning_rate": 0.0005786903792908424,
      "loss": 2.1275,
      "step": 27809
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.7447078824043274,
      "learning_rate": 0.0005786888651015709,
      "loss": 2.3646,
      "step": 27810
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6517224907875061,
      "learning_rate": 0.0005786873508604861,
      "loss": 2.2032,
      "step": 27811
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6479751467704773,
      "learning_rate": 0.0005786858365675881,
      "loss": 2.2781,
      "step": 27812
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.683894157409668,
      "learning_rate": 0.000578684322222877,
      "loss": 2.4255,
      "step": 27813
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6158778667449951,
      "learning_rate": 0.0005786828078263532,
      "loss": 2.1057,
      "step": 27814
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6868499517440796,
      "learning_rate": 0.0005786812933780171,
      "loss": 2.3652,
      "step": 27815
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6225244402885437,
      "learning_rate": 0.0005786797788778688,
      "loss": 2.2881,
      "step": 27816
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6272533535957336,
      "learning_rate": 0.0005786782643259088,
      "loss": 2.1496,
      "step": 27817
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.5722367167472839,
      "learning_rate": 0.0005786767497221371,
      "loss": 2.4137,
      "step": 27818
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.649827778339386,
      "learning_rate": 0.0005786752350665542,
      "loss": 2.4356,
      "step": 27819
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6147266626358032,
      "learning_rate": 0.0005786737203591603,
      "loss": 2.2451,
      "step": 27820
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6723379492759705,
      "learning_rate": 0.0005786722055999557,
      "loss": 2.3886,
      "step": 27821
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6134493350982666,
      "learning_rate": 0.0005786706907889407,
      "loss": 2.3375,
      "step": 27822
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6121564507484436,
      "learning_rate": 0.0005786691759261156,
      "loss": 2.4462,
      "step": 27823
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.5749025940895081,
      "learning_rate": 0.0005786676610114805,
      "loss": 2.2052,
      "step": 27824
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6158670783042908,
      "learning_rate": 0.0005786661460450359,
      "loss": 2.3518,
      "step": 27825
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6264457106590271,
      "learning_rate": 0.000578664631026782,
      "loss": 2.1732,
      "step": 27826
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6370219588279724,
      "learning_rate": 0.0005786631159567191,
      "loss": 2.3763,
      "step": 27827
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.675667405128479,
      "learning_rate": 0.0005786616008348475,
      "loss": 2.3692,
      "step": 27828
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.621312141418457,
      "learning_rate": 0.0005786600856611674,
      "loss": 2.2774,
      "step": 27829
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6353198289871216,
      "learning_rate": 0.0005786585704356791,
      "loss": 2.5727,
      "step": 27830
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.5982691049575806,
      "learning_rate": 0.000578657055158383,
      "loss": 2.3114,
      "step": 27831
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6627662181854248,
      "learning_rate": 0.0005786555398292792,
      "loss": 2.4673,
      "step": 27832
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6378557682037354,
      "learning_rate": 0.0005786540244483681,
      "loss": 2.254,
      "step": 27833
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.7319316267967224,
      "learning_rate": 0.00057865250901565,
      "loss": 2.33,
      "step": 27834
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6419297456741333,
      "learning_rate": 0.0005786509935311251,
      "loss": 2.5173,
      "step": 27835
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6686010360717773,
      "learning_rate": 0.0005786494779947938,
      "loss": 2.2606,
      "step": 27836
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6167565584182739,
      "learning_rate": 0.0005786479624066562,
      "loss": 2.472,
      "step": 27837
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6156690716743469,
      "learning_rate": 0.0005786464467667128,
      "loss": 2.5049,
      "step": 27838
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6640374660491943,
      "learning_rate": 0.0005786449310749636,
      "loss": 2.6351,
      "step": 27839
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.5775731205940247,
      "learning_rate": 0.0005786434153314092,
      "loss": 2.2425,
      "step": 27840
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6491432785987854,
      "learning_rate": 0.0005786418995360497,
      "loss": 2.3974,
      "step": 27841
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6008294820785522,
      "learning_rate": 0.0005786403836888853,
      "loss": 2.1812,
      "step": 27842
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6093000769615173,
      "learning_rate": 0.0005786388677899164,
      "loss": 2.378,
      "step": 27843
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6134883165359497,
      "learning_rate": 0.0005786373518391434,
      "loss": 2.5574,
      "step": 27844
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6643854379653931,
      "learning_rate": 0.0005786358358365664,
      "loss": 2.2887,
      "step": 27845
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6892435550689697,
      "learning_rate": 0.0005786343197821857,
      "loss": 2.3859,
      "step": 27846
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6498056054115295,
      "learning_rate": 0.0005786328036760018,
      "loss": 2.3226,
      "step": 27847
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6555158495903015,
      "learning_rate": 0.0005786312875180146,
      "loss": 2.6194,
      "step": 27848
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6761205792427063,
      "learning_rate": 0.0005786297713082246,
      "loss": 2.3861,
      "step": 27849
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6770347952842712,
      "learning_rate": 0.0005786282550466321,
      "loss": 2.4699,
      "step": 27850
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6378076076507568,
      "learning_rate": 0.0005786267387332373,
      "loss": 2.2033,
      "step": 27851
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6323301792144775,
      "learning_rate": 0.0005786252223680406,
      "loss": 2.1755,
      "step": 27852
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.767561137676239,
      "learning_rate": 0.0005786237059510422,
      "loss": 2.3924,
      "step": 27853
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6100536584854126,
      "learning_rate": 0.0005786221894822424,
      "loss": 2.4554,
      "step": 27854
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6258146166801453,
      "learning_rate": 0.0005786206729616416,
      "loss": 2.399,
      "step": 27855
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6868932247161865,
      "learning_rate": 0.0005786191563892397,
      "loss": 2.2883,
      "step": 27856
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.700551450252533,
      "learning_rate": 0.0005786176397650374,
      "loss": 2.3309,
      "step": 27857
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6748934984207153,
      "learning_rate": 0.0005786161230890347,
      "loss": 2.3671,
      "step": 27858
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6657612919807434,
      "learning_rate": 0.0005786146063612321,
      "loss": 2.5401,
      "step": 27859
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.7986941933631897,
      "learning_rate": 0.0005786130895816298,
      "loss": 2.1224,
      "step": 27860
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.7166428565979004,
      "learning_rate": 0.000578611572750228,
      "loss": 2.3495,
      "step": 27861
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6758357882499695,
      "learning_rate": 0.0005786100558670272,
      "loss": 2.1392,
      "step": 27862
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6577600240707397,
      "learning_rate": 0.0005786085389320274,
      "loss": 2.4156,
      "step": 27863
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.7014693021774292,
      "learning_rate": 0.000578607021945229,
      "loss": 2.2456,
      "step": 27864
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.7030181884765625,
      "learning_rate": 0.0005786055049066323,
      "loss": 2.4101,
      "step": 27865
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6889173984527588,
      "learning_rate": 0.0005786039878162377,
      "loss": 2.1756,
      "step": 27866
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6187517046928406,
      "learning_rate": 0.0005786024706740453,
      "loss": 2.267,
      "step": 27867
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.7716091871261597,
      "learning_rate": 0.0005786009534800554,
      "loss": 2.414,
      "step": 27868
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6918323636054993,
      "learning_rate": 0.0005785994362342683,
      "loss": 2.4368,
      "step": 27869
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.669106662273407,
      "learning_rate": 0.0005785979189366843,
      "loss": 2.325,
      "step": 27870
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.757985532283783,
      "learning_rate": 0.0005785964015873038,
      "loss": 2.2981,
      "step": 27871
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.7240273952484131,
      "learning_rate": 0.0005785948841861269,
      "loss": 2.2144,
      "step": 27872
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.712926983833313,
      "learning_rate": 0.0005785933667331538,
      "loss": 2.3851,
      "step": 27873
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.9944437146186829,
      "learning_rate": 0.0005785918492283852,
      "loss": 2.4981,
      "step": 27874
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6012210845947266,
      "learning_rate": 0.000578590331671821,
      "loss": 2.1408,
      "step": 27875
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6288678646087646,
      "learning_rate": 0.0005785888140634615,
      "loss": 2.1238,
      "step": 27876
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.7279298305511475,
      "learning_rate": 0.0005785872964033073,
      "loss": 2.4441,
      "step": 27877
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.810366690158844,
      "learning_rate": 0.0005785857786913583,
      "loss": 2.2226,
      "step": 27878
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.599165678024292,
      "learning_rate": 0.000578584260927615,
      "loss": 2.2098,
      "step": 27879
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.8352980017662048,
      "learning_rate": 0.0005785827431120776,
      "loss": 2.2685,
      "step": 27880
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.7658840417861938,
      "learning_rate": 0.0005785812252447465,
      "loss": 2.4062,
      "step": 27881
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.5975106954574585,
      "learning_rate": 0.0005785797073256218,
      "loss": 2.4092,
      "step": 27882
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6585174202919006,
      "learning_rate": 0.0005785781893547038,
      "loss": 2.2873,
      "step": 27883
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.8090313076972961,
      "learning_rate": 0.000578576671331993,
      "loss": 2.2578,
      "step": 27884
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6563068628311157,
      "learning_rate": 0.0005785751532574895,
      "loss": 2.4514,
      "step": 27885
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6388887763023376,
      "learning_rate": 0.0005785736351311936,
      "loss": 2.1938,
      "step": 27886
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.8801082372665405,
      "learning_rate": 0.0005785721169531056,
      "loss": 2.4902,
      "step": 27887
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.7158915996551514,
      "learning_rate": 0.0005785705987232258,
      "loss": 2.3953,
      "step": 27888
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6035506129264832,
      "learning_rate": 0.0005785690804415545,
      "loss": 2.6609,
      "step": 27889
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.9630897641181946,
      "learning_rate": 0.0005785675621080918,
      "loss": 2.3758,
      "step": 27890
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.866025984287262,
      "learning_rate": 0.0005785660437228383,
      "loss": 2.3498,
      "step": 27891
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6226928234100342,
      "learning_rate": 0.000578564525285794,
      "loss": 2.4232,
      "step": 27892
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.8801444172859192,
      "learning_rate": 0.0005785630067969594,
      "loss": 2.0335,
      "step": 27893
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.9106794595718384,
      "learning_rate": 0.0005785614882563346,
      "loss": 2.3552,
      "step": 27894
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.7163737416267395,
      "learning_rate": 0.00057855996966392,
      "loss": 2.2177,
      "step": 27895
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.8118970990180969,
      "learning_rate": 0.0005785584510197159,
      "loss": 2.4458,
      "step": 27896
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6746031641960144,
      "learning_rate": 0.0005785569323237224,
      "loss": 2.2434,
      "step": 27897
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.7114272713661194,
      "learning_rate": 0.00057855541357594,
      "loss": 2.21,
      "step": 27898
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6768204569816589,
      "learning_rate": 0.0005785538947763689,
      "loss": 2.4051,
      "step": 27899
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.616574227809906,
      "learning_rate": 0.0005785523759250093,
      "loss": 2.3606,
      "step": 27900
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6334913372993469,
      "learning_rate": 0.0005785508570218616,
      "loss": 2.3067,
      "step": 27901
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6073115468025208,
      "learning_rate": 0.0005785493380669259,
      "loss": 2.2561,
      "step": 27902
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6467400789260864,
      "learning_rate": 0.0005785478190602028,
      "loss": 2.3774,
      "step": 27903
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6404222846031189,
      "learning_rate": 0.0005785463000016924,
      "loss": 2.0901,
      "step": 27904
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.643311619758606,
      "learning_rate": 0.0005785447808913949,
      "loss": 2.4033,
      "step": 27905
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6815385818481445,
      "learning_rate": 0.0005785432617293107,
      "loss": 2.5565,
      "step": 27906
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6617143750190735,
      "learning_rate": 0.0005785417425154401,
      "loss": 2.3676,
      "step": 27907
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6551686525344849,
      "learning_rate": 0.0005785402232497832,
      "loss": 2.2193,
      "step": 27908
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6371405720710754,
      "learning_rate": 0.0005785387039323406,
      "loss": 2.433,
      "step": 27909
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6207729578018188,
      "learning_rate": 0.0005785371845631123,
      "loss": 2.2172,
      "step": 27910
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.63065105676651,
      "learning_rate": 0.0005785356651420987,
      "loss": 2.3377,
      "step": 27911
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6322156190872192,
      "learning_rate": 0.0005785341456693,
      "loss": 2.2137,
      "step": 27912
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6293078064918518,
      "learning_rate": 0.0005785326261447165,
      "loss": 2.4224,
      "step": 27913
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6586890816688538,
      "learning_rate": 0.0005785311065683486,
      "loss": 2.3786,
      "step": 27914
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.7073611617088318,
      "learning_rate": 0.0005785295869401966,
      "loss": 2.5403,
      "step": 27915
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.7271679043769836,
      "learning_rate": 0.0005785280672602605,
      "loss": 2.4918,
      "step": 27916
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6206167340278625,
      "learning_rate": 0.000578526547528541,
      "loss": 2.0835,
      "step": 27917
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.7337155938148499,
      "learning_rate": 0.0005785250277450379,
      "loss": 2.5107,
      "step": 27918
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6396299600601196,
      "learning_rate": 0.000578523507909752,
      "loss": 2.2482,
      "step": 27919
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6275799870491028,
      "learning_rate": 0.0005785219880226831,
      "loss": 2.3736,
      "step": 27920
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6205373406410217,
      "learning_rate": 0.0005785204680838319,
      "loss": 2.3808,
      "step": 27921
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6310408711433411,
      "learning_rate": 0.0005785189480931984,
      "loss": 2.4649,
      "step": 27922
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.7265506982803345,
      "learning_rate": 0.0005785174280507829,
      "loss": 2.1495,
      "step": 27923
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6252872943878174,
      "learning_rate": 0.0005785159079565858,
      "loss": 2.293,
      "step": 27924
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6324518322944641,
      "learning_rate": 0.0005785143878106073,
      "loss": 2.1309,
      "step": 27925
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6286365985870361,
      "learning_rate": 0.0005785128676128479,
      "loss": 2.1473,
      "step": 27926
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6291888356208801,
      "learning_rate": 0.0005785113473633075,
      "loss": 2.306,
      "step": 27927
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6686140894889832,
      "learning_rate": 0.0005785098270619867,
      "loss": 2.4799,
      "step": 27928
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.7005850672721863,
      "learning_rate": 0.0005785083067088855,
      "loss": 2.3129,
      "step": 27929
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.7225909233093262,
      "learning_rate": 0.0005785067863040046,
      "loss": 2.6008,
      "step": 27930
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.691691517829895,
      "learning_rate": 0.0005785052658473438,
      "loss": 2.3319,
      "step": 27931
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6876548528671265,
      "learning_rate": 0.0005785037453389037,
      "loss": 2.2465,
      "step": 27932
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6693435311317444,
      "learning_rate": 0.0005785022247786847,
      "loss": 2.5317,
      "step": 27933
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6497507095336914,
      "learning_rate": 0.0005785007041666866,
      "loss": 2.4416,
      "step": 27934
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6714847683906555,
      "learning_rate": 0.00057849918350291,
      "loss": 1.9215,
      "step": 27935
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6073291301727295,
      "learning_rate": 0.0005784976627873552,
      "loss": 2.5204,
      "step": 27936
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6956767439842224,
      "learning_rate": 0.0005784961420200224,
      "loss": 2.1191,
      "step": 27937
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6478415727615356,
      "learning_rate": 0.000578494621200912,
      "loss": 2.3841,
      "step": 27938
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6669503450393677,
      "learning_rate": 0.0005784931003300241,
      "loss": 2.1043,
      "step": 27939
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6725537776947021,
      "learning_rate": 0.0005784915794073591,
      "loss": 2.3555,
      "step": 27940
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6670308113098145,
      "learning_rate": 0.0005784900584329173,
      "loss": 2.0815,
      "step": 27941
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.7034432888031006,
      "learning_rate": 0.0005784885374066988,
      "loss": 2.3415,
      "step": 27942
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6668187379837036,
      "learning_rate": 0.0005784870163287043,
      "loss": 2.5236,
      "step": 27943
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6375360488891602,
      "learning_rate": 0.0005784854951989335,
      "loss": 2.5789,
      "step": 27944
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6813894510269165,
      "learning_rate": 0.0005784839740173871,
      "loss": 2.4425,
      "step": 27945
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.8128301501274109,
      "learning_rate": 0.0005784824527840654,
      "loss": 2.5828,
      "step": 27946
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.7046841979026794,
      "learning_rate": 0.0005784809314989684,
      "loss": 2.2854,
      "step": 27947
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6932001709938049,
      "learning_rate": 0.0005784794101620966,
      "loss": 2.3344,
      "step": 27948
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.5996909737586975,
      "learning_rate": 0.0005784778887734502,
      "loss": 2.2103,
      "step": 27949
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6174036264419556,
      "learning_rate": 0.0005784763673330295,
      "loss": 2.434,
      "step": 27950
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6078070402145386,
      "learning_rate": 0.0005784748458408348,
      "loss": 2.1827,
      "step": 27951
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6619384288787842,
      "learning_rate": 0.0005784733242968664,
      "loss": 2.0449,
      "step": 27952
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.5980150699615479,
      "learning_rate": 0.0005784718027011245,
      "loss": 2.2714,
      "step": 27953
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6142206192016602,
      "learning_rate": 0.0005784702810536094,
      "loss": 2.3986,
      "step": 27954
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6189777851104736,
      "learning_rate": 0.0005784687593543216,
      "loss": 2.175,
      "step": 27955
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6712581515312195,
      "learning_rate": 0.000578467237603261,
      "loss": 2.4539,
      "step": 27956
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6218519806861877,
      "learning_rate": 0.0005784657158004283,
      "loss": 2.1816,
      "step": 27957
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6254136562347412,
      "learning_rate": 0.0005784641939458234,
      "loss": 2.2687,
      "step": 27958
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.7129623293876648,
      "learning_rate": 0.0005784626720394467,
      "loss": 2.474,
      "step": 27959
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6995939612388611,
      "learning_rate": 0.0005784611500812986,
      "loss": 2.3675,
      "step": 27960
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6517598628997803,
      "learning_rate": 0.0005784596280713793,
      "loss": 2.2434,
      "step": 27961
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.5784583687782288,
      "learning_rate": 0.0005784581060096891,
      "loss": 1.9746,
      "step": 27962
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6029255986213684,
      "learning_rate": 0.0005784565838962283,
      "loss": 2.2057,
      "step": 27963
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6791149973869324,
      "learning_rate": 0.0005784550617309972,
      "loss": 2.3463,
      "step": 27964
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.622114360332489,
      "learning_rate": 0.000578453539513996,
      "loss": 2.4445,
      "step": 27965
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6251111030578613,
      "learning_rate": 0.0005784520172452251,
      "loss": 2.3335,
      "step": 27966
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6579917073249817,
      "learning_rate": 0.0005784504949246847,
      "loss": 2.4601,
      "step": 27967
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.735559344291687,
      "learning_rate": 0.000578448972552375,
      "loss": 2.2645,
      "step": 27968
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6291172504425049,
      "learning_rate": 0.0005784474501282965,
      "loss": 2.3964,
      "step": 27969
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6253905296325684,
      "learning_rate": 0.0005784459276524492,
      "loss": 2.2541,
      "step": 27970
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6250237822532654,
      "learning_rate": 0.0005784444051248338,
      "loss": 2.4049,
      "step": 27971
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6526738405227661,
      "learning_rate": 0.0005784428825454501,
      "loss": 2.4203,
      "step": 27972
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6075162291526794,
      "learning_rate": 0.0005784413599142987,
      "loss": 2.2378,
      "step": 27973
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6324951648712158,
      "learning_rate": 0.0005784398372313797,
      "loss": 2.354,
      "step": 27974
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6005429625511169,
      "learning_rate": 0.0005784383144966937,
      "loss": 2.1948,
      "step": 27975
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.7083269357681274,
      "learning_rate": 0.0005784367917102406,
      "loss": 2.2813,
      "step": 27976
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.5945920944213867,
      "learning_rate": 0.0005784352688720208,
      "loss": 2.4977,
      "step": 27977
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6873810887336731,
      "learning_rate": 0.0005784337459820347,
      "loss": 2.1957,
      "step": 27978
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6434441804885864,
      "learning_rate": 0.0005784322230402826,
      "loss": 2.3947,
      "step": 27979
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6221089959144592,
      "learning_rate": 0.0005784307000467646,
      "loss": 2.1925,
      "step": 27980
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6501102447509766,
      "learning_rate": 0.0005784291770014811,
      "loss": 2.413,
      "step": 27981
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6084650158882141,
      "learning_rate": 0.0005784276539044323,
      "loss": 2.5013,
      "step": 27982
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.7027196288108826,
      "learning_rate": 0.0005784261307556185,
      "loss": 2.0318,
      "step": 27983
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6208900213241577,
      "learning_rate": 0.0005784246075550401,
      "loss": 2.3635,
      "step": 27984
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.5927395224571228,
      "learning_rate": 0.0005784230843026974,
      "loss": 2.2216,
      "step": 27985
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.5937693119049072,
      "learning_rate": 0.0005784215609985904,
      "loss": 2.306,
      "step": 27986
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6153709888458252,
      "learning_rate": 0.0005784200376427198,
      "loss": 2.2366,
      "step": 27987
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.693638265132904,
      "learning_rate": 0.0005784185142350855,
      "loss": 2.2012,
      "step": 27988
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.677189826965332,
      "learning_rate": 0.0005784169907756881,
      "loss": 2.4749,
      "step": 27989
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6176124215126038,
      "learning_rate": 0.0005784154672645275,
      "loss": 2.3261,
      "step": 27990
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.8603108525276184,
      "learning_rate": 0.0005784139437016043,
      "loss": 2.268,
      "step": 27991
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6631823778152466,
      "learning_rate": 0.0005784124200869187,
      "loss": 2.2361,
      "step": 27992
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6069934368133545,
      "learning_rate": 0.0005784108964204711,
      "loss": 2.3314,
      "step": 27993
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6544259786605835,
      "learning_rate": 0.0005784093727022616,
      "loss": 2.5782,
      "step": 27994
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6021704077720642,
      "learning_rate": 0.0005784078489322904,
      "loss": 2.3104,
      "step": 27995
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6579701900482178,
      "learning_rate": 0.000578406325110558,
      "loss": 2.1921,
      "step": 27996
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6795846223831177,
      "learning_rate": 0.0005784048012370647,
      "loss": 2.3388,
      "step": 27997
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.630495011806488,
      "learning_rate": 0.0005784032773118106,
      "loss": 2.229,
      "step": 27998
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.692944347858429,
      "learning_rate": 0.0005784017533347961,
      "loss": 2.5969,
      "step": 27999
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6922175884246826,
      "learning_rate": 0.0005784002293060214,
      "loss": 2.1664,
      "step": 28000
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6832270622253418,
      "learning_rate": 0.000578398705225487,
      "loss": 2.306,
      "step": 28001
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6118653416633606,
      "learning_rate": 0.0005783971810931929,
      "loss": 2.361,
      "step": 28002
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6324133276939392,
      "learning_rate": 0.0005783956569091395,
      "loss": 2.2262,
      "step": 28003
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.744517982006073,
      "learning_rate": 0.0005783941326733272,
      "loss": 2.4903,
      "step": 28004
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6705502867698669,
      "learning_rate": 0.000578392608385756,
      "loss": 2.5852,
      "step": 28005
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6013936996459961,
      "learning_rate": 0.0005783910840464265,
      "loss": 2.2174,
      "step": 28006
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6809194684028625,
      "learning_rate": 0.0005783895596553388,
      "loss": 2.0706,
      "step": 28007
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6238375902175903,
      "learning_rate": 0.0005783880352124932,
      "loss": 2.5799,
      "step": 28008
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6317022442817688,
      "learning_rate": 0.00057838651071789,
      "loss": 2.345,
      "step": 28009
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6741857528686523,
      "learning_rate": 0.0005783849861715296,
      "loss": 2.3389,
      "step": 28010
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6886394619941711,
      "learning_rate": 0.0005783834615734121,
      "loss": 2.469,
      "step": 28011
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.5383511781692505,
      "learning_rate": 0.0005783819369235379,
      "loss": 2.2826,
      "step": 28012
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6937966346740723,
      "learning_rate": 0.0005783804122219072,
      "loss": 2.4958,
      "step": 28013
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.7363414168357849,
      "learning_rate": 0.0005783788874685203,
      "loss": 2.4087,
      "step": 28014
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.719847559928894,
      "learning_rate": 0.0005783773626633776,
      "loss": 2.4944,
      "step": 28015
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6598153114318848,
      "learning_rate": 0.0005783758378064791,
      "loss": 2.512,
      "step": 28016
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.7394014596939087,
      "learning_rate": 0.0005783743128978256,
      "loss": 2.4109,
      "step": 28017
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6321637034416199,
      "learning_rate": 0.0005783727879374168,
      "loss": 2.3921,
      "step": 28018
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6356180906295776,
      "learning_rate": 0.0005783712629252533,
      "loss": 2.1714,
      "step": 28019
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6425736546516418,
      "learning_rate": 0.0005783697378613353,
      "loss": 2.3704,
      "step": 28020
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6632546782493591,
      "learning_rate": 0.0005783682127456632,
      "loss": 2.4916,
      "step": 28021
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.661798894405365,
      "learning_rate": 0.0005783666875782372,
      "loss": 2.4172,
      "step": 28022
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6705686450004578,
      "learning_rate": 0.0005783651623590575,
      "loss": 2.2094,
      "step": 28023
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6207029223442078,
      "learning_rate": 0.0005783636370881245,
      "loss": 2.1009,
      "step": 28024
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6120396256446838,
      "learning_rate": 0.0005783621117654385,
      "loss": 2.3031,
      "step": 28025
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6849578619003296,
      "learning_rate": 0.0005783605863909995,
      "loss": 2.4208,
      "step": 28026
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6274306774139404,
      "learning_rate": 0.0005783590609648083,
      "loss": 2.1219,
      "step": 28027
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.7057597637176514,
      "learning_rate": 0.0005783575354868647,
      "loss": 2.354,
      "step": 28028
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6594246029853821,
      "learning_rate": 0.0005783560099571692,
      "loss": 2.1961,
      "step": 28029
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6130130290985107,
      "learning_rate": 0.0005783544843757223,
      "loss": 2.2083,
      "step": 28030
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6198726892471313,
      "learning_rate": 0.0005783529587425238,
      "loss": 2.2907,
      "step": 28031
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.7834486961364746,
      "learning_rate": 0.0005783514330575743,
      "loss": 2.1681,
      "step": 28032
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6396796703338623,
      "learning_rate": 0.0005783499073208741,
      "loss": 2.5176,
      "step": 28033
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.8711926341056824,
      "learning_rate": 0.0005783483815324233,
      "loss": 2.3345,
      "step": 28034
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6333935260772705,
      "learning_rate": 0.0005783468556922222,
      "loss": 2.3212,
      "step": 28035
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6501927971839905,
      "learning_rate": 0.0005783453298002713,
      "loss": 2.4614,
      "step": 28036
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6063046455383301,
      "learning_rate": 0.0005783438038565707,
      "loss": 2.3289,
      "step": 28037
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6887978315353394,
      "learning_rate": 0.0005783422778611207,
      "loss": 2.4167,
      "step": 28038
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6219501495361328,
      "learning_rate": 0.0005783407518139217,
      "loss": 2.1878,
      "step": 28039
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.7042416930198669,
      "learning_rate": 0.0005783392257149738,
      "loss": 2.4081,
      "step": 28040
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6563705801963806,
      "learning_rate": 0.0005783376995642774,
      "loss": 2.4553,
      "step": 28041
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6683205962181091,
      "learning_rate": 0.0005783361733618329,
      "loss": 2.5032,
      "step": 28042
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6303480863571167,
      "learning_rate": 0.0005783346471076403,
      "loss": 2.4883,
      "step": 28043
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6518864035606384,
      "learning_rate": 0.0005783331208017001,
      "loss": 2.1147,
      "step": 28044
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6374754905700684,
      "learning_rate": 0.0005783315944440125,
      "loss": 2.448,
      "step": 28045
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.5977316498756409,
      "learning_rate": 0.0005783300680345778,
      "loss": 2.3777,
      "step": 28046
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.5897573232650757,
      "learning_rate": 0.0005783285415733964,
      "loss": 2.216,
      "step": 28047
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.7479609251022339,
      "learning_rate": 0.0005783270150604683,
      "loss": 2.3395,
      "step": 28048
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.7021430730819702,
      "learning_rate": 0.000578325488495794,
      "loss": 2.471,
      "step": 28049
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6002153158187866,
      "learning_rate": 0.0005783239618793738,
      "loss": 2.2842,
      "step": 28050
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6723382472991943,
      "learning_rate": 0.0005783224352112079,
      "loss": 2.4703,
      "step": 28051
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6626808047294617,
      "learning_rate": 0.0005783209084912965,
      "loss": 2.2843,
      "step": 28052
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.5890997052192688,
      "learning_rate": 0.0005783193817196402,
      "loss": 2.4781,
      "step": 28053
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6244314312934875,
      "learning_rate": 0.0005783178548962389,
      "loss": 2.1067,
      "step": 28054
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6604047417640686,
      "learning_rate": 0.0005783163280210931,
      "loss": 2.4116,
      "step": 28055
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6125998497009277,
      "learning_rate": 0.000578314801094203,
      "loss": 2.4888,
      "step": 28056
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6093679070472717,
      "learning_rate": 0.0005783132741155691,
      "loss": 2.2782,
      "step": 28057
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.647550642490387,
      "learning_rate": 0.0005783117470851913,
      "loss": 2.5344,
      "step": 28058
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6302580237388611,
      "learning_rate": 0.0005783102200030702,
      "loss": 2.4018,
      "step": 28059
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.7845141887664795,
      "learning_rate": 0.000578308692869206,
      "loss": 2.3368,
      "step": 28060
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6514440774917603,
      "learning_rate": 0.0005783071656835989,
      "loss": 2.1707,
      "step": 28061
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6393366456031799,
      "learning_rate": 0.0005783056384462492,
      "loss": 2.159,
      "step": 28062
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6277065873146057,
      "learning_rate": 0.0005783041111571573,
      "loss": 2.3048,
      "step": 28063
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.766321063041687,
      "learning_rate": 0.0005783025838163235,
      "loss": 2.2867,
      "step": 28064
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6437609195709229,
      "learning_rate": 0.0005783010564237479,
      "loss": 1.8751,
      "step": 28065
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6664299964904785,
      "learning_rate": 0.0005782995289794308,
      "loss": 2.2384,
      "step": 28066
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6146891713142395,
      "learning_rate": 0.0005782980014833727,
      "loss": 2.1477,
      "step": 28067
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.7451208829879761,
      "learning_rate": 0.0005782964739355737,
      "loss": 2.4297,
      "step": 28068
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6558067798614502,
      "learning_rate": 0.0005782949463360341,
      "loss": 2.1854,
      "step": 28069
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6818090081214905,
      "learning_rate": 0.0005782934186847542,
      "loss": 2.2861,
      "step": 28070
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.627577543258667,
      "learning_rate": 0.0005782918909817342,
      "loss": 2.2491,
      "step": 28071
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.5947011709213257,
      "learning_rate": 0.0005782903632269747,
      "loss": 2.2414,
      "step": 28072
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6435753107070923,
      "learning_rate": 0.0005782888354204757,
      "loss": 2.2743,
      "step": 28073
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6655462384223938,
      "learning_rate": 0.0005782873075622375,
      "loss": 2.2975,
      "step": 28074
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.7526729106903076,
      "learning_rate": 0.0005782857796522606,
      "loss": 2.4621,
      "step": 28075
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.5764285326004028,
      "learning_rate": 0.0005782842516905449,
      "loss": 2.1657,
      "step": 28076
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6331673264503479,
      "learning_rate": 0.0005782827236770911,
      "loss": 2.2128,
      "step": 28077
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6606079339981079,
      "learning_rate": 0.0005782811956118991,
      "loss": 2.3843,
      "step": 28078
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.7064279317855835,
      "learning_rate": 0.0005782796674949695,
      "loss": 2.3454,
      "step": 28079
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.7117314338684082,
      "learning_rate": 0.0005782781393263024,
      "loss": 2.2935,
      "step": 28080
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.7579371333122253,
      "learning_rate": 0.0005782766111058982,
      "loss": 2.3252,
      "step": 28081
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.9322913885116577,
      "learning_rate": 0.000578275082833757,
      "loss": 2.3806,
      "step": 28082
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6829087734222412,
      "learning_rate": 0.0005782735545098793,
      "loss": 2.3117,
      "step": 28083
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.7572195529937744,
      "learning_rate": 0.0005782720261342653,
      "loss": 2.3112,
      "step": 28084
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.7874695658683777,
      "learning_rate": 0.0005782704977069154,
      "loss": 2.4989,
      "step": 28085
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6456989049911499,
      "learning_rate": 0.0005782689692278296,
      "loss": 2.3265,
      "step": 28086
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.5986990928649902,
      "learning_rate": 0.0005782674406970083,
      "loss": 2.3341,
      "step": 28087
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6634830236434937,
      "learning_rate": 0.0005782659121144519,
      "loss": 2.2413,
      "step": 28088
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6199949383735657,
      "learning_rate": 0.0005782643834801607,
      "loss": 2.4858,
      "step": 28089
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6807820200920105,
      "learning_rate": 0.0005782628547941347,
      "loss": 2.2903,
      "step": 28090
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.5790101885795593,
      "learning_rate": 0.0005782613260563747,
      "loss": 2.5136,
      "step": 28091
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.702881932258606,
      "learning_rate": 0.0005782597972668804,
      "loss": 2.3393,
      "step": 28092
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6835522651672363,
      "learning_rate": 0.0005782582684256525,
      "loss": 2.4574,
      "step": 28093
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.5940566062927246,
      "learning_rate": 0.0005782567395326909,
      "loss": 2.2935,
      "step": 28094
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6131184101104736,
      "learning_rate": 0.0005782552105879963,
      "loss": 2.2504,
      "step": 28095
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6742697358131409,
      "learning_rate": 0.0005782536815915688,
      "loss": 2.1927,
      "step": 28096
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6536211371421814,
      "learning_rate": 0.0005782521525434087,
      "loss": 2.3643,
      "step": 28097
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6115769147872925,
      "learning_rate": 0.0005782506234435162,
      "loss": 2.4115,
      "step": 28098
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6078362464904785,
      "learning_rate": 0.0005782490942918916,
      "loss": 2.2729,
      "step": 28099
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6705940365791321,
      "learning_rate": 0.0005782475650885355,
      "loss": 2.4776,
      "step": 28100
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.5854184031486511,
      "learning_rate": 0.0005782460358334477,
      "loss": 2.3862,
      "step": 28101
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6290711760520935,
      "learning_rate": 0.0005782445065266287,
      "loss": 2.1892,
      "step": 28102
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6739020943641663,
      "learning_rate": 0.000578242977168079,
      "loss": 2.2619,
      "step": 28103
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6512495279312134,
      "learning_rate": 0.0005782414477577986,
      "loss": 2.4382,
      "step": 28104
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6090394258499146,
      "learning_rate": 0.0005782399182957876,
      "loss": 2.5398,
      "step": 28105
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6291326284408569,
      "learning_rate": 0.0005782383887820469,
      "loss": 2.1581,
      "step": 28106
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.0389047861099243,
      "learning_rate": 0.0005782368592165762,
      "loss": 2.5033,
      "step": 28107
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6785077452659607,
      "learning_rate": 0.0005782353295993761,
      "loss": 2.4709,
      "step": 28108
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6907792091369629,
      "learning_rate": 0.0005782337999304468,
      "loss": 2.4636,
      "step": 28109
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6381286382675171,
      "learning_rate": 0.0005782322702097886,
      "loss": 2.5255,
      "step": 28110
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.5944383144378662,
      "learning_rate": 0.0005782307404374018,
      "loss": 2.3211,
      "step": 28111
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6046452522277832,
      "learning_rate": 0.0005782292106132866,
      "loss": 2.3885,
      "step": 28112
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.5902220010757446,
      "learning_rate": 0.0005782276807374434,
      "loss": 2.3128,
      "step": 28113
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6548307538032532,
      "learning_rate": 0.0005782261508098723,
      "loss": 2.4226,
      "step": 28114
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.0632994174957275,
      "learning_rate": 0.0005782246208305737,
      "loss": 2.0942,
      "step": 28115
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.577899694442749,
      "learning_rate": 0.000578223090799548,
      "loss": 2.2655,
      "step": 28116
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6602180600166321,
      "learning_rate": 0.0005782215607167953,
      "loss": 2.2917,
      "step": 28117
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6769029498100281,
      "learning_rate": 0.0005782200305823161,
      "loss": 2.3361,
      "step": 28118
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6385593414306641,
      "learning_rate": 0.0005782185003961104,
      "loss": 2.0831,
      "step": 28119
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6007773280143738,
      "learning_rate": 0.0005782169701581787,
      "loss": 2.2528,
      "step": 28120
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.641244113445282,
      "learning_rate": 0.0005782154398685211,
      "loss": 2.2948,
      "step": 28121
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6933686137199402,
      "learning_rate": 0.0005782139095271381,
      "loss": 2.3487,
      "step": 28122
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.5796805024147034,
      "learning_rate": 0.00057821237913403,
      "loss": 2.3506,
      "step": 28123
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.654240071773529,
      "learning_rate": 0.0005782108486891968,
      "loss": 2.2286,
      "step": 28124
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6836888790130615,
      "learning_rate": 0.000578209318192639,
      "loss": 2.1856,
      "step": 28125
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6429418921470642,
      "learning_rate": 0.0005782077876443568,
      "loss": 2.1995,
      "step": 28126
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.7482316493988037,
      "learning_rate": 0.0005782062570443505,
      "loss": 2.2445,
      "step": 28127
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6348981261253357,
      "learning_rate": 0.0005782047263926204,
      "loss": 2.4869,
      "step": 28128
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.7447454333305359,
      "learning_rate": 0.000578203195689167,
      "loss": 2.2806,
      "step": 28129
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.675654411315918,
      "learning_rate": 0.0005782016649339901,
      "loss": 2.1883,
      "step": 28130
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6613315939903259,
      "learning_rate": 0.0005782001341270905,
      "loss": 2.3285,
      "step": 28131
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6873018741607666,
      "learning_rate": 0.0005781986032684681,
      "loss": 2.3504,
      "step": 28132
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.7084499597549438,
      "learning_rate": 0.0005781970723581233,
      "loss": 2.1754,
      "step": 28133
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.5984115600585938,
      "learning_rate": 0.0005781955413960564,
      "loss": 2.3033,
      "step": 28134
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6848400235176086,
      "learning_rate": 0.0005781940103822677,
      "loss": 2.2652,
      "step": 28135
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6913815140724182,
      "learning_rate": 0.0005781924793167577,
      "loss": 2.3037,
      "step": 28136
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.5977043509483337,
      "learning_rate": 0.0005781909481995263,
      "loss": 2.2465,
      "step": 28137
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.633646547794342,
      "learning_rate": 0.0005781894170305739,
      "loss": 2.5657,
      "step": 28138
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.7384877800941467,
      "learning_rate": 0.0005781878858099009,
      "loss": 2.2373,
      "step": 28139
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.7061854004859924,
      "learning_rate": 0.0005781863545375075,
      "loss": 2.5636,
      "step": 28140
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6207507848739624,
      "learning_rate": 0.000578184823213394,
      "loss": 2.2308,
      "step": 28141
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6676060557365417,
      "learning_rate": 0.0005781832918375608,
      "loss": 2.3375,
      "step": 28142
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6559650897979736,
      "learning_rate": 0.0005781817604100078,
      "loss": 2.3632,
      "step": 28143
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6125949621200562,
      "learning_rate": 0.0005781802289307359,
      "loss": 2.3896,
      "step": 28144
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.5653013586997986,
      "learning_rate": 0.0005781786973997449,
      "loss": 2.3621,
      "step": 28145
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.620373010635376,
      "learning_rate": 0.0005781771658170352,
      "loss": 2.2808,
      "step": 28146
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6047700643539429,
      "learning_rate": 0.000578175634182607,
      "loss": 2.5831,
      "step": 28147
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6263941526412964,
      "learning_rate": 0.0005781741024964608,
      "loss": 2.2947,
      "step": 28148
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.5865669846534729,
      "learning_rate": 0.0005781725707585969,
      "loss": 2.1626,
      "step": 28149
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6867908835411072,
      "learning_rate": 0.0005781710389690153,
      "loss": 2.3711,
      "step": 28150
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6409068703651428,
      "learning_rate": 0.0005781695071277167,
      "loss": 2.2053,
      "step": 28151
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.638980507850647,
      "learning_rate": 0.0005781679752347008,
      "loss": 2.3613,
      "step": 28152
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6246773600578308,
      "learning_rate": 0.0005781664432899685,
      "loss": 2.3447,
      "step": 28153
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.591778039932251,
      "learning_rate": 0.0005781649112935197,
      "loss": 2.3417,
      "step": 28154
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6232059597969055,
      "learning_rate": 0.0005781633792453549,
      "loss": 2.3276,
      "step": 28155
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6046044230461121,
      "learning_rate": 0.0005781618471454741,
      "loss": 2.2839,
      "step": 28156
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6692380905151367,
      "learning_rate": 0.0005781603149938779,
      "loss": 2.3169,
      "step": 28157
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6340814232826233,
      "learning_rate": 0.0005781587827905663,
      "loss": 2.1179,
      "step": 28158
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6474685072898865,
      "learning_rate": 0.0005781572505355398,
      "loss": 2.2212,
      "step": 28159
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6155269145965576,
      "learning_rate": 0.0005781557182287986,
      "loss": 2.3005,
      "step": 28160
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6924031376838684,
      "learning_rate": 0.000578154185870343,
      "loss": 2.1217,
      "step": 28161
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6739959120750427,
      "learning_rate": 0.0005781526534601733,
      "loss": 2.5004,
      "step": 28162
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.62598717212677,
      "learning_rate": 0.0005781511209982898,
      "loss": 2.4093,
      "step": 28163
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6226180195808411,
      "learning_rate": 0.0005781495884846927,
      "loss": 2.1046,
      "step": 28164
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6660616993904114,
      "learning_rate": 0.0005781480559193825,
      "loss": 2.4441,
      "step": 28165
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6295341849327087,
      "learning_rate": 0.0005781465233023592,
      "loss": 2.3625,
      "step": 28166
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.62281733751297,
      "learning_rate": 0.0005781449906336231,
      "loss": 2.3548,
      "step": 28167
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6252296566963196,
      "learning_rate": 0.0005781434579131749,
      "loss": 2.414,
      "step": 28168
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6546737551689148,
      "learning_rate": 0.0005781419251410143,
      "loss": 2.3563,
      "step": 28169
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6319957375526428,
      "learning_rate": 0.000578140392317142,
      "loss": 2.3572,
      "step": 28170
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6742763519287109,
      "learning_rate": 0.0005781388594415581,
      "loss": 2.2963,
      "step": 28171
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6429132223129272,
      "learning_rate": 0.0005781373265142629,
      "loss": 2.2541,
      "step": 28172
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6243970394134521,
      "learning_rate": 0.0005781357935352567,
      "loss": 2.306,
      "step": 28173
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6657538414001465,
      "learning_rate": 0.00057813426050454,
      "loss": 2.3856,
      "step": 28174
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6741505265235901,
      "learning_rate": 0.0005781327274221128,
      "loss": 2.2076,
      "step": 28175
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6045727729797363,
      "learning_rate": 0.0005781311942879753,
      "loss": 2.1604,
      "step": 28176
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6121450066566467,
      "learning_rate": 0.0005781296611021282,
      "loss": 2.4273,
      "step": 28177
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.628275454044342,
      "learning_rate": 0.0005781281278645715,
      "loss": 2.1732,
      "step": 28178
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.615247905254364,
      "learning_rate": 0.0005781265945753054,
      "loss": 2.3699,
      "step": 28179
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.5716782808303833,
      "learning_rate": 0.0005781250612343304,
      "loss": 2.2743,
      "step": 28180
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.5899711847305298,
      "learning_rate": 0.0005781235278416467,
      "loss": 2.3498,
      "step": 28181
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6161669492721558,
      "learning_rate": 0.0005781219943972547,
      "loss": 2.5009,
      "step": 28182
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6673171520233154,
      "learning_rate": 0.0005781204609011544,
      "loss": 2.2676,
      "step": 28183
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6239805817604065,
      "learning_rate": 0.0005781189273533462,
      "loss": 2.2037,
      "step": 28184
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6530880928039551,
      "learning_rate": 0.0005781173937538306,
      "loss": 2.2556,
      "step": 28185
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.5988574624061584,
      "learning_rate": 0.0005781158601026078,
      "loss": 2.3078,
      "step": 28186
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6481305956840515,
      "learning_rate": 0.0005781143263996777,
      "loss": 2.3454,
      "step": 28187
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6814941167831421,
      "learning_rate": 0.0005781127926450412,
      "loss": 2.484,
      "step": 28188
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.7062338590621948,
      "learning_rate": 0.0005781112588386982,
      "loss": 2.2303,
      "step": 28189
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6657737493515015,
      "learning_rate": 0.000578109724980649,
      "loss": 2.4056,
      "step": 28190
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6858015060424805,
      "learning_rate": 0.0005781081910708939,
      "loss": 2.2519,
      "step": 28191
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6176591515541077,
      "learning_rate": 0.0005781066571094334,
      "loss": 2.3679,
      "step": 28192
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.7100871205329895,
      "learning_rate": 0.0005781051230962675,
      "loss": 2.5482,
      "step": 28193
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.8508296012878418,
      "learning_rate": 0.0005781035890313967,
      "loss": 2.1462,
      "step": 28194
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6429218053817749,
      "learning_rate": 0.0005781020549148212,
      "loss": 2.3324,
      "step": 28195
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6316356658935547,
      "learning_rate": 0.0005781005207465412,
      "loss": 2.4273,
      "step": 28196
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.7790406346321106,
      "learning_rate": 0.000578098986526557,
      "loss": 2.3376,
      "step": 28197
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6864020824432373,
      "learning_rate": 0.000578097452254869,
      "loss": 2.6605,
      "step": 28198
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.5757447481155396,
      "learning_rate": 0.0005780959179314775,
      "loss": 2.1659,
      "step": 28199
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.8883839845657349,
      "learning_rate": 0.0005780943835563828,
      "loss": 2.4965,
      "step": 28200
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.0253416299819946,
      "learning_rate": 0.000578092849129585,
      "loss": 2.5399,
      "step": 28201
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6442285776138306,
      "learning_rate": 0.0005780913146510845,
      "loss": 2.4752,
      "step": 28202
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.7574382424354553,
      "learning_rate": 0.0005780897801208815,
      "loss": 2.2578,
      "step": 28203
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.7588161826133728,
      "learning_rate": 0.0005780882455389765,
      "loss": 2.6455,
      "step": 28204
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.8240292072296143,
      "learning_rate": 0.0005780867109053695,
      "loss": 2.4472,
      "step": 28205
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.635283887386322,
      "learning_rate": 0.0005780851762200611,
      "loss": 2.4192,
      "step": 28206
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.8773770928382874,
      "learning_rate": 0.0005780836414830512,
      "loss": 2.448,
      "step": 28207
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.018059253692627,
      "learning_rate": 0.0005780821066943405,
      "loss": 2.324,
      "step": 28208
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6409215927124023,
      "learning_rate": 0.0005780805718539289,
      "loss": 2.3376,
      "step": 28209
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.7017132639884949,
      "learning_rate": 0.0005780790369618171,
      "loss": 2.2926,
      "step": 28210
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.8311566114425659,
      "learning_rate": 0.0005780775020180049,
      "loss": 2.3066,
      "step": 28211
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.8108379244804382,
      "learning_rate": 0.0005780759670224931,
      "loss": 2.3475,
      "step": 28212
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6730977892875671,
      "learning_rate": 0.0005780744319752816,
      "loss": 2.4912,
      "step": 28213
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.8026037812232971,
      "learning_rate": 0.0005780728968763708,
      "loss": 2.3083,
      "step": 28214
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6476418972015381,
      "learning_rate": 0.0005780713617257611,
      "loss": 2.0528,
      "step": 28215
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.0894713401794434,
      "learning_rate": 0.0005780698265234526,
      "loss": 2.774,
      "step": 28216
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.679326057434082,
      "learning_rate": 0.0005780682912694457,
      "loss": 2.2322,
      "step": 28217
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6918943524360657,
      "learning_rate": 0.0005780667559637405,
      "loss": 2.1029,
      "step": 28218
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6205414533615112,
      "learning_rate": 0.0005780652206063376,
      "loss": 2.3578,
      "step": 28219
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6555792093276978,
      "learning_rate": 0.0005780636851972372,
      "loss": 2.3204,
      "step": 28220
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.7318408489227295,
      "learning_rate": 0.0005780621497364393,
      "loss": 2.5114,
      "step": 28221
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6605221629142761,
      "learning_rate": 0.0005780606142239446,
      "loss": 2.4241,
      "step": 28222
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6370821595191956,
      "learning_rate": 0.0005780590786597531,
      "loss": 2.5333,
      "step": 28223
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6582337021827698,
      "learning_rate": 0.0005780575430438652,
      "loss": 2.3932,
      "step": 28224
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.5633981823921204,
      "learning_rate": 0.0005780560073762811,
      "loss": 2.1363,
      "step": 28225
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6127316951751709,
      "learning_rate": 0.0005780544716570012,
      "loss": 2.3301,
      "step": 28226
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6447532773017883,
      "learning_rate": 0.0005780529358860256,
      "loss": 2.3404,
      "step": 28227
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.633236289024353,
      "learning_rate": 0.0005780514000633547,
      "loss": 2.3314,
      "step": 28228
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6425197124481201,
      "learning_rate": 0.0005780498641889889,
      "loss": 2.2099,
      "step": 28229
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6574013829231262,
      "learning_rate": 0.0005780483282629285,
      "loss": 2.3047,
      "step": 28230
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6855834126472473,
      "learning_rate": 0.0005780467922851735,
      "loss": 2.3589,
      "step": 28231
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.5842131972312927,
      "learning_rate": 0.0005780452562557243,
      "loss": 2.2234,
      "step": 28232
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6331129670143127,
      "learning_rate": 0.0005780437201745814,
      "loss": 2.2066,
      "step": 28233
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.7984635233879089,
      "learning_rate": 0.0005780421840417449,
      "loss": 2.4595,
      "step": 28234
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.637274980545044,
      "learning_rate": 0.0005780406478572151,
      "loss": 2.0985,
      "step": 28235
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6371493339538574,
      "learning_rate": 0.0005780391116209921,
      "loss": 2.3078,
      "step": 28236
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6328555941581726,
      "learning_rate": 0.0005780375753330766,
      "loss": 2.2099,
      "step": 28237
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.636613667011261,
      "learning_rate": 0.0005780360389934687,
      "loss": 2.5015,
      "step": 28238
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6229625344276428,
      "learning_rate": 0.0005780345026021685,
      "loss": 2.323,
      "step": 28239
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.5896660685539246,
      "learning_rate": 0.0005780329661591765,
      "loss": 2.4712,
      "step": 28240
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6543837189674377,
      "learning_rate": 0.0005780314296644929,
      "loss": 2.457,
      "step": 28241
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.7023831009864807,
      "learning_rate": 0.0005780298931181181,
      "loss": 2.2265,
      "step": 28242
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6370149850845337,
      "learning_rate": 0.0005780283565200522,
      "loss": 2.213,
      "step": 28243
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6411039233207703,
      "learning_rate": 0.0005780268198702955,
      "loss": 2.1092,
      "step": 28244
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6418254971504211,
      "learning_rate": 0.0005780252831688485,
      "loss": 2.205,
      "step": 28245
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6013768911361694,
      "learning_rate": 0.0005780237464157113,
      "loss": 2.26,
      "step": 28246
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6964576244354248,
      "learning_rate": 0.0005780222096108842,
      "loss": 2.1863,
      "step": 28247
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6821284294128418,
      "learning_rate": 0.0005780206727543677,
      "loss": 2.1212,
      "step": 28248
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6001781821250916,
      "learning_rate": 0.0005780191358461617,
      "loss": 2.1663,
      "step": 28249
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.8125172853469849,
      "learning_rate": 0.0005780175988862668,
      "loss": 2.416,
      "step": 28250
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.7055167555809021,
      "learning_rate": 0.0005780160618746832,
      "loss": 2.4366,
      "step": 28251
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.7502137422561646,
      "learning_rate": 0.0005780145248114111,
      "loss": 2.4972,
      "step": 28252
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6391963362693787,
      "learning_rate": 0.0005780129876964509,
      "loss": 2.2712,
      "step": 28253
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.666198194026947,
      "learning_rate": 0.0005780114505298028,
      "loss": 2.4654,
      "step": 28254
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6585684418678284,
      "learning_rate": 0.0005780099133114671,
      "loss": 2.1935,
      "step": 28255
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.7366546392440796,
      "learning_rate": 0.0005780083760414442,
      "loss": 2.2675,
      "step": 28256
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6270344257354736,
      "learning_rate": 0.0005780068387197342,
      "loss": 2.2511,
      "step": 28257
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.663452684879303,
      "learning_rate": 0.0005780053013463377,
      "loss": 2.4644,
      "step": 28258
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6914403438568115,
      "learning_rate": 0.0005780037639212545,
      "loss": 2.1766,
      "step": 28259
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.5986143350601196,
      "learning_rate": 0.0005780022264444853,
      "loss": 2.363,
      "step": 28260
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.5995640158653259,
      "learning_rate": 0.0005780006889160301,
      "loss": 2.323,
      "step": 28261
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6021575927734375,
      "learning_rate": 0.0005779991513358894,
      "loss": 2.3285,
      "step": 28262
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6217944622039795,
      "learning_rate": 0.0005779976137040635,
      "loss": 2.259,
      "step": 28263
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.61638343334198,
      "learning_rate": 0.0005779960760205525,
      "loss": 2.2083,
      "step": 28264
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6466181874275208,
      "learning_rate": 0.0005779945382853568,
      "loss": 2.2662,
      "step": 28265
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6594060063362122,
      "learning_rate": 0.0005779930004984766,
      "loss": 2.4678,
      "step": 28266
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.8305715918540955,
      "learning_rate": 0.0005779914626599124,
      "loss": 2.3786,
      "step": 28267
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6156581044197083,
      "learning_rate": 0.0005779899247696642,
      "loss": 2.3648,
      "step": 28268
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6256808638572693,
      "learning_rate": 0.0005779883868277326,
      "loss": 2.3722,
      "step": 28269
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.5868071913719177,
      "learning_rate": 0.0005779868488341175,
      "loss": 2.2759,
      "step": 28270
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6123999357223511,
      "learning_rate": 0.0005779853107888195,
      "loss": 2.1943,
      "step": 28271
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.7070750594139099,
      "learning_rate": 0.0005779837726918388,
      "loss": 2.4668,
      "step": 28272
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6603877544403076,
      "learning_rate": 0.0005779822345431756,
      "loss": 2.1912,
      "step": 28273
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6342329978942871,
      "learning_rate": 0.0005779806963428303,
      "loss": 2.2701,
      "step": 28274
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.589604377746582,
      "learning_rate": 0.0005779791580908032,
      "loss": 2.3277,
      "step": 28275
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6661734580993652,
      "learning_rate": 0.0005779776197870944,
      "loss": 2.4822,
      "step": 28276
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6408271789550781,
      "learning_rate": 0.0005779760814317044,
      "loss": 2.3893,
      "step": 28277
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6158925294876099,
      "learning_rate": 0.0005779745430246334,
      "loss": 2.3359,
      "step": 28278
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6187160611152649,
      "learning_rate": 0.0005779730045658817,
      "loss": 2.1514,
      "step": 28279
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6366308927536011,
      "learning_rate": 0.0005779714660554495,
      "loss": 2.2825,
      "step": 28280
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6593812108039856,
      "learning_rate": 0.0005779699274933372,
      "loss": 2.4205,
      "step": 28281
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6486663222312927,
      "learning_rate": 0.000577968388879545,
      "loss": 2.3838,
      "step": 28282
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.589171290397644,
      "learning_rate": 0.0005779668502140733,
      "loss": 2.1843,
      "step": 28283
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6290215849876404,
      "learning_rate": 0.0005779653114969223,
      "loss": 2.0655,
      "step": 28284
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6660401225090027,
      "learning_rate": 0.0005779637727280924,
      "loss": 2.2346,
      "step": 28285
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6111165285110474,
      "learning_rate": 0.0005779622339075835,
      "loss": 2.2087,
      "step": 28286
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6629317402839661,
      "learning_rate": 0.0005779606950353964,
      "loss": 2.4173,
      "step": 28287
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6327890157699585,
      "learning_rate": 0.0005779591561115311,
      "loss": 2.3035,
      "step": 28288
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.660221517086029,
      "learning_rate": 0.000577957617135988,
      "loss": 2.422,
      "step": 28289
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6144137382507324,
      "learning_rate": 0.0005779560781087673,
      "loss": 2.4701,
      "step": 28290
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.5915302038192749,
      "learning_rate": 0.0005779545390298692,
      "loss": 2.4073,
      "step": 28291
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6862678527832031,
      "learning_rate": 0.0005779529998992942,
      "loss": 2.3581,
      "step": 28292
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6533157825469971,
      "learning_rate": 0.0005779514607170426,
      "loss": 2.3484,
      "step": 28293
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6147283315658569,
      "learning_rate": 0.0005779499214831143,
      "loss": 2.192,
      "step": 28294
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6558911204338074,
      "learning_rate": 0.0005779483821975102,
      "loss": 2.238,
      "step": 28295
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.7184257507324219,
      "learning_rate": 0.00057794684286023,
      "loss": 2.3492,
      "step": 28296
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.71868497133255,
      "learning_rate": 0.0005779453034712744,
      "loss": 2.2041,
      "step": 28297
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.5887919664382935,
      "learning_rate": 0.0005779437640306435,
      "loss": 2.179,
      "step": 28298
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.7934316396713257,
      "learning_rate": 0.0005779422245383375,
      "loss": 2.1778,
      "step": 28299
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.7136216759681702,
      "learning_rate": 0.0005779406849943568,
      "loss": 2.2329,
      "step": 28300
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6385284662246704,
      "learning_rate": 0.0005779391453987017,
      "loss": 2.3827,
      "step": 28301
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6757574081420898,
      "learning_rate": 0.0005779376057513725,
      "loss": 2.0442,
      "step": 28302
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6441436409950256,
      "learning_rate": 0.0005779360660523696,
      "loss": 2.4406,
      "step": 28303
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.619361937046051,
      "learning_rate": 0.0005779345263016929,
      "loss": 2.25,
      "step": 28304
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6774872541427612,
      "learning_rate": 0.0005779329864993429,
      "loss": 2.4395,
      "step": 28305
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6276401281356812,
      "learning_rate": 0.0005779314466453201,
      "loss": 2.4482,
      "step": 28306
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6934226155281067,
      "learning_rate": 0.0005779299067396245,
      "loss": 2.3771,
      "step": 28307
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6847056746482849,
      "learning_rate": 0.0005779283667822565,
      "loss": 2.2912,
      "step": 28308
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.7043931484222412,
      "learning_rate": 0.0005779268267732165,
      "loss": 2.4342,
      "step": 28309
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6372161507606506,
      "learning_rate": 0.0005779252867125044,
      "loss": 2.4021,
      "step": 28310
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.801455557346344,
      "learning_rate": 0.000577923746600121,
      "loss": 2.1,
      "step": 28311
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6168875098228455,
      "learning_rate": 0.0005779222064360661,
      "loss": 2.253,
      "step": 28312
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.7007005214691162,
      "learning_rate": 0.0005779206662203404,
      "loss": 2.4307,
      "step": 28313
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6078698039054871,
      "learning_rate": 0.0005779191259529438,
      "loss": 2.3723,
      "step": 28314
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6555256843566895,
      "learning_rate": 0.000577917585633877,
      "loss": 2.4354,
      "step": 28315
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6079689860343933,
      "learning_rate": 0.00057791604526314,
      "loss": 2.2238,
      "step": 28316
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6242054104804993,
      "learning_rate": 0.000577914504840733,
      "loss": 2.269,
      "step": 28317
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6228974461555481,
      "learning_rate": 0.0005779129643666567,
      "loss": 2.2417,
      "step": 28318
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6874188184738159,
      "learning_rate": 0.000577911423840911,
      "loss": 2.1369,
      "step": 28319
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6116076707839966,
      "learning_rate": 0.0005779098832634963,
      "loss": 2.3563,
      "step": 28320
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6668540239334106,
      "learning_rate": 0.000577908342634413,
      "loss": 2.3306,
      "step": 28321
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.665532648563385,
      "learning_rate": 0.0005779068019536612,
      "loss": 2.5353,
      "step": 28322
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6502915620803833,
      "learning_rate": 0.0005779052612212413,
      "loss": 2.3399,
      "step": 28323
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6160585880279541,
      "learning_rate": 0.0005779037204371537,
      "loss": 2.4944,
      "step": 28324
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.8186346292495728,
      "learning_rate": 0.0005779021796013985,
      "loss": 2.4841,
      "step": 28325
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6371109485626221,
      "learning_rate": 0.0005779006387139758,
      "loss": 2.3807,
      "step": 28326
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.771499514579773,
      "learning_rate": 0.0005778990977748864,
      "loss": 2.4193,
      "step": 28327
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.7229793071746826,
      "learning_rate": 0.0005778975567841303,
      "loss": 2.3577,
      "step": 28328
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.563615083694458,
      "learning_rate": 0.0005778960157417076,
      "loss": 2.3687,
      "step": 28329
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.7710747122764587,
      "learning_rate": 0.0005778944746476189,
      "loss": 2.3776,
      "step": 28330
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.80335932970047,
      "learning_rate": 0.0005778929335018644,
      "loss": 2.272,
      "step": 28331
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6566783785820007,
      "learning_rate": 0.0005778913923044443,
      "loss": 2.5104,
      "step": 28332
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6262904405593872,
      "learning_rate": 0.000577889851055359,
      "loss": 2.2044,
      "step": 28333
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.8300247192382812,
      "learning_rate": 0.0005778883097546087,
      "loss": 2.2785,
      "step": 28334
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.938463032245636,
      "learning_rate": 0.0005778867684021938,
      "loss": 2.1592,
      "step": 28335
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6569384336471558,
      "learning_rate": 0.0005778852269981143,
      "loss": 2.3363,
      "step": 28336
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.090153455734253,
      "learning_rate": 0.0005778836855423709,
      "loss": 2.4173,
      "step": 28337
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6375340223312378,
      "learning_rate": 0.0005778821440349636,
      "loss": 2.4742,
      "step": 28338
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6594363451004028,
      "learning_rate": 0.0005778806024758928,
      "loss": 2.3349,
      "step": 28339
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.7861872911453247,
      "learning_rate": 0.0005778790608651586,
      "loss": 2.3037,
      "step": 28340
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.7032135725021362,
      "learning_rate": 0.0005778775192027616,
      "loss": 2.3234,
      "step": 28341
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6103619933128357,
      "learning_rate": 0.0005778759774887019,
      "loss": 2.4522,
      "step": 28342
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6466366052627563,
      "learning_rate": 0.0005778744357229798,
      "loss": 2.3671,
      "step": 28343
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6248505711555481,
      "learning_rate": 0.0005778728939055956,
      "loss": 2.2383,
      "step": 28344
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.5982246994972229,
      "learning_rate": 0.0005778713520365495,
      "loss": 2.3798,
      "step": 28345
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.602198600769043,
      "learning_rate": 0.0005778698101158419,
      "loss": 2.2262,
      "step": 28346
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.669860303401947,
      "learning_rate": 0.0005778682681434732,
      "loss": 2.2936,
      "step": 28347
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.670619010925293,
      "learning_rate": 0.0005778667261194434,
      "loss": 2.2947,
      "step": 28348
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6366400122642517,
      "learning_rate": 0.0005778651840437529,
      "loss": 1.9852,
      "step": 28349
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6434930562973022,
      "learning_rate": 0.0005778636419164022,
      "loss": 2.351,
      "step": 28350
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.610363781452179,
      "learning_rate": 0.0005778620997373913,
      "loss": 2.3874,
      "step": 28351
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.7604305744171143,
      "learning_rate": 0.0005778605575067205,
      "loss": 2.2189,
      "step": 28352
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.8177043199539185,
      "learning_rate": 0.0005778590152243902,
      "loss": 2.407,
      "step": 28353
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6861039400100708,
      "learning_rate": 0.0005778574728904008,
      "loss": 2.3996,
      "step": 28354
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6493266224861145,
      "learning_rate": 0.0005778559305047523,
      "loss": 2.5557,
      "step": 28355
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.675594687461853,
      "learning_rate": 0.0005778543880674453,
      "loss": 2.2383,
      "step": 28356
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.66996169090271,
      "learning_rate": 0.0005778528455784797,
      "loss": 2.2531,
      "step": 28357
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.673852264881134,
      "learning_rate": 0.0005778513030378562,
      "loss": 2.1953,
      "step": 28358
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.7144457697868347,
      "learning_rate": 0.0005778497604455749,
      "loss": 2.3064,
      "step": 28359
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6332646012306213,
      "learning_rate": 0.0005778482178016359,
      "loss": 2.1335,
      "step": 28360
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.7143704891204834,
      "learning_rate": 0.0005778466751060399,
      "loss": 2.1567,
      "step": 28361
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.7301064729690552,
      "learning_rate": 0.0005778451323587869,
      "loss": 2.3453,
      "step": 28362
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.5861207842826843,
      "learning_rate": 0.0005778435895598772,
      "loss": 2.3663,
      "step": 28363
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.7305867075920105,
      "learning_rate": 0.0005778420467093111,
      "loss": 2.3741,
      "step": 28364
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6823252439498901,
      "learning_rate": 0.000577840503807089,
      "loss": 2.2151,
      "step": 28365
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6653623580932617,
      "learning_rate": 0.000577838960853211,
      "loss": 2.4782,
      "step": 28366
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6381778717041016,
      "learning_rate": 0.0005778374178476775,
      "loss": 2.4695,
      "step": 28367
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6928441524505615,
      "learning_rate": 0.0005778358747904889,
      "loss": 2.2047,
      "step": 28368
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.7156713604927063,
      "learning_rate": 0.0005778343316816453,
      "loss": 2.2113,
      "step": 28369
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.7226927876472473,
      "learning_rate": 0.0005778327885211469,
      "loss": 2.2123,
      "step": 28370
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6815701127052307,
      "learning_rate": 0.0005778312453089944,
      "loss": 2.5087,
      "step": 28371
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6802576780319214,
      "learning_rate": 0.0005778297020451878,
      "loss": 2.2441,
      "step": 28372
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6747092604637146,
      "learning_rate": 0.0005778281587297272,
      "loss": 2.3494,
      "step": 28373
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.64385986328125,
      "learning_rate": 0.0005778266153626133,
      "loss": 2.4443,
      "step": 28374
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6769381761550903,
      "learning_rate": 0.0005778250719438461,
      "loss": 2.2647,
      "step": 28375
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6511815190315247,
      "learning_rate": 0.0005778235284734259,
      "loss": 2.2003,
      "step": 28376
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6231364607810974,
      "learning_rate": 0.0005778219849513533,
      "loss": 2.2753,
      "step": 28377
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.5915906429290771,
      "learning_rate": 0.0005778204413776282,
      "loss": 2.3128,
      "step": 28378
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6290453672409058,
      "learning_rate": 0.000577818897752251,
      "loss": 2.3894,
      "step": 28379
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6158751845359802,
      "learning_rate": 0.000577817354075222,
      "loss": 2.4382,
      "step": 28380
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.7126724123954773,
      "learning_rate": 0.0005778158103465417,
      "loss": 2.3461,
      "step": 28381
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.7191523909568787,
      "learning_rate": 0.00057781426656621,
      "loss": 2.2574,
      "step": 28382
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6370009183883667,
      "learning_rate": 0.0005778127227342274,
      "loss": 2.3757,
      "step": 28383
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.7104634642601013,
      "learning_rate": 0.0005778111788505943,
      "loss": 2.2309,
      "step": 28384
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.7164820432662964,
      "learning_rate": 0.0005778096349153107,
      "loss": 2.4746,
      "step": 28385
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.7137026786804199,
      "learning_rate": 0.0005778080909283772,
      "loss": 2.3086,
      "step": 28386
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6392715573310852,
      "learning_rate": 0.0005778065468897939,
      "loss": 2.3827,
      "step": 28387
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6054926514625549,
      "learning_rate": 0.000577805002799561,
      "loss": 2.0939,
      "step": 28388
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.9425567388534546,
      "learning_rate": 0.000577803458657679,
      "loss": 2.4764,
      "step": 28389
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.5936091542243958,
      "learning_rate": 0.0005778019144641481,
      "loss": 2.4988,
      "step": 28390
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6883305311203003,
      "learning_rate": 0.0005778003702189685,
      "loss": 2.1484,
      "step": 28391
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6771200299263,
      "learning_rate": 0.0005777988259221407,
      "loss": 2.2599,
      "step": 28392
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.1043694019317627,
      "learning_rate": 0.0005777972815736648,
      "loss": 2.2114,
      "step": 28393
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6982874870300293,
      "learning_rate": 0.0005777957371735412,
      "loss": 2.2766,
      "step": 28394
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.7204428911209106,
      "learning_rate": 0.0005777941927217699,
      "loss": 2.0846,
      "step": 28395
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6118941903114319,
      "learning_rate": 0.0005777926482183516,
      "loss": 2.3143,
      "step": 28396
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6346865892410278,
      "learning_rate": 0.0005777911036632863,
      "loss": 2.3878,
      "step": 28397
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6744676828384399,
      "learning_rate": 0.0005777895590565746,
      "loss": 2.3996,
      "step": 28398
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.620323896408081,
      "learning_rate": 0.0005777880143982163,
      "loss": 2.5716,
      "step": 28399
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6556680798530579,
      "learning_rate": 0.0005777864696882121,
      "loss": 2.4842,
      "step": 28400
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6185534000396729,
      "learning_rate": 0.0005777849249265621,
      "loss": 2.2919,
      "step": 28401
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.5983186364173889,
      "learning_rate": 0.0005777833801132667,
      "loss": 2.3883,
      "step": 28402
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6159899234771729,
      "learning_rate": 0.0005777818352483262,
      "loss": 2.2497,
      "step": 28403
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6419121623039246,
      "learning_rate": 0.0005777802903317407,
      "loss": 2.4068,
      "step": 28404
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6330294609069824,
      "learning_rate": 0.0005777787453635106,
      "loss": 2.1635,
      "step": 28405
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.7632527947425842,
      "learning_rate": 0.0005777772003436362,
      "loss": 2.4243,
      "step": 28406
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6423560380935669,
      "learning_rate": 0.0005777756552721179,
      "loss": 2.1827,
      "step": 28407
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6443159580230713,
      "learning_rate": 0.0005777741101489557,
      "loss": 2.419,
      "step": 28408
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6997010707855225,
      "learning_rate": 0.0005777725649741501,
      "loss": 2.5819,
      "step": 28409
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.5849893093109131,
      "learning_rate": 0.0005777710197477013,
      "loss": 2.3173,
      "step": 28410
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.5864876508712769,
      "learning_rate": 0.0005777694744696098,
      "loss": 2.3193,
      "step": 28411
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6381588578224182,
      "learning_rate": 0.0005777679291398755,
      "loss": 2.2894,
      "step": 28412
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6457977294921875,
      "learning_rate": 0.000577766383758499,
      "loss": 2.338,
      "step": 28413
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.5898308753967285,
      "learning_rate": 0.0005777648383254805,
      "loss": 2.3698,
      "step": 28414
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6696941256523132,
      "learning_rate": 0.0005777632928408202,
      "loss": 2.1743,
      "step": 28415
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6857132911682129,
      "learning_rate": 0.0005777617473045185,
      "loss": 2.46,
      "step": 28416
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6257239580154419,
      "learning_rate": 0.0005777602017165756,
      "loss": 2.0332,
      "step": 28417
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6404608488082886,
      "learning_rate": 0.0005777586560769919,
      "loss": 2.5013,
      "step": 28418
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.7502403259277344,
      "learning_rate": 0.0005777571103857676,
      "loss": 2.3239,
      "step": 28419
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.5869673490524292,
      "learning_rate": 0.000577755564642903,
      "loss": 2.2671,
      "step": 28420
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6978550553321838,
      "learning_rate": 0.0005777540188483984,
      "loss": 2.3857,
      "step": 28421
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6785383224487305,
      "learning_rate": 0.0005777524730022541,
      "loss": 2.3313,
      "step": 28422
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.7069365978240967,
      "learning_rate": 0.0005777509271044704,
      "loss": 2.4499,
      "step": 28423
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6849677562713623,
      "learning_rate": 0.0005777493811550474,
      "loss": 2.4074,
      "step": 28424
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6303771734237671,
      "learning_rate": 0.0005777478351539858,
      "loss": 2.3141,
      "step": 28425
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6690380573272705,
      "learning_rate": 0.0005777462891012855,
      "loss": 2.1075,
      "step": 28426
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6942635178565979,
      "learning_rate": 0.0005777447429969469,
      "loss": 2.2941,
      "step": 28427
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.5930776000022888,
      "learning_rate": 0.0005777431968409703,
      "loss": 2.4834,
      "step": 28428
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6811596155166626,
      "learning_rate": 0.0005777416506333561,
      "loss": 2.1577,
      "step": 28429
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.617310643196106,
      "learning_rate": 0.0005777401043741044,
      "loss": 2.3064,
      "step": 28430
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6323198080062866,
      "learning_rate": 0.0005777385580632156,
      "loss": 2.4579,
      "step": 28431
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.5991901159286499,
      "learning_rate": 0.0005777370117006899,
      "loss": 2.3044,
      "step": 28432
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.663901686668396,
      "learning_rate": 0.0005777354652865277,
      "loss": 2.2482,
      "step": 28433
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6905991435050964,
      "learning_rate": 0.0005777339188207293,
      "loss": 2.3397,
      "step": 28434
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6855937838554382,
      "learning_rate": 0.0005777323723032948,
      "loss": 2.2889,
      "step": 28435
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6240450739860535,
      "learning_rate": 0.0005777308257342246,
      "loss": 2.2698,
      "step": 28436
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.596125602722168,
      "learning_rate": 0.0005777292791135192,
      "loss": 2.3633,
      "step": 28437
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6019448041915894,
      "learning_rate": 0.0005777277324411784,
      "loss": 2.327,
      "step": 28438
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6179091334342957,
      "learning_rate": 0.000577726185717203,
      "loss": 2.3262,
      "step": 28439
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.584441602230072,
      "learning_rate": 0.000577724638941593,
      "loss": 2.0948,
      "step": 28440
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.5815931558609009,
      "learning_rate": 0.0005777230921143487,
      "loss": 2.1814,
      "step": 28441
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6677603125572205,
      "learning_rate": 0.0005777215452354705,
      "loss": 2.3512,
      "step": 28442
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.5848307013511658,
      "learning_rate": 0.0005777199983049585,
      "loss": 2.2728,
      "step": 28443
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6572555303573608,
      "learning_rate": 0.0005777184513228133,
      "loss": 2.2726,
      "step": 28444
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6840546727180481,
      "learning_rate": 0.0005777169042890349,
      "loss": 2.6647,
      "step": 28445
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.629409670829773,
      "learning_rate": 0.0005777153572036236,
      "loss": 2.3268,
      "step": 28446
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6152670979499817,
      "learning_rate": 0.0005777138100665799,
      "loss": 2.1813,
      "step": 28447
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6571470499038696,
      "learning_rate": 0.0005777122628779039,
      "loss": 2.2441,
      "step": 28448
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.7594197988510132,
      "learning_rate": 0.0005777107156375959,
      "loss": 2.2736,
      "step": 28449
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.5759336352348328,
      "learning_rate": 0.0005777091683456562,
      "loss": 2.1991,
      "step": 28450
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6680780053138733,
      "learning_rate": 0.0005777076210020853,
      "loss": 2.24,
      "step": 28451
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6575466990470886,
      "learning_rate": 0.0005777060736068833,
      "loss": 2.3128,
      "step": 28452
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.5830191969871521,
      "learning_rate": 0.0005777045261600504,
      "loss": 2.2309,
      "step": 28453
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6137071251869202,
      "learning_rate": 0.000577702978661587,
      "loss": 2.4933,
      "step": 28454
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6785317063331604,
      "learning_rate": 0.0005777014311114933,
      "loss": 2.3828,
      "step": 28455
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.7230276465415955,
      "learning_rate": 0.0005776998835097698,
      "loss": 2.3587,
      "step": 28456
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6977041959762573,
      "learning_rate": 0.0005776983358564166,
      "loss": 2.276,
      "step": 28457
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6311801075935364,
      "learning_rate": 0.000577696788151434,
      "loss": 2.3752,
      "step": 28458
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.7495848536491394,
      "learning_rate": 0.0005776952403948225,
      "loss": 2.0306,
      "step": 28459
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6320036053657532,
      "learning_rate": 0.0005776936925865819,
      "loss": 2.1268,
      "step": 28460
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6241058111190796,
      "learning_rate": 0.000577692144726713,
      "loss": 2.1754,
      "step": 28461
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.759119987487793,
      "learning_rate": 0.0005776905968152159,
      "loss": 2.1772,
      "step": 28462
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6397352814674377,
      "learning_rate": 0.0005776890488520908,
      "loss": 2.4414,
      "step": 28463
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6073212027549744,
      "learning_rate": 0.0005776875008373382,
      "loss": 2.2746,
      "step": 28464
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6513562798500061,
      "learning_rate": 0.0005776859527709581,
      "loss": 2.3949,
      "step": 28465
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.7784797549247742,
      "learning_rate": 0.000577684404652951,
      "loss": 2.2661,
      "step": 28466
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.5764042139053345,
      "learning_rate": 0.000577682856483317,
      "loss": 2.3562,
      "step": 28467
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.7360349297523499,
      "learning_rate": 0.0005776813082620568,
      "loss": 2.3034,
      "step": 28468
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.7200576663017273,
      "learning_rate": 0.0005776797599891702,
      "loss": 2.2083,
      "step": 28469
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6705102324485779,
      "learning_rate": 0.0005776782116646577,
      "loss": 2.3216,
      "step": 28470
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.601272702217102,
      "learning_rate": 0.0005776766632885197,
      "loss": 2.1869,
      "step": 28471
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.7514594793319702,
      "learning_rate": 0.0005776751148607563,
      "loss": 2.2549,
      "step": 28472
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6196475625038147,
      "learning_rate": 0.0005776735663813678,
      "loss": 2.4275,
      "step": 28473
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6915422081947327,
      "learning_rate": 0.0005776720178503544,
      "loss": 2.534,
      "step": 28474
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.7672297954559326,
      "learning_rate": 0.0005776704692677168,
      "loss": 2.4853,
      "step": 28475
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6612936854362488,
      "learning_rate": 0.0005776689206334549,
      "loss": 2.5399,
      "step": 28476
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.609878420829773,
      "learning_rate": 0.0005776673719475692,
      "loss": 2.4394,
      "step": 28477
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.765455961227417,
      "learning_rate": 0.0005776658232100598,
      "loss": 2.2467,
      "step": 28478
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6671837568283081,
      "learning_rate": 0.000577664274420927,
      "loss": 2.4016,
      "step": 28479
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.5870144367218018,
      "learning_rate": 0.0005776627255801713,
      "loss": 2.2093,
      "step": 28480
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.5761433243751526,
      "learning_rate": 0.0005776611766877929,
      "loss": 2.3932,
      "step": 28481
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.7655895352363586,
      "learning_rate": 0.000577659627743792,
      "loss": 2.3597,
      "step": 28482
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6078428626060486,
      "learning_rate": 0.0005776580787481689,
      "loss": 2.5018,
      "step": 28483
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.5869490504264832,
      "learning_rate": 0.0005776565297009239,
      "loss": 2.2288,
      "step": 28484
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6868324875831604,
      "learning_rate": 0.0005776549806020574,
      "loss": 2.2388,
      "step": 28485
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6985095143318176,
      "learning_rate": 0.0005776534314515695,
      "loss": 2.291,
      "step": 28486
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6447032690048218,
      "learning_rate": 0.0005776518822494607,
      "loss": 2.3314,
      "step": 28487
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.0598877668380737,
      "learning_rate": 0.0005776503329957311,
      "loss": 2.3539,
      "step": 28488
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.5831316709518433,
      "learning_rate": 0.000577648783690381,
      "loss": 2.1206,
      "step": 28489
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6479723453521729,
      "learning_rate": 0.0005776472343334108,
      "loss": 2.3802,
      "step": 28490
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6937894821166992,
      "learning_rate": 0.0005776456849248207,
      "loss": 2.2768,
      "step": 28491
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6940656900405884,
      "learning_rate": 0.0005776441354646112,
      "loss": 2.3429,
      "step": 28492
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.5948827862739563,
      "learning_rate": 0.0005776425859527822,
      "loss": 2.1017,
      "step": 28493
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6878772974014282,
      "learning_rate": 0.0005776410363893344,
      "loss": 2.0776,
      "step": 28494
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6541242003440857,
      "learning_rate": 0.0005776394867742677,
      "loss": 2.5952,
      "step": 28495
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6657929420471191,
      "learning_rate": 0.0005776379371075827,
      "loss": 2.1445,
      "step": 28496
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.5720400810241699,
      "learning_rate": 0.0005776363873892795,
      "loss": 2.2345,
      "step": 28497
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.7067803144454956,
      "learning_rate": 0.0005776348376193585,
      "loss": 2.2548,
      "step": 28498
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6293997168540955,
      "learning_rate": 0.0005776332877978199,
      "loss": 2.2126,
      "step": 28499
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.5965307950973511,
      "learning_rate": 0.0005776317379246641,
      "loss": 2.4876,
      "step": 28500
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6341919898986816,
      "learning_rate": 0.0005776301879998912,
      "loss": 2.0791,
      "step": 28501
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6394579410552979,
      "learning_rate": 0.0005776286380235017,
      "loss": 2.1269,
      "step": 28502
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6911501288414001,
      "learning_rate": 0.0005776270879954958,
      "loss": 2.3982,
      "step": 28503
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.714688241481781,
      "learning_rate": 0.0005776255379158737,
      "loss": 2.4372,
      "step": 28504
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6030741333961487,
      "learning_rate": 0.000577623987784636,
      "loss": 2.3155,
      "step": 28505
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6834115982055664,
      "learning_rate": 0.0005776224376017825,
      "loss": 2.4547,
      "step": 28506
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.60916668176651,
      "learning_rate": 0.0005776208873673139,
      "loss": 2.2985,
      "step": 28507
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6646416783332825,
      "learning_rate": 0.0005776193370812303,
      "loss": 2.6846,
      "step": 28508
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6500452756881714,
      "learning_rate": 0.000577617786743532,
      "loss": 2.1263,
      "step": 28509
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6850194334983826,
      "learning_rate": 0.0005776162363542193,
      "loss": 2.2031,
      "step": 28510
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.5776261687278748,
      "learning_rate": 0.0005776146859132926,
      "loss": 2.4301,
      "step": 28511
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.7752026915550232,
      "learning_rate": 0.0005776131354207519,
      "loss": 2.3803,
      "step": 28512
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.7002710103988647,
      "learning_rate": 0.0005776115848765978,
      "loss": 2.4471,
      "step": 28513
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6854209899902344,
      "learning_rate": 0.0005776100342808303,
      "loss": 2.3411,
      "step": 28514
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6444520354270935,
      "learning_rate": 0.0005776084836334502,
      "loss": 2.1937,
      "step": 28515
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.9134219288825989,
      "learning_rate": 0.0005776069329344572,
      "loss": 2.5422,
      "step": 28516
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6028645038604736,
      "learning_rate": 0.0005776053821838519,
      "loss": 2.1543,
      "step": 28517
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6320432424545288,
      "learning_rate": 0.0005776038313816345,
      "loss": 2.262,
      "step": 28518
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.8993105888366699,
      "learning_rate": 0.0005776022805278053,
      "loss": 2.1196,
      "step": 28519
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6762747168540955,
      "learning_rate": 0.0005776007296223646,
      "loss": 2.0828,
      "step": 28520
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.707611083984375,
      "learning_rate": 0.0005775991786653127,
      "loss": 2.3114,
      "step": 28521
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.628484845161438,
      "learning_rate": 0.0005775976276566499,
      "loss": 2.5463,
      "step": 28522
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6691364049911499,
      "learning_rate": 0.0005775960765963764,
      "loss": 2.2761,
      "step": 28523
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.638879120349884,
      "learning_rate": 0.0005775945254844926,
      "loss": 2.2576,
      "step": 28524
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6830522418022156,
      "learning_rate": 0.0005775929743209987,
      "loss": 2.207,
      "step": 28525
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.7288385033607483,
      "learning_rate": 0.000577591423105895,
      "loss": 2.279,
      "step": 28526
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.7547867894172668,
      "learning_rate": 0.0005775898718391819,
      "loss": 2.3689,
      "step": 28527
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.7534800171852112,
      "learning_rate": 0.0005775883205208595,
      "loss": 2.2754,
      "step": 28528
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6472681164741516,
      "learning_rate": 0.0005775867691509282,
      "loss": 2.4934,
      "step": 28529
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6998923420906067,
      "learning_rate": 0.0005775852177293884,
      "loss": 2.4007,
      "step": 28530
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6647664308547974,
      "learning_rate": 0.0005775836662562402,
      "loss": 2.6229,
      "step": 28531
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6025149822235107,
      "learning_rate": 0.000577582114731484,
      "loss": 2.5003,
      "step": 28532
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6368935108184814,
      "learning_rate": 0.0005775805631551199,
      "loss": 2.6336,
      "step": 28533
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6163456439971924,
      "learning_rate": 0.0005775790115271484,
      "loss": 2.3639,
      "step": 28534
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6780623197555542,
      "learning_rate": 0.0005775774598475698,
      "loss": 2.0073,
      "step": 28535
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.7428637742996216,
      "learning_rate": 0.0005775759081163842,
      "loss": 2.3219,
      "step": 28536
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6575360298156738,
      "learning_rate": 0.000577574356333592,
      "loss": 2.3576,
      "step": 28537
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.5548654794692993,
      "learning_rate": 0.0005775728044991937,
      "loss": 2.1729,
      "step": 28538
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.7456112504005432,
      "learning_rate": 0.0005775712526131891,
      "loss": 2.3048,
      "step": 28539
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6074552536010742,
      "learning_rate": 0.000577569700675579,
      "loss": 2.3621,
      "step": 28540
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6488989591598511,
      "learning_rate": 0.0005775681486863633,
      "loss": 2.5798,
      "step": 28541
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.7645621299743652,
      "learning_rate": 0.0005775665966455425,
      "loss": 2.4619,
      "step": 28542
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6500369906425476,
      "learning_rate": 0.0005775650445531168,
      "loss": 2.1351,
      "step": 28543
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.755843997001648,
      "learning_rate": 0.0005775634924090865,
      "loss": 2.3679,
      "step": 28544
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.757785975933075,
      "learning_rate": 0.0005775619402134519,
      "loss": 2.3182,
      "step": 28545
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6306139230728149,
      "learning_rate": 0.0005775603879662135,
      "loss": 2.421,
      "step": 28546
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6918650269508362,
      "learning_rate": 0.0005775588356673711,
      "loss": 2.404,
      "step": 28547
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.7451707720756531,
      "learning_rate": 0.0005775572833169254,
      "loss": 2.1633,
      "step": 28548
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.7308000326156616,
      "learning_rate": 0.0005775557309148766,
      "loss": 2.3851,
      "step": 28549
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6502431035041809,
      "learning_rate": 0.0005775541784612248,
      "loss": 2.3319,
      "step": 28550
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.7015661597251892,
      "learning_rate": 0.0005775526259559706,
      "loss": 2.4011,
      "step": 28551
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.7775217890739441,
      "learning_rate": 0.0005775510733991141,
      "loss": 2.1234,
      "step": 28552
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6464515328407288,
      "learning_rate": 0.0005775495207906556,
      "loss": 2.1877,
      "step": 28553
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6523109078407288,
      "learning_rate": 0.0005775479681305953,
      "loss": 2.1298,
      "step": 28554
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.7897355556488037,
      "learning_rate": 0.0005775464154189337,
      "loss": 2.1811,
      "step": 28555
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.7512413859367371,
      "learning_rate": 0.000577544862655671,
      "loss": 2.3344,
      "step": 28556
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6191144585609436,
      "learning_rate": 0.0005775433098408074,
      "loss": 2.2573,
      "step": 28557
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.7031162977218628,
      "learning_rate": 0.0005775417569743433,
      "loss": 2.4549,
      "step": 28558
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.7419270277023315,
      "learning_rate": 0.000577540204056279,
      "loss": 2.1991,
      "step": 28559
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6242654323577881,
      "learning_rate": 0.0005775386510866146,
      "loss": 2.6422,
      "step": 28560
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.5954598188400269,
      "learning_rate": 0.0005775370980653507,
      "loss": 2.1238,
      "step": 28561
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6876155138015747,
      "learning_rate": 0.0005775355449924873,
      "loss": 2.1282,
      "step": 28562
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.7394809126853943,
      "learning_rate": 0.0005775339918680249,
      "loss": 2.4111,
      "step": 28563
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.5985767841339111,
      "learning_rate": 0.0005775324386919636,
      "loss": 2.2284,
      "step": 28564
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6176304817199707,
      "learning_rate": 0.0005775308854643038,
      "loss": 2.2604,
      "step": 28565
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6219142079353333,
      "learning_rate": 0.0005775293321850457,
      "loss": 2.3806,
      "step": 28566
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6480650901794434,
      "learning_rate": 0.0005775277788541898,
      "loss": 2.3411,
      "step": 28567
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.8268342614173889,
      "learning_rate": 0.0005775262254717362,
      "loss": 2.3999,
      "step": 28568
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6541481018066406,
      "learning_rate": 0.0005775246720376852,
      "loss": 2.0507,
      "step": 28569
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6329594850540161,
      "learning_rate": 0.0005775231185520371,
      "loss": 2.4613,
      "step": 28570
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6222198009490967,
      "learning_rate": 0.0005775215650147922,
      "loss": 2.2489,
      "step": 28571
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.7818828225135803,
      "learning_rate": 0.0005775200114259509,
      "loss": 2.3392,
      "step": 28572
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6312073469161987,
      "learning_rate": 0.0005775184577855133,
      "loss": 2.3507,
      "step": 28573
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6776655912399292,
      "learning_rate": 0.0005775169040934798,
      "loss": 2.2311,
      "step": 28574
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6739767789840698,
      "learning_rate": 0.0005775153503498507,
      "loss": 2.135,
      "step": 28575
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6385213136672974,
      "learning_rate": 0.0005775137965546263,
      "loss": 2.3069,
      "step": 28576
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6557830572128296,
      "learning_rate": 0.0005775122427078067,
      "loss": 2.4387,
      "step": 28577
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.720812201499939,
      "learning_rate": 0.0005775106888093926,
      "loss": 2.2694,
      "step": 28578
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6636291146278381,
      "learning_rate": 0.0005775091348593838,
      "loss": 2.2331,
      "step": 28579
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6957070827484131,
      "learning_rate": 0.0005775075808577809,
      "loss": 2.2825,
      "step": 28580
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.7485185861587524,
      "learning_rate": 0.000577506026804584,
      "loss": 2.464,
      "step": 28581
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.7012758851051331,
      "learning_rate": 0.0005775044726997936,
      "loss": 2.4349,
      "step": 28582
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.923842191696167,
      "learning_rate": 0.0005775029185434099,
      "loss": 2.1419,
      "step": 28583
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6141646504402161,
      "learning_rate": 0.000577501364335433,
      "loss": 2.3971,
      "step": 28584
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6741915345191956,
      "learning_rate": 0.0005774998100758636,
      "loss": 2.3719,
      "step": 28585
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6205662488937378,
      "learning_rate": 0.0005774982557647016,
      "loss": 2.3186,
      "step": 28586
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.7115229368209839,
      "learning_rate": 0.0005774967014019474,
      "loss": 2.3058,
      "step": 28587
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6034542918205261,
      "learning_rate": 0.0005774951469876014,
      "loss": 2.1217,
      "step": 28588
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.7718707323074341,
      "learning_rate": 0.0005774935925216638,
      "loss": 2.4331,
      "step": 28589
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.7379949688911438,
      "learning_rate": 0.0005774920380041349,
      "loss": 2.0707,
      "step": 28590
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.666483998298645,
      "learning_rate": 0.000577490483435015,
      "loss": 2.4154,
      "step": 28591
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6387113332748413,
      "learning_rate": 0.0005774889288143044,
      "loss": 2.3578,
      "step": 28592
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.8438649773597717,
      "learning_rate": 0.0005774873741420033,
      "loss": 2.3802,
      "step": 28593
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.7106316685676575,
      "learning_rate": 0.0005774858194181121,
      "loss": 2.4247,
      "step": 28594
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.617985725402832,
      "learning_rate": 0.000577484264642631,
      "loss": 2.1942,
      "step": 28595
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.7301636934280396,
      "learning_rate": 0.0005774827098155604,
      "loss": 2.0437,
      "step": 28596
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.7343389391899109,
      "learning_rate": 0.0005774811549369006,
      "loss": 2.5201,
      "step": 28597
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.7318040728569031,
      "learning_rate": 0.0005774796000066517,
      "loss": 2.4107,
      "step": 28598
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.7079798579216003,
      "learning_rate": 0.0005774780450248141,
      "loss": 2.136,
      "step": 28599
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.7163739204406738,
      "learning_rate": 0.0005774764899913881,
      "loss": 2.4883,
      "step": 28600
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6439816355705261,
      "learning_rate": 0.000577474934906374,
      "loss": 2.5254,
      "step": 28601
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.707426130771637,
      "learning_rate": 0.0005774733797697722,
      "loss": 2.2429,
      "step": 28602
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6948139667510986,
      "learning_rate": 0.0005774718245815827,
      "loss": 2.2538,
      "step": 28603
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.5703102350234985,
      "learning_rate": 0.0005774702693418059,
      "loss": 2.4001,
      "step": 28604
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6213733553886414,
      "learning_rate": 0.0005774687140504423,
      "loss": 2.4574,
      "step": 28605
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.7106990814208984,
      "learning_rate": 0.0005774671587074919,
      "loss": 2.3873,
      "step": 28606
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6224285364151001,
      "learning_rate": 0.0005774656033129552,
      "loss": 2.2794,
      "step": 28607
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.5713821649551392,
      "learning_rate": 0.0005774640478668324,
      "loss": 2.0372,
      "step": 28608
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6331007480621338,
      "learning_rate": 0.0005774624923691238,
      "loss": 2.3518,
      "step": 28609
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.7307685017585754,
      "learning_rate": 0.0005774609368198296,
      "loss": 2.4137,
      "step": 28610
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.5821893215179443,
      "learning_rate": 0.0005774593812189503,
      "loss": 2.4373,
      "step": 28611
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6382943987846375,
      "learning_rate": 0.000577457825566486,
      "loss": 2.2756,
      "step": 28612
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6694843173027039,
      "learning_rate": 0.000577456269862437,
      "loss": 2.0841,
      "step": 28613
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.7016935348510742,
      "learning_rate": 0.0005774547141068037,
      "loss": 2.4098,
      "step": 28614
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6164502501487732,
      "learning_rate": 0.0005774531582995863,
      "loss": 2.2407,
      "step": 28615
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6711859703063965,
      "learning_rate": 0.0005774516024407853,
      "loss": 2.3334,
      "step": 28616
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.7028961777687073,
      "learning_rate": 0.0005774500465304006,
      "loss": 2.3421,
      "step": 28617
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6298618316650391,
      "learning_rate": 0.0005774484905684328,
      "loss": 2.4123,
      "step": 28618
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6791338324546814,
      "learning_rate": 0.000577446934554882,
      "loss": 2.2028,
      "step": 28619
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6501148343086243,
      "learning_rate": 0.0005774453784897485,
      "loss": 2.3625,
      "step": 28620
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6591660976409912,
      "learning_rate": 0.0005774438223730329,
      "loss": 2.248,
      "step": 28621
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6358774900436401,
      "learning_rate": 0.0005774422662047351,
      "loss": 2.5882,
      "step": 28622
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.5994897484779358,
      "learning_rate": 0.0005774407099848557,
      "loss": 2.3447,
      "step": 28623
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6586755514144897,
      "learning_rate": 0.0005774391537133947,
      "loss": 2.4057,
      "step": 28624
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6427969932556152,
      "learning_rate": 0.0005774375973903526,
      "loss": 2.2765,
      "step": 28625
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.7282207608222961,
      "learning_rate": 0.0005774360410157295,
      "loss": 2.1783,
      "step": 28626
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.7154486179351807,
      "learning_rate": 0.0005774344845895259,
      "loss": 2.3765,
      "step": 28627
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.7069638967514038,
      "learning_rate": 0.0005774329281117419,
      "loss": 2.129,
      "step": 28628
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6018998026847839,
      "learning_rate": 0.0005774313715823779,
      "loss": 2.3906,
      "step": 28629
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.625809907913208,
      "learning_rate": 0.0005774298150014343,
      "loss": 2.3247,
      "step": 28630
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6324695944786072,
      "learning_rate": 0.0005774282583689112,
      "loss": 2.4837,
      "step": 28631
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6949299573898315,
      "learning_rate": 0.0005774267016848089,
      "loss": 2.4014,
      "step": 28632
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6530888676643372,
      "learning_rate": 0.0005774251449491277,
      "loss": 2.1262,
      "step": 28633
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.7063595652580261,
      "learning_rate": 0.000577423588161868,
      "loss": 2.2805,
      "step": 28634
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6667545437812805,
      "learning_rate": 0.0005774220313230301,
      "loss": 2.2336,
      "step": 28635
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6437304019927979,
      "learning_rate": 0.000577420474432614,
      "loss": 2.2377,
      "step": 28636
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6615824699401855,
      "learning_rate": 0.0005774189174906203,
      "loss": 2.2722,
      "step": 28637
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.7049592137336731,
      "learning_rate": 0.0005774173604970493,
      "loss": 2.3487,
      "step": 28638
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.7465110421180725,
      "learning_rate": 0.000577415803451901,
      "loss": 2.4873,
      "step": 28639
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6697222590446472,
      "learning_rate": 0.000577414246355176,
      "loss": 2.3817,
      "step": 28640
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6722185015678406,
      "learning_rate": 0.0005774126892068743,
      "loss": 2.0773,
      "step": 28641
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.590427577495575,
      "learning_rate": 0.0005774111320069965,
      "loss": 2.3569,
      "step": 28642
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.676878035068512,
      "learning_rate": 0.0005774095747555427,
      "loss": 2.2102,
      "step": 28643
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.7923230528831482,
      "learning_rate": 0.0005774080174525132,
      "loss": 2.3638,
      "step": 28644
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6919757723808289,
      "learning_rate": 0.0005774064600979083,
      "loss": 2.3397,
      "step": 28645
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.7279717326164246,
      "learning_rate": 0.0005774049026917283,
      "loss": 2.3853,
      "step": 28646
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.7658500075340271,
      "learning_rate": 0.0005774033452339736,
      "loss": 2.1527,
      "step": 28647
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6558818221092224,
      "learning_rate": 0.0005774017877246443,
      "loss": 2.2954,
      "step": 28648
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6305725574493408,
      "learning_rate": 0.0005774002301637407,
      "loss": 2.2619,
      "step": 28649
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6159110069274902,
      "learning_rate": 0.0005773986725512633,
      "loss": 2.3931,
      "step": 28650
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.7172472476959229,
      "learning_rate": 0.000577397114887212,
      "loss": 2.1954,
      "step": 28651
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6265065670013428,
      "learning_rate": 0.0005773955571715876,
      "loss": 1.9955,
      "step": 28652
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6520388126373291,
      "learning_rate": 0.00057739399940439,
      "loss": 2.1453,
      "step": 28653
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6353492736816406,
      "learning_rate": 0.0005773924415856197,
      "loss": 2.1983,
      "step": 28654
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6179034113883972,
      "learning_rate": 0.0005773908837152768,
      "loss": 2.28,
      "step": 28655
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6337947249412537,
      "learning_rate": 0.0005773893257933619,
      "loss": 2.333,
      "step": 28656
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6884483098983765,
      "learning_rate": 0.0005773877678198749,
      "loss": 2.5729,
      "step": 28657
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6785327196121216,
      "learning_rate": 0.0005773862097948163,
      "loss": 2.0982,
      "step": 28658
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6965813636779785,
      "learning_rate": 0.0005773846517181864,
      "loss": 2.526,
      "step": 28659
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.5983858704566956,
      "learning_rate": 0.0005773830935899855,
      "loss": 2.3811,
      "step": 28660
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6744416952133179,
      "learning_rate": 0.0005773815354102139,
      "loss": 2.1757,
      "step": 28661
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6033819913864136,
      "learning_rate": 0.0005773799771788716,
      "loss": 2.2018,
      "step": 28662
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6390690803527832,
      "learning_rate": 0.0005773784188959593,
      "loss": 2.1356,
      "step": 28663
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6951947808265686,
      "learning_rate": 0.0005773768605614771,
      "loss": 2.286,
      "step": 28664
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.64775550365448,
      "learning_rate": 0.0005773753021754253,
      "loss": 2.4816,
      "step": 28665
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.66609126329422,
      "learning_rate": 0.0005773737437378041,
      "loss": 2.4714,
      "step": 28666
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6788056492805481,
      "learning_rate": 0.000577372185248614,
      "loss": 2.1775,
      "step": 28667
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.5934638381004333,
      "learning_rate": 0.0005773706267078552,
      "loss": 2.1242,
      "step": 28668
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.7036795020103455,
      "learning_rate": 0.0005773690681155279,
      "loss": 2.2932,
      "step": 28669
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6254200339317322,
      "learning_rate": 0.0005773675094716325,
      "loss": 2.4446,
      "step": 28670
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6234767436981201,
      "learning_rate": 0.0005773659507761692,
      "loss": 2.1689,
      "step": 28671
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.682000994682312,
      "learning_rate": 0.0005773643920291383,
      "loss": 2.153,
      "step": 28672
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.562171220779419,
      "learning_rate": 0.0005773628332305402,
      "loss": 2.3618,
      "step": 28673
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6176614761352539,
      "learning_rate": 0.0005773612743803751,
      "loss": 2.2699,
      "step": 28674
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6226609349250793,
      "learning_rate": 0.0005773597154786433,
      "loss": 2.3634,
      "step": 28675
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6618995666503906,
      "learning_rate": 0.000577358156525345,
      "loss": 2.0388,
      "step": 28676
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6378462314605713,
      "learning_rate": 0.0005773565975204807,
      "loss": 2.3233,
      "step": 28677
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6435603499412537,
      "learning_rate": 0.0005773550384640505,
      "loss": 2.2146,
      "step": 28678
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6840113401412964,
      "learning_rate": 0.0005773534793560548,
      "loss": 2.1378,
      "step": 28679
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6409479379653931,
      "learning_rate": 0.0005773519201964939,
      "loss": 2.1944,
      "step": 28680
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6317422389984131,
      "learning_rate": 0.000577350360985368,
      "loss": 2.3411,
      "step": 28681
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6371908783912659,
      "learning_rate": 0.0005773488017226774,
      "loss": 2.2422,
      "step": 28682
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6711732149124146,
      "learning_rate": 0.0005773472424084225,
      "loss": 2.5239,
      "step": 28683
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.7046377658843994,
      "learning_rate": 0.0005773456830426033,
      "loss": 2.3937,
      "step": 28684
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6243964433670044,
      "learning_rate": 0.0005773441236252206,
      "loss": 2.2703,
      "step": 28685
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6528034806251526,
      "learning_rate": 0.0005773425641562742,
      "loss": 2.4298,
      "step": 28686
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6128789782524109,
      "learning_rate": 0.0005773410046357646,
      "loss": 2.2121,
      "step": 28687
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6951131820678711,
      "learning_rate": 0.0005773394450636922,
      "loss": 2.7118,
      "step": 28688
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.708100438117981,
      "learning_rate": 0.000577337885440057,
      "loss": 2.5178,
      "step": 28689
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6051755547523499,
      "learning_rate": 0.0005773363257648595,
      "loss": 2.1852,
      "step": 28690
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6735630631446838,
      "learning_rate": 0.0005773347660381,
      "loss": 2.0839,
      "step": 28691
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6156628727912903,
      "learning_rate": 0.0005773332062597788,
      "loss": 2.326,
      "step": 28692
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.7337954640388489,
      "learning_rate": 0.000577331646429896,
      "loss": 2.3859,
      "step": 28693
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6497856974601746,
      "learning_rate": 0.000577330086548452,
      "loss": 2.2402,
      "step": 28694
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6366687417030334,
      "learning_rate": 0.0005773285266154471,
      "loss": 2.3032,
      "step": 28695
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6499537229537964,
      "learning_rate": 0.0005773269666308816,
      "loss": 2.3502,
      "step": 28696
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.685350775718689,
      "learning_rate": 0.0005773254065947559,
      "loss": 2.1786,
      "step": 28697
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.5842157602310181,
      "learning_rate": 0.00057732384650707,
      "loss": 2.3333,
      "step": 28698
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6363821029663086,
      "learning_rate": 0.0005773222863678244,
      "loss": 2.2358,
      "step": 28699
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6719704866409302,
      "learning_rate": 0.0005773207261770194,
      "loss": 2.0731,
      "step": 28700
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6407309770584106,
      "learning_rate": 0.0005773191659346553,
      "loss": 2.364,
      "step": 28701
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6775667071342468,
      "learning_rate": 0.0005773176056407323,
      "loss": 2.2638,
      "step": 28702
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6842876672744751,
      "learning_rate": 0.0005773160452952507,
      "loss": 2.2427,
      "step": 28703
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6117223501205444,
      "learning_rate": 0.0005773144848982108,
      "loss": 2.3079,
      "step": 28704
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6584370136260986,
      "learning_rate": 0.0005773129244496129,
      "loss": 2.2919,
      "step": 28705
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6248869895935059,
      "learning_rate": 0.0005773113639494572,
      "loss": 2.2527,
      "step": 28706
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6553084254264832,
      "learning_rate": 0.0005773098033977442,
      "loss": 2.2174,
      "step": 28707
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6305575370788574,
      "learning_rate": 0.0005773082427944741,
      "loss": 2.3365,
      "step": 28708
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6258593201637268,
      "learning_rate": 0.0005773066821396472,
      "loss": 2.1188,
      "step": 28709
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6213112473487854,
      "learning_rate": 0.0005773051214332636,
      "loss": 2.1733,
      "step": 28710
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6346842646598816,
      "learning_rate": 0.0005773035606753238,
      "loss": 2.2526,
      "step": 28711
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6390604376792908,
      "learning_rate": 0.000577301999865828,
      "loss": 2.4962,
      "step": 28712
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6697396636009216,
      "learning_rate": 0.0005773004390047765,
      "loss": 2.4687,
      "step": 28713
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6640049815177917,
      "learning_rate": 0.0005772988780921698,
      "loss": 2.4351,
      "step": 28714
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6088953614234924,
      "learning_rate": 0.0005772973171280079,
      "loss": 2.2878,
      "step": 28715
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6266587376594543,
      "learning_rate": 0.0005772957561122912,
      "loss": 2.0734,
      "step": 28716
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6921460032463074,
      "learning_rate": 0.0005772941950450199,
      "loss": 2.5487,
      "step": 28717
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.5798591375350952,
      "learning_rate": 0.0005772926339261945,
      "loss": 2.1098,
      "step": 28718
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6882199048995972,
      "learning_rate": 0.000577291072755815,
      "loss": 2.3546,
      "step": 28719
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6821247339248657,
      "learning_rate": 0.000577289511533882,
      "loss": 2.318,
      "step": 28720
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6692227125167847,
      "learning_rate": 0.0005772879502603957,
      "loss": 2.2932,
      "step": 28721
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6950647830963135,
      "learning_rate": 0.0005772863889353561,
      "loss": 2.2379,
      "step": 28722
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.8109463453292847,
      "learning_rate": 0.0005772848275587639,
      "loss": 2.1657,
      "step": 28723
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6832598447799683,
      "learning_rate": 0.0005772832661306192,
      "loss": 2.1481,
      "step": 28724
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6536679267883301,
      "learning_rate": 0.0005772817046509224,
      "loss": 2.1915,
      "step": 28725
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.8656509518623352,
      "learning_rate": 0.0005772801431196734,
      "loss": 2.3641,
      "step": 28726
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.7124715447425842,
      "learning_rate": 0.0005772785815368731,
      "loss": 2.2252,
      "step": 28727
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.7629411220550537,
      "learning_rate": 0.0005772770199025213,
      "loss": 2.3986,
      "step": 28728
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.7322023510932922,
      "learning_rate": 0.0005772754582166185,
      "loss": 2.508,
      "step": 28729
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.738541841506958,
      "learning_rate": 0.000577273896479165,
      "loss": 2.3455,
      "step": 28730
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6346291303634644,
      "learning_rate": 0.0005772723346901611,
      "loss": 2.3239,
      "step": 28731
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6777920722961426,
      "learning_rate": 0.000577270772849607,
      "loss": 2.3657,
      "step": 28732
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.8781419992446899,
      "learning_rate": 0.0005772692109575029,
      "loss": 2.4619,
      "step": 28733
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6540825963020325,
      "learning_rate": 0.0005772676490138493,
      "loss": 2.4296,
      "step": 28734
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.7295377254486084,
      "learning_rate": 0.0005772660870186465,
      "loss": 2.4481,
      "step": 28735
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.65711909532547,
      "learning_rate": 0.0005772645249718946,
      "loss": 2.1894,
      "step": 28736
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6486925482749939,
      "learning_rate": 0.0005772629628735939,
      "loss": 2.521,
      "step": 28737
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.7185559272766113,
      "learning_rate": 0.000577261400723745,
      "loss": 2.2873,
      "step": 28738
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6363452672958374,
      "learning_rate": 0.0005772598385223479,
      "loss": 2.4202,
      "step": 28739
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.7085652947425842,
      "learning_rate": 0.000577258276269403,
      "loss": 2.328,
      "step": 28740
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6717226505279541,
      "learning_rate": 0.0005772567139649105,
      "loss": 2.3595,
      "step": 28741
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.5878704190254211,
      "learning_rate": 0.0005772551516088707,
      "loss": 2.3357,
      "step": 28742
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6550354361534119,
      "learning_rate": 0.0005772535892012839,
      "loss": 2.2255,
      "step": 28743
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6120133399963379,
      "learning_rate": 0.0005772520267421505,
      "loss": 2.193,
      "step": 28744
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.8772456645965576,
      "learning_rate": 0.0005772504642314707,
      "loss": 2.3299,
      "step": 28745
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6468178629875183,
      "learning_rate": 0.0005772489016692448,
      "loss": 2.2885,
      "step": 28746
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6758897304534912,
      "learning_rate": 0.0005772473390554732,
      "loss": 2.1478,
      "step": 28747
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6881729960441589,
      "learning_rate": 0.0005772457763901559,
      "loss": 2.2759,
      "step": 28748
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6801629662513733,
      "learning_rate": 0.0005772442136732934,
      "loss": 2.4583,
      "step": 28749
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6131265163421631,
      "learning_rate": 0.0005772426509048861,
      "loss": 2.3986,
      "step": 28750
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6725318431854248,
      "learning_rate": 0.000577241088084934,
      "loss": 2.3885,
      "step": 28751
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.7771304249763489,
      "learning_rate": 0.0005772395252134377,
      "loss": 2.1768,
      "step": 28752
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6641445755958557,
      "learning_rate": 0.0005772379622903973,
      "loss": 2.3318,
      "step": 28753
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6145978569984436,
      "learning_rate": 0.0005772363993158131,
      "loss": 2.2188,
      "step": 28754
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.8361321091651917,
      "learning_rate": 0.0005772348362896854,
      "loss": 2.3656,
      "step": 28755
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6589754223823547,
      "learning_rate": 0.0005772332732120147,
      "loss": 2.2547,
      "step": 28756
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6357221603393555,
      "learning_rate": 0.0005772317100828008,
      "loss": 2.4894,
      "step": 28757
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6621749997138977,
      "learning_rate": 0.0005772301469020445,
      "loss": 2.3921,
      "step": 28758
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.7262077331542969,
      "learning_rate": 0.0005772285836697458,
      "loss": 2.3873,
      "step": 28759
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.7129717469215393,
      "learning_rate": 0.000577227020385905,
      "loss": 2.5399,
      "step": 28760
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.5953466892242432,
      "learning_rate": 0.0005772254570505225,
      "loss": 2.3374,
      "step": 28761
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.700272798538208,
      "learning_rate": 0.0005772238936635987,
      "loss": 2.4994,
      "step": 28762
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6309398412704468,
      "learning_rate": 0.0005772223302251338,
      "loss": 2.1887,
      "step": 28763
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6325092315673828,
      "learning_rate": 0.0005772207667351278,
      "loss": 2.2106,
      "step": 28764
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.7690398097038269,
      "learning_rate": 0.0005772192031935813,
      "loss": 2.2115,
      "step": 28765
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.594453752040863,
      "learning_rate": 0.0005772176396004945,
      "loss": 2.285,
      "step": 28766
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.641402006149292,
      "learning_rate": 0.0005772160759558678,
      "loss": 2.2228,
      "step": 28767
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6453348398208618,
      "learning_rate": 0.0005772145122597014,
      "loss": 2.1735,
      "step": 28768
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.7346216440200806,
      "learning_rate": 0.0005772129485119955,
      "loss": 2.4038,
      "step": 28769
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6321568489074707,
      "learning_rate": 0.0005772113847127505,
      "loss": 2.2222,
      "step": 28770
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.5865139961242676,
      "learning_rate": 0.0005772098208619667,
      "loss": 2.0566,
      "step": 28771
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.5779989361763,
      "learning_rate": 0.0005772082569596443,
      "loss": 2.1499,
      "step": 28772
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6399452686309814,
      "learning_rate": 0.0005772066930057836,
      "loss": 2.3544,
      "step": 28773
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6431454420089722,
      "learning_rate": 0.0005772051290003851,
      "loss": 2.3343,
      "step": 28774
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.5998740196228027,
      "learning_rate": 0.0005772035649434489,
      "loss": 2.2192,
      "step": 28775
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.634297251701355,
      "learning_rate": 0.0005772020008349753,
      "loss": 2.3816,
      "step": 28776
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6590192317962646,
      "learning_rate": 0.0005772004366749646,
      "loss": 2.4222,
      "step": 28777
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6673269867897034,
      "learning_rate": 0.0005771988724634171,
      "loss": 2.3901,
      "step": 28778
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6651288270950317,
      "learning_rate": 0.0005771973082003331,
      "loss": 2.4123,
      "step": 28779
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6851856112480164,
      "learning_rate": 0.000577195743885713,
      "loss": 2.7566,
      "step": 28780
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.8039079308509827,
      "learning_rate": 0.0005771941795195568,
      "loss": 2.4197,
      "step": 28781
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6324911713600159,
      "learning_rate": 0.0005771926151018651,
      "loss": 2.5454,
      "step": 28782
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6620563864707947,
      "learning_rate": 0.000577191050632638,
      "loss": 2.3807,
      "step": 28783
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.7702207565307617,
      "learning_rate": 0.0005771894861118758,
      "loss": 2.2905,
      "step": 28784
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6697844862937927,
      "learning_rate": 0.0005771879215395789,
      "loss": 2.1024,
      "step": 28785
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6640157103538513,
      "learning_rate": 0.0005771863569157476,
      "loss": 2.4256,
      "step": 28786
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6566082835197449,
      "learning_rate": 0.000577184792240382,
      "loss": 2.4498,
      "step": 28787
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6032999753952026,
      "learning_rate": 0.0005771832275134827,
      "loss": 2.2643,
      "step": 28788
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6171989440917969,
      "learning_rate": 0.0005771816627350495,
      "loss": 2.2133,
      "step": 28789
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6619374752044678,
      "learning_rate": 0.0005771800979050833,
      "loss": 2.3715,
      "step": 28790
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.7152361869812012,
      "learning_rate": 0.000577178533023584,
      "loss": 2.2835,
      "step": 28791
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6792706847190857,
      "learning_rate": 0.0005771769680905519,
      "loss": 2.232,
      "step": 28792
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6016092300415039,
      "learning_rate": 0.0005771754031059874,
      "loss": 2.2798,
      "step": 28793
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6457588076591492,
      "learning_rate": 0.0005771738380698907,
      "loss": 2.2703,
      "step": 28794
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6342126131057739,
      "learning_rate": 0.0005771722729822623,
      "loss": 2.2987,
      "step": 28795
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.5905577540397644,
      "learning_rate": 0.0005771707078431023,
      "loss": 2.3846,
      "step": 28796
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6979331970214844,
      "learning_rate": 0.0005771691426524109,
      "loss": 2.5163,
      "step": 28797
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6567838788032532,
      "learning_rate": 0.0005771675774101887,
      "loss": 2.498,
      "step": 28798
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.7200313210487366,
      "learning_rate": 0.0005771660121164357,
      "loss": 2.5938,
      "step": 28799
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.625063419342041,
      "learning_rate": 0.0005771644467711523,
      "loss": 2.2767,
      "step": 28800
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.633140504360199,
      "learning_rate": 0.000577162881374339,
      "loss": 2.2759,
      "step": 28801
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6075100302696228,
      "learning_rate": 0.0005771613159259956,
      "loss": 2.2268,
      "step": 28802
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.602238118648529,
      "learning_rate": 0.0005771597504261229,
      "loss": 2.3443,
      "step": 28803
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6326549053192139,
      "learning_rate": 0.0005771581848747208,
      "loss": 2.2995,
      "step": 28804
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6440476179122925,
      "learning_rate": 0.0005771566192717899,
      "loss": 2.383,
      "step": 28805
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.622073233127594,
      "learning_rate": 0.0005771550536173303,
      "loss": 2.3073,
      "step": 28806
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6330123543739319,
      "learning_rate": 0.0005771534879113423,
      "loss": 2.2871,
      "step": 28807
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.5854056477546692,
      "learning_rate": 0.0005771519221538263,
      "loss": 2.3276,
      "step": 28808
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6353126764297485,
      "learning_rate": 0.0005771503563447825,
      "loss": 2.1617,
      "step": 28809
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.5996323227882385,
      "learning_rate": 0.0005771487904842112,
      "loss": 2.4921,
      "step": 28810
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6287961602210999,
      "learning_rate": 0.0005771472245721127,
      "loss": 2.2113,
      "step": 28811
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6802610158920288,
      "learning_rate": 0.0005771456586084873,
      "loss": 2.2317,
      "step": 28812
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6216834187507629,
      "learning_rate": 0.0005771440925933354,
      "loss": 2.492,
      "step": 28813
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6056867241859436,
      "learning_rate": 0.0005771425265266571,
      "loss": 2.3324,
      "step": 28814
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.638169527053833,
      "learning_rate": 0.0005771409604084528,
      "loss": 2.2493,
      "step": 28815
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6554783582687378,
      "learning_rate": 0.0005771393942387227,
      "loss": 2.4234,
      "step": 28816
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.600606381893158,
      "learning_rate": 0.0005771378280174672,
      "loss": 2.3909,
      "step": 28817
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.7054585218429565,
      "learning_rate": 0.0005771362617446865,
      "loss": 2.3707,
      "step": 28818
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.695422351360321,
      "learning_rate": 0.0005771346954203811,
      "loss": 2.3489,
      "step": 28819
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6004781723022461,
      "learning_rate": 0.0005771331290445509,
      "loss": 2.2466,
      "step": 28820
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6323478817939758,
      "learning_rate": 0.0005771315626171966,
      "loss": 2.4147,
      "step": 28821
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6925303936004639,
      "learning_rate": 0.0005771299961383181,
      "loss": 2.5423,
      "step": 28822
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.705274760723114,
      "learning_rate": 0.0005771284296079161,
      "loss": 2.2944,
      "step": 28823
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6022634506225586,
      "learning_rate": 0.0005771268630259907,
      "loss": 2.2697,
      "step": 28824
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6504960060119629,
      "learning_rate": 0.0005771252963925421,
      "loss": 2.3127,
      "step": 28825
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.7076340913772583,
      "learning_rate": 0.0005771237297075707,
      "loss": 2.226,
      "step": 28826
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.9414168000221252,
      "learning_rate": 0.0005771221629710767,
      "loss": 2.3529,
      "step": 28827
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6720606088638306,
      "learning_rate": 0.0005771205961830606,
      "loss": 2.2726,
      "step": 28828
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.7778528928756714,
      "learning_rate": 0.0005771190293435224,
      "loss": 2.4256,
      "step": 28829
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.7114375233650208,
      "learning_rate": 0.0005771174624524626,
      "loss": 2.3138,
      "step": 28830
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6406732201576233,
      "learning_rate": 0.0005771158955098815,
      "loss": 2.4965,
      "step": 28831
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.7142686247825623,
      "learning_rate": 0.0005771143285157792,
      "loss": 2.5564,
      "step": 28832
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.7487132549285889,
      "learning_rate": 0.0005771127614701562,
      "loss": 2.1537,
      "step": 28833
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.5978951454162598,
      "learning_rate": 0.0005771111943730127,
      "loss": 2.308,
      "step": 28834
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.7594525218009949,
      "learning_rate": 0.000577109627224349,
      "loss": 2.4696,
      "step": 28835
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.7615347504615784,
      "learning_rate": 0.0005771080600241653,
      "loss": 2.1817,
      "step": 28836
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6621307134628296,
      "learning_rate": 0.0005771064927724621,
      "loss": 2.411,
      "step": 28837
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6465635895729065,
      "learning_rate": 0.0005771049254692396,
      "loss": 2.2916,
      "step": 28838
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.8626833558082581,
      "learning_rate": 0.0005771033581144979,
      "loss": 2.2469,
      "step": 28839
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.7528290748596191,
      "learning_rate": 0.0005771017907082376,
      "loss": 2.2218,
      "step": 28840
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.639141857624054,
      "learning_rate": 0.0005771002232504588,
      "loss": 2.5375,
      "step": 28841
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.754686713218689,
      "learning_rate": 0.0005770986557411618,
      "loss": 2.4322,
      "step": 28842
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.7635363936424255,
      "learning_rate": 0.000577097088180347,
      "loss": 2.2529,
      "step": 28843
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6370103359222412,
      "learning_rate": 0.0005770955205680146,
      "loss": 2.2074,
      "step": 28844
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.5924755930900574,
      "learning_rate": 0.0005770939529041648,
      "loss": 2.2074,
      "step": 28845
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.7715235352516174,
      "learning_rate": 0.0005770923851887983,
      "loss": 2.3102,
      "step": 28846
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.7605066895484924,
      "learning_rate": 0.0005770908174219148,
      "loss": 2.3205,
      "step": 28847
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6213414669036865,
      "learning_rate": 0.000577089249603515,
      "loss": 2.2842,
      "step": 28848
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.7168876528739929,
      "learning_rate": 0.000577087681733599,
      "loss": 2.1968,
      "step": 28849
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.7294461727142334,
      "learning_rate": 0.0005770861138121672,
      "loss": 2.2326,
      "step": 28850
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6325283050537109,
      "learning_rate": 0.0005770845458392199,
      "loss": 2.282,
      "step": 28851
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.734786331653595,
      "learning_rate": 0.0005770829778147574,
      "loss": 2.4345,
      "step": 28852
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6530753970146179,
      "learning_rate": 0.0005770814097387798,
      "loss": 2.3081,
      "step": 28853
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6342902183532715,
      "learning_rate": 0.0005770798416112877,
      "loss": 2.2267,
      "step": 28854
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.7091478109359741,
      "learning_rate": 0.0005770782734322811,
      "loss": 2.0235,
      "step": 28855
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.7125817537307739,
      "learning_rate": 0.0005770767052017605,
      "loss": 2.4278,
      "step": 28856
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6651865243911743,
      "learning_rate": 0.000577075136919726,
      "loss": 2.6732,
      "step": 28857
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.5946300029754639,
      "learning_rate": 0.0005770735685861781,
      "loss": 2.2886,
      "step": 28858
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6546531319618225,
      "learning_rate": 0.000577072000201117,
      "loss": 2.2062,
      "step": 28859
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6479920148849487,
      "learning_rate": 0.000577070431764543,
      "loss": 2.4687,
      "step": 28860
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6389589309692383,
      "learning_rate": 0.0005770688632764563,
      "loss": 2.2201,
      "step": 28861
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.7274736762046814,
      "learning_rate": 0.0005770672947368573,
      "loss": 2.5286,
      "step": 28862
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6133642792701721,
      "learning_rate": 0.0005770657261457463,
      "loss": 2.3771,
      "step": 28863
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6297001242637634,
      "learning_rate": 0.0005770641575031235,
      "loss": 2.1712,
      "step": 28864
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6306612491607666,
      "learning_rate": 0.0005770625888089892,
      "loss": 2.2985,
      "step": 28865
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6479396820068359,
      "learning_rate": 0.0005770610200633439,
      "loss": 2.3362,
      "step": 28866
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6752665042877197,
      "learning_rate": 0.0005770594512661876,
      "loss": 2.3136,
      "step": 28867
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.670941948890686,
      "learning_rate": 0.0005770578824175207,
      "loss": 2.4329,
      "step": 28868
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6458055973052979,
      "learning_rate": 0.0005770563135173437,
      "loss": 2.3072,
      "step": 28869
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6103219985961914,
      "learning_rate": 0.0005770547445656563,
      "loss": 2.2235,
      "step": 28870
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.7496943473815918,
      "learning_rate": 0.0005770531755624595,
      "loss": 2.2561,
      "step": 28871
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6158789396286011,
      "learning_rate": 0.0005770516065077533,
      "loss": 2.3218,
      "step": 28872
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6162179708480835,
      "learning_rate": 0.0005770500374015379,
      "loss": 2.1097,
      "step": 28873
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6414710283279419,
      "learning_rate": 0.0005770484682438137,
      "loss": 2.2889,
      "step": 28874
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6395383477210999,
      "learning_rate": 0.0005770468990345809,
      "loss": 2.3533,
      "step": 28875
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6088885068893433,
      "learning_rate": 0.0005770453297738398,
      "loss": 2.3617,
      "step": 28876
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.5981021523475647,
      "learning_rate": 0.0005770437604615909,
      "loss": 2.366,
      "step": 28877
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.636438250541687,
      "learning_rate": 0.0005770421910978342,
      "loss": 2.1732,
      "step": 28878
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.5768280625343323,
      "learning_rate": 0.0005770406216825702,
      "loss": 2.0649,
      "step": 28879
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6393447518348694,
      "learning_rate": 0.0005770390522157992,
      "loss": 2.281,
      "step": 28880
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6470363140106201,
      "learning_rate": 0.0005770374826975212,
      "loss": 2.5611,
      "step": 28881
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6114509701728821,
      "learning_rate": 0.000577035913127737,
      "loss": 2.2244,
      "step": 28882
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6514571309089661,
      "learning_rate": 0.0005770343435064463,
      "loss": 2.2979,
      "step": 28883
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6784536838531494,
      "learning_rate": 0.0005770327738336498,
      "loss": 2.3916,
      "step": 28884
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6067877411842346,
      "learning_rate": 0.0005770312041093477,
      "loss": 2.3655,
      "step": 28885
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6632479429244995,
      "learning_rate": 0.0005770296343335402,
      "loss": 2.2015,
      "step": 28886
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6265417337417603,
      "learning_rate": 0.0005770280645062278,
      "loss": 2.2491,
      "step": 28887
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.5708823204040527,
      "learning_rate": 0.0005770264946274104,
      "loss": 2.1228,
      "step": 28888
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.8169800639152527,
      "learning_rate": 0.0005770249246970888,
      "loss": 2.3073,
      "step": 28889
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6301298141479492,
      "learning_rate": 0.000577023354715263,
      "loss": 2.3271,
      "step": 28890
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.685756504535675,
      "learning_rate": 0.0005770217846819333,
      "loss": 2.32,
      "step": 28891
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6862484812736511,
      "learning_rate": 0.0005770202145970999,
      "loss": 2.1808,
      "step": 28892
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6726877093315125,
      "learning_rate": 0.0005770186444607634,
      "loss": 2.2981,
      "step": 28893
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6069682836532593,
      "learning_rate": 0.0005770170742729237,
      "loss": 2.3685,
      "step": 28894
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6606578230857849,
      "learning_rate": 0.0005770155040335814,
      "loss": 2.1999,
      "step": 28895
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6380652785301208,
      "learning_rate": 0.0005770139337427368,
      "loss": 2.3386,
      "step": 28896
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.624204695224762,
      "learning_rate": 0.00057701236340039,
      "loss": 2.5365,
      "step": 28897
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6868918538093567,
      "learning_rate": 0.0005770107930065413,
      "loss": 2.2772,
      "step": 28898
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.7150828838348389,
      "learning_rate": 0.0005770092225611911,
      "loss": 2.3431,
      "step": 28899
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6755093336105347,
      "learning_rate": 0.0005770076520643397,
      "loss": 2.4022,
      "step": 28900
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6376550197601318,
      "learning_rate": 0.0005770060815159873,
      "loss": 2.3238,
      "step": 28901
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6946691870689392,
      "learning_rate": 0.0005770045109161343,
      "loss": 2.087,
      "step": 28902
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6498968005180359,
      "learning_rate": 0.0005770029402647809,
      "loss": 2.1507,
      "step": 28903
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6743413209915161,
      "learning_rate": 0.0005770013695619274,
      "loss": 2.6067,
      "step": 28904
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6478803157806396,
      "learning_rate": 0.0005769997988075741,
      "loss": 2.2061,
      "step": 28905
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6505470871925354,
      "learning_rate": 0.0005769982280017215,
      "loss": 2.0763,
      "step": 28906
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6373049020767212,
      "learning_rate": 0.0005769966571443695,
      "loss": 2.178,
      "step": 28907
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6255633234977722,
      "learning_rate": 0.0005769950862355186,
      "loss": 2.3327,
      "step": 28908
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6214558482170105,
      "learning_rate": 0.0005769935152751691,
      "loss": 2.101,
      "step": 28909
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6089237332344055,
      "learning_rate": 0.0005769919442633213,
      "loss": 2.1554,
      "step": 28910
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6491646766662598,
      "learning_rate": 0.0005769903731999755,
      "loss": 2.3153,
      "step": 28911
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6294799447059631,
      "learning_rate": 0.0005769888020851319,
      "loss": 2.4596,
      "step": 28912
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6253291964530945,
      "learning_rate": 0.0005769872309187909,
      "loss": 2.3718,
      "step": 28913
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6475580930709839,
      "learning_rate": 0.0005769856597009527,
      "loss": 2.3065,
      "step": 28914
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6835643649101257,
      "learning_rate": 0.0005769840884316177,
      "loss": 2.2359,
      "step": 28915
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6048765182495117,
      "learning_rate": 0.0005769825171107861,
      "loss": 2.3238,
      "step": 28916
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6651504039764404,
      "learning_rate": 0.0005769809457384582,
      "loss": 2.2961,
      "step": 28917
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6950404047966003,
      "learning_rate": 0.0005769793743146343,
      "loss": 2.2344,
      "step": 28918
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6599883437156677,
      "learning_rate": 0.0005769778028393148,
      "loss": 2.2715,
      "step": 28919
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6880359649658203,
      "learning_rate": 0.0005769762313124998,
      "loss": 2.2799,
      "step": 28920
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.7100219130516052,
      "learning_rate": 0.0005769746597341898,
      "loss": 2.4714,
      "step": 28921
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6883043050765991,
      "learning_rate": 0.000576973088104385,
      "loss": 2.4371,
      "step": 28922
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.7317346930503845,
      "learning_rate": 0.0005769715164230854,
      "loss": 2.3363,
      "step": 28923
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.8153489828109741,
      "learning_rate": 0.0005769699446902918,
      "loss": 2.2329,
      "step": 28924
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.7264679670333862,
      "learning_rate": 0.0005769683729060042,
      "loss": 2.4667,
      "step": 28925
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6740686893463135,
      "learning_rate": 0.0005769668010702229,
      "loss": 2.1441,
      "step": 28926
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.624826192855835,
      "learning_rate": 0.0005769652291829484,
      "loss": 2.2855,
      "step": 28927
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6543669700622559,
      "learning_rate": 0.0005769636572441806,
      "loss": 2.2091,
      "step": 28928
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.726658284664154,
      "learning_rate": 0.0005769620852539203,
      "loss": 2.1012,
      "step": 28929
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6952409148216248,
      "learning_rate": 0.0005769605132121674,
      "loss": 2.5558,
      "step": 28930
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.7621489763259888,
      "learning_rate": 0.0005769589411189223,
      "loss": 2.4036,
      "step": 28931
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6293886303901672,
      "learning_rate": 0.0005769573689741853,
      "loss": 1.9686,
      "step": 28932
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6120564341545105,
      "learning_rate": 0.0005769557967779566,
      "loss": 2.331,
      "step": 28933
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6257121562957764,
      "learning_rate": 0.0005769542245302367,
      "loss": 2.3986,
      "step": 28934
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6581952571868896,
      "learning_rate": 0.0005769526522310258,
      "loss": 2.236,
      "step": 28935
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6467458605766296,
      "learning_rate": 0.0005769510798803242,
      "loss": 2.1845,
      "step": 28936
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6877515912055969,
      "learning_rate": 0.0005769495074781321,
      "loss": 2.2548,
      "step": 28937
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6074120402336121,
      "learning_rate": 0.0005769479350244499,
      "loss": 2.2014,
      "step": 28938
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6548745632171631,
      "learning_rate": 0.0005769463625192778,
      "loss": 2.4073,
      "step": 28939
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6680818796157837,
      "learning_rate": 0.0005769447899626162,
      "loss": 2.1669,
      "step": 28940
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.7175577282905579,
      "learning_rate": 0.0005769432173544653,
      "loss": 2.4372,
      "step": 28941
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6960780024528503,
      "learning_rate": 0.0005769416446948254,
      "loss": 2.4886,
      "step": 28942
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.7391253113746643,
      "learning_rate": 0.0005769400719836969,
      "loss": 2.5407,
      "step": 28943
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6460922360420227,
      "learning_rate": 0.0005769384992210799,
      "loss": 2.3511,
      "step": 28944
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.5940620303153992,
      "learning_rate": 0.0005769369264069749,
      "loss": 2.3537,
      "step": 28945
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.5915679335594177,
      "learning_rate": 0.0005769353535413821,
      "loss": 2.415,
      "step": 28946
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.63373863697052,
      "learning_rate": 0.0005769337806243017,
      "loss": 2.4032,
      "step": 28947
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.586257815361023,
      "learning_rate": 0.0005769322076557341,
      "loss": 2.2863,
      "step": 28948
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6272820830345154,
      "learning_rate": 0.0005769306346356796,
      "loss": 2.3994,
      "step": 28949
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6600862145423889,
      "learning_rate": 0.0005769290615641386,
      "loss": 2.5372,
      "step": 28950
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6352044939994812,
      "learning_rate": 0.0005769274884411111,
      "loss": 2.3022,
      "step": 28951
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.7261090874671936,
      "learning_rate": 0.0005769259152665976,
      "loss": 2.3778,
      "step": 28952
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6392658352851868,
      "learning_rate": 0.0005769243420405984,
      "loss": 2.2316,
      "step": 28953
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.584554135799408,
      "learning_rate": 0.0005769227687631136,
      "loss": 2.3516,
      "step": 28954
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6735396981239319,
      "learning_rate": 0.0005769211954341438,
      "loss": 2.3512,
      "step": 28955
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6288568377494812,
      "learning_rate": 0.0005769196220536889,
      "loss": 2.4017,
      "step": 28956
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6000200510025024,
      "learning_rate": 0.0005769180486217497,
      "loss": 2.2007,
      "step": 28957
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6230291724205017,
      "learning_rate": 0.0005769164751383259,
      "loss": 2.1136,
      "step": 28958
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.707669198513031,
      "learning_rate": 0.0005769149016034183,
      "loss": 2.169,
      "step": 28959
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6905189752578735,
      "learning_rate": 0.0005769133280170269,
      "loss": 2.2053,
      "step": 28960
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6189415454864502,
      "learning_rate": 0.0005769117543791522,
      "loss": 2.2158,
      "step": 28961
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6723277568817139,
      "learning_rate": 0.0005769101806897943,
      "loss": 2.4334,
      "step": 28962
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6574307084083557,
      "learning_rate": 0.0005769086069489536,
      "loss": 2.4015,
      "step": 28963
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6530323028564453,
      "learning_rate": 0.0005769070331566303,
      "loss": 2.3499,
      "step": 28964
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.7889243364334106,
      "learning_rate": 0.0005769054593128248,
      "loss": 2.3791,
      "step": 28965
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6870569586753845,
      "learning_rate": 0.0005769038854175374,
      "loss": 2.2164,
      "step": 28966
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.8351798057556152,
      "learning_rate": 0.0005769023114707682,
      "loss": 2.2864,
      "step": 28967
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.7730582356452942,
      "learning_rate": 0.0005769007374725178,
      "loss": 2.654,
      "step": 28968
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6521185636520386,
      "learning_rate": 0.0005768991634227862,
      "loss": 2.2837,
      "step": 28969
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6737373471260071,
      "learning_rate": 0.0005768975893215739,
      "loss": 2.3703,
      "step": 28970
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.7475262880325317,
      "learning_rate": 0.000576896015168881,
      "loss": 2.334,
      "step": 28971
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6243593692779541,
      "learning_rate": 0.000576894440964708,
      "loss": 2.4384,
      "step": 28972
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6523253917694092,
      "learning_rate": 0.0005768928667090551,
      "loss": 2.3685,
      "step": 28973
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.5834090709686279,
      "learning_rate": 0.0005768912924019226,
      "loss": 2.3837,
      "step": 28974
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6579153537750244,
      "learning_rate": 0.0005768897180433107,
      "loss": 2.2358,
      "step": 28975
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6605518460273743,
      "learning_rate": 0.0005768881436332199,
      "loss": 2.4693,
      "step": 28976
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6516912579536438,
      "learning_rate": 0.0005768865691716502,
      "loss": 2.4232,
      "step": 28977
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6504937410354614,
      "learning_rate": 0.0005768849946586022,
      "loss": 2.4115,
      "step": 28978
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.5951379537582397,
      "learning_rate": 0.000576883420094076,
      "loss": 2.231,
      "step": 28979
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6580698490142822,
      "learning_rate": 0.0005768818454780719,
      "loss": 2.2906,
      "step": 28980
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.5792244076728821,
      "learning_rate": 0.0005768802708105903,
      "loss": 2.4365,
      "step": 28981
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.597119152545929,
      "learning_rate": 0.0005768786960916315,
      "loss": 2.1833,
      "step": 28982
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6754339337348938,
      "learning_rate": 0.0005768771213211955,
      "loss": 2.2788,
      "step": 28983
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.680476188659668,
      "learning_rate": 0.0005768755464992831,
      "loss": 2.3353,
      "step": 28984
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.5924699306488037,
      "learning_rate": 0.0005768739716258941,
      "loss": 2.4114,
      "step": 28985
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.5958264470100403,
      "learning_rate": 0.0005768723967010291,
      "loss": 2.1357,
      "step": 28986
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.7931926250457764,
      "learning_rate": 0.0005768708217246882,
      "loss": 2.5021,
      "step": 28987
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.7301259636878967,
      "learning_rate": 0.0005768692466968718,
      "loss": 2.3527,
      "step": 28988
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6481860280036926,
      "learning_rate": 0.0005768676716175803,
      "loss": 2.665,
      "step": 28989
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6284952759742737,
      "learning_rate": 0.0005768660964868139,
      "loss": 2.3136,
      "step": 28990
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6504234671592712,
      "learning_rate": 0.0005768645213045727,
      "loss": 2.3545,
      "step": 28991
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6412996053695679,
      "learning_rate": 0.0005768629460708572,
      "loss": 2.5944,
      "step": 28992
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6715093851089478,
      "learning_rate": 0.0005768613707856677,
      "loss": 2.3335,
      "step": 28993
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.5927068591117859,
      "learning_rate": 0.0005768597954490045,
      "loss": 2.3691,
      "step": 28994
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.7018805146217346,
      "learning_rate": 0.0005768582200608677,
      "loss": 2.3174,
      "step": 28995
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6683371663093567,
      "learning_rate": 0.0005768566446212579,
      "loss": 2.2082,
      "step": 28996
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6349066495895386,
      "learning_rate": 0.0005768550691301751,
      "loss": 2.4066,
      "step": 28997
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6569281816482544,
      "learning_rate": 0.0005768534935876198,
      "loss": 2.2741,
      "step": 28998
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6618912220001221,
      "learning_rate": 0.0005768519179935921,
      "loss": 2.2718,
      "step": 28999
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6766603589057922,
      "learning_rate": 0.0005768503423480924,
      "loss": 2.2213,
      "step": 29000
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.7418363690376282,
      "learning_rate": 0.0005768487666511211,
      "loss": 2.3999,
      "step": 29001
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6445176005363464,
      "learning_rate": 0.0005768471909026783,
      "loss": 2.3148,
      "step": 29002
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.8146662712097168,
      "learning_rate": 0.0005768456151027645,
      "loss": 2.22,
      "step": 29003
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6216662526130676,
      "learning_rate": 0.0005768440392513798,
      "loss": 2.2891,
      "step": 29004
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6963326930999756,
      "learning_rate": 0.0005768424633485244,
      "loss": 2.1246,
      "step": 29005
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.8786529898643494,
      "learning_rate": 0.000576840887394199,
      "loss": 2.1816,
      "step": 29006
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6260676980018616,
      "learning_rate": 0.0005768393113884036,
      "loss": 2.5095,
      "step": 29007
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.9289641976356506,
      "learning_rate": 0.0005768377353311385,
      "loss": 2.2366,
      "step": 29008
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.7162064909934998,
      "learning_rate": 0.000576836159222404,
      "loss": 2.1823,
      "step": 29009
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6696396470069885,
      "learning_rate": 0.0005768345830622005,
      "loss": 2.2963,
      "step": 29010
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.8039302229881287,
      "learning_rate": 0.0005768330068505282,
      "loss": 2.38,
      "step": 29011
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.8149538040161133,
      "learning_rate": 0.0005768314305873876,
      "loss": 2.4583,
      "step": 29012
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6428437232971191,
      "learning_rate": 0.0005768298542727786,
      "loss": 2.3191,
      "step": 29013
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.8856045007705688,
      "learning_rate": 0.0005768282779067017,
      "loss": 2.4557,
      "step": 29014
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.8731096982955933,
      "learning_rate": 0.0005768267014891571,
      "loss": 2.2662,
      "step": 29015
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6267855763435364,
      "learning_rate": 0.0005768251250201455,
      "loss": 2.3057,
      "step": 29016
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6578879356384277,
      "learning_rate": 0.0005768235484996667,
      "loss": 2.3026,
      "step": 29017
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.9686912298202515,
      "learning_rate": 0.0005768219719277212,
      "loss": 2.4002,
      "step": 29018
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6577351093292236,
      "learning_rate": 0.0005768203953043093,
      "loss": 2.201,
      "step": 29019
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6786640882492065,
      "learning_rate": 0.0005768188186294312,
      "loss": 2.2293,
      "step": 29020
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.653462827205658,
      "learning_rate": 0.0005768172419030873,
      "loss": 2.4288,
      "step": 29021
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.86440110206604,
      "learning_rate": 0.0005768156651252778,
      "loss": 2.2045,
      "step": 29022
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6464223861694336,
      "learning_rate": 0.0005768140882960031,
      "loss": 2.1181,
      "step": 29023
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.5957608819007874,
      "learning_rate": 0.0005768125114152634,
      "loss": 2.3563,
      "step": 29024
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.7241095900535583,
      "learning_rate": 0.000576810934483059,
      "loss": 2.4012,
      "step": 29025
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.7183818817138672,
      "learning_rate": 0.0005768093574993903,
      "loss": 2.3317,
      "step": 29026
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6956241726875305,
      "learning_rate": 0.0005768077804642575,
      "loss": 2.2565,
      "step": 29027
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6577485799789429,
      "learning_rate": 0.0005768062033776609,
      "loss": 2.5309,
      "step": 29028
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.7486410737037659,
      "learning_rate": 0.0005768046262396007,
      "loss": 2.3257,
      "step": 29029
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6928936243057251,
      "learning_rate": 0.0005768030490500774,
      "loss": 2.2876,
      "step": 29030
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.5883214473724365,
      "learning_rate": 0.0005768014718090912,
      "loss": 2.1008,
      "step": 29031
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.630246639251709,
      "learning_rate": 0.0005767998945166424,
      "loss": 2.1001,
      "step": 29032
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.7156790494918823,
      "learning_rate": 0.0005767983171727312,
      "loss": 2.4971,
      "step": 29033
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6596051454544067,
      "learning_rate": 0.0005767967397773579,
      "loss": 2.161,
      "step": 29034
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6614871621131897,
      "learning_rate": 0.0005767951623305231,
      "loss": 2.3348,
      "step": 29035
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.647345244884491,
      "learning_rate": 0.0005767935848322267,
      "loss": 2.3529,
      "step": 29036
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.7147456407546997,
      "learning_rate": 0.0005767920072824692,
      "loss": 2.2996,
      "step": 29037
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.7050433158874512,
      "learning_rate": 0.0005767904296812507,
      "loss": 2.1356,
      "step": 29038
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.5796345472335815,
      "learning_rate": 0.0005767888520285718,
      "loss": 2.2478,
      "step": 29039
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6398019194602966,
      "learning_rate": 0.0005767872743244326,
      "loss": 2.1998,
      "step": 29040
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.7474173307418823,
      "learning_rate": 0.0005767856965688333,
      "loss": 2.3358,
      "step": 29041
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.642169713973999,
      "learning_rate": 0.0005767841187617744,
      "loss": 2.3218,
      "step": 29042
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6225684285163879,
      "learning_rate": 0.0005767825409032562,
      "loss": 2.2179,
      "step": 29043
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.8278326988220215,
      "learning_rate": 0.0005767809629932788,
      "loss": 2.2736,
      "step": 29044
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.708182692527771,
      "learning_rate": 0.0005767793850318426,
      "loss": 2.315,
      "step": 29045
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6379746794700623,
      "learning_rate": 0.0005767778070189479,
      "loss": 2.3162,
      "step": 29046
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.7087400555610657,
      "learning_rate": 0.000576776228954595,
      "loss": 2.3026,
      "step": 29047
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.7185099720954895,
      "learning_rate": 0.0005767746508387841,
      "loss": 2.4464,
      "step": 29048
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6890617609024048,
      "learning_rate": 0.0005767730726715155,
      "loss": 2.3408,
      "step": 29049
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.7445822954177856,
      "learning_rate": 0.0005767714944527898,
      "loss": 2.2841,
      "step": 29050
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6089953780174255,
      "learning_rate": 0.0005767699161826069,
      "loss": 2.2608,
      "step": 29051
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6090492010116577,
      "learning_rate": 0.0005767683378609672,
      "loss": 2.2089,
      "step": 29052
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6486918926239014,
      "learning_rate": 0.0005767667594878712,
      "loss": 2.4763,
      "step": 29053
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6345382928848267,
      "learning_rate": 0.000576765181063319,
      "loss": 2.3618,
      "step": 29054
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.7406172156333923,
      "learning_rate": 0.0005767636025873109,
      "loss": 2.3532,
      "step": 29055
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6463865637779236,
      "learning_rate": 0.0005767620240598472,
      "loss": 2.3245,
      "step": 29056
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.7009931802749634,
      "learning_rate": 0.0005767604454809281,
      "loss": 2.5607,
      "step": 29057
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6337999105453491,
      "learning_rate": 0.0005767588668505543,
      "loss": 2.252,
      "step": 29058
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6489060521125793,
      "learning_rate": 0.0005767572881687255,
      "loss": 2.3074,
      "step": 29059
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.629217803478241,
      "learning_rate": 0.0005767557094354424,
      "loss": 2.3534,
      "step": 29060
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.735877513885498,
      "learning_rate": 0.0005767541306507053,
      "loss": 2.3167,
      "step": 29061
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6251345276832581,
      "learning_rate": 0.0005767525518145143,
      "loss": 2.3912,
      "step": 29062
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6383100748062134,
      "learning_rate": 0.0005767509729268698,
      "loss": 2.2152,
      "step": 29063
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.7006266713142395,
      "learning_rate": 0.0005767493939877721,
      "loss": 2.2106,
      "step": 29064
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.7840138077735901,
      "learning_rate": 0.0005767478149972214,
      "loss": 2.1474,
      "step": 29065
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6056737303733826,
      "learning_rate": 0.000576746235955218,
      "loss": 2.3284,
      "step": 29066
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.7408056855201721,
      "learning_rate": 0.0005767446568617623,
      "loss": 2.4309,
      "step": 29067
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.8222500681877136,
      "learning_rate": 0.0005767430777168545,
      "loss": 2.2742,
      "step": 29068
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.7311285138130188,
      "learning_rate": 0.000576741498520495,
      "loss": 2.3702,
      "step": 29069
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6843715310096741,
      "learning_rate": 0.0005767399192726841,
      "loss": 2.2133,
      "step": 29070
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6582217812538147,
      "learning_rate": 0.000576738339973422,
      "loss": 2.3543,
      "step": 29071
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.7027262449264526,
      "learning_rate": 0.0005767367606227089,
      "loss": 2.3859,
      "step": 29072
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6431116461753845,
      "learning_rate": 0.0005767351812205453,
      "loss": 2.2961,
      "step": 29073
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.666487455368042,
      "learning_rate": 0.0005767336017669314,
      "loss": 2.5986,
      "step": 29074
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.7342921495437622,
      "learning_rate": 0.0005767320222618675,
      "loss": 2.2715,
      "step": 29075
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.7592676281929016,
      "learning_rate": 0.0005767304427053538,
      "loss": 2.531,
      "step": 29076
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6821624040603638,
      "learning_rate": 0.000576728863097391,
      "loss": 2.4333,
      "step": 29077
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6672846078872681,
      "learning_rate": 0.0005767272834379787,
      "loss": 2.3478,
      "step": 29078
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6739605665206909,
      "learning_rate": 0.0005767257037271177,
      "loss": 2.2057,
      "step": 29079
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.7255631685256958,
      "learning_rate": 0.0005767241239648083,
      "loss": 2.315,
      "step": 29080
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.65365070104599,
      "learning_rate": 0.0005767225441510505,
      "loss": 2.2822,
      "step": 29081
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6148008704185486,
      "learning_rate": 0.0005767209642858448,
      "loss": 2.2355,
      "step": 29082
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.7061678171157837,
      "learning_rate": 0.0005767193843691915,
      "loss": 2.3905,
      "step": 29083
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.7091997861862183,
      "learning_rate": 0.0005767178044010908,
      "loss": 2.2099,
      "step": 29084
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6583979725837708,
      "learning_rate": 0.0005767162243815431,
      "loss": 2.3432,
      "step": 29085
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6698991060256958,
      "learning_rate": 0.0005767146443105485,
      "loss": 2.1689,
      "step": 29086
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.7106232047080994,
      "learning_rate": 0.0005767130641881076,
      "loss": 2.213,
      "step": 29087
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6603208184242249,
      "learning_rate": 0.0005767114840142203,
      "loss": 2.3635,
      "step": 29088
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.7310968041419983,
      "learning_rate": 0.0005767099037888873,
      "loss": 2.2543,
      "step": 29089
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6706331968307495,
      "learning_rate": 0.0005767083235121086,
      "loss": 2.2142,
      "step": 29090
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.5770325660705566,
      "learning_rate": 0.0005767067431838846,
      "loss": 2.1591,
      "step": 29091
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6276244521141052,
      "learning_rate": 0.0005767051628042157,
      "loss": 2.3239,
      "step": 29092
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6139776110649109,
      "learning_rate": 0.000576703582373102,
      "loss": 2.0891,
      "step": 29093
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6261215806007385,
      "learning_rate": 0.0005767020018905438,
      "loss": 2.5803,
      "step": 29094
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6536431312561035,
      "learning_rate": 0.0005767004213565416,
      "loss": 2.2661,
      "step": 29095
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.622279942035675,
      "learning_rate": 0.0005766988407710956,
      "loss": 2.1091,
      "step": 29096
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6102525591850281,
      "learning_rate": 0.000576697260134206,
      "loss": 2.5469,
      "step": 29097
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6196103096008301,
      "learning_rate": 0.0005766956794458732,
      "loss": 2.2252,
      "step": 29098
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.8043172359466553,
      "learning_rate": 0.0005766940987060973,
      "loss": 2.3381,
      "step": 29099
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6359255313873291,
      "learning_rate": 0.0005766925179148789,
      "loss": 2.5593,
      "step": 29100
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.653697669506073,
      "learning_rate": 0.0005766909370722181,
      "loss": 2.3631,
      "step": 29101
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.8369396924972534,
      "learning_rate": 0.0005766893561781152,
      "loss": 2.3077,
      "step": 29102
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6997760534286499,
      "learning_rate": 0.0005766877752325706,
      "loss": 2.4432,
      "step": 29103
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6108224987983704,
      "learning_rate": 0.0005766861942355844,
      "loss": 2.3595,
      "step": 29104
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.615565836429596,
      "learning_rate": 0.0005766846131871571,
      "loss": 2.3308,
      "step": 29105
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6807066202163696,
      "learning_rate": 0.0005766830320872889,
      "loss": 2.2294,
      "step": 29106
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6515559554100037,
      "learning_rate": 0.0005766814509359801,
      "loss": 2.3697,
      "step": 29107
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6605409979820251,
      "learning_rate": 0.000576679869733231,
      "loss": 2.5117,
      "step": 29108
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6788201332092285,
      "learning_rate": 0.0005766782884790418,
      "loss": 2.3385,
      "step": 29109
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.5761340260505676,
      "learning_rate": 0.000576676707173413,
      "loss": 2.111,
      "step": 29110
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6897692084312439,
      "learning_rate": 0.0005766751258163448,
      "loss": 2.3479,
      "step": 29111
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6734988689422607,
      "learning_rate": 0.0005766735444078374,
      "loss": 2.3222,
      "step": 29112
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6758120656013489,
      "learning_rate": 0.0005766719629478911,
      "loss": 2.4752,
      "step": 29113
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6390287280082703,
      "learning_rate": 0.0005766703814365063,
      "loss": 2.249,
      "step": 29114
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6227551102638245,
      "learning_rate": 0.0005766687998736834,
      "loss": 2.5581,
      "step": 29115
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6128634810447693,
      "learning_rate": 0.0005766672182594224,
      "loss": 2.4647,
      "step": 29116
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6497339606285095,
      "learning_rate": 0.0005766656365937238,
      "loss": 2.3069,
      "step": 29117
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6865933537483215,
      "learning_rate": 0.0005766640548765877,
      "loss": 2.3894,
      "step": 29118
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6313251852989197,
      "learning_rate": 0.0005766624731080147,
      "loss": 2.4798,
      "step": 29119
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6464540362358093,
      "learning_rate": 0.0005766608912880049,
      "loss": 2.2197,
      "step": 29120
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6243706345558167,
      "learning_rate": 0.0005766593094165584,
      "loss": 2.2166,
      "step": 29121
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6751813888549805,
      "learning_rate": 0.000576657727493676,
      "loss": 2.2703,
      "step": 29122
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6441845893859863,
      "learning_rate": 0.0005766561455193576,
      "loss": 2.3222,
      "step": 29123
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.7773994207382202,
      "learning_rate": 0.0005766545634936036,
      "loss": 2.5351,
      "step": 29124
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6262161135673523,
      "learning_rate": 0.0005766529814164144,
      "loss": 2.2379,
      "step": 29125
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.7008112072944641,
      "learning_rate": 0.00057665139928779,
      "loss": 2.3527,
      "step": 29126
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6955101490020752,
      "learning_rate": 0.000576649817107731,
      "loss": 2.3798,
      "step": 29127
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6933193206787109,
      "learning_rate": 0.0005766482348762375,
      "loss": 2.2744,
      "step": 29128
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.5880160331726074,
      "learning_rate": 0.00057664665259331,
      "loss": 2.3779,
      "step": 29129
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6549177765846252,
      "learning_rate": 0.0005766450702589486,
      "loss": 2.2286,
      "step": 29130
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6464308500289917,
      "learning_rate": 0.0005766434878731536,
      "loss": 2.2957,
      "step": 29131
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6776045560836792,
      "learning_rate": 0.0005766419054359253,
      "loss": 2.3193,
      "step": 29132
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6875136494636536,
      "learning_rate": 0.0005766403229472641,
      "loss": 2.4116,
      "step": 29133
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6690537929534912,
      "learning_rate": 0.0005766387404071703,
      "loss": 2.3618,
      "step": 29134
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.630226731300354,
      "learning_rate": 0.0005766371578156441,
      "loss": 2.2611,
      "step": 29135
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.7565705180168152,
      "learning_rate": 0.0005766355751726859,
      "loss": 2.1151,
      "step": 29136
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6478314399719238,
      "learning_rate": 0.0005766339924782958,
      "loss": 2.1486,
      "step": 29137
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6825256943702698,
      "learning_rate": 0.0005766324097324744,
      "loss": 2.2238,
      "step": 29138
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6279507875442505,
      "learning_rate": 0.0005766308269352216,
      "loss": 2.3844,
      "step": 29139
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.7018426060676575,
      "learning_rate": 0.0005766292440865381,
      "loss": 2.4148,
      "step": 29140
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6142566204071045,
      "learning_rate": 0.0005766276611864239,
      "loss": 2.3086,
      "step": 29141
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6382827758789062,
      "learning_rate": 0.0005766260782348794,
      "loss": 2.3902,
      "step": 29142
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6816795468330383,
      "learning_rate": 0.0005766244952319049,
      "loss": 2.3443,
      "step": 29143
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6617851257324219,
      "learning_rate": 0.0005766229121775008,
      "loss": 2.2968,
      "step": 29144
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6537097692489624,
      "learning_rate": 0.0005766213290716671,
      "loss": 2.4009,
      "step": 29145
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6495149731636047,
      "learning_rate": 0.0005766197459144042,
      "loss": 2.5055,
      "step": 29146
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.60542893409729,
      "learning_rate": 0.0005766181627057127,
      "loss": 2.362,
      "step": 29147
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6550726890563965,
      "learning_rate": 0.0005766165794455926,
      "loss": 2.2794,
      "step": 29148
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6068239808082581,
      "learning_rate": 0.0005766149961340442,
      "loss": 2.2693,
      "step": 29149
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6270835399627686,
      "learning_rate": 0.0005766134127710679,
      "loss": 2.3793,
      "step": 29150
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.5867936611175537,
      "learning_rate": 0.0005766118293566638,
      "loss": 2.3801,
      "step": 29151
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6550157070159912,
      "learning_rate": 0.0005766102458908325,
      "loss": 2.703,
      "step": 29152
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6161425709724426,
      "learning_rate": 0.0005766086623735742,
      "loss": 2.4458,
      "step": 29153
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6670656204223633,
      "learning_rate": 0.000576607078804889,
      "loss": 2.0559,
      "step": 29154
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6300875544548035,
      "learning_rate": 0.0005766054951847773,
      "loss": 2.1544,
      "step": 29155
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.5602298378944397,
      "learning_rate": 0.0005766039115132394,
      "loss": 2.2202,
      "step": 29156
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6428955793380737,
      "learning_rate": 0.0005766023277902758,
      "loss": 2.4025,
      "step": 29157
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6445236206054688,
      "learning_rate": 0.0005766007440158864,
      "loss": 2.3037,
      "step": 29158
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6399112939834595,
      "learning_rate": 0.0005765991601900718,
      "loss": 2.2554,
      "step": 29159
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6367592215538025,
      "learning_rate": 0.0005765975763128322,
      "loss": 2.4334,
      "step": 29160
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6335732340812683,
      "learning_rate": 0.0005765959923841678,
      "loss": 2.2188,
      "step": 29161
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.7070369124412537,
      "learning_rate": 0.000576594408404079,
      "loss": 2.523,
      "step": 29162
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6513569951057434,
      "learning_rate": 0.0005765928243725662,
      "loss": 2.3487,
      "step": 29163
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6322657465934753,
      "learning_rate": 0.0005765912402896294,
      "loss": 2.2719,
      "step": 29164
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6214627623558044,
      "learning_rate": 0.0005765896561552692,
      "loss": 2.1644,
      "step": 29165
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6400529742240906,
      "learning_rate": 0.0005765880719694858,
      "loss": 2.3424,
      "step": 29166
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6571546792984009,
      "learning_rate": 0.0005765864877322794,
      "loss": 2.2848,
      "step": 29167
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6629324555397034,
      "learning_rate": 0.0005765849034436503,
      "loss": 2.4605,
      "step": 29168
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6024477481842041,
      "learning_rate": 0.0005765833191035989,
      "loss": 2.1962,
      "step": 29169
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.7883132100105286,
      "learning_rate": 0.0005765817347121254,
      "loss": 2.2449,
      "step": 29170
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6376121044158936,
      "learning_rate": 0.0005765801502692301,
      "loss": 2.2096,
      "step": 29171
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.5934451818466187,
      "learning_rate": 0.0005765785657749135,
      "loss": 2.4864,
      "step": 29172
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.841719925403595,
      "learning_rate": 0.0005765769812291756,
      "loss": 2.382,
      "step": 29173
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6948018074035645,
      "learning_rate": 0.0005765753966320168,
      "loss": 2.2668,
      "step": 29174
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6138802766799927,
      "learning_rate": 0.0005765738119834373,
      "loss": 2.2485,
      "step": 29175
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6977214813232422,
      "learning_rate": 0.0005765722272834378,
      "loss": 2.2624,
      "step": 29176
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.7142877578735352,
      "learning_rate": 0.000576570642532018,
      "loss": 2.5149,
      "step": 29177
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6685702800750732,
      "learning_rate": 0.0005765690577291787,
      "loss": 2.1868,
      "step": 29178
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6307108402252197,
      "learning_rate": 0.0005765674728749199,
      "loss": 2.323,
      "step": 29179
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6561660766601562,
      "learning_rate": 0.0005765658879692421,
      "loss": 2.1514,
      "step": 29180
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6877186298370361,
      "learning_rate": 0.0005765643030121454,
      "loss": 2.1989,
      "step": 29181
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6650254726409912,
      "learning_rate": 0.0005765627180036302,
      "loss": 2.4574,
      "step": 29182
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.7345796823501587,
      "learning_rate": 0.0005765611329436967,
      "loss": 2.3805,
      "step": 29183
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6686243414878845,
      "learning_rate": 0.0005765595478323454,
      "loss": 2.3812,
      "step": 29184
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6439919471740723,
      "learning_rate": 0.0005765579626695763,
      "loss": 2.4911,
      "step": 29185
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.61788409948349,
      "learning_rate": 0.00057655637745539,
      "loss": 2.5258,
      "step": 29186
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.7273604273796082,
      "learning_rate": 0.0005765547921897865,
      "loss": 2.2849,
      "step": 29187
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.8146103620529175,
      "learning_rate": 0.0005765532068727664,
      "loss": 2.3219,
      "step": 29188
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.7855877876281738,
      "learning_rate": 0.0005765516215043297,
      "loss": 2.1546,
      "step": 29189
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.8667243123054504,
      "learning_rate": 0.000576550036084477,
      "loss": 2.16,
      "step": 29190
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.7626936435699463,
      "learning_rate": 0.0005765484506132083,
      "loss": 2.3257,
      "step": 29191
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6765891909599304,
      "learning_rate": 0.000576546865090524,
      "loss": 2.3094,
      "step": 29192
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6448924541473389,
      "learning_rate": 0.0005765452795164245,
      "loss": 2.4938,
      "step": 29193
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6443763971328735,
      "learning_rate": 0.0005765436938909099,
      "loss": 2.3226,
      "step": 29194
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.9608222246170044,
      "learning_rate": 0.0005765421082139807,
      "loss": 2.212,
      "step": 29195
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.67153000831604,
      "learning_rate": 0.0005765405224856371,
      "loss": 2.4136,
      "step": 29196
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.7822055220603943,
      "learning_rate": 0.0005765389367058793,
      "loss": 2.3364,
      "step": 29197
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.7228398323059082,
      "learning_rate": 0.0005765373508747078,
      "loss": 2.207,
      "step": 29198
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6353830695152283,
      "learning_rate": 0.0005765357649921228,
      "loss": 2.3728,
      "step": 29199
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.7340894341468811,
      "learning_rate": 0.0005765341790581245,
      "loss": 2.2152,
      "step": 29200
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.7442187070846558,
      "learning_rate": 0.0005765325930727134,
      "loss": 2.2169,
      "step": 29201
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.9351126551628113,
      "learning_rate": 0.0005765310070358896,
      "loss": 2.2146,
      "step": 29202
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6551125049591064,
      "learning_rate": 0.0005765294209476534,
      "loss": 2.2449,
      "step": 29203
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.648362398147583,
      "learning_rate": 0.0005765278348080051,
      "loss": 2.408,
      "step": 29204
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.7490688562393188,
      "learning_rate": 0.0005765262486169452,
      "loss": 2.272,
      "step": 29205
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6600948572158813,
      "learning_rate": 0.0005765246623744739,
      "loss": 2.3981,
      "step": 29206
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6628472208976746,
      "learning_rate": 0.0005765230760805913,
      "loss": 2.2935,
      "step": 29207
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6421636939048767,
      "learning_rate": 0.000576521489735298,
      "loss": 2.7365,
      "step": 29208
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.7626791596412659,
      "learning_rate": 0.000576519903338594,
      "loss": 2.3897,
      "step": 29209
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.7393081188201904,
      "learning_rate": 0.0005765183168904797,
      "loss": 2.4725,
      "step": 29210
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.7104976177215576,
      "learning_rate": 0.0005765167303909555,
      "loss": 1.989,
      "step": 29211
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6078600287437439,
      "learning_rate": 0.0005765151438400217,
      "loss": 2.0421,
      "step": 29212
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.7343937754631042,
      "learning_rate": 0.0005765135572376784,
      "loss": 2.2362,
      "step": 29213
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.7888069748878479,
      "learning_rate": 0.0005765119705839261,
      "loss": 2.2043,
      "step": 29214
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6241970658302307,
      "learning_rate": 0.0005765103838787648,
      "loss": 2.4202,
      "step": 29215
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.7099238038063049,
      "learning_rate": 0.0005765087971221953,
      "loss": 2.4127,
      "step": 29216
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.682198166847229,
      "learning_rate": 0.0005765072103142174,
      "loss": 2.213,
      "step": 29217
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6488996744155884,
      "learning_rate": 0.0005765056234548316,
      "loss": 2.3478,
      "step": 29218
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.7349332571029663,
      "learning_rate": 0.0005765040365440382,
      "loss": 2.3895,
      "step": 29219
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.7399359941482544,
      "learning_rate": 0.0005765024495818375,
      "loss": 2.5143,
      "step": 29220
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6602608561515808,
      "learning_rate": 0.0005765008625682298,
      "loss": 2.2637,
      "step": 29221
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6472809314727783,
      "learning_rate": 0.0005764992755032154,
      "loss": 2.424,
      "step": 29222
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6817535161972046,
      "learning_rate": 0.0005764976883867944,
      "loss": 2.2431,
      "step": 29223
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6759322881698608,
      "learning_rate": 0.0005764961012189674,
      "loss": 2.39,
      "step": 29224
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.8996555209159851,
      "learning_rate": 0.0005764945139997345,
      "loss": 2.393,
      "step": 29225
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6552823781967163,
      "learning_rate": 0.0005764929267290961,
      "loss": 2.3154,
      "step": 29226
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.5987188220024109,
      "learning_rate": 0.0005764913394070524,
      "loss": 2.2282,
      "step": 29227
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6218541264533997,
      "learning_rate": 0.0005764897520336037,
      "loss": 2.412,
      "step": 29228
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.66538405418396,
      "learning_rate": 0.0005764881646087505,
      "loss": 2.3297,
      "step": 29229
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6601483821868896,
      "learning_rate": 0.0005764865771324928,
      "loss": 2.1365,
      "step": 29230
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6598038077354431,
      "learning_rate": 0.0005764849896048311,
      "loss": 2.3776,
      "step": 29231
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6069675087928772,
      "learning_rate": 0.0005764834020257655,
      "loss": 2.3268,
      "step": 29232
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.698783814907074,
      "learning_rate": 0.0005764818143952965,
      "loss": 2.3687,
      "step": 29233
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6511617302894592,
      "learning_rate": 0.0005764802267134243,
      "loss": 2.217,
      "step": 29234
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6026568412780762,
      "learning_rate": 0.0005764786389801492,
      "loss": 2.3971,
      "step": 29235
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.663378894329071,
      "learning_rate": 0.0005764770511954716,
      "loss": 2.3797,
      "step": 29236
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.7207851409912109,
      "learning_rate": 0.0005764754633593916,
      "loss": 2.4182,
      "step": 29237
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.7642713189125061,
      "learning_rate": 0.0005764738754719096,
      "loss": 2.2264,
      "step": 29238
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6576052904129028,
      "learning_rate": 0.0005764722875330259,
      "loss": 2.2716,
      "step": 29239
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.7148841023445129,
      "learning_rate": 0.0005764706995427407,
      "loss": 2.2944,
      "step": 29240
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.9403054118156433,
      "learning_rate": 0.0005764691115010545,
      "loss": 2.1541,
      "step": 29241
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6687734723091125,
      "learning_rate": 0.0005764675234079675,
      "loss": 2.3715,
      "step": 29242
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.8562464714050293,
      "learning_rate": 0.0005764659352634798,
      "loss": 2.2327,
      "step": 29243
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.8396531343460083,
      "learning_rate": 0.000576464347067592,
      "loss": 2.2643,
      "step": 29244
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6275461316108704,
      "learning_rate": 0.0005764627588203041,
      "loss": 2.3464,
      "step": 29245
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.7065944075584412,
      "learning_rate": 0.0005764611705216167,
      "loss": 2.4144,
      "step": 29246
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.887771725654602,
      "learning_rate": 0.0005764595821715299,
      "loss": 2.3482,
      "step": 29247
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.7638875842094421,
      "learning_rate": 0.0005764579937700441,
      "loss": 2.1664,
      "step": 29248
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6897531747817993,
      "learning_rate": 0.0005764564053171595,
      "loss": 2.1004,
      "step": 29249
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.7822694778442383,
      "learning_rate": 0.0005764548168128765,
      "loss": 2.3841,
      "step": 29250
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6457822918891907,
      "learning_rate": 0.0005764532282571953,
      "loss": 2.3116,
      "step": 29251
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6404581069946289,
      "learning_rate": 0.0005764516396501161,
      "loss": 2.0764,
      "step": 29252
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6404291391372681,
      "learning_rate": 0.0005764500509916395,
      "loss": 2.3673,
      "step": 29253
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.7170881628990173,
      "learning_rate": 0.0005764484622817655,
      "loss": 2.6033,
      "step": 29254
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.5981370210647583,
      "learning_rate": 0.0005764468735204945,
      "loss": 2.1569,
      "step": 29255
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6867061853408813,
      "learning_rate": 0.000576445284707827,
      "loss": 2.3242,
      "step": 29256
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6481103897094727,
      "learning_rate": 0.0005764436958437629,
      "loss": 2.1851,
      "step": 29257
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6791784167289734,
      "learning_rate": 0.0005764421069283027,
      "loss": 2.2224,
      "step": 29258
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6595406532287598,
      "learning_rate": 0.0005764405179614468,
      "loss": 2.4572,
      "step": 29259
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6427028775215149,
      "learning_rate": 0.0005764389289431953,
      "loss": 2.4511,
      "step": 29260
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6026971936225891,
      "learning_rate": 0.0005764373398735486,
      "loss": 2.113,
      "step": 29261
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6291519999504089,
      "learning_rate": 0.0005764357507525071,
      "loss": 2.2231,
      "step": 29262
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6155453324317932,
      "learning_rate": 0.0005764341615800708,
      "loss": 2.4311,
      "step": 29263
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6222410798072815,
      "learning_rate": 0.0005764325723562403,
      "loss": 2.1887,
      "step": 29264
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6333084106445312,
      "learning_rate": 0.0005764309830810157,
      "loss": 2.2545,
      "step": 29265
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6089112162590027,
      "learning_rate": 0.0005764293937543974,
      "loss": 2.2246,
      "step": 29266
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6898400187492371,
      "learning_rate": 0.0005764278043763857,
      "loss": 2.3136,
      "step": 29267
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.7738808989524841,
      "learning_rate": 0.0005764262149469808,
      "loss": 2.2984,
      "step": 29268
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.5903893709182739,
      "learning_rate": 0.000576424625466183,
      "loss": 2.3051,
      "step": 29269
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.680462658405304,
      "learning_rate": 0.0005764230359339927,
      "loss": 2.3637,
      "step": 29270
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.8074467778205872,
      "learning_rate": 0.0005764214463504101,
      "loss": 2.2915,
      "step": 29271
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.63178551197052,
      "learning_rate": 0.0005764198567154356,
      "loss": 2.3531,
      "step": 29272
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6387341022491455,
      "learning_rate": 0.0005764182670290694,
      "loss": 2.3374,
      "step": 29273
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6291741132736206,
      "learning_rate": 0.0005764166772913119,
      "loss": 2.1424,
      "step": 29274
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.7697428464889526,
      "learning_rate": 0.0005764150875021631,
      "loss": 2.161,
      "step": 29275
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6877894401550293,
      "learning_rate": 0.0005764134976616237,
      "loss": 2.5478,
      "step": 29276
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.589533269405365,
      "learning_rate": 0.0005764119077696938,
      "loss": 2.3968,
      "step": 29277
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.8061283826828003,
      "learning_rate": 0.0005764103178263737,
      "loss": 2.5274,
      "step": 29278
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6552085280418396,
      "learning_rate": 0.0005764087278316637,
      "loss": 2.366,
      "step": 29279
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.694032609462738,
      "learning_rate": 0.000576407137785564,
      "loss": 2.136,
      "step": 29280
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6230174899101257,
      "learning_rate": 0.0005764055476880751,
      "loss": 2.3145,
      "step": 29281
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6486471891403198,
      "learning_rate": 0.0005764039575391972,
      "loss": 2.3979,
      "step": 29282
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.7385129928588867,
      "learning_rate": 0.0005764023673389305,
      "loss": 2.428,
      "step": 29283
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6631051301956177,
      "learning_rate": 0.0005764007770872755,
      "loss": 2.2294,
      "step": 29284
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6335906386375427,
      "learning_rate": 0.0005763991867842324,
      "loss": 2.3498,
      "step": 29285
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6628738641738892,
      "learning_rate": 0.0005763975964298012,
      "loss": 2.4632,
      "step": 29286
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6648592352867126,
      "learning_rate": 0.0005763960060239828,
      "loss": 2.3385,
      "step": 29287
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.643492579460144,
      "learning_rate": 0.000576394415566777,
      "loss": 2.4346,
      "step": 29288
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6452503204345703,
      "learning_rate": 0.0005763928250581843,
      "loss": 2.3197,
      "step": 29289
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6879420876502991,
      "learning_rate": 0.0005763912344982049,
      "loss": 2.6854,
      "step": 29290
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6321626901626587,
      "learning_rate": 0.0005763896438868392,
      "loss": 2.3949,
      "step": 29291
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6421700119972229,
      "learning_rate": 0.0005763880532240875,
      "loss": 2.2477,
      "step": 29292
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6387452483177185,
      "learning_rate": 0.00057638646250995,
      "loss": 2.4027,
      "step": 29293
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6597832441329956,
      "learning_rate": 0.000576384871744427,
      "loss": 2.2997,
      "step": 29294
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.70590740442276,
      "learning_rate": 0.0005763832809275189,
      "loss": 2.4527,
      "step": 29295
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6206006407737732,
      "learning_rate": 0.0005763816900592258,
      "loss": 2.3096,
      "step": 29296
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.618890106678009,
      "learning_rate": 0.0005763800991395483,
      "loss": 2.3956,
      "step": 29297
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.669674813747406,
      "learning_rate": 0.0005763785081684865,
      "loss": 2.318,
      "step": 29298
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6538531184196472,
      "learning_rate": 0.0005763769171460406,
      "loss": 2.2976,
      "step": 29299
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6223263740539551,
      "learning_rate": 0.0005763753260722111,
      "loss": 2.1224,
      "step": 29300
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.713692307472229,
      "learning_rate": 0.0005763737349469981,
      "loss": 2.2461,
      "step": 29301
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6022208333015442,
      "learning_rate": 0.0005763721437704022,
      "loss": 2.3644,
      "step": 29302
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6513155102729797,
      "learning_rate": 0.0005763705525424234,
      "loss": 2.3501,
      "step": 29303
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.621561586856842,
      "learning_rate": 0.000576368961263062,
      "loss": 2.3923,
      "step": 29304
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6578028798103333,
      "learning_rate": 0.0005763673699323185,
      "loss": 2.1992,
      "step": 29305
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.7808408737182617,
      "learning_rate": 0.0005763657785501931,
      "loss": 2.5478,
      "step": 29306
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.7413775324821472,
      "learning_rate": 0.0005763641871166861,
      "loss": 2.2982,
      "step": 29307
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6638205051422119,
      "learning_rate": 0.0005763625956317976,
      "loss": 2.2607,
      "step": 29308
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6512035727500916,
      "learning_rate": 0.0005763610040955283,
      "loss": 2.1287,
      "step": 29309
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.7394565939903259,
      "learning_rate": 0.0005763594125078781,
      "loss": 2.3298,
      "step": 29310
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6749569773674011,
      "learning_rate": 0.0005763578208688476,
      "loss": 2.3416,
      "step": 29311
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6584434509277344,
      "learning_rate": 0.000576356229178437,
      "loss": 2.2187,
      "step": 29312
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6411421895027161,
      "learning_rate": 0.0005763546374366465,
      "loss": 2.443,
      "step": 29313
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6347505450248718,
      "learning_rate": 0.0005763530456434763,
      "loss": 2.3167,
      "step": 29314
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6444552540779114,
      "learning_rate": 0.000576351453798927,
      "loss": 2.3404,
      "step": 29315
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.7419344782829285,
      "learning_rate": 0.0005763498619029988,
      "loss": 2.5454,
      "step": 29316
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6591961979866028,
      "learning_rate": 0.0005763482699556919,
      "loss": 2.208,
      "step": 29317
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6444370150566101,
      "learning_rate": 0.0005763466779570065,
      "loss": 2.547,
      "step": 29318
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6730200052261353,
      "learning_rate": 0.0005763450859069433,
      "loss": 2.4138,
      "step": 29319
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6396892070770264,
      "learning_rate": 0.0005763434938055021,
      "loss": 2.3744,
      "step": 29320
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6422283053398132,
      "learning_rate": 0.0005763419016526836,
      "loss": 2.2785,
      "step": 29321
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6984236836433411,
      "learning_rate": 0.0005763403094484878,
      "loss": 2.3772,
      "step": 29322
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.5885054469108582,
      "learning_rate": 0.0005763387171929152,
      "loss": 2.3181,
      "step": 29323
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6383662819862366,
      "learning_rate": 0.0005763371248859659,
      "loss": 2.3932,
      "step": 29324
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.691279947757721,
      "learning_rate": 0.0005763355325276406,
      "loss": 2.386,
      "step": 29325
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6387674808502197,
      "learning_rate": 0.000576333940117939,
      "loss": 2.3154,
      "step": 29326
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6159008741378784,
      "learning_rate": 0.0005763323476568618,
      "loss": 2.2471,
      "step": 29327
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6160016655921936,
      "learning_rate": 0.0005763307551444092,
      "loss": 2.1793,
      "step": 29328
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6847724318504333,
      "learning_rate": 0.0005763291625805815,
      "loss": 2.2464,
      "step": 29329
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6142379641532898,
      "learning_rate": 0.000576327569965379,
      "loss": 2.1831,
      "step": 29330
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6470287442207336,
      "learning_rate": 0.0005763259772988019,
      "loss": 2.293,
      "step": 29331
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6690388917922974,
      "learning_rate": 0.0005763243845808507,
      "loss": 2.2474,
      "step": 29332
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6818495392799377,
      "learning_rate": 0.0005763227918115256,
      "loss": 2.3183,
      "step": 29333
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6711164116859436,
      "learning_rate": 0.0005763211989908268,
      "loss": 2.6079,
      "step": 29334
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.5736005902290344,
      "learning_rate": 0.0005763196061187546,
      "loss": 2.2617,
      "step": 29335
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6289584636688232,
      "learning_rate": 0.0005763180131953096,
      "loss": 2.3162,
      "step": 29336
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6398501396179199,
      "learning_rate": 0.0005763164202204918,
      "loss": 2.4537,
      "step": 29337
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.7211530208587646,
      "learning_rate": 0.0005763148271943014,
      "loss": 2.4798,
      "step": 29338
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.7020646929740906,
      "learning_rate": 0.0005763132341167389,
      "loss": 2.4056,
      "step": 29339
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6519070267677307,
      "learning_rate": 0.0005763116409878046,
      "loss": 2.2694,
      "step": 29340
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6025976538658142,
      "learning_rate": 0.0005763100478074989,
      "loss": 2.1911,
      "step": 29341
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.619577169418335,
      "learning_rate": 0.0005763084545758217,
      "loss": 2.3254,
      "step": 29342
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6318924427032471,
      "learning_rate": 0.0005763068612927737,
      "loss": 2.52,
      "step": 29343
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.661697506904602,
      "learning_rate": 0.0005763052679583549,
      "loss": 2.277,
      "step": 29344
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.605919361114502,
      "learning_rate": 0.0005763036745725659,
      "loss": 2.6549,
      "step": 29345
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6894167065620422,
      "learning_rate": 0.0005763020811354067,
      "loss": 2.2952,
      "step": 29346
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6002272367477417,
      "learning_rate": 0.0005763004876468778,
      "loss": 2.0857,
      "step": 29347
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6140027046203613,
      "learning_rate": 0.0005762988941069794,
      "loss": 2.1943,
      "step": 29348
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6381782293319702,
      "learning_rate": 0.0005762973005157117,
      "loss": 2.2706,
      "step": 29349
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6559239029884338,
      "learning_rate": 0.0005762957068730754,
      "loss": 2.1895,
      "step": 29350
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6400637626647949,
      "learning_rate": 0.0005762941131790703,
      "loss": 2.4836,
      "step": 29351
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.7253413200378418,
      "learning_rate": 0.000576292519433697,
      "loss": 2.3977,
      "step": 29352
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6280955076217651,
      "learning_rate": 0.0005762909256369557,
      "loss": 2.2459,
      "step": 29353
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.712674617767334,
      "learning_rate": 0.0005762893317888466,
      "loss": 2.2531,
      "step": 29354
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6489461064338684,
      "learning_rate": 0.0005762877378893702,
      "loss": 2.1606,
      "step": 29355
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6420145034790039,
      "learning_rate": 0.0005762861439385266,
      "loss": 2.1641,
      "step": 29356
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.8874557018280029,
      "learning_rate": 0.0005762845499363163,
      "loss": 2.3049,
      "step": 29357
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6560174226760864,
      "learning_rate": 0.0005762829558827394,
      "loss": 2.197,
      "step": 29358
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6122981905937195,
      "learning_rate": 0.0005762813617777963,
      "loss": 2.2546,
      "step": 29359
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6268851161003113,
      "learning_rate": 0.0005762797676214873,
      "loss": 2.3257,
      "step": 29360
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6436678171157837,
      "learning_rate": 0.0005762781734138127,
      "loss": 2.4046,
      "step": 29361
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6963983774185181,
      "learning_rate": 0.0005762765791547728,
      "loss": 2.2987,
      "step": 29362
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6168307662010193,
      "learning_rate": 0.0005762749848443678,
      "loss": 2.3456,
      "step": 29363
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6926388740539551,
      "learning_rate": 0.0005762733904825981,
      "loss": 2.4307,
      "step": 29364
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6929655075073242,
      "learning_rate": 0.0005762717960694639,
      "loss": 2.5739,
      "step": 29365
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6213747262954712,
      "learning_rate": 0.0005762702016049656,
      "loss": 2.3087,
      "step": 29366
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.5792533159255981,
      "learning_rate": 0.0005762686070891035,
      "loss": 2.4278,
      "step": 29367
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6692723035812378,
      "learning_rate": 0.0005762670125218777,
      "loss": 2.5046,
      "step": 29368
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6234319806098938,
      "learning_rate": 0.0005762654179032889,
      "loss": 2.2846,
      "step": 29369
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.5808460116386414,
      "learning_rate": 0.0005762638232333369,
      "loss": 2.4335,
      "step": 29370
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6676647663116455,
      "learning_rate": 0.0005762622285120224,
      "loss": 2.1916,
      "step": 29371
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.7624301910400391,
      "learning_rate": 0.0005762606337393454,
      "loss": 2.3503,
      "step": 29372
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.610937237739563,
      "learning_rate": 0.0005762590389153065,
      "loss": 2.2242,
      "step": 29373
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6570045351982117,
      "learning_rate": 0.0005762574440399057,
      "loss": 2.4248,
      "step": 29374
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.685113251209259,
      "learning_rate": 0.0005762558491131434,
      "loss": 2.1867,
      "step": 29375
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6080610156059265,
      "learning_rate": 0.00057625425413502,
      "loss": 2.4567,
      "step": 29376
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6265543103218079,
      "learning_rate": 0.0005762526591055357,
      "loss": 2.3238,
      "step": 29377
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6799845099449158,
      "learning_rate": 0.0005762510640246909,
      "loss": 2.358,
      "step": 29378
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6145320534706116,
      "learning_rate": 0.0005762494688924857,
      "loss": 2.5009,
      "step": 29379
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6388925313949585,
      "learning_rate": 0.0005762478737089205,
      "loss": 2.2311,
      "step": 29380
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.9234294891357422,
      "learning_rate": 0.0005762462784739957,
      "loss": 2.2968,
      "step": 29381
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.621818482875824,
      "learning_rate": 0.0005762446831877114,
      "loss": 2.4544,
      "step": 29382
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6837759613990784,
      "learning_rate": 0.0005762430878500681,
      "loss": 2.1388,
      "step": 29383
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6132795810699463,
      "learning_rate": 0.000576241492461066,
      "loss": 2.4441,
      "step": 29384
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6843236684799194,
      "learning_rate": 0.0005762398970207053,
      "loss": 2.4701,
      "step": 29385
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6230437755584717,
      "learning_rate": 0.0005762383015289864,
      "loss": 2.2876,
      "step": 29386
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6697450876235962,
      "learning_rate": 0.0005762367059859096,
      "loss": 2.2209,
      "step": 29387
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6958906650543213,
      "learning_rate": 0.0005762351103914752,
      "loss": 2.298,
      "step": 29388
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.7074089050292969,
      "learning_rate": 0.0005762335147456836,
      "loss": 2.2883,
      "step": 29389
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6242141127586365,
      "learning_rate": 0.0005762319190485347,
      "loss": 2.2353,
      "step": 29390
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6387965679168701,
      "learning_rate": 0.0005762303233000293,
      "loss": 2.2407,
      "step": 29391
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6682329177856445,
      "learning_rate": 0.0005762287275001674,
      "loss": 2.1912,
      "step": 29392
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.7529619336128235,
      "learning_rate": 0.0005762271316489492,
      "loss": 2.3587,
      "step": 29393
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6885293126106262,
      "learning_rate": 0.0005762255357463755,
      "loss": 2.3619,
      "step": 29394
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6512991786003113,
      "learning_rate": 0.000576223939792446,
      "loss": 2.4325,
      "step": 29395
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6405097246170044,
      "learning_rate": 0.0005762223437871613,
      "loss": 2.1914,
      "step": 29396
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.7038384675979614,
      "learning_rate": 0.0005762207477305217,
      "loss": 2.288,
      "step": 29397
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6359337568283081,
      "learning_rate": 0.0005762191516225275,
      "loss": 2.2665,
      "step": 29398
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6725749969482422,
      "learning_rate": 0.0005762175554631788,
      "loss": 2.3198,
      "step": 29399
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6596242785453796,
      "learning_rate": 0.0005762159592524761,
      "loss": 2.2296,
      "step": 29400
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6665075421333313,
      "learning_rate": 0.0005762143629904198,
      "loss": 2.303,
      "step": 29401
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6694417595863342,
      "learning_rate": 0.0005762127666770098,
      "loss": 2.5837,
      "step": 29402
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.8304759860038757,
      "learning_rate": 0.0005762111703122467,
      "loss": 2.3506,
      "step": 29403
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6744779944419861,
      "learning_rate": 0.0005762095738961308,
      "loss": 2.351,
      "step": 29404
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.5940569043159485,
      "learning_rate": 0.0005762079774286623,
      "loss": 2.2246,
      "step": 29405
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.7281913757324219,
      "learning_rate": 0.0005762063809098414,
      "loss": 2.4301,
      "step": 29406
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6437307000160217,
      "learning_rate": 0.0005762047843396687,
      "loss": 2.3041,
      "step": 29407
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6594735383987427,
      "learning_rate": 0.0005762031877181443,
      "loss": 2.0104,
      "step": 29408
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6180585622787476,
      "learning_rate": 0.0005762015910452684,
      "loss": 2.4291,
      "step": 29409
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.7170729637145996,
      "learning_rate": 0.0005761999943210415,
      "loss": 2.4987,
      "step": 29410
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.7301099300384521,
      "learning_rate": 0.0005761983975454638,
      "loss": 2.1873,
      "step": 29411
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6388943195343018,
      "learning_rate": 0.0005761968007185356,
      "loss": 2.2413,
      "step": 29412
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6899154186248779,
      "learning_rate": 0.0005761952038402571,
      "loss": 2.2874,
      "step": 29413
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6613898873329163,
      "learning_rate": 0.0005761936069106289,
      "loss": 2.3455,
      "step": 29414
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6032116413116455,
      "learning_rate": 0.0005761920099296509,
      "loss": 2.1585,
      "step": 29415
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6332996487617493,
      "learning_rate": 0.0005761904128973236,
      "loss": 2.2275,
      "step": 29416
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6226364374160767,
      "learning_rate": 0.0005761888158136474,
      "loss": 2.3396,
      "step": 29417
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6200070977210999,
      "learning_rate": 0.0005761872186786224,
      "loss": 2.1851,
      "step": 29418
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6169576644897461,
      "learning_rate": 0.000576185621492249,
      "loss": 2.3913,
      "step": 29419
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6813810467720032,
      "learning_rate": 0.0005761840242545275,
      "loss": 2.2698,
      "step": 29420
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6448442339897156,
      "learning_rate": 0.0005761824269654583,
      "loss": 2.2601,
      "step": 29421
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6634610295295715,
      "learning_rate": 0.0005761808296250413,
      "loss": 2.5656,
      "step": 29422
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6196497678756714,
      "learning_rate": 0.0005761792322332773,
      "loss": 2.2931,
      "step": 29423
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6330896019935608,
      "learning_rate": 0.0005761776347901663,
      "loss": 2.3125,
      "step": 29424
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.679835319519043,
      "learning_rate": 0.0005761760372957087,
      "loss": 2.2306,
      "step": 29425
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6063193678855896,
      "learning_rate": 0.0005761744397499047,
      "loss": 2.4277,
      "step": 29426
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.629679262638092,
      "learning_rate": 0.0005761728421527547,
      "loss": 2.1381,
      "step": 29427
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6188501119613647,
      "learning_rate": 0.0005761712445042588,
      "loss": 2.556,
      "step": 29428
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.660496711730957,
      "learning_rate": 0.0005761696468044176,
      "loss": 2.2525,
      "step": 29429
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6992001533508301,
      "learning_rate": 0.0005761680490532313,
      "loss": 2.3167,
      "step": 29430
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6032506823539734,
      "learning_rate": 0.0005761664512507001,
      "loss": 2.1424,
      "step": 29431
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6024749279022217,
      "learning_rate": 0.0005761648533968243,
      "loss": 2.0598,
      "step": 29432
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.7224184274673462,
      "learning_rate": 0.0005761632554916043,
      "loss": 2.3355,
      "step": 29433
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.7291755080223083,
      "learning_rate": 0.0005761616575350403,
      "loss": 2.3876,
      "step": 29434
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6491556167602539,
      "learning_rate": 0.0005761600595271326,
      "loss": 2.2542,
      "step": 29435
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6314032077789307,
      "learning_rate": 0.0005761584614678815,
      "loss": 2.1187,
      "step": 29436
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6005769968032837,
      "learning_rate": 0.0005761568633572874,
      "loss": 2.1224,
      "step": 29437
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6741588115692139,
      "learning_rate": 0.0005761552651953507,
      "loss": 2.5646,
      "step": 29438
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.703361451625824,
      "learning_rate": 0.0005761536669820713,
      "loss": 2.2949,
      "step": 29439
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6200869679450989,
      "learning_rate": 0.0005761520687174497,
      "loss": 2.2838,
      "step": 29440
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6120800971984863,
      "learning_rate": 0.0005761504704014863,
      "loss": 2.3185,
      "step": 29441
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.7594819068908691,
      "learning_rate": 0.0005761488720341813,
      "loss": 2.4664,
      "step": 29442
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.659696638584137,
      "learning_rate": 0.000576147273615535,
      "loss": 2.4365,
      "step": 29443
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.7708103060722351,
      "learning_rate": 0.0005761456751455478,
      "loss": 2.3908,
      "step": 29444
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.5931903719902039,
      "learning_rate": 0.0005761440766242198,
      "loss": 2.305,
      "step": 29445
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6232693791389465,
      "learning_rate": 0.0005761424780515514,
      "loss": 2.0681,
      "step": 29446
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6212260127067566,
      "learning_rate": 0.000576140879427543,
      "loss": 2.407,
      "step": 29447
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6715528964996338,
      "learning_rate": 0.0005761392807521946,
      "loss": 2.1407,
      "step": 29448
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6512277126312256,
      "learning_rate": 0.0005761376820255069,
      "loss": 2.3004,
      "step": 29449
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6744031310081482,
      "learning_rate": 0.0005761360832474799,
      "loss": 2.384,
      "step": 29450
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6190856099128723,
      "learning_rate": 0.000576134484418114,
      "loss": 2.4196,
      "step": 29451
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.5963948965072632,
      "learning_rate": 0.0005761328855374095,
      "loss": 2.1245,
      "step": 29452
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6698551177978516,
      "learning_rate": 0.0005761312866053667,
      "loss": 2.5214,
      "step": 29453
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.5914215445518494,
      "learning_rate": 0.0005761296876219858,
      "loss": 2.3108,
      "step": 29454
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.7431963086128235,
      "learning_rate": 0.0005761280885872673,
      "loss": 2.2835,
      "step": 29455
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.7831793427467346,
      "learning_rate": 0.0005761264895012112,
      "loss": 2.3462,
      "step": 29456
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.5978549122810364,
      "learning_rate": 0.0005761248903638182,
      "loss": 2.4005,
      "step": 29457
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6233375668525696,
      "learning_rate": 0.0005761232911750883,
      "loss": 2.4336,
      "step": 29458
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6415920257568359,
      "learning_rate": 0.0005761216919350217,
      "loss": 2.372,
      "step": 29459
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6749498248100281,
      "learning_rate": 0.0005761200926436191,
      "loss": 2.2691,
      "step": 29460
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6686205863952637,
      "learning_rate": 0.0005761184933008804,
      "loss": 2.5978,
      "step": 29461
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6219525337219238,
      "learning_rate": 0.0005761168939068061,
      "loss": 2.0691,
      "step": 29462
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6017541885375977,
      "learning_rate": 0.0005761152944613964,
      "loss": 2.1262,
      "step": 29463
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6391639709472656,
      "learning_rate": 0.0005761136949646519,
      "loss": 2.461,
      "step": 29464
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6616321206092834,
      "learning_rate": 0.0005761120954165724,
      "loss": 2.2779,
      "step": 29465
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.664486289024353,
      "learning_rate": 0.0005761104958171585,
      "loss": 2.486,
      "step": 29466
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6402473449707031,
      "learning_rate": 0.0005761088961664105,
      "loss": 2.5668,
      "step": 29467
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.599353015422821,
      "learning_rate": 0.0005761072964643286,
      "loss": 2.163,
      "step": 29468
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.5858947038650513,
      "learning_rate": 0.0005761056967109132,
      "loss": 2.2638,
      "step": 29469
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6401993036270142,
      "learning_rate": 0.0005761040969061645,
      "loss": 2.2613,
      "step": 29470
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6574158668518066,
      "learning_rate": 0.0005761024970500828,
      "loss": 2.3742,
      "step": 29471
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.5977338552474976,
      "learning_rate": 0.0005761008971426685,
      "loss": 2.2093,
      "step": 29472
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6562843322753906,
      "learning_rate": 0.0005760992971839217,
      "loss": 2.3503,
      "step": 29473
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6377376317977905,
      "learning_rate": 0.000576097697173843,
      "loss": 2.2019,
      "step": 29474
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6664105653762817,
      "learning_rate": 0.0005760960971124325,
      "loss": 2.3485,
      "step": 29475
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.621412456035614,
      "learning_rate": 0.0005760944969996905,
      "loss": 2.247,
      "step": 29476
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6393666863441467,
      "learning_rate": 0.0005760928968356172,
      "loss": 2.4862,
      "step": 29477
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.7095544934272766,
      "learning_rate": 0.0005760912966202131,
      "loss": 2.4539,
      "step": 29478
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6090511679649353,
      "learning_rate": 0.0005760896963534786,
      "loss": 2.2613,
      "step": 29479
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6361498236656189,
      "learning_rate": 0.0005760880960354136,
      "loss": 2.4622,
      "step": 29480
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6067290306091309,
      "learning_rate": 0.0005760864956660186,
      "loss": 2.338,
      "step": 29481
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.5969593524932861,
      "learning_rate": 0.000576084895245294,
      "loss": 2.2403,
      "step": 29482
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6754380464553833,
      "learning_rate": 0.00057608329477324,
      "loss": 2.1727,
      "step": 29483
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6204646825790405,
      "learning_rate": 0.0005760816942498568,
      "loss": 2.1856,
      "step": 29484
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6050822734832764,
      "learning_rate": 0.0005760800936751449,
      "loss": 2.1925,
      "step": 29485
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6917590498924255,
      "learning_rate": 0.0005760784930491045,
      "loss": 2.5367,
      "step": 29486
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.7549626231193542,
      "learning_rate": 0.0005760768923717359,
      "loss": 2.3532,
      "step": 29487
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.7041736841201782,
      "learning_rate": 0.0005760752916430394,
      "loss": 2.1598,
      "step": 29488
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6210139989852905,
      "learning_rate": 0.0005760736908630152,
      "loss": 2.1576,
      "step": 29489
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.7196828126907349,
      "learning_rate": 0.0005760720900316638,
      "loss": 2.3973,
      "step": 29490
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6303645968437195,
      "learning_rate": 0.0005760704891489853,
      "loss": 2.2332,
      "step": 29491
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6955553293228149,
      "learning_rate": 0.0005760688882149802,
      "loss": 2.4454,
      "step": 29492
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6269170045852661,
      "learning_rate": 0.0005760672872296486,
      "loss": 2.2067,
      "step": 29493
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6185799241065979,
      "learning_rate": 0.0005760656861929908,
      "loss": 2.5492,
      "step": 29494
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.5841174125671387,
      "learning_rate": 0.0005760640851050073,
      "loss": 2.0919,
      "step": 29495
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6708308458328247,
      "learning_rate": 0.0005760624839656983,
      "loss": 2.4188,
      "step": 29496
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.604577362537384,
      "learning_rate": 0.000576060882775064,
      "loss": 2.0144,
      "step": 29497
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6413286328315735,
      "learning_rate": 0.0005760592815331047,
      "loss": 2.3972,
      "step": 29498
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.7056108117103577,
      "learning_rate": 0.0005760576802398209,
      "loss": 2.369,
      "step": 29499
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.7233229279518127,
      "learning_rate": 0.0005760560788952128,
      "loss": 2.5656,
      "step": 29500
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6571568846702576,
      "learning_rate": 0.0005760544774992806,
      "loss": 2.388,
      "step": 29501
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6061375737190247,
      "learning_rate": 0.0005760528760520246,
      "loss": 2.5735,
      "step": 29502
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6114950180053711,
      "learning_rate": 0.0005760512745534451,
      "loss": 2.3422,
      "step": 29503
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6883352994918823,
      "learning_rate": 0.0005760496730035426,
      "loss": 2.4051,
      "step": 29504
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6008345484733582,
      "learning_rate": 0.0005760480714023172,
      "loss": 2.426,
      "step": 29505
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6358914971351624,
      "learning_rate": 0.0005760464697497693,
      "loss": 2.306,
      "step": 29506
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.7139653563499451,
      "learning_rate": 0.0005760448680458992,
      "loss": 2.4079,
      "step": 29507
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6296643018722534,
      "learning_rate": 0.0005760432662907071,
      "loss": 2.4674,
      "step": 29508
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6036222577095032,
      "learning_rate": 0.0005760416644841933,
      "loss": 2.3202,
      "step": 29509
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.7579277753829956,
      "learning_rate": 0.0005760400626263582,
      "loss": 2.4474,
      "step": 29510
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.5950761437416077,
      "learning_rate": 0.0005760384607172019,
      "loss": 2.3652,
      "step": 29511
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6792060136795044,
      "learning_rate": 0.000576036858756725,
      "loss": 2.4936,
      "step": 29512
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.7047445774078369,
      "learning_rate": 0.0005760352567449275,
      "loss": 2.4028,
      "step": 29513
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.7135204672813416,
      "learning_rate": 0.00057603365468181,
      "loss": 2.1726,
      "step": 29514
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6262597441673279,
      "learning_rate": 0.0005760320525673724,
      "loss": 2.4935,
      "step": 29515
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.624233603477478,
      "learning_rate": 0.0005760304504016155,
      "loss": 2.2251,
      "step": 29516
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.7425118684768677,
      "learning_rate": 0.0005760288481845392,
      "loss": 2.1129,
      "step": 29517
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6353165507316589,
      "learning_rate": 0.0005760272459161439,
      "loss": 2.0378,
      "step": 29518
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6232986450195312,
      "learning_rate": 0.0005760256435964299,
      "loss": 2.4318,
      "step": 29519
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.7286052107810974,
      "learning_rate": 0.0005760240412253976,
      "loss": 2.3916,
      "step": 29520
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.632517397403717,
      "learning_rate": 0.0005760224388030473,
      "loss": 2.3804,
      "step": 29521
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.7649911642074585,
      "learning_rate": 0.0005760208363293791,
      "loss": 2.4505,
      "step": 29522
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6901332139968872,
      "learning_rate": 0.0005760192338043934,
      "loss": 2.3697,
      "step": 29523
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.670495867729187,
      "learning_rate": 0.0005760176312280906,
      "loss": 2.143,
      "step": 29524
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6223575472831726,
      "learning_rate": 0.0005760160286004708,
      "loss": 2.367,
      "step": 29525
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6051712036132812,
      "learning_rate": 0.0005760144259215345,
      "loss": 2.3618,
      "step": 29526
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6299101710319519,
      "learning_rate": 0.0005760128231912818,
      "loss": 2.2591,
      "step": 29527
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.7418399453163147,
      "learning_rate": 0.0005760112204097132,
      "loss": 2.2655,
      "step": 29528
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6358602046966553,
      "learning_rate": 0.000576009617576829,
      "loss": 2.2469,
      "step": 29529
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.8011335730552673,
      "learning_rate": 0.0005760080146926292,
      "loss": 2.2932,
      "step": 29530
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.65975022315979,
      "learning_rate": 0.0005760064117571144,
      "loss": 2.4734,
      "step": 29531
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.771517813205719,
      "learning_rate": 0.0005760048087702849,
      "loss": 2.3475,
      "step": 29532
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6934500336647034,
      "learning_rate": 0.0005760032057321407,
      "loss": 2.2065,
      "step": 29533
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6828629970550537,
      "learning_rate": 0.0005760016026426823,
      "loss": 2.36,
      "step": 29534
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6946513056755066,
      "learning_rate": 0.0005759999995019102,
      "loss": 2.3828,
      "step": 29535
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6325433254241943,
      "learning_rate": 0.0005759983963098244,
      "loss": 2.3246,
      "step": 29536
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6361360549926758,
      "learning_rate": 0.0005759967930664252,
      "loss": 2.208,
      "step": 29537
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6520164608955383,
      "learning_rate": 0.000575995189771713,
      "loss": 2.4182,
      "step": 29538
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.670138418674469,
      "learning_rate": 0.0005759935864256882,
      "loss": 2.4327,
      "step": 29539
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.7281248569488525,
      "learning_rate": 0.0005759919830283509,
      "loss": 2.5091,
      "step": 29540
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.7553063035011292,
      "learning_rate": 0.0005759903795797014,
      "loss": 2.3744,
      "step": 29541
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.691943347454071,
      "learning_rate": 0.0005759887760797402,
      "loss": 2.1964,
      "step": 29542
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6155949831008911,
      "learning_rate": 0.0005759871725284675,
      "loss": 2.2432,
      "step": 29543
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.626360297203064,
      "learning_rate": 0.0005759855689258834,
      "loss": 2.602,
      "step": 29544
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6667084693908691,
      "learning_rate": 0.0005759839652719886,
      "loss": 2.3767,
      "step": 29545
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.5976426005363464,
      "learning_rate": 0.000575982361566783,
      "loss": 2.46,
      "step": 29546
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6105793714523315,
      "learning_rate": 0.0005759807578102671,
      "loss": 2.3087,
      "step": 29547
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6663665771484375,
      "learning_rate": 0.0005759791540024412,
      "loss": 2.3783,
      "step": 29548
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.621798574924469,
      "learning_rate": 0.0005759775501433057,
      "loss": 2.3777,
      "step": 29549
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.7321014404296875,
      "learning_rate": 0.0005759759462328605,
      "loss": 2.2871,
      "step": 29550
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6643330454826355,
      "learning_rate": 0.0005759743422711062,
      "loss": 2.495,
      "step": 29551
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.5634965300559998,
      "learning_rate": 0.0005759727382580432,
      "loss": 2.3545,
      "step": 29552
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6496848464012146,
      "learning_rate": 0.0005759711341936717,
      "loss": 2.2253,
      "step": 29553
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.7039823532104492,
      "learning_rate": 0.0005759695300779918,
      "loss": 2.5464,
      "step": 29554
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.7153568863868713,
      "learning_rate": 0.0005759679259110039,
      "loss": 2.2497,
      "step": 29555
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6410485506057739,
      "learning_rate": 0.0005759663216927086,
      "loss": 2.0986,
      "step": 29556
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.7106240391731262,
      "learning_rate": 0.0005759647174231058,
      "loss": 2.2417,
      "step": 29557
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6670598983764648,
      "learning_rate": 0.000575963113102196,
      "loss": 2.4268,
      "step": 29558
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6557602882385254,
      "learning_rate": 0.0005759615087299794,
      "loss": 2.3264,
      "step": 29559
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6433734893798828,
      "learning_rate": 0.0005759599043064563,
      "loss": 2.3162,
      "step": 29560
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.7972868084907532,
      "learning_rate": 0.0005759582998316272,
      "loss": 2.1846,
      "step": 29561
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6246787905693054,
      "learning_rate": 0.0005759566953054922,
      "loss": 2.1946,
      "step": 29562
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6615707874298096,
      "learning_rate": 0.0005759550907280515,
      "loss": 2.1516,
      "step": 29563
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6808860301971436,
      "learning_rate": 0.0005759534860993057,
      "loss": 2.3293,
      "step": 29564
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6710101962089539,
      "learning_rate": 0.0005759518814192549,
      "loss": 2.4472,
      "step": 29565
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6433428525924683,
      "learning_rate": 0.0005759502766878994,
      "loss": 2.3558,
      "step": 29566
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.7128418684005737,
      "learning_rate": 0.0005759486719052395,
      "loss": 2.1566,
      "step": 29567
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.7246811389923096,
      "learning_rate": 0.0005759470670712756,
      "loss": 2.2875,
      "step": 29568
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.7103403210639954,
      "learning_rate": 0.0005759454621860079,
      "loss": 2.2547,
      "step": 29569
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.5983313322067261,
      "learning_rate": 0.0005759438572494367,
      "loss": 2.2575,
      "step": 29570
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.7756986021995544,
      "learning_rate": 0.0005759422522615624,
      "loss": 2.3197,
      "step": 29571
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6857500672340393,
      "learning_rate": 0.0005759406472223852,
      "loss": 2.2477,
      "step": 29572
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6372089982032776,
      "learning_rate": 0.0005759390421319054,
      "loss": 2.2997,
      "step": 29573
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6213577389717102,
      "learning_rate": 0.0005759374369901233,
      "loss": 2.1923,
      "step": 29574
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6337519288063049,
      "learning_rate": 0.0005759358317970393,
      "loss": 2.3951,
      "step": 29575
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.619418203830719,
      "learning_rate": 0.0005759342265526536,
      "loss": 2.3801,
      "step": 29576
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6222786903381348,
      "learning_rate": 0.0005759326212569666,
      "loss": 2.3158,
      "step": 29577
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6577648520469666,
      "learning_rate": 0.0005759310159099783,
      "loss": 2.3306,
      "step": 29578
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.7271621227264404,
      "learning_rate": 0.0005759294105116894,
      "loss": 2.258,
      "step": 29579
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6121007800102234,
      "learning_rate": 0.0005759278050620999,
      "loss": 2.2242,
      "step": 29580
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6968039870262146,
      "learning_rate": 0.0005759261995612103,
      "loss": 2.3433,
      "step": 29581
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.7245671153068542,
      "learning_rate": 0.0005759245940090207,
      "loss": 2.2229,
      "step": 29582
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.7131038904190063,
      "learning_rate": 0.0005759229884055317,
      "loss": 2.3875,
      "step": 29583
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.7411971688270569,
      "learning_rate": 0.0005759213827507432,
      "loss": 2.2322,
      "step": 29584
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6200737953186035,
      "learning_rate": 0.0005759197770446558,
      "loss": 2.3659,
      "step": 29585
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6752151846885681,
      "learning_rate": 0.0005759181712872698,
      "loss": 2.5536,
      "step": 29586
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.642586886882782,
      "learning_rate": 0.0005759165654785852,
      "loss": 2.2204,
      "step": 29587
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6657537817955017,
      "learning_rate": 0.0005759149596186026,
      "loss": 2.5035,
      "step": 29588
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.7244916558265686,
      "learning_rate": 0.0005759133537073223,
      "loss": 2.219,
      "step": 29589
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6385734677314758,
      "learning_rate": 0.0005759117477447444,
      "loss": 2.6094,
      "step": 29590
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.7025530934333801,
      "learning_rate": 0.0005759101417308693,
      "loss": 2.271,
      "step": 29591
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.8666951656341553,
      "learning_rate": 0.0005759085356656972,
      "loss": 2.2313,
      "step": 29592
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6959090232849121,
      "learning_rate": 0.0005759069295492287,
      "loss": 2.2637,
      "step": 29593
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.7772642970085144,
      "learning_rate": 0.0005759053233814637,
      "loss": 2.364,
      "step": 29594
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.8468689918518066,
      "learning_rate": 0.0005759037171624028,
      "loss": 2.2638,
      "step": 29595
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6806913614273071,
      "learning_rate": 0.0005759021108920462,
      "loss": 2.401,
      "step": 29596
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.8139169812202454,
      "learning_rate": 0.0005759005045703942,
      "loss": 2.2757,
      "step": 29597
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.9117382764816284,
      "learning_rate": 0.000575898898197447,
      "loss": 2.3837,
      "step": 29598
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6747374534606934,
      "learning_rate": 0.000575897291773205,
      "loss": 2.2584,
      "step": 29599
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.8915109634399414,
      "learning_rate": 0.0005758956852976685,
      "loss": 2.336,
      "step": 29600
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.8174502849578857,
      "learning_rate": 0.0005758940787708379,
      "loss": 2.3438,
      "step": 29601
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6515853404998779,
      "learning_rate": 0.0005758924721927133,
      "loss": 2.2534,
      "step": 29602
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6345339417457581,
      "learning_rate": 0.000575890865563295,
      "loss": 2.1013,
      "step": 29603
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.7982191443443298,
      "learning_rate": 0.0005758892588825836,
      "loss": 2.4696,
      "step": 29604
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.7697859406471252,
      "learning_rate": 0.000575887652150579,
      "loss": 2.4795,
      "step": 29605
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.7210440039634705,
      "learning_rate": 0.0005758860453672816,
      "loss": 2.5384,
      "step": 29606
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.7031500339508057,
      "learning_rate": 0.000575884438532692,
      "loss": 2.2078,
      "step": 29607
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.8531973361968994,
      "learning_rate": 0.0005758828316468101,
      "loss": 2.2479,
      "step": 29608
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6849308013916016,
      "learning_rate": 0.0005758812247096364,
      "loss": 2.2625,
      "step": 29609
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.585695207118988,
      "learning_rate": 0.0005758796177211713,
      "loss": 2.3273,
      "step": 29610
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.7367717027664185,
      "learning_rate": 0.0005758780106814148,
      "loss": 2.0263,
      "step": 29611
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.761197566986084,
      "learning_rate": 0.0005758764035903674,
      "loss": 2.4472,
      "step": 29612
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6314600706100464,
      "learning_rate": 0.0005758747964480295,
      "loss": 2.374,
      "step": 29613
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6141054034233093,
      "learning_rate": 0.0005758731892544012,
      "loss": 2.2973,
      "step": 29614
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.7265406250953674,
      "learning_rate": 0.0005758715820094829,
      "loss": 2.2602,
      "step": 29615
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6768630146980286,
      "learning_rate": 0.0005758699747132747,
      "loss": 2.235,
      "step": 29616
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.596333920955658,
      "learning_rate": 0.0005758683673657772,
      "loss": 2.3163,
      "step": 29617
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.639228105545044,
      "learning_rate": 0.0005758667599669906,
      "loss": 2.3898,
      "step": 29618
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.7007051706314087,
      "learning_rate": 0.0005758651525169151,
      "loss": 2.1547,
      "step": 29619
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6506069302558899,
      "learning_rate": 0.000575863545015551,
      "loss": 2.2364,
      "step": 29620
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6792998313903809,
      "learning_rate": 0.0005758619374628988,
      "loss": 2.2694,
      "step": 29621
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6666871905326843,
      "learning_rate": 0.0005758603298589586,
      "loss": 2.2929,
      "step": 29622
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6226899027824402,
      "learning_rate": 0.0005758587222037307,
      "loss": 2.2924,
      "step": 29623
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6429777145385742,
      "learning_rate": 0.0005758571144972155,
      "loss": 2.1764,
      "step": 29624
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.5988396406173706,
      "learning_rate": 0.0005758555067394132,
      "loss": 2.2469,
      "step": 29625
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6458725333213806,
      "learning_rate": 0.0005758538989303243,
      "loss": 2.5856,
      "step": 29626
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6387912034988403,
      "learning_rate": 0.0005758522910699488,
      "loss": 2.1035,
      "step": 29627
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6462841033935547,
      "learning_rate": 0.0005758506831582872,
      "loss": 2.2539,
      "step": 29628
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6731981635093689,
      "learning_rate": 0.0005758490751953398,
      "loss": 2.215,
      "step": 29629
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6715444326400757,
      "learning_rate": 0.0005758474671811068,
      "loss": 2.3462,
      "step": 29630
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.7712700366973877,
      "learning_rate": 0.0005758458591155886,
      "loss": 2.1797,
      "step": 29631
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6501625180244446,
      "learning_rate": 0.0005758442509987854,
      "loss": 2.4091,
      "step": 29632
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.7723162174224854,
      "learning_rate": 0.0005758426428306975,
      "loss": 2.4491,
      "step": 29633
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6928550004959106,
      "learning_rate": 0.0005758410346113253,
      "loss": 2.132,
      "step": 29634
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6591159701347351,
      "learning_rate": 0.0005758394263406691,
      "loss": 2.2214,
      "step": 29635
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.7020530104637146,
      "learning_rate": 0.0005758378180187291,
      "loss": 2.247,
      "step": 29636
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6440013647079468,
      "learning_rate": 0.0005758362096455056,
      "loss": 2.2271,
      "step": 29637
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6484421491622925,
      "learning_rate": 0.000575834601220999,
      "loss": 2.2862,
      "step": 29638
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6140764951705933,
      "learning_rate": 0.0005758329927452096,
      "loss": 2.3478,
      "step": 29639
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6078905463218689,
      "learning_rate": 0.0005758313842181375,
      "loss": 2.2007,
      "step": 29640
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6329815983772278,
      "learning_rate": 0.0005758297756397832,
      "loss": 2.4611,
      "step": 29641
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.7012696266174316,
      "learning_rate": 0.0005758281670101469,
      "loss": 2.3409,
      "step": 29642
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.7807575464248657,
      "learning_rate": 0.000575826558329229,
      "loss": 2.2614,
      "step": 29643
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.689680278301239,
      "learning_rate": 0.0005758249495970297,
      "loss": 2.314,
      "step": 29644
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6834075450897217,
      "learning_rate": 0.0005758233408135493,
      "loss": 2.4082,
      "step": 29645
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6432178020477295,
      "learning_rate": 0.0005758217319787883,
      "loss": 2.2068,
      "step": 29646
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6440913081169128,
      "learning_rate": 0.0005758201230927466,
      "loss": 2.264,
      "step": 29647
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6168311834335327,
      "learning_rate": 0.0005758185141554248,
      "loss": 2.2578,
      "step": 29648
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6922505497932434,
      "learning_rate": 0.0005758169051668233,
      "loss": 2.337,
      "step": 29649
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6518219709396362,
      "learning_rate": 0.0005758152961269421,
      "loss": 2.5488,
      "step": 29650
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6507477760314941,
      "learning_rate": 0.0005758136870357817,
      "loss": 2.4062,
      "step": 29651
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.649448037147522,
      "learning_rate": 0.0005758120778933423,
      "loss": 2.2242,
      "step": 29652
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.7259600162506104,
      "learning_rate": 0.0005758104686996241,
      "loss": 2.3655,
      "step": 29653
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6472911834716797,
      "learning_rate": 0.0005758088594546276,
      "loss": 2.2637,
      "step": 29654
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6516228318214417,
      "learning_rate": 0.000575807250158353,
      "loss": 2.2975,
      "step": 29655
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6901260614395142,
      "learning_rate": 0.0005758056408108007,
      "loss": 2.3873,
      "step": 29656
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.5995168685913086,
      "learning_rate": 0.0005758040314119709,
      "loss": 2.0881,
      "step": 29657
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6282639503479004,
      "learning_rate": 0.0005758024219618639,
      "loss": 2.297,
      "step": 29658
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.7623682618141174,
      "learning_rate": 0.0005758008124604801,
      "loss": 2.3463,
      "step": 29659
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6565189957618713,
      "learning_rate": 0.0005757992029078197,
      "loss": 2.3616,
      "step": 29660
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6948307156562805,
      "learning_rate": 0.0005757975933038829,
      "loss": 2.458,
      "step": 29661
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.763116717338562,
      "learning_rate": 0.0005757959836486703,
      "loss": 2.4404,
      "step": 29662
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.7193318605422974,
      "learning_rate": 0.0005757943739421818,
      "loss": 2.411,
      "step": 29663
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6036995053291321,
      "learning_rate": 0.0005757927641844182,
      "loss": 2.2993,
      "step": 29664
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.7077129483222961,
      "learning_rate": 0.0005757911543753793,
      "loss": 2.3703,
      "step": 29665
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.757570743560791,
      "learning_rate": 0.0005757895445150657,
      "loss": 2.4356,
      "step": 29666
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.7033982872962952,
      "learning_rate": 0.0005757879346034776,
      "loss": 2.4736,
      "step": 29667
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6322725415229797,
      "learning_rate": 0.0005757863246406153,
      "loss": 2.4794,
      "step": 29668
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.601384699344635,
      "learning_rate": 0.0005757847146264792,
      "loss": 2.198,
      "step": 29669
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6670771241188049,
      "learning_rate": 0.0005757831045610694,
      "loss": 2.2075,
      "step": 29670
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.5929322838783264,
      "learning_rate": 0.0005757814944443864,
      "loss": 2.4581,
      "step": 29671
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.691801905632019,
      "learning_rate": 0.0005757798842764304,
      "loss": 2.3731,
      "step": 29672
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6636702418327332,
      "learning_rate": 0.0005757782740572017,
      "loss": 2.2133,
      "step": 29673
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6379684805870056,
      "learning_rate": 0.0005757766637867006,
      "loss": 2.3452,
      "step": 29674
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6480910181999207,
      "learning_rate": 0.0005757750534649275,
      "loss": 2.4281,
      "step": 29675
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6497192978858948,
      "learning_rate": 0.0005757734430918825,
      "loss": 2.3104,
      "step": 29676
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6878387928009033,
      "learning_rate": 0.000575771832667566,
      "loss": 2.2739,
      "step": 29677
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6848784685134888,
      "learning_rate": 0.0005757702221919784,
      "loss": 2.6825,
      "step": 29678
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6040094494819641,
      "learning_rate": 0.0005757686116651199,
      "loss": 2.2616,
      "step": 29679
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6305306553840637,
      "learning_rate": 0.0005757670010869908,
      "loss": 2.2057,
      "step": 29680
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6278333067893982,
      "learning_rate": 0.0005757653904575913,
      "loss": 2.3992,
      "step": 29681
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6245113015174866,
      "learning_rate": 0.0005757637797769219,
      "loss": 2.457,
      "step": 29682
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6021548509597778,
      "learning_rate": 0.0005757621690449828,
      "loss": 2.2305,
      "step": 29683
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.748942494392395,
      "learning_rate": 0.0005757605582617744,
      "loss": 2.4388,
      "step": 29684
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6452861428260803,
      "learning_rate": 0.0005757589474272968,
      "loss": 2.2884,
      "step": 29685
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6586354970932007,
      "learning_rate": 0.0005757573365415505,
      "loss": 2.4811,
      "step": 29686
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6095783114433289,
      "learning_rate": 0.0005757557256045355,
      "loss": 2.1573,
      "step": 29687
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6430941224098206,
      "learning_rate": 0.0005757541146162525,
      "loss": 2.2809,
      "step": 29688
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6721890568733215,
      "learning_rate": 0.0005757525035767015,
      "loss": 2.2595,
      "step": 29689
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.668851912021637,
      "learning_rate": 0.0005757508924858829,
      "loss": 2.3991,
      "step": 29690
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6768303513526917,
      "learning_rate": 0.0005757492813437971,
      "loss": 2.2687,
      "step": 29691
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.591629147529602,
      "learning_rate": 0.0005757476701504443,
      "loss": 2.1868,
      "step": 29692
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.627921462059021,
      "learning_rate": 0.0005757460589058248,
      "loss": 2.2103,
      "step": 29693
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6283673644065857,
      "learning_rate": 0.0005757444476099388,
      "loss": 2.3967,
      "step": 29694
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6873449683189392,
      "learning_rate": 0.0005757428362627867,
      "loss": 2.3955,
      "step": 29695
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6484238505363464,
      "learning_rate": 0.0005757412248643689,
      "loss": 2.1946,
      "step": 29696
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6394240856170654,
      "learning_rate": 0.0005757396134146855,
      "loss": 2.1682,
      "step": 29697
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6885178685188293,
      "learning_rate": 0.0005757380019137371,
      "loss": 2.1476,
      "step": 29698
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.7678884863853455,
      "learning_rate": 0.0005757363903615236,
      "loss": 2.3525,
      "step": 29699
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6821905374526978,
      "learning_rate": 0.0005757347787580455,
      "loss": 2.2838,
      "step": 29700
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.7019528150558472,
      "learning_rate": 0.0005757331671033032,
      "loss": 2.287,
      "step": 29701
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.7277547121047974,
      "learning_rate": 0.000575731555397297,
      "loss": 2.4488,
      "step": 29702
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6273932456970215,
      "learning_rate": 0.0005757299436400269,
      "loss": 2.1623,
      "step": 29703
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6403059959411621,
      "learning_rate": 0.0005757283318314934,
      "loss": 2.3963,
      "step": 29704
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6414499282836914,
      "learning_rate": 0.0005757267199716969,
      "loss": 2.1238,
      "step": 29705
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6647287011146545,
      "learning_rate": 0.0005757251080606376,
      "loss": 2.0947,
      "step": 29706
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.785743236541748,
      "learning_rate": 0.0005757234960983157,
      "loss": 2.2016,
      "step": 29707
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.7027668356895447,
      "learning_rate": 0.0005757218840847317,
      "loss": 2.2143,
      "step": 29708
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6583982706069946,
      "learning_rate": 0.0005757202720198857,
      "loss": 2.0611,
      "step": 29709
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6463476419448853,
      "learning_rate": 0.0005757186599037781,
      "loss": 2.2546,
      "step": 29710
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6227352619171143,
      "learning_rate": 0.0005757170477364093,
      "loss": 2.308,
      "step": 29711
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6153554916381836,
      "learning_rate": 0.0005757154355177795,
      "loss": 2.2464,
      "step": 29712
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6585201621055603,
      "learning_rate": 0.0005757138232478889,
      "loss": 2.1372,
      "step": 29713
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.7380784749984741,
      "learning_rate": 0.0005757122109267379,
      "loss": 2.5316,
      "step": 29714
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6822606921195984,
      "learning_rate": 0.0005757105985543269,
      "loss": 2.1882,
      "step": 29715
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6775773763656616,
      "learning_rate": 0.0005757089861306559,
      "loss": 2.2775,
      "step": 29716
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.7031049132347107,
      "learning_rate": 0.0005757073736557256,
      "loss": 2.18,
      "step": 29717
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6288254261016846,
      "learning_rate": 0.000575705761129536,
      "loss": 2.3491,
      "step": 29718
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6372284889221191,
      "learning_rate": 0.0005757041485520875,
      "loss": 2.3356,
      "step": 29719
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.7761808633804321,
      "learning_rate": 0.0005757025359233804,
      "loss": 2.3937,
      "step": 29720
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6487364768981934,
      "learning_rate": 0.000575700923243415,
      "loss": 2.3308,
      "step": 29721
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6457905769348145,
      "learning_rate": 0.0005756993105121916,
      "loss": 2.2069,
      "step": 29722
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.603303849697113,
      "learning_rate": 0.0005756976977297105,
      "loss": 2.5102,
      "step": 29723
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6573808193206787,
      "learning_rate": 0.000575696084895972,
      "loss": 2.2149,
      "step": 29724
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6641608476638794,
      "learning_rate": 0.0005756944720109764,
      "loss": 2.2495,
      "step": 29725
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6541653871536255,
      "learning_rate": 0.0005756928590747239,
      "loss": 2.1902,
      "step": 29726
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.67343670129776,
      "learning_rate": 0.000575691246087215,
      "loss": 2.2172,
      "step": 29727
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6090858578681946,
      "learning_rate": 0.0005756896330484498,
      "loss": 2.1909,
      "step": 29728
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6704344749450684,
      "learning_rate": 0.0005756880199584288,
      "loss": 2.1559,
      "step": 29729
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6677989959716797,
      "learning_rate": 0.0005756864068171521,
      "loss": 2.2468,
      "step": 29730
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6360737681388855,
      "learning_rate": 0.0005756847936246201,
      "loss": 2.279,
      "step": 29731
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.592011034488678,
      "learning_rate": 0.0005756831803808332,
      "loss": 2.318,
      "step": 29732
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.605312168598175,
      "learning_rate": 0.0005756815670857915,
      "loss": 2.3988,
      "step": 29733
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.595480740070343,
      "learning_rate": 0.0005756799537394954,
      "loss": 2.1458,
      "step": 29734
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6187176704406738,
      "learning_rate": 0.0005756783403419452,
      "loss": 2.2407,
      "step": 29735
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.5966929197311401,
      "learning_rate": 0.0005756767268931412,
      "loss": 2.3547,
      "step": 29736
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6482514142990112,
      "learning_rate": 0.0005756751133930837,
      "loss": 2.3876,
      "step": 29737
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6842986345291138,
      "learning_rate": 0.000575673499841773,
      "loss": 2.4312,
      "step": 29738
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.7348819971084595,
      "learning_rate": 0.0005756718862392093,
      "loss": 2.3843,
      "step": 29739
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6981739401817322,
      "learning_rate": 0.0005756702725853931,
      "loss": 2.3767,
      "step": 29740
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.5703333616256714,
      "learning_rate": 0.0005756686588803245,
      "loss": 2.2978,
      "step": 29741
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6018308401107788,
      "learning_rate": 0.000575667045124004,
      "loss": 2.4143,
      "step": 29742
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.7452878952026367,
      "learning_rate": 0.0005756654313164317,
      "loss": 2.2348,
      "step": 29743
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6175393462181091,
      "learning_rate": 0.0005756638174576081,
      "loss": 2.293,
      "step": 29744
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6259042620658875,
      "learning_rate": 0.0005756622035475333,
      "loss": 2.4532,
      "step": 29745
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6395639777183533,
      "learning_rate": 0.0005756605895862077,
      "loss": 2.3198,
      "step": 29746
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6941041946411133,
      "learning_rate": 0.0005756589755736317,
      "loss": 2.3104,
      "step": 29747
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.5694405436515808,
      "learning_rate": 0.0005756573615098054,
      "loss": 2.3152,
      "step": 29748
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6768239736557007,
      "learning_rate": 0.0005756557473947292,
      "loss": 2.2289,
      "step": 29749
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6326014995574951,
      "learning_rate": 0.0005756541332284033,
      "loss": 2.161,
      "step": 29750
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6047893166542053,
      "learning_rate": 0.0005756525190108282,
      "loss": 2.2283,
      "step": 29751
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6545565724372864,
      "learning_rate": 0.0005756509047420041,
      "loss": 2.6002,
      "step": 29752
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6312915086746216,
      "learning_rate": 0.0005756492904219313,
      "loss": 2.3272,
      "step": 29753
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.7227054238319397,
      "learning_rate": 0.0005756476760506101,
      "loss": 2.2216,
      "step": 29754
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6127873063087463,
      "learning_rate": 0.0005756460616280407,
      "loss": 2.2185,
      "step": 29755
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6644479632377625,
      "learning_rate": 0.0005756444471542236,
      "loss": 2.0738,
      "step": 29756
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6612353324890137,
      "learning_rate": 0.0005756428326291589,
      "loss": 2.0741,
      "step": 29757
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6119970679283142,
      "learning_rate": 0.000575641218052847,
      "loss": 2.2685,
      "step": 29758
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6784480810165405,
      "learning_rate": 0.0005756396034252883,
      "loss": 2.303,
      "step": 29759
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6545847654342651,
      "learning_rate": 0.0005756379887464831,
      "loss": 2.1816,
      "step": 29760
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.7602438926696777,
      "learning_rate": 0.0005756363740164314,
      "loss": 2.4282,
      "step": 29761
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6430938243865967,
      "learning_rate": 0.0005756347592351336,
      "loss": 2.3899,
      "step": 29762
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6392872929573059,
      "learning_rate": 0.0005756331444025902,
      "loss": 2.4625,
      "step": 29763
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6118485331535339,
      "learning_rate": 0.0005756315295188015,
      "loss": 2.4116,
      "step": 29764
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.7570196390151978,
      "learning_rate": 0.0005756299145837676,
      "loss": 2.4127,
      "step": 29765
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6494843363761902,
      "learning_rate": 0.0005756282995974889,
      "loss": 2.3793,
      "step": 29766
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6080301403999329,
      "learning_rate": 0.0005756266845599657,
      "loss": 2.2486,
      "step": 29767
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.8785832524299622,
      "learning_rate": 0.0005756250694711983,
      "loss": 2.3298,
      "step": 29768
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.7440791726112366,
      "learning_rate": 0.000575623454331187,
      "loss": 1.9526,
      "step": 29769
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6791956424713135,
      "learning_rate": 0.000575621839139932,
      "loss": 2.2242,
      "step": 29770
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.8997264504432678,
      "learning_rate": 0.0005756202238974339,
      "loss": 2.3678,
      "step": 29771
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.7835190296173096,
      "learning_rate": 0.0005756186086036927,
      "loss": 2.3576,
      "step": 29772
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6565882563591003,
      "learning_rate": 0.0005756169932587087,
      "loss": 2.1287,
      "step": 29773
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.7636826634407043,
      "learning_rate": 0.0005756153778624824,
      "loss": 2.3237,
      "step": 29774
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.7952946424484253,
      "learning_rate": 0.0005756137624150139,
      "loss": 2.4636,
      "step": 29775
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6834359169006348,
      "learning_rate": 0.0005756121469163039,
      "loss": 2.3772,
      "step": 29776
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6231377124786377,
      "learning_rate": 0.000575610531366352,
      "loss": 2.4738,
      "step": 29777
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.7931034564971924,
      "learning_rate": 0.0005756089157651591,
      "loss": 2.2056,
      "step": 29778
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.7579266428947449,
      "learning_rate": 0.0005756073001127254,
      "loss": 2.2153,
      "step": 29779
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6482366323471069,
      "learning_rate": 0.0005756056844090509,
      "loss": 2.2517,
      "step": 29780
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6268588304519653,
      "learning_rate": 0.0005756040686541362,
      "loss": 2.3489,
      "step": 29781
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.8291475772857666,
      "learning_rate": 0.0005756024528479815,
      "loss": 2.3327,
      "step": 29782
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6994389891624451,
      "learning_rate": 0.0005756008369905871,
      "loss": 2.594,
      "step": 29783
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.7006228566169739,
      "learning_rate": 0.0005755992210819532,
      "loss": 2.2512,
      "step": 29784
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.7011534571647644,
      "learning_rate": 0.0005755976051220803,
      "loss": 2.1639,
      "step": 29785
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6806946396827698,
      "learning_rate": 0.0005755959891109686,
      "loss": 2.3615,
      "step": 29786
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6287594437599182,
      "learning_rate": 0.0005755943730486185,
      "loss": 2.3665,
      "step": 29787
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6282912492752075,
      "learning_rate": 0.00057559275693503,
      "loss": 2.2458,
      "step": 29788
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6368639469146729,
      "learning_rate": 0.0005755911407702037,
      "loss": 2.6489,
      "step": 29789
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6783267259597778,
      "learning_rate": 0.0005755895245541399,
      "loss": 2.6064,
      "step": 29790
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6268712878227234,
      "learning_rate": 0.0005755879082868387,
      "loss": 2.3752,
      "step": 29791
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.650667130947113,
      "learning_rate": 0.0005755862919683004,
      "loss": 2.387,
      "step": 29792
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6369243264198303,
      "learning_rate": 0.0005755846755985256,
      "loss": 2.344,
      "step": 29793
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6449193358421326,
      "learning_rate": 0.0005755830591775143,
      "loss": 2.4052,
      "step": 29794
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.5913776159286499,
      "learning_rate": 0.000575581442705267,
      "loss": 2.1272,
      "step": 29795
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.5899738669395447,
      "learning_rate": 0.0005755798261817838,
      "loss": 2.2246,
      "step": 29796
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6132709980010986,
      "learning_rate": 0.0005755782096070651,
      "loss": 1.9661,
      "step": 29797
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6644496917724609,
      "learning_rate": 0.0005755765929811113,
      "loss": 2.2397,
      "step": 29798
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6810440421104431,
      "learning_rate": 0.0005755749763039226,
      "loss": 2.291,
      "step": 29799
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6789728999137878,
      "learning_rate": 0.0005755733595754993,
      "loss": 2.3218,
      "step": 29800
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.7103634476661682,
      "learning_rate": 0.0005755717427958416,
      "loss": 2.0909,
      "step": 29801
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.8094508051872253,
      "learning_rate": 0.0005755701259649501,
      "loss": 2.3277,
      "step": 29802
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6349567174911499,
      "learning_rate": 0.0005755685090828249,
      "loss": 2.3384,
      "step": 29803
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.8309019207954407,
      "learning_rate": 0.0005755668921494662,
      "loss": 2.2082,
      "step": 29804
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.7872403860092163,
      "learning_rate": 0.0005755652751648744,
      "loss": 2.5284,
      "step": 29805
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.5770077109336853,
      "learning_rate": 0.0005755636581290498,
      "loss": 2.2514,
      "step": 29806
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.739902138710022,
      "learning_rate": 0.0005755620410419929,
      "loss": 2.4195,
      "step": 29807
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.8852580189704895,
      "learning_rate": 0.0005755604239037036,
      "loss": 2.2942,
      "step": 29808
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.636690080165863,
      "learning_rate": 0.0005755588067141825,
      "loss": 2.2827,
      "step": 29809
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.7512302398681641,
      "learning_rate": 0.0005755571894734299,
      "loss": 2.2417,
      "step": 29810
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.7724112868309021,
      "learning_rate": 0.0005755555721814459,
      "loss": 2.297,
      "step": 29811
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.8291906118392944,
      "learning_rate": 0.000575553954838231,
      "loss": 2.0998,
      "step": 29812
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6967895030975342,
      "learning_rate": 0.0005755523374437853,
      "loss": 2.4891,
      "step": 29813
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.7847270369529724,
      "learning_rate": 0.0005755507199981093,
      "loss": 2.4268,
      "step": 29814
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.007364273071289,
      "learning_rate": 0.0005755491025012031,
      "loss": 2.2996,
      "step": 29815
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6448459625244141,
      "learning_rate": 0.0005755474849530673,
      "loss": 2.2781,
      "step": 29816
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.047595739364624,
      "learning_rate": 0.0005755458673537019,
      "loss": 2.3594,
      "step": 29817
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.868702232837677,
      "learning_rate": 0.0005755442497031074,
      "loss": 2.4695,
      "step": 29818
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6546096205711365,
      "learning_rate": 0.0005755426320012841,
      "loss": 2.406,
      "step": 29819
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.7633438110351562,
      "learning_rate": 0.0005755410142482321,
      "loss": 2.2731,
      "step": 29820
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.8010419011116028,
      "learning_rate": 0.0005755393964439518,
      "loss": 2.5137,
      "step": 29821
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.7469469904899597,
      "learning_rate": 0.0005755377785884435,
      "loss": 2.1689,
      "step": 29822
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6699827909469604,
      "learning_rate": 0.0005755361606817076,
      "loss": 2.2882,
      "step": 29823
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6715394258499146,
      "learning_rate": 0.0005755345427237445,
      "loss": 2.2471,
      "step": 29824
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.7313259840011597,
      "learning_rate": 0.0005755329247145541,
      "loss": 2.3092,
      "step": 29825
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.7905275821685791,
      "learning_rate": 0.000575531306654137,
      "loss": 2.3412,
      "step": 29826
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6254984736442566,
      "learning_rate": 0.0005755296885424933,
      "loss": 2.2569,
      "step": 29827
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.7070889472961426,
      "learning_rate": 0.0005755280703796235,
      "loss": 2.3752,
      "step": 29828
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6733625531196594,
      "learning_rate": 0.0005755264521655279,
      "loss": 2.3674,
      "step": 29829
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.7719777226448059,
      "learning_rate": 0.0005755248339002067,
      "loss": 2.2779,
      "step": 29830
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6577627658843994,
      "learning_rate": 0.0005755232155836603,
      "loss": 2.3363,
      "step": 29831
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6849983334541321,
      "learning_rate": 0.0005755215972158888,
      "loss": 2.197,
      "step": 29832
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.9631433486938477,
      "learning_rate": 0.0005755199787968927,
      "loss": 2.4594,
      "step": 29833
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.7851040363311768,
      "learning_rate": 0.0005755183603266722,
      "loss": 2.1211,
      "step": 29834
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6926098465919495,
      "learning_rate": 0.0005755167418052277,
      "loss": 2.4887,
      "step": 29835
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.9320820569992065,
      "learning_rate": 0.0005755151232325594,
      "loss": 2.426,
      "step": 29836
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.7551162838935852,
      "learning_rate": 0.0005755135046086676,
      "loss": 2.446,
      "step": 29837
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6866965889930725,
      "learning_rate": 0.0005755118859335527,
      "loss": 2.3044,
      "step": 29838
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6939185857772827,
      "learning_rate": 0.0005755102672072149,
      "loss": 2.1993,
      "step": 29839
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.9005032777786255,
      "learning_rate": 0.0005755086484296546,
      "loss": 2.2329,
      "step": 29840
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6553856134414673,
      "learning_rate": 0.000575507029600872,
      "loss": 2.383,
      "step": 29841
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6347984671592712,
      "learning_rate": 0.0005755054107208675,
      "loss": 2.2881,
      "step": 29842
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6473729014396667,
      "learning_rate": 0.0005755037917896412,
      "loss": 1.9811,
      "step": 29843
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.7820661067962646,
      "learning_rate": 0.0005755021728071937,
      "loss": 2.4476,
      "step": 29844
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.8298563957214355,
      "learning_rate": 0.000575500553773525,
      "loss": 2.2448,
      "step": 29845
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.784029483795166,
      "learning_rate": 0.0005754989346886357,
      "loss": 2.3398,
      "step": 29846
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.0606482028961182,
      "learning_rate": 0.0005754973155525259,
      "loss": 2.5319,
      "step": 29847
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.103208303451538,
      "learning_rate": 0.0005754956963651958,
      "loss": 2.3164,
      "step": 29848
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6506218314170837,
      "learning_rate": 0.0005754940771266461,
      "loss": 2.2888,
      "step": 29849
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.1730576753616333,
      "learning_rate": 0.0005754924578368766,
      "loss": 2.1654,
      "step": 29850
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.7579604983329773,
      "learning_rate": 0.0005754908384958881,
      "loss": 2.4551,
      "step": 29851
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.8346493244171143,
      "learning_rate": 0.0005754892191036804,
      "loss": 2.2377,
      "step": 29852
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6218888759613037,
      "learning_rate": 0.0005754875996602542,
      "loss": 2.2772,
      "step": 29853
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.8862309455871582,
      "learning_rate": 0.0005754859801656097,
      "loss": 2.4904,
      "step": 29854
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.8820880651473999,
      "learning_rate": 0.0005754843606197472,
      "loss": 2.1518,
      "step": 29855
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.677090048789978,
      "learning_rate": 0.0005754827410226667,
      "loss": 2.3736,
      "step": 29856
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6162078976631165,
      "learning_rate": 0.000575481121374369,
      "loss": 2.1526,
      "step": 29857
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.781032145023346,
      "learning_rate": 0.000575479501674854,
      "loss": 2.096,
      "step": 29858
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.7180500626564026,
      "learning_rate": 0.0005754778819241222,
      "loss": 2.2995,
      "step": 29859
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6999027132987976,
      "learning_rate": 0.0005754762621221739,
      "loss": 2.173,
      "step": 29860
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6650757193565369,
      "learning_rate": 0.0005754746422690094,
      "loss": 2.3256,
      "step": 29861
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6854044198989868,
      "learning_rate": 0.0005754730223646289,
      "loss": 2.4156,
      "step": 29862
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.7001088857650757,
      "learning_rate": 0.0005754714024090328,
      "loss": 2.301,
      "step": 29863
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6927860379219055,
      "learning_rate": 0.0005754697824022214,
      "loss": 2.3724,
      "step": 29864
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.7363414168357849,
      "learning_rate": 0.0005754681623441948,
      "loss": 2.2556,
      "step": 29865
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6861193180084229,
      "learning_rate": 0.0005754665422349536,
      "loss": 2.2291,
      "step": 29866
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6406322717666626,
      "learning_rate": 0.000575464922074498,
      "loss": 2.2769,
      "step": 29867
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6729788184165955,
      "learning_rate": 0.0005754633018628283,
      "loss": 2.3658,
      "step": 29868
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.7643662095069885,
      "learning_rate": 0.0005754616815999447,
      "loss": 2.2288,
      "step": 29869
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.642045259475708,
      "learning_rate": 0.0005754600612858476,
      "loss": 2.0878,
      "step": 29870
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6616353392601013,
      "learning_rate": 0.0005754584409205371,
      "loss": 2.418,
      "step": 29871
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.7017512917518616,
      "learning_rate": 0.0005754568205040139,
      "loss": 2.1448,
      "step": 29872
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.7048478722572327,
      "learning_rate": 0.000575455200036278,
      "loss": 2.4347,
      "step": 29873
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6834139227867126,
      "learning_rate": 0.0005754535795173298,
      "loss": 2.1719,
      "step": 29874
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.5757218599319458,
      "learning_rate": 0.0005754519589471694,
      "loss": 2.3148,
      "step": 29875
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6849898099899292,
      "learning_rate": 0.0005754503383257975,
      "loss": 2.5959,
      "step": 29876
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6972465515136719,
      "learning_rate": 0.0005754487176532142,
      "loss": 2.4838,
      "step": 29877
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.7234869003295898,
      "learning_rate": 0.0005754470969294198,
      "loss": 2.4312,
      "step": 29878
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6943584680557251,
      "learning_rate": 0.0005754454761544143,
      "loss": 1.9887,
      "step": 29879
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6764422655105591,
      "learning_rate": 0.0005754438553281986,
      "loss": 2.3045,
      "step": 29880
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6620385646820068,
      "learning_rate": 0.0005754422344507725,
      "loss": 2.0956,
      "step": 29881
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6178314685821533,
      "learning_rate": 0.0005754406135221365,
      "loss": 2.2549,
      "step": 29882
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.7109761238098145,
      "learning_rate": 0.000575438992542291,
      "loss": 2.4418,
      "step": 29883
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.8174643516540527,
      "learning_rate": 0.0005754373715112361,
      "loss": 2.243,
      "step": 29884
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6526234149932861,
      "learning_rate": 0.0005754357504289722,
      "loss": 2.3199,
      "step": 29885
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6558012366294861,
      "learning_rate": 0.0005754341292954998,
      "loss": 2.2213,
      "step": 29886
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.576694130897522,
      "learning_rate": 0.0005754325081108188,
      "loss": 2.0852,
      "step": 29887
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.8729432821273804,
      "learning_rate": 0.0005754308868749297,
      "loss": 2.5063,
      "step": 29888
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6048679947853088,
      "learning_rate": 0.0005754292655878328,
      "loss": 2.3516,
      "step": 29889
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.7607631683349609,
      "learning_rate": 0.0005754276442495284,
      "loss": 2.3642,
      "step": 29890
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6693382263183594,
      "learning_rate": 0.0005754260228600168,
      "loss": 2.4313,
      "step": 29891
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6213985681533813,
      "learning_rate": 0.0005754244014192984,
      "loss": 2.2707,
      "step": 29892
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.5790231823921204,
      "learning_rate": 0.0005754227799273732,
      "loss": 2.33,
      "step": 29893
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6497175097465515,
      "learning_rate": 0.0005754211583842419,
      "loss": 2.2408,
      "step": 29894
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6210301518440247,
      "learning_rate": 0.0005754195367899046,
      "loss": 2.134,
      "step": 29895
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6401147246360779,
      "learning_rate": 0.0005754179151443615,
      "loss": 2.2326,
      "step": 29896
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.5988796353340149,
      "learning_rate": 0.0005754162934476131,
      "loss": 2.3587,
      "step": 29897
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6891996264457703,
      "learning_rate": 0.0005754146716996595,
      "loss": 2.1971,
      "step": 29898
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6091803908348083,
      "learning_rate": 0.0005754130499005013,
      "loss": 2.3873,
      "step": 29899
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.629907488822937,
      "learning_rate": 0.0005754114280501385,
      "loss": 2.5645,
      "step": 29900
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6451677083969116,
      "learning_rate": 0.0005754098061485714,
      "loss": 2.2115,
      "step": 29901
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6404264569282532,
      "learning_rate": 0.0005754081841958007,
      "loss": 2.2084,
      "step": 29902
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6163884997367859,
      "learning_rate": 0.0005754065621918262,
      "loss": 2.0941,
      "step": 29903
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.7275456786155701,
      "learning_rate": 0.0005754049401366484,
      "loss": 2.2656,
      "step": 29904
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.5927055478096008,
      "learning_rate": 0.0005754033180302677,
      "loss": 2.371,
      "step": 29905
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6318150162696838,
      "learning_rate": 0.0005754016958726844,
      "loss": 2.4032,
      "step": 29906
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.5732226371765137,
      "learning_rate": 0.0005754000736638987,
      "loss": 2.2645,
      "step": 29907
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6407284140586853,
      "learning_rate": 0.0005753984514039109,
      "loss": 2.5662,
      "step": 29908
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.7300575971603394,
      "learning_rate": 0.0005753968290927213,
      "loss": 2.3924,
      "step": 29909
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6807156205177307,
      "learning_rate": 0.0005753952067303301,
      "loss": 2.3038,
      "step": 29910
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6138509511947632,
      "learning_rate": 0.0005753935843167379,
      "loss": 2.4511,
      "step": 29911
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.630761981010437,
      "learning_rate": 0.0005753919618519448,
      "loss": 2.2298,
      "step": 29912
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.5980399250984192,
      "learning_rate": 0.0005753903393359512,
      "loss": 2.2467,
      "step": 29913
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6138767600059509,
      "learning_rate": 0.0005753887167687573,
      "loss": 2.4891,
      "step": 29914
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6528798341751099,
      "learning_rate": 0.0005753870941503634,
      "loss": 2.2116,
      "step": 29915
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6341097950935364,
      "learning_rate": 0.0005753854714807699,
      "loss": 2.5167,
      "step": 29916
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.7222516536712646,
      "learning_rate": 0.000575383848759977,
      "loss": 2.2595,
      "step": 29917
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6177414059638977,
      "learning_rate": 0.000575382225987985,
      "loss": 2.1604,
      "step": 29918
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.7410787343978882,
      "learning_rate": 0.0005753806031647943,
      "loss": 2.4311,
      "step": 29919
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6946361660957336,
      "learning_rate": 0.0005753789802904051,
      "loss": 2.4406,
      "step": 29920
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6949864029884338,
      "learning_rate": 0.0005753773573648178,
      "loss": 2.2706,
      "step": 29921
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.5979471802711487,
      "learning_rate": 0.0005753757343880327,
      "loss": 2.1572,
      "step": 29922
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6085194945335388,
      "learning_rate": 0.00057537411136005,
      "loss": 2.2879,
      "step": 29923
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6369509100914001,
      "learning_rate": 0.00057537248828087,
      "loss": 2.5197,
      "step": 29924
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.674597978591919,
      "learning_rate": 0.0005753708651504932,
      "loss": 2.3368,
      "step": 29925
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6703494787216187,
      "learning_rate": 0.0005753692419689197,
      "loss": 2.3008,
      "step": 29926
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.5948329567909241,
      "learning_rate": 0.0005753676187361497,
      "loss": 2.1652,
      "step": 29927
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6623444557189941,
      "learning_rate": 0.0005753659954521839,
      "loss": 2.3072,
      "step": 29928
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6400050520896912,
      "learning_rate": 0.0005753643721170222,
      "loss": 2.2015,
      "step": 29929
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6272059679031372,
      "learning_rate": 0.0005753627487306651,
      "loss": 2.2357,
      "step": 29930
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6586174368858337,
      "learning_rate": 0.0005753611252931129,
      "loss": 2.5413,
      "step": 29931
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.608909547328949,
      "learning_rate": 0.0005753595018043659,
      "loss": 2.1409,
      "step": 29932
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6425281763076782,
      "learning_rate": 0.0005753578782644243,
      "loss": 2.2823,
      "step": 29933
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6706607937812805,
      "learning_rate": 0.0005753562546732884,
      "loss": 2.2685,
      "step": 29934
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6808770298957825,
      "learning_rate": 0.0005753546310309587,
      "loss": 2.3783,
      "step": 29935
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6572399139404297,
      "learning_rate": 0.0005753530073374354,
      "loss": 2.3482,
      "step": 29936
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6338856220245361,
      "learning_rate": 0.0005753513835927188,
      "loss": 2.3065,
      "step": 29937
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6385894417762756,
      "learning_rate": 0.000575349759796809,
      "loss": 2.3725,
      "step": 29938
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.607223391532898,
      "learning_rate": 0.0005753481359497066,
      "loss": 2.3799,
      "step": 29939
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.7701824307441711,
      "learning_rate": 0.0005753465120514117,
      "loss": 2.3787,
      "step": 29940
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6344553232192993,
      "learning_rate": 0.0005753448881019248,
      "loss": 2.3249,
      "step": 29941
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.7207696437835693,
      "learning_rate": 0.000575343264101246,
      "loss": 2.5511,
      "step": 29942
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6951619982719421,
      "learning_rate": 0.0005753416400493757,
      "loss": 2.3519,
      "step": 29943
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.7002811431884766,
      "learning_rate": 0.0005753400159463143,
      "loss": 2.3487,
      "step": 29944
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.7403700351715088,
      "learning_rate": 0.000575338391792062,
      "loss": 2.2852,
      "step": 29945
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.8433571457862854,
      "learning_rate": 0.0005753367675866189,
      "loss": 2.106,
      "step": 29946
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6657026410102844,
      "learning_rate": 0.0005753351433299856,
      "loss": 2.5464,
      "step": 29947
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6306635141372681,
      "learning_rate": 0.0005753335190221624,
      "loss": 2.2574,
      "step": 29948
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.7043790221214294,
      "learning_rate": 0.0005753318946631493,
      "loss": 2.3278,
      "step": 29949
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6977135539054871,
      "learning_rate": 0.0005753302702529471,
      "loss": 2.3114,
      "step": 29950
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.7029199004173279,
      "learning_rate": 0.0005753286457915555,
      "loss": 2.1711,
      "step": 29951
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6292842626571655,
      "learning_rate": 0.0005753270212789753,
      "loss": 2.3586,
      "step": 29952
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6810554265975952,
      "learning_rate": 0.0005753253967152065,
      "loss": 2.4928,
      "step": 29953
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6786947846412659,
      "learning_rate": 0.0005753237721002496,
      "loss": 2.206,
      "step": 29954
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6751472353935242,
      "learning_rate": 0.0005753221474341047,
      "loss": 2.4886,
      "step": 29955
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6123555302619934,
      "learning_rate": 0.0005753205227167723,
      "loss": 2.1724,
      "step": 29956
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.7011033296585083,
      "learning_rate": 0.0005753188979482525,
      "loss": 2.435,
      "step": 29957
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.7786841988563538,
      "learning_rate": 0.0005753172731285459,
      "loss": 2.3864,
      "step": 29958
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.576011598110199,
      "learning_rate": 0.0005753156482576525,
      "loss": 2.2164,
      "step": 29959
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6410913467407227,
      "learning_rate": 0.0005753140233355728,
      "loss": 2.1424,
      "step": 29960
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6816244721412659,
      "learning_rate": 0.0005753123983623069,
      "loss": 2.3651,
      "step": 29961
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.7129164934158325,
      "learning_rate": 0.0005753107733378554,
      "loss": 2.3207,
      "step": 29962
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6582503318786621,
      "learning_rate": 0.0005753091482622183,
      "loss": 2.4578,
      "step": 29963
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6600995659828186,
      "learning_rate": 0.0005753075231353962,
      "loss": 2.4248,
      "step": 29964
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.663451611995697,
      "learning_rate": 0.000575305897957389,
      "loss": 2.2653,
      "step": 29965
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6274003386497498,
      "learning_rate": 0.0005753042727281973,
      "loss": 2.3543,
      "step": 29966
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6225204467773438,
      "learning_rate": 0.0005753026474478215,
      "loss": 2.5371,
      "step": 29967
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.7817510962486267,
      "learning_rate": 0.0005753010221162616,
      "loss": 2.3255,
      "step": 29968
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.7083775401115417,
      "learning_rate": 0.0005752993967335181,
      "loss": 2.3865,
      "step": 29969
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6225127577781677,
      "learning_rate": 0.0005752977712995913,
      "loss": 2.0354,
      "step": 29970
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.8148614168167114,
      "learning_rate": 0.0005752961458144812,
      "loss": 2.2994,
      "step": 29971
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6733555197715759,
      "learning_rate": 0.0005752945202781887,
      "loss": 2.6001,
      "step": 29972
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6980842351913452,
      "learning_rate": 0.0005752928946907135,
      "loss": 2.2318,
      "step": 29973
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6901339292526245,
      "learning_rate": 0.0005752912690520562,
      "loss": 2.3,
      "step": 29974
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6559435725212097,
      "learning_rate": 0.0005752896433622171,
      "loss": 2.5614,
      "step": 29975
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.585951566696167,
      "learning_rate": 0.0005752880176211965,
      "loss": 2.3373,
      "step": 29976
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.7705246210098267,
      "learning_rate": 0.0005752863918289945,
      "loss": 2.3395,
      "step": 29977
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6922686696052551,
      "learning_rate": 0.0005752847659856117,
      "loss": 2.313,
      "step": 29978
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.8020562529563904,
      "learning_rate": 0.0005752831400910482,
      "loss": 2.5019,
      "step": 29979
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6164003014564514,
      "learning_rate": 0.0005752815141453044,
      "loss": 2.1975,
      "step": 29980
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6826968789100647,
      "learning_rate": 0.0005752798881483806,
      "loss": 2.2658,
      "step": 29981
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.7716333866119385,
      "learning_rate": 0.0005752782621002771,
      "loss": 2.4751,
      "step": 29982
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6410987973213196,
      "learning_rate": 0.0005752766360009941,
      "loss": 2.2595,
      "step": 29983
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6156099438667297,
      "learning_rate": 0.0005752750098505319,
      "loss": 2.1598,
      "step": 29984
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6231635808944702,
      "learning_rate": 0.000575273383648891,
      "loss": 2.3682,
      "step": 29985
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6459521055221558,
      "learning_rate": 0.0005752717573960716,
      "loss": 2.2844,
      "step": 29986
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.5951159596443176,
      "learning_rate": 0.000575270131092074,
      "loss": 2.4881,
      "step": 29987
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6591587662696838,
      "learning_rate": 0.0005752685047368983,
      "loss": 2.4099,
      "step": 29988
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.667610228061676,
      "learning_rate": 0.0005752668783305452,
      "loss": 2.4345,
      "step": 29989
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.7500665187835693,
      "learning_rate": 0.0005752652518730146,
      "loss": 2.5318,
      "step": 29990
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6066729426383972,
      "learning_rate": 0.0005752636253643071,
      "loss": 2.2804,
      "step": 29991
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6808688044548035,
      "learning_rate": 0.0005752619988044229,
      "loss": 2.2054,
      "step": 29992
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6195389628410339,
      "learning_rate": 0.0005752603721933624,
      "loss": 2.3532,
      "step": 29993
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6226502060890198,
      "learning_rate": 0.0005752587455311256,
      "loss": 2.171,
      "step": 29994
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.625870406627655,
      "learning_rate": 0.0005752571188177131,
      "loss": 2.3391,
      "step": 29995
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6596253514289856,
      "learning_rate": 0.0005752554920531251,
      "loss": 2.2846,
      "step": 29996
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6142097115516663,
      "learning_rate": 0.000575253865237362,
      "loss": 2.1928,
      "step": 29997
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6583409309387207,
      "learning_rate": 0.0005752522383704239,
      "loss": 2.3719,
      "step": 29998
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6558692455291748,
      "learning_rate": 0.0005752506114523112,
      "loss": 2.2956,
      "step": 29999
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.599704384803772,
      "learning_rate": 0.0005752489844830243,
      "loss": 2.2776,
      "step": 30000
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6493820548057556,
      "learning_rate": 0.0005752473574625633,
      "loss": 2.2601,
      "step": 30001
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6610499620437622,
      "learning_rate": 0.0005752457303909287,
      "loss": 2.5564,
      "step": 30002
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.5957750082015991,
      "learning_rate": 0.0005752441032681208,
      "loss": 2.3633,
      "step": 30003
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.7103159427642822,
      "learning_rate": 0.0005752424760941397,
      "loss": 2.2202,
      "step": 30004
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6518768072128296,
      "learning_rate": 0.0005752408488689858,
      "loss": 2.3209,
      "step": 30005
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.5758997201919556,
      "learning_rate": 0.0005752392215926595,
      "loss": 2.391,
      "step": 30006
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.5990768074989319,
      "learning_rate": 0.0005752375942651611,
      "loss": 2.264,
      "step": 30007
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6555601358413696,
      "learning_rate": 0.0005752359668864907,
      "loss": 2.1546,
      "step": 30008
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6228715181350708,
      "learning_rate": 0.0005752343394566489,
      "loss": 2.3446,
      "step": 30009
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.62786465883255,
      "learning_rate": 0.0005752327119756356,
      "loss": 2.5113,
      "step": 30010
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6908722519874573,
      "learning_rate": 0.0005752310844434515,
      "loss": 2.2255,
      "step": 30011
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6038776636123657,
      "learning_rate": 0.0005752294568600968,
      "loss": 2.2479,
      "step": 30012
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6228377819061279,
      "learning_rate": 0.0005752278292255717,
      "loss": 2.2072,
      "step": 30013
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6922584176063538,
      "learning_rate": 0.0005752262015398766,
      "loss": 2.457,
      "step": 30014
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.5895347595214844,
      "learning_rate": 0.0005752245738030116,
      "loss": 2.1158,
      "step": 30015
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6515071392059326,
      "learning_rate": 0.0005752229460149772,
      "loss": 2.2654,
      "step": 30016
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6871113777160645,
      "learning_rate": 0.0005752213181757737,
      "loss": 2.2434,
      "step": 30017
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.7151250243186951,
      "learning_rate": 0.0005752196902854014,
      "loss": 2.1696,
      "step": 30018
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.7132129669189453,
      "learning_rate": 0.0005752180623438605,
      "loss": 2.3293,
      "step": 30019
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.7363296151161194,
      "learning_rate": 0.0005752164343511514,
      "loss": 2.5075,
      "step": 30020
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6993515491485596,
      "learning_rate": 0.0005752148063072744,
      "loss": 2.5793,
      "step": 30021
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6515552997589111,
      "learning_rate": 0.0005752131782122298,
      "loss": 2.4894,
      "step": 30022
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6758151650428772,
      "learning_rate": 0.0005752115500660178,
      "loss": 2.3781,
      "step": 30023
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6428196430206299,
      "learning_rate": 0.0005752099218686387,
      "loss": 2.4209,
      "step": 30024
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6698493361473083,
      "learning_rate": 0.000575208293620093,
      "loss": 2.1426,
      "step": 30025
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6431262493133545,
      "learning_rate": 0.0005752066653203808,
      "loss": 2.2899,
      "step": 30026
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6701699495315552,
      "learning_rate": 0.0005752050369695026,
      "loss": 2.5345,
      "step": 30027
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.692731499671936,
      "learning_rate": 0.0005752034085674586,
      "loss": 2.3821,
      "step": 30028
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6220204830169678,
      "learning_rate": 0.0005752017801142489,
      "loss": 2.2449,
      "step": 30029
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.5962275266647339,
      "learning_rate": 0.0005752001516098742,
      "loss": 2.197,
      "step": 30030
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.7229761481285095,
      "learning_rate": 0.0005751985230543345,
      "loss": 2.3334,
      "step": 30031
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.695564866065979,
      "learning_rate": 0.0005751968944476303,
      "loss": 2.5046,
      "step": 30032
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6281720995903015,
      "learning_rate": 0.0005751952657897616,
      "loss": 2.2278,
      "step": 30033
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.697176456451416,
      "learning_rate": 0.0005751936370807291,
      "loss": 2.3753,
      "step": 30034
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6386656165122986,
      "learning_rate": 0.0005751920083205327,
      "loss": 2.5635,
      "step": 30035
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6795275211334229,
      "learning_rate": 0.0005751903795091731,
      "loss": 2.1218,
      "step": 30036
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6702499985694885,
      "learning_rate": 0.0005751887506466503,
      "loss": 2.4611,
      "step": 30037
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6595489978790283,
      "learning_rate": 0.0005751871217329649,
      "loss": 2.6547,
      "step": 30038
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.577268123626709,
      "learning_rate": 0.0005751854927681169,
      "loss": 2.2164,
      "step": 30039
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.61738520860672,
      "learning_rate": 0.0005751838637521067,
      "loss": 2.3851,
      "step": 30040
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6203767657279968,
      "learning_rate": 0.0005751822346849346,
      "loss": 2.389,
      "step": 30041
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.7233378887176514,
      "learning_rate": 0.0005751806055666009,
      "loss": 2.4045,
      "step": 30042
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6536446809768677,
      "learning_rate": 0.000575178976397106,
      "loss": 2.1708,
      "step": 30043
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6360883116722107,
      "learning_rate": 0.0005751773471764501,
      "loss": 2.1865,
      "step": 30044
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.7189996242523193,
      "learning_rate": 0.0005751757179046336,
      "loss": 2.1969,
      "step": 30045
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6420816779136658,
      "learning_rate": 0.0005751740885816567,
      "loss": 2.4121,
      "step": 30046
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6863670945167542,
      "learning_rate": 0.0005751724592075198,
      "loss": 2.12,
      "step": 30047
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.7142459154129028,
      "learning_rate": 0.000575170829782223,
      "loss": 2.3432,
      "step": 30048
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.626705527305603,
      "learning_rate": 0.0005751692003057669,
      "loss": 2.1345,
      "step": 30049
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.7132204174995422,
      "learning_rate": 0.0005751675707781516,
      "loss": 2.2132,
      "step": 30050
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6797134876251221,
      "learning_rate": 0.0005751659411993774,
      "loss": 2.377,
      "step": 30051
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6245877742767334,
      "learning_rate": 0.0005751643115694448,
      "loss": 2.3492,
      "step": 30052
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6025981307029724,
      "learning_rate": 0.0005751626818883538,
      "loss": 2.1172,
      "step": 30053
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6461952328681946,
      "learning_rate": 0.0005751610521561049,
      "loss": 2.5538,
      "step": 30054
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6136649250984192,
      "learning_rate": 0.0005751594223726984,
      "loss": 2.4888,
      "step": 30055
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6679800748825073,
      "learning_rate": 0.0005751577925381346,
      "loss": 2.1764,
      "step": 30056
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6501963138580322,
      "learning_rate": 0.0005751561626524138,
      "loss": 2.3354,
      "step": 30057
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.7055522203445435,
      "learning_rate": 0.0005751545327155362,
      "loss": 2.4397,
      "step": 30058
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.5968123078346252,
      "learning_rate": 0.0005751529027275022,
      "loss": 2.1352,
      "step": 30059
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6037437319755554,
      "learning_rate": 0.000575151272688312,
      "loss": 2.329,
      "step": 30060
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6549440622329712,
      "learning_rate": 0.0005751496425979661,
      "loss": 2.1404,
      "step": 30061
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6189646124839783,
      "learning_rate": 0.0005751480124564646,
      "loss": 2.4915,
      "step": 30062
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6660516262054443,
      "learning_rate": 0.000575146382263808,
      "loss": 2.1454,
      "step": 30063
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.602911114692688,
      "learning_rate": 0.0005751447520199965,
      "loss": 2.4011,
      "step": 30064
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6278093457221985,
      "learning_rate": 0.0005751431217250302,
      "loss": 2.1747,
      "step": 30065
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.632357656955719,
      "learning_rate": 0.0005751414913789097,
      "loss": 2.4461,
      "step": 30066
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6748117208480835,
      "learning_rate": 0.0005751398609816354,
      "loss": 2.1326,
      "step": 30067
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6055691242218018,
      "learning_rate": 0.0005751382305332071,
      "loss": 2.5633,
      "step": 30068
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6926959156990051,
      "learning_rate": 0.0005751366000336256,
      "loss": 2.0961,
      "step": 30069
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6988457441329956,
      "learning_rate": 0.0005751349694828909,
      "loss": 2.5478,
      "step": 30070
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6653711795806885,
      "learning_rate": 0.0005751333388810035,
      "loss": 2.1221,
      "step": 30071
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.741124153137207,
      "learning_rate": 0.0005751317082279636,
      "loss": 2.4223,
      "step": 30072
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6222506165504456,
      "learning_rate": 0.0005751300775237715,
      "loss": 2.4191,
      "step": 30073
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.7278262376785278,
      "learning_rate": 0.0005751284467684275,
      "loss": 2.4344,
      "step": 30074
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6770552396774292,
      "learning_rate": 0.000575126815961932,
      "loss": 2.2186,
      "step": 30075
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6441064476966858,
      "learning_rate": 0.0005751251851042852,
      "loss": 2.2952,
      "step": 30076
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6188047528266907,
      "learning_rate": 0.0005751235541954874,
      "loss": 2.2915,
      "step": 30077
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6096639633178711,
      "learning_rate": 0.000575121923235539,
      "loss": 2.2119,
      "step": 30078
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6229467988014221,
      "learning_rate": 0.0005751202922244403,
      "loss": 2.3031,
      "step": 30079
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.7832093834877014,
      "learning_rate": 0.0005751186611621914,
      "loss": 2.4584,
      "step": 30080
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6119762063026428,
      "learning_rate": 0.0005751170300487928,
      "loss": 2.383,
      "step": 30081
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6388528347015381,
      "learning_rate": 0.0005751153988842447,
      "loss": 2.4639,
      "step": 30082
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6608702540397644,
      "learning_rate": 0.0005751137676685476,
      "loss": 2.3022,
      "step": 30083
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.7861617803573608,
      "learning_rate": 0.0005751121364017015,
      "loss": 2.0554,
      "step": 30084
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6416260600090027,
      "learning_rate": 0.0005751105050837069,
      "loss": 2.3091,
      "step": 30085
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6533446907997131,
      "learning_rate": 0.0005751088737145641,
      "loss": 2.3452,
      "step": 30086
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.7243051528930664,
      "learning_rate": 0.0005751072422942733,
      "loss": 2.2591,
      "step": 30087
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6433428525924683,
      "learning_rate": 0.000575105610822835,
      "loss": 2.4644,
      "step": 30088
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6472607254981995,
      "learning_rate": 0.0005751039793002494,
      "loss": 2.1413,
      "step": 30089
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.7449659705162048,
      "learning_rate": 0.0005751023477265166,
      "loss": 2.2715,
      "step": 30090
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.8067467212677002,
      "learning_rate": 0.0005751007161016372,
      "loss": 2.4005,
      "step": 30091
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6800596117973328,
      "learning_rate": 0.0005750990844256113,
      "loss": 2.6093,
      "step": 30092
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.675256073474884,
      "learning_rate": 0.0005750974526984394,
      "loss": 2.271,
      "step": 30093
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.7816242575645447,
      "learning_rate": 0.0005750958209201216,
      "loss": 2.411,
      "step": 30094
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6814267039299011,
      "learning_rate": 0.0005750941890906583,
      "loss": 2.3971,
      "step": 30095
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6289166808128357,
      "learning_rate": 0.0005750925572100498,
      "loss": 2.3047,
      "step": 30096
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.9017943143844604,
      "learning_rate": 0.0005750909252782965,
      "loss": 2.2525,
      "step": 30097
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6585627794265747,
      "learning_rate": 0.0005750892932953985,
      "loss": 2.4228,
      "step": 30098
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6751285195350647,
      "learning_rate": 0.0005750876612613562,
      "loss": 2.2816,
      "step": 30099
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6273354291915894,
      "learning_rate": 0.00057508602917617,
      "loss": 2.2923,
      "step": 30100
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6107890009880066,
      "learning_rate": 0.00057508439703984,
      "loss": 2.3074,
      "step": 30101
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.7180174589157104,
      "learning_rate": 0.0005750827648523666,
      "loss": 2.1868,
      "step": 30102
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.7225481271743774,
      "learning_rate": 0.0005750811326137502,
      "loss": 2.3108,
      "step": 30103
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6181595325469971,
      "learning_rate": 0.000575079500323991,
      "loss": 2.1985,
      "step": 30104
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6756465435028076,
      "learning_rate": 0.0005750778679830894,
      "loss": 2.3186,
      "step": 30105
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.661155104637146,
      "learning_rate": 0.0005750762355910456,
      "loss": 2.2105,
      "step": 30106
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.678080677986145,
      "learning_rate": 0.00057507460314786,
      "loss": 2.2915,
      "step": 30107
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.771068811416626,
      "learning_rate": 0.0005750729706535326,
      "loss": 2.2905,
      "step": 30108
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6241371631622314,
      "learning_rate": 0.0005750713381080641,
      "loss": 2.2934,
      "step": 30109
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6050981879234314,
      "learning_rate": 0.0005750697055114547,
      "loss": 2.2668,
      "step": 30110
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.7035614848136902,
      "learning_rate": 0.0005750680728637045,
      "loss": 2.2838,
      "step": 30111
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6409116387367249,
      "learning_rate": 0.0005750664401648142,
      "loss": 2.4043,
      "step": 30112
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.7025530338287354,
      "learning_rate": 0.0005750648074147836,
      "loss": 2.6375,
      "step": 30113
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6664889454841614,
      "learning_rate": 0.0005750631746136134,
      "loss": 2.4325,
      "step": 30114
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6290243268013,
      "learning_rate": 0.0005750615417613038,
      "loss": 2.1715,
      "step": 30115
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.5954633355140686,
      "learning_rate": 0.0005750599088578549,
      "loss": 2.1888,
      "step": 30116
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.7189886569976807,
      "learning_rate": 0.0005750582759032673,
      "loss": 2.4544,
      "step": 30117
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6037909388542175,
      "learning_rate": 0.0005750566428975411,
      "loss": 2.3323,
      "step": 30118
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6574503183364868,
      "learning_rate": 0.0005750550098406767,
      "loss": 2.3066,
      "step": 30119
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6984748244285583,
      "learning_rate": 0.0005750533767326744,
      "loss": 2.3104,
      "step": 30120
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6619157791137695,
      "learning_rate": 0.0005750517435735345,
      "loss": 2.2329,
      "step": 30121
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6512411236763,
      "learning_rate": 0.0005750501103632572,
      "loss": 2.1942,
      "step": 30122
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6570216417312622,
      "learning_rate": 0.0005750484771018429,
      "loss": 2.4687,
      "step": 30123
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6578079462051392,
      "learning_rate": 0.0005750468437892921,
      "loss": 2.2859,
      "step": 30124
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6855969429016113,
      "learning_rate": 0.0005750452104256047,
      "loss": 2.2762,
      "step": 30125
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.8615889549255371,
      "learning_rate": 0.0005750435770107813,
      "loss": 2.2107,
      "step": 30126
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6538885235786438,
      "learning_rate": 0.000575041943544822,
      "loss": 2.3151,
      "step": 30127
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6214568018913269,
      "learning_rate": 0.0005750403100277273,
      "loss": 2.2424,
      "step": 30128
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6993882656097412,
      "learning_rate": 0.0005750386764594974,
      "loss": 2.3592,
      "step": 30129
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.597042977809906,
      "learning_rate": 0.0005750370428401326,
      "loss": 2.4848,
      "step": 30130
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6734188199043274,
      "learning_rate": 0.0005750354091696333,
      "loss": 2.2162,
      "step": 30131
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6147833466529846,
      "learning_rate": 0.0005750337754479996,
      "loss": 2.189,
      "step": 30132
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6304011344909668,
      "learning_rate": 0.000575032141675232,
      "loss": 2.2099,
      "step": 30133
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.657746434211731,
      "learning_rate": 0.0005750305078513307,
      "loss": 2.2975,
      "step": 30134
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.645918607711792,
      "learning_rate": 0.0005750288739762961,
      "loss": 2.2331,
      "step": 30135
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6210399866104126,
      "learning_rate": 0.0005750272400501284,
      "loss": 2.5512,
      "step": 30136
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6129852533340454,
      "learning_rate": 0.0005750256060728279,
      "loss": 2.3151,
      "step": 30137
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6513414978981018,
      "learning_rate": 0.000575023972044395,
      "loss": 2.2486,
      "step": 30138
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6255864500999451,
      "learning_rate": 0.00057502233796483,
      "loss": 2.3261,
      "step": 30139
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6485009789466858,
      "learning_rate": 0.0005750207038341331,
      "loss": 2.234,
      "step": 30140
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6280471682548523,
      "learning_rate": 0.0005750190696523046,
      "loss": 2.2552,
      "step": 30141
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6909289956092834,
      "learning_rate": 0.000575017435419345,
      "loss": 2.5414,
      "step": 30142
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6025984883308411,
      "learning_rate": 0.0005750158011352544,
      "loss": 2.3716,
      "step": 30143
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6646190285682678,
      "learning_rate": 0.0005750141668000331,
      "loss": 2.3889,
      "step": 30144
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.7475692629814148,
      "learning_rate": 0.0005750125324136816,
      "loss": 2.3715,
      "step": 30145
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.767439603805542,
      "learning_rate": 0.0005750108979762,
      "loss": 2.305,
      "step": 30146
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6363946795463562,
      "learning_rate": 0.0005750092634875887,
      "loss": 2.5189,
      "step": 30147
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6579523682594299,
      "learning_rate": 0.000575007628947848,
      "loss": 2.3921,
      "step": 30148
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.8242512345314026,
      "learning_rate": 0.0005750059943569781,
      "loss": 2.439,
      "step": 30149
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6040031313896179,
      "learning_rate": 0.0005750043597149794,
      "loss": 2.5548,
      "step": 30150
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.8016980886459351,
      "learning_rate": 0.0005750027250218524,
      "loss": 2.1938,
      "step": 30151
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.646966814994812,
      "learning_rate": 0.000575001090277597,
      "loss": 2.3769,
      "step": 30152
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6761722564697266,
      "learning_rate": 0.0005749994554822139,
      "loss": 2.1852,
      "step": 30153
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6261885166168213,
      "learning_rate": 0.000574997820635703,
      "loss": 2.3336,
      "step": 30154
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6455771327018738,
      "learning_rate": 0.0005749961857380649,
      "loss": 2.4502,
      "step": 30155
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6137776970863342,
      "learning_rate": 0.0005749945507892998,
      "loss": 2.4096,
      "step": 30156
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.669308066368103,
      "learning_rate": 0.000574992915789408,
      "loss": 2.5571,
      "step": 30157
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6348302960395813,
      "learning_rate": 0.00057499128073839,
      "loss": 2.2764,
      "step": 30158
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.5888347625732422,
      "learning_rate": 0.0005749896456362458,
      "loss": 2.4697,
      "step": 30159
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6731411218643188,
      "learning_rate": 0.0005749880104829758,
      "loss": 2.4427,
      "step": 30160
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6682611107826233,
      "learning_rate": 0.0005749863752785803,
      "loss": 2.4384,
      "step": 30161
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6918933987617493,
      "learning_rate": 0.0005749847400230597,
      "loss": 2.4337,
      "step": 30162
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6063499450683594,
      "learning_rate": 0.0005749831047164143,
      "loss": 2.4306,
      "step": 30163
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6668082475662231,
      "learning_rate": 0.0005749814693586443,
      "loss": 2.4088,
      "step": 30164
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6859844923019409,
      "learning_rate": 0.0005749798339497501,
      "loss": 2.238,
      "step": 30165
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.7553271055221558,
      "learning_rate": 0.0005749781984897319,
      "loss": 2.2846,
      "step": 30166
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6789812445640564,
      "learning_rate": 0.0005749765629785901,
      "loss": 1.8835,
      "step": 30167
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6316097378730774,
      "learning_rate": 0.000574974927416325,
      "loss": 2.2139,
      "step": 30168
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6875348091125488,
      "learning_rate": 0.0005749732918029367,
      "loss": 2.3785,
      "step": 30169
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.7012543678283691,
      "learning_rate": 0.0005749716561384259,
      "loss": 2.3101,
      "step": 30170
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.8107516765594482,
      "learning_rate": 0.0005749700204227925,
      "loss": 2.3668,
      "step": 30171
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6759005188941956,
      "learning_rate": 0.0005749683846560371,
      "loss": 2.6282,
      "step": 30172
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.9447885155677795,
      "learning_rate": 0.0005749667488381599,
      "loss": 2.252,
      "step": 30173
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6270396709442139,
      "learning_rate": 0.000574965112969161,
      "loss": 2.1875,
      "step": 30174
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6403630375862122,
      "learning_rate": 0.0005749634770490412,
      "loss": 2.259,
      "step": 30175
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6982108950614929,
      "learning_rate": 0.0005749618410778002,
      "loss": 2.3387,
      "step": 30176
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6836267709732056,
      "learning_rate": 0.0005749602050554388,
      "loss": 2.2487,
      "step": 30177
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.763663649559021,
      "learning_rate": 0.000574958568981957,
      "loss": 2.3169,
      "step": 30178
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.740202784538269,
      "learning_rate": 0.0005749569328573552,
      "loss": 2.4636,
      "step": 30179
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6281188130378723,
      "learning_rate": 0.0005749552966816338,
      "loss": 2.2057,
      "step": 30180
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.583663821220398,
      "learning_rate": 0.000574953660454793,
      "loss": 2.2552,
      "step": 30181
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.0958726406097412,
      "learning_rate": 0.0005749520241768332,
      "loss": 2.2081,
      "step": 30182
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6803581118583679,
      "learning_rate": 0.0005749503878477544,
      "loss": 2.5346,
      "step": 30183
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6716189980506897,
      "learning_rate": 0.0005749487514675574,
      "loss": 2.4018,
      "step": 30184
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6458092927932739,
      "learning_rate": 0.0005749471150362421,
      "loss": 2.2944,
      "step": 30185
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.694259524345398,
      "learning_rate": 0.0005749454785538089,
      "loss": 2.3073,
      "step": 30186
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6746082901954651,
      "learning_rate": 0.0005749438420202583,
      "loss": 2.3479,
      "step": 30187
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.5795888900756836,
      "learning_rate": 0.0005749422054355903,
      "loss": 1.929,
      "step": 30188
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6715743541717529,
      "learning_rate": 0.0005749405687998053,
      "loss": 2.1955,
      "step": 30189
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6723313927650452,
      "learning_rate": 0.0005749389321129038,
      "loss": 2.4379,
      "step": 30190
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.5906250476837158,
      "learning_rate": 0.0005749372953748859,
      "loss": 2.3664,
      "step": 30191
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6625637412071228,
      "learning_rate": 0.0005749356585857521,
      "loss": 2.236,
      "step": 30192
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.7625451683998108,
      "learning_rate": 0.0005749340217455024,
      "loss": 2.5086,
      "step": 30193
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.7209582328796387,
      "learning_rate": 0.0005749323848541374,
      "loss": 2.3159,
      "step": 30194
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6442596316337585,
      "learning_rate": 0.0005749307479116572,
      "loss": 2.2444,
      "step": 30195
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.8140220046043396,
      "learning_rate": 0.0005749291109180621,
      "loss": 2.1698,
      "step": 30196
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.5791084170341492,
      "learning_rate": 0.0005749274738733527,
      "loss": 2.4788,
      "step": 30197
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6579012870788574,
      "learning_rate": 0.000574925836777529,
      "loss": 2.4557,
      "step": 30198
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.664764404296875,
      "learning_rate": 0.0005749241996305913,
      "loss": 2.1357,
      "step": 30199
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6187741160392761,
      "learning_rate": 0.0005749225624325401,
      "loss": 2.2246,
      "step": 30200
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.7483767867088318,
      "learning_rate": 0.0005749209251833756,
      "loss": 2.5317,
      "step": 30201
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6518932580947876,
      "learning_rate": 0.000574919287883098,
      "loss": 2.1091,
      "step": 30202
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.651103675365448,
      "learning_rate": 0.0005749176505317079,
      "loss": 2.4713,
      "step": 30203
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6308247447013855,
      "learning_rate": 0.0005749160131292054,
      "loss": 2.2542,
      "step": 30204
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.5960618257522583,
      "learning_rate": 0.0005749143756755906,
      "loss": 2.1095,
      "step": 30205
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6966584920883179,
      "learning_rate": 0.0005749127381708643,
      "loss": 2.3686,
      "step": 30206
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.584467351436615,
      "learning_rate": 0.0005749111006150264,
      "loss": 2.1802,
      "step": 30207
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6167392134666443,
      "learning_rate": 0.0005749094630080773,
      "loss": 2.3178,
      "step": 30208
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6207652688026428,
      "learning_rate": 0.0005749078253500174,
      "loss": 2.4952,
      "step": 30209
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6518980860710144,
      "learning_rate": 0.0005749061876408469,
      "loss": 2.3075,
      "step": 30210
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.7304666042327881,
      "learning_rate": 0.0005749045498805662,
      "loss": 2.3295,
      "step": 30211
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6336327195167542,
      "learning_rate": 0.0005749029120691756,
      "loss": 2.4901,
      "step": 30212
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6097535490989685,
      "learning_rate": 0.0005749012742066753,
      "loss": 2.3687,
      "step": 30213
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6839411854743958,
      "learning_rate": 0.0005748996362930657,
      "loss": 2.1948,
      "step": 30214
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6646852493286133,
      "learning_rate": 0.000574897998328347,
      "loss": 2.3063,
      "step": 30215
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.5891554355621338,
      "learning_rate": 0.0005748963603125196,
      "loss": 2.1872,
      "step": 30216
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.8120322823524475,
      "learning_rate": 0.0005748947222455838,
      "loss": 2.2154,
      "step": 30217
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6677576303482056,
      "learning_rate": 0.0005748930841275398,
      "loss": 2.3195,
      "step": 30218
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6336280703544617,
      "learning_rate": 0.0005748914459583882,
      "loss": 2.3115,
      "step": 30219
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.8552976250648499,
      "learning_rate": 0.0005748898077381289,
      "loss": 2.0555,
      "step": 30220
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.7425848245620728,
      "learning_rate": 0.0005748881694667624,
      "loss": 2.3694,
      "step": 30221
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.676864504814148,
      "learning_rate": 0.0005748865311442891,
      "loss": 2.4536,
      "step": 30222
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.7139713168144226,
      "learning_rate": 0.0005748848927707091,
      "loss": 2.2744,
      "step": 30223
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.7419037818908691,
      "learning_rate": 0.0005748832543460229,
      "loss": 2.2945,
      "step": 30224
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6849862337112427,
      "learning_rate": 0.0005748816158702306,
      "loss": 2.5864,
      "step": 30225
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6915128827095032,
      "learning_rate": 0.0005748799773433327,
      "loss": 2.5604,
      "step": 30226
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6902121901512146,
      "learning_rate": 0.0005748783387653294,
      "loss": 2.4774,
      "step": 30227
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.5955599546432495,
      "learning_rate": 0.0005748767001362211,
      "loss": 2.2639,
      "step": 30228
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6152703762054443,
      "learning_rate": 0.000574875061456008,
      "loss": 2.27,
      "step": 30229
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6225623488426208,
      "learning_rate": 0.0005748734227246904,
      "loss": 2.122,
      "step": 30230
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.596601665019989,
      "learning_rate": 0.0005748717839422687,
      "loss": 2.1137,
      "step": 30231
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6411452293395996,
      "learning_rate": 0.0005748701451087431,
      "loss": 2.3142,
      "step": 30232
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.5860279202461243,
      "learning_rate": 0.0005748685062241139,
      "loss": 2.204,
      "step": 30233
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.8003945350646973,
      "learning_rate": 0.0005748668672883816,
      "loss": 2.2476,
      "step": 30234
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6328306794166565,
      "learning_rate": 0.0005748652283015461,
      "loss": 2.278,
      "step": 30235
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.7850856781005859,
      "learning_rate": 0.0005748635892636083,
      "loss": 2.4049,
      "step": 30236
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.7349221110343933,
      "learning_rate": 0.0005748619501745679,
      "loss": 2.2372,
      "step": 30237
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.655764102935791,
      "learning_rate": 0.0005748603110344256,
      "loss": 2.5352,
      "step": 30238
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6879677772521973,
      "learning_rate": 0.0005748586718431816,
      "loss": 2.2407,
      "step": 30239
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6679127812385559,
      "learning_rate": 0.0005748570326008361,
      "loss": 2.2996,
      "step": 30240
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.7587930560112,
      "learning_rate": 0.0005748553933073896,
      "loss": 2.1879,
      "step": 30241
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6025793552398682,
      "learning_rate": 0.0005748537539628422,
      "loss": 2.4574,
      "step": 30242
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6549901962280273,
      "learning_rate": 0.0005748521145671943,
      "loss": 2.1713,
      "step": 30243
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6715293526649475,
      "learning_rate": 0.0005748504751204463,
      "loss": 2.4493,
      "step": 30244
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.648828387260437,
      "learning_rate": 0.0005748488356225983,
      "loss": 2.3208,
      "step": 30245
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6485559344291687,
      "learning_rate": 0.0005748471960736508,
      "loss": 2.3836,
      "step": 30246
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6304303407669067,
      "learning_rate": 0.000574845556473604,
      "loss": 2.1733,
      "step": 30247
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6274328231811523,
      "learning_rate": 0.0005748439168224581,
      "loss": 2.2456,
      "step": 30248
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6125553250312805,
      "learning_rate": 0.0005748422771202136,
      "loss": 2.1181,
      "step": 30249
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6227473616600037,
      "learning_rate": 0.0005748406373668708,
      "loss": 2.2024,
      "step": 30250
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6902329921722412,
      "learning_rate": 0.00057483899756243,
      "loss": 2.2998,
      "step": 30251
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.615557849407196,
      "learning_rate": 0.0005748373577068913,
      "loss": 2.4281,
      "step": 30252
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6673429608345032,
      "learning_rate": 0.0005748357178002551,
      "loss": 2.2853,
      "step": 30253
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.7065671682357788,
      "learning_rate": 0.0005748340778425218,
      "loss": 2.6128,
      "step": 30254
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6807841062545776,
      "learning_rate": 0.0005748324378336917,
      "loss": 2.2961,
      "step": 30255
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6205148696899414,
      "learning_rate": 0.0005748307977737652,
      "loss": 2.2976,
      "step": 30256
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6415671110153198,
      "learning_rate": 0.0005748291576627421,
      "loss": 2.3892,
      "step": 30257
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.618338942527771,
      "learning_rate": 0.0005748275175006234,
      "loss": 2.2857,
      "step": 30258
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6220616102218628,
      "learning_rate": 0.0005748258772874089,
      "loss": 2.1567,
      "step": 30259
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6670863032341003,
      "learning_rate": 0.0005748242370230992,
      "loss": 2.1707,
      "step": 30260
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6135380864143372,
      "learning_rate": 0.0005748225967076944,
      "loss": 2.326,
      "step": 30261
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6706348657608032,
      "learning_rate": 0.000574820956341195,
      "loss": 2.5095,
      "step": 30262
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6269475817680359,
      "learning_rate": 0.0005748193159236011,
      "loss": 2.2497,
      "step": 30263
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.653655469417572,
      "learning_rate": 0.0005748176754549131,
      "loss": 2.3079,
      "step": 30264
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.7145944237709045,
      "learning_rate": 0.0005748160349351313,
      "loss": 2.3826,
      "step": 30265
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6575790643692017,
      "learning_rate": 0.0005748143943642561,
      "loss": 2.2752,
      "step": 30266
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6408429145812988,
      "learning_rate": 0.0005748127537422876,
      "loss": 2.2596,
      "step": 30267
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6320503354072571,
      "learning_rate": 0.0005748111130692263,
      "loss": 2.3771,
      "step": 30268
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.7919650673866272,
      "learning_rate": 0.0005748094723450725,
      "loss": 2.3807,
      "step": 30269
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6530030369758606,
      "learning_rate": 0.0005748078315698263,
      "loss": 2.2276,
      "step": 30270
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.651296079158783,
      "learning_rate": 0.0005748061907434881,
      "loss": 2.1931,
      "step": 30271
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.9054716229438782,
      "learning_rate": 0.0005748045498660584,
      "loss": 2.2371,
      "step": 30272
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.7251612544059753,
      "learning_rate": 0.0005748029089375373,
      "loss": 2.4117,
      "step": 30273
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.7374428510665894,
      "learning_rate": 0.0005748012679579252,
      "loss": 2.3902,
      "step": 30274
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.7374331951141357,
      "learning_rate": 0.0005747996269272222,
      "loss": 2.4455,
      "step": 30275
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6611601710319519,
      "learning_rate": 0.0005747979858454288,
      "loss": 2.0783,
      "step": 30276
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6459950804710388,
      "learning_rate": 0.0005747963447125453,
      "loss": 2.2898,
      "step": 30277
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6294000148773193,
      "learning_rate": 0.000574794703528572,
      "loss": 2.1544,
      "step": 30278
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.65814208984375,
      "learning_rate": 0.0005747930622935092,
      "loss": 2.5107,
      "step": 30279
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6253845691680908,
      "learning_rate": 0.0005747914210073571,
      "loss": 2.2888,
      "step": 30280
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.5916144847869873,
      "learning_rate": 0.0005747897796701161,
      "loss": 2.2472,
      "step": 30281
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.7087454795837402,
      "learning_rate": 0.0005747881382817865,
      "loss": 2.2303,
      "step": 30282
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6628603935241699,
      "learning_rate": 0.0005747864968423687,
      "loss": 2.3525,
      "step": 30283
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6241610646247864,
      "learning_rate": 0.0005747848553518627,
      "loss": 2.3329,
      "step": 30284
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.7277154922485352,
      "learning_rate": 0.0005747832138102691,
      "loss": 2.3124,
      "step": 30285
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.5765339136123657,
      "learning_rate": 0.0005747815722175882,
      "loss": 2.3324,
      "step": 30286
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.653823971748352,
      "learning_rate": 0.0005747799305738202,
      "loss": 2.4575,
      "step": 30287
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.5943088531494141,
      "learning_rate": 0.0005747782888789654,
      "loss": 2.2039,
      "step": 30288
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6252635717391968,
      "learning_rate": 0.0005747766471330242,
      "loss": 2.1456,
      "step": 30289
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6392985582351685,
      "learning_rate": 0.0005747750053359966,
      "loss": 2.1224,
      "step": 30290
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6125579476356506,
      "learning_rate": 0.0005747733634878833,
      "loss": 2.3219,
      "step": 30291
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.609323263168335,
      "learning_rate": 0.0005747717215886845,
      "loss": 2.2077,
      "step": 30292
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6105009913444519,
      "learning_rate": 0.0005747700796384004,
      "loss": 2.3984,
      "step": 30293
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6867361068725586,
      "learning_rate": 0.0005747684376370312,
      "loss": 2.2969,
      "step": 30294
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6905669569969177,
      "learning_rate": 0.0005747667955845777,
      "loss": 2.458,
      "step": 30295
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6481603384017944,
      "learning_rate": 0.0005747651534810396,
      "loss": 2.2612,
      "step": 30296
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6168192028999329,
      "learning_rate": 0.0005747635113264176,
      "loss": 2.709,
      "step": 30297
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6216854453086853,
      "learning_rate": 0.0005747618691207117,
      "loss": 2.1151,
      "step": 30298
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6606526970863342,
      "learning_rate": 0.0005747602268639225,
      "loss": 2.3024,
      "step": 30299
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6480032205581665,
      "learning_rate": 0.0005747585845560502,
      "loss": 2.2911,
      "step": 30300
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.7081460952758789,
      "learning_rate": 0.000574756942197095,
      "loss": 2.3692,
      "step": 30301
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6292390823364258,
      "learning_rate": 0.0005747552997870575,
      "loss": 2.3998,
      "step": 30302
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6708055138587952,
      "learning_rate": 0.0005747536573259376,
      "loss": 2.2968,
      "step": 30303
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.633573591709137,
      "learning_rate": 0.000574752014813736,
      "loss": 2.204,
      "step": 30304
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6640349626541138,
      "learning_rate": 0.0005747503722504526,
      "loss": 2.2348,
      "step": 30305
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6680883765220642,
      "learning_rate": 0.000574748729636088,
      "loss": 2.2543,
      "step": 30306
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6144492030143738,
      "learning_rate": 0.0005747470869706425,
      "loss": 2.393,
      "step": 30307
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6444442868232727,
      "learning_rate": 0.0005747454442541162,
      "loss": 2.4029,
      "step": 30308
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6688840985298157,
      "learning_rate": 0.0005747438014865096,
      "loss": 2.3099,
      "step": 30309
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6035844683647156,
      "learning_rate": 0.0005747421586678231,
      "loss": 2.2441,
      "step": 30310
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.5987590551376343,
      "learning_rate": 0.0005747405157980566,
      "loss": 2.1358,
      "step": 30311
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6377812623977661,
      "learning_rate": 0.0005747388728772108,
      "loss": 2.3497,
      "step": 30312
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.5946065187454224,
      "learning_rate": 0.0005747372299052858,
      "loss": 2.1778,
      "step": 30313
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6057883501052856,
      "learning_rate": 0.000574735586882282,
      "loss": 2.2116,
      "step": 30314
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6238240599632263,
      "learning_rate": 0.0005747339438081996,
      "loss": 2.1683,
      "step": 30315
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.601810872554779,
      "learning_rate": 0.000574732300683039,
      "loss": 2.4271,
      "step": 30316
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6656852960586548,
      "learning_rate": 0.0005747306575068005,
      "loss": 2.1853,
      "step": 30317
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6312374472618103,
      "learning_rate": 0.0005747290142794843,
      "loss": 2.4087,
      "step": 30318
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6495707035064697,
      "learning_rate": 0.0005747273710010909,
      "loss": 2.4018,
      "step": 30319
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.658547043800354,
      "learning_rate": 0.0005747257276716205,
      "loss": 2.2723,
      "step": 30320
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.611998975276947,
      "learning_rate": 0.0005747240842910733,
      "loss": 2.0699,
      "step": 30321
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6462926864624023,
      "learning_rate": 0.0005747224408594498,
      "loss": 2.0962,
      "step": 30322
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.7381742596626282,
      "learning_rate": 0.0005747207973767503,
      "loss": 2.3459,
      "step": 30323
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.658690333366394,
      "learning_rate": 0.0005747191538429748,
      "loss": 2.2954,
      "step": 30324
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.7568420171737671,
      "learning_rate": 0.0005747175102581239,
      "loss": 2.1145,
      "step": 30325
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6231062412261963,
      "learning_rate": 0.0005747158666221979,
      "loss": 2.4249,
      "step": 30326
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6806812286376953,
      "learning_rate": 0.000574714222935197,
      "loss": 2.5436,
      "step": 30327
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6567290425300598,
      "learning_rate": 0.0005747125791971215,
      "loss": 2.2294,
      "step": 30328
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6078907251358032,
      "learning_rate": 0.0005747109354079718,
      "loss": 2.4263,
      "step": 30329
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6092864274978638,
      "learning_rate": 0.0005747092915677481,
      "loss": 2.3492,
      "step": 30330
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6494137644767761,
      "learning_rate": 0.0005747076476764508,
      "loss": 2.3858,
      "step": 30331
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.8144659399986267,
      "learning_rate": 0.0005747060037340801,
      "loss": 2.3123,
      "step": 30332
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.7060120701789856,
      "learning_rate": 0.0005747043597406365,
      "loss": 2.2494,
      "step": 30333
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6059406995773315,
      "learning_rate": 0.0005747027156961201,
      "loss": 2.2875,
      "step": 30334
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6436241269111633,
      "learning_rate": 0.0005747010716005312,
      "loss": 2.2702,
      "step": 30335
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6297721862792969,
      "learning_rate": 0.0005746994274538703,
      "loss": 2.2333,
      "step": 30336
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6135494709014893,
      "learning_rate": 0.0005746977832561376,
      "loss": 2.2369,
      "step": 30337
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6242430210113525,
      "learning_rate": 0.0005746961390073333,
      "loss": 2.1792,
      "step": 30338
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6399765014648438,
      "learning_rate": 0.0005746944947074579,
      "loss": 2.4476,
      "step": 30339
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.682893693447113,
      "learning_rate": 0.0005746928503565115,
      "loss": 2.351,
      "step": 30340
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.608643114566803,
      "learning_rate": 0.0005746912059544946,
      "loss": 2.3682,
      "step": 30341
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6949208974838257,
      "learning_rate": 0.0005746895615014074,
      "loss": 2.2887,
      "step": 30342
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.7318418025970459,
      "learning_rate": 0.0005746879169972501,
      "loss": 2.4471,
      "step": 30343
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.5994358658790588,
      "learning_rate": 0.0005746862724420233,
      "loss": 1.8623,
      "step": 30344
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.622447669506073,
      "learning_rate": 0.0005746846278357271,
      "loss": 2.313,
      "step": 30345
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.7190094590187073,
      "learning_rate": 0.0005746829831783617,
      "loss": 2.3529,
      "step": 30346
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6608619093894958,
      "learning_rate": 0.0005746813384699277,
      "loss": 2.3149,
      "step": 30347
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6566250920295715,
      "learning_rate": 0.0005746796937104253,
      "loss": 2.2507,
      "step": 30348
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6438612937927246,
      "learning_rate": 0.0005746780488998547,
      "loss": 2.3023,
      "step": 30349
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6340318918228149,
      "learning_rate": 0.0005746764040382162,
      "loss": 2.3782,
      "step": 30350
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.636193037033081,
      "learning_rate": 0.0005746747591255102,
      "loss": 2.2957,
      "step": 30351
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6448231339454651,
      "learning_rate": 0.0005746731141617371,
      "loss": 2.4459,
      "step": 30352
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.7623216509819031,
      "learning_rate": 0.0005746714691468969,
      "loss": 2.0857,
      "step": 30353
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.7415675520896912,
      "learning_rate": 0.0005746698240809902,
      "loss": 2.4186,
      "step": 30354
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.5980328917503357,
      "learning_rate": 0.0005746681789640172,
      "loss": 2.2571,
      "step": 30355
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6659983992576599,
      "learning_rate": 0.0005746665337959782,
      "loss": 2.2616,
      "step": 30356
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6151237487792969,
      "learning_rate": 0.0005746648885768733,
      "loss": 2.2228,
      "step": 30357
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6036814451217651,
      "learning_rate": 0.0005746632433067032,
      "loss": 2.2925,
      "step": 30358
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6887078881263733,
      "learning_rate": 0.0005746615979854681,
      "loss": 2.2744,
      "step": 30359
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6005997061729431,
      "learning_rate": 0.0005746599526131682,
      "loss": 2.2749,
      "step": 30360
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6841828227043152,
      "learning_rate": 0.0005746583071898037,
      "loss": 2.4233,
      "step": 30361
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.7025030851364136,
      "learning_rate": 0.0005746566617153751,
      "loss": 2.0682,
      "step": 30362
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6693009734153748,
      "learning_rate": 0.0005746550161898826,
      "loss": 2.2798,
      "step": 30363
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6326125860214233,
      "learning_rate": 0.0005746533706133265,
      "loss": 2.2977,
      "step": 30364
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6901618242263794,
      "learning_rate": 0.0005746517249857073,
      "loss": 2.6096,
      "step": 30365
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6258352398872375,
      "learning_rate": 0.0005746500793070251,
      "loss": 2.4812,
      "step": 30366
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6749882698059082,
      "learning_rate": 0.0005746484335772804,
      "loss": 2.1956,
      "step": 30367
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6889775991439819,
      "learning_rate": 0.0005746467877964731,
      "loss": 2.4178,
      "step": 30368
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6534614562988281,
      "learning_rate": 0.000574645141964604,
      "loss": 2.3728,
      "step": 30369
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.7938738465309143,
      "learning_rate": 0.000574643496081673,
      "loss": 2.3223,
      "step": 30370
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6635642647743225,
      "learning_rate": 0.0005746418501476807,
      "loss": 2.4494,
      "step": 30371
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.5993123054504395,
      "learning_rate": 0.0005746402041626273,
      "loss": 2.3759,
      "step": 30372
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6975660920143127,
      "learning_rate": 0.000574638558126513,
      "loss": 2.1349,
      "step": 30373
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6566330194473267,
      "learning_rate": 0.0005746369120393384,
      "loss": 2.2515,
      "step": 30374
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6405287384986877,
      "learning_rate": 0.0005746352659011035,
      "loss": 2.4989,
      "step": 30375
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.7454926371574402,
      "learning_rate": 0.0005746336197118087,
      "loss": 2.4567,
      "step": 30376
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.7615500092506409,
      "learning_rate": 0.0005746319734714544,
      "loss": 2.381,
      "step": 30377
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6181784868240356,
      "learning_rate": 0.0005746303271800408,
      "loss": 2.3063,
      "step": 30378
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6185728311538696,
      "learning_rate": 0.0005746286808375682,
      "loss": 2.4476,
      "step": 30379
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6389281153678894,
      "learning_rate": 0.0005746270344440369,
      "loss": 2.409,
      "step": 30380
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6559807658195496,
      "learning_rate": 0.0005746253879994474,
      "loss": 2.3738,
      "step": 30381
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6532807350158691,
      "learning_rate": 0.0005746237415037998,
      "loss": 2.2425,
      "step": 30382
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6261717081069946,
      "learning_rate": 0.0005746220949570944,
      "loss": 2.2702,
      "step": 30383
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6763250827789307,
      "learning_rate": 0.0005746204483593316,
      "loss": 2.058,
      "step": 30384
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6463891863822937,
      "learning_rate": 0.0005746188017105117,
      "loss": 2.4617,
      "step": 30385
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.7008724808692932,
      "learning_rate": 0.0005746171550106349,
      "loss": 2.3331,
      "step": 30386
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.630010187625885,
      "learning_rate": 0.0005746155082597017,
      "loss": 2.4046,
      "step": 30387
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6615968346595764,
      "learning_rate": 0.0005746138614577123,
      "loss": 2.4663,
      "step": 30388
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.7421820759773254,
      "learning_rate": 0.0005746122146046669,
      "loss": 2.333,
      "step": 30389
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6373552680015564,
      "learning_rate": 0.000574610567700566,
      "loss": 2.3268,
      "step": 30390
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6541493535041809,
      "learning_rate": 0.0005746089207454096,
      "loss": 2.294,
      "step": 30391
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6738887429237366,
      "learning_rate": 0.0005746072737391985,
      "loss": 2.5782,
      "step": 30392
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6420307159423828,
      "learning_rate": 0.0005746056266819326,
      "loss": 2.423,
      "step": 30393
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6860201358795166,
      "learning_rate": 0.0005746039795736124,
      "loss": 2.0815,
      "step": 30394
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6452472805976868,
      "learning_rate": 0.000574602332414238,
      "loss": 2.2117,
      "step": 30395
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6162142157554626,
      "learning_rate": 0.0005746006852038099,
      "loss": 2.329,
      "step": 30396
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6342434883117676,
      "learning_rate": 0.0005745990379423284,
      "loss": 2.3341,
      "step": 30397
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6702728271484375,
      "learning_rate": 0.0005745973906297936,
      "loss": 2.3279,
      "step": 30398
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6443105936050415,
      "learning_rate": 0.0005745957432662061,
      "loss": 2.2578,
      "step": 30399
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6365202069282532,
      "learning_rate": 0.000574594095851566,
      "loss": 2.4953,
      "step": 30400
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.7101220488548279,
      "learning_rate": 0.0005745924483858737,
      "loss": 2.3963,
      "step": 30401
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.7165387868881226,
      "learning_rate": 0.0005745908008691295,
      "loss": 2.3624,
      "step": 30402
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6345232725143433,
      "learning_rate": 0.0005745891533013338,
      "loss": 2.3383,
      "step": 30403
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.5852439999580383,
      "learning_rate": 0.0005745875056824867,
      "loss": 2.1861,
      "step": 30404
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6008633375167847,
      "learning_rate": 0.0005745858580125886,
      "loss": 2.0797,
      "step": 30405
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6648377776145935,
      "learning_rate": 0.0005745842102916397,
      "loss": 2.2627,
      "step": 30406
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6445051431655884,
      "learning_rate": 0.0005745825625196406,
      "loss": 2.2445,
      "step": 30407
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6659411787986755,
      "learning_rate": 0.0005745809146965912,
      "loss": 2.0835,
      "step": 30408
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.636095404624939,
      "learning_rate": 0.0005745792668224921,
      "loss": 2.1199,
      "step": 30409
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6649637222290039,
      "learning_rate": 0.0005745776188973437,
      "loss": 2.4031,
      "step": 30410
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.7367107272148132,
      "learning_rate": 0.0005745759709211459,
      "loss": 2.213,
      "step": 30411
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6894656419754028,
      "learning_rate": 0.0005745743228938994,
      "loss": 2.3421,
      "step": 30412
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6901484131813049,
      "learning_rate": 0.0005745726748156044,
      "loss": 2.3727,
      "step": 30413
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.7525221109390259,
      "learning_rate": 0.0005745710266862609,
      "loss": 2.1678,
      "step": 30414
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6939774751663208,
      "learning_rate": 0.0005745693785058697,
      "loss": 2.3905,
      "step": 30415
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6634995341300964,
      "learning_rate": 0.0005745677302744308,
      "loss": 2.0925,
      "step": 30416
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6374601125717163,
      "learning_rate": 0.0005745660819919446,
      "loss": 2.3738,
      "step": 30417
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6885788440704346,
      "learning_rate": 0.0005745644336584113,
      "loss": 2.2051,
      "step": 30418
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6335166692733765,
      "learning_rate": 0.0005745627852738313,
      "loss": 2.363,
      "step": 30419
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6500526666641235,
      "learning_rate": 0.000574561136838205,
      "loss": 2.3237,
      "step": 30420
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6497078537940979,
      "learning_rate": 0.0005745594883515326,
      "loss": 2.3436,
      "step": 30421
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.690294623374939,
      "learning_rate": 0.0005745578398138143,
      "loss": 2.0615,
      "step": 30422
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6924198269844055,
      "learning_rate": 0.0005745561912250505,
      "loss": 2.2988,
      "step": 30423
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6656761765480042,
      "learning_rate": 0.0005745545425852416,
      "loss": 2.3558,
      "step": 30424
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.7166184782981873,
      "learning_rate": 0.0005745528938943879,
      "loss": 2.333,
      "step": 30425
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.7691289186477661,
      "learning_rate": 0.0005745512451524895,
      "loss": 2.2161,
      "step": 30426
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6247862577438354,
      "learning_rate": 0.0005745495963595469,
      "loss": 2.096,
      "step": 30427
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.7708715796470642,
      "learning_rate": 0.0005745479475155603,
      "loss": 2.2862,
      "step": 30428
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.7924925088882446,
      "learning_rate": 0.0005745462986205301,
      "loss": 2.3126,
      "step": 30429
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6294156312942505,
      "learning_rate": 0.0005745446496744566,
      "loss": 2.2027,
      "step": 30430
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.7594109177589417,
      "learning_rate": 0.00057454300067734,
      "loss": 2.0928,
      "step": 30431
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.792750358581543,
      "learning_rate": 0.0005745413516291807,
      "loss": 2.2471,
      "step": 30432
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.7996000051498413,
      "learning_rate": 0.000574539702529979,
      "loss": 2.6548,
      "step": 30433
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.7901574969291687,
      "learning_rate": 0.0005745380533797351,
      "loss": 2.2473,
      "step": 30434
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.7296348214149475,
      "learning_rate": 0.0005745364041784494,
      "loss": 2.0679,
      "step": 30435
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6685686111450195,
      "learning_rate": 0.0005745347549261223,
      "loss": 2.3312,
      "step": 30436
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.7715604901313782,
      "learning_rate": 0.0005745331056227539,
      "loss": 2.3205,
      "step": 30437
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.7415469288825989,
      "learning_rate": 0.0005745314562683447,
      "loss": 2.4687,
      "step": 30438
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6781367063522339,
      "learning_rate": 0.0005745298068628949,
      "loss": 2.3473,
      "step": 30439
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.7005327939987183,
      "learning_rate": 0.0005745281574064049,
      "loss": 2.4188,
      "step": 30440
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6345427632331848,
      "learning_rate": 0.0005745265078988749,
      "loss": 2.3906,
      "step": 30441
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.7069318294525146,
      "learning_rate": 0.0005745248583403051,
      "loss": 2.4555,
      "step": 30442
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6558820009231567,
      "learning_rate": 0.000574523208730696,
      "loss": 2.3158,
      "step": 30443
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6108163595199585,
      "learning_rate": 0.0005745215590700479,
      "loss": 2.3889,
      "step": 30444
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6697880625724792,
      "learning_rate": 0.0005745199093583611,
      "loss": 2.4441,
      "step": 30445
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6146711111068726,
      "learning_rate": 0.0005745182595956358,
      "loss": 2.2083,
      "step": 30446
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.7810991406440735,
      "learning_rate": 0.0005745166097818724,
      "loss": 2.3213,
      "step": 30447
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6535294055938721,
      "learning_rate": 0.0005745149599170711,
      "loss": 2.2577,
      "step": 30448
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6270866394042969,
      "learning_rate": 0.0005745133100012324,
      "loss": 2.2023,
      "step": 30449
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6519522666931152,
      "learning_rate": 0.0005745116600343564,
      "loss": 2.3669,
      "step": 30450
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.7335303425788879,
      "learning_rate": 0.0005745100100164436,
      "loss": 2.4137,
      "step": 30451
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6574758887290955,
      "learning_rate": 0.0005745083599474942,
      "loss": 2.4538,
      "step": 30452
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.592981219291687,
      "learning_rate": 0.0005745067098275084,
      "loss": 2.0771,
      "step": 30453
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6570961475372314,
      "learning_rate": 0.0005745050596564868,
      "loss": 2.2993,
      "step": 30454
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6226920485496521,
      "learning_rate": 0.0005745034094344294,
      "loss": 2.4125,
      "step": 30455
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6844644546508789,
      "learning_rate": 0.0005745017591613366,
      "loss": 2.3047,
      "step": 30456
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6072752475738525,
      "learning_rate": 0.0005745001088372088,
      "loss": 2.3357,
      "step": 30457
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6503261923789978,
      "learning_rate": 0.0005744984584620463,
      "loss": 2.3692,
      "step": 30458
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.7090410590171814,
      "learning_rate": 0.0005744968080358492,
      "loss": 2.6533,
      "step": 30459
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6212301254272461,
      "learning_rate": 0.0005744951575586181,
      "loss": 2.2319,
      "step": 30460
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6896753311157227,
      "learning_rate": 0.0005744935070303532,
      "loss": 2.2232,
      "step": 30461
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6589064002037048,
      "learning_rate": 0.0005744918564510546,
      "loss": 2.5042,
      "step": 30462
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6697238087654114,
      "learning_rate": 0.000574490205820723,
      "loss": 2.2635,
      "step": 30463
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.671253502368927,
      "learning_rate": 0.0005744885551393584,
      "loss": 2.1463,
      "step": 30464
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6529405117034912,
      "learning_rate": 0.0005744869044069611,
      "loss": 2.3147,
      "step": 30465
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6236769556999207,
      "learning_rate": 0.0005744852536235316,
      "loss": 2.4286,
      "step": 30466
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.7075810432434082,
      "learning_rate": 0.0005744836027890702,
      "loss": 2.3246,
      "step": 30467
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6726329326629639,
      "learning_rate": 0.0005744819519035769,
      "loss": 2.4571,
      "step": 30468
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.5891017317771912,
      "learning_rate": 0.0005744803009670523,
      "loss": 2.16,
      "step": 30469
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6311726570129395,
      "learning_rate": 0.0005744786499794967,
      "loss": 2.2223,
      "step": 30470
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6482125520706177,
      "learning_rate": 0.0005744769989409103,
      "loss": 2.2767,
      "step": 30471
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6020681858062744,
      "learning_rate": 0.0005744753478512934,
      "loss": 2.1569,
      "step": 30472
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6537536382675171,
      "learning_rate": 0.0005744736967106465,
      "loss": 2.1892,
      "step": 30473
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.709411084651947,
      "learning_rate": 0.0005744720455189696,
      "loss": 2.3828,
      "step": 30474
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.7082326412200928,
      "learning_rate": 0.0005744703942762632,
      "loss": 2.3415,
      "step": 30475
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6352560520172119,
      "learning_rate": 0.0005744687429825277,
      "loss": 2.3892,
      "step": 30476
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.7170787453651428,
      "learning_rate": 0.0005744670916377631,
      "loss": 2.1324,
      "step": 30477
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.7029104828834534,
      "learning_rate": 0.00057446544024197,
      "loss": 2.3466,
      "step": 30478
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.596823513507843,
      "learning_rate": 0.0005744637887951485,
      "loss": 2.2402,
      "step": 30479
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6963446736335754,
      "learning_rate": 0.000574462137297299,
      "loss": 2.1627,
      "step": 30480
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.5896899104118347,
      "learning_rate": 0.0005744604857484219,
      "loss": 2.2993,
      "step": 30481
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6887056231498718,
      "learning_rate": 0.0005744588341485174,
      "loss": 2.2891,
      "step": 30482
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6604493856430054,
      "learning_rate": 0.0005744571824975858,
      "loss": 2.4326,
      "step": 30483
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6189132332801819,
      "learning_rate": 0.0005744555307956274,
      "loss": 2.457,
      "step": 30484
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.5988743901252747,
      "learning_rate": 0.0005744538790426425,
      "loss": 2.5001,
      "step": 30485
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6099427938461304,
      "learning_rate": 0.0005744522272386316,
      "loss": 2.1635,
      "step": 30486
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6882455945014954,
      "learning_rate": 0.0005744505753835947,
      "loss": 2.3506,
      "step": 30487
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6572281122207642,
      "learning_rate": 0.0005744489234775324,
      "loss": 2.1186,
      "step": 30488
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6282869577407837,
      "learning_rate": 0.0005744472715204448,
      "loss": 2.2272,
      "step": 30489
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.671082615852356,
      "learning_rate": 0.0005744456195123322,
      "loss": 2.1803,
      "step": 30490
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6533429622650146,
      "learning_rate": 0.000574443967453195,
      "loss": 2.415,
      "step": 30491
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6436111330986023,
      "learning_rate": 0.0005744423153430336,
      "loss": 2.099,
      "step": 30492
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6546927690505981,
      "learning_rate": 0.000574440663181848,
      "loss": 2.2955,
      "step": 30493
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6800495982170105,
      "learning_rate": 0.0005744390109696389,
      "loss": 2.3407,
      "step": 30494
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6056201457977295,
      "learning_rate": 0.0005744373587064063,
      "loss": 2.258,
      "step": 30495
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6636220216751099,
      "learning_rate": 0.0005744357063921506,
      "loss": 2.3278,
      "step": 30496
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6076624393463135,
      "learning_rate": 0.0005744340540268721,
      "loss": 2.4978,
      "step": 30497
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.7164968848228455,
      "learning_rate": 0.0005744324016105712,
      "loss": 2.3453,
      "step": 30498
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6342151761054993,
      "learning_rate": 0.0005744307491432481,
      "loss": 2.3488,
      "step": 30499
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6669543385505676,
      "learning_rate": 0.0005744290966249033,
      "loss": 2.5133,
      "step": 30500
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.615949273109436,
      "learning_rate": 0.0005744274440555368,
      "loss": 2.4439,
      "step": 30501
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6482452154159546,
      "learning_rate": 0.000574425791435149,
      "loss": 2.2165,
      "step": 30502
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.8187167048454285,
      "learning_rate": 0.0005744241387637403,
      "loss": 2.6378,
      "step": 30503
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.628395140171051,
      "learning_rate": 0.000574422486041311,
      "loss": 2.1653,
      "step": 30504
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.7107965350151062,
      "learning_rate": 0.0005744208332678615,
      "loss": 2.1408,
      "step": 30505
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.698173463344574,
      "learning_rate": 0.0005744191804433918,
      "loss": 2.4219,
      "step": 30506
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6591811776161194,
      "learning_rate": 0.0005744175275679025,
      "loss": 2.3024,
      "step": 30507
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.7271239161491394,
      "learning_rate": 0.0005744158746413938,
      "loss": 2.3402,
      "step": 30508
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6535618901252747,
      "learning_rate": 0.000574414221663866,
      "loss": 2.2487,
      "step": 30509
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6710436344146729,
      "learning_rate": 0.0005744125686353193,
      "loss": 2.3264,
      "step": 30510
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6124226450920105,
      "learning_rate": 0.0005744109155557543,
      "loss": 2.1881,
      "step": 30511
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.5675469040870667,
      "learning_rate": 0.0005744092624251709,
      "loss": 2.3647,
      "step": 30512
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.5793419480323792,
      "learning_rate": 0.0005744076092435699,
      "loss": 2.3153,
      "step": 30513
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.5711320042610168,
      "learning_rate": 0.0005744059560109512,
      "loss": 2.3669,
      "step": 30514
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.7067462801933289,
      "learning_rate": 0.0005744043027273154,
      "loss": 2.3223,
      "step": 30515
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6587713360786438,
      "learning_rate": 0.0005744026493926625,
      "loss": 2.2135,
      "step": 30516
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6661335229873657,
      "learning_rate": 0.000574400996006993,
      "loss": 2.1728,
      "step": 30517
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6755728721618652,
      "learning_rate": 0.0005743993425703072,
      "loss": 2.2152,
      "step": 30518
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.5906411409378052,
      "learning_rate": 0.0005743976890826054,
      "loss": 2.3646,
      "step": 30519
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6803874969482422,
      "learning_rate": 0.0005743960355438877,
      "loss": 2.415,
      "step": 30520
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6782050132751465,
      "learning_rate": 0.0005743943819541548,
      "loss": 2.3101,
      "step": 30521
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6363824605941772,
      "learning_rate": 0.0005743927283134068,
      "loss": 2.1892,
      "step": 30522
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6267421245574951,
      "learning_rate": 0.000574391074621644,
      "loss": 2.1692,
      "step": 30523
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6170772910118103,
      "learning_rate": 0.0005743894208788667,
      "loss": 2.2585,
      "step": 30524
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.7105995416641235,
      "learning_rate": 0.0005743877670850752,
      "loss": 2.3514,
      "step": 30525
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.5887627601623535,
      "learning_rate": 0.0005743861132402699,
      "loss": 2.3519,
      "step": 30526
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.624298095703125,
      "learning_rate": 0.000574384459344451,
      "loss": 2.261,
      "step": 30527
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.600307047367096,
      "learning_rate": 0.0005743828053976188,
      "loss": 2.1908,
      "step": 30528
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6473154425621033,
      "learning_rate": 0.0005743811513997738,
      "loss": 2.5917,
      "step": 30529
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6174888014793396,
      "learning_rate": 0.0005743794973509161,
      "loss": 2.1401,
      "step": 30530
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6736774444580078,
      "learning_rate": 0.000574377843251046,
      "loss": 2.1587,
      "step": 30531
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.7357923984527588,
      "learning_rate": 0.0005743761891001638,
      "loss": 2.2302,
      "step": 30532
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6300590634346008,
      "learning_rate": 0.0005743745348982701,
      "loss": 2.2572,
      "step": 30533
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.636393129825592,
      "learning_rate": 0.0005743728806453649,
      "loss": 2.313,
      "step": 30534
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.7251119017601013,
      "learning_rate": 0.0005743712263414486,
      "loss": 2.2139,
      "step": 30535
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6454655528068542,
      "learning_rate": 0.0005743695719865215,
      "loss": 2.3237,
      "step": 30536
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6578091382980347,
      "learning_rate": 0.0005743679175805839,
      "loss": 2.4305,
      "step": 30537
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.60800701379776,
      "learning_rate": 0.0005743662631236363,
      "loss": 2.3674,
      "step": 30538
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6533193588256836,
      "learning_rate": 0.0005743646086156786,
      "loss": 2.3835,
      "step": 30539
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.61873459815979,
      "learning_rate": 0.0005743629540567114,
      "loss": 2.5485,
      "step": 30540
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6640404462814331,
      "learning_rate": 0.000574361299446735,
      "loss": 2.2264,
      "step": 30541
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.705348551273346,
      "learning_rate": 0.0005743596447857497,
      "loss": 2.1717,
      "step": 30542
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6457149982452393,
      "learning_rate": 0.0005743579900737557,
      "loss": 2.3884,
      "step": 30543
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6765996217727661,
      "learning_rate": 0.0005743563353107533,
      "loss": 2.2148,
      "step": 30544
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6502789258956909,
      "learning_rate": 0.000574354680496743,
      "loss": 2.3826,
      "step": 30545
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.8027374744415283,
      "learning_rate": 0.0005743530256317249,
      "loss": 2.143,
      "step": 30546
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.684627115726471,
      "learning_rate": 0.0005743513707156994,
      "loss": 2.1441,
      "step": 30547
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.7759801149368286,
      "learning_rate": 0.0005743497157486669,
      "loss": 2.3844,
      "step": 30548
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.8227739930152893,
      "learning_rate": 0.0005743480607306275,
      "loss": 2.3689,
      "step": 30549
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6422262191772461,
      "learning_rate": 0.0005743464056615817,
      "loss": 2.2594,
      "step": 30550
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.658408522605896,
      "learning_rate": 0.0005743447505415296,
      "loss": 2.2817,
      "step": 30551
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.7066799998283386,
      "learning_rate": 0.0005743430953704718,
      "loss": 2.3654,
      "step": 30552
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6942632794380188,
      "learning_rate": 0.0005743414401484083,
      "loss": 2.2988,
      "step": 30553
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6451921463012695,
      "learning_rate": 0.0005743397848753396,
      "loss": 2.5748,
      "step": 30554
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6961818337440491,
      "learning_rate": 0.000574338129551266,
      "loss": 2.2162,
      "step": 30555
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.7430323958396912,
      "learning_rate": 0.0005743364741761876,
      "loss": 2.2173,
      "step": 30556
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6352354884147644,
      "learning_rate": 0.0005743348187501051,
      "loss": 2.4337,
      "step": 30557
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.628714382648468,
      "learning_rate": 0.0005743331632730185,
      "loss": 2.1323,
      "step": 30558
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6297891736030579,
      "learning_rate": 0.0005743315077449281,
      "loss": 2.3299,
      "step": 30559
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6330798864364624,
      "learning_rate": 0.0005743298521658344,
      "loss": 2.4271,
      "step": 30560
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.667633593082428,
      "learning_rate": 0.0005743281965357375,
      "loss": 2.3723,
      "step": 30561
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6940308213233948,
      "learning_rate": 0.0005743265408546379,
      "loss": 2.1995,
      "step": 30562
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.7320073843002319,
      "learning_rate": 0.0005743248851225358,
      "loss": 2.5558,
      "step": 30563
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.7448933124542236,
      "learning_rate": 0.0005743232293394314,
      "loss": 2.3694,
      "step": 30564
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.7495929598808289,
      "learning_rate": 0.0005743215735053253,
      "loss": 2.2784,
      "step": 30565
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.7740578651428223,
      "learning_rate": 0.0005743199176202176,
      "loss": 2.2949,
      "step": 30566
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6445364952087402,
      "learning_rate": 0.0005743182616841086,
      "loss": 2.354,
      "step": 30567
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6609946489334106,
      "learning_rate": 0.0005743166056969987,
      "loss": 2.4767,
      "step": 30568
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.7997153401374817,
      "learning_rate": 0.0005743149496588881,
      "loss": 2.5862,
      "step": 30569
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6203352212905884,
      "learning_rate": 0.0005743132935697772,
      "loss": 2.4808,
      "step": 30570
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6370694637298584,
      "learning_rate": 0.0005743116374296663,
      "loss": 2.2296,
      "step": 30571
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6445669531822205,
      "learning_rate": 0.0005743099812385557,
      "loss": 2.322,
      "step": 30572
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.5888316035270691,
      "learning_rate": 0.0005743083249964457,
      "loss": 2.2902,
      "step": 30573
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6822840571403503,
      "learning_rate": 0.0005743066687033366,
      "loss": 2.5069,
      "step": 30574
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6911417841911316,
      "learning_rate": 0.0005743050123592288,
      "loss": 2.4086,
      "step": 30575
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6520154476165771,
      "learning_rate": 0.0005743033559641223,
      "loss": 2.3578,
      "step": 30576
    },
    {
      "epoch": 0.4,
      "grad_norm": 2.167609691619873,
      "learning_rate": 0.0005743016995180178,
      "loss": 2.6003,
      "step": 30577
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6276889443397522,
      "learning_rate": 0.0005743000430209153,
      "loss": 1.919,
      "step": 30578
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6576700210571289,
      "learning_rate": 0.0005742983864728154,
      "loss": 2.2764,
      "step": 30579
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.642059862613678,
      "learning_rate": 0.0005742967298737182,
      "loss": 2.3012,
      "step": 30580
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6677065491676331,
      "learning_rate": 0.000574295073223624,
      "loss": 2.3736,
      "step": 30581
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6225016117095947,
      "learning_rate": 0.0005742934165225333,
      "loss": 2.2046,
      "step": 30582
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6062545776367188,
      "learning_rate": 0.0005742917597704462,
      "loss": 2.5732,
      "step": 30583
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6535045504570007,
      "learning_rate": 0.0005742901029673631,
      "loss": 2.3192,
      "step": 30584
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6579896211624146,
      "learning_rate": 0.0005742884461132842,
      "loss": 2.6162,
      "step": 30585
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6317556500434875,
      "learning_rate": 0.00057428678920821,
      "loss": 2.23,
      "step": 30586
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6326186656951904,
      "learning_rate": 0.0005742851322521407,
      "loss": 2.2834,
      "step": 30587
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6745091080665588,
      "learning_rate": 0.0005742834752450765,
      "loss": 2.4026,
      "step": 30588
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.661729633808136,
      "learning_rate": 0.000574281818187018,
      "loss": 2.5605,
      "step": 30589
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6417601704597473,
      "learning_rate": 0.0005742801610779652,
      "loss": 2.4975,
      "step": 30590
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6389750838279724,
      "learning_rate": 0.0005742785039179186,
      "loss": 2.2693,
      "step": 30591
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6163130402565002,
      "learning_rate": 0.0005742768467068784,
      "loss": 1.8565,
      "step": 30592
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.7116318941116333,
      "learning_rate": 0.0005742751894448451,
      "loss": 2.1978,
      "step": 30593
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6721615195274353,
      "learning_rate": 0.0005742735321318187,
      "loss": 2.4776,
      "step": 30594
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.8101214170455933,
      "learning_rate": 0.0005742718747677998,
      "loss": 2.4637,
      "step": 30595
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.7000184059143066,
      "learning_rate": 0.0005742702173527885,
      "loss": 2.3127,
      "step": 30596
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.7765582799911499,
      "learning_rate": 0.0005742685598867851,
      "loss": 2.3384,
      "step": 30597
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.886316180229187,
      "learning_rate": 0.0005742669023697901,
      "loss": 2.1618,
      "step": 30598
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.579980731010437,
      "learning_rate": 0.0005742652448018037,
      "loss": 2.4172,
      "step": 30599
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.7518712282180786,
      "learning_rate": 0.0005742635871828262,
      "loss": 2.1491,
      "step": 30600
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.791361927986145,
      "learning_rate": 0.000574261929512858,
      "loss": 2.4549,
      "step": 30601
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.5986749529838562,
      "learning_rate": 0.0005742602717918993,
      "loss": 2.3211,
      "step": 30602
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6343684196472168,
      "learning_rate": 0.0005742586140199504,
      "loss": 2.3724,
      "step": 30603
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6399199366569519,
      "learning_rate": 0.0005742569561970116,
      "loss": 2.088,
      "step": 30604
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6205902099609375,
      "learning_rate": 0.0005742552983230834,
      "loss": 2.2168,
      "step": 30605
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6377909779548645,
      "learning_rate": 0.0005742536403981659,
      "loss": 2.2905,
      "step": 30606
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6429881453514099,
      "learning_rate": 0.0005742519824222594,
      "loss": 2.2682,
      "step": 30607
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.690266489982605,
      "learning_rate": 0.0005742503243953644,
      "loss": 2.3902,
      "step": 30608
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.992680013179779,
      "learning_rate": 0.000574248666317481,
      "loss": 2.3994,
      "step": 30609
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6507886648178101,
      "learning_rate": 0.0005742470081886096,
      "loss": 2.5049,
      "step": 30610
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.7152866721153259,
      "learning_rate": 0.0005742453500087504,
      "loss": 2.5146,
      "step": 30611
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.670985996723175,
      "learning_rate": 0.0005742436917779041,
      "loss": 2.2563,
      "step": 30612
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6440481543540955,
      "learning_rate": 0.0005742420334960705,
      "loss": 2.3953,
      "step": 30613
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6550607681274414,
      "learning_rate": 0.0005742403751632501,
      "loss": 2.2619,
      "step": 30614
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6310955286026001,
      "learning_rate": 0.0005742387167794433,
      "loss": 2.4172,
      "step": 30615
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6065002679824829,
      "learning_rate": 0.0005742370583446504,
      "loss": 2.3807,
      "step": 30616
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6730052828788757,
      "learning_rate": 0.0005742353998588716,
      "loss": 2.1875,
      "step": 30617
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6414511799812317,
      "learning_rate": 0.0005742337413221073,
      "loss": 2.3455,
      "step": 30618
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6567838191986084,
      "learning_rate": 0.0005742320827343577,
      "loss": 2.3329,
      "step": 30619
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6053808331489563,
      "learning_rate": 0.0005742304240956233,
      "loss": 2.4121,
      "step": 30620
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6826205849647522,
      "learning_rate": 0.0005742287654059042,
      "loss": 2.2737,
      "step": 30621
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6733106970787048,
      "learning_rate": 0.0005742271066652008,
      "loss": 2.4282,
      "step": 30622
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6564099192619324,
      "learning_rate": 0.0005742254478735134,
      "loss": 2.3179,
      "step": 30623
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.630298376083374,
      "learning_rate": 0.0005742237890308423,
      "loss": 2.3395,
      "step": 30624
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.715096116065979,
      "learning_rate": 0.0005742221301371879,
      "loss": 2.3608,
      "step": 30625
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6883538961410522,
      "learning_rate": 0.0005742204711925503,
      "loss": 2.4829,
      "step": 30626
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.621881902217865,
      "learning_rate": 0.0005742188121969301,
      "loss": 2.1375,
      "step": 30627
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6243833303451538,
      "learning_rate": 0.0005742171531503273,
      "loss": 2.3938,
      "step": 30628
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.7216113209724426,
      "learning_rate": 0.0005742154940527425,
      "loss": 2.277,
      "step": 30629
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.7523855566978455,
      "learning_rate": 0.0005742138349041757,
      "loss": 2.2821,
      "step": 30630
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6260091662406921,
      "learning_rate": 0.0005742121757046275,
      "loss": 2.3703,
      "step": 30631
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6599233150482178,
      "learning_rate": 0.000574210516454098,
      "loss": 2.369,
      "step": 30632
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.8104179501533508,
      "learning_rate": 0.0005742088571525877,
      "loss": 2.3305,
      "step": 30633
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.7901190519332886,
      "learning_rate": 0.0005742071978000967,
      "loss": 2.4519,
      "step": 30634
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6069920063018799,
      "learning_rate": 0.0005742055383966254,
      "loss": 2.1042,
      "step": 30635
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6921800971031189,
      "learning_rate": 0.0005742038789421742,
      "loss": 2.5944,
      "step": 30636
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.7184095978736877,
      "learning_rate": 0.0005742022194367433,
      "loss": 2.2943,
      "step": 30637
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6846596002578735,
      "learning_rate": 0.0005742005598803329,
      "loss": 2.2784,
      "step": 30638
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6468807458877563,
      "learning_rate": 0.0005741989002729436,
      "loss": 2.1843,
      "step": 30639
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.7742666006088257,
      "learning_rate": 0.0005741972406145754,
      "loss": 2.1907,
      "step": 30640
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.7438437342643738,
      "learning_rate": 0.000574195580905229,
      "loss": 2.2978,
      "step": 30641
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6256628632545471,
      "learning_rate": 0.0005741939211449041,
      "loss": 2.3623,
      "step": 30642
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.660180389881134,
      "learning_rate": 0.0005741922613336017,
      "loss": 2.2951,
      "step": 30643
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.718572199344635,
      "learning_rate": 0.0005741906014713217,
      "loss": 2.3631,
      "step": 30644
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6500464081764221,
      "learning_rate": 0.0005741889415580645,
      "loss": 2.3064,
      "step": 30645
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6268280148506165,
      "learning_rate": 0.0005741872815938303,
      "loss": 2.3729,
      "step": 30646
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6420456767082214,
      "learning_rate": 0.0005741856215786196,
      "loss": 2.3127,
      "step": 30647
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.649760901927948,
      "learning_rate": 0.0005741839615124325,
      "loss": 2.2563,
      "step": 30648
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.7309603095054626,
      "learning_rate": 0.0005741823013952695,
      "loss": 2.3238,
      "step": 30649
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.7446863651275635,
      "learning_rate": 0.0005741806412271308,
      "loss": 2.2149,
      "step": 30650
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.7408673763275146,
      "learning_rate": 0.0005741789810080168,
      "loss": 2.5009,
      "step": 30651
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.9397552609443665,
      "learning_rate": 0.0005741773207379278,
      "loss": 2.4388,
      "step": 30652
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6610990166664124,
      "learning_rate": 0.000574175660416864,
      "loss": 2.3593,
      "step": 30653
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.7105762958526611,
      "learning_rate": 0.0005741740000448258,
      "loss": 2.0843,
      "step": 30654
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.7469114661216736,
      "learning_rate": 0.0005741723396218134,
      "loss": 2.2995,
      "step": 30655
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.8197286128997803,
      "learning_rate": 0.0005741706791478272,
      "loss": 2.4044,
      "step": 30656
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.712142288684845,
      "learning_rate": 0.0005741690186228676,
      "loss": 2.0903,
      "step": 30657
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.8028407096862793,
      "learning_rate": 0.0005741673580469347,
      "loss": 2.2654,
      "step": 30658
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6785517334938049,
      "learning_rate": 0.0005741656974200289,
      "loss": 2.4717,
      "step": 30659
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6256293058395386,
      "learning_rate": 0.0005741640367421506,
      "loss": 2.3486,
      "step": 30660
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.7423029541969299,
      "learning_rate": 0.0005741623760133,
      "loss": 2.3025,
      "step": 30661
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6565030217170715,
      "learning_rate": 0.0005741607152334775,
      "loss": 2.3188,
      "step": 30662
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.7844802737236023,
      "learning_rate": 0.0005741590544026833,
      "loss": 2.4842,
      "step": 30663
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.657886266708374,
      "learning_rate": 0.0005741573935209177,
      "loss": 2.1872,
      "step": 30664
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.8376899361610413,
      "learning_rate": 0.0005741557325881811,
      "loss": 2.2822,
      "step": 30665
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6750772595405579,
      "learning_rate": 0.0005741540716044738,
      "loss": 2.0811,
      "step": 30666
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.7124064564704895,
      "learning_rate": 0.000574152410569796,
      "loss": 2.2772,
      "step": 30667
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.7272862195968628,
      "learning_rate": 0.0005741507494841482,
      "loss": 2.584,
      "step": 30668
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6383306980133057,
      "learning_rate": 0.0005741490883475305,
      "loss": 2.1675,
      "step": 30669
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6717202067375183,
      "learning_rate": 0.0005741474271599434,
      "loss": 2.2793,
      "step": 30670
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6855956315994263,
      "learning_rate": 0.000574145765921387,
      "loss": 2.1233,
      "step": 30671
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.8133711814880371,
      "learning_rate": 0.0005741441046318618,
      "loss": 2.1456,
      "step": 30672
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6398322582244873,
      "learning_rate": 0.0005741424432913682,
      "loss": 2.2552,
      "step": 30673
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.9165957570075989,
      "learning_rate": 0.0005741407818999062,
      "loss": 2.2361,
      "step": 30674
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.7927693128585815,
      "learning_rate": 0.0005741391204574762,
      "loss": 2.1599,
      "step": 30675
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6708686947822571,
      "learning_rate": 0.0005741374589640786,
      "loss": 2.2723,
      "step": 30676
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.7525461316108704,
      "learning_rate": 0.0005741357974197137,
      "loss": 2.3346,
      "step": 30677
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.7293826937675476,
      "learning_rate": 0.0005741341358243817,
      "loss": 2.3848,
      "step": 30678
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6058170199394226,
      "learning_rate": 0.0005741324741780831,
      "loss": 2.347,
      "step": 30679
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6374409794807434,
      "learning_rate": 0.000574130812480818,
      "loss": 2.4205,
      "step": 30680
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.7123531699180603,
      "learning_rate": 0.0005741291507325868,
      "loss": 2.5737,
      "step": 30681
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.7539791464805603,
      "learning_rate": 0.00057412748893339,
      "loss": 2.3298,
      "step": 30682
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6279462575912476,
      "learning_rate": 0.0005741258270832275,
      "loss": 2.348,
      "step": 30683
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6678004860877991,
      "learning_rate": 0.0005741241651821001,
      "loss": 2.1555,
      "step": 30684
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6648818254470825,
      "learning_rate": 0.0005741225032300075,
      "loss": 2.4322,
      "step": 30685
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6933547258377075,
      "learning_rate": 0.0005741208412269506,
      "loss": 2.1181,
      "step": 30686
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.5900693535804749,
      "learning_rate": 0.0005741191791729293,
      "loss": 2.2751,
      "step": 30687
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6768429279327393,
      "learning_rate": 0.0005741175170679442,
      "loss": 2.3029,
      "step": 30688
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.7723510265350342,
      "learning_rate": 0.0005741158549119954,
      "loss": 2.2774,
      "step": 30689
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6250858902931213,
      "learning_rate": 0.0005741141927050834,
      "loss": 2.5053,
      "step": 30690
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6372314691543579,
      "learning_rate": 0.0005741125304472083,
      "loss": 2.172,
      "step": 30691
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6932801008224487,
      "learning_rate": 0.0005741108681383706,
      "loss": 2.4285,
      "step": 30692
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6458691358566284,
      "learning_rate": 0.0005741092057785703,
      "loss": 2.1119,
      "step": 30693
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6036798357963562,
      "learning_rate": 0.0005741075433678081,
      "loss": 2.1649,
      "step": 30694
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6470610499382019,
      "learning_rate": 0.0005741058809060842,
      "loss": 2.3992,
      "step": 30695
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.7580357789993286,
      "learning_rate": 0.0005741042183933987,
      "loss": 2.3673,
      "step": 30696
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.578953206539154,
      "learning_rate": 0.0005741025558297521,
      "loss": 2.1538,
      "step": 30697
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.612086296081543,
      "learning_rate": 0.0005741008932151447,
      "loss": 2.2036,
      "step": 30698
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6463651061058044,
      "learning_rate": 0.0005740992305495767,
      "loss": 2.2485,
      "step": 30699
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6730058193206787,
      "learning_rate": 0.0005740975678330485,
      "loss": 2.3963,
      "step": 30700
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6241481304168701,
      "learning_rate": 0.0005740959050655604,
      "loss": 2.2274,
      "step": 30701
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6710466742515564,
      "learning_rate": 0.0005740942422471127,
      "loss": 2.4799,
      "step": 30702
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.7228113412857056,
      "learning_rate": 0.0005740925793777056,
      "loss": 2.5286,
      "step": 30703
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.592471182346344,
      "learning_rate": 0.0005740909164573397,
      "loss": 2.179,
      "step": 30704
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6413553357124329,
      "learning_rate": 0.000574089253486015,
      "loss": 2.3001,
      "step": 30705
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6676985621452332,
      "learning_rate": 0.000574087590463732,
      "loss": 2.2988,
      "step": 30706
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6656195521354675,
      "learning_rate": 0.000574085927390491,
      "loss": 2.0511,
      "step": 30707
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6076804399490356,
      "learning_rate": 0.0005740842642662921,
      "loss": 2.435,
      "step": 30708
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6174045205116272,
      "learning_rate": 0.0005740826010911358,
      "loss": 2.2525,
      "step": 30709
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6450605988502502,
      "learning_rate": 0.0005740809378650224,
      "loss": 2.1619,
      "step": 30710
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.7395772933959961,
      "learning_rate": 0.0005740792745879522,
      "loss": 2.3578,
      "step": 30711
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6545548439025879,
      "learning_rate": 0.0005740776112599255,
      "loss": 2.1913,
      "step": 30712
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6951753497123718,
      "learning_rate": 0.0005740759478809425,
      "loss": 2.522,
      "step": 30713
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6888433694839478,
      "learning_rate": 0.0005740742844510037,
      "loss": 2.4961,
      "step": 30714
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6713395714759827,
      "learning_rate": 0.0005740726209701093,
      "loss": 2.2079,
      "step": 30715
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6889990568161011,
      "learning_rate": 0.0005740709574382596,
      "loss": 2.1341,
      "step": 30716
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6702421307563782,
      "learning_rate": 0.000574069293855455,
      "loss": 2.123,
      "step": 30717
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6587888598442078,
      "learning_rate": 0.0005740676302216957,
      "loss": 2.3923,
      "step": 30718
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.653634786605835,
      "learning_rate": 0.000574065966536982,
      "loss": 2.2466,
      "step": 30719
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6598747372627258,
      "learning_rate": 0.0005740643028013143,
      "loss": 2.5014,
      "step": 30720
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6964394450187683,
      "learning_rate": 0.0005740626390146929,
      "loss": 2.3297,
      "step": 30721
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.7098795175552368,
      "learning_rate": 0.0005740609751771181,
      "loss": 2.358,
      "step": 30722
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.7634263634681702,
      "learning_rate": 0.0005740593112885902,
      "loss": 2.2589,
      "step": 30723
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.703129768371582,
      "learning_rate": 0.0005740576473491094,
      "loss": 2.3115,
      "step": 30724
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.659250795841217,
      "learning_rate": 0.0005740559833586762,
      "loss": 2.2601,
      "step": 30725
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.7475352883338928,
      "learning_rate": 0.0005740543193172909,
      "loss": 2.3143,
      "step": 30726
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.689987301826477,
      "learning_rate": 0.0005740526552249535,
      "loss": 2.4118,
      "step": 30727
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6711349487304688,
      "learning_rate": 0.0005740509910816646,
      "loss": 2.3372,
      "step": 30728
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6217588782310486,
      "learning_rate": 0.0005740493268874246,
      "loss": 2.3093,
      "step": 30729
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6454101800918579,
      "learning_rate": 0.0005740476626422336,
      "loss": 2.127,
      "step": 30730
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6418198943138123,
      "learning_rate": 0.0005740459983460919,
      "loss": 2.2257,
      "step": 30731
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.688018798828125,
      "learning_rate": 0.0005740443339989999,
      "loss": 2.2038,
      "step": 30732
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.7611494064331055,
      "learning_rate": 0.0005740426696009579,
      "loss": 2.1868,
      "step": 30733
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.601814866065979,
      "learning_rate": 0.0005740410051519662,
      "loss": 2.5112,
      "step": 30734
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6309415698051453,
      "learning_rate": 0.0005740393406520251,
      "loss": 2.6378,
      "step": 30735
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6637667417526245,
      "learning_rate": 0.0005740376761011349,
      "loss": 2.1405,
      "step": 30736
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.7958967089653015,
      "learning_rate": 0.0005740360114992959,
      "loss": 2.3702,
      "step": 30737
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6151422262191772,
      "learning_rate": 0.0005740343468465085,
      "loss": 2.4411,
      "step": 30738
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.7553966045379639,
      "learning_rate": 0.0005740326821427729,
      "loss": 2.388,
      "step": 30739
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.7228662371635437,
      "learning_rate": 0.0005740310173880895,
      "loss": 2.2133,
      "step": 30740
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.668376624584198,
      "learning_rate": 0.0005740293525824586,
      "loss": 2.2911,
      "step": 30741
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6665043830871582,
      "learning_rate": 0.0005740276877258802,
      "loss": 1.9427,
      "step": 30742
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.7645853161811829,
      "learning_rate": 0.0005740260228183551,
      "loss": 2.3312,
      "step": 30743
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.76657634973526,
      "learning_rate": 0.0005740243578598833,
      "loss": 2.3658,
      "step": 30744
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6342324018478394,
      "learning_rate": 0.0005740226928504652,
      "loss": 2.1827,
      "step": 30745
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.8797340989112854,
      "learning_rate": 0.0005740210277901014,
      "loss": 2.5489,
      "step": 30746
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.8252055644989014,
      "learning_rate": 0.0005740193626787916,
      "loss": 2.3099,
      "step": 30747
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.7229159474372864,
      "learning_rate": 0.0005740176975165365,
      "loss": 2.5293,
      "step": 30748
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.7444681525230408,
      "learning_rate": 0.0005740160323033363,
      "loss": 2.2501,
      "step": 30749
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.9013618230819702,
      "learning_rate": 0.0005740143670391914,
      "loss": 2.4203,
      "step": 30750
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6494297981262207,
      "learning_rate": 0.0005740127017241022,
      "loss": 2.3019,
      "step": 30751
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6613859534263611,
      "learning_rate": 0.0005740110363580686,
      "loss": 2.6311,
      "step": 30752
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.68845134973526,
      "learning_rate": 0.0005740093709410912,
      "loss": 2.4838,
      "step": 30753
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6477363109588623,
      "learning_rate": 0.0005740077054731705,
      "loss": 1.9763,
      "step": 30754
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6782238483428955,
      "learning_rate": 0.0005740060399543064,
      "loss": 2.5765,
      "step": 30755
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6743537187576294,
      "learning_rate": 0.0005740043743844996,
      "loss": 2.2654,
      "step": 30756
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6399242281913757,
      "learning_rate": 0.0005740027087637499,
      "loss": 2.525,
      "step": 30757
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.7216593623161316,
      "learning_rate": 0.0005740010430920582,
      "loss": 2.3437,
      "step": 30758
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6684930324554443,
      "learning_rate": 0.0005739993773694245,
      "loss": 2.3245,
      "step": 30759
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6480767726898193,
      "learning_rate": 0.000573997711595849,
      "loss": 2.2544,
      "step": 30760
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6197147369384766,
      "learning_rate": 0.0005739960457713323,
      "loss": 2.2965,
      "step": 30761
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.629330039024353,
      "learning_rate": 0.0005739943798958746,
      "loss": 2.4146,
      "step": 30762
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.763857364654541,
      "learning_rate": 0.000573992713969476,
      "loss": 2.1843,
      "step": 30763
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6311519742012024,
      "learning_rate": 0.0005739910479921371,
      "loss": 2.3501,
      "step": 30764
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6326076984405518,
      "learning_rate": 0.0005739893819638581,
      "loss": 2.0331,
      "step": 30765
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6866296529769897,
      "learning_rate": 0.0005739877158846392,
      "loss": 2.1918,
      "step": 30766
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6051292419433594,
      "learning_rate": 0.0005739860497544809,
      "loss": 2.1222,
      "step": 30767
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6347329020500183,
      "learning_rate": 0.0005739843835733834,
      "loss": 2.1844,
      "step": 30768
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.7299121618270874,
      "learning_rate": 0.000573982717341347,
      "loss": 2.5822,
      "step": 30769
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6256373524665833,
      "learning_rate": 0.000573981051058372,
      "loss": 2.34,
      "step": 30770
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6553982496261597,
      "learning_rate": 0.000573979384724459,
      "loss": 2.2469,
      "step": 30771
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6302608847618103,
      "learning_rate": 0.0005739777183396078,
      "loss": 2.2986,
      "step": 30772
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.834627091884613,
      "learning_rate": 0.000573976051903819,
      "loss": 2.4709,
      "step": 30773
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6813021898269653,
      "learning_rate": 0.000573974385417093,
      "loss": 2.3574,
      "step": 30774
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6097844243049622,
      "learning_rate": 0.0005739727188794299,
      "loss": 2.2612,
      "step": 30775
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6419326066970825,
      "learning_rate": 0.0005739710522908301,
      "loss": 2.1743,
      "step": 30776
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6523931622505188,
      "learning_rate": 0.0005739693856512939,
      "loss": 2.2134,
      "step": 30777
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6319078803062439,
      "learning_rate": 0.0005739677189608217,
      "loss": 2.1771,
      "step": 30778
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6112979650497437,
      "learning_rate": 0.0005739660522194137,
      "loss": 2.4505,
      "step": 30779
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6107465028762817,
      "learning_rate": 0.0005739643854270702,
      "loss": 2.0745,
      "step": 30780
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.7140432596206665,
      "learning_rate": 0.0005739627185837916,
      "loss": 2.1548,
      "step": 30781
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6999642252922058,
      "learning_rate": 0.0005739610516895781,
      "loss": 2.4015,
      "step": 30782
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.7538795471191406,
      "learning_rate": 0.0005739593847444301,
      "loss": 2.4502,
      "step": 30783
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6205816268920898,
      "learning_rate": 0.0005739577177483478,
      "loss": 2.1811,
      "step": 30784
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6687310934066772,
      "learning_rate": 0.0005739560507013317,
      "loss": 2.337,
      "step": 30785
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6437564492225647,
      "learning_rate": 0.000573954383603382,
      "loss": 2.3256,
      "step": 30786
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6827681660652161,
      "learning_rate": 0.000573952716454499,
      "loss": 2.2451,
      "step": 30787
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6458321213722229,
      "learning_rate": 0.000573951049254683,
      "loss": 2.5653,
      "step": 30788
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6935288310050964,
      "learning_rate": 0.0005739493820039343,
      "loss": 2.3082,
      "step": 30789
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6936275362968445,
      "learning_rate": 0.0005739477147022532,
      "loss": 2.2076,
      "step": 30790
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.5990422964096069,
      "learning_rate": 0.0005739460473496401,
      "loss": 2.2651,
      "step": 30791
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6285587549209595,
      "learning_rate": 0.0005739443799460953,
      "loss": 2.4145,
      "step": 30792
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6106969118118286,
      "learning_rate": 0.0005739427124916191,
      "loss": 2.197,
      "step": 30793
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6459618806838989,
      "learning_rate": 0.0005739410449862116,
      "loss": 2.0946,
      "step": 30794
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6997732520103455,
      "learning_rate": 0.0005739393774298735,
      "loss": 2.4591,
      "step": 30795
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6874006986618042,
      "learning_rate": 0.0005739377098226047,
      "loss": 2.3994,
      "step": 30796
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6444332003593445,
      "learning_rate": 0.0005739360421644058,
      "loss": 2.1446,
      "step": 30797
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.671806812286377,
      "learning_rate": 0.0005739343744552771,
      "loss": 2.2685,
      "step": 30798
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6995415091514587,
      "learning_rate": 0.0005739327066952187,
      "loss": 2.3156,
      "step": 30799
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6750442981719971,
      "learning_rate": 0.0005739310388842311,
      "loss": 2.4567,
      "step": 30800
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6293703317642212,
      "learning_rate": 0.0005739293710223146,
      "loss": 2.5428,
      "step": 30801
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6070089340209961,
      "learning_rate": 0.0005739277031094694,
      "loss": 2.7258,
      "step": 30802
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6169217228889465,
      "learning_rate": 0.0005739260351456959,
      "loss": 2.4707,
      "step": 30803
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.61203932762146,
      "learning_rate": 0.0005739243671309943,
      "loss": 1.9367,
      "step": 30804
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6561952829360962,
      "learning_rate": 0.0005739226990653651,
      "loss": 2.1635,
      "step": 30805
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.651054859161377,
      "learning_rate": 0.0005739210309488084,
      "loss": 2.3874,
      "step": 30806
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6433196663856506,
      "learning_rate": 0.0005739193627813246,
      "loss": 2.2117,
      "step": 30807
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.7010897994041443,
      "learning_rate": 0.0005739176945629141,
      "loss": 2.5234,
      "step": 30808
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6175284385681152,
      "learning_rate": 0.0005739160262935772,
      "loss": 2.5542,
      "step": 30809
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6625723242759705,
      "learning_rate": 0.000573914357973314,
      "loss": 2.1683,
      "step": 30810
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.620960533618927,
      "learning_rate": 0.000573912689602125,
      "loss": 2.2649,
      "step": 30811
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6169801950454712,
      "learning_rate": 0.0005739110211800104,
      "loss": 2.4147,
      "step": 30812
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6892377138137817,
      "learning_rate": 0.0005739093527069707,
      "loss": 2.2868,
      "step": 30813
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6531577706336975,
      "learning_rate": 0.000573907684183006,
      "loss": 2.1029,
      "step": 30814
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6965304017066956,
      "learning_rate": 0.0005739060156081166,
      "loss": 2.1954,
      "step": 30815
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6416721940040588,
      "learning_rate": 0.000573904346982303,
      "loss": 2.2672,
      "step": 30816
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6339949369430542,
      "learning_rate": 0.0005739026783055654,
      "loss": 2.2491,
      "step": 30817
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.721790611743927,
      "learning_rate": 0.0005739010095779042,
      "loss": 2.2859,
      "step": 30818
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6847296953201294,
      "learning_rate": 0.0005738993407993195,
      "loss": 2.4615,
      "step": 30819
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6047105193138123,
      "learning_rate": 0.0005738976719698118,
      "loss": 2.2331,
      "step": 30820
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.7929213643074036,
      "learning_rate": 0.0005738960030893815,
      "loss": 2.0301,
      "step": 30821
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.7076390981674194,
      "learning_rate": 0.0005738943341580285,
      "loss": 2.3935,
      "step": 30822
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.8418184518814087,
      "learning_rate": 0.0005738926651757536,
      "loss": 2.2774,
      "step": 30823
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6707469820976257,
      "learning_rate": 0.0005738909961425566,
      "loss": 2.1083,
      "step": 30824
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6384414434432983,
      "learning_rate": 0.0005738893270584383,
      "loss": 2.2636,
      "step": 30825
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.5998393893241882,
      "learning_rate": 0.0005738876579233987,
      "loss": 2.282,
      "step": 30826
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6721541881561279,
      "learning_rate": 0.0005738859887374384,
      "loss": 2.3415,
      "step": 30827
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6745461821556091,
      "learning_rate": 0.0005738843195005574,
      "loss": 2.2127,
      "step": 30828
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.7196496725082397,
      "learning_rate": 0.0005738826502127561,
      "loss": 2.5277,
      "step": 30829
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6255340576171875,
      "learning_rate": 0.0005738809808740349,
      "loss": 2.379,
      "step": 30830
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6439524292945862,
      "learning_rate": 0.0005738793114843941,
      "loss": 2.2391,
      "step": 30831
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6730895638465881,
      "learning_rate": 0.0005738776420438339,
      "loss": 2.2316,
      "step": 30832
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.7111883163452148,
      "learning_rate": 0.0005738759725523547,
      "loss": 2.6221,
      "step": 30833
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6518901586532593,
      "learning_rate": 0.0005738743030099568,
      "loss": 2.3562,
      "step": 30834
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6424496173858643,
      "learning_rate": 0.0005738726334166404,
      "loss": 2.4699,
      "step": 30835
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.7217243909835815,
      "learning_rate": 0.000573870963772406,
      "loss": 2.3154,
      "step": 30836
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.7129813432693481,
      "learning_rate": 0.0005738692940772538,
      "loss": 2.0532,
      "step": 30837
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6543235182762146,
      "learning_rate": 0.0005738676243311841,
      "loss": 2.4125,
      "step": 30838
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6573771834373474,
      "learning_rate": 0.0005738659545341974,
      "loss": 2.3757,
      "step": 30839
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6160045862197876,
      "learning_rate": 0.0005738642846862936,
      "loss": 2.2444,
      "step": 30840
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6777757406234741,
      "learning_rate": 0.0005738626147874734,
      "loss": 2.1926,
      "step": 30841
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6991518139839172,
      "learning_rate": 0.0005738609448377371,
      "loss": 2.3479,
      "step": 30842
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6743288040161133,
      "learning_rate": 0.0005738592748370845,
      "loss": 2.2673,
      "step": 30843
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6007387042045593,
      "learning_rate": 0.0005738576047855166,
      "loss": 2.1576,
      "step": 30844
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6665674448013306,
      "learning_rate": 0.0005738559346830334,
      "loss": 2.3617,
      "step": 30845
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6273189783096313,
      "learning_rate": 0.0005738542645296351,
      "loss": 2.1321,
      "step": 30846
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6458927392959595,
      "learning_rate": 0.0005738525943253222,
      "loss": 2.3508,
      "step": 30847
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.656657338142395,
      "learning_rate": 0.0005738509240700948,
      "loss": 2.2849,
      "step": 30848
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6408944129943848,
      "learning_rate": 0.0005738492537639535,
      "loss": 2.3829,
      "step": 30849
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.7328681349754333,
      "learning_rate": 0.0005738475834068984,
      "loss": 2.4826,
      "step": 30850
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.732956051826477,
      "learning_rate": 0.0005738459129989298,
      "loss": 2.2252,
      "step": 30851
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6585466861724854,
      "learning_rate": 0.000573844242540048,
      "loss": 2.2445,
      "step": 30852
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.653009831905365,
      "learning_rate": 0.0005738425720302536,
      "loss": 2.4578,
      "step": 30853
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.7766101360321045,
      "learning_rate": 0.0005738409014695466,
      "loss": 2.3037,
      "step": 30854
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6203768253326416,
      "learning_rate": 0.0005738392308579274,
      "loss": 2.0933,
      "step": 30855
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.7441579699516296,
      "learning_rate": 0.0005738375601953964,
      "loss": 2.3255,
      "step": 30856
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6659943461418152,
      "learning_rate": 0.0005738358894819538,
      "loss": 2.4834,
      "step": 30857
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.7386094331741333,
      "learning_rate": 0.0005738342187175999,
      "loss": 2.1767,
      "step": 30858
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6236302256584167,
      "learning_rate": 0.0005738325479023351,
      "loss": 2.2069,
      "step": 30859
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6625546813011169,
      "learning_rate": 0.0005738308770361595,
      "loss": 2.1227,
      "step": 30860
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.7085561752319336,
      "learning_rate": 0.0005738292061190737,
      "loss": 2.0615,
      "step": 30861
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6915957927703857,
      "learning_rate": 0.0005738275351510779,
      "loss": 2.2956,
      "step": 30862
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.7395426034927368,
      "learning_rate": 0.0005738258641321724,
      "loss": 2.4726,
      "step": 30863
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6446634531021118,
      "learning_rate": 0.0005738241930623575,
      "loss": 2.2649,
      "step": 30864
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6839785575866699,
      "learning_rate": 0.0005738225219416335,
      "loss": 2.2314,
      "step": 30865
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6750379204750061,
      "learning_rate": 0.0005738208507700006,
      "loss": 2.3859,
      "step": 30866
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6582058668136597,
      "learning_rate": 0.0005738191795474593,
      "loss": 2.443,
      "step": 30867
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6756871342658997,
      "learning_rate": 0.0005738175082740099,
      "loss": 2.1355,
      "step": 30868
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.8555009365081787,
      "learning_rate": 0.0005738158369496527,
      "loss": 2.3102,
      "step": 30869
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6825917959213257,
      "learning_rate": 0.0005738141655743879,
      "loss": 2.3839,
      "step": 30870
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.5955584645271301,
      "learning_rate": 0.0005738124941482159,
      "loss": 2.2792,
      "step": 30871
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.689822256565094,
      "learning_rate": 0.000573810822671137,
      "loss": 2.2418,
      "step": 30872
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.656762421131134,
      "learning_rate": 0.0005738091511431514,
      "loss": 2.1717,
      "step": 30873
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.5795559883117676,
      "learning_rate": 0.0005738074795642596,
      "loss": 2.2359,
      "step": 30874
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.7652784585952759,
      "learning_rate": 0.0005738058079344618,
      "loss": 2.1158,
      "step": 30875
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.7035320997238159,
      "learning_rate": 0.0005738041362537583,
      "loss": 2.5994,
      "step": 30876
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6678807735443115,
      "learning_rate": 0.0005738024645221495,
      "loss": 2.2096,
      "step": 30877
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.7125513553619385,
      "learning_rate": 0.0005738007927396356,
      "loss": 2.416,
      "step": 30878
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.7518397569656372,
      "learning_rate": 0.0005737991209062171,
      "loss": 2.336,
      "step": 30879
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6517199277877808,
      "learning_rate": 0.0005737974490218939,
      "loss": 2.1079,
      "step": 30880
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.7363319396972656,
      "learning_rate": 0.0005737957770866669,
      "loss": 2.322,
      "step": 30881
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6692901849746704,
      "learning_rate": 0.0005737941051005359,
      "loss": 2.4399,
      "step": 30882
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.5867049098014832,
      "learning_rate": 0.0005737924330635015,
      "loss": 2.2419,
      "step": 30883
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6458703279495239,
      "learning_rate": 0.0005737907609755638,
      "loss": 2.0981,
      "step": 30884
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6227084994316101,
      "learning_rate": 0.0005737890888367234,
      "loss": 2.2553,
      "step": 30885
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6173256635665894,
      "learning_rate": 0.0005737874166469803,
      "loss": 2.2957,
      "step": 30886
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6340566277503967,
      "learning_rate": 0.000573785744406335,
      "loss": 2.5363,
      "step": 30887
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6660873293876648,
      "learning_rate": 0.0005737840721147878,
      "loss": 2.5753,
      "step": 30888
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.5742993950843811,
      "learning_rate": 0.0005737823997723389,
      "loss": 2.1505,
      "step": 30889
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.7531279921531677,
      "learning_rate": 0.0005737807273789887,
      "loss": 2.115,
      "step": 30890
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.617901623249054,
      "learning_rate": 0.0005737790549347375,
      "loss": 2.2294,
      "step": 30891
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6657348275184631,
      "learning_rate": 0.0005737773824395856,
      "loss": 2.5437,
      "step": 30892
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.5849869847297668,
      "learning_rate": 0.0005737757098935334,
      "loss": 2.385,
      "step": 30893
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6622028946876526,
      "learning_rate": 0.0005737740372965811,
      "loss": 2.312,
      "step": 30894
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6825892925262451,
      "learning_rate": 0.0005737723646487289,
      "loss": 2.5374,
      "step": 30895
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6523518562316895,
      "learning_rate": 0.0005737706919499774,
      "loss": 2.215,
      "step": 30896
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6312899589538574,
      "learning_rate": 0.0005737690192003267,
      "loss": 2.3492,
      "step": 30897
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6470059752464294,
      "learning_rate": 0.0005737673463997772,
      "loss": 2.1607,
      "step": 30898
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6420384645462036,
      "learning_rate": 0.0005737656735483292,
      "loss": 2.0325,
      "step": 30899
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6269915103912354,
      "learning_rate": 0.000573764000645983,
      "loss": 2.2945,
      "step": 30900
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.7238867878913879,
      "learning_rate": 0.0005737623276927388,
      "loss": 2.2327,
      "step": 30901
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6574763059616089,
      "learning_rate": 0.0005737606546885972,
      "loss": 2.1021,
      "step": 30902
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.666217565536499,
      "learning_rate": 0.000573758981633558,
      "loss": 2.5181,
      "step": 30903
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.7154296040534973,
      "learning_rate": 0.0005737573085276221,
      "loss": 2.1093,
      "step": 30904
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6585714817047119,
      "learning_rate": 0.0005737556353707895,
      "loss": 2.1164,
      "step": 30905
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.7039358615875244,
      "learning_rate": 0.0005737539621630606,
      "loss": 2.154,
      "step": 30906
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.7230240702629089,
      "learning_rate": 0.0005737522889044356,
      "loss": 2.3308,
      "step": 30907
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.7967312335968018,
      "learning_rate": 0.0005737506155949149,
      "loss": 2.7685,
      "step": 30908
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6546398401260376,
      "learning_rate": 0.0005737489422344988,
      "loss": 2.2739,
      "step": 30909
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6881098747253418,
      "learning_rate": 0.0005737472688231876,
      "loss": 2.3439,
      "step": 30910
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6957219839096069,
      "learning_rate": 0.0005737455953609816,
      "loss": 2.3079,
      "step": 30911
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6556115746498108,
      "learning_rate": 0.0005737439218478812,
      "loss": 2.2854,
      "step": 30912
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6669348478317261,
      "learning_rate": 0.0005737422482838865,
      "loss": 2.5722,
      "step": 30913
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6267386674880981,
      "learning_rate": 0.000573740574668998,
      "loss": 2.3197,
      "step": 30914
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6000357270240784,
      "learning_rate": 0.0005737389010032161,
      "loss": 1.9579,
      "step": 30915
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6818804740905762,
      "learning_rate": 0.0005737372272865408,
      "loss": 2.1391,
      "step": 30916
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6514747142791748,
      "learning_rate": 0.0005737355535189726,
      "loss": 2.5153,
      "step": 30917
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6516950726509094,
      "learning_rate": 0.0005737338797005119,
      "loss": 2.3552,
      "step": 30918
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6592419743537903,
      "learning_rate": 0.0005737322058311588,
      "loss": 2.1828,
      "step": 30919
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.5727185010910034,
      "learning_rate": 0.0005737305319109138,
      "loss": 2.124,
      "step": 30920
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.627382218837738,
      "learning_rate": 0.0005737288579397771,
      "loss": 2.188,
      "step": 30921
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6397131085395813,
      "learning_rate": 0.000573727183917749,
      "loss": 2.3934,
      "step": 30922
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6932563781738281,
      "learning_rate": 0.0005737255098448298,
      "loss": 2.498,
      "step": 30923
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.657501220703125,
      "learning_rate": 0.0005737238357210201,
      "loss": 2.2858,
      "step": 30924
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.7010071873664856,
      "learning_rate": 0.0005737221615463198,
      "loss": 2.1807,
      "step": 30925
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.7013047933578491,
      "learning_rate": 0.0005737204873207292,
      "loss": 2.4735,
      "step": 30926
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.5944458246231079,
      "learning_rate": 0.0005737188130442491,
      "loss": 2.2319,
      "step": 30927
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.631216287612915,
      "learning_rate": 0.0005737171387168793,
      "loss": 2.4465,
      "step": 30928
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6273143291473389,
      "learning_rate": 0.0005737154643386205,
      "loss": 2.3855,
      "step": 30929
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6692051887512207,
      "learning_rate": 0.0005737137899094728,
      "loss": 2.5135,
      "step": 30930
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6101582646369934,
      "learning_rate": 0.0005737121154294364,
      "loss": 2.3291,
      "step": 30931
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.619799792766571,
      "learning_rate": 0.0005737104408985118,
      "loss": 2.433,
      "step": 30932
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.5795872807502747,
      "learning_rate": 0.0005737087663166994,
      "loss": 2.4802,
      "step": 30933
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6436322927474976,
      "learning_rate": 0.0005737070916839992,
      "loss": 2.3017,
      "step": 30934
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.681423008441925,
      "learning_rate": 0.0005737054170004117,
      "loss": 2.255,
      "step": 30935
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.5960571765899658,
      "learning_rate": 0.0005737037422659374,
      "loss": 2.4968,
      "step": 30936
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.7373514771461487,
      "learning_rate": 0.0005737020674805761,
      "loss": 2.2449,
      "step": 30937
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6160097122192383,
      "learning_rate": 0.0005737003926443285,
      "loss": 2.2988,
      "step": 30938
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6242725253105164,
      "learning_rate": 0.000573698717757195,
      "loss": 2.37,
      "step": 30939
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6927655339241028,
      "learning_rate": 0.0005736970428191757,
      "loss": 2.3219,
      "step": 30940
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6602180600166321,
      "learning_rate": 0.0005736953678302708,
      "loss": 2.3383,
      "step": 30941
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6941307783126831,
      "learning_rate": 0.0005736936927904808,
      "loss": 2.3972,
      "step": 30942
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6709675192832947,
      "learning_rate": 0.0005736920176998061,
      "loss": 2.4755,
      "step": 30943
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6059378981590271,
      "learning_rate": 0.0005736903425582468,
      "loss": 2.617,
      "step": 30944
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6479863524436951,
      "learning_rate": 0.0005736886673658034,
      "loss": 2.4964,
      "step": 30945
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.5861757397651672,
      "learning_rate": 0.000573686992122476,
      "loss": 2.2521,
      "step": 30946
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6727263927459717,
      "learning_rate": 0.000573685316828265,
      "loss": 2.3734,
      "step": 30947
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.5967127084732056,
      "learning_rate": 0.0005736836414831709,
      "loss": 2.4355,
      "step": 30948
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.7864766716957092,
      "learning_rate": 0.0005736819660871937,
      "loss": 2.4566,
      "step": 30949
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.7057957649230957,
      "learning_rate": 0.0005736802906403339,
      "loss": 2.1536,
      "step": 30950
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6637113690376282,
      "learning_rate": 0.0005736786151425917,
      "loss": 2.5081,
      "step": 30951
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.7881098985671997,
      "learning_rate": 0.0005736769395939676,
      "loss": 2.2907,
      "step": 30952
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6756545901298523,
      "learning_rate": 0.0005736752639944618,
      "loss": 2.3103,
      "step": 30953
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6624512076377869,
      "learning_rate": 0.0005736735883440744,
      "loss": 2.3309,
      "step": 30954
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6885396242141724,
      "learning_rate": 0.000573671912642806,
      "loss": 2.1757,
      "step": 30955
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.738515317440033,
      "learning_rate": 0.0005736702368906569,
      "loss": 2.1538,
      "step": 30956
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.83769690990448,
      "learning_rate": 0.0005736685610876274,
      "loss": 2.1445,
      "step": 30957
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6562678217887878,
      "learning_rate": 0.0005736668852337176,
      "loss": 2.3886,
      "step": 30958
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.7045539617538452,
      "learning_rate": 0.000573665209328928,
      "loss": 2.2674,
      "step": 30959
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.7895631790161133,
      "learning_rate": 0.0005736635333732589,
      "loss": 2.2622,
      "step": 30960
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.7453635931015015,
      "learning_rate": 0.0005736618573667107,
      "loss": 2.1578,
      "step": 30961
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.8184843063354492,
      "learning_rate": 0.0005736601813092834,
      "loss": 2.3296,
      "step": 30962
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6491727232933044,
      "learning_rate": 0.0005736585052009775,
      "loss": 2.0173,
      "step": 30963
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.7353441119194031,
      "learning_rate": 0.0005736568290417934,
      "loss": 2.2953,
      "step": 30964
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6377979516983032,
      "learning_rate": 0.0005736551528317313,
      "loss": 2.319,
      "step": 30965
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.8315731883049011,
      "learning_rate": 0.0005736534765707917,
      "loss": 2.2349,
      "step": 30966
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6293266415596008,
      "learning_rate": 0.0005736518002589745,
      "loss": 2.3455,
      "step": 30967
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6056536436080933,
      "learning_rate": 0.0005736501238962804,
      "loss": 2.1494,
      "step": 30968
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6406365633010864,
      "learning_rate": 0.0005736484474827095,
      "loss": 2.1687,
      "step": 30969
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6119071841239929,
      "learning_rate": 0.0005736467710182623,
      "loss": 2.6016,
      "step": 30970
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.5940008163452148,
      "learning_rate": 0.0005736450945029389,
      "loss": 2.3081,
      "step": 30971
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6263171434402466,
      "learning_rate": 0.0005736434179367398,
      "loss": 2.4284,
      "step": 30972
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6850789785385132,
      "learning_rate": 0.0005736417413196652,
      "loss": 2.3258,
      "step": 30973
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6669180393218994,
      "learning_rate": 0.0005736400646517153,
      "loss": 2.484,
      "step": 30974
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.730996310710907,
      "learning_rate": 0.0005736383879328908,
      "loss": 2.3471,
      "step": 30975
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.7871134877204895,
      "learning_rate": 0.0005736367111631916,
      "loss": 2.3649,
      "step": 30976
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.8127910494804382,
      "learning_rate": 0.0005736350343426181,
      "loss": 2.3166,
      "step": 30977
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.8128570318222046,
      "learning_rate": 0.0005736333574711708,
      "loss": 2.0544,
      "step": 30978
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6364457607269287,
      "learning_rate": 0.0005736316805488499,
      "loss": 2.5964,
      "step": 30979
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.646607518196106,
      "learning_rate": 0.0005736300035756556,
      "loss": 2.3515,
      "step": 30980
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.746526300907135,
      "learning_rate": 0.0005736283265515884,
      "loss": 2.2965,
      "step": 30981
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.7871113419532776,
      "learning_rate": 0.0005736266494766486,
      "loss": 2.4887,
      "step": 30982
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.5888394117355347,
      "learning_rate": 0.0005736249723508364,
      "loss": 2.1188,
      "step": 30983
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.7037126421928406,
      "learning_rate": 0.000573623295174152,
      "loss": 2.3819,
      "step": 30984
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.8932756185531616,
      "learning_rate": 0.000573621617946596,
      "loss": 2.4901,
      "step": 30985
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6804355382919312,
      "learning_rate": 0.0005736199406681685,
      "loss": 2.3622,
      "step": 30986
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.7016650438308716,
      "learning_rate": 0.0005736182633388699,
      "loss": 2.2678,
      "step": 30987
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.7633816003799438,
      "learning_rate": 0.0005736165859587006,
      "loss": 2.0529,
      "step": 30988
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6354910135269165,
      "learning_rate": 0.0005736149085276608,
      "loss": 2.4362,
      "step": 30989
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.7341347932815552,
      "learning_rate": 0.0005736132310457507,
      "loss": 2.2899,
      "step": 30990
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6454067230224609,
      "learning_rate": 0.0005736115535129708,
      "loss": 2.3918,
      "step": 30991
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.7530201077461243,
      "learning_rate": 0.0005736098759293215,
      "loss": 2.2185,
      "step": 30992
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.618750274181366,
      "learning_rate": 0.0005736081982948026,
      "loss": 2.4538,
      "step": 30993
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.682692289352417,
      "learning_rate": 0.0005736065206094151,
      "loss": 2.222,
      "step": 30994
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6486653089523315,
      "learning_rate": 0.0005736048428731589,
      "loss": 2.4935,
      "step": 30995
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.617268443107605,
      "learning_rate": 0.0005736031650860344,
      "loss": 2.1348,
      "step": 30996
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.7145793437957764,
      "learning_rate": 0.0005736014872480418,
      "loss": 2.2744,
      "step": 30997
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6386895775794983,
      "learning_rate": 0.0005735998093591816,
      "loss": 2.2407,
      "step": 30998
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6701033115386963,
      "learning_rate": 0.000573598131419454,
      "loss": 2.2484,
      "step": 30999
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6178617477416992,
      "learning_rate": 0.0005735964534288595,
      "loss": 2.393,
      "step": 31000
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6688418388366699,
      "learning_rate": 0.000573594775387398,
      "loss": 2.0587,
      "step": 31001
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6243117451667786,
      "learning_rate": 0.0005735930972950702,
      "loss": 2.4296,
      "step": 31002
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6631291508674622,
      "learning_rate": 0.0005735914191518763,
      "loss": 2.5543,
      "step": 31003
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6456300616264343,
      "learning_rate": 0.0005735897409578166,
      "loss": 2.4333,
      "step": 31004
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6831706166267395,
      "learning_rate": 0.0005735880627128913,
      "loss": 2.2533,
      "step": 31005
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.7293919920921326,
      "learning_rate": 0.0005735863844171009,
      "loss": 1.9265,
      "step": 31006
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6344160437583923,
      "learning_rate": 0.0005735847060704455,
      "loss": 2.2729,
      "step": 31007
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.7020188570022583,
      "learning_rate": 0.0005735830276729257,
      "loss": 2.583,
      "step": 31008
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6557525396347046,
      "learning_rate": 0.0005735813492245416,
      "loss": 2.2996,
      "step": 31009
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6687334179878235,
      "learning_rate": 0.0005735796707252935,
      "loss": 2.1061,
      "step": 31010
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6856352090835571,
      "learning_rate": 0.0005735779921751819,
      "loss": 2.2522,
      "step": 31011
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6508729457855225,
      "learning_rate": 0.0005735763135742069,
      "loss": 1.9395,
      "step": 31012
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6802907586097717,
      "learning_rate": 0.0005735746349223688,
      "loss": 2.5596,
      "step": 31013
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.8743677139282227,
      "learning_rate": 0.0005735729562196683,
      "loss": 2.2921,
      "step": 31014
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.668679416179657,
      "learning_rate": 0.0005735712774661051,
      "loss": 2.2917,
      "step": 31015
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.780335009098053,
      "learning_rate": 0.00057356959866168,
      "loss": 2.2378,
      "step": 31016
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6213598251342773,
      "learning_rate": 0.0005735679198063931,
      "loss": 2.292,
      "step": 31017
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6672201156616211,
      "learning_rate": 0.0005735662409002448,
      "loss": 2.1822,
      "step": 31018
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6616458892822266,
      "learning_rate": 0.0005735645619432354,
      "loss": 2.1609,
      "step": 31019
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.802211344242096,
      "learning_rate": 0.0005735628829353651,
      "loss": 2.3231,
      "step": 31020
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.7176770567893982,
      "learning_rate": 0.0005735612038766344,
      "loss": 2.4691,
      "step": 31021
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6568515300750732,
      "learning_rate": 0.0005735595247670435,
      "loss": 2.4011,
      "step": 31022
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.8467571139335632,
      "learning_rate": 0.0005735578456065927,
      "loss": 2.0773,
      "step": 31023
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.7785478234291077,
      "learning_rate": 0.0005735561663952822,
      "loss": 2.3556,
      "step": 31024
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6523662209510803,
      "learning_rate": 0.0005735544871331126,
      "loss": 2.3451,
      "step": 31025
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.8446966409683228,
      "learning_rate": 0.000573552807820084,
      "loss": 2.2341,
      "step": 31026
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.7311245203018188,
      "learning_rate": 0.0005735511284561968,
      "loss": 2.19,
      "step": 31027
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6198866963386536,
      "learning_rate": 0.0005735494490414512,
      "loss": 2.2355,
      "step": 31028
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.8439937233924866,
      "learning_rate": 0.0005735477695758477,
      "loss": 2.0221,
      "step": 31029
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6565948724746704,
      "learning_rate": 0.0005735460900593865,
      "loss": 2.4839,
      "step": 31030
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6767556667327881,
      "learning_rate": 0.0005735444104920678,
      "loss": 2.3222,
      "step": 31031
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.7083750367164612,
      "learning_rate": 0.0005735427308738921,
      "loss": 2.1412,
      "step": 31032
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.7138035297393799,
      "learning_rate": 0.0005735410512048596,
      "loss": 2.4762,
      "step": 31033
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6662606596946716,
      "learning_rate": 0.0005735393714849708,
      "loss": 2.3557,
      "step": 31034
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.7153778672218323,
      "learning_rate": 0.0005735376917142258,
      "loss": 2.5045,
      "step": 31035
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.7447075247764587,
      "learning_rate": 0.0005735360118926249,
      "loss": 2.4318,
      "step": 31036
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6962849497795105,
      "learning_rate": 0.0005735343320201685,
      "loss": 2.3993,
      "step": 31037
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6529227495193481,
      "learning_rate": 0.000573532652096857,
      "loss": 2.2078,
      "step": 31038
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.7161298394203186,
      "learning_rate": 0.0005735309721226906,
      "loss": 2.3102,
      "step": 31039
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6295125484466553,
      "learning_rate": 0.0005735292920976695,
      "loss": 2.3463,
      "step": 31040
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6273733973503113,
      "learning_rate": 0.0005735276120217942,
      "loss": 2.2768,
      "step": 31041
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6885399222373962,
      "learning_rate": 0.0005735259318950651,
      "loss": 2.168,
      "step": 31042
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6889652609825134,
      "learning_rate": 0.0005735242517174822,
      "loss": 2.3488,
      "step": 31043
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.702961266040802,
      "learning_rate": 0.000573522571489046,
      "loss": 2.2653,
      "step": 31044
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6585432291030884,
      "learning_rate": 0.0005735208912097568,
      "loss": 2.1426,
      "step": 31045
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.7232186198234558,
      "learning_rate": 0.0005735192108796151,
      "loss": 2.4599,
      "step": 31046
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.612136721611023,
      "learning_rate": 0.0005735175304986208,
      "loss": 2.2129,
      "step": 31047
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.7646514177322388,
      "learning_rate": 0.0005735158500667744,
      "loss": 2.5699,
      "step": 31048
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6500151753425598,
      "learning_rate": 0.0005735141695840763,
      "loss": 2.2602,
      "step": 31049
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6669064164161682,
      "learning_rate": 0.0005735124890505267,
      "loss": 2.1279,
      "step": 31050
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.666191577911377,
      "learning_rate": 0.000573510808466126,
      "loss": 2.1887,
      "step": 31051
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.656938910484314,
      "learning_rate": 0.0005735091278308746,
      "loss": 2.3872,
      "step": 31052
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6790696382522583,
      "learning_rate": 0.0005735074471447726,
      "loss": 2.285,
      "step": 31053
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6741310358047485,
      "learning_rate": 0.0005735057664078203,
      "loss": 2.4432,
      "step": 31054
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6830424666404724,
      "learning_rate": 0.0005735040856200183,
      "loss": 2.2685,
      "step": 31055
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6185069680213928,
      "learning_rate": 0.0005735024047813667,
      "loss": 2.3519,
      "step": 31056
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6665610074996948,
      "learning_rate": 0.0005735007238918657,
      "loss": 2.2967,
      "step": 31057
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6765965819358826,
      "learning_rate": 0.0005734990429515159,
      "loss": 2.4344,
      "step": 31058
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6440751552581787,
      "learning_rate": 0.0005734973619603174,
      "loss": 2.5268,
      "step": 31059
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.5993959307670593,
      "learning_rate": 0.0005734956809182706,
      "loss": 2.2927,
      "step": 31060
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.5899458527565002,
      "learning_rate": 0.0005734939998253757,
      "loss": 2.314,
      "step": 31061
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.7197297811508179,
      "learning_rate": 0.0005734923186816333,
      "loss": 2.4581,
      "step": 31062
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6662071347236633,
      "learning_rate": 0.0005734906374870433,
      "loss": 2.4156,
      "step": 31063
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6148014068603516,
      "learning_rate": 0.0005734889562416063,
      "loss": 2.3693,
      "step": 31064
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6202930212020874,
      "learning_rate": 0.0005734872749453226,
      "loss": 2.5699,
      "step": 31065
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.592673659324646,
      "learning_rate": 0.0005734855935981923,
      "loss": 2.2206,
      "step": 31066
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6482453346252441,
      "learning_rate": 0.000573483912200216,
      "loss": 2.5983,
      "step": 31067
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6368158459663391,
      "learning_rate": 0.0005734822307513938,
      "loss": 2.3529,
      "step": 31068
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.7012951374053955,
      "learning_rate": 0.0005734805492517263,
      "loss": 2.3537,
      "step": 31069
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.616992175579071,
      "learning_rate": 0.0005734788677012134,
      "loss": 2.1867,
      "step": 31070
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.7290173768997192,
      "learning_rate": 0.0005734771860998557,
      "loss": 2.2515,
      "step": 31071
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6027730107307434,
      "learning_rate": 0.0005734755044476533,
      "loss": 2.1493,
      "step": 31072
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6763419508934021,
      "learning_rate": 0.0005734738227446067,
      "loss": 2.2951,
      "step": 31073
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6126947999000549,
      "learning_rate": 0.0005734721409907163,
      "loss": 2.286,
      "step": 31074
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.683138370513916,
      "learning_rate": 0.0005734704591859821,
      "loss": 2.1655,
      "step": 31075
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6714671850204468,
      "learning_rate": 0.0005734687773304046,
      "loss": 2.4344,
      "step": 31076
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.7344369888305664,
      "learning_rate": 0.0005734670954239842,
      "loss": 2.4175,
      "step": 31077
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6161763668060303,
      "learning_rate": 0.0005734654134667209,
      "loss": 2.3983,
      "step": 31078
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6357816457748413,
      "learning_rate": 0.0005734637314586154,
      "loss": 2.3214,
      "step": 31079
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6680130958557129,
      "learning_rate": 0.0005734620493996678,
      "loss": 2.2823,
      "step": 31080
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6654557585716248,
      "learning_rate": 0.0005734603672898784,
      "loss": 2.4008,
      "step": 31081
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.5540706515312195,
      "learning_rate": 0.0005734586851292477,
      "loss": 2.5161,
      "step": 31082
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6397289633750916,
      "learning_rate": 0.0005734570029177757,
      "loss": 2.0864,
      "step": 31083
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6387592554092407,
      "learning_rate": 0.0005734553206554629,
      "loss": 2.3668,
      "step": 31084
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6499170660972595,
      "learning_rate": 0.0005734536383423096,
      "loss": 2.2695,
      "step": 31085
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.7475513219833374,
      "learning_rate": 0.0005734519559783161,
      "loss": 2.2886,
      "step": 31086
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6270866990089417,
      "learning_rate": 0.0005734502735634828,
      "loss": 2.3,
      "step": 31087
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6150203943252563,
      "learning_rate": 0.0005734485910978099,
      "loss": 2.2523,
      "step": 31088
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6451670527458191,
      "learning_rate": 0.0005734469085812977,
      "loss": 2.3795,
      "step": 31089
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6338139176368713,
      "learning_rate": 0.0005734452260139466,
      "loss": 2.2791,
      "step": 31090
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6406729221343994,
      "learning_rate": 0.0005734435433957569,
      "loss": 2.6105,
      "step": 31091
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6693191528320312,
      "learning_rate": 0.0005734418607267288,
      "loss": 2.2016,
      "step": 31092
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6788350939750671,
      "learning_rate": 0.0005734401780068627,
      "loss": 2.2166,
      "step": 31093
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6516745090484619,
      "learning_rate": 0.000573438495236159,
      "loss": 2.3303,
      "step": 31094
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6261032819747925,
      "learning_rate": 0.0005734368124146178,
      "loss": 2.3603,
      "step": 31095
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6222416758537292,
      "learning_rate": 0.0005734351295422396,
      "loss": 2.4456,
      "step": 31096
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6313802599906921,
      "learning_rate": 0.0005734334466190247,
      "loss": 2.3844,
      "step": 31097
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.7062538862228394,
      "learning_rate": 0.0005734317636449733,
      "loss": 2.4449,
      "step": 31098
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6655527949333191,
      "learning_rate": 0.0005734300806200858,
      "loss": 2.3336,
      "step": 31099
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6680992245674133,
      "learning_rate": 0.0005734283975443624,
      "loss": 2.4142,
      "step": 31100
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6831476092338562,
      "learning_rate": 0.0005734267144178036,
      "loss": 2.348,
      "step": 31101
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6125192642211914,
      "learning_rate": 0.0005734250312404096,
      "loss": 2.2338,
      "step": 31102
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.660756528377533,
      "learning_rate": 0.0005734233480121806,
      "loss": 2.3139,
      "step": 31103
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.7080127596855164,
      "learning_rate": 0.0005734216647331172,
      "loss": 2.1211,
      "step": 31104
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6775235533714294,
      "learning_rate": 0.0005734199814032196,
      "loss": 2.2194,
      "step": 31105
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6788938045501709,
      "learning_rate": 0.0005734182980224879,
      "loss": 2.3269,
      "step": 31106
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6499666571617126,
      "learning_rate": 0.0005734166145909226,
      "loss": 2.4534,
      "step": 31107
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6603359580039978,
      "learning_rate": 0.000573414931108524,
      "loss": 2.5837,
      "step": 31108
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6824548244476318,
      "learning_rate": 0.0005734132475752925,
      "loss": 2.4553,
      "step": 31109
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6398637890815735,
      "learning_rate": 0.0005734115639912282,
      "loss": 2.3404,
      "step": 31110
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6565589904785156,
      "learning_rate": 0.0005734098803563315,
      "loss": 2.1513,
      "step": 31111
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6484686732292175,
      "learning_rate": 0.0005734081966706028,
      "loss": 2.2416,
      "step": 31112
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6262905597686768,
      "learning_rate": 0.0005734065129340425,
      "loss": 2.6237,
      "step": 31113
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6594515442848206,
      "learning_rate": 0.0005734048291466507,
      "loss": 2.4076,
      "step": 31114
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6265988349914551,
      "learning_rate": 0.0005734031453084276,
      "loss": 2.3008,
      "step": 31115
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.680929958820343,
      "learning_rate": 0.0005734014614193738,
      "loss": 2.4159,
      "step": 31116
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6669438481330872,
      "learning_rate": 0.0005733997774794896,
      "loss": 2.329,
      "step": 31117
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6879364252090454,
      "learning_rate": 0.0005733980934887752,
      "loss": 2.385,
      "step": 31118
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6268215179443359,
      "learning_rate": 0.0005733964094472309,
      "loss": 2.214,
      "step": 31119
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6821433305740356,
      "learning_rate": 0.000573394725354857,
      "loss": 2.1339,
      "step": 31120
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6913339495658875,
      "learning_rate": 0.0005733930412116538,
      "loss": 2.2719,
      "step": 31121
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6150659322738647,
      "learning_rate": 0.0005733913570176219,
      "loss": 2.1171,
      "step": 31122
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6243483424186707,
      "learning_rate": 0.0005733896727727611,
      "loss": 2.2308,
      "step": 31123
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6810407042503357,
      "learning_rate": 0.0005733879884770722,
      "loss": 2.2882,
      "step": 31124
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6976286172866821,
      "learning_rate": 0.0005733863041305552,
      "loss": 2.5119,
      "step": 31125
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6067504286766052,
      "learning_rate": 0.0005733846197332106,
      "loss": 2.1639,
      "step": 31126
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6536661386489868,
      "learning_rate": 0.0005733829352850386,
      "loss": 2.2586,
      "step": 31127
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6669735312461853,
      "learning_rate": 0.0005733812507860396,
      "loss": 2.164,
      "step": 31128
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6527238488197327,
      "learning_rate": 0.0005733795662362139,
      "loss": 2.3292,
      "step": 31129
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.7326814532279968,
      "learning_rate": 0.0005733778816355617,
      "loss": 2.5473,
      "step": 31130
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6434949040412903,
      "learning_rate": 0.0005733761969840833,
      "loss": 2.2654,
      "step": 31131
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6276819109916687,
      "learning_rate": 0.0005733745122817793,
      "loss": 2.5555,
      "step": 31132
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.686525285243988,
      "learning_rate": 0.0005733728275286497,
      "loss": 2.3718,
      "step": 31133
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6934534907341003,
      "learning_rate": 0.0005733711427246951,
      "loss": 2.4533,
      "step": 31134
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.5912068486213684,
      "learning_rate": 0.0005733694578699154,
      "loss": 2.1246,
      "step": 31135
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6612497568130493,
      "learning_rate": 0.0005733677729643112,
      "loss": 2.4596,
      "step": 31136
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6968916654586792,
      "learning_rate": 0.000573366088007883,
      "loss": 2.3392,
      "step": 31137
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.724201500415802,
      "learning_rate": 0.0005733644030006306,
      "loss": 2.4066,
      "step": 31138
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6227759122848511,
      "learning_rate": 0.0005733627179425547,
      "loss": 2.2735,
      "step": 31139
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6365683078765869,
      "learning_rate": 0.0005733610328336556,
      "loss": 2.2841,
      "step": 31140
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6010524034500122,
      "learning_rate": 0.0005733593476739334,
      "loss": 2.2098,
      "step": 31141
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.7276496887207031,
      "learning_rate": 0.0005733576624633886,
      "loss": 2.1257,
      "step": 31142
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6581752300262451,
      "learning_rate": 0.0005733559772020215,
      "loss": 2.2075,
      "step": 31143
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.666657567024231,
      "learning_rate": 0.0005733542918898322,
      "loss": 2.3574,
      "step": 31144
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6895654797554016,
      "learning_rate": 0.0005733526065268213,
      "loss": 2.4009,
      "step": 31145
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6242275238037109,
      "learning_rate": 0.000573350921112989,
      "loss": 2.3671,
      "step": 31146
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.8356873989105225,
      "learning_rate": 0.0005733492356483355,
      "loss": 2.1445,
      "step": 31147
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.7735244035720825,
      "learning_rate": 0.0005733475501328613,
      "loss": 2.3859,
      "step": 31148
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6498472690582275,
      "learning_rate": 0.0005733458645665666,
      "loss": 2.0088,
      "step": 31149
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.7270126342773438,
      "learning_rate": 0.0005733441789494519,
      "loss": 2.3496,
      "step": 31150
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.7716292142868042,
      "learning_rate": 0.0005733424932815171,
      "loss": 2.5334,
      "step": 31151
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6521389484405518,
      "learning_rate": 0.000573340807562763,
      "loss": 2.5209,
      "step": 31152
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.7084344625473022,
      "learning_rate": 0.0005733391217931894,
      "loss": 2.4856,
      "step": 31153
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.9086648225784302,
      "learning_rate": 0.0005733374359727972,
      "loss": 2.2849,
      "step": 31154
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.5802969932556152,
      "learning_rate": 0.0005733357501015862,
      "loss": 2.2837,
      "step": 31155
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.7133983969688416,
      "learning_rate": 0.0005733340641795571,
      "loss": 2.2444,
      "step": 31156
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.7885593771934509,
      "learning_rate": 0.00057333237820671,
      "loss": 2.4854,
      "step": 31157
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.5947861671447754,
      "learning_rate": 0.0005733306921830452,
      "loss": 2.4344,
      "step": 31158
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.7383235096931458,
      "learning_rate": 0.000573329006108563,
      "loss": 2.2332,
      "step": 31159
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.8387277722358704,
      "learning_rate": 0.0005733273199832638,
      "loss": 2.3974,
      "step": 31160
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.5854518413543701,
      "learning_rate": 0.000573325633807148,
      "loss": 2.0703,
      "step": 31161
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.73320072889328,
      "learning_rate": 0.0005733239475802157,
      "loss": 2.2423,
      "step": 31162
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.7135179042816162,
      "learning_rate": 0.0005733222613024673,
      "loss": 2.1228,
      "step": 31163
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6648552417755127,
      "learning_rate": 0.0005733205749739032,
      "loss": 2.4133,
      "step": 31164
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.5957982540130615,
      "learning_rate": 0.0005733188885945237,
      "loss": 2.3981,
      "step": 31165
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.846135675907135,
      "learning_rate": 0.0005733172021643291,
      "loss": 2.2032,
      "step": 31166
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.7053189873695374,
      "learning_rate": 0.0005733155156833194,
      "loss": 2.4909,
      "step": 31167
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6460169553756714,
      "learning_rate": 0.0005733138291514954,
      "loss": 2.3761,
      "step": 31168
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.7031312584877014,
      "learning_rate": 0.0005733121425688572,
      "loss": 2.2569,
      "step": 31169
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6248025298118591,
      "learning_rate": 0.000573310455935405,
      "loss": 2.0678,
      "step": 31170
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6321693062782288,
      "learning_rate": 0.0005733087692511394,
      "loss": 2.5708,
      "step": 31171
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.5800639390945435,
      "learning_rate": 0.0005733070825160604,
      "loss": 2.0887,
      "step": 31172
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.7207503318786621,
      "learning_rate": 0.0005733053957301684,
      "loss": 2.2954,
      "step": 31173
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.7211912870407104,
      "learning_rate": 0.0005733037088934639,
      "loss": 2.1506,
      "step": 31174
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.652826189994812,
      "learning_rate": 0.000573302022005947,
      "loss": 2.5586,
      "step": 31175
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6214460134506226,
      "learning_rate": 0.0005733003350676182,
      "loss": 2.3843,
      "step": 31176
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.7911031246185303,
      "learning_rate": 0.0005732986480784777,
      "loss": 2.213,
      "step": 31177
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.64778733253479,
      "learning_rate": 0.0005732969610385257,
      "loss": 2.4162,
      "step": 31178
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6348686218261719,
      "learning_rate": 0.0005732952739477626,
      "loss": 2.1926,
      "step": 31179
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.77240389585495,
      "learning_rate": 0.0005732935868061889,
      "loss": 2.416,
      "step": 31180
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6913983821868896,
      "learning_rate": 0.0005732918996138049,
      "loss": 2.3983,
      "step": 31181
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6778676509857178,
      "learning_rate": 0.0005732902123706104,
      "loss": 2.1399,
      "step": 31182
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.8084318041801453,
      "learning_rate": 0.0005732885250766064,
      "loss": 2.4468,
      "step": 31183
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.7664639949798584,
      "learning_rate": 0.0005732868377317928,
      "loss": 2.4459,
      "step": 31184
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6823339462280273,
      "learning_rate": 0.0005732851503361699,
      "loss": 2.2604,
      "step": 31185
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6572117805480957,
      "learning_rate": 0.0005732834628897382,
      "loss": 2.3828,
      "step": 31186
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.7182655930519104,
      "learning_rate": 0.0005732817753924981,
      "loss": 2.0837,
      "step": 31187
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.7391439080238342,
      "learning_rate": 0.0005732800878444496,
      "loss": 2.3623,
      "step": 31188
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.756040632724762,
      "learning_rate": 0.0005732784002455933,
      "loss": 2.4007,
      "step": 31189
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6092743277549744,
      "learning_rate": 0.0005732767125959293,
      "loss": 2.3424,
      "step": 31190
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.7079396843910217,
      "learning_rate": 0.0005732750248954579,
      "loss": 2.2092,
      "step": 31191
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.8546048402786255,
      "learning_rate": 0.0005732733371441797,
      "loss": 2.4864,
      "step": 31192
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6753716468811035,
      "learning_rate": 0.0005732716493420946,
      "loss": 2.1562,
      "step": 31193
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6847745180130005,
      "learning_rate": 0.0005732699614892033,
      "loss": 2.2378,
      "step": 31194
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.732020378112793,
      "learning_rate": 0.000573268273585506,
      "loss": 2.3362,
      "step": 31195
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6651498675346375,
      "learning_rate": 0.0005732665856310029,
      "loss": 2.3476,
      "step": 31196
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.7439402937889099,
      "learning_rate": 0.0005732648976256945,
      "loss": 2.4327,
      "step": 31197
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.7831635475158691,
      "learning_rate": 0.0005732632095695808,
      "loss": 2.4964,
      "step": 31198
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.7176942229270935,
      "learning_rate": 0.0005732615214626624,
      "loss": 2.4538,
      "step": 31199
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6745740175247192,
      "learning_rate": 0.0005732598333049395,
      "loss": 2.2349,
      "step": 31200
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6186906695365906,
      "learning_rate": 0.0005732581450964125,
      "loss": 2.1769,
      "step": 31201
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6419808268547058,
      "learning_rate": 0.0005732564568370817,
      "loss": 2.2277,
      "step": 31202
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6861484050750732,
      "learning_rate": 0.0005732547685269472,
      "loss": 2.3811,
      "step": 31203
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.7104219794273376,
      "learning_rate": 0.0005732530801660096,
      "loss": 2.5178,
      "step": 31204
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6960875391960144,
      "learning_rate": 0.0005732513917542691,
      "loss": 2.2511,
      "step": 31205
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6823962330818176,
      "learning_rate": 0.0005732497032917259,
      "loss": 2.2903,
      "step": 31206
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.685608983039856,
      "learning_rate": 0.0005732480147783805,
      "loss": 2.0497,
      "step": 31207
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6591611504554749,
      "learning_rate": 0.0005732463262142332,
      "loss": 2.1249,
      "step": 31208
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.691474974155426,
      "learning_rate": 0.0005732446375992841,
      "loss": 2.3492,
      "step": 31209
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6654725074768066,
      "learning_rate": 0.0005732429489335338,
      "loss": 2.3708,
      "step": 31210
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6611117124557495,
      "learning_rate": 0.0005732412602169825,
      "loss": 2.2914,
      "step": 31211
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6778265833854675,
      "learning_rate": 0.0005732395714496303,
      "loss": 2.3034,
      "step": 31212
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6502302289009094,
      "learning_rate": 0.0005732378826314779,
      "loss": 2.3901,
      "step": 31213
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6763079166412354,
      "learning_rate": 0.0005732361937625253,
      "loss": 2.368,
      "step": 31214
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6533514261245728,
      "learning_rate": 0.000573234504842773,
      "loss": 2.4055,
      "step": 31215
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.67551589012146,
      "learning_rate": 0.0005732328158722211,
      "loss": 2.3756,
      "step": 31216
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6636033058166504,
      "learning_rate": 0.0005732311268508702,
      "loss": 2.2116,
      "step": 31217
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6429091095924377,
      "learning_rate": 0.0005732294377787205,
      "loss": 2.0649,
      "step": 31218
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6632034182548523,
      "learning_rate": 0.0005732277486557723,
      "loss": 2.2933,
      "step": 31219
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6554911136627197,
      "learning_rate": 0.0005732260594820259,
      "loss": 2.4436,
      "step": 31220
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6844475269317627,
      "learning_rate": 0.0005732243702574815,
      "loss": 2.2554,
      "step": 31221
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6561638712882996,
      "learning_rate": 0.0005732226809821396,
      "loss": 2.3176,
      "step": 31222
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.7643623352050781,
      "learning_rate": 0.0005732209916560004,
      "loss": 2.3428,
      "step": 31223
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6943076252937317,
      "learning_rate": 0.0005732193022790643,
      "loss": 2.1766,
      "step": 31224
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6227458715438843,
      "learning_rate": 0.0005732176128513316,
      "loss": 2.2848,
      "step": 31225
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6275932192802429,
      "learning_rate": 0.0005732159233728026,
      "loss": 2.2769,
      "step": 31226
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6254160404205322,
      "learning_rate": 0.0005732142338434776,
      "loss": 2.1461,
      "step": 31227
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.695888102054596,
      "learning_rate": 0.0005732125442633568,
      "loss": 2.2932,
      "step": 31228
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.625690221786499,
      "learning_rate": 0.0005732108546324408,
      "loss": 2.329,
      "step": 31229
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6658254861831665,
      "learning_rate": 0.0005732091649507296,
      "loss": 2.3585,
      "step": 31230
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6772403717041016,
      "learning_rate": 0.0005732074752182238,
      "loss": 2.6209,
      "step": 31231
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.685515820980072,
      "learning_rate": 0.0005732057854349235,
      "loss": 2.2078,
      "step": 31232
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6224311590194702,
      "learning_rate": 0.000573204095600829,
      "loss": 2.0575,
      "step": 31233
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.7175979614257812,
      "learning_rate": 0.0005732024057159409,
      "loss": 2.3373,
      "step": 31234
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6554394960403442,
      "learning_rate": 0.0005732007157802591,
      "loss": 2.4719,
      "step": 31235
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6172284483909607,
      "learning_rate": 0.0005731990257937842,
      "loss": 2.2562,
      "step": 31236
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6938462853431702,
      "learning_rate": 0.0005731973357565166,
      "loss": 2.1521,
      "step": 31237
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6967849731445312,
      "learning_rate": 0.0005731956456684563,
      "loss": 2.254,
      "step": 31238
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.78810715675354,
      "learning_rate": 0.0005731939555296037,
      "loss": 2.3348,
      "step": 31239
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6476951837539673,
      "learning_rate": 0.0005731922653399594,
      "loss": 2.061,
      "step": 31240
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6827951669692993,
      "learning_rate": 0.0005731905750995233,
      "loss": 2.0961,
      "step": 31241
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6310334801673889,
      "learning_rate": 0.0005731888848082962,
      "loss": 2.0926,
      "step": 31242
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6735228896141052,
      "learning_rate": 0.0005731871944662778,
      "loss": 2.2293,
      "step": 31243
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6961188316345215,
      "learning_rate": 0.0005731855040734689,
      "loss": 2.3782,
      "step": 31244
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6626983284950256,
      "learning_rate": 0.0005731838136298696,
      "loss": 2.1804,
      "step": 31245
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6785886287689209,
      "learning_rate": 0.0005731821231354803,
      "loss": 2.3431,
      "step": 31246
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.7207052111625671,
      "learning_rate": 0.0005731804325903014,
      "loss": 2.3386,
      "step": 31247
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6079649329185486,
      "learning_rate": 0.000573178741994333,
      "loss": 2.4824,
      "step": 31248
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6457898020744324,
      "learning_rate": 0.0005731770513475755,
      "loss": 2.168,
      "step": 31249
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.7274050712585449,
      "learning_rate": 0.0005731753606500293,
      "loss": 2.3491,
      "step": 31250
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6851877570152283,
      "learning_rate": 0.0005731736699016946,
      "loss": 2.2456,
      "step": 31251
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.646981954574585,
      "learning_rate": 0.0005731719791025716,
      "loss": 2.4159,
      "step": 31252
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6632049679756165,
      "learning_rate": 0.000573170288252661,
      "loss": 2.4237,
      "step": 31253
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6771642565727234,
      "learning_rate": 0.0005731685973519628,
      "loss": 2.2552,
      "step": 31254
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6951766610145569,
      "learning_rate": 0.0005731669064004774,
      "loss": 2.2685,
      "step": 31255
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.673265814781189,
      "learning_rate": 0.0005731652153982052,
      "loss": 2.5211,
      "step": 31256
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6712045073509216,
      "learning_rate": 0.0005731635243451463,
      "loss": 2.2917,
      "step": 31257
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6559250354766846,
      "learning_rate": 0.0005731618332413012,
      "loss": 2.2515,
      "step": 31258
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.7299274206161499,
      "learning_rate": 0.0005731601420866701,
      "loss": 2.2702,
      "step": 31259
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.7280707955360413,
      "learning_rate": 0.0005731584508812535,
      "loss": 2.3976,
      "step": 31260
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6815271377563477,
      "learning_rate": 0.0005731567596250515,
      "loss": 2.2246,
      "step": 31261
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.7722386121749878,
      "learning_rate": 0.0005731550683180645,
      "loss": 2.2949,
      "step": 31262
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6456601023674011,
      "learning_rate": 0.0005731533769602929,
      "loss": 2.3137,
      "step": 31263
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.656304121017456,
      "learning_rate": 0.0005731516855517368,
      "loss": 2.2559,
      "step": 31264
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6722200512886047,
      "learning_rate": 0.0005731499940923967,
      "loss": 2.2672,
      "step": 31265
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.7156818509101868,
      "learning_rate": 0.0005731483025822729,
      "loss": 2.5124,
      "step": 31266
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6920783519744873,
      "learning_rate": 0.0005731466110213656,
      "loss": 2.0994,
      "step": 31267
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6400395631790161,
      "learning_rate": 0.0005731449194096753,
      "loss": 2.2894,
      "step": 31268
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6205806136131287,
      "learning_rate": 0.0005731432277472022,
      "loss": 2.3496,
      "step": 31269
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6685386300086975,
      "learning_rate": 0.0005731415360339465,
      "loss": 2.1473,
      "step": 31270
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.7660852074623108,
      "learning_rate": 0.0005731398442699087,
      "loss": 2.3372,
      "step": 31271
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6430100202560425,
      "learning_rate": 0.0005731381524550891,
      "loss": 2.2639,
      "step": 31272
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6957123279571533,
      "learning_rate": 0.0005731364605894878,
      "loss": 2.1344,
      "step": 31273
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6899083852767944,
      "learning_rate": 0.0005731347686731054,
      "loss": 2.3361,
      "step": 31274
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6951159834861755,
      "learning_rate": 0.000573133076705942,
      "loss": 2.4895,
      "step": 31275
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6370092034339905,
      "learning_rate": 0.0005731313846879982,
      "loss": 2.2899,
      "step": 31276
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.7467067837715149,
      "learning_rate": 0.000573129692619274,
      "loss": 2.3056,
      "step": 31277
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.7178422212600708,
      "learning_rate": 0.0005731280004997697,
      "loss": 2.2328,
      "step": 31278
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6620581746101379,
      "learning_rate": 0.0005731263083294859,
      "loss": 2.4332,
      "step": 31279
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.7750993967056274,
      "learning_rate": 0.0005731246161084228,
      "loss": 2.5361,
      "step": 31280
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6799381375312805,
      "learning_rate": 0.0005731229238365806,
      "loss": 2.3445,
      "step": 31281
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6349753737449646,
      "learning_rate": 0.0005731212315139597,
      "loss": 2.3337,
      "step": 31282
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.7256988286972046,
      "learning_rate": 0.0005731195391405604,
      "loss": 2.2272,
      "step": 31283
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6452035307884216,
      "learning_rate": 0.0005731178467163831,
      "loss": 2.2087,
      "step": 31284
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6635401844978333,
      "learning_rate": 0.0005731161542414279,
      "loss": 2.1939,
      "step": 31285
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6593509316444397,
      "learning_rate": 0.0005731144617156953,
      "loss": 2.2223,
      "step": 31286
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6353408098220825,
      "learning_rate": 0.0005731127691391857,
      "loss": 2.2332,
      "step": 31287
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.622542679309845,
      "learning_rate": 0.0005731110765118991,
      "loss": 2.1457,
      "step": 31288
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6593855619430542,
      "learning_rate": 0.0005731093838338361,
      "loss": 2.3136,
      "step": 31289
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6282706260681152,
      "learning_rate": 0.0005731076911049969,
      "loss": 2.1819,
      "step": 31290
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6204907298088074,
      "learning_rate": 0.0005731059983253819,
      "loss": 2.2689,
      "step": 31291
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.7084484100341797,
      "learning_rate": 0.0005731043054949911,
      "loss": 2.4575,
      "step": 31292
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.7069023251533508,
      "learning_rate": 0.0005731026126138253,
      "loss": 2.442,
      "step": 31293
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6001332998275757,
      "learning_rate": 0.0005731009196818845,
      "loss": 2.1561,
      "step": 31294
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6822465658187866,
      "learning_rate": 0.000573099226699169,
      "loss": 2.2043,
      "step": 31295
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6567630767822266,
      "learning_rate": 0.0005730975336656792,
      "loss": 2.4288,
      "step": 31296
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.667288064956665,
      "learning_rate": 0.0005730958405814155,
      "loss": 2.4067,
      "step": 31297
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6247636079788208,
      "learning_rate": 0.0005730941474463781,
      "loss": 2.5562,
      "step": 31298
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.675155520439148,
      "learning_rate": 0.0005730924542605674,
      "loss": 2.4041,
      "step": 31299
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.5985844135284424,
      "learning_rate": 0.0005730907610239835,
      "loss": 2.2299,
      "step": 31300
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6008226275444031,
      "learning_rate": 0.000573089067736627,
      "loss": 2.405,
      "step": 31301
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.5971868634223938,
      "learning_rate": 0.000573087374398498,
      "loss": 2.1676,
      "step": 31302
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6702218055725098,
      "learning_rate": 0.000573085681009597,
      "loss": 2.2177,
      "step": 31303
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.667773425579071,
      "learning_rate": 0.0005730839875699242,
      "loss": 2.4253,
      "step": 31304
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6223222017288208,
      "learning_rate": 0.0005730822940794799,
      "loss": 2.5185,
      "step": 31305
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6773566603660583,
      "learning_rate": 0.0005730806005382644,
      "loss": 2.4583,
      "step": 31306
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6277980804443359,
      "learning_rate": 0.0005730789069462781,
      "loss": 2.3245,
      "step": 31307
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6334816813468933,
      "learning_rate": 0.0005730772133035213,
      "loss": 2.4094,
      "step": 31308
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6540951728820801,
      "learning_rate": 0.0005730755196099943,
      "loss": 2.2634,
      "step": 31309
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.7691770195960999,
      "learning_rate": 0.0005730738258656973,
      "loss": 2.3561,
      "step": 31310
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6593034863471985,
      "learning_rate": 0.0005730721320706309,
      "loss": 2.3836,
      "step": 31311
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6704745888710022,
      "learning_rate": 0.0005730704382247952,
      "loss": 2.4428,
      "step": 31312
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6640346050262451,
      "learning_rate": 0.0005730687443281905,
      "loss": 2.4023,
      "step": 31313
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6544334888458252,
      "learning_rate": 0.000573067050380817,
      "loss": 2.3778,
      "step": 31314
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6389174461364746,
      "learning_rate": 0.0005730653563826754,
      "loss": 2.2279,
      "step": 31315
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6596701145172119,
      "learning_rate": 0.0005730636623337658,
      "loss": 2.2653,
      "step": 31316
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.7003281712532043,
      "learning_rate": 0.0005730619682340884,
      "loss": 2.3338,
      "step": 31317
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6780429482460022,
      "learning_rate": 0.0005730602740836437,
      "loss": 2.59,
      "step": 31318
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.7659967541694641,
      "learning_rate": 0.0005730585798824319,
      "loss": 2.3994,
      "step": 31319
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.636588454246521,
      "learning_rate": 0.0005730568856304534,
      "loss": 2.3715,
      "step": 31320
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6329331398010254,
      "learning_rate": 0.0005730551913277084,
      "loss": 2.2507,
      "step": 31321
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.7000955939292908,
      "learning_rate": 0.0005730534969741974,
      "loss": 2.5332,
      "step": 31322
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6848188638687134,
      "learning_rate": 0.0005730518025699204,
      "loss": 2.3597,
      "step": 31323
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6947100162506104,
      "learning_rate": 0.0005730501081148781,
      "loss": 2.338,
      "step": 31324
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6728481650352478,
      "learning_rate": 0.0005730484136090707,
      "loss": 2.4551,
      "step": 31325
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6220880746841431,
      "learning_rate": 0.0005730467190524982,
      "loss": 2.2885,
      "step": 31326
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.7075985670089722,
      "learning_rate": 0.0005730450244451613,
      "loss": 2.3198,
      "step": 31327
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6098732948303223,
      "learning_rate": 0.0005730433297870602,
      "loss": 2.0388,
      "step": 31328
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6324723362922668,
      "learning_rate": 0.000573041635078195,
      "loss": 2.1931,
      "step": 31329
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.774204671382904,
      "learning_rate": 0.0005730399403185664,
      "loss": 2.2854,
      "step": 31330
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6387521028518677,
      "learning_rate": 0.0005730382455081746,
      "loss": 2.2838,
      "step": 31331
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6356930136680603,
      "learning_rate": 0.0005730365506470196,
      "loss": 2.1299,
      "step": 31332
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6902785897254944,
      "learning_rate": 0.0005730348557351021,
      "loss": 2.382,
      "step": 31333
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6994406580924988,
      "learning_rate": 0.0005730331607724222,
      "loss": 2.7006,
      "step": 31334
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6119370460510254,
      "learning_rate": 0.0005730314657589803,
      "loss": 2.2455,
      "step": 31335
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6231346726417542,
      "learning_rate": 0.0005730297706947767,
      "loss": 2.1002,
      "step": 31336
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6544083952903748,
      "learning_rate": 0.0005730280755798118,
      "loss": 2.4709,
      "step": 31337
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6748391389846802,
      "learning_rate": 0.0005730263804140857,
      "loss": 2.2706,
      "step": 31338
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.7173226475715637,
      "learning_rate": 0.0005730246851975988,
      "loss": 2.3295,
      "step": 31339
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6346267461776733,
      "learning_rate": 0.0005730229899303516,
      "loss": 2.3171,
      "step": 31340
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.7017590999603271,
      "learning_rate": 0.0005730212946123442,
      "loss": 2.1455,
      "step": 31341
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6539033055305481,
      "learning_rate": 0.000573019599243577,
      "loss": 2.3529,
      "step": 31342
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6795399188995361,
      "learning_rate": 0.0005730179038240504,
      "loss": 2.2569,
      "step": 31343
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.7673994898796082,
      "learning_rate": 0.0005730162083537644,
      "loss": 2.256,
      "step": 31344
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6573729515075684,
      "learning_rate": 0.0005730145128327197,
      "loss": 2.1669,
      "step": 31345
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6617002487182617,
      "learning_rate": 0.0005730128172609165,
      "loss": 2.4223,
      "step": 31346
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6805531978607178,
      "learning_rate": 0.0005730111216383549,
      "loss": 2.3836,
      "step": 31347
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6721923351287842,
      "learning_rate": 0.0005730094259650355,
      "loss": 2.1679,
      "step": 31348
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6376718282699585,
      "learning_rate": 0.0005730077302409584,
      "loss": 2.3092,
      "step": 31349
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6423225998878479,
      "learning_rate": 0.000573006034466124,
      "loss": 2.3769,
      "step": 31350
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6112431287765503,
      "learning_rate": 0.0005730043386405327,
      "loss": 2.2445,
      "step": 31351
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.7109572291374207,
      "learning_rate": 0.0005730026427641847,
      "loss": 2.5816,
      "step": 31352
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6604889035224915,
      "learning_rate": 0.0005730009468370805,
      "loss": 2.4072,
      "step": 31353
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.7332929372787476,
      "learning_rate": 0.00057299925085922,
      "loss": 2.3856,
      "step": 31354
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.603825032711029,
      "learning_rate": 0.000572997554830604,
      "loss": 2.1827,
      "step": 31355
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6239510178565979,
      "learning_rate": 0.0005729958587512325,
      "loss": 2.2062,
      "step": 31356
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.606936514377594,
      "learning_rate": 0.000572994162621106,
      "loss": 2.3062,
      "step": 31357
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6483262181282043,
      "learning_rate": 0.0005729924664402247,
      "loss": 2.1371,
      "step": 31358
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6682211756706238,
      "learning_rate": 0.0005729907702085889,
      "loss": 2.2189,
      "step": 31359
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6139503717422485,
      "learning_rate": 0.0005729890739261989,
      "loss": 2.0742,
      "step": 31360
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6798850893974304,
      "learning_rate": 0.0005729873775930553,
      "loss": 2.1956,
      "step": 31361
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6350980997085571,
      "learning_rate": 0.000572985681209158,
      "loss": 2.4322,
      "step": 31362
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6486425399780273,
      "learning_rate": 0.0005729839847745076,
      "loss": 2.4348,
      "step": 31363
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.738954484462738,
      "learning_rate": 0.0005729822882891043,
      "loss": 2.4073,
      "step": 31364
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6641376614570618,
      "learning_rate": 0.0005729805917529484,
      "loss": 2.1968,
      "step": 31365
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.7666944265365601,
      "learning_rate": 0.0005729788951660403,
      "loss": 2.2851,
      "step": 31366
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6820623874664307,
      "learning_rate": 0.0005729771985283803,
      "loss": 2.2554,
      "step": 31367
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6989865899085999,
      "learning_rate": 0.0005729755018399687,
      "loss": 2.2919,
      "step": 31368
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.687566339969635,
      "learning_rate": 0.0005729738051008056,
      "loss": 2.3884,
      "step": 31369
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6679155230522156,
      "learning_rate": 0.0005729721083108917,
      "loss": 2.3384,
      "step": 31370
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6409580707550049,
      "learning_rate": 0.000572970411470227,
      "loss": 2.4802,
      "step": 31371
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6580262184143066,
      "learning_rate": 0.0005729687145788121,
      "loss": 2.4596,
      "step": 31372
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6666450500488281,
      "learning_rate": 0.0005729670176366471,
      "loss": 2.3027,
      "step": 31373
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6386538147926331,
      "learning_rate": 0.0005729653206437324,
      "loss": 2.0774,
      "step": 31374
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.660061240196228,
      "learning_rate": 0.0005729636236000683,
      "loss": 2.1595,
      "step": 31375
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6446110010147095,
      "learning_rate": 0.0005729619265056552,
      "loss": 2.4736,
      "step": 31376
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.656446635723114,
      "learning_rate": 0.0005729602293604931,
      "loss": 2.3683,
      "step": 31377
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6847703456878662,
      "learning_rate": 0.0005729585321645826,
      "loss": 2.2751,
      "step": 31378
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.7442610263824463,
      "learning_rate": 0.0005729568349179241,
      "loss": 2.4761,
      "step": 31379
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6797969937324524,
      "learning_rate": 0.0005729551376205177,
      "loss": 2.3216,
      "step": 31380
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6804021596908569,
      "learning_rate": 0.0005729534402723637,
      "loss": 2.3416,
      "step": 31381
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6693112254142761,
      "learning_rate": 0.0005729517428734627,
      "loss": 2.2643,
      "step": 31382
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.7731470465660095,
      "learning_rate": 0.0005729500454238147,
      "loss": 2.4731,
      "step": 31383
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.7836915850639343,
      "learning_rate": 0.0005729483479234201,
      "loss": 2.1235,
      "step": 31384
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6740207672119141,
      "learning_rate": 0.0005729466503722793,
      "loss": 2.3257,
      "step": 31385
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6797532439231873,
      "learning_rate": 0.0005729449527703925,
      "loss": 2.5539,
      "step": 31386
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.7725704312324524,
      "learning_rate": 0.0005729432551177602,
      "loss": 2.4651,
      "step": 31387
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6307283639907837,
      "learning_rate": 0.0005729415574143824,
      "loss": 2.2634,
      "step": 31388
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6921427845954895,
      "learning_rate": 0.0005729398596602598,
      "loss": 2.1308,
      "step": 31389
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.699618935585022,
      "learning_rate": 0.0005729381618553924,
      "loss": 2.49,
      "step": 31390
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6814497113227844,
      "learning_rate": 0.0005729364639997807,
      "loss": 2.5009,
      "step": 31391
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.7211318612098694,
      "learning_rate": 0.000572934766093425,
      "loss": 2.426,
      "step": 31392
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.695440948009491,
      "learning_rate": 0.0005729330681363254,
      "loss": 2.3855,
      "step": 31393
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6584886312484741,
      "learning_rate": 0.0005729313701284826,
      "loss": 2.1902,
      "step": 31394
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.7398767471313477,
      "learning_rate": 0.0005729296720698967,
      "loss": 2.2524,
      "step": 31395
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.677984893321991,
      "learning_rate": 0.000572927973960568,
      "loss": 2.1721,
      "step": 31396
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6808719038963318,
      "learning_rate": 0.0005729262758004968,
      "loss": 2.3249,
      "step": 31397
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.754846453666687,
      "learning_rate": 0.0005729245775896834,
      "loss": 2.3042,
      "step": 31398
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6698136329650879,
      "learning_rate": 0.0005729228793281282,
      "loss": 2.351,
      "step": 31399
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6328748464584351,
      "learning_rate": 0.0005729211810158316,
      "loss": 2.3679,
      "step": 31400
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6436738967895508,
      "learning_rate": 0.0005729194826527936,
      "loss": 2.0924,
      "step": 31401
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6549003720283508,
      "learning_rate": 0.0005729177842390148,
      "loss": 2.4578,
      "step": 31402
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6421467065811157,
      "learning_rate": 0.0005729160857744955,
      "loss": 2.3034,
      "step": 31403
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6475361585617065,
      "learning_rate": 0.0005729143872592359,
      "loss": 2.2136,
      "step": 31404
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6462585926055908,
      "learning_rate": 0.0005729126886932364,
      "loss": 2.3958,
      "step": 31405
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6104583144187927,
      "learning_rate": 0.0005729109900764972,
      "loss": 2.2711,
      "step": 31406
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6217634677886963,
      "learning_rate": 0.0005729092914090187,
      "loss": 2.4237,
      "step": 31407
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.7094323039054871,
      "learning_rate": 0.0005729075926908013,
      "loss": 2.3165,
      "step": 31408
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.664637565612793,
      "learning_rate": 0.0005729058939218451,
      "loss": 2.2192,
      "step": 31409
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6942546963691711,
      "learning_rate": 0.0005729041951021506,
      "loss": 2.2232,
      "step": 31410
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6465075016021729,
      "learning_rate": 0.0005729024962317182,
      "loss": 2.4339,
      "step": 31411
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6841621398925781,
      "learning_rate": 0.0005729007973105479,
      "loss": 2.2944,
      "step": 31412
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.7080556154251099,
      "learning_rate": 0.0005728990983386403,
      "loss": 2.3892,
      "step": 31413
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.7094151377677917,
      "learning_rate": 0.0005728973993159955,
      "loss": 2.3297,
      "step": 31414
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6287119388580322,
      "learning_rate": 0.000572895700242614,
      "loss": 2.2363,
      "step": 31415
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.7073306441307068,
      "learning_rate": 0.000572894001118496,
      "loss": 2.3324,
      "step": 31416
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.92380291223526,
      "learning_rate": 0.0005728923019436418,
      "loss": 2.2992,
      "step": 31417
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.674507200717926,
      "learning_rate": 0.0005728906027180518,
      "loss": 2.3342,
      "step": 31418
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6886898875236511,
      "learning_rate": 0.0005728889034417264,
      "loss": 2.3212,
      "step": 31419
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6372742652893066,
      "learning_rate": 0.0005728872041146657,
      "loss": 2.2138,
      "step": 31420
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6713815331459045,
      "learning_rate": 0.0005728855047368702,
      "loss": 2.3168,
      "step": 31421
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.7327558398246765,
      "learning_rate": 0.00057288380530834,
      "loss": 2.3059,
      "step": 31422
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6903253197669983,
      "learning_rate": 0.0005728821058290757,
      "loss": 2.4286,
      "step": 31423
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6296640038490295,
      "learning_rate": 0.0005728804062990773,
      "loss": 2.2531,
      "step": 31424
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6844133734703064,
      "learning_rate": 0.0005728787067183454,
      "loss": 2.3278,
      "step": 31425
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6187381148338318,
      "learning_rate": 0.0005728770070868802,
      "loss": 2.3524,
      "step": 31426
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6417437791824341,
      "learning_rate": 0.0005728753074046819,
      "loss": 2.3836,
      "step": 31427
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.652642548084259,
      "learning_rate": 0.000572873607671751,
      "loss": 2.3593,
      "step": 31428
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.7160079479217529,
      "learning_rate": 0.0005728719078880878,
      "loss": 2.2685,
      "step": 31429
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6388072371482849,
      "learning_rate": 0.0005728702080536925,
      "loss": 2.1949,
      "step": 31430
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6670088171958923,
      "learning_rate": 0.0005728685081685655,
      "loss": 2.3045,
      "step": 31431
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6349270343780518,
      "learning_rate": 0.0005728668082327071,
      "loss": 2.2169,
      "step": 31432
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6684441566467285,
      "learning_rate": 0.0005728651082461176,
      "loss": 2.3705,
      "step": 31433
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6585982441902161,
      "learning_rate": 0.0005728634082087973,
      "loss": 2.1907,
      "step": 31434
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6019290685653687,
      "learning_rate": 0.0005728617081207466,
      "loss": 2.1961,
      "step": 31435
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.7260528802871704,
      "learning_rate": 0.0005728600079819658,
      "loss": 2.2561,
      "step": 31436
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.5863932967185974,
      "learning_rate": 0.000572858307792455,
      "loss": 2.2262,
      "step": 31437
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6542316675186157,
      "learning_rate": 0.0005728566075522148,
      "loss": 2.451,
      "step": 31438
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.7130244970321655,
      "learning_rate": 0.0005728549072612454,
      "loss": 2.2952,
      "step": 31439
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.7132940292358398,
      "learning_rate": 0.0005728532069195472,
      "loss": 2.4131,
      "step": 31440
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.5738420486450195,
      "learning_rate": 0.0005728515065271203,
      "loss": 2.2092,
      "step": 31441
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.7032821178436279,
      "learning_rate": 0.0005728498060839652,
      "loss": 2.1152,
      "step": 31442
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.7279995083808899,
      "learning_rate": 0.0005728481055900822,
      "loss": 2.4464,
      "step": 31443
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6574634909629822,
      "learning_rate": 0.0005728464050454715,
      "loss": 2.3108,
      "step": 31444
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.8426153063774109,
      "learning_rate": 0.0005728447044501336,
      "loss": 2.3318,
      "step": 31445
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6080814599990845,
      "learning_rate": 0.0005728430038040687,
      "loss": 2.3124,
      "step": 31446
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6928046941757202,
      "learning_rate": 0.0005728413031072771,
      "loss": 2.4798,
      "step": 31447
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6714719533920288,
      "learning_rate": 0.0005728396023597592,
      "loss": 2.3084,
      "step": 31448
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6154125928878784,
      "learning_rate": 0.0005728379015615152,
      "loss": 2.3176,
      "step": 31449
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6893122792243958,
      "learning_rate": 0.0005728362007125455,
      "loss": 2.3146,
      "step": 31450
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.8677617311477661,
      "learning_rate": 0.0005728344998128504,
      "loss": 2.3196,
      "step": 31451
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.7257934212684631,
      "learning_rate": 0.0005728327988624303,
      "loss": 2.4664,
      "step": 31452
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.596747875213623,
      "learning_rate": 0.0005728310978612854,
      "loss": 2.2272,
      "step": 31453
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6761406064033508,
      "learning_rate": 0.000572829396809416,
      "loss": 2.3138,
      "step": 31454
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6830227375030518,
      "learning_rate": 0.0005728276957068225,
      "loss": 2.2824,
      "step": 31455
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6947154998779297,
      "learning_rate": 0.0005728259945535052,
      "loss": 2.4458,
      "step": 31456
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6742477416992188,
      "learning_rate": 0.0005728242933494644,
      "loss": 2.6153,
      "step": 31457
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6457191109657288,
      "learning_rate": 0.0005728225920947003,
      "loss": 2.2561,
      "step": 31458
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.5778645873069763,
      "learning_rate": 0.0005728208907892134,
      "loss": 2.5738,
      "step": 31459
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.7133086323738098,
      "learning_rate": 0.0005728191894330039,
      "loss": 2.3144,
      "step": 31460
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6620266437530518,
      "learning_rate": 0.0005728174880260722,
      "loss": 2.5471,
      "step": 31461
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6966380476951599,
      "learning_rate": 0.0005728157865684186,
      "loss": 2.2329,
      "step": 31462
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6635578870773315,
      "learning_rate": 0.0005728140850600435,
      "loss": 2.6429,
      "step": 31463
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.5842204093933105,
      "learning_rate": 0.0005728123835009469,
      "loss": 2.3811,
      "step": 31464
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6917840838432312,
      "learning_rate": 0.0005728106818911295,
      "loss": 2.2043,
      "step": 31465
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6372070908546448,
      "learning_rate": 0.0005728089802305914,
      "loss": 2.3099,
      "step": 31466
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.65451580286026,
      "learning_rate": 0.000572807278519333,
      "loss": 2.2695,
      "step": 31467
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6945058703422546,
      "learning_rate": 0.0005728055767573545,
      "loss": 2.3216,
      "step": 31468
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6312007904052734,
      "learning_rate": 0.0005728038749446562,
      "loss": 2.1162,
      "step": 31469
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6169517040252686,
      "learning_rate": 0.0005728021730812387,
      "loss": 2.2102,
      "step": 31470
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.587721049785614,
      "learning_rate": 0.000572800471167102,
      "loss": 2.3859,
      "step": 31471
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6266893148422241,
      "learning_rate": 0.0005727987692022466,
      "loss": 2.2682,
      "step": 31472
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6978000998497009,
      "learning_rate": 0.0005727970671866728,
      "loss": 2.3644,
      "step": 31473
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6617335081100464,
      "learning_rate": 0.0005727953651203808,
      "loss": 2.4169,
      "step": 31474
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6660628914833069,
      "learning_rate": 0.000572793663003371,
      "loss": 2.3094,
      "step": 31475
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.660522997379303,
      "learning_rate": 0.0005727919608356438,
      "loss": 2.3742,
      "step": 31476
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6522355675697327,
      "learning_rate": 0.0005727902586171994,
      "loss": 2.3679,
      "step": 31477
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6313214302062988,
      "learning_rate": 0.000572788556348038,
      "loss": 2.155,
      "step": 31478
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6135173439979553,
      "learning_rate": 0.0005727868540281601,
      "loss": 2.315,
      "step": 31479
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6463488936424255,
      "learning_rate": 0.0005727851516575661,
      "loss": 2.0344,
      "step": 31480
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.5993614792823792,
      "learning_rate": 0.000572783449236256,
      "loss": 2.3344,
      "step": 31481
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6046724319458008,
      "learning_rate": 0.0005727817467642305,
      "loss": 2.2384,
      "step": 31482
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.7158229947090149,
      "learning_rate": 0.0005727800442414897,
      "loss": 2.2585,
      "step": 31483
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6883325576782227,
      "learning_rate": 0.0005727783416680338,
      "loss": 2.1399,
      "step": 31484
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6835439801216125,
      "learning_rate": 0.0005727766390438634,
      "loss": 2.3791,
      "step": 31485
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6243674159049988,
      "learning_rate": 0.0005727749363689786,
      "loss": 2.1957,
      "step": 31486
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6366450786590576,
      "learning_rate": 0.0005727732336433798,
      "loss": 2.5975,
      "step": 31487
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.7016605734825134,
      "learning_rate": 0.0005727715308670673,
      "loss": 2.1957,
      "step": 31488
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6213163137435913,
      "learning_rate": 0.0005727698280400415,
      "loss": 2.1253,
      "step": 31489
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6855854988098145,
      "learning_rate": 0.0005727681251623026,
      "loss": 2.2842,
      "step": 31490
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6391335129737854,
      "learning_rate": 0.0005727664222338509,
      "loss": 2.6017,
      "step": 31491
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.7094929814338684,
      "learning_rate": 0.0005727647192546868,
      "loss": 2.3322,
      "step": 31492
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.7046436071395874,
      "learning_rate": 0.0005727630162248106,
      "loss": 2.305,
      "step": 31493
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6820918321609497,
      "learning_rate": 0.0005727613131442225,
      "loss": 2.3903,
      "step": 31494
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6853892803192139,
      "learning_rate": 0.0005727596100129231,
      "loss": 2.2725,
      "step": 31495
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6981908679008484,
      "learning_rate": 0.0005727579068309125,
      "loss": 2.1534,
      "step": 31496
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6216048002243042,
      "learning_rate": 0.0005727562035981912,
      "loss": 2.225,
      "step": 31497
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.8030561804771423,
      "learning_rate": 0.0005727545003147591,
      "loss": 2.3504,
      "step": 31498
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.712558388710022,
      "learning_rate": 0.0005727527969806167,
      "loss": 2.1969,
      "step": 31499
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6752310991287231,
      "learning_rate": 0.0005727510935957647,
      "loss": 2.4316,
      "step": 31500
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.7827302813529968,
      "learning_rate": 0.000572749390160203,
      "loss": 2.2321,
      "step": 31501
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.5906562805175781,
      "learning_rate": 0.0005727476866739321,
      "loss": 2.2111,
      "step": 31502
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6362890005111694,
      "learning_rate": 0.0005727459831369522,
      "loss": 2.2441,
      "step": 31503
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.7449247241020203,
      "learning_rate": 0.0005727442795492638,
      "loss": 2.3577,
      "step": 31504
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.694600522518158,
      "learning_rate": 0.0005727425759108668,
      "loss": 2.2472,
      "step": 31505
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6177834868431091,
      "learning_rate": 0.000572740872221762,
      "loss": 2.1683,
      "step": 31506
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.020612120628357,
      "learning_rate": 0.0005727391684819496,
      "loss": 2.2038,
      "step": 31507
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.8067914843559265,
      "learning_rate": 0.0005727374646914298,
      "loss": 2.3713,
      "step": 31508
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.7335782647132874,
      "learning_rate": 0.0005727357608502028,
      "loss": 2.1714,
      "step": 31509
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.1138602495193481,
      "learning_rate": 0.0005727340569582692,
      "loss": 2.6543,
      "step": 31510
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.7971373200416565,
      "learning_rate": 0.0005727323530156293,
      "loss": 2.4011,
      "step": 31511
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6950727105140686,
      "learning_rate": 0.0005727306490222831,
      "loss": 2.4232,
      "step": 31512
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.8165121674537659,
      "learning_rate": 0.0005727289449782312,
      "loss": 2.4669,
      "step": 31513
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.7309545278549194,
      "learning_rate": 0.0005727272408834737,
      "loss": 2.3493,
      "step": 31514
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6653085350990295,
      "learning_rate": 0.0005727255367380113,
      "loss": 2.3601,
      "step": 31515
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6667670011520386,
      "learning_rate": 0.0005727238325418441,
      "loss": 2.3134,
      "step": 31516
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.8736110329627991,
      "learning_rate": 0.0005727221282949721,
      "loss": 2.3934,
      "step": 31517
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.750137209892273,
      "learning_rate": 0.0005727204239973962,
      "loss": 2.4535,
      "step": 31518
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6976584196090698,
      "learning_rate": 0.0005727187196491163,
      "loss": 2.2424,
      "step": 31519
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6316179037094116,
      "learning_rate": 0.0005727170152501328,
      "loss": 2.5808,
      "step": 31520
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.8867489099502563,
      "learning_rate": 0.0005727153108004462,
      "loss": 2.2705,
      "step": 31521
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6309951543807983,
      "learning_rate": 0.0005727136063000565,
      "loss": 2.3348,
      "step": 31522
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6565919518470764,
      "learning_rate": 0.0005727119017489643,
      "loss": 2.3961,
      "step": 31523
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6995534896850586,
      "learning_rate": 0.0005727101971471697,
      "loss": 2.6509,
      "step": 31524
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6506801247596741,
      "learning_rate": 0.0005727084924946733,
      "loss": 2.5051,
      "step": 31525
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.5956746339797974,
      "learning_rate": 0.000572706787791475,
      "loss": 2.2075,
      "step": 31526
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6906282305717468,
      "learning_rate": 0.0005727050830375756,
      "loss": 2.4319,
      "step": 31527
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6786606311798096,
      "learning_rate": 0.000572703378232975,
      "loss": 2.2286,
      "step": 31528
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.8301591277122498,
      "learning_rate": 0.0005727016733776738,
      "loss": 2.2223,
      "step": 31529
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6634534001350403,
      "learning_rate": 0.0005726999684716722,
      "loss": 2.2704,
      "step": 31530
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6256877183914185,
      "learning_rate": 0.0005726982635149705,
      "loss": 2.153,
      "step": 31531
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.7622934579849243,
      "learning_rate": 0.0005726965585075691,
      "loss": 2.1259,
      "step": 31532
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6598362326622009,
      "learning_rate": 0.0005726948534494683,
      "loss": 2.3549,
      "step": 31533
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6192879676818848,
      "learning_rate": 0.0005726931483406683,
      "loss": 2.0936,
      "step": 31534
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6916964054107666,
      "learning_rate": 0.0005726914431811695,
      "loss": 2.2654,
      "step": 31535
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.7421740889549255,
      "learning_rate": 0.0005726897379709721,
      "loss": 2.5455,
      "step": 31536
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.707809329032898,
      "learning_rate": 0.0005726880327100767,
      "loss": 2.211,
      "step": 31537
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6464436054229736,
      "learning_rate": 0.0005726863273984834,
      "loss": 2.2514,
      "step": 31538
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.7334502339363098,
      "learning_rate": 0.0005726846220361926,
      "loss": 2.1913,
      "step": 31539
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6706071496009827,
      "learning_rate": 0.0005726829166232046,
      "loss": 2.4134,
      "step": 31540
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6892953515052795,
      "learning_rate": 0.0005726812111595196,
      "loss": 2.4379,
      "step": 31541
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.7611459493637085,
      "learning_rate": 0.000572679505645138,
      "loss": 2.2829,
      "step": 31542
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6401415467262268,
      "learning_rate": 0.0005726778000800602,
      "loss": 2.221,
      "step": 31543
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.7445157766342163,
      "learning_rate": 0.0005726760944642864,
      "loss": 2.3154,
      "step": 31544
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.8004333972930908,
      "learning_rate": 0.000572674388797817,
      "loss": 2.3113,
      "step": 31545
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.633996844291687,
      "learning_rate": 0.0005726726830806522,
      "loss": 2.2098,
      "step": 31546
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.7522575855255127,
      "learning_rate": 0.0005726709773127927,
      "loss": 2.3454,
      "step": 31547
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.7634231448173523,
      "learning_rate": 0.0005726692714942383,
      "loss": 2.2599,
      "step": 31548
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6943742632865906,
      "learning_rate": 0.0005726675656249895,
      "loss": 2.2618,
      "step": 31549
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6916515827178955,
      "learning_rate": 0.0005726658597050467,
      "loss": 2.1891,
      "step": 31550
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.7875994443893433,
      "learning_rate": 0.0005726641537344101,
      "loss": 2.3309,
      "step": 31551
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.7692628502845764,
      "learning_rate": 0.0005726624477130802,
      "loss": 2.3474,
      "step": 31552
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6744043231010437,
      "learning_rate": 0.0005726607416410572,
      "loss": 2.1681,
      "step": 31553
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.7401121854782104,
      "learning_rate": 0.0005726590355183414,
      "loss": 2.2511,
      "step": 31554
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.9215782880783081,
      "learning_rate": 0.000572657329344933,
      "loss": 2.0795,
      "step": 31555
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6551084518432617,
      "learning_rate": 0.0005726556231208326,
      "loss": 2.3394,
      "step": 31556
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.7170730829238892,
      "learning_rate": 0.0005726539168460404,
      "loss": 2.3334,
      "step": 31557
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.7451813220977783,
      "learning_rate": 0.0005726522105205566,
      "loss": 2.2215,
      "step": 31558
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.7073327302932739,
      "learning_rate": 0.0005726505041443817,
      "loss": 2.3384,
      "step": 31559
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6709718704223633,
      "learning_rate": 0.0005726487977175158,
      "loss": 2.2072,
      "step": 31560
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.7981426119804382,
      "learning_rate": 0.0005726470912399594,
      "loss": 2.2265,
      "step": 31561
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6582807302474976,
      "learning_rate": 0.0005726453847117128,
      "loss": 2.1302,
      "step": 31562
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6380605101585388,
      "learning_rate": 0.0005726436781327762,
      "loss": 2.061,
      "step": 31563
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.7228025197982788,
      "learning_rate": 0.0005726419715031501,
      "loss": 2.0673,
      "step": 31564
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6603953242301941,
      "learning_rate": 0.0005726402648228346,
      "loss": 2.4905,
      "step": 31565
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6388161182403564,
      "learning_rate": 0.0005726385580918301,
      "loss": 2.455,
      "step": 31566
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6613481640815735,
      "learning_rate": 0.000572636851310137,
      "loss": 2.2884,
      "step": 31567
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6600552797317505,
      "learning_rate": 0.0005726351444777557,
      "loss": 2.4795,
      "step": 31568
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.7088596820831299,
      "learning_rate": 0.0005726334375946862,
      "loss": 2.3517,
      "step": 31569
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6452282667160034,
      "learning_rate": 0.0005726317306609291,
      "loss": 2.2849,
      "step": 31570
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6475120782852173,
      "learning_rate": 0.0005726300236764846,
      "loss": 2.4995,
      "step": 31571
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.7073305249214172,
      "learning_rate": 0.0005726283166413529,
      "loss": 2.2952,
      "step": 31572
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.7563067674636841,
      "learning_rate": 0.0005726266095555346,
      "loss": 2.4301,
      "step": 31573
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6646019220352173,
      "learning_rate": 0.0005726249024190299,
      "loss": 2.3039,
      "step": 31574
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6296245455741882,
      "learning_rate": 0.000572623195231839,
      "loss": 2.1429,
      "step": 31575
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6158019304275513,
      "learning_rate": 0.0005726214879939623,
      "loss": 2.3193,
      "step": 31576
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6210277676582336,
      "learning_rate": 0.0005726197807054002,
      "loss": 2.2396,
      "step": 31577
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6678676605224609,
      "learning_rate": 0.0005726180733661528,
      "loss": 2.1644,
      "step": 31578
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.7378937602043152,
      "learning_rate": 0.0005726163659762206,
      "loss": 2.3364,
      "step": 31579
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6720608472824097,
      "learning_rate": 0.0005726146585356039,
      "loss": 2.0851,
      "step": 31580
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6410098671913147,
      "learning_rate": 0.000572612951044303,
      "loss": 2.3448,
      "step": 31581
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6626404523849487,
      "learning_rate": 0.0005726112435023182,
      "loss": 2.2396,
      "step": 31582
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6054611206054688,
      "learning_rate": 0.0005726095359096498,
      "loss": 2.2355,
      "step": 31583
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.7600394487380981,
      "learning_rate": 0.0005726078282662981,
      "loss": 2.1322,
      "step": 31584
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.7188370823860168,
      "learning_rate": 0.0005726061205722636,
      "loss": 2.3185,
      "step": 31585
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.7080008387565613,
      "learning_rate": 0.0005726044128275463,
      "loss": 2.3977,
      "step": 31586
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6534675359725952,
      "learning_rate": 0.0005726027050321469,
      "loss": 2.3791,
      "step": 31587
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6309990286827087,
      "learning_rate": 0.0005726009971860653,
      "loss": 2.1997,
      "step": 31588
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.635205090045929,
      "learning_rate": 0.0005725992892893022,
      "loss": 2.4099,
      "step": 31589
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6698161363601685,
      "learning_rate": 0.0005725975813418577,
      "loss": 2.5313,
      "step": 31590
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6378223299980164,
      "learning_rate": 0.0005725958733437321,
      "loss": 2.3961,
      "step": 31591
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6130802631378174,
      "learning_rate": 0.0005725941652949258,
      "loss": 2.5315,
      "step": 31592
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6460835337638855,
      "learning_rate": 0.0005725924571954391,
      "loss": 2.3214,
      "step": 31593
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.621218204498291,
      "learning_rate": 0.0005725907490452723,
      "loss": 2.4351,
      "step": 31594
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6642074584960938,
      "learning_rate": 0.0005725890408444258,
      "loss": 2.3271,
      "step": 31595
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.7611773014068604,
      "learning_rate": 0.0005725873325928999,
      "loss": 2.1322,
      "step": 31596
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.745069682598114,
      "learning_rate": 0.0005725856242906948,
      "loss": 2.5567,
      "step": 31597
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6189777851104736,
      "learning_rate": 0.0005725839159378108,
      "loss": 2.059,
      "step": 31598
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6520496606826782,
      "learning_rate": 0.0005725822075342483,
      "loss": 2.1561,
      "step": 31599
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.9402774572372437,
      "learning_rate": 0.0005725804990800077,
      "loss": 2.5193,
      "step": 31600
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6872125864028931,
      "learning_rate": 0.0005725787905750894,
      "loss": 2.5014,
      "step": 31601
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.670036256313324,
      "learning_rate": 0.0005725770820194933,
      "loss": 2.4084,
      "step": 31602
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.7754738926887512,
      "learning_rate": 0.0005725753734132201,
      "loss": 2.3241,
      "step": 31603
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6817620992660522,
      "learning_rate": 0.00057257366475627,
      "loss": 2.1287,
      "step": 31604
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.7453113198280334,
      "learning_rate": 0.0005725719560486433,
      "loss": 2.2689,
      "step": 31605
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.7046607136726379,
      "learning_rate": 0.0005725702472903402,
      "loss": 2.4168,
      "step": 31606
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6221550107002258,
      "learning_rate": 0.0005725685384813614,
      "loss": 2.1736,
      "step": 31607
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6421116590499878,
      "learning_rate": 0.0005725668296217068,
      "loss": 2.2968,
      "step": 31608
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6481674313545227,
      "learning_rate": 0.0005725651207113768,
      "loss": 2.3904,
      "step": 31609
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6407135725021362,
      "learning_rate": 0.000572563411750372,
      "loss": 2.326,
      "step": 31610
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.7198361158370972,
      "learning_rate": 0.0005725617027386924,
      "loss": 2.3154,
      "step": 31611
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6811168193817139,
      "learning_rate": 0.0005725599936763384,
      "loss": 2.4934,
      "step": 31612
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6145004034042358,
      "learning_rate": 0.0005725582845633104,
      "loss": 2.344,
      "step": 31613
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6406784057617188,
      "learning_rate": 0.0005725565753996086,
      "loss": 2.1075,
      "step": 31614
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6795601844787598,
      "learning_rate": 0.0005725548661852335,
      "loss": 2.2032,
      "step": 31615
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6161883473396301,
      "learning_rate": 0.0005725531569201852,
      "loss": 2.3603,
      "step": 31616
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6316697597503662,
      "learning_rate": 0.0005725514476044643,
      "loss": 2.2885,
      "step": 31617
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.7416883707046509,
      "learning_rate": 0.0005725497382380708,
      "loss": 2.2972,
      "step": 31618
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6583704948425293,
      "learning_rate": 0.0005725480288210051,
      "loss": 2.4434,
      "step": 31619
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.7043258547782898,
      "learning_rate": 0.0005725463193532677,
      "loss": 2.4041,
      "step": 31620
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6968516707420349,
      "learning_rate": 0.0005725446098348589,
      "loss": 2.4293,
      "step": 31621
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6943168044090271,
      "learning_rate": 0.0005725429002657787,
      "loss": 2.2059,
      "step": 31622
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.7187613844871521,
      "learning_rate": 0.0005725411906460277,
      "loss": 2.1279,
      "step": 31623
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6590277552604675,
      "learning_rate": 0.0005725394809756061,
      "loss": 2.4989,
      "step": 31624
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.8579692244529724,
      "learning_rate": 0.0005725377712545143,
      "loss": 2.3969,
      "step": 31625
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6427634954452515,
      "learning_rate": 0.0005725360614827526,
      "loss": 2.3578,
      "step": 31626
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.7399582862854004,
      "learning_rate": 0.0005725343516603213,
      "loss": 2.0422,
      "step": 31627
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.7196022272109985,
      "learning_rate": 0.0005725326417872206,
      "loss": 2.3347,
      "step": 31628
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6920958757400513,
      "learning_rate": 0.0005725309318634512,
      "loss": 2.5467,
      "step": 31629
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.7239266633987427,
      "learning_rate": 0.000572529221889013,
      "loss": 2.2205,
      "step": 31630
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.633636474609375,
      "learning_rate": 0.0005725275118639064,
      "loss": 2.2223,
      "step": 31631
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6179029941558838,
      "learning_rate": 0.0005725258017881319,
      "loss": 2.1953,
      "step": 31632
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.646281898021698,
      "learning_rate": 0.0005725240916616898,
      "loss": 2.1732,
      "step": 31633
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6349753737449646,
      "learning_rate": 0.0005725223814845801,
      "loss": 2.234,
      "step": 31634
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6054032444953918,
      "learning_rate": 0.0005725206712568035,
      "loss": 2.2743,
      "step": 31635
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6494733095169067,
      "learning_rate": 0.0005725189609783601,
      "loss": 2.4326,
      "step": 31636
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6276440024375916,
      "learning_rate": 0.0005725172506492503,
      "loss": 2.2531,
      "step": 31637
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6659000515937805,
      "learning_rate": 0.0005725155402694743,
      "loss": 2.4749,
      "step": 31638
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6517196893692017,
      "learning_rate": 0.0005725138298390326,
      "loss": 2.3719,
      "step": 31639
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.606742799282074,
      "learning_rate": 0.0005725121193579255,
      "loss": 2.3135,
      "step": 31640
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6446213126182556,
      "learning_rate": 0.0005725104088261533,
      "loss": 2.5026,
      "step": 31641
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6129499673843384,
      "learning_rate": 0.0005725086982437161,
      "loss": 2.1681,
      "step": 31642
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.702800452709198,
      "learning_rate": 0.0005725069876106145,
      "loss": 2.317,
      "step": 31643
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6601566076278687,
      "learning_rate": 0.0005725052769268487,
      "loss": 2.3028,
      "step": 31644
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6949124932289124,
      "learning_rate": 0.000572503566192419,
      "loss": 2.3529,
      "step": 31645
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.5698099732398987,
      "learning_rate": 0.0005725018554073257,
      "loss": 2.128,
      "step": 31646
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6797316074371338,
      "learning_rate": 0.0005725001445715692,
      "loss": 2.2367,
      "step": 31647
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6219270825386047,
      "learning_rate": 0.0005724984336851498,
      "loss": 2.2257,
      "step": 31648
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6233777403831482,
      "learning_rate": 0.0005724967227480678,
      "loss": 2.2062,
      "step": 31649
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6530818939208984,
      "learning_rate": 0.0005724950117603235,
      "loss": 2.5127,
      "step": 31650
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6228453516960144,
      "learning_rate": 0.0005724933007219172,
      "loss": 2.272,
      "step": 31651
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6702476143836975,
      "learning_rate": 0.0005724915896328493,
      "loss": 2.2937,
      "step": 31652
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.7132824659347534,
      "learning_rate": 0.0005724898784931201,
      "loss": 2.423,
      "step": 31653
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6550750136375427,
      "learning_rate": 0.0005724881673027297,
      "loss": 2.2261,
      "step": 31654
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6588416695594788,
      "learning_rate": 0.0005724864560616787,
      "loss": 2.3111,
      "step": 31655
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6599531769752502,
      "learning_rate": 0.0005724847447699675,
      "loss": 2.4093,
      "step": 31656
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6689267158508301,
      "learning_rate": 0.0005724830334275961,
      "loss": 2.2301,
      "step": 31657
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6998701095581055,
      "learning_rate": 0.0005724813220345649,
      "loss": 2.2135,
      "step": 31658
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6567801833152771,
      "learning_rate": 0.0005724796105908744,
      "loss": 2.3592,
      "step": 31659
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.7357443571090698,
      "learning_rate": 0.0005724778990965247,
      "loss": 2.187,
      "step": 31660
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.7183245420455933,
      "learning_rate": 0.0005724761875515163,
      "loss": 2.489,
      "step": 31661
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6777524948120117,
      "learning_rate": 0.0005724744759558494,
      "loss": 2.1151,
      "step": 31662
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6277376413345337,
      "learning_rate": 0.0005724727643095243,
      "loss": 2.2209,
      "step": 31663
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6931312680244446,
      "learning_rate": 0.0005724710526125415,
      "loss": 2.2139,
      "step": 31664
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.7434828877449036,
      "learning_rate": 0.0005724693408649011,
      "loss": 2.1823,
      "step": 31665
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.7079793214797974,
      "learning_rate": 0.0005724676290666034,
      "loss": 2.189,
      "step": 31666
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6831129193305969,
      "learning_rate": 0.000572465917217649,
      "loss": 2.2111,
      "step": 31667
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6718563437461853,
      "learning_rate": 0.000572464205318038,
      "loss": 2.3969,
      "step": 31668
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6464375853538513,
      "learning_rate": 0.0005724624933677707,
      "loss": 2.0915,
      "step": 31669
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.7208226323127747,
      "learning_rate": 0.0005724607813668475,
      "loss": 2.1373,
      "step": 31670
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.7982525825500488,
      "learning_rate": 0.0005724590693152688,
      "loss": 2.3643,
      "step": 31671
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6676533222198486,
      "learning_rate": 0.0005724573572130346,
      "loss": 2.0952,
      "step": 31672
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6059037446975708,
      "learning_rate": 0.0005724556450601456,
      "loss": 2.4404,
      "step": 31673
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6956950426101685,
      "learning_rate": 0.000572453932856602,
      "loss": 2.3435,
      "step": 31674
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.7177020311355591,
      "learning_rate": 0.000572452220602404,
      "loss": 2.3673,
      "step": 31675
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6947600245475769,
      "learning_rate": 0.0005724505082975519,
      "loss": 2.3308,
      "step": 31676
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6418324708938599,
      "learning_rate": 0.0005724487959420462,
      "loss": 2.187,
      "step": 31677
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.7314725518226624,
      "learning_rate": 0.0005724470835358871,
      "loss": 2.5951,
      "step": 31678
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.7645397782325745,
      "learning_rate": 0.000572445371079075,
      "loss": 2.4013,
      "step": 31679
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6873208284378052,
      "learning_rate": 0.0005724436585716101,
      "loss": 2.408,
      "step": 31680
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6562317609786987,
      "learning_rate": 0.0005724419460134928,
      "loss": 2.3771,
      "step": 31681
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6966412663459778,
      "learning_rate": 0.0005724402334047234,
      "loss": 2.3128,
      "step": 31682
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6721004843711853,
      "learning_rate": 0.0005724385207453022,
      "loss": 2.3758,
      "step": 31683
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6210860013961792,
      "learning_rate": 0.0005724368080352296,
      "loss": 2.3534,
      "step": 31684
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6449518799781799,
      "learning_rate": 0.0005724350952745058,
      "loss": 2.2265,
      "step": 31685
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6228625774383545,
      "learning_rate": 0.0005724333824631312,
      "loss": 2.2988,
      "step": 31686
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6668691635131836,
      "learning_rate": 0.000572431669601106,
      "loss": 2.1423,
      "step": 31687
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6511327624320984,
      "learning_rate": 0.0005724299566884307,
      "loss": 2.3822,
      "step": 31688
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6260639429092407,
      "learning_rate": 0.0005724282437251056,
      "loss": 2.3592,
      "step": 31689
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6687813997268677,
      "learning_rate": 0.0005724265307111308,
      "loss": 2.3338,
      "step": 31690
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.7033154964447021,
      "learning_rate": 0.0005724248176465069,
      "loss": 2.3024,
      "step": 31691
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.7289125919342041,
      "learning_rate": 0.0005724231045312339,
      "loss": 2.2223,
      "step": 31692
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6255946755409241,
      "learning_rate": 0.0005724213913653124,
      "loss": 2.2063,
      "step": 31693
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6923772692680359,
      "learning_rate": 0.0005724196781487427,
      "loss": 2.4111,
      "step": 31694
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6281554698944092,
      "learning_rate": 0.0005724179648815249,
      "loss": 2.1821,
      "step": 31695
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6690658330917358,
      "learning_rate": 0.0005724162515636595,
      "loss": 2.2641,
      "step": 31696
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.658085823059082,
      "learning_rate": 0.0005724145381951469,
      "loss": 2.3138,
      "step": 31697
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6132538318634033,
      "learning_rate": 0.0005724128247759871,
      "loss": 2.1121,
      "step": 31698
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6163793802261353,
      "learning_rate": 0.0005724111113061808,
      "loss": 2.1582,
      "step": 31699
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6192386150360107,
      "learning_rate": 0.000572409397785728,
      "loss": 2.2,
      "step": 31700
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6795157194137573,
      "learning_rate": 0.0005724076842146292,
      "loss": 2.2988,
      "step": 31701
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6624670028686523,
      "learning_rate": 0.0005724059705928847,
      "loss": 2.3475,
      "step": 31702
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.7788852453231812,
      "learning_rate": 0.0005724042569204947,
      "loss": 2.3723,
      "step": 31703
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6799054741859436,
      "learning_rate": 0.0005724025431974597,
      "loss": 2.2917,
      "step": 31704
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6426421999931335,
      "learning_rate": 0.0005724008294237798,
      "loss": 2.1433,
      "step": 31705
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6584509015083313,
      "learning_rate": 0.0005723991155994556,
      "loss": 2.3138,
      "step": 31706
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6667274832725525,
      "learning_rate": 0.0005723974017244872,
      "loss": 2.4537,
      "step": 31707
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6559573411941528,
      "learning_rate": 0.000572395687798875,
      "loss": 2.1755,
      "step": 31708
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6623638868331909,
      "learning_rate": 0.0005723939738226193,
      "loss": 2.2146,
      "step": 31709
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.7134301066398621,
      "learning_rate": 0.0005723922597957203,
      "loss": 2.3357,
      "step": 31710
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.5996288657188416,
      "learning_rate": 0.0005723905457181785,
      "loss": 2.177,
      "step": 31711
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6247649192810059,
      "learning_rate": 0.0005723888315899942,
      "loss": 2.4014,
      "step": 31712
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6012862324714661,
      "learning_rate": 0.0005723871174111676,
      "loss": 2.1374,
      "step": 31713
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.7512204051017761,
      "learning_rate": 0.0005723854031816993,
      "loss": 2.132,
      "step": 31714
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6917539834976196,
      "learning_rate": 0.0005723836889015892,
      "loss": 2.1962,
      "step": 31715
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6144638061523438,
      "learning_rate": 0.0005723819745708379,
      "loss": 2.4188,
      "step": 31716
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6598542332649231,
      "learning_rate": 0.0005723802601894457,
      "loss": 2.418,
      "step": 31717
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6496521234512329,
      "learning_rate": 0.0005723785457574128,
      "loss": 2.3214,
      "step": 31718
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6810579299926758,
      "learning_rate": 0.0005723768312747396,
      "loss": 2.3694,
      "step": 31719
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6056556701660156,
      "learning_rate": 0.0005723751167414264,
      "loss": 2.1135,
      "step": 31720
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.5926324129104614,
      "learning_rate": 0.0005723734021574735,
      "loss": 2.2244,
      "step": 31721
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.650213360786438,
      "learning_rate": 0.0005723716875228813,
      "loss": 2.3779,
      "step": 31722
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6054427623748779,
      "learning_rate": 0.00057236997283765,
      "loss": 2.2061,
      "step": 31723
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6407096982002258,
      "learning_rate": 0.00057236825810178,
      "loss": 2.1968,
      "step": 31724
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6966454386711121,
      "learning_rate": 0.0005723665433152717,
      "loss": 2.3166,
      "step": 31725
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6555847525596619,
      "learning_rate": 0.0005723648284781252,
      "loss": 2.0311,
      "step": 31726
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6187293529510498,
      "learning_rate": 0.0005723631135903409,
      "loss": 2.0821,
      "step": 31727
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6778780817985535,
      "learning_rate": 0.0005723613986519193,
      "loss": 2.3467,
      "step": 31728
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.8098867535591125,
      "learning_rate": 0.0005723596836628605,
      "loss": 2.2921,
      "step": 31729
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.7005521655082703,
      "learning_rate": 0.0005723579686231649,
      "loss": 2.3643,
      "step": 31730
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6928200721740723,
      "learning_rate": 0.0005723562535328327,
      "loss": 2.3504,
      "step": 31731
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6538974046707153,
      "learning_rate": 0.0005723545383918645,
      "loss": 2.1082,
      "step": 31732
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6820217967033386,
      "learning_rate": 0.0005723528232002604,
      "loss": 2.3034,
      "step": 31733
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6043439507484436,
      "learning_rate": 0.0005723511079580208,
      "loss": 2.3123,
      "step": 31734
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6787883639335632,
      "learning_rate": 0.0005723493926651458,
      "loss": 2.0585,
      "step": 31735
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6368515491485596,
      "learning_rate": 0.0005723476773216361,
      "loss": 2.2581,
      "step": 31736
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6068649888038635,
      "learning_rate": 0.0005723459619274918,
      "loss": 2.4125,
      "step": 31737
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6743599772453308,
      "learning_rate": 0.0005723442464827133,
      "loss": 2.2632,
      "step": 31738
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.8933371305465698,
      "learning_rate": 0.0005723425309873008,
      "loss": 2.3572,
      "step": 31739
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6611570715904236,
      "learning_rate": 0.0005723408154412546,
      "loss": 2.0766,
      "step": 31740
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.7925621271133423,
      "learning_rate": 0.0005723390998445751,
      "loss": 2.2458,
      "step": 31741
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.7148882150650024,
      "learning_rate": 0.0005723373841972627,
      "loss": 2.4446,
      "step": 31742
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.656773567199707,
      "learning_rate": 0.0005723356684993176,
      "loss": 2.2087,
      "step": 31743
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.7726501822471619,
      "learning_rate": 0.0005723339527507403,
      "loss": 2.3604,
      "step": 31744
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.740920901298523,
      "learning_rate": 0.0005723322369515308,
      "loss": 2.4851,
      "step": 31745
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.641118586063385,
      "learning_rate": 0.0005723305211016897,
      "loss": 2.1674,
      "step": 31746
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.7912932634353638,
      "learning_rate": 0.0005723288052012171,
      "loss": 2.3735,
      "step": 31747
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.7325773239135742,
      "learning_rate": 0.0005723270892501136,
      "loss": 2.3057,
      "step": 31748
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6118312478065491,
      "learning_rate": 0.0005723253732483792,
      "loss": 2.4284,
      "step": 31749
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.7378655076026917,
      "learning_rate": 0.0005723236571960145,
      "loss": 2.2872,
      "step": 31750
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.7329961061477661,
      "learning_rate": 0.0005723219410930196,
      "loss": 2.471,
      "step": 31751
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6176314353942871,
      "learning_rate": 0.000572320224939395,
      "loss": 2.3992,
      "step": 31752
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6531740427017212,
      "learning_rate": 0.0005723185087351409,
      "loss": 2.201,
      "step": 31753
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6146471500396729,
      "learning_rate": 0.0005723167924802575,
      "loss": 2.5175,
      "step": 31754
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.7347555160522461,
      "learning_rate": 0.0005723150761747454,
      "loss": 2.2794,
      "step": 31755
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.660775899887085,
      "learning_rate": 0.0005723133598186048,
      "loss": 2.2813,
      "step": 31756
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.7496140599250793,
      "learning_rate": 0.000572311643411836,
      "loss": 2.2243,
      "step": 31757
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.7188494801521301,
      "learning_rate": 0.0005723099269544393,
      "loss": 2.4593,
      "step": 31758
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6545806527137756,
      "learning_rate": 0.0005723082104464151,
      "loss": 2.318,
      "step": 31759
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6802395582199097,
      "learning_rate": 0.0005723064938877636,
      "loss": 2.1716,
      "step": 31760
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6987963914871216,
      "learning_rate": 0.0005723047772784851,
      "loss": 2.458,
      "step": 31761
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.7828826904296875,
      "learning_rate": 0.0005723030606185802,
      "loss": 2.0588,
      "step": 31762
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.7258837223052979,
      "learning_rate": 0.000572301343908049,
      "loss": 2.3266,
      "step": 31763
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6854941248893738,
      "learning_rate": 0.0005722996271468917,
      "loss": 2.3544,
      "step": 31764
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6929807066917419,
      "learning_rate": 0.0005722979103351088,
      "loss": 2.4492,
      "step": 31765
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.7599812150001526,
      "learning_rate": 0.0005722961934727006,
      "loss": 2.2342,
      "step": 31766
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6594407558441162,
      "learning_rate": 0.0005722944765596674,
      "loss": 2.33,
      "step": 31767
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.70716792345047,
      "learning_rate": 0.0005722927595960094,
      "loss": 2.3473,
      "step": 31768
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.7926201224327087,
      "learning_rate": 0.0005722910425817273,
      "loss": 2.2622,
      "step": 31769
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6245784759521484,
      "learning_rate": 0.000572289325516821,
      "loss": 2.296,
      "step": 31770
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6282920837402344,
      "learning_rate": 0.0005722876084012909,
      "loss": 2.4454,
      "step": 31771
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.7047657370567322,
      "learning_rate": 0.0005722858912351374,
      "loss": 2.5213,
      "step": 31772
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6921951174736023,
      "learning_rate": 0.000572284174018361,
      "loss": 2.1801,
      "step": 31773
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6235188841819763,
      "learning_rate": 0.0005722824567509616,
      "loss": 2.493,
      "step": 31774
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6360793113708496,
      "learning_rate": 0.00057228073943294,
      "loss": 1.9318,
      "step": 31775
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.9941116571426392,
      "learning_rate": 0.000572279022064296,
      "loss": 2.1979,
      "step": 31776
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6306089162826538,
      "learning_rate": 0.0005722773046450305,
      "loss": 2.2743,
      "step": 31777
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.7361078262329102,
      "learning_rate": 0.0005722755871751432,
      "loss": 2.2982,
      "step": 31778
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.7056979537010193,
      "learning_rate": 0.0005722738696546349,
      "loss": 2.2244,
      "step": 31779
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6745496988296509,
      "learning_rate": 0.0005722721520835057,
      "loss": 2.2584,
      "step": 31780
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6803721785545349,
      "learning_rate": 0.000572270434461756,
      "loss": 2.4635,
      "step": 31781
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.7360202670097351,
      "learning_rate": 0.0005722687167893861,
      "loss": 2.5287,
      "step": 31782
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.7385777831077576,
      "learning_rate": 0.0005722669990663962,
      "loss": 2.2981,
      "step": 31783
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6203628182411194,
      "learning_rate": 0.0005722652812927869,
      "loss": 2.2158,
      "step": 31784
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6614797115325928,
      "learning_rate": 0.0005722635634685582,
      "loss": 2.4463,
      "step": 31785
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.8005649447441101,
      "learning_rate": 0.0005722618455937106,
      "loss": 2.3922,
      "step": 31786
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.5877828001976013,
      "learning_rate": 0.0005722601276682445,
      "loss": 2.1766,
      "step": 31787
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6481817960739136,
      "learning_rate": 0.0005722584096921598,
      "loss": 2.3926,
      "step": 31788
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.8890119791030884,
      "learning_rate": 0.0005722566916654574,
      "loss": 2.3803,
      "step": 31789
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.7164904475212097,
      "learning_rate": 0.0005722549735881373,
      "loss": 2.2023,
      "step": 31790
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6698004603385925,
      "learning_rate": 0.0005722532554601999,
      "loss": 2.295,
      "step": 31791
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.7257091403007507,
      "learning_rate": 0.0005722515372816454,
      "loss": 2.1111,
      "step": 31792
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6992787718772888,
      "learning_rate": 0.0005722498190524741,
      "loss": 2.271,
      "step": 31793
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6865267157554626,
      "learning_rate": 0.0005722481007726866,
      "loss": 2.4452,
      "step": 31794
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6583786010742188,
      "learning_rate": 0.0005722463824422829,
      "loss": 2.5028,
      "step": 31795
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.7223701477050781,
      "learning_rate": 0.0005722446640612635,
      "loss": 2.3706,
      "step": 31796
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.7104156613349915,
      "learning_rate": 0.0005722429456296287,
      "loss": 2.2648,
      "step": 31797
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6558613181114197,
      "learning_rate": 0.0005722412271473788,
      "loss": 2.4967,
      "step": 31798
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6426772475242615,
      "learning_rate": 0.0005722395086145141,
      "loss": 2.3576,
      "step": 31799
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6810122728347778,
      "learning_rate": 0.0005722377900310349,
      "loss": 2.446,
      "step": 31800
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.9511269330978394,
      "learning_rate": 0.0005722360713969417,
      "loss": 2.2722,
      "step": 31801
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6523920893669128,
      "learning_rate": 0.0005722343527122346,
      "loss": 2.444,
      "step": 31802
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.600094199180603,
      "learning_rate": 0.0005722326339769139,
      "loss": 2.4287,
      "step": 31803
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6443785429000854,
      "learning_rate": 0.00057223091519098,
      "loss": 2.2525,
      "step": 31804
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6494942307472229,
      "learning_rate": 0.0005722291963544334,
      "loss": 2.4368,
      "step": 31805
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.682962954044342,
      "learning_rate": 0.0005722274774672743,
      "loss": 2.5077,
      "step": 31806
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6177726984024048,
      "learning_rate": 0.0005722257585295028,
      "loss": 2.5325,
      "step": 31807
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6184218525886536,
      "learning_rate": 0.0005722240395411194,
      "loss": 2.207,
      "step": 31808
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6370054483413696,
      "learning_rate": 0.0005722223205021245,
      "loss": 2.3668,
      "step": 31809
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.7409444451332092,
      "learning_rate": 0.0005722206014125182,
      "loss": 2.3289,
      "step": 31810
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6057871580123901,
      "learning_rate": 0.0005722188822723011,
      "loss": 2.2049,
      "step": 31811
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6148126721382141,
      "learning_rate": 0.0005722171630814733,
      "loss": 2.1903,
      "step": 31812
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6469029784202576,
      "learning_rate": 0.0005722154438400352,
      "loss": 2.2763,
      "step": 31813
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.639525830745697,
      "learning_rate": 0.0005722137245479871,
      "loss": 2.3361,
      "step": 31814
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.7220330238342285,
      "learning_rate": 0.0005722120052053293,
      "loss": 2.385,
      "step": 31815
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.7032841444015503,
      "learning_rate": 0.0005722102858120622,
      "loss": 2.3863,
      "step": 31816
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6214620471000671,
      "learning_rate": 0.0005722085663681861,
      "loss": 2.1228,
      "step": 31817
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.7746533751487732,
      "learning_rate": 0.0005722068468737013,
      "loss": 2.2418,
      "step": 31818
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6796015501022339,
      "learning_rate": 0.000572205127328608,
      "loss": 2.196,
      "step": 31819
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.7512415051460266,
      "learning_rate": 0.0005722034077329067,
      "loss": 2.2614,
      "step": 31820
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6549338698387146,
      "learning_rate": 0.0005722016880865976,
      "loss": 2.202,
      "step": 31821
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.8907344937324524,
      "learning_rate": 0.0005721999683896811,
      "loss": 2.1645,
      "step": 31822
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6900830864906311,
      "learning_rate": 0.0005721982486421575,
      "loss": 2.0961,
      "step": 31823
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.659345805644989,
      "learning_rate": 0.0005721965288440271,
      "loss": 2.3425,
      "step": 31824
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.641835629940033,
      "learning_rate": 0.0005721948089952903,
      "loss": 2.2287,
      "step": 31825
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.7195644378662109,
      "learning_rate": 0.0005721930890959472,
      "loss": 2.2163,
      "step": 31826
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6490104794502258,
      "learning_rate": 0.0005721913691459983,
      "loss": 2.469,
      "step": 31827
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6310521364212036,
      "learning_rate": 0.000572189649145444,
      "loss": 2.4756,
      "step": 31828
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.7106572389602661,
      "learning_rate": 0.0005721879290942844,
      "loss": 2.5787,
      "step": 31829
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6900045871734619,
      "learning_rate": 0.0005721862089925199,
      "loss": 2.4934,
      "step": 31830
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6443061232566833,
      "learning_rate": 0.000572184488840151,
      "loss": 2.4604,
      "step": 31831
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.7300883531570435,
      "learning_rate": 0.0005721827686371777,
      "loss": 2.268,
      "step": 31832
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.7449756860733032,
      "learning_rate": 0.0005721810483836006,
      "loss": 2.5487,
      "step": 31833
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6857239603996277,
      "learning_rate": 0.0005721793280794199,
      "loss": 2.525,
      "step": 31834
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6285954713821411,
      "learning_rate": 0.0005721776077246359,
      "loss": 2.0225,
      "step": 31835
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.7157647609710693,
      "learning_rate": 0.000572175887319249,
      "loss": 2.3612,
      "step": 31836
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6943503618240356,
      "learning_rate": 0.0005721741668632593,
      "loss": 2.3744,
      "step": 31837
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.7576119303703308,
      "learning_rate": 0.0005721724463566674,
      "loss": 2.3002,
      "step": 31838
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6762143969535828,
      "learning_rate": 0.0005721707257994735,
      "loss": 2.3597,
      "step": 31839
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.7004931569099426,
      "learning_rate": 0.0005721690051916779,
      "loss": 2.3112,
      "step": 31840
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.75040203332901,
      "learning_rate": 0.0005721672845332809,
      "loss": 2.3994,
      "step": 31841
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6087857484817505,
      "learning_rate": 0.000572165563824283,
      "loss": 2.2735,
      "step": 31842
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6143274307250977,
      "learning_rate": 0.0005721638430646842,
      "loss": 2.2226,
      "step": 31843
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6980924606323242,
      "learning_rate": 0.0005721621222544852,
      "loss": 2.3986,
      "step": 31844
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6904900074005127,
      "learning_rate": 0.000572160401393686,
      "loss": 2.3776,
      "step": 31845
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6749035716056824,
      "learning_rate": 0.000572158680482287,
      "loss": 2.2601,
      "step": 31846
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.7616857886314392,
      "learning_rate": 0.0005721569595202887,
      "loss": 2.3491,
      "step": 31847
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6872797012329102,
      "learning_rate": 0.0005721552385076912,
      "loss": 2.2258,
      "step": 31848
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6319987177848816,
      "learning_rate": 0.0005721535174444949,
      "loss": 2.2204,
      "step": 31849
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6926339268684387,
      "learning_rate": 0.0005721517963307002,
      "loss": 2.2618,
      "step": 31850
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6891388893127441,
      "learning_rate": 0.0005721500751663073,
      "loss": 2.1206,
      "step": 31851
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6941758394241333,
      "learning_rate": 0.0005721483539513166,
      "loss": 2.272,
      "step": 31852
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6949778199195862,
      "learning_rate": 0.0005721466326857283,
      "loss": 2.4401,
      "step": 31853
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6769109964370728,
      "learning_rate": 0.0005721449113695428,
      "loss": 2.2152,
      "step": 31854
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6621813178062439,
      "learning_rate": 0.0005721431900027604,
      "loss": 2.1163,
      "step": 31855
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6703301668167114,
      "learning_rate": 0.0005721414685853816,
      "loss": 2.1112,
      "step": 31856
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6788991093635559,
      "learning_rate": 0.0005721397471174064,
      "loss": 2.2408,
      "step": 31857
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6744707822799683,
      "learning_rate": 0.0005721380255988354,
      "loss": 2.3331,
      "step": 31858
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6148263812065125,
      "learning_rate": 0.0005721363040296688,
      "loss": 2.1579,
      "step": 31859
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.7017614245414734,
      "learning_rate": 0.0005721345824099068,
      "loss": 2.2791,
      "step": 31860
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6398202776908875,
      "learning_rate": 0.00057213286073955,
      "loss": 2.4525,
      "step": 31861
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.67269366979599,
      "learning_rate": 0.0005721311390185985,
      "loss": 2.3965,
      "step": 31862
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6536967754364014,
      "learning_rate": 0.0005721294172470527,
      "loss": 2.2619,
      "step": 31863
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6327229142189026,
      "learning_rate": 0.0005721276954249128,
      "loss": 2.2551,
      "step": 31864
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6414032578468323,
      "learning_rate": 0.0005721259735521793,
      "loss": 2.1595,
      "step": 31865
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.7209487557411194,
      "learning_rate": 0.0005721242516288525,
      "loss": 2.2539,
      "step": 31866
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.7694794535636902,
      "learning_rate": 0.0005721225296549326,
      "loss": 2.4802,
      "step": 31867
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6640094518661499,
      "learning_rate": 0.0005721208076304201,
      "loss": 2.3306,
      "step": 31868
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.8013903498649597,
      "learning_rate": 0.0005721190855553151,
      "loss": 2.2714,
      "step": 31869
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6104701161384583,
      "learning_rate": 0.000572117363429618,
      "loss": 2.3856,
      "step": 31870
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.7393004894256592,
      "learning_rate": 0.0005721156412533292,
      "loss": 2.3272,
      "step": 31871
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6719484925270081,
      "learning_rate": 0.000572113919026449,
      "loss": 2.3867,
      "step": 31872
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6911748051643372,
      "learning_rate": 0.0005721121967489775,
      "loss": 2.4762,
      "step": 31873
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6884002685546875,
      "learning_rate": 0.0005721104744209153,
      "loss": 2.2903,
      "step": 31874
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6533371806144714,
      "learning_rate": 0.0005721087520422628,
      "loss": 2.3863,
      "step": 31875
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.689517617225647,
      "learning_rate": 0.00057210702961302,
      "loss": 2.4708,
      "step": 31876
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.7120790481567383,
      "learning_rate": 0.0005721053071331874,
      "loss": 2.3823,
      "step": 31877
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.7251201272010803,
      "learning_rate": 0.0005721035846027652,
      "loss": 2.2925,
      "step": 31878
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6542404890060425,
      "learning_rate": 0.000572101862021754,
      "loss": 2.3779,
      "step": 31879
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6918370127677917,
      "learning_rate": 0.0005721001393901537,
      "loss": 2.5762,
      "step": 31880
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6950814127922058,
      "learning_rate": 0.000572098416707965,
      "loss": 2.2409,
      "step": 31881
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6313438415527344,
      "learning_rate": 0.000572096693975188,
      "loss": 2.4147,
      "step": 31882
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.7688749432563782,
      "learning_rate": 0.0005720949711918231,
      "loss": 2.361,
      "step": 31883
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6653174757957458,
      "learning_rate": 0.0005720932483578706,
      "loss": 2.403,
      "step": 31884
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6486937999725342,
      "learning_rate": 0.0005720915254733309,
      "loss": 2.0717,
      "step": 31885
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6689137816429138,
      "learning_rate": 0.0005720898025382042,
      "loss": 2.4423,
      "step": 31886
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6931657791137695,
      "learning_rate": 0.0005720880795524908,
      "loss": 2.31,
      "step": 31887
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.7163695096969604,
      "learning_rate": 0.0005720863565161911,
      "loss": 2.4389,
      "step": 31888
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.7678937911987305,
      "learning_rate": 0.0005720846334293055,
      "loss": 2.4168,
      "step": 31889
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6842554211616516,
      "learning_rate": 0.0005720829102918341,
      "loss": 2.2435,
      "step": 31890
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.7287994623184204,
      "learning_rate": 0.0005720811871037775,
      "loss": 2.4616,
      "step": 31891
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6925568580627441,
      "learning_rate": 0.0005720794638651358,
      "loss": 2.2356,
      "step": 31892
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6868734955787659,
      "learning_rate": 0.0005720777405759095,
      "loss": 2.4628,
      "step": 31893
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6980385184288025,
      "learning_rate": 0.0005720760172360986,
      "loss": 2.3915,
      "step": 31894
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.8785348534584045,
      "learning_rate": 0.0005720742938457038,
      "loss": 2.557,
      "step": 31895
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6419517397880554,
      "learning_rate": 0.0005720725704047253,
      "loss": 2.2206,
      "step": 31896
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6495891809463501,
      "learning_rate": 0.0005720708469131632,
      "loss": 2.3386,
      "step": 31897
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6999849081039429,
      "learning_rate": 0.0005720691233710181,
      "loss": 2.2999,
      "step": 31898
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.7025177478790283,
      "learning_rate": 0.0005720673997782902,
      "loss": 2.2044,
      "step": 31899
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6569740176200867,
      "learning_rate": 0.0005720656761349798,
      "loss": 2.3999,
      "step": 31900
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6497693657875061,
      "learning_rate": 0.0005720639524410872,
      "loss": 2.3958,
      "step": 31901
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.8193406462669373,
      "learning_rate": 0.000572062228696613,
      "loss": 2.2293,
      "step": 31902
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6402329802513123,
      "learning_rate": 0.0005720605049015571,
      "loss": 2.5411,
      "step": 31903
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6684139966964722,
      "learning_rate": 0.00057205878105592,
      "loss": 2.0886,
      "step": 31904
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.7681622505187988,
      "learning_rate": 0.0005720570571597022,
      "loss": 2.6134,
      "step": 31905
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6821379065513611,
      "learning_rate": 0.0005720553332129037,
      "loss": 2.5007,
      "step": 31906
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.7612812519073486,
      "learning_rate": 0.000572053609215525,
      "loss": 2.2636,
      "step": 31907
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.8224797248840332,
      "learning_rate": 0.0005720518851675665,
      "loss": 2.4601,
      "step": 31908
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6095239520072937,
      "learning_rate": 0.0005720501610690283,
      "loss": 2.2035,
      "step": 31909
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.7268157005310059,
      "learning_rate": 0.000572048436919911,
      "loss": 2.2186,
      "step": 31910
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.7169010639190674,
      "learning_rate": 0.0005720467127202145,
      "loss": 2.2828,
      "step": 31911
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6377960443496704,
      "learning_rate": 0.0005720449884699396,
      "loss": 2.1743,
      "step": 31912
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.5963848233222961,
      "learning_rate": 0.0005720432641690863,
      "loss": 2.2384,
      "step": 31913
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6637971997261047,
      "learning_rate": 0.0005720415398176551,
      "loss": 2.2093,
      "step": 31914
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.7014187574386597,
      "learning_rate": 0.0005720398154156461,
      "loss": 2.4824,
      "step": 31915
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6065280437469482,
      "learning_rate": 0.00057203809096306,
      "loss": 2.1859,
      "step": 31916
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6996181011199951,
      "learning_rate": 0.0005720363664598967,
      "loss": 2.2979,
      "step": 31917
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.7077986001968384,
      "learning_rate": 0.0005720346419061567,
      "loss": 2.3595,
      "step": 31918
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.8969176411628723,
      "learning_rate": 0.0005720329173018404,
      "loss": 2.3466,
      "step": 31919
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.7601103782653809,
      "learning_rate": 0.0005720311926469481,
      "loss": 2.2672,
      "step": 31920
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.9843137860298157,
      "learning_rate": 0.00057202946794148,
      "loss": 2.4742,
      "step": 31921
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6259127259254456,
      "learning_rate": 0.0005720277431854364,
      "loss": 2.3033,
      "step": 31922
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.752841055393219,
      "learning_rate": 0.0005720260183788178,
      "loss": 2.2347,
      "step": 31923
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.8648379445075989,
      "learning_rate": 0.0005720242935216245,
      "loss": 2.3392,
      "step": 31924
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6853956580162048,
      "learning_rate": 0.0005720225686138566,
      "loss": 2.3042,
      "step": 31925
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.731049656867981,
      "learning_rate": 0.0005720208436555146,
      "loss": 2.1874,
      "step": 31926
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.8260321617126465,
      "learning_rate": 0.0005720191186465988,
      "loss": 2.0942,
      "step": 31927
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.809149444103241,
      "learning_rate": 0.0005720173935871095,
      "loss": 2.1853,
      "step": 31928
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.8784154057502747,
      "learning_rate": 0.0005720156684770471,
      "loss": 2.2424,
      "step": 31929
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6040236949920654,
      "learning_rate": 0.0005720139433164119,
      "loss": 2.3749,
      "step": 31930
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.5733463168144226,
      "learning_rate": 0.000572012218105204,
      "loss": 2.3616,
      "step": 31931
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6625584363937378,
      "learning_rate": 0.000572010492843424,
      "loss": 2.3358,
      "step": 31932
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6636163592338562,
      "learning_rate": 0.0005720087675310722,
      "loss": 2.2086,
      "step": 31933
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6293941736221313,
      "learning_rate": 0.0005720070421681487,
      "loss": 2.048,
      "step": 31934
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.5940877795219421,
      "learning_rate": 0.000572005316754654,
      "loss": 2.3891,
      "step": 31935
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6970279812812805,
      "learning_rate": 0.0005720035912905884,
      "loss": 2.4374,
      "step": 31936
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.7383607029914856,
      "learning_rate": 0.0005720018657759521,
      "loss": 2.1616,
      "step": 31937
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6746649742126465,
      "learning_rate": 0.0005720001402107455,
      "loss": 2.4766,
      "step": 31938
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.7135897874832153,
      "learning_rate": 0.0005719984145949691,
      "loss": 2.2167,
      "step": 31939
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.7959843873977661,
      "learning_rate": 0.000571996688928623,
      "loss": 2.3175,
      "step": 31940
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6818773150444031,
      "learning_rate": 0.0005719949632117075,
      "loss": 2.2876,
      "step": 31941
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6896228790283203,
      "learning_rate": 0.0005719932374442232,
      "loss": 2.5182,
      "step": 31942
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.7757795453071594,
      "learning_rate": 0.00057199151162617,
      "loss": 2.1785,
      "step": 31943
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.694273829460144,
      "learning_rate": 0.0005719897857575485,
      "loss": 2.2779,
      "step": 31944
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6570324897766113,
      "learning_rate": 0.000571988059838359,
      "loss": 2.2318,
      "step": 31945
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6777750253677368,
      "learning_rate": 0.0005719863338686018,
      "loss": 2.4416,
      "step": 31946
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.7010220289230347,
      "learning_rate": 0.0005719846078482771,
      "loss": 2.2097,
      "step": 31947
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6743559241294861,
      "learning_rate": 0.0005719828817773854,
      "loss": 2.3236,
      "step": 31948
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.5751869082450867,
      "learning_rate": 0.0005719811556559271,
      "loss": 2.0729,
      "step": 31949
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6814852356910706,
      "learning_rate": 0.0005719794294839022,
      "loss": 2.2725,
      "step": 31950
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.7532133460044861,
      "learning_rate": 0.0005719777032613113,
      "loss": 2.4139,
      "step": 31951
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6913440227508545,
      "learning_rate": 0.0005719759769881544,
      "loss": 2.5132,
      "step": 31952
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6301581263542175,
      "learning_rate": 0.0005719742506644322,
      "loss": 2.2618,
      "step": 31953
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.7817589044570923,
      "learning_rate": 0.0005719725242901448,
      "loss": 2.3259,
      "step": 31954
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6558117270469666,
      "learning_rate": 0.0005719707978652927,
      "loss": 2.1849,
      "step": 31955
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.7318716645240784,
      "learning_rate": 0.000571969071389876,
      "loss": 2.3244,
      "step": 31956
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.64723140001297,
      "learning_rate": 0.0005719673448638951,
      "loss": 2.3634,
      "step": 31957
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.7359058856964111,
      "learning_rate": 0.0005719656182873503,
      "loss": 2.2315,
      "step": 31958
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.76241135597229,
      "learning_rate": 0.000571963891660242,
      "loss": 2.174,
      "step": 31959
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.5997025370597839,
      "learning_rate": 0.0005719621649825705,
      "loss": 2.3886,
      "step": 31960
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6881023049354553,
      "learning_rate": 0.0005719604382543361,
      "loss": 2.521,
      "step": 31961
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.7662340998649597,
      "learning_rate": 0.0005719587114755391,
      "loss": 2.4703,
      "step": 31962
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6931371688842773,
      "learning_rate": 0.0005719569846461798,
      "loss": 2.4083,
      "step": 31963
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6335762739181519,
      "learning_rate": 0.0005719552577662586,
      "loss": 2.4091,
      "step": 31964
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.7156837582588196,
      "learning_rate": 0.0005719535308357758,
      "loss": 2.1894,
      "step": 31965
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6427946090698242,
      "learning_rate": 0.0005719518038547317,
      "loss": 2.3054,
      "step": 31966
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.7557548880577087,
      "learning_rate": 0.0005719500768231265,
      "loss": 2.3129,
      "step": 31967
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6235799193382263,
      "learning_rate": 0.0005719483497409609,
      "loss": 2.1793,
      "step": 31968
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.5938585996627808,
      "learning_rate": 0.0005719466226082349,
      "loss": 2.1709,
      "step": 31969
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6506211161613464,
      "learning_rate": 0.0005719448954249487,
      "loss": 2.0227,
      "step": 31970
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.7279802560806274,
      "learning_rate": 0.000571943168191103,
      "loss": 2.1,
      "step": 31971
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.648658037185669,
      "learning_rate": 0.0005719414409066978,
      "loss": 2.3059,
      "step": 31972
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6851908564567566,
      "learning_rate": 0.0005719397135717336,
      "loss": 2.1906,
      "step": 31973
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6282424330711365,
      "learning_rate": 0.0005719379861862108,
      "loss": 2.2439,
      "step": 31974
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6320230960845947,
      "learning_rate": 0.0005719362587501294,
      "loss": 2.397,
      "step": 31975
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6055604815483093,
      "learning_rate": 0.0005719345312634901,
      "loss": 2.3323,
      "step": 31976
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6354066729545593,
      "learning_rate": 0.0005719328037262929,
      "loss": 2.4622,
      "step": 31977
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6180475950241089,
      "learning_rate": 0.0005719310761385383,
      "loss": 2.2469,
      "step": 31978
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.606179416179657,
      "learning_rate": 0.0005719293485002266,
      "loss": 2.458,
      "step": 31979
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6105279326438904,
      "learning_rate": 0.0005719276208113581,
      "loss": 2.34,
      "step": 31980
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.7270410060882568,
      "learning_rate": 0.0005719258930719331,
      "loss": 2.3295,
      "step": 31981
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6273944973945618,
      "learning_rate": 0.0005719241652819519,
      "loss": 2.1091,
      "step": 31982
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.5863141417503357,
      "learning_rate": 0.0005719224374414149,
      "loss": 2.3666,
      "step": 31983
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6783373951911926,
      "learning_rate": 0.0005719207095503224,
      "loss": 2.2732,
      "step": 31984
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.5902495980262756,
      "learning_rate": 0.0005719189816086747,
      "loss": 2.1429,
      "step": 31985
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.7106481194496155,
      "learning_rate": 0.0005719172536164721,
      "loss": 2.3394,
      "step": 31986
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.65208500623703,
      "learning_rate": 0.0005719155255737151,
      "loss": 2.3166,
      "step": 31987
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.581756591796875,
      "learning_rate": 0.0005719137974804037,
      "loss": 2.3434,
      "step": 31988
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6486421227455139,
      "learning_rate": 0.0005719120693365384,
      "loss": 2.4332,
      "step": 31989
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.5948606729507446,
      "learning_rate": 0.0005719103411421197,
      "loss": 2.0975,
      "step": 31990
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.5770957469940186,
      "learning_rate": 0.0005719086128971475,
      "loss": 2.3263,
      "step": 31991
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6293858885765076,
      "learning_rate": 0.0005719068846016225,
      "loss": 2.4502,
      "step": 31992
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.658146321773529,
      "learning_rate": 0.0005719051562555448,
      "loss": 2.3201,
      "step": 31993
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.7813382148742676,
      "learning_rate": 0.0005719034278589149,
      "loss": 2.2536,
      "step": 31994
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6139860153198242,
      "learning_rate": 0.0005719016994117329,
      "loss": 2.2376,
      "step": 31995
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.7054372429847717,
      "learning_rate": 0.0005718999709139993,
      "loss": 2.3581,
      "step": 31996
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6581553220748901,
      "learning_rate": 0.0005718982423657143,
      "loss": 2.2705,
      "step": 31997
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.661493182182312,
      "learning_rate": 0.0005718965137668783,
      "loss": 2.5406,
      "step": 31998
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6505923867225647,
      "learning_rate": 0.0005718947851174917,
      "loss": 2.3824,
      "step": 31999
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6545260548591614,
      "learning_rate": 0.0005718930564175547,
      "loss": 2.4078,
      "step": 32000
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6685318350791931,
      "learning_rate": 0.0005718913276670676,
      "loss": 2.454,
      "step": 32001
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.668411374092102,
      "learning_rate": 0.0005718895988660308,
      "loss": 2.2225,
      "step": 32002
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6272162199020386,
      "learning_rate": 0.0005718878700144446,
      "loss": 2.2023,
      "step": 32003
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6911470890045166,
      "learning_rate": 0.0005718861411123093,
      "loss": 2.4048,
      "step": 32004
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6323380470275879,
      "learning_rate": 0.0005718844121596252,
      "loss": 2.3938,
      "step": 32005
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6091195344924927,
      "learning_rate": 0.0005718826831563927,
      "loss": 2.0829,
      "step": 32006
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6747851371765137,
      "learning_rate": 0.000571880954102612,
      "loss": 2.3013,
      "step": 32007
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6623813509941101,
      "learning_rate": 0.0005718792249982835,
      "loss": 2.2958,
      "step": 32008
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6925426125526428,
      "learning_rate": 0.0005718774958434076,
      "loss": 2.0939,
      "step": 32009
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.7371603846549988,
      "learning_rate": 0.0005718757666379845,
      "loss": 2.4714,
      "step": 32010
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.621153712272644,
      "learning_rate": 0.0005718740373820146,
      "loss": 2.1775,
      "step": 32011
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6074244379997253,
      "learning_rate": 0.0005718723080754982,
      "loss": 2.3237,
      "step": 32012
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.643517017364502,
      "learning_rate": 0.0005718705787184356,
      "loss": 2.2978,
      "step": 32013
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.727768063545227,
      "learning_rate": 0.0005718688493108271,
      "loss": 2.1821,
      "step": 32014
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6774075627326965,
      "learning_rate": 0.000571867119852673,
      "loss": 2.3043,
      "step": 32015
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.7197876572608948,
      "learning_rate": 0.0005718653903439738,
      "loss": 2.4074,
      "step": 32016
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.7185415029525757,
      "learning_rate": 0.0005718636607847296,
      "loss": 2.2231,
      "step": 32017
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6049776077270508,
      "learning_rate": 0.0005718619311749408,
      "loss": 2.272,
      "step": 32018
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.7060830593109131,
      "learning_rate": 0.0005718602015146078,
      "loss": 2.4208,
      "step": 32019
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6311477422714233,
      "learning_rate": 0.0005718584718037308,
      "loss": 2.4141,
      "step": 32020
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.7102562785148621,
      "learning_rate": 0.0005718567420423101,
      "loss": 2.4579,
      "step": 32021
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6944526433944702,
      "learning_rate": 0.0005718550122303463,
      "loss": 2.387,
      "step": 32022
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6851134896278381,
      "learning_rate": 0.0005718532823678394,
      "loss": 2.1971,
      "step": 32023
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6989448666572571,
      "learning_rate": 0.0005718515524547899,
      "loss": 2.3967,
      "step": 32024
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.7958575487136841,
      "learning_rate": 0.000571849822491198,
      "loss": 2.4795,
      "step": 32025
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.7098931670188904,
      "learning_rate": 0.000571848092477064,
      "loss": 2.6621,
      "step": 32026
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6248765587806702,
      "learning_rate": 0.0005718463624123885,
      "loss": 2.3776,
      "step": 32027
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.7731048464775085,
      "learning_rate": 0.0005718446322971716,
      "loss": 2.1811,
      "step": 32028
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.730506181716919,
      "learning_rate": 0.0005718429021314135,
      "loss": 2.5693,
      "step": 32029
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6648112535476685,
      "learning_rate": 0.0005718411719151148,
      "loss": 2.3989,
      "step": 32030
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.7197253108024597,
      "learning_rate": 0.0005718394416482757,
      "loss": 2.3776,
      "step": 32031
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.9867809414863586,
      "learning_rate": 0.0005718377113308965,
      "loss": 2.3323,
      "step": 32032
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6556035876274109,
      "learning_rate": 0.0005718359809629774,
      "loss": 2.2159,
      "step": 32033
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6753393411636353,
      "learning_rate": 0.0005718342505445191,
      "loss": 2.2436,
      "step": 32034
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.773594081401825,
      "learning_rate": 0.0005718325200755215,
      "loss": 2.4464,
      "step": 32035
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.7731587290763855,
      "learning_rate": 0.0005718307895559852,
      "loss": 2.4434,
      "step": 32036
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6430684328079224,
      "learning_rate": 0.0005718290589859104,
      "loss": 2.4951,
      "step": 32037
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6665039658546448,
      "learning_rate": 0.0005718273283652974,
      "loss": 2.3134,
      "step": 32038
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.7155477404594421,
      "learning_rate": 0.0005718255976941467,
      "loss": 2.4528,
      "step": 32039
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6710201501846313,
      "learning_rate": 0.0005718238669724584,
      "loss": 2.0651,
      "step": 32040
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6610179543495178,
      "learning_rate": 0.0005718221362002328,
      "loss": 2.2761,
      "step": 32041
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.7408264875411987,
      "learning_rate": 0.0005718204053774705,
      "loss": 2.4826,
      "step": 32042
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.670663595199585,
      "learning_rate": 0.0005718186745041717,
      "loss": 2.3093,
      "step": 32043
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6965833306312561,
      "learning_rate": 0.0005718169435803365,
      "loss": 2.1337,
      "step": 32044
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.7305876016616821,
      "learning_rate": 0.0005718152126059654,
      "loss": 2.3114,
      "step": 32045
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6523951292037964,
      "learning_rate": 0.0005718134815810589,
      "loss": 2.3565,
      "step": 32046
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.713822603225708,
      "learning_rate": 0.000571811750505617,
      "loss": 2.1683,
      "step": 32047
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6085225939750671,
      "learning_rate": 0.0005718100193796402,
      "loss": 2.2251,
      "step": 32048
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6594578623771667,
      "learning_rate": 0.0005718082882031288,
      "loss": 2.2513,
      "step": 32049
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6596841812133789,
      "learning_rate": 0.0005718065569760831,
      "loss": 2.5644,
      "step": 32050
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.7843074798583984,
      "learning_rate": 0.0005718048256985035,
      "loss": 2.3255,
      "step": 32051
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6232589483261108,
      "learning_rate": 0.0005718030943703902,
      "loss": 2.2675,
      "step": 32052
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.572169840335846,
      "learning_rate": 0.0005718013629917435,
      "loss": 2.1523,
      "step": 32053
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6886155605316162,
      "learning_rate": 0.0005717996315625639,
      "loss": 2.4527,
      "step": 32054
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.7649244070053101,
      "learning_rate": 0.0005717979000828515,
      "loss": 2.1004,
      "step": 32055
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6719003915786743,
      "learning_rate": 0.0005717961685526067,
      "loss": 2.3693,
      "step": 32056
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.7507323622703552,
      "learning_rate": 0.0005717944369718301,
      "loss": 2.3185,
      "step": 32057
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.7633970975875854,
      "learning_rate": 0.0005717927053405217,
      "loss": 2.4068,
      "step": 32058
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6062174439430237,
      "learning_rate": 0.0005717909736586818,
      "loss": 2.33,
      "step": 32059
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6205474138259888,
      "learning_rate": 0.0005717892419263109,
      "loss": 2.1453,
      "step": 32060
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.7188135981559753,
      "learning_rate": 0.0005717875101434092,
      "loss": 2.331,
      "step": 32061
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6378275752067566,
      "learning_rate": 0.0005717857783099771,
      "loss": 2.2972,
      "step": 32062
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.7171370983123779,
      "learning_rate": 0.0005717840464260149,
      "loss": 2.3177,
      "step": 32063
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6765181422233582,
      "learning_rate": 0.000571782314491523,
      "loss": 2.3125,
      "step": 32064
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6876214146614075,
      "learning_rate": 0.0005717805825065015,
      "loss": 2.4743,
      "step": 32065
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6347004771232605,
      "learning_rate": 0.0005717788504709509,
      "loss": 2.307,
      "step": 32066
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.614254355430603,
      "learning_rate": 0.0005717771183848716,
      "loss": 2.1894,
      "step": 32067
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6593526005744934,
      "learning_rate": 0.0005717753862482636,
      "loss": 2.2979,
      "step": 32068
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.7037262916564941,
      "learning_rate": 0.0005717736540611276,
      "loss": 2.3265,
      "step": 32069
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6738418936729431,
      "learning_rate": 0.0005717719218234636,
      "loss": 2.2679,
      "step": 32070
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6854364275932312,
      "learning_rate": 0.0005717701895352722,
      "loss": 2.3042,
      "step": 32071
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.7729782462120056,
      "learning_rate": 0.0005717684571965535,
      "loss": 2.3825,
      "step": 32072
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.614166796207428,
      "learning_rate": 0.000571766724807308,
      "loss": 2.3575,
      "step": 32073
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6859658360481262,
      "learning_rate": 0.0005717649923675358,
      "loss": 2.1877,
      "step": 32074
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6919586062431335,
      "learning_rate": 0.0005717632598772374,
      "loss": 2.0839,
      "step": 32075
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.636369526386261,
      "learning_rate": 0.0005717615273364132,
      "loss": 2.2141,
      "step": 32076
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.7242621779441833,
      "learning_rate": 0.0005717597947450633,
      "loss": 2.1277,
      "step": 32077
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.7748643755912781,
      "learning_rate": 0.0005717580621031882,
      "loss": 2.2322,
      "step": 32078
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6189572811126709,
      "learning_rate": 0.0005717563294107881,
      "loss": 1.9637,
      "step": 32079
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.7087200284004211,
      "learning_rate": 0.0005717545966678634,
      "loss": 2.4095,
      "step": 32080
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.841663658618927,
      "learning_rate": 0.0005717528638744143,
      "loss": 2.2488,
      "step": 32081
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6836140155792236,
      "learning_rate": 0.0005717511310304413,
      "loss": 2.3294,
      "step": 32082
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.0679162740707397,
      "learning_rate": 0.0005717493981359445,
      "loss": 2.2624,
      "step": 32083
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.7238181829452515,
      "learning_rate": 0.0005717476651909245,
      "loss": 2.1911,
      "step": 32084
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6231971979141235,
      "learning_rate": 0.0005717459321953815,
      "loss": 2.4991,
      "step": 32085
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.8284510374069214,
      "learning_rate": 0.0005717441991493158,
      "loss": 2.1856,
      "step": 32086
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6458067297935486,
      "learning_rate": 0.0005717424660527276,
      "loss": 2.1218,
      "step": 32087
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6723542213439941,
      "learning_rate": 0.0005717407329056174,
      "loss": 2.2149,
      "step": 32088
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6398596167564392,
      "learning_rate": 0.0005717389997079855,
      "loss": 2.4982,
      "step": 32089
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.7117432951927185,
      "learning_rate": 0.0005717372664598322,
      "loss": 2.1947,
      "step": 32090
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.5946319699287415,
      "learning_rate": 0.0005717355331611578,
      "loss": 2.2027,
      "step": 32091
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6237478852272034,
      "learning_rate": 0.0005717337998119626,
      "loss": 2.246,
      "step": 32092
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6647538542747498,
      "learning_rate": 0.0005717320664122469,
      "loss": 2.1397,
      "step": 32093
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.7035808563232422,
      "learning_rate": 0.0005717303329620112,
      "loss": 2.2099,
      "step": 32094
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6362962126731873,
      "learning_rate": 0.0005717285994612557,
      "loss": 2.2912,
      "step": 32095
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6574828028678894,
      "learning_rate": 0.0005717268659099806,
      "loss": 2.2387,
      "step": 32096
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.7005489468574524,
      "learning_rate": 0.0005717251323081865,
      "loss": 2.426,
      "step": 32097
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6486418843269348,
      "learning_rate": 0.0005717233986558734,
      "loss": 2.3279,
      "step": 32098
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6732982993125916,
      "learning_rate": 0.0005717216649530419,
      "loss": 2.3975,
      "step": 32099
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6531926393508911,
      "learning_rate": 0.0005717199311996922,
      "loss": 2.3602,
      "step": 32100
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6488668918609619,
      "learning_rate": 0.0005717181973958247,
      "loss": 2.1958,
      "step": 32101
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6731057167053223,
      "learning_rate": 0.0005717164635414395,
      "loss": 2.3498,
      "step": 32102
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6765093207359314,
      "learning_rate": 0.0005717147296365372,
      "loss": 2.4702,
      "step": 32103
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6569932103157043,
      "learning_rate": 0.000571712995681118,
      "loss": 2.2937,
      "step": 32104
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6866936683654785,
      "learning_rate": 0.0005717112616751822,
      "loss": 2.2769,
      "step": 32105
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6820205450057983,
      "learning_rate": 0.0005717095276187301,
      "loss": 2.2082,
      "step": 32106
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6462850570678711,
      "learning_rate": 0.0005717077935117621,
      "loss": 2.1803,
      "step": 32107
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.7056246399879456,
      "learning_rate": 0.0005717060593542786,
      "loss": 2.5046,
      "step": 32108
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6930593848228455,
      "learning_rate": 0.0005717043251462796,
      "loss": 2.2005,
      "step": 32109
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6676016449928284,
      "learning_rate": 0.0005717025908877659,
      "loss": 2.2559,
      "step": 32110
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6460157632827759,
      "learning_rate": 0.0005717008565787374,
      "loss": 2.267,
      "step": 32111
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.7090137004852295,
      "learning_rate": 0.0005716991222191947,
      "loss": 2.4976,
      "step": 32112
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.658694326877594,
      "learning_rate": 0.0005716973878091378,
      "loss": 2.3314,
      "step": 32113
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6109852194786072,
      "learning_rate": 0.0005716956533485673,
      "loss": 2.096,
      "step": 32114
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6344072818756104,
      "learning_rate": 0.0005716939188374835,
      "loss": 2.4294,
      "step": 32115
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6834247708320618,
      "learning_rate": 0.0005716921842758866,
      "loss": 2.2383,
      "step": 32116
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6824423670768738,
      "learning_rate": 0.000571690449663777,
      "loss": 2.5028,
      "step": 32117
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.5811241269111633,
      "learning_rate": 0.0005716887150011552,
      "loss": 2.2673,
      "step": 32118
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.7833662629127502,
      "learning_rate": 0.0005716869802880212,
      "loss": 2.6291,
      "step": 32119
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.8121705651283264,
      "learning_rate": 0.0005716852455243755,
      "loss": 2.3655,
      "step": 32120
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.7190194129943848,
      "learning_rate": 0.0005716835107102184,
      "loss": 2.3488,
      "step": 32121
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.7580857872962952,
      "learning_rate": 0.0005716817758455501,
      "loss": 2.3346,
      "step": 32122
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.7015305757522583,
      "learning_rate": 0.000571680040930371,
      "loss": 2.3807,
      "step": 32123
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.7106849551200867,
      "learning_rate": 0.0005716783059646816,
      "loss": 2.2903,
      "step": 32124
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6335483193397522,
      "learning_rate": 0.0005716765709484821,
      "loss": 2.1905,
      "step": 32125
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.7567780613899231,
      "learning_rate": 0.0005716748358817726,
      "loss": 2.3211,
      "step": 32126
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6802037358283997,
      "learning_rate": 0.0005716731007645538,
      "loss": 2.2556,
      "step": 32127
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6305330991744995,
      "learning_rate": 0.0005716713655968258,
      "loss": 2.2837,
      "step": 32128
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.7622405290603638,
      "learning_rate": 0.0005716696303785889,
      "loss": 2.1583,
      "step": 32129
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.7271397709846497,
      "learning_rate": 0.0005716678951098435,
      "loss": 2.481,
      "step": 32130
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.5933282375335693,
      "learning_rate": 0.00057166615979059,
      "loss": 2.3519,
      "step": 32131
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6623939871788025,
      "learning_rate": 0.0005716644244208285,
      "loss": 2.3537,
      "step": 32132
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6804381608963013,
      "learning_rate": 0.0005716626890005595,
      "loss": 2.0899,
      "step": 32133
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6671538949012756,
      "learning_rate": 0.0005716609535297834,
      "loss": 2.2317,
      "step": 32134
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6364361047744751,
      "learning_rate": 0.0005716592180085003,
      "loss": 2.3281,
      "step": 32135
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.7025483250617981,
      "learning_rate": 0.0005716574824367105,
      "loss": 2.4223,
      "step": 32136
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6781449913978577,
      "learning_rate": 0.0005716557468144147,
      "loss": 2.2953,
      "step": 32137
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6855984926223755,
      "learning_rate": 0.0005716540111416128,
      "loss": 2.3708,
      "step": 32138
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.690549910068512,
      "learning_rate": 0.0005716522754183054,
      "loss": 2.4419,
      "step": 32139
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.5870022773742676,
      "learning_rate": 0.0005716505396444927,
      "loss": 2.0443,
      "step": 32140
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.7054250836372375,
      "learning_rate": 0.0005716488038201749,
      "loss": 2.261,
      "step": 32141
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6116982698440552,
      "learning_rate": 0.0005716470679453526,
      "loss": 2.0822,
      "step": 32142
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6806920766830444,
      "learning_rate": 0.0005716453320200259,
      "loss": 2.4082,
      "step": 32143
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.760511577129364,
      "learning_rate": 0.0005716435960441953,
      "loss": 2.2853,
      "step": 32144
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6720314621925354,
      "learning_rate": 0.0005716418600178609,
      "loss": 2.5199,
      "step": 32145
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6401680707931519,
      "learning_rate": 0.0005716401239410232,
      "loss": 2.2026,
      "step": 32146
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6280673146247864,
      "learning_rate": 0.0005716383878136825,
      "loss": 2.289,
      "step": 32147
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6360148191452026,
      "learning_rate": 0.0005716366516358391,
      "loss": 2.4415,
      "step": 32148
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6643483638763428,
      "learning_rate": 0.0005716349154074932,
      "loss": 2.3304,
      "step": 32149
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.697169303894043,
      "learning_rate": 0.0005716331791286454,
      "loss": 2.2265,
      "step": 32150
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6400793194770813,
      "learning_rate": 0.0005716314427992959,
      "loss": 2.2221,
      "step": 32151
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.7246385812759399,
      "learning_rate": 0.0005716297064194448,
      "loss": 2.336,
      "step": 32152
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.7333021759986877,
      "learning_rate": 0.0005716279699890927,
      "loss": 2.3746,
      "step": 32153
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.7307521104812622,
      "learning_rate": 0.0005716262335082399,
      "loss": 2.1785,
      "step": 32154
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6212826371192932,
      "learning_rate": 0.0005716244969768866,
      "loss": 2.1968,
      "step": 32155
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.7363834381103516,
      "learning_rate": 0.0005716227603950331,
      "loss": 2.2664,
      "step": 32156
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6704673767089844,
      "learning_rate": 0.0005716210237626799,
      "loss": 2.11,
      "step": 32157
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6899068355560303,
      "learning_rate": 0.0005716192870798271,
      "loss": 2.4203,
      "step": 32158
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6461880803108215,
      "learning_rate": 0.0005716175503464753,
      "loss": 2.5038,
      "step": 32159
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6304767727851868,
      "learning_rate": 0.0005716158135626246,
      "loss": 2.4692,
      "step": 32160
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6419144868850708,
      "learning_rate": 0.0005716140767282754,
      "loss": 2.3039,
      "step": 32161
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6593001484870911,
      "learning_rate": 0.0005716123398434281,
      "loss": 2.2483,
      "step": 32162
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6420608162879944,
      "learning_rate": 0.0005716106029080829,
      "loss": 2.1685,
      "step": 32163
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.7656771540641785,
      "learning_rate": 0.0005716088659222401,
      "loss": 2.1327,
      "step": 32164
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6770532131195068,
      "learning_rate": 0.0005716071288859001,
      "loss": 2.2785,
      "step": 32165
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6220195293426514,
      "learning_rate": 0.0005716053917990632,
      "loss": 2.5178,
      "step": 32166
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6341894268989563,
      "learning_rate": 0.0005716036546617298,
      "loss": 2.469,
      "step": 32167
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.65888512134552,
      "learning_rate": 0.0005716019174739,
      "loss": 2.1483,
      "step": 32168
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.7033528089523315,
      "learning_rate": 0.0005716001802355745,
      "loss": 2.3127,
      "step": 32169
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.7008150815963745,
      "learning_rate": 0.0005715984429467533,
      "loss": 2.2844,
      "step": 32170
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6170913577079773,
      "learning_rate": 0.0005715967056074367,
      "loss": 2.4318,
      "step": 32171
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6072471141815186,
      "learning_rate": 0.0005715949682176252,
      "loss": 2.1664,
      "step": 32172
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.694648265838623,
      "learning_rate": 0.0005715932307773191,
      "loss": 2.408,
      "step": 32173
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.5969669222831726,
      "learning_rate": 0.0005715914932865188,
      "loss": 2.4216,
      "step": 32174
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6004573702812195,
      "learning_rate": 0.0005715897557452244,
      "loss": 2.2471,
      "step": 32175
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6787561178207397,
      "learning_rate": 0.0005715880181534364,
      "loss": 2.3018,
      "step": 32176
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.7685699462890625,
      "learning_rate": 0.000571586280511155,
      "loss": 2.5094,
      "step": 32177
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.645889163017273,
      "learning_rate": 0.0005715845428183805,
      "loss": 2.3566,
      "step": 32178
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.7149627208709717,
      "learning_rate": 0.0005715828050751135,
      "loss": 2.3441,
      "step": 32179
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.7013083100318909,
      "learning_rate": 0.000571581067281354,
      "loss": 2.0512,
      "step": 32180
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.645520806312561,
      "learning_rate": 0.0005715793294371025,
      "loss": 2.2516,
      "step": 32181
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6605992913246155,
      "learning_rate": 0.0005715775915423593,
      "loss": 2.1029,
      "step": 32182
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6399250030517578,
      "learning_rate": 0.0005715758535971247,
      "loss": 2.3077,
      "step": 32183
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.629907488822937,
      "learning_rate": 0.0005715741156013989,
      "loss": 2.3633,
      "step": 32184
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.7696505784988403,
      "learning_rate": 0.0005715723775551825,
      "loss": 2.3241,
      "step": 32185
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.7794349789619446,
      "learning_rate": 0.0005715706394584756,
      "loss": 2.5403,
      "step": 32186
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.7358105778694153,
      "learning_rate": 0.0005715689013112786,
      "loss": 2.2028,
      "step": 32187
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6997547149658203,
      "learning_rate": 0.0005715671631135918,
      "loss": 2.4562,
      "step": 32188
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.7773988842964172,
      "learning_rate": 0.0005715654248654155,
      "loss": 2.2086,
      "step": 32189
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6935365200042725,
      "learning_rate": 0.0005715636865667502,
      "loss": 2.1594,
      "step": 32190
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6902902722358704,
      "learning_rate": 0.000571561948217596,
      "loss": 2.3996,
      "step": 32191
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.7088010311126709,
      "learning_rate": 0.0005715602098179533,
      "loss": 2.2913,
      "step": 32192
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6735863089561462,
      "learning_rate": 0.0005715584713678224,
      "loss": 2.4082,
      "step": 32193
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6554372310638428,
      "learning_rate": 0.0005715567328672037,
      "loss": 2.3461,
      "step": 32194
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6935067772865295,
      "learning_rate": 0.0005715549943160975,
      "loss": 2.3679,
      "step": 32195
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.7227579951286316,
      "learning_rate": 0.0005715532557145041,
      "loss": 2.2587,
      "step": 32196
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.7291141152381897,
      "learning_rate": 0.0005715515170624238,
      "loss": 2.2239,
      "step": 32197
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6973947882652283,
      "learning_rate": 0.0005715497783598569,
      "loss": 2.1686,
      "step": 32198
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.7399954795837402,
      "learning_rate": 0.0005715480396068038,
      "loss": 2.3792,
      "step": 32199
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.8000969886779785,
      "learning_rate": 0.0005715463008032647,
      "loss": 2.3972,
      "step": 32200
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6427513957023621,
      "learning_rate": 0.0005715445619492403,
      "loss": 2.3122,
      "step": 32201
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6334254741668701,
      "learning_rate": 0.0005715428230447304,
      "loss": 2.4866,
      "step": 32202
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6590074300765991,
      "learning_rate": 0.0005715410840897356,
      "loss": 2.2879,
      "step": 32203
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.748401939868927,
      "learning_rate": 0.0005715393450842562,
      "loss": 1.9981,
      "step": 32204
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6943451762199402,
      "learning_rate": 0.0005715376060282925,
      "loss": 2.4554,
      "step": 32205
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6715142726898193,
      "learning_rate": 0.0005715358669218449,
      "loss": 2.2346,
      "step": 32206
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6317419409751892,
      "learning_rate": 0.0005715341277649135,
      "loss": 2.3592,
      "step": 32207
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6403551697731018,
      "learning_rate": 0.0005715323885574989,
      "loss": 2.3537,
      "step": 32208
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6876906156539917,
      "learning_rate": 0.0005715306492996014,
      "loss": 2.2848,
      "step": 32209
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6103476285934448,
      "learning_rate": 0.0005715289099912211,
      "loss": 2.4316,
      "step": 32210
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6768937110900879,
      "learning_rate": 0.0005715271706323585,
      "loss": 2.3342,
      "step": 32211
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.7324227094650269,
      "learning_rate": 0.0005715254312230138,
      "loss": 2.2358,
      "step": 32212
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6759381890296936,
      "learning_rate": 0.0005715236917631874,
      "loss": 2.5315,
      "step": 32213
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6468250155448914,
      "learning_rate": 0.0005715219522528797,
      "loss": 2.3549,
      "step": 32214
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.69557124376297,
      "learning_rate": 0.0005715202126920909,
      "loss": 2.4891,
      "step": 32215
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.7818008661270142,
      "learning_rate": 0.0005715184730808214,
      "loss": 2.242,
      "step": 32216
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6322148442268372,
      "learning_rate": 0.0005715167334190715,
      "loss": 2.4435,
      "step": 32217
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6370491981506348,
      "learning_rate": 0.0005715149937068415,
      "loss": 2.2462,
      "step": 32218
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6959673762321472,
      "learning_rate": 0.0005715132539441316,
      "loss": 2.2074,
      "step": 32219
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6553263664245605,
      "learning_rate": 0.0005715115141309425,
      "loss": 2.2909,
      "step": 32220
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.7346813678741455,
      "learning_rate": 0.0005715097742672741,
      "loss": 2.2999,
      "step": 32221
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.8095971345901489,
      "learning_rate": 0.000571508034353127,
      "loss": 2.5338,
      "step": 32222
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.66330885887146,
      "learning_rate": 0.0005715062943885014,
      "loss": 2.4633,
      "step": 32223
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6393035054206848,
      "learning_rate": 0.0005715045543733977,
      "loss": 2.3275,
      "step": 32224
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.7776737809181213,
      "learning_rate": 0.000571502814307816,
      "loss": 2.3484,
      "step": 32225
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.7200323343276978,
      "learning_rate": 0.000571501074191757,
      "loss": 2.257,
      "step": 32226
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6144558787345886,
      "learning_rate": 0.0005714993340252207,
      "loss": 2.0557,
      "step": 32227
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.7056804895401001,
      "learning_rate": 0.0005714975938082077,
      "loss": 2.3669,
      "step": 32228
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6972220540046692,
      "learning_rate": 0.000571495853540718,
      "loss": 2.0023,
      "step": 32229
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6678051948547363,
      "learning_rate": 0.0005714941132227522,
      "loss": 2.2884,
      "step": 32230
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6815411448478699,
      "learning_rate": 0.0005714923728543105,
      "loss": 2.0266,
      "step": 32231
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.7881183624267578,
      "learning_rate": 0.0005714906324353932,
      "loss": 2.6014,
      "step": 32232
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.624442994594574,
      "learning_rate": 0.0005714888919660007,
      "loss": 2.3714,
      "step": 32233
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6660968065261841,
      "learning_rate": 0.0005714871514461333,
      "loss": 2.4486,
      "step": 32234
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.7257723212242126,
      "learning_rate": 0.0005714854108757913,
      "loss": 2.1703,
      "step": 32235
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6556735038757324,
      "learning_rate": 0.0005714836702549751,
      "loss": 2.419,
      "step": 32236
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.7223790884017944,
      "learning_rate": 0.0005714819295836849,
      "loss": 2.3138,
      "step": 32237
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6663012504577637,
      "learning_rate": 0.0005714801888619211,
      "loss": 2.163,
      "step": 32238
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6575232148170471,
      "learning_rate": 0.000571478448089684,
      "loss": 2.1849,
      "step": 32239
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.62752366065979,
      "learning_rate": 0.000571476707266974,
      "loss": 2.2024,
      "step": 32240
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6170885562896729,
      "learning_rate": 0.0005714749663937913,
      "loss": 2.2199,
      "step": 32241
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.655951976776123,
      "learning_rate": 0.0005714732254701361,
      "loss": 2.3681,
      "step": 32242
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6648052930831909,
      "learning_rate": 0.0005714714844960091,
      "loss": 2.2812,
      "step": 32243
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6814670562744141,
      "learning_rate": 0.0005714697434714104,
      "loss": 2.2091,
      "step": 32244
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6339695453643799,
      "learning_rate": 0.0005714680023963404,
      "loss": 2.2428,
      "step": 32245
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6063396334648132,
      "learning_rate": 0.0005714662612707993,
      "loss": 2.1398,
      "step": 32246
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6107755899429321,
      "learning_rate": 0.0005714645200947876,
      "loss": 2.3486,
      "step": 32247
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.5958898663520813,
      "learning_rate": 0.0005714627788683054,
      "loss": 2.3459,
      "step": 32248
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.667070209980011,
      "learning_rate": 0.0005714610375913531,
      "loss": 2.1525,
      "step": 32249
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.7615095376968384,
      "learning_rate": 0.0005714592962639312,
      "loss": 2.3864,
      "step": 32250
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.640748918056488,
      "learning_rate": 0.0005714575548860398,
      "loss": 2.4188,
      "step": 32251
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.5675365924835205,
      "learning_rate": 0.0005714558134576794,
      "loss": 2.3426,
      "step": 32252
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.7594522833824158,
      "learning_rate": 0.0005714540719788502,
      "loss": 2.3197,
      "step": 32253
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6537696719169617,
      "learning_rate": 0.0005714523304495527,
      "loss": 2.3159,
      "step": 32254
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6128390431404114,
      "learning_rate": 0.0005714505888697868,
      "loss": 2.3168,
      "step": 32255
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6228113174438477,
      "learning_rate": 0.0005714488472395533,
      "loss": 2.6706,
      "step": 32256
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6720985770225525,
      "learning_rate": 0.0005714471055588522,
      "loss": 2.3808,
      "step": 32257
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6858435273170471,
      "learning_rate": 0.0005714453638276842,
      "loss": 2.4683,
      "step": 32258
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6910293698310852,
      "learning_rate": 0.0005714436220460493,
      "loss": 2.2243,
      "step": 32259
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6279575228691101,
      "learning_rate": 0.0005714418802139477,
      "loss": 2.4349,
      "step": 32260
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6692288517951965,
      "learning_rate": 0.0005714401383313801,
      "loss": 2.3814,
      "step": 32261
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6618451476097107,
      "learning_rate": 0.0005714383963983467,
      "loss": 2.1809,
      "step": 32262
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6111489534378052,
      "learning_rate": 0.0005714366544148477,
      "loss": 2.1475,
      "step": 32263
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6612722873687744,
      "learning_rate": 0.0005714349123808835,
      "loss": 2.2854,
      "step": 32264
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6887754201889038,
      "learning_rate": 0.0005714331702964544,
      "loss": 2.4352,
      "step": 32265
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.7586331367492676,
      "learning_rate": 0.0005714314281615609,
      "loss": 2.3859,
      "step": 32266
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6778078675270081,
      "learning_rate": 0.000571429685976203,
      "loss": 2.4888,
      "step": 32267
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.7120853662490845,
      "learning_rate": 0.0005714279437403812,
      "loss": 2.2134,
      "step": 32268
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.7007701396942139,
      "learning_rate": 0.000571426201454096,
      "loss": 2.2928,
      "step": 32269
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6749770641326904,
      "learning_rate": 0.0005714244591173474,
      "loss": 2.1542,
      "step": 32270
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6138307452201843,
      "learning_rate": 0.000571422716730136,
      "loss": 2.1178,
      "step": 32271
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.7407337427139282,
      "learning_rate": 0.0005714209742924618,
      "loss": 2.1946,
      "step": 32272
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6419659852981567,
      "learning_rate": 0.0005714192318043254,
      "loss": 2.4788,
      "step": 32273
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.7729620933532715,
      "learning_rate": 0.0005714174892657271,
      "loss": 2.3804,
      "step": 32274
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.8175081610679626,
      "learning_rate": 0.000571415746676667,
      "loss": 2.3684,
      "step": 32275
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6285266876220703,
      "learning_rate": 0.0005714140040371458,
      "loss": 2.0846,
      "step": 32276
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6531391143798828,
      "learning_rate": 0.0005714122613471635,
      "loss": 2.0765,
      "step": 32277
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.7226482629776001,
      "learning_rate": 0.0005714105186067206,
      "loss": 2.0999,
      "step": 32278
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6616166830062866,
      "learning_rate": 0.0005714087758158173,
      "loss": 2.3476,
      "step": 32279
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.797278106212616,
      "learning_rate": 0.000571407032974454,
      "loss": 2.2767,
      "step": 32280
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.7008540630340576,
      "learning_rate": 0.000571405290082631,
      "loss": 2.3038,
      "step": 32281
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6289063692092896,
      "learning_rate": 0.0005714035471403487,
      "loss": 2.2258,
      "step": 32282
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6189079880714417,
      "learning_rate": 0.0005714018041476073,
      "loss": 2.4091,
      "step": 32283
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.8286280632019043,
      "learning_rate": 0.0005714000611044071,
      "loss": 2.2042,
      "step": 32284
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.691105306148529,
      "learning_rate": 0.0005713983180107486,
      "loss": 2.2147,
      "step": 32285
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6923825740814209,
      "learning_rate": 0.000571396574866632,
      "loss": 2.234,
      "step": 32286
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.8695960640907288,
      "learning_rate": 0.0005713948316720577,
      "loss": 2.2015,
      "step": 32287
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6490801572799683,
      "learning_rate": 0.0005713930884270259,
      "loss": 2.2702,
      "step": 32288
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.7194511294364929,
      "learning_rate": 0.0005713913451315372,
      "loss": 2.2717,
      "step": 32289
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.7610693573951721,
      "learning_rate": 0.0005713896017855916,
      "loss": 2.2891,
      "step": 32290
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6439123749732971,
      "learning_rate": 0.0005713878583891894,
      "loss": 2.2553,
      "step": 32291
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.7083842158317566,
      "learning_rate": 0.0005713861149423313,
      "loss": 2.2787,
      "step": 32292
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.630435049533844,
      "learning_rate": 0.0005713843714450173,
      "loss": 2.4543,
      "step": 32293
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.9023119807243347,
      "learning_rate": 0.0005713826278972478,
      "loss": 2.2677,
      "step": 32294
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.7375211119651794,
      "learning_rate": 0.0005713808842990231,
      "loss": 2.1913,
      "step": 32295
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.7796372175216675,
      "learning_rate": 0.0005713791406503437,
      "loss": 2.1605,
      "step": 32296
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.8467001914978027,
      "learning_rate": 0.0005713773969512098,
      "loss": 2.3095,
      "step": 32297
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.7098126411437988,
      "learning_rate": 0.0005713756532016216,
      "loss": 2.2456,
      "step": 32298
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.5919068455696106,
      "learning_rate": 0.0005713739094015797,
      "loss": 2.2489,
      "step": 32299
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.7129660248756409,
      "learning_rate": 0.0005713721655510841,
      "loss": 2.4797,
      "step": 32300
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.717958390712738,
      "learning_rate": 0.0005713704216501354,
      "loss": 2.21,
      "step": 32301
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.7215579152107239,
      "learning_rate": 0.0005713686776987338,
      "loss": 2.3212,
      "step": 32302
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6794831156730652,
      "learning_rate": 0.0005713669336968796,
      "loss": 2.373,
      "step": 32303
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.7033449411392212,
      "learning_rate": 0.0005713651896445732,
      "loss": 2.3807,
      "step": 32304
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6692275404930115,
      "learning_rate": 0.0005713634455418148,
      "loss": 2.4658,
      "step": 32305
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6601129770278931,
      "learning_rate": 0.0005713617013886048,
      "loss": 2.231,
      "step": 32306
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.7657373547554016,
      "learning_rate": 0.0005713599571849437,
      "loss": 2.1831,
      "step": 32307
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.9147326946258545,
      "learning_rate": 0.0005713582129308316,
      "loss": 2.2579,
      "step": 32308
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6572837829589844,
      "learning_rate": 0.0005713564686262688,
      "loss": 2.3744,
      "step": 32309
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.687547504901886,
      "learning_rate": 0.0005713547242712559,
      "loss": 2.3959,
      "step": 32310
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.0875847339630127,
      "learning_rate": 0.0005713529798657929,
      "loss": 2.3948,
      "step": 32311
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6651904582977295,
      "learning_rate": 0.0005713512354098803,
      "loss": 2.183,
      "step": 32312
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6854314804077148,
      "learning_rate": 0.0005713494909035183,
      "loss": 2.2486,
      "step": 32313
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.7418965101242065,
      "learning_rate": 0.0005713477463467074,
      "loss": 2.1794,
      "step": 32314
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.8440613150596619,
      "learning_rate": 0.0005713460017394477,
      "loss": 2.1546,
      "step": 32315
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.630004346370697,
      "learning_rate": 0.0005713442570817398,
      "loss": 2.1643,
      "step": 32316
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.7028512954711914,
      "learning_rate": 0.0005713425123735838,
      "loss": 2.1551,
      "step": 32317
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.7758232951164246,
      "learning_rate": 0.0005713407676149802,
      "loss": 2.1869,
      "step": 32318
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6245988607406616,
      "learning_rate": 0.0005713390228059292,
      "loss": 2.3202,
      "step": 32319
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6484389305114746,
      "learning_rate": 0.0005713372779464311,
      "loss": 2.1538,
      "step": 32320
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6809162497520447,
      "learning_rate": 0.0005713355330364863,
      "loss": 2.4023,
      "step": 32321
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.7814383506774902,
      "learning_rate": 0.0005713337880760951,
      "loss": 2.5436,
      "step": 32322
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6905731558799744,
      "learning_rate": 0.0005713320430652578,
      "loss": 2.4612,
      "step": 32323
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6316004395484924,
      "learning_rate": 0.0005713302980039748,
      "loss": 2.3519,
      "step": 32324
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6884388327598572,
      "learning_rate": 0.0005713285528922464,
      "loss": 2.4471,
      "step": 32325
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6615322232246399,
      "learning_rate": 0.0005713268077300728,
      "loss": 2.2775,
      "step": 32326
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6759481430053711,
      "learning_rate": 0.0005713250625174545,
      "loss": 2.2522,
      "step": 32327
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.679980993270874,
      "learning_rate": 0.0005713233172543917,
      "loss": 2.3567,
      "step": 32328
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.7919862270355225,
      "learning_rate": 0.0005713215719408848,
      "loss": 2.4515,
      "step": 32329
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6814256310462952,
      "learning_rate": 0.0005713198265769341,
      "loss": 2.1972,
      "step": 32330
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6570677757263184,
      "learning_rate": 0.0005713180811625399,
      "loss": 2.1757,
      "step": 32331
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6824288964271545,
      "learning_rate": 0.0005713163356977026,
      "loss": 2.2839,
      "step": 32332
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6588274240493774,
      "learning_rate": 0.0005713145901824224,
      "loss": 2.2942,
      "step": 32333
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6802038550376892,
      "learning_rate": 0.0005713128446166997,
      "loss": 2.4955,
      "step": 32334
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6473526954650879,
      "learning_rate": 0.000571311099000535,
      "loss": 2.2098,
      "step": 32335
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.734626829624176,
      "learning_rate": 0.0005713093533339282,
      "loss": 2.384,
      "step": 32336
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6005306839942932,
      "learning_rate": 0.0005713076076168799,
      "loss": 2.3304,
      "step": 32337
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.7040048837661743,
      "learning_rate": 0.0005713058618493906,
      "loss": 2.4916,
      "step": 32338
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.8257836699485779,
      "learning_rate": 0.0005713041160314603,
      "loss": 2.1171,
      "step": 32339
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6350681781768799,
      "learning_rate": 0.0005713023701630894,
      "loss": 2.1864,
      "step": 32340
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6899385452270508,
      "learning_rate": 0.0005713006242442783,
      "loss": 1.9918,
      "step": 32341
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6888870000839233,
      "learning_rate": 0.0005712988782750273,
      "loss": 2.1655,
      "step": 32342
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.7811228632926941,
      "learning_rate": 0.0005712971322553367,
      "loss": 2.2142,
      "step": 32343
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.7222144603729248,
      "learning_rate": 0.0005712953861852068,
      "loss": 2.4673,
      "step": 32344
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.7128061056137085,
      "learning_rate": 0.0005712936400646381,
      "loss": 2.2491,
      "step": 32345
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6875178813934326,
      "learning_rate": 0.0005712918938936307,
      "loss": 2.4163,
      "step": 32346
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.7089011669158936,
      "learning_rate": 0.000571290147672185,
      "loss": 2.1836,
      "step": 32347
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.5911710262298584,
      "learning_rate": 0.0005712884014003013,
      "loss": 2.3784,
      "step": 32348
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.7037937045097351,
      "learning_rate": 0.0005712866550779801,
      "loss": 2.1238,
      "step": 32349
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6756162047386169,
      "learning_rate": 0.0005712849087052216,
      "loss": 2.0155,
      "step": 32350
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6198444366455078,
      "learning_rate": 0.000571283162282026,
      "loss": 2.2785,
      "step": 32351
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6679441928863525,
      "learning_rate": 0.0005712814158083938,
      "loss": 2.4589,
      "step": 32352
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.697127640247345,
      "learning_rate": 0.0005712796692843252,
      "loss": 2.2817,
      "step": 32353
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.8384247422218323,
      "learning_rate": 0.0005712779227098206,
      "loss": 2.1773,
      "step": 32354
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.5716528296470642,
      "learning_rate": 0.0005712761760848804,
      "loss": 2.3289,
      "step": 32355
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6477783918380737,
      "learning_rate": 0.0005712744294095048,
      "loss": 2.123,
      "step": 32356
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6303723454475403,
      "learning_rate": 0.0005712726826836941,
      "loss": 2.2631,
      "step": 32357
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.7396898865699768,
      "learning_rate": 0.0005712709359074487,
      "loss": 2.3121,
      "step": 32358
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.651396632194519,
      "learning_rate": 0.000571269189080769,
      "loss": 2.5395,
      "step": 32359
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.711785078048706,
      "learning_rate": 0.0005712674422036551,
      "loss": 2.1524,
      "step": 32360
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.7624623775482178,
      "learning_rate": 0.0005712656952761076,
      "loss": 2.1992,
      "step": 32361
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.713416576385498,
      "learning_rate": 0.0005712639482981267,
      "loss": 2.2661,
      "step": 32362
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.8352981209754944,
      "learning_rate": 0.0005712622012697126,
      "loss": 2.2111,
      "step": 32363
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.8275866508483887,
      "learning_rate": 0.0005712604541908658,
      "loss": 2.405,
      "step": 32364
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6493760943412781,
      "learning_rate": 0.0005712587070615864,
      "loss": 2.2443,
      "step": 32365
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.7757260799407959,
      "learning_rate": 0.0005712569598818751,
      "loss": 2.1626,
      "step": 32366
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.0035638809204102,
      "learning_rate": 0.000571255212651732,
      "loss": 2.4925,
      "step": 32367
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6077263951301575,
      "learning_rate": 0.0005712534653711573,
      "loss": 2.3351,
      "step": 32368
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6994255185127258,
      "learning_rate": 0.0005712517180401515,
      "loss": 2.134,
      "step": 32369
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.8102447390556335,
      "learning_rate": 0.000571249970658715,
      "loss": 2.4757,
      "step": 32370
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.73291015625,
      "learning_rate": 0.0005712482232268479,
      "loss": 2.3383,
      "step": 32371
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6783652305603027,
      "learning_rate": 0.0005712464757445506,
      "loss": 2.498,
      "step": 32372
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6778112649917603,
      "learning_rate": 0.0005712447282118236,
      "loss": 2.4365,
      "step": 32373
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.0136598348617554,
      "learning_rate": 0.000571242980628667,
      "loss": 2.1059,
      "step": 32374
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.5941076874732971,
      "learning_rate": 0.0005712412329950811,
      "loss": 2.6146,
      "step": 32375
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6465873718261719,
      "learning_rate": 0.0005712394853110665,
      "loss": 2.2791,
      "step": 32376
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6610643267631531,
      "learning_rate": 0.0005712377375766233,
      "loss": 2.152,
      "step": 32377
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.7200633883476257,
      "learning_rate": 0.0005712359897917519,
      "loss": 2.4954,
      "step": 32378
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.7445335984230042,
      "learning_rate": 0.0005712342419564526,
      "loss": 2.2441,
      "step": 32379
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.7223411798477173,
      "learning_rate": 0.0005712324940707258,
      "loss": 2.0701,
      "step": 32380
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.7432184219360352,
      "learning_rate": 0.0005712307461345717,
      "loss": 2.0811,
      "step": 32381
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.7731726765632629,
      "learning_rate": 0.0005712289981479906,
      "loss": 2.2536,
      "step": 32382
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.731504499912262,
      "learning_rate": 0.0005712272501109831,
      "loss": 2.2742,
      "step": 32383
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.646340548992157,
      "learning_rate": 0.0005712255020235492,
      "loss": 2.0359,
      "step": 32384
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6460008025169373,
      "learning_rate": 0.0005712237538856894,
      "loss": 2.3112,
      "step": 32385
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.7956874370574951,
      "learning_rate": 0.000571222005697404,
      "loss": 2.3232,
      "step": 32386
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.7155167460441589,
      "learning_rate": 0.0005712202574586932,
      "loss": 2.4767,
      "step": 32387
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6413207650184631,
      "learning_rate": 0.0005712185091695575,
      "loss": 2.2815,
      "step": 32388
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6536709666252136,
      "learning_rate": 0.0005712167608299971,
      "loss": 2.2486,
      "step": 32389
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.9647409319877625,
      "learning_rate": 0.0005712150124400125,
      "loss": 2.1424,
      "step": 32390
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6559401154518127,
      "learning_rate": 0.000571213263999604,
      "loss": 2.5889,
      "step": 32391
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.9507277011871338,
      "learning_rate": 0.0005712115155087717,
      "loss": 2.3522,
      "step": 32392
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6879463195800781,
      "learning_rate": 0.000571209766967516,
      "loss": 2.1574,
      "step": 32393
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6387650966644287,
      "learning_rate": 0.0005712080183758373,
      "loss": 2.2489,
      "step": 32394
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.7464394569396973,
      "learning_rate": 0.000571206269733736,
      "loss": 2.2809,
      "step": 32395
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6738227009773254,
      "learning_rate": 0.0005712045210412123,
      "loss": 2.3422,
      "step": 32396
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.9356345534324646,
      "learning_rate": 0.0005712027722982665,
      "loss": 2.6238,
      "step": 32397
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6690847277641296,
      "learning_rate": 0.0005712010235048991,
      "loss": 2.2912,
      "step": 32398
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.7792332172393799,
      "learning_rate": 0.0005711992746611103,
      "loss": 2.4274,
      "step": 32399
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6970400214195251,
      "learning_rate": 0.0005711975257669004,
      "loss": 2.347,
      "step": 32400
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6430476903915405,
      "learning_rate": 0.0005711957768222696,
      "loss": 2.3116,
      "step": 32401
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.7213853001594543,
      "learning_rate": 0.0005711940278272185,
      "loss": 2.206,
      "step": 32402
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6316695809364319,
      "learning_rate": 0.0005711922787817475,
      "loss": 2.1889,
      "step": 32403
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.7266687154769897,
      "learning_rate": 0.0005711905296858565,
      "loss": 2.4423,
      "step": 32404
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.7464798092842102,
      "learning_rate": 0.0005711887805395462,
      "loss": 2.4751,
      "step": 32405
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6624904274940491,
      "learning_rate": 0.0005711870313428167,
      "loss": 2.5399,
      "step": 32406
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6283560395240784,
      "learning_rate": 0.0005711852820956684,
      "loss": 2.2429,
      "step": 32407
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6153599619865417,
      "learning_rate": 0.0005711835327981017,
      "loss": 2.3377,
      "step": 32408
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6825348734855652,
      "learning_rate": 0.0005711817834501169,
      "loss": 2.2672,
      "step": 32409
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6661182045936584,
      "learning_rate": 0.0005711800340517142,
      "loss": 2.0685,
      "step": 32410
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6229424476623535,
      "learning_rate": 0.000571178284602894,
      "loss": 2.3418,
      "step": 32411
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6289379000663757,
      "learning_rate": 0.0005711765351036567,
      "loss": 2.2261,
      "step": 32412
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6681696772575378,
      "learning_rate": 0.0005711747855540025,
      "loss": 2.3938,
      "step": 32413
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6940783858299255,
      "learning_rate": 0.0005711730359539318,
      "loss": 2.3911,
      "step": 32414
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6453670263290405,
      "learning_rate": 0.0005711712863034449,
      "loss": 2.135,
      "step": 32415
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.694276750087738,
      "learning_rate": 0.0005711695366025423,
      "loss": 2.2791,
      "step": 32416
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6234314441680908,
      "learning_rate": 0.0005711677868512241,
      "loss": 2.1044,
      "step": 32417
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6391426920890808,
      "learning_rate": 0.0005711660370494906,
      "loss": 2.0802,
      "step": 32418
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.7064297795295715,
      "learning_rate": 0.0005711642871973422,
      "loss": 2.3189,
      "step": 32419
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6527315974235535,
      "learning_rate": 0.0005711625372947793,
      "loss": 2.3485,
      "step": 32420
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6645267605781555,
      "learning_rate": 0.0005711607873418021,
      "loss": 2.2959,
      "step": 32421
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.715579092502594,
      "learning_rate": 0.0005711590373384111,
      "loss": 2.4969,
      "step": 32422
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.5925572514533997,
      "learning_rate": 0.0005711572872846065,
      "loss": 2.179,
      "step": 32423
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6589568853378296,
      "learning_rate": 0.0005711555371803886,
      "loss": 2.2688,
      "step": 32424
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6133939623832703,
      "learning_rate": 0.0005711537870257576,
      "loss": 2.05,
      "step": 32425
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.634028971195221,
      "learning_rate": 0.0005711520368207142,
      "loss": 2.2966,
      "step": 32426
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6856959462165833,
      "learning_rate": 0.0005711502865652586,
      "loss": 2.5303,
      "step": 32427
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6713538765907288,
      "learning_rate": 0.0005711485362593909,
      "loss": 2.0507,
      "step": 32428
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.746862530708313,
      "learning_rate": 0.0005711467859031116,
      "loss": 2.1225,
      "step": 32429
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6666002869606018,
      "learning_rate": 0.0005711450354964209,
      "loss": 2.1706,
      "step": 32430
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.8073069453239441,
      "learning_rate": 0.0005711432850393194,
      "loss": 2.3341,
      "step": 32431
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.7456822991371155,
      "learning_rate": 0.0005711415345318071,
      "loss": 2.36,
      "step": 32432
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6423914432525635,
      "learning_rate": 0.0005711397839738845,
      "loss": 2.2962,
      "step": 32433
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6526744961738586,
      "learning_rate": 0.000571138033365552,
      "loss": 2.282,
      "step": 32434
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6967973709106445,
      "learning_rate": 0.0005711362827068098,
      "loss": 2.3032,
      "step": 32435
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6475844979286194,
      "learning_rate": 0.0005711345319976581,
      "loss": 2.1484,
      "step": 32436
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6276755928993225,
      "learning_rate": 0.0005711327812380975,
      "loss": 2.213,
      "step": 32437
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6197258830070496,
      "learning_rate": 0.0005711310304281281,
      "loss": 2.0526,
      "step": 32438
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6544364094734192,
      "learning_rate": 0.0005711292795677504,
      "loss": 2.3336,
      "step": 32439
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.598642110824585,
      "learning_rate": 0.0005711275286569646,
      "loss": 2.2654,
      "step": 32440
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6071107387542725,
      "learning_rate": 0.0005711257776957711,
      "loss": 2.3203,
      "step": 32441
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6388678550720215,
      "learning_rate": 0.0005711240266841703,
      "loss": 2.2926,
      "step": 32442
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6871987581253052,
      "learning_rate": 0.0005711222756221622,
      "loss": 2.3597,
      "step": 32443
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.740527868270874,
      "learning_rate": 0.0005711205245097475,
      "loss": 2.4017,
      "step": 32444
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6199372410774231,
      "learning_rate": 0.0005711187733469264,
      "loss": 2.4498,
      "step": 32445
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6610651016235352,
      "learning_rate": 0.0005711170221336992,
      "loss": 2.2683,
      "step": 32446
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6528065800666809,
      "learning_rate": 0.0005711152708700661,
      "loss": 2.2895,
      "step": 32447
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6322718858718872,
      "learning_rate": 0.0005711135195560277,
      "loss": 2.2763,
      "step": 32448
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6234006285667419,
      "learning_rate": 0.000571111768191584,
      "loss": 2.2011,
      "step": 32449
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6735838651657104,
      "learning_rate": 0.0005711100167767357,
      "loss": 2.2812,
      "step": 32450
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6754698753356934,
      "learning_rate": 0.0005711082653114829,
      "loss": 2.4766,
      "step": 32451
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6934701800346375,
      "learning_rate": 0.0005711065137958258,
      "loss": 2.0786,
      "step": 32452
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6564037203788757,
      "learning_rate": 0.0005711047622297651,
      "loss": 2.0333,
      "step": 32453
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6029623746871948,
      "learning_rate": 0.0005711030106133008,
      "loss": 2.441,
      "step": 32454
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.711216151714325,
      "learning_rate": 0.0005711012589464333,
      "loss": 2.1407,
      "step": 32455
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6292426586151123,
      "learning_rate": 0.000571099507229163,
      "loss": 2.3357,
      "step": 32456
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6537116765975952,
      "learning_rate": 0.0005710977554614903,
      "loss": 2.4613,
      "step": 32457
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6808401942253113,
      "learning_rate": 0.0005710960036434152,
      "loss": 2.1855,
      "step": 32458
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6399842500686646,
      "learning_rate": 0.0005710942517749382,
      "loss": 2.2255,
      "step": 32459
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6435319185256958,
      "learning_rate": 0.0005710924998560599,
      "loss": 2.1783,
      "step": 32460
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6383877396583557,
      "learning_rate": 0.0005710907478867803,
      "loss": 2.2078,
      "step": 32461
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6605584025382996,
      "learning_rate": 0.0005710889958670997,
      "loss": 2.4474,
      "step": 32462
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6128920912742615,
      "learning_rate": 0.0005710872437970187,
      "loss": 2.2942,
      "step": 32463
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.597043514251709,
      "learning_rate": 0.0005710854916765373,
      "loss": 2.3178,
      "step": 32464
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6614452004432678,
      "learning_rate": 0.0005710837395056562,
      "loss": 2.4261,
      "step": 32465
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.607246994972229,
      "learning_rate": 0.0005710819872843753,
      "loss": 2.3112,
      "step": 32466
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6777719259262085,
      "learning_rate": 0.0005710802350126952,
      "loss": 2.3823,
      "step": 32467
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.8088304400444031,
      "learning_rate": 0.0005710784826906162,
      "loss": 2.3605,
      "step": 32468
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6430881023406982,
      "learning_rate": 0.0005710767303181387,
      "loss": 2.4272,
      "step": 32469
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6465181708335876,
      "learning_rate": 0.0005710749778952628,
      "loss": 2.1931,
      "step": 32470
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.7063296437263489,
      "learning_rate": 0.000571073225421989,
      "loss": 2.233,
      "step": 32471
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6578213572502136,
      "learning_rate": 0.0005710714728983174,
      "loss": 2.3043,
      "step": 32472
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6258843541145325,
      "learning_rate": 0.0005710697203242486,
      "loss": 2.2639,
      "step": 32473
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6909924149513245,
      "learning_rate": 0.0005710679676997829,
      "loss": 2.0969,
      "step": 32474
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6299789547920227,
      "learning_rate": 0.0005710662150249206,
      "loss": 2.1053,
      "step": 32475
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.651125967502594,
      "learning_rate": 0.0005710644622996618,
      "loss": 2.2573,
      "step": 32476
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.8006042242050171,
      "learning_rate": 0.000571062709524007,
      "loss": 2.2858,
      "step": 32477
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6590927839279175,
      "learning_rate": 0.0005710609566979566,
      "loss": 2.5172,
      "step": 32478
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6504667401313782,
      "learning_rate": 0.0005710592038215108,
      "loss": 2.0634,
      "step": 32479
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.5916337370872498,
      "learning_rate": 0.0005710574508946702,
      "loss": 2.3258,
      "step": 32480
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6966269612312317,
      "learning_rate": 0.0005710556979174346,
      "loss": 2.4297,
      "step": 32481
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6617240309715271,
      "learning_rate": 0.0005710539448898048,
      "loss": 2.4532,
      "step": 32482
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.5855353474617004,
      "learning_rate": 0.0005710521918117809,
      "loss": 2.4949,
      "step": 32483
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.8245205879211426,
      "learning_rate": 0.0005710504386833632,
      "loss": 2.3343,
      "step": 32484
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.7298128008842468,
      "learning_rate": 0.0005710486855045523,
      "loss": 2.1751,
      "step": 32485
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6774377822875977,
      "learning_rate": 0.0005710469322753482,
      "loss": 2.2553,
      "step": 32486
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6757306456565857,
      "learning_rate": 0.0005710451789957514,
      "loss": 2.1556,
      "step": 32487
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.7753324508666992,
      "learning_rate": 0.0005710434256657622,
      "loss": 2.3719,
      "step": 32488
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6656482815742493,
      "learning_rate": 0.0005710416722853809,
      "loss": 2.2339,
      "step": 32489
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6729296445846558,
      "learning_rate": 0.0005710399188546078,
      "loss": 2.2274,
      "step": 32490
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6707813739776611,
      "learning_rate": 0.0005710381653734433,
      "loss": 2.3723,
      "step": 32491
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6672598123550415,
      "learning_rate": 0.0005710364118418877,
      "loss": 2.4537,
      "step": 32492
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.671726405620575,
      "learning_rate": 0.0005710346582599412,
      "loss": 2.1891,
      "step": 32493
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6288390159606934,
      "learning_rate": 0.0005710329046276043,
      "loss": 2.1136,
      "step": 32494
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.5697085857391357,
      "learning_rate": 0.0005710311509448773,
      "loss": 2.2207,
      "step": 32495
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6921485662460327,
      "learning_rate": 0.0005710293972117606,
      "loss": 2.3109,
      "step": 32496
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6658703088760376,
      "learning_rate": 0.0005710276434282542,
      "loss": 2.0884,
      "step": 32497
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.7175642251968384,
      "learning_rate": 0.0005710258895943589,
      "loss": 2.2156,
      "step": 32498
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.7647016644477844,
      "learning_rate": 0.0005710241357100745,
      "loss": 2.1992,
      "step": 32499
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.690100908279419,
      "learning_rate": 0.0005710223817754018,
      "loss": 2.1303,
      "step": 32500
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6372561454772949,
      "learning_rate": 0.0005710206277903408,
      "loss": 2.2658,
      "step": 32501
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.7474521994590759,
      "learning_rate": 0.000571018873754892,
      "loss": 2.4834,
      "step": 32502
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6798719167709351,
      "learning_rate": 0.0005710171196690556,
      "loss": 2.3526,
      "step": 32503
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.616588294506073,
      "learning_rate": 0.0005710153655328322,
      "loss": 2.2604,
      "step": 32504
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6411285400390625,
      "learning_rate": 0.0005710136113462217,
      "loss": 2.6553,
      "step": 32505
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.632841169834137,
      "learning_rate": 0.0005710118571092248,
      "loss": 2.2341,
      "step": 32506
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.7211920022964478,
      "learning_rate": 0.0005710101028218416,
      "loss": 2.1114,
      "step": 32507
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6111659407615662,
      "learning_rate": 0.0005710083484840726,
      "loss": 2.338,
      "step": 32508
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.7054287791252136,
      "learning_rate": 0.000571006594095918,
      "loss": 2.2073,
      "step": 32509
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6553893089294434,
      "learning_rate": 0.0005710048396573781,
      "loss": 2.3381,
      "step": 32510
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6376656293869019,
      "learning_rate": 0.0005710030851684533,
      "loss": 2.2036,
      "step": 32511
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6868120431900024,
      "learning_rate": 0.0005710013306291439,
      "loss": 2.3357,
      "step": 32512
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6086673736572266,
      "learning_rate": 0.0005709995760394502,
      "loss": 1.9408,
      "step": 32513
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6477493643760681,
      "learning_rate": 0.0005709978213993727,
      "loss": 2.3239,
      "step": 32514
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6947087049484253,
      "learning_rate": 0.0005709960667089116,
      "loss": 2.4487,
      "step": 32515
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.7742170691490173,
      "learning_rate": 0.0005709943119680671,
      "loss": 2.3322,
      "step": 32516
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6049395799636841,
      "learning_rate": 0.0005709925571768396,
      "loss": 2.1881,
      "step": 32517
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.8632681965827942,
      "learning_rate": 0.0005709908023352296,
      "loss": 2.1429,
      "step": 32518
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6856908798217773,
      "learning_rate": 0.0005709890474432372,
      "loss": 2.2067,
      "step": 32519
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6778435111045837,
      "learning_rate": 0.0005709872925008628,
      "loss": 2.0853,
      "step": 32520
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6817034482955933,
      "learning_rate": 0.0005709855375081068,
      "loss": 2.1553,
      "step": 32521
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.8526940941810608,
      "learning_rate": 0.0005709837824649695,
      "loss": 2.1986,
      "step": 32522
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.7236844301223755,
      "learning_rate": 0.0005709820273714512,
      "loss": 2.4264,
      "step": 32523
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.732198178768158,
      "learning_rate": 0.0005709802722275521,
      "loss": 2.1919,
      "step": 32524
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.7112894058227539,
      "learning_rate": 0.0005709785170332728,
      "loss": 2.1511,
      "step": 32525
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.7632378339767456,
      "learning_rate": 0.0005709767617886135,
      "loss": 2.2777,
      "step": 32526
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6730270385742188,
      "learning_rate": 0.0005709750064935744,
      "loss": 2.4264,
      "step": 32527
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.7968250513076782,
      "learning_rate": 0.000570973251148156,
      "loss": 2.436,
      "step": 32528
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.8599715828895569,
      "learning_rate": 0.0005709714957523585,
      "loss": 2.0976,
      "step": 32529
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.7097349762916565,
      "learning_rate": 0.0005709697403061824,
      "loss": 2.4114,
      "step": 32530
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.814918577671051,
      "learning_rate": 0.0005709679848096278,
      "loss": 2.2596,
      "step": 32531
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.7387372851371765,
      "learning_rate": 0.0005709662292626951,
      "loss": 2.3373,
      "step": 32532
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.7153056263923645,
      "learning_rate": 0.0005709644736653848,
      "loss": 2.1809,
      "step": 32533
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.7008968591690063,
      "learning_rate": 0.0005709627180176969,
      "loss": 2.4399,
      "step": 32534
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6560304760932922,
      "learning_rate": 0.0005709609623196321,
      "loss": 2.1108,
      "step": 32535
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.66291743516922,
      "learning_rate": 0.0005709592065711904,
      "loss": 2.247,
      "step": 32536
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6671019792556763,
      "learning_rate": 0.0005709574507723723,
      "loss": 2.5038,
      "step": 32537
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6423349380493164,
      "learning_rate": 0.0005709556949231783,
      "loss": 2.2928,
      "step": 32538
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6032874584197998,
      "learning_rate": 0.0005709539390236082,
      "loss": 2.1866,
      "step": 32539
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.7225180864334106,
      "learning_rate": 0.0005709521830736629,
      "loss": 2.2242,
      "step": 32540
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.7095532417297363,
      "learning_rate": 0.0005709504270733423,
      "loss": 2.3618,
      "step": 32541
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6324561238288879,
      "learning_rate": 0.000570948671022647,
      "loss": 2.4625,
      "step": 32542
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.7188125848770142,
      "learning_rate": 0.0005709469149215772,
      "loss": 2.2691,
      "step": 32543
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.684299111366272,
      "learning_rate": 0.0005709451587701333,
      "loss": 2.3886,
      "step": 32544
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.7585755586624146,
      "learning_rate": 0.0005709434025683156,
      "loss": 2.2523,
      "step": 32545
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6658661365509033,
      "learning_rate": 0.0005709416463161243,
      "loss": 2.2152,
      "step": 32546
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6366802453994751,
      "learning_rate": 0.0005709398900135598,
      "loss": 2.1649,
      "step": 32547
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6571148037910461,
      "learning_rate": 0.0005709381336606226,
      "loss": 2.4869,
      "step": 32548
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6223183274269104,
      "learning_rate": 0.0005709363772573128,
      "loss": 2.5099,
      "step": 32549
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6290537118911743,
      "learning_rate": 0.0005709346208036308,
      "loss": 2.3339,
      "step": 32550
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6622973680496216,
      "learning_rate": 0.000570932864299577,
      "loss": 2.184,
      "step": 32551
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.7393441796302795,
      "learning_rate": 0.0005709311077451517,
      "loss": 2.3523,
      "step": 32552
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.7258162498474121,
      "learning_rate": 0.0005709293511403552,
      "loss": 2.3581,
      "step": 32553
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6347821950912476,
      "learning_rate": 0.0005709275944851877,
      "loss": 2.2736,
      "step": 32554
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.7547203898429871,
      "learning_rate": 0.0005709258377796496,
      "loss": 2.3072,
      "step": 32555
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.652392566204071,
      "learning_rate": 0.0005709240810237414,
      "loss": 2.0376,
      "step": 32556
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6679765582084656,
      "learning_rate": 0.0005709223242174634,
      "loss": 2.3191,
      "step": 32557
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6454550623893738,
      "learning_rate": 0.0005709205673608156,
      "loss": 2.4178,
      "step": 32558
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6560977697372437,
      "learning_rate": 0.0005709188104537987,
      "loss": 2.2056,
      "step": 32559
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6444557309150696,
      "learning_rate": 0.0005709170534964128,
      "loss": 2.3367,
      "step": 32560
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.7085298895835876,
      "learning_rate": 0.0005709152964886583,
      "loss": 2.4914,
      "step": 32561
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.7333923578262329,
      "learning_rate": 0.0005709135394305357,
      "loss": 2.2906,
      "step": 32562
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.7493019700050354,
      "learning_rate": 0.0005709117823220449,
      "loss": 2.2771,
      "step": 32563
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.7304552793502808,
      "learning_rate": 0.0005709100251631867,
      "loss": 2.2306,
      "step": 32564
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.7719317078590393,
      "learning_rate": 0.0005709082679539612,
      "loss": 2.4362,
      "step": 32565
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.8043520450592041,
      "learning_rate": 0.0005709065106943686,
      "loss": 2.2676,
      "step": 32566
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6526541113853455,
      "learning_rate": 0.0005709047533844093,
      "loss": 2.3158,
      "step": 32567
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.7209911942481995,
      "learning_rate": 0.0005709029960240839,
      "loss": 2.4191,
      "step": 32568
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.9511565566062927,
      "learning_rate": 0.0005709012386133925,
      "loss": 2.3403,
      "step": 32569
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6920099854469299,
      "learning_rate": 0.0005708994811523354,
      "loss": 2.1936,
      "step": 32570
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.7927360534667969,
      "learning_rate": 0.000570897723640913,
      "loss": 2.3165,
      "step": 32571
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.2814134359359741,
      "learning_rate": 0.0005708959660791255,
      "loss": 2.1815,
      "step": 32572
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6783475875854492,
      "learning_rate": 0.0005708942084669734,
      "loss": 2.3525,
      "step": 32573
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.7775275707244873,
      "learning_rate": 0.000570892450804457,
      "loss": 2.5114,
      "step": 32574
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.8751131296157837,
      "learning_rate": 0.0005708906930915765,
      "loss": 2.4393,
      "step": 32575
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6047451496124268,
      "learning_rate": 0.0005708889353283323,
      "loss": 2.223,
      "step": 32576
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.7141001224517822,
      "learning_rate": 0.0005708871775147248,
      "loss": 2.2562,
      "step": 32577
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.8529679775238037,
      "learning_rate": 0.0005708854196507542,
      "loss": 2.2727,
      "step": 32578
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.642143189907074,
      "learning_rate": 0.0005708836617364209,
      "loss": 2.2532,
      "step": 32579
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6455323100090027,
      "learning_rate": 0.0005708819037717252,
      "loss": 2.165,
      "step": 32580
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.690305233001709,
      "learning_rate": 0.0005708801457566675,
      "loss": 2.2556,
      "step": 32581
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.8538049459457397,
      "learning_rate": 0.0005708783876912481,
      "loss": 2.2855,
      "step": 32582
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.7761378884315491,
      "learning_rate": 0.0005708766295754672,
      "loss": 2.2922,
      "step": 32583
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.72411048412323,
      "learning_rate": 0.0005708748714093252,
      "loss": 2.3006,
      "step": 32584
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.9724639654159546,
      "learning_rate": 0.0005708731131928225,
      "loss": 2.3491,
      "step": 32585
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.8988666534423828,
      "learning_rate": 0.0005708713549259594,
      "loss": 2.5265,
      "step": 32586
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6441743969917297,
      "learning_rate": 0.0005708695966087362,
      "loss": 2.0883,
      "step": 32587
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.8999764323234558,
      "learning_rate": 0.0005708678382411532,
      "loss": 2.264,
      "step": 32588
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.8485264182090759,
      "learning_rate": 0.0005708660798232107,
      "loss": 2.349,
      "step": 32589
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6579653024673462,
      "learning_rate": 0.0005708643213549092,
      "loss": 2.1671,
      "step": 32590
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.7415853142738342,
      "learning_rate": 0.0005708625628362489,
      "loss": 2.231,
      "step": 32591
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.1158795356750488,
      "learning_rate": 0.0005708608042672299,
      "loss": 2.2389,
      "step": 32592
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6606153249740601,
      "learning_rate": 0.0005708590456478531,
      "loss": 2.0841,
      "step": 32593
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6745033860206604,
      "learning_rate": 0.0005708572869781184,
      "loss": 2.4516,
      "step": 32594
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.9070999622344971,
      "learning_rate": 0.0005708555282580261,
      "loss": 2.4677,
      "step": 32595
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.755001962184906,
      "learning_rate": 0.0005708537694875767,
      "loss": 2.2841,
      "step": 32596
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6573953032493591,
      "learning_rate": 0.0005708520106667705,
      "loss": 2.2021,
      "step": 32597
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.646064043045044,
      "learning_rate": 0.0005708502517956077,
      "loss": 2.219,
      "step": 32598
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6750495433807373,
      "learning_rate": 0.0005708484928740889,
      "loss": 2.3864,
      "step": 32599
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6959015727043152,
      "learning_rate": 0.000570846733902214,
      "loss": 2.5264,
      "step": 32600
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6856193542480469,
      "learning_rate": 0.0005708449748799837,
      "loss": 2.3119,
      "step": 32601
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.5788787603378296,
      "learning_rate": 0.0005708432158073983,
      "loss": 2.1719,
      "step": 32602
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6190058588981628,
      "learning_rate": 0.000570841456684458,
      "loss": 2.5562,
      "step": 32603
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.7145385146141052,
      "learning_rate": 0.0005708396975111632,
      "loss": 2.5413,
      "step": 32604
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.709729015827179,
      "learning_rate": 0.0005708379382875141,
      "loss": 2.2675,
      "step": 32605
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6830922961235046,
      "learning_rate": 0.0005708361790135112,
      "loss": 2.2619,
      "step": 32606
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6326397061347961,
      "learning_rate": 0.0005708344196891546,
      "loss": 2.2873,
      "step": 32607
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6660932898521423,
      "learning_rate": 0.0005708326603144449,
      "loss": 2.2139,
      "step": 32608
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6140298843383789,
      "learning_rate": 0.0005708309008893822,
      "loss": 2.3823,
      "step": 32609
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.737425684928894,
      "learning_rate": 0.0005708291414139669,
      "loss": 2.2781,
      "step": 32610
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.7181492447853088,
      "learning_rate": 0.0005708273818881994,
      "loss": 2.4143,
      "step": 32611
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6731500029563904,
      "learning_rate": 0.0005708256223120801,
      "loss": 2.2557,
      "step": 32612
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6331713795661926,
      "learning_rate": 0.0005708238626856091,
      "loss": 2.0723,
      "step": 32613
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.7399502992630005,
      "learning_rate": 0.0005708221030087868,
      "loss": 2.3509,
      "step": 32614
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6494930982589722,
      "learning_rate": 0.0005708203432816136,
      "loss": 2.4558,
      "step": 32615
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.7433727383613586,
      "learning_rate": 0.0005708185835040898,
      "loss": 2.2729,
      "step": 32616
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6433513164520264,
      "learning_rate": 0.0005708168236762157,
      "loss": 2.2562,
      "step": 32617
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6259446144104004,
      "learning_rate": 0.0005708150637979916,
      "loss": 2.4809,
      "step": 32618
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6870527267456055,
      "learning_rate": 0.000570813303869418,
      "loss": 2.4238,
      "step": 32619
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6695189476013184,
      "learning_rate": 0.000570811543890495,
      "loss": 2.2429,
      "step": 32620
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6629283428192139,
      "learning_rate": 0.0005708097838612229,
      "loss": 2.3611,
      "step": 32621
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6762330532073975,
      "learning_rate": 0.0005708080237816024,
      "loss": 2.275,
      "step": 32622
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.636203408241272,
      "learning_rate": 0.0005708062636516334,
      "loss": 2.1814,
      "step": 32623
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6958888173103333,
      "learning_rate": 0.0005708045034713163,
      "loss": 2.2449,
      "step": 32624
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6696907877922058,
      "learning_rate": 0.0005708027432406518,
      "loss": 2.3798,
      "step": 32625
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6721940040588379,
      "learning_rate": 0.0005708009829596398,
      "loss": 2.3469,
      "step": 32626
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6368098855018616,
      "learning_rate": 0.0005707992226282808,
      "loss": 2.4039,
      "step": 32627
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6534785628318787,
      "learning_rate": 0.0005707974622465751,
      "loss": 2.392,
      "step": 32628
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.633111834526062,
      "learning_rate": 0.0005707957018145231,
      "loss": 2.2042,
      "step": 32629
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6573236584663391,
      "learning_rate": 0.0005707939413321251,
      "loss": 2.2003,
      "step": 32630
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6697527766227722,
      "learning_rate": 0.0005707921807993812,
      "loss": 2.165,
      "step": 32631
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6845117807388306,
      "learning_rate": 0.0005707904202162922,
      "loss": 2.3906,
      "step": 32632
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6965922713279724,
      "learning_rate": 0.000570788659582858,
      "loss": 2.3875,
      "step": 32633
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.660275936126709,
      "learning_rate": 0.0005707868988990791,
      "loss": 2.3264,
      "step": 32634
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.7295790314674377,
      "learning_rate": 0.0005707851381649557,
      "loss": 2.3478,
      "step": 32635
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.62840336561203,
      "learning_rate": 0.0005707833773804883,
      "loss": 2.3609,
      "step": 32636
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.9186174869537354,
      "learning_rate": 0.0005707816165456772,
      "loss": 2.1833,
      "step": 32637
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.7247406840324402,
      "learning_rate": 0.0005707798556605227,
      "loss": 2.0981,
      "step": 32638
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6520365476608276,
      "learning_rate": 0.000570778094725025,
      "loss": 2.4722,
      "step": 32639
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.7407788038253784,
      "learning_rate": 0.0005707763337391847,
      "loss": 2.4813,
      "step": 32640
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6994321942329407,
      "learning_rate": 0.0005707745727030018,
      "loss": 2.3913,
      "step": 32641
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6873042583465576,
      "learning_rate": 0.0005707728116164769,
      "loss": 2.4528,
      "step": 32642
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6310214996337891,
      "learning_rate": 0.0005707710504796102,
      "loss": 2.3169,
      "step": 32643
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6511291861534119,
      "learning_rate": 0.000570769289292402,
      "loss": 2.2294,
      "step": 32644
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6426193118095398,
      "learning_rate": 0.0005707675280548528,
      "loss": 2.077,
      "step": 32645
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.7568691372871399,
      "learning_rate": 0.0005707657667669627,
      "loss": 2.4659,
      "step": 32646
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6299399733543396,
      "learning_rate": 0.0005707640054287322,
      "loss": 2.2766,
      "step": 32647
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6216219663619995,
      "learning_rate": 0.0005707622440401615,
      "loss": 2.1849,
      "step": 32648
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.7020405530929565,
      "learning_rate": 0.0005707604826012511,
      "loss": 2.3466,
      "step": 32649
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6627683639526367,
      "learning_rate": 0.000570758721112001,
      "loss": 2.3529,
      "step": 32650
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6648905873298645,
      "learning_rate": 0.0005707569595724119,
      "loss": 2.1173,
      "step": 32651
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6560262441635132,
      "learning_rate": 0.000570755197982484,
      "loss": 2.3195,
      "step": 32652
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.701019823551178,
      "learning_rate": 0.0005707534363422176,
      "loss": 2.5504,
      "step": 32653
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6590316295623779,
      "learning_rate": 0.0005707516746516129,
      "loss": 2.103,
      "step": 32654
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.8254535794258118,
      "learning_rate": 0.0005707499129106704,
      "loss": 2.3769,
      "step": 32655
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.7140050530433655,
      "learning_rate": 0.0005707481511193904,
      "loss": 2.283,
      "step": 32656
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6039051413536072,
      "learning_rate": 0.0005707463892777732,
      "loss": 2.3607,
      "step": 32657
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6342026591300964,
      "learning_rate": 0.0005707446273858192,
      "loss": 2.2665,
      "step": 32658
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6356203556060791,
      "learning_rate": 0.0005707428654435285,
      "loss": 2.3344,
      "step": 32659
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.7079406976699829,
      "learning_rate": 0.0005707411034509019,
      "loss": 2.3089,
      "step": 32660
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6202808022499084,
      "learning_rate": 0.000570739341407939,
      "loss": 2.3448,
      "step": 32661
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6744188666343689,
      "learning_rate": 0.0005707375793146408,
      "loss": 2.5052,
      "step": 32662
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6295250654220581,
      "learning_rate": 0.0005707358171710074,
      "loss": 2.1327,
      "step": 32663
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.7464444041252136,
      "learning_rate": 0.000570734054977039,
      "loss": 2.1822,
      "step": 32664
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.7461233735084534,
      "learning_rate": 0.0005707322927327361,
      "loss": 2.3305,
      "step": 32665
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6529682278633118,
      "learning_rate": 0.0005707305304380989,
      "loss": 2.1003,
      "step": 32666
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.643238365650177,
      "learning_rate": 0.0005707287680931279,
      "loss": 2.1643,
      "step": 32667
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.7075506448745728,
      "learning_rate": 0.0005707270056978232,
      "loss": 2.2768,
      "step": 32668
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.7341639995574951,
      "learning_rate": 0.0005707252432521852,
      "loss": 2.2113,
      "step": 32669
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6489492654800415,
      "learning_rate": 0.0005707234807562143,
      "loss": 2.2423,
      "step": 32670
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6673746705055237,
      "learning_rate": 0.0005707217182099109,
      "loss": 2.2692,
      "step": 32671
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.7117733359336853,
      "learning_rate": 0.0005707199556132752,
      "loss": 2.2868,
      "step": 32672
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6404719948768616,
      "learning_rate": 0.0005707181929663074,
      "loss": 2.3973,
      "step": 32673
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6414111852645874,
      "learning_rate": 0.0005707164302690081,
      "loss": 2.3072,
      "step": 32674
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6450518369674683,
      "learning_rate": 0.0005707146675213774,
      "loss": 2.0781,
      "step": 32675
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.7186715602874756,
      "learning_rate": 0.0005707129047234159,
      "loss": 2.4009,
      "step": 32676
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6480646729469299,
      "learning_rate": 0.0005707111418751236,
      "loss": 2.3825,
      "step": 32677
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.7009403705596924,
      "learning_rate": 0.0005707093789765011,
      "loss": 2.0965,
      "step": 32678
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6469658017158508,
      "learning_rate": 0.0005707076160275486,
      "loss": 2.5657,
      "step": 32679
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6672420501708984,
      "learning_rate": 0.0005707058530282664,
      "loss": 2.3475,
      "step": 32680
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.7553593516349792,
      "learning_rate": 0.0005707040899786548,
      "loss": 2.0817,
      "step": 32681
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.665262758731842,
      "learning_rate": 0.0005707023268787144,
      "loss": 2.2823,
      "step": 32682
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6715499758720398,
      "learning_rate": 0.0005707005637284452,
      "loss": 2.2401,
      "step": 32683
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6027911901473999,
      "learning_rate": 0.0005706988005278477,
      "loss": 2.3303,
      "step": 32684
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6218065619468689,
      "learning_rate": 0.0005706970372769221,
      "loss": 2.3643,
      "step": 32685
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6311399936676025,
      "learning_rate": 0.0005706952739756689,
      "loss": 2.1483,
      "step": 32686
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.635505199432373,
      "learning_rate": 0.0005706935106240883,
      "loss": 2.433,
      "step": 32687
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6837604641914368,
      "learning_rate": 0.0005706917472221806,
      "loss": 2.2386,
      "step": 32688
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6375572681427002,
      "learning_rate": 0.0005706899837699463,
      "loss": 2.2598,
      "step": 32689
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.5948781371116638,
      "learning_rate": 0.0005706882202673856,
      "loss": 2.0667,
      "step": 32690
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.7294152975082397,
      "learning_rate": 0.0005706864567144988,
      "loss": 2.458,
      "step": 32691
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6430273652076721,
      "learning_rate": 0.0005706846931112864,
      "loss": 2.1931,
      "step": 32692
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6551876068115234,
      "learning_rate": 0.0005706829294577485,
      "loss": 2.3841,
      "step": 32693
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.650840699672699,
      "learning_rate": 0.0005706811657538855,
      "loss": 2.4629,
      "step": 32694
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6732448935508728,
      "learning_rate": 0.0005706794019996979,
      "loss": 2.5735,
      "step": 32695
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6627185940742493,
      "learning_rate": 0.0005706776381951857,
      "loss": 2.2753,
      "step": 32696
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6635132431983948,
      "learning_rate": 0.0005706758743403496,
      "loss": 2.4548,
      "step": 32697
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.011592984199524,
      "learning_rate": 0.0005706741104351897,
      "loss": 2.11,
      "step": 32698
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6342777013778687,
      "learning_rate": 0.0005706723464797063,
      "loss": 2.2151,
      "step": 32699
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6443994045257568,
      "learning_rate": 0.0005706705824738999,
      "loss": 2.3205,
      "step": 32700
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6577926874160767,
      "learning_rate": 0.0005706688184177707,
      "loss": 2.4167,
      "step": 32701
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6428296566009521,
      "learning_rate": 0.000570667054311319,
      "loss": 2.2586,
      "step": 32702
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6931096911430359,
      "learning_rate": 0.0005706652901545452,
      "loss": 2.2859,
      "step": 32703
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6942499876022339,
      "learning_rate": 0.0005706635259474496,
      "loss": 2.3649,
      "step": 32704
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6549360752105713,
      "learning_rate": 0.0005706617616900326,
      "loss": 2.0968,
      "step": 32705
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.7489290833473206,
      "learning_rate": 0.0005706599973822944,
      "loss": 2.4703,
      "step": 32706
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.7060158252716064,
      "learning_rate": 0.0005706582330242355,
      "loss": 2.3223,
      "step": 32707
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6104739904403687,
      "learning_rate": 0.0005706564686158561,
      "loss": 2.0816,
      "step": 32708
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6600925326347351,
      "learning_rate": 0.0005706547041571565,
      "loss": 2.0932,
      "step": 32709
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6350250840187073,
      "learning_rate": 0.0005706529396481371,
      "loss": 2.2545,
      "step": 32710
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6378743648529053,
      "learning_rate": 0.0005706511750887982,
      "loss": 2.2085,
      "step": 32711
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6494932174682617,
      "learning_rate": 0.0005706494104791402,
      "loss": 2.0982,
      "step": 32712
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.7003688812255859,
      "learning_rate": 0.0005706476458191632,
      "loss": 2.225,
      "step": 32713
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.7256497144699097,
      "learning_rate": 0.0005706458811088679,
      "loss": 2.2741,
      "step": 32714
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6456936001777649,
      "learning_rate": 0.0005706441163482542,
      "loss": 2.4252,
      "step": 32715
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6101385354995728,
      "learning_rate": 0.0005706423515373228,
      "loss": 2.1556,
      "step": 32716
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6304404735565186,
      "learning_rate": 0.0005706405866760739,
      "loss": 2.1602,
      "step": 32717
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6200286746025085,
      "learning_rate": 0.0005706388217645077,
      "loss": 2.2475,
      "step": 32718
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6424967050552368,
      "learning_rate": 0.0005706370568026248,
      "loss": 2.3249,
      "step": 32719
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6561921238899231,
      "learning_rate": 0.0005706352917904253,
      "loss": 2.1201,
      "step": 32720
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.600313127040863,
      "learning_rate": 0.0005706335267279095,
      "loss": 2.1884,
      "step": 32721
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.5817702412605286,
      "learning_rate": 0.0005706317616150779,
      "loss": 2.1894,
      "step": 32722
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6782060265541077,
      "learning_rate": 0.0005706299964519306,
      "loss": 2.3533,
      "step": 32723
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6002444624900818,
      "learning_rate": 0.0005706282312384683,
      "loss": 2.238,
      "step": 32724
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.5970260500907898,
      "learning_rate": 0.000570626465974691,
      "loss": 2.1982,
      "step": 32725
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6026697754859924,
      "learning_rate": 0.0005706247006605991,
      "loss": 1.9075,
      "step": 32726
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6309201717376709,
      "learning_rate": 0.0005706229352961929,
      "loss": 2.4166,
      "step": 32727
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6008369326591492,
      "learning_rate": 0.0005706211698814729,
      "loss": 1.9697,
      "step": 32728
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6814479827880859,
      "learning_rate": 0.0005706194044164394,
      "loss": 2.154,
      "step": 32729
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6695585250854492,
      "learning_rate": 0.0005706176389010924,
      "loss": 2.3377,
      "step": 32730
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.7116018533706665,
      "learning_rate": 0.0005706158733354327,
      "loss": 2.3247,
      "step": 32731
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6963883638381958,
      "learning_rate": 0.0005706141077194603,
      "loss": 2.2816,
      "step": 32732
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6345521807670593,
      "learning_rate": 0.0005706123420531756,
      "loss": 2.2209,
      "step": 32733
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6130853891372681,
      "learning_rate": 0.0005706105763365791,
      "loss": 2.246,
      "step": 32734
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.7223308682441711,
      "learning_rate": 0.0005706088105696707,
      "loss": 2.4303,
      "step": 32735
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6282598972320557,
      "learning_rate": 0.0005706070447524513,
      "loss": 2.162,
      "step": 32736
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6527354121208191,
      "learning_rate": 0.0005706052788849209,
      "loss": 2.3123,
      "step": 32737
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.5781487226486206,
      "learning_rate": 0.0005706035129670798,
      "loss": 2.2409,
      "step": 32738
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.7187482714653015,
      "learning_rate": 0.0005706017469989283,
      "loss": 2.3982,
      "step": 32739
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6208854913711548,
      "learning_rate": 0.000570599980980467,
      "loss": 2.1375,
      "step": 32740
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6204132437705994,
      "learning_rate": 0.000570598214911696,
      "loss": 2.2184,
      "step": 32741
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6768783926963806,
      "learning_rate": 0.0005705964487926157,
      "loss": 2.4246,
      "step": 32742
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6713395714759827,
      "learning_rate": 0.0005705946826232263,
      "loss": 2.2819,
      "step": 32743
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6789229512214661,
      "learning_rate": 0.0005705929164035283,
      "loss": 2.3712,
      "step": 32744
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.61814945936203,
      "learning_rate": 0.0005705911501335221,
      "loss": 2.2796,
      "step": 32745
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6151573061943054,
      "learning_rate": 0.0005705893838132077,
      "loss": 2.2992,
      "step": 32746
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6427228450775146,
      "learning_rate": 0.0005705876174425858,
      "loss": 2.2478,
      "step": 32747
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6440072655677795,
      "learning_rate": 0.0005705858510216564,
      "loss": 2.4251,
      "step": 32748
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6560040712356567,
      "learning_rate": 0.0005705840845504201,
      "loss": 2.4487,
      "step": 32749
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6288289427757263,
      "learning_rate": 0.0005705823180288771,
      "loss": 2.188,
      "step": 32750
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6132702827453613,
      "learning_rate": 0.0005705805514570276,
      "loss": 2.3356,
      "step": 32751
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6051550507545471,
      "learning_rate": 0.0005705787848348723,
      "loss": 2.19,
      "step": 32752
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6953943371772766,
      "learning_rate": 0.0005705770181624112,
      "loss": 2.1724,
      "step": 32753
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.630767822265625,
      "learning_rate": 0.0005705752514396446,
      "loss": 2.2988,
      "step": 32754
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6546463966369629,
      "learning_rate": 0.000570573484666573,
      "loss": 2.3151,
      "step": 32755
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.7455509305000305,
      "learning_rate": 0.0005705717178431969,
      "loss": 1.9937,
      "step": 32756
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.715417206287384,
      "learning_rate": 0.0005705699509695161,
      "loss": 2.2907,
      "step": 32757
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.7809932827949524,
      "learning_rate": 0.0005705681840455314,
      "loss": 2.1091,
      "step": 32758
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.8515731692314148,
      "learning_rate": 0.0005705664170712429,
      "loss": 2.1696,
      "step": 32759
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6925246119499207,
      "learning_rate": 0.000570564650046651,
      "loss": 2.2369,
      "step": 32760
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.7208357453346252,
      "learning_rate": 0.0005705628829717561,
      "loss": 2.3884,
      "step": 32761
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.7752251029014587,
      "learning_rate": 0.0005705611158465585,
      "loss": 2.2723,
      "step": 32762
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6846054792404175,
      "learning_rate": 0.0005705593486710584,
      "loss": 2.3569,
      "step": 32763
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6785243153572083,
      "learning_rate": 0.0005705575814452562,
      "loss": 2.4842,
      "step": 32764
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6858835220336914,
      "learning_rate": 0.0005705558141691522,
      "loss": 2.262,
      "step": 32765
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6793990135192871,
      "learning_rate": 0.0005705540468427468,
      "loss": 2.3477,
      "step": 32766
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.7187842726707458,
      "learning_rate": 0.0005705522794660404,
      "loss": 2.3626,
      "step": 32767
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.7163474559783936,
      "learning_rate": 0.0005705505120390331,
      "loss": 2.4745,
      "step": 32768
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.7004907131195068,
      "learning_rate": 0.0005705487445617254,
      "loss": 2.2997,
      "step": 32769
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.670173704624176,
      "learning_rate": 0.0005705469770341175,
      "loss": 2.2961,
      "step": 32770
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6447883248329163,
      "learning_rate": 0.00057054520945621,
      "loss": 2.1488,
      "step": 32771
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.626898467540741,
      "learning_rate": 0.0005705434418280029,
      "loss": 2.2804,
      "step": 32772
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6731022000312805,
      "learning_rate": 0.0005705416741494967,
      "loss": 2.406,
      "step": 32773
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6863632798194885,
      "learning_rate": 0.0005705399064206917,
      "loss": 2.2022,
      "step": 32774
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6850524544715881,
      "learning_rate": 0.0005705381386415881,
      "loss": 2.5519,
      "step": 32775
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.671094536781311,
      "learning_rate": 0.0005705363708121865,
      "loss": 2.1069,
      "step": 32776
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6557014584541321,
      "learning_rate": 0.0005705346029324871,
      "loss": 2.2739,
      "step": 32777
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.7540039420127869,
      "learning_rate": 0.00057053283500249,
      "loss": 2.1272,
      "step": 32778
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6847739219665527,
      "learning_rate": 0.000570531067022196,
      "loss": 2.099,
      "step": 32779
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.7826758623123169,
      "learning_rate": 0.000570529298991605,
      "loss": 2.3637,
      "step": 32780
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.671716570854187,
      "learning_rate": 0.0005705275309107176,
      "loss": 2.4392,
      "step": 32781
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6447989344596863,
      "learning_rate": 0.000570525762779534,
      "loss": 2.4733,
      "step": 32782
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.7316668629646301,
      "learning_rate": 0.0005705239945980545,
      "loss": 2.4964,
      "step": 32783
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6686208248138428,
      "learning_rate": 0.0005705222263662795,
      "loss": 2.2946,
      "step": 32784
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6952928900718689,
      "learning_rate": 0.0005705204580842094,
      "loss": 2.2056,
      "step": 32785
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.7282442450523376,
      "learning_rate": 0.0005705186897518443,
      "loss": 2.0599,
      "step": 32786
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6933706998825073,
      "learning_rate": 0.0005705169213691848,
      "loss": 2.3546,
      "step": 32787
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6434769034385681,
      "learning_rate": 0.000570515152936231,
      "loss": 2.3702,
      "step": 32788
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6107842922210693,
      "learning_rate": 0.0005705133844529834,
      "loss": 2.2418,
      "step": 32789
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6698116064071655,
      "learning_rate": 0.0005705116159194422,
      "loss": 2.424,
      "step": 32790
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.5872765779495239,
      "learning_rate": 0.0005705098473356078,
      "loss": 2.2034,
      "step": 32791
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6474820375442505,
      "learning_rate": 0.0005705080787014805,
      "loss": 2.2951,
      "step": 32792
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6149112582206726,
      "learning_rate": 0.0005705063100170606,
      "loss": 2.3255,
      "step": 32793
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.5989972352981567,
      "learning_rate": 0.0005705045412823486,
      "loss": 2.1538,
      "step": 32794
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.7553077340126038,
      "learning_rate": 0.0005705027724973446,
      "loss": 2.4,
      "step": 32795
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6259501576423645,
      "learning_rate": 0.000570501003662049,
      "loss": 2.1427,
      "step": 32796
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6936280727386475,
      "learning_rate": 0.0005704992347764623,
      "loss": 2.3003,
      "step": 32797
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6513113379478455,
      "learning_rate": 0.0005704974658405845,
      "loss": 2.1083,
      "step": 32798
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6952993273735046,
      "learning_rate": 0.0005704956968544163,
      "loss": 2.2641,
      "step": 32799
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.7345446348190308,
      "learning_rate": 0.0005704939278179577,
      "loss": 2.1468,
      "step": 32800
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.7744765281677246,
      "learning_rate": 0.0005704921587312093,
      "loss": 2.2329,
      "step": 32801
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6433254480361938,
      "learning_rate": 0.0005704903895941712,
      "loss": 2.1813,
      "step": 32802
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6333087086677551,
      "learning_rate": 0.0005704886204068439,
      "loss": 2.3336,
      "step": 32803
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6502842903137207,
      "learning_rate": 0.0005704868511692276,
      "loss": 2.4991,
      "step": 32804
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.7213677167892456,
      "learning_rate": 0.0005704850818813227,
      "loss": 2.1962,
      "step": 32805
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.7150327563285828,
      "learning_rate": 0.0005704833125431295,
      "loss": 2.1248,
      "step": 32806
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.7680779695510864,
      "learning_rate": 0.0005704815431546485,
      "loss": 2.4043,
      "step": 32807
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6677519679069519,
      "learning_rate": 0.0005704797737158796,
      "loss": 2.1519,
      "step": 32808
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6212025880813599,
      "learning_rate": 0.0005704780042268237,
      "loss": 2.1854,
      "step": 32809
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6395608186721802,
      "learning_rate": 0.0005704762346874806,
      "loss": 2.1283,
      "step": 32810
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6773537993431091,
      "learning_rate": 0.000570474465097851,
      "loss": 2.2762,
      "step": 32811
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.7318385243415833,
      "learning_rate": 0.0005704726954579351,
      "loss": 2.4367,
      "step": 32812
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.652121901512146,
      "learning_rate": 0.0005704709257677331,
      "loss": 2.4502,
      "step": 32813
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6372647881507874,
      "learning_rate": 0.0005704691560272455,
      "loss": 2.0948,
      "step": 32814
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6065282225608826,
      "learning_rate": 0.0005704673862364726,
      "loss": 2.3247,
      "step": 32815
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.7079456448554993,
      "learning_rate": 0.0005704656163954147,
      "loss": 2.2751,
      "step": 32816
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.758211076259613,
      "learning_rate": 0.000570463846504072,
      "loss": 2.203,
      "step": 32817
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.7132178544998169,
      "learning_rate": 0.0005704620765624452,
      "loss": 2.5359,
      "step": 32818
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6642484664916992,
      "learning_rate": 0.0005704603065705343,
      "loss": 2.5026,
      "step": 32819
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6107035279273987,
      "learning_rate": 0.0005704585365283398,
      "loss": 2.3178,
      "step": 32820
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.7052861452102661,
      "learning_rate": 0.0005704567664358618,
      "loss": 2.542,
      "step": 32821
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6735411286354065,
      "learning_rate": 0.0005704549962931008,
      "loss": 2.3556,
      "step": 32822
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6463158130645752,
      "learning_rate": 0.0005704532261000573,
      "loss": 2.1779,
      "step": 32823
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6309137940406799,
      "learning_rate": 0.0005704514558567312,
      "loss": 2.4228,
      "step": 32824
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6383640766143799,
      "learning_rate": 0.0005704496855631232,
      "loss": 2.1188,
      "step": 32825
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6349663734436035,
      "learning_rate": 0.0005704479152192335,
      "loss": 2.3231,
      "step": 32826
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6443928480148315,
      "learning_rate": 0.0005704461448250624,
      "loss": 2.3312,
      "step": 32827
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6788437962532043,
      "learning_rate": 0.0005704443743806102,
      "loss": 2.1881,
      "step": 32828
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6277301907539368,
      "learning_rate": 0.0005704426038858773,
      "loss": 2.1637,
      "step": 32829
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.5777165293693542,
      "learning_rate": 0.0005704408333408642,
      "loss": 2.3477,
      "step": 32830
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6795932650566101,
      "learning_rate": 0.0005704390627455708,
      "loss": 2.5621,
      "step": 32831
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6856009364128113,
      "learning_rate": 0.0005704372920999978,
      "loss": 2.2014,
      "step": 32832
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.7522737979888916,
      "learning_rate": 0.0005704355214041453,
      "loss": 2.2967,
      "step": 32833
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.8939910531044006,
      "learning_rate": 0.0005704337506580138,
      "loss": 2.4139,
      "step": 32834
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.7103987336158752,
      "learning_rate": 0.0005704319798616037,
      "loss": 2.298,
      "step": 32835
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6860007047653198,
      "learning_rate": 0.0005704302090149151,
      "loss": 2.3626,
      "step": 32836
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.8357385396957397,
      "learning_rate": 0.0005704284381179483,
      "loss": 2.4592,
      "step": 32837
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.7828933596611023,
      "learning_rate": 0.0005704266671707038,
      "loss": 2.0943,
      "step": 32838
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6510005593299866,
      "learning_rate": 0.0005704248961731819,
      "loss": 2.3085,
      "step": 32839
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.7424818277359009,
      "learning_rate": 0.000570423125125383,
      "loss": 2.2402,
      "step": 32840
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.7981142401695251,
      "learning_rate": 0.0005704213540273072,
      "loss": 2.3015,
      "step": 32841
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6500662565231323,
      "learning_rate": 0.000570419582878955,
      "loss": 2.3953,
      "step": 32842
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6945694088935852,
      "learning_rate": 0.0005704178116803267,
      "loss": 2.3558,
      "step": 32843
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.7051419019699097,
      "learning_rate": 0.0005704160404314227,
      "loss": 2.4856,
      "step": 32844
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.671863853931427,
      "learning_rate": 0.0005704142691322432,
      "loss": 2.4072,
      "step": 32845
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6426540613174438,
      "learning_rate": 0.0005704124977827886,
      "loss": 2.1683,
      "step": 32846
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.757252037525177,
      "learning_rate": 0.0005704107263830591,
      "loss": 2.5739,
      "step": 32847
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.9034474492073059,
      "learning_rate": 0.0005704089549330553,
      "loss": 2.4534,
      "step": 32848
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.664779543876648,
      "learning_rate": 0.0005704071834327772,
      "loss": 2.2883,
      "step": 32849
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.7075361013412476,
      "learning_rate": 0.0005704054118822254,
      "loss": 2.1699,
      "step": 32850
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.7958452105522156,
      "learning_rate": 0.0005704036402814001,
      "loss": 2.2244,
      "step": 32851
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.7115005254745483,
      "learning_rate": 0.0005704018686303017,
      "loss": 2.3549,
      "step": 32852
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.7181411981582642,
      "learning_rate": 0.0005704000969289305,
      "loss": 2.3827,
      "step": 32853
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.8471544981002808,
      "learning_rate": 0.0005703983251772869,
      "loss": 2.4886,
      "step": 32854
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6318312883377075,
      "learning_rate": 0.000570396553375371,
      "loss": 2.2357,
      "step": 32855
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6495124101638794,
      "learning_rate": 0.0005703947815231832,
      "loss": 2.1474,
      "step": 32856
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6996840834617615,
      "learning_rate": 0.000570393009620724,
      "loss": 2.2524,
      "step": 32857
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.7144520878791809,
      "learning_rate": 0.0005703912376679937,
      "loss": 2.2579,
      "step": 32858
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6873116493225098,
      "learning_rate": 0.0005703894656649925,
      "loss": 2.3883,
      "step": 32859
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.5963112115859985,
      "learning_rate": 0.0005703876936117209,
      "loss": 2.1958,
      "step": 32860
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.7376095056533813,
      "learning_rate": 0.000570385921508179,
      "loss": 2.2868,
      "step": 32861
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6993283033370972,
      "learning_rate": 0.0005703841493543674,
      "loss": 2.3255,
      "step": 32862
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6273356676101685,
      "learning_rate": 0.0005703823771502861,
      "loss": 2.309,
      "step": 32863
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6465873122215271,
      "learning_rate": 0.0005703806048959356,
      "loss": 2.2984,
      "step": 32864
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.849127471446991,
      "learning_rate": 0.0005703788325913164,
      "loss": 2.2099,
      "step": 32865
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6257649064064026,
      "learning_rate": 0.0005703770602364285,
      "loss": 2.1387,
      "step": 32866
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.5832728743553162,
      "learning_rate": 0.0005703752878312725,
      "loss": 2.115,
      "step": 32867
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6571158766746521,
      "learning_rate": 0.0005703735153758487,
      "loss": 2.3803,
      "step": 32868
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6367626786231995,
      "learning_rate": 0.0005703717428701573,
      "loss": 2.1341,
      "step": 32869
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6839366555213928,
      "learning_rate": 0.0005703699703141987,
      "loss": 2.3948,
      "step": 32870
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6274393796920776,
      "learning_rate": 0.0005703681977079732,
      "loss": 2.3257,
      "step": 32871
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.8485584855079651,
      "learning_rate": 0.0005703664250514811,
      "loss": 2.0563,
      "step": 32872
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.7106054425239563,
      "learning_rate": 0.0005703646523447228,
      "loss": 2.2491,
      "step": 32873
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6312880516052246,
      "learning_rate": 0.0005703628795876987,
      "loss": 2.1055,
      "step": 32874
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.860838770866394,
      "learning_rate": 0.000570361106780409,
      "loss": 2.3084,
      "step": 32875
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6880171895027161,
      "learning_rate": 0.000570359333922854,
      "loss": 2.2707,
      "step": 32876
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.8649153709411621,
      "learning_rate": 0.0005703575610150342,
      "loss": 2.3775,
      "step": 32877
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6119951009750366,
      "learning_rate": 0.0005703557880569497,
      "loss": 2.2201,
      "step": 32878
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.670842170715332,
      "learning_rate": 0.0005703540150486011,
      "loss": 2.3311,
      "step": 32879
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.7127838730812073,
      "learning_rate": 0.0005703522419899884,
      "loss": 1.975,
      "step": 32880
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.7229475975036621,
      "learning_rate": 0.0005703504688811122,
      "loss": 2.3604,
      "step": 32881
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6342235803604126,
      "learning_rate": 0.0005703486957219728,
      "loss": 2.4218,
      "step": 32882
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.7006807923316956,
      "learning_rate": 0.0005703469225125704,
      "loss": 2.1519,
      "step": 32883
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.8169094324111938,
      "learning_rate": 0.0005703451492529054,
      "loss": 2.5096,
      "step": 32884
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6715247631072998,
      "learning_rate": 0.0005703433759429782,
      "loss": 2.1657,
      "step": 32885
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.7215819954872131,
      "learning_rate": 0.000570341602582789,
      "loss": 2.2711,
      "step": 32886
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.8632984161376953,
      "learning_rate": 0.0005703398291723383,
      "loss": 2.1924,
      "step": 32887
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.7259985208511353,
      "learning_rate": 0.0005703380557116263,
      "loss": 2.2562,
      "step": 32888
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6605538129806519,
      "learning_rate": 0.0005703362822006532,
      "loss": 2.3537,
      "step": 32889
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.7881776094436646,
      "learning_rate": 0.0005703345086394197,
      "loss": 2.2188,
      "step": 32890
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.7724890112876892,
      "learning_rate": 0.0005703327350279257,
      "loss": 2.2532,
      "step": 32891
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6364285349845886,
      "learning_rate": 0.0005703309613661719,
      "loss": 2.4821,
      "step": 32892
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.758141815662384,
      "learning_rate": 0.0005703291876541583,
      "loss": 2.3527,
      "step": 32893
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.7068024277687073,
      "learning_rate": 0.0005703274138918856,
      "loss": 2.2858,
      "step": 32894
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6842737197875977,
      "learning_rate": 0.0005703256400793539,
      "loss": 2.4144,
      "step": 32895
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.637981116771698,
      "learning_rate": 0.0005703238662165635,
      "loss": 2.1895,
      "step": 32896
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6690579056739807,
      "learning_rate": 0.0005703220923035147,
      "loss": 2.0657,
      "step": 32897
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6800231337547302,
      "learning_rate": 0.0005703203183402081,
      "loss": 2.4604,
      "step": 32898
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6444557309150696,
      "learning_rate": 0.0005703185443266437,
      "loss": 2.1062,
      "step": 32899
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6491919755935669,
      "learning_rate": 0.000570316770262822,
      "loss": 2.2579,
      "step": 32900
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6548004150390625,
      "learning_rate": 0.0005703149961487433,
      "loss": 2.3453,
      "step": 32901
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6668572425842285,
      "learning_rate": 0.0005703132219844081,
      "loss": 2.4532,
      "step": 32902
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6602540612220764,
      "learning_rate": 0.0005703114477698164,
      "loss": 2.3005,
      "step": 32903
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.8338480591773987,
      "learning_rate": 0.0005703096735049688,
      "loss": 2.3414,
      "step": 32904
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6862542033195496,
      "learning_rate": 0.0005703078991898655,
      "loss": 2.1329,
      "step": 32905
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6659103035926819,
      "learning_rate": 0.0005703061248245069,
      "loss": 2.194,
      "step": 32906
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6543338298797607,
      "learning_rate": 0.0005703043504088932,
      "loss": 2.2698,
      "step": 32907
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6641258597373962,
      "learning_rate": 0.0005703025759430249,
      "loss": 2.1927,
      "step": 32908
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.7247211933135986,
      "learning_rate": 0.0005703008014269022,
      "loss": 2.4014,
      "step": 32909
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6163074374198914,
      "learning_rate": 0.0005702990268605254,
      "loss": 2.2192,
      "step": 32910
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6818710565567017,
      "learning_rate": 0.000570297252243895,
      "loss": 2.4032,
      "step": 32911
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.7156208753585815,
      "learning_rate": 0.0005702954775770112,
      "loss": 2.2934,
      "step": 32912
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6755503416061401,
      "learning_rate": 0.0005702937028598745,
      "loss": 2.3552,
      "step": 32913
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6854671239852905,
      "learning_rate": 0.0005702919280924849,
      "loss": 2.2529,
      "step": 32914
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.7570205926895142,
      "learning_rate": 0.000570290153274843,
      "loss": 2.693,
      "step": 32915
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6809123754501343,
      "learning_rate": 0.0005702883784069492,
      "loss": 2.1507,
      "step": 32916
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6400548815727234,
      "learning_rate": 0.0005702866034888036,
      "loss": 1.9635,
      "step": 32917
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.781238317489624,
      "learning_rate": 0.0005702848285204066,
      "loss": 2.2906,
      "step": 32918
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.7033447027206421,
      "learning_rate": 0.0005702830535017586,
      "loss": 2.2389,
      "step": 32919
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6807308197021484,
      "learning_rate": 0.0005702812784328597,
      "loss": 2.4819,
      "step": 32920
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6807154417037964,
      "learning_rate": 0.0005702795033137105,
      "loss": 2.2264,
      "step": 32921
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6119442582130432,
      "learning_rate": 0.0005702777281443113,
      "loss": 2.3672,
      "step": 32922
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6423656344413757,
      "learning_rate": 0.0005702759529246623,
      "loss": 2.3676,
      "step": 32923
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6457709074020386,
      "learning_rate": 0.0005702741776547639,
      "loss": 2.2909,
      "step": 32924
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6308207511901855,
      "learning_rate": 0.0005702724023346165,
      "loss": 2.4151,
      "step": 32925
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6786571741104126,
      "learning_rate": 0.0005702706269642203,
      "loss": 2.2846,
      "step": 32926
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6442371010780334,
      "learning_rate": 0.0005702688515435757,
      "loss": 2.3787,
      "step": 32927
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6358290910720825,
      "learning_rate": 0.000570267076072683,
      "loss": 2.3415,
      "step": 32928
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6651293635368347,
      "learning_rate": 0.0005702653005515427,
      "loss": 2.2274,
      "step": 32929
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6944925785064697,
      "learning_rate": 0.0005702635249801548,
      "loss": 2.3136,
      "step": 32930
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.697149395942688,
      "learning_rate": 0.00057026174935852,
      "loss": 2.4432,
      "step": 32931
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6554835438728333,
      "learning_rate": 0.0005702599736866382,
      "loss": 2.4472,
      "step": 32932
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6092352867126465,
      "learning_rate": 0.00057025819796451,
      "loss": 2.3309,
      "step": 32933
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.766800582408905,
      "learning_rate": 0.0005702564221921358,
      "loss": 2.3317,
      "step": 32934
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6454518437385559,
      "learning_rate": 0.0005702546463695159,
      "loss": 2.2264,
      "step": 32935
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.8003473281860352,
      "learning_rate": 0.0005702528704966504,
      "loss": 2.458,
      "step": 32936
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.5800830721855164,
      "learning_rate": 0.0005702510945735399,
      "loss": 2.2019,
      "step": 32937
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6845188140869141,
      "learning_rate": 0.0005702493186001846,
      "loss": 2.1707,
      "step": 32938
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.7289955615997314,
      "learning_rate": 0.0005702475425765848,
      "loss": 2.2028,
      "step": 32939
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.7059706449508667,
      "learning_rate": 0.0005702457665027409,
      "loss": 2.3538,
      "step": 32940
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.62237149477005,
      "learning_rate": 0.0005702439903786531,
      "loss": 2.1851,
      "step": 32941
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6756473183631897,
      "learning_rate": 0.000570242214204322,
      "loss": 2.3794,
      "step": 32942
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6764270067214966,
      "learning_rate": 0.0005702404379797478,
      "loss": 2.1225,
      "step": 32943
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.667237401008606,
      "learning_rate": 0.0005702386617049307,
      "loss": 1.9768,
      "step": 32944
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6664875149726868,
      "learning_rate": 0.0005702368853798712,
      "loss": 2.0671,
      "step": 32945
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6932315230369568,
      "learning_rate": 0.0005702351090045695,
      "loss": 2.2542,
      "step": 32946
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6904445886611938,
      "learning_rate": 0.000570233332579026,
      "loss": 2.1944,
      "step": 32947
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6989412903785706,
      "learning_rate": 0.0005702315561032411,
      "loss": 2.3848,
      "step": 32948
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6344401836395264,
      "learning_rate": 0.0005702297795772149,
      "loss": 2.3849,
      "step": 32949
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6779447197914124,
      "learning_rate": 0.000570228003000948,
      "loss": 2.4026,
      "step": 32950
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6678987741470337,
      "learning_rate": 0.0005702262263744406,
      "loss": 2.045,
      "step": 32951
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6896328330039978,
      "learning_rate": 0.000570224449697693,
      "loss": 2.2015,
      "step": 32952
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.7149592041969299,
      "learning_rate": 0.0005702226729707056,
      "loss": 2.2081,
      "step": 32953
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.694947361946106,
      "learning_rate": 0.0005702208961934787,
      "loss": 2.1199,
      "step": 32954
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6733587384223938,
      "learning_rate": 0.0005702191193660127,
      "loss": 2.2755,
      "step": 32955
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6197234392166138,
      "learning_rate": 0.0005702173424883077,
      "loss": 1.9303,
      "step": 32956
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6672050356864929,
      "learning_rate": 0.0005702155655603644,
      "loss": 2.4115,
      "step": 32957
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.7013231515884399,
      "learning_rate": 0.0005702137885821828,
      "loss": 1.9771,
      "step": 32958
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.64182049036026,
      "learning_rate": 0.0005702120115537634,
      "loss": 2.4097,
      "step": 32959
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6370505094528198,
      "learning_rate": 0.0005702102344751065,
      "loss": 2.2588,
      "step": 32960
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.7337844371795654,
      "learning_rate": 0.0005702084573462123,
      "loss": 2.1965,
      "step": 32961
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6334541440010071,
      "learning_rate": 0.0005702066801670814,
      "loss": 2.1295,
      "step": 32962
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.7361952066421509,
      "learning_rate": 0.0005702049029377139,
      "loss": 2.2221,
      "step": 32963
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6497778296470642,
      "learning_rate": 0.0005702031256581102,
      "loss": 2.196,
      "step": 32964
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.677274227142334,
      "learning_rate": 0.0005702013483282706,
      "loss": 2.3868,
      "step": 32965
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6566999554634094,
      "learning_rate": 0.0005701995709481956,
      "loss": 2.2907,
      "step": 32966
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.7309216260910034,
      "learning_rate": 0.0005701977935178852,
      "loss": 2.1039,
      "step": 32967
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.7436085939407349,
      "learning_rate": 0.0005701960160373401,
      "loss": 2.1798,
      "step": 32968
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.7075580358505249,
      "learning_rate": 0.0005701942385065605,
      "loss": 2.3074,
      "step": 32969
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6447978019714355,
      "learning_rate": 0.0005701924609255466,
      "loss": 2.2244,
      "step": 32970
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6110379099845886,
      "learning_rate": 0.0005701906832942989,
      "loss": 2.4951,
      "step": 32971
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.7082169651985168,
      "learning_rate": 0.0005701889056128175,
      "loss": 2.2661,
      "step": 32972
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6698182225227356,
      "learning_rate": 0.0005701871278811029,
      "loss": 2.4497,
      "step": 32973
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6455262303352356,
      "learning_rate": 0.0005701853500991556,
      "loss": 2.1521,
      "step": 32974
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.7344611287117004,
      "learning_rate": 0.0005701835722669755,
      "loss": 2.6519,
      "step": 32975
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6908275485038757,
      "learning_rate": 0.0005701817943845633,
      "loss": 2.4145,
      "step": 32976
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.7093927264213562,
      "learning_rate": 0.0005701800164519192,
      "loss": 2.299,
      "step": 32977
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.7223813533782959,
      "learning_rate": 0.0005701782384690436,
      "loss": 2.0328,
      "step": 32978
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.7288183569908142,
      "learning_rate": 0.0005701764604359367,
      "loss": 2.3121,
      "step": 32979
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6079853773117065,
      "learning_rate": 0.0005701746823525988,
      "loss": 2.484,
      "step": 32980
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.7001233100891113,
      "learning_rate": 0.0005701729042190305,
      "loss": 2.5861,
      "step": 32981
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.7481938600540161,
      "learning_rate": 0.0005701711260352318,
      "loss": 2.1648,
      "step": 32982
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6466388702392578,
      "learning_rate": 0.0005701693478012032,
      "loss": 2.326,
      "step": 32983
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6892321705818176,
      "learning_rate": 0.0005701675695169451,
      "loss": 2.4008,
      "step": 32984
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.661065936088562,
      "learning_rate": 0.0005701657911824578,
      "loss": 2.3438,
      "step": 32985
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.7666879296302795,
      "learning_rate": 0.0005701640127977416,
      "loss": 2.2094,
      "step": 32986
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6205534934997559,
      "learning_rate": 0.0005701622343627967,
      "loss": 2.2803,
      "step": 32987
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6847100257873535,
      "learning_rate": 0.0005701604558776235,
      "loss": 2.5309,
      "step": 32988
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.8068661093711853,
      "learning_rate": 0.0005701586773422224,
      "loss": 2.4057,
      "step": 32989
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.014193058013916,
      "learning_rate": 0.0005701568987565938,
      "loss": 2.1398,
      "step": 32990
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6890790462493896,
      "learning_rate": 0.0005701551201207379,
      "loss": 2.4851,
      "step": 32991
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.8085187673568726,
      "learning_rate": 0.000570153341434655,
      "loss": 2.3903,
      "step": 32992
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.8434954881668091,
      "learning_rate": 0.0005701515626983455,
      "loss": 2.3513,
      "step": 32993
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6984544396400452,
      "learning_rate": 0.0005701497839118098,
      "loss": 2.514,
      "step": 32994
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6717032790184021,
      "learning_rate": 0.000570148005075048,
      "loss": 2.279,
      "step": 32995
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.8590960502624512,
      "learning_rate": 0.0005701462261880608,
      "loss": 2.237,
      "step": 32996
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.7142683267593384,
      "learning_rate": 0.0005701444472508482,
      "loss": 2.5512,
      "step": 32997
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6798954606056213,
      "learning_rate": 0.0005701426682634107,
      "loss": 2.4901,
      "step": 32998
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.7729719877243042,
      "learning_rate": 0.0005701408892257485,
      "loss": 2.1148,
      "step": 32999
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6546603441238403,
      "learning_rate": 0.000570139110137862,
      "loss": 2.2471,
      "step": 33000
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6302304267883301,
      "learning_rate": 0.0005701373309997517,
      "loss": 2.3235,
      "step": 33001
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6870596408843994,
      "learning_rate": 0.0005701355518114177,
      "loss": 2.3583,
      "step": 33002
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6522809863090515,
      "learning_rate": 0.0005701337725728602,
      "loss": 2.4533,
      "step": 33003
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.7420893907546997,
      "learning_rate": 0.0005701319932840799,
      "loss": 2.3126,
      "step": 33004
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6529439687728882,
      "learning_rate": 0.000570130213945077,
      "loss": 2.3075,
      "step": 33005
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6213623881340027,
      "learning_rate": 0.0005701284345558517,
      "loss": 2.4728,
      "step": 33006
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.7475330233573914,
      "learning_rate": 0.0005701266551164045,
      "loss": 2.4546,
      "step": 33007
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.5825481414794922,
      "learning_rate": 0.0005701248756267357,
      "loss": 2.2577,
      "step": 33008
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6594709157943726,
      "learning_rate": 0.0005701230960868455,
      "loss": 2.3991,
      "step": 33009
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6825723648071289,
      "learning_rate": 0.0005701213164967344,
      "loss": 2.1839,
      "step": 33010
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6542614698410034,
      "learning_rate": 0.0005701195368564024,
      "loss": 2.2459,
      "step": 33011
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6577020883560181,
      "learning_rate": 0.0005701177571658503,
      "loss": 2.387,
      "step": 33012
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6030840277671814,
      "learning_rate": 0.0005701159774250782,
      "loss": 2.1019,
      "step": 33013
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6343216300010681,
      "learning_rate": 0.0005701141976340864,
      "loss": 2.1544,
      "step": 33014
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6414927244186401,
      "learning_rate": 0.0005701124177928754,
      "loss": 2.4298,
      "step": 33015
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.7196327447891235,
      "learning_rate": 0.0005701106379014453,
      "loss": 2.1927,
      "step": 33016
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.7792220711708069,
      "learning_rate": 0.0005701088579597964,
      "loss": 2.3147,
      "step": 33017
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6314793229103088,
      "learning_rate": 0.0005701070779679292,
      "loss": 2.0216,
      "step": 33018
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6476446986198425,
      "learning_rate": 0.0005701052979258442,
      "loss": 2.2465,
      "step": 33019
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.7754208445549011,
      "learning_rate": 0.0005701035178335414,
      "loss": 2.5324,
      "step": 33020
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.7144209742546082,
      "learning_rate": 0.0005701017376910213,
      "loss": 2.0192,
      "step": 33021
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.8071452379226685,
      "learning_rate": 0.0005700999574982841,
      "loss": 2.4853,
      "step": 33022
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.689154863357544,
      "learning_rate": 0.0005700981772553303,
      "loss": 2.2272,
      "step": 33023
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6684480309486389,
      "learning_rate": 0.00057009639696216,
      "loss": 2.3925,
      "step": 33024
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6078165173530579,
      "learning_rate": 0.0005700946166187739,
      "loss": 2.4204,
      "step": 33025
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.661794900894165,
      "learning_rate": 0.000570092836225172,
      "loss": 2.2939,
      "step": 33026
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.8381797075271606,
      "learning_rate": 0.0005700910557813548,
      "loss": 2.3046,
      "step": 33027
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.7334146499633789,
      "learning_rate": 0.0005700892752873225,
      "loss": 2.1928,
      "step": 33028
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6340002417564392,
      "learning_rate": 0.0005700874947430755,
      "loss": 2.2747,
      "step": 33029
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.7382001876831055,
      "learning_rate": 0.0005700857141486143,
      "loss": 2.4038,
      "step": 33030
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.8875411748886108,
      "learning_rate": 0.0005700839335039388,
      "loss": 2.1913,
      "step": 33031
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.8038643598556519,
      "learning_rate": 0.0005700821528090499,
      "loss": 2.4245,
      "step": 33032
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6370940804481506,
      "learning_rate": 0.0005700803720639474,
      "loss": 2.3753,
      "step": 33033
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.7725825905799866,
      "learning_rate": 0.0005700785912686319,
      "loss": 2.1946,
      "step": 33034
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.730210542678833,
      "learning_rate": 0.0005700768104231038,
      "loss": 2.4045,
      "step": 33035
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6620138883590698,
      "learning_rate": 0.0005700750295273633,
      "loss": 2.2936,
      "step": 33036
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6839022636413574,
      "learning_rate": 0.0005700732485814107,
      "loss": 2.2751,
      "step": 33037
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.7729943990707397,
      "learning_rate": 0.0005700714675852464,
      "loss": 1.9472,
      "step": 33038
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.8612713813781738,
      "learning_rate": 0.0005700696865388707,
      "loss": 2.4229,
      "step": 33039
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6443866491317749,
      "learning_rate": 0.0005700679054422839,
      "loss": 2.31,
      "step": 33040
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6810159087181091,
      "learning_rate": 0.0005700661242954864,
      "loss": 2.2692,
      "step": 33041
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.804791271686554,
      "learning_rate": 0.0005700643430984785,
      "loss": 2.2652,
      "step": 33042
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.7419309616088867,
      "learning_rate": 0.0005700625618512607,
      "loss": 2.3618,
      "step": 33043
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6366571187973022,
      "learning_rate": 0.000570060780553833,
      "loss": 2.467,
      "step": 33044
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6511828899383545,
      "learning_rate": 0.000570058999206196,
      "loss": 2.3252,
      "step": 33045
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.8710397481918335,
      "learning_rate": 0.0005700572178083498,
      "loss": 2.235,
      "step": 33046
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6602710485458374,
      "learning_rate": 0.000570055436360295,
      "loss": 2.2027,
      "step": 33047
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.7086889743804932,
      "learning_rate": 0.0005700536548620316,
      "loss": 2.3075,
      "step": 33048
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.7254994511604309,
      "learning_rate": 0.0005700518733135604,
      "loss": 2.319,
      "step": 33049
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6213517785072327,
      "learning_rate": 0.0005700500917148814,
      "loss": 2.3278,
      "step": 33050
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6261430978775024,
      "learning_rate": 0.0005700483100659949,
      "loss": 2.4005,
      "step": 33051
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6610519886016846,
      "learning_rate": 0.0005700465283669013,
      "loss": 2.1578,
      "step": 33052
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6732816696166992,
      "learning_rate": 0.000570044746617601,
      "loss": 2.2995,
      "step": 33053
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.7306255102157593,
      "learning_rate": 0.0005700429648180943,
      "loss": 2.352,
      "step": 33054
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6671055555343628,
      "learning_rate": 0.0005700411829683815,
      "loss": 2.4786,
      "step": 33055
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.7167569398880005,
      "learning_rate": 0.000570039401068463,
      "loss": 2.3354,
      "step": 33056
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6568601727485657,
      "learning_rate": 0.000570037619118339,
      "loss": 2.4146,
      "step": 33057
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6287283301353455,
      "learning_rate": 0.0005700358371180099,
      "loss": 2.3575,
      "step": 33058
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6194731593132019,
      "learning_rate": 0.000570034055067476,
      "loss": 2.3407,
      "step": 33059
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6967337131500244,
      "learning_rate": 0.0005700322729667379,
      "loss": 2.3632,
      "step": 33060
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6027197241783142,
      "learning_rate": 0.0005700304908157955,
      "loss": 2.7428,
      "step": 33061
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.651366114616394,
      "learning_rate": 0.0005700287086146494,
      "loss": 2.129,
      "step": 33062
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6286380290985107,
      "learning_rate": 0.0005700269263632998,
      "loss": 2.0476,
      "step": 33063
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6222167611122131,
      "learning_rate": 0.0005700251440617472,
      "loss": 2.3332,
      "step": 33064
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6582309007644653,
      "learning_rate": 0.0005700233617099917,
      "loss": 2.2119,
      "step": 33065
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6109291911125183,
      "learning_rate": 0.0005700215793080338,
      "loss": 2.2015,
      "step": 33066
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6298537850379944,
      "learning_rate": 0.0005700197968558739,
      "loss": 2.24,
      "step": 33067
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6552506685256958,
      "learning_rate": 0.0005700180143535121,
      "loss": 2.2488,
      "step": 33068
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.7154690623283386,
      "learning_rate": 0.000570016231800949,
      "loss": 2.4295,
      "step": 33069
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.5635026097297668,
      "learning_rate": 0.0005700144491981847,
      "loss": 2.1548,
      "step": 33070
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.7410127520561218,
      "learning_rate": 0.0005700126665452196,
      "loss": 2.3068,
      "step": 33071
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6658688187599182,
      "learning_rate": 0.000570010883842054,
      "loss": 2.1795,
      "step": 33072
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6773008108139038,
      "learning_rate": 0.0005700091010886883,
      "loss": 2.3522,
      "step": 33073
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.7259719371795654,
      "learning_rate": 0.0005700073182851228,
      "loss": 2.4969,
      "step": 33074
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.7351318597793579,
      "learning_rate": 0.000570005535431358,
      "loss": 2.3917,
      "step": 33075
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6850709319114685,
      "learning_rate": 0.0005700037525273939,
      "loss": 2.3111,
      "step": 33076
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.7184998989105225,
      "learning_rate": 0.0005700019695732312,
      "loss": 2.1392,
      "step": 33077
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6890105605125427,
      "learning_rate": 0.0005700001865688699,
      "loss": 2.3878,
      "step": 33078
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6819241046905518,
      "learning_rate": 0.0005699984035143105,
      "loss": 2.2156,
      "step": 33079
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6150093078613281,
      "learning_rate": 0.0005699966204095532,
      "loss": 2.2906,
      "step": 33080
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6738259792327881,
      "learning_rate": 0.0005699948372545984,
      "loss": 2.2177,
      "step": 33081
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6518822312355042,
      "learning_rate": 0.0005699930540494466,
      "loss": 2.2339,
      "step": 33082
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6711459159851074,
      "learning_rate": 0.000569991270794098,
      "loss": 2.3364,
      "step": 33083
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6538423895835876,
      "learning_rate": 0.000569989487488553,
      "loss": 2.0972,
      "step": 33084
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.617311418056488,
      "learning_rate": 0.0005699877041328117,
      "loss": 2.1573,
      "step": 33085
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6512206792831421,
      "learning_rate": 0.0005699859207268746,
      "loss": 2.2088,
      "step": 33086
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6525938510894775,
      "learning_rate": 0.000569984137270742,
      "loss": 2.2555,
      "step": 33087
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.647998571395874,
      "learning_rate": 0.0005699823537644144,
      "loss": 2.3984,
      "step": 33088
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.582822322845459,
      "learning_rate": 0.0005699805702078919,
      "loss": 2.1585,
      "step": 33089
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6135560870170593,
      "learning_rate": 0.0005699787866011749,
      "loss": 2.4091,
      "step": 33090
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6395660638809204,
      "learning_rate": 0.0005699770029442638,
      "loss": 2.4926,
      "step": 33091
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6169586181640625,
      "learning_rate": 0.0005699752192371588,
      "loss": 2.4997,
      "step": 33092
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6112475395202637,
      "learning_rate": 0.0005699734354798604,
      "loss": 2.4027,
      "step": 33093
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.686681866645813,
      "learning_rate": 0.0005699716516723687,
      "loss": 2.431,
      "step": 33094
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.5843517184257507,
      "learning_rate": 0.0005699698678146844,
      "loss": 2.0611,
      "step": 33095
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6734287142753601,
      "learning_rate": 0.0005699680839068074,
      "loss": 2.369,
      "step": 33096
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6853426694869995,
      "learning_rate": 0.0005699662999487384,
      "loss": 2.1306,
      "step": 33097
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6674806475639343,
      "learning_rate": 0.0005699645159404774,
      "loss": 2.3148,
      "step": 33098
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6965973377227783,
      "learning_rate": 0.000569962731882025,
      "loss": 2.2772,
      "step": 33099
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6472269296646118,
      "learning_rate": 0.0005699609477733814,
      "loss": 2.2093,
      "step": 33100
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.751054048538208,
      "learning_rate": 0.0005699591636145469,
      "loss": 2.1501,
      "step": 33101
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6494491696357727,
      "learning_rate": 0.000569957379405522,
      "loss": 2.2406,
      "step": 33102
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6893275380134583,
      "learning_rate": 0.0005699555951463069,
      "loss": 2.239,
      "step": 33103
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.7512690424919128,
      "learning_rate": 0.0005699538108369021,
      "loss": 2.3106,
      "step": 33104
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.9786357283592224,
      "learning_rate": 0.0005699520264773075,
      "loss": 2.1725,
      "step": 33105
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6667916774749756,
      "learning_rate": 0.0005699502420675239,
      "loss": 2.3975,
      "step": 33106
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.0145900249481201,
      "learning_rate": 0.0005699484576075515,
      "loss": 2.1285,
      "step": 33107
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.8018466234207153,
      "learning_rate": 0.0005699466730973904,
      "loss": 2.1767,
      "step": 33108
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6457071900367737,
      "learning_rate": 0.0005699448885370412,
      "loss": 2.3841,
      "step": 33109
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.7911092042922974,
      "learning_rate": 0.0005699431039265043,
      "loss": 2.01,
      "step": 33110
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.9769337773323059,
      "learning_rate": 0.0005699413192657796,
      "loss": 2.0605,
      "step": 33111
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6383607387542725,
      "learning_rate": 0.0005699395345548679,
      "loss": 2.1648,
      "step": 33112
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6878769993782043,
      "learning_rate": 0.0005699377497937694,
      "loss": 2.4448,
      "step": 33113
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.7680072784423828,
      "learning_rate": 0.0005699359649824843,
      "loss": 2.2844,
      "step": 33114
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.7360825538635254,
      "learning_rate": 0.000569934180121013,
      "loss": 2.2872,
      "step": 33115
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6642758846282959,
      "learning_rate": 0.0005699323952093558,
      "loss": 2.0578,
      "step": 33116
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.7133007645606995,
      "learning_rate": 0.0005699306102475132,
      "loss": 2.3534,
      "step": 33117
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6792699098587036,
      "learning_rate": 0.0005699288252354852,
      "loss": 2.3493,
      "step": 33118
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6205081939697266,
      "learning_rate": 0.0005699270401732726,
      "loss": 2.2354,
      "step": 33119
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6823005080223083,
      "learning_rate": 0.0005699252550608753,
      "loss": 2.3197,
      "step": 33120
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.7573674917221069,
      "learning_rate": 0.0005699234698982937,
      "loss": 2.3906,
      "step": 33121
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.8832736611366272,
      "learning_rate": 0.0005699216846855284,
      "loss": 2.3777,
      "step": 33122
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6448279023170471,
      "learning_rate": 0.0005699198994225795,
      "loss": 2.2414,
      "step": 33123
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6776245832443237,
      "learning_rate": 0.0005699181141094475,
      "loss": 2.4201,
      "step": 33124
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.7811824083328247,
      "learning_rate": 0.0005699163287461325,
      "loss": 2.065,
      "step": 33125
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.655215322971344,
      "learning_rate": 0.000569914543332635,
      "loss": 2.3483,
      "step": 33126
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6671246290206909,
      "learning_rate": 0.0005699127578689553,
      "loss": 2.4063,
      "step": 33127
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6786909103393555,
      "learning_rate": 0.0005699109723550937,
      "loss": 2.1345,
      "step": 33128
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6556946635246277,
      "learning_rate": 0.0005699091867910506,
      "loss": 2.3715,
      "step": 33129
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.7931499481201172,
      "learning_rate": 0.0005699074011768263,
      "loss": 2.3566,
      "step": 33130
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.8001559972763062,
      "learning_rate": 0.000569905615512421,
      "loss": 2.423,
      "step": 33131
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.7982110381126404,
      "learning_rate": 0.0005699038297978352,
      "loss": 2.1275,
      "step": 33132
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6827203631401062,
      "learning_rate": 0.0005699020440330692,
      "loss": 2.1309,
      "step": 33133
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.7773775458335876,
      "learning_rate": 0.0005699002582181233,
      "loss": 2.3442,
      "step": 33134
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6704099178314209,
      "learning_rate": 0.000569898472352998,
      "loss": 2.0578,
      "step": 33135
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.7144174575805664,
      "learning_rate": 0.0005698966864376933,
      "loss": 2.5468,
      "step": 33136
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.7712196111679077,
      "learning_rate": 0.0005698949004722098,
      "loss": 2.3195,
      "step": 33137
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.7765796184539795,
      "learning_rate": 0.0005698931144565477,
      "loss": 2.2201,
      "step": 33138
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.7150713801383972,
      "learning_rate": 0.0005698913283907074,
      "loss": 2.4099,
      "step": 33139
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.679774820804596,
      "learning_rate": 0.0005698895422746892,
      "loss": 2.2084,
      "step": 33140
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.849873423576355,
      "learning_rate": 0.0005698877561084933,
      "loss": 2.3559,
      "step": 33141
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.761985182762146,
      "learning_rate": 0.0005698859698921204,
      "loss": 2.112,
      "step": 33142
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.7118210196495056,
      "learning_rate": 0.0005698841836255705,
      "loss": 2.2696,
      "step": 33143
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.7168862819671631,
      "learning_rate": 0.000569882397308844,
      "loss": 2.4614,
      "step": 33144
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.8656013607978821,
      "learning_rate": 0.0005698806109419414,
      "loss": 2.4138,
      "step": 33145
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.7405866384506226,
      "learning_rate": 0.0005698788245248628,
      "loss": 2.4815,
      "step": 33146
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.7312613725662231,
      "learning_rate": 0.0005698770380576086,
      "loss": 2.232,
      "step": 33147
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.7125496864318848,
      "learning_rate": 0.0005698752515401793,
      "loss": 2.2971,
      "step": 33148
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.8851495981216431,
      "learning_rate": 0.0005698734649725749,
      "loss": 2.2657,
      "step": 33149
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.7434159517288208,
      "learning_rate": 0.0005698716783547961,
      "loss": 2.2799,
      "step": 33150
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6982072591781616,
      "learning_rate": 0.0005698698916868429,
      "loss": 2.3405,
      "step": 33151
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6892526745796204,
      "learning_rate": 0.000569868104968716,
      "loss": 2.4278,
      "step": 33152
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.7523108124732971,
      "learning_rate": 0.0005698663182004154,
      "loss": 2.151,
      "step": 33153
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.7095564007759094,
      "learning_rate": 0.0005698645313819415,
      "loss": 2.3394,
      "step": 33154
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6673392057418823,
      "learning_rate": 0.0005698627445132947,
      "loss": 2.4805,
      "step": 33155
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6314550042152405,
      "learning_rate": 0.0005698609575944754,
      "loss": 2.3034,
      "step": 33156
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.7869697213172913,
      "learning_rate": 0.0005698591706254838,
      "loss": 2.2777,
      "step": 33157
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.7041158080101013,
      "learning_rate": 0.0005698573836063203,
      "loss": 2.4028,
      "step": 33158
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.7094170451164246,
      "learning_rate": 0.0005698555965369854,
      "loss": 2.2759,
      "step": 33159
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.7106199860572815,
      "learning_rate": 0.000569853809417479,
      "loss": 2.2365,
      "step": 33160
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.9059372544288635,
      "learning_rate": 0.0005698520222478018,
      "loss": 2.4803,
      "step": 33161
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6764277219772339,
      "learning_rate": 0.000569850235027954,
      "loss": 2.3033,
      "step": 33162
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.9313251376152039,
      "learning_rate": 0.000569848447757936,
      "loss": 2.5551,
      "step": 33163
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.8759704232215881,
      "learning_rate": 0.0005698466604377479,
      "loss": 2.2399,
      "step": 33164
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6400274634361267,
      "learning_rate": 0.0005698448730673903,
      "loss": 2.5139,
      "step": 33165
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.7071784138679504,
      "learning_rate": 0.0005698430856468635,
      "loss": 2.2387,
      "step": 33166
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.7244735360145569,
      "learning_rate": 0.0005698412981761678,
      "loss": 2.2433,
      "step": 33167
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.7773872017860413,
      "learning_rate": 0.0005698395106553035,
      "loss": 2.1031,
      "step": 33168
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.7294595241546631,
      "learning_rate": 0.0005698377230842707,
      "loss": 2.3453,
      "step": 33169
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6835429072380066,
      "learning_rate": 0.0005698359354630703,
      "loss": 2.4966,
      "step": 33170
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.7972235679626465,
      "learning_rate": 0.0005698341477917021,
      "loss": 2.4934,
      "step": 33171
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.7366416454315186,
      "learning_rate": 0.0005698323600701668,
      "loss": 2.4259,
      "step": 33172
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6978464722633362,
      "learning_rate": 0.0005698305722984645,
      "loss": 2.0879,
      "step": 33173
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6966409087181091,
      "learning_rate": 0.0005698287844765955,
      "loss": 2.3857,
      "step": 33174
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.7794549465179443,
      "learning_rate": 0.0005698269966045604,
      "loss": 2.3083,
      "step": 33175
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6605175137519836,
      "learning_rate": 0.0005698252086823593,
      "loss": 2.1517,
      "step": 33176
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6649250984191895,
      "learning_rate": 0.0005698234207099926,
      "loss": 2.39,
      "step": 33177
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.7410992980003357,
      "learning_rate": 0.0005698216326874607,
      "loss": 2.1795,
      "step": 33178
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.7968180179595947,
      "learning_rate": 0.0005698198446147637,
      "loss": 2.2304,
      "step": 33179
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6902332305908203,
      "learning_rate": 0.0005698180564919021,
      "loss": 2.4589,
      "step": 33180
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.764725923538208,
      "learning_rate": 0.0005698162683188763,
      "loss": 2.2295,
      "step": 33181
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6312152147293091,
      "learning_rate": 0.0005698144800956867,
      "loss": 2.1551,
      "step": 33182
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.7602201700210571,
      "learning_rate": 0.0005698126918223334,
      "loss": 2.1452,
      "step": 33183
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6968722343444824,
      "learning_rate": 0.0005698109034988168,
      "loss": 2.2431,
      "step": 33184
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6760663986206055,
      "learning_rate": 0.0005698091151251372,
      "loss": 2.6288,
      "step": 33185
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6546074151992798,
      "learning_rate": 0.0005698073267012951,
      "loss": 2.2213,
      "step": 33186
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6708280444145203,
      "learning_rate": 0.0005698055382272907,
      "loss": 2.3018,
      "step": 33187
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6663974523544312,
      "learning_rate": 0.0005698037497031244,
      "loss": 2.3143,
      "step": 33188
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6274539828300476,
      "learning_rate": 0.0005698019611287963,
      "loss": 2.144,
      "step": 33189
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6324418187141418,
      "learning_rate": 0.000569800172504307,
      "loss": 2.334,
      "step": 33190
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.7027426958084106,
      "learning_rate": 0.0005697983838296569,
      "loss": 2.2112,
      "step": 33191
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6416164040565491,
      "learning_rate": 0.0005697965951048462,
      "loss": 2.3819,
      "step": 33192
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6851269602775574,
      "learning_rate": 0.000569794806329875,
      "loss": 2.292,
      "step": 33193
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.7295204401016235,
      "learning_rate": 0.0005697930175047439,
      "loss": 2.4412,
      "step": 33194
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6254141330718994,
      "learning_rate": 0.0005697912286294533,
      "loss": 2.1695,
      "step": 33195
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.7260001301765442,
      "learning_rate": 0.0005697894397040034,
      "loss": 2.2364,
      "step": 33196
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.687614381313324,
      "learning_rate": 0.0005697876507283945,
      "loss": 2.1834,
      "step": 33197
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6351355314254761,
      "learning_rate": 0.0005697858617026271,
      "loss": 2.377,
      "step": 33198
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6892365217208862,
      "learning_rate": 0.0005697840726267012,
      "loss": 2.4151,
      "step": 33199
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6127223968505859,
      "learning_rate": 0.0005697822835006175,
      "loss": 2.2672,
      "step": 33200
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.7362662553787231,
      "learning_rate": 0.0005697804943243762,
      "loss": 2.3274,
      "step": 33201
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6128473877906799,
      "learning_rate": 0.0005697787050979776,
      "loss": 2.3757,
      "step": 33202
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6713333129882812,
      "learning_rate": 0.000569776915821422,
      "loss": 2.2858,
      "step": 33203
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6572028398513794,
      "learning_rate": 0.0005697751264947097,
      "loss": 2.4042,
      "step": 33204
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.7193668484687805,
      "learning_rate": 0.0005697733371178412,
      "loss": 2.4886,
      "step": 33205
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.613105833530426,
      "learning_rate": 0.0005697715476908167,
      "loss": 2.2873,
      "step": 33206
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.7421178221702576,
      "learning_rate": 0.0005697697582136367,
      "loss": 2.3942,
      "step": 33207
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.7761504054069519,
      "learning_rate": 0.0005697679686863013,
      "loss": 2.1827,
      "step": 33208
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.5959526300430298,
      "learning_rate": 0.000569766179108811,
      "loss": 2.2506,
      "step": 33209
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6465826034545898,
      "learning_rate": 0.000569764389481166,
      "loss": 2.6329,
      "step": 33210
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6416730880737305,
      "learning_rate": 0.0005697625998033668,
      "loss": 2.3535,
      "step": 33211
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.5763314366340637,
      "learning_rate": 0.0005697608100754136,
      "loss": 2.2317,
      "step": 33212
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.65486741065979,
      "learning_rate": 0.0005697590202973067,
      "loss": 2.2818,
      "step": 33213
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6122965216636658,
      "learning_rate": 0.0005697572304690467,
      "loss": 2.1149,
      "step": 33214
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6562069654464722,
      "learning_rate": 0.0005697554405906334,
      "loss": 2.5856,
      "step": 33215
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6410079002380371,
      "learning_rate": 0.0005697536506620678,
      "loss": 2.5343,
      "step": 33216
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6190462112426758,
      "learning_rate": 0.0005697518606833499,
      "loss": 2.3648,
      "step": 33217
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6094770431518555,
      "learning_rate": 0.0005697500706544799,
      "loss": 2.4484,
      "step": 33218
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6341766119003296,
      "learning_rate": 0.0005697482805754582,
      "loss": 2.3217,
      "step": 33219
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6295506954193115,
      "learning_rate": 0.0005697464904462854,
      "loss": 2.2186,
      "step": 33220
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6320799589157104,
      "learning_rate": 0.0005697447002669615,
      "loss": 2.1072,
      "step": 33221
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6772138476371765,
      "learning_rate": 0.000569742910037487,
      "loss": 2.458,
      "step": 33222
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6973821520805359,
      "learning_rate": 0.0005697411197578622,
      "loss": 2.2056,
      "step": 33223
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.670528769493103,
      "learning_rate": 0.0005697393294280874,
      "loss": 2.3565,
      "step": 33224
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.7075932025909424,
      "learning_rate": 0.000569737539048163,
      "loss": 2.456,
      "step": 33225
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6476991176605225,
      "learning_rate": 0.0005697357486180893,
      "loss": 2.38,
      "step": 33226
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6956098079681396,
      "learning_rate": 0.0005697339581378667,
      "loss": 2.4128,
      "step": 33227
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6635351181030273,
      "learning_rate": 0.0005697321676074954,
      "loss": 2.4582,
      "step": 33228
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6763717532157898,
      "learning_rate": 0.0005697303770269758,
      "loss": 2.3088,
      "step": 33229
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6267282366752625,
      "learning_rate": 0.0005697285863963082,
      "loss": 2.0662,
      "step": 33230
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6413576602935791,
      "learning_rate": 0.0005697267957154931,
      "loss": 2.2998,
      "step": 33231
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.7538350820541382,
      "learning_rate": 0.0005697250049845305,
      "loss": 2.3093,
      "step": 33232
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6458447575569153,
      "learning_rate": 0.000569723214203421,
      "loss": 2.0736,
      "step": 33233
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.7479848861694336,
      "learning_rate": 0.0005697214233721649,
      "loss": 2.1864,
      "step": 33234
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.700874924659729,
      "learning_rate": 0.0005697196324907626,
      "loss": 2.248,
      "step": 33235
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.7170717120170593,
      "learning_rate": 0.0005697178415592141,
      "loss": 2.5844,
      "step": 33236
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6736727356910706,
      "learning_rate": 0.0005697160505775201,
      "loss": 2.3488,
      "step": 33237
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6564576625823975,
      "learning_rate": 0.0005697142595456808,
      "loss": 2.3498,
      "step": 33238
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.8403686881065369,
      "learning_rate": 0.0005697124684636964,
      "loss": 2.3458,
      "step": 33239
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.8533400297164917,
      "learning_rate": 0.0005697106773315675,
      "loss": 2.0809,
      "step": 33240
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.7286823391914368,
      "learning_rate": 0.0005697088861492943,
      "loss": 2.3885,
      "step": 33241
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.7086895108222961,
      "learning_rate": 0.000569707094916877,
      "loss": 2.5162,
      "step": 33242
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.8176812529563904,
      "learning_rate": 0.0005697053036343163,
      "loss": 2.1384,
      "step": 33243
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6381121873855591,
      "learning_rate": 0.000569703512301612,
      "loss": 2.3716,
      "step": 33244
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.5663110017776489,
      "learning_rate": 0.0005697017209187649,
      "loss": 2.5155,
      "step": 33245
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6307039856910706,
      "learning_rate": 0.0005696999294857751,
      "loss": 2.1719,
      "step": 33246
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.7449934482574463,
      "learning_rate": 0.0005696981380026429,
      "loss": 2.191,
      "step": 33247
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6734147071838379,
      "learning_rate": 0.000569696346469369,
      "loss": 2.3217,
      "step": 33248
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6308204531669617,
      "learning_rate": 0.0005696945548859532,
      "loss": 2.6534,
      "step": 33249
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6196898221969604,
      "learning_rate": 0.0005696927632523963,
      "loss": 2.5358,
      "step": 33250
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6762380599975586,
      "learning_rate": 0.0005696909715686983,
      "loss": 2.45,
      "step": 33251
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.7301334142684937,
      "learning_rate": 0.0005696891798348596,
      "loss": 2.0897,
      "step": 33252
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6216720938682556,
      "learning_rate": 0.0005696873880508806,
      "loss": 2.2925,
      "step": 33253
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6047382354736328,
      "learning_rate": 0.0005696855962167617,
      "loss": 2.076,
      "step": 33254
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.7364232540130615,
      "learning_rate": 0.0005696838043325031,
      "loss": 2.3068,
      "step": 33255
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6841675639152527,
      "learning_rate": 0.0005696820123981053,
      "loss": 2.2523,
      "step": 33256
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6900175213813782,
      "learning_rate": 0.0005696802204135684,
      "loss": 2.5048,
      "step": 33257
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.7775070667266846,
      "learning_rate": 0.000569678428378893,
      "loss": 2.3259,
      "step": 33258
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6989985108375549,
      "learning_rate": 0.0005696766362940791,
      "loss": 2.3088,
      "step": 33259
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.719353973865509,
      "learning_rate": 0.0005696748441591274,
      "loss": 2.2505,
      "step": 33260
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6455236077308655,
      "learning_rate": 0.000569673051974038,
      "loss": 2.2389,
      "step": 33261
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.7300071120262146,
      "learning_rate": 0.0005696712597388112,
      "loss": 2.3614,
      "step": 33262
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6985184550285339,
      "learning_rate": 0.0005696694674534476,
      "loss": 2.0735,
      "step": 33263
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6646224856376648,
      "learning_rate": 0.0005696676751179472,
      "loss": 2.3482,
      "step": 33264
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6346637010574341,
      "learning_rate": 0.0005696658827323105,
      "loss": 2.5217,
      "step": 33265
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6855340003967285,
      "learning_rate": 0.0005696640902965378,
      "loss": 2.3785,
      "step": 33266
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.7272427678108215,
      "learning_rate": 0.0005696622978106296,
      "loss": 2.4665,
      "step": 33267
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.8247542381286621,
      "learning_rate": 0.000569660505274586,
      "loss": 2.2275,
      "step": 33268
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6879956722259521,
      "learning_rate": 0.0005696587126884074,
      "loss": 2.2203,
      "step": 33269
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.7305409908294678,
      "learning_rate": 0.0005696569200520942,
      "loss": 2.1657,
      "step": 33270
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.5954713821411133,
      "learning_rate": 0.0005696551273656466,
      "loss": 2.2199,
      "step": 33271
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6733417510986328,
      "learning_rate": 0.0005696533346290651,
      "loss": 2.2824,
      "step": 33272
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6123059391975403,
      "learning_rate": 0.00056965154184235,
      "loss": 2.3369,
      "step": 33273
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.7052836418151855,
      "learning_rate": 0.0005696497490055015,
      "loss": 2.2893,
      "step": 33274
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6596969962120056,
      "learning_rate": 0.00056964795611852,
      "loss": 2.3014,
      "step": 33275
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.8480443358421326,
      "learning_rate": 0.0005696461631814059,
      "loss": 2.3708,
      "step": 33276
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.7044782042503357,
      "learning_rate": 0.0005696443701941595,
      "loss": 2.2449,
      "step": 33277
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.5800887942314148,
      "learning_rate": 0.0005696425771567811,
      "loss": 2.1302,
      "step": 33278
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.5977724194526672,
      "learning_rate": 0.000569640784069271,
      "loss": 2.472,
      "step": 33279
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.667289674282074,
      "learning_rate": 0.0005696389909316297,
      "loss": 2.198,
      "step": 33280
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.7405929565429688,
      "learning_rate": 0.0005696371977438575,
      "loss": 2.1953,
      "step": 33281
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.669421911239624,
      "learning_rate": 0.0005696354045059545,
      "loss": 2.3497,
      "step": 33282
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.7035943269729614,
      "learning_rate": 0.0005696336112179212,
      "loss": 2.4466,
      "step": 33283
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6987738013267517,
      "learning_rate": 0.000569631817879758,
      "loss": 2.3184,
      "step": 33284
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.5955469608306885,
      "learning_rate": 0.0005696300244914651,
      "loss": 2.3591,
      "step": 33285
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6017998456954956,
      "learning_rate": 0.0005696282310530429,
      "loss": 2.3456,
      "step": 33286
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6335872411727905,
      "learning_rate": 0.0005696264375644917,
      "loss": 2.3768,
      "step": 33287
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.812674343585968,
      "learning_rate": 0.0005696246440258119,
      "loss": 2.253,
      "step": 33288
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6575721502304077,
      "learning_rate": 0.0005696228504370037,
      "loss": 2.2068,
      "step": 33289
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.765839695930481,
      "learning_rate": 0.0005696210567980675,
      "loss": 2.3626,
      "step": 33290
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.7567142248153687,
      "learning_rate": 0.0005696192631090038,
      "loss": 2.1838,
      "step": 33291
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.7721896767616272,
      "learning_rate": 0.0005696174693698127,
      "loss": 2.0221,
      "step": 33292
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6236363649368286,
      "learning_rate": 0.0005696156755804946,
      "loss": 2.2978,
      "step": 33293
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6566622853279114,
      "learning_rate": 0.0005696138817410499,
      "loss": 2.3301,
      "step": 33294
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6981834769248962,
      "learning_rate": 0.000569612087851479,
      "loss": 2.2735,
      "step": 33295
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.644644021987915,
      "learning_rate": 0.000569610293911782,
      "loss": 2.2723,
      "step": 33296
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.7231488823890686,
      "learning_rate": 0.0005696084999219594,
      "loss": 2.3351,
      "step": 33297
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.7533949613571167,
      "learning_rate": 0.0005696067058820114,
      "loss": 2.2514,
      "step": 33298
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6034095883369446,
      "learning_rate": 0.0005696049117919385,
      "loss": 2.2347,
      "step": 33299
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6978083252906799,
      "learning_rate": 0.000569603117651741,
      "loss": 2.3469,
      "step": 33300
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6666735410690308,
      "learning_rate": 0.0005696013234614191,
      "loss": 1.9949,
      "step": 33301
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.7162773013114929,
      "learning_rate": 0.0005695995292209733,
      "loss": 2.5749,
      "step": 33302
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.705922544002533,
      "learning_rate": 0.0005695977349304038,
      "loss": 2.5721,
      "step": 33303
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6100335717201233,
      "learning_rate": 0.0005695959405897109,
      "loss": 2.2494,
      "step": 33304
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6455125212669373,
      "learning_rate": 0.0005695941461988952,
      "loss": 2.3777,
      "step": 33305
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.743818461894989,
      "learning_rate": 0.0005695923517579568,
      "loss": 2.2345,
      "step": 33306
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6343221664428711,
      "learning_rate": 0.0005695905572668962,
      "loss": 2.2871,
      "step": 33307
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6741090416908264,
      "learning_rate": 0.0005695887627257135,
      "loss": 2.2716,
      "step": 33308
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6642642617225647,
      "learning_rate": 0.0005695869681344092,
      "loss": 2.1586,
      "step": 33309
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6371359825134277,
      "learning_rate": 0.0005695851734929835,
      "loss": 2.2616,
      "step": 33310
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6556206345558167,
      "learning_rate": 0.000569583378801437,
      "loss": 2.3766,
      "step": 33311
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6527436375617981,
      "learning_rate": 0.0005695815840597697,
      "loss": 2.1633,
      "step": 33312
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.640598714351654,
      "learning_rate": 0.0005695797892679821,
      "loss": 2.4591,
      "step": 33313
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6454119682312012,
      "learning_rate": 0.0005695779944260746,
      "loss": 2.3059,
      "step": 33314
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6483353972434998,
      "learning_rate": 0.0005695761995340475,
      "loss": 2.2363,
      "step": 33315
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.629550039768219,
      "learning_rate": 0.000569574404591901,
      "loss": 2.1602,
      "step": 33316
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.7042959928512573,
      "learning_rate": 0.0005695726095996356,
      "loss": 2.1532,
      "step": 33317
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6396886706352234,
      "learning_rate": 0.0005695708145572515,
      "loss": 2.0663,
      "step": 33318
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6715009212493896,
      "learning_rate": 0.0005695690194647492,
      "loss": 2.3374,
      "step": 33319
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.7407647967338562,
      "learning_rate": 0.0005695672243221288,
      "loss": 2.1732,
      "step": 33320
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.5933812856674194,
      "learning_rate": 0.0005695654291293908,
      "loss": 2.1477,
      "step": 33321
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.674738883972168,
      "learning_rate": 0.0005695636338865355,
      "loss": 2.2458,
      "step": 33322
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.8089740872383118,
      "learning_rate": 0.0005695618385935633,
      "loss": 2.1818,
      "step": 33323
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6749461889266968,
      "learning_rate": 0.0005695600432504745,
      "loss": 2.2584,
      "step": 33324
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.7054223418235779,
      "learning_rate": 0.0005695582478572693,
      "loss": 2.3626,
      "step": 33325
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6900982856750488,
      "learning_rate": 0.0005695564524139481,
      "loss": 2.4283,
      "step": 33326
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6717131733894348,
      "learning_rate": 0.0005695546569205114,
      "loss": 2.1641,
      "step": 33327
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.7120504975318909,
      "learning_rate": 0.0005695528613769593,
      "loss": 2.4002,
      "step": 33328
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6909406185150146,
      "learning_rate": 0.0005695510657832922,
      "loss": 2.3167,
      "step": 33329
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.666398286819458,
      "learning_rate": 0.0005695492701395105,
      "loss": 2.0636,
      "step": 33330
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.7346190214157104,
      "learning_rate": 0.0005695474744456146,
      "loss": 2.2989,
      "step": 33331
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.7577620148658752,
      "learning_rate": 0.0005695456787016047,
      "loss": 2.4891,
      "step": 33332
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.659326434135437,
      "learning_rate": 0.0005695438829074811,
      "loss": 2.1934,
      "step": 33333
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6827726364135742,
      "learning_rate": 0.0005695420870632442,
      "loss": 2.2404,
      "step": 33334
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.7241412401199341,
      "learning_rate": 0.0005695402911688944,
      "loss": 2.3202,
      "step": 33335
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.8872131705284119,
      "learning_rate": 0.000569538495224432,
      "loss": 2.2504,
      "step": 33336
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6296292543411255,
      "learning_rate": 0.0005695366992298572,
      "loss": 2.1747,
      "step": 33337
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.8249791264533997,
      "learning_rate": 0.0005695349031851705,
      "loss": 2.1759,
      "step": 33338
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.8928187489509583,
      "learning_rate": 0.0005695331070903723,
      "loss": 2.3413,
      "step": 33339
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6996395587921143,
      "learning_rate": 0.0005695313109454626,
      "loss": 2.2255,
      "step": 33340
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.7712095975875854,
      "learning_rate": 0.0005695295147504421,
      "loss": 2.363,
      "step": 33341
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.9712343215942383,
      "learning_rate": 0.0005695277185053108,
      "loss": 2.3245,
      "step": 33342
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6158835291862488,
      "learning_rate": 0.0005695259222100694,
      "loss": 2.228,
      "step": 33343
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.649563729763031,
      "learning_rate": 0.0005695241258647178,
      "loss": 2.2838,
      "step": 33344
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.7085578441619873,
      "learning_rate": 0.0005695223294692568,
      "loss": 2.3096,
      "step": 33345
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.7905003428459167,
      "learning_rate": 0.0005695205330236864,
      "loss": 2.3376,
      "step": 33346
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6986607909202576,
      "learning_rate": 0.0005695187365280071,
      "loss": 2.2897,
      "step": 33347
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.7819268107414246,
      "learning_rate": 0.0005695169399822192,
      "loss": 2.291,
      "step": 33348
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.7555844783782959,
      "learning_rate": 0.0005695151433863229,
      "loss": 2.2762,
      "step": 33349
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.792524516582489,
      "learning_rate": 0.0005695133467403188,
      "loss": 2.4141,
      "step": 33350
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6317756175994873,
      "learning_rate": 0.000569511550044207,
      "loss": 2.3138,
      "step": 33351
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.8016167283058167,
      "learning_rate": 0.0005695097532979878,
      "loss": 2.339,
      "step": 33352
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6423292756080627,
      "learning_rate": 0.0005695079565016619,
      "loss": 2.3139,
      "step": 33353
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6384962797164917,
      "learning_rate": 0.0005695061596552291,
      "loss": 2.2646,
      "step": 33354
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6045129895210266,
      "learning_rate": 0.0005695043627586902,
      "loss": 2.3754,
      "step": 33355
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6072821021080017,
      "learning_rate": 0.0005695025658120454,
      "loss": 2.2822,
      "step": 33356
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.7029136419296265,
      "learning_rate": 0.0005695007688152949,
      "loss": 2.2994,
      "step": 33357
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6164661049842834,
      "learning_rate": 0.0005694989717684391,
      "loss": 2.2109,
      "step": 33358
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6483907699584961,
      "learning_rate": 0.0005694971746714784,
      "loss": 2.6105,
      "step": 33359
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6374027132987976,
      "learning_rate": 0.000569495377524413,
      "loss": 2.3477,
      "step": 33360
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.7101073265075684,
      "learning_rate": 0.0005694935803272434,
      "loss": 2.342,
      "step": 33361
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6560577154159546,
      "learning_rate": 0.0005694917830799699,
      "loss": 2.188,
      "step": 33362
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6535358428955078,
      "learning_rate": 0.0005694899857825928,
      "loss": 2.3846,
      "step": 33363
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.7396068572998047,
      "learning_rate": 0.0005694881884351124,
      "loss": 2.3962,
      "step": 33364
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.7011464834213257,
      "learning_rate": 0.0005694863910375289,
      "loss": 2.3233,
      "step": 33365
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6799129843711853,
      "learning_rate": 0.000569484593589843,
      "loss": 2.1087,
      "step": 33366
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.7052147388458252,
      "learning_rate": 0.0005694827960920546,
      "loss": 2.5659,
      "step": 33367
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6893421411514282,
      "learning_rate": 0.0005694809985441645,
      "loss": 2.2686,
      "step": 33368
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.7392269372940063,
      "learning_rate": 0.0005694792009461727,
      "loss": 2.4035,
      "step": 33369
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.758396565914154,
      "learning_rate": 0.0005694774032980796,
      "loss": 2.1891,
      "step": 33370
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.7021269798278809,
      "learning_rate": 0.0005694756055998856,
      "loss": 2.2041,
      "step": 33371
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.7533462047576904,
      "learning_rate": 0.000569473807851591,
      "loss": 2.2775,
      "step": 33372
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.8019042015075684,
      "learning_rate": 0.0005694720100531961,
      "loss": 2.1713,
      "step": 33373
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6572269201278687,
      "learning_rate": 0.0005694702122047013,
      "loss": 2.3998,
      "step": 33374
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.673693835735321,
      "learning_rate": 0.000569468414306107,
      "loss": 2.2549,
      "step": 33375
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.7869241833686829,
      "learning_rate": 0.0005694666163574133,
      "loss": 2.2346,
      "step": 33376
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.7696240544319153,
      "learning_rate": 0.0005694648183586207,
      "loss": 2.2863,
      "step": 33377
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6547664403915405,
      "learning_rate": 0.0005694630203097295,
      "loss": 2.2225,
      "step": 33378
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.7454348802566528,
      "learning_rate": 0.0005694612222107399,
      "loss": 2.3743,
      "step": 33379
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.696442723274231,
      "learning_rate": 0.0005694594240616526,
      "loss": 2.6231,
      "step": 33380
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.7664307355880737,
      "learning_rate": 0.0005694576258624675,
      "loss": 2.3287,
      "step": 33381
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6677624583244324,
      "learning_rate": 0.0005694558276131853,
      "loss": 2.3185,
      "step": 33382
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6613977551460266,
      "learning_rate": 0.0005694540293138062,
      "loss": 2.4577,
      "step": 33383
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.7746266722679138,
      "learning_rate": 0.0005694522309643304,
      "loss": 2.1681,
      "step": 33384
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6611276268959045,
      "learning_rate": 0.0005694504325647584,
      "loss": 2.5226,
      "step": 33385
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.7331912517547607,
      "learning_rate": 0.0005694486341150905,
      "loss": 2.2807,
      "step": 33386
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.83029705286026,
      "learning_rate": 0.000569446835615327,
      "loss": 2.3593,
      "step": 33387
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.9708970785140991,
      "learning_rate": 0.0005694450370654682,
      "loss": 2.3614,
      "step": 33388
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6517114639282227,
      "learning_rate": 0.0005694432384655145,
      "loss": 2.0917,
      "step": 33389
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6486279368400574,
      "learning_rate": 0.0005694414398154661,
      "loss": 2.2551,
      "step": 33390
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.7983576655387878,
      "learning_rate": 0.0005694396411153236,
      "loss": 2.1596,
      "step": 33391
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6639607548713684,
      "learning_rate": 0.0005694378423650873,
      "loss": 2.1483,
      "step": 33392
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.7419265508651733,
      "learning_rate": 0.0005694360435647573,
      "loss": 2.2842,
      "step": 33393
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.7936859726905823,
      "learning_rate": 0.0005694342447143339,
      "loss": 2.0914,
      "step": 33394
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.7902534604072571,
      "learning_rate": 0.0005694324458138178,
      "loss": 2.501,
      "step": 33395
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6401944160461426,
      "learning_rate": 0.0005694306468632089,
      "loss": 2.2966,
      "step": 33396
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.7761878967285156,
      "learning_rate": 0.0005694288478625079,
      "loss": 2.2285,
      "step": 33397
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6891712546348572,
      "learning_rate": 0.0005694270488117151,
      "loss": 2.134,
      "step": 33398
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.7238163352012634,
      "learning_rate": 0.0005694252497108306,
      "loss": 2.2401,
      "step": 33399
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.7678583860397339,
      "learning_rate": 0.0005694234505598548,
      "loss": 2.2645,
      "step": 33400
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.9626293182373047,
      "learning_rate": 0.0005694216513587883,
      "loss": 2.672,
      "step": 33401
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.648259699344635,
      "learning_rate": 0.0005694198521076309,
      "loss": 2.4888,
      "step": 33402
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6894230842590332,
      "learning_rate": 0.0005694180528063835,
      "loss": 2.2043,
      "step": 33403
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.9073963761329651,
      "learning_rate": 0.0005694162534550462,
      "loss": 2.3841,
      "step": 33404
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6516061425209045,
      "learning_rate": 0.0005694144540536193,
      "loss": 2.4783,
      "step": 33405
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.5968701243400574,
      "learning_rate": 0.0005694126546021031,
      "loss": 2.6568,
      "step": 33406
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.7292360067367554,
      "learning_rate": 0.0005694108551004981,
      "loss": 2.566,
      "step": 33407
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.754626452922821,
      "learning_rate": 0.0005694090555488045,
      "loss": 2.3195,
      "step": 33408
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6581361889839172,
      "learning_rate": 0.0005694072559470227,
      "loss": 2.3136,
      "step": 33409
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6353950500488281,
      "learning_rate": 0.0005694054562951531,
      "loss": 2.178,
      "step": 33410
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.7181198000907898,
      "learning_rate": 0.0005694036565931957,
      "loss": 2.337,
      "step": 33411
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.7160370945930481,
      "learning_rate": 0.0005694018568411512,
      "loss": 2.4138,
      "step": 33412
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.7526153922080994,
      "learning_rate": 0.0005694000570390197,
      "loss": 2.3111,
      "step": 33413
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.709891140460968,
      "learning_rate": 0.0005693982571868018,
      "loss": 2.3377,
      "step": 33414
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6672982573509216,
      "learning_rate": 0.0005693964572844977,
      "loss": 2.0997,
      "step": 33415
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.7048078179359436,
      "learning_rate": 0.0005693946573321077,
      "loss": 2.4302,
      "step": 33416
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.5911093950271606,
      "learning_rate": 0.0005693928573296319,
      "loss": 2.3685,
      "step": 33417
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.658514678478241,
      "learning_rate": 0.0005693910572770711,
      "loss": 2.2802,
      "step": 33418
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6094104051589966,
      "learning_rate": 0.0005693892571744254,
      "loss": 2.2842,
      "step": 33419
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.7366769313812256,
      "learning_rate": 0.0005693874570216952,
      "loss": 2.2133,
      "step": 33420
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6838945150375366,
      "learning_rate": 0.0005693856568188807,
      "loss": 2.2239,
      "step": 33421
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6613828539848328,
      "learning_rate": 0.0005693838565659823,
      "loss": 2.2515,
      "step": 33422
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6535586714744568,
      "learning_rate": 0.0005693820562630004,
      "loss": 2.3885,
      "step": 33423
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6630151271820068,
      "learning_rate": 0.0005693802559099353,
      "loss": 2.4393,
      "step": 33424
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6972423791885376,
      "learning_rate": 0.0005693784555067873,
      "loss": 2.2734,
      "step": 33425
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6191674470901489,
      "learning_rate": 0.0005693766550535569,
      "loss": 2.2327,
      "step": 33426
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.7181165218353271,
      "learning_rate": 0.0005693748545502441,
      "loss": 2.4964,
      "step": 33427
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.687184751033783,
      "learning_rate": 0.0005693730539968495,
      "loss": 2.2052,
      "step": 33428
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6463192105293274,
      "learning_rate": 0.0005693712533933733,
      "loss": 2.0153,
      "step": 33429
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6869107484817505,
      "learning_rate": 0.0005693694527398159,
      "loss": 2.0848,
      "step": 33430
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6896921396255493,
      "learning_rate": 0.0005693676520361777,
      "loss": 2.3122,
      "step": 33431
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6445651650428772,
      "learning_rate": 0.0005693658512824591,
      "loss": 2.1958,
      "step": 33432
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6187265515327454,
      "learning_rate": 0.0005693640504786601,
      "loss": 2.3007,
      "step": 33433
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6792159080505371,
      "learning_rate": 0.0005693622496247814,
      "loss": 2.3787,
      "step": 33434
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6115691065788269,
      "learning_rate": 0.0005693604487208231,
      "loss": 2.1795,
      "step": 33435
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6538807153701782,
      "learning_rate": 0.0005693586477667855,
      "loss": 2.2387,
      "step": 33436
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6632831692695618,
      "learning_rate": 0.0005693568467626691,
      "loss": 2.3244,
      "step": 33437
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6171356439590454,
      "learning_rate": 0.0005693550457084743,
      "loss": 2.2875,
      "step": 33438
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6999641060829163,
      "learning_rate": 0.0005693532446042011,
      "loss": 2.1466,
      "step": 33439
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.621296763420105,
      "learning_rate": 0.0005693514434498502,
      "loss": 2.2424,
      "step": 33440
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6542606353759766,
      "learning_rate": 0.0005693496422454218,
      "loss": 2.3018,
      "step": 33441
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6362195014953613,
      "learning_rate": 0.0005693478409909161,
      "loss": 2.0092,
      "step": 33442
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6455013751983643,
      "learning_rate": 0.0005693460396863337,
      "loss": 2.4985,
      "step": 33443
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6837364435195923,
      "learning_rate": 0.0005693442383316747,
      "loss": 2.4861,
      "step": 33444
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6611385345458984,
      "learning_rate": 0.0005693424369269395,
      "loss": 2.2258,
      "step": 33445
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.7509466409683228,
      "learning_rate": 0.0005693406354721285,
      "loss": 2.2093,
      "step": 33446
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.654197096824646,
      "learning_rate": 0.000569338833967242,
      "loss": 2.2103,
      "step": 33447
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6086367964744568,
      "learning_rate": 0.0005693370324122803,
      "loss": 2.3266,
      "step": 33448
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.8102024793624878,
      "learning_rate": 0.0005693352308072438,
      "loss": 2.3626,
      "step": 33449
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.7288011312484741,
      "learning_rate": 0.0005693334291521328,
      "loss": 2.2204,
      "step": 33450
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.7427820563316345,
      "learning_rate": 0.0005693316274469475,
      "loss": 2.4881,
      "step": 33451
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.7309442162513733,
      "learning_rate": 0.0005693298256916885,
      "loss": 2.3863,
      "step": 33452
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6534525752067566,
      "learning_rate": 0.0005693280238863559,
      "loss": 2.2783,
      "step": 33453
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6696270108222961,
      "learning_rate": 0.0005693262220309503,
      "loss": 2.2682,
      "step": 33454
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6314418911933899,
      "learning_rate": 0.0005693244201254718,
      "loss": 2.2717,
      "step": 33455
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.733011782169342,
      "learning_rate": 0.0005693226181699208,
      "loss": 2.2884,
      "step": 33456
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6464257836341858,
      "learning_rate": 0.0005693208161642975,
      "loss": 2.2826,
      "step": 33457
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.644985020160675,
      "learning_rate": 0.0005693190141086025,
      "loss": 2.1921,
      "step": 33458
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6917603015899658,
      "learning_rate": 0.000569317212002836,
      "loss": 2.3299,
      "step": 33459
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6796150803565979,
      "learning_rate": 0.0005693154098469984,
      "loss": 2.3182,
      "step": 33460
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6298119425773621,
      "learning_rate": 0.0005693136076410899,
      "loss": 2.1565,
      "step": 33461
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6571289300918579,
      "learning_rate": 0.0005693118053851109,
      "loss": 2.1298,
      "step": 33462
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.7133551239967346,
      "learning_rate": 0.0005693100030790619,
      "loss": 2.2605,
      "step": 33463
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.7527438402175903,
      "learning_rate": 0.000569308200722943,
      "loss": 2.1646,
      "step": 33464
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.662380576133728,
      "learning_rate": 0.0005693063983167546,
      "loss": 2.4065,
      "step": 33465
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6221052408218384,
      "learning_rate": 0.0005693045958604971,
      "loss": 2.4064,
      "step": 33466
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.7510748505592346,
      "learning_rate": 0.0005693027933541707,
      "loss": 2.4334,
      "step": 33467
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.7416015863418579,
      "learning_rate": 0.0005693009907977759,
      "loss": 2.332,
      "step": 33468
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6583740711212158,
      "learning_rate": 0.0005692991881913129,
      "loss": 2.281,
      "step": 33469
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6865202784538269,
      "learning_rate": 0.0005692973855347822,
      "loss": 2.3668,
      "step": 33470
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.7671056389808655,
      "learning_rate": 0.000569295582828184,
      "loss": 2.2104,
      "step": 33471
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6678270101547241,
      "learning_rate": 0.0005692937800715186,
      "loss": 2.2811,
      "step": 33472
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6570692658424377,
      "learning_rate": 0.0005692919772647865,
      "loss": 2.2425,
      "step": 33473
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.7201582193374634,
      "learning_rate": 0.0005692901744079878,
      "loss": 2.2263,
      "step": 33474
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.7205349206924438,
      "learning_rate": 0.0005692883715011231,
      "loss": 2.1333,
      "step": 33475
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6447793245315552,
      "learning_rate": 0.0005692865685441925,
      "loss": 2.0626,
      "step": 33476
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6180112957954407,
      "learning_rate": 0.0005692847655371965,
      "loss": 2.19,
      "step": 33477
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6058257222175598,
      "learning_rate": 0.0005692829624801355,
      "loss": 2.2425,
      "step": 33478
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.7137778997421265,
      "learning_rate": 0.0005692811593730096,
      "loss": 2.2296,
      "step": 33479
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.7121599912643433,
      "learning_rate": 0.0005692793562158192,
      "loss": 2.2614,
      "step": 33480
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.712548553943634,
      "learning_rate": 0.0005692775530085648,
      "loss": 2.4231,
      "step": 33481
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.7400464415550232,
      "learning_rate": 0.0005692757497512465,
      "loss": 2.556,
      "step": 33482
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.7515975832939148,
      "learning_rate": 0.0005692739464438648,
      "loss": 2.3224,
      "step": 33483
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6454315781593323,
      "learning_rate": 0.0005692721430864201,
      "loss": 2.3046,
      "step": 33484
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6387001872062683,
      "learning_rate": 0.0005692703396789126,
      "loss": 2.2804,
      "step": 33485
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6999014616012573,
      "learning_rate": 0.0005692685362213426,
      "loss": 2.2037,
      "step": 33486
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.7565458416938782,
      "learning_rate": 0.0005692667327137104,
      "loss": 2.4125,
      "step": 33487
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.7191683053970337,
      "learning_rate": 0.0005692649291560165,
      "loss": 2.1536,
      "step": 33488
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6921905279159546,
      "learning_rate": 0.0005692631255482612,
      "loss": 2.2985,
      "step": 33489
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6742521524429321,
      "learning_rate": 0.0005692613218904448,
      "loss": 2.3865,
      "step": 33490
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.7924867272377014,
      "learning_rate": 0.0005692595181825676,
      "loss": 2.2493,
      "step": 33491
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.8660054206848145,
      "learning_rate": 0.00056925771442463,
      "loss": 2.1816,
      "step": 33492
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.7156333327293396,
      "learning_rate": 0.0005692559106166323,
      "loss": 2.3712,
      "step": 33493
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.8413417339324951,
      "learning_rate": 0.0005692541067585749,
      "loss": 2.2721,
      "step": 33494
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.9686576724052429,
      "learning_rate": 0.0005692523028504579,
      "loss": 2.3259,
      "step": 33495
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.7586185932159424,
      "learning_rate": 0.0005692504988922821,
      "loss": 2.2608,
      "step": 33496
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.8734990358352661,
      "learning_rate": 0.0005692486948840474,
      "loss": 2.2272,
      "step": 33497
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.8008045554161072,
      "learning_rate": 0.0005692468908257542,
      "loss": 2.1434,
      "step": 33498
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.8100643157958984,
      "learning_rate": 0.000569245086717403,
      "loss": 2.5317,
      "step": 33499
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.7262048721313477,
      "learning_rate": 0.0005692432825589941,
      "loss": 2.2627,
      "step": 33500
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.7779178619384766,
      "learning_rate": 0.0005692414783505278,
      "loss": 2.2023,
      "step": 33501
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.8070164918899536,
      "learning_rate": 0.0005692396740920044,
      "loss": 2.1173,
      "step": 33502
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.7364604473114014,
      "learning_rate": 0.0005692378697834242,
      "loss": 2.2422,
      "step": 33503
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6505603194236755,
      "learning_rate": 0.0005692360654247875,
      "loss": 2.2453,
      "step": 33504
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6789957284927368,
      "learning_rate": 0.000569234261016095,
      "loss": 2.2963,
      "step": 33505
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6652055382728577,
      "learning_rate": 0.0005692324565573466,
      "loss": 2.2266,
      "step": 33506
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.688534677028656,
      "learning_rate": 0.0005692306520485429,
      "loss": 2.2991,
      "step": 33507
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6506085395812988,
      "learning_rate": 0.0005692288474896841,
      "loss": 2.3376,
      "step": 33508
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.7700023651123047,
      "learning_rate": 0.0005692270428807705,
      "loss": 2.3719,
      "step": 33509
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6336703300476074,
      "learning_rate": 0.0005692252382218026,
      "loss": 2.2738,
      "step": 33510
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6334267258644104,
      "learning_rate": 0.0005692234335127807,
      "loss": 2.219,
      "step": 33511
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6733125448226929,
      "learning_rate": 0.0005692216287537049,
      "loss": 2.1223,
      "step": 33512
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6153138875961304,
      "learning_rate": 0.0005692198239445758,
      "loss": 2.433,
      "step": 33513
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6764212846755981,
      "learning_rate": 0.0005692180190853937,
      "loss": 2.4335,
      "step": 33514
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.66721510887146,
      "learning_rate": 0.0005692162141761588,
      "loss": 2.1317,
      "step": 33515
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6887144446372986,
      "learning_rate": 0.0005692144092168716,
      "loss": 2.3682,
      "step": 33516
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6657105088233948,
      "learning_rate": 0.0005692126042075324,
      "loss": 2.389,
      "step": 33517
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6741120219230652,
      "learning_rate": 0.0005692107991481413,
      "loss": 2.3572,
      "step": 33518
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6086811423301697,
      "learning_rate": 0.000569208994038699,
      "loss": 2.1908,
      "step": 33519
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.5962571501731873,
      "learning_rate": 0.0005692071888792056,
      "loss": 2.2762,
      "step": 33520
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.706365168094635,
      "learning_rate": 0.0005692053836696615,
      "loss": 2.1795,
      "step": 33521
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6546503901481628,
      "learning_rate": 0.0005692035784100672,
      "loss": 2.4181,
      "step": 33522
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6849912405014038,
      "learning_rate": 0.0005692017731004227,
      "loss": 2.297,
      "step": 33523
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.655727207660675,
      "learning_rate": 0.0005691999677407285,
      "loss": 2.1849,
      "step": 33524
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.667630136013031,
      "learning_rate": 0.0005691981623309851,
      "loss": 2.2576,
      "step": 33525
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6560155153274536,
      "learning_rate": 0.0005691963568711925,
      "loss": 2.4551,
      "step": 33526
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.7211126685142517,
      "learning_rate": 0.0005691945513613513,
      "loss": 2.1546,
      "step": 33527
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.673215925693512,
      "learning_rate": 0.0005691927458014618,
      "loss": 2.263,
      "step": 33528
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6747457385063171,
      "learning_rate": 0.0005691909401915242,
      "loss": 2.3365,
      "step": 33529
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6227056980133057,
      "learning_rate": 0.0005691891345315389,
      "loss": 2.2076,
      "step": 33530
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6804394721984863,
      "learning_rate": 0.0005691873288215063,
      "loss": 2.3021,
      "step": 33531
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.7030587792396545,
      "learning_rate": 0.0005691855230614268,
      "loss": 2.1193,
      "step": 33532
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6750191450119019,
      "learning_rate": 0.0005691837172513005,
      "loss": 2.1785,
      "step": 33533
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6927124261856079,
      "learning_rate": 0.0005691819113911279,
      "loss": 2.5001,
      "step": 33534
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.639468789100647,
      "learning_rate": 0.0005691801054809093,
      "loss": 2.3506,
      "step": 33535
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.7519547343254089,
      "learning_rate": 0.0005691782995206451,
      "loss": 2.2272,
      "step": 33536
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6405320167541504,
      "learning_rate": 0.0005691764935103355,
      "loss": 2.4037,
      "step": 33537
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.8191795945167542,
      "learning_rate": 0.0005691746874499809,
      "loss": 2.4207,
      "step": 33538
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.7546148300170898,
      "learning_rate": 0.0005691728813395817,
      "loss": 1.9802,
      "step": 33539
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6980270147323608,
      "learning_rate": 0.000569171075179138,
      "loss": 2.4764,
      "step": 33540
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.9604061841964722,
      "learning_rate": 0.0005691692689686505,
      "loss": 2.3549,
      "step": 33541
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.7709290981292725,
      "learning_rate": 0.0005691674627081193,
      "loss": 2.2055,
      "step": 33542
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6456505060195923,
      "learning_rate": 0.0005691656563975449,
      "loss": 2.4394,
      "step": 33543
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.7117475271224976,
      "learning_rate": 0.0005691638500369274,
      "loss": 2.3675,
      "step": 33544
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.7137454152107239,
      "learning_rate": 0.0005691620436262673,
      "loss": 2.3032,
      "step": 33545
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.7508400082588196,
      "learning_rate": 0.0005691602371655648,
      "loss": 2.4052,
      "step": 33546
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6729004979133606,
      "learning_rate": 0.0005691584306548205,
      "loss": 2.2766,
      "step": 33547
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.8343063592910767,
      "learning_rate": 0.0005691566240940344,
      "loss": 2.2828,
      "step": 33548
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6545051336288452,
      "learning_rate": 0.0005691548174832072,
      "loss": 2.3754,
      "step": 33549
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6381449103355408,
      "learning_rate": 0.0005691530108223389,
      "loss": 2.3623,
      "step": 33550
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6232459545135498,
      "learning_rate": 0.00056915120411143,
      "loss": 2.4952,
      "step": 33551
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.7889977097511292,
      "learning_rate": 0.0005691493973504808,
      "loss": 2.1832,
      "step": 33552
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.002776861190796,
      "learning_rate": 0.0005691475905394916,
      "loss": 2.273,
      "step": 33553
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.5957300066947937,
      "learning_rate": 0.0005691457836784628,
      "loss": 2.0757,
      "step": 33554
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.7451996803283691,
      "learning_rate": 0.0005691439767673949,
      "loss": 2.3443,
      "step": 33555
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6931688785552979,
      "learning_rate": 0.0005691421698062879,
      "loss": 2.187,
      "step": 33556
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6639559268951416,
      "learning_rate": 0.0005691403627951422,
      "loss": 2.34,
      "step": 33557
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6161927580833435,
      "learning_rate": 0.0005691385557339584,
      "loss": 2.3086,
      "step": 33558
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.748290479183197,
      "learning_rate": 0.0005691367486227365,
      "loss": 2.2004,
      "step": 33559
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.637117862701416,
      "learning_rate": 0.0005691349414614771,
      "loss": 2.1961,
      "step": 33560
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6643804311752319,
      "learning_rate": 0.0005691331342501804,
      "loss": 2.1668,
      "step": 33561
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6521227359771729,
      "learning_rate": 0.0005691313269888469,
      "loss": 2.3681,
      "step": 33562
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6643301844596863,
      "learning_rate": 0.0005691295196774766,
      "loss": 2.3812,
      "step": 33563
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.7133492231369019,
      "learning_rate": 0.0005691277123160702,
      "loss": 2.3644,
      "step": 33564
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6421463489532471,
      "learning_rate": 0.0005691259049046278,
      "loss": 2.2059,
      "step": 33565
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.7133013010025024,
      "learning_rate": 0.0005691240974431497,
      "loss": 2.5054,
      "step": 33566
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.7130225896835327,
      "learning_rate": 0.0005691222899316366,
      "loss": 2.2622,
      "step": 33567
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6363502144813538,
      "learning_rate": 0.0005691204823700884,
      "loss": 2.2859,
      "step": 33568
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.5758503079414368,
      "learning_rate": 0.0005691186747585058,
      "loss": 2.1545,
      "step": 33569
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6970293521881104,
      "learning_rate": 0.0005691168670968888,
      "loss": 2.2162,
      "step": 33570
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.721691370010376,
      "learning_rate": 0.000569115059385238,
      "loss": 2.3086,
      "step": 33571
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6773955225944519,
      "learning_rate": 0.0005691132516235535,
      "loss": 2.5802,
      "step": 33572
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6482338905334473,
      "learning_rate": 0.0005691114438118359,
      "loss": 2.1209,
      "step": 33573
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6814771890640259,
      "learning_rate": 0.0005691096359500853,
      "loss": 2.5214,
      "step": 33574
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6388270854949951,
      "learning_rate": 0.0005691078280383021,
      "loss": 2.1592,
      "step": 33575
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6212226748466492,
      "learning_rate": 0.0005691060200764868,
      "loss": 2.2905,
      "step": 33576
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.7214416265487671,
      "learning_rate": 0.0005691042120646396,
      "loss": 2.4407,
      "step": 33577
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6090372204780579,
      "learning_rate": 0.0005691024040027608,
      "loss": 2.1683,
      "step": 33578
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6469995379447937,
      "learning_rate": 0.0005691005958908508,
      "loss": 2.2153,
      "step": 33579
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.5828123092651367,
      "learning_rate": 0.0005690987877289099,
      "loss": 2.2774,
      "step": 33580
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.64048832654953,
      "learning_rate": 0.0005690969795169385,
      "loss": 2.3585,
      "step": 33581
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6637582182884216,
      "learning_rate": 0.0005690951712549368,
      "loss": 2.3756,
      "step": 33582
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.7061551213264465,
      "learning_rate": 0.0005690933629429053,
      "loss": 2.3639,
      "step": 33583
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.694983184337616,
      "learning_rate": 0.0005690915545808443,
      "loss": 2.2597,
      "step": 33584
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.7730859518051147,
      "learning_rate": 0.000569089746168754,
      "loss": 2.2763,
      "step": 33585
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6304630637168884,
      "learning_rate": 0.0005690879377066349,
      "loss": 2.1465,
      "step": 33586
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6693868637084961,
      "learning_rate": 0.0005690861291944873,
      "loss": 2.4506,
      "step": 33587
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.7028031945228577,
      "learning_rate": 0.0005690843206323114,
      "loss": 2.3151,
      "step": 33588
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6798774003982544,
      "learning_rate": 0.0005690825120201078,
      "loss": 2.2565,
      "step": 33589
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.7162624001502991,
      "learning_rate": 0.0005690807033578765,
      "loss": 2.3978,
      "step": 33590
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.7420763969421387,
      "learning_rate": 0.0005690788946456181,
      "loss": 2.3023,
      "step": 33591
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.7148808240890503,
      "learning_rate": 0.000569077085883333,
      "loss": 2.3871,
      "step": 33592
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6357962489128113,
      "learning_rate": 0.0005690752770710212,
      "loss": 2.2003,
      "step": 33593
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.7235339879989624,
      "learning_rate": 0.0005690734682086833,
      "loss": 2.3618,
      "step": 33594
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6512503623962402,
      "learning_rate": 0.0005690716592963196,
      "loss": 2.3633,
      "step": 33595
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6710609197616577,
      "learning_rate": 0.0005690698503339303,
      "loss": 2.3753,
      "step": 33596
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.7475990056991577,
      "learning_rate": 0.0005690680413215159,
      "loss": 2.3356,
      "step": 33597
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.8680306673049927,
      "learning_rate": 0.0005690662322590767,
      "loss": 2.4184,
      "step": 33598
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6478850841522217,
      "learning_rate": 0.0005690644231466129,
      "loss": 2.4695,
      "step": 33599
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.7284267544746399,
      "learning_rate": 0.000569062613984125,
      "loss": 2.3294,
      "step": 33600
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6413936614990234,
      "learning_rate": 0.0005690608047716134,
      "loss": 2.1841,
      "step": 33601
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6346105337142944,
      "learning_rate": 0.0005690589955090782,
      "loss": 2.2495,
      "step": 33602
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.5892882347106934,
      "learning_rate": 0.0005690571861965198,
      "loss": 2.1335,
      "step": 33603
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.686960756778717,
      "learning_rate": 0.0005690553768339387,
      "loss": 2.3522,
      "step": 33604
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6840874552726746,
      "learning_rate": 0.0005690535674213351,
      "loss": 2.3231,
      "step": 33605
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6599024534225464,
      "learning_rate": 0.0005690517579587094,
      "loss": 2.286,
      "step": 33606
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6830524206161499,
      "learning_rate": 0.0005690499484460618,
      "loss": 2.2682,
      "step": 33607
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.7194493412971497,
      "learning_rate": 0.0005690481388833927,
      "loss": 2.1883,
      "step": 33608
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6167882084846497,
      "learning_rate": 0.0005690463292707026,
      "loss": 2.3093,
      "step": 33609
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6355524659156799,
      "learning_rate": 0.0005690445196079916,
      "loss": 2.1606,
      "step": 33610
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6504687666893005,
      "learning_rate": 0.0005690427098952603,
      "loss": 2.3303,
      "step": 33611
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6750519871711731,
      "learning_rate": 0.0005690409001325087,
      "loss": 2.2296,
      "step": 33612
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6593489646911621,
      "learning_rate": 0.0005690390903197374,
      "loss": 2.3388,
      "step": 33613
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.7260294556617737,
      "learning_rate": 0.0005690372804569467,
      "loss": 2.5911,
      "step": 33614
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6582414507865906,
      "learning_rate": 0.0005690354705441368,
      "loss": 2.3902,
      "step": 33615
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6253514289855957,
      "learning_rate": 0.0005690336605813083,
      "loss": 2.2373,
      "step": 33616
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6046665906906128,
      "learning_rate": 0.0005690318505684613,
      "loss": 2.507,
      "step": 33617
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6783434152603149,
      "learning_rate": 0.000569030040505596,
      "loss": 2.3018,
      "step": 33618
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6500803828239441,
      "learning_rate": 0.0005690282303927131,
      "loss": 2.2108,
      "step": 33619
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6918622851371765,
      "learning_rate": 0.0005690264202298128,
      "loss": 2.2877,
      "step": 33620
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6156821846961975,
      "learning_rate": 0.0005690246100168954,
      "loss": 2.1,
      "step": 33621
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.7179767489433289,
      "learning_rate": 0.0005690227997539613,
      "loss": 2.4046,
      "step": 33622
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.772849977016449,
      "learning_rate": 0.0005690209894410106,
      "loss": 2.4744,
      "step": 33623
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6853955388069153,
      "learning_rate": 0.0005690191790780439,
      "loss": 2.2935,
      "step": 33624
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6697778105735779,
      "learning_rate": 0.0005690173686650615,
      "loss": 2.2298,
      "step": 33625
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6622191071510315,
      "learning_rate": 0.0005690155582020636,
      "loss": 2.3806,
      "step": 33626
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6341180801391602,
      "learning_rate": 0.0005690137476890507,
      "loss": 2.3026,
      "step": 33627
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6715442538261414,
      "learning_rate": 0.0005690119371260231,
      "loss": 2.6459,
      "step": 33628
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6536057591438293,
      "learning_rate": 0.0005690101265129811,
      "loss": 2.3216,
      "step": 33629
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.7222805619239807,
      "learning_rate": 0.000569008315849925,
      "loss": 2.3353,
      "step": 33630
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.5703774094581604,
      "learning_rate": 0.0005690065051368551,
      "loss": 2.2265,
      "step": 33631
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6356734037399292,
      "learning_rate": 0.0005690046943737719,
      "loss": 2.3253,
      "step": 33632
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6448776125907898,
      "learning_rate": 0.0005690028835606758,
      "loss": 2.2509,
      "step": 33633
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6432698965072632,
      "learning_rate": 0.0005690010726975668,
      "loss": 2.4868,
      "step": 33634
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.695624053478241,
      "learning_rate": 0.0005689992617844453,
      "loss": 2.3833,
      "step": 33635
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6300537586212158,
      "learning_rate": 0.000568997450821312,
      "loss": 2.3828,
      "step": 33636
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6606866121292114,
      "learning_rate": 0.0005689956398081669,
      "loss": 2.2302,
      "step": 33637
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6908755302429199,
      "learning_rate": 0.0005689938287450105,
      "loss": 2.1763,
      "step": 33638
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6856730580329895,
      "learning_rate": 0.0005689920176318429,
      "loss": 2.3376,
      "step": 33639
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.7223724722862244,
      "learning_rate": 0.0005689902064686648,
      "loss": 2.3995,
      "step": 33640
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6545154452323914,
      "learning_rate": 0.0005689883952554762,
      "loss": 2.0765,
      "step": 33641
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6382490396499634,
      "learning_rate": 0.0005689865839922777,
      "loss": 2.4812,
      "step": 33642
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6686307787895203,
      "learning_rate": 0.0005689847726790694,
      "loss": 2.3407,
      "step": 33643
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.7111655473709106,
      "learning_rate": 0.0005689829613158518,
      "loss": 2.3189,
      "step": 33644
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.7316573858261108,
      "learning_rate": 0.0005689811499026252,
      "loss": 2.2958,
      "step": 33645
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6897193789482117,
      "learning_rate": 0.00056897933843939,
      "loss": 2.2896,
      "step": 33646
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6922396421432495,
      "learning_rate": 0.0005689775269261463,
      "loss": 2.3318,
      "step": 33647
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.7061281800270081,
      "learning_rate": 0.0005689757153628947,
      "loss": 2.5276,
      "step": 33648
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.650638222694397,
      "learning_rate": 0.0005689739037496354,
      "loss": 2.2619,
      "step": 33649
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6716392636299133,
      "learning_rate": 0.0005689720920863689,
      "loss": 2.3882,
      "step": 33650
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6419953107833862,
      "learning_rate": 0.0005689702803730952,
      "loss": 2.1812,
      "step": 33651
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6684615015983582,
      "learning_rate": 0.000568968468609815,
      "loss": 2.2696,
      "step": 33652
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.652523398399353,
      "learning_rate": 0.0005689666567965285,
      "loss": 2.4589,
      "step": 33653
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6732662916183472,
      "learning_rate": 0.0005689648449332359,
      "loss": 2.3916,
      "step": 33654
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6328821182250977,
      "learning_rate": 0.0005689630330199378,
      "loss": 2.3122,
      "step": 33655
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.623914897441864,
      "learning_rate": 0.0005689612210566342,
      "loss": 2.125,
      "step": 33656
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.7200480103492737,
      "learning_rate": 0.0005689594090433259,
      "loss": 2.273,
      "step": 33657
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6824347972869873,
      "learning_rate": 0.0005689575969800128,
      "loss": 2.287,
      "step": 33658
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.7248309254646301,
      "learning_rate": 0.0005689557848666954,
      "loss": 2.3502,
      "step": 33659
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.7338035702705383,
      "learning_rate": 0.0005689539727033741,
      "loss": 2.2302,
      "step": 33660
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6100096106529236,
      "learning_rate": 0.000568952160490049,
      "loss": 2.1761,
      "step": 33661
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.7202712893486023,
      "learning_rate": 0.0005689503482267209,
      "loss": 2.3053,
      "step": 33662
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.7120503187179565,
      "learning_rate": 0.0005689485359133897,
      "loss": 2.3121,
      "step": 33663
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6636177897453308,
      "learning_rate": 0.0005689467235500559,
      "loss": 2.208,
      "step": 33664
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6129133105278015,
      "learning_rate": 0.0005689449111367198,
      "loss": 1.9117,
      "step": 33665
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6496002078056335,
      "learning_rate": 0.0005689430986733818,
      "loss": 2.4612,
      "step": 33666
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6304978132247925,
      "learning_rate": 0.0005689412861600423,
      "loss": 2.3395,
      "step": 33667
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6478864550590515,
      "learning_rate": 0.0005689394735967013,
      "loss": 2.3684,
      "step": 33668
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6188907027244568,
      "learning_rate": 0.0005689376609833596,
      "loss": 2.1724,
      "step": 33669
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.7306366562843323,
      "learning_rate": 0.0005689358483200172,
      "loss": 2.4163,
      "step": 33670
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6731482148170471,
      "learning_rate": 0.0005689340356066745,
      "loss": 2.1067,
      "step": 33671
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6000182628631592,
      "learning_rate": 0.0005689322228433321,
      "loss": 2.1487,
      "step": 33672
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.7205415964126587,
      "learning_rate": 0.0005689304100299899,
      "loss": 2.2174,
      "step": 33673
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6566365957260132,
      "learning_rate": 0.0005689285971666486,
      "loss": 2.292,
      "step": 33674
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6558485627174377,
      "learning_rate": 0.0005689267842533083,
      "loss": 2.1847,
      "step": 33675
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6664535999298096,
      "learning_rate": 0.0005689249712899694,
      "loss": 2.3143,
      "step": 33676
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.63728928565979,
      "learning_rate": 0.0005689231582766325,
      "loss": 2.1724,
      "step": 33677
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6039085388183594,
      "learning_rate": 0.0005689213452132975,
      "loss": 2.399,
      "step": 33678
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6339864730834961,
      "learning_rate": 0.0005689195320999651,
      "loss": 2.1431,
      "step": 33679
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6127442717552185,
      "learning_rate": 0.0005689177189366353,
      "loss": 2.2454,
      "step": 33680
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6267457604408264,
      "learning_rate": 0.0005689159057233087,
      "loss": 2.3057,
      "step": 33681
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6755617260932922,
      "learning_rate": 0.0005689140924599855,
      "loss": 2.2824,
      "step": 33682
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6528506278991699,
      "learning_rate": 0.0005689122791466663,
      "loss": 2.3482,
      "step": 33683
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.621207058429718,
      "learning_rate": 0.000568910465783351,
      "loss": 2.1331,
      "step": 33684
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6103946566581726,
      "learning_rate": 0.0005689086523700403,
      "loss": 2.1143,
      "step": 33685
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6415345668792725,
      "learning_rate": 0.0005689068389067343,
      "loss": 2.0285,
      "step": 33686
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6513963937759399,
      "learning_rate": 0.0005689050253934335,
      "loss": 2.3087,
      "step": 33687
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6768801808357239,
      "learning_rate": 0.0005689032118301382,
      "loss": 2.2845,
      "step": 33688
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6690851449966431,
      "learning_rate": 0.0005689013982168487,
      "loss": 2.4216,
      "step": 33689
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6605457663536072,
      "learning_rate": 0.0005688995845535654,
      "loss": 2.1321,
      "step": 33690
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6656420826911926,
      "learning_rate": 0.0005688977708402886,
      "loss": 2.0805,
      "step": 33691
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6677218675613403,
      "learning_rate": 0.0005688959570770185,
      "loss": 2.3925,
      "step": 33692
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6431638598442078,
      "learning_rate": 0.0005688941432637557,
      "loss": 2.2832,
      "step": 33693
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6171888709068298,
      "learning_rate": 0.0005688923294005004,
      "loss": 2.4304,
      "step": 33694
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.681509256362915,
      "learning_rate": 0.0005688905154872528,
      "loss": 2.2913,
      "step": 33695
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6377732157707214,
      "learning_rate": 0.0005688887015240135,
      "loss": 2.1572,
      "step": 33696
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.7104106545448303,
      "learning_rate": 0.0005688868875107826,
      "loss": 2.4794,
      "step": 33697
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6211843490600586,
      "learning_rate": 0.0005688850734475606,
      "loss": 2.3151,
      "step": 33698
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6957201957702637,
      "learning_rate": 0.0005688832593343479,
      "loss": 2.4388,
      "step": 33699
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6316558718681335,
      "learning_rate": 0.0005688814451711446,
      "loss": 2.1584,
      "step": 33700
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6658332943916321,
      "learning_rate": 0.0005688796309579513,
      "loss": 1.835,
      "step": 33701
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6827359795570374,
      "learning_rate": 0.000568877816694768,
      "loss": 2.3398,
      "step": 33702
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.8540219068527222,
      "learning_rate": 0.0005688760023815953,
      "loss": 2.1632,
      "step": 33703
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6709910035133362,
      "learning_rate": 0.0005688741880184336,
      "loss": 2.4365,
      "step": 33704
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.7916355133056641,
      "learning_rate": 0.000568872373605283,
      "loss": 2.1895,
      "step": 33705
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6437349915504456,
      "learning_rate": 0.000568870559142144,
      "loss": 2.3135,
      "step": 33706
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6765745878219604,
      "learning_rate": 0.0005688687446290169,
      "loss": 2.2477,
      "step": 33707
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.700940728187561,
      "learning_rate": 0.000568866930065902,
      "loss": 2.5076,
      "step": 33708
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.8355583548545837,
      "learning_rate": 0.0005688651154527997,
      "loss": 2.5011,
      "step": 33709
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6831992268562317,
      "learning_rate": 0.0005688633007897103,
      "loss": 2.2317,
      "step": 33710
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.7091566324234009,
      "learning_rate": 0.0005688614860766341,
      "loss": 2.321,
      "step": 33711
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.7597169280052185,
      "learning_rate": 0.0005688596713135715,
      "loss": 2.1942,
      "step": 33712
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.7022892832756042,
      "learning_rate": 0.0005688578565005227,
      "loss": 2.3345,
      "step": 33713
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6558612585067749,
      "learning_rate": 0.0005688560416374882,
      "loss": 2.3083,
      "step": 33714
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.7185836434364319,
      "learning_rate": 0.0005688542267244683,
      "loss": 2.4457,
      "step": 33715
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.7552570104598999,
      "learning_rate": 0.0005688524117614634,
      "loss": 2.1376,
      "step": 33716
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6603764295578003,
      "learning_rate": 0.0005688505967484736,
      "loss": 2.376,
      "step": 33717
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.7034316062927246,
      "learning_rate": 0.0005688487816854996,
      "loss": 2.3379,
      "step": 33718
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.99333655834198,
      "learning_rate": 0.0005688469665725414,
      "loss": 2.3578,
      "step": 33719
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6734426617622375,
      "learning_rate": 0.0005688451514095995,
      "loss": 2.2386,
      "step": 33720
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.7205378413200378,
      "learning_rate": 0.0005688433361966743,
      "loss": 2.3433,
      "step": 33721
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.640265703201294,
      "learning_rate": 0.0005688415209337659,
      "loss": 2.0919,
      "step": 33722
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.7512133717536926,
      "learning_rate": 0.000568839705620875,
      "loss": 2.3192,
      "step": 33723
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.605973482131958,
      "learning_rate": 0.0005688378902580015,
      "loss": 2.2332,
      "step": 33724
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.7401033639907837,
      "learning_rate": 0.0005688360748451461,
      "loss": 2.2267,
      "step": 33725
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.7210761308670044,
      "learning_rate": 0.0005688342593823089,
      "loss": 2.3291,
      "step": 33726
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6813453435897827,
      "learning_rate": 0.0005688324438694903,
      "loss": 2.2654,
      "step": 33727
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.7757946252822876,
      "learning_rate": 0.0005688306283066908,
      "loss": 2.417,
      "step": 33728
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.688269317150116,
      "learning_rate": 0.0005688288126939107,
      "loss": 2.3512,
      "step": 33729
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6407666802406311,
      "learning_rate": 0.0005688269970311501,
      "loss": 2.1965,
      "step": 33730
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6454861164093018,
      "learning_rate": 0.0005688251813184094,
      "loss": 2.3557,
      "step": 33731
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6115732789039612,
      "learning_rate": 0.0005688233655556892,
      "loss": 2.1241,
      "step": 33732
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6612826585769653,
      "learning_rate": 0.0005688215497429896,
      "loss": 2.3632,
      "step": 33733
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6328810453414917,
      "learning_rate": 0.0005688197338803109,
      "loss": 2.1439,
      "step": 33734
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6142268776893616,
      "learning_rate": 0.0005688179179676537,
      "loss": 2.3673,
      "step": 33735
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6392490863800049,
      "learning_rate": 0.0005688161020050181,
      "loss": 2.33,
      "step": 33736
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6038960218429565,
      "learning_rate": 0.0005688142859924046,
      "loss": 2.3628,
      "step": 33737
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6664600968360901,
      "learning_rate": 0.0005688124699298133,
      "loss": 2.1758,
      "step": 33738
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6694104075431824,
      "learning_rate": 0.0005688106538172447,
      "loss": 2.4099,
      "step": 33739
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.7015334963798523,
      "learning_rate": 0.0005688088376546993,
      "loss": 2.3082,
      "step": 33740
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6932795643806458,
      "learning_rate": 0.0005688070214421771,
      "loss": 2.3861,
      "step": 33741
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6594898700714111,
      "learning_rate": 0.0005688052051796787,
      "loss": 2.3301,
      "step": 33742
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6510486602783203,
      "learning_rate": 0.0005688033888672043,
      "loss": 2.2637,
      "step": 33743
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.7851746678352356,
      "learning_rate": 0.0005688015725047542,
      "loss": 2.0884,
      "step": 33744
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.7451440691947937,
      "learning_rate": 0.0005687997560923288,
      "loss": 2.0596,
      "step": 33745
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6349549293518066,
      "learning_rate": 0.0005687979396299286,
      "loss": 2.3861,
      "step": 33746
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.7431861162185669,
      "learning_rate": 0.0005687961231175537,
      "loss": 2.4592,
      "step": 33747
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.7786773443222046,
      "learning_rate": 0.0005687943065552045,
      "loss": 2.2152,
      "step": 33748
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6679547429084778,
      "learning_rate": 0.0005687924899428814,
      "loss": 2.3019,
      "step": 33749
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.7446616291999817,
      "learning_rate": 0.0005687906732805847,
      "loss": 2.1469,
      "step": 33750
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.7270280122756958,
      "learning_rate": 0.0005687888565683147,
      "loss": 2.3135,
      "step": 33751
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6486196517944336,
      "learning_rate": 0.0005687870398060719,
      "loss": 2.1182,
      "step": 33752
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.7034943699836731,
      "learning_rate": 0.0005687852229938564,
      "loss": 2.3098,
      "step": 33753
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6233118772506714,
      "learning_rate": 0.0005687834061316687,
      "loss": 2.3878,
      "step": 33754
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.7327359914779663,
      "learning_rate": 0.0005687815892195091,
      "loss": 2.3147,
      "step": 33755
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6485295295715332,
      "learning_rate": 0.000568779772257378,
      "loss": 2.3782,
      "step": 33756
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6270470023155212,
      "learning_rate": 0.0005687779552452755,
      "loss": 2.506,
      "step": 33757
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6503524780273438,
      "learning_rate": 0.0005687761381832022,
      "loss": 2.3896,
      "step": 33758
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6072289943695068,
      "learning_rate": 0.0005687743210711583,
      "loss": 2.1949,
      "step": 33759
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6518279910087585,
      "learning_rate": 0.0005687725039091442,
      "loss": 2.3972,
      "step": 33760
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6436772346496582,
      "learning_rate": 0.0005687706866971602,
      "loss": 2.4564,
      "step": 33761
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6959797739982605,
      "learning_rate": 0.0005687688694352068,
      "loss": 2.2922,
      "step": 33762
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6121798157691956,
      "learning_rate": 0.000568767052123284,
      "loss": 2.3824,
      "step": 33763
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.5817094445228577,
      "learning_rate": 0.0005687652347613924,
      "loss": 1.8893,
      "step": 33764
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6646634936332703,
      "learning_rate": 0.0005687634173495322,
      "loss": 2.0954,
      "step": 33765
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6546993255615234,
      "learning_rate": 0.000568761599887704,
      "loss": 2.3015,
      "step": 33766
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.688288152217865,
      "learning_rate": 0.0005687597823759078,
      "loss": 2.3937,
      "step": 33767
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6696935892105103,
      "learning_rate": 0.0005687579648141441,
      "loss": 2.4158,
      "step": 33768
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.7084726691246033,
      "learning_rate": 0.0005687561472024133,
      "loss": 2.166,
      "step": 33769
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6687182784080505,
      "learning_rate": 0.0005687543295407156,
      "loss": 2.2392,
      "step": 33770
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.7449108362197876,
      "learning_rate": 0.0005687525118290513,
      "loss": 2.3312,
      "step": 33771
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6503695249557495,
      "learning_rate": 0.000568750694067421,
      "loss": 2.3322,
      "step": 33772
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6589252948760986,
      "learning_rate": 0.0005687488762558247,
      "loss": 2.1739,
      "step": 33773
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6072435975074768,
      "learning_rate": 0.000568747058394263,
      "loss": 1.9307,
      "step": 33774
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.7248337268829346,
      "learning_rate": 0.0005687452404827362,
      "loss": 2.2208,
      "step": 33775
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6693934798240662,
      "learning_rate": 0.0005687434225212446,
      "loss": 2.1656,
      "step": 33776
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6679741740226746,
      "learning_rate": 0.0005687416045097885,
      "loss": 2.096,
      "step": 33777
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6429283618927002,
      "learning_rate": 0.0005687397864483681,
      "loss": 2.1853,
      "step": 33778
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6072399020195007,
      "learning_rate": 0.0005687379683369841,
      "loss": 2.1703,
      "step": 33779
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6509642601013184,
      "learning_rate": 0.0005687361501756365,
      "loss": 2.3581,
      "step": 33780
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6651714444160461,
      "learning_rate": 0.0005687343319643259,
      "loss": 2.2183,
      "step": 33781
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6334033012390137,
      "learning_rate": 0.0005687325137030524,
      "loss": 2.2373,
      "step": 33782
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6194397807121277,
      "learning_rate": 0.0005687306953918165,
      "loss": 2.1545,
      "step": 33783
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6791542768478394,
      "learning_rate": 0.0005687288770306185,
      "loss": 2.3945,
      "step": 33784
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6473374962806702,
      "learning_rate": 0.0005687270586194588,
      "loss": 2.1681,
      "step": 33785
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6553272604942322,
      "learning_rate": 0.0005687252401583376,
      "loss": 2.2355,
      "step": 33786
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6779212355613708,
      "learning_rate": 0.0005687234216472553,
      "loss": 2.2253,
      "step": 33787
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6823680996894836,
      "learning_rate": 0.0005687216030862123,
      "loss": 2.3406,
      "step": 33788
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6443562507629395,
      "learning_rate": 0.0005687197844752088,
      "loss": 2.2994,
      "step": 33789
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.7218703627586365,
      "learning_rate": 0.0005687179658142453,
      "loss": 2.2382,
      "step": 33790
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.7492017149925232,
      "learning_rate": 0.000568716147103322,
      "loss": 2.5043,
      "step": 33791
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6802859306335449,
      "learning_rate": 0.0005687143283424393,
      "loss": 2.1648,
      "step": 33792
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6554695963859558,
      "learning_rate": 0.0005687125095315974,
      "loss": 2.4184,
      "step": 33793
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6443294882774353,
      "learning_rate": 0.000568710690670797,
      "loss": 2.1529,
      "step": 33794
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6848245859146118,
      "learning_rate": 0.0005687088717600381,
      "loss": 2.2081,
      "step": 33795
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6215364933013916,
      "learning_rate": 0.0005687070527993211,
      "loss": 2.4433,
      "step": 33796
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6022897362709045,
      "learning_rate": 0.0005687052337886465,
      "loss": 2.5869,
      "step": 33797
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6952463984489441,
      "learning_rate": 0.0005687034147280146,
      "loss": 2.4192,
      "step": 33798
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6285470724105835,
      "learning_rate": 0.0005687015956174254,
      "loss": 2.4141,
      "step": 33799
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6576305031776428,
      "learning_rate": 0.0005686997764568797,
      "loss": 2.3477,
      "step": 33800
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.7413997650146484,
      "learning_rate": 0.0005686979572463776,
      "loss": 2.2333,
      "step": 33801
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6410600543022156,
      "learning_rate": 0.0005686961379859195,
      "loss": 2.4265,
      "step": 33802
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.7105410099029541,
      "learning_rate": 0.0005686943186755056,
      "loss": 2.3634,
      "step": 33803
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6549822688102722,
      "learning_rate": 0.0005686924993151364,
      "loss": 2.2724,
      "step": 33804
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6609894633293152,
      "learning_rate": 0.0005686906799048123,
      "loss": 2.2339,
      "step": 33805
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6683440208435059,
      "learning_rate": 0.0005686888604445335,
      "loss": 2.3368,
      "step": 33806
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6743080019950867,
      "learning_rate": 0.0005686870409343003,
      "loss": 2.2261,
      "step": 33807
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6505542397499084,
      "learning_rate": 0.0005686852213741131,
      "loss": 2.4357,
      "step": 33808
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6978282332420349,
      "learning_rate": 0.0005686834017639723,
      "loss": 2.4039,
      "step": 33809
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.7186084389686584,
      "learning_rate": 0.0005686815821038781,
      "loss": 2.1263,
      "step": 33810
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6581296324729919,
      "learning_rate": 0.0005686797623938311,
      "loss": 2.0026,
      "step": 33811
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6620804667472839,
      "learning_rate": 0.0005686779426338313,
      "loss": 2.43,
      "step": 33812
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.8229654431343079,
      "learning_rate": 0.0005686761228238793,
      "loss": 2.3816,
      "step": 33813
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.7207303047180176,
      "learning_rate": 0.0005686743029639752,
      "loss": 2.1953,
      "step": 33814
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6277611255645752,
      "learning_rate": 0.0005686724830541195,
      "loss": 2.3613,
      "step": 33815
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.7022078633308411,
      "learning_rate": 0.0005686706630943127,
      "loss": 2.1971,
      "step": 33816
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.7243244051933289,
      "learning_rate": 0.0005686688430845548,
      "loss": 2.4699,
      "step": 33817
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.7117270231246948,
      "learning_rate": 0.0005686670230248463,
      "loss": 1.9982,
      "step": 33818
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6748045086860657,
      "learning_rate": 0.0005686652029151876,
      "loss": 2.172,
      "step": 33819
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6829957962036133,
      "learning_rate": 0.0005686633827555788,
      "loss": 2.1592,
      "step": 33820
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.7014092206954956,
      "learning_rate": 0.0005686615625460205,
      "loss": 2.5221,
      "step": 33821
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6610058546066284,
      "learning_rate": 0.000568659742286513,
      "loss": 2.3516,
      "step": 33822
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6690674424171448,
      "learning_rate": 0.0005686579219770565,
      "loss": 2.39,
      "step": 33823
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6841062307357788,
      "learning_rate": 0.0005686561016176515,
      "loss": 2.16,
      "step": 33824
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.7654158473014832,
      "learning_rate": 0.0005686542812082982,
      "loss": 2.2174,
      "step": 33825
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6204593181610107,
      "learning_rate": 0.000568652460748997,
      "loss": 2.2639,
      "step": 33826
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6551927924156189,
      "learning_rate": 0.0005686506402397483,
      "loss": 2.1808,
      "step": 33827
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.7165198922157288,
      "learning_rate": 0.0005686488196805523,
      "loss": 2.3711,
      "step": 33828
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.7380436658859253,
      "learning_rate": 0.0005686469990714095,
      "loss": 2.1728,
      "step": 33829
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6859889030456543,
      "learning_rate": 0.0005686451784123201,
      "loss": 2.2806,
      "step": 33830
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6128023862838745,
      "learning_rate": 0.0005686433577032844,
      "loss": 2.1733,
      "step": 33831
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6494951844215393,
      "learning_rate": 0.000568641536944303,
      "loss": 2.1667,
      "step": 33832
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6160128712654114,
      "learning_rate": 0.000568639716135376,
      "loss": 2.2156,
      "step": 33833
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6328158974647522,
      "learning_rate": 0.0005686378952765037,
      "loss": 2.3891,
      "step": 33834
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.739634096622467,
      "learning_rate": 0.0005686360743676867,
      "loss": 2.3548,
      "step": 33835
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6482198238372803,
      "learning_rate": 0.0005686342534089251,
      "loss": 2.502,
      "step": 33836
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.5981199741363525,
      "learning_rate": 0.0005686324324002193,
      "loss": 2.3381,
      "step": 33837
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.7013772130012512,
      "learning_rate": 0.0005686306113415697,
      "loss": 2.1853,
      "step": 33838
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.66254723072052,
      "learning_rate": 0.0005686287902329767,
      "loss": 2.4603,
      "step": 33839
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.635362982749939,
      "learning_rate": 0.0005686269690744403,
      "loss": 2.3466,
      "step": 33840
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.7039538025856018,
      "learning_rate": 0.0005686251478659612,
      "loss": 2.1999,
      "step": 33841
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.7399082779884338,
      "learning_rate": 0.0005686233266075396,
      "loss": 2.1328,
      "step": 33842
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6067267060279846,
      "learning_rate": 0.0005686215052991759,
      "loss": 2.204,
      "step": 33843
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6842986941337585,
      "learning_rate": 0.0005686196839408704,
      "loss": 2.1083,
      "step": 33844
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.7698951959609985,
      "learning_rate": 0.0005686178625326233,
      "loss": 2.1498,
      "step": 33845
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.7462179064750671,
      "learning_rate": 0.0005686160410744353,
      "loss": 2.0979,
      "step": 33846
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6635816097259521,
      "learning_rate": 0.0005686142195663064,
      "loss": 2.5373,
      "step": 33847
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.7317016124725342,
      "learning_rate": 0.000568612398008237,
      "loss": 2.22,
      "step": 33848
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.7713143229484558,
      "learning_rate": 0.0005686105764002275,
      "loss": 2.3447,
      "step": 33849
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.714754045009613,
      "learning_rate": 0.0005686087547422782,
      "loss": 2.4903,
      "step": 33850
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6670452356338501,
      "learning_rate": 0.0005686069330343895,
      "loss": 2.3978,
      "step": 33851
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.7704976797103882,
      "learning_rate": 0.0005686051112765617,
      "loss": 2.2655,
      "step": 33852
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6860505938529968,
      "learning_rate": 0.0005686032894687951,
      "loss": 2.4756,
      "step": 33853
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.853354275226593,
      "learning_rate": 0.0005686014676110901,
      "loss": 2.307,
      "step": 33854
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.691580593585968,
      "learning_rate": 0.000568599645703447,
      "loss": 2.1517,
      "step": 33855
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.748364269733429,
      "learning_rate": 0.0005685978237458662,
      "loss": 2.2909,
      "step": 33856
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.669488251209259,
      "learning_rate": 0.000568596001738348,
      "loss": 2.1521,
      "step": 33857
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.7183117270469666,
      "learning_rate": 0.0005685941796808927,
      "loss": 2.4535,
      "step": 33858
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.7494966387748718,
      "learning_rate": 0.0005685923575735007,
      "loss": 2.4514,
      "step": 33859
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.5997403860092163,
      "learning_rate": 0.0005685905354161722,
      "loss": 2.561,
      "step": 33860
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.7075631022453308,
      "learning_rate": 0.0005685887132089078,
      "loss": 2.1039,
      "step": 33861
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6543291211128235,
      "learning_rate": 0.0005685868909517076,
      "loss": 2.4173,
      "step": 33862
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6317607164382935,
      "learning_rate": 0.0005685850686445722,
      "loss": 2.451,
      "step": 33863
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.5943480134010315,
      "learning_rate": 0.0005685832462875016,
      "loss": 2.4131,
      "step": 33864
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.7062545418739319,
      "learning_rate": 0.0005685814238804962,
      "loss": 2.4519,
      "step": 33865
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6375641822814941,
      "learning_rate": 0.0005685796014235566,
      "loss": 2.1906,
      "step": 33866
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6897320747375488,
      "learning_rate": 0.0005685777789166829,
      "loss": 2.0785,
      "step": 33867
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6352145671844482,
      "learning_rate": 0.0005685759563598756,
      "loss": 2.2209,
      "step": 33868
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.7147576212882996,
      "learning_rate": 0.000568574133753135,
      "loss": 2.235,
      "step": 33869
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.7453622221946716,
      "learning_rate": 0.0005685723110964612,
      "loss": 2.4143,
      "step": 33870
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6709615588188171,
      "learning_rate": 0.0005685704883898549,
      "loss": 2.1915,
      "step": 33871
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.701708197593689,
      "learning_rate": 0.0005685686656333164,
      "loss": 2.5004,
      "step": 33872
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6764970421791077,
      "learning_rate": 0.0005685668428268456,
      "loss": 2.4368,
      "step": 33873
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6918326616287231,
      "learning_rate": 0.0005685650199704434,
      "loss": 2.1509,
      "step": 33874
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.5950064659118652,
      "learning_rate": 0.0005685631970641098,
      "loss": 2.1726,
      "step": 33875
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6983931660652161,
      "learning_rate": 0.0005685613741078453,
      "loss": 2.3565,
      "step": 33876
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.7349680066108704,
      "learning_rate": 0.00056855955110165,
      "loss": 2.291,
      "step": 33877
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6329238414764404,
      "learning_rate": 0.0005685577280455246,
      "loss": 2.2715,
      "step": 33878
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6688233017921448,
      "learning_rate": 0.0005685559049394692,
      "loss": 2.3528,
      "step": 33879
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6802188754081726,
      "learning_rate": 0.000568554081783484,
      "loss": 2.3377,
      "step": 33880
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.7451556324958801,
      "learning_rate": 0.0005685522585775698,
      "loss": 2.3794,
      "step": 33881
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.7299973964691162,
      "learning_rate": 0.0005685504353217265,
      "loss": 2.416,
      "step": 33882
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6558232307434082,
      "learning_rate": 0.0005685486120159547,
      "loss": 2.2839,
      "step": 33883
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6824935078620911,
      "learning_rate": 0.0005685467886602546,
      "loss": 2.0378,
      "step": 33884
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.718282163143158,
      "learning_rate": 0.0005685449652546266,
      "loss": 2.1764,
      "step": 33885
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.7429079413414001,
      "learning_rate": 0.000568543141799071,
      "loss": 2.3876,
      "step": 33886
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.7224545478820801,
      "learning_rate": 0.0005685413182935881,
      "loss": 2.26,
      "step": 33887
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6252009868621826,
      "learning_rate": 0.0005685394947381783,
      "loss": 2.254,
      "step": 33888
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6827819347381592,
      "learning_rate": 0.0005685376711328422,
      "loss": 2.4429,
      "step": 33889
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.7259310483932495,
      "learning_rate": 0.0005685358474775797,
      "loss": 2.1819,
      "step": 33890
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6406369209289551,
      "learning_rate": 0.0005685340237723912,
      "loss": 2.1902,
      "step": 33891
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6438977122306824,
      "learning_rate": 0.0005685322000172772,
      "loss": 2.4607,
      "step": 33892
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.9946226477622986,
      "learning_rate": 0.0005685303762122382,
      "loss": 2.2604,
      "step": 33893
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.7699950933456421,
      "learning_rate": 0.0005685285523572741,
      "loss": 2.4477,
      "step": 33894
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.675877034664154,
      "learning_rate": 0.0005685267284523855,
      "loss": 2.4484,
      "step": 33895
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.712777853012085,
      "learning_rate": 0.0005685249044975728,
      "loss": 2.2926,
      "step": 33896
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.7082032561302185,
      "learning_rate": 0.0005685230804928363,
      "loss": 2.1387,
      "step": 33897
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.7875264286994934,
      "learning_rate": 0.0005685212564381761,
      "loss": 2.1902,
      "step": 33898
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6608676314353943,
      "learning_rate": 0.0005685194323335928,
      "loss": 2.4114,
      "step": 33899
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.7677172422409058,
      "learning_rate": 0.0005685176081790868,
      "loss": 2.2505,
      "step": 33900
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6378914713859558,
      "learning_rate": 0.0005685157839746582,
      "loss": 2.306,
      "step": 33901
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6787290573120117,
      "learning_rate": 0.0005685139597203074,
      "loss": 2.2809,
      "step": 33902
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6653850674629211,
      "learning_rate": 0.0005685121354160349,
      "loss": 2.4322,
      "step": 33903
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.816057026386261,
      "learning_rate": 0.0005685103110618409,
      "loss": 2.5254,
      "step": 33904
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.8042615652084351,
      "learning_rate": 0.0005685084866577257,
      "loss": 2.102,
      "step": 33905
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.7100601196289062,
      "learning_rate": 0.0005685066622036898,
      "loss": 2.303,
      "step": 33906
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.9697349071502686,
      "learning_rate": 0.0005685048376997333,
      "loss": 2.3063,
      "step": 33907
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6606914401054382,
      "learning_rate": 0.0005685030131458568,
      "loss": 2.3815,
      "step": 33908
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.9354897141456604,
      "learning_rate": 0.0005685011885420606,
      "loss": 2.3145,
      "step": 33909
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6183494925498962,
      "learning_rate": 0.0005684993638883448,
      "loss": 2.2977,
      "step": 33910
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.7379355430603027,
      "learning_rate": 0.00056849753918471,
      "loss": 2.3279,
      "step": 33911
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.7391715049743652,
      "learning_rate": 0.0005684957144311563,
      "loss": 2.3725,
      "step": 33912
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.681027352809906,
      "learning_rate": 0.0005684938896276843,
      "loss": 2.3836,
      "step": 33913
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.7304408550262451,
      "learning_rate": 0.0005684920647742942,
      "loss": 2.3201,
      "step": 33914
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.7316154837608337,
      "learning_rate": 0.0005684902398709863,
      "loss": 2.2594,
      "step": 33915
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6301367282867432,
      "learning_rate": 0.0005684884149177611,
      "loss": 2.1858,
      "step": 33916
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.7812976241111755,
      "learning_rate": 0.0005684865899146188,
      "loss": 2.4855,
      "step": 33917
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6376495361328125,
      "learning_rate": 0.0005684847648615597,
      "loss": 2.4774,
      "step": 33918
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6255210638046265,
      "learning_rate": 0.0005684829397585842,
      "loss": 2.0653,
      "step": 33919
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6449114680290222,
      "learning_rate": 0.0005684811146056928,
      "loss": 2.4753,
      "step": 33920
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6376368999481201,
      "learning_rate": 0.0005684792894028857,
      "loss": 2.4341,
      "step": 33921
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.7120148539543152,
      "learning_rate": 0.000568477464150163,
      "loss": 2.3322,
      "step": 33922
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6522119641304016,
      "learning_rate": 0.0005684756388475255,
      "loss": 2.3842,
      "step": 33923
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6819939017295837,
      "learning_rate": 0.0005684738134949732,
      "loss": 2.4513,
      "step": 33924
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6525633335113525,
      "learning_rate": 0.0005684719880925066,
      "loss": 2.5539,
      "step": 33925
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.5961601138114929,
      "learning_rate": 0.0005684701626401259,
      "loss": 2.0652,
      "step": 33926
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6884132027626038,
      "learning_rate": 0.0005684683371378317,
      "loss": 2.5171,
      "step": 33927
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6269963383674622,
      "learning_rate": 0.000568466511585624,
      "loss": 2.1676,
      "step": 33928
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6449739933013916,
      "learning_rate": 0.0005684646859835034,
      "loss": 2.2654,
      "step": 33929
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6952023506164551,
      "learning_rate": 0.0005684628603314701,
      "loss": 2.3843,
      "step": 33930
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6970304846763611,
      "learning_rate": 0.0005684610346295244,
      "loss": 2.3761,
      "step": 33931
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6582965850830078,
      "learning_rate": 0.0005684592088776669,
      "loss": 2.3395,
      "step": 33932
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6785601377487183,
      "learning_rate": 0.0005684573830758978,
      "loss": 2.5192,
      "step": 33933
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6170819401741028,
      "learning_rate": 0.0005684555572242172,
      "loss": 2.3051,
      "step": 33934
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.7514334917068481,
      "learning_rate": 0.0005684537313226257,
      "loss": 2.1295,
      "step": 33935
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.7135763168334961,
      "learning_rate": 0.0005684519053711238,
      "loss": 2.3431,
      "step": 33936
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6386234164237976,
      "learning_rate": 0.0005684500793697114,
      "loss": 2.4401,
      "step": 33937
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6850574016571045,
      "learning_rate": 0.0005684482533183891,
      "loss": 2.2506,
      "step": 33938
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.7357304692268372,
      "learning_rate": 0.0005684464272171572,
      "loss": 2.197,
      "step": 33939
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.682658314704895,
      "learning_rate": 0.0005684446010660161,
      "loss": 2.2305,
      "step": 33940
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.7023311853408813,
      "learning_rate": 0.0005684427748649661,
      "loss": 2.1715,
      "step": 33941
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6937623023986816,
      "learning_rate": 0.0005684409486140074,
      "loss": 2.2816,
      "step": 33942
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6578390598297119,
      "learning_rate": 0.0005684391223131405,
      "loss": 2.2228,
      "step": 33943
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6803364157676697,
      "learning_rate": 0.0005684372959623657,
      "loss": 2.1859,
      "step": 33944
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.7092854380607605,
      "learning_rate": 0.0005684354695616833,
      "loss": 2.295,
      "step": 33945
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.7582664489746094,
      "learning_rate": 0.0005684336431110938,
      "loss": 2.2948,
      "step": 33946
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6735652089118958,
      "learning_rate": 0.0005684318166105974,
      "loss": 2.1358,
      "step": 33947
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.7102999091148376,
      "learning_rate": 0.0005684299900601944,
      "loss": 2.4276,
      "step": 33948
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.7558717131614685,
      "learning_rate": 0.0005684281634598852,
      "loss": 2.2443,
      "step": 33949
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6484412550926208,
      "learning_rate": 0.0005684263368096701,
      "loss": 2.3158,
      "step": 33950
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6884157657623291,
      "learning_rate": 0.0005684245101095496,
      "loss": 2.0383,
      "step": 33951
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.5827451348304749,
      "learning_rate": 0.0005684226833595239,
      "loss": 2.2569,
      "step": 33952
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6909515857696533,
      "learning_rate": 0.0005684208565595933,
      "loss": 2.5447,
      "step": 33953
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.637213408946991,
      "learning_rate": 0.0005684190297097581,
      "loss": 2.0582,
      "step": 33954
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6723353266716003,
      "learning_rate": 0.0005684172028100189,
      "loss": 2.3077,
      "step": 33955
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6868422627449036,
      "learning_rate": 0.0005684153758603759,
      "loss": 2.432,
      "step": 33956
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6591261029243469,
      "learning_rate": 0.0005684135488608292,
      "loss": 2.4733,
      "step": 33957
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6395053267478943,
      "learning_rate": 0.0005684117218113796,
      "loss": 2.303,
      "step": 33958
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6777539253234863,
      "learning_rate": 0.000568409894712027,
      "loss": 2.3683,
      "step": 33959
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6578139066696167,
      "learning_rate": 0.000568408067562772,
      "loss": 2.3303,
      "step": 33960
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6835055947303772,
      "learning_rate": 0.0005684062403636149,
      "loss": 2.3215,
      "step": 33961
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6865967512130737,
      "learning_rate": 0.0005684044131145559,
      "loss": 2.2306,
      "step": 33962
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6990640759468079,
      "learning_rate": 0.0005684025858155957,
      "loss": 2.2958,
      "step": 33963
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6767627000808716,
      "learning_rate": 0.0005684007584667342,
      "loss": 2.4902,
      "step": 33964
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.7178865075111389,
      "learning_rate": 0.0005683989310679719,
      "loss": 2.1237,
      "step": 33965
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6813569664955139,
      "learning_rate": 0.0005683971036193093,
      "loss": 2.2691,
      "step": 33966
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6631339192390442,
      "learning_rate": 0.0005683952761207466,
      "loss": 2.0288,
      "step": 33967
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6723495125770569,
      "learning_rate": 0.0005683934485722841,
      "loss": 2.1391,
      "step": 33968
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.7470648288726807,
      "learning_rate": 0.0005683916209739221,
      "loss": 2.2714,
      "step": 33969
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6740603446960449,
      "learning_rate": 0.0005683897933256612,
      "loss": 2.1029,
      "step": 33970
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6633943915367126,
      "learning_rate": 0.0005683879656275014,
      "loss": 2.4534,
      "step": 33971
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.7306972146034241,
      "learning_rate": 0.0005683861378794434,
      "loss": 2.2541,
      "step": 33972
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.5989126563072205,
      "learning_rate": 0.0005683843100814873,
      "loss": 1.9413,
      "step": 33973
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6526817083358765,
      "learning_rate": 0.0005683824822336334,
      "loss": 2.2729,
      "step": 33974
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6757026314735413,
      "learning_rate": 0.0005683806543358822,
      "loss": 2.3112,
      "step": 33975
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.7808806300163269,
      "learning_rate": 0.0005683788263882339,
      "loss": 2.4311,
      "step": 33976
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6173394918441772,
      "learning_rate": 0.000568376998390689,
      "loss": 2.4045,
      "step": 33977
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.8929571509361267,
      "learning_rate": 0.0005683751703432477,
      "loss": 2.3671,
      "step": 33978
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6442717909812927,
      "learning_rate": 0.0005683733422459105,
      "loss": 2.3626,
      "step": 33979
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.7105184197425842,
      "learning_rate": 0.0005683715140986775,
      "loss": 2.216,
      "step": 33980
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.640481173992157,
      "learning_rate": 0.0005683696859015492,
      "loss": 2.3584,
      "step": 33981
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.7174156308174133,
      "learning_rate": 0.0005683678576545259,
      "loss": 2.4568,
      "step": 33982
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6287569403648376,
      "learning_rate": 0.000568366029357608,
      "loss": 2.2634,
      "step": 33983
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6568157076835632,
      "learning_rate": 0.0005683642010107957,
      "loss": 2.2513,
      "step": 33984
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6075953841209412,
      "learning_rate": 0.0005683623726140895,
      "loss": 2.1643,
      "step": 33985
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6068766117095947,
      "learning_rate": 0.0005683605441674897,
      "loss": 2.3459,
      "step": 33986
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.684640645980835,
      "learning_rate": 0.0005683587156709966,
      "loss": 2.1792,
      "step": 33987
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.639310896396637,
      "learning_rate": 0.0005683568871246104,
      "loss": 2.3371,
      "step": 33988
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6324070692062378,
      "learning_rate": 0.0005683550585283317,
      "loss": 2.2468,
      "step": 33989
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6421200037002563,
      "learning_rate": 0.0005683532298821608,
      "loss": 2.3617,
      "step": 33990
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6782963871955872,
      "learning_rate": 0.0005683514011860978,
      "loss": 2.1655,
      "step": 33991
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6044484376907349,
      "learning_rate": 0.0005683495724401434,
      "loss": 2.1962,
      "step": 33992
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.8277497887611389,
      "learning_rate": 0.0005683477436442976,
      "loss": 2.4586,
      "step": 33993
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.711161732673645,
      "learning_rate": 0.000568345914798561,
      "loss": 2.4629,
      "step": 33994
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6887698173522949,
      "learning_rate": 0.0005683440859029338,
      "loss": 2.2664,
      "step": 33995
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6438309550285339,
      "learning_rate": 0.0005683422569574163,
      "loss": 2.2629,
      "step": 33996
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6065730452537537,
      "learning_rate": 0.000568340427962009,
      "loss": 2.2323,
      "step": 33997
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.7218989729881287,
      "learning_rate": 0.0005683385989167121,
      "loss": 2.5036,
      "step": 33998
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6339399814605713,
      "learning_rate": 0.000568336769821526,
      "loss": 2.2878,
      "step": 33999
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.7329067587852478,
      "learning_rate": 0.0005683349406764509,
      "loss": 2.343,
      "step": 34000
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.685801088809967,
      "learning_rate": 0.0005683331114814874,
      "loss": 2.1593,
      "step": 34001
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6762842535972595,
      "learning_rate": 0.0005683312822366358,
      "loss": 2.2985,
      "step": 34002
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6873797178268433,
      "learning_rate": 0.0005683294529418962,
      "loss": 2.2102,
      "step": 34003
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6742249131202698,
      "learning_rate": 0.0005683276235972692,
      "loss": 2.3709,
      "step": 34004
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6743032336235046,
      "learning_rate": 0.0005683257942027549,
      "loss": 2.2265,
      "step": 34005
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6173861026763916,
      "learning_rate": 0.0005683239647583539,
      "loss": 2.481,
      "step": 34006
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6473914980888367,
      "learning_rate": 0.0005683221352640663,
      "loss": 2.4649,
      "step": 34007
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.7098000645637512,
      "learning_rate": 0.0005683203057198925,
      "loss": 2.3117,
      "step": 34008
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.7334755063056946,
      "learning_rate": 0.0005683184761258332,
      "loss": 2.3105,
      "step": 34009
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6275269985198975,
      "learning_rate": 0.0005683166464818881,
      "loss": 2.2773,
      "step": 34010
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6633690595626831,
      "learning_rate": 0.000568314816788058,
      "loss": 2.429,
      "step": 34011
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6784063577651978,
      "learning_rate": 0.0005683129870443431,
      "loss": 2.2551,
      "step": 34012
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.7086652517318726,
      "learning_rate": 0.0005683111572507437,
      "loss": 2.2042,
      "step": 34013
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.5902895331382751,
      "learning_rate": 0.0005683093274072603,
      "loss": 2.497,
      "step": 34014
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6852162480354309,
      "learning_rate": 0.000568307497513893,
      "loss": 2.3641,
      "step": 34015
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.7032548189163208,
      "learning_rate": 0.0005683056675706424,
      "loss": 2.5706,
      "step": 34016
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6256945729255676,
      "learning_rate": 0.0005683038375775087,
      "loss": 2.0732,
      "step": 34017
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6713510751724243,
      "learning_rate": 0.0005683020075344923,
      "loss": 2.3005,
      "step": 34018
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6788626313209534,
      "learning_rate": 0.0005683001774415933,
      "loss": 2.4727,
      "step": 34019
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6708599925041199,
      "learning_rate": 0.0005682983472988124,
      "loss": 2.3283,
      "step": 34020
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6347842216491699,
      "learning_rate": 0.0005682965171061496,
      "loss": 2.3633,
      "step": 34021
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.7192392945289612,
      "learning_rate": 0.0005682946868636055,
      "loss": 2.3842,
      "step": 34022
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6929574012756348,
      "learning_rate": 0.0005682928565711805,
      "loss": 2.3811,
      "step": 34023
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.8161721229553223,
      "learning_rate": 0.0005682910262288746,
      "loss": 2.219,
      "step": 34024
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.7105990648269653,
      "learning_rate": 0.0005682891958366884,
      "loss": 2.3283,
      "step": 34025
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.7285244464874268,
      "learning_rate": 0.0005682873653946222,
      "loss": 2.2898,
      "step": 34026
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.7131457328796387,
      "learning_rate": 0.0005682855349026763,
      "loss": 2.2819,
      "step": 34027
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.8048223853111267,
      "learning_rate": 0.0005682837043608509,
      "loss": 2.1186,
      "step": 34028
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6476991176605225,
      "learning_rate": 0.0005682818737691468,
      "loss": 2.2079,
      "step": 34029
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.7938110828399658,
      "learning_rate": 0.0005682800431275638,
      "loss": 2.3922,
      "step": 34030
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6420378684997559,
      "learning_rate": 0.0005682782124361025,
      "loss": 2.2445,
      "step": 34031
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6232519745826721,
      "learning_rate": 0.0005682763816947633,
      "loss": 2.2893,
      "step": 34032
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.8787339329719543,
      "learning_rate": 0.0005682745509035464,
      "loss": 2.2728,
      "step": 34033
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.7170550227165222,
      "learning_rate": 0.0005682727200624522,
      "loss": 2.5684,
      "step": 34034
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6442448496818542,
      "learning_rate": 0.0005682708891714811,
      "loss": 2.4593,
      "step": 34035
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.7379124164581299,
      "learning_rate": 0.0005682690582306332,
      "loss": 2.2659,
      "step": 34036
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.926280677318573,
      "learning_rate": 0.0005682672272399093,
      "loss": 2.3447,
      "step": 34037
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6385349631309509,
      "learning_rate": 0.0005682653961993093,
      "loss": 2.284,
      "step": 34038
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.7170746326446533,
      "learning_rate": 0.0005682635651088337,
      "loss": 2.2098,
      "step": 34039
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.7807600498199463,
      "learning_rate": 0.0005682617339684828,
      "loss": 2.0919,
      "step": 34040
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.658591091632843,
      "learning_rate": 0.0005682599027782571,
      "loss": 2.3597,
      "step": 34041
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6299332976341248,
      "learning_rate": 0.0005682580715381567,
      "loss": 2.3393,
      "step": 34042
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.8217611908912659,
      "learning_rate": 0.0005682562402481821,
      "loss": 2.0162,
      "step": 34043
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.7931960821151733,
      "learning_rate": 0.0005682544089083337,
      "loss": 2.1568,
      "step": 34044
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6888946294784546,
      "learning_rate": 0.0005682525775186116,
      "loss": 2.004,
      "step": 34045
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.8488011360168457,
      "learning_rate": 0.0005682507460790163,
      "loss": 2.2234,
      "step": 34046
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6948363780975342,
      "learning_rate": 0.0005682489145895482,
      "loss": 2.4277,
      "step": 34047
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6476600766181946,
      "learning_rate": 0.0005682470830502076,
      "loss": 2.1917,
      "step": 34048
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.603777289390564,
      "learning_rate": 0.0005682452514609946,
      "loss": 2.3341,
      "step": 34049
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.7252187728881836,
      "learning_rate": 0.0005682434198219099,
      "loss": 2.0963,
      "step": 34050
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.7300589084625244,
      "learning_rate": 0.0005682415881329536,
      "loss": 2.409,
      "step": 34051
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6605852246284485,
      "learning_rate": 0.0005682397563941263,
      "loss": 2.4073,
      "step": 34052
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6850965023040771,
      "learning_rate": 0.000568237924605428,
      "loss": 2.1402,
      "step": 34053
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6827855110168457,
      "learning_rate": 0.0005682360927668593,
      "loss": 2.2884,
      "step": 34054
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.5773234963417053,
      "learning_rate": 0.0005682342608784204,
      "loss": 2.2198,
      "step": 34055
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6418043375015259,
      "learning_rate": 0.0005682324289401116,
      "loss": 2.6023,
      "step": 34056
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6879342198371887,
      "learning_rate": 0.0005682305969519334,
      "loss": 2.5273,
      "step": 34057
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.8481895923614502,
      "learning_rate": 0.0005682287649138862,
      "loss": 2.4428,
      "step": 34058
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.7010824084281921,
      "learning_rate": 0.00056822693282597,
      "loss": 2.3685,
      "step": 34059
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.782676637172699,
      "learning_rate": 0.0005682251006881854,
      "loss": 2.2009,
      "step": 34060
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.7026550769805908,
      "learning_rate": 0.0005682232685005328,
      "loss": 2.1393,
      "step": 34061
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.5939462780952454,
      "learning_rate": 0.0005682214362630124,
      "loss": 2.2033,
      "step": 34062
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6359228491783142,
      "learning_rate": 0.0005682196039756244,
      "loss": 2.173,
      "step": 34063
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.7050970792770386,
      "learning_rate": 0.0005682177716383696,
      "loss": 2.2572,
      "step": 34064
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6677871942520142,
      "learning_rate": 0.0005682159392512478,
      "loss": 2.3544,
      "step": 34065
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6969617605209351,
      "learning_rate": 0.0005682141068142597,
      "loss": 2.3382,
      "step": 34066
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6496680378913879,
      "learning_rate": 0.0005682122743274055,
      "loss": 2.2578,
      "step": 34067
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.633736789226532,
      "learning_rate": 0.0005682104417906857,
      "loss": 2.3016,
      "step": 34068
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6555033922195435,
      "learning_rate": 0.0005682086092041004,
      "loss": 2.3818,
      "step": 34069
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6565552353858948,
      "learning_rate": 0.0005682067765676501,
      "loss": 2.4426,
      "step": 34070
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.7334428429603577,
      "learning_rate": 0.0005682049438813351,
      "loss": 2.463,
      "step": 34071
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6596307158470154,
      "learning_rate": 0.0005682031111451557,
      "loss": 2.2751,
      "step": 34072
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6606257557868958,
      "learning_rate": 0.0005682012783591123,
      "loss": 2.4872,
      "step": 34073
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.7947878241539001,
      "learning_rate": 0.0005681994455232052,
      "loss": 2.5313,
      "step": 34074
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6629871726036072,
      "learning_rate": 0.0005681976126374347,
      "loss": 2.3889,
      "step": 34075
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6721230745315552,
      "learning_rate": 0.0005681957797018014,
      "loss": 2.0942,
      "step": 34076
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6609817147254944,
      "learning_rate": 0.0005681939467163052,
      "loss": 2.3576,
      "step": 34077
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.9087684750556946,
      "learning_rate": 0.0005681921136809468,
      "loss": 2.2853,
      "step": 34078
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6930809617042542,
      "learning_rate": 0.0005681902805957264,
      "loss": 2.375,
      "step": 34079
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6774478554725647,
      "learning_rate": 0.0005681884474606445,
      "loss": 2.5867,
      "step": 34080
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6944125890731812,
      "learning_rate": 0.0005681866142757011,
      "loss": 2.1417,
      "step": 34081
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.8637771010398865,
      "learning_rate": 0.0005681847810408969,
      "loss": 2.3715,
      "step": 34082
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.7028162479400635,
      "learning_rate": 0.000568182947756232,
      "loss": 2.2821,
      "step": 34083
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.84614098072052,
      "learning_rate": 0.0005681811144217069,
      "loss": 2.3027,
      "step": 34084
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.9110349416732788,
      "learning_rate": 0.0005681792810373218,
      "loss": 2.3038,
      "step": 34085
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.646382212638855,
      "learning_rate": 0.0005681774476030771,
      "loss": 2.3909,
      "step": 34086
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6154125332832336,
      "learning_rate": 0.0005681756141189732,
      "loss": 2.3508,
      "step": 34087
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.7458646297454834,
      "learning_rate": 0.0005681737805850103,
      "loss": 2.2879,
      "step": 34088
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6922720074653625,
      "learning_rate": 0.000568171947001189,
      "loss": 2.5377,
      "step": 34089
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6508892774581909,
      "learning_rate": 0.0005681701133675093,
      "loss": 2.2775,
      "step": 34090
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.611613929271698,
      "learning_rate": 0.0005681682796839718,
      "loss": 2.2074,
      "step": 34091
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6641829609870911,
      "learning_rate": 0.0005681664459505768,
      "loss": 2.2211,
      "step": 34092
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.7280232310295105,
      "learning_rate": 0.0005681646121673244,
      "loss": 2.2202,
      "step": 34093
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.619800329208374,
      "learning_rate": 0.0005681627783342153,
      "loss": 2.244,
      "step": 34094
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.7845434546470642,
      "learning_rate": 0.0005681609444512497,
      "loss": 2.3057,
      "step": 34095
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.7704920172691345,
      "learning_rate": 0.0005681591105184279,
      "loss": 2.1479,
      "step": 34096
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6552501320838928,
      "learning_rate": 0.0005681572765357503,
      "loss": 2.273,
      "step": 34097
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.7685052752494812,
      "learning_rate": 0.0005681554425032171,
      "loss": 2.3064,
      "step": 34098
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6976200938224792,
      "learning_rate": 0.0005681536084208288,
      "loss": 2.3972,
      "step": 34099
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6771081686019897,
      "learning_rate": 0.0005681517742885855,
      "loss": 2.4896,
      "step": 34100
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6209388375282288,
      "learning_rate": 0.000568149940106488,
      "loss": 2.2713,
      "step": 34101
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.7029153108596802,
      "learning_rate": 0.0005681481058745362,
      "loss": 2.4085,
      "step": 34102
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.952730119228363,
      "learning_rate": 0.0005681462715927307,
      "loss": 2.2432,
      "step": 34103
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6313182711601257,
      "learning_rate": 0.0005681444372610717,
      "loss": 2.4982,
      "step": 34104
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6828628778457642,
      "learning_rate": 0.0005681426028795595,
      "loss": 2.1478,
      "step": 34105
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6805288791656494,
      "learning_rate": 0.0005681407684481947,
      "loss": 2.3024,
      "step": 34106
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.7130433917045593,
      "learning_rate": 0.0005681389339669774,
      "loss": 2.3899,
      "step": 34107
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6528364419937134,
      "learning_rate": 0.000568137099435908,
      "loss": 2.2815,
      "step": 34108
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6390784382820129,
      "learning_rate": 0.0005681352648549868,
      "loss": 2.3718,
      "step": 34109
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6600193381309509,
      "learning_rate": 0.0005681334302242142,
      "loss": 2.2065,
      "step": 34110
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6358272433280945,
      "learning_rate": 0.0005681315955435906,
      "loss": 2.3009,
      "step": 34111
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.7029947638511658,
      "learning_rate": 0.0005681297608131162,
      "loss": 2.3634,
      "step": 34112
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.607305109500885,
      "learning_rate": 0.0005681279260327914,
      "loss": 2.1835,
      "step": 34113
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.5993157625198364,
      "learning_rate": 0.0005681260912026167,
      "loss": 2.2714,
      "step": 34114
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6706365942955017,
      "learning_rate": 0.0005681242563225922,
      "loss": 2.365,
      "step": 34115
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.7463381886482239,
      "learning_rate": 0.0005681224213927184,
      "loss": 2.2492,
      "step": 34116
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6439096331596375,
      "learning_rate": 0.0005681205864129955,
      "loss": 2.1491,
      "step": 34117
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.7847349643707275,
      "learning_rate": 0.000568118751383424,
      "loss": 2.223,
      "step": 34118
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6692578792572021,
      "learning_rate": 0.0005681169163040042,
      "loss": 2.2489,
      "step": 34119
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6622895002365112,
      "learning_rate": 0.0005681150811747362,
      "loss": 2.3302,
      "step": 34120
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6842542886734009,
      "learning_rate": 0.0005681132459956207,
      "loss": 2.5613,
      "step": 34121
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.7041063904762268,
      "learning_rate": 0.0005681114107666578,
      "loss": 2.1067,
      "step": 34122
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6611006259918213,
      "learning_rate": 0.0005681095754878481,
      "loss": 2.181,
      "step": 34123
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6316160559654236,
      "learning_rate": 0.0005681077401591916,
      "loss": 2.1256,
      "step": 34124
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6396278738975525,
      "learning_rate": 0.0005681059047806889,
      "loss": 2.1077,
      "step": 34125
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6369318962097168,
      "learning_rate": 0.0005681040693523402,
      "loss": 2.3491,
      "step": 34126
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.5897183418273926,
      "learning_rate": 0.000568102233874146,
      "loss": 2.089,
      "step": 34127
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6716546416282654,
      "learning_rate": 0.0005681003983461064,
      "loss": 2.233,
      "step": 34128
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.7488184571266174,
      "learning_rate": 0.0005680985627682218,
      "loss": 2.5689,
      "step": 34129
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6937360167503357,
      "learning_rate": 0.0005680967271404928,
      "loss": 2.1544,
      "step": 34130
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.657382607460022,
      "learning_rate": 0.0005680948914629195,
      "loss": 2.0127,
      "step": 34131
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6260006427764893,
      "learning_rate": 0.0005680930557355024,
      "loss": 2.3494,
      "step": 34132
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6881178617477417,
      "learning_rate": 0.0005680912199582416,
      "loss": 2.273,
      "step": 34133
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6538113355636597,
      "learning_rate": 0.0005680893841311375,
      "loss": 2.1847,
      "step": 34134
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6251406669616699,
      "learning_rate": 0.0005680875482541907,
      "loss": 2.2201,
      "step": 34135
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6425092220306396,
      "learning_rate": 0.0005680857123274012,
      "loss": 2.1613,
      "step": 34136
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6685848236083984,
      "learning_rate": 0.0005680838763507697,
      "loss": 2.1746,
      "step": 34137
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.724280059337616,
      "learning_rate": 0.0005680820403242962,
      "loss": 2.2016,
      "step": 34138
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6475447416305542,
      "learning_rate": 0.0005680802042479814,
      "loss": 2.3134,
      "step": 34139
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.7826218605041504,
      "learning_rate": 0.0005680783681218252,
      "loss": 2.638,
      "step": 34140
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.7945188879966736,
      "learning_rate": 0.0005680765319458281,
      "loss": 2.329,
      "step": 34141
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6921018958091736,
      "learning_rate": 0.0005680746957199907,
      "loss": 2.2649,
      "step": 34142
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.8708277344703674,
      "learning_rate": 0.0005680728594443131,
      "loss": 2.2057,
      "step": 34143
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.674678385257721,
      "learning_rate": 0.0005680710231187958,
      "loss": 2.182,
      "step": 34144
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6597174406051636,
      "learning_rate": 0.0005680691867434389,
      "loss": 2.1879,
      "step": 34145
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.7078427076339722,
      "learning_rate": 0.000568067350318243,
      "loss": 2.4409,
      "step": 34146
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.7398450374603271,
      "learning_rate": 0.0005680655138432082,
      "loss": 2.129,
      "step": 34147
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.7185831069946289,
      "learning_rate": 0.0005680636773183349,
      "loss": 2.2698,
      "step": 34148
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.7654100060462952,
      "learning_rate": 0.0005680618407436236,
      "loss": 2.3246,
      "step": 34149
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6737341284751892,
      "learning_rate": 0.0005680600041190747,
      "loss": 2.3745,
      "step": 34150
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.7388089299201965,
      "learning_rate": 0.0005680581674446881,
      "loss": 2.1383,
      "step": 34151
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6238320469856262,
      "learning_rate": 0.0005680563307204646,
      "loss": 2.1629,
      "step": 34152
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6836376786231995,
      "learning_rate": 0.0005680544939464043,
      "loss": 2.2619,
      "step": 34153
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.7384613752365112,
      "learning_rate": 0.0005680526571225077,
      "loss": 2.2923,
      "step": 34154
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6561002731323242,
      "learning_rate": 0.000568050820248775,
      "loss": 2.4755,
      "step": 34155
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.8468746542930603,
      "learning_rate": 0.0005680489833252066,
      "loss": 2.2558,
      "step": 34156
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6945206522941589,
      "learning_rate": 0.0005680471463518028,
      "loss": 2.2373,
      "step": 34157
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.7073342204093933,
      "learning_rate": 0.0005680453093285639,
      "loss": 2.2248,
      "step": 34158
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.5946212410926819,
      "learning_rate": 0.0005680434722554904,
      "loss": 2.0649,
      "step": 34159
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.7212139964103699,
      "learning_rate": 0.0005680416351325825,
      "loss": 2.2143,
      "step": 34160
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6735243797302246,
      "learning_rate": 0.0005680397979598408,
      "loss": 2.1609,
      "step": 34161
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.7513108849525452,
      "learning_rate": 0.0005680379607372652,
      "loss": 2.1742,
      "step": 34162
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6452295184135437,
      "learning_rate": 0.0005680361234648565,
      "loss": 2.1662,
      "step": 34163
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6605923771858215,
      "learning_rate": 0.0005680342861426146,
      "loss": 2.1461,
      "step": 34164
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.7778968811035156,
      "learning_rate": 0.0005680324487705403,
      "loss": 2.4548,
      "step": 34165
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.609234094619751,
      "learning_rate": 0.0005680306113486336,
      "loss": 2.242,
      "step": 34166
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6803743839263916,
      "learning_rate": 0.0005680287738768949,
      "loss": 2.0866,
      "step": 34167
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6848134398460388,
      "learning_rate": 0.0005680269363553246,
      "loss": 2.2132,
      "step": 34168
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.7701480388641357,
      "learning_rate": 0.0005680250987839231,
      "loss": 2.3866,
      "step": 34169
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6795380115509033,
      "learning_rate": 0.0005680232611626906,
      "loss": 2.3507,
      "step": 34170
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.5979772806167603,
      "learning_rate": 0.0005680214234916276,
      "loss": 2.1946,
      "step": 34171
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6971716284751892,
      "learning_rate": 0.0005680195857707343,
      "loss": 2.405,
      "step": 34172
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6774393320083618,
      "learning_rate": 0.0005680177480000111,
      "loss": 2.3791,
      "step": 34173
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6285507678985596,
      "learning_rate": 0.0005680159101794583,
      "loss": 2.3735,
      "step": 34174
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6472957134246826,
      "learning_rate": 0.0005680140723090764,
      "loss": 2.5273,
      "step": 34175
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6597388982772827,
      "learning_rate": 0.0005680122343888656,
      "loss": 2.2937,
      "step": 34176
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6396329402923584,
      "learning_rate": 0.0005680103964188261,
      "loss": 2.1267,
      "step": 34177
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6242459416389465,
      "learning_rate": 0.0005680085583989586,
      "loss": 2.4856,
      "step": 34178
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6575056910514832,
      "learning_rate": 0.0005680067203292633,
      "loss": 2.2029,
      "step": 34179
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.652833104133606,
      "learning_rate": 0.0005680048822097403,
      "loss": 2.1681,
      "step": 34180
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6841036081314087,
      "learning_rate": 0.0005680030440403901,
      "loss": 2.0095,
      "step": 34181
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.7780545949935913,
      "learning_rate": 0.0005680012058212133,
      "loss": 2.5869,
      "step": 34182
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.7578401565551758,
      "learning_rate": 0.0005679993675522098,
      "loss": 2.5562,
      "step": 34183
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6246929168701172,
      "learning_rate": 0.0005679975292333803,
      "loss": 1.9222,
      "step": 34184
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6367581486701965,
      "learning_rate": 0.000567995690864725,
      "loss": 2.4999,
      "step": 34185
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.8030651807785034,
      "learning_rate": 0.0005679938524462443,
      "loss": 2.3016,
      "step": 34186
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6942577362060547,
      "learning_rate": 0.0005679920139779384,
      "loss": 2.271,
      "step": 34187
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.7163202166557312,
      "learning_rate": 0.0005679901754598076,
      "loss": 2.5544,
      "step": 34188
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.7193905711174011,
      "learning_rate": 0.0005679883368918525,
      "loss": 2.3749,
      "step": 34189
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.7817379236221313,
      "learning_rate": 0.0005679864982740733,
      "loss": 2.4082,
      "step": 34190
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6727473735809326,
      "learning_rate": 0.0005679846596064704,
      "loss": 2.2159,
      "step": 34191
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6768316626548767,
      "learning_rate": 0.000567982820889044,
      "loss": 2.4638,
      "step": 34192
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6663030982017517,
      "learning_rate": 0.0005679809821217945,
      "loss": 2.3408,
      "step": 34193
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.9951509833335876,
      "learning_rate": 0.0005679791433047224,
      "loss": 2.1875,
      "step": 34194
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6860628128051758,
      "learning_rate": 0.0005679773044378279,
      "loss": 2.08,
      "step": 34195
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.9229660034179688,
      "learning_rate": 0.0005679754655211113,
      "loss": 2.1586,
      "step": 34196
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.030504584312439,
      "learning_rate": 0.000567973626554573,
      "loss": 2.545,
      "step": 34197
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.7575298547744751,
      "learning_rate": 0.0005679717875382134,
      "loss": 2.4395,
      "step": 34198
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6865803003311157,
      "learning_rate": 0.0005679699484720328,
      "loss": 2.1634,
      "step": 34199
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.8859550952911377,
      "learning_rate": 0.0005679681093560315,
      "loss": 2.4256,
      "step": 34200
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.7698156237602234,
      "learning_rate": 0.0005679662701902099,
      "loss": 2.4097,
      "step": 34201
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.7315306663513184,
      "learning_rate": 0.0005679644309745682,
      "loss": 2.127,
      "step": 34202
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6878027319908142,
      "learning_rate": 0.0005679625917091069,
      "loss": 2.2182,
      "step": 34203
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.7686203122138977,
      "learning_rate": 0.0005679607523938263,
      "loss": 2.4314,
      "step": 34204
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.7005183100700378,
      "learning_rate": 0.0005679589130287269,
      "loss": 2.3631,
      "step": 34205
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6349085569381714,
      "learning_rate": 0.0005679570736138086,
      "loss": 2.2314,
      "step": 34206
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6803942322731018,
      "learning_rate": 0.0005679552341490723,
      "loss": 2.3409,
      "step": 34207
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.7128424048423767,
      "learning_rate": 0.0005679533946345178,
      "loss": 2.5619,
      "step": 34208
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.7706748843193054,
      "learning_rate": 0.0005679515550701459,
      "loss": 2.3811,
      "step": 34209
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6059108376502991,
      "learning_rate": 0.0005679497154559567,
      "loss": 2.2747,
      "step": 34210
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6273067593574524,
      "learning_rate": 0.0005679478757919505,
      "loss": 2.3922,
      "step": 34211
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.8132330179214478,
      "learning_rate": 0.0005679460360781278,
      "loss": 2.1712,
      "step": 34212
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.5650640726089478,
      "learning_rate": 0.0005679441963144889,
      "loss": 2.2626,
      "step": 34213
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6416505575180054,
      "learning_rate": 0.0005679423565010341,
      "loss": 2.5218,
      "step": 34214
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.7085990905761719,
      "learning_rate": 0.0005679405166377638,
      "loss": 2.463,
      "step": 34215
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6962586045265198,
      "learning_rate": 0.0005679386767246782,
      "loss": 2.4667,
      "step": 34216
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.5867428779602051,
      "learning_rate": 0.0005679368367617778,
      "loss": 2.1261,
      "step": 34217
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6378267407417297,
      "learning_rate": 0.0005679349967490628,
      "loss": 2.1215,
      "step": 34218
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6604383587837219,
      "learning_rate": 0.0005679331566865338,
      "loss": 2.1598,
      "step": 34219
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.7272496819496155,
      "learning_rate": 0.0005679313165741908,
      "loss": 2.5369,
      "step": 34220
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.8767026662826538,
      "learning_rate": 0.0005679294764120344,
      "loss": 2.1516,
      "step": 34221
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.742364227771759,
      "learning_rate": 0.0005679276362000649,
      "loss": 2.4705,
      "step": 34222
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.8730983734130859,
      "learning_rate": 0.0005679257959382826,
      "loss": 2.4832,
      "step": 34223
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6269422173500061,
      "learning_rate": 0.0005679239556266877,
      "loss": 2.2396,
      "step": 34224
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6516112089157104,
      "learning_rate": 0.0005679221152652807,
      "loss": 2.241,
      "step": 34225
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.7460156083106995,
      "learning_rate": 0.000567920274854062,
      "loss": 2.4554,
      "step": 34226
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.0922328233718872,
      "learning_rate": 0.0005679184343930318,
      "loss": 2.479,
      "step": 34227
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.7013742923736572,
      "learning_rate": 0.0005679165938821905,
      "loss": 2.4236,
      "step": 34228
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6827887892723083,
      "learning_rate": 0.0005679147533215386,
      "loss": 2.121,
      "step": 34229
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.7721990942955017,
      "learning_rate": 0.0005679129127110762,
      "loss": 2.4287,
      "step": 34230
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.7140790820121765,
      "learning_rate": 0.0005679110720508037,
      "loss": 2.4301,
      "step": 34231
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6418187618255615,
      "learning_rate": 0.0005679092313407214,
      "loss": 2.2352,
      "step": 34232
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.7111985087394714,
      "learning_rate": 0.0005679073905808298,
      "loss": 2.439,
      "step": 34233
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.7065623998641968,
      "learning_rate": 0.0005679055497711292,
      "loss": 2.2449,
      "step": 34234
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6593415141105652,
      "learning_rate": 0.0005679037089116199,
      "loss": 2.4233,
      "step": 34235
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6371908187866211,
      "learning_rate": 0.0005679018680023021,
      "loss": 2.3264,
      "step": 34236
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6397290229797363,
      "learning_rate": 0.0005679000270431764,
      "loss": 2.1444,
      "step": 34237
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.7361193895339966,
      "learning_rate": 0.0005678981860342431,
      "loss": 2.3136,
      "step": 34238
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6372268795967102,
      "learning_rate": 0.0005678963449755023,
      "loss": 2.1948,
      "step": 34239
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6513452529907227,
      "learning_rate": 0.0005678945038669547,
      "loss": 2.2902,
      "step": 34240
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6396981477737427,
      "learning_rate": 0.0005678926627086002,
      "loss": 2.1495,
      "step": 34241
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6112691164016724,
      "learning_rate": 0.0005678908215004397,
      "loss": 2.2711,
      "step": 34242
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6280509829521179,
      "learning_rate": 0.000567888980242473,
      "loss": 2.1487,
      "step": 34243
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6114875674247742,
      "learning_rate": 0.0005678871389347008,
      "loss": 2.2386,
      "step": 34244
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6657570600509644,
      "learning_rate": 0.0005678852975771233,
      "loss": 2.3867,
      "step": 34245
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6826448440551758,
      "learning_rate": 0.0005678834561697408,
      "loss": 2.056,
      "step": 34246
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6681627035140991,
      "learning_rate": 0.0005678816147125539,
      "loss": 2.3077,
      "step": 34247
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6490887403488159,
      "learning_rate": 0.0005678797732055625,
      "loss": 2.2627,
      "step": 34248
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6610524654388428,
      "learning_rate": 0.0005678779316487673,
      "loss": 2.1102,
      "step": 34249
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6740993857383728,
      "learning_rate": 0.0005678760900421686,
      "loss": 2.4063,
      "step": 34250
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6050018072128296,
      "learning_rate": 0.0005678742483857666,
      "loss": 2.2714,
      "step": 34251
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.7174972891807556,
      "learning_rate": 0.0005678724066795617,
      "loss": 2.2649,
      "step": 34252
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.637913167476654,
      "learning_rate": 0.0005678705649235544,
      "loss": 2.3841,
      "step": 34253
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6779441833496094,
      "learning_rate": 0.0005678687231177448,
      "loss": 2.2534,
      "step": 34254
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6725576519966125,
      "learning_rate": 0.0005678668812621334,
      "loss": 2.3275,
      "step": 34255
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6775634288787842,
      "learning_rate": 0.0005678650393567204,
      "loss": 2.385,
      "step": 34256
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6924273371696472,
      "learning_rate": 0.0005678631974015064,
      "loss": 2.3178,
      "step": 34257
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6527905464172363,
      "learning_rate": 0.0005678613553964914,
      "loss": 2.2004,
      "step": 34258
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6660528182983398,
      "learning_rate": 0.000567859513341676,
      "loss": 2.3392,
      "step": 34259
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6415175795555115,
      "learning_rate": 0.0005678576712370605,
      "loss": 2.1439,
      "step": 34260
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6188262104988098,
      "learning_rate": 0.000567855829082645,
      "loss": 2.2371,
      "step": 34261
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6773682832717896,
      "learning_rate": 0.0005678539868784303,
      "loss": 2.026,
      "step": 34262
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6513376832008362,
      "learning_rate": 0.0005678521446244164,
      "loss": 2.3389,
      "step": 34263
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6756674647331238,
      "learning_rate": 0.0005678503023206037,
      "loss": 2.4928,
      "step": 34264
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.710810661315918,
      "learning_rate": 0.0005678484599669926,
      "loss": 2.2036,
      "step": 34265
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6286771893501282,
      "learning_rate": 0.0005678466175635834,
      "loss": 2.1673,
      "step": 34266
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.7217957377433777,
      "learning_rate": 0.0005678447751103765,
      "loss": 2.7996,
      "step": 34267
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6380185484886169,
      "learning_rate": 0.000567842932607372,
      "loss": 2.2593,
      "step": 34268
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.7839307188987732,
      "learning_rate": 0.0005678410900545706,
      "loss": 2.1402,
      "step": 34269
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6868163347244263,
      "learning_rate": 0.0005678392474519725,
      "loss": 2.237,
      "step": 34270
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6175583600997925,
      "learning_rate": 0.000567837404799578,
      "loss": 2.3847,
      "step": 34271
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.7212433218955994,
      "learning_rate": 0.0005678355620973876,
      "loss": 2.3303,
      "step": 34272
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6738077998161316,
      "learning_rate": 0.0005678337193454014,
      "loss": 2.2657,
      "step": 34273
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6538188457489014,
      "learning_rate": 0.0005678318765436198,
      "loss": 2.5422,
      "step": 34274
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6707850098609924,
      "learning_rate": 0.0005678300336920433,
      "loss": 2.1839,
      "step": 34275
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.7829603552818298,
      "learning_rate": 0.000567828190790672,
      "loss": 2.3541,
      "step": 34276
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.673799991607666,
      "learning_rate": 0.0005678263478395066,
      "loss": 2.2512,
      "step": 34277
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.7426391243934631,
      "learning_rate": 0.000567824504838547,
      "loss": 2.3561,
      "step": 34278
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6531442999839783,
      "learning_rate": 0.0005678226617877939,
      "loss": 2.3832,
      "step": 34279
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6647301912307739,
      "learning_rate": 0.0005678208186872475,
      "loss": 2.07,
      "step": 34280
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6706183552742004,
      "learning_rate": 0.0005678189755369081,
      "loss": 2.2667,
      "step": 34281
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6741617321968079,
      "learning_rate": 0.0005678171323367762,
      "loss": 2.2141,
      "step": 34282
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.7490078210830688,
      "learning_rate": 0.000567815289086852,
      "loss": 2.4343,
      "step": 34283
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6374332308769226,
      "learning_rate": 0.0005678134457871357,
      "loss": 2.313,
      "step": 34284
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6547130346298218,
      "learning_rate": 0.000567811602437628,
      "loss": 2.3522,
      "step": 34285
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6976949572563171,
      "learning_rate": 0.0005678097590383291,
      "loss": 2.4944,
      "step": 34286
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6389250755310059,
      "learning_rate": 0.0005678079155892393,
      "loss": 2.2551,
      "step": 34287
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.7066693902015686,
      "learning_rate": 0.0005678060720903589,
      "loss": 2.4102,
      "step": 34288
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.722051739692688,
      "learning_rate": 0.0005678042285416883,
      "loss": 2.2873,
      "step": 34289
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6721893548965454,
      "learning_rate": 0.0005678023849432279,
      "loss": 2.2807,
      "step": 34290
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.8074743151664734,
      "learning_rate": 0.0005678005412949778,
      "loss": 2.338,
      "step": 34291
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6810413599014282,
      "learning_rate": 0.0005677986975969388,
      "loss": 2.4393,
      "step": 34292
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.66953045129776,
      "learning_rate": 0.0005677968538491107,
      "loss": 2.0981,
      "step": 34293
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6587085723876953,
      "learning_rate": 0.0005677950100514943,
      "loss": 2.007,
      "step": 34294
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6469148993492126,
      "learning_rate": 0.0005677931662040897,
      "loss": 2.4001,
      "step": 34295
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6979413628578186,
      "learning_rate": 0.0005677913223068972,
      "loss": 2.4322,
      "step": 34296
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6513697504997253,
      "learning_rate": 0.0005677894783599174,
      "loss": 2.2935,
      "step": 34297
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6143029928207397,
      "learning_rate": 0.0005677876343631504,
      "loss": 2.1889,
      "step": 34298
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.7316300272941589,
      "learning_rate": 0.0005677857903165966,
      "loss": 2.5398,
      "step": 34299
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.7229989767074585,
      "learning_rate": 0.0005677839462202563,
      "loss": 2.3282,
      "step": 34300
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6590989828109741,
      "learning_rate": 0.00056778210207413,
      "loss": 2.1147,
      "step": 34301
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.695955216884613,
      "learning_rate": 0.0005677802578782181,
      "loss": 2.3864,
      "step": 34302
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6590467691421509,
      "learning_rate": 0.0005677784136325205,
      "loss": 2.2331,
      "step": 34303
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6065016984939575,
      "learning_rate": 0.0005677765693370381,
      "loss": 2.251,
      "step": 34304
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6406316161155701,
      "learning_rate": 0.0005677747249917708,
      "loss": 2.3319,
      "step": 34305
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6360499262809753,
      "learning_rate": 0.0005677728805967192,
      "loss": 2.4276,
      "step": 34306
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6693363785743713,
      "learning_rate": 0.0005677710361518836,
      "loss": 2.1818,
      "step": 34307
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6413488388061523,
      "learning_rate": 0.0005677691916572641,
      "loss": 2.3017,
      "step": 34308
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6899759769439697,
      "learning_rate": 0.0005677673471128615,
      "loss": 2.3723,
      "step": 34309
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.5897797346115112,
      "learning_rate": 0.0005677655025186758,
      "loss": 2.2492,
      "step": 34310
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6127133965492249,
      "learning_rate": 0.0005677636578747075,
      "loss": 2.1745,
      "step": 34311
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.7275320291519165,
      "learning_rate": 0.0005677618131809568,
      "loss": 2.2642,
      "step": 34312
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.7102771401405334,
      "learning_rate": 0.0005677599684374241,
      "loss": 2.2594,
      "step": 34313
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.7147589921951294,
      "learning_rate": 0.00056775812364411,
      "loss": 2.1809,
      "step": 34314
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6548415422439575,
      "learning_rate": 0.0005677562788010144,
      "loss": 2.2941,
      "step": 34315
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.7383063435554504,
      "learning_rate": 0.0005677544339081378,
      "loss": 2.3314,
      "step": 34316
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.7507918477058411,
      "learning_rate": 0.0005677525889654807,
      "loss": 2.3091,
      "step": 34317
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6963257193565369,
      "learning_rate": 0.0005677507439730433,
      "loss": 2.2681,
      "step": 34318
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.7962176203727722,
      "learning_rate": 0.000567748898930826,
      "loss": 2.2008,
      "step": 34319
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.7336779832839966,
      "learning_rate": 0.0005677470538388292,
      "loss": 2.1327,
      "step": 34320
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.7910181283950806,
      "learning_rate": 0.000567745208697053,
      "loss": 2.1719,
      "step": 34321
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6366648077964783,
      "learning_rate": 0.0005677433635054981,
      "loss": 2.1873,
      "step": 34322
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.5885040760040283,
      "learning_rate": 0.0005677415182641645,
      "loss": 2.2513,
      "step": 34323
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.7042156457901001,
      "learning_rate": 0.0005677396729730529,
      "loss": 2.3543,
      "step": 34324
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.7353740334510803,
      "learning_rate": 0.0005677378276321632,
      "loss": 2.3577,
      "step": 34325
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.685949981212616,
      "learning_rate": 0.0005677359822414961,
      "loss": 2.5426,
      "step": 34326
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.7037392854690552,
      "learning_rate": 0.0005677341368010518,
      "loss": 2.1841,
      "step": 34327
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6842558979988098,
      "learning_rate": 0.0005677322913108308,
      "loss": 2.2744,
      "step": 34328
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6908900141716003,
      "learning_rate": 0.0005677304457708331,
      "loss": 2.3021,
      "step": 34329
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6381868124008179,
      "learning_rate": 0.0005677286001810595,
      "loss": 2.3891,
      "step": 34330
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6632308959960938,
      "learning_rate": 0.0005677267545415098,
      "loss": 2.0958,
      "step": 34331
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6269432902336121,
      "learning_rate": 0.0005677249088521848,
      "loss": 2.0898,
      "step": 34332
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6304634213447571,
      "learning_rate": 0.0005677230631130848,
      "loss": 2.4111,
      "step": 34333
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6997814774513245,
      "learning_rate": 0.0005677212173242098,
      "loss": 2.3152,
      "step": 34334
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6390604972839355,
      "learning_rate": 0.0005677193714855606,
      "loss": 2.2746,
      "step": 34335
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6469402313232422,
      "learning_rate": 0.0005677175255971371,
      "loss": 2.2344,
      "step": 34336
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.7071729302406311,
      "learning_rate": 0.00056771567965894,
      "loss": 2.3222,
      "step": 34337
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6362072229385376,
      "learning_rate": 0.0005677138336709695,
      "loss": 2.6811,
      "step": 34338
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.733214795589447,
      "learning_rate": 0.0005677119876332259,
      "loss": 2.5067,
      "step": 34339
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.7225192189216614,
      "learning_rate": 0.0005677101415457095,
      "loss": 2.542,
      "step": 34340
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.7745264768600464,
      "learning_rate": 0.0005677082954084209,
      "loss": 2.4222,
      "step": 34341
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6961007714271545,
      "learning_rate": 0.0005677064492213602,
      "loss": 2.2744,
      "step": 34342
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6482267379760742,
      "learning_rate": 0.0005677046029845278,
      "loss": 2.2393,
      "step": 34343
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6768101453781128,
      "learning_rate": 0.0005677027566979242,
      "loss": 2.1525,
      "step": 34344
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6361023783683777,
      "learning_rate": 0.0005677009103615494,
      "loss": 2.1951,
      "step": 34345
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6588609218597412,
      "learning_rate": 0.000567699063975404,
      "loss": 2.0972,
      "step": 34346
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.7079130411148071,
      "learning_rate": 0.0005676972175394883,
      "loss": 2.3025,
      "step": 34347
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.7302139401435852,
      "learning_rate": 0.0005676953710538027,
      "loss": 2.0794,
      "step": 34348
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6404639482498169,
      "learning_rate": 0.0005676935245183475,
      "loss": 2.306,
      "step": 34349
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.8184380531311035,
      "learning_rate": 0.0005676916779331229,
      "loss": 2.2453,
      "step": 34350
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.667550802230835,
      "learning_rate": 0.0005676898312981294,
      "loss": 2.4445,
      "step": 34351
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.667652428150177,
      "learning_rate": 0.0005676879846133673,
      "loss": 2.2869,
      "step": 34352
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.7193573713302612,
      "learning_rate": 0.000567686137878837,
      "loss": 2.3971,
      "step": 34353
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.7171987891197205,
      "learning_rate": 0.0005676842910945387,
      "loss": 2.4718,
      "step": 34354
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6528981328010559,
      "learning_rate": 0.000567682444260473,
      "loss": 2.363,
      "step": 34355
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.692047119140625,
      "learning_rate": 0.0005676805973766399,
      "loss": 2.0751,
      "step": 34356
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6696884632110596,
      "learning_rate": 0.0005676787504430401,
      "loss": 2.2973,
      "step": 34357
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6102609038352966,
      "learning_rate": 0.0005676769034596736,
      "loss": 2.326,
      "step": 34358
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6221790909767151,
      "learning_rate": 0.0005676750564265411,
      "loss": 2.1077,
      "step": 34359
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6814869046211243,
      "learning_rate": 0.0005676732093436426,
      "loss": 2.0894,
      "step": 34360
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6585378050804138,
      "learning_rate": 0.0005676713622109786,
      "loss": 2.349,
      "step": 34361
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6675277352333069,
      "learning_rate": 0.0005676695150285496,
      "loss": 2.308,
      "step": 34362
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6379508972167969,
      "learning_rate": 0.0005676676677963555,
      "loss": 2.2042,
      "step": 34363
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.7114145159721375,
      "learning_rate": 0.0005676658205143972,
      "loss": 2.3464,
      "step": 34364
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.7243300676345825,
      "learning_rate": 0.0005676639731826747,
      "loss": 2.3468,
      "step": 34365
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6752748489379883,
      "learning_rate": 0.0005676621258011883,
      "loss": 2.1294,
      "step": 34366
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.7082778215408325,
      "learning_rate": 0.0005676602783699385,
      "loss": 2.3855,
      "step": 34367
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6963145136833191,
      "learning_rate": 0.0005676584308889257,
      "loss": 2.4755,
      "step": 34368
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6603294014930725,
      "learning_rate": 0.00056765658335815,
      "loss": 2.1441,
      "step": 34369
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.657868504524231,
      "learning_rate": 0.000567654735777612,
      "loss": 2.2141,
      "step": 34370
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.7201794981956482,
      "learning_rate": 0.000567652888147312,
      "loss": 2.2866,
      "step": 34371
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6786518096923828,
      "learning_rate": 0.0005676510404672502,
      "loss": 2.281,
      "step": 34372
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.7032731175422668,
      "learning_rate": 0.000567649192737427,
      "loss": 2.4263,
      "step": 34373
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6727845668792725,
      "learning_rate": 0.0005676473449578428,
      "loss": 2.2834,
      "step": 34374
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6857239603996277,
      "learning_rate": 0.0005676454971284977,
      "loss": 2.4855,
      "step": 34375
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6131510138511658,
      "learning_rate": 0.0005676436492493925,
      "loss": 2.5389,
      "step": 34376
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.7448967695236206,
      "learning_rate": 0.0005676418013205272,
      "loss": 2.3413,
      "step": 34377
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.7050143480300903,
      "learning_rate": 0.0005676399533419023,
      "loss": 2.264,
      "step": 34378
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.7003405094146729,
      "learning_rate": 0.0005676381053135181,
      "loss": 2.4609,
      "step": 34379
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.673017144203186,
      "learning_rate": 0.0005676362572353748,
      "loss": 2.1885,
      "step": 34380
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.7105779647827148,
      "learning_rate": 0.000567634409107473,
      "loss": 2.2185,
      "step": 34381
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6602106094360352,
      "learning_rate": 0.0005676325609298127,
      "loss": 2.1671,
      "step": 34382
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.7360346913337708,
      "learning_rate": 0.0005676307127023947,
      "loss": 2.3443,
      "step": 34383
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6945136785507202,
      "learning_rate": 0.0005676288644252189,
      "loss": 2.4504,
      "step": 34384
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.654744565486908,
      "learning_rate": 0.000567627016098286,
      "loss": 2.3235,
      "step": 34385
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.652353823184967,
      "learning_rate": 0.000567625167721596,
      "loss": 2.1275,
      "step": 34386
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.7480977177619934,
      "learning_rate": 0.0005676233192951496,
      "loss": 2.4499,
      "step": 34387
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6784921288490295,
      "learning_rate": 0.0005676214708189469,
      "loss": 2.3656,
      "step": 34388
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.7415252327919006,
      "learning_rate": 0.0005676196222929884,
      "loss": 2.268,
      "step": 34389
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.7102857232093811,
      "learning_rate": 0.0005676177737172742,
      "loss": 2.3585,
      "step": 34390
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6682571768760681,
      "learning_rate": 0.000567615925091805,
      "loss": 2.1896,
      "step": 34391
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6757438778877258,
      "learning_rate": 0.0005676140764165808,
      "loss": 2.3711,
      "step": 34392
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6467634439468384,
      "learning_rate": 0.0005676122276916022,
      "loss": 2.0796,
      "step": 34393
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.7783365845680237,
      "learning_rate": 0.0005676103789168693,
      "loss": 2.2099,
      "step": 34394
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6953967809677124,
      "learning_rate": 0.0005676085300923827,
      "loss": 2.3374,
      "step": 34395
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.832390308380127,
      "learning_rate": 0.0005676066812181426,
      "loss": 2.2179,
      "step": 34396
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6612670421600342,
      "learning_rate": 0.0005676048322941494,
      "loss": 2.3498,
      "step": 34397
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.5954563021659851,
      "learning_rate": 0.0005676029833204034,
      "loss": 2.1788,
      "step": 34398
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.7954577207565308,
      "learning_rate": 0.0005676011342969048,
      "loss": 2.1483,
      "step": 34399
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6958011984825134,
      "learning_rate": 0.0005675992852236542,
      "loss": 2.1765,
      "step": 34400
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.5958744287490845,
      "learning_rate": 0.0005675974361006519,
      "loss": 2.4431,
      "step": 34401
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6472558379173279,
      "learning_rate": 0.0005675955869278982,
      "loss": 2.2447,
      "step": 34402
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6760592460632324,
      "learning_rate": 0.0005675937377053932,
      "loss": 2.6629,
      "step": 34403
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6826289892196655,
      "learning_rate": 0.0005675918884331377,
      "loss": 2.4707,
      "step": 34404
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6694596409797668,
      "learning_rate": 0.0005675900391111317,
      "loss": 2.6728,
      "step": 34405
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6469776034355164,
      "learning_rate": 0.0005675881897393757,
      "loss": 2.0691,
      "step": 34406
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.7508313655853271,
      "learning_rate": 0.0005675863403178702,
      "loss": 2.2442,
      "step": 34407
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.7156417369842529,
      "learning_rate": 0.0005675844908466151,
      "loss": 2.0702,
      "step": 34408
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.675631582736969,
      "learning_rate": 0.0005675826413256111,
      "loss": 2.2281,
      "step": 34409
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.7678704857826233,
      "learning_rate": 0.0005675807917548582,
      "loss": 2.3022,
      "step": 34410
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6462979912757874,
      "learning_rate": 0.0005675789421343572,
      "loss": 2.3655,
      "step": 34411
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.7731310129165649,
      "learning_rate": 0.0005675770924641083,
      "loss": 2.4387,
      "step": 34412
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.7160441279411316,
      "learning_rate": 0.0005675752427441117,
      "loss": 2.4272,
      "step": 34413
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.7447006106376648,
      "learning_rate": 0.0005675733929743677,
      "loss": 2.263,
      "step": 34414
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6086636781692505,
      "learning_rate": 0.0005675715431548768,
      "loss": 2.116,
      "step": 34415
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6662493348121643,
      "learning_rate": 0.0005675696932856394,
      "loss": 2.274,
      "step": 34416
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6478682160377502,
      "learning_rate": 0.0005675678433666556,
      "loss": 2.3216,
      "step": 34417
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.7266704440116882,
      "learning_rate": 0.000567565993397926,
      "loss": 2.3386,
      "step": 34418
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.676357626914978,
      "learning_rate": 0.0005675641433794508,
      "loss": 2.4763,
      "step": 34419
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.9085774421691895,
      "learning_rate": 0.0005675622933112304,
      "loss": 2.1971,
      "step": 34420
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6646645665168762,
      "learning_rate": 0.0005675604431932651,
      "loss": 2.4502,
      "step": 34421
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6746050119400024,
      "learning_rate": 0.0005675585930255552,
      "loss": 2.4819,
      "step": 34422
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.7153358459472656,
      "learning_rate": 0.0005675567428081011,
      "loss": 2.3348,
      "step": 34423
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6854162216186523,
      "learning_rate": 0.0005675548925409032,
      "loss": 2.5213,
      "step": 34424
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.7061719298362732,
      "learning_rate": 0.0005675530422239619,
      "loss": 2.2433,
      "step": 34425
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.7251614928245544,
      "learning_rate": 0.0005675511918572774,
      "loss": 2.2453,
      "step": 34426
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.7521108984947205,
      "learning_rate": 0.00056754934144085,
      "loss": 2.4345,
      "step": 34427
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.7928237318992615,
      "learning_rate": 0.0005675474909746801,
      "loss": 2.3717,
      "step": 34428
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6461632251739502,
      "learning_rate": 0.0005675456404587681,
      "loss": 2.3076,
      "step": 34429
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6832857131958008,
      "learning_rate": 0.0005675437898931143,
      "loss": 2.4261,
      "step": 34430
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6310522556304932,
      "learning_rate": 0.0005675419392777192,
      "loss": 2.1962,
      "step": 34431
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.7423737049102783,
      "learning_rate": 0.0005675400886125828,
      "loss": 2.2496,
      "step": 34432
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6392616033554077,
      "learning_rate": 0.0005675382378977059,
      "loss": 2.3008,
      "step": 34433
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6419064998626709,
      "learning_rate": 0.0005675363871330884,
      "loss": 2.1251,
      "step": 34434
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.7629125714302063,
      "learning_rate": 0.0005675345363187308,
      "loss": 2.2517,
      "step": 34435
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.7535486817359924,
      "learning_rate": 0.0005675326854546336,
      "loss": 2.2618,
      "step": 34436
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6971542239189148,
      "learning_rate": 0.000567530834540797,
      "loss": 2.2819,
      "step": 34437
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.7228991389274597,
      "learning_rate": 0.0005675289835772213,
      "loss": 2.4207,
      "step": 34438
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.7447221279144287,
      "learning_rate": 0.000567527132563907,
      "loss": 2.4004,
      "step": 34439
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6742976903915405,
      "learning_rate": 0.0005675252815008543,
      "loss": 2.2516,
      "step": 34440
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6588144302368164,
      "learning_rate": 0.0005675234303880636,
      "loss": 2.4239,
      "step": 34441
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6978256702423096,
      "learning_rate": 0.0005675215792255352,
      "loss": 2.279,
      "step": 34442
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6922926306724548,
      "learning_rate": 0.0005675197280132696,
      "loss": 2.271,
      "step": 34443
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.7692676782608032,
      "learning_rate": 0.000567517876751267,
      "loss": 2.2795,
      "step": 34444
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6864952445030212,
      "learning_rate": 0.0005675160254395278,
      "loss": 2.0721,
      "step": 34445
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.649269700050354,
      "learning_rate": 0.0005675141740780523,
      "loss": 2.2739,
      "step": 34446
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.7054382562637329,
      "learning_rate": 0.0005675123226668408,
      "loss": 2.2585,
      "step": 34447
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6553884148597717,
      "learning_rate": 0.0005675104712058938,
      "loss": 2.3385,
      "step": 34448
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.666968047618866,
      "learning_rate": 0.0005675086196952114,
      "loss": 2.1998,
      "step": 34449
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6343622207641602,
      "learning_rate": 0.0005675067681347942,
      "loss": 2.3945,
      "step": 34450
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6942745447158813,
      "learning_rate": 0.0005675049165246425,
      "loss": 2.2025,
      "step": 34451
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6595121026039124,
      "learning_rate": 0.0005675030648647565,
      "loss": 2.4101,
      "step": 34452
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6139227151870728,
      "learning_rate": 0.0005675012131551366,
      "loss": 2.2855,
      "step": 34453
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.7355943918228149,
      "learning_rate": 0.0005674993613957834,
      "loss": 2.4447,
      "step": 34454
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6475505232810974,
      "learning_rate": 0.0005674975095866968,
      "loss": 2.3511,
      "step": 34455
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.675264835357666,
      "learning_rate": 0.0005674956577278774,
      "loss": 2.3829,
      "step": 34456
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6383558511734009,
      "learning_rate": 0.0005674938058193255,
      "loss": 2.5191,
      "step": 34457
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6441044211387634,
      "learning_rate": 0.0005674919538610415,
      "loss": 2.3062,
      "step": 34458
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6490716338157654,
      "learning_rate": 0.0005674901018530257,
      "loss": 2.3901,
      "step": 34459
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.666573703289032,
      "learning_rate": 0.0005674882497952784,
      "loss": 2.4718,
      "step": 34460
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6933552026748657,
      "learning_rate": 0.0005674863976878,
      "loss": 2.4345,
      "step": 34461
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6494178771972656,
      "learning_rate": 0.0005674845455305909,
      "loss": 2.1068,
      "step": 34462
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6326073408126831,
      "learning_rate": 0.0005674826933236512,
      "loss": 2.3857,
      "step": 34463
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6958487629890442,
      "learning_rate": 0.0005674808410669816,
      "loss": 2.0991,
      "step": 34464
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6686784625053406,
      "learning_rate": 0.0005674789887605821,
      "loss": 2.3847,
      "step": 34465
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.7036904096603394,
      "learning_rate": 0.0005674771364044534,
      "loss": 2.2675,
      "step": 34466
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6614261269569397,
      "learning_rate": 0.0005674752839985955,
      "loss": 2.3476,
      "step": 34467
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6391754746437073,
      "learning_rate": 0.000567473431543009,
      "loss": 2.5378,
      "step": 34468
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6859075427055359,
      "learning_rate": 0.000567471579037694,
      "loss": 2.4031,
      "step": 34469
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6897574067115784,
      "learning_rate": 0.0005674697264826511,
      "loss": 2.4171,
      "step": 34470
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6147283911705017,
      "learning_rate": 0.0005674678738778805,
      "loss": 2.243,
      "step": 34471
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6658892035484314,
      "learning_rate": 0.0005674660212233826,
      "loss": 2.0793,
      "step": 34472
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6956170797348022,
      "learning_rate": 0.0005674641685191577,
      "loss": 2.2271,
      "step": 34473
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.621929943561554,
      "learning_rate": 0.0005674623157652061,
      "loss": 2.1686,
      "step": 34474
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6435209512710571,
      "learning_rate": 0.0005674604629615283,
      "loss": 2.2344,
      "step": 34475
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.7579225301742554,
      "learning_rate": 0.0005674586101081245,
      "loss": 2.3703,
      "step": 34476
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.872083306312561,
      "learning_rate": 0.000567456757204995,
      "loss": 2.5891,
      "step": 34477
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6607376337051392,
      "learning_rate": 0.0005674549042521404,
      "loss": 2.3824,
      "step": 34478
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6345958113670349,
      "learning_rate": 0.0005674530512495609,
      "loss": 2.3253,
      "step": 34479
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6790000200271606,
      "learning_rate": 0.0005674511981972567,
      "loss": 2.2613,
      "step": 34480
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6595168113708496,
      "learning_rate": 0.0005674493450952284,
      "loss": 2.4131,
      "step": 34481
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.645666778087616,
      "learning_rate": 0.0005674474919434761,
      "loss": 2.1846,
      "step": 34482
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6331266760826111,
      "learning_rate": 0.0005674456387420003,
      "loss": 2.1458,
      "step": 34483
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6509013175964355,
      "learning_rate": 0.0005674437854908012,
      "loss": 2.3801,
      "step": 34484
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6241033673286438,
      "learning_rate": 0.0005674419321898795,
      "loss": 2.5161,
      "step": 34485
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6562449336051941,
      "learning_rate": 0.0005674400788392352,
      "loss": 2.2613,
      "step": 34486
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6312967538833618,
      "learning_rate": 0.0005674382254388687,
      "loss": 2.4996,
      "step": 34487
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6596043705940247,
      "learning_rate": 0.0005674363719887803,
      "loss": 2.1504,
      "step": 34488
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6343567967414856,
      "learning_rate": 0.0005674345184889706,
      "loss": 2.4181,
      "step": 34489
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6819106340408325,
      "learning_rate": 0.0005674326649394396,
      "loss": 2.3929,
      "step": 34490
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6970973014831543,
      "learning_rate": 0.000567430811340188,
      "loss": 2.1967,
      "step": 34491
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6516939997673035,
      "learning_rate": 0.000567428957691216,
      "loss": 2.2483,
      "step": 34492
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6262858510017395,
      "learning_rate": 0.0005674271039925237,
      "loss": 2.3775,
      "step": 34493
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.7469517588615417,
      "learning_rate": 0.0005674252502441118,
      "loss": 2.2688,
      "step": 34494
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.749605119228363,
      "learning_rate": 0.0005674233964459804,
      "loss": 2.1433,
      "step": 34495
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.7183961272239685,
      "learning_rate": 0.00056742154259813,
      "loss": 2.4265,
      "step": 34496
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6421123743057251,
      "learning_rate": 0.0005674196887005609,
      "loss": 2.2221,
      "step": 34497
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.670177161693573,
      "learning_rate": 0.0005674178347532733,
      "loss": 2.4127,
      "step": 34498
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6288686990737915,
      "learning_rate": 0.0005674159807562678,
      "loss": 2.2664,
      "step": 34499
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6724554896354675,
      "learning_rate": 0.0005674141267095446,
      "loss": 2.2772,
      "step": 34500
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.682793378829956,
      "learning_rate": 0.0005674122726131041,
      "loss": 2.1537,
      "step": 34501
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6189277172088623,
      "learning_rate": 0.0005674104184669466,
      "loss": 2.1061,
      "step": 34502
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.644250750541687,
      "learning_rate": 0.0005674085642710724,
      "loss": 1.9581,
      "step": 34503
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6651155352592468,
      "learning_rate": 0.000567406710025482,
      "loss": 2.287,
      "step": 34504
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.89980548620224,
      "learning_rate": 0.0005674048557301756,
      "loss": 2.4011,
      "step": 34505
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.7633140683174133,
      "learning_rate": 0.0005674030013851535,
      "loss": 2.1391,
      "step": 34506
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.9480150938034058,
      "learning_rate": 0.0005674011469904162,
      "loss": 2.2993,
      "step": 34507
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6798726320266724,
      "learning_rate": 0.000567399292545964,
      "loss": 2.3477,
      "step": 34508
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6449140310287476,
      "learning_rate": 0.0005673974380517972,
      "loss": 2.2483,
      "step": 34509
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.7296728491783142,
      "learning_rate": 0.0005673955835079162,
      "loss": 2.2288,
      "step": 34510
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.7589389085769653,
      "learning_rate": 0.0005673937289143213,
      "loss": 2.2451,
      "step": 34511
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6467874050140381,
      "learning_rate": 0.0005673918742710128,
      "loss": 2.1448,
      "step": 34512
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.70428866147995,
      "learning_rate": 0.0005673900195779912,
      "loss": 2.3666,
      "step": 34513
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6938491463661194,
      "learning_rate": 0.0005673881648352566,
      "loss": 2.3435,
      "step": 34514
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6836172938346863,
      "learning_rate": 0.0005673863100428097,
      "loss": 2.3243,
      "step": 34515
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.626937747001648,
      "learning_rate": 0.0005673844552006505,
      "loss": 2.4133,
      "step": 34516
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6663854718208313,
      "learning_rate": 0.0005673826003087795,
      "loss": 2.2343,
      "step": 34517
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6615282893180847,
      "learning_rate": 0.000567380745367197,
      "loss": 2.3366,
      "step": 34518
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.7079686522483826,
      "learning_rate": 0.0005673788903759034,
      "loss": 2.4024,
      "step": 34519
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6701104044914246,
      "learning_rate": 0.0005673770353348991,
      "loss": 2.2895,
      "step": 34520
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6617783904075623,
      "learning_rate": 0.0005673751802441843,
      "loss": 2.2633,
      "step": 34521
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6658505797386169,
      "learning_rate": 0.0005673733251037594,
      "loss": 2.3422,
      "step": 34522
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6402133107185364,
      "learning_rate": 0.0005673714699136247,
      "loss": 2.5166,
      "step": 34523
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.5751636624336243,
      "learning_rate": 0.0005673696146737807,
      "loss": 2.2414,
      "step": 34524
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.7320370078086853,
      "learning_rate": 0.0005673677593842275,
      "loss": 2.3399,
      "step": 34525
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.7206665873527527,
      "learning_rate": 0.0005673659040449657,
      "loss": 2.3743,
      "step": 34526
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.711746871471405,
      "learning_rate": 0.0005673640486559956,
      "loss": 2.2804,
      "step": 34527
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6081081628799438,
      "learning_rate": 0.0005673621932173173,
      "loss": 2.2037,
      "step": 34528
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6259236931800842,
      "learning_rate": 0.0005673603377289315,
      "loss": 2.1391,
      "step": 34529
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6893460154533386,
      "learning_rate": 0.0005673584821908382,
      "loss": 2.2673,
      "step": 34530
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.7276840806007385,
      "learning_rate": 0.000567356626603038,
      "loss": 2.2504,
      "step": 34531
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.7627342939376831,
      "learning_rate": 0.0005673547709655312,
      "loss": 2.296,
      "step": 34532
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.7071145176887512,
      "learning_rate": 0.000567352915278318,
      "loss": 2.4444,
      "step": 34533
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6169103980064392,
      "learning_rate": 0.0005673510595413988,
      "loss": 2.338,
      "step": 34534
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6846855282783508,
      "learning_rate": 0.0005673492037547741,
      "loss": 2.3161,
      "step": 34535
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.8015573024749756,
      "learning_rate": 0.0005673473479184441,
      "loss": 2.1732,
      "step": 34536
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.7647144794464111,
      "learning_rate": 0.0005673454920324092,
      "loss": 2.0492,
      "step": 34537
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.7291778922080994,
      "learning_rate": 0.0005673436360966697,
      "loss": 2.2837,
      "step": 34538
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.7077193856239319,
      "learning_rate": 0.000567341780111226,
      "loss": 2.1785,
      "step": 34539
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6756022572517395,
      "learning_rate": 0.0005673399240760785,
      "loss": 2.4517,
      "step": 34540
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.913232684135437,
      "learning_rate": 0.0005673380679912274,
      "loss": 2.3026,
      "step": 34541
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6860756278038025,
      "learning_rate": 0.000567336211856673,
      "loss": 2.1752,
      "step": 34542
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6723076701164246,
      "learning_rate": 0.0005673343556724158,
      "loss": 2.363,
      "step": 34543
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.7600238919258118,
      "learning_rate": 0.0005673324994384561,
      "loss": 2.3423,
      "step": 34544
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.7169156670570374,
      "learning_rate": 0.0005673306431547944,
      "loss": 2.462,
      "step": 34545
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6453255414962769,
      "learning_rate": 0.0005673287868214307,
      "loss": 2.1674,
      "step": 34546
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6111176013946533,
      "learning_rate": 0.0005673269304383655,
      "loss": 2.3345,
      "step": 34547
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.7748019695281982,
      "learning_rate": 0.0005673250740055993,
      "loss": 2.0938,
      "step": 34548
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6207349896430969,
      "learning_rate": 0.0005673232175231323,
      "loss": 2.4701,
      "step": 34549
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.7226444482803345,
      "learning_rate": 0.0005673213609909648,
      "loss": 2.1553,
      "step": 34550
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6179142594337463,
      "learning_rate": 0.0005673195044090973,
      "loss": 2.3138,
      "step": 34551
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.707035481929779,
      "learning_rate": 0.0005673176477775301,
      "loss": 2.5181,
      "step": 34552
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6668399572372437,
      "learning_rate": 0.0005673157910962635,
      "loss": 2.3112,
      "step": 34553
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.7994719743728638,
      "learning_rate": 0.0005673139343652978,
      "loss": 2.5048,
      "step": 34554
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.8929895758628845,
      "learning_rate": 0.0005673120775846333,
      "loss": 2.5003,
      "step": 34555
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6197819113731384,
      "learning_rate": 0.0005673102207542705,
      "loss": 2.1652,
      "step": 34556
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.7215643525123596,
      "learning_rate": 0.0005673083638742097,
      "loss": 2.3408,
      "step": 34557
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.762405276298523,
      "learning_rate": 0.0005673065069444512,
      "loss": 2.3347,
      "step": 34558
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.667369544506073,
      "learning_rate": 0.0005673046499649954,
      "loss": 2.5978,
      "step": 34559
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6351240873336792,
      "learning_rate": 0.0005673027929358426,
      "loss": 2.2947,
      "step": 34560
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6756678819656372,
      "learning_rate": 0.0005673009358569932,
      "loss": 2.2878,
      "step": 34561
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.7480361461639404,
      "learning_rate": 0.0005672990787284474,
      "loss": 2.5365,
      "step": 34562
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6745439767837524,
      "learning_rate": 0.0005672972215502058,
      "loss": 2.1227,
      "step": 34563
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.7074769139289856,
      "learning_rate": 0.0005672953643222685,
      "loss": 2.3365,
      "step": 34564
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.7174203395843506,
      "learning_rate": 0.0005672935070446361,
      "loss": 2.2929,
      "step": 34565
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.7093020081520081,
      "learning_rate": 0.0005672916497173086,
      "loss": 2.194,
      "step": 34566
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6166104078292847,
      "learning_rate": 0.0005672897923402867,
      "loss": 2.2388,
      "step": 34567
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.689505398273468,
      "learning_rate": 0.0005672879349135705,
      "loss": 2.4026,
      "step": 34568
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6571313738822937,
      "learning_rate": 0.0005672860774371604,
      "loss": 2.5047,
      "step": 34569
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6607669591903687,
      "learning_rate": 0.0005672842199110569,
      "loss": 2.2448,
      "step": 34570
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6354832649230957,
      "learning_rate": 0.0005672823623352601,
      "loss": 2.4763,
      "step": 34571
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.604641318321228,
      "learning_rate": 0.0005672805047097705,
      "loss": 2.2162,
      "step": 34572
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.7071534991264343,
      "learning_rate": 0.0005672786470345884,
      "loss": 2.2166,
      "step": 34573
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6662706732749939,
      "learning_rate": 0.0005672767893097141,
      "loss": 2.4156,
      "step": 34574
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6282561421394348,
      "learning_rate": 0.000567274931535148,
      "loss": 2.1209,
      "step": 34575
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.725814938545227,
      "learning_rate": 0.0005672730737108906,
      "loss": 2.2967,
      "step": 34576
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.7056785225868225,
      "learning_rate": 0.0005672712158369419,
      "loss": 2.28,
      "step": 34577
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6877421736717224,
      "learning_rate": 0.0005672693579133026,
      "loss": 2.5715,
      "step": 34578
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.8015261888504028,
      "learning_rate": 0.0005672674999399728,
      "loss": 2.3156,
      "step": 34579
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6761755347251892,
      "learning_rate": 0.0005672656419169529,
      "loss": 2.3326,
      "step": 34580
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6325564384460449,
      "learning_rate": 0.0005672637838442433,
      "loss": 2.169,
      "step": 34581
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.7292106747627258,
      "learning_rate": 0.0005672619257218443,
      "loss": 2.3807,
      "step": 34582
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6829279065132141,
      "learning_rate": 0.0005672600675497564,
      "loss": 2.2023,
      "step": 34583
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6178924441337585,
      "learning_rate": 0.0005672582093279796,
      "loss": 2.2016,
      "step": 34584
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6531779170036316,
      "learning_rate": 0.0005672563510565146,
      "loss": 2.2991,
      "step": 34585
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.7040098309516907,
      "learning_rate": 0.0005672544927353615,
      "loss": 2.359,
      "step": 34586
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6686004400253296,
      "learning_rate": 0.0005672526343645208,
      "loss": 2.3348,
      "step": 34587
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.645750105381012,
      "learning_rate": 0.0005672507759439928,
      "loss": 2.4615,
      "step": 34588
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6409333348274231,
      "learning_rate": 0.0005672489174737777,
      "loss": 2.2601,
      "step": 34589
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6113208532333374,
      "learning_rate": 0.0005672470589538762,
      "loss": 2.2491,
      "step": 34590
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.7244918346405029,
      "learning_rate": 0.0005672452003842883,
      "loss": 2.3913,
      "step": 34591
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6573916673660278,
      "learning_rate": 0.0005672433417650146,
      "loss": 2.2693,
      "step": 34592
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6764267683029175,
      "learning_rate": 0.0005672414830960551,
      "loss": 2.1263,
      "step": 34593
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6612220406532288,
      "learning_rate": 0.0005672396243774105,
      "loss": 2.2965,
      "step": 34594
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.7155401110649109,
      "learning_rate": 0.000567237765609081,
      "loss": 2.2822,
      "step": 34595
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6921924352645874,
      "learning_rate": 0.000567235906791067,
      "loss": 2.5707,
      "step": 34596
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6573867201805115,
      "learning_rate": 0.0005672340479233687,
      "loss": 2.2257,
      "step": 34597
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6738526225090027,
      "learning_rate": 0.0005672321890059866,
      "loss": 2.0858,
      "step": 34598
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.7157226204872131,
      "learning_rate": 0.000567230330038921,
      "loss": 2.3026,
      "step": 34599
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.7542508840560913,
      "learning_rate": 0.0005672284710221722,
      "loss": 2.2689,
      "step": 34600
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6084601283073425,
      "learning_rate": 0.0005672266119557406,
      "loss": 2.1193,
      "step": 34601
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6457217335700989,
      "learning_rate": 0.0005672247528396265,
      "loss": 2.347,
      "step": 34602
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6439046859741211,
      "learning_rate": 0.0005672228936738304,
      "loss": 2.3484,
      "step": 34603
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.7210121154785156,
      "learning_rate": 0.0005672210344583524,
      "loss": 2.3093,
      "step": 34604
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6732489466667175,
      "learning_rate": 0.000567219175193193,
      "loss": 2.3808,
      "step": 34605
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6647782921791077,
      "learning_rate": 0.0005672173158783525,
      "loss": 2.1136,
      "step": 34606
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.7216984629631042,
      "learning_rate": 0.0005672154565138312,
      "loss": 2.266,
      "step": 34607
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6351798176765442,
      "learning_rate": 0.0005672135970996296,
      "loss": 2.2674,
      "step": 34608
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6379204392433167,
      "learning_rate": 0.0005672117376357479,
      "loss": 2.1558,
      "step": 34609
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6358300447463989,
      "learning_rate": 0.0005672098781221866,
      "loss": 2.3644,
      "step": 34610
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.7971452474594116,
      "learning_rate": 0.0005672080185589458,
      "loss": 2.3232,
      "step": 34611
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.5962719917297363,
      "learning_rate": 0.0005672061589460259,
      "loss": 2.3364,
      "step": 34612
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6593006253242493,
      "learning_rate": 0.0005672042992834275,
      "loss": 2.5051,
      "step": 34613
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6640075445175171,
      "learning_rate": 0.0005672024395711508,
      "loss": 2.2275,
      "step": 34614
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.7161383628845215,
      "learning_rate": 0.0005672005798091961,
      "loss": 2.0385,
      "step": 34615
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6548115611076355,
      "learning_rate": 0.0005671987199975636,
      "loss": 2.137,
      "step": 34616
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.7322176098823547,
      "learning_rate": 0.0005671968601362541,
      "loss": 2.3159,
      "step": 34617
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.7387984991073608,
      "learning_rate": 0.0005671950002252674,
      "loss": 2.0973,
      "step": 34618
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.5895375609397888,
      "learning_rate": 0.0005671931402646042,
      "loss": 2.2511,
      "step": 34619
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6064035892486572,
      "learning_rate": 0.0005671912802542648,
      "loss": 2.2536,
      "step": 34620
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.685920774936676,
      "learning_rate": 0.0005671894201942495,
      "loss": 2.253,
      "step": 34621
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6299330592155457,
      "learning_rate": 0.0005671875600845585,
      "loss": 2.2178,
      "step": 34622
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6162596344947815,
      "learning_rate": 0.0005671856999251925,
      "loss": 2.3515,
      "step": 34623
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6813088655471802,
      "learning_rate": 0.0005671838397161515,
      "loss": 2.341,
      "step": 34624
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.7559042572975159,
      "learning_rate": 0.0005671819794574359,
      "loss": 2.3425,
      "step": 34625
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6006073951721191,
      "learning_rate": 0.0005671801191490463,
      "loss": 2.2444,
      "step": 34626
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6173654198646545,
      "learning_rate": 0.0005671782587909828,
      "loss": 2.4419,
      "step": 34627
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.637808620929718,
      "learning_rate": 0.0005671763983832458,
      "loss": 2.4219,
      "step": 34628
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6411055326461792,
      "learning_rate": 0.0005671745379258356,
      "loss": 2.6001,
      "step": 34629
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.7093291282653809,
      "learning_rate": 0.0005671726774187528,
      "loss": 2.1968,
      "step": 34630
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.7537139654159546,
      "learning_rate": 0.0005671708168619974,
      "loss": 2.4934,
      "step": 34631
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6897212862968445,
      "learning_rate": 0.00056716895625557,
      "loss": 2.4004,
      "step": 34632
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.8251761198043823,
      "learning_rate": 0.0005671670955994709,
      "loss": 2.4702,
      "step": 34633
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.7028350830078125,
      "learning_rate": 0.0005671652348937002,
      "loss": 2.3415,
      "step": 34634
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.7331756353378296,
      "learning_rate": 0.0005671633741382585,
      "loss": 2.374,
      "step": 34635
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.8022285103797913,
      "learning_rate": 0.0005671615133331461,
      "loss": 2.0817,
      "step": 34636
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.7204460501670837,
      "learning_rate": 0.0005671596524783635,
      "loss": 2.0716,
      "step": 34637
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6519054770469666,
      "learning_rate": 0.0005671577915739107,
      "loss": 2.3286,
      "step": 34638
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.7408285737037659,
      "learning_rate": 0.0005671559306197882,
      "loss": 2.3617,
      "step": 34639
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6903679370880127,
      "learning_rate": 0.0005671540696159964,
      "loss": 2.4894,
      "step": 34640
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.7444426417350769,
      "learning_rate": 0.0005671522085625356,
      "loss": 2.2108,
      "step": 34641
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6565137505531311,
      "learning_rate": 0.0005671503474594063,
      "loss": 2.3198,
      "step": 34642
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6120706796646118,
      "learning_rate": 0.0005671484863066087,
      "loss": 2.3517,
      "step": 34643
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6874878406524658,
      "learning_rate": 0.000567146625104143,
      "loss": 2.4515,
      "step": 34644
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6199076175689697,
      "learning_rate": 0.0005671447638520098,
      "loss": 2.2465,
      "step": 34645
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6433330178260803,
      "learning_rate": 0.0005671429025502092,
      "loss": 2.1093,
      "step": 34646
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.7080862522125244,
      "learning_rate": 0.0005671410411987419,
      "loss": 2.0968,
      "step": 34647
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6491456627845764,
      "learning_rate": 0.0005671391797976078,
      "loss": 2.264,
      "step": 34648
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.714235246181488,
      "learning_rate": 0.0005671373183468077,
      "loss": 2.54,
      "step": 34649
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6495269536972046,
      "learning_rate": 0.0005671354568463417,
      "loss": 2.2318,
      "step": 34650
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6935330033302307,
      "learning_rate": 0.00056713359529621,
      "loss": 2.2727,
      "step": 34651
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.5937506556510925,
      "learning_rate": 0.0005671317336964133,
      "loss": 2.1693,
      "step": 34652
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.7382925748825073,
      "learning_rate": 0.0005671298720469518,
      "loss": 2.0481,
      "step": 34653
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6571764349937439,
      "learning_rate": 0.0005671280103478256,
      "loss": 2.349,
      "step": 34654
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.7096192240715027,
      "learning_rate": 0.0005671261485990355,
      "loss": 2.2609,
      "step": 34655
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6807627081871033,
      "learning_rate": 0.0005671242868005815,
      "loss": 2.5098,
      "step": 34656
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.7017543315887451,
      "learning_rate": 0.000567122424952464,
      "loss": 2.3627,
      "step": 34657
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.675238847732544,
      "learning_rate": 0.0005671205630546834,
      "loss": 2.1478,
      "step": 34658
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6684949398040771,
      "learning_rate": 0.0005671187011072401,
      "loss": 2.1497,
      "step": 34659
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.697784423828125,
      "learning_rate": 0.0005671168391101343,
      "loss": 2.3873,
      "step": 34660
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6978986859321594,
      "learning_rate": 0.0005671149770633667,
      "loss": 2.3323,
      "step": 34661
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.70200115442276,
      "learning_rate": 0.0005671131149669372,
      "loss": 2.4133,
      "step": 34662
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.7229363918304443,
      "learning_rate": 0.0005671112528208463,
      "loss": 2.3072,
      "step": 34663
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6431347727775574,
      "learning_rate": 0.0005671093906250944,
      "loss": 2.2857,
      "step": 34664
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6715978980064392,
      "learning_rate": 0.0005671075283796819,
      "loss": 2.2153,
      "step": 34665
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.7791070342063904,
      "learning_rate": 0.000567105666084609,
      "loss": 2.4479,
      "step": 34666
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.7361433506011963,
      "learning_rate": 0.0005671038037398761,
      "loss": 2.2289,
      "step": 34667
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6293907165527344,
      "learning_rate": 0.0005671019413454835,
      "loss": 2.2074,
      "step": 34668
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.7372571229934692,
      "learning_rate": 0.0005671000789014318,
      "loss": 2.2029,
      "step": 34669
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6142002940177917,
      "learning_rate": 0.000567098216407721,
      "loss": 2.2356,
      "step": 34670
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6416676640510559,
      "learning_rate": 0.0005670963538643517,
      "loss": 2.2902,
      "step": 34671
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6038126349449158,
      "learning_rate": 0.0005670944912713239,
      "loss": 2.1374,
      "step": 34672
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.7381772398948669,
      "learning_rate": 0.0005670926286286385,
      "loss": 2.1179,
      "step": 34673
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6923967003822327,
      "learning_rate": 0.0005670907659362954,
      "loss": 2.083,
      "step": 34674
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6573684215545654,
      "learning_rate": 0.0005670889031942952,
      "loss": 2.1938,
      "step": 34675
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.647915780544281,
      "learning_rate": 0.0005670870404026379,
      "loss": 2.2363,
      "step": 34676
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.7194434404373169,
      "learning_rate": 0.0005670851775613242,
      "loss": 2.2062,
      "step": 34677
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.8052641749382019,
      "learning_rate": 0.0005670833146703544,
      "loss": 2.128,
      "step": 34678
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.7219235301017761,
      "learning_rate": 0.0005670814517297287,
      "loss": 2.4606,
      "step": 34679
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.7490527033805847,
      "learning_rate": 0.0005670795887394474,
      "loss": 2.4075,
      "step": 34680
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.839529812335968,
      "learning_rate": 0.0005670777256995111,
      "loss": 2.3801,
      "step": 34681
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.7593011260032654,
      "learning_rate": 0.0005670758626099201,
      "loss": 2.337,
      "step": 34682
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6265972256660461,
      "learning_rate": 0.0005670739994706745,
      "loss": 2.4255,
      "step": 34683
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.8451895117759705,
      "learning_rate": 0.0005670721362817749,
      "loss": 2.2001,
      "step": 34684
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6812644004821777,
      "learning_rate": 0.0005670702730432215,
      "loss": 2.3061,
      "step": 34685
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6958006620407104,
      "learning_rate": 0.0005670684097550146,
      "loss": 2.2533,
      "step": 34686
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.7072268724441528,
      "learning_rate": 0.0005670665464171548,
      "loss": 2.3642,
      "step": 34687
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.7446666359901428,
      "learning_rate": 0.0005670646830296422,
      "loss": 2.2259,
      "step": 34688
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.66982102394104,
      "learning_rate": 0.0005670628195924772,
      "loss": 2.5281,
      "step": 34689
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.7405992746353149,
      "learning_rate": 0.0005670609561056603,
      "loss": 2.6726,
      "step": 34690
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6563610434532166,
      "learning_rate": 0.0005670590925691916,
      "loss": 2.1949,
      "step": 34691
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.7476971745491028,
      "learning_rate": 0.0005670572289830717,
      "loss": 2.4904,
      "step": 34692
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6712468266487122,
      "learning_rate": 0.0005670553653473007,
      "loss": 2.5492,
      "step": 34693
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.655076801776886,
      "learning_rate": 0.0005670535016618792,
      "loss": 2.3102,
      "step": 34694
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.7114708423614502,
      "learning_rate": 0.0005670516379268072,
      "loss": 2.3299,
      "step": 34695
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6469763517379761,
      "learning_rate": 0.0005670497741420855,
      "loss": 2.183,
      "step": 34696
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.669295608997345,
      "learning_rate": 0.0005670479103077142,
      "loss": 2.2504,
      "step": 34697
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6798837184906006,
      "learning_rate": 0.0005670460464236934,
      "loss": 2.2904,
      "step": 34698
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6456013321876526,
      "learning_rate": 0.0005670441824900239,
      "loss": 2.0248,
      "step": 34699
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6288231015205383,
      "learning_rate": 0.0005670423185067058,
      "loss": 2.0765,
      "step": 34700
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6338881254196167,
      "learning_rate": 0.0005670404544737396,
      "loss": 2.2891,
      "step": 34701
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.7205584049224854,
      "learning_rate": 0.0005670385903911254,
      "loss": 2.3873,
      "step": 34702
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.7099311947822571,
      "learning_rate": 0.0005670367262588636,
      "loss": 2.5283,
      "step": 34703
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.7046894431114197,
      "learning_rate": 0.0005670348620769549,
      "loss": 2.3512,
      "step": 34704
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.7059882283210754,
      "learning_rate": 0.0005670329978453992,
      "loss": 2.1915,
      "step": 34705
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6721441745758057,
      "learning_rate": 0.000567031133564197,
      "loss": 2.4957,
      "step": 34706
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6399803757667542,
      "learning_rate": 0.0005670292692333488,
      "loss": 2.3651,
      "step": 34707
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6909265518188477,
      "learning_rate": 0.0005670274048528547,
      "loss": 2.1724,
      "step": 34708
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6980966329574585,
      "learning_rate": 0.0005670255404227152,
      "loss": 2.4281,
      "step": 34709
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6899728775024414,
      "learning_rate": 0.0005670236759429306,
      "loss": 2.2863,
      "step": 34710
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.7001033425331116,
      "learning_rate": 0.0005670218114135013,
      "loss": 2.5087,
      "step": 34711
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.8210636973381042,
      "learning_rate": 0.0005670199468344276,
      "loss": 2.2169,
      "step": 34712
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.611944854259491,
      "learning_rate": 0.0005670180822057098,
      "loss": 2.1628,
      "step": 34713
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.7144454717636108,
      "learning_rate": 0.0005670162175273484,
      "loss": 2.4627,
      "step": 34714
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6948007941246033,
      "learning_rate": 0.0005670143527993436,
      "loss": 2.1815,
      "step": 34715
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6708950400352478,
      "learning_rate": 0.0005670124880216956,
      "loss": 2.2835,
      "step": 34716
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6611586809158325,
      "learning_rate": 0.0005670106231944052,
      "loss": 2.3515,
      "step": 34717
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.7584460377693176,
      "learning_rate": 0.0005670087583174723,
      "loss": 2.387,
      "step": 34718
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.7065868377685547,
      "learning_rate": 0.0005670068933908975,
      "loss": 2.502,
      "step": 34719
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6753739714622498,
      "learning_rate": 0.0005670050284146812,
      "loss": 2.2523,
      "step": 34720
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.8490135669708252,
      "learning_rate": 0.0005670031633888236,
      "loss": 2.4522,
      "step": 34721
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6957083940505981,
      "learning_rate": 0.0005670012983133249,
      "loss": 2.4451,
      "step": 34722
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.64883953332901,
      "learning_rate": 0.0005669994331881857,
      "loss": 2.0436,
      "step": 34723
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.7167089581489563,
      "learning_rate": 0.0005669975680134062,
      "loss": 2.3827,
      "step": 34724
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6379024982452393,
      "learning_rate": 0.000566995702788987,
      "loss": 2.3538,
      "step": 34725
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6776820421218872,
      "learning_rate": 0.0005669938375149282,
      "loss": 2.4509,
      "step": 34726
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6904377937316895,
      "learning_rate": 0.0005669919721912302,
      "loss": 2.2676,
      "step": 34727
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6333205103874207,
      "learning_rate": 0.0005669901068178933,
      "loss": 2.2666,
      "step": 34728
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6357371211051941,
      "learning_rate": 0.0005669882413949179,
      "loss": 2.3087,
      "step": 34729
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6331722140312195,
      "learning_rate": 0.0005669863759223044,
      "loss": 2.2819,
      "step": 34730
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.7790703177452087,
      "learning_rate": 0.0005669845104000529,
      "loss": 2.5986,
      "step": 34731
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6787206530570984,
      "learning_rate": 0.0005669826448281642,
      "loss": 2.3186,
      "step": 34732
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6831234097480774,
      "learning_rate": 0.0005669807792066383,
      "loss": 2.2869,
      "step": 34733
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.7146177291870117,
      "learning_rate": 0.0005669789135354756,
      "loss": 2.3787,
      "step": 34734
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6631264686584473,
      "learning_rate": 0.0005669770478146765,
      "loss": 2.4066,
      "step": 34735
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.7056631445884705,
      "learning_rate": 0.0005669751820442413,
      "loss": 2.2598,
      "step": 34736
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6297789216041565,
      "learning_rate": 0.0005669733162241704,
      "loss": 2.2522,
      "step": 34737
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.7288181781768799,
      "learning_rate": 0.0005669714503544641,
      "loss": 2.3215,
      "step": 34738
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.7375115156173706,
      "learning_rate": 0.0005669695844351228,
      "loss": 2.1695,
      "step": 34739
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.7887977957725525,
      "learning_rate": 0.0005669677184661467,
      "loss": 2.2448,
      "step": 34740
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.7258148193359375,
      "learning_rate": 0.0005669658524475364,
      "loss": 2.2593,
      "step": 34741
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6672478318214417,
      "learning_rate": 0.000566963986379292,
      "loss": 2.2365,
      "step": 34742
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.9525560736656189,
      "learning_rate": 0.0005669621202614141,
      "loss": 2.3896,
      "step": 34743
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.683748722076416,
      "learning_rate": 0.0005669602540939029,
      "loss": 2.2888,
      "step": 34744
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.7021905779838562,
      "learning_rate": 0.0005669583878767586,
      "loss": 2.1838,
      "step": 34745
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.9061443209648132,
      "learning_rate": 0.0005669565216099817,
      "loss": 2.2957,
      "step": 34746
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6654444932937622,
      "learning_rate": 0.0005669546552935727,
      "loss": 2.4055,
      "step": 34747
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.650793731212616,
      "learning_rate": 0.0005669527889275316,
      "loss": 2.2703,
      "step": 34748
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.781135618686676,
      "learning_rate": 0.0005669509225118591,
      "loss": 2.222,
      "step": 34749
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.77381432056427,
      "learning_rate": 0.0005669490560465553,
      "loss": 2.4042,
      "step": 34750
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6105101704597473,
      "learning_rate": 0.0005669471895316207,
      "loss": 2.3838,
      "step": 34751
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6917226314544678,
      "learning_rate": 0.0005669453229670556,
      "loss": 2.5744,
      "step": 34752
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.7032342553138733,
      "learning_rate": 0.0005669434563528602,
      "loss": 2.3517,
      "step": 34753
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6692589521408081,
      "learning_rate": 0.000566941589689035,
      "loss": 2.2892,
      "step": 34754
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6313486099243164,
      "learning_rate": 0.0005669397229755804,
      "loss": 2.2625,
      "step": 34755
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.691217839717865,
      "learning_rate": 0.0005669378562124966,
      "loss": 2.3668,
      "step": 34756
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.7661077380180359,
      "learning_rate": 0.000566935989399784,
      "loss": 2.5469,
      "step": 34757
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6937757730484009,
      "learning_rate": 0.000566934122537443,
      "loss": 2.3943,
      "step": 34758
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.7037566900253296,
      "learning_rate": 0.0005669322556254739,
      "loss": 2.4382,
      "step": 34759
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.7079178094863892,
      "learning_rate": 0.000566930388663877,
      "loss": 2.4094,
      "step": 34760
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.7011724710464478,
      "learning_rate": 0.0005669285216526529,
      "loss": 2.5604,
      "step": 34761
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.7059173583984375,
      "learning_rate": 0.0005669266545918015,
      "loss": 2.2518,
      "step": 34762
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6630992889404297,
      "learning_rate": 0.0005669247874813235,
      "loss": 2.3403,
      "step": 34763
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6085999608039856,
      "learning_rate": 0.0005669229203212192,
      "loss": 2.3736,
      "step": 34764
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6100518107414246,
      "learning_rate": 0.0005669210531114888,
      "loss": 2.2501,
      "step": 34765
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6687663793563843,
      "learning_rate": 0.0005669191858521328,
      "loss": 2.2358,
      "step": 34766
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.7518497705459595,
      "learning_rate": 0.0005669173185431514,
      "loss": 2.3257,
      "step": 34767
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6711109280586243,
      "learning_rate": 0.0005669154511845451,
      "loss": 2.2599,
      "step": 34768
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.7427060604095459,
      "learning_rate": 0.0005669135837763141,
      "loss": 2.2885,
      "step": 34769
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6803866028785706,
      "learning_rate": 0.000566911716318459,
      "loss": 2.2586,
      "step": 34770
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6466091871261597,
      "learning_rate": 0.0005669098488109799,
      "loss": 2.4078,
      "step": 34771
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6532696485519409,
      "learning_rate": 0.0005669079812538772,
      "loss": 2.0931,
      "step": 34772
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.711111307144165,
      "learning_rate": 0.0005669061136471513,
      "loss": 2.3336,
      "step": 34773
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.7128779292106628,
      "learning_rate": 0.0005669042459908024,
      "loss": 2.3105,
      "step": 34774
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.741656482219696,
      "learning_rate": 0.0005669023782848311,
      "loss": 2.4241,
      "step": 34775
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.663245439529419,
      "learning_rate": 0.0005669005105292375,
      "loss": 2.3106,
      "step": 34776
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.762140691280365,
      "learning_rate": 0.0005668986427240221,
      "loss": 2.2727,
      "step": 34777
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6595812439918518,
      "learning_rate": 0.0005668967748691852,
      "loss": 2.4865,
      "step": 34778
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.7130886912345886,
      "learning_rate": 0.0005668949069647271,
      "loss": 2.2788,
      "step": 34779
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.7466031908988953,
      "learning_rate": 0.0005668930390106483,
      "loss": 2.3601,
      "step": 34780
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6325902938842773,
      "learning_rate": 0.0005668911710069491,
      "loss": 2.3035,
      "step": 34781
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.621534526348114,
      "learning_rate": 0.0005668893029536296,
      "loss": 2.5286,
      "step": 34782
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.8050834536552429,
      "learning_rate": 0.0005668874348506905,
      "loss": 2.1956,
      "step": 34783
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6574848890304565,
      "learning_rate": 0.0005668855666981319,
      "loss": 2.2435,
      "step": 34784
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.8548883199691772,
      "learning_rate": 0.0005668836984959543,
      "loss": 2.1603,
      "step": 34785
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6009894013404846,
      "learning_rate": 0.0005668818302441579,
      "loss": 2.1504,
      "step": 34786
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6554783582687378,
      "learning_rate": 0.0005668799619427432,
      "loss": 2.2502,
      "step": 34787
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.5981587767601013,
      "learning_rate": 0.0005668780935917105,
      "loss": 2.4614,
      "step": 34788
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6975033283233643,
      "learning_rate": 0.00056687622519106,
      "loss": 2.1279,
      "step": 34789
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.7470691800117493,
      "learning_rate": 0.0005668743567407923,
      "loss": 2.4097,
      "step": 34790
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6973981261253357,
      "learning_rate": 0.0005668724882409075,
      "loss": 2.1968,
      "step": 34791
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.682857096195221,
      "learning_rate": 0.0005668706196914061,
      "loss": 2.1872,
      "step": 34792
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6339879631996155,
      "learning_rate": 0.0005668687510922885,
      "loss": 2.212,
      "step": 34793
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.7351874709129333,
      "learning_rate": 0.0005668668824435549,
      "loss": 2.3067,
      "step": 34794
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6501777768135071,
      "learning_rate": 0.0005668650137452058,
      "loss": 2.2101,
      "step": 34795
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6481756567955017,
      "learning_rate": 0.0005668631449972413,
      "loss": 2.441,
      "step": 34796
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6849495768547058,
      "learning_rate": 0.000566861276199662,
      "loss": 2.3314,
      "step": 34797
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6497293710708618,
      "learning_rate": 0.0005668594073524681,
      "loss": 2.4052,
      "step": 34798
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6919543147087097,
      "learning_rate": 0.0005668575384556601,
      "loss": 2.2801,
      "step": 34799
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.774179995059967,
      "learning_rate": 0.0005668556695092382,
      "loss": 2.2736,
      "step": 34800
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6391472816467285,
      "learning_rate": 0.0005668538005132028,
      "loss": 2.1419,
      "step": 34801
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6246402263641357,
      "learning_rate": 0.0005668519314675541,
      "loss": 2.3804,
      "step": 34802
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.7043949365615845,
      "learning_rate": 0.0005668500623722927,
      "loss": 2.3365,
      "step": 34803
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.7263571619987488,
      "learning_rate": 0.0005668481932274189,
      "loss": 2.3349,
      "step": 34804
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.652263879776001,
      "learning_rate": 0.000566846324032933,
      "loss": 2.1985,
      "step": 34805
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6946554183959961,
      "learning_rate": 0.0005668444547888351,
      "loss": 2.2453,
      "step": 34806
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.690796434879303,
      "learning_rate": 0.000566842585495126,
      "loss": 2.4637,
      "step": 34807
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6972289681434631,
      "learning_rate": 0.0005668407161518057,
      "loss": 2.1351,
      "step": 34808
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6563349962234497,
      "learning_rate": 0.0005668388467588748,
      "loss": 2.2599,
      "step": 34809
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.7096195220947266,
      "learning_rate": 0.0005668369773163335,
      "loss": 2.1725,
      "step": 34810
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.7185564041137695,
      "learning_rate": 0.0005668351078241821,
      "loss": 2.4789,
      "step": 34811
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6283711791038513,
      "learning_rate": 0.0005668332382824209,
      "loss": 2.1954,
      "step": 34812
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6847926378250122,
      "learning_rate": 0.0005668313686910505,
      "loss": 2.524,
      "step": 34813
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6329970359802246,
      "learning_rate": 0.0005668294990500712,
      "loss": 2.5241,
      "step": 34814
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6413826942443848,
      "learning_rate": 0.0005668276293594831,
      "loss": 2.2903,
      "step": 34815
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6983715295791626,
      "learning_rate": 0.0005668257596192868,
      "loss": 2.3838,
      "step": 34816
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6623250246047974,
      "learning_rate": 0.0005668238898294826,
      "loss": 2.1349,
      "step": 34817
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6741746068000793,
      "learning_rate": 0.0005668220199900707,
      "loss": 2.3842,
      "step": 34818
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6618444323539734,
      "learning_rate": 0.0005668201501010516,
      "loss": 2.1342,
      "step": 34819
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.7129660248756409,
      "learning_rate": 0.0005668182801624256,
      "loss": 2.1465,
      "step": 34820
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6334959864616394,
      "learning_rate": 0.0005668164101741929,
      "loss": 2.3966,
      "step": 34821
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6584728956222534,
      "learning_rate": 0.0005668145401363543,
      "loss": 2.3303,
      "step": 34822
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6668667793273926,
      "learning_rate": 0.0005668126700489097,
      "loss": 2.3554,
      "step": 34823
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.713732898235321,
      "learning_rate": 0.0005668107999118595,
      "loss": 2.4799,
      "step": 34824
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6237350106239319,
      "learning_rate": 0.0005668089297252042,
      "loss": 2.4164,
      "step": 34825
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6559430956840515,
      "learning_rate": 0.000566807059488944,
      "loss": 2.2213,
      "step": 34826
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6418710350990295,
      "learning_rate": 0.0005668051892030794,
      "loss": 2.3309,
      "step": 34827
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6734462976455688,
      "learning_rate": 0.0005668033188676108,
      "loss": 2.0069,
      "step": 34828
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.5821629166603088,
      "learning_rate": 0.0005668014484825382,
      "loss": 2.3197,
      "step": 34829
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.629152774810791,
      "learning_rate": 0.0005667995780478623,
      "loss": 2.3054,
      "step": 34830
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.713415801525116,
      "learning_rate": 0.0005667977075635833,
      "loss": 2.4546,
      "step": 34831
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6259493231773376,
      "learning_rate": 0.0005667958370297016,
      "loss": 2.1515,
      "step": 34832
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6753761172294617,
      "learning_rate": 0.0005667939664462174,
      "loss": 2.266,
      "step": 34833
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.7078245282173157,
      "learning_rate": 0.0005667920958131313,
      "loss": 2.3842,
      "step": 34834
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.7148733735084534,
      "learning_rate": 0.0005667902251304435,
      "loss": 2.4043,
      "step": 34835
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.631321132183075,
      "learning_rate": 0.0005667883543981543,
      "loss": 2.2773,
      "step": 34836
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.7520961761474609,
      "learning_rate": 0.0005667864836162641,
      "loss": 2.243,
      "step": 34837
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.7112865447998047,
      "learning_rate": 0.0005667846127847733,
      "loss": 2.1317,
      "step": 34838
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6756699085235596,
      "learning_rate": 0.0005667827419036822,
      "loss": 2.4796,
      "step": 34839
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6587023138999939,
      "learning_rate": 0.0005667808709729912,
      "loss": 2.2673,
      "step": 34840
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.852508008480072,
      "learning_rate": 0.0005667789999927006,
      "loss": 2.0298,
      "step": 34841
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6573759913444519,
      "learning_rate": 0.0005667771289628107,
      "loss": 2.1739,
      "step": 34842
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.8180722594261169,
      "learning_rate": 0.0005667752578833218,
      "loss": 2.352,
      "step": 34843
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6718841195106506,
      "learning_rate": 0.0005667733867542344,
      "loss": 2.1348,
      "step": 34844
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6369504928588867,
      "learning_rate": 0.0005667715155755489,
      "loss": 2.3588,
      "step": 34845
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.7941845059394836,
      "learning_rate": 0.0005667696443472655,
      "loss": 2.2746,
      "step": 34846
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.7213273644447327,
      "learning_rate": 0.0005667677730693846,
      "loss": 2.0954,
      "step": 34847
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.5975999236106873,
      "learning_rate": 0.0005667659017419065,
      "loss": 2.4418,
      "step": 34848
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6815730929374695,
      "learning_rate": 0.0005667640303648315,
      "loss": 1.9846,
      "step": 34849
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.8633471131324768,
      "learning_rate": 0.0005667621589381601,
      "loss": 2.3167,
      "step": 34850
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6926096677780151,
      "learning_rate": 0.0005667602874618928,
      "loss": 2.4495,
      "step": 34851
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.7248791456222534,
      "learning_rate": 0.0005667584159360293,
      "loss": 2.2556,
      "step": 34852
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.8464428186416626,
      "learning_rate": 0.0005667565443605707,
      "loss": 2.1158,
      "step": 34853
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6998944878578186,
      "learning_rate": 0.0005667546727355169,
      "loss": 2.2125,
      "step": 34854
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6755159497261047,
      "learning_rate": 0.0005667528010608684,
      "loss": 2.2714,
      "step": 34855
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.8829164505004883,
      "learning_rate": 0.0005667509293366254,
      "loss": 2.2622,
      "step": 34856
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.856199324131012,
      "learning_rate": 0.0005667490575627886,
      "loss": 2.3002,
      "step": 34857
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.7416653037071228,
      "learning_rate": 0.000566747185739358,
      "loss": 2.2436,
      "step": 34858
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.7860561013221741,
      "learning_rate": 0.000566745313866334,
      "loss": 2.3434,
      "step": 34859
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.7065668702125549,
      "learning_rate": 0.0005667434419437171,
      "loss": 2.3753,
      "step": 34860
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6968448758125305,
      "learning_rate": 0.0005667415699715074,
      "loss": 2.1192,
      "step": 34861
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.652250349521637,
      "learning_rate": 0.0005667396979497056,
      "loss": 2.2825,
      "step": 34862
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6848474144935608,
      "learning_rate": 0.0005667378258783117,
      "loss": 2.2049,
      "step": 34863
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.8853832483291626,
      "learning_rate": 0.0005667359537573263,
      "loss": 2.0769,
      "step": 34864
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6977698802947998,
      "learning_rate": 0.0005667340815867498,
      "loss": 2.1953,
      "step": 34865
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.7206510305404663,
      "learning_rate": 0.0005667322093665821,
      "loss": 2.1295,
      "step": 34866
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.9591212868690491,
      "learning_rate": 0.000566730337096824,
      "loss": 2.398,
      "step": 34867
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.7840954661369324,
      "learning_rate": 0.0005667284647774757,
      "loss": 2.0725,
      "step": 34868
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6110488176345825,
      "learning_rate": 0.0005667265924085375,
      "loss": 2.2799,
      "step": 34869
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6569175720214844,
      "learning_rate": 0.0005667247199900098,
      "loss": 2.3198,
      "step": 34870
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.8704648613929749,
      "learning_rate": 0.0005667228475218929,
      "loss": 2.4395,
      "step": 34871
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6825242638587952,
      "learning_rate": 0.0005667209750041872,
      "loss": 2.4768,
      "step": 34872
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6246609687805176,
      "learning_rate": 0.0005667191024368932,
      "loss": 2.1992,
      "step": 34873
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.8738495707511902,
      "learning_rate": 0.000566717229820011,
      "loss": 2.159,
      "step": 34874
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6497191190719604,
      "learning_rate": 0.000566715357153541,
      "loss": 2.3065,
      "step": 34875
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.7531185746192932,
      "learning_rate": 0.0005667134844374834,
      "loss": 2.4092,
      "step": 34876
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6639055609703064,
      "learning_rate": 0.000566711611671839,
      "loss": 2.3893,
      "step": 34877
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6877617239952087,
      "learning_rate": 0.0005667097388566078,
      "loss": 2.2042,
      "step": 34878
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.7620705962181091,
      "learning_rate": 0.0005667078659917902,
      "loss": 2.298,
      "step": 34879
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.8757444620132446,
      "learning_rate": 0.0005667059930773866,
      "loss": 2.4493,
      "step": 34880
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6114325523376465,
      "learning_rate": 0.0005667041201133973,
      "loss": 2.0576,
      "step": 34881
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.7779975533485413,
      "learning_rate": 0.0005667022470998227,
      "loss": 2.2953,
      "step": 34882
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.8387266397476196,
      "learning_rate": 0.000566700374036663,
      "loss": 2.1565,
      "step": 34883
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6447649002075195,
      "learning_rate": 0.0005666985009239188,
      "loss": 2.2303,
      "step": 34884
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.7741183042526245,
      "learning_rate": 0.0005666966277615902,
      "loss": 2.4958,
      "step": 34885
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.9250270128250122,
      "learning_rate": 0.0005666947545496778,
      "loss": 2.3149,
      "step": 34886
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6488602757453918,
      "learning_rate": 0.0005666928812881817,
      "loss": 2.0787,
      "step": 34887
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.8713482022285461,
      "learning_rate": 0.0005666910079771024,
      "loss": 2.4301,
      "step": 34888
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.7384233474731445,
      "learning_rate": 0.0005666891346164402,
      "loss": 2.1534,
      "step": 34889
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6349632143974304,
      "learning_rate": 0.0005666872612061955,
      "loss": 2.3618,
      "step": 34890
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.7649579048156738,
      "learning_rate": 0.0005666853877463686,
      "loss": 2.411,
      "step": 34891
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.8622618913650513,
      "learning_rate": 0.0005666835142369598,
      "loss": 2.4192,
      "step": 34892
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6905128359794617,
      "learning_rate": 0.0005666816406779695,
      "loss": 2.2846,
      "step": 34893
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.7334088087081909,
      "learning_rate": 0.000566679767069398,
      "loss": 2.2975,
      "step": 34894
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.7308374643325806,
      "learning_rate": 0.0005666778934112458,
      "loss": 2.329,
      "step": 34895
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.7112100720405579,
      "learning_rate": 0.0005666760197035131,
      "loss": 2.2396,
      "step": 34896
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.7461674213409424,
      "learning_rate": 0.0005666741459462003,
      "loss": 2.2297,
      "step": 34897
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6739590167999268,
      "learning_rate": 0.0005666722721393078,
      "loss": 2.149,
      "step": 34898
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.7157001495361328,
      "learning_rate": 0.0005666703982828359,
      "loss": 2.1257,
      "step": 34899
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.7834823131561279,
      "learning_rate": 0.0005666685243767848,
      "loss": 2.1572,
      "step": 34900
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6658782958984375,
      "learning_rate": 0.000566666650421155,
      "loss": 2.3508,
      "step": 34901
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6394687294960022,
      "learning_rate": 0.0005666647764159469,
      "loss": 2.0924,
      "step": 34902
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.7527577877044678,
      "learning_rate": 0.000566662902361161,
      "loss": 2.4891,
      "step": 34903
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6845336556434631,
      "learning_rate": 0.0005666610282567971,
      "loss": 2.3184,
      "step": 34904
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6884883642196655,
      "learning_rate": 0.000566659154102856,
      "loss": 2.2373,
      "step": 34905
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.7179741263389587,
      "learning_rate": 0.0005666572798993379,
      "loss": 2.2349,
      "step": 34906
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.7216191291809082,
      "learning_rate": 0.0005666554056462433,
      "loss": 2.4148,
      "step": 34907
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.680812418460846,
      "learning_rate": 0.0005666535313435723,
      "loss": 2.2382,
      "step": 34908
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.7086821794509888,
      "learning_rate": 0.0005666516569913253,
      "loss": 2.442,
      "step": 34909
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6581041216850281,
      "learning_rate": 0.0005666497825895029,
      "loss": 2.1654,
      "step": 34910
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.7360798716545105,
      "learning_rate": 0.0005666479081381052,
      "loss": 2.1396,
      "step": 34911
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6914260387420654,
      "learning_rate": 0.0005666460336371327,
      "loss": 2.2848,
      "step": 34912
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6483240127563477,
      "learning_rate": 0.0005666441590865855,
      "loss": 2.2497,
      "step": 34913
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.667966902256012,
      "learning_rate": 0.0005666422844864642,
      "loss": 2.1011,
      "step": 34914
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.66237473487854,
      "learning_rate": 0.0005666404098367691,
      "loss": 2.2739,
      "step": 34915
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6117228865623474,
      "learning_rate": 0.0005666385351375005,
      "loss": 2.3657,
      "step": 34916
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6799163818359375,
      "learning_rate": 0.0005666366603886587,
      "loss": 2.0989,
      "step": 34917
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6615424156188965,
      "learning_rate": 0.0005666347855902442,
      "loss": 2.1631,
      "step": 34918
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6549327373504639,
      "learning_rate": 0.0005666329107422573,
      "loss": 2.4903,
      "step": 34919
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6811257600784302,
      "learning_rate": 0.0005666310358446981,
      "loss": 2.4537,
      "step": 34920
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.7138115167617798,
      "learning_rate": 0.0005666291608975673,
      "loss": 2.2815,
      "step": 34921
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6613209843635559,
      "learning_rate": 0.0005666272859008652,
      "loss": 2.0997,
      "step": 34922
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.7252492308616638,
      "learning_rate": 0.0005666254108545918,
      "loss": 2.3195,
      "step": 34923
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.724245011806488,
      "learning_rate": 0.000566623535758748,
      "loss": 2.2814,
      "step": 34924
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6848180890083313,
      "learning_rate": 0.0005666216606133337,
      "loss": 2.131,
      "step": 34925
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6640955805778503,
      "learning_rate": 0.0005666197854183495,
      "loss": 2.5516,
      "step": 34926
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.7381076812744141,
      "learning_rate": 0.0005666179101737954,
      "loss": 2.2938,
      "step": 34927
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.8448066711425781,
      "learning_rate": 0.0005666160348796722,
      "loss": 2.103,
      "step": 34928
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.7281341552734375,
      "learning_rate": 0.00056661415953598,
      "loss": 2.1897,
      "step": 34929
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.8336824774742126,
      "learning_rate": 0.0005666122841427192,
      "loss": 2.2741,
      "step": 34930
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.0437120199203491,
      "learning_rate": 0.0005666104086998902,
      "loss": 2.4125,
      "step": 34931
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.5997422337532043,
      "learning_rate": 0.0005666085332074932,
      "loss": 2.1781,
      "step": 34932
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6654552221298218,
      "learning_rate": 0.0005666066576655287,
      "loss": 2.2062,
      "step": 34933
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.764961838722229,
      "learning_rate": 0.000566604782073997,
      "loss": 2.331,
      "step": 34934
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.0051935911178589,
      "learning_rate": 0.0005666029064328983,
      "loss": 2.301,
      "step": 34935
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.778925895690918,
      "learning_rate": 0.0005666010307422332,
      "loss": 2.1984,
      "step": 34936
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.7197971343994141,
      "learning_rate": 0.0005665991550020019,
      "loss": 2.2525,
      "step": 34937
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.2375476360321045,
      "learning_rate": 0.0005665972792122049,
      "loss": 2.1815,
      "step": 34938
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.63304603099823,
      "learning_rate": 0.0005665954033728423,
      "loss": 2.2136,
      "step": 34939
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6496390104293823,
      "learning_rate": 0.0005665935274839146,
      "loss": 2.2391,
      "step": 34940
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6910710334777832,
      "learning_rate": 0.0005665916515454222,
      "loss": 2.1041,
      "step": 34941
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.7229723334312439,
      "learning_rate": 0.0005665897755573653,
      "loss": 2.2579,
      "step": 34942
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6681534647941589,
      "learning_rate": 0.0005665878995197444,
      "loss": 2.2747,
      "step": 34943
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.7362374067306519,
      "learning_rate": 0.0005665860234325597,
      "loss": 2.5295,
      "step": 34944
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6656594276428223,
      "learning_rate": 0.0005665841472958117,
      "loss": 2.1615,
      "step": 34945
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6936377286911011,
      "learning_rate": 0.0005665822711095007,
      "loss": 2.2652,
      "step": 34946
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.685766339302063,
      "learning_rate": 0.0005665803948736269,
      "loss": 2.1958,
      "step": 34947
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6551594138145447,
      "learning_rate": 0.0005665785185881908,
      "loss": 2.1176,
      "step": 34948
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.7252333760261536,
      "learning_rate": 0.0005665766422531927,
      "loss": 2.1653,
      "step": 34949
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6597106456756592,
      "learning_rate": 0.0005665747658686332,
      "loss": 2.4778,
      "step": 34950
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6461002230644226,
      "learning_rate": 0.0005665728894345121,
      "loss": 2.0618,
      "step": 34951
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6259665489196777,
      "learning_rate": 0.0005665710129508303,
      "loss": 2.2518,
      "step": 34952
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.7221367955207825,
      "learning_rate": 0.0005665691364175879,
      "loss": 2.4901,
      "step": 34953
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.649243950843811,
      "learning_rate": 0.0005665672598347851,
      "loss": 2.3708,
      "step": 34954
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6460580229759216,
      "learning_rate": 0.0005665653832024225,
      "loss": 2.3679,
      "step": 34955
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6788938641548157,
      "learning_rate": 0.0005665635065205004,
      "loss": 2.3711,
      "step": 34956
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6546839475631714,
      "learning_rate": 0.0005665616297890192,
      "loss": 2.4539,
      "step": 34957
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.7162798047065735,
      "learning_rate": 0.0005665597530079789,
      "loss": 2.4641,
      "step": 34958
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6387679576873779,
      "learning_rate": 0.0005665578761773803,
      "loss": 2.179,
      "step": 34959
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6886379718780518,
      "learning_rate": 0.0005665559992972236,
      "loss": 2.3169,
      "step": 34960
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.7244577407836914,
      "learning_rate": 0.000566554122367509,
      "loss": 2.0847,
      "step": 34961
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.5971755385398865,
      "learning_rate": 0.000566552245388237,
      "loss": 2.3336,
      "step": 34962
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6641483306884766,
      "learning_rate": 0.0005665503683594078,
      "loss": 1.9477,
      "step": 34963
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6838640570640564,
      "learning_rate": 0.000566548491281022,
      "loss": 2.6363,
      "step": 34964
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6445682048797607,
      "learning_rate": 0.0005665466141530798,
      "loss": 2.2862,
      "step": 34965
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6424752473831177,
      "learning_rate": 0.0005665447369755815,
      "loss": 2.2681,
      "step": 34966
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.7059574127197266,
      "learning_rate": 0.0005665428597485275,
      "loss": 2.2057,
      "step": 34967
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.7586324214935303,
      "learning_rate": 0.0005665409824719181,
      "loss": 2.3319,
      "step": 34968
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.741202712059021,
      "learning_rate": 0.0005665391051457538,
      "loss": 2.5219,
      "step": 34969
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6365834474563599,
      "learning_rate": 0.0005665372277700348,
      "loss": 2.4078,
      "step": 34970
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6590028405189514,
      "learning_rate": 0.0005665353503447615,
      "loss": 2.2957,
      "step": 34971
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6441038846969604,
      "learning_rate": 0.0005665334728699343,
      "loss": 2.3095,
      "step": 34972
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.683502197265625,
      "learning_rate": 0.0005665315953455533,
      "loss": 2.0838,
      "step": 34973
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6965879797935486,
      "learning_rate": 0.0005665297177716193,
      "loss": 2.3487,
      "step": 34974
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.7027139663696289,
      "learning_rate": 0.0005665278401481323,
      "loss": 2.3432,
      "step": 34975
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6407710313796997,
      "learning_rate": 0.0005665259624750928,
      "loss": 2.3234,
      "step": 34976
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.7655818462371826,
      "learning_rate": 0.0005665240847525009,
      "loss": 2.2894,
      "step": 34977
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6839878559112549,
      "learning_rate": 0.0005665222069803573,
      "loss": 2.3445,
      "step": 34978
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6519657373428345,
      "learning_rate": 0.0005665203291586622,
      "loss": 2.5983,
      "step": 34979
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.772357702255249,
      "learning_rate": 0.0005665184512874159,
      "loss": 2.3514,
      "step": 34980
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6344962120056152,
      "learning_rate": 0.0005665165733666187,
      "loss": 2.2541,
      "step": 34981
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6490282416343689,
      "learning_rate": 0.0005665146953962711,
      "loss": 2.2677,
      "step": 34982
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.631941556930542,
      "learning_rate": 0.0005665128173763735,
      "loss": 2.5343,
      "step": 34983
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6795027852058411,
      "learning_rate": 0.000566510939306926,
      "loss": 2.2454,
      "step": 34984
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.7464766502380371,
      "learning_rate": 0.0005665090611879292,
      "loss": 2.5406,
      "step": 34985
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6132107377052307,
      "learning_rate": 0.0005665071830193832,
      "loss": 2.4035,
      "step": 34986
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6853214502334595,
      "learning_rate": 0.0005665053048012885,
      "loss": 2.4733,
      "step": 34987
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6174662113189697,
      "learning_rate": 0.0005665034265336455,
      "loss": 2.2165,
      "step": 34988
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.7811551094055176,
      "learning_rate": 0.0005665015482164545,
      "loss": 2.3493,
      "step": 34989
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.7399964928627014,
      "learning_rate": 0.0005664996698497159,
      "loss": 2.4322,
      "step": 34990
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6302775144577026,
      "learning_rate": 0.0005664977914334299,
      "loss": 2.421,
      "step": 34991
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.747189462184906,
      "learning_rate": 0.0005664959129675968,
      "loss": 2.1961,
      "step": 34992
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.7163386940956116,
      "learning_rate": 0.0005664940344522172,
      "loss": 2.0513,
      "step": 34993
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.692262589931488,
      "learning_rate": 0.0005664921558872913,
      "loss": 2.2619,
      "step": 34994
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6755757331848145,
      "learning_rate": 0.0005664902772728195,
      "loss": 2.3978,
      "step": 34995
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6429259181022644,
      "learning_rate": 0.0005664883986088021,
      "loss": 2.3381,
      "step": 34996
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.7248324155807495,
      "learning_rate": 0.0005664865198952395,
      "loss": 2.407,
      "step": 34997
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6506653428077698,
      "learning_rate": 0.0005664846411321321,
      "loss": 2.3502,
      "step": 34998
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.7138881683349609,
      "learning_rate": 0.00056648276231948,
      "loss": 2.3445,
      "step": 34999
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.8367691040039062,
      "learning_rate": 0.000566480883457284,
      "loss": 2.1343,
      "step": 35000
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6746078729629517,
      "learning_rate": 0.0005664790045455439,
      "loss": 2.3518,
      "step": 35001
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.7095664739608765,
      "learning_rate": 0.0005664771255842603,
      "loss": 2.1414,
      "step": 35002
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.7828887104988098,
      "learning_rate": 0.0005664752465734339,
      "loss": 2.2553,
      "step": 35003
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6703692674636841,
      "learning_rate": 0.0005664733675130644,
      "loss": 2.0461,
      "step": 35004
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6898223161697388,
      "learning_rate": 0.0005664714884031526,
      "loss": 2.1742,
      "step": 35005
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.8391319513320923,
      "learning_rate": 0.0005664696092436987,
      "loss": 2.2648,
      "step": 35006
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.67450350522995,
      "learning_rate": 0.000566467730034703,
      "loss": 2.4808,
      "step": 35007
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.638500988483429,
      "learning_rate": 0.000566465850776166,
      "loss": 2.147,
      "step": 35008
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6855686902999878,
      "learning_rate": 0.0005664639714680881,
      "loss": 2.3716,
      "step": 35009
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6837514042854309,
      "learning_rate": 0.0005664620921104693,
      "loss": 2.1923,
      "step": 35010
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6740825176239014,
      "learning_rate": 0.0005664602127033102,
      "loss": 2.4872,
      "step": 35011
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.5966416001319885,
      "learning_rate": 0.0005664583332466112,
      "loss": 2.3891,
      "step": 35012
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6577666401863098,
      "learning_rate": 0.0005664564537403725,
      "loss": 2.1534,
      "step": 35013
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6742962598800659,
      "learning_rate": 0.0005664545741845945,
      "loss": 2.2084,
      "step": 35014
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.7512149810791016,
      "learning_rate": 0.0005664526945792777,
      "loss": 2.2894,
      "step": 35015
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6961948871612549,
      "learning_rate": 0.0005664508149244222,
      "loss": 2.2726,
      "step": 35016
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6441028714179993,
      "learning_rate": 0.0005664489352200285,
      "loss": 2.5442,
      "step": 35017
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.7322866320610046,
      "learning_rate": 0.0005664470554660969,
      "loss": 2.3835,
      "step": 35018
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6679612398147583,
      "learning_rate": 0.0005664451756626277,
      "loss": 2.4144,
      "step": 35019
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.7567586898803711,
      "learning_rate": 0.0005664432958096214,
      "loss": 2.2264,
      "step": 35020
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6349384188652039,
      "learning_rate": 0.0005664414159070783,
      "loss": 2.5127,
      "step": 35021
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6305831074714661,
      "learning_rate": 0.0005664395359549986,
      "loss": 2.3042,
      "step": 35022
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.647691011428833,
      "learning_rate": 0.0005664376559533828,
      "loss": 2.3784,
      "step": 35023
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.7235662937164307,
      "learning_rate": 0.0005664357759022313,
      "loss": 2.4415,
      "step": 35024
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6822302341461182,
      "learning_rate": 0.0005664338958015442,
      "loss": 2.4746,
      "step": 35025
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6613776683807373,
      "learning_rate": 0.0005664320156513221,
      "loss": 2.325,
      "step": 35026
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6817737817764282,
      "learning_rate": 0.0005664301354515653,
      "loss": 2.2713,
      "step": 35027
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6095275282859802,
      "learning_rate": 0.000566428255202274,
      "loss": 2.0942,
      "step": 35028
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.5848317742347717,
      "learning_rate": 0.0005664263749034488,
      "loss": 2.4044,
      "step": 35029
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.697415828704834,
      "learning_rate": 0.0005664244945550897,
      "loss": 2.2507,
      "step": 35030
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6536204218864441,
      "learning_rate": 0.0005664226141571974,
      "loss": 2.0554,
      "step": 35031
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6960012912750244,
      "learning_rate": 0.0005664207337097721,
      "loss": 2.1484,
      "step": 35032
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6459046602249146,
      "learning_rate": 0.0005664188532128142,
      "loss": 2.009,
      "step": 35033
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6069297194480896,
      "learning_rate": 0.0005664169726663239,
      "loss": 2.094,
      "step": 35034
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.7053136825561523,
      "learning_rate": 0.0005664150920703017,
      "loss": 2.0881,
      "step": 35035
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6999396085739136,
      "learning_rate": 0.0005664132114247479,
      "loss": 2.3161,
      "step": 35036
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.7939384579658508,
      "learning_rate": 0.0005664113307296628,
      "loss": 2.3895,
      "step": 35037
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6797547340393066,
      "learning_rate": 0.000566409449985047,
      "loss": 2.131,
      "step": 35038
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.7308445572853088,
      "learning_rate": 0.0005664075691909005,
      "loss": 2.1467,
      "step": 35039
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6785426139831543,
      "learning_rate": 0.0005664056883472238,
      "loss": 2.3947,
      "step": 35040
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6168601512908936,
      "learning_rate": 0.0005664038074540173,
      "loss": 2.2143,
      "step": 35041
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.7409515380859375,
      "learning_rate": 0.0005664019265112813,
      "loss": 2.2399,
      "step": 35042
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6706860065460205,
      "learning_rate": 0.0005664000455190162,
      "loss": 2.1251,
      "step": 35043
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6684502363204956,
      "learning_rate": 0.0005663981644772222,
      "loss": 2.1718,
      "step": 35044
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.8309780955314636,
      "learning_rate": 0.0005663962833858998,
      "loss": 2.2972,
      "step": 35045
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6995241641998291,
      "learning_rate": 0.0005663944022450492,
      "loss": 2.3668,
      "step": 35046
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.7128171920776367,
      "learning_rate": 0.000566392521054671,
      "loss": 2.4033,
      "step": 35047
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.8865798115730286,
      "learning_rate": 0.0005663906398147654,
      "loss": 2.5788,
      "step": 35048
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6492826342582703,
      "learning_rate": 0.0005663887585253326,
      "loss": 2.3817,
      "step": 35049
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.713553249835968,
      "learning_rate": 0.0005663868771863732,
      "loss": 2.3867,
      "step": 35050
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.734774112701416,
      "learning_rate": 0.0005663849957978875,
      "loss": 2.3458,
      "step": 35051
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6206803321838379,
      "learning_rate": 0.0005663831143598758,
      "loss": 2.3447,
      "step": 35052
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6186136603355408,
      "learning_rate": 0.0005663812328723384,
      "loss": 2.2401,
      "step": 35053
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.7079740166664124,
      "learning_rate": 0.0005663793513352758,
      "loss": 2.2359,
      "step": 35054
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.650191605091095,
      "learning_rate": 0.000566377469748688,
      "loss": 2.4294,
      "step": 35055
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6856265664100647,
      "learning_rate": 0.0005663755881125758,
      "loss": 2.4377,
      "step": 35056
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6612809896469116,
      "learning_rate": 0.0005663737064269393,
      "loss": 2.2874,
      "step": 35057
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.7497949600219727,
      "learning_rate": 0.0005663718246917789,
      "loss": 2.3612,
      "step": 35058
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.7929273843765259,
      "learning_rate": 0.0005663699429070949,
      "loss": 2.2726,
      "step": 35059
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.640546441078186,
      "learning_rate": 0.0005663680610728878,
      "loss": 2.3806,
      "step": 35060
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6532005667686462,
      "learning_rate": 0.0005663661791891579,
      "loss": 2.5011,
      "step": 35061
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.7071626782417297,
      "learning_rate": 0.0005663642972559053,
      "loss": 2.1963,
      "step": 35062
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6396497488021851,
      "learning_rate": 0.0005663624152731307,
      "loss": 2.1409,
      "step": 35063
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6929618716239929,
      "learning_rate": 0.0005663605332408342,
      "loss": 2.237,
      "step": 35064
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.7702742218971252,
      "learning_rate": 0.0005663586511590163,
      "loss": 2.2858,
      "step": 35065
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.7123693823814392,
      "learning_rate": 0.0005663567690276773,
      "loss": 2.1486,
      "step": 35066
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6479884386062622,
      "learning_rate": 0.0005663548868468176,
      "loss": 2.0429,
      "step": 35067
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.7071377038955688,
      "learning_rate": 0.0005663530046164375,
      "loss": 2.2337,
      "step": 35068
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6898757815361023,
      "learning_rate": 0.0005663511223365373,
      "loss": 2.5838,
      "step": 35069
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.7325287461280823,
      "learning_rate": 0.0005663492400071173,
      "loss": 2.2546,
      "step": 35070
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6874409914016724,
      "learning_rate": 0.0005663473576281781,
      "loss": 2.3824,
      "step": 35071
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.674623966217041,
      "learning_rate": 0.0005663454751997199,
      "loss": 2.2734,
      "step": 35072
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6675642728805542,
      "learning_rate": 0.000566343592721743,
      "loss": 2.1466,
      "step": 35073
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6646366119384766,
      "learning_rate": 0.0005663417101942477,
      "loss": 2.318,
      "step": 35074
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6519581079483032,
      "learning_rate": 0.0005663398276172345,
      "loss": 2.339,
      "step": 35075
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6657224297523499,
      "learning_rate": 0.0005663379449907038,
      "loss": 2.4102,
      "step": 35076
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.7896743416786194,
      "learning_rate": 0.0005663360623146558,
      "loss": 2.0094,
      "step": 35077
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6806663274765015,
      "learning_rate": 0.0005663341795890908,
      "loss": 2.4071,
      "step": 35078
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.7305108308792114,
      "learning_rate": 0.0005663322968140094,
      "loss": 2.1504,
      "step": 35079
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6769608855247498,
      "learning_rate": 0.0005663304139894116,
      "loss": 2.1062,
      "step": 35080
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.7787375450134277,
      "learning_rate": 0.0005663285311152982,
      "loss": 2.3881,
      "step": 35081
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.662624180316925,
      "learning_rate": 0.0005663266481916692,
      "loss": 2.2935,
      "step": 35082
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6074809432029724,
      "learning_rate": 0.000566324765218525,
      "loss": 2.2838,
      "step": 35083
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6853370070457458,
      "learning_rate": 0.000566322882195866,
      "loss": 2.2487,
      "step": 35084
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.7536832094192505,
      "learning_rate": 0.0005663209991236926,
      "loss": 2.438,
      "step": 35085
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6498608589172363,
      "learning_rate": 0.000566319116002005,
      "loss": 2.1725,
      "step": 35086
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.64602130651474,
      "learning_rate": 0.0005663172328308038,
      "loss": 2.156,
      "step": 35087
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6570662260055542,
      "learning_rate": 0.0005663153496100891,
      "loss": 2.4337,
      "step": 35088
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6197119355201721,
      "learning_rate": 0.0005663134663398614,
      "loss": 2.1126,
      "step": 35089
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6502849459648132,
      "learning_rate": 0.000566311583020121,
      "loss": 2.3921,
      "step": 35090
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6702070832252502,
      "learning_rate": 0.0005663096996508681,
      "loss": 2.2366,
      "step": 35091
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.7013112306594849,
      "learning_rate": 0.0005663078162321033,
      "loss": 2.5298,
      "step": 35092
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.694894015789032,
      "learning_rate": 0.0005663059327638269,
      "loss": 2.3458,
      "step": 35093
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.673269510269165,
      "learning_rate": 0.0005663040492460392,
      "loss": 2.2544,
      "step": 35094
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.685960590839386,
      "learning_rate": 0.0005663021656787406,
      "loss": 2.3472,
      "step": 35095
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6381310820579529,
      "learning_rate": 0.0005663002820619313,
      "loss": 2.4658,
      "step": 35096
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6735011339187622,
      "learning_rate": 0.0005662983983956118,
      "loss": 2.1792,
      "step": 35097
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6374209523200989,
      "learning_rate": 0.0005662965146797823,
      "loss": 2.3823,
      "step": 35098
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.7667437791824341,
      "learning_rate": 0.0005662946309144434,
      "loss": 2.3663,
      "step": 35099
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.7026516199111938,
      "learning_rate": 0.0005662927470995951,
      "loss": 2.2587,
      "step": 35100
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6256258487701416,
      "learning_rate": 0.0005662908632352382,
      "loss": 2.1436,
      "step": 35101
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.7377362847328186,
      "learning_rate": 0.0005662889793213727,
      "loss": 2.1177,
      "step": 35102
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6313369870185852,
      "learning_rate": 0.0005662870953579989,
      "loss": 2.3033,
      "step": 35103
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6744939088821411,
      "learning_rate": 0.0005662852113451175,
      "loss": 2.2454,
      "step": 35104
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6492701768875122,
      "learning_rate": 0.0005662833272827286,
      "loss": 2.3069,
      "step": 35105
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6771419644355774,
      "learning_rate": 0.0005662814431708326,
      "loss": 2.3818,
      "step": 35106
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.7242342233657837,
      "learning_rate": 0.0005662795590094298,
      "loss": 2.1151,
      "step": 35107
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6475381851196289,
      "learning_rate": 0.0005662776747985207,
      "loss": 2.4046,
      "step": 35108
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.822943389415741,
      "learning_rate": 0.0005662757905381054,
      "loss": 2.4298,
      "step": 35109
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6698201298713684,
      "learning_rate": 0.0005662739062281846,
      "loss": 2.4996,
      "step": 35110
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6428489089012146,
      "learning_rate": 0.0005662720218687583,
      "loss": 2.2779,
      "step": 35111
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.654479444026947,
      "learning_rate": 0.0005662701374598272,
      "loss": 2.3688,
      "step": 35112
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.722771406173706,
      "learning_rate": 0.0005662682530013912,
      "loss": 2.1306,
      "step": 35113
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6810539960861206,
      "learning_rate": 0.0005662663684934511,
      "loss": 2.2348,
      "step": 35114
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.672452986240387,
      "learning_rate": 0.000566264483936007,
      "loss": 2.3691,
      "step": 35115
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6676496863365173,
      "learning_rate": 0.0005662625993290592,
      "loss": 2.1915,
      "step": 35116
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.7147614359855652,
      "learning_rate": 0.0005662607146726083,
      "loss": 2.3351,
      "step": 35117
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6657976508140564,
      "learning_rate": 0.0005662588299666544,
      "loss": 2.2897,
      "step": 35118
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6773353219032288,
      "learning_rate": 0.0005662569452111981,
      "loss": 2.2656,
      "step": 35119
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.629811704158783,
      "learning_rate": 0.0005662550604062394,
      "loss": 2.1882,
      "step": 35120
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6202906370162964,
      "learning_rate": 0.000566253175551779,
      "loss": 2.3273,
      "step": 35121
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.64728844165802,
      "learning_rate": 0.000566251290647817,
      "loss": 2.1372,
      "step": 35122
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6668899059295654,
      "learning_rate": 0.000566249405694354,
      "loss": 2.4837,
      "step": 35123
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6549572944641113,
      "learning_rate": 0.0005662475206913902,
      "loss": 2.3277,
      "step": 35124
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.7888430953025818,
      "learning_rate": 0.0005662456356389259,
      "loss": 2.2911,
      "step": 35125
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6631022095680237,
      "learning_rate": 0.0005662437505369615,
      "loss": 2.3542,
      "step": 35126
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6630979776382446,
      "learning_rate": 0.0005662418653854974,
      "loss": 2.3006,
      "step": 35127
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.7046641707420349,
      "learning_rate": 0.0005662399801845339,
      "loss": 2.3992,
      "step": 35128
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6510074734687805,
      "learning_rate": 0.0005662380949340713,
      "loss": 2.2549,
      "step": 35129
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.734389066696167,
      "learning_rate": 0.0005662362096341101,
      "loss": 2.2627,
      "step": 35130
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6721212267875671,
      "learning_rate": 0.0005662343242846506,
      "loss": 2.1905,
      "step": 35131
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6539868116378784,
      "learning_rate": 0.000566232438885693,
      "loss": 2.198,
      "step": 35132
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6815302968025208,
      "learning_rate": 0.0005662305534372378,
      "loss": 2.2514,
      "step": 35133
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6296755075454712,
      "learning_rate": 0.0005662286679392854,
      "loss": 2.4844,
      "step": 35134
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6168690919876099,
      "learning_rate": 0.000566226782391836,
      "loss": 2.1938,
      "step": 35135
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6517292261123657,
      "learning_rate": 0.0005662248967948899,
      "loss": 2.2092,
      "step": 35136
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6651149988174438,
      "learning_rate": 0.0005662230111484478,
      "loss": 2.1943,
      "step": 35137
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6852433085441589,
      "learning_rate": 0.0005662211254525096,
      "loss": 2.5072,
      "step": 35138
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6330865621566772,
      "learning_rate": 0.000566219239707076,
      "loss": 2.2145,
      "step": 35139
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6649575233459473,
      "learning_rate": 0.0005662173539121471,
      "loss": 2.1859,
      "step": 35140
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6319140791893005,
      "learning_rate": 0.0005662154680677234,
      "loss": 2.2408,
      "step": 35141
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6554328799247742,
      "learning_rate": 0.0005662135821738053,
      "loss": 2.2139,
      "step": 35142
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.7186988592147827,
      "learning_rate": 0.0005662116962303931,
      "loss": 2.0155,
      "step": 35143
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.7287616729736328,
      "learning_rate": 0.0005662098102374869,
      "loss": 2.5685,
      "step": 35144
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6095768809318542,
      "learning_rate": 0.0005662079241950875,
      "loss": 2.2145,
      "step": 35145
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6160100102424622,
      "learning_rate": 0.000566206038103195,
      "loss": 2.2695,
      "step": 35146
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6318041086196899,
      "learning_rate": 0.0005662041519618097,
      "loss": 2.2415,
      "step": 35147
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.7297921776771545,
      "learning_rate": 0.000566202265770932,
      "loss": 2.015,
      "step": 35148
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6521238684654236,
      "learning_rate": 0.0005662003795305624,
      "loss": 2.2017,
      "step": 35149
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6143633127212524,
      "learning_rate": 0.0005661984932407011,
      "loss": 2.2476,
      "step": 35150
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.7000152468681335,
      "learning_rate": 0.0005661966069013483,
      "loss": 2.3054,
      "step": 35151
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.7911442518234253,
      "learning_rate": 0.0005661947205125046,
      "loss": 2.3296,
      "step": 35152
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.7517988085746765,
      "learning_rate": 0.0005661928340741704,
      "loss": 2.1383,
      "step": 35153
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.7228608727455139,
      "learning_rate": 0.0005661909475863458,
      "loss": 2.3348,
      "step": 35154
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6738242506980896,
      "learning_rate": 0.0005661890610490313,
      "loss": 2.442,
      "step": 35155
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6885198354721069,
      "learning_rate": 0.0005661871744622273,
      "loss": 2.4008,
      "step": 35156
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.663868248462677,
      "learning_rate": 0.000566185287825934,
      "loss": 2.3515,
      "step": 35157
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.7494562268257141,
      "learning_rate": 0.0005661834011401519,
      "loss": 2.5575,
      "step": 35158
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6816982626914978,
      "learning_rate": 0.0005661815144048812,
      "loss": 2.1535,
      "step": 35159
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.712212860584259,
      "learning_rate": 0.0005661796276201224,
      "loss": 2.2339,
      "step": 35160
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.7009434103965759,
      "learning_rate": 0.0005661777407858757,
      "loss": 2.5673,
      "step": 35161
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6064338088035583,
      "learning_rate": 0.0005661758539021417,
      "loss": 2.341,
      "step": 35162
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.7114320397377014,
      "learning_rate": 0.0005661739669689204,
      "loss": 2.4428,
      "step": 35163
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.633562445640564,
      "learning_rate": 0.0005661720799862125,
      "loss": 2.2636,
      "step": 35164
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.7069577574729919,
      "learning_rate": 0.0005661701929540181,
      "loss": 2.2879,
      "step": 35165
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6363495588302612,
      "learning_rate": 0.0005661683058723376,
      "loss": 2.1787,
      "step": 35166
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.7735701203346252,
      "learning_rate": 0.0005661664187411714,
      "loss": 2.059,
      "step": 35167
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6059213280677795,
      "learning_rate": 0.0005661645315605199,
      "loss": 2.2109,
      "step": 35168
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6539510488510132,
      "learning_rate": 0.0005661626443303833,
      "loss": 2.1481,
      "step": 35169
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.7019992470741272,
      "learning_rate": 0.000566160757050762,
      "loss": 2.4341,
      "step": 35170
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.7912783622741699,
      "learning_rate": 0.0005661588697216567,
      "loss": 2.3356,
      "step": 35171
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.8359492421150208,
      "learning_rate": 0.0005661569823430671,
      "loss": 2.1737,
      "step": 35172
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6638502478599548,
      "learning_rate": 0.0005661550949149941,
      "loss": 2.2197,
      "step": 35173
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6131653785705566,
      "learning_rate": 0.0005661532074374377,
      "loss": 2.4926,
      "step": 35174
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.7493083477020264,
      "learning_rate": 0.0005661513199103985,
      "loss": 2.3121,
      "step": 35175
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6927367448806763,
      "learning_rate": 0.0005661494323338767,
      "loss": 2.3982,
      "step": 35176
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6895735859870911,
      "learning_rate": 0.0005661475447078727,
      "loss": 2.1503,
      "step": 35177
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.653925359249115,
      "learning_rate": 0.0005661456570323867,
      "loss": 2.3702,
      "step": 35178
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.7203997373580933,
      "learning_rate": 0.0005661437693074194,
      "loss": 2.5228,
      "step": 35179
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6773401498794556,
      "learning_rate": 0.0005661418815329709,
      "loss": 2.4273,
      "step": 35180
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6620983481407166,
      "learning_rate": 0.0005661399937090416,
      "loss": 2.2631,
      "step": 35181
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6965424418449402,
      "learning_rate": 0.0005661381058356318,
      "loss": 2.3878,
      "step": 35182
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.8059207201004028,
      "learning_rate": 0.000566136217912742,
      "loss": 2.5957,
      "step": 35183
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.734607994556427,
      "learning_rate": 0.0005661343299403724,
      "loss": 2.41,
      "step": 35184
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.7761783599853516,
      "learning_rate": 0.0005661324419185234,
      "loss": 2.3646,
      "step": 35185
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6592671275138855,
      "learning_rate": 0.0005661305538471953,
      "loss": 2.2864,
      "step": 35186
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.7688543200492859,
      "learning_rate": 0.0005661286657263887,
      "loss": 2.2974,
      "step": 35187
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6560115218162537,
      "learning_rate": 0.0005661267775561035,
      "loss": 2.3477,
      "step": 35188
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6525179147720337,
      "learning_rate": 0.0005661248893363405,
      "loss": 2.2921,
      "step": 35189
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6931495666503906,
      "learning_rate": 0.0005661230010670997,
      "loss": 2.0198,
      "step": 35190
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.692196786403656,
      "learning_rate": 0.0005661211127483818,
      "loss": 2.4019,
      "step": 35191
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.7256932854652405,
      "learning_rate": 0.0005661192243801869,
      "loss": 2.0873,
      "step": 35192
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.7490075826644897,
      "learning_rate": 0.0005661173359625154,
      "loss": 2.3831,
      "step": 35193
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.7000446319580078,
      "learning_rate": 0.0005661154474953675,
      "loss": 2.0873,
      "step": 35194
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6693248152732849,
      "learning_rate": 0.0005661135589787439,
      "loss": 2.2968,
      "step": 35195
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6692816019058228,
      "learning_rate": 0.0005661116704126448,
      "loss": 2.4119,
      "step": 35196
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.7036852240562439,
      "learning_rate": 0.0005661097817970704,
      "loss": 2.116,
      "step": 35197
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6585788726806641,
      "learning_rate": 0.0005661078931320212,
      "loss": 2.2141,
      "step": 35198
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.043171763420105,
      "learning_rate": 0.0005661060044174975,
      "loss": 2.4489,
      "step": 35199
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6921644806861877,
      "learning_rate": 0.0005661041156534997,
      "loss": 2.1761,
      "step": 35200
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.7230849266052246,
      "learning_rate": 0.0005661022268400281,
      "loss": 2.5209,
      "step": 35201
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.668276846408844,
      "learning_rate": 0.0005661003379770831,
      "loss": 2.1845,
      "step": 35202
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6500931978225708,
      "learning_rate": 0.000566098449064665,
      "loss": 2.3522,
      "step": 35203
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.713065505027771,
      "learning_rate": 0.0005660965601027743,
      "loss": 2.3352,
      "step": 35204
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.7355061769485474,
      "learning_rate": 0.000566094671091411,
      "loss": 2.294,
      "step": 35205
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6908473372459412,
      "learning_rate": 0.0005660927820305758,
      "loss": 2.221,
      "step": 35206
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6891180276870728,
      "learning_rate": 0.000566090892920269,
      "loss": 2.3157,
      "step": 35207
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.7394642233848572,
      "learning_rate": 0.0005660890037604907,
      "loss": 2.4222,
      "step": 35208
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6840315461158752,
      "learning_rate": 0.0005660871145512416,
      "loss": 2.3748,
      "step": 35209
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6942214965820312,
      "learning_rate": 0.0005660852252925219,
      "loss": 2.3324,
      "step": 35210
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6384931802749634,
      "learning_rate": 0.0005660833359843317,
      "loss": 2.1201,
      "step": 35211
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.7070624828338623,
      "learning_rate": 0.0005660814466266717,
      "loss": 2.3673,
      "step": 35212
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6822625398635864,
      "learning_rate": 0.0005660795572195422,
      "loss": 2.3596,
      "step": 35213
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6654779314994812,
      "learning_rate": 0.0005660776677629436,
      "loss": 2.3974,
      "step": 35214
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6213129162788391,
      "learning_rate": 0.0005660757782568759,
      "loss": 2.2429,
      "step": 35215
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.653096616268158,
      "learning_rate": 0.0005660738887013398,
      "loss": 2.1124,
      "step": 35216
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6284763216972351,
      "learning_rate": 0.0005660719990963355,
      "loss": 2.3091,
      "step": 35217
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.7013089060783386,
      "learning_rate": 0.0005660701094418635,
      "loss": 2.1614,
      "step": 35218
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6555612683296204,
      "learning_rate": 0.0005660682197379239,
      "loss": 2.5241,
      "step": 35219
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6954016089439392,
      "learning_rate": 0.0005660663299845173,
      "loss": 2.109,
      "step": 35220
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6549499034881592,
      "learning_rate": 0.0005660644401816439,
      "loss": 2.3962,
      "step": 35221
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.7562828660011292,
      "learning_rate": 0.0005660625503293041,
      "loss": 2.3486,
      "step": 35222
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.7976483106613159,
      "learning_rate": 0.0005660606604274982,
      "loss": 2.1907,
      "step": 35223
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6904311776161194,
      "learning_rate": 0.0005660587704762268,
      "loss": 2.2098,
      "step": 35224
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6380312442779541,
      "learning_rate": 0.0005660568804754898,
      "loss": 2.1566,
      "step": 35225
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.7595271468162537,
      "learning_rate": 0.000566054990425288,
      "loss": 2.3699,
      "step": 35226
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6811081767082214,
      "learning_rate": 0.0005660531003256215,
      "loss": 2.3587,
      "step": 35227
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.7269172072410583,
      "learning_rate": 0.0005660512101764906,
      "loss": 2.0322,
      "step": 35228
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.8007369041442871,
      "learning_rate": 0.0005660493199778959,
      "loss": 2.1196,
      "step": 35229
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6643386483192444,
      "learning_rate": 0.0005660474297298374,
      "loss": 2.3348,
      "step": 35230
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.713385283946991,
      "learning_rate": 0.0005660455394323159,
      "loss": 2.3065,
      "step": 35231
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6540377140045166,
      "learning_rate": 0.0005660436490853315,
      "loss": 2.0748,
      "step": 35232
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.778988778591156,
      "learning_rate": 0.0005660417586888844,
      "loss": 2.378,
      "step": 35233
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.608197808265686,
      "learning_rate": 0.0005660398682429752,
      "loss": 2.146,
      "step": 35234
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.5998210310935974,
      "learning_rate": 0.0005660379777476042,
      "loss": 2.3155,
      "step": 35235
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.5914604067802429,
      "learning_rate": 0.0005660360872027716,
      "loss": 2.0842,
      "step": 35236
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6724472045898438,
      "learning_rate": 0.000566034196608478,
      "loss": 2.0824,
      "step": 35237
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6211223006248474,
      "learning_rate": 0.0005660323059647236,
      "loss": 2.2524,
      "step": 35238
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6193487048149109,
      "learning_rate": 0.0005660304152715087,
      "loss": 2.1708,
      "step": 35239
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6792862415313721,
      "learning_rate": 0.0005660285245288338,
      "loss": 2.4442,
      "step": 35240
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6817934513092041,
      "learning_rate": 0.0005660266337366991,
      "loss": 2.2009,
      "step": 35241
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6358282566070557,
      "learning_rate": 0.0005660247428951051,
      "loss": 2.3822,
      "step": 35242
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6492099165916443,
      "learning_rate": 0.0005660228520040521,
      "loss": 2.3061,
      "step": 35243
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6727924346923828,
      "learning_rate": 0.0005660209610635404,
      "loss": 2.3163,
      "step": 35244
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6849318742752075,
      "learning_rate": 0.0005660190700735703,
      "loss": 2.1078,
      "step": 35245
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6129905581474304,
      "learning_rate": 0.0005660171790341423,
      "loss": 2.2244,
      "step": 35246
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6425696611404419,
      "learning_rate": 0.0005660152879452566,
      "loss": 2.1315,
      "step": 35247
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6694719195365906,
      "learning_rate": 0.0005660133968069138,
      "loss": 2.1933,
      "step": 35248
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6546847820281982,
      "learning_rate": 0.0005660115056191139,
      "loss": 2.2625,
      "step": 35249
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6770456433296204,
      "learning_rate": 0.0005660096143818577,
      "loss": 2.3583,
      "step": 35250
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.7561637163162231,
      "learning_rate": 0.0005660077230951451,
      "loss": 2.2304,
      "step": 35251
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.7328956127166748,
      "learning_rate": 0.0005660058317589766,
      "loss": 2.1814,
      "step": 35252
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6547801494598389,
      "learning_rate": 0.0005660039403733528,
      "loss": 2.1625,
      "step": 35253
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6292161345481873,
      "learning_rate": 0.0005660020489382737,
      "loss": 2.1782,
      "step": 35254
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.7129989862442017,
      "learning_rate": 0.0005660001574537397,
      "loss": 2.0513,
      "step": 35255
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.7659342288970947,
      "learning_rate": 0.0005659982659197515,
      "loss": 2.3164,
      "step": 35256
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6582627296447754,
      "learning_rate": 0.000565996374336309,
      "loss": 2.1762,
      "step": 35257
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.68599933385849,
      "learning_rate": 0.0005659944827034128,
      "loss": 2.3683,
      "step": 35258
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6469460725784302,
      "learning_rate": 0.0005659925910210632,
      "loss": 2.3647,
      "step": 35259
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6671594977378845,
      "learning_rate": 0.0005659906992892607,
      "loss": 2.2806,
      "step": 35260
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6979934573173523,
      "learning_rate": 0.0005659888075080052,
      "loss": 2.4373,
      "step": 35261
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6075490117073059,
      "learning_rate": 0.0005659869156772978,
      "loss": 2.3762,
      "step": 35262
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6874788999557495,
      "learning_rate": 0.000565985023797138,
      "loss": 2.3453,
      "step": 35263
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6655565500259399,
      "learning_rate": 0.0005659831318675267,
      "loss": 2.2477,
      "step": 35264
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6977876424789429,
      "learning_rate": 0.0005659812398884643,
      "loss": 2.3164,
      "step": 35265
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6442160606384277,
      "learning_rate": 0.0005659793478599507,
      "loss": 2.4282,
      "step": 35266
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6146386861801147,
      "learning_rate": 0.0005659774557819866,
      "loss": 2.2872,
      "step": 35267
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6817792654037476,
      "learning_rate": 0.0005659755636545724,
      "loss": 2.3029,
      "step": 35268
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6491501927375793,
      "learning_rate": 0.0005659736714777082,
      "loss": 2.2514,
      "step": 35269
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6893108487129211,
      "learning_rate": 0.0005659717792513945,
      "loss": 2.2174,
      "step": 35270
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.7401677966117859,
      "learning_rate": 0.0005659698869756315,
      "loss": 2.4165,
      "step": 35271
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.63519686460495,
      "learning_rate": 0.0005659679946504199,
      "loss": 2.3434,
      "step": 35272
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6455432772636414,
      "learning_rate": 0.0005659661022757597,
      "loss": 2.2297,
      "step": 35273
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.7870273590087891,
      "learning_rate": 0.0005659642098516514,
      "loss": 2.0616,
      "step": 35274
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6801372766494751,
      "learning_rate": 0.0005659623173780953,
      "loss": 2.3607,
      "step": 35275
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.7111009955406189,
      "learning_rate": 0.0005659604248550919,
      "loss": 2.384,
      "step": 35276
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6832528114318848,
      "learning_rate": 0.0005659585322826413,
      "loss": 2.0962,
      "step": 35277
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6771093010902405,
      "learning_rate": 0.000565956639660744,
      "loss": 2.5551,
      "step": 35278
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.7191579937934875,
      "learning_rate": 0.0005659547469894005,
      "loss": 2.2339,
      "step": 35279
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.9012952446937561,
      "learning_rate": 0.0005659528542686108,
      "loss": 2.3764,
      "step": 35280
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6565174460411072,
      "learning_rate": 0.0005659509614983756,
      "loss": 2.3282,
      "step": 35281
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.7198140025138855,
      "learning_rate": 0.0005659490686786949,
      "loss": 2.1677,
      "step": 35282
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6821417808532715,
      "learning_rate": 0.0005659471758095695,
      "loss": 2.125,
      "step": 35283
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.7150912880897522,
      "learning_rate": 0.0005659452828909992,
      "loss": 2.1961,
      "step": 35284
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6986872553825378,
      "learning_rate": 0.0005659433899229849,
      "loss": 2.2215,
      "step": 35285
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.7758783102035522,
      "learning_rate": 0.0005659414969055266,
      "loss": 2.2835,
      "step": 35286
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6622164249420166,
      "learning_rate": 0.0005659396038386248,
      "loss": 2.1059,
      "step": 35287
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.7254616618156433,
      "learning_rate": 0.0005659377107222798,
      "loss": 2.47,
      "step": 35288
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6588828563690186,
      "learning_rate": 0.000565935817556492,
      "loss": 2.2478,
      "step": 35289
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.7765381932258606,
      "learning_rate": 0.0005659339243412616,
      "loss": 2.5467,
      "step": 35290
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6459435224533081,
      "learning_rate": 0.0005659320310765891,
      "loss": 2.3345,
      "step": 35291
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6160252690315247,
      "learning_rate": 0.0005659301377624749,
      "loss": 2.0812,
      "step": 35292
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6883399486541748,
      "learning_rate": 0.0005659282443989192,
      "loss": 2.327,
      "step": 35293
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.8663725852966309,
      "learning_rate": 0.0005659263509859225,
      "loss": 2.3603,
      "step": 35294
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6250141263008118,
      "learning_rate": 0.0005659244575234849,
      "loss": 2.0144,
      "step": 35295
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.7405836582183838,
      "learning_rate": 0.000565922564011607,
      "loss": 2.2958,
      "step": 35296
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.7085137367248535,
      "learning_rate": 0.0005659206704502892,
      "loss": 2.4149,
      "step": 35297
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6547596454620361,
      "learning_rate": 0.0005659187768395316,
      "loss": 2.1559,
      "step": 35298
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.7353000640869141,
      "learning_rate": 0.0005659168831793346,
      "loss": 2.6273,
      "step": 35299
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6070367097854614,
      "learning_rate": 0.0005659149894696988,
      "loss": 2.1715,
      "step": 35300
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.7375035881996155,
      "learning_rate": 0.0005659130957106242,
      "loss": 2.134,
      "step": 35301
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6623288989067078,
      "learning_rate": 0.0005659112019021116,
      "loss": 2.1637,
      "step": 35302
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6996580362319946,
      "learning_rate": 0.0005659093080441609,
      "loss": 2.3211,
      "step": 35303
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.689038097858429,
      "learning_rate": 0.0005659074141367727,
      "loss": 2.3073,
      "step": 35304
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6864548921585083,
      "learning_rate": 0.0005659055201799473,
      "loss": 2.1444,
      "step": 35305
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6674540042877197,
      "learning_rate": 0.000565903626173685,
      "loss": 2.0637,
      "step": 35306
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6831672787666321,
      "learning_rate": 0.0005659017321179862,
      "loss": 2.2172,
      "step": 35307
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.7073701024055481,
      "learning_rate": 0.0005658998380128513,
      "loss": 2.1819,
      "step": 35308
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6544737219810486,
      "learning_rate": 0.0005658979438582806,
      "loss": 2.1177,
      "step": 35309
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.7269435524940491,
      "learning_rate": 0.0005658960496542744,
      "loss": 2.3558,
      "step": 35310
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.7722931504249573,
      "learning_rate": 0.0005658941554008331,
      "loss": 2.2029,
      "step": 35311
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.6985037326812744,
      "learning_rate": 0.0005658922610979571,
      "loss": 2.3021,
      "step": 35312
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.642300546169281,
      "learning_rate": 0.0005658903667456466,
      "loss": 2.6592,
      "step": 35313
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6749134659767151,
      "learning_rate": 0.0005658884723439022,
      "loss": 2.2687,
      "step": 35314
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.7041371464729309,
      "learning_rate": 0.000565886577892724,
      "loss": 2.2531,
      "step": 35315
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6299489140510559,
      "learning_rate": 0.0005658846833921127,
      "loss": 2.2469,
      "step": 35316
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.7482243180274963,
      "learning_rate": 0.0005658827888420681,
      "loss": 2.3876,
      "step": 35317
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.7189609408378601,
      "learning_rate": 0.000565880894242591,
      "loss": 2.1738,
      "step": 35318
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6447356939315796,
      "learning_rate": 0.0005658789995936817,
      "loss": 2.1903,
      "step": 35319
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6967765688896179,
      "learning_rate": 0.0005658771048953405,
      "loss": 2.063,
      "step": 35320
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.9297560453414917,
      "learning_rate": 0.0005658752101475675,
      "loss": 2.2191,
      "step": 35321
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6168643832206726,
      "learning_rate": 0.0005658733153503635,
      "loss": 2.4354,
      "step": 35322
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.7025323510169983,
      "learning_rate": 0.0005658714205037285,
      "loss": 2.2446,
      "step": 35323
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.8347092866897583,
      "learning_rate": 0.0005658695256076632,
      "loss": 2.3492,
      "step": 35324
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.7926472425460815,
      "learning_rate": 0.0005658676306621675,
      "loss": 2.418,
      "step": 35325
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6353578567504883,
      "learning_rate": 0.0005658657356672421,
      "loss": 2.3222,
      "step": 35326
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.7521144151687622,
      "learning_rate": 0.0005658638406228872,
      "loss": 2.1794,
      "step": 35327
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.7541881799697876,
      "learning_rate": 0.0005658619455291032,
      "loss": 2.3664,
      "step": 35328
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6655171513557434,
      "learning_rate": 0.0005658600503858905,
      "loss": 2.2646,
      "step": 35329
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6419463157653809,
      "learning_rate": 0.0005658581551932492,
      "loss": 2.2687,
      "step": 35330
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.8687868714332581,
      "learning_rate": 0.0005658562599511802,
      "loss": 2.3704,
      "step": 35331
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6317678689956665,
      "learning_rate": 0.0005658543646596832,
      "loss": 2.313,
      "step": 35332
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6117756962776184,
      "learning_rate": 0.000565852469318759,
      "loss": 2.2386,
      "step": 35333
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6423740983009338,
      "learning_rate": 0.0005658505739284077,
      "loss": 2.2623,
      "step": 35334
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.8081698417663574,
      "learning_rate": 0.0005658486784886298,
      "loss": 2.2855,
      "step": 35335
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6755533814430237,
      "learning_rate": 0.0005658467829994258,
      "loss": 2.2766,
      "step": 35336
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6406537890434265,
      "learning_rate": 0.0005658448874607956,
      "loss": 2.2581,
      "step": 35337
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.7757881879806519,
      "learning_rate": 0.00056584299187274,
      "loss": 2.0416,
      "step": 35338
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6787272095680237,
      "learning_rate": 0.000565841096235259,
      "loss": 2.3192,
      "step": 35339
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6467111110687256,
      "learning_rate": 0.0005658392005483533,
      "loss": 2.1287,
      "step": 35340
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.8377770781517029,
      "learning_rate": 0.0005658373048120229,
      "loss": 2.217,
      "step": 35341
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.7119868993759155,
      "learning_rate": 0.0005658354090262684,
      "loss": 2.2368,
      "step": 35342
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6122598052024841,
      "learning_rate": 0.0005658335131910901,
      "loss": 2.284,
      "step": 35343
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6542171835899353,
      "learning_rate": 0.0005658316173064883,
      "loss": 2.1607,
      "step": 35344
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.8478094339370728,
      "learning_rate": 0.0005658297213724634,
      "loss": 2.1416,
      "step": 35345
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6531829237937927,
      "learning_rate": 0.0005658278253890158,
      "loss": 2.3478,
      "step": 35346
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.7759741544723511,
      "learning_rate": 0.0005658259293561457,
      "loss": 2.3498,
      "step": 35347
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6920945644378662,
      "learning_rate": 0.0005658240332738535,
      "loss": 2.3436,
      "step": 35348
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6838459968566895,
      "learning_rate": 0.0005658221371421398,
      "loss": 2.3304,
      "step": 35349
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.7273210883140564,
      "learning_rate": 0.0005658202409610045,
      "loss": 2.2258,
      "step": 35350
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.7778716683387756,
      "learning_rate": 0.0005658183447304483,
      "loss": 2.2334,
      "step": 35351
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6792696118354797,
      "learning_rate": 0.0005658164484504715,
      "loss": 2.245,
      "step": 35352
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6998555064201355,
      "learning_rate": 0.0005658145521210744,
      "loss": 2.2038,
      "step": 35353
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.7400544285774231,
      "learning_rate": 0.0005658126557422573,
      "loss": 2.1512,
      "step": 35354
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6480051279067993,
      "learning_rate": 0.0005658107593140207,
      "loss": 2.5924,
      "step": 35355
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6612870693206787,
      "learning_rate": 0.0005658088628363647,
      "loss": 2.2756,
      "step": 35356
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.673713207244873,
      "learning_rate": 0.0005658069663092899,
      "loss": 2.3048,
      "step": 35357
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.7309256792068481,
      "learning_rate": 0.0005658050697327966,
      "loss": 2.4966,
      "step": 35358
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6442828178405762,
      "learning_rate": 0.0005658031731068852,
      "loss": 2.3261,
      "step": 35359
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6575813889503479,
      "learning_rate": 0.0005658012764315559,
      "loss": 2.1685,
      "step": 35360
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.672418475151062,
      "learning_rate": 0.000565799379706809,
      "loss": 2.4425,
      "step": 35361
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6659457087516785,
      "learning_rate": 0.0005657974829326452,
      "loss": 2.3114,
      "step": 35362
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6767550706863403,
      "learning_rate": 0.0005657955861090644,
      "loss": 2.5179,
      "step": 35363
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.688107430934906,
      "learning_rate": 0.0005657936892360674,
      "loss": 2.3514,
      "step": 35364
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6545053124427795,
      "learning_rate": 0.0005657917923136542,
      "loss": 2.2562,
      "step": 35365
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6490263342857361,
      "learning_rate": 0.0005657898953418252,
      "loss": 2.3745,
      "step": 35366
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.7117917537689209,
      "learning_rate": 0.000565787998320581,
      "loss": 2.261,
      "step": 35367
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.7530788779258728,
      "learning_rate": 0.0005657861012499218,
      "loss": 2.286,
      "step": 35368
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.684129536151886,
      "learning_rate": 0.000565784204129848,
      "loss": 2.5524,
      "step": 35369
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6411764025688171,
      "learning_rate": 0.0005657823069603597,
      "loss": 2.2089,
      "step": 35370
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.7447623014450073,
      "learning_rate": 0.0005657804097414576,
      "loss": 2.3031,
      "step": 35371
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.7171819806098938,
      "learning_rate": 0.0005657785124731418,
      "loss": 2.5428,
      "step": 35372
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6596527695655823,
      "learning_rate": 0.0005657766151554129,
      "loss": 2.1773,
      "step": 35373
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.7349377870559692,
      "learning_rate": 0.000565774717788271,
      "loss": 2.2048,
      "step": 35374
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.7071487903594971,
      "learning_rate": 0.0005657728203717166,
      "loss": 2.2235,
      "step": 35375
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6664377450942993,
      "learning_rate": 0.0005657709229057499,
      "loss": 2.4342,
      "step": 35376
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6627541184425354,
      "learning_rate": 0.0005657690253903715,
      "loss": 2.1345,
      "step": 35377
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6799466013908386,
      "learning_rate": 0.0005657671278255816,
      "loss": 2.3227,
      "step": 35378
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.0358468294143677,
      "learning_rate": 0.0005657652302113805,
      "loss": 2.2099,
      "step": 35379
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6408713459968567,
      "learning_rate": 0.0005657633325477688,
      "loss": 2.3336,
      "step": 35380
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.712855339050293,
      "learning_rate": 0.0005657614348347466,
      "loss": 2.4477,
      "step": 35381
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.7200725078582764,
      "learning_rate": 0.0005657595370723142,
      "loss": 2.3882,
      "step": 35382
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.7154445648193359,
      "learning_rate": 0.0005657576392604723,
      "loss": 2.1942,
      "step": 35383
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.7070412635803223,
      "learning_rate": 0.0005657557413992209,
      "loss": 2.7173,
      "step": 35384
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6414517164230347,
      "learning_rate": 0.0005657538434885606,
      "loss": 2.3681,
      "step": 35385
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.712379515171051,
      "learning_rate": 0.0005657519455284914,
      "loss": 2.4159,
      "step": 35386
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.8140910267829895,
      "learning_rate": 0.0005657500475190141,
      "loss": 2.3303,
      "step": 35387
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6697630882263184,
      "learning_rate": 0.0005657481494601289,
      "loss": 2.0827,
      "step": 35388
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.771248996257782,
      "learning_rate": 0.0005657462513518359,
      "loss": 2.2117,
      "step": 35389
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.7240718007087708,
      "learning_rate": 0.0005657443531941358,
      "loss": 2.103,
      "step": 35390
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6808772683143616,
      "learning_rate": 0.0005657424549870288,
      "loss": 2.298,
      "step": 35391
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.7350404262542725,
      "learning_rate": 0.0005657405567305153,
      "loss": 2.1768,
      "step": 35392
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6805586814880371,
      "learning_rate": 0.0005657386584245955,
      "loss": 2.235,
      "step": 35393
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.7360802888870239,
      "learning_rate": 0.0005657367600692698,
      "loss": 2.2294,
      "step": 35394
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6667882800102234,
      "learning_rate": 0.0005657348616645389,
      "loss": 2.3412,
      "step": 35395
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.7762010097503662,
      "learning_rate": 0.0005657329632104027,
      "loss": 2.1233,
      "step": 35396
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6871045231819153,
      "learning_rate": 0.0005657310647068617,
      "loss": 2.0957,
      "step": 35397
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6581728458404541,
      "learning_rate": 0.0005657291661539163,
      "loss": 2.2176,
      "step": 35398
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6812810301780701,
      "learning_rate": 0.000565727267551567,
      "loss": 2.3415,
      "step": 35399
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.8304958343505859,
      "learning_rate": 0.0005657253688998137,
      "loss": 2.3273,
      "step": 35400
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.8762506246566772,
      "learning_rate": 0.0005657234701986572,
      "loss": 2.2468,
      "step": 35401
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6369541883468628,
      "learning_rate": 0.0005657215714480976,
      "loss": 2.1577,
      "step": 35402
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.7391273975372314,
      "learning_rate": 0.0005657196726481354,
      "loss": 2.4535,
      "step": 35403
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.7491919994354248,
      "learning_rate": 0.000565717773798771,
      "loss": 2.5428,
      "step": 35404
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6812593340873718,
      "learning_rate": 0.0005657158749000045,
      "loss": 2.1669,
      "step": 35405
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6537267565727234,
      "learning_rate": 0.0005657139759518365,
      "loss": 2.307,
      "step": 35406
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.7631018161773682,
      "learning_rate": 0.0005657120769542672,
      "loss": 2.1522,
      "step": 35407
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6546555161476135,
      "learning_rate": 0.000565710177907297,
      "loss": 2.2999,
      "step": 35408
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6186525821685791,
      "learning_rate": 0.0005657082788109263,
      "loss": 2.1909,
      "step": 35409
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6948843598365784,
      "learning_rate": 0.0005657063796651555,
      "loss": 2.3929,
      "step": 35410
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.7820211052894592,
      "learning_rate": 0.0005657044804699848,
      "loss": 2.365,
      "step": 35411
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.7956331372261047,
      "learning_rate": 0.0005657025812254144,
      "loss": 2.3155,
      "step": 35412
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.7044397592544556,
      "learning_rate": 0.0005657006819314452,
      "loss": 2.2897,
      "step": 35413
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.7658891677856445,
      "learning_rate": 0.0005656987825880771,
      "loss": 2.3779,
      "step": 35414
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6827353835105896,
      "learning_rate": 0.0005656968831953106,
      "loss": 2.0574,
      "step": 35415
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6876406669616699,
      "learning_rate": 0.0005656949837531459,
      "loss": 2.2719,
      "step": 35416
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6821773052215576,
      "learning_rate": 0.0005656930842615836,
      "loss": 2.3105,
      "step": 35417
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6347643136978149,
      "learning_rate": 0.000565691184720624,
      "loss": 1.9555,
      "step": 35418
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6871092319488525,
      "learning_rate": 0.0005656892851302673,
      "loss": 2.2989,
      "step": 35419
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.7506490349769592,
      "learning_rate": 0.000565687385490514,
      "loss": 2.3868,
      "step": 35420
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6864736676216125,
      "learning_rate": 0.0005656854858013643,
      "loss": 2.2496,
      "step": 35421
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.9161120653152466,
      "learning_rate": 0.0005656835860628188,
      "loss": 2.4677,
      "step": 35422
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6358354091644287,
      "learning_rate": 0.0005656816862748777,
      "loss": 2.2609,
      "step": 35423
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6213020086288452,
      "learning_rate": 0.0005656797864375412,
      "loss": 2.0483,
      "step": 35424
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6530936360359192,
      "learning_rate": 0.00056567788655081,
      "loss": 2.4481,
      "step": 35425
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6603441834449768,
      "learning_rate": 0.0005656759866146841,
      "loss": 2.107,
      "step": 35426
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6277174353599548,
      "learning_rate": 0.0005656740866291642,
      "loss": 2.1594,
      "step": 35427
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6224789023399353,
      "learning_rate": 0.0005656721865942503,
      "loss": 2.3576,
      "step": 35428
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6550918817520142,
      "learning_rate": 0.0005656702865099429,
      "loss": 2.3164,
      "step": 35429
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.7009689807891846,
      "learning_rate": 0.0005656683863762425,
      "loss": 2.0254,
      "step": 35430
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6759136915206909,
      "learning_rate": 0.0005656664861931494,
      "loss": 2.1495,
      "step": 35431
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6969684958457947,
      "learning_rate": 0.0005656645859606638,
      "loss": 2.2069,
      "step": 35432
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.7089580297470093,
      "learning_rate": 0.0005656626856787861,
      "loss": 2.3616,
      "step": 35433
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6632055044174194,
      "learning_rate": 0.0005656607853475167,
      "loss": 2.5467,
      "step": 35434
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6444494724273682,
      "learning_rate": 0.0005656588849668559,
      "loss": 2.035,
      "step": 35435
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.716224730014801,
      "learning_rate": 0.0005656569845368042,
      "loss": 2.4314,
      "step": 35436
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6982794404029846,
      "learning_rate": 0.0005656550840573619,
      "loss": 2.1296,
      "step": 35437
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6614632606506348,
      "learning_rate": 0.0005656531835285291,
      "loss": 2.366,
      "step": 35438
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6707803606987,
      "learning_rate": 0.0005656512829503065,
      "loss": 2.4159,
      "step": 35439
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6576154828071594,
      "learning_rate": 0.0005656493823226943,
      "loss": 2.3686,
      "step": 35440
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6410094499588013,
      "learning_rate": 0.0005656474816456928,
      "loss": 2.2606,
      "step": 35441
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6867085099220276,
      "learning_rate": 0.0005656455809193026,
      "loss": 2.4684,
      "step": 35442
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.7356037497520447,
      "learning_rate": 0.0005656436801435236,
      "loss": 2.354,
      "step": 35443
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6210066080093384,
      "learning_rate": 0.0005656417793183567,
      "loss": 2.2851,
      "step": 35444
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6555635929107666,
      "learning_rate": 0.0005656398784438018,
      "loss": 2.3126,
      "step": 35445
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6796082854270935,
      "learning_rate": 0.0005656379775198595,
      "loss": 2.1809,
      "step": 35446
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6992048025131226,
      "learning_rate": 0.0005656360765465301,
      "loss": 2.1683,
      "step": 35447
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6622158288955688,
      "learning_rate": 0.0005656341755238139,
      "loss": 2.2204,
      "step": 35448
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6523889899253845,
      "learning_rate": 0.0005656322744517113,
      "loss": 2.1822,
      "step": 35449
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.7033731341362,
      "learning_rate": 0.0005656303733302227,
      "loss": 2.4708,
      "step": 35450
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6743374466896057,
      "learning_rate": 0.0005656284721593483,
      "loss": 2.3477,
      "step": 35451
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6762815713882446,
      "learning_rate": 0.0005656265709390886,
      "loss": 2.5009,
      "step": 35452
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6608911156654358,
      "learning_rate": 0.0005656246696694438,
      "loss": 2.321,
      "step": 35453
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.7060365080833435,
      "learning_rate": 0.0005656227683504145,
      "loss": 2.3171,
      "step": 35454
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.5971441864967346,
      "learning_rate": 0.0005656208669820009,
      "loss": 2.2535,
      "step": 35455
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6131190657615662,
      "learning_rate": 0.0005656189655642034,
      "loss": 2.4118,
      "step": 35456
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.8079591393470764,
      "learning_rate": 0.0005656170640970222,
      "loss": 2.3097,
      "step": 35457
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.7013736963272095,
      "learning_rate": 0.0005656151625804579,
      "loss": 2.36,
      "step": 35458
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6363208293914795,
      "learning_rate": 0.0005656132610145106,
      "loss": 2.3914,
      "step": 35459
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.7872263193130493,
      "learning_rate": 0.0005656113593991809,
      "loss": 2.2838,
      "step": 35460
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.7739970684051514,
      "learning_rate": 0.0005656094577344689,
      "loss": 2.4233,
      "step": 35461
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6467483043670654,
      "learning_rate": 0.0005656075560203752,
      "loss": 2.1481,
      "step": 35462
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6443352699279785,
      "learning_rate": 0.0005656056542569,
      "loss": 2.3028,
      "step": 35463
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.7275155782699585,
      "learning_rate": 0.0005656037524440437,
      "loss": 2.3584,
      "step": 35464
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.9404756426811218,
      "learning_rate": 0.0005656018505818067,
      "loss": 2.3452,
      "step": 35465
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.7322482466697693,
      "learning_rate": 0.0005655999486701892,
      "loss": 2.3845,
      "step": 35466
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.7724816799163818,
      "learning_rate": 0.0005655980467091918,
      "loss": 2.4945,
      "step": 35467
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.7880923748016357,
      "learning_rate": 0.0005655961446988147,
      "loss": 2.477,
      "step": 35468
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.9177581667900085,
      "learning_rate": 0.000565594242639058,
      "loss": 2.1734,
      "step": 35469
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6395202875137329,
      "learning_rate": 0.0005655923405299226,
      "loss": 2.2804,
      "step": 35470
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.8194028735160828,
      "learning_rate": 0.0005655904383714084,
      "loss": 2.4567,
      "step": 35471
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.764998733997345,
      "learning_rate": 0.000565588536163516,
      "loss": 2.289,
      "step": 35472
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.7824713587760925,
      "learning_rate": 0.0005655866339062457,
      "loss": 2.2066,
      "step": 35473
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6126126050949097,
      "learning_rate": 0.0005655847315995979,
      "loss": 2.284,
      "step": 35474
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6626967787742615,
      "learning_rate": 0.0005655828292435728,
      "loss": 2.1434,
      "step": 35475
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.9403309226036072,
      "learning_rate": 0.0005655809268381708,
      "loss": 2.2928,
      "step": 35476
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6666468381881714,
      "learning_rate": 0.0005655790243833923,
      "loss": 2.1835,
      "step": 35477
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.7597939968109131,
      "learning_rate": 0.0005655771218792377,
      "loss": 2.1941,
      "step": 35478
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.7313436269760132,
      "learning_rate": 0.0005655752193257072,
      "loss": 2.5194,
      "step": 35479
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6911666989326477,
      "learning_rate": 0.0005655733167228014,
      "loss": 2.1108,
      "step": 35480
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6309380531311035,
      "learning_rate": 0.0005655714140705203,
      "loss": 2.421,
      "step": 35481
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6525343656539917,
      "learning_rate": 0.0005655695113688647,
      "loss": 2.1939,
      "step": 35482
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6845937967300415,
      "learning_rate": 0.0005655676086178346,
      "loss": 2.2549,
      "step": 35483
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.7309908270835876,
      "learning_rate": 0.0005655657058174304,
      "loss": 2.3062,
      "step": 35484
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6407964825630188,
      "learning_rate": 0.0005655638029676527,
      "loss": 2.4343,
      "step": 35485
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6397050619125366,
      "learning_rate": 0.0005655619000685015,
      "loss": 2.4689,
      "step": 35486
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6775479912757874,
      "learning_rate": 0.0005655599971199774,
      "loss": 2.3634,
      "step": 35487
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.8494307994842529,
      "learning_rate": 0.0005655580941220808,
      "loss": 2.4501,
      "step": 35488
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.700892448425293,
      "learning_rate": 0.0005655561910748117,
      "loss": 2.1858,
      "step": 35489
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6868441700935364,
      "learning_rate": 0.0005655542879781708,
      "loss": 2.2886,
      "step": 35490
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.7337262034416199,
      "learning_rate": 0.0005655523848321584,
      "loss": 2.0659,
      "step": 35491
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6882655620574951,
      "learning_rate": 0.0005655504816367748,
      "loss": 2.1358,
      "step": 35492
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6583343148231506,
      "learning_rate": 0.0005655485783920203,
      "loss": 2.4523,
      "step": 35493
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6924512982368469,
      "learning_rate": 0.0005655466750978953,
      "loss": 2.3538,
      "step": 35494
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6849538683891296,
      "learning_rate": 0.0005655447717544002,
      "loss": 2.3638,
      "step": 35495
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6745437383651733,
      "learning_rate": 0.0005655428683615353,
      "loss": 2.3204,
      "step": 35496
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6613354086875916,
      "learning_rate": 0.0005655409649193009,
      "loss": 2.21,
      "step": 35497
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.7392359375953674,
      "learning_rate": 0.0005655390614276976,
      "loss": 2.2956,
      "step": 35498
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.7659130096435547,
      "learning_rate": 0.0005655371578867255,
      "loss": 2.2043,
      "step": 35499
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6895366907119751,
      "learning_rate": 0.0005655352542963849,
      "loss": 2.472,
      "step": 35500
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6404862403869629,
      "learning_rate": 0.0005655333506566763,
      "loss": 2.483,
      "step": 35501
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.5831575989723206,
      "learning_rate": 0.0005655314469676002,
      "loss": 2.0723,
      "step": 35502
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6565300822257996,
      "learning_rate": 0.0005655295432291567,
      "loss": 2.5485,
      "step": 35503
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6663715839385986,
      "learning_rate": 0.0005655276394413462,
      "loss": 2.2921,
      "step": 35504
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6547670960426331,
      "learning_rate": 0.0005655257356041691,
      "loss": 2.1813,
      "step": 35505
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6299989223480225,
      "learning_rate": 0.0005655238317176258,
      "loss": 2.1045,
      "step": 35506
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6686801314353943,
      "learning_rate": 0.0005655219277817167,
      "loss": 2.1801,
      "step": 35507
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.7183594107627869,
      "learning_rate": 0.0005655200237964419,
      "loss": 2.2504,
      "step": 35508
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6921826004981995,
      "learning_rate": 0.000565518119761802,
      "loss": 2.3086,
      "step": 35509
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6496779322624207,
      "learning_rate": 0.0005655162156777973,
      "loss": 2.3307,
      "step": 35510
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.752861499786377,
      "learning_rate": 0.0005655143115444281,
      "loss": 2.2137,
      "step": 35511
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.8218972682952881,
      "learning_rate": 0.0005655124073616947,
      "loss": 2.1398,
      "step": 35512
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.654738187789917,
      "learning_rate": 0.0005655105031295976,
      "loss": 2.2993,
      "step": 35513
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.9761565327644348,
      "learning_rate": 0.0005655085988481371,
      "loss": 2.483,
      "step": 35514
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6907267570495605,
      "learning_rate": 0.0005655066945173135,
      "loss": 2.3677,
      "step": 35515
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6346783638000488,
      "learning_rate": 0.0005655047901371273,
      "loss": 2.5211,
      "step": 35516
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.7207716703414917,
      "learning_rate": 0.0005655028857075786,
      "loss": 2.3017,
      "step": 35517
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.7695400714874268,
      "learning_rate": 0.0005655009812286679,
      "loss": 2.3776,
      "step": 35518
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6730092167854309,
      "learning_rate": 0.0005654990767003956,
      "loss": 2.237,
      "step": 35519
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6736600399017334,
      "learning_rate": 0.000565497172122762,
      "loss": 2.3303,
      "step": 35520
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.7935599088668823,
      "learning_rate": 0.0005654952674957676,
      "loss": 2.2119,
      "step": 35521
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.7140854001045227,
      "learning_rate": 0.0005654933628194125,
      "loss": 2.2172,
      "step": 35522
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6936950087547302,
      "learning_rate": 0.0005654914580936972,
      "loss": 2.3217,
      "step": 35523
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.7256315350532532,
      "learning_rate": 0.0005654895533186219,
      "loss": 2.2722,
      "step": 35524
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.7655079960823059,
      "learning_rate": 0.0005654876484941872,
      "loss": 2.1292,
      "step": 35525
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.7007485032081604,
      "learning_rate": 0.0005654857436203934,
      "loss": 2.493,
      "step": 35526
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6966292858123779,
      "learning_rate": 0.0005654838386972407,
      "loss": 2.3974,
      "step": 35527
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.8255247473716736,
      "learning_rate": 0.0005654819337247295,
      "loss": 2.2702,
      "step": 35528
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.7086870670318604,
      "learning_rate": 0.0005654800287028602,
      "loss": 2.2033,
      "step": 35529
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6707996129989624,
      "learning_rate": 0.0005654781236316333,
      "loss": 2.3166,
      "step": 35530
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6716998815536499,
      "learning_rate": 0.0005654762185110489,
      "loss": 2.4554,
      "step": 35531
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6396287083625793,
      "learning_rate": 0.0005654743133411075,
      "loss": 2.1697,
      "step": 35532
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6522142887115479,
      "learning_rate": 0.0005654724081218093,
      "loss": 2.3397,
      "step": 35533
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.8911333084106445,
      "learning_rate": 0.0005654705028531548,
      "loss": 2.2865,
      "step": 35534
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6377482414245605,
      "learning_rate": 0.0005654685975351445,
      "loss": 2.2526,
      "step": 35535
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6405201554298401,
      "learning_rate": 0.0005654666921677783,
      "loss": 2.357,
      "step": 35536
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6496091485023499,
      "learning_rate": 0.000565464786751057,
      "loss": 2.1965,
      "step": 35537
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.7238649129867554,
      "learning_rate": 0.0005654628812849808,
      "loss": 2.3494,
      "step": 35538
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6667011380195618,
      "learning_rate": 0.0005654609757695501,
      "loss": 2.3185,
      "step": 35539
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6860635280609131,
      "learning_rate": 0.000565459070204765,
      "loss": 2.4996,
      "step": 35540
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6475871205329895,
      "learning_rate": 0.0005654571645906262,
      "loss": 2.0682,
      "step": 35541
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6436081528663635,
      "learning_rate": 0.0005654552589271338,
      "loss": 2.3355,
      "step": 35542
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6658452749252319,
      "learning_rate": 0.0005654533532142884,
      "loss": 2.5376,
      "step": 35543
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6774945259094238,
      "learning_rate": 0.00056545144745209,
      "loss": 2.2326,
      "step": 35544
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6767246723175049,
      "learning_rate": 0.0005654495416405393,
      "loss": 2.3873,
      "step": 35545
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.7820090055465698,
      "learning_rate": 0.0005654476357796365,
      "loss": 2.6162,
      "step": 35546
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6542949080467224,
      "learning_rate": 0.000565445729869382,
      "loss": 2.3044,
      "step": 35547
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6699367761611938,
      "learning_rate": 0.000565443823909776,
      "loss": 2.2781,
      "step": 35548
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.7491821646690369,
      "learning_rate": 0.0005654419179008191,
      "loss": 2.336,
      "step": 35549
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.7694914937019348,
      "learning_rate": 0.0005654400118425116,
      "loss": 2.316,
      "step": 35550
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6481150984764099,
      "learning_rate": 0.0005654381057348537,
      "loss": 2.3928,
      "step": 35551
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6976972222328186,
      "learning_rate": 0.0005654361995778459,
      "loss": 2.2951,
      "step": 35552
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.7065452933311462,
      "learning_rate": 0.0005654342933714884,
      "loss": 2.318,
      "step": 35553
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.692125678062439,
      "learning_rate": 0.0005654323871157816,
      "loss": 2.2857,
      "step": 35554
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6696075797080994,
      "learning_rate": 0.0005654304808107261,
      "loss": 2.2485,
      "step": 35555
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.8145807385444641,
      "learning_rate": 0.000565428574456322,
      "loss": 2.1243,
      "step": 35556
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.7869498133659363,
      "learning_rate": 0.0005654266680525697,
      "loss": 2.281,
      "step": 35557
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6823799014091492,
      "learning_rate": 0.0005654247615994695,
      "loss": 2.2185,
      "step": 35558
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.7250123023986816,
      "learning_rate": 0.000565422855097022,
      "loss": 2.2068,
      "step": 35559
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.0697288513183594,
      "learning_rate": 0.0005654209485452272,
      "loss": 2.2081,
      "step": 35560
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.8587592840194702,
      "learning_rate": 0.0005654190419440857,
      "loss": 2.3155,
      "step": 35561
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.7171779870986938,
      "learning_rate": 0.0005654171352935979,
      "loss": 2.3425,
      "step": 35562
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.7770671248435974,
      "learning_rate": 0.0005654152285937639,
      "loss": 2.1572,
      "step": 35563
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.7221051454544067,
      "learning_rate": 0.0005654133218445843,
      "loss": 2.7135,
      "step": 35564
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.7939700484275818,
      "learning_rate": 0.0005654114150460593,
      "loss": 2.0499,
      "step": 35565
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.7179582715034485,
      "learning_rate": 0.0005654095081981893,
      "loss": 2.2027,
      "step": 35566
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.8796750903129578,
      "learning_rate": 0.0005654076013009747,
      "loss": 2.1911,
      "step": 35567
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6840811967849731,
      "learning_rate": 0.0005654056943544158,
      "loss": 2.2351,
      "step": 35568
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.717374324798584,
      "learning_rate": 0.0005654037873585129,
      "loss": 2.1973,
      "step": 35569
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.7844551801681519,
      "learning_rate": 0.0005654018803132665,
      "loss": 2.3336,
      "step": 35570
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.7731530666351318,
      "learning_rate": 0.0005653999732186768,
      "loss": 2.2157,
      "step": 35571
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.7272152304649353,
      "learning_rate": 0.0005653980660747444,
      "loss": 2.2162,
      "step": 35572
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.694014310836792,
      "learning_rate": 0.0005653961588814693,
      "loss": 2.606,
      "step": 35573
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.674493670463562,
      "learning_rate": 0.0005653942516388521,
      "loss": 2.3386,
      "step": 35574
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.7510790824890137,
      "learning_rate": 0.0005653923443468933,
      "loss": 2.3194,
      "step": 35575
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6773305535316467,
      "learning_rate": 0.0005653904370055928,
      "loss": 2.1773,
      "step": 35576
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.7557963132858276,
      "learning_rate": 0.0005653885296149513,
      "loss": 2.2009,
      "step": 35577
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6805627346038818,
      "learning_rate": 0.0005653866221749691,
      "loss": 2.2589,
      "step": 35578
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6436289548873901,
      "learning_rate": 0.0005653847146856464,
      "loss": 2.4463,
      "step": 35579
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.66508549451828,
      "learning_rate": 0.0005653828071469838,
      "loss": 2.3356,
      "step": 35580
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.7186981439590454,
      "learning_rate": 0.0005653808995589814,
      "loss": 2.255,
      "step": 35581
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6959295868873596,
      "learning_rate": 0.0005653789919216398,
      "loss": 2.4428,
      "step": 35582
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6889892816543579,
      "learning_rate": 0.0005653770842349592,
      "loss": 2.382,
      "step": 35583
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6818506121635437,
      "learning_rate": 0.00056537517649894,
      "loss": 2.4266,
      "step": 35584
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6879606246948242,
      "learning_rate": 0.0005653732687135826,
      "loss": 2.2986,
      "step": 35585
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.7373321056365967,
      "learning_rate": 0.0005653713608788872,
      "loss": 2.2177,
      "step": 35586
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6238737106323242,
      "learning_rate": 0.0005653694529948542,
      "loss": 2.2782,
      "step": 35587
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6735102534294128,
      "learning_rate": 0.0005653675450614841,
      "loss": 2.2479,
      "step": 35588
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.7027131915092468,
      "learning_rate": 0.0005653656370787772,
      "loss": 2.4384,
      "step": 35589
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.7097833156585693,
      "learning_rate": 0.0005653637290467337,
      "loss": 2.365,
      "step": 35590
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6797823309898376,
      "learning_rate": 0.0005653618209653543,
      "loss": 2.2648,
      "step": 35591
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6488977670669556,
      "learning_rate": 0.0005653599128346389,
      "loss": 2.4018,
      "step": 35592
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.744804322719574,
      "learning_rate": 0.0005653580046545882,
      "loss": 2.1798,
      "step": 35593
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6439215540885925,
      "learning_rate": 0.0005653560964252024,
      "loss": 2.1851,
      "step": 35594
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.666659951210022,
      "learning_rate": 0.0005653541881464819,
      "loss": 2.3199,
      "step": 35595
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6508718729019165,
      "learning_rate": 0.000565352279818427,
      "loss": 2.2248,
      "step": 35596
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.7481127381324768,
      "learning_rate": 0.0005653503714410381,
      "loss": 2.5892,
      "step": 35597
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.710972785949707,
      "learning_rate": 0.0005653484630143156,
      "loss": 2.3295,
      "step": 35598
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.7209494709968567,
      "learning_rate": 0.0005653465545382598,
      "loss": 2.2639,
      "step": 35599
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.7086877822875977,
      "learning_rate": 0.0005653446460128711,
      "loss": 2.3995,
      "step": 35600
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.7235702276229858,
      "learning_rate": 0.0005653427374381498,
      "loss": 2.1087,
      "step": 35601
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6727982759475708,
      "learning_rate": 0.0005653408288140964,
      "loss": 2.295,
      "step": 35602
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.7072234153747559,
      "learning_rate": 0.0005653389201407109,
      "loss": 2.1561,
      "step": 35603
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6448248624801636,
      "learning_rate": 0.000565337011417994,
      "loss": 2.2555,
      "step": 35604
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6751402020454407,
      "learning_rate": 0.0005653351026459459,
      "loss": 1.9605,
      "step": 35605
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6999319195747375,
      "learning_rate": 0.0005653331938245671,
      "loss": 2.2647,
      "step": 35606
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6705896258354187,
      "learning_rate": 0.0005653312849538577,
      "loss": 2.3754,
      "step": 35607
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.7094058990478516,
      "learning_rate": 0.0005653293760338183,
      "loss": 2.1901,
      "step": 35608
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.7131985425949097,
      "learning_rate": 0.0005653274670644492,
      "loss": 2.3802,
      "step": 35609
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6952202916145325,
      "learning_rate": 0.0005653255580457507,
      "loss": 2.0818,
      "step": 35610
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.8179839849472046,
      "learning_rate": 0.0005653236489777231,
      "loss": 2.3615,
      "step": 35611
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.7369546890258789,
      "learning_rate": 0.0005653217398603668,
      "loss": 2.2802,
      "step": 35612
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6703979969024658,
      "learning_rate": 0.0005653198306936822,
      "loss": 2.2665,
      "step": 35613
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6193094849586487,
      "learning_rate": 0.0005653179214776697,
      "loss": 2.4155,
      "step": 35614
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.7268566489219666,
      "learning_rate": 0.0005653160122123295,
      "loss": 2.3892,
      "step": 35615
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6515614986419678,
      "learning_rate": 0.0005653141028976622,
      "loss": 2.2741,
      "step": 35616
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6502383947372437,
      "learning_rate": 0.0005653121935336679,
      "loss": 2.0368,
      "step": 35617
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6823809742927551,
      "learning_rate": 0.0005653102841203471,
      "loss": 2.418,
      "step": 35618
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.7364628911018372,
      "learning_rate": 0.0005653083746577,
      "loss": 2.5306,
      "step": 35619
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6694746613502502,
      "learning_rate": 0.0005653064651457272,
      "loss": 2.2269,
      "step": 35620
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6722795963287354,
      "learning_rate": 0.0005653045555844288,
      "loss": 2.3201,
      "step": 35621
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6883134841918945,
      "learning_rate": 0.0005653026459738053,
      "loss": 2.436,
      "step": 35622
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6987122297286987,
      "learning_rate": 0.0005653007363138571,
      "loss": 2.2883,
      "step": 35623
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6400719285011292,
      "learning_rate": 0.0005652988266045845,
      "loss": 2.3519,
      "step": 35624
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.678229033946991,
      "learning_rate": 0.0005652969168459878,
      "loss": 2.3278,
      "step": 35625
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.5917131900787354,
      "learning_rate": 0.0005652950070380673,
      "loss": 2.4335,
      "step": 35626
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.685236394405365,
      "learning_rate": 0.0005652930971808236,
      "loss": 2.3566,
      "step": 35627
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.7034621238708496,
      "learning_rate": 0.0005652911872742569,
      "loss": 2.1731,
      "step": 35628
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6285598278045654,
      "learning_rate": 0.0005652892773183674,
      "loss": 2.1973,
      "step": 35629
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.7584246397018433,
      "learning_rate": 0.0005652873673131557,
      "loss": 2.4661,
      "step": 35630
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.639430582523346,
      "learning_rate": 0.0005652854572586221,
      "loss": 2.5059,
      "step": 35631
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6566319465637207,
      "learning_rate": 0.000565283547154767,
      "loss": 2.2444,
      "step": 35632
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.7446306943893433,
      "learning_rate": 0.0005652816370015906,
      "loss": 2.2723,
      "step": 35633
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.7421389818191528,
      "learning_rate": 0.0005652797267990933,
      "loss": 2.1388,
      "step": 35634
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.5926917791366577,
      "learning_rate": 0.0005652778165472755,
      "loss": 2.2428,
      "step": 35635
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.668285071849823,
      "learning_rate": 0.0005652759062461377,
      "loss": 2.1013,
      "step": 35636
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.8502413630485535,
      "learning_rate": 0.0005652739958956799,
      "loss": 2.2038,
      "step": 35637
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6882644295692444,
      "learning_rate": 0.0005652720854959027,
      "loss": 2.2866,
      "step": 35638
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6494699716567993,
      "learning_rate": 0.0005652701750468064,
      "loss": 2.1045,
      "step": 35639
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.7433904409408569,
      "learning_rate": 0.0005652682645483914,
      "loss": 2.3167,
      "step": 35640
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6872592568397522,
      "learning_rate": 0.000565266354000658,
      "loss": 2.4303,
      "step": 35641
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.7466116547584534,
      "learning_rate": 0.0005652644434036066,
      "loss": 2.3889,
      "step": 35642
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6972972750663757,
      "learning_rate": 0.0005652625327572376,
      "loss": 2.3949,
      "step": 35643
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6518115401268005,
      "learning_rate": 0.0005652606220615513,
      "loss": 2.4805,
      "step": 35644
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6648131608963013,
      "learning_rate": 0.0005652587113165478,
      "loss": 2.2073,
      "step": 35645
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.7965183854103088,
      "learning_rate": 0.0005652568005222279,
      "loss": 2.5613,
      "step": 35646
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6594915986061096,
      "learning_rate": 0.0005652548896785917,
      "loss": 2.104,
      "step": 35647
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6506693363189697,
      "learning_rate": 0.0005652529787856397,
      "loss": 2.3415,
      "step": 35648
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.7060573697090149,
      "learning_rate": 0.0005652510678433721,
      "loss": 2.4103,
      "step": 35649
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6797043681144714,
      "learning_rate": 0.0005652491568517892,
      "loss": 2.397,
      "step": 35650
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.659545361995697,
      "learning_rate": 0.0005652472458108917,
      "loss": 2.46,
      "step": 35651
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6603894829750061,
      "learning_rate": 0.0005652453347206796,
      "loss": 2.4794,
      "step": 35652
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6754115223884583,
      "learning_rate": 0.0005652434235811534,
      "loss": 2.2457,
      "step": 35653
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.700142502784729,
      "learning_rate": 0.0005652415123923134,
      "loss": 2.2226,
      "step": 35654
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.7530637383460999,
      "learning_rate": 0.0005652396011541601,
      "loss": 2.3068,
      "step": 35655
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6838621497154236,
      "learning_rate": 0.0005652376898666936,
      "loss": 2.4618,
      "step": 35656
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.7098304629325867,
      "learning_rate": 0.0005652357785299146,
      "loss": 2.2902,
      "step": 35657
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6001248359680176,
      "learning_rate": 0.0005652338671438232,
      "loss": 2.3095,
      "step": 35658
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6417641043663025,
      "learning_rate": 0.0005652319557084198,
      "loss": 2.1945,
      "step": 35659
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6903370022773743,
      "learning_rate": 0.0005652300442237047,
      "loss": 2.1396,
      "step": 35660
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6736688613891602,
      "learning_rate": 0.0005652281326896783,
      "loss": 2.3245,
      "step": 35661
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6583737134933472,
      "learning_rate": 0.0005652262211063411,
      "loss": 2.5674,
      "step": 35662
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6844966411590576,
      "learning_rate": 0.0005652243094736933,
      "loss": 2.2652,
      "step": 35663
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.7709034085273743,
      "learning_rate": 0.0005652223977917353,
      "loss": 2.3833,
      "step": 35664
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.7176604866981506,
      "learning_rate": 0.0005652204860604674,
      "loss": 2.3254,
      "step": 35665
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6619623899459839,
      "learning_rate": 0.0005652185742798901,
      "loss": 2.1409,
      "step": 35666
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.7236087322235107,
      "learning_rate": 0.0005652166624500036,
      "loss": 2.4183,
      "step": 35667
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6783744692802429,
      "learning_rate": 0.0005652147505708083,
      "loss": 2.3931,
      "step": 35668
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6651982665061951,
      "learning_rate": 0.0005652128386423046,
      "loss": 2.434,
      "step": 35669
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6347401142120361,
      "learning_rate": 0.0005652109266644929,
      "loss": 2.2229,
      "step": 35670
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.7320294976234436,
      "learning_rate": 0.0005652090146373733,
      "loss": 2.3424,
      "step": 35671
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6402605772018433,
      "learning_rate": 0.0005652071025609465,
      "loss": 2.4397,
      "step": 35672
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6432849168777466,
      "learning_rate": 0.0005652051904352126,
      "loss": 2.2087,
      "step": 35673
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.7278282642364502,
      "learning_rate": 0.0005652032782601722,
      "loss": 2.0332,
      "step": 35674
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6372876763343811,
      "learning_rate": 0.0005652013660358253,
      "loss": 2.4099,
      "step": 35675
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.659281313419342,
      "learning_rate": 0.0005651994537621725,
      "loss": 2.0565,
      "step": 35676
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.7478809356689453,
      "learning_rate": 0.0005651975414392143,
      "loss": 2.3391,
      "step": 35677
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6911638379096985,
      "learning_rate": 0.0005651956290669507,
      "loss": 2.3867,
      "step": 35678
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6379761099815369,
      "learning_rate": 0.0005651937166453822,
      "loss": 2.519,
      "step": 35679
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6774699687957764,
      "learning_rate": 0.0005651918041745094,
      "loss": 2.2716,
      "step": 35680
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6444521546363831,
      "learning_rate": 0.0005651898916543322,
      "loss": 1.9913,
      "step": 35681
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.7121098041534424,
      "learning_rate": 0.0005651879790848513,
      "loss": 2.3321,
      "step": 35682
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6541879773139954,
      "learning_rate": 0.0005651860664660669,
      "loss": 2.4243,
      "step": 35683
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6694767475128174,
      "learning_rate": 0.0005651841537979794,
      "loss": 2.226,
      "step": 35684
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.66983962059021,
      "learning_rate": 0.0005651822410805892,
      "loss": 2.1406,
      "step": 35685
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.713189959526062,
      "learning_rate": 0.0005651803283138967,
      "loss": 2.4713,
      "step": 35686
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6668662428855896,
      "learning_rate": 0.000565178415497902,
      "loss": 2.3698,
      "step": 35687
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6580095887184143,
      "learning_rate": 0.0005651765026326057,
      "loss": 2.2625,
      "step": 35688
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6596013307571411,
      "learning_rate": 0.0005651745897180081,
      "loss": 2.4177,
      "step": 35689
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.7294360995292664,
      "learning_rate": 0.0005651726767541095,
      "loss": 2.3536,
      "step": 35690
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6103489995002747,
      "learning_rate": 0.0005651707637409104,
      "loss": 2.1403,
      "step": 35691
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6379361748695374,
      "learning_rate": 0.0005651688506784109,
      "loss": 2.2319,
      "step": 35692
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6667768955230713,
      "learning_rate": 0.0005651669375666117,
      "loss": 2.2045,
      "step": 35693
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6656742691993713,
      "learning_rate": 0.0005651650244055127,
      "loss": 2.1314,
      "step": 35694
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.710555911064148,
      "learning_rate": 0.0005651631111951147,
      "loss": 2.2764,
      "step": 35695
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6896600723266602,
      "learning_rate": 0.0005651611979354178,
      "loss": 2.0861,
      "step": 35696
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6602007150650024,
      "learning_rate": 0.0005651592846264224,
      "loss": 2.0875,
      "step": 35697
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6652681827545166,
      "learning_rate": 0.000565157371268129,
      "loss": 2.2903,
      "step": 35698
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.7072902917861938,
      "learning_rate": 0.0005651554578605378,
      "loss": 2.2779,
      "step": 35699
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.7005916833877563,
      "learning_rate": 0.0005651535444036492,
      "loss": 2.176,
      "step": 35700
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.7027247548103333,
      "learning_rate": 0.0005651516308974635,
      "loss": 2.2318,
      "step": 35701
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.7588793039321899,
      "learning_rate": 0.0005651497173419811,
      "loss": 2.0336,
      "step": 35702
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6265660524368286,
      "learning_rate": 0.0005651478037372023,
      "loss": 2.073,
      "step": 35703
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.7169646620750427,
      "learning_rate": 0.0005651458900831277,
      "loss": 2.2854,
      "step": 35704
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.640208899974823,
      "learning_rate": 0.0005651439763797574,
      "loss": 2.0979,
      "step": 35705
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.7676445245742798,
      "learning_rate": 0.0005651420626270918,
      "loss": 2.1973,
      "step": 35706
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.675001323223114,
      "learning_rate": 0.0005651401488251314,
      "loss": 2.1897,
      "step": 35707
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.7122179269790649,
      "learning_rate": 0.0005651382349738763,
      "loss": 2.4207,
      "step": 35708
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6375358700752258,
      "learning_rate": 0.000565136321073327,
      "loss": 2.2977,
      "step": 35709
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6309075355529785,
      "learning_rate": 0.0005651344071234839,
      "loss": 2.2753,
      "step": 35710
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6407337188720703,
      "learning_rate": 0.0005651324931243473,
      "loss": 2.3579,
      "step": 35711
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6623150110244751,
      "learning_rate": 0.0005651305790759176,
      "loss": 2.1339,
      "step": 35712
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6959637403488159,
      "learning_rate": 0.000565128664978195,
      "loss": 2.3592,
      "step": 35713
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6103479862213135,
      "learning_rate": 0.0005651267508311801,
      "loss": 2.0463,
      "step": 35714
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.783368706703186,
      "learning_rate": 0.0005651248366348731,
      "loss": 2.4674,
      "step": 35715
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.67523592710495,
      "learning_rate": 0.0005651229223892744,
      "loss": 2.2356,
      "step": 35716
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.7272155284881592,
      "learning_rate": 0.0005651210080943843,
      "loss": 2.2996,
      "step": 35717
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6861841678619385,
      "learning_rate": 0.0005651190937502032,
      "loss": 2.2067,
      "step": 35718
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.7596644163131714,
      "learning_rate": 0.0005651171793567315,
      "loss": 2.0903,
      "step": 35719
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6441349387168884,
      "learning_rate": 0.0005651152649139695,
      "loss": 2.112,
      "step": 35720
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.743107259273529,
      "learning_rate": 0.0005651133504219176,
      "loss": 2.1266,
      "step": 35721
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.7025794386863708,
      "learning_rate": 0.0005651114358805761,
      "loss": 2.5283,
      "step": 35722
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6076042056083679,
      "learning_rate": 0.0005651095212899453,
      "loss": 2.404,
      "step": 35723
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.7294290661811829,
      "learning_rate": 0.0005651076066500258,
      "loss": 2.2463,
      "step": 35724
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.750727117061615,
      "learning_rate": 0.0005651056919608176,
      "loss": 2.0782,
      "step": 35725
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.7145500779151917,
      "learning_rate": 0.0005651037772223214,
      "loss": 2.0717,
      "step": 35726
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.698927640914917,
      "learning_rate": 0.0005651018624345373,
      "loss": 2.3593,
      "step": 35727
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.7063083052635193,
      "learning_rate": 0.0005650999475974659,
      "loss": 2.0546,
      "step": 35728
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6711869835853577,
      "learning_rate": 0.0005650980327111072,
      "loss": 2.0007,
      "step": 35729
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6163240075111389,
      "learning_rate": 0.000565096117775462,
      "loss": 2.3002,
      "step": 35730
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6736705303192139,
      "learning_rate": 0.0005650942027905303,
      "loss": 2.3434,
      "step": 35731
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.7085778713226318,
      "learning_rate": 0.0005650922877563125,
      "loss": 2.1245,
      "step": 35732
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6458241939544678,
      "learning_rate": 0.0005650903726728093,
      "loss": 2.1885,
      "step": 35733
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6153748035430908,
      "learning_rate": 0.0005650884575400206,
      "loss": 2.1947,
      "step": 35734
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6744362711906433,
      "learning_rate": 0.0005650865423579469,
      "loss": 2.0654,
      "step": 35735
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6508287191390991,
      "learning_rate": 0.0005650846271265887,
      "loss": 2.3259,
      "step": 35736
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6935011744499207,
      "learning_rate": 0.0005650827118459463,
      "loss": 2.3934,
      "step": 35737
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6607697606086731,
      "learning_rate": 0.0005650807965160201,
      "loss": 2.1746,
      "step": 35738
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.7584920525550842,
      "learning_rate": 0.0005650788811368101,
      "loss": 2.3745,
      "step": 35739
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6711649298667908,
      "learning_rate": 0.0005650769657083171,
      "loss": 2.2999,
      "step": 35740
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.696735680103302,
      "learning_rate": 0.0005650750502305412,
      "loss": 2.3447,
      "step": 35741
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.7138048410415649,
      "learning_rate": 0.0005650731347034831,
      "loss": 2.3928,
      "step": 35742
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6892640590667725,
      "learning_rate": 0.0005650712191271428,
      "loss": 2.3898,
      "step": 35743
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6469964981079102,
      "learning_rate": 0.0005650693035015205,
      "loss": 2.2864,
      "step": 35744
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.8324707746505737,
      "learning_rate": 0.0005650673878266171,
      "loss": 2.4006,
      "step": 35745
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6795507669448853,
      "learning_rate": 0.0005650654721024325,
      "loss": 2.2205,
      "step": 35746
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.7191525101661682,
      "learning_rate": 0.0005650635563289672,
      "loss": 2.1212,
      "step": 35747
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.7251082062721252,
      "learning_rate": 0.0005650616405062217,
      "loss": 2.4005,
      "step": 35748
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6115285754203796,
      "learning_rate": 0.0005650597246341962,
      "loss": 2.0368,
      "step": 35749
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6739197969436646,
      "learning_rate": 0.0005650578087128911,
      "loss": 2.4461,
      "step": 35750
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.7228196859359741,
      "learning_rate": 0.0005650558927423068,
      "loss": 2.2803,
      "step": 35751
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6443017721176147,
      "learning_rate": 0.0005650539767224436,
      "loss": 2.0868,
      "step": 35752
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6259632110595703,
      "learning_rate": 0.0005650520606533018,
      "loss": 2.3642,
      "step": 35753
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6896595358848572,
      "learning_rate": 0.0005650501445348818,
      "loss": 2.2032,
      "step": 35754
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6165391802787781,
      "learning_rate": 0.0005650482283671841,
      "loss": 2.2484,
      "step": 35755
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6832812428474426,
      "learning_rate": 0.0005650463121502088,
      "loss": 2.3975,
      "step": 35756
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6652600169181824,
      "learning_rate": 0.0005650443958839564,
      "loss": 2.2249,
      "step": 35757
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6543270945549011,
      "learning_rate": 0.0005650424795684274,
      "loss": 2.3782,
      "step": 35758
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6655768752098083,
      "learning_rate": 0.0005650405632036218,
      "loss": 2.6755,
      "step": 35759
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.718205451965332,
      "learning_rate": 0.0005650386467895402,
      "loss": 2.325,
      "step": 35760
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6340285539627075,
      "learning_rate": 0.0005650367303261829,
      "loss": 2.3214,
      "step": 35761
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.7196236848831177,
      "learning_rate": 0.0005650348138135504,
      "loss": 2.6076,
      "step": 35762
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.671963095664978,
      "learning_rate": 0.0005650328972516428,
      "loss": 2.4623,
      "step": 35763
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6867666840553284,
      "learning_rate": 0.0005650309806404605,
      "loss": 2.4082,
      "step": 35764
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6559109687805176,
      "learning_rate": 0.0005650290639800041,
      "loss": 2.344,
      "step": 35765
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.660129964351654,
      "learning_rate": 0.0005650271472702738,
      "loss": 2.4205,
      "step": 35766
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6551200747489929,
      "learning_rate": 0.0005650252305112699,
      "loss": 2.3151,
      "step": 35767
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.7406945824623108,
      "learning_rate": 0.0005650233137029928,
      "loss": 2.3444,
      "step": 35768
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.7564792037010193,
      "learning_rate": 0.0005650213968454429,
      "loss": 2.0174,
      "step": 35769
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.602445662021637,
      "learning_rate": 0.0005650194799386205,
      "loss": 2.364,
      "step": 35770
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6524810791015625,
      "learning_rate": 0.000565017562982526,
      "loss": 2.1956,
      "step": 35771
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6934581995010376,
      "learning_rate": 0.0005650156459771597,
      "loss": 2.3879,
      "step": 35772
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.7387363910675049,
      "learning_rate": 0.000565013728922522,
      "loss": 2.3644,
      "step": 35773
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.5923560261726379,
      "learning_rate": 0.0005650118118186132,
      "loss": 2.1287,
      "step": 35774
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6701387763023376,
      "learning_rate": 0.0005650098946654337,
      "loss": 2.3702,
      "step": 35775
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6907190680503845,
      "learning_rate": 0.0005650079774629839,
      "loss": 2.3707,
      "step": 35776
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6810689568519592,
      "learning_rate": 0.0005650060602112641,
      "loss": 2.3483,
      "step": 35777
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.5980839729309082,
      "learning_rate": 0.0005650041429102747,
      "loss": 2.5036,
      "step": 35778
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.674005389213562,
      "learning_rate": 0.000565002225560016,
      "loss": 2.4759,
      "step": 35779
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.7279223203659058,
      "learning_rate": 0.0005650003081604884,
      "loss": 2.3237,
      "step": 35780
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6708645820617676,
      "learning_rate": 0.0005649983907116922,
      "loss": 2.2874,
      "step": 35781
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6401708722114563,
      "learning_rate": 0.0005649964732136279,
      "loss": 2.2314,
      "step": 35782
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.660350501537323,
      "learning_rate": 0.0005649945556662957,
      "loss": 2.202,
      "step": 35783
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6671704053878784,
      "learning_rate": 0.0005649926380696959,
      "loss": 2.2861,
      "step": 35784
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6326870918273926,
      "learning_rate": 0.0005649907204238291,
      "loss": 2.1294,
      "step": 35785
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6085804104804993,
      "learning_rate": 0.0005649888027286955,
      "loss": 2.2349,
      "step": 35786
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.7929406762123108,
      "learning_rate": 0.0005649868849842954,
      "loss": 2.4675,
      "step": 35787
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6388344168663025,
      "learning_rate": 0.0005649849671906294,
      "loss": 2.3883,
      "step": 35788
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6486337780952454,
      "learning_rate": 0.0005649830493476976,
      "loss": 2.6187,
      "step": 35789
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6160445213317871,
      "learning_rate": 0.0005649811314555003,
      "loss": 2.1807,
      "step": 35790
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.7143592238426208,
      "learning_rate": 0.0005649792135140382,
      "loss": 2.318,
      "step": 35791
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6475204825401306,
      "learning_rate": 0.0005649772955233115,
      "loss": 2.1894,
      "step": 35792
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6270825266838074,
      "learning_rate": 0.0005649753774833204,
      "loss": 2.1993,
      "step": 35793
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6823700666427612,
      "learning_rate": 0.0005649734593940654,
      "loss": 2.2973,
      "step": 35794
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.7102822661399841,
      "learning_rate": 0.0005649715412555468,
      "loss": 2.4027,
      "step": 35795
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6300728917121887,
      "learning_rate": 0.0005649696230677649,
      "loss": 2.3136,
      "step": 35796
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6062650084495544,
      "learning_rate": 0.0005649677048307205,
      "loss": 2.4105,
      "step": 35797
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.7108599543571472,
      "learning_rate": 0.0005649657865444133,
      "loss": 2.2876,
      "step": 35798
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.646463930606842,
      "learning_rate": 0.0005649638682088439,
      "loss": 2.223,
      "step": 35799
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6632016897201538,
      "learning_rate": 0.000564961949824013,
      "loss": 2.429,
      "step": 35800
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6346660852432251,
      "learning_rate": 0.0005649600313899205,
      "loss": 2.3691,
      "step": 35801
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6656777858734131,
      "learning_rate": 0.0005649581129065669,
      "loss": 2.1814,
      "step": 35802
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.600157618522644,
      "learning_rate": 0.0005649561943739526,
      "loss": 2.2178,
      "step": 35803
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6144561171531677,
      "learning_rate": 0.000564954275792078,
      "loss": 2.4077,
      "step": 35804
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6602174043655396,
      "learning_rate": 0.0005649523571609434,
      "loss": 2.2166,
      "step": 35805
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6776084899902344,
      "learning_rate": 0.0005649504384805491,
      "loss": 2.2359,
      "step": 35806
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6470098495483398,
      "learning_rate": 0.0005649485197508955,
      "loss": 2.362,
      "step": 35807
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6243839859962463,
      "learning_rate": 0.000564946600971983,
      "loss": 2.1632,
      "step": 35808
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6364357471466064,
      "learning_rate": 0.000564944682143812,
      "loss": 2.1949,
      "step": 35809
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6592386364936829,
      "learning_rate": 0.0005649427632663827,
      "loss": 2.1683,
      "step": 35810
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.760293185710907,
      "learning_rate": 0.0005649408443396956,
      "loss": 2.3163,
      "step": 35811
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6738857626914978,
      "learning_rate": 0.0005649389253637509,
      "loss": 2.4053,
      "step": 35812
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.646084725856781,
      "learning_rate": 0.0005649370063385491,
      "loss": 2.58,
      "step": 35813
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.7168493866920471,
      "learning_rate": 0.0005649350872640905,
      "loss": 2.2855,
      "step": 35814
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6462148427963257,
      "learning_rate": 0.0005649331681403755,
      "loss": 2.2228,
      "step": 35815
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6780037879943848,
      "learning_rate": 0.0005649312489674044,
      "loss": 2.3375,
      "step": 35816
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.7033398151397705,
      "learning_rate": 0.0005649293297451776,
      "loss": 2.2193,
      "step": 35817
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6444820165634155,
      "learning_rate": 0.0005649274104736954,
      "loss": 2.2245,
      "step": 35818
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6408208608627319,
      "learning_rate": 0.0005649254911529582,
      "loss": 2.3309,
      "step": 35819
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6884050965309143,
      "learning_rate": 0.0005649235717829664,
      "loss": 2.4921,
      "step": 35820
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6441910266876221,
      "learning_rate": 0.0005649216523637202,
      "loss": 2.1958,
      "step": 35821
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6754339933395386,
      "learning_rate": 0.0005649197328952202,
      "loss": 2.2869,
      "step": 35822
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.7040720582008362,
      "learning_rate": 0.0005649178133774665,
      "loss": 2.339,
      "step": 35823
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.7064585089683533,
      "learning_rate": 0.0005649158938104597,
      "loss": 2.4788,
      "step": 35824
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.7525694966316223,
      "learning_rate": 0.0005649139741941999,
      "loss": 2.2743,
      "step": 35825
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6668149828910828,
      "learning_rate": 0.0005649120545286878,
      "loss": 2.4597,
      "step": 35826
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6488742232322693,
      "learning_rate": 0.0005649101348139233,
      "loss": 2.4522,
      "step": 35827
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6955538988113403,
      "learning_rate": 0.0005649082150499071,
      "loss": 2.0497,
      "step": 35828
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6901032328605652,
      "learning_rate": 0.0005649062952366395,
      "loss": 2.0684,
      "step": 35829
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6821514964103699,
      "learning_rate": 0.0005649043753741209,
      "loss": 2.3813,
      "step": 35830
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6900049448013306,
      "learning_rate": 0.0005649024554623514,
      "loss": 2.4992,
      "step": 35831
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6401146650314331,
      "learning_rate": 0.0005649005355013316,
      "loss": 2.4916,
      "step": 35832
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.7064625024795532,
      "learning_rate": 0.0005648986154910617,
      "loss": 2.6066,
      "step": 35833
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6793861985206604,
      "learning_rate": 0.0005648966954315423,
      "loss": 2.3737,
      "step": 35834
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6705800890922546,
      "learning_rate": 0.0005648947753227735,
      "loss": 2.1279,
      "step": 35835
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.685820460319519,
      "learning_rate": 0.0005648928551647558,
      "loss": 2.3635,
      "step": 35836
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6250001788139343,
      "learning_rate": 0.0005648909349574895,
      "loss": 2.3529,
      "step": 35837
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.717165470123291,
      "learning_rate": 0.0005648890147009751,
      "loss": 2.5255,
      "step": 35838
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6281662583351135,
      "learning_rate": 0.0005648870943952127,
      "loss": 2.1902,
      "step": 35839
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.8671293258666992,
      "learning_rate": 0.0005648851740402028,
      "loss": 2.1647,
      "step": 35840
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.693925678730011,
      "learning_rate": 0.0005648832536359459,
      "loss": 2.1551,
      "step": 35841
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.7028467655181885,
      "learning_rate": 0.000564881333182442,
      "loss": 2.3373,
      "step": 35842
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.757808268070221,
      "learning_rate": 0.0005648794126796918,
      "loss": 2.1753,
      "step": 35843
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6161782145500183,
      "learning_rate": 0.0005648774921276953,
      "loss": 2.3711,
      "step": 35844
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.7286176681518555,
      "learning_rate": 0.0005648755715264533,
      "loss": 2.1304,
      "step": 35845
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.7017325162887573,
      "learning_rate": 0.0005648736508759658,
      "loss": 2.436,
      "step": 35846
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6217334866523743,
      "learning_rate": 0.0005648717301762334,
      "loss": 2.3404,
      "step": 35847
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6515896320343018,
      "learning_rate": 0.0005648698094272562,
      "loss": 2.4431,
      "step": 35848
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6413387060165405,
      "learning_rate": 0.0005648678886290348,
      "loss": 2.1564,
      "step": 35849
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6456472277641296,
      "learning_rate": 0.0005648659677815694,
      "loss": 2.2528,
      "step": 35850
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.7507211565971375,
      "learning_rate": 0.0005648640468848605,
      "loss": 2.0742,
      "step": 35851
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6815484762191772,
      "learning_rate": 0.0005648621259389084,
      "loss": 2.1435,
      "step": 35852
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.7292650938034058,
      "learning_rate": 0.0005648602049437133,
      "loss": 2.2585,
      "step": 35853
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.7191407084465027,
      "learning_rate": 0.0005648582838992757,
      "loss": 2.1891,
      "step": 35854
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6784477829933167,
      "learning_rate": 0.000564856362805596,
      "loss": 2.2206,
      "step": 35855
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.7032521367073059,
      "learning_rate": 0.0005648544416626746,
      "loss": 2.4599,
      "step": 35856
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.7128095030784607,
      "learning_rate": 0.0005648525204705117,
      "loss": 2.3104,
      "step": 35857
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6027387976646423,
      "learning_rate": 0.0005648505992291076,
      "loss": 2.1771,
      "step": 35858
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6431760787963867,
      "learning_rate": 0.0005648486779384629,
      "loss": 2.1002,
      "step": 35859
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.7692467570304871,
      "learning_rate": 0.0005648467565985778,
      "loss": 2.306,
      "step": 35860
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6363232731819153,
      "learning_rate": 0.0005648448352094526,
      "loss": 1.9588,
      "step": 35861
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6678155064582825,
      "learning_rate": 0.0005648429137710878,
      "loss": 2.2546,
      "step": 35862
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.8094498515129089,
      "learning_rate": 0.0005648409922834838,
      "loss": 2.4752,
      "step": 35863
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.7014102935791016,
      "learning_rate": 0.0005648390707466408,
      "loss": 2.3895,
      "step": 35864
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.7290606498718262,
      "learning_rate": 0.0005648371491605592,
      "loss": 2.2316,
      "step": 35865
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6830683350563049,
      "learning_rate": 0.0005648352275252394,
      "loss": 2.2293,
      "step": 35866
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6370179057121277,
      "learning_rate": 0.0005648333058406817,
      "loss": 2.4322,
      "step": 35867
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6930547952651978,
      "learning_rate": 0.0005648313841068865,
      "loss": 2.3086,
      "step": 35868
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6359462738037109,
      "learning_rate": 0.0005648294623238542,
      "loss": 2.2148,
      "step": 35869
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.7409873008728027,
      "learning_rate": 0.0005648275404915849,
      "loss": 2.3052,
      "step": 35870
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6142020225524902,
      "learning_rate": 0.0005648256186100794,
      "loss": 2.2168,
      "step": 35871
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6858701109886169,
      "learning_rate": 0.0005648236966793377,
      "loss": 2.4462,
      "step": 35872
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6565147042274475,
      "learning_rate": 0.0005648217746993604,
      "loss": 2.3049,
      "step": 35873
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6994454264640808,
      "learning_rate": 0.0005648198526701475,
      "loss": 2.3693,
      "step": 35874
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.7505038380622864,
      "learning_rate": 0.0005648179305916997,
      "loss": 2.3754,
      "step": 35875
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6679704785346985,
      "learning_rate": 0.0005648160084640175,
      "loss": 2.3344,
      "step": 35876
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.693250298500061,
      "learning_rate": 0.0005648140862871007,
      "loss": 2.1393,
      "step": 35877
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6751648783683777,
      "learning_rate": 0.00056481216406095,
      "loss": 2.3992,
      "step": 35878
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6332964301109314,
      "learning_rate": 0.0005648102417855658,
      "loss": 2.2259,
      "step": 35879
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.7177199125289917,
      "learning_rate": 0.0005648083194609483,
      "loss": 2.448,
      "step": 35880
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6486526727676392,
      "learning_rate": 0.000564806397087098,
      "loss": 2.18,
      "step": 35881
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6818844676017761,
      "learning_rate": 0.0005648044746640151,
      "loss": 2.0788,
      "step": 35882
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6560901999473572,
      "learning_rate": 0.0005648025521917002,
      "loss": 2.3598,
      "step": 35883
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.7167326211929321,
      "learning_rate": 0.0005648006296701534,
      "loss": 2.4707,
      "step": 35884
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6720175743103027,
      "learning_rate": 0.0005647987070993752,
      "loss": 2.4189,
      "step": 35885
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6233528852462769,
      "learning_rate": 0.0005647967844793659,
      "loss": 2.4558,
      "step": 35886
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6483690738677979,
      "learning_rate": 0.0005647948618101258,
      "loss": 2.5467,
      "step": 35887
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6817283034324646,
      "learning_rate": 0.0005647929390916555,
      "loss": 2.2963,
      "step": 35888
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.5914231538772583,
      "learning_rate": 0.0005647910163239551,
      "loss": 2.1505,
      "step": 35889
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6482473015785217,
      "learning_rate": 0.000564789093507025,
      "loss": 2.439,
      "step": 35890
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6175252795219421,
      "learning_rate": 0.0005647871706408659,
      "loss": 2.2791,
      "step": 35891
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.708148717880249,
      "learning_rate": 0.0005647852477254775,
      "loss": 2.3611,
      "step": 35892
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6261049509048462,
      "learning_rate": 0.0005647833247608607,
      "loss": 2.1621,
      "step": 35893
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6688054203987122,
      "learning_rate": 0.0005647814017470157,
      "loss": 2.1686,
      "step": 35894
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6893400549888611,
      "learning_rate": 0.0005647794786839428,
      "loss": 2.3536,
      "step": 35895
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6172738671302795,
      "learning_rate": 0.0005647775555716423,
      "loss": 2.1163,
      "step": 35896
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6717872023582458,
      "learning_rate": 0.0005647756324101148,
      "loss": 2.3009,
      "step": 35897
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6737233996391296,
      "learning_rate": 0.0005647737091993604,
      "loss": 2.6207,
      "step": 35898
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6317808628082275,
      "learning_rate": 0.0005647717859393797,
      "loss": 2.2275,
      "step": 35899
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.7183879017829895,
      "learning_rate": 0.0005647698626301728,
      "loss": 2.18,
      "step": 35900
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6964318156242371,
      "learning_rate": 0.0005647679392717402,
      "loss": 2.2581,
      "step": 35901
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6512233018875122,
      "learning_rate": 0.0005647660158640824,
      "loss": 2.2696,
      "step": 35902
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.7174254655838013,
      "learning_rate": 0.0005647640924071994,
      "loss": 2.1746,
      "step": 35903
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.692685067653656,
      "learning_rate": 0.0005647621689010919,
      "loss": 2.4698,
      "step": 35904
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.694755494594574,
      "learning_rate": 0.0005647602453457601,
      "loss": 2.33,
      "step": 35905
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.7553712129592896,
      "learning_rate": 0.0005647583217412043,
      "loss": 2.087,
      "step": 35906
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6896414160728455,
      "learning_rate": 0.0005647563980874249,
      "loss": 2.2217,
      "step": 35907
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6688714623451233,
      "learning_rate": 0.0005647544743844224,
      "loss": 2.2147,
      "step": 35908
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6655116081237793,
      "learning_rate": 0.0005647525506321969,
      "loss": 2.2841,
      "step": 35909
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6336759924888611,
      "learning_rate": 0.0005647506268307491,
      "loss": 2.2716,
      "step": 35910
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6403465867042542,
      "learning_rate": 0.000564748702980079,
      "loss": 2.1426,
      "step": 35911
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.5916748642921448,
      "learning_rate": 0.0005647467790801872,
      "loss": 2.1229,
      "step": 35912
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.7820551991462708,
      "learning_rate": 0.000564744855131074,
      "loss": 2.4778,
      "step": 35913
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6500037908554077,
      "learning_rate": 0.0005647429311327396,
      "loss": 2.4727,
      "step": 35914
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6565709710121155,
      "learning_rate": 0.0005647410070851847,
      "loss": 2.0337,
      "step": 35915
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.700839102268219,
      "learning_rate": 0.0005647390829884092,
      "loss": 2.3646,
      "step": 35916
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6341371536254883,
      "learning_rate": 0.0005647371588424139,
      "loss": 2.4466,
      "step": 35917
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6230559945106506,
      "learning_rate": 0.0005647352346471988,
      "loss": 2.3362,
      "step": 35918
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6408242583274841,
      "learning_rate": 0.0005647333104027646,
      "loss": 2.1794,
      "step": 35919
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6683906316757202,
      "learning_rate": 0.0005647313861091114,
      "loss": 2.1628,
      "step": 35920
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.677190363407135,
      "learning_rate": 0.0005647294617662396,
      "loss": 2.5359,
      "step": 35921
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.8229984045028687,
      "learning_rate": 0.0005647275373741496,
      "loss": 2.3419,
      "step": 35922
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6732827425003052,
      "learning_rate": 0.0005647256129328418,
      "loss": 2.3533,
      "step": 35923
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.7560814023017883,
      "learning_rate": 0.0005647236884423165,
      "loss": 2.2088,
      "step": 35924
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.8091729283332825,
      "learning_rate": 0.0005647217639025741,
      "loss": 2.3661,
      "step": 35925
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6657609939575195,
      "learning_rate": 0.0005647198393136149,
      "loss": 2.1605,
      "step": 35926
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.8221121430397034,
      "learning_rate": 0.0005647179146754393,
      "loss": 2.2787,
      "step": 35927
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.622741162776947,
      "learning_rate": 0.0005647159899880476,
      "loss": 2.4521,
      "step": 35928
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6785603761672974,
      "learning_rate": 0.0005647140652514402,
      "loss": 2.2264,
      "step": 35929
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.9156689047813416,
      "learning_rate": 0.0005647121404656175,
      "loss": 2.2337,
      "step": 35930
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.7222716212272644,
      "learning_rate": 0.0005647102156305799,
      "loss": 2.3957,
      "step": 35931
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.7032995223999023,
      "learning_rate": 0.0005647082907463275,
      "loss": 2.2269,
      "step": 35932
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.7763568758964539,
      "learning_rate": 0.0005647063658128609,
      "loss": 2.3161,
      "step": 35933
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.7309248447418213,
      "learning_rate": 0.0005647044408301804,
      "loss": 2.5019,
      "step": 35934
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6724643707275391,
      "learning_rate": 0.0005647025157982863,
      "loss": 2.228,
      "step": 35935
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.7681671380996704,
      "learning_rate": 0.0005647005907171791,
      "loss": 2.4372,
      "step": 35936
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.7870886325836182,
      "learning_rate": 0.0005646986655868589,
      "loss": 2.3065,
      "step": 35937
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6439613103866577,
      "learning_rate": 0.0005646967404073264,
      "loss": 2.1497,
      "step": 35938
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.688596785068512,
      "learning_rate": 0.0005646948151785818,
      "loss": 2.2508,
      "step": 35939
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6658830046653748,
      "learning_rate": 0.0005646928899006253,
      "loss": 2.3472,
      "step": 35940
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6475558280944824,
      "learning_rate": 0.0005646909645734574,
      "loss": 2.2745,
      "step": 35941
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6392848491668701,
      "learning_rate": 0.0005646890391970786,
      "loss": 2.055,
      "step": 35942
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.7784947156906128,
      "learning_rate": 0.0005646871137714889,
      "loss": 2.4075,
      "step": 35943
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.7365067005157471,
      "learning_rate": 0.000564685188296689,
      "loss": 2.2568,
      "step": 35944
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6650791168212891,
      "learning_rate": 0.0005646832627726791,
      "loss": 2.5644,
      "step": 35945
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.8138774037361145,
      "learning_rate": 0.0005646813371994596,
      "loss": 2.2417,
      "step": 35946
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6088520288467407,
      "learning_rate": 0.0005646794115770308,
      "loss": 2.1851,
      "step": 35947
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.7741196751594543,
      "learning_rate": 0.0005646774859053933,
      "loss": 2.2639,
      "step": 35948
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.664291501045227,
      "learning_rate": 0.0005646755601845469,
      "loss": 2.2709,
      "step": 35949
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6574741005897522,
      "learning_rate": 0.0005646736344144926,
      "loss": 2.3167,
      "step": 35950
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6341136693954468,
      "learning_rate": 0.0005646717085952304,
      "loss": 2.2367,
      "step": 35951
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.7352648377418518,
      "learning_rate": 0.0005646697827267607,
      "loss": 2.3585,
      "step": 35952
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.7121124863624573,
      "learning_rate": 0.0005646678568090839,
      "loss": 2.2218,
      "step": 35953
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6650214195251465,
      "learning_rate": 0.0005646659308422004,
      "loss": 2.2654,
      "step": 35954
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6514323949813843,
      "learning_rate": 0.0005646640048261104,
      "loss": 2.3427,
      "step": 35955
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6647486090660095,
      "learning_rate": 0.0005646620787608144,
      "loss": 2.1841,
      "step": 35956
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.7013094425201416,
      "learning_rate": 0.0005646601526463128,
      "loss": 2.311,
      "step": 35957
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.674685537815094,
      "learning_rate": 0.0005646582264826058,
      "loss": 2.4519,
      "step": 35958
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6865732669830322,
      "learning_rate": 0.000564656300269694,
      "loss": 1.9626,
      "step": 35959
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.700458824634552,
      "learning_rate": 0.0005646543740075773,
      "loss": 2.3734,
      "step": 35960
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6547806262969971,
      "learning_rate": 0.0005646524476962567,
      "loss": 2.3276,
      "step": 35961
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.8235174417495728,
      "learning_rate": 0.0005646505213357319,
      "loss": 2.214,
      "step": 35962
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6719157695770264,
      "learning_rate": 0.0005646485949260038,
      "loss": 2.1958,
      "step": 35963
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.671069860458374,
      "learning_rate": 0.0005646466684670725,
      "loss": 2.4435,
      "step": 35964
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.7835944890975952,
      "learning_rate": 0.0005646447419589383,
      "loss": 2.3463,
      "step": 35965
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.641567051410675,
      "learning_rate": 0.0005646428154016017,
      "loss": 2.1969,
      "step": 35966
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6993581056594849,
      "learning_rate": 0.000564640888795063,
      "loss": 2.2698,
      "step": 35967
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6909701824188232,
      "learning_rate": 0.0005646389621393225,
      "loss": 2.2609,
      "step": 35968
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.7630739808082581,
      "learning_rate": 0.0005646370354343807,
      "loss": 2.288,
      "step": 35969
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.7109383940696716,
      "learning_rate": 0.0005646351086802378,
      "loss": 2.244,
      "step": 35970
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.7871822118759155,
      "learning_rate": 0.0005646331818768944,
      "loss": 2.2818,
      "step": 35971
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.9146016240119934,
      "learning_rate": 0.0005646312550243506,
      "loss": 2.2982,
      "step": 35972
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.660540759563446,
      "learning_rate": 0.0005646293281226069,
      "loss": 2.2875,
      "step": 35973
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6940610408782959,
      "learning_rate": 0.0005646274011716635,
      "loss": 2.1911,
      "step": 35974
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.7038997411727905,
      "learning_rate": 0.0005646254741715209,
      "loss": 2.2978,
      "step": 35975
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.72770094871521,
      "learning_rate": 0.0005646235471221795,
      "loss": 2.4097,
      "step": 35976
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6715110540390015,
      "learning_rate": 0.0005646216200236395,
      "loss": 2.5544,
      "step": 35977
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.7471268773078918,
      "learning_rate": 0.0005646196928759014,
      "loss": 2.3258,
      "step": 35978
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6611287593841553,
      "learning_rate": 0.0005646177656789655,
      "loss": 2.2468,
      "step": 35979
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6671069264411926,
      "learning_rate": 0.0005646158384328321,
      "loss": 2.2288,
      "step": 35980
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.7840360403060913,
      "learning_rate": 0.0005646139111375018,
      "loss": 2.2615,
      "step": 35981
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6674193143844604,
      "learning_rate": 0.0005646119837929746,
      "loss": 2.3706,
      "step": 35982
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6300215125083923,
      "learning_rate": 0.000564610056399251,
      "loss": 2.2218,
      "step": 35983
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6592704653739929,
      "learning_rate": 0.0005646081289563316,
      "loss": 2.2694,
      "step": 35984
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.669833779335022,
      "learning_rate": 0.0005646062014642165,
      "loss": 2.1776,
      "step": 35985
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.7470497488975525,
      "learning_rate": 0.000564604273922906,
      "loss": 2.1874,
      "step": 35986
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6679551005363464,
      "learning_rate": 0.0005646023463324006,
      "loss": 2.1888,
      "step": 35987
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.772423505783081,
      "learning_rate": 0.0005646004186927006,
      "loss": 2.1915,
      "step": 35988
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.662077009677887,
      "learning_rate": 0.0005645984910038065,
      "loss": 2.3835,
      "step": 35989
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6657321453094482,
      "learning_rate": 0.0005645965632657185,
      "loss": 2.2911,
      "step": 35990
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.8701056838035583,
      "learning_rate": 0.000564594635478437,
      "loss": 2.132,
      "step": 35991
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6542301774024963,
      "learning_rate": 0.0005645927076419623,
      "loss": 2.4845,
      "step": 35992
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.8042254447937012,
      "learning_rate": 0.0005645907797562949,
      "loss": 2.0454,
      "step": 35993
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.8078737258911133,
      "learning_rate": 0.000564588851821435,
      "loss": 2.356,
      "step": 35994
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6458622813224792,
      "learning_rate": 0.0005645869238373832,
      "loss": 2.088,
      "step": 35995
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.7735845446586609,
      "learning_rate": 0.0005645849958041396,
      "loss": 2.2127,
      "step": 35996
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6917937397956848,
      "learning_rate": 0.0005645830677217046,
      "loss": 2.4865,
      "step": 35997
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6924459934234619,
      "learning_rate": 0.0005645811395900787,
      "loss": 2.0331,
      "step": 35998
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.709302544593811,
      "learning_rate": 0.0005645792114092621,
      "loss": 2.4109,
      "step": 35999
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6760686039924622,
      "learning_rate": 0.0005645772831792552,
      "loss": 2.4927,
      "step": 36000
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.7502350211143494,
      "learning_rate": 0.0005645753549000584,
      "loss": 2.151,
      "step": 36001
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6909816265106201,
      "learning_rate": 0.0005645734265716722,
      "loss": 2.2568,
      "step": 36002
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.8059422373771667,
      "learning_rate": 0.0005645714981940967,
      "loss": 2.4791,
      "step": 36003
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6791498064994812,
      "learning_rate": 0.0005645695697673324,
      "loss": 2.0956,
      "step": 36004
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.5967602729797363,
      "learning_rate": 0.0005645676412913795,
      "loss": 2.4429,
      "step": 36005
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.8744308948516846,
      "learning_rate": 0.0005645657127662386,
      "loss": 2.405,
      "step": 36006
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.7476184964179993,
      "learning_rate": 0.00056456378419191,
      "loss": 2.312,
      "step": 36007
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6932419538497925,
      "learning_rate": 0.0005645618555683939,
      "loss": 2.4444,
      "step": 36008
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.7618317008018494,
      "learning_rate": 0.0005645599268956907,
      "loss": 2.2881,
      "step": 36009
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.7303771376609802,
      "learning_rate": 0.0005645579981738009,
      "loss": 2.3069,
      "step": 36010
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.7973247170448303,
      "learning_rate": 0.0005645560694027247,
      "loss": 2.5223,
      "step": 36011
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.7112969756126404,
      "learning_rate": 0.0005645541405824627,
      "loss": 2.2834,
      "step": 36012
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6465855240821838,
      "learning_rate": 0.0005645522117130149,
      "loss": 2.4119,
      "step": 36013
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6519352793693542,
      "learning_rate": 0.000564550282794382,
      "loss": 2.2664,
      "step": 36014
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6627589464187622,
      "learning_rate": 0.0005645483538265641,
      "loss": 2.3598,
      "step": 36015
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6539306044578552,
      "learning_rate": 0.0005645464248095617,
      "loss": 2.4947,
      "step": 36016
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6438047885894775,
      "learning_rate": 0.0005645444957433751,
      "loss": 2.0069,
      "step": 36017
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6473283767700195,
      "learning_rate": 0.0005645425666280048,
      "loss": 2.0977,
      "step": 36018
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6747720837593079,
      "learning_rate": 0.0005645406374634509,
      "loss": 2.217,
      "step": 36019
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.616256594657898,
      "learning_rate": 0.0005645387082497139,
      "loss": 2.4315,
      "step": 36020
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6458840370178223,
      "learning_rate": 0.0005645367789867943,
      "loss": 2.2822,
      "step": 36021
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.669643759727478,
      "learning_rate": 0.0005645348496746923,
      "loss": 2.1714,
      "step": 36022
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6126983165740967,
      "learning_rate": 0.0005645329203134081,
      "loss": 2.3717,
      "step": 36023
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.7167374491691589,
      "learning_rate": 0.0005645309909029424,
      "loss": 2.4682,
      "step": 36024
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6291319727897644,
      "learning_rate": 0.0005645290614432954,
      "loss": 2.5393,
      "step": 36025
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6561164259910583,
      "learning_rate": 0.0005645271319344674,
      "loss": 2.4737,
      "step": 36026
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6992452144622803,
      "learning_rate": 0.0005645252023764588,
      "loss": 2.2889,
      "step": 36027
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6587750315666199,
      "learning_rate": 0.00056452327276927,
      "loss": 2.2944,
      "step": 36028
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6836420297622681,
      "learning_rate": 0.0005645213431129013,
      "loss": 2.4069,
      "step": 36029
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6309877038002014,
      "learning_rate": 0.0005645194134073531,
      "loss": 2.2485,
      "step": 36030
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.7165863513946533,
      "learning_rate": 0.0005645174836526258,
      "loss": 2.2601,
      "step": 36031
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6659544706344604,
      "learning_rate": 0.0005645155538487196,
      "loss": 2.2079,
      "step": 36032
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6521768569946289,
      "learning_rate": 0.000564513623995635,
      "loss": 2.289,
      "step": 36033
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.7531570792198181,
      "learning_rate": 0.0005645116940933725,
      "loss": 2.4103,
      "step": 36034
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.678866982460022,
      "learning_rate": 0.0005645097641419321,
      "loss": 2.1179,
      "step": 36035
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6365140080451965,
      "learning_rate": 0.0005645078341413143,
      "loss": 2.3979,
      "step": 36036
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.702393114566803,
      "learning_rate": 0.0005645059040915196,
      "loss": 2.2607,
      "step": 36037
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6574251651763916,
      "learning_rate": 0.0005645039739925482,
      "loss": 2.2924,
      "step": 36038
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6757113933563232,
      "learning_rate": 0.0005645020438444007,
      "loss": 2.2179,
      "step": 36039
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6468434929847717,
      "learning_rate": 0.0005645001136470771,
      "loss": 2.1861,
      "step": 36040
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.653869092464447,
      "learning_rate": 0.0005644981834005779,
      "loss": 2.306,
      "step": 36041
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.7432371973991394,
      "learning_rate": 0.0005644962531049035,
      "loss": 2.3159,
      "step": 36042
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6162365674972534,
      "learning_rate": 0.0005644943227600544,
      "loss": 2.357,
      "step": 36043
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.7134935855865479,
      "learning_rate": 0.0005644923923660307,
      "loss": 2.4265,
      "step": 36044
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.7191276550292969,
      "learning_rate": 0.000564490461922833,
      "loss": 2.4077,
      "step": 36045
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6957350969314575,
      "learning_rate": 0.0005644885314304613,
      "loss": 2.2127,
      "step": 36046
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6263139247894287,
      "learning_rate": 0.0005644866008889164,
      "loss": 2.2019,
      "step": 36047
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6254746913909912,
      "learning_rate": 0.0005644846702981983,
      "loss": 2.3455,
      "step": 36048
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6598895788192749,
      "learning_rate": 0.0005644827396583076,
      "loss": 2.1219,
      "step": 36049
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.7003505825996399,
      "learning_rate": 0.0005644808089692446,
      "loss": 2.2696,
      "step": 36050
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6300163865089417,
      "learning_rate": 0.0005644788782310096,
      "loss": 2.31,
      "step": 36051
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6607488393783569,
      "learning_rate": 0.0005644769474436029,
      "loss": 2.2992,
      "step": 36052
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6619144082069397,
      "learning_rate": 0.000564475016607025,
      "loss": 2.4256,
      "step": 36053
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6093520522117615,
      "learning_rate": 0.0005644730857212761,
      "loss": 2.182,
      "step": 36054
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6776993870735168,
      "learning_rate": 0.0005644711547863569,
      "loss": 2.0371,
      "step": 36055
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.7127712965011597,
      "learning_rate": 0.0005644692238022674,
      "loss": 2.3478,
      "step": 36056
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.620614767074585,
      "learning_rate": 0.0005644672927690081,
      "loss": 2.2714,
      "step": 36057
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.653486430644989,
      "learning_rate": 0.0005644653616865792,
      "loss": 2.2681,
      "step": 36058
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6910120248794556,
      "learning_rate": 0.0005644634305549813,
      "loss": 2.1895,
      "step": 36059
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6447100043296814,
      "learning_rate": 0.0005644614993742147,
      "loss": 2.1989,
      "step": 36060
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.7680718898773193,
      "learning_rate": 0.0005644595681442796,
      "loss": 2.3041,
      "step": 36061
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.7067611217498779,
      "learning_rate": 0.0005644576368651766,
      "loss": 2.526,
      "step": 36062
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6259340643882751,
      "learning_rate": 0.0005644557055369058,
      "loss": 2.1103,
      "step": 36063
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6407124996185303,
      "learning_rate": 0.0005644537741594678,
      "loss": 2.3331,
      "step": 36064
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6350987553596497,
      "learning_rate": 0.0005644518427328628,
      "loss": 2.2644,
      "step": 36065
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6991509795188904,
      "learning_rate": 0.0005644499112570912,
      "loss": 2.3702,
      "step": 36066
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.701014518737793,
      "learning_rate": 0.0005644479797321534,
      "loss": 2.2416,
      "step": 36067
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6493358612060547,
      "learning_rate": 0.0005644460481580497,
      "loss": 2.3236,
      "step": 36068
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6239422559738159,
      "learning_rate": 0.0005644441165347806,
      "loss": 2.2735,
      "step": 36069
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6666993498802185,
      "learning_rate": 0.0005644421848623462,
      "loss": 2.2446,
      "step": 36070
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6724241971969604,
      "learning_rate": 0.0005644402531407471,
      "loss": 2.2506,
      "step": 36071
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6579481959342957,
      "learning_rate": 0.0005644383213699836,
      "loss": 2.035,
      "step": 36072
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.7362481355667114,
      "learning_rate": 0.0005644363895500559,
      "loss": 2.1799,
      "step": 36073
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6871358156204224,
      "learning_rate": 0.0005644344576809646,
      "loss": 2.0733,
      "step": 36074
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.7435281276702881,
      "learning_rate": 0.0005644325257627099,
      "loss": 2.2253,
      "step": 36075
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.7095939517021179,
      "learning_rate": 0.0005644305937952921,
      "loss": 2.4298,
      "step": 36076
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.676878035068512,
      "learning_rate": 0.0005644286617787118,
      "loss": 2.3624,
      "step": 36077
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6843816041946411,
      "learning_rate": 0.0005644267297129691,
      "loss": 2.23,
      "step": 36078
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6316015720367432,
      "learning_rate": 0.0005644247975980645,
      "loss": 2.3314,
      "step": 36079
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6928966641426086,
      "learning_rate": 0.0005644228654339985,
      "loss": 2.4173,
      "step": 36080
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6964516043663025,
      "learning_rate": 0.0005644209332207712,
      "loss": 2.3768,
      "step": 36081
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6536906361579895,
      "learning_rate": 0.0005644190009583829,
      "loss": 2.2732,
      "step": 36082
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.734857439994812,
      "learning_rate": 0.0005644170686468343,
      "loss": 2.3849,
      "step": 36083
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.7231912016868591,
      "learning_rate": 0.0005644151362861256,
      "loss": 2.3299,
      "step": 36084
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6616020798683167,
      "learning_rate": 0.0005644132038762571,
      "loss": 2.2838,
      "step": 36085
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.7323097586631775,
      "learning_rate": 0.0005644112714172291,
      "loss": 2.5648,
      "step": 36086
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.637708842754364,
      "learning_rate": 0.0005644093389090421,
      "loss": 2.0323,
      "step": 36087
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.7251572608947754,
      "learning_rate": 0.0005644074063516965,
      "loss": 2.4278,
      "step": 36088
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.869871199131012,
      "learning_rate": 0.0005644054737451924,
      "loss": 2.2839,
      "step": 36089
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6496292352676392,
      "learning_rate": 0.0005644035410895305,
      "loss": 2.2253,
      "step": 36090
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6766074895858765,
      "learning_rate": 0.0005644016083847108,
      "loss": 2.3187,
      "step": 36091
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6410994529724121,
      "learning_rate": 0.000564399675630734,
      "loss": 2.3473,
      "step": 36092
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.7843943238258362,
      "learning_rate": 0.0005643977428276003,
      "loss": 2.2459,
      "step": 36093
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6839876770973206,
      "learning_rate": 0.0005643958099753101,
      "loss": 2.249,
      "step": 36094
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.645820677280426,
      "learning_rate": 0.0005643938770738637,
      "loss": 2.3211,
      "step": 36095
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6629029512405396,
      "learning_rate": 0.0005643919441232615,
      "loss": 2.3178,
      "step": 36096
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6164912581443787,
      "learning_rate": 0.0005643900111235038,
      "loss": 2.279,
      "step": 36097
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6722131371498108,
      "learning_rate": 0.000564388078074591,
      "loss": 2.5652,
      "step": 36098
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6873205900192261,
      "learning_rate": 0.0005643861449765235,
      "loss": 2.2801,
      "step": 36099
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6718103289604187,
      "learning_rate": 0.0005643842118293016,
      "loss": 2.4102,
      "step": 36100
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6648507714271545,
      "learning_rate": 0.0005643822786329256,
      "loss": 2.2399,
      "step": 36101
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6439846158027649,
      "learning_rate": 0.000564380345387396,
      "loss": 2.1326,
      "step": 36102
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.5959927439689636,
      "learning_rate": 0.0005643784120927133,
      "loss": 2.4568,
      "step": 36103
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6842124462127686,
      "learning_rate": 0.0005643764787488775,
      "loss": 2.1878,
      "step": 36104
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6498392820358276,
      "learning_rate": 0.0005643745453558891,
      "loss": 2.2855,
      "step": 36105
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.7014263272285461,
      "learning_rate": 0.0005643726119137485,
      "loss": 2.1588,
      "step": 36106
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.713904619216919,
      "learning_rate": 0.000564370678422456,
      "loss": 2.1975,
      "step": 36107
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6376514434814453,
      "learning_rate": 0.0005643687448820121,
      "loss": 2.3041,
      "step": 36108
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6748192310333252,
      "learning_rate": 0.000564366811292417,
      "loss": 2.2044,
      "step": 36109
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6410304307937622,
      "learning_rate": 0.0005643648776536711,
      "loss": 2.4584,
      "step": 36110
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.7105172872543335,
      "learning_rate": 0.0005643629439657748,
      "loss": 2.4177,
      "step": 36111
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6625507473945618,
      "learning_rate": 0.0005643610102287284,
      "loss": 2.1361,
      "step": 36112
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.7025267481803894,
      "learning_rate": 0.0005643590764425325,
      "loss": 2.3567,
      "step": 36113
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6645371317863464,
      "learning_rate": 0.0005643571426071869,
      "loss": 2.4995,
      "step": 36114
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6218323707580566,
      "learning_rate": 0.0005643552087226926,
      "loss": 2.4094,
      "step": 36115
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.67085200548172,
      "learning_rate": 0.0005643532747890497,
      "loss": 2.5168,
      "step": 36116
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6551882028579712,
      "learning_rate": 0.0005643513408062584,
      "loss": 2.2792,
      "step": 36117
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6736687421798706,
      "learning_rate": 0.0005643494067743192,
      "loss": 2.3217,
      "step": 36118
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6628764867782593,
      "learning_rate": 0.0005643474726932325,
      "loss": 2.412,
      "step": 36119
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.7595902681350708,
      "learning_rate": 0.0005643455385629985,
      "loss": 2.4231,
      "step": 36120
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.725042462348938,
      "learning_rate": 0.0005643436043836178,
      "loss": 2.2071,
      "step": 36121
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6071219444274902,
      "learning_rate": 0.0005643416701550906,
      "loss": 2.1242,
      "step": 36122
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6696736216545105,
      "learning_rate": 0.0005643397358774172,
      "loss": 2.1077,
      "step": 36123
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6780552864074707,
      "learning_rate": 0.0005643378015505983,
      "loss": 2.3091,
      "step": 36124
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6368656158447266,
      "learning_rate": 0.0005643358671746339,
      "loss": 2.3021,
      "step": 36125
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.623878002166748,
      "learning_rate": 0.0005643339327495244,
      "loss": 2.1701,
      "step": 36126
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6982681751251221,
      "learning_rate": 0.0005643319982752702,
      "loss": 2.2239,
      "step": 36127
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6595891118049622,
      "learning_rate": 0.0005643300637518718,
      "loss": 2.4461,
      "step": 36128
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6566624641418457,
      "learning_rate": 0.0005643281291793295,
      "loss": 2.2092,
      "step": 36129
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6720011830329895,
      "learning_rate": 0.0005643261945576434,
      "loss": 2.1877,
      "step": 36130
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6622061729431152,
      "learning_rate": 0.0005643242598868143,
      "loss": 2.6848,
      "step": 36131
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6552825570106506,
      "learning_rate": 0.0005643223251668422,
      "loss": 2.3139,
      "step": 36132
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6476895809173584,
      "learning_rate": 0.0005643203903977276,
      "loss": 2.1817,
      "step": 36133
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.7776474952697754,
      "learning_rate": 0.0005643184555794708,
      "loss": 2.0262,
      "step": 36134
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6957667469978333,
      "learning_rate": 0.0005643165207120723,
      "loss": 2.1526,
      "step": 36135
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.741262674331665,
      "learning_rate": 0.0005643145857955323,
      "loss": 2.4271,
      "step": 36136
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6298760175704956,
      "learning_rate": 0.0005643126508298514,
      "loss": 2.3909,
      "step": 36137
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.7586355805397034,
      "learning_rate": 0.0005643107158150296,
      "loss": 2.3806,
      "step": 36138
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6771268248558044,
      "learning_rate": 0.0005643087807510675,
      "loss": 2.1064,
      "step": 36139
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.7278632521629333,
      "learning_rate": 0.0005643068456379655,
      "loss": 2.2725,
      "step": 36140
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6857710480690002,
      "learning_rate": 0.0005643049104757237,
      "loss": 2.3991,
      "step": 36141
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.7032364010810852,
      "learning_rate": 0.0005643029752643427,
      "loss": 2.1107,
      "step": 36142
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.8096326589584351,
      "learning_rate": 0.0005643010400038229,
      "loss": 2.3643,
      "step": 36143
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6430250406265259,
      "learning_rate": 0.0005642991046941645,
      "loss": 2.1634,
      "step": 36144
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.7422122955322266,
      "learning_rate": 0.0005642971693353677,
      "loss": 2.0753,
      "step": 36145
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6568222641944885,
      "learning_rate": 0.0005642952339274332,
      "loss": 2.3488,
      "step": 36146
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6796136498451233,
      "learning_rate": 0.0005642932984703614,
      "loss": 2.2735,
      "step": 36147
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6749089956283569,
      "learning_rate": 0.0005642913629641523,
      "loss": 2.2248,
      "step": 36148
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.7643105387687683,
      "learning_rate": 0.0005642894274088064,
      "loss": 2.2963,
      "step": 36149
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.690632164478302,
      "learning_rate": 0.0005642874918043243,
      "loss": 2.3629,
      "step": 36150
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6803138256072998,
      "learning_rate": 0.000564285556150706,
      "loss": 2.1595,
      "step": 36151
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6783117651939392,
      "learning_rate": 0.0005642836204479521,
      "loss": 2.3886,
      "step": 36152
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6310924887657166,
      "learning_rate": 0.0005642816846960629,
      "loss": 2.2353,
      "step": 36153
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.7687686681747437,
      "learning_rate": 0.0005642797488950386,
      "loss": 2.1331,
      "step": 36154
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.642978310585022,
      "learning_rate": 0.00056427781304488,
      "loss": 2.2713,
      "step": 36155
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.7284579277038574,
      "learning_rate": 0.000564275877145587,
      "loss": 2.316,
      "step": 36156
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.663604736328125,
      "learning_rate": 0.00056427394119716,
      "loss": 2.3843,
      "step": 36157
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.7027704119682312,
      "learning_rate": 0.0005642720051995996,
      "loss": 2.1199,
      "step": 36158
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.7081993222236633,
      "learning_rate": 0.0005642700691529061,
      "loss": 2.3605,
      "step": 36159
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6525512337684631,
      "learning_rate": 0.0005642681330570797,
      "loss": 2.2967,
      "step": 36160
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6677007079124451,
      "learning_rate": 0.000564266196912121,
      "loss": 2.4252,
      "step": 36161
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.729825496673584,
      "learning_rate": 0.0005642642607180301,
      "loss": 2.5443,
      "step": 36162
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6498260498046875,
      "learning_rate": 0.0005642623244748075,
      "loss": 2.2196,
      "step": 36163
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6879275441169739,
      "learning_rate": 0.0005642603881824537,
      "loss": 2.3054,
      "step": 36164
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6662365198135376,
      "learning_rate": 0.0005642584518409687,
      "loss": 2.1661,
      "step": 36165
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6883111000061035,
      "learning_rate": 0.0005642565154503531,
      "loss": 2.3482,
      "step": 36166
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6850963234901428,
      "learning_rate": 0.0005642545790106073,
      "loss": 2.2165,
      "step": 36167
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.7397286891937256,
      "learning_rate": 0.0005642526425217316,
      "loss": 2.1626,
      "step": 36168
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.7078873515129089,
      "learning_rate": 0.0005642507059837262,
      "loss": 2.2171,
      "step": 36169
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6336501240730286,
      "learning_rate": 0.0005642487693965917,
      "loss": 2.2007,
      "step": 36170
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.8293735384941101,
      "learning_rate": 0.0005642468327603285,
      "loss": 2.1025,
      "step": 36171
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6936759352684021,
      "learning_rate": 0.0005642448960749367,
      "loss": 2.3071,
      "step": 36172
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.658585250377655,
      "learning_rate": 0.0005642429593404167,
      "loss": 2.3595,
      "step": 36173
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.7572301030158997,
      "learning_rate": 0.0005642410225567691,
      "loss": 2.2499,
      "step": 36174
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6509605646133423,
      "learning_rate": 0.0005642390857239941,
      "loss": 2.2525,
      "step": 36175
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.693880021572113,
      "learning_rate": 0.000564237148842092,
      "loss": 2.3162,
      "step": 36176
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.7448678612709045,
      "learning_rate": 0.0005642352119110632,
      "loss": 2.1922,
      "step": 36177
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.8288370966911316,
      "learning_rate": 0.000564233274930908,
      "loss": 2.3719,
      "step": 36178
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6590167284011841,
      "learning_rate": 0.000564231337901627,
      "loss": 2.217,
      "step": 36179
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.7434893846511841,
      "learning_rate": 0.0005642294008232204,
      "loss": 2.2172,
      "step": 36180
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6657791137695312,
      "learning_rate": 0.0005642274636956886,
      "loss": 2.5455,
      "step": 36181
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.5995309352874756,
      "learning_rate": 0.0005642255265190317,
      "loss": 2.265,
      "step": 36182
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.7664969563484192,
      "learning_rate": 0.0005642235892932505,
      "loss": 2.3459,
      "step": 36183
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6533293128013611,
      "learning_rate": 0.0005642216520183452,
      "loss": 2.2839,
      "step": 36184
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6774500012397766,
      "learning_rate": 0.000564219714694316,
      "loss": 2.2999,
      "step": 36185
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6462364792823792,
      "learning_rate": 0.0005642177773211633,
      "loss": 2.2602,
      "step": 36186
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.7261425852775574,
      "learning_rate": 0.0005642158398988876,
      "loss": 2.4146,
      "step": 36187
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6294736266136169,
      "learning_rate": 0.0005642139024274892,
      "loss": 2.3458,
      "step": 36188
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6775413751602173,
      "learning_rate": 0.0005642119649069685,
      "loss": 2.3979,
      "step": 36189
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6476636528968811,
      "learning_rate": 0.0005642100273373256,
      "loss": 2.2856,
      "step": 36190
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6639556884765625,
      "learning_rate": 0.0005642080897185613,
      "loss": 2.1791,
      "step": 36191
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.7385218143463135,
      "learning_rate": 0.0005642061520506757,
      "loss": 2.1474,
      "step": 36192
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6200379133224487,
      "learning_rate": 0.0005642042143336691,
      "loss": 1.9936,
      "step": 36193
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6331365704536438,
      "learning_rate": 0.000564202276567542,
      "loss": 1.9952,
      "step": 36194
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.7426766753196716,
      "learning_rate": 0.0005642003387522946,
      "loss": 2.371,
      "step": 36195
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.8221641182899475,
      "learning_rate": 0.0005641984008879276,
      "loss": 2.3174,
      "step": 36196
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6610701084136963,
      "learning_rate": 0.0005641964629744409,
      "loss": 2.4131,
      "step": 36197
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.827855110168457,
      "learning_rate": 0.0005641945250118352,
      "loss": 2.2675,
      "step": 36198
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.7736599445343018,
      "learning_rate": 0.0005641925870001108,
      "loss": 2.2123,
      "step": 36199
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.7308722138404846,
      "learning_rate": 0.0005641906489392679,
      "loss": 2.1267,
      "step": 36200
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.5787455439567566,
      "learning_rate": 0.000564188710829307,
      "loss": 2.1797,
      "step": 36201
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.8622502684593201,
      "learning_rate": 0.0005641867726702284,
      "loss": 2.078,
      "step": 36202
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.7248939275741577,
      "learning_rate": 0.0005641848344620326,
      "loss": 2.4548,
      "step": 36203
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.7007642388343811,
      "learning_rate": 0.0005641828962047198,
      "loss": 2.2527,
      "step": 36204
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.7536633610725403,
      "learning_rate": 0.0005641809578982904,
      "loss": 2.2726,
      "step": 36205
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.0168124437332153,
      "learning_rate": 0.0005641790195427447,
      "loss": 2.4937,
      "step": 36206
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6707767248153687,
      "learning_rate": 0.0005641770811380832,
      "loss": 2.246,
      "step": 36207
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.772335410118103,
      "learning_rate": 0.0005641751426843062,
      "loss": 2.4507,
      "step": 36208
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.7510514259338379,
      "learning_rate": 0.0005641732041814141,
      "loss": 2.421,
      "step": 36209
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.8612430095672607,
      "learning_rate": 0.0005641712656294072,
      "loss": 2.3991,
      "step": 36210
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.659501850605011,
      "learning_rate": 0.0005641693270282859,
      "loss": 2.5484,
      "step": 36211
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.8353171348571777,
      "learning_rate": 0.0005641673883780504,
      "loss": 2.3205,
      "step": 36212
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.8743764758110046,
      "learning_rate": 0.0005641654496787013,
      "loss": 2.0058,
      "step": 36213
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6270079016685486,
      "learning_rate": 0.0005641635109302389,
      "loss": 2.1298,
      "step": 36214
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6759881377220154,
      "learning_rate": 0.0005641615721326634,
      "loss": 2.4892,
      "step": 36215
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.7633004784584045,
      "learning_rate": 0.0005641596332859753,
      "loss": 2.2485,
      "step": 36216
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.8534736633300781,
      "learning_rate": 0.0005641576943901749,
      "loss": 2.4145,
      "step": 36217
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6260151863098145,
      "learning_rate": 0.0005641557554452628,
      "loss": 2.2862,
      "step": 36218
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.7915287017822266,
      "learning_rate": 0.000564153816451239,
      "loss": 2.2939,
      "step": 36219
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.8977476954460144,
      "learning_rate": 0.000564151877408104,
      "loss": 2.2142,
      "step": 36220
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6305214166641235,
      "learning_rate": 0.0005641499383158582,
      "loss": 2.2571,
      "step": 36221
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6809085011482239,
      "learning_rate": 0.000564147999174502,
      "loss": 2.2921,
      "step": 36222
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.8165113925933838,
      "learning_rate": 0.0005641460599840357,
      "loss": 2.0701,
      "step": 36223
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.678900957107544,
      "learning_rate": 0.0005641441207444596,
      "loss": 2.2618,
      "step": 36224
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6395171880722046,
      "learning_rate": 0.0005641421814557741,
      "loss": 2.1671,
      "step": 36225
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6397145390510559,
      "learning_rate": 0.0005641402421179797,
      "loss": 2.137,
      "step": 36226
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.8038987517356873,
      "learning_rate": 0.0005641383027310765,
      "loss": 2.1824,
      "step": 36227
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.9207824468612671,
      "learning_rate": 0.0005641363632950651,
      "loss": 2.535,
      "step": 36228
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6654220223426819,
      "learning_rate": 0.0005641344238099458,
      "loss": 2.1348,
      "step": 36229
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6267595887184143,
      "learning_rate": 0.0005641324842757189,
      "loss": 2.2942,
      "step": 36230
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.9195178151130676,
      "learning_rate": 0.0005641305446923847,
      "loss": 2.1416,
      "step": 36231
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6969221830368042,
      "learning_rate": 0.0005641286050599437,
      "loss": 2.0804,
      "step": 36232
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6137305498123169,
      "learning_rate": 0.0005641266653783962,
      "loss": 2.1228,
      "step": 36233
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.7101507186889648,
      "learning_rate": 0.0005641247256477425,
      "loss": 2.3792,
      "step": 36234
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.7370818853378296,
      "learning_rate": 0.0005641227858679831,
      "loss": 2.1963,
      "step": 36235
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.649214506149292,
      "learning_rate": 0.0005641208460391183,
      "loss": 2.1393,
      "step": 36236
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6648110747337341,
      "learning_rate": 0.0005641189061611485,
      "loss": 2.3647,
      "step": 36237
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6692270636558533,
      "learning_rate": 0.0005641169662340739,
      "loss": 2.3121,
      "step": 36238
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.7381832003593445,
      "learning_rate": 0.000564115026257895,
      "loss": 2.0631,
      "step": 36239
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6327792406082153,
      "learning_rate": 0.0005641130862326121,
      "loss": 2.4082,
      "step": 36240
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.700498104095459,
      "learning_rate": 0.0005641111461582257,
      "loss": 2.2038,
      "step": 36241
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6304435729980469,
      "learning_rate": 0.000564109206034736,
      "loss": 2.1607,
      "step": 36242
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.8277453184127808,
      "learning_rate": 0.0005641072658621434,
      "loss": 2.1565,
      "step": 36243
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6431701183319092,
      "learning_rate": 0.0005641053256404483,
      "loss": 2.1949,
      "step": 36244
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6860752701759338,
      "learning_rate": 0.000564103385369651,
      "loss": 2.5368,
      "step": 36245
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.8026779890060425,
      "learning_rate": 0.000564101445049752,
      "loss": 2.1934,
      "step": 36246
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.7023630142211914,
      "learning_rate": 0.0005640995046807514,
      "loss": 2.4752,
      "step": 36247
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6697254776954651,
      "learning_rate": 0.0005640975642626497,
      "loss": 2.2615,
      "step": 36248
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.7262238264083862,
      "learning_rate": 0.0005640956237954474,
      "loss": 2.3308,
      "step": 36249
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.7958067655563354,
      "learning_rate": 0.0005640936832791446,
      "loss": 2.4612,
      "step": 36250
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6902934908866882,
      "learning_rate": 0.0005640917427137419,
      "loss": 2.3037,
      "step": 36251
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6794776320457458,
      "learning_rate": 0.0005640898020992397,
      "loss": 2.0946,
      "step": 36252
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.7188226580619812,
      "learning_rate": 0.000564087861435638,
      "loss": 2.1485,
      "step": 36253
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6159114241600037,
      "learning_rate": 0.0005640859207229374,
      "loss": 2.4955,
      "step": 36254
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6169997453689575,
      "learning_rate": 0.0005640839799611383,
      "loss": 2.2853,
      "step": 36255
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6242812871932983,
      "learning_rate": 0.0005640820391502411,
      "loss": 2.1958,
      "step": 36256
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.653832733631134,
      "learning_rate": 0.000564080098290246,
      "loss": 2.3526,
      "step": 36257
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.707963764667511,
      "learning_rate": 0.0005640781573811533,
      "loss": 2.2626,
      "step": 36258
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6617307662963867,
      "learning_rate": 0.0005640762164229636,
      "loss": 2.3135,
      "step": 36259
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6806774735450745,
      "learning_rate": 0.0005640742754156772,
      "loss": 2.3639,
      "step": 36260
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.63047194480896,
      "learning_rate": 0.0005640723343592943,
      "loss": 2.0064,
      "step": 36261
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.7366271615028381,
      "learning_rate": 0.0005640703932538153,
      "loss": 2.4821,
      "step": 36262
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6591772437095642,
      "learning_rate": 0.0005640684520992408,
      "loss": 2.3589,
      "step": 36263
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.8834503889083862,
      "learning_rate": 0.0005640665108955709,
      "loss": 2.0948,
      "step": 36264
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6235363483428955,
      "learning_rate": 0.0005640645696428061,
      "loss": 2.4452,
      "step": 36265
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.7550014853477478,
      "learning_rate": 0.0005640626283409467,
      "loss": 2.3197,
      "step": 36266
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.8105127811431885,
      "learning_rate": 0.000564060686989993,
      "loss": 2.3983,
      "step": 36267
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6812363862991333,
      "learning_rate": 0.0005640587455899455,
      "loss": 2.5701,
      "step": 36268
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.8648712038993835,
      "learning_rate": 0.0005640568041408045,
      "loss": 2.564,
      "step": 36269
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6737399101257324,
      "learning_rate": 0.0005640548626425704,
      "loss": 2.3034,
      "step": 36270
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.7781624794006348,
      "learning_rate": 0.0005640529210952434,
      "loss": 2.455,
      "step": 36271
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.7320011258125305,
      "learning_rate": 0.000564050979498824,
      "loss": 2.3373,
      "step": 36272
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.8139793872833252,
      "learning_rate": 0.0005640490378533126,
      "loss": 2.137,
      "step": 36273
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6871272921562195,
      "learning_rate": 0.0005640470961587095,
      "loss": 2.138,
      "step": 36274
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6821083426475525,
      "learning_rate": 0.000564045154415015,
      "loss": 2.1722,
      "step": 36275
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6413348913192749,
      "learning_rate": 0.0005640432126222295,
      "loss": 2.1907,
      "step": 36276
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.7287179827690125,
      "learning_rate": 0.0005640412707803535,
      "loss": 2.3374,
      "step": 36277
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.7235474586486816,
      "learning_rate": 0.0005640393288893871,
      "loss": 2.4644,
      "step": 36278
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.668178915977478,
      "learning_rate": 0.000564037386949331,
      "loss": 2.2523,
      "step": 36279
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6710696220397949,
      "learning_rate": 0.0005640354449601852,
      "loss": 2.4088,
      "step": 36280
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6877472996711731,
      "learning_rate": 0.0005640335029219502,
      "loss": 2.3358,
      "step": 36281
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.7716516852378845,
      "learning_rate": 0.0005640315608346266,
      "loss": 2.1482,
      "step": 36282
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.651473879814148,
      "learning_rate": 0.0005640296186982144,
      "loss": 2.4588,
      "step": 36283
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6711680889129639,
      "learning_rate": 0.000564027676512714,
      "loss": 2.4405,
      "step": 36284
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.7734348177909851,
      "learning_rate": 0.000564025734278126,
      "loss": 2.2816,
      "step": 36285
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.700149416923523,
      "learning_rate": 0.0005640237919944507,
      "loss": 2.522,
      "step": 36286
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6581568717956543,
      "learning_rate": 0.0005640218496616883,
      "loss": 2.1074,
      "step": 36287
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6294565796852112,
      "learning_rate": 0.0005640199072798393,
      "loss": 2.2333,
      "step": 36288
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.7782076001167297,
      "learning_rate": 0.0005640179648489039,
      "loss": 2.2878,
      "step": 36289
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.880999743938446,
      "learning_rate": 0.0005640160223688828,
      "loss": 2.1676,
      "step": 36290
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.609468936920166,
      "learning_rate": 0.0005640140798397759,
      "loss": 2.1783,
      "step": 36291
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.9139342308044434,
      "learning_rate": 0.000564012137261584,
      "loss": 2.1822,
      "step": 36292
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.9991297125816345,
      "learning_rate": 0.000564010194634307,
      "loss": 2.0965,
      "step": 36293
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6740251183509827,
      "learning_rate": 0.0005640082519579458,
      "loss": 2.1869,
      "step": 36294
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.8029143214225769,
      "learning_rate": 0.0005640063092325003,
      "loss": 2.1046,
      "step": 36295
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.0471324920654297,
      "learning_rate": 0.0005640043664579711,
      "loss": 2.1692,
      "step": 36296
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.7260205745697021,
      "learning_rate": 0.0005640024236343586,
      "loss": 2.43,
      "step": 36297
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.7523892521858215,
      "learning_rate": 0.000564000480761663,
      "loss": 2.4354,
      "step": 36298
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.7277366518974304,
      "learning_rate": 0.0005639985378398846,
      "loss": 2.2036,
      "step": 36299
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.8277688026428223,
      "learning_rate": 0.000563996594869024,
      "loss": 2.2868,
      "step": 36300
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6869853138923645,
      "learning_rate": 0.0005639946518490815,
      "loss": 2.5161,
      "step": 36301
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.7132418751716614,
      "learning_rate": 0.0005639927087800574,
      "loss": 2.2696,
      "step": 36302
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.7931271195411682,
      "learning_rate": 0.0005639907656619519,
      "loss": 2.1565,
      "step": 36303
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.8520572781562805,
      "learning_rate": 0.0005639888224947657,
      "loss": 2.159,
      "step": 36304
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6693607568740845,
      "learning_rate": 0.000563986879278499,
      "loss": 2.271,
      "step": 36305
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.7519809603691101,
      "learning_rate": 0.0005639849360131521,
      "loss": 2.4632,
      "step": 36306
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.8815537691116333,
      "learning_rate": 0.0005639829926987254,
      "loss": 2.2311,
      "step": 36307
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6702574491500854,
      "learning_rate": 0.0005639810493352193,
      "loss": 2.3011,
      "step": 36308
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.73589688539505,
      "learning_rate": 0.0005639791059226342,
      "loss": 2.0749,
      "step": 36309
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.7584020495414734,
      "learning_rate": 0.0005639771624609704,
      "loss": 2.1325,
      "step": 36310
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.7050313949584961,
      "learning_rate": 0.0005639752189502282,
      "loss": 2.3716,
      "step": 36311
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6393021941184998,
      "learning_rate": 0.0005639732753904079,
      "loss": 2.271,
      "step": 36312
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6614353060722351,
      "learning_rate": 0.0005639713317815103,
      "loss": 2.5155,
      "step": 36313
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.7105831503868103,
      "learning_rate": 0.0005639693881235351,
      "loss": 2.2715,
      "step": 36314
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.7492263317108154,
      "learning_rate": 0.0005639674444164833,
      "loss": 2.3314,
      "step": 36315
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.7322497963905334,
      "learning_rate": 0.0005639655006603548,
      "loss": 2.3031,
      "step": 36316
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6636761426925659,
      "learning_rate": 0.0005639635568551502,
      "loss": 2.0972,
      "step": 36317
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.7627521753311157,
      "learning_rate": 0.0005639616130008698,
      "loss": 2.093,
      "step": 36318
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.8190116286277771,
      "learning_rate": 0.0005639596690975139,
      "loss": 2.4726,
      "step": 36319
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.7097468972206116,
      "learning_rate": 0.000563957725145083,
      "loss": 2.1815,
      "step": 36320
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.699443519115448,
      "learning_rate": 0.0005639557811435772,
      "loss": 2.332,
      "step": 36321
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.7793692946434021,
      "learning_rate": 0.0005639538370929972,
      "loss": 2.5942,
      "step": 36322
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.685370683670044,
      "learning_rate": 0.0005639518929933432,
      "loss": 2.4351,
      "step": 36323
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.7365633249282837,
      "learning_rate": 0.0005639499488446154,
      "loss": 2.2684,
      "step": 36324
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.7009458541870117,
      "learning_rate": 0.0005639480046468145,
      "loss": 2.4634,
      "step": 36325
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6702237129211426,
      "learning_rate": 0.0005639460603999406,
      "loss": 2.2789,
      "step": 36326
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6580926775932312,
      "learning_rate": 0.0005639441161039941,
      "loss": 2.4099,
      "step": 36327
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6789479851722717,
      "learning_rate": 0.0005639421717589755,
      "loss": 2.2651,
      "step": 36328
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.7402456998825073,
      "learning_rate": 0.000563940227364885,
      "loss": 2.3531,
      "step": 36329
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.7144144773483276,
      "learning_rate": 0.0005639382829217232,
      "loss": 2.3359,
      "step": 36330
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.7266632318496704,
      "learning_rate": 0.0005639363384294901,
      "loss": 2.4229,
      "step": 36331
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6815871000289917,
      "learning_rate": 0.0005639343938881863,
      "loss": 2.4313,
      "step": 36332
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.7506164312362671,
      "learning_rate": 0.0005639324492978121,
      "loss": 2.316,
      "step": 36333
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6362336277961731,
      "learning_rate": 0.0005639305046583679,
      "loss": 2.2446,
      "step": 36334
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6779451966285706,
      "learning_rate": 0.000563928559969854,
      "loss": 2.2414,
      "step": 36335
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.7006142735481262,
      "learning_rate": 0.0005639266152322708,
      "loss": 2.1775,
      "step": 36336
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.665635347366333,
      "learning_rate": 0.0005639246704456188,
      "loss": 2.4562,
      "step": 36337
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6396192312240601,
      "learning_rate": 0.0005639227256098981,
      "loss": 2.3137,
      "step": 36338
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6696494817733765,
      "learning_rate": 0.0005639207807251091,
      "loss": 2.2622,
      "step": 36339
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6964598298072815,
      "learning_rate": 0.0005639188357912523,
      "loss": 2.299,
      "step": 36340
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.790037989616394,
      "learning_rate": 0.0005639168908083282,
      "loss": 2.4076,
      "step": 36341
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6952895522117615,
      "learning_rate": 0.0005639149457763368,
      "loss": 2.4605,
      "step": 36342
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.733893632888794,
      "learning_rate": 0.0005639130006952787,
      "loss": 2.2318,
      "step": 36343
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.7537714242935181,
      "learning_rate": 0.0005639110555651541,
      "loss": 2.2872,
      "step": 36344
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6609136462211609,
      "learning_rate": 0.0005639091103859635,
      "loss": 2.299,
      "step": 36345
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6601184606552124,
      "learning_rate": 0.0005639071651577072,
      "loss": 2.1025,
      "step": 36346
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.7672944664955139,
      "learning_rate": 0.0005639052198803856,
      "loss": 2.4223,
      "step": 36347
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.7709934711456299,
      "learning_rate": 0.0005639032745539989,
      "loss": 2.4217,
      "step": 36348
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6896519660949707,
      "learning_rate": 0.0005639013291785478,
      "loss": 2.3383,
      "step": 36349
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.7201541662216187,
      "learning_rate": 0.0005638993837540324,
      "loss": 2.3096,
      "step": 36350
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.7575642466545105,
      "learning_rate": 0.0005638974382804531,
      "loss": 2.4933,
      "step": 36351
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.7831138968467712,
      "learning_rate": 0.0005638954927578103,
      "loss": 2.2559,
      "step": 36352
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.7048823237419128,
      "learning_rate": 0.0005638935471861044,
      "loss": 2.1859,
      "step": 36353
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.7236130237579346,
      "learning_rate": 0.0005638916015653356,
      "loss": 2.2893,
      "step": 36354
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6865200996398926,
      "learning_rate": 0.0005638896558955044,
      "loss": 2.3748,
      "step": 36355
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.7772822380065918,
      "learning_rate": 0.0005638877101766111,
      "loss": 2.0797,
      "step": 36356
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6903663277626038,
      "learning_rate": 0.0005638857644086562,
      "loss": 2.3039,
      "step": 36357
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.7334630489349365,
      "learning_rate": 0.0005638838185916399,
      "loss": 2.2794,
      "step": 36358
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.649173378944397,
      "learning_rate": 0.0005638818727255625,
      "loss": 2.177,
      "step": 36359
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.7155869603157043,
      "learning_rate": 0.0005638799268104247,
      "loss": 2.1736,
      "step": 36360
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.7032793760299683,
      "learning_rate": 0.0005638779808462266,
      "loss": 2.3526,
      "step": 36361
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6744583249092102,
      "learning_rate": 0.0005638760348329684,
      "loss": 2.3375,
      "step": 36362
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6848230361938477,
      "learning_rate": 0.0005638740887706509,
      "loss": 2.2468,
      "step": 36363
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6968305706977844,
      "learning_rate": 0.000563872142659274,
      "loss": 2.2788,
      "step": 36364
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6501424312591553,
      "learning_rate": 0.0005638701964988385,
      "loss": 2.3253,
      "step": 36365
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6983605623245239,
      "learning_rate": 0.0005638682502893444,
      "loss": 2.353,
      "step": 36366
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.7019464373588562,
      "learning_rate": 0.0005638663040307923,
      "loss": 2.3323,
      "step": 36367
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.7540310621261597,
      "learning_rate": 0.0005638643577231825,
      "loss": 2.2828,
      "step": 36368
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6236746907234192,
      "learning_rate": 0.0005638624113665152,
      "loss": 2.2322,
      "step": 36369
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.670379638671875,
      "learning_rate": 0.000563860464960791,
      "loss": 2.3577,
      "step": 36370
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.693877100944519,
      "learning_rate": 0.0005638585185060101,
      "loss": 2.3395,
      "step": 36371
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6858661770820618,
      "learning_rate": 0.000563856572002173,
      "loss": 2.5131,
      "step": 36372
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.7417614459991455,
      "learning_rate": 0.00056385462544928,
      "loss": 2.3047,
      "step": 36373
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.7861376404762268,
      "learning_rate": 0.0005638526788473313,
      "loss": 2.2863,
      "step": 36374
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6339395642280579,
      "learning_rate": 0.0005638507321963276,
      "loss": 2.3842,
      "step": 36375
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.7647690176963806,
      "learning_rate": 0.000563848785496269,
      "loss": 2.2838,
      "step": 36376
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6378148198127747,
      "learning_rate": 0.0005638468387471558,
      "loss": 2.1677,
      "step": 36377
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6656678915023804,
      "learning_rate": 0.0005638448919489886,
      "loss": 2.205,
      "step": 36378
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6537644267082214,
      "learning_rate": 0.0005638429451017676,
      "loss": 2.3923,
      "step": 36379
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6207597255706787,
      "learning_rate": 0.0005638409982054932,
      "loss": 2.2884,
      "step": 36380
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6545083522796631,
      "learning_rate": 0.0005638390512601659,
      "loss": 2.2239,
      "step": 36381
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6679072380065918,
      "learning_rate": 0.0005638371042657859,
      "loss": 2.133,
      "step": 36382
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.7442247271537781,
      "learning_rate": 0.0005638351572223536,
      "loss": 2.5268,
      "step": 36383
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.714313268661499,
      "learning_rate": 0.0005638332101298693,
      "loss": 2.4124,
      "step": 36384
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6789960265159607,
      "learning_rate": 0.0005638312629883335,
      "loss": 2.1112,
      "step": 36385
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6413059234619141,
      "learning_rate": 0.0005638293157977463,
      "loss": 2.2336,
      "step": 36386
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.7284976243972778,
      "learning_rate": 0.0005638273685581084,
      "loss": 2.5493,
      "step": 36387
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.691944420337677,
      "learning_rate": 0.0005638254212694201,
      "loss": 2.3348,
      "step": 36388
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.714675784111023,
      "learning_rate": 0.0005638234739316816,
      "loss": 2.3135,
      "step": 36389
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6606578230857849,
      "learning_rate": 0.0005638215265448933,
      "loss": 2.3786,
      "step": 36390
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.619174599647522,
      "learning_rate": 0.0005638195791090556,
      "loss": 2.0158,
      "step": 36391
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6647834181785583,
      "learning_rate": 0.0005638176316241689,
      "loss": 2.1288,
      "step": 36392
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6961115598678589,
      "learning_rate": 0.0005638156840902335,
      "loss": 2.4578,
      "step": 36393
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6229215264320374,
      "learning_rate": 0.0005638137365072499,
      "loss": 2.4166,
      "step": 36394
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.7463759779930115,
      "learning_rate": 0.0005638117888752181,
      "loss": 2.2367,
      "step": 36395
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6078621745109558,
      "learning_rate": 0.0005638098411941388,
      "loss": 2.1912,
      "step": 36396
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6553921103477478,
      "learning_rate": 0.0005638078934640123,
      "loss": 1.939,
      "step": 36397
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6915431618690491,
      "learning_rate": 0.000563805945684839,
      "loss": 2.2105,
      "step": 36398
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6843986511230469,
      "learning_rate": 0.0005638039978566191,
      "loss": 2.1962,
      "step": 36399
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6864753365516663,
      "learning_rate": 0.0005638020499793532,
      "loss": 2.2196,
      "step": 36400
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6837088465690613,
      "learning_rate": 0.0005638001020530414,
      "loss": 2.3694,
      "step": 36401
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6387485265731812,
      "learning_rate": 0.0005637981540776842,
      "loss": 2.4299,
      "step": 36402
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.8123103380203247,
      "learning_rate": 0.000563796206053282,
      "loss": 2.3646,
      "step": 36403
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.800537645816803,
      "learning_rate": 0.000563794257979835,
      "loss": 2.3651,
      "step": 36404
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.7783415913581848,
      "learning_rate": 0.0005637923098573437,
      "loss": 2.4709,
      "step": 36405
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.9227256178855896,
      "learning_rate": 0.0005637903616858085,
      "loss": 2.5006,
      "step": 36406
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6487983465194702,
      "learning_rate": 0.0005637884134652297,
      "loss": 2.215,
      "step": 36407
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.7575986981391907,
      "learning_rate": 0.0005637864651956074,
      "loss": 2.3162,
      "step": 36408
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.7116217017173767,
      "learning_rate": 0.0005637845168769425,
      "loss": 2.3565,
      "step": 36409
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6871475577354431,
      "learning_rate": 0.000563782568509235,
      "loss": 2.1875,
      "step": 36410
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.7301583886146545,
      "learning_rate": 0.0005637806200924853,
      "loss": 2.1325,
      "step": 36411
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6728177666664124,
      "learning_rate": 0.0005637786716266938,
      "loss": 2.3203,
      "step": 36412
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6377668380737305,
      "learning_rate": 0.000563776723111861,
      "loss": 2.0957,
      "step": 36413
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.7041703462600708,
      "learning_rate": 0.000563774774547987,
      "loss": 2.3683,
      "step": 36414
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6708130836486816,
      "learning_rate": 0.0005637728259350723,
      "loss": 2.2343,
      "step": 36415
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6226471066474915,
      "learning_rate": 0.0005637708772731172,
      "loss": 2.2846,
      "step": 36416
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6804802417755127,
      "learning_rate": 0.0005637689285621222,
      "loss": 2.2408,
      "step": 36417
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6676617860794067,
      "learning_rate": 0.0005637669798020876,
      "loss": 2.1602,
      "step": 36418
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6719253659248352,
      "learning_rate": 0.0005637650309930137,
      "loss": 2.2093,
      "step": 36419
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.688709557056427,
      "learning_rate": 0.0005637630821349009,
      "loss": 2.3259,
      "step": 36420
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.675443172454834,
      "learning_rate": 0.0005637611332277496,
      "loss": 2.2134,
      "step": 36421
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6871898174285889,
      "learning_rate": 0.0005637591842715601,
      "loss": 2.1757,
      "step": 36422
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6831821203231812,
      "learning_rate": 0.0005637572352663327,
      "loss": 2.4473,
      "step": 36423
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6756443381309509,
      "learning_rate": 0.000563755286212068,
      "loss": 2.487,
      "step": 36424
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.692740261554718,
      "learning_rate": 0.0005637533371087661,
      "loss": 2.0835,
      "step": 36425
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.7281158566474915,
      "learning_rate": 0.0005637513879564275,
      "loss": 2.5198,
      "step": 36426
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6931083798408508,
      "learning_rate": 0.0005637494387550527,
      "loss": 2.5474,
      "step": 36427
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.7010185718536377,
      "learning_rate": 0.0005637474895046418,
      "loss": 2.5005,
      "step": 36428
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6442718505859375,
      "learning_rate": 0.0005637455402051951,
      "loss": 2.3043,
      "step": 36429
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.7426183819770813,
      "learning_rate": 0.0005637435908567133,
      "loss": 2.2421,
      "step": 36430
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.7316254377365112,
      "learning_rate": 0.0005637416414591965,
      "loss": 2.2719,
      "step": 36431
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.7033008337020874,
      "learning_rate": 0.0005637396920126453,
      "loss": 2.3431,
      "step": 36432
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6204631328582764,
      "learning_rate": 0.0005637377425170598,
      "loss": 2.2991,
      "step": 36433
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.940307080745697,
      "learning_rate": 0.0005637357929724404,
      "loss": 2.2983,
      "step": 36434
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6451287269592285,
      "learning_rate": 0.0005637338433787877,
      "loss": 2.1541,
      "step": 36435
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6866536736488342,
      "learning_rate": 0.0005637318937361018,
      "loss": 2.0229,
      "step": 36436
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.7550148367881775,
      "learning_rate": 0.000563729944044383,
      "loss": 2.0969,
      "step": 36437
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6780451536178589,
      "learning_rate": 0.0005637279943036321,
      "loss": 2.2301,
      "step": 36438
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.7015228867530823,
      "learning_rate": 0.0005637260445138492,
      "loss": 2.4059,
      "step": 36439
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.7819444537162781,
      "learning_rate": 0.0005637240946750345,
      "loss": 2.34,
      "step": 36440
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.684532105922699,
      "learning_rate": 0.0005637221447871884,
      "loss": 2.1973,
      "step": 36441
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.7034986615180969,
      "learning_rate": 0.0005637201948503116,
      "loss": 2.3227,
      "step": 36442
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.7073531150817871,
      "learning_rate": 0.0005637182448644042,
      "loss": 2.3485,
      "step": 36443
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.7136020660400391,
      "learning_rate": 0.0005637162948294665,
      "loss": 2.3758,
      "step": 36444
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.8527194857597351,
      "learning_rate": 0.0005637143447454992,
      "loss": 2.1097,
      "step": 36445
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6893103122711182,
      "learning_rate": 0.0005637123946125021,
      "loss": 2.1347,
      "step": 36446
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6155161261558533,
      "learning_rate": 0.0005637104444304761,
      "loss": 2.2186,
      "step": 36447
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.7709179520606995,
      "learning_rate": 0.0005637084941994213,
      "loss": 2.3661,
      "step": 36448
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6432580351829529,
      "learning_rate": 0.0005637065439193381,
      "loss": 2.1135,
      "step": 36449
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6431018710136414,
      "learning_rate": 0.0005637045935902269,
      "loss": 2.1835,
      "step": 36450
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6799545884132385,
      "learning_rate": 0.000563702643212088,
      "loss": 2.2305,
      "step": 36451
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.1522692441940308,
      "learning_rate": 0.0005637006927849218,
      "loss": 2.2703,
      "step": 36452
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6532832980155945,
      "learning_rate": 0.0005636987423087287,
      "loss": 2.2303,
      "step": 36453
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6027187705039978,
      "learning_rate": 0.000563696791783509,
      "loss": 2.2846,
      "step": 36454
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6872610449790955,
      "learning_rate": 0.0005636948412092631,
      "loss": 2.1897,
      "step": 36455
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6598705649375916,
      "learning_rate": 0.0005636928905859914,
      "loss": 2.39,
      "step": 36456
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.665212869644165,
      "learning_rate": 0.0005636909399136941,
      "loss": 2.289,
      "step": 36457
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.7160019874572754,
      "learning_rate": 0.0005636889891923718,
      "loss": 2.4409,
      "step": 36458
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6486143469810486,
      "learning_rate": 0.0005636870384220247,
      "loss": 2.2748,
      "step": 36459
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6197910904884338,
      "learning_rate": 0.0005636850876026531,
      "loss": 2.3103,
      "step": 36460
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6489624977111816,
      "learning_rate": 0.0005636831367342576,
      "loss": 2.3697,
      "step": 36461
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6597340703010559,
      "learning_rate": 0.0005636811858168384,
      "loss": 2.2462,
      "step": 36462
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.7428938746452332,
      "learning_rate": 0.0005636792348503958,
      "loss": 2.4032,
      "step": 36463
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6346182823181152,
      "learning_rate": 0.0005636772838349303,
      "loss": 2.3575,
      "step": 36464
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6501525044441223,
      "learning_rate": 0.0005636753327704423,
      "loss": 2.4868,
      "step": 36465
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6797511577606201,
      "learning_rate": 0.000563673381656932,
      "loss": 2.234,
      "step": 36466
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6915962100028992,
      "learning_rate": 0.0005636714304943999,
      "loss": 2.1495,
      "step": 36467
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6777521371841431,
      "learning_rate": 0.0005636694792828462,
      "loss": 2.0912,
      "step": 36468
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6786648035049438,
      "learning_rate": 0.0005636675280222714,
      "loss": 2.3731,
      "step": 36469
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6399146318435669,
      "learning_rate": 0.0005636655767126759,
      "loss": 2.439,
      "step": 36470
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.601372480392456,
      "learning_rate": 0.0005636636253540599,
      "loss": 1.9923,
      "step": 36471
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6850584149360657,
      "learning_rate": 0.000563661673946424,
      "loss": 2.3613,
      "step": 36472
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.9591765403747559,
      "learning_rate": 0.0005636597224897682,
      "loss": 2.2599,
      "step": 36473
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6657029390335083,
      "learning_rate": 0.0005636577709840933,
      "loss": 2.2796,
      "step": 36474
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.7560908198356628,
      "learning_rate": 0.0005636558194293994,
      "loss": 2.2857,
      "step": 36475
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6797101497650146,
      "learning_rate": 0.0005636538678256866,
      "loss": 2.1812,
      "step": 36476
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6510411500930786,
      "learning_rate": 0.0005636519161729559,
      "loss": 2.5248,
      "step": 36477
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6949231028556824,
      "learning_rate": 0.0005636499644712073,
      "loss": 2.3613,
      "step": 36478
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.7985618114471436,
      "learning_rate": 0.0005636480127204412,
      "loss": 2.2282,
      "step": 36479
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.7234842777252197,
      "learning_rate": 0.0005636460609206578,
      "loss": 2.291,
      "step": 36480
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6580400466918945,
      "learning_rate": 0.0005636441090718577,
      "loss": 2.2325,
      "step": 36481
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.703252375125885,
      "learning_rate": 0.0005636421571740413,
      "loss": 2.2617,
      "step": 36482
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.706195592880249,
      "learning_rate": 0.0005636402052272087,
      "loss": 2.3755,
      "step": 36483
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6209449172019958,
      "learning_rate": 0.0005636382532313605,
      "loss": 2.4856,
      "step": 36484
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6446207761764526,
      "learning_rate": 0.0005636363011864969,
      "loss": 2.3217,
      "step": 36485
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.7143926620483398,
      "learning_rate": 0.0005636343490926185,
      "loss": 2.364,
      "step": 36486
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.625714898109436,
      "learning_rate": 0.0005636323969497253,
      "loss": 2.1958,
      "step": 36487
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.7102421522140503,
      "learning_rate": 0.0005636304447578179,
      "loss": 2.275,
      "step": 36488
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.7218449711799622,
      "learning_rate": 0.0005636284925168966,
      "loss": 2.3964,
      "step": 36489
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6939915418624878,
      "learning_rate": 0.0005636265402269618,
      "loss": 2.2771,
      "step": 36490
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.7028450965881348,
      "learning_rate": 0.000563624587888014,
      "loss": 1.9669,
      "step": 36491
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6695407032966614,
      "learning_rate": 0.0005636226355000532,
      "loss": 2.4665,
      "step": 36492
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.7426700592041016,
      "learning_rate": 0.0005636206830630802,
      "loss": 2.1854,
      "step": 36493
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.60196453332901,
      "learning_rate": 0.000563618730577095,
      "loss": 2.1947,
      "step": 36494
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6986981630325317,
      "learning_rate": 0.000563616778042098,
      "loss": 2.4685,
      "step": 36495
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.680903434753418,
      "learning_rate": 0.0005636148254580898,
      "loss": 2.1003,
      "step": 36496
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6605600714683533,
      "learning_rate": 0.0005636128728250707,
      "loss": 2.3533,
      "step": 36497
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6399988532066345,
      "learning_rate": 0.0005636109201430408,
      "loss": 2.3029,
      "step": 36498
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.645453691482544,
      "learning_rate": 0.0005636089674120008,
      "loss": 2.0867,
      "step": 36499
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.7697583436965942,
      "learning_rate": 0.000563607014631951,
      "loss": 2.2128,
      "step": 36500
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6491093039512634,
      "learning_rate": 0.0005636050618028915,
      "loss": 2.0801,
      "step": 36501
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.674136221408844,
      "learning_rate": 0.0005636031089248229,
      "loss": 2.3593,
      "step": 36502
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6742996573448181,
      "learning_rate": 0.0005636011559977455,
      "loss": 2.4523,
      "step": 36503
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.737041175365448,
      "learning_rate": 0.0005635992030216596,
      "loss": 2.2436,
      "step": 36504
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.7108150720596313,
      "learning_rate": 0.0005635972499965658,
      "loss": 2.4052,
      "step": 36505
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6570100784301758,
      "learning_rate": 0.0005635952969224641,
      "loss": 2.2373,
      "step": 36506
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.7150996327400208,
      "learning_rate": 0.0005635933437993551,
      "loss": 2.3197,
      "step": 36507
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6524344682693481,
      "learning_rate": 0.0005635913906272392,
      "loss": 2.4714,
      "step": 36508
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6402462124824524,
      "learning_rate": 0.0005635894374061167,
      "loss": 2.2905,
      "step": 36509
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6083586812019348,
      "learning_rate": 0.000563587484135988,
      "loss": 2.5046,
      "step": 36510
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6459023356437683,
      "learning_rate": 0.0005635855308168533,
      "loss": 2.2633,
      "step": 36511
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6900836825370789,
      "learning_rate": 0.0005635835774487131,
      "loss": 2.2289,
      "step": 36512
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6983367800712585,
      "learning_rate": 0.0005635816240315677,
      "loss": 2.3842,
      "step": 36513
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.7347694039344788,
      "learning_rate": 0.0005635796705654176,
      "loss": 2.4135,
      "step": 36514
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6760650277137756,
      "learning_rate": 0.000563577717050263,
      "loss": 2.2698,
      "step": 36515
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6795472502708435,
      "learning_rate": 0.0005635757634861043,
      "loss": 2.4524,
      "step": 36516
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6399080157279968,
      "learning_rate": 0.0005635738098729419,
      "loss": 2.3119,
      "step": 36517
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6560256481170654,
      "learning_rate": 0.0005635718562107762,
      "loss": 2.2058,
      "step": 36518
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6487269997596741,
      "learning_rate": 0.0005635699024996075,
      "loss": 2.3549,
      "step": 36519
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6791326999664307,
      "learning_rate": 0.0005635679487394361,
      "loss": 2.497,
      "step": 36520
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.667679488658905,
      "learning_rate": 0.0005635659949302626,
      "loss": 2.123,
      "step": 36521
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6834304332733154,
      "learning_rate": 0.000563564041072087,
      "loss": 2.5316,
      "step": 36522
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6121498942375183,
      "learning_rate": 0.00056356208716491,
      "loss": 2.1634,
      "step": 36523
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.9177005887031555,
      "learning_rate": 0.0005635601332087319,
      "loss": 2.2589,
      "step": 36524
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.7110331654548645,
      "learning_rate": 0.0005635581792035528,
      "loss": 2.1998,
      "step": 36525
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.8089677691459656,
      "learning_rate": 0.0005635562251493734,
      "loss": 2.2484,
      "step": 36526
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.771676778793335,
      "learning_rate": 0.0005635542710461939,
      "loss": 2.3602,
      "step": 36527
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.7185361981391907,
      "learning_rate": 0.0005635523168940147,
      "loss": 2.5667,
      "step": 36528
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.672262966632843,
      "learning_rate": 0.0005635503626928362,
      "loss": 2.1821,
      "step": 36529
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6561068296432495,
      "learning_rate": 0.0005635484084426586,
      "loss": 2.2871,
      "step": 36530
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6242940425872803,
      "learning_rate": 0.0005635464541434824,
      "loss": 1.9579,
      "step": 36531
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6232393383979797,
      "learning_rate": 0.0005635444997953079,
      "loss": 2.3328,
      "step": 36532
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.767313539981842,
      "learning_rate": 0.0005635425453981356,
      "loss": 2.3865,
      "step": 36533
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.7148499488830566,
      "learning_rate": 0.0005635405909519657,
      "loss": 2.3493,
      "step": 36534
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6455715298652649,
      "learning_rate": 0.0005635386364567987,
      "loss": 2.1871,
      "step": 36535
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.803280234336853,
      "learning_rate": 0.0005635366819126349,
      "loss": 2.2045,
      "step": 36536
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.7423843741416931,
      "learning_rate": 0.0005635347273194746,
      "loss": 2.1326,
      "step": 36537
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.7059385776519775,
      "learning_rate": 0.0005635327726773182,
      "loss": 2.3254,
      "step": 36538
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.7598845362663269,
      "learning_rate": 0.0005635308179861661,
      "loss": 2.5582,
      "step": 36539
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.7010864615440369,
      "learning_rate": 0.0005635288632460187,
      "loss": 2.2718,
      "step": 36540
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6451449394226074,
      "learning_rate": 0.0005635269084568763,
      "loss": 2.3761,
      "step": 36541
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6349475979804993,
      "learning_rate": 0.0005635249536187391,
      "loss": 1.9517,
      "step": 36542
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.8054138422012329,
      "learning_rate": 0.0005635229987316078,
      "loss": 2.3159,
      "step": 36543
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6715527772903442,
      "learning_rate": 0.0005635210437954826,
      "loss": 2.0066,
      "step": 36544
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.5981935858726501,
      "learning_rate": 0.0005635190888103639,
      "loss": 2.1602,
      "step": 36545
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.7584526538848877,
      "learning_rate": 0.0005635171337762519,
      "loss": 2.3667,
      "step": 36546
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.7344563007354736,
      "learning_rate": 0.0005635151786931472,
      "loss": 2.4288,
      "step": 36547
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6392746567726135,
      "learning_rate": 0.00056351322356105,
      "loss": 2.2543,
      "step": 36548
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6954455375671387,
      "learning_rate": 0.0005635112683799607,
      "loss": 2.4039,
      "step": 36549
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.62291419506073,
      "learning_rate": 0.0005635093131498797,
      "loss": 2.3017,
      "step": 36550
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.7977260947227478,
      "learning_rate": 0.0005635073578708074,
      "loss": 2.1978,
      "step": 36551
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.7038689851760864,
      "learning_rate": 0.0005635054025427441,
      "loss": 2.3496,
      "step": 36552
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.897344708442688,
      "learning_rate": 0.00056350344716569,
      "loss": 2.3448,
      "step": 36553
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.7168551087379456,
      "learning_rate": 0.0005635014917396459,
      "loss": 2.2551,
      "step": 36554
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.688728392124176,
      "learning_rate": 0.0005634995362646118,
      "loss": 2.33,
      "step": 36555
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.8010635375976562,
      "learning_rate": 0.000563497580740588,
      "loss": 2.0691,
      "step": 36556
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.700922429561615,
      "learning_rate": 0.0005634956251675752,
      "loss": 2.1938,
      "step": 36557
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.7309650778770447,
      "learning_rate": 0.0005634936695455736,
      "loss": 2.4172,
      "step": 36558
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.022412657737732,
      "learning_rate": 0.0005634917138745834,
      "loss": 2.1667,
      "step": 36559
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.7260527610778809,
      "learning_rate": 0.0005634897581546053,
      "loss": 2.2727,
      "step": 36560
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6645829081535339,
      "learning_rate": 0.0005634878023856394,
      "loss": 2.1956,
      "step": 36561
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.8659345507621765,
      "learning_rate": 0.0005634858465676861,
      "loss": 2.2972,
      "step": 36562
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.7104679346084595,
      "learning_rate": 0.0005634838907007458,
      "loss": 2.2408,
      "step": 36563
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6732611060142517,
      "learning_rate": 0.000563481934784819,
      "loss": 2.0361,
      "step": 36564
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.7561599016189575,
      "learning_rate": 0.0005634799788199057,
      "loss": 2.2594,
      "step": 36565
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6931480169296265,
      "learning_rate": 0.0005634780228060068,
      "loss": 2.1658,
      "step": 36566
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.7069803476333618,
      "learning_rate": 0.0005634760667431221,
      "loss": 2.2986,
      "step": 36567
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6694048047065735,
      "learning_rate": 0.0005634741106312523,
      "loss": 2.3559,
      "step": 36568
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6765676736831665,
      "learning_rate": 0.0005634721544703978,
      "loss": 2.2075,
      "step": 36569
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.7462122440338135,
      "learning_rate": 0.0005634701982605588,
      "loss": 2.4067,
      "step": 36570
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.7242693901062012,
      "learning_rate": 0.0005634682420017357,
      "loss": 2.0159,
      "step": 36571
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6216524243354797,
      "learning_rate": 0.0005634662856939289,
      "loss": 2.1328,
      "step": 36572
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6921514272689819,
      "learning_rate": 0.0005634643293371388,
      "loss": 2.3604,
      "step": 36573
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6406574249267578,
      "learning_rate": 0.0005634623729313655,
      "loss": 2.3474,
      "step": 36574
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6201566457748413,
      "learning_rate": 0.0005634604164766097,
      "loss": 2.5592,
      "step": 36575
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.7484852075576782,
      "learning_rate": 0.0005634584599728717,
      "loss": 2.2361,
      "step": 36576
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.7193686366081238,
      "learning_rate": 0.0005634565034201518,
      "loss": 2.2308,
      "step": 36577
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.7113999724388123,
      "learning_rate": 0.0005634545468184502,
      "loss": 2.2489,
      "step": 36578
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6720778942108154,
      "learning_rate": 0.0005634525901677675,
      "loss": 2.2377,
      "step": 36579
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6876112818717957,
      "learning_rate": 0.000563450633468104,
      "loss": 2.3964,
      "step": 36580
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6620567440986633,
      "learning_rate": 0.0005634486767194601,
      "loss": 2.1982,
      "step": 36581
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.8490893840789795,
      "learning_rate": 0.0005634467199218361,
      "loss": 2.1999,
      "step": 36582
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6508756875991821,
      "learning_rate": 0.0005634447630752323,
      "loss": 2.084,
      "step": 36583
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6706458330154419,
      "learning_rate": 0.0005634428061796493,
      "loss": 2.1721,
      "step": 36584
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.7187790274620056,
      "learning_rate": 0.0005634408492350872,
      "loss": 2.3397,
      "step": 36585
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.8474462032318115,
      "learning_rate": 0.0005634388922415466,
      "loss": 2.4519,
      "step": 36586
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.707334578037262,
      "learning_rate": 0.0005634369351990275,
      "loss": 2.1307,
      "step": 36587
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6770284175872803,
      "learning_rate": 0.0005634349781075308,
      "loss": 2.3889,
      "step": 36588
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6500150561332703,
      "learning_rate": 0.0005634330209670563,
      "loss": 2.3686,
      "step": 36589
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.7097293734550476,
      "learning_rate": 0.0005634310637776047,
      "loss": 2.3754,
      "step": 36590
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6835433840751648,
      "learning_rate": 0.0005634291065391763,
      "loss": 2.3603,
      "step": 36591
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.7232665419578552,
      "learning_rate": 0.0005634271492517716,
      "loss": 2.4351,
      "step": 36592
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6279451251029968,
      "learning_rate": 0.0005634251919153907,
      "loss": 2.1542,
      "step": 36593
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6515175700187683,
      "learning_rate": 0.000563423234530034,
      "loss": 2.3224,
      "step": 36594
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6291717886924744,
      "learning_rate": 0.0005634212770957021,
      "loss": 2.2858,
      "step": 36595
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.637797474861145,
      "learning_rate": 0.000563419319612395,
      "loss": 2.2057,
      "step": 36596
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6226249933242798,
      "learning_rate": 0.0005634173620801135,
      "loss": 2.4077,
      "step": 36597
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6469582915306091,
      "learning_rate": 0.0005634154044988576,
      "loss": 2.1749,
      "step": 36598
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6649371981620789,
      "learning_rate": 0.000563413446868628,
      "loss": 2.3069,
      "step": 36599
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6688264012336731,
      "learning_rate": 0.0005634114891894246,
      "loss": 2.372,
      "step": 36600
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6757612824440002,
      "learning_rate": 0.0005634095314612482,
      "loss": 2.1341,
      "step": 36601
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6945117115974426,
      "learning_rate": 0.0005634075736840991,
      "loss": 2.1927,
      "step": 36602
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6308683156967163,
      "learning_rate": 0.0005634056158579772,
      "loss": 2.1691,
      "step": 36603
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6456308364868164,
      "learning_rate": 0.0005634036579828835,
      "loss": 2.1751,
      "step": 36604
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.7109354734420776,
      "learning_rate": 0.0005634017000588181,
      "loss": 2.3419,
      "step": 36605
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6538608074188232,
      "learning_rate": 0.0005633997420857813,
      "loss": 2.0482,
      "step": 36606
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.7333763241767883,
      "learning_rate": 0.0005633977840637734,
      "loss": 2.2696,
      "step": 36607
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.7496581673622131,
      "learning_rate": 0.0005633958259927951,
      "loss": 2.381,
      "step": 36608
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6735325455665588,
      "learning_rate": 0.0005633938678728464,
      "loss": 2.1242,
      "step": 36609
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.7224481105804443,
      "learning_rate": 0.0005633919097039279,
      "loss": 2.0506,
      "step": 36610
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6188814640045166,
      "learning_rate": 0.0005633899514860398,
      "loss": 2.156,
      "step": 36611
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6632323265075684,
      "learning_rate": 0.0005633879932191825,
      "loss": 2.2545,
      "step": 36612
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6716423630714417,
      "learning_rate": 0.0005633860349033564,
      "loss": 2.074,
      "step": 36613
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6765773296356201,
      "learning_rate": 0.0005633840765385619,
      "loss": 2.4723,
      "step": 36614
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.627199113368988,
      "learning_rate": 0.0005633821181247993,
      "loss": 2.1093,
      "step": 36615
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6533066034317017,
      "learning_rate": 0.0005633801596620691,
      "loss": 2.3545,
      "step": 36616
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6830061078071594,
      "learning_rate": 0.0005633782011503714,
      "loss": 2.3423,
      "step": 36617
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.7124523520469666,
      "learning_rate": 0.0005633762425897068,
      "loss": 2.2646,
      "step": 36618
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6877037882804871,
      "learning_rate": 0.0005633742839800756,
      "loss": 2.1185,
      "step": 36619
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.5975678563117981,
      "learning_rate": 0.0005633723253214781,
      "loss": 2.1539,
      "step": 36620
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6236370205879211,
      "learning_rate": 0.0005633703666139148,
      "loss": 2.5817,
      "step": 36621
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6696792840957642,
      "learning_rate": 0.000563368407857386,
      "loss": 2.2226,
      "step": 36622
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.8629102110862732,
      "learning_rate": 0.0005633664490518918,
      "loss": 2.3088,
      "step": 36623
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6073732376098633,
      "learning_rate": 0.0005633644901974331,
      "loss": 2.3055,
      "step": 36624
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.668746829032898,
      "learning_rate": 0.0005633625312940098,
      "loss": 2.4182,
      "step": 36625
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6664109826087952,
      "learning_rate": 0.0005633605723416225,
      "loss": 2.1915,
      "step": 36626
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.7124621272087097,
      "learning_rate": 0.0005633586133402715,
      "loss": 2.335,
      "step": 36627
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.7019113302230835,
      "learning_rate": 0.0005633566542899571,
      "loss": 2.2732,
      "step": 36628
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6689795851707458,
      "learning_rate": 0.0005633546951906797,
      "loss": 2.4476,
      "step": 36629
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6017826795578003,
      "learning_rate": 0.0005633527360424398,
      "loss": 2.2115,
      "step": 36630
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6175782084465027,
      "learning_rate": 0.0005633507768452377,
      "loss": 2.1322,
      "step": 36631
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6587755680084229,
      "learning_rate": 0.0005633488175990736,
      "loss": 2.0466,
      "step": 36632
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6718366146087646,
      "learning_rate": 0.000563346858303948,
      "loss": 2.2413,
      "step": 36633
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.7069177031517029,
      "learning_rate": 0.0005633448989598612,
      "loss": 2.1567,
      "step": 36634
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6695288419723511,
      "learning_rate": 0.0005633429395668138,
      "loss": 2.1513,
      "step": 36635
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6479403972625732,
      "learning_rate": 0.0005633409801248058,
      "loss": 2.1855,
      "step": 36636
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.7390580177307129,
      "learning_rate": 0.0005633390206338378,
      "loss": 2.1729,
      "step": 36637
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.624733567237854,
      "learning_rate": 0.0005633370610939101,
      "loss": 2.3379,
      "step": 36638
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6276768445968628,
      "learning_rate": 0.0005633351015050231,
      "loss": 2.2266,
      "step": 36639
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6879162788391113,
      "learning_rate": 0.000563333141867177,
      "loss": 2.2704,
      "step": 36640
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.7322236895561218,
      "learning_rate": 0.0005633311821803725,
      "loss": 2.3111,
      "step": 36641
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.8285963535308838,
      "learning_rate": 0.0005633292224446096,
      "loss": 2.4096,
      "step": 36642
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.645505428314209,
      "learning_rate": 0.0005633272626598889,
      "loss": 2.2853,
      "step": 36643
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.8181477785110474,
      "learning_rate": 0.0005633253028262107,
      "loss": 2.0786,
      "step": 36644
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.7317414283752441,
      "learning_rate": 0.0005633233429435752,
      "loss": 2.2654,
      "step": 36645
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6344284415245056,
      "learning_rate": 0.0005633213830119832,
      "loss": 2.3044,
      "step": 36646
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.7058482766151428,
      "learning_rate": 0.0005633194230314346,
      "loss": 2.2042,
      "step": 36647
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.7559038996696472,
      "learning_rate": 0.00056331746300193,
      "loss": 2.2822,
      "step": 36648
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.658298909664154,
      "learning_rate": 0.0005633155029234697,
      "loss": 2.274,
      "step": 36649
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.694388210773468,
      "learning_rate": 0.000563313542796054,
      "loss": 2.1017,
      "step": 36650
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.7413324117660522,
      "learning_rate": 0.0005633115826196836,
      "loss": 2.3097,
      "step": 36651
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.7093188762664795,
      "learning_rate": 0.0005633096223943584,
      "loss": 2.3241,
      "step": 36652
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.7126840949058533,
      "learning_rate": 0.0005633076621200789,
      "loss": 2.3331,
      "step": 36653
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.7357802987098694,
      "learning_rate": 0.0005633057017968456,
      "loss": 2.261,
      "step": 36654
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6753698587417603,
      "learning_rate": 0.0005633037414246589,
      "loss": 2.2009,
      "step": 36655
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.8108394742012024,
      "learning_rate": 0.000563301781003519,
      "loss": 2.171,
      "step": 36656
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6530439853668213,
      "learning_rate": 0.0005632998205334263,
      "loss": 1.9386,
      "step": 36657
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6830167174339294,
      "learning_rate": 0.0005632978600143813,
      "loss": 2.0911,
      "step": 36658
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6608974933624268,
      "learning_rate": 0.0005632958994463841,
      "loss": 2.3011,
      "step": 36659
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.7063028812408447,
      "learning_rate": 0.0005632939388294354,
      "loss": 2.4958,
      "step": 36660
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.8390982151031494,
      "learning_rate": 0.0005632919781635351,
      "loss": 2.3199,
      "step": 36661
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.7140257954597473,
      "learning_rate": 0.0005632900174486841,
      "loss": 2.1888,
      "step": 36662
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.7236959338188171,
      "learning_rate": 0.0005632880566848825,
      "loss": 2.1374,
      "step": 36663
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6764554381370544,
      "learning_rate": 0.0005632860958721306,
      "loss": 2.3433,
      "step": 36664
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6371918320655823,
      "learning_rate": 0.0005632841350104289,
      "loss": 2.3173,
      "step": 36665
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6000369787216187,
      "learning_rate": 0.0005632821740997777,
      "loss": 2.0028,
      "step": 36666
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6609606146812439,
      "learning_rate": 0.0005632802131401774,
      "loss": 2.2634,
      "step": 36667
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.609595000743866,
      "learning_rate": 0.0005632782521316282,
      "loss": 2.3994,
      "step": 36668
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.7069463729858398,
      "learning_rate": 0.0005632762910741308,
      "loss": 2.273,
      "step": 36669
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6343265175819397,
      "learning_rate": 0.0005632743299676852,
      "loss": 2.3025,
      "step": 36670
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6301723718643188,
      "learning_rate": 0.0005632723688122921,
      "loss": 2.3888,
      "step": 36671
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6567074060440063,
      "learning_rate": 0.0005632704076079516,
      "loss": 2.2088,
      "step": 36672
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6967792510986328,
      "learning_rate": 0.0005632684463546641,
      "loss": 2.1554,
      "step": 36673
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6614879965782166,
      "learning_rate": 0.0005632664850524302,
      "loss": 2.3649,
      "step": 36674
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.8427964448928833,
      "learning_rate": 0.0005632645237012499,
      "loss": 2.1352,
      "step": 36675
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6551538705825806,
      "learning_rate": 0.0005632625623011238,
      "loss": 2.1612,
      "step": 36676
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.615760326385498,
      "learning_rate": 0.0005632606008520522,
      "loss": 2.1886,
      "step": 36677
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6527551412582397,
      "learning_rate": 0.0005632586393540356,
      "loss": 2.2324,
      "step": 36678
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.033265233039856,
      "learning_rate": 0.0005632566778070742,
      "loss": 2.305,
      "step": 36679
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6413501501083374,
      "learning_rate": 0.0005632547162111684,
      "loss": 2.3416,
      "step": 36680
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.1407220363616943,
      "learning_rate": 0.0005632527545663186,
      "loss": 2.1313,
      "step": 36681
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6532609462738037,
      "learning_rate": 0.0005632507928725251,
      "loss": 2.0795,
      "step": 36682
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.7046386003494263,
      "learning_rate": 0.0005632488311297884,
      "loss": 2.4315,
      "step": 36683
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.8071029782295227,
      "learning_rate": 0.0005632468693381087,
      "loss": 2.1247,
      "step": 36684
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6768676042556763,
      "learning_rate": 0.0005632449074974864,
      "loss": 2.4528,
      "step": 36685
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.7089820504188538,
      "learning_rate": 0.000563242945607922,
      "loss": 2.4635,
      "step": 36686
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6587494611740112,
      "learning_rate": 0.0005632409836694157,
      "loss": 2.1102,
      "step": 36687
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.7511174082756042,
      "learning_rate": 0.000563239021681968,
      "loss": 2.3401,
      "step": 36688
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6923274993896484,
      "learning_rate": 0.000563237059645579,
      "loss": 2.2945,
      "step": 36689
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.7484356760978699,
      "learning_rate": 0.0005632350975602494,
      "loss": 2.3404,
      "step": 36690
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6801719665527344,
      "learning_rate": 0.0005632331354259795,
      "loss": 2.2326,
      "step": 36691
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.7351812720298767,
      "learning_rate": 0.0005632311732427695,
      "loss": 2.3086,
      "step": 36692
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.7494656443595886,
      "learning_rate": 0.0005632292110106198,
      "loss": 2.6625,
      "step": 36693
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6140223145484924,
      "learning_rate": 0.000563227248729531,
      "loss": 2.3495,
      "step": 36694
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.7705028057098389,
      "learning_rate": 0.0005632252863995032,
      "loss": 2.36,
      "step": 36695
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.7838059663772583,
      "learning_rate": 0.0005632233240205368,
      "loss": 2.2367,
      "step": 36696
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6400755047798157,
      "learning_rate": 0.0005632213615926323,
      "loss": 1.9938,
      "step": 36697
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6790797710418701,
      "learning_rate": 0.0005632193991157899,
      "loss": 2.3807,
      "step": 36698
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.7231727838516235,
      "learning_rate": 0.0005632174365900101,
      "loss": 2.1922,
      "step": 36699
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.7487170100212097,
      "learning_rate": 0.0005632154740152931,
      "loss": 2.4371,
      "step": 36700
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6992782354354858,
      "learning_rate": 0.0005632135113916396,
      "loss": 2.4809,
      "step": 36701
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6217548251152039,
      "learning_rate": 0.0005632115487190495,
      "loss": 2.0756,
      "step": 36702
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6883103847503662,
      "learning_rate": 0.0005632095859975235,
      "loss": 2.4229,
      "step": 36703
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6975172758102417,
      "learning_rate": 0.0005632076232270619,
      "loss": 2.2294,
      "step": 36704
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.7192726135253906,
      "learning_rate": 0.000563205660407665,
      "loss": 2.2959,
      "step": 36705
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6529010534286499,
      "learning_rate": 0.0005632036975393332,
      "loss": 2.3261,
      "step": 36706
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6833510994911194,
      "learning_rate": 0.0005632017346220669,
      "loss": 2.3453,
      "step": 36707
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.8521193265914917,
      "learning_rate": 0.0005631997716558664,
      "loss": 2.5329,
      "step": 36708
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6657468676567078,
      "learning_rate": 0.000563197808640732,
      "loss": 2.1293,
      "step": 36709
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6961477398872375,
      "learning_rate": 0.0005631958455766644,
      "loss": 2.2315,
      "step": 36710
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6776368618011475,
      "learning_rate": 0.0005631938824636636,
      "loss": 2.0727,
      "step": 36711
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6131128668785095,
      "learning_rate": 0.00056319191930173,
      "loss": 2.3505,
      "step": 36712
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.661867618560791,
      "learning_rate": 0.0005631899560908641,
      "loss": 2.4424,
      "step": 36713
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6497253179550171,
      "learning_rate": 0.0005631879928310662,
      "loss": 2.303,
      "step": 36714
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.7232799530029297,
      "learning_rate": 0.0005631860295223367,
      "loss": 2.2949,
      "step": 36715
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6898436546325684,
      "learning_rate": 0.000563184066164676,
      "loss": 2.2238,
      "step": 36716
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.7072188258171082,
      "learning_rate": 0.0005631821027580844,
      "loss": 2.4791,
      "step": 36717
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.654971718788147,
      "learning_rate": 0.0005631801393025623,
      "loss": 2.179,
      "step": 36718
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.686622679233551,
      "learning_rate": 0.00056317817579811,
      "loss": 2.4192,
      "step": 36719
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6912194490432739,
      "learning_rate": 0.000563176212244728,
      "loss": 2.3632,
      "step": 36720
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.7416368722915649,
      "learning_rate": 0.0005631742486424165,
      "loss": 2.3053,
      "step": 36721
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6372824311256409,
      "learning_rate": 0.0005631722849911759,
      "loss": 2.2243,
      "step": 36722
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6955330967903137,
      "learning_rate": 0.0005631703212910067,
      "loss": 2.2336,
      "step": 36723
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.7060268521308899,
      "learning_rate": 0.0005631683575419091,
      "loss": 2.2402,
      "step": 36724
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.7649140357971191,
      "learning_rate": 0.0005631663937438835,
      "loss": 2.2227,
      "step": 36725
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.670632541179657,
      "learning_rate": 0.0005631644298969303,
      "loss": 2.2268,
      "step": 36726
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6887732148170471,
      "learning_rate": 0.0005631624660010499,
      "loss": 2.2352,
      "step": 36727
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6778631806373596,
      "learning_rate": 0.0005631605020562428,
      "loss": 2.3847,
      "step": 36728
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6767047643661499,
      "learning_rate": 0.000563158538062509,
      "loss": 2.3667,
      "step": 36729
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.7414425015449524,
      "learning_rate": 0.000563156574019849,
      "loss": 2.266,
      "step": 36730
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.68790602684021,
      "learning_rate": 0.0005631546099282633,
      "loss": 2.3679,
      "step": 36731
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.7064036130905151,
      "learning_rate": 0.0005631526457877522,
      "loss": 2.2262,
      "step": 36732
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.7669580578804016,
      "learning_rate": 0.0005631506815983161,
      "loss": 2.2238,
      "step": 36733
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6608229279518127,
      "learning_rate": 0.0005631487173599553,
      "loss": 2.3301,
      "step": 36734
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6555101871490479,
      "learning_rate": 0.0005631467530726701,
      "loss": 2.3207,
      "step": 36735
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.7045110464096069,
      "learning_rate": 0.0005631447887364611,
      "loss": 2.4277,
      "step": 36736
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.7019830346107483,
      "learning_rate": 0.0005631428243513284,
      "loss": 2.3359,
      "step": 36737
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.7575980424880981,
      "learning_rate": 0.0005631408599172726,
      "loss": 2.2318,
      "step": 36738
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6509377956390381,
      "learning_rate": 0.0005631388954342938,
      "loss": 2.2565,
      "step": 36739
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.8937901854515076,
      "learning_rate": 0.0005631369309023926,
      "loss": 2.1771,
      "step": 36740
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.7286866307258606,
      "learning_rate": 0.0005631349663215693,
      "loss": 2.3182,
      "step": 36741
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6520659327507019,
      "learning_rate": 0.0005631330016918241,
      "loss": 2.3311,
      "step": 36742
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.7343124151229858,
      "learning_rate": 0.0005631310370131576,
      "loss": 2.098,
      "step": 36743
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.8562471270561218,
      "learning_rate": 0.0005631290722855701,
      "loss": 2.3056,
      "step": 36744
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.7039974927902222,
      "learning_rate": 0.000563127107509062,
      "loss": 2.2283,
      "step": 36745
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6714129447937012,
      "learning_rate": 0.0005631251426836335,
      "loss": 2.3449,
      "step": 36746
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.9294330477714539,
      "learning_rate": 0.0005631231778092851,
      "loss": 2.3491,
      "step": 36747
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6163346171379089,
      "learning_rate": 0.0005631212128860171,
      "loss": 2.2255,
      "step": 36748
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6433631777763367,
      "learning_rate": 0.00056311924791383,
      "loss": 2.4309,
      "step": 36749
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6947631239891052,
      "learning_rate": 0.000563117282892724,
      "loss": 2.2419,
      "step": 36750
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6295453906059265,
      "learning_rate": 0.0005631153178226995,
      "loss": 2.452,
      "step": 36751
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6087015271186829,
      "learning_rate": 0.0005631133527037569,
      "loss": 2.2475,
      "step": 36752
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.7374100685119629,
      "learning_rate": 0.0005631113875358966,
      "loss": 2.4215,
      "step": 36753
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.7173593044281006,
      "learning_rate": 0.0005631094223191189,
      "loss": 2.1014,
      "step": 36754
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6698627471923828,
      "learning_rate": 0.0005631074570534243,
      "loss": 2.3405,
      "step": 36755
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6967988014221191,
      "learning_rate": 0.0005631054917388128,
      "loss": 2.4247,
      "step": 36756
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.8544946908950806,
      "learning_rate": 0.0005631035263752853,
      "loss": 2.1559,
      "step": 36757
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6274371147155762,
      "learning_rate": 0.0005631015609628417,
      "loss": 2.0351,
      "step": 36758
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.7258495688438416,
      "learning_rate": 0.0005630995955014827,
      "loss": 2.3581,
      "step": 36759
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.9490389823913574,
      "learning_rate": 0.0005630976299912084,
      "loss": 2.2423,
      "step": 36760
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.9950833916664124,
      "learning_rate": 0.0005630956644320193,
      "loss": 2.1969,
      "step": 36761
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.703677773475647,
      "learning_rate": 0.0005630936988239158,
      "loss": 2.4412,
      "step": 36762
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.7147214412689209,
      "learning_rate": 0.0005630917331668981,
      "loss": 2.4052,
      "step": 36763
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.691287100315094,
      "learning_rate": 0.0005630897674609669,
      "loss": 2.0992,
      "step": 36764
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6888741254806519,
      "learning_rate": 0.0005630878017061222,
      "loss": 2.1547,
      "step": 36765
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6054742336273193,
      "learning_rate": 0.0005630858359023644,
      "loss": 2.3422,
      "step": 36766
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6244943737983704,
      "learning_rate": 0.0005630838700496942,
      "loss": 2.2468,
      "step": 36767
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.7346753478050232,
      "learning_rate": 0.0005630819041481116,
      "loss": 2.5232,
      "step": 36768
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.7160912156105042,
      "learning_rate": 0.0005630799381976172,
      "loss": 2.1151,
      "step": 36769
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6324069499969482,
      "learning_rate": 0.0005630779721982112,
      "loss": 2.2599,
      "step": 36770
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.8298054337501526,
      "learning_rate": 0.0005630760061498941,
      "loss": 2.4359,
      "step": 36771
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.8037493228912354,
      "learning_rate": 0.0005630740400526661,
      "loss": 2.1942,
      "step": 36772
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.615254819393158,
      "learning_rate": 0.0005630720739065278,
      "loss": 2.2461,
      "step": 36773
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.7977954745292664,
      "learning_rate": 0.0005630701077114793,
      "loss": 2.3653,
      "step": 36774
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6128422617912292,
      "learning_rate": 0.0005630681414675212,
      "loss": 2.0762,
      "step": 36775
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.655007541179657,
      "learning_rate": 0.0005630661751746537,
      "loss": 2.4073,
      "step": 36776
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6400910019874573,
      "learning_rate": 0.0005630642088328774,
      "loss": 2.1246,
      "step": 36777
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6656218767166138,
      "learning_rate": 0.0005630622424421923,
      "loss": 2.4674,
      "step": 36778
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.8091616034507751,
      "learning_rate": 0.000563060276002599,
      "loss": 2.3565,
      "step": 36779
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.7057444453239441,
      "learning_rate": 0.0005630583095140979,
      "loss": 2.2268,
      "step": 36780
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6762486100196838,
      "learning_rate": 0.0005630563429766893,
      "loss": 2.4209,
      "step": 36781
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.8105860948562622,
      "learning_rate": 0.0005630543763903735,
      "loss": 2.3018,
      "step": 36782
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.7437679767608643,
      "learning_rate": 0.000563052409755151,
      "loss": 2.4508,
      "step": 36783
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6466784477233887,
      "learning_rate": 0.000563050443071022,
      "loss": 2.154,
      "step": 36784
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.663904070854187,
      "learning_rate": 0.000563048476337987,
      "loss": 2.5408,
      "step": 36785
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6615020036697388,
      "learning_rate": 0.0005630465095560463,
      "loss": 2.5111,
      "step": 36786
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6501235961914062,
      "learning_rate": 0.0005630445427252004,
      "loss": 2.5317,
      "step": 36787
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6432948708534241,
      "learning_rate": 0.0005630425758454494,
      "loss": 2.3014,
      "step": 36788
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.615104079246521,
      "learning_rate": 0.0005630406089167938,
      "loss": 2.2384,
      "step": 36789
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6576407551765442,
      "learning_rate": 0.0005630386419392342,
      "loss": 2.4362,
      "step": 36790
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6825218200683594,
      "learning_rate": 0.0005630366749127706,
      "loss": 2.2356,
      "step": 36791
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.8375471234321594,
      "learning_rate": 0.0005630347078374035,
      "loss": 2.5426,
      "step": 36792
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6443555355072021,
      "learning_rate": 0.0005630327407131333,
      "loss": 2.1928,
      "step": 36793
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.9519506096839905,
      "learning_rate": 0.0005630307735399603,
      "loss": 2.3075,
      "step": 36794
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.7047451138496399,
      "learning_rate": 0.0005630288063178851,
      "loss": 2.4719,
      "step": 36795
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6302468776702881,
      "learning_rate": 0.0005630268390469078,
      "loss": 2.0508,
      "step": 36796
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.8168497085571289,
      "learning_rate": 0.0005630248717270287,
      "loss": 2.4297,
      "step": 36797
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6082898378372192,
      "learning_rate": 0.0005630229043582485,
      "loss": 2.2501,
      "step": 36798
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.695016622543335,
      "learning_rate": 0.0005630209369405672,
      "loss": 2.2863,
      "step": 36799
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.7005714774131775,
      "learning_rate": 0.0005630189694739856,
      "loss": 2.135,
      "step": 36800
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6681121587753296,
      "learning_rate": 0.0005630170019585035,
      "loss": 2.3754,
      "step": 36801
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.706671953201294,
      "learning_rate": 0.0005630150343941217,
      "loss": 2.2385,
      "step": 36802
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.685766339302063,
      "learning_rate": 0.0005630130667808405,
      "loss": 2.3191,
      "step": 36803
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.7617892026901245,
      "learning_rate": 0.0005630110991186602,
      "loss": 2.4001,
      "step": 36804
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6630953550338745,
      "learning_rate": 0.0005630091314075811,
      "loss": 2.4234,
      "step": 36805
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6698949933052063,
      "learning_rate": 0.0005630071636476036,
      "loss": 2.2518,
      "step": 36806
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.7121171355247498,
      "learning_rate": 0.0005630051958387283,
      "loss": 2.3558,
      "step": 36807
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.697198748588562,
      "learning_rate": 0.0005630032279809553,
      "loss": 2.3022,
      "step": 36808
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.667023241519928,
      "learning_rate": 0.000563001260074285,
      "loss": 2.4394,
      "step": 36809
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6873176693916321,
      "learning_rate": 0.0005629992921187177,
      "loss": 2.2821,
      "step": 36810
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.641878068447113,
      "learning_rate": 0.000562997324114254,
      "loss": 2.3971,
      "step": 36811
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.619231641292572,
      "learning_rate": 0.0005629953560608941,
      "loss": 2.2789,
      "step": 36812
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.7501187324523926,
      "learning_rate": 0.0005629933879586383,
      "loss": 2.3198,
      "step": 36813
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.7084789872169495,
      "learning_rate": 0.0005629914198074872,
      "loss": 2.3738,
      "step": 36814
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6821922063827515,
      "learning_rate": 0.000562989451607441,
      "loss": 2.4287,
      "step": 36815
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6355603933334351,
      "learning_rate": 0.0005629874833585001,
      "loss": 2.2827,
      "step": 36816
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6801495552062988,
      "learning_rate": 0.0005629855150606648,
      "loss": 2.2045,
      "step": 36817
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6583185195922852,
      "learning_rate": 0.0005629835467139355,
      "loss": 2.4272,
      "step": 36818
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6564228534698486,
      "learning_rate": 0.0005629815783183128,
      "loss": 2.2947,
      "step": 36819
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6801841259002686,
      "learning_rate": 0.0005629796098737967,
      "loss": 2.211,
      "step": 36820
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.7327837944030762,
      "learning_rate": 0.0005629776413803877,
      "loss": 2.4626,
      "step": 36821
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.7087026834487915,
      "learning_rate": 0.0005629756728380862,
      "loss": 2.2977,
      "step": 36822
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6451501250267029,
      "learning_rate": 0.0005629737042468926,
      "loss": 2.1938,
      "step": 36823
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6420702934265137,
      "learning_rate": 0.0005629717356068072,
      "loss": 1.9871,
      "step": 36824
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6700690388679504,
      "learning_rate": 0.0005629697669178304,
      "loss": 2.0666,
      "step": 36825
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6743909120559692,
      "learning_rate": 0.0005629677981799626,
      "loss": 2.3825,
      "step": 36826
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6422399282455444,
      "learning_rate": 0.0005629658293932042,
      "loss": 2.2872,
      "step": 36827
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6397433280944824,
      "learning_rate": 0.0005629638605575553,
      "loss": 2.1913,
      "step": 36828
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6040495038032532,
      "learning_rate": 0.0005629618916730165,
      "loss": 2.3321,
      "step": 36829
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.9850736856460571,
      "learning_rate": 0.0005629599227395882,
      "loss": 2.4119,
      "step": 36830
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6614665985107422,
      "learning_rate": 0.0005629579537572706,
      "loss": 2.4378,
      "step": 36831
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.7232536673545837,
      "learning_rate": 0.0005629559847260643,
      "loss": 2.2738,
      "step": 36832
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6933951377868652,
      "learning_rate": 0.0005629540156459694,
      "loss": 2.5101,
      "step": 36833
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.7554787397384644,
      "learning_rate": 0.0005629520465169864,
      "loss": 2.4477,
      "step": 36834
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6875302791595459,
      "learning_rate": 0.0005629500773391156,
      "loss": 2.3186,
      "step": 36835
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.5955039858818054,
      "learning_rate": 0.0005629481081123575,
      "loss": 2.2711,
      "step": 36836
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.7448116540908813,
      "learning_rate": 0.0005629461388367124,
      "loss": 2.3569,
      "step": 36837
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.7149994373321533,
      "learning_rate": 0.0005629441695121806,
      "loss": 2.1755,
      "step": 36838
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6415284276008606,
      "learning_rate": 0.0005629422001387625,
      "loss": 2.4255,
      "step": 36839
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.7281581163406372,
      "learning_rate": 0.0005629402307164585,
      "loss": 2.316,
      "step": 36840
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6659916639328003,
      "learning_rate": 0.0005629382612452689,
      "loss": 2.3829,
      "step": 36841
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6549358367919922,
      "learning_rate": 0.0005629362917251941,
      "loss": 2.1936,
      "step": 36842
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6579662561416626,
      "learning_rate": 0.0005629343221562346,
      "loss": 2.2774,
      "step": 36843
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.7009457945823669,
      "learning_rate": 0.0005629323525383906,
      "loss": 2.1996,
      "step": 36844
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6287349462509155,
      "learning_rate": 0.0005629303828716624,
      "loss": 2.2859,
      "step": 36845
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6745288968086243,
      "learning_rate": 0.0005629284131560507,
      "loss": 2.2113,
      "step": 36846
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6952493786811829,
      "learning_rate": 0.0005629264433915555,
      "loss": 2.28,
      "step": 36847
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.7305325269699097,
      "learning_rate": 0.0005629244735781773,
      "loss": 2.1948,
      "step": 36848
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6800849437713623,
      "learning_rate": 0.0005629225037159165,
      "loss": 2.2206,
      "step": 36849
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6279010772705078,
      "learning_rate": 0.0005629205338047735,
      "loss": 2.487,
      "step": 36850
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.8028319478034973,
      "learning_rate": 0.0005629185638447485,
      "loss": 2.5807,
      "step": 36851
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.7524060010910034,
      "learning_rate": 0.0005629165938358422,
      "loss": 2.3629,
      "step": 36852
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6361504793167114,
      "learning_rate": 0.0005629146237780545,
      "loss": 2.214,
      "step": 36853
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.7305916547775269,
      "learning_rate": 0.0005629126536713861,
      "loss": 2.3809,
      "step": 36854
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.7524386048316956,
      "learning_rate": 0.0005629106835158373,
      "loss": 2.5776,
      "step": 36855
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.7459536790847778,
      "learning_rate": 0.0005629087133114083,
      "loss": 2.3984,
      "step": 36856
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.693475604057312,
      "learning_rate": 0.0005629067430580998,
      "loss": 2.3331,
      "step": 36857
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.7563840746879578,
      "learning_rate": 0.0005629047727559118,
      "loss": 2.391,
      "step": 36858
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6949837803840637,
      "learning_rate": 0.000562902802404845,
      "loss": 2.3487,
      "step": 36859
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.7116804122924805,
      "learning_rate": 0.0005629008320048995,
      "loss": 2.2543,
      "step": 36860
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.7608654499053955,
      "learning_rate": 0.0005628988615560759,
      "loss": 2.2615,
      "step": 36861
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6964634656906128,
      "learning_rate": 0.0005628968910583743,
      "loss": 2.2863,
      "step": 36862
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6151759028434753,
      "learning_rate": 0.0005628949205117952,
      "loss": 2.2988,
      "step": 36863
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6861935257911682,
      "learning_rate": 0.000562892949916339,
      "loss": 2.2828,
      "step": 36864
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6996137499809265,
      "learning_rate": 0.0005628909792720062,
      "loss": 2.5152,
      "step": 36865
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.662205696105957,
      "learning_rate": 0.0005628890085787968,
      "loss": 2.1624,
      "step": 36866
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6920711994171143,
      "learning_rate": 0.0005628870378367115,
      "loss": 2.2387,
      "step": 36867
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6469413042068481,
      "learning_rate": 0.0005628850670457504,
      "loss": 2.3695,
      "step": 36868
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6808059811592102,
      "learning_rate": 0.0005628830962059141,
      "loss": 2.0795,
      "step": 36869
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.7437086701393127,
      "learning_rate": 0.0005628811253172028,
      "loss": 2.4078,
      "step": 36870
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6323073506355286,
      "learning_rate": 0.000562879154379617,
      "loss": 2.1202,
      "step": 36871
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6613857746124268,
      "learning_rate": 0.0005628771833931569,
      "loss": 2.4327,
      "step": 36872
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6728459000587463,
      "learning_rate": 0.0005628752123578232,
      "loss": 2.2758,
      "step": 36873
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.9358309507369995,
      "learning_rate": 0.0005628732412736159,
      "loss": 2.3849,
      "step": 36874
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.653834879398346,
      "learning_rate": 0.0005628712701405354,
      "loss": 2.382,
      "step": 36875
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.7382810115814209,
      "learning_rate": 0.0005628692989585823,
      "loss": 2.3751,
      "step": 36876
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.7101888656616211,
      "learning_rate": 0.0005628673277277567,
      "loss": 2.0997,
      "step": 36877
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6561559438705444,
      "learning_rate": 0.0005628653564480593,
      "loss": 2.3226,
      "step": 36878
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6056394577026367,
      "learning_rate": 0.0005628633851194902,
      "loss": 2.0655,
      "step": 36879
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.850534200668335,
      "learning_rate": 0.0005628614137420497,
      "loss": 2.4018,
      "step": 36880
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6630686521530151,
      "learning_rate": 0.0005628594423157385,
      "loss": 2.3131,
      "step": 36881
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6542278528213501,
      "learning_rate": 0.0005628574708405566,
      "loss": 2.2761,
      "step": 36882
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6686838865280151,
      "learning_rate": 0.0005628554993165047,
      "loss": 2.4163,
      "step": 36883
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.9295037388801575,
      "learning_rate": 0.0005628535277435829,
      "loss": 2.2607,
      "step": 36884
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6363762021064758,
      "learning_rate": 0.0005628515561217916,
      "loss": 2.3968,
      "step": 36885
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6507828831672668,
      "learning_rate": 0.0005628495844511313,
      "loss": 2.2267,
      "step": 36886
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.7616819739341736,
      "learning_rate": 0.0005628476127316023,
      "loss": 2.3839,
      "step": 36887
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.7568260431289673,
      "learning_rate": 0.000562845640963205,
      "loss": 2.3193,
      "step": 36888
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6402776837348938,
      "learning_rate": 0.0005628436691459398,
      "loss": 2.0407,
      "step": 36889
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.7112708687782288,
      "learning_rate": 0.0005628416972798069,
      "loss": 2.4337,
      "step": 36890
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.7075754404067993,
      "learning_rate": 0.0005628397253648067,
      "loss": 2.1561,
      "step": 36891
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.758624255657196,
      "learning_rate": 0.0005628377534009397,
      "loss": 2.0851,
      "step": 36892
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6111050248146057,
      "learning_rate": 0.0005628357813882063,
      "loss": 2.2447,
      "step": 36893
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.7915289998054504,
      "learning_rate": 0.0005628338093266067,
      "loss": 1.9396,
      "step": 36894
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.7418314218521118,
      "learning_rate": 0.0005628318372161412,
      "loss": 2.2044,
      "step": 36895
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6785867810249329,
      "learning_rate": 0.0005628298650568104,
      "loss": 2.2849,
      "step": 36896
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.7269787788391113,
      "learning_rate": 0.0005628278928486146,
      "loss": 2.3009,
      "step": 36897
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6527889370918274,
      "learning_rate": 0.0005628259205915541,
      "loss": 2.4107,
      "step": 36898
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6589279770851135,
      "learning_rate": 0.0005628239482856293,
      "loss": 2.3979,
      "step": 36899
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.7277275919914246,
      "learning_rate": 0.0005628219759308406,
      "loss": 2.4786,
      "step": 36900
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6569832563400269,
      "learning_rate": 0.0005628200035271883,
      "loss": 2.2995,
      "step": 36901
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.7860515117645264,
      "learning_rate": 0.0005628180310746729,
      "loss": 2.3066,
      "step": 36902
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6774383187294006,
      "learning_rate": 0.0005628160585732945,
      "loss": 2.5285,
      "step": 36903
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.8133544325828552,
      "learning_rate": 0.0005628140860230537,
      "loss": 2.2415,
      "step": 36904
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6280348896980286,
      "learning_rate": 0.0005628121134239509,
      "loss": 2.0718,
      "step": 36905
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6344318985939026,
      "learning_rate": 0.0005628101407759862,
      "loss": 2.3665,
      "step": 36906
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6687475442886353,
      "learning_rate": 0.0005628081680791603,
      "loss": 2.4726,
      "step": 36907
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.8250974416732788,
      "learning_rate": 0.0005628061953334734,
      "loss": 2.1335,
      "step": 36908
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.7621241211891174,
      "learning_rate": 0.0005628042225389256,
      "loss": 2.3585,
      "step": 36909
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.8682082295417786,
      "learning_rate": 0.0005628022496955178,
      "loss": 2.3043,
      "step": 36910
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.7466908693313599,
      "learning_rate": 0.00056280027680325,
      "loss": 2.2955,
      "step": 36911
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6523352265357971,
      "learning_rate": 0.0005627983038621227,
      "loss": 2.2563,
      "step": 36912
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.698847234249115,
      "learning_rate": 0.0005627963308721363,
      "loss": 2.4704,
      "step": 36913
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.7422312498092651,
      "learning_rate": 0.0005627943578332909,
      "loss": 2.2689,
      "step": 36914
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6527363657951355,
      "learning_rate": 0.0005627923847455873,
      "loss": 2.0403,
      "step": 36915
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6808061003684998,
      "learning_rate": 0.0005627904116090255,
      "loss": 2.4359,
      "step": 36916
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.7445205450057983,
      "learning_rate": 0.000562788438423606,
      "loss": 2.2667,
      "step": 36917
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6642299294471741,
      "learning_rate": 0.0005627864651893291,
      "loss": 2.2552,
      "step": 36918
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6813169121742249,
      "learning_rate": 0.0005627844919061954,
      "loss": 2.2861,
      "step": 36919
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.7677903771400452,
      "learning_rate": 0.000562782518574205,
      "loss": 2.2665,
      "step": 36920
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.7014808058738708,
      "learning_rate": 0.0005627805451933585,
      "loss": 2.2759,
      "step": 36921
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6606802344322205,
      "learning_rate": 0.000562778571763656,
      "loss": 2.2401,
      "step": 36922
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6476398706436157,
      "learning_rate": 0.000562776598285098,
      "loss": 2.4495,
      "step": 36923
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.7661548852920532,
      "learning_rate": 0.000562774624757685,
      "loss": 2.2102,
      "step": 36924
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6740710139274597,
      "learning_rate": 0.0005627726511814171,
      "loss": 2.1085,
      "step": 36925
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.8126741647720337,
      "learning_rate": 0.0005627706775562949,
      "loss": 2.3089,
      "step": 36926
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.8625802397727966,
      "learning_rate": 0.0005627687038823186,
      "loss": 2.3889,
      "step": 36927
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6188275814056396,
      "learning_rate": 0.0005627667301594887,
      "loss": 2.3589,
      "step": 36928
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.8608431220054626,
      "learning_rate": 0.0005627647563878054,
      "loss": 2.1389,
      "step": 36929
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.7646406292915344,
      "learning_rate": 0.0005627627825672693,
      "loss": 2.4004,
      "step": 36930
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6506785750389099,
      "learning_rate": 0.0005627608086978806,
      "loss": 2.19,
      "step": 36931
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6716992855072021,
      "learning_rate": 0.0005627588347796396,
      "loss": 2.2289,
      "step": 36932
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.7525502443313599,
      "learning_rate": 0.0005627568608125469,
      "loss": 2.3904,
      "step": 36933
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6367920637130737,
      "learning_rate": 0.0005627548867966027,
      "loss": 2.2959,
      "step": 36934
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6357089281082153,
      "learning_rate": 0.0005627529127318074,
      "loss": 2.1459,
      "step": 36935
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6325045824050903,
      "learning_rate": 0.0005627509386181613,
      "loss": 2.3609,
      "step": 36936
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6534104347229004,
      "learning_rate": 0.0005627489644556649,
      "loss": 2.4236,
      "step": 36937
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.7093142867088318,
      "learning_rate": 0.0005627469902443187,
      "loss": 2.3648,
      "step": 36938
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.64570552110672,
      "learning_rate": 0.0005627450159841227,
      "loss": 2.1631,
      "step": 36939
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6827941536903381,
      "learning_rate": 0.0005627430416750773,
      "loss": 2.3448,
      "step": 36940
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6765193343162537,
      "learning_rate": 0.0005627410673171833,
      "loss": 2.3418,
      "step": 36941
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.7233288288116455,
      "learning_rate": 0.0005627390929104406,
      "loss": 2.5126,
      "step": 36942
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.8853988647460938,
      "learning_rate": 0.0005627371184548497,
      "loss": 2.4663,
      "step": 36943
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.7239260077476501,
      "learning_rate": 0.0005627351439504111,
      "loss": 2.2648,
      "step": 36944
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6908420324325562,
      "learning_rate": 0.000562733169397125,
      "loss": 2.5081,
      "step": 36945
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6770761609077454,
      "learning_rate": 0.0005627311947949919,
      "loss": 2.2467,
      "step": 36946
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.7033790946006775,
      "learning_rate": 0.0005627292201440122,
      "loss": 2.2562,
      "step": 36947
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.8347897529602051,
      "learning_rate": 0.000562727245444186,
      "loss": 2.2722,
      "step": 36948
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6642468571662903,
      "learning_rate": 0.000562725270695514,
      "loss": 2.2571,
      "step": 36949
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.8220186829566956,
      "learning_rate": 0.0005627232958979963,
      "loss": 2.2882,
      "step": 36950
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.751023530960083,
      "learning_rate": 0.0005627213210516335,
      "loss": 2.3899,
      "step": 36951
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.9806159138679504,
      "learning_rate": 0.0005627193461564257,
      "loss": 2.1855,
      "step": 36952
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6689879894256592,
      "learning_rate": 0.0005627173712123736,
      "loss": 2.3687,
      "step": 36953
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.7850974798202515,
      "learning_rate": 0.0005627153962194772,
      "loss": 2.2831,
      "step": 36954
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.7243646383285522,
      "learning_rate": 0.0005627134211777371,
      "loss": 2.3451,
      "step": 36955
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.7039268016815186,
      "learning_rate": 0.0005627114460871537,
      "loss": 2.1981,
      "step": 36956
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.7025136351585388,
      "learning_rate": 0.0005627094709477272,
      "loss": 2.2895,
      "step": 36957
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6818183064460754,
      "learning_rate": 0.0005627074957594581,
      "loss": 2.2642,
      "step": 36958
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.9503957033157349,
      "learning_rate": 0.0005627055205223466,
      "loss": 2.3447,
      "step": 36959
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6774488687515259,
      "learning_rate": 0.0005627035452363933,
      "loss": 2.2743,
      "step": 36960
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.7115071415901184,
      "learning_rate": 0.0005627015699015984,
      "loss": 2.2708,
      "step": 36961
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.836621105670929,
      "learning_rate": 0.0005626995945179624,
      "loss": 2.1539,
      "step": 36962
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6569264531135559,
      "learning_rate": 0.0005626976190854855,
      "loss": 2.1855,
      "step": 36963
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.7021500468254089,
      "learning_rate": 0.0005626956436041682,
      "loss": 2.3286,
      "step": 36964
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.8311257362365723,
      "learning_rate": 0.0005626936680740108,
      "loss": 2.3538,
      "step": 36965
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.5742389559745789,
      "learning_rate": 0.0005626916924950137,
      "loss": 2.3338,
      "step": 36966
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6554042100906372,
      "learning_rate": 0.0005626897168671772,
      "loss": 2.2197,
      "step": 36967
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.7189942002296448,
      "learning_rate": 0.0005626877411905019,
      "loss": 2.4886,
      "step": 36968
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6054475903511047,
      "learning_rate": 0.0005626857654649878,
      "loss": 2.1153,
      "step": 36969
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6527659296989441,
      "learning_rate": 0.0005626837896906355,
      "loss": 2.0491,
      "step": 36970
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.5909543633460999,
      "learning_rate": 0.0005626818138674453,
      "loss": 2.1068,
      "step": 36971
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.7533969283103943,
      "learning_rate": 0.0005626798379954177,
      "loss": 2.2217,
      "step": 36972
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.8021429777145386,
      "learning_rate": 0.0005626778620745528,
      "loss": 2.3711,
      "step": 36973
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.7547793984413147,
      "learning_rate": 0.0005626758861048513,
      "loss": 2.4201,
      "step": 36974
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.7599573135375977,
      "learning_rate": 0.0005626739100863134,
      "loss": 2.2232,
      "step": 36975
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.824604868888855,
      "learning_rate": 0.0005626719340189393,
      "loss": 2.3164,
      "step": 36976
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.7016379237174988,
      "learning_rate": 0.0005626699579027297,
      "loss": 2.359,
      "step": 36977
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.7227339744567871,
      "learning_rate": 0.0005626679817376846,
      "loss": 2.3789,
      "step": 36978
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.668668270111084,
      "learning_rate": 0.0005626660055238047,
      "loss": 2.2313,
      "step": 36979
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6763903498649597,
      "learning_rate": 0.0005626640292610901,
      "loss": 2.361,
      "step": 36980
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.8383083939552307,
      "learning_rate": 0.0005626620529495415,
      "loss": 2.417,
      "step": 36981
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6487935185432434,
      "learning_rate": 0.0005626600765891589,
      "loss": 2.4158,
      "step": 36982
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6963415741920471,
      "learning_rate": 0.000562658100179943,
      "loss": 2.2037,
      "step": 36983
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6362873911857605,
      "learning_rate": 0.0005626561237218939,
      "loss": 2.2587,
      "step": 36984
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.7266238331794739,
      "learning_rate": 0.0005626541472150121,
      "loss": 2.1987,
      "step": 36985
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6858068108558655,
      "learning_rate": 0.0005626521706592979,
      "loss": 2.3084,
      "step": 36986
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6855447888374329,
      "learning_rate": 0.0005626501940547517,
      "loss": 2.3291,
      "step": 36987
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.7022820115089417,
      "learning_rate": 0.0005626482174013738,
      "loss": 2.2954,
      "step": 36988
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6800145506858826,
      "learning_rate": 0.0005626462406991648,
      "loss": 2.2468,
      "step": 36989
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.7655016779899597,
      "learning_rate": 0.0005626442639481248,
      "loss": 2.3298,
      "step": 36990
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.7119553089141846,
      "learning_rate": 0.0005626422871482543,
      "loss": 2.2722,
      "step": 36991
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6783682703971863,
      "learning_rate": 0.0005626403102995536,
      "loss": 2.2241,
      "step": 36992
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.7337063550949097,
      "learning_rate": 0.0005626383334020233,
      "loss": 2.3668,
      "step": 36993
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6572455167770386,
      "learning_rate": 0.0005626363564556633,
      "loss": 2.227,
      "step": 36994
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6662794351577759,
      "learning_rate": 0.0005626343794604744,
      "loss": 2.2463,
      "step": 36995
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.7068114280700684,
      "learning_rate": 0.0005626324024164568,
      "loss": 2.4464,
      "step": 36996
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.7372244596481323,
      "learning_rate": 0.0005626304253236109,
      "loss": 2.3381,
      "step": 36997
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.685596764087677,
      "learning_rate": 0.000562628448181937,
      "loss": 2.356,
      "step": 36998
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6625380516052246,
      "learning_rate": 0.0005626264709914354,
      "loss": 2.2515,
      "step": 36999
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.725084662437439,
      "learning_rate": 0.0005626244937521068,
      "loss": 2.4045,
      "step": 37000
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.7249109745025635,
      "learning_rate": 0.0005626225164639512,
      "loss": 2.3316,
      "step": 37001
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.7606365084648132,
      "learning_rate": 0.0005626205391269691,
      "loss": 2.4051,
      "step": 37002
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6152970790863037,
      "learning_rate": 0.000562618561741161,
      "loss": 2.317,
      "step": 37003
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.776779294013977,
      "learning_rate": 0.0005626165843065271,
      "loss": 2.5566,
      "step": 37004
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6137863993644714,
      "learning_rate": 0.0005626146068230678,
      "loss": 2.1803,
      "step": 37005
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6625291705131531,
      "learning_rate": 0.0005626126292907835,
      "loss": 2.2931,
      "step": 37006
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6812633872032166,
      "learning_rate": 0.0005626106517096747,
      "loss": 2.1403,
      "step": 37007
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.7532830238342285,
      "learning_rate": 0.0005626086740797414,
      "loss": 2.292,
      "step": 37008
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6521835923194885,
      "learning_rate": 0.0005626066964009843,
      "loss": 2.29,
      "step": 37009
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6638951897621155,
      "learning_rate": 0.0005626047186734036,
      "loss": 2.3879,
      "step": 37010
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.033260464668274,
      "learning_rate": 0.0005626027408969998,
      "loss": 2.1369,
      "step": 37011
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6377221345901489,
      "learning_rate": 0.0005626007630717732,
      "loss": 2.0984,
      "step": 37012
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.7905413508415222,
      "learning_rate": 0.0005625987851977241,
      "loss": 2.2856,
      "step": 37013
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.1329275369644165,
      "learning_rate": 0.0005625968072748529,
      "loss": 2.2221,
      "step": 37014
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6579122543334961,
      "learning_rate": 0.00056259482930316,
      "loss": 2.2202,
      "step": 37015
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.7853554487228394,
      "learning_rate": 0.0005625928512826458,
      "loss": 2.3911,
      "step": 37016
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.7733492851257324,
      "learning_rate": 0.0005625908732133108,
      "loss": 2.4077,
      "step": 37017
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.8136579990386963,
      "learning_rate": 0.0005625888950951549,
      "loss": 2.1889,
      "step": 37018
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6189915537834167,
      "learning_rate": 0.000562586916928179,
      "loss": 2.1008,
      "step": 37019
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.7619757056236267,
      "learning_rate": 0.0005625849387123831,
      "loss": 2.1778,
      "step": 37020
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.0950405597686768,
      "learning_rate": 0.0005625829604477677,
      "loss": 2.401,
      "step": 37021
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.7141613364219666,
      "learning_rate": 0.0005625809821343333,
      "loss": 2.3644,
      "step": 37022
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.7007049918174744,
      "learning_rate": 0.0005625790037720801,
      "loss": 2.3368,
      "step": 37023
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.9271655678749084,
      "learning_rate": 0.0005625770253610084,
      "loss": 2.353,
      "step": 37024
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6971442103385925,
      "learning_rate": 0.0005625750469011187,
      "loss": 2.5682,
      "step": 37025
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6532210111618042,
      "learning_rate": 0.0005625730683924113,
      "loss": 2.3371,
      "step": 37026
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.7841643691062927,
      "learning_rate": 0.0005625710898348867,
      "loss": 2.3418,
      "step": 37027
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.7722675204277039,
      "learning_rate": 0.0005625691112285452,
      "loss": 2.1845,
      "step": 37028
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.7528517246246338,
      "learning_rate": 0.0005625671325733872,
      "loss": 2.2256,
      "step": 37029
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.7051732540130615,
      "learning_rate": 0.0005625651538694129,
      "loss": 2.2507,
      "step": 37030
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.7684483528137207,
      "learning_rate": 0.0005625631751166228,
      "loss": 2.3223,
      "step": 37031
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.7411359548568726,
      "learning_rate": 0.0005625611963150172,
      "loss": 2.1579,
      "step": 37032
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6815564632415771,
      "learning_rate": 0.0005625592174645967,
      "loss": 2.2624,
      "step": 37033
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.7459393739700317,
      "learning_rate": 0.0005625572385653612,
      "loss": 2.1084,
      "step": 37034
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.7988075017929077,
      "learning_rate": 0.0005625552596173115,
      "loss": 2.3077,
      "step": 37035
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.665229856967926,
      "learning_rate": 0.000562553280620448,
      "loss": 2.3792,
      "step": 37036
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.699097752571106,
      "learning_rate": 0.0005625513015747707,
      "loss": 2.15,
      "step": 37037
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6889110803604126,
      "learning_rate": 0.0005625493224802802,
      "loss": 2.3052,
      "step": 37038
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.7117302417755127,
      "learning_rate": 0.0005625473433369767,
      "loss": 2.3037,
      "step": 37039
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.7606853246688843,
      "learning_rate": 0.0005625453641448609,
      "loss": 2.1874,
      "step": 37040
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6819958090782166,
      "learning_rate": 0.0005625433849039328,
      "loss": 2.489,
      "step": 37041
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6353190541267395,
      "learning_rate": 0.000562541405614193,
      "loss": 2.0866,
      "step": 37042
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6854920387268066,
      "learning_rate": 0.0005625394262756418,
      "loss": 2.5302,
      "step": 37043
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6415756940841675,
      "learning_rate": 0.0005625374468882795,
      "loss": 2.4193,
      "step": 37044
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.9065499305725098,
      "learning_rate": 0.0005625354674521067,
      "loss": 2.4251,
      "step": 37045
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.7000470757484436,
      "learning_rate": 0.0005625334879671235,
      "loss": 2.3142,
      "step": 37046
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.7764001488685608,
      "learning_rate": 0.0005625315084333303,
      "loss": 2.3114,
      "step": 37047
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6764816045761108,
      "learning_rate": 0.0005625295288507275,
      "loss": 2.4349,
      "step": 37048
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6180932521820068,
      "learning_rate": 0.0005625275492193156,
      "loss": 2.3216,
      "step": 37049
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6948026418685913,
      "learning_rate": 0.000562525569539095,
      "loss": 2.2083,
      "step": 37050
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6550835967063904,
      "learning_rate": 0.0005625235898100657,
      "loss": 1.996,
      "step": 37051
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6770547032356262,
      "learning_rate": 0.0005625216100322285,
      "loss": 2.2063,
      "step": 37052
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6803203225135803,
      "learning_rate": 0.0005625196302055834,
      "loss": 2.1768,
      "step": 37053
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.660833477973938,
      "learning_rate": 0.000562517650330131,
      "loss": 2.34,
      "step": 37054
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.665701150894165,
      "learning_rate": 0.0005625156704058716,
      "loss": 2.1922,
      "step": 37055
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6410934329032898,
      "learning_rate": 0.0005625136904328058,
      "loss": 2.3572,
      "step": 37056
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6476294994354248,
      "learning_rate": 0.0005625117104109336,
      "loss": 2.4106,
      "step": 37057
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6786699891090393,
      "learning_rate": 0.0005625097303402554,
      "loss": 2.4034,
      "step": 37058
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6394832730293274,
      "learning_rate": 0.0005625077502207717,
      "loss": 2.1938,
      "step": 37059
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.703650176525116,
      "learning_rate": 0.0005625057700524831,
      "loss": 2.3574,
      "step": 37060
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.7055221796035767,
      "learning_rate": 0.0005625037898353895,
      "loss": 2.3749,
      "step": 37061
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.689507246017456,
      "learning_rate": 0.0005625018095694915,
      "loss": 2.1618,
      "step": 37062
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6541344523429871,
      "learning_rate": 0.0005624998292547896,
      "loss": 2.2638,
      "step": 37063
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6609915494918823,
      "learning_rate": 0.0005624978488912839,
      "loss": 2.2191,
      "step": 37064
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.7317599654197693,
      "learning_rate": 0.000562495868478975,
      "loss": 2.2957,
      "step": 37065
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6615692973136902,
      "learning_rate": 0.000562493888017863,
      "loss": 2.3806,
      "step": 37066
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.67258620262146,
      "learning_rate": 0.0005624919075079486,
      "loss": 2.4659,
      "step": 37067
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6444868445396423,
      "learning_rate": 0.0005624899269492319,
      "loss": 2.2707,
      "step": 37068
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6790850162506104,
      "learning_rate": 0.0005624879463417134,
      "loss": 2.3894,
      "step": 37069
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6896510124206543,
      "learning_rate": 0.0005624859656853935,
      "loss": 2.1831,
      "step": 37070
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6396039128303528,
      "learning_rate": 0.0005624839849802724,
      "loss": 2.3491,
      "step": 37071
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6396558284759521,
      "learning_rate": 0.0005624820042263506,
      "loss": 2.3022,
      "step": 37072
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6810011267662048,
      "learning_rate": 0.0005624800234236285,
      "loss": 2.3129,
      "step": 37073
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.630469799041748,
      "learning_rate": 0.0005624780425721064,
      "loss": 2.1187,
      "step": 37074
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.9851778149604797,
      "learning_rate": 0.0005624760616717846,
      "loss": 2.3534,
      "step": 37075
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6772754192352295,
      "learning_rate": 0.0005624740807226636,
      "loss": 2.2878,
      "step": 37076
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.7098382115364075,
      "learning_rate": 0.0005624720997247436,
      "loss": 2.3344,
      "step": 37077
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6745458841323853,
      "learning_rate": 0.0005624701186780251,
      "loss": 2.3673,
      "step": 37078
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6525108814239502,
      "learning_rate": 0.0005624681375825086,
      "loss": 2.0146,
      "step": 37079
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6344110369682312,
      "learning_rate": 0.0005624661564381943,
      "loss": 2.1731,
      "step": 37080
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.7242334485054016,
      "learning_rate": 0.0005624641752450825,
      "loss": 2.107,
      "step": 37081
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6812325119972229,
      "learning_rate": 0.0005624621940031737,
      "loss": 2.3015,
      "step": 37082
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6608221530914307,
      "learning_rate": 0.0005624602127124682,
      "loss": 2.2341,
      "step": 37083
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.7050170302391052,
      "learning_rate": 0.0005624582313729664,
      "loss": 2.5251,
      "step": 37084
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6432563066482544,
      "learning_rate": 0.0005624562499846687,
      "loss": 2.3157,
      "step": 37085
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.8651770353317261,
      "learning_rate": 0.0005624542685475753,
      "loss": 2.2319,
      "step": 37086
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6118800640106201,
      "learning_rate": 0.0005624522870616869,
      "loss": 2.1668,
      "step": 37087
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.795139491558075,
      "learning_rate": 0.0005624503055270035,
      "loss": 2.5427,
      "step": 37088
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6999680995941162,
      "learning_rate": 0.0005624483239435258,
      "loss": 2.3467,
      "step": 37089
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.7176931500434875,
      "learning_rate": 0.0005624463423112538,
      "loss": 2.4915,
      "step": 37090
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6475757360458374,
      "learning_rate": 0.0005624443606301882,
      "loss": 2.113,
      "step": 37091
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.667858898639679,
      "learning_rate": 0.0005624423789003292,
      "loss": 2.0865,
      "step": 37092
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6630118489265442,
      "learning_rate": 0.0005624403971216772,
      "loss": 2.3532,
      "step": 37093
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6859622001647949,
      "learning_rate": 0.0005624384152942326,
      "loss": 2.3086,
      "step": 37094
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6462097764015198,
      "learning_rate": 0.0005624364334179956,
      "loss": 2.2051,
      "step": 37095
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6966227293014526,
      "learning_rate": 0.0005624344514929669,
      "loss": 2.3073,
      "step": 37096
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6790993809700012,
      "learning_rate": 0.0005624324695191466,
      "loss": 2.2571,
      "step": 37097
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.721348762512207,
      "learning_rate": 0.0005624304874965353,
      "loss": 2.248,
      "step": 37098
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.690743088722229,
      "learning_rate": 0.0005624285054251331,
      "loss": 2.0534,
      "step": 37099
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.657221794128418,
      "learning_rate": 0.0005624265233049405,
      "loss": 2.3491,
      "step": 37100
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.7675933837890625,
      "learning_rate": 0.0005624245411359577,
      "loss": 2.2782,
      "step": 37101
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6950293183326721,
      "learning_rate": 0.0005624225589181854,
      "loss": 2.339,
      "step": 37102
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.7243102192878723,
      "learning_rate": 0.0005624205766516237,
      "loss": 2.0023,
      "step": 37103
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.8383566737174988,
      "learning_rate": 0.0005624185943362732,
      "loss": 2.2131,
      "step": 37104
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6360610127449036,
      "learning_rate": 0.000562416611972134,
      "loss": 2.4681,
      "step": 37105
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.7387647032737732,
      "learning_rate": 0.0005624146295592066,
      "loss": 2.2909,
      "step": 37106
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6650710105895996,
      "learning_rate": 0.0005624126470974915,
      "loss": 2.3466,
      "step": 37107
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6889503598213196,
      "learning_rate": 0.0005624106645869888,
      "loss": 2.3032,
      "step": 37108
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6610259413719177,
      "learning_rate": 0.0005624086820276991,
      "loss": 2.367,
      "step": 37109
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6618399620056152,
      "learning_rate": 0.0005624066994196226,
      "loss": 1.9129,
      "step": 37110
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.7642412185668945,
      "learning_rate": 0.0005624047167627598,
      "loss": 2.2643,
      "step": 37111
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6692593693733215,
      "learning_rate": 0.0005624027340571109,
      "loss": 2.4194,
      "step": 37112
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6470916271209717,
      "learning_rate": 0.0005624007513026765,
      "loss": 2.2967,
      "step": 37113
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.664104163646698,
      "learning_rate": 0.0005623987684994569,
      "loss": 2.4021,
      "step": 37114
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.7732409834861755,
      "learning_rate": 0.0005623967856474522,
      "loss": 2.2608,
      "step": 37115
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6492826342582703,
      "learning_rate": 0.0005623948027466631,
      "loss": 2.2338,
      "step": 37116
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.7051246762275696,
      "learning_rate": 0.0005623928197970899,
      "loss": 2.2686,
      "step": 37117
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.7202702760696411,
      "learning_rate": 0.0005623908367987329,
      "loss": 2.1757,
      "step": 37118
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.8162990212440491,
      "learning_rate": 0.0005623888537515925,
      "loss": 2.4869,
      "step": 37119
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6762630343437195,
      "learning_rate": 0.000562386870655669,
      "loss": 2.2804,
      "step": 37120
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.7981861233711243,
      "learning_rate": 0.0005623848875109628,
      "loss": 2.4671,
      "step": 37121
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.7739861011505127,
      "learning_rate": 0.0005623829043174744,
      "loss": 2.036,
      "step": 37122
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.8044999837875366,
      "learning_rate": 0.000562380921075204,
      "loss": 2.1961,
      "step": 37123
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.7839338183403015,
      "learning_rate": 0.000562378937784152,
      "loss": 2.2095,
      "step": 37124
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.8035277128219604,
      "learning_rate": 0.0005623769544443189,
      "loss": 2.2366,
      "step": 37125
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.7231222987174988,
      "learning_rate": 0.000562374971055705,
      "loss": 2.0921,
      "step": 37126
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6440231800079346,
      "learning_rate": 0.0005623729876183105,
      "loss": 2.3668,
      "step": 37127
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6779725551605225,
      "learning_rate": 0.0005623710041321359,
      "loss": 2.3252,
      "step": 37128
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.7737768888473511,
      "learning_rate": 0.0005623690205971817,
      "loss": 2.1323,
      "step": 37129
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6449865698814392,
      "learning_rate": 0.000562367037013448,
      "loss": 2.2429,
      "step": 37130
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.7098497748374939,
      "learning_rate": 0.0005623650533809355,
      "loss": 2.5508,
      "step": 37131
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.7891730070114136,
      "learning_rate": 0.0005623630696996444,
      "loss": 2.3138,
      "step": 37132
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.662699818611145,
      "learning_rate": 0.0005623610859695748,
      "loss": 2.1696,
      "step": 37133
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6724075675010681,
      "learning_rate": 0.0005623591021907275,
      "loss": 2.2011,
      "step": 37134
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6946178674697876,
      "learning_rate": 0.0005623571183631026,
      "loss": 2.3145,
      "step": 37135
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6794362664222717,
      "learning_rate": 0.0005623551344867007,
      "loss": 2.3201,
      "step": 37136
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.7467452883720398,
      "learning_rate": 0.0005623531505615219,
      "loss": 2.2611,
      "step": 37137
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.8849288821220398,
      "learning_rate": 0.0005623511665875668,
      "loss": 2.2659,
      "step": 37138
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6803920865058899,
      "learning_rate": 0.0005623491825648356,
      "loss": 2.2543,
      "step": 37139
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6976219415664673,
      "learning_rate": 0.0005623471984933287,
      "loss": 2.3531,
      "step": 37140
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6727557182312012,
      "learning_rate": 0.0005623452143730466,
      "loss": 2.3465,
      "step": 37141
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6800811290740967,
      "learning_rate": 0.0005623432302039894,
      "loss": 2.2167,
      "step": 37142
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6696632504463196,
      "learning_rate": 0.0005623412459861579,
      "loss": 1.9698,
      "step": 37143
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6822752952575684,
      "learning_rate": 0.0005623392617195521,
      "loss": 2.3053,
      "step": 37144
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6511951088905334,
      "learning_rate": 0.0005623372774041724,
      "loss": 2.1897,
      "step": 37145
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6036028861999512,
      "learning_rate": 0.0005623352930400193,
      "loss": 2.2058,
      "step": 37146
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.5826804041862488,
      "learning_rate": 0.0005623333086270931,
      "loss": 2.2049,
      "step": 37147
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6360996961593628,
      "learning_rate": 0.0005623313241653942,
      "loss": 2.0096,
      "step": 37148
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6892315149307251,
      "learning_rate": 0.000562329339654923,
      "loss": 2.2388,
      "step": 37149
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.7295739054679871,
      "learning_rate": 0.0005623273550956797,
      "loss": 2.0712,
      "step": 37150
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6787919402122498,
      "learning_rate": 0.000562325370487665,
      "loss": 2.4462,
      "step": 37151
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.7197316288948059,
      "learning_rate": 0.0005623233858308788,
      "loss": 2.231,
      "step": 37152
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6537156701087952,
      "learning_rate": 0.0005623214011253219,
      "loss": 2.3226,
      "step": 37153
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.7050217986106873,
      "learning_rate": 0.0005623194163709945,
      "loss": 2.4363,
      "step": 37154
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.7318947911262512,
      "learning_rate": 0.000562317431567897,
      "loss": 2.2838,
      "step": 37155
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6980078816413879,
      "learning_rate": 0.0005623154467160296,
      "loss": 2.4076,
      "step": 37156
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.68771892786026,
      "learning_rate": 0.0005623134618153929,
      "loss": 2.4619,
      "step": 37157
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6353570222854614,
      "learning_rate": 0.0005623114768659871,
      "loss": 2.4273,
      "step": 37158
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.7010742425918579,
      "learning_rate": 0.0005623094918678127,
      "loss": 2.3883,
      "step": 37159
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6289911270141602,
      "learning_rate": 0.0005623075068208701,
      "loss": 2.4198,
      "step": 37160
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.7845630645751953,
      "learning_rate": 0.0005623055217251595,
      "loss": 2.3832,
      "step": 37161
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6303892135620117,
      "learning_rate": 0.0005623035365806814,
      "loss": 2.3303,
      "step": 37162
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.7486765384674072,
      "learning_rate": 0.000562301551387436,
      "loss": 2.042,
      "step": 37163
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6205809712409973,
      "learning_rate": 0.0005622995661454239,
      "loss": 2.2052,
      "step": 37164
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6893899440765381,
      "learning_rate": 0.0005622975808546453,
      "loss": 2.1293,
      "step": 37165
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.7985210418701172,
      "learning_rate": 0.0005622955955151006,
      "loss": 2.2647,
      "step": 37166
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6677672266960144,
      "learning_rate": 0.0005622936101267904,
      "loss": 2.3109,
      "step": 37167
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6507992744445801,
      "learning_rate": 0.0005622916246897146,
      "loss": 2.4017,
      "step": 37168
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.7754212617874146,
      "learning_rate": 0.000562289639203874,
      "loss": 2.2369,
      "step": 37169
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.725153923034668,
      "learning_rate": 0.0005622876536692687,
      "loss": 2.1986,
      "step": 37170
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6722660064697266,
      "learning_rate": 0.0005622856680858993,
      "loss": 2.2964,
      "step": 37171
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.7795050144195557,
      "learning_rate": 0.0005622836824537659,
      "loss": 2.2871,
      "step": 37172
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6567238569259644,
      "learning_rate": 0.0005622816967728691,
      "loss": 2.3268,
      "step": 37173
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6541146636009216,
      "learning_rate": 0.0005622797110432092,
      "loss": 2.3541,
      "step": 37174
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6790732145309448,
      "learning_rate": 0.0005622777252647865,
      "loss": 2.3743,
      "step": 37175
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.7104918956756592,
      "learning_rate": 0.0005622757394376015,
      "loss": 2.278,
      "step": 37176
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.7509050369262695,
      "learning_rate": 0.0005622737535616543,
      "loss": 2.1532,
      "step": 37177
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6870271563529968,
      "learning_rate": 0.0005622717676369457,
      "loss": 2.2425,
      "step": 37178
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.7406368255615234,
      "learning_rate": 0.0005622697816634756,
      "loss": 2.6124,
      "step": 37179
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.7922090291976929,
      "learning_rate": 0.0005622677956412448,
      "loss": 2.2176,
      "step": 37180
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6482413411140442,
      "learning_rate": 0.0005622658095702533,
      "loss": 2.106,
      "step": 37181
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6446921229362488,
      "learning_rate": 0.0005622638234505018,
      "loss": 2.2071,
      "step": 37182
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.7072780728340149,
      "learning_rate": 0.0005622618372819904,
      "loss": 2.2879,
      "step": 37183
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6457874774932861,
      "learning_rate": 0.0005622598510647196,
      "loss": 2.1392,
      "step": 37184
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6720286011695862,
      "learning_rate": 0.0005622578647986897,
      "loss": 2.2735,
      "step": 37185
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.7120798826217651,
      "learning_rate": 0.0005622558784839012,
      "loss": 2.4159,
      "step": 37186
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6118577718734741,
      "learning_rate": 0.0005622538921203543,
      "loss": 2.2249,
      "step": 37187
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6379430294036865,
      "learning_rate": 0.0005622519057080494,
      "loss": 2.3151,
      "step": 37188
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6587619781494141,
      "learning_rate": 0.000562249919246987,
      "loss": 2.2352,
      "step": 37189
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.647095799446106,
      "learning_rate": 0.0005622479327371674,
      "loss": 2.1889,
      "step": 37190
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.65053790807724,
      "learning_rate": 0.0005622459461785909,
      "loss": 2.2427,
      "step": 37191
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6891515254974365,
      "learning_rate": 0.0005622439595712581,
      "loss": 2.176,
      "step": 37192
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6615890264511108,
      "learning_rate": 0.0005622419729151691,
      "loss": 2.5199,
      "step": 37193
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.7240766286849976,
      "learning_rate": 0.0005622399862103243,
      "loss": 2.1192,
      "step": 37194
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.7003827095031738,
      "learning_rate": 0.0005622379994567241,
      "loss": 2.3588,
      "step": 37195
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.7463648319244385,
      "learning_rate": 0.000562236012654369,
      "loss": 2.2753,
      "step": 37196
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.7649025321006775,
      "learning_rate": 0.0005622340258032594,
      "loss": 2.0743,
      "step": 37197
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6902799010276794,
      "learning_rate": 0.0005622320389033953,
      "loss": 2.2962,
      "step": 37198
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6820987462997437,
      "learning_rate": 0.0005622300519547774,
      "loss": 2.2722,
      "step": 37199
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6815254092216492,
      "learning_rate": 0.000562228064957406,
      "loss": 2.4002,
      "step": 37200
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6312999725341797,
      "learning_rate": 0.0005622260779112815,
      "loss": 2.3507,
      "step": 37201
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.650493860244751,
      "learning_rate": 0.0005622240908164042,
      "loss": 2.4054,
      "step": 37202
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.7733779549598694,
      "learning_rate": 0.0005622221036727744,
      "loss": 2.2571,
      "step": 37203
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.7901785373687744,
      "learning_rate": 0.0005622201164803926,
      "loss": 2.2291,
      "step": 37204
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.7405014634132385,
      "learning_rate": 0.0005622181292392592,
      "loss": 2.1713,
      "step": 37205
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.703124463558197,
      "learning_rate": 0.0005622161419493745,
      "loss": 2.2646,
      "step": 37206
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6649913191795349,
      "learning_rate": 0.0005622141546107388,
      "loss": 2.4131,
      "step": 37207
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6801469326019287,
      "learning_rate": 0.0005622121672233525,
      "loss": 2.2952,
      "step": 37208
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6803823113441467,
      "learning_rate": 0.0005622101797872161,
      "loss": 2.2211,
      "step": 37209
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6614253520965576,
      "learning_rate": 0.0005622081923023299,
      "loss": 2.0523,
      "step": 37210
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6936367750167847,
      "learning_rate": 0.0005622062047686942,
      "loss": 2.4039,
      "step": 37211
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.9722142219543457,
      "learning_rate": 0.0005622042171863094,
      "loss": 2.3292,
      "step": 37212
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.7109942436218262,
      "learning_rate": 0.0005622022295551759,
      "loss": 2.1456,
      "step": 37213
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.7140098810195923,
      "learning_rate": 0.000562200241875294,
      "loss": 2.4599,
      "step": 37214
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.606420636177063,
      "learning_rate": 0.0005621982541466642,
      "loss": 2.2172,
      "step": 37215
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.5947619676589966,
      "learning_rate": 0.0005621962663692868,
      "loss": 2.2692,
      "step": 37216
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6850314736366272,
      "learning_rate": 0.0005621942785431622,
      "loss": 2.1825,
      "step": 37217
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.7049466371536255,
      "learning_rate": 0.0005621922906682907,
      "loss": 2.24,
      "step": 37218
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.7781533002853394,
      "learning_rate": 0.0005621903027446726,
      "loss": 2.1269,
      "step": 37219
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.7608000040054321,
      "learning_rate": 0.0005621883147723085,
      "loss": 2.31,
      "step": 37220
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6567938923835754,
      "learning_rate": 0.0005621863267511986,
      "loss": 2.2945,
      "step": 37221
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.8094834089279175,
      "learning_rate": 0.0005621843386813434,
      "loss": 2.1923,
      "step": 37222
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.751168966293335,
      "learning_rate": 0.0005621823505627429,
      "loss": 2.3092,
      "step": 37223
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.8826397657394409,
      "learning_rate": 0.0005621803623953981,
      "loss": 2.2048,
      "step": 37224
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.8791239857673645,
      "learning_rate": 0.0005621783741793088,
      "loss": 2.3744,
      "step": 37225
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.7138242125511169,
      "learning_rate": 0.0005621763859144757,
      "loss": 2.127,
      "step": 37226
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6612316966056824,
      "learning_rate": 0.0005621743976008991,
      "loss": 2.4037,
      "step": 37227
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.7117884159088135,
      "learning_rate": 0.0005621724092385791,
      "loss": 2.3801,
      "step": 37228
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.7385649681091309,
      "learning_rate": 0.0005621704208275165,
      "loss": 2.183,
      "step": 37229
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.8543297052383423,
      "learning_rate": 0.0005621684323677114,
      "loss": 2.2658,
      "step": 37230
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6928002834320068,
      "learning_rate": 0.0005621664438591643,
      "loss": 2.4268,
      "step": 37231
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6800654530525208,
      "learning_rate": 0.0005621644553018754,
      "loss": 2.2512,
      "step": 37232
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.8213757276535034,
      "learning_rate": 0.0005621624666958453,
      "loss": 2.4636,
      "step": 37233
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.7452155947685242,
      "learning_rate": 0.0005621604780410742,
      "loss": 2.2813,
      "step": 37234
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6406759023666382,
      "learning_rate": 0.0005621584893375625,
      "loss": 2.2233,
      "step": 37235
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.7393092513084412,
      "learning_rate": 0.0005621565005853106,
      "loss": 2.1183,
      "step": 37236
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.7966977953910828,
      "learning_rate": 0.0005621545117843188,
      "loss": 2.0948,
      "step": 37237
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6133833527565002,
      "learning_rate": 0.0005621525229345877,
      "loss": 2.2949,
      "step": 37238
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6856333613395691,
      "learning_rate": 0.0005621505340361174,
      "loss": 2.2324,
      "step": 37239
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6932956576347351,
      "learning_rate": 0.0005621485450889083,
      "loss": 2.1837,
      "step": 37240
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.7745771408081055,
      "learning_rate": 0.0005621465560929609,
      "loss": 2.152,
      "step": 37241
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6902766227722168,
      "learning_rate": 0.0005621445670482755,
      "loss": 2.2136,
      "step": 37242
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.7946873307228088,
      "learning_rate": 0.0005621425779548526,
      "loss": 2.2088,
      "step": 37243
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.7164348363876343,
      "learning_rate": 0.0005621405888126922,
      "loss": 2.1785,
      "step": 37244
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6855255961418152,
      "learning_rate": 0.000562138599621795,
      "loss": 2.431,
      "step": 37245
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.7359362840652466,
      "learning_rate": 0.0005621366103821615,
      "loss": 2.4391,
      "step": 37246
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.7173928618431091,
      "learning_rate": 0.0005621346210937916,
      "loss": 2.2528,
      "step": 37247
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.644781768321991,
      "learning_rate": 0.0005621326317566859,
      "loss": 2.4187,
      "step": 37248
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6834074854850769,
      "learning_rate": 0.000562130642370845,
      "loss": 2.5407,
      "step": 37249
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6891473531723022,
      "learning_rate": 0.0005621286529362689,
      "loss": 2.2266,
      "step": 37250
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6816656589508057,
      "learning_rate": 0.0005621266634529583,
      "loss": 2.3519,
      "step": 37251
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6486101150512695,
      "learning_rate": 0.0005621246739209132,
      "loss": 2.3451,
      "step": 37252
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.655600905418396,
      "learning_rate": 0.0005621226843401343,
      "loss": 2.1335,
      "step": 37253
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6812199354171753,
      "learning_rate": 0.0005621206947106219,
      "loss": 2.3132,
      "step": 37254
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.7820536494255066,
      "learning_rate": 0.0005621187050323762,
      "loss": 2.1962,
      "step": 37255
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6802406907081604,
      "learning_rate": 0.0005621167153053978,
      "loss": 2.3735,
      "step": 37256
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6935713291168213,
      "learning_rate": 0.0005621147255296868,
      "loss": 2.2857,
      "step": 37257
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6859859824180603,
      "learning_rate": 0.0005621127357052438,
      "loss": 2.2425,
      "step": 37258
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.7025429606437683,
      "learning_rate": 0.0005621107458320691,
      "loss": 2.5003,
      "step": 37259
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6696215867996216,
      "learning_rate": 0.0005621087559101631,
      "loss": 2.2945,
      "step": 37260
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.7607541084289551,
      "learning_rate": 0.0005621067659395261,
      "loss": 2.243,
      "step": 37261
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6769835352897644,
      "learning_rate": 0.0005621047759201585,
      "loss": 2.1902,
      "step": 37262
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6656277179718018,
      "learning_rate": 0.0005621027858520607,
      "loss": 2.2992,
      "step": 37263
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.698469877243042,
      "learning_rate": 0.0005621007957352331,
      "loss": 2.2607,
      "step": 37264
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6977904438972473,
      "learning_rate": 0.0005620988055696759,
      "loss": 2.2155,
      "step": 37265
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6360577344894409,
      "learning_rate": 0.0005620968153553895,
      "loss": 2.3306,
      "step": 37266
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6745490431785583,
      "learning_rate": 0.0005620948250923746,
      "loss": 2.1938,
      "step": 37267
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.706531286239624,
      "learning_rate": 0.0005620928347806313,
      "loss": 2.261,
      "step": 37268
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.672453761100769,
      "learning_rate": 0.0005620908444201597,
      "loss": 2.3943,
      "step": 37269
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6575247645378113,
      "learning_rate": 0.0005620888540109608,
      "loss": 2.375,
      "step": 37270
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6154513955116272,
      "learning_rate": 0.0005620868635530345,
      "loss": 2.3152,
      "step": 37271
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.636638343334198,
      "learning_rate": 0.0005620848730463812,
      "loss": 2.3236,
      "step": 37272
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.7454746961593628,
      "learning_rate": 0.0005620828824910015,
      "loss": 2.2426,
      "step": 37273
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6202527284622192,
      "learning_rate": 0.0005620808918868957,
      "loss": 2.2104,
      "step": 37274
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6460668444633484,
      "learning_rate": 0.000562078901234064,
      "loss": 1.9831,
      "step": 37275
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6743189096450806,
      "learning_rate": 0.0005620769105325069,
      "loss": 2.355,
      "step": 37276
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6699526906013489,
      "learning_rate": 0.0005620749197822247,
      "loss": 2.3294,
      "step": 37277
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6733719706535339,
      "learning_rate": 0.000562072928983218,
      "loss": 2.2789,
      "step": 37278
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.7287729382514954,
      "learning_rate": 0.0005620709381354868,
      "loss": 2.3101,
      "step": 37279
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6938009858131409,
      "learning_rate": 0.0005620689472390318,
      "loss": 2.3804,
      "step": 37280
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6690453886985779,
      "learning_rate": 0.0005620669562938532,
      "loss": 2.3835,
      "step": 37281
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6512295007705688,
      "learning_rate": 0.0005620649652999514,
      "loss": 2.2705,
      "step": 37282
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6095039248466492,
      "learning_rate": 0.0005620629742573268,
      "loss": 2.1926,
      "step": 37283
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.7041597366333008,
      "learning_rate": 0.0005620609831659798,
      "loss": 2.2595,
      "step": 37284
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6268837451934814,
      "learning_rate": 0.0005620589920259106,
      "loss": 2.3037,
      "step": 37285
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.7085612416267395,
      "learning_rate": 0.0005620570008371198,
      "loss": 2.2364,
      "step": 37286
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.697486937046051,
      "learning_rate": 0.0005620550095996075,
      "loss": 2.2802,
      "step": 37287
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6709431409835815,
      "learning_rate": 0.0005620530183133743,
      "loss": 2.4258,
      "step": 37288
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6939753890037537,
      "learning_rate": 0.0005620510269784205,
      "loss": 2.4428,
      "step": 37289
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6350076794624329,
      "learning_rate": 0.0005620490355947466,
      "loss": 2.0948,
      "step": 37290
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.7085374593734741,
      "learning_rate": 0.0005620470441623526,
      "loss": 2.318,
      "step": 37291
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6702136397361755,
      "learning_rate": 0.0005620450526812393,
      "loss": 2.2415,
      "step": 37292
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.7203327417373657,
      "learning_rate": 0.0005620430611514068,
      "loss": 1.8958,
      "step": 37293
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6562618613243103,
      "learning_rate": 0.0005620410695728555,
      "loss": 2.4603,
      "step": 37294
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6941836476325989,
      "learning_rate": 0.0005620390779455859,
      "loss": 2.4572,
      "step": 37295
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6733242869377136,
      "learning_rate": 0.0005620370862695983,
      "loss": 2.2867,
      "step": 37296
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.7204827070236206,
      "learning_rate": 0.000562035094544893,
      "loss": 2.0577,
      "step": 37297
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6685001850128174,
      "learning_rate": 0.0005620331027714705,
      "loss": 2.293,
      "step": 37298
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.7467536330223083,
      "learning_rate": 0.000562031110949331,
      "loss": 2.1207,
      "step": 37299
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.9278137683868408,
      "learning_rate": 0.000562029119078475,
      "loss": 2.493,
      "step": 37300
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6358217000961304,
      "learning_rate": 0.0005620271271589029,
      "loss": 2.4157,
      "step": 37301
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.7369260191917419,
      "learning_rate": 0.000562025135190615,
      "loss": 2.4257,
      "step": 37302
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.8285662531852722,
      "learning_rate": 0.0005620231431736116,
      "loss": 2.429,
      "step": 37303
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6826592087745667,
      "learning_rate": 0.0005620211511078932,
      "loss": 2.2975,
      "step": 37304
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.7741941213607788,
      "learning_rate": 0.0005620191589934601,
      "loss": 2.1201,
      "step": 37305
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6739228367805481,
      "learning_rate": 0.0005620171668303127,
      "loss": 2.1384,
      "step": 37306
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6584441661834717,
      "learning_rate": 0.0005620151746184514,
      "loss": 2.1912,
      "step": 37307
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.7208250761032104,
      "learning_rate": 0.0005620131823578765,
      "loss": 2.2799,
      "step": 37308
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.7008322477340698,
      "learning_rate": 0.0005620111900485884,
      "loss": 2.3497,
      "step": 37309
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6611384749412537,
      "learning_rate": 0.0005620091976905876,
      "loss": 2.2188,
      "step": 37310
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6432989239692688,
      "learning_rate": 0.0005620072052838742,
      "loss": 2.011,
      "step": 37311
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6261820197105408,
      "learning_rate": 0.0005620052128284487,
      "loss": 2.1952,
      "step": 37312
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.7285248637199402,
      "learning_rate": 0.0005620032203243116,
      "loss": 2.1992,
      "step": 37313
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6642245650291443,
      "learning_rate": 0.0005620012277714631,
      "loss": 2.2032,
      "step": 37314
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.5935694575309753,
      "learning_rate": 0.0005619992351699035,
      "loss": 2.0841,
      "step": 37315
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.7417163252830505,
      "learning_rate": 0.0005619972425196335,
      "loss": 2.3116,
      "step": 37316
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.8377932906150818,
      "learning_rate": 0.0005619952498206532,
      "loss": 2.3292,
      "step": 37317
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6196041703224182,
      "learning_rate": 0.000561993257072963,
      "loss": 2.4363,
      "step": 37318
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.707838237285614,
      "learning_rate": 0.0005619912642765633,
      "loss": 2.3217,
      "step": 37319
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.7114533185958862,
      "learning_rate": 0.0005619892714314545,
      "loss": 2.3641,
      "step": 37320
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.7148754000663757,
      "learning_rate": 0.000561987278537637,
      "loss": 2.2421,
      "step": 37321
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6876563429832458,
      "learning_rate": 0.000561985285595111,
      "loss": 2.3308,
      "step": 37322
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6953153014183044,
      "learning_rate": 0.0005619832926038771,
      "loss": 2.6216,
      "step": 37323
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.7078351378440857,
      "learning_rate": 0.0005619812995639355,
      "loss": 2.5283,
      "step": 37324
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.7527421116828918,
      "learning_rate": 0.0005619793064752867,
      "loss": 2.5069,
      "step": 37325
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.7441895008087158,
      "learning_rate": 0.000561977313337931,
      "loss": 2.1847,
      "step": 37326
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.728201150894165,
      "learning_rate": 0.0005619753201518687,
      "loss": 2.2528,
      "step": 37327
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6484631896018982,
      "learning_rate": 0.0005619733269171003,
      "loss": 2.0667,
      "step": 37328
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.7473706603050232,
      "learning_rate": 0.000561971333633626,
      "loss": 2.0969,
      "step": 37329
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6377789974212646,
      "learning_rate": 0.0005619693403014465,
      "loss": 2.5272,
      "step": 37330
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6348448395729065,
      "learning_rate": 0.0005619673469205617,
      "loss": 2.2481,
      "step": 37331
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.8763612508773804,
      "learning_rate": 0.0005619653534909724,
      "loss": 2.1672,
      "step": 37332
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6052632331848145,
      "learning_rate": 0.0005619633600126788,
      "loss": 2.4851,
      "step": 37333
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6882766485214233,
      "learning_rate": 0.0005619613664856811,
      "loss": 2.0406,
      "step": 37334
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.7372097373008728,
      "learning_rate": 0.0005619593729099799,
      "loss": 2.2121,
      "step": 37335
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.678242564201355,
      "learning_rate": 0.0005619573792855756,
      "loss": 2.2703,
      "step": 37336
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6461930274963379,
      "learning_rate": 0.0005619553856124684,
      "loss": 2.3443,
      "step": 37337
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.9294577836990356,
      "learning_rate": 0.0005619533918906588,
      "loss": 2.3,
      "step": 37338
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6850562691688538,
      "learning_rate": 0.0005619513981201472,
      "loss": 2.397,
      "step": 37339
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6989485025405884,
      "learning_rate": 0.0005619494043009337,
      "loss": 2.3795,
      "step": 37340
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.7207039594650269,
      "learning_rate": 0.000561947410433019,
      "loss": 2.1623,
      "step": 37341
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.652633011341095,
      "learning_rate": 0.0005619454165164033,
      "loss": 2.3735,
      "step": 37342
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.7650569677352905,
      "learning_rate": 0.000561943422551087,
      "loss": 2.3036,
      "step": 37343
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6453307867050171,
      "learning_rate": 0.0005619414285370705,
      "loss": 2.3955,
      "step": 37344
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6125298738479614,
      "learning_rate": 0.0005619394344743541,
      "loss": 2.4122,
      "step": 37345
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6701716780662537,
      "learning_rate": 0.0005619374403629382,
      "loss": 2.1841,
      "step": 37346
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.708441972732544,
      "learning_rate": 0.0005619354462028232,
      "loss": 2.3111,
      "step": 37347
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.8129386305809021,
      "learning_rate": 0.0005619334519940094,
      "loss": 2.3224,
      "step": 37348
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6973317265510559,
      "learning_rate": 0.0005619314577364974,
      "loss": 2.2639,
      "step": 37349
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6720190644264221,
      "learning_rate": 0.0005619294634302872,
      "loss": 2.0411,
      "step": 37350
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6201164722442627,
      "learning_rate": 0.0005619274690753795,
      "loss": 2.3939,
      "step": 37351
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6818984150886536,
      "learning_rate": 0.0005619254746717744,
      "loss": 2.1797,
      "step": 37352
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.7103416323661804,
      "learning_rate": 0.0005619234802194727,
      "loss": 2.3442,
      "step": 37353
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6301873922348022,
      "learning_rate": 0.0005619214857184742,
      "loss": 2.437,
      "step": 37354
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6786655783653259,
      "learning_rate": 0.0005619194911687796,
      "loss": 2.4983,
      "step": 37355
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6123299598693848,
      "learning_rate": 0.0005619174965703893,
      "loss": 2.2196,
      "step": 37356
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.7901924848556519,
      "learning_rate": 0.0005619155019233036,
      "loss": 2.4672,
      "step": 37357
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.7687651515007019,
      "learning_rate": 0.0005619135072275227,
      "loss": 2.3028,
      "step": 37358
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.7067634463310242,
      "learning_rate": 0.0005619115124830473,
      "loss": 2.2535,
      "step": 37359
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.724346935749054,
      "learning_rate": 0.0005619095176898776,
      "loss": 2.1481,
      "step": 37360
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6852306127548218,
      "learning_rate": 0.0005619075228480139,
      "loss": 2.5309,
      "step": 37361
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.7671427726745605,
      "learning_rate": 0.0005619055279574566,
      "loss": 2.1167,
      "step": 37362
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.8039476275444031,
      "learning_rate": 0.0005619035330182063,
      "loss": 2.5407,
      "step": 37363
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6483153104782104,
      "learning_rate": 0.000561901538030263,
      "loss": 2.1596,
      "step": 37364
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.8232914805412292,
      "learning_rate": 0.0005618995429936274,
      "loss": 2.2019,
      "step": 37365
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6535434722900391,
      "learning_rate": 0.0005618975479082997,
      "loss": 2.1962,
      "step": 37366
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6442552208900452,
      "learning_rate": 0.0005618955527742803,
      "loss": 2.1386,
      "step": 37367
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6372513771057129,
      "learning_rate": 0.0005618935575915695,
      "loss": 1.9845,
      "step": 37368
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6356849670410156,
      "learning_rate": 0.0005618915623601679,
      "loss": 2.3685,
      "step": 37369
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6352542638778687,
      "learning_rate": 0.0005618895670800755,
      "loss": 2.4117,
      "step": 37370
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.7004671692848206,
      "learning_rate": 0.0005618875717512931,
      "loss": 2.3376,
      "step": 37371
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.7036993503570557,
      "learning_rate": 0.0005618855763738208,
      "loss": 2.3396,
      "step": 37372
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.7695114612579346,
      "learning_rate": 0.0005618835809476589,
      "loss": 2.1945,
      "step": 37373
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6978541016578674,
      "learning_rate": 0.000561881585472808,
      "loss": 2.4085,
      "step": 37374
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6746752262115479,
      "learning_rate": 0.0005618795899492683,
      "loss": 2.549,
      "step": 37375
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.7211438417434692,
      "learning_rate": 0.0005618775943770403,
      "loss": 2.3098,
      "step": 37376
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6863551735877991,
      "learning_rate": 0.0005618755987561243,
      "loss": 2.3024,
      "step": 37377
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.7265716195106506,
      "learning_rate": 0.0005618736030865207,
      "loss": 2.375,
      "step": 37378
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6941573619842529,
      "learning_rate": 0.0005618716073682298,
      "loss": 2.3139,
      "step": 37379
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6916932463645935,
      "learning_rate": 0.0005618696116012522,
      "loss": 2.6117,
      "step": 37380
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.7178367376327515,
      "learning_rate": 0.000561867615785588,
      "loss": 2.3253,
      "step": 37381
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.7376992702484131,
      "learning_rate": 0.0005618656199212375,
      "loss": 2.4346,
      "step": 37382
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.8382599353790283,
      "learning_rate": 0.0005618636240082015,
      "loss": 2.354,
      "step": 37383
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6861879825592041,
      "learning_rate": 0.0005618616280464799,
      "loss": 2.2785,
      "step": 37384
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6998604536056519,
      "learning_rate": 0.0005618596320360734,
      "loss": 2.1739,
      "step": 37385
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6752282977104187,
      "learning_rate": 0.0005618576359769821,
      "loss": 2.2413,
      "step": 37386
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6762723326683044,
      "learning_rate": 0.0005618556398692067,
      "loss": 2.3061,
      "step": 37387
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6354871392250061,
      "learning_rate": 0.0005618536437127473,
      "loss": 2.1647,
      "step": 37388
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6739105582237244,
      "learning_rate": 0.0005618516475076044,
      "loss": 2.3314,
      "step": 37389
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6653521060943604,
      "learning_rate": 0.0005618496512537783,
      "loss": 2.3361,
      "step": 37390
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6432218551635742,
      "learning_rate": 0.0005618476549512695,
      "loss": 2.2648,
      "step": 37391
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6910777688026428,
      "learning_rate": 0.0005618456586000782,
      "loss": 2.4252,
      "step": 37392
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.661038339138031,
      "learning_rate": 0.0005618436622002048,
      "loss": 2.2544,
      "step": 37393
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.7234492897987366,
      "learning_rate": 0.0005618416657516499,
      "loss": 2.3355,
      "step": 37394
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6502491235733032,
      "learning_rate": 0.0005618396692544135,
      "loss": 2.2589,
      "step": 37395
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.7099590301513672,
      "learning_rate": 0.0005618376727084963,
      "loss": 2.2285,
      "step": 37396
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6818845868110657,
      "learning_rate": 0.0005618356761138985,
      "loss": 2.0429,
      "step": 37397
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.7310870289802551,
      "learning_rate": 0.0005618336794706204,
      "loss": 2.5127,
      "step": 37398
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6597025394439697,
      "learning_rate": 0.0005618316827786625,
      "loss": 2.2124,
      "step": 37399
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6490763425827026,
      "learning_rate": 0.0005618296860380252,
      "loss": 2.3447,
      "step": 37400
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.651029109954834,
      "learning_rate": 0.0005618276892487089,
      "loss": 2.1498,
      "step": 37401
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6417135000228882,
      "learning_rate": 0.0005618256924107138,
      "loss": 2.2652,
      "step": 37402
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.7418727278709412,
      "learning_rate": 0.0005618236955240404,
      "loss": 2.2397,
      "step": 37403
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.7468366026878357,
      "learning_rate": 0.000561821698588689,
      "loss": 2.0637,
      "step": 37404
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.7449350357055664,
      "learning_rate": 0.00056181970160466,
      "loss": 2.5049,
      "step": 37405
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6294070482254028,
      "learning_rate": 0.0005618177045719538,
      "loss": 2.2835,
      "step": 37406
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6915972232818604,
      "learning_rate": 0.0005618157074905707,
      "loss": 2.1453,
      "step": 37407
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6637709140777588,
      "learning_rate": 0.0005618137103605112,
      "loss": 2.1331,
      "step": 37408
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.8081411719322205,
      "learning_rate": 0.0005618117131817755,
      "loss": 2.3167,
      "step": 37409
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.7274287939071655,
      "learning_rate": 0.0005618097159543641,
      "loss": 2.2779,
      "step": 37410
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.683329164981842,
      "learning_rate": 0.0005618077186782774,
      "loss": 2.5135,
      "step": 37411
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.8222454190254211,
      "learning_rate": 0.0005618057213535155,
      "loss": 2.3517,
      "step": 37412
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.8084933161735535,
      "learning_rate": 0.0005618037239800792,
      "loss": 2.5477,
      "step": 37413
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6313285827636719,
      "learning_rate": 0.0005618017265579685,
      "loss": 2.1243,
      "step": 37414
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.7192068099975586,
      "learning_rate": 0.0005617997290871839,
      "loss": 2.4732,
      "step": 37415
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6807236671447754,
      "learning_rate": 0.0005617977315677259,
      "loss": 2.1451,
      "step": 37416
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6987778544425964,
      "learning_rate": 0.0005617957339995947,
      "loss": 2.1859,
      "step": 37417
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.652538537979126,
      "learning_rate": 0.0005617937363827909,
      "loss": 2.3479,
      "step": 37418
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6598678231239319,
      "learning_rate": 0.0005617917387173145,
      "loss": 2.1826,
      "step": 37419
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6940543055534363,
      "learning_rate": 0.0005617897410031661,
      "loss": 2.1319,
      "step": 37420
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6578735709190369,
      "learning_rate": 0.0005617877432403461,
      "loss": 2.3991,
      "step": 37421
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.5995215773582458,
      "learning_rate": 0.0005617857454288548,
      "loss": 2.2296,
      "step": 37422
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.7873808145523071,
      "learning_rate": 0.0005617837475686925,
      "loss": 2.2101,
      "step": 37423
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6659824252128601,
      "learning_rate": 0.0005617817496598597,
      "loss": 2.2935,
      "step": 37424
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6626363396644592,
      "learning_rate": 0.0005617797517023568,
      "loss": 2.1652,
      "step": 37425
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6145860552787781,
      "learning_rate": 0.0005617777536961841,
      "loss": 2.1754,
      "step": 37426
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.7645061612129211,
      "learning_rate": 0.000561775755641342,
      "loss": 2.2643,
      "step": 37427
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6500228643417358,
      "learning_rate": 0.0005617737575378308,
      "loss": 2.2366,
      "step": 37428
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6969577670097351,
      "learning_rate": 0.0005617717593856509,
      "loss": 2.0891,
      "step": 37429
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.7757415175437927,
      "learning_rate": 0.0005617697611848026,
      "loss": 2.2975,
      "step": 37430
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6811338067054749,
      "learning_rate": 0.0005617677629352865,
      "loss": 2.5605,
      "step": 37431
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.663511335849762,
      "learning_rate": 0.0005617657646371029,
      "loss": 2.2822,
      "step": 37432
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6676194071769714,
      "learning_rate": 0.0005617637662902519,
      "loss": 2.3633,
      "step": 37433
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6461963057518005,
      "learning_rate": 0.0005617617678947342,
      "loss": 2.3303,
      "step": 37434
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6500813961029053,
      "learning_rate": 0.0005617597694505499,
      "loss": 2.0609,
      "step": 37435
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6325893402099609,
      "learning_rate": 0.0005617577709576997,
      "loss": 2.2011,
      "step": 37436
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.7081273198127747,
      "learning_rate": 0.0005617557724161838,
      "loss": 2.3304,
      "step": 37437
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6370201110839844,
      "learning_rate": 0.0005617537738260024,
      "loss": 2.2682,
      "step": 37438
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6698643565177917,
      "learning_rate": 0.0005617517751871562,
      "loss": 2.2891,
      "step": 37439
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.7444953918457031,
      "learning_rate": 0.0005617497764996452,
      "loss": 2.1375,
      "step": 37440
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.8518021702766418,
      "learning_rate": 0.0005617477777634701,
      "loss": 2.1729,
      "step": 37441
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.7987304329872131,
      "learning_rate": 0.0005617457789786312,
      "loss": 1.9712,
      "step": 37442
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6854231357574463,
      "learning_rate": 0.0005617437801451288,
      "loss": 2.2439,
      "step": 37443
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.7489503622055054,
      "learning_rate": 0.0005617417812629631,
      "loss": 2.4153,
      "step": 37444
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.8272913098335266,
      "learning_rate": 0.0005617397823321348,
      "loss": 2.3935,
      "step": 37445
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6475668549537659,
      "learning_rate": 0.0005617377833526442,
      "loss": 2.169,
      "step": 37446
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.031069278717041,
      "learning_rate": 0.0005617357843244916,
      "loss": 2.2034,
      "step": 37447
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6493325233459473,
      "learning_rate": 0.0005617337852476773,
      "loss": 1.9568,
      "step": 37448
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6816415190696716,
      "learning_rate": 0.0005617317861222018,
      "loss": 2.3796,
      "step": 37449
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.7514593005180359,
      "learning_rate": 0.0005617297869480654,
      "loss": 2.2707,
      "step": 37450
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.7316779494285583,
      "learning_rate": 0.0005617277877252685,
      "loss": 2.2772,
      "step": 37451
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.624300479888916,
      "learning_rate": 0.0005617257884538115,
      "loss": 2.0362,
      "step": 37452
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6394225358963013,
      "learning_rate": 0.0005617237891336946,
      "loss": 2.109,
      "step": 37453
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.7591556310653687,
      "learning_rate": 0.0005617217897649184,
      "loss": 2.1261,
      "step": 37454
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.5650200843811035,
      "learning_rate": 0.0005617197903474832,
      "loss": 2.361,
      "step": 37455
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6566960215568542,
      "learning_rate": 0.0005617177908813893,
      "loss": 2.3416,
      "step": 37456
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.9403752684593201,
      "learning_rate": 0.0005617157913666371,
      "loss": 2.0791,
      "step": 37457
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6076168417930603,
      "learning_rate": 0.0005617137918032271,
      "loss": 2.1801,
      "step": 37458
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.7808860540390015,
      "learning_rate": 0.0005617117921911595,
      "loss": 2.0196,
      "step": 37459
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.8713744282722473,
      "learning_rate": 0.0005617097925304348,
      "loss": 2.1879,
      "step": 37460
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6790605187416077,
      "learning_rate": 0.0005617077928210532,
      "loss": 2.1548,
      "step": 37461
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.679663896560669,
      "learning_rate": 0.0005617057930630152,
      "loss": 2.3043,
      "step": 37462
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.9069917798042297,
      "learning_rate": 0.0005617037932563212,
      "loss": 2.1011,
      "step": 37463
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.8680837750434875,
      "learning_rate": 0.0005617017934009716,
      "loss": 2.2569,
      "step": 37464
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6684203743934631,
      "learning_rate": 0.0005616997934969665,
      "loss": 2.2771,
      "step": 37465
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.0285297632217407,
      "learning_rate": 0.0005616977935443067,
      "loss": 2.2068,
      "step": 37466
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.9553290605545044,
      "learning_rate": 0.0005616957935429922,
      "loss": 2.3708,
      "step": 37467
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.7144915461540222,
      "learning_rate": 0.0005616937934930235,
      "loss": 2.2188,
      "step": 37468
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.7644768953323364,
      "learning_rate": 0.000561691793394401,
      "loss": 2.3027,
      "step": 37469
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.9037529826164246,
      "learning_rate": 0.0005616897932471251,
      "loss": 2.2269,
      "step": 37470
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.0668832063674927,
      "learning_rate": 0.0005616877930511961,
      "loss": 2.3042,
      "step": 37471
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.5907306671142578,
      "learning_rate": 0.0005616857928066144,
      "loss": 2.2671,
      "step": 37472
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.1501799821853638,
      "learning_rate": 0.0005616837925133803,
      "loss": 2.4172,
      "step": 37473
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.1845951080322266,
      "learning_rate": 0.0005616817921714943,
      "loss": 2.2502,
      "step": 37474
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.7378461360931396,
      "learning_rate": 0.0005616797917809568,
      "loss": 2.4463,
      "step": 37475
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6617720723152161,
      "learning_rate": 0.000561677791341768,
      "loss": 2.3376,
      "step": 37476
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.8546270728111267,
      "learning_rate": 0.0005616757908539283,
      "loss": 2.3287,
      "step": 37477
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.9104153513908386,
      "learning_rate": 0.0005616737903174383,
      "loss": 2.3498,
      "step": 37478
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6723834872245789,
      "learning_rate": 0.000561671789732298,
      "loss": 2.4542,
      "step": 37479
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6990742683410645,
      "learning_rate": 0.0005616697890985082,
      "loss": 2.1426,
      "step": 37480
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.7610651254653931,
      "learning_rate": 0.0005616677884160689,
      "loss": 2.2784,
      "step": 37481
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.3895068168640137,
      "learning_rate": 0.0005616657876849806,
      "loss": 2.0592,
      "step": 37482
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.8259090781211853,
      "learning_rate": 0.0005616637869052439,
      "loss": 2.2749,
      "step": 37483
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6760368943214417,
      "learning_rate": 0.0005616617860768587,
      "loss": 2.4275,
      "step": 37484
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.7546126246452332,
      "learning_rate": 0.0005616597851998257,
      "loss": 2.3113,
      "step": 37485
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.061814785003662,
      "learning_rate": 0.0005616577842741452,
      "loss": 2.3713,
      "step": 37486
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.9017364978790283,
      "learning_rate": 0.0005616557832998177,
      "loss": 2.3462,
      "step": 37487
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6473388671875,
      "learning_rate": 0.0005616537822768434,
      "loss": 2.421,
      "step": 37488
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.7851340174674988,
      "learning_rate": 0.0005616517812052226,
      "loss": 2.2993,
      "step": 37489
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.5402268171310425,
      "learning_rate": 0.0005616497800849559,
      "loss": 2.2603,
      "step": 37490
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.7513819336891174,
      "learning_rate": 0.0005616477789160436,
      "loss": 2.2184,
      "step": 37491
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6746188402175903,
      "learning_rate": 0.0005616457776984861,
      "loss": 2.325,
      "step": 37492
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.8865496516227722,
      "learning_rate": 0.0005616437764322835,
      "loss": 2.1833,
      "step": 37493
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.3587554693222046,
      "learning_rate": 0.0005616417751174366,
      "loss": 2.1761,
      "step": 37494
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6836293935775757,
      "learning_rate": 0.0005616397737539453,
      "loss": 2.2335,
      "step": 37495
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.7308820486068726,
      "learning_rate": 0.0005616377723418104,
      "loss": 2.4353,
      "step": 37496
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.8109415173530579,
      "learning_rate": 0.0005616357708810321,
      "loss": 2.47,
      "step": 37497
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.8805670738220215,
      "learning_rate": 0.0005616337693716107,
      "loss": 2.4564,
      "step": 37498
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6721478700637817,
      "learning_rate": 0.0005616317678135467,
      "loss": 2.5488,
      "step": 37499
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.7245645523071289,
      "learning_rate": 0.0005616297662068404,
      "loss": 2.2509,
      "step": 37500
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.7709256410598755,
      "learning_rate": 0.0005616277645514922,
      "loss": 2.3526,
      "step": 37501
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.7879222631454468,
      "learning_rate": 0.0005616257628475025,
      "loss": 2.2528,
      "step": 37502
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6309002637863159,
      "learning_rate": 0.0005616237610948716,
      "loss": 2.2851,
      "step": 37503
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6473422050476074,
      "learning_rate": 0.0005616217592935999,
      "loss": 2.2501,
      "step": 37504
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6723617911338806,
      "learning_rate": 0.0005616197574436877,
      "loss": 2.2735,
      "step": 37505
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.8327340483665466,
      "learning_rate": 0.0005616177555451355,
      "loss": 2.4661,
      "step": 37506
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.8070492744445801,
      "learning_rate": 0.0005616157535979437,
      "loss": 2.0417,
      "step": 37507
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6892330050468445,
      "learning_rate": 0.0005616137516021124,
      "loss": 2.2562,
      "step": 37508
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.7596879005432129,
      "learning_rate": 0.0005616117495576424,
      "loss": 2.1116,
      "step": 37509
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.739268958568573,
      "learning_rate": 0.0005616097474645338,
      "loss": 2.1167,
      "step": 37510
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.7117944359779358,
      "learning_rate": 0.0005616077453227869,
      "loss": 2.1027,
      "step": 37511
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.668738603591919,
      "learning_rate": 0.0005616057431324021,
      "loss": 2.274,
      "step": 37512
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.7658613920211792,
      "learning_rate": 0.00056160374089338,
      "loss": 2.353,
      "step": 37513
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.8894312381744385,
      "learning_rate": 0.0005616017386057209,
      "loss": 2.4147,
      "step": 37514
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6613083481788635,
      "learning_rate": 0.000561599736269425,
      "loss": 2.3156,
      "step": 37515
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6977331638336182,
      "learning_rate": 0.0005615977338844927,
      "loss": 2.4126,
      "step": 37516
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.9392895698547363,
      "learning_rate": 0.0005615957314509245,
      "loss": 2.5686,
      "step": 37517
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.7118521332740784,
      "learning_rate": 0.0005615937289687208,
      "loss": 2.3063,
      "step": 37518
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6601245999336243,
      "learning_rate": 0.0005615917264378818,
      "loss": 2.149,
      "step": 37519
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6812667846679688,
      "learning_rate": 0.0005615897238584079,
      "loss": 2.2152,
      "step": 37520
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6451578736305237,
      "learning_rate": 0.0005615877212302997,
      "loss": 2.3578,
      "step": 37521
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.8993154168128967,
      "learning_rate": 0.0005615857185535572,
      "loss": 2.5478,
      "step": 37522
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6314488053321838,
      "learning_rate": 0.0005615837158281812,
      "loss": 2.0855,
      "step": 37523
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.7076296210289001,
      "learning_rate": 0.0005615817130541717,
      "loss": 2.3807,
      "step": 37524
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.8078005313873291,
      "learning_rate": 0.0005615797102315293,
      "loss": 2.3236,
      "step": 37525
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.7340859174728394,
      "learning_rate": 0.0005615777073602542,
      "loss": 2.3103,
      "step": 37526
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6322150826454163,
      "learning_rate": 0.0005615757044403469,
      "loss": 2.2106,
      "step": 37527
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.7159342765808105,
      "learning_rate": 0.0005615737014718078,
      "loss": 2.3767,
      "step": 37528
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.7623336911201477,
      "learning_rate": 0.0005615716984546372,
      "loss": 2.3936,
      "step": 37529
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.5958757400512695,
      "learning_rate": 0.0005615696953888354,
      "loss": 2.3175,
      "step": 37530
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.7119414210319519,
      "learning_rate": 0.000561567692274403,
      "loss": 2.3716,
      "step": 37531
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.702409565448761,
      "learning_rate": 0.0005615656891113401,
      "loss": 2.3796,
      "step": 37532
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6978137493133545,
      "learning_rate": 0.0005615636858996472,
      "loss": 2.1826,
      "step": 37533
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.710901141166687,
      "learning_rate": 0.0005615616826393248,
      "loss": 2.4178,
      "step": 37534
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.7601301670074463,
      "learning_rate": 0.000561559679330373,
      "loss": 2.1444,
      "step": 37535
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6166180372238159,
      "learning_rate": 0.0005615576759727924,
      "loss": 2.3881,
      "step": 37536
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6391898989677429,
      "learning_rate": 0.0005615556725665832,
      "loss": 2.2355,
      "step": 37537
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6806313395500183,
      "learning_rate": 0.000561553669111746,
      "loss": 2.0648,
      "step": 37538
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6600664258003235,
      "learning_rate": 0.0005615516656082809,
      "loss": 2.2615,
      "step": 37539
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6964203715324402,
      "learning_rate": 0.0005615496620561886,
      "loss": 2.5754,
      "step": 37540
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6882619261741638,
      "learning_rate": 0.0005615476584554691,
      "loss": 2.4234,
      "step": 37541
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.7403654456138611,
      "learning_rate": 0.000561545654806123,
      "loss": 2.3007,
      "step": 37542
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.7403901815414429,
      "learning_rate": 0.0005615436511081506,
      "loss": 2.434,
      "step": 37543
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6438923478126526,
      "learning_rate": 0.0005615416473615523,
      "loss": 2.1231,
      "step": 37544
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6715177297592163,
      "learning_rate": 0.0005615396435663286,
      "loss": 2.4972,
      "step": 37545
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6992174386978149,
      "learning_rate": 0.0005615376397224796,
      "loss": 2.2386,
      "step": 37546
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.7129865884780884,
      "learning_rate": 0.0005615356358300059,
      "loss": 2.1052,
      "step": 37547
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6793238520622253,
      "learning_rate": 0.0005615336318889078,
      "loss": 2.2217,
      "step": 37548
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.75076824426651,
      "learning_rate": 0.0005615316278991855,
      "loss": 2.1698,
      "step": 37549
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.7256721258163452,
      "learning_rate": 0.0005615296238608397,
      "loss": 2.1427,
      "step": 37550
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6801184415817261,
      "learning_rate": 0.0005615276197738706,
      "loss": 2.484,
      "step": 37551
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6637787818908691,
      "learning_rate": 0.0005615256156382784,
      "loss": 2.145,
      "step": 37552
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6975098252296448,
      "learning_rate": 0.0005615236114540639,
      "loss": 2.3251,
      "step": 37553
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.7320744395256042,
      "learning_rate": 0.0005615216072212271,
      "loss": 2.2177,
      "step": 37554
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.7443321347236633,
      "learning_rate": 0.0005615196029397685,
      "loss": 2.2542,
      "step": 37555
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.7318131327629089,
      "learning_rate": 0.0005615175986096885,
      "loss": 2.3733,
      "step": 37556
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.722251296043396,
      "learning_rate": 0.0005615155942309874,
      "loss": 2.2916,
      "step": 37557
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.8153268098831177,
      "learning_rate": 0.0005615135898036657,
      "loss": 2.2744,
      "step": 37558
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.751599133014679,
      "learning_rate": 0.0005615115853277236,
      "loss": 2.3064,
      "step": 37559
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6544331908226013,
      "learning_rate": 0.0005615095808031615,
      "loss": 2.1805,
      "step": 37560
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6819833517074585,
      "learning_rate": 0.00056150757622998,
      "loss": 2.2579,
      "step": 37561
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.7180221080780029,
      "learning_rate": 0.0005615055716081791,
      "loss": 2.3234,
      "step": 37562
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.7170984745025635,
      "learning_rate": 0.0005615035669377595,
      "loss": 2.3198,
      "step": 37563
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6824626922607422,
      "learning_rate": 0.0005615015622187215,
      "loss": 2.3188,
      "step": 37564
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.641045093536377,
      "learning_rate": 0.0005614995574510654,
      "loss": 2.0888,
      "step": 37565
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.7932104468345642,
      "learning_rate": 0.0005614975526347915,
      "loss": 2.1898,
      "step": 37566
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.8160841464996338,
      "learning_rate": 0.0005614955477699004,
      "loss": 2.4624,
      "step": 37567
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6287897229194641,
      "learning_rate": 0.0005614935428563923,
      "loss": 2.0147,
      "step": 37568
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.7724172472953796,
      "learning_rate": 0.0005614915378942675,
      "loss": 2.5086,
      "step": 37569
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.7187211513519287,
      "learning_rate": 0.0005614895328835267,
      "loss": 2.2947,
      "step": 37570
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.8320391774177551,
      "learning_rate": 0.0005614875278241698,
      "loss": 2.1392,
      "step": 37571
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6208714842796326,
      "learning_rate": 0.0005614855227161976,
      "loss": 2.3593,
      "step": 37572
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.7138100862503052,
      "learning_rate": 0.0005614835175596103,
      "loss": 2.1515,
      "step": 37573
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.84343421459198,
      "learning_rate": 0.0005614815123544082,
      "loss": 2.3032,
      "step": 37574
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.8972658514976501,
      "learning_rate": 0.0005614795071005918,
      "loss": 2.2565,
      "step": 37575
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.7412922382354736,
      "learning_rate": 0.0005614775017981615,
      "loss": 2.2959,
      "step": 37576
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6544908285140991,
      "learning_rate": 0.0005614754964471174,
      "loss": 2.2326,
      "step": 37577
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.7441381812095642,
      "learning_rate": 0.0005614734910474603,
      "loss": 2.1237,
      "step": 37578
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.7062798738479614,
      "learning_rate": 0.0005614714855991902,
      "loss": 2.2243,
      "step": 37579
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6814462542533875,
      "learning_rate": 0.0005614694801023076,
      "loss": 2.3122,
      "step": 37580
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.7242015600204468,
      "learning_rate": 0.0005614674745568131,
      "loss": 2.3681,
      "step": 37581
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.8181013464927673,
      "learning_rate": 0.0005614654689627066,
      "loss": 2.0823,
      "step": 37582
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6281712651252747,
      "learning_rate": 0.0005614634633199888,
      "loss": 2.4528,
      "step": 37583
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.7072504162788391,
      "learning_rate": 0.0005614614576286601,
      "loss": 2.362,
      "step": 37584
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.895676851272583,
      "learning_rate": 0.0005614594518887206,
      "loss": 2.105,
      "step": 37585
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.7215112447738647,
      "learning_rate": 0.000561457446100171,
      "loss": 2.4738,
      "step": 37586
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.8788513541221619,
      "learning_rate": 0.0005614554402630116,
      "loss": 2.4409,
      "step": 37587
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.8077143430709839,
      "learning_rate": 0.0005614534343772426,
      "loss": 2.4103,
      "step": 37588
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.7543089389801025,
      "learning_rate": 0.0005614514284428644,
      "loss": 2.2992,
      "step": 37589
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.7179213762283325,
      "learning_rate": 0.0005614494224598775,
      "loss": 2.4479,
      "step": 37590
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.7345895767211914,
      "learning_rate": 0.0005614474164282822,
      "loss": 2.2548,
      "step": 37591
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.7222464084625244,
      "learning_rate": 0.0005614454103480789,
      "loss": 2.317,
      "step": 37592
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6817598938941956,
      "learning_rate": 0.000561443404219268,
      "loss": 2.1567,
      "step": 37593
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6951867341995239,
      "learning_rate": 0.0005614413980418499,
      "loss": 2.3009,
      "step": 37594
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.8224248290061951,
      "learning_rate": 0.0005614393918158247,
      "loss": 2.1682,
      "step": 37595
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6601114273071289,
      "learning_rate": 0.0005614373855411932,
      "loss": 2.4224,
      "step": 37596
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.65484619140625,
      "learning_rate": 0.0005614353792179554,
      "loss": 2.1817,
      "step": 37597
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.7115432024002075,
      "learning_rate": 0.000561433372846112,
      "loss": 2.116,
      "step": 37598
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6958001255989075,
      "learning_rate": 0.0005614313664256631,
      "loss": 2.2636,
      "step": 37599
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.771108865737915,
      "learning_rate": 0.0005614293599566091,
      "loss": 2.455,
      "step": 37600
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.7005071640014648,
      "learning_rate": 0.0005614273534389506,
      "loss": 2.2833,
      "step": 37601
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6950106024742126,
      "learning_rate": 0.0005614253468726876,
      "loss": 2.1347,
      "step": 37602
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6609636545181274,
      "learning_rate": 0.000561423340257821,
      "loss": 2.3506,
      "step": 37603
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.7691659331321716,
      "learning_rate": 0.0005614213335943506,
      "loss": 2.0063,
      "step": 37604
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6451236605644226,
      "learning_rate": 0.0005614193268822772,
      "loss": 2.1045,
      "step": 37605
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.7004042267799377,
      "learning_rate": 0.0005614173201216009,
      "loss": 2.2294,
      "step": 37606
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.767817497253418,
      "learning_rate": 0.0005614153133123223,
      "loss": 2.3386,
      "step": 37607
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.7353700399398804,
      "learning_rate": 0.0005614133064544414,
      "loss": 2.2417,
      "step": 37608
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6565096974372864,
      "learning_rate": 0.0005614112995479591,
      "loss": 2.113,
      "step": 37609
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6744306087493896,
      "learning_rate": 0.0005614092925928755,
      "loss": 2.2077,
      "step": 37610
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.7313067317008972,
      "learning_rate": 0.0005614072855891908,
      "loss": 2.0084,
      "step": 37611
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6488051414489746,
      "learning_rate": 0.0005614052785369057,
      "loss": 2.32,
      "step": 37612
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.7397459149360657,
      "learning_rate": 0.0005614032714360204,
      "loss": 2.5646,
      "step": 37613
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6788555979728699,
      "learning_rate": 0.0005614012642865353,
      "loss": 2.4445,
      "step": 37614
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.642288863658905,
      "learning_rate": 0.0005613992570884508,
      "loss": 2.1425,
      "step": 37615
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.7690557837486267,
      "learning_rate": 0.0005613972498417671,
      "loss": 2.5331,
      "step": 37616
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.690282940864563,
      "learning_rate": 0.0005613952425464848,
      "loss": 2.4202,
      "step": 37617
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.7119702100753784,
      "learning_rate": 0.0005613932352026042,
      "loss": 2.14,
      "step": 37618
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.707245409488678,
      "learning_rate": 0.0005613912278101257,
      "loss": 2.3657,
      "step": 37619
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6154983639717102,
      "learning_rate": 0.0005613892203690495,
      "loss": 2.2645,
      "step": 37620
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6752554774284363,
      "learning_rate": 0.0005613872128793762,
      "loss": 2.3772,
      "step": 37621
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6252918243408203,
      "learning_rate": 0.0005613852053411062,
      "loss": 2.0835,
      "step": 37622
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6681968569755554,
      "learning_rate": 0.0005613831977542396,
      "loss": 2.4392,
      "step": 37623
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.686260998249054,
      "learning_rate": 0.000561381190118777,
      "loss": 2.1455,
      "step": 37624
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6550688147544861,
      "learning_rate": 0.0005613791824347187,
      "loss": 2.2716,
      "step": 37625
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6850203275680542,
      "learning_rate": 0.000561377174702065,
      "loss": 2.2703,
      "step": 37626
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6803669333457947,
      "learning_rate": 0.0005613751669208164,
      "loss": 2.4293,
      "step": 37627
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.722575843334198,
      "learning_rate": 0.0005613731590909732,
      "loss": 2.4235,
      "step": 37628
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.7327587604522705,
      "learning_rate": 0.0005613711512125358,
      "loss": 2.5227,
      "step": 37629
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.7092573046684265,
      "learning_rate": 0.0005613691432855047,
      "loss": 2.2346,
      "step": 37630
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6433370113372803,
      "learning_rate": 0.0005613671353098801,
      "loss": 2.5327,
      "step": 37631
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6811557412147522,
      "learning_rate": 0.0005613651272856622,
      "loss": 2.2982,
      "step": 37632
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.7866306900978088,
      "learning_rate": 0.0005613631192128518,
      "loss": 2.2866,
      "step": 37633
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.7796523571014404,
      "learning_rate": 0.0005613611110914491,
      "loss": 2.306,
      "step": 37634
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6627756357192993,
      "learning_rate": 0.0005613591029214542,
      "loss": 2.5519,
      "step": 37635
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.8361085653305054,
      "learning_rate": 0.0005613570947028679,
      "loss": 2.1036,
      "step": 37636
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.7948198318481445,
      "learning_rate": 0.0005613550864356903,
      "loss": 2.2763,
      "step": 37637
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6196587085723877,
      "learning_rate": 0.0005613530781199219,
      "loss": 2.3971,
      "step": 37638
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.7002207040786743,
      "learning_rate": 0.0005613510697555631,
      "loss": 2.4148,
      "step": 37639
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.8020808696746826,
      "learning_rate": 0.0005613490613426141,
      "loss": 2.4939,
      "step": 37640
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6919488906860352,
      "learning_rate": 0.0005613470528810754,
      "loss": 2.334,
      "step": 37641
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6984049081802368,
      "learning_rate": 0.0005613450443709473,
      "loss": 2.2749,
      "step": 37642
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6758089065551758,
      "learning_rate": 0.0005613430358122303,
      "loss": 2.3683,
      "step": 37643
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.7783827781677246,
      "learning_rate": 0.0005613410272049247,
      "loss": 2.2317,
      "step": 37644
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.7954320907592773,
      "learning_rate": 0.0005613390185490308,
      "loss": 2.0076,
      "step": 37645
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.7836663126945496,
      "learning_rate": 0.0005613370098445491,
      "loss": 2.4499,
      "step": 37646
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.8604839444160461,
      "learning_rate": 0.0005613350010914799,
      "loss": 2.0986,
      "step": 37647
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.8625642657279968,
      "learning_rate": 0.0005613329922898235,
      "loss": 2.1322,
      "step": 37648
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6425504684448242,
      "learning_rate": 0.0005613309834395805,
      "loss": 2.2992,
      "step": 37649
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.7715316414833069,
      "learning_rate": 0.000561328974540751,
      "loss": 2.2398,
      "step": 37650
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.7021366357803345,
      "learning_rate": 0.0005613269655933357,
      "loss": 2.5462,
      "step": 37651
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6908699870109558,
      "learning_rate": 0.0005613249565973346,
      "loss": 2.1508,
      "step": 37652
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.7576121687889099,
      "learning_rate": 0.0005613229475527483,
      "loss": 2.217,
      "step": 37653
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6568792462348938,
      "learning_rate": 0.0005613209384595772,
      "loss": 2.3037,
      "step": 37654
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.7730284333229065,
      "learning_rate": 0.0005613189293178216,
      "loss": 2.0797,
      "step": 37655
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.7166836857795715,
      "learning_rate": 0.0005613169201274819,
      "loss": 2.2198,
      "step": 37656
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.7254109382629395,
      "learning_rate": 0.0005613149108885583,
      "loss": 2.3531,
      "step": 37657
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.7895412445068359,
      "learning_rate": 0.0005613129016010515,
      "loss": 2.2314,
      "step": 37658
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6936690807342529,
      "learning_rate": 0.0005613108922649617,
      "loss": 1.9953,
      "step": 37659
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6916988492012024,
      "learning_rate": 0.0005613088828802891,
      "loss": 2.2597,
      "step": 37660
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.7395544648170471,
      "learning_rate": 0.0005613068734470344,
      "loss": 2.39,
      "step": 37661
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.665184736251831,
      "learning_rate": 0.0005613048639651978,
      "loss": 2.2675,
      "step": 37662
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6185871362686157,
      "learning_rate": 0.0005613028544347795,
      "loss": 1.9554,
      "step": 37663
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6333345770835876,
      "learning_rate": 0.0005613008448557803,
      "loss": 2.2742,
      "step": 37664
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6694096922874451,
      "learning_rate": 0.0005612988352282003,
      "loss": 2.2139,
      "step": 37665
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6185722351074219,
      "learning_rate": 0.0005612968255520399,
      "loss": 2.2959,
      "step": 37666
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6639317274093628,
      "learning_rate": 0.0005612948158272995,
      "loss": 2.3074,
      "step": 37667
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.7369340062141418,
      "learning_rate": 0.0005612928060539794,
      "loss": 2.4082,
      "step": 37668
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.7051348090171814,
      "learning_rate": 0.00056129079623208,
      "loss": 2.4426,
      "step": 37669
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6785673499107361,
      "learning_rate": 0.0005612887863616019,
      "loss": 2.265,
      "step": 37670
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.7337421178817749,
      "learning_rate": 0.0005612867764425451,
      "loss": 2.2251,
      "step": 37671
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6692274212837219,
      "learning_rate": 0.0005612847664749102,
      "loss": 2.5214,
      "step": 37672
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6808459758758545,
      "learning_rate": 0.0005612827564586975,
      "loss": 2.1081,
      "step": 37673
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6455959677696228,
      "learning_rate": 0.0005612807463939074,
      "loss": 2.2287,
      "step": 37674
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.657727837562561,
      "learning_rate": 0.0005612787362805405,
      "loss": 2.3146,
      "step": 37675
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.662799060344696,
      "learning_rate": 0.0005612767261185967,
      "loss": 2.3891,
      "step": 37676
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.7213904857635498,
      "learning_rate": 0.0005612747159080767,
      "loss": 2.3035,
      "step": 37677
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.7179238200187683,
      "learning_rate": 0.0005612727056489808,
      "loss": 2.347,
      "step": 37678
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6609368920326233,
      "learning_rate": 0.0005612706953413094,
      "loss": 2.3789,
      "step": 37679
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6271763443946838,
      "learning_rate": 0.0005612686849850628,
      "loss": 2.1736,
      "step": 37680
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.656855583190918,
      "learning_rate": 0.0005612666745802414,
      "loss": 2.2012,
      "step": 37681
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.8032815456390381,
      "learning_rate": 0.0005612646641268457,
      "loss": 2.3099,
      "step": 37682
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6896350383758545,
      "learning_rate": 0.0005612626536248758,
      "loss": 2.3565,
      "step": 37683
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.7473463416099548,
      "learning_rate": 0.0005612606430743323,
      "loss": 2.3134,
      "step": 37684
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6489664912223816,
      "learning_rate": 0.0005612586324752156,
      "loss": 2.1518,
      "step": 37685
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6232582330703735,
      "learning_rate": 0.0005612566218275259,
      "loss": 2.2736,
      "step": 37686
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6936407685279846,
      "learning_rate": 0.0005612546111312638,
      "loss": 2.2456,
      "step": 37687
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6681749224662781,
      "learning_rate": 0.0005612526003864294,
      "loss": 2.1149,
      "step": 37688
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.7014468312263489,
      "learning_rate": 0.0005612505895930232,
      "loss": 2.0384,
      "step": 37689
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.655021071434021,
      "learning_rate": 0.0005612485787510457,
      "loss": 2.1204,
      "step": 37690
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6828858852386475,
      "learning_rate": 0.0005612465678604971,
      "loss": 2.3889,
      "step": 37691
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.7553310990333557,
      "learning_rate": 0.0005612445569213777,
      "loss": 2.4498,
      "step": 37692
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.7155932188034058,
      "learning_rate": 0.0005612425459336882,
      "loss": 2.3249,
      "step": 37693
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6770194172859192,
      "learning_rate": 0.0005612405348974287,
      "loss": 2.2803,
      "step": 37694
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.7401971817016602,
      "learning_rate": 0.0005612385238125997,
      "loss": 2.1273,
      "step": 37695
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6428577899932861,
      "learning_rate": 0.0005612365126792014,
      "loss": 2.0853,
      "step": 37696
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.7372482419013977,
      "learning_rate": 0.0005612345014972344,
      "loss": 2.3942,
      "step": 37697
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.715090274810791,
      "learning_rate": 0.000561232490266699,
      "loss": 2.1179,
      "step": 37698
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6806427836418152,
      "learning_rate": 0.0005612304789875955,
      "loss": 2.121,
      "step": 37699
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6847140192985535,
      "learning_rate": 0.0005612284676599244,
      "loss": 2.3727,
      "step": 37700
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6975948810577393,
      "learning_rate": 0.0005612264562836859,
      "loss": 2.1327,
      "step": 37701
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6208162307739258,
      "learning_rate": 0.0005612244448588806,
      "loss": 2.2495,
      "step": 37702
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6183698177337646,
      "learning_rate": 0.0005612224333855085,
      "loss": 2.3206,
      "step": 37703
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.625568151473999,
      "learning_rate": 0.0005612204218635704,
      "loss": 2.3425,
      "step": 37704
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6533669233322144,
      "learning_rate": 0.0005612184102930665,
      "loss": 2.5056,
      "step": 37705
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6614344716072083,
      "learning_rate": 0.000561216398673997,
      "loss": 2.2946,
      "step": 37706
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6096552014350891,
      "learning_rate": 0.0005612143870063626,
      "loss": 2.2825,
      "step": 37707
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6200394034385681,
      "learning_rate": 0.0005612123752901635,
      "loss": 2.26,
      "step": 37708
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.622339129447937,
      "learning_rate": 0.0005612103635254,
      "loss": 2.405,
      "step": 37709
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.652124285697937,
      "learning_rate": 0.0005612083517120728,
      "loss": 2.174,
      "step": 37710
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.7149948477745056,
      "learning_rate": 0.0005612063398501818,
      "loss": 2.2084,
      "step": 37711
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6404076218605042,
      "learning_rate": 0.0005612043279397277,
      "loss": 2.3846,
      "step": 37712
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6842264533042908,
      "learning_rate": 0.0005612023159807108,
      "loss": 2.0727,
      "step": 37713
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6471533179283142,
      "learning_rate": 0.0005612003039731314,
      "loss": 2.3278,
      "step": 37714
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6657001376152039,
      "learning_rate": 0.0005611982919169899,
      "loss": 2.2011,
      "step": 37715
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.7306596636772156,
      "learning_rate": 0.0005611962798122868,
      "loss": 2.3875,
      "step": 37716
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.7108176946640015,
      "learning_rate": 0.0005611942676590223,
      "loss": 2.3482,
      "step": 37717
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6639987826347351,
      "learning_rate": 0.0005611922554571969,
      "loss": 2.3959,
      "step": 37718
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6749515533447266,
      "learning_rate": 0.000561190243206811,
      "loss": 2.3731,
      "step": 37719
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6767362952232361,
      "learning_rate": 0.0005611882309078648,
      "loss": 2.3661,
      "step": 37720
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.7260933518409729,
      "learning_rate": 0.0005611862185603587,
      "loss": 2.267,
      "step": 37721
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6500540375709534,
      "learning_rate": 0.0005611842061642934,
      "loss": 2.234,
      "step": 37722
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.7256674766540527,
      "learning_rate": 0.0005611821937196688,
      "loss": 2.3766,
      "step": 37723
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6536265015602112,
      "learning_rate": 0.0005611801812264856,
      "loss": 2.1113,
      "step": 37724
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6421492695808411,
      "learning_rate": 0.0005611781686847441,
      "loss": 2.28,
      "step": 37725
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6665130257606506,
      "learning_rate": 0.0005611761560944446,
      "loss": 2.0624,
      "step": 37726
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6653071045875549,
      "learning_rate": 0.0005611741434555876,
      "loss": 2.2442,
      "step": 37727
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.658584475517273,
      "learning_rate": 0.0005611721307681733,
      "loss": 2.2394,
      "step": 37728
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6273598670959473,
      "learning_rate": 0.0005611701180322022,
      "loss": 2.1559,
      "step": 37729
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.670012354850769,
      "learning_rate": 0.0005611681052476746,
      "loss": 2.2311,
      "step": 37730
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.7089611887931824,
      "learning_rate": 0.0005611660924145909,
      "loss": 2.0717,
      "step": 37731
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6873772144317627,
      "learning_rate": 0.0005611640795329517,
      "loss": 2.6036,
      "step": 37732
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6909991502761841,
      "learning_rate": 0.000561162066602757,
      "loss": 2.358,
      "step": 37733
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6309951543807983,
      "learning_rate": 0.0005611600536240074,
      "loss": 2.3576,
      "step": 37734
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6105477809906006,
      "learning_rate": 0.0005611580405967032,
      "loss": 2.0531,
      "step": 37735
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.7139399647712708,
      "learning_rate": 0.0005611560275208447,
      "loss": 2.3195,
      "step": 37736
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.7150512337684631,
      "learning_rate": 0.0005611540143964324,
      "loss": 2.3592,
      "step": 37737
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6440985202789307,
      "learning_rate": 0.0005611520012234667,
      "loss": 2.158,
      "step": 37738
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6771897673606873,
      "learning_rate": 0.0005611499880019479,
      "loss": 2.1753,
      "step": 37739
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.7214100360870361,
      "learning_rate": 0.0005611479747318764,
      "loss": 2.4234,
      "step": 37740
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.7522709369659424,
      "learning_rate": 0.0005611459614132526,
      "loss": 2.4919,
      "step": 37741
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6984639167785645,
      "learning_rate": 0.0005611439480460767,
      "loss": 2.3285,
      "step": 37742
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.7137762308120728,
      "learning_rate": 0.0005611419346303493,
      "loss": 2.3182,
      "step": 37743
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6483027338981628,
      "learning_rate": 0.0005611399211660707,
      "loss": 2.2489,
      "step": 37744
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.7827891111373901,
      "learning_rate": 0.0005611379076532412,
      "loss": 2.2412,
      "step": 37745
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6894048452377319,
      "learning_rate": 0.0005611358940918612,
      "loss": 2.0276,
      "step": 37746
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6949002146720886,
      "learning_rate": 0.0005611338804819313,
      "loss": 2.2796,
      "step": 37747
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.7020936608314514,
      "learning_rate": 0.0005611318668234514,
      "loss": 2.2536,
      "step": 37748
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.7439521551132202,
      "learning_rate": 0.0005611298531164224,
      "loss": 2.2983,
      "step": 37749
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6550353765487671,
      "learning_rate": 0.0005611278393608443,
      "loss": 2.5105,
      "step": 37750
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6391058564186096,
      "learning_rate": 0.0005611258255567177,
      "loss": 2.2817,
      "step": 37751
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.7326897382736206,
      "learning_rate": 0.0005611238117040428,
      "loss": 2.2705,
      "step": 37752
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.730648398399353,
      "learning_rate": 0.0005611217978028202,
      "loss": 2.1842,
      "step": 37753
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6691539883613586,
      "learning_rate": 0.0005611197838530499,
      "loss": 2.1031,
      "step": 37754
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6538702845573425,
      "learning_rate": 0.0005611177698547327,
      "loss": 2.282,
      "step": 37755
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6145812273025513,
      "learning_rate": 0.0005611157558078686,
      "loss": 2.2033,
      "step": 37756
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.7361732125282288,
      "learning_rate": 0.0005611137417124583,
      "loss": 2.2425,
      "step": 37757
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6925920844078064,
      "learning_rate": 0.0005611117275685019,
      "loss": 2.2405,
      "step": 37758
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.7001939415931702,
      "learning_rate": 0.000561109713376,
      "loss": 2.2321,
      "step": 37759
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6872130632400513,
      "learning_rate": 0.0005611076991349529,
      "loss": 2.2895,
      "step": 37760
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6421990990638733,
      "learning_rate": 0.0005611056848453609,
      "loss": 2.3334,
      "step": 37761
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.7139462828636169,
      "learning_rate": 0.0005611036705072244,
      "loss": 2.3938,
      "step": 37762
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.670088529586792,
      "learning_rate": 0.0005611016561205438,
      "loss": 2.2414,
      "step": 37763
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.7752105593681335,
      "learning_rate": 0.0005610996416853195,
      "loss": 2.5631,
      "step": 37764
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.7529062032699585,
      "learning_rate": 0.0005610976272015519,
      "loss": 2.4857,
      "step": 37765
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.758622407913208,
      "learning_rate": 0.0005610956126692412,
      "loss": 2.3259,
      "step": 37766
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.9769017696380615,
      "learning_rate": 0.000561093598088388,
      "loss": 2.162,
      "step": 37767
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.7291423678398132,
      "learning_rate": 0.0005610915834589925,
      "loss": 2.3715,
      "step": 37768
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6517329812049866,
      "learning_rate": 0.0005610895687810552,
      "loss": 2.1461,
      "step": 37769
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6663917303085327,
      "learning_rate": 0.0005610875540545764,
      "loss": 2.15,
      "step": 37770
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6585208177566528,
      "learning_rate": 0.0005610855392795565,
      "loss": 2.2424,
      "step": 37771
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6726970076560974,
      "learning_rate": 0.0005610835244559959,
      "loss": 2.39,
      "step": 37772
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6635955572128296,
      "learning_rate": 0.0005610815095838949,
      "loss": 2.5176,
      "step": 37773
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6999069452285767,
      "learning_rate": 0.000561079494663254,
      "loss": 2.1422,
      "step": 37774
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.702017605304718,
      "learning_rate": 0.0005610774796940733,
      "loss": 2.3563,
      "step": 37775
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.7190110087394714,
      "learning_rate": 0.0005610754646763535,
      "loss": 2.2333,
      "step": 37776
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6846656799316406,
      "learning_rate": 0.0005610734496100949,
      "loss": 2.0686,
      "step": 37777
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6897249817848206,
      "learning_rate": 0.0005610714344952976,
      "loss": 2.3596,
      "step": 37778
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6111617684364319,
      "learning_rate": 0.0005610694193319624,
      "loss": 2.336,
      "step": 37779
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6749706864356995,
      "learning_rate": 0.0005610674041200894,
      "loss": 2.1741,
      "step": 37780
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6466999053955078,
      "learning_rate": 0.0005610653888596792,
      "loss": 2.5397,
      "step": 37781
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6406632661819458,
      "learning_rate": 0.0005610633735507318,
      "loss": 2.3949,
      "step": 37782
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.7207496166229248,
      "learning_rate": 0.0005610613581932478,
      "loss": 2.3984,
      "step": 37783
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6437670588493347,
      "learning_rate": 0.0005610593427872277,
      "loss": 2.2213,
      "step": 37784
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6880309581756592,
      "learning_rate": 0.0005610573273326716,
      "loss": 2.3534,
      "step": 37785
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6519973278045654,
      "learning_rate": 0.0005610553118295801,
      "loss": 2.1276,
      "step": 37786
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.7491362690925598,
      "learning_rate": 0.0005610532962779534,
      "loss": 2.2819,
      "step": 37787
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6411536335945129,
      "learning_rate": 0.0005610512806777921,
      "loss": 2.2816,
      "step": 37788
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6839237809181213,
      "learning_rate": 0.0005610492650290964,
      "loss": 2.3934,
      "step": 37789
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6738080978393555,
      "learning_rate": 0.0005610472493318666,
      "loss": 2.2696,
      "step": 37790
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.7031828761100769,
      "learning_rate": 0.0005610452335861033,
      "loss": 2.3383,
      "step": 37791
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6597232818603516,
      "learning_rate": 0.0005610432177918068,
      "loss": 2.3115,
      "step": 37792
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6943860650062561,
      "learning_rate": 0.0005610412019489774,
      "loss": 2.3604,
      "step": 37793
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6343585252761841,
      "learning_rate": 0.0005610391860576154,
      "loss": 2.3629,
      "step": 37794
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6988240480422974,
      "learning_rate": 0.0005610371701177215,
      "loss": 2.4206,
      "step": 37795
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6569712162017822,
      "learning_rate": 0.0005610351541292957,
      "loss": 2.0164,
      "step": 37796
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.651304304599762,
      "learning_rate": 0.0005610331380923385,
      "loss": 2.2235,
      "step": 37797
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6781763434410095,
      "learning_rate": 0.0005610311220068504,
      "loss": 2.0557,
      "step": 37798
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6551381945610046,
      "learning_rate": 0.0005610291058728318,
      "loss": 2.2157,
      "step": 37799
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.7057421207427979,
      "learning_rate": 0.0005610270896902827,
      "loss": 2.4365,
      "step": 37800
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.8366444706916809,
      "learning_rate": 0.0005610250734592039,
      "loss": 2.249,
      "step": 37801
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6540954113006592,
      "learning_rate": 0.0005610230571795956,
      "loss": 2.0863,
      "step": 37802
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6705397367477417,
      "learning_rate": 0.0005610210408514582,
      "loss": 2.1675,
      "step": 37803
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.7108331322669983,
      "learning_rate": 0.000561019024474792,
      "loss": 2.0959,
      "step": 37804
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6695470213890076,
      "learning_rate": 0.0005610170080495975,
      "loss": 2.1034,
      "step": 37805
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.681393563747406,
      "learning_rate": 0.000561014991575875,
      "loss": 2.1477,
      "step": 37806
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6756528615951538,
      "learning_rate": 0.0005610129750536248,
      "loss": 2.5535,
      "step": 37807
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6965740919113159,
      "learning_rate": 0.0005610109584828476,
      "loss": 2.3394,
      "step": 37808
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.7953941822052002,
      "learning_rate": 0.0005610089418635433,
      "loss": 2.3119,
      "step": 37809
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6945398449897766,
      "learning_rate": 0.0005610069251957125,
      "loss": 2.2896,
      "step": 37810
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6996011137962341,
      "learning_rate": 0.0005610049084793558,
      "loss": 2.3393,
      "step": 37811
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6900560855865479,
      "learning_rate": 0.0005610028917144731,
      "loss": 2.4109,
      "step": 37812
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.7260507941246033,
      "learning_rate": 0.0005610008749010652,
      "loss": 2.2026,
      "step": 37813
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6827998757362366,
      "learning_rate": 0.0005609988580391322,
      "loss": 2.5091,
      "step": 37814
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6668738722801208,
      "learning_rate": 0.0005609968411286747,
      "loss": 2.2771,
      "step": 37815
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6496357917785645,
      "learning_rate": 0.000560994824169693,
      "loss": 2.1195,
      "step": 37816
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6774831414222717,
      "learning_rate": 0.0005609928071621873,
      "loss": 2.2781,
      "step": 37817
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6633696556091309,
      "learning_rate": 0.0005609907901061582,
      "loss": 2.1425,
      "step": 37818
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6754912734031677,
      "learning_rate": 0.0005609887730016059,
      "loss": 2.2672,
      "step": 37819
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6378572583198547,
      "learning_rate": 0.0005609867558485309,
      "loss": 2.3439,
      "step": 37820
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6305454969406128,
      "learning_rate": 0.0005609847386469335,
      "loss": 2.0979,
      "step": 37821
    },
    {
      "epoch": 0.49,
      "grad_norm": 2.376490831375122,
      "learning_rate": 0.0005609827213968143,
      "loss": 2.3874,
      "step": 37822
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6624247431755066,
      "learning_rate": 0.0005609807040981732,
      "loss": 2.2564,
      "step": 37823
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.631744921207428,
      "learning_rate": 0.000560978686751011,
      "loss": 2.1551,
      "step": 37824
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.7548249363899231,
      "learning_rate": 0.000560976669355328,
      "loss": 2.4597,
      "step": 37825
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6376689076423645,
      "learning_rate": 0.0005609746519111245,
      "loss": 2.1632,
      "step": 37826
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.7344840168952942,
      "learning_rate": 0.0005609726344184007,
      "loss": 2.4396,
      "step": 37827
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.7131533026695251,
      "learning_rate": 0.0005609706168771574,
      "loss": 2.3621,
      "step": 37828
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6733720898628235,
      "learning_rate": 0.0005609685992873946,
      "loss": 2.2468,
      "step": 37829
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.7417361736297607,
      "learning_rate": 0.0005609665816491129,
      "loss": 2.1656,
      "step": 37830
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6895938515663147,
      "learning_rate": 0.0005609645639623125,
      "loss": 2.2317,
      "step": 37831
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.7005859017372131,
      "learning_rate": 0.0005609625462269938,
      "loss": 2.347,
      "step": 37832
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6396589279174805,
      "learning_rate": 0.0005609605284431574,
      "loss": 2.2457,
      "step": 37833
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.7153144478797913,
      "learning_rate": 0.0005609585106108034,
      "loss": 2.1059,
      "step": 37834
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.7287668585777283,
      "learning_rate": 0.0005609564927299323,
      "loss": 2.1799,
      "step": 37835
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6308786869049072,
      "learning_rate": 0.0005609544748005446,
      "loss": 2.2416,
      "step": 37836
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6306431293487549,
      "learning_rate": 0.0005609524568226403,
      "loss": 2.2942,
      "step": 37837
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.7043731212615967,
      "learning_rate": 0.0005609504387962202,
      "loss": 2.2987,
      "step": 37838
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6864960789680481,
      "learning_rate": 0.0005609484207212843,
      "loss": 2.3334,
      "step": 37839
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.81169593334198,
      "learning_rate": 0.0005609464025978333,
      "loss": 2.369,
      "step": 37840
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.703447699546814,
      "learning_rate": 0.0005609443844258675,
      "loss": 2.1659,
      "step": 37841
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6905799508094788,
      "learning_rate": 0.0005609423662053871,
      "loss": 2.0619,
      "step": 37842
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.693959653377533,
      "learning_rate": 0.0005609403479363927,
      "loss": 2.3043,
      "step": 37843
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6799330115318298,
      "learning_rate": 0.0005609383296188845,
      "loss": 2.1604,
      "step": 37844
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6708940267562866,
      "learning_rate": 0.0005609363112528629,
      "loss": 2.2774,
      "step": 37845
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.7141352891921997,
      "learning_rate": 0.0005609342928383284,
      "loss": 2.2465,
      "step": 37846
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.7605692744255066,
      "learning_rate": 0.0005609322743752813,
      "loss": 2.3786,
      "step": 37847
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6841753125190735,
      "learning_rate": 0.0005609302558637218,
      "loss": 2.1687,
      "step": 37848
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.731119692325592,
      "learning_rate": 0.0005609282373036505,
      "loss": 2.1429,
      "step": 37849
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.8405598998069763,
      "learning_rate": 0.0005609262186950677,
      "loss": 2.3293,
      "step": 37850
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.7326719760894775,
      "learning_rate": 0.0005609242000379739,
      "loss": 2.2868,
      "step": 37851
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.7248973846435547,
      "learning_rate": 0.0005609221813323693,
      "loss": 2.3603,
      "step": 37852
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.801002562046051,
      "learning_rate": 0.0005609201625782544,
      "loss": 2.2181,
      "step": 37853
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.7534433007240295,
      "learning_rate": 0.0005609181437756294,
      "loss": 2.4417,
      "step": 37854
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.7892282009124756,
      "learning_rate": 0.0005609161249244949,
      "loss": 2.2443,
      "step": 37855
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.7234025001525879,
      "learning_rate": 0.0005609141060248511,
      "loss": 2.3267,
      "step": 37856
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.7104951739311218,
      "learning_rate": 0.0005609120870766985,
      "loss": 2.359,
      "step": 37857
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.7055099606513977,
      "learning_rate": 0.0005609100680800375,
      "loss": 2.0158,
      "step": 37858
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.693158745765686,
      "learning_rate": 0.0005609080490348681,
      "loss": 1.9463,
      "step": 37859
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.7047538757324219,
      "learning_rate": 0.0005609060299411912,
      "loss": 2.3103,
      "step": 37860
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.7077449560165405,
      "learning_rate": 0.000560904010799007,
      "loss": 2.1906,
      "step": 37861
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6585671901702881,
      "learning_rate": 0.0005609019916083156,
      "loss": 2.1741,
      "step": 37862
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6188183426856995,
      "learning_rate": 0.0005608999723691177,
      "loss": 2.3161,
      "step": 37863
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.7016057372093201,
      "learning_rate": 0.0005608979530814137,
      "loss": 2.208,
      "step": 37864
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6710892915725708,
      "learning_rate": 0.0005608959337452036,
      "loss": 2.1268,
      "step": 37865
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6548070311546326,
      "learning_rate": 0.0005608939143604882,
      "loss": 2.2377,
      "step": 37866
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6469113826751709,
      "learning_rate": 0.0005608918949272676,
      "loss": 2.0622,
      "step": 37867
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.7160016894340515,
      "learning_rate": 0.0005608898754455423,
      "loss": 2.2907,
      "step": 37868
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6709016561508179,
      "learning_rate": 0.0005608878559153126,
      "loss": 2.2137,
      "step": 37869
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6515932083129883,
      "learning_rate": 0.0005608858363365791,
      "loss": 2.1378,
      "step": 37870
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6724448800086975,
      "learning_rate": 0.0005608838167093418,
      "loss": 2.2158,
      "step": 37871
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.7111849784851074,
      "learning_rate": 0.0005608817970336014,
      "loss": 2.2756,
      "step": 37872
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6712284684181213,
      "learning_rate": 0.0005608797773093581,
      "loss": 2.1779,
      "step": 37873
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.8086271286010742,
      "learning_rate": 0.0005608777575366123,
      "loss": 2.1936,
      "step": 37874
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.730462372303009,
      "learning_rate": 0.0005608757377153644,
      "loss": 2.4401,
      "step": 37875
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.7134233713150024,
      "learning_rate": 0.0005608737178456149,
      "loss": 2.3682,
      "step": 37876
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.8283994793891907,
      "learning_rate": 0.000560871697927364,
      "loss": 2.327,
      "step": 37877
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.617110013961792,
      "learning_rate": 0.000560869677960612,
      "loss": 2.2907,
      "step": 37878
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.8840574026107788,
      "learning_rate": 0.0005608676579453595,
      "loss": 2.441,
      "step": 37879
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.9232277870178223,
      "learning_rate": 0.0005608656378816068,
      "loss": 2.4543,
      "step": 37880
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.7399465441703796,
      "learning_rate": 0.0005608636177693542,
      "loss": 2.1873,
      "step": 37881
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.7945360541343689,
      "learning_rate": 0.0005608615976086021,
      "loss": 2.4083,
      "step": 37882
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.9939969778060913,
      "learning_rate": 0.000560859577399351,
      "loss": 2.1918,
      "step": 37883
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6732154488563538,
      "learning_rate": 0.0005608575571416011,
      "loss": 2.2979,
      "step": 37884
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.7827268838882446,
      "learning_rate": 0.0005608555368353529,
      "loss": 2.3095,
      "step": 37885
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.7701940536499023,
      "learning_rate": 0.0005608535164806068,
      "loss": 2.2971,
      "step": 37886
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6810305714607239,
      "learning_rate": 0.000560851496077363,
      "loss": 2.1719,
      "step": 37887
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.686882734298706,
      "learning_rate": 0.000560849475625622,
      "loss": 2.4211,
      "step": 37888
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6822036504745483,
      "learning_rate": 0.0005608474551253842,
      "loss": 1.9929,
      "step": 37889
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.7151395082473755,
      "learning_rate": 0.0005608454345766499,
      "loss": 2.1073,
      "step": 37890
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6688547134399414,
      "learning_rate": 0.0005608434139794194,
      "loss": 2.253,
      "step": 37891
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6954883933067322,
      "learning_rate": 0.0005608413933336934,
      "loss": 2.3459,
      "step": 37892
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.7736191749572754,
      "learning_rate": 0.000560839372639472,
      "loss": 2.4265,
      "step": 37893
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6585035920143127,
      "learning_rate": 0.0005608373518967556,
      "loss": 2.1921,
      "step": 37894
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6429094672203064,
      "learning_rate": 0.0005608353311055445,
      "loss": 2.1947,
      "step": 37895
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6669584512710571,
      "learning_rate": 0.0005608333102658393,
      "loss": 2.3749,
      "step": 37896
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.7543801069259644,
      "learning_rate": 0.0005608312893776404,
      "loss": 2.4991,
      "step": 37897
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6670674681663513,
      "learning_rate": 0.0005608292684409478,
      "loss": 2.4132,
      "step": 37898
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.8043797612190247,
      "learning_rate": 0.0005608272474557622,
      "loss": 2.3146,
      "step": 37899
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6518601775169373,
      "learning_rate": 0.0005608252264220839,
      "loss": 2.4199,
      "step": 37900
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.7542564868927002,
      "learning_rate": 0.0005608232053399133,
      "loss": 2.2881,
      "step": 37901
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.7807925343513489,
      "learning_rate": 0.0005608211842092507,
      "loss": 2.4562,
      "step": 37902
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6397162079811096,
      "learning_rate": 0.0005608191630300965,
      "loss": 2.2996,
      "step": 37903
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6760733723640442,
      "learning_rate": 0.0005608171418024512,
      "loss": 2.2335,
      "step": 37904
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6817347407341003,
      "learning_rate": 0.000560815120526315,
      "loss": 2.314,
      "step": 37905
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6310295462608337,
      "learning_rate": 0.0005608130992016884,
      "loss": 2.383,
      "step": 37906
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.7221367955207825,
      "learning_rate": 0.0005608110778285716,
      "loss": 2.14,
      "step": 37907
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6404227614402771,
      "learning_rate": 0.0005608090564069651,
      "loss": 2.2168,
      "step": 37908
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6561281681060791,
      "learning_rate": 0.0005608070349368695,
      "loss": 2.164,
      "step": 37909
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.7480064034461975,
      "learning_rate": 0.0005608050134182848,
      "loss": 2.3689,
      "step": 37910
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.7070189118385315,
      "learning_rate": 0.0005608029918512114,
      "loss": 2.4257,
      "step": 37911
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.7110480666160583,
      "learning_rate": 0.0005608009702356499,
      "loss": 2.5074,
      "step": 37912
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.7188807725906372,
      "learning_rate": 0.0005607989485716005,
      "loss": 2.2188,
      "step": 37913
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6995511054992676,
      "learning_rate": 0.0005607969268590639,
      "loss": 2.2223,
      "step": 37914
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.7994273900985718,
      "learning_rate": 0.0005607949050980401,
      "loss": 2.27,
      "step": 37915
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.7978991270065308,
      "learning_rate": 0.0005607928832885295,
      "loss": 2.3282,
      "step": 37916
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.0190820693969727,
      "learning_rate": 0.0005607908614305327,
      "loss": 2.1427,
      "step": 37917
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.8095455169677734,
      "learning_rate": 0.0005607888395240499,
      "loss": 2.4645,
      "step": 37918
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.8960317373275757,
      "learning_rate": 0.0005607868175690815,
      "loss": 2.2797,
      "step": 37919
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.7538712024688721,
      "learning_rate": 0.000560784795565628,
      "loss": 2.3994,
      "step": 37920
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.8501900434494019,
      "learning_rate": 0.0005607827735136896,
      "loss": 2.208,
      "step": 37921
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.7784247398376465,
      "learning_rate": 0.0005607807514132668,
      "loss": 2.3878,
      "step": 37922
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.7176558971405029,
      "learning_rate": 0.00056077872926436,
      "loss": 2.1769,
      "step": 37923
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.747873842716217,
      "learning_rate": 0.0005607767070669695,
      "loss": 2.1865,
      "step": 37924
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6445989608764648,
      "learning_rate": 0.0005607746848210956,
      "loss": 2.2699,
      "step": 37925
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6755197644233704,
      "learning_rate": 0.0005607726625267387,
      "loss": 2.1454,
      "step": 37926
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6779192686080933,
      "learning_rate": 0.0005607706401838993,
      "loss": 2.2014,
      "step": 37927
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6742244362831116,
      "learning_rate": 0.0005607686177925778,
      "loss": 2.3329,
      "step": 37928
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.7183065414428711,
      "learning_rate": 0.0005607665953527744,
      "loss": 2.1683,
      "step": 37929
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.5997470617294312,
      "learning_rate": 0.0005607645728644897,
      "loss": 2.2098,
      "step": 37930
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.7178153991699219,
      "learning_rate": 0.0005607625503277238,
      "loss": 2.474,
      "step": 37931
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.7314560413360596,
      "learning_rate": 0.0005607605277424773,
      "loss": 2.2071,
      "step": 37932
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6647787094116211,
      "learning_rate": 0.0005607585051087505,
      "loss": 2.1344,
      "step": 37933
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6550585031509399,
      "learning_rate": 0.0005607564824265437,
      "loss": 2.0904,
      "step": 37934
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6977106332778931,
      "learning_rate": 0.0005607544596958573,
      "loss": 2.2134,
      "step": 37935
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.716064989566803,
      "learning_rate": 0.0005607524369166918,
      "loss": 2.21,
      "step": 37936
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6658298373222351,
      "learning_rate": 0.0005607504140890476,
      "loss": 2.2054,
      "step": 37937
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6438281536102295,
      "learning_rate": 0.0005607483912129249,
      "loss": 2.2283,
      "step": 37938
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.701921284198761,
      "learning_rate": 0.0005607463682883242,
      "loss": 2.349,
      "step": 37939
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6317594647407532,
      "learning_rate": 0.0005607443453152456,
      "loss": 2.2997,
      "step": 37940
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.7047269344329834,
      "learning_rate": 0.0005607423222936899,
      "loss": 2.4227,
      "step": 37941
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6255375742912292,
      "learning_rate": 0.0005607402992236574,
      "loss": 2.2383,
      "step": 37942
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.7792580723762512,
      "learning_rate": 0.000560738276105148,
      "loss": 2.1553,
      "step": 37943
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6767216324806213,
      "learning_rate": 0.0005607362529381628,
      "loss": 2.3091,
      "step": 37944
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6873194575309753,
      "learning_rate": 0.0005607342297227016,
      "loss": 2.0058,
      "step": 37945
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.7130531668663025,
      "learning_rate": 0.000560732206458765,
      "loss": 2.1779,
      "step": 37946
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.7956541180610657,
      "learning_rate": 0.0005607301831463535,
      "loss": 2.3248,
      "step": 37947
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.715013325214386,
      "learning_rate": 0.0005607281597854672,
      "loss": 2.2416,
      "step": 37948
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.7878881096839905,
      "learning_rate": 0.0005607261363761066,
      "loss": 2.0327,
      "step": 37949
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.8198874592781067,
      "learning_rate": 0.0005607241129182721,
      "loss": 2.2442,
      "step": 37950
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.7011236548423767,
      "learning_rate": 0.0005607220894119641,
      "loss": 2.3501,
      "step": 37951
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.8257647156715393,
      "learning_rate": 0.000560720065857183,
      "loss": 2.23,
      "step": 37952
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.7428568005561829,
      "learning_rate": 0.0005607180422539291,
      "loss": 2.2192,
      "step": 37953
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6099473834037781,
      "learning_rate": 0.0005607160186022027,
      "loss": 2.1282,
      "step": 37954
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.7591986656188965,
      "learning_rate": 0.0005607139949020043,
      "loss": 2.2789,
      "step": 37955
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.7646490335464478,
      "learning_rate": 0.0005607119711533344,
      "loss": 2.436,
      "step": 37956
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6378135681152344,
      "learning_rate": 0.000560709947356193,
      "loss": 2.2561,
      "step": 37957
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6554786562919617,
      "learning_rate": 0.0005607079235105809,
      "loss": 2.1242,
      "step": 37958
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6798957586288452,
      "learning_rate": 0.0005607058996164982,
      "loss": 2.0455,
      "step": 37959
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.7649333477020264,
      "learning_rate": 0.0005607038756739452,
      "loss": 2.2864,
      "step": 37960
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6657949686050415,
      "learning_rate": 0.0005607018516829226,
      "loss": 2.2605,
      "step": 37961
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6384091377258301,
      "learning_rate": 0.0005606998276434306,
      "loss": 2.445,
      "step": 37962
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.7313396334648132,
      "learning_rate": 0.0005606978035554695,
      "loss": 2.2266,
      "step": 37963
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.629669189453125,
      "learning_rate": 0.0005606957794190398,
      "loss": 2.127,
      "step": 37964
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6910532116889954,
      "learning_rate": 0.0005606937552341418,
      "loss": 2.2897,
      "step": 37965
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6757238507270813,
      "learning_rate": 0.000560691731000776,
      "loss": 2.302,
      "step": 37966
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6712712049484253,
      "learning_rate": 0.0005606897067189425,
      "loss": 2.1815,
      "step": 37967
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.7434363961219788,
      "learning_rate": 0.000560687682388642,
      "loss": 2.2894,
      "step": 37968
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6579272747039795,
      "learning_rate": 0.0005606856580098748,
      "loss": 2.245,
      "step": 37969
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6301105618476868,
      "learning_rate": 0.0005606836335826412,
      "loss": 2.3654,
      "step": 37970
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6369659304618835,
      "learning_rate": 0.0005606816091069415,
      "loss": 2.3012,
      "step": 37971
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6518303751945496,
      "learning_rate": 0.0005606795845827762,
      "loss": 2.1118,
      "step": 37972
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6272497177124023,
      "learning_rate": 0.0005606775600101456,
      "loss": 2.2964,
      "step": 37973
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6801986694335938,
      "learning_rate": 0.0005606755353890502,
      "loss": 2.2555,
      "step": 37974
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6875463724136353,
      "learning_rate": 0.0005606735107194903,
      "loss": 2.2724,
      "step": 37975
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6363338232040405,
      "learning_rate": 0.0005606714860014663,
      "loss": 2.4728,
      "step": 37976
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6545178294181824,
      "learning_rate": 0.0005606694612349785,
      "loss": 2.4087,
      "step": 37977
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6618234515190125,
      "learning_rate": 0.0005606674364200272,
      "loss": 2.3144,
      "step": 37978
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.7209722995758057,
      "learning_rate": 0.000560665411556613,
      "loss": 2.298,
      "step": 37979
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6909998059272766,
      "learning_rate": 0.0005606633866447363,
      "loss": 2.2081,
      "step": 37980
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6028268933296204,
      "learning_rate": 0.0005606613616843972,
      "loss": 2.2595,
      "step": 37981
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6548976302146912,
      "learning_rate": 0.0005606593366755962,
      "loss": 2.1125,
      "step": 37982
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6812949776649475,
      "learning_rate": 0.0005606573116183339,
      "loss": 2.3803,
      "step": 37983
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6486790180206299,
      "learning_rate": 0.0005606552865126103,
      "loss": 2.3328,
      "step": 37984
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6186292767524719,
      "learning_rate": 0.000560653261358426,
      "loss": 2.2198,
      "step": 37985
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6643728017807007,
      "learning_rate": 0.0005606512361557815,
      "loss": 2.2035,
      "step": 37986
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.7690510153770447,
      "learning_rate": 0.0005606492109046768,
      "loss": 2.3418,
      "step": 37987
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.66590815782547,
      "learning_rate": 0.0005606471856051127,
      "loss": 2.4667,
      "step": 37988
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.7209389805793762,
      "learning_rate": 0.0005606451602570892,
      "loss": 2.3209,
      "step": 37989
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.7173844575881958,
      "learning_rate": 0.0005606431348606069,
      "loss": 2.299,
      "step": 37990
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.7635062336921692,
      "learning_rate": 0.0005606411094156662,
      "loss": 2.2354,
      "step": 37991
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.5959138870239258,
      "learning_rate": 0.0005606390839222673,
      "loss": 2.4953,
      "step": 37992
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6267769932746887,
      "learning_rate": 0.0005606370583804106,
      "loss": 2.2453,
      "step": 37993
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6716712713241577,
      "learning_rate": 0.0005606350327900967,
      "loss": 2.5467,
      "step": 37994
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.7226990461349487,
      "learning_rate": 0.0005606330071513258,
      "loss": 2.5779,
      "step": 37995
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6426090002059937,
      "learning_rate": 0.0005606309814640983,
      "loss": 2.1331,
      "step": 37996
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.7209743857383728,
      "learning_rate": 0.0005606289557284145,
      "loss": 2.1151,
      "step": 37997
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.7034037113189697,
      "learning_rate": 0.000560626929944275,
      "loss": 2.3203,
      "step": 37998
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6288338899612427,
      "learning_rate": 0.0005606249041116799,
      "loss": 2.0337,
      "step": 37999
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6535874009132385,
      "learning_rate": 0.0005606228782306297,
      "loss": 2.1805,
      "step": 38000
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.799278974533081,
      "learning_rate": 0.0005606208523011249,
      "loss": 2.3839,
      "step": 38001
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6957241296768188,
      "learning_rate": 0.0005606188263231657,
      "loss": 2.4006,
      "step": 38002
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.7095125317573547,
      "learning_rate": 0.0005606168002967526,
      "loss": 2.3662,
      "step": 38003
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.7493277788162231,
      "learning_rate": 0.0005606147742218858,
      "loss": 2.2797,
      "step": 38004
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.8253780603408813,
      "learning_rate": 0.0005606127480985659,
      "loss": 2.2154,
      "step": 38005
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.720408022403717,
      "learning_rate": 0.0005606107219267931,
      "loss": 2.3602,
      "step": 38006
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.7194609642028809,
      "learning_rate": 0.0005606086957065679,
      "loss": 2.123,
      "step": 38007
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.9008349776268005,
      "learning_rate": 0.0005606066694378907,
      "loss": 2.3848,
      "step": 38008
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.8858307600021362,
      "learning_rate": 0.0005606046431207616,
      "loss": 2.1166,
      "step": 38009
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.757698655128479,
      "learning_rate": 0.0005606026167551813,
      "loss": 2.4245,
      "step": 38010
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.8281235098838806,
      "learning_rate": 0.0005606005903411501,
      "loss": 2.3337,
      "step": 38011
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.7787169814109802,
      "learning_rate": 0.0005605985638786682,
      "loss": 2.3381,
      "step": 38012
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.8173893094062805,
      "learning_rate": 0.0005605965373677361,
      "loss": 2.3783,
      "step": 38013
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.7651525735855103,
      "learning_rate": 0.0005605945108083544,
      "loss": 2.3077,
      "step": 38014
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.7321875691413879,
      "learning_rate": 0.0005605924842005232,
      "loss": 2.2056,
      "step": 38015
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6885012984275818,
      "learning_rate": 0.0005605904575442428,
      "loss": 2.3443,
      "step": 38016
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.7073653936386108,
      "learning_rate": 0.0005605884308395138,
      "loss": 2.3731,
      "step": 38017
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.728943407535553,
      "learning_rate": 0.0005605864040863364,
      "loss": 2.2758,
      "step": 38018
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.7465516924858093,
      "learning_rate": 0.0005605843772847111,
      "loss": 2.4101,
      "step": 38019
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.7982501983642578,
      "learning_rate": 0.0005605823504346382,
      "loss": 2.2349,
      "step": 38020
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6699472665786743,
      "learning_rate": 0.0005605803235361183,
      "loss": 2.1806,
      "step": 38021
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6428279876708984,
      "learning_rate": 0.0005605782965891515,
      "loss": 2.3181,
      "step": 38022
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.682067334651947,
      "learning_rate": 0.0005605762695937383,
      "loss": 2.1878,
      "step": 38023
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6699610352516174,
      "learning_rate": 0.000560574242549879,
      "loss": 2.3875,
      "step": 38024
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.8892242908477783,
      "learning_rate": 0.000560572215457574,
      "loss": 2.3688,
      "step": 38025
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6742346882820129,
      "learning_rate": 0.0005605701883168239,
      "loss": 2.1777,
      "step": 38026
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.9178432822227478,
      "learning_rate": 0.0005605681611276286,
      "loss": 2.3118,
      "step": 38027
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6580013036727905,
      "learning_rate": 0.000560566133889989,
      "loss": 2.1684,
      "step": 38028
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.70588219165802,
      "learning_rate": 0.0005605641066039051,
      "loss": 2.228,
      "step": 38029
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.7281204462051392,
      "learning_rate": 0.0005605620792693775,
      "loss": 2.2831,
      "step": 38030
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.9772213101387024,
      "learning_rate": 0.0005605600518864065,
      "loss": 2.445,
      "step": 38031
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.8450093865394592,
      "learning_rate": 0.0005605580244549925,
      "loss": 2.1523,
      "step": 38032
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6834134459495544,
      "learning_rate": 0.0005605559969751356,
      "loss": 2.5118,
      "step": 38033
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.8634653091430664,
      "learning_rate": 0.0005605539694468367,
      "loss": 1.9639,
      "step": 38034
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.8614579439163208,
      "learning_rate": 0.0005605519418700958,
      "loss": 2.2951,
      "step": 38035
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.8680568933486938,
      "learning_rate": 0.0005605499142449133,
      "loss": 2.4523,
      "step": 38036
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.7196474075317383,
      "learning_rate": 0.0005605478865712897,
      "loss": 2.2324,
      "step": 38037
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.7616041898727417,
      "learning_rate": 0.0005605458588492254,
      "loss": 2.2523,
      "step": 38038
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.7668750286102295,
      "learning_rate": 0.0005605438310787206,
      "loss": 2.1774,
      "step": 38039
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.774235725402832,
      "learning_rate": 0.0005605418032597759,
      "loss": 2.4759,
      "step": 38040
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.7541597485542297,
      "learning_rate": 0.0005605397753923914,
      "loss": 2.1218,
      "step": 38041
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6963039040565491,
      "learning_rate": 0.0005605377474765678,
      "loss": 2.2523,
      "step": 38042
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6945394277572632,
      "learning_rate": 0.0005605357195123052,
      "loss": 2.1903,
      "step": 38043
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6874582767486572,
      "learning_rate": 0.0005605336914996042,
      "loss": 2.3586,
      "step": 38044
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6733995079994202,
      "learning_rate": 0.0005605316634384649,
      "loss": 2.4377,
      "step": 38045
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.688798725605011,
      "learning_rate": 0.000560529635328888,
      "loss": 2.0926,
      "step": 38046
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.7025111317634583,
      "learning_rate": 0.0005605276071708737,
      "loss": 2.2603,
      "step": 38047
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6853458285331726,
      "learning_rate": 0.0005605255789644224,
      "loss": 2.2471,
      "step": 38048
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6493936777114868,
      "learning_rate": 0.0005605235507095345,
      "loss": 2.3534,
      "step": 38049
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6453539133071899,
      "learning_rate": 0.0005605215224062102,
      "loss": 2.3304,
      "step": 38050
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.7568485140800476,
      "learning_rate": 0.0005605194940544502,
      "loss": 2.4481,
      "step": 38051
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6897435784339905,
      "learning_rate": 0.0005605174656542546,
      "loss": 1.9939,
      "step": 38052
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6747114658355713,
      "learning_rate": 0.000560515437205624,
      "loss": 2.3306,
      "step": 38053
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.7086760997772217,
      "learning_rate": 0.0005605134087085585,
      "loss": 2.3862,
      "step": 38054
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6777170300483704,
      "learning_rate": 0.0005605113801630588,
      "loss": 2.1388,
      "step": 38055
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.701189398765564,
      "learning_rate": 0.0005605093515691251,
      "loss": 2.0211,
      "step": 38056
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.629469096660614,
      "learning_rate": 0.0005605073229267577,
      "loss": 2.0453,
      "step": 38057
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6889302134513855,
      "learning_rate": 0.0005605052942359571,
      "loss": 2.3365,
      "step": 38058
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6443819403648376,
      "learning_rate": 0.0005605032654967237,
      "loss": 2.1527,
      "step": 38059
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6974906921386719,
      "learning_rate": 0.0005605012367090578,
      "loss": 2.2741,
      "step": 38060
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6681499481201172,
      "learning_rate": 0.0005604992078729598,
      "loss": 2.1321,
      "step": 38061
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6890664100646973,
      "learning_rate": 0.0005604971789884301,
      "loss": 2.3271,
      "step": 38062
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.7313154339790344,
      "learning_rate": 0.000560495150055469,
      "loss": 2.2607,
      "step": 38063
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.7119373083114624,
      "learning_rate": 0.000560493121074077,
      "loss": 2.4343,
      "step": 38064
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6859877109527588,
      "learning_rate": 0.0005604910920442544,
      "loss": 2.5338,
      "step": 38065
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.7234519720077515,
      "learning_rate": 0.0005604890629660015,
      "loss": 2.1751,
      "step": 38066
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6642618775367737,
      "learning_rate": 0.0005604870338393189,
      "loss": 2.4162,
      "step": 38067
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6513638496398926,
      "learning_rate": 0.0005604850046642067,
      "loss": 2.48,
      "step": 38068
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6761201620101929,
      "learning_rate": 0.0005604829754406655,
      "loss": 2.3682,
      "step": 38069
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6752604246139526,
      "learning_rate": 0.0005604809461686956,
      "loss": 2.3097,
      "step": 38070
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6805171966552734,
      "learning_rate": 0.0005604789168482974,
      "loss": 2.2441,
      "step": 38071
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.5977177023887634,
      "learning_rate": 0.0005604768874794712,
      "loss": 2.2779,
      "step": 38072
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.7675999402999878,
      "learning_rate": 0.0005604748580622175,
      "loss": 2.3424,
      "step": 38073
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6921404600143433,
      "learning_rate": 0.0005604728285965366,
      "loss": 2.3031,
      "step": 38074
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.7917766571044922,
      "learning_rate": 0.0005604707990824288,
      "loss": 2.1095,
      "step": 38075
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.670674741268158,
      "learning_rate": 0.0005604687695198946,
      "loss": 2.0272,
      "step": 38076
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6549336314201355,
      "learning_rate": 0.0005604667399089344,
      "loss": 2.2366,
      "step": 38077
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6774708032608032,
      "learning_rate": 0.0005604647102495484,
      "loss": 2.3251,
      "step": 38078
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.7052452564239502,
      "learning_rate": 0.0005604626805417372,
      "loss": 2.2369,
      "step": 38079
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6584556102752686,
      "learning_rate": 0.0005604606507855009,
      "loss": 2.0319,
      "step": 38080
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6689745187759399,
      "learning_rate": 0.0005604586209808402,
      "loss": 2.2587,
      "step": 38081
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.8339603543281555,
      "learning_rate": 0.0005604565911277553,
      "loss": 2.2812,
      "step": 38082
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6881440281867981,
      "learning_rate": 0.0005604545612262466,
      "loss": 2.4358,
      "step": 38083
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.710896909236908,
      "learning_rate": 0.0005604525312763145,
      "loss": 2.3357,
      "step": 38084
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.7025466561317444,
      "learning_rate": 0.0005604505012779593,
      "loss": 2.3233,
      "step": 38085
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6527224779129028,
      "learning_rate": 0.0005604484712311815,
      "loss": 2.2388,
      "step": 38086
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6482433080673218,
      "learning_rate": 0.0005604464411359814,
      "loss": 2.1618,
      "step": 38087
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6280649304389954,
      "learning_rate": 0.0005604444109923594,
      "loss": 2.3244,
      "step": 38088
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6365891098976135,
      "learning_rate": 0.0005604423808003158,
      "loss": 2.253,
      "step": 38089
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.7399970293045044,
      "learning_rate": 0.0005604403505598512,
      "loss": 2.2638,
      "step": 38090
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.7193847298622131,
      "learning_rate": 0.0005604383202709657,
      "loss": 2.2645,
      "step": 38091
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6688013672828674,
      "learning_rate": 0.0005604362899336597,
      "loss": 2.2792,
      "step": 38092
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6973856687545776,
      "learning_rate": 0.0005604342595479339,
      "loss": 2.2428,
      "step": 38093
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6174642443656921,
      "learning_rate": 0.0005604322291137883,
      "loss": 2.202,
      "step": 38094
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6176730394363403,
      "learning_rate": 0.0005604301986312236,
      "loss": 2.3081,
      "step": 38095
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.7787083983421326,
      "learning_rate": 0.0005604281681002399,
      "loss": 2.2957,
      "step": 38096
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6346474289894104,
      "learning_rate": 0.0005604261375208376,
      "loss": 2.1672,
      "step": 38097
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.7408667206764221,
      "learning_rate": 0.0005604241068930171,
      "loss": 2.2624,
      "step": 38098
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6416932344436646,
      "learning_rate": 0.0005604220762167791,
      "loss": 2.1096,
      "step": 38099
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.7307824492454529,
      "learning_rate": 0.0005604200454921236,
      "loss": 2.1627,
      "step": 38100
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6812415719032288,
      "learning_rate": 0.0005604180147190511,
      "loss": 2.4631,
      "step": 38101
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6848996877670288,
      "learning_rate": 0.0005604159838975619,
      "loss": 2.2956,
      "step": 38102
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6487802267074585,
      "learning_rate": 0.0005604139530276566,
      "loss": 2.1189,
      "step": 38103
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6722686886787415,
      "learning_rate": 0.0005604119221093354,
      "loss": 2.348,
      "step": 38104
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.656960129737854,
      "learning_rate": 0.0005604098911425987,
      "loss": 2.0503,
      "step": 38105
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.8137362599372864,
      "learning_rate": 0.0005604078601274469,
      "loss": 2.0659,
      "step": 38106
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6428796648979187,
      "learning_rate": 0.0005604058290638803,
      "loss": 2.4491,
      "step": 38107
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.653541624546051,
      "learning_rate": 0.0005604037979518993,
      "loss": 2.4552,
      "step": 38108
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6734550595283508,
      "learning_rate": 0.0005604017667915044,
      "loss": 2.3216,
      "step": 38109
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.5670548677444458,
      "learning_rate": 0.0005603997355826958,
      "loss": 2.3757,
      "step": 38110
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.7659775018692017,
      "learning_rate": 0.0005603977043254742,
      "loss": 2.4327,
      "step": 38111
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6975340247154236,
      "learning_rate": 0.0005603956730198395,
      "loss": 2.4247,
      "step": 38112
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6791172027587891,
      "learning_rate": 0.0005603936416657924,
      "loss": 2.1474,
      "step": 38113
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6728511452674866,
      "learning_rate": 0.0005603916102633333,
      "loss": 2.2668,
      "step": 38114
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.7094720602035522,
      "learning_rate": 0.0005603895788124624,
      "loss": 2.3477,
      "step": 38115
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.7771416902542114,
      "learning_rate": 0.0005603875473131802,
      "loss": 2.3426,
      "step": 38116
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6451831459999084,
      "learning_rate": 0.000560385515765487,
      "loss": 2.2807,
      "step": 38117
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.5967247486114502,
      "learning_rate": 0.0005603834841693832,
      "loss": 2.0265,
      "step": 38118
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6844086647033691,
      "learning_rate": 0.0005603814525248692,
      "loss": 2.4831,
      "step": 38119
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.7142614126205444,
      "learning_rate": 0.0005603794208319454,
      "loss": 2.346,
      "step": 38120
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6771838068962097,
      "learning_rate": 0.0005603773890906122,
      "loss": 2.3554,
      "step": 38121
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6974306106567383,
      "learning_rate": 0.00056037535730087,
      "loss": 2.2392,
      "step": 38122
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.7278448939323425,
      "learning_rate": 0.0005603733254627189,
      "loss": 2.2957,
      "step": 38123
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6344622373580933,
      "learning_rate": 0.0005603712935761596,
      "loss": 2.3001,
      "step": 38124
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.8331124186515808,
      "learning_rate": 0.0005603692616411923,
      "loss": 2.3565,
      "step": 38125
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.7459753155708313,
      "learning_rate": 0.0005603672296578175,
      "loss": 2.1755,
      "step": 38126
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.9785122871398926,
      "learning_rate": 0.0005603651976260355,
      "loss": 2.2527,
      "step": 38127
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6546716094017029,
      "learning_rate": 0.0005603631655458467,
      "loss": 2.1803,
      "step": 38128
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.7762349247932434,
      "learning_rate": 0.0005603611334172514,
      "loss": 2.2198,
      "step": 38129
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.9013329744338989,
      "learning_rate": 0.0005603591012402502,
      "loss": 2.267,
      "step": 38130
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.688679575920105,
      "learning_rate": 0.0005603570690148433,
      "loss": 2.1852,
      "step": 38131
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6678965091705322,
      "learning_rate": 0.000560355036741031,
      "loss": 2.2405,
      "step": 38132
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.7685136795043945,
      "learning_rate": 0.0005603530044188138,
      "loss": 2.0091,
      "step": 38133
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.8436872959136963,
      "learning_rate": 0.0005603509720481921,
      "loss": 1.9152,
      "step": 38134
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6670327186584473,
      "learning_rate": 0.0005603489396291662,
      "loss": 2.0183,
      "step": 38135
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.9876212477684021,
      "learning_rate": 0.0005603469071617367,
      "loss": 2.3038,
      "step": 38136
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.8273606300354004,
      "learning_rate": 0.0005603448746459035,
      "loss": 2.2975,
      "step": 38137
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.757053554058075,
      "learning_rate": 0.0005603428420816674,
      "loss": 2.1211,
      "step": 38138
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6959875226020813,
      "learning_rate": 0.0005603408094690287,
      "loss": 2.4136,
      "step": 38139
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.8497105836868286,
      "learning_rate": 0.0005603387768079877,
      "loss": 2.2483,
      "step": 38140
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.7851129174232483,
      "learning_rate": 0.0005603367440985449,
      "loss": 2.2986,
      "step": 38141
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.7290197014808655,
      "learning_rate": 0.0005603347113407004,
      "loss": 2.3315,
      "step": 38142
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.7705696821212769,
      "learning_rate": 0.0005603326785344549,
      "loss": 2.5175,
      "step": 38143
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6984646916389465,
      "learning_rate": 0.0005603306456798085,
      "loss": 2.2428,
      "step": 38144
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.807989239692688,
      "learning_rate": 0.0005603286127767619,
      "loss": 2.0211,
      "step": 38145
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.7431900501251221,
      "learning_rate": 0.0005603265798253152,
      "loss": 2.3858,
      "step": 38146
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.7761949896812439,
      "learning_rate": 0.0005603245468254689,
      "loss": 2.3002,
      "step": 38147
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.948056161403656,
      "learning_rate": 0.0005603225137772234,
      "loss": 2.1282,
      "step": 38148
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6445595026016235,
      "learning_rate": 0.0005603204806805789,
      "loss": 2.4198,
      "step": 38149
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.8953972458839417,
      "learning_rate": 0.000560318447535536,
      "loss": 2.2847,
      "step": 38150
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.7746130228042603,
      "learning_rate": 0.000560316414342095,
      "loss": 2.2036,
      "step": 38151
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.7312358617782593,
      "learning_rate": 0.0005603143811002562,
      "loss": 2.1811,
      "step": 38152
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.7153947949409485,
      "learning_rate": 0.00056031234781002,
      "loss": 2.331,
      "step": 38153
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.7222707271575928,
      "learning_rate": 0.0005603103144713869,
      "loss": 2.3068,
      "step": 38154
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.7365224361419678,
      "learning_rate": 0.0005603082810843572,
      "loss": 2.1942,
      "step": 38155
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6480008363723755,
      "learning_rate": 0.0005603062476489313,
      "loss": 2.0745,
      "step": 38156
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6621220111846924,
      "learning_rate": 0.0005603042141651096,
      "loss": 2.3427,
      "step": 38157
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.687257707118988,
      "learning_rate": 0.0005603021806328924,
      "loss": 2.1709,
      "step": 38158
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.1938072443008423,
      "learning_rate": 0.00056030014705228,
      "loss": 2.2356,
      "step": 38159
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.7152337431907654,
      "learning_rate": 0.0005602981134232729,
      "loss": 2.3087,
      "step": 38160
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.9376441240310669,
      "learning_rate": 0.0005602960797458715,
      "loss": 2.34,
      "step": 38161
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.7157139182090759,
      "learning_rate": 0.0005602940460200762,
      "loss": 2.3373,
      "step": 38162
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6856179237365723,
      "learning_rate": 0.0005602920122458874,
      "loss": 2.4468,
      "step": 38163
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6941990256309509,
      "learning_rate": 0.0005602899784233052,
      "loss": 2.0829,
      "step": 38164
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.8727289438247681,
      "learning_rate": 0.0005602879445523304,
      "loss": 2.1599,
      "step": 38165
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.7658370137214661,
      "learning_rate": 0.0005602859106329629,
      "loss": 2.2551,
      "step": 38166
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.7060549259185791,
      "learning_rate": 0.0005602838766652035,
      "loss": 2.2926,
      "step": 38167
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6747076511383057,
      "learning_rate": 0.0005602818426490524,
      "loss": 2.3147,
      "step": 38168
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.7599310874938965,
      "learning_rate": 0.0005602798085845099,
      "loss": 2.2596,
      "step": 38169
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.7325190901756287,
      "learning_rate": 0.0005602777744715767,
      "loss": 2.3719,
      "step": 38170
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6710454821586609,
      "learning_rate": 0.0005602757403102527,
      "loss": 2.0835,
      "step": 38171
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6781418919563293,
      "learning_rate": 0.0005602737061005386,
      "loss": 2.3517,
      "step": 38172
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.7386615872383118,
      "learning_rate": 0.0005602716718424348,
      "loss": 2.1999,
      "step": 38173
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.7745988965034485,
      "learning_rate": 0.0005602696375359415,
      "loss": 2.3146,
      "step": 38174
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.7454653382301331,
      "learning_rate": 0.0005602676031810592,
      "loss": 2.402,
      "step": 38175
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.723376452922821,
      "learning_rate": 0.0005602655687777882,
      "loss": 2.3178,
      "step": 38176
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.7630828022956848,
      "learning_rate": 0.0005602635343261289,
      "loss": 1.9862,
      "step": 38177
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6991607546806335,
      "learning_rate": 0.0005602614998260816,
      "loss": 2.4164,
      "step": 38178
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6603198051452637,
      "learning_rate": 0.000560259465277647,
      "loss": 2.2825,
      "step": 38179
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6990208029747009,
      "learning_rate": 0.0005602574306808252,
      "loss": 2.2942,
      "step": 38180
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.8208426237106323,
      "learning_rate": 0.0005602553960356166,
      "loss": 2.2027,
      "step": 38181
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.699377715587616,
      "learning_rate": 0.0005602533613420215,
      "loss": 2.3684,
      "step": 38182
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.0473227500915527,
      "learning_rate": 0.0005602513266000405,
      "loss": 2.106,
      "step": 38183
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6434025168418884,
      "learning_rate": 0.0005602492918096738,
      "loss": 2.3525,
      "step": 38184
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6622870564460754,
      "learning_rate": 0.0005602472569709219,
      "loss": 2.1134,
      "step": 38185
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6586174368858337,
      "learning_rate": 0.0005602452220837852,
      "loss": 2.2261,
      "step": 38186
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6488991379737854,
      "learning_rate": 0.0005602431871482639,
      "loss": 2.3727,
      "step": 38187
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.748666524887085,
      "learning_rate": 0.0005602411521643584,
      "loss": 2.0805,
      "step": 38188
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.721082329750061,
      "learning_rate": 0.0005602391171320693,
      "loss": 2.07,
      "step": 38189
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.7689367532730103,
      "learning_rate": 0.0005602370820513968,
      "loss": 2.3791,
      "step": 38190
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.7068344354629517,
      "learning_rate": 0.0005602350469223412,
      "loss": 2.1489,
      "step": 38191
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6605749726295471,
      "learning_rate": 0.0005602330117449032,
      "loss": 2.1955,
      "step": 38192
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6610681414604187,
      "learning_rate": 0.0005602309765190828,
      "loss": 2.2203,
      "step": 38193
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.781521201133728,
      "learning_rate": 0.0005602289412448807,
      "loss": 2.4837,
      "step": 38194
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.683354377746582,
      "learning_rate": 0.0005602269059222971,
      "loss": 2.3765,
      "step": 38195
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.731208086013794,
      "learning_rate": 0.0005602248705513322,
      "loss": 2.3824,
      "step": 38196
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.695405900478363,
      "learning_rate": 0.0005602228351319868,
      "loss": 2.126,
      "step": 38197
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.7715744376182556,
      "learning_rate": 0.000560220799664261,
      "loss": 2.2534,
      "step": 38198
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6920599341392517,
      "learning_rate": 0.0005602187641481553,
      "loss": 2.3594,
      "step": 38199
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.7624616622924805,
      "learning_rate": 0.0005602167285836698,
      "loss": 2.067,
      "step": 38200
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6771604418754578,
      "learning_rate": 0.0005602146929708053,
      "loss": 2.2006,
      "step": 38201
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.736115038394928,
      "learning_rate": 0.000560212657309562,
      "loss": 2.3229,
      "step": 38202
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6749690175056458,
      "learning_rate": 0.00056021062159994,
      "loss": 2.4131,
      "step": 38203
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.7375388741493225,
      "learning_rate": 0.0005602085858419403,
      "loss": 2.3996,
      "step": 38204
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.7488123178482056,
      "learning_rate": 0.0005602065500355627,
      "loss": 2.5357,
      "step": 38205
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.8619928956031799,
      "learning_rate": 0.0005602045141808078,
      "loss": 2.2772,
      "step": 38206
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6354703307151794,
      "learning_rate": 0.0005602024782776759,
      "loss": 2.1098,
      "step": 38207
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.7418613433837891,
      "learning_rate": 0.0005602004423261675,
      "loss": 2.3436,
      "step": 38208
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6661838889122009,
      "learning_rate": 0.0005601984063262831,
      "loss": 2.301,
      "step": 38209
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6756309866905212,
      "learning_rate": 0.0005601963702780227,
      "loss": 2.0661,
      "step": 38210
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6922445297241211,
      "learning_rate": 0.000560194334181387,
      "loss": 2.1533,
      "step": 38211
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6923369765281677,
      "learning_rate": 0.0005601922980363761,
      "loss": 2.581,
      "step": 38212
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6006876826286316,
      "learning_rate": 0.0005601902618429908,
      "loss": 2.2291,
      "step": 38213
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.7303407788276672,
      "learning_rate": 0.000560188225601231,
      "loss": 2.2135,
      "step": 38214
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6509413123130798,
      "learning_rate": 0.0005601861893110973,
      "loss": 2.2272,
      "step": 38215
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6827479600906372,
      "learning_rate": 0.0005601841529725902,
      "loss": 2.3401,
      "step": 38216
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.8234387636184692,
      "learning_rate": 0.0005601821165857099,
      "loss": 2.2741,
      "step": 38217
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.73253333568573,
      "learning_rate": 0.0005601800801504569,
      "loss": 2.2173,
      "step": 38218
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.7029889822006226,
      "learning_rate": 0.0005601780436668314,
      "loss": 2.4332,
      "step": 38219
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6836662292480469,
      "learning_rate": 0.0005601760071348339,
      "loss": 2.2457,
      "step": 38220
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.7134904861450195,
      "learning_rate": 0.0005601739705544648,
      "loss": 2.2542,
      "step": 38221
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.7554231882095337,
      "learning_rate": 0.0005601719339257246,
      "loss": 2.3833,
      "step": 38222
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6843733787536621,
      "learning_rate": 0.0005601698972486133,
      "loss": 2.2667,
      "step": 38223
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6826975345611572,
      "learning_rate": 0.0005601678605231316,
      "loss": 2.1966,
      "step": 38224
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.7564788460731506,
      "learning_rate": 0.0005601658237492797,
      "loss": 2.0321,
      "step": 38225
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6202488541603088,
      "learning_rate": 0.0005601637869270582,
      "loss": 2.1885,
      "step": 38226
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6780035495758057,
      "learning_rate": 0.0005601617500564672,
      "loss": 2.1912,
      "step": 38227
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6877077221870422,
      "learning_rate": 0.0005601597131375072,
      "loss": 2.389,
      "step": 38228
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6582145094871521,
      "learning_rate": 0.0005601576761701788,
      "loss": 2.2634,
      "step": 38229
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6702370643615723,
      "learning_rate": 0.0005601556391544821,
      "loss": 2.1949,
      "step": 38230
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6632032990455627,
      "learning_rate": 0.0005601536020904174,
      "loss": 2.2373,
      "step": 38231
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.7273357510566711,
      "learning_rate": 0.0005601515649779855,
      "loss": 2.098,
      "step": 38232
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6463205814361572,
      "learning_rate": 0.0005601495278171862,
      "loss": 2.1704,
      "step": 38233
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6751410961151123,
      "learning_rate": 0.0005601474906080205,
      "loss": 2.3133,
      "step": 38234
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6969277262687683,
      "learning_rate": 0.0005601454533504883,
      "loss": 2.3807,
      "step": 38235
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6582635045051575,
      "learning_rate": 0.0005601434160445902,
      "loss": 2.1709,
      "step": 38236
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6903122067451477,
      "learning_rate": 0.0005601413786903265,
      "loss": 2.3857,
      "step": 38237
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6633694767951965,
      "learning_rate": 0.0005601393412876976,
      "loss": 2.2716,
      "step": 38238
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6669872999191284,
      "learning_rate": 0.0005601373038367039,
      "loss": 2.0704,
      "step": 38239
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.686991274356842,
      "learning_rate": 0.0005601352663373457,
      "loss": 2.3796,
      "step": 38240
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6915185451507568,
      "learning_rate": 0.0005601332287896235,
      "loss": 2.4387,
      "step": 38241
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.7707267999649048,
      "learning_rate": 0.0005601311911935377,
      "loss": 2.2242,
      "step": 38242
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6209866404533386,
      "learning_rate": 0.0005601291535490885,
      "loss": 2.1292,
      "step": 38243
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6802529692649841,
      "learning_rate": 0.0005601271158562764,
      "loss": 2.3739,
      "step": 38244
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6408732533454895,
      "learning_rate": 0.0005601250781151018,
      "loss": 2.2594,
      "step": 38245
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6387840509414673,
      "learning_rate": 0.000560123040325565,
      "loss": 2.0643,
      "step": 38246
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.7014530897140503,
      "learning_rate": 0.0005601210024876664,
      "loss": 2.192,
      "step": 38247
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.7622758150100708,
      "learning_rate": 0.0005601189646014065,
      "loss": 2.3933,
      "step": 38248
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6822715401649475,
      "learning_rate": 0.0005601169266667854,
      "loss": 2.1261,
      "step": 38249
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.7227959036827087,
      "learning_rate": 0.0005601148886838037,
      "loss": 2.1349,
      "step": 38250
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.7087448239326477,
      "learning_rate": 0.0005601128506524619,
      "loss": 2.4354,
      "step": 38251
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6606361865997314,
      "learning_rate": 0.0005601108125727601,
      "loss": 2.1366,
      "step": 38252
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.7148611545562744,
      "learning_rate": 0.0005601087744446988,
      "loss": 2.1683,
      "step": 38253
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.7232391238212585,
      "learning_rate": 0.0005601067362682783,
      "loss": 2.3078,
      "step": 38254
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6450928449630737,
      "learning_rate": 0.0005601046980434992,
      "loss": 2.2318,
      "step": 38255
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.7601343989372253,
      "learning_rate": 0.0005601026597703615,
      "loss": 2.2601,
      "step": 38256
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.7496931552886963,
      "learning_rate": 0.000560100621448866,
      "loss": 2.3756,
      "step": 38257
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6869646310806274,
      "learning_rate": 0.0005600985830790128,
      "loss": 2.0927,
      "step": 38258
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.7159438133239746,
      "learning_rate": 0.0005600965446608024,
      "loss": 2.3862,
      "step": 38259
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6533892750740051,
      "learning_rate": 0.0005600945061942351,
      "loss": 2.4946,
      "step": 38260
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6315227150917053,
      "learning_rate": 0.0005600924676793114,
      "loss": 2.1224,
      "step": 38261
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6323261260986328,
      "learning_rate": 0.0005600904291160315,
      "loss": 2.3844,
      "step": 38262
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.661626398563385,
      "learning_rate": 0.0005600883905043959,
      "loss": 2.0927,
      "step": 38263
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6822765469551086,
      "learning_rate": 0.0005600863518444051,
      "loss": 2.1735,
      "step": 38264
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6773364543914795,
      "learning_rate": 0.0005600843131360592,
      "loss": 2.3028,
      "step": 38265
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.7095553874969482,
      "learning_rate": 0.0005600822743793587,
      "loss": 2.2831,
      "step": 38266
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6412763595581055,
      "learning_rate": 0.000560080235574304,
      "loss": 2.0893,
      "step": 38267
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6466487646102905,
      "learning_rate": 0.0005600781967208955,
      "loss": 2.4599,
      "step": 38268
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.677972137928009,
      "learning_rate": 0.0005600761578191336,
      "loss": 2.0702,
      "step": 38269
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.7078608274459839,
      "learning_rate": 0.0005600741188690186,
      "loss": 2.2566,
      "step": 38270
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.7202282547950745,
      "learning_rate": 0.0005600720798705508,
      "loss": 2.2392,
      "step": 38271
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.7211991548538208,
      "learning_rate": 0.0005600700408237309,
      "loss": 2.3121,
      "step": 38272
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6867405772209167,
      "learning_rate": 0.000560068001728559,
      "loss": 2.2359,
      "step": 38273
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6753488183021545,
      "learning_rate": 0.0005600659625850356,
      "loss": 2.1779,
      "step": 38274
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.622356116771698,
      "learning_rate": 0.0005600639233931608,
      "loss": 2.1467,
      "step": 38275
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.7346248030662537,
      "learning_rate": 0.0005600618841529354,
      "loss": 2.4139,
      "step": 38276
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6917015910148621,
      "learning_rate": 0.0005600598448643596,
      "loss": 2.5199,
      "step": 38277
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.7141131162643433,
      "learning_rate": 0.0005600578055274336,
      "loss": 2.1546,
      "step": 38278
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.7875279784202576,
      "learning_rate": 0.0005600557661421581,
      "loss": 2.4879,
      "step": 38279
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.7511889934539795,
      "learning_rate": 0.0005600537267085332,
      "loss": 2.4144,
      "step": 38280
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6692548990249634,
      "learning_rate": 0.0005600516872265595,
      "loss": 2.0765,
      "step": 38281
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6056667566299438,
      "learning_rate": 0.0005600496476962373,
      "loss": 2.1355,
      "step": 38282
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.8736351132392883,
      "learning_rate": 0.000560047608117567,
      "loss": 2.4627,
      "step": 38283
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6486617922782898,
      "learning_rate": 0.0005600455684905488,
      "loss": 2.3348,
      "step": 38284
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6649940609931946,
      "learning_rate": 0.0005600435288151832,
      "loss": 2.1988,
      "step": 38285
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.669894278049469,
      "learning_rate": 0.0005600414890914707,
      "loss": 2.2658,
      "step": 38286
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.8241235017776489,
      "learning_rate": 0.0005600394493194117,
      "loss": 2.316,
      "step": 38287
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6542044281959534,
      "learning_rate": 0.0005600374094990062,
      "loss": 2.165,
      "step": 38288
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.690847635269165,
      "learning_rate": 0.000560035369630255,
      "loss": 2.2514,
      "step": 38289
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6521286368370056,
      "learning_rate": 0.0005600333297131583,
      "loss": 2.1435,
      "step": 38290
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.7495136857032776,
      "learning_rate": 0.0005600312897477165,
      "loss": 2.3413,
      "step": 38291
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.7255734801292419,
      "learning_rate": 0.0005600292497339299,
      "loss": 2.2557,
      "step": 38292
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6518486142158508,
      "learning_rate": 0.000560027209671799,
      "loss": 2.1962,
      "step": 38293
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6926791071891785,
      "learning_rate": 0.0005600251695613242,
      "loss": 2.3639,
      "step": 38294
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6484814286231995,
      "learning_rate": 0.0005600231294025056,
      "loss": 2.3262,
      "step": 38295
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6988217234611511,
      "learning_rate": 0.000560021089195344,
      "loss": 2.4858,
      "step": 38296
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.756618082523346,
      "learning_rate": 0.0005600190489398395,
      "loss": 2.2769,
      "step": 38297
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.731367826461792,
      "learning_rate": 0.0005600170086359925,
      "loss": 2.4543,
      "step": 38298
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6580649018287659,
      "learning_rate": 0.0005600149682838036,
      "loss": 2.176,
      "step": 38299
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.664588212966919,
      "learning_rate": 0.0005600129278832729,
      "loss": 2.0756,
      "step": 38300
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.664766788482666,
      "learning_rate": 0.0005600108874344008,
      "loss": 2.1831,
      "step": 38301
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6709396839141846,
      "learning_rate": 0.0005600088469371879,
      "loss": 2.1191,
      "step": 38302
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.7073716521263123,
      "learning_rate": 0.0005600068063916344,
      "loss": 2.165,
      "step": 38303
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6652688980102539,
      "learning_rate": 0.0005600047657977407,
      "loss": 2.4557,
      "step": 38304
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6476961970329285,
      "learning_rate": 0.0005600027251555072,
      "loss": 2.3362,
      "step": 38305
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6268797516822815,
      "learning_rate": 0.0005600006844649344,
      "loss": 2.0454,
      "step": 38306
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.7107133865356445,
      "learning_rate": 0.0005599986437260224,
      "loss": 2.2122,
      "step": 38307
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6512203216552734,
      "learning_rate": 0.0005599966029387719,
      "loss": 2.3914,
      "step": 38308
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6418280601501465,
      "learning_rate": 0.0005599945621031831,
      "loss": 2.0992,
      "step": 38309
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6598442792892456,
      "learning_rate": 0.0005599925212192564,
      "loss": 2.0723,
      "step": 38310
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.811742901802063,
      "learning_rate": 0.000559990480286992,
      "loss": 2.2013,
      "step": 38311
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6647353768348694,
      "learning_rate": 0.0005599884393063906,
      "loss": 2.3336,
      "step": 38312
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6984134316444397,
      "learning_rate": 0.0005599863982774525,
      "loss": 2.3269,
      "step": 38313
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6983197331428528,
      "learning_rate": 0.000559984357200178,
      "loss": 2.3989,
      "step": 38314
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.7095704078674316,
      "learning_rate": 0.0005599823160745674,
      "loss": 2.549,
      "step": 38315
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.7045446038246155,
      "learning_rate": 0.0005599802749006212,
      "loss": 2.402,
      "step": 38316
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.8556553721427917,
      "learning_rate": 0.0005599782336783398,
      "loss": 2.3483,
      "step": 38317
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.7414289116859436,
      "learning_rate": 0.0005599761924077235,
      "loss": 2.2055,
      "step": 38318
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.7896122932434082,
      "learning_rate": 0.0005599741510887729,
      "loss": 2.331,
      "step": 38319
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6529331207275391,
      "learning_rate": 0.000559972109721488,
      "loss": 2.1329,
      "step": 38320
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.7510815858840942,
      "learning_rate": 0.0005599700683058693,
      "loss": 2.221,
      "step": 38321
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6464280486106873,
      "learning_rate": 0.0005599680268419175,
      "loss": 2.3738,
      "step": 38322
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.7511785626411438,
      "learning_rate": 0.0005599659853296327,
      "loss": 2.182,
      "step": 38323
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6842761039733887,
      "learning_rate": 0.0005599639437690152,
      "loss": 2.3984,
      "step": 38324
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.7119500041007996,
      "learning_rate": 0.0005599619021600656,
      "loss": 2.2293,
      "step": 38325
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.7378665208816528,
      "learning_rate": 0.000559959860502784,
      "loss": 2.3283,
      "step": 38326
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.7003901600837708,
      "learning_rate": 0.0005599578187971712,
      "loss": 2.3372,
      "step": 38327
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.7229912877082825,
      "learning_rate": 0.0005599557770432272,
      "loss": 2.1486,
      "step": 38328
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.7046256065368652,
      "learning_rate": 0.0005599537352409524,
      "loss": 2.3944,
      "step": 38329
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6558573246002197,
      "learning_rate": 0.0005599516933903475,
      "loss": 2.1889,
      "step": 38330
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.7017770409584045,
      "learning_rate": 0.0005599496514914126,
      "loss": 2.3196,
      "step": 38331
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6843120455741882,
      "learning_rate": 0.0005599476095441481,
      "loss": 2.4541,
      "step": 38332
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6521773338317871,
      "learning_rate": 0.0005599455675485544,
      "loss": 2.1556,
      "step": 38333
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6867221593856812,
      "learning_rate": 0.000559943525504632,
      "loss": 2.3282,
      "step": 38334
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.7535300254821777,
      "learning_rate": 0.0005599414834123812,
      "loss": 2.5457,
      "step": 38335
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.7374148368835449,
      "learning_rate": 0.0005599394412718024,
      "loss": 2.2405,
      "step": 38336
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.7321304082870483,
      "learning_rate": 0.0005599373990828959,
      "loss": 2.2958,
      "step": 38337
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.7659525275230408,
      "learning_rate": 0.0005599353568456621,
      "loss": 2.4051,
      "step": 38338
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6873064041137695,
      "learning_rate": 0.0005599333145601015,
      "loss": 2.3041,
      "step": 38339
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6794083714485168,
      "learning_rate": 0.0005599312722262143,
      "loss": 2.1748,
      "step": 38340
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.9571002125740051,
      "learning_rate": 0.000559929229844001,
      "loss": 2.2643,
      "step": 38341
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6811787486076355,
      "learning_rate": 0.0005599271874134619,
      "loss": 2.3582,
      "step": 38342
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.7055040597915649,
      "learning_rate": 0.0005599251449345975,
      "loss": 2.2394,
      "step": 38343
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.7251301407814026,
      "learning_rate": 0.0005599231024074081,
      "loss": 2.4194,
      "step": 38344
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.8460227847099304,
      "learning_rate": 0.000559921059831894,
      "loss": 2.3108,
      "step": 38345
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6731708645820618,
      "learning_rate": 0.0005599190172080557,
      "loss": 2.307,
      "step": 38346
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.813349187374115,
      "learning_rate": 0.0005599169745358936,
      "loss": 2.3871,
      "step": 38347
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.7538438439369202,
      "learning_rate": 0.000559914931815408,
      "loss": 2.2333,
      "step": 38348
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6629835963249207,
      "learning_rate": 0.0005599128890465994,
      "loss": 2.1965,
      "step": 38349
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6661075949668884,
      "learning_rate": 0.000559910846229468,
      "loss": 2.2044,
      "step": 38350
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6902899742126465,
      "learning_rate": 0.0005599088033640142,
      "loss": 2.4647,
      "step": 38351
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.7579968571662903,
      "learning_rate": 0.0005599067604502387,
      "loss": 2.4282,
      "step": 38352
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6615963578224182,
      "learning_rate": 0.0005599047174881414,
      "loss": 2.2882,
      "step": 38353
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6717126965522766,
      "learning_rate": 0.000559902674477723,
      "loss": 2.511,
      "step": 38354
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.8272601962089539,
      "learning_rate": 0.0005599006314189837,
      "loss": 2.5701,
      "step": 38355
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6973999738693237,
      "learning_rate": 0.0005598985883119241,
      "loss": 2.5321,
      "step": 38356
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.7051076889038086,
      "learning_rate": 0.0005598965451565443,
      "loss": 2.2728,
      "step": 38357
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.7702716588973999,
      "learning_rate": 0.0005598945019528449,
      "loss": 2.0545,
      "step": 38358
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6821914315223694,
      "learning_rate": 0.0005598924587008262,
      "loss": 2.3094,
      "step": 38359
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6995260715484619,
      "learning_rate": 0.0005598904154004886,
      "loss": 1.9152,
      "step": 38360
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.7399185299873352,
      "learning_rate": 0.0005598883720518324,
      "loss": 2.3912,
      "step": 38361
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6830017566680908,
      "learning_rate": 0.0005598863286548581,
      "loss": 2.4705,
      "step": 38362
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.765065610408783,
      "learning_rate": 0.000559884285209566,
      "loss": 2.2218,
      "step": 38363
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.7626076340675354,
      "learning_rate": 0.0005598822417159565,
      "loss": 2.4117,
      "step": 38364
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.7591395378112793,
      "learning_rate": 0.00055988019817403,
      "loss": 2.4238,
      "step": 38365
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.7390096783638,
      "learning_rate": 0.0005598781545837869,
      "loss": 2.2454,
      "step": 38366
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.7687476277351379,
      "learning_rate": 0.0005598761109452276,
      "loss": 2.3751,
      "step": 38367
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.7358322143554688,
      "learning_rate": 0.0005598740672583522,
      "loss": 2.4605,
      "step": 38368
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6407173871994019,
      "learning_rate": 0.0005598720235231616,
      "loss": 2.445,
      "step": 38369
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.740475594997406,
      "learning_rate": 0.0005598699797396557,
      "loss": 2.1986,
      "step": 38370
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6356092691421509,
      "learning_rate": 0.0005598679359078351,
      "loss": 2.2023,
      "step": 38371
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6948220133781433,
      "learning_rate": 0.0005598658920277002,
      "loss": 2.4012,
      "step": 38372
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.7351694703102112,
      "learning_rate": 0.0005598638480992513,
      "loss": 2.3101,
      "step": 38373
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.789947509765625,
      "learning_rate": 0.0005598618041224887,
      "loss": 2.3193,
      "step": 38374
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.7377157211303711,
      "learning_rate": 0.000559859760097413,
      "loss": 2.4008,
      "step": 38375
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6879349946975708,
      "learning_rate": 0.0005598577160240244,
      "loss": 2.3282,
      "step": 38376
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.7460539937019348,
      "learning_rate": 0.0005598556719023234,
      "loss": 2.4394,
      "step": 38377
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.8460278511047363,
      "learning_rate": 0.0005598536277323104,
      "loss": 2.1139,
      "step": 38378
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6463940739631653,
      "learning_rate": 0.0005598515835139855,
      "loss": 2.15,
      "step": 38379
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.8104152679443359,
      "learning_rate": 0.0005598495392473494,
      "loss": 2.3056,
      "step": 38380
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6861836314201355,
      "learning_rate": 0.0005598474949324024,
      "loss": 2.1958,
      "step": 38381
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6050475835800171,
      "learning_rate": 0.0005598454505691448,
      "loss": 2.376,
      "step": 38382
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6252676844596863,
      "learning_rate": 0.000559843406157577,
      "loss": 2.1488,
      "step": 38383
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6788150072097778,
      "learning_rate": 0.0005598413616976994,
      "loss": 2.5808,
      "step": 38384
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6380694508552551,
      "learning_rate": 0.0005598393171895125,
      "loss": 2.2238,
      "step": 38385
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.7094825506210327,
      "learning_rate": 0.0005598372726330164,
      "loss": 2.2736,
      "step": 38386
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.7648456692695618,
      "learning_rate": 0.0005598352280282118,
      "loss": 2.34,
      "step": 38387
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6479441523551941,
      "learning_rate": 0.0005598331833750989,
      "loss": 2.6061,
      "step": 38388
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6275137066841125,
      "learning_rate": 0.000559831138673678,
      "loss": 2.3824,
      "step": 38389
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.703018844127655,
      "learning_rate": 0.0005598290939239496,
      "loss": 2.2961,
      "step": 38390
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6393917798995972,
      "learning_rate": 0.0005598270491259142,
      "loss": 2.2075,
      "step": 38391
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6155433654785156,
      "learning_rate": 0.0005598250042795719,
      "loss": 2.1901,
      "step": 38392
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.676679790019989,
      "learning_rate": 0.0005598229593849234,
      "loss": 2.2981,
      "step": 38393
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6873540282249451,
      "learning_rate": 0.0005598209144419687,
      "loss": 2.2353,
      "step": 38394
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6803606152534485,
      "learning_rate": 0.0005598188694507085,
      "loss": 2.2935,
      "step": 38395
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.7176887392997742,
      "learning_rate": 0.0005598168244111431,
      "loss": 2.1741,
      "step": 38396
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6306059956550598,
      "learning_rate": 0.0005598147793232727,
      "loss": 2.2644,
      "step": 38397
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6612670421600342,
      "learning_rate": 0.0005598127341870981,
      "loss": 2.2118,
      "step": 38398
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.7090371251106262,
      "learning_rate": 0.0005598106890026192,
      "loss": 2.1457,
      "step": 38399
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.785518229007721,
      "learning_rate": 0.0005598086437698366,
      "loss": 2.1533,
      "step": 38400
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.7968292832374573,
      "learning_rate": 0.0005598065984887507,
      "loss": 2.4207,
      "step": 38401
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.709646999835968,
      "learning_rate": 0.0005598045531593619,
      "loss": 2.2353,
      "step": 38402
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6795495748519897,
      "learning_rate": 0.0005598025077816704,
      "loss": 2.5385,
      "step": 38403
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.7480353116989136,
      "learning_rate": 0.0005598004623556768,
      "loss": 2.3444,
      "step": 38404
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.7026129364967346,
      "learning_rate": 0.0005597984168813814,
      "loss": 2.2534,
      "step": 38405
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6460071802139282,
      "learning_rate": 0.0005597963713587845,
      "loss": 2.4133,
      "step": 38406
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.7085450291633606,
      "learning_rate": 0.0005597943257878866,
      "loss": 2.551,
      "step": 38407
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.7052595615386963,
      "learning_rate": 0.0005597922801686881,
      "loss": 2.3223,
      "step": 38408
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6889573335647583,
      "learning_rate": 0.0005597902345011892,
      "loss": 2.2118,
      "step": 38409
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6589077115058899,
      "learning_rate": 0.0005597881887853904,
      "loss": 2.4845,
      "step": 38410
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.944371223449707,
      "learning_rate": 0.0005597861430212921,
      "loss": 2.362,
      "step": 38411
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6324644684791565,
      "learning_rate": 0.0005597840972088947,
      "loss": 2.2681,
      "step": 38412
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6413339376449585,
      "learning_rate": 0.0005597820513481985,
      "loss": 2.3138,
      "step": 38413
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6729437112808228,
      "learning_rate": 0.0005597800054392038,
      "loss": 2.3565,
      "step": 38414
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.7965511083602905,
      "learning_rate": 0.0005597779594819113,
      "loss": 2.5427,
      "step": 38415
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6409533023834229,
      "learning_rate": 0.000559775913476321,
      "loss": 2.343,
      "step": 38416
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.8123684525489807,
      "learning_rate": 0.0005597738674224335,
      "loss": 2.2588,
      "step": 38417
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6219335198402405,
      "learning_rate": 0.0005597718213202493,
      "loss": 2.3791,
      "step": 38418
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.7329214811325073,
      "learning_rate": 0.0005597697751697684,
      "loss": 2.2849,
      "step": 38419
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6206537485122681,
      "learning_rate": 0.0005597677289709915,
      "loss": 2.2143,
      "step": 38420
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.658667802810669,
      "learning_rate": 0.0005597656827239188,
      "loss": 2.3256,
      "step": 38421
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.7089849710464478,
      "learning_rate": 0.0005597636364285508,
      "loss": 2.4461,
      "step": 38422
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6967931389808655,
      "learning_rate": 0.0005597615900848878,
      "loss": 2.351,
      "step": 38423
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.7196982502937317,
      "learning_rate": 0.0005597595436929303,
      "loss": 2.3876,
      "step": 38424
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.7127160429954529,
      "learning_rate": 0.0005597574972526785,
      "loss": 2.0545,
      "step": 38425
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6206629872322083,
      "learning_rate": 0.0005597554507641329,
      "loss": 2.1952,
      "step": 38426
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6818075180053711,
      "learning_rate": 0.000559753404227294,
      "loss": 2.1635,
      "step": 38427
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6771770715713501,
      "learning_rate": 0.0005597513576421619,
      "loss": 2.1474,
      "step": 38428
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6779127717018127,
      "learning_rate": 0.0005597493110087371,
      "loss": 2.2102,
      "step": 38429
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.7469311356544495,
      "learning_rate": 0.00055974726432702,
      "loss": 2.3069,
      "step": 38430
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6627864837646484,
      "learning_rate": 0.0005597452175970111,
      "loss": 2.1426,
      "step": 38431
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.7841155529022217,
      "learning_rate": 0.0005597431708187105,
      "loss": 2.2776,
      "step": 38432
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.7531728148460388,
      "learning_rate": 0.0005597411239921189,
      "loss": 2.2122,
      "step": 38433
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6750405430793762,
      "learning_rate": 0.0005597390771172365,
      "loss": 2.4592,
      "step": 38434
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.7192479968070984,
      "learning_rate": 0.0005597370301940637,
      "loss": 2.3446,
      "step": 38435
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.701991617679596,
      "learning_rate": 0.0005597349832226008,
      "loss": 2.2545,
      "step": 38436
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6820940971374512,
      "learning_rate": 0.0005597329362028483,
      "loss": 2.2568,
      "step": 38437
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.7040464282035828,
      "learning_rate": 0.0005597308891348065,
      "loss": 2.3307,
      "step": 38438
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.7559356689453125,
      "learning_rate": 0.0005597288420184759,
      "loss": 2.375,
      "step": 38439
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6392416954040527,
      "learning_rate": 0.0005597267948538568,
      "loss": 2.0704,
      "step": 38440
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6193296909332275,
      "learning_rate": 0.0005597247476409494,
      "loss": 1.9488,
      "step": 38441
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6919043064117432,
      "learning_rate": 0.0005597227003797544,
      "loss": 2.3569,
      "step": 38442
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6626009345054626,
      "learning_rate": 0.0005597206530702722,
      "loss": 2.1602,
      "step": 38443
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.638225793838501,
      "learning_rate": 0.0005597186057125029,
      "loss": 2.3413,
      "step": 38444
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.7670114636421204,
      "learning_rate": 0.000559716558306447,
      "loss": 2.4839,
      "step": 38445
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6756072044372559,
      "learning_rate": 0.0005597145108521049,
      "loss": 2.4667,
      "step": 38446
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.7274692058563232,
      "learning_rate": 0.000559712463349477,
      "loss": 2.4307,
      "step": 38447
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.7311459183692932,
      "learning_rate": 0.0005597104157985636,
      "loss": 2.4831,
      "step": 38448
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.749439001083374,
      "learning_rate": 0.0005597083681993652,
      "loss": 2.3668,
      "step": 38449
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.678768515586853,
      "learning_rate": 0.000559706320551882,
      "loss": 2.1268,
      "step": 38450
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6776394844055176,
      "learning_rate": 0.0005597042728561145,
      "loss": 2.3746,
      "step": 38451
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6542670130729675,
      "learning_rate": 0.0005597022251120631,
      "loss": 2.2158,
      "step": 38452
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6453137993812561,
      "learning_rate": 0.0005597001773197282,
      "loss": 2.4204,
      "step": 38453
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.690144419670105,
      "learning_rate": 0.0005596981294791102,
      "loss": 2.2036,
      "step": 38454
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.7222486138343811,
      "learning_rate": 0.0005596960815902093,
      "loss": 2.3596,
      "step": 38455
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6729128956794739,
      "learning_rate": 0.000559694033653026,
      "loss": 2.1891,
      "step": 38456
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6269869208335876,
      "learning_rate": 0.0005596919856675608,
      "loss": 2.0515,
      "step": 38457
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6446307301521301,
      "learning_rate": 0.0005596899376338138,
      "loss": 2.1803,
      "step": 38458
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6545964479446411,
      "learning_rate": 0.0005596878895517856,
      "loss": 2.4232,
      "step": 38459
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6994324922561646,
      "learning_rate": 0.0005596858414214765,
      "loss": 2.3553,
      "step": 38460
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6440518498420715,
      "learning_rate": 0.000559683793242887,
      "loss": 2.2532,
      "step": 38461
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.7043164372444153,
      "learning_rate": 0.0005596817450160173,
      "loss": 2.4129,
      "step": 38462
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.8259658217430115,
      "learning_rate": 0.0005596796967408679,
      "loss": 2.1281,
      "step": 38463
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6589394211769104,
      "learning_rate": 0.0005596776484174391,
      "loss": 2.2075,
      "step": 38464
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.7022648453712463,
      "learning_rate": 0.0005596756000457312,
      "loss": 2.175,
      "step": 38465
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6893284916877747,
      "learning_rate": 0.0005596735516257449,
      "loss": 2.2248,
      "step": 38466
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.689850389957428,
      "learning_rate": 0.0005596715031574803,
      "loss": 2.235,
      "step": 38467
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6147331595420837,
      "learning_rate": 0.0005596694546409379,
      "loss": 2.1186,
      "step": 38468
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.7145128846168518,
      "learning_rate": 0.0005596674060761181,
      "loss": 1.9942,
      "step": 38469
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6545491218566895,
      "learning_rate": 0.0005596653574630211,
      "loss": 2.0847,
      "step": 38470
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6367554664611816,
      "learning_rate": 0.0005596633088016475,
      "loss": 2.0716,
      "step": 38471
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6662807464599609,
      "learning_rate": 0.0005596612600919976,
      "loss": 2.5488,
      "step": 38472
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6573594808578491,
      "learning_rate": 0.0005596592113340717,
      "loss": 2.1053,
      "step": 38473
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6870547533035278,
      "learning_rate": 0.0005596571625278703,
      "loss": 2.3413,
      "step": 38474
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6512036323547363,
      "learning_rate": 0.0005596551136733937,
      "loss": 2.1296,
      "step": 38475
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.7139388918876648,
      "learning_rate": 0.0005596530647706423,
      "loss": 2.3165,
      "step": 38476
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6661390066146851,
      "learning_rate": 0.0005596510158196164,
      "loss": 2.3906,
      "step": 38477
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6906548738479614,
      "learning_rate": 0.0005596489668203167,
      "loss": 2.4439,
      "step": 38478
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6914466619491577,
      "learning_rate": 0.0005596469177727432,
      "loss": 2.3747,
      "step": 38479
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6714949011802673,
      "learning_rate": 0.0005596448686768966,
      "loss": 2.3156,
      "step": 38480
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.8084232211112976,
      "learning_rate": 0.000559642819532777,
      "loss": 2.2619,
      "step": 38481
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6023194193840027,
      "learning_rate": 0.0005596407703403848,
      "loss": 2.0909,
      "step": 38482
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.7763378620147705,
      "learning_rate": 0.0005596387210997206,
      "loss": 2.3166,
      "step": 38483
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6724815964698792,
      "learning_rate": 0.0005596366718107846,
      "loss": 2.2733,
      "step": 38484
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.912976086139679,
      "learning_rate": 0.0005596346224735773,
      "loss": 2.2551,
      "step": 38485
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6741409301757812,
      "learning_rate": 0.000559632573088099,
      "loss": 2.2283,
      "step": 38486
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.0786200761795044,
      "learning_rate": 0.0005596305236543501,
      "loss": 2.2052,
      "step": 38487
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.714326024055481,
      "learning_rate": 0.000559628474172331,
      "loss": 2.3513,
      "step": 38488
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6512306332588196,
      "learning_rate": 0.0005596264246420421,
      "loss": 1.9208,
      "step": 38489
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.8952144384384155,
      "learning_rate": 0.0005596243750634837,
      "loss": 2.2371,
      "step": 38490
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6714885234832764,
      "learning_rate": 0.0005596223254366563,
      "loss": 2.2927,
      "step": 38491
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6974746584892273,
      "learning_rate": 0.0005596202757615601,
      "loss": 2.3314,
      "step": 38492
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.7202181816101074,
      "learning_rate": 0.0005596182260381957,
      "loss": 2.2349,
      "step": 38493
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.8082852959632874,
      "learning_rate": 0.0005596161762665633,
      "loss": 2.3521,
      "step": 38494
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6571692824363708,
      "learning_rate": 0.0005596141264466634,
      "loss": 2.1088,
      "step": 38495
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.702710747718811,
      "learning_rate": 0.0005596120765784962,
      "loss": 2.421,
      "step": 38496
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.7916853427886963,
      "learning_rate": 0.0005596100266620624,
      "loss": 2.4136,
      "step": 38497
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.756915271282196,
      "learning_rate": 0.000559607976697362,
      "loss": 2.219,
      "step": 38498
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6666256189346313,
      "learning_rate": 0.0005596059266843957,
      "loss": 2.1471,
      "step": 38499
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.762451708316803,
      "learning_rate": 0.0005596038766231637,
      "loss": 2.1995,
      "step": 38500
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.8034055829048157,
      "learning_rate": 0.0005596018265136665,
      "loss": 2.0924,
      "step": 38501
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.7837749123573303,
      "learning_rate": 0.0005595997763559044,
      "loss": 2.5267,
      "step": 38502
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6725034117698669,
      "learning_rate": 0.0005595977261498779,
      "loss": 2.392,
      "step": 38503
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.9873330593109131,
      "learning_rate": 0.0005595956758955871,
      "loss": 2.4019,
      "step": 38504
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.702610969543457,
      "learning_rate": 0.0005595936255930327,
      "loss": 2.3422,
      "step": 38505
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6662399768829346,
      "learning_rate": 0.0005595915752422149,
      "loss": 2.048,
      "step": 38506
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.751240611076355,
      "learning_rate": 0.0005595895248431341,
      "loss": 2.3701,
      "step": 38507
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.7051445245742798,
      "learning_rate": 0.0005595874743957907,
      "loss": 2.4222,
      "step": 38508
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.8214354515075684,
      "learning_rate": 0.0005595854239001851,
      "loss": 2.274,
      "step": 38509
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.8672863841056824,
      "learning_rate": 0.0005595833733563176,
      "loss": 2.4113,
      "step": 38510
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6987462639808655,
      "learning_rate": 0.0005595813227641888,
      "loss": 2.486,
      "step": 38511
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.7523326873779297,
      "learning_rate": 0.0005595792721237988,
      "loss": 2.2502,
      "step": 38512
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.7586158514022827,
      "learning_rate": 0.0005595772214351482,
      "loss": 2.2131,
      "step": 38513
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6533793210983276,
      "learning_rate": 0.0005595751706982372,
      "loss": 1.9776,
      "step": 38514
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.7451609373092651,
      "learning_rate": 0.0005595731199130664,
      "loss": 2.5395,
      "step": 38515
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6544786095619202,
      "learning_rate": 0.000559571069079636,
      "loss": 2.3726,
      "step": 38516
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.7103780508041382,
      "learning_rate": 0.0005595690181979464,
      "loss": 2.386,
      "step": 38517
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6750357151031494,
      "learning_rate": 0.0005595669672679982,
      "loss": 2.2383,
      "step": 38518
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6438168883323669,
      "learning_rate": 0.0005595649162897915,
      "loss": 2.2284,
      "step": 38519
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6457141041755676,
      "learning_rate": 0.0005595628652633266,
      "loss": 2.1671,
      "step": 38520
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.7435819506645203,
      "learning_rate": 0.0005595608141886043,
      "loss": 1.9527,
      "step": 38521
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6920743584632874,
      "learning_rate": 0.0005595587630656245,
      "loss": 2.3035,
      "step": 38522
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.7064840793609619,
      "learning_rate": 0.0005595567118943881,
      "loss": 2.3746,
      "step": 38523
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6841019988059998,
      "learning_rate": 0.000559554660674895,
      "loss": 2.4049,
      "step": 38524
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6842148303985596,
      "learning_rate": 0.0005595526094071459,
      "loss": 2.3464,
      "step": 38525
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6986136436462402,
      "learning_rate": 0.0005595505580911411,
      "loss": 2.1875,
      "step": 38526
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6939737796783447,
      "learning_rate": 0.0005595485067268809,
      "loss": 2.357,
      "step": 38527
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.7556809186935425,
      "learning_rate": 0.0005595464553143657,
      "loss": 2.4513,
      "step": 38528
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6780194640159607,
      "learning_rate": 0.0005595444038535959,
      "loss": 2.3168,
      "step": 38529
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6865414381027222,
      "learning_rate": 0.000559542352344572,
      "loss": 2.1885,
      "step": 38530
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6564705967903137,
      "learning_rate": 0.0005595403007872942,
      "loss": 2.2084,
      "step": 38531
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.7168181538581848,
      "learning_rate": 0.0005595382491817631,
      "loss": 2.2218,
      "step": 38532
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.7005332112312317,
      "learning_rate": 0.0005595361975279788,
      "loss": 2.1821,
      "step": 38533
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6766069531440735,
      "learning_rate": 0.0005595341458259418,
      "loss": 2.3238,
      "step": 38534
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6606670618057251,
      "learning_rate": 0.0005595320940756526,
      "loss": 2.3684,
      "step": 38535
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.9625547528266907,
      "learning_rate": 0.0005595300422771113,
      "loss": 2.2354,
      "step": 38536
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6916911602020264,
      "learning_rate": 0.0005595279904303186,
      "loss": 2.2189,
      "step": 38537
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6672014594078064,
      "learning_rate": 0.0005595259385352747,
      "loss": 2.2817,
      "step": 38538
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.8008283972740173,
      "learning_rate": 0.0005595238865919801,
      "loss": 2.4227,
      "step": 38539
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.7948957085609436,
      "learning_rate": 0.0005595218346004352,
      "loss": 2.1964,
      "step": 38540
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.7119712233543396,
      "learning_rate": 0.0005595197825606402,
      "loss": 2.3696,
      "step": 38541
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.8908401131629944,
      "learning_rate": 0.0005595177304725954,
      "loss": 2.4399,
      "step": 38542
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.76338130235672,
      "learning_rate": 0.0005595156783363015,
      "loss": 2.1908,
      "step": 38543
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.693138599395752,
      "learning_rate": 0.0005595136261517588,
      "loss": 2.2087,
      "step": 38544
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.8671765923500061,
      "learning_rate": 0.0005595115739189675,
      "loss": 2.324,
      "step": 38545
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.0065202713012695,
      "learning_rate": 0.0005595095216379283,
      "loss": 2.3329,
      "step": 38546
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6936181783676147,
      "learning_rate": 0.0005595074693086412,
      "loss": 2.0578,
      "step": 38547
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.7029526233673096,
      "learning_rate": 0.0005595054169311068,
      "loss": 2.0957,
      "step": 38548
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.9200924038887024,
      "learning_rate": 0.0005595033645053254,
      "loss": 2.4171,
      "step": 38549
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.7958976626396179,
      "learning_rate": 0.0005595013120312975,
      "loss": 1.9856,
      "step": 38550
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.7210371494293213,
      "learning_rate": 0.0005594992595090233,
      "loss": 2.4112,
      "step": 38551
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.7599987387657166,
      "learning_rate": 0.0005594972069385034,
      "loss": 2.217,
      "step": 38552
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.708055317401886,
      "learning_rate": 0.0005594951543197381,
      "loss": 2.3768,
      "step": 38553
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6619176268577576,
      "learning_rate": 0.0005594931016527277,
      "loss": 2.0651,
      "step": 38554
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6182949542999268,
      "learning_rate": 0.0005594910489374725,
      "loss": 2.1666,
      "step": 38555
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.7141823172569275,
      "learning_rate": 0.0005594889961739732,
      "loss": 2.3692,
      "step": 38556
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6940404772758484,
      "learning_rate": 0.00055948694336223,
      "loss": 2.2619,
      "step": 38557
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.7368637919425964,
      "learning_rate": 0.0005594848905022432,
      "loss": 2.7018,
      "step": 38558
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.713690459728241,
      "learning_rate": 0.0005594828375940133,
      "loss": 2.2927,
      "step": 38559
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6834654808044434,
      "learning_rate": 0.0005594807846375406,
      "loss": 1.9824,
      "step": 38560
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6827869415283203,
      "learning_rate": 0.0005594787316328255,
      "loss": 2.1837,
      "step": 38561
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.758787989616394,
      "learning_rate": 0.0005594766785798685,
      "loss": 2.5363,
      "step": 38562
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.7418597340583801,
      "learning_rate": 0.0005594746254786699,
      "loss": 2.3503,
      "step": 38563
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.7396204471588135,
      "learning_rate": 0.00055947257232923,
      "loss": 2.1793,
      "step": 38564
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6959874033927917,
      "learning_rate": 0.0005594705191315491,
      "loss": 2.3663,
      "step": 38565
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6668288707733154,
      "learning_rate": 0.0005594684658856281,
      "loss": 2.3278,
      "step": 38566
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.686301589012146,
      "learning_rate": 0.0005594664125914667,
      "loss": 2.3595,
      "step": 38567
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6651320457458496,
      "learning_rate": 0.0005594643592490658,
      "loss": 2.0988,
      "step": 38568
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.7080366015434265,
      "learning_rate": 0.0005594623058584254,
      "loss": 2.3378,
      "step": 38569
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6955103874206543,
      "learning_rate": 0.0005594602524195461,
      "loss": 2.16,
      "step": 38570
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.7547996044158936,
      "learning_rate": 0.0005594581989324283,
      "loss": 2.3435,
      "step": 38571
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.772840678691864,
      "learning_rate": 0.0005594561453970723,
      "loss": 2.0799,
      "step": 38572
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.7594558596611023,
      "learning_rate": 0.0005594540918134785,
      "loss": 2.1704,
      "step": 38573
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.7627964019775391,
      "learning_rate": 0.0005594520381816473,
      "loss": 2.3624,
      "step": 38574
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6797218322753906,
      "learning_rate": 0.0005594499845015791,
      "loss": 2.1437,
      "step": 38575
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6729196906089783,
      "learning_rate": 0.0005594479307732742,
      "loss": 2.2588,
      "step": 38576
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.7322214841842651,
      "learning_rate": 0.000559445876996733,
      "loss": 2.3259,
      "step": 38577
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.678813636302948,
      "learning_rate": 0.000559443823171956,
      "loss": 2.115,
      "step": 38578
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.7442527413368225,
      "learning_rate": 0.0005594417692989435,
      "loss": 2.325,
      "step": 38579
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6593942642211914,
      "learning_rate": 0.0005594397153776958,
      "loss": 2.1645,
      "step": 38580
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.7188473343849182,
      "learning_rate": 0.0005594376614082134,
      "loss": 2.4879,
      "step": 38581
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.7056527137756348,
      "learning_rate": 0.0005594356073904967,
      "loss": 2.2573,
      "step": 38582
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6873767375946045,
      "learning_rate": 0.0005594335533245459,
      "loss": 2.221,
      "step": 38583
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.7005608677864075,
      "learning_rate": 0.0005594314992103615,
      "loss": 2.2524,
      "step": 38584
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.645180344581604,
      "learning_rate": 0.0005594294450479441,
      "loss": 2.2454,
      "step": 38585
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.7589175701141357,
      "learning_rate": 0.0005594273908372937,
      "loss": 2.5484,
      "step": 38586
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6835634708404541,
      "learning_rate": 0.0005594253365784108,
      "loss": 2.262,
      "step": 38587
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.676506757736206,
      "learning_rate": 0.0005594232822712959,
      "loss": 2.4041,
      "step": 38588
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6616753935813904,
      "learning_rate": 0.0005594212279159493,
      "loss": 2.0519,
      "step": 38589
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6830675005912781,
      "learning_rate": 0.0005594191735123714,
      "loss": 2.1413,
      "step": 38590
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6712913513183594,
      "learning_rate": 0.0005594171190605626,
      "loss": 2.2429,
      "step": 38591
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6602123379707336,
      "learning_rate": 0.0005594150645605233,
      "loss": 2.3682,
      "step": 38592
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6372560262680054,
      "learning_rate": 0.0005594130100122538,
      "loss": 2.3344,
      "step": 38593
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6532660126686096,
      "learning_rate": 0.0005594109554157545,
      "loss": 2.0158,
      "step": 38594
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6487715244293213,
      "learning_rate": 0.0005594089007710259,
      "loss": 2.2299,
      "step": 38595
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6716013550758362,
      "learning_rate": 0.0005594068460780682,
      "loss": 2.4207,
      "step": 38596
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6922296285629272,
      "learning_rate": 0.0005594047913368819,
      "loss": 2.2069,
      "step": 38597
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.7694524526596069,
      "learning_rate": 0.0005594027365474673,
      "loss": 2.0978,
      "step": 38598
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.7108725905418396,
      "learning_rate": 0.0005594006817098249,
      "loss": 2.0434,
      "step": 38599
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6430910229682922,
      "learning_rate": 0.0005593986268239551,
      "loss": 2.0696,
      "step": 38600
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.7105593085289001,
      "learning_rate": 0.000559396571889858,
      "loss": 2.2946,
      "step": 38601
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.7423733472824097,
      "learning_rate": 0.0005593945169075342,
      "loss": 2.1917,
      "step": 38602
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6377955675125122,
      "learning_rate": 0.0005593924618769842,
      "loss": 1.931,
      "step": 38603
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.7126961946487427,
      "learning_rate": 0.0005593904067982082,
      "loss": 2.3693,
      "step": 38604
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.7373737692832947,
      "learning_rate": 0.0005593883516712066,
      "loss": 2.4575,
      "step": 38605
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6866686940193176,
      "learning_rate": 0.0005593862964959797,
      "loss": 2.2395,
      "step": 38606
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.651517927646637,
      "learning_rate": 0.0005593842412725281,
      "loss": 2.3061,
      "step": 38607
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6275423765182495,
      "learning_rate": 0.0005593821860008522,
      "loss": 1.8875,
      "step": 38608
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.7824596762657166,
      "learning_rate": 0.0005593801306809521,
      "loss": 2.2483,
      "step": 38609
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6567779183387756,
      "learning_rate": 0.0005593780753128284,
      "loss": 2.3861,
      "step": 38610
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6874179244041443,
      "learning_rate": 0.0005593760198964814,
      "loss": 2.4674,
      "step": 38611
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.7006148099899292,
      "learning_rate": 0.0005593739644319116,
      "loss": 2.081,
      "step": 38612
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.7623172998428345,
      "learning_rate": 0.000559371908919119,
      "loss": 2.3849,
      "step": 38613
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.743966817855835,
      "learning_rate": 0.0005593698533581045,
      "loss": 2.307,
      "step": 38614
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.7032042741775513,
      "learning_rate": 0.0005593677977488682,
      "loss": 2.3173,
      "step": 38615
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.689605176448822,
      "learning_rate": 0.0005593657420914105,
      "loss": 2.1583,
      "step": 38616
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.748035728931427,
      "learning_rate": 0.0005593636863857319,
      "loss": 2.4446,
      "step": 38617
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.7102162837982178,
      "learning_rate": 0.0005593616306318326,
      "loss": 2.4003,
      "step": 38618
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6650283336639404,
      "learning_rate": 0.0005593595748297131,
      "loss": 2.2113,
      "step": 38619
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.7678616046905518,
      "learning_rate": 0.0005593575189793737,
      "loss": 2.1574,
      "step": 38620
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6241934895515442,
      "learning_rate": 0.000559355463080815,
      "loss": 2.4092,
      "step": 38621
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.7151004076004028,
      "learning_rate": 0.0005593534071340372,
      "loss": 2.3549,
      "step": 38622
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6586014628410339,
      "learning_rate": 0.0005593513511390405,
      "loss": 2.1722,
      "step": 38623
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6963439583778381,
      "learning_rate": 0.0005593492950958257,
      "loss": 2.3073,
      "step": 38624
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6794204711914062,
      "learning_rate": 0.0005593472390043929,
      "loss": 2.6082,
      "step": 38625
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.7181997895240784,
      "learning_rate": 0.0005593451828647425,
      "loss": 2.0548,
      "step": 38626
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.701278805732727,
      "learning_rate": 0.0005593431266768751,
      "loss": 2.3428,
      "step": 38627
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6875413656234741,
      "learning_rate": 0.0005593410704407907,
      "loss": 2.3309,
      "step": 38628
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6494967341423035,
      "learning_rate": 0.0005593390141564901,
      "loss": 2.1692,
      "step": 38629
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.809781551361084,
      "learning_rate": 0.0005593369578239734,
      "loss": 2.1295,
      "step": 38630
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.7350251078605652,
      "learning_rate": 0.000559334901443241,
      "loss": 2.3358,
      "step": 38631
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.7660307884216309,
      "learning_rate": 0.0005593328450142933,
      "loss": 2.3451,
      "step": 38632
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.9266886711120605,
      "learning_rate": 0.0005593307885371309,
      "loss": 2.1119,
      "step": 38633
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6858447790145874,
      "learning_rate": 0.0005593287320117539,
      "loss": 2.3869,
      "step": 38634
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.7042403221130371,
      "learning_rate": 0.0005593266754381628,
      "loss": 2.3306,
      "step": 38635
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.8369943499565125,
      "learning_rate": 0.000559324618816358,
      "loss": 2.3173,
      "step": 38636
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6961526274681091,
      "learning_rate": 0.0005593225621463399,
      "loss": 2.1813,
      "step": 38637
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.9102303981781006,
      "learning_rate": 0.0005593205054281088,
      "loss": 2.4292,
      "step": 38638
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.8259094953536987,
      "learning_rate": 0.0005593184486616652,
      "loss": 2.2273,
      "step": 38639
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6443530917167664,
      "learning_rate": 0.0005593163918470093,
      "loss": 2.2187,
      "step": 38640
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.8040115833282471,
      "learning_rate": 0.0005593143349841417,
      "loss": 2.3091,
      "step": 38641
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.7583508491516113,
      "learning_rate": 0.0005593122780730625,
      "loss": 2.2858,
      "step": 38642
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6416403651237488,
      "learning_rate": 0.0005593102211137723,
      "loss": 2.1908,
      "step": 38643
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.704031229019165,
      "learning_rate": 0.0005593081641062714,
      "loss": 2.5222,
      "step": 38644
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6924042105674744,
      "learning_rate": 0.0005593061070505603,
      "loss": 2.1933,
      "step": 38645
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6691151261329651,
      "learning_rate": 0.0005593040499466393,
      "loss": 2.0734,
      "step": 38646
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.7030514478683472,
      "learning_rate": 0.0005593019927945088,
      "loss": 2.2947,
      "step": 38647
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6727451086044312,
      "learning_rate": 0.0005592999355941692,
      "loss": 2.568,
      "step": 38648
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.7220269441604614,
      "learning_rate": 0.0005592978783456208,
      "loss": 1.9563,
      "step": 38649
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6776315569877625,
      "learning_rate": 0.000559295821048864,
      "loss": 2.2064,
      "step": 38650
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6505590081214905,
      "learning_rate": 0.0005592937637038992,
      "loss": 2.3627,
      "step": 38651
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.7085984945297241,
      "learning_rate": 0.0005592917063107268,
      "loss": 2.4189,
      "step": 38652
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.675595223903656,
      "learning_rate": 0.0005592896488693473,
      "loss": 2.1336,
      "step": 38653
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.7692010402679443,
      "learning_rate": 0.0005592875913797608,
      "loss": 2.2556,
      "step": 38654
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6584293246269226,
      "learning_rate": 0.0005592855338419679,
      "loss": 2.1662,
      "step": 38655
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6344857811927795,
      "learning_rate": 0.000559283476255969,
      "loss": 2.2151,
      "step": 38656
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6993483901023865,
      "learning_rate": 0.0005592814186217643,
      "loss": 2.2995,
      "step": 38657
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.785576343536377,
      "learning_rate": 0.0005592793609393543,
      "loss": 2.135,
      "step": 38658
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6599603891372681,
      "learning_rate": 0.0005592773032087394,
      "loss": 2.2559,
      "step": 38659
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.7160370945930481,
      "learning_rate": 0.00055927524542992,
      "loss": 2.0648,
      "step": 38660
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6509510278701782,
      "learning_rate": 0.0005592731876028964,
      "loss": 2.2985,
      "step": 38661
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.702496349811554,
      "learning_rate": 0.000559271129727669,
      "loss": 2.3484,
      "step": 38662
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6757904291152954,
      "learning_rate": 0.0005592690718042382,
      "loss": 2.2439,
      "step": 38663
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6843082904815674,
      "learning_rate": 0.0005592670138326045,
      "loss": 2.2017,
      "step": 38664
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.7385247945785522,
      "learning_rate": 0.000559264955812768,
      "loss": 2.5745,
      "step": 38665
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6947309374809265,
      "learning_rate": 0.0005592628977447293,
      "loss": 2.1738,
      "step": 38666
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.721213698387146,
      "learning_rate": 0.0005592608396284888,
      "loss": 2.4507,
      "step": 38667
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6692251563072205,
      "learning_rate": 0.0005592587814640467,
      "loss": 2.2179,
      "step": 38668
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6471080780029297,
      "learning_rate": 0.0005592567232514036,
      "loss": 2.3756,
      "step": 38669
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.668846070766449,
      "learning_rate": 0.0005592546649905597,
      "loss": 2.5131,
      "step": 38670
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6398253440856934,
      "learning_rate": 0.0005592526066815156,
      "loss": 2.0852,
      "step": 38671
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6522176265716553,
      "learning_rate": 0.0005592505483242715,
      "loss": 2.1878,
      "step": 38672
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6771922707557678,
      "learning_rate": 0.0005592484899188278,
      "loss": 2.3201,
      "step": 38673
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6739504933357239,
      "learning_rate": 0.0005592464314651848,
      "loss": 2.3309,
      "step": 38674
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6630104184150696,
      "learning_rate": 0.0005592443729633431,
      "loss": 2.2394,
      "step": 38675
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.613054096698761,
      "learning_rate": 0.000559242314413303,
      "loss": 2.4431,
      "step": 38676
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6803172826766968,
      "learning_rate": 0.0005592402558150648,
      "loss": 2.3612,
      "step": 38677
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6371217966079712,
      "learning_rate": 0.000559238197168629,
      "loss": 2.3795,
      "step": 38678
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6693997979164124,
      "learning_rate": 0.0005592361384739959,
      "loss": 2.3852,
      "step": 38679
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6410285234451294,
      "learning_rate": 0.0005592340797311659,
      "loss": 2.1912,
      "step": 38680
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6080919504165649,
      "learning_rate": 0.0005592320209401394,
      "loss": 2.1282,
      "step": 38681
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6226646900177002,
      "learning_rate": 0.0005592299621009168,
      "loss": 2.2435,
      "step": 38682
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6152526140213013,
      "learning_rate": 0.0005592279032134984,
      "loss": 2.3467,
      "step": 38683
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.7652521729469299,
      "learning_rate": 0.0005592258442778847,
      "loss": 2.0,
      "step": 38684
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.707428514957428,
      "learning_rate": 0.0005592237852940759,
      "loss": 2.422,
      "step": 38685
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.7085235118865967,
      "learning_rate": 0.0005592217262620727,
      "loss": 2.2733,
      "step": 38686
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6568328738212585,
      "learning_rate": 0.0005592196671818751,
      "loss": 2.3377,
      "step": 38687
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.738558828830719,
      "learning_rate": 0.0005592176080534838,
      "loss": 2.3861,
      "step": 38688
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6617802977561951,
      "learning_rate": 0.000559215548876899,
      "loss": 2.2453,
      "step": 38689
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6112197637557983,
      "learning_rate": 0.0005592134896521212,
      "loss": 2.2452,
      "step": 38690
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.7368168234825134,
      "learning_rate": 0.0005592114303791506,
      "loss": 2.2719,
      "step": 38691
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6171866655349731,
      "learning_rate": 0.0005592093710579878,
      "loss": 2.2175,
      "step": 38692
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.7369301319122314,
      "learning_rate": 0.0005592073116886331,
      "loss": 2.2911,
      "step": 38693
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.755631685256958,
      "learning_rate": 0.0005592052522710867,
      "loss": 2.3675,
      "step": 38694
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6578329801559448,
      "learning_rate": 0.0005592031928053494,
      "loss": 2.4388,
      "step": 38695
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.7204320430755615,
      "learning_rate": 0.0005592011332914211,
      "loss": 2.2057,
      "step": 38696
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.707902193069458,
      "learning_rate": 0.0005591990737293025,
      "loss": 2.1022,
      "step": 38697
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.7122653722763062,
      "learning_rate": 0.000559197014118994,
      "loss": 2.1993,
      "step": 38698
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.684563398361206,
      "learning_rate": 0.0005591949544604958,
      "loss": 2.1934,
      "step": 38699
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.7187482714653015,
      "learning_rate": 0.0005591928947538084,
      "loss": 2.1838,
      "step": 38700
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.7084661722183228,
      "learning_rate": 0.0005591908349989321,
      "loss": 2.2417,
      "step": 38701
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.7489786744117737,
      "learning_rate": 0.0005591887751958673,
      "loss": 2.2563,
      "step": 38702
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.7073109745979309,
      "learning_rate": 0.0005591867153446145,
      "loss": 2.1324,
      "step": 38703
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.7651098966598511,
      "learning_rate": 0.0005591846554451739,
      "loss": 2.322,
      "step": 38704
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.8673226237297058,
      "learning_rate": 0.0005591825954975461,
      "loss": 2.4234,
      "step": 38705
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.943519651889801,
      "learning_rate": 0.0005591805355017312,
      "loss": 2.2522,
      "step": 38706
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.7526267170906067,
      "learning_rate": 0.00055917847545773,
      "loss": 2.3158,
      "step": 38707
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6815983653068542,
      "learning_rate": 0.0005591764153655424,
      "loss": 2.138,
      "step": 38708
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.8032360076904297,
      "learning_rate": 0.0005591743552251691,
      "loss": 2.2719,
      "step": 38709
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6954429745674133,
      "learning_rate": 0.0005591722950366105,
      "loss": 2.3,
      "step": 38710
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6966077089309692,
      "learning_rate": 0.0005591702347998667,
      "loss": 2.1316,
      "step": 38711
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.640042245388031,
      "learning_rate": 0.0005591681745149384,
      "loss": 2.4403,
      "step": 38712
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6717579364776611,
      "learning_rate": 0.0005591661141818257,
      "loss": 2.224,
      "step": 38713
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6599724292755127,
      "learning_rate": 0.0005591640538005292,
      "loss": 2.3474,
      "step": 38714
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.635308563709259,
      "learning_rate": 0.0005591619933710492,
      "loss": 2.2861,
      "step": 38715
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6725031733512878,
      "learning_rate": 0.0005591599328933861,
      "loss": 2.2537,
      "step": 38716
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.7292619943618774,
      "learning_rate": 0.0005591578723675402,
      "loss": 2.0188,
      "step": 38717
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.7178628444671631,
      "learning_rate": 0.0005591558117935121,
      "loss": 2.3851,
      "step": 38718
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.7043046951293945,
      "learning_rate": 0.000559153751171302,
      "loss": 2.4782,
      "step": 38719
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6178948879241943,
      "learning_rate": 0.0005591516905009103,
      "loss": 2.483,
      "step": 38720
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6910507678985596,
      "learning_rate": 0.0005591496297823374,
      "loss": 2.1364,
      "step": 38721
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6769295930862427,
      "learning_rate": 0.0005591475690155836,
      "loss": 2.3273,
      "step": 38722
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6668289303779602,
      "learning_rate": 0.0005591455082006496,
      "loss": 2.2441,
      "step": 38723
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6979066133499146,
      "learning_rate": 0.0005591434473375354,
      "loss": 2.3453,
      "step": 38724
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.7169561386108398,
      "learning_rate": 0.0005591413864262416,
      "loss": 2.5861,
      "step": 38725
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6072290539741516,
      "learning_rate": 0.0005591393254667685,
      "loss": 2.2696,
      "step": 38726
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.7121720314025879,
      "learning_rate": 0.0005591372644591165,
      "loss": 2.4368,
      "step": 38727
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6061352491378784,
      "learning_rate": 0.0005591352034032861,
      "loss": 2.2172,
      "step": 38728
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6923290491104126,
      "learning_rate": 0.0005591331422992775,
      "loss": 2.2143,
      "step": 38729
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.7649573087692261,
      "learning_rate": 0.000559131081147091,
      "loss": 2.3121,
      "step": 38730
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6820746660232544,
      "learning_rate": 0.0005591290199467273,
      "loss": 2.4356,
      "step": 38731
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.7379562854766846,
      "learning_rate": 0.0005591269586981867,
      "loss": 2.2264,
      "step": 38732
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.7098322510719299,
      "learning_rate": 0.0005591248974014694,
      "loss": 2.3737,
      "step": 38733
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6974220871925354,
      "learning_rate": 0.0005591228360565758,
      "loss": 2.2507,
      "step": 38734
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6594680547714233,
      "learning_rate": 0.0005591207746635065,
      "loss": 2.35,
      "step": 38735
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.7157738208770752,
      "learning_rate": 0.0005591187132222617,
      "loss": 2.3939,
      "step": 38736
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.7431190013885498,
      "learning_rate": 0.0005591166517328419,
      "loss": 2.5312,
      "step": 38737
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.736077606678009,
      "learning_rate": 0.0005591145901952474,
      "loss": 2.1649,
      "step": 38738
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6671558618545532,
      "learning_rate": 0.0005591125286094786,
      "loss": 2.2509,
      "step": 38739
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.7515292763710022,
      "learning_rate": 0.0005591104669755359,
      "loss": 2.2948,
      "step": 38740
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.7281774282455444,
      "learning_rate": 0.0005591084052934196,
      "loss": 2.2431,
      "step": 38741
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6955920457839966,
      "learning_rate": 0.0005591063435631303,
      "loss": 2.0933,
      "step": 38742
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.7811431288719177,
      "learning_rate": 0.0005591042817846681,
      "loss": 2.2712,
      "step": 38743
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.8493810296058655,
      "learning_rate": 0.0005591022199580336,
      "loss": 2.2568,
      "step": 38744
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6914254426956177,
      "learning_rate": 0.000559100158083227,
      "loss": 2.1067,
      "step": 38745
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6998646259307861,
      "learning_rate": 0.000559098096160249,
      "loss": 2.234,
      "step": 38746
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6800134181976318,
      "learning_rate": 0.0005590960341890996,
      "loss": 2.1508,
      "step": 38747
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.7467674016952515,
      "learning_rate": 0.0005590939721697793,
      "loss": 2.3427,
      "step": 38748
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6766002178192139,
      "learning_rate": 0.0005590919101022887,
      "loss": 2.2501,
      "step": 38749
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.8258422613143921,
      "learning_rate": 0.000559089847986628,
      "loss": 2.3815,
      "step": 38750
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.700555145740509,
      "learning_rate": 0.0005590877858227976,
      "loss": 2.1591,
      "step": 38751
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6753339767456055,
      "learning_rate": 0.0005590857236107978,
      "loss": 2.4624,
      "step": 38752
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.7226613163948059,
      "learning_rate": 0.0005590836613506291,
      "loss": 2.0761,
      "step": 38753
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6552157998085022,
      "learning_rate": 0.000559081599042292,
      "loss": 2.38,
      "step": 38754
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6149161458015442,
      "learning_rate": 0.0005590795366857866,
      "loss": 2.1952,
      "step": 38755
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.7039092779159546,
      "learning_rate": 0.0005590774742811135,
      "loss": 2.086,
      "step": 38756
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.7150452136993408,
      "learning_rate": 0.0005590754118282729,
      "loss": 2.2631,
      "step": 38757
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6761785745620728,
      "learning_rate": 0.0005590733493272653,
      "loss": 1.947,
      "step": 38758
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.7251183390617371,
      "learning_rate": 0.0005590712867780913,
      "loss": 2.1708,
      "step": 38759
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6681898832321167,
      "learning_rate": 0.0005590692241807507,
      "loss": 2.1619,
      "step": 38760
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.734191358089447,
      "learning_rate": 0.0005590671615352445,
      "loss": 2.1587,
      "step": 38761
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.632667064666748,
      "learning_rate": 0.0005590650988415727,
      "loss": 2.2278,
      "step": 38762
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.7184704542160034,
      "learning_rate": 0.000559063036099736,
      "loss": 2.0987,
      "step": 38763
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6866652369499207,
      "learning_rate": 0.0005590609733097344,
      "loss": 2.1164,
      "step": 38764
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6870722770690918,
      "learning_rate": 0.0005590589104715686,
      "loss": 2.3027,
      "step": 38765
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.7019772529602051,
      "learning_rate": 0.0005590568475852387,
      "loss": 2.1595,
      "step": 38766
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6832450032234192,
      "learning_rate": 0.0005590547846507453,
      "loss": 2.3476,
      "step": 38767
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6746577024459839,
      "learning_rate": 0.0005590527216680888,
      "loss": 2.4373,
      "step": 38768
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.5795153975486755,
      "learning_rate": 0.0005590506586372695,
      "loss": 2.0061,
      "step": 38769
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.8877397775650024,
      "learning_rate": 0.0005590485955582877,
      "loss": 2.382,
      "step": 38770
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6452001929283142,
      "learning_rate": 0.000559046532431144,
      "loss": 2.2249,
      "step": 38771
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6788631677627563,
      "learning_rate": 0.0005590444692558386,
      "loss": 2.056,
      "step": 38772
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.7718449234962463,
      "learning_rate": 0.0005590424060323718,
      "loss": 2.5042,
      "step": 38773
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.8872328400611877,
      "learning_rate": 0.0005590403427607443,
      "loss": 2.4537,
      "step": 38774
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.667675256729126,
      "learning_rate": 0.0005590382794409563,
      "loss": 2.2989,
      "step": 38775
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6896356344223022,
      "learning_rate": 0.0005590362160730082,
      "loss": 2.1576,
      "step": 38776
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.7544156312942505,
      "learning_rate": 0.0005590341526569003,
      "loss": 2.2054,
      "step": 38777
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6102579236030579,
      "learning_rate": 0.000559032089192633,
      "loss": 2.246,
      "step": 38778
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6542252898216248,
      "learning_rate": 0.0005590300256802068,
      "loss": 2.414,
      "step": 38779
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6623474955558777,
      "learning_rate": 0.0005590279621196222,
      "loss": 2.371,
      "step": 38780
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.7250574231147766,
      "learning_rate": 0.0005590258985108793,
      "loss": 2.409,
      "step": 38781
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6513857841491699,
      "learning_rate": 0.0005590238348539786,
      "loss": 2.367,
      "step": 38782
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.8963853120803833,
      "learning_rate": 0.0005590217711489204,
      "loss": 2.2925,
      "step": 38783
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.7225130200386047,
      "learning_rate": 0.0005590197073957053,
      "loss": 2.3751,
      "step": 38784
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.716817319393158,
      "learning_rate": 0.0005590176435943335,
      "loss": 2.2542,
      "step": 38785
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6976175904273987,
      "learning_rate": 0.0005590155797448054,
      "loss": 2.3837,
      "step": 38786
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6843535900115967,
      "learning_rate": 0.0005590135158471213,
      "loss": 2.227,
      "step": 38787
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.670815110206604,
      "learning_rate": 0.0005590114519012819,
      "loss": 2.397,
      "step": 38788
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.5967589616775513,
      "learning_rate": 0.0005590093879072873,
      "loss": 2.4145,
      "step": 38789
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.7154949307441711,
      "learning_rate": 0.000559007323865138,
      "loss": 2.2861,
      "step": 38790
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.7426646947860718,
      "learning_rate": 0.0005590052597748342,
      "loss": 2.304,
      "step": 38791
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6977558135986328,
      "learning_rate": 0.0005590031956363766,
      "loss": 2.1449,
      "step": 38792
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6475809216499329,
      "learning_rate": 0.0005590011314497654,
      "loss": 2.2639,
      "step": 38793
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.7580709457397461,
      "learning_rate": 0.0005589990672150009,
      "loss": 2.2999,
      "step": 38794
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.8592355251312256,
      "learning_rate": 0.0005589970029320837,
      "loss": 2.3541,
      "step": 38795
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.7023661732673645,
      "learning_rate": 0.000558994938601014,
      "loss": 2.2898,
      "step": 38796
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6874511241912842,
      "learning_rate": 0.0005589928742217923,
      "loss": 2.5335,
      "step": 38797
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.7702730894088745,
      "learning_rate": 0.0005589908097944189,
      "loss": 2.0714,
      "step": 38798
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6657934784889221,
      "learning_rate": 0.0005589887453188942,
      "loss": 2.4429,
      "step": 38799
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.878625214099884,
      "learning_rate": 0.0005589866807952186,
      "loss": 2.5228,
      "step": 38800
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.7344892621040344,
      "learning_rate": 0.0005589846162233925,
      "loss": 2.1918,
      "step": 38801
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.7296226620674133,
      "learning_rate": 0.0005589825516034163,
      "loss": 2.2567,
      "step": 38802
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.7729061245918274,
      "learning_rate": 0.0005589804869352904,
      "loss": 2.2741,
      "step": 38803
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.7202446460723877,
      "learning_rate": 0.0005589784222190152,
      "loss": 2.2567,
      "step": 38804
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6963604092597961,
      "learning_rate": 0.0005589763574545908,
      "loss": 2.0907,
      "step": 38805
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6909793615341187,
      "learning_rate": 0.0005589742926420179,
      "loss": 2.1989,
      "step": 38806
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.7226160168647766,
      "learning_rate": 0.0005589722277812968,
      "loss": 2.2232,
      "step": 38807
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.7901099920272827,
      "learning_rate": 0.0005589701628724279,
      "loss": 2.2956,
      "step": 38808
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6434945464134216,
      "learning_rate": 0.0005589680979154115,
      "loss": 2.215,
      "step": 38809
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.7112414836883545,
      "learning_rate": 0.0005589660329102481,
      "loss": 2.4426,
      "step": 38810
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.7350051999092102,
      "learning_rate": 0.000558963967856938,
      "loss": 2.3159,
      "step": 38811
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6679667234420776,
      "learning_rate": 0.0005589619027554817,
      "loss": 2.3165,
      "step": 38812
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.682057797908783,
      "learning_rate": 0.0005589598376058793,
      "loss": 2.251,
      "step": 38813
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6655476093292236,
      "learning_rate": 0.0005589577724081315,
      "loss": 2.1744,
      "step": 38814
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.764840841293335,
      "learning_rate": 0.0005589557071622386,
      "loss": 2.3264,
      "step": 38815
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6564813256263733,
      "learning_rate": 0.0005589536418682009,
      "loss": 2.2887,
      "step": 38816
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.8000612854957581,
      "learning_rate": 0.0005589515765260188,
      "loss": 2.2361,
      "step": 38817
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.7514024972915649,
      "learning_rate": 0.0005589495111356927,
      "loss": 2.323,
      "step": 38818
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.7060465216636658,
      "learning_rate": 0.000558947445697223,
      "loss": 2.2561,
      "step": 38819
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.7002527117729187,
      "learning_rate": 0.0005589453802106102,
      "loss": 2.4341,
      "step": 38820
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.720729649066925,
      "learning_rate": 0.0005589433146758544,
      "loss": 2.2142,
      "step": 38821
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.756718099117279,
      "learning_rate": 0.0005589412490929563,
      "loss": 2.4681,
      "step": 38822
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6213138699531555,
      "learning_rate": 0.000558939183461916,
      "loss": 2.4068,
      "step": 38823
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.7157274484634399,
      "learning_rate": 0.0005589371177827341,
      "loss": 2.3979,
      "step": 38824
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.7012889385223389,
      "learning_rate": 0.0005589350520554109,
      "loss": 2.1008,
      "step": 38825
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6966363787651062,
      "learning_rate": 0.0005589329862799468,
      "loss": 2.3297,
      "step": 38826
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6985730528831482,
      "learning_rate": 0.0005589309204563421,
      "loss": 2.3091,
      "step": 38827
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6253502368927002,
      "learning_rate": 0.0005589288545845973,
      "loss": 2.6478,
      "step": 38828
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6212183237075806,
      "learning_rate": 0.0005589267886647127,
      "loss": 2.2816,
      "step": 38829
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6745285391807556,
      "learning_rate": 0.0005589247226966888,
      "loss": 2.5246,
      "step": 38830
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.7068139910697937,
      "learning_rate": 0.0005589226566805258,
      "loss": 2.3967,
      "step": 38831
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.7110483050346375,
      "learning_rate": 0.0005589205906162243,
      "loss": 2.3637,
      "step": 38832
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6291517615318298,
      "learning_rate": 0.0005589185245037845,
      "loss": 2.3675,
      "step": 38833
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6482730507850647,
      "learning_rate": 0.0005589164583432069,
      "loss": 2.0956,
      "step": 38834
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6975287795066833,
      "learning_rate": 0.0005589143921344919,
      "loss": 1.9897,
      "step": 38835
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.8069513440132141,
      "learning_rate": 0.0005589123258776397,
      "loss": 2.2898,
      "step": 38836
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6436265110969543,
      "learning_rate": 0.0005589102595726508,
      "loss": 2.2828,
      "step": 38837
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.7229249477386475,
      "learning_rate": 0.0005589081932195257,
      "loss": 2.3352,
      "step": 38838
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.7885556817054749,
      "learning_rate": 0.0005589061268182647,
      "loss": 2.1115,
      "step": 38839
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6682722568511963,
      "learning_rate": 0.000558904060368868,
      "loss": 2.0558,
      "step": 38840
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.7175440788269043,
      "learning_rate": 0.0005589019938713364,
      "loss": 2.1643,
      "step": 38841
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6296047568321228,
      "learning_rate": 0.0005588999273256699,
      "loss": 2.2463,
      "step": 38842
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.2050377130508423,
      "learning_rate": 0.000558897860731869,
      "loss": 2.2857,
      "step": 38843
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6916912794113159,
      "learning_rate": 0.000558895794089934,
      "loss": 2.1677,
      "step": 38844
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.704197347164154,
      "learning_rate": 0.0005588937273998656,
      "loss": 2.1087,
      "step": 38845
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6906378269195557,
      "learning_rate": 0.0005588916606616638,
      "loss": 2.1585,
      "step": 38846
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.7177673578262329,
      "learning_rate": 0.0005588895938753291,
      "loss": 2.4193,
      "step": 38847
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6897974014282227,
      "learning_rate": 0.0005588875270408621,
      "loss": 2.306,
      "step": 38848
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6476647853851318,
      "learning_rate": 0.0005588854601582631,
      "loss": 1.9716,
      "step": 38849
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.5906103849411011,
      "learning_rate": 0.0005588833932275323,
      "loss": 2.072,
      "step": 38850
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.718989372253418,
      "learning_rate": 0.00055888132624867,
      "loss": 2.3566,
      "step": 38851
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.7065452337265015,
      "learning_rate": 0.000558879259221677,
      "loss": 2.1012,
      "step": 38852
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6877858638763428,
      "learning_rate": 0.0005588771921465534,
      "loss": 2.2696,
      "step": 38853
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6316595673561096,
      "learning_rate": 0.0005588751250232997,
      "loss": 2.3636,
      "step": 38854
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.7966728806495667,
      "learning_rate": 0.0005588730578519161,
      "loss": 2.4074,
      "step": 38855
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.7127631306648254,
      "learning_rate": 0.0005588709906324032,
      "loss": 2.2488,
      "step": 38856
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6110243797302246,
      "learning_rate": 0.0005588689233647613,
      "loss": 2.0471,
      "step": 38857
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.972999095916748,
      "learning_rate": 0.0005588668560489908,
      "loss": 2.344,
      "step": 38858
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6951794624328613,
      "learning_rate": 0.000558864788685092,
      "loss": 2.2626,
      "step": 38859
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6588084697723389,
      "learning_rate": 0.0005588627212730654,
      "loss": 2.3641,
      "step": 38860
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.0273680686950684,
      "learning_rate": 0.0005588606538129113,
      "loss": 2.1461,
      "step": 38861
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.7377768754959106,
      "learning_rate": 0.0005588585863046301,
      "loss": 2.3665,
      "step": 38862
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.9189233183860779,
      "learning_rate": 0.0005588565187482223,
      "loss": 2.3745,
      "step": 38863
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.7491429448127747,
      "learning_rate": 0.0005588544511436881,
      "loss": 2.2077,
      "step": 38864
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6633448004722595,
      "learning_rate": 0.0005588523834910279,
      "loss": 2.2741,
      "step": 38865
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6892273426055908,
      "learning_rate": 0.0005588503157902422,
      "loss": 2.3012,
      "step": 38866
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.7369005680084229,
      "learning_rate": 0.0005588482480413315,
      "loss": 2.0319,
      "step": 38867
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6386417150497437,
      "learning_rate": 0.0005588461802442958,
      "loss": 2.1908,
      "step": 38868
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.705169677734375,
      "learning_rate": 0.0005588441123991359,
      "loss": 2.0645,
      "step": 38869
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.7856740355491638,
      "learning_rate": 0.000558842044505852,
      "loss": 2.4511,
      "step": 38870
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6972308158874512,
      "learning_rate": 0.0005588399765644443,
      "loss": 2.0908,
      "step": 38871
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.7420215010643005,
      "learning_rate": 0.0005588379085749134,
      "loss": 2.202,
      "step": 38872
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.7713679075241089,
      "learning_rate": 0.0005588358405372597,
      "loss": 2.1894,
      "step": 38873
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.7422381043434143,
      "learning_rate": 0.0005588337724514835,
      "loss": 2.1476,
      "step": 38874
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.7277743220329285,
      "learning_rate": 0.0005588317043175852,
      "loss": 2.1986,
      "step": 38875
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.7316532731056213,
      "learning_rate": 0.0005588296361355653,
      "loss": 2.3138,
      "step": 38876
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.7713601589202881,
      "learning_rate": 0.0005588275679054241,
      "loss": 2.2982,
      "step": 38877
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.697876513004303,
      "learning_rate": 0.0005588254996271619,
      "loss": 2.4518,
      "step": 38878
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.8044924736022949,
      "learning_rate": 0.0005588234313007791,
      "loss": 2.3762,
      "step": 38879
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6718019843101501,
      "learning_rate": 0.0005588213629262762,
      "loss": 2.4758,
      "step": 38880
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6985352039337158,
      "learning_rate": 0.0005588192945036536,
      "loss": 2.3308,
      "step": 38881
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6797931790351868,
      "learning_rate": 0.0005588172260329114,
      "loss": 2.2306,
      "step": 38882
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6671309471130371,
      "learning_rate": 0.0005588151575140504,
      "loss": 2.0436,
      "step": 38883
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6259787082672119,
      "learning_rate": 0.0005588130889470706,
      "loss": 2.2926,
      "step": 38884
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.9232447147369385,
      "learning_rate": 0.0005588110203319727,
      "loss": 2.353,
      "step": 38885
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6806836128234863,
      "learning_rate": 0.0005588089516687569,
      "loss": 2.2417,
      "step": 38886
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.7308631539344788,
      "learning_rate": 0.0005588068829574237,
      "loss": 2.211,
      "step": 38887
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.8973677158355713,
      "learning_rate": 0.0005588048141979732,
      "loss": 2.2762,
      "step": 38888
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.7358025908470154,
      "learning_rate": 0.0005588027453904063,
      "loss": 2.6103,
      "step": 38889
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6890686750411987,
      "learning_rate": 0.0005588006765347229,
      "loss": 2.3638,
      "step": 38890
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.7698992490768433,
      "learning_rate": 0.0005587986076309235,
      "loss": 2.1712,
      "step": 38891
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6637818217277527,
      "learning_rate": 0.0005587965386790087,
      "loss": 2.1016,
      "step": 38892
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.7048127055168152,
      "learning_rate": 0.0005587944696789787,
      "loss": 2.2742,
      "step": 38893
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.7018910646438599,
      "learning_rate": 0.0005587924006308339,
      "loss": 2.1923,
      "step": 38894
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.8169146180152893,
      "learning_rate": 0.0005587903315345747,
      "loss": 2.2524,
      "step": 38895
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6975212693214417,
      "learning_rate": 0.0005587882623902015,
      "loss": 2.0472,
      "step": 38896
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6672688126564026,
      "learning_rate": 0.0005587861931977148,
      "loss": 2.139,
      "step": 38897
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6254123449325562,
      "learning_rate": 0.0005587841239571146,
      "loss": 2.2461,
      "step": 38898
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.9276823401451111,
      "learning_rate": 0.0005587820546684018,
      "loss": 2.1277,
      "step": 38899
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.8805751204490662,
      "learning_rate": 0.0005587799853315764,
      "loss": 2.172,
      "step": 38900
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.7276805639266968,
      "learning_rate": 0.000558777915946639,
      "loss": 2.1824,
      "step": 38901
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6127278804779053,
      "learning_rate": 0.0005587758465135899,
      "loss": 2.0999,
      "step": 38902
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6808339953422546,
      "learning_rate": 0.0005587737770324294,
      "loss": 2.1612,
      "step": 38903
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.8052239418029785,
      "learning_rate": 0.000558771707503158,
      "loss": 2.1932,
      "step": 38904
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.8239191770553589,
      "learning_rate": 0.000558769637925776,
      "loss": 2.3658,
      "step": 38905
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.657998263835907,
      "learning_rate": 0.000558767568300284,
      "loss": 1.988,
      "step": 38906
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.9595863223075867,
      "learning_rate": 0.000558765498626682,
      "loss": 2.2608,
      "step": 38907
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6488298177719116,
      "learning_rate": 0.0005587634289049708,
      "loss": 2.4129,
      "step": 38908
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6338115334510803,
      "learning_rate": 0.0005587613591351505,
      "loss": 2.3212,
      "step": 38909
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.669988214969635,
      "learning_rate": 0.0005587592893172216,
      "loss": 2.2698,
      "step": 38910
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.647087574005127,
      "learning_rate": 0.0005587572194511843,
      "loss": 2.197,
      "step": 38911
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6982393860816956,
      "learning_rate": 0.0005587551495370394,
      "loss": 2.1918,
      "step": 38912
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6345334649085999,
      "learning_rate": 0.0005587530795747869,
      "loss": 2.29,
      "step": 38913
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6451272964477539,
      "learning_rate": 0.0005587510095644274,
      "loss": 2.3596,
      "step": 38914
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.7247034907341003,
      "learning_rate": 0.0005587489395059612,
      "loss": 2.2721,
      "step": 38915
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6503580808639526,
      "learning_rate": 0.0005587468693993886,
      "loss": 2.2157,
      "step": 38916
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.5753293037414551,
      "learning_rate": 0.0005587447992447101,
      "loss": 1.9234,
      "step": 38917
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.636425793170929,
      "learning_rate": 0.0005587427290419261,
      "loss": 2.373,
      "step": 38918
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.7148193120956421,
      "learning_rate": 0.000558740658791037,
      "loss": 2.2423,
      "step": 38919
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.7279472947120667,
      "learning_rate": 0.000558738588492043,
      "loss": 2.294,
      "step": 38920
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6861666440963745,
      "learning_rate": 0.0005587365181449448,
      "loss": 2.435,
      "step": 38921
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6802473664283752,
      "learning_rate": 0.0005587344477497424,
      "loss": 1.9795,
      "step": 38922
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6680415868759155,
      "learning_rate": 0.0005587323773064364,
      "loss": 2.2085,
      "step": 38923
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.7505602836608887,
      "learning_rate": 0.0005587303068150273,
      "loss": 2.3846,
      "step": 38924
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.695839524269104,
      "learning_rate": 0.0005587282362755152,
      "loss": 2.269,
      "step": 38925
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.627122163772583,
      "learning_rate": 0.0005587261656879007,
      "loss": 2.2399,
      "step": 38926
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.7255228161811829,
      "learning_rate": 0.0005587240950521841,
      "loss": 2.2505,
      "step": 38927
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6520739197731018,
      "learning_rate": 0.0005587220243683658,
      "loss": 2.3543,
      "step": 38928
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6539474129676819,
      "learning_rate": 0.0005587199536364463,
      "loss": 2.247,
      "step": 38929
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6440900564193726,
      "learning_rate": 0.0005587178828564258,
      "loss": 2.2897,
      "step": 38930
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.8592984676361084,
      "learning_rate": 0.0005587158120283048,
      "loss": 2.2555,
      "step": 38931
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6777201890945435,
      "learning_rate": 0.0005587137411520837,
      "loss": 2.3714,
      "step": 38932
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.7705410718917847,
      "learning_rate": 0.0005587116702277627,
      "loss": 2.2211,
      "step": 38933
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.699612021446228,
      "learning_rate": 0.0005587095992553423,
      "loss": 2.2583,
      "step": 38934
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6669695973396301,
      "learning_rate": 0.0005587075282348229,
      "loss": 2.1341,
      "step": 38935
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6408336758613586,
      "learning_rate": 0.000558705457166205,
      "loss": 2.175,
      "step": 38936
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6750351786613464,
      "learning_rate": 0.0005587033860494888,
      "loss": 1.9468,
      "step": 38937
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.7197229862213135,
      "learning_rate": 0.0005587013148846749,
      "loss": 2.2743,
      "step": 38938
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.7980313897132874,
      "learning_rate": 0.0005586992436717633,
      "loss": 2.2302,
      "step": 38939
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.8208560347557068,
      "learning_rate": 0.0005586971724107548,
      "loss": 2.2369,
      "step": 38940
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.9439864754676819,
      "learning_rate": 0.0005586951011016495,
      "loss": 2.2115,
      "step": 38941
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.608212411403656,
      "learning_rate": 0.000558693029744448,
      "loss": 2.1987,
      "step": 38942
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.7132753729820251,
      "learning_rate": 0.0005586909583391506,
      "loss": 2.021,
      "step": 38943
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.7199951410293579,
      "learning_rate": 0.0005586888868857575,
      "loss": 2.2787,
      "step": 38944
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.7400840520858765,
      "learning_rate": 0.0005586868153842694,
      "loss": 2.3386,
      "step": 38945
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.7252143621444702,
      "learning_rate": 0.0005586847438346865,
      "loss": 2.2765,
      "step": 38946
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6542911529541016,
      "learning_rate": 0.0005586826722370092,
      "loss": 2.116,
      "step": 38947
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6922786235809326,
      "learning_rate": 0.000558680600591238,
      "loss": 2.3889,
      "step": 38948
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.7261157631874084,
      "learning_rate": 0.000558678528897373,
      "loss": 2.2608,
      "step": 38949
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6487961411476135,
      "learning_rate": 0.0005586764571554149,
      "loss": 2.1897,
      "step": 38950
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6729921102523804,
      "learning_rate": 0.000558674385365364,
      "loss": 2.3841,
      "step": 38951
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6490383148193359,
      "learning_rate": 0.0005586723135272206,
      "loss": 2.17,
      "step": 38952
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.7265528440475464,
      "learning_rate": 0.0005586702416409852,
      "loss": 2.439,
      "step": 38953
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6676310300827026,
      "learning_rate": 0.0005586681697066581,
      "loss": 2.3213,
      "step": 38954
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6705345511436462,
      "learning_rate": 0.0005586660977242396,
      "loss": 2.3755,
      "step": 38955
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6382991075515747,
      "learning_rate": 0.0005586640256937303,
      "loss": 2.153,
      "step": 38956
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6435081362724304,
      "learning_rate": 0.0005586619536151304,
      "loss": 2.1295,
      "step": 38957
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.7547958493232727,
      "learning_rate": 0.0005586598814884404,
      "loss": 2.2675,
      "step": 38958
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6786555647850037,
      "learning_rate": 0.0005586578093136606,
      "loss": 2.1295,
      "step": 38959
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.7405670285224915,
      "learning_rate": 0.0005586557370907915,
      "loss": 2.5113,
      "step": 38960
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6616686582565308,
      "learning_rate": 0.0005586536648198334,
      "loss": 2.1501,
      "step": 38961
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.7271254658699036,
      "learning_rate": 0.0005586515925007867,
      "loss": 2.2479,
      "step": 38962
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6708195805549622,
      "learning_rate": 0.0005586495201336518,
      "loss": 2.1342,
      "step": 38963
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.7279259562492371,
      "learning_rate": 0.0005586474477184291,
      "loss": 2.2801,
      "step": 38964
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6995972394943237,
      "learning_rate": 0.0005586453752551188,
      "loss": 2.2744,
      "step": 38965
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.63297438621521,
      "learning_rate": 0.0005586433027437215,
      "loss": 2.2698,
      "step": 38966
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6855058073997498,
      "learning_rate": 0.0005586412301842377,
      "loss": 2.4374,
      "step": 38967
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.7125771641731262,
      "learning_rate": 0.0005586391575766674,
      "loss": 2.013,
      "step": 38968
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6840186715126038,
      "learning_rate": 0.0005586370849210114,
      "loss": 2.3329,
      "step": 38969
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.68524169921875,
      "learning_rate": 0.0005586350122172697,
      "loss": 2.3556,
      "step": 38970
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.7130311131477356,
      "learning_rate": 0.0005586329394654429,
      "loss": 2.2692,
      "step": 38971
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6724147796630859,
      "learning_rate": 0.0005586308666655315,
      "loss": 2.202,
      "step": 38972
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6619961857795715,
      "learning_rate": 0.0005586287938175356,
      "loss": 2.4542,
      "step": 38973
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.7583202123641968,
      "learning_rate": 0.0005586267209214558,
      "loss": 2.3387,
      "step": 38974
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6571915149688721,
      "learning_rate": 0.0005586246479772923,
      "loss": 2.2243,
      "step": 38975
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6777662038803101,
      "learning_rate": 0.0005586225749850457,
      "loss": 2.7481,
      "step": 38976
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6855906248092651,
      "learning_rate": 0.0005586205019447162,
      "loss": 2.3207,
      "step": 38977
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6432825922966003,
      "learning_rate": 0.0005586184288563043,
      "loss": 2.1334,
      "step": 38978
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.7594640851020813,
      "learning_rate": 0.0005586163557198104,
      "loss": 2.1829,
      "step": 38979
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6684479117393494,
      "learning_rate": 0.0005586142825352348,
      "loss": 2.1415,
      "step": 38980
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6055747866630554,
      "learning_rate": 0.000558612209302578,
      "loss": 2.3461,
      "step": 38981
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6855810284614563,
      "learning_rate": 0.0005586101360218401,
      "loss": 2.3604,
      "step": 38982
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.7330384254455566,
      "learning_rate": 0.0005586080626930218,
      "loss": 2.3454,
      "step": 38983
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.795107364654541,
      "learning_rate": 0.0005586059893161235,
      "loss": 2.555,
      "step": 38984
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.683949887752533,
      "learning_rate": 0.0005586039158911454,
      "loss": 2.2878,
      "step": 38985
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.663275957107544,
      "learning_rate": 0.0005586018424180879,
      "loss": 2.3059,
      "step": 38986
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6754855513572693,
      "learning_rate": 0.0005585997688969514,
      "loss": 2.0683,
      "step": 38987
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6723487377166748,
      "learning_rate": 0.0005585976953277364,
      "loss": 2.3784,
      "step": 38988
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.662467360496521,
      "learning_rate": 0.0005585956217104432,
      "loss": 2.2904,
      "step": 38989
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6973745226860046,
      "learning_rate": 0.0005585935480450722,
      "loss": 2.3256,
      "step": 38990
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6858046054840088,
      "learning_rate": 0.0005585914743316238,
      "loss": 2.2149,
      "step": 38991
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6472172141075134,
      "learning_rate": 0.0005585894005700983,
      "loss": 2.2059,
      "step": 38992
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.0107265710830688,
      "learning_rate": 0.0005585873267604962,
      "loss": 2.3271,
      "step": 38993
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6755781769752502,
      "learning_rate": 0.0005585852529028177,
      "loss": 2.3032,
      "step": 38994
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.7160507440567017,
      "learning_rate": 0.0005585831789970636,
      "loss": 2.3361,
      "step": 38995
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6555432677268982,
      "learning_rate": 0.0005585811050432337,
      "loss": 2.4294,
      "step": 38996
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.7447100877761841,
      "learning_rate": 0.0005585790310413289,
      "loss": 2.1965,
      "step": 38997
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6761254072189331,
      "learning_rate": 0.0005585769569913493,
      "loss": 2.1811,
      "step": 38998
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.7002180218696594,
      "learning_rate": 0.0005585748828932953,
      "loss": 2.363,
      "step": 38999
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.7350544929504395,
      "learning_rate": 0.0005585728087471675,
      "loss": 2.3994,
      "step": 39000
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6645104885101318,
      "learning_rate": 0.000558570734552966,
      "loss": 2.2507,
      "step": 39001
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.7360220551490784,
      "learning_rate": 0.0005585686603106914,
      "loss": 2.2104,
      "step": 39002
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.7753613591194153,
      "learning_rate": 0.0005585665860203439,
      "loss": 2.3348,
      "step": 39003
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.7559295892715454,
      "learning_rate": 0.0005585645116819242,
      "loss": 2.1271,
      "step": 39004
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6858935952186584,
      "learning_rate": 0.0005585624372954323,
      "loss": 2.4827,
      "step": 39005
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.7602289319038391,
      "learning_rate": 0.0005585603628608688,
      "loss": 2.4529,
      "step": 39006
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.0468063354492188,
      "learning_rate": 0.000558558288378234,
      "loss": 2.443,
      "step": 39007
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.7317548394203186,
      "learning_rate": 0.0005585562138475284,
      "loss": 2.1534,
      "step": 39008
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.8951811194419861,
      "learning_rate": 0.0005585541392687523,
      "loss": 2.5228,
      "step": 39009
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.0607212781906128,
      "learning_rate": 0.0005585520646419061,
      "loss": 2.151,
      "step": 39010
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.9850965738296509,
      "learning_rate": 0.0005585499899669902,
      "loss": 2.4627,
      "step": 39011
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.8025457859039307,
      "learning_rate": 0.000558547915244005,
      "loss": 2.3224,
      "step": 39012
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.3811347484588623,
      "learning_rate": 0.0005585458404729508,
      "loss": 2.183,
      "step": 39013
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.0669584274291992,
      "learning_rate": 0.0005585437656538281,
      "loss": 2.5204,
      "step": 39014
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.7072175741195679,
      "learning_rate": 0.0005585416907866372,
      "loss": 2.0486,
      "step": 39015
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.115919589996338,
      "learning_rate": 0.0005585396158713786,
      "loss": 2.1676,
      "step": 39016
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.8677635788917542,
      "learning_rate": 0.0005585375409080524,
      "loss": 2.297,
      "step": 39017
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.7437757849693298,
      "learning_rate": 0.0005585354658966594,
      "loss": 2.2064,
      "step": 39018
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.676875114440918,
      "learning_rate": 0.0005585333908371998,
      "loss": 2.2897,
      "step": 39019
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.7912808060646057,
      "learning_rate": 0.0005585313157296738,
      "loss": 2.5712,
      "step": 39020
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.9141845107078552,
      "learning_rate": 0.000558529240574082,
      "loss": 2.1854,
      "step": 39021
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6397297382354736,
      "learning_rate": 0.0005585271653704248,
      "loss": 2.2102,
      "step": 39022
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6886895895004272,
      "learning_rate": 0.0005585250901187025,
      "loss": 2.1143,
      "step": 39023
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6706212759017944,
      "learning_rate": 0.0005585230148189154,
      "loss": 2.3817,
      "step": 39024
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.709352970123291,
      "learning_rate": 0.0005585209394710642,
      "loss": 2.2961,
      "step": 39025
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.677746593952179,
      "learning_rate": 0.0005585188640751489,
      "loss": 2.5729,
      "step": 39026
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6527053117752075,
      "learning_rate": 0.00055851678863117,
      "loss": 2.3353,
      "step": 39027
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.7136927247047424,
      "learning_rate": 0.0005585147131391281,
      "loss": 2.1604,
      "step": 39028
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6415774822235107,
      "learning_rate": 0.0005585126375990234,
      "loss": 2.316,
      "step": 39029
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.7568106055259705,
      "learning_rate": 0.0005585105620108563,
      "loss": 2.1484,
      "step": 39030
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6733812689781189,
      "learning_rate": 0.0005585084863746273,
      "loss": 2.151,
      "step": 39031
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.7408007383346558,
      "learning_rate": 0.0005585064106903366,
      "loss": 2.0705,
      "step": 39032
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.7540163993835449,
      "learning_rate": 0.0005585043349579847,
      "loss": 2.4136,
      "step": 39033
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6364305019378662,
      "learning_rate": 0.000558502259177572,
      "loss": 2.1651,
      "step": 39034
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.8178157806396484,
      "learning_rate": 0.0005585001833490987,
      "loss": 2.313,
      "step": 39035
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6853535771369934,
      "learning_rate": 0.0005584981074725655,
      "loss": 2.0615,
      "step": 39036
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6447882056236267,
      "learning_rate": 0.0005584960315479726,
      "loss": 2.4125,
      "step": 39037
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.7004520297050476,
      "learning_rate": 0.0005584939555753203,
      "loss": 2.1301,
      "step": 39038
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.7165592312812805,
      "learning_rate": 0.0005584918795546092,
      "loss": 2.1665,
      "step": 39039
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.7185097336769104,
      "learning_rate": 0.0005584898034858395,
      "loss": 2.2474,
      "step": 39040
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6221884489059448,
      "learning_rate": 0.0005584877273690118,
      "loss": 2.16,
      "step": 39041
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.7422216534614563,
      "learning_rate": 0.0005584856512041263,
      "loss": 2.1881,
      "step": 39042
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6849607229232788,
      "learning_rate": 0.0005584835749911834,
      "loss": 2.1841,
      "step": 39043
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6735509634017944,
      "learning_rate": 0.0005584814987301834,
      "loss": 2.3367,
      "step": 39044
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6606247425079346,
      "learning_rate": 0.000558479422421127,
      "loss": 2.3476,
      "step": 39045
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.7116726636886597,
      "learning_rate": 0.0005584773460640144,
      "loss": 2.3802,
      "step": 39046
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.7453805804252625,
      "learning_rate": 0.0005584752696588459,
      "loss": 2.3945,
      "step": 39047
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6720770597457886,
      "learning_rate": 0.0005584731932056219,
      "loss": 2.2757,
      "step": 39048
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.7504279017448425,
      "learning_rate": 0.0005584711167043431,
      "loss": 2.2899,
      "step": 39049
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.7165343761444092,
      "learning_rate": 0.0005584690401550093,
      "loss": 2.3617,
      "step": 39050
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6248443722724915,
      "learning_rate": 0.0005584669635576216,
      "loss": 2.1775,
      "step": 39051
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.7052504420280457,
      "learning_rate": 0.0005584648869121797,
      "loss": 2.1359,
      "step": 39052
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6371657848358154,
      "learning_rate": 0.0005584628102186844,
      "loss": 1.9699,
      "step": 39053
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6596759557723999,
      "learning_rate": 0.0005584607334771361,
      "loss": 2.1859,
      "step": 39054
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.689916729927063,
      "learning_rate": 0.0005584586566875349,
      "loss": 2.2372,
      "step": 39055
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6617133617401123,
      "learning_rate": 0.0005584565798498815,
      "loss": 2.2661,
      "step": 39056
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.7197753190994263,
      "learning_rate": 0.0005584545029641761,
      "loss": 2.1684,
      "step": 39057
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6554197072982788,
      "learning_rate": 0.000558452426030419,
      "loss": 2.2744,
      "step": 39058
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.678753137588501,
      "learning_rate": 0.0005584503490486109,
      "loss": 2.4165,
      "step": 39059
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6088939905166626,
      "learning_rate": 0.0005584482720187519,
      "loss": 2.4471,
      "step": 39060
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6554864048957825,
      "learning_rate": 0.0005584461949408425,
      "loss": 2.2308,
      "step": 39061
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6286147236824036,
      "learning_rate": 0.000558444117814883,
      "loss": 2.3208,
      "step": 39062
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6862502694129944,
      "learning_rate": 0.000558442040640874,
      "loss": 2.1673,
      "step": 39063
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6831972599029541,
      "learning_rate": 0.0005584399634188157,
      "loss": 2.3317,
      "step": 39064
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6723240613937378,
      "learning_rate": 0.0005584378861487085,
      "loss": 2.398,
      "step": 39065
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6575225591659546,
      "learning_rate": 0.0005584358088305529,
      "loss": 1.9744,
      "step": 39066
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.7613211274147034,
      "learning_rate": 0.0005584337314643492,
      "loss": 2.4605,
      "step": 39067
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6864929795265198,
      "learning_rate": 0.0005584316540500977,
      "loss": 2.6006,
      "step": 39068
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6284805536270142,
      "learning_rate": 0.0005584295765877989,
      "loss": 2.1153,
      "step": 39069
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6540307402610779,
      "learning_rate": 0.0005584274990774531,
      "loss": 2.1341,
      "step": 39070
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6994385123252869,
      "learning_rate": 0.0005584254215190608,
      "loss": 2.3588,
      "step": 39071
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.7379612922668457,
      "learning_rate": 0.0005584233439126225,
      "loss": 2.3081,
      "step": 39072
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.7077512741088867,
      "learning_rate": 0.0005584212662581382,
      "loss": 2.1229,
      "step": 39073
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.7611021399497986,
      "learning_rate": 0.0005584191885556086,
      "loss": 1.9889,
      "step": 39074
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.67007976770401,
      "learning_rate": 0.0005584171108050341,
      "loss": 2.2404,
      "step": 39075
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6864631175994873,
      "learning_rate": 0.0005584150330064148,
      "loss": 2.2787,
      "step": 39076
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.7121659517288208,
      "learning_rate": 0.0005584129551597514,
      "loss": 2.2942,
      "step": 39077
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.9748547673225403,
      "learning_rate": 0.000558410877265044,
      "loss": 2.2747,
      "step": 39078
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6404505968093872,
      "learning_rate": 0.0005584087993222933,
      "loss": 2.0597,
      "step": 39079
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6456778645515442,
      "learning_rate": 0.0005584067213314994,
      "loss": 2.2881,
      "step": 39080
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6488718390464783,
      "learning_rate": 0.0005584046432926629,
      "loss": 2.0768,
      "step": 39081
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.8169863820075989,
      "learning_rate": 0.000558402565205784,
      "loss": 2.3959,
      "step": 39082
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.7228847742080688,
      "learning_rate": 0.0005584004870708633,
      "loss": 2.1097,
      "step": 39083
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6958941221237183,
      "learning_rate": 0.000558398408887901,
      "loss": 2.3891,
      "step": 39084
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6643959879875183,
      "learning_rate": 0.0005583963306568977,
      "loss": 2.2363,
      "step": 39085
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.7495208978652954,
      "learning_rate": 0.0005583942523778535,
      "loss": 2.314,
      "step": 39086
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6247769594192505,
      "learning_rate": 0.0005583921740507689,
      "loss": 2.1559,
      "step": 39087
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6863092184066772,
      "learning_rate": 0.0005583900956756445,
      "loss": 2.3011,
      "step": 39088
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6724929809570312,
      "learning_rate": 0.0005583880172524803,
      "loss": 2.2625,
      "step": 39089
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.633400559425354,
      "learning_rate": 0.000558385938781277,
      "loss": 2.3374,
      "step": 39090
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.7094230651855469,
      "learning_rate": 0.0005583838602620349,
      "loss": 2.3348,
      "step": 39091
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6661656498908997,
      "learning_rate": 0.0005583817816947542,
      "loss": 2.3657,
      "step": 39092
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.8416235446929932,
      "learning_rate": 0.0005583797030794355,
      "loss": 2.2133,
      "step": 39093
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6226400136947632,
      "learning_rate": 0.0005583776244160793,
      "loss": 2.3998,
      "step": 39094
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6917751431465149,
      "learning_rate": 0.0005583755457046857,
      "loss": 2.3519,
      "step": 39095
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6821883916854858,
      "learning_rate": 0.0005583734669452552,
      "loss": 1.9933,
      "step": 39096
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.7516145706176758,
      "learning_rate": 0.0005583713881377883,
      "loss": 2.2061,
      "step": 39097
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6480256915092468,
      "learning_rate": 0.0005583693092822852,
      "loss": 2.3558,
      "step": 39098
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6642807126045227,
      "learning_rate": 0.0005583672303787463,
      "loss": 2.2719,
      "step": 39099
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6877418756484985,
      "learning_rate": 0.0005583651514271721,
      "loss": 2.4913,
      "step": 39100
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.7570053935050964,
      "learning_rate": 0.0005583630724275631,
      "loss": 2.2477,
      "step": 39101
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.7524172067642212,
      "learning_rate": 0.0005583609933799193,
      "loss": 2.275,
      "step": 39102
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.7170252203941345,
      "learning_rate": 0.0005583589142842414,
      "loss": 2.3162,
      "step": 39103
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.7939436435699463,
      "learning_rate": 0.0005583568351405297,
      "loss": 2.1751,
      "step": 39104
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.8615910410881042,
      "learning_rate": 0.0005583547559487846,
      "loss": 2.3389,
      "step": 39105
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6767799258232117,
      "learning_rate": 0.0005583526767090066,
      "loss": 2.4057,
      "step": 39106
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.8757940530776978,
      "learning_rate": 0.0005583505974211958,
      "loss": 2.4526,
      "step": 39107
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.8252397775650024,
      "learning_rate": 0.0005583485180853527,
      "loss": 2.1459,
      "step": 39108
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6549055576324463,
      "learning_rate": 0.0005583464387014779,
      "loss": 2.4741,
      "step": 39109
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.7276586890220642,
      "learning_rate": 0.0005583443592695715,
      "loss": 2.3028,
      "step": 39110
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.1498842239379883,
      "learning_rate": 0.000558342279789634,
      "loss": 2.2635,
      "step": 39111
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6538006663322449,
      "learning_rate": 0.0005583402002616658,
      "loss": 2.1531,
      "step": 39112
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.8912661075592041,
      "learning_rate": 0.0005583381206856673,
      "loss": 2.3806,
      "step": 39113
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.8789196014404297,
      "learning_rate": 0.0005583360410616389,
      "loss": 2.4169,
      "step": 39114
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6463189721107483,
      "learning_rate": 0.000558333961389581,
      "loss": 2.37,
      "step": 39115
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.8515452742576599,
      "learning_rate": 0.0005583318816694939,
      "loss": 2.3075,
      "step": 39116
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.7383289933204651,
      "learning_rate": 0.000558329801901378,
      "loss": 2.2626,
      "step": 39117
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6307544112205505,
      "learning_rate": 0.0005583277220852337,
      "loss": 2.3582,
      "step": 39118
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.7056729197502136,
      "learning_rate": 0.0005583256422210614,
      "loss": 2.1066,
      "step": 39119
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.8384297490119934,
      "learning_rate": 0.0005583235623088615,
      "loss": 2.3549,
      "step": 39120
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.66677325963974,
      "learning_rate": 0.0005583214823486344,
      "loss": 2.0285,
      "step": 39121
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.7581608891487122,
      "learning_rate": 0.0005583194023403805,
      "loss": 2.3061,
      "step": 39122
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6882111430168152,
      "learning_rate": 0.0005583173222841,
      "loss": 2.3175,
      "step": 39123
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6782733201980591,
      "learning_rate": 0.0005583152421797936,
      "loss": 2.258,
      "step": 39124
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.9920295476913452,
      "learning_rate": 0.0005583131620274614,
      "loss": 2.2962,
      "step": 39125
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6431775093078613,
      "learning_rate": 0.0005583110818271039,
      "loss": 2.0388,
      "step": 39126
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.7355175614356995,
      "learning_rate": 0.0005583090015787215,
      "loss": 2.3427,
      "step": 39127
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6680994033813477,
      "learning_rate": 0.0005583069212823146,
      "loss": 2.2398,
      "step": 39128
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.7859325408935547,
      "learning_rate": 0.0005583048409378836,
      "loss": 2.356,
      "step": 39129
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6569604873657227,
      "learning_rate": 0.0005583027605454289,
      "loss": 2.0755,
      "step": 39130
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.7494506239891052,
      "learning_rate": 0.0005583006801049507,
      "loss": 2.2322,
      "step": 39131
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6894451975822449,
      "learning_rate": 0.0005582985996164497,
      "loss": 2.2998,
      "step": 39132
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.7062436938285828,
      "learning_rate": 0.0005582965190799261,
      "loss": 2.2885,
      "step": 39133
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.7298474311828613,
      "learning_rate": 0.0005582944384953801,
      "loss": 2.27,
      "step": 39134
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.7400541305541992,
      "learning_rate": 0.0005582923578628125,
      "loss": 1.9922,
      "step": 39135
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6191822290420532,
      "learning_rate": 0.0005582902771822234,
      "loss": 2.2448,
      "step": 39136
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6771389245986938,
      "learning_rate": 0.0005582881964536131,
      "loss": 2.3575,
      "step": 39137
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6678776144981384,
      "learning_rate": 0.0005582861156769825,
      "loss": 2.3157,
      "step": 39138
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.7058044672012329,
      "learning_rate": 0.0005582840348523313,
      "loss": 2.258,
      "step": 39139
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.7944058775901794,
      "learning_rate": 0.0005582819539796603,
      "loss": 2.5525,
      "step": 39140
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.641567051410675,
      "learning_rate": 0.0005582798730589699,
      "loss": 2.3726,
      "step": 39141
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6640930771827698,
      "learning_rate": 0.0005582777920902605,
      "loss": 2.2941,
      "step": 39142
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6670256853103638,
      "learning_rate": 0.0005582757110735322,
      "loss": 2.3444,
      "step": 39143
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6347119212150574,
      "learning_rate": 0.0005582736300087857,
      "loss": 2.5232,
      "step": 39144
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.7940190434455872,
      "learning_rate": 0.0005582715488960211,
      "loss": 2.1971,
      "step": 39145
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6326985359191895,
      "learning_rate": 0.000558269467735239,
      "loss": 2.4453,
      "step": 39146
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.0039091110229492,
      "learning_rate": 0.0005582673865264397,
      "loss": 2.226,
      "step": 39147
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.7073647379875183,
      "learning_rate": 0.0005582653052696237,
      "loss": 2.325,
      "step": 39148
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6466993689537048,
      "learning_rate": 0.0005582632239647913,
      "loss": 2.1681,
      "step": 39149
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.7091610431671143,
      "learning_rate": 0.0005582611426119428,
      "loss": 2.3845,
      "step": 39150
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6515629887580872,
      "learning_rate": 0.0005582590612110787,
      "loss": 2.3158,
      "step": 39151
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.8488723039627075,
      "learning_rate": 0.0005582569797621995,
      "loss": 2.3202,
      "step": 39152
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6789752840995789,
      "learning_rate": 0.0005582548982653053,
      "loss": 2.2851,
      "step": 39153
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6357629299163818,
      "learning_rate": 0.0005582528167203968,
      "loss": 2.1083,
      "step": 39154
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.7989513874053955,
      "learning_rate": 0.000558250735127474,
      "loss": 2.1413,
      "step": 39155
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.665656328201294,
      "learning_rate": 0.0005582486534865378,
      "loss": 2.2453,
      "step": 39156
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.7833971977233887,
      "learning_rate": 0.000558246571797588,
      "loss": 2.2817,
      "step": 39157
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.764798104763031,
      "learning_rate": 0.0005582444900606255,
      "loss": 2.2128,
      "step": 39158
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6586429476737976,
      "learning_rate": 0.0005582424082756504,
      "loss": 2.3673,
      "step": 39159
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6924682855606079,
      "learning_rate": 0.0005582403264426632,
      "loss": 2.2862,
      "step": 39160
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6535101532936096,
      "learning_rate": 0.0005582382445616642,
      "loss": 2.3291,
      "step": 39161
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.7388415932655334,
      "learning_rate": 0.0005582361626326538,
      "loss": 2.3018,
      "step": 39162
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.667338490486145,
      "learning_rate": 0.0005582340806556325,
      "loss": 2.3201,
      "step": 39163
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.9670575261116028,
      "learning_rate": 0.0005582319986306006,
      "loss": 2.4118,
      "step": 39164
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6707223057746887,
      "learning_rate": 0.0005582299165575585,
      "loss": 2.1759,
      "step": 39165
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.7752508521080017,
      "learning_rate": 0.0005582278344365064,
      "loss": 2.5461,
      "step": 39166
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.7068593502044678,
      "learning_rate": 0.0005582257522674451,
      "loss": 2.15,
      "step": 39167
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.657048761844635,
      "learning_rate": 0.0005582236700503747,
      "loss": 2.4506,
      "step": 39168
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6538491249084473,
      "learning_rate": 0.0005582215877852957,
      "loss": 2.1365,
      "step": 39169
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6096116304397583,
      "learning_rate": 0.0005582195054722084,
      "loss": 2.2119,
      "step": 39170
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6822035908699036,
      "learning_rate": 0.000558217423111113,
      "loss": 2.1972,
      "step": 39171
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6442293524742126,
      "learning_rate": 0.0005582153407020104,
      "loss": 2.1963,
      "step": 39172
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6900222897529602,
      "learning_rate": 0.0005582132582449006,
      "loss": 2.1728,
      "step": 39173
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6629997491836548,
      "learning_rate": 0.000558211175739784,
      "loss": 2.4526,
      "step": 39174
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.7083736062049866,
      "learning_rate": 0.0005582090931866612,
      "loss": 2.3669,
      "step": 39175
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.8560879826545715,
      "learning_rate": 0.0005582070105855324,
      "loss": 2.3384,
      "step": 39176
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.7759683728218079,
      "learning_rate": 0.000558204927936398,
      "loss": 2.3963,
      "step": 39177
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.8031644821166992,
      "learning_rate": 0.0005582028452392585,
      "loss": 2.4049,
      "step": 39178
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.7486039996147156,
      "learning_rate": 0.0005582007624941141,
      "loss": 2.3689,
      "step": 39179
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.7379202842712402,
      "learning_rate": 0.0005581986797009654,
      "loss": 2.4512,
      "step": 39180
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6492629051208496,
      "learning_rate": 0.0005581965968598127,
      "loss": 2.2563,
      "step": 39181
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.7218391299247742,
      "learning_rate": 0.0005581945139706564,
      "loss": 2.249,
      "step": 39182
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6387468576431274,
      "learning_rate": 0.0005581924310334969,
      "loss": 1.9403,
      "step": 39183
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6599516272544861,
      "learning_rate": 0.0005581903480483344,
      "loss": 1.9846,
      "step": 39184
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.7000138163566589,
      "learning_rate": 0.0005581882650151695,
      "loss": 2.3612,
      "step": 39185
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.7828881144523621,
      "learning_rate": 0.0005581861819340026,
      "loss": 2.2989,
      "step": 39186
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6350340843200684,
      "learning_rate": 0.000558184098804834,
      "loss": 2.5037,
      "step": 39187
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6964656710624695,
      "learning_rate": 0.000558182015627664,
      "loss": 1.9967,
      "step": 39188
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.7003605365753174,
      "learning_rate": 0.0005581799324024931,
      "loss": 2.4297,
      "step": 39189
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.7247943878173828,
      "learning_rate": 0.0005581778491293218,
      "loss": 2.3707,
      "step": 39190
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6048872470855713,
      "learning_rate": 0.0005581757658081502,
      "loss": 2.3584,
      "step": 39191
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.65907883644104,
      "learning_rate": 0.000558173682438979,
      "loss": 2.3124,
      "step": 39192
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6840347647666931,
      "learning_rate": 0.0005581715990218084,
      "loss": 2.3146,
      "step": 39193
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.7743973731994629,
      "learning_rate": 0.0005581695155566388,
      "loss": 2.297,
      "step": 39194
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6868780851364136,
      "learning_rate": 0.0005581674320434707,
      "loss": 2.1649,
      "step": 39195
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.7479152083396912,
      "learning_rate": 0.0005581653484823043,
      "loss": 2.1619,
      "step": 39196
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6199613809585571,
      "learning_rate": 0.0005581632648731401,
      "loss": 2.1487,
      "step": 39197
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6715490221977234,
      "learning_rate": 0.0005581611812159785,
      "loss": 2.199,
      "step": 39198
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.669389009475708,
      "learning_rate": 0.0005581590975108199,
      "loss": 2.3537,
      "step": 39199
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6953641176223755,
      "learning_rate": 0.0005581570137576645,
      "loss": 2.3021,
      "step": 39200
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6453751921653748,
      "learning_rate": 0.000558154929956513,
      "loss": 2.4509,
      "step": 39201
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6807607412338257,
      "learning_rate": 0.0005581528461073655,
      "loss": 2.2391,
      "step": 39202
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6096529960632324,
      "learning_rate": 0.0005581507622102227,
      "loss": 2.3464,
      "step": 39203
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6731548309326172,
      "learning_rate": 0.0005581486782650846,
      "loss": 2.2516,
      "step": 39204
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.7950406074523926,
      "learning_rate": 0.0005581465942719519,
      "loss": 2.2218,
      "step": 39205
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.7220920324325562,
      "learning_rate": 0.0005581445102308248,
      "loss": 2.5645,
      "step": 39206
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6753000020980835,
      "learning_rate": 0.0005581424261417038,
      "loss": 2.2899,
      "step": 39207
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.8218724727630615,
      "learning_rate": 0.0005581403420045893,
      "loss": 2.1505,
      "step": 39208
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.7019622921943665,
      "learning_rate": 0.0005581382578194816,
      "loss": 2.4935,
      "step": 39209
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6787266135215759,
      "learning_rate": 0.0005581361735863811,
      "loss": 2.2204,
      "step": 39210
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6415203213691711,
      "learning_rate": 0.0005581340893052883,
      "loss": 2.0874,
      "step": 39211
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.7609876990318298,
      "learning_rate": 0.0005581320049762034,
      "loss": 2.3368,
      "step": 39212
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.648389458656311,
      "learning_rate": 0.0005581299205991269,
      "loss": 2.2217,
      "step": 39213
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6131319403648376,
      "learning_rate": 0.0005581278361740591,
      "loss": 2.182,
      "step": 39214
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.8673056364059448,
      "learning_rate": 0.0005581257517010006,
      "loss": 2.1327,
      "step": 39215
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.657346785068512,
      "learning_rate": 0.0005581236671799516,
      "loss": 2.1149,
      "step": 39216
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6773852109909058,
      "learning_rate": 0.0005581215826109126,
      "loss": 2.2541,
      "step": 39217
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6895866990089417,
      "learning_rate": 0.0005581194979938839,
      "loss": 2.2918,
      "step": 39218
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.7263248562812805,
      "learning_rate": 0.0005581174133288658,
      "loss": 2.366,
      "step": 39219
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6979341506958008,
      "learning_rate": 0.0005581153286158589,
      "loss": 2.3963,
      "step": 39220
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6806773543357849,
      "learning_rate": 0.0005581132438548634,
      "loss": 2.1805,
      "step": 39221
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.7000793218612671,
      "learning_rate": 0.0005581111590458799,
      "loss": 1.9917,
      "step": 39222
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6912707090377808,
      "learning_rate": 0.0005581090741889087,
      "loss": 2.2141,
      "step": 39223
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6313791275024414,
      "learning_rate": 0.0005581069892839501,
      "loss": 1.9778,
      "step": 39224
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6568761467933655,
      "learning_rate": 0.0005581049043310044,
      "loss": 2.1787,
      "step": 39225
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.7709694504737854,
      "learning_rate": 0.0005581028193300722,
      "loss": 2.1144,
      "step": 39226
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6410406231880188,
      "learning_rate": 0.000558100734281154,
      "loss": 2.5002,
      "step": 39227
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.70118647813797,
      "learning_rate": 0.0005580986491842497,
      "loss": 2.3633,
      "step": 39228
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.7239158749580383,
      "learning_rate": 0.0005580965640393601,
      "loss": 2.3487,
      "step": 39229
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6172539591789246,
      "learning_rate": 0.0005580944788464855,
      "loss": 2.276,
      "step": 39230
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6370904445648193,
      "learning_rate": 0.0005580923936056263,
      "loss": 2.3168,
      "step": 39231
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6182091236114502,
      "learning_rate": 0.0005580903083167828,
      "loss": 2.2854,
      "step": 39232
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6871834397315979,
      "learning_rate": 0.0005580882229799556,
      "loss": 2.5903,
      "step": 39233
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.713516116142273,
      "learning_rate": 0.0005580861375951448,
      "loss": 2.1564,
      "step": 39234
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6683725714683533,
      "learning_rate": 0.0005580840521623509,
      "loss": 2.243,
      "step": 39235
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6522251963615417,
      "learning_rate": 0.0005580819666815743,
      "loss": 2.5551,
      "step": 39236
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.8144567012786865,
      "learning_rate": 0.0005580798811528154,
      "loss": 2.3289,
      "step": 39237
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6308128833770752,
      "learning_rate": 0.0005580777955760745,
      "loss": 2.1354,
      "step": 39238
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.741862952709198,
      "learning_rate": 0.0005580757099513522,
      "loss": 2.1766,
      "step": 39239
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6173581480979919,
      "learning_rate": 0.0005580736242786487,
      "loss": 2.3415,
      "step": 39240
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6763243079185486,
      "learning_rate": 0.0005580715385579644,
      "loss": 2.1691,
      "step": 39241
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6736131906509399,
      "learning_rate": 0.0005580694527892998,
      "loss": 2.2818,
      "step": 39242
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.7194678783416748,
      "learning_rate": 0.0005580673669726553,
      "loss": 2.2914,
      "step": 39243
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6597394347190857,
      "learning_rate": 0.000558065281108031,
      "loss": 2.1982,
      "step": 39244
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.639279842376709,
      "learning_rate": 0.0005580631951954277,
      "loss": 2.1482,
      "step": 39245
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6474337577819824,
      "learning_rate": 0.0005580611092348453,
      "loss": 2.1708,
      "step": 39246
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.7733057141304016,
      "learning_rate": 0.0005580590232262847,
      "loss": 2.298,
      "step": 39247
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.7934513092041016,
      "learning_rate": 0.0005580569371697459,
      "loss": 2.3301,
      "step": 39248
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6582211852073669,
      "learning_rate": 0.0005580548510652297,
      "loss": 2.2431,
      "step": 39249
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.661465048789978,
      "learning_rate": 0.000558052764912736,
      "loss": 2.2608,
      "step": 39250
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.7321262955665588,
      "learning_rate": 0.0005580506787122655,
      "loss": 2.2917,
      "step": 39251
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6675732135772705,
      "learning_rate": 0.0005580485924638185,
      "loss": 2.3156,
      "step": 39252
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.7248867154121399,
      "learning_rate": 0.0005580465061673953,
      "loss": 2.5328,
      "step": 39253
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.7101352214813232,
      "learning_rate": 0.0005580444198229965,
      "loss": 2.2906,
      "step": 39254
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6604380011558533,
      "learning_rate": 0.0005580423334306223,
      "loss": 2.3358,
      "step": 39255
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6727027893066406,
      "learning_rate": 0.0005580402469902733,
      "loss": 2.305,
      "step": 39256
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6417905688285828,
      "learning_rate": 0.0005580381605019496,
      "loss": 2.1243,
      "step": 39257
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6567795276641846,
      "learning_rate": 0.0005580360739656519,
      "loss": 2.1437,
      "step": 39258
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6926352977752686,
      "learning_rate": 0.0005580339873813802,
      "loss": 2.2892,
      "step": 39259
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6613802313804626,
      "learning_rate": 0.0005580319007491353,
      "loss": 2.1513,
      "step": 39260
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.7558916211128235,
      "learning_rate": 0.0005580298140689173,
      "loss": 2.1685,
      "step": 39261
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6764965653419495,
      "learning_rate": 0.0005580277273407267,
      "loss": 2.1221,
      "step": 39262
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6714943647384644,
      "learning_rate": 0.0005580256405645638,
      "loss": 2.2286,
      "step": 39263
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6692304015159607,
      "learning_rate": 0.0005580235537404292,
      "loss": 2.2312,
      "step": 39264
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.7626955509185791,
      "learning_rate": 0.0005580214668683231,
      "loss": 2.5044,
      "step": 39265
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6518080234527588,
      "learning_rate": 0.0005580193799482459,
      "loss": 2.0515,
      "step": 39266
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6590995788574219,
      "learning_rate": 0.0005580172929801981,
      "loss": 2.0333,
      "step": 39267
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6085749268531799,
      "learning_rate": 0.00055801520596418,
      "loss": 2.1933,
      "step": 39268
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.7543085813522339,
      "learning_rate": 0.000558013118900192,
      "loss": 2.3881,
      "step": 39269
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6723199486732483,
      "learning_rate": 0.0005580110317882344,
      "loss": 2.4376,
      "step": 39270
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.7442053556442261,
      "learning_rate": 0.0005580089446283078,
      "loss": 2.1266,
      "step": 39271
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6494528651237488,
      "learning_rate": 0.0005580068574204124,
      "loss": 2.2302,
      "step": 39272
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6735026240348816,
      "learning_rate": 0.0005580047701645487,
      "loss": 2.3061,
      "step": 39273
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.7327182292938232,
      "learning_rate": 0.000558002682860717,
      "loss": 2.5347,
      "step": 39274
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6622711420059204,
      "learning_rate": 0.0005580005955089178,
      "loss": 2.6106,
      "step": 39275
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6713538765907288,
      "learning_rate": 0.0005579985081091513,
      "loss": 2.2208,
      "step": 39276
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.7086421847343445,
      "learning_rate": 0.0005579964206614181,
      "loss": 2.3021,
      "step": 39277
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.7333781719207764,
      "learning_rate": 0.0005579943331657185,
      "loss": 2.4567,
      "step": 39278
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6448187232017517,
      "learning_rate": 0.0005579922456220529,
      "loss": 2.0564,
      "step": 39279
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6492561101913452,
      "learning_rate": 0.0005579901580304215,
      "loss": 2.1542,
      "step": 39280
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.7117918133735657,
      "learning_rate": 0.000557988070390825,
      "loss": 2.2325,
      "step": 39281
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6887861490249634,
      "learning_rate": 0.0005579859827032637,
      "loss": 2.5111,
      "step": 39282
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6693130135536194,
      "learning_rate": 0.0005579838949677379,
      "loss": 2.2138,
      "step": 39283
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6347677707672119,
      "learning_rate": 0.0005579818071842479,
      "loss": 2.4233,
      "step": 39284
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.7151471972465515,
      "learning_rate": 0.0005579797193527944,
      "loss": 2.219,
      "step": 39285
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.7013528347015381,
      "learning_rate": 0.0005579776314733775,
      "loss": 2.3662,
      "step": 39286
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6289030909538269,
      "learning_rate": 0.0005579755435459975,
      "loss": 2.2793,
      "step": 39287
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6509454846382141,
      "learning_rate": 0.0005579734555706552,
      "loss": 2.3536,
      "step": 39288
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.7040560841560364,
      "learning_rate": 0.0005579713675473508,
      "loss": 2.4986,
      "step": 39289
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.7880474328994751,
      "learning_rate": 0.0005579692794760846,
      "loss": 2.3142,
      "step": 39290
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.7194850444793701,
      "learning_rate": 0.000557967191356857,
      "loss": 2.045,
      "step": 39291
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6730425357818604,
      "learning_rate": 0.0005579651031896685,
      "loss": 2.3657,
      "step": 39292
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6644591689109802,
      "learning_rate": 0.0005579630149745193,
      "loss": 2.0406,
      "step": 39293
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6616653203964233,
      "learning_rate": 0.0005579609267114101,
      "loss": 2.0442,
      "step": 39294
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6429470777511597,
      "learning_rate": 0.0005579588384003409,
      "loss": 2.2276,
      "step": 39295
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.7070764899253845,
      "learning_rate": 0.0005579567500413124,
      "loss": 2.1742,
      "step": 39296
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.7882255911827087,
      "learning_rate": 0.0005579546616343248,
      "loss": 2.516,
      "step": 39297
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.7235275506973267,
      "learning_rate": 0.0005579525731793786,
      "loss": 2.3945,
      "step": 39298
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.7652301788330078,
      "learning_rate": 0.0005579504846764742,
      "loss": 2.2539,
      "step": 39299
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.7608026266098022,
      "learning_rate": 0.0005579483961256118,
      "loss": 2.1002,
      "step": 39300
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.7294735908508301,
      "learning_rate": 0.000557946307526792,
      "loss": 2.3002,
      "step": 39301
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6641372442245483,
      "learning_rate": 0.0005579442188800151,
      "loss": 2.3276,
      "step": 39302
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.777241587638855,
      "learning_rate": 0.0005579421301852815,
      "loss": 2.2262,
      "step": 39303
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.7676024436950684,
      "learning_rate": 0.0005579400414425916,
      "loss": 2.3974,
      "step": 39304
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6915013790130615,
      "learning_rate": 0.0005579379526519457,
      "loss": 2.208,
      "step": 39305
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6471375226974487,
      "learning_rate": 0.0005579358638133444,
      "loss": 2.3662,
      "step": 39306
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6820258498191833,
      "learning_rate": 0.0005579337749267879,
      "loss": 2.1972,
      "step": 39307
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.7036057710647583,
      "learning_rate": 0.0005579316859922767,
      "loss": 2.1191,
      "step": 39308
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.8341397643089294,
      "learning_rate": 0.000557929597009811,
      "loss": 2.1891,
      "step": 39309
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.7524347305297852,
      "learning_rate": 0.0005579275079793914,
      "loss": 2.2939,
      "step": 39310
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6901466250419617,
      "learning_rate": 0.0005579254189010183,
      "loss": 2.1887,
      "step": 39311
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.7139888405799866,
      "learning_rate": 0.0005579233297746918,
      "loss": 2.1856,
      "step": 39312
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6682708263397217,
      "learning_rate": 0.0005579212406004126,
      "loss": 2.4789,
      "step": 39313
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6790138483047485,
      "learning_rate": 0.0005579191513781809,
      "loss": 2.1852,
      "step": 39314
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6878737211227417,
      "learning_rate": 0.0005579170621079971,
      "loss": 2.1329,
      "step": 39315
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.7164804339408875,
      "learning_rate": 0.0005579149727898619,
      "loss": 2.4033,
      "step": 39316
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.731363832950592,
      "learning_rate": 0.0005579128834237752,
      "loss": 2.3412,
      "step": 39317
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.7173728346824646,
      "learning_rate": 0.0005579107940097378,
      "loss": 2.1956,
      "step": 39318
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6653929948806763,
      "learning_rate": 0.0005579087045477499,
      "loss": 2.3457,
      "step": 39319
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.7795987129211426,
      "learning_rate": 0.0005579066150378117,
      "loss": 2.2732,
      "step": 39320
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6952000856399536,
      "learning_rate": 0.000557904525479924,
      "loss": 2.4958,
      "step": 39321
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6671530604362488,
      "learning_rate": 0.0005579024358740869,
      "loss": 2.4403,
      "step": 39322
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.7592647671699524,
      "learning_rate": 0.0005579003462203009,
      "loss": 2.429,
      "step": 39323
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.7018429636955261,
      "learning_rate": 0.0005578982565185663,
      "loss": 2.1604,
      "step": 39324
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.7110336422920227,
      "learning_rate": 0.0005578961667688836,
      "loss": 2.3719,
      "step": 39325
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.686399519443512,
      "learning_rate": 0.0005578940769712531,
      "loss": 2.1284,
      "step": 39326
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.7334420084953308,
      "learning_rate": 0.0005578919871256753,
      "loss": 2.1949,
      "step": 39327
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6820403933525085,
      "learning_rate": 0.0005578898972321504,
      "loss": 1.9725,
      "step": 39328
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6703087091445923,
      "learning_rate": 0.0005578878072906789,
      "loss": 2.4153,
      "step": 39329
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.868112325668335,
      "learning_rate": 0.0005578857173012614,
      "loss": 2.258,
      "step": 39330
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.7623746991157532,
      "learning_rate": 0.0005578836272638978,
      "loss": 2.4506,
      "step": 39331
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6639470458030701,
      "learning_rate": 0.0005578815371785889,
      "loss": 2.237,
      "step": 39332
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.7066469192504883,
      "learning_rate": 0.0005578794470453349,
      "loss": 2.3581,
      "step": 39333
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6784598231315613,
      "learning_rate": 0.0005578773568641365,
      "loss": 2.1495,
      "step": 39334
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.7195086479187012,
      "learning_rate": 0.0005578752666349935,
      "loss": 2.4739,
      "step": 39335
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6528233289718628,
      "learning_rate": 0.0005578731763579066,
      "loss": 2.2058,
      "step": 39336
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.7538480162620544,
      "learning_rate": 0.0005578710860328764,
      "loss": 2.375,
      "step": 39337
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.657730758190155,
      "learning_rate": 0.0005578689956599031,
      "loss": 2.1518,
      "step": 39338
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6505822539329529,
      "learning_rate": 0.000557866905238987,
      "loss": 2.29,
      "step": 39339
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.7123932838439941,
      "learning_rate": 0.0005578648147701287,
      "loss": 2.6237,
      "step": 39340
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.7164159417152405,
      "learning_rate": 0.0005578627242533283,
      "loss": 2.4362,
      "step": 39341
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.7047423720359802,
      "learning_rate": 0.0005578606336885864,
      "loss": 2.3229,
      "step": 39342
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6764883399009705,
      "learning_rate": 0.0005578585430759033,
      "loss": 2.4637,
      "step": 39343
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6745099425315857,
      "learning_rate": 0.0005578564524152795,
      "loss": 2.0827,
      "step": 39344
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.7671043872833252,
      "learning_rate": 0.0005578543617067153,
      "loss": 2.1981,
      "step": 39345
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.643829882144928,
      "learning_rate": 0.0005578522709502112,
      "loss": 2.1782,
      "step": 39346
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.642877995967865,
      "learning_rate": 0.0005578501801457674,
      "loss": 2.11,
      "step": 39347
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6724568009376526,
      "learning_rate": 0.0005578480892933844,
      "loss": 2.3386,
      "step": 39348
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.675275444984436,
      "learning_rate": 0.0005578459983930625,
      "loss": 2.1919,
      "step": 39349
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.7261727452278137,
      "learning_rate": 0.0005578439074448023,
      "loss": 2.3626,
      "step": 39350
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.7288211584091187,
      "learning_rate": 0.0005578418164486039,
      "loss": 2.3059,
      "step": 39351
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6428430080413818,
      "learning_rate": 0.0005578397254044679,
      "loss": 2.4806,
      "step": 39352
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6829562783241272,
      "learning_rate": 0.0005578376343123946,
      "loss": 2.3094,
      "step": 39353
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.676272988319397,
      "learning_rate": 0.0005578355431723846,
      "loss": 2.4701,
      "step": 39354
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6980093121528625,
      "learning_rate": 0.0005578334519844379,
      "loss": 2.3046,
      "step": 39355
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.7605375051498413,
      "learning_rate": 0.0005578313607485552,
      "loss": 2.2118,
      "step": 39356
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.7445191144943237,
      "learning_rate": 0.0005578292694647367,
      "loss": 2.2107,
      "step": 39357
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.8041889667510986,
      "learning_rate": 0.0005578271781329829,
      "loss": 2.3926,
      "step": 39358
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.7474885582923889,
      "learning_rate": 0.0005578250867532943,
      "loss": 2.2819,
      "step": 39359
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.7283506393432617,
      "learning_rate": 0.0005578229953256709,
      "loss": 2.4345,
      "step": 39360
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6780716776847839,
      "learning_rate": 0.0005578209038501135,
      "loss": 2.2882,
      "step": 39361
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6814529299736023,
      "learning_rate": 0.0005578188123266223,
      "loss": 2.4041,
      "step": 39362
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6790963411331177,
      "learning_rate": 0.0005578167207551976,
      "loss": 2.1682,
      "step": 39363
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6765947341918945,
      "learning_rate": 0.0005578146291358401,
      "loss": 2.2682,
      "step": 39364
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.7258523106575012,
      "learning_rate": 0.0005578125374685499,
      "loss": 2.5275,
      "step": 39365
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.7316235303878784,
      "learning_rate": 0.0005578104457533275,
      "loss": 2.2377,
      "step": 39366
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6807077527046204,
      "learning_rate": 0.0005578083539901732,
      "loss": 2.1764,
      "step": 39367
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6667801141738892,
      "learning_rate": 0.0005578062621790876,
      "loss": 2.2386,
      "step": 39368
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.7477734088897705,
      "learning_rate": 0.0005578041703200708,
      "loss": 2.4408,
      "step": 39369
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.7053999900817871,
      "learning_rate": 0.0005578020784131235,
      "loss": 2.2819,
      "step": 39370
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.654834508895874,
      "learning_rate": 0.0005577999864582458,
      "loss": 2.3785,
      "step": 39371
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.7328099012374878,
      "learning_rate": 0.0005577978944554383,
      "loss": 2.5109,
      "step": 39372
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.7198640704154968,
      "learning_rate": 0.0005577958024047012,
      "loss": 2.3434,
      "step": 39373
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.7315211296081543,
      "learning_rate": 0.0005577937103060351,
      "loss": 2.0161,
      "step": 39374
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6166415810585022,
      "learning_rate": 0.0005577916181594402,
      "loss": 2.385,
      "step": 39375
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.647796094417572,
      "learning_rate": 0.0005577895259649171,
      "loss": 2.0766,
      "step": 39376
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6577857136726379,
      "learning_rate": 0.000557787433722466,
      "loss": 2.2084,
      "step": 39377
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.7136090397834778,
      "learning_rate": 0.0005577853414320872,
      "loss": 2.3912,
      "step": 39378
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.7188982367515564,
      "learning_rate": 0.0005577832490937814,
      "loss": 2.5166,
      "step": 39379
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.8675153255462646,
      "learning_rate": 0.0005577811567075488,
      "loss": 2.2713,
      "step": 39380
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.703376829624176,
      "learning_rate": 0.0005577790642733898,
      "loss": 2.3845,
      "step": 39381
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6327173113822937,
      "learning_rate": 0.0005577769717913047,
      "loss": 2.3718,
      "step": 39382
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.7361733317375183,
      "learning_rate": 0.0005577748792612941,
      "loss": 2.599,
      "step": 39383
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6822105646133423,
      "learning_rate": 0.0005577727866833583,
      "loss": 2.3793,
      "step": 39384
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6977435946464539,
      "learning_rate": 0.0005577706940574976,
      "loss": 2.3048,
      "step": 39385
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.8474568128585815,
      "learning_rate": 0.0005577686013837126,
      "loss": 2.0224,
      "step": 39386
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6721578240394592,
      "learning_rate": 0.0005577665086620035,
      "loss": 2.1519,
      "step": 39387
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6500333547592163,
      "learning_rate": 0.0005577644158923708,
      "loss": 2.2859,
      "step": 39388
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.8042227625846863,
      "learning_rate": 0.0005577623230748147,
      "loss": 2.2773,
      "step": 39389
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.7242050766944885,
      "learning_rate": 0.0005577602302093357,
      "loss": 2.1694,
      "step": 39390
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6946156024932861,
      "learning_rate": 0.0005577581372959343,
      "loss": 2.0879,
      "step": 39391
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.670700192451477,
      "learning_rate": 0.0005577560443346107,
      "loss": 2.1904,
      "step": 39392
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6822032928466797,
      "learning_rate": 0.0005577539513253654,
      "loss": 2.2868,
      "step": 39393
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6847561597824097,
      "learning_rate": 0.0005577518582681989,
      "loss": 2.4904,
      "step": 39394
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.690196692943573,
      "learning_rate": 0.0005577497651631114,
      "loss": 2.0603,
      "step": 39395
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.683268129825592,
      "learning_rate": 0.0005577476720101033,
      "loss": 2.1746,
      "step": 39396
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.642707884311676,
      "learning_rate": 0.0005577455788091751,
      "loss": 2.3108,
      "step": 39397
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6372348666191101,
      "learning_rate": 0.0005577434855603271,
      "loss": 2.4144,
      "step": 39398
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.7318875193595886,
      "learning_rate": 0.0005577413922635598,
      "loss": 2.2993,
      "step": 39399
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6287277936935425,
      "learning_rate": 0.0005577392989188735,
      "loss": 2.3182,
      "step": 39400
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.7012573480606079,
      "learning_rate": 0.0005577372055262684,
      "loss": 2.1917,
      "step": 39401
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6568717360496521,
      "learning_rate": 0.0005577351120857453,
      "loss": 2.3781,
      "step": 39402
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6806849837303162,
      "learning_rate": 0.0005577330185973043,
      "loss": 2.1854,
      "step": 39403
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.7116484045982361,
      "learning_rate": 0.0005577309250609458,
      "loss": 2.2773,
      "step": 39404
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6317102909088135,
      "learning_rate": 0.0005577288314766705,
      "loss": 2.067,
      "step": 39405
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6986637711524963,
      "learning_rate": 0.0005577267378444783,
      "loss": 2.5023,
      "step": 39406
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.8030158877372742,
      "learning_rate": 0.00055772464416437,
      "loss": 2.319,
      "step": 39407
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.7593262195587158,
      "learning_rate": 0.0005577225504363457,
      "loss": 2.1262,
      "step": 39408
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.661506175994873,
      "learning_rate": 0.0005577204566604059,
      "loss": 2.2558,
      "step": 39409
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.8911968469619751,
      "learning_rate": 0.0005577183628365511,
      "loss": 2.3359,
      "step": 39410
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.7810278534889221,
      "learning_rate": 0.0005577162689647815,
      "loss": 2.1073,
      "step": 39411
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.708756685256958,
      "learning_rate": 0.0005577141750450976,
      "loss": 2.2432,
      "step": 39412
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.7679393887519836,
      "learning_rate": 0.0005577120810774998,
      "loss": 1.9437,
      "step": 39413
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.7076098322868347,
      "learning_rate": 0.0005577099870619886,
      "loss": 2.2116,
      "step": 39414
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.7064935564994812,
      "learning_rate": 0.000557707892998564,
      "loss": 2.239,
      "step": 39415
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.7133185863494873,
      "learning_rate": 0.0005577057988872268,
      "loss": 2.1834,
      "step": 39416
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.9799114465713501,
      "learning_rate": 0.0005577037047279771,
      "loss": 2.4039,
      "step": 39417
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6803395748138428,
      "learning_rate": 0.0005577016105208155,
      "loss": 2.5494,
      "step": 39418
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.920920729637146,
      "learning_rate": 0.0005576995162657423,
      "loss": 1.9854,
      "step": 39419
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.7494502067565918,
      "learning_rate": 0.0005576974219627578,
      "loss": 2.1282,
      "step": 39420
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.665165364742279,
      "learning_rate": 0.0005576953276118626,
      "loss": 2.1622,
      "step": 39421
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.7427276968955994,
      "learning_rate": 0.000557693233213057,
      "loss": 2.333,
      "step": 39422
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.9021129608154297,
      "learning_rate": 0.0005576911387663413,
      "loss": 2.4111,
      "step": 39423
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6727831363677979,
      "learning_rate": 0.0005576890442717161,
      "loss": 2.3997,
      "step": 39424
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.634586751461029,
      "learning_rate": 0.0005576869497291814,
      "loss": 2.305,
      "step": 39425
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.7163503766059875,
      "learning_rate": 0.000557684855138738,
      "loss": 2.2769,
      "step": 39426
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6890376210212708,
      "learning_rate": 0.000557682760500386,
      "loss": 2.2022,
      "step": 39427
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6319466233253479,
      "learning_rate": 0.0005576806658141259,
      "loss": 2.0769,
      "step": 39428
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6936174631118774,
      "learning_rate": 0.0005576785710799582,
      "loss": 2.0491,
      "step": 39429
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6315758228302002,
      "learning_rate": 0.0005576764762978832,
      "loss": 2.2567,
      "step": 39430
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6307544112205505,
      "learning_rate": 0.0005576743814679013,
      "loss": 2.3217,
      "step": 39431
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6366649270057678,
      "learning_rate": 0.0005576722865900127,
      "loss": 2.2632,
      "step": 39432
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.7305315136909485,
      "learning_rate": 0.0005576701916642182,
      "loss": 2.2231,
      "step": 39433
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6971582174301147,
      "learning_rate": 0.0005576680966905178,
      "loss": 2.0823,
      "step": 39434
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.649949312210083,
      "learning_rate": 0.000557666001668912,
      "loss": 2.3972,
      "step": 39435
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6614997982978821,
      "learning_rate": 0.0005576639065994013,
      "loss": 2.3898,
      "step": 39436
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.7364609241485596,
      "learning_rate": 0.000557661811481986,
      "loss": 2.2111,
      "step": 39437
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6211559772491455,
      "learning_rate": 0.0005576597163166664,
      "loss": 2.3146,
      "step": 39438
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.7393221259117126,
      "learning_rate": 0.0005576576211034431,
      "loss": 2.3169,
      "step": 39439
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.7045652866363525,
      "learning_rate": 0.0005576555258423164,
      "loss": 2.2941,
      "step": 39440
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6550592184066772,
      "learning_rate": 0.0005576534305332866,
      "loss": 2.2048,
      "step": 39441
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6881512403488159,
      "learning_rate": 0.0005576513351763543,
      "loss": 2.1375,
      "step": 39442
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.7106161713600159,
      "learning_rate": 0.0005576492397715196,
      "loss": 2.0853,
      "step": 39443
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.7117427587509155,
      "learning_rate": 0.0005576471443187831,
      "loss": 2.4314,
      "step": 39444
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6349583864212036,
      "learning_rate": 0.0005576450488181452,
      "loss": 2.1626,
      "step": 39445
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.7090372443199158,
      "learning_rate": 0.0005576429532696062,
      "loss": 2.1728,
      "step": 39446
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6809964179992676,
      "learning_rate": 0.0005576408576731664,
      "loss": 2.2981,
      "step": 39447
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.7551742196083069,
      "learning_rate": 0.0005576387620288265,
      "loss": 2.3461,
      "step": 39448
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6427809000015259,
      "learning_rate": 0.0005576366663365865,
      "loss": 2.2164,
      "step": 39449
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6835008263587952,
      "learning_rate": 0.000557634570596447,
      "loss": 2.4268,
      "step": 39450
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6355711221694946,
      "learning_rate": 0.0005576324748084085,
      "loss": 2.3735,
      "step": 39451
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.7036811113357544,
      "learning_rate": 0.0005576303789724712,
      "loss": 2.4405,
      "step": 39452
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6720027923583984,
      "learning_rate": 0.0005576282830886354,
      "loss": 2.3528,
      "step": 39453
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6977025866508484,
      "learning_rate": 0.0005576261871569018,
      "loss": 2.2357,
      "step": 39454
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.661323606967926,
      "learning_rate": 0.0005576240911772706,
      "loss": 2.0384,
      "step": 39455
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.7063615322113037,
      "learning_rate": 0.0005576219951497423,
      "loss": 2.1992,
      "step": 39456
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.8271217942237854,
      "learning_rate": 0.000557619899074317,
      "loss": 2.4423,
      "step": 39457
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6591618061065674,
      "learning_rate": 0.0005576178029509955,
      "loss": 2.3199,
      "step": 39458
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6455486416816711,
      "learning_rate": 0.000557615706779778,
      "loss": 2.3427,
      "step": 39459
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6628019213676453,
      "learning_rate": 0.0005576136105606647,
      "loss": 2.1045,
      "step": 39460
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.7634662389755249,
      "learning_rate": 0.0005576115142936562,
      "loss": 2.2503,
      "step": 39461
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.7322859764099121,
      "learning_rate": 0.000557609417978753,
      "loss": 2.281,
      "step": 39462
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.7290686964988708,
      "learning_rate": 0.0005576073216159552,
      "loss": 2.3129,
      "step": 39463
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6550069451332092,
      "learning_rate": 0.0005576052252052635,
      "loss": 2.2686,
      "step": 39464
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6417675018310547,
      "learning_rate": 0.000557603128746678,
      "loss": 2.4259,
      "step": 39465
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.7358752489089966,
      "learning_rate": 0.0005576010322401991,
      "loss": 2.2933,
      "step": 39466
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6842812299728394,
      "learning_rate": 0.0005575989356858275,
      "loss": 2.3752,
      "step": 39467
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6917148232460022,
      "learning_rate": 0.0005575968390835633,
      "loss": 2.3995,
      "step": 39468
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6628757119178772,
      "learning_rate": 0.0005575947424334071,
      "loss": 2.1768,
      "step": 39469
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6981054544448853,
      "learning_rate": 0.000557592645735359,
      "loss": 2.1432,
      "step": 39470
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.7238413691520691,
      "learning_rate": 0.0005575905489894196,
      "loss": 2.3902,
      "step": 39471
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6933252811431885,
      "learning_rate": 0.0005575884521955894,
      "loss": 2.4303,
      "step": 39472
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6824935078620911,
      "learning_rate": 0.0005575863553538686,
      "loss": 2.3311,
      "step": 39473
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.5987774729728699,
      "learning_rate": 0.0005575842584642576,
      "loss": 2.1552,
      "step": 39474
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.8662638068199158,
      "learning_rate": 0.0005575821615267569,
      "loss": 2.3223,
      "step": 39475
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6516382098197937,
      "learning_rate": 0.0005575800645413666,
      "loss": 2.3178,
      "step": 39476
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.7484164834022522,
      "learning_rate": 0.0005575779675080875,
      "loss": 2.4554,
      "step": 39477
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6853044033050537,
      "learning_rate": 0.0005575758704269197,
      "loss": 2.2332,
      "step": 39478
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.7076470255851746,
      "learning_rate": 0.0005575737732978637,
      "loss": 2.1202,
      "step": 39479
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6555594205856323,
      "learning_rate": 0.00055757167612092,
      "loss": 2.4608,
      "step": 39480
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6899303197860718,
      "learning_rate": 0.0005575695788960887,
      "loss": 2.4574,
      "step": 39481
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.7105581164360046,
      "learning_rate": 0.0005575674816233704,
      "loss": 2.4397,
      "step": 39482
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6471251845359802,
      "learning_rate": 0.0005575653843027654,
      "loss": 2.0771,
      "step": 39483
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.7949692606925964,
      "learning_rate": 0.0005575632869342742,
      "loss": 2.12,
      "step": 39484
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6837689280509949,
      "learning_rate": 0.0005575611895178971,
      "loss": 2.2032,
      "step": 39485
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.7100957036018372,
      "learning_rate": 0.0005575590920536346,
      "loss": 2.3273,
      "step": 39486
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.8179112672805786,
      "learning_rate": 0.0005575569945414869,
      "loss": 2.3088,
      "step": 39487
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.672989010810852,
      "learning_rate": 0.0005575548969814543,
      "loss": 2.5094,
      "step": 39488
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.714104175567627,
      "learning_rate": 0.0005575527993735376,
      "loss": 2.0313,
      "step": 39489
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6971514821052551,
      "learning_rate": 0.0005575507017177371,
      "loss": 2.4646,
      "step": 39490
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.7186192870140076,
      "learning_rate": 0.0005575486040140529,
      "loss": 2.1009,
      "step": 39491
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.912036120891571,
      "learning_rate": 0.0005575465062624855,
      "loss": 2.0671,
      "step": 39492
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6904293894767761,
      "learning_rate": 0.0005575444084630355,
      "loss": 2.3743,
      "step": 39493
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.7201026678085327,
      "learning_rate": 0.000557542310615703,
      "loss": 2.2985,
      "step": 39494
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.7421960830688477,
      "learning_rate": 0.0005575402127204886,
      "loss": 2.418,
      "step": 39495
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6573203802108765,
      "learning_rate": 0.0005575381147773926,
      "loss": 2.3529,
      "step": 39496
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6908204555511475,
      "learning_rate": 0.0005575360167864154,
      "loss": 2.1277,
      "step": 39497
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6082168221473694,
      "learning_rate": 0.0005575339187475574,
      "loss": 2.0489,
      "step": 39498
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.7849243879318237,
      "learning_rate": 0.000557531820660819,
      "loss": 2.2682,
      "step": 39499
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6576635241508484,
      "learning_rate": 0.0005575297225262005,
      "loss": 2.2547,
      "step": 39500
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.7555839419364929,
      "learning_rate": 0.0005575276243437024,
      "loss": 2.221,
      "step": 39501
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6694751977920532,
      "learning_rate": 0.0005575255261133251,
      "loss": 2.1448,
      "step": 39502
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.8826578855514526,
      "learning_rate": 0.000557523427835069,
      "loss": 2.294,
      "step": 39503
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6646015048027039,
      "learning_rate": 0.0005575213295089342,
      "loss": 2.3753,
      "step": 39504
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6752191781997681,
      "learning_rate": 0.0005575192311349216,
      "loss": 2.2919,
      "step": 39505
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6881930828094482,
      "learning_rate": 0.0005575171327130311,
      "loss": 2.429,
      "step": 39506
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6951220035552979,
      "learning_rate": 0.0005575150342432634,
      "loss": 2.3234,
      "step": 39507
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.7019882202148438,
      "learning_rate": 0.0005575129357256188,
      "loss": 2.3452,
      "step": 39508
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.7510757446289062,
      "learning_rate": 0.0005575108371600976,
      "loss": 2.4727,
      "step": 39509
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.8075811266899109,
      "learning_rate": 0.0005575087385467003,
      "loss": 2.3612,
      "step": 39510
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.7243099212646484,
      "learning_rate": 0.0005575066398854273,
      "loss": 2.2719,
      "step": 39511
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6731258034706116,
      "learning_rate": 0.000557504541176279,
      "loss": 2.3069,
      "step": 39512
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.9783525466918945,
      "learning_rate": 0.0005575024424192557,
      "loss": 2.2962,
      "step": 39513
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.7279084324836731,
      "learning_rate": 0.0005575003436143578,
      "loss": 2.2248,
      "step": 39514
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.7259491086006165,
      "learning_rate": 0.0005574982447615858,
      "loss": 2.1474,
      "step": 39515
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6652047038078308,
      "learning_rate": 0.00055749614586094,
      "loss": 2.1744,
      "step": 39516
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6807411909103394,
      "learning_rate": 0.0005574940469124208,
      "loss": 2.1278,
      "step": 39517
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6697371006011963,
      "learning_rate": 0.0005574919479160286,
      "loss": 2.1587,
      "step": 39518
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6412414312362671,
      "learning_rate": 0.0005574898488717637,
      "loss": 2.27,
      "step": 39519
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.7314954400062561,
      "learning_rate": 0.0005574877497796267,
      "loss": 2.1598,
      "step": 39520
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6464900374412537,
      "learning_rate": 0.0005574856506396177,
      "loss": 2.2561,
      "step": 39521
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.652328610420227,
      "learning_rate": 0.0005574835514517375,
      "loss": 2.2691,
      "step": 39522
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.7081809043884277,
      "learning_rate": 0.0005574814522159861,
      "loss": 2.1653,
      "step": 39523
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.7106841206550598,
      "learning_rate": 0.0005574793529323639,
      "loss": 2.3022,
      "step": 39524
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6450153589248657,
      "learning_rate": 0.0005574772536008718,
      "loss": 2.1832,
      "step": 39525
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6852353811264038,
      "learning_rate": 0.0005574751542215095,
      "loss": 2.414,
      "step": 39526
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.7095264792442322,
      "learning_rate": 0.0005574730547942778,
      "loss": 2.5082,
      "step": 39527
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.668832004070282,
      "learning_rate": 0.000557470955319177,
      "loss": 2.2492,
      "step": 39528
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6698973774909973,
      "learning_rate": 0.0005574688557962074,
      "loss": 2.1405,
      "step": 39529
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6935722827911377,
      "learning_rate": 0.0005574667562253697,
      "loss": 2.1941,
      "step": 39530
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.642378032207489,
      "learning_rate": 0.0005574646566066638,
      "loss": 2.2059,
      "step": 39531
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.70035719871521,
      "learning_rate": 0.0005574625569400906,
      "loss": 2.1897,
      "step": 39532
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6978331804275513,
      "learning_rate": 0.0005574604572256501,
      "loss": 2.2418,
      "step": 39533
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6547723412513733,
      "learning_rate": 0.0005574583574633429,
      "loss": 2.4465,
      "step": 39534
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6835533976554871,
      "learning_rate": 0.0005574562576531694,
      "loss": 2.4242,
      "step": 39535
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.7019733190536499,
      "learning_rate": 0.0005574541577951298,
      "loss": 2.1556,
      "step": 39536
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6734395027160645,
      "learning_rate": 0.0005574520578892247,
      "loss": 2.4954,
      "step": 39537
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.7015010118484497,
      "learning_rate": 0.0005574499579354543,
      "loss": 2.4026,
      "step": 39538
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.7391200661659241,
      "learning_rate": 0.0005574478579338192,
      "loss": 2.1445,
      "step": 39539
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6850032806396484,
      "learning_rate": 0.0005574457578843196,
      "loss": 2.1921,
      "step": 39540
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.619982898235321,
      "learning_rate": 0.0005574436577869561,
      "loss": 2.57,
      "step": 39541
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.8322488069534302,
      "learning_rate": 0.0005574415576417288,
      "loss": 2.2481,
      "step": 39542
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6869107484817505,
      "learning_rate": 0.0005574394574486384,
      "loss": 2.3878,
      "step": 39543
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6740704774856567,
      "learning_rate": 0.000557437357207685,
      "loss": 2.053,
      "step": 39544
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.7465034127235413,
      "learning_rate": 0.0005574352569188692,
      "loss": 2.2077,
      "step": 39545
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.8274627923965454,
      "learning_rate": 0.0005574331565821914,
      "loss": 2.2908,
      "step": 39546
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6895930767059326,
      "learning_rate": 0.0005574310561976518,
      "loss": 2.056,
      "step": 39547
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6822492480278015,
      "learning_rate": 0.000557428955765251,
      "loss": 2.4233,
      "step": 39548
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.908334493637085,
      "learning_rate": 0.0005574268552849893,
      "loss": 2.1931,
      "step": 39549
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.9358207583427429,
      "learning_rate": 0.000557424754756867,
      "loss": 2.2693,
      "step": 39550
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.02354896068573,
      "learning_rate": 0.0005574226541808846,
      "loss": 2.2326,
      "step": 39551
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.8895465731620789,
      "learning_rate": 0.0005574205535570425,
      "loss": 2.0908,
      "step": 39552
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6735683679580688,
      "learning_rate": 0.000557418452885341,
      "loss": 2.1995,
      "step": 39553
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.7225437164306641,
      "learning_rate": 0.0005574163521657807,
      "loss": 2.1921,
      "step": 39554
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6906633377075195,
      "learning_rate": 0.0005574142513983618,
      "loss": 2.4061,
      "step": 39555
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.7973791360855103,
      "learning_rate": 0.0005574121505830846,
      "loss": 2.253,
      "step": 39556
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6921986937522888,
      "learning_rate": 0.0005574100497199498,
      "loss": 2.2442,
      "step": 39557
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.7635995745658875,
      "learning_rate": 0.0005574079488089574,
      "loss": 2.3569,
      "step": 39558
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.736157238483429,
      "learning_rate": 0.0005574058478501082,
      "loss": 2.1606,
      "step": 39559
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.7520092725753784,
      "learning_rate": 0.0005574037468434023,
      "loss": 2.1535,
      "step": 39560
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6606640219688416,
      "learning_rate": 0.0005574016457888403,
      "loss": 2.3406,
      "step": 39561
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.8085278272628784,
      "learning_rate": 0.0005573995446864222,
      "loss": 2.3172,
      "step": 39562
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.7985082864761353,
      "learning_rate": 0.000557397443536149,
      "loss": 2.1318,
      "step": 39563
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.0103436708450317,
      "learning_rate": 0.0005573953423380206,
      "loss": 2.2134,
      "step": 39564
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6694188714027405,
      "learning_rate": 0.0005573932410920375,
      "loss": 2.1505,
      "step": 39565
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.7541879415512085,
      "learning_rate": 0.0005573911397982001,
      "loss": 2.2969,
      "step": 39566
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.7375331521034241,
      "learning_rate": 0.000557389038456509,
      "loss": 2.246,
      "step": 39567
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6747769713401794,
      "learning_rate": 0.0005573869370669643,
      "loss": 2.4372,
      "step": 39568
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.7250047922134399,
      "learning_rate": 0.0005573848356295665,
      "loss": 2.3009,
      "step": 39569
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.7433093190193176,
      "learning_rate": 0.000557382734144316,
      "loss": 2.5896,
      "step": 39570
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.7551977038383484,
      "learning_rate": 0.0005573806326112133,
      "loss": 2.3028,
      "step": 39571
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6977797150611877,
      "learning_rate": 0.0005573785310302586,
      "loss": 2.2799,
      "step": 39572
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6447157263755798,
      "learning_rate": 0.0005573764294014524,
      "loss": 2.3788,
      "step": 39573
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.7243600487709045,
      "learning_rate": 0.000557374327724795,
      "loss": 2.4469,
      "step": 39574
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.7762601375579834,
      "learning_rate": 0.0005573722260002869,
      "loss": 2.3003,
      "step": 39575
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.9485530257225037,
      "learning_rate": 0.0005573701242279285,
      "loss": 2.2183,
      "step": 39576
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6643067598342896,
      "learning_rate": 0.00055736802240772,
      "loss": 2.448,
      "step": 39577
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.7916122078895569,
      "learning_rate": 0.0005573659205396621,
      "loss": 2.3553,
      "step": 39578
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.864108681678772,
      "learning_rate": 0.0005573638186237549,
      "loss": 2.4059,
      "step": 39579
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.660731852054596,
      "learning_rate": 0.000557361716659999,
      "loss": 2.245,
      "step": 39580
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6832422018051147,
      "learning_rate": 0.0005573596146483946,
      "loss": 2.3696,
      "step": 39581
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.7488052845001221,
      "learning_rate": 0.0005573575125889422,
      "loss": 2.2449,
      "step": 39582
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6733046174049377,
      "learning_rate": 0.0005573554104816422,
      "loss": 2.2802,
      "step": 39583
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6827792525291443,
      "learning_rate": 0.000557353308326495,
      "loss": 2.2499,
      "step": 39584
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.7246673107147217,
      "learning_rate": 0.0005573512061235008,
      "loss": 2.2296,
      "step": 39585
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.7217071652412415,
      "learning_rate": 0.0005573491038726604,
      "loss": 2.1992,
      "step": 39586
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6543413996696472,
      "learning_rate": 0.0005573470015739738,
      "loss": 2.4685,
      "step": 39587
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.622123658657074,
      "learning_rate": 0.0005573448992274416,
      "loss": 2.1537,
      "step": 39588
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6296374201774597,
      "learning_rate": 0.0005573427968330641,
      "loss": 2.1886,
      "step": 39589
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6665666103363037,
      "learning_rate": 0.0005573406943908417,
      "loss": 2.4514,
      "step": 39590
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6813581585884094,
      "learning_rate": 0.0005573385919007748,
      "loss": 2.1455,
      "step": 39591
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6252139210700989,
      "learning_rate": 0.0005573364893628638,
      "loss": 2.504,
      "step": 39592
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.722620964050293,
      "learning_rate": 0.0005573343867771091,
      "loss": 2.201,
      "step": 39593
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.7145314812660217,
      "learning_rate": 0.000557332284143511,
      "loss": 2.3291,
      "step": 39594
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6425639986991882,
      "learning_rate": 0.0005573301814620702,
      "loss": 2.2055,
      "step": 39595
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.622047483921051,
      "learning_rate": 0.0005573280787327867,
      "loss": 2.3258,
      "step": 39596
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6635404229164124,
      "learning_rate": 0.000557325975955661,
      "loss": 2.1051,
      "step": 39597
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6878830790519714,
      "learning_rate": 0.0005573238731306936,
      "loss": 2.0013,
      "step": 39598
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6922977566719055,
      "learning_rate": 0.000557321770257885,
      "loss": 2.4939,
      "step": 39599
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.7243798971176147,
      "learning_rate": 0.0005573196673372352,
      "loss": 2.2467,
      "step": 39600
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6837687492370605,
      "learning_rate": 0.0005573175643687449,
      "loss": 2.3116,
      "step": 39601
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6495379209518433,
      "learning_rate": 0.0005573154613524143,
      "loss": 2.3418,
      "step": 39602
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6949172616004944,
      "learning_rate": 0.000557313358288244,
      "loss": 2.0908,
      "step": 39603
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.7234641909599304,
      "learning_rate": 0.0005573112551762343,
      "loss": 2.334,
      "step": 39604
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6486110091209412,
      "learning_rate": 0.0005573091520163857,
      "loss": 2.3704,
      "step": 39605
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.7709421515464783,
      "learning_rate": 0.0005573070488086982,
      "loss": 2.2712,
      "step": 39606
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.745087742805481,
      "learning_rate": 0.0005573049455531727,
      "loss": 2.1148,
      "step": 39607
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.7187830805778503,
      "learning_rate": 0.0005573028422498093,
      "loss": 2.3223,
      "step": 39608
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.7014771103858948,
      "learning_rate": 0.0005573007388986083,
      "loss": 2.2325,
      "step": 39609
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6653192639350891,
      "learning_rate": 0.0005572986354995703,
      "loss": 2.2844,
      "step": 39610
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.7033281326293945,
      "learning_rate": 0.0005572965320526957,
      "loss": 2.3781,
      "step": 39611
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.7652438879013062,
      "learning_rate": 0.0005572944285579847,
      "loss": 2.1291,
      "step": 39612
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.7988574504852295,
      "learning_rate": 0.000557292325015438,
      "loss": 2.3198,
      "step": 39613
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6369831562042236,
      "learning_rate": 0.0005572902214250557,
      "loss": 2.6069,
      "step": 39614
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6561675071716309,
      "learning_rate": 0.0005572881177868382,
      "loss": 2.3641,
      "step": 39615
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.7663372755050659,
      "learning_rate": 0.0005572860141007861,
      "loss": 2.0449,
      "step": 39616
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.7945507168769836,
      "learning_rate": 0.0005572839103668996,
      "loss": 2.326,
      "step": 39617
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.7913299202919006,
      "learning_rate": 0.0005572818065851793,
      "loss": 2.0859,
      "step": 39618
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.7119308114051819,
      "learning_rate": 0.0005572797027556253,
      "loss": 2.3205,
      "step": 39619
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6383460164070129,
      "learning_rate": 0.0005572775988782381,
      "loss": 2.4424,
      "step": 39620
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.687130331993103,
      "learning_rate": 0.0005572754949530182,
      "loss": 2.2976,
      "step": 39621
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6635661125183105,
      "learning_rate": 0.0005572733909799661,
      "loss": 2.4143,
      "step": 39622
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.7538244128227234,
      "learning_rate": 0.0005572712869590818,
      "loss": 2.3955,
      "step": 39623
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6594605445861816,
      "learning_rate": 0.0005572691828903661,
      "loss": 2.329,
      "step": 39624
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6255273222923279,
      "learning_rate": 0.0005572670787738191,
      "loss": 2.3969,
      "step": 39625
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6810084581375122,
      "learning_rate": 0.0005572649746094414,
      "loss": 2.1007,
      "step": 39626
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.7206432819366455,
      "learning_rate": 0.0005572628703972331,
      "loss": 2.4328,
      "step": 39627
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.7100096940994263,
      "learning_rate": 0.0005572607661371948,
      "loss": 2.3999,
      "step": 39628
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.7105889916419983,
      "learning_rate": 0.000557258661829327,
      "loss": 2.2756,
      "step": 39629
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.7222037315368652,
      "learning_rate": 0.0005572565574736299,
      "loss": 2.3147,
      "step": 39630
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.8193293213844299,
      "learning_rate": 0.0005572544530701039,
      "loss": 2.0929,
      "step": 39631
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.7310396432876587,
      "learning_rate": 0.0005572523486187494,
      "loss": 2.2848,
      "step": 39632
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.8176039457321167,
      "learning_rate": 0.0005572502441195669,
      "loss": 2.5036,
      "step": 39633
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.8575143218040466,
      "learning_rate": 0.0005572481395725567,
      "loss": 2.2756,
      "step": 39634
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6491712331771851,
      "learning_rate": 0.0005572460349777192,
      "loss": 2.4505,
      "step": 39635
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.7567831873893738,
      "learning_rate": 0.000557243930335055,
      "loss": 2.2877,
      "step": 39636
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.7264404296875,
      "learning_rate": 0.0005572418256445641,
      "loss": 2.3688,
      "step": 39637
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.645036518573761,
      "learning_rate": 0.0005572397209062472,
      "loss": 2.4253,
      "step": 39638
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.7581160664558411,
      "learning_rate": 0.0005572376161201045,
      "loss": 2.3583,
      "step": 39639
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.7200393080711365,
      "learning_rate": 0.0005572355112861365,
      "loss": 2.3776,
      "step": 39640
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6776188611984253,
      "learning_rate": 0.0005572334064043434,
      "loss": 2.4162,
      "step": 39641
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6424809098243713,
      "learning_rate": 0.0005572313014747261,
      "loss": 2.278,
      "step": 39642
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.7277021408081055,
      "learning_rate": 0.0005572291964972844,
      "loss": 2.1636,
      "step": 39643
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.782838761806488,
      "learning_rate": 0.000557227091472019,
      "loss": 2.2339,
      "step": 39644
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6937515139579773,
      "learning_rate": 0.0005572249863989301,
      "loss": 2.1753,
      "step": 39645
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.8168734908103943,
      "learning_rate": 0.0005572228812780185,
      "loss": 2.3343,
      "step": 39646
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.7706378102302551,
      "learning_rate": 0.000557220776109284,
      "loss": 2.2775,
      "step": 39647
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.7128843069076538,
      "learning_rate": 0.0005572186708927275,
      "loss": 2.5029,
      "step": 39648
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6501462459564209,
      "learning_rate": 0.0005572165656283492,
      "loss": 2.4049,
      "step": 39649
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.780637264251709,
      "learning_rate": 0.0005572144603161493,
      "loss": 2.2868,
      "step": 39650
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.704138457775116,
      "learning_rate": 0.0005572123549561286,
      "loss": 2.1716,
      "step": 39651
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6213682293891907,
      "learning_rate": 0.0005572102495482871,
      "loss": 2.3608,
      "step": 39652
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.8130653500556946,
      "learning_rate": 0.0005572081440926255,
      "loss": 2.3883,
      "step": 39653
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6599932909011841,
      "learning_rate": 0.0005572060385891439,
      "loss": 1.9701,
      "step": 39654
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.7441160082817078,
      "learning_rate": 0.000557203933037843,
      "loss": 2.4478,
      "step": 39655
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6689562797546387,
      "learning_rate": 0.0005572018274387229,
      "loss": 2.467,
      "step": 39656
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6717811226844788,
      "learning_rate": 0.0005571997217917842,
      "loss": 2.3509,
      "step": 39657
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.0253950357437134,
      "learning_rate": 0.0005571976160970273,
      "loss": 2.2041,
      "step": 39658
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.7187959551811218,
      "learning_rate": 0.0005571955103544523,
      "loss": 2.0453,
      "step": 39659
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.7687831521034241,
      "learning_rate": 0.00055719340456406,
      "loss": 2.3369,
      "step": 39660
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.8115823268890381,
      "learning_rate": 0.0005571912987258504,
      "loss": 2.343,
      "step": 39661
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.687995195388794,
      "learning_rate": 0.0005571891928398242,
      "loss": 2.489,
      "step": 39662
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6782007217407227,
      "learning_rate": 0.0005571870869059817,
      "loss": 2.2688,
      "step": 39663
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.7641661763191223,
      "learning_rate": 0.0005571849809243232,
      "loss": 2.5434,
      "step": 39664
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6512486338615417,
      "learning_rate": 0.0005571828748948492,
      "loss": 2.288,
      "step": 39665
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6860416531562805,
      "learning_rate": 0.0005571807688175601,
      "loss": 2.2478,
      "step": 39666
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6814953684806824,
      "learning_rate": 0.0005571786626924561,
      "loss": 2.2066,
      "step": 39667
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.7636734247207642,
      "learning_rate": 0.0005571765565195377,
      "loss": 2.5021,
      "step": 39668
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6619327068328857,
      "learning_rate": 0.0005571744502988055,
      "loss": 2.2103,
      "step": 39669
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.7380762100219727,
      "learning_rate": 0.0005571723440302597,
      "loss": 2.3274,
      "step": 39670
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.698997974395752,
      "learning_rate": 0.0005571702377139006,
      "loss": 2.5385,
      "step": 39671
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.7175939083099365,
      "learning_rate": 0.0005571681313497288,
      "loss": 2.2549,
      "step": 39672
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6541736721992493,
      "learning_rate": 0.0005571660249377445,
      "loss": 2.4341,
      "step": 39673
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6567292809486389,
      "learning_rate": 0.0005571639184779482,
      "loss": 2.1605,
      "step": 39674
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6765546798706055,
      "learning_rate": 0.0005571618119703403,
      "loss": 2.18,
      "step": 39675
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.8926012516021729,
      "learning_rate": 0.0005571597054149211,
      "loss": 2.2255,
      "step": 39676
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6906937956809998,
      "learning_rate": 0.0005571575988116912,
      "loss": 2.3496,
      "step": 39677
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.639767050743103,
      "learning_rate": 0.0005571554921606507,
      "loss": 2.0715,
      "step": 39678
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.8002676963806152,
      "learning_rate": 0.0005571533854618002,
      "loss": 2.2146,
      "step": 39679
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.8539811372756958,
      "learning_rate": 0.00055715127871514,
      "loss": 2.1053,
      "step": 39680
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6936371326446533,
      "learning_rate": 0.0005571491719206706,
      "loss": 2.097,
      "step": 39681
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.7669573426246643,
      "learning_rate": 0.0005571470650783922,
      "loss": 2.2289,
      "step": 39682
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.7730388045310974,
      "learning_rate": 0.0005571449581883054,
      "loss": 2.0333,
      "step": 39683
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6652560830116272,
      "learning_rate": 0.0005571428512504105,
      "loss": 2.3406,
      "step": 39684
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.82962566614151,
      "learning_rate": 0.0005571407442647078,
      "loss": 2.3553,
      "step": 39685
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.7278316020965576,
      "learning_rate": 0.0005571386372311979,
      "loss": 2.4234,
      "step": 39686
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6755643486976624,
      "learning_rate": 0.0005571365301498811,
      "loss": 2.3815,
      "step": 39687
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6730893850326538,
      "learning_rate": 0.0005571344230207576,
      "loss": 2.2683,
      "step": 39688
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6404981017112732,
      "learning_rate": 0.0005571323158438281,
      "loss": 2.1647,
      "step": 39689
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.773973822593689,
      "learning_rate": 0.0005571302086190928,
      "loss": 1.9971,
      "step": 39690
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.7600671052932739,
      "learning_rate": 0.0005571281013465521,
      "loss": 2.2879,
      "step": 39691
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6267024874687195,
      "learning_rate": 0.0005571259940262065,
      "loss": 2.063,
      "step": 39692
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.7986043095588684,
      "learning_rate": 0.0005571238866580562,
      "loss": 2.1218,
      "step": 39693
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.7175190448760986,
      "learning_rate": 0.0005571217792421018,
      "loss": 2.3448,
      "step": 39694
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6498540043830872,
      "learning_rate": 0.0005571196717783436,
      "loss": 2.2642,
      "step": 39695
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.7285140156745911,
      "learning_rate": 0.000557117564266782,
      "loss": 2.4425,
      "step": 39696
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.7948716282844543,
      "learning_rate": 0.0005571154567074175,
      "loss": 2.3296,
      "step": 39697
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.7835065722465515,
      "learning_rate": 0.0005571133491002503,
      "loss": 2.3233,
      "step": 39698
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.649797797203064,
      "learning_rate": 0.0005571112414452808,
      "loss": 2.336,
      "step": 39699
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.723580002784729,
      "learning_rate": 0.0005571091337425095,
      "loss": 2.0362,
      "step": 39700
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6404179930686951,
      "learning_rate": 0.0005571070259919368,
      "loss": 2.2084,
      "step": 39701
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.767659604549408,
      "learning_rate": 0.0005571049181935631,
      "loss": 2.2015,
      "step": 39702
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6326661109924316,
      "learning_rate": 0.0005571028103473887,
      "loss": 2.1163,
      "step": 39703
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6697148680686951,
      "learning_rate": 0.000557100702453414,
      "loss": 2.2708,
      "step": 39704
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.7488128542900085,
      "learning_rate": 0.0005570985945116394,
      "loss": 2.3222,
      "step": 39705
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6873449087142944,
      "learning_rate": 0.0005570964865220652,
      "loss": 2.4424,
      "step": 39706
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6610639691352844,
      "learning_rate": 0.0005570943784846922,
      "loss": 2.0801,
      "step": 39707
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6995626091957092,
      "learning_rate": 0.0005570922703995203,
      "loss": 2.3037,
      "step": 39708
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.7528946995735168,
      "learning_rate": 0.0005570901622665502,
      "loss": 2.2128,
      "step": 39709
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6846575736999512,
      "learning_rate": 0.0005570880540857821,
      "loss": 2.2713,
      "step": 39710
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.680458664894104,
      "learning_rate": 0.0005570859458572166,
      "loss": 2.3108,
      "step": 39711
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6901174187660217,
      "learning_rate": 0.000557083837580854,
      "loss": 2.4084,
      "step": 39712
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.7260013818740845,
      "learning_rate": 0.0005570817292566944,
      "loss": 2.5533,
      "step": 39713
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6502994894981384,
      "learning_rate": 0.0005570796208847386,
      "loss": 2.0074,
      "step": 39714
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6698377728462219,
      "learning_rate": 0.0005570775124649869,
      "loss": 2.281,
      "step": 39715
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.7228807210922241,
      "learning_rate": 0.0005570754039974396,
      "loss": 2.2506,
      "step": 39716
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.8616564273834229,
      "learning_rate": 0.0005570732954820971,
      "loss": 2.3178,
      "step": 39717
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6376227140426636,
      "learning_rate": 0.0005570711869189598,
      "loss": 2.4139,
      "step": 39718
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.7434744238853455,
      "learning_rate": 0.0005570690783080282,
      "loss": 2.4096,
      "step": 39719
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.7420892715454102,
      "learning_rate": 0.0005570669696493025,
      "loss": 2.2377,
      "step": 39720
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6734316349029541,
      "learning_rate": 0.0005570648609427833,
      "loss": 2.3421,
      "step": 39721
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6859153509140015,
      "learning_rate": 0.0005570627521884709,
      "loss": 2.412,
      "step": 39722
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.7431240677833557,
      "learning_rate": 0.0005570606433863656,
      "loss": 2.439,
      "step": 39723
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6711957454681396,
      "learning_rate": 0.000557058534536468,
      "loss": 2.3323,
      "step": 39724
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6391695737838745,
      "learning_rate": 0.0005570564256387782,
      "loss": 2.2017,
      "step": 39725
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.7132248878479004,
      "learning_rate": 0.0005570543166932969,
      "loss": 2.2475,
      "step": 39726
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6333386301994324,
      "learning_rate": 0.0005570522077000242,
      "loss": 2.2618,
      "step": 39727
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6321162581443787,
      "learning_rate": 0.0005570500986589608,
      "loss": 2.1568,
      "step": 39728
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6608678698539734,
      "learning_rate": 0.0005570479895701068,
      "loss": 2.411,
      "step": 39729
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6901422739028931,
      "learning_rate": 0.0005570458804334628,
      "loss": 2.3929,
      "step": 39730
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6385588645935059,
      "learning_rate": 0.0005570437712490291,
      "loss": 2.1136,
      "step": 39731
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6626220345497131,
      "learning_rate": 0.0005570416620168062,
      "loss": 2.2132,
      "step": 39732
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6854562759399414,
      "learning_rate": 0.0005570395527367942,
      "loss": 2.4875,
      "step": 39733
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.7888649702072144,
      "learning_rate": 0.0005570374434089939,
      "loss": 2.3205,
      "step": 39734
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6970335245132446,
      "learning_rate": 0.0005570353340334054,
      "loss": 2.3519,
      "step": 39735
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6824108958244324,
      "learning_rate": 0.0005570332246100291,
      "loss": 2.4272,
      "step": 39736
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6409052610397339,
      "learning_rate": 0.0005570311151388656,
      "loss": 2.1321,
      "step": 39737
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6556654572486877,
      "learning_rate": 0.0005570290056199151,
      "loss": 2.3101,
      "step": 39738
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6549684405326843,
      "learning_rate": 0.0005570268960531781,
      "loss": 2.2701,
      "step": 39739
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6636161804199219,
      "learning_rate": 0.0005570247864386549,
      "loss": 2.2239,
      "step": 39740
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6662521958351135,
      "learning_rate": 0.000557022676776346,
      "loss": 2.1606,
      "step": 39741
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6393499970436096,
      "learning_rate": 0.0005570205670662517,
      "loss": 2.0585,
      "step": 39742
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.711696445941925,
      "learning_rate": 0.0005570184573083724,
      "loss": 2.2178,
      "step": 39743
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6849641799926758,
      "learning_rate": 0.0005570163475027085,
      "loss": 2.2917,
      "step": 39744
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.7137271761894226,
      "learning_rate": 0.0005570142376492605,
      "loss": 2.1591,
      "step": 39745
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6737106442451477,
      "learning_rate": 0.0005570121277480287,
      "loss": 2.0635,
      "step": 39746
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6824328303337097,
      "learning_rate": 0.0005570100177990134,
      "loss": 2.3708,
      "step": 39747
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.7041174173355103,
      "learning_rate": 0.0005570079078022151,
      "loss": 2.1262,
      "step": 39748
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.7153530716896057,
      "learning_rate": 0.0005570057977576342,
      "loss": 2.2107,
      "step": 39749
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.7266848087310791,
      "learning_rate": 0.000557003687665271,
      "loss": 2.4681,
      "step": 39750
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6868345737457275,
      "learning_rate": 0.0005570015775251261,
      "loss": 2.2989,
      "step": 39751
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6855387687683105,
      "learning_rate": 0.0005569994673371997,
      "loss": 2.4708,
      "step": 39752
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.7360922694206238,
      "learning_rate": 0.0005569973571014922,
      "loss": 2.5231,
      "step": 39753
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.7768763303756714,
      "learning_rate": 0.0005569952468180042,
      "loss": 2.2867,
      "step": 39754
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.7939938902854919,
      "learning_rate": 0.0005569931364867357,
      "loss": 2.3787,
      "step": 39755
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.7375327944755554,
      "learning_rate": 0.0005569910261076875,
      "loss": 2.4187,
      "step": 39756
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.7043344378471375,
      "learning_rate": 0.0005569889156808597,
      "loss": 2.2235,
      "step": 39757
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.629162609577179,
      "learning_rate": 0.0005569868052062529,
      "loss": 2.2714,
      "step": 39758
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.8946489691734314,
      "learning_rate": 0.0005569846946838675,
      "loss": 2.2048,
      "step": 39759
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.8101338148117065,
      "learning_rate": 0.0005569825841137036,
      "loss": 2.2133,
      "step": 39760
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.627178430557251,
      "learning_rate": 0.0005569804734957618,
      "loss": 2.3082,
      "step": 39761
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.786677896976471,
      "learning_rate": 0.0005569783628300426,
      "loss": 2.2454,
      "step": 39762
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.7467975616455078,
      "learning_rate": 0.0005569762521165462,
      "loss": 2.2793,
      "step": 39763
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6824052333831787,
      "learning_rate": 0.000556974141355273,
      "loss": 2.136,
      "step": 39764
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6353722810745239,
      "learning_rate": 0.0005569720305462235,
      "loss": 2.2331,
      "step": 39765
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.7018542289733887,
      "learning_rate": 0.000556969919689398,
      "loss": 2.2485,
      "step": 39766
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.7016906142234802,
      "learning_rate": 0.0005569678087847971,
      "loss": 2.5135,
      "step": 39767
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.7890077829360962,
      "learning_rate": 0.000556965697832421,
      "loss": 2.2624,
      "step": 39768
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.7321780323982239,
      "learning_rate": 0.00055696358683227,
      "loss": 2.3331,
      "step": 39769
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6501492261886597,
      "learning_rate": 0.0005569614757843448,
      "loss": 2.14,
      "step": 39770
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.8539415001869202,
      "learning_rate": 0.0005569593646886455,
      "loss": 2.1302,
      "step": 39771
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6876341104507446,
      "learning_rate": 0.0005569572535451726,
      "loss": 2.1773,
      "step": 39772
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6454915404319763,
      "learning_rate": 0.0005569551423539264,
      "loss": 2.4138,
      "step": 39773
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.7948508858680725,
      "learning_rate": 0.0005569530311149075,
      "loss": 2.0496,
      "step": 39774
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.8409600853919983,
      "learning_rate": 0.0005569509198281162,
      "loss": 2.2218,
      "step": 39775
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.7300565242767334,
      "learning_rate": 0.0005569488084935528,
      "loss": 2.2151,
      "step": 39776
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.7508371472358704,
      "learning_rate": 0.0005569466971112179,
      "loss": 2.245,
      "step": 39777
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.7143235802650452,
      "learning_rate": 0.0005569445856811117,
      "loss": 2.1461,
      "step": 39778
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.8022832870483398,
      "learning_rate": 0.0005569424742032346,
      "loss": 2.3398,
      "step": 39779
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6998946666717529,
      "learning_rate": 0.0005569403626775871,
      "loss": 2.3508,
      "step": 39780
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6994722485542297,
      "learning_rate": 0.0005569382511041695,
      "loss": 2.3593,
      "step": 39781
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.7123031616210938,
      "learning_rate": 0.0005569361394829821,
      "loss": 2.3734,
      "step": 39782
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6839370727539062,
      "learning_rate": 0.0005569340278140256,
      "loss": 2.2875,
      "step": 39783
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6730503439903259,
      "learning_rate": 0.0005569319160973001,
      "loss": 2.3672,
      "step": 39784
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6734417676925659,
      "learning_rate": 0.0005569298043328063,
      "loss": 2.3328,
      "step": 39785
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.7079734802246094,
      "learning_rate": 0.0005569276925205442,
      "loss": 2.1436,
      "step": 39786
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.680402398109436,
      "learning_rate": 0.0005569255806605145,
      "loss": 2.1573,
      "step": 39787
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.8021984100341797,
      "learning_rate": 0.0005569234687527174,
      "loss": 2.3442,
      "step": 39788
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6080526113510132,
      "learning_rate": 0.0005569213567971535,
      "loss": 2.1681,
      "step": 39789
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6537806987762451,
      "learning_rate": 0.0005569192447938229,
      "loss": 2.1453,
      "step": 39790
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.8148815035820007,
      "learning_rate": 0.0005569171327427263,
      "loss": 2.569,
      "step": 39791
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.7834347486495972,
      "learning_rate": 0.000556915020643864,
      "loss": 2.2882,
      "step": 39792
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.7170102000236511,
      "learning_rate": 0.0005569129084972362,
      "loss": 2.1543,
      "step": 39793
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.9432059526443481,
      "learning_rate": 0.0005569107963028435,
      "loss": 2.2672,
      "step": 39794
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.7144389152526855,
      "learning_rate": 0.0005569086840606862,
      "loss": 2.2989,
      "step": 39795
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.7775499224662781,
      "learning_rate": 0.0005569065717707648,
      "loss": 2.3926,
      "step": 39796
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6826828718185425,
      "learning_rate": 0.0005569044594330795,
      "loss": 2.386,
      "step": 39797
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6529734134674072,
      "learning_rate": 0.0005569023470476309,
      "loss": 2.3801,
      "step": 39798
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6441653966903687,
      "learning_rate": 0.0005569002346144193,
      "loss": 2.3368,
      "step": 39799
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6160901188850403,
      "learning_rate": 0.0005568981221334451,
      "loss": 2.241,
      "step": 39800
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6991111636161804,
      "learning_rate": 0.0005568960096047087,
      "loss": 2.296,
      "step": 39801
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6833149194717407,
      "learning_rate": 0.0005568938970282105,
      "loss": 2.298,
      "step": 39802
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6557127237319946,
      "learning_rate": 0.0005568917844039509,
      "loss": 2.23,
      "step": 39803
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.7389007806777954,
      "learning_rate": 0.0005568896717319301,
      "loss": 2.4445,
      "step": 39804
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6617119312286377,
      "learning_rate": 0.0005568875590121488,
      "loss": 2.0937,
      "step": 39805
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.682580828666687,
      "learning_rate": 0.0005568854462446071,
      "loss": 2.3276,
      "step": 39806
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.7329707145690918,
      "learning_rate": 0.0005568833334293059,
      "loss": 2.3458,
      "step": 39807
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6611601710319519,
      "learning_rate": 0.0005568812205662449,
      "loss": 2.2821,
      "step": 39808
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.7154085636138916,
      "learning_rate": 0.000556879107655425,
      "loss": 2.2406,
      "step": 39809
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.7333797216415405,
      "learning_rate": 0.0005568769946968464,
      "loss": 2.4441,
      "step": 39810
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6613203287124634,
      "learning_rate": 0.0005568748816905094,
      "loss": 2.3111,
      "step": 39811
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.7020727396011353,
      "learning_rate": 0.0005568727686364146,
      "loss": 2.3959,
      "step": 39812
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6818769574165344,
      "learning_rate": 0.0005568706555345623,
      "loss": 2.2407,
      "step": 39813
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6575005054473877,
      "learning_rate": 0.0005568685423849529,
      "loss": 2.0995,
      "step": 39814
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6483529210090637,
      "learning_rate": 0.0005568664291875868,
      "loss": 2.2329,
      "step": 39815
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6445338129997253,
      "learning_rate": 0.0005568643159424644,
      "loss": 2.1704,
      "step": 39816
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6568291783332825,
      "learning_rate": 0.000556862202649586,
      "loss": 2.4499,
      "step": 39817
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.7739246487617493,
      "learning_rate": 0.0005568600893089521,
      "loss": 2.1879,
      "step": 39818
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6775502562522888,
      "learning_rate": 0.000556857975920563,
      "loss": 2.4882,
      "step": 39819
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6934255361557007,
      "learning_rate": 0.0005568558624844194,
      "loss": 2.053,
      "step": 39820
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.7393532991409302,
      "learning_rate": 0.0005568537490005212,
      "loss": 2.4286,
      "step": 39821
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.7169396281242371,
      "learning_rate": 0.0005568516354688691,
      "loss": 2.3091,
      "step": 39822
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.7581129670143127,
      "learning_rate": 0.0005568495218894635,
      "loss": 2.4827,
      "step": 39823
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.7089429497718811,
      "learning_rate": 0.0005568474082623046,
      "loss": 2.2309,
      "step": 39824
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6696142554283142,
      "learning_rate": 0.0005568452945873929,
      "loss": 2.4121,
      "step": 39825
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6804126501083374,
      "learning_rate": 0.000556843180864729,
      "loss": 2.2093,
      "step": 39826
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.749904215335846,
      "learning_rate": 0.0005568410670943129,
      "loss": 2.2628,
      "step": 39827
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.7025558948516846,
      "learning_rate": 0.0005568389532761453,
      "loss": 2.3746,
      "step": 39828
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.698051929473877,
      "learning_rate": 0.0005568368394102266,
      "loss": 2.2576,
      "step": 39829
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.7152224183082581,
      "learning_rate": 0.000556834725496557,
      "loss": 2.0839,
      "step": 39830
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6932985782623291,
      "learning_rate": 0.0005568326115351369,
      "loss": 2.2733,
      "step": 39831
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.7497408390045166,
      "learning_rate": 0.0005568304975259668,
      "loss": 2.4596,
      "step": 39832
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.7180967330932617,
      "learning_rate": 0.0005568283834690472,
      "loss": 2.4526,
      "step": 39833
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.7246482968330383,
      "learning_rate": 0.0005568262693643782,
      "loss": 2.4659,
      "step": 39834
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6842982769012451,
      "learning_rate": 0.0005568241552119604,
      "loss": 2.3669,
      "step": 39835
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6854373216629028,
      "learning_rate": 0.000556822041011794,
      "loss": 2.2862,
      "step": 39836
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.711269736289978,
      "learning_rate": 0.0005568199267638798,
      "loss": 2.1833,
      "step": 39837
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.7244740724563599,
      "learning_rate": 0.0005568178124682176,
      "loss": 2.2932,
      "step": 39838
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6590456366539001,
      "learning_rate": 0.0005568156981248084,
      "loss": 2.1401,
      "step": 39839
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.7011553049087524,
      "learning_rate": 0.0005568135837336522,
      "loss": 2.3657,
      "step": 39840
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.7415750622749329,
      "learning_rate": 0.0005568114692947496,
      "loss": 2.6081,
      "step": 39841
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6359907984733582,
      "learning_rate": 0.0005568093548081008,
      "loss": 2.1569,
      "step": 39842
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.7913838624954224,
      "learning_rate": 0.0005568072402737064,
      "loss": 2.2309,
      "step": 39843
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.8380227088928223,
      "learning_rate": 0.0005568051256915666,
      "loss": 2.1931,
      "step": 39844
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6560083627700806,
      "learning_rate": 0.0005568030110616819,
      "loss": 2.2862,
      "step": 39845
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.7218148708343506,
      "learning_rate": 0.0005568008963840526,
      "loss": 2.445,
      "step": 39846
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6930971145629883,
      "learning_rate": 0.0005567987816586792,
      "loss": 2.3563,
      "step": 39847
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6424285173416138,
      "learning_rate": 0.0005567966668855621,
      "loss": 2.353,
      "step": 39848
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6806643009185791,
      "learning_rate": 0.0005567945520647017,
      "loss": 2.1645,
      "step": 39849
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6639282703399658,
      "learning_rate": 0.0005567924371960983,
      "loss": 2.4015,
      "step": 39850
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6420656442642212,
      "learning_rate": 0.0005567903222797523,
      "loss": 2.1961,
      "step": 39851
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.7092220783233643,
      "learning_rate": 0.0005567882073156641,
      "loss": 2.0726,
      "step": 39852
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.7063464522361755,
      "learning_rate": 0.0005567860923038342,
      "loss": 2.4096,
      "step": 39853
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6759513020515442,
      "learning_rate": 0.0005567839772442629,
      "loss": 2.3855,
      "step": 39854
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6508781909942627,
      "learning_rate": 0.0005567818621369506,
      "loss": 2.1556,
      "step": 39855
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6599644422531128,
      "learning_rate": 0.0005567797469818978,
      "loss": 2.4093,
      "step": 39856
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6776238679885864,
      "learning_rate": 0.0005567776317791047,
      "loss": 2.3349,
      "step": 39857
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.7150996327400208,
      "learning_rate": 0.0005567755165285718,
      "loss": 2.0926,
      "step": 39858
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.7055464386940002,
      "learning_rate": 0.0005567734012302994,
      "loss": 2.3221,
      "step": 39859
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6852743625640869,
      "learning_rate": 0.0005567712858842881,
      "loss": 2.3884,
      "step": 39860
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.8037324547767639,
      "learning_rate": 0.0005567691704905382,
      "loss": 2.3078,
      "step": 39861
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.7181708216667175,
      "learning_rate": 0.00055676705504905,
      "loss": 2.4342,
      "step": 39862
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.7695279121398926,
      "learning_rate": 0.0005567649395598239,
      "loss": 2.3198,
      "step": 39863
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6869370937347412,
      "learning_rate": 0.0005567628240228604,
      "loss": 2.1956,
      "step": 39864
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6772776246070862,
      "learning_rate": 0.0005567607084381599,
      "loss": 2.4926,
      "step": 39865
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6364886164665222,
      "learning_rate": 0.0005567585928057227,
      "loss": 2.4943,
      "step": 39866
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.7218289375305176,
      "learning_rate": 0.0005567564771255492,
      "loss": 2.2182,
      "step": 39867
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.779471218585968,
      "learning_rate": 0.0005567543613976398,
      "loss": 2.204,
      "step": 39868
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6748028993606567,
      "learning_rate": 0.0005567522456219951,
      "loss": 2.4585,
      "step": 39869
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.7117692828178406,
      "learning_rate": 0.0005567501297986151,
      "loss": 2.4541,
      "step": 39870
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.7085400223731995,
      "learning_rate": 0.0005567480139275005,
      "loss": 2.2312,
      "step": 39871
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6795132756233215,
      "learning_rate": 0.0005567458980086515,
      "loss": 2.3366,
      "step": 39872
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.7379907369613647,
      "learning_rate": 0.0005567437820420687,
      "loss": 2.2729,
      "step": 39873
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.7202892303466797,
      "learning_rate": 0.0005567416660277523,
      "loss": 2.3712,
      "step": 39874
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6870927810668945,
      "learning_rate": 0.0005567395499657029,
      "loss": 2.2514,
      "step": 39875
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6952093243598938,
      "learning_rate": 0.0005567374338559207,
      "loss": 2.1686,
      "step": 39876
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.7540605664253235,
      "learning_rate": 0.0005567353176984063,
      "loss": 2.2993,
      "step": 39877
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.8127297759056091,
      "learning_rate": 0.0005567332014931597,
      "loss": 2.3682,
      "step": 39878
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6525990962982178,
      "learning_rate": 0.0005567310852401818,
      "loss": 2.2121,
      "step": 39879
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.7326436638832092,
      "learning_rate": 0.0005567289689394725,
      "loss": 2.0837,
      "step": 39880
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.693908154964447,
      "learning_rate": 0.0005567268525910326,
      "loss": 2.3692,
      "step": 39881
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6901671886444092,
      "learning_rate": 0.0005567247361948623,
      "loss": 2.2417,
      "step": 39882
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.7198260426521301,
      "learning_rate": 0.000556722619750962,
      "loss": 2.4182,
      "step": 39883
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6790831685066223,
      "learning_rate": 0.0005567205032593321,
      "loss": 2.3936,
      "step": 39884
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.7028412818908691,
      "learning_rate": 0.0005567183867199731,
      "loss": 2.0651,
      "step": 39885
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6404677033424377,
      "learning_rate": 0.0005567162701328852,
      "loss": 2.0787,
      "step": 39886
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.666740357875824,
      "learning_rate": 0.000556714153498069,
      "loss": 2.1968,
      "step": 39887
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6616307497024536,
      "learning_rate": 0.0005567120368155248,
      "loss": 2.2859,
      "step": 39888
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6642329096794128,
      "learning_rate": 0.000556709920085253,
      "loss": 2.2183,
      "step": 39889
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6277256011962891,
      "learning_rate": 0.0005567078033072539,
      "loss": 2.4002,
      "step": 39890
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.5923824310302734,
      "learning_rate": 0.0005567056864815279,
      "loss": 2.4212,
      "step": 39891
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6884660124778748,
      "learning_rate": 0.0005567035696080756,
      "loss": 2.3754,
      "step": 39892
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.7641395926475525,
      "learning_rate": 0.0005567014526868971,
      "loss": 2.057,
      "step": 39893
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6284443736076355,
      "learning_rate": 0.0005566993357179931,
      "loss": 2.3004,
      "step": 39894
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.7268925309181213,
      "learning_rate": 0.0005566972187013639,
      "loss": 2.1564,
      "step": 39895
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.7451715469360352,
      "learning_rate": 0.0005566951016370097,
      "loss": 2.2817,
      "step": 39896
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.7002559900283813,
      "learning_rate": 0.0005566929845249311,
      "loss": 2.3492,
      "step": 39897
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6883137226104736,
      "learning_rate": 0.0005566908673651285,
      "loss": 2.3124,
      "step": 39898
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.7032965421676636,
      "learning_rate": 0.0005566887501576021,
      "loss": 2.2575,
      "step": 39899
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.7027909755706787,
      "learning_rate": 0.0005566866329023525,
      "loss": 2.3957,
      "step": 39900
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.7282582521438599,
      "learning_rate": 0.0005566845155993799,
      "loss": 2.27,
      "step": 39901
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.7160139083862305,
      "learning_rate": 0.0005566823982486849,
      "loss": 2.4392,
      "step": 39902
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6548214554786682,
      "learning_rate": 0.0005566802808502678,
      "loss": 2.4788,
      "step": 39903
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.7331069707870483,
      "learning_rate": 0.0005566781634041289,
      "loss": 2.2771,
      "step": 39904
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6529485583305359,
      "learning_rate": 0.0005566760459102688,
      "loss": 2.4773,
      "step": 39905
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.7075919508934021,
      "learning_rate": 0.0005566739283686876,
      "loss": 2.4511,
      "step": 39906
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.7470900416374207,
      "learning_rate": 0.0005566718107793861,
      "loss": 2.3303,
      "step": 39907
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.7214161157608032,
      "learning_rate": 0.0005566696931423644,
      "loss": 2.2945,
      "step": 39908
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6739645004272461,
      "learning_rate": 0.0005566675754576229,
      "loss": 2.3357,
      "step": 39909
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.7158689498901367,
      "learning_rate": 0.0005566654577251622,
      "loss": 2.265,
      "step": 39910
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6838754415512085,
      "learning_rate": 0.0005566633399449823,
      "loss": 2.1534,
      "step": 39911
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.5773195624351501,
      "learning_rate": 0.000556661222117084,
      "loss": 2.2503,
      "step": 39912
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6395968198776245,
      "learning_rate": 0.0005566591042414675,
      "loss": 2.0682,
      "step": 39913
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6292380690574646,
      "learning_rate": 0.0005566569863181332,
      "loss": 2.3937,
      "step": 39914
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.7367310523986816,
      "learning_rate": 0.0005566548683470815,
      "loss": 2.2347,
      "step": 39915
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.7248936295509338,
      "learning_rate": 0.0005566527503283129,
      "loss": 2.1761,
      "step": 39916
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6436375379562378,
      "learning_rate": 0.0005566506322618276,
      "loss": 2.2768,
      "step": 39917
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.7475285530090332,
      "learning_rate": 0.0005566485141476261,
      "loss": 2.3285,
      "step": 39918
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.7514943480491638,
      "learning_rate": 0.000556646395985709,
      "loss": 1.9932,
      "step": 39919
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.7024145722389221,
      "learning_rate": 0.0005566442777760764,
      "loss": 2.3421,
      "step": 39920
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6936362385749817,
      "learning_rate": 0.0005566421595187287,
      "loss": 2.2857,
      "step": 39921
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6730011105537415,
      "learning_rate": 0.0005566400412136664,
      "loss": 2.402,
      "step": 39922
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6877206563949585,
      "learning_rate": 0.0005566379228608899,
      "loss": 2.3679,
      "step": 39923
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6720536947250366,
      "learning_rate": 0.0005566358044603995,
      "loss": 2.3771,
      "step": 39924
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.7832421660423279,
      "learning_rate": 0.0005566336860121958,
      "loss": 2.2839,
      "step": 39925
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.7007130980491638,
      "learning_rate": 0.000556631567516279,
      "loss": 2.3284,
      "step": 39926
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6248365044593811,
      "learning_rate": 0.0005566294489726494,
      "loss": 2.3377,
      "step": 39927
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.7319793701171875,
      "learning_rate": 0.0005566273303813078,
      "loss": 2.0547,
      "step": 39928
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6506639122962952,
      "learning_rate": 0.0005566252117422541,
      "loss": 2.2921,
      "step": 39929
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.7724391222000122,
      "learning_rate": 0.0005566230930554891,
      "loss": 2.0993,
      "step": 39930
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.7255648374557495,
      "learning_rate": 0.0005566209743210128,
      "loss": 2.2563,
      "step": 39931
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.7452461123466492,
      "learning_rate": 0.0005566188555388261,
      "loss": 2.1553,
      "step": 39932
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.7126902937889099,
      "learning_rate": 0.000556616736708929,
      "loss": 2.4341,
      "step": 39933
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.7979218363761902,
      "learning_rate": 0.0005566146178313219,
      "loss": 2.315,
      "step": 39934
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.7678552269935608,
      "learning_rate": 0.0005566124989060054,
      "loss": 2.1396,
      "step": 39935
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.7756898403167725,
      "learning_rate": 0.0005566103799329797,
      "loss": 2.4698,
      "step": 39936
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6674593687057495,
      "learning_rate": 0.0005566082609122454,
      "loss": 2.2412,
      "step": 39937
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6292733550071716,
      "learning_rate": 0.0005566061418438028,
      "loss": 2.3628,
      "step": 39938
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.658799946308136,
      "learning_rate": 0.0005566040227276522,
      "loss": 2.2834,
      "step": 39939
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.7038922905921936,
      "learning_rate": 0.0005566019035637941,
      "loss": 2.309,
      "step": 39940
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6668065786361694,
      "learning_rate": 0.0005565997843522289,
      "loss": 2.2896,
      "step": 39941
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6673969626426697,
      "learning_rate": 0.0005565976650929568,
      "loss": 2.3807,
      "step": 39942
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6812837719917297,
      "learning_rate": 0.0005565955457859785,
      "loss": 2.2357,
      "step": 39943
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6572301387786865,
      "learning_rate": 0.0005565934264312941,
      "loss": 2.2405,
      "step": 39944
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6909244060516357,
      "learning_rate": 0.0005565913070289043,
      "loss": 2.2083,
      "step": 39945
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6635010242462158,
      "learning_rate": 0.0005565891875788093,
      "loss": 2.3055,
      "step": 39946
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6500863432884216,
      "learning_rate": 0.0005565870680810095,
      "loss": 2.2019,
      "step": 39947
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6632886528968811,
      "learning_rate": 0.0005565849485355053,
      "loss": 2.4081,
      "step": 39948
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6759406924247742,
      "learning_rate": 0.0005565828289422971,
      "loss": 2.3082,
      "step": 39949
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.7249971032142639,
      "learning_rate": 0.0005565807093013853,
      "loss": 2.5541,
      "step": 39950
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.688931405544281,
      "learning_rate": 0.0005565785896127703,
      "loss": 2.2869,
      "step": 39951
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.697097659111023,
      "learning_rate": 0.0005565764698764526,
      "loss": 2.3323,
      "step": 39952
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6618892550468445,
      "learning_rate": 0.0005565743500924324,
      "loss": 2.4174,
      "step": 39953
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.7019771337509155,
      "learning_rate": 0.0005565722302607102,
      "loss": 2.1263,
      "step": 39954
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6609507203102112,
      "learning_rate": 0.0005565701103812864,
      "loss": 2.2931,
      "step": 39955
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6312598586082458,
      "learning_rate": 0.0005565679904541613,
      "loss": 2.4241,
      "step": 39956
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.7217223644256592,
      "learning_rate": 0.0005565658704793355,
      "loss": 2.4013,
      "step": 39957
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.7104731798171997,
      "learning_rate": 0.0005565637504568092,
      "loss": 2.3521,
      "step": 39958
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.7039015293121338,
      "learning_rate": 0.0005565616303865829,
      "loss": 2.4062,
      "step": 39959
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6103391051292419,
      "learning_rate": 0.0005565595102686568,
      "loss": 2.2224,
      "step": 39960
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6445378065109253,
      "learning_rate": 0.0005565573901030316,
      "loss": 2.4175,
      "step": 39961
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6869528293609619,
      "learning_rate": 0.0005565552698897075,
      "loss": 2.1409,
      "step": 39962
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6640816330909729,
      "learning_rate": 0.000556553149628685,
      "loss": 2.3212,
      "step": 39963
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.7572781443595886,
      "learning_rate": 0.0005565510293199642,
      "loss": 2.3235,
      "step": 39964
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6476484537124634,
      "learning_rate": 0.0005565489089635458,
      "loss": 2.2381,
      "step": 39965
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6878963112831116,
      "learning_rate": 0.0005565467885594303,
      "loss": 2.1561,
      "step": 39966
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.7001088857650757,
      "learning_rate": 0.0005565446681076177,
      "loss": 2.2983,
      "step": 39967
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6770575046539307,
      "learning_rate": 0.0005565425476081087,
      "loss": 2.3709,
      "step": 39968
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.8352444171905518,
      "learning_rate": 0.0005565404270609037,
      "loss": 2.3946,
      "step": 39969
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.7437409162521362,
      "learning_rate": 0.0005565383064660027,
      "loss": 2.2815,
      "step": 39970
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6764615774154663,
      "learning_rate": 0.0005565361858234067,
      "loss": 2.5843,
      "step": 39971
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.7282742261886597,
      "learning_rate": 0.0005565340651331156,
      "loss": 2.2083,
      "step": 39972
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6942625045776367,
      "learning_rate": 0.00055653194439513,
      "loss": 2.4058,
      "step": 39973
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6792436838150024,
      "learning_rate": 0.0005565298236094503,
      "loss": 2.2925,
      "step": 39974
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.7153542637825012,
      "learning_rate": 0.0005565277027760768,
      "loss": 2.1591,
      "step": 39975
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6919055581092834,
      "learning_rate": 0.0005565255818950101,
      "loss": 2.2108,
      "step": 39976
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6115350127220154,
      "learning_rate": 0.0005565234609662504,
      "loss": 2.2529,
      "step": 39977
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.7151932716369629,
      "learning_rate": 0.0005565213399897981,
      "loss": 2.5382,
      "step": 39978
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6811935901641846,
      "learning_rate": 0.0005565192189656536,
      "loss": 2.2226,
      "step": 39979
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.5889165997505188,
      "learning_rate": 0.0005565170978938174,
      "loss": 2.0756,
      "step": 39980
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6627148985862732,
      "learning_rate": 0.0005565149767742897,
      "loss": 2.2383,
      "step": 39981
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.7449453473091125,
      "learning_rate": 0.0005565128556070712,
      "loss": 2.2645,
      "step": 39982
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.7015286684036255,
      "learning_rate": 0.0005565107343921621,
      "loss": 2.2107,
      "step": 39983
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6801328063011169,
      "learning_rate": 0.0005565086131295627,
      "loss": 2.5614,
      "step": 39984
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.7049450278282166,
      "learning_rate": 0.0005565064918192736,
      "loss": 2.3236,
      "step": 39985
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6835364699363708,
      "learning_rate": 0.0005565043704612951,
      "loss": 2.422,
      "step": 39986
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6892072558403015,
      "learning_rate": 0.0005565022490556275,
      "loss": 2.0669,
      "step": 39987
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.7243703007698059,
      "learning_rate": 0.0005565001276022714,
      "loss": 2.2341,
      "step": 39988
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6674599051475525,
      "learning_rate": 0.000556498006101227,
      "loss": 2.2025,
      "step": 39989
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.7242921590805054,
      "learning_rate": 0.0005564958845524949,
      "loss": 2.1685,
      "step": 39990
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6674038767814636,
      "learning_rate": 0.0005564937629560753,
      "loss": 2.3078,
      "step": 39991
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6500576138496399,
      "learning_rate": 0.0005564916413119687,
      "loss": 2.2676,
      "step": 39992
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6769278049468994,
      "learning_rate": 0.0005564895196201754,
      "loss": 2.2173,
      "step": 39993
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.7750228047370911,
      "learning_rate": 0.0005564873978806959,
      "loss": 2.2573,
      "step": 39994
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6982447504997253,
      "learning_rate": 0.0005564852760935305,
      "loss": 2.103,
      "step": 39995
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.7297115921974182,
      "learning_rate": 0.0005564831542586797,
      "loss": 2.1779,
      "step": 39996
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.7304863929748535,
      "learning_rate": 0.0005564810323761438,
      "loss": 2.1973,
      "step": 39997
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6799125671386719,
      "learning_rate": 0.0005564789104459233,
      "loss": 2.3585,
      "step": 39998
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6854064464569092,
      "learning_rate": 0.0005564767884680185,
      "loss": 2.1434,
      "step": 39999
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.8565124869346619,
      "learning_rate": 0.0005564746664424298,
      "loss": 2.4337,
      "step": 40000
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.632851779460907,
      "learning_rate": 0.0005564725443691576,
      "loss": 2.0004,
      "step": 40001
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6547630429267883,
      "learning_rate": 0.0005564704222482023,
      "loss": 2.2766,
      "step": 40002
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.71522057056427,
      "learning_rate": 0.0005564683000795645,
      "loss": 2.3846,
      "step": 40003
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6366719007492065,
      "learning_rate": 0.0005564661778632441,
      "loss": 2.1221,
      "step": 40004
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6973688006401062,
      "learning_rate": 0.000556464055599242,
      "loss": 2.3243,
      "step": 40005
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6997036337852478,
      "learning_rate": 0.0005564619332875584,
      "loss": 2.3086,
      "step": 40006
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.7034834027290344,
      "learning_rate": 0.0005564598109281936,
      "loss": 2.2015,
      "step": 40007
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6860028505325317,
      "learning_rate": 0.000556457688521148,
      "loss": 2.4047,
      "step": 40008
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6513795852661133,
      "learning_rate": 0.0005564555660664222,
      "loss": 2.2168,
      "step": 40009
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6702552437782288,
      "learning_rate": 0.0005564534435640165,
      "loss": 2.2978,
      "step": 40010
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.692899227142334,
      "learning_rate": 0.0005564513210139312,
      "loss": 2.493,
      "step": 40011
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.665700376033783,
      "learning_rate": 0.0005564491984161667,
      "loss": 2.0634,
      "step": 40012
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.7936832904815674,
      "learning_rate": 0.0005564470757707235,
      "loss": 2.2166,
      "step": 40013
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6191203594207764,
      "learning_rate": 0.000556444953077602,
      "loss": 2.1278,
      "step": 40014
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.7113662362098694,
      "learning_rate": 0.0005564428303368027,
      "loss": 2.2274,
      "step": 40015
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.0569233894348145,
      "learning_rate": 0.0005564407075483254,
      "loss": 2.094,
      "step": 40016
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6854428648948669,
      "learning_rate": 0.0005564385847121713,
      "loss": 2.3966,
      "step": 40017
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.7257582545280457,
      "learning_rate": 0.0005564364618283403,
      "loss": 2.3462,
      "step": 40018
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6888575553894043,
      "learning_rate": 0.0005564343388968329,
      "loss": 2.1679,
      "step": 40019
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6871417760848999,
      "learning_rate": 0.0005564322159176496,
      "loss": 2.1824,
      "step": 40020
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.7067593932151794,
      "learning_rate": 0.0005564300928907907,
      "loss": 2.3859,
      "step": 40021
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.7125250101089478,
      "learning_rate": 0.0005564279698162565,
      "loss": 2.5796,
      "step": 40022
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.701666533946991,
      "learning_rate": 0.0005564258466940475,
      "loss": 2.1672,
      "step": 40023
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.7053881287574768,
      "learning_rate": 0.0005564237235241642,
      "loss": 2.2719,
      "step": 40024
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.7577143907546997,
      "learning_rate": 0.0005564216003066068,
      "loss": 2.303,
      "step": 40025
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6581554412841797,
      "learning_rate": 0.0005564194770413759,
      "loss": 2.2836,
      "step": 40026
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6328091025352478,
      "learning_rate": 0.0005564173537284717,
      "loss": 2.0812,
      "step": 40027
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.7392528057098389,
      "learning_rate": 0.0005564152303678947,
      "loss": 2.4257,
      "step": 40028
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.7674012780189514,
      "learning_rate": 0.0005564131069596452,
      "loss": 2.2101,
      "step": 40029
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.678051769733429,
      "learning_rate": 0.0005564109835037238,
      "loss": 2.1842,
      "step": 40030
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6255333423614502,
      "learning_rate": 0.0005564088600001306,
      "loss": 2.0471,
      "step": 40031
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.9009752869606018,
      "learning_rate": 0.0005564067364488663,
      "loss": 2.458,
      "step": 40032
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.7312791347503662,
      "learning_rate": 0.0005564046128499311,
      "loss": 2.2645,
      "step": 40033
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.7053887844085693,
      "learning_rate": 0.0005564024892033253,
      "loss": 2.2973,
      "step": 40034
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.7713528871536255,
      "learning_rate": 0.0005564003655090495,
      "loss": 2.1406,
      "step": 40035
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6684420704841614,
      "learning_rate": 0.0005563982417671041,
      "loss": 2.2181,
      "step": 40036
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6899215579032898,
      "learning_rate": 0.0005563961179774894,
      "loss": 2.3149,
      "step": 40037
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.7161946892738342,
      "learning_rate": 0.0005563939941402059,
      "loss": 2.1537,
      "step": 40038
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6778402924537659,
      "learning_rate": 0.0005563918702552539,
      "loss": 2.3406,
      "step": 40039
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6787219643592834,
      "learning_rate": 0.0005563897463226337,
      "loss": 2.241,
      "step": 40040
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.7336425185203552,
      "learning_rate": 0.0005563876223423459,
      "loss": 2.2727,
      "step": 40041
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.7176132202148438,
      "learning_rate": 0.0005563854983143908,
      "loss": 2.186,
      "step": 40042
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6023632287979126,
      "learning_rate": 0.0005563833742387688,
      "loss": 2.2784,
      "step": 40043
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6837670207023621,
      "learning_rate": 0.0005563812501154802,
      "loss": 2.3951,
      "step": 40044
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6747270822525024,
      "learning_rate": 0.0005563791259445256,
      "loss": 2.2333,
      "step": 40045
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6412444710731506,
      "learning_rate": 0.000556377001725905,
      "loss": 2.4268,
      "step": 40046
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6688358187675476,
      "learning_rate": 0.0005563748774596194,
      "loss": 2.3832,
      "step": 40047
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6898407936096191,
      "learning_rate": 0.0005563727531456688,
      "loss": 2.1914,
      "step": 40048
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6532241702079773,
      "learning_rate": 0.0005563706287840536,
      "loss": 2.3334,
      "step": 40049
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6563129425048828,
      "learning_rate": 0.0005563685043747744,
      "loss": 2.2541,
      "step": 40050
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.7192974090576172,
      "learning_rate": 0.0005563663799178313,
      "loss": 2.2046,
      "step": 40051
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.7458522319793701,
      "learning_rate": 0.0005563642554132249,
      "loss": 2.09,
      "step": 40052
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6433721780776978,
      "learning_rate": 0.0005563621308609556,
      "loss": 2.3043,
      "step": 40053
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.7756160497665405,
      "learning_rate": 0.0005563600062610237,
      "loss": 2.3234,
      "step": 40054
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6817682385444641,
      "learning_rate": 0.0005563578816134296,
      "loss": 2.241,
      "step": 40055
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.7055508494377136,
      "learning_rate": 0.0005563557569181737,
      "loss": 2.0866,
      "step": 40056
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.9525611996650696,
      "learning_rate": 0.0005563536321752565,
      "loss": 2.3669,
      "step": 40057
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.7651941776275635,
      "learning_rate": 0.0005563515073846783,
      "loss": 2.0484,
      "step": 40058
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.7878032326698303,
      "learning_rate": 0.0005563493825464394,
      "loss": 2.2911,
      "step": 40059
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.8897990584373474,
      "learning_rate": 0.0005563472576605405,
      "loss": 2.4067,
      "step": 40060
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.7919256687164307,
      "learning_rate": 0.0005563451327269818,
      "loss": 2.0553,
      "step": 40061
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.7551004886627197,
      "learning_rate": 0.0005563430077457636,
      "loss": 2.1677,
      "step": 40062
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.7630205154418945,
      "learning_rate": 0.0005563408827168865,
      "loss": 2.1665,
      "step": 40063
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.9275009632110596,
      "learning_rate": 0.0005563387576403507,
      "loss": 2.3141,
      "step": 40064
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.7911143898963928,
      "learning_rate": 0.0005563366325161567,
      "loss": 2.4345,
      "step": 40065
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6835479140281677,
      "learning_rate": 0.0005563345073443048,
      "loss": 2.1935,
      "step": 40066
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.8223118185997009,
      "learning_rate": 0.0005563323821247956,
      "loss": 2.2079,
      "step": 40067
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.3415029048919678,
      "learning_rate": 0.0005563302568576293,
      "loss": 2.0557,
      "step": 40068
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6792449355125427,
      "learning_rate": 0.0005563281315428064,
      "loss": 2.4404,
      "step": 40069
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.9266813397407532,
      "learning_rate": 0.0005563260061803273,
      "loss": 2.2861,
      "step": 40070
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.9169555902481079,
      "learning_rate": 0.0005563238807701924,
      "loss": 2.181,
      "step": 40071
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6594277024269104,
      "learning_rate": 0.000556321755312402,
      "loss": 2.1972,
      "step": 40072
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.8156999349594116,
      "learning_rate": 0.0005563196298069564,
      "loss": 2.1495,
      "step": 40073
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.8624613285064697,
      "learning_rate": 0.0005563175042538563,
      "loss": 2.2133,
      "step": 40074
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.7383627891540527,
      "learning_rate": 0.0005563153786531019,
      "loss": 2.1996,
      "step": 40075
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6722114086151123,
      "learning_rate": 0.0005563132530046936,
      "loss": 2.1402,
      "step": 40076
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.7137354612350464,
      "learning_rate": 0.0005563111273086319,
      "loss": 2.4292,
      "step": 40077
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.726789116859436,
      "learning_rate": 0.0005563090015649171,
      "loss": 2.1552,
      "step": 40078
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.7998080253601074,
      "learning_rate": 0.0005563068757735495,
      "loss": 2.3469,
      "step": 40079
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6908833980560303,
      "learning_rate": 0.0005563047499345298,
      "loss": 2.3358,
      "step": 40080
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.691589891910553,
      "learning_rate": 0.0005563026240478581,
      "loss": 2.2731,
      "step": 40081
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.9978593587875366,
      "learning_rate": 0.0005563004981135349,
      "loss": 2.3757,
      "step": 40082
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6877949833869934,
      "learning_rate": 0.0005562983721315607,
      "loss": 2.5091,
      "step": 40083
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.759601354598999,
      "learning_rate": 0.0005562962461019357,
      "loss": 2.2504,
      "step": 40084
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.7348016500473022,
      "learning_rate": 0.0005562941200246604,
      "loss": 2.3198,
      "step": 40085
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.69441819190979,
      "learning_rate": 0.0005562919938997352,
      "loss": 2.2074,
      "step": 40086
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.802641749382019,
      "learning_rate": 0.0005562898677271606,
      "loss": 2.1974,
      "step": 40087
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.7591481804847717,
      "learning_rate": 0.0005562877415069366,
      "loss": 2.2362,
      "step": 40088
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.7057657241821289,
      "learning_rate": 0.0005562856152390641,
      "loss": 2.2905,
      "step": 40089
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.7040449976921082,
      "learning_rate": 0.0005562834889235432,
      "loss": 2.2036,
      "step": 40090
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6875885725021362,
      "learning_rate": 0.0005562813625603744,
      "loss": 1.9341,
      "step": 40091
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.7287554144859314,
      "learning_rate": 0.0005562792361495579,
      "loss": 2.3979,
      "step": 40092
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6469164490699768,
      "learning_rate": 0.0005562771096910944,
      "loss": 2.2462,
      "step": 40093
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6645330190658569,
      "learning_rate": 0.0005562749831849841,
      "loss": 2.1478,
      "step": 40094
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6973733305931091,
      "learning_rate": 0.0005562728566312274,
      "loss": 2.4797,
      "step": 40095
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6630513668060303,
      "learning_rate": 0.0005562707300298249,
      "loss": 2.4287,
      "step": 40096
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.7265143394470215,
      "learning_rate": 0.0005562686033807767,
      "loss": 2.446,
      "step": 40097
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6241964101791382,
      "learning_rate": 0.0005562664766840834,
      "loss": 2.2179,
      "step": 40098
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6699742078781128,
      "learning_rate": 0.0005562643499397453,
      "loss": 2.0969,
      "step": 40099
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6886035799980164,
      "learning_rate": 0.0005562622231477628,
      "loss": 2.1832,
      "step": 40100
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6921819448471069,
      "learning_rate": 0.0005562600963081362,
      "loss": 2.249,
      "step": 40101
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6385670900344849,
      "learning_rate": 0.0005562579694208662,
      "loss": 2.0263,
      "step": 40102
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.660815954208374,
      "learning_rate": 0.0005562558424859529,
      "loss": 2.2758,
      "step": 40103
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.7655864357948303,
      "learning_rate": 0.0005562537155033968,
      "loss": 2.411,
      "step": 40104
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.7186431884765625,
      "learning_rate": 0.0005562515884731984,
      "loss": 2.2411,
      "step": 40105
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.7011926770210266,
      "learning_rate": 0.000556249461395358,
      "loss": 2.1991,
      "step": 40106
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.7036615610122681,
      "learning_rate": 0.0005562473342698758,
      "loss": 2.1115,
      "step": 40107
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6485072374343872,
      "learning_rate": 0.0005562452070967525,
      "loss": 2.2001,
      "step": 40108
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6726380586624146,
      "learning_rate": 0.0005562430798759883,
      "loss": 2.197,
      "step": 40109
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.7273842692375183,
      "learning_rate": 0.0005562409526075839,
      "loss": 2.3583,
      "step": 40110
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.7025638222694397,
      "learning_rate": 0.0005562388252915392,
      "loss": 2.0683,
      "step": 40111
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6538728475570679,
      "learning_rate": 0.000556236697927855,
      "loss": 2.3001,
      "step": 40112
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6534461379051208,
      "learning_rate": 0.0005562345705165315,
      "loss": 2.3653,
      "step": 40113
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6735146045684814,
      "learning_rate": 0.0005562324430575693,
      "loss": 2.2627,
      "step": 40114
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.8021653294563293,
      "learning_rate": 0.0005562303155509686,
      "loss": 2.3805,
      "step": 40115
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6591125726699829,
      "learning_rate": 0.0005562281879967297,
      "loss": 2.3838,
      "step": 40116
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.623059093952179,
      "learning_rate": 0.0005562260603948532,
      "loss": 2.1833,
      "step": 40117
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.5984055399894714,
      "learning_rate": 0.0005562239327453396,
      "loss": 2.261,
      "step": 40118
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6918245553970337,
      "learning_rate": 0.000556221805048189,
      "loss": 2.3417,
      "step": 40119
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.7245990633964539,
      "learning_rate": 0.000556219677303402,
      "loss": 2.3661,
      "step": 40120
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.7295213341712952,
      "learning_rate": 0.0005562175495109788,
      "loss": 2.4938,
      "step": 40121
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.8523014187812805,
      "learning_rate": 0.00055621542167092,
      "loss": 2.166,
      "step": 40122
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.7090939283370972,
      "learning_rate": 0.0005562132937832259,
      "loss": 2.6529,
      "step": 40123
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6462988257408142,
      "learning_rate": 0.000556211165847897,
      "loss": 2.1209,
      "step": 40124
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.7376822233200073,
      "learning_rate": 0.0005562090378649335,
      "loss": 2.3735,
      "step": 40125
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6934565305709839,
      "learning_rate": 0.0005562069098343359,
      "loss": 2.2658,
      "step": 40126
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.769443154335022,
      "learning_rate": 0.0005562047817561047,
      "loss": 2.3799,
      "step": 40127
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.7067482471466064,
      "learning_rate": 0.0005562026536302401,
      "loss": 2.3036,
      "step": 40128
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.7156384587287903,
      "learning_rate": 0.0005562005254567426,
      "loss": 2.319,
      "step": 40129
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.7085840702056885,
      "learning_rate": 0.0005561983972356126,
      "loss": 2.2835,
      "step": 40130
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6350188255310059,
      "learning_rate": 0.0005561962689668504,
      "loss": 2.3535,
      "step": 40131
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.790554940700531,
      "learning_rate": 0.0005561941406504566,
      "loss": 2.4307,
      "step": 40132
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6896087527275085,
      "learning_rate": 0.0005561920122864315,
      "loss": 2.3259,
      "step": 40133
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.641825258731842,
      "learning_rate": 0.0005561898838747755,
      "loss": 2.4134,
      "step": 40134
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.691832959651947,
      "learning_rate": 0.0005561877554154888,
      "loss": 2.1892,
      "step": 40135
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.7307023406028748,
      "learning_rate": 0.000556185626908572,
      "loss": 2.3196,
      "step": 40136
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6576041579246521,
      "learning_rate": 0.0005561834983540255,
      "loss": 2.1648,
      "step": 40137
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6706557869911194,
      "learning_rate": 0.0005561813697518497,
      "loss": 2.4648,
      "step": 40138
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.732571005821228,
      "learning_rate": 0.0005561792411020449,
      "loss": 2.4509,
      "step": 40139
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.7729184627532959,
      "learning_rate": 0.0005561771124046115,
      "loss": 2.3638,
      "step": 40140
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.614088237285614,
      "learning_rate": 0.00055617498365955,
      "loss": 2.2885,
      "step": 40141
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6497586965560913,
      "learning_rate": 0.0005561728548668607,
      "loss": 2.2981,
      "step": 40142
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6497154831886292,
      "learning_rate": 0.000556170726026544,
      "loss": 2.2175,
      "step": 40143
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6933636665344238,
      "learning_rate": 0.0005561685971386003,
      "loss": 2.4049,
      "step": 40144
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6155117154121399,
      "learning_rate": 0.0005561664682030301,
      "loss": 2.1667,
      "step": 40145
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.7079467177391052,
      "learning_rate": 0.0005561643392198338,
      "loss": 2.497,
      "step": 40146
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6632336378097534,
      "learning_rate": 0.0005561622101890115,
      "loss": 2.5272,
      "step": 40147
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6704077124595642,
      "learning_rate": 0.000556160081110564,
      "loss": 2.3378,
      "step": 40148
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6251567602157593,
      "learning_rate": 0.0005561579519844914,
      "loss": 1.9653,
      "step": 40149
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6773827075958252,
      "learning_rate": 0.0005561558228107942,
      "loss": 2.2412,
      "step": 40150
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6623477339744568,
      "learning_rate": 0.0005561536935894728,
      "loss": 2.5503,
      "step": 40151
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6436901092529297,
      "learning_rate": 0.0005561515643205277,
      "loss": 2.1548,
      "step": 40152
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6698703765869141,
      "learning_rate": 0.0005561494350039592,
      "loss": 2.3248,
      "step": 40153
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6492403149604797,
      "learning_rate": 0.0005561473056397675,
      "loss": 2.1271,
      "step": 40154
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6920177936553955,
      "learning_rate": 0.0005561451762279533,
      "loss": 2.1043,
      "step": 40155
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6517442464828491,
      "learning_rate": 0.0005561430467685168,
      "loss": 2.2118,
      "step": 40156
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6563259959220886,
      "learning_rate": 0.0005561409172614586,
      "loss": 2.2585,
      "step": 40157
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.7176297903060913,
      "learning_rate": 0.0005561387877067789,
      "loss": 2.3005,
      "step": 40158
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.7373538613319397,
      "learning_rate": 0.0005561366581044782,
      "loss": 2.2743,
      "step": 40159
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.658293604850769,
      "learning_rate": 0.0005561345284545567,
      "loss": 2.3691,
      "step": 40160
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6609867811203003,
      "learning_rate": 0.0005561323987570151,
      "loss": 2.4479,
      "step": 40161
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.738251268863678,
      "learning_rate": 0.0005561302690118536,
      "loss": 2.3518,
      "step": 40162
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.7376965284347534,
      "learning_rate": 0.0005561281392190727,
      "loss": 2.2565,
      "step": 40163
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.7067654728889465,
      "learning_rate": 0.0005561260093786727,
      "loss": 2.3975,
      "step": 40164
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.7062034010887146,
      "learning_rate": 0.0005561238794906541,
      "loss": 2.255,
      "step": 40165
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6790770292282104,
      "learning_rate": 0.0005561217495550173,
      "loss": 2.0883,
      "step": 40166
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.7116197347640991,
      "learning_rate": 0.0005561196195717624,
      "loss": 2.205,
      "step": 40167
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.661422848701477,
      "learning_rate": 0.0005561174895408902,
      "loss": 2.3103,
      "step": 40168
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6887289881706238,
      "learning_rate": 0.0005561153594624009,
      "loss": 2.2575,
      "step": 40169
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.7116853594779968,
      "learning_rate": 0.0005561132293362949,
      "loss": 2.38,
      "step": 40170
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.7033806443214417,
      "learning_rate": 0.0005561110991625727,
      "loss": 2.2122,
      "step": 40171
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.7043763399124146,
      "learning_rate": 0.0005561089689412345,
      "loss": 2.261,
      "step": 40172
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.775747537612915,
      "learning_rate": 0.0005561068386722808,
      "loss": 2.2927,
      "step": 40173
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.641115128993988,
      "learning_rate": 0.000556104708355712,
      "loss": 2.1716,
      "step": 40174
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.7532142400741577,
      "learning_rate": 0.0005561025779915286,
      "loss": 2.2778,
      "step": 40175
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6985424160957336,
      "learning_rate": 0.0005561004475797308,
      "loss": 2.2065,
      "step": 40176
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.638918936252594,
      "learning_rate": 0.0005560983171203191,
      "loss": 2.3034,
      "step": 40177
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.689881443977356,
      "learning_rate": 0.000556096186613294,
      "loss": 2.6006,
      "step": 40178
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.9074930548667908,
      "learning_rate": 0.0005560940560586557,
      "loss": 2.2346,
      "step": 40179
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.7125569581985474,
      "learning_rate": 0.0005560919254564046,
      "loss": 2.2905,
      "step": 40180
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.7136727571487427,
      "learning_rate": 0.0005560897948065413,
      "loss": 2.0645,
      "step": 40181
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.7346981763839722,
      "learning_rate": 0.0005560876641090661,
      "loss": 2.2929,
      "step": 40182
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6911985874176025,
      "learning_rate": 0.0005560855333639793,
      "loss": 2.2249,
      "step": 40183
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.73271644115448,
      "learning_rate": 0.0005560834025712813,
      "loss": 2.3222,
      "step": 40184
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6856634020805359,
      "learning_rate": 0.0005560812717309726,
      "loss": 2.397,
      "step": 40185
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.673528790473938,
      "learning_rate": 0.0005560791408430536,
      "loss": 2.1574,
      "step": 40186
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.7625306844711304,
      "learning_rate": 0.0005560770099075247,
      "loss": 2.227,
      "step": 40187
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6631233096122742,
      "learning_rate": 0.0005560748789243861,
      "loss": 2.1144,
      "step": 40188
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.9255581498146057,
      "learning_rate": 0.0005560727478936385,
      "loss": 2.1265,
      "step": 40189
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.7329496145248413,
      "learning_rate": 0.0005560706168152821,
      "loss": 2.1707,
      "step": 40190
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6636311411857605,
      "learning_rate": 0.0005560684856893174,
      "loss": 2.5537,
      "step": 40191
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.7167924046516418,
      "learning_rate": 0.0005560663545157446,
      "loss": 2.3237,
      "step": 40192
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.7174550890922546,
      "learning_rate": 0.0005560642232945643,
      "loss": 2.2932,
      "step": 40193
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6517220139503479,
      "learning_rate": 0.0005560620920257767,
      "loss": 2.3867,
      "step": 40194
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6352704763412476,
      "learning_rate": 0.0005560599607093825,
      "loss": 2.0616,
      "step": 40195
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6818063259124756,
      "learning_rate": 0.0005560578293453819,
      "loss": 2.4727,
      "step": 40196
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.713749885559082,
      "learning_rate": 0.0005560556979337753,
      "loss": 2.2376,
      "step": 40197
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.7638587951660156,
      "learning_rate": 0.000556053566474563,
      "loss": 2.3378,
      "step": 40198
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6880000233650208,
      "learning_rate": 0.0005560514349677456,
      "loss": 2.1978,
      "step": 40199
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.7280901074409485,
      "learning_rate": 0.0005560493034133234,
      "loss": 2.4094,
      "step": 40200
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.8089499473571777,
      "learning_rate": 0.0005560471718112968,
      "loss": 2.1826,
      "step": 40201
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6110448241233826,
      "learning_rate": 0.0005560450401616662,
      "loss": 2.253,
      "step": 40202
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.8384561538696289,
      "learning_rate": 0.000556042908464432,
      "loss": 2.5133,
      "step": 40203
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.7898208498954773,
      "learning_rate": 0.0005560407767195946,
      "loss": 2.2015,
      "step": 40204
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6842586398124695,
      "learning_rate": 0.0005560386449271543,
      "loss": 2.3155,
      "step": 40205
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.8012814521789551,
      "learning_rate": 0.0005560365130871117,
      "loss": 2.2949,
      "step": 40206
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.811134397983551,
      "learning_rate": 0.000556034381199467,
      "loss": 2.4407,
      "step": 40207
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6144810318946838,
      "learning_rate": 0.0005560322492642208,
      "loss": 2.1402,
      "step": 40208
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6622112989425659,
      "learning_rate": 0.0005560301172813732,
      "loss": 2.346,
      "step": 40209
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.7271400690078735,
      "learning_rate": 0.0005560279852509249,
      "loss": 2.114,
      "step": 40210
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.7540724277496338,
      "learning_rate": 0.0005560258531728761,
      "loss": 2.1601,
      "step": 40211
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6610909104347229,
      "learning_rate": 0.0005560237210472272,
      "loss": 2.2572,
      "step": 40212
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6676123738288879,
      "learning_rate": 0.0005560215888739787,
      "loss": 2.2111,
      "step": 40213
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6399754285812378,
      "learning_rate": 0.0005560194566531312,
      "loss": 2.204,
      "step": 40214
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6757937669754028,
      "learning_rate": 0.0005560173243846846,
      "loss": 2.2501,
      "step": 40215
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.682366669178009,
      "learning_rate": 0.0005560151920686395,
      "loss": 2.0435,
      "step": 40216
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6557901501655579,
      "learning_rate": 0.0005560130597049964,
      "loss": 2.4185,
      "step": 40217
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6679835915565491,
      "learning_rate": 0.0005560109272937557,
      "loss": 2.402,
      "step": 40218
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.7597160935401917,
      "learning_rate": 0.0005560087948349177,
      "loss": 2.3413,
      "step": 40219
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6972303986549377,
      "learning_rate": 0.0005560066623284828,
      "loss": 2.536,
      "step": 40220
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.5847466588020325,
      "learning_rate": 0.0005560045297744514,
      "loss": 2.2321,
      "step": 40221
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.8312222361564636,
      "learning_rate": 0.0005560023971728241,
      "loss": 2.254,
      "step": 40222
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.7940203547477722,
      "learning_rate": 0.000556000264523601,
      "loss": 2.5432,
      "step": 40223
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6318079233169556,
      "learning_rate": 0.0005559981318267826,
      "loss": 2.2396,
      "step": 40224
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6386604905128479,
      "learning_rate": 0.0005559959990823693,
      "loss": 2.3336,
      "step": 40225
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6814496517181396,
      "learning_rate": 0.0005559938662903616,
      "loss": 2.1365,
      "step": 40226
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.7808137536048889,
      "learning_rate": 0.0005559917334507598,
      "loss": 2.0775,
      "step": 40227
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.698681652545929,
      "learning_rate": 0.0005559896005635643,
      "loss": 2.2328,
      "step": 40228
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.7409719824790955,
      "learning_rate": 0.0005559874676287753,
      "loss": 2.2371,
      "step": 40229
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.7121326923370361,
      "learning_rate": 0.0005559853346463937,
      "loss": 2.4764,
      "step": 40230
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.7276303172111511,
      "learning_rate": 0.0005559832016164194,
      "loss": 2.3243,
      "step": 40231
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6242434978485107,
      "learning_rate": 0.0005559810685388531,
      "loss": 2.2247,
      "step": 40232
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.7371541261672974,
      "learning_rate": 0.000555978935413695,
      "loss": 2.1803,
      "step": 40233
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.9557065963745117,
      "learning_rate": 0.0005559768022409457,
      "loss": 2.2873,
      "step": 40234
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.7611674666404724,
      "learning_rate": 0.0005559746690206053,
      "loss": 2.2868,
      "step": 40235
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6748312711715698,
      "learning_rate": 0.0005559725357526745,
      "loss": 2.0762,
      "step": 40236
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.8106098771095276,
      "learning_rate": 0.0005559704024371535,
      "loss": 2.421,
      "step": 40237
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.7557589411735535,
      "learning_rate": 0.0005559682690740429,
      "loss": 2.3138,
      "step": 40238
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6584919691085815,
      "learning_rate": 0.0005559661356633428,
      "loss": 2.209,
      "step": 40239
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.7562587857246399,
      "learning_rate": 0.0005559640022050539,
      "loss": 2.3972,
      "step": 40240
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.8137063980102539,
      "learning_rate": 0.0005559618686991765,
      "loss": 2.194,
      "step": 40241
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.7124682068824768,
      "learning_rate": 0.0005559597351457108,
      "loss": 2.2564,
      "step": 40242
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.7054683566093445,
      "learning_rate": 0.0005559576015446574,
      "loss": 2.2884,
      "step": 40243
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.7136515974998474,
      "learning_rate": 0.0005559554678960168,
      "loss": 2.2715,
      "step": 40244
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6989582777023315,
      "learning_rate": 0.000555953334199789,
      "loss": 2.391,
      "step": 40245
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6485477685928345,
      "learning_rate": 0.0005559512004559748,
      "loss": 2.1245,
      "step": 40246
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.9131109118461609,
      "learning_rate": 0.0005559490666645743,
      "loss": 2.3877,
      "step": 40247
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6849687099456787,
      "learning_rate": 0.0005559469328255882,
      "loss": 2.1287,
      "step": 40248
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.7753490209579468,
      "learning_rate": 0.0005559447989390167,
      "loss": 2.3741,
      "step": 40249
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6531676054000854,
      "learning_rate": 0.0005559426650048602,
      "loss": 2.3529,
      "step": 40250
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.630579948425293,
      "learning_rate": 0.0005559405310231191,
      "loss": 2.1784,
      "step": 40251
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6983497142791748,
      "learning_rate": 0.0005559383969937939,
      "loss": 2.2003,
      "step": 40252
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6541600823402405,
      "learning_rate": 0.0005559362629168849,
      "loss": 2.305,
      "step": 40253
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6784106492996216,
      "learning_rate": 0.0005559341287923924,
      "loss": 2.3164,
      "step": 40254
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.635535717010498,
      "learning_rate": 0.0005559319946203171,
      "loss": 2.0891,
      "step": 40255
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.692304253578186,
      "learning_rate": 0.0005559298604006591,
      "loss": 2.3192,
      "step": 40256
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6379939317703247,
      "learning_rate": 0.0005559277261334189,
      "loss": 2.1325,
      "step": 40257
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6901984810829163,
      "learning_rate": 0.0005559255918185969,
      "loss": 2.2028,
      "step": 40258
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.709883451461792,
      "learning_rate": 0.0005559234574561936,
      "loss": 2.3167,
      "step": 40259
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6687037348747253,
      "learning_rate": 0.0005559213230462093,
      "loss": 2.2449,
      "step": 40260
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.7046892642974854,
      "learning_rate": 0.0005559191885886443,
      "loss": 2.2128,
      "step": 40261
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6689760088920593,
      "learning_rate": 0.000555917054083499,
      "loss": 2.2216,
      "step": 40262
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6876367330551147,
      "learning_rate": 0.0005559149195307741,
      "loss": 2.2394,
      "step": 40263
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.7287224531173706,
      "learning_rate": 0.0005559127849304697,
      "loss": 2.2922,
      "step": 40264
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6546453833580017,
      "learning_rate": 0.0005559106502825863,
      "loss": 2.2067,
      "step": 40265
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6894428133964539,
      "learning_rate": 0.0005559085155871242,
      "loss": 2.2595,
      "step": 40266
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.7288317680358887,
      "learning_rate": 0.0005559063808440839,
      "loss": 2.2004,
      "step": 40267
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.7668352723121643,
      "learning_rate": 0.0005559042460534659,
      "loss": 2.3499,
      "step": 40268
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.793522298336029,
      "learning_rate": 0.0005559021112152703,
      "loss": 2.4082,
      "step": 40269
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.7172695994377136,
      "learning_rate": 0.0005558999763294978,
      "loss": 2.2362,
      "step": 40270
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.8278971910476685,
      "learning_rate": 0.0005558978413961486,
      "loss": 2.1855,
      "step": 40271
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.7169864177703857,
      "learning_rate": 0.0005558957064152231,
      "loss": 2.4,
      "step": 40272
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6781932711601257,
      "learning_rate": 0.0005558935713867219,
      "loss": 2.2197,
      "step": 40273
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.7877388000488281,
      "learning_rate": 0.0005558914363106451,
      "loss": 2.1741,
      "step": 40274
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6062111258506775,
      "learning_rate": 0.0005558893011869934,
      "loss": 2.1493,
      "step": 40275
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6875113844871521,
      "learning_rate": 0.000555887166015767,
      "loss": 2.0584,
      "step": 40276
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.678308367729187,
      "learning_rate": 0.0005558850307969663,
      "loss": 2.3911,
      "step": 40277
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6969539523124695,
      "learning_rate": 0.0005558828955305918,
      "loss": 2.4235,
      "step": 40278
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.709435760974884,
      "learning_rate": 0.0005558807602166438,
      "loss": 2.191,
      "step": 40279
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.7369253039360046,
      "learning_rate": 0.0005558786248551227,
      "loss": 2.5239,
      "step": 40280
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6414477229118347,
      "learning_rate": 0.000555876489446029,
      "loss": 2.1453,
      "step": 40281
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6797871589660645,
      "learning_rate": 0.000555874353989363,
      "loss": 2.2906,
      "step": 40282
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6718193888664246,
      "learning_rate": 0.0005558722184851252,
      "loss": 2.2653,
      "step": 40283
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.675029456615448,
      "learning_rate": 0.0005558700829333158,
      "loss": 2.1801,
      "step": 40284
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6490569114685059,
      "learning_rate": 0.0005558679473339354,
      "loss": 2.1831,
      "step": 40285
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.788343071937561,
      "learning_rate": 0.0005558658116869843,
      "loss": 2.1742,
      "step": 40286
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.7103695273399353,
      "learning_rate": 0.0005558636759924629,
      "loss": 2.4928,
      "step": 40287
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6544703841209412,
      "learning_rate": 0.0005558615402503716,
      "loss": 2.0096,
      "step": 40288
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.7786440253257751,
      "learning_rate": 0.0005558594044607109,
      "loss": 2.4332,
      "step": 40289
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.8161718249320984,
      "learning_rate": 0.0005558572686234811,
      "loss": 2.2744,
      "step": 40290
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.885570228099823,
      "learning_rate": 0.0005558551327386825,
      "loss": 2.2787,
      "step": 40291
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6764482259750366,
      "learning_rate": 0.0005558529968063157,
      "loss": 2.3222,
      "step": 40292
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.7623985409736633,
      "learning_rate": 0.000555850860826381,
      "loss": 2.35,
      "step": 40293
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.710477352142334,
      "learning_rate": 0.0005558487247988786,
      "loss": 2.201,
      "step": 40294
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.7138403654098511,
      "learning_rate": 0.0005558465887238093,
      "loss": 2.1649,
      "step": 40295
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.7115699052810669,
      "learning_rate": 0.0005558444526011732,
      "loss": 2.2745,
      "step": 40296
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6424833536148071,
      "learning_rate": 0.0005558423164309708,
      "loss": 2.2946,
      "step": 40297
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6377993822097778,
      "learning_rate": 0.0005558401802132025,
      "loss": 2.2442,
      "step": 40298
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.7382377982139587,
      "learning_rate": 0.0005558380439478687,
      "loss": 2.3115,
      "step": 40299
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.7074511647224426,
      "learning_rate": 0.0005558359076349698,
      "loss": 2.3115,
      "step": 40300
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.7240904569625854,
      "learning_rate": 0.0005558337712745061,
      "loss": 2.1746,
      "step": 40301
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6966482996940613,
      "learning_rate": 0.0005558316348664782,
      "loss": 2.2805,
      "step": 40302
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.7966887354850769,
      "learning_rate": 0.0005558294984108862,
      "loss": 2.5693,
      "step": 40303
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.7755759358406067,
      "learning_rate": 0.0005558273619077308,
      "loss": 2.274,
      "step": 40304
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6645599603652954,
      "learning_rate": 0.0005558252253570122,
      "loss": 2.4119,
      "step": 40305
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6959390640258789,
      "learning_rate": 0.0005558230887587308,
      "loss": 2.4798,
      "step": 40306
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6689738035202026,
      "learning_rate": 0.0005558209521128872,
      "loss": 2.2232,
      "step": 40307
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.8138702511787415,
      "learning_rate": 0.0005558188154194815,
      "loss": 2.2288,
      "step": 40308
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.7567909955978394,
      "learning_rate": 0.0005558166786785144,
      "loss": 2.4902,
      "step": 40309
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.781407356262207,
      "learning_rate": 0.0005558145418899861,
      "loss": 2.2189,
      "step": 40310
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6363760828971863,
      "learning_rate": 0.000555812405053897,
      "loss": 2.1658,
      "step": 40311
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.7780405879020691,
      "learning_rate": 0.0005558102681702475,
      "loss": 2.4714,
      "step": 40312
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.7283036708831787,
      "learning_rate": 0.0005558081312390383,
      "loss": 2.3007,
      "step": 40313
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.7019343376159668,
      "learning_rate": 0.0005558059942602692,
      "loss": 2.4107,
      "step": 40314
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.644893229007721,
      "learning_rate": 0.0005558038572339412,
      "loss": 2.0748,
      "step": 40315
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.8480789065361023,
      "learning_rate": 0.0005558017201600543,
      "loss": 2.3674,
      "step": 40316
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.7604448795318604,
      "learning_rate": 0.0005557995830386091,
      "loss": 2.2042,
      "step": 40317
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.7216063141822815,
      "learning_rate": 0.0005557974458696059,
      "loss": 2.1713,
      "step": 40318
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6564291715621948,
      "learning_rate": 0.0005557953086530452,
      "loss": 2.2421,
      "step": 40319
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.733451247215271,
      "learning_rate": 0.0005557931713889272,
      "loss": 2.2426,
      "step": 40320
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.7808341383934021,
      "learning_rate": 0.0005557910340772524,
      "loss": 2.1947,
      "step": 40321
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.7709169983863831,
      "learning_rate": 0.0005557888967180212,
      "loss": 2.2054,
      "step": 40322
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6756733655929565,
      "learning_rate": 0.0005557867593112342,
      "loss": 2.1837,
      "step": 40323
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6909315586090088,
      "learning_rate": 0.0005557846218568915,
      "loss": 2.3483,
      "step": 40324
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.7358032464981079,
      "learning_rate": 0.0005557824843549936,
      "loss": 2.4544,
      "step": 40325
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.7306388020515442,
      "learning_rate": 0.000555780346805541,
      "loss": 2.2146,
      "step": 40326
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6394204497337341,
      "learning_rate": 0.0005557782092085339,
      "loss": 2.2175,
      "step": 40327
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.7863181829452515,
      "learning_rate": 0.0005557760715639729,
      "loss": 2.1377,
      "step": 40328
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.7225774526596069,
      "learning_rate": 0.0005557739338718582,
      "loss": 2.2399,
      "step": 40329
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.7107920050621033,
      "learning_rate": 0.0005557717961321903,
      "loss": 2.1889,
      "step": 40330
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.7635735869407654,
      "learning_rate": 0.0005557696583449697,
      "loss": 1.8047,
      "step": 40331
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.8910387754440308,
      "learning_rate": 0.0005557675205101967,
      "loss": 2.093,
      "step": 40332
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.7161161303520203,
      "learning_rate": 0.0005557653826278715,
      "loss": 2.2489,
      "step": 40333
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.8483116030693054,
      "learning_rate": 0.0005557632446979947,
      "loss": 1.923,
      "step": 40334
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.8041942715644836,
      "learning_rate": 0.0005557611067205668,
      "loss": 2.2681,
      "step": 40335
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6912711262702942,
      "learning_rate": 0.0005557589686955882,
      "loss": 2.5513,
      "step": 40336
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.7893434762954712,
      "learning_rate": 0.000555756830623059,
      "loss": 2.1729,
      "step": 40337
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.8817539215087891,
      "learning_rate": 0.0005557546925029799,
      "loss": 2.3727,
      "step": 40338
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.7668164372444153,
      "learning_rate": 0.000555752554335351,
      "loss": 2.2611,
      "step": 40339
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.7390257120132446,
      "learning_rate": 0.000555750416120173,
      "loss": 2.3313,
      "step": 40340
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.8390271067619324,
      "learning_rate": 0.0005557482778574462,
      "loss": 2.0934,
      "step": 40341
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6561506390571594,
      "learning_rate": 0.0005557461395471709,
      "loss": 2.1291,
      "step": 40342
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.7044817805290222,
      "learning_rate": 0.0005557440011893475,
      "loss": 2.2964,
      "step": 40343
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.7993459105491638,
      "learning_rate": 0.0005557418627839765,
      "loss": 2.1321,
      "step": 40344
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.865630030632019,
      "learning_rate": 0.0005557397243310583,
      "loss": 2.3652,
      "step": 40345
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.7275331020355225,
      "learning_rate": 0.0005557375858305934,
      "loss": 2.3066,
      "step": 40346
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.8959073424339294,
      "learning_rate": 0.0005557354472825819,
      "loss": 2.074,
      "step": 40347
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6501060128211975,
      "learning_rate": 0.0005557333086870243,
      "loss": 2.1642,
      "step": 40348
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.7721526026725769,
      "learning_rate": 0.0005557311700439212,
      "loss": 2.5567,
      "step": 40349
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.696609616279602,
      "learning_rate": 0.0005557290313532728,
      "loss": 2.3387,
      "step": 40350
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6417539119720459,
      "learning_rate": 0.0005557268926150794,
      "loss": 2.145,
      "step": 40351
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.7033727765083313,
      "learning_rate": 0.0005557247538293417,
      "loss": 2.2835,
      "step": 40352
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6977561116218567,
      "learning_rate": 0.0005557226149960599,
      "loss": 2.3515,
      "step": 40353
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6939007639884949,
      "learning_rate": 0.0005557204761152344,
      "loss": 2.1668,
      "step": 40354
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6772475242614746,
      "learning_rate": 0.0005557183371868656,
      "loss": 2.2967,
      "step": 40355
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6777262091636658,
      "learning_rate": 0.0005557161982109541,
      "loss": 2.276,
      "step": 40356
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.7187551856040955,
      "learning_rate": 0.0005557140591875,
      "loss": 2.2636,
      "step": 40357
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6431846022605896,
      "learning_rate": 0.0005557119201165039,
      "loss": 2.3551,
      "step": 40358
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6471856832504272,
      "learning_rate": 0.0005557097809979661,
      "loss": 2.1639,
      "step": 40359
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6868547201156616,
      "learning_rate": 0.000555707641831887,
      "loss": 2.1993,
      "step": 40360
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.7597500085830688,
      "learning_rate": 0.0005557055026182671,
      "loss": 2.3951,
      "step": 40361
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.7522802352905273,
      "learning_rate": 0.0005557033633571066,
      "loss": 2.0547,
      "step": 40362
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6720454692840576,
      "learning_rate": 0.000555701224048406,
      "loss": 2.0936,
      "step": 40363
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.8609398603439331,
      "learning_rate": 0.0005556990846921658,
      "loss": 2.3021,
      "step": 40364
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6666563153266907,
      "learning_rate": 0.0005556969452883864,
      "loss": 2.1998,
      "step": 40365
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.655023992061615,
      "learning_rate": 0.000555694805837068,
      "loss": 2.2908,
      "step": 40366
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6741692423820496,
      "learning_rate": 0.0005556926663382112,
      "loss": 2.1731,
      "step": 40367
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6855180263519287,
      "learning_rate": 0.0005556905267918162,
      "loss": 2.0512,
      "step": 40368
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.7392942309379578,
      "learning_rate": 0.0005556883871978835,
      "loss": 2.5293,
      "step": 40369
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6574175953865051,
      "learning_rate": 0.0005556862475564135,
      "loss": 2.2178,
      "step": 40370
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6454691886901855,
      "learning_rate": 0.0005556841078674067,
      "loss": 2.3213,
      "step": 40371
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.75602787733078,
      "learning_rate": 0.0005556819681308633,
      "loss": 2.2688,
      "step": 40372
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6442784070968628,
      "learning_rate": 0.0005556798283467839,
      "loss": 2.2349,
      "step": 40373
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6700108647346497,
      "learning_rate": 0.0005556776885151687,
      "loss": 2.3038,
      "step": 40374
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6953100562095642,
      "learning_rate": 0.0005556755486360182,
      "loss": 2.4022,
      "step": 40375
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6434609889984131,
      "learning_rate": 0.0005556734087093328,
      "loss": 2.092,
      "step": 40376
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6606646776199341,
      "learning_rate": 0.0005556712687351129,
      "loss": 2.2948,
      "step": 40377
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.633983850479126,
      "learning_rate": 0.0005556691287133588,
      "loss": 2.1794,
      "step": 40378
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6928922533988953,
      "learning_rate": 0.0005556669886440711,
      "loss": 2.3911,
      "step": 40379
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6525603532791138,
      "learning_rate": 0.0005556648485272501,
      "loss": 2.2431,
      "step": 40380
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.9624375700950623,
      "learning_rate": 0.0005556627083628962,
      "loss": 2.3451,
      "step": 40381
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.8350123763084412,
      "learning_rate": 0.0005556605681510096,
      "loss": 2.4301,
      "step": 40382
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6888072490692139,
      "learning_rate": 0.000555658427891591,
      "loss": 2.4311,
      "step": 40383
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.7453694939613342,
      "learning_rate": 0.0005556562875846406,
      "loss": 2.1872,
      "step": 40384
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.801685094833374,
      "learning_rate": 0.000555654147230159,
      "loss": 2.0233,
      "step": 40385
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.7076000571250916,
      "learning_rate": 0.0005556520068281464,
      "loss": 1.9922,
      "step": 40386
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.7986331582069397,
      "learning_rate": 0.0005556498663786033,
      "loss": 2.303,
      "step": 40387
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.7632180452346802,
      "learning_rate": 0.00055564772588153,
      "loss": 2.1243,
      "step": 40388
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.7440948486328125,
      "learning_rate": 0.0005556455853369268,
      "loss": 2.2936,
      "step": 40389
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.7737279534339905,
      "learning_rate": 0.0005556434447447945,
      "loss": 2.4016,
      "step": 40390
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.9981600642204285,
      "learning_rate": 0.0005556413041051333,
      "loss": 2.1369,
      "step": 40391
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6655021905899048,
      "learning_rate": 0.0005556391634179434,
      "loss": 2.106,
      "step": 40392
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6220017075538635,
      "learning_rate": 0.0005556370226832254,
      "loss": 2.1514,
      "step": 40393
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.978492259979248,
      "learning_rate": 0.0005556348819009796,
      "loss": 2.1346,
      "step": 40394
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.7661998867988586,
      "learning_rate": 0.0005556327410712065,
      "loss": 2.218,
      "step": 40395
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6436378955841064,
      "learning_rate": 0.0005556306001939065,
      "loss": 2.0726,
      "step": 40396
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.896906852722168,
      "learning_rate": 0.00055562845926908,
      "loss": 2.2299,
      "step": 40397
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.914182186126709,
      "learning_rate": 0.0005556263182967272,
      "loss": 2.336,
      "step": 40398
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6475067734718323,
      "learning_rate": 0.0005556241772768486,
      "loss": 2.1961,
      "step": 40399
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.8630412817001343,
      "learning_rate": 0.0005556220362094448,
      "loss": 2.2657,
      "step": 40400
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.9020551443099976,
      "learning_rate": 0.0005556198950945159,
      "loss": 2.5113,
      "step": 40401
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.7212061882019043,
      "learning_rate": 0.0005556177539320625,
      "loss": 2.6354,
      "step": 40402
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.8482975959777832,
      "learning_rate": 0.000555615612722085,
      "loss": 2.3385,
      "step": 40403
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.8904997110366821,
      "learning_rate": 0.0005556134714645837,
      "loss": 2.3284,
      "step": 40404
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.7232847809791565,
      "learning_rate": 0.0005556113301595589,
      "loss": 2.3235,
      "step": 40405
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6767598986625671,
      "learning_rate": 0.0005556091888070112,
      "loss": 2.3156,
      "step": 40406
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.7018606662750244,
      "learning_rate": 0.000555607047406941,
      "loss": 2.1551,
      "step": 40407
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6906190514564514,
      "learning_rate": 0.0005556049059593486,
      "loss": 2.5307,
      "step": 40408
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.7040809988975525,
      "learning_rate": 0.0005556027644642344,
      "loss": 2.2028,
      "step": 40409
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.7975486516952515,
      "learning_rate": 0.0005556006229215989,
      "loss": 2.0615,
      "step": 40410
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.8459628224372864,
      "learning_rate": 0.0005555984813314423,
      "loss": 2.5102,
      "step": 40411
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.7085611820220947,
      "learning_rate": 0.0005555963396937651,
      "loss": 2.1654,
      "step": 40412
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.7257030606269836,
      "learning_rate": 0.0005555941980085678,
      "loss": 2.3142,
      "step": 40413
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.611481785774231,
      "learning_rate": 0.0005555920562758507,
      "loss": 2.2961,
      "step": 40414
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.7578496336936951,
      "learning_rate": 0.0005555899144956142,
      "loss": 2.1835,
      "step": 40415
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6766419410705566,
      "learning_rate": 0.0005555877726678587,
      "loss": 2.1705,
      "step": 40416
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6511169075965881,
      "learning_rate": 0.0005555856307925847,
      "loss": 2.3041,
      "step": 40417
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6779433488845825,
      "learning_rate": 0.0005555834888697924,
      "loss": 2.4977,
      "step": 40418
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6828387379646301,
      "learning_rate": 0.0005555813468994822,
      "loss": 2.2721,
      "step": 40419
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6496050953865051,
      "learning_rate": 0.0005555792048816549,
      "loss": 2.1663,
      "step": 40420
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6891112327575684,
      "learning_rate": 0.0005555770628163103,
      "loss": 2.4895,
      "step": 40421
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6565261483192444,
      "learning_rate": 0.0005555749207034493,
      "loss": 2.1071,
      "step": 40422
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6509358882904053,
      "learning_rate": 0.000555572778543072,
      "loss": 2.2272,
      "step": 40423
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6742473840713501,
      "learning_rate": 0.0005555706363351789,
      "loss": 2.1346,
      "step": 40424
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.9206438660621643,
      "learning_rate": 0.0005555684940797706,
      "loss": 2.2174,
      "step": 40425
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6871518492698669,
      "learning_rate": 0.0005555663517768471,
      "loss": 2.2179,
      "step": 40426
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6635408401489258,
      "learning_rate": 0.000555564209426409,
      "loss": 2.2594,
      "step": 40427
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6979947090148926,
      "learning_rate": 0.0005555620670284567,
      "loss": 2.3444,
      "step": 40428
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.7004439830780029,
      "learning_rate": 0.0005555599245829905,
      "loss": 2.4205,
      "step": 40429
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.7266519069671631,
      "learning_rate": 0.000555557782090011,
      "loss": 2.2212,
      "step": 40430
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.7208753824234009,
      "learning_rate": 0.0005555556395495184,
      "loss": 2.3656,
      "step": 40431
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6539238095283508,
      "learning_rate": 0.0005555534969615133,
      "loss": 2.0385,
      "step": 40432
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6990638971328735,
      "learning_rate": 0.0005555513543259959,
      "loss": 2.4688,
      "step": 40433
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.8046720027923584,
      "learning_rate": 0.0005555492116429667,
      "loss": 2.1176,
      "step": 40434
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.7301727533340454,
      "learning_rate": 0.0005555470689124261,
      "loss": 2.5669,
      "step": 40435
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.7378323078155518,
      "learning_rate": 0.0005555449261343744,
      "loss": 2.2371,
      "step": 40436
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.7275251746177673,
      "learning_rate": 0.0005555427833088121,
      "loss": 2.1639,
      "step": 40437
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6178264617919922,
      "learning_rate": 0.0005555406404357397,
      "loss": 2.2051,
      "step": 40438
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.7833581566810608,
      "learning_rate": 0.0005555384975151574,
      "loss": 2.357,
      "step": 40439
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6671722531318665,
      "learning_rate": 0.0005555363545470656,
      "loss": 2.2237,
      "step": 40440
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.730074942111969,
      "learning_rate": 0.0005555342115314648,
      "loss": 2.245,
      "step": 40441
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6723818182945251,
      "learning_rate": 0.0005555320684683554,
      "loss": 2.2316,
      "step": 40442
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.7274028658866882,
      "learning_rate": 0.0005555299253577377,
      "loss": 2.2215,
      "step": 40443
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.7469397783279419,
      "learning_rate": 0.0005555277821996123,
      "loss": 2.4976,
      "step": 40444
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6389766335487366,
      "learning_rate": 0.0005555256389939793,
      "loss": 2.2904,
      "step": 40445
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.8146456480026245,
      "learning_rate": 0.0005555234957408394,
      "loss": 2.2573,
      "step": 40446
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.663468599319458,
      "learning_rate": 0.0005555213524401927,
      "loss": 2.279,
      "step": 40447
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.7358670830726624,
      "learning_rate": 0.0005555192090920399,
      "loss": 2.4036,
      "step": 40448
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.7624762058258057,
      "learning_rate": 0.0005555170656963811,
      "loss": 2.2275,
      "step": 40449
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.8038187026977539,
      "learning_rate": 0.000555514922253217,
      "loss": 2.4001,
      "step": 40450
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.9054683446884155,
      "learning_rate": 0.0005555127787625478,
      "loss": 2.3894,
      "step": 40451
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.7073875665664673,
      "learning_rate": 0.0005555106352243739,
      "loss": 2.1393,
      "step": 40452
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6646339893341064,
      "learning_rate": 0.0005555084916386958,
      "loss": 2.3211,
      "step": 40453
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.8084825873374939,
      "learning_rate": 0.0005555063480055139,
      "loss": 2.2607,
      "step": 40454
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6752246022224426,
      "learning_rate": 0.0005555042043248285,
      "loss": 2.2109,
      "step": 40455
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.7308364510536194,
      "learning_rate": 0.00055550206059664,
      "loss": 2.2098,
      "step": 40456
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.7547129392623901,
      "learning_rate": 0.0005554999168209489,
      "loss": 2.1006,
      "step": 40457
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.7132425904273987,
      "learning_rate": 0.0005554977729977556,
      "loss": 2.316,
      "step": 40458
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.8674628734588623,
      "learning_rate": 0.0005554956291270604,
      "loss": 2.2439,
      "step": 40459
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6565993428230286,
      "learning_rate": 0.0005554934852088636,
      "loss": 2.318,
      "step": 40460
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.7318620681762695,
      "learning_rate": 0.000555491341243166,
      "loss": 2.3953,
      "step": 40461
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.7503175735473633,
      "learning_rate": 0.0005554891972299675,
      "loss": 2.2393,
      "step": 40462
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6837207674980164,
      "learning_rate": 0.0005554870531692688,
      "loss": 2.3921,
      "step": 40463
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.7696427702903748,
      "learning_rate": 0.0005554849090610703,
      "loss": 2.259,
      "step": 40464
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.7050545811653137,
      "learning_rate": 0.0005554827649053723,
      "loss": 2.3699,
      "step": 40465
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6639383435249329,
      "learning_rate": 0.0005554806207021752,
      "loss": 2.4272,
      "step": 40466
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.8454093933105469,
      "learning_rate": 0.0005554784764514795,
      "loss": 2.2599,
      "step": 40467
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6952928304672241,
      "learning_rate": 0.0005554763321532854,
      "loss": 2.1597,
      "step": 40468
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.7106459140777588,
      "learning_rate": 0.0005554741878075936,
      "loss": 2.1299,
      "step": 40469
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6235015988349915,
      "learning_rate": 0.0005554720434144042,
      "loss": 2.2962,
      "step": 40470
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6653393507003784,
      "learning_rate": 0.0005554698989737177,
      "loss": 2.4765,
      "step": 40471
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.7034355998039246,
      "learning_rate": 0.0005554677544855347,
      "loss": 2.1446,
      "step": 40472
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.7216731309890747,
      "learning_rate": 0.0005554656099498553,
      "loss": 2.2694,
      "step": 40473
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6627145409584045,
      "learning_rate": 0.0005554634653666799,
      "loss": 2.3551,
      "step": 40474
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6670940518379211,
      "learning_rate": 0.0005554613207360092,
      "loss": 2.0164,
      "step": 40475
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6928395628929138,
      "learning_rate": 0.0005554591760578434,
      "loss": 2.349,
      "step": 40476
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6560922861099243,
      "learning_rate": 0.0005554570313321829,
      "loss": 2.2811,
      "step": 40477
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6936256289482117,
      "learning_rate": 0.000555454886559028,
      "loss": 2.1581,
      "step": 40478
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.7154948115348816,
      "learning_rate": 0.0005554527417383794,
      "loss": 2.2382,
      "step": 40479
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6632410287857056,
      "learning_rate": 0.0005554505968702372,
      "loss": 2.2712,
      "step": 40480
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.7036781311035156,
      "learning_rate": 0.0005554484519546019,
      "loss": 2.3367,
      "step": 40481
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6578360199928284,
      "learning_rate": 0.000555446306991474,
      "loss": 2.3417,
      "step": 40482
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6711034178733826,
      "learning_rate": 0.0005554441619808537,
      "loss": 2.2936,
      "step": 40483
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.649608314037323,
      "learning_rate": 0.0005554420169227416,
      "loss": 2.0014,
      "step": 40484
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6494495868682861,
      "learning_rate": 0.000555439871817138,
      "loss": 2.2907,
      "step": 40485
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6780592799186707,
      "learning_rate": 0.0005554377266640433,
      "loss": 2.4374,
      "step": 40486
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.678519606590271,
      "learning_rate": 0.0005554355814634579,
      "loss": 2.3595,
      "step": 40487
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.7074331045150757,
      "learning_rate": 0.0005554334362153821,
      "loss": 2.2984,
      "step": 40488
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.7719473838806152,
      "learning_rate": 0.0005554312909198166,
      "loss": 2.2604,
      "step": 40489
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.7078140377998352,
      "learning_rate": 0.0005554291455767614,
      "loss": 2.2075,
      "step": 40490
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6672765016555786,
      "learning_rate": 0.0005554270001862172,
      "loss": 2.2751,
      "step": 40491
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6977410316467285,
      "learning_rate": 0.0005554248547481843,
      "loss": 2.3787,
      "step": 40492
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6806853413581848,
      "learning_rate": 0.0005554227092626631,
      "loss": 2.1511,
      "step": 40493
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6298106908798218,
      "learning_rate": 0.0005554205637296539,
      "loss": 2.3375,
      "step": 40494
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6627689599990845,
      "learning_rate": 0.0005554184181491573,
      "loss": 2.3098,
      "step": 40495
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.7273954749107361,
      "learning_rate": 0.0005554162725211737,
      "loss": 2.3927,
      "step": 40496
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.7247177958488464,
      "learning_rate": 0.0005554141268457032,
      "loss": 2.0508,
      "step": 40497
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.7142760157585144,
      "learning_rate": 0.0005554119811227465,
      "loss": 2.3794,
      "step": 40498
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6809550523757935,
      "learning_rate": 0.0005554098353523038,
      "loss": 2.2922,
      "step": 40499
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.7144165635108948,
      "learning_rate": 0.0005554076895343757,
      "loss": 2.1028,
      "step": 40500
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6886250376701355,
      "learning_rate": 0.0005554055436689624,
      "loss": 2.364,
      "step": 40501
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6582044959068298,
      "learning_rate": 0.0005554033977560644,
      "loss": 2.3153,
      "step": 40502
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6755949854850769,
      "learning_rate": 0.0005554012517956821,
      "loss": 2.2526,
      "step": 40503
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.7927257418632507,
      "learning_rate": 0.0005553991057878159,
      "loss": 2.2057,
      "step": 40504
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6242644786834717,
      "learning_rate": 0.0005553969597324662,
      "loss": 2.2447,
      "step": 40505
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.731945276260376,
      "learning_rate": 0.0005553948136296334,
      "loss": 2.1713,
      "step": 40506
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6204497814178467,
      "learning_rate": 0.0005553926674793178,
      "loss": 2.2118,
      "step": 40507
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.8039458990097046,
      "learning_rate": 0.0005553905212815199,
      "loss": 2.2185,
      "step": 40508
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.7312890291213989,
      "learning_rate": 0.0005553883750362402,
      "loss": 2.2576,
      "step": 40509
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6927792429924011,
      "learning_rate": 0.0005553862287434787,
      "loss": 2.1853,
      "step": 40510
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.9047132730484009,
      "learning_rate": 0.0005553840824032365,
      "loss": 2.1757,
      "step": 40511
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.669715166091919,
      "learning_rate": 0.0005553819360155132,
      "loss": 2.3463,
      "step": 40512
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.7234418392181396,
      "learning_rate": 0.0005553797895803098,
      "loss": 2.2847,
      "step": 40513
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.8274936676025391,
      "learning_rate": 0.0005553776430976264,
      "loss": 2.303,
      "step": 40514
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6787160634994507,
      "learning_rate": 0.0005553754965674635,
      "loss": 2.2474,
      "step": 40515
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.7949504256248474,
      "learning_rate": 0.0005553733499898214,
      "loss": 2.3551,
      "step": 40516
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6597537994384766,
      "learning_rate": 0.0005553712033647007,
      "loss": 2.3392,
      "step": 40517
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6317415833473206,
      "learning_rate": 0.0005553690566921015,
      "loss": 2.2666,
      "step": 40518
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.7216281294822693,
      "learning_rate": 0.0005553669099720246,
      "loss": 2.1766,
      "step": 40519
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.688869297504425,
      "learning_rate": 0.00055536476320447,
      "loss": 2.4596,
      "step": 40520
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6953223347663879,
      "learning_rate": 0.0005553626163894382,
      "loss": 2.2029,
      "step": 40521
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.69504314661026,
      "learning_rate": 0.0005553604695269299,
      "loss": 2.3442,
      "step": 40522
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6485916972160339,
      "learning_rate": 0.000555358322616945,
      "loss": 2.2316,
      "step": 40523
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.748028576374054,
      "learning_rate": 0.0005553561756594845,
      "loss": 2.3514,
      "step": 40524
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.8088598251342773,
      "learning_rate": 0.0005553540286545482,
      "loss": 2.0834,
      "step": 40525
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.7266623973846436,
      "learning_rate": 0.0005553518816021369,
      "loss": 2.5338,
      "step": 40526
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6900333762168884,
      "learning_rate": 0.0005553497345022508,
      "loss": 2.0252,
      "step": 40527
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.7745590209960938,
      "learning_rate": 0.0005553475873548904,
      "loss": 2.2657,
      "step": 40528
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.626142144203186,
      "learning_rate": 0.000555345440160056,
      "loss": 2.1614,
      "step": 40529
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.7073546648025513,
      "learning_rate": 0.0005553432929177481,
      "loss": 2.1839,
      "step": 40530
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.7068687081336975,
      "learning_rate": 0.0005553411456279671,
      "loss": 2.2132,
      "step": 40531
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.7122154831886292,
      "learning_rate": 0.0005553389982907133,
      "loss": 2.0311,
      "step": 40532
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.7176229357719421,
      "learning_rate": 0.0005553368509059872,
      "loss": 2.5009,
      "step": 40533
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6662561893463135,
      "learning_rate": 0.0005553347034737891,
      "loss": 2.7137,
      "step": 40534
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.7320588827133179,
      "learning_rate": 0.0005553325559941195,
      "loss": 2.5468,
      "step": 40535
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.8154261708259583,
      "learning_rate": 0.0005553304084669788,
      "loss": 2.2911,
      "step": 40536
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.7278373837471008,
      "learning_rate": 0.0005553282608923673,
      "loss": 2.2452,
      "step": 40537
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.7912527322769165,
      "learning_rate": 0.0005553261132702855,
      "loss": 2.1714,
      "step": 40538
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.7730419635772705,
      "learning_rate": 0.0005553239656007337,
      "loss": 2.4852,
      "step": 40539
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.7474532127380371,
      "learning_rate": 0.0005553218178837125,
      "loss": 2.3217,
      "step": 40540
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.7406105399131775,
      "learning_rate": 0.000555319670119222,
      "loss": 2.3056,
      "step": 40541
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.8708299398422241,
      "learning_rate": 0.0005553175223072628,
      "loss": 2.363,
      "step": 40542
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.8901105523109436,
      "learning_rate": 0.0005553153744478353,
      "loss": 2.0814,
      "step": 40543
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6988227963447571,
      "learning_rate": 0.0005553132265409397,
      "loss": 2.2587,
      "step": 40544
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.7993443012237549,
      "learning_rate": 0.0005553110785865768,
      "loss": 2.3168,
      "step": 40545
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.0198073387145996,
      "learning_rate": 0.0005553089305847464,
      "loss": 2.0777,
      "step": 40546
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.850067675113678,
      "learning_rate": 0.0005553067825354495,
      "loss": 2.2985,
      "step": 40547
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.806035041809082,
      "learning_rate": 0.0005553046344386862,
      "loss": 2.2525,
      "step": 40548
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.0737262964248657,
      "learning_rate": 0.0005553024862944571,
      "loss": 2.2785,
      "step": 40549
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.9788966774940491,
      "learning_rate": 0.0005553003381027623,
      "loss": 2.2722,
      "step": 40550
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.5981993079185486,
      "learning_rate": 0.0005552981898636022,
      "loss": 2.2478,
      "step": 40551
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.7554683685302734,
      "learning_rate": 0.0005552960415769776,
      "loss": 2.2567,
      "step": 40552
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.0027624368667603,
      "learning_rate": 0.0005552938932428885,
      "loss": 2.2481,
      "step": 40553
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.7816415429115295,
      "learning_rate": 0.0005552917448613356,
      "loss": 2.3258,
      "step": 40554
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.7629172205924988,
      "learning_rate": 0.0005552895964323189,
      "loss": 2.2437,
      "step": 40555
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.8490954637527466,
      "learning_rate": 0.0005552874479558392,
      "loss": 2.2586,
      "step": 40556
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.8845479488372803,
      "learning_rate": 0.0005552852994318968,
      "loss": 2.3968,
      "step": 40557
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6974360346794128,
      "learning_rate": 0.0005552831508604919,
      "loss": 2.2826,
      "step": 40558
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6446869969367981,
      "learning_rate": 0.0005552810022416252,
      "loss": 2.3058,
      "step": 40559
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.8418254256248474,
      "learning_rate": 0.0005552788535752968,
      "loss": 2.277,
      "step": 40560
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.7264614105224609,
      "learning_rate": 0.0005552767048615073,
      "loss": 2.3215,
      "step": 40561
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.7096428275108337,
      "learning_rate": 0.0005552745561002571,
      "loss": 2.3434,
      "step": 40562
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6860139966011047,
      "learning_rate": 0.0005552724072915464,
      "loss": 2.2437,
      "step": 40563
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.7206318378448486,
      "learning_rate": 0.0005552702584353759,
      "loss": 2.4894,
      "step": 40564
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.7306801676750183,
      "learning_rate": 0.0005552681095317457,
      "loss": 2.2336,
      "step": 40565
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.7202398777008057,
      "learning_rate": 0.0005552659605806565,
      "loss": 2.1547,
      "step": 40566
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6809082627296448,
      "learning_rate": 0.0005552638115821084,
      "loss": 2.3797,
      "step": 40567
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.7609078288078308,
      "learning_rate": 0.000555261662536102,
      "loss": 2.3255,
      "step": 40568
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6917853951454163,
      "learning_rate": 0.0005552595134426376,
      "loss": 2.2798,
      "step": 40569
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.7209949493408203,
      "learning_rate": 0.0005552573643017156,
      "loss": 2.3109,
      "step": 40570
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6575571894645691,
      "learning_rate": 0.0005552552151133365,
      "loss": 2.3156,
      "step": 40571
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.7444162368774414,
      "learning_rate": 0.0005552530658775007,
      "loss": 2.27,
      "step": 40572
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.7248349785804749,
      "learning_rate": 0.0005552509165942085,
      "loss": 2.2853,
      "step": 40573
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.666204035282135,
      "learning_rate": 0.0005552487672634603,
      "loss": 2.1412,
      "step": 40574
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.7095512747764587,
      "learning_rate": 0.0005552466178852565,
      "loss": 2.2395,
      "step": 40575
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.7207381725311279,
      "learning_rate": 0.0005552444684595976,
      "loss": 2.2367,
      "step": 40576
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6735413074493408,
      "learning_rate": 0.0005552423189864839,
      "loss": 2.198,
      "step": 40577
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6650813817977905,
      "learning_rate": 0.0005552401694659159,
      "loss": 2.0323,
      "step": 40578
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.7110785841941833,
      "learning_rate": 0.0005552380198978939,
      "loss": 2.2448,
      "step": 40579
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.7060825228691101,
      "learning_rate": 0.0005552358702824183,
      "loss": 2.3907,
      "step": 40580
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6709717512130737,
      "learning_rate": 0.0005552337206194896,
      "loss": 2.1562,
      "step": 40581
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.7306498289108276,
      "learning_rate": 0.000555231570909108,
      "loss": 2.3192,
      "step": 40582
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.7889584898948669,
      "learning_rate": 0.0005552294211512741,
      "loss": 2.1923,
      "step": 40583
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6473093032836914,
      "learning_rate": 0.0005552272713459884,
      "loss": 2.4464,
      "step": 40584
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.7352566123008728,
      "learning_rate": 0.0005552251214932509,
      "loss": 2.1314,
      "step": 40585
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6891076564788818,
      "learning_rate": 0.0005552229715930623,
      "loss": 2.3799,
      "step": 40586
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.7340739965438843,
      "learning_rate": 0.000555220821645423,
      "loss": 2.1503,
      "step": 40587
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6355089545249939,
      "learning_rate": 0.0005552186716503332,
      "loss": 2.0923,
      "step": 40588
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6967906951904297,
      "learning_rate": 0.0005552165216077935,
      "loss": 2.2935,
      "step": 40589
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.7759230732917786,
      "learning_rate": 0.0005552143715178043,
      "loss": 2.1671,
      "step": 40590
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6544115543365479,
      "learning_rate": 0.0005552122213803658,
      "loss": 2.0875,
      "step": 40591
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.7913029193878174,
      "learning_rate": 0.0005552100711954786,
      "loss": 2.3842,
      "step": 40592
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6860687732696533,
      "learning_rate": 0.000555207920963143,
      "loss": 2.078,
      "step": 40593
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6931827068328857,
      "learning_rate": 0.0005552057706833595,
      "loss": 2.2989,
      "step": 40594
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.7118539810180664,
      "learning_rate": 0.0005552036203561284,
      "loss": 2.4097,
      "step": 40595
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6687862873077393,
      "learning_rate": 0.0005552014699814501,
      "loss": 2.3958,
      "step": 40596
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6576436161994934,
      "learning_rate": 0.0005551993195593251,
      "loss": 2.4426,
      "step": 40597
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.7150963544845581,
      "learning_rate": 0.0005551971690897537,
      "loss": 2.1905,
      "step": 40598
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.747573971748352,
      "learning_rate": 0.0005551950185727363,
      "loss": 2.1636,
      "step": 40599
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.7573517560958862,
      "learning_rate": 0.0005551928680082733,
      "loss": 2.2453,
      "step": 40600
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6874925494194031,
      "learning_rate": 0.0005551907173963653,
      "loss": 2.1564,
      "step": 40601
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6453484892845154,
      "learning_rate": 0.0005551885667370122,
      "loss": 2.4482,
      "step": 40602
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6349872350692749,
      "learning_rate": 0.0005551864160302151,
      "loss": 2.1206,
      "step": 40603
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6738230586051941,
      "learning_rate": 0.0005551842652759738,
      "loss": 2.1904,
      "step": 40604
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.7426722049713135,
      "learning_rate": 0.000555182114474289,
      "loss": 2.2916,
      "step": 40605
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6697371602058411,
      "learning_rate": 0.0005551799636251609,
      "loss": 2.3253,
      "step": 40606
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6829714775085449,
      "learning_rate": 0.0005551778127285903,
      "loss": 2.3578,
      "step": 40607
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6366955041885376,
      "learning_rate": 0.0005551756617845771,
      "loss": 2.1757,
      "step": 40608
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.611878514289856,
      "learning_rate": 0.0005551735107931219,
      "loss": 2.3462,
      "step": 40609
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.7441445589065552,
      "learning_rate": 0.0005551713597542254,
      "loss": 2.5361,
      "step": 40610
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6916086077690125,
      "learning_rate": 0.0005551692086678874,
      "loss": 2.288,
      "step": 40611
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.7050301432609558,
      "learning_rate": 0.0005551670575341088,
      "loss": 2.0426,
      "step": 40612
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6394495368003845,
      "learning_rate": 0.0005551649063528898,
      "loss": 2.3935,
      "step": 40613
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.7404155135154724,
      "learning_rate": 0.0005551627551242308,
      "loss": 2.3822,
      "step": 40614
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.640802264213562,
      "learning_rate": 0.0005551606038481324,
      "loss": 2.3731,
      "step": 40615
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6973078846931458,
      "learning_rate": 0.0005551584525245947,
      "loss": 2.1913,
      "step": 40616
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6627985239028931,
      "learning_rate": 0.0005551563011536181,
      "loss": 2.2341,
      "step": 40617
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6872201561927795,
      "learning_rate": 0.0005551541497352033,
      "loss": 2.0313,
      "step": 40618
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6803763508796692,
      "learning_rate": 0.0005551519982693505,
      "loss": 2.3222,
      "step": 40619
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.7118785381317139,
      "learning_rate": 0.00055514984675606,
      "loss": 2.1007,
      "step": 40620
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.7068328261375427,
      "learning_rate": 0.0005551476951953325,
      "loss": 2.1784,
      "step": 40621
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.650010883808136,
      "learning_rate": 0.0005551455435871681,
      "loss": 2.1607,
      "step": 40622
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.7102729678153992,
      "learning_rate": 0.0005551433919315673,
      "loss": 2.2726,
      "step": 40623
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.9504648447036743,
      "learning_rate": 0.0005551412402285306,
      "loss": 2.291,
      "step": 40624
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.742310643196106,
      "learning_rate": 0.0005551390884780584,
      "loss": 2.2979,
      "step": 40625
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.8296348452568054,
      "learning_rate": 0.000555136936680151,
      "loss": 2.2304,
      "step": 40626
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.9636280536651611,
      "learning_rate": 0.0005551347848348088,
      "loss": 2.4445,
      "step": 40627
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6983606219291687,
      "learning_rate": 0.0005551326329420322,
      "loss": 2.4163,
      "step": 40628
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.7039216160774231,
      "learning_rate": 0.0005551304810018216,
      "loss": 2.3413,
      "step": 40629
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.7566622495651245,
      "learning_rate": 0.0005551283290141774,
      "loss": 2.1766,
      "step": 40630
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.7602301836013794,
      "learning_rate": 0.0005551261769791002,
      "loss": 2.2936,
      "step": 40631
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6441393494606018,
      "learning_rate": 0.0005551240248965902,
      "loss": 2.2456,
      "step": 40632
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.7850387096405029,
      "learning_rate": 0.0005551218727666476,
      "loss": 2.2575,
      "step": 40633
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.7300142049789429,
      "learning_rate": 0.0005551197205892732,
      "loss": 2.1901,
      "step": 40634
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.663773775100708,
      "learning_rate": 0.0005551175683644672,
      "loss": 2.419,
      "step": 40635
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.7024748921394348,
      "learning_rate": 0.00055511541609223,
      "loss": 2.1392,
      "step": 40636
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6278901100158691,
      "learning_rate": 0.0005551132637725622,
      "loss": 2.195,
      "step": 40637
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.8721565008163452,
      "learning_rate": 0.0005551111114054638,
      "loss": 2.1517,
      "step": 40638
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6921263337135315,
      "learning_rate": 0.0005551089589909355,
      "loss": 2.0796,
      "step": 40639
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6495863199234009,
      "learning_rate": 0.0005551068065289776,
      "loss": 2.1791,
      "step": 40640
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6871827840805054,
      "learning_rate": 0.0005551046540195906,
      "loss": 2.4439,
      "step": 40641
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.7649796009063721,
      "learning_rate": 0.0005551025014627749,
      "loss": 2.1379,
      "step": 40642
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6572553515434265,
      "learning_rate": 0.0005551003488585307,
      "loss": 2.2625,
      "step": 40643
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.7385649681091309,
      "learning_rate": 0.0005550981962068586,
      "loss": 2.3374,
      "step": 40644
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.7986342310905457,
      "learning_rate": 0.0005550960435077589,
      "loss": 2.1231,
      "step": 40645
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6407912969589233,
      "learning_rate": 0.000555093890761232,
      "loss": 2.232,
      "step": 40646
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6927366852760315,
      "learning_rate": 0.0005550917379672784,
      "loss": 2.1012,
      "step": 40647
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.8056381344795227,
      "learning_rate": 0.0005550895851258983,
      "loss": 2.265,
      "step": 40648
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6640526056289673,
      "learning_rate": 0.0005550874322370924,
      "loss": 2.4126,
      "step": 40649
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6720440983772278,
      "learning_rate": 0.0005550852793008608,
      "loss": 2.39,
      "step": 40650
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.7196059226989746,
      "learning_rate": 0.000555083126317204,
      "loss": 2.2089,
      "step": 40651
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.7524423003196716,
      "learning_rate": 0.0005550809732861226,
      "loss": 2.1856,
      "step": 40652
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6697684526443481,
      "learning_rate": 0.0005550788202076168,
      "loss": 2.3649,
      "step": 40653
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6721004843711853,
      "learning_rate": 0.0005550766670816869,
      "loss": 2.4002,
      "step": 40654
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.7348284125328064,
      "learning_rate": 0.0005550745139083336,
      "loss": 2.3805,
      "step": 40655
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.830915629863739,
      "learning_rate": 0.000555072360687557,
      "loss": 2.2547,
      "step": 40656
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6770071387290955,
      "learning_rate": 0.0005550702074193577,
      "loss": 2.2077,
      "step": 40657
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6474073529243469,
      "learning_rate": 0.000555068054103736,
      "loss": 2.0862,
      "step": 40658
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.696395218372345,
      "learning_rate": 0.0005550659007406923,
      "loss": 2.3787,
      "step": 40659
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6910848617553711,
      "learning_rate": 0.0005550637473302271,
      "loss": 2.2931,
      "step": 40660
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.7690514922142029,
      "learning_rate": 0.0005550615938723408,
      "loss": 2.4905,
      "step": 40661
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6565483212471008,
      "learning_rate": 0.0005550594403670336,
      "loss": 2.0064,
      "step": 40662
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.7350010275840759,
      "learning_rate": 0.0005550572868143061,
      "loss": 2.4449,
      "step": 40663
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6193144917488098,
      "learning_rate": 0.0005550551332141586,
      "loss": 2.272,
      "step": 40664
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.7155214548110962,
      "learning_rate": 0.0005550529795665915,
      "loss": 2.3737,
      "step": 40665
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.8008400797843933,
      "learning_rate": 0.0005550508258716052,
      "loss": 2.3302,
      "step": 40666
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.7895508408546448,
      "learning_rate": 0.0005550486721292003,
      "loss": 2.3484,
      "step": 40667
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.8915249705314636,
      "learning_rate": 0.0005550465183393771,
      "loss": 2.3033,
      "step": 40668
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.7067022323608398,
      "learning_rate": 0.0005550443645021357,
      "loss": 1.9391,
      "step": 40669
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.744426965713501,
      "learning_rate": 0.000555042210617477,
      "loss": 2.3724,
      "step": 40670
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.7190151214599609,
      "learning_rate": 0.0005550400566854009,
      "loss": 2.1964,
      "step": 40671
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.8109835982322693,
      "learning_rate": 0.0005550379027059082,
      "loss": 2.2723,
      "step": 40672
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6766642332077026,
      "learning_rate": 0.000555035748678999,
      "loss": 2.2877,
      "step": 40673
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.665820837020874,
      "learning_rate": 0.0005550335946046738,
      "loss": 2.1266,
      "step": 40674
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6549139618873596,
      "learning_rate": 0.0005550314404829332,
      "loss": 2.1491,
      "step": 40675
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.7007425427436829,
      "learning_rate": 0.0005550292863137773,
      "loss": 2.2907,
      "step": 40676
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6576360464096069,
      "learning_rate": 0.0005550271320972068,
      "loss": 2.2683,
      "step": 40677
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6477534174919128,
      "learning_rate": 0.0005550249778332218,
      "loss": 2.2224,
      "step": 40678
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.7547932863235474,
      "learning_rate": 0.000555022823521823,
      "loss": 2.0762,
      "step": 40679
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6737746000289917,
      "learning_rate": 0.0005550206691630105,
      "loss": 2.3303,
      "step": 40680
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6390202045440674,
      "learning_rate": 0.0005550185147567849,
      "loss": 2.2337,
      "step": 40681
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.7055999636650085,
      "learning_rate": 0.0005550163603031466,
      "loss": 2.3843,
      "step": 40682
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.7315444946289062,
      "learning_rate": 0.0005550142058020958,
      "loss": 2.3027,
      "step": 40683
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.687375009059906,
      "learning_rate": 0.0005550120512536331,
      "loss": 2.2952,
      "step": 40684
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6871019601821899,
      "learning_rate": 0.000555009896657759,
      "loss": 2.1072,
      "step": 40685
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6414837837219238,
      "learning_rate": 0.0005550077420144737,
      "loss": 2.2709,
      "step": 40686
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6261892914772034,
      "learning_rate": 0.0005550055873237774,
      "loss": 2.3318,
      "step": 40687
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.7469545602798462,
      "learning_rate": 0.0005550034325856709,
      "loss": 2.4263,
      "step": 40688
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6654090285301208,
      "learning_rate": 0.0005550012778001545,
      "loss": 2.217,
      "step": 40689
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6934425830841064,
      "learning_rate": 0.0005549991229672286,
      "loss": 2.6093,
      "step": 40690
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6429967880249023,
      "learning_rate": 0.0005549969680868934,
      "loss": 2.0779,
      "step": 40691
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6573289036750793,
      "learning_rate": 0.0005549948131591496,
      "loss": 2.2083,
      "step": 40692
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6736220121383667,
      "learning_rate": 0.0005549926581839974,
      "loss": 2.365,
      "step": 40693
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6580214500427246,
      "learning_rate": 0.0005549905031614371,
      "loss": 2.7734,
      "step": 40694
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6213995814323425,
      "learning_rate": 0.0005549883480914695,
      "loss": 2.2142,
      "step": 40695
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6675651669502258,
      "learning_rate": 0.0005549861929740944,
      "loss": 2.2794,
      "step": 40696
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.7295473217964172,
      "learning_rate": 0.0005549840378093129,
      "loss": 2.2208,
      "step": 40697
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6866121292114258,
      "learning_rate": 0.0005549818825971249,
      "loss": 2.5401,
      "step": 40698
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.7148102521896362,
      "learning_rate": 0.000554979727337531,
      "loss": 2.3876,
      "step": 40699
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6044485569000244,
      "learning_rate": 0.0005549775720305316,
      "loss": 2.064,
      "step": 40700
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6693169474601746,
      "learning_rate": 0.0005549754166761268,
      "loss": 2.4741,
      "step": 40701
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6303896307945251,
      "learning_rate": 0.0005549732612743176,
      "loss": 2.3858,
      "step": 40702
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6420987248420715,
      "learning_rate": 0.0005549711058251039,
      "loss": 2.093,
      "step": 40703
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6559629440307617,
      "learning_rate": 0.0005549689503284862,
      "loss": 2.4365,
      "step": 40704
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6457946300506592,
      "learning_rate": 0.0005549667947844651,
      "loss": 1.9899,
      "step": 40705
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6747126579284668,
      "learning_rate": 0.0005549646391930407,
      "loss": 2.1649,
      "step": 40706
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6060351729393005,
      "learning_rate": 0.0005549624835542137,
      "loss": 2.285,
      "step": 40707
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.7059582471847534,
      "learning_rate": 0.0005549603278679843,
      "loss": 2.275,
      "step": 40708
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.7349716424942017,
      "learning_rate": 0.000554958172134353,
      "loss": 2.404,
      "step": 40709
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6582940816879272,
      "learning_rate": 0.0005549560163533201,
      "loss": 2.2024,
      "step": 40710
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.7076020836830139,
      "learning_rate": 0.0005549538605248861,
      "loss": 2.3199,
      "step": 40711
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6989403963088989,
      "learning_rate": 0.0005549517046490514,
      "loss": 2.2461,
      "step": 40712
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.7148800492286682,
      "learning_rate": 0.0005549495487258164,
      "loss": 2.2309,
      "step": 40713
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.8081128597259521,
      "learning_rate": 0.0005549473927551813,
      "loss": 2.2438,
      "step": 40714
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6977174282073975,
      "learning_rate": 0.0005549452367371469,
      "loss": 2.4538,
      "step": 40715
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.7808442115783691,
      "learning_rate": 0.0005549430806717131,
      "loss": 2.3944,
      "step": 40716
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.7226912975311279,
      "learning_rate": 0.0005549409245588808,
      "loss": 2.2609,
      "step": 40717
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6222845315933228,
      "learning_rate": 0.00055493876839865,
      "loss": 2.3813,
      "step": 40718
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.8059555888175964,
      "learning_rate": 0.0005549366121910214,
      "loss": 2.1012,
      "step": 40719
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6484667062759399,
      "learning_rate": 0.0005549344559359952,
      "loss": 2.1431,
      "step": 40720
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.7176300287246704,
      "learning_rate": 0.0005549322996335719,
      "loss": 2.282,
      "step": 40721
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.7123073935508728,
      "learning_rate": 0.0005549301432837519,
      "loss": 2.3185,
      "step": 40722
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.65324467420578,
      "learning_rate": 0.0005549279868865355,
      "loss": 2.2535,
      "step": 40723
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.7037706971168518,
      "learning_rate": 0.0005549258304419232,
      "loss": 2.1844,
      "step": 40724
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6813123226165771,
      "learning_rate": 0.0005549236739499153,
      "loss": 2.1761,
      "step": 40725
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6725337505340576,
      "learning_rate": 0.0005549215174105124,
      "loss": 2.0813,
      "step": 40726
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6703853011131287,
      "learning_rate": 0.0005549193608237147,
      "loss": 2.3337,
      "step": 40727
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6679802536964417,
      "learning_rate": 0.0005549172041895228,
      "loss": 2.3152,
      "step": 40728
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6710672378540039,
      "learning_rate": 0.0005549150475079368,
      "loss": 2.1361,
      "step": 40729
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6655517220497131,
      "learning_rate": 0.0005549128907789575,
      "loss": 2.0631,
      "step": 40730
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6441001892089844,
      "learning_rate": 0.0005549107340025849,
      "loss": 2.1938,
      "step": 40731
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6433948874473572,
      "learning_rate": 0.0005549085771788196,
      "loss": 2.2109,
      "step": 40732
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6255745887756348,
      "learning_rate": 0.000554906420307662,
      "loss": 2.2571,
      "step": 40733
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.7278568744659424,
      "learning_rate": 0.0005549042633891127,
      "loss": 2.3101,
      "step": 40734
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.7306512594223022,
      "learning_rate": 0.0005549021064231716,
      "loss": 2.2957,
      "step": 40735
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.7085460424423218,
      "learning_rate": 0.0005548999494098395,
      "loss": 2.1808,
      "step": 40736
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.7173884510993958,
      "learning_rate": 0.0005548977923491167,
      "loss": 2.2928,
      "step": 40737
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.7105095982551575,
      "learning_rate": 0.0005548956352410035,
      "loss": 2.1429,
      "step": 40738
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6885315775871277,
      "learning_rate": 0.0005548934780855004,
      "loss": 2.2795,
      "step": 40739
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.7126823663711548,
      "learning_rate": 0.0005548913208826079,
      "loss": 2.0857,
      "step": 40740
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6915606260299683,
      "learning_rate": 0.0005548891636323263,
      "loss": 2.1996,
      "step": 40741
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.688008725643158,
      "learning_rate": 0.0005548870063346559,
      "loss": 2.1165,
      "step": 40742
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.8355945348739624,
      "learning_rate": 0.0005548848489895972,
      "loss": 2.2545,
      "step": 40743
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.666206955909729,
      "learning_rate": 0.0005548826915971507,
      "loss": 2.0603,
      "step": 40744
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6388005018234253,
      "learning_rate": 0.0005548805341573165,
      "loss": 2.189,
      "step": 40745
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.8575243949890137,
      "learning_rate": 0.0005548783766700954,
      "loss": 2.3418,
      "step": 40746
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.7379090785980225,
      "learning_rate": 0.0005548762191354875,
      "loss": 2.347,
      "step": 40747
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6630125641822815,
      "learning_rate": 0.0005548740615534932,
      "loss": 2.1586,
      "step": 40748
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.7026627063751221,
      "learning_rate": 0.0005548719039241132,
      "loss": 2.339,
      "step": 40749
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6881237626075745,
      "learning_rate": 0.0005548697462473476,
      "loss": 2.3646,
      "step": 40750
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6700462102890015,
      "learning_rate": 0.0005548675885231969,
      "loss": 2.1767,
      "step": 40751
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.7529748678207397,
      "learning_rate": 0.0005548654307516615,
      "loss": 2.4399,
      "step": 40752
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6817252039909363,
      "learning_rate": 0.0005548632729327418,
      "loss": 2.4646,
      "step": 40753
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6497015953063965,
      "learning_rate": 0.0005548611150664383,
      "loss": 2.1227,
      "step": 40754
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6938951015472412,
      "learning_rate": 0.0005548589571527511,
      "loss": 2.0733,
      "step": 40755
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.7006768584251404,
      "learning_rate": 0.0005548567991916811,
      "loss": 2.5629,
      "step": 40756
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6961459517478943,
      "learning_rate": 0.0005548546411832282,
      "loss": 2.3882,
      "step": 40757
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6775323152542114,
      "learning_rate": 0.0005548524831273931,
      "loss": 2.4084,
      "step": 40758
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.8268710970878601,
      "learning_rate": 0.000554850325024176,
      "loss": 2.3437,
      "step": 40759
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.7072980999946594,
      "learning_rate": 0.0005548481668735775,
      "loss": 2.2143,
      "step": 40760
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.8252496123313904,
      "learning_rate": 0.000554846008675598,
      "loss": 2.0385,
      "step": 40761
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.7177759408950806,
      "learning_rate": 0.0005548438504302377,
      "loss": 2.2497,
      "step": 40762
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6733856797218323,
      "learning_rate": 0.0005548416921374971,
      "loss": 2.3531,
      "step": 40763
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.7094200849533081,
      "learning_rate": 0.0005548395337973766,
      "loss": 2.3099,
      "step": 40764
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.7178698182106018,
      "learning_rate": 0.0005548373754098768,
      "loss": 2.2478,
      "step": 40765
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6572310924530029,
      "learning_rate": 0.0005548352169749978,
      "loss": 2.2024,
      "step": 40766
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.7670502662658691,
      "learning_rate": 0.0005548330584927402,
      "loss": 2.2687,
      "step": 40767
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6357461810112,
      "learning_rate": 0.0005548308999631041,
      "loss": 2.2255,
      "step": 40768
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6989627480506897,
      "learning_rate": 0.0005548287413860903,
      "loss": 2.4807,
      "step": 40769
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.701151967048645,
      "learning_rate": 0.000554826582761699,
      "loss": 2.3133,
      "step": 40770
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.691058874130249,
      "learning_rate": 0.0005548244240899307,
      "loss": 2.386,
      "step": 40771
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.7597849369049072,
      "learning_rate": 0.0005548222653707856,
      "loss": 2.2292,
      "step": 40772
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6433612704277039,
      "learning_rate": 0.0005548201066042644,
      "loss": 2.236,
      "step": 40773
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.7144506573677063,
      "learning_rate": 0.0005548179477903671,
      "loss": 2.3334,
      "step": 40774
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.7475773096084595,
      "learning_rate": 0.0005548157889290945,
      "loss": 2.0588,
      "step": 40775
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.7026098370552063,
      "learning_rate": 0.0005548136300204469,
      "loss": 2.3126,
      "step": 40776
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.7258775234222412,
      "learning_rate": 0.0005548114710644244,
      "loss": 2.318,
      "step": 40777
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.7399753928184509,
      "learning_rate": 0.0005548093120610278,
      "loss": 2.2877,
      "step": 40778
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.7841933965682983,
      "learning_rate": 0.0005548071530102573,
      "loss": 2.4451,
      "step": 40779
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.5882627964019775,
      "learning_rate": 0.0005548049939121133,
      "loss": 2.1766,
      "step": 40780
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6133606433868408,
      "learning_rate": 0.0005548028347665963,
      "loss": 2.3542,
      "step": 40781
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.658793032169342,
      "learning_rate": 0.0005548006755737067,
      "loss": 2.1625,
      "step": 40782
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6658139824867249,
      "learning_rate": 0.0005547985163334448,
      "loss": 2.3813,
      "step": 40783
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6414158940315247,
      "learning_rate": 0.000554796357045811,
      "loss": 2.2899,
      "step": 40784
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.7309260964393616,
      "learning_rate": 0.0005547941977108058,
      "loss": 2.1803,
      "step": 40785
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6944669485092163,
      "learning_rate": 0.0005547920383284294,
      "loss": 2.2326,
      "step": 40786
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.7032339572906494,
      "learning_rate": 0.0005547898788986825,
      "loss": 2.1462,
      "step": 40787
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.7044783234596252,
      "learning_rate": 0.0005547877194215653,
      "loss": 2.3543,
      "step": 40788
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6849888563156128,
      "learning_rate": 0.0005547855598970784,
      "loss": 2.1253,
      "step": 40789
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6353728175163269,
      "learning_rate": 0.0005547834003252219,
      "loss": 2.1748,
      "step": 40790
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.7327690124511719,
      "learning_rate": 0.0005547812407059963,
      "loss": 2.3405,
      "step": 40791
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6640893220901489,
      "learning_rate": 0.0005547790810394022,
      "loss": 2.2989,
      "step": 40792
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6633888483047485,
      "learning_rate": 0.0005547769213254398,
      "loss": 2.3206,
      "step": 40793
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6679086685180664,
      "learning_rate": 0.0005547747615641096,
      "loss": 2.2026,
      "step": 40794
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.7468762397766113,
      "learning_rate": 0.0005547726017554118,
      "loss": 2.2179,
      "step": 40795
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.7434672713279724,
      "learning_rate": 0.0005547704418993472,
      "loss": 2.2618,
      "step": 40796
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6996728777885437,
      "learning_rate": 0.0005547682819959158,
      "loss": 2.2974,
      "step": 40797
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6365908980369568,
      "learning_rate": 0.0005547661220451183,
      "loss": 2.2253,
      "step": 40798
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6718396544456482,
      "learning_rate": 0.0005547639620469549,
      "loss": 2.298,
      "step": 40799
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.8047976493835449,
      "learning_rate": 0.0005547618020014261,
      "loss": 2.2044,
      "step": 40800
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.9638206362724304,
      "learning_rate": 0.0005547596419085322,
      "loss": 2.3946,
      "step": 40801
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.7592094540596008,
      "learning_rate": 0.0005547574817682736,
      "loss": 2.1791,
      "step": 40802
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.984326183795929,
      "learning_rate": 0.000554755321580651,
      "loss": 2.1892,
      "step": 40803
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.779036819934845,
      "learning_rate": 0.0005547531613456644,
      "loss": 2.0587,
      "step": 40804
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.9104996919631958,
      "learning_rate": 0.0005547510010633144,
      "loss": 2.3002,
      "step": 40805
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.8755958676338196,
      "learning_rate": 0.0005547488407336014,
      "loss": 2.3491,
      "step": 40806
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6399625539779663,
      "learning_rate": 0.0005547466803565258,
      "loss": 2.358,
      "step": 40807
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6870179176330566,
      "learning_rate": 0.0005547445199320881,
      "loss": 2.2336,
      "step": 40808
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.7211663722991943,
      "learning_rate": 0.0005547423594602885,
      "loss": 2.3088,
      "step": 40809
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6873470544815063,
      "learning_rate": 0.0005547401989411274,
      "loss": 2.493,
      "step": 40810
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.7047834396362305,
      "learning_rate": 0.0005547380383746054,
      "loss": 2.3056,
      "step": 40811
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.7014618515968323,
      "learning_rate": 0.0005547358777607226,
      "loss": 2.3783,
      "step": 40812
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.7147481441497803,
      "learning_rate": 0.0005547337170994798,
      "loss": 2.1693,
      "step": 40813
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6528352499008179,
      "learning_rate": 0.0005547315563908771,
      "loss": 2.1305,
      "step": 40814
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6927462816238403,
      "learning_rate": 0.000554729395634915,
      "loss": 2.3469,
      "step": 40815
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.675240695476532,
      "learning_rate": 0.0005547272348315939,
      "loss": 2.0645,
      "step": 40816
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.7071858644485474,
      "learning_rate": 0.0005547250739809142,
      "loss": 2.4482,
      "step": 40817
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.7389718294143677,
      "learning_rate": 0.0005547229130828763,
      "loss": 2.3094,
      "step": 40818
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6600592732429504,
      "learning_rate": 0.0005547207521374807,
      "loss": 2.5621,
      "step": 40819
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.7738163471221924,
      "learning_rate": 0.0005547185911447275,
      "loss": 2.3209,
      "step": 40820
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.7195767164230347,
      "learning_rate": 0.0005547164301046175,
      "loss": 2.2822,
      "step": 40821
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.7449058890342712,
      "learning_rate": 0.0005547142690171508,
      "loss": 2.2791,
      "step": 40822
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.7712033987045288,
      "learning_rate": 0.0005547121078823279,
      "loss": 2.3809,
      "step": 40823
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.7233248353004456,
      "learning_rate": 0.0005547099467001494,
      "loss": 2.3107,
      "step": 40824
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6663939356803894,
      "learning_rate": 0.0005547077854706153,
      "loss": 2.2335,
      "step": 40825
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6834742426872253,
      "learning_rate": 0.0005547056241937263,
      "loss": 2.1882,
      "step": 40826
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6403356194496155,
      "learning_rate": 0.0005547034628694827,
      "loss": 2.2548,
      "step": 40827
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.643502414226532,
      "learning_rate": 0.0005547013014978849,
      "loss": 2.2465,
      "step": 40828
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6758027672767639,
      "learning_rate": 0.0005546991400789333,
      "loss": 2.2481,
      "step": 40829
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6907966136932373,
      "learning_rate": 0.0005546969786126284,
      "loss": 2.483,
      "step": 40830
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.711775004863739,
      "learning_rate": 0.0005546948170989704,
      "loss": 2.3826,
      "step": 40831
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.7605414390563965,
      "learning_rate": 0.00055469265553796,
      "loss": 2.3796,
      "step": 40832
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.7167529463768005,
      "learning_rate": 0.0005546904939295974,
      "loss": 2.3116,
      "step": 40833
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.769379198551178,
      "learning_rate": 0.0005546883322738829,
      "loss": 2.2872,
      "step": 40834
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.8378868699073792,
      "learning_rate": 0.0005546861705708171,
      "loss": 2.1748,
      "step": 40835
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.7215064167976379,
      "learning_rate": 0.0005546840088204003,
      "loss": 2.2832,
      "step": 40836
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.7374468445777893,
      "learning_rate": 0.000554681847022633,
      "loss": 2.1875,
      "step": 40837
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6665530204772949,
      "learning_rate": 0.0005546796851775154,
      "loss": 2.2045,
      "step": 40838
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6513815522193909,
      "learning_rate": 0.0005546775232850483,
      "loss": 2.2166,
      "step": 40839
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6971470713615417,
      "learning_rate": 0.0005546753613452315,
      "loss": 1.9955,
      "step": 40840
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.7094449996948242,
      "learning_rate": 0.0005546731993580661,
      "loss": 2.1271,
      "step": 40841
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6825091242790222,
      "learning_rate": 0.0005546710373235519,
      "loss": 2.2882,
      "step": 40842
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.632879912853241,
      "learning_rate": 0.0005546688752416896,
      "loss": 2.3616,
      "step": 40843
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6615034937858582,
      "learning_rate": 0.0005546667131124795,
      "loss": 2.3619,
      "step": 40844
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.647237241268158,
      "learning_rate": 0.0005546645509359222,
      "loss": 2.5291,
      "step": 40845
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.7041781544685364,
      "learning_rate": 0.0005546623887120178,
      "loss": 2.2858,
      "step": 40846
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.7281615138053894,
      "learning_rate": 0.000554660226440767,
      "loss": 2.4505,
      "step": 40847
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.5850164890289307,
      "learning_rate": 0.00055465806412217,
      "loss": 2.1531,
      "step": 40848
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6229977011680603,
      "learning_rate": 0.0005546559017562271,
      "loss": 2.3788,
      "step": 40849
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.7681839466094971,
      "learning_rate": 0.0005546537393429391,
      "loss": 2.3442,
      "step": 40850
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.7305585145950317,
      "learning_rate": 0.000554651576882306,
      "loss": 2.2813,
      "step": 40851
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6681297421455383,
      "learning_rate": 0.0005546494143743284,
      "loss": 2.1178,
      "step": 40852
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.7161800861358643,
      "learning_rate": 0.0005546472518190067,
      "loss": 2.1039,
      "step": 40853
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.7729951739311218,
      "learning_rate": 0.0005546450892163413,
      "loss": 2.4516,
      "step": 40854
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6247565150260925,
      "learning_rate": 0.0005546429265663326,
      "loss": 2.4727,
      "step": 40855
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.684312641620636,
      "learning_rate": 0.0005546407638689809,
      "loss": 2.1177,
      "step": 40856
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.764602541923523,
      "learning_rate": 0.0005546386011242866,
      "loss": 2.3793,
      "step": 40857
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.7708762884140015,
      "learning_rate": 0.0005546364383322503,
      "loss": 2.3147,
      "step": 40858
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6754156351089478,
      "learning_rate": 0.0005546342754928723,
      "loss": 2.1673,
      "step": 40859
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.774365246295929,
      "learning_rate": 0.0005546321126061528,
      "loss": 2.1305,
      "step": 40860
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.5985473990440369,
      "learning_rate": 0.0005546299496720925,
      "loss": 2.1797,
      "step": 40861
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.793973982334137,
      "learning_rate": 0.0005546277866906917,
      "loss": 2.2597,
      "step": 40862
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.8073055148124695,
      "learning_rate": 0.0005546256236619509,
      "loss": 2.332,
      "step": 40863
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6835448145866394,
      "learning_rate": 0.0005546234605858702,
      "loss": 2.2954,
      "step": 40864
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.7452186942100525,
      "learning_rate": 0.0005546212974624503,
      "loss": 2.2903,
      "step": 40865
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.7479708194732666,
      "learning_rate": 0.0005546191342916914,
      "loss": 2.2021,
      "step": 40866
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6619210243225098,
      "learning_rate": 0.0005546169710735941,
      "loss": 2.283,
      "step": 40867
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6597967743873596,
      "learning_rate": 0.0005546148078081586,
      "loss": 2.5163,
      "step": 40868
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.8108438849449158,
      "learning_rate": 0.0005546126444953855,
      "loss": 2.1614,
      "step": 40869
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.752814531326294,
      "learning_rate": 0.000554610481135275,
      "loss": 2.2381,
      "step": 40870
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6659705638885498,
      "learning_rate": 0.0005546083177278277,
      "loss": 2.281,
      "step": 40871
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.7199180722236633,
      "learning_rate": 0.0005546061542730439,
      "loss": 2.1815,
      "step": 40872
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6878988742828369,
      "learning_rate": 0.0005546039907709238,
      "loss": 2.407,
      "step": 40873
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.647890567779541,
      "learning_rate": 0.0005546018272214682,
      "loss": 2.2166,
      "step": 40874
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6389339566230774,
      "learning_rate": 0.0005545996636246775,
      "loss": 2.0734,
      "step": 40875
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6976295113563538,
      "learning_rate": 0.0005545974999805516,
      "loss": 2.2037,
      "step": 40876
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6870991587638855,
      "learning_rate": 0.0005545953362890914,
      "loss": 2.2895,
      "step": 40877
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.7311981320381165,
      "learning_rate": 0.0005545931725502971,
      "loss": 2.4551,
      "step": 40878
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.7792907357215881,
      "learning_rate": 0.000554591008764169,
      "loss": 2.287,
      "step": 40879
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6636326313018799,
      "learning_rate": 0.0005545888449307077,
      "loss": 2.5733,
      "step": 40880
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.8897956013679504,
      "learning_rate": 0.0005545866810499135,
      "loss": 2.2745,
      "step": 40881
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6589862704277039,
      "learning_rate": 0.0005545845171217869,
      "loss": 2.1597,
      "step": 40882
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6859024167060852,
      "learning_rate": 0.0005545823531463282,
      "loss": 2.0372,
      "step": 40883
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.7008334398269653,
      "learning_rate": 0.000554580189123538,
      "loss": 2.2344,
      "step": 40884
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.7453963756561279,
      "learning_rate": 0.0005545780250534163,
      "loss": 2.0967,
      "step": 40885
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.7136139869689941,
      "learning_rate": 0.0005545758609359638,
      "loss": 2.0902,
      "step": 40886
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.9101283550262451,
      "learning_rate": 0.0005545736967711808,
      "loss": 2.1479,
      "step": 40887
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.7130299210548401,
      "learning_rate": 0.0005545715325590678,
      "loss": 2.1557,
      "step": 40888
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6818475723266602,
      "learning_rate": 0.0005545693682996252,
      "loss": 2.34,
      "step": 40889
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.784278392791748,
      "learning_rate": 0.0005545672039928533,
      "loss": 2.4164,
      "step": 40890
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.7003558874130249,
      "learning_rate": 0.0005545650396387525,
      "loss": 2.3021,
      "step": 40891
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6587842702865601,
      "learning_rate": 0.0005545628752373234,
      "loss": 2.4079,
      "step": 40892
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.7401265501976013,
      "learning_rate": 0.0005545607107885661,
      "loss": 2.4417,
      "step": 40893
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6611360311508179,
      "learning_rate": 0.0005545585462924811,
      "loss": 2.4014,
      "step": 40894
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.8510105609893799,
      "learning_rate": 0.000554556381749069,
      "loss": 2.2679,
      "step": 40895
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6648901104927063,
      "learning_rate": 0.0005545542171583301,
      "loss": 2.4,
      "step": 40896
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6786354780197144,
      "learning_rate": 0.0005545520525202648,
      "loss": 2.4678,
      "step": 40897
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.7418239116668701,
      "learning_rate": 0.0005545498878348733,
      "loss": 2.1654,
      "step": 40898
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.8295999765396118,
      "learning_rate": 0.0005545477231021562,
      "loss": 2.3436,
      "step": 40899
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6838045716285706,
      "learning_rate": 0.000554545558322114,
      "loss": 2.2907,
      "step": 40900
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.7233403921127319,
      "learning_rate": 0.0005545433934947469,
      "loss": 2.1133,
      "step": 40901
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.7935061454772949,
      "learning_rate": 0.0005545412286200554,
      "loss": 2.2402,
      "step": 40902
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.721284806728363,
      "learning_rate": 0.0005545390636980397,
      "loss": 2.2102,
      "step": 40903
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.725782573223114,
      "learning_rate": 0.0005545368987287006,
      "loss": 2.3825,
      "step": 40904
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6966110467910767,
      "learning_rate": 0.0005545347337120381,
      "loss": 2.254,
      "step": 40905
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.7964962124824524,
      "learning_rate": 0.0005545325686480531,
      "loss": 2.3209,
      "step": 40906
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6701962351799011,
      "learning_rate": 0.0005545304035367455,
      "loss": 2.1579,
      "step": 40907
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.7457334995269775,
      "learning_rate": 0.0005545282383781158,
      "loss": 2.4851,
      "step": 40908
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.665036678314209,
      "learning_rate": 0.0005545260731721646,
      "loss": 2.1252,
      "step": 40909
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6654692888259888,
      "learning_rate": 0.0005545239079188922,
      "loss": 2.1775,
      "step": 40910
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6741310954093933,
      "learning_rate": 0.000554521742618299,
      "loss": 2.2356,
      "step": 40911
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.7461392283439636,
      "learning_rate": 0.0005545195772703854,
      "loss": 2.1838,
      "step": 40912
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.7021007537841797,
      "learning_rate": 0.0005545174118751517,
      "loss": 1.9786,
      "step": 40913
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6623768210411072,
      "learning_rate": 0.0005545152464325986,
      "loss": 2.2793,
      "step": 40914
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6375683546066284,
      "learning_rate": 0.0005545130809427262,
      "loss": 2.5486,
      "step": 40915
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6749999523162842,
      "learning_rate": 0.000554510915405535,
      "loss": 2.2397,
      "step": 40916
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6701016426086426,
      "learning_rate": 0.0005545087498210255,
      "loss": 2.1378,
      "step": 40917
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6435489654541016,
      "learning_rate": 0.000554506584189198,
      "loss": 2.2071,
      "step": 40918
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6164337992668152,
      "learning_rate": 0.0005545044185100528,
      "loss": 2.0797,
      "step": 40919
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6728721857070923,
      "learning_rate": 0.0005545022527835906,
      "loss": 2.3392,
      "step": 40920
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.8220095038414001,
      "learning_rate": 0.0005545000870098116,
      "loss": 2.2822,
      "step": 40921
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6634799838066101,
      "learning_rate": 0.0005544979211887162,
      "loss": 2.1782,
      "step": 40922
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.7160131335258484,
      "learning_rate": 0.0005544957553203047,
      "loss": 2.0732,
      "step": 40923
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.8513820171356201,
      "learning_rate": 0.0005544935894045777,
      "loss": 2.1305,
      "step": 40924
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6411675810813904,
      "learning_rate": 0.0005544914234415356,
      "loss": 2.1821,
      "step": 40925
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6451576948165894,
      "learning_rate": 0.0005544892574311787,
      "loss": 2.3309,
      "step": 40926
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.7972051501274109,
      "learning_rate": 0.0005544870913735075,
      "loss": 2.3616,
      "step": 40927
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.7169713377952576,
      "learning_rate": 0.0005544849252685222,
      "loss": 2.3212,
      "step": 40928
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6180921792984009,
      "learning_rate": 0.0005544827591162236,
      "loss": 2.3083,
      "step": 40929
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.8364983201026917,
      "learning_rate": 0.0005544805929166117,
      "loss": 2.5586,
      "step": 40930
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6065216064453125,
      "learning_rate": 0.000554478426669687,
      "loss": 2.18,
      "step": 40931
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.7282964587211609,
      "learning_rate": 0.00055447626037545,
      "loss": 2.4785,
      "step": 40932
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.7058263421058655,
      "learning_rate": 0.0005544740940339011,
      "loss": 2.1627,
      "step": 40933
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.7094464302062988,
      "learning_rate": 0.0005544719276450406,
      "loss": 2.0972,
      "step": 40934
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6105328798294067,
      "learning_rate": 0.0005544697612088691,
      "loss": 2.1123,
      "step": 40935
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.7727511525154114,
      "learning_rate": 0.0005544675947253868,
      "loss": 2.0055,
      "step": 40936
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6091210246086121,
      "learning_rate": 0.0005544654281945941,
      "loss": 2.0059,
      "step": 40937
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6931883096694946,
      "learning_rate": 0.0005544632616164914,
      "loss": 2.2869,
      "step": 40938
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.8097257018089294,
      "learning_rate": 0.0005544610949910794,
      "loss": 2.43,
      "step": 40939
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.7435895204544067,
      "learning_rate": 0.0005544589283183581,
      "loss": 2.3179,
      "step": 40940
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.7471653819084167,
      "learning_rate": 0.0005544567615983282,
      "loss": 2.1091,
      "step": 40941
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.698235809803009,
      "learning_rate": 0.0005544545948309899,
      "loss": 2.3776,
      "step": 40942
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.8557838201522827,
      "learning_rate": 0.0005544524280163438,
      "loss": 2.2252,
      "step": 40943
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6684276461601257,
      "learning_rate": 0.0005544502611543901,
      "loss": 2.117,
      "step": 40944
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.7661815881729126,
      "learning_rate": 0.0005544480942451293,
      "loss": 2.0132,
      "step": 40945
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6477927565574646,
      "learning_rate": 0.0005544459272885618,
      "loss": 2.391,
      "step": 40946
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.7010215520858765,
      "learning_rate": 0.000554443760284688,
      "loss": 2.3758,
      "step": 40947
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6713586449623108,
      "learning_rate": 0.0005544415932335084,
      "loss": 1.9778,
      "step": 40948
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6845287084579468,
      "learning_rate": 0.0005544394261350232,
      "loss": 2.1342,
      "step": 40949
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.7180739045143127,
      "learning_rate": 0.0005544372589892329,
      "loss": 2.2132,
      "step": 40950
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.748357892036438,
      "learning_rate": 0.0005544350917961379,
      "loss": 2.3557,
      "step": 40951
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.720190703868866,
      "learning_rate": 0.0005544329245557386,
      "loss": 2.2347,
      "step": 40952
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6341838836669922,
      "learning_rate": 0.0005544307572680355,
      "loss": 2.0344,
      "step": 40953
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6990377306938171,
      "learning_rate": 0.0005544285899330289,
      "loss": 2.1391,
      "step": 40954
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6659300327301025,
      "learning_rate": 0.0005544264225507191,
      "loss": 2.5076,
      "step": 40955
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.875616192817688,
      "learning_rate": 0.0005544242551211068,
      "loss": 2.3642,
      "step": 40956
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6893391013145447,
      "learning_rate": 0.0005544220876441922,
      "loss": 2.2747,
      "step": 40957
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6931488513946533,
      "learning_rate": 0.0005544199201199757,
      "loss": 2.2168,
      "step": 40958
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.7232412695884705,
      "learning_rate": 0.0005544177525484577,
      "loss": 2.1858,
      "step": 40959
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6877449750900269,
      "learning_rate": 0.0005544155849296386,
      "loss": 2.3652,
      "step": 40960
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.7120299935340881,
      "learning_rate": 0.0005544134172635189,
      "loss": 2.1405,
      "step": 40961
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6775887608528137,
      "learning_rate": 0.000554411249550099,
      "loss": 2.2587,
      "step": 40962
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.7439510226249695,
      "learning_rate": 0.0005544090817893792,
      "loss": 2.5696,
      "step": 40963
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.7074671983718872,
      "learning_rate": 0.0005544069139813599,
      "loss": 2.268,
      "step": 40964
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.7448282241821289,
      "learning_rate": 0.0005544047461260416,
      "loss": 2.2068,
      "step": 40965
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.7560412883758545,
      "learning_rate": 0.0005544025782234247,
      "loss": 2.3055,
      "step": 40966
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6919718980789185,
      "learning_rate": 0.0005544004102735094,
      "loss": 2.265,
      "step": 40967
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.658613383769989,
      "learning_rate": 0.0005543982422762964,
      "loss": 2.3465,
      "step": 40968
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6647763848304749,
      "learning_rate": 0.0005543960742317859,
      "loss": 2.2678,
      "step": 40969
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.751865029335022,
      "learning_rate": 0.0005543939061399784,
      "loss": 2.2239,
      "step": 40970
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6749767065048218,
      "learning_rate": 0.0005543917380008743,
      "loss": 2.1808,
      "step": 40971
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.7791240811347961,
      "learning_rate": 0.000554389569814474,
      "loss": 2.1602,
      "step": 40972
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6476127505302429,
      "learning_rate": 0.0005543874015807778,
      "loss": 2.5965,
      "step": 40973
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.7610296607017517,
      "learning_rate": 0.0005543852332997862,
      "loss": 2.0934,
      "step": 40974
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.65566086769104,
      "learning_rate": 0.0005543830649714996,
      "loss": 2.1855,
      "step": 40975
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.7001519203186035,
      "learning_rate": 0.0005543808965959183,
      "loss": 2.1328,
      "step": 40976
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.7382607460021973,
      "learning_rate": 0.0005543787281730429,
      "loss": 2.2475,
      "step": 40977
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.7012184858322144,
      "learning_rate": 0.0005543765597028737,
      "loss": 2.27,
      "step": 40978
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6994919776916504,
      "learning_rate": 0.000554374391185411,
      "loss": 2.3511,
      "step": 40979
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.7654501795768738,
      "learning_rate": 0.0005543722226206553,
      "loss": 2.5298,
      "step": 40980
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6134331226348877,
      "learning_rate": 0.0005543700540086071,
      "loss": 2.2856,
      "step": 40981
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6862902641296387,
      "learning_rate": 0.0005543678853492666,
      "loss": 2.3649,
      "step": 40982
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.7335780262947083,
      "learning_rate": 0.0005543657166426345,
      "loss": 2.1228,
      "step": 40983
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.7035095691680908,
      "learning_rate": 0.0005543635478887108,
      "loss": 2.2014,
      "step": 40984
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.640970766544342,
      "learning_rate": 0.0005543613790874963,
      "loss": 2.0667,
      "step": 40985
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6822078227996826,
      "learning_rate": 0.0005543592102389911,
      "loss": 2.2466,
      "step": 40986
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6619325280189514,
      "learning_rate": 0.0005543570413431956,
      "loss": 2.0302,
      "step": 40987
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.7016177773475647,
      "learning_rate": 0.0005543548724001104,
      "loss": 2.4526,
      "step": 40988
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6907315850257874,
      "learning_rate": 0.000554352703409736,
      "loss": 2.245,
      "step": 40989
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6728759407997131,
      "learning_rate": 0.0005543505343720725,
      "loss": 2.4126,
      "step": 40990
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6635167002677917,
      "learning_rate": 0.0005543483652871204,
      "loss": 2.3647,
      "step": 40991
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6789321899414062,
      "learning_rate": 0.0005543461961548803,
      "loss": 2.4457,
      "step": 40992
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.7042654156684875,
      "learning_rate": 0.0005543440269753522,
      "loss": 2.0291,
      "step": 40993
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6749272346496582,
      "learning_rate": 0.0005543418577485369,
      "loss": 2.3412,
      "step": 40994
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6416124105453491,
      "learning_rate": 0.0005543396884744346,
      "loss": 2.2231,
      "step": 40995
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6821879744529724,
      "learning_rate": 0.0005543375191530458,
      "loss": 2.3051,
      "step": 40996
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.7008573412895203,
      "learning_rate": 0.0005543353497843708,
      "loss": 2.3071,
      "step": 40997
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.7282172441482544,
      "learning_rate": 0.00055433318036841,
      "loss": 2.2198,
      "step": 40998
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6877637505531311,
      "learning_rate": 0.0005543310109051641,
      "loss": 2.3855,
      "step": 40999
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6731247305870056,
      "learning_rate": 0.0005543288413946331,
      "loss": 2.3122,
      "step": 41000
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.7236661911010742,
      "learning_rate": 0.0005543266718368176,
      "loss": 2.2578,
      "step": 41001
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.773521363735199,
      "learning_rate": 0.0005543245022317179,
      "loss": 2.4516,
      "step": 41002
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6622071266174316,
      "learning_rate": 0.0005543223325793345,
      "loss": 2.0773,
      "step": 41003
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.8599395751953125,
      "learning_rate": 0.0005543201628796678,
      "loss": 2.3356,
      "step": 41004
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.7085492610931396,
      "learning_rate": 0.0005543179931327181,
      "loss": 2.264,
      "step": 41005
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.7540958523750305,
      "learning_rate": 0.000554315823338486,
      "loss": 2.3009,
      "step": 41006
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6744333505630493,
      "learning_rate": 0.0005543136534969718,
      "loss": 2.382,
      "step": 41007
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.715731143951416,
      "learning_rate": 0.0005543114836081758,
      "loss": 2.4107,
      "step": 41008
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.7485503554344177,
      "learning_rate": 0.0005543093136720985,
      "loss": 2.2213,
      "step": 41009
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.7782658338546753,
      "learning_rate": 0.0005543071436887404,
      "loss": 2.1433,
      "step": 41010
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6776101589202881,
      "learning_rate": 0.0005543049736581017,
      "loss": 2.2166,
      "step": 41011
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.8581128716468811,
      "learning_rate": 0.000554302803580183,
      "loss": 2.294,
      "step": 41012
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6973155736923218,
      "learning_rate": 0.0005543006334549845,
      "loss": 2.1524,
      "step": 41013
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6922496557235718,
      "learning_rate": 0.0005542984632825068,
      "loss": 2.1904,
      "step": 41014
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.7117066383361816,
      "learning_rate": 0.0005542962930627502,
      "loss": 2.2264,
      "step": 41015
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6808304190635681,
      "learning_rate": 0.0005542941227957151,
      "loss": 2.5166,
      "step": 41016
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6807321906089783,
      "learning_rate": 0.0005542919524814019,
      "loss": 2.4398,
      "step": 41017
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.714905858039856,
      "learning_rate": 0.000554289782119811,
      "loss": 2.0456,
      "step": 41018
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6945802569389343,
      "learning_rate": 0.000554287611710943,
      "loss": 2.3193,
      "step": 41019
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.7295540571212769,
      "learning_rate": 0.000554285441254798,
      "loss": 2.2263,
      "step": 41020
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6952190399169922,
      "learning_rate": 0.0005542832707513764,
      "loss": 2.306,
      "step": 41021
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.7082826495170593,
      "learning_rate": 0.0005542811002006789,
      "loss": 2.2192,
      "step": 41022
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6972872018814087,
      "learning_rate": 0.0005542789296027057,
      "loss": 2.3058,
      "step": 41023
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.7655072808265686,
      "learning_rate": 0.0005542767589574573,
      "loss": 2.3386,
      "step": 41024
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.7427130341529846,
      "learning_rate": 0.000554274588264934,
      "loss": 2.5566,
      "step": 41025
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.7100929617881775,
      "learning_rate": 0.0005542724175251363,
      "loss": 2.3705,
      "step": 41026
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6753875613212585,
      "learning_rate": 0.0005542702467380646,
      "loss": 2.2114,
      "step": 41027
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6578081846237183,
      "learning_rate": 0.0005542680759037192,
      "loss": 2.2098,
      "step": 41028
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.7082244753837585,
      "learning_rate": 0.0005542659050221005,
      "loss": 2.4032,
      "step": 41029
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.7236810922622681,
      "learning_rate": 0.0005542637340932091,
      "loss": 2.4104,
      "step": 41030
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6565132141113281,
      "learning_rate": 0.0005542615631170452,
      "loss": 2.4092,
      "step": 41031
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6524496674537659,
      "learning_rate": 0.0005542593920936092,
      "loss": 2.1414,
      "step": 41032
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.7956929802894592,
      "learning_rate": 0.0005542572210229018,
      "loss": 2.4155,
      "step": 41033
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6497326493263245,
      "learning_rate": 0.000554255049904923,
      "loss": 2.2346,
      "step": 41034
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6657895445823669,
      "learning_rate": 0.0005542528787396733,
      "loss": 2.2801,
      "step": 41035
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6968101859092712,
      "learning_rate": 0.0005542507075271533,
      "loss": 2.1301,
      "step": 41036
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6430697441101074,
      "learning_rate": 0.0005542485362673634,
      "loss": 1.9329,
      "step": 41037
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6706250309944153,
      "learning_rate": 0.0005542463649603038,
      "loss": 2.4437,
      "step": 41038
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6958552598953247,
      "learning_rate": 0.000554244193605975,
      "loss": 2.1079,
      "step": 41039
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.7079275846481323,
      "learning_rate": 0.0005542420222043775,
      "loss": 2.413,
      "step": 41040
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.791583240032196,
      "learning_rate": 0.0005542398507555115,
      "loss": 2.2772,
      "step": 41041
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6601667404174805,
      "learning_rate": 0.0005542376792593775,
      "loss": 2.1892,
      "step": 41042
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.8325493931770325,
      "learning_rate": 0.000554235507715976,
      "loss": 2.2494,
      "step": 41043
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6928982734680176,
      "learning_rate": 0.0005542333361253072,
      "loss": 2.2025,
      "step": 41044
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6778910160064697,
      "learning_rate": 0.0005542311644873717,
      "loss": 2.3514,
      "step": 41045
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.8780085444450378,
      "learning_rate": 0.0005542289928021698,
      "loss": 2.2367,
      "step": 41046
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.746803879737854,
      "learning_rate": 0.000554226821069702,
      "loss": 2.245,
      "step": 41047
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.668640673160553,
      "learning_rate": 0.0005542246492899687,
      "loss": 2.3366,
      "step": 41048
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.7105277180671692,
      "learning_rate": 0.00055422247746297,
      "loss": 2.454,
      "step": 41049
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6502312421798706,
      "learning_rate": 0.0005542203055887067,
      "loss": 2.2964,
      "step": 41050
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6862422823905945,
      "learning_rate": 0.0005542181336671791,
      "loss": 2.1389,
      "step": 41051
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.70562744140625,
      "learning_rate": 0.0005542159616983874,
      "loss": 2.2448,
      "step": 41052
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6905227303504944,
      "learning_rate": 0.0005542137896823322,
      "loss": 2.0735,
      "step": 41053
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.66887366771698,
      "learning_rate": 0.000554211617619014,
      "loss": 2.1243,
      "step": 41054
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.626431405544281,
      "learning_rate": 0.0005542094455084328,
      "loss": 2.466,
      "step": 41055
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6393661499023438,
      "learning_rate": 0.0005542072733505895,
      "loss": 2.4214,
      "step": 41056
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6833779811859131,
      "learning_rate": 0.0005542051011454841,
      "loss": 2.2661,
      "step": 41057
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.7069329023361206,
      "learning_rate": 0.0005542029288931173,
      "loss": 2.2834,
      "step": 41058
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6556822657585144,
      "learning_rate": 0.0005542007565934894,
      "loss": 2.3659,
      "step": 41059
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.7360222935676575,
      "learning_rate": 0.0005541985842466006,
      "loss": 2.1933,
      "step": 41060
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6781147122383118,
      "learning_rate": 0.0005541964118524516,
      "loss": 2.2484,
      "step": 41061
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.748098611831665,
      "learning_rate": 0.0005541942394110428,
      "loss": 2.2381,
      "step": 41062
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.7942196130752563,
      "learning_rate": 0.0005541920669223743,
      "loss": 2.1453,
      "step": 41063
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6591213941574097,
      "learning_rate": 0.0005541898943864468,
      "loss": 2.2075,
      "step": 41064
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6974844932556152,
      "learning_rate": 0.0005541877218032606,
      "loss": 2.052,
      "step": 41065
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6728929281234741,
      "learning_rate": 0.0005541855491728161,
      "loss": 2.1979,
      "step": 41066
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.7010693550109863,
      "learning_rate": 0.0005541833764951136,
      "loss": 2.0939,
      "step": 41067
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6844531297683716,
      "learning_rate": 0.0005541812037701538,
      "loss": 2.4581,
      "step": 41068
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6356546878814697,
      "learning_rate": 0.0005541790309979368,
      "loss": 2.2589,
      "step": 41069
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6247965097427368,
      "learning_rate": 0.0005541768581784632,
      "loss": 2.1999,
      "step": 41070
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6683622598648071,
      "learning_rate": 0.0005541746853117331,
      "loss": 2.417,
      "step": 41071
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.7941470742225647,
      "learning_rate": 0.0005541725123977473,
      "loss": 2.1035,
      "step": 41072
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.732874870300293,
      "learning_rate": 0.0005541703394365061,
      "loss": 2.3977,
      "step": 41073
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6220638155937195,
      "learning_rate": 0.0005541681664280097,
      "loss": 2.3032,
      "step": 41074
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6807776689529419,
      "learning_rate": 0.0005541659933722587,
      "loss": 2.5311,
      "step": 41075
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.7737424373626709,
      "learning_rate": 0.0005541638202692534,
      "loss": 2.1859,
      "step": 41076
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6412448883056641,
      "learning_rate": 0.0005541616471189943,
      "loss": 2.168,
      "step": 41077
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.7133446335792542,
      "learning_rate": 0.0005541594739214818,
      "loss": 2.3792,
      "step": 41078
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.7076550722122192,
      "learning_rate": 0.0005541573006767162,
      "loss": 2.2197,
      "step": 41079
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.7696231603622437,
      "learning_rate": 0.0005541551273846979,
      "loss": 2.1783,
      "step": 41080
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.7133907079696655,
      "learning_rate": 0.0005541529540454274,
      "loss": 2.4231,
      "step": 41081
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6994274258613586,
      "learning_rate": 0.0005541507806589051,
      "loss": 2.1804,
      "step": 41082
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.79635089635849,
      "learning_rate": 0.0005541486072251314,
      "loss": 2.2473,
      "step": 41083
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.7251141667366028,
      "learning_rate": 0.0005541464337441066,
      "loss": 2.3068,
      "step": 41084
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.7022745013237,
      "learning_rate": 0.0005541442602158312,
      "loss": 2.2763,
      "step": 41085
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.7439581155776978,
      "learning_rate": 0.0005541420866403057,
      "loss": 2.199,
      "step": 41086
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.8663318753242493,
      "learning_rate": 0.0005541399130175302,
      "loss": 2.379,
      "step": 41087
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6652877926826477,
      "learning_rate": 0.0005541377393475054,
      "loss": 2.2634,
      "step": 41088
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6796842813491821,
      "learning_rate": 0.0005541355656302317,
      "loss": 2.3284,
      "step": 41089
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.8177391886711121,
      "learning_rate": 0.0005541333918657092,
      "loss": 2.1119,
      "step": 41090
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.7075925469398499,
      "learning_rate": 0.0005541312180539385,
      "loss": 2.2065,
      "step": 41091
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.7348743081092834,
      "learning_rate": 0.0005541290441949202,
      "loss": 2.2376,
      "step": 41092
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.7561613917350769,
      "learning_rate": 0.0005541268702886543,
      "loss": 2.2218,
      "step": 41093
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.939929723739624,
      "learning_rate": 0.0005541246963351417,
      "loss": 2.3563,
      "step": 41094
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6807554364204407,
      "learning_rate": 0.0005541225223343822,
      "loss": 2.1895,
      "step": 41095
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6708826422691345,
      "learning_rate": 0.0005541203482863767,
      "loss": 2.138,
      "step": 41096
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.9650269746780396,
      "learning_rate": 0.0005541181741911254,
      "loss": 2.2159,
      "step": 41097
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.628058135509491,
      "learning_rate": 0.0005541160000486289,
      "loss": 2.1573,
      "step": 41098
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.9171163439750671,
      "learning_rate": 0.0005541138258588872,
      "loss": 2.2162,
      "step": 41099
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.7337902188301086,
      "learning_rate": 0.000554111651621901,
      "loss": 2.3161,
      "step": 41100
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.7164806723594666,
      "learning_rate": 0.0005541094773376707,
      "loss": 2.1237,
      "step": 41101
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6445800065994263,
      "learning_rate": 0.0005541073030061966,
      "loss": 2.2808,
      "step": 41102
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.7649118900299072,
      "learning_rate": 0.0005541051286274792,
      "loss": 2.2946,
      "step": 41103
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.7250392436981201,
      "learning_rate": 0.0005541029542015188,
      "loss": 2.4838,
      "step": 41104
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6814699769020081,
      "learning_rate": 0.000554100779728316,
      "loss": 2.3682,
      "step": 41105
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6894041299819946,
      "learning_rate": 0.0005540986052078708,
      "loss": 2.1411,
      "step": 41106
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.7254297733306885,
      "learning_rate": 0.0005540964306401841,
      "loss": 2.3433,
      "step": 41107
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6255182027816772,
      "learning_rate": 0.0005540942560252561,
      "loss": 2.3438,
      "step": 41108
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.674397349357605,
      "learning_rate": 0.0005540920813630871,
      "loss": 2.2557,
      "step": 41109
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.7181665897369385,
      "learning_rate": 0.0005540899066536775,
      "loss": 1.9654,
      "step": 41110
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6348194479942322,
      "learning_rate": 0.0005540877318970279,
      "loss": 2.0892,
      "step": 41111
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6797197461128235,
      "learning_rate": 0.0005540855570931386,
      "loss": 2.3833,
      "step": 41112
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6357208490371704,
      "learning_rate": 0.00055408338224201,
      "loss": 2.3844,
      "step": 41113
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6401693820953369,
      "learning_rate": 0.0005540812073436425,
      "loss": 2.5339,
      "step": 41114
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6885771751403809,
      "learning_rate": 0.0005540790323980365,
      "loss": 2.0975,
      "step": 41115
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.5976910591125488,
      "learning_rate": 0.0005540768574051923,
      "loss": 2.2982,
      "step": 41116
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.699543297290802,
      "learning_rate": 0.0005540746823651106,
      "loss": 2.2533,
      "step": 41117
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6494537591934204,
      "learning_rate": 0.0005540725072777915,
      "loss": 2.35,
      "step": 41118
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.636890709400177,
      "learning_rate": 0.0005540703321432355,
      "loss": 2.2348,
      "step": 41119
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.7810191512107849,
      "learning_rate": 0.0005540681569614431,
      "loss": 2.4875,
      "step": 41120
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.7844941020011902,
      "learning_rate": 0.0005540659817324147,
      "loss": 2.4812,
      "step": 41121
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6427976489067078,
      "learning_rate": 0.0005540638064561505,
      "loss": 2.2002,
      "step": 41122
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6517565846443176,
      "learning_rate": 0.000554061631132651,
      "loss": 2.2608,
      "step": 41123
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.7509560585021973,
      "learning_rate": 0.0005540594557619168,
      "loss": 2.4714,
      "step": 41124
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6451922059059143,
      "learning_rate": 0.0005540572803439481,
      "loss": 2.3106,
      "step": 41125
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6967496871948242,
      "learning_rate": 0.0005540551048787454,
      "loss": 2.1694,
      "step": 41126
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6541076898574829,
      "learning_rate": 0.000554052929366309,
      "loss": 2.2475,
      "step": 41127
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.7138645052909851,
      "learning_rate": 0.0005540507538066393,
      "loss": 2.3256,
      "step": 41128
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6309321522712708,
      "learning_rate": 0.0005540485781997369,
      "loss": 2.29,
      "step": 41129
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6885138154029846,
      "learning_rate": 0.0005540464025456021,
      "loss": 2.1418,
      "step": 41130
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6978387832641602,
      "learning_rate": 0.0005540442268442351,
      "loss": 2.3362,
      "step": 41131
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6752834916114807,
      "learning_rate": 0.0005540420510956367,
      "loss": 2.316,
      "step": 41132
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6970670223236084,
      "learning_rate": 0.0005540398752998068,
      "loss": 2.4178,
      "step": 41133
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.8268173336982727,
      "learning_rate": 0.0005540376994567463,
      "loss": 2.1543,
      "step": 41134
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6693944334983826,
      "learning_rate": 0.0005540355235664554,
      "loss": 2.1231,
      "step": 41135
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6613200306892395,
      "learning_rate": 0.0005540333476289344,
      "loss": 2.0482,
      "step": 41136
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.7267056107521057,
      "learning_rate": 0.000554031171644184,
      "loss": 2.3489,
      "step": 41137
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.7323919534683228,
      "learning_rate": 0.0005540289956122041,
      "loss": 2.2581,
      "step": 41138
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.7166581749916077,
      "learning_rate": 0.0005540268195329957,
      "loss": 2.1536,
      "step": 41139
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6470939517021179,
      "learning_rate": 0.0005540246434065588,
      "loss": 2.1199,
      "step": 41140
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6848522424697876,
      "learning_rate": 0.0005540224672328939,
      "loss": 2.3245,
      "step": 41141
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6981673836708069,
      "learning_rate": 0.0005540202910120015,
      "loss": 2.3254,
      "step": 41142
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6894050240516663,
      "learning_rate": 0.0005540181147438819,
      "loss": 2.4927,
      "step": 41143
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.7777244448661804,
      "learning_rate": 0.0005540159384285354,
      "loss": 2.2176,
      "step": 41144
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.7461414933204651,
      "learning_rate": 0.0005540137620659628,
      "loss": 2.3047,
      "step": 41145
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.7477971315383911,
      "learning_rate": 0.0005540115856561642,
      "loss": 2.225,
      "step": 41146
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.7121537923812866,
      "learning_rate": 0.0005540094091991399,
      "loss": 2.505,
      "step": 41147
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.8421118259429932,
      "learning_rate": 0.0005540072326948905,
      "loss": 2.2001,
      "step": 41148
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.8022881746292114,
      "learning_rate": 0.0005540050561434164,
      "loss": 2.2831,
      "step": 41149
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.7680308818817139,
      "learning_rate": 0.0005540028795447179,
      "loss": 2.3503,
      "step": 41150
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.7004534006118774,
      "learning_rate": 0.0005540007028987955,
      "loss": 2.1008,
      "step": 41151
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6971276998519897,
      "learning_rate": 0.0005539985262056497,
      "loss": 2.4246,
      "step": 41152
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.63764488697052,
      "learning_rate": 0.0005539963494652807,
      "loss": 2.3637,
      "step": 41153
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6787756085395813,
      "learning_rate": 0.0005539941726776889,
      "loss": 2.4052,
      "step": 41154
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.718527615070343,
      "learning_rate": 0.0005539919958428748,
      "loss": 2.593,
      "step": 41155
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.703419029712677,
      "learning_rate": 0.0005539898189608389,
      "loss": 2.3453,
      "step": 41156
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.7496581077575684,
      "learning_rate": 0.0005539876420315815,
      "loss": 2.2112,
      "step": 41157
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.7325448393821716,
      "learning_rate": 0.0005539854650551029,
      "loss": 2.2595,
      "step": 41158
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6916411519050598,
      "learning_rate": 0.0005539832880314036,
      "loss": 2.4282,
      "step": 41159
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6806238293647766,
      "learning_rate": 0.0005539811109604841,
      "loss": 2.4947,
      "step": 41160
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.7412835359573364,
      "learning_rate": 0.0005539789338423447,
      "loss": 2.301,
      "step": 41161
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6705780029296875,
      "learning_rate": 0.0005539767566769858,
      "loss": 2.1916,
      "step": 41162
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.633968710899353,
      "learning_rate": 0.0005539745794644078,
      "loss": 2.2277,
      "step": 41163
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.7562239766120911,
      "learning_rate": 0.0005539724022046112,
      "loss": 2.0464,
      "step": 41164
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.7001277208328247,
      "learning_rate": 0.0005539702248975963,
      "loss": 2.1171,
      "step": 41165
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.691890299320221,
      "learning_rate": 0.0005539680475433635,
      "loss": 2.1973,
      "step": 41166
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.7043986916542053,
      "learning_rate": 0.0005539658701419133,
      "loss": 2.0937,
      "step": 41167
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6721038222312927,
      "learning_rate": 0.0005539636926932461,
      "loss": 2.308,
      "step": 41168
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.7636759281158447,
      "learning_rate": 0.0005539615151973622,
      "loss": 2.2998,
      "step": 41169
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.7771124243736267,
      "learning_rate": 0.000553959337654262,
      "loss": 2.3208,
      "step": 41170
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6676696538925171,
      "learning_rate": 0.0005539571600639459,
      "loss": 2.2752,
      "step": 41171
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6648821830749512,
      "learning_rate": 0.0005539549824264146,
      "loss": 2.3109,
      "step": 41172
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6453331708908081,
      "learning_rate": 0.0005539528047416682,
      "loss": 2.3547,
      "step": 41173
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6624367833137512,
      "learning_rate": 0.0005539506270097072,
      "loss": 2.3936,
      "step": 41174
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.7101569175720215,
      "learning_rate": 0.0005539484492305319,
      "loss": 2.3417,
      "step": 41175
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.7058696150779724,
      "learning_rate": 0.0005539462714041428,
      "loss": 2.3439,
      "step": 41176
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6912406086921692,
      "learning_rate": 0.0005539440935305402,
      "loss": 2.2764,
      "step": 41177
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6594180464744568,
      "learning_rate": 0.0005539419156097247,
      "loss": 2.3275,
      "step": 41178
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.708721935749054,
      "learning_rate": 0.0005539397376416966,
      "loss": 2.3876,
      "step": 41179
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6457678079605103,
      "learning_rate": 0.0005539375596264564,
      "loss": 2.3054,
      "step": 41180
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6558716893196106,
      "learning_rate": 0.0005539353815640043,
      "loss": 2.1867,
      "step": 41181
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.7807238698005676,
      "learning_rate": 0.0005539332034543409,
      "loss": 2.2932,
      "step": 41182
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.7083799242973328,
      "learning_rate": 0.0005539310252974664,
      "loss": 2.2904,
      "step": 41183
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.618586540222168,
      "learning_rate": 0.0005539288470933814,
      "loss": 2.0074,
      "step": 41184
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.638900637626648,
      "learning_rate": 0.0005539266688420864,
      "loss": 2.3878,
      "step": 41185
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.7506908774375916,
      "learning_rate": 0.0005539244905435814,
      "loss": 2.1755,
      "step": 41186
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.717061460018158,
      "learning_rate": 0.0005539223121978672,
      "loss": 2.3324,
      "step": 41187
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.7103369832038879,
      "learning_rate": 0.0005539201338049439,
      "loss": 2.2735,
      "step": 41188
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.7541316747665405,
      "learning_rate": 0.0005539179553648121,
      "loss": 2.2734,
      "step": 41189
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6786980628967285,
      "learning_rate": 0.0005539157768774722,
      "loss": 2.3332,
      "step": 41190
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.7762269973754883,
      "learning_rate": 0.0005539135983429246,
      "loss": 2.4173,
      "step": 41191
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6159541010856628,
      "learning_rate": 0.0005539114197611696,
      "loss": 2.2602,
      "step": 41192
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6376085877418518,
      "learning_rate": 0.0005539092411322077,
      "loss": 2.1403,
      "step": 41193
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.7420288920402527,
      "learning_rate": 0.0005539070624560393,
      "loss": 2.2013,
      "step": 41194
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6779146790504456,
      "learning_rate": 0.0005539048837326648,
      "loss": 2.3524,
      "step": 41195
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6150272488594055,
      "learning_rate": 0.0005539027049620845,
      "loss": 2.2091,
      "step": 41196
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.7862902283668518,
      "learning_rate": 0.000553900526144299,
      "loss": 2.2633,
      "step": 41197
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6874133348464966,
      "learning_rate": 0.0005538983472793084,
      "loss": 2.3355,
      "step": 41198
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.809031069278717,
      "learning_rate": 0.0005538961683671136,
      "loss": 2.1165,
      "step": 41199
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.7251360416412354,
      "learning_rate": 0.0005538939894077145,
      "loss": 2.2141,
      "step": 41200
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6266138553619385,
      "learning_rate": 0.0005538918104011118,
      "loss": 2.3302,
      "step": 41201
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.7008433938026428,
      "learning_rate": 0.0005538896313473058,
      "loss": 2.433,
      "step": 41202
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.7292768955230713,
      "learning_rate": 0.0005538874522462969,
      "loss": 2.3071,
      "step": 41203
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.7201571464538574,
      "learning_rate": 0.0005538852730980855,
      "loss": 2.3767,
      "step": 41204
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6551132798194885,
      "learning_rate": 0.0005538830939026721,
      "loss": 2.1947,
      "step": 41205
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.793840229511261,
      "learning_rate": 0.000553880914660057,
      "loss": 2.2182,
      "step": 41206
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.7600812315940857,
      "learning_rate": 0.0005538787353702408,
      "loss": 2.3178,
      "step": 41207
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6404733657836914,
      "learning_rate": 0.0005538765560332235,
      "loss": 2.3996,
      "step": 41208
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6557099223136902,
      "learning_rate": 0.0005538743766490059,
      "loss": 2.2879,
      "step": 41209
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6172621846199036,
      "learning_rate": 0.0005538721972175882,
      "loss": 2.4116,
      "step": 41210
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.7033743262290955,
      "learning_rate": 0.0005538700177389709,
      "loss": 2.5647,
      "step": 41211
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.7046166062355042,
      "learning_rate": 0.0005538678382131545,
      "loss": 2.3177,
      "step": 41212
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6796086430549622,
      "learning_rate": 0.0005538656586401391,
      "loss": 2.1406,
      "step": 41213
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6908532381057739,
      "learning_rate": 0.0005538634790199253,
      "loss": 2.3429,
      "step": 41214
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.7803115844726562,
      "learning_rate": 0.0005538612993525136,
      "loss": 2.1942,
      "step": 41215
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6705235242843628,
      "learning_rate": 0.0005538591196379042,
      "loss": 2.407,
      "step": 41216
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6808181405067444,
      "learning_rate": 0.0005538569398760976,
      "loss": 2.371,
      "step": 41217
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.9984342455863953,
      "learning_rate": 0.0005538547600670941,
      "loss": 2.2767,
      "step": 41218
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6928049921989441,
      "learning_rate": 0.0005538525802108944,
      "loss": 2.5843,
      "step": 41219
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6747945547103882,
      "learning_rate": 0.0005538504003074987,
      "loss": 2.37,
      "step": 41220
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.7206078767776489,
      "learning_rate": 0.0005538482203569073,
      "loss": 2.0811,
      "step": 41221
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6903308033943176,
      "learning_rate": 0.0005538460403591208,
      "loss": 2.2489,
      "step": 41222
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6860519051551819,
      "learning_rate": 0.0005538438603141394,
      "loss": 2.0963,
      "step": 41223
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.7834827303886414,
      "learning_rate": 0.0005538416802219638,
      "loss": 2.5744,
      "step": 41224
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.7145708203315735,
      "learning_rate": 0.0005538395000825941,
      "loss": 2.4052,
      "step": 41225
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6612057089805603,
      "learning_rate": 0.0005538373198960311,
      "loss": 2.3931,
      "step": 41226
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.8877300024032593,
      "learning_rate": 0.0005538351396622747,
      "loss": 2.2352,
      "step": 41227
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6714639067649841,
      "learning_rate": 0.0005538329593813257,
      "loss": 2.1288,
      "step": 41228
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6491308808326721,
      "learning_rate": 0.0005538307790531842,
      "loss": 2.4198,
      "step": 41229
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6948254108428955,
      "learning_rate": 0.0005538285986778509,
      "loss": 2.4401,
      "step": 41230
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.7091322541236877,
      "learning_rate": 0.000553826418255326,
      "loss": 2.105,
      "step": 41231
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.7462610006332397,
      "learning_rate": 0.00055382423778561,
      "loss": 2.3215,
      "step": 41232
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6756764650344849,
      "learning_rate": 0.0005538220572687034,
      "loss": 2.23,
      "step": 41233
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.668971598148346,
      "learning_rate": 0.0005538198767046064,
      "loss": 2.4037,
      "step": 41234
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6745400428771973,
      "learning_rate": 0.0005538176960933193,
      "loss": 2.1807,
      "step": 41235
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.7853789329528809,
      "learning_rate": 0.000553815515434843,
      "loss": 2.4539,
      "step": 41236
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.724959671497345,
      "learning_rate": 0.0005538133347291774,
      "loss": 2.374,
      "step": 41237
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.7283960580825806,
      "learning_rate": 0.0005538111539763232,
      "loss": 2.237,
      "step": 41238
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6580329537391663,
      "learning_rate": 0.0005538089731762808,
      "loss": 2.4193,
      "step": 41239
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6951119899749756,
      "learning_rate": 0.0005538067923290504,
      "loss": 2.1937,
      "step": 41240
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6257993578910828,
      "learning_rate": 0.0005538046114346325,
      "loss": 2.0544,
      "step": 41241
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.711752712726593,
      "learning_rate": 0.0005538024304930277,
      "loss": 2.2369,
      "step": 41242
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6947631239891052,
      "learning_rate": 0.0005538002495042361,
      "loss": 2.3152,
      "step": 41243
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6649774312973022,
      "learning_rate": 0.0005537980684682582,
      "loss": 2.4126,
      "step": 41244
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.7691895365715027,
      "learning_rate": 0.0005537958873850946,
      "loss": 2.2146,
      "step": 41245
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6392555236816406,
      "learning_rate": 0.0005537937062547455,
      "loss": 2.2635,
      "step": 41246
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.7586721777915955,
      "learning_rate": 0.0005537915250772112,
      "loss": 2.2929,
      "step": 41247
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.688230574131012,
      "learning_rate": 0.0005537893438524923,
      "loss": 2.3658,
      "step": 41248
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6948111057281494,
      "learning_rate": 0.0005537871625805893,
      "loss": 2.0149,
      "step": 41249
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.7000382542610168,
      "learning_rate": 0.0005537849812615025,
      "loss": 2.0451,
      "step": 41250
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.8188217282295227,
      "learning_rate": 0.0005537827998952321,
      "loss": 2.2458,
      "step": 41251
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.7527724504470825,
      "learning_rate": 0.000553780618481779,
      "loss": 2.0841,
      "step": 41252
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.7763763070106506,
      "learning_rate": 0.000553778437021143,
      "loss": 2.29,
      "step": 41253
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.7169129252433777,
      "learning_rate": 0.0005537762555133249,
      "loss": 2.3168,
      "step": 41254
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6991714835166931,
      "learning_rate": 0.000553774073958325,
      "loss": 2.4036,
      "step": 41255
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.7979167103767395,
      "learning_rate": 0.0005537718923561437,
      "loss": 2.3886,
      "step": 41256
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.7969529628753662,
      "learning_rate": 0.0005537697107067815,
      "loss": 2.1878,
      "step": 41257
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6912667155265808,
      "learning_rate": 0.0005537675290102385,
      "loss": 2.28,
      "step": 41258
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6883258819580078,
      "learning_rate": 0.0005537653472665156,
      "loss": 2.2585,
      "step": 41259
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.7063941359519958,
      "learning_rate": 0.0005537631654756128,
      "loss": 2.4443,
      "step": 41260
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.7449245452880859,
      "learning_rate": 0.0005537609836375306,
      "loss": 2.4055,
      "step": 41261
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6765190958976746,
      "learning_rate": 0.0005537588017522695,
      "loss": 2.2839,
      "step": 41262
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6309977769851685,
      "learning_rate": 0.0005537566198198298,
      "loss": 2.3766,
      "step": 41263
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6599905490875244,
      "learning_rate": 0.0005537544378402121,
      "loss": 2.2023,
      "step": 41264
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.8313420414924622,
      "learning_rate": 0.0005537522558134165,
      "loss": 2.2153,
      "step": 41265
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.7954403758049011,
      "learning_rate": 0.0005537500737394436,
      "loss": 2.223,
      "step": 41266
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.7971444725990295,
      "learning_rate": 0.0005537478916182938,
      "loss": 2.3823,
      "step": 41267
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.7168864011764526,
      "learning_rate": 0.0005537457094499674,
      "loss": 2.3933,
      "step": 41268
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6832042932510376,
      "learning_rate": 0.000553743527234465,
      "loss": 2.3035,
      "step": 41269
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6729474663734436,
      "learning_rate": 0.0005537413449717868,
      "loss": 2.4974,
      "step": 41270
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.7527444362640381,
      "learning_rate": 0.0005537391626619333,
      "loss": 2.0508,
      "step": 41271
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6996421813964844,
      "learning_rate": 0.0005537369803049049,
      "loss": 2.1386,
      "step": 41272
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.774543821811676,
      "learning_rate": 0.0005537347979007021,
      "loss": 2.4238,
      "step": 41273
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6735937595367432,
      "learning_rate": 0.0005537326154493251,
      "loss": 2.5195,
      "step": 41274
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6491267085075378,
      "learning_rate": 0.0005537304329507745,
      "loss": 2.4012,
      "step": 41275
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6612878441810608,
      "learning_rate": 0.0005537282504050506,
      "loss": 2.352,
      "step": 41276
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6198475956916809,
      "learning_rate": 0.0005537260678121538,
      "loss": 2.1876,
      "step": 41277
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.8268315196037292,
      "learning_rate": 0.0005537238851720845,
      "loss": 2.4534,
      "step": 41278
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.7652065753936768,
      "learning_rate": 0.0005537217024848433,
      "loss": 2.3254,
      "step": 41279
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6459944248199463,
      "learning_rate": 0.0005537195197504302,
      "loss": 2.1298,
      "step": 41280
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6321645379066467,
      "learning_rate": 0.000553717336968846,
      "loss": 2.2451,
      "step": 41281
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6976720094680786,
      "learning_rate": 0.000553715154140091,
      "loss": 2.4008,
      "step": 41282
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6531875729560852,
      "learning_rate": 0.0005537129712641654,
      "loss": 2.1528,
      "step": 41283
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.7141768336296082,
      "learning_rate": 0.00055371078834107,
      "loss": 2.3315,
      "step": 41284
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6388683319091797,
      "learning_rate": 0.0005537086053708048,
      "loss": 2.098,
      "step": 41285
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6495742201805115,
      "learning_rate": 0.0005537064223533704,
      "loss": 2.2891,
      "step": 41286
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.7238900661468506,
      "learning_rate": 0.0005537042392887672,
      "loss": 2.3405,
      "step": 41287
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.677158534526825,
      "learning_rate": 0.0005537020561769957,
      "loss": 2.231,
      "step": 41288
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6712056398391724,
      "learning_rate": 0.0005536998730180561,
      "loss": 2.2478,
      "step": 41289
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.7954574823379517,
      "learning_rate": 0.0005536976898119489,
      "loss": 2.4074,
      "step": 41290
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6813507080078125,
      "learning_rate": 0.0005536955065586746,
      "loss": 2.2334,
      "step": 41291
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.7909543514251709,
      "learning_rate": 0.0005536933232582333,
      "loss": 2.3039,
      "step": 41292
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.7097969651222229,
      "learning_rate": 0.0005536911399106258,
      "loss": 2.3607,
      "step": 41293
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.7936986684799194,
      "learning_rate": 0.0005536889565158523,
      "loss": 2.1706,
      "step": 41294
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.7178007364273071,
      "learning_rate": 0.0005536867730739132,
      "loss": 2.2057,
      "step": 41295
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.7118842005729675,
      "learning_rate": 0.000553684589584809,
      "loss": 2.1128,
      "step": 41296
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.7568524479866028,
      "learning_rate": 0.00055368240604854,
      "loss": 2.316,
      "step": 41297
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.8090082406997681,
      "learning_rate": 0.0005536802224651067,
      "loss": 2.1939,
      "step": 41298
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6914533972740173,
      "learning_rate": 0.0005536780388345094,
      "loss": 2.2228,
      "step": 41299
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.7348207235336304,
      "learning_rate": 0.0005536758551567486,
      "loss": 2.0855,
      "step": 41300
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.7330595254898071,
      "learning_rate": 0.0005536736714318246,
      "loss": 2.3368,
      "step": 41301
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6653368473052979,
      "learning_rate": 0.0005536714876597379,
      "loss": 2.4291,
      "step": 41302
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.7214062213897705,
      "learning_rate": 0.000553669303840489,
      "loss": 2.1824,
      "step": 41303
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.7171985507011414,
      "learning_rate": 0.0005536671199740781,
      "loss": 2.3436,
      "step": 41304
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.7834936380386353,
      "learning_rate": 0.0005536649360605056,
      "loss": 1.9735,
      "step": 41305
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.7448120713233948,
      "learning_rate": 0.0005536627520997722,
      "loss": 2.3191,
      "step": 41306
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.695020854473114,
      "learning_rate": 0.000553660568091878,
      "loss": 2.3892,
      "step": 41307
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.673693060874939,
      "learning_rate": 0.0005536583840368236,
      "loss": 2.3156,
      "step": 41308
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.944007396697998,
      "learning_rate": 0.0005536561999346092,
      "loss": 2.2253,
      "step": 41309
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.7344761490821838,
      "learning_rate": 0.0005536540157852353,
      "loss": 2.1522,
      "step": 41310
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.8533888459205627,
      "learning_rate": 0.0005536518315887024,
      "loss": 2.2618,
      "step": 41311
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.0886738300323486,
      "learning_rate": 0.0005536496473450108,
      "loss": 2.1605,
      "step": 41312
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.9641183614730835,
      "learning_rate": 0.000553647463054161,
      "loss": 2.2159,
      "step": 41313
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.7177318334579468,
      "learning_rate": 0.0005536452787161533,
      "loss": 2.274,
      "step": 41314
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.080040693283081,
      "learning_rate": 0.0005536430943309883,
      "loss": 2.3244,
      "step": 41315
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.803714394569397,
      "learning_rate": 0.0005536409098986661,
      "loss": 2.3039,
      "step": 41316
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6507725119590759,
      "learning_rate": 0.0005536387254191872,
      "loss": 2.2568,
      "step": 41317
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.8691866397857666,
      "learning_rate": 0.0005536365408925521,
      "loss": 2.3866,
      "step": 41318
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.7145665287971497,
      "learning_rate": 0.0005536343563187614,
      "loss": 2.3391,
      "step": 41319
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6609121561050415,
      "learning_rate": 0.0005536321716978151,
      "loss": 2.0318,
      "step": 41320
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.7118706703186035,
      "learning_rate": 0.0005536299870297139,
      "loss": 2.1424,
      "step": 41321
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.7685236930847168,
      "learning_rate": 0.0005536278023144579,
      "loss": 2.1603,
      "step": 41322
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6668601632118225,
      "learning_rate": 0.0005536256175520479,
      "loss": 2.3676,
      "step": 41323
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.701994776725769,
      "learning_rate": 0.000553623432742484,
      "loss": 2.3914,
      "step": 41324
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.729459285736084,
      "learning_rate": 0.0005536212478857668,
      "loss": 2.2391,
      "step": 41325
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6882487535476685,
      "learning_rate": 0.0005536190629818966,
      "loss": 2.2763,
      "step": 41326
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6816298961639404,
      "learning_rate": 0.0005536168780308737,
      "loss": 2.1561,
      "step": 41327
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.7522366642951965,
      "learning_rate": 0.0005536146930326987,
      "loss": 2.3477,
      "step": 41328
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.668643057346344,
      "learning_rate": 0.0005536125079873721,
      "loss": 2.5148,
      "step": 41329
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.7774386405944824,
      "learning_rate": 0.000553610322894894,
      "loss": 2.3202,
      "step": 41330
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.7273107171058655,
      "learning_rate": 0.000553608137755265,
      "loss": 2.1921,
      "step": 41331
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6951687335968018,
      "learning_rate": 0.0005536059525684854,
      "loss": 2.1286,
      "step": 41332
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.7130635976791382,
      "learning_rate": 0.0005536037673345558,
      "loss": 2.3388,
      "step": 41333
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6834187507629395,
      "learning_rate": 0.0005536015820534764,
      "loss": 2.3054,
      "step": 41334
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6946737170219421,
      "learning_rate": 0.0005535993967252476,
      "loss": 2.1972,
      "step": 41335
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6758764982223511,
      "learning_rate": 0.00055359721134987,
      "loss": 2.0803,
      "step": 41336
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.725504994392395,
      "learning_rate": 0.0005535950259273439,
      "loss": 2.2819,
      "step": 41337
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.7273632884025574,
      "learning_rate": 0.0005535928404576695,
      "loss": 2.3991,
      "step": 41338
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.7338296175003052,
      "learning_rate": 0.0005535906549408476,
      "loss": 2.2347,
      "step": 41339
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.8689759373664856,
      "learning_rate": 0.0005535884693768783,
      "loss": 2.3852,
      "step": 41340
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.666766881942749,
      "learning_rate": 0.0005535862837657623,
      "loss": 2.4095,
      "step": 41341
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.7423827052116394,
      "learning_rate": 0.0005535840981074996,
      "loss": 2.5041,
      "step": 41342
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6742237210273743,
      "learning_rate": 0.000553581912402091,
      "loss": 2.2494,
      "step": 41343
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6870861053466797,
      "learning_rate": 0.0005535797266495366,
      "loss": 2.3665,
      "step": 41344
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6652631163597107,
      "learning_rate": 0.000553577540849837,
      "loss": 2.3874,
      "step": 41345
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.7312646508216858,
      "learning_rate": 0.0005535753550029925,
      "loss": 2.412,
      "step": 41346
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.7012574076652527,
      "learning_rate": 0.0005535731691090037,
      "loss": 2.2637,
      "step": 41347
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6038685441017151,
      "learning_rate": 0.0005535709831678707,
      "loss": 2.1876,
      "step": 41348
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.7927626371383667,
      "learning_rate": 0.0005535687971795941,
      "loss": 2.5444,
      "step": 41349
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.669960618019104,
      "learning_rate": 0.0005535666111441743,
      "loss": 2.3351,
      "step": 41350
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6287882328033447,
      "learning_rate": 0.0005535644250616117,
      "loss": 2.3678,
      "step": 41351
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.677425742149353,
      "learning_rate": 0.0005535622389319067,
      "loss": 2.2062,
      "step": 41352
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.7557121515274048,
      "learning_rate": 0.0005535600527550597,
      "loss": 2.351,
      "step": 41353
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6541540622711182,
      "learning_rate": 0.0005535578665310711,
      "loss": 2.175,
      "step": 41354
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6620411276817322,
      "learning_rate": 0.0005535556802599412,
      "loss": 2.3244,
      "step": 41355
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.668181836605072,
      "learning_rate": 0.0005535534939416706,
      "loss": 2.2949,
      "step": 41356
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6631242036819458,
      "learning_rate": 0.0005535513075762596,
      "loss": 2.2935,
      "step": 41357
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.7037243843078613,
      "learning_rate": 0.0005535491211637086,
      "loss": 2.2301,
      "step": 41358
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6592944860458374,
      "learning_rate": 0.0005535469347040181,
      "loss": 2.2156,
      "step": 41359
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6916143298149109,
      "learning_rate": 0.0005535447481971883,
      "loss": 2.0426,
      "step": 41360
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6921262741088867,
      "learning_rate": 0.00055354256164322,
      "loss": 2.4111,
      "step": 41361
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6887478232383728,
      "learning_rate": 0.0005535403750421131,
      "loss": 2.1418,
      "step": 41362
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6656447052955627,
      "learning_rate": 0.0005535381883938683,
      "loss": 2.1807,
      "step": 41363
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.705905556678772,
      "learning_rate": 0.0005535360016984862,
      "loss": 2.2005,
      "step": 41364
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6883155107498169,
      "learning_rate": 0.0005535338149559667,
      "loss": 2.3082,
      "step": 41365
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6937254667282104,
      "learning_rate": 0.0005535316281663105,
      "loss": 2.4257,
      "step": 41366
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.7216683030128479,
      "learning_rate": 0.0005535294413295182,
      "loss": 2.3498,
      "step": 41367
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6454392671585083,
      "learning_rate": 0.0005535272544455898,
      "loss": 2.1403,
      "step": 41368
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.7450825572013855,
      "learning_rate": 0.0005535250675145259,
      "loss": 2.2723,
      "step": 41369
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6546170711517334,
      "learning_rate": 0.000553522880536327,
      "loss": 2.2763,
      "step": 41370
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6900003552436829,
      "learning_rate": 0.0005535206935109933,
      "loss": 2.134,
      "step": 41371
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.7907087802886963,
      "learning_rate": 0.0005535185064385254,
      "loss": 2.228,
      "step": 41372
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.7871303558349609,
      "learning_rate": 0.0005535163193189236,
      "loss": 2.4266,
      "step": 41373
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.7654173374176025,
      "learning_rate": 0.0005535141321521883,
      "loss": 2.3003,
      "step": 41374
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.7034382820129395,
      "learning_rate": 0.0005535119449383201,
      "loss": 2.1619,
      "step": 41375
    },
    {
      "epoch": 0.54,
      "grad_norm": 2.0049188137054443,
      "learning_rate": 0.0005535097576773191,
      "loss": 2.3974,
      "step": 41376
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.685958981513977,
      "learning_rate": 0.0005535075703691858,
      "loss": 2.3121,
      "step": 41377
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.7486352920532227,
      "learning_rate": 0.0005535053830139207,
      "loss": 2.2208,
      "step": 41378
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.7600340247154236,
      "learning_rate": 0.0005535031956115242,
      "loss": 2.0319,
      "step": 41379
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6821401119232178,
      "learning_rate": 0.0005535010081619967,
      "loss": 2.5853,
      "step": 41380
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6549109220504761,
      "learning_rate": 0.0005534988206653385,
      "loss": 2.2785,
      "step": 41381
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.7190935015678406,
      "learning_rate": 0.0005534966331215501,
      "loss": 2.2127,
      "step": 41382
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.7043619155883789,
      "learning_rate": 0.000553494445530632,
      "loss": 2.1936,
      "step": 41383
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.7510155439376831,
      "learning_rate": 0.0005534922578925843,
      "loss": 2.4312,
      "step": 41384
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.660385012626648,
      "learning_rate": 0.0005534900702074077,
      "loss": 2.1641,
      "step": 41385
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.7002528309822083,
      "learning_rate": 0.0005534878824751026,
      "loss": 2.2708,
      "step": 41386
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.7319542169570923,
      "learning_rate": 0.0005534856946956692,
      "loss": 2.1644,
      "step": 41387
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.7066983580589294,
      "learning_rate": 0.000553483506869108,
      "loss": 2.221,
      "step": 41388
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.8188304901123047,
      "learning_rate": 0.0005534813189954196,
      "loss": 2.4957,
      "step": 41389
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.7356176972389221,
      "learning_rate": 0.0005534791310746041,
      "loss": 2.3289,
      "step": 41390
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6868988275527954,
      "learning_rate": 0.0005534769431066621,
      "loss": 2.4381,
      "step": 41391
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.7043784260749817,
      "learning_rate": 0.0005534747550915939,
      "loss": 2.4399,
      "step": 41392
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6850630044937134,
      "learning_rate": 0.0005534725670293999,
      "loss": 2.2619,
      "step": 41393
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.696341872215271,
      "learning_rate": 0.0005534703789200808,
      "loss": 2.2962,
      "step": 41394
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.7712381482124329,
      "learning_rate": 0.0005534681907636365,
      "loss": 2.5451,
      "step": 41395
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.7512627840042114,
      "learning_rate": 0.0005534660025600679,
      "loss": 2.1768,
      "step": 41396
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6432065963745117,
      "learning_rate": 0.000553463814309375,
      "loss": 2.339,
      "step": 41397
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6488487720489502,
      "learning_rate": 0.0005534616260115586,
      "loss": 2.2187,
      "step": 41398
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6991082429885864,
      "learning_rate": 0.0005534594376666187,
      "loss": 2.1021,
      "step": 41399
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6788536906242371,
      "learning_rate": 0.000553457249274556,
      "loss": 2.2312,
      "step": 41400
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6710082292556763,
      "learning_rate": 0.0005534550608353708,
      "loss": 2.1904,
      "step": 41401
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.7122067213058472,
      "learning_rate": 0.0005534528723490635,
      "loss": 2.3648,
      "step": 41402
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.8077697157859802,
      "learning_rate": 0.0005534506838156346,
      "loss": 2.2338,
      "step": 41403
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.757091760635376,
      "learning_rate": 0.0005534484952350844,
      "loss": 2.3543,
      "step": 41404
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.7057427167892456,
      "learning_rate": 0.0005534463066074133,
      "loss": 2.2647,
      "step": 41405
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6742790937423706,
      "learning_rate": 0.0005534441179326217,
      "loss": 2.1329,
      "step": 41406
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6643416881561279,
      "learning_rate": 0.0005534419292107101,
      "loss": 2.2241,
      "step": 41407
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.678070068359375,
      "learning_rate": 0.0005534397404416789,
      "loss": 2.3637,
      "step": 41408
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6905609965324402,
      "learning_rate": 0.0005534375516255285,
      "loss": 2.3452,
      "step": 41409
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.7598956823348999,
      "learning_rate": 0.0005534353627622593,
      "loss": 2.3226,
      "step": 41410
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.7639802694320679,
      "learning_rate": 0.0005534331738518716,
      "loss": 2.2837,
      "step": 41411
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6953903436660767,
      "learning_rate": 0.0005534309848943659,
      "loss": 2.4791,
      "step": 41412
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.693702220916748,
      "learning_rate": 0.0005534287958897426,
      "loss": 2.3597,
      "step": 41413
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6701464056968689,
      "learning_rate": 0.000553426606838002,
      "loss": 2.5701,
      "step": 41414
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6963911652565002,
      "learning_rate": 0.0005534244177391448,
      "loss": 2.2783,
      "step": 41415
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.7229307889938354,
      "learning_rate": 0.0005534222285931711,
      "loss": 2.0143,
      "step": 41416
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6594786643981934,
      "learning_rate": 0.0005534200394000814,
      "loss": 2.4624,
      "step": 41417
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.8020468950271606,
      "learning_rate": 0.0005534178501598763,
      "loss": 2.2844,
      "step": 41418
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6662436127662659,
      "learning_rate": 0.0005534156608725558,
      "loss": 2.3685,
      "step": 41419
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6742522120475769,
      "learning_rate": 0.0005534134715381206,
      "loss": 2.3741,
      "step": 41420
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.7849698662757874,
      "learning_rate": 0.0005534112821565712,
      "loss": 2.1548,
      "step": 41421
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.722801923751831,
      "learning_rate": 0.0005534090927279077,
      "loss": 2.5102,
      "step": 41422
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.647104799747467,
      "learning_rate": 0.0005534069032521307,
      "loss": 2.181,
      "step": 41423
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6952993869781494,
      "learning_rate": 0.0005534047137292407,
      "loss": 2.2577,
      "step": 41424
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.7553892135620117,
      "learning_rate": 0.0005534025241592379,
      "loss": 2.2217,
      "step": 41425
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6443173289299011,
      "learning_rate": 0.0005534003345421229,
      "loss": 2.4856,
      "step": 41426
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6578149795532227,
      "learning_rate": 0.0005533981448778958,
      "loss": 2.3162,
      "step": 41427
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.8555161952972412,
      "learning_rate": 0.0005533959551665572,
      "loss": 2.2755,
      "step": 41428
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.7087304592132568,
      "learning_rate": 0.0005533937654081077,
      "loss": 2.1169,
      "step": 41429
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6741294264793396,
      "learning_rate": 0.0005533915756025474,
      "loss": 2.253,
      "step": 41430
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6873598694801331,
      "learning_rate": 0.0005533893857498768,
      "loss": 2.1681,
      "step": 41431
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.7385127544403076,
      "learning_rate": 0.0005533871958500964,
      "loss": 2.0376,
      "step": 41432
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.7229893207550049,
      "learning_rate": 0.0005533850059032064,
      "loss": 2.5122,
      "step": 41433
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6421549916267395,
      "learning_rate": 0.0005533828159092074,
      "loss": 2.462,
      "step": 41434
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6588277220726013,
      "learning_rate": 0.0005533806258680998,
      "loss": 2.3365,
      "step": 41435
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.705818772315979,
      "learning_rate": 0.000553378435779884,
      "loss": 2.2461,
      "step": 41436
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6696246266365051,
      "learning_rate": 0.0005533762456445603,
      "loss": 2.3653,
      "step": 41437
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.7063809633255005,
      "learning_rate": 0.0005533740554621292,
      "loss": 2.2916,
      "step": 41438
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.7497774362564087,
      "learning_rate": 0.000553371865232591,
      "loss": 2.5083,
      "step": 41439
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6655479073524475,
      "learning_rate": 0.0005533696749559463,
      "loss": 2.1745,
      "step": 41440
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.681767463684082,
      "learning_rate": 0.0005533674846321953,
      "loss": 2.0508,
      "step": 41441
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.7684099078178406,
      "learning_rate": 0.0005533652942613386,
      "loss": 2.1943,
      "step": 41442
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6746224164962769,
      "learning_rate": 0.0005533631038433765,
      "loss": 2.4441,
      "step": 41443
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6601641178131104,
      "learning_rate": 0.0005533609133783093,
      "loss": 2.2875,
      "step": 41444
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.7128636837005615,
      "learning_rate": 0.0005533587228661376,
      "loss": 2.3312,
      "step": 41445
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6477115154266357,
      "learning_rate": 0.0005533565323068617,
      "loss": 2.2706,
      "step": 41446
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6839130520820618,
      "learning_rate": 0.0005533543417004821,
      "loss": 2.4047,
      "step": 41447
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6226806640625,
      "learning_rate": 0.0005533521510469991,
      "loss": 2.0257,
      "step": 41448
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.7319103479385376,
      "learning_rate": 0.0005533499603464131,
      "loss": 2.4279,
      "step": 41449
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6944910287857056,
      "learning_rate": 0.0005533477695987247,
      "loss": 2.2711,
      "step": 41450
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.9186910390853882,
      "learning_rate": 0.0005533455788039342,
      "loss": 2.6761,
      "step": 41451
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6220367550849915,
      "learning_rate": 0.0005533433879620417,
      "loss": 2.2993,
      "step": 41452
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6320483684539795,
      "learning_rate": 0.0005533411970730481,
      "loss": 2.1837,
      "step": 41453
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.8277488350868225,
      "learning_rate": 0.0005533390061369536,
      "loss": 2.3831,
      "step": 41454
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6244562864303589,
      "learning_rate": 0.0005533368151537585,
      "loss": 2.0624,
      "step": 41455
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.7313286066055298,
      "learning_rate": 0.0005533346241234632,
      "loss": 2.1057,
      "step": 41456
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6984906792640686,
      "learning_rate": 0.0005533324330460684,
      "loss": 2.4342,
      "step": 41457
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.7702509164810181,
      "learning_rate": 0.0005533302419215743,
      "loss": 2.3796,
      "step": 41458
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6951795816421509,
      "learning_rate": 0.0005533280507499813,
      "loss": 2.3881,
      "step": 41459
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6783215999603271,
      "learning_rate": 0.0005533258595312899,
      "loss": 2.2583,
      "step": 41460
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6656495332717896,
      "learning_rate": 0.0005533236682655003,
      "loss": 2.1715,
      "step": 41461
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.7268514037132263,
      "learning_rate": 0.0005533214769526132,
      "loss": 2.2249,
      "step": 41462
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.9528253674507141,
      "learning_rate": 0.0005533192855926288,
      "loss": 2.1906,
      "step": 41463
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.7400808334350586,
      "learning_rate": 0.0005533170941855475,
      "loss": 2.3025,
      "step": 41464
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.9076765179634094,
      "learning_rate": 0.0005533149027313698,
      "loss": 2.2057,
      "step": 41465
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.8742280602455139,
      "learning_rate": 0.0005533127112300961,
      "loss": 2.3708,
      "step": 41466
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.671520471572876,
      "learning_rate": 0.0005533105196817268,
      "loss": 2.4204,
      "step": 41467
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6640814542770386,
      "learning_rate": 0.0005533083280862622,
      "loss": 2.4385,
      "step": 41468
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.7624460458755493,
      "learning_rate": 0.0005533061364437028,
      "loss": 2.2471,
      "step": 41469
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.7262119650840759,
      "learning_rate": 0.0005533039447540492,
      "loss": 2.2443,
      "step": 41470
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.5975128412246704,
      "learning_rate": 0.0005533017530173015,
      "loss": 2.2329,
      "step": 41471
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6427943110466003,
      "learning_rate": 0.0005532995612334601,
      "loss": 2.3409,
      "step": 41472
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6910547018051147,
      "learning_rate": 0.0005532973694025256,
      "loss": 2.1247,
      "step": 41473
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6663945913314819,
      "learning_rate": 0.0005532951775244984,
      "loss": 2.3585,
      "step": 41474
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6160604357719421,
      "learning_rate": 0.0005532929855993788,
      "loss": 2.3675,
      "step": 41475
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.7684045433998108,
      "learning_rate": 0.0005532907936271674,
      "loss": 2.378,
      "step": 41476
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6753557324409485,
      "learning_rate": 0.0005532886016078643,
      "loss": 2.2479,
      "step": 41477
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.7264819145202637,
      "learning_rate": 0.0005532864095414701,
      "loss": 2.3061,
      "step": 41478
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6801318526268005,
      "learning_rate": 0.0005532842174279851,
      "loss": 1.9275,
      "step": 41479
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6662907004356384,
      "learning_rate": 0.0005532820252674099,
      "loss": 2.1941,
      "step": 41480
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6829830408096313,
      "learning_rate": 0.0005532798330597448,
      "loss": 2.0771,
      "step": 41481
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.668575644493103,
      "learning_rate": 0.0005532776408049901,
      "loss": 2.1716,
      "step": 41482
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6929787397384644,
      "learning_rate": 0.0005532754485031464,
      "loss": 2.3387,
      "step": 41483
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.7156487703323364,
      "learning_rate": 0.000553273256154214,
      "loss": 2.1413,
      "step": 41484
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.688301682472229,
      "learning_rate": 0.0005532710637581932,
      "loss": 2.3627,
      "step": 41485
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6078683137893677,
      "learning_rate": 0.0005532688713150847,
      "loss": 2.4421,
      "step": 41486
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.7230920791625977,
      "learning_rate": 0.0005532666788248886,
      "loss": 2.3952,
      "step": 41487
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.7283351421356201,
      "learning_rate": 0.0005532644862876055,
      "loss": 2.4,
      "step": 41488
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6934891939163208,
      "learning_rate": 0.0005532622937032358,
      "loss": 2.3359,
      "step": 41489
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6403757929801941,
      "learning_rate": 0.0005532601010717798,
      "loss": 2.3237,
      "step": 41490
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6751313805580139,
      "learning_rate": 0.0005532579083932379,
      "loss": 2.3321,
      "step": 41491
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6935389637947083,
      "learning_rate": 0.0005532557156676107,
      "loss": 2.5748,
      "step": 41492
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.7401835918426514,
      "learning_rate": 0.0005532535228948985,
      "loss": 2.2545,
      "step": 41493
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.713791012763977,
      "learning_rate": 0.0005532513300751015,
      "loss": 2.32,
      "step": 41494
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.7335045337677002,
      "learning_rate": 0.0005532491372082205,
      "loss": 1.9633,
      "step": 41495
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.7993860244750977,
      "learning_rate": 0.0005532469442942555,
      "loss": 2.2797,
      "step": 41496
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6687060594558716,
      "learning_rate": 0.0005532447513332073,
      "loss": 2.346,
      "step": 41497
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6907631754875183,
      "learning_rate": 0.0005532425583250761,
      "loss": 2.3234,
      "step": 41498
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6505352854728699,
      "learning_rate": 0.0005532403652698623,
      "loss": 2.3504,
      "step": 41499
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.7016174793243408,
      "learning_rate": 0.0005532381721675662,
      "loss": 2.078,
      "step": 41500
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6822671294212341,
      "learning_rate": 0.0005532359790181886,
      "loss": 2.2463,
      "step": 41501
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.730480968952179,
      "learning_rate": 0.0005532337858217294,
      "loss": 2.3851,
      "step": 41502
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6734231114387512,
      "learning_rate": 0.0005532315925781895,
      "loss": 2.1588,
      "step": 41503
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6551550030708313,
      "learning_rate": 0.0005532293992875689,
      "loss": 2.2917,
      "step": 41504
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.7112996578216553,
      "learning_rate": 0.0005532272059498682,
      "loss": 2.2395,
      "step": 41505
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6232491135597229,
      "learning_rate": 0.0005532250125650878,
      "loss": 2.182,
      "step": 41506
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.849410355091095,
      "learning_rate": 0.0005532228191332281,
      "loss": 2.2182,
      "step": 41507
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.738882839679718,
      "learning_rate": 0.0005532206256542894,
      "loss": 2.2777,
      "step": 41508
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.7276636362075806,
      "learning_rate": 0.0005532184321282724,
      "loss": 2.5436,
      "step": 41509
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.7701058983802795,
      "learning_rate": 0.0005532162385551772,
      "loss": 2.4552,
      "step": 41510
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6996845602989197,
      "learning_rate": 0.0005532140449350043,
      "loss": 2.2946,
      "step": 41511
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.784711480140686,
      "learning_rate": 0.0005532118512677542,
      "loss": 2.2975,
      "step": 41512
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.7419773936271667,
      "learning_rate": 0.0005532096575534271,
      "loss": 2.5879,
      "step": 41513
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.7856775522232056,
      "learning_rate": 0.0005532074637920237,
      "loss": 2.3768,
      "step": 41514
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.7119930386543274,
      "learning_rate": 0.0005532052699835442,
      "loss": 2.2142,
      "step": 41515
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.85357266664505,
      "learning_rate": 0.000553203076127989,
      "loss": 2.3016,
      "step": 41516
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.7687886953353882,
      "learning_rate": 0.0005532008822253586,
      "loss": 2.3237,
      "step": 41517
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.7688018679618835,
      "learning_rate": 0.0005531986882756535,
      "loss": 2.2802,
      "step": 41518
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.7364839911460876,
      "learning_rate": 0.0005531964942788737,
      "loss": 2.0813,
      "step": 41519
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.9418123364448547,
      "learning_rate": 0.0005531943002350201,
      "loss": 2.3801,
      "step": 41520
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6983370780944824,
      "learning_rate": 0.000553192106144093,
      "loss": 2.2112,
      "step": 41521
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.7930088043212891,
      "learning_rate": 0.0005531899120060924,
      "loss": 2.5224,
      "step": 41522
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.8515605330467224,
      "learning_rate": 0.0005531877178210193,
      "loss": 2.0654,
      "step": 41523
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.7731490135192871,
      "learning_rate": 0.0005531855235888736,
      "loss": 2.3124,
      "step": 41524
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6664767861366272,
      "learning_rate": 0.0005531833293096561,
      "loss": 2.0298,
      "step": 41525
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.7594797611236572,
      "learning_rate": 0.0005531811349833671,
      "loss": 2.1447,
      "step": 41526
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.7532381415367126,
      "learning_rate": 0.0005531789406100069,
      "loss": 2.4585,
      "step": 41527
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.7471349835395813,
      "learning_rate": 0.0005531767461895758,
      "loss": 2.3591,
      "step": 41528
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.653581976890564,
      "learning_rate": 0.0005531745517220744,
      "loss": 2.2535,
      "step": 41529
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.7593129873275757,
      "learning_rate": 0.0005531723572075032,
      "loss": 2.1312,
      "step": 41530
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.7190727591514587,
      "learning_rate": 0.0005531701626458623,
      "loss": 2.2763,
      "step": 41531
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6818674206733704,
      "learning_rate": 0.0005531679680371525,
      "loss": 2.3402,
      "step": 41532
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.7395414113998413,
      "learning_rate": 0.0005531657733813739,
      "loss": 2.0695,
      "step": 41533
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6871117949485779,
      "learning_rate": 0.000553163578678527,
      "loss": 2.2421,
      "step": 41534
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6876775622367859,
      "learning_rate": 0.0005531613839286121,
      "loss": 2.3409,
      "step": 41535
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6848615407943726,
      "learning_rate": 0.00055315918913163,
      "loss": 2.3161,
      "step": 41536
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.8057745099067688,
      "learning_rate": 0.0005531569942875805,
      "loss": 2.1239,
      "step": 41537
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6993764638900757,
      "learning_rate": 0.0005531547993964646,
      "loss": 2.2901,
      "step": 41538
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.7529723644256592,
      "learning_rate": 0.0005531526044582824,
      "loss": 2.4537,
      "step": 41539
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6520448327064514,
      "learning_rate": 0.0005531504094730342,
      "loss": 2.2061,
      "step": 41540
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.7636060118675232,
      "learning_rate": 0.0005531482144407207,
      "loss": 2.2965,
      "step": 41541
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.7874470949172974,
      "learning_rate": 0.0005531460193613421,
      "loss": 2.3277,
      "step": 41542
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6829655766487122,
      "learning_rate": 0.0005531438242348988,
      "loss": 2.5382,
      "step": 41543
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.7066493630409241,
      "learning_rate": 0.0005531416290613915,
      "loss": 2.3036,
      "step": 41544
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6767809987068176,
      "learning_rate": 0.0005531394338408203,
      "loss": 2.1551,
      "step": 41545
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6528900265693665,
      "learning_rate": 0.0005531372385731856,
      "loss": 2.2571,
      "step": 41546
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.7122672200202942,
      "learning_rate": 0.000553135043258488,
      "loss": 2.234,
      "step": 41547
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.7218824028968811,
      "learning_rate": 0.0005531328478967278,
      "loss": 2.4583,
      "step": 41548
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.7632819414138794,
      "learning_rate": 0.0005531306524879054,
      "loss": 2.1146,
      "step": 41549
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.7492988705635071,
      "learning_rate": 0.0005531284570320212,
      "loss": 2.2207,
      "step": 41550
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.7118426561355591,
      "learning_rate": 0.0005531262615290758,
      "loss": 2.1967,
      "step": 41551
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.7241488099098206,
      "learning_rate": 0.0005531240659790693,
      "loss": 2.2962,
      "step": 41552
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.7351499199867249,
      "learning_rate": 0.0005531218703820023,
      "loss": 2.3868,
      "step": 41553
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6977037787437439,
      "learning_rate": 0.0005531196747378752,
      "loss": 2.1733,
      "step": 41554
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6587879061698914,
      "learning_rate": 0.0005531174790466884,
      "loss": 1.9767,
      "step": 41555
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.631870448589325,
      "learning_rate": 0.0005531152833084422,
      "loss": 2.513,
      "step": 41556
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.7560025453567505,
      "learning_rate": 0.0005531130875231371,
      "loss": 2.4247,
      "step": 41557
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.7311245799064636,
      "learning_rate": 0.0005531108916907735,
      "loss": 2.4971,
      "step": 41558
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.7086047530174255,
      "learning_rate": 0.0005531086958113519,
      "loss": 2.0999,
      "step": 41559
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6943731904029846,
      "learning_rate": 0.0005531064998848726,
      "loss": 2.0419,
      "step": 41560
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.7144019603729248,
      "learning_rate": 0.0005531043039113359,
      "loss": 2.4634,
      "step": 41561
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6747415065765381,
      "learning_rate": 0.0005531021078907424,
      "loss": 2.3095,
      "step": 41562
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6964226365089417,
      "learning_rate": 0.0005530999118230924,
      "loss": 2.3146,
      "step": 41563
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6329938173294067,
      "learning_rate": 0.0005530977157083864,
      "loss": 2.3182,
      "step": 41564
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6966624855995178,
      "learning_rate": 0.0005530955195466248,
      "loss": 2.1532,
      "step": 41565
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.7083837389945984,
      "learning_rate": 0.0005530933233378079,
      "loss": 2.4522,
      "step": 41566
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.7032982110977173,
      "learning_rate": 0.0005530911270819361,
      "loss": 2.4378,
      "step": 41567
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6717379689216614,
      "learning_rate": 0.00055308893077901,
      "loss": 2.2358,
      "step": 41568
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.7228714227676392,
      "learning_rate": 0.0005530867344290299,
      "loss": 2.4491,
      "step": 41569
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6627957820892334,
      "learning_rate": 0.0005530845380319961,
      "loss": 2.3207,
      "step": 41570
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6816267967224121,
      "learning_rate": 0.0005530823415879092,
      "loss": 1.9662,
      "step": 41571
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.8117899894714355,
      "learning_rate": 0.0005530801450967694,
      "loss": 2.3255,
      "step": 41572
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6382364630699158,
      "learning_rate": 0.0005530779485585774,
      "loss": 1.9213,
      "step": 41573
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.7080362439155579,
      "learning_rate": 0.0005530757519733333,
      "loss": 2.055,
      "step": 41574
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.7537822723388672,
      "learning_rate": 0.0005530735553410377,
      "loss": 2.226,
      "step": 41575
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6920315623283386,
      "learning_rate": 0.000553071358661691,
      "loss": 2.4762,
      "step": 41576
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.7994702458381653,
      "learning_rate": 0.0005530691619352934,
      "loss": 2.3342,
      "step": 41577
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.7257805466651917,
      "learning_rate": 0.0005530669651618456,
      "loss": 2.3368,
      "step": 41578
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.7496269345283508,
      "learning_rate": 0.0005530647683413478,
      "loss": 2.1827,
      "step": 41579
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.7640451192855835,
      "learning_rate": 0.0005530625714738005,
      "loss": 2.3987,
      "step": 41580
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6760320663452148,
      "learning_rate": 0.0005530603745592041,
      "loss": 2.3522,
      "step": 41581
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6237241625785828,
      "learning_rate": 0.000553058177597559,
      "loss": 2.4829,
      "step": 41582
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6660571098327637,
      "learning_rate": 0.0005530559805888656,
      "loss": 2.3794,
      "step": 41583
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.7473587393760681,
      "learning_rate": 0.0005530537835331243,
      "loss": 2.3253,
      "step": 41584
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6705911755561829,
      "learning_rate": 0.0005530515864303355,
      "loss": 2.35,
      "step": 41585
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6329800486564636,
      "learning_rate": 0.0005530493892804998,
      "loss": 2.2223,
      "step": 41586
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.661146879196167,
      "learning_rate": 0.0005530471920836172,
      "loss": 2.2403,
      "step": 41587
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.7082335352897644,
      "learning_rate": 0.0005530449948396885,
      "loss": 2.1916,
      "step": 41588
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.7131029367446899,
      "learning_rate": 0.0005530427975487139,
      "loss": 2.322,
      "step": 41589
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.7766759991645813,
      "learning_rate": 0.000553040600210694,
      "loss": 2.284,
      "step": 41590
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6993991136550903,
      "learning_rate": 0.0005530384028256289,
      "loss": 2.3134,
      "step": 41591
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6607526540756226,
      "learning_rate": 0.0005530362053935193,
      "loss": 2.3029,
      "step": 41592
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.703345775604248,
      "learning_rate": 0.0005530340079143654,
      "loss": 2.4765,
      "step": 41593
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.7107737064361572,
      "learning_rate": 0.0005530318103881678,
      "loss": 2.2323,
      "step": 41594
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.7215384244918823,
      "learning_rate": 0.0005530296128149268,
      "loss": 2.2627,
      "step": 41595
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6550741195678711,
      "learning_rate": 0.0005530274151946428,
      "loss": 2.3158,
      "step": 41596
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.696714460849762,
      "learning_rate": 0.0005530252175273161,
      "loss": 2.1293,
      "step": 41597
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6312780976295471,
      "learning_rate": 0.0005530230198129473,
      "loss": 2.2792,
      "step": 41598
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.7634027600288391,
      "learning_rate": 0.0005530208220515369,
      "loss": 2.1956,
      "step": 41599
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6663509607315063,
      "learning_rate": 0.000553018624243085,
      "loss": 2.1992,
      "step": 41600
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.7512010931968689,
      "learning_rate": 0.0005530164263875922,
      "loss": 2.169,
      "step": 41601
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6233484148979187,
      "learning_rate": 0.0005530142284850589,
      "loss": 2.2619,
      "step": 41602
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6345477104187012,
      "learning_rate": 0.0005530120305354855,
      "loss": 2.3924,
      "step": 41603
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.674333930015564,
      "learning_rate": 0.0005530098325388722,
      "loss": 2.2967,
      "step": 41604
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.7155988216400146,
      "learning_rate": 0.0005530076344952198,
      "loss": 2.2162,
      "step": 41605
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.9023895859718323,
      "learning_rate": 0.0005530054364045285,
      "loss": 2.4006,
      "step": 41606
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6698325276374817,
      "learning_rate": 0.0005530032382667986,
      "loss": 2.3595,
      "step": 41607
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.768100917339325,
      "learning_rate": 0.0005530010400820307,
      "loss": 2.2049,
      "step": 41608
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.7385203838348389,
      "learning_rate": 0.000552998841850225,
      "loss": 2.2274,
      "step": 41609
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6568924784660339,
      "learning_rate": 0.0005529966435713822,
      "loss": 2.2508,
      "step": 41610
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.759987473487854,
      "learning_rate": 0.0005529944452455025,
      "loss": 2.3459,
      "step": 41611
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.7004620432853699,
      "learning_rate": 0.0005529922468725864,
      "loss": 2.3269,
      "step": 41612
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.800417959690094,
      "learning_rate": 0.0005529900484526342,
      "loss": 2.0753,
      "step": 41613
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.7460920810699463,
      "learning_rate": 0.0005529878499856463,
      "loss": 2.2661,
      "step": 41614
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.7424648404121399,
      "learning_rate": 0.0005529856514716234,
      "loss": 2.4734,
      "step": 41615
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.7367540597915649,
      "learning_rate": 0.0005529834529105656,
      "loss": 2.1258,
      "step": 41616
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.7161271572113037,
      "learning_rate": 0.0005529812543024733,
      "loss": 2.4047,
      "step": 41617
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6820421814918518,
      "learning_rate": 0.0005529790556473471,
      "loss": 2.3545,
      "step": 41618
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.7421757578849792,
      "learning_rate": 0.0005529768569451873,
      "loss": 2.5024,
      "step": 41619
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.686444878578186,
      "learning_rate": 0.0005529746581959943,
      "loss": 2.3349,
      "step": 41620
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.8025076985359192,
      "learning_rate": 0.0005529724593997685,
      "loss": 2.2124,
      "step": 41621
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.7347205877304077,
      "learning_rate": 0.0005529702605565105,
      "loss": 2.4214,
      "step": 41622
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6956730484962463,
      "learning_rate": 0.0005529680616662203,
      "loss": 2.1877,
      "step": 41623
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.9415897727012634,
      "learning_rate": 0.0005529658627288988,
      "loss": 2.3645,
      "step": 41624
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.7682538628578186,
      "learning_rate": 0.000552963663744546,
      "loss": 2.3113,
      "step": 41625
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6770060658454895,
      "learning_rate": 0.0005529614647131627,
      "loss": 2.2266,
      "step": 41626
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.0114291906356812,
      "learning_rate": 0.0005529592656347489,
      "loss": 2.2534,
      "step": 41627
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6688549518585205,
      "learning_rate": 0.0005529570665093053,
      "loss": 2.3299,
      "step": 41628
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.890518069267273,
      "learning_rate": 0.0005529548673368322,
      "loss": 2.0546,
      "step": 41629
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6222070455551147,
      "learning_rate": 0.0005529526681173299,
      "loss": 2.1852,
      "step": 41630
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6621318459510803,
      "learning_rate": 0.0005529504688507991,
      "loss": 2.2214,
      "step": 41631
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.7606369853019714,
      "learning_rate": 0.0005529482695372399,
      "loss": 2.0646,
      "step": 41632
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6772820949554443,
      "learning_rate": 0.0005529460701766529,
      "loss": 2.3375,
      "step": 41633
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6815006136894226,
      "learning_rate": 0.0005529438707690384,
      "loss": 2.388,
      "step": 41634
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6934979557991028,
      "learning_rate": 0.0005529416713143969,
      "loss": 2.1655,
      "step": 41635
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.7956241965293884,
      "learning_rate": 0.0005529394718127287,
      "loss": 2.4482,
      "step": 41636
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6060658097267151,
      "learning_rate": 0.0005529372722640343,
      "loss": 2.1175,
      "step": 41637
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6831956505775452,
      "learning_rate": 0.0005529350726683142,
      "loss": 2.2546,
      "step": 41638
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.7360616326332092,
      "learning_rate": 0.0005529328730255686,
      "loss": 2.2646,
      "step": 41639
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.0071407556533813,
      "learning_rate": 0.000552930673335798,
      "loss": 2.3434,
      "step": 41640
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.70285564661026,
      "learning_rate": 0.0005529284735990028,
      "loss": 2.2156,
      "step": 41641
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.7708256840705872,
      "learning_rate": 0.0005529262738151836,
      "loss": 2.061,
      "step": 41642
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6899302005767822,
      "learning_rate": 0.0005529240739843403,
      "loss": 2.3202,
      "step": 41643
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6908610463142395,
      "learning_rate": 0.000552921874106474,
      "loss": 2.1813,
      "step": 41644
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6709072589874268,
      "learning_rate": 0.0005529196741815845,
      "loss": 2.1729,
      "step": 41645
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.755388617515564,
      "learning_rate": 0.0005529174742096725,
      "loss": 2.1264,
      "step": 41646
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6603060960769653,
      "learning_rate": 0.0005529152741907384,
      "loss": 2.29,
      "step": 41647
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6926239132881165,
      "learning_rate": 0.0005529130741247826,
      "loss": 2.3181,
      "step": 41648
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.7065444588661194,
      "learning_rate": 0.0005529108740118055,
      "loss": 2.2215,
      "step": 41649
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.7568498849868774,
      "learning_rate": 0.0005529086738518074,
      "loss": 2.2673,
      "step": 41650
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.7117425799369812,
      "learning_rate": 0.0005529064736447889,
      "loss": 2.3072,
      "step": 41651
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.7240858674049377,
      "learning_rate": 0.0005529042733907503,
      "loss": 2.2021,
      "step": 41652
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.7618235945701599,
      "learning_rate": 0.000552902073089692,
      "loss": 2.3516,
      "step": 41653
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6978466510772705,
      "learning_rate": 0.0005528998727416144,
      "loss": 2.0821,
      "step": 41654
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.7083352208137512,
      "learning_rate": 0.000552897672346518,
      "loss": 2.2883,
      "step": 41655
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.677259087562561,
      "learning_rate": 0.0005528954719044031,
      "loss": 2.1532,
      "step": 41656
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6699725389480591,
      "learning_rate": 0.0005528932714152702,
      "loss": 2.3377,
      "step": 41657
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.7641085982322693,
      "learning_rate": 0.0005528910708791195,
      "loss": 2.1123,
      "step": 41658
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.7671491503715515,
      "learning_rate": 0.0005528888702959518,
      "loss": 2.2197,
      "step": 41659
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.71574866771698,
      "learning_rate": 0.0005528866696657672,
      "loss": 2.2787,
      "step": 41660
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.7793870568275452,
      "learning_rate": 0.0005528844689885662,
      "loss": 2.1276,
      "step": 41661
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.794909656047821,
      "learning_rate": 0.0005528822682643492,
      "loss": 2.3831,
      "step": 41662
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.7397525906562805,
      "learning_rate": 0.0005528800674931166,
      "loss": 2.2898,
      "step": 41663
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.7438803315162659,
      "learning_rate": 0.0005528778666748688,
      "loss": 2.2662,
      "step": 41664
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6973676681518555,
      "learning_rate": 0.0005528756658096063,
      "loss": 2.2851,
      "step": 41665
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.7296698689460754,
      "learning_rate": 0.0005528734648973295,
      "loss": 2.1624,
      "step": 41666
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.7368128895759583,
      "learning_rate": 0.0005528712639380386,
      "loss": 2.2752,
      "step": 41667
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.7971636056900024,
      "learning_rate": 0.0005528690629317343,
      "loss": 2.3846,
      "step": 41668
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.7187473177909851,
      "learning_rate": 0.0005528668618784167,
      "loss": 2.4022,
      "step": 41669
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.7000054717063904,
      "learning_rate": 0.0005528646607780865,
      "loss": 2.3671,
      "step": 41670
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.7727893590927124,
      "learning_rate": 0.0005528624596307439,
      "loss": 2.1919,
      "step": 41671
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6915243268013,
      "learning_rate": 0.0005528602584363894,
      "loss": 2.245,
      "step": 41672
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.7213504314422607,
      "learning_rate": 0.0005528580571950235,
      "loss": 2.2937,
      "step": 41673
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6680023670196533,
      "learning_rate": 0.0005528558559066464,
      "loss": 2.1638,
      "step": 41674
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.7520648837089539,
      "learning_rate": 0.0005528536545712587,
      "loss": 2.2353,
      "step": 41675
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6524231433868408,
      "learning_rate": 0.0005528514531888607,
      "loss": 2.2296,
      "step": 41676
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.7157332301139832,
      "learning_rate": 0.0005528492517594529,
      "loss": 2.4575,
      "step": 41677
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6954784989356995,
      "learning_rate": 0.0005528470502830356,
      "loss": 2.2749,
      "step": 41678
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.7194119691848755,
      "learning_rate": 0.0005528448487596093,
      "loss": 2.3406,
      "step": 41679
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.7532763481140137,
      "learning_rate": 0.0005528426471891743,
      "loss": 2.2378,
      "step": 41680
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6695229411125183,
      "learning_rate": 0.0005528404455717312,
      "loss": 2.4557,
      "step": 41681
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6187922954559326,
      "learning_rate": 0.0005528382439072802,
      "loss": 2.4346,
      "step": 41682
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.7619982361793518,
      "learning_rate": 0.0005528360421958217,
      "loss": 2.5225,
      "step": 41683
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.7538423538208008,
      "learning_rate": 0.0005528338404373564,
      "loss": 2.483,
      "step": 41684
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6806983947753906,
      "learning_rate": 0.0005528316386318844,
      "loss": 2.1616,
      "step": 41685
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6464446783065796,
      "learning_rate": 0.0005528294367794063,
      "loss": 2.2529,
      "step": 41686
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.7276580333709717,
      "learning_rate": 0.0005528272348799224,
      "loss": 2.2528,
      "step": 41687
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6904659867286682,
      "learning_rate": 0.0005528250329334331,
      "loss": 2.2475,
      "step": 41688
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.7410268187522888,
      "learning_rate": 0.0005528228309399388,
      "loss": 2.1404,
      "step": 41689
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.686558723449707,
      "learning_rate": 0.0005528206288994401,
      "loss": 2.2659,
      "step": 41690
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.757855236530304,
      "learning_rate": 0.0005528184268119372,
      "loss": 2.2433,
      "step": 41691
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6996372938156128,
      "learning_rate": 0.0005528162246774307,
      "loss": 2.2873,
      "step": 41692
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.653536319732666,
      "learning_rate": 0.0005528140224959208,
      "loss": 2.2932,
      "step": 41693
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.709854781627655,
      "learning_rate": 0.000552811820267408,
      "loss": 2.5384,
      "step": 41694
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.8282160758972168,
      "learning_rate": 0.0005528096179918927,
      "loss": 2.2829,
      "step": 41695
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6894688606262207,
      "learning_rate": 0.0005528074156693753,
      "loss": 2.1955,
      "step": 41696
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.7515633702278137,
      "learning_rate": 0.0005528052132998563,
      "loss": 2.4514,
      "step": 41697
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.8121048808097839,
      "learning_rate": 0.0005528030108833359,
      "loss": 2.2351,
      "step": 41698
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.7826786637306213,
      "learning_rate": 0.0005528008084198148,
      "loss": 2.3096,
      "step": 41699
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.7206777334213257,
      "learning_rate": 0.0005527986059092932,
      "loss": 2.4337,
      "step": 41700
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6932011842727661,
      "learning_rate": 0.0005527964033517715,
      "loss": 2.4041,
      "step": 41701
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.921981930732727,
      "learning_rate": 0.0005527942007472503,
      "loss": 2.1341,
      "step": 41702
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6596224308013916,
      "learning_rate": 0.0005527919980957298,
      "loss": 2.345,
      "step": 41703
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.705189049243927,
      "learning_rate": 0.0005527897953972106,
      "loss": 2.0402,
      "step": 41704
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.9183304309844971,
      "learning_rate": 0.0005527875926516929,
      "loss": 2.2839,
      "step": 41705
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6449701189994812,
      "learning_rate": 0.0005527853898591772,
      "loss": 2.1193,
      "step": 41706
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.7525328993797302,
      "learning_rate": 0.000552783187019664,
      "loss": 2.2105,
      "step": 41707
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.7210420370101929,
      "learning_rate": 0.0005527809841331536,
      "loss": 2.2372,
      "step": 41708
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.7055017948150635,
      "learning_rate": 0.0005527787811996465,
      "loss": 2.1598,
      "step": 41709
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.7183243036270142,
      "learning_rate": 0.0005527765782191431,
      "loss": 2.258,
      "step": 41710
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.7475981116294861,
      "learning_rate": 0.0005527743751916435,
      "loss": 2.1395,
      "step": 41711
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.7281330227851868,
      "learning_rate": 0.0005527721721171485,
      "loss": 2.4198,
      "step": 41712
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.666000247001648,
      "learning_rate": 0.0005527699689956585,
      "loss": 2.2207,
      "step": 41713
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.71181321144104,
      "learning_rate": 0.0005527677658271737,
      "loss": 2.2029,
      "step": 41714
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.7180813550949097,
      "learning_rate": 0.0005527655626116946,
      "loss": 2.2033,
      "step": 41715
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6717357635498047,
      "learning_rate": 0.0005527633593492217,
      "loss": 2.3678,
      "step": 41716
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6717910170555115,
      "learning_rate": 0.0005527611560397553,
      "loss": 2.3407,
      "step": 41717
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.7104036808013916,
      "learning_rate": 0.0005527589526832956,
      "loss": 2.5918,
      "step": 41718
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.7724294066429138,
      "learning_rate": 0.0005527567492798435,
      "loss": 2.2887,
      "step": 41719
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6940104961395264,
      "learning_rate": 0.0005527545458293991,
      "loss": 2.4968,
      "step": 41720
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.7183694839477539,
      "learning_rate": 0.0005527523423319628,
      "loss": 2.3603,
      "step": 41721
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.742778480052948,
      "learning_rate": 0.0005527501387875351,
      "loss": 2.261,
      "step": 41722
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6509994864463806,
      "learning_rate": 0.0005527479351961163,
      "loss": 2.2978,
      "step": 41723
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.699377715587616,
      "learning_rate": 0.000552745731557707,
      "loss": 2.2903,
      "step": 41724
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.7663684487342834,
      "learning_rate": 0.0005527435278723075,
      "loss": 2.4161,
      "step": 41725
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6464784741401672,
      "learning_rate": 0.0005527413241399181,
      "loss": 2.2439,
      "step": 41726
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.7152295112609863,
      "learning_rate": 0.0005527391203605394,
      "loss": 2.1888,
      "step": 41727
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6679145097732544,
      "learning_rate": 0.0005527369165341717,
      "loss": 2.3808,
      "step": 41728
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6572046875953674,
      "learning_rate": 0.0005527347126608154,
      "loss": 2.2542,
      "step": 41729
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6763384342193604,
      "learning_rate": 0.000552732508740471,
      "loss": 2.1969,
      "step": 41730
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.7104123830795288,
      "learning_rate": 0.0005527303047731389,
      "loss": 2.1525,
      "step": 41731
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6683359146118164,
      "learning_rate": 0.0005527281007588194,
      "loss": 2.3783,
      "step": 41732
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6805333495140076,
      "learning_rate": 0.000552725896697513,
      "loss": 2.3107,
      "step": 41733
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6720899939537048,
      "learning_rate": 0.00055272369258922,
      "loss": 2.3419,
      "step": 41734
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6979380249977112,
      "learning_rate": 0.000552721488433941,
      "loss": 2.5589,
      "step": 41735
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6564452648162842,
      "learning_rate": 0.0005527192842316763,
      "loss": 2.209,
      "step": 41736
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.8952282667160034,
      "learning_rate": 0.0005527170799824263,
      "loss": 2.4597,
      "step": 41737
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.7191557288169861,
      "learning_rate": 0.0005527148756861914,
      "loss": 2.5523,
      "step": 41738
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.8348619341850281,
      "learning_rate": 0.000552712671342972,
      "loss": 2.2426,
      "step": 41739
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6924886107444763,
      "learning_rate": 0.0005527104669527686,
      "loss": 2.3121,
      "step": 41740
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6482589840888977,
      "learning_rate": 0.0005527082625155816,
      "loss": 2.2537,
      "step": 41741
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.7201988697052002,
      "learning_rate": 0.0005527060580314113,
      "loss": 2.2847,
      "step": 41742
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6982204914093018,
      "learning_rate": 0.0005527038535002581,
      "loss": 2.2934,
      "step": 41743
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.7358422875404358,
      "learning_rate": 0.0005527016489221226,
      "loss": 2.4164,
      "step": 41744
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.7165572047233582,
      "learning_rate": 0.000552699444297005,
      "loss": 2.4062,
      "step": 41745
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.7231374979019165,
      "learning_rate": 0.0005526972396249059,
      "loss": 2.1632,
      "step": 41746
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6507026553153992,
      "learning_rate": 0.0005526950349058255,
      "loss": 2.2263,
      "step": 41747
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6887990236282349,
      "learning_rate": 0.0005526928301397645,
      "loss": 2.4253,
      "step": 41748
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.8001307845115662,
      "learning_rate": 0.000552690625326723,
      "loss": 2.157,
      "step": 41749
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.7592028975486755,
      "learning_rate": 0.0005526884204667016,
      "loss": 2.0807,
      "step": 41750
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.8037431836128235,
      "learning_rate": 0.0005526862155597006,
      "loss": 2.1482,
      "step": 41751
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.625415563583374,
      "learning_rate": 0.0005526840106057204,
      "loss": 1.9156,
      "step": 41752
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6589580178260803,
      "learning_rate": 0.0005526818056047616,
      "loss": 2.2994,
      "step": 41753
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.786228597164154,
      "learning_rate": 0.0005526796005568245,
      "loss": 2.3483,
      "step": 41754
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.7205541729927063,
      "learning_rate": 0.0005526773954619093,
      "loss": 2.2265,
      "step": 41755
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.7089498043060303,
      "learning_rate": 0.0005526751903200168,
      "loss": 2.4539,
      "step": 41756
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6782941222190857,
      "learning_rate": 0.0005526729851311472,
      "loss": 2.2031,
      "step": 41757
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6536290049552917,
      "learning_rate": 0.0005526707798953008,
      "loss": 1.9802,
      "step": 41758
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6848011612892151,
      "learning_rate": 0.0005526685746124782,
      "loss": 2.3201,
      "step": 41759
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6617887020111084,
      "learning_rate": 0.0005526663692826798,
      "loss": 2.1563,
      "step": 41760
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6503896117210388,
      "learning_rate": 0.0005526641639059059,
      "loss": 2.3826,
      "step": 41761
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.7887890934944153,
      "learning_rate": 0.000552661958482157,
      "loss": 2.298,
      "step": 41762
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.7452752590179443,
      "learning_rate": 0.0005526597530114334,
      "loss": 2.3117,
      "step": 41763
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.7020488381385803,
      "learning_rate": 0.0005526575474937356,
      "loss": 2.2569,
      "step": 41764
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.846689760684967,
      "learning_rate": 0.000552655341929064,
      "loss": 2.3075,
      "step": 41765
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.7580943703651428,
      "learning_rate": 0.000552653136317419,
      "loss": 2.1613,
      "step": 41766
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6731683611869812,
      "learning_rate": 0.000552650930658801,
      "loss": 2.1023,
      "step": 41767
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.8008383512496948,
      "learning_rate": 0.0005526487249532105,
      "loss": 2.2458,
      "step": 41768
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.7145570516586304,
      "learning_rate": 0.0005526465192006477,
      "loss": 2.142,
      "step": 41769
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.7739966511726379,
      "learning_rate": 0.0005526443134011134,
      "loss": 2.0956,
      "step": 41770
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6706695556640625,
      "learning_rate": 0.0005526421075546075,
      "loss": 2.4033,
      "step": 41771
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.8628923892974854,
      "learning_rate": 0.0005526399016611307,
      "loss": 2.2823,
      "step": 41772
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.652706503868103,
      "learning_rate": 0.0005526376957206835,
      "loss": 2.3746,
      "step": 41773
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.7090942859649658,
      "learning_rate": 0.000552635489733266,
      "loss": 2.2106,
      "step": 41774
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.733460545539856,
      "learning_rate": 0.0005526332836988789,
      "loss": 2.0604,
      "step": 41775
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6328767538070679,
      "learning_rate": 0.0005526310776175225,
      "loss": 2.3464,
      "step": 41776
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6545847058296204,
      "learning_rate": 0.0005526288714891972,
      "loss": 2.3076,
      "step": 41777
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.7225081920623779,
      "learning_rate": 0.0005526266653139034,
      "loss": 2.5613,
      "step": 41778
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6855237483978271,
      "learning_rate": 0.0005526244590916416,
      "loss": 2.2711,
      "step": 41779
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.7509642839431763,
      "learning_rate": 0.0005526222528224121,
      "loss": 2.3944,
      "step": 41780
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6473350524902344,
      "learning_rate": 0.0005526200465062154,
      "loss": 2.1906,
      "step": 41781
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6677408814430237,
      "learning_rate": 0.0005526178401430517,
      "loss": 2.3638,
      "step": 41782
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6021743416786194,
      "learning_rate": 0.0005526156337329218,
      "loss": 2.498,
      "step": 41783
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6802158951759338,
      "learning_rate": 0.0005526134272758258,
      "loss": 2.381,
      "step": 41784
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.7973182797431946,
      "learning_rate": 0.0005526112207717642,
      "loss": 2.2885,
      "step": 41785
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6812132596969604,
      "learning_rate": 0.0005526090142207373,
      "loss": 2.0495,
      "step": 41786
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6461286544799805,
      "learning_rate": 0.0005526068076227458,
      "loss": 2.1122,
      "step": 41787
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.7048869132995605,
      "learning_rate": 0.0005526046009777898,
      "loss": 2.149,
      "step": 41788
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.7321058511734009,
      "learning_rate": 0.0005526023942858699,
      "loss": 2.1813,
      "step": 41789
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6849994659423828,
      "learning_rate": 0.0005526001875469864,
      "loss": 2.2483,
      "step": 41790
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.703906774520874,
      "learning_rate": 0.0005525979807611398,
      "loss": 2.2762,
      "step": 41791
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.8207423686981201,
      "learning_rate": 0.0005525957739283304,
      "loss": 2.2279,
      "step": 41792
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.7153573036193848,
      "learning_rate": 0.0005525935670485587,
      "loss": 2.2679,
      "step": 41793
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.7065172791481018,
      "learning_rate": 0.0005525913601218251,
      "loss": 2.4547,
      "step": 41794
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6903895139694214,
      "learning_rate": 0.0005525891531481301,
      "loss": 2.1992,
      "step": 41795
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6776970028877258,
      "learning_rate": 0.0005525869461274739,
      "loss": 2.2213,
      "step": 41796
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.7563678026199341,
      "learning_rate": 0.0005525847390598569,
      "loss": 2.0742,
      "step": 41797
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.7174379825592041,
      "learning_rate": 0.0005525825319452798,
      "loss": 2.2876,
      "step": 41798
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.7009995579719543,
      "learning_rate": 0.0005525803247837428,
      "loss": 2.1713,
      "step": 41799
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.7068844437599182,
      "learning_rate": 0.0005525781175752464,
      "loss": 2.0936,
      "step": 41800
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.644235372543335,
      "learning_rate": 0.0005525759103197909,
      "loss": 2.1241,
      "step": 41801
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.7508050203323364,
      "learning_rate": 0.0005525737030173767,
      "loss": 2.2302,
      "step": 41802
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.7716891765594482,
      "learning_rate": 0.0005525714956680044,
      "loss": 2.3106,
      "step": 41803
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.7359300851821899,
      "learning_rate": 0.0005525692882716743,
      "loss": 2.249,
      "step": 41804
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.7410554885864258,
      "learning_rate": 0.0005525670808283866,
      "loss": 2.1519,
      "step": 41805
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.7285242080688477,
      "learning_rate": 0.0005525648733381422,
      "loss": 2.099,
      "step": 41806
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.7203130722045898,
      "learning_rate": 0.000552562665800941,
      "loss": 2.6457,
      "step": 41807
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.8284500241279602,
      "learning_rate": 0.0005525604582167839,
      "loss": 2.2174,
      "step": 41808
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.7025387287139893,
      "learning_rate": 0.0005525582505856707,
      "loss": 2.2447,
      "step": 41809
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.7171606421470642,
      "learning_rate": 0.0005525560429076023,
      "loss": 2.6117,
      "step": 41810
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6741509437561035,
      "learning_rate": 0.000552553835182579,
      "loss": 2.0557,
      "step": 41811
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.72605961561203,
      "learning_rate": 0.0005525516274106013,
      "loss": 2.0464,
      "step": 41812
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.653903603553772,
      "learning_rate": 0.0005525494195916693,
      "loss": 2.2374,
      "step": 41813
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.7283159494400024,
      "learning_rate": 0.0005525472117257836,
      "loss": 2.331,
      "step": 41814
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.7028117179870605,
      "learning_rate": 0.0005525450038129446,
      "loss": 2.5079,
      "step": 41815
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.7656901478767395,
      "learning_rate": 0.0005525427958531529,
      "loss": 2.3674,
      "step": 41816
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.7345918416976929,
      "learning_rate": 0.0005525405878464085,
      "loss": 2.5008,
      "step": 41817
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.8835929036140442,
      "learning_rate": 0.0005525383797927121,
      "loss": 2.2928,
      "step": 41818
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.8043086528778076,
      "learning_rate": 0.0005525361716920642,
      "loss": 2.3366,
      "step": 41819
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.7212681174278259,
      "learning_rate": 0.0005525339635444648,
      "loss": 2.59,
      "step": 41820
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.7095746397972107,
      "learning_rate": 0.0005525317553499148,
      "loss": 2.2336,
      "step": 41821
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6612937450408936,
      "learning_rate": 0.0005525295471084142,
      "loss": 2.2968,
      "step": 41822
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6396859884262085,
      "learning_rate": 0.0005525273388199638,
      "loss": 2.1643,
      "step": 41823
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6763834953308105,
      "learning_rate": 0.0005525251304845636,
      "loss": 2.108,
      "step": 41824
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.8809905052185059,
      "learning_rate": 0.0005525229221022142,
      "loss": 2.2425,
      "step": 41825
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6814358234405518,
      "learning_rate": 0.0005525207136729161,
      "loss": 2.2101,
      "step": 41826
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.7071101665496826,
      "learning_rate": 0.0005525185051966697,
      "loss": 2.1609,
      "step": 41827
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6639364957809448,
      "learning_rate": 0.0005525162966734752,
      "loss": 2.2076,
      "step": 41828
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6784404516220093,
      "learning_rate": 0.0005525140881033333,
      "loss": 2.2978,
      "step": 41829
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.7367494106292725,
      "learning_rate": 0.0005525118794862441,
      "loss": 2.3765,
      "step": 41830
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.7962791323661804,
      "learning_rate": 0.0005525096708222084,
      "loss": 2.4274,
      "step": 41831
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6769587993621826,
      "learning_rate": 0.0005525074621112263,
      "loss": 2.0652,
      "step": 41832
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6602870225906372,
      "learning_rate": 0.0005525052533532982,
      "loss": 2.3114,
      "step": 41833
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.7119183540344238,
      "learning_rate": 0.0005525030445484247,
      "loss": 2.3596,
      "step": 41834
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.7954736351966858,
      "learning_rate": 0.000552500835696606,
      "loss": 2.2026,
      "step": 41835
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.7206346988677979,
      "learning_rate": 0.0005524986267978427,
      "loss": 2.1341,
      "step": 41836
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6608253717422485,
      "learning_rate": 0.0005524964178521351,
      "loss": 2.24,
      "step": 41837
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.8751728534698486,
      "learning_rate": 0.0005524942088594838,
      "loss": 2.4412,
      "step": 41838
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.8441786766052246,
      "learning_rate": 0.0005524919998198889,
      "loss": 2.3087,
      "step": 41839
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.7242068648338318,
      "learning_rate": 0.0005524897907333511,
      "loss": 2.4553,
      "step": 41840
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.881788432598114,
      "learning_rate": 0.0005524875815998705,
      "loss": 2.1297,
      "step": 41841
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.7600477337837219,
      "learning_rate": 0.000552485372419448,
      "loss": 2.0881,
      "step": 41842
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6444898247718811,
      "learning_rate": 0.0005524831631920834,
      "loss": 2.2079,
      "step": 41843
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.7108889818191528,
      "learning_rate": 0.0005524809539177776,
      "loss": 2.1994,
      "step": 41844
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.7476075291633606,
      "learning_rate": 0.0005524787445965307,
      "loss": 2.3801,
      "step": 41845
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6979448795318604,
      "learning_rate": 0.0005524765352283432,
      "loss": 2.4158,
      "step": 41846
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6495497822761536,
      "learning_rate": 0.0005524743258132157,
      "loss": 2.3788,
      "step": 41847
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.742956817150116,
      "learning_rate": 0.0005524721163511484,
      "loss": 2.613,
      "step": 41848
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6686856150627136,
      "learning_rate": 0.0005524699068421417,
      "loss": 2.2455,
      "step": 41849
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.7442359328269958,
      "learning_rate": 0.0005524676972861962,
      "loss": 2.5032,
      "step": 41850
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.7703474760055542,
      "learning_rate": 0.0005524654876833121,
      "loss": 2.28,
      "step": 41851
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.7254239916801453,
      "learning_rate": 0.0005524632780334899,
      "loss": 2.3114,
      "step": 41852
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.647801399230957,
      "learning_rate": 0.00055246106833673,
      "loss": 2.2389,
      "step": 41853
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.7071083784103394,
      "learning_rate": 0.000552458858593033,
      "loss": 2.2431,
      "step": 41854
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.8031278252601624,
      "learning_rate": 0.0005524566488023988,
      "loss": 2.1559,
      "step": 41855
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.7109824419021606,
      "learning_rate": 0.0005524544389648284,
      "loss": 2.2853,
      "step": 41856
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.7996611595153809,
      "learning_rate": 0.0005524522290803218,
      "loss": 2.4213,
      "step": 41857
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.689177393913269,
      "learning_rate": 0.0005524500191488798,
      "loss": 2.0793,
      "step": 41858
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.6904777884483337,
      "learning_rate": 0.0005524478091705024,
      "loss": 2.1029,
      "step": 41859
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.7055807113647461,
      "learning_rate": 0.0005524455991451902,
      "loss": 2.1545,
      "step": 41860
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.7434815764427185,
      "learning_rate": 0.0005524433890729436,
      "loss": 2.4587,
      "step": 41861
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.8291016817092896,
      "learning_rate": 0.000552441178953763,
      "loss": 2.1721,
      "step": 41862
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.7351923584938049,
      "learning_rate": 0.0005524389687876488,
      "loss": 2.1634,
      "step": 41863
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.0438644886016846,
      "learning_rate": 0.0005524367585746015,
      "loss": 2.3669,
      "step": 41864
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.6797118186950684,
      "learning_rate": 0.0005524345483146214,
      "loss": 2.2667,
      "step": 41865
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.6561341285705566,
      "learning_rate": 0.000552432338007709,
      "loss": 2.3989,
      "step": 41866
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.7348978519439697,
      "learning_rate": 0.0005524301276538646,
      "loss": 2.2564,
      "step": 41867
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.7133581042289734,
      "learning_rate": 0.0005524279172530888,
      "loss": 2.2476,
      "step": 41868
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.6648618578910828,
      "learning_rate": 0.0005524257068053817,
      "loss": 2.2744,
      "step": 41869
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.7089489102363586,
      "learning_rate": 0.0005524234963107439,
      "loss": 2.1112,
      "step": 41870
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.7053005695343018,
      "learning_rate": 0.000552421285769176,
      "loss": 2.1862,
      "step": 41871
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.64069664478302,
      "learning_rate": 0.000552419075180678,
      "loss": 1.9787,
      "step": 41872
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.6779802441596985,
      "learning_rate": 0.0005524168645452507,
      "loss": 2.3499,
      "step": 41873
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.6658482551574707,
      "learning_rate": 0.0005524146538628944,
      "loss": 2.18,
      "step": 41874
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.6686668395996094,
      "learning_rate": 0.0005524124431336092,
      "loss": 2.0841,
      "step": 41875
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.6910772919654846,
      "learning_rate": 0.0005524102323573959,
      "loss": 2.2584,
      "step": 41876
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.7001650333404541,
      "learning_rate": 0.0005524080215342547,
      "loss": 2.1957,
      "step": 41877
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.7384697198867798,
      "learning_rate": 0.0005524058106641862,
      "loss": 2.2563,
      "step": 41878
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.7308487296104431,
      "learning_rate": 0.0005524035997471906,
      "loss": 2.5248,
      "step": 41879
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.7034355401992798,
      "learning_rate": 0.0005524013887832685,
      "loss": 2.3421,
      "step": 41880
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.6798911094665527,
      "learning_rate": 0.0005523991777724201,
      "loss": 2.3666,
      "step": 41881
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.6990649700164795,
      "learning_rate": 0.0005523969667146461,
      "loss": 2.1109,
      "step": 41882
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.6793997883796692,
      "learning_rate": 0.0005523947556099466,
      "loss": 2.3927,
      "step": 41883
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.6814647912979126,
      "learning_rate": 0.0005523925444583221,
      "loss": 2.1273,
      "step": 41884
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.7939126491546631,
      "learning_rate": 0.0005523903332597732,
      "loss": 2.2335,
      "step": 41885
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.7667938470840454,
      "learning_rate": 0.0005523881220143002,
      "loss": 2.0803,
      "step": 41886
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.6785852313041687,
      "learning_rate": 0.0005523859107219033,
      "loss": 2.331,
      "step": 41887
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.6574698090553284,
      "learning_rate": 0.0005523836993825832,
      "loss": 2.1354,
      "step": 41888
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.6685755848884583,
      "learning_rate": 0.0005523814879963403,
      "loss": 2.2372,
      "step": 41889
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.6484618782997131,
      "learning_rate": 0.0005523792765631748,
      "loss": 2.0312,
      "step": 41890
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.6367879509925842,
      "learning_rate": 0.0005523770650830872,
      "loss": 2.0926,
      "step": 41891
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.7009397745132446,
      "learning_rate": 0.000552374853556078,
      "loss": 2.3259,
      "step": 41892
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.6934835314750671,
      "learning_rate": 0.0005523726419821475,
      "loss": 2.2633,
      "step": 41893
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.665072500705719,
      "learning_rate": 0.0005523704303612962,
      "loss": 2.1692,
      "step": 41894
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.6543398499488831,
      "learning_rate": 0.0005523682186935246,
      "loss": 2.1435,
      "step": 41895
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.6441652774810791,
      "learning_rate": 0.0005523660069788329,
      "loss": 2.2571,
      "step": 41896
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.7079554200172424,
      "learning_rate": 0.0005523637952172214,
      "loss": 2.5356,
      "step": 41897
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.7250428795814514,
      "learning_rate": 0.000552361583408691,
      "loss": 2.2001,
      "step": 41898
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.7028688788414001,
      "learning_rate": 0.0005523593715532416,
      "loss": 2.4214,
      "step": 41899
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.6917188763618469,
      "learning_rate": 0.000552357159650874,
      "loss": 2.3202,
      "step": 41900
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.680427610874176,
      "learning_rate": 0.0005523549477015882,
      "loss": 2.4655,
      "step": 41901
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.6976291537284851,
      "learning_rate": 0.0005523527357053851,
      "loss": 2.2743,
      "step": 41902
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.6946786642074585,
      "learning_rate": 0.0005523505236622648,
      "loss": 2.2847,
      "step": 41903
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.6859903931617737,
      "learning_rate": 0.0005523483115722277,
      "loss": 2.3476,
      "step": 41904
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.655180037021637,
      "learning_rate": 0.0005523460994352744,
      "loss": 2.1487,
      "step": 41905
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.6797287464141846,
      "learning_rate": 0.000552343887251405,
      "loss": 2.1632,
      "step": 41906
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.6143772602081299,
      "learning_rate": 0.0005523416750206202,
      "loss": 2.3184,
      "step": 41907
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.7277035713195801,
      "learning_rate": 0.0005523394627429204,
      "loss": 2.3865,
      "step": 41908
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.9687227606773376,
      "learning_rate": 0.0005523372504183059,
      "loss": 2.1539,
      "step": 41909
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.6919167637825012,
      "learning_rate": 0.0005523350380467771,
      "loss": 2.3762,
      "step": 41910
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.8203825354576111,
      "learning_rate": 0.0005523328256283345,
      "loss": 2.4096,
      "step": 41911
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.6986881494522095,
      "learning_rate": 0.0005523306131629785,
      "loss": 2.1386,
      "step": 41912
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.7327856421470642,
      "learning_rate": 0.0005523284006507094,
      "loss": 2.3672,
      "step": 41913
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.7206160426139832,
      "learning_rate": 0.0005523261880915276,
      "loss": 2.2946,
      "step": 41914
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.726757824420929,
      "learning_rate": 0.0005523239754854338,
      "loss": 2.3168,
      "step": 41915
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.697433590888977,
      "learning_rate": 0.000552321762832428,
      "loss": 2.212,
      "step": 41916
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.7386558055877686,
      "learning_rate": 0.0005523195501325111,
      "loss": 2.3096,
      "step": 41917
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.6766619682312012,
      "learning_rate": 0.000552317337385683,
      "loss": 2.1541,
      "step": 41918
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.6798357367515564,
      "learning_rate": 0.0005523151245919443,
      "loss": 2.3825,
      "step": 41919
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.7330812811851501,
      "learning_rate": 0.0005523129117512956,
      "loss": 2.4114,
      "step": 41920
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.6943531036376953,
      "learning_rate": 0.000552310698863737,
      "loss": 2.2038,
      "step": 41921
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.7518090605735779,
      "learning_rate": 0.0005523084859292693,
      "loss": 2.2053,
      "step": 41922
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.7590051889419556,
      "learning_rate": 0.0005523062729478926,
      "loss": 2.1231,
      "step": 41923
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.680081844329834,
      "learning_rate": 0.0005523040599196075,
      "loss": 2.0027,
      "step": 41924
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.7352327108383179,
      "learning_rate": 0.0005523018468444141,
      "loss": 2.4206,
      "step": 41925
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.6999675631523132,
      "learning_rate": 0.0005522996337223131,
      "loss": 2.4063,
      "step": 41926
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.6351516246795654,
      "learning_rate": 0.000552297420553305,
      "loss": 2.3763,
      "step": 41927
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.7537173628807068,
      "learning_rate": 0.0005522952073373899,
      "loss": 2.2506,
      "step": 41928
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.6670343279838562,
      "learning_rate": 0.0005522929940745683,
      "loss": 2.1908,
      "step": 41929
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.6348719000816345,
      "learning_rate": 0.0005522907807648408,
      "loss": 2.2565,
      "step": 41930
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.740465521812439,
      "learning_rate": 0.0005522885674082077,
      "loss": 2.4065,
      "step": 41931
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.7664778232574463,
      "learning_rate": 0.0005522863540046693,
      "loss": 2.1567,
      "step": 41932
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.7359693646430969,
      "learning_rate": 0.0005522841405542261,
      "loss": 2.414,
      "step": 41933
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.6726933121681213,
      "learning_rate": 0.0005522819270568786,
      "loss": 2.4449,
      "step": 41934
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.6705275774002075,
      "learning_rate": 0.0005522797135126269,
      "loss": 2.1558,
      "step": 41935
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.7784181237220764,
      "learning_rate": 0.000552277499921472,
      "loss": 2.2913,
      "step": 41936
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.7027032375335693,
      "learning_rate": 0.0005522752862834137,
      "loss": 2.3578,
      "step": 41937
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.7060868740081787,
      "learning_rate": 0.0005522730725984527,
      "loss": 2.3105,
      "step": 41938
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.6710401773452759,
      "learning_rate": 0.0005522708588665894,
      "loss": 2.1681,
      "step": 41939
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.6767838597297668,
      "learning_rate": 0.0005522686450878242,
      "loss": 2.4117,
      "step": 41940
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.6464818120002747,
      "learning_rate": 0.0005522664312621575,
      "loss": 2.0237,
      "step": 41941
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.6538450717926025,
      "learning_rate": 0.0005522642173895896,
      "loss": 2.3321,
      "step": 41942
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.7327671647071838,
      "learning_rate": 0.0005522620034701212,
      "loss": 2.3776,
      "step": 41943
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.6625581383705139,
      "learning_rate": 0.0005522597895037523,
      "loss": 2.2261,
      "step": 41944
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.7113929986953735,
      "learning_rate": 0.0005522575754904838,
      "loss": 2.3115,
      "step": 41945
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.7041043639183044,
      "learning_rate": 0.0005522553614303158,
      "loss": 2.3378,
      "step": 41946
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.7153629660606384,
      "learning_rate": 0.0005522531473232486,
      "loss": 2.2593,
      "step": 41947
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.6925229430198669,
      "learning_rate": 0.0005522509331692828,
      "loss": 2.4808,
      "step": 41948
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.6968785524368286,
      "learning_rate": 0.000552248718968419,
      "loss": 2.2277,
      "step": 41949
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.8155184388160706,
      "learning_rate": 0.0005522465047206572,
      "loss": 2.2593,
      "step": 41950
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.6723272204399109,
      "learning_rate": 0.000552244290425998,
      "loss": 2.2338,
      "step": 41951
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.6628913283348083,
      "learning_rate": 0.0005522420760844421,
      "loss": 2.2441,
      "step": 41952
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.01380455493927,
      "learning_rate": 0.0005522398616959894,
      "loss": 2.3885,
      "step": 41953
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.6565245389938354,
      "learning_rate": 0.0005522376472606406,
      "loss": 2.4397,
      "step": 41954
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.6910865306854248,
      "learning_rate": 0.000552235432778396,
      "loss": 2.3107,
      "step": 41955
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.6911131739616394,
      "learning_rate": 0.0005522332182492562,
      "loss": 2.1552,
      "step": 41956
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.6869274973869324,
      "learning_rate": 0.0005522310036732214,
      "loss": 2.2447,
      "step": 41957
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.6730567216873169,
      "learning_rate": 0.0005522287890502921,
      "loss": 2.3063,
      "step": 41958
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.670242965221405,
      "learning_rate": 0.0005522265743804687,
      "loss": 2.2493,
      "step": 41959
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.6999282836914062,
      "learning_rate": 0.0005522243596637516,
      "loss": 2.3084,
      "step": 41960
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.6516987681388855,
      "learning_rate": 0.0005522221449001414,
      "loss": 2.0788,
      "step": 41961
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.7948766350746155,
      "learning_rate": 0.0005522199300896381,
      "loss": 2.0824,
      "step": 41962
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.6498615145683289,
      "learning_rate": 0.0005522177152322425,
      "loss": 2.3551,
      "step": 41963
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.6474988460540771,
      "learning_rate": 0.0005522155003279547,
      "loss": 2.4784,
      "step": 41964
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.6678487062454224,
      "learning_rate": 0.0005522132853767756,
      "loss": 2.159,
      "step": 41965
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.8557220101356506,
      "learning_rate": 0.0005522110703787051,
      "loss": 2.1373,
      "step": 41966
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.7157498598098755,
      "learning_rate": 0.0005522088553337437,
      "loss": 2.1775,
      "step": 41967
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.7571476101875305,
      "learning_rate": 0.000552206640241892,
      "loss": 2.3555,
      "step": 41968
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.6907142996788025,
      "learning_rate": 0.0005522044251031503,
      "loss": 2.2041,
      "step": 41969
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.6204506754875183,
      "learning_rate": 0.0005522022099175191,
      "loss": 2.2134,
      "step": 41970
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.6652551293373108,
      "learning_rate": 0.0005521999946849987,
      "loss": 2.2087,
      "step": 41971
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.7061558961868286,
      "learning_rate": 0.0005521977794055896,
      "loss": 2.5993,
      "step": 41972
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.8010873794555664,
      "learning_rate": 0.0005521955640792922,
      "loss": 2.3445,
      "step": 41973
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.6916240453720093,
      "learning_rate": 0.0005521933487061067,
      "loss": 2.2552,
      "step": 41974
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.7860300540924072,
      "learning_rate": 0.000552191133286034,
      "loss": 2.2404,
      "step": 41975
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.669434130191803,
      "learning_rate": 0.000552188917819074,
      "loss": 2.419,
      "step": 41976
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.6680963039398193,
      "learning_rate": 0.0005521867023052273,
      "loss": 2.3265,
      "step": 41977
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.741066038608551,
      "learning_rate": 0.0005521844867444944,
      "loss": 2.4039,
      "step": 41978
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.7261665463447571,
      "learning_rate": 0.0005521822711368757,
      "loss": 2.1899,
      "step": 41979
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.6604165434837341,
      "learning_rate": 0.0005521800554823714,
      "loss": 2.3431,
      "step": 41980
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.7036683559417725,
      "learning_rate": 0.0005521778397809821,
      "loss": 2.3256,
      "step": 41981
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.6924464702606201,
      "learning_rate": 0.0005521756240327083,
      "loss": 2.365,
      "step": 41982
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.6954430937767029,
      "learning_rate": 0.0005521734082375503,
      "loss": 2.4601,
      "step": 41983
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.7510690689086914,
      "learning_rate": 0.0005521711923955083,
      "loss": 2.3424,
      "step": 41984
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.6863144040107727,
      "learning_rate": 0.0005521689765065832,
      "loss": 2.3695,
      "step": 41985
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.6790976524353027,
      "learning_rate": 0.0005521667605707749,
      "loss": 2.0414,
      "step": 41986
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.7432891130447388,
      "learning_rate": 0.0005521645445880841,
      "loss": 2.1953,
      "step": 41987
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.6876479387283325,
      "learning_rate": 0.000552162328558511,
      "loss": 2.1749,
      "step": 41988
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.6392235159873962,
      "learning_rate": 0.0005521601124820565,
      "loss": 2.253,
      "step": 41989
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.6627472639083862,
      "learning_rate": 0.0005521578963587205,
      "loss": 2.1213,
      "step": 41990
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.7374812960624695,
      "learning_rate": 0.0005521556801885035,
      "loss": 2.4411,
      "step": 41991
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.7105194330215454,
      "learning_rate": 0.0005521534639714061,
      "loss": 2.1885,
      "step": 41992
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.657818078994751,
      "learning_rate": 0.0005521512477074286,
      "loss": 2.3328,
      "step": 41993
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.7333478331565857,
      "learning_rate": 0.0005521490313965715,
      "loss": 2.2644,
      "step": 41994
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.739484965801239,
      "learning_rate": 0.000552146815038835,
      "loss": 2.4503,
      "step": 41995
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.6521703600883484,
      "learning_rate": 0.0005521445986342197,
      "loss": 2.176,
      "step": 41996
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.6404620409011841,
      "learning_rate": 0.0005521423821827259,
      "loss": 2.3856,
      "step": 41997
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.7723190784454346,
      "learning_rate": 0.0005521401656843543,
      "loss": 2.2275,
      "step": 41998
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.6871780157089233,
      "learning_rate": 0.0005521379491391048,
      "loss": 2.1983,
      "step": 41999
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.734989583492279,
      "learning_rate": 0.0005521357325469783,
      "loss": 2.3838,
      "step": 42000
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.7025517821311951,
      "learning_rate": 0.0005521335159079749,
      "loss": 2.4809,
      "step": 42001
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.8313845992088318,
      "learning_rate": 0.0005521312992220951,
      "loss": 2.2792,
      "step": 42002
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.7041850686073303,
      "learning_rate": 0.0005521290824893394,
      "loss": 2.3172,
      "step": 42003
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.6937659978866577,
      "learning_rate": 0.0005521268657097081,
      "loss": 2.4713,
      "step": 42004
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.7679965496063232,
      "learning_rate": 0.0005521246488832017,
      "loss": 2.1471,
      "step": 42005
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.6630342602729797,
      "learning_rate": 0.0005521224320098205,
      "loss": 2.3493,
      "step": 42006
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.6654844880104065,
      "learning_rate": 0.0005521202150895651,
      "loss": 2.4037,
      "step": 42007
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.7367474436759949,
      "learning_rate": 0.0005521179981224357,
      "loss": 2.2288,
      "step": 42008
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.6940218806266785,
      "learning_rate": 0.0005521157811084328,
      "loss": 2.3649,
      "step": 42009
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.8346299529075623,
      "learning_rate": 0.0005521135640475568,
      "loss": 2.2929,
      "step": 42010
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.732915461063385,
      "learning_rate": 0.0005521113469398081,
      "loss": 2.2602,
      "step": 42011
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.7254041433334351,
      "learning_rate": 0.0005521091297851873,
      "loss": 2.2758,
      "step": 42012
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.7030171155929565,
      "learning_rate": 0.0005521069125836945,
      "loss": 2.0394,
      "step": 42013
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.7872864603996277,
      "learning_rate": 0.0005521046953353304,
      "loss": 2.4269,
      "step": 42014
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.6398378014564514,
      "learning_rate": 0.0005521024780400952,
      "loss": 2.1168,
      "step": 42015
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.783742368221283,
      "learning_rate": 0.0005521002606979894,
      "loss": 2.2237,
      "step": 42016
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.8110136389732361,
      "learning_rate": 0.0005520980433090135,
      "loss": 2.3052,
      "step": 42017
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.6568477153778076,
      "learning_rate": 0.0005520958258731677,
      "loss": 2.2754,
      "step": 42018
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.0656534433364868,
      "learning_rate": 0.0005520936083904525,
      "loss": 2.1913,
      "step": 42019
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.667057454586029,
      "learning_rate": 0.0005520913908608684,
      "loss": 2.2292,
      "step": 42020
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.6326492428779602,
      "learning_rate": 0.0005520891732844158,
      "loss": 2.3227,
      "step": 42021
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.8246812224388123,
      "learning_rate": 0.0005520869556610949,
      "loss": 2.1748,
      "step": 42022
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.711670994758606,
      "learning_rate": 0.0005520847379909063,
      "loss": 2.1707,
      "step": 42023
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.647340714931488,
      "learning_rate": 0.0005520825202738505,
      "loss": 2.1856,
      "step": 42024
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.670668363571167,
      "learning_rate": 0.0005520803025099278,
      "loss": 2.0733,
      "step": 42025
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.730286180973053,
      "learning_rate": 0.0005520780846991386,
      "loss": 2.417,
      "step": 42026
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.6672574281692505,
      "learning_rate": 0.0005520758668414832,
      "loss": 2.3823,
      "step": 42027
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.7752571105957031,
      "learning_rate": 0.0005520736489369624,
      "loss": 2.379,
      "step": 42028
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.6826125979423523,
      "learning_rate": 0.0005520714309855761,
      "loss": 2.3178,
      "step": 42029
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.6417522430419922,
      "learning_rate": 0.000552069212987325,
      "loss": 2.2991,
      "step": 42030
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.7808308601379395,
      "learning_rate": 0.0005520669949422096,
      "loss": 2.2492,
      "step": 42031
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.7035848498344421,
      "learning_rate": 0.0005520647768502301,
      "loss": 2.0767,
      "step": 42032
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.8019646406173706,
      "learning_rate": 0.000552062558711387,
      "loss": 2.2242,
      "step": 42033
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.7095528244972229,
      "learning_rate": 0.0005520603405256808,
      "loss": 2.2281,
      "step": 42034
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.6640965938568115,
      "learning_rate": 0.0005520581222931117,
      "loss": 2.3252,
      "step": 42035
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.7018221020698547,
      "learning_rate": 0.0005520559040136803,
      "loss": 2.3165,
      "step": 42036
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.9397322535514832,
      "learning_rate": 0.000552053685687387,
      "loss": 2.3723,
      "step": 42037
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.6971773505210876,
      "learning_rate": 0.0005520514673142321,
      "loss": 2.353,
      "step": 42038
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.6905346512794495,
      "learning_rate": 0.0005520492488942161,
      "loss": 2.0494,
      "step": 42039
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.7085086107254028,
      "learning_rate": 0.0005520470304273394,
      "loss": 2.1943,
      "step": 42040
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.7079015374183655,
      "learning_rate": 0.0005520448119136022,
      "loss": 2.3869,
      "step": 42041
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.7329228520393372,
      "learning_rate": 0.0005520425933530054,
      "loss": 2.3693,
      "step": 42042
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.6423704624176025,
      "learning_rate": 0.000552040374745549,
      "loss": 2.1848,
      "step": 42043
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.763759195804596,
      "learning_rate": 0.0005520381560912336,
      "loss": 2.5759,
      "step": 42044
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.7531652450561523,
      "learning_rate": 0.0005520359373900594,
      "loss": 2.3816,
      "step": 42045
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.7380744814872742,
      "learning_rate": 0.0005520337186420271,
      "loss": 2.4098,
      "step": 42046
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.8551572561264038,
      "learning_rate": 0.0005520314998471369,
      "loss": 2.1935,
      "step": 42047
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.7024012207984924,
      "learning_rate": 0.0005520292810053893,
      "loss": 2.3561,
      "step": 42048
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.7196593284606934,
      "learning_rate": 0.0005520270621167848,
      "loss": 2.1523,
      "step": 42049
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.8293755054473877,
      "learning_rate": 0.0005520248431813235,
      "loss": 2.4867,
      "step": 42050
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.779841423034668,
      "learning_rate": 0.0005520226241990062,
      "loss": 2.2271,
      "step": 42051
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.8614265322685242,
      "learning_rate": 0.0005520204051698331,
      "loss": 2.3739,
      "step": 42052
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.776587963104248,
      "learning_rate": 0.0005520181860938046,
      "loss": 2.364,
      "step": 42053
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.6584134697914124,
      "learning_rate": 0.0005520159669709212,
      "loss": 2.2368,
      "step": 42054
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.6946700811386108,
      "learning_rate": 0.0005520137478011832,
      "loss": 2.1433,
      "step": 42055
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.8238826394081116,
      "learning_rate": 0.0005520115285845912,
      "loss": 2.2688,
      "step": 42056
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.6410667896270752,
      "learning_rate": 0.0005520093093211454,
      "loss": 2.303,
      "step": 42057
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.6254749894142151,
      "learning_rate": 0.0005520070900108465,
      "loss": 2.0134,
      "step": 42058
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.6966226100921631,
      "learning_rate": 0.0005520048706536946,
      "loss": 2.3203,
      "step": 42059
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.6810694336891174,
      "learning_rate": 0.0005520026512496901,
      "loss": 2.2908,
      "step": 42060
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.739843487739563,
      "learning_rate": 0.0005520004317988336,
      "loss": 2.3486,
      "step": 42061
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.7709816098213196,
      "learning_rate": 0.0005519982123011256,
      "loss": 2.3923,
      "step": 42062
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.6705959439277649,
      "learning_rate": 0.0005519959927565664,
      "loss": 2.1004,
      "step": 42063
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.7074775099754333,
      "learning_rate": 0.0005519937731651563,
      "loss": 2.482,
      "step": 42064
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.6477487087249756,
      "learning_rate": 0.0005519915535268958,
      "loss": 2.2927,
      "step": 42065
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.6522454619407654,
      "learning_rate": 0.0005519893338417853,
      "loss": 2.2258,
      "step": 42066
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.7337061166763306,
      "learning_rate": 0.0005519871141098253,
      "loss": 2.5434,
      "step": 42067
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.7556411623954773,
      "learning_rate": 0.0005519848943310161,
      "loss": 2.1488,
      "step": 42068
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.583554744720459,
      "learning_rate": 0.000551982674505358,
      "loss": 2.1294,
      "step": 42069
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.7510461211204529,
      "learning_rate": 0.0005519804546328519,
      "loss": 2.5322,
      "step": 42070
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.699367105960846,
      "learning_rate": 0.0005519782347134976,
      "loss": 2.3647,
      "step": 42071
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.7602986097335815,
      "learning_rate": 0.0005519760147472958,
      "loss": 2.2248,
      "step": 42072
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.6979914307594299,
      "learning_rate": 0.000551973794734247,
      "loss": 2.0224,
      "step": 42073
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.6785837411880493,
      "learning_rate": 0.0005519715746743515,
      "loss": 2.5657,
      "step": 42074
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.6744847893714905,
      "learning_rate": 0.0005519693545676097,
      "loss": 2.2809,
      "step": 42075
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.654001772403717,
      "learning_rate": 0.0005519671344140221,
      "loss": 2.301,
      "step": 42076
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.7335836887359619,
      "learning_rate": 0.000551964914213589,
      "loss": 2.339,
      "step": 42077
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.6999269723892212,
      "learning_rate": 0.0005519626939663109,
      "loss": 2.2016,
      "step": 42078
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.7333914041519165,
      "learning_rate": 0.0005519604736721881,
      "loss": 2.0271,
      "step": 42079
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.621703028678894,
      "learning_rate": 0.0005519582533312212,
      "loss": 2.331,
      "step": 42080
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.7177910804748535,
      "learning_rate": 0.0005519560329434104,
      "loss": 2.509,
      "step": 42081
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.6957383751869202,
      "learning_rate": 0.0005519538125087562,
      "loss": 2.0362,
      "step": 42082
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.6813667416572571,
      "learning_rate": 0.0005519515920272592,
      "loss": 2.1877,
      "step": 42083
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.7688540816307068,
      "learning_rate": 0.0005519493714989196,
      "loss": 2.2789,
      "step": 42084
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.6855077743530273,
      "learning_rate": 0.0005519471509237377,
      "loss": 2.2412,
      "step": 42085
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.7013890743255615,
      "learning_rate": 0.0005519449303017141,
      "loss": 2.3922,
      "step": 42086
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.8795271515846252,
      "learning_rate": 0.0005519427096328494,
      "loss": 2.3376,
      "step": 42087
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.67962646484375,
      "learning_rate": 0.0005519404889171435,
      "loss": 2.3761,
      "step": 42088
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.778822660446167,
      "learning_rate": 0.0005519382681545973,
      "loss": 2.2187,
      "step": 42089
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.8096737265586853,
      "learning_rate": 0.000551936047345211,
      "loss": 2.4408,
      "step": 42090
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.6997770071029663,
      "learning_rate": 0.000551933826488985,
      "loss": 2.3384,
      "step": 42091
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.6425469517707825,
      "learning_rate": 0.0005519316055859197,
      "loss": 2.2449,
      "step": 42092
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.629459023475647,
      "learning_rate": 0.0005519293846360157,
      "loss": 2.2074,
      "step": 42093
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.846670389175415,
      "learning_rate": 0.0005519271636392732,
      "loss": 2.214,
      "step": 42094
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.7118527293205261,
      "learning_rate": 0.0005519249425956925,
      "loss": 2.1351,
      "step": 42095
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.6633149981498718,
      "learning_rate": 0.0005519227215052744,
      "loss": 2.2997,
      "step": 42096
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.6611599922180176,
      "learning_rate": 0.0005519205003680191,
      "loss": 2.3512,
      "step": 42097
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.6297431588172913,
      "learning_rate": 0.000551918279183927,
      "loss": 2.2616,
      "step": 42098
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.6621682047843933,
      "learning_rate": 0.0005519160579529986,
      "loss": 2.4168,
      "step": 42099
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.7044884562492371,
      "learning_rate": 0.0005519138366752341,
      "loss": 2.3113,
      "step": 42100
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.687494695186615,
      "learning_rate": 0.0005519116153506341,
      "loss": 2.4125,
      "step": 42101
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.8381965756416321,
      "learning_rate": 0.0005519093939791991,
      "loss": 2.342,
      "step": 42102
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.7172794342041016,
      "learning_rate": 0.0005519071725609292,
      "loss": 2.2837,
      "step": 42103
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.6563155055046082,
      "learning_rate": 0.0005519049510958252,
      "loss": 2.3594,
      "step": 42104
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.7047944664955139,
      "learning_rate": 0.0005519027295838873,
      "loss": 2.3407,
      "step": 42105
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.6822264790534973,
      "learning_rate": 0.0005519005080251158,
      "loss": 2.283,
      "step": 42106
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.7302462458610535,
      "learning_rate": 0.0005518982864195113,
      "loss": 2.1557,
      "step": 42107
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.7328212857246399,
      "learning_rate": 0.0005518960647670742,
      "loss": 2.1174,
      "step": 42108
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.6981557607650757,
      "learning_rate": 0.0005518938430678048,
      "loss": 2.0423,
      "step": 42109
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.6904452443122864,
      "learning_rate": 0.0005518916213217035,
      "loss": 2.5363,
      "step": 42110
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.7556701302528381,
      "learning_rate": 0.0005518893995287709,
      "loss": 2.4791,
      "step": 42111
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.7293422222137451,
      "learning_rate": 0.0005518871776890073,
      "loss": 2.2031,
      "step": 42112
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.7041318416595459,
      "learning_rate": 0.0005518849558024131,
      "loss": 2.1092,
      "step": 42113
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.6707445383071899,
      "learning_rate": 0.0005518827338689888,
      "loss": 2.3705,
      "step": 42114
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.7288690209388733,
      "learning_rate": 0.0005518805118887347,
      "loss": 2.0781,
      "step": 42115
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.8613142967224121,
      "learning_rate": 0.0005518782898616512,
      "loss": 2.1537,
      "step": 42116
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.9446648359298706,
      "learning_rate": 0.0005518760677877388,
      "loss": 2.4162,
      "step": 42117
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.7310355305671692,
      "learning_rate": 0.0005518738456669979,
      "loss": 2.4557,
      "step": 42118
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.7693674564361572,
      "learning_rate": 0.0005518716234994288,
      "loss": 2.2783,
      "step": 42119
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.9166837930679321,
      "learning_rate": 0.0005518694012850323,
      "loss": 2.191,
      "step": 42120
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.7100326418876648,
      "learning_rate": 0.0005518671790238083,
      "loss": 2.2997,
      "step": 42121
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.6950097680091858,
      "learning_rate": 0.0005518649567157574,
      "loss": 2.4217,
      "step": 42122
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.8605507016181946,
      "learning_rate": 0.0005518627343608801,
      "loss": 2.221,
      "step": 42123
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.7790413498878479,
      "learning_rate": 0.0005518605119591768,
      "loss": 2.147,
      "step": 42124
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.7216564416885376,
      "learning_rate": 0.0005518582895106479,
      "loss": 2.1251,
      "step": 42125
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.662444531917572,
      "learning_rate": 0.0005518560670152938,
      "loss": 2.3792,
      "step": 42126
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.9233466982841492,
      "learning_rate": 0.0005518538444731147,
      "loss": 2.1977,
      "step": 42127
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.7827080488204956,
      "learning_rate": 0.0005518516218841114,
      "loss": 2.2586,
      "step": 42128
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.71485835313797,
      "learning_rate": 0.0005518493992482842,
      "loss": 2.3994,
      "step": 42129
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.7728436589241028,
      "learning_rate": 0.0005518471765656333,
      "loss": 2.2566,
      "step": 42130
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.8033937811851501,
      "learning_rate": 0.0005518449538361594,
      "loss": 2.384,
      "step": 42131
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.6553564071655273,
      "learning_rate": 0.0005518427310598626,
      "loss": 2.0752,
      "step": 42132
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.7146350741386414,
      "learning_rate": 0.0005518405082367435,
      "loss": 2.307,
      "step": 42133
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.991023063659668,
      "learning_rate": 0.0005518382853668026,
      "loss": 2.1745,
      "step": 42134
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.7485882639884949,
      "learning_rate": 0.0005518360624500401,
      "loss": 2.1343,
      "step": 42135
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.7226417064666748,
      "learning_rate": 0.0005518338394864566,
      "loss": 2.4635,
      "step": 42136
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.8017111420631409,
      "learning_rate": 0.0005518316164760524,
      "loss": 2.1995,
      "step": 42137
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.7504467964172363,
      "learning_rate": 0.000551829393418828,
      "loss": 2.3567,
      "step": 42138
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.708499014377594,
      "learning_rate": 0.0005518271703147837,
      "loss": 2.2451,
      "step": 42139
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.8012408018112183,
      "learning_rate": 0.00055182494716392,
      "loss": 2.1647,
      "step": 42140
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.7006354331970215,
      "learning_rate": 0.0005518227239662373,
      "loss": 2.1423,
      "step": 42141
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.7138344645500183,
      "learning_rate": 0.000551820500721736,
      "loss": 2.332,
      "step": 42142
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.6959271430969238,
      "learning_rate": 0.0005518182774304165,
      "loss": 2.4067,
      "step": 42143
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.7714593410491943,
      "learning_rate": 0.0005518160540922793,
      "loss": 2.5168,
      "step": 42144
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.7249521613121033,
      "learning_rate": 0.0005518138307073245,
      "loss": 2.3087,
      "step": 42145
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.7141045928001404,
      "learning_rate": 0.000551811607275553,
      "loss": 2.1616,
      "step": 42146
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.7972896099090576,
      "learning_rate": 0.0005518093837969649,
      "loss": 2.157,
      "step": 42147
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.6918911337852478,
      "learning_rate": 0.0005518071602715606,
      "loss": 2.3393,
      "step": 42148
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.691878616809845,
      "learning_rate": 0.0005518049366993407,
      "loss": 2.3112,
      "step": 42149
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.755927562713623,
      "learning_rate": 0.0005518027130803055,
      "loss": 2.2968,
      "step": 42150
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.7239494323730469,
      "learning_rate": 0.0005518004894144554,
      "loss": 2.4319,
      "step": 42151
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.7399422526359558,
      "learning_rate": 0.0005517982657017909,
      "loss": 2.275,
      "step": 42152
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.7610582709312439,
      "learning_rate": 0.0005517960419423124,
      "loss": 2.2521,
      "step": 42153
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.770197868347168,
      "learning_rate": 0.00055179381813602,
      "loss": 2.4879,
      "step": 42154
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.8448896408081055,
      "learning_rate": 0.0005517915942829146,
      "loss": 2.021,
      "step": 42155
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.799596905708313,
      "learning_rate": 0.0005517893703829963,
      "loss": 2.4721,
      "step": 42156
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.9297758936882019,
      "learning_rate": 0.0005517871464362657,
      "loss": 2.2436,
      "step": 42157
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.7904684543609619,
      "learning_rate": 0.0005517849224427229,
      "loss": 2.121,
      "step": 42158
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.6944625377655029,
      "learning_rate": 0.0005517826984023687,
      "loss": 2.1931,
      "step": 42159
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.7622085213661194,
      "learning_rate": 0.0005517804743152033,
      "loss": 2.4747,
      "step": 42160
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.9154534935951233,
      "learning_rate": 0.0005517782501812272,
      "loss": 2.4433,
      "step": 42161
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.65113365650177,
      "learning_rate": 0.0005517760260004408,
      "loss": 2.1288,
      "step": 42162
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.6722516417503357,
      "learning_rate": 0.0005517738017728444,
      "loss": 2.0471,
      "step": 42163
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.7924888134002686,
      "learning_rate": 0.0005517715774984386,
      "loss": 2.2034,
      "step": 42164
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.7364499568939209,
      "learning_rate": 0.0005517693531772236,
      "loss": 2.412,
      "step": 42165
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.7416113615036011,
      "learning_rate": 0.0005517671288092,
      "loss": 2.3999,
      "step": 42166
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.650679886341095,
      "learning_rate": 0.000551764904394368,
      "loss": 1.9533,
      "step": 42167
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.6763060092926025,
      "learning_rate": 0.0005517626799327283,
      "loss": 2.4209,
      "step": 42168
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.6980661153793335,
      "learning_rate": 0.0005517604554242812,
      "loss": 2.4652,
      "step": 42169
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.6945353150367737,
      "learning_rate": 0.0005517582308690269,
      "loss": 2.2988,
      "step": 42170
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.8025232553482056,
      "learning_rate": 0.0005517560062669661,
      "loss": 2.2118,
      "step": 42171
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.66679447889328,
      "learning_rate": 0.0005517537816180991,
      "loss": 2.2258,
      "step": 42172
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.7192324995994568,
      "learning_rate": 0.0005517515569224263,
      "loss": 2.4371,
      "step": 42173
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.860919177532196,
      "learning_rate": 0.0005517493321799482,
      "loss": 2.24,
      "step": 42174
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.7641915082931519,
      "learning_rate": 0.000551747107390665,
      "loss": 2.2046,
      "step": 42175
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.6448470950126648,
      "learning_rate": 0.0005517448825545773,
      "loss": 2.2098,
      "step": 42176
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.7763416767120361,
      "learning_rate": 0.0005517426576716857,
      "loss": 2.3033,
      "step": 42177
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.5930473804473877,
      "learning_rate": 0.0005517404327419901,
      "loss": 2.1623,
      "step": 42178
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.7022992372512817,
      "learning_rate": 0.0005517382077654914,
      "loss": 2.4923,
      "step": 42179
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.6993308663368225,
      "learning_rate": 0.0005517359827421897,
      "loss": 2.3856,
      "step": 42180
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.6160598993301392,
      "learning_rate": 0.0005517337576720856,
      "loss": 2.3741,
      "step": 42181
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.6587774157524109,
      "learning_rate": 0.0005517315325551794,
      "loss": 2.4905,
      "step": 42182
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.6281653642654419,
      "learning_rate": 0.0005517293073914716,
      "loss": 2.4683,
      "step": 42183
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.6818265318870544,
      "learning_rate": 0.0005517270821809625,
      "loss": 2.0434,
      "step": 42184
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.653495192527771,
      "learning_rate": 0.0005517248569236526,
      "loss": 2.4428,
      "step": 42185
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.6606438755989075,
      "learning_rate": 0.0005517226316195424,
      "loss": 2.094,
      "step": 42186
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.7224717736244202,
      "learning_rate": 0.000551720406268632,
      "loss": 2.3136,
      "step": 42187
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.6688154339790344,
      "learning_rate": 0.0005517181808709222,
      "loss": 2.4589,
      "step": 42188
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.6833460330963135,
      "learning_rate": 0.0005517159554264133,
      "loss": 2.2483,
      "step": 42189
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.683292031288147,
      "learning_rate": 0.0005517137299351054,
      "loss": 2.261,
      "step": 42190
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.6377330422401428,
      "learning_rate": 0.0005517115043969993,
      "loss": 2.1354,
      "step": 42191
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.6822888851165771,
      "learning_rate": 0.0005517092788120953,
      "loss": 2.3918,
      "step": 42192
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.7373728156089783,
      "learning_rate": 0.0005517070531803938,
      "loss": 2.455,
      "step": 42193
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.675836980342865,
      "learning_rate": 0.0005517048275018952,
      "loss": 2.4024,
      "step": 42194
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.6832268238067627,
      "learning_rate": 0.0005517026017765998,
      "loss": 2.094,
      "step": 42195
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.6832877397537231,
      "learning_rate": 0.0005517003760045084,
      "loss": 2.2142,
      "step": 42196
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.7094905376434326,
      "learning_rate": 0.000551698150185621,
      "loss": 2.3557,
      "step": 42197
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.6714866757392883,
      "learning_rate": 0.0005516959243199381,
      "loss": 2.3762,
      "step": 42198
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.6837754845619202,
      "learning_rate": 0.0005516936984074602,
      "loss": 2.1015,
      "step": 42199
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.728691816329956,
      "learning_rate": 0.0005516914724481878,
      "loss": 2.5917,
      "step": 42200
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.7746064066886902,
      "learning_rate": 0.000551689246442121,
      "loss": 2.3861,
      "step": 42201
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.7088947296142578,
      "learning_rate": 0.0005516870203892606,
      "loss": 2.5446,
      "step": 42202
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.7068116664886475,
      "learning_rate": 0.0005516847942896068,
      "loss": 2.4606,
      "step": 42203
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.8713075518608093,
      "learning_rate": 0.00055168256814316,
      "loss": 2.3152,
      "step": 42204
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.6906656622886658,
      "learning_rate": 0.0005516803419499208,
      "loss": 2.1829,
      "step": 42205
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.7633633017539978,
      "learning_rate": 0.0005516781157098893,
      "loss": 2.1332,
      "step": 42206
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.868163526058197,
      "learning_rate": 0.0005516758894230661,
      "loss": 2.22,
      "step": 42207
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.6992200016975403,
      "learning_rate": 0.0005516736630894516,
      "loss": 2.256,
      "step": 42208
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.7218912243843079,
      "learning_rate": 0.0005516714367090463,
      "loss": 2.1686,
      "step": 42209
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.7299289703369141,
      "learning_rate": 0.0005516692102818505,
      "loss": 2.138,
      "step": 42210
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.7468902468681335,
      "learning_rate": 0.0005516669838078646,
      "loss": 2.2577,
      "step": 42211
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.6647409200668335,
      "learning_rate": 0.0005516647572870892,
      "loss": 2.2998,
      "step": 42212
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.8302561044692993,
      "learning_rate": 0.0005516625307195244,
      "loss": 2.1661,
      "step": 42213
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.6597766280174255,
      "learning_rate": 0.0005516603041051709,
      "loss": 2.3196,
      "step": 42214
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.6793602705001831,
      "learning_rate": 0.0005516580774440288,
      "loss": 2.4324,
      "step": 42215
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.7553501725196838,
      "learning_rate": 0.0005516558507360989,
      "loss": 2.5483,
      "step": 42216
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.7276190519332886,
      "learning_rate": 0.0005516536239813814,
      "loss": 2.0012,
      "step": 42217
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.6138545870780945,
      "learning_rate": 0.0005516513971798768,
      "loss": 2.3488,
      "step": 42218
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.6890594959259033,
      "learning_rate": 0.0005516491703315854,
      "loss": 2.4822,
      "step": 42219
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.8304433822631836,
      "learning_rate": 0.0005516469434365076,
      "loss": 2.1873,
      "step": 42220
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.6664363145828247,
      "learning_rate": 0.0005516447164946439,
      "loss": 2.4201,
      "step": 42221
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.7133288383483887,
      "learning_rate": 0.0005516424895059947,
      "loss": 2.341,
      "step": 42222
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.712568461894989,
      "learning_rate": 0.0005516402624705605,
      "loss": 2.2026,
      "step": 42223
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.8303015828132629,
      "learning_rate": 0.0005516380353883415,
      "loss": 2.6334,
      "step": 42224
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.6363911032676697,
      "learning_rate": 0.0005516358082593383,
      "loss": 2.0407,
      "step": 42225
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.6701568961143494,
      "learning_rate": 0.0005516335810835513,
      "loss": 2.2508,
      "step": 42226
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.8003566861152649,
      "learning_rate": 0.0005516313538609808,
      "loss": 2.4414,
      "step": 42227
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.6980895400047302,
      "learning_rate": 0.0005516291265916274,
      "loss": 2.2342,
      "step": 42228
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.658704936504364,
      "learning_rate": 0.0005516268992754913,
      "loss": 2.141,
      "step": 42229
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.6456117033958435,
      "learning_rate": 0.0005516246719125729,
      "loss": 1.9453,
      "step": 42230
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.8120989799499512,
      "learning_rate": 0.0005516224445028729,
      "loss": 2.5148,
      "step": 42231
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.63909512758255,
      "learning_rate": 0.0005516202170463914,
      "loss": 2.1838,
      "step": 42232
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.6606488227844238,
      "learning_rate": 0.0005516179895431291,
      "loss": 2.0915,
      "step": 42233
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.6578055620193481,
      "learning_rate": 0.0005516157619930861,
      "loss": 2.131,
      "step": 42234
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.6607282161712646,
      "learning_rate": 0.0005516135343962632,
      "loss": 2.2843,
      "step": 42235
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.7297627329826355,
      "learning_rate": 0.0005516113067526604,
      "loss": 2.2529,
      "step": 42236
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.6540560722351074,
      "learning_rate": 0.0005516090790622784,
      "loss": 2.2478,
      "step": 42237
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.661406934261322,
      "learning_rate": 0.0005516068513251176,
      "loss": 2.5189,
      "step": 42238
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.697195291519165,
      "learning_rate": 0.0005516046235411781,
      "loss": 2.3182,
      "step": 42239
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.7475652098655701,
      "learning_rate": 0.0005516023957104608,
      "loss": 2.4106,
      "step": 42240
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.8256678581237793,
      "learning_rate": 0.0005516001678329657,
      "loss": 2.1533,
      "step": 42241
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.7356605529785156,
      "learning_rate": 0.0005515979399086934,
      "loss": 2.2981,
      "step": 42242
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.6675333976745605,
      "learning_rate": 0.0005515957119376444,
      "loss": 2.0931,
      "step": 42243
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.7036066055297852,
      "learning_rate": 0.0005515934839198189,
      "loss": 2.4261,
      "step": 42244
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.7257797122001648,
      "learning_rate": 0.0005515912558552175,
      "loss": 2.1103,
      "step": 42245
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.709857165813446,
      "learning_rate": 0.0005515890277438404,
      "loss": 2.28,
      "step": 42246
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.7447788715362549,
      "learning_rate": 0.0005515867995856882,
      "loss": 2.0847,
      "step": 42247
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.6760309934616089,
      "learning_rate": 0.0005515845713807613,
      "loss": 2.1287,
      "step": 42248
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.7179445624351501,
      "learning_rate": 0.0005515823431290602,
      "loss": 1.981,
      "step": 42249
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.7644441723823547,
      "learning_rate": 0.000551580114830585,
      "loss": 2.425,
      "step": 42250
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.7092260122299194,
      "learning_rate": 0.0005515778864853365,
      "loss": 2.1672,
      "step": 42251
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.7676571607589722,
      "learning_rate": 0.0005515756580933147,
      "loss": 2.2776,
      "step": 42252
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.6938890218734741,
      "learning_rate": 0.0005515734296545204,
      "loss": 2.2593,
      "step": 42253
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.7173007130622864,
      "learning_rate": 0.0005515712011689538,
      "loss": 2.2775,
      "step": 42254
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.7038207650184631,
      "learning_rate": 0.0005515689726366153,
      "loss": 2.3264,
      "step": 42255
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.6644437909126282,
      "learning_rate": 0.0005515667440575055,
      "loss": 2.0072,
      "step": 42256
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.7500032186508179,
      "learning_rate": 0.0005515645154316245,
      "loss": 2.5269,
      "step": 42257
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.8056651949882507,
      "learning_rate": 0.0005515622867589732,
      "loss": 2.394,
      "step": 42258
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.7089685797691345,
      "learning_rate": 0.0005515600580395516,
      "loss": 2.3176,
      "step": 42259
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.6730933785438538,
      "learning_rate": 0.0005515578292733601,
      "loss": 2.2246,
      "step": 42260
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.7334414124488831,
      "learning_rate": 0.0005515556004603993,
      "loss": 2.429,
      "step": 42261
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.6393905282020569,
      "learning_rate": 0.0005515533716006697,
      "loss": 2.576,
      "step": 42262
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.7102109789848328,
      "learning_rate": 0.0005515511426941714,
      "loss": 2.5574,
      "step": 42263
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.677758514881134,
      "learning_rate": 0.0005515489137409052,
      "loss": 2.2731,
      "step": 42264
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.7429383993148804,
      "learning_rate": 0.0005515466847408712,
      "loss": 2.3458,
      "step": 42265
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.6762034893035889,
      "learning_rate": 0.0005515444556940699,
      "loss": 2.2163,
      "step": 42266
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.6623457074165344,
      "learning_rate": 0.0005515422266005018,
      "loss": 2.3052,
      "step": 42267
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.643526017665863,
      "learning_rate": 0.0005515399974601672,
      "loss": 2.1051,
      "step": 42268
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.688556432723999,
      "learning_rate": 0.0005515377682730665,
      "loss": 2.2163,
      "step": 42269
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.6715684533119202,
      "learning_rate": 0.0005515355390392002,
      "loss": 2.1521,
      "step": 42270
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.7295352220535278,
      "learning_rate": 0.0005515333097585688,
      "loss": 2.3124,
      "step": 42271
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.8358188271522522,
      "learning_rate": 0.0005515310804311726,
      "loss": 2.2698,
      "step": 42272
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.6360698342323303,
      "learning_rate": 0.000551528851057012,
      "loss": 2.2423,
      "step": 42273
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.7722457647323608,
      "learning_rate": 0.0005515266216360874,
      "loss": 2.3651,
      "step": 42274
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.745771586894989,
      "learning_rate": 0.0005515243921683994,
      "loss": 2.0898,
      "step": 42275
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.7483538389205933,
      "learning_rate": 0.0005515221626539481,
      "loss": 2.2933,
      "step": 42276
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.6646970510482788,
      "learning_rate": 0.0005515199330927341,
      "loss": 2.3018,
      "step": 42277
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.6828956604003906,
      "learning_rate": 0.0005515177034847578,
      "loss": 2.1082,
      "step": 42278
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.7099699974060059,
      "learning_rate": 0.0005515154738300196,
      "loss": 2.1516,
      "step": 42279
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.7208942770957947,
      "learning_rate": 0.00055151324412852,
      "loss": 2.3325,
      "step": 42280
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.8058203458786011,
      "learning_rate": 0.0005515110143802593,
      "loss": 2.4183,
      "step": 42281
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.6921218633651733,
      "learning_rate": 0.000551508784585238,
      "loss": 2.1784,
      "step": 42282
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.6911364793777466,
      "learning_rate": 0.0005515065547434564,
      "loss": 2.2619,
      "step": 42283
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.7109805345535278,
      "learning_rate": 0.0005515043248549151,
      "loss": 2.3802,
      "step": 42284
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.8239279985427856,
      "learning_rate": 0.0005515020949196143,
      "loss": 2.3339,
      "step": 42285
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.74989914894104,
      "learning_rate": 0.0005514998649375545,
      "loss": 2.1271,
      "step": 42286
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.7247152328491211,
      "learning_rate": 0.000551497634908736,
      "loss": 2.1943,
      "step": 42287
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.9522680640220642,
      "learning_rate": 0.0005514954048331596,
      "loss": 2.2372,
      "step": 42288
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.6984752416610718,
      "learning_rate": 0.0005514931747108254,
      "loss": 2.4289,
      "step": 42289
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.7460967898368835,
      "learning_rate": 0.0005514909445417337,
      "loss": 2.1905,
      "step": 42290
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.7661181688308716,
      "learning_rate": 0.0005514887143258853,
      "loss": 2.3332,
      "step": 42291
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.6963400840759277,
      "learning_rate": 0.0005514864840632802,
      "loss": 1.9975,
      "step": 42292
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.6222767233848572,
      "learning_rate": 0.0005514842537539192,
      "loss": 2.2942,
      "step": 42293
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.7060545682907104,
      "learning_rate": 0.0005514820233978025,
      "loss": 2.295,
      "step": 42294
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.6734851598739624,
      "learning_rate": 0.0005514797929949305,
      "loss": 2.5024,
      "step": 42295
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.7230744957923889,
      "learning_rate": 0.0005514775625453036,
      "loss": 2.5059,
      "step": 42296
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.7097035646438599,
      "learning_rate": 0.0005514753320489223,
      "loss": 2.3486,
      "step": 42297
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.7229562401771545,
      "learning_rate": 0.000551473101505787,
      "loss": 2.1901,
      "step": 42298
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.6529744863510132,
      "learning_rate": 0.0005514708709158982,
      "loss": 2.4033,
      "step": 42299
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.6954403519630432,
      "learning_rate": 0.0005514686402792562,
      "loss": 2.2874,
      "step": 42300
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.7047312259674072,
      "learning_rate": 0.0005514664095958613,
      "loss": 2.3299,
      "step": 42301
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.740903913974762,
      "learning_rate": 0.0005514641788657141,
      "loss": 2.5047,
      "step": 42302
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.777539849281311,
      "learning_rate": 0.0005514619480888151,
      "loss": 2.3776,
      "step": 42303
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.7055615782737732,
      "learning_rate": 0.0005514597172651644,
      "loss": 2.4582,
      "step": 42304
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.6636202335357666,
      "learning_rate": 0.0005514574863947627,
      "loss": 2.2838,
      "step": 42305
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.7694829106330872,
      "learning_rate": 0.0005514552554776102,
      "loss": 2.384,
      "step": 42306
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.729458212852478,
      "learning_rate": 0.0005514530245137075,
      "loss": 2.3688,
      "step": 42307
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.8242834210395813,
      "learning_rate": 0.0005514507935030549,
      "loss": 2.1624,
      "step": 42308
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.687251091003418,
      "learning_rate": 0.0005514485624456528,
      "loss": 2.3076,
      "step": 42309
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.7275840640068054,
      "learning_rate": 0.0005514463313415018,
      "loss": 2.4059,
      "step": 42310
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.9209998846054077,
      "learning_rate": 0.0005514441001906021,
      "loss": 2.3934,
      "step": 42311
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.7990274429321289,
      "learning_rate": 0.0005514418689929543,
      "loss": 2.2547,
      "step": 42312
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.6859052777290344,
      "learning_rate": 0.0005514396377485586,
      "loss": 2.2052,
      "step": 42313
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.8511735796928406,
      "learning_rate": 0.0005514374064574156,
      "loss": 2.2659,
      "step": 42314
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.7817888855934143,
      "learning_rate": 0.0005514351751195255,
      "loss": 2.2609,
      "step": 42315
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.6401249766349792,
      "learning_rate": 0.000551432943734889,
      "loss": 2.4392,
      "step": 42316
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.7937461733818054,
      "learning_rate": 0.0005514307123035062,
      "loss": 1.9494,
      "step": 42317
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.725976824760437,
      "learning_rate": 0.0005514284808253779,
      "loss": 2.4165,
      "step": 42318
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.6368350982666016,
      "learning_rate": 0.0005514262493005041,
      "loss": 2.0916,
      "step": 42319
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.7061471939086914,
      "learning_rate": 0.0005514240177288855,
      "loss": 2.1395,
      "step": 42320
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.804506242275238,
      "learning_rate": 0.0005514217861105226,
      "loss": 2.0817,
      "step": 42321
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.7974497675895691,
      "learning_rate": 0.0005514195544454153,
      "loss": 2.2531,
      "step": 42322
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.7055973410606384,
      "learning_rate": 0.0005514173227335646,
      "loss": 2.1886,
      "step": 42323
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.792117178440094,
      "learning_rate": 0.0005514150909749706,
      "loss": 2.2657,
      "step": 42324
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.6532840132713318,
      "learning_rate": 0.0005514128591696338,
      "loss": 2.099,
      "step": 42325
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.7249898314476013,
      "learning_rate": 0.0005514106273175546,
      "loss": 2.4121,
      "step": 42326
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.6565032005310059,
      "learning_rate": 0.0005514083954187334,
      "loss": 2.0115,
      "step": 42327
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.6148461699485779,
      "learning_rate": 0.0005514061634731707,
      "loss": 2.3668,
      "step": 42328
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.6989065408706665,
      "learning_rate": 0.0005514039314808668,
      "loss": 2.1371,
      "step": 42329
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.7428250312805176,
      "learning_rate": 0.0005514016994418221,
      "loss": 2.2674,
      "step": 42330
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.7598682045936584,
      "learning_rate": 0.0005513994673560373,
      "loss": 2.1595,
      "step": 42331
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.6814383864402771,
      "learning_rate": 0.0005513972352235123,
      "loss": 2.0952,
      "step": 42332
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.7036347389221191,
      "learning_rate": 0.000551395003044248,
      "loss": 2.2673,
      "step": 42333
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.6951266527175903,
      "learning_rate": 0.0005513927708182446,
      "loss": 2.3248,
      "step": 42334
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.6920085549354553,
      "learning_rate": 0.0005513905385455025,
      "loss": 2.5488,
      "step": 42335
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.7131156325340271,
      "learning_rate": 0.0005513883062260222,
      "loss": 2.3067,
      "step": 42336
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.6872478127479553,
      "learning_rate": 0.000551386073859804,
      "loss": 2.3203,
      "step": 42337
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.7292324900627136,
      "learning_rate": 0.0005513838414468484,
      "loss": 2.3713,
      "step": 42338
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.7416979074478149,
      "learning_rate": 0.0005513816089871559,
      "loss": 2.2603,
      "step": 42339
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.7943540811538696,
      "learning_rate": 0.0005513793764807266,
      "loss": 2.3174,
      "step": 42340
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.6473451852798462,
      "learning_rate": 0.0005513771439275614,
      "loss": 2.3078,
      "step": 42341
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.6797417402267456,
      "learning_rate": 0.0005513749113276602,
      "loss": 2.2456,
      "step": 42342
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.6499501466751099,
      "learning_rate": 0.0005513726786810239,
      "loss": 2.1372,
      "step": 42343
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.6973513960838318,
      "learning_rate": 0.0005513704459876524,
      "loss": 2.2737,
      "step": 42344
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.7376603484153748,
      "learning_rate": 0.0005513682132475466,
      "loss": 2.3326,
      "step": 42345
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.6772572994232178,
      "learning_rate": 0.0005513659804607066,
      "loss": 2.0324,
      "step": 42346
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.6607681512832642,
      "learning_rate": 0.000551363747627133,
      "loss": 1.9908,
      "step": 42347
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.7385778427124023,
      "learning_rate": 0.0005513615147468261,
      "loss": 2.244,
      "step": 42348
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.8818199634552002,
      "learning_rate": 0.0005513592818197863,
      "loss": 2.3097,
      "step": 42349
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.8366350531578064,
      "learning_rate": 0.0005513570488460141,
      "loss": 2.366,
      "step": 42350
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.6454943418502808,
      "learning_rate": 0.0005513548158255099,
      "loss": 2.3575,
      "step": 42351
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.1213865280151367,
      "learning_rate": 0.0005513525827582741,
      "loss": 2.2517,
      "step": 42352
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.731737494468689,
      "learning_rate": 0.000551350349644307,
      "loss": 2.3513,
      "step": 42353
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.7285383939743042,
      "learning_rate": 0.0005513481164836092,
      "loss": 2.2331,
      "step": 42354
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.9617179036140442,
      "learning_rate": 0.0005513458832761811,
      "loss": 2.1268,
      "step": 42355
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.8163718581199646,
      "learning_rate": 0.000551343650022023,
      "loss": 2.3333,
      "step": 42356
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.6764796376228333,
      "learning_rate": 0.0005513414167211354,
      "loss": 2.2713,
      "step": 42357
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.1234585046768188,
      "learning_rate": 0.0005513391833735186,
      "loss": 2.1492,
      "step": 42358
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.6546319127082825,
      "learning_rate": 0.0005513369499791732,
      "loss": 2.1741,
      "step": 42359
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.6706700921058655,
      "learning_rate": 0.0005513347165380995,
      "loss": 2.2425,
      "step": 42360
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.705829918384552,
      "learning_rate": 0.0005513324830502979,
      "loss": 2.2204,
      "step": 42361
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.7469449639320374,
      "learning_rate": 0.0005513302495157689,
      "loss": 2.2519,
      "step": 42362
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.71206134557724,
      "learning_rate": 0.0005513280159345128,
      "loss": 2.4171,
      "step": 42363
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.6655301451683044,
      "learning_rate": 0.0005513257823065301,
      "loss": 2.1593,
      "step": 42364
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.717936098575592,
      "learning_rate": 0.0005513235486318212,
      "loss": 2.3617,
      "step": 42365
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.8287782669067383,
      "learning_rate": 0.0005513213149103864,
      "loss": 2.1235,
      "step": 42366
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.6707274317741394,
      "learning_rate": 0.0005513190811422264,
      "loss": 2.4476,
      "step": 42367
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.7308635115623474,
      "learning_rate": 0.0005513168473273414,
      "loss": 2.3805,
      "step": 42368
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.9986806511878967,
      "learning_rate": 0.0005513146134657318,
      "loss": 2.0642,
      "step": 42369
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.6461469531059265,
      "learning_rate": 0.0005513123795573981,
      "loss": 2.1963,
      "step": 42370
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.8155930042266846,
      "learning_rate": 0.0005513101456023407,
      "loss": 2.3262,
      "step": 42371
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.8211585283279419,
      "learning_rate": 0.0005513079116005599,
      "loss": 2.229,
      "step": 42372
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.6181225180625916,
      "learning_rate": 0.0005513056775520563,
      "loss": 2.1661,
      "step": 42373
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.7002351880073547,
      "learning_rate": 0.0005513034434568302,
      "loss": 2.3615,
      "step": 42374
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.711600661277771,
      "learning_rate": 0.0005513012093148821,
      "loss": 2.2612,
      "step": 42375
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.7772454619407654,
      "learning_rate": 0.0005512989751262123,
      "loss": 2.0668,
      "step": 42376
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.6900022029876709,
      "learning_rate": 0.0005512967408908213,
      "loss": 2.1447,
      "step": 42377
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.7310569286346436,
      "learning_rate": 0.0005512945066087095,
      "loss": 2.1792,
      "step": 42378
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.0475714206695557,
      "learning_rate": 0.0005512922722798773,
      "loss": 2.3897,
      "step": 42379
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.6585114598274231,
      "learning_rate": 0.0005512900379043251,
      "loss": 1.9878,
      "step": 42380
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.7813503742218018,
      "learning_rate": 0.0005512878034820535,
      "loss": 2.3443,
      "step": 42381
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.7367669939994812,
      "learning_rate": 0.0005512855690130626,
      "loss": 2.2346,
      "step": 42382
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.6966529488563538,
      "learning_rate": 0.0005512833344973528,
      "loss": 2.23,
      "step": 42383
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.653846263885498,
      "learning_rate": 0.000551281099934925,
      "loss": 2.3111,
      "step": 42384
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.695616602897644,
      "learning_rate": 0.0005512788653257791,
      "loss": 2.3488,
      "step": 42385
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.8805540204048157,
      "learning_rate": 0.0005512766306699159,
      "loss": 2.2721,
      "step": 42386
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.6807222366333008,
      "learning_rate": 0.0005512743959673355,
      "loss": 2.3213,
      "step": 42387
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.7217487096786499,
      "learning_rate": 0.0005512721612180385,
      "loss": 2.4677,
      "step": 42388
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.7408615946769714,
      "learning_rate": 0.0005512699264220251,
      "loss": 2.286,
      "step": 42389
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.6914253830909729,
      "learning_rate": 0.000551267691579296,
      "loss": 2.0839,
      "step": 42390
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.7108940482139587,
      "learning_rate": 0.0005512654566898515,
      "loss": 2.1405,
      "step": 42391
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.7444449663162231,
      "learning_rate": 0.0005512632217536921,
      "loss": 2.2279,
      "step": 42392
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.7818928956985474,
      "learning_rate": 0.000551260986770818,
      "loss": 2.3616,
      "step": 42393
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.6486613750457764,
      "learning_rate": 0.0005512587517412298,
      "loss": 2.4042,
      "step": 42394
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.8515828251838684,
      "learning_rate": 0.0005512565166649278,
      "loss": 2.3313,
      "step": 42395
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.636338472366333,
      "learning_rate": 0.0005512542815419126,
      "loss": 2.2217,
      "step": 42396
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.6547828912734985,
      "learning_rate": 0.0005512520463721843,
      "loss": 2.1734,
      "step": 42397
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.6676141619682312,
      "learning_rate": 0.0005512498111557436,
      "loss": 1.9967,
      "step": 42398
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.9478825926780701,
      "learning_rate": 0.0005512475758925907,
      "loss": 2.4357,
      "step": 42399
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.7065176367759705,
      "learning_rate": 0.0005512453405827263,
      "loss": 2.111,
      "step": 42400
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.7121185660362244,
      "learning_rate": 0.0005512431052261506,
      "loss": 2.4612,
      "step": 42401
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.7785927653312683,
      "learning_rate": 0.000551240869822864,
      "loss": 2.1888,
      "step": 42402
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.6855864524841309,
      "learning_rate": 0.0005512386343728671,
      "loss": 2.3542,
      "step": 42403
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.6448788642883301,
      "learning_rate": 0.00055123639887616,
      "loss": 2.0908,
      "step": 42404
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.7342696189880371,
      "learning_rate": 0.0005512341633327435,
      "loss": 2.2936,
      "step": 42405
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.7225639224052429,
      "learning_rate": 0.0005512319277426178,
      "loss": 2.2761,
      "step": 42406
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.8406624794006348,
      "learning_rate": 0.0005512296921057833,
      "loss": 2.2824,
      "step": 42407
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.742440938949585,
      "learning_rate": 0.0005512274564222404,
      "loss": 2.0958,
      "step": 42408
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.8066383600234985,
      "learning_rate": 0.0005512252206919897,
      "loss": 2.2891,
      "step": 42409
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.7174088954925537,
      "learning_rate": 0.0005512229849150315,
      "loss": 2.3678,
      "step": 42410
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.6270967125892639,
      "learning_rate": 0.000551220749091366,
      "loss": 2.078,
      "step": 42411
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.6836211085319519,
      "learning_rate": 0.0005512185132209939,
      "loss": 2.1427,
      "step": 42412
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.7195379734039307,
      "learning_rate": 0.0005512162773039157,
      "loss": 2.2476,
      "step": 42413
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.6779964566230774,
      "learning_rate": 0.0005512140413401316,
      "loss": 2.0586,
      "step": 42414
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.6284322738647461,
      "learning_rate": 0.0005512118053296419,
      "loss": 2.3324,
      "step": 42415
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.7663077712059021,
      "learning_rate": 0.0005512095692724473,
      "loss": 2.1096,
      "step": 42416
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.7308064699172974,
      "learning_rate": 0.0005512073331685482,
      "loss": 2.2824,
      "step": 42417
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.7678995132446289,
      "learning_rate": 0.0005512050970179446,
      "loss": 2.3861,
      "step": 42418
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.6596054434776306,
      "learning_rate": 0.0005512028608206375,
      "loss": 2.2227,
      "step": 42419
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.7130218744277954,
      "learning_rate": 0.000551200624576627,
      "loss": 2.0985,
      "step": 42420
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.7415775656700134,
      "learning_rate": 0.0005511983882859135,
      "loss": 2.2525,
      "step": 42421
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.6712007522583008,
      "learning_rate": 0.0005511961519484975,
      "loss": 2.1586,
      "step": 42422
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.6630855798721313,
      "learning_rate": 0.0005511939155643795,
      "loss": 2.3375,
      "step": 42423
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.6456459760665894,
      "learning_rate": 0.0005511916791335597,
      "loss": 2.0993,
      "step": 42424
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.7263168096542358,
      "learning_rate": 0.0005511894426560387,
      "loss": 2.1149,
      "step": 42425
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.6930454969406128,
      "learning_rate": 0.0005511872061318168,
      "loss": 2.3224,
      "step": 42426
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.6654497981071472,
      "learning_rate": 0.0005511849695608944,
      "loss": 2.2995,
      "step": 42427
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.743388295173645,
      "learning_rate": 0.0005511827329432719,
      "loss": 2.1991,
      "step": 42428
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.7215173840522766,
      "learning_rate": 0.00055118049627895,
      "loss": 2.3183,
      "step": 42429
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.7461299300193787,
      "learning_rate": 0.0005511782595679288,
      "loss": 2.1717,
      "step": 42430
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.7119920253753662,
      "learning_rate": 0.0005511760228102089,
      "loss": 2.3665,
      "step": 42431
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.7271958589553833,
      "learning_rate": 0.0005511737860057906,
      "loss": 2.2004,
      "step": 42432
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.6973679661750793,
      "learning_rate": 0.0005511715491546742,
      "loss": 2.2975,
      "step": 42433
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.6541284918785095,
      "learning_rate": 0.0005511693122568605,
      "loss": 2.186,
      "step": 42434
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.6617321372032166,
      "learning_rate": 0.0005511670753123494,
      "loss": 2.212,
      "step": 42435
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.6720463037490845,
      "learning_rate": 0.0005511648383211418,
      "loss": 2.0745,
      "step": 42436
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.6935226917266846,
      "learning_rate": 0.0005511626012832379,
      "loss": 2.352,
      "step": 42437
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.6313481330871582,
      "learning_rate": 0.0005511603641986382,
      "loss": 2.3427,
      "step": 42438
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.8004462122917175,
      "learning_rate": 0.0005511581270673429,
      "loss": 2.285,
      "step": 42439
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.7101153135299683,
      "learning_rate": 0.0005511558898893527,
      "loss": 2.359,
      "step": 42440
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.6574092507362366,
      "learning_rate": 0.0005511536526646677,
      "loss": 2.2445,
      "step": 42441
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.6890095472335815,
      "learning_rate": 0.0005511514153932886,
      "loss": 2.353,
      "step": 42442
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.6916987299919128,
      "learning_rate": 0.0005511491780752158,
      "loss": 2.2438,
      "step": 42443
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.664280891418457,
      "learning_rate": 0.0005511469407104496,
      "loss": 2.4212,
      "step": 42444
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.7088289856910706,
      "learning_rate": 0.0005511447032989904,
      "loss": 2.2818,
      "step": 42445
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.7260628938674927,
      "learning_rate": 0.0005511424658408386,
      "loss": 2.1371,
      "step": 42446
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.6827343106269836,
      "learning_rate": 0.0005511402283359946,
      "loss": 2.3279,
      "step": 42447
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.676505982875824,
      "learning_rate": 0.0005511379907844591,
      "loss": 2.288,
      "step": 42448
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.7284757494926453,
      "learning_rate": 0.0005511357531862321,
      "loss": 2.3776,
      "step": 42449
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.8250083327293396,
      "learning_rate": 0.0005511335155413144,
      "loss": 2.1086,
      "step": 42450
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.7138622403144836,
      "learning_rate": 0.0005511312778497063,
      "loss": 2.1746,
      "step": 42451
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.9043546915054321,
      "learning_rate": 0.000551129040111408,
      "loss": 2.2164,
      "step": 42452
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.6863371133804321,
      "learning_rate": 0.00055112680232642,
      "loss": 2.3932,
      "step": 42453
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.6651982665061951,
      "learning_rate": 0.000551124564494743,
      "loss": 2.2413,
      "step": 42454
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.9487289786338806,
      "learning_rate": 0.000551122326616377,
      "loss": 2.4834,
      "step": 42455
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.7293195724487305,
      "learning_rate": 0.0005511200886913227,
      "loss": 2.3335,
      "step": 42456
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.7450965642929077,
      "learning_rate": 0.0005511178507195804,
      "loss": 2.196,
      "step": 42457
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.7346744537353516,
      "learning_rate": 0.0005511156127011507,
      "loss": 2.2438,
      "step": 42458
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.70361328125,
      "learning_rate": 0.0005511133746360338,
      "loss": 2.1146,
      "step": 42459
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.7891890406608582,
      "learning_rate": 0.0005511111365242301,
      "loss": 2.2681,
      "step": 42460
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.6907667517662048,
      "learning_rate": 0.0005511088983657401,
      "loss": 2.098,
      "step": 42461
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.9119361639022827,
      "learning_rate": 0.0005511066601605643,
      "loss": 2.1049,
      "step": 42462
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.8156408071517944,
      "learning_rate": 0.000551104421908703,
      "loss": 2.191,
      "step": 42463
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.6673765182495117,
      "learning_rate": 0.0005511021836101567,
      "loss": 2.2498,
      "step": 42464
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.666324257850647,
      "learning_rate": 0.0005510999452649257,
      "loss": 2.1234,
      "step": 42465
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.9435092806816101,
      "learning_rate": 0.0005510977068730105,
      "loss": 2.2713,
      "step": 42466
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.7375255227088928,
      "learning_rate": 0.0005510954684344114,
      "loss": 2.3233,
      "step": 42467
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.7182486653327942,
      "learning_rate": 0.0005510932299491291,
      "loss": 2.3188,
      "step": 42468
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.755953848361969,
      "learning_rate": 0.0005510909914171636,
      "loss": 2.1493,
      "step": 42469
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.897962749004364,
      "learning_rate": 0.0005510887528385157,
      "loss": 2.2973,
      "step": 42470
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.6568105816841125,
      "learning_rate": 0.0005510865142131857,
      "loss": 2.2986,
      "step": 42471
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.9621316194534302,
      "learning_rate": 0.0005510842755411739,
      "loss": 2.2539,
      "step": 42472
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.7091642022132874,
      "learning_rate": 0.0005510820368224808,
      "loss": 2.2609,
      "step": 42473
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.6939765810966492,
      "learning_rate": 0.0005510797980571068,
      "loss": 2.3501,
      "step": 42474
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.7337583899497986,
      "learning_rate": 0.0005510775592450524,
      "loss": 2.3533,
      "step": 42475
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.7149887084960938,
      "learning_rate": 0.0005510753203863179,
      "loss": 2.6071,
      "step": 42476
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.718082845211029,
      "learning_rate": 0.0005510730814809037,
      "loss": 2.1432,
      "step": 42477
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.7319403290748596,
      "learning_rate": 0.0005510708425288102,
      "loss": 2.372,
      "step": 42478
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.8504955172538757,
      "learning_rate": 0.0005510686035300382,
      "loss": 2.2987,
      "step": 42479
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.8092408776283264,
      "learning_rate": 0.0005510663644845875,
      "loss": 2.248,
      "step": 42480
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.741691529750824,
      "learning_rate": 0.000551064125392459,
      "loss": 2.2692,
      "step": 42481
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.7403760552406311,
      "learning_rate": 0.0005510618862536529,
      "loss": 2.2566,
      "step": 42482
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.6418494582176208,
      "learning_rate": 0.0005510596470681696,
      "loss": 1.976,
      "step": 42483
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.6810159683227539,
      "learning_rate": 0.0005510574078360095,
      "loss": 2.0788,
      "step": 42484
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.6841731071472168,
      "learning_rate": 0.0005510551685571733,
      "loss": 2.2527,
      "step": 42485
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.699650764465332,
      "learning_rate": 0.000551052929231661,
      "loss": 2.3874,
      "step": 42486
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.7001851797103882,
      "learning_rate": 0.0005510506898594733,
      "loss": 2.343,
      "step": 42487
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.7521947622299194,
      "learning_rate": 0.0005510484504406106,
      "loss": 2.3713,
      "step": 42488
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.6969462633132935,
      "learning_rate": 0.0005510462109750733,
      "loss": 2.307,
      "step": 42489
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.7961838841438293,
      "learning_rate": 0.0005510439714628616,
      "loss": 2.4749,
      "step": 42490
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.9762333035469055,
      "learning_rate": 0.0005510417319039763,
      "loss": 2.1752,
      "step": 42491
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.7617326974868774,
      "learning_rate": 0.0005510394922984173,
      "loss": 2.0879,
      "step": 42492
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.7887697219848633,
      "learning_rate": 0.0005510372526461855,
      "loss": 2.1158,
      "step": 42493
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.6706022024154663,
      "learning_rate": 0.0005510350129472812,
      "loss": 2.1373,
      "step": 42494
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.7191025018692017,
      "learning_rate": 0.0005510327732017047,
      "loss": 2.4172,
      "step": 42495
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.8339072465896606,
      "learning_rate": 0.0005510305334094565,
      "loss": 2.2836,
      "step": 42496
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.7234290838241577,
      "learning_rate": 0.0005510282935705369,
      "loss": 2.1909,
      "step": 42497
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.7278472781181335,
      "learning_rate": 0.0005510260536849464,
      "loss": 2.171,
      "step": 42498
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.6761277318000793,
      "learning_rate": 0.0005510238137526854,
      "loss": 2.089,
      "step": 42499
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.6838981509208679,
      "learning_rate": 0.0005510215737737545,
      "loss": 2.423,
      "step": 42500
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.755763590335846,
      "learning_rate": 0.0005510193337481538,
      "loss": 2.3983,
      "step": 42501
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.6798708438873291,
      "learning_rate": 0.0005510170936758839,
      "loss": 2.3525,
      "step": 42502
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.66700279712677,
      "learning_rate": 0.0005510148535569452,
      "loss": 2.4577,
      "step": 42503
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.7028853893280029,
      "learning_rate": 0.0005510126133913381,
      "loss": 2.4384,
      "step": 42504
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.6840161085128784,
      "learning_rate": 0.000551010373179063,
      "loss": 2.0515,
      "step": 42505
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.6982190608978271,
      "learning_rate": 0.0005510081329201203,
      "loss": 2.3402,
      "step": 42506
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.652085542678833,
      "learning_rate": 0.0005510058926145105,
      "loss": 2.1881,
      "step": 42507
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.6580065488815308,
      "learning_rate": 0.000551003652262234,
      "loss": 2.1204,
      "step": 42508
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.7517198324203491,
      "learning_rate": 0.0005510014118632912,
      "loss": 2.2372,
      "step": 42509
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.6590657234191895,
      "learning_rate": 0.0005509991714176824,
      "loss": 2.1462,
      "step": 42510
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.7437541484832764,
      "learning_rate": 0.0005509969309254082,
      "loss": 2.4417,
      "step": 42511
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.680122971534729,
      "learning_rate": 0.0005509946903864689,
      "loss": 2.2474,
      "step": 42512
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.9144816398620605,
      "learning_rate": 0.0005509924498008649,
      "loss": 2.1654,
      "step": 42513
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.6861791014671326,
      "learning_rate": 0.0005509902091685966,
      "loss": 2.1226,
      "step": 42514
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.6775332093238831,
      "learning_rate": 0.0005509879684896646,
      "loss": 2.4328,
      "step": 42515
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.6602488160133362,
      "learning_rate": 0.0005509857277640693,
      "loss": 2.4218,
      "step": 42516
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.716275155544281,
      "learning_rate": 0.0005509834869918109,
      "loss": 2.3793,
      "step": 42517
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.6286061406135559,
      "learning_rate": 0.00055098124617289,
      "loss": 2.3186,
      "step": 42518
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.709674596786499,
      "learning_rate": 0.0005509790053073068,
      "loss": 2.3271,
      "step": 42519
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.7230147123336792,
      "learning_rate": 0.0005509767643950619,
      "loss": 2.244,
      "step": 42520
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.6539015173912048,
      "learning_rate": 0.0005509745234361557,
      "loss": 2.1267,
      "step": 42521
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.8123343586921692,
      "learning_rate": 0.0005509722824305887,
      "loss": 2.3865,
      "step": 42522
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.7942054271697998,
      "learning_rate": 0.0005509700413783612,
      "loss": 2.313,
      "step": 42523
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.6754631996154785,
      "learning_rate": 0.0005509678002794735,
      "loss": 2.4681,
      "step": 42524
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.6880298852920532,
      "learning_rate": 0.0005509655591339263,
      "loss": 2.2366,
      "step": 42525
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.7127578258514404,
      "learning_rate": 0.0005509633179417198,
      "loss": 2.2419,
      "step": 42526
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.6416394710540771,
      "learning_rate": 0.0005509610767028544,
      "loss": 2.0544,
      "step": 42527
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.6868161559104919,
      "learning_rate": 0.0005509588354173306,
      "loss": 2.3323,
      "step": 42528
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.7427076697349548,
      "learning_rate": 0.000550956594085149,
      "loss": 2.4198,
      "step": 42529
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.7710552215576172,
      "learning_rate": 0.0005509543527063098,
      "loss": 2.4449,
      "step": 42530
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.6685965657234192,
      "learning_rate": 0.0005509521112808133,
      "loss": 2.2088,
      "step": 42531
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.6141152381896973,
      "learning_rate": 0.0005509498698086601,
      "loss": 1.9745,
      "step": 42532
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.6758230924606323,
      "learning_rate": 0.0005509476282898506,
      "loss": 2.4045,
      "step": 42533
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.6706002354621887,
      "learning_rate": 0.0005509453867243852,
      "loss": 2.2914,
      "step": 42534
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.6696701049804688,
      "learning_rate": 0.0005509431451122644,
      "loss": 2.1856,
      "step": 42535
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.6790997982025146,
      "learning_rate": 0.0005509409034534884,
      "loss": 2.6093,
      "step": 42536
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.7387083172798157,
      "learning_rate": 0.0005509386617480579,
      "loss": 2.4806,
      "step": 42537
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.6982165575027466,
      "learning_rate": 0.000550936419995973,
      "loss": 2.3814,
      "step": 42538
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.7448173761367798,
      "learning_rate": 0.0005509341781972344,
      "loss": 2.1837,
      "step": 42539
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.6789906620979309,
      "learning_rate": 0.0005509319363518423,
      "loss": 2.3192,
      "step": 42540
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.8391270041465759,
      "learning_rate": 0.0005509296944597973,
      "loss": 2.3813,
      "step": 42541
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.709915816783905,
      "learning_rate": 0.0005509274525210997,
      "loss": 2.4587,
      "step": 42542
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.6999884247779846,
      "learning_rate": 0.0005509252105357499,
      "loss": 2.1624,
      "step": 42543
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.7476174235343933,
      "learning_rate": 0.0005509229685037485,
      "loss": 2.1024,
      "step": 42544
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.6915393471717834,
      "learning_rate": 0.0005509207264250957,
      "loss": 2.366,
      "step": 42545
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.6467667818069458,
      "learning_rate": 0.0005509184842997921,
      "loss": 2.1942,
      "step": 42546
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.7429349422454834,
      "learning_rate": 0.0005509162421278379,
      "loss": 2.393,
      "step": 42547
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.6159615516662598,
      "learning_rate": 0.0005509139999092336,
      "loss": 2.0674,
      "step": 42548
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.6883712410926819,
      "learning_rate": 0.0005509117576439797,
      "loss": 2.3307,
      "step": 42549
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.6751832365989685,
      "learning_rate": 0.0005509095153320765,
      "loss": 2.1915,
      "step": 42550
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.8592265248298645,
      "learning_rate": 0.0005509072729735246,
      "loss": 2.2716,
      "step": 42551
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.6776989698410034,
      "learning_rate": 0.0005509050305683243,
      "loss": 2.2971,
      "step": 42552
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.7725440263748169,
      "learning_rate": 0.000550902788116476,
      "loss": 2.3955,
      "step": 42553
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.7324928641319275,
      "learning_rate": 0.0005509005456179802,
      "loss": 2.3135,
      "step": 42554
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.7073483467102051,
      "learning_rate": 0.0005508983030728371,
      "loss": 2.2164,
      "step": 42555
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.7700434327125549,
      "learning_rate": 0.0005508960604810474,
      "loss": 2.2457,
      "step": 42556
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.6873443126678467,
      "learning_rate": 0.0005508938178426112,
      "loss": 2.082,
      "step": 42557
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.6741026639938354,
      "learning_rate": 0.0005508915751575293,
      "loss": 2.6061,
      "step": 42558
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.7682406306266785,
      "learning_rate": 0.0005508893324258019,
      "loss": 2.3169,
      "step": 42559
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.6394212245941162,
      "learning_rate": 0.0005508870896474294,
      "loss": 2.356,
      "step": 42560
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.6368048787117004,
      "learning_rate": 0.0005508848468224123,
      "loss": 2.6051,
      "step": 42561
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.7325697541236877,
      "learning_rate": 0.0005508826039507508,
      "loss": 2.3553,
      "step": 42562
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.6724783182144165,
      "learning_rate": 0.0005508803610324456,
      "loss": 2.1257,
      "step": 42563
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.666068434715271,
      "learning_rate": 0.000550878118067497,
      "loss": 2.3327,
      "step": 42564
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.7341559529304504,
      "learning_rate": 0.0005508758750559055,
      "loss": 2.2812,
      "step": 42565
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.7415605783462524,
      "learning_rate": 0.0005508736319976713,
      "loss": 2.2069,
      "step": 42566
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.6704504489898682,
      "learning_rate": 0.0005508713888927951,
      "loss": 2.1247,
      "step": 42567
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.896445631980896,
      "learning_rate": 0.000550869145741277,
      "loss": 2.1339,
      "step": 42568
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.6714275479316711,
      "learning_rate": 0.0005508669025431178,
      "loss": 2.2255,
      "step": 42569
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.7014726996421814,
      "learning_rate": 0.0005508646592983175,
      "loss": 2.2201,
      "step": 42570
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.7533894181251526,
      "learning_rate": 0.0005508624160068769,
      "loss": 2.3094,
      "step": 42571
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.9349193572998047,
      "learning_rate": 0.0005508601726687961,
      "loss": 2.376,
      "step": 42572
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.7893264889717102,
      "learning_rate": 0.0005508579292840757,
      "loss": 2.2708,
      "step": 42573
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.8496723771095276,
      "learning_rate": 0.0005508556858527161,
      "loss": 2.3256,
      "step": 42574
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.7169581055641174,
      "learning_rate": 0.0005508534423747178,
      "loss": 2.2247,
      "step": 42575
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.683138906955719,
      "learning_rate": 0.0005508511988500809,
      "loss": 2.1667,
      "step": 42576
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.80726557970047,
      "learning_rate": 0.0005508489552788062,
      "loss": 2.3575,
      "step": 42577
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.7713274955749512,
      "learning_rate": 0.0005508467116608939,
      "loss": 2.2641,
      "step": 42578
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.6673981547355652,
      "learning_rate": 0.0005508444679963444,
      "loss": 2.1763,
      "step": 42579
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.7152774333953857,
      "learning_rate": 0.0005508422242851581,
      "loss": 1.9985,
      "step": 42580
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.8379616737365723,
      "learning_rate": 0.0005508399805273357,
      "loss": 2.2538,
      "step": 42581
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.675242006778717,
      "learning_rate": 0.0005508377367228772,
      "loss": 2.3365,
      "step": 42582
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.6552929282188416,
      "learning_rate": 0.0005508354928717835,
      "loss": 2.4381,
      "step": 42583
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.6287117600440979,
      "learning_rate": 0.0005508332489740545,
      "loss": 2.1631,
      "step": 42584
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.7597419619560242,
      "learning_rate": 0.0005508310050296911,
      "loss": 2.2714,
      "step": 42585
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.6617655158042908,
      "learning_rate": 0.0005508287610386932,
      "loss": 2.2587,
      "step": 42586
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.6409009695053101,
      "learning_rate": 0.0005508265170010617,
      "loss": 2.077,
      "step": 42587
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.6886441707611084,
      "learning_rate": 0.0005508242729167967,
      "loss": 2.3839,
      "step": 42588
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.8124896883964539,
      "learning_rate": 0.0005508220287858988,
      "loss": 2.4967,
      "step": 42589
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.7362368106842041,
      "learning_rate": 0.0005508197846083683,
      "loss": 2.3807,
      "step": 42590
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.6759902834892273,
      "learning_rate": 0.0005508175403842057,
      "loss": 2.1642,
      "step": 42591
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.7095226645469666,
      "learning_rate": 0.0005508152961134114,
      "loss": 2.2594,
      "step": 42592
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.7229483127593994,
      "learning_rate": 0.0005508130517959858,
      "loss": 2.3606,
      "step": 42593
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.7660216093063354,
      "learning_rate": 0.0005508108074319293,
      "loss": 2.3457,
      "step": 42594
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.6410693526268005,
      "learning_rate": 0.0005508085630212424,
      "loss": 2.1896,
      "step": 42595
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.6597742438316345,
      "learning_rate": 0.0005508063185639254,
      "loss": 2.3128,
      "step": 42596
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.6221743226051331,
      "learning_rate": 0.0005508040740599788,
      "loss": 2.254,
      "step": 42597
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.7082403302192688,
      "learning_rate": 0.000550801829509403,
      "loss": 2.1382,
      "step": 42598
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.7700201272964478,
      "learning_rate": 0.0005507995849121984,
      "loss": 2.2092,
      "step": 42599
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.6833880543708801,
      "learning_rate": 0.0005507973402683652,
      "loss": 2.4409,
      "step": 42600
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.7159833908081055,
      "learning_rate": 0.0005507950955779043,
      "loss": 2.4497,
      "step": 42601
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.664889931678772,
      "learning_rate": 0.0005507928508408158,
      "loss": 1.9308,
      "step": 42602
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.6858198642730713,
      "learning_rate": 0.0005507906060571001,
      "loss": 2.1672,
      "step": 42603
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.676088273525238,
      "learning_rate": 0.0005507883612267578,
      "loss": 2.2603,
      "step": 42604
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.7483043670654297,
      "learning_rate": 0.0005507861163497892,
      "loss": 2.3466,
      "step": 42605
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.6755366921424866,
      "learning_rate": 0.0005507838714261947,
      "loss": 2.2864,
      "step": 42606
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.646501362323761,
      "learning_rate": 0.0005507816264559747,
      "loss": 2.2841,
      "step": 42607
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.7116771340370178,
      "learning_rate": 0.0005507793814391297,
      "loss": 2.1755,
      "step": 42608
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.716991126537323,
      "learning_rate": 0.0005507771363756601,
      "loss": 2.3839,
      "step": 42609
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.6645360589027405,
      "learning_rate": 0.0005507748912655663,
      "loss": 2.3161,
      "step": 42610
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.7670716047286987,
      "learning_rate": 0.0005507726461088487,
      "loss": 2.321,
      "step": 42611
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.6802992224693298,
      "learning_rate": 0.0005507704009055077,
      "loss": 2.0033,
      "step": 42612
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.6313685774803162,
      "learning_rate": 0.0005507681556555438,
      "loss": 2.1157,
      "step": 42613
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.7176412343978882,
      "learning_rate": 0.0005507659103589573,
      "loss": 2.3262,
      "step": 42614
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.700762927532196,
      "learning_rate": 0.0005507636650157486,
      "loss": 2.1685,
      "step": 42615
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.6675657629966736,
      "learning_rate": 0.0005507614196259184,
      "loss": 2.3033,
      "step": 42616
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.6989364624023438,
      "learning_rate": 0.0005507591741894668,
      "loss": 2.0554,
      "step": 42617
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.6530536413192749,
      "learning_rate": 0.0005507569287063945,
      "loss": 2.2449,
      "step": 42618
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.7323063611984253,
      "learning_rate": 0.0005507546831767016,
      "loss": 2.3093,
      "step": 42619
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.6736290454864502,
      "learning_rate": 0.0005507524376003886,
      "loss": 2.1773,
      "step": 42620
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.7244822382926941,
      "learning_rate": 0.0005507501919774562,
      "loss": 2.3476,
      "step": 42621
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.6395993232727051,
      "learning_rate": 0.0005507479463079044,
      "loss": 2.18,
      "step": 42622
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.7092534899711609,
      "learning_rate": 0.0005507457005917339,
      "loss": 2.3369,
      "step": 42623
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.6999136805534363,
      "learning_rate": 0.0005507434548289451,
      "loss": 2.2622,
      "step": 42624
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.675165057182312,
      "learning_rate": 0.0005507412090195383,
      "loss": 2.0861,
      "step": 42625
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.8011432886123657,
      "learning_rate": 0.0005507389631635141,
      "loss": 2.467,
      "step": 42626
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.6428893208503723,
      "learning_rate": 0.0005507367172608726,
      "loss": 2.4829,
      "step": 42627
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.7049593329429626,
      "learning_rate": 0.0005507344713116145,
      "loss": 2.2,
      "step": 42628
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.7461371421813965,
      "learning_rate": 0.0005507322253157401,
      "loss": 2.395,
      "step": 42629
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.6959307193756104,
      "learning_rate": 0.00055072997927325,
      "loss": 2.1573,
      "step": 42630
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.7429195046424866,
      "learning_rate": 0.0005507277331841442,
      "loss": 2.0086,
      "step": 42631
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.8026281595230103,
      "learning_rate": 0.0005507254870484236,
      "loss": 2.1521,
      "step": 42632
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.7253570556640625,
      "learning_rate": 0.0005507232408660882,
      "loss": 2.2917,
      "step": 42633
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.7642718553543091,
      "learning_rate": 0.0005507209946371388,
      "loss": 2.2368,
      "step": 42634
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.6822502017021179,
      "learning_rate": 0.0005507187483615756,
      "loss": 2.1998,
      "step": 42635
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.6286950707435608,
      "learning_rate": 0.000550716502039399,
      "loss": 2.2413,
      "step": 42636
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.8096708059310913,
      "learning_rate": 0.0005507142556706095,
      "loss": 2.2523,
      "step": 42637
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.6768482327461243,
      "learning_rate": 0.0005507120092552074,
      "loss": 2.3773,
      "step": 42638
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.7145002484321594,
      "learning_rate": 0.0005507097627931933,
      "loss": 1.9782,
      "step": 42639
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.9179280400276184,
      "learning_rate": 0.0005507075162845675,
      "loss": 2.1907,
      "step": 42640
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.6500728130340576,
      "learning_rate": 0.0005507052697293304,
      "loss": 2.096,
      "step": 42641
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.8871018290519714,
      "learning_rate": 0.0005507030231274825,
      "loss": 2.4055,
      "step": 42642
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.7718073129653931,
      "learning_rate": 0.0005507007764790242,
      "loss": 2.2413,
      "step": 42643
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.734831690788269,
      "learning_rate": 0.0005506985297839559,
      "loss": 2.3573,
      "step": 42644
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.9549051523208618,
      "learning_rate": 0.000550696283042278,
      "loss": 2.0537,
      "step": 42645
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.8132162690162659,
      "learning_rate": 0.0005506940362539909,
      "loss": 2.2327,
      "step": 42646
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.6784384846687317,
      "learning_rate": 0.000550691789419095,
      "loss": 2.3096,
      "step": 42647
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.7754884958267212,
      "learning_rate": 0.0005506895425375908,
      "loss": 2.2018,
      "step": 42648
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.8474719524383545,
      "learning_rate": 0.0005506872956094787,
      "loss": 2.1692,
      "step": 42649
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.6869966387748718,
      "learning_rate": 0.0005506850486347592,
      "loss": 2.2345,
      "step": 42650
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.7269156575202942,
      "learning_rate": 0.0005506828016134325,
      "loss": 2.1004,
      "step": 42651
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.9351339340209961,
      "learning_rate": 0.0005506805545454992,
      "loss": 2.3141,
      "step": 42652
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.7823585867881775,
      "learning_rate": 0.0005506783074309596,
      "loss": 2.4299,
      "step": 42653
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.6739144325256348,
      "learning_rate": 0.0005506760602698141,
      "loss": 2.3066,
      "step": 42654
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.70672607421875,
      "learning_rate": 0.0005506738130620633,
      "loss": 2.1653,
      "step": 42655
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.8323193788528442,
      "learning_rate": 0.0005506715658077075,
      "loss": 2.2336,
      "step": 42656
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.750769317150116,
      "learning_rate": 0.0005506693185067471,
      "loss": 2.2094,
      "step": 42657
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.7415888905525208,
      "learning_rate": 0.0005506670711591825,
      "loss": 2.322,
      "step": 42658
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.7957714796066284,
      "learning_rate": 0.0005506648237650143,
      "loss": 2.2527,
      "step": 42659
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.710079550743103,
      "learning_rate": 0.0005506625763242427,
      "loss": 2.0462,
      "step": 42660
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.7203173637390137,
      "learning_rate": 0.000550660328836868,
      "loss": 2.3721,
      "step": 42661
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.7979182600975037,
      "learning_rate": 0.000550658081302891,
      "loss": 2.108,
      "step": 42662
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.7414897084236145,
      "learning_rate": 0.000550655833722312,
      "loss": 2.2563,
      "step": 42663
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.7940480709075928,
      "learning_rate": 0.0005506535860951312,
      "loss": 2.2837,
      "step": 42664
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.6890988349914551,
      "learning_rate": 0.0005506513384213493,
      "loss": 2.1959,
      "step": 42665
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.8972170948982239,
      "learning_rate": 0.0005506490907009665,
      "loss": 2.4117,
      "step": 42666
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.6557784676551819,
      "learning_rate": 0.0005506468429339833,
      "loss": 2.2177,
      "step": 42667
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.6653571724891663,
      "learning_rate": 0.0005506445951204002,
      "loss": 2.4712,
      "step": 42668
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.804058313369751,
      "learning_rate": 0.0005506423472602175,
      "loss": 2.3814,
      "step": 42669
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.7800530791282654,
      "learning_rate": 0.0005506400993534356,
      "loss": 2.2684,
      "step": 42670
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.6605862379074097,
      "learning_rate": 0.000550637851400055,
      "loss": 2.1624,
      "step": 42671
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.9025932550430298,
      "learning_rate": 0.0005506356034000761,
      "loss": 2.5274,
      "step": 42672
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.6944166421890259,
      "learning_rate": 0.0005506333553534993,
      "loss": 2.277,
      "step": 42673
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.6839698553085327,
      "learning_rate": 0.0005506311072603251,
      "loss": 2.1656,
      "step": 42674
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.672111451625824,
      "learning_rate": 0.0005506288591205537,
      "loss": 2.3136,
      "step": 42675
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.697921097278595,
      "learning_rate": 0.0005506266109341859,
      "loss": 2.3302,
      "step": 42676
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.7513649463653564,
      "learning_rate": 0.0005506243627012216,
      "loss": 2.2442,
      "step": 42677
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.6821721196174622,
      "learning_rate": 0.0005506221144216617,
      "loss": 2.0299,
      "step": 42678
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.6895298957824707,
      "learning_rate": 0.0005506198660955063,
      "loss": 2.2341,
      "step": 42679
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.7063762545585632,
      "learning_rate": 0.000550617617722756,
      "loss": 2.3786,
      "step": 42680
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.7257096767425537,
      "learning_rate": 0.0005506153693034112,
      "loss": 2.3056,
      "step": 42681
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.7403335571289062,
      "learning_rate": 0.0005506131208374722,
      "loss": 2.2202,
      "step": 42682
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.6899837851524353,
      "learning_rate": 0.0005506108723249395,
      "loss": 2.2554,
      "step": 42683
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.7385151982307434,
      "learning_rate": 0.0005506086237658135,
      "loss": 2.2189,
      "step": 42684
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.6724459528923035,
      "learning_rate": 0.0005506063751600945,
      "loss": 2.0647,
      "step": 42685
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.636746346950531,
      "learning_rate": 0.0005506041265077833,
      "loss": 2.3337,
      "step": 42686
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.684549868106842,
      "learning_rate": 0.0005506018778088799,
      "loss": 2.3916,
      "step": 42687
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.6773515939712524,
      "learning_rate": 0.0005505996290633849,
      "loss": 2.1834,
      "step": 42688
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.7838991284370422,
      "learning_rate": 0.0005505973802712987,
      "loss": 2.3667,
      "step": 42689
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.7074155807495117,
      "learning_rate": 0.0005505951314326218,
      "loss": 2.4389,
      "step": 42690
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.6236920952796936,
      "learning_rate": 0.0005505928825473546,
      "loss": 2.1013,
      "step": 42691
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.6820928454399109,
      "learning_rate": 0.0005505906336154973,
      "loss": 2.1933,
      "step": 42692
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.7394906878471375,
      "learning_rate": 0.0005505883846370505,
      "loss": 2.3284,
      "step": 42693
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.7104337215423584,
      "learning_rate": 0.0005505861356120145,
      "loss": 2.284,
      "step": 42694
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.7708858847618103,
      "learning_rate": 0.0005505838865403899,
      "loss": 2.5624,
      "step": 42695
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.7719027996063232,
      "learning_rate": 0.000550581637422177,
      "loss": 2.2648,
      "step": 42696
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.6846997737884521,
      "learning_rate": 0.0005505793882573763,
      "loss": 2.2948,
      "step": 42697
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.656517744064331,
      "learning_rate": 0.0005505771390459881,
      "loss": 2.2077,
      "step": 42698
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.6813169717788696,
      "learning_rate": 0.0005505748897880129,
      "loss": 2.2786,
      "step": 42699
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.8104158043861389,
      "learning_rate": 0.000550572640483451,
      "loss": 2.2369,
      "step": 42700
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.6904524564743042,
      "learning_rate": 0.0005505703911323032,
      "loss": 2.2073,
      "step": 42701
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.7149046063423157,
      "learning_rate": 0.0005505681417345694,
      "loss": 2.3863,
      "step": 42702
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.6967563629150391,
      "learning_rate": 0.0005505658922902503,
      "loss": 2.1309,
      "step": 42703
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.6885780692100525,
      "learning_rate": 0.0005505636427993463,
      "loss": 2.1046,
      "step": 42704
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.7248300313949585,
      "learning_rate": 0.0005505613932618577,
      "loss": 2.3235,
      "step": 42705
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.781758725643158,
      "learning_rate": 0.0005505591436777851,
      "loss": 2.3864,
      "step": 42706
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.6761840581893921,
      "learning_rate": 0.0005505568940471289,
      "loss": 2.3372,
      "step": 42707
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.6680927872657776,
      "learning_rate": 0.0005505546443698892,
      "loss": 2.3841,
      "step": 42708
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.6052345037460327,
      "learning_rate": 0.000550552394646067,
      "loss": 2.0527,
      "step": 42709
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.7386445999145508,
      "learning_rate": 0.000550550144875662,
      "loss": 2.3002,
      "step": 42710
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.7848776578903198,
      "learning_rate": 0.0005505478950586753,
      "loss": 2.3053,
      "step": 42711
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.6504341959953308,
      "learning_rate": 0.0005505456451951069,
      "loss": 2.3594,
      "step": 42712
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.7571744918823242,
      "learning_rate": 0.0005505433952849573,
      "loss": 2.417,
      "step": 42713
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.7828201651573181,
      "learning_rate": 0.000550541145328227,
      "loss": 2.4528,
      "step": 42714
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.7250856757164001,
      "learning_rate": 0.0005505388953249164,
      "loss": 2.286,
      "step": 42715
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.7385913729667664,
      "learning_rate": 0.0005505366452750258,
      "loss": 2.3135,
      "step": 42716
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.8009222745895386,
      "learning_rate": 0.0005505343951785558,
      "loss": 2.2403,
      "step": 42717
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.8331223130226135,
      "learning_rate": 0.0005505321450355067,
      "loss": 2.3684,
      "step": 42718
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.67982417345047,
      "learning_rate": 0.0005505298948458788,
      "loss": 2.3285,
      "step": 42719
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.7657414078712463,
      "learning_rate": 0.0005505276446096729,
      "loss": 2.2683,
      "step": 42720
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.6954218149185181,
      "learning_rate": 0.000550525394326889,
      "loss": 2.4183,
      "step": 42721
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.8295726180076599,
      "learning_rate": 0.0005505231439975279,
      "loss": 2.1177,
      "step": 42722
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.7143111824989319,
      "learning_rate": 0.0005505208936215896,
      "loss": 2.2457,
      "step": 42723
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.929349958896637,
      "learning_rate": 0.0005505186431990749,
      "loss": 2.274,
      "step": 42724
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.7189895510673523,
      "learning_rate": 0.0005505163927299839,
      "loss": 2.3207,
      "step": 42725
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.6912923455238342,
      "learning_rate": 0.0005505141422143172,
      "loss": 2.3286,
      "step": 42726
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.6674326658248901,
      "learning_rate": 0.0005505118916520754,
      "loss": 2.1615,
      "step": 42727
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.7905136346817017,
      "learning_rate": 0.0005505096410432584,
      "loss": 2.4621,
      "step": 42728
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.7442195415496826,
      "learning_rate": 0.0005505073903878672,
      "loss": 2.202,
      "step": 42729
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.6207206845283508,
      "learning_rate": 0.0005505051396859019,
      "loss": 2.139,
      "step": 42730
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.7129926085472107,
      "learning_rate": 0.0005505028889373629,
      "loss": 2.3292,
      "step": 42731
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.7207767963409424,
      "learning_rate": 0.0005505006381422506,
      "loss": 2.3087,
      "step": 42732
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.7287100553512573,
      "learning_rate": 0.0005504983873005657,
      "loss": 2.2199,
      "step": 42733
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.6670235991477966,
      "learning_rate": 0.0005504961364123082,
      "loss": 2.3456,
      "step": 42734
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.7578818202018738,
      "learning_rate": 0.0005504938854774789,
      "loss": 2.4648,
      "step": 42735
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.7029481530189514,
      "learning_rate": 0.000550491634496078,
      "loss": 2.1491,
      "step": 42736
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.6793972849845886,
      "learning_rate": 0.000550489383468106,
      "loss": 2.0518,
      "step": 42737
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.7636272311210632,
      "learning_rate": 0.0005504871323935633,
      "loss": 2.4891,
      "step": 42738
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.7822688221931458,
      "learning_rate": 0.0005504848812724504,
      "loss": 2.3371,
      "step": 42739
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.6483932137489319,
      "learning_rate": 0.0005504826301047673,
      "loss": 2.1814,
      "step": 42740
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.683711588382721,
      "learning_rate": 0.0005504803788905152,
      "loss": 2.1712,
      "step": 42741
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.7212444543838501,
      "learning_rate": 0.0005504781276296938,
      "loss": 2.1477,
      "step": 42742
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.6647822856903076,
      "learning_rate": 0.0005504758763223038,
      "loss": 2.1143,
      "step": 42743
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.6931868195533752,
      "learning_rate": 0.0005504736249683456,
      "loss": 2.187,
      "step": 42744
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.7224649786949158,
      "learning_rate": 0.0005504713735678198,
      "loss": 2.4854,
      "step": 42745
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.69191575050354,
      "learning_rate": 0.0005504691221207264,
      "loss": 2.0535,
      "step": 42746
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.6461462378501892,
      "learning_rate": 0.0005504668706270662,
      "loss": 2.3056,
      "step": 42747
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.7031391859054565,
      "learning_rate": 0.0005504646190868395,
      "loss": 2.2543,
      "step": 42748
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.7372772097587585,
      "learning_rate": 0.0005504623675000466,
      "loss": 2.3322,
      "step": 42749
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.7108338475227356,
      "learning_rate": 0.0005504601158666881,
      "loss": 2.3484,
      "step": 42750
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.6316297054290771,
      "learning_rate": 0.0005504578641867643,
      "loss": 2.2876,
      "step": 42751
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.7729988098144531,
      "learning_rate": 0.0005504556124602758,
      "loss": 2.2825,
      "step": 42752
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.9219883680343628,
      "learning_rate": 0.0005504533606872226,
      "loss": 2.2693,
      "step": 42753
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.6686140894889832,
      "learning_rate": 0.0005504511088676057,
      "loss": 2.2619,
      "step": 42754
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.7802931666374207,
      "learning_rate": 0.0005504488570014249,
      "loss": 2.3727,
      "step": 42755
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.7080293893814087,
      "learning_rate": 0.0005504466050886811,
      "loss": 2.2448,
      "step": 42756
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.6776398420333862,
      "learning_rate": 0.0005504443531293746,
      "loss": 2.2732,
      "step": 42757
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.7176530957221985,
      "learning_rate": 0.0005504421011235056,
      "loss": 2.3207,
      "step": 42758
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.6465449929237366,
      "learning_rate": 0.0005504398490710747,
      "loss": 2.2998,
      "step": 42759
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.6726177930831909,
      "learning_rate": 0.0005504375969720825,
      "loss": 2.3795,
      "step": 42760
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.7420480847358704,
      "learning_rate": 0.0005504353448265291,
      "loss": 2.3846,
      "step": 42761
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.7098219990730286,
      "learning_rate": 0.000550433092634415,
      "loss": 2.3427,
      "step": 42762
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.8168237805366516,
      "learning_rate": 0.0005504308403957407,
      "loss": 2.1531,
      "step": 42763
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.6700173020362854,
      "learning_rate": 0.0005504285881105065,
      "loss": 2.013,
      "step": 42764
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.698127806186676,
      "learning_rate": 0.0005504263357787131,
      "loss": 2.3563,
      "step": 42765
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.6940244436264038,
      "learning_rate": 0.0005504240834003606,
      "loss": 2.0831,
      "step": 42766
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.8014993071556091,
      "learning_rate": 0.0005504218309754495,
      "loss": 2.3389,
      "step": 42767
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.6989393830299377,
      "learning_rate": 0.0005504195785039803,
      "loss": 2.3131,
      "step": 42768
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.6909437775611877,
      "learning_rate": 0.0005504173259859534,
      "loss": 2.3718,
      "step": 42769
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.6845472455024719,
      "learning_rate": 0.0005504150734213692,
      "loss": 2.2766,
      "step": 42770
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.6545066833496094,
      "learning_rate": 0.0005504128208102281,
      "loss": 2.0985,
      "step": 42771
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.7086923122406006,
      "learning_rate": 0.0005504105681525303,
      "loss": 2.364,
      "step": 42772
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.7747541069984436,
      "learning_rate": 0.0005504083154482767,
      "loss": 2.3576,
      "step": 42773
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.761186957359314,
      "learning_rate": 0.0005504060626974675,
      "loss": 2.3727,
      "step": 42774
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.6181201934814453,
      "learning_rate": 0.0005504038099001029,
      "loss": 2.0517,
      "step": 42775
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.9645010232925415,
      "learning_rate": 0.0005504015570561837,
      "loss": 2.298,
      "step": 42776
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.6979835033416748,
      "learning_rate": 0.00055039930416571,
      "loss": 2.0732,
      "step": 42777
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.8282484412193298,
      "learning_rate": 0.0005503970512286823,
      "loss": 2.3532,
      "step": 42778
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.7688621878623962,
      "learning_rate": 0.0005503947982451013,
      "loss": 2.2222,
      "step": 42779
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.739752471446991,
      "learning_rate": 0.000550392545214967,
      "loss": 2.2879,
      "step": 42780
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.698512852191925,
      "learning_rate": 0.00055039029213828,
      "loss": 2.2971,
      "step": 42781
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.7969578504562378,
      "learning_rate": 0.0005503880390150406,
      "loss": 2.2044,
      "step": 42782
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.7204169631004333,
      "learning_rate": 0.0005503857858452496,
      "loss": 2.327,
      "step": 42783
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.6907174587249756,
      "learning_rate": 0.0005503835326289069,
      "loss": 2.3557,
      "step": 42784
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.7266966700553894,
      "learning_rate": 0.0005503812793660134,
      "loss": 2.5033,
      "step": 42785
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.7104089856147766,
      "learning_rate": 0.0005503790260565692,
      "loss": 2.3219,
      "step": 42786
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.8089837431907654,
      "learning_rate": 0.0005503767727005747,
      "loss": 2.2238,
      "step": 42787
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.6768142580986023,
      "learning_rate": 0.0005503745192980306,
      "loss": 2.1106,
      "step": 42788
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.7370222806930542,
      "learning_rate": 0.0005503722658489371,
      "loss": 2.1842,
      "step": 42789
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.8511245846748352,
      "learning_rate": 0.0005503700123532947,
      "loss": 2.4932,
      "step": 42790
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.7148282527923584,
      "learning_rate": 0.0005503677588111036,
      "loss": 2.3012,
      "step": 42791
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.6739543676376343,
      "learning_rate": 0.0005503655052223646,
      "loss": 2.3484,
      "step": 42792
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.8086404204368591,
      "learning_rate": 0.000550363251587078,
      "loss": 2.5219,
      "step": 42793
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.6894761919975281,
      "learning_rate": 0.000550360997905244,
      "loss": 2.3394,
      "step": 42794
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.6393022537231445,
      "learning_rate": 0.0005503587441768632,
      "loss": 2.1904,
      "step": 42795
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.6830849051475525,
      "learning_rate": 0.000550356490401936,
      "loss": 2.3141,
      "step": 42796
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.6847425699234009,
      "learning_rate": 0.0005503542365804628,
      "loss": 2.3951,
      "step": 42797
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.6899234652519226,
      "learning_rate": 0.0005503519827124439,
      "loss": 2.0957,
      "step": 42798
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.675247311592102,
      "learning_rate": 0.0005503497287978799,
      "loss": 2.2013,
      "step": 42799
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.7629833221435547,
      "learning_rate": 0.0005503474748367712,
      "loss": 2.3485,
      "step": 42800
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.6931436061859131,
      "learning_rate": 0.0005503452208291182,
      "loss": 2.2816,
      "step": 42801
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.6781894564628601,
      "learning_rate": 0.0005503429667749213,
      "loss": 2.3035,
      "step": 42802
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.6747809648513794,
      "learning_rate": 0.0005503407126741809,
      "loss": 2.4675,
      "step": 42803
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.7044435143470764,
      "learning_rate": 0.0005503384585268975,
      "loss": 2.2019,
      "step": 42804
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.634425699710846,
      "learning_rate": 0.0005503362043330713,
      "loss": 2.5568,
      "step": 42805
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.6951317191123962,
      "learning_rate": 0.0005503339500927029,
      "loss": 2.2078,
      "step": 42806
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.7205966114997864,
      "learning_rate": 0.0005503316958057928,
      "loss": 2.2231,
      "step": 42807
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.8027887344360352,
      "learning_rate": 0.0005503294414723412,
      "loss": 2.2853,
      "step": 42808
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.7055739164352417,
      "learning_rate": 0.0005503271870923489,
      "loss": 2.0904,
      "step": 42809
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.7278452515602112,
      "learning_rate": 0.0005503249326658158,
      "loss": 2.3426,
      "step": 42810
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.7066052556037903,
      "learning_rate": 0.0005503226781927425,
      "loss": 2.2833,
      "step": 42811
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.7404667139053345,
      "learning_rate": 0.0005503204236731296,
      "loss": 2.2193,
      "step": 42812
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.7623347640037537,
      "learning_rate": 0.0005503181691069774,
      "loss": 1.9295,
      "step": 42813
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.8405235409736633,
      "learning_rate": 0.0005503159144942864,
      "loss": 2.2681,
      "step": 42814
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.703630805015564,
      "learning_rate": 0.0005503136598350569,
      "loss": 2.0806,
      "step": 42815
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.7696812152862549,
      "learning_rate": 0.0005503114051292893,
      "loss": 2.133,
      "step": 42816
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.865803599357605,
      "learning_rate": 0.0005503091503769841,
      "loss": 2.1957,
      "step": 42817
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.7235902547836304,
      "learning_rate": 0.0005503068955781418,
      "loss": 2.1795,
      "step": 42818
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.7886182069778442,
      "learning_rate": 0.0005503046407327627,
      "loss": 2.3025,
      "step": 42819
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.7215784192085266,
      "learning_rate": 0.0005503023858408471,
      "loss": 2.1891,
      "step": 42820
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.7072517275810242,
      "learning_rate": 0.0005503001309023958,
      "loss": 2.3105,
      "step": 42821
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.7652894854545593,
      "learning_rate": 0.0005502978759174089,
      "loss": 2.2388,
      "step": 42822
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.7161871194839478,
      "learning_rate": 0.0005502956208858868,
      "loss": 2.4815,
      "step": 42823
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.6605458855628967,
      "learning_rate": 0.00055029336580783,
      "loss": 2.1779,
      "step": 42824
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.7417550086975098,
      "learning_rate": 0.000550291110683239,
      "loss": 2.4076,
      "step": 42825
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.7288651466369629,
      "learning_rate": 0.0005502888555121142,
      "loss": 2.1901,
      "step": 42826
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.7826097011566162,
      "learning_rate": 0.0005502866002944558,
      "loss": 2.2777,
      "step": 42827
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.6405069828033447,
      "learning_rate": 0.0005502843450302646,
      "loss": 2.2525,
      "step": 42828
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.6803017258644104,
      "learning_rate": 0.0005502820897195409,
      "loss": 2.4281,
      "step": 42829
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.6952850818634033,
      "learning_rate": 0.0005502798343622849,
      "loss": 2.2557,
      "step": 42830
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.8740202188491821,
      "learning_rate": 0.0005502775789584971,
      "loss": 2.3475,
      "step": 42831
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.7303866744041443,
      "learning_rate": 0.000550275323508178,
      "loss": 1.908,
      "step": 42832
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.6714662313461304,
      "learning_rate": 0.000550273068011328,
      "loss": 2.4202,
      "step": 42833
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.6456208825111389,
      "learning_rate": 0.0005502708124679475,
      "loss": 2.2927,
      "step": 42834
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.6799165606498718,
      "learning_rate": 0.000550268556878037,
      "loss": 2.3478,
      "step": 42835
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.6741200089454651,
      "learning_rate": 0.0005502663012415968,
      "loss": 2.3051,
      "step": 42836
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.6940663456916809,
      "learning_rate": 0.0005502640455586274,
      "loss": 2.1094,
      "step": 42837
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.7709901332855225,
      "learning_rate": 0.0005502617898291291,
      "loss": 2.5388,
      "step": 42838
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.7014974355697632,
      "learning_rate": 0.0005502595340531025,
      "loss": 2.3965,
      "step": 42839
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.791553258895874,
      "learning_rate": 0.0005502572782305479,
      "loss": 2.3844,
      "step": 42840
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.7052765488624573,
      "learning_rate": 0.0005502550223614658,
      "loss": 2.2803,
      "step": 42841
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.7054828405380249,
      "learning_rate": 0.0005502527664458564,
      "loss": 2.1103,
      "step": 42842
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.7238202691078186,
      "learning_rate": 0.0005502505104837205,
      "loss": 2.2903,
      "step": 42843
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.7111126184463501,
      "learning_rate": 0.0005502482544750581,
      "loss": 2.1656,
      "step": 42844
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.6854608654975891,
      "learning_rate": 0.00055024599841987,
      "loss": 2.0791,
      "step": 42845
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.6868157982826233,
      "learning_rate": 0.0005502437423181564,
      "loss": 2.2814,
      "step": 42846
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.745693027973175,
      "learning_rate": 0.0005502414861699177,
      "loss": 2.0547,
      "step": 42847
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.6894679665565491,
      "learning_rate": 0.0005502392299751545,
      "loss": 2.1326,
      "step": 42848
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.0536497831344604,
      "learning_rate": 0.000550236973733867,
      "loss": 2.3314,
      "step": 42849
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.6506667137145996,
      "learning_rate": 0.0005502347174460557,
      "loss": 2.1837,
      "step": 42850
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.7881971597671509,
      "learning_rate": 0.0005502324611117212,
      "loss": 2.3316,
      "step": 42851
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.9032886624336243,
      "learning_rate": 0.0005502302047308636,
      "loss": 2.1763,
      "step": 42852
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.6640021204948425,
      "learning_rate": 0.0005502279483034836,
      "loss": 2.3225,
      "step": 42853
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.6912857890129089,
      "learning_rate": 0.0005502256918295815,
      "loss": 2.2549,
      "step": 42854
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.7320406436920166,
      "learning_rate": 0.0005502234353091576,
      "loss": 2.1644,
      "step": 42855
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.7618736028671265,
      "learning_rate": 0.0005502211787422125,
      "loss": 2.3722,
      "step": 42856
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.7261292934417725,
      "learning_rate": 0.0005502189221287465,
      "loss": 2.2532,
      "step": 42857
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.6656633019447327,
      "learning_rate": 0.0005502166654687601,
      "loss": 2.2313,
      "step": 42858
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.7458394765853882,
      "learning_rate": 0.0005502144087622538,
      "loss": 2.0894,
      "step": 42859
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.7368835806846619,
      "learning_rate": 0.0005502121520092279,
      "loss": 2.291,
      "step": 42860
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.6590776443481445,
      "learning_rate": 0.0005502098952096828,
      "loss": 2.237,
      "step": 42861
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.6657084822654724,
      "learning_rate": 0.000550207638363619,
      "loss": 1.932,
      "step": 42862
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.8189603090286255,
      "learning_rate": 0.0005502053814710367,
      "loss": 2.3187,
      "step": 42863
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.6792385578155518,
      "learning_rate": 0.0005502031245319367,
      "loss": 2.0188,
      "step": 42864
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.7207566499710083,
      "learning_rate": 0.0005502008675463191,
      "loss": 2.3354,
      "step": 42865
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.7056212425231934,
      "learning_rate": 0.0005501986105141846,
      "loss": 2.3681,
      "step": 42866
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.7919942140579224,
      "learning_rate": 0.0005501963534355333,
      "loss": 2.53,
      "step": 42867
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.7110952734947205,
      "learning_rate": 0.0005501940963103658,
      "loss": 2.2162,
      "step": 42868
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.7342767715454102,
      "learning_rate": 0.0005501918391386825,
      "loss": 2.1883,
      "step": 42869
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.7802417874336243,
      "learning_rate": 0.0005501895819204839,
      "loss": 2.1816,
      "step": 42870
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.680800199508667,
      "learning_rate": 0.0005501873246557702,
      "loss": 2.1167,
      "step": 42871
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.6935215592384338,
      "learning_rate": 0.0005501850673445421,
      "loss": 2.5016,
      "step": 42872
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.7263532876968384,
      "learning_rate": 0.0005501828099867997,
      "loss": 2.1826,
      "step": 42873
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.0813544988632202,
      "learning_rate": 0.0005501805525825437,
      "loss": 2.2797,
      "step": 42874
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.7493624687194824,
      "learning_rate": 0.0005501782951317743,
      "loss": 2.203,
      "step": 42875
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.7191603779792786,
      "learning_rate": 0.0005501760376344922,
      "loss": 2.2825,
      "step": 42876
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.6396224498748779,
      "learning_rate": 0.0005501737800906976,
      "loss": 2.2077,
      "step": 42877
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.7934576869010925,
      "learning_rate": 0.0005501715225003908,
      "loss": 2.1615,
      "step": 42878
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.6616617441177368,
      "learning_rate": 0.0005501692648635727,
      "loss": 2.4597,
      "step": 42879
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.682305634021759,
      "learning_rate": 0.0005501670071802432,
      "loss": 2.5784,
      "step": 42880
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.7380767464637756,
      "learning_rate": 0.0005501647494504029,
      "loss": 2.0067,
      "step": 42881
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.6413140892982483,
      "learning_rate": 0.0005501624916740524,
      "loss": 2.2124,
      "step": 42882
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.6840128898620605,
      "learning_rate": 0.0005501602338511918,
      "loss": 2.3523,
      "step": 42883
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.7743709087371826,
      "learning_rate": 0.0005501579759818218,
      "loss": 2.4593,
      "step": 42884
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.7484415173530579,
      "learning_rate": 0.0005501557180659427,
      "loss": 2.2262,
      "step": 42885
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.6697841286659241,
      "learning_rate": 0.000550153460103555,
      "loss": 2.2465,
      "step": 42886
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.782191276550293,
      "learning_rate": 0.000550151202094659,
      "loss": 2.3808,
      "step": 42887
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.8386456966400146,
      "learning_rate": 0.0005501489440392552,
      "loss": 2.2183,
      "step": 42888
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.7868819236755371,
      "learning_rate": 0.0005501466859373439,
      "loss": 2.051,
      "step": 42889
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.7030195593833923,
      "learning_rate": 0.0005501444277889256,
      "loss": 2.3911,
      "step": 42890
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.0087367296218872,
      "learning_rate": 0.0005501421695940008,
      "loss": 2.0763,
      "step": 42891
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.9114407300949097,
      "learning_rate": 0.0005501399113525699,
      "loss": 2.3,
      "step": 42892
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.7045570611953735,
      "learning_rate": 0.0005501376530646331,
      "loss": 2.1921,
      "step": 42893
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.9162128567695618,
      "learning_rate": 0.0005501353947301911,
      "loss": 2.229,
      "step": 42894
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.7173637747764587,
      "learning_rate": 0.0005501331363492442,
      "loss": 2.2962,
      "step": 42895
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.7206664681434631,
      "learning_rate": 0.0005501308779217928,
      "loss": 2.3605,
      "step": 42896
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.6774587631225586,
      "learning_rate": 0.0005501286194478374,
      "loss": 2.4145,
      "step": 42897
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.8348682522773743,
      "learning_rate": 0.0005501263609273783,
      "loss": 2.1424,
      "step": 42898
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.7383179664611816,
      "learning_rate": 0.000550124102360416,
      "loss": 2.3468,
      "step": 42899
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.6675275564193726,
      "learning_rate": 0.000550121843746951,
      "loss": 2.332,
      "step": 42900
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.7439825534820557,
      "learning_rate": 0.0005501195850869836,
      "loss": 2.3356,
      "step": 42901
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.6814992427825928,
      "learning_rate": 0.0005501173263805142,
      "loss": 2.1523,
      "step": 42902
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.7218989133834839,
      "learning_rate": 0.0005501150676275432,
      "loss": 2.3048,
      "step": 42903
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.6774598956108093,
      "learning_rate": 0.0005501128088280712,
      "loss": 2.0866,
      "step": 42904
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.7372496128082275,
      "learning_rate": 0.0005501105499820984,
      "loss": 2.1826,
      "step": 42905
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.7045762538909912,
      "learning_rate": 0.0005501082910896255,
      "loss": 2.3532,
      "step": 42906
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.6381622552871704,
      "learning_rate": 0.0005501060321506527,
      "loss": 2.2974,
      "step": 42907
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.6782696843147278,
      "learning_rate": 0.0005501037731651803,
      "loss": 2.184,
      "step": 42908
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.7955156564712524,
      "learning_rate": 0.0005501015141332091,
      "loss": 2.4201,
      "step": 42909
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.8258759379386902,
      "learning_rate": 0.0005500992550547391,
      "loss": 2.5107,
      "step": 42910
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.8416896462440491,
      "learning_rate": 0.0005500969959297712,
      "loss": 2.2977,
      "step": 42911
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.7605114579200745,
      "learning_rate": 0.0005500947367583053,
      "loss": 2.092,
      "step": 42912
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.6484983563423157,
      "learning_rate": 0.0005500924775403421,
      "loss": 2.0763,
      "step": 42913
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.7335663437843323,
      "learning_rate": 0.0005500902182758821,
      "loss": 2.2304,
      "step": 42914
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.709256112575531,
      "learning_rate": 0.0005500879589649255,
      "loss": 2.2963,
      "step": 42915
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.6779664754867554,
      "learning_rate": 0.0005500856996074729,
      "loss": 2.1812,
      "step": 42916
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.6620600819587708,
      "learning_rate": 0.0005500834402035246,
      "loss": 2.5318,
      "step": 42917
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.6872848868370056,
      "learning_rate": 0.0005500811807530811,
      "loss": 2.1751,
      "step": 42918
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.6866400837898254,
      "learning_rate": 0.0005500789212561428,
      "loss": 2.1078,
      "step": 42919
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.6897681951522827,
      "learning_rate": 0.00055007666171271,
      "loss": 2.4733,
      "step": 42920
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.7548987865447998,
      "learning_rate": 0.0005500744021227834,
      "loss": 2.4721,
      "step": 42921
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.7311016917228699,
      "learning_rate": 0.0005500721424863632,
      "loss": 2.1984,
      "step": 42922
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.657608151435852,
      "learning_rate": 0.0005500698828034498,
      "loss": 2.1469,
      "step": 42923
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.7338316440582275,
      "learning_rate": 0.0005500676230740438,
      "loss": 2.1859,
      "step": 42924
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.7047488689422607,
      "learning_rate": 0.0005500653632981453,
      "loss": 2.3334,
      "step": 42925
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.7173285484313965,
      "learning_rate": 0.000550063103475755,
      "loss": 2.3611,
      "step": 42926
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.6690166592597961,
      "learning_rate": 0.0005500608436068734,
      "loss": 2.1341,
      "step": 42927
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.7117436528205872,
      "learning_rate": 0.0005500585836915007,
      "loss": 2.371,
      "step": 42928
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.6827138662338257,
      "learning_rate": 0.0005500563237296373,
      "loss": 2.159,
      "step": 42929
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.8321431875228882,
      "learning_rate": 0.0005500540637212838,
      "loss": 2.1894,
      "step": 42930
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.6604613661766052,
      "learning_rate": 0.0005500518036664405,
      "loss": 2.2091,
      "step": 42931
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.8150764107704163,
      "learning_rate": 0.0005500495435651078,
      "loss": 2.2731,
      "step": 42932
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.68017578125,
      "learning_rate": 0.0005500472834172863,
      "loss": 2.1142,
      "step": 42933
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.752236545085907,
      "learning_rate": 0.0005500450232229763,
      "loss": 2.3901,
      "step": 42934
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.6439617276191711,
      "learning_rate": 0.000550042762982178,
      "loss": 2.2905,
      "step": 42935
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.7580307126045227,
      "learning_rate": 0.0005500405026948921,
      "loss": 2.4131,
      "step": 42936
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.728755533695221,
      "learning_rate": 0.0005500382423611191,
      "loss": 2.3213,
      "step": 42937
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.7414531111717224,
      "learning_rate": 0.0005500359819808591,
      "loss": 2.4751,
      "step": 42938
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.6618906855583191,
      "learning_rate": 0.0005500337215541128,
      "loss": 2.4328,
      "step": 42939
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.7923210859298706,
      "learning_rate": 0.0005500314610808805,
      "loss": 2.193,
      "step": 42940
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.7629625797271729,
      "learning_rate": 0.0005500292005611626,
      "loss": 2.257,
      "step": 42941
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.657648503780365,
      "learning_rate": 0.0005500269399949595,
      "loss": 2.3682,
      "step": 42942
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.8305273056030273,
      "learning_rate": 0.0005500246793822718,
      "loss": 2.5223,
      "step": 42943
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.727520227432251,
      "learning_rate": 0.0005500224187230997,
      "loss": 2.2862,
      "step": 42944
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.6902729272842407,
      "learning_rate": 0.0005500201580174438,
      "loss": 2.347,
      "step": 42945
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.8244689106941223,
      "learning_rate": 0.0005500178972653043,
      "loss": 2.1397,
      "step": 42946
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.6619865298271179,
      "learning_rate": 0.0005500156364666818,
      "loss": 2.2736,
      "step": 42947
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.7713747620582581,
      "learning_rate": 0.0005500133756215767,
      "loss": 2.2804,
      "step": 42948
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.7199332118034363,
      "learning_rate": 0.0005500111147299895,
      "loss": 2.2632,
      "step": 42949
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.7922878265380859,
      "learning_rate": 0.0005500088537919202,
      "loss": 2.2229,
      "step": 42950
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.6820435523986816,
      "learning_rate": 0.0005500065928073699,
      "loss": 2.3228,
      "step": 42951
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.7249335050582886,
      "learning_rate": 0.0005500043317763384,
      "loss": 2.2528,
      "step": 42952
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.7712740302085876,
      "learning_rate": 0.0005500020706988265,
      "loss": 2.3525,
      "step": 42953
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.6683923602104187,
      "learning_rate": 0.0005499998095748345,
      "loss": 2.3811,
      "step": 42954
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.6947718262672424,
      "learning_rate": 0.0005499975484043628,
      "loss": 2.2272,
      "step": 42955
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.7104913592338562,
      "learning_rate": 0.0005499952871874118,
      "loss": 2.3565,
      "step": 42956
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.7376508712768555,
      "learning_rate": 0.000549993025923982,
      "loss": 2.4096,
      "step": 42957
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.6990509033203125,
      "learning_rate": 0.0005499907646140738,
      "loss": 2.3544,
      "step": 42958
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.8937097191810608,
      "learning_rate": 0.0005499885032576874,
      "loss": 2.3994,
      "step": 42959
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.7304819822311401,
      "learning_rate": 0.0005499862418548236,
      "loss": 2.0617,
      "step": 42960
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.7588337063789368,
      "learning_rate": 0.0005499839804054827,
      "loss": 2.352,
      "step": 42961
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.8510565757751465,
      "learning_rate": 0.0005499817189096649,
      "loss": 2.2836,
      "step": 42962
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.7005515098571777,
      "learning_rate": 0.0005499794573673709,
      "loss": 2.1778,
      "step": 42963
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.7962385416030884,
      "learning_rate": 0.0005499771957786009,
      "loss": 2.2059,
      "step": 42964
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.7770209312438965,
      "learning_rate": 0.0005499749341433555,
      "loss": 2.4248,
      "step": 42965
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.8340985178947449,
      "learning_rate": 0.000549972672461635,
      "loss": 2.1707,
      "step": 42966
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.7182008028030396,
      "learning_rate": 0.0005499704107334399,
      "loss": 2.1552,
      "step": 42967
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.8372421860694885,
      "learning_rate": 0.0005499681489587705,
      "loss": 2.2935,
      "step": 42968
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.7798536419868469,
      "learning_rate": 0.0005499658871376274,
      "loss": 2.3196,
      "step": 42969
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.8710371851921082,
      "learning_rate": 0.0005499636252700109,
      "loss": 2.1518,
      "step": 42970
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.727460503578186,
      "learning_rate": 0.0005499613633559214,
      "loss": 2.3462,
      "step": 42971
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.7833576798439026,
      "learning_rate": 0.0005499591013953595,
      "loss": 2.3212,
      "step": 42972
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.7770208120346069,
      "learning_rate": 0.0005499568393883252,
      "loss": 2.1696,
      "step": 42973
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.6462256908416748,
      "learning_rate": 0.0005499545773348195,
      "loss": 2.301,
      "step": 42974
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.7358672618865967,
      "learning_rate": 0.0005499523152348423,
      "loss": 2.2829,
      "step": 42975
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.7197186946868896,
      "learning_rate": 0.0005499500530883944,
      "loss": 2.5525,
      "step": 42976
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.8963853716850281,
      "learning_rate": 0.000549947790895476,
      "loss": 2.2065,
      "step": 42977
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.6802173256874084,
      "learning_rate": 0.0005499455286560876,
      "loss": 2.1075,
      "step": 42978
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.7456079721450806,
      "learning_rate": 0.0005499432663702296,
      "loss": 2.4112,
      "step": 42979
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.7537379860877991,
      "learning_rate": 0.0005499410040379025,
      "loss": 2.3106,
      "step": 42980
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.6955567002296448,
      "learning_rate": 0.0005499387416591066,
      "loss": 2.2482,
      "step": 42981
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.7076419591903687,
      "learning_rate": 0.0005499364792338423,
      "loss": 2.1486,
      "step": 42982
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.6540254354476929,
      "learning_rate": 0.0005499342167621102,
      "loss": 2.0905,
      "step": 42983
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.6984699368476868,
      "learning_rate": 0.0005499319542439105,
      "loss": 2.2637,
      "step": 42984
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.7446070313453674,
      "learning_rate": 0.0005499296916792438,
      "loss": 2.1938,
      "step": 42985
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.6487333178520203,
      "learning_rate": 0.0005499274290681104,
      "loss": 2.2556,
      "step": 42986
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.6810197234153748,
      "learning_rate": 0.0005499251664105109,
      "loss": 2.3364,
      "step": 42987
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.6987234950065613,
      "learning_rate": 0.0005499229037064455,
      "loss": 2.4478,
      "step": 42988
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.675534188747406,
      "learning_rate": 0.0005499206409559147,
      "loss": 2.5796,
      "step": 42989
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.6485956311225891,
      "learning_rate": 0.0005499183781589189,
      "loss": 2.3146,
      "step": 42990
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.6357097625732422,
      "learning_rate": 0.0005499161153154587,
      "loss": 2.3841,
      "step": 42991
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.6436858773231506,
      "learning_rate": 0.0005499138524255343,
      "loss": 2.3365,
      "step": 42992
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.618667721748352,
      "learning_rate": 0.0005499115894891461,
      "loss": 2.1882,
      "step": 42993
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.918553352355957,
      "learning_rate": 0.0005499093265062948,
      "loss": 2.1574,
      "step": 42994
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.7553752064704895,
      "learning_rate": 0.0005499070634769805,
      "loss": 2.2533,
      "step": 42995
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.6833319664001465,
      "learning_rate": 0.0005499048004012039,
      "loss": 2.1805,
      "step": 42996
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.7164521813392639,
      "learning_rate": 0.0005499025372789652,
      "loss": 2.3343,
      "step": 42997
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.7291693091392517,
      "learning_rate": 0.0005499002741102648,
      "loss": 2.4512,
      "step": 42998
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.9023930430412292,
      "learning_rate": 0.0005498980108951033,
      "loss": 2.2618,
      "step": 42999
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.690519392490387,
      "learning_rate": 0.0005498957476334811,
      "loss": 2.2875,
      "step": 43000
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.7464544177055359,
      "learning_rate": 0.0005498934843253986,
      "loss": 2.2623,
      "step": 43001
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.7694873213768005,
      "learning_rate": 0.000549891220970856,
      "loss": 2.3768,
      "step": 43002
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.7387239336967468,
      "learning_rate": 0.000549888957569854,
      "loss": 2.191,
      "step": 43003
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.731837809085846,
      "learning_rate": 0.0005498866941223929,
      "loss": 2.2425,
      "step": 43004
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.711563229560852,
      "learning_rate": 0.0005498844306284733,
      "loss": 2.1833,
      "step": 43005
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.7842252254486084,
      "learning_rate": 0.0005498821670880953,
      "loss": 2.0298,
      "step": 43006
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.8207446932792664,
      "learning_rate": 0.0005498799035012596,
      "loss": 2.2453,
      "step": 43007
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.6595190763473511,
      "learning_rate": 0.0005498776398679664,
      "loss": 2.1967,
      "step": 43008
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.759406566619873,
      "learning_rate": 0.0005498753761882163,
      "loss": 2.1723,
      "step": 43009
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.7243785858154297,
      "learning_rate": 0.0005498731124620096,
      "loss": 2.0279,
      "step": 43010
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.7262460589408875,
      "learning_rate": 0.0005498708486893468,
      "loss": 2.3607,
      "step": 43011
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.6909098625183105,
      "learning_rate": 0.0005498685848702283,
      "loss": 2.2782,
      "step": 43012
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.6640931963920593,
      "learning_rate": 0.0005498663210046545,
      "loss": 2.3613,
      "step": 43013
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.7985612750053406,
      "learning_rate": 0.0005498640570926258,
      "loss": 2.0662,
      "step": 43014
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.6424942016601562,
      "learning_rate": 0.0005498617931341427,
      "loss": 2.144,
      "step": 43015
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.6762004494667053,
      "learning_rate": 0.0005498595291292056,
      "loss": 2.345,
      "step": 43016
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.6683256030082703,
      "learning_rate": 0.0005498572650778149,
      "loss": 1.9735,
      "step": 43017
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.685515820980072,
      "learning_rate": 0.0005498550009799709,
      "loss": 2.387,
      "step": 43018
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.677422821521759,
      "learning_rate": 0.0005498527368356743,
      "loss": 2.3838,
      "step": 43019
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.7089558839797974,
      "learning_rate": 0.0005498504726449253,
      "loss": 2.2962,
      "step": 43020
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.7164463400840759,
      "learning_rate": 0.0005498482084077244,
      "loss": 2.3618,
      "step": 43021
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.7072516083717346,
      "learning_rate": 0.0005498459441240719,
      "loss": 2.2313,
      "step": 43022
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.6276382803916931,
      "learning_rate": 0.0005498436797939684,
      "loss": 2.2965,
      "step": 43023
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.738639771938324,
      "learning_rate": 0.0005498414154174142,
      "loss": 2.378,
      "step": 43024
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.7323949337005615,
      "learning_rate": 0.0005498391509944098,
      "loss": 2.1745,
      "step": 43025
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.6853955388069153,
      "learning_rate": 0.0005498368865249556,
      "loss": 2.5242,
      "step": 43026
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.7419643998146057,
      "learning_rate": 0.0005498346220090519,
      "loss": 2.296,
      "step": 43027
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.6840011477470398,
      "learning_rate": 0.0005498323574466994,
      "loss": 2.4292,
      "step": 43028
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.8250702023506165,
      "learning_rate": 0.0005498300928378982,
      "loss": 2.3067,
      "step": 43029
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.6497909426689148,
      "learning_rate": 0.000549827828182649,
      "loss": 2.3068,
      "step": 43030
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.6577886939048767,
      "learning_rate": 0.000549825563480952,
      "loss": 2.0393,
      "step": 43031
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.7612890601158142,
      "learning_rate": 0.0005498232987328078,
      "loss": 2.1372,
      "step": 43032
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.7368078827857971,
      "learning_rate": 0.0005498210339382166,
      "loss": 2.2023,
      "step": 43033
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.7103472948074341,
      "learning_rate": 0.000549818769097179,
      "loss": 2.3582,
      "step": 43034
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.6860312223434448,
      "learning_rate": 0.0005498165042096954,
      "loss": 2.1308,
      "step": 43035
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.7171280980110168,
      "learning_rate": 0.0005498142392757662,
      "loss": 2.2928,
      "step": 43036
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.6744970083236694,
      "learning_rate": 0.0005498119742953919,
      "loss": 2.2694,
      "step": 43037
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.68299400806427,
      "learning_rate": 0.0005498097092685728,
      "loss": 2.3104,
      "step": 43038
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.7258290648460388,
      "learning_rate": 0.0005498074441953093,
      "loss": 2.1937,
      "step": 43039
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.7130995392799377,
      "learning_rate": 0.0005498051790756019,
      "loss": 2.1719,
      "step": 43040
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.6675199270248413,
      "learning_rate": 0.000549802913909451,
      "loss": 2.4858,
      "step": 43041
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.691080629825592,
      "learning_rate": 0.0005498006486968569,
      "loss": 2.2771,
      "step": 43042
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.6741182208061218,
      "learning_rate": 0.0005497983834378203,
      "loss": 2.3833,
      "step": 43043
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.745566725730896,
      "learning_rate": 0.0005497961181323414,
      "loss": 2.2299,
      "step": 43044
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.7284832000732422,
      "learning_rate": 0.0005497938527804207,
      "loss": 2.5303,
      "step": 43045
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.7124274373054504,
      "learning_rate": 0.0005497915873820587,
      "loss": 2.1217,
      "step": 43046
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.6445256471633911,
      "learning_rate": 0.0005497893219372556,
      "loss": 2.2429,
      "step": 43047
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.7486974596977234,
      "learning_rate": 0.0005497870564460121,
      "loss": 2.1084,
      "step": 43048
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.7187952995300293,
      "learning_rate": 0.0005497847909083283,
      "loss": 2.2736,
      "step": 43049
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.7350819110870361,
      "learning_rate": 0.0005497825253242049,
      "loss": 2.3455,
      "step": 43050
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.7233168482780457,
      "learning_rate": 0.0005497802596936422,
      "loss": 2.1323,
      "step": 43051
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.7600288391113281,
      "learning_rate": 0.0005497779940166406,
      "loss": 2.4258,
      "step": 43052
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.6680011749267578,
      "learning_rate": 0.0005497757282932006,
      "loss": 2.1249,
      "step": 43053
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.7571635842323303,
      "learning_rate": 0.0005497734625233224,
      "loss": 2.1346,
      "step": 43054
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.6959699392318726,
      "learning_rate": 0.0005497711967070068,
      "loss": 2.2287,
      "step": 43055
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.6792492866516113,
      "learning_rate": 0.000549768930844254,
      "loss": 2.0596,
      "step": 43056
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.789111316204071,
      "learning_rate": 0.0005497666649350643,
      "loss": 2.2423,
      "step": 43057
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.753459632396698,
      "learning_rate": 0.0005497643989794385,
      "loss": 2.2781,
      "step": 43058
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.6910347938537598,
      "learning_rate": 0.0005497621329773766,
      "loss": 2.279,
      "step": 43059
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.6608757972717285,
      "learning_rate": 0.0005497598669288792,
      "loss": 1.893,
      "step": 43060
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.808634340763092,
      "learning_rate": 0.0005497576008339468,
      "loss": 2.4155,
      "step": 43061
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.6878712177276611,
      "learning_rate": 0.0005497553346925798,
      "loss": 2.4248,
      "step": 43062
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.8981375098228455,
      "learning_rate": 0.0005497530685047784,
      "loss": 2.301,
      "step": 43063
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.6534541249275208,
      "learning_rate": 0.0005497508022705433,
      "loss": 2.4464,
      "step": 43064
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.6839404106140137,
      "learning_rate": 0.0005497485359898748,
      "loss": 2.3396,
      "step": 43065
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.8812821507453918,
      "learning_rate": 0.0005497462696627734,
      "loss": 2.3292,
      "step": 43066
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.6199762225151062,
      "learning_rate": 0.0005497440032892393,
      "loss": 2.3489,
      "step": 43067
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.7940693497657776,
      "learning_rate": 0.0005497417368692732,
      "loss": 2.2072,
      "step": 43068
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.7009017467498779,
      "learning_rate": 0.0005497394704028753,
      "loss": 2.3359,
      "step": 43069
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.6758749485015869,
      "learning_rate": 0.0005497372038900462,
      "loss": 2.4184,
      "step": 43070
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.6402197480201721,
      "learning_rate": 0.0005497349373307862,
      "loss": 2.0193,
      "step": 43071
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.6920008659362793,
      "learning_rate": 0.0005497326707250958,
      "loss": 2.3343,
      "step": 43072
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.7039710283279419,
      "learning_rate": 0.0005497304040729754,
      "loss": 2.2203,
      "step": 43073
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.7615364789962769,
      "learning_rate": 0.0005497281373744254,
      "loss": 2.3003,
      "step": 43074
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.7693875432014465,
      "learning_rate": 0.0005497258706294462,
      "loss": 2.3018,
      "step": 43075
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.6684352159500122,
      "learning_rate": 0.0005497236038380382,
      "loss": 2.2857,
      "step": 43076
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.671963632106781,
      "learning_rate": 0.0005497213370002019,
      "loss": 2.3541,
      "step": 43077
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.6917016506195068,
      "learning_rate": 0.0005497190701159376,
      "loss": 2.1072,
      "step": 43078
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.7114253640174866,
      "learning_rate": 0.000549716803185246,
      "loss": 2.4832,
      "step": 43079
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.6448368430137634,
      "learning_rate": 0.0005497145362081272,
      "loss": 2.198,
      "step": 43080
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.7049986720085144,
      "learning_rate": 0.0005497122691845818,
      "loss": 2.2819,
      "step": 43081
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.7140931487083435,
      "learning_rate": 0.0005497100021146102,
      "loss": 2.2853,
      "step": 43082
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.726324200630188,
      "learning_rate": 0.0005497077349982128,
      "loss": 2.1685,
      "step": 43083
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.9506115913391113,
      "learning_rate": 0.0005497054678353898,
      "loss": 2.2522,
      "step": 43084
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.0036884546279907,
      "learning_rate": 0.0005497032006261421,
      "loss": 2.3635,
      "step": 43085
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.6773006916046143,
      "learning_rate": 0.0005497009333704699,
      "loss": 2.508,
      "step": 43086
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.7903699278831482,
      "learning_rate": 0.0005496986660683734,
      "loss": 2.2286,
      "step": 43087
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.7764244079589844,
      "learning_rate": 0.0005496963987198533,
      "loss": 2.3928,
      "step": 43088
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.091931700706482,
      "learning_rate": 0.0005496941313249099,
      "loss": 2.1319,
      "step": 43089
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.8231282234191895,
      "learning_rate": 0.0005496918638835436,
      "loss": 2.4332,
      "step": 43090
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.185462474822998,
      "learning_rate": 0.0005496895963957548,
      "loss": 2.4495,
      "step": 43091
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.8054810762405396,
      "learning_rate": 0.0005496873288615442,
      "loss": 2.3956,
      "step": 43092
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.6715225577354431,
      "learning_rate": 0.0005496850612809118,
      "loss": 2.2791,
      "step": 43093
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.9988039135932922,
      "learning_rate": 0.0005496827936538585,
      "loss": 2.0264,
      "step": 43094
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.7466795444488525,
      "learning_rate": 0.0005496805259803842,
      "loss": 2.1795,
      "step": 43095
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.7587131261825562,
      "learning_rate": 0.0005496782582604898,
      "loss": 2.3416,
      "step": 43096
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.7632759213447571,
      "learning_rate": 0.0005496759904941753,
      "loss": 2.3108,
      "step": 43097
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.821614682674408,
      "learning_rate": 0.0005496737226814415,
      "loss": 2.2211,
      "step": 43098
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.7745955586433411,
      "learning_rate": 0.0005496714548222885,
      "loss": 2.0662,
      "step": 43099
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.6729641556739807,
      "learning_rate": 0.0005496691869167168,
      "loss": 2.1688,
      "step": 43100
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.7354579567909241,
      "learning_rate": 0.0005496669189647271,
      "loss": 2.3066,
      "step": 43101
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.7251558303833008,
      "learning_rate": 0.0005496646509663195,
      "loss": 2.3719,
      "step": 43102
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.710482120513916,
      "learning_rate": 0.0005496623829214946,
      "loss": 2.3512,
      "step": 43103
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.7418417930603027,
      "learning_rate": 0.0005496601148302526,
      "loss": 2.249,
      "step": 43104
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.6942059993743896,
      "learning_rate": 0.0005496578466925942,
      "loss": 2.2965,
      "step": 43105
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.7028114199638367,
      "learning_rate": 0.0005496555785085196,
      "loss": 2.1081,
      "step": 43106
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.6475133299827576,
      "learning_rate": 0.0005496533102780294,
      "loss": 2.168,
      "step": 43107
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.7326790690422058,
      "learning_rate": 0.0005496510420011239,
      "loss": 2.3224,
      "step": 43108
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.6687158942222595,
      "learning_rate": 0.0005496487736778035,
      "loss": 2.4848,
      "step": 43109
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.7467741966247559,
      "learning_rate": 0.0005496465053080687,
      "loss": 2.1698,
      "step": 43110
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.742090106010437,
      "learning_rate": 0.00054964423689192,
      "loss": 2.3108,
      "step": 43111
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.7143953442573547,
      "learning_rate": 0.0005496419684293577,
      "loss": 2.259,
      "step": 43112
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.7213353514671326,
      "learning_rate": 0.0005496396999203822,
      "loss": 2.2161,
      "step": 43113
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.7351602911949158,
      "learning_rate": 0.0005496374313649939,
      "loss": 2.3452,
      "step": 43114
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.7506316900253296,
      "learning_rate": 0.0005496351627631934,
      "loss": 2.3068,
      "step": 43115
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.6639910340309143,
      "learning_rate": 0.0005496328941149809,
      "loss": 2.1722,
      "step": 43116
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.7422268390655518,
      "learning_rate": 0.0005496306254203571,
      "loss": 2.1425,
      "step": 43117
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.7200874090194702,
      "learning_rate": 0.0005496283566793222,
      "loss": 2.2987,
      "step": 43118
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.6958484649658203,
      "learning_rate": 0.0005496260878918766,
      "loss": 2.4858,
      "step": 43119
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.7876444458961487,
      "learning_rate": 0.0005496238190580208,
      "loss": 2.3576,
      "step": 43120
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.9467384815216064,
      "learning_rate": 0.0005496215501777553,
      "loss": 2.1086,
      "step": 43121
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.6920480132102966,
      "learning_rate": 0.0005496192812510804,
      "loss": 2.3436,
      "step": 43122
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.7523747682571411,
      "learning_rate": 0.0005496170122779965,
      "loss": 2.2711,
      "step": 43123
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.7501529455184937,
      "learning_rate": 0.0005496147432585042,
      "loss": 2.2069,
      "step": 43124
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.8097500205039978,
      "learning_rate": 0.0005496124741926037,
      "loss": 2.3069,
      "step": 43125
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.7112904787063599,
      "learning_rate": 0.0005496102050802957,
      "loss": 2.2399,
      "step": 43126
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.779080867767334,
      "learning_rate": 0.0005496079359215803,
      "loss": 2.1975,
      "step": 43127
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.8325525522232056,
      "learning_rate": 0.0005496056667164581,
      "loss": 2.3826,
      "step": 43128
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.6464806795120239,
      "learning_rate": 0.0005496033974649295,
      "loss": 2.2425,
      "step": 43129
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.9159905910491943,
      "learning_rate": 0.000549601128166995,
      "loss": 2.4417,
      "step": 43130
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.7592617869377136,
      "learning_rate": 0.0005495988588226548,
      "loss": 2.1125,
      "step": 43131
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.6859946250915527,
      "learning_rate": 0.0005495965894319095,
      "loss": 2.0478,
      "step": 43132
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.7407442331314087,
      "learning_rate": 0.0005495943199947595,
      "loss": 2.1414,
      "step": 43133
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.6889174580574036,
      "learning_rate": 0.0005495920505112053,
      "loss": 2.3702,
      "step": 43134
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.6870259642601013,
      "learning_rate": 0.0005495897809812471,
      "loss": 2.4003,
      "step": 43135
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.8100033402442932,
      "learning_rate": 0.0005495875114048854,
      "loss": 2.2254,
      "step": 43136
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.6967387795448303,
      "learning_rate": 0.0005495852417821208,
      "loss": 2.0094,
      "step": 43137
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.8041760325431824,
      "learning_rate": 0.0005495829721129536,
      "loss": 2.4187,
      "step": 43138
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.6886605024337769,
      "learning_rate": 0.000549580702397384,
      "loss": 2.3133,
      "step": 43139
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.6695495247840881,
      "learning_rate": 0.0005495784326354128,
      "loss": 2.202,
      "step": 43140
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.7289758324623108,
      "learning_rate": 0.0005495761628270403,
      "loss": 2.4511,
      "step": 43141
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.6573546528816223,
      "learning_rate": 0.0005495738929722668,
      "loss": 2.3329,
      "step": 43142
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.6588670015335083,
      "learning_rate": 0.0005495716230710929,
      "loss": 2.4309,
      "step": 43143
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.7231160998344421,
      "learning_rate": 0.0005495693531235189,
      "loss": 2.5983,
      "step": 43144
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.671119749546051,
      "learning_rate": 0.0005495670831295452,
      "loss": 1.984,
      "step": 43145
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.6700963973999023,
      "learning_rate": 0.0005495648130891722,
      "loss": 2.2146,
      "step": 43146
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.6818835139274597,
      "learning_rate": 0.0005495625430024005,
      "loss": 2.3596,
      "step": 43147
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.7403977513313293,
      "learning_rate": 0.0005495602728692303,
      "loss": 2.1562,
      "step": 43148
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.6601003408432007,
      "learning_rate": 0.0005495580026896623,
      "loss": 2.1702,
      "step": 43149
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.6986052393913269,
      "learning_rate": 0.0005495557324636966,
      "loss": 2.2042,
      "step": 43150
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.7546675205230713,
      "learning_rate": 0.0005495534621913339,
      "loss": 2.3082,
      "step": 43151
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.628488302230835,
      "learning_rate": 0.0005495511918725745,
      "loss": 2.3742,
      "step": 43152
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.6388651132583618,
      "learning_rate": 0.0005495489215074187,
      "loss": 2.1128,
      "step": 43153
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.8749098777770996,
      "learning_rate": 0.0005495466510958671,
      "loss": 2.2554,
      "step": 43154
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.7682097554206848,
      "learning_rate": 0.0005495443806379202,
      "loss": 2.1444,
      "step": 43155
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.7441569566726685,
      "learning_rate": 0.000549542110133578,
      "loss": 2.2242,
      "step": 43156
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.675751805305481,
      "learning_rate": 0.0005495398395828414,
      "loss": 2.5936,
      "step": 43157
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.7070342302322388,
      "learning_rate": 0.0005495375689857107,
      "loss": 2.1007,
      "step": 43158
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.8012633323669434,
      "learning_rate": 0.0005495352983421862,
      "loss": 2.1991,
      "step": 43159
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.722459077835083,
      "learning_rate": 0.0005495330276522683,
      "loss": 2.2743,
      "step": 43160
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.7135249972343445,
      "learning_rate": 0.0005495307569159575,
      "loss": 2.2854,
      "step": 43161
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.7146398425102234,
      "learning_rate": 0.0005495284861332543,
      "loss": 2.2717,
      "step": 43162
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.7764624953269958,
      "learning_rate": 0.000549526215304159,
      "loss": 2.1963,
      "step": 43163
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.7022436857223511,
      "learning_rate": 0.0005495239444286721,
      "loss": 2.3041,
      "step": 43164
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.8253841400146484,
      "learning_rate": 0.000549521673506794,
      "loss": 2.3161,
      "step": 43165
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.809341549873352,
      "learning_rate": 0.0005495194025385251,
      "loss": 2.3081,
      "step": 43166
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.6891997456550598,
      "learning_rate": 0.0005495171315238658,
      "loss": 2.1956,
      "step": 43167
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.6593963503837585,
      "learning_rate": 0.0005495148604628165,
      "loss": 2.3406,
      "step": 43168
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.7751864790916443,
      "learning_rate": 0.0005495125893553778,
      "loss": 2.2158,
      "step": 43169
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.631773829460144,
      "learning_rate": 0.00054951031820155,
      "loss": 2.1822,
      "step": 43170
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.7412121295928955,
      "learning_rate": 0.0005495080470013334,
      "loss": 2.434,
      "step": 43171
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.7461463809013367,
      "learning_rate": 0.0005495057757547286,
      "loss": 2.1801,
      "step": 43172
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.6849039196968079,
      "learning_rate": 0.0005495035044617361,
      "loss": 2.0165,
      "step": 43173
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.7100605368614197,
      "learning_rate": 0.000549501233122356,
      "loss": 2.0013,
      "step": 43174
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.8332968354225159,
      "learning_rate": 0.000549498961736589,
      "loss": 2.3185,
      "step": 43175
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.7291807532310486,
      "learning_rate": 0.0005494966903044354,
      "loss": 2.1019,
      "step": 43176
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.8308221101760864,
      "learning_rate": 0.0005494944188258956,
      "loss": 2.2521,
      "step": 43177
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.6822531223297119,
      "learning_rate": 0.0005494921473009702,
      "loss": 2.2297,
      "step": 43178
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.6928606033325195,
      "learning_rate": 0.0005494898757296595,
      "loss": 2.3386,
      "step": 43179
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.8988938927650452,
      "learning_rate": 0.0005494876041119638,
      "loss": 2.1249,
      "step": 43180
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.7238871455192566,
      "learning_rate": 0.0005494853324478837,
      "loss": 2.2129,
      "step": 43181
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.6458896994590759,
      "learning_rate": 0.0005494830607374196,
      "loss": 2.0803,
      "step": 43182
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.8892576098442078,
      "learning_rate": 0.0005494807889805718,
      "loss": 2.1713,
      "step": 43183
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.8728708028793335,
      "learning_rate": 0.000549478517177341,
      "loss": 2.2701,
      "step": 43184
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.6882877945899963,
      "learning_rate": 0.0005494762453277273,
      "loss": 2.2099,
      "step": 43185
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.7881877422332764,
      "learning_rate": 0.0005494739734317312,
      "loss": 2.1926,
      "step": 43186
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.6964231133460999,
      "learning_rate": 0.0005494717014893533,
      "loss": 2.148,
      "step": 43187
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.6671605706214905,
      "learning_rate": 0.0005494694295005939,
      "loss": 2.2594,
      "step": 43188
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.7081008553504944,
      "learning_rate": 0.0005494671574654533,
      "loss": 2.3074,
      "step": 43189
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.8443422913551331,
      "learning_rate": 0.0005494648853839321,
      "loss": 2.379,
      "step": 43190
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.6959161162376404,
      "learning_rate": 0.0005494626132560307,
      "loss": 2.4651,
      "step": 43191
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.6707978248596191,
      "learning_rate": 0.0005494603410817495,
      "loss": 1.995,
      "step": 43192
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.8942024111747742,
      "learning_rate": 0.000549458068861089,
      "loss": 2.1366,
      "step": 43193
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.6679838299751282,
      "learning_rate": 0.0005494557965940494,
      "loss": 2.1168,
      "step": 43194
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.6446121335029602,
      "learning_rate": 0.0005494535242806314,
      "loss": 2.2218,
      "step": 43195
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.669948160648346,
      "learning_rate": 0.0005494512519208351,
      "loss": 2.2125,
      "step": 43196
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.6291391253471375,
      "learning_rate": 0.0005494489795146611,
      "loss": 2.2076,
      "step": 43197
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.705642819404602,
      "learning_rate": 0.0005494467070621099,
      "loss": 2.2808,
      "step": 43198
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.7045721411705017,
      "learning_rate": 0.0005494444345631819,
      "loss": 2.3144,
      "step": 43199
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.6729510426521301,
      "learning_rate": 0.0005494421620178774,
      "loss": 2.085,
      "step": 43200
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.7259923219680786,
      "learning_rate": 0.0005494398894261968,
      "loss": 2.2386,
      "step": 43201
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.7876562476158142,
      "learning_rate": 0.0005494376167881409,
      "loss": 2.423,
      "step": 43202
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.6607381701469421,
      "learning_rate": 0.0005494353441037096,
      "loss": 2.3129,
      "step": 43203
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.6876407265663147,
      "learning_rate": 0.0005494330713729037,
      "loss": 2.2086,
      "step": 43204
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.6901463866233826,
      "learning_rate": 0.0005494307985957235,
      "loss": 2.3951,
      "step": 43205
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.6650844812393188,
      "learning_rate": 0.0005494285257721692,
      "loss": 2.1701,
      "step": 43206
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.6483656167984009,
      "learning_rate": 0.0005494262529022416,
      "loss": 2.0412,
      "step": 43207
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.7759782075881958,
      "learning_rate": 0.0005494239799859409,
      "loss": 2.2787,
      "step": 43208
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.7374202609062195,
      "learning_rate": 0.0005494217070232674,
      "loss": 2.2878,
      "step": 43209
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.8053290843963623,
      "learning_rate": 0.0005494194340142219,
      "loss": 2.1182,
      "step": 43210
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.7432953119277954,
      "learning_rate": 0.0005494171609588047,
      "loss": 2.2088,
      "step": 43211
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.690117359161377,
      "learning_rate": 0.000549414887857016,
      "loss": 2.4281,
      "step": 43212
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.7362483739852905,
      "learning_rate": 0.0005494126147088564,
      "loss": 2.2436,
      "step": 43213
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.6760904788970947,
      "learning_rate": 0.0005494103415143263,
      "loss": 2.3892,
      "step": 43214
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.648600161075592,
      "learning_rate": 0.0005494080682734261,
      "loss": 2.3288,
      "step": 43215
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.6861203908920288,
      "learning_rate": 0.0005494057949861562,
      "loss": 2.1415,
      "step": 43216
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.7064000368118286,
      "learning_rate": 0.0005494035216525171,
      "loss": 2.2886,
      "step": 43217
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.7290548086166382,
      "learning_rate": 0.0005494012482725091,
      "loss": 2.1954,
      "step": 43218
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.7391183376312256,
      "learning_rate": 0.0005493989748461328,
      "loss": 2.2959,
      "step": 43219
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.6250030398368835,
      "learning_rate": 0.0005493967013733885,
      "loss": 2.0341,
      "step": 43220
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.6939297914505005,
      "learning_rate": 0.0005493944278542767,
      "loss": 2.1517,
      "step": 43221
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.735817551612854,
      "learning_rate": 0.0005493921542887977,
      "loss": 2.3792,
      "step": 43222
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.7164406180381775,
      "learning_rate": 0.0005493898806769519,
      "loss": 2.2726,
      "step": 43223
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.6670630574226379,
      "learning_rate": 0.0005493876070187399,
      "loss": 2.2965,
      "step": 43224
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.6623457670211792,
      "learning_rate": 0.0005493853333141621,
      "loss": 2.5071,
      "step": 43225
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.736334502696991,
      "learning_rate": 0.0005493830595632187,
      "loss": 2.0724,
      "step": 43226
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.7364544868469238,
      "learning_rate": 0.0005493807857659104,
      "loss": 2.2745,
      "step": 43227
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.6764693260192871,
      "learning_rate": 0.0005493785119222376,
      "loss": 2.1918,
      "step": 43228
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.7203229665756226,
      "learning_rate": 0.0005493762380322004,
      "loss": 2.2431,
      "step": 43229
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.671448290348053,
      "learning_rate": 0.0005493739640957996,
      "loss": 2.2666,
      "step": 43230
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.7071559429168701,
      "learning_rate": 0.0005493716901130354,
      "loss": 2.0858,
      "step": 43231
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.6170870065689087,
      "learning_rate": 0.0005493694160839083,
      "loss": 2.2459,
      "step": 43232
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.9305136799812317,
      "learning_rate": 0.0005493671420084187,
      "loss": 2.0128,
      "step": 43233
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.7693148851394653,
      "learning_rate": 0.0005493648678865672,
      "loss": 2.2193,
      "step": 43234
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.7243789434432983,
      "learning_rate": 0.0005493625937183538,
      "loss": 2.2925,
      "step": 43235
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.7612221837043762,
      "learning_rate": 0.0005493603195037795,
      "loss": 2.3444,
      "step": 43236
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.7702760100364685,
      "learning_rate": 0.0005493580452428442,
      "loss": 2.4059,
      "step": 43237
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.7264482378959656,
      "learning_rate": 0.0005493557709355487,
      "loss": 2.4687,
      "step": 43238
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.7092511653900146,
      "learning_rate": 0.000549353496581893,
      "loss": 2.1391,
      "step": 43239
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.8060606122016907,
      "learning_rate": 0.0005493512221818779,
      "loss": 2.3398,
      "step": 43240
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.732132613658905,
      "learning_rate": 0.0005493489477355037,
      "loss": 2.6416,
      "step": 43241
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.7388646006584167,
      "learning_rate": 0.0005493466732427709,
      "loss": 2.5007,
      "step": 43242
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.686124861240387,
      "learning_rate": 0.0005493443987036798,
      "loss": 2.2369,
      "step": 43243
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.633677065372467,
      "learning_rate": 0.0005493421241182309,
      "loss": 2.4344,
      "step": 43244
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.6555941104888916,
      "learning_rate": 0.0005493398494864245,
      "loss": 2.3432,
      "step": 43245
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.6664395332336426,
      "learning_rate": 0.0005493375748082612,
      "loss": 2.4087,
      "step": 43246
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.6665794849395752,
      "learning_rate": 0.0005493353000837412,
      "loss": 2.4767,
      "step": 43247
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.665016770362854,
      "learning_rate": 0.0005493330253128651,
      "loss": 2.2827,
      "step": 43248
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.6653527021408081,
      "learning_rate": 0.0005493307504956334,
      "loss": 2.3011,
      "step": 43249
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.6731242537498474,
      "learning_rate": 0.0005493284756320464,
      "loss": 2.4204,
      "step": 43250
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.6627224087715149,
      "learning_rate": 0.0005493262007221045,
      "loss": 2.4337,
      "step": 43251
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.7766281962394714,
      "learning_rate": 0.0005493239257658081,
      "loss": 2.1616,
      "step": 43252
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.6621471643447876,
      "learning_rate": 0.0005493216507631577,
      "loss": 2.4467,
      "step": 43253
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.7688771486282349,
      "learning_rate": 0.0005493193757141537,
      "loss": 2.2208,
      "step": 43254
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.817640483379364,
      "learning_rate": 0.0005493171006187965,
      "loss": 2.3323,
      "step": 43255
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.6724271178245544,
      "learning_rate": 0.0005493148254770866,
      "loss": 2.3252,
      "step": 43256
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.7535417079925537,
      "learning_rate": 0.0005493125502890244,
      "loss": 2.4233,
      "step": 43257
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.889476478099823,
      "learning_rate": 0.0005493102750546102,
      "loss": 2.2097,
      "step": 43258
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.6810131669044495,
      "learning_rate": 0.0005493079997738445,
      "loss": 2.18,
      "step": 43259
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.8493314385414124,
      "learning_rate": 0.0005493057244467278,
      "loss": 2.1981,
      "step": 43260
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.0196291208267212,
      "learning_rate": 0.0005493034490732605,
      "loss": 2.3367,
      "step": 43261
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.68490070104599,
      "learning_rate": 0.000549301173653443,
      "loss": 2.2839,
      "step": 43262
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.8201106786727905,
      "learning_rate": 0.0005492988981872756,
      "loss": 2.6086,
      "step": 43263
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.7558485269546509,
      "learning_rate": 0.0005492966226747588,
      "loss": 2.2763,
      "step": 43264
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.9082675576210022,
      "learning_rate": 0.0005492943471158931,
      "loss": 2.3411,
      "step": 43265
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.659095287322998,
      "learning_rate": 0.0005492920715106788,
      "loss": 2.1826,
      "step": 43266
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.7133359909057617,
      "learning_rate": 0.0005492897958591166,
      "loss": 2.4624,
      "step": 43267
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.1792038679122925,
      "learning_rate": 0.0005492875201612066,
      "loss": 2.4449,
      "step": 43268
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.9105897545814514,
      "learning_rate": 0.0005492852444169493,
      "loss": 2.1446,
      "step": 43269
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.6454830169677734,
      "learning_rate": 0.0005492829686263453,
      "loss": 2.3895,
      "step": 43270
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.3283367156982422,
      "learning_rate": 0.0005492806927893948,
      "loss": 2.4088,
      "step": 43271
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.018781304359436,
      "learning_rate": 0.0005492784169060983,
      "loss": 2.4271,
      "step": 43272
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.6160194277763367,
      "learning_rate": 0.0005492761409764563,
      "loss": 1.9455,
      "step": 43273
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.684363067150116,
      "learning_rate": 0.0005492738650004691,
      "loss": 2.3918,
      "step": 43274
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.9699714779853821,
      "learning_rate": 0.0005492715889781373,
      "loss": 2.4481,
      "step": 43275
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.874248743057251,
      "learning_rate": 0.000549269312909461,
      "loss": 2.2161,
      "step": 43276
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.6653486490249634,
      "learning_rate": 0.000549267036794441,
      "loss": 2.1399,
      "step": 43277
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.8202645182609558,
      "learning_rate": 0.0005492647606330774,
      "loss": 2.303,
      "step": 43278
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.7954267859458923,
      "learning_rate": 0.0005492624844253709,
      "loss": 2.0739,
      "step": 43279
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.7910961508750916,
      "learning_rate": 0.0005492602081713217,
      "loss": 2.3466,
      "step": 43280
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.6534162163734436,
      "learning_rate": 0.0005492579318709305,
      "loss": 2.2057,
      "step": 43281
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.7463506460189819,
      "learning_rate": 0.0005492556555241973,
      "loss": 2.2948,
      "step": 43282
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.8706713914871216,
      "learning_rate": 0.0005492533791311231,
      "loss": 2.2529,
      "step": 43283
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.8146232962608337,
      "learning_rate": 0.0005492511026917077,
      "loss": 2.1807,
      "step": 43284
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.667671799659729,
      "learning_rate": 0.000549248826205952,
      "loss": 2.4329,
      "step": 43285
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.7751926779747009,
      "learning_rate": 0.0005492465496738561,
      "loss": 2.2956,
      "step": 43286
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.1399704217910767,
      "learning_rate": 0.0005492442730954206,
      "loss": 2.1611,
      "step": 43287
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.8162915706634521,
      "learning_rate": 0.0005492419964706459,
      "loss": 2.4337,
      "step": 43288
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.7080253958702087,
      "learning_rate": 0.0005492397197995323,
      "loss": 2.0199,
      "step": 43289
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.7975649833679199,
      "learning_rate": 0.0005492374430820805,
      "loss": 2.3377,
      "step": 43290
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.8191112875938416,
      "learning_rate": 0.0005492351663182908,
      "loss": 2.4157,
      "step": 43291
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.7510530352592468,
      "learning_rate": 0.0005492328895081634,
      "loss": 2.1334,
      "step": 43292
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.754936695098877,
      "learning_rate": 0.000549230612651699,
      "loss": 2.2337,
      "step": 43293
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.8858052492141724,
      "learning_rate": 0.0005492283357488979,
      "loss": 2.1555,
      "step": 43294
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.7820952534675598,
      "learning_rate": 0.0005492260587997605,
      "loss": 2.3495,
      "step": 43295
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.7786719799041748,
      "learning_rate": 0.0005492237818042873,
      "loss": 2.4978,
      "step": 43296
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.8841004967689514,
      "learning_rate": 0.0005492215047624787,
      "loss": 2.2394,
      "step": 43297
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.8534592390060425,
      "learning_rate": 0.0005492192276743351,
      "loss": 2.5136,
      "step": 43298
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.8577565550804138,
      "learning_rate": 0.000549216950539857,
      "loss": 2.2299,
      "step": 43299
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.923730731010437,
      "learning_rate": 0.0005492146733590446,
      "loss": 2.1041,
      "step": 43300
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.7396661043167114,
      "learning_rate": 0.0005492123961318988,
      "loss": 2.1073,
      "step": 43301
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.6844340562820435,
      "learning_rate": 0.0005492101188584195,
      "loss": 2.4788,
      "step": 43302
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.7304432392120361,
      "learning_rate": 0.0005492078415386074,
      "loss": 2.0357,
      "step": 43303
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.8387026190757751,
      "learning_rate": 0.0005492055641724627,
      "loss": 2.1981,
      "step": 43304
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.7318989038467407,
      "learning_rate": 0.0005492032867599862,
      "loss": 2.1688,
      "step": 43305
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.8072382807731628,
      "learning_rate": 0.0005492010093011779,
      "loss": 2.5086,
      "step": 43306
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.8624008893966675,
      "learning_rate": 0.0005491987317960387,
      "loss": 2.3634,
      "step": 43307
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.8195739388465881,
      "learning_rate": 0.0005491964542445685,
      "loss": 2.1518,
      "step": 43308
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.700772762298584,
      "learning_rate": 0.0005491941766467681,
      "loss": 2.3735,
      "step": 43309
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.6635563373565674,
      "learning_rate": 0.0005491918990026377,
      "loss": 2.1609,
      "step": 43310
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.781857967376709,
      "learning_rate": 0.0005491896213121779,
      "loss": 2.1538,
      "step": 43311
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.7959250807762146,
      "learning_rate": 0.000549187343575389,
      "loss": 2.2408,
      "step": 43312
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.6951711177825928,
      "learning_rate": 0.0005491850657922715,
      "loss": 2.1527,
      "step": 43313
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.6772757172584534,
      "learning_rate": 0.0005491827879628258,
      "loss": 2.4295,
      "step": 43314
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.7435051798820496,
      "learning_rate": 0.0005491805100870523,
      "loss": 2.2288,
      "step": 43315
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.6724593043327332,
      "learning_rate": 0.0005491782321649514,
      "loss": 2.137,
      "step": 43316
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.675507128238678,
      "learning_rate": 0.0005491759541965237,
      "loss": 2.3206,
      "step": 43317
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.6825176477432251,
      "learning_rate": 0.0005491736761817695,
      "loss": 2.2273,
      "step": 43318
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.797836422920227,
      "learning_rate": 0.000549171398120689,
      "loss": 2.3138,
      "step": 43319
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.6618643403053284,
      "learning_rate": 0.0005491691200132829,
      "loss": 2.119,
      "step": 43320
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.6715333461761475,
      "learning_rate": 0.0005491668418595517,
      "loss": 2.3803,
      "step": 43321
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.8599417209625244,
      "learning_rate": 0.0005491645636594955,
      "loss": 2.2638,
      "step": 43322
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.755219042301178,
      "learning_rate": 0.000549162285413115,
      "loss": 2.361,
      "step": 43323
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.7262292504310608,
      "learning_rate": 0.0005491600071204105,
      "loss": 2.656,
      "step": 43324
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.7179700136184692,
      "learning_rate": 0.0005491577287813825,
      "loss": 2.385,
      "step": 43325
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.7403808236122131,
      "learning_rate": 0.0005491554503960313,
      "loss": 2.3572,
      "step": 43326
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.6581765413284302,
      "learning_rate": 0.0005491531719643575,
      "loss": 2.3382,
      "step": 43327
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.8300436735153198,
      "learning_rate": 0.0005491508934863613,
      "loss": 2.1065,
      "step": 43328
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.6914095878601074,
      "learning_rate": 0.0005491486149620433,
      "loss": 2.2606,
      "step": 43329
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.6616238355636597,
      "learning_rate": 0.0005491463363914039,
      "loss": 2.4055,
      "step": 43330
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.6839609146118164,
      "learning_rate": 0.0005491440577744434,
      "loss": 2.1219,
      "step": 43331
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.71208256483078,
      "learning_rate": 0.0005491417791111624,
      "loss": 2.1458,
      "step": 43332
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.7822561860084534,
      "learning_rate": 0.0005491395004015613,
      "loss": 2.3098,
      "step": 43333
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.6428763270378113,
      "learning_rate": 0.0005491372216456403,
      "loss": 2.0771,
      "step": 43334
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.6785741448402405,
      "learning_rate": 0.0005491349428434001,
      "loss": 2.2445,
      "step": 43335
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.6394723653793335,
      "learning_rate": 0.000549132663994841,
      "loss": 2.2665,
      "step": 43336
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.6928409337997437,
      "learning_rate": 0.0005491303850999634,
      "loss": 2.4386,
      "step": 43337
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.7626000642776489,
      "learning_rate": 0.0005491281061587678,
      "loss": 2.3544,
      "step": 43338
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.8379860520362854,
      "learning_rate": 0.0005491258271712546,
      "loss": 2.4343,
      "step": 43339
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.736301064491272,
      "learning_rate": 0.0005491235481374241,
      "loss": 2.2824,
      "step": 43340
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.7485452890396118,
      "learning_rate": 0.0005491212690572769,
      "loss": 2.1526,
      "step": 43341
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.9687157869338989,
      "learning_rate": 0.0005491189899308135,
      "loss": 2.3702,
      "step": 43342
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.6595783829689026,
      "learning_rate": 0.0005491167107580338,
      "loss": 2.4275,
      "step": 43343
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.6652487516403198,
      "learning_rate": 0.0005491144315389389,
      "loss": 2.455,
      "step": 43344
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.6345334053039551,
      "learning_rate": 0.0005491121522735289,
      "loss": 2.166,
      "step": 43345
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.6570379734039307,
      "learning_rate": 0.0005491098729618042,
      "loss": 2.2348,
      "step": 43346
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.7450543642044067,
      "learning_rate": 0.0005491075936037652,
      "loss": 2.2913,
      "step": 43347
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.7514005899429321,
      "learning_rate": 0.0005491053141994125,
      "loss": 2.3695,
      "step": 43348
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.6817735433578491,
      "learning_rate": 0.0005491030347487464,
      "loss": 2.1298,
      "step": 43349
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.7284563183784485,
      "learning_rate": 0.0005491007552517673,
      "loss": 2.3905,
      "step": 43350
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.7386649250984192,
      "learning_rate": 0.0005490984757084758,
      "loss": 2.0702,
      "step": 43351
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.6575617790222168,
      "learning_rate": 0.000549096196118872,
      "loss": 2.5039,
      "step": 43352
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.780407190322876,
      "learning_rate": 0.0005490939164829566,
      "loss": 2.4655,
      "step": 43353
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.7184612154960632,
      "learning_rate": 0.00054909163680073,
      "loss": 2.2109,
      "step": 43354
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.6725492477416992,
      "learning_rate": 0.0005490893570721925,
      "loss": 2.393,
      "step": 43355
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.6502846479415894,
      "learning_rate": 0.0005490870772973446,
      "loss": 2.3117,
      "step": 43356
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.7043170928955078,
      "learning_rate": 0.0005490847974761867,
      "loss": 2.3977,
      "step": 43357
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.7084116339683533,
      "learning_rate": 0.0005490825176087192,
      "loss": 2.2419,
      "step": 43358
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.7442262172698975,
      "learning_rate": 0.0005490802376949426,
      "loss": 2.3756,
      "step": 43359
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.6607348918914795,
      "learning_rate": 0.0005490779577348574,
      "loss": 2.1825,
      "step": 43360
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.6841340065002441,
      "learning_rate": 0.0005490756777284637,
      "loss": 2.0754,
      "step": 43361
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.6646318435668945,
      "learning_rate": 0.0005490733976757622,
      "loss": 2.1986,
      "step": 43362
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.6557016968727112,
      "learning_rate": 0.0005490711175767533,
      "loss": 2.5433,
      "step": 43363
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.6956528425216675,
      "learning_rate": 0.0005490688374314374,
      "loss": 2.2817,
      "step": 43364
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.6422584652900696,
      "learning_rate": 0.0005490665572398148,
      "loss": 2.1403,
      "step": 43365
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.6619652509689331,
      "learning_rate": 0.0005490642770018861,
      "loss": 2.2097,
      "step": 43366
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.6598905324935913,
      "learning_rate": 0.0005490619967176515,
      "loss": 2.3304,
      "step": 43367
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.7345667481422424,
      "learning_rate": 0.0005490597163871118,
      "loss": 2.2458,
      "step": 43368
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.6843665838241577,
      "learning_rate": 0.0005490574360102671,
      "loss": 2.2217,
      "step": 43369
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.6609147191047668,
      "learning_rate": 0.000549055155587118,
      "loss": 2.2477,
      "step": 43370
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.7463124990463257,
      "learning_rate": 0.0005490528751176647,
      "loss": 2.1772,
      "step": 43371
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.6404315829277039,
      "learning_rate": 0.0005490505946019079,
      "loss": 2.2156,
      "step": 43372
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.7535818219184875,
      "learning_rate": 0.0005490483140398477,
      "loss": 2.1356,
      "step": 43373
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.6794283390045166,
      "learning_rate": 0.0005490460334314849,
      "loss": 2.5954,
      "step": 43374
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.661686360836029,
      "learning_rate": 0.0005490437527768196,
      "loss": 2.1874,
      "step": 43375
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.709926187992096,
      "learning_rate": 0.0005490414720758525,
      "loss": 2.1901,
      "step": 43376
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.7472578287124634,
      "learning_rate": 0.0005490391913285839,
      "loss": 2.2714,
      "step": 43377
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.6818003058433533,
      "learning_rate": 0.0005490369105350142,
      "loss": 2.2241,
      "step": 43378
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.6973049640655518,
      "learning_rate": 0.0005490346296951437,
      "loss": 2.4186,
      "step": 43379
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.9150241613388062,
      "learning_rate": 0.0005490323488089731,
      "loss": 2.1981,
      "step": 43380
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.6640646457672119,
      "learning_rate": 0.0005490300678765025,
      "loss": 2.3741,
      "step": 43381
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.7370967268943787,
      "learning_rate": 0.0005490277868977327,
      "loss": 2.3538,
      "step": 43382
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.8778579831123352,
      "learning_rate": 0.0005490255058726639,
      "loss": 2.1849,
      "step": 43383
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.6537800431251526,
      "learning_rate": 0.0005490232248012965,
      "loss": 2.3268,
      "step": 43384
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.7940323948860168,
      "learning_rate": 0.0005490209436836311,
      "loss": 2.383,
      "step": 43385
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.6879245042800903,
      "learning_rate": 0.0005490186625196679,
      "loss": 2.1477,
      "step": 43386
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.6823185682296753,
      "learning_rate": 0.0005490163813094075,
      "loss": 2.3677,
      "step": 43387
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.6815378665924072,
      "learning_rate": 0.0005490141000528502,
      "loss": 2.1692,
      "step": 43388
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.6692392230033875,
      "learning_rate": 0.0005490118187499965,
      "loss": 2.1968,
      "step": 43389
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.7273299694061279,
      "learning_rate": 0.0005490095374008467,
      "loss": 2.3984,
      "step": 43390
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.6796921491622925,
      "learning_rate": 0.0005490072560054015,
      "loss": 2.2602,
      "step": 43391
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.761752724647522,
      "learning_rate": 0.0005490049745636611,
      "loss": 2.3427,
      "step": 43392
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.6659413576126099,
      "learning_rate": 0.0005490026930756259,
      "loss": 2.1237,
      "step": 43393
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.6754850745201111,
      "learning_rate": 0.0005490004115412964,
      "loss": 2.3138,
      "step": 43394
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7027930617332458,
      "learning_rate": 0.0005489981299606732,
      "loss": 2.3309,
      "step": 43395
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.6618067026138306,
      "learning_rate": 0.0005489958483337563,
      "loss": 2.278,
      "step": 43396
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7346439957618713,
      "learning_rate": 0.0005489935666605465,
      "loss": 2.2768,
      "step": 43397
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.6728821396827698,
      "learning_rate": 0.000548991284941044,
      "loss": 2.0853,
      "step": 43398
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.6727637648582458,
      "learning_rate": 0.0005489890031752494,
      "loss": 2.2771,
      "step": 43399
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7480404376983643,
      "learning_rate": 0.0005489867213631631,
      "loss": 2.219,
      "step": 43400
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.8015201687812805,
      "learning_rate": 0.0005489844395047854,
      "loss": 2.2176,
      "step": 43401
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.6946426033973694,
      "learning_rate": 0.0005489821576001168,
      "loss": 2.2565,
      "step": 43402
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.6427614688873291,
      "learning_rate": 0.0005489798756491578,
      "loss": 2.3265,
      "step": 43403
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7320873141288757,
      "learning_rate": 0.0005489775936519086,
      "loss": 2.3601,
      "step": 43404
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.6872004866600037,
      "learning_rate": 0.0005489753116083698,
      "loss": 2.2139,
      "step": 43405
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7346745729446411,
      "learning_rate": 0.0005489730295185419,
      "loss": 2.2072,
      "step": 43406
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7098615169525146,
      "learning_rate": 0.000548970747382425,
      "loss": 2.3085,
      "step": 43407
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.682951033115387,
      "learning_rate": 0.00054896846520002,
      "loss": 2.2305,
      "step": 43408
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7822166085243225,
      "learning_rate": 0.0005489661829713269,
      "loss": 2.1619,
      "step": 43409
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.6494044065475464,
      "learning_rate": 0.0005489639006963463,
      "loss": 2.2017,
      "step": 43410
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7816382646560669,
      "learning_rate": 0.0005489616183750786,
      "loss": 2.3459,
      "step": 43411
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.6958832144737244,
      "learning_rate": 0.0005489593360075243,
      "loss": 2.3884,
      "step": 43412
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7136519551277161,
      "learning_rate": 0.0005489570535936837,
      "loss": 2.1712,
      "step": 43413
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.6484608054161072,
      "learning_rate": 0.0005489547711335573,
      "loss": 2.4593,
      "step": 43414
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.6390289664268494,
      "learning_rate": 0.0005489524886271456,
      "loss": 2.1209,
      "step": 43415
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.770211935043335,
      "learning_rate": 0.0005489502060744489,
      "loss": 2.2684,
      "step": 43416
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7251085042953491,
      "learning_rate": 0.0005489479234754676,
      "loss": 2.4798,
      "step": 43417
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7100120782852173,
      "learning_rate": 0.0005489456408302021,
      "loss": 2.4431,
      "step": 43418
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.6550353765487671,
      "learning_rate": 0.000548943358138653,
      "loss": 2.1343,
      "step": 43419
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.72661292552948,
      "learning_rate": 0.0005489410754008206,
      "loss": 2.4059,
      "step": 43420
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.6987733840942383,
      "learning_rate": 0.0005489387926167055,
      "loss": 2.0785,
      "step": 43421
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7146062254905701,
      "learning_rate": 0.0005489365097863078,
      "loss": 2.1555,
      "step": 43422
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.8668389320373535,
      "learning_rate": 0.0005489342269096282,
      "loss": 2.3869,
      "step": 43423
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7095049023628235,
      "learning_rate": 0.000548931943986667,
      "loss": 2.2862,
      "step": 43424
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.6636188626289368,
      "learning_rate": 0.0005489296610174247,
      "loss": 2.3721,
      "step": 43425
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.986633837223053,
      "learning_rate": 0.0005489273780019015,
      "loss": 2.349,
      "step": 43426
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7804768085479736,
      "learning_rate": 0.0005489250949400982,
      "loss": 2.4625,
      "step": 43427
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7567647099494934,
      "learning_rate": 0.0005489228118320149,
      "loss": 2.2655,
      "step": 43428
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.001977562904358,
      "learning_rate": 0.0005489205286776523,
      "loss": 2.2051,
      "step": 43429
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.716312825679779,
      "learning_rate": 0.0005489182454770106,
      "loss": 2.276,
      "step": 43430
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7329973578453064,
      "learning_rate": 0.0005489159622300902,
      "loss": 2.2779,
      "step": 43431
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.1517802476882935,
      "learning_rate": 0.0005489136789368917,
      "loss": 2.2123,
      "step": 43432
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.8468422293663025,
      "learning_rate": 0.0005489113955974154,
      "loss": 2.2306,
      "step": 43433
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7694106698036194,
      "learning_rate": 0.0005489091122116618,
      "loss": 2.3294,
      "step": 43434
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.8574157357215881,
      "learning_rate": 0.0005489068287796314,
      "loss": 2.3089,
      "step": 43435
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.0052365064620972,
      "learning_rate": 0.0005489045453013244,
      "loss": 2.2787,
      "step": 43436
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.6465701460838318,
      "learning_rate": 0.0005489022617767414,
      "loss": 1.9926,
      "step": 43437
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7198811173439026,
      "learning_rate": 0.0005488999782058828,
      "loss": 2.5065,
      "step": 43438
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7607279419898987,
      "learning_rate": 0.0005488976945887489,
      "loss": 2.1812,
      "step": 43439
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.8115781545639038,
      "learning_rate": 0.0005488954109253404,
      "loss": 2.2576,
      "step": 43440
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.6926937103271484,
      "learning_rate": 0.0005488931272156573,
      "loss": 2.3748,
      "step": 43441
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7054333090782166,
      "learning_rate": 0.0005488908434597004,
      "loss": 2.4899,
      "step": 43442
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7701871395111084,
      "learning_rate": 0.00054888855965747,
      "loss": 2.4158,
      "step": 43443
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7989141941070557,
      "learning_rate": 0.0005488862758089665,
      "loss": 2.1903,
      "step": 43444
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7454965710639954,
      "learning_rate": 0.0005488839919141904,
      "loss": 2.5478,
      "step": 43445
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.8569928407669067,
      "learning_rate": 0.000548881707973142,
      "loss": 2.2632,
      "step": 43446
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.8677440881729126,
      "learning_rate": 0.0005488794239858219,
      "loss": 1.8801,
      "step": 43447
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7201974987983704,
      "learning_rate": 0.0005488771399522303,
      "loss": 2.2714,
      "step": 43448
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.6811642646789551,
      "learning_rate": 0.0005488748558723677,
      "loss": 2.3896,
      "step": 43449
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7608428597450256,
      "learning_rate": 0.0005488725717462347,
      "loss": 2.0526,
      "step": 43450
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.6947026252746582,
      "learning_rate": 0.0005488702875738314,
      "loss": 2.0206,
      "step": 43451
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7502464056015015,
      "learning_rate": 0.0005488680033551586,
      "loss": 2.1223,
      "step": 43452
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.6881131529808044,
      "learning_rate": 0.0005488657190902165,
      "loss": 2.1044,
      "step": 43453
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.6462079286575317,
      "learning_rate": 0.0005488634347790056,
      "loss": 2.3074,
      "step": 43454
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.8033325672149658,
      "learning_rate": 0.0005488611504215261,
      "loss": 2.2915,
      "step": 43455
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7199784517288208,
      "learning_rate": 0.0005488588660177788,
      "loss": 2.0938,
      "step": 43456
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7951146960258484,
      "learning_rate": 0.0005488565815677639,
      "loss": 2.2447,
      "step": 43457
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.672188401222229,
      "learning_rate": 0.0005488542970714819,
      "loss": 2.1623,
      "step": 43458
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7576016187667847,
      "learning_rate": 0.0005488520125289331,
      "loss": 2.2373,
      "step": 43459
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7212622165679932,
      "learning_rate": 0.0005488497279401181,
      "loss": 2.1586,
      "step": 43460
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.789950430393219,
      "learning_rate": 0.0005488474433050371,
      "loss": 2.1598,
      "step": 43461
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.6932496428489685,
      "learning_rate": 0.0005488451586236909,
      "loss": 2.055,
      "step": 43462
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7280361652374268,
      "learning_rate": 0.0005488428738960794,
      "loss": 2.2303,
      "step": 43463
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7326741218566895,
      "learning_rate": 0.0005488405891222036,
      "loss": 2.2487,
      "step": 43464
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7685874104499817,
      "learning_rate": 0.0005488383043020634,
      "loss": 2.2035,
      "step": 43465
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.677215039730072,
      "learning_rate": 0.0005488360194356596,
      "loss": 2.2142,
      "step": 43466
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.607308030128479,
      "learning_rate": 0.0005488337345229923,
      "loss": 2.3587,
      "step": 43467
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.6315316557884216,
      "learning_rate": 0.0005488314495640624,
      "loss": 2.0755,
      "step": 43468
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7566019296646118,
      "learning_rate": 0.0005488291645588698,
      "loss": 2.2351,
      "step": 43469
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7334315776824951,
      "learning_rate": 0.0005488268795074153,
      "loss": 2.4166,
      "step": 43470
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.6939820051193237,
      "learning_rate": 0.000548824594409699,
      "loss": 2.102,
      "step": 43471
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.6972839832305908,
      "learning_rate": 0.0005488223092657218,
      "loss": 2.3132,
      "step": 43472
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7221665978431702,
      "learning_rate": 0.0005488200240754837,
      "loss": 2.2412,
      "step": 43473
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7195730805397034,
      "learning_rate": 0.0005488177388389852,
      "loss": 2.4155,
      "step": 43474
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7369786500930786,
      "learning_rate": 0.0005488154535562268,
      "loss": 2.1588,
      "step": 43475
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.6475892066955566,
      "learning_rate": 0.000548813168227209,
      "loss": 2.4256,
      "step": 43476
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7510969042778015,
      "learning_rate": 0.000548810882851932,
      "loss": 2.2511,
      "step": 43477
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7592940926551819,
      "learning_rate": 0.0005488085974303965,
      "loss": 2.2966,
      "step": 43478
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.6764962077140808,
      "learning_rate": 0.0005488063119626028,
      "loss": 2.2466,
      "step": 43479
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7171692848205566,
      "learning_rate": 0.0005488040264485511,
      "loss": 2.0708,
      "step": 43480
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.6926925182342529,
      "learning_rate": 0.0005488017408882423,
      "loss": 2.3168,
      "step": 43481
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.6857607960700989,
      "learning_rate": 0.0005487994552816764,
      "loss": 2.0478,
      "step": 43482
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.6514334678649902,
      "learning_rate": 0.0005487971696288539,
      "loss": 2.123,
      "step": 43483
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.798073947429657,
      "learning_rate": 0.0005487948839297754,
      "loss": 2.4422,
      "step": 43484
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.6451551914215088,
      "learning_rate": 0.0005487925981844413,
      "loss": 2.1012,
      "step": 43485
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7105004787445068,
      "learning_rate": 0.0005487903123928519,
      "loss": 2.2962,
      "step": 43486
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7822600603103638,
      "learning_rate": 0.0005487880265550077,
      "loss": 2.1791,
      "step": 43487
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.8799409866333008,
      "learning_rate": 0.0005487857406709091,
      "loss": 2.0896,
      "step": 43488
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.684834361076355,
      "learning_rate": 0.0005487834547405566,
      "loss": 2.3007,
      "step": 43489
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.8144212961196899,
      "learning_rate": 0.0005487811687639505,
      "loss": 2.1436,
      "step": 43490
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.8011850714683533,
      "learning_rate": 0.0005487788827410913,
      "loss": 2.0794,
      "step": 43491
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.718611478805542,
      "learning_rate": 0.0005487765966719794,
      "loss": 2.2199,
      "step": 43492
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7085124850273132,
      "learning_rate": 0.0005487743105566153,
      "loss": 2.2461,
      "step": 43493
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.8491836190223694,
      "learning_rate": 0.0005487720243949991,
      "loss": 2.3965,
      "step": 43494
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7159740924835205,
      "learning_rate": 0.0005487697381871318,
      "loss": 2.2909,
      "step": 43495
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.6568403840065002,
      "learning_rate": 0.0005487674519330134,
      "loss": 2.0906,
      "step": 43496
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.6798595786094666,
      "learning_rate": 0.0005487651656326444,
      "loss": 2.4787,
      "step": 43497
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7588481903076172,
      "learning_rate": 0.0005487628792860253,
      "loss": 2.1809,
      "step": 43498
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.720149040222168,
      "learning_rate": 0.0005487605928931565,
      "loss": 2.3522,
      "step": 43499
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.6792547702789307,
      "learning_rate": 0.0005487583064540384,
      "loss": 2.0257,
      "step": 43500
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.8364324569702148,
      "learning_rate": 0.0005487560199686713,
      "loss": 2.588,
      "step": 43501
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7562100291252136,
      "learning_rate": 0.000548753733437056,
      "loss": 2.3105,
      "step": 43502
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7155599594116211,
      "learning_rate": 0.0005487514468591925,
      "loss": 2.2833,
      "step": 43503
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.6651417016983032,
      "learning_rate": 0.0005487491602350815,
      "loss": 2.3236,
      "step": 43504
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.6437116861343384,
      "learning_rate": 0.0005487468735647233,
      "loss": 2.1471,
      "step": 43505
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.6382081508636475,
      "learning_rate": 0.0005487445868481184,
      "loss": 2.3578,
      "step": 43506
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.6986783742904663,
      "learning_rate": 0.000548742300085267,
      "loss": 2.0875,
      "step": 43507
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7712810039520264,
      "learning_rate": 0.0005487400132761699,
      "loss": 2.3049,
      "step": 43508
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7245868444442749,
      "learning_rate": 0.0005487377264208273,
      "loss": 2.3165,
      "step": 43509
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7528132200241089,
      "learning_rate": 0.0005487354395192398,
      "loss": 2.4937,
      "step": 43510
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.6993482708930969,
      "learning_rate": 0.0005487331525714075,
      "loss": 2.3047,
      "step": 43511
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7060975432395935,
      "learning_rate": 0.0005487308655773309,
      "loss": 2.3504,
      "step": 43512
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7003400921821594,
      "learning_rate": 0.0005487285785370108,
      "loss": 2.2972,
      "step": 43513
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7373392581939697,
      "learning_rate": 0.0005487262914504472,
      "loss": 2.2617,
      "step": 43514
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.8384801149368286,
      "learning_rate": 0.0005487240043176408,
      "loss": 2.2903,
      "step": 43515
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.8333895206451416,
      "learning_rate": 0.0005487217171385918,
      "loss": 2.2349,
      "step": 43516
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.6495532393455505,
      "learning_rate": 0.0005487194299133008,
      "loss": 2.2464,
      "step": 43517
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.6679736375808716,
      "learning_rate": 0.0005487171426417683,
      "loss": 2.2083,
      "step": 43518
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7096244692802429,
      "learning_rate": 0.0005487148553239943,
      "loss": 2.2122,
      "step": 43519
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.6800112128257751,
      "learning_rate": 0.0005487125679599796,
      "loss": 2.215,
      "step": 43520
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.6765493750572205,
      "learning_rate": 0.0005487102805497247,
      "loss": 2.5447,
      "step": 43521
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.6542453765869141,
      "learning_rate": 0.0005487079930932298,
      "loss": 2.3122,
      "step": 43522
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.695614755153656,
      "learning_rate": 0.0005487057055904952,
      "loss": 2.2423,
      "step": 43523
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7905808091163635,
      "learning_rate": 0.0005487034180415217,
      "loss": 2.4902,
      "step": 43524
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7365114688873291,
      "learning_rate": 0.0005487011304463096,
      "loss": 2.2741,
      "step": 43525
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7146175503730774,
      "learning_rate": 0.0005486988428048591,
      "loss": 2.1755,
      "step": 43526
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.777074933052063,
      "learning_rate": 0.0005486965551171709,
      "loss": 2.1567,
      "step": 43527
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7172577977180481,
      "learning_rate": 0.0005486942673832453,
      "loss": 2.3297,
      "step": 43528
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.845856785774231,
      "learning_rate": 0.0005486919796030827,
      "loss": 2.3436,
      "step": 43529
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7592577934265137,
      "learning_rate": 0.0005486896917766836,
      "loss": 2.6904,
      "step": 43530
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7315928936004639,
      "learning_rate": 0.0005486874039040484,
      "loss": 2.1505,
      "step": 43531
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.832169234752655,
      "learning_rate": 0.0005486851159851776,
      "loss": 2.1941,
      "step": 43532
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.6612213850021362,
      "learning_rate": 0.0005486828280200713,
      "loss": 2.1614,
      "step": 43533
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.6870533227920532,
      "learning_rate": 0.0005486805400087303,
      "loss": 2.2391,
      "step": 43534
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.6936732530593872,
      "learning_rate": 0.0005486782519511549,
      "loss": 2.2605,
      "step": 43535
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.6868476867675781,
      "learning_rate": 0.0005486759638473456,
      "loss": 2.112,
      "step": 43536
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.6566592454910278,
      "learning_rate": 0.0005486736756973026,
      "loss": 2.2582,
      "step": 43537
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.813851535320282,
      "learning_rate": 0.0005486713875010265,
      "loss": 2.463,
      "step": 43538
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.8087391257286072,
      "learning_rate": 0.0005486690992585177,
      "loss": 2.1781,
      "step": 43539
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.6928377747535706,
      "learning_rate": 0.0005486668109697767,
      "loss": 1.983,
      "step": 43540
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.6790900826454163,
      "learning_rate": 0.0005486645226348038,
      "loss": 2.2744,
      "step": 43541
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7344571948051453,
      "learning_rate": 0.0005486622342535995,
      "loss": 2.2455,
      "step": 43542
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7297144532203674,
      "learning_rate": 0.0005486599458261641,
      "loss": 2.356,
      "step": 43543
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7575849890708923,
      "learning_rate": 0.0005486576573524982,
      "loss": 2.4096,
      "step": 43544
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7804555892944336,
      "learning_rate": 0.0005486553688326022,
      "loss": 2.4221,
      "step": 43545
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7544769048690796,
      "learning_rate": 0.0005486530802664764,
      "loss": 2.358,
      "step": 43546
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.6859474182128906,
      "learning_rate": 0.0005486507916541213,
      "loss": 2.2269,
      "step": 43547
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.6487029790878296,
      "learning_rate": 0.0005486485029955373,
      "loss": 2.2907,
      "step": 43548
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.6972043514251709,
      "learning_rate": 0.0005486462142907248,
      "loss": 2.2055,
      "step": 43549
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7022216320037842,
      "learning_rate": 0.0005486439255396844,
      "loss": 2.067,
      "step": 43550
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.5961694717407227,
      "learning_rate": 0.0005486416367424164,
      "loss": 2.283,
      "step": 43551
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.6817389726638794,
      "learning_rate": 0.0005486393478989211,
      "loss": 2.2455,
      "step": 43552
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7735093235969543,
      "learning_rate": 0.0005486370590091992,
      "loss": 2.6299,
      "step": 43553
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.6375136375427246,
      "learning_rate": 0.0005486347700732509,
      "loss": 2.1538,
      "step": 43554
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7390077710151672,
      "learning_rate": 0.0005486324810910767,
      "loss": 2.3503,
      "step": 43555
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.6799494624137878,
      "learning_rate": 0.000548630192062677,
      "loss": 1.9957,
      "step": 43556
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7636147141456604,
      "learning_rate": 0.0005486279029880524,
      "loss": 2.2134,
      "step": 43557
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.6861314177513123,
      "learning_rate": 0.000548625613867203,
      "loss": 2.3166,
      "step": 43558
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7002568244934082,
      "learning_rate": 0.0005486233247001294,
      "loss": 2.5167,
      "step": 43559
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7618376612663269,
      "learning_rate": 0.0005486210354868321,
      "loss": 2.1378,
      "step": 43560
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.6384215950965881,
      "learning_rate": 0.0005486187462273116,
      "loss": 2.3211,
      "step": 43561
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.6122434139251709,
      "learning_rate": 0.000548616456921568,
      "loss": 2.0463,
      "step": 43562
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.6485782861709595,
      "learning_rate": 0.0005486141675696019,
      "loss": 2.4224,
      "step": 43563
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.6583068370819092,
      "learning_rate": 0.0005486118781714138,
      "loss": 2.3994,
      "step": 43564
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.8689704537391663,
      "learning_rate": 0.000548609588727004,
      "loss": 2.2583,
      "step": 43565
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.6760944128036499,
      "learning_rate": 0.0005486072992363731,
      "loss": 2.3243,
      "step": 43566
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.6633405089378357,
      "learning_rate": 0.0005486050096995213,
      "loss": 2.5572,
      "step": 43567
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7242544889450073,
      "learning_rate": 0.0005486027201164492,
      "loss": 2.5452,
      "step": 43568
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.6881847381591797,
      "learning_rate": 0.0005486004304871571,
      "loss": 2.2908,
      "step": 43569
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.6961220502853394,
      "learning_rate": 0.0005485981408116455,
      "loss": 2.1501,
      "step": 43570
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7219257950782776,
      "learning_rate": 0.0005485958510899149,
      "loss": 2.3431,
      "step": 43571
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.6981592774391174,
      "learning_rate": 0.0005485935613219656,
      "loss": 2.3268,
      "step": 43572
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7239639759063721,
      "learning_rate": 0.0005485912715077981,
      "loss": 2.1221,
      "step": 43573
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7441341280937195,
      "learning_rate": 0.0005485889816474127,
      "loss": 2.2122,
      "step": 43574
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.6992890238761902,
      "learning_rate": 0.0005485866917408099,
      "loss": 2.2012,
      "step": 43575
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.0433237552642822,
      "learning_rate": 0.0005485844017879902,
      "loss": 2.2825,
      "step": 43576
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.8686555027961731,
      "learning_rate": 0.000548582111788954,
      "loss": 2.259,
      "step": 43577
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.6850494742393494,
      "learning_rate": 0.0005485798217437016,
      "loss": 2.2946,
      "step": 43578
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.6992732882499695,
      "learning_rate": 0.0005485775316522336,
      "loss": 2.2624,
      "step": 43579
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.6549968719482422,
      "learning_rate": 0.0005485752415145504,
      "loss": 2.1974,
      "step": 43580
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.8242031931877136,
      "learning_rate": 0.0005485729513306523,
      "loss": 2.2295,
      "step": 43581
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7528101205825806,
      "learning_rate": 0.0005485706611005397,
      "loss": 2.3299,
      "step": 43582
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7834619283676147,
      "learning_rate": 0.0005485683708242132,
      "loss": 2.1314,
      "step": 43583
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7060104012489319,
      "learning_rate": 0.0005485660805016732,
      "loss": 2.4248,
      "step": 43584
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7520434260368347,
      "learning_rate": 0.0005485637901329199,
      "loss": 2.4669,
      "step": 43585
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7368192076683044,
      "learning_rate": 0.0005485614997179541,
      "loss": 2.3256,
      "step": 43586
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7904452085494995,
      "learning_rate": 0.000548559209256776,
      "loss": 2.067,
      "step": 43587
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.8780986070632935,
      "learning_rate": 0.0005485569187493859,
      "loss": 2.0782,
      "step": 43588
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.6971650123596191,
      "learning_rate": 0.0005485546281957846,
      "loss": 2.1957,
      "step": 43589
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.6289418935775757,
      "learning_rate": 0.0005485523375959721,
      "loss": 2.1201,
      "step": 43590
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.9537366032600403,
      "learning_rate": 0.0005485500469499491,
      "loss": 2.128,
      "step": 43591
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7541112899780273,
      "learning_rate": 0.000548547756257716,
      "loss": 2.2991,
      "step": 43592
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7727174758911133,
      "learning_rate": 0.0005485454655192731,
      "loss": 2.2823,
      "step": 43593
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7296972274780273,
      "learning_rate": 0.000548543174734621,
      "loss": 2.1529,
      "step": 43594
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.8180187344551086,
      "learning_rate": 0.0005485408839037599,
      "loss": 2.4206,
      "step": 43595
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.760329008102417,
      "learning_rate": 0.0005485385930266905,
      "loss": 2.3168,
      "step": 43596
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.6821985244750977,
      "learning_rate": 0.000548536302103413,
      "loss": 2.3023,
      "step": 43597
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7388675808906555,
      "learning_rate": 0.0005485340111339279,
      "loss": 2.3213,
      "step": 43598
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7534734010696411,
      "learning_rate": 0.0005485317201182357,
      "loss": 2.2208,
      "step": 43599
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.6460971236228943,
      "learning_rate": 0.0005485294290563368,
      "loss": 2.2194,
      "step": 43600
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.6858938336372375,
      "learning_rate": 0.0005485271379482314,
      "loss": 2.1574,
      "step": 43601
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.77028888463974,
      "learning_rate": 0.0005485248467939204,
      "loss": 2.2275,
      "step": 43602
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.0284541845321655,
      "learning_rate": 0.0005485225555934037,
      "loss": 2.0848,
      "step": 43603
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7131406664848328,
      "learning_rate": 0.0005485202643466821,
      "loss": 2.1119,
      "step": 43604
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7354188561439514,
      "learning_rate": 0.0005485179730537557,
      "loss": 2.3526,
      "step": 43605
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7164523601531982,
      "learning_rate": 0.0005485156817146253,
      "loss": 2.1825,
      "step": 43606
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.6862111687660217,
      "learning_rate": 0.0005485133903292911,
      "loss": 2.4323,
      "step": 43607
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7682908177375793,
      "learning_rate": 0.0005485110988977536,
      "loss": 2.1347,
      "step": 43608
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7318482995033264,
      "learning_rate": 0.0005485088074200131,
      "loss": 2.1662,
      "step": 43609
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.8234038949012756,
      "learning_rate": 0.0005485065158960701,
      "loss": 2.4972,
      "step": 43610
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.763702392578125,
      "learning_rate": 0.0005485042243259252,
      "loss": 2.2428,
      "step": 43611
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.669494092464447,
      "learning_rate": 0.0005485019327095786,
      "loss": 2.0961,
      "step": 43612
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.6399058699607849,
      "learning_rate": 0.0005484996410470308,
      "loss": 2.1602,
      "step": 43613
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.6908183097839355,
      "learning_rate": 0.0005484973493382822,
      "loss": 2.2878,
      "step": 43614
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7343409657478333,
      "learning_rate": 0.0005484950575833333,
      "loss": 2.3034,
      "step": 43615
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.6613020300865173,
      "learning_rate": 0.0005484927657821844,
      "loss": 2.1244,
      "step": 43616
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7652273178100586,
      "learning_rate": 0.0005484904739348361,
      "loss": 2.1622,
      "step": 43617
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.6934854984283447,
      "learning_rate": 0.0005484881820412887,
      "loss": 2.1308,
      "step": 43618
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.6700518727302551,
      "learning_rate": 0.0005484858901015426,
      "loss": 2.1482,
      "step": 43619
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7103755474090576,
      "learning_rate": 0.0005484835981155984,
      "loss": 2.3442,
      "step": 43620
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7302517294883728,
      "learning_rate": 0.0005484813060834564,
      "loss": 2.2937,
      "step": 43621
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.6711630821228027,
      "learning_rate": 0.0005484790140051169,
      "loss": 2.2207,
      "step": 43622
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7116839289665222,
      "learning_rate": 0.0005484767218805805,
      "loss": 2.2586,
      "step": 43623
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7194367051124573,
      "learning_rate": 0.0005484744297098477,
      "loss": 2.1304,
      "step": 43624
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7298157811164856,
      "learning_rate": 0.0005484721374929187,
      "loss": 2.331,
      "step": 43625
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7578105330467224,
      "learning_rate": 0.0005484698452297941,
      "loss": 2.2898,
      "step": 43626
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.6862451434135437,
      "learning_rate": 0.0005484675529204742,
      "loss": 2.2708,
      "step": 43627
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.6789001822471619,
      "learning_rate": 0.0005484652605649595,
      "loss": 2.0636,
      "step": 43628
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.6910184621810913,
      "learning_rate": 0.0005484629681632505,
      "loss": 2.1771,
      "step": 43629
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7033625841140747,
      "learning_rate": 0.0005484606757153475,
      "loss": 2.3399,
      "step": 43630
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7703083753585815,
      "learning_rate": 0.000548458383221251,
      "loss": 2.2341,
      "step": 43631
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.6888644099235535,
      "learning_rate": 0.0005484560906809612,
      "loss": 2.3316,
      "step": 43632
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.6794075965881348,
      "learning_rate": 0.000548453798094479,
      "loss": 2.041,
      "step": 43633
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.6855987906455994,
      "learning_rate": 0.0005484515054618044,
      "loss": 2.306,
      "step": 43634
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.6726277470588684,
      "learning_rate": 0.000548449212782938,
      "loss": 2.3281,
      "step": 43635
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7188710570335388,
      "learning_rate": 0.0005484469200578802,
      "loss": 2.4386,
      "step": 43636
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7015278339385986,
      "learning_rate": 0.0005484446272866315,
      "loss": 2.1578,
      "step": 43637
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7558027505874634,
      "learning_rate": 0.0005484423344691921,
      "loss": 2.3704,
      "step": 43638
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.6975756883621216,
      "learning_rate": 0.0005484400416055627,
      "loss": 2.2707,
      "step": 43639
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.6616050004959106,
      "learning_rate": 0.0005484377486957436,
      "loss": 2.354,
      "step": 43640
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.6970941424369812,
      "learning_rate": 0.0005484354557397352,
      "loss": 2.2238,
      "step": 43641
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.6921879053115845,
      "learning_rate": 0.0005484331627375379,
      "loss": 2.3651,
      "step": 43642
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.6919475197792053,
      "learning_rate": 0.0005484308696891523,
      "loss": 2.1341,
      "step": 43643
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.6976680159568787,
      "learning_rate": 0.0005484285765945787,
      "loss": 2.2423,
      "step": 43644
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7768071293830872,
      "learning_rate": 0.0005484262834538176,
      "loss": 2.1862,
      "step": 43645
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.6875760555267334,
      "learning_rate": 0.0005484239902668692,
      "loss": 2.2853,
      "step": 43646
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7117722034454346,
      "learning_rate": 0.0005484216970337341,
      "loss": 2.1336,
      "step": 43647
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.74592125415802,
      "learning_rate": 0.0005484194037544128,
      "loss": 2.2933,
      "step": 43648
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7029221057891846,
      "learning_rate": 0.0005484171104289056,
      "loss": 2.369,
      "step": 43649
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.735837996006012,
      "learning_rate": 0.000548414817057213,
      "loss": 2.2746,
      "step": 43650
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.6907771229743958,
      "learning_rate": 0.0005484125236393354,
      "loss": 2.2757,
      "step": 43651
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7616079449653625,
      "learning_rate": 0.0005484102301752732,
      "loss": 2.3935,
      "step": 43652
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.9408535957336426,
      "learning_rate": 0.0005484079366650269,
      "loss": 2.3626,
      "step": 43653
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.687009871006012,
      "learning_rate": 0.0005484056431085968,
      "loss": 2.0793,
      "step": 43654
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7685450911521912,
      "learning_rate": 0.0005484033495059834,
      "loss": 2.3225,
      "step": 43655
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7504435777664185,
      "learning_rate": 0.0005484010558571872,
      "loss": 1.9984,
      "step": 43656
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7848415374755859,
      "learning_rate": 0.0005483987621622085,
      "loss": 2.2413,
      "step": 43657
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7338558435440063,
      "learning_rate": 0.0005483964684210479,
      "loss": 2.1361,
      "step": 43658
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7310439944267273,
      "learning_rate": 0.0005483941746337055,
      "loss": 2.2036,
      "step": 43659
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7289137840270996,
      "learning_rate": 0.0005483918808001821,
      "loss": 2.4299,
      "step": 43660
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.8565520644187927,
      "learning_rate": 0.000548389586920478,
      "loss": 2.4511,
      "step": 43661
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.6300093531608582,
      "learning_rate": 0.0005483872929945933,
      "loss": 2.0881,
      "step": 43662
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7253202795982361,
      "learning_rate": 0.0005483849990225289,
      "loss": 2.0358,
      "step": 43663
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.9902393817901611,
      "learning_rate": 0.0005483827050042851,
      "loss": 2.0399,
      "step": 43664
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.6971258521080017,
      "learning_rate": 0.0005483804109398622,
      "loss": 2.2445,
      "step": 43665
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.8195917010307312,
      "learning_rate": 0.0005483781168292606,
      "loss": 2.221,
      "step": 43666
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7581101655960083,
      "learning_rate": 0.0005483758226724811,
      "loss": 1.9763,
      "step": 43667
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7686347365379333,
      "learning_rate": 0.0005483735284695236,
      "loss": 2.2272,
      "step": 43668
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7299849390983582,
      "learning_rate": 0.0005483712342203888,
      "loss": 2.2775,
      "step": 43669
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.9130483269691467,
      "learning_rate": 0.0005483689399250771,
      "loss": 2.3875,
      "step": 43670
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7775365114212036,
      "learning_rate": 0.0005483666455835891,
      "loss": 2.2574,
      "step": 43671
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.718859076499939,
      "learning_rate": 0.000548364351195925,
      "loss": 2.1821,
      "step": 43672
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.6680192947387695,
      "learning_rate": 0.0005483620567620851,
      "loss": 2.1238,
      "step": 43673
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.6989573240280151,
      "learning_rate": 0.0005483597622820701,
      "loss": 1.9784,
      "step": 43674
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7786949872970581,
      "learning_rate": 0.0005483574677558803,
      "loss": 2.2788,
      "step": 43675
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.6862706542015076,
      "learning_rate": 0.0005483551731835162,
      "loss": 2.24,
      "step": 43676
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7131536602973938,
      "learning_rate": 0.0005483528785649783,
      "loss": 2.3618,
      "step": 43677
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7018651962280273,
      "learning_rate": 0.0005483505839002666,
      "loss": 2.2135,
      "step": 43678
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.6554352045059204,
      "learning_rate": 0.0005483482891893821,
      "loss": 2.4458,
      "step": 43679
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7403125762939453,
      "learning_rate": 0.000548345994432325,
      "loss": 2.2797,
      "step": 43680
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7441144585609436,
      "learning_rate": 0.0005483436996290955,
      "loss": 2.3459,
      "step": 43681
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7286447882652283,
      "learning_rate": 0.0005483414047796944,
      "loss": 2.1805,
      "step": 43682
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.755803108215332,
      "learning_rate": 0.0005483391098841218,
      "loss": 2.1535,
      "step": 43683
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.6955599188804626,
      "learning_rate": 0.0005483368149423783,
      "loss": 2.304,
      "step": 43684
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.6916868090629578,
      "learning_rate": 0.0005483345199544643,
      "loss": 2.2927,
      "step": 43685
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.6438132524490356,
      "learning_rate": 0.0005483322249203803,
      "loss": 2.2162,
      "step": 43686
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7063099145889282,
      "learning_rate": 0.0005483299298401265,
      "loss": 2.1492,
      "step": 43687
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7579742074012756,
      "learning_rate": 0.0005483276347137036,
      "loss": 2.3206,
      "step": 43688
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.6718144416809082,
      "learning_rate": 0.0005483253395411119,
      "loss": 2.3831,
      "step": 43689
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.6812702417373657,
      "learning_rate": 0.0005483230443223519,
      "loss": 2.0829,
      "step": 43690
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.6598026156425476,
      "learning_rate": 0.0005483207490574238,
      "loss": 2.2217,
      "step": 43691
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.6707660555839539,
      "learning_rate": 0.0005483184537463283,
      "loss": 2.1531,
      "step": 43692
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7461099028587341,
      "learning_rate": 0.0005483161583890656,
      "loss": 2.3097,
      "step": 43693
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.6742525100708008,
      "learning_rate": 0.0005483138629856363,
      "loss": 2.31,
      "step": 43694
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.6926015019416809,
      "learning_rate": 0.0005483115675360408,
      "loss": 2.1948,
      "step": 43695
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7289026975631714,
      "learning_rate": 0.0005483092720402795,
      "loss": 2.1195,
      "step": 43696
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.6795610785484314,
      "learning_rate": 0.0005483069764983527,
      "loss": 2.1713,
      "step": 43697
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.6918401718139648,
      "learning_rate": 0.000548304680910261,
      "loss": 2.3321,
      "step": 43698
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7254922986030579,
      "learning_rate": 0.0005483023852760049,
      "loss": 2.0658,
      "step": 43699
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7787299752235413,
      "learning_rate": 0.0005483000895955846,
      "loss": 2.5587,
      "step": 43700
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.658594012260437,
      "learning_rate": 0.0005482977938690006,
      "loss": 2.1999,
      "step": 43701
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.733089029788971,
      "learning_rate": 0.0005482954980962533,
      "loss": 2.3485,
      "step": 43702
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7026514410972595,
      "learning_rate": 0.0005482932022773434,
      "loss": 2.4286,
      "step": 43703
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.6857211589813232,
      "learning_rate": 0.0005482909064122709,
      "loss": 2.2355,
      "step": 43704
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.738292396068573,
      "learning_rate": 0.0005482886105010365,
      "loss": 2.2287,
      "step": 43705
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7355906367301941,
      "learning_rate": 0.0005482863145436405,
      "loss": 2.2596,
      "step": 43706
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7179631590843201,
      "learning_rate": 0.0005482840185400835,
      "loss": 2.3089,
      "step": 43707
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.6644614934921265,
      "learning_rate": 0.0005482817224903658,
      "loss": 2.3307,
      "step": 43708
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.6743863224983215,
      "learning_rate": 0.0005482794263944878,
      "loss": 2.1782,
      "step": 43709
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7072059512138367,
      "learning_rate": 0.0005482771302524499,
      "loss": 2.0944,
      "step": 43710
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7055342793464661,
      "learning_rate": 0.0005482748340642527,
      "loss": 2.1948,
      "step": 43711
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.670866072177887,
      "learning_rate": 0.0005482725378298964,
      "loss": 2.2755,
      "step": 43712
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.6864045858383179,
      "learning_rate": 0.0005482702415493817,
      "loss": 1.9515,
      "step": 43713
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7375550866127014,
      "learning_rate": 0.0005482679452227089,
      "loss": 2.2117,
      "step": 43714
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.6986312866210938,
      "learning_rate": 0.0005482656488498783,
      "loss": 2.1153,
      "step": 43715
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7124404311180115,
      "learning_rate": 0.0005482633524308905,
      "loss": 2.3149,
      "step": 43716
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.6407338380813599,
      "learning_rate": 0.0005482610559657457,
      "loss": 2.2378,
      "step": 43717
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7054151892662048,
      "learning_rate": 0.0005482587594544447,
      "loss": 2.1962,
      "step": 43718
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.6725289225578308,
      "learning_rate": 0.0005482564628969876,
      "loss": 2.235,
      "step": 43719
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.6569308638572693,
      "learning_rate": 0.0005482541662933748,
      "loss": 2.1521,
      "step": 43720
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.729194700717926,
      "learning_rate": 0.000548251869643607,
      "loss": 2.1541,
      "step": 43721
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7236759662628174,
      "learning_rate": 0.0005482495729476845,
      "loss": 2.3696,
      "step": 43722
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.6759232878684998,
      "learning_rate": 0.0005482472762056077,
      "loss": 2.3348,
      "step": 43723
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7134134769439697,
      "learning_rate": 0.0005482449794173771,
      "loss": 2.2291,
      "step": 43724
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.658962070941925,
      "learning_rate": 0.000548242682582993,
      "loss": 2.2341,
      "step": 43725
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.6677605509757996,
      "learning_rate": 0.0005482403857024559,
      "loss": 2.2785,
      "step": 43726
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7052050232887268,
      "learning_rate": 0.0005482380887757663,
      "loss": 2.3419,
      "step": 43727
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.6524079442024231,
      "learning_rate": 0.0005482357918029245,
      "loss": 2.242,
      "step": 43728
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7539637088775635,
      "learning_rate": 0.000548233494783931,
      "loss": 2.2342,
      "step": 43729
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.6799604892730713,
      "learning_rate": 0.0005482311977187862,
      "loss": 2.3853,
      "step": 43730
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7153341174125671,
      "learning_rate": 0.0005482289006074906,
      "loss": 2.1383,
      "step": 43731
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7211261987686157,
      "learning_rate": 0.0005482266034500445,
      "loss": 2.0792,
      "step": 43732
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7818402051925659,
      "learning_rate": 0.0005482243062464483,
      "loss": 2.2823,
      "step": 43733
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.6492633819580078,
      "learning_rate": 0.0005482220089967026,
      "loss": 2.3539,
      "step": 43734
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.6556297540664673,
      "learning_rate": 0.0005482197117008076,
      "loss": 2.2987,
      "step": 43735
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.702178955078125,
      "learning_rate": 0.0005482174143587641,
      "loss": 2.0794,
      "step": 43736
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7052367329597473,
      "learning_rate": 0.0005482151169705722,
      "loss": 2.0496,
      "step": 43737
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7000748515129089,
      "learning_rate": 0.0005482128195362325,
      "loss": 2.235,
      "step": 43738
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.649867832660675,
      "learning_rate": 0.0005482105220557452,
      "loss": 2.097,
      "step": 43739
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7753893733024597,
      "learning_rate": 0.000548208224529111,
      "loss": 2.2011,
      "step": 43740
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7443494200706482,
      "learning_rate": 0.0005482059269563302,
      "loss": 2.3542,
      "step": 43741
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7459290027618408,
      "learning_rate": 0.0005482036293374032,
      "loss": 2.2939,
      "step": 43742
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7946731448173523,
      "learning_rate": 0.0005482013316723304,
      "loss": 2.4301,
      "step": 43743
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.6698272228240967,
      "learning_rate": 0.0005481990339611123,
      "loss": 2.0956,
      "step": 43744
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7157350182533264,
      "learning_rate": 0.0005481967362037494,
      "loss": 2.217,
      "step": 43745
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7142696976661682,
      "learning_rate": 0.000548194438400242,
      "loss": 2.2855,
      "step": 43746
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7383583188056946,
      "learning_rate": 0.0005481921405505906,
      "loss": 2.0325,
      "step": 43747
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7197808027267456,
      "learning_rate": 0.0005481898426547956,
      "loss": 2.5404,
      "step": 43748
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.6948209404945374,
      "learning_rate": 0.0005481875447128573,
      "loss": 2.2494,
      "step": 43749
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7095921635627747,
      "learning_rate": 0.0005481852467247763,
      "loss": 2.103,
      "step": 43750
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.8306027054786682,
      "learning_rate": 0.0005481829486905531,
      "loss": 2.3637,
      "step": 43751
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7227576971054077,
      "learning_rate": 0.0005481806506101879,
      "loss": 2.1948,
      "step": 43752
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.6588733792304993,
      "learning_rate": 0.0005481783524836812,
      "loss": 2.1562,
      "step": 43753
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.9525344967842102,
      "learning_rate": 0.0005481760543110336,
      "loss": 2.2361,
      "step": 43754
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7048699259757996,
      "learning_rate": 0.0005481737560922453,
      "loss": 2.2636,
      "step": 43755
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7939780354499817,
      "learning_rate": 0.0005481714578273168,
      "loss": 2.3916,
      "step": 43756
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.6759689450263977,
      "learning_rate": 0.0005481691595162486,
      "loss": 2.2275,
      "step": 43757
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.6567614078521729,
      "learning_rate": 0.000548166861159041,
      "loss": 2.3129,
      "step": 43758
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7398102879524231,
      "learning_rate": 0.0005481645627556945,
      "loss": 2.4254,
      "step": 43759
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.6943357586860657,
      "learning_rate": 0.0005481622643062096,
      "loss": 2.2008,
      "step": 43760
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7624650001525879,
      "learning_rate": 0.0005481599658105864,
      "loss": 2.1784,
      "step": 43761
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7358159422874451,
      "learning_rate": 0.0005481576672688259,
      "loss": 2.2165,
      "step": 43762
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7459999322891235,
      "learning_rate": 0.0005481553686809281,
      "loss": 2.4153,
      "step": 43763
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.6715757250785828,
      "learning_rate": 0.0005481530700468935,
      "loss": 2.413,
      "step": 43764
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.6342846751213074,
      "learning_rate": 0.0005481507713667225,
      "loss": 2.1082,
      "step": 43765
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.8203867077827454,
      "learning_rate": 0.0005481484726404158,
      "loss": 2.3296,
      "step": 43766
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.8003010749816895,
      "learning_rate": 0.0005481461738679734,
      "loss": 2.2417,
      "step": 43767
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7186444401741028,
      "learning_rate": 0.000548143875049396,
      "loss": 2.2153,
      "step": 43768
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7363731861114502,
      "learning_rate": 0.000548141576184684,
      "loss": 2.2474,
      "step": 43769
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7436690926551819,
      "learning_rate": 0.0005481392772738378,
      "loss": 2.1858,
      "step": 43770
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.654177188873291,
      "learning_rate": 0.0005481369783168578,
      "loss": 2.3611,
      "step": 43771
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.754277765750885,
      "learning_rate": 0.0005481346793137445,
      "loss": 2.398,
      "step": 43772
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7417693734169006,
      "learning_rate": 0.0005481323802644982,
      "loss": 2.3553,
      "step": 43773
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7381545901298523,
      "learning_rate": 0.0005481300811691195,
      "loss": 2.4653,
      "step": 43774
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.8703324794769287,
      "learning_rate": 0.0005481277820276086,
      "loss": 2.3257,
      "step": 43775
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.705100953578949,
      "learning_rate": 0.0005481254828399661,
      "loss": 2.4839,
      "step": 43776
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7043420076370239,
      "learning_rate": 0.0005481231836061925,
      "loss": 2.0808,
      "step": 43777
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.8709792494773865,
      "learning_rate": 0.000548120884326288,
      "loss": 2.3939,
      "step": 43778
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7230272889137268,
      "learning_rate": 0.0005481185850002532,
      "loss": 2.4354,
      "step": 43779
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.632402777671814,
      "learning_rate": 0.0005481162856280884,
      "loss": 1.9242,
      "step": 43780
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.0017493963241577,
      "learning_rate": 0.0005481139862097943,
      "loss": 2.1029,
      "step": 43781
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.9532985687255859,
      "learning_rate": 0.000548111686745371,
      "loss": 2.2154,
      "step": 43782
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7188363075256348,
      "learning_rate": 0.0005481093872348189,
      "loss": 2.2756,
      "step": 43783
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.8938896656036377,
      "learning_rate": 0.0005481070876781388,
      "loss": 2.2718,
      "step": 43784
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.9092408418655396,
      "learning_rate": 0.0005481047880753308,
      "loss": 2.5227,
      "step": 43785
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7035336494445801,
      "learning_rate": 0.0005481024884263954,
      "loss": 2.5225,
      "step": 43786
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7064115405082703,
      "learning_rate": 0.0005481001887313331,
      "loss": 2.1935,
      "step": 43787
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7779688239097595,
      "learning_rate": 0.0005480978889901444,
      "loss": 2.4314,
      "step": 43788
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.070915937423706,
      "learning_rate": 0.0005480955892028296,
      "loss": 2.3004,
      "step": 43789
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7131266593933105,
      "learning_rate": 0.0005480932893693891,
      "loss": 2.1874,
      "step": 43790
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7959633469581604,
      "learning_rate": 0.0005480909894898233,
      "loss": 2.1868,
      "step": 43791
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.193036675453186,
      "learning_rate": 0.0005480886895641328,
      "loss": 2.2545,
      "step": 43792
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.9224624037742615,
      "learning_rate": 0.0005480863895923179,
      "loss": 2.1098,
      "step": 43793
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7898185849189758,
      "learning_rate": 0.0005480840895743791,
      "loss": 2.4268,
      "step": 43794
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.0200598239898682,
      "learning_rate": 0.0005480817895103167,
      "loss": 2.2603,
      "step": 43795
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.9615764617919922,
      "learning_rate": 0.0005480794894001312,
      "loss": 2.1445,
      "step": 43796
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7184892892837524,
      "learning_rate": 0.0005480771892438231,
      "loss": 2.0579,
      "step": 43797
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.717556893825531,
      "learning_rate": 0.0005480748890413928,
      "loss": 2.5178,
      "step": 43798
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.8911111950874329,
      "learning_rate": 0.0005480725887928407,
      "loss": 2.0235,
      "step": 43799
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.8932613134384155,
      "learning_rate": 0.0005480702884981671,
      "loss": 2.2774,
      "step": 43800
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.6680035591125488,
      "learning_rate": 0.0005480679881573726,
      "loss": 2.3831,
      "step": 43801
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.8447600603103638,
      "learning_rate": 0.0005480656877704576,
      "loss": 2.3382,
      "step": 43802
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.8268601894378662,
      "learning_rate": 0.0005480633873374226,
      "loss": 2.148,
      "step": 43803
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.8588496446609497,
      "learning_rate": 0.0005480610868582677,
      "loss": 2.3277,
      "step": 43804
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.6536086201667786,
      "learning_rate": 0.0005480587863329938,
      "loss": 2.1645,
      "step": 43805
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.678163468837738,
      "learning_rate": 0.000548056485761601,
      "loss": 2.1779,
      "step": 43806
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.8453348875045776,
      "learning_rate": 0.0005480541851440898,
      "loss": 2.1863,
      "step": 43807
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7370221614837646,
      "learning_rate": 0.0005480518844804606,
      "loss": 2.2912,
      "step": 43808
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7401001453399658,
      "learning_rate": 0.000548049583770714,
      "loss": 2.1428,
      "step": 43809
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7239211201667786,
      "learning_rate": 0.0005480472830148502,
      "loss": 2.3092,
      "step": 43810
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.8650437593460083,
      "learning_rate": 0.0005480449822128698,
      "loss": 2.2044,
      "step": 43811
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.6700002551078796,
      "learning_rate": 0.0005480426813647731,
      "loss": 2.2172,
      "step": 43812
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.8552078008651733,
      "learning_rate": 0.0005480403804705606,
      "loss": 2.2447,
      "step": 43813
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.9461405277252197,
      "learning_rate": 0.0005480380795302327,
      "loss": 2.2464,
      "step": 43814
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.8032853603363037,
      "learning_rate": 0.0005480357785437899,
      "loss": 2.3449,
      "step": 43815
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.8234861493110657,
      "learning_rate": 0.0005480334775112324,
      "loss": 2.2939,
      "step": 43816
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.764678418636322,
      "learning_rate": 0.000548031176432561,
      "loss": 2.4807,
      "step": 43817
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.653683066368103,
      "learning_rate": 0.0005480288753077757,
      "loss": 2.3227,
      "step": 43818
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.6789129972457886,
      "learning_rate": 0.0005480265741368773,
      "loss": 2.0642,
      "step": 43819
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7278159856796265,
      "learning_rate": 0.0005480242729198661,
      "loss": 2.4029,
      "step": 43820
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7742688059806824,
      "learning_rate": 0.0005480219716567425,
      "loss": 2.3478,
      "step": 43821
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.6407052278518677,
      "learning_rate": 0.0005480196703475068,
      "loss": 2.3613,
      "step": 43822
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7905930280685425,
      "learning_rate": 0.0005480173689921597,
      "loss": 2.3216,
      "step": 43823
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.616984486579895,
      "learning_rate": 0.0005480150675907015,
      "loss": 2.0439,
      "step": 43824
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.728094220161438,
      "learning_rate": 0.0005480127661431325,
      "loss": 2.3481,
      "step": 43825
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.758892297744751,
      "learning_rate": 0.0005480104646494533,
      "loss": 2.3258,
      "step": 43826
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7444454431533813,
      "learning_rate": 0.0005480081631096643,
      "loss": 2.46,
      "step": 43827
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.6733790040016174,
      "learning_rate": 0.0005480058615237659,
      "loss": 2.2514,
      "step": 43828
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7090297341346741,
      "learning_rate": 0.0005480035598917585,
      "loss": 2.3682,
      "step": 43829
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7903578877449036,
      "learning_rate": 0.0005480012582136425,
      "loss": 2.1999,
      "step": 43830
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.6599437594413757,
      "learning_rate": 0.0005479989564894185,
      "loss": 2.166,
      "step": 43831
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7566716074943542,
      "learning_rate": 0.0005479966547190867,
      "loss": 2.2389,
      "step": 43832
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7178643345832825,
      "learning_rate": 0.0005479943529026479,
      "loss": 2.1997,
      "step": 43833
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7668174505233765,
      "learning_rate": 0.000547992051040102,
      "loss": 2.4564,
      "step": 43834
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7352849841117859,
      "learning_rate": 0.0005479897491314497,
      "loss": 2.0981,
      "step": 43835
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7823075652122498,
      "learning_rate": 0.0005479874471766915,
      "loss": 2.3253,
      "step": 43836
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7465499639511108,
      "learning_rate": 0.0005479851451758277,
      "loss": 2.1725,
      "step": 43837
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.8047687411308289,
      "learning_rate": 0.0005479828431288589,
      "loss": 2.2525,
      "step": 43838
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.6659645438194275,
      "learning_rate": 0.0005479805410357853,
      "loss": 2.0872,
      "step": 43839
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.672110915184021,
      "learning_rate": 0.0005479782388966075,
      "loss": 2.0659,
      "step": 43840
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.6316168308258057,
      "learning_rate": 0.0005479759367113259,
      "loss": 2.1839,
      "step": 43841
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7209184765815735,
      "learning_rate": 0.0005479736344799408,
      "loss": 2.2985,
      "step": 43842
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.6972655653953552,
      "learning_rate": 0.0005479713322024528,
      "loss": 2.3286,
      "step": 43843
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7552549242973328,
      "learning_rate": 0.0005479690298788622,
      "loss": 2.3207,
      "step": 43844
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7005817890167236,
      "learning_rate": 0.0005479667275091694,
      "loss": 2.2176,
      "step": 43845
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7097814083099365,
      "learning_rate": 0.0005479644250933751,
      "loss": 2.1901,
      "step": 43846
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.6462945342063904,
      "learning_rate": 0.0005479621226314793,
      "loss": 2.5021,
      "step": 43847
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7386767864227295,
      "learning_rate": 0.0005479598201234828,
      "loss": 2.132,
      "step": 43848
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.6893935799598694,
      "learning_rate": 0.0005479575175693859,
      "loss": 2.3352,
      "step": 43849
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7165333032608032,
      "learning_rate": 0.0005479552149691889,
      "loss": 2.314,
      "step": 43850
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.6787026524543762,
      "learning_rate": 0.0005479529123228925,
      "loss": 2.3151,
      "step": 43851
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7006615400314331,
      "learning_rate": 0.0005479506096304969,
      "loss": 2.415,
      "step": 43852
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.6614544987678528,
      "learning_rate": 0.0005479483068920026,
      "loss": 2.1242,
      "step": 43853
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.6545891165733337,
      "learning_rate": 0.00054794600410741,
      "loss": 2.2167,
      "step": 43854
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7469241619110107,
      "learning_rate": 0.0005479437012767196,
      "loss": 2.2723,
      "step": 43855
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.6360827088356018,
      "learning_rate": 0.0005479413983999318,
      "loss": 2.1807,
      "step": 43856
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.824631929397583,
      "learning_rate": 0.000547939095477047,
      "loss": 2.1618,
      "step": 43857
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.6769078373908997,
      "learning_rate": 0.0005479367925080655,
      "loss": 2.3714,
      "step": 43858
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7033084630966187,
      "learning_rate": 0.0005479344894929881,
      "loss": 2.3764,
      "step": 43859
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.670628547668457,
      "learning_rate": 0.0005479321864318149,
      "loss": 2.2498,
      "step": 43860
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7490832209587097,
      "learning_rate": 0.0005479298833245464,
      "loss": 2.2377,
      "step": 43861
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7275843620300293,
      "learning_rate": 0.0005479275801711832,
      "loss": 2.2777,
      "step": 43862
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.6624696254730225,
      "learning_rate": 0.0005479252769717254,
      "loss": 2.2823,
      "step": 43863
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7509506940841675,
      "learning_rate": 0.0005479229737261736,
      "loss": 2.1978,
      "step": 43864
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.8415368795394897,
      "learning_rate": 0.0005479206704345283,
      "loss": 2.3676,
      "step": 43865
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.658279299736023,
      "learning_rate": 0.0005479183670967899,
      "loss": 2.16,
      "step": 43866
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7043430209159851,
      "learning_rate": 0.0005479160637129588,
      "loss": 2.1879,
      "step": 43867
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.6899982690811157,
      "learning_rate": 0.0005479137602830353,
      "loss": 2.248,
      "step": 43868
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.773283839225769,
      "learning_rate": 0.0005479114568070202,
      "loss": 2.4798,
      "step": 43869
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7023143768310547,
      "learning_rate": 0.0005479091532849135,
      "loss": 2.3215,
      "step": 43870
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.6897959113121033,
      "learning_rate": 0.0005479068497167157,
      "loss": 2.435,
      "step": 43871
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.6420837640762329,
      "learning_rate": 0.0005479045461024275,
      "loss": 2.2914,
      "step": 43872
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.681918740272522,
      "learning_rate": 0.0005479022424420493,
      "loss": 2.1847,
      "step": 43873
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.6913795471191406,
      "learning_rate": 0.0005478999387355812,
      "loss": 2.0698,
      "step": 43874
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.6263729333877563,
      "learning_rate": 0.0005478976349830238,
      "loss": 2.1369,
      "step": 43875
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.647783637046814,
      "learning_rate": 0.0005478953311843776,
      "loss": 2.2586,
      "step": 43876
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.6528430581092834,
      "learning_rate": 0.000547893027339643,
      "loss": 1.9858,
      "step": 43877
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7765883803367615,
      "learning_rate": 0.0005478907234488204,
      "loss": 2.2094,
      "step": 43878
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.6722115874290466,
      "learning_rate": 0.0005478884195119102,
      "loss": 2.1512,
      "step": 43879
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7683990001678467,
      "learning_rate": 0.0005478861155289129,
      "loss": 2.2387,
      "step": 43880
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7129387259483337,
      "learning_rate": 0.0005478838114998289,
      "loss": 2.0819,
      "step": 43881
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7443661689758301,
      "learning_rate": 0.0005478815074246586,
      "loss": 2.2267,
      "step": 43882
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.6556043028831482,
      "learning_rate": 0.0005478792033034025,
      "loss": 2.24,
      "step": 43883
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7659624814987183,
      "learning_rate": 0.0005478768991360608,
      "loss": 2.2581,
      "step": 43884
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7204155921936035,
      "learning_rate": 0.0005478745949226343,
      "loss": 2.0505,
      "step": 43885
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.732404887676239,
      "learning_rate": 0.0005478722906631233,
      "loss": 2.2071,
      "step": 43886
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.6873507499694824,
      "learning_rate": 0.000547869986357528,
      "loss": 2.4796,
      "step": 43887
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7089651823043823,
      "learning_rate": 0.000547867682005849,
      "loss": 2.3211,
      "step": 43888
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.6763629913330078,
      "learning_rate": 0.0005478653776080866,
      "loss": 2.3218,
      "step": 43889
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.6957488656044006,
      "learning_rate": 0.0005478630731642416,
      "loss": 2.2286,
      "step": 43890
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7353280186653137,
      "learning_rate": 0.000547860768674314,
      "loss": 2.2126,
      "step": 43891
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.6683446168899536,
      "learning_rate": 0.0005478584641383045,
      "loss": 2.2467,
      "step": 43892
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7175434827804565,
      "learning_rate": 0.0005478561595562134,
      "loss": 2.0985,
      "step": 43893
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7265470623970032,
      "learning_rate": 0.0005478538549280412,
      "loss": 2.3395,
      "step": 43894
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.6992363333702087,
      "learning_rate": 0.0005478515502537882,
      "loss": 2.2263,
      "step": 43895
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7195507884025574,
      "learning_rate": 0.000547849245533455,
      "loss": 2.5337,
      "step": 43896
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.6374915242195129,
      "learning_rate": 0.0005478469407670419,
      "loss": 2.1973,
      "step": 43897
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.6386519074440002,
      "learning_rate": 0.0005478446359545495,
      "loss": 2.3548,
      "step": 43898
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.6481034755706787,
      "learning_rate": 0.000547842331095978,
      "loss": 2.1859,
      "step": 43899
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.6499325633049011,
      "learning_rate": 0.0005478400261913279,
      "loss": 2.0433,
      "step": 43900
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7208526730537415,
      "learning_rate": 0.0005478377212405996,
      "loss": 2.3162,
      "step": 43901
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.5925438404083252,
      "learning_rate": 0.0005478354162437937,
      "loss": 2.328,
      "step": 43902
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7008898854255676,
      "learning_rate": 0.0005478331112009105,
      "loss": 2.1138,
      "step": 43903
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.682518720626831,
      "learning_rate": 0.0005478308061119504,
      "loss": 2.3501,
      "step": 43904
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.6486510634422302,
      "learning_rate": 0.000547828500976914,
      "loss": 2.2551,
      "step": 43905
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7120299935340881,
      "learning_rate": 0.0005478261957958014,
      "loss": 2.3962,
      "step": 43906
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7091740369796753,
      "learning_rate": 0.0005478238905686134,
      "loss": 2.3381,
      "step": 43907
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7197520732879639,
      "learning_rate": 0.0005478215852953503,
      "loss": 2.467,
      "step": 43908
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.6142759323120117,
      "learning_rate": 0.0005478192799760124,
      "loss": 2.4313,
      "step": 43909
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.6746588349342346,
      "learning_rate": 0.0005478169746106001,
      "loss": 2.3756,
      "step": 43910
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7303913831710815,
      "learning_rate": 0.0005478146691991142,
      "loss": 2.2145,
      "step": 43911
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.6958669424057007,
      "learning_rate": 0.0005478123637415546,
      "loss": 2.1556,
      "step": 43912
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7306602597236633,
      "learning_rate": 0.0005478100582379222,
      "loss": 2.2644,
      "step": 43913
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7257452607154846,
      "learning_rate": 0.0005478077526882173,
      "loss": 2.2873,
      "step": 43914
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.69492506980896,
      "learning_rate": 0.0005478054470924401,
      "loss": 2.3936,
      "step": 43915
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.6848210096359253,
      "learning_rate": 0.0005478031414505913,
      "loss": 2.3158,
      "step": 43916
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7070311903953552,
      "learning_rate": 0.0005478008357626711,
      "loss": 2.2537,
      "step": 43917
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7593816518783569,
      "learning_rate": 0.0005477985300286801,
      "loss": 2.2256,
      "step": 43918
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.700408399105072,
      "learning_rate": 0.0005477962242486186,
      "loss": 2.38,
      "step": 43919
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.6706836223602295,
      "learning_rate": 0.0005477939184224872,
      "loss": 2.4234,
      "step": 43920
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.724640965461731,
      "learning_rate": 0.0005477916125502862,
      "loss": 2.4293,
      "step": 43921
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.6744024753570557,
      "learning_rate": 0.000547789306632016,
      "loss": 2.3345,
      "step": 43922
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.6387510895729065,
      "learning_rate": 0.0005477870006676772,
      "loss": 2.367,
      "step": 43923
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.6932413578033447,
      "learning_rate": 0.0005477846946572701,
      "loss": 2.2131,
      "step": 43924
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.6818865537643433,
      "learning_rate": 0.0005477823886007951,
      "loss": 2.3122,
      "step": 43925
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.6979476809501648,
      "learning_rate": 0.0005477800824982527,
      "loss": 2.2206,
      "step": 43926
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7324457764625549,
      "learning_rate": 0.0005477777763496434,
      "loss": 2.4488,
      "step": 43927
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.6362886428833008,
      "learning_rate": 0.0005477754701549674,
      "loss": 2.0343,
      "step": 43928
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7796920537948608,
      "learning_rate": 0.0005477731639142253,
      "loss": 2.2665,
      "step": 43929
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7041966915130615,
      "learning_rate": 0.0005477708576274175,
      "loss": 2.3803,
      "step": 43930
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.6577704548835754,
      "learning_rate": 0.0005477685512945444,
      "loss": 2.2385,
      "step": 43931
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.692030668258667,
      "learning_rate": 0.0005477662449156064,
      "loss": 2.3612,
      "step": 43932
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7026275396347046,
      "learning_rate": 0.0005477639384906042,
      "loss": 2.1455,
      "step": 43933
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.6681921482086182,
      "learning_rate": 0.0005477616320195377,
      "loss": 2.3239,
      "step": 43934
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7476757764816284,
      "learning_rate": 0.0005477593255024079,
      "loss": 2.1741,
      "step": 43935
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.661382794380188,
      "learning_rate": 0.0005477570189392148,
      "loss": 2.2242,
      "step": 43936
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7195073366165161,
      "learning_rate": 0.0005477547123299591,
      "loss": 2.4683,
      "step": 43937
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7140665054321289,
      "learning_rate": 0.0005477524056746411,
      "loss": 2.5103,
      "step": 43938
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7486923933029175,
      "learning_rate": 0.0005477500989732612,
      "loss": 2.3279,
      "step": 43939
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.703613817691803,
      "learning_rate": 0.0005477477922258199,
      "loss": 2.231,
      "step": 43940
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.8417519927024841,
      "learning_rate": 0.0005477454854323176,
      "loss": 2.6091,
      "step": 43941
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.6988170146942139,
      "learning_rate": 0.0005477431785927547,
      "loss": 2.3417,
      "step": 43942
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.6829235553741455,
      "learning_rate": 0.0005477408717071318,
      "loss": 2.0644,
      "step": 43943
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.6869906783103943,
      "learning_rate": 0.000547738564775449,
      "loss": 2.1192,
      "step": 43944
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7192161679267883,
      "learning_rate": 0.0005477362577977072,
      "loss": 2.3365,
      "step": 43945
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7359111309051514,
      "learning_rate": 0.0005477339507739062,
      "loss": 2.2444,
      "step": 43946
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7067000269889832,
      "learning_rate": 0.0005477316437040471,
      "loss": 2.3077,
      "step": 43947
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.6277413368225098,
      "learning_rate": 0.0005477293365881298,
      "loss": 2.1286,
      "step": 43948
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7785463333129883,
      "learning_rate": 0.0005477270294261551,
      "loss": 2.1845,
      "step": 43949
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.6455017924308777,
      "learning_rate": 0.0005477247222181232,
      "loss": 2.2073,
      "step": 43950
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.785876452922821,
      "learning_rate": 0.0005477224149640345,
      "loss": 2.3129,
      "step": 43951
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.639162003993988,
      "learning_rate": 0.0005477201076638897,
      "loss": 2.1782,
      "step": 43952
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7860509157180786,
      "learning_rate": 0.0005477178003176889,
      "loss": 2.295,
      "step": 43953
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7407128214836121,
      "learning_rate": 0.0005477154929254328,
      "loss": 2.2895,
      "step": 43954
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7347254753112793,
      "learning_rate": 0.0005477131854871217,
      "loss": 2.3939,
      "step": 43955
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7100722193717957,
      "learning_rate": 0.000547710878002756,
      "loss": 2.1494,
      "step": 43956
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.8526698350906372,
      "learning_rate": 0.0005477085704723361,
      "loss": 2.5462,
      "step": 43957
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7890607714653015,
      "learning_rate": 0.0005477062628958626,
      "loss": 2.1724,
      "step": 43958
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.6476700305938721,
      "learning_rate": 0.0005477039552733359,
      "loss": 2.0034,
      "step": 43959
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7198718786239624,
      "learning_rate": 0.0005477016476047562,
      "loss": 2.3176,
      "step": 43960
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7325571775436401,
      "learning_rate": 0.0005476993398901242,
      "loss": 2.3064,
      "step": 43961
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7752858400344849,
      "learning_rate": 0.0005476970321294402,
      "loss": 2.1318,
      "step": 43962
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7063807845115662,
      "learning_rate": 0.0005476947243227045,
      "loss": 2.2449,
      "step": 43963
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.6898366212844849,
      "learning_rate": 0.0005476924164699179,
      "loss": 2.1591,
      "step": 43964
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7846615314483643,
      "learning_rate": 0.0005476901085710805,
      "loss": 2.3643,
      "step": 43965
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.6933990716934204,
      "learning_rate": 0.0005476878006261929,
      "loss": 2.2743,
      "step": 43966
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.6854868531227112,
      "learning_rate": 0.0005476854926352553,
      "loss": 2.1281,
      "step": 43967
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.6670771837234497,
      "learning_rate": 0.0005476831845982684,
      "loss": 2.1897,
      "step": 43968
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7203892469406128,
      "learning_rate": 0.0005476808765152325,
      "loss": 2.1497,
      "step": 43969
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.6736630797386169,
      "learning_rate": 0.0005476785683861482,
      "loss": 2.3061,
      "step": 43970
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.6852396726608276,
      "learning_rate": 0.0005476762602110156,
      "loss": 2.2639,
      "step": 43971
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.6819261908531189,
      "learning_rate": 0.0005476739519898353,
      "loss": 2.5494,
      "step": 43972
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.6859497427940369,
      "learning_rate": 0.0005476716437226078,
      "loss": 2.3463,
      "step": 43973
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7207888960838318,
      "learning_rate": 0.0005476693354093334,
      "loss": 2.6061,
      "step": 43974
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7308072447776794,
      "learning_rate": 0.0005476670270500127,
      "loss": 2.4063,
      "step": 43975
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.6205573678016663,
      "learning_rate": 0.000547664718644646,
      "loss": 2.1956,
      "step": 43976
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.6755220293998718,
      "learning_rate": 0.0005476624101932336,
      "loss": 2.238,
      "step": 43977
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.6752177476882935,
      "learning_rate": 0.0005476601016957763,
      "loss": 2.1361,
      "step": 43978
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.6690553426742554,
      "learning_rate": 0.0005476577931522742,
      "loss": 2.0016,
      "step": 43979
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.6919255256652832,
      "learning_rate": 0.0005476554845627279,
      "loss": 2.2668,
      "step": 43980
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.6653348803520203,
      "learning_rate": 0.0005476531759271377,
      "loss": 2.4098,
      "step": 43981
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7692352533340454,
      "learning_rate": 0.0005476508672455041,
      "loss": 2.1197,
      "step": 43982
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.6483243107795715,
      "learning_rate": 0.0005476485585178277,
      "loss": 2.1035,
      "step": 43983
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7351872324943542,
      "learning_rate": 0.0005476462497441085,
      "loss": 2.2215,
      "step": 43984
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.8879562020301819,
      "learning_rate": 0.0005476439409243473,
      "loss": 2.2278,
      "step": 43985
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7531020045280457,
      "learning_rate": 0.0005476416320585445,
      "loss": 2.1697,
      "step": 43986
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7065218091011047,
      "learning_rate": 0.0005476393231467003,
      "loss": 2.106,
      "step": 43987
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7969585061073303,
      "learning_rate": 0.0005476370141888154,
      "loss": 2.2143,
      "step": 43988
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7613599300384521,
      "learning_rate": 0.00054763470518489,
      "loss": 2.5411,
      "step": 43989
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.6393290162086487,
      "learning_rate": 0.0005476323961349247,
      "loss": 2.1339,
      "step": 43990
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7082887291908264,
      "learning_rate": 0.0005476300870389198,
      "loss": 2.2965,
      "step": 43991
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7792696952819824,
      "learning_rate": 0.0005476277778968759,
      "loss": 2.2589,
      "step": 43992
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7062004804611206,
      "learning_rate": 0.0005476254687087933,
      "loss": 2.0468,
      "step": 43993
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.6338661909103394,
      "learning_rate": 0.0005476231594746724,
      "loss": 2.21,
      "step": 43994
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7652676105499268,
      "learning_rate": 0.0005476208501945136,
      "loss": 2.4397,
      "step": 43995
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.9582868814468384,
      "learning_rate": 0.0005476185408683176,
      "loss": 2.1117,
      "step": 43996
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.755653440952301,
      "learning_rate": 0.0005476162314960846,
      "loss": 2.2351,
      "step": 43997
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7111148238182068,
      "learning_rate": 0.0005476139220778149,
      "loss": 2.261,
      "step": 43998
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7698434591293335,
      "learning_rate": 0.0005476116126135092,
      "loss": 2.3062,
      "step": 43999
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.6459094882011414,
      "learning_rate": 0.0005476093031031679,
      "loss": 2.0391,
      "step": 44000
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.6591730117797852,
      "learning_rate": 0.0005476069935467912,
      "loss": 2.3727,
      "step": 44001
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.6469961404800415,
      "learning_rate": 0.0005476046839443798,
      "loss": 2.2139,
      "step": 44002
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.6922880411148071,
      "learning_rate": 0.000547602374295934,
      "loss": 2.1669,
      "step": 44003
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7252997159957886,
      "learning_rate": 0.0005476000646014543,
      "loss": 2.1031,
      "step": 44004
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7244155406951904,
      "learning_rate": 0.0005475977548609409,
      "loss": 2.2314,
      "step": 44005
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7561745643615723,
      "learning_rate": 0.0005475954450743947,
      "loss": 2.3899,
      "step": 44006
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7920387983322144,
      "learning_rate": 0.0005475931352418155,
      "loss": 2.3087,
      "step": 44007
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7036452889442444,
      "learning_rate": 0.0005475908253632042,
      "loss": 2.353,
      "step": 44008
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7224031686782837,
      "learning_rate": 0.0005475885154385612,
      "loss": 2.1564,
      "step": 44009
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7253536581993103,
      "learning_rate": 0.0005475862054678867,
      "loss": 2.1407,
      "step": 44010
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7260507941246033,
      "learning_rate": 0.0005475838954511812,
      "loss": 2.2306,
      "step": 44011
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7188934683799744,
      "learning_rate": 0.0005475815853884453,
      "loss": 2.3186,
      "step": 44012
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.6953834295272827,
      "learning_rate": 0.0005475792752796793,
      "loss": 2.248,
      "step": 44013
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7272042036056519,
      "learning_rate": 0.0005475769651248836,
      "loss": 1.9115,
      "step": 44014
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.8017436265945435,
      "learning_rate": 0.0005475746549240587,
      "loss": 2.3099,
      "step": 44015
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.6688223481178284,
      "learning_rate": 0.0005475723446772049,
      "loss": 2.1327,
      "step": 44016
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7430937886238098,
      "learning_rate": 0.0005475700343843228,
      "loss": 2.3246,
      "step": 44017
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7752900123596191,
      "learning_rate": 0.0005475677240454128,
      "loss": 2.0326,
      "step": 44018
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.6489748358726501,
      "learning_rate": 0.0005475654136604752,
      "loss": 2.3587,
      "step": 44019
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.6977865099906921,
      "learning_rate": 0.0005475631032295106,
      "loss": 2.3659,
      "step": 44020
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7918246388435364,
      "learning_rate": 0.0005475607927525193,
      "loss": 2.4156,
      "step": 44021
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.69728022813797,
      "learning_rate": 0.0005475584822295017,
      "loss": 2.3249,
      "step": 44022
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.6898410320281982,
      "learning_rate": 0.0005475561716604585,
      "loss": 2.2068,
      "step": 44023
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.810505747795105,
      "learning_rate": 0.0005475538610453897,
      "loss": 2.3778,
      "step": 44024
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.8556594848632812,
      "learning_rate": 0.0005475515503842963,
      "loss": 2.5094,
      "step": 44025
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.6413958668708801,
      "learning_rate": 0.000547549239677178,
      "loss": 2.3188,
      "step": 44026
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.6789307594299316,
      "learning_rate": 0.0005475469289240359,
      "loss": 2.2993,
      "step": 44027
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7942556142807007,
      "learning_rate": 0.0005475446181248699,
      "loss": 2.4366,
      "step": 44028
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7421041131019592,
      "learning_rate": 0.0005475423072796809,
      "loss": 2.343,
      "step": 44029
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.6917088031768799,
      "learning_rate": 0.000547539996388469,
      "loss": 2.3922,
      "step": 44030
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7692320942878723,
      "learning_rate": 0.0005475376854512349,
      "loss": 2.0968,
      "step": 44031
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.6795138716697693,
      "learning_rate": 0.0005475353744679787,
      "loss": 2.3812,
      "step": 44032
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.691089391708374,
      "learning_rate": 0.0005475330634387011,
      "loss": 2.3193,
      "step": 44033
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.827331006526947,
      "learning_rate": 0.0005475307523634024,
      "loss": 2.4041,
      "step": 44034
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7099063992500305,
      "learning_rate": 0.000547528441242083,
      "loss": 2.3863,
      "step": 44035
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.6558375358581543,
      "learning_rate": 0.0005475261300747435,
      "loss": 2.3404,
      "step": 44036
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.6871591210365295,
      "learning_rate": 0.0005475238188613841,
      "loss": 2.18,
      "step": 44037
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7667639851570129,
      "learning_rate": 0.0005475215076020055,
      "loss": 2.0109,
      "step": 44038
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.8672636151313782,
      "learning_rate": 0.0005475191962966079,
      "loss": 2.2678,
      "step": 44039
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7237831354141235,
      "learning_rate": 0.0005475168849451918,
      "loss": 2.3856,
      "step": 44040
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7492804527282715,
      "learning_rate": 0.0005475145735477576,
      "loss": 2.3739,
      "step": 44041
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.9800491333007812,
      "learning_rate": 0.0005475122621043059,
      "loss": 1.9229,
      "step": 44042
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.671546459197998,
      "learning_rate": 0.0005475099506148369,
      "loss": 2.2341,
      "step": 44043
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.9535865783691406,
      "learning_rate": 0.0005475076390793511,
      "loss": 2.2086,
      "step": 44044
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.780630886554718,
      "learning_rate": 0.000547505327497849,
      "loss": 2.1825,
      "step": 44045
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7637236714363098,
      "learning_rate": 0.0005475030158703311,
      "loss": 2.2274,
      "step": 44046
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7520928382873535,
      "learning_rate": 0.0005475007041967976,
      "loss": 2.2154,
      "step": 44047
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.793560266494751,
      "learning_rate": 0.0005474983924772489,
      "loss": 2.5176,
      "step": 44048
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.6259064078330994,
      "learning_rate": 0.0005474960807116859,
      "loss": 2.1604,
      "step": 44049
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7108761668205261,
      "learning_rate": 0.0005474937689001084,
      "loss": 2.2292,
      "step": 44050
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7631110548973083,
      "learning_rate": 0.0005474914570425173,
      "loss": 2.3234,
      "step": 44051
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.6557855010032654,
      "learning_rate": 0.0005474891451389128,
      "loss": 2.3169,
      "step": 44052
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.6685329675674438,
      "learning_rate": 0.0005474868331892955,
      "loss": 2.2785,
      "step": 44053
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.6319500803947449,
      "learning_rate": 0.0005474845211936656,
      "loss": 2.3589,
      "step": 44054
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.724300742149353,
      "learning_rate": 0.0005474822091520237,
      "loss": 2.4188,
      "step": 44055
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.6694303154945374,
      "learning_rate": 0.0005474798970643703,
      "loss": 2.3393,
      "step": 44056
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7734153866767883,
      "learning_rate": 0.0005474775849307056,
      "loss": 2.335,
      "step": 44057
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.753130316734314,
      "learning_rate": 0.0005474752727510302,
      "loss": 2.4037,
      "step": 44058
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.675165593624115,
      "learning_rate": 0.0005474729605253444,
      "loss": 2.3242,
      "step": 44059
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.6920179724693298,
      "learning_rate": 0.0005474706482536487,
      "loss": 2.0158,
      "step": 44060
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.6663835048675537,
      "learning_rate": 0.0005474683359359436,
      "loss": 2.2208,
      "step": 44061
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.6941440105438232,
      "learning_rate": 0.0005474660235722294,
      "loss": 2.3753,
      "step": 44062
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.6637396216392517,
      "learning_rate": 0.0005474637111625067,
      "loss": 2.4788,
      "step": 44063
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.6910862326622009,
      "learning_rate": 0.0005474613987067758,
      "loss": 2.2481,
      "step": 44064
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7037630677223206,
      "learning_rate": 0.0005474590862050371,
      "loss": 2.324,
      "step": 44065
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.6801458597183228,
      "learning_rate": 0.000547456773657291,
      "loss": 2.2946,
      "step": 44066
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.6396640539169312,
      "learning_rate": 0.0005474544610635382,
      "loss": 2.5116,
      "step": 44067
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.6278761029243469,
      "learning_rate": 0.0005474521484237788,
      "loss": 2.0435,
      "step": 44068
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.699403703212738,
      "learning_rate": 0.0005474498357380135,
      "loss": 2.2696,
      "step": 44069
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7338674068450928,
      "learning_rate": 0.0005474475230062425,
      "loss": 2.4545,
      "step": 44070
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.6937226057052612,
      "learning_rate": 0.0005474452102284664,
      "loss": 2.3767,
      "step": 44071
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.6887484192848206,
      "learning_rate": 0.0005474428974046855,
      "loss": 2.3887,
      "step": 44072
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.6410611867904663,
      "learning_rate": 0.0005474405845349004,
      "loss": 2.159,
      "step": 44073
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.758401095867157,
      "learning_rate": 0.0005474382716191113,
      "loss": 2.3654,
      "step": 44074
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7072795629501343,
      "learning_rate": 0.0005474359586573189,
      "loss": 2.1906,
      "step": 44075
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.671341598033905,
      "learning_rate": 0.0005474336456495234,
      "loss": 2.5072,
      "step": 44076
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.6709108948707581,
      "learning_rate": 0.0005474313325957253,
      "loss": 2.0624,
      "step": 44077
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.6482585072517395,
      "learning_rate": 0.000547429019495925,
      "loss": 2.3505,
      "step": 44078
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.6844265460968018,
      "learning_rate": 0.0005474267063501231,
      "loss": 2.2252,
      "step": 44079
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.6477813124656677,
      "learning_rate": 0.0005474243931583198,
      "loss": 2.2318,
      "step": 44080
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7200924158096313,
      "learning_rate": 0.0005474220799205156,
      "loss": 2.3254,
      "step": 44081
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7976807355880737,
      "learning_rate": 0.0005474197666367111,
      "loss": 2.1802,
      "step": 44082
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7617908716201782,
      "learning_rate": 0.0005474174533069066,
      "loss": 2.2638,
      "step": 44083
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.8135398030281067,
      "learning_rate": 0.0005474151399311024,
      "loss": 2.0846,
      "step": 44084
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.751470685005188,
      "learning_rate": 0.0005474128265092991,
      "loss": 2.3901,
      "step": 44085
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7689827680587769,
      "learning_rate": 0.0005474105130414971,
      "loss": 2.2253,
      "step": 44086
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7011088132858276,
      "learning_rate": 0.0005474081995276969,
      "loss": 2.2493,
      "step": 44087
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.8661419749259949,
      "learning_rate": 0.0005474058859678986,
      "loss": 2.1578,
      "step": 44088
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.714881181716919,
      "learning_rate": 0.000547403572362103,
      "loss": 2.2352,
      "step": 44089
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7719456553459167,
      "learning_rate": 0.0005474012587103105,
      "loss": 2.4205,
      "step": 44090
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7484944462776184,
      "learning_rate": 0.0005473989450125214,
      "loss": 2.3613,
      "step": 44091
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.8345449566841125,
      "learning_rate": 0.0005473966312687362,
      "loss": 2.329,
      "step": 44092
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.6979323625564575,
      "learning_rate": 0.0005473943174789552,
      "loss": 2.5923,
      "step": 44093
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7953617572784424,
      "learning_rate": 0.0005473920036431789,
      "loss": 2.1316,
      "step": 44094
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7249249815940857,
      "learning_rate": 0.0005473896897614078,
      "loss": 2.276,
      "step": 44095
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.6761424541473389,
      "learning_rate": 0.0005473873758336424,
      "loss": 2.1815,
      "step": 44096
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.6715067028999329,
      "learning_rate": 0.0005473850618598828,
      "loss": 2.2405,
      "step": 44097
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7796538472175598,
      "learning_rate": 0.0005473827478401299,
      "loss": 2.6111,
      "step": 44098
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7870969772338867,
      "learning_rate": 0.0005473804337743837,
      "loss": 2.3639,
      "step": 44099
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7060675621032715,
      "learning_rate": 0.0005473781196626449,
      "loss": 2.2797,
      "step": 44100
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.6298427581787109,
      "learning_rate": 0.0005473758055049138,
      "loss": 2.2815,
      "step": 44101
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.8161662220954895,
      "learning_rate": 0.0005473734913011908,
      "loss": 2.3369,
      "step": 44102
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.6406161189079285,
      "learning_rate": 0.0005473711770514765,
      "loss": 2.0995,
      "step": 44103
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7584018707275391,
      "learning_rate": 0.0005473688627557711,
      "loss": 2.2065,
      "step": 44104
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7280204892158508,
      "learning_rate": 0.0005473665484140753,
      "loss": 2.3381,
      "step": 44105
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.6692070960998535,
      "learning_rate": 0.0005473642340263894,
      "loss": 2.3467,
      "step": 44106
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7397196888923645,
      "learning_rate": 0.0005473619195927137,
      "loss": 2.2416,
      "step": 44107
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.6551218032836914,
      "learning_rate": 0.0005473596051130488,
      "loss": 2.2931,
      "step": 44108
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.6929377317428589,
      "learning_rate": 0.0005473572905873951,
      "loss": 2.3682,
      "step": 44109
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.6575645208358765,
      "learning_rate": 0.000547354976015753,
      "loss": 2.3008,
      "step": 44110
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7683097124099731,
      "learning_rate": 0.0005473526613981229,
      "loss": 2.4757,
      "step": 44111
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.6981987953186035,
      "learning_rate": 0.0005473503467345053,
      "loss": 2.2069,
      "step": 44112
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7045239210128784,
      "learning_rate": 0.0005473480320249007,
      "loss": 2.3442,
      "step": 44113
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.812065601348877,
      "learning_rate": 0.0005473457172693093,
      "loss": 2.1396,
      "step": 44114
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.6881484389305115,
      "learning_rate": 0.0005473434024677317,
      "loss": 2.3714,
      "step": 44115
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7458156943321228,
      "learning_rate": 0.0005473410876201683,
      "loss": 2.2211,
      "step": 44116
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.6689450144767761,
      "learning_rate": 0.0005473387727266195,
      "loss": 2.2042,
      "step": 44117
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7134537100791931,
      "learning_rate": 0.0005473364577870858,
      "loss": 2.154,
      "step": 44118
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.6791983842849731,
      "learning_rate": 0.0005473341428015676,
      "loss": 2.2762,
      "step": 44119
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.793831467628479,
      "learning_rate": 0.0005473318277700652,
      "loss": 2.4091,
      "step": 44120
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7099527716636658,
      "learning_rate": 0.0005473295126925793,
      "loss": 2.3683,
      "step": 44121
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.694101870059967,
      "learning_rate": 0.0005473271975691101,
      "loss": 2.2755,
      "step": 44122
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.753346860408783,
      "learning_rate": 0.0005473248823996581,
      "loss": 2.3827,
      "step": 44123
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7625045776367188,
      "learning_rate": 0.0005473225671842237,
      "loss": 2.2932,
      "step": 44124
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.6272250413894653,
      "learning_rate": 0.0005473202519228074,
      "loss": 2.2139,
      "step": 44125
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7863317131996155,
      "learning_rate": 0.0005473179366154095,
      "loss": 2.2555,
      "step": 44126
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.67332524061203,
      "learning_rate": 0.0005473156212620307,
      "loss": 2.3158,
      "step": 44127
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7113655805587769,
      "learning_rate": 0.0005473133058626711,
      "loss": 2.257,
      "step": 44128
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7075148820877075,
      "learning_rate": 0.0005473109904173314,
      "loss": 2.1549,
      "step": 44129
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.6695768237113953,
      "learning_rate": 0.0005473086749260119,
      "loss": 2.431,
      "step": 44130
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7285133600234985,
      "learning_rate": 0.000547306359388713,
      "loss": 2.2811,
      "step": 44131
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.6827030181884766,
      "learning_rate": 0.0005473040438054353,
      "loss": 2.304,
      "step": 44132
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.698248028755188,
      "learning_rate": 0.000547301728176179,
      "loss": 2.0904,
      "step": 44133
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.6643303036689758,
      "learning_rate": 0.0005472994125009446,
      "loss": 2.1194,
      "step": 44134
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.6562657952308655,
      "learning_rate": 0.0005472970967797327,
      "loss": 2.1713,
      "step": 44135
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7312234044075012,
      "learning_rate": 0.0005472947810125435,
      "loss": 2.2878,
      "step": 44136
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7153456807136536,
      "learning_rate": 0.0005472924651993776,
      "loss": 2.2993,
      "step": 44137
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7512909770011902,
      "learning_rate": 0.0005472901493402354,
      "loss": 2.2889,
      "step": 44138
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7306193709373474,
      "learning_rate": 0.0005472878334351172,
      "loss": 2.1139,
      "step": 44139
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7430717349052429,
      "learning_rate": 0.0005472855174840235,
      "loss": 2.2759,
      "step": 44140
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.6819320321083069,
      "learning_rate": 0.0005472832014869548,
      "loss": 2.2099,
      "step": 44141
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7058495283126831,
      "learning_rate": 0.0005472808854439116,
      "loss": 2.1458,
      "step": 44142
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.6942834258079529,
      "learning_rate": 0.0005472785693548941,
      "loss": 2.3867,
      "step": 44143
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.6313217878341675,
      "learning_rate": 0.0005472762532199028,
      "loss": 2.0953,
      "step": 44144
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7614870071411133,
      "learning_rate": 0.0005472739370389381,
      "loss": 2.3279,
      "step": 44145
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7014086842536926,
      "learning_rate": 0.0005472716208120008,
      "loss": 2.043,
      "step": 44146
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.6426452994346619,
      "learning_rate": 0.0005472693045390909,
      "loss": 2.3035,
      "step": 44147
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.648621678352356,
      "learning_rate": 0.0005472669882202088,
      "loss": 2.3722,
      "step": 44148
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.737217366695404,
      "learning_rate": 0.0005472646718553553,
      "loss": 2.2206,
      "step": 44149
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7894011735916138,
      "learning_rate": 0.0005472623554445305,
      "loss": 2.2026,
      "step": 44150
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7423797249794006,
      "learning_rate": 0.0005472600389877351,
      "loss": 2.148,
      "step": 44151
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7083181142807007,
      "learning_rate": 0.0005472577224849693,
      "loss": 2.4649,
      "step": 44152
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7550252079963684,
      "learning_rate": 0.0005472554059362337,
      "loss": 2.3438,
      "step": 44153
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7337856888771057,
      "learning_rate": 0.0005472530893415286,
      "loss": 2.2834,
      "step": 44154
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.6669383645057678,
      "learning_rate": 0.0005472507727008545,
      "loss": 2.4616,
      "step": 44155
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.686158299446106,
      "learning_rate": 0.0005472484560142117,
      "loss": 2.2577,
      "step": 44156
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.783467710018158,
      "learning_rate": 0.0005472461392816008,
      "loss": 2.1947,
      "step": 44157
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7700796723365784,
      "learning_rate": 0.0005472438225030223,
      "loss": 2.1393,
      "step": 44158
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7041695713996887,
      "learning_rate": 0.0005472415056784765,
      "loss": 2.158,
      "step": 44159
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7065062522888184,
      "learning_rate": 0.0005472391888079637,
      "loss": 1.9976,
      "step": 44160
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7282696962356567,
      "learning_rate": 0.0005472368718914845,
      "loss": 2.2006,
      "step": 44161
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.6815348267555237,
      "learning_rate": 0.0005472345549290394,
      "loss": 2.1705,
      "step": 44162
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7230067849159241,
      "learning_rate": 0.0005472322379206287,
      "loss": 2.0814,
      "step": 44163
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7009341716766357,
      "learning_rate": 0.0005472299208662527,
      "loss": 2.2761,
      "step": 44164
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7652205228805542,
      "learning_rate": 0.0005472276037659121,
      "loss": 2.2925,
      "step": 44165
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.6501162648200989,
      "learning_rate": 0.0005472252866196072,
      "loss": 1.9684,
      "step": 44166
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.6588784456253052,
      "learning_rate": 0.0005472229694273385,
      "loss": 2.0727,
      "step": 44167
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7449289560317993,
      "learning_rate": 0.0005472206521891064,
      "loss": 2.2658,
      "step": 44168
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7495128512382507,
      "learning_rate": 0.0005472183349049113,
      "loss": 2.3457,
      "step": 44169
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7224423289299011,
      "learning_rate": 0.0005472160175747536,
      "loss": 2.3938,
      "step": 44170
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7338704466819763,
      "learning_rate": 0.0005472137001986338,
      "loss": 2.3356,
      "step": 44171
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7649018168449402,
      "learning_rate": 0.0005472113827765523,
      "loss": 2.4255,
      "step": 44172
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7179310917854309,
      "learning_rate": 0.0005472090653085095,
      "loss": 2.3931,
      "step": 44173
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.6442874073982239,
      "learning_rate": 0.000547206747794506,
      "loss": 2.0566,
      "step": 44174
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7015128135681152,
      "learning_rate": 0.0005472044302345419,
      "loss": 2.4176,
      "step": 44175
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.8248568773269653,
      "learning_rate": 0.000547202112628618,
      "loss": 2.1717,
      "step": 44176
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7424539923667908,
      "learning_rate": 0.0005471997949767345,
      "loss": 2.3846,
      "step": 44177
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.6955704092979431,
      "learning_rate": 0.0005471974772788919,
      "loss": 2.1149,
      "step": 44178
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.9080748558044434,
      "learning_rate": 0.0005471951595350906,
      "loss": 2.3646,
      "step": 44179
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.720043420791626,
      "learning_rate": 0.000547192841745331,
      "loss": 2.3149,
      "step": 44180
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7937076091766357,
      "learning_rate": 0.0005471905239096137,
      "loss": 2.2003,
      "step": 44181
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.788139283657074,
      "learning_rate": 0.000547188206027939,
      "loss": 2.0697,
      "step": 44182
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.6923220753669739,
      "learning_rate": 0.0005471858881003073,
      "loss": 2.2604,
      "step": 44183
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7323798537254333,
      "learning_rate": 0.0005471835701267191,
      "loss": 2.1975,
      "step": 44184
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.9737318754196167,
      "learning_rate": 0.0005471812521071748,
      "loss": 2.0297,
      "step": 44185
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7060547471046448,
      "learning_rate": 0.0005471789340416748,
      "loss": 2.1759,
      "step": 44186
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.1072555780410767,
      "learning_rate": 0.0005471766159302196,
      "loss": 2.2701,
      "step": 44187
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7864511013031006,
      "learning_rate": 0.0005471742977728097,
      "loss": 2.3499,
      "step": 44188
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.6410092115402222,
      "learning_rate": 0.0005471719795694454,
      "loss": 2.0506,
      "step": 44189
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.8503660559654236,
      "learning_rate": 0.000547169661320127,
      "loss": 2.1082,
      "step": 44190
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.8535822033882141,
      "learning_rate": 0.0005471673430248553,
      "loss": 2.3453,
      "step": 44191
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.6868439316749573,
      "learning_rate": 0.0005471650246836304,
      "loss": 2.2264,
      "step": 44192
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7667388319969177,
      "learning_rate": 0.0005471627062964529,
      "loss": 2.5506,
      "step": 44193
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7733988761901855,
      "learning_rate": 0.0005471603878633232,
      "loss": 2.3564,
      "step": 44194
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7060700058937073,
      "learning_rate": 0.0005471580693842416,
      "loss": 2.0648,
      "step": 44195
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7321368455886841,
      "learning_rate": 0.0005471557508592087,
      "loss": 2.4448,
      "step": 44196
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7683581113815308,
      "learning_rate": 0.000547153432288225,
      "loss": 2.2419,
      "step": 44197
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7506665587425232,
      "learning_rate": 0.0005471511136712906,
      "loss": 2.1941,
      "step": 44198
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.6595674753189087,
      "learning_rate": 0.0005471487950084065,
      "loss": 2.0493,
      "step": 44199
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.6697030663490295,
      "learning_rate": 0.0005471464762995724,
      "loss": 2.3385,
      "step": 44200
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7309979200363159,
      "learning_rate": 0.0005471441575447893,
      "loss": 2.1783,
      "step": 44201
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.691350519657135,
      "learning_rate": 0.0005471418387440573,
      "loss": 2.3569,
      "step": 44202
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.6853436827659607,
      "learning_rate": 0.0005471395198973771,
      "loss": 2.3712,
      "step": 44203
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.8318737149238586,
      "learning_rate": 0.000547137201004749,
      "loss": 2.2715,
      "step": 44204
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.6877542734146118,
      "learning_rate": 0.0005471348820661733,
      "loss": 2.2515,
      "step": 44205
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.6617299914360046,
      "learning_rate": 0.0005471325630816507,
      "loss": 2.3732,
      "step": 44206
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7713225483894348,
      "learning_rate": 0.0005471302440511814,
      "loss": 2.1682,
      "step": 44207
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.6903896331787109,
      "learning_rate": 0.0005471279249747661,
      "loss": 2.3915,
      "step": 44208
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.8033844232559204,
      "learning_rate": 0.0005471256058524049,
      "loss": 2.3828,
      "step": 44209
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7019890546798706,
      "learning_rate": 0.0005471232866840983,
      "loss": 2.1292,
      "step": 44210
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7705867290496826,
      "learning_rate": 0.000547120967469847,
      "loss": 2.2552,
      "step": 44211
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.701263427734375,
      "learning_rate": 0.0005471186482096511,
      "loss": 2.2805,
      "step": 44212
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.6934764981269836,
      "learning_rate": 0.0005471163289035112,
      "loss": 2.1559,
      "step": 44213
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.8546295762062073,
      "learning_rate": 0.0005471140095514278,
      "loss": 2.3486,
      "step": 44214
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.6887341141700745,
      "learning_rate": 0.0005471116901534012,
      "loss": 2.1245,
      "step": 44215
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7867177128791809,
      "learning_rate": 0.0005471093707094318,
      "loss": 2.3835,
      "step": 44216
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.8784992694854736,
      "learning_rate": 0.0005471070512195202,
      "loss": 2.1483,
      "step": 44217
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.8353797197341919,
      "learning_rate": 0.0005471047316836666,
      "loss": 2.111,
      "step": 44218
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7906714081764221,
      "learning_rate": 0.0005471024121018718,
      "loss": 2.1224,
      "step": 44219
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.856730043888092,
      "learning_rate": 0.0005471000924741359,
      "loss": 2.5045,
      "step": 44220
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.1098798513412476,
      "learning_rate": 0.0005470977728004593,
      "loss": 2.0569,
      "step": 44221
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7811220288276672,
      "learning_rate": 0.0005470954530808426,
      "loss": 2.306,
      "step": 44222
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.9820960164070129,
      "learning_rate": 0.0005470931333152862,
      "loss": 2.366,
      "step": 44223
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.988760769367218,
      "learning_rate": 0.0005470908135037907,
      "loss": 2.211,
      "step": 44224
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.6854972839355469,
      "learning_rate": 0.0005470884936463562,
      "loss": 2.4288,
      "step": 44225
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.6431185603141785,
      "learning_rate": 0.0005470861737429833,
      "loss": 2.3574,
      "step": 44226
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.9245073795318604,
      "learning_rate": 0.0005470838537936723,
      "loss": 2.6049,
      "step": 44227
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.531626582145691,
      "learning_rate": 0.0005470815337984239,
      "loss": 2.3387,
      "step": 44228
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.8366885781288147,
      "learning_rate": 0.0005470792137572384,
      "loss": 2.3815,
      "step": 44229
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.0531182289123535,
      "learning_rate": 0.0005470768936701162,
      "loss": 2.2177,
      "step": 44230
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.9647327065467834,
      "learning_rate": 0.0005470745735370576,
      "loss": 2.1273,
      "step": 44231
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.0192135572433472,
      "learning_rate": 0.0005470722533580634,
      "loss": 2.3352,
      "step": 44232
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7035426497459412,
      "learning_rate": 0.0005470699331331337,
      "loss": 2.1405,
      "step": 44233
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7440783381462097,
      "learning_rate": 0.0005470676128622689,
      "loss": 2.1824,
      "step": 44234
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.974004328250885,
      "learning_rate": 0.0005470652925454696,
      "loss": 2.4221,
      "step": 44235
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.0006991624832153,
      "learning_rate": 0.0005470629721827364,
      "loss": 2.2543,
      "step": 44236
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.6470783948898315,
      "learning_rate": 0.0005470606517740693,
      "loss": 2.2352,
      "step": 44237
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.9914100766181946,
      "learning_rate": 0.0005470583313194692,
      "loss": 2.2641,
      "step": 44238
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.221127986907959,
      "learning_rate": 0.0005470560108189361,
      "loss": 2.2574,
      "step": 44239
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.820097029209137,
      "learning_rate": 0.0005470536902724706,
      "loss": 2.3449,
      "step": 44240
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7027374505996704,
      "learning_rate": 0.0005470513696800733,
      "loss": 2.0387,
      "step": 44241
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7905511856079102,
      "learning_rate": 0.0005470490490417445,
      "loss": 2.3266,
      "step": 44242
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.0083345174789429,
      "learning_rate": 0.0005470467283574845,
      "loss": 2.3999,
      "step": 44243
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.0341947078704834,
      "learning_rate": 0.0005470444076272939,
      "loss": 2.4281,
      "step": 44244
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7545471787452698,
      "learning_rate": 0.000547042086851173,
      "loss": 2.1238,
      "step": 44245
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.9098302125930786,
      "learning_rate": 0.0005470397660291224,
      "loss": 2.1385,
      "step": 44246
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.8114100098609924,
      "learning_rate": 0.0005470374451611425,
      "loss": 2.2722,
      "step": 44247
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7635465860366821,
      "learning_rate": 0.0005470351242472336,
      "loss": 2.2431,
      "step": 44248
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7856378555297852,
      "learning_rate": 0.0005470328032873962,
      "loss": 2.1759,
      "step": 44249
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7164816856384277,
      "learning_rate": 0.0005470304822816308,
      "loss": 2.2276,
      "step": 44250
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.8427121639251709,
      "learning_rate": 0.0005470281612299377,
      "loss": 2.0976,
      "step": 44251
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7553021311759949,
      "learning_rate": 0.0005470258401323174,
      "loss": 1.8854,
      "step": 44252
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.687516450881958,
      "learning_rate": 0.0005470235189887704,
      "loss": 2.1082,
      "step": 44253
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7900754809379578,
      "learning_rate": 0.000547021197799297,
      "loss": 2.2199,
      "step": 44254
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7391010522842407,
      "learning_rate": 0.0005470188765638977,
      "loss": 2.0739,
      "step": 44255
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.6873355507850647,
      "learning_rate": 0.000547016555282573,
      "loss": 2.3709,
      "step": 44256
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.644870400428772,
      "learning_rate": 0.0005470142339553232,
      "loss": 2.3241,
      "step": 44257
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.6903090476989746,
      "learning_rate": 0.0005470119125821488,
      "loss": 2.3676,
      "step": 44258
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.6773470044136047,
      "learning_rate": 0.0005470095911630503,
      "loss": 2.0911,
      "step": 44259
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.6973685622215271,
      "learning_rate": 0.0005470072696980279,
      "loss": 2.2797,
      "step": 44260
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7149935960769653,
      "learning_rate": 0.0005470049481870823,
      "loss": 2.3888,
      "step": 44261
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7094032764434814,
      "learning_rate": 0.0005470026266302137,
      "loss": 2.1926,
      "step": 44262
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.6865929961204529,
      "learning_rate": 0.0005470003050274228,
      "loss": 2.1262,
      "step": 44263
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.6816116571426392,
      "learning_rate": 0.0005469979833787098,
      "loss": 2.2931,
      "step": 44264
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7335239052772522,
      "learning_rate": 0.0005469956616840751,
      "loss": 2.2463,
      "step": 44265
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7917966246604919,
      "learning_rate": 0.0005469933399435194,
      "loss": 2.0115,
      "step": 44266
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.6109262704849243,
      "learning_rate": 0.000546991018157043,
      "loss": 2.1552,
      "step": 44267
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.6749688982963562,
      "learning_rate": 0.0005469886963246462,
      "loss": 2.2799,
      "step": 44268
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.6929721236228943,
      "learning_rate": 0.0005469863744463297,
      "loss": 2.4442,
      "step": 44269
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.6732223629951477,
      "learning_rate": 0.0005469840525220936,
      "loss": 2.3162,
      "step": 44270
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.6908687949180603,
      "learning_rate": 0.0005469817305519385,
      "loss": 2.4108,
      "step": 44271
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.6524379849433899,
      "learning_rate": 0.0005469794085358649,
      "loss": 2.3375,
      "step": 44272
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7614993453025818,
      "learning_rate": 0.0005469770864738731,
      "loss": 2.1913,
      "step": 44273
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7267707586288452,
      "learning_rate": 0.0005469747643659637,
      "loss": 2.1032,
      "step": 44274
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7133493423461914,
      "learning_rate": 0.0005469724422121371,
      "loss": 2.3382,
      "step": 44275
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.6534004211425781,
      "learning_rate": 0.0005469701200123935,
      "loss": 2.1574,
      "step": 44276
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7145503163337708,
      "learning_rate": 0.0005469677977667335,
      "loss": 2.4699,
      "step": 44277
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7076576948165894,
      "learning_rate": 0.0005469654754751576,
      "loss": 2.5295,
      "step": 44278
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.6936297416687012,
      "learning_rate": 0.0005469631531376661,
      "loss": 2.4141,
      "step": 44279
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.6501360535621643,
      "learning_rate": 0.0005469608307542595,
      "loss": 2.2699,
      "step": 44280
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.6638122797012329,
      "learning_rate": 0.0005469585083249383,
      "loss": 2.326,
      "step": 44281
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7396140098571777,
      "learning_rate": 0.0005469561858497027,
      "loss": 2.1909,
      "step": 44282
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.6931940317153931,
      "learning_rate": 0.0005469538633285534,
      "loss": 2.0719,
      "step": 44283
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.6707834005355835,
      "learning_rate": 0.0005469515407614907,
      "loss": 2.1476,
      "step": 44284
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.6540733575820923,
      "learning_rate": 0.000546949218148515,
      "loss": 2.1893,
      "step": 44285
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7130618095397949,
      "learning_rate": 0.0005469468954896269,
      "loss": 2.3765,
      "step": 44286
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7453294396400452,
      "learning_rate": 0.0005469445727848266,
      "loss": 2.122,
      "step": 44287
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7463821172714233,
      "learning_rate": 0.0005469422500341145,
      "loss": 2.2108,
      "step": 44288
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7534680962562561,
      "learning_rate": 0.0005469399272374914,
      "loss": 2.2229,
      "step": 44289
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7111890316009521,
      "learning_rate": 0.0005469376043949574,
      "loss": 2.2128,
      "step": 44290
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7715507745742798,
      "learning_rate": 0.0005469352815065132,
      "loss": 2.2524,
      "step": 44291
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7843762040138245,
      "learning_rate": 0.0005469329585721589,
      "loss": 2.427,
      "step": 44292
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7450952529907227,
      "learning_rate": 0.0005469306355918951,
      "loss": 2.3556,
      "step": 44293
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.6816306114196777,
      "learning_rate": 0.0005469283125657223,
      "loss": 2.3463,
      "step": 44294
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.6839346289634705,
      "learning_rate": 0.0005469259894936409,
      "loss": 2.248,
      "step": 44295
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7050228118896484,
      "learning_rate": 0.0005469236663756513,
      "loss": 2.0586,
      "step": 44296
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.6423255801200867,
      "learning_rate": 0.0005469213432117538,
      "loss": 2.2167,
      "step": 44297
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7196608185768127,
      "learning_rate": 0.000546919020001949,
      "loss": 2.2778,
      "step": 44298
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.6852784752845764,
      "learning_rate": 0.0005469166967462375,
      "loss": 2.0708,
      "step": 44299
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.6515970230102539,
      "learning_rate": 0.0005469143734446192,
      "loss": 2.5014,
      "step": 44300
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.6548599004745483,
      "learning_rate": 0.0005469120500970951,
      "loss": 2.1779,
      "step": 44301
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.6816426515579224,
      "learning_rate": 0.0005469097267036653,
      "loss": 2.172,
      "step": 44302
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7673636674880981,
      "learning_rate": 0.0005469074032643304,
      "loss": 2.211,
      "step": 44303
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7011223435401917,
      "learning_rate": 0.0005469050797790905,
      "loss": 2.2454,
      "step": 44304
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.69736248254776,
      "learning_rate": 0.0005469027562479465,
      "loss": 2.1804,
      "step": 44305
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.6457359790802002,
      "learning_rate": 0.0005469004326708987,
      "loss": 2.2341,
      "step": 44306
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7140259146690369,
      "learning_rate": 0.0005468981090479473,
      "loss": 2.0476,
      "step": 44307
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7115989923477173,
      "learning_rate": 0.0005468957853790929,
      "loss": 2.2224,
      "step": 44308
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7001177072525024,
      "learning_rate": 0.0005468934616643359,
      "loss": 2.4283,
      "step": 44309
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.6845905184745789,
      "learning_rate": 0.0005468911379036767,
      "loss": 2.2754,
      "step": 44310
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.6951780915260315,
      "learning_rate": 0.0005468888140971158,
      "loss": 2.0859,
      "step": 44311
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7464083433151245,
      "learning_rate": 0.0005468864902446537,
      "loss": 2.4549,
      "step": 44312
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.67439204454422,
      "learning_rate": 0.0005468841663462907,
      "loss": 2.0651,
      "step": 44313
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7855283617973328,
      "learning_rate": 0.0005468818424020273,
      "loss": 2.2228,
      "step": 44314
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7925752401351929,
      "learning_rate": 0.0005468795184118638,
      "loss": 2.2841,
      "step": 44315
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.8038747906684875,
      "learning_rate": 0.0005468771943758008,
      "loss": 2.2508,
      "step": 44316
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7968342900276184,
      "learning_rate": 0.0005468748702938388,
      "loss": 2.2603,
      "step": 44317
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.6313679814338684,
      "learning_rate": 0.0005468725461659779,
      "loss": 2.3665,
      "step": 44318
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7311863303184509,
      "learning_rate": 0.0005468702219922189,
      "loss": 2.3035,
      "step": 44319
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.656337320804596,
      "learning_rate": 0.0005468678977725618,
      "loss": 2.2016,
      "step": 44320
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.6722167730331421,
      "learning_rate": 0.0005468655735070076,
      "loss": 2.2233,
      "step": 44321
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.6690144538879395,
      "learning_rate": 0.0005468632491955563,
      "loss": 2.3453,
      "step": 44322
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7499256134033203,
      "learning_rate": 0.0005468609248382085,
      "loss": 2.3517,
      "step": 44323
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7232590317726135,
      "learning_rate": 0.0005468586004349644,
      "loss": 2.2532,
      "step": 44324
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.6673406958580017,
      "learning_rate": 0.0005468562759858249,
      "loss": 2.3077,
      "step": 44325
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7113764882087708,
      "learning_rate": 0.00054685395149079,
      "loss": 2.1967,
      "step": 44326
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7229601740837097,
      "learning_rate": 0.0005468516269498604,
      "loss": 2.4208,
      "step": 44327
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.734682559967041,
      "learning_rate": 0.0005468493023630364,
      "loss": 2.3431,
      "step": 44328
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.6957941055297852,
      "learning_rate": 0.0005468469777303184,
      "loss": 2.0949,
      "step": 44329
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.6370887756347656,
      "learning_rate": 0.0005468446530517068,
      "loss": 2.3494,
      "step": 44330
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.775384783744812,
      "learning_rate": 0.0005468423283272023,
      "loss": 2.271,
      "step": 44331
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7512542009353638,
      "learning_rate": 0.0005468400035568051,
      "loss": 2.2466,
      "step": 44332
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7449319958686829,
      "learning_rate": 0.0005468376787405156,
      "loss": 2.2049,
      "step": 44333
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.6961383819580078,
      "learning_rate": 0.0005468353538783344,
      "loss": 2.384,
      "step": 44334
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7371636033058167,
      "learning_rate": 0.0005468330289702618,
      "loss": 2.3622,
      "step": 44335
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7432698011398315,
      "learning_rate": 0.0005468307040162984,
      "loss": 2.378,
      "step": 44336
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.6917681097984314,
      "learning_rate": 0.0005468283790164444,
      "loss": 2.1409,
      "step": 44337
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.6580255031585693,
      "learning_rate": 0.0005468260539707003,
      "loss": 2.2109,
      "step": 44338
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.673022449016571,
      "learning_rate": 0.0005468237288790667,
      "loss": 2.2079,
      "step": 44339
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.6470780372619629,
      "learning_rate": 0.0005468214037415437,
      "loss": 2.1494,
      "step": 44340
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.8247833251953125,
      "learning_rate": 0.0005468190785581323,
      "loss": 2.2302,
      "step": 44341
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7114359736442566,
      "learning_rate": 0.0005468167533288322,
      "loss": 2.2349,
      "step": 44342
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7415615916252136,
      "learning_rate": 0.0005468144280536445,
      "loss": 2.3012,
      "step": 44343
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7543076872825623,
      "learning_rate": 0.0005468121027325691,
      "loss": 2.1964,
      "step": 44344
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7315526008605957,
      "learning_rate": 0.0005468097773656069,
      "loss": 1.9607,
      "step": 44345
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.6375980377197266,
      "learning_rate": 0.0005468074519527578,
      "loss": 2.2428,
      "step": 44346
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.6892926692962646,
      "learning_rate": 0.0005468051264940229,
      "loss": 2.1103,
      "step": 44347
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7488250136375427,
      "learning_rate": 0.000546802800989402,
      "loss": 2.2745,
      "step": 44348
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7491846084594727,
      "learning_rate": 0.0005468004754388959,
      "loss": 2.2382,
      "step": 44349
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.6555929780006409,
      "learning_rate": 0.000546798149842505,
      "loss": 2.4082,
      "step": 44350
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.6822742819786072,
      "learning_rate": 0.0005467958242002295,
      "loss": 2.1715,
      "step": 44351
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7113298177719116,
      "learning_rate": 0.00054679349851207,
      "loss": 2.2989,
      "step": 44352
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7148479223251343,
      "learning_rate": 0.0005467911727780272,
      "loss": 2.256,
      "step": 44353
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7415528297424316,
      "learning_rate": 0.000546788846998101,
      "loss": 2.3205,
      "step": 44354
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.6665322184562683,
      "learning_rate": 0.0005467865211722922,
      "loss": 2.1702,
      "step": 44355
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.6576457619667053,
      "learning_rate": 0.000546784195300601,
      "loss": 2.1391,
      "step": 44356
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.5957069396972656,
      "learning_rate": 0.0005467818693830281,
      "loss": 2.153,
      "step": 44357
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.6854935884475708,
      "learning_rate": 0.0005467795434195738,
      "loss": 2.5072,
      "step": 44358
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7802026271820068,
      "learning_rate": 0.0005467772174102384,
      "loss": 2.3775,
      "step": 44359
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7261138558387756,
      "learning_rate": 0.0005467748913550225,
      "loss": 2.2303,
      "step": 44360
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.692494809627533,
      "learning_rate": 0.0005467725652539266,
      "loss": 2.262,
      "step": 44361
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7609063386917114,
      "learning_rate": 0.000546770239106951,
      "loss": 2.2557,
      "step": 44362
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7892004251480103,
      "learning_rate": 0.0005467679129140961,
      "loss": 2.6192,
      "step": 44363
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7310909628868103,
      "learning_rate": 0.0005467655866753623,
      "loss": 2.426,
      "step": 44364
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7011798620223999,
      "learning_rate": 0.0005467632603907502,
      "loss": 2.4021,
      "step": 44365
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7867111563682556,
      "learning_rate": 0.0005467609340602602,
      "loss": 2.5037,
      "step": 44366
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.6467297077178955,
      "learning_rate": 0.0005467586076838926,
      "loss": 2.1795,
      "step": 44367
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7307488918304443,
      "learning_rate": 0.000546756281261648,
      "loss": 2.1291,
      "step": 44368
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.6763005256652832,
      "learning_rate": 0.0005467539547935267,
      "loss": 2.1821,
      "step": 44369
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7219390869140625,
      "learning_rate": 0.0005467516282795293,
      "loss": 2.2238,
      "step": 44370
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.6878841519355774,
      "learning_rate": 0.0005467493017196559,
      "loss": 2.1947,
      "step": 44371
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.6752516031265259,
      "learning_rate": 0.0005467469751139072,
      "loss": 2.116,
      "step": 44372
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7921931147575378,
      "learning_rate": 0.0005467446484622837,
      "loss": 2.2186,
      "step": 44373
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.8606274127960205,
      "learning_rate": 0.0005467423217647856,
      "loss": 2.2533,
      "step": 44374
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.6588691473007202,
      "learning_rate": 0.0005467399950214134,
      "loss": 2.1364,
      "step": 44375
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7103167176246643,
      "learning_rate": 0.0005467376682321678,
      "loss": 2.1649,
      "step": 44376
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.738575279712677,
      "learning_rate": 0.0005467353413970488,
      "loss": 2.3501,
      "step": 44377
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.8187311887741089,
      "learning_rate": 0.0005467330145160572,
      "loss": 2.3783,
      "step": 44378
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7348834276199341,
      "learning_rate": 0.0005467306875891932,
      "loss": 2.1427,
      "step": 44379
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.6653665900230408,
      "learning_rate": 0.0005467283606164573,
      "loss": 2.4093,
      "step": 44380
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.6471176743507385,
      "learning_rate": 0.0005467260335978499,
      "loss": 2.2006,
      "step": 44381
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.6403255462646484,
      "learning_rate": 0.0005467237065333716,
      "loss": 2.2309,
      "step": 44382
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7171839475631714,
      "learning_rate": 0.0005467213794230225,
      "loss": 2.1267,
      "step": 44383
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7022835612297058,
      "learning_rate": 0.0005467190522668034,
      "loss": 2.3525,
      "step": 44384
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.734584391117096,
      "learning_rate": 0.0005467167250647146,
      "loss": 2.4287,
      "step": 44385
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7014817595481873,
      "learning_rate": 0.0005467143978167564,
      "loss": 2.3576,
      "step": 44386
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.6735900640487671,
      "learning_rate": 0.0005467120705229294,
      "loss": 2.3229,
      "step": 44387
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.6571516394615173,
      "learning_rate": 0.0005467097431832339,
      "loss": 2.5231,
      "step": 44388
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.6910914182662964,
      "learning_rate": 0.0005467074157976704,
      "loss": 2.3846,
      "step": 44389
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7737420201301575,
      "learning_rate": 0.0005467050883662394,
      "loss": 2.1192,
      "step": 44390
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.8115123510360718,
      "learning_rate": 0.0005467027608889412,
      "loss": 2.299,
      "step": 44391
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.6451818346977234,
      "learning_rate": 0.0005467004333657763,
      "loss": 2.1879,
      "step": 44392
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7822633385658264,
      "learning_rate": 0.0005466981057967453,
      "loss": 2.3096,
      "step": 44393
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.8127570748329163,
      "learning_rate": 0.0005466957781818483,
      "loss": 2.3965,
      "step": 44394
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.8062898516654968,
      "learning_rate": 0.0005466934505210859,
      "loss": 2.1313,
      "step": 44395
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.6533293724060059,
      "learning_rate": 0.0005466911228144587,
      "loss": 1.9972,
      "step": 44396
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.6421183943748474,
      "learning_rate": 0.0005466887950619667,
      "loss": 2.1257,
      "step": 44397
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7249456644058228,
      "learning_rate": 0.0005466864672636107,
      "loss": 2.2251,
      "step": 44398
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.6413895487785339,
      "learning_rate": 0.000546684139419391,
      "loss": 2.2437,
      "step": 44399
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.6677693724632263,
      "learning_rate": 0.0005466818115293083,
      "loss": 2.3065,
      "step": 44400
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.739301860332489,
      "learning_rate": 0.0005466794835933625,
      "loss": 2.5305,
      "step": 44401
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7454068660736084,
      "learning_rate": 0.0005466771556115545,
      "loss": 2.2598,
      "step": 44402
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.6561325788497925,
      "learning_rate": 0.0005466748275838845,
      "loss": 2.2927,
      "step": 44403
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.9113451838493347,
      "learning_rate": 0.000546672499510353,
      "loss": 2.2326,
      "step": 44404
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.6744853854179382,
      "learning_rate": 0.0005466701713909604,
      "loss": 2.5227,
      "step": 44405
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7576702833175659,
      "learning_rate": 0.0005466678432257072,
      "loss": 2.1864,
      "step": 44406
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7944976687431335,
      "learning_rate": 0.0005466655150145938,
      "loss": 2.234,
      "step": 44407
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.9094618558883667,
      "learning_rate": 0.0005466631867576205,
      "loss": 2.2125,
      "step": 44408
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.6107233762741089,
      "learning_rate": 0.000546660858454788,
      "loss": 2.412,
      "step": 44409
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7147114276885986,
      "learning_rate": 0.0005466585301060966,
      "loss": 2.2694,
      "step": 44410
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7888392806053162,
      "learning_rate": 0.0005466562017115466,
      "loss": 2.2578,
      "step": 44411
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.8663447499275208,
      "learning_rate": 0.0005466538732711387,
      "loss": 2.3303,
      "step": 44412
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.6766402125358582,
      "learning_rate": 0.000546651544784873,
      "loss": 2.2317,
      "step": 44413
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7379620671272278,
      "learning_rate": 0.0005466492162527503,
      "loss": 2.3585,
      "step": 44414
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.0670981407165527,
      "learning_rate": 0.0005466468876747707,
      "loss": 2.113,
      "step": 44415
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.9221939444541931,
      "learning_rate": 0.000546644559050935,
      "loss": 2.2831,
      "step": 44416
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7304043769836426,
      "learning_rate": 0.0005466422303812432,
      "loss": 2.2746,
      "step": 44417
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.8090671300888062,
      "learning_rate": 0.000546639901665696,
      "loss": 2.1444,
      "step": 44418
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7008694410324097,
      "learning_rate": 0.000546637572904294,
      "loss": 2.2995,
      "step": 44419
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.8304724097251892,
      "learning_rate": 0.0005466352440970371,
      "loss": 2.3829,
      "step": 44420
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.6717390418052673,
      "learning_rate": 0.0005466329152439262,
      "loss": 2.2097,
      "step": 44421
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.8079005479812622,
      "learning_rate": 0.0005466305863449617,
      "loss": 2.2117,
      "step": 44422
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.6927369832992554,
      "learning_rate": 0.0005466282574001438,
      "loss": 2.116,
      "step": 44423
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.6596949100494385,
      "learning_rate": 0.0005466259284094731,
      "loss": 2.2261,
      "step": 44424
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7064184546470642,
      "learning_rate": 0.0005466235993729499,
      "loss": 2.3668,
      "step": 44425
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7892845273017883,
      "learning_rate": 0.0005466212702905748,
      "loss": 2.249,
      "step": 44426
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7569597959518433,
      "learning_rate": 0.0005466189411623482,
      "loss": 2.2931,
      "step": 44427
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7039141058921814,
      "learning_rate": 0.0005466166119882704,
      "loss": 2.3863,
      "step": 44428
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.9527533650398254,
      "learning_rate": 0.0005466142827683419,
      "loss": 2.2361,
      "step": 44429
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7677422165870667,
      "learning_rate": 0.000546611953502563,
      "loss": 2.2575,
      "step": 44430
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.6679669618606567,
      "learning_rate": 0.0005466096241909346,
      "loss": 2.28,
      "step": 44431
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7824707627296448,
      "learning_rate": 0.0005466072948334566,
      "loss": 2.3082,
      "step": 44432
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7017785906791687,
      "learning_rate": 0.0005466049654301297,
      "loss": 2.4388,
      "step": 44433
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7045599222183228,
      "learning_rate": 0.0005466026359809543,
      "loss": 2.0516,
      "step": 44434
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7302470207214355,
      "learning_rate": 0.0005466003064859309,
      "loss": 2.2988,
      "step": 44435
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7008727788925171,
      "learning_rate": 0.0005465979769450597,
      "loss": 2.2833,
      "step": 44436
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7980291247367859,
      "learning_rate": 0.0005465956473583414,
      "loss": 2.2756,
      "step": 44437
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.6335923671722412,
      "learning_rate": 0.0005465933177257762,
      "loss": 2.0723,
      "step": 44438
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.6655712127685547,
      "learning_rate": 0.0005465909880473647,
      "loss": 2.228,
      "step": 44439
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.8147209882736206,
      "learning_rate": 0.0005465886583231073,
      "loss": 2.2484,
      "step": 44440
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7033302187919617,
      "learning_rate": 0.0005465863285530044,
      "loss": 2.0781,
      "step": 44441
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7569107413291931,
      "learning_rate": 0.0005465839987370565,
      "loss": 2.081,
      "step": 44442
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.656690239906311,
      "learning_rate": 0.0005465816688752639,
      "loss": 2.4358,
      "step": 44443
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7718221545219421,
      "learning_rate": 0.0005465793389676271,
      "loss": 2.503,
      "step": 44444
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.625667929649353,
      "learning_rate": 0.0005465770090141465,
      "loss": 2.2749,
      "step": 44445
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.6923503279685974,
      "learning_rate": 0.0005465746790148228,
      "loss": 2.2494,
      "step": 44446
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7198349833488464,
      "learning_rate": 0.0005465723489696561,
      "loss": 2.3188,
      "step": 44447
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7733396887779236,
      "learning_rate": 0.0005465700188786468,
      "loss": 2.2743,
      "step": 44448
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.6720625162124634,
      "learning_rate": 0.0005465676887417957,
      "loss": 2.3788,
      "step": 44449
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.6923291683197021,
      "learning_rate": 0.0005465653585591028,
      "loss": 2.3389,
      "step": 44450
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.6699967980384827,
      "learning_rate": 0.0005465630283305689,
      "loss": 2.0756,
      "step": 44451
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7312616109848022,
      "learning_rate": 0.0005465606980561943,
      "loss": 2.452,
      "step": 44452
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.6340108513832092,
      "learning_rate": 0.0005465583677359793,
      "loss": 2.4029,
      "step": 44453
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.6965870261192322,
      "learning_rate": 0.0005465560373699246,
      "loss": 2.1621,
      "step": 44454
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.6832846403121948,
      "learning_rate": 0.0005465537069580303,
      "loss": 2.1325,
      "step": 44455
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7606250643730164,
      "learning_rate": 0.000546551376500297,
      "loss": 2.4667,
      "step": 44456
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7576968669891357,
      "learning_rate": 0.0005465490459967252,
      "loss": 2.1424,
      "step": 44457
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.6619609594345093,
      "learning_rate": 0.0005465467154473154,
      "loss": 2.2658,
      "step": 44458
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7607673406600952,
      "learning_rate": 0.0005465443848520678,
      "loss": 2.323,
      "step": 44459
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7019917368888855,
      "learning_rate": 0.000546542054210983,
      "loss": 2.37,
      "step": 44460
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7317976951599121,
      "learning_rate": 0.0005465397235240613,
      "loss": 2.2408,
      "step": 44461
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7649379968643188,
      "learning_rate": 0.0005465373927913033,
      "loss": 2.2682,
      "step": 44462
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7485652565956116,
      "learning_rate": 0.0005465350620127093,
      "loss": 2.2594,
      "step": 44463
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7086687684059143,
      "learning_rate": 0.0005465327311882797,
      "loss": 2.2625,
      "step": 44464
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7356670498847961,
      "learning_rate": 0.0005465304003180152,
      "loss": 2.4503,
      "step": 44465
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.6668757796287537,
      "learning_rate": 0.0005465280694019159,
      "loss": 2.1887,
      "step": 44466
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7303990125656128,
      "learning_rate": 0.0005465257384399824,
      "loss": 1.9873,
      "step": 44467
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7062836289405823,
      "learning_rate": 0.0005465234074322152,
      "loss": 2.0883,
      "step": 44468
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7545006275177002,
      "learning_rate": 0.0005465210763786145,
      "loss": 2.3941,
      "step": 44469
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7761594653129578,
      "learning_rate": 0.0005465187452791809,
      "loss": 2.3707,
      "step": 44470
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.703502893447876,
      "learning_rate": 0.0005465164141339149,
      "loss": 2.3936,
      "step": 44471
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.669228196144104,
      "learning_rate": 0.0005465140829428168,
      "loss": 2.1919,
      "step": 44472
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.6984822154045105,
      "learning_rate": 0.0005465117517058871,
      "loss": 2.4077,
      "step": 44473
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.6540584564208984,
      "learning_rate": 0.0005465094204231263,
      "loss": 2.5284,
      "step": 44474
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.6797922253608704,
      "learning_rate": 0.0005465070890945346,
      "loss": 2.3988,
      "step": 44475
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.8539208173751831,
      "learning_rate": 0.0005465047577201125,
      "loss": 2.1852,
      "step": 44476
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.6655030250549316,
      "learning_rate": 0.0005465024262998608,
      "loss": 2.2069,
      "step": 44477
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.6938247680664062,
      "learning_rate": 0.0005465000948337794,
      "loss": 2.075,
      "step": 44478
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.6872619986534119,
      "learning_rate": 0.000546497763321869,
      "loss": 2.3295,
      "step": 44479
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.6313886046409607,
      "learning_rate": 0.0005464954317641301,
      "loss": 2.2406,
      "step": 44480
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.6684648990631104,
      "learning_rate": 0.000546493100160563,
      "loss": 2.1393,
      "step": 44481
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7751221656799316,
      "learning_rate": 0.0005464907685111682,
      "loss": 2.2572,
      "step": 44482
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.6724159121513367,
      "learning_rate": 0.0005464884368159461,
      "loss": 2.0786,
      "step": 44483
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7374940514564514,
      "learning_rate": 0.0005464861050748972,
      "loss": 2.1569,
      "step": 44484
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7189217805862427,
      "learning_rate": 0.0005464837732880218,
      "loss": 2.2415,
      "step": 44485
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7021557688713074,
      "learning_rate": 0.0005464814414553205,
      "loss": 2.1669,
      "step": 44486
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7874598503112793,
      "learning_rate": 0.0005464791095767936,
      "loss": 2.3796,
      "step": 44487
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.8290987610816956,
      "learning_rate": 0.0005464767776524416,
      "loss": 2.591,
      "step": 44488
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7066132426261902,
      "learning_rate": 0.0005464744456822649,
      "loss": 2.435,
      "step": 44489
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.6480889916419983,
      "learning_rate": 0.0005464721136662639,
      "loss": 2.2388,
      "step": 44490
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.0505080223083496,
      "learning_rate": 0.0005464697816044391,
      "loss": 2.2705,
      "step": 44491
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.6741110682487488,
      "learning_rate": 0.0005464674494967909,
      "loss": 2.125,
      "step": 44492
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7409980297088623,
      "learning_rate": 0.0005464651173433198,
      "loss": 2.2984,
      "step": 44493
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.102382779121399,
      "learning_rate": 0.0005464627851440262,
      "loss": 2.3695,
      "step": 44494
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7780052423477173,
      "learning_rate": 0.0005464604528989105,
      "loss": 2.314,
      "step": 44495
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.651344358921051,
      "learning_rate": 0.0005464581206079731,
      "loss": 1.9528,
      "step": 44496
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.6891727447509766,
      "learning_rate": 0.0005464557882712145,
      "loss": 2.045,
      "step": 44497
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.68300461769104,
      "learning_rate": 0.0005464534558886351,
      "loss": 2.3565,
      "step": 44498
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7120451927185059,
      "learning_rate": 0.0005464511234602354,
      "loss": 2.1813,
      "step": 44499
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.685350239276886,
      "learning_rate": 0.0005464487909860158,
      "loss": 2.1143,
      "step": 44500
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.8466361165046692,
      "learning_rate": 0.0005464464584659768,
      "loss": 2.1113,
      "step": 44501
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.655524492263794,
      "learning_rate": 0.0005464441259001186,
      "loss": 2.2071,
      "step": 44502
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.6579974293708801,
      "learning_rate": 0.0005464417932884418,
      "loss": 2.4864,
      "step": 44503
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.6894309520721436,
      "learning_rate": 0.0005464394606309468,
      "loss": 2.3786,
      "step": 44504
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.6198410987854004,
      "learning_rate": 0.0005464371279276341,
      "loss": 2.2186,
      "step": 44505
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7065416574478149,
      "learning_rate": 0.0005464347951785041,
      "loss": 2.3638,
      "step": 44506
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.6464793086051941,
      "learning_rate": 0.0005464324623835572,
      "loss": 2.41,
      "step": 44507
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.6554287075996399,
      "learning_rate": 0.0005464301295427939,
      "loss": 2.4522,
      "step": 44508
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7159216403961182,
      "learning_rate": 0.0005464277966562145,
      "loss": 2.5164,
      "step": 44509
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.8757937550544739,
      "learning_rate": 0.0005464254637238196,
      "loss": 2.2674,
      "step": 44510
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7943888306617737,
      "learning_rate": 0.0005464231307456095,
      "loss": 2.4354,
      "step": 44511
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7644777894020081,
      "learning_rate": 0.0005464207977215847,
      "loss": 2.2836,
      "step": 44512
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.8596352338790894,
      "learning_rate": 0.0005464184646517456,
      "loss": 2.2822,
      "step": 44513
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.683529794216156,
      "learning_rate": 0.0005464161315360927,
      "loss": 2.0206,
      "step": 44514
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.6668747067451477,
      "learning_rate": 0.0005464137983746263,
      "loss": 2.0925,
      "step": 44515
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7948773503303528,
      "learning_rate": 0.0005464114651673471,
      "loss": 2.5605,
      "step": 44516
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7973371148109436,
      "learning_rate": 0.0005464091319142552,
      "loss": 2.4576,
      "step": 44517
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7320377826690674,
      "learning_rate": 0.0005464067986153513,
      "loss": 2.2126,
      "step": 44518
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.6929908394813538,
      "learning_rate": 0.0005464044652706356,
      "loss": 2.2742,
      "step": 44519
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.9865420460700989,
      "learning_rate": 0.0005464021318801088,
      "loss": 2.0512,
      "step": 44520
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.8666732311248779,
      "learning_rate": 0.0005463997984437711,
      "loss": 2.3574,
      "step": 44521
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.706039547920227,
      "learning_rate": 0.000546397464961623,
      "loss": 2.1356,
      "step": 44522
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7611104846000671,
      "learning_rate": 0.000546395131433665,
      "loss": 2.5359,
      "step": 44523
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.9994605183601379,
      "learning_rate": 0.0005463927978598975,
      "loss": 2.389,
      "step": 44524
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.702198326587677,
      "learning_rate": 0.000546390464240321,
      "loss": 2.3804,
      "step": 44525
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7207252979278564,
      "learning_rate": 0.0005463881305749358,
      "loss": 2.424,
      "step": 44526
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.9119575023651123,
      "learning_rate": 0.0005463857968637423,
      "loss": 2.0642,
      "step": 44527
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.8121318817138672,
      "learning_rate": 0.0005463834631067412,
      "loss": 2.4369,
      "step": 44528
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.6514591574668884,
      "learning_rate": 0.0005463811293039327,
      "loss": 2.1553,
      "step": 44529
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.9438997507095337,
      "learning_rate": 0.0005463787954553172,
      "loss": 2.2494,
      "step": 44530
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.75678551197052,
      "learning_rate": 0.0005463764615608953,
      "loss": 2.3452,
      "step": 44531
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.6827924847602844,
      "learning_rate": 0.0005463741276206675,
      "loss": 2.237,
      "step": 44532
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.6712067723274231,
      "learning_rate": 0.0005463717936346338,
      "loss": 2.3755,
      "step": 44533
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.6335271000862122,
      "learning_rate": 0.0005463694596027952,
      "loss": 2.1823,
      "step": 44534
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7142989635467529,
      "learning_rate": 0.0005463671255251518,
      "loss": 2.1344,
      "step": 44535
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7162178158760071,
      "learning_rate": 0.0005463647914017041,
      "loss": 2.4521,
      "step": 44536
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.6657367944717407,
      "learning_rate": 0.0005463624572324524,
      "loss": 2.4135,
      "step": 44537
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7458229660987854,
      "learning_rate": 0.0005463601230173975,
      "loss": 2.3427,
      "step": 44538
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.6744242310523987,
      "learning_rate": 0.0005463577887565394,
      "loss": 2.2897,
      "step": 44539
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7358576059341431,
      "learning_rate": 0.0005463554544498788,
      "loss": 2.2934,
      "step": 44540
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7112663388252258,
      "learning_rate": 0.0005463531200974161,
      "loss": 2.0916,
      "step": 44541
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.8367668390274048,
      "learning_rate": 0.0005463507856991517,
      "loss": 2.1436,
      "step": 44542
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.8416169881820679,
      "learning_rate": 0.000546348451255086,
      "loss": 2.2047,
      "step": 44543
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.6480495929718018,
      "learning_rate": 0.0005463461167652195,
      "loss": 2.1837,
      "step": 44544
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7315731048583984,
      "learning_rate": 0.0005463437822295526,
      "loss": 2.149,
      "step": 44545
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7270505428314209,
      "learning_rate": 0.0005463414476480857,
      "loss": 2.5642,
      "step": 44546
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7365325093269348,
      "learning_rate": 0.0005463391130208193,
      "loss": 2.3433,
      "step": 44547
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7750242948532104,
      "learning_rate": 0.000546336778347754,
      "loss": 2.3044,
      "step": 44548
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.6734107136726379,
      "learning_rate": 0.0005463344436288898,
      "loss": 2.2628,
      "step": 44549
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7043740153312683,
      "learning_rate": 0.0005463321088642275,
      "loss": 2.2998,
      "step": 44550
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7450931072235107,
      "learning_rate": 0.0005463297740537673,
      "loss": 2.1663,
      "step": 44551
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.6544045805931091,
      "learning_rate": 0.0005463274391975098,
      "loss": 2.3032,
      "step": 44552
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7353530526161194,
      "learning_rate": 0.0005463251042954554,
      "loss": 2.0438,
      "step": 44553
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7176347970962524,
      "learning_rate": 0.0005463227693476045,
      "loss": 2.382,
      "step": 44554
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.673984706401825,
      "learning_rate": 0.0005463204343539576,
      "loss": 2.2447,
      "step": 44555
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7914706468582153,
      "learning_rate": 0.000546318099314515,
      "loss": 2.3592,
      "step": 44556
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.6830021739006042,
      "learning_rate": 0.0005463157642292772,
      "loss": 2.5519,
      "step": 44557
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7905114889144897,
      "learning_rate": 0.0005463134290982448,
      "loss": 2.2417,
      "step": 44558
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.6587149500846863,
      "learning_rate": 0.0005463110939214179,
      "loss": 2.1694,
      "step": 44559
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7033205628395081,
      "learning_rate": 0.0005463087586987972,
      "loss": 2.3871,
      "step": 44560
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.6939679980278015,
      "learning_rate": 0.0005463064234303831,
      "loss": 2.194,
      "step": 44561
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.6393133997917175,
      "learning_rate": 0.0005463040881161759,
      "loss": 2.3499,
      "step": 44562
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.6575694680213928,
      "learning_rate": 0.0005463017527561762,
      "loss": 2.2761,
      "step": 44563
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7210869193077087,
      "learning_rate": 0.0005462994173503843,
      "loss": 2.1807,
      "step": 44564
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.6957069635391235,
      "learning_rate": 0.0005462970818988006,
      "loss": 2.4446,
      "step": 44565
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7099423408508301,
      "learning_rate": 0.0005462947464014258,
      "loss": 2.3096,
      "step": 44566
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7022475600242615,
      "learning_rate": 0.00054629241085826,
      "loss": 2.3951,
      "step": 44567
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7347371578216553,
      "learning_rate": 0.000546290075269304,
      "loss": 2.0795,
      "step": 44568
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7461672425270081,
      "learning_rate": 0.0005462877396345578,
      "loss": 2.255,
      "step": 44569
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.6747148633003235,
      "learning_rate": 0.0005462854039540222,
      "loss": 2.3092,
      "step": 44570
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7194972634315491,
      "learning_rate": 0.0005462830682276975,
      "loss": 2.3164,
      "step": 44571
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.6892818808555603,
      "learning_rate": 0.000546280732455584,
      "loss": 2.3094,
      "step": 44572
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7255939245223999,
      "learning_rate": 0.0005462783966376824,
      "loss": 2.3219,
      "step": 44573
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.6754827499389648,
      "learning_rate": 0.000546276060773993,
      "loss": 2.1628,
      "step": 44574
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.6740314364433289,
      "learning_rate": 0.0005462737248645161,
      "loss": 2.2389,
      "step": 44575
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7580535411834717,
      "learning_rate": 0.0005462713889092523,
      "loss": 2.2993,
      "step": 44576
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.6120090484619141,
      "learning_rate": 0.0005462690529082022,
      "loss": 2.1525,
      "step": 44577
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7034280896186829,
      "learning_rate": 0.0005462667168613657,
      "loss": 2.3016,
      "step": 44578
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.6691897511482239,
      "learning_rate": 0.0005462643807687439,
      "loss": 2.1998,
      "step": 44579
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.799318790435791,
      "learning_rate": 0.0005462620446303367,
      "loss": 2.2865,
      "step": 44580
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.6668564677238464,
      "learning_rate": 0.0005462597084461448,
      "loss": 2.3199,
      "step": 44581
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.6573787927627563,
      "learning_rate": 0.0005462573722161686,
      "loss": 1.915,
      "step": 44582
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.6838036775588989,
      "learning_rate": 0.0005462550359404084,
      "loss": 2.115,
      "step": 44583
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.8028521537780762,
      "learning_rate": 0.0005462526996188648,
      "loss": 2.1542,
      "step": 44584
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.6846592426300049,
      "learning_rate": 0.0005462503632515381,
      "loss": 2.414,
      "step": 44585
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7030662894248962,
      "learning_rate": 0.000546248026838429,
      "loss": 2.0959,
      "step": 44586
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.8291717767715454,
      "learning_rate": 0.0005462456903795376,
      "loss": 2.3311,
      "step": 44587
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.6469507813453674,
      "learning_rate": 0.0005462433538748645,
      "loss": 2.4851,
      "step": 44588
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7988024353981018,
      "learning_rate": 0.0005462410173244101,
      "loss": 2.1609,
      "step": 44589
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.6707012057304382,
      "learning_rate": 0.0005462386807281749,
      "loss": 2.5224,
      "step": 44590
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.6546673774719238,
      "learning_rate": 0.0005462363440861593,
      "loss": 2.1227,
      "step": 44591
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7240211367607117,
      "learning_rate": 0.0005462340073983636,
      "loss": 2.1659,
      "step": 44592
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7337819337844849,
      "learning_rate": 0.0005462316706647883,
      "loss": 2.0885,
      "step": 44593
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7438878417015076,
      "learning_rate": 0.0005462293338854341,
      "loss": 2.2863,
      "step": 44594
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.8014870882034302,
      "learning_rate": 0.0005462269970603011,
      "loss": 2.2318,
      "step": 44595
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7666529417037964,
      "learning_rate": 0.0005462246601893899,
      "loss": 2.1904,
      "step": 44596
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7361413836479187,
      "learning_rate": 0.0005462223232727009,
      "loss": 2.541,
      "step": 44597
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7408473491668701,
      "learning_rate": 0.0005462199863102345,
      "loss": 2.2071,
      "step": 44598
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7576781511306763,
      "learning_rate": 0.0005462176493019911,
      "loss": 2.0885,
      "step": 44599
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.6656572222709656,
      "learning_rate": 0.0005462153122479711,
      "loss": 2.5048,
      "step": 44600
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7638875842094421,
      "learning_rate": 0.0005462129751481752,
      "loss": 2.3245,
      "step": 44601
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.681236207485199,
      "learning_rate": 0.0005462106380026036,
      "loss": 2.1714,
      "step": 44602
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.6907824873924255,
      "learning_rate": 0.0005462083008112569,
      "loss": 2.1322,
      "step": 44603
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.8527239561080933,
      "learning_rate": 0.0005462059635741353,
      "loss": 2.3785,
      "step": 44604
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.6744069457054138,
      "learning_rate": 0.0005462036262912395,
      "loss": 2.3036,
      "step": 44605
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.6943528652191162,
      "learning_rate": 0.0005462012889625696,
      "loss": 2.2277,
      "step": 44606
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7122798562049866,
      "learning_rate": 0.0005461989515881263,
      "loss": 2.121,
      "step": 44607
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.6714604496955872,
      "learning_rate": 0.0005461966141679101,
      "loss": 2.211,
      "step": 44608
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7147833108901978,
      "learning_rate": 0.0005461942767019212,
      "loss": 2.1587,
      "step": 44609
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7168949246406555,
      "learning_rate": 0.0005461919391901601,
      "loss": 2.4959,
      "step": 44610
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.6950842142105103,
      "learning_rate": 0.0005461896016326273,
      "loss": 2.078,
      "step": 44611
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.6764770150184631,
      "learning_rate": 0.0005461872640293233,
      "loss": 2.3515,
      "step": 44612
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.6546424031257629,
      "learning_rate": 0.0005461849263802482,
      "loss": 2.1848,
      "step": 44613
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.6924923062324524,
      "learning_rate": 0.0005461825886854029,
      "loss": 2.2009,
      "step": 44614
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7241561412811279,
      "learning_rate": 0.0005461802509447876,
      "loss": 2.315,
      "step": 44615
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.8399633169174194,
      "learning_rate": 0.0005461779131584026,
      "loss": 2.3112,
      "step": 44616
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.9028435349464417,
      "learning_rate": 0.0005461755753262486,
      "loss": 2.171,
      "step": 44617
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.6701582074165344,
      "learning_rate": 0.0005461732374483259,
      "loss": 2.208,
      "step": 44618
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7024791836738586,
      "learning_rate": 0.0005461708995246349,
      "loss": 2.5495,
      "step": 44619
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7474980354309082,
      "learning_rate": 0.0005461685615551761,
      "loss": 2.0709,
      "step": 44620
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7375984787940979,
      "learning_rate": 0.00054616622353995,
      "loss": 2.0933,
      "step": 44621
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.8117192387580872,
      "learning_rate": 0.0005461638854789569,
      "loss": 2.3371,
      "step": 44622
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7618316411972046,
      "learning_rate": 0.0005461615473721972,
      "loss": 2.4082,
      "step": 44623
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.6800745129585266,
      "learning_rate": 0.0005461592092196715,
      "loss": 2.3123,
      "step": 44624
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.8361275792121887,
      "learning_rate": 0.0005461568710213801,
      "loss": 2.0846,
      "step": 44625
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.695895791053772,
      "learning_rate": 0.0005461545327773235,
      "loss": 2.1932,
      "step": 44626
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7249265313148499,
      "learning_rate": 0.0005461521944875022,
      "loss": 2.2694,
      "step": 44627
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7432135343551636,
      "learning_rate": 0.0005461498561519166,
      "loss": 2.4149,
      "step": 44628
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.6405037045478821,
      "learning_rate": 0.0005461475177705671,
      "loss": 2.2889,
      "step": 44629
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7829313278198242,
      "learning_rate": 0.000546145179343454,
      "loss": 2.3198,
      "step": 44630
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.6801192760467529,
      "learning_rate": 0.0005461428408705779,
      "loss": 2.4596,
      "step": 44631
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7593603730201721,
      "learning_rate": 0.0005461405023519391,
      "loss": 2.3476,
      "step": 44632
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7713959813117981,
      "learning_rate": 0.0005461381637875384,
      "loss": 2.3963,
      "step": 44633
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7449427843093872,
      "learning_rate": 0.0005461358251773758,
      "loss": 2.3739,
      "step": 44634
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.6606417298316956,
      "learning_rate": 0.0005461334865214519,
      "loss": 2.2963,
      "step": 44635
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.925497829914093,
      "learning_rate": 0.0005461311478197671,
      "loss": 2.3722,
      "step": 44636
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.8054125308990479,
      "learning_rate": 0.0005461288090723219,
      "loss": 2.3101,
      "step": 44637
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7508938908576965,
      "learning_rate": 0.0005461264702791168,
      "loss": 2.0542,
      "step": 44638
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7649844884872437,
      "learning_rate": 0.0005461241314401521,
      "loss": 2.2974,
      "step": 44639
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7237818837165833,
      "learning_rate": 0.0005461217925554283,
      "loss": 2.0164,
      "step": 44640
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7234633564949036,
      "learning_rate": 0.0005461194536249457,
      "loss": 2.3172,
      "step": 44641
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7133338451385498,
      "learning_rate": 0.000546117114648705,
      "loss": 2.3917,
      "step": 44642
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.661809504032135,
      "learning_rate": 0.0005461147756267063,
      "loss": 2.2318,
      "step": 44643
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7001239061355591,
      "learning_rate": 0.0005461124365589504,
      "loss": 2.6669,
      "step": 44644
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7022026181221008,
      "learning_rate": 0.0005461100974454374,
      "loss": 2.2946,
      "step": 44645
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.6528114676475525,
      "learning_rate": 0.000546107758286168,
      "loss": 2.3226,
      "step": 44646
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7881191968917847,
      "learning_rate": 0.0005461054190811425,
      "loss": 2.3686,
      "step": 44647
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.6843011379241943,
      "learning_rate": 0.0005461030798303613,
      "loss": 2.1898,
      "step": 44648
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.6516548991203308,
      "learning_rate": 0.0005461007405338249,
      "loss": 2.4545,
      "step": 44649
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.6516534686088562,
      "learning_rate": 0.0005460984011915338,
      "loss": 2.1292,
      "step": 44650
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.6650488376617432,
      "learning_rate": 0.0005460960618034884,
      "loss": 1.9945,
      "step": 44651
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.743994414806366,
      "learning_rate": 0.000546093722369689,
      "loss": 2.3025,
      "step": 44652
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.694254994392395,
      "learning_rate": 0.0005460913828901361,
      "loss": 2.3279,
      "step": 44653
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7222999334335327,
      "learning_rate": 0.0005460890433648302,
      "loss": 2.5198,
      "step": 44654
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7447350025177002,
      "learning_rate": 0.0005460867037937717,
      "loss": 2.3201,
      "step": 44655
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.6315982341766357,
      "learning_rate": 0.0005460843641769611,
      "loss": 2.4082,
      "step": 44656
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7457296252250671,
      "learning_rate": 0.0005460820245143987,
      "loss": 2.4032,
      "step": 44657
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.6894057393074036,
      "learning_rate": 0.0005460796848060851,
      "loss": 2.1704,
      "step": 44658
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.6999388337135315,
      "learning_rate": 0.0005460773450520205,
      "loss": 2.3844,
      "step": 44659
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.6958493590354919,
      "learning_rate": 0.0005460750052522056,
      "loss": 2.2154,
      "step": 44660
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.726588785648346,
      "learning_rate": 0.0005460726654066407,
      "loss": 2.3576,
      "step": 44661
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7815188765525818,
      "learning_rate": 0.0005460703255153261,
      "loss": 2.4661,
      "step": 44662
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.8110120296478271,
      "learning_rate": 0.0005460679855782626,
      "loss": 2.1483,
      "step": 44663
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.8443973660469055,
      "learning_rate": 0.0005460656455954502,
      "loss": 2.3637,
      "step": 44664
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7709261775016785,
      "learning_rate": 0.0005460633055668897,
      "loss": 2.3133,
      "step": 44665
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7126791477203369,
      "learning_rate": 0.0005460609654925814,
      "loss": 2.2016,
      "step": 44666
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.1161853075027466,
      "learning_rate": 0.0005460586253725257,
      "loss": 2.2304,
      "step": 44667
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7441393136978149,
      "learning_rate": 0.000546056285206723,
      "loss": 2.3514,
      "step": 44668
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7203887701034546,
      "learning_rate": 0.0005460539449951738,
      "loss": 2.2069,
      "step": 44669
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.8780645728111267,
      "learning_rate": 0.0005460516047378786,
      "loss": 2.3821,
      "step": 44670
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7834069728851318,
      "learning_rate": 0.0005460492644348376,
      "loss": 2.1437,
      "step": 44671
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.8230350613594055,
      "learning_rate": 0.0005460469240860515,
      "loss": 2.442,
      "step": 44672
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7343173623085022,
      "learning_rate": 0.0005460445836915207,
      "loss": 2.2041,
      "step": 44673
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.8949748277664185,
      "learning_rate": 0.0005460422432512455,
      "loss": 2.2481,
      "step": 44674
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7303478121757507,
      "learning_rate": 0.0005460399027652263,
      "loss": 2.2206,
      "step": 44675
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.722494900226593,
      "learning_rate": 0.0005460375622334638,
      "loss": 2.3687,
      "step": 44676
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.6838162541389465,
      "learning_rate": 0.0005460352216559583,
      "loss": 2.0992,
      "step": 44677
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.8012865781784058,
      "learning_rate": 0.00054603288103271,
      "loss": 2.2901,
      "step": 44678
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.6610217094421387,
      "learning_rate": 0.0005460305403637198,
      "loss": 2.14,
      "step": 44679
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7457250952720642,
      "learning_rate": 0.0005460281996489877,
      "loss": 2.1421,
      "step": 44680
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.8237011432647705,
      "learning_rate": 0.0005460258588885144,
      "loss": 2.2542,
      "step": 44681
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.6756654381752014,
      "learning_rate": 0.0005460235180823001,
      "loss": 2.3794,
      "step": 44682
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7209292650222778,
      "learning_rate": 0.0005460211772303455,
      "loss": 2.0981,
      "step": 44683
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7995787858963013,
      "learning_rate": 0.0005460188363326509,
      "loss": 2.283,
      "step": 44684
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7140093445777893,
      "learning_rate": 0.0005460164953892168,
      "loss": 2.4908,
      "step": 44685
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.693683385848999,
      "learning_rate": 0.0005460141544000436,
      "loss": 2.284,
      "step": 44686
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7445428371429443,
      "learning_rate": 0.0005460118133651316,
      "loss": 2.2736,
      "step": 44687
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.8159094452857971,
      "learning_rate": 0.0005460094722844815,
      "loss": 2.4707,
      "step": 44688
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7178479433059692,
      "learning_rate": 0.0005460071311580935,
      "loss": 2.5195,
      "step": 44689
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.6850244998931885,
      "learning_rate": 0.0005460047899859681,
      "loss": 2.2779,
      "step": 44690
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7810165286064148,
      "learning_rate": 0.000546002448768106,
      "loss": 2.0163,
      "step": 44691
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.6877259612083435,
      "learning_rate": 0.0005460001075045072,
      "loss": 2.2436,
      "step": 44692
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7582355737686157,
      "learning_rate": 0.0005459977661951723,
      "loss": 2.1459,
      "step": 44693
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.6478493809700012,
      "learning_rate": 0.0005459954248401019,
      "loss": 2.4537,
      "step": 44694
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.6669360995292664,
      "learning_rate": 0.0005459930834392962,
      "loss": 2.1513,
      "step": 44695
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7616193890571594,
      "learning_rate": 0.0005459907419927558,
      "loss": 2.2268,
      "step": 44696
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7127373814582825,
      "learning_rate": 0.000545988400500481,
      "loss": 2.1825,
      "step": 44697
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.6652586460113525,
      "learning_rate": 0.0005459860589624724,
      "loss": 2.1638,
      "step": 44698
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7233163118362427,
      "learning_rate": 0.0005459837173787303,
      "loss": 2.3379,
      "step": 44699
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7161201238632202,
      "learning_rate": 0.0005459813757492552,
      "loss": 2.3983,
      "step": 44700
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7035424709320068,
      "learning_rate": 0.0005459790340740477,
      "loss": 2.2668,
      "step": 44701
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.6226714253425598,
      "learning_rate": 0.0005459766923531078,
      "loss": 1.9926,
      "step": 44702
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.6574909687042236,
      "learning_rate": 0.0005459743505864363,
      "loss": 2.1058,
      "step": 44703
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.6503353714942932,
      "learning_rate": 0.0005459720087740335,
      "loss": 2.2536,
      "step": 44704
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.0023245811462402,
      "learning_rate": 0.0005459696669158998,
      "loss": 2.512,
      "step": 44705
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7419370412826538,
      "learning_rate": 0.0005459673250120358,
      "loss": 2.2645,
      "step": 44706
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.6874369382858276,
      "learning_rate": 0.0005459649830624418,
      "loss": 2.0135,
      "step": 44707
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7055485844612122,
      "learning_rate": 0.0005459626410671182,
      "loss": 2.2254,
      "step": 44708
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.6888337731361389,
      "learning_rate": 0.0005459602990260656,
      "loss": 2.3733,
      "step": 44709
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.6789257526397705,
      "learning_rate": 0.0005459579569392842,
      "loss": 2.4993,
      "step": 44710
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.6744623184204102,
      "learning_rate": 0.0005459556148067747,
      "loss": 2.2535,
      "step": 44711
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7241822481155396,
      "learning_rate": 0.0005459532726285374,
      "loss": 2.0244,
      "step": 44712
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7049328684806824,
      "learning_rate": 0.0005459509304045727,
      "loss": 2.3289,
      "step": 44713
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7141597867012024,
      "learning_rate": 0.0005459485881348812,
      "loss": 2.3848,
      "step": 44714
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.6451884508132935,
      "learning_rate": 0.000545946245819463,
      "loss": 2.1406,
      "step": 44715
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.699639081954956,
      "learning_rate": 0.0005459439034583189,
      "loss": 2.2862,
      "step": 44716
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.6394093036651611,
      "learning_rate": 0.0005459415610514491,
      "loss": 2.2904,
      "step": 44717
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.6723127365112305,
      "learning_rate": 0.0005459392185988542,
      "loss": 2.1418,
      "step": 44718
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.8120065331459045,
      "learning_rate": 0.0005459368761005346,
      "loss": 2.2591,
      "step": 44719
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7169106006622314,
      "learning_rate": 0.0005459345335564906,
      "loss": 2.2638,
      "step": 44720
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.627815842628479,
      "learning_rate": 0.0005459321909667227,
      "loss": 1.9217,
      "step": 44721
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7346104979515076,
      "learning_rate": 0.0005459298483312314,
      "loss": 2.4004,
      "step": 44722
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7261823415756226,
      "learning_rate": 0.0005459275056500172,
      "loss": 2.2835,
      "step": 44723
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7483221888542175,
      "learning_rate": 0.0005459251629230804,
      "loss": 2.4023,
      "step": 44724
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.6774993538856506,
      "learning_rate": 0.0005459228201504213,
      "loss": 2.2527,
      "step": 44725
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.6981993317604065,
      "learning_rate": 0.0005459204773320406,
      "loss": 2.2033,
      "step": 44726
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7040091156959534,
      "learning_rate": 0.0005459181344679387,
      "loss": 2.313,
      "step": 44727
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7715952396392822,
      "learning_rate": 0.0005459157915581158,
      "loss": 2.0986,
      "step": 44728
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.710487425327301,
      "learning_rate": 0.0005459134486025727,
      "loss": 2.2064,
      "step": 44729
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7128447890281677,
      "learning_rate": 0.0005459111056013097,
      "loss": 2.3262,
      "step": 44730
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.874363899230957,
      "learning_rate": 0.0005459087625543271,
      "loss": 2.3854,
      "step": 44731
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.6632083058357239,
      "learning_rate": 0.0005459064194616253,
      "loss": 2.3121,
      "step": 44732
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7694461345672607,
      "learning_rate": 0.0005459040763232051,
      "loss": 2.2988,
      "step": 44733
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7216692566871643,
      "learning_rate": 0.0005459017331390665,
      "loss": 2.173,
      "step": 44734
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7031180262565613,
      "learning_rate": 0.0005458993899092101,
      "loss": 2.0991,
      "step": 44735
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.64413982629776,
      "learning_rate": 0.0005458970466336365,
      "loss": 2.2939,
      "step": 44736
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.6326048970222473,
      "learning_rate": 0.0005458947033123459,
      "loss": 2.2593,
      "step": 44737
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7444490790367126,
      "learning_rate": 0.0005458923599453389,
      "loss": 2.2874,
      "step": 44738
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7503328919410706,
      "learning_rate": 0.0005458900165326158,
      "loss": 2.0899,
      "step": 44739
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.691208004951477,
      "learning_rate": 0.0005458876730741771,
      "loss": 2.1494,
      "step": 44740
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.706589937210083,
      "learning_rate": 0.0005458853295700233,
      "loss": 2.4481,
      "step": 44741
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7147359848022461,
      "learning_rate": 0.0005458829860201548,
      "loss": 2.2884,
      "step": 44742
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.8266469836235046,
      "learning_rate": 0.0005458806424245719,
      "loss": 2.2948,
      "step": 44743
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.8265838623046875,
      "learning_rate": 0.0005458782987832752,
      "loss": 2.2499,
      "step": 44744
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.6915833353996277,
      "learning_rate": 0.0005458759550962651,
      "loss": 2.269,
      "step": 44745
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.8542684316635132,
      "learning_rate": 0.000545873611363542,
      "loss": 2.2164,
      "step": 44746
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7633109092712402,
      "learning_rate": 0.0005458712675851064,
      "loss": 2.2821,
      "step": 44747
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7067098617553711,
      "learning_rate": 0.0005458689237609586,
      "loss": 2.372,
      "step": 44748
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.6838030815124512,
      "learning_rate": 0.0005458665798910992,
      "loss": 2.1831,
      "step": 44749
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7764460444450378,
      "learning_rate": 0.0005458642359755286,
      "loss": 2.1843,
      "step": 44750
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7408024668693542,
      "learning_rate": 0.0005458618920142472,
      "loss": 2.1764,
      "step": 44751
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.726857602596283,
      "learning_rate": 0.0005458595480072553,
      "loss": 2.1699,
      "step": 44752
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7519277930259705,
      "learning_rate": 0.0005458572039545535,
      "loss": 2.3097,
      "step": 44753
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7034433484077454,
      "learning_rate": 0.0005458548598561423,
      "loss": 2.5677,
      "step": 44754
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.6951050162315369,
      "learning_rate": 0.000545852515712022,
      "loss": 2.3125,
      "step": 44755
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.6608263850212097,
      "learning_rate": 0.0005458501715221931,
      "loss": 2.2668,
      "step": 44756
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7155313491821289,
      "learning_rate": 0.000545847827286656,
      "loss": 2.1591,
      "step": 44757
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7661635875701904,
      "learning_rate": 0.000545845483005411,
      "loss": 2.1318,
      "step": 44758
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7392632961273193,
      "learning_rate": 0.000545843138678459,
      "loss": 2.1825,
      "step": 44759
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.8868210315704346,
      "learning_rate": 0.0005458407943057999,
      "loss": 2.3199,
      "step": 44760
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7167781591415405,
      "learning_rate": 0.0005458384498874343,
      "loss": 2.2242,
      "step": 44761
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.6002246141433716,
      "learning_rate": 0.0005458361054233628,
      "loss": 2.45,
      "step": 44762
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7469065189361572,
      "learning_rate": 0.0005458337609135857,
      "loss": 2.417,
      "step": 44763
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.6895849108695984,
      "learning_rate": 0.0005458314163581035,
      "loss": 2.1005,
      "step": 44764
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.6765480041503906,
      "learning_rate": 0.0005458290717569166,
      "loss": 2.2508,
      "step": 44765
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.764735221862793,
      "learning_rate": 0.0005458267271100253,
      "loss": 2.3456,
      "step": 44766
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.6830411553382874,
      "learning_rate": 0.0005458243824174304,
      "loss": 2.2366,
      "step": 44767
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.6960754990577698,
      "learning_rate": 0.0005458220376791319,
      "loss": 2.0765,
      "step": 44768
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7356778979301453,
      "learning_rate": 0.0005458196928951305,
      "loss": 2.2719,
      "step": 44769
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.6821801066398621,
      "learning_rate": 0.0005458173480654266,
      "loss": 2.2024,
      "step": 44770
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.667545735836029,
      "learning_rate": 0.0005458150031900207,
      "loss": 1.9812,
      "step": 44771
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7349767684936523,
      "learning_rate": 0.000545812658268913,
      "loss": 2.3072,
      "step": 44772
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7113093733787537,
      "learning_rate": 0.0005458103133021042,
      "loss": 2.292,
      "step": 44773
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7538228631019592,
      "learning_rate": 0.0005458079682895946,
      "loss": 2.3525,
      "step": 44774
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7249446511268616,
      "learning_rate": 0.0005458056232313846,
      "loss": 2.2277,
      "step": 44775
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7196481227874756,
      "learning_rate": 0.0005458032781274747,
      "loss": 2.3059,
      "step": 44776
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7451549172401428,
      "learning_rate": 0.0005458009329778655,
      "loss": 2.1844,
      "step": 44777
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.6655228137969971,
      "learning_rate": 0.0005457985877825571,
      "loss": 2.1596,
      "step": 44778
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.6415429711341858,
      "learning_rate": 0.0005457962425415501,
      "loss": 2.2904,
      "step": 44779
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.6889739632606506,
      "learning_rate": 0.000545793897254845,
      "loss": 2.3224,
      "step": 44780
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.6523335576057434,
      "learning_rate": 0.0005457915519224421,
      "loss": 2.3458,
      "step": 44781
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7322390675544739,
      "learning_rate": 0.0005457892065443419,
      "loss": 2.2107,
      "step": 44782
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.6423497796058655,
      "learning_rate": 0.000545786861120545,
      "loss": 2.12,
      "step": 44783
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7200359106063843,
      "learning_rate": 0.0005457845156510516,
      "loss": 2.4975,
      "step": 44784
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7361637949943542,
      "learning_rate": 0.0005457821701358622,
      "loss": 2.3743,
      "step": 44785
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.6619354486465454,
      "learning_rate": 0.0005457798245749772,
      "loss": 2.3113,
      "step": 44786
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7744871973991394,
      "learning_rate": 0.0005457774789683972,
      "loss": 2.2094,
      "step": 44787
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7116063237190247,
      "learning_rate": 0.0005457751333161224,
      "loss": 2.2091,
      "step": 44788
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.685309886932373,
      "learning_rate": 0.0005457727876181535,
      "loss": 2.2101,
      "step": 44789
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7348514795303345,
      "learning_rate": 0.0005457704418744908,
      "loss": 2.5049,
      "step": 44790
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7328616976737976,
      "learning_rate": 0.0005457680960851346,
      "loss": 2.1546,
      "step": 44791
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7508512139320374,
      "learning_rate": 0.0005457657502500855,
      "loss": 2.3754,
      "step": 44792
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7341334223747253,
      "learning_rate": 0.000545763404369344,
      "loss": 2.4538,
      "step": 44793
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.6976367831230164,
      "learning_rate": 0.0005457610584429104,
      "loss": 2.5423,
      "step": 44794
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.6786804795265198,
      "learning_rate": 0.0005457587124707852,
      "loss": 2.5074,
      "step": 44795
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.6408016085624695,
      "learning_rate": 0.0005457563664529688,
      "loss": 2.3143,
      "step": 44796
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7079795598983765,
      "learning_rate": 0.0005457540203894616,
      "loss": 2.0741,
      "step": 44797
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.8420699834823608,
      "learning_rate": 0.0005457516742802641,
      "loss": 2.1946,
      "step": 44798
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.6511228680610657,
      "learning_rate": 0.0005457493281253769,
      "loss": 2.2112,
      "step": 44799
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.8436301350593567,
      "learning_rate": 0.0005457469819248001,
      "loss": 2.441,
      "step": 44800
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7911840677261353,
      "learning_rate": 0.0005457446356785343,
      "loss": 2.3388,
      "step": 44801
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.6843822002410889,
      "learning_rate": 0.00054574228938658,
      "loss": 2.1327,
      "step": 44802
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.6951562762260437,
      "learning_rate": 0.0005457399430489375,
      "loss": 2.2728,
      "step": 44803
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7422076463699341,
      "learning_rate": 0.0005457375966656074,
      "loss": 2.4431,
      "step": 44804
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7358139753341675,
      "learning_rate": 0.0005457352502365899,
      "loss": 2.3915,
      "step": 44805
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.6624081134796143,
      "learning_rate": 0.0005457329037618856,
      "loss": 2.4341,
      "step": 44806
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.67742520570755,
      "learning_rate": 0.000545730557241495,
      "loss": 2.2904,
      "step": 44807
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.6751037836074829,
      "learning_rate": 0.0005457282106754184,
      "loss": 2.419,
      "step": 44808
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.6544645428657532,
      "learning_rate": 0.0005457258640636563,
      "loss": 2.1067,
      "step": 44809
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.6995534896850586,
      "learning_rate": 0.0005457235174062092,
      "loss": 2.3463,
      "step": 44810
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.6686811447143555,
      "learning_rate": 0.0005457211707030773,
      "loss": 2.3713,
      "step": 44811
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.724818766117096,
      "learning_rate": 0.0005457188239542613,
      "loss": 2.211,
      "step": 44812
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.6899328827857971,
      "learning_rate": 0.0005457164771597616,
      "loss": 2.2171,
      "step": 44813
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.6309057474136353,
      "learning_rate": 0.0005457141303195785,
      "loss": 2.3193,
      "step": 44814
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7044705748558044,
      "learning_rate": 0.0005457117834337125,
      "loss": 2.1669,
      "step": 44815
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.6653148531913757,
      "learning_rate": 0.000545709436502164,
      "loss": 2.4165,
      "step": 44816
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7863739728927612,
      "learning_rate": 0.0005457070895249335,
      "loss": 2.0489,
      "step": 44817
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.6578167676925659,
      "learning_rate": 0.0005457047425020214,
      "loss": 2.3135,
      "step": 44818
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.6972320675849915,
      "learning_rate": 0.0005457023954334282,
      "loss": 2.0456,
      "step": 44819
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7509982585906982,
      "learning_rate": 0.0005457000483191542,
      "loss": 2.4542,
      "step": 44820
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7024261951446533,
      "learning_rate": 0.0005456977011592,
      "loss": 2.2605,
      "step": 44821
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7024047374725342,
      "learning_rate": 0.000545695353953566,
      "loss": 2.528,
      "step": 44822
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7004929184913635,
      "learning_rate": 0.0005456930067022526,
      "loss": 2.0263,
      "step": 44823
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.6798997521400452,
      "learning_rate": 0.00054569065940526,
      "loss": 2.0576,
      "step": 44824
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.6333369612693787,
      "learning_rate": 0.0005456883120625891,
      "loss": 2.2684,
      "step": 44825
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7520152926445007,
      "learning_rate": 0.0005456859646742401,
      "loss": 2.3558,
      "step": 44826
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.686702311038971,
      "learning_rate": 0.0005456836172402133,
      "loss": 2.3966,
      "step": 44827
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.6709418892860413,
      "learning_rate": 0.0005456812697605093,
      "loss": 2.3344,
      "step": 44828
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.6637357473373413,
      "learning_rate": 0.0005456789222351286,
      "loss": 2.2925,
      "step": 44829
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.6671899557113647,
      "learning_rate": 0.0005456765746640716,
      "loss": 1.9143,
      "step": 44830
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.8839724063873291,
      "learning_rate": 0.0005456742270473385,
      "loss": 2.0999,
      "step": 44831
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7704799771308899,
      "learning_rate": 0.0005456718793849302,
      "loss": 2.2354,
      "step": 44832
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7115027904510498,
      "learning_rate": 0.0005456695316768467,
      "loss": 2.2208,
      "step": 44833
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7315904498100281,
      "learning_rate": 0.0005456671839230886,
      "loss": 2.2615,
      "step": 44834
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7476598620414734,
      "learning_rate": 0.0005456648361236563,
      "loss": 2.0536,
      "step": 44835
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.6929086446762085,
      "learning_rate": 0.0005456624882785503,
      "loss": 2.0125,
      "step": 44836
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7684639692306519,
      "learning_rate": 0.000545660140387771,
      "loss": 2.3032,
      "step": 44837
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7255077958106995,
      "learning_rate": 0.0005456577924513188,
      "loss": 2.2412,
      "step": 44838
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.6618762016296387,
      "learning_rate": 0.0005456554444691943,
      "loss": 2.1131,
      "step": 44839
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.6920608878135681,
      "learning_rate": 0.0005456530964413977,
      "loss": 2.3534,
      "step": 44840
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.6858688592910767,
      "learning_rate": 0.0005456507483679296,
      "loss": 2.4656,
      "step": 44841
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7093908190727234,
      "learning_rate": 0.0005456484002487904,
      "loss": 2.2597,
      "step": 44842
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7009816765785217,
      "learning_rate": 0.0005456460520839805,
      "loss": 2.3027,
      "step": 44843
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.666274905204773,
      "learning_rate": 0.0005456437038735003,
      "loss": 2.5038,
      "step": 44844
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7623775005340576,
      "learning_rate": 0.0005456413556173505,
      "loss": 2.3102,
      "step": 44845
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.6566244959831238,
      "learning_rate": 0.0005456390073155312,
      "loss": 2.3077,
      "step": 44846
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7792502045631409,
      "learning_rate": 0.000545636658968043,
      "loss": 2.5348,
      "step": 44847
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7094908952713013,
      "learning_rate": 0.0005456343105748863,
      "loss": 2.4133,
      "step": 44848
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7662608623504639,
      "learning_rate": 0.0005456319621360613,
      "loss": 2.1678,
      "step": 44849
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.8002660870552063,
      "learning_rate": 0.0005456296136515691,
      "loss": 2.1853,
      "step": 44850
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7070325016975403,
      "learning_rate": 0.0005456272651214094,
      "loss": 2.2165,
      "step": 44851
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.6263552308082581,
      "learning_rate": 0.0005456249165455831,
      "loss": 2.262,
      "step": 44852
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.6999699473381042,
      "learning_rate": 0.0005456225679240905,
      "loss": 2.3043,
      "step": 44853
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7170262336730957,
      "learning_rate": 0.000545620219256932,
      "loss": 2.0558,
      "step": 44854
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.6668564081192017,
      "learning_rate": 0.000545617870544108,
      "loss": 2.2834,
      "step": 44855
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7540294528007507,
      "learning_rate": 0.0005456155217856191,
      "loss": 2.2016,
      "step": 44856
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7021688222885132,
      "learning_rate": 0.0005456131729814658,
      "loss": 2.5504,
      "step": 44857
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7339368462562561,
      "learning_rate": 0.0005456108241316481,
      "loss": 2.3524,
      "step": 44858
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.742358922958374,
      "learning_rate": 0.0005456084752361668,
      "loss": 2.2301,
      "step": 44859
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7130964994430542,
      "learning_rate": 0.0005456061262950222,
      "loss": 2.407,
      "step": 44860
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.6870601773262024,
      "learning_rate": 0.0005456037773082148,
      "loss": 2.2502,
      "step": 44861
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.6276056170463562,
      "learning_rate": 0.0005456014282757451,
      "loss": 2.1036,
      "step": 44862
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7320499420166016,
      "learning_rate": 0.0005455990791976134,
      "loss": 2.3241,
      "step": 44863
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7481369376182556,
      "learning_rate": 0.0005455967300738203,
      "loss": 2.2187,
      "step": 44864
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7340972423553467,
      "learning_rate": 0.0005455943809043659,
      "loss": 2.395,
      "step": 44865
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.6814285516738892,
      "learning_rate": 0.0005455920316892511,
      "loss": 2.3868,
      "step": 44866
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7526198029518127,
      "learning_rate": 0.000545589682428476,
      "loss": 2.1539,
      "step": 44867
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.744037389755249,
      "learning_rate": 0.0005455873331220412,
      "loss": 2.2377,
      "step": 44868
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7328324317932129,
      "learning_rate": 0.0005455849837699469,
      "loss": 2.1707,
      "step": 44869
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7063367962837219,
      "learning_rate": 0.000545582634372194,
      "loss": 2.3208,
      "step": 44870
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.718467652797699,
      "learning_rate": 0.0005455802849287824,
      "loss": 2.2866,
      "step": 44871
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.6763586401939392,
      "learning_rate": 0.0005455779354397129,
      "loss": 2.1115,
      "step": 44872
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7761644124984741,
      "learning_rate": 0.0005455755859049857,
      "loss": 2.4269,
      "step": 44873
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.784768283367157,
      "learning_rate": 0.0005455732363246016,
      "loss": 2.1602,
      "step": 44874
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.6689361333847046,
      "learning_rate": 0.0005455708866985607,
      "loss": 1.9895,
      "step": 44875
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7632946968078613,
      "learning_rate": 0.0005455685370268635,
      "loss": 2.1892,
      "step": 44876
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.8721001744270325,
      "learning_rate": 0.0005455661873095104,
      "loss": 2.5495,
      "step": 44877
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7288481593132019,
      "learning_rate": 0.0005455638375465021,
      "loss": 2.2188,
      "step": 44878
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.8259783387184143,
      "learning_rate": 0.0005455614877378388,
      "loss": 2.3981,
      "step": 44879
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.9203808307647705,
      "learning_rate": 0.0005455591378835208,
      "loss": 2.0852,
      "step": 44880
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7172853350639343,
      "learning_rate": 0.0005455567879835488,
      "loss": 2.4699,
      "step": 44881
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.6721705198287964,
      "learning_rate": 0.0005455544380379233,
      "loss": 2.2148,
      "step": 44882
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7223548293113708,
      "learning_rate": 0.0005455520880466446,
      "loss": 2.2143,
      "step": 44883
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.752269983291626,
      "learning_rate": 0.0005455497380097129,
      "loss": 2.0906,
      "step": 44884
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.6833389401435852,
      "learning_rate": 0.0005455473879271291,
      "loss": 2.2556,
      "step": 44885
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7297079563140869,
      "learning_rate": 0.0005455450377988932,
      "loss": 2.2919,
      "step": 44886
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7029249668121338,
      "learning_rate": 0.000545542687625006,
      "loss": 2.3378,
      "step": 44887
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.6853296160697937,
      "learning_rate": 0.0005455403374054677,
      "loss": 2.2991,
      "step": 44888
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7036734223365784,
      "learning_rate": 0.0005455379871402788,
      "loss": 2.3486,
      "step": 44889
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7254583835601807,
      "learning_rate": 0.0005455356368294399,
      "loss": 2.1906,
      "step": 44890
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.6692661643028259,
      "learning_rate": 0.0005455332864729511,
      "loss": 2.1582,
      "step": 44891
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7343945503234863,
      "learning_rate": 0.0005455309360708131,
      "loss": 2.2574,
      "step": 44892
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7343534231185913,
      "learning_rate": 0.0005455285856230263,
      "loss": 2.4211,
      "step": 44893
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.6451090574264526,
      "learning_rate": 0.0005455262351295911,
      "loss": 2.115,
      "step": 44894
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.6842931509017944,
      "learning_rate": 0.000545523884590508,
      "loss": 2.3459,
      "step": 44895
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7131528854370117,
      "learning_rate": 0.0005455215340057773,
      "loss": 2.3373,
      "step": 44896
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.893921434879303,
      "learning_rate": 0.0005455191833753995,
      "loss": 2.2929,
      "step": 44897
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.6981210112571716,
      "learning_rate": 0.000545516832699375,
      "loss": 2.3446,
      "step": 44898
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7609028816223145,
      "learning_rate": 0.0005455144819777044,
      "loss": 2.2164,
      "step": 44899
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.6805180907249451,
      "learning_rate": 0.0005455121312103879,
      "loss": 2.2113,
      "step": 44900
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7441197037696838,
      "learning_rate": 0.0005455097803974262,
      "loss": 2.2186,
      "step": 44901
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.716524600982666,
      "learning_rate": 0.0005455074295388194,
      "loss": 2.234,
      "step": 44902
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7109008431434631,
      "learning_rate": 0.0005455050786345683,
      "loss": 2.2966,
      "step": 44903
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.6976449489593506,
      "learning_rate": 0.0005455027276846732,
      "loss": 2.284,
      "step": 44904
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7213167548179626,
      "learning_rate": 0.0005455003766891344,
      "loss": 2.2555,
      "step": 44905
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.6480620503425598,
      "learning_rate": 0.0005454980256479525,
      "loss": 2.1944,
      "step": 44906
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7425051331520081,
      "learning_rate": 0.0005454956745611279,
      "loss": 2.3167,
      "step": 44907
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.6479749083518982,
      "learning_rate": 0.000545493323428661,
      "loss": 2.0084,
      "step": 44908
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7096628546714783,
      "learning_rate": 0.0005454909722505522,
      "loss": 2.2272,
      "step": 44909
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7022477984428406,
      "learning_rate": 0.000545488621026802,
      "loss": 2.5173,
      "step": 44910
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.6763998866081238,
      "learning_rate": 0.0005454862697574109,
      "loss": 2.2294,
      "step": 44911
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7163780927658081,
      "learning_rate": 0.0005454839184423794,
      "loss": 2.4252,
      "step": 44912
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.734993040561676,
      "learning_rate": 0.0005454815670817076,
      "loss": 2.2316,
      "step": 44913
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.6844390630722046,
      "learning_rate": 0.0005454792156753961,
      "loss": 2.273,
      "step": 44914
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.6452129483222961,
      "learning_rate": 0.0005454768642234456,
      "loss": 2.1748,
      "step": 44915
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7829742431640625,
      "learning_rate": 0.0005454745127258561,
      "loss": 2.2329,
      "step": 44916
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.6895620822906494,
      "learning_rate": 0.0005454721611826283,
      "loss": 2.328,
      "step": 44917
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.663265585899353,
      "learning_rate": 0.0005454698095937626,
      "loss": 2.1887,
      "step": 44918
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7430252432823181,
      "learning_rate": 0.0005454674579592595,
      "loss": 2.3801,
      "step": 44919
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.6889898180961609,
      "learning_rate": 0.0005454651062791194,
      "loss": 2.2358,
      "step": 44920
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.6521801352500916,
      "learning_rate": 0.0005454627545533426,
      "loss": 2.2381,
      "step": 44921
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7849100828170776,
      "learning_rate": 0.0005454604027819297,
      "loss": 2.0902,
      "step": 44922
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.6972372531890869,
      "learning_rate": 0.000545458050964881,
      "loss": 2.2403,
      "step": 44923
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7274559140205383,
      "learning_rate": 0.000545455699102197,
      "loss": 2.2828,
      "step": 44924
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7072954177856445,
      "learning_rate": 0.0005454533471938782,
      "loss": 2.1689,
      "step": 44925
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.6758613586425781,
      "learning_rate": 0.000545450995239925,
      "loss": 2.1523,
      "step": 44926
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.734575092792511,
      "learning_rate": 0.000545448643240338,
      "loss": 2.4481,
      "step": 44927
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.8975232243537903,
      "learning_rate": 0.0005454462911951173,
      "loss": 2.2119,
      "step": 44928
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7921855449676514,
      "learning_rate": 0.0005454439391042634,
      "loss": 2.4094,
      "step": 44929
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7506876587867737,
      "learning_rate": 0.0005454415869677769,
      "loss": 2.2879,
      "step": 44930
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.0086569786071777,
      "learning_rate": 0.0005454392347856582,
      "loss": 2.3386,
      "step": 44931
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7341873049736023,
      "learning_rate": 0.0005454368825579078,
      "loss": 2.1211,
      "step": 44932
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7262780070304871,
      "learning_rate": 0.000545434530284526,
      "loss": 2.2111,
      "step": 44933
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.8428944945335388,
      "learning_rate": 0.0005454321779655132,
      "loss": 2.3232,
      "step": 44934
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7908545732498169,
      "learning_rate": 0.00054542982560087,
      "loss": 2.3194,
      "step": 44935
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.6815522313117981,
      "learning_rate": 0.0005454274731905968,
      "loss": 2.4949,
      "step": 44936
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7489272356033325,
      "learning_rate": 0.0005454251207346939,
      "loss": 2.5666,
      "step": 44937
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7790519595146179,
      "learning_rate": 0.0005454227682331618,
      "loss": 2.3257,
      "step": 44938
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.6894511580467224,
      "learning_rate": 0.0005454204156860012,
      "loss": 2.2409,
      "step": 44939
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7202558517456055,
      "learning_rate": 0.0005454180630932121,
      "loss": 2.0937,
      "step": 44940
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7292210459709167,
      "learning_rate": 0.0005454157104547953,
      "loss": 2.2533,
      "step": 44941
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.8370069861412048,
      "learning_rate": 0.000545413357770751,
      "loss": 2.2534,
      "step": 44942
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7316955924034119,
      "learning_rate": 0.0005454110050410797,
      "loss": 2.3892,
      "step": 44943
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.8011602163314819,
      "learning_rate": 0.0005454086522657819,
      "loss": 2.4575,
      "step": 44944
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7042779326438904,
      "learning_rate": 0.000545406299444858,
      "loss": 2.1188,
      "step": 44945
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7201080918312073,
      "learning_rate": 0.0005454039465783084,
      "loss": 2.0976,
      "step": 44946
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7030012607574463,
      "learning_rate": 0.0005454015936661337,
      "loss": 2.1347,
      "step": 44947
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7193364500999451,
      "learning_rate": 0.000545399240708334,
      "loss": 1.9835,
      "step": 44948
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.6609574556350708,
      "learning_rate": 0.0005453968877049101,
      "loss": 2.4371,
      "step": 44949
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7010514140129089,
      "learning_rate": 0.0005453945346558623,
      "loss": 2.3923,
      "step": 44950
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7813109755516052,
      "learning_rate": 0.0005453921815611909,
      "loss": 2.1915,
      "step": 44951
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.6888014674186707,
      "learning_rate": 0.0005453898284208966,
      "loss": 2.2977,
      "step": 44952
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7734188437461853,
      "learning_rate": 0.0005453874752349797,
      "loss": 2.3011,
      "step": 44953
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7926947474479675,
      "learning_rate": 0.0005453851220034406,
      "loss": 2.3012,
      "step": 44954
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7507987022399902,
      "learning_rate": 0.0005453827687262796,
      "loss": 2.351,
      "step": 44955
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7071405053138733,
      "learning_rate": 0.0005453804154034976,
      "loss": 2.3373,
      "step": 44956
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.6974765658378601,
      "learning_rate": 0.0005453780620350945,
      "loss": 2.3412,
      "step": 44957
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.6877410411834717,
      "learning_rate": 0.0005453757086210711,
      "loss": 2.2467,
      "step": 44958
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.69734787940979,
      "learning_rate": 0.0005453733551614278,
      "loss": 2.3128,
      "step": 44959
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.6650380492210388,
      "learning_rate": 0.0005453710016561649,
      "loss": 2.3034,
      "step": 44960
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.727826714515686,
      "learning_rate": 0.0005453686481052828,
      "loss": 2.1368,
      "step": 44961
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.6443892121315002,
      "learning_rate": 0.0005453662945087821,
      "loss": 2.1187,
      "step": 44962
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.6584264636039734,
      "learning_rate": 0.0005453639408666632,
      "loss": 2.1564,
      "step": 44963
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.5960028767585754,
      "learning_rate": 0.0005453615871789265,
      "loss": 2.0399,
      "step": 44964
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7308539748191833,
      "learning_rate": 0.0005453592334455725,
      "loss": 2.3342,
      "step": 44965
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7225608825683594,
      "learning_rate": 0.0005453568796666015,
      "loss": 2.1507,
      "step": 44966
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7454875111579895,
      "learning_rate": 0.0005453545258420141,
      "loss": 2.3966,
      "step": 44967
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7500330805778503,
      "learning_rate": 0.0005453521719718106,
      "loss": 2.4094,
      "step": 44968
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.6964388489723206,
      "learning_rate": 0.0005453498180559915,
      "loss": 2.4544,
      "step": 44969
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.655697762966156,
      "learning_rate": 0.0005453474640945572,
      "loss": 2.276,
      "step": 44970
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.6759264469146729,
      "learning_rate": 0.0005453451100875084,
      "loss": 2.2481,
      "step": 44971
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.6496176719665527,
      "learning_rate": 0.0005453427560348451,
      "loss": 2.2578,
      "step": 44972
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7303159832954407,
      "learning_rate": 0.000545340401936568,
      "loss": 2.3729,
      "step": 44973
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.6216704845428467,
      "learning_rate": 0.0005453380477926774,
      "loss": 2.2959,
      "step": 44974
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.6749699115753174,
      "learning_rate": 0.000545335693603174,
      "loss": 2.2757,
      "step": 44975
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.6449532508850098,
      "learning_rate": 0.000545333339368058,
      "loss": 2.1177,
      "step": 44976
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7140427231788635,
      "learning_rate": 0.0005453309850873299,
      "loss": 2.3837,
      "step": 44977
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.699363648891449,
      "learning_rate": 0.00054532863076099,
      "loss": 2.2886,
      "step": 44978
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7224795818328857,
      "learning_rate": 0.000545326276389039,
      "loss": 2.0751,
      "step": 44979
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7353420853614807,
      "learning_rate": 0.0005453239219714773,
      "loss": 2.3345,
      "step": 44980
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.8126862645149231,
      "learning_rate": 0.0005453215675083051,
      "loss": 2.4474,
      "step": 44981
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.6923041939735413,
      "learning_rate": 0.000545319212999523,
      "loss": 2.2535,
      "step": 44982
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7200814485549927,
      "learning_rate": 0.0005453168584451315,
      "loss": 2.283,
      "step": 44983
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7766902446746826,
      "learning_rate": 0.0005453145038451311,
      "loss": 2.4686,
      "step": 44984
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.6645956039428711,
      "learning_rate": 0.0005453121491995218,
      "loss": 2.3298,
      "step": 44985
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.6552565693855286,
      "learning_rate": 0.0005453097945083045,
      "loss": 2.0114,
      "step": 44986
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7451673150062561,
      "learning_rate": 0.0005453074397714795,
      "loss": 2.2911,
      "step": 44987
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7030603289604187,
      "learning_rate": 0.0005453050849890471,
      "loss": 2.5707,
      "step": 44988
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.6359556317329407,
      "learning_rate": 0.0005453027301610079,
      "loss": 2.1081,
      "step": 44989
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.9903889298439026,
      "learning_rate": 0.0005453003752873625,
      "loss": 2.2433,
      "step": 44990
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.6775433421134949,
      "learning_rate": 0.0005452980203681109,
      "loss": 2.247,
      "step": 44991
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7606726288795471,
      "learning_rate": 0.0005452956654032538,
      "loss": 2.2439,
      "step": 44992
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.6871526837348938,
      "learning_rate": 0.0005452933103927916,
      "loss": 2.3283,
      "step": 44993
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.6630075573921204,
      "learning_rate": 0.0005452909553367248,
      "loss": 2.1698,
      "step": 44994
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.6126575469970703,
      "learning_rate": 0.0005452886002350537,
      "loss": 2.2691,
      "step": 44995
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.6510599851608276,
      "learning_rate": 0.0005452862450877789,
      "loss": 1.9282,
      "step": 44996
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7691417932510376,
      "learning_rate": 0.0005452838898949007,
      "loss": 2.1623,
      "step": 44997
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.6918908953666687,
      "learning_rate": 0.0005452815346564197,
      "loss": 2.1137,
      "step": 44998
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7805203795433044,
      "learning_rate": 0.0005452791793723362,
      "loss": 2.1391,
      "step": 44999
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7746860384941101,
      "learning_rate": 0.0005452768240426507,
      "loss": 2.3192,
      "step": 45000
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.6561573147773743,
      "learning_rate": 0.0005452744686673634,
      "loss": 2.2357,
      "step": 45001
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7731606960296631,
      "learning_rate": 0.0005452721132464751,
      "loss": 2.3989,
      "step": 45002
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.661277174949646,
      "learning_rate": 0.0005452697577799861,
      "loss": 2.0487,
      "step": 45003
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.8052117228507996,
      "learning_rate": 0.0005452674022678967,
      "loss": 2.3365,
      "step": 45004
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7539721727371216,
      "learning_rate": 0.0005452650467102076,
      "loss": 2.4437,
      "step": 45005
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7672541737556458,
      "learning_rate": 0.0005452626911069191,
      "loss": 2.3906,
      "step": 45006
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.6917367577552795,
      "learning_rate": 0.0005452603354580316,
      "loss": 2.2719,
      "step": 45007
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.0702155828475952,
      "learning_rate": 0.0005452579797635456,
      "loss": 2.323,
      "step": 45008
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.705554723739624,
      "learning_rate": 0.0005452556240234614,
      "loss": 2.348,
      "step": 45009
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.6982343196868896,
      "learning_rate": 0.0005452532682377796,
      "loss": 2.2919,
      "step": 45010
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.8148386478424072,
      "learning_rate": 0.0005452509124065007,
      "loss": 2.1435,
      "step": 45011
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7265744209289551,
      "learning_rate": 0.0005452485565296249,
      "loss": 2.3731,
      "step": 45012
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7304444909095764,
      "learning_rate": 0.0005452462006071527,
      "loss": 2.1758,
      "step": 45013
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.648158073425293,
      "learning_rate": 0.0005452438446390849,
      "loss": 2.0377,
      "step": 45014
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7026556134223938,
      "learning_rate": 0.0005452414886254213,
      "loss": 2.2211,
      "step": 45015
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7787572741508484,
      "learning_rate": 0.0005452391325661628,
      "loss": 2.4141,
      "step": 45016
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7292689085006714,
      "learning_rate": 0.0005452367764613098,
      "loss": 2.2473,
      "step": 45017
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.8897809982299805,
      "learning_rate": 0.0005452344203108626,
      "loss": 2.7072,
      "step": 45018
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7276944518089294,
      "learning_rate": 0.0005452320641148216,
      "loss": 2.4257,
      "step": 45019
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.6721927523612976,
      "learning_rate": 0.0005452297078731874,
      "loss": 2.3483,
      "step": 45020
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.6846593618392944,
      "learning_rate": 0.0005452273515859604,
      "loss": 2.4034,
      "step": 45021
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7252881526947021,
      "learning_rate": 0.0005452249952531409,
      "loss": 2.1844,
      "step": 45022
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7753407955169678,
      "learning_rate": 0.0005452226388747296,
      "loss": 2.0528,
      "step": 45023
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.6293372511863708,
      "learning_rate": 0.0005452202824507268,
      "loss": 2.3557,
      "step": 45024
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7161511182785034,
      "learning_rate": 0.0005452179259811328,
      "loss": 2.4139,
      "step": 45025
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7199897170066833,
      "learning_rate": 0.0005452155694659482,
      "loss": 2.2148,
      "step": 45026
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.739449143409729,
      "learning_rate": 0.0005452132129051733,
      "loss": 2.3947,
      "step": 45027
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7254217267036438,
      "learning_rate": 0.0005452108562988088,
      "loss": 2.4722,
      "step": 45028
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.8325021266937256,
      "learning_rate": 0.0005452084996468549,
      "loss": 2.2538,
      "step": 45029
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7314591407775879,
      "learning_rate": 0.0005452061429493121,
      "loss": 2.0093,
      "step": 45030
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7294636964797974,
      "learning_rate": 0.0005452037862061809,
      "loss": 2.2591,
      "step": 45031
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.6549742221832275,
      "learning_rate": 0.0005452014294174616,
      "loss": 2.3082,
      "step": 45032
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.6530523896217346,
      "learning_rate": 0.0005451990725831548,
      "loss": 2.2078,
      "step": 45033
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.695223867893219,
      "learning_rate": 0.0005451967157032608,
      "loss": 2.2449,
      "step": 45034
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7506452202796936,
      "learning_rate": 0.0005451943587777801,
      "loss": 2.3169,
      "step": 45035
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.6714919805526733,
      "learning_rate": 0.0005451920018067132,
      "loss": 2.055,
      "step": 45036
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.8936988711357117,
      "learning_rate": 0.0005451896447900605,
      "loss": 2.2799,
      "step": 45037
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.757020890712738,
      "learning_rate": 0.0005451872877278224,
      "loss": 2.1291,
      "step": 45038
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.6968920230865479,
      "learning_rate": 0.0005451849306199992,
      "loss": 2.4824,
      "step": 45039
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.9582210779190063,
      "learning_rate": 0.0005451825734665917,
      "loss": 2.2703,
      "step": 45040
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.6990836262702942,
      "learning_rate": 0.0005451802162676001,
      "loss": 2.3687,
      "step": 45041
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.647962212562561,
      "learning_rate": 0.0005451778590230248,
      "loss": 2.3084,
      "step": 45042
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7208752036094666,
      "learning_rate": 0.0005451755017328663,
      "loss": 2.3743,
      "step": 45043
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.724114179611206,
      "learning_rate": 0.0005451731443971251,
      "loss": 2.239,
      "step": 45044
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7221434116363525,
      "learning_rate": 0.0005451707870158015,
      "loss": 2.0003,
      "step": 45045
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.798086404800415,
      "learning_rate": 0.0005451684295888961,
      "loss": 2.2038,
      "step": 45046
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7269535660743713,
      "learning_rate": 0.0005451660721164092,
      "loss": 2.2158,
      "step": 45047
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.651240348815918,
      "learning_rate": 0.0005451637145983414,
      "loss": 2.5135,
      "step": 45048
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7614392638206482,
      "learning_rate": 0.000545161357034693,
      "loss": 2.4701,
      "step": 45049
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.8257277011871338,
      "learning_rate": 0.0005451589994254645,
      "loss": 2.2334,
      "step": 45050
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7066203951835632,
      "learning_rate": 0.0005451566417706562,
      "loss": 2.3681,
      "step": 45051
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.6445902585983276,
      "learning_rate": 0.0005451542840702688,
      "loss": 2.4096,
      "step": 45052
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7488601803779602,
      "learning_rate": 0.0005451519263243024,
      "loss": 2.244,
      "step": 45053
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7832868099212646,
      "learning_rate": 0.0005451495685327579,
      "loss": 2.2124,
      "step": 45054
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7491664886474609,
      "learning_rate": 0.0005451472106956353,
      "loss": 2.4626,
      "step": 45055
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.6533704400062561,
      "learning_rate": 0.0005451448528129352,
      "loss": 2.417,
      "step": 45056
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7342856526374817,
      "learning_rate": 0.0005451424948846581,
      "loss": 2.3456,
      "step": 45057
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7510638236999512,
      "learning_rate": 0.0005451401369108043,
      "loss": 2.2073,
      "step": 45058
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7263953685760498,
      "learning_rate": 0.0005451377788913743,
      "loss": 2.3329,
      "step": 45059
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.6546970009803772,
      "learning_rate": 0.0005451354208263686,
      "loss": 2.263,
      "step": 45060
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.6580758094787598,
      "learning_rate": 0.0005451330627157877,
      "loss": 2.2284,
      "step": 45061
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7033275365829468,
      "learning_rate": 0.0005451307045596319,
      "loss": 2.1821,
      "step": 45062
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7186300754547119,
      "learning_rate": 0.0005451283463579016,
      "loss": 2.3859,
      "step": 45063
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7551771402359009,
      "learning_rate": 0.0005451259881105974,
      "loss": 1.9293,
      "step": 45064
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.6650205850601196,
      "learning_rate": 0.0005451236298177196,
      "loss": 2.3916,
      "step": 45065
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7473512291908264,
      "learning_rate": 0.0005451212714792686,
      "loss": 2.2815,
      "step": 45066
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.6664790511131287,
      "learning_rate": 0.000545118913095245,
      "loss": 2.4401,
      "step": 45067
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7212168574333191,
      "learning_rate": 0.0005451165546656492,
      "loss": 2.1492,
      "step": 45068
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.6842755675315857,
      "learning_rate": 0.0005451141961904816,
      "loss": 2.0482,
      "step": 45069
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7277598977088928,
      "learning_rate": 0.0005451118376697426,
      "loss": 2.5067,
      "step": 45070
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.6521105170249939,
      "learning_rate": 0.0005451094791034327,
      "loss": 2.4215,
      "step": 45071
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.695604681968689,
      "learning_rate": 0.0005451071204915524,
      "loss": 2.3471,
      "step": 45072
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.6959479451179504,
      "learning_rate": 0.000545104761834102,
      "loss": 2.1341,
      "step": 45073
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7661812901496887,
      "learning_rate": 0.000545102403131082,
      "loss": 2.0277,
      "step": 45074
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7197527885437012,
      "learning_rate": 0.0005451000443824929,
      "loss": 2.0102,
      "step": 45075
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.6394932270050049,
      "learning_rate": 0.0005450976855883349,
      "loss": 2.2238,
      "step": 45076
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7650903463363647,
      "learning_rate": 0.0005450953267486089,
      "loss": 2.3276,
      "step": 45077
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.8092329502105713,
      "learning_rate": 0.0005450929678633147,
      "loss": 2.6941,
      "step": 45078
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7223415374755859,
      "learning_rate": 0.0005450906089324533,
      "loss": 2.4327,
      "step": 45079
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.6997250318527222,
      "learning_rate": 0.0005450882499560248,
      "loss": 2.2412,
      "step": 45080
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.8016356825828552,
      "learning_rate": 0.00054508589093403,
      "loss": 2.2987,
      "step": 45081
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.6431779265403748,
      "learning_rate": 0.0005450835318664689,
      "loss": 2.2618,
      "step": 45082
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.6702612042427063,
      "learning_rate": 0.0005450811727533422,
      "loss": 2.3191,
      "step": 45083
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.666696310043335,
      "learning_rate": 0.0005450788135946503,
      "loss": 2.1335,
      "step": 45084
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7253313064575195,
      "learning_rate": 0.0005450764543903937,
      "loss": 2.3058,
      "step": 45085
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.6609973311424255,
      "learning_rate": 0.0005450740951405726,
      "loss": 2.1005,
      "step": 45086
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7077212333679199,
      "learning_rate": 0.0005450717358451877,
      "loss": 2.3752,
      "step": 45087
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.8166616559028625,
      "learning_rate": 0.0005450693765042393,
      "loss": 2.4001,
      "step": 45088
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.6945900917053223,
      "learning_rate": 0.0005450670171177279,
      "loss": 2.1144,
      "step": 45089
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7157211899757385,
      "learning_rate": 0.0005450646576856539,
      "loss": 2.1738,
      "step": 45090
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7671686410903931,
      "learning_rate": 0.0005450622982080179,
      "loss": 2.3977,
      "step": 45091
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7350260615348816,
      "learning_rate": 0.0005450599386848201,
      "loss": 2.6836,
      "step": 45092
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.6229022145271301,
      "learning_rate": 0.0005450575791160608,
      "loss": 2.0877,
      "step": 45093
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7226430773735046,
      "learning_rate": 0.0005450552195017409,
      "loss": 2.3777,
      "step": 45094
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7219686508178711,
      "learning_rate": 0.0005450528598418606,
      "loss": 2.1033,
      "step": 45095
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7737059593200684,
      "learning_rate": 0.0005450505001364204,
      "loss": 2.4223,
      "step": 45096
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.720984160900116,
      "learning_rate": 0.0005450481403854205,
      "loss": 2.1561,
      "step": 45097
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7201855182647705,
      "learning_rate": 0.0005450457805888616,
      "loss": 2.3242,
      "step": 45098
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.8163513541221619,
      "learning_rate": 0.0005450434207467441,
      "loss": 2.2786,
      "step": 45099
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.6377155184745789,
      "learning_rate": 0.0005450410608590685,
      "loss": 2.227,
      "step": 45100
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.6108102202415466,
      "learning_rate": 0.0005450387009258349,
      "loss": 2.365,
      "step": 45101
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.9301210045814514,
      "learning_rate": 0.0005450363409470442,
      "loss": 2.5266,
      "step": 45102
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.722501814365387,
      "learning_rate": 0.0005450339809226965,
      "loss": 2.4956,
      "step": 45103
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7903653979301453,
      "learning_rate": 0.0005450316208527924,
      "loss": 2.3932,
      "step": 45104
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7177960276603699,
      "learning_rate": 0.0005450292607373324,
      "loss": 2.2058,
      "step": 45105
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7337515354156494,
      "learning_rate": 0.0005450269005763166,
      "loss": 2.2814,
      "step": 45106
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7763016819953918,
      "learning_rate": 0.0005450245403697459,
      "loss": 2.3506,
      "step": 45107
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7740914225578308,
      "learning_rate": 0.0005450221801176203,
      "loss": 2.3382,
      "step": 45108
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7278286218643188,
      "learning_rate": 0.0005450198198199407,
      "loss": 2.1138,
      "step": 45109
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7556023597717285,
      "learning_rate": 0.0005450174594767071,
      "loss": 2.4731,
      "step": 45110
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.6913398504257202,
      "learning_rate": 0.0005450150990879201,
      "loss": 2.1955,
      "step": 45111
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7116596698760986,
      "learning_rate": 0.0005450127386535802,
      "loss": 2.3904,
      "step": 45112
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7085727453231812,
      "learning_rate": 0.000545010378173688,
      "loss": 1.9718,
      "step": 45113
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.6898225545883179,
      "learning_rate": 0.0005450080176482436,
      "loss": 2.1944,
      "step": 45114
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.749991238117218,
      "learning_rate": 0.0005450056570772475,
      "loss": 2.061,
      "step": 45115
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7526400089263916,
      "learning_rate": 0.0005450032964607002,
      "loss": 2.3038,
      "step": 45116
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7245990633964539,
      "learning_rate": 0.0005450009357986024,
      "loss": 2.0611,
      "step": 45117
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7352301478385925,
      "learning_rate": 0.0005449985750909542,
      "loss": 2.1795,
      "step": 45118
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.4202218055725098,
      "learning_rate": 0.000544996214337756,
      "loss": 2.0884,
      "step": 45119
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.6391956806182861,
      "learning_rate": 0.0005449938535390086,
      "loss": 2.0361,
      "step": 45120
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7353700995445251,
      "learning_rate": 0.0005449914926947121,
      "loss": 2.447,
      "step": 45121
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.6695125102996826,
      "learning_rate": 0.0005449891318048671,
      "loss": 2.0537,
      "step": 45122
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7887928485870361,
      "learning_rate": 0.000544986770869474,
      "loss": 2.3576,
      "step": 45123
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.6421737670898438,
      "learning_rate": 0.0005449844098885331,
      "loss": 2.3501,
      "step": 45124
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.61724853515625,
      "learning_rate": 0.0005449820488620451,
      "loss": 2.3023,
      "step": 45125
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.6173955798149109,
      "learning_rate": 0.0005449796877900103,
      "loss": 2.1797,
      "step": 45126
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7015454173088074,
      "learning_rate": 0.0005449773266724291,
      "loss": 2.3242,
      "step": 45127
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7558948397636414,
      "learning_rate": 0.0005449749655093021,
      "loss": 2.3837,
      "step": 45128
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7752649188041687,
      "learning_rate": 0.0005449726043006295,
      "loss": 2.3377,
      "step": 45129
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7063431739807129,
      "learning_rate": 0.000544970243046412,
      "loss": 2.3016,
      "step": 45130
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7558278441429138,
      "learning_rate": 0.0005449678817466498,
      "loss": 2.1431,
      "step": 45131
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.718022346496582,
      "learning_rate": 0.0005449655204013434,
      "loss": 2.5886,
      "step": 45132
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7470662593841553,
      "learning_rate": 0.0005449631590104934,
      "loss": 2.183,
      "step": 45133
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.8113352656364441,
      "learning_rate": 0.0005449607975741001,
      "loss": 2.4025,
      "step": 45134
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.8226607441902161,
      "learning_rate": 0.000544958436092164,
      "loss": 2.3513,
      "step": 45135
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.803857147693634,
      "learning_rate": 0.0005449560745646855,
      "loss": 2.4392,
      "step": 45136
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.6876635551452637,
      "learning_rate": 0.0005449537129916649,
      "loss": 2.4396,
      "step": 45137
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.753734290599823,
      "learning_rate": 0.0005449513513731029,
      "loss": 2.2285,
      "step": 45138
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.6838038563728333,
      "learning_rate": 0.0005449489897089998,
      "loss": 2.2853,
      "step": 45139
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.6822800040245056,
      "learning_rate": 0.000544946627999356,
      "loss": 2.1632,
      "step": 45140
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.8758164644241333,
      "learning_rate": 0.0005449442662441721,
      "loss": 2.4861,
      "step": 45141
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.6387295126914978,
      "learning_rate": 0.0005449419044434485,
      "loss": 1.7991,
      "step": 45142
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.6988415718078613,
      "learning_rate": 0.0005449395425971854,
      "loss": 2.3968,
      "step": 45143
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.9520338177680969,
      "learning_rate": 0.0005449371807053835,
      "loss": 2.2736,
      "step": 45144
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.6800631284713745,
      "learning_rate": 0.0005449348187680431,
      "loss": 2.5543,
      "step": 45145
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7087194919586182,
      "learning_rate": 0.0005449324567851647,
      "loss": 2.1677,
      "step": 45146
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.8861400485038757,
      "learning_rate": 0.0005449300947567488,
      "loss": 2.2845,
      "step": 45147
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.6883180737495422,
      "learning_rate": 0.0005449277326827958,
      "loss": 2.4795,
      "step": 45148
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.8370886445045471,
      "learning_rate": 0.000544925370563306,
      "loss": 2.3613,
      "step": 45149
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7627177238464355,
      "learning_rate": 0.00054492300839828,
      "loss": 2.2158,
      "step": 45150
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.6984080076217651,
      "learning_rate": 0.0005449206461877182,
      "loss": 2.3751,
      "step": 45151
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.6686806082725525,
      "learning_rate": 0.000544918283931621,
      "loss": 2.2226,
      "step": 45152
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7660834193229675,
      "learning_rate": 0.0005449159216299889,
      "loss": 2.3887,
      "step": 45153
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.6799121499061584,
      "learning_rate": 0.0005449135592828223,
      "loss": 2.3203,
      "step": 45154
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7107633352279663,
      "learning_rate": 0.0005449111968901216,
      "loss": 2.3158,
      "step": 45155
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.707382321357727,
      "learning_rate": 0.0005449088344518875,
      "loss": 2.4094,
      "step": 45156
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7342867851257324,
      "learning_rate": 0.0005449064719681199,
      "loss": 2.5214,
      "step": 45157
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7855249643325806,
      "learning_rate": 0.0005449041094388197,
      "loss": 2.4475,
      "step": 45158
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.6570300459861755,
      "learning_rate": 0.0005449017468639872,
      "loss": 2.1619,
      "step": 45159
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.641385555267334,
      "learning_rate": 0.0005448993842436229,
      "loss": 2.3203,
      "step": 45160
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.8261722326278687,
      "learning_rate": 0.0005448970215777272,
      "loss": 2.3137,
      "step": 45161
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.6776121854782104,
      "learning_rate": 0.0005448946588663005,
      "loss": 2.2014,
      "step": 45162
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.8585237264633179,
      "learning_rate": 0.0005448922961093432,
      "loss": 2.1743,
      "step": 45163
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.8078722357749939,
      "learning_rate": 0.0005448899333068557,
      "loss": 2.243,
      "step": 45164
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.8970404863357544,
      "learning_rate": 0.0005448875704588387,
      "loss": 2.3543,
      "step": 45165
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7168957591056824,
      "learning_rate": 0.0005448852075652924,
      "loss": 2.1175,
      "step": 45166
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7550939917564392,
      "learning_rate": 0.0005448828446262174,
      "loss": 2.3648,
      "step": 45167
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.871604859828949,
      "learning_rate": 0.000544880481641614,
      "loss": 2.2175,
      "step": 45168
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.677924633026123,
      "learning_rate": 0.0005448781186114827,
      "loss": 2.2876,
      "step": 45169
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.8693972826004028,
      "learning_rate": 0.000544875755535824,
      "loss": 2.1133,
      "step": 45170
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.9206741452217102,
      "learning_rate": 0.0005448733924146382,
      "loss": 2.3271,
      "step": 45171
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.691407322883606,
      "learning_rate": 0.0005448710292479257,
      "loss": 2.2859,
      "step": 45172
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7897167801856995,
      "learning_rate": 0.0005448686660356873,
      "loss": 2.2604,
      "step": 45173
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.9034469127655029,
      "learning_rate": 0.000544866302777923,
      "loss": 2.3119,
      "step": 45174
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.6645832657814026,
      "learning_rate": 0.0005448639394746336,
      "loss": 2.3022,
      "step": 45175
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7292057275772095,
      "learning_rate": 0.0005448615761258193,
      "loss": 2.5085,
      "step": 45176
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.8420695662498474,
      "learning_rate": 0.0005448592127314806,
      "loss": 2.6068,
      "step": 45177
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.8263386487960815,
      "learning_rate": 0.0005448568492916179,
      "loss": 2.4024,
      "step": 45178
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.6755277514457703,
      "learning_rate": 0.0005448544858062317,
      "loss": 2.2134,
      "step": 45179
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7076242566108704,
      "learning_rate": 0.0005448521222753225,
      "loss": 2.0567,
      "step": 45180
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7987642884254456,
      "learning_rate": 0.0005448497586988906,
      "loss": 2.3786,
      "step": 45181
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7331791520118713,
      "learning_rate": 0.0005448473950769365,
      "loss": 2.1982,
      "step": 45182
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.747179388999939,
      "learning_rate": 0.0005448450314094607,
      "loss": 2.3888,
      "step": 45183
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7829628586769104,
      "learning_rate": 0.0005448426676964637,
      "loss": 2.5198,
      "step": 45184
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.8299219608306885,
      "learning_rate": 0.0005448403039379457,
      "loss": 2.1103,
      "step": 45185
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7162683010101318,
      "learning_rate": 0.0005448379401339072,
      "loss": 2.4938,
      "step": 45186
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.6921520829200745,
      "learning_rate": 0.0005448355762843488,
      "loss": 2.3849,
      "step": 45187
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.713782787322998,
      "learning_rate": 0.0005448332123892709,
      "loss": 2.2127,
      "step": 45188
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.6942436695098877,
      "learning_rate": 0.0005448308484486737,
      "loss": 2.2316,
      "step": 45189
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7843104600906372,
      "learning_rate": 0.0005448284844625579,
      "loss": 2.0331,
      "step": 45190
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7210665345191956,
      "learning_rate": 0.0005448261204309239,
      "loss": 2.1907,
      "step": 45191
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7294902801513672,
      "learning_rate": 0.0005448237563537721,
      "loss": 2.3616,
      "step": 45192
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7566381692886353,
      "learning_rate": 0.0005448213922311029,
      "loss": 2.369,
      "step": 45193
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.6568309664726257,
      "learning_rate": 0.0005448190280629168,
      "loss": 2.1445,
      "step": 45194
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7522078156471252,
      "learning_rate": 0.0005448166638492143,
      "loss": 2.3586,
      "step": 45195
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7669596076011658,
      "learning_rate": 0.0005448142995899956,
      "loss": 2.1506,
      "step": 45196
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.662381112575531,
      "learning_rate": 0.0005448119352852614,
      "loss": 2.1003,
      "step": 45197
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.6914109587669373,
      "learning_rate": 0.000544809570935012,
      "loss": 2.3372,
      "step": 45198
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.6769742965698242,
      "learning_rate": 0.000544807206539248,
      "loss": 2.3464,
      "step": 45199
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.6721193790435791,
      "learning_rate": 0.0005448048420979697,
      "loss": 2.3784,
      "step": 45200
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7266843318939209,
      "learning_rate": 0.0005448024776111774,
      "loss": 2.2793,
      "step": 45201
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.6691672801971436,
      "learning_rate": 0.0005448001130788718,
      "loss": 2.2466,
      "step": 45202
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.6567296385765076,
      "learning_rate": 0.0005447977485010533,
      "loss": 2.0416,
      "step": 45203
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.6962956786155701,
      "learning_rate": 0.0005447953838777222,
      "loss": 2.4361,
      "step": 45204
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.6820784211158752,
      "learning_rate": 0.000544793019208879,
      "loss": 2.2367,
      "step": 45205
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.6873133182525635,
      "learning_rate": 0.0005447906544945242,
      "loss": 2.2762,
      "step": 45206
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7023615837097168,
      "learning_rate": 0.0005447882897346581,
      "loss": 2.1939,
      "step": 45207
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7216231226921082,
      "learning_rate": 0.0005447859249292814,
      "loss": 2.2192,
      "step": 45208
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.715814471244812,
      "learning_rate": 0.0005447835600783942,
      "loss": 2.3421,
      "step": 45209
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.6512824892997742,
      "learning_rate": 0.0005447811951819972,
      "loss": 2.2085,
      "step": 45210
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7083009481430054,
      "learning_rate": 0.0005447788302400909,
      "loss": 2.4428,
      "step": 45211
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7797820568084717,
      "learning_rate": 0.0005447764652526753,
      "loss": 2.3756,
      "step": 45212
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7059767842292786,
      "learning_rate": 0.0005447741002197513,
      "loss": 2.0881,
      "step": 45213
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7084043025970459,
      "learning_rate": 0.0005447717351413192,
      "loss": 2.4007,
      "step": 45214
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7078381776809692,
      "learning_rate": 0.0005447693700173793,
      "loss": 2.2922,
      "step": 45215
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.6658864617347717,
      "learning_rate": 0.0005447670048479323,
      "loss": 2.2608,
      "step": 45216
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.6545359492301941,
      "learning_rate": 0.0005447646396329784,
      "loss": 2.0306,
      "step": 45217
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.6501155495643616,
      "learning_rate": 0.0005447622743725183,
      "loss": 2.3986,
      "step": 45218
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.6388604640960693,
      "learning_rate": 0.0005447599090665521,
      "loss": 2.2867,
      "step": 45219
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.6650403738021851,
      "learning_rate": 0.0005447575437150803,
      "loss": 2.3081,
      "step": 45220
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7232499718666077,
      "learning_rate": 0.0005447551783181037,
      "loss": 2.3586,
      "step": 45221
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7347670197486877,
      "learning_rate": 0.0005447528128756224,
      "loss": 2.2173,
      "step": 45222
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7115324139595032,
      "learning_rate": 0.0005447504473876369,
      "loss": 2.0198,
      "step": 45223
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.6967105865478516,
      "learning_rate": 0.0005447480818541477,
      "loss": 2.6661,
      "step": 45224
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7272136807441711,
      "learning_rate": 0.0005447457162751552,
      "loss": 2.2272,
      "step": 45225
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.714262068271637,
      "learning_rate": 0.0005447433506506599,
      "loss": 2.0823,
      "step": 45226
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7183489203453064,
      "learning_rate": 0.0005447409849806622,
      "loss": 2.2603,
      "step": 45227
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.662962794303894,
      "learning_rate": 0.0005447386192651624,
      "loss": 2.183,
      "step": 45228
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7048143744468689,
      "learning_rate": 0.0005447362535041611,
      "loss": 2.3978,
      "step": 45229
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.6621763110160828,
      "learning_rate": 0.0005447338876976588,
      "loss": 2.2821,
      "step": 45230
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.6473255753517151,
      "learning_rate": 0.0005447315218456558,
      "loss": 2.3614,
      "step": 45231
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.6538583636283875,
      "learning_rate": 0.0005447291559481526,
      "loss": 2.3296,
      "step": 45232
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.8227397203445435,
      "learning_rate": 0.0005447267900051497,
      "loss": 2.0981,
      "step": 45233
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7288206815719604,
      "learning_rate": 0.0005447244240166473,
      "loss": 2.2037,
      "step": 45234
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7972478270530701,
      "learning_rate": 0.0005447220579826461,
      "loss": 2.2663,
      "step": 45235
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7144483923912048,
      "learning_rate": 0.0005447196919031466,
      "loss": 2.4056,
      "step": 45236
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.6713389158248901,
      "learning_rate": 0.0005447173257781489,
      "loss": 2.1172,
      "step": 45237
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.730118989944458,
      "learning_rate": 0.0005447149596076536,
      "loss": 2.3527,
      "step": 45238
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7370797991752625,
      "learning_rate": 0.0005447125933916613,
      "loss": 2.1986,
      "step": 45239
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.703200101852417,
      "learning_rate": 0.0005447102271301722,
      "loss": 2.2691,
      "step": 45240
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.6510740518569946,
      "learning_rate": 0.000544707860823187,
      "loss": 2.2768,
      "step": 45241
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.8036366701126099,
      "learning_rate": 0.0005447054944707058,
      "loss": 2.37,
      "step": 45242
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.6927487254142761,
      "learning_rate": 0.0005447031280727294,
      "loss": 2.1262,
      "step": 45243
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7000091671943665,
      "learning_rate": 0.000544700761629258,
      "loss": 2.2451,
      "step": 45244
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.6815060377120972,
      "learning_rate": 0.0005446983951402922,
      "loss": 2.3755,
      "step": 45245
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7330851554870605,
      "learning_rate": 0.0005446960286058321,
      "loss": 2.3116,
      "step": 45246
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.9153919219970703,
      "learning_rate": 0.0005446936620258786,
      "loss": 2.1407,
      "step": 45247
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7087584733963013,
      "learning_rate": 0.0005446912954004318,
      "loss": 2.2636,
      "step": 45248
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7743087410926819,
      "learning_rate": 0.0005446889287294924,
      "loss": 1.9141,
      "step": 45249
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7908170819282532,
      "learning_rate": 0.0005446865620130607,
      "loss": 2.2068,
      "step": 45250
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.6836799383163452,
      "learning_rate": 0.0005446841952511371,
      "loss": 2.2062,
      "step": 45251
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7379628419876099,
      "learning_rate": 0.000544681828443722,
      "loss": 2.4435,
      "step": 45252
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7215782403945923,
      "learning_rate": 0.0005446794615908161,
      "loss": 2.3623,
      "step": 45253
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7138835787773132,
      "learning_rate": 0.0005446770946924195,
      "loss": 2.4428,
      "step": 45254
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7515108585357666,
      "learning_rate": 0.000544674727748533,
      "loss": 2.5232,
      "step": 45255
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.6956409215927124,
      "learning_rate": 0.0005446723607591568,
      "loss": 2.3169,
      "step": 45256
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7075790762901306,
      "learning_rate": 0.0005446699937242912,
      "loss": 2.3704,
      "step": 45257
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.6950167417526245,
      "learning_rate": 0.000544667626643937,
      "loss": 2.1056,
      "step": 45258
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7924774885177612,
      "learning_rate": 0.0005446652595180945,
      "loss": 2.3789,
      "step": 45259
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.6718085408210754,
      "learning_rate": 0.000544662892346764,
      "loss": 2.1869,
      "step": 45260
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.8202382922172546,
      "learning_rate": 0.0005446605251299461,
      "loss": 2.3721,
      "step": 45261
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.6585615873336792,
      "learning_rate": 0.0005446581578676412,
      "loss": 2.1492,
      "step": 45262
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.6717619895935059,
      "learning_rate": 0.0005446557905598497,
      "loss": 2.0545,
      "step": 45263
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.6832002997398376,
      "learning_rate": 0.0005446534232065721,
      "loss": 2.2646,
      "step": 45264
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7222286462783813,
      "learning_rate": 0.0005446510558078088,
      "loss": 2.3424,
      "step": 45265
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.6789904832839966,
      "learning_rate": 0.0005446486883635603,
      "loss": 2.246,
      "step": 45266
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.6796305179595947,
      "learning_rate": 0.0005446463208738269,
      "loss": 2.3153,
      "step": 45267
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.6555665731430054,
      "learning_rate": 0.0005446439533386092,
      "loss": 2.2487,
      "step": 45268
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7605152726173401,
      "learning_rate": 0.0005446415857579076,
      "loss": 2.3983,
      "step": 45269
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7289299368858337,
      "learning_rate": 0.0005446392181317224,
      "loss": 2.2344,
      "step": 45270
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7360315322875977,
      "learning_rate": 0.0005446368504600542,
      "loss": 2.2046,
      "step": 45271
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.6701710820198059,
      "learning_rate": 0.0005446344827429034,
      "loss": 2.5305,
      "step": 45272
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.6989472508430481,
      "learning_rate": 0.0005446321149802704,
      "loss": 2.2748,
      "step": 45273
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.6722258925437927,
      "learning_rate": 0.0005446297471721558,
      "loss": 2.0107,
      "step": 45274
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.6649636030197144,
      "learning_rate": 0.0005446273793185597,
      "loss": 2.4883,
      "step": 45275
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.163399577140808,
      "learning_rate": 0.000544625011419483,
      "loss": 2.0409,
      "step": 45276
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.6961351037025452,
      "learning_rate": 0.0005446226434749257,
      "loss": 2.2684,
      "step": 45277
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.9171726703643799,
      "learning_rate": 0.0005446202754848884,
      "loss": 2.2579,
      "step": 45278
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.8423285484313965,
      "learning_rate": 0.0005446179074493718,
      "loss": 2.2977,
      "step": 45279
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7095463275909424,
      "learning_rate": 0.0005446155393683758,
      "loss": 2.1609,
      "step": 45280
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.727523148059845,
      "learning_rate": 0.0005446131712419015,
      "loss": 2.1427,
      "step": 45281
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7625054717063904,
      "learning_rate": 0.0005446108030699488,
      "loss": 2.298,
      "step": 45282
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7750279903411865,
      "learning_rate": 0.0005446084348525183,
      "loss": 2.3977,
      "step": 45283
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7138870358467102,
      "learning_rate": 0.0005446060665896106,
      "loss": 2.5929,
      "step": 45284
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7665200233459473,
      "learning_rate": 0.0005446036982812259,
      "loss": 2.2798,
      "step": 45285
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.712292492389679,
      "learning_rate": 0.0005446013299273649,
      "loss": 2.3368,
      "step": 45286
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.8655468225479126,
      "learning_rate": 0.0005445989615280278,
      "loss": 2.3298,
      "step": 45287
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.6351306438446045,
      "learning_rate": 0.0005445965930832151,
      "loss": 2.3174,
      "step": 45288
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7418907284736633,
      "learning_rate": 0.0005445942245929273,
      "loss": 2.1326,
      "step": 45289
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7097384929656982,
      "learning_rate": 0.0005445918560571648,
      "loss": 2.0755,
      "step": 45290
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.6034875512123108,
      "learning_rate": 0.0005445894874759281,
      "loss": 2.4483,
      "step": 45291
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7044628858566284,
      "learning_rate": 0.0005445871188492175,
      "loss": 2.1933,
      "step": 45292
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7346844673156738,
      "learning_rate": 0.0005445847501770336,
      "loss": 2.5675,
      "step": 45293
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7139987349510193,
      "learning_rate": 0.0005445823814593768,
      "loss": 2.4981,
      "step": 45294
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.8294224739074707,
      "learning_rate": 0.0005445800126962476,
      "loss": 2.2912,
      "step": 45295
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7011265754699707,
      "learning_rate": 0.0005445776438876463,
      "loss": 2.1692,
      "step": 45296
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7329970598220825,
      "learning_rate": 0.0005445752750335733,
      "loss": 2.2972,
      "step": 45297
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7238184213638306,
      "learning_rate": 0.0005445729061340293,
      "loss": 2.297,
      "step": 45298
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.6704439520835876,
      "learning_rate": 0.0005445705371890145,
      "loss": 2.0542,
      "step": 45299
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7176331281661987,
      "learning_rate": 0.0005445681681985294,
      "loss": 2.3046,
      "step": 45300
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7013192176818848,
      "learning_rate": 0.0005445657991625744,
      "loss": 2.447,
      "step": 45301
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.684341549873352,
      "learning_rate": 0.0005445634300811501,
      "loss": 2.3548,
      "step": 45302
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.6978092789649963,
      "learning_rate": 0.0005445610609542569,
      "loss": 2.4403,
      "step": 45303
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7090893387794495,
      "learning_rate": 0.000544558691781895,
      "loss": 2.3026,
      "step": 45304
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.6630246639251709,
      "learning_rate": 0.0005445563225640653,
      "loss": 2.2994,
      "step": 45305
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.828118622303009,
      "learning_rate": 0.0005445539533007677,
      "loss": 1.9939,
      "step": 45306
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.6418860554695129,
      "learning_rate": 0.000544551583992003,
      "loss": 2.2527,
      "step": 45307
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.6901715993881226,
      "learning_rate": 0.0005445492146377715,
      "loss": 2.0973,
      "step": 45308
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.6975268721580505,
      "learning_rate": 0.0005445468452380738,
      "loss": 2.3623,
      "step": 45309
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.710122287273407,
      "learning_rate": 0.0005445444757929101,
      "loss": 2.1848,
      "step": 45310
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.6403421759605408,
      "learning_rate": 0.000544542106302281,
      "loss": 2.0848,
      "step": 45311
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7291409373283386,
      "learning_rate": 0.0005445397367661869,
      "loss": 2.2612,
      "step": 45312
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.6565730571746826,
      "learning_rate": 0.0005445373671846283,
      "loss": 2.1898,
      "step": 45313
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.6485264897346497,
      "learning_rate": 0.0005445349975576056,
      "loss": 2.1303,
      "step": 45314
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7519777417182922,
      "learning_rate": 0.0005445326278851191,
      "loss": 2.4155,
      "step": 45315
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7352376580238342,
      "learning_rate": 0.0005445302581671695,
      "loss": 2.1175,
      "step": 45316
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7061375379562378,
      "learning_rate": 0.0005445278884037571,
      "loss": 2.0888,
      "step": 45317
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7414904832839966,
      "learning_rate": 0.0005445255185948822,
      "loss": 2.159,
      "step": 45318
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7267264127731323,
      "learning_rate": 0.0005445231487405456,
      "loss": 2.3498,
      "step": 45319
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7318135499954224,
      "learning_rate": 0.0005445207788407474,
      "loss": 2.3337,
      "step": 45320
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.6954681277275085,
      "learning_rate": 0.0005445184088954882,
      "loss": 2.362,
      "step": 45321
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7299493551254272,
      "learning_rate": 0.0005445160389047684,
      "loss": 2.2594,
      "step": 45322
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7821182012557983,
      "learning_rate": 0.0005445136688685885,
      "loss": 2.2276,
      "step": 45323
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.655665934085846,
      "learning_rate": 0.0005445112987869488,
      "loss": 2.1791,
      "step": 45324
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.6692842841148376,
      "learning_rate": 0.0005445089286598499,
      "loss": 2.3763,
      "step": 45325
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7020354866981506,
      "learning_rate": 0.0005445065584872922,
      "loss": 2.4167,
      "step": 45326
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7016215324401855,
      "learning_rate": 0.000544504188269276,
      "loss": 2.3297,
      "step": 45327
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.6880149841308594,
      "learning_rate": 0.000544501818005802,
      "loss": 2.3279,
      "step": 45328
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7360478043556213,
      "learning_rate": 0.0005444994476968704,
      "loss": 2.3743,
      "step": 45329
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7777884602546692,
      "learning_rate": 0.0005444970773424818,
      "loss": 2.3005,
      "step": 45330
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7027008533477783,
      "learning_rate": 0.0005444947069426366,
      "loss": 2.3084,
      "step": 45331
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7035503387451172,
      "learning_rate": 0.000544492336497335,
      "loss": 2.48,
      "step": 45332
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.6758074760437012,
      "learning_rate": 0.0005444899660065779,
      "loss": 2.4317,
      "step": 45333
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7199244499206543,
      "learning_rate": 0.0005444875954703655,
      "loss": 2.2255,
      "step": 45334
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7070344686508179,
      "learning_rate": 0.0005444852248886982,
      "loss": 2.345,
      "step": 45335
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7378895282745361,
      "learning_rate": 0.0005444828542615764,
      "loss": 2.1516,
      "step": 45336
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.6911367177963257,
      "learning_rate": 0.0005444804835890006,
      "loss": 2.2893,
      "step": 45337
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7219586968421936,
      "learning_rate": 0.0005444781128709714,
      "loss": 2.3638,
      "step": 45338
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.6953999400138855,
      "learning_rate": 0.0005444757421074891,
      "loss": 2.297,
      "step": 45339
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.6655656099319458,
      "learning_rate": 0.000544473371298554,
      "loss": 2.1622,
      "step": 45340
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.6938156485557556,
      "learning_rate": 0.0005444710004441669,
      "loss": 2.2167,
      "step": 45341
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.684999942779541,
      "learning_rate": 0.0005444686295443279,
      "loss": 2.3013,
      "step": 45342
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.6910728812217712,
      "learning_rate": 0.0005444662585990375,
      "loss": 2.3803,
      "step": 45343
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.6825423836708069,
      "learning_rate": 0.0005444638876082962,
      "loss": 2.2919,
      "step": 45344
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7124438285827637,
      "learning_rate": 0.0005444615165721046,
      "loss": 2.3598,
      "step": 45345
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7470538020133972,
      "learning_rate": 0.000544459145490463,
      "loss": 2.1631,
      "step": 45346
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7602262496948242,
      "learning_rate": 0.0005444567743633716,
      "loss": 2.2531,
      "step": 45347
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.759791910648346,
      "learning_rate": 0.0005444544031908313,
      "loss": 2.2694,
      "step": 45348
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7764075994491577,
      "learning_rate": 0.0005444520319728422,
      "loss": 2.3931,
      "step": 45349
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7846602201461792,
      "learning_rate": 0.0005444496607094048,
      "loss": 2.1403,
      "step": 45350
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.776012659072876,
      "learning_rate": 0.0005444472894005197,
      "loss": 2.5993,
      "step": 45351
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.8735262751579285,
      "learning_rate": 0.0005444449180461873,
      "loss": 2.4712,
      "step": 45352
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.6491332054138184,
      "learning_rate": 0.0005444425466464078,
      "loss": 2.2371,
      "step": 45353
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.6413045525550842,
      "learning_rate": 0.0005444401752011819,
      "loss": 2.4646,
      "step": 45354
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7743551731109619,
      "learning_rate": 0.00054443780371051,
      "loss": 2.4015,
      "step": 45355
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7239559292793274,
      "learning_rate": 0.0005444354321743924,
      "loss": 2.2006,
      "step": 45356
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.6866633892059326,
      "learning_rate": 0.0005444330605928297,
      "loss": 2.4579,
      "step": 45357
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7983174324035645,
      "learning_rate": 0.0005444306889658223,
      "loss": 2.4273,
      "step": 45358
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7138120532035828,
      "learning_rate": 0.0005444283172933706,
      "loss": 2.151,
      "step": 45359
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.6657349467277527,
      "learning_rate": 0.0005444259455754751,
      "loss": 1.9666,
      "step": 45360
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.8368076086044312,
      "learning_rate": 0.0005444235738121361,
      "loss": 2.2939,
      "step": 45361
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.8178200721740723,
      "learning_rate": 0.0005444212020033543,
      "loss": 2.2741,
      "step": 45362
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.8909131288528442,
      "learning_rate": 0.0005444188301491298,
      "loss": 2.1336,
      "step": 45363
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.740386962890625,
      "learning_rate": 0.0005444164582494634,
      "loss": 2.3145,
      "step": 45364
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.6722273826599121,
      "learning_rate": 0.0005444140863043552,
      "loss": 2.1726,
      "step": 45365
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7168483734130859,
      "learning_rate": 0.0005444117143138059,
      "loss": 2.3737,
      "step": 45366
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7703740000724792,
      "learning_rate": 0.0005444093422778158,
      "loss": 2.566,
      "step": 45367
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.6354094743728638,
      "learning_rate": 0.0005444069701963855,
      "loss": 2.2961,
      "step": 45368
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.6953826546669006,
      "learning_rate": 0.0005444045980695152,
      "loss": 2.1756,
      "step": 45369
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.8106463551521301,
      "learning_rate": 0.0005444022258972056,
      "loss": 2.2266,
      "step": 45370
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.6362101435661316,
      "learning_rate": 0.000544399853679457,
      "loss": 2.4662,
      "step": 45371
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7005284428596497,
      "learning_rate": 0.0005443974814162698,
      "loss": 2.3316,
      "step": 45372
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7406900525093079,
      "learning_rate": 0.0005443951091076444,
      "loss": 2.287,
      "step": 45373
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7788083553314209,
      "learning_rate": 0.0005443927367535814,
      "loss": 2.3564,
      "step": 45374
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.6801301836967468,
      "learning_rate": 0.0005443903643540812,
      "loss": 2.4744,
      "step": 45375
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.8367108702659607,
      "learning_rate": 0.0005443879919091442,
      "loss": 2.3755,
      "step": 45376
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7116593718528748,
      "learning_rate": 0.0005443856194187709,
      "loss": 2.122,
      "step": 45377
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.653975784778595,
      "learning_rate": 0.0005443832468829616,
      "loss": 2.1616,
      "step": 45378
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7431994080543518,
      "learning_rate": 0.000544380874301717,
      "loss": 2.2568,
      "step": 45379
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.8384870290756226,
      "learning_rate": 0.0005443785016750373,
      "loss": 2.3035,
      "step": 45380
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7204433083534241,
      "learning_rate": 0.0005443761290029231,
      "loss": 2.1008,
      "step": 45381
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7966304421424866,
      "learning_rate": 0.0005443737562853746,
      "loss": 2.2675,
      "step": 45382
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7092794179916382,
      "learning_rate": 0.0005443713835223925,
      "loss": 2.1538,
      "step": 45383
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.6798006296157837,
      "learning_rate": 0.0005443690107139772,
      "loss": 2.2286,
      "step": 45384
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.8391019701957703,
      "learning_rate": 0.000544366637860129,
      "loss": 2.4563,
      "step": 45385
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.8052781820297241,
      "learning_rate": 0.0005443642649608485,
      "loss": 2.3044,
      "step": 45386
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.6940578818321228,
      "learning_rate": 0.000544361892016136,
      "loss": 2.3317,
      "step": 45387
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.6551464796066284,
      "learning_rate": 0.000544359519025992,
      "loss": 2.0515,
      "step": 45388
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.6653830409049988,
      "learning_rate": 0.000544357145990417,
      "loss": 2.2388,
      "step": 45389
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.6537920236587524,
      "learning_rate": 0.0005443547729094114,
      "loss": 2.2014,
      "step": 45390
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7775511145591736,
      "learning_rate": 0.0005443523997829757,
      "loss": 2.0247,
      "step": 45391
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7058229446411133,
      "learning_rate": 0.0005443500266111101,
      "loss": 2.2259,
      "step": 45392
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7066851258277893,
      "learning_rate": 0.0005443476533938154,
      "loss": 2.3664,
      "step": 45393
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7222789525985718,
      "learning_rate": 0.0005443452801310917,
      "loss": 2.1269,
      "step": 45394
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7219680547714233,
      "learning_rate": 0.0005443429068229397,
      "loss": 2.0821,
      "step": 45395
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7054089307785034,
      "learning_rate": 0.0005443405334693598,
      "loss": 2.5176,
      "step": 45396
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.716296374797821,
      "learning_rate": 0.0005443381600703523,
      "loss": 2.2554,
      "step": 45397
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7031478881835938,
      "learning_rate": 0.0005443357866259177,
      "loss": 2.3194,
      "step": 45398
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.6807116866111755,
      "learning_rate": 0.0005443334131360564,
      "loss": 2.2109,
      "step": 45399
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7459506988525391,
      "learning_rate": 0.0005443310396007691,
      "loss": 2.3024,
      "step": 45400
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.6641572713851929,
      "learning_rate": 0.0005443286660200559,
      "loss": 2.4173,
      "step": 45401
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.6724855899810791,
      "learning_rate": 0.0005443262923939175,
      "loss": 2.217,
      "step": 45402
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.6953956484794617,
      "learning_rate": 0.0005443239187223541,
      "loss": 2.1362,
      "step": 45403
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.6450598239898682,
      "learning_rate": 0.0005443215450053664,
      "loss": 2.132,
      "step": 45404
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.6595102548599243,
      "learning_rate": 0.0005443191712429545,
      "loss": 2.3552,
      "step": 45405
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7357807159423828,
      "learning_rate": 0.0005443167974351193,
      "loss": 2.3056,
      "step": 45406
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7673431038856506,
      "learning_rate": 0.0005443144235818609,
      "loss": 2.1936,
      "step": 45407
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.6568432450294495,
      "learning_rate": 0.0005443120496831799,
      "loss": 2.4241,
      "step": 45408
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.6542638540267944,
      "learning_rate": 0.0005443096757390765,
      "loss": 2.2517,
      "step": 45409
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.6679314970970154,
      "learning_rate": 0.0005443073017495515,
      "loss": 2.1779,
      "step": 45410
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.881659746170044,
      "learning_rate": 0.0005443049277146051,
      "loss": 2.348,
      "step": 45411
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7565373778343201,
      "learning_rate": 0.0005443025536342377,
      "loss": 2.4591,
      "step": 45412
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.6774353384971619,
      "learning_rate": 0.00054430017950845,
      "loss": 2.4581,
      "step": 45413
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.6528006196022034,
      "learning_rate": 0.0005442978053372423,
      "loss": 2.0301,
      "step": 45414
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.6913399696350098,
      "learning_rate": 0.0005442954311206149,
      "loss": 2.3294,
      "step": 45415
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7374823093414307,
      "learning_rate": 0.0005442930568585685,
      "loss": 2.1189,
      "step": 45416
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.6831468343734741,
      "learning_rate": 0.0005442906825511032,
      "loss": 2.1903,
      "step": 45417
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7104930877685547,
      "learning_rate": 0.0005442883081982199,
      "loss": 2.2614,
      "step": 45418
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7101612091064453,
      "learning_rate": 0.0005442859337999186,
      "loss": 2.2049,
      "step": 45419
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.6682112812995911,
      "learning_rate": 0.0005442835593562,
      "loss": 2.2614,
      "step": 45420
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.6890681385993958,
      "learning_rate": 0.0005442811848670644,
      "loss": 2.2081,
      "step": 45421
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7010055184364319,
      "learning_rate": 0.0005442788103325125,
      "loss": 2.2941,
      "step": 45422
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.6859009861946106,
      "learning_rate": 0.0005442764357525444,
      "loss": 2.4474,
      "step": 45423
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.709675669670105,
      "learning_rate": 0.0005442740611271608,
      "loss": 2.2572,
      "step": 45424
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.763942003250122,
      "learning_rate": 0.0005442716864563621,
      "loss": 2.2961,
      "step": 45425
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7161461710929871,
      "learning_rate": 0.0005442693117401486,
      "loss": 2.2997,
      "step": 45426
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.810495138168335,
      "learning_rate": 0.0005442669369785207,
      "loss": 2.1962,
      "step": 45427
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7438540458679199,
      "learning_rate": 0.0005442645621714791,
      "loss": 2.3921,
      "step": 45428
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7529118061065674,
      "learning_rate": 0.000544262187319024,
      "loss": 2.473,
      "step": 45429
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.6914166212081909,
      "learning_rate": 0.000544259812421156,
      "loss": 2.288,
      "step": 45430
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.6358933448791504,
      "learning_rate": 0.0005442574374778756,
      "loss": 2.3369,
      "step": 45431
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.9042178988456726,
      "learning_rate": 0.000544255062489183,
      "loss": 2.2224,
      "step": 45432
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7656285166740417,
      "learning_rate": 0.0005442526874550788,
      "loss": 2.2865,
      "step": 45433
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.6890520453453064,
      "learning_rate": 0.0005442503123755635,
      "loss": 2.4221,
      "step": 45434
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7866694927215576,
      "learning_rate": 0.0005442479372506373,
      "loss": 2.3657,
      "step": 45435
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.6934879422187805,
      "learning_rate": 0.000544245562080301,
      "loss": 2.2867,
      "step": 45436
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7235968708992004,
      "learning_rate": 0.0005442431868645547,
      "loss": 2.0638,
      "step": 45437
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.6479379534721375,
      "learning_rate": 0.0005442408116033989,
      "loss": 2.1828,
      "step": 45438
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7221183776855469,
      "learning_rate": 0.0005442384362968343,
      "loss": 2.1324,
      "step": 45439
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.8559706807136536,
      "learning_rate": 0.000544236060944861,
      "loss": 2.6409,
      "step": 45440
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.6870532631874084,
      "learning_rate": 0.0005442336855474798,
      "loss": 2.5415,
      "step": 45441
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7151297926902771,
      "learning_rate": 0.0005442313101046909,
      "loss": 2.3132,
      "step": 45442
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.6441948413848877,
      "learning_rate": 0.0005442289346164948,
      "loss": 2.1911,
      "step": 45443
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7698947191238403,
      "learning_rate": 0.0005442265590828918,
      "loss": 2.2258,
      "step": 45444
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7233268022537231,
      "learning_rate": 0.0005442241835038825,
      "loss": 2.1853,
      "step": 45445
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.6542342901229858,
      "learning_rate": 0.0005442218078794673,
      "loss": 2.2146,
      "step": 45446
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7381162643432617,
      "learning_rate": 0.0005442194322096468,
      "loss": 2.1606,
      "step": 45447
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7473454475402832,
      "learning_rate": 0.0005442170564944212,
      "loss": 2.4966,
      "step": 45448
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7627376914024353,
      "learning_rate": 0.000544214680733791,
      "loss": 2.2734,
      "step": 45449
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.696505069732666,
      "learning_rate": 0.0005442123049277568,
      "loss": 2.2397,
      "step": 45450
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7313539981842041,
      "learning_rate": 0.0005442099290763188,
      "loss": 2.2557,
      "step": 45451
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7147855758666992,
      "learning_rate": 0.0005442075531794776,
      "loss": 2.2837,
      "step": 45452
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7995299100875854,
      "learning_rate": 0.0005442051772372336,
      "loss": 2.3509,
      "step": 45453
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7574194669723511,
      "learning_rate": 0.0005442028012495873,
      "loss": 2.269,
      "step": 45454
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.6796947121620178,
      "learning_rate": 0.0005442004252165392,
      "loss": 2.2522,
      "step": 45455
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.6755115389823914,
      "learning_rate": 0.0005441980491380894,
      "loss": 2.1846,
      "step": 45456
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.6762566566467285,
      "learning_rate": 0.0005441956730142387,
      "loss": 2.3419,
      "step": 45457
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.6885209083557129,
      "learning_rate": 0.0005441932968449873,
      "loss": 2.2693,
      "step": 45458
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.682460367679596,
      "learning_rate": 0.0005441909206303359,
      "loss": 2.271,
      "step": 45459
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7110214233398438,
      "learning_rate": 0.0005441885443702848,
      "loss": 1.9834,
      "step": 45460
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7780501246452332,
      "learning_rate": 0.0005441861680648343,
      "loss": 2.0898,
      "step": 45461
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7805249691009521,
      "learning_rate": 0.000544183791713985,
      "loss": 2.0628,
      "step": 45462
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7839930653572083,
      "learning_rate": 0.0005441814153177374,
      "loss": 2.3128,
      "step": 45463
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7780398726463318,
      "learning_rate": 0.0005441790388760919,
      "loss": 2.2574,
      "step": 45464
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.6798415184020996,
      "learning_rate": 0.0005441766623890488,
      "loss": 2.3786,
      "step": 45465
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.6750006079673767,
      "learning_rate": 0.0005441742858566087,
      "loss": 2.343,
      "step": 45466
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7006856203079224,
      "learning_rate": 0.000544171909278772,
      "loss": 2.3778,
      "step": 45467
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7002543210983276,
      "learning_rate": 0.0005441695326555392,
      "loss": 2.2007,
      "step": 45468
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7164534330368042,
      "learning_rate": 0.0005441671559869106,
      "loss": 2.3225,
      "step": 45469
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.744249165058136,
      "learning_rate": 0.0005441647792728866,
      "loss": 2.4522,
      "step": 45470
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.6681188344955444,
      "learning_rate": 0.0005441624025134679,
      "loss": 2.2002,
      "step": 45471
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7328798770904541,
      "learning_rate": 0.0005441600257086547,
      "loss": 2.2915,
      "step": 45472
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7630165219306946,
      "learning_rate": 0.0005441576488584477,
      "loss": 2.1934,
      "step": 45473
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7599602937698364,
      "learning_rate": 0.0005441552719628472,
      "loss": 2.1667,
      "step": 45474
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.8284958004951477,
      "learning_rate": 0.0005441528950218535,
      "loss": 2.3661,
      "step": 45475
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.6331518888473511,
      "learning_rate": 0.0005441505180354671,
      "loss": 2.1846,
      "step": 45476
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.048707127571106,
      "learning_rate": 0.0005441481410036887,
      "loss": 2.2501,
      "step": 45477
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.6839073300361633,
      "learning_rate": 0.0005441457639265184,
      "loss": 2.3369,
      "step": 45478
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.9302040934562683,
      "learning_rate": 0.0005441433868039569,
      "loss": 2.384,
      "step": 45479
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7973890900611877,
      "learning_rate": 0.0005441410096360045,
      "loss": 2.3052,
      "step": 45480
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.6376107931137085,
      "learning_rate": 0.0005441386324226617,
      "loss": 2.0362,
      "step": 45481
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.6807172894477844,
      "learning_rate": 0.0005441362551639289,
      "loss": 2.3363,
      "step": 45482
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7525897026062012,
      "learning_rate": 0.0005441338778598065,
      "loss": 2.3298,
      "step": 45483
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.726459264755249,
      "learning_rate": 0.000544131500510295,
      "loss": 2.1562,
      "step": 45484
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.8446177244186401,
      "learning_rate": 0.000544129123115395,
      "loss": 2.3566,
      "step": 45485
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.8246825933456421,
      "learning_rate": 0.0005441267456751067,
      "loss": 2.3265,
      "step": 45486
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.6979565620422363,
      "learning_rate": 0.0005441243681894305,
      "loss": 2.3116,
      "step": 45487
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.6713582873344421,
      "learning_rate": 0.0005441219906583671,
      "loss": 2.3747,
      "step": 45488
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7875845432281494,
      "learning_rate": 0.0005441196130819169,
      "loss": 2.3312,
      "step": 45489
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.6861170530319214,
      "learning_rate": 0.00054411723546008,
      "loss": 2.1964,
      "step": 45490
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7101329565048218,
      "learning_rate": 0.0005441148577928573,
      "loss": 2.1532,
      "step": 45491
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.698380172252655,
      "learning_rate": 0.0005441124800802491,
      "loss": 2.6604,
      "step": 45492
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7237759828567505,
      "learning_rate": 0.0005441101023222557,
      "loss": 2.3192,
      "step": 45493
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.6650364398956299,
      "learning_rate": 0.0005441077245188776,
      "loss": 2.1512,
      "step": 45494
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7957738637924194,
      "learning_rate": 0.0005441053466701152,
      "loss": 2.3266,
      "step": 45495
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7531262040138245,
      "learning_rate": 0.000544102968775969,
      "loss": 2.1826,
      "step": 45496
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7219875454902649,
      "learning_rate": 0.0005441005908364396,
      "loss": 2.2043,
      "step": 45497
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.6994971632957458,
      "learning_rate": 0.0005440982128515272,
      "loss": 2.4703,
      "step": 45498
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7689657807350159,
      "learning_rate": 0.0005440958348212324,
      "loss": 2.0138,
      "step": 45499
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7173407673835754,
      "learning_rate": 0.0005440934567455555,
      "loss": 2.4037,
      "step": 45500
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.8222644329071045,
      "learning_rate": 0.000544091078624497,
      "loss": 2.2599,
      "step": 45501
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.8401618599891663,
      "learning_rate": 0.0005440887004580574,
      "loss": 2.2312,
      "step": 45502
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7047402262687683,
      "learning_rate": 0.0005440863222462372,
      "loss": 2.2231,
      "step": 45503
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7621013522148132,
      "learning_rate": 0.0005440839439890366,
      "loss": 2.3268,
      "step": 45504
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.6892667412757874,
      "learning_rate": 0.0005440815656864563,
      "loss": 2.386,
      "step": 45505
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.6581245064735413,
      "learning_rate": 0.0005440791873384965,
      "loss": 2.3063,
      "step": 45506
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7825301289558411,
      "learning_rate": 0.0005440768089451579,
      "loss": 2.303,
      "step": 45507
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.6991109848022461,
      "learning_rate": 0.0005440744305064407,
      "loss": 2.3314,
      "step": 45508
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.1596845388412476,
      "learning_rate": 0.0005440720520223456,
      "loss": 2.3382,
      "step": 45509
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7741931676864624,
      "learning_rate": 0.0005440696734928728,
      "loss": 2.4612,
      "step": 45510
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.0524852275848389,
      "learning_rate": 0.0005440672949180228,
      "loss": 2.3492,
      "step": 45511
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.1777914762496948,
      "learning_rate": 0.0005440649162977961,
      "loss": 2.314,
      "step": 45512
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.6865884065628052,
      "learning_rate": 0.0005440625376321932,
      "loss": 2.5008,
      "step": 45513
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.9057422280311584,
      "learning_rate": 0.0005440601589212145,
      "loss": 2.0995,
      "step": 45514
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7826802730560303,
      "learning_rate": 0.0005440577801648603,
      "loss": 2.3771,
      "step": 45515
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7026862502098083,
      "learning_rate": 0.0005440554013631312,
      "loss": 2.0942,
      "step": 45516
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.6693249940872192,
      "learning_rate": 0.0005440530225160276,
      "loss": 2.103,
      "step": 45517
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.8120289444923401,
      "learning_rate": 0.0005440506436235499,
      "loss": 2.3201,
      "step": 45518
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7771075963973999,
      "learning_rate": 0.0005440482646856986,
      "loss": 2.1173,
      "step": 45519
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7942773103713989,
      "learning_rate": 0.0005440458857024741,
      "loss": 2.5206,
      "step": 45520
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.6912121176719666,
      "learning_rate": 0.0005440435066738768,
      "loss": 2.0374,
      "step": 45521
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.9183928370475769,
      "learning_rate": 0.0005440411275999074,
      "loss": 2.1984,
      "step": 45522
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.6612002849578857,
      "learning_rate": 0.000544038748480566,
      "loss": 2.0339,
      "step": 45523
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.6936126351356506,
      "learning_rate": 0.0005440363693158532,
      "loss": 2.1209,
      "step": 45524
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.8403648734092712,
      "learning_rate": 0.0005440339901057695,
      "loss": 2.567,
      "step": 45525
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.6959254741668701,
      "learning_rate": 0.0005440316108503153,
      "loss": 2.3986,
      "step": 45526
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.6877262592315674,
      "learning_rate": 0.0005440292315494909,
      "loss": 2.2229,
      "step": 45527
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7075034976005554,
      "learning_rate": 0.0005440268522032969,
      "loss": 2.3026,
      "step": 45528
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.6890959143638611,
      "learning_rate": 0.0005440244728117337,
      "loss": 2.2201,
      "step": 45529
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.701871395111084,
      "learning_rate": 0.0005440220933748017,
      "loss": 2.2472,
      "step": 45530
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.8025308847427368,
      "learning_rate": 0.0005440197138925015,
      "loss": 2.1459,
      "step": 45531
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.9765541553497314,
      "learning_rate": 0.0005440173343648334,
      "loss": 2.4468,
      "step": 45532
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.6948211789131165,
      "learning_rate": 0.0005440149547917978,
      "loss": 2.3622,
      "step": 45533
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.6837143301963806,
      "learning_rate": 0.0005440125751733952,
      "loss": 2.2821,
      "step": 45534
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.6474882364273071,
      "learning_rate": 0.0005440101955096262,
      "loss": 2.2732,
      "step": 45535
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7200354933738708,
      "learning_rate": 0.000544007815800491,
      "loss": 2.3996,
      "step": 45536
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.8369287252426147,
      "learning_rate": 0.0005440054360459902,
      "loss": 2.174,
      "step": 45537
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.6522208452224731,
      "learning_rate": 0.0005440030562461241,
      "loss": 2.1077,
      "step": 45538
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7893528342247009,
      "learning_rate": 0.0005440006764008933,
      "loss": 2.1469,
      "step": 45539
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7049874067306519,
      "learning_rate": 0.000543998296510298,
      "loss": 2.1416,
      "step": 45540
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7698708772659302,
      "learning_rate": 0.000543995916574339,
      "loss": 2.4125,
      "step": 45541
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.6902058124542236,
      "learning_rate": 0.0005439935365930165,
      "loss": 2.0853,
      "step": 45542
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.8446957468986511,
      "learning_rate": 0.000543991156566331,
      "loss": 2.0473,
      "step": 45543
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.9020742774009705,
      "learning_rate": 0.000543988776494283,
      "loss": 2.0119,
      "step": 45544
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7442538142204285,
      "learning_rate": 0.0005439863963768727,
      "loss": 2.2859,
      "step": 45545
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.9421591758728027,
      "learning_rate": 0.0005439840162141009,
      "loss": 2.5191,
      "step": 45546
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.8444653153419495,
      "learning_rate": 0.0005439816360059679,
      "loss": 2.5321,
      "step": 45547
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7946355938911438,
      "learning_rate": 0.0005439792557524739,
      "loss": 2.3444,
      "step": 45548
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.876751184463501,
      "learning_rate": 0.0005439768754536197,
      "loss": 2.5697,
      "step": 45549
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.0905998945236206,
      "learning_rate": 0.0005439744951094055,
      "loss": 2.2274,
      "step": 45550
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7946195602416992,
      "learning_rate": 0.0005439721147198319,
      "loss": 2.2387,
      "step": 45551
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7268058061599731,
      "learning_rate": 0.0005439697342848993,
      "loss": 2.1506,
      "step": 45552
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.8181207776069641,
      "learning_rate": 0.000543967353804608,
      "loss": 2.2403,
      "step": 45553
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.839262068271637,
      "learning_rate": 0.0005439649732789587,
      "loss": 2.2498,
      "step": 45554
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7730296850204468,
      "learning_rate": 0.0005439625927079517,
      "loss": 2.4539,
      "step": 45555
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7495759129524231,
      "learning_rate": 0.0005439602120915874,
      "loss": 2.2576,
      "step": 45556
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7416378855705261,
      "learning_rate": 0.0005439578314298662,
      "loss": 2.4258,
      "step": 45557
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7718064188957214,
      "learning_rate": 0.0005439554507227888,
      "loss": 2.3643,
      "step": 45558
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7236809730529785,
      "learning_rate": 0.0005439530699703554,
      "loss": 2.5175,
      "step": 45559
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7414790987968445,
      "learning_rate": 0.0005439506891725665,
      "loss": 2.2013,
      "step": 45560
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7773333191871643,
      "learning_rate": 0.0005439483083294227,
      "loss": 2.2068,
      "step": 45561
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7429558038711548,
      "learning_rate": 0.0005439459274409241,
      "loss": 2.2322,
      "step": 45562
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7691649198532104,
      "learning_rate": 0.0005439435465070714,
      "loss": 2.1478,
      "step": 45563
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7131608724594116,
      "learning_rate": 0.0005439411655278651,
      "loss": 2.1171,
      "step": 45564
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.6744388341903687,
      "learning_rate": 0.0005439387845033055,
      "loss": 2.2881,
      "step": 45565
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.8132792711257935,
      "learning_rate": 0.000543936403433393,
      "loss": 2.2029,
      "step": 45566
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7002090215682983,
      "learning_rate": 0.0005439340223181282,
      "loss": 2.3846,
      "step": 45567
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7454506754875183,
      "learning_rate": 0.0005439316411575114,
      "loss": 2.3945,
      "step": 45568
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.6423481106758118,
      "learning_rate": 0.0005439292599515431,
      "loss": 2.1415,
      "step": 45569
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7706181406974792,
      "learning_rate": 0.0005439268787002238,
      "loss": 2.2383,
      "step": 45570
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.6776573657989502,
      "learning_rate": 0.0005439244974035539,
      "loss": 2.2079,
      "step": 45571
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.001465082168579,
      "learning_rate": 0.0005439221160615337,
      "loss": 2.1106,
      "step": 45572
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7464613318443298,
      "learning_rate": 0.0005439197346741639,
      "loss": 2.278,
      "step": 45573
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.6782883405685425,
      "learning_rate": 0.0005439173532414448,
      "loss": 2.3136,
      "step": 45574
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.6949065923690796,
      "learning_rate": 0.0005439149717633767,
      "loss": 2.3429,
      "step": 45575
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.637950599193573,
      "learning_rate": 0.0005439125902399604,
      "loss": 2.5437,
      "step": 45576
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.6875051259994507,
      "learning_rate": 0.0005439102086711961,
      "loss": 2.339,
      "step": 45577
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7181955575942993,
      "learning_rate": 0.0005439078270570843,
      "loss": 2.0898,
      "step": 45578
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.64201819896698,
      "learning_rate": 0.0005439054453976254,
      "loss": 2.1662,
      "step": 45579
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.6535115838050842,
      "learning_rate": 0.0005439030636928199,
      "loss": 2.3511,
      "step": 45580
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.6650540232658386,
      "learning_rate": 0.0005439006819426682,
      "loss": 2.087,
      "step": 45581
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7196215987205505,
      "learning_rate": 0.0005438983001471707,
      "loss": 2.2141,
      "step": 45582
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.8177695870399475,
      "learning_rate": 0.000543895918306328,
      "loss": 2.2755,
      "step": 45583
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.6940461993217468,
      "learning_rate": 0.0005438935364201404,
      "loss": 2.2673,
      "step": 45584
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.8293153643608093,
      "learning_rate": 0.0005438911544886083,
      "loss": 2.35,
      "step": 45585
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.8244814276695251,
      "learning_rate": 0.0005438887725117324,
      "loss": 2.0283,
      "step": 45586
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.6668856739997864,
      "learning_rate": 0.000543886390489513,
      "loss": 2.1991,
      "step": 45587
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7160289287567139,
      "learning_rate": 0.0005438840084219503,
      "loss": 2.2644,
      "step": 45588
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.9554432034492493,
      "learning_rate": 0.0005438816263090451,
      "loss": 2.0958,
      "step": 45589
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7075430154800415,
      "learning_rate": 0.0005438792441507976,
      "loss": 2.3586,
      "step": 45590
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7056288719177246,
      "learning_rate": 0.0005438768619472085,
      "loss": 2.455,
      "step": 45591
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7997410297393799,
      "learning_rate": 0.000543874479698278,
      "loss": 2.1456,
      "step": 45592
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.65318763256073,
      "learning_rate": 0.0005438720974040066,
      "loss": 2.1104,
      "step": 45593
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.6363823413848877,
      "learning_rate": 0.0005438697150643948,
      "loss": 2.3716,
      "step": 45594
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7338547706604004,
      "learning_rate": 0.0005438673326794431,
      "loss": 2.0143,
      "step": 45595
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.6898996829986572,
      "learning_rate": 0.0005438649502491516,
      "loss": 2.0922,
      "step": 45596
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.6880847215652466,
      "learning_rate": 0.0005438625677735213,
      "loss": 2.3138,
      "step": 45597
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7761890292167664,
      "learning_rate": 0.0005438601852525522,
      "loss": 2.376,
      "step": 45598
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7803677916526794,
      "learning_rate": 0.000543857802686245,
      "loss": 2.3079,
      "step": 45599
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.6751161217689514,
      "learning_rate": 0.0005438554200745999,
      "loss": 2.2324,
      "step": 45600
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7487549781799316,
      "learning_rate": 0.0005438530374176175,
      "loss": 2.2699,
      "step": 45601
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7050512433052063,
      "learning_rate": 0.0005438506547152983,
      "loss": 2.2587,
      "step": 45602
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.6522430181503296,
      "learning_rate": 0.0005438482719676426,
      "loss": 2.0278,
      "step": 45603
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7038490176200867,
      "learning_rate": 0.0005438458891746508,
      "loss": 2.2727,
      "step": 45604
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.6802762746810913,
      "learning_rate": 0.0005438435063363236,
      "loss": 2.0522,
      "step": 45605
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7077404260635376,
      "learning_rate": 0.0005438411234526613,
      "loss": 2.2412,
      "step": 45606
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7149590849876404,
      "learning_rate": 0.0005438387405236642,
      "loss": 1.9452,
      "step": 45607
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7433879971504211,
      "learning_rate": 0.000543836357549333,
      "loss": 2.2803,
      "step": 45608
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.6869075298309326,
      "learning_rate": 0.000543833974529668,
      "loss": 2.3041,
      "step": 45609
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.6932998895645142,
      "learning_rate": 0.0005438315914646696,
      "loss": 2.3168,
      "step": 45610
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7085170149803162,
      "learning_rate": 0.0005438292083543384,
      "loss": 2.2147,
      "step": 45611
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.6668121218681335,
      "learning_rate": 0.0005438268251986747,
      "loss": 2.4574,
      "step": 45612
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7164260745048523,
      "learning_rate": 0.0005438244419976788,
      "loss": 2.0278,
      "step": 45613
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.6194619536399841,
      "learning_rate": 0.0005438220587513516,
      "loss": 2.3508,
      "step": 45614
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7110164761543274,
      "learning_rate": 0.0005438196754596931,
      "loss": 2.3781,
      "step": 45615
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7803772687911987,
      "learning_rate": 0.0005438172921227041,
      "loss": 2.2028,
      "step": 45616
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7022218704223633,
      "learning_rate": 0.0005438149087403848,
      "loss": 2.119,
      "step": 45617
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7231503129005432,
      "learning_rate": 0.0005438125253127357,
      "loss": 1.9942,
      "step": 45618
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.6777170896530151,
      "learning_rate": 0.0005438101418397571,
      "loss": 2.065,
      "step": 45619
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7141894698143005,
      "learning_rate": 0.0005438077583214498,
      "loss": 2.2904,
      "step": 45620
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.6632763743400574,
      "learning_rate": 0.0005438053747578139,
      "loss": 2.3265,
      "step": 45621
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7386813163757324,
      "learning_rate": 0.00054380299114885,
      "loss": 2.4735,
      "step": 45622
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7254332304000854,
      "learning_rate": 0.0005438006074945586,
      "loss": 2.574,
      "step": 45623
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7072517275810242,
      "learning_rate": 0.00054379822379494,
      "loss": 2.3802,
      "step": 45624
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.8755053281784058,
      "learning_rate": 0.0005437958400499948,
      "loss": 2.3742,
      "step": 45625
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7442500591278076,
      "learning_rate": 0.0005437934562597231,
      "loss": 2.2736,
      "step": 45626
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7687379121780396,
      "learning_rate": 0.0005437910724241259,
      "loss": 2.1464,
      "step": 45627
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.9187772870063782,
      "learning_rate": 0.0005437886885432032,
      "loss": 2.5049,
      "step": 45628
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7290201187133789,
      "learning_rate": 0.0005437863046169556,
      "loss": 2.1494,
      "step": 45629
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7178642153739929,
      "learning_rate": 0.0005437839206453834,
      "loss": 2.1877,
      "step": 45630
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.6480646729469299,
      "learning_rate": 0.0005437815366284873,
      "loss": 2.2931,
      "step": 45631
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.8445007801055908,
      "learning_rate": 0.0005437791525662677,
      "loss": 2.5291,
      "step": 45632
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.6762719750404358,
      "learning_rate": 0.0005437767684587248,
      "loss": 2.3361,
      "step": 45633
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.705947995185852,
      "learning_rate": 0.0005437743843058593,
      "loss": 2.5478,
      "step": 45634
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7298749685287476,
      "learning_rate": 0.0005437720001076715,
      "loss": 2.1693,
      "step": 45635
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.6841930747032166,
      "learning_rate": 0.0005437696158641618,
      "loss": 2.5584,
      "step": 45636
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.6636483669281006,
      "learning_rate": 0.0005437672315753308,
      "loss": 2.1378,
      "step": 45637
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7259510159492493,
      "learning_rate": 0.0005437648472411788,
      "loss": 2.1518,
      "step": 45638
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7155489921569824,
      "learning_rate": 0.0005437624628617064,
      "loss": 2.1901,
      "step": 45639
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.6645383834838867,
      "learning_rate": 0.0005437600784369139,
      "loss": 2.3084,
      "step": 45640
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7060385346412659,
      "learning_rate": 0.0005437576939668018,
      "loss": 2.1563,
      "step": 45641
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.6944172978401184,
      "learning_rate": 0.0005437553094513706,
      "loss": 2.2839,
      "step": 45642
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.6652764678001404,
      "learning_rate": 0.0005437529248906207,
      "loss": 2.2137,
      "step": 45643
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7495677471160889,
      "learning_rate": 0.0005437505402845525,
      "loss": 2.2512,
      "step": 45644
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.6969311237335205,
      "learning_rate": 0.0005437481556331664,
      "loss": 2.2363,
      "step": 45645
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7359083890914917,
      "learning_rate": 0.000543745770936463,
      "loss": 2.2253,
      "step": 45646
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.6781530976295471,
      "learning_rate": 0.0005437433861944426,
      "loss": 2.2852,
      "step": 45647
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.8106445670127869,
      "learning_rate": 0.0005437410014071058,
      "loss": 2.0848,
      "step": 45648
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.9457045793533325,
      "learning_rate": 0.0005437386165744529,
      "loss": 2.1631,
      "step": 45649
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.8277490735054016,
      "learning_rate": 0.0005437362316964844,
      "loss": 2.194,
      "step": 45650
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7390843033790588,
      "learning_rate": 0.0005437338467732007,
      "loss": 2.3619,
      "step": 45651
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7365818619728088,
      "learning_rate": 0.0005437314618046023,
      "loss": 2.2462,
      "step": 45652
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.6775662302970886,
      "learning_rate": 0.0005437290767906896,
      "loss": 2.0889,
      "step": 45653
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.6897047162055969,
      "learning_rate": 0.0005437266917314631,
      "loss": 2.4851,
      "step": 45654
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7482145428657532,
      "learning_rate": 0.0005437243066269232,
      "loss": 2.2022,
      "step": 45655
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7066938281059265,
      "learning_rate": 0.0005437219214770703,
      "loss": 2.2813,
      "step": 45656
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.6836010813713074,
      "learning_rate": 0.0005437195362819049,
      "loss": 2.1044,
      "step": 45657
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.671569287776947,
      "learning_rate": 0.0005437171510414275,
      "loss": 2.235,
      "step": 45658
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7013264894485474,
      "learning_rate": 0.0005437147657556385,
      "loss": 2.403,
      "step": 45659
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7293945550918579,
      "learning_rate": 0.0005437123804245382,
      "loss": 2.5514,
      "step": 45660
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.6444916725158691,
      "learning_rate": 0.0005437099950481272,
      "loss": 2.1537,
      "step": 45661
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.6983782649040222,
      "learning_rate": 0.000543707609626406,
      "loss": 2.2296,
      "step": 45662
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.703347384929657,
      "learning_rate": 0.0005437052241593749,
      "loss": 2.0672,
      "step": 45663
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7459237575531006,
      "learning_rate": 0.0005437028386470344,
      "loss": 2.0712,
      "step": 45664
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7633787393569946,
      "learning_rate": 0.000543700453089385,
      "loss": 2.3499,
      "step": 45665
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7141441106796265,
      "learning_rate": 0.0005436980674864269,
      "loss": 2.3759,
      "step": 45666
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7474645376205444,
      "learning_rate": 0.000543695681838161,
      "loss": 2.1602,
      "step": 45667
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.9255700707435608,
      "learning_rate": 0.0005436932961445873,
      "loss": 2.1965,
      "step": 45668
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7632657289505005,
      "learning_rate": 0.0005436909104057065,
      "loss": 2.1957,
      "step": 45669
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7044353485107422,
      "learning_rate": 0.0005436885246215188,
      "loss": 2.2253,
      "step": 45670
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.8436059355735779,
      "learning_rate": 0.000543686138792025,
      "loss": 2.2425,
      "step": 45671
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.8696678876876831,
      "learning_rate": 0.0005436837529172253,
      "loss": 2.2489,
      "step": 45672
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7623080611228943,
      "learning_rate": 0.00054368136699712,
      "loss": 2.0822,
      "step": 45673
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.8697737455368042,
      "learning_rate": 0.00054367898103171,
      "loss": 2.3384,
      "step": 45674
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.000527262687683,
      "learning_rate": 0.0005436765950209954,
      "loss": 2.2743,
      "step": 45675
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7252081036567688,
      "learning_rate": 0.0005436742089649767,
      "loss": 2.2035,
      "step": 45676
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.6548424363136292,
      "learning_rate": 0.0005436718228636543,
      "loss": 1.9737,
      "step": 45677
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7476993799209595,
      "learning_rate": 0.0005436694367170287,
      "loss": 2.2128,
      "step": 45678
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7470564842224121,
      "learning_rate": 0.0005436670505251004,
      "loss": 2.3354,
      "step": 45679
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7507882118225098,
      "learning_rate": 0.0005436646642878698,
      "loss": 2.1931,
      "step": 45680
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.6509478092193604,
      "learning_rate": 0.0005436622780053375,
      "loss": 1.9662,
      "step": 45681
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.6943311095237732,
      "learning_rate": 0.0005436598916775035,
      "loss": 2.3778,
      "step": 45682
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.8554462790489197,
      "learning_rate": 0.0005436575053043687,
      "loss": 2.0471,
      "step": 45683
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.6672885417938232,
      "learning_rate": 0.0005436551188859333,
      "loss": 2.046,
      "step": 45684
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.8172805905342102,
      "learning_rate": 0.0005436527324221979,
      "loss": 2.4505,
      "step": 45685
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7854565978050232,
      "learning_rate": 0.0005436503459131627,
      "loss": 2.0385,
      "step": 45686
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.6977800726890564,
      "learning_rate": 0.0005436479593588285,
      "loss": 2.2414,
      "step": 45687
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.9551343321800232,
      "learning_rate": 0.0005436455727591954,
      "loss": 2.2882,
      "step": 45688
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.676746666431427,
      "learning_rate": 0.0005436431861142641,
      "loss": 2.0683,
      "step": 45689
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.6482614278793335,
      "learning_rate": 0.0005436407994240347,
      "loss": 2.2928,
      "step": 45690
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7680632472038269,
      "learning_rate": 0.0005436384126885081,
      "loss": 2.3151,
      "step": 45691
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7460297346115112,
      "learning_rate": 0.0005436360259076845,
      "loss": 2.4014,
      "step": 45692
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7595162391662598,
      "learning_rate": 0.0005436336390815643,
      "loss": 2.2654,
      "step": 45693
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.660071074962616,
      "learning_rate": 0.0005436312522101481,
      "loss": 2.1,
      "step": 45694
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7018402814865112,
      "learning_rate": 0.0005436288652934362,
      "loss": 2.2545,
      "step": 45695
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.6870164275169373,
      "learning_rate": 0.000543626478331429,
      "loss": 2.2645,
      "step": 45696
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.636763334274292,
      "learning_rate": 0.0005436240913241271,
      "loss": 2.385,
      "step": 45697
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.6791229248046875,
      "learning_rate": 0.000543621704271531,
      "loss": 2.3548,
      "step": 45698
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.8455894589424133,
      "learning_rate": 0.0005436193171736408,
      "loss": 2.4332,
      "step": 45699
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7287036776542664,
      "learning_rate": 0.0005436169300304573,
      "loss": 2.3841,
      "step": 45700
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.6792547702789307,
      "learning_rate": 0.0005436145428419809,
      "loss": 2.2774,
      "step": 45701
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7079198360443115,
      "learning_rate": 0.0005436121556082118,
      "loss": 2.2001,
      "step": 45702
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.6419703364372253,
      "learning_rate": 0.0005436097683291508,
      "loss": 2.3708,
      "step": 45703
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7704774737358093,
      "learning_rate": 0.0005436073810047979,
      "loss": 2.1946,
      "step": 45704
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7362657189369202,
      "learning_rate": 0.0005436049936351539,
      "loss": 2.5575,
      "step": 45705
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.654496431350708,
      "learning_rate": 0.0005436026062202191,
      "loss": 2.3189,
      "step": 45706
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7447599172592163,
      "learning_rate": 0.0005436002187599942,
      "loss": 2.1986,
      "step": 45707
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7140571475028992,
      "learning_rate": 0.0005435978312544791,
      "loss": 2.1508,
      "step": 45708
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7844418883323669,
      "learning_rate": 0.0005435954437036747,
      "loss": 2.1773,
      "step": 45709
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7103595733642578,
      "learning_rate": 0.0005435930561075813,
      "loss": 2.235,
      "step": 45710
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7398917078971863,
      "learning_rate": 0.0005435906684661994,
      "loss": 2.3502,
      "step": 45711
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7425373196601868,
      "learning_rate": 0.0005435882807795294,
      "loss": 2.3929,
      "step": 45712
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7118502855300903,
      "learning_rate": 0.0005435858930475717,
      "loss": 2.44,
      "step": 45713
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.6249401569366455,
      "learning_rate": 0.0005435835052703268,
      "loss": 2.2371,
      "step": 45714
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.8260560631752014,
      "learning_rate": 0.0005435811174477951,
      "loss": 2.2812,
      "step": 45715
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7076142430305481,
      "learning_rate": 0.0005435787295799771,
      "loss": 2.1661,
      "step": 45716
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7430626153945923,
      "learning_rate": 0.0005435763416668732,
      "loss": 2.1802,
      "step": 45717
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7300789952278137,
      "learning_rate": 0.0005435739537084838,
      "loss": 2.2777,
      "step": 45718
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.6148079037666321,
      "learning_rate": 0.0005435715657048095,
      "loss": 2.1468,
      "step": 45719
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7666390538215637,
      "learning_rate": 0.0005435691776558507,
      "loss": 2.2301,
      "step": 45720
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7121397256851196,
      "learning_rate": 0.0005435667895616076,
      "loss": 2.0167,
      "step": 45721
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.6491742134094238,
      "learning_rate": 0.000543564401422081,
      "loss": 2.1151,
      "step": 45722
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.681565523147583,
      "learning_rate": 0.000543562013237271,
      "loss": 2.2912,
      "step": 45723
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7135611772537231,
      "learning_rate": 0.0005435596250071785,
      "loss": 2.2076,
      "step": 45724
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.8183755278587341,
      "learning_rate": 0.0005435572367318035,
      "loss": 2.3043,
      "step": 45725
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.640417754650116,
      "learning_rate": 0.0005435548484111466,
      "loss": 2.3185,
      "step": 45726
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.6803309917449951,
      "learning_rate": 0.0005435524600452083,
      "loss": 2.3667,
      "step": 45727
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.6859673261642456,
      "learning_rate": 0.000543550071633989,
      "loss": 2.2715,
      "step": 45728
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.6849029064178467,
      "learning_rate": 0.0005435476831774891,
      "loss": 2.2241,
      "step": 45729
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7293452620506287,
      "learning_rate": 0.0005435452946757091,
      "loss": 2.45,
      "step": 45730
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.6955270767211914,
      "learning_rate": 0.0005435429061286495,
      "loss": 2.2543,
      "step": 45731
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7043671607971191,
      "learning_rate": 0.0005435405175363106,
      "loss": 2.148,
      "step": 45732
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.721003532409668,
      "learning_rate": 0.0005435381288986929,
      "loss": 2.3358,
      "step": 45733
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.6845194697380066,
      "learning_rate": 0.0005435357402157969,
      "loss": 2.4758,
      "step": 45734
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.6828582882881165,
      "learning_rate": 0.000543533351487623,
      "loss": 2.1585,
      "step": 45735
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.6968381404876709,
      "learning_rate": 0.0005435309627141717,
      "loss": 2.3521,
      "step": 45736
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.733189046382904,
      "learning_rate": 0.0005435285738954434,
      "loss": 2.2741,
      "step": 45737
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7074993252754211,
      "learning_rate": 0.0005435261850314385,
      "loss": 2.0861,
      "step": 45738
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7701911330223083,
      "learning_rate": 0.0005435237961221575,
      "loss": 2.2738,
      "step": 45739
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.6528182625770569,
      "learning_rate": 0.0005435214071676008,
      "loss": 2.2578,
      "step": 45740
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.692676842212677,
      "learning_rate": 0.000543519018167769,
      "loss": 2.1366,
      "step": 45741
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.70378577709198,
      "learning_rate": 0.0005435166291226622,
      "loss": 2.057,
      "step": 45742
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7399398684501648,
      "learning_rate": 0.0005435142400322813,
      "loss": 2.1107,
      "step": 45743
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7267073392868042,
      "learning_rate": 0.0005435118508966264,
      "loss": 2.4525,
      "step": 45744
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7434991598129272,
      "learning_rate": 0.000543509461715698,
      "loss": 2.0605,
      "step": 45745
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7221654653549194,
      "learning_rate": 0.0005435070724894966,
      "loss": 2.3056,
      "step": 45746
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.707472562789917,
      "learning_rate": 0.0005435046832180226,
      "loss": 2.5202,
      "step": 45747
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.8041200637817383,
      "learning_rate": 0.0005435022939012765,
      "loss": 2.1246,
      "step": 45748
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7985724806785583,
      "learning_rate": 0.0005434999045392588,
      "loss": 2.2142,
      "step": 45749
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7529407143592834,
      "learning_rate": 0.00054349751513197,
      "loss": 2.2026,
      "step": 45750
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.6903209090232849,
      "learning_rate": 0.0005434951256794102,
      "loss": 2.4415,
      "step": 45751
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.702633261680603,
      "learning_rate": 0.0005434927361815801,
      "loss": 2.1302,
      "step": 45752
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7318461537361145,
      "learning_rate": 0.0005434903466384802,
      "loss": 2.3244,
      "step": 45753
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.8369781374931335,
      "learning_rate": 0.0005434879570501108,
      "loss": 2.347,
      "step": 45754
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.6879624724388123,
      "learning_rate": 0.0005434855674164723,
      "loss": 2.3053,
      "step": 45755
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7284192442893982,
      "learning_rate": 0.0005434831777375653,
      "loss": 2.6306,
      "step": 45756
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7448039650917053,
      "learning_rate": 0.0005434807880133902,
      "loss": 2.1768,
      "step": 45757
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.6644452810287476,
      "learning_rate": 0.0005434783982439475,
      "loss": 2.4426,
      "step": 45758
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.6895758509635925,
      "learning_rate": 0.0005434760084292374,
      "loss": 2.2938,
      "step": 45759
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.6656456589698792,
      "learning_rate": 0.0005434736185692606,
      "loss": 2.1421,
      "step": 45760
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7518264651298523,
      "learning_rate": 0.0005434712286640174,
      "loss": 2.1273,
      "step": 45761
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.6708759665489197,
      "learning_rate": 0.0005434688387135084,
      "loss": 2.1426,
      "step": 45762
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7213081121444702,
      "learning_rate": 0.0005434664487177339,
      "loss": 2.0644,
      "step": 45763
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.6898430585861206,
      "learning_rate": 0.0005434640586766945,
      "loss": 2.157,
      "step": 45764
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.740522563457489,
      "learning_rate": 0.0005434616685903904,
      "loss": 2.598,
      "step": 45765
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7006681561470032,
      "learning_rate": 0.0005434592784588222,
      "loss": 2.2651,
      "step": 45766
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7427464723587036,
      "learning_rate": 0.0005434568882819903,
      "loss": 2.2924,
      "step": 45767
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7093484997749329,
      "learning_rate": 0.0005434544980598952,
      "loss": 2.1831,
      "step": 45768
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7017993330955505,
      "learning_rate": 0.0005434521077925374,
      "loss": 2.3325,
      "step": 45769
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.6367404460906982,
      "learning_rate": 0.0005434497174799172,
      "loss": 2.2265,
      "step": 45770
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7426519989967346,
      "learning_rate": 0.000543447327122035,
      "loss": 2.2393,
      "step": 45771
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7145007252693176,
      "learning_rate": 0.0005434449367188915,
      "loss": 2.3088,
      "step": 45772
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7179461717605591,
      "learning_rate": 0.0005434425462704868,
      "loss": 2.32,
      "step": 45773
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7231781482696533,
      "learning_rate": 0.0005434401557768217,
      "loss": 2.1587,
      "step": 45774
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7249732613563538,
      "learning_rate": 0.0005434377652378964,
      "loss": 2.1749,
      "step": 45775
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.6623517870903015,
      "learning_rate": 0.0005434353746537115,
      "loss": 2.3935,
      "step": 45776
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.722577691078186,
      "learning_rate": 0.0005434329840242673,
      "loss": 2.3261,
      "step": 45777
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7157939672470093,
      "learning_rate": 0.0005434305933495644,
      "loss": 2.1322,
      "step": 45778
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7371517419815063,
      "learning_rate": 0.0005434282026296031,
      "loss": 1.9788,
      "step": 45779
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7055179476737976,
      "learning_rate": 0.0005434258118643838,
      "loss": 2.2924,
      "step": 45780
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7234471440315247,
      "learning_rate": 0.0005434234210539071,
      "loss": 2.159,
      "step": 45781
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.6923347115516663,
      "learning_rate": 0.0005434210301981735,
      "loss": 2.0359,
      "step": 45782
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.6766456961631775,
      "learning_rate": 0.0005434186392971833,
      "loss": 2.2431,
      "step": 45783
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.677419126033783,
      "learning_rate": 0.0005434162483509369,
      "loss": 2.3325,
      "step": 45784
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7555345296859741,
      "learning_rate": 0.0005434138573594349,
      "loss": 2.4135,
      "step": 45785
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.6952663660049438,
      "learning_rate": 0.0005434114663226777,
      "loss": 2.2738,
      "step": 45786
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7096775770187378,
      "learning_rate": 0.0005434090752406656,
      "loss": 2.4236,
      "step": 45787
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.745909571647644,
      "learning_rate": 0.0005434066841133993,
      "loss": 2.4086,
      "step": 45788
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7149842977523804,
      "learning_rate": 0.000543404292940879,
      "loss": 2.225,
      "step": 45789
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7688727378845215,
      "learning_rate": 0.0005434019017231051,
      "loss": 2.3373,
      "step": 45790
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.8408730626106262,
      "learning_rate": 0.0005433995104600786,
      "loss": 2.5615,
      "step": 45791
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.6515539884567261,
      "learning_rate": 0.0005433971191517991,
      "loss": 2.4319,
      "step": 45792
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.910768985748291,
      "learning_rate": 0.0005433947277982678,
      "loss": 2.2584,
      "step": 45793
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.6936249732971191,
      "learning_rate": 0.0005433923363994846,
      "loss": 2.2231,
      "step": 45794
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.6927105188369751,
      "learning_rate": 0.0005433899449554504,
      "loss": 2.6186,
      "step": 45795
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.9226663708686829,
      "learning_rate": 0.0005433875534661653,
      "loss": 2.1483,
      "step": 45796
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7092804312705994,
      "learning_rate": 0.0005433851619316299,
      "loss": 2.1716,
      "step": 45797
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7937191724777222,
      "learning_rate": 0.0005433827703518444,
      "loss": 2.1629,
      "step": 45798
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7883421778678894,
      "learning_rate": 0.0005433803787268097,
      "loss": 2.4381,
      "step": 45799
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.6763474941253662,
      "learning_rate": 0.0005433779870565259,
      "loss": 2.2356,
      "step": 45800
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.8168469071388245,
      "learning_rate": 0.0005433755953409934,
      "loss": 2.4177,
      "step": 45801
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.740630030632019,
      "learning_rate": 0.000543373203580213,
      "loss": 2.234,
      "step": 45802
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.8252732157707214,
      "learning_rate": 0.0005433708117741848,
      "loss": 2.3019,
      "step": 45803
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7596940994262695,
      "learning_rate": 0.0005433684199229094,
      "loss": 2.3336,
      "step": 45804
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7848809361457825,
      "learning_rate": 0.0005433660280263873,
      "loss": 2.2389,
      "step": 45805
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.6651548743247986,
      "learning_rate": 0.0005433636360846188,
      "loss": 2.2991,
      "step": 45806
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7620632648468018,
      "learning_rate": 0.0005433612440976043,
      "loss": 2.2133,
      "step": 45807
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.795878529548645,
      "learning_rate": 0.0005433588520653445,
      "loss": 2.1735,
      "step": 45808
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.6722381114959717,
      "learning_rate": 0.0005433564599878396,
      "loss": 2.2224,
      "step": 45809
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7266121506690979,
      "learning_rate": 0.0005433540678650902,
      "loss": 2.3897,
      "step": 45810
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.6504825353622437,
      "learning_rate": 0.0005433516756970967,
      "loss": 2.2578,
      "step": 45811
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7135067582130432,
      "learning_rate": 0.0005433492834838595,
      "loss": 2.3964,
      "step": 45812
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.6837726831436157,
      "learning_rate": 0.0005433468912253791,
      "loss": 2.4068,
      "step": 45813
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.6462016105651855,
      "learning_rate": 0.0005433444989216559,
      "loss": 2.1675,
      "step": 45814
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.6618162989616394,
      "learning_rate": 0.0005433421065726904,
      "loss": 2.0643,
      "step": 45815
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.6950786113739014,
      "learning_rate": 0.0005433397141784828,
      "loss": 2.0958,
      "step": 45816
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7052890658378601,
      "learning_rate": 0.000543337321739034,
      "loss": 2.1298,
      "step": 45817
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7462231516838074,
      "learning_rate": 0.0005433349292543442,
      "loss": 2.0768,
      "step": 45818
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7854385375976562,
      "learning_rate": 0.0005433325367244137,
      "loss": 2.3652,
      "step": 45819
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.6799994707107544,
      "learning_rate": 0.0005433301441492432,
      "loss": 2.3036,
      "step": 45820
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.688666820526123,
      "learning_rate": 0.0005433277515288329,
      "loss": 2.2029,
      "step": 45821
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.6841837763786316,
      "learning_rate": 0.0005433253588631835,
      "loss": 2.1617,
      "step": 45822
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7138550877571106,
      "learning_rate": 0.0005433229661522953,
      "loss": 2.3374,
      "step": 45823
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7366800308227539,
      "learning_rate": 0.0005433205733961688,
      "loss": 2.1277,
      "step": 45824
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7221220135688782,
      "learning_rate": 0.0005433181805948042,
      "loss": 2.2649,
      "step": 45825
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.8185142278671265,
      "learning_rate": 0.0005433157877482024,
      "loss": 2.0563,
      "step": 45826
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.735897421836853,
      "learning_rate": 0.0005433133948563634,
      "loss": 2.2924,
      "step": 45827
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7132818698883057,
      "learning_rate": 0.0005433110019192881,
      "loss": 2.1205,
      "step": 45828
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7758775353431702,
      "learning_rate": 0.0005433086089369765,
      "loss": 1.9602,
      "step": 45829
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.8137350678443909,
      "learning_rate": 0.0005433062159094294,
      "loss": 2.1422,
      "step": 45830
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7513675093650818,
      "learning_rate": 0.0005433038228366469,
      "loss": 2.0364,
      "step": 45831
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.8250401020050049,
      "learning_rate": 0.0005433014297186298,
      "loss": 2.335,
      "step": 45832
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7605966925621033,
      "learning_rate": 0.0005432990365553782,
      "loss": 2.1791,
      "step": 45833
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7154225707054138,
      "learning_rate": 0.0005432966433468929,
      "loss": 2.2768,
      "step": 45834
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7775304913520813,
      "learning_rate": 0.000543294250093174,
      "loss": 2.0883,
      "step": 45835
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.855664849281311,
      "learning_rate": 0.0005432918567942222,
      "loss": 2.3135,
      "step": 45836
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.9352271556854248,
      "learning_rate": 0.0005432894634500379,
      "loss": 2.1907,
      "step": 45837
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7205047607421875,
      "learning_rate": 0.0005432870700606213,
      "loss": 2.3256,
      "step": 45838
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7467912435531616,
      "learning_rate": 0.0005432846766259731,
      "loss": 2.157,
      "step": 45839
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.9695811867713928,
      "learning_rate": 0.0005432822831460937,
      "loss": 2.1788,
      "step": 45840
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.6820196509361267,
      "learning_rate": 0.0005432798896209836,
      "loss": 2.0255,
      "step": 45841
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.6830863356590271,
      "learning_rate": 0.0005432774960506432,
      "loss": 2.0025,
      "step": 45842
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.0612921714782715,
      "learning_rate": 0.0005432751024350729,
      "loss": 2.0219,
      "step": 45843
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.6353856921195984,
      "learning_rate": 0.0005432727087742731,
      "loss": 1.9659,
      "step": 45844
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.9214428067207336,
      "learning_rate": 0.0005432703150682443,
      "loss": 1.964,
      "step": 45845
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.733497679233551,
      "learning_rate": 0.0005432679213169871,
      "loss": 2.1391,
      "step": 45846
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.747800886631012,
      "learning_rate": 0.0005432655275205017,
      "loss": 2.2475,
      "step": 45847
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7364044785499573,
      "learning_rate": 0.0005432631336787885,
      "loss": 2.3142,
      "step": 45848
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7933785319328308,
      "learning_rate": 0.0005432607397918483,
      "loss": 2.1506,
      "step": 45849
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.729983925819397,
      "learning_rate": 0.0005432583458596813,
      "loss": 2.2603,
      "step": 45850
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.6737359762191772,
      "learning_rate": 0.0005432559518822879,
      "loss": 2.1861,
      "step": 45851
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.8131904602050781,
      "learning_rate": 0.0005432535578596688,
      "loss": 2.2703,
      "step": 45852
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7002347707748413,
      "learning_rate": 0.0005432511637918241,
      "loss": 2.1628,
      "step": 45853
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7855413556098938,
      "learning_rate": 0.0005432487696787544,
      "loss": 2.3789,
      "step": 45854
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7209798693656921,
      "learning_rate": 0.0005432463755204603,
      "loss": 2.1013,
      "step": 45855
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7123457789421082,
      "learning_rate": 0.0005432439813169421,
      "loss": 2.1951,
      "step": 45856
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7027974128723145,
      "learning_rate": 0.0005432415870682001,
      "loss": 2.401,
      "step": 45857
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.6680698394775391,
      "learning_rate": 0.000543239192774235,
      "loss": 2.2107,
      "step": 45858
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7009525895118713,
      "learning_rate": 0.0005432367984350472,
      "loss": 2.3486,
      "step": 45859
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7134054899215698,
      "learning_rate": 0.000543234404050637,
      "loss": 2.2776,
      "step": 45860
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7870684266090393,
      "learning_rate": 0.000543232009621005,
      "loss": 2.3896,
      "step": 45861
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7825015187263489,
      "learning_rate": 0.0005432296151461515,
      "loss": 2.1822,
      "step": 45862
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.6755107045173645,
      "learning_rate": 0.000543227220626077,
      "loss": 2.1171,
      "step": 45863
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.8020227551460266,
      "learning_rate": 0.0005432248260607821,
      "loss": 2.4826,
      "step": 45864
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.9929565191268921,
      "learning_rate": 0.0005432224314502671,
      "loss": 2.3443,
      "step": 45865
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.6682844758033752,
      "learning_rate": 0.0005432200367945323,
      "loss": 2.2545,
      "step": 45866
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7819428443908691,
      "learning_rate": 0.0005432176420935784,
      "loss": 2.3359,
      "step": 45867
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.6898670792579651,
      "learning_rate": 0.0005432152473474058,
      "loss": 2.4391,
      "step": 45868
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.6536076068878174,
      "learning_rate": 0.0005432128525560148,
      "loss": 2.1343,
      "step": 45869
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7878853678703308,
      "learning_rate": 0.000543210457719406,
      "loss": 2.3467,
      "step": 45870
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7166067361831665,
      "learning_rate": 0.0005432080628375798,
      "loss": 2.0978,
      "step": 45871
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.8359286785125732,
      "learning_rate": 0.0005432056679105366,
      "loss": 2.2095,
      "step": 45872
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.701231837272644,
      "learning_rate": 0.0005432032729382769,
      "loss": 2.5526,
      "step": 45873
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7375995516777039,
      "learning_rate": 0.000543200877920801,
      "loss": 2.0869,
      "step": 45874
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.698691189289093,
      "learning_rate": 0.0005431984828581097,
      "loss": 2.2448,
      "step": 45875
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7752489447593689,
      "learning_rate": 0.0005431960877502031,
      "loss": 2.2666,
      "step": 45876
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7624173760414124,
      "learning_rate": 0.0005431936925970815,
      "loss": 2.2519,
      "step": 45877
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.6969292759895325,
      "learning_rate": 0.0005431912973987459,
      "loss": 2.6339,
      "step": 45878
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7804326415061951,
      "learning_rate": 0.0005431889021551964,
      "loss": 2.2292,
      "step": 45879
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.6738428473472595,
      "learning_rate": 0.0005431865068664335,
      "loss": 2.2833,
      "step": 45880
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.6996961236000061,
      "learning_rate": 0.0005431841115324577,
      "loss": 2.3462,
      "step": 45881
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7494253516197205,
      "learning_rate": 0.0005431817161532693,
      "loss": 2.2173,
      "step": 45882
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7834216356277466,
      "learning_rate": 0.0005431793207288688,
      "loss": 2.4109,
      "step": 45883
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7310279011726379,
      "learning_rate": 0.0005431769252592566,
      "loss": 2.3209,
      "step": 45884
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7127630114555359,
      "learning_rate": 0.0005431745297444333,
      "loss": 2.0708,
      "step": 45885
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7000077366828918,
      "learning_rate": 0.0005431721341843993,
      "loss": 2.2493,
      "step": 45886
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7954592108726501,
      "learning_rate": 0.000543169738579155,
      "loss": 2.3179,
      "step": 45887
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.691547691822052,
      "learning_rate": 0.000543167342928701,
      "loss": 2.2647,
      "step": 45888
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.725662112236023,
      "learning_rate": 0.0005431649472330373,
      "loss": 2.3899,
      "step": 45889
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7251359224319458,
      "learning_rate": 0.0005431625514921648,
      "loss": 2.3646,
      "step": 45890
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.6604259610176086,
      "learning_rate": 0.0005431601557060838,
      "loss": 2.2699,
      "step": 45891
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7373802065849304,
      "learning_rate": 0.0005431577598747948,
      "loss": 2.4262,
      "step": 45892
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7103350758552551,
      "learning_rate": 0.000543155363998298,
      "loss": 2.2954,
      "step": 45893
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.6081085205078125,
      "learning_rate": 0.0005431529680765941,
      "loss": 2.0133,
      "step": 45894
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.6471406817436218,
      "learning_rate": 0.0005431505721096836,
      "loss": 2.3507,
      "step": 45895
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7231257557868958,
      "learning_rate": 0.0005431481760975667,
      "loss": 2.1805,
      "step": 45896
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7032644748687744,
      "learning_rate": 0.000543145780040244,
      "loss": 2.3664,
      "step": 45897
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.74141925573349,
      "learning_rate": 0.0005431433839377159,
      "loss": 2.1473,
      "step": 45898
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.6848357319831848,
      "learning_rate": 0.0005431409877899828,
      "loss": 2.1885,
      "step": 45899
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7526175379753113,
      "learning_rate": 0.0005431385915970453,
      "loss": 2.2123,
      "step": 45900
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7248474955558777,
      "learning_rate": 0.0005431361953589037,
      "loss": 2.4966,
      "step": 45901
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.8541720509529114,
      "learning_rate": 0.0005431337990755585,
      "loss": 2.3384,
      "step": 45902
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.066953420639038,
      "learning_rate": 0.0005431314027470101,
      "loss": 2.2144,
      "step": 45903
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.657345175743103,
      "learning_rate": 0.000543129006373259,
      "loss": 2.3517,
      "step": 45904
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.8684794902801514,
      "learning_rate": 0.0005431266099543057,
      "loss": 2.2795,
      "step": 45905
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.9016759991645813,
      "learning_rate": 0.0005431242134901504,
      "loss": 2.3124,
      "step": 45906
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.6051374077796936,
      "learning_rate": 0.0005431218169807939,
      "loss": 2.3383,
      "step": 45907
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.6925665140151978,
      "learning_rate": 0.0005431194204262365,
      "loss": 2.2743,
      "step": 45908
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.6876119375228882,
      "learning_rate": 0.0005431170238264784,
      "loss": 2.3649,
      "step": 45909
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.8054169416427612,
      "learning_rate": 0.0005431146271815204,
      "loss": 2.4684,
      "step": 45910
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.6605544090270996,
      "learning_rate": 0.0005431122304913626,
      "loss": 2.3352,
      "step": 45911
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.6828722953796387,
      "learning_rate": 0.000543109833756006,
      "loss": 2.3237,
      "step": 45912
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7015798687934875,
      "learning_rate": 0.0005431074369754505,
      "loss": 2.3774,
      "step": 45913
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7897151112556458,
      "learning_rate": 0.0005431050401496967,
      "loss": 2.2565,
      "step": 45914
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7566322684288025,
      "learning_rate": 0.0005431026432787452,
      "loss": 2.3434,
      "step": 45915
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7066665291786194,
      "learning_rate": 0.0005431002463625963,
      "loss": 2.3391,
      "step": 45916
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.8162945508956909,
      "learning_rate": 0.0005430978494012504,
      "loss": 2.2348,
      "step": 45917
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.727281391620636,
      "learning_rate": 0.000543095452394708,
      "loss": 2.374,
      "step": 45918
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.6580910086631775,
      "learning_rate": 0.0005430930553429696,
      "loss": 2.4157,
      "step": 45919
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.6924219727516174,
      "learning_rate": 0.0005430906582460357,
      "loss": 2.2795,
      "step": 45920
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7162759900093079,
      "learning_rate": 0.0005430882611039065,
      "loss": 2.2823,
      "step": 45921
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.6920410990715027,
      "learning_rate": 0.0005430858639165827,
      "loss": 2.3406,
      "step": 45922
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7647935152053833,
      "learning_rate": 0.0005430834666840647,
      "loss": 2.4519,
      "step": 45923
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.691831648349762,
      "learning_rate": 0.0005430810694063528,
      "loss": 2.3791,
      "step": 45924
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7932907938957214,
      "learning_rate": 0.0005430786720834476,
      "loss": 2.302,
      "step": 45925
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7275428175926208,
      "learning_rate": 0.0005430762747153495,
      "loss": 2.2244,
      "step": 45926
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7110625505447388,
      "learning_rate": 0.0005430738773020589,
      "loss": 1.9868,
      "step": 45927
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7153303027153015,
      "learning_rate": 0.0005430714798435762,
      "loss": 2.1124,
      "step": 45928
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7151195406913757,
      "learning_rate": 0.000543069082339902,
      "loss": 2.3096,
      "step": 45929
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7342796325683594,
      "learning_rate": 0.0005430666847910366,
      "loss": 2.2644,
      "step": 45930
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7179582118988037,
      "learning_rate": 0.0005430642871969807,
      "loss": 2.216,
      "step": 45931
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.684813380241394,
      "learning_rate": 0.0005430618895577344,
      "loss": 2.2106,
      "step": 45932
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.6580467820167542,
      "learning_rate": 0.0005430594918732984,
      "loss": 2.3367,
      "step": 45933
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7840414643287659,
      "learning_rate": 0.0005430570941436731,
      "loss": 2.3568,
      "step": 45934
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.720770001411438,
      "learning_rate": 0.0005430546963688587,
      "loss": 2.3609,
      "step": 45935
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.6771062612533569,
      "learning_rate": 0.0005430522985488559,
      "loss": 2.2079,
      "step": 45936
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.6946414113044739,
      "learning_rate": 0.0005430499006836652,
      "loss": 2.3059,
      "step": 45937
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.8187048435211182,
      "learning_rate": 0.0005430475027732869,
      "loss": 2.2214,
      "step": 45938
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.6751086711883545,
      "learning_rate": 0.0005430451048177215,
      "loss": 2.3117,
      "step": 45939
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7915266156196594,
      "learning_rate": 0.0005430427068169693,
      "loss": 2.0877,
      "step": 45940
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.8380581736564636,
      "learning_rate": 0.0005430403087710309,
      "loss": 2.0456,
      "step": 45941
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.6908159852027893,
      "learning_rate": 0.0005430379106799068,
      "loss": 2.151,
      "step": 45942
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7710651159286499,
      "learning_rate": 0.0005430355125435974,
      "loss": 2.378,
      "step": 45943
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.8184790015220642,
      "learning_rate": 0.0005430331143621031,
      "loss": 2.2816,
      "step": 45944
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7184902429580688,
      "learning_rate": 0.0005430307161354242,
      "loss": 1.9614,
      "step": 45945
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7315855622291565,
      "learning_rate": 0.0005430283178635615,
      "loss": 2.1662,
      "step": 45946
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.8414426445960999,
      "learning_rate": 0.000543025919546515,
      "loss": 2.1349,
      "step": 45947
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7540003061294556,
      "learning_rate": 0.0005430235211842856,
      "loss": 2.2987,
      "step": 45948
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7349799871444702,
      "learning_rate": 0.0005430211227768736,
      "loss": 2.0582,
      "step": 45949
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.6977233290672302,
      "learning_rate": 0.0005430187243242792,
      "loss": 2.1529,
      "step": 45950
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7378800511360168,
      "learning_rate": 0.0005430163258265032,
      "loss": 1.989,
      "step": 45951
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.8424441814422607,
      "learning_rate": 0.0005430139272835457,
      "loss": 2.199,
      "step": 45952
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7029076814651489,
      "learning_rate": 0.0005430115286954074,
      "loss": 2.3218,
      "step": 45953
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.8628395199775696,
      "learning_rate": 0.0005430091300620887,
      "loss": 2.0455,
      "step": 45954
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7919850945472717,
      "learning_rate": 0.0005430067313835899,
      "loss": 2.0934,
      "step": 45955
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7419233322143555,
      "learning_rate": 0.0005430043326599117,
      "loss": 2.1924,
      "step": 45956
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.8107555508613586,
      "learning_rate": 0.0005430019338910544,
      "loss": 2.3357,
      "step": 45957
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.791999876499176,
      "learning_rate": 0.0005429995350770183,
      "loss": 2.4488,
      "step": 45958
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7622981071472168,
      "learning_rate": 0.0005429971362178042,
      "loss": 2.5501,
      "step": 45959
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.9705753922462463,
      "learning_rate": 0.0005429947373134122,
      "loss": 2.3908,
      "step": 45960
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7518180012702942,
      "learning_rate": 0.000542992338363843,
      "loss": 2.4605,
      "step": 45961
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7158810496330261,
      "learning_rate": 0.000542989939369097,
      "loss": 2.4754,
      "step": 45962
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.756443977355957,
      "learning_rate": 0.0005429875403291744,
      "loss": 2.2238,
      "step": 45963
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.9403918981552124,
      "learning_rate": 0.000542985141244076,
      "loss": 2.4292,
      "step": 45964
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7416672110557556,
      "learning_rate": 0.0005429827421138018,
      "loss": 2.4173,
      "step": 45965
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7227804660797119,
      "learning_rate": 0.0005429803429383527,
      "loss": 2.2367,
      "step": 45966
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.9858960509300232,
      "learning_rate": 0.000542977943717729,
      "loss": 2.281,
      "step": 45967
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.68289715051651,
      "learning_rate": 0.0005429755444519311,
      "loss": 2.2522,
      "step": 45968
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7552261352539062,
      "learning_rate": 0.0005429731451409595,
      "loss": 2.27,
      "step": 45969
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7285750508308411,
      "learning_rate": 0.0005429707457848145,
      "loss": 2.1721,
      "step": 45970
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.9178253412246704,
      "learning_rate": 0.0005429683463834968,
      "loss": 2.3626,
      "step": 45971
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7141332626342773,
      "learning_rate": 0.0005429659469370065,
      "loss": 2.26,
      "step": 45972
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7054517865180969,
      "learning_rate": 0.0005429635474453444,
      "loss": 2.1207,
      "step": 45973
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7532625794410706,
      "learning_rate": 0.0005429611479085106,
      "loss": 2.2642,
      "step": 45974
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7310280799865723,
      "learning_rate": 0.000542958748326506,
      "loss": 2.1796,
      "step": 45975
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7309885621070862,
      "learning_rate": 0.0005429563486993305,
      "loss": 2.3618,
      "step": 45976
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7339574694633484,
      "learning_rate": 0.000542953949026985,
      "loss": 2.4682,
      "step": 45977
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7476788759231567,
      "learning_rate": 0.0005429515493094698,
      "loss": 2.262,
      "step": 45978
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7461860179901123,
      "learning_rate": 0.0005429491495467854,
      "loss": 2.2011,
      "step": 45979
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.6903083920478821,
      "learning_rate": 0.000542946749738932,
      "loss": 2.2554,
      "step": 45980
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.667140781879425,
      "learning_rate": 0.0005429443498859102,
      "loss": 2.2183,
      "step": 45981
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7147269248962402,
      "learning_rate": 0.0005429419499877206,
      "loss": 2.2881,
      "step": 45982
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7563881278038025,
      "learning_rate": 0.0005429395500443633,
      "loss": 2.1564,
      "step": 45983
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.6921879649162292,
      "learning_rate": 0.0005429371500558391,
      "loss": 2.1408,
      "step": 45984
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7275959849357605,
      "learning_rate": 0.0005429347500221484,
      "loss": 2.3607,
      "step": 45985
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7345103025436401,
      "learning_rate": 0.0005429323499432913,
      "loss": 2.3505,
      "step": 45986
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7865944504737854,
      "learning_rate": 0.0005429299498192686,
      "loss": 2.2429,
      "step": 45987
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7058544754981995,
      "learning_rate": 0.0005429275496500807,
      "loss": 2.2711,
      "step": 45988
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.6881567239761353,
      "learning_rate": 0.0005429251494357279,
      "loss": 2.5188,
      "step": 45989
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.676160991191864,
      "learning_rate": 0.0005429227491762107,
      "loss": 2.1865,
      "step": 45990
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.6562017202377319,
      "learning_rate": 0.0005429203488715297,
      "loss": 2.0654,
      "step": 45991
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7744596004486084,
      "learning_rate": 0.0005429179485216852,
      "loss": 2.094,
      "step": 45992
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.8976073861122131,
      "learning_rate": 0.0005429155481266776,
      "loss": 2.2876,
      "step": 45993
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.6869946122169495,
      "learning_rate": 0.0005429131476865073,
      "loss": 2.3499,
      "step": 45994
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.6963444352149963,
      "learning_rate": 0.000542910747201175,
      "loss": 2.4342,
      "step": 45995
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7135627865791321,
      "learning_rate": 0.000542908346670681,
      "loss": 2.2709,
      "step": 45996
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.6999461650848389,
      "learning_rate": 0.0005429059460950258,
      "loss": 2.1817,
      "step": 45997
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7862212061882019,
      "learning_rate": 0.0005429035454742097,
      "loss": 2.2337,
      "step": 45998
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.6941041350364685,
      "learning_rate": 0.0005429011448082334,
      "loss": 2.315,
      "step": 45999
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7098286151885986,
      "learning_rate": 0.0005428987440970971,
      "loss": 2.1686,
      "step": 46000
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7053884267807007,
      "learning_rate": 0.0005428963433408013,
      "loss": 2.3798,
      "step": 46001
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.6936838030815125,
      "learning_rate": 0.0005428939425393465,
      "loss": 2.3129,
      "step": 46002
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.8049421310424805,
      "learning_rate": 0.0005428915416927331,
      "loss": 2.4237,
      "step": 46003
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.723976731300354,
      "learning_rate": 0.0005428891408009616,
      "loss": 2.2629,
      "step": 46004
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.6782933473587036,
      "learning_rate": 0.0005428867398640325,
      "loss": 2.2374,
      "step": 46005
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7413604855537415,
      "learning_rate": 0.0005428843388819461,
      "loss": 2.28,
      "step": 46006
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7253943681716919,
      "learning_rate": 0.0005428819378547029,
      "loss": 2.3653,
      "step": 46007
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7381651401519775,
      "learning_rate": 0.0005428795367823034,
      "loss": 2.2866,
      "step": 46008
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.6608441472053528,
      "learning_rate": 0.0005428771356647481,
      "loss": 2.2936,
      "step": 46009
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.6550148725509644,
      "learning_rate": 0.0005428747345020371,
      "loss": 2.2746,
      "step": 46010
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7337337136268616,
      "learning_rate": 0.0005428723332941714,
      "loss": 2.4395,
      "step": 46011
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.6946761012077332,
      "learning_rate": 0.0005428699320411509,
      "loss": 2.2414,
      "step": 46012
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7348043322563171,
      "learning_rate": 0.0005428675307429765,
      "loss": 2.1726,
      "step": 46013
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7557495832443237,
      "learning_rate": 0.0005428651293996483,
      "loss": 2.4692,
      "step": 46014
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7330526113510132,
      "learning_rate": 0.0005428627280111669,
      "loss": 2.3244,
      "step": 46015
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.6792862415313721,
      "learning_rate": 0.0005428603265775329,
      "loss": 2.1179,
      "step": 46016
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7341019511222839,
      "learning_rate": 0.0005428579250987464,
      "loss": 2.1845,
      "step": 46017
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7186928987503052,
      "learning_rate": 0.000542855523574808,
      "loss": 2.2638,
      "step": 46018
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.8042271137237549,
      "learning_rate": 0.0005428531220057183,
      "loss": 2.3025,
      "step": 46019
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.6886171698570251,
      "learning_rate": 0.0005428507203914776,
      "loss": 2.3409,
      "step": 46020
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7669689655303955,
      "learning_rate": 0.0005428483187320864,
      "loss": 2.3994,
      "step": 46021
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7575465440750122,
      "learning_rate": 0.000542845917027545,
      "loss": 2.3898,
      "step": 46022
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.6807407140731812,
      "learning_rate": 0.000542843515277854,
      "loss": 2.3042,
      "step": 46023
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7469301819801331,
      "learning_rate": 0.0005428411134830137,
      "loss": 2.2407,
      "step": 46024
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.6877549886703491,
      "learning_rate": 0.0005428387116430249,
      "loss": 2.0973,
      "step": 46025
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7303130626678467,
      "learning_rate": 0.0005428363097578877,
      "loss": 2.3664,
      "step": 46026
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7575492262840271,
      "learning_rate": 0.0005428339078276026,
      "loss": 2.5851,
      "step": 46027
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.6633937358856201,
      "learning_rate": 0.0005428315058521701,
      "loss": 2.033,
      "step": 46028
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.677659809589386,
      "learning_rate": 0.0005428291038315907,
      "loss": 2.3992,
      "step": 46029
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.6764299273490906,
      "learning_rate": 0.0005428267017658647,
      "loss": 2.2682,
      "step": 46030
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.8381155729293823,
      "learning_rate": 0.0005428242996549927,
      "loss": 2.1067,
      "step": 46031
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7116925120353699,
      "learning_rate": 0.000542821897498975,
      "loss": 2.4344,
      "step": 46032
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7122374773025513,
      "learning_rate": 0.0005428194952978123,
      "loss": 2.0832,
      "step": 46033
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.6882476806640625,
      "learning_rate": 0.0005428170930515046,
      "loss": 2.2129,
      "step": 46034
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7486759424209595,
      "learning_rate": 0.0005428146907600528,
      "loss": 2.38,
      "step": 46035
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7638042569160461,
      "learning_rate": 0.0005428122884234571,
      "loss": 2.4521,
      "step": 46036
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.6844130754470825,
      "learning_rate": 0.0005428098860417181,
      "loss": 2.248,
      "step": 46037
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7100807428359985,
      "learning_rate": 0.0005428074836148361,
      "loss": 2.2958,
      "step": 46038
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7606484889984131,
      "learning_rate": 0.0005428050811428116,
      "loss": 2.4736,
      "step": 46039
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7834721207618713,
      "learning_rate": 0.000542802678625645,
      "loss": 2.0644,
      "step": 46040
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.665979266166687,
      "learning_rate": 0.0005428002760633369,
      "loss": 2.4976,
      "step": 46041
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7416830062866211,
      "learning_rate": 0.0005427978734558876,
      "loss": 2.1917,
      "step": 46042
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7155402898788452,
      "learning_rate": 0.0005427954708032976,
      "loss": 2.3332,
      "step": 46043
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.6764664053916931,
      "learning_rate": 0.0005427930681055673,
      "loss": 2.4336,
      "step": 46044
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.6944342851638794,
      "learning_rate": 0.0005427906653626973,
      "loss": 2.3375,
      "step": 46045
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7237893342971802,
      "learning_rate": 0.0005427882625746878,
      "loss": 2.1407,
      "step": 46046
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7011657953262329,
      "learning_rate": 0.0005427858597415394,
      "loss": 2.3288,
      "step": 46047
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.6748349666595459,
      "learning_rate": 0.0005427834568632525,
      "loss": 2.1759,
      "step": 46048
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7306126356124878,
      "learning_rate": 0.0005427810539398276,
      "loss": 2.1747,
      "step": 46049
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.6348070502281189,
      "learning_rate": 0.0005427786509712652,
      "loss": 2.2488,
      "step": 46050
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.704658567905426,
      "learning_rate": 0.0005427762479575656,
      "loss": 2.2431,
      "step": 46051
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.6927461624145508,
      "learning_rate": 0.0005427738448987294,
      "loss": 2.2731,
      "step": 46052
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.8076956868171692,
      "learning_rate": 0.0005427714417947569,
      "loss": 2.4902,
      "step": 46053
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7423313856124878,
      "learning_rate": 0.0005427690386456486,
      "loss": 2.1553,
      "step": 46054
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7967370748519897,
      "learning_rate": 0.0005427666354514049,
      "loss": 2.3327,
      "step": 46055
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.738592803478241,
      "learning_rate": 0.0005427642322120264,
      "loss": 2.3683,
      "step": 46056
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7692068815231323,
      "learning_rate": 0.0005427618289275133,
      "loss": 2.1088,
      "step": 46057
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.801712691783905,
      "learning_rate": 0.0005427594255978664,
      "loss": 2.2085,
      "step": 46058
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.690805971622467,
      "learning_rate": 0.0005427570222230857,
      "loss": 2.3151,
      "step": 46059
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.6993643045425415,
      "learning_rate": 0.000542754618803172,
      "loss": 2.4073,
      "step": 46060
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7979633808135986,
      "learning_rate": 0.0005427522153381258,
      "loss": 2.3526,
      "step": 46061
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7126687169075012,
      "learning_rate": 0.0005427498118279472,
      "loss": 2.1234,
      "step": 46062
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.6724451184272766,
      "learning_rate": 0.0005427474082726368,
      "loss": 2.296,
      "step": 46063
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7404060959815979,
      "learning_rate": 0.0005427450046721951,
      "loss": 2.2294,
      "step": 46064
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7199402451515198,
      "learning_rate": 0.0005427426010266226,
      "loss": 2.1966,
      "step": 46065
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.8006207346916199,
      "learning_rate": 0.0005427401973359196,
      "loss": 2.4796,
      "step": 46066
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.6099620461463928,
      "learning_rate": 0.0005427377936000866,
      "loss": 2.1168,
      "step": 46067
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7339810729026794,
      "learning_rate": 0.0005427353898191242,
      "loss": 2.2538,
      "step": 46068
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7161070108413696,
      "learning_rate": 0.0005427329859930325,
      "loss": 2.2998,
      "step": 46069
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7535717487335205,
      "learning_rate": 0.0005427305821218123,
      "loss": 2.2341,
      "step": 46070
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.6707137823104858,
      "learning_rate": 0.0005427281782054639,
      "loss": 2.0787,
      "step": 46071
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.6976972818374634,
      "learning_rate": 0.0005427257742439877,
      "loss": 2.2168,
      "step": 46072
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.6785927414894104,
      "learning_rate": 0.0005427233702373842,
      "loss": 2.2367,
      "step": 46073
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.6524147987365723,
      "learning_rate": 0.0005427209661856538,
      "loss": 2.3151,
      "step": 46074
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.6731507778167725,
      "learning_rate": 0.0005427185620887971,
      "loss": 2.0903,
      "step": 46075
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.6594619154930115,
      "learning_rate": 0.0005427161579468144,
      "loss": 2.267,
      "step": 46076
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7650724053382874,
      "learning_rate": 0.0005427137537597062,
      "loss": 2.3537,
      "step": 46077
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.6888459324836731,
      "learning_rate": 0.0005427113495274729,
      "loss": 2.1881,
      "step": 46078
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7218086123466492,
      "learning_rate": 0.000542708945250115,
      "loss": 2.4023,
      "step": 46079
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7939414978027344,
      "learning_rate": 0.0005427065409276329,
      "loss": 2.3545,
      "step": 46080
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.6595481634140015,
      "learning_rate": 0.000542704136560027,
      "loss": 2.2501,
      "step": 46081
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7565575242042542,
      "learning_rate": 0.0005427017321472979,
      "loss": 2.2697,
      "step": 46082
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.809822142124176,
      "learning_rate": 0.000542699327689446,
      "loss": 2.3665,
      "step": 46083
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.656831681728363,
      "learning_rate": 0.0005426969231864717,
      "loss": 2.0919,
      "step": 46084
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7648680806159973,
      "learning_rate": 0.0005426945186383754,
      "loss": 2.4411,
      "step": 46085
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7094770073890686,
      "learning_rate": 0.0005426921140451576,
      "loss": 2.4349,
      "step": 46086
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.6505691409111023,
      "learning_rate": 0.0005426897094068188,
      "loss": 2.2619,
      "step": 46087
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.6165012121200562,
      "learning_rate": 0.0005426873047233594,
      "loss": 2.1879,
      "step": 46088
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7107431292533875,
      "learning_rate": 0.00054268489999478,
      "loss": 2.2745,
      "step": 46089
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.6979294419288635,
      "learning_rate": 0.0005426824952210807,
      "loss": 2.4418,
      "step": 46090
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.6761119961738586,
      "learning_rate": 0.0005426800904022621,
      "loss": 2.0681,
      "step": 46091
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.6975985169410706,
      "learning_rate": 0.0005426776855383249,
      "loss": 2.1667,
      "step": 46092
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.6968285441398621,
      "learning_rate": 0.0005426752806292691,
      "loss": 2.4999,
      "step": 46093
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.6968562006950378,
      "learning_rate": 0.0005426728756750956,
      "loss": 2.3149,
      "step": 46094
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7594400644302368,
      "learning_rate": 0.0005426704706758045,
      "loss": 2.1977,
      "step": 46095
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7505611181259155,
      "learning_rate": 0.0005426680656313964,
      "loss": 2.1282,
      "step": 46096
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7026036381721497,
      "learning_rate": 0.0005426656605418718,
      "loss": 2.2062,
      "step": 46097
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7316515445709229,
      "learning_rate": 0.000542663255407231,
      "loss": 2.2958,
      "step": 46098
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7487376928329468,
      "learning_rate": 0.0005426608502274745,
      "loss": 2.1441,
      "step": 46099
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7572367191314697,
      "learning_rate": 0.0005426584450026028,
      "loss": 2.2441,
      "step": 46100
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.8275096416473389,
      "learning_rate": 0.0005426560397326163,
      "loss": 1.9973,
      "step": 46101
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7234947681427002,
      "learning_rate": 0.0005426536344175155,
      "loss": 2.2092,
      "step": 46102
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.6948674917221069,
      "learning_rate": 0.0005426512290573008,
      "loss": 2.1203,
      "step": 46103
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.6547001004219055,
      "learning_rate": 0.0005426488236519725,
      "loss": 2.2021,
      "step": 46104
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.6972776651382446,
      "learning_rate": 0.0005426464182015315,
      "loss": 2.3561,
      "step": 46105
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7758288383483887,
      "learning_rate": 0.0005426440127059778,
      "loss": 2.3076,
      "step": 46106
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7772716879844666,
      "learning_rate": 0.0005426416071653119,
      "loss": 2.223,
      "step": 46107
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.6756067872047424,
      "learning_rate": 0.0005426392015795345,
      "loss": 2.5819,
      "step": 46108
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.6698662638664246,
      "learning_rate": 0.0005426367959486457,
      "loss": 2.2405,
      "step": 46109
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.6834144592285156,
      "learning_rate": 0.0005426343902726464,
      "loss": 2.0365,
      "step": 46110
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7328171730041504,
      "learning_rate": 0.0005426319845515367,
      "loss": 2.2922,
      "step": 46111
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7165507078170776,
      "learning_rate": 0.0005426295787853171,
      "loss": 2.1064,
      "step": 46112
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7210580110549927,
      "learning_rate": 0.000542627172973988,
      "loss": 2.4517,
      "step": 46113
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.8650026917457581,
      "learning_rate": 0.0005426247671175499,
      "loss": 2.1639,
      "step": 46114
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.6140697002410889,
      "learning_rate": 0.0005426223612160035,
      "loss": 2.2291,
      "step": 46115
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.83859783411026,
      "learning_rate": 0.0005426199552693488,
      "loss": 2.3461,
      "step": 46116
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.8020689487457275,
      "learning_rate": 0.0005426175492775866,
      "loss": 2.1415,
      "step": 46117
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.6758301854133606,
      "learning_rate": 0.0005426151432407172,
      "loss": 2.3164,
      "step": 46118
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7039169073104858,
      "learning_rate": 0.0005426127371587411,
      "loss": 2.1751,
      "step": 46119
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.6951965093612671,
      "learning_rate": 0.0005426103310316587,
      "loss": 2.2289,
      "step": 46120
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7280144095420837,
      "learning_rate": 0.0005426079248594704,
      "loss": 2.0893,
      "step": 46121
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7078955173492432,
      "learning_rate": 0.0005426055186421767,
      "loss": 2.3621,
      "step": 46122
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7007211446762085,
      "learning_rate": 0.0005426031123797781,
      "loss": 2.2434,
      "step": 46123
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.6702489852905273,
      "learning_rate": 0.000542600706072275,
      "loss": 2.3109,
      "step": 46124
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7528963088989258,
      "learning_rate": 0.0005425982997196679,
      "loss": 2.1818,
      "step": 46125
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7863224744796753,
      "learning_rate": 0.0005425958933219571,
      "loss": 2.352,
      "step": 46126
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.846878170967102,
      "learning_rate": 0.0005425934868791431,
      "loss": 2.2576,
      "step": 46127
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.6976178288459778,
      "learning_rate": 0.0005425910803912265,
      "loss": 2.3717,
      "step": 46128
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7980198264122009,
      "learning_rate": 0.0005425886738582076,
      "loss": 2.1861,
      "step": 46129
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.6600844264030457,
      "learning_rate": 0.0005425862672800869,
      "loss": 2.2542,
      "step": 46130
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.8192774653434753,
      "learning_rate": 0.0005425838606568648,
      "loss": 2.293,
      "step": 46131
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.8154765963554382,
      "learning_rate": 0.0005425814539885418,
      "loss": 2.3121,
      "step": 46132
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7622020840644836,
      "learning_rate": 0.0005425790472751183,
      "loss": 2.2742,
      "step": 46133
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.784411609172821,
      "learning_rate": 0.0005425766405165949,
      "loss": 2.1977,
      "step": 46134
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7205573320388794,
      "learning_rate": 0.0005425742337129719,
      "loss": 2.5463,
      "step": 46135
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7350767254829407,
      "learning_rate": 0.0005425718268642496,
      "loss": 2.2456,
      "step": 46136
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7035632133483887,
      "learning_rate": 0.0005425694199704288,
      "loss": 2.0512,
      "step": 46137
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.6648561954498291,
      "learning_rate": 0.0005425670130315097,
      "loss": 2.2449,
      "step": 46138
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.6469144821166992,
      "learning_rate": 0.0005425646060474927,
      "loss": 2.3715,
      "step": 46139
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7142561078071594,
      "learning_rate": 0.0005425621990183785,
      "loss": 2.491,
      "step": 46140
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7083126902580261,
      "learning_rate": 0.0005425597919441674,
      "loss": 2.2483,
      "step": 46141
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.6493303775787354,
      "learning_rate": 0.0005425573848248598,
      "loss": 2.1582,
      "step": 46142
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7215946316719055,
      "learning_rate": 0.0005425549776604563,
      "loss": 2.0966,
      "step": 46143
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.6635704040527344,
      "learning_rate": 0.0005425525704509573,
      "loss": 2.3533,
      "step": 46144
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.6927452683448792,
      "learning_rate": 0.0005425501631963629,
      "loss": 2.041,
      "step": 46145
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.6543493866920471,
      "learning_rate": 0.0005425477558966741,
      "loss": 2.2173,
      "step": 46146
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7454565763473511,
      "learning_rate": 0.000542545348551891,
      "loss": 2.1483,
      "step": 46147
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7697619795799255,
      "learning_rate": 0.0005425429411620142,
      "loss": 2.2777,
      "step": 46148
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.6891181468963623,
      "learning_rate": 0.000542540533727044,
      "loss": 2.0974,
      "step": 46149
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.8078967332839966,
      "learning_rate": 0.000542538126246981,
      "loss": 2.3561,
      "step": 46150
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7815735936164856,
      "learning_rate": 0.0005425357187218257,
      "loss": 2.2252,
      "step": 46151
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.8910087943077087,
      "learning_rate": 0.0005425333111515782,
      "loss": 2.3292,
      "step": 46152
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7134448885917664,
      "learning_rate": 0.0005425309035362394,
      "loss": 2.1371,
      "step": 46153
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.6958548426628113,
      "learning_rate": 0.0005425284958758093,
      "loss": 2.2248,
      "step": 46154
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7154803276062012,
      "learning_rate": 0.0005425260881702888,
      "loss": 2.4162,
      "step": 46155
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.6578805446624756,
      "learning_rate": 0.000542523680419678,
      "loss": 2.2219,
      "step": 46156
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7656646370887756,
      "learning_rate": 0.0005425212726239774,
      "loss": 2.2483,
      "step": 46157
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.8035087585449219,
      "learning_rate": 0.0005425188647831876,
      "loss": 2.3707,
      "step": 46158
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7245272994041443,
      "learning_rate": 0.000542516456897309,
      "loss": 2.3516,
      "step": 46159
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.6369174718856812,
      "learning_rate": 0.000542514048966342,
      "loss": 2.247,
      "step": 46160
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.700455904006958,
      "learning_rate": 0.000542511640990287,
      "loss": 2.1241,
      "step": 46161
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7901318669319153,
      "learning_rate": 0.0005425092329691446,
      "loss": 2.4313,
      "step": 46162
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7497623562812805,
      "learning_rate": 0.0005425068249029151,
      "loss": 2.1925,
      "step": 46163
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7439061999320984,
      "learning_rate": 0.0005425044167915989,
      "loss": 2.2432,
      "step": 46164
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.6770280599594116,
      "learning_rate": 0.0005425020086351968,
      "loss": 2.3368,
      "step": 46165
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7592806220054626,
      "learning_rate": 0.0005424996004337087,
      "loss": 2.2309,
      "step": 46166
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7513667345046997,
      "learning_rate": 0.0005424971921871355,
      "loss": 2.0096,
      "step": 46167
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7680751085281372,
      "learning_rate": 0.0005424947838954775,
      "loss": 2.1796,
      "step": 46168
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.8852894902229309,
      "learning_rate": 0.0005424923755587353,
      "loss": 2.6161,
      "step": 46169
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.8400829434394836,
      "learning_rate": 0.0005424899671769089,
      "loss": 2.31,
      "step": 46170
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.6700804233551025,
      "learning_rate": 0.0005424875587499992,
      "loss": 2.2105,
      "step": 46171
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.8741535544395447,
      "learning_rate": 0.0005424851502780065,
      "loss": 2.3375,
      "step": 46172
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.698840856552124,
      "learning_rate": 0.0005424827417609312,
      "loss": 2.3254,
      "step": 46173
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7607682347297668,
      "learning_rate": 0.0005424803331987737,
      "loss": 2.0272,
      "step": 46174
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.725406289100647,
      "learning_rate": 0.0005424779245915345,
      "loss": 2.2218,
      "step": 46175
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7360382676124573,
      "learning_rate": 0.0005424755159392143,
      "loss": 2.4694,
      "step": 46176
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7001734375953674,
      "learning_rate": 0.0005424731072418131,
      "loss": 2.1181,
      "step": 46177
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7926666140556335,
      "learning_rate": 0.0005424706984993317,
      "loss": 2.252,
      "step": 46178
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.647728681564331,
      "learning_rate": 0.0005424682897117704,
      "loss": 2.0173,
      "step": 46179
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.6633938550949097,
      "learning_rate": 0.0005424658808791297,
      "loss": 2.0491,
      "step": 46180
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.6996508240699768,
      "learning_rate": 0.0005424634720014099,
      "loss": 2.3341,
      "step": 46181
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7316819429397583,
      "learning_rate": 0.0005424610630786117,
      "loss": 2.2787,
      "step": 46182
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7062531113624573,
      "learning_rate": 0.0005424586541107353,
      "loss": 2.2659,
      "step": 46183
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.675642728805542,
      "learning_rate": 0.0005424562450977813,
      "loss": 2.0038,
      "step": 46184
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.6598694920539856,
      "learning_rate": 0.0005424538360397503,
      "loss": 2.0302,
      "step": 46185
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.8264822959899902,
      "learning_rate": 0.0005424514269366422,
      "loss": 2.3346,
      "step": 46186
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7286071181297302,
      "learning_rate": 0.000542449017788458,
      "loss": 2.36,
      "step": 46187
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.757447361946106,
      "learning_rate": 0.0005424466085951979,
      "loss": 2.1608,
      "step": 46188
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.6987543702125549,
      "learning_rate": 0.0005424441993568624,
      "loss": 2.2174,
      "step": 46189
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.6323418021202087,
      "learning_rate": 0.000542441790073452,
      "loss": 2.1763,
      "step": 46190
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7275710701942444,
      "learning_rate": 0.000542439380744967,
      "loss": 2.3079,
      "step": 46191
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.6771353483200073,
      "learning_rate": 0.0005424369713714081,
      "loss": 2.3785,
      "step": 46192
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.6845318078994751,
      "learning_rate": 0.0005424345619527755,
      "loss": 2.2507,
      "step": 46193
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.6981039047241211,
      "learning_rate": 0.0005424321524890698,
      "loss": 2.4112,
      "step": 46194
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7410791516304016,
      "learning_rate": 0.0005424297429802912,
      "loss": 2.0473,
      "step": 46195
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.6854482293128967,
      "learning_rate": 0.0005424273334264405,
      "loss": 2.4343,
      "step": 46196
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.6455633640289307,
      "learning_rate": 0.000542424923827518,
      "loss": 2.2308,
      "step": 46197
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7235878109931946,
      "learning_rate": 0.0005424225141835241,
      "loss": 2.2434,
      "step": 46198
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.695540726184845,
      "learning_rate": 0.0005424201044944591,
      "loss": 2.2356,
      "step": 46199
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7229272127151489,
      "learning_rate": 0.0005424176947603238,
      "loss": 2.3068,
      "step": 46200
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7595946788787842,
      "learning_rate": 0.0005424152849811185,
      "loss": 2.2306,
      "step": 46201
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.8129263520240784,
      "learning_rate": 0.0005424128751568436,
      "loss": 2.2432,
      "step": 46202
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7300429344177246,
      "learning_rate": 0.0005424104652874996,
      "loss": 2.4818,
      "step": 46203
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7209764719009399,
      "learning_rate": 0.0005424080553730867,
      "loss": 2.109,
      "step": 46204
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7099881768226624,
      "learning_rate": 0.0005424056454136058,
      "loss": 2.0091,
      "step": 46205
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7232478857040405,
      "learning_rate": 0.000542403235409057,
      "loss": 2.1644,
      "step": 46206
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7556726336479187,
      "learning_rate": 0.000542400825359441,
      "loss": 2.2607,
      "step": 46207
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7191267013549805,
      "learning_rate": 0.000542398415264758,
      "loss": 2.3436,
      "step": 46208
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.708886444568634,
      "learning_rate": 0.0005423960051250086,
      "loss": 2.2012,
      "step": 46209
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7792587876319885,
      "learning_rate": 0.0005423935949401931,
      "loss": 2.4496,
      "step": 46210
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.6889855861663818,
      "learning_rate": 0.0005423911847103122,
      "loss": 2.1427,
      "step": 46211
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.690700352191925,
      "learning_rate": 0.0005423887744353661,
      "loss": 2.0799,
      "step": 46212
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.685045599937439,
      "learning_rate": 0.0005423863641153554,
      "loss": 2.2494,
      "step": 46213
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7464091777801514,
      "learning_rate": 0.0005423839537502806,
      "loss": 1.9572,
      "step": 46214
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7115891575813293,
      "learning_rate": 0.0005423815433401419,
      "loss": 2.5051,
      "step": 46215
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.710728645324707,
      "learning_rate": 0.0005423791328849399,
      "loss": 2.3536,
      "step": 46216
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7836296558380127,
      "learning_rate": 0.000542376722384675,
      "loss": 2.3505,
      "step": 46217
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.719552218914032,
      "learning_rate": 0.0005423743118393478,
      "loss": 2.2503,
      "step": 46218
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.719399094581604,
      "learning_rate": 0.0005423719012489586,
      "loss": 2.0649,
      "step": 46219
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7287847399711609,
      "learning_rate": 0.000542369490613508,
      "loss": 2.3233,
      "step": 46220
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.6592283844947815,
      "learning_rate": 0.0005423670799329962,
      "loss": 2.158,
      "step": 46221
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7308275103569031,
      "learning_rate": 0.0005423646692074238,
      "loss": 2.0875,
      "step": 46222
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.8088465929031372,
      "learning_rate": 0.0005423622584367913,
      "loss": 2.1485,
      "step": 46223
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.6268547773361206,
      "learning_rate": 0.000542359847621099,
      "loss": 2.3377,
      "step": 46224
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7577593326568604,
      "learning_rate": 0.0005423574367603475,
      "loss": 2.556,
      "step": 46225
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7388266921043396,
      "learning_rate": 0.000542355025854537,
      "loss": 2.2683,
      "step": 46226
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.8553369641304016,
      "learning_rate": 0.0005423526149036684,
      "loss": 2.422,
      "step": 46227
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7306105494499207,
      "learning_rate": 0.0005423502039077417,
      "loss": 2.0976,
      "step": 46228
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.901525616645813,
      "learning_rate": 0.0005423477928667576,
      "loss": 2.5293,
      "step": 46229
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7123805284500122,
      "learning_rate": 0.0005423453817807165,
      "loss": 2.3006,
      "step": 46230
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.915679931640625,
      "learning_rate": 0.0005423429706496188,
      "loss": 2.078,
      "step": 46231
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.6997784376144409,
      "learning_rate": 0.0005423405594734648,
      "loss": 2.2078,
      "step": 46232
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7277599573135376,
      "learning_rate": 0.0005423381482522553,
      "loss": 2.214,
      "step": 46233
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7153210043907166,
      "learning_rate": 0.0005423357369859905,
      "loss": 2.4319,
      "step": 46234
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7386265993118286,
      "learning_rate": 0.0005423333256746708,
      "loss": 2.4469,
      "step": 46235
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7190628051757812,
      "learning_rate": 0.0005423309143182969,
      "loss": 2.3605,
      "step": 46236
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.6948771476745605,
      "learning_rate": 0.0005423285029168692,
      "loss": 2.3107,
      "step": 46237
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7060410380363464,
      "learning_rate": 0.0005423260914703879,
      "loss": 2.2157,
      "step": 46238
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.6675916910171509,
      "learning_rate": 0.0005423236799788537,
      "loss": 2.2567,
      "step": 46239
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7257641553878784,
      "learning_rate": 0.0005423212684422669,
      "loss": 2.507,
      "step": 46240
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.689712643623352,
      "learning_rate": 0.000542318856860628,
      "loss": 2.2842,
      "step": 46241
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.8817535638809204,
      "learning_rate": 0.0005423164452339373,
      "loss": 2.1979,
      "step": 46242
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.6733435988426208,
      "learning_rate": 0.0005423140335621957,
      "loss": 2.1228,
      "step": 46243
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.8204017281532288,
      "learning_rate": 0.0005423116218454032,
      "loss": 2.5149,
      "step": 46244
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7211716175079346,
      "learning_rate": 0.0005423092100835603,
      "loss": 2.2494,
      "step": 46245
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7316737771034241,
      "learning_rate": 0.0005423067982766676,
      "loss": 2.1207,
      "step": 46246
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.6544981598854065,
      "learning_rate": 0.0005423043864247255,
      "loss": 2.1531,
      "step": 46247
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.8711000680923462,
      "learning_rate": 0.0005423019745277346,
      "loss": 2.1934,
      "step": 46248
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.6913752555847168,
      "learning_rate": 0.000542299562585695,
      "loss": 2.1897,
      "step": 46249
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7100590467453003,
      "learning_rate": 0.0005422971505986074,
      "loss": 2.4704,
      "step": 46250
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.8695182204246521,
      "learning_rate": 0.0005422947385664722,
      "loss": 2.3677,
      "step": 46251
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.6869609355926514,
      "learning_rate": 0.0005422923264892899,
      "loss": 2.1816,
      "step": 46252
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7529831528663635,
      "learning_rate": 0.0005422899143670607,
      "loss": 2.147,
      "step": 46253
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.771038830280304,
      "learning_rate": 0.0005422875021997854,
      "loss": 2.4543,
      "step": 46254
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.765636682510376,
      "learning_rate": 0.0005422850899874642,
      "loss": 2.6778,
      "step": 46255
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7667096853256226,
      "learning_rate": 0.0005422826777300976,
      "loss": 2.4485,
      "step": 46256
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7291097044944763,
      "learning_rate": 0.0005422802654276861,
      "loss": 2.2291,
      "step": 46257
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.9201961755752563,
      "learning_rate": 0.0005422778530802301,
      "loss": 2.166,
      "step": 46258
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7504900097846985,
      "learning_rate": 0.0005422754406877302,
      "loss": 2.4986,
      "step": 46259
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.6967511177062988,
      "learning_rate": 0.0005422730282501866,
      "loss": 2.3484,
      "step": 46260
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.6440110802650452,
      "learning_rate": 0.0005422706157675999,
      "loss": 2.2675,
      "step": 46261
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.8590155839920044,
      "learning_rate": 0.0005422682032399706,
      "loss": 2.3376,
      "step": 46262
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7349302768707275,
      "learning_rate": 0.0005422657906672989,
      "loss": 2.1873,
      "step": 46263
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.8001108765602112,
      "learning_rate": 0.0005422633780495855,
      "loss": 2.2323,
      "step": 46264
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.726201593875885,
      "learning_rate": 0.0005422609653868308,
      "loss": 2.2197,
      "step": 46265
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7121500968933105,
      "learning_rate": 0.0005422585526790352,
      "loss": 2.4251,
      "step": 46266
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7576766014099121,
      "learning_rate": 0.0005422561399261992,
      "loss": 2.2468,
      "step": 46267
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.6712670922279358,
      "learning_rate": 0.0005422537271283232,
      "loss": 2.3057,
      "step": 46268
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.761281430721283,
      "learning_rate": 0.0005422513142854076,
      "loss": 2.3047,
      "step": 46269
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.9376311302185059,
      "learning_rate": 0.0005422489013974528,
      "loss": 2.1265,
      "step": 46270
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.6563177704811096,
      "learning_rate": 0.0005422464884644596,
      "loss": 2.1443,
      "step": 46271
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7974942922592163,
      "learning_rate": 0.0005422440754864281,
      "loss": 2.2939,
      "step": 46272
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.6947484016418457,
      "learning_rate": 0.0005422416624633587,
      "loss": 2.3051,
      "step": 46273
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.6725686192512512,
      "learning_rate": 0.0005422392493952522,
      "loss": 2.336,
      "step": 46274
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.662220299243927,
      "learning_rate": 0.0005422368362821089,
      "loss": 2.4092,
      "step": 46275
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.6646396517753601,
      "learning_rate": 0.0005422344231239291,
      "loss": 2.2002,
      "step": 46276
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7100399136543274,
      "learning_rate": 0.0005422320099207134,
      "loss": 2.4996,
      "step": 46277
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.669403076171875,
      "learning_rate": 0.0005422295966724621,
      "loss": 2.1217,
      "step": 46278
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.6535384654998779,
      "learning_rate": 0.0005422271833791757,
      "loss": 2.2883,
      "step": 46279
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.6706488132476807,
      "learning_rate": 0.0005422247700408548,
      "loss": 2.309,
      "step": 46280
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.6704768538475037,
      "learning_rate": 0.0005422223566574997,
      "loss": 2.1621,
      "step": 46281
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7330270409584045,
      "learning_rate": 0.000542219943229111,
      "loss": 2.1138,
      "step": 46282
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.6973379254341125,
      "learning_rate": 0.000542217529755689,
      "loss": 1.9906,
      "step": 46283
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.6986653208732605,
      "learning_rate": 0.0005422151162372341,
      "loss": 2.1037,
      "step": 46284
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7265270352363586,
      "learning_rate": 0.0005422127026737468,
      "loss": 2.274,
      "step": 46285
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7047773599624634,
      "learning_rate": 0.0005422102890652277,
      "loss": 2.5928,
      "step": 46286
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.6578201055526733,
      "learning_rate": 0.0005422078754116771,
      "loss": 2.2205,
      "step": 46287
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.6850128769874573,
      "learning_rate": 0.0005422054617130955,
      "loss": 2.3879,
      "step": 46288
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.704818844795227,
      "learning_rate": 0.0005422030479694833,
      "loss": 2.3105,
      "step": 46289
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7462974190711975,
      "learning_rate": 0.0005422006341808411,
      "loss": 2.4138,
      "step": 46290
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.6603686213493347,
      "learning_rate": 0.000542198220347169,
      "loss": 2.2395,
      "step": 46291
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.6876329779624939,
      "learning_rate": 0.0005421958064684679,
      "loss": 2.3308,
      "step": 46292
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7650659084320068,
      "learning_rate": 0.0005421933925447379,
      "loss": 2.0501,
      "step": 46293
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7537336945533752,
      "learning_rate": 0.0005421909785759796,
      "loss": 2.077,
      "step": 46294
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7191691398620605,
      "learning_rate": 0.0005421885645621935,
      "loss": 2.2473,
      "step": 46295
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.6433762907981873,
      "learning_rate": 0.0005421861505033799,
      "loss": 2.2323,
      "step": 46296
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.6685315370559692,
      "learning_rate": 0.0005421837363995393,
      "loss": 2.3367,
      "step": 46297
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7205055952072144,
      "learning_rate": 0.0005421813222506723,
      "loss": 2.4367,
      "step": 46298
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.6504010558128357,
      "learning_rate": 0.0005421789080567791,
      "loss": 2.2639,
      "step": 46299
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7570973634719849,
      "learning_rate": 0.0005421764938178603,
      "loss": 2.2291,
      "step": 46300
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.683441698551178,
      "learning_rate": 0.0005421740795339163,
      "loss": 2.1488,
      "step": 46301
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.6579183340072632,
      "learning_rate": 0.0005421716652049476,
      "loss": 2.5037,
      "step": 46302
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7052640318870544,
      "learning_rate": 0.0005421692508309546,
      "loss": 2.4368,
      "step": 46303
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7328047752380371,
      "learning_rate": 0.0005421668364119378,
      "loss": 2.1617,
      "step": 46304
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.6730205416679382,
      "learning_rate": 0.0005421644219478976,
      "loss": 2.1669,
      "step": 46305
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7044074535369873,
      "learning_rate": 0.0005421620074388344,
      "loss": 2.3933,
      "step": 46306
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7681037783622742,
      "learning_rate": 0.0005421595928847487,
      "loss": 2.4135,
      "step": 46307
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.6601420044898987,
      "learning_rate": 0.0005421571782856411,
      "loss": 2.2886,
      "step": 46308
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.750004231929779,
      "learning_rate": 0.0005421547636415118,
      "loss": 2.3451,
      "step": 46309
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.137792706489563,
      "learning_rate": 0.0005421523489523614,
      "loss": 2.3339,
      "step": 46310
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.9058400988578796,
      "learning_rate": 0.0005421499342181902,
      "loss": 2.2712,
      "step": 46311
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7645283341407776,
      "learning_rate": 0.000542147519438999,
      "loss": 2.3393,
      "step": 46312
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.0936551094055176,
      "learning_rate": 0.0005421451046147878,
      "loss": 2.2303,
      "step": 46313
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7422693371772766,
      "learning_rate": 0.0005421426897455572,
      "loss": 2.2299,
      "step": 46314
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7247369289398193,
      "learning_rate": 0.0005421402748313079,
      "loss": 2.1525,
      "step": 46315
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.0188456773757935,
      "learning_rate": 0.00054213785987204,
      "loss": 2.2849,
      "step": 46316
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7143513560295105,
      "learning_rate": 0.0005421354448677543,
      "loss": 2.1643,
      "step": 46317
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7645766735076904,
      "learning_rate": 0.0005421330298184509,
      "loss": 2.4046,
      "step": 46318
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.6627634167671204,
      "learning_rate": 0.0005421306147241304,
      "loss": 2.2158,
      "step": 46319
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.8056313991546631,
      "learning_rate": 0.0005421281995847933,
      "loss": 2.3242,
      "step": 46320
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7370327115058899,
      "learning_rate": 0.0005421257844004399,
      "loss": 2.3866,
      "step": 46321
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.6976856589317322,
      "learning_rate": 0.0005421233691710708,
      "loss": 2.4015,
      "step": 46322
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.8484482765197754,
      "learning_rate": 0.0005421209538966864,
      "loss": 2.41,
      "step": 46323
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.8569095134735107,
      "learning_rate": 0.0005421185385772871,
      "loss": 2.2449,
      "step": 46324
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7165634632110596,
      "learning_rate": 0.0005421161232128734,
      "loss": 2.1819,
      "step": 46325
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.723975658416748,
      "learning_rate": 0.0005421137078034458,
      "loss": 2.2616,
      "step": 46326
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.1551947593688965,
      "learning_rate": 0.0005421112923490048,
      "loss": 2.1196,
      "step": 46327
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7317739725112915,
      "learning_rate": 0.0005421088768495506,
      "loss": 2.2743,
      "step": 46328
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.6976521611213684,
      "learning_rate": 0.0005421064613050837,
      "loss": 2.172,
      "step": 46329
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.8117530941963196,
      "learning_rate": 0.0005421040457156048,
      "loss": 2.3287,
      "step": 46330
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7366952896118164,
      "learning_rate": 0.0005421016300811141,
      "loss": 2.1432,
      "step": 46331
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7846903204917908,
      "learning_rate": 0.0005420992144016122,
      "loss": 2.3893,
      "step": 46332
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.9172620177268982,
      "learning_rate": 0.0005420967986770994,
      "loss": 2.0446,
      "step": 46333
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.6388676166534424,
      "learning_rate": 0.0005420943829075763,
      "loss": 2.2789,
      "step": 46334
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.6848170757293701,
      "learning_rate": 0.0005420919670930433,
      "loss": 2.3031,
      "step": 46335
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7784958481788635,
      "learning_rate": 0.0005420895512335008,
      "loss": 2.0069,
      "step": 46336
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7492499351501465,
      "learning_rate": 0.0005420871353289493,
      "loss": 2.2784,
      "step": 46337
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.6791442632675171,
      "learning_rate": 0.0005420847193793893,
      "loss": 2.0439,
      "step": 46338
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.823836088180542,
      "learning_rate": 0.0005420823033848211,
      "loss": 2.3919,
      "step": 46339
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7524839639663696,
      "learning_rate": 0.0005420798873452451,
      "loss": 2.1852,
      "step": 46340
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7422601580619812,
      "learning_rate": 0.0005420774712606621,
      "loss": 2.2349,
      "step": 46341
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7028920650482178,
      "learning_rate": 0.0005420750551310722,
      "loss": 2.3576,
      "step": 46342
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.6662891507148743,
      "learning_rate": 0.000542072638956476,
      "loss": 2.2555,
      "step": 46343
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7252817153930664,
      "learning_rate": 0.000542070222736874,
      "loss": 2.3049,
      "step": 46344
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7253223657608032,
      "learning_rate": 0.0005420678064722665,
      "loss": 2.228,
      "step": 46345
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.6597030758857727,
      "learning_rate": 0.000542065390162654,
      "loss": 2.3152,
      "step": 46346
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.679490327835083,
      "learning_rate": 0.0005420629738080371,
      "loss": 2.217,
      "step": 46347
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.6583091020584106,
      "learning_rate": 0.000542060557408416,
      "loss": 2.2516,
      "step": 46348
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.6950977444648743,
      "learning_rate": 0.0005420581409637914,
      "loss": 2.5555,
      "step": 46349
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7370688915252686,
      "learning_rate": 0.0005420557244741634,
      "loss": 2.1903,
      "step": 46350
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7142402529716492,
      "learning_rate": 0.0005420533079395328,
      "loss": 2.312,
      "step": 46351
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.75340336561203,
      "learning_rate": 0.0005420508913599,
      "loss": 2.0384,
      "step": 46352
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7074925899505615,
      "learning_rate": 0.0005420484747352653,
      "loss": 2.342,
      "step": 46353
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7073105573654175,
      "learning_rate": 0.0005420460580656293,
      "loss": 2.379,
      "step": 46354
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.691425085067749,
      "learning_rate": 0.0005420436413509923,
      "loss": 2.3361,
      "step": 46355
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.709877073764801,
      "learning_rate": 0.0005420412245913547,
      "loss": 2.2337,
      "step": 46356
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7306937575340271,
      "learning_rate": 0.0005420388077867172,
      "loss": 2.0696,
      "step": 46357
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.6823617815971375,
      "learning_rate": 0.0005420363909370801,
      "loss": 2.6193,
      "step": 46358
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.6557251214981079,
      "learning_rate": 0.0005420339740424438,
      "loss": 2.3385,
      "step": 46359
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.6800428628921509,
      "learning_rate": 0.0005420315571028089,
      "loss": 2.3882,
      "step": 46360
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.6863123774528503,
      "learning_rate": 0.0005420291401181757,
      "loss": 2.1543,
      "step": 46361
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7388787865638733,
      "learning_rate": 0.0005420267230885448,
      "loss": 2.4153,
      "step": 46362
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.6539710164070129,
      "learning_rate": 0.0005420243060139165,
      "loss": 2.1691,
      "step": 46363
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7031132578849792,
      "learning_rate": 0.0005420218888942914,
      "loss": 2.3151,
      "step": 46364
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.6668642163276672,
      "learning_rate": 0.0005420194717296698,
      "loss": 2.3524,
      "step": 46365
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.620206356048584,
      "learning_rate": 0.0005420170545200521,
      "loss": 2.2498,
      "step": 46366
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.6369227766990662,
      "learning_rate": 0.0005420146372654389,
      "loss": 2.1477,
      "step": 46367
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.6785948872566223,
      "learning_rate": 0.0005420122199658307,
      "loss": 2.1389,
      "step": 46368
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.6773160099983215,
      "learning_rate": 0.0005420098026212279,
      "loss": 2.2808,
      "step": 46369
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.6642740368843079,
      "learning_rate": 0.0005420073852316308,
      "loss": 2.3152,
      "step": 46370
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.6650174856185913,
      "learning_rate": 0.0005420049677970401,
      "loss": 2.3833,
      "step": 46371
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.708594799041748,
      "learning_rate": 0.000542002550317456,
      "loss": 2.1813,
      "step": 46372
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.6942468881607056,
      "learning_rate": 0.0005420001327928791,
      "loss": 1.978,
      "step": 46373
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7649774551391602,
      "learning_rate": 0.0005419977152233097,
      "loss": 2.3365,
      "step": 46374
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7526630759239197,
      "learning_rate": 0.0005419952976087485,
      "loss": 2.4001,
      "step": 46375
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.9897559285163879,
      "learning_rate": 0.0005419928799491957,
      "loss": 2.1073,
      "step": 46376
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7477601766586304,
      "learning_rate": 0.0005419904622446519,
      "loss": 2.3186,
      "step": 46377
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.6748821139335632,
      "learning_rate": 0.0005419880444951175,
      "loss": 2.4095,
      "step": 46378
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.6898544430732727,
      "learning_rate": 0.000541985626700593,
      "loss": 2.4366,
      "step": 46379
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7266316413879395,
      "learning_rate": 0.0005419832088610787,
      "loss": 2.1953,
      "step": 46380
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7428990602493286,
      "learning_rate": 0.0005419807909765752,
      "loss": 2.5062,
      "step": 46381
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.6137309670448303,
      "learning_rate": 0.000541978373047083,
      "loss": 2.2432,
      "step": 46382
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.9323803782463074,
      "learning_rate": 0.0005419759550726024,
      "loss": 2.1136,
      "step": 46383
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7161407470703125,
      "learning_rate": 0.0005419735370531338,
      "loss": 2.5116,
      "step": 46384
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7099364995956421,
      "learning_rate": 0.0005419711189886779,
      "loss": 2.2229,
      "step": 46385
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7964617609977722,
      "learning_rate": 0.000541968700879235,
      "loss": 2.3055,
      "step": 46386
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7312468886375427,
      "learning_rate": 0.0005419662827248053,
      "loss": 2.0846,
      "step": 46387
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7230038046836853,
      "learning_rate": 0.0005419638645253897,
      "loss": 2.2741,
      "step": 46388
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.6971989274024963,
      "learning_rate": 0.0005419614462809885,
      "loss": 2.1566,
      "step": 46389
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7239482402801514,
      "learning_rate": 0.0005419590279916019,
      "loss": 2.0659,
      "step": 46390
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7478430271148682,
      "learning_rate": 0.0005419566096572307,
      "loss": 2.2404,
      "step": 46391
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.6342710256576538,
      "learning_rate": 0.0005419541912778751,
      "loss": 2.2051,
      "step": 46392
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7267043590545654,
      "learning_rate": 0.0005419517728535359,
      "loss": 2.3464,
      "step": 46393
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7071566581726074,
      "learning_rate": 0.0005419493543842131,
      "loss": 2.1976,
      "step": 46394
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7001959681510925,
      "learning_rate": 0.0005419469358699073,
      "loss": 2.2079,
      "step": 46395
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.8359093070030212,
      "learning_rate": 0.0005419445173106191,
      "loss": 2.1573,
      "step": 46396
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.6537637114524841,
      "learning_rate": 0.0005419420987063488,
      "loss": 2.1978,
      "step": 46397
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.6789496541023254,
      "learning_rate": 0.0005419396800570969,
      "loss": 2.1661,
      "step": 46398
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7293165922164917,
      "learning_rate": 0.0005419372613628639,
      "loss": 2.5052,
      "step": 46399
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.928034782409668,
      "learning_rate": 0.0005419348426236501,
      "loss": 2.4669,
      "step": 46400
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.8266295194625854,
      "learning_rate": 0.0005419324238394561,
      "loss": 2.19,
      "step": 46401
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7099931836128235,
      "learning_rate": 0.0005419300050102824,
      "loss": 2.1766,
      "step": 46402
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.6902527809143066,
      "learning_rate": 0.0005419275861361292,
      "loss": 2.1429,
      "step": 46403
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7383880019187927,
      "learning_rate": 0.0005419251672169972,
      "loss": 2.3886,
      "step": 46404
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7298640012741089,
      "learning_rate": 0.0005419227482528865,
      "loss": 2.3633,
      "step": 46405
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7206423282623291,
      "learning_rate": 0.000541920329243798,
      "loss": 2.2729,
      "step": 46406
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7859734892845154,
      "learning_rate": 0.0005419179101897319,
      "loss": 2.2196,
      "step": 46407
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7301807999610901,
      "learning_rate": 0.0005419154910906889,
      "loss": 2.1002,
      "step": 46408
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7186474800109863,
      "learning_rate": 0.0005419130719466689,
      "loss": 2.2713,
      "step": 46409
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.8675461411476135,
      "learning_rate": 0.0005419106527576728,
      "loss": 2.1634,
      "step": 46410
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7022513747215271,
      "learning_rate": 0.000541908233523701,
      "loss": 2.3877,
      "step": 46411
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7128016948699951,
      "learning_rate": 0.0005419058142447541,
      "loss": 2.2548,
      "step": 46412
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7821671962738037,
      "learning_rate": 0.000541903394920832,
      "loss": 2.2577,
      "step": 46413
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7606220245361328,
      "learning_rate": 0.0005419009755519356,
      "loss": 2.2605,
      "step": 46414
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.664462685585022,
      "learning_rate": 0.0005418985561380653,
      "loss": 2.4183,
      "step": 46415
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.6984138488769531,
      "learning_rate": 0.0005418961366792215,
      "loss": 2.2946,
      "step": 46416
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7540326714515686,
      "learning_rate": 0.0005418937171754046,
      "loss": 2.2404,
      "step": 46417
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7835192084312439,
      "learning_rate": 0.000541891297626615,
      "loss": 2.3542,
      "step": 46418
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7375985383987427,
      "learning_rate": 0.0005418888780328534,
      "loss": 2.3445,
      "step": 46419
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.6784453392028809,
      "learning_rate": 0.00054188645839412,
      "loss": 2.0936,
      "step": 46420
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.8662909865379333,
      "learning_rate": 0.0005418840387104154,
      "loss": 2.1322,
      "step": 46421
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.6943345665931702,
      "learning_rate": 0.0005418816189817399,
      "loss": 2.2031,
      "step": 46422
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.8717727661132812,
      "learning_rate": 0.0005418791992080941,
      "loss": 2.0877,
      "step": 46423
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7306239604949951,
      "learning_rate": 0.0005418767793894785,
      "loss": 2.2354,
      "step": 46424
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7143635749816895,
      "learning_rate": 0.0005418743595258932,
      "loss": 2.1793,
      "step": 46425
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7464645504951477,
      "learning_rate": 0.000541871939617339,
      "loss": 1.9515,
      "step": 46426
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.6696249842643738,
      "learning_rate": 0.0005418695196638163,
      "loss": 2.2871,
      "step": 46427
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.8323885798454285,
      "learning_rate": 0.0005418670996653255,
      "loss": 2.3468,
      "step": 46428
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7381043434143066,
      "learning_rate": 0.000541864679621867,
      "loss": 2.3502,
      "step": 46429
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7039854526519775,
      "learning_rate": 0.0005418622595334412,
      "loss": 2.3242,
      "step": 46430
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7395527958869934,
      "learning_rate": 0.0005418598394000487,
      "loss": 2.3868,
      "step": 46431
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.73291015625,
      "learning_rate": 0.00054185741922169,
      "loss": 2.1191,
      "step": 46432
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.6879366636276245,
      "learning_rate": 0.0005418549989983653,
      "loss": 2.2444,
      "step": 46433
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7828722596168518,
      "learning_rate": 0.0005418525787300753,
      "loss": 2.2709,
      "step": 46434
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7310610413551331,
      "learning_rate": 0.0005418501584168202,
      "loss": 2.1529,
      "step": 46435
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7370925545692444,
      "learning_rate": 0.0005418477380586008,
      "loss": 2.0886,
      "step": 46436
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.9020059704780579,
      "learning_rate": 0.0005418453176554172,
      "loss": 2.1379,
      "step": 46437
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.6807796359062195,
      "learning_rate": 0.0005418428972072699,
      "loss": 2.4823,
      "step": 46438
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.706338107585907,
      "learning_rate": 0.0005418404767141596,
      "loss": 2.3034,
      "step": 46439
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7207480072975159,
      "learning_rate": 0.0005418380561760866,
      "loss": 2.3003,
      "step": 46440
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.8543062210083008,
      "learning_rate": 0.0005418356355930512,
      "loss": 2.5222,
      "step": 46441
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.6755179762840271,
      "learning_rate": 0.000541833214965054,
      "loss": 2.2183,
      "step": 46442
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7187542915344238,
      "learning_rate": 0.0005418307942920956,
      "loss": 2.2595,
      "step": 46443
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7667885422706604,
      "learning_rate": 0.0005418283735741761,
      "loss": 2.346,
      "step": 46444
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.6701594591140747,
      "learning_rate": 0.0005418259528112962,
      "loss": 2.3448,
      "step": 46445
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7143561244010925,
      "learning_rate": 0.0005418235320034565,
      "loss": 2.2157,
      "step": 46446
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.6975789070129395,
      "learning_rate": 0.0005418211111506571,
      "loss": 2.3423,
      "step": 46447
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7410518527030945,
      "learning_rate": 0.0005418186902528985,
      "loss": 2.3834,
      "step": 46448
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.8087750673294067,
      "learning_rate": 0.0005418162693101813,
      "loss": 2.1542,
      "step": 46449
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7109761834144592,
      "learning_rate": 0.0005418138483225059,
      "loss": 2.1537,
      "step": 46450
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7547305822372437,
      "learning_rate": 0.0005418114272898728,
      "loss": 2.2774,
      "step": 46451
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.6896275281906128,
      "learning_rate": 0.0005418090062122822,
      "loss": 1.9719,
      "step": 46452
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7493883371353149,
      "learning_rate": 0.0005418065850897349,
      "loss": 1.9997,
      "step": 46453
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7570146322250366,
      "learning_rate": 0.0005418041639222312,
      "loss": 2.1719,
      "step": 46454
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.6557998657226562,
      "learning_rate": 0.0005418017427097715,
      "loss": 2.1442,
      "step": 46455
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7262337803840637,
      "learning_rate": 0.0005417993214523563,
      "loss": 2.1969,
      "step": 46456
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.0685616731643677,
      "learning_rate": 0.000541796900149986,
      "loss": 2.2895,
      "step": 46457
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7690775990486145,
      "learning_rate": 0.000541794478802661,
      "loss": 2.2844,
      "step": 46458
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.780236005783081,
      "learning_rate": 0.0005417920574103821,
      "loss": 2.2213,
      "step": 46459
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7021288275718689,
      "learning_rate": 0.0005417896359731493,
      "loss": 2.1795,
      "step": 46460
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7838359475135803,
      "learning_rate": 0.0005417872144909632,
      "loss": 2.3679,
      "step": 46461
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7685806155204773,
      "learning_rate": 0.0005417847929638245,
      "loss": 2.2841,
      "step": 46462
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.8541831374168396,
      "learning_rate": 0.0005417823713917333,
      "loss": 2.1233,
      "step": 46463
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.6651095747947693,
      "learning_rate": 0.0005417799497746904,
      "loss": 2.3068,
      "step": 46464
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.718217670917511,
      "learning_rate": 0.0005417775281126957,
      "loss": 2.4103,
      "step": 46465
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.739676296710968,
      "learning_rate": 0.0005417751064057502,
      "loss": 2.1783,
      "step": 46466
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7749015092849731,
      "learning_rate": 0.0005417726846538542,
      "loss": 2.1402,
      "step": 46467
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7058967351913452,
      "learning_rate": 0.000541770262857008,
      "loss": 2.3678,
      "step": 46468
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.870756983757019,
      "learning_rate": 0.0005417678410152121,
      "loss": 2.3432,
      "step": 46469
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.79356849193573,
      "learning_rate": 0.000541765419128467,
      "loss": 2.2015,
      "step": 46470
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7009608745574951,
      "learning_rate": 0.0005417629971967732,
      "loss": 2.3492,
      "step": 46471
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.6500691175460815,
      "learning_rate": 0.0005417605752201311,
      "loss": 2.391,
      "step": 46472
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.6919491291046143,
      "learning_rate": 0.0005417581531985411,
      "loss": 1.9228,
      "step": 46473
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.6480857133865356,
      "learning_rate": 0.0005417557311320038,
      "loss": 2.3913,
      "step": 46474
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7347082495689392,
      "learning_rate": 0.0005417533090205194,
      "loss": 2.289,
      "step": 46475
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7781204581260681,
      "learning_rate": 0.0005417508868640886,
      "loss": 2.2627,
      "step": 46476
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7385596632957458,
      "learning_rate": 0.0005417484646627116,
      "loss": 2.2826,
      "step": 46477
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.6558150053024292,
      "learning_rate": 0.0005417460424163893,
      "loss": 2.5618,
      "step": 46478
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7634941935539246,
      "learning_rate": 0.0005417436201251216,
      "loss": 2.5237,
      "step": 46479
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.6680857539176941,
      "learning_rate": 0.0005417411977889093,
      "loss": 2.3361,
      "step": 46480
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7041684985160828,
      "learning_rate": 0.0005417387754077526,
      "loss": 2.4447,
      "step": 46481
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7326423525810242,
      "learning_rate": 0.0005417363529816523,
      "loss": 2.5025,
      "step": 46482
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.6924216151237488,
      "learning_rate": 0.0005417339305106085,
      "loss": 2.3239,
      "step": 46483
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.731333315372467,
      "learning_rate": 0.0005417315079946218,
      "loss": 2.4106,
      "step": 46484
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.6920583248138428,
      "learning_rate": 0.0005417290854336927,
      "loss": 2.2246,
      "step": 46485
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.703830361366272,
      "learning_rate": 0.0005417266628278216,
      "loss": 2.4396,
      "step": 46486
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7046782970428467,
      "learning_rate": 0.000541724240177009,
      "loss": 2.2314,
      "step": 46487
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.6929818987846375,
      "learning_rate": 0.0005417218174812552,
      "loss": 2.2574,
      "step": 46488
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7100718021392822,
      "learning_rate": 0.0005417193947405608,
      "loss": 2.2491,
      "step": 46489
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7713786363601685,
      "learning_rate": 0.0005417169719549262,
      "loss": 2.3533,
      "step": 46490
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.6612808108329773,
      "learning_rate": 0.0005417145491243519,
      "loss": 2.2597,
      "step": 46491
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.709454357624054,
      "learning_rate": 0.0005417121262488383,
      "loss": 2.1865,
      "step": 46492
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7505807876586914,
      "learning_rate": 0.0005417097033283858,
      "loss": 2.2826,
      "step": 46493
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7189956903457642,
      "learning_rate": 0.0005417072803629949,
      "loss": 2.196,
      "step": 46494
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.712836503982544,
      "learning_rate": 0.000541704857352666,
      "loss": 2.3625,
      "step": 46495
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.6917122602462769,
      "learning_rate": 0.0005417024342973997,
      "loss": 2.2158,
      "step": 46496
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7037789225578308,
      "learning_rate": 0.0005417000111971964,
      "loss": 2.2255,
      "step": 46497
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7202780842781067,
      "learning_rate": 0.0005416975880520563,
      "loss": 2.2998,
      "step": 46498
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.6832450032234192,
      "learning_rate": 0.0005416951648619803,
      "loss": 2.406,
      "step": 46499
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7140265107154846,
      "learning_rate": 0.0005416927416269685,
      "loss": 2.278,
      "step": 46500
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7053765654563904,
      "learning_rate": 0.0005416903183470214,
      "loss": 2.3539,
      "step": 46501
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7401317358016968,
      "learning_rate": 0.0005416878950221396,
      "loss": 2.4073,
      "step": 46502
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.8096128702163696,
      "learning_rate": 0.0005416854716523234,
      "loss": 2.2552,
      "step": 46503
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.6703451871871948,
      "learning_rate": 0.0005416830482375734,
      "loss": 2.2972,
      "step": 46504
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7045425772666931,
      "learning_rate": 0.0005416806247778898,
      "loss": 2.3975,
      "step": 46505
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.6997509598731995,
      "learning_rate": 0.0005416782012732733,
      "loss": 2.1146,
      "step": 46506
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.6915993094444275,
      "learning_rate": 0.0005416757777237243,
      "loss": 2.1474,
      "step": 46507
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.8531639575958252,
      "learning_rate": 0.0005416733541292432,
      "loss": 2.1319,
      "step": 46508
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7424329519271851,
      "learning_rate": 0.0005416709304898304,
      "loss": 2.3529,
      "step": 46509
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7246691584587097,
      "learning_rate": 0.0005416685068054864,
      "loss": 2.4465,
      "step": 46510
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7605170011520386,
      "learning_rate": 0.0005416660830762118,
      "loss": 2.01,
      "step": 46511
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.670009434223175,
      "learning_rate": 0.0005416636593020067,
      "loss": 2.1466,
      "step": 46512
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.6437318325042725,
      "learning_rate": 0.000541661235482872,
      "loss": 2.3209,
      "step": 46513
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.6530793309211731,
      "learning_rate": 0.0005416588116188077,
      "loss": 2.2323,
      "step": 46514
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.6591701507568359,
      "learning_rate": 0.0005416563877098146,
      "loss": 2.2424,
      "step": 46515
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7286093235015869,
      "learning_rate": 0.000541653963755893,
      "loss": 1.9427,
      "step": 46516
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.6889538168907166,
      "learning_rate": 0.0005416515397570434,
      "loss": 2.1604,
      "step": 46517
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.716115415096283,
      "learning_rate": 0.0005416491157132661,
      "loss": 2.2986,
      "step": 46518
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7122958302497864,
      "learning_rate": 0.0005416466916245617,
      "loss": 2.1239,
      "step": 46519
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.6667617559432983,
      "learning_rate": 0.0005416442674909306,
      "loss": 2.351,
      "step": 46520
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7575826644897461,
      "learning_rate": 0.0005416418433123734,
      "loss": 2.3749,
      "step": 46521
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7045247554779053,
      "learning_rate": 0.0005416394190888903,
      "loss": 2.1969,
      "step": 46522
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7362864017486572,
      "learning_rate": 0.0005416369948204819,
      "loss": 2.2576,
      "step": 46523
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.6766162514686584,
      "learning_rate": 0.0005416345705071486,
      "loss": 1.9863,
      "step": 46524
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.6988179683685303,
      "learning_rate": 0.000541632146148891,
      "loss": 1.9925,
      "step": 46525
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7140522599220276,
      "learning_rate": 0.0005416297217457093,
      "loss": 2.242,
      "step": 46526
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7382950782775879,
      "learning_rate": 0.0005416272972976041,
      "loss": 2.429,
      "step": 46527
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.6549876928329468,
      "learning_rate": 0.0005416248728045758,
      "loss": 2.3376,
      "step": 46528
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7223557829856873,
      "learning_rate": 0.0005416224482666248,
      "loss": 2.2028,
      "step": 46529
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7150915265083313,
      "learning_rate": 0.0005416200236837518,
      "loss": 2.3418,
      "step": 46530
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7178907990455627,
      "learning_rate": 0.0005416175990559571,
      "loss": 2.1926,
      "step": 46531
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7091696262359619,
      "learning_rate": 0.000541615174383241,
      "loss": 2.237,
      "step": 46532
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7160081267356873,
      "learning_rate": 0.000541612749665604,
      "loss": 2.0774,
      "step": 46533
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.6351714730262756,
      "learning_rate": 0.0005416103249030468,
      "loss": 2.1141,
      "step": 46534
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.6786306500434875,
      "learning_rate": 0.0005416079000955696,
      "loss": 1.9603,
      "step": 46535
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7395288348197937,
      "learning_rate": 0.000541605475243173,
      "loss": 2.3896,
      "step": 46536
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7064781188964844,
      "learning_rate": 0.0005416030503458572,
      "loss": 2.3883,
      "step": 46537
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.6546066403388977,
      "learning_rate": 0.000541600625403623,
      "loss": 2.2543,
      "step": 46538
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7211216688156128,
      "learning_rate": 0.0005415982004164708,
      "loss": 2.3223,
      "step": 46539
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.6693703532218933,
      "learning_rate": 0.0005415957753844006,
      "loss": 2.4866,
      "step": 46540
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.6995104551315308,
      "learning_rate": 0.0005415933503074134,
      "loss": 2.3275,
      "step": 46541
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7525931000709534,
      "learning_rate": 0.0005415909251855095,
      "loss": 2.3366,
      "step": 46542
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7239152789115906,
      "learning_rate": 0.0005415885000186892,
      "loss": 2.2953,
      "step": 46543
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7937795519828796,
      "learning_rate": 0.0005415860748069529,
      "loss": 2.267,
      "step": 46544
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.6808226704597473,
      "learning_rate": 0.0005415836495503013,
      "loss": 2.1954,
      "step": 46545
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7648397088050842,
      "learning_rate": 0.0005415812242487348,
      "loss": 2.5147,
      "step": 46546
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7205386161804199,
      "learning_rate": 0.0005415787989022537,
      "loss": 2.3288,
      "step": 46547
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7964221239089966,
      "learning_rate": 0.0005415763735108586,
      "loss": 2.2981,
      "step": 46548
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.6947271823883057,
      "learning_rate": 0.0005415739480745499,
      "loss": 2.0865,
      "step": 46549
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7349903583526611,
      "learning_rate": 0.000541571522593328,
      "loss": 2.3397,
      "step": 46550
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.727070689201355,
      "learning_rate": 0.0005415690970671934,
      "loss": 2.4072,
      "step": 46551
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.6957157254219055,
      "learning_rate": 0.0005415666714961466,
      "loss": 2.2402,
      "step": 46552
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.686029314994812,
      "learning_rate": 0.000541564245880188,
      "loss": 1.9683,
      "step": 46553
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.661018431186676,
      "learning_rate": 0.0005415618202193179,
      "loss": 2.1306,
      "step": 46554
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7754442095756531,
      "learning_rate": 0.000541559394513537,
      "loss": 2.0421,
      "step": 46555
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7462182641029358,
      "learning_rate": 0.0005415569687628457,
      "loss": 2.2016,
      "step": 46556
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7192927002906799,
      "learning_rate": 0.0005415545429672443,
      "loss": 2.2505,
      "step": 46557
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7040623426437378,
      "learning_rate": 0.0005415521171267334,
      "loss": 2.2212,
      "step": 46558
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7025357484817505,
      "learning_rate": 0.0005415496912413134,
      "loss": 2.1245,
      "step": 46559
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7931419610977173,
      "learning_rate": 0.0005415472653109847,
      "loss": 2.066,
      "step": 46560
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7935452461242676,
      "learning_rate": 0.0005415448393357479,
      "loss": 2.2561,
      "step": 46561
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.9253461360931396,
      "learning_rate": 0.0005415424133156033,
      "loss": 2.246,
      "step": 46562
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7149497270584106,
      "learning_rate": 0.0005415399872505514,
      "loss": 2.0529,
      "step": 46563
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7613407969474792,
      "learning_rate": 0.0005415375611405927,
      "loss": 2.2202,
      "step": 46564
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7715578675270081,
      "learning_rate": 0.0005415351349857277,
      "loss": 2.4418,
      "step": 46565
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.6571616530418396,
      "learning_rate": 0.0005415327087859565,
      "loss": 1.9178,
      "step": 46566
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7476023435592651,
      "learning_rate": 0.00054153028254128,
      "loss": 2.1593,
      "step": 46567
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7301928400993347,
      "learning_rate": 0.0005415278562516984,
      "loss": 2.3002,
      "step": 46568
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7549305558204651,
      "learning_rate": 0.0005415254299172122,
      "loss": 2.2307,
      "step": 46569
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7932860851287842,
      "learning_rate": 0.000541523003537822,
      "loss": 2.1075,
      "step": 46570
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.8841125965118408,
      "learning_rate": 0.000541520577113528,
      "loss": 2.1942,
      "step": 46571
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.687147855758667,
      "learning_rate": 0.0005415181506443306,
      "loss": 2.1174,
      "step": 46572
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7609744668006897,
      "learning_rate": 0.0005415157241302307,
      "loss": 2.428,
      "step": 46573
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7861348986625671,
      "learning_rate": 0.0005415132975712283,
      "loss": 2.0578,
      "step": 46574
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7035468220710754,
      "learning_rate": 0.000541510870967324,
      "loss": 2.3154,
      "step": 46575
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.6779117584228516,
      "learning_rate": 0.0005415084443185184,
      "loss": 2.4024,
      "step": 46576
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.8261622786521912,
      "learning_rate": 0.0005415060176248117,
      "loss": 2.3977,
      "step": 46577
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7136803269386292,
      "learning_rate": 0.0005415035908862046,
      "loss": 2.2801,
      "step": 46578
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.738544225692749,
      "learning_rate": 0.0005415011641026973,
      "loss": 2.5155,
      "step": 46579
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.8225641846656799,
      "learning_rate": 0.0005414987372742905,
      "loss": 2.1211,
      "step": 46580
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7579811215400696,
      "learning_rate": 0.0005414963104009845,
      "loss": 2.1453,
      "step": 46581
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.6998944282531738,
      "learning_rate": 0.0005414938834827797,
      "loss": 2.4578,
      "step": 46582
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7150428891181946,
      "learning_rate": 0.0005414914565196766,
      "loss": 1.9914,
      "step": 46583
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.706213116645813,
      "learning_rate": 0.0005414890295116758,
      "loss": 2.253,
      "step": 46584
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7605195045471191,
      "learning_rate": 0.0005414866024587776,
      "loss": 2.3956,
      "step": 46585
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7487369179725647,
      "learning_rate": 0.0005414841753609824,
      "loss": 2.1961,
      "step": 46586
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7183136940002441,
      "learning_rate": 0.0005414817482182908,
      "loss": 2.137,
      "step": 46587
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.855864942073822,
      "learning_rate": 0.0005414793210307032,
      "loss": 2.4975,
      "step": 46588
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7232981324195862,
      "learning_rate": 0.0005414768937982199,
      "loss": 2.2651,
      "step": 46589
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.8908661007881165,
      "learning_rate": 0.0005414744665208416,
      "loss": 2.2151,
      "step": 46590
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7492514848709106,
      "learning_rate": 0.0005414720391985687,
      "loss": 2.2624,
      "step": 46591
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7636227607727051,
      "learning_rate": 0.0005414696118314016,
      "loss": 2.239,
      "step": 46592
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.6992093920707703,
      "learning_rate": 0.0005414671844193406,
      "loss": 2.1624,
      "step": 46593
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7187300324440002,
      "learning_rate": 0.0005414647569623864,
      "loss": 2.2566,
      "step": 46594
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.6934862732887268,
      "learning_rate": 0.0005414623294605393,
      "loss": 2.314,
      "step": 46595
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.6513301134109497,
      "learning_rate": 0.0005414599019137998,
      "loss": 2.1794,
      "step": 46596
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7685180306434631,
      "learning_rate": 0.0005414574743221683,
      "loss": 2.192,
      "step": 46597
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.735772967338562,
      "learning_rate": 0.0005414550466856453,
      "loss": 2.1761,
      "step": 46598
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7216960191726685,
      "learning_rate": 0.0005414526190042313,
      "loss": 2.2602,
      "step": 46599
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.8103559613227844,
      "learning_rate": 0.0005414501912779267,
      "loss": 2.3005,
      "step": 46600
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7890390753746033,
      "learning_rate": 0.000541447763506732,
      "loss": 2.4206,
      "step": 46601
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7123221158981323,
      "learning_rate": 0.0005414453356906475,
      "loss": 2.2178,
      "step": 46602
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7518011331558228,
      "learning_rate": 0.0005414429078296739,
      "loss": 2.2276,
      "step": 46603
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7551032304763794,
      "learning_rate": 0.0005414404799238113,
      "loss": 2.1985,
      "step": 46604
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.678245484828949,
      "learning_rate": 0.0005414380519730605,
      "loss": 1.8897,
      "step": 46605
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.8942261934280396,
      "learning_rate": 0.0005414356239774218,
      "loss": 2.2308,
      "step": 46606
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.818254292011261,
      "learning_rate": 0.0005414331959368956,
      "loss": 2.0932,
      "step": 46607
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.6387813091278076,
      "learning_rate": 0.0005414307678514825,
      "loss": 2.0423,
      "step": 46608
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.6860682368278503,
      "learning_rate": 0.0005414283397211828,
      "loss": 2.1356,
      "step": 46609
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.8256871700286865,
      "learning_rate": 0.000541425911545997,
      "loss": 2.3414,
      "step": 46610
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.6701550483703613,
      "learning_rate": 0.0005414234833259256,
      "loss": 2.343,
      "step": 46611
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7466825246810913,
      "learning_rate": 0.000541421055060969,
      "loss": 2.0887,
      "step": 46612
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7237319946289062,
      "learning_rate": 0.0005414186267511277,
      "loss": 2.3665,
      "step": 46613
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.6781230568885803,
      "learning_rate": 0.000541416198396402,
      "loss": 2.281,
      "step": 46614
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.6619880199432373,
      "learning_rate": 0.0005414137699967926,
      "loss": 1.9965,
      "step": 46615
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.6784308552742004,
      "learning_rate": 0.0005414113415522997,
      "loss": 2.0701,
      "step": 46616
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.6563624143600464,
      "learning_rate": 0.000541408913062924,
      "loss": 2.2484,
      "step": 46617
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.6582245230674744,
      "learning_rate": 0.0005414064845286658,
      "loss": 2.0403,
      "step": 46618
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7152423858642578,
      "learning_rate": 0.0005414040559495256,
      "loss": 2.256,
      "step": 46619
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.8566771149635315,
      "learning_rate": 0.0005414016273255038,
      "loss": 2.3715,
      "step": 46620
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.6788341999053955,
      "learning_rate": 0.000541399198656601,
      "loss": 2.3233,
      "step": 46621
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7224428057670593,
      "learning_rate": 0.0005413967699428173,
      "loss": 1.9834,
      "step": 46622
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7193295359611511,
      "learning_rate": 0.0005413943411841535,
      "loss": 2.3561,
      "step": 46623
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.6760554909706116,
      "learning_rate": 0.0005413919123806099,
      "loss": 2.1478,
      "step": 46624
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.6771955490112305,
      "learning_rate": 0.000541389483532187,
      "loss": 2.3745,
      "step": 46625
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.696681559085846,
      "learning_rate": 0.0005413870546388852,
      "loss": 2.286,
      "step": 46626
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7783323526382446,
      "learning_rate": 0.0005413846257007051,
      "loss": 2.3147,
      "step": 46627
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.6501486301422119,
      "learning_rate": 0.000541382196717647,
      "loss": 2.2618,
      "step": 46628
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7233433127403259,
      "learning_rate": 0.0005413797676897114,
      "loss": 2.3439,
      "step": 46629
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7584272623062134,
      "learning_rate": 0.0005413773386168986,
      "loss": 2.4897,
      "step": 46630
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7685487270355225,
      "learning_rate": 0.0005413749094992094,
      "loss": 2.1901,
      "step": 46631
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.744515061378479,
      "learning_rate": 0.000541372480336644,
      "loss": 2.4429,
      "step": 46632
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7359245419502258,
      "learning_rate": 0.0005413700511292029,
      "loss": 2.0743,
      "step": 46633
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7165008187294006,
      "learning_rate": 0.0005413676218768864,
      "loss": 2.3058,
      "step": 46634
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7259074449539185,
      "learning_rate": 0.0005413651925796953,
      "loss": 2.4072,
      "step": 46635
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7529022097587585,
      "learning_rate": 0.0005413627632376298,
      "loss": 2.248,
      "step": 46636
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7235498428344727,
      "learning_rate": 0.0005413603338506903,
      "loss": 2.1738,
      "step": 46637
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.6815610527992249,
      "learning_rate": 0.0005413579044188775,
      "loss": 2.4037,
      "step": 46638
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.8412957787513733,
      "learning_rate": 0.0005413554749421917,
      "loss": 2.2226,
      "step": 46639
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7341932654380798,
      "learning_rate": 0.0005413530454206332,
      "loss": 2.3806,
      "step": 46640
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.6952933073043823,
      "learning_rate": 0.0005413506158542028,
      "loss": 2.3179,
      "step": 46641
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7554706931114197,
      "learning_rate": 0.0005413481862429006,
      "loss": 2.3918,
      "step": 46642
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.8448486328125,
      "learning_rate": 0.0005413457565867273,
      "loss": 2.0394,
      "step": 46643
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.6708329319953918,
      "learning_rate": 0.0005413433268856834,
      "loss": 2.1837,
      "step": 46644
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.9146744608879089,
      "learning_rate": 0.000541340897139769,
      "loss": 2.3596,
      "step": 46645
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.8867771625518799,
      "learning_rate": 0.0005413384673489849,
      "loss": 2.1813,
      "step": 46646
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.82965487241745,
      "learning_rate": 0.0005413360375133313,
      "loss": 2.2809,
      "step": 46647
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.9209051132202148,
      "learning_rate": 0.0005413336076328089,
      "loss": 2.2194,
      "step": 46648
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7421661615371704,
      "learning_rate": 0.000541331177707418,
      "loss": 2.0128,
      "step": 46649
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7142453193664551,
      "learning_rate": 0.0005413287477371591,
      "loss": 2.4185,
      "step": 46650
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7273688316345215,
      "learning_rate": 0.0005413263177220326,
      "loss": 2.1537,
      "step": 46651
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.708664059638977,
      "learning_rate": 0.000541323887662039,
      "loss": 2.3278,
      "step": 46652
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7776938080787659,
      "learning_rate": 0.0005413214575571787,
      "loss": 2.3625,
      "step": 46653
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.6553853750228882,
      "learning_rate": 0.0005413190274074522,
      "loss": 2.0217,
      "step": 46654
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.6784049868583679,
      "learning_rate": 0.0005413165972128599,
      "loss": 2.4294,
      "step": 46655
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.8799314498901367,
      "learning_rate": 0.0005413141669734023,
      "loss": 2.2908,
      "step": 46656
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.6418302059173584,
      "learning_rate": 0.0005413117366890799,
      "loss": 2.1953,
      "step": 46657
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7924010157585144,
      "learning_rate": 0.0005413093063598931,
      "loss": 2.331,
      "step": 46658
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7551506757736206,
      "learning_rate": 0.0005413068759858423,
      "loss": 2.2875,
      "step": 46659
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7765339016914368,
      "learning_rate": 0.0005413044455669279,
      "loss": 2.243,
      "step": 46660
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.6431248188018799,
      "learning_rate": 0.0005413020151031506,
      "loss": 2.3286,
      "step": 46661
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.0184590816497803,
      "learning_rate": 0.0005412995845945107,
      "loss": 2.0442,
      "step": 46662
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.65312659740448,
      "learning_rate": 0.0005412971540410086,
      "loss": 2.0566,
      "step": 46663
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.8306090831756592,
      "learning_rate": 0.0005412947234426448,
      "loss": 2.1661,
      "step": 46664
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7657603025436401,
      "learning_rate": 0.0005412922927994198,
      "loss": 2.1861,
      "step": 46665
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.6747577786445618,
      "learning_rate": 0.0005412898621113338,
      "loss": 2.3037,
      "step": 46666
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7430068254470825,
      "learning_rate": 0.0005412874313783878,
      "loss": 2.1658,
      "step": 46667
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7910383939743042,
      "learning_rate": 0.0005412850006005817,
      "loss": 2.2393,
      "step": 46668
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7264306545257568,
      "learning_rate": 0.0005412825697779162,
      "loss": 2.3273,
      "step": 46669
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7269579768180847,
      "learning_rate": 0.0005412801389103918,
      "loss": 2.2041,
      "step": 46670
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7742516398429871,
      "learning_rate": 0.0005412777079980088,
      "loss": 2.1655,
      "step": 46671
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.6647537350654602,
      "learning_rate": 0.0005412752770407676,
      "loss": 2.2947,
      "step": 46672
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.6851726770401001,
      "learning_rate": 0.0005412728460386689,
      "loss": 2.1843,
      "step": 46673
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.6620179414749146,
      "learning_rate": 0.000541270414991713,
      "loss": 2.2821,
      "step": 46674
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.9084179997444153,
      "learning_rate": 0.0005412679838999005,
      "loss": 2.1686,
      "step": 46675
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7318814992904663,
      "learning_rate": 0.0005412655527632315,
      "loss": 2.1782,
      "step": 46676
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.8194682002067566,
      "learning_rate": 0.0005412631215817068,
      "loss": 2.3179,
      "step": 46677
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7986129522323608,
      "learning_rate": 0.0005412606903553266,
      "loss": 2.5268,
      "step": 46678
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7328477501869202,
      "learning_rate": 0.0005412582590840917,
      "loss": 2.3844,
      "step": 46679
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.808323323726654,
      "learning_rate": 0.0005412558277680023,
      "loss": 2.1244,
      "step": 46680
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.6997485160827637,
      "learning_rate": 0.0005412533964070588,
      "loss": 2.1012,
      "step": 46681
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7316906452178955,
      "learning_rate": 0.0005412509650012618,
      "loss": 2.3373,
      "step": 46682
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.668660044670105,
      "learning_rate": 0.0005412485335506116,
      "loss": 2.3668,
      "step": 46683
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7346406579017639,
      "learning_rate": 0.0005412461020551088,
      "loss": 2.2982,
      "step": 46684
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7567669153213501,
      "learning_rate": 0.0005412436705147538,
      "loss": 2.2714,
      "step": 46685
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7139257788658142,
      "learning_rate": 0.0005412412389295471,
      "loss": 2.396,
      "step": 46686
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.6439856886863708,
      "learning_rate": 0.000541238807299489,
      "loss": 1.948,
      "step": 46687
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.662419855594635,
      "learning_rate": 0.0005412363756245801,
      "loss": 2.1323,
      "step": 46688
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.8377622365951538,
      "learning_rate": 0.0005412339439048209,
      "loss": 2.3499,
      "step": 46689
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7238803505897522,
      "learning_rate": 0.0005412315121402115,
      "loss": 2.1821,
      "step": 46690
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.6822713613510132,
      "learning_rate": 0.0005412290803307529,
      "loss": 2.2058,
      "step": 46691
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.6515188217163086,
      "learning_rate": 0.000541226648476445,
      "loss": 2.2897,
      "step": 46692
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.6690176129341125,
      "learning_rate": 0.0005412242165772887,
      "loss": 2.1603,
      "step": 46693
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.6825780272483826,
      "learning_rate": 0.0005412217846332842,
      "loss": 2.2381,
      "step": 46694
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.8076930046081543,
      "learning_rate": 0.000541219352644432,
      "loss": 2.2359,
      "step": 46695
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.742232084274292,
      "learning_rate": 0.0005412169206107326,
      "loss": 2.0914,
      "step": 46696
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.6636621356010437,
      "learning_rate": 0.0005412144885321864,
      "loss": 2.3657,
      "step": 46697
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7794264554977417,
      "learning_rate": 0.0005412120564087939,
      "loss": 2.4162,
      "step": 46698
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.6569973230361938,
      "learning_rate": 0.0005412096242405555,
      "loss": 2.269,
      "step": 46699
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7070808410644531,
      "learning_rate": 0.0005412071920274718,
      "loss": 2.3696,
      "step": 46700
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.6939905285835266,
      "learning_rate": 0.000541204759769543,
      "loss": 2.1478,
      "step": 46701
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7089270353317261,
      "learning_rate": 0.0005412023274667696,
      "loss": 2.1786,
      "step": 46702
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7703996896743774,
      "learning_rate": 0.0005411998951191522,
      "loss": 2.1568,
      "step": 46703
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.6387091875076294,
      "learning_rate": 0.0005411974627266913,
      "loss": 2.2116,
      "step": 46704
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7750310301780701,
      "learning_rate": 0.0005411950302893872,
      "loss": 2.3964,
      "step": 46705
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.6903188824653625,
      "learning_rate": 0.0005411925978072403,
      "loss": 2.0391,
      "step": 46706
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.6982248425483704,
      "learning_rate": 0.0005411901652802512,
      "loss": 2.4825,
      "step": 46707
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7185529470443726,
      "learning_rate": 0.0005411877327084202,
      "loss": 2.1275,
      "step": 46708
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.8189519047737122,
      "learning_rate": 0.000541185300091748,
      "loss": 2.3285,
      "step": 46709
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.6980510354042053,
      "learning_rate": 0.0005411828674302347,
      "loss": 2.307,
      "step": 46710
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7657620906829834,
      "learning_rate": 0.0005411804347238812,
      "loss": 2.2689,
      "step": 46711
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.6854966878890991,
      "learning_rate": 0.0005411780019726877,
      "loss": 2.2289,
      "step": 46712
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.6743300557136536,
      "learning_rate": 0.0005411755691766545,
      "loss": 2.2127,
      "step": 46713
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.8062958717346191,
      "learning_rate": 0.0005411731363357822,
      "loss": 2.3177,
      "step": 46714
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.6401434540748596,
      "learning_rate": 0.0005411707034500714,
      "loss": 2.0879,
      "step": 46715
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.8725517988204956,
      "learning_rate": 0.0005411682705195222,
      "loss": 2.1185,
      "step": 46716
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7225383520126343,
      "learning_rate": 0.0005411658375441354,
      "loss": 2.2822,
      "step": 46717
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.9252380132675171,
      "learning_rate": 0.0005411634045239113,
      "loss": 2.2333,
      "step": 46718
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.6841160655021667,
      "learning_rate": 0.0005411609714588505,
      "loss": 2.4526,
      "step": 46719
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.8269199728965759,
      "learning_rate": 0.0005411585383489532,
      "loss": 2.0926,
      "step": 46720
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7628110647201538,
      "learning_rate": 0.00054115610519422,
      "loss": 2.2347,
      "step": 46721
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.8823485970497131,
      "learning_rate": 0.0005411536719946513,
      "loss": 2.1645,
      "step": 46722
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7012640237808228,
      "learning_rate": 0.0005411512387502476,
      "loss": 2.3026,
      "step": 46723
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7036888599395752,
      "learning_rate": 0.0005411488054610093,
      "loss": 2.248,
      "step": 46724
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.6375115513801575,
      "learning_rate": 0.000541146372126937,
      "loss": 2.2446,
      "step": 46725
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7000635862350464,
      "learning_rate": 0.0005411439387480308,
      "loss": 2.2341,
      "step": 46726
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.6502084732055664,
      "learning_rate": 0.0005411415053242916,
      "loss": 2.3716,
      "step": 46727
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.697698175907135,
      "learning_rate": 0.0005411390718557196,
      "loss": 2.4178,
      "step": 46728
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.8282325863838196,
      "learning_rate": 0.0005411366383423153,
      "loss": 2.3639,
      "step": 46729
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7403396964073181,
      "learning_rate": 0.000541134204784079,
      "loss": 2.1951,
      "step": 46730
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.6519961357116699,
      "learning_rate": 0.0005411317711810114,
      "loss": 2.0667,
      "step": 46731
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.6859798431396484,
      "learning_rate": 0.000541129337533113,
      "loss": 2.1642,
      "step": 46732
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7086515426635742,
      "learning_rate": 0.000541126903840384,
      "loss": 2.421,
      "step": 46733
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7544359564781189,
      "learning_rate": 0.0005411244701028249,
      "loss": 2.2942,
      "step": 46734
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7127543687820435,
      "learning_rate": 0.0005411220363204362,
      "loss": 2.3133,
      "step": 46735
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.805901825428009,
      "learning_rate": 0.0005411196024932185,
      "loss": 2.4651,
      "step": 46736
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.6391494274139404,
      "learning_rate": 0.000541117168621172,
      "loss": 2.352,
      "step": 46737
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7689998149871826,
      "learning_rate": 0.0005411147347042973,
      "loss": 2.166,
      "step": 46738
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7075815200805664,
      "learning_rate": 0.0005411123007425948,
      "loss": 2.2079,
      "step": 46739
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7557204365730286,
      "learning_rate": 0.000541109866736065,
      "loss": 2.2883,
      "step": 46740
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.8518755435943604,
      "learning_rate": 0.0005411074326847082,
      "loss": 2.2653,
      "step": 46741
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7305139303207397,
      "learning_rate": 0.0005411049985885252,
      "loss": 2.1868,
      "step": 46742
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.8078579306602478,
      "learning_rate": 0.0005411025644475161,
      "loss": 2.2344,
      "step": 46743
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7011294960975647,
      "learning_rate": 0.0005411001302616815,
      "loss": 2.2889,
      "step": 46744
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.6440159678459167,
      "learning_rate": 0.0005410976960310218,
      "loss": 2.2119,
      "step": 46745
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.706438422203064,
      "learning_rate": 0.0005410952617555375,
      "loss": 2.3086,
      "step": 46746
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.8103753924369812,
      "learning_rate": 0.000541092827435229,
      "loss": 2.2405,
      "step": 46747
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7208762168884277,
      "learning_rate": 0.0005410903930700969,
      "loss": 2.3347,
      "step": 46748
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.6721839308738708,
      "learning_rate": 0.0005410879586601415,
      "loss": 2.1676,
      "step": 46749
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7044333815574646,
      "learning_rate": 0.0005410855242053633,
      "loss": 2.235,
      "step": 46750
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.6889853477478027,
      "learning_rate": 0.0005410830897057627,
      "loss": 2.1344,
      "step": 46751
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.6579216122627258,
      "learning_rate": 0.0005410806551613403,
      "loss": 2.0501,
      "step": 46752
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7678750157356262,
      "learning_rate": 0.0005410782205720964,
      "loss": 2.2646,
      "step": 46753
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7322285175323486,
      "learning_rate": 0.0005410757859380315,
      "loss": 2.4463,
      "step": 46754
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7181456089019775,
      "learning_rate": 0.000541073351259146,
      "loss": 2.471,
      "step": 46755
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7517589926719666,
      "learning_rate": 0.0005410709165354405,
      "loss": 2.2843,
      "step": 46756
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.6789011359214783,
      "learning_rate": 0.0005410684817669154,
      "loss": 2.1856,
      "step": 46757
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7283867001533508,
      "learning_rate": 0.000541066046953571,
      "loss": 2.3414,
      "step": 46758
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7107311487197876,
      "learning_rate": 0.0005410636120954079,
      "loss": 2.1522,
      "step": 46759
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7067207098007202,
      "learning_rate": 0.0005410611771924267,
      "loss": 2.1833,
      "step": 46760
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.6914608478546143,
      "learning_rate": 0.0005410587422446275,
      "loss": 2.1974,
      "step": 46761
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7108147740364075,
      "learning_rate": 0.0005410563072520109,
      "loss": 2.273,
      "step": 46762
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7027382254600525,
      "learning_rate": 0.0005410538722145775,
      "loss": 2.2374,
      "step": 46763
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7671127915382385,
      "learning_rate": 0.0005410514371323276,
      "loss": 2.4192,
      "step": 46764
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7088439464569092,
      "learning_rate": 0.0005410490020052617,
      "loss": 2.2666,
      "step": 46765
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.8028104901313782,
      "learning_rate": 0.0005410465668333801,
      "loss": 2.109,
      "step": 46766
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.680680513381958,
      "learning_rate": 0.0005410441316166837,
      "loss": 2.2039,
      "step": 46767
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.6606121063232422,
      "learning_rate": 0.0005410416963551723,
      "loss": 2.1158,
      "step": 46768
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.6580414772033691,
      "learning_rate": 0.0005410392610488468,
      "loss": 2.2635,
      "step": 46769
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.6697556376457214,
      "learning_rate": 0.0005410368256977077,
      "loss": 2.058,
      "step": 46770
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.6194577813148499,
      "learning_rate": 0.0005410343903017552,
      "loss": 2.0083,
      "step": 46771
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7759096622467041,
      "learning_rate": 0.0005410319548609898,
      "loss": 2.2352,
      "step": 46772
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7894774079322815,
      "learning_rate": 0.0005410295193754121,
      "loss": 2.1047,
      "step": 46773
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7931929230690002,
      "learning_rate": 0.0005410270838450225,
      "loss": 2.4107,
      "step": 46774
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.6904681324958801,
      "learning_rate": 0.0005410246482698214,
      "loss": 2.1477,
      "step": 46775
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7768964171409607,
      "learning_rate": 0.0005410222126498092,
      "loss": 2.2849,
      "step": 46776
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7049208283424377,
      "learning_rate": 0.0005410197769849864,
      "loss": 2.1416,
      "step": 46777
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7108160853385925,
      "learning_rate": 0.0005410173412753534,
      "loss": 2.38,
      "step": 46778
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7964509129524231,
      "learning_rate": 0.0005410149055209109,
      "loss": 2.3851,
      "step": 46779
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.67305588722229,
      "learning_rate": 0.0005410124697216591,
      "loss": 2.3479,
      "step": 46780
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7391013503074646,
      "learning_rate": 0.0005410100338775986,
      "loss": 2.1731,
      "step": 46781
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.9678106904029846,
      "learning_rate": 0.0005410075979887297,
      "loss": 1.9667,
      "step": 46782
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.6585325598716736,
      "learning_rate": 0.0005410051620550531,
      "loss": 2.045,
      "step": 46783
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7634423971176147,
      "learning_rate": 0.0005410027260765689,
      "loss": 2.1944,
      "step": 46784
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.9617829322814941,
      "learning_rate": 0.0005410002900532779,
      "loss": 2.2748,
      "step": 46785
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.684373676776886,
      "learning_rate": 0.0005409978539851802,
      "loss": 2.2486,
      "step": 46786
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.805751621723175,
      "learning_rate": 0.0005409954178722766,
      "loss": 2.4508,
      "step": 46787
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.8404096364974976,
      "learning_rate": 0.0005409929817145674,
      "loss": 2.1281,
      "step": 46788
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.8421888947486877,
      "learning_rate": 0.0005409905455120529,
      "loss": 2.0831,
      "step": 46789
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7117310166358948,
      "learning_rate": 0.0005409881092647339,
      "loss": 2.2647,
      "step": 46790
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.8259810209274292,
      "learning_rate": 0.0005409856729726105,
      "loss": 2.1232,
      "step": 46791
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.0723761320114136,
      "learning_rate": 0.0005409832366356835,
      "loss": 2.0891,
      "step": 46792
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7414964437484741,
      "learning_rate": 0.000540980800253953,
      "loss": 2.3934,
      "step": 46793
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.6924619078636169,
      "learning_rate": 0.0005409783638274197,
      "loss": 2.0673,
      "step": 46794
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.8569322824478149,
      "learning_rate": 0.0005409759273560839,
      "loss": 2.3066,
      "step": 46795
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.11940336227417,
      "learning_rate": 0.0005409734908399463,
      "loss": 2.3233,
      "step": 46796
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.6900089979171753,
      "learning_rate": 0.000540971054279007,
      "loss": 2.3265,
      "step": 46797
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.75847989320755,
      "learning_rate": 0.0005409686176732668,
      "loss": 2.2984,
      "step": 46798
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.0419909954071045,
      "learning_rate": 0.0005409661810227258,
      "loss": 2.181,
      "step": 46799
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7446755170822144,
      "learning_rate": 0.0005409637443273848,
      "loss": 2.1723,
      "step": 46800
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.6984530687332153,
      "learning_rate": 0.000540961307587244,
      "loss": 2.2615,
      "step": 46801
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.9825125336647034,
      "learning_rate": 0.0005409588708023039,
      "loss": 2.1607,
      "step": 46802
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7484095096588135,
      "learning_rate": 0.0005409564339725651,
      "loss": 2.2832,
      "step": 46803
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.6468603610992432,
      "learning_rate": 0.0005409539970980279,
      "loss": 2.0573,
      "step": 46804
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.756881058216095,
      "learning_rate": 0.0005409515601786927,
      "loss": 2.4809,
      "step": 46805
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7980300188064575,
      "learning_rate": 0.0005409491232145603,
      "loss": 2.244,
      "step": 46806
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7810183167457581,
      "learning_rate": 0.0005409466862056308,
      "loss": 2.2522,
      "step": 46807
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.6848453283309937,
      "learning_rate": 0.0005409442491519046,
      "loss": 2.2581,
      "step": 46808
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7163529992103577,
      "learning_rate": 0.0005409418120533825,
      "loss": 2.1506,
      "step": 46809
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.8338438868522644,
      "learning_rate": 0.0005409393749100647,
      "loss": 2.2128,
      "step": 46810
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7173776030540466,
      "learning_rate": 0.0005409369377219518,
      "loss": 2.2077,
      "step": 46811
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.6388956308364868,
      "learning_rate": 0.0005409345004890441,
      "loss": 2.3632,
      "step": 46812
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.8032368421554565,
      "learning_rate": 0.0005409320632113421,
      "loss": 2.2999,
      "step": 46813
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7571279406547546,
      "learning_rate": 0.0005409296258888465,
      "loss": 2.3581,
      "step": 46814
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7111808061599731,
      "learning_rate": 0.0005409271885215573,
      "loss": 2.2036,
      "step": 46815
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7380946278572083,
      "learning_rate": 0.0005409247511094752,
      "loss": 2.5267,
      "step": 46816
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.693728506565094,
      "learning_rate": 0.0005409223136526007,
      "loss": 2.018,
      "step": 46817
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.9439711570739746,
      "learning_rate": 0.0005409198761509343,
      "loss": 2.3319,
      "step": 46818
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.8238323330879211,
      "learning_rate": 0.0005409174386044762,
      "loss": 2.3338,
      "step": 46819
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7151706218719482,
      "learning_rate": 0.000540915001013227,
      "loss": 2.6194,
      "step": 46820
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7987887859344482,
      "learning_rate": 0.0005409125633771872,
      "loss": 2.179,
      "step": 46821
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7390913963317871,
      "learning_rate": 0.0005409101256963572,
      "loss": 2.228,
      "step": 46822
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.704239547252655,
      "learning_rate": 0.0005409076879707374,
      "loss": 2.2584,
      "step": 46823
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7563365697860718,
      "learning_rate": 0.0005409052502003284,
      "loss": 2.1525,
      "step": 46824
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.822751522064209,
      "learning_rate": 0.0005409028123851305,
      "loss": 2.2633,
      "step": 46825
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.6655423045158386,
      "learning_rate": 0.0005409003745251443,
      "loss": 2.1687,
      "step": 46826
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7568264603614807,
      "learning_rate": 0.0005408979366203702,
      "loss": 2.139,
      "step": 46827
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.8493149280548096,
      "learning_rate": 0.0005408954986708085,
      "loss": 2.3864,
      "step": 46828
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7707500457763672,
      "learning_rate": 0.0005408930606764599,
      "loss": 2.3106,
      "step": 46829
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7829288244247437,
      "learning_rate": 0.0005408906226373247,
      "loss": 2.231,
      "step": 46830
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.8192917704582214,
      "learning_rate": 0.0005408881845534033,
      "loss": 2.5139,
      "step": 46831
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.6977155804634094,
      "learning_rate": 0.0005408857464246963,
      "loss": 2.3326,
      "step": 46832
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7478809952735901,
      "learning_rate": 0.0005408833082512041,
      "loss": 2.2056,
      "step": 46833
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.8469233512878418,
      "learning_rate": 0.0005408808700329271,
      "loss": 2.0748,
      "step": 46834
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.684327244758606,
      "learning_rate": 0.0005408784317698658,
      "loss": 2.252,
      "step": 46835
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.8365804553031921,
      "learning_rate": 0.0005408759934620207,
      "loss": 2.3554,
      "step": 46836
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.9468001127243042,
      "learning_rate": 0.0005408735551093922,
      "loss": 2.2279,
      "step": 46837
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7319895029067993,
      "learning_rate": 0.0005408711167119807,
      "loss": 2.3261,
      "step": 46838
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7311575412750244,
      "learning_rate": 0.0005408686782697868,
      "loss": 2.2175,
      "step": 46839
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7219207882881165,
      "learning_rate": 0.0005408662397828109,
      "loss": 2.4826,
      "step": 46840
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.9866169691085815,
      "learning_rate": 0.0005408638012510533,
      "loss": 2.2268,
      "step": 46841
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7248252034187317,
      "learning_rate": 0.0005408613626745147,
      "loss": 2.3355,
      "step": 46842
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7522042989730835,
      "learning_rate": 0.0005408589240531953,
      "loss": 2.5035,
      "step": 46843
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.9193492531776428,
      "learning_rate": 0.0005408564853870957,
      "loss": 2.2189,
      "step": 46844
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7387586832046509,
      "learning_rate": 0.0005408540466762164,
      "loss": 2.4425,
      "step": 46845
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7055785059928894,
      "learning_rate": 0.0005408516079205577,
      "loss": 2.3678,
      "step": 46846
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.9154310822486877,
      "learning_rate": 0.0005408491691201202,
      "loss": 2.1911,
      "step": 46847
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.713746190071106,
      "learning_rate": 0.0005408467302749042,
      "loss": 2.2938,
      "step": 46848
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.6418527364730835,
      "learning_rate": 0.0005408442913849104,
      "loss": 1.8684,
      "step": 46849
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.0100210905075073,
      "learning_rate": 0.000540841852450139,
      "loss": 2.3205,
      "step": 46850
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7532503008842468,
      "learning_rate": 0.0005408394134705907,
      "loss": 2.4106,
      "step": 46851
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.6517637968063354,
      "learning_rate": 0.0005408369744462656,
      "loss": 2.3218,
      "step": 46852
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7182909846305847,
      "learning_rate": 0.0005408345353771644,
      "loss": 2.1639,
      "step": 46853
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.8093889951705933,
      "learning_rate": 0.0005408320962632876,
      "loss": 2.2296,
      "step": 46854
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.6988685131072998,
      "learning_rate": 0.0005408296571046355,
      "loss": 2.1728,
      "step": 46855
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7308318018913269,
      "learning_rate": 0.0005408272179012087,
      "loss": 2.2093,
      "step": 46856
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.870206892490387,
      "learning_rate": 0.0005408247786530075,
      "loss": 2.2539,
      "step": 46857
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7186105251312256,
      "learning_rate": 0.0005408223393600324,
      "loss": 2.1757,
      "step": 46858
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.702643632888794,
      "learning_rate": 0.000540819900022284,
      "loss": 2.1772,
      "step": 46859
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7237061858177185,
      "learning_rate": 0.0005408174606397625,
      "loss": 2.3228,
      "step": 46860
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7237396240234375,
      "learning_rate": 0.0005408150212124687,
      "loss": 2.3557,
      "step": 46861
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7343801259994507,
      "learning_rate": 0.0005408125817404026,
      "loss": 2.013,
      "step": 46862
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7263050675392151,
      "learning_rate": 0.0005408101422235649,
      "loss": 2.1588,
      "step": 46863
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7834439277648926,
      "learning_rate": 0.0005408077026619562,
      "loss": 2.1029,
      "step": 46864
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7626723647117615,
      "learning_rate": 0.0005408052630555768,
      "loss": 2.2204,
      "step": 46865
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7711566090583801,
      "learning_rate": 0.0005408028234044271,
      "loss": 2.2979,
      "step": 46866
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.8211483955383301,
      "learning_rate": 0.0005408003837085076,
      "loss": 2.363,
      "step": 46867
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7773460745811462,
      "learning_rate": 0.0005407979439678188,
      "loss": 2.442,
      "step": 46868
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.740129828453064,
      "learning_rate": 0.0005407955041823611,
      "loss": 2.2008,
      "step": 46869
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.6855103969573975,
      "learning_rate": 0.0005407930643521349,
      "loss": 2.1684,
      "step": 46870
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.6883289217948914,
      "learning_rate": 0.0005407906244771408,
      "loss": 2.1145,
      "step": 46871
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.6722116470336914,
      "learning_rate": 0.0005407881845573791,
      "loss": 2.1798,
      "step": 46872
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7134261727333069,
      "learning_rate": 0.0005407857445928504,
      "loss": 2.3889,
      "step": 46873
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.691010594367981,
      "learning_rate": 0.0005407833045835552,
      "loss": 2.3094,
      "step": 46874
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.6657585501670837,
      "learning_rate": 0.0005407808645294937,
      "loss": 2.2728,
      "step": 46875
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7405645847320557,
      "learning_rate": 0.0005407784244306664,
      "loss": 2.1181,
      "step": 46876
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7262507677078247,
      "learning_rate": 0.000540775984287074,
      "loss": 2.1066,
      "step": 46877
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.6936752796173096,
      "learning_rate": 0.0005407735440987167,
      "loss": 2.1854,
      "step": 46878
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.6226344108581543,
      "learning_rate": 0.0005407711038655952,
      "loss": 2.1127,
      "step": 46879
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7792284488677979,
      "learning_rate": 0.0005407686635877097,
      "loss": 2.3809,
      "step": 46880
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.745213508605957,
      "learning_rate": 0.0005407662232650607,
      "loss": 2.3511,
      "step": 46881
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7912240028381348,
      "learning_rate": 0.0005407637828976488,
      "loss": 2.1785,
      "step": 46882
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7550411820411682,
      "learning_rate": 0.0005407613424854743,
      "loss": 2.2971,
      "step": 46883
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.8224928379058838,
      "learning_rate": 0.0005407589020285378,
      "loss": 2.3736,
      "step": 46884
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.6716500520706177,
      "learning_rate": 0.0005407564615268396,
      "loss": 2.2825,
      "step": 46885
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7951668500900269,
      "learning_rate": 0.0005407540209803804,
      "loss": 2.1818,
      "step": 46886
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7358453273773193,
      "learning_rate": 0.0005407515803891604,
      "loss": 2.1707,
      "step": 46887
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.6871157288551331,
      "learning_rate": 0.00054074913975318,
      "loss": 2.2796,
      "step": 46888
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7805806398391724,
      "learning_rate": 0.00054074669907244,
      "loss": 2.3071,
      "step": 46889
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7053356766700745,
      "learning_rate": 0.0005407442583469405,
      "loss": 2.3708,
      "step": 46890
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.9358527660369873,
      "learning_rate": 0.000540741817576682,
      "loss": 2.1038,
      "step": 46891
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.6795746088027954,
      "learning_rate": 0.0005407393767616653,
      "loss": 2.1459,
      "step": 46892
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.802482008934021,
      "learning_rate": 0.0005407369359018905,
      "loss": 2.1385,
      "step": 46893
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.9418951272964478,
      "learning_rate": 0.0005407344949973581,
      "loss": 2.438,
      "step": 46894
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.8034261465072632,
      "learning_rate": 0.0005407320540480687,
      "loss": 2.3406,
      "step": 46895
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7508407235145569,
      "learning_rate": 0.0005407296130540226,
      "loss": 2.2099,
      "step": 46896
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.8609816431999207,
      "learning_rate": 0.0005407271720152203,
      "loss": 2.4503,
      "step": 46897
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7490358352661133,
      "learning_rate": 0.0005407247309316623,
      "loss": 2.3384,
      "step": 46898
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.6464259624481201,
      "learning_rate": 0.0005407222898033492,
      "loss": 2.4071,
      "step": 46899
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.8710575699806213,
      "learning_rate": 0.000540719848630281,
      "loss": 2.2802,
      "step": 46900
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.8063095808029175,
      "learning_rate": 0.0005407174074124586,
      "loss": 2.4911,
      "step": 46901
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7213379740715027,
      "learning_rate": 0.0005407149661498823,
      "loss": 2.3178,
      "step": 46902
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.8047701716423035,
      "learning_rate": 0.0005407125248425525,
      "loss": 2.1269,
      "step": 46903
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7566007971763611,
      "learning_rate": 0.0005407100834904698,
      "loss": 2.1242,
      "step": 46904
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.664253294467926,
      "learning_rate": 0.0005407076420936344,
      "loss": 2.2317,
      "step": 46905
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7515243887901306,
      "learning_rate": 0.000540705200652047,
      "loss": 2.3545,
      "step": 46906
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.8286752700805664,
      "learning_rate": 0.0005407027591657078,
      "loss": 2.3506,
      "step": 46907
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7139478325843811,
      "learning_rate": 0.0005407003176346176,
      "loss": 2.2196,
      "step": 46908
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7279188632965088,
      "learning_rate": 0.0005406978760587767,
      "loss": 2.3524,
      "step": 46909
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.6938947439193726,
      "learning_rate": 0.0005406954344381853,
      "loss": 2.3064,
      "step": 46910
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.8179240226745605,
      "learning_rate": 0.0005406929927728442,
      "loss": 2.3847,
      "step": 46911
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7263664603233337,
      "learning_rate": 0.0005406905510627538,
      "loss": 2.2474,
      "step": 46912
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7980432510375977,
      "learning_rate": 0.0005406881093079144,
      "loss": 2.398,
      "step": 46913
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.6543785929679871,
      "learning_rate": 0.0005406856675083266,
      "loss": 2.1948,
      "step": 46914
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7653446793556213,
      "learning_rate": 0.0005406832256639907,
      "loss": 2.5463,
      "step": 46915
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7392100691795349,
      "learning_rate": 0.0005406807837749073,
      "loss": 2.1644,
      "step": 46916
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7906302809715271,
      "learning_rate": 0.0005406783418410768,
      "loss": 2.3281,
      "step": 46917
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7060312628746033,
      "learning_rate": 0.0005406758998624996,
      "loss": 2.3298,
      "step": 46918
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7333802580833435,
      "learning_rate": 0.0005406734578391762,
      "loss": 2.3375,
      "step": 46919
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.663722574710846,
      "learning_rate": 0.0005406710157711071,
      "loss": 2.1559,
      "step": 46920
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7802658677101135,
      "learning_rate": 0.0005406685736582928,
      "loss": 2.22,
      "step": 46921
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.66694575548172,
      "learning_rate": 0.0005406661315007336,
      "loss": 2.0898,
      "step": 46922
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.6962028741836548,
      "learning_rate": 0.00054066368929843,
      "loss": 2.5093,
      "step": 46923
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.6434093117713928,
      "learning_rate": 0.0005406612470513825,
      "loss": 2.0629,
      "step": 46924
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.8277764320373535,
      "learning_rate": 0.0005406588047595915,
      "loss": 2.275,
      "step": 46925
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.726263165473938,
      "learning_rate": 0.0005406563624230575,
      "loss": 2.3053,
      "step": 46926
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7820366621017456,
      "learning_rate": 0.0005406539200417809,
      "loss": 2.2413,
      "step": 46927
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.6275814175605774,
      "learning_rate": 0.0005406514776157623,
      "loss": 2.169,
      "step": 46928
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7395961880683899,
      "learning_rate": 0.000540649035145002,
      "loss": 2.2098,
      "step": 46929
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.696574866771698,
      "learning_rate": 0.0005406465926295004,
      "loss": 2.2048,
      "step": 46930
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7337846755981445,
      "learning_rate": 0.0005406441500692582,
      "loss": 2.1501,
      "step": 46931
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7003219723701477,
      "learning_rate": 0.0005406417074642757,
      "loss": 2.1622,
      "step": 46932
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7444531917572021,
      "learning_rate": 0.0005406392648145533,
      "loss": 2.396,
      "step": 46933
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.0353689193725586,
      "learning_rate": 0.0005406368221200916,
      "loss": 2.4414,
      "step": 46934
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7000924944877625,
      "learning_rate": 0.0005406343793808909,
      "loss": 2.0832,
      "step": 46935
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7462719082832336,
      "learning_rate": 0.0005406319365969517,
      "loss": 2.3009,
      "step": 46936
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.872373104095459,
      "learning_rate": 0.0005406294937682745,
      "loss": 2.2444,
      "step": 46937
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7644875645637512,
      "learning_rate": 0.0005406270508948596,
      "loss": 2.2051,
      "step": 46938
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.8421856760978699,
      "learning_rate": 0.0005406246079767079,
      "loss": 2.3779,
      "step": 46939
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.89434814453125,
      "learning_rate": 0.0005406221650138194,
      "loss": 2.3446,
      "step": 46940
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.745379626750946,
      "learning_rate": 0.0005406197220061946,
      "loss": 2.0456,
      "step": 46941
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.6547701358795166,
      "learning_rate": 0.0005406172789538342,
      "loss": 2.0877,
      "step": 46942
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7238325476646423,
      "learning_rate": 0.0005406148358567384,
      "loss": 2.1287,
      "step": 46943
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.9005459547042847,
      "learning_rate": 0.0005406123927149078,
      "loss": 2.1735,
      "step": 46944
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.8272057175636292,
      "learning_rate": 0.0005406099495283427,
      "loss": 2.2935,
      "step": 46945
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7544915676116943,
      "learning_rate": 0.0005406075062970439,
      "loss": 2.2696,
      "step": 46946
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.8558833599090576,
      "learning_rate": 0.0005406050630210115,
      "loss": 2.2197,
      "step": 46947
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7065342664718628,
      "learning_rate": 0.0005406026197002461,
      "loss": 2.3193,
      "step": 46948
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.8189666867256165,
      "learning_rate": 0.0005406001763347481,
      "loss": 2.3817,
      "step": 46949
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7877883911132812,
      "learning_rate": 0.0005405977329245179,
      "loss": 2.2618,
      "step": 46950
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7124367952346802,
      "learning_rate": 0.0005405952894695562,
      "loss": 2.2623,
      "step": 46951
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.8157014846801758,
      "learning_rate": 0.0005405928459698633,
      "loss": 2.2592,
      "step": 46952
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.6966643929481506,
      "learning_rate": 0.0005405904024254395,
      "loss": 2.265,
      "step": 46953
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.6757414937019348,
      "learning_rate": 0.0005405879588362856,
      "loss": 2.4285,
      "step": 46954
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.6555812954902649,
      "learning_rate": 0.0005405855152024017,
      "loss": 2.1958,
      "step": 46955
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.8452660441398621,
      "learning_rate": 0.0005405830715237885,
      "loss": 2.278,
      "step": 46956
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.694881021976471,
      "learning_rate": 0.0005405806278004463,
      "loss": 2.2912,
      "step": 46957
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7281473278999329,
      "learning_rate": 0.0005405781840323756,
      "loss": 2.3375,
      "step": 46958
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7600839734077454,
      "learning_rate": 0.0005405757402195769,
      "loss": 2.3742,
      "step": 46959
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.8118951916694641,
      "learning_rate": 0.0005405732963620507,
      "loss": 2.3586,
      "step": 46960
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.6668752431869507,
      "learning_rate": 0.0005405708524597973,
      "loss": 2.0882,
      "step": 46961
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7209969162940979,
      "learning_rate": 0.0005405684085128172,
      "loss": 2.1091,
      "step": 46962
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7219737768173218,
      "learning_rate": 0.000540565964521111,
      "loss": 2.2465,
      "step": 46963
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.6951939463615417,
      "learning_rate": 0.000540563520484679,
      "loss": 2.552,
      "step": 46964
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.6950849890708923,
      "learning_rate": 0.0005405610764035217,
      "loss": 2.3686,
      "step": 46965
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.6798505187034607,
      "learning_rate": 0.0005405586322776395,
      "loss": 2.0466,
      "step": 46966
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.6791510581970215,
      "learning_rate": 0.000540556188107033,
      "loss": 2.1243,
      "step": 46967
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.702394962310791,
      "learning_rate": 0.0005405537438917025,
      "loss": 2.1765,
      "step": 46968
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.70428466796875,
      "learning_rate": 0.0005405512996316486,
      "loss": 2.3748,
      "step": 46969
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7197948694229126,
      "learning_rate": 0.0005405488553268716,
      "loss": 2.1913,
      "step": 46970
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.6343956589698792,
      "learning_rate": 0.000540546410977372,
      "loss": 2.2552,
      "step": 46971
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7748360633850098,
      "learning_rate": 0.0005405439665831504,
      "loss": 2.2783,
      "step": 46972
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7162119746208191,
      "learning_rate": 0.0005405415221442071,
      "loss": 2.3672,
      "step": 46973
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.6619309186935425,
      "learning_rate": 0.0005405390776605425,
      "loss": 2.3596,
      "step": 46974
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.6436206698417664,
      "learning_rate": 0.0005405366331321572,
      "loss": 2.1909,
      "step": 46975
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.6982041597366333,
      "learning_rate": 0.0005405341885590516,
      "loss": 2.2809,
      "step": 46976
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.68417888879776,
      "learning_rate": 0.0005405317439412261,
      "loss": 2.3509,
      "step": 46977
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7103133201599121,
      "learning_rate": 0.0005405292992786813,
      "loss": 2.2073,
      "step": 46978
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.703777015209198,
      "learning_rate": 0.0005405268545714175,
      "loss": 2.2203,
      "step": 46979
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7171511650085449,
      "learning_rate": 0.0005405244098194352,
      "loss": 2.3953,
      "step": 46980
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.714044451713562,
      "learning_rate": 0.000540521965022735,
      "loss": 2.2188,
      "step": 46981
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7795478105545044,
      "learning_rate": 0.0005405195201813171,
      "loss": 2.3488,
      "step": 46982
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7312068939208984,
      "learning_rate": 0.0005405170752951822,
      "loss": 2.1344,
      "step": 46983
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7398974895477295,
      "learning_rate": 0.0005405146303643306,
      "loss": 2.2411,
      "step": 46984
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.6618854999542236,
      "learning_rate": 0.0005405121853887627,
      "loss": 2.3682,
      "step": 46985
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.728500485420227,
      "learning_rate": 0.000540509740368479,
      "loss": 2.3102,
      "step": 46986
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7511248588562012,
      "learning_rate": 0.0005405072953034801,
      "loss": 2.3002,
      "step": 46987
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.6323150396347046,
      "learning_rate": 0.0005405048501937663,
      "loss": 2.3974,
      "step": 46988
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7739845514297485,
      "learning_rate": 0.0005405024050393382,
      "loss": 2.2449,
      "step": 46989
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.716566264629364,
      "learning_rate": 0.0005404999598401961,
      "loss": 2.4022,
      "step": 46990
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.6841937303543091,
      "learning_rate": 0.0005404975145963406,
      "loss": 2.4049,
      "step": 46991
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.8012043833732605,
      "learning_rate": 0.000540495069307772,
      "loss": 1.9481,
      "step": 46992
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7644671201705933,
      "learning_rate": 0.0005404926239744908,
      "loss": 2.2485,
      "step": 46993
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.8753490447998047,
      "learning_rate": 0.0005404901785964975,
      "loss": 2.3978,
      "step": 46994
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.762444257736206,
      "learning_rate": 0.0005404877331737926,
      "loss": 2.4673,
      "step": 46995
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.6927464604377747,
      "learning_rate": 0.0005404852877063764,
      "loss": 2.2514,
      "step": 46996
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.715697705745697,
      "learning_rate": 0.0005404828421942496,
      "loss": 2.2594,
      "step": 46997
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7399930953979492,
      "learning_rate": 0.0005404803966374123,
      "loss": 2.3424,
      "step": 46998
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7394384741783142,
      "learning_rate": 0.0005404779510358652,
      "loss": 2.1182,
      "step": 46999
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.775050938129425,
      "learning_rate": 0.0005404755053896087,
      "loss": 2.3227,
      "step": 47000
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7862280011177063,
      "learning_rate": 0.0005404730596986435,
      "loss": 2.1583,
      "step": 47001
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.6768062710762024,
      "learning_rate": 0.0005404706139629696,
      "loss": 2.3277,
      "step": 47002
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7244856357574463,
      "learning_rate": 0.0005404681681825877,
      "loss": 2.3035,
      "step": 47003
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.809762179851532,
      "learning_rate": 0.0005404657223574982,
      "loss": 2.2749,
      "step": 47004
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.6474451422691345,
      "learning_rate": 0.0005404632764877016,
      "loss": 2.0726,
      "step": 47005
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.739208996295929,
      "learning_rate": 0.0005404608305731983,
      "loss": 2.2844,
      "step": 47006
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7814844846725464,
      "learning_rate": 0.0005404583846139889,
      "loss": 2.2815,
      "step": 47007
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.770642101764679,
      "learning_rate": 0.0005404559386100738,
      "loss": 2.3666,
      "step": 47008
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.6820759177207947,
      "learning_rate": 0.0005404534925614532,
      "loss": 2.0305,
      "step": 47009
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.9922272562980652,
      "learning_rate": 0.0005404510464681279,
      "loss": 2.0222,
      "step": 47010
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.6669414043426514,
      "learning_rate": 0.0005404486003300981,
      "loss": 2.2504,
      "step": 47011
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7455993294715881,
      "learning_rate": 0.0005404461541473645,
      "loss": 2.2965,
      "step": 47012
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7878243327140808,
      "learning_rate": 0.0005404437079199274,
      "loss": 2.2665,
      "step": 47013
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7821688055992126,
      "learning_rate": 0.0005404412616477871,
      "loss": 2.5312,
      "step": 47014
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7057772278785706,
      "learning_rate": 0.0005404388153309444,
      "loss": 2.1966,
      "step": 47015
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.8141469955444336,
      "learning_rate": 0.0005404363689693996,
      "loss": 2.2227,
      "step": 47016
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.8552327156066895,
      "learning_rate": 0.000540433922563153,
      "loss": 2.2369,
      "step": 47017
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.707137942314148,
      "learning_rate": 0.0005404314761122053,
      "loss": 2.0675,
      "step": 47018
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.6366252303123474,
      "learning_rate": 0.0005404290296165569,
      "loss": 2.1588,
      "step": 47019
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7914252281188965,
      "learning_rate": 0.0005404265830762082,
      "loss": 2.1972,
      "step": 47020
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7372151613235474,
      "learning_rate": 0.0005404241364911595,
      "loss": 2.2393,
      "step": 47021
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.6858569979667664,
      "learning_rate": 0.0005404216898614115,
      "loss": 2.0128,
      "step": 47022
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7998175621032715,
      "learning_rate": 0.0005404192431869646,
      "loss": 2.2834,
      "step": 47023
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7338938117027283,
      "learning_rate": 0.0005404167964678192,
      "loss": 2.3414,
      "step": 47024
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.9312999844551086,
      "learning_rate": 0.0005404143497039758,
      "loss": 2.1574,
      "step": 47025
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.6891579031944275,
      "learning_rate": 0.0005404119028954348,
      "loss": 2.5155,
      "step": 47026
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7768384218215942,
      "learning_rate": 0.0005404094560421967,
      "loss": 2.4789,
      "step": 47027
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.8275748491287231,
      "learning_rate": 0.0005404070091442619,
      "loss": 2.287,
      "step": 47028
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.8754575848579407,
      "learning_rate": 0.000540404562201631,
      "loss": 2.3649,
      "step": 47029
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.6686606407165527,
      "learning_rate": 0.0005404021152143043,
      "loss": 2.064,
      "step": 47030
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.6890689730644226,
      "learning_rate": 0.0005403996681822824,
      "loss": 2.1743,
      "step": 47031
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7179516553878784,
      "learning_rate": 0.0005403972211055656,
      "loss": 2.1281,
      "step": 47032
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.9739667177200317,
      "learning_rate": 0.0005403947739841544,
      "loss": 2.2816,
      "step": 47033
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7920061945915222,
      "learning_rate": 0.0005403923268180492,
      "loss": 2.2414,
      "step": 47034
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7391668558120728,
      "learning_rate": 0.0005403898796072506,
      "loss": 2.1969,
      "step": 47035
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.8318431973457336,
      "learning_rate": 0.0005403874323517589,
      "loss": 2.2062,
      "step": 47036
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7205803990364075,
      "learning_rate": 0.0005403849850515748,
      "loss": 2.2362,
      "step": 47037
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7288422584533691,
      "learning_rate": 0.0005403825377066985,
      "loss": 2.1409,
      "step": 47038
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.9441274404525757,
      "learning_rate": 0.0005403800903171305,
      "loss": 2.2374,
      "step": 47039
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.8106662631034851,
      "learning_rate": 0.0005403776428828715,
      "loss": 2.2672,
      "step": 47040
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7694059014320374,
      "learning_rate": 0.0005403751954039215,
      "loss": 2.4118,
      "step": 47041
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7624847292900085,
      "learning_rate": 0.0005403727478802814,
      "loss": 2.3016,
      "step": 47042
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7822750210762024,
      "learning_rate": 0.0005403703003119514,
      "loss": 2.1064,
      "step": 47043
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7536131143569946,
      "learning_rate": 0.000540367852698932,
      "loss": 2.4184,
      "step": 47044
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.6380589604377747,
      "learning_rate": 0.0005403654050412238,
      "loss": 2.2621,
      "step": 47045
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.8479553461074829,
      "learning_rate": 0.000540362957338827,
      "loss": 2.3002,
      "step": 47046
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7097246646881104,
      "learning_rate": 0.0005403605095917422,
      "loss": 2.2017,
      "step": 47047
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7204167246818542,
      "learning_rate": 0.0005403580617999699,
      "loss": 2.3922,
      "step": 47048
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7184621095657349,
      "learning_rate": 0.0005403556139635105,
      "loss": 2.2975,
      "step": 47049
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7497143149375916,
      "learning_rate": 0.0005403531660823645,
      "loss": 2.3812,
      "step": 47050
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7275131940841675,
      "learning_rate": 0.0005403507181565322,
      "loss": 2.1941,
      "step": 47051
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7097691297531128,
      "learning_rate": 0.0005403482701860143,
      "loss": 2.2905,
      "step": 47052
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7179152965545654,
      "learning_rate": 0.0005403458221708109,
      "loss": 2.2867,
      "step": 47053
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.6818517446517944,
      "learning_rate": 0.0005403433741109228,
      "loss": 2.0966,
      "step": 47054
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7451489567756653,
      "learning_rate": 0.0005403409260063504,
      "loss": 2.1764,
      "step": 47055
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7158611416816711,
      "learning_rate": 0.000540338477857094,
      "loss": 2.1767,
      "step": 47056
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7101566195487976,
      "learning_rate": 0.0005403360296631542,
      "loss": 2.2421,
      "step": 47057
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.666645348072052,
      "learning_rate": 0.0005403335814245313,
      "loss": 2.2259,
      "step": 47058
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.6973479986190796,
      "learning_rate": 0.0005403311331412258,
      "loss": 2.1497,
      "step": 47059
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7026190161705017,
      "learning_rate": 0.0005403286848132384,
      "loss": 2.4083,
      "step": 47060
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7185637950897217,
      "learning_rate": 0.0005403262364405693,
      "loss": 2.2074,
      "step": 47061
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7076919674873352,
      "learning_rate": 0.0005403237880232189,
      "loss": 2.3129,
      "step": 47062
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7520106434822083,
      "learning_rate": 0.0005403213395611878,
      "loss": 2.2493,
      "step": 47063
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7157024145126343,
      "learning_rate": 0.0005403188910544765,
      "loss": 2.4838,
      "step": 47064
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.6732160449028015,
      "learning_rate": 0.0005403164425030855,
      "loss": 2.3062,
      "step": 47065
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7298057675361633,
      "learning_rate": 0.0005403139939070149,
      "loss": 2.0239,
      "step": 47066
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.6437395811080933,
      "learning_rate": 0.0005403115452662655,
      "loss": 2.0923,
      "step": 47067
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7136744260787964,
      "learning_rate": 0.0005403090965808376,
      "loss": 2.178,
      "step": 47068
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7001488208770752,
      "learning_rate": 0.0005403066478507319,
      "loss": 2.1143,
      "step": 47069
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.6655015349388123,
      "learning_rate": 0.0005403041990759485,
      "loss": 2.3133,
      "step": 47070
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.6552194952964783,
      "learning_rate": 0.000540301750256488,
      "loss": 2.152,
      "step": 47071
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7744035124778748,
      "learning_rate": 0.0005402993013923508,
      "loss": 2.3204,
      "step": 47072
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7091845273971558,
      "learning_rate": 0.0005402968524835375,
      "loss": 2.2837,
      "step": 47073
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.6674473881721497,
      "learning_rate": 0.0005402944035300485,
      "loss": 2.1461,
      "step": 47074
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7421608567237854,
      "learning_rate": 0.0005402919545318842,
      "loss": 2.0838,
      "step": 47075
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.6821697354316711,
      "learning_rate": 0.0005402895054890453,
      "loss": 2.3825,
      "step": 47076
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7330201864242554,
      "learning_rate": 0.0005402870564015317,
      "loss": 2.3171,
      "step": 47077
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.6917909979820251,
      "learning_rate": 0.0005402846072693443,
      "loss": 2.2456,
      "step": 47078
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.643561840057373,
      "learning_rate": 0.0005402821580924836,
      "loss": 2.3633,
      "step": 47079
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.6499173045158386,
      "learning_rate": 0.0005402797088709498,
      "loss": 2.2314,
      "step": 47080
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.652640700340271,
      "learning_rate": 0.0005402772596047435,
      "loss": 2.3873,
      "step": 47081
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.6853933334350586,
      "learning_rate": 0.0005402748102938652,
      "loss": 2.0188,
      "step": 47082
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.6955296397209167,
      "learning_rate": 0.0005402723609383151,
      "loss": 2.2184,
      "step": 47083
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.649806022644043,
      "learning_rate": 0.0005402699115380939,
      "loss": 2.0318,
      "step": 47084
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7137224078178406,
      "learning_rate": 0.000540267462093202,
      "loss": 2.3506,
      "step": 47085
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.8039960265159607,
      "learning_rate": 0.0005402650126036399,
      "loss": 2.2448,
      "step": 47086
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.5922810435295105,
      "learning_rate": 0.000540262563069408,
      "loss": 2.2601,
      "step": 47087
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.8097473978996277,
      "learning_rate": 0.0005402601134905067,
      "loss": 2.2943,
      "step": 47088
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.628227174282074,
      "learning_rate": 0.0005402576638669365,
      "loss": 2.1219,
      "step": 47089
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7026053071022034,
      "learning_rate": 0.0005402552141986979,
      "loss": 2.1628,
      "step": 47090
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.838583767414093,
      "learning_rate": 0.0005402527644857913,
      "loss": 2.3679,
      "step": 47091
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7671038508415222,
      "learning_rate": 0.0005402503147282172,
      "loss": 2.2585,
      "step": 47092
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.8111296892166138,
      "learning_rate": 0.000540247864925976,
      "loss": 2.1762,
      "step": 47093
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7051941752433777,
      "learning_rate": 0.0005402454150790681,
      "loss": 2.4092,
      "step": 47094
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.6871632933616638,
      "learning_rate": 0.0005402429651874942,
      "loss": 2.322,
      "step": 47095
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.6812929511070251,
      "learning_rate": 0.0005402405152512545,
      "loss": 2.3346,
      "step": 47096
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.6913114786148071,
      "learning_rate": 0.0005402380652703496,
      "loss": 2.2442,
      "step": 47097
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.6327908039093018,
      "learning_rate": 0.00054023561524478,
      "loss": 2.2083,
      "step": 47098
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.9785196185112,
      "learning_rate": 0.0005402331651745459,
      "loss": 2.3238,
      "step": 47099
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.6454055905342102,
      "learning_rate": 0.0005402307150596479,
      "loss": 2.4224,
      "step": 47100
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.6964001655578613,
      "learning_rate": 0.0005402282649000866,
      "loss": 2.0858,
      "step": 47101
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7091565132141113,
      "learning_rate": 0.0005402258146958623,
      "loss": 2.2129,
      "step": 47102
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7140097618103027,
      "learning_rate": 0.0005402233644469755,
      "loss": 2.1993,
      "step": 47103
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.6666122078895569,
      "learning_rate": 0.0005402209141534266,
      "loss": 2.2479,
      "step": 47104
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.623053789138794,
      "learning_rate": 0.0005402184638152161,
      "loss": 2.1331,
      "step": 47105
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7419663071632385,
      "learning_rate": 0.0005402160134323445,
      "loss": 2.3105,
      "step": 47106
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.6947274804115295,
      "learning_rate": 0.0005402135630048121,
      "loss": 2.268,
      "step": 47107
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.65267014503479,
      "learning_rate": 0.0005402111125326196,
      "loss": 2.0536,
      "step": 47108
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7068420648574829,
      "learning_rate": 0.0005402086620157672,
      "loss": 2.1491,
      "step": 47109
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7326367497444153,
      "learning_rate": 0.0005402062114542556,
      "loss": 2.2874,
      "step": 47110
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.6821478009223938,
      "learning_rate": 0.0005402037608480851,
      "loss": 2.3125,
      "step": 47111
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.6935425996780396,
      "learning_rate": 0.0005402013101972561,
      "loss": 2.3098,
      "step": 47112
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7479387521743774,
      "learning_rate": 0.0005401988595017692,
      "loss": 2.4752,
      "step": 47113
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7183694243431091,
      "learning_rate": 0.0005401964087616248,
      "loss": 2.3091,
      "step": 47114
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7014402151107788,
      "learning_rate": 0.0005401939579768233,
      "loss": 2.2195,
      "step": 47115
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7665628790855408,
      "learning_rate": 0.0005401915071473652,
      "loss": 2.4149,
      "step": 47116
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7190272808074951,
      "learning_rate": 0.0005401890562732511,
      "loss": 2.3847,
      "step": 47117
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.6531596779823303,
      "learning_rate": 0.0005401866053544813,
      "loss": 2.029,
      "step": 47118
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.6408197283744812,
      "learning_rate": 0.0005401841543910562,
      "loss": 2.2021,
      "step": 47119
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.6600286364555359,
      "learning_rate": 0.0005401817033829764,
      "loss": 2.3557,
      "step": 47120
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.6786080598831177,
      "learning_rate": 0.0005401792523302422,
      "loss": 2.2042,
      "step": 47121
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7042741179466248,
      "learning_rate": 0.0005401768012328542,
      "loss": 2.1392,
      "step": 47122
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.6954942941665649,
      "learning_rate": 0.0005401743500908127,
      "loss": 2.1545,
      "step": 47123
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7002567648887634,
      "learning_rate": 0.0005401718989041183,
      "loss": 2.4005,
      "step": 47124
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7403092384338379,
      "learning_rate": 0.0005401694476727715,
      "loss": 2.15,
      "step": 47125
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.6851475834846497,
      "learning_rate": 0.0005401669963967726,
      "loss": 2.1787,
      "step": 47126
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.6645302176475525,
      "learning_rate": 0.0005401645450761221,
      "loss": 2.2327,
      "step": 47127
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7018216848373413,
      "learning_rate": 0.0005401620937108206,
      "loss": 2.4706,
      "step": 47128
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.679588794708252,
      "learning_rate": 0.0005401596423008683,
      "loss": 2.2519,
      "step": 47129
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.6783128380775452,
      "learning_rate": 0.0005401571908462659,
      "loss": 2.1426,
      "step": 47130
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7664605379104614,
      "learning_rate": 0.0005401547393470137,
      "loss": 2.2066,
      "step": 47131
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.647793710231781,
      "learning_rate": 0.0005401522878031121,
      "loss": 2.3484,
      "step": 47132
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7277635931968689,
      "learning_rate": 0.0005401498362145618,
      "loss": 2.3458,
      "step": 47133
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.8081222772598267,
      "learning_rate": 0.000540147384581363,
      "loss": 2.4383,
      "step": 47134
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7365807890892029,
      "learning_rate": 0.0005401449329035164,
      "loss": 2.1623,
      "step": 47135
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.6495561599731445,
      "learning_rate": 0.0005401424811810221,
      "loss": 2.2347,
      "step": 47136
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.6407170295715332,
      "learning_rate": 0.000540140029413881,
      "loss": 2.1517,
      "step": 47137
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7232040166854858,
      "learning_rate": 0.0005401375776020932,
      "loss": 2.2562,
      "step": 47138
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.6350343823432922,
      "learning_rate": 0.0005401351257456593,
      "loss": 2.1466,
      "step": 47139
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7742947340011597,
      "learning_rate": 0.0005401326738445799,
      "loss": 2.257,
      "step": 47140
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7284936308860779,
      "learning_rate": 0.0005401302218988552,
      "loss": 2.4328,
      "step": 47141
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7419198155403137,
      "learning_rate": 0.0005401277699084857,
      "loss": 2.2974,
      "step": 47142
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.6892749071121216,
      "learning_rate": 0.000540125317873472,
      "loss": 2.1944,
      "step": 47143
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.8068373799324036,
      "learning_rate": 0.0005401228657938144,
      "loss": 2.3069,
      "step": 47144
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.822772204875946,
      "learning_rate": 0.0005401204136695134,
      "loss": 2.3557,
      "step": 47145
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7140839695930481,
      "learning_rate": 0.0005401179615005696,
      "loss": 2.2855,
      "step": 47146
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7234206199645996,
      "learning_rate": 0.0005401155092869833,
      "loss": 2.1637,
      "step": 47147
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.6795086860656738,
      "learning_rate": 0.000540113057028755,
      "loss": 2.1965,
      "step": 47148
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7209305167198181,
      "learning_rate": 0.0005401106047258851,
      "loss": 2.1867,
      "step": 47149
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.6664864420890808,
      "learning_rate": 0.0005401081523783742,
      "loss": 2.2925,
      "step": 47150
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.6773321628570557,
      "learning_rate": 0.0005401056999862226,
      "loss": 2.4348,
      "step": 47151
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.6987074613571167,
      "learning_rate": 0.0005401032475494308,
      "loss": 2.3807,
      "step": 47152
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.6890178918838501,
      "learning_rate": 0.0005401007950679993,
      "loss": 2.0255,
      "step": 47153
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7347354292869568,
      "learning_rate": 0.0005400983425419286,
      "loss": 2.5162,
      "step": 47154
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.6637850999832153,
      "learning_rate": 0.000540095889971219,
      "loss": 2.2493,
      "step": 47155
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.6659584045410156,
      "learning_rate": 0.000540093437355871,
      "loss": 2.2226,
      "step": 47156
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.8006020188331604,
      "learning_rate": 0.0005400909846958851,
      "loss": 2.4263,
      "step": 47157
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.6938930749893188,
      "learning_rate": 0.0005400885319912619,
      "loss": 2.0465,
      "step": 47158
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7303542494773865,
      "learning_rate": 0.0005400860792420016,
      "loss": 2.2509,
      "step": 47159
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.6697003245353699,
      "learning_rate": 0.0005400836264481048,
      "loss": 2.0155,
      "step": 47160
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7635682821273804,
      "learning_rate": 0.0005400811736095718,
      "loss": 2.1706,
      "step": 47161
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7082152962684631,
      "learning_rate": 0.0005400787207264033,
      "loss": 2.388,
      "step": 47162
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.6974512934684753,
      "learning_rate": 0.0005400762677985996,
      "loss": 2.1994,
      "step": 47163
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.8752977848052979,
      "learning_rate": 0.0005400738148261612,
      "loss": 2.3313,
      "step": 47164
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7261055707931519,
      "learning_rate": 0.0005400713618090885,
      "loss": 2.2528,
      "step": 47165
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.677954375743866,
      "learning_rate": 0.0005400689087473821,
      "loss": 2.3239,
      "step": 47166
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7343921661376953,
      "learning_rate": 0.0005400664556410423,
      "loss": 2.1031,
      "step": 47167
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.0239851474761963,
      "learning_rate": 0.0005400640024900697,
      "loss": 2.5206,
      "step": 47168
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.758083164691925,
      "learning_rate": 0.0005400615492944645,
      "loss": 2.0161,
      "step": 47169
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.8553615212440491,
      "learning_rate": 0.0005400590960542275,
      "loss": 2.1709,
      "step": 47170
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.800462007522583,
      "learning_rate": 0.0005400566427693589,
      "loss": 2.1486,
      "step": 47171
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.8150081634521484,
      "learning_rate": 0.0005400541894398593,
      "loss": 2.2644,
      "step": 47172
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.1246172189712524,
      "learning_rate": 0.0005400517360657289,
      "loss": 2.0984,
      "step": 47173
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7519605159759521,
      "learning_rate": 0.0005400492826469686,
      "loss": 2.317,
      "step": 47174
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.753383457660675,
      "learning_rate": 0.0005400468291835784,
      "loss": 2.2222,
      "step": 47175
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7839964628219604,
      "learning_rate": 0.0005400443756755591,
      "loss": 2.287,
      "step": 47176
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7255538105964661,
      "learning_rate": 0.000540041922122911,
      "loss": 2.3883,
      "step": 47177
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7357727885246277,
      "learning_rate": 0.0005400394685256346,
      "loss": 2.2819,
      "step": 47178
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7311283946037292,
      "learning_rate": 0.0005400370148837301,
      "loss": 2.0987,
      "step": 47179
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7491709589958191,
      "learning_rate": 0.0005400345611971984,
      "loss": 2.336,
      "step": 47180
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.6746770143508911,
      "learning_rate": 0.0005400321074660397,
      "loss": 2.2608,
      "step": 47181
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.901819109916687,
      "learning_rate": 0.0005400296536902546,
      "loss": 2.268,
      "step": 47182
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.6548084020614624,
      "learning_rate": 0.0005400271998698434,
      "loss": 2.107,
      "step": 47183
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.922516405582428,
      "learning_rate": 0.0005400247460048066,
      "loss": 2.0997,
      "step": 47184
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.723311722278595,
      "learning_rate": 0.0005400222920951446,
      "loss": 2.2009,
      "step": 47185
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7024356126785278,
      "learning_rate": 0.000540019838140858,
      "loss": 2.1352,
      "step": 47186
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.8227646350860596,
      "learning_rate": 0.0005400173841419472,
      "loss": 2.382,
      "step": 47187
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7031937837600708,
      "learning_rate": 0.0005400149300984126,
      "loss": 2.3205,
      "step": 47188
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.6884818077087402,
      "learning_rate": 0.0005400124760102546,
      "loss": 2.5436,
      "step": 47189
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.8417328596115112,
      "learning_rate": 0.000540010021877474,
      "loss": 2.314,
      "step": 47190
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.8015644550323486,
      "learning_rate": 0.0005400075677000708,
      "loss": 2.317,
      "step": 47191
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7276374697685242,
      "learning_rate": 0.0005400051134780457,
      "loss": 2.1318,
      "step": 47192
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7262181639671326,
      "learning_rate": 0.0005400026592113992,
      "loss": 2.4401,
      "step": 47193
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.6911435723304749,
      "learning_rate": 0.0005400002049001314,
      "loss": 2.0947,
      "step": 47194
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.8090222477912903,
      "learning_rate": 0.0005399977505442433,
      "loss": 2.1561,
      "step": 47195
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.8078710436820984,
      "learning_rate": 0.0005399952961437351,
      "loss": 2.0741,
      "step": 47196
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7406045198440552,
      "learning_rate": 0.0005399928416986072,
      "loss": 2.2578,
      "step": 47197
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7149465680122375,
      "learning_rate": 0.00053999038720886,
      "loss": 2.3365,
      "step": 47198
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7196362018585205,
      "learning_rate": 0.0005399879326744941,
      "loss": 2.3986,
      "step": 47199
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.0540695190429688,
      "learning_rate": 0.0005399854780955099,
      "loss": 2.2714,
      "step": 47200
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.696365237236023,
      "learning_rate": 0.000539983023471908,
      "loss": 2.2916,
      "step": 47201
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7682743072509766,
      "learning_rate": 0.0005399805688036886,
      "loss": 2.2688,
      "step": 47202
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.9637796878814697,
      "learning_rate": 0.0005399781140908523,
      "loss": 2.3282,
      "step": 47203
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.6879151463508606,
      "learning_rate": 0.0005399756593333997,
      "loss": 2.1288,
      "step": 47204
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7230072617530823,
      "learning_rate": 0.0005399732045313308,
      "loss": 2.0148,
      "step": 47205
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7462756633758545,
      "learning_rate": 0.0005399707496846466,
      "loss": 2.3007,
      "step": 47206
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.812312662601471,
      "learning_rate": 0.0005399682947933471,
      "loss": 2.5977,
      "step": 47207
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7593120336532593,
      "learning_rate": 0.0005399658398574332,
      "loss": 2.4104,
      "step": 47208
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7856101989746094,
      "learning_rate": 0.0005399633848769049,
      "loss": 2.4157,
      "step": 47209
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.1091753244400024,
      "learning_rate": 0.000539960929851763,
      "loss": 2.169,
      "step": 47210
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.6907060146331787,
      "learning_rate": 0.0005399584747820078,
      "loss": 2.2252,
      "step": 47211
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.8406491279602051,
      "learning_rate": 0.0005399560196676398,
      "loss": 2.4916,
      "step": 47212
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.8481477499008179,
      "learning_rate": 0.0005399535645086595,
      "loss": 2.486,
      "step": 47213
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7258812785148621,
      "learning_rate": 0.0005399511093050672,
      "loss": 2.2369,
      "step": 47214
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7547587752342224,
      "learning_rate": 0.0005399486540568634,
      "loss": 2.1288,
      "step": 47215
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.6440291404724121,
      "learning_rate": 0.0005399461987640488,
      "loss": 1.9715,
      "step": 47216
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7495824694633484,
      "learning_rate": 0.0005399437434266236,
      "loss": 2.3542,
      "step": 47217
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7545822262763977,
      "learning_rate": 0.0005399412880445883,
      "loss": 2.3508,
      "step": 47218
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.6797018647193909,
      "learning_rate": 0.0005399388326179435,
      "loss": 2.3979,
      "step": 47219
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.722317636013031,
      "learning_rate": 0.0005399363771466894,
      "loss": 2.2525,
      "step": 47220
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7097633481025696,
      "learning_rate": 0.0005399339216308266,
      "loss": 2.2852,
      "step": 47221
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.8165228962898254,
      "learning_rate": 0.0005399314660703556,
      "loss": 2.3113,
      "step": 47222
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.6766301393508911,
      "learning_rate": 0.0005399290104652768,
      "loss": 2.1386,
      "step": 47223
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7539858818054199,
      "learning_rate": 0.0005399265548155907,
      "loss": 2.1613,
      "step": 47224
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.8872091174125671,
      "learning_rate": 0.0005399240991212977,
      "loss": 2.0385,
      "step": 47225
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.6785825490951538,
      "learning_rate": 0.0005399216433823983,
      "loss": 2.041,
      "step": 47226
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7941638827323914,
      "learning_rate": 0.0005399191875988929,
      "loss": 2.2062,
      "step": 47227
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.8920553922653198,
      "learning_rate": 0.0005399167317707821,
      "loss": 2.3223,
      "step": 47228
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.6624460816383362,
      "learning_rate": 0.0005399142758980662,
      "loss": 2.2359,
      "step": 47229
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7133057117462158,
      "learning_rate": 0.0005399118199807456,
      "loss": 2.2311,
      "step": 47230
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.9574804306030273,
      "learning_rate": 0.0005399093640188209,
      "loss": 2.1061,
      "step": 47231
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7983794212341309,
      "learning_rate": 0.0005399069080122927,
      "loss": 2.4745,
      "step": 47232
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.6426024436950684,
      "learning_rate": 0.000539904451961161,
      "loss": 2.1647,
      "step": 47233
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.9359349012374878,
      "learning_rate": 0.0005399019958654267,
      "loss": 2.3603,
      "step": 47234
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7189372181892395,
      "learning_rate": 0.0005398995397250901,
      "loss": 2.2462,
      "step": 47235
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.742630124092102,
      "learning_rate": 0.0005398970835401516,
      "loss": 2.3072,
      "step": 47236
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.694435715675354,
      "learning_rate": 0.0005398946273106117,
      "loss": 2.2905,
      "step": 47237
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.8548702001571655,
      "learning_rate": 0.0005398921710364708,
      "loss": 2.0467,
      "step": 47238
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.8177291750907898,
      "learning_rate": 0.0005398897147177295,
      "loss": 2.3032,
      "step": 47239
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7126863598823547,
      "learning_rate": 0.0005398872583543882,
      "loss": 2.3439,
      "step": 47240
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.9011496305465698,
      "learning_rate": 0.0005398848019464473,
      "loss": 2.4292,
      "step": 47241
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7849264144897461,
      "learning_rate": 0.0005398823454939072,
      "loss": 2.2513,
      "step": 47242
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7162663340568542,
      "learning_rate": 0.0005398798889967686,
      "loss": 2.1933,
      "step": 47243
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7283070683479309,
      "learning_rate": 0.0005398774324550318,
      "loss": 2.1546,
      "step": 47244
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7233208417892456,
      "learning_rate": 0.0005398749758686971,
      "loss": 2.2768,
      "step": 47245
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.8933306932449341,
      "learning_rate": 0.0005398725192377651,
      "loss": 2.2492,
      "step": 47246
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.703254759311676,
      "learning_rate": 0.0005398700625622366,
      "loss": 2.1216,
      "step": 47247
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7612406611442566,
      "learning_rate": 0.0005398676058421114,
      "loss": 2.34,
      "step": 47248
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.8073580861091614,
      "learning_rate": 0.0005398651490773903,
      "loss": 2.0648,
      "step": 47249
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.703449547290802,
      "learning_rate": 0.0005398626922680738,
      "loss": 2.3132,
      "step": 47250
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7682996392250061,
      "learning_rate": 0.0005398602354141625,
      "loss": 2.2952,
      "step": 47251
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.77101731300354,
      "learning_rate": 0.0005398577785156564,
      "loss": 2.1953,
      "step": 47252
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7716392874717712,
      "learning_rate": 0.0005398553215725563,
      "loss": 2.4458,
      "step": 47253
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.6847072839736938,
      "learning_rate": 0.0005398528645848625,
      "loss": 2.0884,
      "step": 47254
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7065532803535461,
      "learning_rate": 0.0005398504075525756,
      "loss": 2.0308,
      "step": 47255
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.681712806224823,
      "learning_rate": 0.000539847950475696,
      "loss": 1.9805,
      "step": 47256
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7094728946685791,
      "learning_rate": 0.0005398454933542241,
      "loss": 2.2734,
      "step": 47257
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.6733455657958984,
      "learning_rate": 0.0005398430361881604,
      "loss": 2.0577,
      "step": 47258
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.6868409514427185,
      "learning_rate": 0.0005398405789775054,
      "loss": 2.2596,
      "step": 47259
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7053988575935364,
      "learning_rate": 0.0005398381217222594,
      "loss": 2.3654,
      "step": 47260
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7656774520874023,
      "learning_rate": 0.000539835664422423,
      "loss": 2.241,
      "step": 47261
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7612229585647583,
      "learning_rate": 0.0005398332070779967,
      "loss": 2.3475,
      "step": 47262
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.6429193615913391,
      "learning_rate": 0.0005398307496889807,
      "loss": 1.9559,
      "step": 47263
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7116706967353821,
      "learning_rate": 0.0005398282922553758,
      "loss": 2.3501,
      "step": 47264
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7172706127166748,
      "learning_rate": 0.0005398258347771822,
      "loss": 2.1554,
      "step": 47265
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.6952751874923706,
      "learning_rate": 0.0005398233772544006,
      "loss": 2.3857,
      "step": 47266
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7105976343154907,
      "learning_rate": 0.0005398209196870312,
      "loss": 2.5078,
      "step": 47267
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7119994759559631,
      "learning_rate": 0.0005398184620750746,
      "loss": 2.3957,
      "step": 47268
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.6740389466285706,
      "learning_rate": 0.0005398160044185311,
      "loss": 2.3296,
      "step": 47269
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.6732121109962463,
      "learning_rate": 0.0005398135467174014,
      "loss": 2.2607,
      "step": 47270
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7712035775184631,
      "learning_rate": 0.0005398110889716859,
      "loss": 2.2779,
      "step": 47271
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.6699088215827942,
      "learning_rate": 0.0005398086311813848,
      "loss": 2.1993,
      "step": 47272
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.6434279084205627,
      "learning_rate": 0.0005398061733464988,
      "loss": 2.4504,
      "step": 47273
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.6357215642929077,
      "learning_rate": 0.0005398037154670284,
      "loss": 2.3517,
      "step": 47274
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7816653847694397,
      "learning_rate": 0.0005398012575429739,
      "loss": 2.2079,
      "step": 47275
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7532714605331421,
      "learning_rate": 0.000539798799574336,
      "loss": 2.5187,
      "step": 47276
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.8098825812339783,
      "learning_rate": 0.0005397963415611147,
      "loss": 2.3902,
      "step": 47277
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7415671944618225,
      "learning_rate": 0.0005397938835033109,
      "loss": 2.1798,
      "step": 47278
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.6670512557029724,
      "learning_rate": 0.0005397914254009248,
      "loss": 2.3704,
      "step": 47279
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7582992911338806,
      "learning_rate": 0.000539788967253957,
      "loss": 2.2225,
      "step": 47280
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7975329756736755,
      "learning_rate": 0.0005397865090624079,
      "loss": 2.3892,
      "step": 47281
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7297185659408569,
      "learning_rate": 0.0005397840508262779,
      "loss": 2.2813,
      "step": 47282
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.0623804330825806,
      "learning_rate": 0.0005397815925455677,
      "loss": 2.1564,
      "step": 47283
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.738192617893219,
      "learning_rate": 0.0005397791342202774,
      "loss": 2.3111,
      "step": 47284
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7692714333534241,
      "learning_rate": 0.0005397766758504076,
      "loss": 2.3423,
      "step": 47285
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.9180765151977539,
      "learning_rate": 0.0005397742174359588,
      "loss": 2.3326,
      "step": 47286
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7811992764472961,
      "learning_rate": 0.0005397717589769315,
      "loss": 2.2607,
      "step": 47287
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7149621248245239,
      "learning_rate": 0.0005397693004733262,
      "loss": 2.1085,
      "step": 47288
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.8163263201713562,
      "learning_rate": 0.0005397668419251432,
      "loss": 2.1985,
      "step": 47289
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.9175174236297607,
      "learning_rate": 0.000539764383332383,
      "loss": 2.2072,
      "step": 47290
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7724214196205139,
      "learning_rate": 0.000539761924695046,
      "loss": 2.1657,
      "step": 47291
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7055222988128662,
      "learning_rate": 0.0005397594660131328,
      "loss": 2.0772,
      "step": 47292
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.686618983745575,
      "learning_rate": 0.0005397570072866438,
      "loss": 2.259,
      "step": 47293
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7709233164787292,
      "learning_rate": 0.0005397545485155795,
      "loss": 2.3133,
      "step": 47294
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.8298953175544739,
      "learning_rate": 0.0005397520896999403,
      "loss": 2.0825,
      "step": 47295
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.6845980286598206,
      "learning_rate": 0.0005397496308397266,
      "loss": 2.351,
      "step": 47296
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7299198508262634,
      "learning_rate": 0.000539747171934939,
      "loss": 2.2882,
      "step": 47297
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.6815566420555115,
      "learning_rate": 0.0005397447129855777,
      "loss": 2.063,
      "step": 47298
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7764337062835693,
      "learning_rate": 0.0005397422539916434,
      "loss": 2.5189,
      "step": 47299
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.6413537263870239,
      "learning_rate": 0.0005397397949531365,
      "loss": 2.3794,
      "step": 47300
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.750830352306366,
      "learning_rate": 0.0005397373358700576,
      "loss": 2.1076,
      "step": 47301
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7782809734344482,
      "learning_rate": 0.0005397348767424068,
      "loss": 1.9954,
      "step": 47302
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.6543981432914734,
      "learning_rate": 0.0005397324175701848,
      "loss": 2.2686,
      "step": 47303
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7228389382362366,
      "learning_rate": 0.0005397299583533921,
      "loss": 2.0614,
      "step": 47304
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7114226222038269,
      "learning_rate": 0.000539727499092029,
      "loss": 2.2047,
      "step": 47305
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.6632533669471741,
      "learning_rate": 0.0005397250397860961,
      "loss": 2.1914,
      "step": 47306
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7718634009361267,
      "learning_rate": 0.0005397225804355937,
      "loss": 2.2036,
      "step": 47307
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.661862850189209,
      "learning_rate": 0.0005397201210405224,
      "loss": 2.1522,
      "step": 47308
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7101560235023499,
      "learning_rate": 0.0005397176616008827,
      "loss": 2.4821,
      "step": 47309
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7126891613006592,
      "learning_rate": 0.0005397152021166747,
      "loss": 2.1513,
      "step": 47310
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.704173743724823,
      "learning_rate": 0.0005397127425878993,
      "loss": 2.456,
      "step": 47311
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.6830158233642578,
      "learning_rate": 0.0005397102830145569,
      "loss": 2.3395,
      "step": 47312
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.763532817363739,
      "learning_rate": 0.0005397078233966477,
      "loss": 2.1922,
      "step": 47313
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.6659402847290039,
      "learning_rate": 0.0005397053637341723,
      "loss": 2.2335,
      "step": 47314
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7167143225669861,
      "learning_rate": 0.000539702904027131,
      "loss": 2.26,
      "step": 47315
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.8270685076713562,
      "learning_rate": 0.0005397004442755245,
      "loss": 2.3095,
      "step": 47316
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7781575322151184,
      "learning_rate": 0.0005396979844793533,
      "loss": 2.249,
      "step": 47317
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.8964592814445496,
      "learning_rate": 0.0005396955246386176,
      "loss": 2.0068,
      "step": 47318
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7946022748947144,
      "learning_rate": 0.0005396930647533179,
      "loss": 2.3962,
      "step": 47319
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.6814014911651611,
      "learning_rate": 0.0005396906048234549,
      "loss": 2.2483,
      "step": 47320
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7006391882896423,
      "learning_rate": 0.0005396881448490288,
      "loss": 2.172,
      "step": 47321
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7778957486152649,
      "learning_rate": 0.0005396856848300402,
      "loss": 2.4233,
      "step": 47322
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7372874617576599,
      "learning_rate": 0.0005396832247664895,
      "loss": 2.254,
      "step": 47323
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.6985418200492859,
      "learning_rate": 0.000539680764658377,
      "loss": 2.3033,
      "step": 47324
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.6969889998435974,
      "learning_rate": 0.0005396783045057036,
      "loss": 2.1945,
      "step": 47325
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7936925888061523,
      "learning_rate": 0.0005396758443084693,
      "loss": 2.1098,
      "step": 47326
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.651849627494812,
      "learning_rate": 0.0005396733840666748,
      "loss": 2.3082,
      "step": 47327
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.8043107390403748,
      "learning_rate": 0.0005396709237803205,
      "loss": 2.1164,
      "step": 47328
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7250389456748962,
      "learning_rate": 0.0005396684634494068,
      "loss": 2.2308,
      "step": 47329
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7665805816650391,
      "learning_rate": 0.0005396660030739342,
      "loss": 2.3047,
      "step": 47330
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.8537851572036743,
      "learning_rate": 0.0005396635426539033,
      "loss": 2.4937,
      "step": 47331
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.6553287506103516,
      "learning_rate": 0.0005396610821893143,
      "loss": 2.1495,
      "step": 47332
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.6900384426116943,
      "learning_rate": 0.0005396586216801678,
      "loss": 2.1623,
      "step": 47333
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7258108258247375,
      "learning_rate": 0.0005396561611264642,
      "loss": 2.1976,
      "step": 47334
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7613016366958618,
      "learning_rate": 0.000539653700528204,
      "loss": 2.2588,
      "step": 47335
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.70316082239151,
      "learning_rate": 0.0005396512398853877,
      "loss": 2.3672,
      "step": 47336
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.6983526349067688,
      "learning_rate": 0.0005396487791980157,
      "loss": 2.1702,
      "step": 47337
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.6856344938278198,
      "learning_rate": 0.0005396463184660885,
      "loss": 2.3387,
      "step": 47338
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7447001338005066,
      "learning_rate": 0.0005396438576896065,
      "loss": 2.0659,
      "step": 47339
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.8047009706497192,
      "learning_rate": 0.0005396413968685701,
      "loss": 2.3455,
      "step": 47340
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.6512965559959412,
      "learning_rate": 0.0005396389360029799,
      "loss": 2.1938,
      "step": 47341
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.8428917527198792,
      "learning_rate": 0.0005396364750928364,
      "loss": 2.1798,
      "step": 47342
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7607927918434143,
      "learning_rate": 0.0005396340141381399,
      "loss": 2.1712,
      "step": 47343
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.6369814872741699,
      "learning_rate": 0.0005396315531388909,
      "loss": 2.4243,
      "step": 47344
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7943565249443054,
      "learning_rate": 0.0005396290920950898,
      "loss": 2.2674,
      "step": 47345
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.8052994608879089,
      "learning_rate": 0.0005396266310067373,
      "loss": 2.0217,
      "step": 47346
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7374798655509949,
      "learning_rate": 0.0005396241698738335,
      "loss": 2.6345,
      "step": 47347
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.6938746571540833,
      "learning_rate": 0.0005396217086963791,
      "loss": 2.2214,
      "step": 47348
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.8864126801490784,
      "learning_rate": 0.0005396192474743745,
      "loss": 2.1904,
      "step": 47349
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.6915985941886902,
      "learning_rate": 0.0005396167862078202,
      "loss": 2.2012,
      "step": 47350
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.8059869408607483,
      "learning_rate": 0.0005396143248967166,
      "loss": 2.2665,
      "step": 47351
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7029668688774109,
      "learning_rate": 0.0005396118635410641,
      "loss": 2.1737,
      "step": 47352
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.6804673075675964,
      "learning_rate": 0.0005396094021408634,
      "loss": 2.2682,
      "step": 47353
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7696487903594971,
      "learning_rate": 0.0005396069406961146,
      "loss": 2.0363,
      "step": 47354
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7275354862213135,
      "learning_rate": 0.0005396044792068185,
      "loss": 1.9448,
      "step": 47355
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7633432745933533,
      "learning_rate": 0.0005396020176729753,
      "loss": 2.2393,
      "step": 47356
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7153149247169495,
      "learning_rate": 0.0005395995560945856,
      "loss": 2.2332,
      "step": 47357
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.6845450401306152,
      "learning_rate": 0.0005395970944716499,
      "loss": 2.2971,
      "step": 47358
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7116446495056152,
      "learning_rate": 0.0005395946328041685,
      "loss": 2.138,
      "step": 47359
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7148217558860779,
      "learning_rate": 0.0005395921710921418,
      "loss": 2.489,
      "step": 47360
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7100386023521423,
      "learning_rate": 0.0005395897093355706,
      "loss": 2.4454,
      "step": 47361
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.6857984662055969,
      "learning_rate": 0.0005395872475344551,
      "loss": 1.8721,
      "step": 47362
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.6580294370651245,
      "learning_rate": 0.0005395847856887958,
      "loss": 2.3421,
      "step": 47363
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.6899370551109314,
      "learning_rate": 0.0005395823237985931,
      "loss": 2.1305,
      "step": 47364
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7352292537689209,
      "learning_rate": 0.0005395798618638475,
      "loss": 2.7237,
      "step": 47365
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.6760072112083435,
      "learning_rate": 0.0005395773998845597,
      "loss": 2.2142,
      "step": 47366
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.6754968762397766,
      "learning_rate": 0.0005395749378607297,
      "loss": 2.1921,
      "step": 47367
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7063196897506714,
      "learning_rate": 0.0005395724757923583,
      "loss": 2.3676,
      "step": 47368
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.650508463382721,
      "learning_rate": 0.0005395700136794459,
      "loss": 2.3216,
      "step": 47369
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7432682514190674,
      "learning_rate": 0.0005395675515219928,
      "loss": 2.0873,
      "step": 47370
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.747613251209259,
      "learning_rate": 0.0005395650893199996,
      "loss": 2.3408,
      "step": 47371
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7044409513473511,
      "learning_rate": 0.0005395626270734668,
      "loss": 2.081,
      "step": 47372
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.6278873085975647,
      "learning_rate": 0.0005395601647823948,
      "loss": 2.4913,
      "step": 47373
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.6908448338508606,
      "learning_rate": 0.0005395577024467839,
      "loss": 2.3033,
      "step": 47374
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.8445104956626892,
      "learning_rate": 0.0005395552400666348,
      "loss": 2.5756,
      "step": 47375
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.6963281631469727,
      "learning_rate": 0.0005395527776419478,
      "loss": 2.3848,
      "step": 47376
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.8023238182067871,
      "learning_rate": 0.0005395503151727235,
      "loss": 2.2628,
      "step": 47377
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7039731740951538,
      "learning_rate": 0.0005395478526589622,
      "loss": 2.2472,
      "step": 47378
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.704922616481781,
      "learning_rate": 0.0005395453901006644,
      "loss": 2.1637,
      "step": 47379
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7171282172203064,
      "learning_rate": 0.0005395429274978308,
      "loss": 2.327,
      "step": 47380
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.6789330244064331,
      "learning_rate": 0.0005395404648504614,
      "loss": 2.3817,
      "step": 47381
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7691433429718018,
      "learning_rate": 0.0005395380021585571,
      "loss": 2.3271,
      "step": 47382
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.6894406080245972,
      "learning_rate": 0.000539535539422118,
      "loss": 2.2161,
      "step": 47383
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7016479969024658,
      "learning_rate": 0.0005395330766411448,
      "loss": 2.2776,
      "step": 47384
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7302258610725403,
      "learning_rate": 0.0005395306138156379,
      "loss": 2.2538,
      "step": 47385
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.9774817228317261,
      "learning_rate": 0.0005395281509455978,
      "loss": 2.3807,
      "step": 47386
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.6873440742492676,
      "learning_rate": 0.0005395256880310247,
      "loss": 2.3638,
      "step": 47387
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.0382109880447388,
      "learning_rate": 0.0005395232250719194,
      "loss": 2.2042,
      "step": 47388
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.9000463485717773,
      "learning_rate": 0.0005395207620682822,
      "loss": 2.0829,
      "step": 47389
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7130746841430664,
      "learning_rate": 0.0005395182990201136,
      "loss": 2.3663,
      "step": 47390
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.230956792831421,
      "learning_rate": 0.0005395158359274138,
      "loss": 2.3623,
      "step": 47391
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.8912383317947388,
      "learning_rate": 0.0005395133727901837,
      "loss": 2.1468,
      "step": 47392
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.677468478679657,
      "learning_rate": 0.0005395109096084235,
      "loss": 2.3355,
      "step": 47393
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.8134469389915466,
      "learning_rate": 0.0005395084463821337,
      "loss": 2.4578,
      "step": 47394
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7711706161499023,
      "learning_rate": 0.0005395059831113147,
      "loss": 2.4082,
      "step": 47395
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.8670631051063538,
      "learning_rate": 0.0005395035197959672,
      "loss": 2.2713,
      "step": 47396
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7051356434822083,
      "learning_rate": 0.0005395010564360913,
      "loss": 2.2883,
      "step": 47397
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7385806441307068,
      "learning_rate": 0.0005394985930316877,
      "loss": 2.5787,
      "step": 47398
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.9634677767753601,
      "learning_rate": 0.0005394961295827567,
      "loss": 2.3787,
      "step": 47399
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7979293465614319,
      "learning_rate": 0.000539493666089299,
      "loss": 2.1995,
      "step": 47400
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.663934588432312,
      "learning_rate": 0.0005394912025513148,
      "loss": 2.1902,
      "step": 47401
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7717431783676147,
      "learning_rate": 0.0005394887389688047,
      "loss": 2.5834,
      "step": 47402
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7855328917503357,
      "learning_rate": 0.0005394862753417691,
      "loss": 2.1263,
      "step": 47403
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7415040731430054,
      "learning_rate": 0.0005394838116702085,
      "loss": 2.4065,
      "step": 47404
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.6396051049232483,
      "learning_rate": 0.0005394813479541234,
      "loss": 2.4838,
      "step": 47405
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7221536040306091,
      "learning_rate": 0.0005394788841935141,
      "loss": 2.3592,
      "step": 47406
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.679815948009491,
      "learning_rate": 0.0005394764203883812,
      "loss": 2.1729,
      "step": 47407
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.9062364101409912,
      "learning_rate": 0.000539473956538725,
      "loss": 2.0874,
      "step": 47408
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7151018977165222,
      "learning_rate": 0.0005394714926445463,
      "loss": 2.2028,
      "step": 47409
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.8457064032554626,
      "learning_rate": 0.0005394690287058451,
      "loss": 2.4274,
      "step": 47410
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7183831930160522,
      "learning_rate": 0.0005394665647226222,
      "loss": 2.0847,
      "step": 47411
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7431687116622925,
      "learning_rate": 0.0005394641006948779,
      "loss": 2.305,
      "step": 47412
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7620176672935486,
      "learning_rate": 0.0005394616366226127,
      "loss": 2.2771,
      "step": 47413
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.6784026026725769,
      "learning_rate": 0.0005394591725058272,
      "loss": 1.9394,
      "step": 47414
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.8892937898635864,
      "learning_rate": 0.0005394567083445216,
      "loss": 2.2688,
      "step": 47415
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.9325964450836182,
      "learning_rate": 0.0005394542441386964,
      "loss": 2.1406,
      "step": 47416
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7863539457321167,
      "learning_rate": 0.0005394517798883523,
      "loss": 2.4911,
      "step": 47417
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.0141100883483887,
      "learning_rate": 0.0005394493155934895,
      "loss": 2.2572,
      "step": 47418
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.8464198708534241,
      "learning_rate": 0.0005394468512541086,
      "loss": 2.1267,
      "step": 47419
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7751665115356445,
      "learning_rate": 0.00053944438687021,
      "loss": 2.3273,
      "step": 47420
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.0588712692260742,
      "learning_rate": 0.0005394419224417942,
      "loss": 2.4179,
      "step": 47421
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.8459197878837585,
      "learning_rate": 0.0005394394579688616,
      "loss": 2.3118,
      "step": 47422
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7684988975524902,
      "learning_rate": 0.0005394369934514126,
      "loss": 2.4154,
      "step": 47423
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.6777157187461853,
      "learning_rate": 0.0005394345288894477,
      "loss": 2.3305,
      "step": 47424
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7741177082061768,
      "learning_rate": 0.0005394320642829676,
      "loss": 2.2414,
      "step": 47425
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7933813333511353,
      "learning_rate": 0.0005394295996319725,
      "loss": 2.1644,
      "step": 47426
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7426643371582031,
      "learning_rate": 0.0005394271349364628,
      "loss": 2.1154,
      "step": 47427
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.6785365343093872,
      "learning_rate": 0.0005394246701964393,
      "loss": 2.374,
      "step": 47428
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7260122299194336,
      "learning_rate": 0.000539422205411902,
      "loss": 2.3181,
      "step": 47429
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7589905261993408,
      "learning_rate": 0.0005394197405828516,
      "loss": 2.2622,
      "step": 47430
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.6571440696716309,
      "learning_rate": 0.0005394172757092888,
      "loss": 2.4085,
      "step": 47431
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.6796718239784241,
      "learning_rate": 0.0005394148107912136,
      "loss": 2.4766,
      "step": 47432
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.6757720112800598,
      "learning_rate": 0.0005394123458286267,
      "loss": 2.0833,
      "step": 47433
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.6884275674819946,
      "learning_rate": 0.0005394098808215286,
      "loss": 2.3789,
      "step": 47434
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7061168551445007,
      "learning_rate": 0.0005394074157699196,
      "loss": 2.1211,
      "step": 47435
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7254127264022827,
      "learning_rate": 0.0005394049506738003,
      "loss": 2.2843,
      "step": 47436
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.6792722940444946,
      "learning_rate": 0.0005394024855331711,
      "loss": 2.2792,
      "step": 47437
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7000517845153809,
      "learning_rate": 0.0005394000203480324,
      "loss": 2.2862,
      "step": 47438
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7112562656402588,
      "learning_rate": 0.0005393975551183847,
      "loss": 2.4711,
      "step": 47439
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.6701682209968567,
      "learning_rate": 0.0005393950898442285,
      "loss": 2.1557,
      "step": 47440
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.6303479075431824,
      "learning_rate": 0.0005393926245255643,
      "loss": 2.1637,
      "step": 47441
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7681673169136047,
      "learning_rate": 0.0005393901591623925,
      "loss": 2.4554,
      "step": 47442
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7629607319831848,
      "learning_rate": 0.0005393876937547134,
      "loss": 2.3948,
      "step": 47443
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.6449705958366394,
      "learning_rate": 0.0005393852283025278,
      "loss": 2.3946,
      "step": 47444
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7051197290420532,
      "learning_rate": 0.0005393827628058358,
      "loss": 2.0295,
      "step": 47445
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.6699883937835693,
      "learning_rate": 0.0005393802972646381,
      "loss": 1.8784,
      "step": 47446
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.682762861251831,
      "learning_rate": 0.0005393778316789351,
      "loss": 2.3089,
      "step": 47447
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.6839554905891418,
      "learning_rate": 0.0005393753660487273,
      "loss": 2.25,
      "step": 47448
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.6751677393913269,
      "learning_rate": 0.000539372900374015,
      "loss": 2.1759,
      "step": 47449
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.6435685157775879,
      "learning_rate": 0.0005393704346547988,
      "loss": 2.3124,
      "step": 47450
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.6642139554023743,
      "learning_rate": 0.0005393679688910791,
      "loss": 1.9471,
      "step": 47451
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7603597640991211,
      "learning_rate": 0.0005393655030828564,
      "loss": 2.0765,
      "step": 47452
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7516347765922546,
      "learning_rate": 0.0005393630372301311,
      "loss": 2.2949,
      "step": 47453
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.6622229218482971,
      "learning_rate": 0.0005393605713329037,
      "loss": 2.204,
      "step": 47454
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7689065337181091,
      "learning_rate": 0.0005393581053911747,
      "loss": 2.4016,
      "step": 47455
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.6962535977363586,
      "learning_rate": 0.0005393556394049444,
      "loss": 2.2497,
      "step": 47456
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.8460249900817871,
      "learning_rate": 0.0005393531733742134,
      "loss": 2.1796,
      "step": 47457
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7084128856658936,
      "learning_rate": 0.0005393507072989823,
      "loss": 2.0516,
      "step": 47458
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7383518218994141,
      "learning_rate": 0.0005393482411792512,
      "loss": 2.3293,
      "step": 47459
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.703889787197113,
      "learning_rate": 0.0005393457750150208,
      "loss": 2.4266,
      "step": 47460
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7557158470153809,
      "learning_rate": 0.0005393433088062915,
      "loss": 2.468,
      "step": 47461
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.716070294380188,
      "learning_rate": 0.0005393408425530637,
      "loss": 2.552,
      "step": 47462
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.6576438546180725,
      "learning_rate": 0.0005393383762553381,
      "loss": 2.1934,
      "step": 47463
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.9674762487411499,
      "learning_rate": 0.0005393359099131147,
      "loss": 2.0816,
      "step": 47464
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.6456344127655029,
      "learning_rate": 0.0005393334435263945,
      "loss": 2.119,
      "step": 47465
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.6372312307357788,
      "learning_rate": 0.0005393309770951775,
      "loss": 2.432,
      "step": 47466
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.8406158089637756,
      "learning_rate": 0.0005393285106194645,
      "loss": 2.0181,
      "step": 47467
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7531337141990662,
      "learning_rate": 0.0005393260440992557,
      "loss": 2.1015,
      "step": 47468
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.816205620765686,
      "learning_rate": 0.0005393235775345517,
      "loss": 2.2553,
      "step": 47469
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.70904141664505,
      "learning_rate": 0.000539321110925353,
      "loss": 2.2389,
      "step": 47470
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7825751900672913,
      "learning_rate": 0.00053931864427166,
      "loss": 2.3478,
      "step": 47471
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.6737815737724304,
      "learning_rate": 0.000539316177573473,
      "loss": 2.278,
      "step": 47472
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7888731360435486,
      "learning_rate": 0.0005393137108307927,
      "loss": 2.3094,
      "step": 47473
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7503556609153748,
      "learning_rate": 0.0005393112440436195,
      "loss": 2.4595,
      "step": 47474
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.6730144619941711,
      "learning_rate": 0.0005393087772119537,
      "loss": 2.1566,
      "step": 47475
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7820779085159302,
      "learning_rate": 0.000539306310335796,
      "loss": 2.2163,
      "step": 47476
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7314186692237854,
      "learning_rate": 0.0005393038434151467,
      "loss": 2.1756,
      "step": 47477
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.8248562812805176,
      "learning_rate": 0.0005393013764500063,
      "loss": 2.5865,
      "step": 47478
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7194589972496033,
      "learning_rate": 0.0005392989094403753,
      "loss": 2.0856,
      "step": 47479
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.8049576878547668,
      "learning_rate": 0.0005392964423862541,
      "loss": 2.2825,
      "step": 47480
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.6641057133674622,
      "learning_rate": 0.0005392939752876431,
      "loss": 2.1136,
      "step": 47481
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.729796290397644,
      "learning_rate": 0.000539291508144543,
      "loss": 2.1609,
      "step": 47482
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.6811248660087585,
      "learning_rate": 0.0005392890409569538,
      "loss": 1.9849,
      "step": 47483
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.6899553537368774,
      "learning_rate": 0.0005392865737248765,
      "loss": 2.1058,
      "step": 47484
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7367105484008789,
      "learning_rate": 0.0005392841064483112,
      "loss": 2.2595,
      "step": 47485
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.6690071225166321,
      "learning_rate": 0.0005392816391272586,
      "loss": 1.9974,
      "step": 47486
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7191110253334045,
      "learning_rate": 0.0005392791717617188,
      "loss": 2.3179,
      "step": 47487
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7859702110290527,
      "learning_rate": 0.0005392767043516926,
      "loss": 2.155,
      "step": 47488
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.8254308700561523,
      "learning_rate": 0.0005392742368971804,
      "loss": 2.3934,
      "step": 47489
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.6889240741729736,
      "learning_rate": 0.0005392717693981826,
      "loss": 2.1817,
      "step": 47490
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.749515950679779,
      "learning_rate": 0.0005392693018546995,
      "loss": 2.2418,
      "step": 47491
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.6856369376182556,
      "learning_rate": 0.0005392668342667319,
      "loss": 2.1591,
      "step": 47492
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.6810199022293091,
      "learning_rate": 0.00053926436663428,
      "loss": 2.2259,
      "step": 47493
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7379439473152161,
      "learning_rate": 0.0005392618989573444,
      "loss": 2.0898,
      "step": 47494
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7089071869850159,
      "learning_rate": 0.0005392594312359255,
      "loss": 2.3639,
      "step": 47495
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7725975513458252,
      "learning_rate": 0.0005392569634700236,
      "loss": 2.0583,
      "step": 47496
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.6568478345870972,
      "learning_rate": 0.0005392544956596395,
      "loss": 2.4272,
      "step": 47497
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7593003511428833,
      "learning_rate": 0.0005392520278047734,
      "loss": 2.3413,
      "step": 47498
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7318364977836609,
      "learning_rate": 0.0005392495599054258,
      "loss": 2.2871,
      "step": 47499
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.682682454586029,
      "learning_rate": 0.0005392470919615972,
      "loss": 2.1978,
      "step": 47500
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.9530633687973022,
      "learning_rate": 0.000539244623973288,
      "loss": 2.158,
      "step": 47501
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7968272566795349,
      "learning_rate": 0.0005392421559404988,
      "loss": 2.1676,
      "step": 47502
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7694467306137085,
      "learning_rate": 0.00053923968786323,
      "loss": 2.3572,
      "step": 47503
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.702643871307373,
      "learning_rate": 0.0005392372197414819,
      "loss": 2.0769,
      "step": 47504
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.6584604978561401,
      "learning_rate": 0.0005392347515752551,
      "loss": 2.2565,
      "step": 47505
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7203978300094604,
      "learning_rate": 0.0005392322833645502,
      "loss": 2.1241,
      "step": 47506
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.6908733248710632,
      "learning_rate": 0.0005392298151093674,
      "loss": 2.4111,
      "step": 47507
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7025145888328552,
      "learning_rate": 0.0005392273468097072,
      "loss": 2.639,
      "step": 47508
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7100827097892761,
      "learning_rate": 0.0005392248784655702,
      "loss": 2.1438,
      "step": 47509
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.642699658870697,
      "learning_rate": 0.0005392224100769568,
      "loss": 2.3528,
      "step": 47510
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.6610684990882874,
      "learning_rate": 0.0005392199416438674,
      "loss": 2.3552,
      "step": 47511
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.6652430891990662,
      "learning_rate": 0.0005392174731663025,
      "loss": 2.0194,
      "step": 47512
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.6700835824012756,
      "learning_rate": 0.0005392150046442625,
      "loss": 2.3892,
      "step": 47513
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.6834883689880371,
      "learning_rate": 0.000539212536077748,
      "loss": 2.1501,
      "step": 47514
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7933629155158997,
      "learning_rate": 0.0005392100674667593,
      "loss": 2.1961,
      "step": 47515
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7300044298171997,
      "learning_rate": 0.000539207598811297,
      "loss": 2.4753,
      "step": 47516
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.740039587020874,
      "learning_rate": 0.0005392051301113614,
      "loss": 2.2695,
      "step": 47517
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.6517267823219299,
      "learning_rate": 0.0005392026613669532,
      "loss": 2.3165,
      "step": 47518
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.693607747554779,
      "learning_rate": 0.0005392001925780727,
      "loss": 2.1819,
      "step": 47519
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.6805629730224609,
      "learning_rate": 0.0005391977237447202,
      "loss": 2.3622,
      "step": 47520
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.775799572467804,
      "learning_rate": 0.0005391952548668965,
      "loss": 2.3426,
      "step": 47521
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7015694975852966,
      "learning_rate": 0.0005391927859446017,
      "loss": 2.2139,
      "step": 47522
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.6866123676300049,
      "learning_rate": 0.0005391903169778366,
      "loss": 2.265,
      "step": 47523
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.6926783919334412,
      "learning_rate": 0.0005391878479666016,
      "loss": 2.0965,
      "step": 47524
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.6952455043792725,
      "learning_rate": 0.0005391853789108968,
      "loss": 2.1769,
      "step": 47525
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7278457880020142,
      "learning_rate": 0.000539182909810723,
      "loss": 2.1432,
      "step": 47526
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7693431377410889,
      "learning_rate": 0.0005391804406660807,
      "loss": 2.1826,
      "step": 47527
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.6933726072311401,
      "learning_rate": 0.0005391779714769702,
      "loss": 2.2521,
      "step": 47528
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.6754959225654602,
      "learning_rate": 0.0005391755022433919,
      "loss": 2.4401,
      "step": 47529
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.6129214763641357,
      "learning_rate": 0.0005391730329653465,
      "loss": 2.0491,
      "step": 47530
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.8824237585067749,
      "learning_rate": 0.0005391705636428342,
      "loss": 2.3312,
      "step": 47531
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7235530018806458,
      "learning_rate": 0.0005391680942758556,
      "loss": 2.3913,
      "step": 47532
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7169225811958313,
      "learning_rate": 0.0005391656248644112,
      "loss": 2.132,
      "step": 47533
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.697371244430542,
      "learning_rate": 0.0005391631554085013,
      "loss": 2.185,
      "step": 47534
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.8207804560661316,
      "learning_rate": 0.0005391606859081265,
      "loss": 2.13,
      "step": 47535
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7553054094314575,
      "learning_rate": 0.0005391582163632873,
      "loss": 2.3024,
      "step": 47536
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.6575946807861328,
      "learning_rate": 0.000539155746773984,
      "loss": 2.0857,
      "step": 47537
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.81545490026474,
      "learning_rate": 0.0005391532771402171,
      "loss": 2.5174,
      "step": 47538
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7797410488128662,
      "learning_rate": 0.0005391508074619872,
      "loss": 2.3976,
      "step": 47539
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7245855927467346,
      "learning_rate": 0.0005391483377392946,
      "loss": 2.3052,
      "step": 47540
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7282978892326355,
      "learning_rate": 0.0005391458679721397,
      "loss": 2.1319,
      "step": 47541
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7128398418426514,
      "learning_rate": 0.0005391433981605233,
      "loss": 2.2083,
      "step": 47542
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7866306900978088,
      "learning_rate": 0.0005391409283044454,
      "loss": 2.2719,
      "step": 47543
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7646054625511169,
      "learning_rate": 0.0005391384584039067,
      "loss": 2.4676,
      "step": 47544
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7719995975494385,
      "learning_rate": 0.0005391359884589078,
      "loss": 2.3873,
      "step": 47545
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.780762255191803,
      "learning_rate": 0.0005391335184694489,
      "loss": 2.4815,
      "step": 47546
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7722100615501404,
      "learning_rate": 0.0005391310484355306,
      "loss": 2.0928,
      "step": 47547
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.6857485771179199,
      "learning_rate": 0.0005391285783571532,
      "loss": 2.1959,
      "step": 47548
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.802718997001648,
      "learning_rate": 0.0005391261082343175,
      "loss": 2.3155,
      "step": 47549
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7394214868545532,
      "learning_rate": 0.0005391236380670237,
      "loss": 2.2872,
      "step": 47550
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.6453489661216736,
      "learning_rate": 0.0005391211678552722,
      "loss": 2.1353,
      "step": 47551
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.655202329158783,
      "learning_rate": 0.0005391186975990636,
      "loss": 2.1904,
      "step": 47552
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.6877284646034241,
      "learning_rate": 0.0005391162272983983,
      "loss": 2.3306,
      "step": 47553
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.6000182628631592,
      "learning_rate": 0.0005391137569532768,
      "loss": 2.3917,
      "step": 47554
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7219585180282593,
      "learning_rate": 0.0005391112865636995,
      "loss": 2.3517,
      "step": 47555
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7094311714172363,
      "learning_rate": 0.0005391088161296669,
      "loss": 2.2565,
      "step": 47556
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7042653560638428,
      "learning_rate": 0.0005391063456511796,
      "loss": 1.9842,
      "step": 47557
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.6766456961631775,
      "learning_rate": 0.0005391038751282378,
      "loss": 2.3043,
      "step": 47558
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.6760864853858948,
      "learning_rate": 0.000539101404560842,
      "loss": 2.2664,
      "step": 47559
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.663362443447113,
      "learning_rate": 0.0005390989339489929,
      "loss": 2.2027,
      "step": 47560
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7060986161231995,
      "learning_rate": 0.0005390964632926908,
      "loss": 2.4385,
      "step": 47561
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7007886171340942,
      "learning_rate": 0.000539093992591936,
      "loss": 2.326,
      "step": 47562
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7124809622764587,
      "learning_rate": 0.0005390915218467291,
      "loss": 2.3698,
      "step": 47563
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7621006965637207,
      "learning_rate": 0.0005390890510570707,
      "loss": 2.3627,
      "step": 47564
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.6910313963890076,
      "learning_rate": 0.0005390865802229611,
      "loss": 2.1863,
      "step": 47565
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.6592235565185547,
      "learning_rate": 0.0005390841093444007,
      "loss": 2.4522,
      "step": 47566
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.6610517501831055,
      "learning_rate": 0.0005390816384213901,
      "loss": 2.142,
      "step": 47567
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7236111760139465,
      "learning_rate": 0.0005390791674539299,
      "loss": 2.4971,
      "step": 47568
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.6992931962013245,
      "learning_rate": 0.0005390766964420201,
      "loss": 2.2449,
      "step": 47569
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7021885514259338,
      "learning_rate": 0.0005390742253856616,
      "loss": 2.3486,
      "step": 47570
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7281048893928528,
      "learning_rate": 0.0005390717542848546,
      "loss": 2.2239,
      "step": 47571
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7600398659706116,
      "learning_rate": 0.0005390692831395996,
      "loss": 2.3432,
      "step": 47572
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.6565827131271362,
      "learning_rate": 0.0005390668119498973,
      "loss": 2.2096,
      "step": 47573
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7659140229225159,
      "learning_rate": 0.000539064340715748,
      "loss": 2.4334,
      "step": 47574
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7295283079147339,
      "learning_rate": 0.0005390618694371519,
      "loss": 2.1141,
      "step": 47575
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.6887519955635071,
      "learning_rate": 0.0005390593981141096,
      "loss": 2.2093,
      "step": 47576
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7401912808418274,
      "learning_rate": 0.0005390569267466219,
      "loss": 2.009,
      "step": 47577
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7100705504417419,
      "learning_rate": 0.000539054455334689,
      "loss": 2.1752,
      "step": 47578
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.6268723011016846,
      "learning_rate": 0.0005390519838783112,
      "loss": 2.1859,
      "step": 47579
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7035090923309326,
      "learning_rate": 0.0005390495123774893,
      "loss": 2.2414,
      "step": 47580
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.6379181742668152,
      "learning_rate": 0.0005390470408322234,
      "loss": 2.2287,
      "step": 47581
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.6486665606498718,
      "learning_rate": 0.0005390445692425143,
      "loss": 2.243,
      "step": 47582
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.6644269227981567,
      "learning_rate": 0.0005390420976083622,
      "loss": 2.2185,
      "step": 47583
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.6640521883964539,
      "learning_rate": 0.0005390396259297677,
      "loss": 2.5428,
      "step": 47584
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.6813840270042419,
      "learning_rate": 0.0005390371542067313,
      "loss": 2.2507,
      "step": 47585
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.6643818020820618,
      "learning_rate": 0.0005390346824392532,
      "loss": 2.0365,
      "step": 47586
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.6340229511260986,
      "learning_rate": 0.0005390322106273341,
      "loss": 2.1508,
      "step": 47587
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.0755993127822876,
      "learning_rate": 0.0005390297387709744,
      "loss": 2.3013,
      "step": 47588
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.714304506778717,
      "learning_rate": 0.0005390272668701747,
      "loss": 2.0593,
      "step": 47589
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.8813723921775818,
      "learning_rate": 0.0005390247949249352,
      "loss": 2.316,
      "step": 47590
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.696071445941925,
      "learning_rate": 0.0005390223229352564,
      "loss": 2.2514,
      "step": 47591
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.8100925087928772,
      "learning_rate": 0.000539019850901139,
      "loss": 2.2778,
      "step": 47592
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7343260645866394,
      "learning_rate": 0.0005390173788225833,
      "loss": 2.0033,
      "step": 47593
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7472203969955444,
      "learning_rate": 0.0005390149066995896,
      "loss": 2.2558,
      "step": 47594
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7118469476699829,
      "learning_rate": 0.0005390124345321586,
      "loss": 2.2038,
      "step": 47595
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7481414675712585,
      "learning_rate": 0.0005390099623202907,
      "loss": 2.4014,
      "step": 47596
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.6898460984230042,
      "learning_rate": 0.0005390074900639863,
      "loss": 2.4545,
      "step": 47597
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.776668131351471,
      "learning_rate": 0.0005390050177632459,
      "loss": 2.1695,
      "step": 47598
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7294499278068542,
      "learning_rate": 0.00053900254541807,
      "loss": 2.2493,
      "step": 47599
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7192529439926147,
      "learning_rate": 0.000539000073028459,
      "loss": 2.4517,
      "step": 47600
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7019298672676086,
      "learning_rate": 0.0005389976005944134,
      "loss": 2.3873,
      "step": 47601
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.6980485320091248,
      "learning_rate": 0.0005389951281159336,
      "loss": 2.1647,
      "step": 47602
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7214747667312622,
      "learning_rate": 0.00053899265559302,
      "loss": 2.3543,
      "step": 47603
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7799333333969116,
      "learning_rate": 0.0005389901830256732,
      "loss": 2.4701,
      "step": 47604
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7621312737464905,
      "learning_rate": 0.0005389877104138937,
      "loss": 2.2367,
      "step": 47605
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.732391893863678,
      "learning_rate": 0.0005389852377576819,
      "loss": 1.9985,
      "step": 47606
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7262861132621765,
      "learning_rate": 0.0005389827650570381,
      "loss": 2.0141,
      "step": 47607
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.6841475963592529,
      "learning_rate": 0.0005389802923119629,
      "loss": 2.2668,
      "step": 47608
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7130717635154724,
      "learning_rate": 0.000538977819522457,
      "loss": 2.402,
      "step": 47609
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7538174390792847,
      "learning_rate": 0.0005389753466885203,
      "loss": 2.0883,
      "step": 47610
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7405424118041992,
      "learning_rate": 0.0005389728738101538,
      "loss": 2.4402,
      "step": 47611
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7315951585769653,
      "learning_rate": 0.0005389704008873575,
      "loss": 2.2102,
      "step": 47612
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7340880036354065,
      "learning_rate": 0.0005389679279201322,
      "loss": 2.1704,
      "step": 47613
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7861325740814209,
      "learning_rate": 0.0005389654549084784,
      "loss": 1.9116,
      "step": 47614
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7603403925895691,
      "learning_rate": 0.0005389629818523962,
      "loss": 2.3572,
      "step": 47615
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.6861973404884338,
      "learning_rate": 0.0005389605087518863,
      "loss": 2.091,
      "step": 47616
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.6473090648651123,
      "learning_rate": 0.0005389580356069492,
      "loss": 2.1597,
      "step": 47617
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.8139391541481018,
      "learning_rate": 0.0005389555624175853,
      "loss": 2.2125,
      "step": 47618
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.6892910003662109,
      "learning_rate": 0.000538953089183795,
      "loss": 2.3076,
      "step": 47619
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.783829927444458,
      "learning_rate": 0.0005389506159055789,
      "loss": 2.3149,
      "step": 47620
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.739152193069458,
      "learning_rate": 0.0005389481425829372,
      "loss": 2.1696,
      "step": 47621
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7386969327926636,
      "learning_rate": 0.0005389456692158707,
      "loss": 2.3293,
      "step": 47622
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.103466272354126,
      "learning_rate": 0.0005389431958043796,
      "loss": 2.177,
      "step": 47623
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7709661722183228,
      "learning_rate": 0.0005389407223484645,
      "loss": 2.1408,
      "step": 47624
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.1406021118164062,
      "learning_rate": 0.0005389382488481258,
      "loss": 2.2251,
      "step": 47625
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.9652073979377747,
      "learning_rate": 0.0005389357753033639,
      "loss": 2.147,
      "step": 47626
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7387517094612122,
      "learning_rate": 0.0005389333017141794,
      "loss": 2.293,
      "step": 47627
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.866954505443573,
      "learning_rate": 0.0005389308280805726,
      "loss": 2.1929,
      "step": 47628
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.0733405351638794,
      "learning_rate": 0.0005389283544025441,
      "loss": 2.2635,
      "step": 47629
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.8228845596313477,
      "learning_rate": 0.0005389258806800944,
      "loss": 1.9982,
      "step": 47630
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.750220000743866,
      "learning_rate": 0.0005389234069132238,
      "loss": 2.233,
      "step": 47631
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7732475399971008,
      "learning_rate": 0.0005389209331019328,
      "loss": 1.9956,
      "step": 47632
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.74916672706604,
      "learning_rate": 0.0005389184592462219,
      "loss": 2.2043,
      "step": 47633
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.6748642921447754,
      "learning_rate": 0.0005389159853460917,
      "loss": 1.9754,
      "step": 47634
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.8538085222244263,
      "learning_rate": 0.0005389135114015423,
      "loss": 2.2938,
      "step": 47635
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.6857236623764038,
      "learning_rate": 0.0005389110374125744,
      "loss": 2.2851,
      "step": 47636
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.6927763223648071,
      "learning_rate": 0.0005389085633791884,
      "loss": 2.1314,
      "step": 47637
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.707814633846283,
      "learning_rate": 0.0005389060893013849,
      "loss": 2.2249,
      "step": 47638
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.6380699872970581,
      "learning_rate": 0.0005389036151791642,
      "loss": 2.496,
      "step": 47639
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7223202586174011,
      "learning_rate": 0.0005389011410125268,
      "loss": 2.3941,
      "step": 47640
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.6648939251899719,
      "learning_rate": 0.0005388986668014732,
      "loss": 2.3908,
      "step": 47641
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7916499972343445,
      "learning_rate": 0.0005388961925460039,
      "loss": 2.2647,
      "step": 47642
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.6441022157669067,
      "learning_rate": 0.0005388937182461191,
      "loss": 2.1118,
      "step": 47643
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.6652364134788513,
      "learning_rate": 0.0005388912439018195,
      "loss": 2.1952,
      "step": 47644
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.6693872213363647,
      "learning_rate": 0.0005388887695131055,
      "loss": 2.2165,
      "step": 47645
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7286673784255981,
      "learning_rate": 0.0005388862950799777,
      "loss": 2.2134,
      "step": 47646
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.6962376832962036,
      "learning_rate": 0.0005388838206024364,
      "loss": 2.3342,
      "step": 47647
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7399526834487915,
      "learning_rate": 0.0005388813460804819,
      "loss": 2.1886,
      "step": 47648
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.698829174041748,
      "learning_rate": 0.0005388788715141151,
      "loss": 2.3924,
      "step": 47649
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.6924476623535156,
      "learning_rate": 0.000538876396903336,
      "loss": 2.4899,
      "step": 47650
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7004448175430298,
      "learning_rate": 0.0005388739222481454,
      "loss": 2.0813,
      "step": 47651
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7297046184539795,
      "learning_rate": 0.0005388714475485437,
      "loss": 2.2848,
      "step": 47652
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7676190733909607,
      "learning_rate": 0.0005388689728045311,
      "loss": 2.2737,
      "step": 47653
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.6966756582260132,
      "learning_rate": 0.0005388664980161083,
      "loss": 2.2912,
      "step": 47654
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7598403096199036,
      "learning_rate": 0.0005388640231832757,
      "loss": 2.1142,
      "step": 47655
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7037874460220337,
      "learning_rate": 0.0005388615483060339,
      "loss": 2.1996,
      "step": 47656
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.671966016292572,
      "learning_rate": 0.0005388590733843831,
      "loss": 2.2606,
      "step": 47657
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.8116848468780518,
      "learning_rate": 0.000538856598418324,
      "loss": 2.3253,
      "step": 47658
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.6610090136528015,
      "learning_rate": 0.0005388541234078568,
      "loss": 2.3367,
      "step": 47659
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.6763656139373779,
      "learning_rate": 0.0005388516483529822,
      "loss": 2.2189,
      "step": 47660
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7915821671485901,
      "learning_rate": 0.0005388491732537005,
      "loss": 2.1305,
      "step": 47661
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.847520649433136,
      "learning_rate": 0.0005388466981100123,
      "loss": 2.1502,
      "step": 47662
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.6594181060791016,
      "learning_rate": 0.000538844222921918,
      "loss": 2.038,
      "step": 47663
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.0461821556091309,
      "learning_rate": 0.0005388417476894181,
      "loss": 2.2838,
      "step": 47664
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.9237328767776489,
      "learning_rate": 0.0005388392724125128,
      "loss": 2.2756,
      "step": 47665
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7459109425544739,
      "learning_rate": 0.0005388367970912029,
      "loss": 2.2307,
      "step": 47666
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.9669155478477478,
      "learning_rate": 0.0005388343217254888,
      "loss": 2.338,
      "step": 47667
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.8978312611579895,
      "learning_rate": 0.0005388318463153708,
      "loss": 2.1869,
      "step": 47668
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.6509953141212463,
      "learning_rate": 0.0005388293708608495,
      "loss": 2.2376,
      "step": 47669
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7070625424385071,
      "learning_rate": 0.0005388268953619253,
      "loss": 2.3045,
      "step": 47670
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.1887749433517456,
      "learning_rate": 0.0005388244198185986,
      "loss": 2.3533,
      "step": 47671
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7360971570014954,
      "learning_rate": 0.00053882194423087,
      "loss": 2.0987,
      "step": 47672
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7692835330963135,
      "learning_rate": 0.00053881946859874,
      "loss": 2.326,
      "step": 47673
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7571901679039001,
      "learning_rate": 0.0005388169929222088,
      "loss": 2.4259,
      "step": 47674
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.9702666401863098,
      "learning_rate": 0.000538814517201277,
      "loss": 2.3604,
      "step": 47675
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7467765808105469,
      "learning_rate": 0.0005388120414359452,
      "loss": 2.2121,
      "step": 47676
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.6523159146308899,
      "learning_rate": 0.0005388095656262137,
      "loss": 2.2728,
      "step": 47677
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.0744681358337402,
      "learning_rate": 0.0005388070897720829,
      "loss": 2.3831,
      "step": 47678
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.67310631275177,
      "learning_rate": 0.0005388046138735534,
      "loss": 2.3048,
      "step": 47679
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.6553718447685242,
      "learning_rate": 0.0005388021379306256,
      "loss": 2.2119,
      "step": 47680
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.760680615901947,
      "learning_rate": 0.0005387996619433,
      "loss": 2.2155,
      "step": 47681
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7156012654304504,
      "learning_rate": 0.0005387971859115771,
      "loss": 2.4397,
      "step": 47682
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7489907145500183,
      "learning_rate": 0.0005387947098354572,
      "loss": 2.3208,
      "step": 47683
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.6675415635108948,
      "learning_rate": 0.0005387922337149408,
      "loss": 2.3694,
      "step": 47684
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.6556859612464905,
      "learning_rate": 0.0005387897575500286,
      "loss": 2.4432,
      "step": 47685
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.6498328447341919,
      "learning_rate": 0.0005387872813407208,
      "loss": 2.3082,
      "step": 47686
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.869991660118103,
      "learning_rate": 0.0005387848050870179,
      "loss": 2.1543,
      "step": 47687
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7832018733024597,
      "learning_rate": 0.0005387823287889204,
      "loss": 2.3099,
      "step": 47688
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.6378514766693115,
      "learning_rate": 0.0005387798524464287,
      "loss": 2.3148,
      "step": 47689
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7522411346435547,
      "learning_rate": 0.0005387773760595435,
      "loss": 2.3482,
      "step": 47690
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.045983076095581,
      "learning_rate": 0.0005387748996282649,
      "loss": 2.2689,
      "step": 47691
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7194831371307373,
      "learning_rate": 0.0005387724231525937,
      "loss": 2.1639,
      "step": 47692
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.846358060836792,
      "learning_rate": 0.0005387699466325301,
      "loss": 2.1506,
      "step": 47693
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7458698153495789,
      "learning_rate": 0.0005387674700680745,
      "loss": 2.2832,
      "step": 47694
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7452294230461121,
      "learning_rate": 0.0005387649934592277,
      "loss": 2.0686,
      "step": 47695
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7456879019737244,
      "learning_rate": 0.0005387625168059901,
      "loss": 2.423,
      "step": 47696
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.0131295919418335,
      "learning_rate": 0.0005387600401083618,
      "loss": 2.2929,
      "step": 47697
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.752543032169342,
      "learning_rate": 0.0005387575633663437,
      "loss": 2.0694,
      "step": 47698
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.6543457508087158,
      "learning_rate": 0.0005387550865799358,
      "loss": 2.1613,
      "step": 47699
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.8033189177513123,
      "learning_rate": 0.0005387526097491391,
      "loss": 2.2106,
      "step": 47700
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7569392323493958,
      "learning_rate": 0.0005387501328739537,
      "loss": 1.8622,
      "step": 47701
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7143566012382507,
      "learning_rate": 0.0005387476559543801,
      "loss": 2.2388,
      "step": 47702
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7252045273780823,
      "learning_rate": 0.0005387451789904188,
      "loss": 1.9483,
      "step": 47703
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.8550092577934265,
      "learning_rate": 0.0005387427019820703,
      "loss": 2.2685,
      "step": 47704
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.8962618708610535,
      "learning_rate": 0.000538740224929335,
      "loss": 2.3092,
      "step": 47705
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.8450567722320557,
      "learning_rate": 0.0005387377478322134,
      "loss": 2.3402,
      "step": 47706
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.8138571977615356,
      "learning_rate": 0.000538735270690706,
      "loss": 2.075,
      "step": 47707
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7178171277046204,
      "learning_rate": 0.0005387327935048132,
      "loss": 2.0438,
      "step": 47708
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.6532812714576721,
      "learning_rate": 0.0005387303162745354,
      "loss": 2.0858,
      "step": 47709
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7391273975372314,
      "learning_rate": 0.0005387278389998732,
      "loss": 2.0979,
      "step": 47710
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7882538437843323,
      "learning_rate": 0.0005387253616808269,
      "loss": 2.259,
      "step": 47711
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7105428576469421,
      "learning_rate": 0.000538722884317397,
      "loss": 2.3033,
      "step": 47712
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.8071495890617371,
      "learning_rate": 0.0005387204069095842,
      "loss": 2.3459,
      "step": 47713
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7193654179573059,
      "learning_rate": 0.0005387179294573885,
      "loss": 2.1058,
      "step": 47714
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.6512110829353333,
      "learning_rate": 0.000538715451960811,
      "loss": 2.22,
      "step": 47715
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.6889597177505493,
      "learning_rate": 0.0005387129744198514,
      "loss": 2.2993,
      "step": 47716
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7173628211021423,
      "learning_rate": 0.0005387104968345109,
      "loss": 2.1189,
      "step": 47717
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.6084082126617432,
      "learning_rate": 0.0005387080192047893,
      "loss": 2.1198,
      "step": 47718
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7008640766143799,
      "learning_rate": 0.0005387055415306876,
      "loss": 2.3048,
      "step": 47719
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.748375654220581,
      "learning_rate": 0.0005387030638122059,
      "loss": 2.4109,
      "step": 47720
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7288034558296204,
      "learning_rate": 0.0005387005860493449,
      "loss": 2.4055,
      "step": 47721
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7126214504241943,
      "learning_rate": 0.000538698108242105,
      "loss": 2.1517,
      "step": 47722
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.6891472339630127,
      "learning_rate": 0.0005386956303904864,
      "loss": 2.2672,
      "step": 47723
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7222043871879578,
      "learning_rate": 0.0005386931524944899,
      "loss": 2.5102,
      "step": 47724
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.758172333240509,
      "learning_rate": 0.0005386906745541159,
      "loss": 2.1459,
      "step": 47725
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.6921451091766357,
      "learning_rate": 0.0005386881965693647,
      "loss": 2.1876,
      "step": 47726
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.680718183517456,
      "learning_rate": 0.0005386857185402369,
      "loss": 2.0812,
      "step": 47727
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.8955540657043457,
      "learning_rate": 0.0005386832404667329,
      "loss": 2.2392,
      "step": 47728
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7487866878509521,
      "learning_rate": 0.0005386807623488531,
      "loss": 2.4687,
      "step": 47729
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7660711407661438,
      "learning_rate": 0.0005386782841865982,
      "loss": 2.2509,
      "step": 47730
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7143102884292603,
      "learning_rate": 0.0005386758059799685,
      "loss": 2.1763,
      "step": 47731
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7412593960762024,
      "learning_rate": 0.0005386733277289643,
      "loss": 2.1115,
      "step": 47732
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7069639563560486,
      "learning_rate": 0.0005386708494335864,
      "loss": 2.1355,
      "step": 47733
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.757575273513794,
      "learning_rate": 0.000538668371093835,
      "loss": 2.293,
      "step": 47734
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7464485168457031,
      "learning_rate": 0.0005386658927097106,
      "loss": 2.2781,
      "step": 47735
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7445665597915649,
      "learning_rate": 0.0005386634142812138,
      "loss": 2.3692,
      "step": 47736
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7711139917373657,
      "learning_rate": 0.0005386609358083448,
      "loss": 2.1268,
      "step": 47737
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7210028767585754,
      "learning_rate": 0.0005386584572911045,
      "loss": 2.4197,
      "step": 47738
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.8048333525657654,
      "learning_rate": 0.0005386559787294929,
      "loss": 2.1456,
      "step": 47739
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7758327126502991,
      "learning_rate": 0.0005386535001235107,
      "loss": 2.0529,
      "step": 47740
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7062515020370483,
      "learning_rate": 0.0005386510214731583,
      "loss": 2.1162,
      "step": 47741
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.6925951838493347,
      "learning_rate": 0.0005386485427784362,
      "loss": 2.1798,
      "step": 47742
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7208421230316162,
      "learning_rate": 0.0005386460640393447,
      "loss": 2.4258,
      "step": 47743
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.6543005704879761,
      "learning_rate": 0.0005386435852558844,
      "loss": 1.9698,
      "step": 47744
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.8639440536499023,
      "learning_rate": 0.000538641106428056,
      "loss": 2.2966,
      "step": 47745
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.751125156879425,
      "learning_rate": 0.0005386386275558595,
      "loss": 2.2758,
      "step": 47746
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7124252915382385,
      "learning_rate": 0.0005386361486392956,
      "loss": 2.2546,
      "step": 47747
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.6940580010414124,
      "learning_rate": 0.0005386336696783648,
      "loss": 1.9365,
      "step": 47748
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7179316282272339,
      "learning_rate": 0.0005386311906730674,
      "loss": 2.203,
      "step": 47749
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7784541845321655,
      "learning_rate": 0.0005386287116234041,
      "loss": 2.1492,
      "step": 47750
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.6784375309944153,
      "learning_rate": 0.0005386262325293752,
      "loss": 2.2599,
      "step": 47751
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7834284901618958,
      "learning_rate": 0.000538623753390981,
      "loss": 2.2546,
      "step": 47752
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.6761525273323059,
      "learning_rate": 0.0005386212742082223,
      "loss": 2.1987,
      "step": 47753
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7009780406951904,
      "learning_rate": 0.0005386187949810994,
      "loss": 2.2493,
      "step": 47754
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.768837571144104,
      "learning_rate": 0.0005386163157096126,
      "loss": 2.3153,
      "step": 47755
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.727364718914032,
      "learning_rate": 0.0005386138363937627,
      "loss": 2.153,
      "step": 47756
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7710139751434326,
      "learning_rate": 0.0005386113570335499,
      "loss": 2.3226,
      "step": 47757
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.6585003137588501,
      "learning_rate": 0.0005386088776289748,
      "loss": 2.1884,
      "step": 47758
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7207208275794983,
      "learning_rate": 0.0005386063981800377,
      "loss": 2.2854,
      "step": 47759
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7829113006591797,
      "learning_rate": 0.0005386039186867393,
      "loss": 2.3282,
      "step": 47760
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.6861773729324341,
      "learning_rate": 0.0005386014391490798,
      "loss": 2.2602,
      "step": 47761
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.6868209838867188,
      "learning_rate": 0.0005385989595670599,
      "loss": 2.1062,
      "step": 47762
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7239835858345032,
      "learning_rate": 0.0005385964799406799,
      "loss": 2.3378,
      "step": 47763
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.6905370354652405,
      "learning_rate": 0.0005385940002699403,
      "loss": 2.3222,
      "step": 47764
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.6965954303741455,
      "learning_rate": 0.0005385915205548416,
      "loss": 2.3197,
      "step": 47765
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7406200766563416,
      "learning_rate": 0.0005385890407953842,
      "loss": 2.6829,
      "step": 47766
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7123302817344666,
      "learning_rate": 0.0005385865609915686,
      "loss": 2.1667,
      "step": 47767
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7162466049194336,
      "learning_rate": 0.0005385840811433953,
      "loss": 2.1994,
      "step": 47768
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.675335705280304,
      "learning_rate": 0.0005385816012508647,
      "loss": 2.3197,
      "step": 47769
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7119290232658386,
      "learning_rate": 0.0005385791213139773,
      "loss": 2.3446,
      "step": 47770
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7509356737136841,
      "learning_rate": 0.0005385766413327335,
      "loss": 2.2403,
      "step": 47771
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.6955049633979797,
      "learning_rate": 0.0005385741613071338,
      "loss": 2.0317,
      "step": 47772
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7755587100982666,
      "learning_rate": 0.0005385716812371786,
      "loss": 2.3812,
      "step": 47773
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.6908864974975586,
      "learning_rate": 0.0005385692011228685,
      "loss": 2.1086,
      "step": 47774
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7119665741920471,
      "learning_rate": 0.0005385667209642038,
      "loss": 2.2089,
      "step": 47775
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.6593850255012512,
      "learning_rate": 0.0005385642407611851,
      "loss": 2.4039,
      "step": 47776
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.649093747138977,
      "learning_rate": 0.0005385617605138129,
      "loss": 2.2188,
      "step": 47777
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7002624869346619,
      "learning_rate": 0.0005385592802220875,
      "loss": 2.0688,
      "step": 47778
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.6964015960693359,
      "learning_rate": 0.0005385567998860094,
      "loss": 2.1988,
      "step": 47779
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.6555872559547424,
      "learning_rate": 0.000538554319505579,
      "loss": 2.2821,
      "step": 47780
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.703289270401001,
      "learning_rate": 0.000538551839080797,
      "loss": 2.1527,
      "step": 47781
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7993993163108826,
      "learning_rate": 0.0005385493586116635,
      "loss": 2.4254,
      "step": 47782
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.6785447597503662,
      "learning_rate": 0.0005385468780981794,
      "loss": 2.3236,
      "step": 47783
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7285563945770264,
      "learning_rate": 0.0005385443975403449,
      "loss": 2.4535,
      "step": 47784
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.8098021149635315,
      "learning_rate": 0.0005385419169381604,
      "loss": 2.3135,
      "step": 47785
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.735936164855957,
      "learning_rate": 0.0005385394362916265,
      "loss": 2.529,
      "step": 47786
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7389758229255676,
      "learning_rate": 0.0005385369556007437,
      "loss": 2.341,
      "step": 47787
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7166969180107117,
      "learning_rate": 0.0005385344748655123,
      "loss": 2.3935,
      "step": 47788
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.8365055322647095,
      "learning_rate": 0.0005385319940859328,
      "loss": 2.1988,
      "step": 47789
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.6734153032302856,
      "learning_rate": 0.0005385295132620056,
      "loss": 2.3189,
      "step": 47790
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.6698499917984009,
      "learning_rate": 0.0005385270323937315,
      "loss": 2.2795,
      "step": 47791
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7047507166862488,
      "learning_rate": 0.0005385245514811106,
      "loss": 2.4009,
      "step": 47792
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7094506621360779,
      "learning_rate": 0.0005385220705241434,
      "loss": 2.4104,
      "step": 47793
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7362462282180786,
      "learning_rate": 0.0005385195895228306,
      "loss": 2.0898,
      "step": 47794
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7120075225830078,
      "learning_rate": 0.0005385171084771725,
      "loss": 2.1851,
      "step": 47795
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7326950430870056,
      "learning_rate": 0.0005385146273871695,
      "loss": 2.1028,
      "step": 47796
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.9655389785766602,
      "learning_rate": 0.0005385121462528221,
      "loss": 2.2598,
      "step": 47797
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7205555438995361,
      "learning_rate": 0.0005385096650741308,
      "loss": 2.4012,
      "step": 47798
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.755534827709198,
      "learning_rate": 0.0005385071838510962,
      "loss": 2.1456,
      "step": 47799
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.8278211355209351,
      "learning_rate": 0.0005385047025837184,
      "loss": 2.3942,
      "step": 47800
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.655987560749054,
      "learning_rate": 0.0005385022212719983,
      "loss": 2.1232,
      "step": 47801
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7130791544914246,
      "learning_rate": 0.0005384997399159359,
      "loss": 2.0915,
      "step": 47802
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7140718698501587,
      "learning_rate": 0.000538497258515532,
      "loss": 2.3196,
      "step": 47803
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7134538292884827,
      "learning_rate": 0.000538494777070787,
      "loss": 2.1904,
      "step": 47804
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.8123939037322998,
      "learning_rate": 0.0005384922955817013,
      "loss": 2.1278,
      "step": 47805
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.77324378490448,
      "learning_rate": 0.0005384898140482754,
      "loss": 2.3495,
      "step": 47806
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.6471400260925293,
      "learning_rate": 0.0005384873324705096,
      "loss": 2.3689,
      "step": 47807
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.8768206238746643,
      "learning_rate": 0.0005384848508484046,
      "loss": 2.0856,
      "step": 47808
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7408009767532349,
      "learning_rate": 0.0005384823691819608,
      "loss": 2.2964,
      "step": 47809
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7759712934494019,
      "learning_rate": 0.0005384798874711786,
      "loss": 2.4738,
      "step": 47810
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.6831647753715515,
      "learning_rate": 0.0005384774057160585,
      "loss": 2.1812,
      "step": 47811
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.8406659364700317,
      "learning_rate": 0.0005384749239166008,
      "loss": 2.1843,
      "step": 47812
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.668912947177887,
      "learning_rate": 0.0005384724420728063,
      "loss": 2.144,
      "step": 47813
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7152033448219299,
      "learning_rate": 0.0005384699601846752,
      "loss": 1.9972,
      "step": 47814
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7481762766838074,
      "learning_rate": 0.0005384674782522081,
      "loss": 2.3122,
      "step": 47815
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.6650925874710083,
      "learning_rate": 0.0005384649962754054,
      "loss": 2.2206,
      "step": 47816
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7117534875869751,
      "learning_rate": 0.0005384625142542674,
      "loss": 2.3397,
      "step": 47817
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.6973717212677002,
      "learning_rate": 0.0005384600321887948,
      "loss": 2.2218,
      "step": 47818
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7099733948707581,
      "learning_rate": 0.000538457550078988,
      "loss": 2.1757,
      "step": 47819
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7467100620269775,
      "learning_rate": 0.0005384550679248474,
      "loss": 2.2489,
      "step": 47820
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.6770565509796143,
      "learning_rate": 0.0005384525857263736,
      "loss": 2.1948,
      "step": 47821
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7148637175559998,
      "learning_rate": 0.0005384501034835669,
      "loss": 2.5562,
      "step": 47822
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.6902004480361938,
      "learning_rate": 0.0005384476211964278,
      "loss": 2.4021,
      "step": 47823
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7520566582679749,
      "learning_rate": 0.0005384451388649568,
      "loss": 2.391,
      "step": 47824
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7570692896842957,
      "learning_rate": 0.0005384426564891544,
      "loss": 2.3293,
      "step": 47825
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7897074222564697,
      "learning_rate": 0.0005384401740690209,
      "loss": 2.5811,
      "step": 47826
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7093380689620972,
      "learning_rate": 0.0005384376916045569,
      "loss": 2.2053,
      "step": 47827
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.8248404860496521,
      "learning_rate": 0.0005384352090957629,
      "loss": 2.2568,
      "step": 47828
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.8745578527450562,
      "learning_rate": 0.0005384327265426392,
      "loss": 2.0239,
      "step": 47829
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.692619800567627,
      "learning_rate": 0.0005384302439451864,
      "loss": 2.0974,
      "step": 47830
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7318117022514343,
      "learning_rate": 0.0005384277613034049,
      "loss": 2.423,
      "step": 47831
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7522783875465393,
      "learning_rate": 0.0005384252786172954,
      "loss": 2.2286,
      "step": 47832
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7723643183708191,
      "learning_rate": 0.0005384227958868578,
      "loss": 2.3086,
      "step": 47833
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7757976055145264,
      "learning_rate": 0.0005384203131120932,
      "loss": 2.3265,
      "step": 47834
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7496877908706665,
      "learning_rate": 0.0005384178302930016,
      "loss": 2.3545,
      "step": 47835
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7959699630737305,
      "learning_rate": 0.0005384153474295836,
      "loss": 2.2779,
      "step": 47836
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.6894733905792236,
      "learning_rate": 0.0005384128645218397,
      "loss": 2.1636,
      "step": 47837
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7815635204315186,
      "learning_rate": 0.0005384103815697704,
      "loss": 2.315,
      "step": 47838
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.767356812953949,
      "learning_rate": 0.000538407898573376,
      "loss": 2.2552,
      "step": 47839
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7551903128623962,
      "learning_rate": 0.0005384054155326572,
      "loss": 1.9589,
      "step": 47840
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.8531439304351807,
      "learning_rate": 0.0005384029324476142,
      "loss": 2.0442,
      "step": 47841
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.6779876351356506,
      "learning_rate": 0.0005384004493182477,
      "loss": 2.3359,
      "step": 47842
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.8229594230651855,
      "learning_rate": 0.0005383979661445582,
      "loss": 2.143,
      "step": 47843
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.8135993480682373,
      "learning_rate": 0.0005383954829265458,
      "loss": 2.4317,
      "step": 47844
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7121075987815857,
      "learning_rate": 0.0005383929996642113,
      "loss": 2.3715,
      "step": 47845
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.6773037910461426,
      "learning_rate": 0.0005383905163575548,
      "loss": 2.4125,
      "step": 47846
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.6834444403648376,
      "learning_rate": 0.0005383880330065773,
      "loss": 2.1279,
      "step": 47847
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7159237861633301,
      "learning_rate": 0.0005383855496112789,
      "loss": 2.17,
      "step": 47848
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7019633054733276,
      "learning_rate": 0.00053838306617166,
      "loss": 2.3373,
      "step": 47849
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.8202184438705444,
      "learning_rate": 0.0005383805826877212,
      "loss": 2.0957,
      "step": 47850
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.8161994814872742,
      "learning_rate": 0.0005383780991594631,
      "loss": 2.5213,
      "step": 47851
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.8503976464271545,
      "learning_rate": 0.0005383756155868859,
      "loss": 2.2543,
      "step": 47852
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.6864861249923706,
      "learning_rate": 0.0005383731319699901,
      "loss": 2.3539,
      "step": 47853
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.9232320189476013,
      "learning_rate": 0.0005383706483087764,
      "loss": 2.355,
      "step": 47854
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.741744875907898,
      "learning_rate": 0.000538368164603245,
      "loss": 2.2487,
      "step": 47855
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.6985235810279846,
      "learning_rate": 0.0005383656808533966,
      "loss": 2.4149,
      "step": 47856
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.6977781057357788,
      "learning_rate": 0.0005383631970592313,
      "loss": 2.1699,
      "step": 47857
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7735164761543274,
      "learning_rate": 0.0005383607132207499,
      "loss": 2.5296,
      "step": 47858
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.799181342124939,
      "learning_rate": 0.0005383582293379528,
      "loss": 2.3261,
      "step": 47859
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7170060873031616,
      "learning_rate": 0.0005383557454108403,
      "loss": 2.3536,
      "step": 47860
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7778720259666443,
      "learning_rate": 0.000538353261439413,
      "loss": 2.3022,
      "step": 47861
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7396160364151001,
      "learning_rate": 0.0005383507774236713,
      "loss": 2.1863,
      "step": 47862
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.8152223229408264,
      "learning_rate": 0.0005383482933636158,
      "loss": 2.3299,
      "step": 47863
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.6256932616233826,
      "learning_rate": 0.0005383458092592467,
      "loss": 2.2798,
      "step": 47864
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7675406336784363,
      "learning_rate": 0.0005383433251105647,
      "loss": 2.0408,
      "step": 47865
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.8181597590446472,
      "learning_rate": 0.0005383408409175702,
      "loss": 2.2721,
      "step": 47866
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.689606249332428,
      "learning_rate": 0.0005383383566802636,
      "loss": 2.2078,
      "step": 47867
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.719051718711853,
      "learning_rate": 0.0005383358723986454,
      "loss": 2.1892,
      "step": 47868
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7660133242607117,
      "learning_rate": 0.0005383333880727161,
      "loss": 2.3969,
      "step": 47869
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.6910654902458191,
      "learning_rate": 0.0005383309037024761,
      "loss": 2.2611,
      "step": 47870
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.6677418947219849,
      "learning_rate": 0.0005383284192879258,
      "loss": 2.2361,
      "step": 47871
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.6851732134819031,
      "learning_rate": 0.0005383259348290659,
      "loss": 2.3331,
      "step": 47872
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7005308270454407,
      "learning_rate": 0.0005383234503258966,
      "loss": 2.1303,
      "step": 47873
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7598885297775269,
      "learning_rate": 0.0005383209657784186,
      "loss": 2.0223,
      "step": 47874
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.6996088027954102,
      "learning_rate": 0.0005383184811866321,
      "loss": 2.2464,
      "step": 47875
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7930610775947571,
      "learning_rate": 0.0005383159965505378,
      "loss": 2.2647,
      "step": 47876
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7395461201667786,
      "learning_rate": 0.000538313511870136,
      "loss": 2.2535,
      "step": 47877
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.6822978258132935,
      "learning_rate": 0.0005383110271454272,
      "loss": 2.3059,
      "step": 47878
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.6997304558753967,
      "learning_rate": 0.0005383085423764119,
      "loss": 2.2983,
      "step": 47879
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.8111201524734497,
      "learning_rate": 0.0005383060575630906,
      "loss": 2.2711,
      "step": 47880
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7885364890098572,
      "learning_rate": 0.0005383035727054635,
      "loss": 2.5877,
      "step": 47881
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7045928239822388,
      "learning_rate": 0.0005383010878035315,
      "loss": 2.1249,
      "step": 47882
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7135710716247559,
      "learning_rate": 0.0005382986028572948,
      "loss": 2.5572,
      "step": 47883
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7622220516204834,
      "learning_rate": 0.0005382961178667539,
      "loss": 2.3694,
      "step": 47884
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7871705293655396,
      "learning_rate": 0.0005382936328319091,
      "loss": 2.4044,
      "step": 47885
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.6829351186752319,
      "learning_rate": 0.0005382911477527611,
      "loss": 2.3067,
      "step": 47886
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7823113203048706,
      "learning_rate": 0.0005382886626293103,
      "loss": 2.3995,
      "step": 47887
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.8486976027488708,
      "learning_rate": 0.0005382861774615571,
      "loss": 2.4293,
      "step": 47888
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7930685877799988,
      "learning_rate": 0.000538283692249502,
      "loss": 2.2308,
      "step": 47889
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.8290566802024841,
      "learning_rate": 0.0005382812069931455,
      "loss": 2.5235,
      "step": 47890
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7339546084403992,
      "learning_rate": 0.0005382787216924881,
      "loss": 2.1741,
      "step": 47891
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.8721022605895996,
      "learning_rate": 0.0005382762363475301,
      "loss": 2.4603,
      "step": 47892
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.745608925819397,
      "learning_rate": 0.000538273750958272,
      "loss": 2.0279,
      "step": 47893
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.6873564720153809,
      "learning_rate": 0.0005382712655247144,
      "loss": 2.4373,
      "step": 47894
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7925155162811279,
      "learning_rate": 0.0005382687800468577,
      "loss": 2.1208,
      "step": 47895
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7261807918548584,
      "learning_rate": 0.0005382662945247022,
      "loss": 2.2643,
      "step": 47896
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.8589199185371399,
      "learning_rate": 0.0005382638089582487,
      "loss": 2.3022,
      "step": 47897
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.8106638193130493,
      "learning_rate": 0.0005382613233474972,
      "loss": 2.1402,
      "step": 47898
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.6610816121101379,
      "learning_rate": 0.0005382588376924485,
      "loss": 2.4756,
      "step": 47899
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.890651285648346,
      "learning_rate": 0.0005382563519931032,
      "loss": 2.4193,
      "step": 47900
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.686908483505249,
      "learning_rate": 0.0005382538662494613,
      "loss": 2.1893,
      "step": 47901
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.8517360091209412,
      "learning_rate": 0.0005382513804615236,
      "loss": 2.2557,
      "step": 47902
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.8002977967262268,
      "learning_rate": 0.0005382488946292904,
      "loss": 2.3698,
      "step": 47903
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.65175461769104,
      "learning_rate": 0.0005382464087527624,
      "loss": 2.3084,
      "step": 47904
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7968487739562988,
      "learning_rate": 0.0005382439228319398,
      "loss": 2.3964,
      "step": 47905
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7235867381095886,
      "learning_rate": 0.0005382414368668231,
      "loss": 2.3853,
      "step": 47906
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.6731414794921875,
      "learning_rate": 0.0005382389508574129,
      "loss": 2.4239,
      "step": 47907
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.6957719326019287,
      "learning_rate": 0.0005382364648037096,
      "loss": 1.8937,
      "step": 47908
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.8734566569328308,
      "learning_rate": 0.0005382339787057136,
      "loss": 2.4335,
      "step": 47909
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.6816579699516296,
      "learning_rate": 0.0005382314925634254,
      "loss": 2.3917,
      "step": 47910
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.6848872900009155,
      "learning_rate": 0.0005382290063768454,
      "loss": 2.3358,
      "step": 47911
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7604688405990601,
      "learning_rate": 0.0005382265201459743,
      "loss": 2.3774,
      "step": 47912
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.6508193612098694,
      "learning_rate": 0.0005382240338708123,
      "loss": 2.4363,
      "step": 47913
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7137274146080017,
      "learning_rate": 0.0005382215475513599,
      "loss": 2.2458,
      "step": 47914
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7967737913131714,
      "learning_rate": 0.0005382190611876176,
      "loss": 2.252,
      "step": 47915
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.6393994688987732,
      "learning_rate": 0.000538216574779586,
      "loss": 2.3251,
      "step": 47916
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7875162363052368,
      "learning_rate": 0.0005382140883272654,
      "loss": 2.0591,
      "step": 47917
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7674839496612549,
      "learning_rate": 0.0005382116018306563,
      "loss": 2.3383,
      "step": 47918
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.6890183091163635,
      "learning_rate": 0.0005382091152897592,
      "loss": 2.259,
      "step": 47919
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.6861823797225952,
      "learning_rate": 0.0005382066287045745,
      "loss": 2.2069,
      "step": 47920
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.6997016072273254,
      "learning_rate": 0.0005382041420751027,
      "loss": 2.2843,
      "step": 47921
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7310020327568054,
      "learning_rate": 0.0005382016554013443,
      "loss": 2.3103,
      "step": 47922
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.6852651238441467,
      "learning_rate": 0.0005381991686832996,
      "loss": 2.2294,
      "step": 47923
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.6936535239219666,
      "learning_rate": 0.0005381966819209692,
      "loss": 2.1996,
      "step": 47924
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7057925462722778,
      "learning_rate": 0.0005381941951143536,
      "loss": 2.0857,
      "step": 47925
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7976027131080627,
      "learning_rate": 0.0005381917082634531,
      "loss": 2.3235,
      "step": 47926
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7860084772109985,
      "learning_rate": 0.0005381892213682684,
      "loss": 2.3419,
      "step": 47927
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.6956639885902405,
      "learning_rate": 0.0005381867344287997,
      "loss": 2.0147,
      "step": 47928
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7652022242546082,
      "learning_rate": 0.0005381842474450477,
      "loss": 2.294,
      "step": 47929
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7478209733963013,
      "learning_rate": 0.0005381817604170127,
      "loss": 2.1055,
      "step": 47930
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.721237301826477,
      "learning_rate": 0.0005381792733446953,
      "loss": 2.4498,
      "step": 47931
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7301779389381409,
      "learning_rate": 0.0005381767862280958,
      "loss": 1.9531,
      "step": 47932
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7229287028312683,
      "learning_rate": 0.0005381742990672147,
      "loss": 2.1552,
      "step": 47933
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.6976489424705505,
      "learning_rate": 0.0005381718118620526,
      "loss": 2.4582,
      "step": 47934
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7211489677429199,
      "learning_rate": 0.0005381693246126097,
      "loss": 2.3389,
      "step": 47935
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7641828656196594,
      "learning_rate": 0.0005381668373188867,
      "loss": 2.1997,
      "step": 47936
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.8291428089141846,
      "learning_rate": 0.0005381643499808842,
      "loss": 2.142,
      "step": 47937
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.6814629435539246,
      "learning_rate": 0.0005381618625986021,
      "loss": 2.2756,
      "step": 47938
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.6587422490119934,
      "learning_rate": 0.0005381593751720415,
      "loss": 2.1739,
      "step": 47939
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.6529449224472046,
      "learning_rate": 0.0005381568877012025,
      "loss": 2.1917,
      "step": 47940
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.6666479110717773,
      "learning_rate": 0.0005381544001860855,
      "loss": 2.2645,
      "step": 47941
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.6519051194190979,
      "learning_rate": 0.0005381519126266913,
      "loss": 2.3035,
      "step": 47942
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7304227948188782,
      "learning_rate": 0.0005381494250230201,
      "loss": 2.2273,
      "step": 47943
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.6715197563171387,
      "learning_rate": 0.0005381469373750723,
      "loss": 2.3674,
      "step": 47944
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7252240777015686,
      "learning_rate": 0.0005381444496828486,
      "loss": 2.308,
      "step": 47945
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.6768603324890137,
      "learning_rate": 0.0005381419619463493,
      "loss": 2.4211,
      "step": 47946
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.6982229948043823,
      "learning_rate": 0.0005381394741655751,
      "loss": 2.1683,
      "step": 47947
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.6930302381515503,
      "learning_rate": 0.000538136986340526,
      "loss": 2.3814,
      "step": 47948
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.700530469417572,
      "learning_rate": 0.0005381344984712029,
      "loss": 2.3476,
      "step": 47949
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7546276450157166,
      "learning_rate": 0.0005381320105576061,
      "loss": 2.1856,
      "step": 47950
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7022258043289185,
      "learning_rate": 0.000538129522599736,
      "loss": 2.2958,
      "step": 47951
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7219752669334412,
      "learning_rate": 0.0005381270345975931,
      "loss": 2.2999,
      "step": 47952
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7938530445098877,
      "learning_rate": 0.0005381245465511781,
      "loss": 2.2505,
      "step": 47953
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7599708437919617,
      "learning_rate": 0.0005381220584604909,
      "loss": 2.3104,
      "step": 47954
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7418441772460938,
      "learning_rate": 0.0005381195703255327,
      "loss": 2.1974,
      "step": 47955
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7807738780975342,
      "learning_rate": 0.0005381170821463033,
      "loss": 1.9323,
      "step": 47956
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.6990248560905457,
      "learning_rate": 0.0005381145939228035,
      "loss": 2.2407,
      "step": 47957
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7149854302406311,
      "learning_rate": 0.0005381121056550338,
      "loss": 2.4222,
      "step": 47958
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.6284695863723755,
      "learning_rate": 0.0005381096173429945,
      "loss": 2.116,
      "step": 47959
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.6863368153572083,
      "learning_rate": 0.0005381071289866861,
      "loss": 2.162,
      "step": 47960
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.8069948554039001,
      "learning_rate": 0.0005381046405861091,
      "loss": 2.0261,
      "step": 47961
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.6426630020141602,
      "learning_rate": 0.0005381021521412639,
      "loss": 2.2159,
      "step": 47962
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.8686776757240295,
      "learning_rate": 0.0005380996636521512,
      "loss": 2.1781,
      "step": 47963
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.747908353805542,
      "learning_rate": 0.0005380971751187711,
      "loss": 2.2886,
      "step": 47964
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7122995853424072,
      "learning_rate": 0.0005380946865411243,
      "loss": 2.1299,
      "step": 47965
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7072219252586365,
      "learning_rate": 0.0005380921979192113,
      "loss": 2.1009,
      "step": 47966
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.8653513789176941,
      "learning_rate": 0.0005380897092530323,
      "loss": 2.3083,
      "step": 47967
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.6598172783851624,
      "learning_rate": 0.0005380872205425881,
      "loss": 2.2429,
      "step": 47968
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7171536684036255,
      "learning_rate": 0.0005380847317878789,
      "loss": 2.089,
      "step": 47969
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7454169392585754,
      "learning_rate": 0.0005380822429889052,
      "loss": 2.25,
      "step": 47970
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7098622918128967,
      "learning_rate": 0.0005380797541456676,
      "loss": 2.2608,
      "step": 47971
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.8441593647003174,
      "learning_rate": 0.0005380772652581664,
      "loss": 2.5266,
      "step": 47972
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.6989297270774841,
      "learning_rate": 0.0005380747763264022,
      "loss": 2.5155,
      "step": 47973
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7996929287910461,
      "learning_rate": 0.0005380722873503754,
      "loss": 2.0953,
      "step": 47974
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7129692435264587,
      "learning_rate": 0.0005380697983300865,
      "loss": 2.2935,
      "step": 47975
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7198497653007507,
      "learning_rate": 0.0005380673092655359,
      "loss": 2.3787,
      "step": 47976
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.8198608756065369,
      "learning_rate": 0.000538064820156724,
      "loss": 2.1663,
      "step": 47977
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.8484891057014465,
      "learning_rate": 0.0005380623310036514,
      "loss": 2.0685,
      "step": 47978
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.703696608543396,
      "learning_rate": 0.0005380598418063186,
      "loss": 2.2803,
      "step": 47979
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7051458954811096,
      "learning_rate": 0.0005380573525647259,
      "loss": 2.434,
      "step": 47980
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.8553736209869385,
      "learning_rate": 0.0005380548632788739,
      "loss": 2.2747,
      "step": 47981
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.819490373134613,
      "learning_rate": 0.000538052373948763,
      "loss": 2.3432,
      "step": 47982
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.6917182207107544,
      "learning_rate": 0.0005380498845743936,
      "loss": 2.107,
      "step": 47983
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.681094765663147,
      "learning_rate": 0.0005380473951557664,
      "loss": 2.2623,
      "step": 47984
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.8247582316398621,
      "learning_rate": 0.0005380449056928816,
      "loss": 2.5896,
      "step": 47985
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.8796190023422241,
      "learning_rate": 0.0005380424161857397,
      "loss": 2.4741,
      "step": 47986
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7356824278831482,
      "learning_rate": 0.0005380399266343413,
      "loss": 2.071,
      "step": 47987
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.0229572057724,
      "learning_rate": 0.0005380374370386867,
      "loss": 2.245,
      "step": 47988
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7062954306602478,
      "learning_rate": 0.0005380349473987766,
      "loss": 2.1445,
      "step": 47989
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.80293208360672,
      "learning_rate": 0.0005380324577146112,
      "loss": 2.074,
      "step": 47990
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.8552610874176025,
      "learning_rate": 0.0005380299679861911,
      "loss": 2.2523,
      "step": 47991
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.9124740958213806,
      "learning_rate": 0.0005380274782135167,
      "loss": 2.524,
      "step": 47992
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.6886508464813232,
      "learning_rate": 0.0005380249883965884,
      "loss": 2.4872,
      "step": 47993
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.8017407655715942,
      "learning_rate": 0.000538022498535407,
      "loss": 2.2845,
      "step": 47994
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7562505006790161,
      "learning_rate": 0.0005380200086299726,
      "loss": 2.3102,
      "step": 47995
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7027673721313477,
      "learning_rate": 0.0005380175186802858,
      "loss": 2.3826,
      "step": 47996
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7707659006118774,
      "learning_rate": 0.000538015028686347,
      "loss": 2.0915,
      "step": 47997
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7380536198616028,
      "learning_rate": 0.0005380125386481568,
      "loss": 2.1017,
      "step": 47998
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7278327345848083,
      "learning_rate": 0.0005380100485657155,
      "loss": 2.3183,
      "step": 47999
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.6671949028968811,
      "learning_rate": 0.0005380075584390236,
      "loss": 2.2826,
      "step": 48000
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.829494833946228,
      "learning_rate": 0.0005380050682680818,
      "loss": 2.2059,
      "step": 48001
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.655063807964325,
      "learning_rate": 0.0005380025780528903,
      "loss": 2.3139,
      "step": 48002
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7919803857803345,
      "learning_rate": 0.0005380000877934494,
      "loss": 2.4618,
      "step": 48003
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.6758016347885132,
      "learning_rate": 0.00053799759748976,
      "loss": 2.087,
      "step": 48004
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7044278979301453,
      "learning_rate": 0.0005379951071418223,
      "loss": 2.4443,
      "step": 48005
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.659094512462616,
      "learning_rate": 0.0005379926167496369,
      "loss": 2.2421,
      "step": 48006
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7483286261558533,
      "learning_rate": 0.0005379901263132041,
      "loss": 2.4035,
      "step": 48007
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.6900240182876587,
      "learning_rate": 0.0005379876358325244,
      "loss": 2.314,
      "step": 48008
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.695501983165741,
      "learning_rate": 0.0005379851453075984,
      "loss": 2.5416,
      "step": 48009
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7250109910964966,
      "learning_rate": 0.0005379826547384265,
      "loss": 2.3429,
      "step": 48010
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.6751247644424438,
      "learning_rate": 0.000537980164125009,
      "loss": 2.2428,
      "step": 48011
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.789749026298523,
      "learning_rate": 0.0005379776734673465,
      "loss": 2.2164,
      "step": 48012
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7798411846160889,
      "learning_rate": 0.0005379751827654395,
      "loss": 2.371,
      "step": 48013
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7156492471694946,
      "learning_rate": 0.0005379726920192885,
      "loss": 2.1392,
      "step": 48014
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.821927011013031,
      "learning_rate": 0.0005379702012288938,
      "loss": 2.2443,
      "step": 48015
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.6856947541236877,
      "learning_rate": 0.000537967710394256,
      "loss": 2.2119,
      "step": 48016
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7274891138076782,
      "learning_rate": 0.0005379652195153756,
      "loss": 2.1629,
      "step": 48017
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.9758622646331787,
      "learning_rate": 0.0005379627285922526,
      "loss": 2.026,
      "step": 48018
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.701362669467926,
      "learning_rate": 0.0005379602376248882,
      "loss": 2.0507,
      "step": 48019
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7250381112098694,
      "learning_rate": 0.0005379577466132824,
      "loss": 2.4138,
      "step": 48020
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.8141327500343323,
      "learning_rate": 0.0005379552555574357,
      "loss": 2.154,
      "step": 48021
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7986637949943542,
      "learning_rate": 0.0005379527644573487,
      "loss": 2.1487,
      "step": 48022
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.719256579875946,
      "learning_rate": 0.0005379502733130218,
      "loss": 2.5456,
      "step": 48023
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.6967686414718628,
      "learning_rate": 0.0005379477821244554,
      "loss": 2.3211,
      "step": 48024
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.8068724274635315,
      "learning_rate": 0.00053794529089165,
      "loss": 2.4914,
      "step": 48025
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.778792142868042,
      "learning_rate": 0.0005379427996146061,
      "loss": 2.1844,
      "step": 48026
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7112554907798767,
      "learning_rate": 0.0005379403082933242,
      "loss": 2.2774,
      "step": 48027
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.9036374688148499,
      "learning_rate": 0.0005379378169278048,
      "loss": 2.6435,
      "step": 48028
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7564917206764221,
      "learning_rate": 0.000537935325518048,
      "loss": 2.1197,
      "step": 48029
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7169679999351501,
      "learning_rate": 0.0005379328340640547,
      "loss": 2.34,
      "step": 48030
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.8824545741081238,
      "learning_rate": 0.0005379303425658251,
      "loss": 2.418,
      "step": 48031
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7446458339691162,
      "learning_rate": 0.0005379278510233599,
      "loss": 2.1753,
      "step": 48032
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.6768729090690613,
      "learning_rate": 0.0005379253594366592,
      "loss": 2.1146,
      "step": 48033
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.0163302421569824,
      "learning_rate": 0.0005379228678057239,
      "loss": 2.1236,
      "step": 48034
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7156599760055542,
      "learning_rate": 0.0005379203761305541,
      "loss": 2.4918,
      "step": 48035
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.6869760751724243,
      "learning_rate": 0.0005379178844111506,
      "loss": 2.3585,
      "step": 48036
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.1662957668304443,
      "learning_rate": 0.0005379153926475135,
      "loss": 2.1193,
      "step": 48037
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7887230515480042,
      "learning_rate": 0.0005379129008396435,
      "loss": 2.1517,
      "step": 48038
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.6570448279380798,
      "learning_rate": 0.0005379104089875409,
      "loss": 2.3547,
      "step": 48039
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.0358469486236572,
      "learning_rate": 0.0005379079170912064,
      "loss": 2.3034,
      "step": 48040
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7899594902992249,
      "learning_rate": 0.0005379054251506404,
      "loss": 2.2135,
      "step": 48041
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7842464447021484,
      "learning_rate": 0.000537902933165843,
      "loss": 2.2262,
      "step": 48042
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7209431529045105,
      "learning_rate": 0.0005379004411368152,
      "loss": 2.1161,
      "step": 48043
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7858969569206238,
      "learning_rate": 0.0005378979490635572,
      "loss": 2.1013,
      "step": 48044
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7148969173431396,
      "learning_rate": 0.0005378954569460693,
      "loss": 2.3958,
      "step": 48045
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.6306449174880981,
      "learning_rate": 0.0005378929647843523,
      "loss": 2.2609,
      "step": 48046
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.8016672730445862,
      "learning_rate": 0.0005378904725784065,
      "loss": 2.2364,
      "step": 48047
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.6877157092094421,
      "learning_rate": 0.0005378879803282324,
      "loss": 2.3287,
      "step": 48048
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.6924954056739807,
      "learning_rate": 0.0005378854880338304,
      "loss": 2.0239,
      "step": 48049
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.648063600063324,
      "learning_rate": 0.0005378829956952009,
      "loss": 2.088,
      "step": 48050
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7096579074859619,
      "learning_rate": 0.0005378805033123446,
      "loss": 2.238,
      "step": 48051
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7258210778236389,
      "learning_rate": 0.0005378780108852618,
      "loss": 2.2847,
      "step": 48052
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.6696073412895203,
      "learning_rate": 0.000537875518413953,
      "loss": 2.0229,
      "step": 48053
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.714134931564331,
      "learning_rate": 0.0005378730258984185,
      "loss": 2.2315,
      "step": 48054
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.6658422350883484,
      "learning_rate": 0.000537870533338659,
      "loss": 2.191,
      "step": 48055
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.6661248207092285,
      "learning_rate": 0.0005378680407346749,
      "loss": 2.4121,
      "step": 48056
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7540587782859802,
      "learning_rate": 0.0005378655480864666,
      "loss": 2.3151,
      "step": 48057
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.648231029510498,
      "learning_rate": 0.0005378630553940347,
      "loss": 2.3141,
      "step": 48058
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.6233908534049988,
      "learning_rate": 0.0005378605626573795,
      "loss": 2.1117,
      "step": 48059
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.6980363726615906,
      "learning_rate": 0.0005378580698765015,
      "loss": 2.0573,
      "step": 48060
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7103904485702515,
      "learning_rate": 0.0005378555770514013,
      "loss": 2.1757,
      "step": 48061
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.6569483280181885,
      "learning_rate": 0.0005378530841820791,
      "loss": 2.1198,
      "step": 48062
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.6548816561698914,
      "learning_rate": 0.0005378505912685356,
      "loss": 2.182,
      "step": 48063
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.6343979239463806,
      "learning_rate": 0.0005378480983107714,
      "loss": 2.2079,
      "step": 48064
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7445253729820251,
      "learning_rate": 0.0005378456053087865,
      "loss": 2.4045,
      "step": 48065
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7540557384490967,
      "learning_rate": 0.0005378431122625816,
      "loss": 2.0492,
      "step": 48066
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.6550554633140564,
      "learning_rate": 0.0005378406191721572,
      "loss": 2.1608,
      "step": 48067
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7417587041854858,
      "learning_rate": 0.0005378381260375138,
      "loss": 2.1983,
      "step": 48068
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7179747223854065,
      "learning_rate": 0.0005378356328586518,
      "loss": 2.2564,
      "step": 48069
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7182890176773071,
      "learning_rate": 0.0005378331396355716,
      "loss": 2.3663,
      "step": 48070
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.6982024312019348,
      "learning_rate": 0.0005378306463682738,
      "loss": 2.2302,
      "step": 48071
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.8233548998832703,
      "learning_rate": 0.0005378281530567587,
      "loss": 2.2065,
      "step": 48072
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7207088470458984,
      "learning_rate": 0.000537825659701027,
      "loss": 2.4513,
      "step": 48073
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7935565710067749,
      "learning_rate": 0.0005378231663010789,
      "loss": 2.3971,
      "step": 48074
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.738532543182373,
      "learning_rate": 0.0005378206728569151,
      "loss": 2.1955,
      "step": 48075
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7127077579498291,
      "learning_rate": 0.0005378181793685359,
      "loss": 2.1752,
      "step": 48076
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.8183267712593079,
      "learning_rate": 0.0005378156858359417,
      "loss": 2.1775,
      "step": 48077
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.6611160635948181,
      "learning_rate": 0.0005378131922591331,
      "loss": 2.3453,
      "step": 48078
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.6693755388259888,
      "learning_rate": 0.0005378106986381107,
      "loss": 2.3569,
      "step": 48079
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7311933040618896,
      "learning_rate": 0.0005378082049728747,
      "loss": 2.0376,
      "step": 48080
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.6850547194480896,
      "learning_rate": 0.0005378057112634257,
      "loss": 2.0941,
      "step": 48081
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.6877458691596985,
      "learning_rate": 0.0005378032175097641,
      "loss": 2.0931,
      "step": 48082
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.765835702419281,
      "learning_rate": 0.0005378007237118904,
      "loss": 2.2194,
      "step": 48083
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7242946028709412,
      "learning_rate": 0.0005377982298698051,
      "loss": 2.4193,
      "step": 48084
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.6703225374221802,
      "learning_rate": 0.0005377957359835085,
      "loss": 2.2138,
      "step": 48085
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.6693829298019409,
      "learning_rate": 0.0005377932420530012,
      "loss": 2.4611,
      "step": 48086
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.6800598502159119,
      "learning_rate": 0.0005377907480782837,
      "loss": 2.2141,
      "step": 48087
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.676922082901001,
      "learning_rate": 0.0005377882540593564,
      "loss": 2.2862,
      "step": 48088
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.6584028601646423,
      "learning_rate": 0.0005377857599962199,
      "loss": 2.1035,
      "step": 48089
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7962435483932495,
      "learning_rate": 0.0005377832658888744,
      "loss": 2.1514,
      "step": 48090
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.6966335773468018,
      "learning_rate": 0.0005377807717373205,
      "loss": 2.2764,
      "step": 48091
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.6776567697525024,
      "learning_rate": 0.0005377782775415587,
      "loss": 2.3352,
      "step": 48092
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.6800535321235657,
      "learning_rate": 0.0005377757833015894,
      "loss": 2.1462,
      "step": 48093
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.6837084889411926,
      "learning_rate": 0.0005377732890174132,
      "loss": 2.3268,
      "step": 48094
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7597448229789734,
      "learning_rate": 0.0005377707946890302,
      "loss": 2.1346,
      "step": 48095
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.812871515750885,
      "learning_rate": 0.0005377683003164414,
      "loss": 2.3445,
      "step": 48096
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.6849643588066101,
      "learning_rate": 0.0005377658058996469,
      "loss": 2.3525,
      "step": 48097
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.659375011920929,
      "learning_rate": 0.0005377633114386471,
      "loss": 2.2724,
      "step": 48098
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7585414052009583,
      "learning_rate": 0.0005377608169334429,
      "loss": 2.0088,
      "step": 48099
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7161027193069458,
      "learning_rate": 0.0005377583223840342,
      "loss": 2.2057,
      "step": 48100
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.720960795879364,
      "learning_rate": 0.0005377558277904219,
      "loss": 2.1316,
      "step": 48101
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7465013861656189,
      "learning_rate": 0.0005377533331526063,
      "loss": 2.306,
      "step": 48102
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7475529313087463,
      "learning_rate": 0.0005377508384705878,
      "loss": 2.2821,
      "step": 48103
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.709050178527832,
      "learning_rate": 0.000537748343744367,
      "loss": 2.2384,
      "step": 48104
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.6773355603218079,
      "learning_rate": 0.0005377458489739443,
      "loss": 2.1759,
      "step": 48105
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7155480980873108,
      "learning_rate": 0.0005377433541593201,
      "loss": 2.1093,
      "step": 48106
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.6496835350990295,
      "learning_rate": 0.0005377408593004949,
      "loss": 2.0231,
      "step": 48107
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.6965588331222534,
      "learning_rate": 0.0005377383643974692,
      "loss": 2.2963,
      "step": 48108
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7348068356513977,
      "learning_rate": 0.0005377358694502435,
      "loss": 2.2533,
      "step": 48109
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.731086015701294,
      "learning_rate": 0.0005377333744588182,
      "loss": 2.3422,
      "step": 48110
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7165906429290771,
      "learning_rate": 0.0005377308794231937,
      "loss": 2.1639,
      "step": 48111
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7160425782203674,
      "learning_rate": 0.0005377283843433706,
      "loss": 2.1077,
      "step": 48112
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7135857343673706,
      "learning_rate": 0.0005377258892193494,
      "loss": 2.199,
      "step": 48113
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7361653447151184,
      "learning_rate": 0.0005377233940511303,
      "loss": 2.2232,
      "step": 48114
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7692310214042664,
      "learning_rate": 0.000537720898838714,
      "loss": 2.405,
      "step": 48115
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.74561607837677,
      "learning_rate": 0.000537718403582101,
      "loss": 2.2534,
      "step": 48116
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.6593990325927734,
      "learning_rate": 0.0005377159082812915,
      "loss": 2.4299,
      "step": 48117
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.6804971098899841,
      "learning_rate": 0.0005377134129362862,
      "loss": 2.3266,
      "step": 48118
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7386177778244019,
      "learning_rate": 0.0005377109175470856,
      "loss": 2.3864,
      "step": 48119
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.6622135043144226,
      "learning_rate": 0.0005377084221136898,
      "loss": 2.2016,
      "step": 48120
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.6498107314109802,
      "learning_rate": 0.0005377059266360997,
      "loss": 2.2503,
      "step": 48121
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.6079248189926147,
      "learning_rate": 0.0005377034311143156,
      "loss": 2.3849,
      "step": 48122
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7559836506843567,
      "learning_rate": 0.0005377009355483379,
      "loss": 2.3869,
      "step": 48123
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.6744511127471924,
      "learning_rate": 0.0005376984399381671,
      "loss": 2.1968,
      "step": 48124
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7411935925483704,
      "learning_rate": 0.0005376959442838036,
      "loss": 2.3942,
      "step": 48125
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7137604355812073,
      "learning_rate": 0.0005376934485852479,
      "loss": 2.3036,
      "step": 48126
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.6789156198501587,
      "learning_rate": 0.0005376909528425007,
      "loss": 2.4652,
      "step": 48127
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.6970728039741516,
      "learning_rate": 0.0005376884570555622,
      "loss": 2.1782,
      "step": 48128
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.6836255192756653,
      "learning_rate": 0.0005376859612244329,
      "loss": 2.1666,
      "step": 48129
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.8121829628944397,
      "learning_rate": 0.0005376834653491133,
      "loss": 2.1481,
      "step": 48130
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7155270576477051,
      "learning_rate": 0.0005376809694296038,
      "loss": 2.0534,
      "step": 48131
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7204529047012329,
      "learning_rate": 0.000537678473465905,
      "loss": 2.3984,
      "step": 48132
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7823584079742432,
      "learning_rate": 0.0005376759774580173,
      "loss": 2.4999,
      "step": 48133
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7331134676933289,
      "learning_rate": 0.0005376734814059411,
      "loss": 2.2007,
      "step": 48134
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7290168404579163,
      "learning_rate": 0.0005376709853096769,
      "loss": 2.2364,
      "step": 48135
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.6706401705741882,
      "learning_rate": 0.0005376684891692252,
      "loss": 2.3426,
      "step": 48136
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7977559566497803,
      "learning_rate": 0.0005376659929845865,
      "loss": 2.2489,
      "step": 48137
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7136301398277283,
      "learning_rate": 0.0005376634967557611,
      "loss": 2.3526,
      "step": 48138
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.9374087452888489,
      "learning_rate": 0.0005376610004827496,
      "loss": 2.258,
      "step": 48139
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7243181467056274,
      "learning_rate": 0.0005376585041655525,
      "loss": 2.1986,
      "step": 48140
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.6726170778274536,
      "learning_rate": 0.0005376560078041701,
      "loss": 2.0891,
      "step": 48141
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7541870474815369,
      "learning_rate": 0.0005376535113986031,
      "loss": 2.2875,
      "step": 48142
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.0418579578399658,
      "learning_rate": 0.0005376510149488516,
      "loss": 2.3015,
      "step": 48143
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.6456816792488098,
      "learning_rate": 0.0005376485184549165,
      "loss": 2.2137,
      "step": 48144
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7756588459014893,
      "learning_rate": 0.0005376460219167979,
      "loss": 2.2185,
      "step": 48145
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.871487557888031,
      "learning_rate": 0.0005376435253344966,
      "loss": 2.3537,
      "step": 48146
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7353245615959167,
      "learning_rate": 0.0005376410287080128,
      "loss": 2.3473,
      "step": 48147
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.6811081767082214,
      "learning_rate": 0.000537638532037347,
      "loss": 2.2363,
      "step": 48148
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.6897520422935486,
      "learning_rate": 0.0005376360353224997,
      "loss": 2.3376,
      "step": 48149
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.8839662671089172,
      "learning_rate": 0.0005376335385634714,
      "loss": 2.1539,
      "step": 48150
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.6539217829704285,
      "learning_rate": 0.0005376310417602626,
      "loss": 2.4174,
      "step": 48151
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.756900429725647,
      "learning_rate": 0.0005376285449128737,
      "loss": 2.4097,
      "step": 48152
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.9236127734184265,
      "learning_rate": 0.0005376260480213052,
      "loss": 2.2273,
      "step": 48153
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.6988005042076111,
      "learning_rate": 0.0005376235510855574,
      "loss": 2.2492,
      "step": 48154
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7392920255661011,
      "learning_rate": 0.000537621054105631,
      "loss": 2.2494,
      "step": 48155
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.2575733661651611,
      "learning_rate": 0.0005376185570815262,
      "loss": 2.0656,
      "step": 48156
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.946675181388855,
      "learning_rate": 0.0005376160600132438,
      "loss": 2.3345,
      "step": 48157
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.8530867099761963,
      "learning_rate": 0.000537613562900784,
      "loss": 2.5748,
      "step": 48158
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.2456742525100708,
      "learning_rate": 0.0005376110657441474,
      "loss": 2.3476,
      "step": 48159
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.9295170903205872,
      "learning_rate": 0.0005376085685433343,
      "loss": 2.3628,
      "step": 48160
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.97659832239151,
      "learning_rate": 0.0005376060712983454,
      "loss": 2.3679,
      "step": 48161
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7285196781158447,
      "learning_rate": 0.0005376035740091811,
      "loss": 2.3004,
      "step": 48162
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.154201626777649,
      "learning_rate": 0.0005376010766758417,
      "loss": 2.0718,
      "step": 48163
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.8062858581542969,
      "learning_rate": 0.0005375985792983279,
      "loss": 2.2528,
      "step": 48164
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7475299835205078,
      "learning_rate": 0.0005375960818766399,
      "loss": 2.156,
      "step": 48165
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7782238125801086,
      "learning_rate": 0.0005375935844107784,
      "loss": 2.256,
      "step": 48166
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.0423299074172974,
      "learning_rate": 0.0005375910869007437,
      "loss": 2.1467,
      "step": 48167
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7122021913528442,
      "learning_rate": 0.0005375885893465363,
      "loss": 2.0676,
      "step": 48168
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.8570662140846252,
      "learning_rate": 0.0005375860917481568,
      "loss": 2.3788,
      "step": 48169
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7162848711013794,
      "learning_rate": 0.0005375835941056055,
      "loss": 2.5383,
      "step": 48170
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.1887023448944092,
      "learning_rate": 0.0005375810964188829,
      "loss": 2.2875,
      "step": 48171
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7344255447387695,
      "learning_rate": 0.0005375785986879895,
      "loss": 2.2728,
      "step": 48172
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7235684990882874,
      "learning_rate": 0.0005375761009129258,
      "loss": 2.3983,
      "step": 48173
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7718769907951355,
      "learning_rate": 0.0005375736030936922,
      "loss": 2.4195,
      "step": 48174
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.8761985898017883,
      "learning_rate": 0.0005375711052302893,
      "loss": 2.0834,
      "step": 48175
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7823488712310791,
      "learning_rate": 0.0005375686073227173,
      "loss": 2.2363,
      "step": 48176
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7512913346290588,
      "learning_rate": 0.0005375661093709767,
      "loss": 2.2285,
      "step": 48177
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7769789099693298,
      "learning_rate": 0.0005375636113750683,
      "loss": 2.0659,
      "step": 48178
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.6930457353591919,
      "learning_rate": 0.0005375611133349922,
      "loss": 2.3684,
      "step": 48179
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7277833819389343,
      "learning_rate": 0.0005375586152507492,
      "loss": 2.3728,
      "step": 48180
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.681513249874115,
      "learning_rate": 0.0005375561171223394,
      "loss": 2.4395,
      "step": 48181
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.705358624458313,
      "learning_rate": 0.0005375536189497635,
      "loss": 2.205,
      "step": 48182
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.2047066688537598,
      "learning_rate": 0.0005375511207330218,
      "loss": 2.32,
      "step": 48183
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7799208164215088,
      "learning_rate": 0.0005375486224721149,
      "loss": 2.3289,
      "step": 48184
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7588245868682861,
      "learning_rate": 0.0005375461241670433,
      "loss": 2.3619,
      "step": 48185
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7517944574356079,
      "learning_rate": 0.0005375436258178073,
      "loss": 2.3567,
      "step": 48186
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.707535982131958,
      "learning_rate": 0.0005375411274244075,
      "loss": 2.2865,
      "step": 48187
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.8343610167503357,
      "learning_rate": 0.0005375386289868443,
      "loss": 2.1271,
      "step": 48188
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.8440070748329163,
      "learning_rate": 0.0005375361305051182,
      "loss": 2.2741,
      "step": 48189
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7684549689292908,
      "learning_rate": 0.0005375336319792297,
      "loss": 2.1891,
      "step": 48190
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.781264066696167,
      "learning_rate": 0.0005375311334091792,
      "loss": 2.3931,
      "step": 48191
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.8309236168861389,
      "learning_rate": 0.000537528634794967,
      "loss": 2.2269,
      "step": 48192
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7865376472473145,
      "learning_rate": 0.000537526136136594,
      "loss": 2.506,
      "step": 48193
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7605488300323486,
      "learning_rate": 0.0005375236374340602,
      "loss": 2.182,
      "step": 48194
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7791750431060791,
      "learning_rate": 0.0005375211386873664,
      "loss": 2.362,
      "step": 48195
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.8446420431137085,
      "learning_rate": 0.000537518639896513,
      "loss": 2.2282,
      "step": 48196
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7029301524162292,
      "learning_rate": 0.0005375161410615003,
      "loss": 2.3309,
      "step": 48197
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7484715580940247,
      "learning_rate": 0.0005375136421823289,
      "loss": 2.3685,
      "step": 48198
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.9310875535011292,
      "learning_rate": 0.0005375111432589991,
      "loss": 2.0191,
      "step": 48199
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.6545897126197815,
      "learning_rate": 0.0005375086442915117,
      "loss": 2.1428,
      "step": 48200
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7441763877868652,
      "learning_rate": 0.0005375061452798668,
      "loss": 2.1783,
      "step": 48201
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7183501124382019,
      "learning_rate": 0.0005375036462240651,
      "loss": 2.3105,
      "step": 48202
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7104732394218445,
      "learning_rate": 0.000537501147124107,
      "loss": 2.2825,
      "step": 48203
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.6294723749160767,
      "learning_rate": 0.0005374986479799929,
      "loss": 2.3815,
      "step": 48204
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7511519193649292,
      "learning_rate": 0.0005374961487917233,
      "loss": 2.3424,
      "step": 48205
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7451426982879639,
      "learning_rate": 0.0005374936495592987,
      "loss": 2.3761,
      "step": 48206
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7222920060157776,
      "learning_rate": 0.0005374911502827196,
      "loss": 2.4031,
      "step": 48207
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7287424206733704,
      "learning_rate": 0.0005374886509619864,
      "loss": 2.3522,
      "step": 48208
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7219622135162354,
      "learning_rate": 0.0005374861515970997,
      "loss": 2.275,
      "step": 48209
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7025570273399353,
      "learning_rate": 0.0005374836521880596,
      "loss": 2.3656,
      "step": 48210
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.737890899181366,
      "learning_rate": 0.000537481152734867,
      "loss": 2.2843,
      "step": 48211
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7047097682952881,
      "learning_rate": 0.000537478653237522,
      "loss": 2.5261,
      "step": 48212
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.8431066870689392,
      "learning_rate": 0.0005374761536960254,
      "loss": 2.4193,
      "step": 48213
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7278220653533936,
      "learning_rate": 0.0005374736541103773,
      "loss": 2.5376,
      "step": 48214
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7093345522880554,
      "learning_rate": 0.0005374711544805785,
      "loss": 2.1489,
      "step": 48215
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.6485515236854553,
      "learning_rate": 0.0005374686548066293,
      "loss": 2.1435,
      "step": 48216
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.719793975353241,
      "learning_rate": 0.0005374661550885302,
      "loss": 2.2462,
      "step": 48217
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.9013873934745789,
      "learning_rate": 0.0005374636553262817,
      "loss": 2.2733,
      "step": 48218
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.720336377620697,
      "learning_rate": 0.0005374611555198842,
      "loss": 2.2797,
      "step": 48219
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7260076999664307,
      "learning_rate": 0.0005374586556693381,
      "loss": 2.2129,
      "step": 48220
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.9349648952484131,
      "learning_rate": 0.0005374561557746441,
      "loss": 2.1106,
      "step": 48221
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.668880820274353,
      "learning_rate": 0.0005374536558358025,
      "loss": 2.0774,
      "step": 48222
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.8569291234016418,
      "learning_rate": 0.0005374511558528137,
      "loss": 2.3889,
      "step": 48223
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.6969975233078003,
      "learning_rate": 0.0005374486558256783,
      "loss": 2.2158,
      "step": 48224
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7543331980705261,
      "learning_rate": 0.0005374461557543967,
      "loss": 2.5089,
      "step": 48225
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.6717814207077026,
      "learning_rate": 0.0005374436556389693,
      "loss": 2.1716,
      "step": 48226
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7762974500656128,
      "learning_rate": 0.0005374411554793968,
      "loss": 2.3453,
      "step": 48227
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.6950470805168152,
      "learning_rate": 0.0005374386552756793,
      "loss": 2.3235,
      "step": 48228
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.6397305130958557,
      "learning_rate": 0.0005374361550278177,
      "loss": 2.4626,
      "step": 48229
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.8045483827590942,
      "learning_rate": 0.0005374336547358121,
      "loss": 2.5091,
      "step": 48230
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.6867645978927612,
      "learning_rate": 0.0005374311543996632,
      "loss": 2.1658,
      "step": 48231
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.6841068863868713,
      "learning_rate": 0.0005374286540193713,
      "loss": 2.1466,
      "step": 48232
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.9052100777626038,
      "learning_rate": 0.0005374261535949368,
      "loss": 2.3439,
      "step": 48233
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7773740887641907,
      "learning_rate": 0.0005374236531263605,
      "loss": 2.2802,
      "step": 48234
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.741009533405304,
      "learning_rate": 0.0005374211526136426,
      "loss": 2.2146,
      "step": 48235
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7271744608879089,
      "learning_rate": 0.0005374186520567836,
      "loss": 2.2192,
      "step": 48236
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.8039855360984802,
      "learning_rate": 0.000537416151455784,
      "loss": 2.5477,
      "step": 48237
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7996250987052917,
      "learning_rate": 0.0005374136508106443,
      "loss": 2.1404,
      "step": 48238
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.717890739440918,
      "learning_rate": 0.0005374111501213649,
      "loss": 2.3435,
      "step": 48239
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7120651602745056,
      "learning_rate": 0.0005374086493879462,
      "loss": 2.464,
      "step": 48240
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7000109553337097,
      "learning_rate": 0.0005374061486103889,
      "loss": 2.3496,
      "step": 48241
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.8551332950592041,
      "learning_rate": 0.0005374036477886931,
      "loss": 2.0786,
      "step": 48242
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7546128034591675,
      "learning_rate": 0.0005374011469228597,
      "loss": 2.2113,
      "step": 48243
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.8336233496665955,
      "learning_rate": 0.0005373986460128888,
      "loss": 2.2552,
      "step": 48244
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7964819073677063,
      "learning_rate": 0.0005373961450587812,
      "loss": 2.3591,
      "step": 48245
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.6961514353752136,
      "learning_rate": 0.0005373936440605369,
      "loss": 2.2842,
      "step": 48246
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.6489901542663574,
      "learning_rate": 0.0005373911430181569,
      "loss": 2.1061,
      "step": 48247
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7184039354324341,
      "learning_rate": 0.0005373886419316413,
      "loss": 2.4061,
      "step": 48248
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7207110524177551,
      "learning_rate": 0.0005373861408009906,
      "loss": 2.5108,
      "step": 48249
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.6771817803382874,
      "learning_rate": 0.0005373836396262054,
      "loss": 2.0439,
      "step": 48250
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.716023862361908,
      "learning_rate": 0.0005373811384072861,
      "loss": 2.116,
      "step": 48251
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.6889903545379639,
      "learning_rate": 0.0005373786371442331,
      "loss": 2.2451,
      "step": 48252
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7737908363342285,
      "learning_rate": 0.0005373761358370471,
      "loss": 2.2838,
      "step": 48253
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.6533406376838684,
      "learning_rate": 0.0005373736344857282,
      "loss": 2.0791,
      "step": 48254
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.8911177515983582,
      "learning_rate": 0.0005373711330902772,
      "loss": 2.2957,
      "step": 48255
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7971490621566772,
      "learning_rate": 0.0005373686316506944,
      "loss": 2.1661,
      "step": 48256
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7099767923355103,
      "learning_rate": 0.0005373661301669802,
      "loss": 2.4651,
      "step": 48257
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.8687613010406494,
      "learning_rate": 0.0005373636286391352,
      "loss": 2.2198,
      "step": 48258
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7220777273178101,
      "learning_rate": 0.0005373611270671599,
      "loss": 2.1598,
      "step": 48259
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7528868913650513,
      "learning_rate": 0.0005373586254510546,
      "loss": 2.3774,
      "step": 48260
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.6699559688568115,
      "learning_rate": 0.0005373561237908199,
      "loss": 2.0545,
      "step": 48261
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7347177267074585,
      "learning_rate": 0.0005373536220864562,
      "loss": 2.2287,
      "step": 48262
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7559443116188049,
      "learning_rate": 0.000537351120337964,
      "loss": 2.3134,
      "step": 48263
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7508161664009094,
      "learning_rate": 0.0005373486185453437,
      "loss": 2.3784,
      "step": 48264
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7115736603736877,
      "learning_rate": 0.0005373461167085959,
      "loss": 2.3743,
      "step": 48265
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7550621628761292,
      "learning_rate": 0.0005373436148277208,
      "loss": 2.2027,
      "step": 48266
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.8427678942680359,
      "learning_rate": 0.0005373411129027193,
      "loss": 2.2835,
      "step": 48267
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7344374060630798,
      "learning_rate": 0.0005373386109335913,
      "loss": 2.1134,
      "step": 48268
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7391924858093262,
      "learning_rate": 0.0005373361089203378,
      "loss": 2.2821,
      "step": 48269
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.9917026162147522,
      "learning_rate": 0.000537333606862959,
      "loss": 2.4174,
      "step": 48270
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.1677082777023315,
      "learning_rate": 0.0005373311047614554,
      "loss": 2.183,
      "step": 48271
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7212752103805542,
      "learning_rate": 0.0005373286026158274,
      "loss": 2.2779,
      "step": 48272
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7906466722488403,
      "learning_rate": 0.0005373261004260755,
      "loss": 2.414,
      "step": 48273
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.8803567290306091,
      "learning_rate": 0.0005373235981922003,
      "loss": 2.1865,
      "step": 48274
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.8340030312538147,
      "learning_rate": 0.0005373210959142021,
      "loss": 2.165,
      "step": 48275
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.6671171188354492,
      "learning_rate": 0.0005373185935920816,
      "loss": 2.1483,
      "step": 48276
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.8027125000953674,
      "learning_rate": 0.0005373160912258389,
      "loss": 2.2458,
      "step": 48277
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.8072831034660339,
      "learning_rate": 0.0005373135888154747,
      "loss": 2.2844,
      "step": 48278
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7750914692878723,
      "learning_rate": 0.0005373110863609894,
      "loss": 2.2255,
      "step": 48279
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7805519104003906,
      "learning_rate": 0.0005373085838623836,
      "loss": 2.3374,
      "step": 48280
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7778971791267395,
      "learning_rate": 0.0005373060813196576,
      "loss": 2.2088,
      "step": 48281
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7942405343055725,
      "learning_rate": 0.0005373035787328119,
      "loss": 2.1998,
      "step": 48282
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7814923524856567,
      "learning_rate": 0.000537301076101847,
      "loss": 2.3014,
      "step": 48283
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7201085090637207,
      "learning_rate": 0.0005372985734267633,
      "loss": 2.2772,
      "step": 48284
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7272409200668335,
      "learning_rate": 0.0005372960707075613,
      "loss": 2.3801,
      "step": 48285
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.8505871891975403,
      "learning_rate": 0.0005372935679442415,
      "loss": 2.2621,
      "step": 48286
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7021971344947815,
      "learning_rate": 0.0005372910651368044,
      "loss": 1.9917,
      "step": 48287
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.6911613941192627,
      "learning_rate": 0.0005372885622852505,
      "loss": 2.152,
      "step": 48288
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7049108147621155,
      "learning_rate": 0.00053728605938958,
      "loss": 2.1504,
      "step": 48289
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7280848026275635,
      "learning_rate": 0.0005372835564497936,
      "loss": 2.054,
      "step": 48290
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7322297096252441,
      "learning_rate": 0.0005372810534658918,
      "loss": 2.2946,
      "step": 48291
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7914270758628845,
      "learning_rate": 0.0005372785504378748,
      "loss": 2.4449,
      "step": 48292
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.6823282241821289,
      "learning_rate": 0.0005372760473657433,
      "loss": 2.2853,
      "step": 48293
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7051840424537659,
      "learning_rate": 0.0005372735442494978,
      "loss": 1.9997,
      "step": 48294
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7115751504898071,
      "learning_rate": 0.0005372710410891387,
      "loss": 2.3092,
      "step": 48295
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7123939394950867,
      "learning_rate": 0.0005372685378846663,
      "loss": 2.2827,
      "step": 48296
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.6958455443382263,
      "learning_rate": 0.0005372660346360812,
      "loss": 2.2985,
      "step": 48297
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.9342759251594543,
      "learning_rate": 0.000537263531343384,
      "loss": 2.4172,
      "step": 48298
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7500119805335999,
      "learning_rate": 0.0005372610280065748,
      "loss": 2.4458,
      "step": 48299
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.6745488047599792,
      "learning_rate": 0.0005372585246256545,
      "loss": 2.2682,
      "step": 48300
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7626597285270691,
      "learning_rate": 0.0005372560212006233,
      "loss": 2.4791,
      "step": 48301
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7088103294372559,
      "learning_rate": 0.0005372535177314817,
      "loss": 2.1184,
      "step": 48302
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7250253558158875,
      "learning_rate": 0.0005372510142182302,
      "loss": 2.2783,
      "step": 48303
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7277057766914368,
      "learning_rate": 0.0005372485106608693,
      "loss": 2.3893,
      "step": 48304
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7804190516471863,
      "learning_rate": 0.0005372460070593994,
      "loss": 2.0906,
      "step": 48305
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.8404586911201477,
      "learning_rate": 0.000537243503413821,
      "loss": 2.5239,
      "step": 48306
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.8098822832107544,
      "learning_rate": 0.0005372409997241344,
      "loss": 2.3128,
      "step": 48307
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7253651022911072,
      "learning_rate": 0.0005372384959903405,
      "loss": 2.2603,
      "step": 48308
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7970916628837585,
      "learning_rate": 0.0005372359922124393,
      "loss": 2.2164,
      "step": 48309
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7225903272628784,
      "learning_rate": 0.0005372334883904315,
      "loss": 2.524,
      "step": 48310
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.8142581582069397,
      "learning_rate": 0.0005372309845243175,
      "loss": 2.1894,
      "step": 48311
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7431338429450989,
      "learning_rate": 0.0005372284806140978,
      "loss": 2.2911,
      "step": 48312
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7942549586296082,
      "learning_rate": 0.0005372259766597729,
      "loss": 2.2345,
      "step": 48313
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7853639125823975,
      "learning_rate": 0.0005372234726613432,
      "loss": 2.3615,
      "step": 48314
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.6758203506469727,
      "learning_rate": 0.0005372209686188091,
      "loss": 2.0502,
      "step": 48315
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7414682507514954,
      "learning_rate": 0.0005372184645321712,
      "loss": 2.3118,
      "step": 48316
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7189840078353882,
      "learning_rate": 0.0005372159604014299,
      "loss": 2.3659,
      "step": 48317
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7250518202781677,
      "learning_rate": 0.0005372134562265856,
      "loss": 2.305,
      "step": 48318
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.6919180154800415,
      "learning_rate": 0.0005372109520076388,
      "loss": 2.2852,
      "step": 48319
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.6851844191551208,
      "learning_rate": 0.0005372084477445902,
      "loss": 2.2535,
      "step": 48320
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7591838240623474,
      "learning_rate": 0.0005372059434374401,
      "loss": 2.1732,
      "step": 48321
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7458250522613525,
      "learning_rate": 0.0005372034390861887,
      "loss": 2.2933,
      "step": 48322
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.6685504913330078,
      "learning_rate": 0.0005372009346908368,
      "loss": 2.1658,
      "step": 48323
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7565189599990845,
      "learning_rate": 0.0005371984302513849,
      "loss": 2.602,
      "step": 48324
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7034940123558044,
      "learning_rate": 0.0005371959257678331,
      "loss": 2.3438,
      "step": 48325
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7100300192832947,
      "learning_rate": 0.0005371934212401822,
      "loss": 2.0329,
      "step": 48326
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7197984457015991,
      "learning_rate": 0.0005371909166684327,
      "loss": 2.0886,
      "step": 48327
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.629680335521698,
      "learning_rate": 0.0005371884120525848,
      "loss": 2.4094,
      "step": 48328
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7071518301963806,
      "learning_rate": 0.0005371859073926391,
      "loss": 2.3168,
      "step": 48329
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7947045564651489,
      "learning_rate": 0.0005371834026885961,
      "loss": 2.2686,
      "step": 48330
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7331678867340088,
      "learning_rate": 0.0005371808979404562,
      "loss": 2.2606,
      "step": 48331
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.751308262348175,
      "learning_rate": 0.00053717839314822,
      "loss": 2.2978,
      "step": 48332
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7129945158958435,
      "learning_rate": 0.0005371758883118877,
      "loss": 2.1284,
      "step": 48333
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7110052704811096,
      "learning_rate": 0.0005371733834314599,
      "loss": 2.2283,
      "step": 48334
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7558595538139343,
      "learning_rate": 0.0005371708785069372,
      "loss": 2.1582,
      "step": 48335
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.6835957765579224,
      "learning_rate": 0.00053716837353832,
      "loss": 2.0639,
      "step": 48336
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.74625563621521,
      "learning_rate": 0.0005371658685256086,
      "loss": 2.2139,
      "step": 48337
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.8497024774551392,
      "learning_rate": 0.0005371633634688037,
      "loss": 2.2847,
      "step": 48338
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.6680799722671509,
      "learning_rate": 0.0005371608583679056,
      "loss": 2.5568,
      "step": 48339
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7931690216064453,
      "learning_rate": 0.0005371583532229148,
      "loss": 2.4151,
      "step": 48340
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7204897999763489,
      "learning_rate": 0.0005371558480338318,
      "loss": 2.1816,
      "step": 48341
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7420856952667236,
      "learning_rate": 0.0005371533428006571,
      "loss": 2.2101,
      "step": 48342
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.674599289894104,
      "learning_rate": 0.000537150837523391,
      "loss": 2.1471,
      "step": 48343
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.6851029396057129,
      "learning_rate": 0.0005371483322020341,
      "loss": 2.0307,
      "step": 48344
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.80726158618927,
      "learning_rate": 0.000537145826836587,
      "loss": 2.2492,
      "step": 48345
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.6717798709869385,
      "learning_rate": 0.0005371433214270499,
      "loss": 2.3501,
      "step": 48346
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.753742516040802,
      "learning_rate": 0.0005371408159734235,
      "loss": 2.183,
      "step": 48347
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.6482371091842651,
      "learning_rate": 0.0005371383104757079,
      "loss": 2.141,
      "step": 48348
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.6752532720565796,
      "learning_rate": 0.0005371358049339039,
      "loss": 1.9498,
      "step": 48349
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.6813209056854248,
      "learning_rate": 0.0005371332993480121,
      "loss": 2.2338,
      "step": 48350
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.6801143884658813,
      "learning_rate": 0.0005371307937180325,
      "loss": 2.2788,
      "step": 48351
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.6784418225288391,
      "learning_rate": 0.0005371282880439658,
      "loss": 2.3305,
      "step": 48352
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7474308013916016,
      "learning_rate": 0.0005371257823258127,
      "loss": 2.3042,
      "step": 48353
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.6888399124145508,
      "learning_rate": 0.0005371232765635733,
      "loss": 2.1754,
      "step": 48354
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.6126730442047119,
      "learning_rate": 0.0005371207707572483,
      "loss": 2.2502,
      "step": 48355
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.6967737674713135,
      "learning_rate": 0.0005371182649068378,
      "loss": 2.1596,
      "step": 48356
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.6911568641662598,
      "learning_rate": 0.0005371157590123428,
      "loss": 2.286,
      "step": 48357
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7625826001167297,
      "learning_rate": 0.0005371132530737634,
      "loss": 2.3227,
      "step": 48358
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.6933477520942688,
      "learning_rate": 0.0005371107470911002,
      "loss": 2.4232,
      "step": 48359
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.6641339659690857,
      "learning_rate": 0.0005371082410643537,
      "loss": 2.1253,
      "step": 48360
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7060263752937317,
      "learning_rate": 0.0005371057349935241,
      "loss": 2.0113,
      "step": 48361
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.6642794609069824,
      "learning_rate": 0.0005371032288786123,
      "loss": 2.132,
      "step": 48362
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7177780270576477,
      "learning_rate": 0.0005371007227196184,
      "loss": 2.3095,
      "step": 48363
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.6984423995018005,
      "learning_rate": 0.000537098216516543,
      "loss": 2.2876,
      "step": 48364
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.6993805766105652,
      "learning_rate": 0.0005370957102693867,
      "loss": 2.2868,
      "step": 48365
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.634505569934845,
      "learning_rate": 0.0005370932039781498,
      "loss": 2.3331,
      "step": 48366
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7121487855911255,
      "learning_rate": 0.0005370906976428326,
      "loss": 2.1187,
      "step": 48367
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7399156093597412,
      "learning_rate": 0.0005370881912634359,
      "loss": 2.3114,
      "step": 48368
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.6621929407119751,
      "learning_rate": 0.0005370856848399601,
      "loss": 2.211,
      "step": 48369
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7561958432197571,
      "learning_rate": 0.0005370831783724054,
      "loss": 2.2926,
      "step": 48370
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.6483347415924072,
      "learning_rate": 0.0005370806718607727,
      "loss": 2.0817,
      "step": 48371
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7393620014190674,
      "learning_rate": 0.0005370781653050621,
      "loss": 2.1187,
      "step": 48372
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7117483615875244,
      "learning_rate": 0.0005370756587052742,
      "loss": 2.6011,
      "step": 48373
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7313537001609802,
      "learning_rate": 0.0005370731520614095,
      "loss": 2.4346,
      "step": 48374
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7311935424804688,
      "learning_rate": 0.0005370706453734684,
      "loss": 2.2614,
      "step": 48375
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.706593930721283,
      "learning_rate": 0.0005370681386414513,
      "loss": 2.3167,
      "step": 48376
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7566912770271301,
      "learning_rate": 0.0005370656318653588,
      "loss": 2.2662,
      "step": 48377
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.8486065864562988,
      "learning_rate": 0.0005370631250451915,
      "loss": 2.3156,
      "step": 48378
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7166525721549988,
      "learning_rate": 0.0005370606181809495,
      "loss": 2.0526,
      "step": 48379
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.6809962391853333,
      "learning_rate": 0.0005370581112726336,
      "loss": 2.2982,
      "step": 48380
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.6442623734474182,
      "learning_rate": 0.000537055604320244,
      "loss": 2.3438,
      "step": 48381
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.6855435967445374,
      "learning_rate": 0.0005370530973237813,
      "loss": 2.0646,
      "step": 48382
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7413707971572876,
      "learning_rate": 0.0005370505902832459,
      "loss": 2.3383,
      "step": 48383
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.6935708522796631,
      "learning_rate": 0.0005370480831986384,
      "loss": 2.0274,
      "step": 48384
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7076102495193481,
      "learning_rate": 0.0005370455760699592,
      "loss": 2.1953,
      "step": 48385
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.6847947835922241,
      "learning_rate": 0.0005370430688972088,
      "loss": 2.1622,
      "step": 48386
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.8350649476051331,
      "learning_rate": 0.0005370405616803874,
      "loss": 2.1615,
      "step": 48387
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7031118869781494,
      "learning_rate": 0.0005370380544194959,
      "loss": 2.2686,
      "step": 48388
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.6868969202041626,
      "learning_rate": 0.0005370355471145345,
      "loss": 2.2212,
      "step": 48389
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7334348559379578,
      "learning_rate": 0.0005370330397655037,
      "loss": 2.3441,
      "step": 48390
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.704185426235199,
      "learning_rate": 0.0005370305323724041,
      "loss": 2.3752,
      "step": 48391
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7663225531578064,
      "learning_rate": 0.0005370280249352359,
      "loss": 2.1012,
      "step": 48392
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.6776450276374817,
      "learning_rate": 0.0005370255174539998,
      "loss": 2.0865,
      "step": 48393
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7979310750961304,
      "learning_rate": 0.0005370230099286961,
      "loss": 2.3955,
      "step": 48394
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7593691945075989,
      "learning_rate": 0.0005370205023593254,
      "loss": 2.3124,
      "step": 48395
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.8049488067626953,
      "learning_rate": 0.0005370179947458882,
      "loss": 2.3944,
      "step": 48396
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7749049067497253,
      "learning_rate": 0.0005370154870883847,
      "loss": 2.0823,
      "step": 48397
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7377041578292847,
      "learning_rate": 0.0005370129793868156,
      "loss": 2.7563,
      "step": 48398
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7679849863052368,
      "learning_rate": 0.0005370104716411814,
      "loss": 2.0085,
      "step": 48399
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.6422693133354187,
      "learning_rate": 0.0005370079638514824,
      "loss": 2.1773,
      "step": 48400
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7326498031616211,
      "learning_rate": 0.0005370054560177191,
      "loss": 2.4149,
      "step": 48401
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.773022472858429,
      "learning_rate": 0.0005370029481398922,
      "loss": 2.2741,
      "step": 48402
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.6998094916343689,
      "learning_rate": 0.0005370004402180019,
      "loss": 2.3485,
      "step": 48403
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7223818302154541,
      "learning_rate": 0.0005369979322520486,
      "loss": 2.0128,
      "step": 48404
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7682592272758484,
      "learning_rate": 0.0005369954242420331,
      "loss": 2.2459,
      "step": 48405
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.6622401475906372,
      "learning_rate": 0.0005369929161879555,
      "loss": 2.3465,
      "step": 48406
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.0735349655151367,
      "learning_rate": 0.0005369904080898166,
      "loss": 2.4509,
      "step": 48407
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.6953060030937195,
      "learning_rate": 0.0005369878999476166,
      "loss": 1.9335,
      "step": 48408
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7423168420791626,
      "learning_rate": 0.0005369853917613561,
      "loss": 2.3198,
      "step": 48409
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7947908639907837,
      "learning_rate": 0.0005369828835310356,
      "loss": 2.3835,
      "step": 48410
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7888637185096741,
      "learning_rate": 0.0005369803752566554,
      "loss": 2.2231,
      "step": 48411
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7259028553962708,
      "learning_rate": 0.0005369778669382162,
      "loss": 2.321,
      "step": 48412
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.8268247246742249,
      "learning_rate": 0.0005369753585757184,
      "loss": 2.2649,
      "step": 48413
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7028234601020813,
      "learning_rate": 0.0005369728501691622,
      "loss": 2.174,
      "step": 48414
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.8577857613563538,
      "learning_rate": 0.0005369703417185483,
      "loss": 2.1796,
      "step": 48415
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7688354849815369,
      "learning_rate": 0.0005369678332238772,
      "loss": 2.117,
      "step": 48416
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7253965735435486,
      "learning_rate": 0.0005369653246851495,
      "loss": 2.445,
      "step": 48417
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.6923483610153198,
      "learning_rate": 0.0005369628161023653,
      "loss": 2.409,
      "step": 48418
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.9391857385635376,
      "learning_rate": 0.0005369603074755251,
      "loss": 2.3161,
      "step": 48419
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.8938995003700256,
      "learning_rate": 0.0005369577988046297,
      "loss": 2.2229,
      "step": 48420
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7136722803115845,
      "learning_rate": 0.0005369552900896793,
      "loss": 2.3099,
      "step": 48421
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.9553294777870178,
      "learning_rate": 0.0005369527813306744,
      "loss": 2.3211,
      "step": 48422
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7828552722930908,
      "learning_rate": 0.0005369502725276156,
      "loss": 2.5052,
      "step": 48423
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7452002763748169,
      "learning_rate": 0.0005369477636805031,
      "loss": 2.1006,
      "step": 48424
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7559008598327637,
      "learning_rate": 0.0005369452547893378,
      "loss": 2.0706,
      "step": 48425
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7780136466026306,
      "learning_rate": 0.0005369427458541197,
      "loss": 2.2338,
      "step": 48426
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.8551424741744995,
      "learning_rate": 0.0005369402368748496,
      "loss": 2.2088,
      "step": 48427
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7549434304237366,
      "learning_rate": 0.0005369377278515277,
      "loss": 2.2725,
      "step": 48428
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7914323806762695,
      "learning_rate": 0.0005369352187841547,
      "loss": 2.3089,
      "step": 48429
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7884746789932251,
      "learning_rate": 0.000536932709672731,
      "loss": 2.2547,
      "step": 48430
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.8034167289733887,
      "learning_rate": 0.000536930200517257,
      "loss": 2.3328,
      "step": 48431
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.8158984184265137,
      "learning_rate": 0.0005369276913177331,
      "loss": 2.1427,
      "step": 48432
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.8192047476768494,
      "learning_rate": 0.00053692518207416,
      "loss": 2.0637,
      "step": 48433
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7866520285606384,
      "learning_rate": 0.000536922672786538,
      "loss": 1.8896,
      "step": 48434
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.8430932760238647,
      "learning_rate": 0.0005369201634548676,
      "loss": 2.2483,
      "step": 48435
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.950492262840271,
      "learning_rate": 0.0005369176540791492,
      "loss": 2.1253,
      "step": 48436
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.8247407674789429,
      "learning_rate": 0.0005369151446593834,
      "loss": 2.183,
      "step": 48437
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.8766441345214844,
      "learning_rate": 0.0005369126351955706,
      "loss": 2.1871,
      "step": 48438
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7019985318183899,
      "learning_rate": 0.0005369101256877114,
      "loss": 2.2619,
      "step": 48439
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7107511162757874,
      "learning_rate": 0.0005369076161358059,
      "loss": 1.9535,
      "step": 48440
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.8862540125846863,
      "learning_rate": 0.0005369051065398549,
      "loss": 2.456,
      "step": 48441
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7131248712539673,
      "learning_rate": 0.0005369025968998589,
      "loss": 2.4859,
      "step": 48442
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.6985971331596375,
      "learning_rate": 0.000536900087215818,
      "loss": 2.2015,
      "step": 48443
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.8743726015090942,
      "learning_rate": 0.000536897577487733,
      "loss": 2.2968,
      "step": 48444
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.6468346118927002,
      "learning_rate": 0.0005368950677156043,
      "loss": 2.2504,
      "step": 48445
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7407417893409729,
      "learning_rate": 0.0005368925578994323,
      "loss": 2.1093,
      "step": 48446
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.673410177230835,
      "learning_rate": 0.0005368900480392176,
      "loss": 2.014,
      "step": 48447
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.78007572889328,
      "learning_rate": 0.0005368875381349605,
      "loss": 2.207,
      "step": 48448
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7396383285522461,
      "learning_rate": 0.0005368850281866615,
      "loss": 2.1016,
      "step": 48449
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7691720128059387,
      "learning_rate": 0.0005368825181943211,
      "loss": 2.1275,
      "step": 48450
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7425797581672668,
      "learning_rate": 0.0005368800081579398,
      "loss": 2.3754,
      "step": 48451
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.057567834854126,
      "learning_rate": 0.0005368774980775181,
      "loss": 2.2593,
      "step": 48452
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.70721435546875,
      "learning_rate": 0.0005368749879530563,
      "loss": 2.2641,
      "step": 48453
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.733221173286438,
      "learning_rate": 0.0005368724777845551,
      "loss": 2.4707,
      "step": 48454
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7254023551940918,
      "learning_rate": 0.0005368699675720146,
      "loss": 2.2116,
      "step": 48455
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.6643907427787781,
      "learning_rate": 0.0005368674573154356,
      "loss": 2.2945,
      "step": 48456
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7742061614990234,
      "learning_rate": 0.0005368649470148185,
      "loss": 2.0841,
      "step": 48457
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7254528999328613,
      "learning_rate": 0.0005368624366701638,
      "loss": 2.2013,
      "step": 48458
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7447143793106079,
      "learning_rate": 0.0005368599262814719,
      "loss": 2.2078,
      "step": 48459
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.8313835263252258,
      "learning_rate": 0.0005368574158487431,
      "loss": 2.1725,
      "step": 48460
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7161794304847717,
      "learning_rate": 0.0005368549053719783,
      "loss": 2.1196,
      "step": 48461
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7204623222351074,
      "learning_rate": 0.0005368523948511775,
      "loss": 2.3618,
      "step": 48462
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.0015217065811157,
      "learning_rate": 0.0005368498842863414,
      "loss": 1.8472,
      "step": 48463
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.8744908571243286,
      "learning_rate": 0.0005368473736774705,
      "loss": 2.406,
      "step": 48464
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7772862911224365,
      "learning_rate": 0.0005368448630245652,
      "loss": 2.3052,
      "step": 48465
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.6523919701576233,
      "learning_rate": 0.0005368423523276259,
      "loss": 2.24,
      "step": 48466
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7445289492607117,
      "learning_rate": 0.0005368398415866531,
      "loss": 2.0984,
      "step": 48467
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.6884261965751648,
      "learning_rate": 0.0005368373308016473,
      "loss": 2.366,
      "step": 48468
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7553982138633728,
      "learning_rate": 0.000536834819972609,
      "loss": 2.2915,
      "step": 48469
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7696073651313782,
      "learning_rate": 0.0005368323090995387,
      "loss": 2.4194,
      "step": 48470
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7463805675506592,
      "learning_rate": 0.0005368297981824369,
      "loss": 2.3473,
      "step": 48471
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7093303799629211,
      "learning_rate": 0.0005368272872213038,
      "loss": 2.16,
      "step": 48472
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7135709524154663,
      "learning_rate": 0.0005368247762161401,
      "loss": 2.2879,
      "step": 48473
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7253766655921936,
      "learning_rate": 0.0005368222651669461,
      "loss": 2.352,
      "step": 48474
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.8039349317550659,
      "learning_rate": 0.0005368197540737225,
      "loss": 2.337,
      "step": 48475
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.6400713920593262,
      "learning_rate": 0.0005368172429364695,
      "loss": 1.9617,
      "step": 48476
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.6678234934806824,
      "learning_rate": 0.0005368147317551878,
      "loss": 2.2612,
      "step": 48477
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7179883718490601,
      "learning_rate": 0.0005368122205298779,
      "loss": 2.4017,
      "step": 48478
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.6698484420776367,
      "learning_rate": 0.0005368097092605399,
      "loss": 2.2595,
      "step": 48479
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.6808534264564514,
      "learning_rate": 0.0005368071979471746,
      "loss": 2.3781,
      "step": 48480
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7137709856033325,
      "learning_rate": 0.0005368046865897824,
      "loss": 2.2726,
      "step": 48481
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.982297420501709,
      "learning_rate": 0.0005368021751883638,
      "loss": 2.2947,
      "step": 48482
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.0261006355285645,
      "learning_rate": 0.0005367996637429192,
      "loss": 2.1758,
      "step": 48483
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.6532762050628662,
      "learning_rate": 0.0005367971522534489,
      "loss": 2.16,
      "step": 48484
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7641447186470032,
      "learning_rate": 0.0005367946407199536,
      "loss": 2.0434,
      "step": 48485
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.728625476360321,
      "learning_rate": 0.0005367921291424339,
      "loss": 2.2465,
      "step": 48486
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7222715616226196,
      "learning_rate": 0.0005367896175208898,
      "loss": 2.1812,
      "step": 48487
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7943141460418701,
      "learning_rate": 0.0005367871058553222,
      "loss": 2.1341,
      "step": 48488
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.6777143478393555,
      "learning_rate": 0.0005367845941457313,
      "loss": 2.3207,
      "step": 48489
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.6954678297042847,
      "learning_rate": 0.0005367820823921178,
      "loss": 2.2878,
      "step": 48490
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.6620121598243713,
      "learning_rate": 0.000536779570594482,
      "loss": 2.2261,
      "step": 48491
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7752326130867004,
      "learning_rate": 0.0005367770587528243,
      "loss": 2.5344,
      "step": 48492
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7126449942588806,
      "learning_rate": 0.0005367745468671454,
      "loss": 2.1299,
      "step": 48493
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7305529713630676,
      "learning_rate": 0.0005367720349374457,
      "loss": 2.1486,
      "step": 48494
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.740183413028717,
      "learning_rate": 0.0005367695229637255,
      "loss": 2.0981,
      "step": 48495
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7233049273490906,
      "learning_rate": 0.0005367670109459854,
      "loss": 2.1854,
      "step": 48496
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7321231365203857,
      "learning_rate": 0.0005367644988842258,
      "loss": 2.0936,
      "step": 48497
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7210097312927246,
      "learning_rate": 0.0005367619867784472,
      "loss": 2.4042,
      "step": 48498
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7467360496520996,
      "learning_rate": 0.0005367594746286501,
      "loss": 2.4228,
      "step": 48499
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.8872986435890198,
      "learning_rate": 0.000536756962434835,
      "loss": 2.2314,
      "step": 48500
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.2762737274169922,
      "learning_rate": 0.0005367544501970023,
      "loss": 2.5022,
      "step": 48501
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7571202516555786,
      "learning_rate": 0.0005367519379151525,
      "loss": 2.4685,
      "step": 48502
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7357261776924133,
      "learning_rate": 0.000536749425589286,
      "loss": 2.5647,
      "step": 48503
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7342754602432251,
      "learning_rate": 0.0005367469132194034,
      "loss": 2.2539,
      "step": 48504
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7712665796279907,
      "learning_rate": 0.0005367444008055049,
      "loss": 2.2297,
      "step": 48505
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7181607484817505,
      "learning_rate": 0.0005367418883475912,
      "loss": 2.2077,
      "step": 48506
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.6642312407493591,
      "learning_rate": 0.0005367393758456628,
      "loss": 2.2415,
      "step": 48507
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.701066792011261,
      "learning_rate": 0.0005367368632997201,
      "loss": 2.2358,
      "step": 48508
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7333288788795471,
      "learning_rate": 0.0005367343507097634,
      "loss": 2.2013,
      "step": 48509
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7313999533653259,
      "learning_rate": 0.0005367318380757935,
      "loss": 2.1932,
      "step": 48510
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7180049419403076,
      "learning_rate": 0.0005367293253978105,
      "loss": 2.1601,
      "step": 48511
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.712354838848114,
      "learning_rate": 0.0005367268126758152,
      "loss": 2.33,
      "step": 48512
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7734352946281433,
      "learning_rate": 0.0005367242999098079,
      "loss": 2.1649,
      "step": 48513
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7299098372459412,
      "learning_rate": 0.000536721787099789,
      "loss": 2.335,
      "step": 48514
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7143123745918274,
      "learning_rate": 0.000536719274245759,
      "loss": 2.245,
      "step": 48515
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.8735546469688416,
      "learning_rate": 0.0005367167613477186,
      "loss": 2.2437,
      "step": 48516
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7639212608337402,
      "learning_rate": 0.000536714248405668,
      "loss": 1.9166,
      "step": 48517
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7545477151870728,
      "learning_rate": 0.0005367117354196076,
      "loss": 2.0155,
      "step": 48518
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7586117386817932,
      "learning_rate": 0.0005367092223895383,
      "loss": 2.2827,
      "step": 48519
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7212578654289246,
      "learning_rate": 0.0005367067093154602,
      "loss": 2.234,
      "step": 48520
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7260521650314331,
      "learning_rate": 0.0005367041961973737,
      "loss": 2.312,
      "step": 48521
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.6930899620056152,
      "learning_rate": 0.0005367016830352796,
      "loss": 2.4354,
      "step": 48522
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7770141959190369,
      "learning_rate": 0.000536699169829178,
      "loss": 2.203,
      "step": 48523
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7103564143180847,
      "learning_rate": 0.0005366966565790698,
      "loss": 2.2483,
      "step": 48524
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.6897030472755432,
      "learning_rate": 0.0005366941432849551,
      "loss": 2.3093,
      "step": 48525
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7176240086555481,
      "learning_rate": 0.0005366916299468344,
      "loss": 2.2487,
      "step": 48526
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.764400064945221,
      "learning_rate": 0.0005366891165647083,
      "loss": 2.1943,
      "step": 48527
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.6958085298538208,
      "learning_rate": 0.0005366866031385774,
      "loss": 2.2445,
      "step": 48528
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7269202470779419,
      "learning_rate": 0.0005366840896684418,
      "loss": 2.1488,
      "step": 48529
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.8150433897972107,
      "learning_rate": 0.0005366815761543022,
      "loss": 2.2802,
      "step": 48530
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.6765289306640625,
      "learning_rate": 0.000536679062596159,
      "loss": 2.0637,
      "step": 48531
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.6942084431648254,
      "learning_rate": 0.0005366765489940129,
      "loss": 2.4324,
      "step": 48532
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7031040191650391,
      "learning_rate": 0.000536674035347864,
      "loss": 2.4234,
      "step": 48533
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7392569780349731,
      "learning_rate": 0.0005366715216577129,
      "loss": 2.3864,
      "step": 48534
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7070748805999756,
      "learning_rate": 0.0005366690079235602,
      "loss": 2.1618,
      "step": 48535
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.634711503982544,
      "learning_rate": 0.0005366664941454061,
      "loss": 2.0846,
      "step": 48536
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.6825990676879883,
      "learning_rate": 0.0005366639803232514,
      "loss": 2.3165,
      "step": 48537
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.6351509690284729,
      "learning_rate": 0.0005366614664570963,
      "loss": 2.2379,
      "step": 48538
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.8185290694236755,
      "learning_rate": 0.0005366589525469414,
      "loss": 2.1031,
      "step": 48539
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7628905177116394,
      "learning_rate": 0.0005366564385927871,
      "loss": 2.5597,
      "step": 48540
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.6685748100280762,
      "learning_rate": 0.0005366539245946339,
      "loss": 2.4158,
      "step": 48541
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7544304728507996,
      "learning_rate": 0.0005366514105524824,
      "loss": 2.0825,
      "step": 48542
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7048640251159668,
      "learning_rate": 0.0005366488964663328,
      "loss": 2.2032,
      "step": 48543
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.703244149684906,
      "learning_rate": 0.0005366463823361857,
      "loss": 1.9835,
      "step": 48544
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7179460525512695,
      "learning_rate": 0.0005366438681620417,
      "loss": 2.1621,
      "step": 48545
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.9765569567680359,
      "learning_rate": 0.000536641353943901,
      "loss": 2.5006,
      "step": 48546
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7204983234405518,
      "learning_rate": 0.0005366388396817642,
      "loss": 2.2568,
      "step": 48547
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.2128621339797974,
      "learning_rate": 0.0005366363253756319,
      "loss": 2.1367,
      "step": 48548
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7664803266525269,
      "learning_rate": 0.0005366338110255044,
      "loss": 2.1271,
      "step": 48549
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.8170470595359802,
      "learning_rate": 0.0005366312966313821,
      "loss": 2.2999,
      "step": 48550
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.9067577123641968,
      "learning_rate": 0.0005366287821932656,
      "loss": 2.0789,
      "step": 48551
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.8142099976539612,
      "learning_rate": 0.0005366262677111554,
      "loss": 2.4183,
      "step": 48552
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7432212233543396,
      "learning_rate": 0.0005366237531850519,
      "loss": 2.2221,
      "step": 48553
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7902822494506836,
      "learning_rate": 0.0005366212386149555,
      "loss": 2.3838,
      "step": 48554
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.835866391658783,
      "learning_rate": 0.000536618724000867,
      "loss": 2.3264,
      "step": 48555
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.8092770576477051,
      "learning_rate": 0.0005366162093427863,
      "loss": 2.3679,
      "step": 48556
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.6662557125091553,
      "learning_rate": 0.0005366136946407143,
      "loss": 2.1564,
      "step": 48557
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.8986936211585999,
      "learning_rate": 0.0005366111798946514,
      "loss": 2.3569,
      "step": 48558
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7850217223167419,
      "learning_rate": 0.000536608665104598,
      "loss": 2.3439,
      "step": 48559
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7450866103172302,
      "learning_rate": 0.0005366061502705546,
      "loss": 2.2164,
      "step": 48560
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7384011149406433,
      "learning_rate": 0.0005366036353925216,
      "loss": 2.3391,
      "step": 48561
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.8350104689598083,
      "learning_rate": 0.0005366011204704995,
      "loss": 2.229,
      "step": 48562
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7634811997413635,
      "learning_rate": 0.0005365986055044888,
      "loss": 2.2103,
      "step": 48563
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.731431245803833,
      "learning_rate": 0.00053659609049449,
      "loss": 2.4406,
      "step": 48564
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7368922233581543,
      "learning_rate": 0.0005365935754405034,
      "loss": 2.3311,
      "step": 48565
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.867622971534729,
      "learning_rate": 0.0005365910603425297,
      "loss": 2.1719,
      "step": 48566
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7061125636100769,
      "learning_rate": 0.0005365885452005693,
      "loss": 2.3776,
      "step": 48567
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.8215956091880798,
      "learning_rate": 0.0005365860300146225,
      "loss": 2.3602,
      "step": 48568
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7808583974838257,
      "learning_rate": 0.00053658351478469,
      "loss": 2.2781,
      "step": 48569
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.8008083701133728,
      "learning_rate": 0.000536580999510772,
      "loss": 2.396,
      "step": 48570
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.8244321942329407,
      "learning_rate": 0.0005365784841928693,
      "loss": 2.1825,
      "step": 48571
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.706552267074585,
      "learning_rate": 0.0005365759688309822,
      "loss": 2.2945,
      "step": 48572
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.6397351026535034,
      "learning_rate": 0.000536573453425111,
      "loss": 2.4029,
      "step": 48573
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7326680421829224,
      "learning_rate": 0.0005365709379752565,
      "loss": 2.2054,
      "step": 48574
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.760886549949646,
      "learning_rate": 0.0005365684224814188,
      "loss": 2.4797,
      "step": 48575
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.6634501814842224,
      "learning_rate": 0.0005365659069435988,
      "loss": 2.2857,
      "step": 48576
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.6396694779396057,
      "learning_rate": 0.0005365633913617966,
      "loss": 2.3294,
      "step": 48577
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.6724259257316589,
      "learning_rate": 0.0005365608757360127,
      "loss": 2.402,
      "step": 48578
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7423829436302185,
      "learning_rate": 0.0005365583600662479,
      "loss": 2.1196,
      "step": 48579
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7417956590652466,
      "learning_rate": 0.0005365558443525023,
      "loss": 2.0784,
      "step": 48580
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7228564620018005,
      "learning_rate": 0.0005365533285947764,
      "loss": 2.321,
      "step": 48581
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.6629030704498291,
      "learning_rate": 0.000536550812793071,
      "loss": 2.293,
      "step": 48582
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.6525373458862305,
      "learning_rate": 0.0005365482969473862,
      "loss": 2.0974,
      "step": 48583
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.6801100969314575,
      "learning_rate": 0.0005365457810577227,
      "loss": 2.3867,
      "step": 48584
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.6969956755638123,
      "learning_rate": 0.0005365432651240807,
      "loss": 2.2796,
      "step": 48585
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.884434163570404,
      "learning_rate": 0.000536540749146461,
      "loss": 2.3378,
      "step": 48586
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7364166975021362,
      "learning_rate": 0.0005365382331248638,
      "loss": 2.5429,
      "step": 48587
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7060596942901611,
      "learning_rate": 0.0005365357170592898,
      "loss": 2.0913,
      "step": 48588
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.6617949604988098,
      "learning_rate": 0.0005365332009497393,
      "loss": 2.0679,
      "step": 48589
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7203173041343689,
      "learning_rate": 0.0005365306847962127,
      "loss": 2.1113,
      "step": 48590
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.8484266400337219,
      "learning_rate": 0.0005365281685987108,
      "loss": 2.5814,
      "step": 48591
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.6607756018638611,
      "learning_rate": 0.0005365256523572337,
      "loss": 2.2908,
      "step": 48592
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7127218246459961,
      "learning_rate": 0.000536523136071782,
      "loss": 2.173,
      "step": 48593
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.8611130714416504,
      "learning_rate": 0.0005365206197423562,
      "loss": 2.1863,
      "step": 48594
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7710520029067993,
      "learning_rate": 0.0005365181033689569,
      "loss": 2.4321,
      "step": 48595
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7205067276954651,
      "learning_rate": 0.0005365155869515842,
      "loss": 2.296,
      "step": 48596
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.813267707824707,
      "learning_rate": 0.0005365130704902389,
      "loss": 2.358,
      "step": 48597
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7881269454956055,
      "learning_rate": 0.0005365105539849213,
      "loss": 2.3473,
      "step": 48598
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.6943551301956177,
      "learning_rate": 0.000536508037435632,
      "loss": 2.1353,
      "step": 48599
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.8093936443328857,
      "learning_rate": 0.0005365055208423713,
      "loss": 2.2612,
      "step": 48600
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7025702595710754,
      "learning_rate": 0.0005365030042051398,
      "loss": 2.3661,
      "step": 48601
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.6902142763137817,
      "learning_rate": 0.0005365004875239381,
      "loss": 2.2525,
      "step": 48602
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7920270562171936,
      "learning_rate": 0.0005364979707987663,
      "loss": 2.2853,
      "step": 48603
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7116127014160156,
      "learning_rate": 0.0005364954540296251,
      "loss": 2.3991,
      "step": 48604
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.6840854287147522,
      "learning_rate": 0.0005364929372165149,
      "loss": 2.1389,
      "step": 48605
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.790922999382019,
      "learning_rate": 0.0005364904203594362,
      "loss": 2.2294,
      "step": 48606
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7251262068748474,
      "learning_rate": 0.0005364879034583895,
      "loss": 2.2931,
      "step": 48607
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7303468585014343,
      "learning_rate": 0.0005364853865133752,
      "loss": 2.0951,
      "step": 48608
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7404744625091553,
      "learning_rate": 0.0005364828695243938,
      "loss": 2.3547,
      "step": 48609
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7038585543632507,
      "learning_rate": 0.000536480352491446,
      "loss": 2.0228,
      "step": 48610
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.8926936388015747,
      "learning_rate": 0.0005364778354145317,
      "loss": 2.1453,
      "step": 48611
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7150723934173584,
      "learning_rate": 0.0005364753182936519,
      "loss": 2.2981,
      "step": 48612
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.8705673217773438,
      "learning_rate": 0.0005364728011288068,
      "loss": 2.2737,
      "step": 48613
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.6400271654129028,
      "learning_rate": 0.0005364702839199969,
      "loss": 2.1756,
      "step": 48614
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7079641222953796,
      "learning_rate": 0.0005364677666672229,
      "loss": 2.4126,
      "step": 48615
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7100024223327637,
      "learning_rate": 0.000536465249370485,
      "loss": 2.0942,
      "step": 48616
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.6580810546875,
      "learning_rate": 0.0005364627320297836,
      "loss": 2.2921,
      "step": 48617
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7366859316825867,
      "learning_rate": 0.0005364602146451194,
      "loss": 2.2,
      "step": 48618
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7331807613372803,
      "learning_rate": 0.0005364576972164928,
      "loss": 2.4571,
      "step": 48619
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.731124222278595,
      "learning_rate": 0.0005364551797439042,
      "loss": 2.2763,
      "step": 48620
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7680312991142273,
      "learning_rate": 0.0005364526622273542,
      "loss": 2.0288,
      "step": 48621
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.6968072056770325,
      "learning_rate": 0.0005364501446668432,
      "loss": 2.3789,
      "step": 48622
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7786259651184082,
      "learning_rate": 0.0005364476270623717,
      "loss": 2.1631,
      "step": 48623
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7294725179672241,
      "learning_rate": 0.00053644510941394,
      "loss": 2.2479,
      "step": 48624
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7107282280921936,
      "learning_rate": 0.0005364425917215486,
      "loss": 2.301,
      "step": 48625
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7094576358795166,
      "learning_rate": 0.0005364400739851983,
      "loss": 2.1558,
      "step": 48626
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.9038664102554321,
      "learning_rate": 0.0005364375562048892,
      "loss": 2.3422,
      "step": 48627
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.6817134618759155,
      "learning_rate": 0.0005364350383806219,
      "loss": 2.1411,
      "step": 48628
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.688382089138031,
      "learning_rate": 0.0005364325205123968,
      "loss": 2.3188,
      "step": 48629
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.6942824125289917,
      "learning_rate": 0.0005364300026002145,
      "loss": 2.3319,
      "step": 48630
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7053917646408081,
      "learning_rate": 0.0005364274846440754,
      "loss": 2.2313,
      "step": 48631
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7962687015533447,
      "learning_rate": 0.00053642496664398,
      "loss": 2.4331,
      "step": 48632
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7115134596824646,
      "learning_rate": 0.0005364224485999287,
      "loss": 2.3473,
      "step": 48633
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.6188604831695557,
      "learning_rate": 0.0005364199305119221,
      "loss": 2.4639,
      "step": 48634
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.6566537022590637,
      "learning_rate": 0.0005364174123799603,
      "loss": 2.3151,
      "step": 48635
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7624225616455078,
      "learning_rate": 0.0005364148942040443,
      "loss": 1.9685,
      "step": 48636
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.6499570608139038,
      "learning_rate": 0.0005364123759841743,
      "loss": 2.2462,
      "step": 48637
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.6292728185653687,
      "learning_rate": 0.0005364098577203508,
      "loss": 2.0844,
      "step": 48638
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.8593209385871887,
      "learning_rate": 0.000536407339412574,
      "loss": 2.296,
      "step": 48639
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7260627150535583,
      "learning_rate": 0.0005364048210608447,
      "loss": 2.2628,
      "step": 48640
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7526006102561951,
      "learning_rate": 0.0005364023026651634,
      "loss": 2.3899,
      "step": 48641
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.6950328946113586,
      "learning_rate": 0.0005363997842255304,
      "loss": 2.2813,
      "step": 48642
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.650160551071167,
      "learning_rate": 0.0005363972657419462,
      "loss": 2.2457,
      "step": 48643
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.6413533687591553,
      "learning_rate": 0.0005363947472144113,
      "loss": 2.2524,
      "step": 48644
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7419453859329224,
      "learning_rate": 0.0005363922286429262,
      "loss": 2.4499,
      "step": 48645
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7470396757125854,
      "learning_rate": 0.0005363897100274913,
      "loss": 2.1527,
      "step": 48646
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.6695097088813782,
      "learning_rate": 0.0005363871913681071,
      "loss": 2.3052,
      "step": 48647
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7478554844856262,
      "learning_rate": 0.0005363846726647739,
      "loss": 2.3553,
      "step": 48648
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.6617410182952881,
      "learning_rate": 0.0005363821539174925,
      "loss": 2.0687,
      "step": 48649
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7427682280540466,
      "learning_rate": 0.0005363796351262631,
      "loss": 2.444,
      "step": 48650
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7577232122421265,
      "learning_rate": 0.0005363771162910863,
      "loss": 2.0144,
      "step": 48651
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7312067747116089,
      "learning_rate": 0.0005363745974119625,
      "loss": 2.2166,
      "step": 48652
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.8123315572738647,
      "learning_rate": 0.0005363720784888923,
      "loss": 2.258,
      "step": 48653
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.749588668346405,
      "learning_rate": 0.000536369559521876,
      "loss": 2.4286,
      "step": 48654
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.6881402134895325,
      "learning_rate": 0.0005363670405109141,
      "loss": 2.4415,
      "step": 48655
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7554064989089966,
      "learning_rate": 0.0005363645214560073,
      "loss": 2.1383,
      "step": 48656
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.8886827230453491,
      "learning_rate": 0.0005363620023571557,
      "loss": 2.2274,
      "step": 48657
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.751504123210907,
      "learning_rate": 0.0005363594832143599,
      "loss": 2.3133,
      "step": 48658
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7329900860786438,
      "learning_rate": 0.0005363569640276205,
      "loss": 2.2623,
      "step": 48659
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7158818244934082,
      "learning_rate": 0.0005363544447969378,
      "loss": 2.3858,
      "step": 48660
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7754558324813843,
      "learning_rate": 0.0005363519255223125,
      "loss": 2.2017,
      "step": 48661
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.6705368757247925,
      "learning_rate": 0.0005363494062037448,
      "loss": 2.171,
      "step": 48662
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7636896967887878,
      "learning_rate": 0.0005363468868412354,
      "loss": 2.2504,
      "step": 48663
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7275585532188416,
      "learning_rate": 0.0005363443674347844,
      "loss": 2.2477,
      "step": 48664
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.6869595050811768,
      "learning_rate": 0.0005363418479843927,
      "loss": 2.1505,
      "step": 48665
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7624626755714417,
      "learning_rate": 0.0005363393284900606,
      "loss": 2.1188,
      "step": 48666
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7628681659698486,
      "learning_rate": 0.0005363368089517885,
      "loss": 2.1011,
      "step": 48667
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7505958676338196,
      "learning_rate": 0.000536334289369577,
      "loss": 2.3035,
      "step": 48668
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7676671147346497,
      "learning_rate": 0.0005363317697434265,
      "loss": 2.4167,
      "step": 48669
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7235437035560608,
      "learning_rate": 0.0005363292500733374,
      "loss": 2.1611,
      "step": 48670
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.6866962909698486,
      "learning_rate": 0.0005363267303593102,
      "loss": 2.2991,
      "step": 48671
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.8105366826057434,
      "learning_rate": 0.0005363242106013454,
      "loss": 2.5117,
      "step": 48672
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.6913003921508789,
      "learning_rate": 0.0005363216907994436,
      "loss": 2.3251,
      "step": 48673
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.6406822800636292,
      "learning_rate": 0.000536319170953605,
      "loss": 2.0254,
      "step": 48674
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.8875573873519897,
      "learning_rate": 0.0005363166510638302,
      "loss": 2.1998,
      "step": 48675
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7139580845832825,
      "learning_rate": 0.0005363141311301198,
      "loss": 2.1676,
      "step": 48676
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.6766366362571716,
      "learning_rate": 0.000536311611152474,
      "loss": 2.2529,
      "step": 48677
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7153919339179993,
      "learning_rate": 0.0005363090911308934,
      "loss": 2.3775,
      "step": 48678
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7147068381309509,
      "learning_rate": 0.0005363065710653786,
      "loss": 2.2014,
      "step": 48679
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7633127570152283,
      "learning_rate": 0.0005363040509559299,
      "loss": 2.3183,
      "step": 48680
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.6932973265647888,
      "learning_rate": 0.0005363015308025478,
      "loss": 2.3633,
      "step": 48681
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.6673999428749084,
      "learning_rate": 0.0005362990106052329,
      "loss": 2.1126,
      "step": 48682
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.6982669234275818,
      "learning_rate": 0.0005362964903639854,
      "loss": 2.3268,
      "step": 48683
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.6860944032669067,
      "learning_rate": 0.0005362939700788061,
      "loss": 2.0821,
      "step": 48684
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7293546795845032,
      "learning_rate": 0.0005362914497496951,
      "loss": 2.4848,
      "step": 48685
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.6768888831138611,
      "learning_rate": 0.0005362889293766531,
      "loss": 2.3427,
      "step": 48686
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.6952911019325256,
      "learning_rate": 0.0005362864089596806,
      "loss": 2.3973,
      "step": 48687
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7187141180038452,
      "learning_rate": 0.0005362838884987779,
      "loss": 2.2625,
      "step": 48688
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7052100896835327,
      "learning_rate": 0.0005362813679939456,
      "loss": 2.3052,
      "step": 48689
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7182190418243408,
      "learning_rate": 0.0005362788474451842,
      "loss": 2.4726,
      "step": 48690
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.9222781658172607,
      "learning_rate": 0.0005362763268524941,
      "loss": 2.0549,
      "step": 48691
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7048540711402893,
      "learning_rate": 0.0005362738062158757,
      "loss": 2.2566,
      "step": 48692
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7333263158798218,
      "learning_rate": 0.0005362712855353296,
      "loss": 2.082,
      "step": 48693
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7031319737434387,
      "learning_rate": 0.0005362687648108561,
      "loss": 2.224,
      "step": 48694
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7397749423980713,
      "learning_rate": 0.0005362662440424559,
      "loss": 2.433,
      "step": 48695
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7507445812225342,
      "learning_rate": 0.0005362637232301294,
      "loss": 2.1494,
      "step": 48696
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.679209291934967,
      "learning_rate": 0.0005362612023738768,
      "loss": 2.1109,
      "step": 48697
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.6809650659561157,
      "learning_rate": 0.0005362586814736988,
      "loss": 2.0777,
      "step": 48698
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7712109088897705,
      "learning_rate": 0.0005362561605295961,
      "loss": 2.2565,
      "step": 48699
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7709034085273743,
      "learning_rate": 0.0005362536395415687,
      "loss": 2.3924,
      "step": 48700
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7761498689651489,
      "learning_rate": 0.0005362511185096173,
      "loss": 2.177,
      "step": 48701
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7596596479415894,
      "learning_rate": 0.0005362485974337425,
      "loss": 2.6014,
      "step": 48702
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7074112892150879,
      "learning_rate": 0.0005362460763139444,
      "loss": 2.1641,
      "step": 48703
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.6642297506332397,
      "learning_rate": 0.0005362435551502239,
      "loss": 2.2394,
      "step": 48704
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.6357218027114868,
      "learning_rate": 0.0005362410339425812,
      "loss": 2.2009,
      "step": 48705
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7398870587348938,
      "learning_rate": 0.0005362385126910169,
      "loss": 2.3734,
      "step": 48706
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.8013543486595154,
      "learning_rate": 0.0005362359913955312,
      "loss": 2.0873,
      "step": 48707
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7129279375076294,
      "learning_rate": 0.000536233470056125,
      "loss": 2.1233,
      "step": 48708
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.6540009379386902,
      "learning_rate": 0.0005362309486727985,
      "loss": 2.1596,
      "step": 48709
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7135989665985107,
      "learning_rate": 0.0005362284272455521,
      "loss": 2.397,
      "step": 48710
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7548192739486694,
      "learning_rate": 0.0005362259057743864,
      "loss": 2.0649,
      "step": 48711
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7173565030097961,
      "learning_rate": 0.0005362233842593018,
      "loss": 2.2597,
      "step": 48712
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7434296607971191,
      "learning_rate": 0.0005362208627002989,
      "loss": 2.3691,
      "step": 48713
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7790074944496155,
      "learning_rate": 0.0005362183410973781,
      "loss": 2.2363,
      "step": 48714
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7573322653770447,
      "learning_rate": 0.0005362158194505398,
      "loss": 2.2294,
      "step": 48715
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.8425291180610657,
      "learning_rate": 0.0005362132977597846,
      "loss": 1.8983,
      "step": 48716
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.6270955204963684,
      "learning_rate": 0.0005362107760251127,
      "loss": 2.4968,
      "step": 48717
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7324729561805725,
      "learning_rate": 0.0005362082542465249,
      "loss": 2.2635,
      "step": 48718
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.6843025088310242,
      "learning_rate": 0.0005362057324240215,
      "loss": 2.3094,
      "step": 48719
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7119276523590088,
      "learning_rate": 0.000536203210557603,
      "loss": 2.3107,
      "step": 48720
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.6718124747276306,
      "learning_rate": 0.0005362006886472699,
      "loss": 2.0503,
      "step": 48721
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.6390523314476013,
      "learning_rate": 0.0005361981666930227,
      "loss": 2.219,
      "step": 48722
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7443454265594482,
      "learning_rate": 0.0005361956446948617,
      "loss": 2.3356,
      "step": 48723
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.6828639507293701,
      "learning_rate": 0.0005361931226527874,
      "loss": 2.2266,
      "step": 48724
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7281203269958496,
      "learning_rate": 0.0005361906005668004,
      "loss": 2.2417,
      "step": 48725
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.6960203647613525,
      "learning_rate": 0.0005361880784369011,
      "loss": 2.3453,
      "step": 48726
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.6697636842727661,
      "learning_rate": 0.0005361855562630901,
      "loss": 2.1477,
      "step": 48727
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7108672261238098,
      "learning_rate": 0.0005361830340453675,
      "loss": 2.3474,
      "step": 48728
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7841151356697083,
      "learning_rate": 0.0005361805117837342,
      "loss": 2.1729,
      "step": 48729
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7514082193374634,
      "learning_rate": 0.0005361779894781904,
      "loss": 2.4531,
      "step": 48730
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7232985496520996,
      "learning_rate": 0.0005361754671287368,
      "loss": 2.2268,
      "step": 48731
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.6589183807373047,
      "learning_rate": 0.0005361729447353737,
      "loss": 2.2532,
      "step": 48732
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.64324951171875,
      "learning_rate": 0.0005361704222981014,
      "loss": 2.2426,
      "step": 48733
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7243078947067261,
      "learning_rate": 0.0005361678998169207,
      "loss": 2.1912,
      "step": 48734
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.6848846077919006,
      "learning_rate": 0.0005361653772918319,
      "loss": 2.3779,
      "step": 48735
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7183387875556946,
      "learning_rate": 0.0005361628547228356,
      "loss": 2.2573,
      "step": 48736
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7535219788551331,
      "learning_rate": 0.000536160332109932,
      "loss": 2.0002,
      "step": 48737
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.755660355091095,
      "learning_rate": 0.0005361578094531217,
      "loss": 2.3278,
      "step": 48738
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.6514770984649658,
      "learning_rate": 0.0005361552867524054,
      "loss": 2.2542,
      "step": 48739
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7495025396347046,
      "learning_rate": 0.0005361527640077833,
      "loss": 2.1152,
      "step": 48740
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7066278457641602,
      "learning_rate": 0.0005361502412192559,
      "loss": 2.4003,
      "step": 48741
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7698219418525696,
      "learning_rate": 0.0005361477183868237,
      "loss": 2.2468,
      "step": 48742
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.6765716671943665,
      "learning_rate": 0.0005361451955104873,
      "loss": 2.2031,
      "step": 48743
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.795204222202301,
      "learning_rate": 0.0005361426725902469,
      "loss": 2.116,
      "step": 48744
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.8583777546882629,
      "learning_rate": 0.0005361401496261033,
      "loss": 1.9656,
      "step": 48745
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7431185841560364,
      "learning_rate": 0.0005361376266180567,
      "loss": 2.2398,
      "step": 48746
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7430948615074158,
      "learning_rate": 0.0005361351035661076,
      "loss": 2.1404,
      "step": 48747
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.759337842464447,
      "learning_rate": 0.0005361325804702565,
      "loss": 2.107,
      "step": 48748
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7400235533714294,
      "learning_rate": 0.000536130057330504,
      "loss": 2.4125,
      "step": 48749
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.6942676305770874,
      "learning_rate": 0.0005361275341468504,
      "loss": 2.4388,
      "step": 48750
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.687862753868103,
      "learning_rate": 0.0005361250109192963,
      "loss": 2.2946,
      "step": 48751
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7262078523635864,
      "learning_rate": 0.000536122487647842,
      "loss": 2.3403,
      "step": 48752
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.738719642162323,
      "learning_rate": 0.0005361199643324882,
      "loss": 2.0763,
      "step": 48753
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.6661561727523804,
      "learning_rate": 0.0005361174409732352,
      "loss": 2.2333,
      "step": 48754
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7617805600166321,
      "learning_rate": 0.0005361149175700836,
      "loss": 2.2625,
      "step": 48755
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.710521936416626,
      "learning_rate": 0.0005361123941230335,
      "loss": 2.1304,
      "step": 48756
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.6941961646080017,
      "learning_rate": 0.0005361098706320858,
      "loss": 2.3014,
      "step": 48757
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.6760689616203308,
      "learning_rate": 0.0005361073470972409,
      "loss": 2.0995,
      "step": 48758
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7041671872138977,
      "learning_rate": 0.0005361048235184992,
      "loss": 2.1633,
      "step": 48759
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.70071941614151,
      "learning_rate": 0.0005361022998958611,
      "loss": 2.3403,
      "step": 48760
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.6461637020111084,
      "learning_rate": 0.000536099776229327,
      "loss": 2.0322,
      "step": 48761
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7671658992767334,
      "learning_rate": 0.0005360972525188977,
      "loss": 2.2175,
      "step": 48762
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7830219864845276,
      "learning_rate": 0.0005360947287645733,
      "loss": 2.6578,
      "step": 48763
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7067950963973999,
      "learning_rate": 0.0005360922049663545,
      "loss": 2.175,
      "step": 48764
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.740195631980896,
      "learning_rate": 0.0005360896811242418,
      "loss": 2.235,
      "step": 48765
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7387135028839111,
      "learning_rate": 0.0005360871572382353,
      "loss": 2.3158,
      "step": 48766
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7107870578765869,
      "learning_rate": 0.000536084633308336,
      "loss": 2.1138,
      "step": 48767
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.6984070539474487,
      "learning_rate": 0.000536082109334544,
      "loss": 2.2622,
      "step": 48768
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7774466872215271,
      "learning_rate": 0.0005360795853168599,
      "loss": 2.4356,
      "step": 48769
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7233102917671204,
      "learning_rate": 0.0005360770612552842,
      "loss": 2.1769,
      "step": 48770
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.678589403629303,
      "learning_rate": 0.0005360745371498173,
      "loss": 2.0965,
      "step": 48771
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.6785895824432373,
      "learning_rate": 0.0005360720130004596,
      "loss": 2.2572,
      "step": 48772
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7582712173461914,
      "learning_rate": 0.0005360694888072117,
      "loss": 2.2807,
      "step": 48773
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7535696029663086,
      "learning_rate": 0.0005360669645700741,
      "loss": 2.5092,
      "step": 48774
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.9841887354850769,
      "learning_rate": 0.0005360644402890471,
      "loss": 2.2606,
      "step": 48775
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.6646439433097839,
      "learning_rate": 0.0005360619159641313,
      "loss": 2.4517,
      "step": 48776
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7361705303192139,
      "learning_rate": 0.0005360593915953272,
      "loss": 2.4388,
      "step": 48777
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7581817507743835,
      "learning_rate": 0.0005360568671826351,
      "loss": 2.2757,
      "step": 48778
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7797307372093201,
      "learning_rate": 0.0005360543427260555,
      "loss": 2.1253,
      "step": 48779
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7389468550682068,
      "learning_rate": 0.0005360518182255891,
      "loss": 2.4992,
      "step": 48780
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7631201148033142,
      "learning_rate": 0.0005360492936812363,
      "loss": 2.1443,
      "step": 48781
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7341111898422241,
      "learning_rate": 0.0005360467690929973,
      "loss": 2.2311,
      "step": 48782
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.6872456073760986,
      "learning_rate": 0.0005360442444608727,
      "loss": 2.2936,
      "step": 48783
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.6904263496398926,
      "learning_rate": 0.0005360417197848632,
      "loss": 2.2159,
      "step": 48784
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.6833555698394775,
      "learning_rate": 0.0005360391950649689,
      "loss": 2.1257,
      "step": 48785
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7696012854576111,
      "learning_rate": 0.0005360366703011906,
      "loss": 2.173,
      "step": 48786
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.6581347584724426,
      "learning_rate": 0.0005360341454935286,
      "loss": 2.2476,
      "step": 48787
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.8259365558624268,
      "learning_rate": 0.0005360316206419834,
      "loss": 2.2702,
      "step": 48788
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7716809511184692,
      "learning_rate": 0.0005360290957465555,
      "loss": 2.1712,
      "step": 48789
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7658668756484985,
      "learning_rate": 0.0005360265708072452,
      "loss": 2.2844,
      "step": 48790
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7287023663520813,
      "learning_rate": 0.0005360240458240532,
      "loss": 2.237,
      "step": 48791
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7603130340576172,
      "learning_rate": 0.0005360215207969799,
      "loss": 2.2668,
      "step": 48792
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7688339352607727,
      "learning_rate": 0.0005360189957260257,
      "loss": 2.1966,
      "step": 48793
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7357058525085449,
      "learning_rate": 0.0005360164706111911,
      "loss": 2.0799,
      "step": 48794
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.6868898272514343,
      "learning_rate": 0.0005360139454524766,
      "loss": 2.2286,
      "step": 48795
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.6980388760566711,
      "learning_rate": 0.0005360114202498826,
      "loss": 2.2294,
      "step": 48796
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.6551035046577454,
      "learning_rate": 0.0005360088950034097,
      "loss": 1.9771,
      "step": 48797
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7565788626670837,
      "learning_rate": 0.0005360063697130583,
      "loss": 2.3737,
      "step": 48798
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.749134361743927,
      "learning_rate": 0.0005360038443788289,
      "loss": 2.2097,
      "step": 48799
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7247757315635681,
      "learning_rate": 0.0005360013190007218,
      "loss": 2.5316,
      "step": 48800
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.6829551458358765,
      "learning_rate": 0.0005359987935787378,
      "loss": 2.1392,
      "step": 48801
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.6905632019042969,
      "learning_rate": 0.0005359962681128769,
      "loss": 2.0534,
      "step": 48802
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.6683205366134644,
      "learning_rate": 0.00053599374260314,
      "loss": 2.2116,
      "step": 48803
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.686240553855896,
      "learning_rate": 0.0005359912170495274,
      "loss": 2.3467,
      "step": 48804
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.670235276222229,
      "learning_rate": 0.0005359886914520396,
      "loss": 2.0877,
      "step": 48805
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.6500953435897827,
      "learning_rate": 0.0005359861658106769,
      "loss": 2.2148,
      "step": 48806
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.738034188747406,
      "learning_rate": 0.00053598364012544,
      "loss": 2.0303,
      "step": 48807
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.8104583621025085,
      "learning_rate": 0.0005359811143963295,
      "loss": 2.2192,
      "step": 48808
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.6871315836906433,
      "learning_rate": 0.0005359785886233454,
      "loss": 2.4122,
      "step": 48809
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.6710947751998901,
      "learning_rate": 0.0005359760628064885,
      "loss": 2.1888,
      "step": 48810
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.8101262450218201,
      "learning_rate": 0.0005359735369457591,
      "loss": 2.3261,
      "step": 48811
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7362467646598816,
      "learning_rate": 0.0005359710110411579,
      "loss": 2.3541,
      "step": 48812
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.8864267468452454,
      "learning_rate": 0.0005359684850926853,
      "loss": 2.4195,
      "step": 48813
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.8191967606544495,
      "learning_rate": 0.0005359659591003415,
      "loss": 2.3033,
      "step": 48814
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7362126708030701,
      "learning_rate": 0.0005359634330641273,
      "loss": 2.1603,
      "step": 48815
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.8489736914634705,
      "learning_rate": 0.0005359609069840429,
      "loss": 2.3554,
      "step": 48816
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7429590225219727,
      "learning_rate": 0.000535958380860089,
      "loss": 2.4903,
      "step": 48817
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.6863986849784851,
      "learning_rate": 0.000535955854692266,
      "loss": 2.215,
      "step": 48818
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7942186594009399,
      "learning_rate": 0.0005359533284805745,
      "loss": 2.2927,
      "step": 48819
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.8634202480316162,
      "learning_rate": 0.0005359508022250146,
      "loss": 2.28,
      "step": 48820
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.6850758194923401,
      "learning_rate": 0.0005359482759255871,
      "loss": 2.1885,
      "step": 48821
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.863477885723114,
      "learning_rate": 0.0005359457495822923,
      "loss": 2.4284,
      "step": 48822
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7662464380264282,
      "learning_rate": 0.0005359432231951308,
      "loss": 2.1522,
      "step": 48823
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.6915603876113892,
      "learning_rate": 0.000535940696764103,
      "loss": 2.5247,
      "step": 48824
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7488515973091125,
      "learning_rate": 0.0005359381702892093,
      "loss": 2.2654,
      "step": 48825
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.6985952258110046,
      "learning_rate": 0.0005359356437704503,
      "loss": 2.2969,
      "step": 48826
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7000725269317627,
      "learning_rate": 0.0005359331172078264,
      "loss": 2.1281,
      "step": 48827
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.8239016532897949,
      "learning_rate": 0.000535930590601338,
      "loss": 2.1481,
      "step": 48828
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7342823147773743,
      "learning_rate": 0.0005359280639509858,
      "loss": 2.2509,
      "step": 48829
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7169699668884277,
      "learning_rate": 0.00053592553725677,
      "loss": 2.364,
      "step": 48830
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7828081250190735,
      "learning_rate": 0.0005359230105186912,
      "loss": 2.1823,
      "step": 48831
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7523835301399231,
      "learning_rate": 0.0005359204837367499,
      "loss": 2.0955,
      "step": 48832
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7220447063446045,
      "learning_rate": 0.0005359179569109465,
      "loss": 2.4114,
      "step": 48833
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.697246789932251,
      "learning_rate": 0.0005359154300412815,
      "loss": 2.5176,
      "step": 48834
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7766938209533691,
      "learning_rate": 0.0005359129031277555,
      "loss": 2.2494,
      "step": 48835
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7637782692909241,
      "learning_rate": 0.0005359103761703686,
      "loss": 2.2045,
      "step": 48836
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7259962558746338,
      "learning_rate": 0.0005359078491691217,
      "loss": 2.3799,
      "step": 48837
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.2196826934814453,
      "learning_rate": 0.0005359053221240149,
      "loss": 2.4172,
      "step": 48838
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7468534111976624,
      "learning_rate": 0.000535902795035049,
      "loss": 2.4069,
      "step": 48839
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.6832401752471924,
      "learning_rate": 0.0005359002679022242,
      "loss": 2.2676,
      "step": 48840
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.8324329257011414,
      "learning_rate": 0.0005358977407255413,
      "loss": 2.2365,
      "step": 48841
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.013178825378418,
      "learning_rate": 0.0005358952135050003,
      "loss": 2.4222,
      "step": 48842
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7054762840270996,
      "learning_rate": 0.0005358926862406022,
      "loss": 2.3758,
      "step": 48843
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7930721640586853,
      "learning_rate": 0.0005358901589323469,
      "loss": 2.2471,
      "step": 48844
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.9016638994216919,
      "learning_rate": 0.0005358876315802353,
      "loss": 2.2498,
      "step": 48845
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7094505429267883,
      "learning_rate": 0.0005358851041842677,
      "loss": 2.1429,
      "step": 48846
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7436691522598267,
      "learning_rate": 0.0005358825767444447,
      "loss": 2.5183,
      "step": 48847
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7754009962081909,
      "learning_rate": 0.0005358800492607667,
      "loss": 2.036,
      "step": 48848
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.8685181736946106,
      "learning_rate": 0.000535877521733234,
      "loss": 2.3123,
      "step": 48849
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7991153597831726,
      "learning_rate": 0.0005358749941618473,
      "loss": 2.2195,
      "step": 48850
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.719346284866333,
      "learning_rate": 0.000535872466546607,
      "loss": 2.155,
      "step": 48851
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.8800346851348877,
      "learning_rate": 0.0005358699388875134,
      "loss": 2.1882,
      "step": 48852
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7512405514717102,
      "learning_rate": 0.0005358674111845673,
      "loss": 2.205,
      "step": 48853
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.8485761284828186,
      "learning_rate": 0.000535864883437769,
      "loss": 2.2535,
      "step": 48854
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.6860007643699646,
      "learning_rate": 0.000535862355647119,
      "loss": 2.2714,
      "step": 48855
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.6331092119216919,
      "learning_rate": 0.0005358598278126176,
      "loss": 2.3516,
      "step": 48856
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7612882852554321,
      "learning_rate": 0.0005358572999342654,
      "loss": 2.2694,
      "step": 48857
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7944591641426086,
      "learning_rate": 0.000535854772012063,
      "loss": 2.042,
      "step": 48858
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.702904462814331,
      "learning_rate": 0.0005358522440460107,
      "loss": 2.1729,
      "step": 48859
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.8651835918426514,
      "learning_rate": 0.0005358497160361089,
      "loss": 2.4017,
      "step": 48860
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.718856155872345,
      "learning_rate": 0.0005358471879823583,
      "loss": 2.1608,
      "step": 48861
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7054384350776672,
      "learning_rate": 0.0005358446598847592,
      "loss": 2.5047,
      "step": 48862
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.6717584729194641,
      "learning_rate": 0.0005358421317433122,
      "loss": 2.3685,
      "step": 48863
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.6403812170028687,
      "learning_rate": 0.0005358396035580177,
      "loss": 2.2021,
      "step": 48864
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7120576500892639,
      "learning_rate": 0.0005358370753288761,
      "loss": 2.242,
      "step": 48865
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7462725639343262,
      "learning_rate": 0.000535834547055888,
      "loss": 2.2815,
      "step": 48866
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7221188545227051,
      "learning_rate": 0.0005358320187390538,
      "loss": 2.3373,
      "step": 48867
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.8351529836654663,
      "learning_rate": 0.0005358294903783739,
      "loss": 2.2693,
      "step": 48868
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.8015643358230591,
      "learning_rate": 0.0005358269619738489,
      "loss": 2.2007,
      "step": 48869
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.6800429821014404,
      "learning_rate": 0.0005358244335254791,
      "loss": 2.0641,
      "step": 48870
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.6703552603721619,
      "learning_rate": 0.0005358219050332653,
      "loss": 2.0308,
      "step": 48871
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7161362171173096,
      "learning_rate": 0.0005358193764972076,
      "loss": 2.3496,
      "step": 48872
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.6904305219650269,
      "learning_rate": 0.0005358168479173067,
      "loss": 2.0331,
      "step": 48873
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.713019847869873,
      "learning_rate": 0.000535814319293563,
      "loss": 2.2074,
      "step": 48874
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.677034854888916,
      "learning_rate": 0.0005358117906259769,
      "loss": 2.2069,
      "step": 48875
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7363641262054443,
      "learning_rate": 0.000535809261914549,
      "loss": 2.3809,
      "step": 48876
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.6732969880104065,
      "learning_rate": 0.0005358067331592797,
      "loss": 2.3611,
      "step": 48877
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.6801537275314331,
      "learning_rate": 0.0005358042043601694,
      "loss": 2.2802,
      "step": 48878
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.8250161409378052,
      "learning_rate": 0.0005358016755172187,
      "loss": 2.2178,
      "step": 48879
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7620136141777039,
      "learning_rate": 0.0005357991466304281,
      "loss": 2.2934,
      "step": 48880
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.6765807271003723,
      "learning_rate": 0.0005357966176997977,
      "loss": 2.3425,
      "step": 48881
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.6846750974655151,
      "learning_rate": 0.0005357940887253287,
      "loss": 2.1378,
      "step": 48882
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.6750237345695496,
      "learning_rate": 0.0005357915597070207,
      "loss": 2.361,
      "step": 48883
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7415711283683777,
      "learning_rate": 0.0005357890306448749,
      "loss": 2.1253,
      "step": 48884
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7211240530014038,
      "learning_rate": 0.0005357865015388914,
      "loss": 2.2086,
      "step": 48885
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.6741483211517334,
      "learning_rate": 0.0005357839723890707,
      "loss": 2.2825,
      "step": 48886
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7048454284667969,
      "learning_rate": 0.0005357814431954132,
      "loss": 2.3682,
      "step": 48887
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7214164137840271,
      "learning_rate": 0.0005357789139579196,
      "loss": 2.1504,
      "step": 48888
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.776649534702301,
      "learning_rate": 0.0005357763846765902,
      "loss": 2.2491,
      "step": 48889
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.6752675175666809,
      "learning_rate": 0.0005357738553514256,
      "loss": 2.2093,
      "step": 48890
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.6526601314544678,
      "learning_rate": 0.0005357713259824262,
      "loss": 2.4382,
      "step": 48891
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.6850563883781433,
      "learning_rate": 0.0005357687965695925,
      "loss": 2.2243,
      "step": 48892
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.6856170296669006,
      "learning_rate": 0.0005357662671129247,
      "loss": 2.0531,
      "step": 48893
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.695500910282135,
      "learning_rate": 0.0005357637376124237,
      "loss": 2.3445,
      "step": 48894
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.6695980429649353,
      "learning_rate": 0.0005357612080680897,
      "loss": 2.1731,
      "step": 48895
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.719998836517334,
      "learning_rate": 0.0005357586784799233,
      "loss": 2.1361,
      "step": 48896
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7074130773544312,
      "learning_rate": 0.0005357561488479248,
      "loss": 2.2946,
      "step": 48897
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7324169278144836,
      "learning_rate": 0.0005357536191720948,
      "loss": 2.3096,
      "step": 48898
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7535043358802795,
      "learning_rate": 0.0005357510894524339,
      "loss": 2.1928,
      "step": 48899
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7183082103729248,
      "learning_rate": 0.0005357485596889423,
      "loss": 2.3869,
      "step": 48900
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.6915631890296936,
      "learning_rate": 0.0005357460298816205,
      "loss": 2.2966,
      "step": 48901
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.6627455949783325,
      "learning_rate": 0.0005357435000304693,
      "loss": 2.2651,
      "step": 48902
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7163479924201965,
      "learning_rate": 0.0005357409701354887,
      "loss": 2.3109,
      "step": 48903
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.9234139919281006,
      "learning_rate": 0.0005357384401966796,
      "loss": 2.4155,
      "step": 48904
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7055037617683411,
      "learning_rate": 0.0005357359102140421,
      "loss": 2.0769,
      "step": 48905
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.6530057787895203,
      "learning_rate": 0.000535733380187577,
      "loss": 2.2579,
      "step": 48906
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7328920364379883,
      "learning_rate": 0.0005357308501172845,
      "loss": 2.3489,
      "step": 48907
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.6906182765960693,
      "learning_rate": 0.0005357283200031652,
      "loss": 2.298,
      "step": 48908
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.8898710608482361,
      "learning_rate": 0.0005357257898452196,
      "loss": 2.3912,
      "step": 48909
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7127282619476318,
      "learning_rate": 0.0005357232596434481,
      "loss": 2.2339,
      "step": 48910
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.727056622505188,
      "learning_rate": 0.0005357207293978513,
      "loss": 2.3464,
      "step": 48911
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.6680203080177307,
      "learning_rate": 0.0005357181991084294,
      "loss": 2.4984,
      "step": 48912
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.778387188911438,
      "learning_rate": 0.0005357156687751832,
      "loss": 2.3871,
      "step": 48913
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.6889321804046631,
      "learning_rate": 0.0005357131383981129,
      "loss": 2.394,
      "step": 48914
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.9197803735733032,
      "learning_rate": 0.000535710607977219,
      "loss": 2.2736,
      "step": 48915
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.742656409740448,
      "learning_rate": 0.0005357080775125021,
      "loss": 2.285,
      "step": 48916
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.719829797744751,
      "learning_rate": 0.0005357055470039628,
      "loss": 2.429,
      "step": 48917
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7239642143249512,
      "learning_rate": 0.0005357030164516013,
      "loss": 2.405,
      "step": 48918
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7015807628631592,
      "learning_rate": 0.000535700485855418,
      "loss": 2.3648,
      "step": 48919
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.690436840057373,
      "learning_rate": 0.0005356979552154136,
      "loss": 2.083,
      "step": 48920
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.688194751739502,
      "learning_rate": 0.0005356954245315886,
      "loss": 2.327,
      "step": 48921
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7451428174972534,
      "learning_rate": 0.0005356928938039433,
      "loss": 2.1335,
      "step": 48922
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7540997862815857,
      "learning_rate": 0.0005356903630324782,
      "loss": 2.2291,
      "step": 48923
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7641006112098694,
      "learning_rate": 0.0005356878322171939,
      "loss": 2.4091,
      "step": 48924
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.8231427073478699,
      "learning_rate": 0.0005356853013580908,
      "loss": 2.1326,
      "step": 48925
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7838988900184631,
      "learning_rate": 0.0005356827704551693,
      "loss": 2.3744,
      "step": 48926
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.9809331297874451,
      "learning_rate": 0.00053568023950843,
      "loss": 2.2239,
      "step": 48927
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7781950235366821,
      "learning_rate": 0.0005356777085178731,
      "loss": 2.0507,
      "step": 48928
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7301579713821411,
      "learning_rate": 0.0005356751774834995,
      "loss": 2.4678,
      "step": 48929
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.6881037950515747,
      "learning_rate": 0.0005356726464053092,
      "loss": 2.128,
      "step": 48930
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7846614122390747,
      "learning_rate": 0.0005356701152833031,
      "loss": 2.4049,
      "step": 48931
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7241126298904419,
      "learning_rate": 0.0005356675841174814,
      "loss": 2.0503,
      "step": 48932
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.916799008846283,
      "learning_rate": 0.0005356650529078446,
      "loss": 2.2036,
      "step": 48933
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7864883542060852,
      "learning_rate": 0.0005356625216543934,
      "loss": 2.3506,
      "step": 48934
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7155210375785828,
      "learning_rate": 0.0005356599903571279,
      "loss": 2.4502,
      "step": 48935
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7786173224449158,
      "learning_rate": 0.0005356574590160488,
      "loss": 2.1853,
      "step": 48936
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.8173889517784119,
      "learning_rate": 0.0005356549276311567,
      "loss": 2.3039,
      "step": 48937
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.8647781014442444,
      "learning_rate": 0.0005356523962024517,
      "loss": 2.3833,
      "step": 48938
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.6828774809837341,
      "learning_rate": 0.0005356498647299345,
      "loss": 2.2935,
      "step": 48939
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.6752071976661682,
      "learning_rate": 0.0005356473332136056,
      "loss": 2.206,
      "step": 48940
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7534995675086975,
      "learning_rate": 0.0005356448016534653,
      "loss": 2.1973,
      "step": 48941
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7738804221153259,
      "learning_rate": 0.0005356422700495142,
      "loss": 2.3193,
      "step": 48942
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.6988157629966736,
      "learning_rate": 0.0005356397384017529,
      "loss": 2.2953,
      "step": 48943
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.015737771987915,
      "learning_rate": 0.0005356372067101816,
      "loss": 2.3978,
      "step": 48944
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7014276385307312,
      "learning_rate": 0.000535634674974801,
      "loss": 2.1874,
      "step": 48945
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.6742119193077087,
      "learning_rate": 0.0005356321431956115,
      "loss": 2.0812,
      "step": 48946
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7650135159492493,
      "learning_rate": 0.0005356296113726134,
      "loss": 2.1516,
      "step": 48947
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.6640363335609436,
      "learning_rate": 0.0005356270795058073,
      "loss": 2.0779,
      "step": 48948
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7699441313743591,
      "learning_rate": 0.0005356245475951938,
      "loss": 2.2636,
      "step": 48949
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.71695476770401,
      "learning_rate": 0.0005356220156407731,
      "loss": 2.0878,
      "step": 48950
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.8746936917304993,
      "learning_rate": 0.0005356194836425458,
      "loss": 2.2059,
      "step": 48951
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.737292468547821,
      "learning_rate": 0.0005356169516005126,
      "loss": 2.2702,
      "step": 48952
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.6700608730316162,
      "learning_rate": 0.0005356144195146737,
      "loss": 2.2349,
      "step": 48953
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.697897732257843,
      "learning_rate": 0.0005356118873850295,
      "loss": 2.094,
      "step": 48954
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7533440589904785,
      "learning_rate": 0.0005356093552115807,
      "loss": 2.2157,
      "step": 48955
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7509734630584717,
      "learning_rate": 0.0005356068229943276,
      "loss": 2.237,
      "step": 48956
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7339149713516235,
      "learning_rate": 0.0005356042907332708,
      "loss": 2.3435,
      "step": 48957
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7909101247787476,
      "learning_rate": 0.0005356017584284107,
      "loss": 2.3229,
      "step": 48958
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7438272833824158,
      "learning_rate": 0.0005355992260797478,
      "loss": 2.3136,
      "step": 48959
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.6519727110862732,
      "learning_rate": 0.0005355966936872827,
      "loss": 2.0149,
      "step": 48960
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.714226484298706,
      "learning_rate": 0.0005355941612510154,
      "loss": 2.3098,
      "step": 48961
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7830196022987366,
      "learning_rate": 0.000535591628770947,
      "loss": 2.3619,
      "step": 48962
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7461931109428406,
      "learning_rate": 0.0005355890962470775,
      "loss": 2.0562,
      "step": 48963
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.6214214563369751,
      "learning_rate": 0.0005355865636794076,
      "loss": 2.0178,
      "step": 48964
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7792298197746277,
      "learning_rate": 0.0005355840310679377,
      "loss": 2.3182,
      "step": 48965
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.8274708986282349,
      "learning_rate": 0.0005355814984126683,
      "loss": 2.288,
      "step": 48966
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7011151909828186,
      "learning_rate": 0.0005355789657135998,
      "loss": 2.2426,
      "step": 48967
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.6909758448600769,
      "learning_rate": 0.0005355764329707328,
      "loss": 2.2902,
      "step": 48968
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.0786616802215576,
      "learning_rate": 0.0005355739001840676,
      "loss": 2.1312,
      "step": 48969
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7230919599533081,
      "learning_rate": 0.0005355713673536048,
      "loss": 2.424,
      "step": 48970
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.8748565912246704,
      "learning_rate": 0.0005355688344793448,
      "loss": 2.0937,
      "step": 48971
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.9495267271995544,
      "learning_rate": 0.0005355663015612882,
      "loss": 2.2408,
      "step": 48972
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.6922974586486816,
      "learning_rate": 0.0005355637685994354,
      "loss": 2.2537,
      "step": 48973
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7014470100402832,
      "learning_rate": 0.0005355612355937867,
      "loss": 2.1102,
      "step": 48974
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.8242380023002625,
      "learning_rate": 0.0005355587025443429,
      "loss": 2.2117,
      "step": 48975
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7910634875297546,
      "learning_rate": 0.0005355561694511042,
      "loss": 2.1512,
      "step": 48976
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.6911534070968628,
      "learning_rate": 0.0005355536363140711,
      "loss": 2.2626,
      "step": 48977
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.701346218585968,
      "learning_rate": 0.0005355511031332442,
      "loss": 2.5409,
      "step": 48978
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7728055715560913,
      "learning_rate": 0.0005355485699086239,
      "loss": 2.3131,
      "step": 48979
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7241725921630859,
      "learning_rate": 0.0005355460366402107,
      "loss": 2.1418,
      "step": 48980
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.6552329063415527,
      "learning_rate": 0.000535543503328005,
      "loss": 2.2877,
      "step": 48981
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.791452944278717,
      "learning_rate": 0.0005355409699720074,
      "loss": 2.107,
      "step": 48982
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.8021472096443176,
      "learning_rate": 0.0005355384365722182,
      "loss": 2.2127,
      "step": 48983
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.715657651424408,
      "learning_rate": 0.0005355359031286379,
      "loss": 2.3597,
      "step": 48984
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.6314854621887207,
      "learning_rate": 0.0005355333696412672,
      "loss": 2.1779,
      "step": 48985
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.8070630431175232,
      "learning_rate": 0.0005355308361101063,
      "loss": 2.3596,
      "step": 48986
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7142990827560425,
      "learning_rate": 0.0005355283025351557,
      "loss": 2.259,
      "step": 48987
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7779820561408997,
      "learning_rate": 0.000535525768916416,
      "loss": 2.4033,
      "step": 48988
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7856294512748718,
      "learning_rate": 0.0005355232352538878,
      "loss": 2.1649,
      "step": 48989
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.788154125213623,
      "learning_rate": 0.0005355207015475711,
      "loss": 2.2227,
      "step": 48990
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7484437227249146,
      "learning_rate": 0.0005355181677974668,
      "loss": 2.0319,
      "step": 48991
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.852623701095581,
      "learning_rate": 0.0005355156340035752,
      "loss": 2.0766,
      "step": 48992
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7857159376144409,
      "learning_rate": 0.0005355131001658968,
      "loss": 2.323,
      "step": 48993
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7391760349273682,
      "learning_rate": 0.0005355105662844321,
      "loss": 2.1289,
      "step": 48994
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.6970833539962769,
      "learning_rate": 0.0005355080323591815,
      "loss": 2.3008,
      "step": 48995
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7521211504936218,
      "learning_rate": 0.0005355054983901455,
      "loss": 2.2672,
      "step": 48996
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.8216708302497864,
      "learning_rate": 0.0005355029643773246,
      "loss": 2.2643,
      "step": 48997
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.8047768473625183,
      "learning_rate": 0.0005355004303207192,
      "loss": 2.0812,
      "step": 48998
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7545751333236694,
      "learning_rate": 0.00053549789622033,
      "loss": 2.4774,
      "step": 48999
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.9354382157325745,
      "learning_rate": 0.0005354953620761571,
      "loss": 2.2176,
      "step": 49000
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.8079631924629211,
      "learning_rate": 0.0005354928278882013,
      "loss": 2.2499,
      "step": 49001
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.6981251239776611,
      "learning_rate": 0.0005354902936564628,
      "loss": 2.4316,
      "step": 49002
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7440047860145569,
      "learning_rate": 0.0005354877593809423,
      "loss": 2.0938,
      "step": 49003
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7375481724739075,
      "learning_rate": 0.0005354852250616403,
      "loss": 2.2522,
      "step": 49004
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7883282899856567,
      "learning_rate": 0.0005354826906985569,
      "loss": 2.4901,
      "step": 49005
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7551859617233276,
      "learning_rate": 0.000535480156291693,
      "loss": 2.1385,
      "step": 49006
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.972553014755249,
      "learning_rate": 0.0005354776218410487,
      "loss": 2.3805,
      "step": 49007
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7454710006713867,
      "learning_rate": 0.0005354750873466249,
      "loss": 2.1605,
      "step": 49008
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.6670839786529541,
      "learning_rate": 0.0005354725528084218,
      "loss": 2.0887,
      "step": 49009
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.6770071983337402,
      "learning_rate": 0.0005354700182264398,
      "loss": 2.3444,
      "step": 49010
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7922782897949219,
      "learning_rate": 0.0005354674836006795,
      "loss": 2.3622,
      "step": 49011
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.8889883756637573,
      "learning_rate": 0.0005354649489311414,
      "loss": 2.4338,
      "step": 49012
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7394134998321533,
      "learning_rate": 0.000535462414217826,
      "loss": 2.0594,
      "step": 49013
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7509230375289917,
      "learning_rate": 0.0005354598794607335,
      "loss": 2.4043,
      "step": 49014
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.9090343117713928,
      "learning_rate": 0.0005354573446598646,
      "loss": 2.3585,
      "step": 49015
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7248230576515198,
      "learning_rate": 0.0005354548098152199,
      "loss": 2.2506,
      "step": 49016
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7260429263114929,
      "learning_rate": 0.0005354522749267995,
      "loss": 2.301,
      "step": 49017
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.6915185451507568,
      "learning_rate": 0.0005354497399946042,
      "loss": 2.2623,
      "step": 49018
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.715546727180481,
      "learning_rate": 0.0005354472050186344,
      "loss": 2.229,
      "step": 49019
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7316318154335022,
      "learning_rate": 0.0005354446699988904,
      "loss": 2.2224,
      "step": 49020
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7065305113792419,
      "learning_rate": 0.0005354421349353728,
      "loss": 2.1924,
      "step": 49021
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7326291799545288,
      "learning_rate": 0.0005354395998280821,
      "loss": 2.4063,
      "step": 49022
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7827708721160889,
      "learning_rate": 0.0005354370646770188,
      "loss": 2.3592,
      "step": 49023
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.6700818538665771,
      "learning_rate": 0.0005354345294821832,
      "loss": 1.9884,
      "step": 49024
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7745131254196167,
      "learning_rate": 0.000535431994243576,
      "loss": 2.0687,
      "step": 49025
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.6970371007919312,
      "learning_rate": 0.0005354294589611974,
      "loss": 2.125,
      "step": 49026
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7064777612686157,
      "learning_rate": 0.0005354269236350481,
      "loss": 2.2995,
      "step": 49027
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.8057313561439514,
      "learning_rate": 0.0005354243882651284,
      "loss": 2.2771,
      "step": 49028
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.6839971542358398,
      "learning_rate": 0.000535421852851439,
      "loss": 2.311,
      "step": 49029
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.8300662040710449,
      "learning_rate": 0.0005354193173939802,
      "loss": 2.3002,
      "step": 49030
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7158800959587097,
      "learning_rate": 0.0005354167818927524,
      "loss": 2.1091,
      "step": 49031
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.6739295125007629,
      "learning_rate": 0.0005354142463477563,
      "loss": 2.3459,
      "step": 49032
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.773920476436615,
      "learning_rate": 0.0005354117107589921,
      "loss": 2.3026,
      "step": 49033
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.685522198677063,
      "learning_rate": 0.0005354091751264606,
      "loss": 2.3521,
      "step": 49034
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7077699899673462,
      "learning_rate": 0.000535406639450162,
      "loss": 2.2916,
      "step": 49035
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7509571313858032,
      "learning_rate": 0.0005354041037300968,
      "loss": 2.3765,
      "step": 49036
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7230043411254883,
      "learning_rate": 0.0005354015679662656,
      "loss": 2.3871,
      "step": 49037
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7435222268104553,
      "learning_rate": 0.0005353990321586688,
      "loss": 2.341,
      "step": 49038
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7030384540557861,
      "learning_rate": 0.0005353964963073069,
      "loss": 2.2941,
      "step": 49039
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.6788095831871033,
      "learning_rate": 0.0005353939604121804,
      "loss": 2.0979,
      "step": 49040
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.6986241340637207,
      "learning_rate": 0.0005353914244732897,
      "loss": 2.1161,
      "step": 49041
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.6740913391113281,
      "learning_rate": 0.0005353888884906352,
      "loss": 2.3301,
      "step": 49042
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7056345343589783,
      "learning_rate": 0.0005353863524642174,
      "loss": 2.3198,
      "step": 49043
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7444031834602356,
      "learning_rate": 0.000535383816394037,
      "loss": 2.2982,
      "step": 49044
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7216590046882629,
      "learning_rate": 0.0005353812802800943,
      "loss": 2.0996,
      "step": 49045
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.710666835308075,
      "learning_rate": 0.0005353787441223897,
      "loss": 2.1596,
      "step": 49046
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.762818455696106,
      "learning_rate": 0.0005353762079209238,
      "loss": 2.1578,
      "step": 49047
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7330090999603271,
      "learning_rate": 0.000535373671675697,
      "loss": 2.5654,
      "step": 49048
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.6920520067214966,
      "learning_rate": 0.0005353711353867097,
      "loss": 2.3123,
      "step": 49049
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.6723033785820007,
      "learning_rate": 0.0005353685990539626,
      "loss": 2.2554,
      "step": 49050
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7429284453392029,
      "learning_rate": 0.0005353660626774559,
      "loss": 2.4198,
      "step": 49051
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.6900949478149414,
      "learning_rate": 0.0005353635262571903,
      "loss": 2.0619,
      "step": 49052
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7044937610626221,
      "learning_rate": 0.0005353609897931662,
      "loss": 2.0572,
      "step": 49053
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7210456728935242,
      "learning_rate": 0.000535358453285384,
      "loss": 2.1761,
      "step": 49054
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7293156981468201,
      "learning_rate": 0.0005353559167338442,
      "loss": 2.2294,
      "step": 49055
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7544364929199219,
      "learning_rate": 0.0005353533801385473,
      "loss": 2.4425,
      "step": 49056
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7927235960960388,
      "learning_rate": 0.0005353508434994938,
      "loss": 2.2434,
      "step": 49057
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.8466070890426636,
      "learning_rate": 0.0005353483068166841,
      "loss": 2.37,
      "step": 49058
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7877885103225708,
      "learning_rate": 0.0005353457700901187,
      "loss": 2.1819,
      "step": 49059
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.76670902967453,
      "learning_rate": 0.0005353432333197981,
      "loss": 2.3523,
      "step": 49060
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.6741406917572021,
      "learning_rate": 0.0005353406965057228,
      "loss": 2.3457,
      "step": 49061
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7253823280334473,
      "learning_rate": 0.0005353381596478932,
      "loss": 2.1297,
      "step": 49062
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.6797540783882141,
      "learning_rate": 0.0005353356227463099,
      "loss": 2.3274,
      "step": 49063
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7412298917770386,
      "learning_rate": 0.0005353330858009731,
      "loss": 2.3044,
      "step": 49064
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.6299030780792236,
      "learning_rate": 0.0005353305488118835,
      "loss": 1.9608,
      "step": 49065
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.6826707124710083,
      "learning_rate": 0.0005353280117790415,
      "loss": 1.9749,
      "step": 49066
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.6945266723632812,
      "learning_rate": 0.0005353254747024475,
      "loss": 2.1995,
      "step": 49067
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7218639254570007,
      "learning_rate": 0.0005353229375821021,
      "loss": 2.3094,
      "step": 49068
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.8134039640426636,
      "learning_rate": 0.0005353204004180058,
      "loss": 2.378,
      "step": 49069
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7690136432647705,
      "learning_rate": 0.000535317863210159,
      "loss": 2.2727,
      "step": 49070
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7092934250831604,
      "learning_rate": 0.000535315325958562,
      "loss": 2.0311,
      "step": 49071
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7514459490776062,
      "learning_rate": 0.0005353127886632155,
      "loss": 2.0963,
      "step": 49072
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7771769762039185,
      "learning_rate": 0.00053531025132412,
      "loss": 2.3097,
      "step": 49073
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7036787271499634,
      "learning_rate": 0.000535307713941276,
      "loss": 2.4205,
      "step": 49074
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7357669472694397,
      "learning_rate": 0.0005353051765146835,
      "loss": 2.1402,
      "step": 49075
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7844156622886658,
      "learning_rate": 0.0005353026390443436,
      "loss": 2.1813,
      "step": 49076
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.6832780838012695,
      "learning_rate": 0.0005353001015302564,
      "loss": 2.069,
      "step": 49077
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.6695463061332703,
      "learning_rate": 0.0005352975639724225,
      "loss": 2.2128,
      "step": 49078
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7035385370254517,
      "learning_rate": 0.0005352950263708423,
      "loss": 2.3336,
      "step": 49079
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7710220813751221,
      "learning_rate": 0.0005352924887255164,
      "loss": 2.1663,
      "step": 49080
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.8479697704315186,
      "learning_rate": 0.000535289951036445,
      "loss": 2.3217,
      "step": 49081
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7352281808853149,
      "learning_rate": 0.000535287413303629,
      "loss": 2.254,
      "step": 49082
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7612752318382263,
      "learning_rate": 0.0005352848755270685,
      "loss": 2.1736,
      "step": 49083
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7207712531089783,
      "learning_rate": 0.000535282337706764,
      "loss": 2.3379,
      "step": 49084
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.96537846326828,
      "learning_rate": 0.0005352797998427162,
      "loss": 2.1111,
      "step": 49085
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.6299002170562744,
      "learning_rate": 0.0005352772619349255,
      "loss": 2.0907,
      "step": 49086
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7368427515029907,
      "learning_rate": 0.0005352747239833922,
      "loss": 2.0959,
      "step": 49087
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.864186704158783,
      "learning_rate": 0.000535272185988117,
      "loss": 1.928,
      "step": 49088
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7103257775306702,
      "learning_rate": 0.0005352696479491002,
      "loss": 2.4141,
      "step": 49089
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7704704999923706,
      "learning_rate": 0.0005352671098663422,
      "loss": 2.2065,
      "step": 49090
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.6513926982879639,
      "learning_rate": 0.0005352645717398438,
      "loss": 2.2698,
      "step": 49091
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.861831545829773,
      "learning_rate": 0.0005352620335696052,
      "loss": 2.3734,
      "step": 49092
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7801442742347717,
      "learning_rate": 0.0005352594953556269,
      "loss": 2.39,
      "step": 49093
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.8399136066436768,
      "learning_rate": 0.0005352569570979095,
      "loss": 2.0752,
      "step": 49094
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7679632902145386,
      "learning_rate": 0.0005352544187964533,
      "loss": 2.4113,
      "step": 49095
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7242507934570312,
      "learning_rate": 0.0005352518804512589,
      "loss": 2.4175,
      "step": 49096
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.6969655156135559,
      "learning_rate": 0.0005352493420623268,
      "loss": 2.3015,
      "step": 49097
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7692897319793701,
      "learning_rate": 0.0005352468036296575,
      "loss": 2.1815,
      "step": 49098
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.633030891418457,
      "learning_rate": 0.0005352442651532512,
      "loss": 2.1472,
      "step": 49099
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.777805745601654,
      "learning_rate": 0.0005352417266331085,
      "loss": 2.2588,
      "step": 49100
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7270414233207703,
      "learning_rate": 0.00053523918806923,
      "loss": 2.1964,
      "step": 49101
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.8077296614646912,
      "learning_rate": 0.0005352366494616161,
      "loss": 2.2688,
      "step": 49102
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7097076177597046,
      "learning_rate": 0.0005352341108102673,
      "loss": 2.2486,
      "step": 49103
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.6848281621932983,
      "learning_rate": 0.000535231572115184,
      "loss": 2.3574,
      "step": 49104
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.6531878709793091,
      "learning_rate": 0.0005352290333763667,
      "loss": 2.3266,
      "step": 49105
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7080153226852417,
      "learning_rate": 0.000535226494593816,
      "loss": 2.2604,
      "step": 49106
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.6682561039924622,
      "learning_rate": 0.0005352239557675321,
      "loss": 2.2716,
      "step": 49107
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.6978935599327087,
      "learning_rate": 0.0005352214168975156,
      "loss": 2.171,
      "step": 49108
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.6956207752227783,
      "learning_rate": 0.0005352188779837671,
      "loss": 2.4414,
      "step": 49109
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.6451753377914429,
      "learning_rate": 0.000535216339026287,
      "loss": 2.0622,
      "step": 49110
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7058696746826172,
      "learning_rate": 0.0005352138000250758,
      "loss": 2.3181,
      "step": 49111
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7446126341819763,
      "learning_rate": 0.0005352112609801337,
      "loss": 2.3241,
      "step": 49112
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.666957676410675,
      "learning_rate": 0.0005352087218914614,
      "loss": 2.128,
      "step": 49113
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.8105974197387695,
      "learning_rate": 0.0005352061827590596,
      "loss": 2.0883,
      "step": 49114
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.6891926527023315,
      "learning_rate": 0.0005352036435829284,
      "loss": 2.1841,
      "step": 49115
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7269241213798523,
      "learning_rate": 0.0005352011043630683,
      "loss": 2.0627,
      "step": 49116
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7067160606384277,
      "learning_rate": 0.00053519856509948,
      "loss": 2.2889,
      "step": 49117
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.6914124488830566,
      "learning_rate": 0.0005351960257921639,
      "loss": 2.1364,
      "step": 49118
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.6976813673973083,
      "learning_rate": 0.0005351934864411203,
      "loss": 2.2924,
      "step": 49119
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.6756740212440491,
      "learning_rate": 0.0005351909470463498,
      "loss": 2.344,
      "step": 49120
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.66470867395401,
      "learning_rate": 0.0005351884076078529,
      "loss": 2.2245,
      "step": 49121
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.664164125919342,
      "learning_rate": 0.00053518586812563,
      "loss": 2.3102,
      "step": 49122
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.6590579748153687,
      "learning_rate": 0.0005351833285996817,
      "loss": 2.2756,
      "step": 49123
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.666857123374939,
      "learning_rate": 0.0005351807890300083,
      "loss": 2.2203,
      "step": 49124
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.74948650598526,
      "learning_rate": 0.0005351782494166103,
      "loss": 2.3863,
      "step": 49125
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7105840444564819,
      "learning_rate": 0.0005351757097594883,
      "loss": 2.2874,
      "step": 49126
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.6800532937049866,
      "learning_rate": 0.0005351731700586426,
      "loss": 2.1511,
      "step": 49127
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7568538784980774,
      "learning_rate": 0.000535170630314074,
      "loss": 2.2776,
      "step": 49128
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7243966460227966,
      "learning_rate": 0.0005351680905257825,
      "loss": 2.178,
      "step": 49129
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.6691868305206299,
      "learning_rate": 0.0005351655506937689,
      "loss": 2.1797,
      "step": 49130
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7560250163078308,
      "learning_rate": 0.0005351630108180336,
      "loss": 2.2401,
      "step": 49131
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.6644378304481506,
      "learning_rate": 0.0005351604708985771,
      "loss": 2.1892,
      "step": 49132
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.6913487315177917,
      "learning_rate": 0.0005351579309353997,
      "loss": 2.5048,
      "step": 49133
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.6948457360267639,
      "learning_rate": 0.0005351553909285021,
      "loss": 2.2652,
      "step": 49134
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.6993255615234375,
      "learning_rate": 0.0005351528508778846,
      "loss": 2.2979,
      "step": 49135
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7092933654785156,
      "learning_rate": 0.0005351503107835478,
      "loss": 2.163,
      "step": 49136
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7391199469566345,
      "learning_rate": 0.0005351477706454922,
      "loss": 2.0628,
      "step": 49137
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.6935263872146606,
      "learning_rate": 0.0005351452304637181,
      "loss": 2.0971,
      "step": 49138
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7046499252319336,
      "learning_rate": 0.000535142690238226,
      "loss": 2.1873,
      "step": 49139
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.6887290477752686,
      "learning_rate": 0.0005351401499690166,
      "loss": 2.3144,
      "step": 49140
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.6704413294792175,
      "learning_rate": 0.00053513760965609,
      "loss": 2.3407,
      "step": 49141
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.8198921084403992,
      "learning_rate": 0.0005351350692994471,
      "loss": 2.2738,
      "step": 49142
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.6862990856170654,
      "learning_rate": 0.000535132528899088,
      "loss": 2.1743,
      "step": 49143
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7139899730682373,
      "learning_rate": 0.0005351299884550134,
      "loss": 2.2777,
      "step": 49144
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.8440712094306946,
      "learning_rate": 0.0005351274479672236,
      "loss": 2.3271,
      "step": 49145
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7185291051864624,
      "learning_rate": 0.0005351249074357192,
      "loss": 2.2436,
      "step": 49146
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7307596802711487,
      "learning_rate": 0.0005351223668605007,
      "loss": 2.0859,
      "step": 49147
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7139830589294434,
      "learning_rate": 0.0005351198262415686,
      "loss": 2.1864,
      "step": 49148
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7189980149269104,
      "learning_rate": 0.000535117285578923,
      "loss": 2.3175,
      "step": 49149
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7475433349609375,
      "learning_rate": 0.0005351147448725649,
      "loss": 2.4253,
      "step": 49150
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7023254036903381,
      "learning_rate": 0.0005351122041224945,
      "loss": 2.3049,
      "step": 49151
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7393055558204651,
      "learning_rate": 0.0005351096633287122,
      "loss": 2.332,
      "step": 49152
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.6330044269561768,
      "learning_rate": 0.0005351071224912187,
      "loss": 2.1593,
      "step": 49153
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7650654911994934,
      "learning_rate": 0.0005351045816100143,
      "loss": 2.3741,
      "step": 49154
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7798937559127808,
      "learning_rate": 0.0005351020406850995,
      "loss": 2.1388,
      "step": 49155
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.728756844997406,
      "learning_rate": 0.0005350994997164747,
      "loss": 2.2799,
      "step": 49156
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.6680770516395569,
      "learning_rate": 0.0005350969587041406,
      "loss": 2.3215,
      "step": 49157
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7364671230316162,
      "learning_rate": 0.0005350944176480976,
      "loss": 2.1764,
      "step": 49158
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.686238706111908,
      "learning_rate": 0.000535091876548346,
      "loss": 2.3495,
      "step": 49159
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.9362075924873352,
      "learning_rate": 0.0005350893354048864,
      "loss": 2.1828,
      "step": 49160
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.8442965745925903,
      "learning_rate": 0.0005350867942177193,
      "loss": 2.3116,
      "step": 49161
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7890759706497192,
      "learning_rate": 0.000535084252986845,
      "loss": 2.2695,
      "step": 49162
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.8951292634010315,
      "learning_rate": 0.0005350817117122643,
      "loss": 2.2831,
      "step": 49163
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7841426134109497,
      "learning_rate": 0.0005350791703939772,
      "loss": 2.1542,
      "step": 49164
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.8336747288703918,
      "learning_rate": 0.0005350766290319847,
      "loss": 2.4768,
      "step": 49165
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7903900146484375,
      "learning_rate": 0.0005350740876262869,
      "loss": 2.2271,
      "step": 49166
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7048375010490417,
      "learning_rate": 0.0005350715461768843,
      "loss": 2.4871,
      "step": 49167
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7631810903549194,
      "learning_rate": 0.0005350690046837776,
      "loss": 2.1813,
      "step": 49168
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7386415004730225,
      "learning_rate": 0.0005350664631469671,
      "loss": 2.1598,
      "step": 49169
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.8707942962646484,
      "learning_rate": 0.0005350639215664534,
      "loss": 2.4649,
      "step": 49170
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7371702194213867,
      "learning_rate": 0.0005350613799422368,
      "loss": 2.2463,
      "step": 49171
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7142382264137268,
      "learning_rate": 0.0005350588382743178,
      "loss": 2.3454,
      "step": 49172
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7160491943359375,
      "learning_rate": 0.0005350562965626969,
      "loss": 2.2602,
      "step": 49173
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.694585919380188,
      "learning_rate": 0.0005350537548073748,
      "loss": 1.9974,
      "step": 49174
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7014217972755432,
      "learning_rate": 0.0005350512130083516,
      "loss": 2.2043,
      "step": 49175
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.872317910194397,
      "learning_rate": 0.000535048671165628,
      "loss": 2.2224,
      "step": 49176
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7733250260353088,
      "learning_rate": 0.0005350461292792044,
      "loss": 2.2788,
      "step": 49177
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7555264234542847,
      "learning_rate": 0.0005350435873490812,
      "loss": 2.4104,
      "step": 49178
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7594394683837891,
      "learning_rate": 0.0005350410453752591,
      "loss": 2.1506,
      "step": 49179
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7802370190620422,
      "learning_rate": 0.0005350385033577384,
      "loss": 2.3042,
      "step": 49180
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7890004515647888,
      "learning_rate": 0.0005350359612965195,
      "loss": 2.1779,
      "step": 49181
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7057148814201355,
      "learning_rate": 0.0005350334191916031,
      "loss": 2.2089,
      "step": 49182
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.8316115140914917,
      "learning_rate": 0.0005350308770429894,
      "loss": 2.0933,
      "step": 49183
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.8027527928352356,
      "learning_rate": 0.0005350283348506792,
      "loss": 2.2521,
      "step": 49184
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7333092093467712,
      "learning_rate": 0.0005350257926146727,
      "loss": 2.049,
      "step": 49185
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.8307936787605286,
      "learning_rate": 0.0005350232503349706,
      "loss": 2.2439,
      "step": 49186
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7128195762634277,
      "learning_rate": 0.000535020708011573,
      "loss": 2.2339,
      "step": 49187
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.6988159418106079,
      "learning_rate": 0.0005350181656444808,
      "loss": 2.2658,
      "step": 49188
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7265756130218506,
      "learning_rate": 0.0005350156232336942,
      "loss": 2.3954,
      "step": 49189
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7354841828346252,
      "learning_rate": 0.0005350130807792138,
      "loss": 2.4601,
      "step": 49190
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7000643610954285,
      "learning_rate": 0.00053501053828104,
      "loss": 2.0709,
      "step": 49191
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.6809060573577881,
      "learning_rate": 0.0005350079957391733,
      "loss": 2.1344,
      "step": 49192
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.9460771679878235,
      "learning_rate": 0.0005350054531536143,
      "loss": 2.1177,
      "step": 49193
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.699663519859314,
      "learning_rate": 0.0005350029105243633,
      "loss": 2.3646,
      "step": 49194
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.6174222230911255,
      "learning_rate": 0.0005350003678514208,
      "loss": 2.0661,
      "step": 49195
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7008878588676453,
      "learning_rate": 0.0005349978251347872,
      "loss": 2.2013,
      "step": 49196
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.8726277947425842,
      "learning_rate": 0.0005349952823744631,
      "loss": 2.1582,
      "step": 49197
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.8101000785827637,
      "learning_rate": 0.000534992739570449,
      "loss": 2.0768,
      "step": 49198
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.6853649020195007,
      "learning_rate": 0.0005349901967227455,
      "loss": 2.0908,
      "step": 49199
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7848341464996338,
      "learning_rate": 0.0005349876538313525,
      "loss": 2.1073,
      "step": 49200
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7669215798377991,
      "learning_rate": 0.0005349851108962711,
      "loss": 2.0318,
      "step": 49201
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7411696910858154,
      "learning_rate": 0.0005349825679175015,
      "loss": 2.099,
      "step": 49202
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.6564363837242126,
      "learning_rate": 0.0005349800248950441,
      "loss": 2.2891,
      "step": 49203
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7523587346076965,
      "learning_rate": 0.0005349774818288997,
      "loss": 2.3234,
      "step": 49204
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7371991276741028,
      "learning_rate": 0.0005349749387190683,
      "loss": 2.1322,
      "step": 49205
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7555482983589172,
      "learning_rate": 0.0005349723955655508,
      "loss": 2.2741,
      "step": 49206
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7648858428001404,
      "learning_rate": 0.0005349698523683474,
      "loss": 2.2309,
      "step": 49207
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7966459393501282,
      "learning_rate": 0.0005349673091274588,
      "loss": 2.3457,
      "step": 49208
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7632154226303101,
      "learning_rate": 0.0005349647658428852,
      "loss": 2.3,
      "step": 49209
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.68900465965271,
      "learning_rate": 0.0005349622225146272,
      "loss": 2.0395,
      "step": 49210
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7375913262367249,
      "learning_rate": 0.0005349596791426853,
      "loss": 2.3363,
      "step": 49211
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7323498725891113,
      "learning_rate": 0.0005349571357270601,
      "loss": 2.3629,
      "step": 49212
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7288216948509216,
      "learning_rate": 0.0005349545922677518,
      "loss": 2.3504,
      "step": 49213
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7190176248550415,
      "learning_rate": 0.000534952048764761,
      "loss": 2.4844,
      "step": 49214
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7854437828063965,
      "learning_rate": 0.0005349495052180882,
      "loss": 2.0388,
      "step": 49215
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.8033376932144165,
      "learning_rate": 0.0005349469616277339,
      "loss": 2.2341,
      "step": 49216
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.6909406781196594,
      "learning_rate": 0.0005349444179936986,
      "loss": 2.1392,
      "step": 49217
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7373902797698975,
      "learning_rate": 0.0005349418743159826,
      "loss": 2.2212,
      "step": 49218
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.798932671546936,
      "learning_rate": 0.0005349393305945864,
      "loss": 2.3948,
      "step": 49219
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7707483768463135,
      "learning_rate": 0.0005349367868295107,
      "loss": 2.2375,
      "step": 49220
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7729897499084473,
      "learning_rate": 0.0005349342430207557,
      "loss": 2.0472,
      "step": 49221
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.8445504307746887,
      "learning_rate": 0.0005349316991683221,
      "loss": 2.2466,
      "step": 49222
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.9088152647018433,
      "learning_rate": 0.0005349291552722101,
      "loss": 2.2303,
      "step": 49223
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.741374135017395,
      "learning_rate": 0.0005349266113324204,
      "loss": 2.1899,
      "step": 49224
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.6900179386138916,
      "learning_rate": 0.0005349240673489534,
      "loss": 2.3104,
      "step": 49225
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.8606805205345154,
      "learning_rate": 0.0005349215233218097,
      "loss": 2.3013,
      "step": 49226
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7296563386917114,
      "learning_rate": 0.0005349189792509895,
      "loss": 2.3089,
      "step": 49227
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7078846096992493,
      "learning_rate": 0.0005349164351364935,
      "loss": 2.3023,
      "step": 49228
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7887652516365051,
      "learning_rate": 0.0005349138909783221,
      "loss": 2.4143,
      "step": 49229
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.8283638954162598,
      "learning_rate": 0.0005349113467764758,
      "loss": 2.1842,
      "step": 49230
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7680617570877075,
      "learning_rate": 0.0005349088025309551,
      "loss": 2.3136,
      "step": 49231
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7500590085983276,
      "learning_rate": 0.0005349062582417603,
      "loss": 2.3803,
      "step": 49232
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7768341898918152,
      "learning_rate": 0.000534903713908892,
      "loss": 2.4139,
      "step": 49233
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.8985615968704224,
      "learning_rate": 0.0005349011695323507,
      "loss": 2.2218,
      "step": 49234
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7465084195137024,
      "learning_rate": 0.0005348986251121368,
      "loss": 2.1815,
      "step": 49235
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.8096506595611572,
      "learning_rate": 0.0005348960806482508,
      "loss": 2.1547,
      "step": 49236
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.9505186676979065,
      "learning_rate": 0.0005348935361406932,
      "loss": 2.4801,
      "step": 49237
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7258265018463135,
      "learning_rate": 0.0005348909915894645,
      "loss": 2.4681,
      "step": 49238
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7237085103988647,
      "learning_rate": 0.000534888446994565,
      "loss": 1.9279,
      "step": 49239
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.842460572719574,
      "learning_rate": 0.0005348859023559955,
      "loss": 2.2169,
      "step": 49240
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.78658527135849,
      "learning_rate": 0.0005348833576737562,
      "loss": 2.5301,
      "step": 49241
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7901813387870789,
      "learning_rate": 0.0005348808129478477,
      "loss": 2.3231,
      "step": 49242
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7746882438659668,
      "learning_rate": 0.0005348782681782703,
      "loss": 2.3323,
      "step": 49243
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7226450443267822,
      "learning_rate": 0.0005348757233650246,
      "loss": 2.3629,
      "step": 49244
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7702608108520508,
      "learning_rate": 0.0005348731785081112,
      "loss": 2.2718,
      "step": 49245
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.6781686544418335,
      "learning_rate": 0.0005348706336075304,
      "loss": 2.123,
      "step": 49246
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.6828799247741699,
      "learning_rate": 0.0005348680886632827,
      "loss": 2.559,
      "step": 49247
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.776345431804657,
      "learning_rate": 0.0005348655436753685,
      "loss": 2.2235,
      "step": 49248
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.8396440744400024,
      "learning_rate": 0.0005348629986437885,
      "loss": 2.4436,
      "step": 49249
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7070186734199524,
      "learning_rate": 0.0005348604535685429,
      "loss": 2.0797,
      "step": 49250
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7317822575569153,
      "learning_rate": 0.0005348579084496324,
      "loss": 2.3129,
      "step": 49251
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.9254857301712036,
      "learning_rate": 0.0005348553632870574,
      "loss": 2.2895,
      "step": 49252
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.6865043640136719,
      "learning_rate": 0.0005348528180808183,
      "loss": 2.1199,
      "step": 49253
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7072392702102661,
      "learning_rate": 0.0005348502728309156,
      "loss": 1.906,
      "step": 49254
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.6810905337333679,
      "learning_rate": 0.00053484772753735,
      "loss": 2.4032,
      "step": 49255
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.6744783520698547,
      "learning_rate": 0.0005348451822001215,
      "loss": 2.0608,
      "step": 49256
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7886241674423218,
      "learning_rate": 0.000534842636819231,
      "loss": 2.3043,
      "step": 49257
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7568792700767517,
      "learning_rate": 0.0005348400913946788,
      "loss": 2.1918,
      "step": 49258
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7538442015647888,
      "learning_rate": 0.0005348375459264654,
      "loss": 2.1584,
      "step": 49259
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7890930771827698,
      "learning_rate": 0.0005348350004145912,
      "loss": 2.0899,
      "step": 49260
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7465062737464905,
      "learning_rate": 0.0005348324548590568,
      "loss": 2.129,
      "step": 49261
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7013547420501709,
      "learning_rate": 0.0005348299092598626,
      "loss": 2.3722,
      "step": 49262
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.8929629921913147,
      "learning_rate": 0.0005348273636170092,
      "loss": 2.5856,
      "step": 49263
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7947055697441101,
      "learning_rate": 0.0005348248179304968,
      "loss": 2.4282,
      "step": 49264
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7208077907562256,
      "learning_rate": 0.0005348222722003261,
      "loss": 2.134,
      "step": 49265
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.6849069595336914,
      "learning_rate": 0.0005348197264264975,
      "loss": 2.2774,
      "step": 49266
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7292923331260681,
      "learning_rate": 0.0005348171806090114,
      "loss": 2.3762,
      "step": 49267
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7013426423072815,
      "learning_rate": 0.0005348146347478684,
      "loss": 2.1309,
      "step": 49268
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7203457355499268,
      "learning_rate": 0.000534812088843069,
      "loss": 2.2771,
      "step": 49269
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.6780694127082825,
      "learning_rate": 0.0005348095428946136,
      "loss": 2.2676,
      "step": 49270
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7414162755012512,
      "learning_rate": 0.0005348069969025027,
      "loss": 2.244,
      "step": 49271
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.6797047853469849,
      "learning_rate": 0.0005348044508667366,
      "loss": 2.2363,
      "step": 49272
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7486311793327332,
      "learning_rate": 0.0005348019047873159,
      "loss": 2.2447,
      "step": 49273
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.700630784034729,
      "learning_rate": 0.0005347993586642413,
      "loss": 2.3834,
      "step": 49274
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7135674357414246,
      "learning_rate": 0.0005347968124975129,
      "loss": 2.1493,
      "step": 49275
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7349086999893188,
      "learning_rate": 0.0005347942662871315,
      "loss": 2.2876,
      "step": 49276
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7416761517524719,
      "learning_rate": 0.0005347917200330972,
      "loss": 2.1744,
      "step": 49277
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.6883178353309631,
      "learning_rate": 0.0005347891737354109,
      "loss": 2.077,
      "step": 49278
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.688552975654602,
      "learning_rate": 0.0005347866273940727,
      "loss": 2.2889,
      "step": 49279
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7290963530540466,
      "learning_rate": 0.0005347840810090833,
      "loss": 2.4622,
      "step": 49280
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7092231512069702,
      "learning_rate": 0.0005347815345804431,
      "loss": 2.1183,
      "step": 49281
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7919875979423523,
      "learning_rate": 0.0005347789881081526,
      "loss": 2.2913,
      "step": 49282
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7485458254814148,
      "learning_rate": 0.0005347764415922122,
      "loss": 2.3837,
      "step": 49283
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.8299955725669861,
      "learning_rate": 0.0005347738950326224,
      "loss": 2.3266,
      "step": 49284
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.6898947358131409,
      "learning_rate": 0.0005347713484293838,
      "loss": 2.1733,
      "step": 49285
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.0906745195388794,
      "learning_rate": 0.0005347688017824968,
      "loss": 2.4658,
      "step": 49286
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7084176540374756,
      "learning_rate": 0.0005347662550919618,
      "loss": 2.3847,
      "step": 49287
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.6843371391296387,
      "learning_rate": 0.0005347637083577793,
      "loss": 2.091,
      "step": 49288
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7514649033546448,
      "learning_rate": 0.0005347611615799497,
      "loss": 2.0343,
      "step": 49289
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7434332370758057,
      "learning_rate": 0.0005347586147584738,
      "loss": 2.2734,
      "step": 49290
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7618067264556885,
      "learning_rate": 0.0005347560678933517,
      "loss": 2.0349,
      "step": 49291
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.8165395855903625,
      "learning_rate": 0.000534753520984584,
      "loss": 2.1003,
      "step": 49292
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.8834130764007568,
      "learning_rate": 0.0005347509740321712,
      "loss": 2.3565,
      "step": 49293
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.8875500559806824,
      "learning_rate": 0.0005347484270361138,
      "loss": 2.2792,
      "step": 49294
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.8034009337425232,
      "learning_rate": 0.0005347458799964122,
      "loss": 2.4765,
      "step": 49295
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.8062990307807922,
      "learning_rate": 0.000534743332913067,
      "loss": 2.4026,
      "step": 49296
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.8548597693443298,
      "learning_rate": 0.0005347407857860785,
      "loss": 2.3704,
      "step": 49297
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7314960956573486,
      "learning_rate": 0.0005347382386154472,
      "loss": 2.3354,
      "step": 49298
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.777994692325592,
      "learning_rate": 0.0005347356914011737,
      "loss": 2.2158,
      "step": 49299
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7296764850616455,
      "learning_rate": 0.0005347331441432584,
      "loss": 2.427,
      "step": 49300
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7327643036842346,
      "learning_rate": 0.0005347305968417018,
      "loss": 2.4044,
      "step": 49301
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7727518677711487,
      "learning_rate": 0.0005347280494965044,
      "loss": 2.3107,
      "step": 49302
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7273404598236084,
      "learning_rate": 0.0005347255021076665,
      "loss": 2.4557,
      "step": 49303
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7102333307266235,
      "learning_rate": 0.0005347229546751888,
      "loss": 2.4996,
      "step": 49304
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.8677130341529846,
      "learning_rate": 0.0005347204071990716,
      "loss": 2.1222,
      "step": 49305
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7712680697441101,
      "learning_rate": 0.0005347178596793155,
      "loss": 2.2025,
      "step": 49306
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7237389087677002,
      "learning_rate": 0.0005347153121159208,
      "loss": 2.1651,
      "step": 49307
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.709010899066925,
      "learning_rate": 0.0005347127645088882,
      "loss": 2.2538,
      "step": 49308
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.8736658692359924,
      "learning_rate": 0.0005347102168582181,
      "loss": 2.3463,
      "step": 49309
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7489131689071655,
      "learning_rate": 0.000534707669163911,
      "loss": 2.266,
      "step": 49310
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7522092461585999,
      "learning_rate": 0.0005347051214259672,
      "loss": 2.3241,
      "step": 49311
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7644986510276794,
      "learning_rate": 0.0005347025736443872,
      "loss": 2.157,
      "step": 49312
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.6982353925704956,
      "learning_rate": 0.0005347000258191717,
      "loss": 2.0486,
      "step": 49313
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7435779571533203,
      "learning_rate": 0.000534697477950321,
      "loss": 2.3013,
      "step": 49314
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7071394324302673,
      "learning_rate": 0.0005346949300378355,
      "loss": 2.2297,
      "step": 49315
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7245479822158813,
      "learning_rate": 0.0005346923820817159,
      "loss": 2.2013,
      "step": 49316
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.6754049062728882,
      "learning_rate": 0.0005346898340819625,
      "loss": 2.211,
      "step": 49317
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.69937664270401,
      "learning_rate": 0.0005346872860385759,
      "loss": 2.023,
      "step": 49318
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7633039951324463,
      "learning_rate": 0.0005346847379515565,
      "loss": 2.2248,
      "step": 49319
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.6746139526367188,
      "learning_rate": 0.0005346821898209047,
      "loss": 2.2574,
      "step": 49320
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.722314715385437,
      "learning_rate": 0.0005346796416466211,
      "loss": 2.46,
      "step": 49321
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.8013970255851746,
      "learning_rate": 0.0005346770934287062,
      "loss": 2.3435,
      "step": 49322
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.8147562146186829,
      "learning_rate": 0.0005346745451671603,
      "loss": 2.3794,
      "step": 49323
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.709621787071228,
      "learning_rate": 0.0005346719968619839,
      "loss": 2.3195,
      "step": 49324
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.8426899313926697,
      "learning_rate": 0.0005346694485131776,
      "loss": 2.4439,
      "step": 49325
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7104116678237915,
      "learning_rate": 0.000534666900120742,
      "loss": 2.6052,
      "step": 49326
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7788044810295105,
      "learning_rate": 0.0005346643516846773,
      "loss": 2.2551,
      "step": 49327
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.745979905128479,
      "learning_rate": 0.0005346618032049839,
      "loss": 2.3811,
      "step": 49328
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7357288002967834,
      "learning_rate": 0.0005346592546816625,
      "loss": 2.0648,
      "step": 49329
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7847839593887329,
      "learning_rate": 0.0005346567061147137,
      "loss": 2.1267,
      "step": 49330
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7556657791137695,
      "learning_rate": 0.0005346541575041376,
      "loss": 2.2914,
      "step": 49331
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.6640448570251465,
      "learning_rate": 0.000534651608849935,
      "loss": 2.3376,
      "step": 49332
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.6720400452613831,
      "learning_rate": 0.0005346490601521062,
      "loss": 2.0784,
      "step": 49333
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.6980068683624268,
      "learning_rate": 0.0005346465114106515,
      "loss": 2.2538,
      "step": 49334
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7045598030090332,
      "learning_rate": 0.0005346439626255719,
      "loss": 2.1417,
      "step": 49335
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.9074081778526306,
      "learning_rate": 0.0005346414137968674,
      "loss": 2.4864,
      "step": 49336
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7226550579071045,
      "learning_rate": 0.0005346388649245386,
      "loss": 2.2872,
      "step": 49337
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7901366949081421,
      "learning_rate": 0.0005346363160085861,
      "loss": 2.0722,
      "step": 49338
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7149661183357239,
      "learning_rate": 0.0005346337670490102,
      "loss": 2.1028,
      "step": 49339
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.715815007686615,
      "learning_rate": 0.0005346312180458115,
      "loss": 2.2844,
      "step": 49340
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7333125472068787,
      "learning_rate": 0.0005346286689989904,
      "loss": 2.4959,
      "step": 49341
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.6944038271903992,
      "learning_rate": 0.0005346261199085475,
      "loss": 2.0073,
      "step": 49342
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7653986215591431,
      "learning_rate": 0.000534623570774483,
      "loss": 2.3541,
      "step": 49343
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.6898908019065857,
      "learning_rate": 0.0005346210215967977,
      "loss": 2.5513,
      "step": 49344
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7469379901885986,
      "learning_rate": 0.0005346184723754919,
      "loss": 2.2014,
      "step": 49345
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7306823134422302,
      "learning_rate": 0.000534615923110566,
      "loss": 2.1559,
      "step": 49346
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.6728169322013855,
      "learning_rate": 0.0005346133738020207,
      "loss": 2.1071,
      "step": 49347
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7366938591003418,
      "learning_rate": 0.0005346108244498562,
      "loss": 2.2335,
      "step": 49348
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7569550275802612,
      "learning_rate": 0.0005346082750540732,
      "loss": 2.3183,
      "step": 49349
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7595729827880859,
      "learning_rate": 0.000534605725614672,
      "loss": 2.1297,
      "step": 49350
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7791875600814819,
      "learning_rate": 0.0005346031761316533,
      "loss": 2.0182,
      "step": 49351
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7258638143539429,
      "learning_rate": 0.0005346006266050174,
      "loss": 2.0412,
      "step": 49352
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7135723233222961,
      "learning_rate": 0.0005345980770347649,
      "loss": 2.1206,
      "step": 49353
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.6543652415275574,
      "learning_rate": 0.000534595527420896,
      "loss": 2.3209,
      "step": 49354
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7722561359405518,
      "learning_rate": 0.0005345929777634115,
      "loss": 2.1031,
      "step": 49355
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.700203001499176,
      "learning_rate": 0.0005345904280623118,
      "loss": 2.0903,
      "step": 49356
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.6880724430084229,
      "learning_rate": 0.0005345878783175971,
      "loss": 2.3144,
      "step": 49357
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7001926898956299,
      "learning_rate": 0.0005345853285292682,
      "loss": 2.1498,
      "step": 49358
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7162033319473267,
      "learning_rate": 0.0005345827786973255,
      "loss": 2.1979,
      "step": 49359
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7428116202354431,
      "learning_rate": 0.0005345802288217694,
      "loss": 2.2452,
      "step": 49360
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7325230240821838,
      "learning_rate": 0.0005345776789026004,
      "loss": 2.1879,
      "step": 49361
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.6658944487571716,
      "learning_rate": 0.000534575128939819,
      "loss": 2.2584,
      "step": 49362
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7017856240272522,
      "learning_rate": 0.0005345725789334256,
      "loss": 2.2678,
      "step": 49363
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.8007227778434753,
      "learning_rate": 0.0005345700288834208,
      "loss": 2.097,
      "step": 49364
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7234385013580322,
      "learning_rate": 0.0005345674787898049,
      "loss": 2.4076,
      "step": 49365
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.6295056343078613,
      "learning_rate": 0.0005345649286525786,
      "loss": 1.9666,
      "step": 49366
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7665179967880249,
      "learning_rate": 0.0005345623784717422,
      "loss": 2.3276,
      "step": 49367
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7115759253501892,
      "learning_rate": 0.0005345598282472963,
      "loss": 2.3627,
      "step": 49368
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7519029974937439,
      "learning_rate": 0.0005345572779792411,
      "loss": 2.2497,
      "step": 49369
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.6949905157089233,
      "learning_rate": 0.0005345547276675775,
      "loss": 2.3361,
      "step": 49370
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.744138777256012,
      "learning_rate": 0.0005345521773123057,
      "loss": 2.3471,
      "step": 49371
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7596695423126221,
      "learning_rate": 0.0005345496269134261,
      "loss": 2.1442,
      "step": 49372
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.6861199736595154,
      "learning_rate": 0.0005345470764709393,
      "loss": 2.139,
      "step": 49373
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.6742717027664185,
      "learning_rate": 0.0005345445259848459,
      "loss": 2.3632,
      "step": 49374
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.69622403383255,
      "learning_rate": 0.0005345419754551462,
      "loss": 2.322,
      "step": 49375
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.763785719871521,
      "learning_rate": 0.0005345394248818406,
      "loss": 2.247,
      "step": 49376
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.6868879199028015,
      "learning_rate": 0.0005345368742649298,
      "loss": 2.1705,
      "step": 49377
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.6985883712768555,
      "learning_rate": 0.0005345343236044142,
      "loss": 2.171,
      "step": 49378
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7062036991119385,
      "learning_rate": 0.0005345317729002943,
      "loss": 1.9871,
      "step": 49379
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.701360285282135,
      "learning_rate": 0.0005345292221525704,
      "loss": 2.3615,
      "step": 49380
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7979510426521301,
      "learning_rate": 0.0005345266713612432,
      "loss": 2.2474,
      "step": 49381
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7294542193412781,
      "learning_rate": 0.000534524120526313,
      "loss": 2.2128,
      "step": 49382
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.644691526889801,
      "learning_rate": 0.0005345215696477803,
      "loss": 2.301,
      "step": 49383
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7167947292327881,
      "learning_rate": 0.0005345190187256455,
      "loss": 2.0854,
      "step": 49384
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.6741663217544556,
      "learning_rate": 0.0005345164677599093,
      "loss": 2.4455,
      "step": 49385
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.6678006649017334,
      "learning_rate": 0.0005345139167505722,
      "loss": 2.0491,
      "step": 49386
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.6987239122390747,
      "learning_rate": 0.0005345113656976344,
      "loss": 2.3478,
      "step": 49387
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.6818952560424805,
      "learning_rate": 0.0005345088146010965,
      "loss": 2.1528,
      "step": 49388
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7190518379211426,
      "learning_rate": 0.000534506263460959,
      "loss": 2.1385,
      "step": 49389
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.6948432922363281,
      "learning_rate": 0.0005345037122772224,
      "loss": 2.1585,
      "step": 49390
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.6910456418991089,
      "learning_rate": 0.000534501161049887,
      "loss": 2.2579,
      "step": 49391
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.8427302241325378,
      "learning_rate": 0.0005344986097789536,
      "loss": 2.2029,
      "step": 49392
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7009679675102234,
      "learning_rate": 0.0005344960584644224,
      "loss": 2.1076,
      "step": 49393
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.720446765422821,
      "learning_rate": 0.000534493507106294,
      "loss": 2.2107,
      "step": 49394
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7417675852775574,
      "learning_rate": 0.0005344909557045687,
      "loss": 2.381,
      "step": 49395
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7007192969322205,
      "learning_rate": 0.0005344884042592471,
      "loss": 2.2299,
      "step": 49396
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.005722165107727,
      "learning_rate": 0.0005344858527703298,
      "loss": 2.3775,
      "step": 49397
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7552391886711121,
      "learning_rate": 0.0005344833012378172,
      "loss": 2.3599,
      "step": 49398
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.8522736430168152,
      "learning_rate": 0.0005344807496617096,
      "loss": 2.3131,
      "step": 49399
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7059934735298157,
      "learning_rate": 0.0005344781980420077,
      "loss": 2.2675,
      "step": 49400
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.8160867094993591,
      "learning_rate": 0.0005344756463787118,
      "loss": 2.3912,
      "step": 49401
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7689979076385498,
      "learning_rate": 0.0005344730946718224,
      "loss": 2.2129,
      "step": 49402
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7134351134300232,
      "learning_rate": 0.0005344705429213402,
      "loss": 2.2465,
      "step": 49403
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.6650478839874268,
      "learning_rate": 0.0005344679911272654,
      "loss": 2.2243,
      "step": 49404
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7874841094017029,
      "learning_rate": 0.0005344654392895985,
      "loss": 1.9867,
      "step": 49405
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7154477834701538,
      "learning_rate": 0.0005344628874083402,
      "loss": 2.1205,
      "step": 49406
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7386422753334045,
      "learning_rate": 0.0005344603354834908,
      "loss": 2.3162,
      "step": 49407
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.8291308283805847,
      "learning_rate": 0.0005344577835150507,
      "loss": 2.144,
      "step": 49408
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7185319066047668,
      "learning_rate": 0.0005344552315030206,
      "loss": 2.082,
      "step": 49409
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.698160707950592,
      "learning_rate": 0.0005344526794474006,
      "loss": 2.2736,
      "step": 49410
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.6976423263549805,
      "learning_rate": 0.0005344501273481917,
      "loss": 2.2321,
      "step": 49411
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7261244058609009,
      "learning_rate": 0.000534447575205394,
      "loss": 2.3554,
      "step": 49412
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.8111730813980103,
      "learning_rate": 0.0005344450230190081,
      "loss": 2.3486,
      "step": 49413
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7540765404701233,
      "learning_rate": 0.0005344424707890343,
      "loss": 2.1259,
      "step": 49414
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7385826706886292,
      "learning_rate": 0.0005344399185154734,
      "loss": 2.1506,
      "step": 49415
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.8360835909843445,
      "learning_rate": 0.0005344373661983255,
      "loss": 2.0385,
      "step": 49416
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7419534921646118,
      "learning_rate": 0.0005344348138375914,
      "loss": 2.253,
      "step": 49417
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.6889808773994446,
      "learning_rate": 0.0005344322614332715,
      "loss": 2.0518,
      "step": 49418
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.6591850519180298,
      "learning_rate": 0.0005344297089853661,
      "loss": 2.2465,
      "step": 49419
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7195168137550354,
      "learning_rate": 0.0005344271564938758,
      "loss": 2.1583,
      "step": 49420
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7752081751823425,
      "learning_rate": 0.0005344246039588011,
      "loss": 2.3232,
      "step": 49421
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.6500101685523987,
      "learning_rate": 0.0005344220513801424,
      "loss": 2.1205,
      "step": 49422
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.8003966808319092,
      "learning_rate": 0.0005344194987579003,
      "loss": 2.27,
      "step": 49423
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.8421294093132019,
      "learning_rate": 0.0005344169460920751,
      "loss": 2.2345,
      "step": 49424
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.9870801568031311,
      "learning_rate": 0.0005344143933826675,
      "loss": 2.388,
      "step": 49425
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7835268974304199,
      "learning_rate": 0.0005344118406296777,
      "loss": 2.2335,
      "step": 49426
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7245764136314392,
      "learning_rate": 0.0005344092878331064,
      "loss": 2.4281,
      "step": 49427
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.8508774042129517,
      "learning_rate": 0.0005344067349929539,
      "loss": 2.4711,
      "step": 49428
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.9231813549995422,
      "learning_rate": 0.0005344041821092209,
      "loss": 2.2909,
      "step": 49429
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.6860232949256897,
      "learning_rate": 0.0005344016291819076,
      "loss": 2.5199,
      "step": 49430
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.0259490013122559,
      "learning_rate": 0.0005343990762110147,
      "loss": 2.2853,
      "step": 49431
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.860954225063324,
      "learning_rate": 0.0005343965231965425,
      "loss": 2.2406,
      "step": 49432
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7346188426017761,
      "learning_rate": 0.0005343939701384917,
      "loss": 2.1341,
      "step": 49433
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.8107066750526428,
      "learning_rate": 0.0005343914170368625,
      "loss": 2.3134,
      "step": 49434
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7879626750946045,
      "learning_rate": 0.0005343888638916556,
      "loss": 2.2297,
      "step": 49435
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7616837024688721,
      "learning_rate": 0.0005343863107028713,
      "loss": 2.2892,
      "step": 49436
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7231876850128174,
      "learning_rate": 0.0005343837574705102,
      "loss": 2.3012,
      "step": 49437
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7715752720832825,
      "learning_rate": 0.0005343812041945728,
      "loss": 2.3543,
      "step": 49438
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7268460392951965,
      "learning_rate": 0.0005343786508750594,
      "loss": 2.0214,
      "step": 49439
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.727199375629425,
      "learning_rate": 0.0005343760975119706,
      "loss": 2.4278,
      "step": 49440
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7749480605125427,
      "learning_rate": 0.000534373544105307,
      "loss": 2.4769,
      "step": 49441
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7035388946533203,
      "learning_rate": 0.0005343709906550689,
      "loss": 2.3433,
      "step": 49442
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.6815958619117737,
      "learning_rate": 0.0005343684371612566,
      "loss": 2.3506,
      "step": 49443
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.6666668653488159,
      "learning_rate": 0.000534365883623871,
      "loss": 2.2644,
      "step": 49444
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.6999740600585938,
      "learning_rate": 0.0005343633300429122,
      "loss": 2.0073,
      "step": 49445
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7983253002166748,
      "learning_rate": 0.000534360776418381,
      "loss": 2.1865,
      "step": 49446
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.6716747879981995,
      "learning_rate": 0.0005343582227502776,
      "loss": 2.2934,
      "step": 49447
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7262308597564697,
      "learning_rate": 0.0005343556690386027,
      "loss": 2.3114,
      "step": 49448
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.8492746353149414,
      "learning_rate": 0.0005343531152833564,
      "loss": 2.1876,
      "step": 49449
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7566527724266052,
      "learning_rate": 0.0005343505614845397,
      "loss": 2.4349,
      "step": 49450
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.6989018321037292,
      "learning_rate": 0.0005343480076421527,
      "loss": 2.2097,
      "step": 49451
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7843171954154968,
      "learning_rate": 0.0005343454537561959,
      "loss": 2.3341,
      "step": 49452
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.9067568182945251,
      "learning_rate": 0.00053434289982667,
      "loss": 2.0984,
      "step": 49453
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.728631317615509,
      "learning_rate": 0.0005343403458535753,
      "loss": 2.4595,
      "step": 49454
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.6892359852790833,
      "learning_rate": 0.0005343377918369122,
      "loss": 2.4074,
      "step": 49455
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7242123484611511,
      "learning_rate": 0.0005343352377766814,
      "loss": 2.2081,
      "step": 49456
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7732129693031311,
      "learning_rate": 0.0005343326836728832,
      "loss": 2.1844,
      "step": 49457
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.876798689365387,
      "learning_rate": 0.0005343301295255181,
      "loss": 2.2275,
      "step": 49458
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7456178069114685,
      "learning_rate": 0.0005343275753345866,
      "loss": 2.2053,
      "step": 49459
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.8165233135223389,
      "learning_rate": 0.0005343250211000893,
      "loss": 2.3612,
      "step": 49460
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.0292121171951294,
      "learning_rate": 0.0005343224668220265,
      "loss": 2.5505,
      "step": 49461
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.732401430606842,
      "learning_rate": 0.0005343199125003987,
      "loss": 2.1981,
      "step": 49462
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.8339659571647644,
      "learning_rate": 0.0005343173581352064,
      "loss": 2.5608,
      "step": 49463
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.0290579795837402,
      "learning_rate": 0.0005343148037264501,
      "loss": 2.0421,
      "step": 49464
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.686152994632721,
      "learning_rate": 0.0005343122492741303,
      "loss": 2.3021,
      "step": 49465
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7500369548797607,
      "learning_rate": 0.0005343096947782474,
      "loss": 2.4121,
      "step": 49466
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.9886155128479004,
      "learning_rate": 0.0005343071402388018,
      "loss": 2.2258,
      "step": 49467
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.734870433807373,
      "learning_rate": 0.0005343045856557942,
      "loss": 2.2217,
      "step": 49468
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.8253277540206909,
      "learning_rate": 0.000534302031029225,
      "loss": 2.118,
      "step": 49469
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7557274103164673,
      "learning_rate": 0.0005342994763590945,
      "loss": 2.2676,
      "step": 49470
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.8580445051193237,
      "learning_rate": 0.0005342969216454034,
      "loss": 2.3135,
      "step": 49471
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7592108845710754,
      "learning_rate": 0.0005342943668881521,
      "loss": 2.1978,
      "step": 49472
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7377592921257019,
      "learning_rate": 0.000534291812087341,
      "loss": 2.4275,
      "step": 49473
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.6996518969535828,
      "learning_rate": 0.0005342892572429706,
      "loss": 2.1513,
      "step": 49474
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7395386099815369,
      "learning_rate": 0.0005342867023550413,
      "loss": 2.4005,
      "step": 49475
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.6535159349441528,
      "learning_rate": 0.0005342841474235539,
      "loss": 2.2494,
      "step": 49476
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7365931868553162,
      "learning_rate": 0.0005342815924485085,
      "loss": 2.2238,
      "step": 49477
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7018612027168274,
      "learning_rate": 0.0005342790374299059,
      "loss": 2.509,
      "step": 49478
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.8325151205062866,
      "learning_rate": 0.0005342764823677462,
      "loss": 2.4033,
      "step": 49479
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.6975604891777039,
      "learning_rate": 0.0005342739272620302,
      "loss": 2.5898,
      "step": 49480
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7051188945770264,
      "learning_rate": 0.0005342713721127583,
      "loss": 2.3921,
      "step": 49481
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.6912607550621033,
      "learning_rate": 0.0005342688169199309,
      "loss": 2.2113,
      "step": 49482
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.6844707131385803,
      "learning_rate": 0.0005342662616835483,
      "loss": 2.029,
      "step": 49483
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7089532613754272,
      "learning_rate": 0.0005342637064036114,
      "loss": 2.2861,
      "step": 49484
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.8624730110168457,
      "learning_rate": 0.0005342611510801204,
      "loss": 2.385,
      "step": 49485
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.8308218717575073,
      "learning_rate": 0.0005342585957130759,
      "loss": 2.1178,
      "step": 49486
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7135833501815796,
      "learning_rate": 0.0005342560403024781,
      "loss": 2.1235,
      "step": 49487
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7125445604324341,
      "learning_rate": 0.000534253484848328,
      "loss": 1.9589,
      "step": 49488
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7818940877914429,
      "learning_rate": 0.0005342509293506254,
      "loss": 2.4684,
      "step": 49489
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.9216566681861877,
      "learning_rate": 0.0005342483738093713,
      "loss": 2.0601,
      "step": 49490
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7158178687095642,
      "learning_rate": 0.0005342458182245662,
      "loss": 2.4507,
      "step": 49491
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.6992402076721191,
      "learning_rate": 0.0005342432625962101,
      "loss": 2.4665,
      "step": 49492
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7683115005493164,
      "learning_rate": 0.0005342407069243039,
      "loss": 2.1834,
      "step": 49493
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.0019783973693848,
      "learning_rate": 0.0005342381512088479,
      "loss": 2.5723,
      "step": 49494
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7564752101898193,
      "learning_rate": 0.0005342355954498425,
      "loss": 2.0787,
      "step": 49495
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.686825156211853,
      "learning_rate": 0.0005342330396472883,
      "loss": 2.3459,
      "step": 49496
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.2485917806625366,
      "learning_rate": 0.0005342304838011859,
      "loss": 2.173,
      "step": 49497
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7583873867988586,
      "learning_rate": 0.0005342279279115356,
      "loss": 2.2063,
      "step": 49498
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7238235473632812,
      "learning_rate": 0.0005342253719783377,
      "loss": 2.5207,
      "step": 49499
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7798227071762085,
      "learning_rate": 0.0005342228160015931,
      "loss": 2.346,
      "step": 49500
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.8666817545890808,
      "learning_rate": 0.000534220259981302,
      "loss": 2.3923,
      "step": 49501
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7557163238525391,
      "learning_rate": 0.0005342177039174649,
      "loss": 2.2941,
      "step": 49502
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.700161874294281,
      "learning_rate": 0.0005342151478100823,
      "loss": 2.2696,
      "step": 49503
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.8569424152374268,
      "learning_rate": 0.0005342125916591548,
      "loss": 2.2419,
      "step": 49504
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.8533119559288025,
      "learning_rate": 0.0005342100354646826,
      "loss": 2.3702,
      "step": 49505
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7101240158081055,
      "learning_rate": 0.0005342074792266665,
      "loss": 2.1875,
      "step": 49506
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7459275722503662,
      "learning_rate": 0.0005342049229451067,
      "loss": 2.1739,
      "step": 49507
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.8991293907165527,
      "learning_rate": 0.0005342023666200039,
      "loss": 2.3182,
      "step": 49508
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7797262072563171,
      "learning_rate": 0.0005341998102513583,
      "loss": 2.2487,
      "step": 49509
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.6771803498268127,
      "learning_rate": 0.0005341972538391705,
      "loss": 2.3772,
      "step": 49510
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7195369601249695,
      "learning_rate": 0.0005341946973834411,
      "loss": 2.0779,
      "step": 49511
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.777832567691803,
      "learning_rate": 0.0005341921408841705,
      "loss": 2.2411,
      "step": 49512
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7128555178642273,
      "learning_rate": 0.0005341895843413593,
      "loss": 2.3922,
      "step": 49513
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.6853517293930054,
      "learning_rate": 0.0005341870277550076,
      "loss": 2.1606,
      "step": 49514
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7297350764274597,
      "learning_rate": 0.0005341844711251162,
      "loss": 2.3237,
      "step": 49515
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.888796865940094,
      "learning_rate": 0.0005341819144516855,
      "loss": 2.4558,
      "step": 49516
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7058253288269043,
      "learning_rate": 0.000534179357734716,
      "loss": 2.1566,
      "step": 49517
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.6909129619598389,
      "learning_rate": 0.000534176800974208,
      "loss": 2.4242,
      "step": 49518
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.8429153561592102,
      "learning_rate": 0.0005341742441701622,
      "loss": 2.383,
      "step": 49519
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7620819211006165,
      "learning_rate": 0.0005341716873225789,
      "loss": 2.4369,
      "step": 49520
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.6513209342956543,
      "learning_rate": 0.0005341691304314587,
      "loss": 2.2765,
      "step": 49521
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.8212264180183411,
      "learning_rate": 0.0005341665734968022,
      "loss": 2.06,
      "step": 49522
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7589091658592224,
      "learning_rate": 0.0005341640165186096,
      "loss": 2.2706,
      "step": 49523
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.6929680705070496,
      "learning_rate": 0.0005341614594968815,
      "loss": 2.3308,
      "step": 49524
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7116661667823792,
      "learning_rate": 0.0005341589024316183,
      "loss": 2.2735,
      "step": 49525
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7084628343582153,
      "learning_rate": 0.0005341563453228206,
      "loss": 1.9955,
      "step": 49526
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.6680008172988892,
      "learning_rate": 0.0005341537881704888,
      "loss": 2.3127,
      "step": 49527
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.6702589392662048,
      "learning_rate": 0.0005341512309746234,
      "loss": 2.4781,
      "step": 49528
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7146349549293518,
      "learning_rate": 0.0005341486737352249,
      "loss": 2.2093,
      "step": 49529
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7430357336997986,
      "learning_rate": 0.0005341461164522938,
      "loss": 2.2928,
      "step": 49530
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7027049660682678,
      "learning_rate": 0.0005341435591258303,
      "loss": 2.2616,
      "step": 49531
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.6639545559883118,
      "learning_rate": 0.0005341410017558352,
      "loss": 2.4533,
      "step": 49532
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.6874881386756897,
      "learning_rate": 0.0005341384443423089,
      "loss": 2.3638,
      "step": 49533
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.6195158362388611,
      "learning_rate": 0.0005341358868852519,
      "loss": 2.0379,
      "step": 49534
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7846958637237549,
      "learning_rate": 0.0005341333293846644,
      "loss": 2.4247,
      "step": 49535
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.749573826789856,
      "learning_rate": 0.0005341307718405473,
      "loss": 2.3806,
      "step": 49536
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7562492489814758,
      "learning_rate": 0.0005341282142529008,
      "loss": 1.909,
      "step": 49537
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.6745536923408508,
      "learning_rate": 0.0005341256566217255,
      "loss": 2.3893,
      "step": 49538
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7334070801734924,
      "learning_rate": 0.0005341230989470219,
      "loss": 2.2415,
      "step": 49539
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7514625191688538,
      "learning_rate": 0.0005341205412287901,
      "loss": 2.1511,
      "step": 49540
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.715358555316925,
      "learning_rate": 0.0005341179834670311,
      "loss": 2.0516,
      "step": 49541
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.667453944683075,
      "learning_rate": 0.000534115425661745,
      "loss": 2.2339,
      "step": 49542
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.8279938697814941,
      "learning_rate": 0.0005341128678129326,
      "loss": 2.305,
      "step": 49543
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7199768424034119,
      "learning_rate": 0.000534110309920594,
      "loss": 2.1823,
      "step": 49544
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.9631559252738953,
      "learning_rate": 0.00053410775198473,
      "loss": 2.3812,
      "step": 49545
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7655327320098877,
      "learning_rate": 0.0005341051940053411,
      "loss": 2.4052,
      "step": 49546
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7197897434234619,
      "learning_rate": 0.0005341026359824274,
      "loss": 2.3016,
      "step": 49547
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7455663084983826,
      "learning_rate": 0.0005341000779159897,
      "loss": 2.3128,
      "step": 49548
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.6805771589279175,
      "learning_rate": 0.0005340975198060284,
      "loss": 2.3672,
      "step": 49549
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7634299993515015,
      "learning_rate": 0.0005340949616525438,
      "loss": 2.1382,
      "step": 49550
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7201173305511475,
      "learning_rate": 0.0005340924034555366,
      "loss": 2.2775,
      "step": 49551
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7120645642280579,
      "learning_rate": 0.0005340898452150072,
      "loss": 2.6085,
      "step": 49552
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7553756833076477,
      "learning_rate": 0.0005340872869309561,
      "loss": 2.307,
      "step": 49553
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7179921269416809,
      "learning_rate": 0.0005340847286033838,
      "loss": 2.273,
      "step": 49554
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7639921307563782,
      "learning_rate": 0.0005340821702322907,
      "loss": 2.3091,
      "step": 49555
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.6358535885810852,
      "learning_rate": 0.0005340796118176773,
      "loss": 2.3167,
      "step": 49556
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7439963221549988,
      "learning_rate": 0.0005340770533595441,
      "loss": 2.2916,
      "step": 49557
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.6776557564735413,
      "learning_rate": 0.0005340744948578915,
      "loss": 2.404,
      "step": 49558
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.6749829053878784,
      "learning_rate": 0.0005340719363127201,
      "loss": 2.2473,
      "step": 49559
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.8259731531143188,
      "learning_rate": 0.0005340693777240303,
      "loss": 2.4258,
      "step": 49560
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.6814109683036804,
      "learning_rate": 0.0005340668190918225,
      "loss": 2.0991,
      "step": 49561
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7544687390327454,
      "learning_rate": 0.0005340642604160972,
      "loss": 2.3047,
      "step": 49562
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7468178868293762,
      "learning_rate": 0.0005340617016968552,
      "loss": 2.1933,
      "step": 49563
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.8525932431221008,
      "learning_rate": 0.0005340591429340966,
      "loss": 2.1719,
      "step": 49564
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7199652194976807,
      "learning_rate": 0.0005340565841278218,
      "loss": 1.9978,
      "step": 49565
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7431723475456238,
      "learning_rate": 0.0005340540252780317,
      "loss": 2.2621,
      "step": 49566
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.6986373662948608,
      "learning_rate": 0.0005340514663847264,
      "loss": 2.1098,
      "step": 49567
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.8181659579277039,
      "learning_rate": 0.0005340489074479066,
      "loss": 2.2386,
      "step": 49568
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.6776007413864136,
      "learning_rate": 0.0005340463484675728,
      "loss": 1.9441,
      "step": 49569
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7221707701683044,
      "learning_rate": 0.0005340437894437251,
      "loss": 2.4015,
      "step": 49570
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7214564681053162,
      "learning_rate": 0.0005340412303763644,
      "loss": 2.4237,
      "step": 49571
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.6882708072662354,
      "learning_rate": 0.0005340386712654909,
      "loss": 2.22,
      "step": 49572
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7118461728096008,
      "learning_rate": 0.0005340361121111053,
      "loss": 2.2968,
      "step": 49573
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7910021543502808,
      "learning_rate": 0.000534033552913208,
      "loss": 2.1579,
      "step": 49574
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7001477479934692,
      "learning_rate": 0.0005340309936717994,
      "loss": 2.2056,
      "step": 49575
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7763957381248474,
      "learning_rate": 0.00053402843438688,
      "loss": 2.0799,
      "step": 49576
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.6749223470687866,
      "learning_rate": 0.0005340258750584503,
      "loss": 2.1349,
      "step": 49577
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.6886096000671387,
      "learning_rate": 0.0005340233156865108,
      "loss": 2.2594,
      "step": 49578
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7737886905670166,
      "learning_rate": 0.0005340207562710619,
      "loss": 2.239,
      "step": 49579
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.694697380065918,
      "learning_rate": 0.0005340181968121042,
      "loss": 1.9552,
      "step": 49580
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7025191783905029,
      "learning_rate": 0.000534015637309638,
      "loss": 2.2634,
      "step": 49581
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7730810046195984,
      "learning_rate": 0.0005340130777636638,
      "loss": 2.1637,
      "step": 49582
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7471359968185425,
      "learning_rate": 0.0005340105181741825,
      "loss": 2.0479,
      "step": 49583
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.6882607936859131,
      "learning_rate": 0.0005340079585411939,
      "loss": 2.2485,
      "step": 49584
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.744904100894928,
      "learning_rate": 0.0005340053988646989,
      "loss": 2.2526,
      "step": 49585
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7230973243713379,
      "learning_rate": 0.000534002839144698,
      "loss": 2.0575,
      "step": 49586
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7133228778839111,
      "learning_rate": 0.0005340002793811913,
      "loss": 2.5225,
      "step": 49587
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.861329972743988,
      "learning_rate": 0.0005339977195741797,
      "loss": 2.4042,
      "step": 49588
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7187262773513794,
      "learning_rate": 0.0005339951597236634,
      "loss": 2.1886,
      "step": 49589
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7315172553062439,
      "learning_rate": 0.0005339925998296432,
      "loss": 2.3366,
      "step": 49590
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7675350904464722,
      "learning_rate": 0.0005339900398921192,
      "loss": 2.1874,
      "step": 49591
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.8725574016571045,
      "learning_rate": 0.000533987479911092,
      "loss": 2.3352,
      "step": 49592
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.6827475428581238,
      "learning_rate": 0.0005339849198865621,
      "loss": 2.1521,
      "step": 49593
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7482829093933105,
      "learning_rate": 0.0005339823598185301,
      "loss": 2.0491,
      "step": 49594
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7377961277961731,
      "learning_rate": 0.0005339797997069963,
      "loss": 2.2644,
      "step": 49595
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.6914811730384827,
      "learning_rate": 0.0005339772395519613,
      "loss": 2.224,
      "step": 49596
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.8399443030357361,
      "learning_rate": 0.0005339746793534255,
      "loss": 2.2494,
      "step": 49597
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7068401575088501,
      "learning_rate": 0.0005339721191113894,
      "loss": 2.4449,
      "step": 49598
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7518014907836914,
      "learning_rate": 0.0005339695588258535,
      "loss": 2.379,
      "step": 49599
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.6653167605400085,
      "learning_rate": 0.0005339669984968181,
      "loss": 2.3935,
      "step": 49600
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.6065623164176941,
      "learning_rate": 0.0005339644381242839,
      "loss": 2.2695,
      "step": 49601
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.6821028590202332,
      "learning_rate": 0.0005339618777082511,
      "loss": 2.3511,
      "step": 49602
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7413526177406311,
      "learning_rate": 0.0005339593172487206,
      "loss": 2.4575,
      "step": 49603
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7124514579772949,
      "learning_rate": 0.0005339567567456926,
      "loss": 2.0884,
      "step": 49604
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7942624688148499,
      "learning_rate": 0.0005339541961991677,
      "loss": 2.2346,
      "step": 49605
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7056112885475159,
      "learning_rate": 0.0005339516356091462,
      "loss": 2.3839,
      "step": 49606
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.8224428296089172,
      "learning_rate": 0.0005339490749756287,
      "loss": 2.2312,
      "step": 49607
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7063701152801514,
      "learning_rate": 0.0005339465142986156,
      "loss": 2.2579,
      "step": 49608
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7219991683959961,
      "learning_rate": 0.0005339439535781076,
      "loss": 2.147,
      "step": 49609
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.6938989758491516,
      "learning_rate": 0.0005339413928141049,
      "loss": 2.426,
      "step": 49610
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.674121618270874,
      "learning_rate": 0.000533938832006608,
      "loss": 2.1852,
      "step": 49611
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7285248041152954,
      "learning_rate": 0.0005339362711556175,
      "loss": 2.1818,
      "step": 49612
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.695112407207489,
      "learning_rate": 0.0005339337102611339,
      "loss": 2.4046,
      "step": 49613
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7030553221702576,
      "learning_rate": 0.0005339311493231574,
      "loss": 2.0668,
      "step": 49614
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7251508831977844,
      "learning_rate": 0.0005339285883416889,
      "loss": 2.072,
      "step": 49615
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.6425052881240845,
      "learning_rate": 0.0005339260273167286,
      "loss": 2.3117,
      "step": 49616
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7666844129562378,
      "learning_rate": 0.000533923466248277,
      "loss": 2.2567,
      "step": 49617
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7237682342529297,
      "learning_rate": 0.0005339209051363347,
      "loss": 2.4283,
      "step": 49618
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7811872363090515,
      "learning_rate": 0.000533918343980902,
      "loss": 2.1826,
      "step": 49619
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7914856672286987,
      "learning_rate": 0.0005339157827819795,
      "loss": 2.3383,
      "step": 49620
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.8510486483573914,
      "learning_rate": 0.0005339132215395676,
      "loss": 2.1186,
      "step": 49621
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7508565187454224,
      "learning_rate": 0.0005339106602536669,
      "loss": 2.1403,
      "step": 49622
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.720856785774231,
      "learning_rate": 0.0005339080989242778,
      "loss": 2.2956,
      "step": 49623
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7020636796951294,
      "learning_rate": 0.0005339055375514006,
      "loss": 2.4365,
      "step": 49624
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7499537467956543,
      "learning_rate": 0.000533902976135036,
      "loss": 2.1791,
      "step": 49625
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.6697725653648376,
      "learning_rate": 0.0005339004146751846,
      "loss": 2.1915,
      "step": 49626
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7226204872131348,
      "learning_rate": 0.0005338978531718465,
      "loss": 2.2036,
      "step": 49627
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.6799259185791016,
      "learning_rate": 0.0005338952916250226,
      "loss": 2.2982,
      "step": 49628
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7106311917304993,
      "learning_rate": 0.0005338927300347129,
      "loss": 2.1867,
      "step": 49629
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7253305912017822,
      "learning_rate": 0.0005338901684009183,
      "loss": 2.3646,
      "step": 49630
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7040061354637146,
      "learning_rate": 0.000533887606723639,
      "loss": 2.1837,
      "step": 49631
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.6489351987838745,
      "learning_rate": 0.0005338850450028756,
      "loss": 2.1813,
      "step": 49632
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.6816534399986267,
      "learning_rate": 0.0005338824832386285,
      "loss": 2.4927,
      "step": 49633
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.6953814029693604,
      "learning_rate": 0.0005338799214308983,
      "loss": 2.2809,
      "step": 49634
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7207192182540894,
      "learning_rate": 0.0005338773595796853,
      "loss": 2.2887,
      "step": 49635
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7480412721633911,
      "learning_rate": 0.0005338747976849904,
      "loss": 2.3578,
      "step": 49636
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.6685507893562317,
      "learning_rate": 0.0005338722357468135,
      "loss": 2.2945,
      "step": 49637
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7467316389083862,
      "learning_rate": 0.0005338696737651554,
      "loss": 2.3925,
      "step": 49638
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.6971195340156555,
      "learning_rate": 0.0005338671117400165,
      "loss": 2.1013,
      "step": 49639
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.8833824992179871,
      "learning_rate": 0.0005338645496713973,
      "loss": 2.2438,
      "step": 49640
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7231757044792175,
      "learning_rate": 0.0005338619875592982,
      "loss": 2.2113,
      "step": 49641
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.6996269226074219,
      "learning_rate": 0.0005338594254037198,
      "loss": 2.2577,
      "step": 49642
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.8113594651222229,
      "learning_rate": 0.0005338568632046625,
      "loss": 2.2506,
      "step": 49643
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7537112236022949,
      "learning_rate": 0.0005338543009621268,
      "loss": 2.3011,
      "step": 49644
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.8361268043518066,
      "learning_rate": 0.0005338517386761132,
      "loss": 2.3898,
      "step": 49645
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.8192920088768005,
      "learning_rate": 0.0005338491763466221,
      "loss": 2.0717,
      "step": 49646
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7148245573043823,
      "learning_rate": 0.0005338466139736541,
      "loss": 2.1352,
      "step": 49647
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7268023490905762,
      "learning_rate": 0.0005338440515572095,
      "loss": 2.2921,
      "step": 49648
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.6801747679710388,
      "learning_rate": 0.0005338414890972889,
      "loss": 2.37,
      "step": 49649
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7484720945358276,
      "learning_rate": 0.0005338389265938929,
      "loss": 2.3639,
      "step": 49650
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7397871613502502,
      "learning_rate": 0.0005338363640470216,
      "loss": 2.4279,
      "step": 49651
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7329080700874329,
      "learning_rate": 0.0005338338014566758,
      "loss": 2.1691,
      "step": 49652
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.8331242799758911,
      "learning_rate": 0.0005338312388228559,
      "loss": 2.1688,
      "step": 49653
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7691935300827026,
      "learning_rate": 0.0005338286761455625,
      "loss": 2.2164,
      "step": 49654
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7222319841384888,
      "learning_rate": 0.0005338261134247956,
      "loss": 2.2373,
      "step": 49655
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7606446146965027,
      "learning_rate": 0.0005338235506605563,
      "loss": 2.1894,
      "step": 49656
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7972294688224792,
      "learning_rate": 0.0005338209878528447,
      "loss": 1.9894,
      "step": 49657
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.8867188692092896,
      "learning_rate": 0.0005338184250016614,
      "loss": 2.3258,
      "step": 49658
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7352639436721802,
      "learning_rate": 0.0005338158621070068,
      "loss": 2.5081,
      "step": 49659
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.9927863478660583,
      "learning_rate": 0.0005338132991688814,
      "loss": 2.3839,
      "step": 49660
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.0244990587234497,
      "learning_rate": 0.0005338107361872856,
      "loss": 1.9446,
      "step": 49661
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7284566164016724,
      "learning_rate": 0.0005338081731622201,
      "loss": 2.3872,
      "step": 49662
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.8967762589454651,
      "learning_rate": 0.0005338056100936853,
      "loss": 2.2143,
      "step": 49663
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.8102685809135437,
      "learning_rate": 0.0005338030469816816,
      "loss": 2.2057,
      "step": 49664
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7598279118537903,
      "learning_rate": 0.0005338004838262094,
      "loss": 2.3542,
      "step": 49665
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.6806544661521912,
      "learning_rate": 0.0005337979206272693,
      "loss": 2.1874,
      "step": 49666
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.9510300159454346,
      "learning_rate": 0.0005337953573848618,
      "loss": 2.2039,
      "step": 49667
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.6972973346710205,
      "learning_rate": 0.0005337927940989873,
      "loss": 2.1353,
      "step": 49668
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7447641491889954,
      "learning_rate": 0.0005337902307696464,
      "loss": 2.1331,
      "step": 49669
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.73734050989151,
      "learning_rate": 0.0005337876673968394,
      "loss": 2.1266,
      "step": 49670
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7867782115936279,
      "learning_rate": 0.0005337851039805667,
      "loss": 2.3247,
      "step": 49671
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.820081889629364,
      "learning_rate": 0.0005337825405208292,
      "loss": 2.2151,
      "step": 49672
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7104217410087585,
      "learning_rate": 0.000533779977017627,
      "loss": 2.1393,
      "step": 49673
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.9894765019416809,
      "learning_rate": 0.0005337774134709607,
      "loss": 2.2495,
      "step": 49674
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.6713946461677551,
      "learning_rate": 0.0005337748498808307,
      "loss": 1.9109,
      "step": 49675
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.6959911584854126,
      "learning_rate": 0.0005337722862472377,
      "loss": 2.2019,
      "step": 49676
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.8817265629768372,
      "learning_rate": 0.0005337697225701819,
      "loss": 2.4014,
      "step": 49677
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7192332744598389,
      "learning_rate": 0.0005337671588496639,
      "loss": 2.2087,
      "step": 49678
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.6719632744789124,
      "learning_rate": 0.0005337645950856841,
      "loss": 2.3175,
      "step": 49679
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7769504189491272,
      "learning_rate": 0.0005337620312782433,
      "loss": 2.1964,
      "step": 49680
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7821126580238342,
      "learning_rate": 0.0005337594674273415,
      "loss": 2.1959,
      "step": 49681
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7236354947090149,
      "learning_rate": 0.0005337569035329795,
      "loss": 2.1987,
      "step": 49682
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.734402596950531,
      "learning_rate": 0.0005337543395951577,
      "loss": 2.4012,
      "step": 49683
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.122086524963379,
      "learning_rate": 0.0005337517756138765,
      "loss": 2.2911,
      "step": 49684
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7140650749206543,
      "learning_rate": 0.0005337492115891363,
      "loss": 2.3648,
      "step": 49685
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.6920543909072876,
      "learning_rate": 0.0005337466475209379,
      "loss": 2.3612,
      "step": 49686
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.8026403188705444,
      "learning_rate": 0.0005337440834092816,
      "loss": 2.0622,
      "step": 49687
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.0523220300674438,
      "learning_rate": 0.0005337415192541678,
      "loss": 2.3652,
      "step": 49688
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.6941566467285156,
      "learning_rate": 0.000533738955055597,
      "loss": 2.1828,
      "step": 49689
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7334073185920715,
      "learning_rate": 0.0005337363908135697,
      "loss": 2.1831,
      "step": 49690
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7168030142784119,
      "learning_rate": 0.0005337338265280865,
      "loss": 2.3501,
      "step": 49691
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.6806365251541138,
      "learning_rate": 0.0005337312621991478,
      "loss": 2.3257,
      "step": 49692
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7577709555625916,
      "learning_rate": 0.0005337286978267539,
      "loss": 2.2835,
      "step": 49693
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7373008131980896,
      "learning_rate": 0.0005337261334109055,
      "loss": 2.3275,
      "step": 49694
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7634150981903076,
      "learning_rate": 0.0005337235689516029,
      "loss": 2.1076,
      "step": 49695
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.792627215385437,
      "learning_rate": 0.0005337210044488468,
      "loss": 2.3137,
      "step": 49696
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7378635406494141,
      "learning_rate": 0.0005337184399026375,
      "loss": 2.2862,
      "step": 49697
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7599552869796753,
      "learning_rate": 0.0005337158753129756,
      "loss": 2.2525,
      "step": 49698
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7487256526947021,
      "learning_rate": 0.0005337133106798614,
      "loss": 2.2041,
      "step": 49699
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.6890132427215576,
      "learning_rate": 0.0005337107460032956,
      "loss": 2.4334,
      "step": 49700
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7280372977256775,
      "learning_rate": 0.0005337081812832784,
      "loss": 2.4026,
      "step": 49701
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7600580453872681,
      "learning_rate": 0.0005337056165198106,
      "loss": 2.245,
      "step": 49702
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.8573101758956909,
      "learning_rate": 0.0005337030517128925,
      "loss": 2.3829,
      "step": 49703
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.6662502884864807,
      "learning_rate": 0.0005337004868625244,
      "loss": 2.3928,
      "step": 49704
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7096352577209473,
      "learning_rate": 0.0005336979219687072,
      "loss": 2.364,
      "step": 49705
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7102556228637695,
      "learning_rate": 0.000533695357031441,
      "loss": 2.0848,
      "step": 49706
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.6198424696922302,
      "learning_rate": 0.0005336927920507264,
      "loss": 2.1511,
      "step": 49707
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7362456917762756,
      "learning_rate": 0.0005336902270265641,
      "loss": 2.2797,
      "step": 49708
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7216787934303284,
      "learning_rate": 0.0005336876619589542,
      "loss": 2.2236,
      "step": 49709
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7198344469070435,
      "learning_rate": 0.0005336850968478974,
      "loss": 2.3205,
      "step": 49710
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7421786189079285,
      "learning_rate": 0.0005336825316933941,
      "loss": 2.3048,
      "step": 49711
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.6968964338302612,
      "learning_rate": 0.0005336799664954449,
      "loss": 2.2183,
      "step": 49712
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7272114753723145,
      "learning_rate": 0.00053367740125405,
      "loss": 2.3624,
      "step": 49713
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.6859056353569031,
      "learning_rate": 0.0005336748359692102,
      "loss": 2.4816,
      "step": 49714
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7209952473640442,
      "learning_rate": 0.0005336722706409259,
      "loss": 2.2625,
      "step": 49715
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7258051633834839,
      "learning_rate": 0.0005336697052691974,
      "loss": 2.359,
      "step": 49716
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7585605978965759,
      "learning_rate": 0.0005336671398540253,
      "loss": 2.2934,
      "step": 49717
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7339417338371277,
      "learning_rate": 0.0005336645743954101,
      "loss": 2.2241,
      "step": 49718
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7078876495361328,
      "learning_rate": 0.0005336620088933523,
      "loss": 2.3619,
      "step": 49719
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7724037766456604,
      "learning_rate": 0.0005336594433478523,
      "loss": 2.2797,
      "step": 49720
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7248582243919373,
      "learning_rate": 0.0005336568777589105,
      "loss": 2.3693,
      "step": 49721
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7798548936843872,
      "learning_rate": 0.0005336543121265275,
      "loss": 2.2956,
      "step": 49722
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7697426080703735,
      "learning_rate": 0.0005336517464507038,
      "loss": 2.4289,
      "step": 49723
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7682325839996338,
      "learning_rate": 0.0005336491807314398,
      "loss": 2.0058,
      "step": 49724
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7418205142021179,
      "learning_rate": 0.000533646614968736,
      "loss": 2.1751,
      "step": 49725
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.738137423992157,
      "learning_rate": 0.0005336440491625929,
      "loss": 2.2648,
      "step": 49726
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.733149528503418,
      "learning_rate": 0.000533641483313011,
      "loss": 2.3281,
      "step": 49727
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.732468843460083,
      "learning_rate": 0.0005336389174199907,
      "loss": 2.0579,
      "step": 49728
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.6788584589958191,
      "learning_rate": 0.0005336363514835325,
      "loss": 2.2248,
      "step": 49729
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7051464915275574,
      "learning_rate": 0.0005336337855036369,
      "loss": 2.3583,
      "step": 49730
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.6811525821685791,
      "learning_rate": 0.0005336312194803044,
      "loss": 2.2809,
      "step": 49731
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.8898867964744568,
      "learning_rate": 0.0005336286534135354,
      "loss": 2.3338,
      "step": 49732
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7221518158912659,
      "learning_rate": 0.0005336260873033304,
      "loss": 2.4372,
      "step": 49733
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.6780140995979309,
      "learning_rate": 0.00053362352114969,
      "loss": 2.1542,
      "step": 49734
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7726898193359375,
      "learning_rate": 0.0005336209549526145,
      "loss": 2.3711,
      "step": 49735
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7159541845321655,
      "learning_rate": 0.0005336183887121045,
      "loss": 2.2117,
      "step": 49736
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7734550833702087,
      "learning_rate": 0.0005336158224281604,
      "loss": 2.1592,
      "step": 49737
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7611820697784424,
      "learning_rate": 0.0005336132561007828,
      "loss": 2.2248,
      "step": 49738
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7151405811309814,
      "learning_rate": 0.0005336106897299719,
      "loss": 2.0542,
      "step": 49739
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7036195993423462,
      "learning_rate": 0.0005336081233157286,
      "loss": 2.267,
      "step": 49740
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7332944869995117,
      "learning_rate": 0.000533605556858053,
      "loss": 2.3849,
      "step": 49741
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.8116405606269836,
      "learning_rate": 0.0005336029903569456,
      "loss": 2.3558,
      "step": 49742
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7623724341392517,
      "learning_rate": 0.0005336004238124072,
      "loss": 2.3306,
      "step": 49743
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.8048142790794373,
      "learning_rate": 0.000533597857224438,
      "loss": 2.0202,
      "step": 49744
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7091258764266968,
      "learning_rate": 0.0005335952905930384,
      "loss": 2.3395,
      "step": 49745
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.6683900952339172,
      "learning_rate": 0.0005335927239182093,
      "loss": 2.006,
      "step": 49746
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7097742557525635,
      "learning_rate": 0.0005335901571999508,
      "loss": 2.2202,
      "step": 49747
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7364819645881653,
      "learning_rate": 0.0005335875904382634,
      "loss": 2.2414,
      "step": 49748
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7487144470214844,
      "learning_rate": 0.0005335850236331477,
      "loss": 2.0724,
      "step": 49749
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.6942132711410522,
      "learning_rate": 0.0005335824567846042,
      "loss": 2.4042,
      "step": 49750
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7486448884010315,
      "learning_rate": 0.0005335798898926333,
      "loss": 2.1992,
      "step": 49751
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.8786065578460693,
      "learning_rate": 0.0005335773229572353,
      "loss": 2.2765,
      "step": 49752
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7216091156005859,
      "learning_rate": 0.0005335747559784111,
      "loss": 2.4357,
      "step": 49753
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.8032443523406982,
      "learning_rate": 0.0005335721889561608,
      "loss": 2.1144,
      "step": 49754
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7884312272071838,
      "learning_rate": 0.0005335696218904851,
      "loss": 2.3962,
      "step": 49755
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7265156507492065,
      "learning_rate": 0.0005335670547813844,
      "loss": 2.1591,
      "step": 49756
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.6812398433685303,
      "learning_rate": 0.0005335644876288591,
      "loss": 2.2732,
      "step": 49757
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.9561648368835449,
      "learning_rate": 0.0005335619204329098,
      "loss": 2.1587,
      "step": 49758
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7768735885620117,
      "learning_rate": 0.000533559353193537,
      "loss": 2.1355,
      "step": 49759
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.6972375512123108,
      "learning_rate": 0.000533556785910741,
      "loss": 2.2488,
      "step": 49760
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.8350309729576111,
      "learning_rate": 0.0005335542185845224,
      "loss": 2.2031,
      "step": 49761
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.6854710578918457,
      "learning_rate": 0.0005335516512148817,
      "loss": 2.1776,
      "step": 49762
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7175114750862122,
      "learning_rate": 0.0005335490838018193,
      "loss": 2.2598,
      "step": 49763
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7757413983345032,
      "learning_rate": 0.0005335465163453358,
      "loss": 2.291,
      "step": 49764
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.6868366003036499,
      "learning_rate": 0.0005335439488454315,
      "loss": 2.0628,
      "step": 49765
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.785556435585022,
      "learning_rate": 0.0005335413813021071,
      "loss": 2.3092,
      "step": 49766
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7837691903114319,
      "learning_rate": 0.0005335388137153627,
      "loss": 2.2827,
      "step": 49767
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7552623748779297,
      "learning_rate": 0.0005335362460851992,
      "loss": 2.2909,
      "step": 49768
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7940680384635925,
      "learning_rate": 0.0005335336784116169,
      "loss": 2.3783,
      "step": 49769
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.8015218377113342,
      "learning_rate": 0.0005335311106946163,
      "loss": 2.0123,
      "step": 49770
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.6812376976013184,
      "learning_rate": 0.0005335285429341978,
      "loss": 2.2056,
      "step": 49771
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.6902989149093628,
      "learning_rate": 0.0005335259751303621,
      "loss": 2.2108,
      "step": 49772
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.626085638999939,
      "learning_rate": 0.0005335234072831093,
      "loss": 2.1832,
      "step": 49773
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7258945107460022,
      "learning_rate": 0.0005335208393924402,
      "loss": 2.1249,
      "step": 49774
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7377926111221313,
      "learning_rate": 0.0005335182714583551,
      "loss": 2.1923,
      "step": 49775
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.6709905862808228,
      "learning_rate": 0.0005335157034808545,
      "loss": 2.1389,
      "step": 49776
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.821185827255249,
      "learning_rate": 0.0005335131354599391,
      "loss": 2.045,
      "step": 49777
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7191416025161743,
      "learning_rate": 0.0005335105673956091,
      "loss": 2.4324,
      "step": 49778
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7516346573829651,
      "learning_rate": 0.0005335079992878651,
      "loss": 2.4997,
      "step": 49779
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.6697781085968018,
      "learning_rate": 0.0005335054311367075,
      "loss": 2.0151,
      "step": 49780
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7789077162742615,
      "learning_rate": 0.000533502862942137,
      "loss": 2.2309,
      "step": 49781
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.8273226022720337,
      "learning_rate": 0.0005335002947041537,
      "loss": 2.202,
      "step": 49782
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7278499603271484,
      "learning_rate": 0.0005334977264227583,
      "loss": 2.1746,
      "step": 49783
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.677608847618103,
      "learning_rate": 0.0005334951580979513,
      "loss": 2.2118,
      "step": 49784
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7089058756828308,
      "learning_rate": 0.0005334925897297332,
      "loss": 2.1892,
      "step": 49785
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7187670469284058,
      "learning_rate": 0.0005334900213181044,
      "loss": 2.4853,
      "step": 49786
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7184407711029053,
      "learning_rate": 0.0005334874528630654,
      "loss": 2.5332,
      "step": 49787
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.6310276985168457,
      "learning_rate": 0.0005334848843646167,
      "loss": 2.0272,
      "step": 49788
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7773270606994629,
      "learning_rate": 0.0005334823158227587,
      "loss": 2.2149,
      "step": 49789
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.729630708694458,
      "learning_rate": 0.0005334797472374919,
      "loss": 2.1785,
      "step": 49790
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.6722791194915771,
      "learning_rate": 0.0005334771786088168,
      "loss": 2.296,
      "step": 49791
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.6781310439109802,
      "learning_rate": 0.0005334746099367341,
      "loss": 2.2025,
      "step": 49792
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.6510741710662842,
      "learning_rate": 0.0005334720412212439,
      "loss": 2.3651,
      "step": 49793
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7005942463874817,
      "learning_rate": 0.0005334694724623468,
      "loss": 2.3902,
      "step": 49794
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.623012363910675,
      "learning_rate": 0.0005334669036600433,
      "loss": 2.1215,
      "step": 49795
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.6843717694282532,
      "learning_rate": 0.000533464334814334,
      "loss": 2.3561,
      "step": 49796
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.669380247592926,
      "learning_rate": 0.0005334617659252192,
      "loss": 2.3737,
      "step": 49797
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.740652322769165,
      "learning_rate": 0.0005334591969926995,
      "loss": 2.1356,
      "step": 49798
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.6984478235244751,
      "learning_rate": 0.0005334566280167753,
      "loss": 2.1926,
      "step": 49799
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.6823782920837402,
      "learning_rate": 0.0005334540589974472,
      "loss": 2.397,
      "step": 49800
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7513936758041382,
      "learning_rate": 0.0005334514899347155,
      "loss": 2.14,
      "step": 49801
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7843366265296936,
      "learning_rate": 0.0005334489208285808,
      "loss": 2.1969,
      "step": 49802
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.6993259191513062,
      "learning_rate": 0.0005334463516790434,
      "loss": 2.2037,
      "step": 49803
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7363229990005493,
      "learning_rate": 0.000533443782486104,
      "loss": 2.1926,
      "step": 49804
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7629838585853577,
      "learning_rate": 0.000533441213249763,
      "loss": 2.2223,
      "step": 49805
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.6849920749664307,
      "learning_rate": 0.000533438643970021,
      "loss": 2.31,
      "step": 49806
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7143815159797668,
      "learning_rate": 0.0005334360746468782,
      "loss": 2.3296,
      "step": 49807
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7229918837547302,
      "learning_rate": 0.0005334335052803352,
      "loss": 2.5022,
      "step": 49808
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.6831212043762207,
      "learning_rate": 0.0005334309358703926,
      "loss": 2.3635,
      "step": 49809
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.6835378408432007,
      "learning_rate": 0.0005334283664170507,
      "loss": 2.4169,
      "step": 49810
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7312472462654114,
      "learning_rate": 0.0005334257969203101,
      "loss": 2.2292,
      "step": 49811
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7294625043869019,
      "learning_rate": 0.0005334232273801713,
      "loss": 2.3056,
      "step": 49812
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7594484686851501,
      "learning_rate": 0.0005334206577966347,
      "loss": 2.2656,
      "step": 49813
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.6867762804031372,
      "learning_rate": 0.0005334180881697007,
      "loss": 2.1821,
      "step": 49814
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7173762917518616,
      "learning_rate": 0.0005334155184993699,
      "loss": 2.0944,
      "step": 49815
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7174468636512756,
      "learning_rate": 0.0005334129487856429,
      "loss": 2.2369,
      "step": 49816
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7714243531227112,
      "learning_rate": 0.0005334103790285198,
      "loss": 2.3755,
      "step": 49817
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7037114500999451,
      "learning_rate": 0.0005334078092280014,
      "loss": 2.3582,
      "step": 49818
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7529077529907227,
      "learning_rate": 0.0005334052393840881,
      "loss": 2.3486,
      "step": 49819
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7093330025672913,
      "learning_rate": 0.0005334026694967804,
      "loss": 2.1095,
      "step": 49820
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7905895709991455,
      "learning_rate": 0.0005334000995660786,
      "loss": 2.3156,
      "step": 49821
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.6678502559661865,
      "learning_rate": 0.0005333975295919835,
      "loss": 2.0631,
      "step": 49822
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7835649847984314,
      "learning_rate": 0.0005333949595744952,
      "loss": 2.4678,
      "step": 49823
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7219322323799133,
      "learning_rate": 0.0005333923895136146,
      "loss": 2.3683,
      "step": 49824
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7356230616569519,
      "learning_rate": 0.0005333898194093417,
      "loss": 2.2053,
      "step": 49825
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7961029410362244,
      "learning_rate": 0.0005333872492616775,
      "loss": 2.3109,
      "step": 49826
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7460397481918335,
      "learning_rate": 0.000533384679070622,
      "loss": 2.2775,
      "step": 49827
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7118346691131592,
      "learning_rate": 0.000533382108836176,
      "loss": 2.2244,
      "step": 49828
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.6706488132476807,
      "learning_rate": 0.0005333795385583397,
      "loss": 2.1442,
      "step": 49829
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.732473611831665,
      "learning_rate": 0.0005333769682371139,
      "loss": 2.2635,
      "step": 49830
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7463409900665283,
      "learning_rate": 0.0005333743978724987,
      "loss": 2.1687,
      "step": 49831
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7593027949333191,
      "learning_rate": 0.000533371827464495,
      "loss": 2.2642,
      "step": 49832
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7166962623596191,
      "learning_rate": 0.000533369257013103,
      "loss": 2.2893,
      "step": 49833
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7139175534248352,
      "learning_rate": 0.0005333666865183233,
      "loss": 2.414,
      "step": 49834
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7626747488975525,
      "learning_rate": 0.0005333641159801563,
      "loss": 2.5879,
      "step": 49835
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.748134195804596,
      "learning_rate": 0.0005333615453986025,
      "loss": 2.1053,
      "step": 49836
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.6416910290718079,
      "learning_rate": 0.0005333589747736624,
      "loss": 2.4532,
      "step": 49837
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7537087798118591,
      "learning_rate": 0.0005333564041053363,
      "loss": 2.0739,
      "step": 49838
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.6642357707023621,
      "learning_rate": 0.0005333538333936251,
      "loss": 2.429,
      "step": 49839
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.6529742479324341,
      "learning_rate": 0.0005333512626385289,
      "loss": 2.1927,
      "step": 49840
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7216897010803223,
      "learning_rate": 0.0005333486918400483,
      "loss": 2.1828,
      "step": 49841
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.8176881074905396,
      "learning_rate": 0.0005333461209981838,
      "loss": 2.0638,
      "step": 49842
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.6470925807952881,
      "learning_rate": 0.0005333435501129357,
      "loss": 2.2004,
      "step": 49843
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7691677212715149,
      "learning_rate": 0.0005333409791843048,
      "loss": 2.3068,
      "step": 49844
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7224409580230713,
      "learning_rate": 0.0005333384082122914,
      "loss": 2.2103,
      "step": 49845
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7291326522827148,
      "learning_rate": 0.0005333358371968959,
      "loss": 2.215,
      "step": 49846
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.6233657002449036,
      "learning_rate": 0.0005333332661381189,
      "loss": 2.1194,
      "step": 49847
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7186859846115112,
      "learning_rate": 0.0005333306950359609,
      "loss": 2.294,
      "step": 49848
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.677291214466095,
      "learning_rate": 0.0005333281238904222,
      "loss": 2.3273,
      "step": 49849
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.679135262966156,
      "learning_rate": 0.0005333255527015035,
      "loss": 2.3186,
      "step": 49850
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.710064172744751,
      "learning_rate": 0.0005333229814692051,
      "loss": 2.1837,
      "step": 49851
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.6789055466651917,
      "learning_rate": 0.0005333204101935277,
      "loss": 2.2264,
      "step": 49852
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7877552509307861,
      "learning_rate": 0.0005333178388744715,
      "loss": 2.2118,
      "step": 49853
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7153369188308716,
      "learning_rate": 0.0005333152675120371,
      "loss": 2.3186,
      "step": 49854
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7722539901733398,
      "learning_rate": 0.0005333126961062249,
      "loss": 2.2275,
      "step": 49855
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7510951161384583,
      "learning_rate": 0.0005333101246570356,
      "loss": 2.1962,
      "step": 49856
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.6641226410865784,
      "learning_rate": 0.0005333075531644695,
      "loss": 2.1537,
      "step": 49857
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7095726132392883,
      "learning_rate": 0.0005333049816285272,
      "loss": 2.0839,
      "step": 49858
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.6530508399009705,
      "learning_rate": 0.0005333024100492091,
      "loss": 2.1361,
      "step": 49859
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7666824460029602,
      "learning_rate": 0.0005332998384265156,
      "loss": 2.2309,
      "step": 49860
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7163316607475281,
      "learning_rate": 0.0005332972667604472,
      "loss": 2.2991,
      "step": 49861
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.8533549308776855,
      "learning_rate": 0.0005332946950510047,
      "loss": 2.2183,
      "step": 49862
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.892621636390686,
      "learning_rate": 0.0005332921232981882,
      "loss": 2.034,
      "step": 49863
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7651025056838989,
      "learning_rate": 0.0005332895515019981,
      "loss": 2.1443,
      "step": 49864
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.8906517624855042,
      "learning_rate": 0.0005332869796624352,
      "loss": 2.2065,
      "step": 49865
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.9022710919380188,
      "learning_rate": 0.0005332844077794998,
      "loss": 2.0987,
      "step": 49866
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.6402943730354309,
      "learning_rate": 0.0005332818358531925,
      "loss": 2.2889,
      "step": 49867
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.6990856528282166,
      "learning_rate": 0.0005332792638835137,
      "loss": 2.2261,
      "step": 49868
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.1258668899536133,
      "learning_rate": 0.0005332766918704638,
      "loss": 2.3349,
      "step": 49869
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.6611765623092651,
      "learning_rate": 0.0005332741198140434,
      "loss": 2.1993,
      "step": 49870
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7087013125419617,
      "learning_rate": 0.000533271547714253,
      "loss": 2.29,
      "step": 49871
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7734848260879517,
      "learning_rate": 0.0005332689755710929,
      "loss": 2.1217,
      "step": 49872
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.8104434013366699,
      "learning_rate": 0.0005332664033845637,
      "loss": 2.2278,
      "step": 49873
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.8554308414459229,
      "learning_rate": 0.000533263831154666,
      "loss": 2.4764,
      "step": 49874
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7511026859283447,
      "learning_rate": 0.0005332612588814001,
      "loss": 2.3243,
      "step": 49875
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.8213986158370972,
      "learning_rate": 0.0005332586865647665,
      "loss": 2.198,
      "step": 49876
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7177872657775879,
      "learning_rate": 0.0005332561142047657,
      "loss": 2.2893,
      "step": 49877
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.907517671585083,
      "learning_rate": 0.0005332535418013983,
      "loss": 2.351,
      "step": 49878
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.6591959595680237,
      "learning_rate": 0.0005332509693546644,
      "loss": 2.2652,
      "step": 49879
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7790653705596924,
      "learning_rate": 0.000533248396864565,
      "loss": 2.2618,
      "step": 49880
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7336204051971436,
      "learning_rate": 0.0005332458243311002,
      "loss": 2.4671,
      "step": 49881
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7123467922210693,
      "learning_rate": 0.0005332432517542706,
      "loss": 2.6034,
      "step": 49882
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7087694406509399,
      "learning_rate": 0.0005332406791340767,
      "loss": 2.3158,
      "step": 49883
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7805606722831726,
      "learning_rate": 0.000533238106470519,
      "loss": 2.4835,
      "step": 49884
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.6647218465805054,
      "learning_rate": 0.000533235533763598,
      "loss": 1.9741,
      "step": 49885
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7101272344589233,
      "learning_rate": 0.0005332329610133139,
      "loss": 2.0496,
      "step": 49886
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7649831771850586,
      "learning_rate": 0.0005332303882196676,
      "loss": 2.3538,
      "step": 49887
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7228033542633057,
      "learning_rate": 0.0005332278153826591,
      "loss": 2.1956,
      "step": 49888
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7049003839492798,
      "learning_rate": 0.0005332252425022893,
      "loss": 2.2195,
      "step": 49889
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7264401912689209,
      "learning_rate": 0.0005332226695785586,
      "loss": 2.564,
      "step": 49890
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.6528550386428833,
      "learning_rate": 0.0005332200966114674,
      "loss": 2.2541,
      "step": 49891
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.6677466630935669,
      "learning_rate": 0.0005332175236010161,
      "loss": 2.1103,
      "step": 49892
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7803651690483093,
      "learning_rate": 0.0005332149505472051,
      "loss": 2.2133,
      "step": 49893
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7599455714225769,
      "learning_rate": 0.0005332123774500353,
      "loss": 2.2578,
      "step": 49894
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.8985555768013,
      "learning_rate": 0.0005332098043095069,
      "loss": 2.2234,
      "step": 49895
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.6894696950912476,
      "learning_rate": 0.0005332072311256202,
      "loss": 2.436,
      "step": 49896
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7500556111335754,
      "learning_rate": 0.0005332046578983761,
      "loss": 2.0647,
      "step": 49897
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.6624525785446167,
      "learning_rate": 0.0005332020846277747,
      "loss": 2.2328,
      "step": 49898
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.6634528040885925,
      "learning_rate": 0.0005331995113138167,
      "loss": 2.3621,
      "step": 49899
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.6745785474777222,
      "learning_rate": 0.0005331969379565025,
      "loss": 2.325,
      "step": 49900
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7157410383224487,
      "learning_rate": 0.0005331943645558325,
      "loss": 2.2218,
      "step": 49901
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.6526656150817871,
      "learning_rate": 0.0005331917911118075,
      "loss": 2.1534,
      "step": 49902
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7143282890319824,
      "learning_rate": 0.0005331892176244275,
      "loss": 2.1644,
      "step": 49903
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7415431141853333,
      "learning_rate": 0.0005331866440936934,
      "loss": 2.237,
      "step": 49904
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7021306753158569,
      "learning_rate": 0.0005331840705196054,
      "loss": 2.3466,
      "step": 49905
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7375777363777161,
      "learning_rate": 0.000533181496902164,
      "loss": 2.1344,
      "step": 49906
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7078282833099365,
      "learning_rate": 0.0005331789232413699,
      "loss": 2.2644,
      "step": 49907
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.6717134714126587,
      "learning_rate": 0.0005331763495372234,
      "loss": 2.2202,
      "step": 49908
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.6379837393760681,
      "learning_rate": 0.000533173775789725,
      "loss": 2.3247,
      "step": 49909
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.8724043965339661,
      "learning_rate": 0.0005331712019988751,
      "loss": 2.4185,
      "step": 49910
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.8113498687744141,
      "learning_rate": 0.0005331686281646745,
      "loss": 2.1368,
      "step": 49911
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7628543972969055,
      "learning_rate": 0.0005331660542871233,
      "loss": 2.1733,
      "step": 49912
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.706754744052887,
      "learning_rate": 0.0005331634803662222,
      "loss": 2.0283,
      "step": 49913
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.6699628233909607,
      "learning_rate": 0.0005331609064019715,
      "loss": 2.4582,
      "step": 49914
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7792750000953674,
      "learning_rate": 0.0005331583323943717,
      "loss": 2.2383,
      "step": 49915
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.8044913411140442,
      "learning_rate": 0.0005331557583434236,
      "loss": 2.1358,
      "step": 49916
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7014675736427307,
      "learning_rate": 0.0005331531842491273,
      "loss": 2.2264,
      "step": 49917
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.9038698673248291,
      "learning_rate": 0.0005331506101114835,
      "loss": 2.3927,
      "step": 49918
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.6472669839859009,
      "learning_rate": 0.0005331480359304926,
      "loss": 2.2541,
      "step": 49919
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.8250529170036316,
      "learning_rate": 0.000533145461706155,
      "loss": 2.395,
      "step": 49920
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7556108236312866,
      "learning_rate": 0.0005331428874384714,
      "loss": 2.4679,
      "step": 49921
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7393746376037598,
      "learning_rate": 0.0005331403131274419,
      "loss": 2.2232,
      "step": 49922
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.6986897587776184,
      "learning_rate": 0.0005331377387730674,
      "loss": 2.2925,
      "step": 49923
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.841118335723877,
      "learning_rate": 0.0005331351643753482,
      "loss": 2.273,
      "step": 49924
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.8025989532470703,
      "learning_rate": 0.0005331325899342847,
      "loss": 2.062,
      "step": 49925
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7479124665260315,
      "learning_rate": 0.0005331300154498775,
      "loss": 2.1489,
      "step": 49926
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7270470857620239,
      "learning_rate": 0.0005331274409221269,
      "loss": 2.2285,
      "step": 49927
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7189992666244507,
      "learning_rate": 0.0005331248663510337,
      "loss": 2.158,
      "step": 49928
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.8235183358192444,
      "learning_rate": 0.0005331222917365981,
      "loss": 2.2035,
      "step": 49929
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7082734704017639,
      "learning_rate": 0.0005331197170788207,
      "loss": 2.2987,
      "step": 49930
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7222757339477539,
      "learning_rate": 0.0005331171423777019,
      "loss": 2.0529,
      "step": 49931
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7022104263305664,
      "learning_rate": 0.0005331145676332423,
      "loss": 2.1866,
      "step": 49932
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.8075495958328247,
      "learning_rate": 0.0005331119928454422,
      "loss": 2.3105,
      "step": 49933
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.8589881658554077,
      "learning_rate": 0.0005331094180143021,
      "loss": 1.8265,
      "step": 49934
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.6733793020248413,
      "learning_rate": 0.0005331068431398228,
      "loss": 2.2394,
      "step": 49935
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7069392204284668,
      "learning_rate": 0.0005331042682220044,
      "loss": 2.2172,
      "step": 49936
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7299283742904663,
      "learning_rate": 0.0005331016932608475,
      "loss": 2.1682,
      "step": 49937
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.6759912967681885,
      "learning_rate": 0.0005330991182563526,
      "loss": 2.0527,
      "step": 49938
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7381405830383301,
      "learning_rate": 0.0005330965432085202,
      "loss": 2.3374,
      "step": 49939
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7587896585464478,
      "learning_rate": 0.0005330939681173507,
      "loss": 2.1954,
      "step": 49940
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7447888851165771,
      "learning_rate": 0.0005330913929828447,
      "loss": 2.459,
      "step": 49941
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.6607433557510376,
      "learning_rate": 0.0005330888178050027,
      "loss": 2.026,
      "step": 49942
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7364915609359741,
      "learning_rate": 0.0005330862425838249,
      "loss": 2.2813,
      "step": 49943
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7851594090461731,
      "learning_rate": 0.0005330836673193121,
      "loss": 2.1695,
      "step": 49944
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.6727442145347595,
      "learning_rate": 0.0005330810920114646,
      "loss": 2.1668,
      "step": 49945
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.8649995923042297,
      "learning_rate": 0.0005330785166602828,
      "loss": 2.3322,
      "step": 49946
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.6909773349761963,
      "learning_rate": 0.0005330759412657674,
      "loss": 2.1731,
      "step": 49947
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.6935697197914124,
      "learning_rate": 0.0005330733658279189,
      "loss": 2.208,
      "step": 49948
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7331412434577942,
      "learning_rate": 0.0005330707903467376,
      "loss": 2.4397,
      "step": 49949
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7823687791824341,
      "learning_rate": 0.000533068214822224,
      "loss": 2.3395,
      "step": 49950
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.741105318069458,
      "learning_rate": 0.0005330656392543786,
      "loss": 2.2309,
      "step": 49951
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7550525665283203,
      "learning_rate": 0.0005330630636432019,
      "loss": 2.3391,
      "step": 49952
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7287142276763916,
      "learning_rate": 0.0005330604879886944,
      "loss": 2.2157,
      "step": 49953
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.6746880412101746,
      "learning_rate": 0.0005330579122908565,
      "loss": 2.4668,
      "step": 49954
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.8062894344329834,
      "learning_rate": 0.0005330553365496889,
      "loss": 2.186,
      "step": 49955
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.8151466250419617,
      "learning_rate": 0.0005330527607651917,
      "loss": 2.382,
      "step": 49956
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.8230727314949036,
      "learning_rate": 0.0005330501849373657,
      "loss": 2.35,
      "step": 49957
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.6473363041877747,
      "learning_rate": 0.0005330476090662113,
      "loss": 2.305,
      "step": 49958
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7168652415275574,
      "learning_rate": 0.000533045033151729,
      "loss": 2.2281,
      "step": 49959
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.6515582799911499,
      "learning_rate": 0.0005330424571939192,
      "loss": 2.1464,
      "step": 49960
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7629051208496094,
      "learning_rate": 0.0005330398811927823,
      "loss": 2.2376,
      "step": 49961
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7302944660186768,
      "learning_rate": 0.0005330373051483189,
      "loss": 2.1838,
      "step": 49962
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.6941796541213989,
      "learning_rate": 0.0005330347290605296,
      "loss": 2.3725,
      "step": 49963
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.6827970743179321,
      "learning_rate": 0.0005330321529294146,
      "loss": 2.3574,
      "step": 49964
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7197142243385315,
      "learning_rate": 0.0005330295767549746,
      "loss": 1.9656,
      "step": 49965
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.704603910446167,
      "learning_rate": 0.00053302700053721,
      "loss": 2.2424,
      "step": 49966
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.8883803486824036,
      "learning_rate": 0.0005330244242761212,
      "loss": 2.1914,
      "step": 49967
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7184898257255554,
      "learning_rate": 0.0005330218479717089,
      "loss": 2.0431,
      "step": 49968
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.6800186038017273,
      "learning_rate": 0.0005330192716239733,
      "loss": 2.1136,
      "step": 49969
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.8706051111221313,
      "learning_rate": 0.0005330166952329151,
      "loss": 2.2716,
      "step": 49970
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.6677778959274292,
      "learning_rate": 0.0005330141187985348,
      "loss": 2.2906,
      "step": 49971
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.709520697593689,
      "learning_rate": 0.0005330115423208326,
      "loss": 2.3232,
      "step": 49972
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7409167289733887,
      "learning_rate": 0.0005330089657998091,
      "loss": 2.2589,
      "step": 49973
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7296558022499084,
      "learning_rate": 0.000533006389235465,
      "loss": 2.2169,
      "step": 49974
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7208848595619202,
      "learning_rate": 0.0005330038126278005,
      "loss": 2.2582,
      "step": 49975
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7174751162528992,
      "learning_rate": 0.0005330012359768164,
      "loss": 2.3612,
      "step": 49976
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7934936285018921,
      "learning_rate": 0.0005329986592825128,
      "loss": 2.1246,
      "step": 49977
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7827550172805786,
      "learning_rate": 0.0005329960825448904,
      "loss": 2.4691,
      "step": 49978
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7980814576148987,
      "learning_rate": 0.0005329935057639495,
      "loss": 2.1897,
      "step": 49979
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7263004779815674,
      "learning_rate": 0.0005329909289396908,
      "loss": 2.2211,
      "step": 49980
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.6663869023323059,
      "learning_rate": 0.0005329883520721147,
      "loss": 2.3515,
      "step": 49981
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.6970306634902954,
      "learning_rate": 0.0005329857751612218,
      "loss": 2.3083,
      "step": 49982
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.6644246578216553,
      "learning_rate": 0.0005329831982070123,
      "loss": 2.4665,
      "step": 49983
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7084671258926392,
      "learning_rate": 0.0005329806212094868,
      "loss": 2.2642,
      "step": 49984
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.6508880853652954,
      "learning_rate": 0.0005329780441686458,
      "loss": 2.1415,
      "step": 49985
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.6690397262573242,
      "learning_rate": 0.0005329754670844899,
      "loss": 2.0989,
      "step": 49986
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.6899683475494385,
      "learning_rate": 0.0005329728899570194,
      "loss": 2.1434,
      "step": 49987
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.8207539916038513,
      "learning_rate": 0.0005329703127862349,
      "loss": 2.5128,
      "step": 49988
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.6492436528205872,
      "learning_rate": 0.0005329677355721367,
      "loss": 2.1471,
      "step": 49989
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7098940014839172,
      "learning_rate": 0.0005329651583147254,
      "loss": 2.2916,
      "step": 49990
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.8289507031440735,
      "learning_rate": 0.0005329625810140016,
      "loss": 2.3221,
      "step": 49991
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.6767268776893616,
      "learning_rate": 0.0005329600036699656,
      "loss": 2.3659,
      "step": 49992
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.6646426916122437,
      "learning_rate": 0.000532957426282618,
      "loss": 2.2889,
      "step": 49993
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7354316711425781,
      "learning_rate": 0.0005329548488519591,
      "loss": 2.3331,
      "step": 49994
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.756851851940155,
      "learning_rate": 0.0005329522713779895,
      "loss": 2.3672,
      "step": 49995
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7902132868766785,
      "learning_rate": 0.0005329496938607098,
      "loss": 2.4079,
      "step": 49996
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.675774097442627,
      "learning_rate": 0.0005329471163001202,
      "loss": 2.3069,
      "step": 49997
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.6729868054389954,
      "learning_rate": 0.0005329445386962215,
      "loss": 2.5184,
      "step": 49998
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.6975084543228149,
      "learning_rate": 0.0005329419610490139,
      "loss": 2.3522,
      "step": 49999
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7208496332168579,
      "learning_rate": 0.000532939383358498,
      "loss": 2.2353,
      "step": 50000
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7008007168769836,
      "learning_rate": 0.0005329368056246743,
      "loss": 2.0763,
      "step": 50001
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.6998897790908813,
      "learning_rate": 0.0005329342278475432,
      "loss": 2.451,
      "step": 50002
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.8003319501876831,
      "learning_rate": 0.0005329316500271053,
      "loss": 2.1883,
      "step": 50003
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.6807101964950562,
      "learning_rate": 0.0005329290721633609,
      "loss": 2.5224,
      "step": 50004
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.6980655193328857,
      "learning_rate": 0.0005329264942563108,
      "loss": 2.283,
      "step": 50005
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7359098792076111,
      "learning_rate": 0.0005329239163059552,
      "loss": 2.2645,
      "step": 50006
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7461017966270447,
      "learning_rate": 0.0005329213383122944,
      "loss": 2.2835,
      "step": 50007
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.710892379283905,
      "learning_rate": 0.0005329187602753294,
      "loss": 2.2095,
      "step": 50008
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7580149173736572,
      "learning_rate": 0.0005329161821950604,
      "loss": 2.0053,
      "step": 50009
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.8046203851699829,
      "learning_rate": 0.0005329136040714877,
      "loss": 2.2247,
      "step": 50010
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7161349654197693,
      "learning_rate": 0.0005329110259046122,
      "loss": 2.16,
      "step": 50011
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7307741045951843,
      "learning_rate": 0.000532908447694434,
      "loss": 2.2498,
      "step": 50012
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.6851939558982849,
      "learning_rate": 0.0005329058694409537,
      "loss": 2.2172,
      "step": 50013
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7617959976196289,
      "learning_rate": 0.0005329032911441719,
      "loss": 2.2594,
      "step": 50014
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7274668216705322,
      "learning_rate": 0.0005329007128040889,
      "loss": 2.3165,
      "step": 50015
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7619709372520447,
      "learning_rate": 0.0005328981344207052,
      "loss": 2.2777,
      "step": 50016
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.6928849816322327,
      "learning_rate": 0.0005328955559940214,
      "loss": 2.0473,
      "step": 50017
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.6742656826972961,
      "learning_rate": 0.0005328929775240379,
      "loss": 2.1696,
      "step": 50018
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.6549320816993713,
      "learning_rate": 0.0005328903990107554,
      "loss": 2.0781,
      "step": 50019
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7529513835906982,
      "learning_rate": 0.0005328878204541739,
      "loss": 2.456,
      "step": 50020
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7123509645462036,
      "learning_rate": 0.0005328852418542943,
      "loss": 2.1243,
      "step": 50021
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.6526802778244019,
      "learning_rate": 0.0005328826632111168,
      "loss": 2.0992,
      "step": 50022
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.730911910533905,
      "learning_rate": 0.0005328800845246423,
      "loss": 2.3125,
      "step": 50023
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7968183755874634,
      "learning_rate": 0.0005328775057948708,
      "loss": 2.346,
      "step": 50024
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.6902251243591309,
      "learning_rate": 0.000532874927021803,
      "loss": 2.2802,
      "step": 50025
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7110227346420288,
      "learning_rate": 0.0005328723482054393,
      "loss": 2.2925,
      "step": 50026
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7388001680374146,
      "learning_rate": 0.0005328697693457803,
      "loss": 2.3823,
      "step": 50027
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.6884439587593079,
      "learning_rate": 0.0005328671904428264,
      "loss": 2.3034,
      "step": 50028
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.680610179901123,
      "learning_rate": 0.0005328646114965781,
      "loss": 2.1022,
      "step": 50029
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.6911253333091736,
      "learning_rate": 0.0005328620325070358,
      "loss": 2.3192,
      "step": 50030
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.6639835238456726,
      "learning_rate": 0.0005328594534742002,
      "loss": 2.4071,
      "step": 50031
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7456450462341309,
      "learning_rate": 0.0005328568743980716,
      "loss": 2.0829,
      "step": 50032
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7004591822624207,
      "learning_rate": 0.0005328542952786504,
      "loss": 2.3652,
      "step": 50033
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7633797526359558,
      "learning_rate": 0.0005328517161159373,
      "loss": 2.1838,
      "step": 50034
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.6749417781829834,
      "learning_rate": 0.0005328491369099327,
      "loss": 2.1604,
      "step": 50035
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.667289674282074,
      "learning_rate": 0.000532846557660637,
      "loss": 2.217,
      "step": 50036
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7245556712150574,
      "learning_rate": 0.0005328439783680506,
      "loss": 2.3206,
      "step": 50037
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7651423811912537,
      "learning_rate": 0.0005328413990321743,
      "loss": 2.2428,
      "step": 50038
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7353946566581726,
      "learning_rate": 0.0005328388196530082,
      "loss": 2.2013,
      "step": 50039
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7583780288696289,
      "learning_rate": 0.0005328362402305532,
      "loss": 1.907,
      "step": 50040
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.6794711947441101,
      "learning_rate": 0.0005328336607648094,
      "loss": 2.1925,
      "step": 50041
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7172942161560059,
      "learning_rate": 0.0005328310812557774,
      "loss": 2.1269,
      "step": 50042
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.6992180347442627,
      "learning_rate": 0.0005328285017034578,
      "loss": 2.3699,
      "step": 50043
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7141591906547546,
      "learning_rate": 0.0005328259221078509,
      "loss": 2.1768,
      "step": 50044
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.741324245929718,
      "learning_rate": 0.0005328233424689574,
      "loss": 2.2478,
      "step": 50045
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.6874564290046692,
      "learning_rate": 0.0005328207627867775,
      "loss": 2.1743,
      "step": 50046
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7499321103096008,
      "learning_rate": 0.0005328181830613119,
      "loss": 2.3351,
      "step": 50047
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7022055983543396,
      "learning_rate": 0.0005328156032925609,
      "loss": 2.2488,
      "step": 50048
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7617379426956177,
      "learning_rate": 0.0005328130234805253,
      "loss": 2.238,
      "step": 50049
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.8947694301605225,
      "learning_rate": 0.0005328104436252052,
      "loss": 2.299,
      "step": 50050
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7305428385734558,
      "learning_rate": 0.0005328078637266013,
      "loss": 2.4202,
      "step": 50051
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7399590611457825,
      "learning_rate": 0.000532805283784714,
      "loss": 2.3895,
      "step": 50052
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7423897981643677,
      "learning_rate": 0.0005328027037995438,
      "loss": 2.5321,
      "step": 50053
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.6965094804763794,
      "learning_rate": 0.0005328001237710912,
      "loss": 2.5171,
      "step": 50054
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7006184458732605,
      "learning_rate": 0.0005327975436993566,
      "loss": 2.1521,
      "step": 50055
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.6972600221633911,
      "learning_rate": 0.0005327949635843405,
      "loss": 2.1387,
      "step": 50056
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.6703349351882935,
      "learning_rate": 0.0005327923834260437,
      "loss": 2.2138,
      "step": 50057
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.6859970092773438,
      "learning_rate": 0.0005327898032244662,
      "loss": 2.443,
      "step": 50058
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.6785403490066528,
      "learning_rate": 0.0005327872229796087,
      "loss": 2.2245,
      "step": 50059
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7085500955581665,
      "learning_rate": 0.0005327846426914716,
      "loss": 2.2839,
      "step": 50060
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.686065137386322,
      "learning_rate": 0.0005327820623600556,
      "loss": 2.2744,
      "step": 50061
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.8411651253700256,
      "learning_rate": 0.0005327794819853608,
      "loss": 1.9631,
      "step": 50062
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7105996012687683,
      "learning_rate": 0.0005327769015673881,
      "loss": 2.1372,
      "step": 50063
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.9431042671203613,
      "learning_rate": 0.0005327743211061376,
      "loss": 1.9653,
      "step": 50064
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.6401776075363159,
      "learning_rate": 0.0005327717406016101,
      "loss": 2.2016,
      "step": 50065
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.8154312372207642,
      "learning_rate": 0.0005327691600538059,
      "loss": 2.3765,
      "step": 50066
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.6956945061683655,
      "learning_rate": 0.0005327665794627256,
      "loss": 2.2415,
      "step": 50067
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.6830713748931885,
      "learning_rate": 0.0005327639988283694,
      "loss": 2.3294,
      "step": 50068
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.75959712266922,
      "learning_rate": 0.0005327614181507382,
      "loss": 2.2576,
      "step": 50069
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7368022203445435,
      "learning_rate": 0.0005327588374298321,
      "loss": 2.0663,
      "step": 50070
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.8330867290496826,
      "learning_rate": 0.0005327562566656518,
      "loss": 2.3498,
      "step": 50071
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7539634108543396,
      "learning_rate": 0.0005327536758581977,
      "loss": 2.2605,
      "step": 50072
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.8596155643463135,
      "learning_rate": 0.0005327510950074704,
      "loss": 2.2021,
      "step": 50073
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7491010427474976,
      "learning_rate": 0.0005327485141134701,
      "loss": 2.361,
      "step": 50074
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7536265254020691,
      "learning_rate": 0.0005327459331761976,
      "loss": 2.6711,
      "step": 50075
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.6897947192192078,
      "learning_rate": 0.0005327433521956532,
      "loss": 2.0439,
      "step": 50076
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7810091376304626,
      "learning_rate": 0.0005327407711718374,
      "loss": 2.1876,
      "step": 50077
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.8009688854217529,
      "learning_rate": 0.0005327381901047507,
      "loss": 2.3671,
      "step": 50078
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7259528040885925,
      "learning_rate": 0.0005327356089943935,
      "loss": 2.2889,
      "step": 50079
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7737375497817993,
      "learning_rate": 0.0005327330278407664,
      "loss": 2.247,
      "step": 50080
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.8203487396240234,
      "learning_rate": 0.00053273044664387,
      "loss": 1.9939,
      "step": 50081
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7056400775909424,
      "learning_rate": 0.0005327278654037044,
      "loss": 2.2314,
      "step": 50082
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.6848273277282715,
      "learning_rate": 0.0005327252841202704,
      "loss": 2.3948,
      "step": 50083
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.6580462455749512,
      "learning_rate": 0.0005327227027935684,
      "loss": 2.1168,
      "step": 50084
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.6722813844680786,
      "learning_rate": 0.0005327201214235988,
      "loss": 2.2041,
      "step": 50085
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.6740105152130127,
      "learning_rate": 0.0005327175400103622,
      "loss": 2.0718,
      "step": 50086
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.8067485094070435,
      "learning_rate": 0.000532714958553859,
      "loss": 2.3162,
      "step": 50087
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7040440440177917,
      "learning_rate": 0.0005327123770540896,
      "loss": 2.217,
      "step": 50088
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.8267340064048767,
      "learning_rate": 0.0005327097955110546,
      "loss": 2.2727,
      "step": 50089
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.6883131265640259,
      "learning_rate": 0.0005327072139247545,
      "loss": 2.0992,
      "step": 50090
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7182831168174744,
      "learning_rate": 0.0005327046322951898,
      "loss": 2.3764,
      "step": 50091
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7817636132240295,
      "learning_rate": 0.0005327020506223608,
      "loss": 2.4763,
      "step": 50092
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7193930149078369,
      "learning_rate": 0.0005326994689062682,
      "loss": 2.2958,
      "step": 50093
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.777960479259491,
      "learning_rate": 0.0005326968871469124,
      "loss": 2.2243,
      "step": 50094
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.6846292614936829,
      "learning_rate": 0.0005326943053442937,
      "loss": 2.3238,
      "step": 50095
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.6909163594245911,
      "learning_rate": 0.0005326917234984128,
      "loss": 2.2146,
      "step": 50096
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7917765378952026,
      "learning_rate": 0.0005326891416092701,
      "loss": 2.2798,
      "step": 50097
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7248749136924744,
      "learning_rate": 0.0005326865596768661,
      "loss": 2.2166,
      "step": 50098
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7853624224662781,
      "learning_rate": 0.0005326839777012014,
      "loss": 2.2668,
      "step": 50099
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7994126081466675,
      "learning_rate": 0.0005326813956822762,
      "loss": 2.2758,
      "step": 50100
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7530993223190308,
      "learning_rate": 0.0005326788136200913,
      "loss": 2.1049,
      "step": 50101
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.8508371710777283,
      "learning_rate": 0.0005326762315146469,
      "loss": 2.3597,
      "step": 50102
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.6741352081298828,
      "learning_rate": 0.0005326736493659435,
      "loss": 2.3499,
      "step": 50103
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.8146366477012634,
      "learning_rate": 0.0005326710671739818,
      "loss": 2.2913,
      "step": 50104
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7093413472175598,
      "learning_rate": 0.0005326684849387621,
      "loss": 2.3328,
      "step": 50105
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.6897331476211548,
      "learning_rate": 0.0005326659026602852,
      "loss": 2.1271,
      "step": 50106
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.6914897561073303,
      "learning_rate": 0.000532663320338551,
      "loss": 2.135,
      "step": 50107
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.766829252243042,
      "learning_rate": 0.0005326607379735604,
      "loss": 2.2544,
      "step": 50108
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7303034663200378,
      "learning_rate": 0.0005326581555653138,
      "loss": 2.4794,
      "step": 50109
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.6890945434570312,
      "learning_rate": 0.0005326555731138117,
      "loss": 2.169,
      "step": 50110
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.8667230010032654,
      "learning_rate": 0.0005326529906190543,
      "loss": 2.1837,
      "step": 50111
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.6471323370933533,
      "learning_rate": 0.0005326504080810426,
      "loss": 2.3069,
      "step": 50112
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7007129788398743,
      "learning_rate": 0.0005326478254997766,
      "loss": 2.1608,
      "step": 50113
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.2493408918380737,
      "learning_rate": 0.000532645242875257,
      "loss": 2.1724,
      "step": 50114
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7272805571556091,
      "learning_rate": 0.0005326426602074844,
      "loss": 2.2072,
      "step": 50115
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7939500212669373,
      "learning_rate": 0.000532640077496459,
      "loss": 2.3579,
      "step": 50116
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7881853580474854,
      "learning_rate": 0.0005326374947421816,
      "loss": 2.3638,
      "step": 50117
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.6887658834457397,
      "learning_rate": 0.0005326349119446523,
      "loss": 2.2167,
      "step": 50118
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.690091609954834,
      "learning_rate": 0.0005326323291038719,
      "loss": 2.1614,
      "step": 50119
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7026370763778687,
      "learning_rate": 0.0005326297462198407,
      "loss": 2.4396,
      "step": 50120
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7363360524177551,
      "learning_rate": 0.0005326271632925592,
      "loss": 2.2,
      "step": 50121
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7040321826934814,
      "learning_rate": 0.0005326245803220279,
      "loss": 2.1044,
      "step": 50122
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.6721693277359009,
      "learning_rate": 0.0005326219973082474,
      "loss": 2.151,
      "step": 50123
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.6454231142997742,
      "learning_rate": 0.000532619414251218,
      "loss": 1.8955,
      "step": 50124
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7268481850624084,
      "learning_rate": 0.0005326168311509404,
      "loss": 2.209,
      "step": 50125
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.716777503490448,
      "learning_rate": 0.0005326142480074149,
      "loss": 2.1363,
      "step": 50126
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7623865008354187,
      "learning_rate": 0.0005326116648206419,
      "loss": 2.2609,
      "step": 50127
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.821145236492157,
      "learning_rate": 0.0005326090815906221,
      "loss": 2.2335,
      "step": 50128
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7611124515533447,
      "learning_rate": 0.0005326064983173558,
      "loss": 2.2909,
      "step": 50129
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7384539246559143,
      "learning_rate": 0.0005326039150008436,
      "loss": 2.4528,
      "step": 50130
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.0657902956008911,
      "learning_rate": 0.000532601331641086,
      "loss": 2.2399,
      "step": 50131
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7537723779678345,
      "learning_rate": 0.0005325987482380834,
      "loss": 2.2507,
      "step": 50132
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.704913854598999,
      "learning_rate": 0.0005325961647918362,
      "loss": 2.1905,
      "step": 50133
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7749239802360535,
      "learning_rate": 0.0005325935813023451,
      "loss": 2.3337,
      "step": 50134
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7835249900817871,
      "learning_rate": 0.0005325909977696104,
      "loss": 2.3137,
      "step": 50135
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7609446048736572,
      "learning_rate": 0.0005325884141936327,
      "loss": 2.302,
      "step": 50136
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7972204685211182,
      "learning_rate": 0.0005325858305744124,
      "loss": 2.158,
      "step": 50137
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7309081554412842,
      "learning_rate": 0.0005325832469119499,
      "loss": 1.9074,
      "step": 50138
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7139919400215149,
      "learning_rate": 0.0005325806632062459,
      "loss": 2.5654,
      "step": 50139
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.6780640482902527,
      "learning_rate": 0.0005325780794573008,
      "loss": 2.3053,
      "step": 50140
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.645369827747345,
      "learning_rate": 0.0005325754956651149,
      "loss": 2.3233,
      "step": 50141
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7496499419212341,
      "learning_rate": 0.0005325729118296889,
      "loss": 2.3114,
      "step": 50142
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.6452721953392029,
      "learning_rate": 0.0005325703279510232,
      "loss": 2.2005,
      "step": 50143
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.6502993106842041,
      "learning_rate": 0.0005325677440291183,
      "loss": 2.2841,
      "step": 50144
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7116087079048157,
      "learning_rate": 0.0005325651600639747,
      "loss": 2.3247,
      "step": 50145
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.9829107522964478,
      "learning_rate": 0.0005325625760555929,
      "loss": 2.319,
      "step": 50146
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7319371700286865,
      "learning_rate": 0.0005325599920039732,
      "loss": 2.1744,
      "step": 50147
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.9842884540557861,
      "learning_rate": 0.0005325574079091162,
      "loss": 2.2698,
      "step": 50148
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.8790618777275085,
      "learning_rate": 0.0005325548237710225,
      "loss": 2.387,
      "step": 50149
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7330570816993713,
      "learning_rate": 0.0005325522395896923,
      "loss": 2.1619,
      "step": 50150
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7830827832221985,
      "learning_rate": 0.0005325496553651264,
      "loss": 2.2683,
      "step": 50151
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.8493123054504395,
      "learning_rate": 0.0005325470710973252,
      "loss": 2.426,
      "step": 50152
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7295597791671753,
      "learning_rate": 0.0005325444867862889,
      "loss": 2.2439,
      "step": 50153
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7241578698158264,
      "learning_rate": 0.0005325419024320182,
      "loss": 2.3999,
      "step": 50154
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.6571833491325378,
      "learning_rate": 0.0005325393180345137,
      "loss": 2.065,
      "step": 50155
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7671559453010559,
      "learning_rate": 0.0005325367335937758,
      "loss": 2.2875,
      "step": 50156
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7018635272979736,
      "learning_rate": 0.0005325341491098048,
      "loss": 2.2116,
      "step": 50157
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.690788984298706,
      "learning_rate": 0.0005325315645826013,
      "loss": 2.1762,
      "step": 50158
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.0934072732925415,
      "learning_rate": 0.0005325289800121658,
      "loss": 2.2271,
      "step": 50159
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7261900305747986,
      "learning_rate": 0.0005325263953984988,
      "loss": 2.4993,
      "step": 50160
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7851223349571228,
      "learning_rate": 0.0005325238107416008,
      "loss": 2.4048,
      "step": 50161
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.6973597407341003,
      "learning_rate": 0.0005325212260414723,
      "loss": 2.411,
      "step": 50162
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.6922281384468079,
      "learning_rate": 0.0005325186412981136,
      "loss": 2.1453,
      "step": 50163
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.720519483089447,
      "learning_rate": 0.0005325160565115253,
      "loss": 2.367,
      "step": 50164
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7437446713447571,
      "learning_rate": 0.0005325134716817079,
      "loss": 2.2323,
      "step": 50165
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.6984475255012512,
      "learning_rate": 0.0005325108868086618,
      "loss": 2.2205,
      "step": 50166
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.8583559989929199,
      "learning_rate": 0.0005325083018923876,
      "loss": 2.3048,
      "step": 50167
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7350152134895325,
      "learning_rate": 0.0005325057169328857,
      "loss": 2.1201,
      "step": 50168
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7429428100585938,
      "learning_rate": 0.0005325031319301566,
      "loss": 2.1189,
      "step": 50169
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.8576043248176575,
      "learning_rate": 0.0005325005468842008,
      "loss": 2.5012,
      "step": 50170
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.729906439781189,
      "learning_rate": 0.0005324979617950188,
      "loss": 2.2343,
      "step": 50171
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.6763548254966736,
      "learning_rate": 0.0005324953766626108,
      "loss": 2.1635,
      "step": 50172
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7001038789749146,
      "learning_rate": 0.0005324927914869778,
      "loss": 2.1068,
      "step": 50173
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7207036018371582,
      "learning_rate": 0.0005324902062681199,
      "loss": 2.3611,
      "step": 50174
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.6977325081825256,
      "learning_rate": 0.0005324876210060376,
      "loss": 2.135,
      "step": 50175
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.6916638612747192,
      "learning_rate": 0.0005324850357007316,
      "loss": 2.4201,
      "step": 50176
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.8841729164123535,
      "learning_rate": 0.0005324824503522022,
      "loss": 2.2922,
      "step": 50177
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.745785653591156,
      "learning_rate": 0.0005324798649604499,
      "loss": 2.2555,
      "step": 50178
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.8285210132598877,
      "learning_rate": 0.0005324772795254752,
      "loss": 2.198,
      "step": 50179
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.6770704388618469,
      "learning_rate": 0.0005324746940472786,
      "loss": 2.371,
      "step": 50180
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7330294251441956,
      "learning_rate": 0.0005324721085258606,
      "loss": 2.2542,
      "step": 50181
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7085524797439575,
      "learning_rate": 0.0005324695229612217,
      "loss": 2.3418,
      "step": 50182
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.6707407236099243,
      "learning_rate": 0.0005324669373533621,
      "loss": 1.9633,
      "step": 50183
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.6961963176727295,
      "learning_rate": 0.0005324643517022827,
      "loss": 2.2473,
      "step": 50184
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7949087619781494,
      "learning_rate": 0.0005324617660079837,
      "loss": 2.5535,
      "step": 50185
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7902485132217407,
      "learning_rate": 0.0005324591802704658,
      "loss": 2.3503,
      "step": 50186
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.6900676488876343,
      "learning_rate": 0.0005324565944897292,
      "loss": 2.2083,
      "step": 50187
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7354286909103394,
      "learning_rate": 0.0005324540086657746,
      "loss": 2.419,
      "step": 50188
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7788710594177246,
      "learning_rate": 0.0005324514227986024,
      "loss": 2.1675,
      "step": 50189
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7213917970657349,
      "learning_rate": 0.0005324488368882131,
      "loss": 2.3549,
      "step": 50190
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.6843517422676086,
      "learning_rate": 0.0005324462509346072,
      "loss": 2.3573,
      "step": 50191
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.8201707005500793,
      "learning_rate": 0.000532443664937785,
      "loss": 2.1847,
      "step": 50192
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7256451845169067,
      "learning_rate": 0.0005324410788977473,
      "loss": 2.2885,
      "step": 50193
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7041860222816467,
      "learning_rate": 0.0005324384928144945,
      "loss": 2.2883,
      "step": 50194
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.830845296382904,
      "learning_rate": 0.0005324359066880268,
      "loss": 2.0336,
      "step": 50195
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7917993664741516,
      "learning_rate": 0.000532433320518345,
      "loss": 2.2876,
      "step": 50196
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7940319180488586,
      "learning_rate": 0.0005324307343054494,
      "loss": 2.2056,
      "step": 50197
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.8807125091552734,
      "learning_rate": 0.0005324281480493405,
      "loss": 2.1259,
      "step": 50198
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.8854358196258545,
      "learning_rate": 0.0005324255617500189,
      "loss": 2.1113,
      "step": 50199
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.8985641002655029,
      "learning_rate": 0.0005324229754074849,
      "loss": 2.1258,
      "step": 50200
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7532393336296082,
      "learning_rate": 0.0005324203890217391,
      "loss": 2.318,
      "step": 50201
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7751653790473938,
      "learning_rate": 0.000532417802592782,
      "loss": 2.0628,
      "step": 50202
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.77024906873703,
      "learning_rate": 0.0005324152161206141,
      "loss": 2.4059,
      "step": 50203
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7031695246696472,
      "learning_rate": 0.0005324126296052358,
      "loss": 2.1308,
      "step": 50204
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.776424765586853,
      "learning_rate": 0.0005324100430466476,
      "loss": 2.3432,
      "step": 50205
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.6892856955528259,
      "learning_rate": 0.00053240745644485,
      "loss": 2.3504,
      "step": 50206
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.6646316051483154,
      "learning_rate": 0.0005324048697998434,
      "loss": 2.0605,
      "step": 50207
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7387459874153137,
      "learning_rate": 0.0005324022831116284,
      "loss": 2.2862,
      "step": 50208
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7328076958656311,
      "learning_rate": 0.0005323996963802054,
      "loss": 2.2158,
      "step": 50209
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7553333044052124,
      "learning_rate": 0.0005323971096055749,
      "loss": 2.3781,
      "step": 50210
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.702072262763977,
      "learning_rate": 0.0005323945227877374,
      "loss": 2.1936,
      "step": 50211
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7327380776405334,
      "learning_rate": 0.0005323919359266934,
      "loss": 2.1872,
      "step": 50212
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7613750100135803,
      "learning_rate": 0.0005323893490224433,
      "loss": 2.2909,
      "step": 50213
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7421261668205261,
      "learning_rate": 0.0005323867620749877,
      "loss": 2.2396,
      "step": 50214
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7521968483924866,
      "learning_rate": 0.000532384175084327,
      "loss": 2.588,
      "step": 50215
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7110356688499451,
      "learning_rate": 0.0005323815880504617,
      "loss": 2.2953,
      "step": 50216
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7291886806488037,
      "learning_rate": 0.0005323790009733922,
      "loss": 2.3799,
      "step": 50217
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7745268940925598,
      "learning_rate": 0.0005323764138531192,
      "loss": 2.354,
      "step": 50218
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.784135639667511,
      "learning_rate": 0.0005323738266896429,
      "loss": 2.3207,
      "step": 50219
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.649480938911438,
      "learning_rate": 0.000532371239482964,
      "loss": 2.0038,
      "step": 50220
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7708218097686768,
      "learning_rate": 0.0005323686522330828,
      "loss": 2.2384,
      "step": 50221
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7301837801933289,
      "learning_rate": 0.0005323660649399999,
      "loss": 1.97,
      "step": 50222
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7440304756164551,
      "learning_rate": 0.0005323634776037159,
      "loss": 2.2869,
      "step": 50223
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7769170999526978,
      "learning_rate": 0.000532360890224231,
      "loss": 2.4027,
      "step": 50224
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.8381377458572388,
      "learning_rate": 0.0005323583028015459,
      "loss": 2.2687,
      "step": 50225
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7378638386726379,
      "learning_rate": 0.000532355715335661,
      "loss": 2.4982,
      "step": 50226
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7575340270996094,
      "learning_rate": 0.0005323531278265767,
      "loss": 2.2813,
      "step": 50227
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.8450540900230408,
      "learning_rate": 0.0005323505402742937,
      "loss": 2.3419,
      "step": 50228
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.6663849949836731,
      "learning_rate": 0.0005323479526788123,
      "loss": 2.2247,
      "step": 50229
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7166910171508789,
      "learning_rate": 0.000532345365040133,
      "loss": 2.1696,
      "step": 50230
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7048253417015076,
      "learning_rate": 0.0005323427773582563,
      "loss": 2.345,
      "step": 50231
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.70987868309021,
      "learning_rate": 0.0005323401896331827,
      "loss": 2.3414,
      "step": 50232
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7687862515449524,
      "learning_rate": 0.0005323376018649127,
      "loss": 2.1115,
      "step": 50233
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.8172751665115356,
      "learning_rate": 0.0005323350140534468,
      "loss": 2.2728,
      "step": 50234
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7456015348434448,
      "learning_rate": 0.0005323324261987854,
      "loss": 2.2614,
      "step": 50235
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.6647520661354065,
      "learning_rate": 0.0005323298383009289,
      "loss": 2.2371,
      "step": 50236
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7028447389602661,
      "learning_rate": 0.000532327250359878,
      "loss": 2.2799,
      "step": 50237
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7166346311569214,
      "learning_rate": 0.0005323246623756331,
      "loss": 2.3701,
      "step": 50238
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.6627569794654846,
      "learning_rate": 0.0005323220743481947,
      "loss": 2.1096,
      "step": 50239
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7248887419700623,
      "learning_rate": 0.0005323194862775632,
      "loss": 2.379,
      "step": 50240
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.6547417044639587,
      "learning_rate": 0.000532316898163739,
      "loss": 2.2452,
      "step": 50241
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.864081621170044,
      "learning_rate": 0.0005323143100067229,
      "loss": 2.1906,
      "step": 50242
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7129330039024353,
      "learning_rate": 0.000532311721806515,
      "loss": 2.3634,
      "step": 50243
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.70456862449646,
      "learning_rate": 0.0005323091335631161,
      "loss": 2.2925,
      "step": 50244
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7630385756492615,
      "learning_rate": 0.0005323065452765265,
      "loss": 2.2602,
      "step": 50245
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7466126084327698,
      "learning_rate": 0.0005323039569467467,
      "loss": 2.5197,
      "step": 50246
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7399722337722778,
      "learning_rate": 0.0005323013685737773,
      "loss": 2.2291,
      "step": 50247
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.6955478191375732,
      "learning_rate": 0.0005322987801576186,
      "loss": 2.4171,
      "step": 50248
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.6959433555603027,
      "learning_rate": 0.0005322961916982712,
      "loss": 2.2293,
      "step": 50249
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7598193883895874,
      "learning_rate": 0.0005322936031957355,
      "loss": 2.0864,
      "step": 50250
    },
    {
      "epoch": 0.65,
      "grad_norm": 2.856221914291382,
      "learning_rate": 0.0005322910146500121,
      "loss": 2.1955,
      "step": 50251
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.6937945485115051,
      "learning_rate": 0.0005322884260611014,
      "loss": 2.2714,
      "step": 50252
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7495242953300476,
      "learning_rate": 0.0005322858374290038,
      "loss": 2.1754,
      "step": 50253
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7865267395973206,
      "learning_rate": 0.0005322832487537201,
      "loss": 2.3313,
      "step": 50254
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7367841005325317,
      "learning_rate": 0.0005322806600352504,
      "loss": 2.0813,
      "step": 50255
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7011784911155701,
      "learning_rate": 0.0005322780712735954,
      "loss": 2.0723,
      "step": 50256
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.8642390370368958,
      "learning_rate": 0.0005322754824687555,
      "loss": 2.2543,
      "step": 50257
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.791955292224884,
      "learning_rate": 0.0005322728936207312,
      "loss": 2.2942,
      "step": 50258
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.6969079375267029,
      "learning_rate": 0.0005322703047295229,
      "loss": 2.2157,
      "step": 50259
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7192642688751221,
      "learning_rate": 0.0005322677157951313,
      "loss": 2.0358,
      "step": 50260
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7277951836585999,
      "learning_rate": 0.0005322651268175568,
      "loss": 2.1824,
      "step": 50261
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7594694495201111,
      "learning_rate": 0.0005322625377967996,
      "loss": 2.2112,
      "step": 50262
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7093203067779541,
      "learning_rate": 0.0005322599487328606,
      "loss": 2.2159,
      "step": 50263
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.6909090876579285,
      "learning_rate": 0.0005322573596257401,
      "loss": 2.3718,
      "step": 50264
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.6941767334938049,
      "learning_rate": 0.0005322547704754385,
      "loss": 1.9887,
      "step": 50265
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.795078694820404,
      "learning_rate": 0.0005322521812819562,
      "loss": 2.3266,
      "step": 50266
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.6966637969017029,
      "learning_rate": 0.000532249592045294,
      "loss": 2.124,
      "step": 50267
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.6976051926612854,
      "learning_rate": 0.0005322470027654523,
      "loss": 2.2248,
      "step": 50268
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7558951377868652,
      "learning_rate": 0.0005322444134424313,
      "loss": 2.4548,
      "step": 50269
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7019570469856262,
      "learning_rate": 0.0005322418240762319,
      "loss": 2.1974,
      "step": 50270
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7468411922454834,
      "learning_rate": 0.0005322392346668542,
      "loss": 2.2512,
      "step": 50271
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7047860622406006,
      "learning_rate": 0.0005322366452142988,
      "loss": 2.3584,
      "step": 50272
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.8837425708770752,
      "learning_rate": 0.0005322340557185664,
      "loss": 2.2672,
      "step": 50273
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7940928936004639,
      "learning_rate": 0.0005322314661796572,
      "loss": 2.3047,
      "step": 50274
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.8320735096931458,
      "learning_rate": 0.0005322288765975719,
      "loss": 2.149,
      "step": 50275
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.6934229731559753,
      "learning_rate": 0.0005322262869723107,
      "loss": 2.3514,
      "step": 50276
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7181086540222168,
      "learning_rate": 0.0005322236973038743,
      "loss": 2.5535,
      "step": 50277
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7197006940841675,
      "learning_rate": 0.0005322211075922632,
      "loss": 2.4234,
      "step": 50278
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.6467331647872925,
      "learning_rate": 0.0005322185178374778,
      "loss": 2.1399,
      "step": 50279
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.8093934655189514,
      "learning_rate": 0.0005322159280395184,
      "loss": 2.173,
      "step": 50280
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7916478514671326,
      "learning_rate": 0.000532213338198386,
      "loss": 2.2462,
      "step": 50281
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7770528793334961,
      "learning_rate": 0.0005322107483140805,
      "loss": 2.4799,
      "step": 50282
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7019871473312378,
      "learning_rate": 0.0005322081583866028,
      "loss": 2.2692,
      "step": 50283
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7843985557556152,
      "learning_rate": 0.0005322055684159531,
      "loss": 2.4406,
      "step": 50284
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.8965232968330383,
      "learning_rate": 0.0005322029784021321,
      "loss": 2.1465,
      "step": 50285
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7027766704559326,
      "learning_rate": 0.0005322003883451402,
      "loss": 2.2958,
      "step": 50286
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.6613538265228271,
      "learning_rate": 0.0005321977982449777,
      "loss": 1.9749,
      "step": 50287
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.0121628046035767,
      "learning_rate": 0.0005321952081016454,
      "loss": 2.2185,
      "step": 50288
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7209998965263367,
      "learning_rate": 0.0005321926179151434,
      "loss": 2.1689,
      "step": 50289
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7713990211486816,
      "learning_rate": 0.0005321900276854727,
      "loss": 2.3381,
      "step": 50290
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.8944351077079773,
      "learning_rate": 0.0005321874374126334,
      "loss": 2.1763,
      "step": 50291
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.6957794427871704,
      "learning_rate": 0.000532184847096626,
      "loss": 2.507,
      "step": 50292
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.746773362159729,
      "learning_rate": 0.0005321822567374511,
      "loss": 2.3364,
      "step": 50293
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7400039434432983,
      "learning_rate": 0.0005321796663351092,
      "loss": 2.24,
      "step": 50294
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7057297229766846,
      "learning_rate": 0.0005321770758896007,
      "loss": 2.0548,
      "step": 50295
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.6822616457939148,
      "learning_rate": 0.000532174485400926,
      "loss": 2.1433,
      "step": 50296
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7360172271728516,
      "learning_rate": 0.0005321718948690856,
      "loss": 2.3193,
      "step": 50297
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.8744921088218689,
      "learning_rate": 0.0005321693042940803,
      "loss": 2.1405,
      "step": 50298
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.6957650780677795,
      "learning_rate": 0.0005321667136759102,
      "loss": 2.2469,
      "step": 50299
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.6623217463493347,
      "learning_rate": 0.000532164123014576,
      "loss": 2.3983,
      "step": 50300
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.770517885684967,
      "learning_rate": 0.0005321615323100781,
      "loss": 2.2148,
      "step": 50301
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7254350781440735,
      "learning_rate": 0.0005321589415624169,
      "loss": 2.0938,
      "step": 50302
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7528771758079529,
      "learning_rate": 0.0005321563507715931,
      "loss": 2.2474,
      "step": 50303
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.6612914800643921,
      "learning_rate": 0.0005321537599376069,
      "loss": 2.3773,
      "step": 50304
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7167834639549255,
      "learning_rate": 0.0005321511690604591,
      "loss": 2.4073,
      "step": 50305
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.6554166078567505,
      "learning_rate": 0.0005321485781401499,
      "loss": 2.2906,
      "step": 50306
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.6779868006706238,
      "learning_rate": 0.00053214598717668,
      "loss": 2.1916,
      "step": 50307
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7643684148788452,
      "learning_rate": 0.0005321433961700497,
      "loss": 2.2603,
      "step": 50308
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.6829580068588257,
      "learning_rate": 0.0005321408051202595,
      "loss": 2.4508,
      "step": 50309
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.6931872963905334,
      "learning_rate": 0.00053213821402731,
      "loss": 2.2934,
      "step": 50310
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.748013436794281,
      "learning_rate": 0.0005321356228912017,
      "loss": 2.3787,
      "step": 50311
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.6668527722358704,
      "learning_rate": 0.0005321330317119349,
      "loss": 2.1709,
      "step": 50312
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7152408957481384,
      "learning_rate": 0.0005321304404895102,
      "loss": 2.249,
      "step": 50313
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.6519579291343689,
      "learning_rate": 0.0005321278492239281,
      "loss": 2.3113,
      "step": 50314
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7419150471687317,
      "learning_rate": 0.000532125257915189,
      "loss": 2.1452,
      "step": 50315
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.6971694231033325,
      "learning_rate": 0.0005321226665632936,
      "loss": 2.3047,
      "step": 50316
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.6815411448478699,
      "learning_rate": 0.000532120075168242,
      "loss": 2.1753,
      "step": 50317
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.8131624460220337,
      "learning_rate": 0.000532117483730035,
      "loss": 2.1388,
      "step": 50318
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7288779616355896,
      "learning_rate": 0.000532114892248673,
      "loss": 2.3168,
      "step": 50319
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.8333279490470886,
      "learning_rate": 0.0005321123007241565,
      "loss": 2.2532,
      "step": 50320
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7196621298789978,
      "learning_rate": 0.0005321097091564858,
      "loss": 2.1065,
      "step": 50321
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.8313068151473999,
      "learning_rate": 0.0005321071175456617,
      "loss": 2.5213,
      "step": 50322
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7620701193809509,
      "learning_rate": 0.0005321045258916843,
      "loss": 2.2511,
      "step": 50323
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7770750522613525,
      "learning_rate": 0.0005321019341945545,
      "loss": 2.1761,
      "step": 50324
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.745367705821991,
      "learning_rate": 0.0005320993424542724,
      "loss": 2.2304,
      "step": 50325
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7385674715042114,
      "learning_rate": 0.0005320967506708387,
      "loss": 2.4357,
      "step": 50326
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.8074543476104736,
      "learning_rate": 0.0005320941588442539,
      "loss": 2.0765,
      "step": 50327
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.9632113575935364,
      "learning_rate": 0.0005320915669745183,
      "loss": 2.2198,
      "step": 50328
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.8022109270095825,
      "learning_rate": 0.0005320889750616326,
      "loss": 2.2329,
      "step": 50329
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.6954237818717957,
      "learning_rate": 0.0005320863831055971,
      "loss": 2.4625,
      "step": 50330
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.8147209882736206,
      "learning_rate": 0.0005320837911064124,
      "loss": 2.2956,
      "step": 50331
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7550657987594604,
      "learning_rate": 0.0005320811990640789,
      "loss": 2.0213,
      "step": 50332
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7418807148933411,
      "learning_rate": 0.0005320786069785972,
      "loss": 2.1987,
      "step": 50333
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.739299476146698,
      "learning_rate": 0.0005320760148499676,
      "loss": 2.245,
      "step": 50334
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7677955031394958,
      "learning_rate": 0.0005320734226781908,
      "loss": 2.3699,
      "step": 50335
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7800153493881226,
      "learning_rate": 0.0005320708304632671,
      "loss": 2.3886,
      "step": 50336
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.8496801853179932,
      "learning_rate": 0.0005320682382051971,
      "loss": 2.3595,
      "step": 50337
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7189459800720215,
      "learning_rate": 0.0005320656459039812,
      "loss": 2.4156,
      "step": 50338
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.8425399661064148,
      "learning_rate": 0.0005320630535596199,
      "loss": 2.3499,
      "step": 50339
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.9546847939491272,
      "learning_rate": 0.0005320604611721138,
      "loss": 2.5733,
      "step": 50340
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7923802733421326,
      "learning_rate": 0.0005320578687414632,
      "loss": 2.3409,
      "step": 50341
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7743640542030334,
      "learning_rate": 0.0005320552762676686,
      "loss": 2.3833,
      "step": 50342
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7059372663497925,
      "learning_rate": 0.0005320526837507306,
      "loss": 2.0976,
      "step": 50343
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7191069722175598,
      "learning_rate": 0.0005320500911906497,
      "loss": 2.2844,
      "step": 50344
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7297969460487366,
      "learning_rate": 0.0005320474985874261,
      "loss": 2.3715,
      "step": 50345
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.8883681297302246,
      "learning_rate": 0.0005320449059410607,
      "loss": 2.3053,
      "step": 50346
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.6564124822616577,
      "learning_rate": 0.0005320423132515538,
      "loss": 2.1404,
      "step": 50347
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7178651690483093,
      "learning_rate": 0.0005320397205189057,
      "loss": 2.2786,
      "step": 50348
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.663234293460846,
      "learning_rate": 0.0005320371277431172,
      "loss": 2.176,
      "step": 50349
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.6488816142082214,
      "learning_rate": 0.0005320345349241885,
      "loss": 2.0627,
      "step": 50350
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.6936500072479248,
      "learning_rate": 0.0005320319420621201,
      "loss": 2.137,
      "step": 50351
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7220854163169861,
      "learning_rate": 0.0005320293491569127,
      "loss": 2.3392,
      "step": 50352
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.6893324255943298,
      "learning_rate": 0.0005320267562085667,
      "loss": 2.0208,
      "step": 50353
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7344450950622559,
      "learning_rate": 0.0005320241632170825,
      "loss": 2.2998,
      "step": 50354
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.6881174445152283,
      "learning_rate": 0.0005320215701824606,
      "loss": 2.3712,
      "step": 50355
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7096810340881348,
      "learning_rate": 0.0005320189771047016,
      "loss": 2.1647,
      "step": 50356
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.6708977222442627,
      "learning_rate": 0.0005320163839838058,
      "loss": 2.3847,
      "step": 50357
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7577720880508423,
      "learning_rate": 0.0005320137908197738,
      "loss": 2.3291,
      "step": 50358
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7422733306884766,
      "learning_rate": 0.0005320111976126061,
      "loss": 2.1777,
      "step": 50359
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7305742502212524,
      "learning_rate": 0.0005320086043623031,
      "loss": 2.2221,
      "step": 50360
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7594938278198242,
      "learning_rate": 0.0005320060110688653,
      "loss": 2.4663,
      "step": 50361
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7456485033035278,
      "learning_rate": 0.0005320034177322933,
      "loss": 2.2024,
      "step": 50362
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.6954340934753418,
      "learning_rate": 0.0005320008243525874,
      "loss": 2.2842,
      "step": 50363
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7110514640808105,
      "learning_rate": 0.0005319982309297482,
      "loss": 2.2968,
      "step": 50364
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7499464154243469,
      "learning_rate": 0.0005319956374637762,
      "loss": 2.4837,
      "step": 50365
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.6890694499015808,
      "learning_rate": 0.0005319930439546716,
      "loss": 2.2364,
      "step": 50366
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7418351173400879,
      "learning_rate": 0.0005319904504024354,
      "loss": 2.1396,
      "step": 50367
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.6484164595603943,
      "learning_rate": 0.0005319878568070676,
      "loss": 2.0113,
      "step": 50368
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7178016901016235,
      "learning_rate": 0.000531985263168569,
      "loss": 2.347,
      "step": 50369
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7009301781654358,
      "learning_rate": 0.0005319826694869399,
      "loss": 2.304,
      "step": 50370
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.6915711164474487,
      "learning_rate": 0.0005319800757621807,
      "loss": 2.5513,
      "step": 50371
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.6808086633682251,
      "learning_rate": 0.0005319774819942923,
      "loss": 2.0587,
      "step": 50372
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7107500433921814,
      "learning_rate": 0.0005319748881832747,
      "loss": 2.4546,
      "step": 50373
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.8046197295188904,
      "learning_rate": 0.0005319722943291287,
      "loss": 2.1987,
      "step": 50374
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.644724428653717,
      "learning_rate": 0.0005319697004318547,
      "loss": 2.1717,
      "step": 50375
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7082769274711609,
      "learning_rate": 0.0005319671064914532,
      "loss": 2.249,
      "step": 50376
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7501966953277588,
      "learning_rate": 0.0005319645125079245,
      "loss": 2.4157,
      "step": 50377
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.770959734916687,
      "learning_rate": 0.0005319619184812692,
      "loss": 2.3802,
      "step": 50378
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.6676561236381531,
      "learning_rate": 0.0005319593244114878,
      "loss": 2.2007,
      "step": 50379
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7275705933570862,
      "learning_rate": 0.0005319567302985809,
      "loss": 2.3095,
      "step": 50380
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7057732343673706,
      "learning_rate": 0.0005319541361425488,
      "loss": 2.4955,
      "step": 50381
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7319905161857605,
      "learning_rate": 0.0005319515419433921,
      "loss": 2.4594,
      "step": 50382
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7072616815567017,
      "learning_rate": 0.0005319489477011111,
      "loss": 2.4157,
      "step": 50383
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7502654790878296,
      "learning_rate": 0.0005319463534157066,
      "loss": 2.1551,
      "step": 50384
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.6173301935195923,
      "learning_rate": 0.0005319437590871787,
      "loss": 2.2861,
      "step": 50385
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.6893715262413025,
      "learning_rate": 0.0005319411647155282,
      "loss": 2.3391,
      "step": 50386
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7652026414871216,
      "learning_rate": 0.0005319385703007554,
      "loss": 2.4319,
      "step": 50387
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.6746026277542114,
      "learning_rate": 0.0005319359758428609,
      "loss": 2.0509,
      "step": 50388
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.6741387844085693,
      "learning_rate": 0.0005319333813418451,
      "loss": 2.2225,
      "step": 50389
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7188100814819336,
      "learning_rate": 0.0005319307867977085,
      "loss": 2.3237,
      "step": 50390
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7984305024147034,
      "learning_rate": 0.0005319281922104516,
      "loss": 2.2214,
      "step": 50391
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.6958145499229431,
      "learning_rate": 0.0005319255975800748,
      "loss": 2.123,
      "step": 50392
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7103517055511475,
      "learning_rate": 0.0005319230029065787,
      "loss": 2.3625,
      "step": 50393
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7277199625968933,
      "learning_rate": 0.0005319204081899638,
      "loss": 2.1804,
      "step": 50394
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7167912125587463,
      "learning_rate": 0.0005319178134302304,
      "loss": 2.3288,
      "step": 50395
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.773231565952301,
      "learning_rate": 0.0005319152186273792,
      "loss": 2.4341,
      "step": 50396
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7132012844085693,
      "learning_rate": 0.0005319126237814106,
      "loss": 2.1911,
      "step": 50397
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7972924709320068,
      "learning_rate": 0.000531910028892325,
      "loss": 2.2344,
      "step": 50398
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7567925453186035,
      "learning_rate": 0.0005319074339601229,
      "loss": 2.1809,
      "step": 50399
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.6598417162895203,
      "learning_rate": 0.000531904838984805,
      "loss": 2.3033,
      "step": 50400
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7837020754814148,
      "learning_rate": 0.0005319022439663714,
      "loss": 2.2849,
      "step": 50401
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7514235973358154,
      "learning_rate": 0.000531899648904823,
      "loss": 2.4591,
      "step": 50402
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.6797884106636047,
      "learning_rate": 0.0005318970538001599,
      "loss": 2.2592,
      "step": 50403
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7939785122871399,
      "learning_rate": 0.0005318944586523829,
      "loss": 2.3563,
      "step": 50404
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.8389747142791748,
      "learning_rate": 0.0005318918634614922,
      "loss": 2.19,
      "step": 50405
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.6706749796867371,
      "learning_rate": 0.0005318892682274885,
      "loss": 2.2529,
      "step": 50406
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7633470296859741,
      "learning_rate": 0.0005318866729503723,
      "loss": 2.1539,
      "step": 50407
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7131566405296326,
      "learning_rate": 0.0005318840776301439,
      "loss": 2.1369,
      "step": 50408
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.6968063712120056,
      "learning_rate": 0.0005318814822668039,
      "loss": 2.0919,
      "step": 50409
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.731951892375946,
      "learning_rate": 0.0005318788868603527,
      "loss": 2.2038,
      "step": 50410
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.8574735522270203,
      "learning_rate": 0.0005318762914107908,
      "loss": 2.2841,
      "step": 50411
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7472202181816101,
      "learning_rate": 0.0005318736959181188,
      "loss": 2.4431,
      "step": 50412
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7123321294784546,
      "learning_rate": 0.0005318711003823372,
      "loss": 2.2938,
      "step": 50413
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.787904679775238,
      "learning_rate": 0.0005318685048034462,
      "loss": 2.4423,
      "step": 50414
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.843063473701477,
      "learning_rate": 0.0005318659091814466,
      "loss": 2.1747,
      "step": 50415
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7038924098014832,
      "learning_rate": 0.0005318633135163387,
      "loss": 2.1649,
      "step": 50416
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7869346737861633,
      "learning_rate": 0.000531860717808123,
      "loss": 2.3649,
      "step": 50417
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.788887083530426,
      "learning_rate": 0.0005318581220568001,
      "loss": 2.1869,
      "step": 50418
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.6668612360954285,
      "learning_rate": 0.0005318555262623703,
      "loss": 2.0901,
      "step": 50419
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.6804497838020325,
      "learning_rate": 0.0005318529304248343,
      "loss": 2.4081,
      "step": 50420
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.8726085424423218,
      "learning_rate": 0.0005318503345441923,
      "loss": 2.4637,
      "step": 50421
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7127935290336609,
      "learning_rate": 0.0005318477386204451,
      "loss": 2.0812,
      "step": 50422
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.710821807384491,
      "learning_rate": 0.000531845142653593,
      "loss": 2.1036,
      "step": 50423
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.6988089680671692,
      "learning_rate": 0.0005318425466436363,
      "loss": 2.1702,
      "step": 50424
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.8278383016586304,
      "learning_rate": 0.000531839950590576,
      "loss": 2.5214,
      "step": 50425
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7074447274208069,
      "learning_rate": 0.0005318373544944122,
      "loss": 2.3829,
      "step": 50426
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7182429432868958,
      "learning_rate": 0.0005318347583551453,
      "loss": 2.1831,
      "step": 50427
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7723375558853149,
      "learning_rate": 0.0005318321621727761,
      "loss": 2.2808,
      "step": 50428
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7740641832351685,
      "learning_rate": 0.0005318295659473049,
      "loss": 2.0223,
      "step": 50429
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7510718703269958,
      "learning_rate": 0.0005318269696787322,
      "loss": 2.3032,
      "step": 50430
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7872682213783264,
      "learning_rate": 0.0005318243733670585,
      "loss": 2.2869,
      "step": 50431
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7508731484413147,
      "learning_rate": 0.0005318217770122841,
      "loss": 2.3195,
      "step": 50432
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.6821520924568176,
      "learning_rate": 0.0005318191806144098,
      "loss": 2.2228,
      "step": 50433
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.6756891012191772,
      "learning_rate": 0.000531816584173436,
      "loss": 2.4504,
      "step": 50434
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7915648221969604,
      "learning_rate": 0.000531813987689363,
      "loss": 2.2025,
      "step": 50435
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7371705770492554,
      "learning_rate": 0.0005318113911621915,
      "loss": 1.9049,
      "step": 50436
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7413057684898376,
      "learning_rate": 0.0005318087945919219,
      "loss": 2.3192,
      "step": 50437
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7167491316795349,
      "learning_rate": 0.0005318061979785547,
      "loss": 2.2613,
      "step": 50438
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7479979395866394,
      "learning_rate": 0.0005318036013220902,
      "loss": 2.6109,
      "step": 50439
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7343336343765259,
      "learning_rate": 0.0005318010046225291,
      "loss": 2.5235,
      "step": 50440
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7782163023948669,
      "learning_rate": 0.0005317984078798719,
      "loss": 2.3699,
      "step": 50441
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7512567043304443,
      "learning_rate": 0.0005317958110941189,
      "loss": 2.1823,
      "step": 50442
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7559558153152466,
      "learning_rate": 0.0005317932142652707,
      "loss": 2.1244,
      "step": 50443
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7801013588905334,
      "learning_rate": 0.0005317906173933277,
      "loss": 2.3409,
      "step": 50444
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.751945436000824,
      "learning_rate": 0.0005317880204782906,
      "loss": 2.2847,
      "step": 50445
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7970023155212402,
      "learning_rate": 0.0005317854235201596,
      "loss": 2.051,
      "step": 50446
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.8097713589668274,
      "learning_rate": 0.0005317828265189355,
      "loss": 2.3534,
      "step": 50447
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7692392468452454,
      "learning_rate": 0.0005317802294746184,
      "loss": 2.3737,
      "step": 50448
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.798323392868042,
      "learning_rate": 0.000531777632387209,
      "loss": 2.1421,
      "step": 50449
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7281590700149536,
      "learning_rate": 0.0005317750352567078,
      "loss": 2.2854,
      "step": 50450
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7602342963218689,
      "learning_rate": 0.0005317724380831154,
      "loss": 2.4365,
      "step": 50451
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7595137357711792,
      "learning_rate": 0.0005317698408664319,
      "loss": 2.3321,
      "step": 50452
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7275425791740417,
      "learning_rate": 0.0005317672436066581,
      "loss": 2.187,
      "step": 50453
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.6840667724609375,
      "learning_rate": 0.0005317646463037943,
      "loss": 2.1004,
      "step": 50454
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.6722426414489746,
      "learning_rate": 0.0005317620489578412,
      "loss": 2.2519,
      "step": 50455
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.8004853129386902,
      "learning_rate": 0.0005317594515687992,
      "loss": 2.2149,
      "step": 50456
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7649012804031372,
      "learning_rate": 0.0005317568541366686,
      "loss": 2.2916,
      "step": 50457
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.6552528142929077,
      "learning_rate": 0.0005317542566614502,
      "loss": 2.0856,
      "step": 50458
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.705046534538269,
      "learning_rate": 0.0005317516591431442,
      "loss": 2.1646,
      "step": 50459
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.63695228099823,
      "learning_rate": 0.0005317490615817512,
      "loss": 2.2444,
      "step": 50460
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.6726247668266296,
      "learning_rate": 0.0005317464639772717,
      "loss": 2.2993,
      "step": 50461
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7078790068626404,
      "learning_rate": 0.0005317438663297061,
      "loss": 2.1926,
      "step": 50462
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.8022422790527344,
      "learning_rate": 0.000531741268639055,
      "loss": 2.56,
      "step": 50463
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.8165222406387329,
      "learning_rate": 0.0005317386709053187,
      "loss": 2.3802,
      "step": 50464
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7754573225975037,
      "learning_rate": 0.0005317360731284979,
      "loss": 2.3102,
      "step": 50465
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7577922344207764,
      "learning_rate": 0.0005317334753085931,
      "loss": 2.3311,
      "step": 50466
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7347902059555054,
      "learning_rate": 0.0005317308774456046,
      "loss": 2.3641,
      "step": 50467
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.767258882522583,
      "learning_rate": 0.0005317282795395328,
      "loss": 2.2436,
      "step": 50468
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.6515543460845947,
      "learning_rate": 0.0005317256815903785,
      "loss": 2.1526,
      "step": 50469
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.698131799697876,
      "learning_rate": 0.000531723083598142,
      "loss": 2.3097,
      "step": 50470
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.8177513480186462,
      "learning_rate": 0.0005317204855628237,
      "loss": 2.3112,
      "step": 50471
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.8326217532157898,
      "learning_rate": 0.0005317178874844243,
      "loss": 2.3178,
      "step": 50472
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.8468207716941833,
      "learning_rate": 0.0005317152893629442,
      "loss": 2.0539,
      "step": 50473
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.8971763253211975,
      "learning_rate": 0.0005317126911983837,
      "loss": 2.3406,
      "step": 50474
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7878299951553345,
      "learning_rate": 0.0005317100929907436,
      "loss": 2.2747,
      "step": 50475
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.9260982871055603,
      "learning_rate": 0.0005317074947400241,
      "loss": 2.3072,
      "step": 50476
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7192459106445312,
      "learning_rate": 0.000531704896446226,
      "loss": 2.2212,
      "step": 50477
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.6968393325805664,
      "learning_rate": 0.0005317022981093494,
      "loss": 2.2258,
      "step": 50478
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7467278838157654,
      "learning_rate": 0.0005316996997293949,
      "loss": 2.285,
      "step": 50479
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7925155162811279,
      "learning_rate": 0.0005316971013063632,
      "loss": 2.2699,
      "step": 50480
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.844142735004425,
      "learning_rate": 0.0005316945028402547,
      "loss": 2.1908,
      "step": 50481
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7149182558059692,
      "learning_rate": 0.0005316919043310696,
      "loss": 2.158,
      "step": 50482
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7116425037384033,
      "learning_rate": 0.0005316893057788088,
      "loss": 2.2174,
      "step": 50483
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.777737021446228,
      "learning_rate": 0.0005316867071834726,
      "loss": 2.1194,
      "step": 50484
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7201696038246155,
      "learning_rate": 0.0005316841085450612,
      "loss": 1.9876,
      "step": 50485
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.76276034116745,
      "learning_rate": 0.0005316815098635756,
      "loss": 2.2349,
      "step": 50486
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.6858976483345032,
      "learning_rate": 0.000531678911139016,
      "loss": 2.3162,
      "step": 50487
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7266268134117126,
      "learning_rate": 0.0005316763123713828,
      "loss": 2.3061,
      "step": 50488
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.6888114213943481,
      "learning_rate": 0.0005316737135606767,
      "loss": 2.0068,
      "step": 50489
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7484995722770691,
      "learning_rate": 0.0005316711147068981,
      "loss": 2.0589,
      "step": 50490
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.6718271374702454,
      "learning_rate": 0.0005316685158100475,
      "loss": 2.1335,
      "step": 50491
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7809686660766602,
      "learning_rate": 0.0005316659168701252,
      "loss": 2.3225,
      "step": 50492
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7290726900100708,
      "learning_rate": 0.0005316633178871319,
      "loss": 1.9244,
      "step": 50493
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7792540788650513,
      "learning_rate": 0.000531660718861068,
      "loss": 2.2652,
      "step": 50494
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.6612460613250732,
      "learning_rate": 0.0005316581197919339,
      "loss": 2.1738,
      "step": 50495
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7876139879226685,
      "learning_rate": 0.0005316555206797304,
      "loss": 2.1348,
      "step": 50496
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7088341116905212,
      "learning_rate": 0.0005316529215244577,
      "loss": 2.2121,
      "step": 50497
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7863563895225525,
      "learning_rate": 0.0005316503223261163,
      "loss": 2.3392,
      "step": 50498
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.712863564491272,
      "learning_rate": 0.0005316477230847068,
      "loss": 2.3263,
      "step": 50499
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7265253663063049,
      "learning_rate": 0.0005316451238002296,
      "loss": 2.2117,
      "step": 50500
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7530960440635681,
      "learning_rate": 0.0005316425244726851,
      "loss": 2.3166,
      "step": 50501
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7131032943725586,
      "learning_rate": 0.0005316399251020739,
      "loss": 1.9495,
      "step": 50502
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.6484360694885254,
      "learning_rate": 0.0005316373256883966,
      "loss": 2.1422,
      "step": 50503
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.80519038438797,
      "learning_rate": 0.0005316347262316535,
      "loss": 2.4732,
      "step": 50504
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7218341827392578,
      "learning_rate": 0.000531632126731845,
      "loss": 2.0772,
      "step": 50505
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7017869353294373,
      "learning_rate": 0.0005316295271889719,
      "loss": 2.1086,
      "step": 50506
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.6994026899337769,
      "learning_rate": 0.0005316269276030344,
      "loss": 2.1672,
      "step": 50507
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.8046655058860779,
      "learning_rate": 0.0005316243279740331,
      "loss": 2.1971,
      "step": 50508
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.6911423802375793,
      "learning_rate": 0.0005316217283019685,
      "loss": 2.4356,
      "step": 50509
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7187726497650146,
      "learning_rate": 0.0005316191285868409,
      "loss": 2.3951,
      "step": 50510
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7001978158950806,
      "learning_rate": 0.0005316165288286511,
      "loss": 2.2452,
      "step": 50511
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7108817100524902,
      "learning_rate": 0.0005316139290273993,
      "loss": 2.1543,
      "step": 50512
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7084252834320068,
      "learning_rate": 0.0005316113291830861,
      "loss": 2.1724,
      "step": 50513
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7314260601997375,
      "learning_rate": 0.0005316087292957121,
      "loss": 2.2905,
      "step": 50514
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.8209052681922913,
      "learning_rate": 0.0005316061293652776,
      "loss": 2.1878,
      "step": 50515
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.672464907169342,
      "learning_rate": 0.0005316035293917831,
      "loss": 2.2435,
      "step": 50516
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.6666325926780701,
      "learning_rate": 0.0005316009293752292,
      "loss": 2.3194,
      "step": 50517
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.768479585647583,
      "learning_rate": 0.0005315983293156163,
      "loss": 2.4802,
      "step": 50518
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.6975958347320557,
      "learning_rate": 0.000531595729212945,
      "loss": 2.1494,
      "step": 50519
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.751849889755249,
      "learning_rate": 0.0005315931290672154,
      "loss": 2.1616,
      "step": 50520
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7858521342277527,
      "learning_rate": 0.0005315905288784284,
      "loss": 2.164,
      "step": 50521
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7501381635665894,
      "learning_rate": 0.0005315879286465845,
      "loss": 2.328,
      "step": 50522
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.6400153636932373,
      "learning_rate": 0.0005315853283716839,
      "loss": 2.2935,
      "step": 50523
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7477025389671326,
      "learning_rate": 0.0005315827280537273,
      "loss": 2.2317,
      "step": 50524
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.8884940147399902,
      "learning_rate": 0.000531580127692715,
      "loss": 2.1878,
      "step": 50525
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.6929625868797302,
      "learning_rate": 0.0005315775272886476,
      "loss": 2.3315,
      "step": 50526
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.6654658317565918,
      "learning_rate": 0.0005315749268415257,
      "loss": 2.2477,
      "step": 50527
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.6839900612831116,
      "learning_rate": 0.0005315723263513495,
      "loss": 2.3169,
      "step": 50528
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.9385388493537903,
      "learning_rate": 0.0005315697258181197,
      "loss": 2.3307,
      "step": 50529
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7502402663230896,
      "learning_rate": 0.0005315671252418366,
      "loss": 2.5232,
      "step": 50530
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.8679714202880859,
      "learning_rate": 0.0005315645246225009,
      "loss": 2.2354,
      "step": 50531
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7519528865814209,
      "learning_rate": 0.000531561923960113,
      "loss": 2.0636,
      "step": 50532
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7933982610702515,
      "learning_rate": 0.0005315593232546733,
      "loss": 2.5038,
      "step": 50533
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.697913408279419,
      "learning_rate": 0.0005315567225061825,
      "loss": 2.3607,
      "step": 50534
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7152289748191833,
      "learning_rate": 0.0005315541217146408,
      "loss": 2.3015,
      "step": 50535
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7253512740135193,
      "learning_rate": 0.0005315515208800488,
      "loss": 2.2765,
      "step": 50536
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.8244321346282959,
      "learning_rate": 0.000531548920002407,
      "loss": 2.3829,
      "step": 50537
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7798596024513245,
      "learning_rate": 0.000531546319081716,
      "loss": 2.3269,
      "step": 50538
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7193201780319214,
      "learning_rate": 0.0005315437181179761,
      "loss": 2.5593,
      "step": 50539
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.8017902374267578,
      "learning_rate": 0.0005315411171111878,
      "loss": 2.2341,
      "step": 50540
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.9393950700759888,
      "learning_rate": 0.0005315385160613517,
      "loss": 2.2363,
      "step": 50541
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.8410625457763672,
      "learning_rate": 0.0005315359149684682,
      "loss": 2.2603,
      "step": 50542
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.6462620496749878,
      "learning_rate": 0.0005315333138325378,
      "loss": 2.1602,
      "step": 50543
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.0134053230285645,
      "learning_rate": 0.000531530712653561,
      "loss": 2.195,
      "step": 50544
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.8297907710075378,
      "learning_rate": 0.0005315281114315382,
      "loss": 2.3303,
      "step": 50545
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7753273844718933,
      "learning_rate": 0.0005315255101664701,
      "loss": 2.3978,
      "step": 50546
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7468100190162659,
      "learning_rate": 0.0005315229088583569,
      "loss": 2.3698,
      "step": 50547
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.8984541893005371,
      "learning_rate": 0.0005315203075071993,
      "loss": 2.4998,
      "step": 50548
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.9097252488136292,
      "learning_rate": 0.0005315177061129977,
      "loss": 2.0227,
      "step": 50549
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7354561686515808,
      "learning_rate": 0.0005315151046757525,
      "loss": 2.3909,
      "step": 50550
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.9291877746582031,
      "learning_rate": 0.0005315125031954644,
      "loss": 2.3376,
      "step": 50551
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.8783636093139648,
      "learning_rate": 0.0005315099016721338,
      "loss": 2.3021,
      "step": 50552
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7362754344940186,
      "learning_rate": 0.000531507300105761,
      "loss": 2.3258,
      "step": 50553
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7341834902763367,
      "learning_rate": 0.0005315046984963465,
      "loss": 2.3208,
      "step": 50554
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.850777268409729,
      "learning_rate": 0.0005315020968438911,
      "loss": 2.4039,
      "step": 50555
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7481169104576111,
      "learning_rate": 0.0005314994951483951,
      "loss": 2.5176,
      "step": 50556
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.6839353442192078,
      "learning_rate": 0.0005314968934098589,
      "loss": 2.4741,
      "step": 50557
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7311921119689941,
      "learning_rate": 0.0005314942916282831,
      "loss": 2.4928,
      "step": 50558
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7031522393226624,
      "learning_rate": 0.0005314916898036681,
      "loss": 2.19,
      "step": 50559
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.8010976910591125,
      "learning_rate": 0.0005314890879360145,
      "loss": 2.407,
      "step": 50560
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.6863464713096619,
      "learning_rate": 0.0005314864860253226,
      "loss": 2.1354,
      "step": 50561
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.697526752948761,
      "learning_rate": 0.0005314838840715931,
      "loss": 2.0506,
      "step": 50562
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7854878306388855,
      "learning_rate": 0.0005314812820748264,
      "loss": 2.259,
      "step": 50563
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7386664152145386,
      "learning_rate": 0.0005314786800350228,
      "loss": 2.3231,
      "step": 50564
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7225016355514526,
      "learning_rate": 0.000531476077952183,
      "loss": 2.0708,
      "step": 50565
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.6906348466873169,
      "learning_rate": 0.0005314734758263075,
      "loss": 2.1475,
      "step": 50566
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.9758308529853821,
      "learning_rate": 0.0005314708736573967,
      "loss": 2.2915,
      "step": 50567
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.9226836562156677,
      "learning_rate": 0.000531468271445451,
      "loss": 1.9846,
      "step": 50568
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.729292631149292,
      "learning_rate": 0.0005314656691904711,
      "loss": 2.12,
      "step": 50569
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.0543776750564575,
      "learning_rate": 0.0005314630668924573,
      "loss": 2.1977,
      "step": 50570
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.847255527973175,
      "learning_rate": 0.0005314604645514101,
      "loss": 2.5795,
      "step": 50571
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7615252733230591,
      "learning_rate": 0.0005314578621673301,
      "loss": 2.2439,
      "step": 50572
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.8168596029281616,
      "learning_rate": 0.0005314552597402177,
      "loss": 2.155,
      "step": 50573
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.8105850219726562,
      "learning_rate": 0.0005314526572700734,
      "loss": 2.2088,
      "step": 50574
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.8729566931724548,
      "learning_rate": 0.0005314500547568977,
      "loss": 2.2508,
      "step": 50575
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7102653980255127,
      "learning_rate": 0.0005314474522006912,
      "loss": 2.4011,
      "step": 50576
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.8240134716033936,
      "learning_rate": 0.0005314448496014539,
      "loss": 2.4691,
      "step": 50577
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7774341702461243,
      "learning_rate": 0.0005314422469591868,
      "loss": 2.351,
      "step": 50578
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7676535248756409,
      "learning_rate": 0.0005314396442738903,
      "loss": 2.3416,
      "step": 50579
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7419998049736023,
      "learning_rate": 0.0005314370415455648,
      "loss": 2.2443,
      "step": 50580
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.721566915512085,
      "learning_rate": 0.0005314344387742107,
      "loss": 2.3329,
      "step": 50581
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7375848293304443,
      "learning_rate": 0.0005314318359598287,
      "loss": 2.393,
      "step": 50582
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.701044499874115,
      "learning_rate": 0.000531429233102419,
      "loss": 2.1152,
      "step": 50583
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.8107144236564636,
      "learning_rate": 0.0005314266302019823,
      "loss": 2.2233,
      "step": 50584
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7835291028022766,
      "learning_rate": 0.000531424027258519,
      "loss": 2.3521,
      "step": 50585
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.6967742443084717,
      "learning_rate": 0.0005314214242720296,
      "loss": 2.2777,
      "step": 50586
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.6394939422607422,
      "learning_rate": 0.0005314188212425146,
      "loss": 2.3793,
      "step": 50587
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7484322786331177,
      "learning_rate": 0.0005314162181699745,
      "loss": 2.2305,
      "step": 50588
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.6735448241233826,
      "learning_rate": 0.0005314136150544097,
      "loss": 2.1713,
      "step": 50589
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7020280361175537,
      "learning_rate": 0.0005314110118958209,
      "loss": 2.1829,
      "step": 50590
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7444983720779419,
      "learning_rate": 0.0005314084086942083,
      "loss": 2.2262,
      "step": 50591
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7721891403198242,
      "learning_rate": 0.0005314058054495725,
      "loss": 2.2191,
      "step": 50592
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.6987614631652832,
      "learning_rate": 0.0005314032021619139,
      "loss": 2.0558,
      "step": 50593
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.6842352747917175,
      "learning_rate": 0.0005314005988312332,
      "loss": 2.0623,
      "step": 50594
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7486119866371155,
      "learning_rate": 0.0005313979954575307,
      "loss": 2.2281,
      "step": 50595
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.699263334274292,
      "learning_rate": 0.000531395392040807,
      "loss": 2.2842,
      "step": 50596
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7271674871444702,
      "learning_rate": 0.0005313927885810625,
      "loss": 2.2487,
      "step": 50597
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.6539902091026306,
      "learning_rate": 0.0005313901850782978,
      "loss": 2.4306,
      "step": 50598
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7199975848197937,
      "learning_rate": 0.0005313875815325133,
      "loss": 2.1302,
      "step": 50599
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7137717604637146,
      "learning_rate": 0.0005313849779437094,
      "loss": 2.327,
      "step": 50600
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7305557727813721,
      "learning_rate": 0.0005313823743118867,
      "loss": 2.3028,
      "step": 50601
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.6898693442344666,
      "learning_rate": 0.0005313797706370455,
      "loss": 2.0164,
      "step": 50602
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.8039661049842834,
      "learning_rate": 0.0005313771669191866,
      "loss": 2.1222,
      "step": 50603
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7076775431632996,
      "learning_rate": 0.0005313745631583103,
      "loss": 2.3518,
      "step": 50604
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7110388278961182,
      "learning_rate": 0.000531371959354417,
      "loss": 2.1554,
      "step": 50605
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7642339468002319,
      "learning_rate": 0.0005313693555075074,
      "loss": 2.4203,
      "step": 50606
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.8274056911468506,
      "learning_rate": 0.0005313667516175818,
      "loss": 2.2659,
      "step": 50607
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.902675449848175,
      "learning_rate": 0.0005313641476846408,
      "loss": 2.2872,
      "step": 50608
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.8285555243492126,
      "learning_rate": 0.0005313615437086848,
      "loss": 2.2427,
      "step": 50609
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7927983999252319,
      "learning_rate": 0.0005313589396897144,
      "loss": 2.392,
      "step": 50610
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7043247818946838,
      "learning_rate": 0.0005313563356277299,
      "loss": 2.295,
      "step": 50611
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.8292932510375977,
      "learning_rate": 0.0005313537315227319,
      "loss": 2.3018,
      "step": 50612
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.770260214805603,
      "learning_rate": 0.0005313511273747209,
      "loss": 2.1333,
      "step": 50613
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7346994280815125,
      "learning_rate": 0.0005313485231836974,
      "loss": 2.1491,
      "step": 50614
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7446240186691284,
      "learning_rate": 0.0005313459189496618,
      "loss": 2.4579,
      "step": 50615
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.9266158938407898,
      "learning_rate": 0.0005313433146726146,
      "loss": 2.2545,
      "step": 50616
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.8399960994720459,
      "learning_rate": 0.0005313407103525563,
      "loss": 2.3629,
      "step": 50617
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7728656530380249,
      "learning_rate": 0.0005313381059894875,
      "loss": 2.0336,
      "step": 50618
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.8829085230827332,
      "learning_rate": 0.0005313355015834086,
      "loss": 2.2333,
      "step": 50619
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.8011751174926758,
      "learning_rate": 0.0005313328971343199,
      "loss": 2.1231,
      "step": 50620
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.9094694256782532,
      "learning_rate": 0.0005313302926422222,
      "loss": 2.4121,
      "step": 50621
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.9032776951789856,
      "learning_rate": 0.0005313276881071157,
      "loss": 2.3709,
      "step": 50622
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7082152366638184,
      "learning_rate": 0.0005313250835290011,
      "loss": 2.2756,
      "step": 50623
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.693315327167511,
      "learning_rate": 0.0005313224789078787,
      "loss": 2.2391,
      "step": 50624
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.8988057971000671,
      "learning_rate": 0.0005313198742437493,
      "loss": 2.045,
      "step": 50625
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.912269651889801,
      "learning_rate": 0.000531317269536613,
      "loss": 2.1173,
      "step": 50626
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7198335528373718,
      "learning_rate": 0.0005313146647864705,
      "loss": 2.3621,
      "step": 50627
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7344604730606079,
      "learning_rate": 0.0005313120599933222,
      "loss": 2.3309,
      "step": 50628
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.8412380218505859,
      "learning_rate": 0.0005313094551571686,
      "loss": 2.4265,
      "step": 50629
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7228931188583374,
      "learning_rate": 0.0005313068502780103,
      "loss": 2.4294,
      "step": 50630
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.8098366856575012,
      "learning_rate": 0.0005313042453558476,
      "loss": 2.3867,
      "step": 50631
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.6757280826568604,
      "learning_rate": 0.0005313016403906811,
      "loss": 2.4275,
      "step": 50632
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.8052802085876465,
      "learning_rate": 0.0005312990353825113,
      "loss": 2.3408,
      "step": 50633
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7461432814598083,
      "learning_rate": 0.0005312964303313387,
      "loss": 2.1186,
      "step": 50634
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7136423587799072,
      "learning_rate": 0.0005312938252371636,
      "loss": 2.2136,
      "step": 50635
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7548028826713562,
      "learning_rate": 0.0005312912200999867,
      "loss": 2.2495,
      "step": 50636
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7626796960830688,
      "learning_rate": 0.0005312886149198083,
      "loss": 2.1149,
      "step": 50637
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.6925551295280457,
      "learning_rate": 0.0005312860096966291,
      "loss": 2.1445,
      "step": 50638
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7505797147750854,
      "learning_rate": 0.0005312834044304493,
      "loss": 2.3057,
      "step": 50639
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.6900762319564819,
      "learning_rate": 0.0005312807991212697,
      "loss": 2.3885,
      "step": 50640
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.910824716091156,
      "learning_rate": 0.0005312781937690906,
      "loss": 2.1672,
      "step": 50641
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7419895529747009,
      "learning_rate": 0.0005312755883739125,
      "loss": 2.269,
      "step": 50642
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.6747270226478577,
      "learning_rate": 0.0005312729829357359,
      "loss": 2.2805,
      "step": 50643
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.6807587742805481,
      "learning_rate": 0.0005312703774545612,
      "loss": 2.1513,
      "step": 50644
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.8089588284492493,
      "learning_rate": 0.0005312677719303892,
      "loss": 2.2401,
      "step": 50645
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7704233527183533,
      "learning_rate": 0.0005312651663632199,
      "loss": 2.3377,
      "step": 50646
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.784529983997345,
      "learning_rate": 0.0005312625607530542,
      "loss": 2.1384,
      "step": 50647
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7190458178520203,
      "learning_rate": 0.0005312599550998925,
      "loss": 2.4092,
      "step": 50648
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7640979290008545,
      "learning_rate": 0.0005312573494037349,
      "loss": 2.2233,
      "step": 50649
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7388137578964233,
      "learning_rate": 0.0005312547436645825,
      "loss": 2.2231,
      "step": 50650
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.698245108127594,
      "learning_rate": 0.0005312521378824354,
      "loss": 2.1072,
      "step": 50651
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.8120949268341064,
      "learning_rate": 0.0005312495320572941,
      "loss": 2.2037,
      "step": 50652
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7533003091812134,
      "learning_rate": 0.0005312469261891591,
      "loss": 2.1395,
      "step": 50653
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.8390014171600342,
      "learning_rate": 0.000531244320278031,
      "loss": 2.3383,
      "step": 50654
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.799629807472229,
      "learning_rate": 0.0005312417143239102,
      "loss": 2.2998,
      "step": 50655
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7661023139953613,
      "learning_rate": 0.0005312391083267974,
      "loss": 2.1366,
      "step": 50656
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.8548364639282227,
      "learning_rate": 0.0005312365022866925,
      "loss": 2.5013,
      "step": 50657
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7456738352775574,
      "learning_rate": 0.0005312338962035967,
      "loss": 2.1593,
      "step": 50658
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.6924521327018738,
      "learning_rate": 0.00053123129007751,
      "loss": 2.4316,
      "step": 50659
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7280488610267639,
      "learning_rate": 0.000531228683908433,
      "loss": 2.0644,
      "step": 50660
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7501577734947205,
      "learning_rate": 0.0005312260776963664,
      "loss": 2.3098,
      "step": 50661
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7264154553413391,
      "learning_rate": 0.0005312234714413103,
      "loss": 2.2508,
      "step": 50662
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7528802156448364,
      "learning_rate": 0.0005312208651432655,
      "loss": 2.5451,
      "step": 50663
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7108902335166931,
      "learning_rate": 0.0005312182588022323,
      "loss": 2.3207,
      "step": 50664
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7566693425178528,
      "learning_rate": 0.0005312156524182114,
      "loss": 2.267,
      "step": 50665
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7209944128990173,
      "learning_rate": 0.000531213045991203,
      "loss": 2.0969,
      "step": 50666
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7112021446228027,
      "learning_rate": 0.0005312104395212078,
      "loss": 2.1965,
      "step": 50667
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7283293008804321,
      "learning_rate": 0.0005312078330082262,
      "loss": 2.2481,
      "step": 50668
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.6822855472564697,
      "learning_rate": 0.0005312052264522588,
      "loss": 2.2453,
      "step": 50669
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7243502140045166,
      "learning_rate": 0.0005312026198533059,
      "loss": 2.0899,
      "step": 50670
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.792449951171875,
      "learning_rate": 0.0005312000132113681,
      "loss": 2.2025,
      "step": 50671
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7006545662879944,
      "learning_rate": 0.0005311974065264457,
      "loss": 2.2199,
      "step": 50672
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.8789005279541016,
      "learning_rate": 0.0005311947997985396,
      "loss": 2.1733,
      "step": 50673
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.6398475766181946,
      "learning_rate": 0.0005311921930276498,
      "loss": 2.1409,
      "step": 50674
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7062423229217529,
      "learning_rate": 0.0005311895862137771,
      "loss": 2.1228,
      "step": 50675
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.728047251701355,
      "learning_rate": 0.000531186979356922,
      "loss": 2.3418,
      "step": 50676
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7521987557411194,
      "learning_rate": 0.0005311843724570847,
      "loss": 2.4936,
      "step": 50677
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.8514487147331238,
      "learning_rate": 0.0005311817655142659,
      "loss": 2.3357,
      "step": 50678
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.6969898343086243,
      "learning_rate": 0.000531179158528466,
      "loss": 2.2604,
      "step": 50679
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7047525644302368,
      "learning_rate": 0.0005311765514996856,
      "loss": 2.289,
      "step": 50680
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7004312872886658,
      "learning_rate": 0.0005311739444279252,
      "loss": 2.2779,
      "step": 50681
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.6990752816200256,
      "learning_rate": 0.0005311713373131852,
      "loss": 2.2072,
      "step": 50682
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7976511120796204,
      "learning_rate": 0.0005311687301554659,
      "loss": 2.0606,
      "step": 50683
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.6965484619140625,
      "learning_rate": 0.0005311661229547681,
      "loss": 2.3078,
      "step": 50684
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7176724076271057,
      "learning_rate": 0.0005311635157110921,
      "loss": 2.3674,
      "step": 50685
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.748901903629303,
      "learning_rate": 0.0005311609084244385,
      "loss": 2.3366,
      "step": 50686
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7348120212554932,
      "learning_rate": 0.0005311583010948076,
      "loss": 2.3615,
      "step": 50687
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7399733066558838,
      "learning_rate": 0.0005311556937222001,
      "loss": 2.2886,
      "step": 50688
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.6957886219024658,
      "learning_rate": 0.0005311530863066164,
      "loss": 2.2435,
      "step": 50689
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7197040915489197,
      "learning_rate": 0.000531150478848057,
      "loss": 2.0994,
      "step": 50690
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.6840088963508606,
      "learning_rate": 0.0005311478713465222,
      "loss": 2.2469,
      "step": 50691
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7957276105880737,
      "learning_rate": 0.0005311452638020128,
      "loss": 2.1826,
      "step": 50692
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7393180727958679,
      "learning_rate": 0.0005311426562145291,
      "loss": 2.1763,
      "step": 50693
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7187314033508301,
      "learning_rate": 0.0005311400485840716,
      "loss": 2.1489,
      "step": 50694
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7144336700439453,
      "learning_rate": 0.0005311374409106409,
      "loss": 2.3469,
      "step": 50695
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.8280335664749146,
      "learning_rate": 0.0005311348331942373,
      "loss": 2.4091,
      "step": 50696
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7354916334152222,
      "learning_rate": 0.0005311322254348613,
      "loss": 2.1567,
      "step": 50697
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.8364405632019043,
      "learning_rate": 0.0005311296176325136,
      "loss": 2.3189,
      "step": 50698
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.712161123752594,
      "learning_rate": 0.0005311270097871945,
      "loss": 2.4731,
      "step": 50699
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.6843024492263794,
      "learning_rate": 0.0005311244018989045,
      "loss": 2.2381,
      "step": 50700
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7229626774787903,
      "learning_rate": 0.0005311217939676441,
      "loss": 2.0238,
      "step": 50701
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7486591339111328,
      "learning_rate": 0.0005311191859934138,
      "loss": 2.0837,
      "step": 50702
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7294169068336487,
      "learning_rate": 0.0005311165779762141,
      "loss": 2.2382,
      "step": 50703
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.6905087232589722,
      "learning_rate": 0.0005311139699160454,
      "loss": 2.2663,
      "step": 50704
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7210891842842102,
      "learning_rate": 0.0005311113618129083,
      "loss": 2.2599,
      "step": 50705
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.8251488208770752,
      "learning_rate": 0.0005311087536668033,
      "loss": 2.294,
      "step": 50706
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.8581119179725647,
      "learning_rate": 0.0005311061454777308,
      "loss": 2.5587,
      "step": 50707
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.8102915287017822,
      "learning_rate": 0.0005311035372456912,
      "loss": 2.2099,
      "step": 50708
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7344481348991394,
      "learning_rate": 0.0005311009289706852,
      "loss": 2.0342,
      "step": 50709
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7934533357620239,
      "learning_rate": 0.0005310983206527131,
      "loss": 2.2316,
      "step": 50710
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7804349064826965,
      "learning_rate": 0.0005310957122917755,
      "loss": 2.4178,
      "step": 50711
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.6954345107078552,
      "learning_rate": 0.0005310931038878729,
      "loss": 2.4081,
      "step": 50712
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.731461763381958,
      "learning_rate": 0.0005310904954410056,
      "loss": 2.264,
      "step": 50713
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.678057849407196,
      "learning_rate": 0.0005310878869511743,
      "loss": 2.2204,
      "step": 50714
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.710405170917511,
      "learning_rate": 0.0005310852784183794,
      "loss": 2.3808,
      "step": 50715
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.6879905462265015,
      "learning_rate": 0.0005310826698426214,
      "loss": 2.1,
      "step": 50716
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7506828904151917,
      "learning_rate": 0.0005310800612239007,
      "loss": 2.3202,
      "step": 50717
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7231478095054626,
      "learning_rate": 0.0005310774525622179,
      "loss": 2.3355,
      "step": 50718
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7140462398529053,
      "learning_rate": 0.0005310748438575734,
      "loss": 1.9312,
      "step": 50719
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.6917905807495117,
      "learning_rate": 0.0005310722351099678,
      "loss": 2.0508,
      "step": 50720
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.6571334004402161,
      "learning_rate": 0.0005310696263194014,
      "loss": 2.0453,
      "step": 50721
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.8434769511222839,
      "learning_rate": 0.0005310670174858749,
      "loss": 2.121,
      "step": 50722
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.6550693511962891,
      "learning_rate": 0.0005310644086093887,
      "loss": 2.2624,
      "step": 50723
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.8103213310241699,
      "learning_rate": 0.0005310617996899432,
      "loss": 2.3727,
      "step": 50724
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.6847221851348877,
      "learning_rate": 0.0005310591907275389,
      "loss": 2.324,
      "step": 50725
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.6658053398132324,
      "learning_rate": 0.0005310565817221763,
      "loss": 2.3074,
      "step": 50726
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7928269505500793,
      "learning_rate": 0.000531053972673856,
      "loss": 1.9969,
      "step": 50727
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7879653573036194,
      "learning_rate": 0.0005310513635825785,
      "loss": 2.3746,
      "step": 50728
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.6993460655212402,
      "learning_rate": 0.0005310487544483441,
      "loss": 2.1476,
      "step": 50729
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7048049569129944,
      "learning_rate": 0.0005310461452711534,
      "loss": 2.2421,
      "step": 50730
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.795171320438385,
      "learning_rate": 0.0005310435360510068,
      "loss": 2.4605,
      "step": 50731
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7374430894851685,
      "learning_rate": 0.0005310409267879049,
      "loss": 2.3783,
      "step": 50732
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.6942369341850281,
      "learning_rate": 0.0005310383174818481,
      "loss": 2.1772,
      "step": 50733
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7425335645675659,
      "learning_rate": 0.000531035708132837,
      "loss": 2.5207,
      "step": 50734
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7435837388038635,
      "learning_rate": 0.0005310330987408719,
      "loss": 2.2752,
      "step": 50735
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.6495669484138489,
      "learning_rate": 0.0005310304893059533,
      "loss": 2.3049,
      "step": 50736
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7352048754692078,
      "learning_rate": 0.000531027879828082,
      "loss": 2.1329,
      "step": 50737
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7602578997612,
      "learning_rate": 0.0005310252703072582,
      "loss": 2.2814,
      "step": 50738
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.6897342801094055,
      "learning_rate": 0.0005310226607434823,
      "loss": 2.2412,
      "step": 50739
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7537461519241333,
      "learning_rate": 0.0005310200511367551,
      "loss": 2.2367,
      "step": 50740
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.733772337436676,
      "learning_rate": 0.0005310174414870767,
      "loss": 2.0445,
      "step": 50741
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7650956511497498,
      "learning_rate": 0.0005310148317944479,
      "loss": 2.4944,
      "step": 50742
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7282882332801819,
      "learning_rate": 0.0005310122220588692,
      "loss": 1.9568,
      "step": 50743
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.66294926404953,
      "learning_rate": 0.0005310096122803408,
      "loss": 2.2096,
      "step": 50744
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7324323654174805,
      "learning_rate": 0.0005310070024588634,
      "loss": 2.3269,
      "step": 50745
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7706943154335022,
      "learning_rate": 0.0005310043925944375,
      "loss": 2.0907,
      "step": 50746
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.6890714764595032,
      "learning_rate": 0.0005310017826870635,
      "loss": 2.409,
      "step": 50747
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7302930355072021,
      "learning_rate": 0.0005309991727367419,
      "loss": 2.3099,
      "step": 50748
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.8493775725364685,
      "learning_rate": 0.0005309965627434732,
      "loss": 2.1918,
      "step": 50749
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.6917411088943481,
      "learning_rate": 0.0005309939527072578,
      "loss": 2.1657,
      "step": 50750
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7166988849639893,
      "learning_rate": 0.0005309913426280963,
      "loss": 2.2002,
      "step": 50751
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.8020893931388855,
      "learning_rate": 0.0005309887325059892,
      "loss": 2.1283,
      "step": 50752
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7838113903999329,
      "learning_rate": 0.000530986122340937,
      "loss": 2.3397,
      "step": 50753
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7050666809082031,
      "learning_rate": 0.00053098351213294,
      "loss": 2.2733,
      "step": 50754
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7327634692192078,
      "learning_rate": 0.0005309809018819988,
      "loss": 2.3669,
      "step": 50755
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7936478853225708,
      "learning_rate": 0.000530978291588114,
      "loss": 2.1936,
      "step": 50756
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7235401272773743,
      "learning_rate": 0.0005309756812512859,
      "loss": 2.2262,
      "step": 50757
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.6927093267440796,
      "learning_rate": 0.0005309730708715149,
      "loss": 2.0581,
      "step": 50758
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7503853440284729,
      "learning_rate": 0.0005309704604488019,
      "loss": 2.2964,
      "step": 50759
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.6867367029190063,
      "learning_rate": 0.000530967849983147,
      "loss": 2.218,
      "step": 50760
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.741052508354187,
      "learning_rate": 0.000530965239474551,
      "loss": 2.3622,
      "step": 50761
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7730188965797424,
      "learning_rate": 0.0005309626289230139,
      "loss": 2.0143,
      "step": 50762
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7105874419212341,
      "learning_rate": 0.0005309600183285367,
      "loss": 2.1769,
      "step": 50763
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7647794485092163,
      "learning_rate": 0.0005309574076911196,
      "loss": 2.3114,
      "step": 50764
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7037101984024048,
      "learning_rate": 0.0005309547970107632,
      "loss": 2.4649,
      "step": 50765
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7572263479232788,
      "learning_rate": 0.0005309521862874679,
      "loss": 2.2941,
      "step": 50766
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.704960286617279,
      "learning_rate": 0.0005309495755212343,
      "loss": 2.1441,
      "step": 50767
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7715782523155212,
      "learning_rate": 0.0005309469647120628,
      "loss": 2.2132,
      "step": 50768
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.8070122599601746,
      "learning_rate": 0.0005309443538599538,
      "loss": 2.3209,
      "step": 50769
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.8308990001678467,
      "learning_rate": 0.000530941742964908,
      "loss": 2.2454,
      "step": 50770
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7094373106956482,
      "learning_rate": 0.0005309391320269257,
      "loss": 1.8492,
      "step": 50771
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.6954032778739929,
      "learning_rate": 0.0005309365210460075,
      "loss": 2.0527,
      "step": 50772
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.8155899047851562,
      "learning_rate": 0.0005309339100221538,
      "loss": 2.5481,
      "step": 50773
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.6920899152755737,
      "learning_rate": 0.0005309312989553651,
      "loss": 2.2287,
      "step": 50774
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7737804651260376,
      "learning_rate": 0.0005309286878456419,
      "loss": 2.1539,
      "step": 50775
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7198363542556763,
      "learning_rate": 0.0005309260766929848,
      "loss": 2.0283,
      "step": 50776
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7331336140632629,
      "learning_rate": 0.0005309234654973942,
      "loss": 2.316,
      "step": 50777
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.702756941318512,
      "learning_rate": 0.0005309208542588704,
      "loss": 2.4372,
      "step": 50778
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.6925843358039856,
      "learning_rate": 0.0005309182429774142,
      "loss": 2.1365,
      "step": 50779
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.6506014466285706,
      "learning_rate": 0.0005309156316530259,
      "loss": 2.0879,
      "step": 50780
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7930925488471985,
      "learning_rate": 0.000530913020285706,
      "loss": 2.2187,
      "step": 50781
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.6710798144340515,
      "learning_rate": 0.0005309104088754551,
      "loss": 2.182,
      "step": 50782
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7374826073646545,
      "learning_rate": 0.0005309077974222735,
      "loss": 2.2659,
      "step": 50783
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7166941165924072,
      "learning_rate": 0.0005309051859261618,
      "loss": 2.2084,
      "step": 50784
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.6254843473434448,
      "learning_rate": 0.0005309025743871205,
      "loss": 2.1179,
      "step": 50785
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.6464178562164307,
      "learning_rate": 0.00053089996280515,
      "loss": 2.1508,
      "step": 50786
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.685705840587616,
      "learning_rate": 0.0005308973511802509,
      "loss": 2.109,
      "step": 50787
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7139134407043457,
      "learning_rate": 0.0005308947395124236,
      "loss": 2.2904,
      "step": 50788
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7009428143501282,
      "learning_rate": 0.0005308921278016686,
      "loss": 2.1932,
      "step": 50789
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.703856348991394,
      "learning_rate": 0.0005308895160479863,
      "loss": 2.2688,
      "step": 50790
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7222641706466675,
      "learning_rate": 0.0005308869042513775,
      "loss": 2.1567,
      "step": 50791
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7108129858970642,
      "learning_rate": 0.0005308842924118424,
      "loss": 2.1905,
      "step": 50792
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7029825448989868,
      "learning_rate": 0.0005308816805293814,
      "loss": 2.1863,
      "step": 50793
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7417383790016174,
      "learning_rate": 0.0005308790686039953,
      "loss": 2.1093,
      "step": 50794
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7331001162528992,
      "learning_rate": 0.0005308764566356842,
      "loss": 2.1916,
      "step": 50795
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.6807174682617188,
      "learning_rate": 0.0005308738446244491,
      "loss": 2.3871,
      "step": 50796
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.807220458984375,
      "learning_rate": 0.00053087123257029,
      "loss": 2.5512,
      "step": 50797
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7189096212387085,
      "learning_rate": 0.0005308686204732076,
      "loss": 2.1678,
      "step": 50798
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.6328315734863281,
      "learning_rate": 0.0005308660083332025,
      "loss": 2.1582,
      "step": 50799
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.6967027187347412,
      "learning_rate": 0.0005308633961502749,
      "loss": 2.0336,
      "step": 50800
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7015097141265869,
      "learning_rate": 0.0005308607839244256,
      "loss": 2.2563,
      "step": 50801
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7938921451568604,
      "learning_rate": 0.0005308581716556547,
      "loss": 2.1441,
      "step": 50802
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7251073122024536,
      "learning_rate": 0.0005308555593439631,
      "loss": 2.0513,
      "step": 50803
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7392946481704712,
      "learning_rate": 0.000530852946989351,
      "loss": 2.3429,
      "step": 50804
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7479269504547119,
      "learning_rate": 0.0005308503345918189,
      "loss": 2.4375,
      "step": 50805
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7474082112312317,
      "learning_rate": 0.0005308477221513675,
      "loss": 2.2614,
      "step": 50806
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7660979628562927,
      "learning_rate": 0.0005308451096679972,
      "loss": 2.4117,
      "step": 50807
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.9965834021568298,
      "learning_rate": 0.0005308424971417083,
      "loss": 2.1398,
      "step": 50808
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7643156051635742,
      "learning_rate": 0.0005308398845725015,
      "loss": 2.2307,
      "step": 50809
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7331172823905945,
      "learning_rate": 0.0005308372719603772,
      "loss": 2.4921,
      "step": 50810
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.8008863925933838,
      "learning_rate": 0.0005308346593053359,
      "loss": 2.2062,
      "step": 50811
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7103745937347412,
      "learning_rate": 0.000530832046607378,
      "loss": 2.3547,
      "step": 50812
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.6331550478935242,
      "learning_rate": 0.0005308294338665042,
      "loss": 2.1788,
      "step": 50813
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7506951689720154,
      "learning_rate": 0.0005308268210827147,
      "loss": 2.2352,
      "step": 50814
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.77066570520401,
      "learning_rate": 0.0005308242082560103,
      "loss": 2.3056,
      "step": 50815
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.6807821989059448,
      "learning_rate": 0.0005308215953863911,
      "loss": 2.2259,
      "step": 50816
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7803356647491455,
      "learning_rate": 0.000530818982473858,
      "loss": 2.3474,
      "step": 50817
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.8321489691734314,
      "learning_rate": 0.0005308163695184112,
      "loss": 2.1305,
      "step": 50818
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.8601935505867004,
      "learning_rate": 0.0005308137565200513,
      "loss": 2.3539,
      "step": 50819
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7687245011329651,
      "learning_rate": 0.0005308111434787787,
      "loss": 2.2636,
      "step": 50820
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7204065918922424,
      "learning_rate": 0.0005308085303945941,
      "loss": 2.2245,
      "step": 50821
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7444491982460022,
      "learning_rate": 0.0005308059172674978,
      "loss": 2.1683,
      "step": 50822
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7530627250671387,
      "learning_rate": 0.0005308033040974903,
      "loss": 2.1317,
      "step": 50823
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7321102619171143,
      "learning_rate": 0.0005308006908845721,
      "loss": 2.3355,
      "step": 50824
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.639599621295929,
      "learning_rate": 0.0005307980776287437,
      "loss": 2.1966,
      "step": 50825
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7140821814537048,
      "learning_rate": 0.0005307954643300056,
      "loss": 2.298,
      "step": 50826
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.800488293170929,
      "learning_rate": 0.0005307928509883582,
      "loss": 2.0967,
      "step": 50827
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7147930264472961,
      "learning_rate": 0.0005307902376038021,
      "loss": 2.0869,
      "step": 50828
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.8047411441802979,
      "learning_rate": 0.0005307876241763377,
      "loss": 2.4317,
      "step": 50829
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7463991045951843,
      "learning_rate": 0.0005307850107059656,
      "loss": 2.1679,
      "step": 50830
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7786501049995422,
      "learning_rate": 0.0005307823971926861,
      "loss": 2.0677,
      "step": 50831
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.6696789264678955,
      "learning_rate": 0.0005307797836365,
      "loss": 2.2042,
      "step": 50832
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7631168365478516,
      "learning_rate": 0.0005307771700374074,
      "loss": 2.1929,
      "step": 50833
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.6664316654205322,
      "learning_rate": 0.0005307745563954089,
      "loss": 2.3767,
      "step": 50834
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7347249388694763,
      "learning_rate": 0.0005307719427105052,
      "loss": 2.1237,
      "step": 50835
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7013487219810486,
      "learning_rate": 0.0005307693289826967,
      "loss": 2.2268,
      "step": 50836
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7074446678161621,
      "learning_rate": 0.0005307667152119836,
      "loss": 2.2533,
      "step": 50837
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7657521963119507,
      "learning_rate": 0.0005307641013983668,
      "loss": 2.2158,
      "step": 50838
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7130953669548035,
      "learning_rate": 0.0005307614875418467,
      "loss": 2.311,
      "step": 50839
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7822784781455994,
      "learning_rate": 0.0005307588736424234,
      "loss": 2.5695,
      "step": 50840
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.770821213722229,
      "learning_rate": 0.0005307562597000978,
      "loss": 2.3402,
      "step": 50841
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.8638968467712402,
      "learning_rate": 0.0005307536457148703,
      "loss": 2.4046,
      "step": 50842
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7174794673919678,
      "learning_rate": 0.0005307510316867412,
      "loss": 2.1437,
      "step": 50843
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.6896023750305176,
      "learning_rate": 0.0005307484176157113,
      "loss": 2.1303,
      "step": 50844
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.6998928189277649,
      "learning_rate": 0.0005307458035017808,
      "loss": 2.1607,
      "step": 50845
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.8146951198577881,
      "learning_rate": 0.0005307431893449503,
      "loss": 1.9958,
      "step": 50846
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7331992387771606,
      "learning_rate": 0.0005307405751452203,
      "loss": 2.169,
      "step": 50847
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7643216848373413,
      "learning_rate": 0.0005307379609025913,
      "loss": 2.1449,
      "step": 50848
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.8274532556533813,
      "learning_rate": 0.0005307353466170638,
      "loss": 2.3447,
      "step": 50849
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7532204985618591,
      "learning_rate": 0.0005307327322886382,
      "loss": 2.3084,
      "step": 50850
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7454721331596375,
      "learning_rate": 0.000530730117917315,
      "loss": 2.1775,
      "step": 50851
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.6605902314186096,
      "learning_rate": 0.0005307275035030947,
      "loss": 2.1383,
      "step": 50852
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7031177878379822,
      "learning_rate": 0.0005307248890459779,
      "loss": 2.1232,
      "step": 50853
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7529407739639282,
      "learning_rate": 0.0005307222745459649,
      "loss": 2.3986,
      "step": 50854
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.795016884803772,
      "learning_rate": 0.0005307196600030562,
      "loss": 2.3685,
      "step": 50855
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7296043038368225,
      "learning_rate": 0.0005307170454172525,
      "loss": 2.3167,
      "step": 50856
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.644559383392334,
      "learning_rate": 0.0005307144307885541,
      "loss": 2.1462,
      "step": 50857
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.6809832453727722,
      "learning_rate": 0.0005307118161169616,
      "loss": 2.2201,
      "step": 50858
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7580923438072205,
      "learning_rate": 0.0005307092014024753,
      "loss": 2.1817,
      "step": 50859
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7374311089515686,
      "learning_rate": 0.0005307065866450958,
      "loss": 2.2451,
      "step": 50860
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.6706498861312866,
      "learning_rate": 0.0005307039718448237,
      "loss": 2.2922,
      "step": 50861
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7065337896347046,
      "learning_rate": 0.0005307013570016593,
      "loss": 2.3707,
      "step": 50862
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7463372349739075,
      "learning_rate": 0.0005306987421156033,
      "loss": 2.3596,
      "step": 50863
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7584038972854614,
      "learning_rate": 0.0005306961271866559,
      "loss": 2.0587,
      "step": 50864
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7072644233703613,
      "learning_rate": 0.0005306935122148178,
      "loss": 2.3552,
      "step": 50865
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.8022677302360535,
      "learning_rate": 0.0005306908972000894,
      "loss": 2.2961,
      "step": 50866
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7989436388015747,
      "learning_rate": 0.0005306882821424711,
      "loss": 2.1684,
      "step": 50867
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7279090285301208,
      "learning_rate": 0.0005306856670419637,
      "loss": 2.2152,
      "step": 50868
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.6724879145622253,
      "learning_rate": 0.0005306830518985674,
      "loss": 2.1727,
      "step": 50869
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.6965845823287964,
      "learning_rate": 0.0005306804367122826,
      "loss": 2.4024,
      "step": 50870
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.9394755363464355,
      "learning_rate": 0.0005306778214831102,
      "loss": 2.1649,
      "step": 50871
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.8122367262840271,
      "learning_rate": 0.0005306752062110503,
      "loss": 2.4262,
      "step": 50872
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7125126719474792,
      "learning_rate": 0.0005306725908961036,
      "loss": 2.3751,
      "step": 50873
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7538294196128845,
      "learning_rate": 0.0005306699755382704,
      "loss": 2.3228,
      "step": 50874
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7954733371734619,
      "learning_rate": 0.0005306673601375515,
      "loss": 2.1655,
      "step": 50875
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7108966708183289,
      "learning_rate": 0.0005306647446939469,
      "loss": 2.3898,
      "step": 50876
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7342315316200256,
      "learning_rate": 0.0005306621292074576,
      "loss": 2.2472,
      "step": 50877
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.755847156047821,
      "learning_rate": 0.0005306595136780837,
      "loss": 2.1601,
      "step": 50878
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.6862035989761353,
      "learning_rate": 0.0005306568981058259,
      "loss": 2.2116,
      "step": 50879
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7543345093727112,
      "learning_rate": 0.0005306542824906847,
      "loss": 2.113,
      "step": 50880
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7421084046363831,
      "learning_rate": 0.0005306516668326605,
      "loss": 2.2851,
      "step": 50881
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.8028141260147095,
      "learning_rate": 0.0005306490511317538,
      "loss": 2.2286,
      "step": 50882
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7352585196495056,
      "learning_rate": 0.000530646435387965,
      "loss": 2.4523,
      "step": 50883
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7960678935050964,
      "learning_rate": 0.0005306438196012947,
      "loss": 2.23,
      "step": 50884
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7265710234642029,
      "learning_rate": 0.0005306412037717434,
      "loss": 2.0726,
      "step": 50885
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.642570972442627,
      "learning_rate": 0.0005306385878993115,
      "loss": 2.0517,
      "step": 50886
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7946502566337585,
      "learning_rate": 0.0005306359719839996,
      "loss": 2.222,
      "step": 50887
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7348916530609131,
      "learning_rate": 0.0005306333560258081,
      "loss": 2.2937,
      "step": 50888
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7292159795761108,
      "learning_rate": 0.0005306307400247375,
      "loss": 2.2978,
      "step": 50889
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.8766440749168396,
      "learning_rate": 0.0005306281239807882,
      "loss": 2.3226,
      "step": 50890
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7340646982192993,
      "learning_rate": 0.0005306255078939609,
      "loss": 2.2652,
      "step": 50891
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7063280344009399,
      "learning_rate": 0.0005306228917642559,
      "loss": 2.3092,
      "step": 50892
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.8125365972518921,
      "learning_rate": 0.0005306202755916738,
      "loss": 2.2929,
      "step": 50893
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7466060519218445,
      "learning_rate": 0.000530617659376215,
      "loss": 2.2426,
      "step": 50894
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.8317223787307739,
      "learning_rate": 0.00053061504311788,
      "loss": 2.4272,
      "step": 50895
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7417041659355164,
      "learning_rate": 0.0005306124268166694,
      "loss": 2.118,
      "step": 50896
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.6503355503082275,
      "learning_rate": 0.0005306098104725836,
      "loss": 2.1903,
      "step": 50897
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7800692915916443,
      "learning_rate": 0.000530607194085623,
      "loss": 2.031,
      "step": 50898
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7751383185386658,
      "learning_rate": 0.0005306045776557882,
      "loss": 2.377,
      "step": 50899
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.8251244425773621,
      "learning_rate": 0.0005306019611830796,
      "loss": 2.1928,
      "step": 50900
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.690520703792572,
      "learning_rate": 0.0005305993446674979,
      "loss": 2.1951,
      "step": 50901
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.661794126033783,
      "learning_rate": 0.0005305967281090433,
      "loss": 2.3305,
      "step": 50902
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.741740345954895,
      "learning_rate": 0.0005305941115077164,
      "loss": 2.159,
      "step": 50903
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7156943082809448,
      "learning_rate": 0.0005305914948635177,
      "loss": 2.0624,
      "step": 50904
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.662732720375061,
      "learning_rate": 0.0005305888781764477,
      "loss": 2.088,
      "step": 50905
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.8127644658088684,
      "learning_rate": 0.000530586261446507,
      "loss": 2.3884,
      "step": 50906
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7707335352897644,
      "learning_rate": 0.0005305836446736959,
      "loss": 2.3619,
      "step": 50907
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7119895219802856,
      "learning_rate": 0.0005305810278580149,
      "loss": 2.1971,
      "step": 50908
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7440228462219238,
      "learning_rate": 0.0005305784109994645,
      "loss": 2.1135,
      "step": 50909
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.9433420896530151,
      "learning_rate": 0.0005305757940980453,
      "loss": 2.36,
      "step": 50910
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7493774890899658,
      "learning_rate": 0.0005305731771537576,
      "loss": 2.2757,
      "step": 50911
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7430376410484314,
      "learning_rate": 0.0005305705601666021,
      "loss": 2.1578,
      "step": 50912
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7679075598716736,
      "learning_rate": 0.0005305679431365792,
      "loss": 2.33,
      "step": 50913
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.1424264907836914,
      "learning_rate": 0.0005305653260636893,
      "loss": 2.3515,
      "step": 50914
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.8523207902908325,
      "learning_rate": 0.000530562708947933,
      "loss": 2.3284,
      "step": 50915
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.9165651202201843,
      "learning_rate": 0.0005305600917893107,
      "loss": 2.2719,
      "step": 50916
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.9608137011528015,
      "learning_rate": 0.0005305574745878229,
      "loss": 2.0547,
      "step": 50917
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.8273130655288696,
      "learning_rate": 0.0005305548573434702,
      "loss": 2.2342,
      "step": 50918
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.9761655926704407,
      "learning_rate": 0.0005305522400562529,
      "loss": 2.1148,
      "step": 50919
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.8344311118125916,
      "learning_rate": 0.0005305496227261716,
      "loss": 2.2951,
      "step": 50920
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7390199899673462,
      "learning_rate": 0.0005305470053532269,
      "loss": 2.14,
      "step": 50921
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7336863875389099,
      "learning_rate": 0.000530544387937419,
      "loss": 2.2274,
      "step": 50922
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7751609086990356,
      "learning_rate": 0.0005305417704787487,
      "loss": 2.1466,
      "step": 50923
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.877059817314148,
      "learning_rate": 0.0005305391529772162,
      "loss": 2.1473,
      "step": 50924
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7331088781356812,
      "learning_rate": 0.0005305365354328222,
      "loss": 2.1002,
      "step": 50925
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.9207700490951538,
      "learning_rate": 0.000530533917845567,
      "loss": 2.3994,
      "step": 50926
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.809212327003479,
      "learning_rate": 0.0005305313002154513,
      "loss": 2.1689,
      "step": 50927
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.6315329074859619,
      "learning_rate": 0.0005305286825424755,
      "loss": 2.138,
      "step": 50928
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.8758095502853394,
      "learning_rate": 0.00053052606482664,
      "loss": 2.3295,
      "step": 50929
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.8390426635742188,
      "learning_rate": 0.0005305234470679454,
      "loss": 2.3549,
      "step": 50930
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7706050276756287,
      "learning_rate": 0.0005305208292663921,
      "loss": 2.3386,
      "step": 50931
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.6809224486351013,
      "learning_rate": 0.0005305182114219806,
      "loss": 2.3762,
      "step": 50932
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.8495795130729675,
      "learning_rate": 0.0005305155935347115,
      "loss": 2.3817,
      "step": 50933
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7995942234992981,
      "learning_rate": 0.0005305129756045851,
      "loss": 2.1348,
      "step": 50934
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.8054885268211365,
      "learning_rate": 0.0005305103576316021,
      "loss": 2.1619,
      "step": 50935
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.8378520011901855,
      "learning_rate": 0.0005305077396157627,
      "loss": 2.3797,
      "step": 50936
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7965981364250183,
      "learning_rate": 0.0005305051215570677,
      "loss": 2.142,
      "step": 50937
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.8235856890678406,
      "learning_rate": 0.0005305025034555175,
      "loss": 2.2756,
      "step": 50938
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.6687524318695068,
      "learning_rate": 0.0005304998853111123,
      "loss": 2.1667,
      "step": 50939
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.637324333190918,
      "learning_rate": 0.000530497267123853,
      "loss": 2.2234,
      "step": 50940
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.893248975276947,
      "learning_rate": 0.0005304946488937399,
      "loss": 2.4824,
      "step": 50941
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7768751978874207,
      "learning_rate": 0.0005304920306207735,
      "loss": 2.1981,
      "step": 50942
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7074700593948364,
      "learning_rate": 0.0005304894123049543,
      "loss": 2.2268,
      "step": 50943
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7127736806869507,
      "learning_rate": 0.0005304867939462827,
      "loss": 2.2377,
      "step": 50944
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.8554255962371826,
      "learning_rate": 0.0005304841755447592,
      "loss": 2.1238,
      "step": 50945
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.782798171043396,
      "learning_rate": 0.0005304815571003845,
      "loss": 2.2471,
      "step": 50946
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7243356108665466,
      "learning_rate": 0.0005304789386131588,
      "loss": 2.29,
      "step": 50947
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.0391957759857178,
      "learning_rate": 0.0005304763200830826,
      "loss": 2.4323,
      "step": 50948
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7337860465049744,
      "learning_rate": 0.0005304737015101568,
      "loss": 2.3788,
      "step": 50949
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.8096234202384949,
      "learning_rate": 0.0005304710828943815,
      "loss": 2.3583,
      "step": 50950
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7901716232299805,
      "learning_rate": 0.0005304684642357571,
      "loss": 2.0895,
      "step": 50951
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.6861870288848877,
      "learning_rate": 0.0005304658455342844,
      "loss": 2.1004,
      "step": 50952
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7336615324020386,
      "learning_rate": 0.0005304632267899638,
      "loss": 2.2617,
      "step": 50953
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.8091813921928406,
      "learning_rate": 0.0005304606080027956,
      "loss": 2.2574,
      "step": 50954
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7540394067764282,
      "learning_rate": 0.0005304579891727806,
      "loss": 2.6071,
      "step": 50955
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.6898424625396729,
      "learning_rate": 0.000530455370299919,
      "loss": 2.2236,
      "step": 50956
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.6513673663139343,
      "learning_rate": 0.0005304527513842113,
      "loss": 2.3456,
      "step": 50957
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.6895139217376709,
      "learning_rate": 0.0005304501324256582,
      "loss": 2.5221,
      "step": 50958
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7132118940353394,
      "learning_rate": 0.0005304475134242601,
      "loss": 2.2767,
      "step": 50959
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.6772200465202332,
      "learning_rate": 0.0005304448943800173,
      "loss": 2.275,
      "step": 50960
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.746360182762146,
      "learning_rate": 0.0005304422752929306,
      "loss": 2.1757,
      "step": 50961
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7424105405807495,
      "learning_rate": 0.0005304396561630003,
      "loss": 2.2444,
      "step": 50962
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.6466232538223267,
      "learning_rate": 0.0005304370369902269,
      "loss": 2.0391,
      "step": 50963
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7311757206916809,
      "learning_rate": 0.0005304344177746109,
      "loss": 2.5077,
      "step": 50964
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7607852220535278,
      "learning_rate": 0.0005304317985161528,
      "loss": 2.0879,
      "step": 50965
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7566898465156555,
      "learning_rate": 0.000530429179214853,
      "loss": 2.3489,
      "step": 50966
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7139564156532288,
      "learning_rate": 0.0005304265598707123,
      "loss": 2.4023,
      "step": 50967
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.6944593191146851,
      "learning_rate": 0.0005304239404837306,
      "loss": 2.2448,
      "step": 50968
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.8056864142417908,
      "learning_rate": 0.000530421321053909,
      "loss": 2.2902,
      "step": 50969
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.0166834592819214,
      "learning_rate": 0.0005304187015812477,
      "loss": 2.3172,
      "step": 50970
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7346978187561035,
      "learning_rate": 0.0005304160820657472,
      "loss": 2.1073,
      "step": 50971
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7767476439476013,
      "learning_rate": 0.000530413462507408,
      "loss": 2.4422,
      "step": 50972
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.8000600934028625,
      "learning_rate": 0.0005304108429062305,
      "loss": 2.4085,
      "step": 50973
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7272638082504272,
      "learning_rate": 0.0005304082232622154,
      "loss": 2.188,
      "step": 50974
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.6895736455917358,
      "learning_rate": 0.0005304056035753629,
      "loss": 2.0913,
      "step": 50975
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7561300992965698,
      "learning_rate": 0.0005304029838456738,
      "loss": 2.4373,
      "step": 50976
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.8105373382568359,
      "learning_rate": 0.0005304003640731484,
      "loss": 2.0709,
      "step": 50977
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7619982957839966,
      "learning_rate": 0.0005303977442577872,
      "loss": 2.2945,
      "step": 50978
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7807155251502991,
      "learning_rate": 0.0005303951243995908,
      "loss": 2.0003,
      "step": 50979
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7221556901931763,
      "learning_rate": 0.0005303925044985594,
      "loss": 2.311,
      "step": 50980
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7411683201789856,
      "learning_rate": 0.0005303898845546938,
      "loss": 2.3286,
      "step": 50981
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.6867998242378235,
      "learning_rate": 0.0005303872645679945,
      "loss": 2.2402,
      "step": 50982
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7248468995094299,
      "learning_rate": 0.0005303846445384616,
      "loss": 2.4256,
      "step": 50983
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.8070587515830994,
      "learning_rate": 0.0005303820244660961,
      "loss": 2.2099,
      "step": 50984
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7477772831916809,
      "learning_rate": 0.000530379404350898,
      "loss": 2.1805,
      "step": 50985
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.8239989280700684,
      "learning_rate": 0.0005303767841928681,
      "loss": 2.2948,
      "step": 50986
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.6968250274658203,
      "learning_rate": 0.000530374163992007,
      "loss": 2.3966,
      "step": 50987
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7240064144134521,
      "learning_rate": 0.0005303715437483148,
      "loss": 2.3925,
      "step": 50988
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7272754311561584,
      "learning_rate": 0.0005303689234617922,
      "loss": 2.1923,
      "step": 50989
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.738305389881134,
      "learning_rate": 0.0005303663031324396,
      "loss": 2.2333,
      "step": 50990
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.800886332988739,
      "learning_rate": 0.0005303636827602577,
      "loss": 2.2325,
      "step": 50991
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7069918513298035,
      "learning_rate": 0.0005303610623452467,
      "loss": 2.3506,
      "step": 50992
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7591716051101685,
      "learning_rate": 0.0005303584418874074,
      "loss": 1.9909,
      "step": 50993
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7686771154403687,
      "learning_rate": 0.00053035582138674,
      "loss": 2.5107,
      "step": 50994
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7916532754898071,
      "learning_rate": 0.0005303532008432451,
      "loss": 2.4593,
      "step": 50995
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7263151407241821,
      "learning_rate": 0.0005303505802569233,
      "loss": 1.9721,
      "step": 50996
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7170690298080444,
      "learning_rate": 0.0005303479596277748,
      "loss": 2.1086,
      "step": 50997
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7945898175239563,
      "learning_rate": 0.0005303453389558004,
      "loss": 2.3337,
      "step": 50998
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7121784687042236,
      "learning_rate": 0.0005303427182410005,
      "loss": 2.4061,
      "step": 50999
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7874629497528076,
      "learning_rate": 0.0005303400974833755,
      "loss": 2.1602,
      "step": 51000
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7260276079177856,
      "learning_rate": 0.0005303374766829259,
      "loss": 2.0763,
      "step": 51001
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.8551456928253174,
      "learning_rate": 0.0005303348558396521,
      "loss": 2.2461,
      "step": 51002
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.8114821314811707,
      "learning_rate": 0.0005303322349535549,
      "loss": 2.3421,
      "step": 51003
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7447844743728638,
      "learning_rate": 0.0005303296140246345,
      "loss": 2.4245,
      "step": 51004
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.840533435344696,
      "learning_rate": 0.0005303269930528913,
      "loss": 2.1685,
      "step": 51005
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7442779541015625,
      "learning_rate": 0.0005303243720383262,
      "loss": 2.2827,
      "step": 51006
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7085054516792297,
      "learning_rate": 0.0005303217509809394,
      "loss": 2.2075,
      "step": 51007
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.9144011735916138,
      "learning_rate": 0.0005303191298807315,
      "loss": 2.1686,
      "step": 51008
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7654657959938049,
      "learning_rate": 0.0005303165087377028,
      "loss": 2.2291,
      "step": 51009
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7169494032859802,
      "learning_rate": 0.000530313887551854,
      "loss": 2.4358,
      "step": 51010
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.8658782839775085,
      "learning_rate": 0.0005303112663231855,
      "loss": 2.1875,
      "step": 51011
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.8083493709564209,
      "learning_rate": 0.0005303086450516978,
      "loss": 2.1451,
      "step": 51012
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7590340375900269,
      "learning_rate": 0.0005303060237373913,
      "loss": 2.3836,
      "step": 51013
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.8403033018112183,
      "learning_rate": 0.0005303034023802666,
      "loss": 2.3031,
      "step": 51014
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.672442615032196,
      "learning_rate": 0.0005303007809803241,
      "loss": 2.2683,
      "step": 51015
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7426661252975464,
      "learning_rate": 0.0005302981595375644,
      "loss": 2.3737,
      "step": 51016
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7295637726783752,
      "learning_rate": 0.0005302955380519879,
      "loss": 2.1111,
      "step": 51017
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.8688253164291382,
      "learning_rate": 0.0005302929165235951,
      "loss": 2.3374,
      "step": 51018
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.6996448040008545,
      "learning_rate": 0.0005302902949523865,
      "loss": 2.2404,
      "step": 51019
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7060850858688354,
      "learning_rate": 0.0005302876733383627,
      "loss": 2.2714,
      "step": 51020
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7542051076889038,
      "learning_rate": 0.0005302850516815239,
      "loss": 2.1704,
      "step": 51021
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7240557074546814,
      "learning_rate": 0.0005302824299818708,
      "loss": 2.4866,
      "step": 51022
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.6580873131752014,
      "learning_rate": 0.0005302798082394039,
      "loss": 2.3541,
      "step": 51023
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.6209801435470581,
      "learning_rate": 0.0005302771864541237,
      "loss": 2.3573,
      "step": 51024
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7586122751235962,
      "learning_rate": 0.0005302745646260306,
      "loss": 2.1826,
      "step": 51025
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.6654248833656311,
      "learning_rate": 0.000530271942755125,
      "loss": 2.1436,
      "step": 51026
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.6604722142219543,
      "learning_rate": 0.0005302693208414076,
      "loss": 2.1004,
      "step": 51027
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.751636803150177,
      "learning_rate": 0.0005302666988848787,
      "loss": 2.4402,
      "step": 51028
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.741069495677948,
      "learning_rate": 0.000530264076885539,
      "loss": 2.2034,
      "step": 51029
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7509323954582214,
      "learning_rate": 0.0005302614548433887,
      "loss": 2.267,
      "step": 51030
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7446638345718384,
      "learning_rate": 0.0005302588327584287,
      "loss": 2.3527,
      "step": 51031
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7036344408988953,
      "learning_rate": 0.000530256210630659,
      "loss": 2.2597,
      "step": 51032
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7555992007255554,
      "learning_rate": 0.0005302535884600804,
      "loss": 2.213,
      "step": 51033
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.9074267148971558,
      "learning_rate": 0.0005302509662466933,
      "loss": 2.241,
      "step": 51034
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7700074911117554,
      "learning_rate": 0.0005302483439904982,
      "loss": 2.4127,
      "step": 51035
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7696967720985413,
      "learning_rate": 0.0005302457216914956,
      "loss": 2.1859,
      "step": 51036
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7075099349021912,
      "learning_rate": 0.0005302430993496859,
      "loss": 2.2333,
      "step": 51037
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7439463138580322,
      "learning_rate": 0.0005302404769650697,
      "loss": 2.1923,
      "step": 51038
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.9307438135147095,
      "learning_rate": 0.0005302378545376476,
      "loss": 2.2531,
      "step": 51039
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.6326881051063538,
      "learning_rate": 0.0005302352320674199,
      "loss": 1.9523,
      "step": 51040
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.715783417224884,
      "learning_rate": 0.000530232609554387,
      "loss": 2.3875,
      "step": 51041
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7837197780609131,
      "learning_rate": 0.0005302299869985494,
      "loss": 2.364,
      "step": 51042
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7824856638908386,
      "learning_rate": 0.0005302273643999079,
      "loss": 2.2187,
      "step": 51043
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.892141580581665,
      "learning_rate": 0.0005302247417584627,
      "loss": 2.4741,
      "step": 51044
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.6781790852546692,
      "learning_rate": 0.0005302221190742145,
      "loss": 2.148,
      "step": 51045
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7769458889961243,
      "learning_rate": 0.0005302194963471635,
      "loss": 2.2767,
      "step": 51046
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.6660463213920593,
      "learning_rate": 0.0005302168735773106,
      "loss": 2.3571,
      "step": 51047
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7093309760093689,
      "learning_rate": 0.0005302142507646557,
      "loss": 2.3541,
      "step": 51048
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.6660062670707703,
      "learning_rate": 0.0005302116279091998,
      "loss": 2.3833,
      "step": 51049
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7195661067962646,
      "learning_rate": 0.0005302090050109433,
      "loss": 2.492,
      "step": 51050
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.6700282096862793,
      "learning_rate": 0.0005302063820698864,
      "loss": 2.1357,
      "step": 51051
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.6962092518806458,
      "learning_rate": 0.00053020375908603,
      "loss": 2.2724,
      "step": 51052
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7094990611076355,
      "learning_rate": 0.0005302011360593743,
      "loss": 2.4141,
      "step": 51053
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7841576337814331,
      "learning_rate": 0.0005301985129899197,
      "loss": 2.3891,
      "step": 51054
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.6996246576309204,
      "learning_rate": 0.000530195889877667,
      "loss": 2.4051,
      "step": 51055
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7645965218544006,
      "learning_rate": 0.0005301932667226165,
      "loss": 2.2741,
      "step": 51056
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7421693205833435,
      "learning_rate": 0.0005301906435247688,
      "loss": 2.2778,
      "step": 51057
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7020835280418396,
      "learning_rate": 0.0005301880202841242,
      "loss": 2.315,
      "step": 51058
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7821236848831177,
      "learning_rate": 0.0005301853970006834,
      "loss": 2.2438,
      "step": 51059
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.6696101427078247,
      "learning_rate": 0.0005301827736744467,
      "loss": 2.0327,
      "step": 51060
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7669756412506104,
      "learning_rate": 0.0005301801503054148,
      "loss": 2.3668,
      "step": 51061
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7047580480575562,
      "learning_rate": 0.000530177526893588,
      "loss": 2.2931,
      "step": 51062
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.6836977601051331,
      "learning_rate": 0.0005301749034389669,
      "loss": 2.2025,
      "step": 51063
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.8181337118148804,
      "learning_rate": 0.0005301722799415518,
      "loss": 2.2822,
      "step": 51064
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7068184614181519,
      "learning_rate": 0.0005301696564013434,
      "loss": 2.3532,
      "step": 51065
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7245960235595703,
      "learning_rate": 0.0005301670328183421,
      "loss": 2.3562,
      "step": 51066
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.2162338495254517,
      "learning_rate": 0.0005301644091925484,
      "loss": 2.3532,
      "step": 51067
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.6840436458587646,
      "learning_rate": 0.0005301617855239628,
      "loss": 2.246,
      "step": 51068
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7444314360618591,
      "learning_rate": 0.0005301591618125859,
      "loss": 2.2753,
      "step": 51069
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7316071391105652,
      "learning_rate": 0.000530156538058418,
      "loss": 2.2184,
      "step": 51070
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7228367924690247,
      "learning_rate": 0.0005301539142614596,
      "loss": 2.2341,
      "step": 51071
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7615293860435486,
      "learning_rate": 0.0005301512904217114,
      "loss": 2.2676,
      "step": 51072
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.6596075892448425,
      "learning_rate": 0.0005301486665391735,
      "loss": 2.2564,
      "step": 51073
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.6969774961471558,
      "learning_rate": 0.0005301460426138467,
      "loss": 2.4426,
      "step": 51074
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.8268441557884216,
      "learning_rate": 0.0005301434186457315,
      "loss": 2.2041,
      "step": 51075
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7676100730895996,
      "learning_rate": 0.0005301407946348281,
      "loss": 2.347,
      "step": 51076
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7261552214622498,
      "learning_rate": 0.0005301381705811374,
      "loss": 2.1107,
      "step": 51077
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7213199734687805,
      "learning_rate": 0.0005301355464846595,
      "loss": 2.4715,
      "step": 51078
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.8022844791412354,
      "learning_rate": 0.0005301329223453952,
      "loss": 2.4246,
      "step": 51079
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.731963574886322,
      "learning_rate": 0.0005301302981633448,
      "loss": 2.323,
      "step": 51080
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7033452391624451,
      "learning_rate": 0.0005301276739385088,
      "loss": 2.2534,
      "step": 51081
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7208845615386963,
      "learning_rate": 0.0005301250496708877,
      "loss": 2.0613,
      "step": 51082
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7148013710975647,
      "learning_rate": 0.000530122425360482,
      "loss": 2.3362,
      "step": 51083
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.6697133779525757,
      "learning_rate": 0.0005301198010072923,
      "loss": 2.3247,
      "step": 51084
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7176044583320618,
      "learning_rate": 0.000530117176611319,
      "loss": 2.3695,
      "step": 51085
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7982349991798401,
      "learning_rate": 0.0005301145521725624,
      "loss": 2.0225,
      "step": 51086
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7230744361877441,
      "learning_rate": 0.0005301119276910232,
      "loss": 2.508,
      "step": 51087
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.6870822310447693,
      "learning_rate": 0.000530109303166702,
      "loss": 2.3515,
      "step": 51088
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7278656363487244,
      "learning_rate": 0.0005301066785995991,
      "loss": 2.4425,
      "step": 51089
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7194103598594666,
      "learning_rate": 0.0005301040539897148,
      "loss": 2.0337,
      "step": 51090
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7657903432846069,
      "learning_rate": 0.00053010142933705,
      "loss": 2.3367,
      "step": 51091
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7271804809570312,
      "learning_rate": 0.000530098804641605,
      "loss": 2.0602,
      "step": 51092
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.9365656971931458,
      "learning_rate": 0.0005300961799033803,
      "loss": 2.2385,
      "step": 51093
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7261733412742615,
      "learning_rate": 0.0005300935551223764,
      "loss": 2.4302,
      "step": 51094
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7807105779647827,
      "learning_rate": 0.0005300909302985936,
      "loss": 2.2918,
      "step": 51095
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7181476354598999,
      "learning_rate": 0.0005300883054320327,
      "loss": 2.0641,
      "step": 51096
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.963050127029419,
      "learning_rate": 0.000530085680522694,
      "loss": 2.2837,
      "step": 51097
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.6802314519882202,
      "learning_rate": 0.0005300830555705781,
      "loss": 2.0969,
      "step": 51098
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7653121948242188,
      "learning_rate": 0.0005300804305756853,
      "loss": 2.5476,
      "step": 51099
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7286203503608704,
      "learning_rate": 0.0005300778055380163,
      "loss": 2.281,
      "step": 51100
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.75840163230896,
      "learning_rate": 0.0005300751804575715,
      "loss": 2.4922,
      "step": 51101
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.6866783499717712,
      "learning_rate": 0.0005300725553343512,
      "loss": 1.9397,
      "step": 51102
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7114363312721252,
      "learning_rate": 0.0005300699301683562,
      "loss": 2.1513,
      "step": 51103
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.8531962037086487,
      "learning_rate": 0.0005300673049595869,
      "loss": 2.2446,
      "step": 51104
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.6757109761238098,
      "learning_rate": 0.0005300646797080437,
      "loss": 2.0989,
      "step": 51105
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7146547436714172,
      "learning_rate": 0.0005300620544137271,
      "loss": 2.2326,
      "step": 51106
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.8979055881500244,
      "learning_rate": 0.0005300594290766378,
      "loss": 2.3128,
      "step": 51107
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7704734802246094,
      "learning_rate": 0.0005300568036967758,
      "loss": 2.3582,
      "step": 51108
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.6560814380645752,
      "learning_rate": 0.0005300541782741422,
      "loss": 2.3081,
      "step": 51109
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7645260691642761,
      "learning_rate": 0.000530051552808737,
      "loss": 2.2037,
      "step": 51110
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7477945685386658,
      "learning_rate": 0.0005300489273005609,
      "loss": 2.1671,
      "step": 51111
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7405162453651428,
      "learning_rate": 0.0005300463017496143,
      "loss": 2.4978,
      "step": 51112
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.719632089138031,
      "learning_rate": 0.0005300436761558979,
      "loss": 2.2887,
      "step": 51113
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7028009295463562,
      "learning_rate": 0.000530041050519412,
      "loss": 2.1486,
      "step": 51114
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.6637985110282898,
      "learning_rate": 0.0005300384248401571,
      "loss": 2.106,
      "step": 51115
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.6597945094108582,
      "learning_rate": 0.0005300357991181337,
      "loss": 2.5084,
      "step": 51116
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7218207120895386,
      "learning_rate": 0.0005300331733533423,
      "loss": 2.2467,
      "step": 51117
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7664011120796204,
      "learning_rate": 0.0005300305475457834,
      "loss": 2.3633,
      "step": 51118
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.6766185760498047,
      "learning_rate": 0.0005300279216954576,
      "loss": 2.2639,
      "step": 51119
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.6767779588699341,
      "learning_rate": 0.0005300252958023651,
      "loss": 2.1512,
      "step": 51120
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7370096445083618,
      "learning_rate": 0.0005300226698665066,
      "loss": 1.9787,
      "step": 51121
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7080309391021729,
      "learning_rate": 0.0005300200438878826,
      "loss": 2.2251,
      "step": 51122
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7441615462303162,
      "learning_rate": 0.0005300174178664934,
      "loss": 2.4035,
      "step": 51123
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7140220403671265,
      "learning_rate": 0.0005300147918023397,
      "loss": 2.1853,
      "step": 51124
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7040667533874512,
      "learning_rate": 0.000530012165695422,
      "loss": 2.0983,
      "step": 51125
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7727128863334656,
      "learning_rate": 0.0005300095395457405,
      "loss": 2.3345,
      "step": 51126
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7112960815429688,
      "learning_rate": 0.0005300069133532961,
      "loss": 2.1743,
      "step": 51127
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7083354592323303,
      "learning_rate": 0.0005300042871180889,
      "loss": 2.1376,
      "step": 51128
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.6219176054000854,
      "learning_rate": 0.0005300016608401196,
      "loss": 2.2407,
      "step": 51129
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.6831967234611511,
      "learning_rate": 0.0005299990345193887,
      "loss": 2.0541,
      "step": 51130
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.6444304585456848,
      "learning_rate": 0.0005299964081558966,
      "loss": 1.9393,
      "step": 51131
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.8067894577980042,
      "learning_rate": 0.0005299937817496438,
      "loss": 2.2202,
      "step": 51132
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.8939506411552429,
      "learning_rate": 0.0005299911553006309,
      "loss": 2.2976,
      "step": 51133
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.6537704467773438,
      "learning_rate": 0.0005299885288088583,
      "loss": 2.2683,
      "step": 51134
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7355934977531433,
      "learning_rate": 0.0005299859022743264,
      "loss": 2.203,
      "step": 51135
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.6806763410568237,
      "learning_rate": 0.0005299832756970359,
      "loss": 2.2004,
      "step": 51136
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7652040123939514,
      "learning_rate": 0.000529980649076987,
      "loss": 2.1098,
      "step": 51137
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7301216125488281,
      "learning_rate": 0.0005299780224141805,
      "loss": 2.1501,
      "step": 51138
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7370928525924683,
      "learning_rate": 0.0005299753957086167,
      "loss": 2.1613,
      "step": 51139
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7115600109100342,
      "learning_rate": 0.0005299727689602961,
      "loss": 2.1401,
      "step": 51140
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.6651456356048584,
      "learning_rate": 0.0005299701421692193,
      "loss": 2.425,
      "step": 51141
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.8289461135864258,
      "learning_rate": 0.0005299675153353866,
      "loss": 2.2345,
      "step": 51142
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.712020993232727,
      "learning_rate": 0.0005299648884587987,
      "loss": 2.2453,
      "step": 51143
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7404515743255615,
      "learning_rate": 0.0005299622615394559,
      "loss": 2.1097,
      "step": 51144
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.6956010460853577,
      "learning_rate": 0.0005299596345773589,
      "loss": 2.2155,
      "step": 51145
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7485642433166504,
      "learning_rate": 0.0005299570075725079,
      "loss": 2.384,
      "step": 51146
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7080000638961792,
      "learning_rate": 0.0005299543805249037,
      "loss": 2.138,
      "step": 51147
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7312002182006836,
      "learning_rate": 0.0005299517534345466,
      "loss": 2.259,
      "step": 51148
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7462464570999146,
      "learning_rate": 0.000529949126301437,
      "loss": 2.2323,
      "step": 51149
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.8302463889122009,
      "learning_rate": 0.0005299464991255756,
      "loss": 2.0892,
      "step": 51150
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7671220898628235,
      "learning_rate": 0.0005299438719069629,
      "loss": 2.2437,
      "step": 51151
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7161979675292969,
      "learning_rate": 0.0005299412446455992,
      "loss": 2.0935,
      "step": 51152
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.0101810693740845,
      "learning_rate": 0.0005299386173414851,
      "loss": 2.2741,
      "step": 51153
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7016957402229309,
      "learning_rate": 0.0005299359899946212,
      "loss": 2.0284,
      "step": 51154
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7057169079780579,
      "learning_rate": 0.0005299333626050077,
      "loss": 2.3062,
      "step": 51155
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.687487781047821,
      "learning_rate": 0.0005299307351726453,
      "loss": 2.2868,
      "step": 51156
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.768644392490387,
      "learning_rate": 0.0005299281076975343,
      "loss": 2.2084,
      "step": 51157
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.781650960445404,
      "learning_rate": 0.0005299254801796754,
      "loss": 2.2023,
      "step": 51158
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7190216183662415,
      "learning_rate": 0.0005299228526190692,
      "loss": 2.1843,
      "step": 51159
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7296572923660278,
      "learning_rate": 0.0005299202250157158,
      "loss": 2.2363,
      "step": 51160
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7164276242256165,
      "learning_rate": 0.000529917597369616,
      "loss": 2.2875,
      "step": 51161
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.6726343035697937,
      "learning_rate": 0.0005299149696807701,
      "loss": 2.2191,
      "step": 51162
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7806500196456909,
      "learning_rate": 0.0005299123419491786,
      "loss": 2.2277,
      "step": 51163
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7367382645606995,
      "learning_rate": 0.0005299097141748422,
      "loss": 2.2013,
      "step": 51164
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.6400642991065979,
      "learning_rate": 0.0005299070863577613,
      "loss": 2.3743,
      "step": 51165
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7027385830879211,
      "learning_rate": 0.0005299044584979361,
      "loss": 2.7931,
      "step": 51166
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7122973203659058,
      "learning_rate": 0.0005299018305953675,
      "loss": 2.2175,
      "step": 51167
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.6879674196243286,
      "learning_rate": 0.0005298992026500557,
      "loss": 2.3935,
      "step": 51168
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7236895561218262,
      "learning_rate": 0.0005298965746620013,
      "loss": 2.2495,
      "step": 51169
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.6877821683883667,
      "learning_rate": 0.0005298939466312049,
      "loss": 2.4168,
      "step": 51170
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7104161381721497,
      "learning_rate": 0.0005298913185576668,
      "loss": 2.3602,
      "step": 51171
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.6715335845947266,
      "learning_rate": 0.0005298886904413875,
      "loss": 2.2286,
      "step": 51172
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.716458261013031,
      "learning_rate": 0.0005298860622823678,
      "loss": 2.3571,
      "step": 51173
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.8189069628715515,
      "learning_rate": 0.0005298834340806077,
      "loss": 2.0346,
      "step": 51174
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7052198648452759,
      "learning_rate": 0.000529880805836108,
      "loss": 2.1807,
      "step": 51175
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.8857327103614807,
      "learning_rate": 0.0005298781775488692,
      "loss": 2.1976,
      "step": 51176
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.8019862174987793,
      "learning_rate": 0.0005298755492188916,
      "loss": 2.3897,
      "step": 51177
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.6694722175598145,
      "learning_rate": 0.0005298729208461757,
      "loss": 2.3184,
      "step": 51178
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.8326345682144165,
      "learning_rate": 0.0005298702924307223,
      "loss": 2.282,
      "step": 51179
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7658780813217163,
      "learning_rate": 0.0005298676639725315,
      "loss": 2.151,
      "step": 51180
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7299163341522217,
      "learning_rate": 0.0005298650354716042,
      "loss": 2.3869,
      "step": 51181
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7291381359100342,
      "learning_rate": 0.0005298624069279403,
      "loss": 2.4983,
      "step": 51182
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7369129061698914,
      "learning_rate": 0.000529859778341541,
      "loss": 2.2606,
      "step": 51183
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7725565433502197,
      "learning_rate": 0.0005298571497124062,
      "loss": 2.1552,
      "step": 51184
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.8203958868980408,
      "learning_rate": 0.0005298545210405367,
      "loss": 2.2514,
      "step": 51185
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7274928092956543,
      "learning_rate": 0.0005298518923259329,
      "loss": 2.5097,
      "step": 51186
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.8424025177955627,
      "learning_rate": 0.0005298492635685952,
      "loss": 2.3032,
      "step": 51187
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7095208168029785,
      "learning_rate": 0.0005298466347685244,
      "loss": 2.4518,
      "step": 51188
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7649399042129517,
      "learning_rate": 0.0005298440059257205,
      "loss": 2.2613,
      "step": 51189
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7137922644615173,
      "learning_rate": 0.0005298413770401845,
      "loss": 2.3438,
      "step": 51190
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7087393999099731,
      "learning_rate": 0.0005298387481119165,
      "loss": 2.2004,
      "step": 51191
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.6816422939300537,
      "learning_rate": 0.0005298361191409172,
      "loss": 2.3206,
      "step": 51192
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7297825217247009,
      "learning_rate": 0.000529833490127187,
      "loss": 2.3702,
      "step": 51193
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7920147180557251,
      "learning_rate": 0.0005298308610707265,
      "loss": 2.3671,
      "step": 51194
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7291107177734375,
      "learning_rate": 0.000529828231971536,
      "loss": 2.2985,
      "step": 51195
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.6926441192626953,
      "learning_rate": 0.0005298256028296161,
      "loss": 2.1752,
      "step": 51196
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.8207092881202698,
      "learning_rate": 0.0005298229736449673,
      "loss": 2.2519,
      "step": 51197
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7803025245666504,
      "learning_rate": 0.0005298203444175901,
      "loss": 2.4454,
      "step": 51198
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.6905470490455627,
      "learning_rate": 0.0005298177151474849,
      "loss": 2.1725,
      "step": 51199
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.8226040005683899,
      "learning_rate": 0.0005298150858346524,
      "loss": 2.3681,
      "step": 51200
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7344483137130737,
      "learning_rate": 0.0005298124564790928,
      "loss": 2.0509,
      "step": 51201
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.8278446197509766,
      "learning_rate": 0.0005298098270808067,
      "loss": 2.2527,
      "step": 51202
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.699100136756897,
      "learning_rate": 0.0005298071976397947,
      "loss": 2.3112,
      "step": 51203
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7038015127182007,
      "learning_rate": 0.0005298045681560571,
      "loss": 2.1707,
      "step": 51204
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7012772560119629,
      "learning_rate": 0.0005298019386295946,
      "loss": 2.4976,
      "step": 51205
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.8453276753425598,
      "learning_rate": 0.0005297993090604077,
      "loss": 2.1884,
      "step": 51206
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7383925318717957,
      "learning_rate": 0.0005297966794484965,
      "loss": 2.4834,
      "step": 51207
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7403895854949951,
      "learning_rate": 0.0005297940497938618,
      "loss": 2.2422,
      "step": 51208
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.8853418827056885,
      "learning_rate": 0.0005297914200965042,
      "loss": 2.2789,
      "step": 51209
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7236633896827698,
      "learning_rate": 0.0005297887903564239,
      "loss": 2.4554,
      "step": 51210
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.696976900100708,
      "learning_rate": 0.0005297861605736216,
      "loss": 2.2763,
      "step": 51211
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.706619918346405,
      "learning_rate": 0.0005297835307480977,
      "loss": 2.4249,
      "step": 51212
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.70441073179245,
      "learning_rate": 0.0005297809008798527,
      "loss": 1.9841,
      "step": 51213
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7008700370788574,
      "learning_rate": 0.0005297782709688871,
      "loss": 2.1397,
      "step": 51214
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7113327980041504,
      "learning_rate": 0.0005297756410152014,
      "loss": 2.1905,
      "step": 51215
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.8216608166694641,
      "learning_rate": 0.0005297730110187961,
      "loss": 2.1032,
      "step": 51216
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.6964347958564758,
      "learning_rate": 0.0005297703809796716,
      "loss": 2.1232,
      "step": 51217
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.8538691401481628,
      "learning_rate": 0.0005297677508978285,
      "loss": 2.2655,
      "step": 51218
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7347792983055115,
      "learning_rate": 0.0005297651207732672,
      "loss": 2.3349,
      "step": 51219
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7298469543457031,
      "learning_rate": 0.0005297624906059883,
      "loss": 2.3334,
      "step": 51220
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.6833609342575073,
      "learning_rate": 0.000529759860395992,
      "loss": 2.247,
      "step": 51221
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.8478155732154846,
      "learning_rate": 0.0005297572301432792,
      "loss": 2.2203,
      "step": 51222
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.769589900970459,
      "learning_rate": 0.0005297545998478501,
      "loss": 2.3463,
      "step": 51223
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7205913066864014,
      "learning_rate": 0.0005297519695097053,
      "loss": 2.2559,
      "step": 51224
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7071001529693604,
      "learning_rate": 0.0005297493391288453,
      "loss": 2.1967,
      "step": 51225
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7219024300575256,
      "learning_rate": 0.0005297467087052706,
      "loss": 2.1111,
      "step": 51226
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7001063227653503,
      "learning_rate": 0.0005297440782389817,
      "loss": 2.2958,
      "step": 51227
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.8068609833717346,
      "learning_rate": 0.0005297414477299789,
      "loss": 2.4199,
      "step": 51228
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.6658817529678345,
      "learning_rate": 0.0005297388171782629,
      "loss": 2.1576,
      "step": 51229
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7196810245513916,
      "learning_rate": 0.000529736186583834,
      "loss": 2.3075,
      "step": 51230
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7247743010520935,
      "learning_rate": 0.000529733555946693,
      "loss": 2.3018,
      "step": 51231
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.6952199935913086,
      "learning_rate": 0.00052973092526684,
      "loss": 2.2387,
      "step": 51232
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7061208486557007,
      "learning_rate": 0.0005297282945442758,
      "loss": 2.0717,
      "step": 51233
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.713904619216919,
      "learning_rate": 0.0005297256637790007,
      "loss": 2.1987,
      "step": 51234
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7872617244720459,
      "learning_rate": 0.0005297230329710154,
      "loss": 2.4387,
      "step": 51235
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.6945088505744934,
      "learning_rate": 0.0005297204021203201,
      "loss": 2.3875,
      "step": 51236
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7604549527168274,
      "learning_rate": 0.0005297177712269155,
      "loss": 2.2719,
      "step": 51237
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.6872361898422241,
      "learning_rate": 0.0005297151402908021,
      "loss": 2.1272,
      "step": 51238
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7227845191955566,
      "learning_rate": 0.0005297125093119803,
      "loss": 2.2028,
      "step": 51239
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7940749526023865,
      "learning_rate": 0.0005297098782904505,
      "loss": 2.1733,
      "step": 51240
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7485359311103821,
      "learning_rate": 0.0005297072472262134,
      "loss": 2.1179,
      "step": 51241
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7491515278816223,
      "learning_rate": 0.0005297046161192693,
      "loss": 2.2717,
      "step": 51242
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7487586140632629,
      "learning_rate": 0.0005297019849696188,
      "loss": 2.1655,
      "step": 51243
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.6833601593971252,
      "learning_rate": 0.0005296993537772625,
      "loss": 2.2189,
      "step": 51244
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7396973967552185,
      "learning_rate": 0.0005296967225422005,
      "loss": 2.3603,
      "step": 51245
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.8432292938232422,
      "learning_rate": 0.0005296940912644338,
      "loss": 2.271,
      "step": 51246
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.6940937638282776,
      "learning_rate": 0.0005296914599439625,
      "loss": 2.3322,
      "step": 51247
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7358604073524475,
      "learning_rate": 0.0005296888285807872,
      "loss": 2.2435,
      "step": 51248
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.1215481758117676,
      "learning_rate": 0.0005296861971749086,
      "loss": 2.3914,
      "step": 51249
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7063992023468018,
      "learning_rate": 0.0005296835657263268,
      "loss": 2.2774,
      "step": 51250
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.9968239665031433,
      "learning_rate": 0.0005296809342350424,
      "loss": 2.2417,
      "step": 51251
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.6867460608482361,
      "learning_rate": 0.0005296783027010562,
      "loss": 2.2499,
      "step": 51252
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7757033705711365,
      "learning_rate": 0.0005296756711243684,
      "loss": 2.3829,
      "step": 51253
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7514273524284363,
      "learning_rate": 0.0005296730395049795,
      "loss": 2.2871,
      "step": 51254
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7132158279418945,
      "learning_rate": 0.0005296704078428902,
      "loss": 2.2573,
      "step": 51255
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7403035163879395,
      "learning_rate": 0.0005296677761381008,
      "loss": 2.5433,
      "step": 51256
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7466885447502136,
      "learning_rate": 0.0005296651443906116,
      "loss": 2.4455,
      "step": 51257
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7966391444206238,
      "learning_rate": 0.0005296625126004236,
      "loss": 2.1659,
      "step": 51258
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.718117892742157,
      "learning_rate": 0.0005296598807675368,
      "loss": 2.082,
      "step": 51259
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7715746760368347,
      "learning_rate": 0.000529657248891952,
      "loss": 2.3592,
      "step": 51260
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7467479705810547,
      "learning_rate": 0.0005296546169736695,
      "loss": 2.3199,
      "step": 51261
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7704089879989624,
      "learning_rate": 0.0005296519850126898,
      "loss": 2.2757,
      "step": 51262
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.748109757900238,
      "learning_rate": 0.0005296493530090136,
      "loss": 2.1621,
      "step": 51263
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7567132711410522,
      "learning_rate": 0.0005296467209626413,
      "loss": 2.3282,
      "step": 51264
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7287124395370483,
      "learning_rate": 0.0005296440888735733,
      "loss": 2.49,
      "step": 51265
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.9226233959197998,
      "learning_rate": 0.0005296414567418101,
      "loss": 2.2884,
      "step": 51266
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.6882290840148926,
      "learning_rate": 0.0005296388245673523,
      "loss": 2.1391,
      "step": 51267
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7272793650627136,
      "learning_rate": 0.0005296361923502001,
      "loss": 2.2905,
      "step": 51268
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7538869380950928,
      "learning_rate": 0.0005296335600903544,
      "loss": 2.4628,
      "step": 51269
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.6913724541664124,
      "learning_rate": 0.0005296309277878154,
      "loss": 2.2073,
      "step": 51270
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.6818656921386719,
      "learning_rate": 0.0005296282954425837,
      "loss": 2.1197,
      "step": 51271
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7009766101837158,
      "learning_rate": 0.0005296256630546597,
      "loss": 2.1181,
      "step": 51272
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7367305755615234,
      "learning_rate": 0.000529623030624044,
      "loss": 2.1168,
      "step": 51273
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7442687749862671,
      "learning_rate": 0.0005296203981507371,
      "loss": 2.33,
      "step": 51274
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7488729953765869,
      "learning_rate": 0.0005296177656347393,
      "loss": 2.2142,
      "step": 51275
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.75734943151474,
      "learning_rate": 0.0005296151330760514,
      "loss": 2.3656,
      "step": 51276
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7489705681800842,
      "learning_rate": 0.0005296125004746736,
      "loss": 2.317,
      "step": 51277
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7632216811180115,
      "learning_rate": 0.0005296098678306065,
      "loss": 2.3695,
      "step": 51278
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7052574157714844,
      "learning_rate": 0.0005296072351438507,
      "loss": 2.4149,
      "step": 51279
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7116189002990723,
      "learning_rate": 0.0005296046024144064,
      "loss": 2.2999,
      "step": 51280
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.6856078505516052,
      "learning_rate": 0.0005296019696422743,
      "loss": 2.3287,
      "step": 51281
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.69376540184021,
      "learning_rate": 0.000529599336827455,
      "loss": 2.3843,
      "step": 51282
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7593057751655579,
      "learning_rate": 0.0005295967039699488,
      "loss": 2.4605,
      "step": 51283
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.66279137134552,
      "learning_rate": 0.0005295940710697562,
      "loss": 2.1625,
      "step": 51284
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.8343583345413208,
      "learning_rate": 0.0005295914381268778,
      "loss": 2.1767,
      "step": 51285
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.6568554639816284,
      "learning_rate": 0.0005295888051413139,
      "loss": 2.4129,
      "step": 51286
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7199745178222656,
      "learning_rate": 0.0005295861721130652,
      "loss": 2.2412,
      "step": 51287
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.838650643825531,
      "learning_rate": 0.000529583539042132,
      "loss": 2.0815,
      "step": 51288
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7399747371673584,
      "learning_rate": 0.000529580905928515,
      "loss": 2.1221,
      "step": 51289
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.8003239035606384,
      "learning_rate": 0.0005295782727722145,
      "loss": 2.2523,
      "step": 51290
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.9297022223472595,
      "learning_rate": 0.0005295756395732311,
      "loss": 2.2743,
      "step": 51291
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.655348539352417,
      "learning_rate": 0.0005295730063315652,
      "loss": 2.2904,
      "step": 51292
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7361499667167664,
      "learning_rate": 0.0005295703730472174,
      "loss": 2.436,
      "step": 51293
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.9830410480499268,
      "learning_rate": 0.0005295677397201881,
      "loss": 2.4507,
      "step": 51294
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7729963064193726,
      "learning_rate": 0.0005295651063504779,
      "loss": 2.2195,
      "step": 51295
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.66204434633255,
      "learning_rate": 0.000529562472938087,
      "loss": 2.1441,
      "step": 51296
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7174729108810425,
      "learning_rate": 0.0005295598394830164,
      "loss": 2.0643,
      "step": 51297
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.76985764503479,
      "learning_rate": 0.0005295572059852661,
      "loss": 2.122,
      "step": 51298
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.6966336369514465,
      "learning_rate": 0.0005295545724448369,
      "loss": 2.1505,
      "step": 51299
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7405824065208435,
      "learning_rate": 0.0005295519388617291,
      "loss": 2.2603,
      "step": 51300
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7593274712562561,
      "learning_rate": 0.0005295493052359432,
      "loss": 2.372,
      "step": 51301
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.2162741422653198,
      "learning_rate": 0.0005295466715674799,
      "loss": 2.3091,
      "step": 51302
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.732586145401001,
      "learning_rate": 0.0005295440378563394,
      "loss": 2.1092,
      "step": 51303
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.8072100877761841,
      "learning_rate": 0.0005295414041025224,
      "loss": 2.1405,
      "step": 51304
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.8891978859901428,
      "learning_rate": 0.0005295387703060293,
      "loss": 2.417,
      "step": 51305
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7403767108917236,
      "learning_rate": 0.0005295361364668608,
      "loss": 2.1355,
      "step": 51306
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7645635604858398,
      "learning_rate": 0.0005295335025850169,
      "loss": 2.2359,
      "step": 51307
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.9011178612709045,
      "learning_rate": 0.0005295308686604985,
      "loss": 2.0941,
      "step": 51308
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.8579666614532471,
      "learning_rate": 0.000529528234693306,
      "loss": 2.2472,
      "step": 51309
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7435696125030518,
      "learning_rate": 0.0005295256006834399,
      "loss": 2.4731,
      "step": 51310
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.9430553913116455,
      "learning_rate": 0.0005295229666309005,
      "loss": 2.3026,
      "step": 51311
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.8035246729850769,
      "learning_rate": 0.0005295203325356887,
      "loss": 2.3946,
      "step": 51312
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.6560527086257935,
      "learning_rate": 0.0005295176983978045,
      "loss": 2.2056,
      "step": 51313
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.8191184401512146,
      "learning_rate": 0.0005295150642172489,
      "loss": 2.3071,
      "step": 51314
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.0442410707473755,
      "learning_rate": 0.0005295124299940218,
      "loss": 2.1784,
      "step": 51315
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.8042902946472168,
      "learning_rate": 0.0005295097957281242,
      "loss": 2.2924,
      "step": 51316
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.8353267908096313,
      "learning_rate": 0.0005295071614195562,
      "loss": 2.1492,
      "step": 51317
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.0419981479644775,
      "learning_rate": 0.0005295045270683188,
      "loss": 2.2011,
      "step": 51318
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.9812262058258057,
      "learning_rate": 0.000529501892674412,
      "loss": 2.3382,
      "step": 51319
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7201859354972839,
      "learning_rate": 0.0005294992582378364,
      "loss": 2.2763,
      "step": 51320
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.8865458369255066,
      "learning_rate": 0.0005294966237585927,
      "loss": 2.3877,
      "step": 51321
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.2261600494384766,
      "learning_rate": 0.0005294939892366812,
      "loss": 2.3728,
      "step": 51322
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7362653613090515,
      "learning_rate": 0.0005294913546721024,
      "loss": 2.4028,
      "step": 51323
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.756862461566925,
      "learning_rate": 0.0005294887200648569,
      "loss": 2.1199,
      "step": 51324
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7683615684509277,
      "learning_rate": 0.000529486085414945,
      "loss": 2.358,
      "step": 51325
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.8904041051864624,
      "learning_rate": 0.0005294834507223673,
      "loss": 2.2077,
      "step": 51326
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.6958092451095581,
      "learning_rate": 0.0005294808159871244,
      "loss": 2.413,
      "step": 51327
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.8324536681175232,
      "learning_rate": 0.0005294781812092167,
      "loss": 2.2498,
      "step": 51328
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7994133830070496,
      "learning_rate": 0.0005294755463886446,
      "loss": 2.3662,
      "step": 51329
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7376834154129028,
      "learning_rate": 0.0005294729115254087,
      "loss": 2.3147,
      "step": 51330
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.6828598380088806,
      "learning_rate": 0.0005294702766195094,
      "loss": 2.2095,
      "step": 51331
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7091315984725952,
      "learning_rate": 0.0005294676416709472,
      "loss": 2.319,
      "step": 51332
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7340081334114075,
      "learning_rate": 0.0005294650066797228,
      "loss": 2.3343,
      "step": 51333
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.6645606160163879,
      "learning_rate": 0.0005294623716458364,
      "loss": 2.3232,
      "step": 51334
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7416062951087952,
      "learning_rate": 0.0005294597365692887,
      "loss": 2.2576,
      "step": 51335
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.6823009848594666,
      "learning_rate": 0.00052945710145008,
      "loss": 2.0118,
      "step": 51336
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.70500248670578,
      "learning_rate": 0.0005294544662882109,
      "loss": 2.3848,
      "step": 51337
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.6934147477149963,
      "learning_rate": 0.0005294518310836819,
      "loss": 2.2562,
      "step": 51338
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.6880940198898315,
      "learning_rate": 0.0005294491958364936,
      "loss": 2.1784,
      "step": 51339
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.6644325256347656,
      "learning_rate": 0.0005294465605466462,
      "loss": 2.2283,
      "step": 51340
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.73529052734375,
      "learning_rate": 0.0005294439252141404,
      "loss": 2.314,
      "step": 51341
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.688114583492279,
      "learning_rate": 0.0005294412898389767,
      "loss": 2.3743,
      "step": 51342
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.6655483245849609,
      "learning_rate": 0.0005294386544211555,
      "loss": 1.9626,
      "step": 51343
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7524941563606262,
      "learning_rate": 0.0005294360189606773,
      "loss": 2.3582,
      "step": 51344
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7110999226570129,
      "learning_rate": 0.0005294333834575427,
      "loss": 2.0388,
      "step": 51345
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.8121650218963623,
      "learning_rate": 0.000529430747911752,
      "loss": 2.1412,
      "step": 51346
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7922027111053467,
      "learning_rate": 0.0005294281123233058,
      "loss": 2.0917,
      "step": 51347
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.6994771361351013,
      "learning_rate": 0.0005294254766922046,
      "loss": 2.2674,
      "step": 51348
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.6758944392204285,
      "learning_rate": 0.0005294228410184489,
      "loss": 2.5304,
      "step": 51349
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.8673185110092163,
      "learning_rate": 0.0005294202053020392,
      "loss": 2.2374,
      "step": 51350
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.749678373336792,
      "learning_rate": 0.0005294175695429758,
      "loss": 2.3725,
      "step": 51351
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.686253547668457,
      "learning_rate": 0.0005294149337412595,
      "loss": 2.1862,
      "step": 51352
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.6846049427986145,
      "learning_rate": 0.0005294122978968906,
      "loss": 2.0756,
      "step": 51353
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7786566019058228,
      "learning_rate": 0.0005294096620098696,
      "loss": 2.2384,
      "step": 51354
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.8138721585273743,
      "learning_rate": 0.000529407026080197,
      "loss": 2.2006,
      "step": 51355
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.6931883096694946,
      "learning_rate": 0.0005294043901078733,
      "loss": 2.2514,
      "step": 51356
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7812845706939697,
      "learning_rate": 0.000529401754092899,
      "loss": 2.2597,
      "step": 51357
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.9456343054771423,
      "learning_rate": 0.0005293991180352746,
      "loss": 2.0621,
      "step": 51358
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7991150617599487,
      "learning_rate": 0.0005293964819350006,
      "loss": 2.343,
      "step": 51359
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7906180620193481,
      "learning_rate": 0.0005293938457920774,
      "loss": 2.264,
      "step": 51360
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.892451822757721,
      "learning_rate": 0.0005293912096065056,
      "loss": 2.4241,
      "step": 51361
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.9606990814208984,
      "learning_rate": 0.0005293885733782856,
      "loss": 2.123,
      "step": 51362
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.653193473815918,
      "learning_rate": 0.000529385937107418,
      "loss": 2.1696,
      "step": 51363
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.4321383237838745,
      "learning_rate": 0.0005293833007939032,
      "loss": 2.094,
      "step": 51364
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.9137653708457947,
      "learning_rate": 0.0005293806644377417,
      "loss": 2.1943,
      "step": 51365
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7633247375488281,
      "learning_rate": 0.000529378028038934,
      "loss": 2.3585,
      "step": 51366
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.8892816305160522,
      "learning_rate": 0.0005293753915974806,
      "loss": 2.1738,
      "step": 51367
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7799375057220459,
      "learning_rate": 0.0005293727551133819,
      "loss": 2.1213,
      "step": 51368
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7577278017997742,
      "learning_rate": 0.0005293701185866387,
      "loss": 2.2922,
      "step": 51369
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.6876881718635559,
      "learning_rate": 0.0005293674820172511,
      "loss": 2.2636,
      "step": 51370
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7786436676979065,
      "learning_rate": 0.0005293648454052197,
      "loss": 2.2248,
      "step": 51371
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.8048564195632935,
      "learning_rate": 0.0005293622087505452,
      "loss": 2.5182,
      "step": 51372
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7158840298652649,
      "learning_rate": 0.0005293595720532279,
      "loss": 2.2808,
      "step": 51373
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.6938843131065369,
      "learning_rate": 0.0005293569353132682,
      "loss": 2.182,
      "step": 51374
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7132437825202942,
      "learning_rate": 0.0005293542985306668,
      "loss": 2.0676,
      "step": 51375
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7769439816474915,
      "learning_rate": 0.0005293516617054241,
      "loss": 2.1214,
      "step": 51376
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.6551626324653625,
      "learning_rate": 0.0005293490248375406,
      "loss": 2.1143,
      "step": 51377
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7065123915672302,
      "learning_rate": 0.0005293463879270168,
      "loss": 2.4197,
      "step": 51378
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.8027898669242859,
      "learning_rate": 0.0005293437509738532,
      "loss": 2.1331,
      "step": 51379
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7020999789237976,
      "learning_rate": 0.0005293411139780502,
      "loss": 2.3155,
      "step": 51380
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7745551466941833,
      "learning_rate": 0.0005293384769396084,
      "loss": 2.3956,
      "step": 51381
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.9992563724517822,
      "learning_rate": 0.0005293358398585283,
      "loss": 2.2895,
      "step": 51382
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.8014311790466309,
      "learning_rate": 0.0005293332027348103,
      "loss": 2.3923,
      "step": 51383
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7297154664993286,
      "learning_rate": 0.000529330565568455,
      "loss": 2.1207,
      "step": 51384
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7116358876228333,
      "learning_rate": 0.0005293279283594627,
      "loss": 2.4383,
      "step": 51385
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.8822799324989319,
      "learning_rate": 0.000529325291107834,
      "loss": 2.1366,
      "step": 51386
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.6895730495452881,
      "learning_rate": 0.0005293226538135694,
      "loss": 2.3395,
      "step": 51387
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.742865264415741,
      "learning_rate": 0.0005293200164766694,
      "loss": 2.0836,
      "step": 51388
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.9834943413734436,
      "learning_rate": 0.0005293173790971346,
      "loss": 2.1648,
      "step": 51389
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7798291444778442,
      "learning_rate": 0.0005293147416749653,
      "loss": 2.4393,
      "step": 51390
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7122181057929993,
      "learning_rate": 0.000529312104210162,
      "loss": 2.5583,
      "step": 51391
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7929888367652893,
      "learning_rate": 0.0005293094667027253,
      "loss": 2.421,
      "step": 51392
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7594298720359802,
      "learning_rate": 0.0005293068291526556,
      "loss": 2.2543,
      "step": 51393
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7013989686965942,
      "learning_rate": 0.0005293041915599535,
      "loss": 2.2545,
      "step": 51394
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.71841961145401,
      "learning_rate": 0.0005293015539246194,
      "loss": 2.0626,
      "step": 51395
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7249143719673157,
      "learning_rate": 0.0005292989162466537,
      "loss": 2.2162,
      "step": 51396
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7878748178482056,
      "learning_rate": 0.0005292962785260572,
      "loss": 2.2329,
      "step": 51397
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7860070466995239,
      "learning_rate": 0.0005292936407628301,
      "loss": 2.2302,
      "step": 51398
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.8736961483955383,
      "learning_rate": 0.000529291002956973,
      "loss": 2.1075,
      "step": 51399
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.698792576789856,
      "learning_rate": 0.0005292883651084864,
      "loss": 2.1382,
      "step": 51400
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.8199203014373779,
      "learning_rate": 0.0005292857272173706,
      "loss": 2.3953,
      "step": 51401
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.6781294941902161,
      "learning_rate": 0.0005292830892836264,
      "loss": 2.2832,
      "step": 51402
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.8821378350257874,
      "learning_rate": 0.0005292804513072541,
      "loss": 2.141,
      "step": 51403
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7278740406036377,
      "learning_rate": 0.0005292778132882543,
      "loss": 2.0495,
      "step": 51404
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.6686420440673828,
      "learning_rate": 0.0005292751752266274,
      "loss": 2.3235,
      "step": 51405
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7323868870735168,
      "learning_rate": 0.0005292725371223738,
      "loss": 2.2694,
      "step": 51406
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.6902598738670349,
      "learning_rate": 0.0005292698989754943,
      "loss": 2.3166,
      "step": 51407
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7604920268058777,
      "learning_rate": 0.0005292672607859891,
      "loss": 2.3201,
      "step": 51408
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7019479870796204,
      "learning_rate": 0.0005292646225538588,
      "loss": 2.3464,
      "step": 51409
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7096783518791199,
      "learning_rate": 0.0005292619842791038,
      "loss": 2.2815,
      "step": 51410
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7880446910858154,
      "learning_rate": 0.0005292593459617248,
      "loss": 2.2172,
      "step": 51411
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7399643659591675,
      "learning_rate": 0.0005292567076017221,
      "loss": 2.4384,
      "step": 51412
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7178963422775269,
      "learning_rate": 0.0005292540691990963,
      "loss": 2.1308,
      "step": 51413
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.8575270771980286,
      "learning_rate": 0.0005292514307538477,
      "loss": 2.3835,
      "step": 51414
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7649911046028137,
      "learning_rate": 0.0005292487922659772,
      "loss": 2.2488,
      "step": 51415
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7393965721130371,
      "learning_rate": 0.0005292461537354848,
      "loss": 2.3032,
      "step": 51416
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7195387482643127,
      "learning_rate": 0.0005292435151623712,
      "loss": 2.2609,
      "step": 51417
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7927114963531494,
      "learning_rate": 0.0005292408765466371,
      "loss": 2.1245,
      "step": 51418
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7403267025947571,
      "learning_rate": 0.0005292382378882825,
      "loss": 2.2964,
      "step": 51419
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7447972893714905,
      "learning_rate": 0.0005292355991873085,
      "loss": 2.0836,
      "step": 51420
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7834299206733704,
      "learning_rate": 0.000529232960443715,
      "loss": 2.3017,
      "step": 51421
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.6800662875175476,
      "learning_rate": 0.0005292303216575029,
      "loss": 2.2046,
      "step": 51422
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.6618945598602295,
      "learning_rate": 0.0005292276828286726,
      "loss": 2.0392,
      "step": 51423
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7783523201942444,
      "learning_rate": 0.0005292250439572245,
      "loss": 2.5398,
      "step": 51424
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.8352106213569641,
      "learning_rate": 0.000529222405043159,
      "loss": 2.2772,
      "step": 51425
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7501404285430908,
      "learning_rate": 0.000529219766086477,
      "loss": 2.2696,
      "step": 51426
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.8308668732643127,
      "learning_rate": 0.0005292171270871785,
      "loss": 2.3819,
      "step": 51427
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.9213682413101196,
      "learning_rate": 0.0005292144880452643,
      "loss": 2.2075,
      "step": 51428
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.6606788635253906,
      "learning_rate": 0.0005292118489607347,
      "loss": 2.1929,
      "step": 51429
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.6763544678688049,
      "learning_rate": 0.0005292092098335905,
      "loss": 2.3077,
      "step": 51430
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.920696496963501,
      "learning_rate": 0.0005292065706638318,
      "loss": 2.3405,
      "step": 51431
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7159145474433899,
      "learning_rate": 0.0005292039314514594,
      "loss": 2.1492,
      "step": 51432
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7645946741104126,
      "learning_rate": 0.0005292012921964735,
      "loss": 2.2911,
      "step": 51433
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7179237008094788,
      "learning_rate": 0.0005291986528988749,
      "loss": 2.1609,
      "step": 51434
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.6818777322769165,
      "learning_rate": 0.0005291960135586637,
      "loss": 2.3087,
      "step": 51435
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.774152398109436,
      "learning_rate": 0.000529193374175841,
      "loss": 2.4447,
      "step": 51436
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7153429985046387,
      "learning_rate": 0.0005291907347504067,
      "loss": 2.1693,
      "step": 51437
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7933000922203064,
      "learning_rate": 0.0005291880952823615,
      "loss": 2.3472,
      "step": 51438
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7879281044006348,
      "learning_rate": 0.000529185455771706,
      "loss": 2.1952,
      "step": 51439
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.783164918422699,
      "learning_rate": 0.0005291828162184406,
      "loss": 2.1018,
      "step": 51440
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.6994486451148987,
      "learning_rate": 0.0005291801766225657,
      "loss": 2.2956,
      "step": 51441
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7270699739456177,
      "learning_rate": 0.000529177536984082,
      "loss": 2.4445,
      "step": 51442
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.6349380016326904,
      "learning_rate": 0.0005291748973029898,
      "loss": 2.1521,
      "step": 51443
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7000855207443237,
      "learning_rate": 0.0005291722575792896,
      "loss": 2.2714,
      "step": 51444
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7465965747833252,
      "learning_rate": 0.0005291696178129821,
      "loss": 2.2355,
      "step": 51445
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7101718187332153,
      "learning_rate": 0.0005291669780040676,
      "loss": 2.2409,
      "step": 51446
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.6210951209068298,
      "learning_rate": 0.0005291643381525465,
      "loss": 2.4965,
      "step": 51447
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.6908970475196838,
      "learning_rate": 0.0005291616982584195,
      "loss": 2.2667,
      "step": 51448
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.691666305065155,
      "learning_rate": 0.000529159058321687,
      "loss": 2.2531,
      "step": 51449
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7065017819404602,
      "learning_rate": 0.0005291564183423496,
      "loss": 2.1241,
      "step": 51450
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7352162599563599,
      "learning_rate": 0.0005291537783204076,
      "loss": 2.1986,
      "step": 51451
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7639642357826233,
      "learning_rate": 0.0005291511382558616,
      "loss": 2.2242,
      "step": 51452
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.9141777753829956,
      "learning_rate": 0.0005291484981487121,
      "loss": 2.2303,
      "step": 51453
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7600030899047852,
      "learning_rate": 0.0005291458579989595,
      "loss": 2.4204,
      "step": 51454
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.8098056316375732,
      "learning_rate": 0.0005291432178066045,
      "loss": 2.2494,
      "step": 51455
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.8799250721931458,
      "learning_rate": 0.0005291405775716473,
      "loss": 2.3254,
      "step": 51456
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7368891835212708,
      "learning_rate": 0.0005291379372940886,
      "loss": 2.2283,
      "step": 51457
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7927060127258301,
      "learning_rate": 0.0005291352969739289,
      "loss": 2.0338,
      "step": 51458
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.9564686417579651,
      "learning_rate": 0.0005291326566111686,
      "loss": 2.2994,
      "step": 51459
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7197956442832947,
      "learning_rate": 0.0005291300162058082,
      "loss": 2.3103,
      "step": 51460
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7267300486564636,
      "learning_rate": 0.0005291273757578481,
      "loss": 2.1586,
      "step": 51461
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.708195149898529,
      "learning_rate": 0.0005291247352672891,
      "loss": 2.1468,
      "step": 51462
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7996042966842651,
      "learning_rate": 0.0005291220947341313,
      "loss": 2.2478,
      "step": 51463
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.694901704788208,
      "learning_rate": 0.0005291194541583755,
      "loss": 2.1063,
      "step": 51464
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7339311838150024,
      "learning_rate": 0.0005291168135400221,
      "loss": 1.9731,
      "step": 51465
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7441409230232239,
      "learning_rate": 0.0005291141728790713,
      "loss": 1.9715,
      "step": 51466
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7695857882499695,
      "learning_rate": 0.0005291115321755241,
      "loss": 2.2365,
      "step": 51467
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7614904642105103,
      "learning_rate": 0.0005291088914293806,
      "loss": 2.2013,
      "step": 51468
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7719453573226929,
      "learning_rate": 0.0005291062506406415,
      "loss": 2.2427,
      "step": 51469
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7334503531455994,
      "learning_rate": 0.0005291036098093073,
      "loss": 2.2808,
      "step": 51470
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.8553913831710815,
      "learning_rate": 0.0005291009689353783,
      "loss": 2.3485,
      "step": 51471
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.9084585905075073,
      "learning_rate": 0.0005290983280188551,
      "loss": 2.2416,
      "step": 51472
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.8469074368476868,
      "learning_rate": 0.0005290956870597384,
      "loss": 2.1382,
      "step": 51473
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.8328067064285278,
      "learning_rate": 0.0005290930460580282,
      "loss": 2.2223,
      "step": 51474
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7820221185684204,
      "learning_rate": 0.0005290904050137254,
      "loss": 2.2176,
      "step": 51475
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.6515762805938721,
      "learning_rate": 0.0005290877639268305,
      "loss": 2.1568,
      "step": 51476
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.6675957441329956,
      "learning_rate": 0.0005290851227973436,
      "loss": 2.2897,
      "step": 51477
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7583856582641602,
      "learning_rate": 0.0005290824816252655,
      "loss": 2.4236,
      "step": 51478
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.8316397070884705,
      "learning_rate": 0.0005290798404105969,
      "loss": 2.2109,
      "step": 51479
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.6628335118293762,
      "learning_rate": 0.0005290771991533378,
      "loss": 2.2033,
      "step": 51480
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7149176001548767,
      "learning_rate": 0.000529074557853489,
      "loss": 2.3141,
      "step": 51481
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.6992847323417664,
      "learning_rate": 0.0005290719165110509,
      "loss": 2.3132,
      "step": 51482
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7333486080169678,
      "learning_rate": 0.000529069275126024,
      "loss": 1.9452,
      "step": 51483
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.6788930892944336,
      "learning_rate": 0.0005290666336984088,
      "loss": 2.1903,
      "step": 51484
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7239118814468384,
      "learning_rate": 0.0005290639922282058,
      "loss": 2.053,
      "step": 51485
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7855795621871948,
      "learning_rate": 0.0005290613507154154,
      "loss": 2.2721,
      "step": 51486
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.6154519319534302,
      "learning_rate": 0.0005290587091600383,
      "loss": 2.2056,
      "step": 51487
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7164384722709656,
      "learning_rate": 0.0005290560675620748,
      "loss": 2.2733,
      "step": 51488
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7046866416931152,
      "learning_rate": 0.0005290534259215255,
      "loss": 2.0205,
      "step": 51489
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7028571963310242,
      "learning_rate": 0.0005290507842383908,
      "loss": 2.3187,
      "step": 51490
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7013031840324402,
      "learning_rate": 0.0005290481425126712,
      "loss": 2.4359,
      "step": 51491
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.6696464419364929,
      "learning_rate": 0.0005290455007443673,
      "loss": 2.1722,
      "step": 51492
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7401914000511169,
      "learning_rate": 0.0005290428589334794,
      "loss": 2.2619,
      "step": 51493
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.6789010167121887,
      "learning_rate": 0.0005290402170800083,
      "loss": 2.3308,
      "step": 51494
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.6733950972557068,
      "learning_rate": 0.0005290375751839542,
      "loss": 2.1347,
      "step": 51495
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.6950365304946899,
      "learning_rate": 0.0005290349332453177,
      "loss": 2.1631,
      "step": 51496
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.6775473952293396,
      "learning_rate": 0.0005290322912640992,
      "loss": 2.3785,
      "step": 51497
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.75005704164505,
      "learning_rate": 0.0005290296492402993,
      "loss": 2.2213,
      "step": 51498
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.8113293051719666,
      "learning_rate": 0.0005290270071739186,
      "loss": 2.0882,
      "step": 51499
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.8001631498336792,
      "learning_rate": 0.0005290243650649573,
      "loss": 2.3339,
      "step": 51500
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.8032782077789307,
      "learning_rate": 0.0005290217229134162,
      "loss": 2.3312,
      "step": 51501
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7460126876831055,
      "learning_rate": 0.0005290190807192954,
      "loss": 2.3416,
      "step": 51502
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7069923877716064,
      "learning_rate": 0.0005290164384825958,
      "loss": 2.1492,
      "step": 51503
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.8395641446113586,
      "learning_rate": 0.0005290137962033178,
      "loss": 2.3459,
      "step": 51504
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.9334272146224976,
      "learning_rate": 0.0005290111538814617,
      "loss": 2.0247,
      "step": 51505
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7742957472801208,
      "learning_rate": 0.0005290085115170281,
      "loss": 2.3857,
      "step": 51506
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7704402208328247,
      "learning_rate": 0.0005290058691100175,
      "loss": 2.195,
      "step": 51507
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7691521048545837,
      "learning_rate": 0.0005290032266604304,
      "loss": 2.2514,
      "step": 51508
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.8297034502029419,
      "learning_rate": 0.0005290005841682673,
      "loss": 2.1516,
      "step": 51509
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.6864030957221985,
      "learning_rate": 0.0005289979416335287,
      "loss": 2.2438,
      "step": 51510
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.8450388312339783,
      "learning_rate": 0.0005289952990562149,
      "loss": 2.2915,
      "step": 51511
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.6692426800727844,
      "learning_rate": 0.0005289926564363267,
      "loss": 2.1033,
      "step": 51512
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7048913836479187,
      "learning_rate": 0.0005289900137738644,
      "loss": 2.1152,
      "step": 51513
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7850584983825684,
      "learning_rate": 0.0005289873710688285,
      "loss": 2.3231,
      "step": 51514
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.6297246813774109,
      "learning_rate": 0.0005289847283212196,
      "loss": 2.2062,
      "step": 51515
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.6971607804298401,
      "learning_rate": 0.0005289820855310381,
      "loss": 2.0504,
      "step": 51516
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7281204462051392,
      "learning_rate": 0.0005289794426982844,
      "loss": 2.2209,
      "step": 51517
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7246126532554626,
      "learning_rate": 0.0005289767998229592,
      "loss": 2.2916,
      "step": 51518
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.6634280681610107,
      "learning_rate": 0.0005289741569050629,
      "loss": 2.3537,
      "step": 51519
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7299849987030029,
      "learning_rate": 0.0005289715139445959,
      "loss": 2.1837,
      "step": 51520
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.6921125650405884,
      "learning_rate": 0.0005289688709415589,
      "loss": 2.4351,
      "step": 51521
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.707646369934082,
      "learning_rate": 0.0005289662278959522,
      "loss": 2.255,
      "step": 51522
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7834100127220154,
      "learning_rate": 0.0005289635848077762,
      "loss": 2.1865,
      "step": 51523
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.711770236492157,
      "learning_rate": 0.0005289609416770318,
      "loss": 2.4347,
      "step": 51524
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7193108797073364,
      "learning_rate": 0.0005289582985037191,
      "loss": 2.2388,
      "step": 51525
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7245481610298157,
      "learning_rate": 0.0005289556552878388,
      "loss": 2.3286,
      "step": 51526
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.69638991355896,
      "learning_rate": 0.0005289530120293912,
      "loss": 2.2675,
      "step": 51527
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7853915691375732,
      "learning_rate": 0.0005289503687283771,
      "loss": 2.4227,
      "step": 51528
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7653801441192627,
      "learning_rate": 0.0005289477253847967,
      "loss": 2.2489,
      "step": 51529
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.6942907571792603,
      "learning_rate": 0.0005289450819986507,
      "loss": 2.3402,
      "step": 51530
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7475093007087708,
      "learning_rate": 0.0005289424385699393,
      "loss": 2.1916,
      "step": 51531
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7161952257156372,
      "learning_rate": 0.0005289397950986632,
      "loss": 2.2888,
      "step": 51532
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7319831252098083,
      "learning_rate": 0.000528937151584823,
      "loss": 2.2072,
      "step": 51533
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7974867820739746,
      "learning_rate": 0.0005289345080284189,
      "loss": 2.3667,
      "step": 51534
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7366024851799011,
      "learning_rate": 0.0005289318644294517,
      "loss": 1.8639,
      "step": 51535
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7521966099739075,
      "learning_rate": 0.0005289292207879217,
      "loss": 2.2028,
      "step": 51536
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.6986535787582397,
      "learning_rate": 0.0005289265771038294,
      "loss": 2.3516,
      "step": 51537
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.9621132016181946,
      "learning_rate": 0.0005289239333771754,
      "loss": 2.2584,
      "step": 51538
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7941729426383972,
      "learning_rate": 0.00052892128960796,
      "loss": 2.3885,
      "step": 51539
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.8443976640701294,
      "learning_rate": 0.000528918645796184,
      "loss": 2.2443,
      "step": 51540
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.8045872449874878,
      "learning_rate": 0.0005289160019418476,
      "loss": 2.2787,
      "step": 51541
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.805076539516449,
      "learning_rate": 0.0005289133580449513,
      "loss": 2.1886,
      "step": 51542
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.6752035617828369,
      "learning_rate": 0.0005289107141054958,
      "loss": 2.2623,
      "step": 51543
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.6836459636688232,
      "learning_rate": 0.0005289080701234815,
      "loss": 2.4197,
      "step": 51544
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.8178463578224182,
      "learning_rate": 0.0005289054260989087,
      "loss": 2.1936,
      "step": 51545
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.9269514083862305,
      "learning_rate": 0.0005289027820317784,
      "loss": 2.228,
      "step": 51546
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7317355871200562,
      "learning_rate": 0.0005289001379220904,
      "loss": 2.0576,
      "step": 51547
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.8029959201812744,
      "learning_rate": 0.0005288974937698457,
      "loss": 2.4141,
      "step": 51548
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7945653796195984,
      "learning_rate": 0.0005288948495750446,
      "loss": 2.3125,
      "step": 51549
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.9171550273895264,
      "learning_rate": 0.0005288922053376877,
      "loss": 2.2997,
      "step": 51550
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7677806615829468,
      "learning_rate": 0.0005288895610577753,
      "loss": 2.2464,
      "step": 51551
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.9514254331588745,
      "learning_rate": 0.0005288869167353081,
      "loss": 2.124,
      "step": 51552
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.8086950182914734,
      "learning_rate": 0.0005288842723702865,
      "loss": 2.5815,
      "step": 51553
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7920604944229126,
      "learning_rate": 0.0005288816279627109,
      "loss": 2.2545,
      "step": 51554
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7238149642944336,
      "learning_rate": 0.0005288789835125819,
      "loss": 2.358,
      "step": 51555
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.816327691078186,
      "learning_rate": 0.0005288763390199,
      "loss": 2.5205,
      "step": 51556
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.6688929796218872,
      "learning_rate": 0.0005288736944846658,
      "loss": 2.2922,
      "step": 51557
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.6967716813087463,
      "learning_rate": 0.0005288710499068796,
      "loss": 2.0237,
      "step": 51558
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.6507314443588257,
      "learning_rate": 0.0005288684052865419,
      "loss": 2.1959,
      "step": 51559
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7517080307006836,
      "learning_rate": 0.0005288657606236531,
      "loss": 2.3861,
      "step": 51560
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.655076801776886,
      "learning_rate": 0.000528863115918214,
      "loss": 2.2002,
      "step": 51561
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7019137740135193,
      "learning_rate": 0.0005288604711702249,
      "loss": 2.1324,
      "step": 51562
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.6859474182128906,
      "learning_rate": 0.0005288578263796863,
      "loss": 2.4838,
      "step": 51563
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.6555172204971313,
      "learning_rate": 0.0005288551815465989,
      "loss": 2.2477,
      "step": 51564
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.6791688203811646,
      "learning_rate": 0.0005288525366709628,
      "loss": 2.4183,
      "step": 51565
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.682314932346344,
      "learning_rate": 0.0005288498917527787,
      "loss": 2.1045,
      "step": 51566
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.6418968439102173,
      "learning_rate": 0.0005288472467920471,
      "loss": 2.193,
      "step": 51567
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7225810885429382,
      "learning_rate": 0.0005288446017887686,
      "loss": 2.2079,
      "step": 51568
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7914049029350281,
      "learning_rate": 0.0005288419567429434,
      "loss": 2.329,
      "step": 51569
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7737236618995667,
      "learning_rate": 0.0005288393116545722,
      "loss": 2.276,
      "step": 51570
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.6952160000801086,
      "learning_rate": 0.0005288366665236555,
      "loss": 2.2283,
      "step": 51571
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.6671199798583984,
      "learning_rate": 0.0005288340213501936,
      "loss": 2.172,
      "step": 51572
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.6903359293937683,
      "learning_rate": 0.0005288313761341874,
      "loss": 2.3871,
      "step": 51573
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.741844117641449,
      "learning_rate": 0.0005288287308756369,
      "loss": 2.2704,
      "step": 51574
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7184650301933289,
      "learning_rate": 0.0005288260855745429,
      "loss": 2.1395,
      "step": 51575
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.8003378510475159,
      "learning_rate": 0.0005288234402309059,
      "loss": 2.2111,
      "step": 51576
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.6913470029830933,
      "learning_rate": 0.0005288207948447261,
      "loss": 2.4037,
      "step": 51577
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.6964665651321411,
      "learning_rate": 0.0005288181494160043,
      "loss": 2.2754,
      "step": 51578
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.8220023512840271,
      "learning_rate": 0.0005288155039447409,
      "loss": 2.0663,
      "step": 51579
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.6982194781303406,
      "learning_rate": 0.0005288128584309365,
      "loss": 1.9072,
      "step": 51580
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.677220344543457,
      "learning_rate": 0.0005288102128745913,
      "loss": 2.3505,
      "step": 51581
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.744453489780426,
      "learning_rate": 0.000528807567275706,
      "loss": 2.2797,
      "step": 51582
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7971954345703125,
      "learning_rate": 0.0005288049216342811,
      "loss": 2.0709,
      "step": 51583
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.6939125061035156,
      "learning_rate": 0.0005288022759503169,
      "loss": 2.1291,
      "step": 51584
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7631717324256897,
      "learning_rate": 0.0005287996302238142,
      "loss": 2.2905,
      "step": 51585
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7593777775764465,
      "learning_rate": 0.0005287969844547732,
      "loss": 2.4512,
      "step": 51586
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7645612359046936,
      "learning_rate": 0.0005287943386431947,
      "loss": 2.2811,
      "step": 51587
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.8204260468482971,
      "learning_rate": 0.000528791692789079,
      "loss": 2.4245,
      "step": 51588
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7742247581481934,
      "learning_rate": 0.0005287890468924264,
      "loss": 2.2966,
      "step": 51589
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.8182246088981628,
      "learning_rate": 0.0005287864009532377,
      "loss": 2.1691,
      "step": 51590
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7506223917007446,
      "learning_rate": 0.0005287837549715134,
      "loss": 2.249,
      "step": 51591
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.8829690217971802,
      "learning_rate": 0.0005287811089472536,
      "loss": 2.2068,
      "step": 51592
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7207500338554382,
      "learning_rate": 0.0005287784628804594,
      "loss": 2.0724,
      "step": 51593
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.6711507439613342,
      "learning_rate": 0.0005287758167711308,
      "loss": 2.3163,
      "step": 51594
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.826752245426178,
      "learning_rate": 0.0005287731706192685,
      "loss": 2.2533,
      "step": 51595
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.8015331625938416,
      "learning_rate": 0.0005287705244248729,
      "loss": 2.3648,
      "step": 51596
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.74315345287323,
      "learning_rate": 0.0005287678781879446,
      "loss": 2.2265,
      "step": 51597
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.8000207543373108,
      "learning_rate": 0.000528765231908484,
      "loss": 2.1155,
      "step": 51598
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.6921528577804565,
      "learning_rate": 0.0005287625855864916,
      "loss": 2.226,
      "step": 51599
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7292273044586182,
      "learning_rate": 0.000528759939221968,
      "loss": 2.3433,
      "step": 51600
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7174094319343567,
      "learning_rate": 0.0005287572928149135,
      "loss": 2.487,
      "step": 51601
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7229806184768677,
      "learning_rate": 0.0005287546463653288,
      "loss": 2.2824,
      "step": 51602
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7595731616020203,
      "learning_rate": 0.0005287519998732142,
      "loss": 2.2834,
      "step": 51603
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7590596079826355,
      "learning_rate": 0.0005287493533385704,
      "loss": 2.289,
      "step": 51604
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.6964886784553528,
      "learning_rate": 0.0005287467067613976,
      "loss": 2.1316,
      "step": 51605
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7493079304695129,
      "learning_rate": 0.0005287440601416967,
      "loss": 2.2418,
      "step": 51606
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7174695730209351,
      "learning_rate": 0.0005287414134794678,
      "loss": 2.3248,
      "step": 51607
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.8142780661582947,
      "learning_rate": 0.0005287387667747116,
      "loss": 2.1166,
      "step": 51608
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.6925923824310303,
      "learning_rate": 0.0005287361200274286,
      "loss": 2.0481,
      "step": 51609
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.794927716255188,
      "learning_rate": 0.0005287334732376191,
      "loss": 2.2533,
      "step": 51610
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.8550218343734741,
      "learning_rate": 0.0005287308264052838,
      "loss": 2.2045,
      "step": 51611
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7320243120193481,
      "learning_rate": 0.0005287281795304232,
      "loss": 2.0978,
      "step": 51612
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7344455122947693,
      "learning_rate": 0.0005287255326130376,
      "loss": 2.5627,
      "step": 51613
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.8075011372566223,
      "learning_rate": 0.0005287228856531277,
      "loss": 2.355,
      "step": 51614
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.798699676990509,
      "learning_rate": 0.0005287202386506938,
      "loss": 2.5427,
      "step": 51615
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7351065874099731,
      "learning_rate": 0.0005287175916057366,
      "loss": 2.2204,
      "step": 51616
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7486585974693298,
      "learning_rate": 0.0005287149445182565,
      "loss": 2.4318,
      "step": 51617
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.8413419127464294,
      "learning_rate": 0.0005287122973882538,
      "loss": 2.197,
      "step": 51618
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7289958596229553,
      "learning_rate": 0.0005287096502157293,
      "loss": 2.0733,
      "step": 51619
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.6871225833892822,
      "learning_rate": 0.0005287070030006833,
      "loss": 2.1515,
      "step": 51620
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.829717218875885,
      "learning_rate": 0.0005287043557431163,
      "loss": 2.1207,
      "step": 51621
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.696767270565033,
      "learning_rate": 0.000528701708443029,
      "loss": 2.1567,
      "step": 51622
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.692887544631958,
      "learning_rate": 0.0005286990611004216,
      "loss": 1.9965,
      "step": 51623
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7109597325325012,
      "learning_rate": 0.0005286964137152947,
      "loss": 2.1764,
      "step": 51624
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7545913457870483,
      "learning_rate": 0.000528693766287649,
      "loss": 2.3851,
      "step": 51625
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7155811190605164,
      "learning_rate": 0.0005286911188174847,
      "loss": 2.1884,
      "step": 51626
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7020562887191772,
      "learning_rate": 0.0005286884713048024,
      "loss": 2.3463,
      "step": 51627
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7556422352790833,
      "learning_rate": 0.0005286858237496027,
      "loss": 2.4531,
      "step": 51628
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.6984529495239258,
      "learning_rate": 0.0005286831761518859,
      "loss": 2.3324,
      "step": 51629
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.6542047262191772,
      "learning_rate": 0.0005286805285116525,
      "loss": 2.3338,
      "step": 51630
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7178425192832947,
      "learning_rate": 0.0005286778808289032,
      "loss": 2.1323,
      "step": 51631
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.6692827939987183,
      "learning_rate": 0.0005286752331036383,
      "loss": 2.1797,
      "step": 51632
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7242037653923035,
      "learning_rate": 0.0005286725853358583,
      "loss": 2.523,
      "step": 51633
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7558392882347107,
      "learning_rate": 0.0005286699375255639,
      "loss": 2.5631,
      "step": 51634
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.6793826222419739,
      "learning_rate": 0.0005286672896727554,
      "loss": 2.0651,
      "step": 51635
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.8269927501678467,
      "learning_rate": 0.0005286646417774333,
      "loss": 2.0856,
      "step": 51636
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7631821036338806,
      "learning_rate": 0.000528661993839598,
      "loss": 2.1792,
      "step": 51637
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7069367170333862,
      "learning_rate": 0.0005286593458592503,
      "loss": 2.2569,
      "step": 51638
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7521448731422424,
      "learning_rate": 0.0005286566978363904,
      "loss": 2.3471,
      "step": 51639
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7856497764587402,
      "learning_rate": 0.000528654049771019,
      "loss": 2.0664,
      "step": 51640
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.8617900013923645,
      "learning_rate": 0.0005286514016631366,
      "loss": 1.9813,
      "step": 51641
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.8608174324035645,
      "learning_rate": 0.0005286487535127434,
      "loss": 2.2682,
      "step": 51642
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7477547526359558,
      "learning_rate": 0.0005286461053198401,
      "loss": 2.2154,
      "step": 51643
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.8223812580108643,
      "learning_rate": 0.0005286434570844272,
      "loss": 2.5531,
      "step": 51644
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.9215220808982849,
      "learning_rate": 0.0005286408088065052,
      "loss": 2.3033,
      "step": 51645
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7038865685462952,
      "learning_rate": 0.0005286381604860745,
      "loss": 2.1272,
      "step": 51646
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.6750054955482483,
      "learning_rate": 0.0005286355121231357,
      "loss": 2.0259,
      "step": 51647
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.9104689955711365,
      "learning_rate": 0.0005286328637176892,
      "loss": 2.2725,
      "step": 51648
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.6582112908363342,
      "learning_rate": 0.0005286302152697355,
      "loss": 2.1331,
      "step": 51649
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.715691089630127,
      "learning_rate": 0.0005286275667792753,
      "loss": 2.1096,
      "step": 51650
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.6991466879844666,
      "learning_rate": 0.0005286249182463087,
      "loss": 2.3319,
      "step": 51651
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7254959940910339,
      "learning_rate": 0.0005286222696708365,
      "loss": 1.9975,
      "step": 51652
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.6729676723480225,
      "learning_rate": 0.0005286196210528591,
      "loss": 2.3379,
      "step": 51653
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7231009006500244,
      "learning_rate": 0.0005286169723923771,
      "loss": 2.2314,
      "step": 51654
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7272545695304871,
      "learning_rate": 0.0005286143236893908,
      "loss": 2.1312,
      "step": 51655
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7700158357620239,
      "learning_rate": 0.0005286116749439007,
      "loss": 2.3612,
      "step": 51656
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7000806927680969,
      "learning_rate": 0.0005286090261559075,
      "loss": 2.3433,
      "step": 51657
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.6421940922737122,
      "learning_rate": 0.0005286063773254116,
      "loss": 2.0579,
      "step": 51658
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.6690587997436523,
      "learning_rate": 0.0005286037284524133,
      "loss": 2.1856,
      "step": 51659
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7705919146537781,
      "learning_rate": 0.0005286010795369134,
      "loss": 2.5303,
      "step": 51660
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.772266685962677,
      "learning_rate": 0.0005285984305789122,
      "loss": 2.3692,
      "step": 51661
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7326856851577759,
      "learning_rate": 0.0005285957815784101,
      "loss": 2.211,
      "step": 51662
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.689784586429596,
      "learning_rate": 0.0005285931325354079,
      "loss": 2.3198,
      "step": 51663
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7176402807235718,
      "learning_rate": 0.0005285904834499059,
      "loss": 2.3065,
      "step": 51664
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.6944625377655029,
      "learning_rate": 0.0005285878343219045,
      "loss": 2.2003,
      "step": 51665
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7517910003662109,
      "learning_rate": 0.0005285851851514044,
      "loss": 2.2371,
      "step": 51666
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7031578421592712,
      "learning_rate": 0.0005285825359384061,
      "loss": 2.1857,
      "step": 51667
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.8599757552146912,
      "learning_rate": 0.0005285798866829098,
      "loss": 2.3893,
      "step": 51668
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.6591641902923584,
      "learning_rate": 0.0005285772373849162,
      "loss": 2.2384,
      "step": 51669
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.6968180537223816,
      "learning_rate": 0.0005285745880444258,
      "loss": 2.0954,
      "step": 51670
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7063650488853455,
      "learning_rate": 0.0005285719386614392,
      "loss": 2.1778,
      "step": 51671
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.6479429602622986,
      "learning_rate": 0.0005285692892359567,
      "loss": 2.1032,
      "step": 51672
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7475886344909668,
      "learning_rate": 0.0005285666397679788,
      "loss": 2.5037,
      "step": 51673
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.747908353805542,
      "learning_rate": 0.000528563990257506,
      "loss": 2.1482,
      "step": 51674
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.6930129528045654,
      "learning_rate": 0.000528561340704539,
      "loss": 2.3471,
      "step": 51675
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7925517559051514,
      "learning_rate": 0.000528558691109078,
      "loss": 2.3702,
      "step": 51676
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.6760520935058594,
      "learning_rate": 0.0005285560414711236,
      "loss": 2.2467,
      "step": 51677
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7492833137512207,
      "learning_rate": 0.0005285533917906765,
      "loss": 2.3024,
      "step": 51678
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7447399497032166,
      "learning_rate": 0.0005285507420677369,
      "loss": 2.0403,
      "step": 51679
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.8288514614105225,
      "learning_rate": 0.0005285480923023054,
      "loss": 2.2862,
      "step": 51680
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.8893485069274902,
      "learning_rate": 0.0005285454424943824,
      "loss": 2.4237,
      "step": 51681
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.8665481209754944,
      "learning_rate": 0.0005285427926439686,
      "loss": 2.19,
      "step": 51682
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7354467511177063,
      "learning_rate": 0.0005285401427510644,
      "loss": 2.2945,
      "step": 51683
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7301546931266785,
      "learning_rate": 0.0005285374928156703,
      "loss": 2.3115,
      "step": 51684
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.8035262823104858,
      "learning_rate": 0.0005285348428377866,
      "loss": 2.3142,
      "step": 51685
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7552059888839722,
      "learning_rate": 0.0005285321928174141,
      "loss": 2.1379,
      "step": 51686
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.9201451539993286,
      "learning_rate": 0.0005285295427545532,
      "loss": 2.2586,
      "step": 51687
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.787700891494751,
      "learning_rate": 0.0005285268926492042,
      "loss": 2.2676,
      "step": 51688
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7309291958808899,
      "learning_rate": 0.0005285242425013678,
      "loss": 2.4296,
      "step": 51689
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7232884168624878,
      "learning_rate": 0.0005285215923110444,
      "loss": 2.3039,
      "step": 51690
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7543854713439941,
      "learning_rate": 0.0005285189420782346,
      "loss": 2.0414,
      "step": 51691
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7059348821640015,
      "learning_rate": 0.0005285162918029387,
      "loss": 1.9945,
      "step": 51692
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7425007820129395,
      "learning_rate": 0.0005285136414851574,
      "loss": 2.3603,
      "step": 51693
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7686278820037842,
      "learning_rate": 0.0005285109911248911,
      "loss": 2.2564,
      "step": 51694
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.9455586075782776,
      "learning_rate": 0.0005285083407221403,
      "loss": 2.238,
      "step": 51695
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7155159115791321,
      "learning_rate": 0.0005285056902769054,
      "loss": 2.4748,
      "step": 51696
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7141332626342773,
      "learning_rate": 0.000528503039789187,
      "loss": 2.3567,
      "step": 51697
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.0056393146514893,
      "learning_rate": 0.0005285003892589856,
      "loss": 2.1361,
      "step": 51698
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7678937315940857,
      "learning_rate": 0.0005284977386863016,
      "loss": 2.1593,
      "step": 51699
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.6944876313209534,
      "learning_rate": 0.0005284950880711357,
      "loss": 2.2711,
      "step": 51700
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7876445055007935,
      "learning_rate": 0.0005284924374134881,
      "loss": 2.2626,
      "step": 51701
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.8120400309562683,
      "learning_rate": 0.0005284897867133595,
      "loss": 2.2221,
      "step": 51702
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.6976425647735596,
      "learning_rate": 0.0005284871359707504,
      "loss": 2.2305,
      "step": 51703
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7554525136947632,
      "learning_rate": 0.0005284844851856613,
      "loss": 2.3028,
      "step": 51704
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7522230744361877,
      "learning_rate": 0.0005284818343580924,
      "loss": 2.4613,
      "step": 51705
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.6918156743049622,
      "learning_rate": 0.0005284791834880446,
      "loss": 2.2135,
      "step": 51706
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7123266458511353,
      "learning_rate": 0.000528476532575518,
      "loss": 2.4901,
      "step": 51707
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.716291606426239,
      "learning_rate": 0.0005284738816205135,
      "loss": 2.2074,
      "step": 51708
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7586440443992615,
      "learning_rate": 0.0005284712306230313,
      "loss": 2.212,
      "step": 51709
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7294325232505798,
      "learning_rate": 0.000528468579583072,
      "loss": 2.3569,
      "step": 51710
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7576107382774353,
      "learning_rate": 0.0005284659285006361,
      "loss": 2.2433,
      "step": 51711
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7545444965362549,
      "learning_rate": 0.000528463277375724,
      "loss": 2.4268,
      "step": 51712
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7668980956077576,
      "learning_rate": 0.0005284606262083364,
      "loss": 2.082,
      "step": 51713
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7838445901870728,
      "learning_rate": 0.0005284579749984735,
      "loss": 2.3825,
      "step": 51714
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7271742820739746,
      "learning_rate": 0.000528455323746136,
      "loss": 1.9032,
      "step": 51715
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.6403931379318237,
      "learning_rate": 0.0005284526724513244,
      "loss": 2.2206,
      "step": 51716
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7914470434188843,
      "learning_rate": 0.0005284500211140391,
      "loss": 2.0819,
      "step": 51717
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.6979643106460571,
      "learning_rate": 0.0005284473697342807,
      "loss": 2.2423,
      "step": 51718
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7356783151626587,
      "learning_rate": 0.0005284447183120495,
      "loss": 2.1033,
      "step": 51719
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7578986883163452,
      "learning_rate": 0.0005284420668473461,
      "loss": 2.4753,
      "step": 51720
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.6688891053199768,
      "learning_rate": 0.0005284394153401711,
      "loss": 2.1322,
      "step": 51721
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.6791452765464783,
      "learning_rate": 0.0005284367637905249,
      "loss": 2.0946,
      "step": 51722
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7479469180107117,
      "learning_rate": 0.0005284341121984079,
      "loss": 2.3914,
      "step": 51723
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7012588977813721,
      "learning_rate": 0.0005284314605638207,
      "loss": 2.082,
      "step": 51724
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7007601261138916,
      "learning_rate": 0.0005284288088867638,
      "loss": 2.4391,
      "step": 51725
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.687002420425415,
      "learning_rate": 0.0005284261571672378,
      "loss": 2.1727,
      "step": 51726
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7507158517837524,
      "learning_rate": 0.000528423505405243,
      "loss": 2.37,
      "step": 51727
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7825727462768555,
      "learning_rate": 0.0005284208536007799,
      "loss": 2.3667,
      "step": 51728
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.9120931625366211,
      "learning_rate": 0.0005284182017538491,
      "loss": 2.2385,
      "step": 51729
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7351581454277039,
      "learning_rate": 0.000528415549864451,
      "loss": 2.4942,
      "step": 51730
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.6730982661247253,
      "learning_rate": 0.0005284128979325862,
      "loss": 2.1611,
      "step": 51731
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7450321316719055,
      "learning_rate": 0.0005284102459582551,
      "loss": 2.2958,
      "step": 51732
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.6884370446205139,
      "learning_rate": 0.0005284075939414581,
      "loss": 2.1975,
      "step": 51733
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.9556664228439331,
      "learning_rate": 0.0005284049418821959,
      "loss": 2.2231,
      "step": 51734
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7062902450561523,
      "learning_rate": 0.0005284022897804691,
      "loss": 2.0701,
      "step": 51735
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7741274833679199,
      "learning_rate": 0.0005283996376362778,
      "loss": 2.119,
      "step": 51736
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.837513267993927,
      "learning_rate": 0.0005283969854496227,
      "loss": 2.2968,
      "step": 51737
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.829778790473938,
      "learning_rate": 0.0005283943332205043,
      "loss": 2.1968,
      "step": 51738
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.8395346999168396,
      "learning_rate": 0.0005283916809489231,
      "loss": 2.3434,
      "step": 51739
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7924641370773315,
      "learning_rate": 0.0005283890286348797,
      "loss": 2.0262,
      "step": 51740
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.8984904289245605,
      "learning_rate": 0.0005283863762783744,
      "loss": 2.1475,
      "step": 51741
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7174007296562195,
      "learning_rate": 0.0005283837238794076,
      "loss": 2.1662,
      "step": 51742
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7763010263442993,
      "learning_rate": 0.0005283810714379801,
      "loss": 2.4765,
      "step": 51743
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.8663968443870544,
      "learning_rate": 0.0005283784189540923,
      "loss": 2.2435,
      "step": 51744
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.8469564318656921,
      "learning_rate": 0.0005283757664277446,
      "loss": 2.0952,
      "step": 51745
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7505071759223938,
      "learning_rate": 0.0005283731138589374,
      "loss": 2.2706,
      "step": 51746
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7793911099433899,
      "learning_rate": 0.0005283704612476715,
      "loss": 2.3943,
      "step": 51747
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7074131965637207,
      "learning_rate": 0.0005283678085939471,
      "loss": 2.3533,
      "step": 51748
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.0976226329803467,
      "learning_rate": 0.0005283651558977649,
      "loss": 2.2322,
      "step": 51749
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.6810604929924011,
      "learning_rate": 0.0005283625031591251,
      "loss": 2.2357,
      "step": 51750
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.6732982397079468,
      "learning_rate": 0.0005283598503780287,
      "loss": 2.387,
      "step": 51751
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.05721914768219,
      "learning_rate": 0.0005283571975544756,
      "loss": 2.2018,
      "step": 51752
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.1416441202163696,
      "learning_rate": 0.0005283545446884668,
      "loss": 2.0909,
      "step": 51753
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.6968336701393127,
      "learning_rate": 0.0005283518917800025,
      "loss": 2.2324,
      "step": 51754
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.0026264190673828,
      "learning_rate": 0.0005283492388290834,
      "loss": 2.2762,
      "step": 51755
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7434213757514954,
      "learning_rate": 0.0005283465858357097,
      "loss": 2.1338,
      "step": 51756
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.9330606460571289,
      "learning_rate": 0.0005283439327998822,
      "loss": 2.0702,
      "step": 51757
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7609621286392212,
      "learning_rate": 0.0005283412797216011,
      "loss": 2.5125,
      "step": 51758
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7764906883239746,
      "learning_rate": 0.0005283386266008672,
      "loss": 2.4575,
      "step": 51759
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.9458906650543213,
      "learning_rate": 0.0005283359734376807,
      "loss": 2.2914,
      "step": 51760
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7199462652206421,
      "learning_rate": 0.0005283333202320422,
      "loss": 2.2762,
      "step": 51761
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.6943562626838684,
      "learning_rate": 0.0005283306669839524,
      "loss": 2.3223,
      "step": 51762
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7518551349639893,
      "learning_rate": 0.0005283280136934115,
      "loss": 2.1168,
      "step": 51763
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7769510746002197,
      "learning_rate": 0.0005283253603604202,
      "loss": 2.2999,
      "step": 51764
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7361121773719788,
      "learning_rate": 0.0005283227069849788,
      "loss": 2.2221,
      "step": 51765
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7274460196495056,
      "learning_rate": 0.0005283200535670879,
      "loss": 2.2701,
      "step": 51766
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7838393449783325,
      "learning_rate": 0.0005283174001067481,
      "loss": 2.2286,
      "step": 51767
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7006072998046875,
      "learning_rate": 0.0005283147466039598,
      "loss": 2.2889,
      "step": 51768
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.820563554763794,
      "learning_rate": 0.0005283120930587233,
      "loss": 2.076,
      "step": 51769
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7463358640670776,
      "learning_rate": 0.0005283094394710394,
      "loss": 2.5528,
      "step": 51770
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.6978147625923157,
      "learning_rate": 0.0005283067858409085,
      "loss": 2.2074,
      "step": 51771
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.9474789500236511,
      "learning_rate": 0.0005283041321683309,
      "loss": 2.3056,
      "step": 51772
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7502709031105042,
      "learning_rate": 0.0005283014784533075,
      "loss": 2.4001,
      "step": 51773
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.787731945514679,
      "learning_rate": 0.0005282988246958384,
      "loss": 2.208,
      "step": 51774
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7706975936889648,
      "learning_rate": 0.0005282961708959242,
      "loss": 2.2505,
      "step": 51775
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7153254747390747,
      "learning_rate": 0.0005282935170535655,
      "loss": 2.1181,
      "step": 51776
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7105838060379028,
      "learning_rate": 0.0005282908631687626,
      "loss": 2.2687,
      "step": 51777
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.621468722820282,
      "learning_rate": 0.0005282882092415163,
      "loss": 2.2755,
      "step": 51778
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.8367773294448853,
      "learning_rate": 0.0005282855552718268,
      "loss": 2.2961,
      "step": 51779
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7639141082763672,
      "learning_rate": 0.0005282829012596947,
      "loss": 2.6011,
      "step": 51780
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7656173706054688,
      "learning_rate": 0.0005282802472051206,
      "loss": 2.2755,
      "step": 51781
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.6997901201248169,
      "learning_rate": 0.0005282775931081049,
      "loss": 2.388,
      "step": 51782
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.6933251023292542,
      "learning_rate": 0.000528274938968648,
      "loss": 2.1996,
      "step": 51783
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7097739577293396,
      "learning_rate": 0.0005282722847867505,
      "loss": 2.3584,
      "step": 51784
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.8486233353614807,
      "learning_rate": 0.0005282696305624128,
      "loss": 2.223,
      "step": 51785
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7150599956512451,
      "learning_rate": 0.0005282669762956356,
      "loss": 2.4311,
      "step": 51786
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.6912281513214111,
      "learning_rate": 0.0005282643219864192,
      "loss": 2.3226,
      "step": 51787
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7106406688690186,
      "learning_rate": 0.0005282616676347641,
      "loss": 2.2951,
      "step": 51788
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.870646595954895,
      "learning_rate": 0.0005282590132406709,
      "loss": 2.192,
      "step": 51789
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.8212926983833313,
      "learning_rate": 0.00052825635880414,
      "loss": 2.3268,
      "step": 51790
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7517428994178772,
      "learning_rate": 0.000528253704325172,
      "loss": 2.4119,
      "step": 51791
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7052474617958069,
      "learning_rate": 0.0005282510498037673,
      "loss": 2.2755,
      "step": 51792
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.8381876349449158,
      "learning_rate": 0.0005282483952399264,
      "loss": 2.2229,
      "step": 51793
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7257786989212036,
      "learning_rate": 0.0005282457406336499,
      "loss": 2.4835,
      "step": 51794
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.6909058690071106,
      "learning_rate": 0.0005282430859849381,
      "loss": 2.0896,
      "step": 51795
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7014180421829224,
      "learning_rate": 0.0005282404312937917,
      "loss": 2.2727,
      "step": 51796
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.8113923668861389,
      "learning_rate": 0.000528237776560211,
      "loss": 2.1998,
      "step": 51797
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.6859421133995056,
      "learning_rate": 0.0005282351217841966,
      "loss": 2.3248,
      "step": 51798
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7219510674476624,
      "learning_rate": 0.0005282324669657491,
      "loss": 2.3097,
      "step": 51799
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.8193363547325134,
      "learning_rate": 0.0005282298121048688,
      "loss": 2.1421,
      "step": 51800
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.6803154349327087,
      "learning_rate": 0.0005282271572015562,
      "loss": 2.07,
      "step": 51801
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7106773257255554,
      "learning_rate": 0.0005282245022558119,
      "loss": 2.2061,
      "step": 51802
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7868915796279907,
      "learning_rate": 0.0005282218472676364,
      "loss": 2.4029,
      "step": 51803
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.8785426020622253,
      "learning_rate": 0.0005282191922370301,
      "loss": 2.2799,
      "step": 51804
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.8348962664604187,
      "learning_rate": 0.0005282165371639936,
      "loss": 2.2058,
      "step": 51805
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7312694191932678,
      "learning_rate": 0.0005282138820485273,
      "loss": 2.3806,
      "step": 51806
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.040860652923584,
      "learning_rate": 0.0005282112268906317,
      "loss": 2.1644,
      "step": 51807
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7908000349998474,
      "learning_rate": 0.0005282085716903075,
      "loss": 2.2597,
      "step": 51808
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7706834673881531,
      "learning_rate": 0.0005282059164475548,
      "loss": 2.1906,
      "step": 51809
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.8907745480537415,
      "learning_rate": 0.0005282032611623744,
      "loss": 2.3946,
      "step": 51810
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.853506863117218,
      "learning_rate": 0.0005282006058347667,
      "loss": 2.0143,
      "step": 51811
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7509332299232483,
      "learning_rate": 0.0005281979504647322,
      "loss": 2.1804,
      "step": 51812
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7376627326011658,
      "learning_rate": 0.0005281952950522714,
      "loss": 2.3207,
      "step": 51813
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.8901149034500122,
      "learning_rate": 0.0005281926395973848,
      "loss": 2.2923,
      "step": 51814
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7677094340324402,
      "learning_rate": 0.0005281899841000729,
      "loss": 2.3079,
      "step": 51815
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.6636930108070374,
      "learning_rate": 0.0005281873285603361,
      "loss": 2.1815,
      "step": 51816
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7106075882911682,
      "learning_rate": 0.000528184672978175,
      "loss": 2.2192,
      "step": 51817
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.8049495816230774,
      "learning_rate": 0.0005281820173535901,
      "loss": 2.2488,
      "step": 51818
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7903134822845459,
      "learning_rate": 0.0005281793616865818,
      "loss": 2.0983,
      "step": 51819
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7341195344924927,
      "learning_rate": 0.0005281767059771508,
      "loss": 2.185,
      "step": 51820
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7074989676475525,
      "learning_rate": 0.0005281740502252972,
      "loss": 2.4645,
      "step": 51821
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.813072681427002,
      "learning_rate": 0.0005281713944310219,
      "loss": 2.461,
      "step": 51822
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7759284973144531,
      "learning_rate": 0.0005281687385943252,
      "loss": 2.1752,
      "step": 51823
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.8275176286697388,
      "learning_rate": 0.0005281660827152076,
      "loss": 2.4351,
      "step": 51824
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7430965900421143,
      "learning_rate": 0.0005281634267936696,
      "loss": 2.2426,
      "step": 51825
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7241299152374268,
      "learning_rate": 0.0005281607708297118,
      "loss": 2.2439,
      "step": 51826
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7704963088035583,
      "learning_rate": 0.0005281581148233346,
      "loss": 2.2654,
      "step": 51827
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.739106297492981,
      "learning_rate": 0.0005281554587745383,
      "loss": 2.059,
      "step": 51828
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7469891309738159,
      "learning_rate": 0.0005281528026833238,
      "loss": 2.1775,
      "step": 51829
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.713853120803833,
      "learning_rate": 0.0005281501465496913,
      "loss": 2.3285,
      "step": 51830
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7792965769767761,
      "learning_rate": 0.0005281474903736415,
      "loss": 2.1005,
      "step": 51831
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7166574001312256,
      "learning_rate": 0.0005281448341551747,
      "loss": 2.1507,
      "step": 51832
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.6519507765769958,
      "learning_rate": 0.0005281421778942914,
      "loss": 2.1803,
      "step": 51833
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.6958761215209961,
      "learning_rate": 0.0005281395215909922,
      "loss": 2.2881,
      "step": 51834
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7423139214515686,
      "learning_rate": 0.0005281368652452776,
      "loss": 2.3068,
      "step": 51835
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.6472216248512268,
      "learning_rate": 0.0005281342088571479,
      "loss": 2.1788,
      "step": 51836
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7200334072113037,
      "learning_rate": 0.000528131552426604,
      "loss": 2.2541,
      "step": 51837
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7488018274307251,
      "learning_rate": 0.000528128895953646,
      "loss": 2.3077,
      "step": 51838
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7000290751457214,
      "learning_rate": 0.0005281262394382745,
      "loss": 2.3744,
      "step": 51839
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.0381591320037842,
      "learning_rate": 0.00052812358288049,
      "loss": 2.056,
      "step": 51840
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7303751707077026,
      "learning_rate": 0.0005281209262802931,
      "loss": 2.2537,
      "step": 51841
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7938959002494812,
      "learning_rate": 0.0005281182696376842,
      "loss": 2.2696,
      "step": 51842
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.9568682909011841,
      "learning_rate": 0.0005281156129526639,
      "loss": 2.2174,
      "step": 51843
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.8043124079704285,
      "learning_rate": 0.0005281129562252324,
      "loss": 2.2118,
      "step": 51844
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.6929249167442322,
      "learning_rate": 0.0005281102994553905,
      "loss": 2.5246,
      "step": 51845
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7545895576477051,
      "learning_rate": 0.0005281076426431387,
      "loss": 2.1922,
      "step": 51846
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7915539741516113,
      "learning_rate": 0.0005281049857884773,
      "loss": 2.4399,
      "step": 51847
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7874593734741211,
      "learning_rate": 0.0005281023288914068,
      "loss": 2.3426,
      "step": 51848
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.6711596250534058,
      "learning_rate": 0.0005280996719519279,
      "loss": 2.3226,
      "step": 51849
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.6947605013847351,
      "learning_rate": 0.0005280970149700408,
      "loss": 2.1075,
      "step": 51850
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7016148567199707,
      "learning_rate": 0.0005280943579457462,
      "loss": 2.2432,
      "step": 51851
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7681600451469421,
      "learning_rate": 0.0005280917008790446,
      "loss": 2.2901,
      "step": 51852
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7442582249641418,
      "learning_rate": 0.0005280890437699365,
      "loss": 2.2833,
      "step": 51853
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.70997554063797,
      "learning_rate": 0.0005280863866184223,
      "loss": 2.1916,
      "step": 51854
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7212405800819397,
      "learning_rate": 0.0005280837294245025,
      "loss": 2.1733,
      "step": 51855
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.9157989025115967,
      "learning_rate": 0.0005280810721881777,
      "loss": 2.2637,
      "step": 51856
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7099199891090393,
      "learning_rate": 0.0005280784149094482,
      "loss": 2.3515,
      "step": 51857
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7263749837875366,
      "learning_rate": 0.0005280757575883147,
      "loss": 2.3752,
      "step": 51858
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7528499364852905,
      "learning_rate": 0.0005280731002247777,
      "loss": 2.3471,
      "step": 51859
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.6701980829238892,
      "learning_rate": 0.0005280704428188374,
      "loss": 2.1286,
      "step": 51860
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.66354900598526,
      "learning_rate": 0.0005280677853704947,
      "loss": 2.2868,
      "step": 51861
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7903393507003784,
      "learning_rate": 0.0005280651278797497,
      "loss": 2.2758,
      "step": 51862
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7285490036010742,
      "learning_rate": 0.0005280624703466033,
      "loss": 2.3221,
      "step": 51863
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.6400792598724365,
      "learning_rate": 0.0005280598127710556,
      "loss": 2.2918,
      "step": 51864
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7231464385986328,
      "learning_rate": 0.0005280571551531073,
      "loss": 2.1823,
      "step": 51865
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.6953729391098022,
      "learning_rate": 0.0005280544974927589,
      "loss": 2.1956,
      "step": 51866
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.6764160394668579,
      "learning_rate": 0.0005280518397900109,
      "loss": 2.2326,
      "step": 51867
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.6306912899017334,
      "learning_rate": 0.0005280491820448638,
      "loss": 2.1418,
      "step": 51868
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7239798903465271,
      "learning_rate": 0.000528046524257318,
      "loss": 2.2358,
      "step": 51869
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7001389861106873,
      "learning_rate": 0.000528043866427374,
      "loss": 2.1314,
      "step": 51870
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7262038588523865,
      "learning_rate": 0.0005280412085550324,
      "loss": 2.0028,
      "step": 51871
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7058932185173035,
      "learning_rate": 0.0005280385506402937,
      "loss": 2.2125,
      "step": 51872
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7063871026039124,
      "learning_rate": 0.0005280358926831581,
      "loss": 2.1317,
      "step": 51873
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7057289481163025,
      "learning_rate": 0.0005280332346836266,
      "loss": 2.2419,
      "step": 51874
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7068365812301636,
      "learning_rate": 0.0005280305766416993,
      "loss": 2.1652,
      "step": 51875
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.8025911450386047,
      "learning_rate": 0.0005280279185573768,
      "loss": 2.3116,
      "step": 51876
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.6781551241874695,
      "learning_rate": 0.0005280252604306596,
      "loss": 2.1778,
      "step": 51877
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7432316541671753,
      "learning_rate": 0.0005280226022615481,
      "loss": 2.2499,
      "step": 51878
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7595558166503906,
      "learning_rate": 0.0005280199440500431,
      "loss": 2.156,
      "step": 51879
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7367031574249268,
      "learning_rate": 0.0005280172857961448,
      "loss": 2.2914,
      "step": 51880
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7737465500831604,
      "learning_rate": 0.0005280146274998537,
      "loss": 2.2969,
      "step": 51881
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7541250586509705,
      "learning_rate": 0.0005280119691611705,
      "loss": 2.1805,
      "step": 51882
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.6815528869628906,
      "learning_rate": 0.0005280093107800955,
      "loss": 2.3686,
      "step": 51883
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.6180835962295532,
      "learning_rate": 0.0005280066523566292,
      "loss": 2.2176,
      "step": 51884
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7666805386543274,
      "learning_rate": 0.0005280039938907723,
      "loss": 2.0366,
      "step": 51885
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7619307041168213,
      "learning_rate": 0.0005280013353825251,
      "loss": 2.2823,
      "step": 51886
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7514050602912903,
      "learning_rate": 0.0005279986768318881,
      "loss": 2.0644,
      "step": 51887
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7203168869018555,
      "learning_rate": 0.0005279960182388619,
      "loss": 2.0361,
      "step": 51888
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7159777283668518,
      "learning_rate": 0.0005279933596034469,
      "loss": 2.2786,
      "step": 51889
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7584484815597534,
      "learning_rate": 0.0005279907009256436,
      "loss": 2.4421,
      "step": 51890
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7383378744125366,
      "learning_rate": 0.0005279880422054526,
      "loss": 2.1922,
      "step": 51891
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7065902352333069,
      "learning_rate": 0.0005279853834428743,
      "loss": 2.1434,
      "step": 51892
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.6741403341293335,
      "learning_rate": 0.0005279827246379092,
      "loss": 2.4527,
      "step": 51893
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7454681396484375,
      "learning_rate": 0.0005279800657905578,
      "loss": 2.3619,
      "step": 51894
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7611892819404602,
      "learning_rate": 0.0005279774069008207,
      "loss": 2.4194,
      "step": 51895
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7817398309707642,
      "learning_rate": 0.0005279747479686981,
      "loss": 2.2165,
      "step": 51896
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.6969716548919678,
      "learning_rate": 0.0005279720889941908,
      "loss": 2.3154,
      "step": 51897
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.8591150641441345,
      "learning_rate": 0.0005279694299772993,
      "loss": 2.117,
      "step": 51898
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7464659214019775,
      "learning_rate": 0.0005279667709180239,
      "loss": 2.0891,
      "step": 51899
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.6653750538825989,
      "learning_rate": 0.0005279641118163651,
      "loss": 2.3029,
      "step": 51900
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7498894333839417,
      "learning_rate": 0.0005279614526723235,
      "loss": 2.2193,
      "step": 51901
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.8223310708999634,
      "learning_rate": 0.0005279587934858995,
      "loss": 2.2442,
      "step": 51902
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7109388709068298,
      "learning_rate": 0.0005279561342570938,
      "loss": 2.2571,
      "step": 51903
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.8310558795928955,
      "learning_rate": 0.0005279534749859067,
      "loss": 2.4067,
      "step": 51904
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.6842622756958008,
      "learning_rate": 0.0005279508156723387,
      "loss": 2.1304,
      "step": 51905
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.6789491176605225,
      "learning_rate": 0.0005279481563163904,
      "loss": 2.363,
      "step": 51906
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7296649813652039,
      "learning_rate": 0.0005279454969180622,
      "loss": 2.2907,
      "step": 51907
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.8085311651229858,
      "learning_rate": 0.0005279428374773546,
      "loss": 2.3573,
      "step": 51908
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7288259863853455,
      "learning_rate": 0.0005279401779942682,
      "loss": 2.2485,
      "step": 51909
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.813414454460144,
      "learning_rate": 0.0005279375184688034,
      "loss": 2.3115,
      "step": 51910
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.6723306775093079,
      "learning_rate": 0.0005279348589009606,
      "loss": 2.1487,
      "step": 51911
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.6773768663406372,
      "learning_rate": 0.0005279321992907405,
      "loss": 2.4467,
      "step": 51912
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.8675252795219421,
      "learning_rate": 0.0005279295396381436,
      "loss": 2.0612,
      "step": 51913
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.6791998147964478,
      "learning_rate": 0.0005279268799431702,
      "loss": 2.1865,
      "step": 51914
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.6902312636375427,
      "learning_rate": 0.0005279242202058208,
      "loss": 2.4715,
      "step": 51915
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.79304438829422,
      "learning_rate": 0.0005279215604260961,
      "loss": 2.4526,
      "step": 51916
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7207325100898743,
      "learning_rate": 0.0005279189006039965,
      "loss": 2.2541,
      "step": 51917
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.8189094066619873,
      "learning_rate": 0.0005279162407395224,
      "loss": 2.2147,
      "step": 51918
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.8782171607017517,
      "learning_rate": 0.0005279135808326744,
      "loss": 2.12,
      "step": 51919
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.8013542294502258,
      "learning_rate": 0.000527910920883453,
      "loss": 2.3409,
      "step": 51920
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.6657043099403381,
      "learning_rate": 0.0005279082608918586,
      "loss": 2.016,
      "step": 51921
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.762821614742279,
      "learning_rate": 0.0005279056008578917,
      "loss": 2.1961,
      "step": 51922
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.8992902040481567,
      "learning_rate": 0.0005279029407815529,
      "loss": 2.1453,
      "step": 51923
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7464696168899536,
      "learning_rate": 0.0005279002806628427,
      "loss": 2.1879,
      "step": 51924
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.66905277967453,
      "learning_rate": 0.0005278976205017615,
      "loss": 1.8085,
      "step": 51925
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7987963557243347,
      "learning_rate": 0.0005278949602983098,
      "loss": 2.325,
      "step": 51926
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.750986635684967,
      "learning_rate": 0.0005278923000524881,
      "loss": 2.3181,
      "step": 51927
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7561323642730713,
      "learning_rate": 0.000527889639764297,
      "loss": 2.3095,
      "step": 51928
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.804865837097168,
      "learning_rate": 0.0005278869794337369,
      "loss": 2.3766,
      "step": 51929
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7388721704483032,
      "learning_rate": 0.0005278843190608082,
      "loss": 2.2062,
      "step": 51930
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7146864533424377,
      "learning_rate": 0.0005278816586455116,
      "loss": 2.1623,
      "step": 51931
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7002440690994263,
      "learning_rate": 0.0005278789981878476,
      "loss": 2.2906,
      "step": 51932
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.6961695551872253,
      "learning_rate": 0.0005278763376878163,
      "loss": 2.2464,
      "step": 51933
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.6604631543159485,
      "learning_rate": 0.0005278736771454188,
      "loss": 2.007,
      "step": 51934
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7235066294670105,
      "learning_rate": 0.0005278710165606552,
      "loss": 2.3038,
      "step": 51935
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7616429328918457,
      "learning_rate": 0.000527868355933526,
      "loss": 2.2534,
      "step": 51936
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7058342695236206,
      "learning_rate": 0.0005278656952640318,
      "loss": 2.2483,
      "step": 51937
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.8322716355323792,
      "learning_rate": 0.000527863034552173,
      "loss": 2.2719,
      "step": 51938
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7226088047027588,
      "learning_rate": 0.0005278603737979503,
      "loss": 2.3359,
      "step": 51939
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.1981925964355469,
      "learning_rate": 0.000527857713001364,
      "loss": 2.2999,
      "step": 51940
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.8081967234611511,
      "learning_rate": 0.0005278550521624145,
      "loss": 2.3171,
      "step": 51941
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7600122690200806,
      "learning_rate": 0.0005278523912811027,
      "loss": 2.3225,
      "step": 51942
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7294259071350098,
      "learning_rate": 0.0005278497303574287,
      "loss": 2.1587,
      "step": 51943
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7729793190956116,
      "learning_rate": 0.0005278470693913931,
      "loss": 2.4273,
      "step": 51944
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7932168841362,
      "learning_rate": 0.0005278444083829964,
      "loss": 2.2185,
      "step": 51945
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7068848013877869,
      "learning_rate": 0.0005278417473322393,
      "loss": 2.2704,
      "step": 51946
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.6916490793228149,
      "learning_rate": 0.000527839086239122,
      "loss": 2.2046,
      "step": 51947
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7710840106010437,
      "learning_rate": 0.000527836425103645,
      "loss": 2.62,
      "step": 51948
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7397581934928894,
      "learning_rate": 0.0005278337639258092,
      "loss": 2.3256,
      "step": 51949
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7532759308815002,
      "learning_rate": 0.0005278311027056146,
      "loss": 2.273,
      "step": 51950
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.6816722750663757,
      "learning_rate": 0.0005278284414430619,
      "loss": 2.0286,
      "step": 51951
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7679751515388489,
      "learning_rate": 0.0005278257801381518,
      "loss": 2.2661,
      "step": 51952
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.8962118625640869,
      "learning_rate": 0.0005278231187908843,
      "loss": 2.3186,
      "step": 51953
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.8630033135414124,
      "learning_rate": 0.0005278204574012603,
      "loss": 2.053,
      "step": 51954
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.8156163692474365,
      "learning_rate": 0.0005278177959692803,
      "loss": 2.0823,
      "step": 51955
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7110943794250488,
      "learning_rate": 0.0005278151344949444,
      "loss": 2.3851,
      "step": 51956
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.008760929107666,
      "learning_rate": 0.0005278124729782536,
      "loss": 2.4365,
      "step": 51957
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7820851802825928,
      "learning_rate": 0.000527809811419208,
      "loss": 2.3181,
      "step": 51958
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7320679426193237,
      "learning_rate": 0.0005278071498178084,
      "loss": 2.426,
      "step": 51959
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.6853466033935547,
      "learning_rate": 0.0005278044881740551,
      "loss": 2.2041,
      "step": 51960
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7849742770195007,
      "learning_rate": 0.0005278018264879486,
      "loss": 2.382,
      "step": 51961
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7235532999038696,
      "learning_rate": 0.0005277991647594894,
      "loss": 2.2149,
      "step": 51962
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.8802248239517212,
      "learning_rate": 0.0005277965029886782,
      "loss": 2.2805,
      "step": 51963
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.6552698612213135,
      "learning_rate": 0.0005277938411755152,
      "loss": 2.2952,
      "step": 51964
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.728600025177002,
      "learning_rate": 0.0005277911793200011,
      "loss": 2.246,
      "step": 51965
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7215644121170044,
      "learning_rate": 0.0005277885174221362,
      "loss": 2.3153,
      "step": 51966
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7702515721321106,
      "learning_rate": 0.0005277858554819212,
      "loss": 2.2589,
      "step": 51967
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7524698972702026,
      "learning_rate": 0.0005277831934993565,
      "loss": 2.1062,
      "step": 51968
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.712628185749054,
      "learning_rate": 0.0005277805314744426,
      "loss": 2.0465,
      "step": 51969
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7741550207138062,
      "learning_rate": 0.00052777786940718,
      "loss": 2.3518,
      "step": 51970
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7701335549354553,
      "learning_rate": 0.0005277752072975693,
      "loss": 2.3453,
      "step": 51971
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7600845694541931,
      "learning_rate": 0.0005277725451456108,
      "loss": 2.3274,
      "step": 51972
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7937719821929932,
      "learning_rate": 0.0005277698829513049,
      "loss": 2.2878,
      "step": 51973
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7715960741043091,
      "learning_rate": 0.0005277672207146526,
      "loss": 2.4777,
      "step": 51974
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.738109827041626,
      "learning_rate": 0.000527764558435654,
      "loss": 2.1608,
      "step": 51975
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7099090218544006,
      "learning_rate": 0.0005277618961143095,
      "loss": 2.4548,
      "step": 51976
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7374202609062195,
      "learning_rate": 0.0005277592337506198,
      "loss": 2.2439,
      "step": 51977
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.8127593398094177,
      "learning_rate": 0.0005277565713445856,
      "loss": 2.2583,
      "step": 51978
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.9060089588165283,
      "learning_rate": 0.0005277539088962068,
      "loss": 2.1918,
      "step": 51979
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7484797835350037,
      "learning_rate": 0.0005277512464054845,
      "loss": 2.4021,
      "step": 51980
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.0055021047592163,
      "learning_rate": 0.0005277485838724188,
      "loss": 2.6024,
      "step": 51981
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.833878219127655,
      "learning_rate": 0.0005277459212970105,
      "loss": 2.187,
      "step": 51982
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7360003590583801,
      "learning_rate": 0.0005277432586792598,
      "loss": 2.0959,
      "step": 51983
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7066109776496887,
      "learning_rate": 0.0005277405960191674,
      "loss": 2.4053,
      "step": 51984
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7593817114830017,
      "learning_rate": 0.0005277379333167337,
      "loss": 2.0611,
      "step": 51985
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.2017128467559814,
      "learning_rate": 0.0005277352705719591,
      "loss": 2.2714,
      "step": 51986
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7167112231254578,
      "learning_rate": 0.0005277326077848444,
      "loss": 2.2689,
      "step": 51987
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.9459967613220215,
      "learning_rate": 0.0005277299449553899,
      "loss": 2.2172,
      "step": 51988
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.9002628326416016,
      "learning_rate": 0.000527727282083596,
      "loss": 2.3396,
      "step": 51989
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7126212120056152,
      "learning_rate": 0.0005277246191694633,
      "loss": 2.1706,
      "step": 51990
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7093673944473267,
      "learning_rate": 0.0005277219562129923,
      "loss": 2.2284,
      "step": 51991
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.6512223482131958,
      "learning_rate": 0.0005277192932141837,
      "loss": 2.2017,
      "step": 51992
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7549351453781128,
      "learning_rate": 0.0005277166301730375,
      "loss": 2.0831,
      "step": 51993
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7884849905967712,
      "learning_rate": 0.0005277139670895546,
      "loss": 2.2432,
      "step": 51994
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.6499119997024536,
      "learning_rate": 0.0005277113039637354,
      "loss": 2.1933,
      "step": 51995
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7265365123748779,
      "learning_rate": 0.0005277086407955803,
      "loss": 2.3761,
      "step": 51996
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.8010738492012024,
      "learning_rate": 0.0005277059775850899,
      "loss": 2.1675,
      "step": 51997
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.6859763264656067,
      "learning_rate": 0.0005277033143322647,
      "loss": 2.2706,
      "step": 51998
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.6980739235877991,
      "learning_rate": 0.0005277006510371051,
      "loss": 2.2771,
      "step": 51999
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.6830967664718628,
      "learning_rate": 0.0005276979876996117,
      "loss": 2.2363,
      "step": 52000
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7550577521324158,
      "learning_rate": 0.0005276953243197849,
      "loss": 1.9593,
      "step": 52001
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.8500865697860718,
      "learning_rate": 0.0005276926608976252,
      "loss": 2.1438,
      "step": 52002
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7138267159461975,
      "learning_rate": 0.0005276899974331333,
      "loss": 2.3281,
      "step": 52003
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7002897262573242,
      "learning_rate": 0.0005276873339263093,
      "loss": 2.1829,
      "step": 52004
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7933881282806396,
      "learning_rate": 0.0005276846703771541,
      "loss": 2.1484,
      "step": 52005
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.678937554359436,
      "learning_rate": 0.000527682006785668,
      "loss": 2.1417,
      "step": 52006
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7226564288139343,
      "learning_rate": 0.0005276793431518515,
      "loss": 2.496,
      "step": 52007
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7771581411361694,
      "learning_rate": 0.0005276766794757051,
      "loss": 2.0178,
      "step": 52008
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.8116739392280579,
      "learning_rate": 0.0005276740157572293,
      "loss": 2.304,
      "step": 52009
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7405073642730713,
      "learning_rate": 0.0005276713519964247,
      "loss": 2.3069,
      "step": 52010
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7140459418296814,
      "learning_rate": 0.0005276686881932915,
      "loss": 2.2035,
      "step": 52011
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.8130918145179749,
      "learning_rate": 0.0005276660243478305,
      "loss": 2.2008,
      "step": 52012
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.6812342405319214,
      "learning_rate": 0.0005276633604600422,
      "loss": 2.0029,
      "step": 52013
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7108038663864136,
      "learning_rate": 0.0005276606965299268,
      "loss": 2.259,
      "step": 52014
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7701746225357056,
      "learning_rate": 0.000527658032557485,
      "loss": 2.2157,
      "step": 52015
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7902392148971558,
      "learning_rate": 0.0005276553685427174,
      "loss": 2.17,
      "step": 52016
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7053056359291077,
      "learning_rate": 0.0005276527044856242,
      "loss": 2.2647,
      "step": 52017
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7374590039253235,
      "learning_rate": 0.0005276500403862063,
      "loss": 2.3172,
      "step": 52018
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7471812963485718,
      "learning_rate": 0.0005276473762444638,
      "loss": 2.0635,
      "step": 52019
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7102420330047607,
      "learning_rate": 0.0005276447120603974,
      "loss": 2.3685,
      "step": 52020
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.8549627065658569,
      "learning_rate": 0.0005276420478340075,
      "loss": 2.1566,
      "step": 52021
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.711323082447052,
      "learning_rate": 0.0005276393835652946,
      "loss": 2.1562,
      "step": 52022
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7734884023666382,
      "learning_rate": 0.0005276367192542594,
      "loss": 2.3607,
      "step": 52023
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7411941289901733,
      "learning_rate": 0.0005276340549009021,
      "loss": 2.4769,
      "step": 52024
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7774087190628052,
      "learning_rate": 0.0005276313905052234,
      "loss": 2.1554,
      "step": 52025
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7087339758872986,
      "learning_rate": 0.0005276287260672237,
      "loss": 2.3699,
      "step": 52026
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.8276612758636475,
      "learning_rate": 0.0005276260615869036,
      "loss": 2.514,
      "step": 52027
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7580741047859192,
      "learning_rate": 0.0005276233970642634,
      "loss": 2.0964,
      "step": 52028
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.6577178239822388,
      "learning_rate": 0.0005276207324993039,
      "loss": 2.2561,
      "step": 52029
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.8015628457069397,
      "learning_rate": 0.0005276180678920252,
      "loss": 2.4915,
      "step": 52030
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.8709797859191895,
      "learning_rate": 0.0005276154032424281,
      "loss": 2.1543,
      "step": 52031
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7329632043838501,
      "learning_rate": 0.000527612738550513,
      "loss": 2.0367,
      "step": 52032
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.74880450963974,
      "learning_rate": 0.0005276100738162803,
      "loss": 2.0849,
      "step": 52033
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.8216313719749451,
      "learning_rate": 0.0005276074090397307,
      "loss": 2.0578,
      "step": 52034
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.9704417586326599,
      "learning_rate": 0.0005276047442208646,
      "loss": 2.006,
      "step": 52035
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7177867293357849,
      "learning_rate": 0.0005276020793596824,
      "loss": 2.3982,
      "step": 52036
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7407001852989197,
      "learning_rate": 0.0005275994144561848,
      "loss": 2.141,
      "step": 52037
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7376348376274109,
      "learning_rate": 0.000527596749510372,
      "loss": 2.357,
      "step": 52038
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7211459279060364,
      "learning_rate": 0.0005275940845222448,
      "loss": 2.2546,
      "step": 52039
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.6991474032402039,
      "learning_rate": 0.0005275914194918035,
      "loss": 2.3092,
      "step": 52040
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.6895773410797119,
      "learning_rate": 0.0005275887544190487,
      "loss": 2.1908,
      "step": 52041
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.8166873455047607,
      "learning_rate": 0.0005275860893039809,
      "loss": 2.286,
      "step": 52042
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.8044959306716919,
      "learning_rate": 0.0005275834241466004,
      "loss": 2.2782,
      "step": 52043
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.8261479735374451,
      "learning_rate": 0.000527580758946908,
      "loss": 2.3181,
      "step": 52044
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.6663519144058228,
      "learning_rate": 0.0005275780937049039,
      "loss": 2.0277,
      "step": 52045
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7515125870704651,
      "learning_rate": 0.0005275754284205888,
      "loss": 2.329,
      "step": 52046
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7505801320075989,
      "learning_rate": 0.0005275727630939632,
      "loss": 2.5637,
      "step": 52047
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.8266546130180359,
      "learning_rate": 0.0005275700977250274,
      "loss": 2.2233,
      "step": 52048
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.8036602139472961,
      "learning_rate": 0.000527567432313782,
      "loss": 2.2972,
      "step": 52049
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.6790658831596375,
      "learning_rate": 0.0005275647668602277,
      "loss": 2.3704,
      "step": 52050
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7912920713424683,
      "learning_rate": 0.0005275621013643647,
      "loss": 2.4894,
      "step": 52051
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7395086288452148,
      "learning_rate": 0.0005275594358261936,
      "loss": 2.2821,
      "step": 52052
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.6881182193756104,
      "learning_rate": 0.000527556770245715,
      "loss": 2.2854,
      "step": 52053
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.746446430683136,
      "learning_rate": 0.0005275541046229293,
      "loss": 2.1716,
      "step": 52054
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.8561513423919678,
      "learning_rate": 0.0005275514389578369,
      "loss": 2.4364,
      "step": 52055
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.6834737658500671,
      "learning_rate": 0.0005275487732504384,
      "loss": 2.1781,
      "step": 52056
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7251130938529968,
      "learning_rate": 0.0005275461075007343,
      "loss": 2.2122,
      "step": 52057
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.940717875957489,
      "learning_rate": 0.0005275434417087252,
      "loss": 2.2018,
      "step": 52058
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7813815474510193,
      "learning_rate": 0.0005275407758744113,
      "loss": 2.1447,
      "step": 52059
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7642827033996582,
      "learning_rate": 0.0005275381099977933,
      "loss": 2.2404,
      "step": 52060
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.9283407330513,
      "learning_rate": 0.0005275354440788717,
      "loss": 2.2377,
      "step": 52061
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.8972592353820801,
      "learning_rate": 0.0005275327781176471,
      "loss": 2.0848,
      "step": 52062
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7523006200790405,
      "learning_rate": 0.0005275301121141198,
      "loss": 2.2105,
      "step": 52063
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.8188791275024414,
      "learning_rate": 0.0005275274460682902,
      "loss": 2.2673,
      "step": 52064
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.8977470993995667,
      "learning_rate": 0.000527524779980159,
      "loss": 2.3606,
      "step": 52065
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7332227826118469,
      "learning_rate": 0.0005275221138497268,
      "loss": 2.0649,
      "step": 52066
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.760757565498352,
      "learning_rate": 0.0005275194476769938,
      "loss": 2.0787,
      "step": 52067
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7846432328224182,
      "learning_rate": 0.0005275167814619607,
      "loss": 2.2117,
      "step": 52068
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.0091359615325928,
      "learning_rate": 0.0005275141152046279,
      "loss": 2.1998,
      "step": 52069
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.715233564376831,
      "learning_rate": 0.000527511448904996,
      "loss": 2.2709,
      "step": 52070
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7510091066360474,
      "learning_rate": 0.0005275087825630654,
      "loss": 2.2574,
      "step": 52071
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.534347653388977,
      "learning_rate": 0.0005275061161788366,
      "loss": 2.0096,
      "step": 52072
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.9299342036247253,
      "learning_rate": 0.0005275034497523102,
      "loss": 2.1241,
      "step": 52073
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7200195789337158,
      "learning_rate": 0.0005275007832834864,
      "loss": 2.1399,
      "step": 52074
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.1424611806869507,
      "learning_rate": 0.0005274981167723661,
      "loss": 2.4291,
      "step": 52075
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.0018486976623535,
      "learning_rate": 0.0005274954502189496,
      "loss": 2.3305,
      "step": 52076
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7119179368019104,
      "learning_rate": 0.0005274927836232374,
      "loss": 2.1727,
      "step": 52077
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7376663684844971,
      "learning_rate": 0.0005274901169852299,
      "loss": 2.1732,
      "step": 52078
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.8617150187492371,
      "learning_rate": 0.0005274874503049278,
      "loss": 2.2694,
      "step": 52079
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.8217859864234924,
      "learning_rate": 0.0005274847835823314,
      "loss": 2.3872,
      "step": 52080
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.6900064945220947,
      "learning_rate": 0.0005274821168174413,
      "loss": 2.0481,
      "step": 52081
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.8136720061302185,
      "learning_rate": 0.0005274794500102581,
      "loss": 2.1375,
      "step": 52082
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.775534451007843,
      "learning_rate": 0.000527476783160782,
      "loss": 2.1099,
      "step": 52083
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7482127547264099,
      "learning_rate": 0.0005274741162690138,
      "loss": 2.2809,
      "step": 52084
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.6935163736343384,
      "learning_rate": 0.0005274714493349539,
      "loss": 2.1515,
      "step": 52085
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.8243829011917114,
      "learning_rate": 0.0005274687823586026,
      "loss": 2.476,
      "step": 52086
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7505190372467041,
      "learning_rate": 0.0005274661153399607,
      "loss": 2.2476,
      "step": 52087
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7681111693382263,
      "learning_rate": 0.0005274634482790284,
      "loss": 2.2829,
      "step": 52088
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.6871037483215332,
      "learning_rate": 0.0005274607811758064,
      "loss": 2.5186,
      "step": 52089
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7376245856285095,
      "learning_rate": 0.0005274581140302953,
      "loss": 2.4633,
      "step": 52090
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.6967019438743591,
      "learning_rate": 0.0005274554468424953,
      "loss": 2.3807,
      "step": 52091
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7334617376327515,
      "learning_rate": 0.000527452779612407,
      "loss": 2.3904,
      "step": 52092
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.6651650667190552,
      "learning_rate": 0.000527450112340031,
      "loss": 2.2898,
      "step": 52093
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7788427472114563,
      "learning_rate": 0.0005274474450253678,
      "loss": 2.1771,
      "step": 52094
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7104447484016418,
      "learning_rate": 0.0005274447776684178,
      "loss": 2.2833,
      "step": 52095
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7176573872566223,
      "learning_rate": 0.0005274421102691815,
      "loss": 2.19,
      "step": 52096
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7874736785888672,
      "learning_rate": 0.0005274394428276594,
      "loss": 2.3247,
      "step": 52097
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7034290432929993,
      "learning_rate": 0.000527436775343852,
      "loss": 2.2141,
      "step": 52098
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7177481055259705,
      "learning_rate": 0.0005274341078177598,
      "loss": 2.2059,
      "step": 52099
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.6541529297828674,
      "learning_rate": 0.0005274314402493834,
      "loss": 2.1965,
      "step": 52100
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.8072791695594788,
      "learning_rate": 0.0005274287726387231,
      "loss": 2.5014,
      "step": 52101
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.6815269589424133,
      "learning_rate": 0.0005274261049857796,
      "loss": 2.2186,
      "step": 52102
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7507996559143066,
      "learning_rate": 0.0005274234372905532,
      "loss": 2.5652,
      "step": 52103
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7388901710510254,
      "learning_rate": 0.0005274207695530446,
      "loss": 2.041,
      "step": 52104
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7184911370277405,
      "learning_rate": 0.000527418101773254,
      "loss": 2.2861,
      "step": 52105
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7419947981834412,
      "learning_rate": 0.0005274154339511822,
      "loss": 2.1509,
      "step": 52106
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7073475122451782,
      "learning_rate": 0.0005274127660868296,
      "loss": 2.1028,
      "step": 52107
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.752430260181427,
      "learning_rate": 0.0005274100981801967,
      "loss": 2.3164,
      "step": 52108
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7163910269737244,
      "learning_rate": 0.0005274074302312839,
      "loss": 2.4169,
      "step": 52109
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.644096851348877,
      "learning_rate": 0.0005274047622400919,
      "loss": 2.2072,
      "step": 52110
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7075099945068359,
      "learning_rate": 0.0005274020942066209,
      "loss": 2.2177,
      "step": 52111
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7591270804405212,
      "learning_rate": 0.0005273994261308715,
      "loss": 2.2984,
      "step": 52112
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7005040049552917,
      "learning_rate": 0.0005273967580128444,
      "loss": 2.1298,
      "step": 52113
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.8910510540008545,
      "learning_rate": 0.0005273940898525398,
      "loss": 2.1628,
      "step": 52114
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.667218804359436,
      "learning_rate": 0.0005273914216499585,
      "loss": 2.3104,
      "step": 52115
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7088950872421265,
      "learning_rate": 0.0005273887534051008,
      "loss": 2.4517,
      "step": 52116
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7277560830116272,
      "learning_rate": 0.0005273860851179673,
      "loss": 2.2802,
      "step": 52117
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7100950479507446,
      "learning_rate": 0.0005273834167885583,
      "loss": 2.2412,
      "step": 52118
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.8055269718170166,
      "learning_rate": 0.0005273807484168745,
      "loss": 2.2148,
      "step": 52119
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.711597740650177,
      "learning_rate": 0.0005273780800029164,
      "loss": 2.3891,
      "step": 52120
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7011733055114746,
      "learning_rate": 0.0005273754115466843,
      "loss": 2.4086,
      "step": 52121
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.6905151009559631,
      "learning_rate": 0.0005273727430481789,
      "loss": 2.2797,
      "step": 52122
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.8205577731132507,
      "learning_rate": 0.0005273700745074006,
      "loss": 2.1959,
      "step": 52123
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7654723525047302,
      "learning_rate": 0.0005273674059243498,
      "loss": 2.3187,
      "step": 52124
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7684164643287659,
      "learning_rate": 0.0005273647372990273,
      "loss": 2.0661,
      "step": 52125
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.696648359298706,
      "learning_rate": 0.0005273620686314333,
      "loss": 2.4894,
      "step": 52126
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.701534628868103,
      "learning_rate": 0.0005273593999215683,
      "loss": 2.4506,
      "step": 52127
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.6748150587081909,
      "learning_rate": 0.000527356731169433,
      "loss": 2.1881,
      "step": 52128
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7375304102897644,
      "learning_rate": 0.0005273540623750278,
      "loss": 2.2165,
      "step": 52129
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7259398698806763,
      "learning_rate": 0.0005273513935383533,
      "loss": 2.1693,
      "step": 52130
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7359548807144165,
      "learning_rate": 0.0005273487246594097,
      "loss": 2.2152,
      "step": 52131
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.8522310256958008,
      "learning_rate": 0.0005273460557381978,
      "loss": 2.3411,
      "step": 52132
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7490527033805847,
      "learning_rate": 0.0005273433867747178,
      "loss": 2.3287,
      "step": 52133
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.8115867376327515,
      "learning_rate": 0.0005273407177689706,
      "loss": 2.0157,
      "step": 52134
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.8273162245750427,
      "learning_rate": 0.0005273380487209563,
      "loss": 2.1401,
      "step": 52135
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7766084671020508,
      "learning_rate": 0.0005273353796306756,
      "loss": 2.4188,
      "step": 52136
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7891420722007751,
      "learning_rate": 0.000527332710498129,
      "loss": 2.359,
      "step": 52137
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.6953263282775879,
      "learning_rate": 0.0005273300413233171,
      "loss": 2.479,
      "step": 52138
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7706308364868164,
      "learning_rate": 0.00052732737210624,
      "loss": 1.9624,
      "step": 52139
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.705066978931427,
      "learning_rate": 0.0005273247028468985,
      "loss": 2.1673,
      "step": 52140
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7450765371322632,
      "learning_rate": 0.0005273220335452932,
      "loss": 2.1066,
      "step": 52141
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.753392219543457,
      "learning_rate": 0.0005273193642014243,
      "loss": 2.246,
      "step": 52142
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.6852813363075256,
      "learning_rate": 0.0005273166948152925,
      "loss": 2.1635,
      "step": 52143
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7312213182449341,
      "learning_rate": 0.0005273140253868983,
      "loss": 2.1997,
      "step": 52144
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7109132409095764,
      "learning_rate": 0.0005273113559162419,
      "loss": 2.1931,
      "step": 52145
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7408182621002197,
      "learning_rate": 0.0005273086864033242,
      "loss": 2.4011,
      "step": 52146
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.6967127323150635,
      "learning_rate": 0.0005273060168481456,
      "loss": 2.3589,
      "step": 52147
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.6950470805168152,
      "learning_rate": 0.0005273033472507064,
      "loss": 2.382,
      "step": 52148
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7179216146469116,
      "learning_rate": 0.0005273006776110073,
      "loss": 2.196,
      "step": 52149
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.6965482234954834,
      "learning_rate": 0.0005272980079290485,
      "loss": 2.1247,
      "step": 52150
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7320868372917175,
      "learning_rate": 0.0005272953382048309,
      "loss": 2.3147,
      "step": 52151
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.6715115904808044,
      "learning_rate": 0.0005272926684383549,
      "loss": 2.4808,
      "step": 52152
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.786960244178772,
      "learning_rate": 0.0005272899986296207,
      "loss": 2.1509,
      "step": 52153
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7324439883232117,
      "learning_rate": 0.0005272873287786291,
      "loss": 2.3978,
      "step": 52154
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.6809563636779785,
      "learning_rate": 0.0005272846588853804,
      "loss": 2.0791,
      "step": 52155
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7223359942436218,
      "learning_rate": 0.0005272819889498753,
      "loss": 2.2791,
      "step": 52156
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7436626553535461,
      "learning_rate": 0.0005272793189721142,
      "loss": 2.0941,
      "step": 52157
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7389787435531616,
      "learning_rate": 0.0005272766489520975,
      "loss": 2.191,
      "step": 52158
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.6963542699813843,
      "learning_rate": 0.0005272739788898258,
      "loss": 2.237,
      "step": 52159
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.862413763999939,
      "learning_rate": 0.0005272713087852995,
      "loss": 2.3423,
      "step": 52160
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7157117128372192,
      "learning_rate": 0.0005272686386385192,
      "loss": 2.3094,
      "step": 52161
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7960439324378967,
      "learning_rate": 0.0005272659684494855,
      "loss": 2.1954,
      "step": 52162
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.8185408115386963,
      "learning_rate": 0.0005272632982181988,
      "loss": 2.3452,
      "step": 52163
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.8016664385795593,
      "learning_rate": 0.0005272606279446595,
      "loss": 2.4917,
      "step": 52164
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.761712908744812,
      "learning_rate": 0.0005272579576288681,
      "loss": 2.194,
      "step": 52165
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.6443120241165161,
      "learning_rate": 0.0005272552872708251,
      "loss": 2.0213,
      "step": 52166
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7192155718803406,
      "learning_rate": 0.0005272526168705311,
      "loss": 2.0586,
      "step": 52167
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.6867431998252869,
      "learning_rate": 0.0005272499464279866,
      "loss": 2.442,
      "step": 52168
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7259256839752197,
      "learning_rate": 0.000527247275943192,
      "loss": 2.1496,
      "step": 52169
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7521314024925232,
      "learning_rate": 0.0005272446054161478,
      "loss": 2.1392,
      "step": 52170
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7293861508369446,
      "learning_rate": 0.0005272419348468547,
      "loss": 2.3488,
      "step": 52171
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7506374716758728,
      "learning_rate": 0.000527239264235313,
      "loss": 2.1744,
      "step": 52172
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.703858494758606,
      "learning_rate": 0.0005272365935815231,
      "loss": 2.3559,
      "step": 52173
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.6758978962898254,
      "learning_rate": 0.0005272339228854857,
      "loss": 2.2881,
      "step": 52174
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.8301745057106018,
      "learning_rate": 0.0005272312521472012,
      "loss": 2.2412,
      "step": 52175
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7824779748916626,
      "learning_rate": 0.0005272285813666702,
      "loss": 2.3306,
      "step": 52176
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.6743761301040649,
      "learning_rate": 0.000527225910543893,
      "loss": 2.4306,
      "step": 52177
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7508247494697571,
      "learning_rate": 0.0005272232396788703,
      "loss": 2.4298,
      "step": 52178
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7769607901573181,
      "learning_rate": 0.0005272205687716024,
      "loss": 2.3471,
      "step": 52179
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.8420594334602356,
      "learning_rate": 0.0005272178978220901,
      "loss": 2.1742,
      "step": 52180
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7646067142486572,
      "learning_rate": 0.0005272152268303336,
      "loss": 2.1589,
      "step": 52181
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7776289582252502,
      "learning_rate": 0.0005272125557963334,
      "loss": 2.3683,
      "step": 52182
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.8473770618438721,
      "learning_rate": 0.0005272098847200903,
      "loss": 2.1846,
      "step": 52183
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7564657926559448,
      "learning_rate": 0.0005272072136016044,
      "loss": 2.1439,
      "step": 52184
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7959662675857544,
      "learning_rate": 0.0005272045424408765,
      "loss": 2.3006,
      "step": 52185
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7073982357978821,
      "learning_rate": 0.000527201871237907,
      "loss": 2.0883,
      "step": 52186
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.6623134016990662,
      "learning_rate": 0.0005271991999926963,
      "loss": 2.0221,
      "step": 52187
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.6755917072296143,
      "learning_rate": 0.000527196528705245,
      "loss": 2.184,
      "step": 52188
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7602079510688782,
      "learning_rate": 0.0005271938573755537,
      "loss": 2.2544,
      "step": 52189
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.701945424079895,
      "learning_rate": 0.0005271911860036226,
      "loss": 2.3372,
      "step": 52190
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7537352442741394,
      "learning_rate": 0.0005271885145894526,
      "loss": 2.2316,
      "step": 52191
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7174667716026306,
      "learning_rate": 0.0005271858431330436,
      "loss": 2.0856,
      "step": 52192
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7121984362602234,
      "learning_rate": 0.0005271831716343969,
      "loss": 2.2004,
      "step": 52193
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.8142674565315247,
      "learning_rate": 0.0005271805000935122,
      "loss": 2.1777,
      "step": 52194
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7455996870994568,
      "learning_rate": 0.0005271778285103905,
      "loss": 2.2152,
      "step": 52195
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7863047122955322,
      "learning_rate": 0.0005271751568850322,
      "loss": 2.5078,
      "step": 52196
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7334001660346985,
      "learning_rate": 0.0005271724852174376,
      "loss": 2.4197,
      "step": 52197
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.6721019744873047,
      "learning_rate": 0.0005271698135076075,
      "loss": 2.2869,
      "step": 52198
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.8031120896339417,
      "learning_rate": 0.0005271671417555421,
      "loss": 2.3085,
      "step": 52199
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.8060513734817505,
      "learning_rate": 0.0005271644699612422,
      "loss": 2.1634,
      "step": 52200
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7054352760314941,
      "learning_rate": 0.0005271617981247081,
      "loss": 2.1413,
      "step": 52201
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.6988580822944641,
      "learning_rate": 0.0005271591262459402,
      "loss": 2.1115,
      "step": 52202
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.8154978156089783,
      "learning_rate": 0.0005271564543249392,
      "loss": 2.24,
      "step": 52203
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7546161413192749,
      "learning_rate": 0.0005271537823617056,
      "loss": 2.3423,
      "step": 52204
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.669798731803894,
      "learning_rate": 0.0005271511103562398,
      "loss": 2.152,
      "step": 52205
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7323758006095886,
      "learning_rate": 0.0005271484383085423,
      "loss": 2.1838,
      "step": 52206
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7363610863685608,
      "learning_rate": 0.0005271457662186136,
      "loss": 2.3297,
      "step": 52207
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.8093339204788208,
      "learning_rate": 0.0005271430940864542,
      "loss": 2.1877,
      "step": 52208
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.6944441199302673,
      "learning_rate": 0.0005271404219120647,
      "loss": 2.3346,
      "step": 52209
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7246477603912354,
      "learning_rate": 0.0005271377496954454,
      "loss": 2.3049,
      "step": 52210
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.8485720753669739,
      "learning_rate": 0.000527135077436597,
      "loss": 2.2363,
      "step": 52211
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7628086805343628,
      "learning_rate": 0.0005271324051355199,
      "loss": 2.1861,
      "step": 52212
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.6853870153427124,
      "learning_rate": 0.0005271297327922146,
      "loss": 2.2582,
      "step": 52213
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7080605626106262,
      "learning_rate": 0.0005271270604066815,
      "loss": 2.2367,
      "step": 52214
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.8292673230171204,
      "learning_rate": 0.0005271243879789213,
      "loss": 2.1423,
      "step": 52215
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.663796067237854,
      "learning_rate": 0.0005271217155089342,
      "loss": 2.1723,
      "step": 52216
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7316792011260986,
      "learning_rate": 0.0005271190429967211,
      "loss": 2.387,
      "step": 52217
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.8000406622886658,
      "learning_rate": 0.0005271163704422821,
      "loss": 2.3504,
      "step": 52218
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.8537704348564148,
      "learning_rate": 0.000527113697845618,
      "loss": 2.2547,
      "step": 52219
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.8167110681533813,
      "learning_rate": 0.0005271110252067292,
      "loss": 2.3654,
      "step": 52220
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.769371747970581,
      "learning_rate": 0.0005271083525256161,
      "loss": 2.3389,
      "step": 52221
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.8797711133956909,
      "learning_rate": 0.0005271056798022794,
      "loss": 2.2779,
      "step": 52222
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.8506084084510803,
      "learning_rate": 0.0005271030070367193,
      "loss": 2.1963,
      "step": 52223
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.6912000775337219,
      "learning_rate": 0.0005271003342289365,
      "loss": 1.8818,
      "step": 52224
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.8650691509246826,
      "learning_rate": 0.0005270976613789316,
      "loss": 2.105,
      "step": 52225
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.8921017646789551,
      "learning_rate": 0.0005270949884867048,
      "loss": 2.1575,
      "step": 52226
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7629163861274719,
      "learning_rate": 0.0005270923155522568,
      "loss": 2.2237,
      "step": 52227
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.8904317617416382,
      "learning_rate": 0.000527089642575588,
      "loss": 2.2582,
      "step": 52228
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.9280523657798767,
      "learning_rate": 0.0005270869695566991,
      "loss": 2.2017,
      "step": 52229
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.6712998151779175,
      "learning_rate": 0.0005270842964955903,
      "loss": 2.1755,
      "step": 52230
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7582321166992188,
      "learning_rate": 0.0005270816233922624,
      "loss": 2.2751,
      "step": 52231
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7780245542526245,
      "learning_rate": 0.0005270789502467156,
      "loss": 2.387,
      "step": 52232
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.8753184080123901,
      "learning_rate": 0.0005270762770589506,
      "loss": 2.2861,
      "step": 52233
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7707610130310059,
      "learning_rate": 0.0005270736038289677,
      "loss": 1.9197,
      "step": 52234
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.6889654994010925,
      "learning_rate": 0.0005270709305567676,
      "loss": 2.3849,
      "step": 52235
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.8741875886917114,
      "learning_rate": 0.0005270682572423508,
      "loss": 2.4975,
      "step": 52236
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7811597585678101,
      "learning_rate": 0.0005270655838857176,
      "loss": 2.2645,
      "step": 52237
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.6763134002685547,
      "learning_rate": 0.0005270629104868689,
      "loss": 2.3339,
      "step": 52238
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7195675373077393,
      "learning_rate": 0.0005270602370458046,
      "loss": 2.1562,
      "step": 52239
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.8391559720039368,
      "learning_rate": 0.0005270575635625257,
      "loss": 2.1267,
      "step": 52240
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7580419778823853,
      "learning_rate": 0.0005270548900370325,
      "loss": 2.2523,
      "step": 52241
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.6698896288871765,
      "learning_rate": 0.0005270522164693254,
      "loss": 2.2047,
      "step": 52242
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.8139658570289612,
      "learning_rate": 0.0005270495428594051,
      "loss": 2.2814,
      "step": 52243
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7895491123199463,
      "learning_rate": 0.0005270468692072719,
      "loss": 2.2162,
      "step": 52244
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7046624422073364,
      "learning_rate": 0.0005270441955129265,
      "loss": 2.3425,
      "step": 52245
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.747063398361206,
      "learning_rate": 0.0005270415217763693,
      "loss": 2.3432,
      "step": 52246
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7558383345603943,
      "learning_rate": 0.0005270388479976008,
      "loss": 2.4074,
      "step": 52247
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.8174681663513184,
      "learning_rate": 0.0005270361741766214,
      "loss": 2.16,
      "step": 52248
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7358463406562805,
      "learning_rate": 0.0005270335003134318,
      "loss": 2.2503,
      "step": 52249
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7506863474845886,
      "learning_rate": 0.0005270308264080323,
      "loss": 2.274,
      "step": 52250
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.865610659122467,
      "learning_rate": 0.0005270281524604235,
      "loss": 2.2516,
      "step": 52251
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7423844933509827,
      "learning_rate": 0.000527025478470606,
      "loss": 2.1884,
      "step": 52252
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7346447706222534,
      "learning_rate": 0.00052702280443858,
      "loss": 2.2206,
      "step": 52253
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.9046958088874817,
      "learning_rate": 0.0005270201303643463,
      "loss": 2.1337,
      "step": 52254
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.6987512111663818,
      "learning_rate": 0.0005270174562479053,
      "loss": 2.2419,
      "step": 52255
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7908718585968018,
      "learning_rate": 0.0005270147820892573,
      "loss": 2.2854,
      "step": 52256
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7518384456634521,
      "learning_rate": 0.0005270121078884031,
      "loss": 2.1979,
      "step": 52257
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7204270362854004,
      "learning_rate": 0.000527009433645343,
      "loss": 2.2322,
      "step": 52258
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7370567321777344,
      "learning_rate": 0.0005270067593600776,
      "loss": 2.2337,
      "step": 52259
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7684109807014465,
      "learning_rate": 0.0005270040850326073,
      "loss": 2.2712,
      "step": 52260
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7255738377571106,
      "learning_rate": 0.0005270014106629328,
      "loss": 2.3095,
      "step": 52261
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7286087274551392,
      "learning_rate": 0.0005269987362510544,
      "loss": 2.1056,
      "step": 52262
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.6588396430015564,
      "learning_rate": 0.0005269960617969725,
      "loss": 2.2488,
      "step": 52263
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.6954437494277954,
      "learning_rate": 0.000526993387300688,
      "loss": 2.3141,
      "step": 52264
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.8489680290222168,
      "learning_rate": 0.0005269907127622009,
      "loss": 2.3301,
      "step": 52265
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7780084609985352,
      "learning_rate": 0.000526988038181512,
      "loss": 2.2009,
      "step": 52266
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7533655166625977,
      "learning_rate": 0.0005269853635586218,
      "loss": 2.2846,
      "step": 52267
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7360234260559082,
      "learning_rate": 0.0005269826888935308,
      "loss": 2.1146,
      "step": 52268
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7416353225708008,
      "learning_rate": 0.0005269800141862393,
      "loss": 2.197,
      "step": 52269
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7513028383255005,
      "learning_rate": 0.000526977339436748,
      "loss": 2.2554,
      "step": 52270
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7736968398094177,
      "learning_rate": 0.0005269746646450573,
      "loss": 2.2899,
      "step": 52271
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7362346053123474,
      "learning_rate": 0.0005269719898111678,
      "loss": 2.4482,
      "step": 52272
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7863070368766785,
      "learning_rate": 0.0005269693149350798,
      "loss": 2.2498,
      "step": 52273
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.8783499002456665,
      "learning_rate": 0.000526966640016794,
      "loss": 2.4417,
      "step": 52274
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7383812069892883,
      "learning_rate": 0.0005269639650563108,
      "loss": 2.1902,
      "step": 52275
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7401190400123596,
      "learning_rate": 0.0005269612900536307,
      "loss": 2.3686,
      "step": 52276
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.8800361752510071,
      "learning_rate": 0.0005269586150087542,
      "loss": 2.3914,
      "step": 52277
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7854601740837097,
      "learning_rate": 0.0005269559399216819,
      "loss": 2.2546,
      "step": 52278
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7743712067604065,
      "learning_rate": 0.000526953264792414,
      "loss": 2.1332,
      "step": 52279
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.6746273040771484,
      "learning_rate": 0.0005269505896209514,
      "loss": 2.267,
      "step": 52280
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.8006870746612549,
      "learning_rate": 0.0005269479144072944,
      "loss": 2.2022,
      "step": 52281
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7056775689125061,
      "learning_rate": 0.0005269452391514435,
      "loss": 1.9523,
      "step": 52282
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7598903179168701,
      "learning_rate": 0.0005269425638533991,
      "loss": 2.1723,
      "step": 52283
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.6838046312332153,
      "learning_rate": 0.0005269398885131618,
      "loss": 2.2944,
      "step": 52284
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.898323118686676,
      "learning_rate": 0.0005269372131307321,
      "loss": 2.3616,
      "step": 52285
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7631338238716125,
      "learning_rate": 0.0005269345377061105,
      "loss": 2.3964,
      "step": 52286
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7598839402198792,
      "learning_rate": 0.0005269318622392976,
      "loss": 2.0259,
      "step": 52287
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7759581208229065,
      "learning_rate": 0.0005269291867302936,
      "loss": 2.1461,
      "step": 52288
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7100967764854431,
      "learning_rate": 0.0005269265111790993,
      "loss": 2.4005,
      "step": 52289
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7131756544113159,
      "learning_rate": 0.0005269238355857151,
      "loss": 2.2984,
      "step": 52290
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7279377579689026,
      "learning_rate": 0.0005269211599501413,
      "loss": 2.3503,
      "step": 52291
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7684811949729919,
      "learning_rate": 0.0005269184842723787,
      "loss": 2.2831,
      "step": 52292
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.6935514211654663,
      "learning_rate": 0.0005269158085524277,
      "loss": 2.3124,
      "step": 52293
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7973402738571167,
      "learning_rate": 0.0005269131327902887,
      "loss": 2.4049,
      "step": 52294
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.6948804259300232,
      "learning_rate": 0.0005269104569859623,
      "loss": 2.5466,
      "step": 52295
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7034627795219421,
      "learning_rate": 0.000526907781139449,
      "loss": 2.2269,
      "step": 52296
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7826994061470032,
      "learning_rate": 0.0005269051052507491,
      "loss": 2.2498,
      "step": 52297
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.685045599937439,
      "learning_rate": 0.0005269024293198634,
      "loss": 2.2671,
      "step": 52298
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.6509308218955994,
      "learning_rate": 0.0005268997533467923,
      "loss": 2.1918,
      "step": 52299
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7205871939659119,
      "learning_rate": 0.0005268970773315362,
      "loss": 2.4265,
      "step": 52300
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7523576021194458,
      "learning_rate": 0.0005268944012740956,
      "loss": 2.2944,
      "step": 52301
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.6603253483772278,
      "learning_rate": 0.0005268917251744711,
      "loss": 2.4659,
      "step": 52302
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7432876825332642,
      "learning_rate": 0.000526889049032663,
      "loss": 2.2444,
      "step": 52303
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.8552563190460205,
      "learning_rate": 0.0005268863728486721,
      "loss": 2.3592,
      "step": 52304
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.6845306754112244,
      "learning_rate": 0.0005268836966224987,
      "loss": 2.2766,
      "step": 52305
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7220690846443176,
      "learning_rate": 0.0005268810203541434,
      "loss": 2.302,
      "step": 52306
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7541275024414062,
      "learning_rate": 0.0005268783440436066,
      "loss": 2.1869,
      "step": 52307
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7193408012390137,
      "learning_rate": 0.0005268756676908888,
      "loss": 2.3106,
      "step": 52308
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.6730112433433533,
      "learning_rate": 0.0005268729912959906,
      "loss": 2.478,
      "step": 52309
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7036665081977844,
      "learning_rate": 0.0005268703148589124,
      "loss": 2.3295,
      "step": 52310
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.693580150604248,
      "learning_rate": 0.0005268676383796546,
      "loss": 2.2783,
      "step": 52311
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7495132684707642,
      "learning_rate": 0.0005268649618582179,
      "loss": 2.2648,
      "step": 52312
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7470788359642029,
      "learning_rate": 0.0005268622852946027,
      "loss": 2.2363,
      "step": 52313
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.8517501950263977,
      "learning_rate": 0.0005268596086888096,
      "loss": 2.1798,
      "step": 52314
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7320108413696289,
      "learning_rate": 0.0005268569320408391,
      "loss": 2.042,
      "step": 52315
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7203292846679688,
      "learning_rate": 0.0005268542553506914,
      "loss": 2.0615,
      "step": 52316
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7221013307571411,
      "learning_rate": 0.0005268515786183674,
      "loss": 2.2515,
      "step": 52317
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7979623079299927,
      "learning_rate": 0.0005268489018438673,
      "loss": 2.3869,
      "step": 52318
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7919586300849915,
      "learning_rate": 0.0005268462250271916,
      "loss": 2.3612,
      "step": 52319
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.6867433786392212,
      "learning_rate": 0.000526843548168341,
      "loss": 2.4348,
      "step": 52320
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.6938782930374146,
      "learning_rate": 0.0005268408712673161,
      "loss": 2.0422,
      "step": 52321
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.8127983212471008,
      "learning_rate": 0.000526838194324117,
      "loss": 2.2349,
      "step": 52322
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7091159224510193,
      "learning_rate": 0.0005268355173387443,
      "loss": 2.304,
      "step": 52323
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7326163649559021,
      "learning_rate": 0.0005268328403111987,
      "loss": 1.9102,
      "step": 52324
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7305842041969299,
      "learning_rate": 0.0005268301632414807,
      "loss": 2.3763,
      "step": 52325
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7599400281906128,
      "learning_rate": 0.0005268274861295907,
      "loss": 2.2859,
      "step": 52326
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.688260018825531,
      "learning_rate": 0.000526824808975529,
      "loss": 1.9567,
      "step": 52327
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7962858080863953,
      "learning_rate": 0.0005268221317792964,
      "loss": 2.2483,
      "step": 52328
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7501775026321411,
      "learning_rate": 0.0005268194545408933,
      "loss": 2.2606,
      "step": 52329
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7137885093688965,
      "learning_rate": 0.0005268167772603201,
      "loss": 2.2592,
      "step": 52330
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.6641136407852173,
      "learning_rate": 0.0005268140999375774,
      "loss": 2.1297,
      "step": 52331
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7335128784179688,
      "learning_rate": 0.0005268114225726657,
      "loss": 2.2176,
      "step": 52332
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7536028027534485,
      "learning_rate": 0.0005268087451655855,
      "loss": 2.3288,
      "step": 52333
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.817048966884613,
      "learning_rate": 0.0005268060677163373,
      "loss": 2.1814,
      "step": 52334
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.6998369097709656,
      "learning_rate": 0.0005268033902249214,
      "loss": 2.2797,
      "step": 52335
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7740216851234436,
      "learning_rate": 0.0005268007126913387,
      "loss": 2.0401,
      "step": 52336
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7364920973777771,
      "learning_rate": 0.0005267980351155893,
      "loss": 2.1609,
      "step": 52337
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7306392788887024,
      "learning_rate": 0.000526795357497674,
      "loss": 2.3406,
      "step": 52338
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7437031269073486,
      "learning_rate": 0.000526792679837593,
      "loss": 2.0662,
      "step": 52339
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7183364629745483,
      "learning_rate": 0.0005267900021353471,
      "loss": 2.2078,
      "step": 52340
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7575241923332214,
      "learning_rate": 0.0005267873243909366,
      "loss": 2.2885,
      "step": 52341
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7013100981712341,
      "learning_rate": 0.000526784646604362,
      "loss": 2.2395,
      "step": 52342
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7134010195732117,
      "learning_rate": 0.000526781968775624,
      "loss": 2.0988,
      "step": 52343
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7459614872932434,
      "learning_rate": 0.0005267792909047228,
      "loss": 2.29,
      "step": 52344
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7602498531341553,
      "learning_rate": 0.0005267766129916591,
      "loss": 2.3354,
      "step": 52345
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7794358730316162,
      "learning_rate": 0.0005267739350364333,
      "loss": 2.1974,
      "step": 52346
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7076540589332581,
      "learning_rate": 0.0005267712570390461,
      "loss": 2.2858,
      "step": 52347
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7296298742294312,
      "learning_rate": 0.0005267685789994978,
      "loss": 2.3828,
      "step": 52348
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.8652993440628052,
      "learning_rate": 0.0005267659009177888,
      "loss": 2.1974,
      "step": 52349
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7141794562339783,
      "learning_rate": 0.0005267632227939198,
      "loss": 2.1591,
      "step": 52350
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.8091114163398743,
      "learning_rate": 0.0005267605446278913,
      "loss": 2.3499,
      "step": 52351
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.0322942733764648,
      "learning_rate": 0.0005267578664197038,
      "loss": 2.5103,
      "step": 52352
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7198524475097656,
      "learning_rate": 0.0005267551881693575,
      "loss": 2.5085,
      "step": 52353
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.004563808441162,
      "learning_rate": 0.0005267525098768532,
      "loss": 2.2995,
      "step": 52354
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.8837019205093384,
      "learning_rate": 0.0005267498315421914,
      "loss": 2.284,
      "step": 52355
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.710120439529419,
      "learning_rate": 0.0005267471531653726,
      "loss": 2.0978,
      "step": 52356
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7757323384284973,
      "learning_rate": 0.0005267444747463971,
      "loss": 2.1355,
      "step": 52357
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.2531538009643555,
      "learning_rate": 0.0005267417962852655,
      "loss": 2.4897,
      "step": 52358
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7533309459686279,
      "learning_rate": 0.0005267391177819784,
      "loss": 2.3109,
      "step": 52359
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.6814499497413635,
      "learning_rate": 0.0005267364392365362,
      "loss": 2.2249,
      "step": 52360
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7594985961914062,
      "learning_rate": 0.0005267337606489394,
      "loss": 2.3045,
      "step": 52361
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.9799268841743469,
      "learning_rate": 0.0005267310820191887,
      "loss": 2.526,
      "step": 52362
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7275819182395935,
      "learning_rate": 0.000526728403347284,
      "loss": 2.1787,
      "step": 52363
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.8383139371871948,
      "learning_rate": 0.0005267257246332266,
      "loss": 2.4904,
      "step": 52364
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.8378196954727173,
      "learning_rate": 0.0005267230458770164,
      "loss": 2.3211,
      "step": 52365
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7866012454032898,
      "learning_rate": 0.0005267203670786542,
      "loss": 2.3579,
      "step": 52366
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.6997387409210205,
      "learning_rate": 0.0005267176882381403,
      "loss": 2.4052,
      "step": 52367
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.832621693611145,
      "learning_rate": 0.0005267150093554754,
      "loss": 2.2904,
      "step": 52368
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.9333682656288147,
      "learning_rate": 0.0005267123304306599,
      "loss": 2.0695,
      "step": 52369
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.731697142124176,
      "learning_rate": 0.0005267096514636942,
      "loss": 1.9821,
      "step": 52370
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.6875461935997009,
      "learning_rate": 0.000526706972454579,
      "loss": 2.2053,
      "step": 52371
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.8330911993980408,
      "learning_rate": 0.0005267042934033147,
      "loss": 2.1884,
      "step": 52372
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.8906260132789612,
      "learning_rate": 0.0005267016143099016,
      "loss": 2.2134,
      "step": 52373
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.9011616110801697,
      "learning_rate": 0.0005266989351743407,
      "loss": 2.4051,
      "step": 52374
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.729376494884491,
      "learning_rate": 0.000526696255996632,
      "loss": 2.2391,
      "step": 52375
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.0343459844589233,
      "learning_rate": 0.0005266935767767761,
      "loss": 2.2961,
      "step": 52376
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7524780035018921,
      "learning_rate": 0.0005266908975147738,
      "loss": 2.1824,
      "step": 52377
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.6907772421836853,
      "learning_rate": 0.0005266882182106252,
      "loss": 2.3555,
      "step": 52378
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7238500118255615,
      "learning_rate": 0.0005266855388643311,
      "loss": 2.4784,
      "step": 52379
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7780410051345825,
      "learning_rate": 0.0005266828594758918,
      "loss": 2.3667,
      "step": 52380
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.8493407368659973,
      "learning_rate": 0.000526680180045308,
      "loss": 2.1913,
      "step": 52381
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7412226796150208,
      "learning_rate": 0.0005266775005725799,
      "loss": 2.1089,
      "step": 52382
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.71431565284729,
      "learning_rate": 0.0005266748210577083,
      "loss": 2.079,
      "step": 52383
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.9275910258293152,
      "learning_rate": 0.0005266721415006936,
      "loss": 2.2361,
      "step": 52384
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.8264786601066589,
      "learning_rate": 0.0005266694619015362,
      "loss": 2.4522,
      "step": 52385
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7906240224838257,
      "learning_rate": 0.0005266667822602367,
      "loss": 2.2256,
      "step": 52386
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7748490571975708,
      "learning_rate": 0.0005266641025767955,
      "loss": 2.2783,
      "step": 52387
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.0522526502609253,
      "learning_rate": 0.0005266614228512132,
      "loss": 2.2696,
      "step": 52388
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7668551206588745,
      "learning_rate": 0.0005266587430834903,
      "loss": 2.2726,
      "step": 52389
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7025208473205566,
      "learning_rate": 0.0005266560632736272,
      "loss": 2.2099,
      "step": 52390
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7860619425773621,
      "learning_rate": 0.0005266533834216246,
      "loss": 2.2114,
      "step": 52391
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.9531616568565369,
      "learning_rate": 0.0005266507035274827,
      "loss": 2.4024,
      "step": 52392
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.808776319026947,
      "learning_rate": 0.0005266480235912021,
      "loss": 2.3321,
      "step": 52393
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.6974766850471497,
      "learning_rate": 0.0005266453436127836,
      "loss": 2.2507,
      "step": 52394
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.8433215618133545,
      "learning_rate": 0.0005266426635922273,
      "loss": 2.3114,
      "step": 52395
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7913037538528442,
      "learning_rate": 0.0005266399835295338,
      "loss": 2.1415,
      "step": 52396
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7756460905075073,
      "learning_rate": 0.0005266373034247038,
      "loss": 2.4116,
      "step": 52397
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7421151995658875,
      "learning_rate": 0.0005266346232777375,
      "loss": 2.2831,
      "step": 52398
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.0421537160873413,
      "learning_rate": 0.0005266319430886357,
      "loss": 2.0353,
      "step": 52399
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.73552006483078,
      "learning_rate": 0.0005266292628573986,
      "loss": 2.1885,
      "step": 52400
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.6594297885894775,
      "learning_rate": 0.0005266265825840268,
      "loss": 2.1499,
      "step": 52401
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7902736067771912,
      "learning_rate": 0.000526623902268521,
      "loss": 2.0745,
      "step": 52402
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7553694248199463,
      "learning_rate": 0.0005266212219108815,
      "loss": 2.1309,
      "step": 52403
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7098574042320251,
      "learning_rate": 0.0005266185415111088,
      "loss": 2.3803,
      "step": 52404
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7240614295005798,
      "learning_rate": 0.0005266158610692034,
      "loss": 2.2364,
      "step": 52405
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7497130036354065,
      "learning_rate": 0.0005266131805851658,
      "loss": 2.3362,
      "step": 52406
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7680678963661194,
      "learning_rate": 0.0005266105000589968,
      "loss": 2.1468,
      "step": 52407
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7106555700302124,
      "learning_rate": 0.0005266078194906963,
      "loss": 2.2412,
      "step": 52408
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.80666583776474,
      "learning_rate": 0.0005266051388802653,
      "loss": 2.5394,
      "step": 52409
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7964200377464294,
      "learning_rate": 0.0005266024582277041,
      "loss": 1.8893,
      "step": 52410
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7636934518814087,
      "learning_rate": 0.0005265997775330132,
      "loss": 2.1955,
      "step": 52411
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7150283455848694,
      "learning_rate": 0.0005265970967961932,
      "loss": 2.181,
      "step": 52412
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7925719022750854,
      "learning_rate": 0.0005265944160172444,
      "loss": 2.4702,
      "step": 52413
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.8814831376075745,
      "learning_rate": 0.0005265917351961675,
      "loss": 2.13,
      "step": 52414
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7556796669960022,
      "learning_rate": 0.000526589054332963,
      "loss": 2.4782,
      "step": 52415
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7483487129211426,
      "learning_rate": 0.0005265863734276312,
      "loss": 2.0943,
      "step": 52416
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7248126864433289,
      "learning_rate": 0.0005265836924801728,
      "loss": 2.2524,
      "step": 52417
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.8031917214393616,
      "learning_rate": 0.000526581011490588,
      "loss": 2.0674,
      "step": 52418
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7528381943702698,
      "learning_rate": 0.0005265783304588778,
      "loss": 2.0026,
      "step": 52419
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.6736144423484802,
      "learning_rate": 0.0005265756493850422,
      "loss": 2.2602,
      "step": 52420
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7689792513847351,
      "learning_rate": 0.000526572968269082,
      "loss": 2.2583,
      "step": 52421
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.6938047409057617,
      "learning_rate": 0.0005265702871109976,
      "loss": 2.4079,
      "step": 52422
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.8918132781982422,
      "learning_rate": 0.0005265676059107895,
      "loss": 2.3725,
      "step": 52423
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7057758569717407,
      "learning_rate": 0.0005265649246684583,
      "loss": 2.4026,
      "step": 52424
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.843597948551178,
      "learning_rate": 0.0005265622433840042,
      "loss": 2.1128,
      "step": 52425
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7586615085601807,
      "learning_rate": 0.0005265595620574281,
      "loss": 2.1647,
      "step": 52426
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7037532925605774,
      "learning_rate": 0.0005265568806887302,
      "loss": 2.2689,
      "step": 52427
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.6891437768936157,
      "learning_rate": 0.0005265541992779112,
      "loss": 2.2439,
      "step": 52428
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7625066637992859,
      "learning_rate": 0.0005265515178249713,
      "loss": 2.2386,
      "step": 52429
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7477262020111084,
      "learning_rate": 0.0005265488363299113,
      "loss": 2.1164,
      "step": 52430
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7419695258140564,
      "learning_rate": 0.0005265461547927316,
      "loss": 2.2415,
      "step": 52431
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.6655066013336182,
      "learning_rate": 0.0005265434732134327,
      "loss": 2.2522,
      "step": 52432
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.8310849666595459,
      "learning_rate": 0.0005265407915920151,
      "loss": 2.2518,
      "step": 52433
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7327730059623718,
      "learning_rate": 0.0005265381099284793,
      "loss": 2.3677,
      "step": 52434
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7442024350166321,
      "learning_rate": 0.0005265354282228258,
      "loss": 2.2706,
      "step": 52435
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7797067165374756,
      "learning_rate": 0.000526532746475055,
      "loss": 2.3417,
      "step": 52436
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.801776647567749,
      "learning_rate": 0.0005265300646851676,
      "loss": 2.1076,
      "step": 52437
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7101714015007019,
      "learning_rate": 0.0005265273828531639,
      "loss": 2.3443,
      "step": 52438
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.8992782831192017,
      "learning_rate": 0.0005265247009790445,
      "loss": 2.3588,
      "step": 52439
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7988971471786499,
      "learning_rate": 0.0005265220190628099,
      "loss": 2.3396,
      "step": 52440
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7372917532920837,
      "learning_rate": 0.0005265193371044606,
      "loss": 2.2742,
      "step": 52441
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.8943454027175903,
      "learning_rate": 0.0005265166551039971,
      "loss": 2.1972,
      "step": 52442
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.9642171859741211,
      "learning_rate": 0.0005265139730614198,
      "loss": 2.2759,
      "step": 52443
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.6864320635795593,
      "learning_rate": 0.0005265112909767293,
      "loss": 2.2023,
      "step": 52444
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7950118780136108,
      "learning_rate": 0.0005265086088499261,
      "loss": 2.194,
      "step": 52445
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.8784319758415222,
      "learning_rate": 0.0005265059266810107,
      "loss": 2.2299,
      "step": 52446
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7192083597183228,
      "learning_rate": 0.0005265032444699834,
      "loss": 2.373,
      "step": 52447
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7798753380775452,
      "learning_rate": 0.0005265005622168451,
      "loss": 2.4363,
      "step": 52448
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7244617938995361,
      "learning_rate": 0.000526497879921596,
      "loss": 1.9969,
      "step": 52449
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7944445013999939,
      "learning_rate": 0.0005264951975842366,
      "loss": 2.3107,
      "step": 52450
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.007224440574646,
      "learning_rate": 0.0005264925152047676,
      "loss": 2.0091,
      "step": 52451
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7206945419311523,
      "learning_rate": 0.0005264898327831892,
      "loss": 2.0635,
      "step": 52452
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.9920211434364319,
      "learning_rate": 0.0005264871503195021,
      "loss": 2.4174,
      "step": 52453
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.8328666090965271,
      "learning_rate": 0.0005264844678137067,
      "loss": 2.2782,
      "step": 52454
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.8403591513633728,
      "learning_rate": 0.0005264817852658037,
      "loss": 2.4443,
      "step": 52455
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7106298804283142,
      "learning_rate": 0.0005264791026757935,
      "loss": 2.0561,
      "step": 52456
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.8650690317153931,
      "learning_rate": 0.0005264764200436765,
      "loss": 2.2816,
      "step": 52457
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.792582094669342,
      "learning_rate": 0.0005264737373694532,
      "loss": 2.3761,
      "step": 52458
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7071003913879395,
      "learning_rate": 0.0005264710546531242,
      "loss": 1.8268,
      "step": 52459
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7976964116096497,
      "learning_rate": 0.0005264683718946899,
      "loss": 2.3352,
      "step": 52460
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7315361499786377,
      "learning_rate": 0.000526465689094151,
      "loss": 2.3612,
      "step": 52461
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.0849443674087524,
      "learning_rate": 0.0005264630062515078,
      "loss": 2.3681,
      "step": 52462
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7579601407051086,
      "learning_rate": 0.0005264603233667607,
      "loss": 2.0595,
      "step": 52463
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7176471948623657,
      "learning_rate": 0.0005264576404399106,
      "loss": 2.1229,
      "step": 52464
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.8599271178245544,
      "learning_rate": 0.0005264549574709576,
      "loss": 2.24,
      "step": 52465
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.8111497759819031,
      "learning_rate": 0.0005264522744599023,
      "loss": 2.3995,
      "step": 52466
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7851267457008362,
      "learning_rate": 0.0005264495914067453,
      "loss": 2.2901,
      "step": 52467
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.8579861521720886,
      "learning_rate": 0.0005264469083114871,
      "loss": 2.4498,
      "step": 52468
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.8512845635414124,
      "learning_rate": 0.0005264442251741281,
      "loss": 2.1925,
      "step": 52469
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.9080467224121094,
      "learning_rate": 0.0005264415419946689,
      "loss": 2.3997,
      "step": 52470
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7008788585662842,
      "learning_rate": 0.0005264388587731099,
      "loss": 2.3517,
      "step": 52471
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7842128872871399,
      "learning_rate": 0.0005264361755094516,
      "loss": 2.2203,
      "step": 52472
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.8872619271278381,
      "learning_rate": 0.0005264334922036947,
      "loss": 2.2198,
      "step": 52473
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7090702056884766,
      "learning_rate": 0.0005264308088558393,
      "loss": 2.1206,
      "step": 52474
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7229543328285217,
      "learning_rate": 0.0005264281254658863,
      "loss": 2.1527,
      "step": 52475
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.8385592103004456,
      "learning_rate": 0.0005264254420338359,
      "loss": 2.5484,
      "step": 52476
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.79432612657547,
      "learning_rate": 0.0005264227585596888,
      "loss": 2.2029,
      "step": 52477
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.6382212042808533,
      "learning_rate": 0.0005264200750434455,
      "loss": 1.9019,
      "step": 52478
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7055817246437073,
      "learning_rate": 0.0005264173914851064,
      "loss": 2.3988,
      "step": 52479
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.8001192808151245,
      "learning_rate": 0.0005264147078846721,
      "loss": 2.3428,
      "step": 52480
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.8445241451263428,
      "learning_rate": 0.000526412024242143,
      "loss": 2.3885,
      "step": 52481
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.6230365037918091,
      "learning_rate": 0.0005264093405575195,
      "loss": 2.3494,
      "step": 52482
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7731276154518127,
      "learning_rate": 0.0005264066568308024,
      "loss": 2.3217,
      "step": 52483
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7742169499397278,
      "learning_rate": 0.0005264039730619918,
      "loss": 2.3528,
      "step": 52484
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.6557273864746094,
      "learning_rate": 0.0005264012892510887,
      "loss": 2.3174,
      "step": 52485
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7017072439193726,
      "learning_rate": 0.0005263986053980932,
      "loss": 2.3139,
      "step": 52486
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7401716709136963,
      "learning_rate": 0.000526395921503006,
      "loss": 2.078,
      "step": 52487
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7148228287696838,
      "learning_rate": 0.0005263932375658275,
      "loss": 2.2085,
      "step": 52488
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.679520845413208,
      "learning_rate": 0.0005263905535865581,
      "loss": 2.1583,
      "step": 52489
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7978009581565857,
      "learning_rate": 0.0005263878695651985,
      "loss": 2.1961,
      "step": 52490
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.8178662657737732,
      "learning_rate": 0.0005263851855017491,
      "loss": 2.2603,
      "step": 52491
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7284256219863892,
      "learning_rate": 0.0005263825013962104,
      "loss": 2.229,
      "step": 52492
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7725634574890137,
      "learning_rate": 0.000526379817248583,
      "loss": 1.9716,
      "step": 52493
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.8019877076148987,
      "learning_rate": 0.0005263771330588673,
      "loss": 2.1449,
      "step": 52494
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.6939594149589539,
      "learning_rate": 0.0005263744488270638,
      "loss": 2.2958,
      "step": 52495
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.6964752674102783,
      "learning_rate": 0.000526371764553173,
      "loss": 2.344,
      "step": 52496
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7236443758010864,
      "learning_rate": 0.0005263690802371954,
      "loss": 2.3804,
      "step": 52497
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.8128368854522705,
      "learning_rate": 0.0005263663958791316,
      "loss": 2.2881,
      "step": 52498
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.6981826424598694,
      "learning_rate": 0.0005263637114789819,
      "loss": 2.2544,
      "step": 52499
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7608402967453003,
      "learning_rate": 0.000526361027036747,
      "loss": 2.2105,
      "step": 52500
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.721717119216919,
      "learning_rate": 0.0005263583425524273,
      "loss": 2.145,
      "step": 52501
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.786247193813324,
      "learning_rate": 0.0005263556580260233,
      "loss": 2.5329,
      "step": 52502
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7771267294883728,
      "learning_rate": 0.0005263529734575355,
      "loss": 2.2616,
      "step": 52503
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.6573849320411682,
      "learning_rate": 0.0005263502888469644,
      "loss": 2.014,
      "step": 52504
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7301244735717773,
      "learning_rate": 0.0005263476041943106,
      "loss": 2.1535,
      "step": 52505
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7178996801376343,
      "learning_rate": 0.0005263449194995745,
      "loss": 2.1435,
      "step": 52506
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.8842126727104187,
      "learning_rate": 0.0005263422347627565,
      "loss": 2.2526,
      "step": 52507
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.8120996356010437,
      "learning_rate": 0.0005263395499838572,
      "loss": 2.0708,
      "step": 52508
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7708507180213928,
      "learning_rate": 0.0005263368651628771,
      "loss": 2.1498,
      "step": 52509
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.8084867596626282,
      "learning_rate": 0.0005263341802998168,
      "loss": 1.9994,
      "step": 52510
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.6577411890029907,
      "learning_rate": 0.0005263314953946767,
      "loss": 2.1534,
      "step": 52511
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.9042302966117859,
      "learning_rate": 0.0005263288104474573,
      "loss": 2.3201,
      "step": 52512
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7670177817344666,
      "learning_rate": 0.0005263261254581591,
      "loss": 2.2411,
      "step": 52513
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7240369915962219,
      "learning_rate": 0.0005263234404267825,
      "loss": 2.3229,
      "step": 52514
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.8202215433120728,
      "learning_rate": 0.0005263207553533281,
      "loss": 2.2888,
      "step": 52515
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.8719095587730408,
      "learning_rate": 0.0005263180702377965,
      "loss": 2.1505,
      "step": 52516
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7188726663589478,
      "learning_rate": 0.0005263153850801881,
      "loss": 2.1172,
      "step": 52517
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.6806894540786743,
      "learning_rate": 0.0005263126998805033,
      "loss": 2.2287,
      "step": 52518
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.9551688432693481,
      "learning_rate": 0.0005263100146387428,
      "loss": 2.0858,
      "step": 52519
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.8713544607162476,
      "learning_rate": 0.0005263073293549069,
      "loss": 2.242,
      "step": 52520
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7237768173217773,
      "learning_rate": 0.0005263046440289963,
      "loss": 2.3383,
      "step": 52521
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.924749493598938,
      "learning_rate": 0.0005263019586610113,
      "loss": 2.1529,
      "step": 52522
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.0960391759872437,
      "learning_rate": 0.0005262992732509526,
      "loss": 2.3137,
      "step": 52523
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7000110745429993,
      "learning_rate": 0.0005262965877988205,
      "loss": 2.1242,
      "step": 52524
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.8667728304862976,
      "learning_rate": 0.0005262939023046156,
      "loss": 2.1617,
      "step": 52525
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.0459303855895996,
      "learning_rate": 0.0005262912167683385,
      "loss": 2.1038,
      "step": 52526
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.6710366606712341,
      "learning_rate": 0.0005262885311899894,
      "loss": 2.4603,
      "step": 52527
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7167118191719055,
      "learning_rate": 0.0005262858455695692,
      "loss": 2.3029,
      "step": 52528
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7966135144233704,
      "learning_rate": 0.000526283159907078,
      "loss": 2.3168,
      "step": 52529
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.5076712369918823,
      "learning_rate": 0.0005262804742025166,
      "loss": 2.1803,
      "step": 52530
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7575048208236694,
      "learning_rate": 0.0005262777884558854,
      "loss": 2.2926,
      "step": 52531
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.75645911693573,
      "learning_rate": 0.0005262751026671849,
      "loss": 2.0236,
      "step": 52532
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.8400801420211792,
      "learning_rate": 0.0005262724168364156,
      "loss": 2.3953,
      "step": 52533
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.958049476146698,
      "learning_rate": 0.0005262697309635779,
      "loss": 1.9257,
      "step": 52534
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7195727825164795,
      "learning_rate": 0.0005262670450486723,
      "loss": 2.1423,
      "step": 52535
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7843804359436035,
      "learning_rate": 0.0005262643590916995,
      "loss": 2.1916,
      "step": 52536
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.82376629114151,
      "learning_rate": 0.0005262616730926599,
      "loss": 2.2048,
      "step": 52537
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7711527347564697,
      "learning_rate": 0.000526258987051554,
      "loss": 2.2061,
      "step": 52538
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7195176482200623,
      "learning_rate": 0.0005262563009683823,
      "loss": 2.1979,
      "step": 52539
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7905930876731873,
      "learning_rate": 0.0005262536148431452,
      "loss": 2.3565,
      "step": 52540
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.6557013392448425,
      "learning_rate": 0.0005262509286758433,
      "loss": 2.3166,
      "step": 52541
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.8554847240447998,
      "learning_rate": 0.0005262482424664771,
      "loss": 2.4567,
      "step": 52542
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.819896399974823,
      "learning_rate": 0.0005262455562150472,
      "loss": 2.1509,
      "step": 52543
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.8307260870933533,
      "learning_rate": 0.0005262428699215537,
      "loss": 2.124,
      "step": 52544
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7558765411376953,
      "learning_rate": 0.0005262401835859976,
      "loss": 2.2103,
      "step": 52545
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7945858836174011,
      "learning_rate": 0.0005262374972083791,
      "loss": 2.0261,
      "step": 52546
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7768086194992065,
      "learning_rate": 0.0005262348107886989,
      "loss": 2.2853,
      "step": 52547
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.6865920424461365,
      "learning_rate": 0.0005262321243269572,
      "loss": 2.2505,
      "step": 52548
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.6918336749076843,
      "learning_rate": 0.0005262294378231547,
      "loss": 2.0918,
      "step": 52549
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.701677680015564,
      "learning_rate": 0.000526226751277292,
      "loss": 2.218,
      "step": 52550
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7575927376747131,
      "learning_rate": 0.0005262240646893695,
      "loss": 2.3038,
      "step": 52551
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.6970164179801941,
      "learning_rate": 0.0005262213780593875,
      "loss": 2.1534,
      "step": 52552
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7159123420715332,
      "learning_rate": 0.0005262186913873467,
      "loss": 2.1819,
      "step": 52553
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.8057989478111267,
      "learning_rate": 0.0005262160046732477,
      "loss": 2.1214,
      "step": 52554
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7576343417167664,
      "learning_rate": 0.0005262133179170909,
      "loss": 2.3275,
      "step": 52555
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7825629711151123,
      "learning_rate": 0.0005262106311188765,
      "loss": 2.1887,
      "step": 52556
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7524517178535461,
      "learning_rate": 0.0005262079442786055,
      "loss": 2.2764,
      "step": 52557
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7685289978981018,
      "learning_rate": 0.0005262052573962782,
      "loss": 2.1592,
      "step": 52558
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.812870979309082,
      "learning_rate": 0.000526202570471895,
      "loss": 2.1455,
      "step": 52559
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.6625479459762573,
      "learning_rate": 0.0005261998835054565,
      "loss": 2.2393,
      "step": 52560
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7311428785324097,
      "learning_rate": 0.0005261971964969631,
      "loss": 2.3287,
      "step": 52561
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7608944177627563,
      "learning_rate": 0.0005261945094464154,
      "loss": 2.3102,
      "step": 52562
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7232862710952759,
      "learning_rate": 0.0005261918223538139,
      "loss": 2.3018,
      "step": 52563
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7116298675537109,
      "learning_rate": 0.000526189135219159,
      "loss": 2.2531,
      "step": 52564
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.6761361360549927,
      "learning_rate": 0.0005261864480424514,
      "loss": 2.4778,
      "step": 52565
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7053866386413574,
      "learning_rate": 0.0005261837608236913,
      "loss": 2.4089,
      "step": 52566
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7875062823295593,
      "learning_rate": 0.0005261810735628794,
      "loss": 2.3972,
      "step": 52567
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7210596799850464,
      "learning_rate": 0.0005261783862600164,
      "loss": 2.4252,
      "step": 52568
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.747523844242096,
      "learning_rate": 0.0005261756989151024,
      "loss": 2.2555,
      "step": 52569
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.6899678111076355,
      "learning_rate": 0.000526173011528138,
      "loss": 2.3541,
      "step": 52570
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.652991533279419,
      "learning_rate": 0.0005261703240991239,
      "loss": 1.9593,
      "step": 52571
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7327809929847717,
      "learning_rate": 0.0005261676366280604,
      "loss": 2.1765,
      "step": 52572
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7227634787559509,
      "learning_rate": 0.0005261649491149481,
      "loss": 2.2322,
      "step": 52573
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7333081364631653,
      "learning_rate": 0.0005261622615597874,
      "loss": 2.2183,
      "step": 52574
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.6749992370605469,
      "learning_rate": 0.000526159573962579,
      "loss": 2.3551,
      "step": 52575
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7743915319442749,
      "learning_rate": 0.0005261568863233231,
      "loss": 2.33,
      "step": 52576
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7508320808410645,
      "learning_rate": 0.0005261541986420205,
      "loss": 2.4026,
      "step": 52577
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7201659679412842,
      "learning_rate": 0.0005261515109186715,
      "loss": 2.318,
      "step": 52578
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7113429307937622,
      "learning_rate": 0.0005261488231532766,
      "loss": 2.2592,
      "step": 52579
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.6755889058113098,
      "learning_rate": 0.0005261461353458364,
      "loss": 2.2295,
      "step": 52580
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.6752930283546448,
      "learning_rate": 0.0005261434474963515,
      "loss": 2.191,
      "step": 52581
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.734408438205719,
      "learning_rate": 0.0005261407596048221,
      "loss": 2.2052,
      "step": 52582
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.6682198643684387,
      "learning_rate": 0.000526138071671249,
      "loss": 2.3078,
      "step": 52583
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7196585536003113,
      "learning_rate": 0.0005261353836956325,
      "loss": 2.0718,
      "step": 52584
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7417873740196228,
      "learning_rate": 0.0005261326956779731,
      "loss": 2.375,
      "step": 52585
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7680883407592773,
      "learning_rate": 0.0005261300076182714,
      "loss": 2.204,
      "step": 52586
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.6748429536819458,
      "learning_rate": 0.0005261273195165279,
      "loss": 2.355,
      "step": 52587
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.8008313179016113,
      "learning_rate": 0.000526124631372743,
      "loss": 2.3427,
      "step": 52588
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.6883481740951538,
      "learning_rate": 0.0005261219431869174,
      "loss": 2.2541,
      "step": 52589
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.6890267133712769,
      "learning_rate": 0.0005261192549590514,
      "loss": 2.4228,
      "step": 52590
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7176741361618042,
      "learning_rate": 0.0005261165666891456,
      "loss": 2.4271,
      "step": 52591
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7860336303710938,
      "learning_rate": 0.0005261138783772003,
      "loss": 2.2909,
      "step": 52592
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7186999917030334,
      "learning_rate": 0.0005261111900232162,
      "loss": 2.4875,
      "step": 52593
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7583852410316467,
      "learning_rate": 0.0005261085016271938,
      "loss": 2.4575,
      "step": 52594
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7218012809753418,
      "learning_rate": 0.0005261058131891337,
      "loss": 2.3535,
      "step": 52595
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7549244165420532,
      "learning_rate": 0.000526103124709036,
      "loss": 2.2272,
      "step": 52596
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7355620861053467,
      "learning_rate": 0.0005261004361869018,
      "loss": 2.3933,
      "step": 52597
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7094995379447937,
      "learning_rate": 0.000526097747622731,
      "loss": 2.3181,
      "step": 52598
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7899034023284912,
      "learning_rate": 0.0005260950590165244,
      "loss": 2.3569,
      "step": 52599
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.6298549771308899,
      "learning_rate": 0.0005260923703682825,
      "loss": 2.396,
      "step": 52600
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7565317749977112,
      "learning_rate": 0.0005260896816780057,
      "loss": 2.2426,
      "step": 52601
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7424192428588867,
      "learning_rate": 0.0005260869929456946,
      "loss": 2.3614,
      "step": 52602
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7123489379882812,
      "learning_rate": 0.0005260843041713497,
      "loss": 2.4409,
      "step": 52603
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.6987264156341553,
      "learning_rate": 0.0005260816153549714,
      "loss": 2.2191,
      "step": 52604
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.8244219422340393,
      "learning_rate": 0.0005260789264965603,
      "loss": 2.2762,
      "step": 52605
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.8058084845542908,
      "learning_rate": 0.0005260762375961168,
      "loss": 2.1746,
      "step": 52606
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7405923008918762,
      "learning_rate": 0.0005260735486536415,
      "loss": 2.3506,
      "step": 52607
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7528248429298401,
      "learning_rate": 0.0005260708596691349,
      "loss": 2.1877,
      "step": 52608
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.8985981345176697,
      "learning_rate": 0.0005260681706425974,
      "loss": 2.3004,
      "step": 52609
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7919903993606567,
      "learning_rate": 0.0005260654815740296,
      "loss": 2.3444,
      "step": 52610
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.6739681959152222,
      "learning_rate": 0.0005260627924634319,
      "loss": 1.9727,
      "step": 52611
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.8186993598937988,
      "learning_rate": 0.000526060103310805,
      "loss": 2.2711,
      "step": 52612
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.8731619715690613,
      "learning_rate": 0.0005260574141161492,
      "loss": 2.0631,
      "step": 52613
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.6678028106689453,
      "learning_rate": 0.0005260547248794649,
      "loss": 2.3238,
      "step": 52614
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.733615517616272,
      "learning_rate": 0.000526052035600753,
      "loss": 2.3354,
      "step": 52615
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.8001127243041992,
      "learning_rate": 0.0005260493462800137,
      "loss": 2.2909,
      "step": 52616
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7426249384880066,
      "learning_rate": 0.0005260466569172474,
      "loss": 2.3179,
      "step": 52617
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.8220246434211731,
      "learning_rate": 0.0005260439675124548,
      "loss": 2.1781,
      "step": 52618
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.8265212774276733,
      "learning_rate": 0.0005260412780656365,
      "loss": 2.3087,
      "step": 52619
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7697346806526184,
      "learning_rate": 0.0005260385885767927,
      "loss": 2.2177,
      "step": 52620
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7017797231674194,
      "learning_rate": 0.0005260358990459242,
      "loss": 2.2615,
      "step": 52621
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.776643693447113,
      "learning_rate": 0.0005260332094730313,
      "loss": 2.3851,
      "step": 52622
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7887623906135559,
      "learning_rate": 0.0005260305198581146,
      "loss": 2.3883,
      "step": 52623
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7839617133140564,
      "learning_rate": 0.0005260278302011745,
      "loss": 2.4386,
      "step": 52624
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7016604542732239,
      "learning_rate": 0.0005260251405022116,
      "loss": 2.344,
      "step": 52625
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.8758454322814941,
      "learning_rate": 0.0005260224507612263,
      "loss": 2.2815,
      "step": 52626
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7770757675170898,
      "learning_rate": 0.0005260197609782192,
      "loss": 2.2682,
      "step": 52627
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7326686382293701,
      "learning_rate": 0.0005260170711531908,
      "loss": 2.1479,
      "step": 52628
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7165445685386658,
      "learning_rate": 0.0005260143812861415,
      "loss": 2.2084,
      "step": 52629
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.735108494758606,
      "learning_rate": 0.0005260116913770719,
      "loss": 2.3626,
      "step": 52630
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.8672468662261963,
      "learning_rate": 0.0005260090014259824,
      "loss": 2.2496,
      "step": 52631
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7825750708580017,
      "learning_rate": 0.0005260063114328736,
      "loss": 2.2966,
      "step": 52632
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.6774574518203735,
      "learning_rate": 0.000526003621397746,
      "loss": 2.4186,
      "step": 52633
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.8347781300544739,
      "learning_rate": 0.0005260009313206,
      "loss": 2.3579,
      "step": 52634
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.8503957986831665,
      "learning_rate": 0.0005259982412014362,
      "loss": 2.2395,
      "step": 52635
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7998604774475098,
      "learning_rate": 0.000525995551040255,
      "loss": 2.1918,
      "step": 52636
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7383777499198914,
      "learning_rate": 0.000525992860837057,
      "loss": 2.3218,
      "step": 52637
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7844898104667664,
      "learning_rate": 0.0005259901705918427,
      "loss": 2.3477,
      "step": 52638
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.8973482847213745,
      "learning_rate": 0.0005259874803046125,
      "loss": 2.1894,
      "step": 52639
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.8579530715942383,
      "learning_rate": 0.000525984789975367,
      "loss": 2.3589,
      "step": 52640
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7376261353492737,
      "learning_rate": 0.0005259820996041067,
      "loss": 2.2493,
      "step": 52641
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.741385281085968,
      "learning_rate": 0.000525979409190832,
      "loss": 2.1768,
      "step": 52642
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.8665420413017273,
      "learning_rate": 0.0005259767187355435,
      "loss": 2.2093,
      "step": 52643
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7358952760696411,
      "learning_rate": 0.0005259740282382415,
      "loss": 2.2913,
      "step": 52644
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7100933194160461,
      "learning_rate": 0.0005259713376989268,
      "loss": 2.4473,
      "step": 52645
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.2186567783355713,
      "learning_rate": 0.0005259686471175998,
      "loss": 2.2442,
      "step": 52646
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.0087337493896484,
      "learning_rate": 0.0005259659564942611,
      "loss": 2.0876,
      "step": 52647
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.6961796283721924,
      "learning_rate": 0.0005259632658289107,
      "loss": 2.2019,
      "step": 52648
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.0992108583450317,
      "learning_rate": 0.0005259605751215497,
      "loss": 2.1829,
      "step": 52649
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.785102128982544,
      "learning_rate": 0.0005259578843721783,
      "loss": 2.2902,
      "step": 52650
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7254629135131836,
      "learning_rate": 0.000525955193580797,
      "loss": 2.1621,
      "step": 52651
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7280879020690918,
      "learning_rate": 0.0005259525027474065,
      "loss": 2.1866,
      "step": 52652
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.0812301635742188,
      "learning_rate": 0.0005259498118720072,
      "loss": 2.1261,
      "step": 52653
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.0165199041366577,
      "learning_rate": 0.0005259471209545994,
      "loss": 2.359,
      "step": 52654
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.6722913980484009,
      "learning_rate": 0.0005259444299951839,
      "loss": 2.3108,
      "step": 52655
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.8395532369613647,
      "learning_rate": 0.000525941738993761,
      "loss": 2.1898,
      "step": 52656
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.096146583557129,
      "learning_rate": 0.0005259390479503313,
      "loss": 2.3155,
      "step": 52657
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.8601599931716919,
      "learning_rate": 0.0005259363568648952,
      "loss": 2.179,
      "step": 52658
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.6743966937065125,
      "learning_rate": 0.0005259336657374533,
      "loss": 2.3173,
      "step": 52659
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.8600103855133057,
      "learning_rate": 0.0005259309745680061,
      "loss": 2.2618,
      "step": 52660
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.09442138671875,
      "learning_rate": 0.000525928283356554,
      "loss": 2.2242,
      "step": 52661
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.725871205329895,
      "learning_rate": 0.0005259255921030977,
      "loss": 2.3963,
      "step": 52662
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.659420907497406,
      "learning_rate": 0.0005259229008076374,
      "loss": 2.2567,
      "step": 52663
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.8745054006576538,
      "learning_rate": 0.0005259202094701738,
      "loss": 2.4183,
      "step": 52664
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.9678267240524292,
      "learning_rate": 0.0005259175180907074,
      "loss": 2.0057,
      "step": 52665
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.6743443012237549,
      "learning_rate": 0.0005259148266692386,
      "loss": 2.3564,
      "step": 52666
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.6992067694664001,
      "learning_rate": 0.0005259121352057681,
      "loss": 2.2589,
      "step": 52667
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.9069302678108215,
      "learning_rate": 0.0005259094437002963,
      "loss": 2.2394,
      "step": 52668
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.6664141416549683,
      "learning_rate": 0.0005259067521528236,
      "loss": 2.0581,
      "step": 52669
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7185280323028564,
      "learning_rate": 0.0005259040605633504,
      "loss": 2.2031,
      "step": 52670
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.6897791624069214,
      "learning_rate": 0.0005259013689318776,
      "loss": 2.3391,
      "step": 52671
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7658784985542297,
      "learning_rate": 0.0005258986772584054,
      "loss": 2.4023,
      "step": 52672
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7727047204971313,
      "learning_rate": 0.0005258959855429344,
      "loss": 2.3859,
      "step": 52673
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7556991577148438,
      "learning_rate": 0.0005258932937854651,
      "loss": 2.4323,
      "step": 52674
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.6447297930717468,
      "learning_rate": 0.0005258906019859978,
      "loss": 2.1339,
      "step": 52675
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.6812108755111694,
      "learning_rate": 0.0005258879101445334,
      "loss": 2.3866,
      "step": 52676
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7353485822677612,
      "learning_rate": 0.000525885218261072,
      "loss": 2.211,
      "step": 52677
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7052185535430908,
      "learning_rate": 0.0005258825263356143,
      "loss": 2.2006,
      "step": 52678
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7665939927101135,
      "learning_rate": 0.0005258798343681609,
      "loss": 2.1754,
      "step": 52679
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.6874117255210876,
      "learning_rate": 0.000525877142358712,
      "loss": 2.3721,
      "step": 52680
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.8340924382209778,
      "learning_rate": 0.0005258744503072683,
      "loss": 2.246,
      "step": 52681
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7380256652832031,
      "learning_rate": 0.0005258717582138305,
      "loss": 2.2042,
      "step": 52682
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7206621170043945,
      "learning_rate": 0.0005258690660783987,
      "loss": 2.3083,
      "step": 52683
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.742056667804718,
      "learning_rate": 0.0005258663739009736,
      "loss": 2.2416,
      "step": 52684
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7570589184761047,
      "learning_rate": 0.0005258636816815556,
      "loss": 1.9414,
      "step": 52685
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.733633279800415,
      "learning_rate": 0.0005258609894201454,
      "loss": 2.1539,
      "step": 52686
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7169069647789001,
      "learning_rate": 0.0005258582971167432,
      "loss": 2.2698,
      "step": 52687
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7584381699562073,
      "learning_rate": 0.0005258556047713498,
      "loss": 2.1766,
      "step": 52688
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.863705575466156,
      "learning_rate": 0.0005258529123839656,
      "loss": 2.2562,
      "step": 52689
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7035551071166992,
      "learning_rate": 0.0005258502199545911,
      "loss": 2.2599,
      "step": 52690
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.8654882907867432,
      "learning_rate": 0.0005258475274832268,
      "loss": 2.2183,
      "step": 52691
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.8442713022232056,
      "learning_rate": 0.0005258448349698731,
      "loss": 2.2719,
      "step": 52692
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.8053242564201355,
      "learning_rate": 0.0005258421424145306,
      "loss": 2.4399,
      "step": 52693
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.799212634563446,
      "learning_rate": 0.0005258394498171997,
      "loss": 2.3562,
      "step": 52694
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.8666043877601624,
      "learning_rate": 0.0005258367571778812,
      "loss": 2.1169,
      "step": 52695
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7214714288711548,
      "learning_rate": 0.0005258340644965753,
      "loss": 2.1386,
      "step": 52696
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7919526100158691,
      "learning_rate": 0.0005258313717732824,
      "loss": 2.1777,
      "step": 52697
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7394213080406189,
      "learning_rate": 0.0005258286790080034,
      "loss": 2.3633,
      "step": 52698
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7177268266677856,
      "learning_rate": 0.0005258259862007384,
      "loss": 2.5158,
      "step": 52699
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.798257052898407,
      "learning_rate": 0.0005258232933514883,
      "loss": 2.2976,
      "step": 52700
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7406299710273743,
      "learning_rate": 0.0005258206004602532,
      "loss": 2.335,
      "step": 52701
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7485952973365784,
      "learning_rate": 0.0005258179075270339,
      "loss": 2.3219,
      "step": 52702
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.9687021970748901,
      "learning_rate": 0.0005258152145518308,
      "loss": 2.1858,
      "step": 52703
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7032213807106018,
      "learning_rate": 0.0005258125215346443,
      "loss": 2.1805,
      "step": 52704
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.6934462785720825,
      "learning_rate": 0.0005258098284754751,
      "loss": 2.1829,
      "step": 52705
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.8042300343513489,
      "learning_rate": 0.0005258071353743235,
      "loss": 2.1359,
      "step": 52706
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7356932163238525,
      "learning_rate": 0.0005258044422311901,
      "loss": 2.3577,
      "step": 52707
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7439282536506653,
      "learning_rate": 0.0005258017490460754,
      "loss": 2.0421,
      "step": 52708
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7754385471343994,
      "learning_rate": 0.0005257990558189799,
      "loss": 2.2942,
      "step": 52709
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.783452033996582,
      "learning_rate": 0.0005257963625499042,
      "loss": 2.244,
      "step": 52710
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7717006206512451,
      "learning_rate": 0.0005257936692388486,
      "loss": 2.3118,
      "step": 52711
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7278455495834351,
      "learning_rate": 0.0005257909758858137,
      "loss": 2.1995,
      "step": 52712
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7482625246047974,
      "learning_rate": 0.0005257882824908,
      "loss": 2.3709,
      "step": 52713
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.6547871232032776,
      "learning_rate": 0.0005257855890538079,
      "loss": 2.0789,
      "step": 52714
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7440409064292908,
      "learning_rate": 0.0005257828955748381,
      "loss": 1.9917,
      "step": 52715
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7019186615943909,
      "learning_rate": 0.000525780202053891,
      "loss": 2.1551,
      "step": 52716
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7271612286567688,
      "learning_rate": 0.0005257775084909671,
      "loss": 2.1964,
      "step": 52717
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7865446209907532,
      "learning_rate": 0.0005257748148860668,
      "loss": 2.3008,
      "step": 52718
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.6880808472633362,
      "learning_rate": 0.0005257721212391908,
      "loss": 2.2905,
      "step": 52719
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.8451194763183594,
      "learning_rate": 0.0005257694275503395,
      "loss": 2.3201,
      "step": 52720
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7021883726119995,
      "learning_rate": 0.0005257667338195133,
      "loss": 2.3069,
      "step": 52721
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.6967076063156128,
      "learning_rate": 0.0005257640400467129,
      "loss": 2.2137,
      "step": 52722
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7141630053520203,
      "learning_rate": 0.0005257613462319388,
      "loss": 2.2357,
      "step": 52723
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.6928333640098572,
      "learning_rate": 0.0005257586523751912,
      "loss": 2.2875,
      "step": 52724
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.6823341250419617,
      "learning_rate": 0.0005257559584764709,
      "loss": 2.0666,
      "step": 52725
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.6378927826881409,
      "learning_rate": 0.0005257532645357783,
      "loss": 2.2365,
      "step": 52726
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.6331731677055359,
      "learning_rate": 0.0005257505705531139,
      "loss": 2.1514,
      "step": 52727
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.661849319934845,
      "learning_rate": 0.0005257478765284782,
      "loss": 2.118,
      "step": 52728
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7054674625396729,
      "learning_rate": 0.0005257451824618717,
      "loss": 2.1608,
      "step": 52729
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7332669496536255,
      "learning_rate": 0.0005257424883532949,
      "loss": 1.9912,
      "step": 52730
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.6897810101509094,
      "learning_rate": 0.0005257397942027484,
      "loss": 2.1174,
      "step": 52731
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.714623749256134,
      "learning_rate": 0.0005257371000102325,
      "loss": 2.2136,
      "step": 52732
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.663186252117157,
      "learning_rate": 0.0005257344057757479,
      "loss": 2.2926,
      "step": 52733
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.731606125831604,
      "learning_rate": 0.0005257317114992949,
      "loss": 2.2591,
      "step": 52734
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7906566262245178,
      "learning_rate": 0.0005257290171808743,
      "loss": 2.1967,
      "step": 52735
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7168853282928467,
      "learning_rate": 0.0005257263228204863,
      "loss": 2.1819,
      "step": 52736
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.8099346160888672,
      "learning_rate": 0.0005257236284181314,
      "loss": 2.2292,
      "step": 52737
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.8167959451675415,
      "learning_rate": 0.0005257209339738103,
      "loss": 2.3816,
      "step": 52738
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.6707557439804077,
      "learning_rate": 0.0005257182394875236,
      "loss": 2.3474,
      "step": 52739
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7097983956336975,
      "learning_rate": 0.0005257155449592715,
      "loss": 2.1598,
      "step": 52740
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.708511233329773,
      "learning_rate": 0.0005257128503890544,
      "loss": 2.1679,
      "step": 52741
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.8312280774116516,
      "learning_rate": 0.0005257101557768733,
      "loss": 2.277,
      "step": 52742
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7646125555038452,
      "learning_rate": 0.0005257074611227284,
      "loss": 2.0159,
      "step": 52743
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.6356247663497925,
      "learning_rate": 0.0005257047664266201,
      "loss": 2.0249,
      "step": 52744
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.6667832732200623,
      "learning_rate": 0.000525702071688549,
      "loss": 2.0598,
      "step": 52745
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7109442353248596,
      "learning_rate": 0.0005256993769085157,
      "loss": 2.3241,
      "step": 52746
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7323475480079651,
      "learning_rate": 0.0005256966820865207,
      "loss": 2.0821,
      "step": 52747
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7640650272369385,
      "learning_rate": 0.0005256939872225644,
      "loss": 2.4209,
      "step": 52748
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7183870673179626,
      "learning_rate": 0.0005256912923166472,
      "loss": 2.3181,
      "step": 52749
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.6692293882369995,
      "learning_rate": 0.0005256885973687699,
      "loss": 2.1734,
      "step": 52750
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7419027090072632,
      "learning_rate": 0.0005256859023789327,
      "loss": 2.1675,
      "step": 52751
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.8332164883613586,
      "learning_rate": 0.0005256832073471363,
      "loss": 2.2154,
      "step": 52752
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7969350814819336,
      "learning_rate": 0.0005256805122733811,
      "loss": 2.3011,
      "step": 52753
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.6932209730148315,
      "learning_rate": 0.0005256778171576676,
      "loss": 2.5042,
      "step": 52754
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7857072949409485,
      "learning_rate": 0.0005256751219999963,
      "loss": 2.0498,
      "step": 52755
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.6928868293762207,
      "learning_rate": 0.0005256724268003679,
      "loss": 2.4401,
      "step": 52756
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7198064923286438,
      "learning_rate": 0.0005256697315587826,
      "loss": 2.5017,
      "step": 52757
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7202314734458923,
      "learning_rate": 0.0005256670362752411,
      "loss": 2.3663,
      "step": 52758
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7021799087524414,
      "learning_rate": 0.0005256643409497438,
      "loss": 2.2115,
      "step": 52759
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7293562889099121,
      "learning_rate": 0.0005256616455822913,
      "loss": 2.2516,
      "step": 52760
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7586187124252319,
      "learning_rate": 0.000525658950172884,
      "loss": 2.4523,
      "step": 52761
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7374430894851685,
      "learning_rate": 0.0005256562547215224,
      "loss": 2.434,
      "step": 52762
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7484518885612488,
      "learning_rate": 0.0005256535592282071,
      "loss": 2.2304,
      "step": 52763
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.6991782188415527,
      "learning_rate": 0.0005256508636929385,
      "loss": 2.3616,
      "step": 52764
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7107918858528137,
      "learning_rate": 0.0005256481681157172,
      "loss": 2.0343,
      "step": 52765
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.6676587462425232,
      "learning_rate": 0.0005256454724965436,
      "loss": 2.2738,
      "step": 52766
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.6796570420265198,
      "learning_rate": 0.0005256427768354183,
      "loss": 2.1298,
      "step": 52767
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7482725977897644,
      "learning_rate": 0.0005256400811323417,
      "loss": 2.2878,
      "step": 52768
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7262206673622131,
      "learning_rate": 0.0005256373853873143,
      "loss": 2.1589,
      "step": 52769
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.6843836307525635,
      "learning_rate": 0.0005256346896003367,
      "loss": 2.2324,
      "step": 52770
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7463020086288452,
      "learning_rate": 0.0005256319937714094,
      "loss": 2.2172,
      "step": 52771
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7049440145492554,
      "learning_rate": 0.0005256292979005328,
      "loss": 2.2036,
      "step": 52772
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7066020965576172,
      "learning_rate": 0.0005256266019877075,
      "loss": 2.4759,
      "step": 52773
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.695053219795227,
      "learning_rate": 0.0005256239060329339,
      "loss": 2.3381,
      "step": 52774
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7304219603538513,
      "learning_rate": 0.0005256212100362126,
      "loss": 2.2081,
      "step": 52775
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.6706132888793945,
      "learning_rate": 0.000525618513997544,
      "loss": 1.9727,
      "step": 52776
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7186582684516907,
      "learning_rate": 0.0005256158179169287,
      "loss": 2.3413,
      "step": 52777
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7216928005218506,
      "learning_rate": 0.0005256131217943671,
      "loss": 2.2576,
      "step": 52778
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7215520739555359,
      "learning_rate": 0.0005256104256298598,
      "loss": 2.3831,
      "step": 52779
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.6572268605232239,
      "learning_rate": 0.0005256077294234073,
      "loss": 2.2164,
      "step": 52780
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.8252370357513428,
      "learning_rate": 0.0005256050331750101,
      "loss": 2.0083,
      "step": 52781
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7403260469436646,
      "learning_rate": 0.0005256023368846686,
      "loss": 2.2417,
      "step": 52782
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7041219472885132,
      "learning_rate": 0.0005255996405523835,
      "loss": 2.3801,
      "step": 52783
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.6926353573799133,
      "learning_rate": 0.000525596944178155,
      "loss": 2.2019,
      "step": 52784
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.839038074016571,
      "learning_rate": 0.0005255942477619838,
      "loss": 2.2139,
      "step": 52785
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7765231728553772,
      "learning_rate": 0.0005255915513038703,
      "loss": 2.2427,
      "step": 52786
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7278420925140381,
      "learning_rate": 0.0005255888548038153,
      "loss": 2.2877,
      "step": 52787
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7438185811042786,
      "learning_rate": 0.0005255861582618189,
      "loss": 2.3856,
      "step": 52788
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.6608299016952515,
      "learning_rate": 0.0005255834616778817,
      "loss": 2.2511,
      "step": 52789
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7533602714538574,
      "learning_rate": 0.0005255807650520045,
      "loss": 2.3508,
      "step": 52790
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.6736526489257812,
      "learning_rate": 0.0005255780683841875,
      "loss": 2.6532,
      "step": 52791
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7019062042236328,
      "learning_rate": 0.0005255753716744313,
      "loss": 2.3191,
      "step": 52792
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.8113275766372681,
      "learning_rate": 0.0005255726749227362,
      "loss": 2.095,
      "step": 52793
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7444059252738953,
      "learning_rate": 0.000525569978129103,
      "loss": 2.1041,
      "step": 52794
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.6994480490684509,
      "learning_rate": 0.0005255672812935321,
      "loss": 2.5892,
      "step": 52795
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.6753302216529846,
      "learning_rate": 0.000525564584416024,
      "loss": 2.3796,
      "step": 52796
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.814437985420227,
      "learning_rate": 0.0005255618874965792,
      "loss": 2.1889,
      "step": 52797
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7571895122528076,
      "learning_rate": 0.0005255591905351981,
      "loss": 2.3193,
      "step": 52798
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.6933687925338745,
      "learning_rate": 0.0005255564935318813,
      "loss": 2.3161,
      "step": 52799
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7070519328117371,
      "learning_rate": 0.0005255537964866293,
      "loss": 2.1685,
      "step": 52800
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.6857032179832458,
      "learning_rate": 0.0005255510993994426,
      "loss": 2.3039,
      "step": 52801
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7449710369110107,
      "learning_rate": 0.0005255484022703218,
      "loss": 2.5489,
      "step": 52802
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7246718406677246,
      "learning_rate": 0.0005255457050992672,
      "loss": 2.0336,
      "step": 52803
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7751195430755615,
      "learning_rate": 0.0005255430078862794,
      "loss": 2.4965,
      "step": 52804
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.6822186708450317,
      "learning_rate": 0.000525540310631359,
      "loss": 2.1493,
      "step": 52805
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.8105347752571106,
      "learning_rate": 0.0005255376133345062,
      "loss": 1.9303,
      "step": 52806
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7175726294517517,
      "learning_rate": 0.0005255349159957218,
      "loss": 2.3019,
      "step": 52807
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.9165680408477783,
      "learning_rate": 0.0005255322186150062,
      "loss": 2.3011,
      "step": 52808
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7667199969291687,
      "learning_rate": 0.0005255295211923599,
      "loss": 2.1799,
      "step": 52809
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7285259962081909,
      "learning_rate": 0.0005255268237277834,
      "loss": 2.1633,
      "step": 52810
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7036505937576294,
      "learning_rate": 0.0005255241262212772,
      "loss": 2.437,
      "step": 52811
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.755406379699707,
      "learning_rate": 0.0005255214286728418,
      "loss": 2.2845,
      "step": 52812
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7653859853744507,
      "learning_rate": 0.0005255187310824778,
      "loss": 2.2432,
      "step": 52813
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7072595953941345,
      "learning_rate": 0.0005255160334501855,
      "loss": 2.3456,
      "step": 52814
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7027662396430969,
      "learning_rate": 0.0005255133357759655,
      "loss": 2.1084,
      "step": 52815
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7900223731994629,
      "learning_rate": 0.0005255106380598184,
      "loss": 2.4371,
      "step": 52816
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.6994370222091675,
      "learning_rate": 0.0005255079403017446,
      "loss": 2.3824,
      "step": 52817
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7381190657615662,
      "learning_rate": 0.0005255052425017446,
      "loss": 2.1924,
      "step": 52818
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.0550624132156372,
      "learning_rate": 0.0005255025446598189,
      "loss": 2.3589,
      "step": 52819
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7191147208213806,
      "learning_rate": 0.000525499846775968,
      "loss": 2.2023,
      "step": 52820
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.6743714213371277,
      "learning_rate": 0.0005254971488501925,
      "loss": 2.1166,
      "step": 52821
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.8473497629165649,
      "learning_rate": 0.0005254944508824928,
      "loss": 2.2542,
      "step": 52822
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.870930552482605,
      "learning_rate": 0.0005254917528728694,
      "loss": 2.4215,
      "step": 52823
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7180376052856445,
      "learning_rate": 0.0005254890548213228,
      "loss": 2.2546,
      "step": 52824
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7224407196044922,
      "learning_rate": 0.0005254863567278536,
      "loss": 2.5527,
      "step": 52825
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.8241426944732666,
      "learning_rate": 0.0005254836585924622,
      "loss": 2.3283,
      "step": 52826
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7507074475288391,
      "learning_rate": 0.0005254809604151491,
      "loss": 2.2064,
      "step": 52827
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.6491382718086243,
      "learning_rate": 0.0005254782621959148,
      "loss": 2.2524,
      "step": 52828
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7579414248466492,
      "learning_rate": 0.0005254755639347601,
      "loss": 2.3096,
      "step": 52829
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7354786396026611,
      "learning_rate": 0.000525472865631685,
      "loss": 2.2697,
      "step": 52830
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7383531928062439,
      "learning_rate": 0.0005254701672866902,
      "loss": 2.2552,
      "step": 52831
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7878382802009583,
      "learning_rate": 0.0005254674688997763,
      "loss": 2.2499,
      "step": 52832
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7184000611305237,
      "learning_rate": 0.0005254647704709438,
      "loss": 2.1531,
      "step": 52833
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.8990772366523743,
      "learning_rate": 0.000525462072000193,
      "loss": 2.2537,
      "step": 52834
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7506809830665588,
      "learning_rate": 0.0005254593734875247,
      "loss": 2.2384,
      "step": 52835
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7738161087036133,
      "learning_rate": 0.0005254566749329392,
      "loss": 2.2341,
      "step": 52836
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.9710226058959961,
      "learning_rate": 0.000525453976336437,
      "loss": 2.0206,
      "step": 52837
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7053461074829102,
      "learning_rate": 0.0005254512776980187,
      "loss": 2.1812,
      "step": 52838
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.8572985529899597,
      "learning_rate": 0.0005254485790176847,
      "loss": 2.1682,
      "step": 52839
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.9218411445617676,
      "learning_rate": 0.0005254458802954356,
      "loss": 2.3532,
      "step": 52840
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7954262495040894,
      "learning_rate": 0.0005254431815312719,
      "loss": 2.1315,
      "step": 52841
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7700623273849487,
      "learning_rate": 0.000525440482725194,
      "loss": 2.3995,
      "step": 52842
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7422444820404053,
      "learning_rate": 0.0005254377838772024,
      "loss": 2.0958,
      "step": 52843
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7535074353218079,
      "learning_rate": 0.0005254350849872978,
      "loss": 2.3808,
      "step": 52844
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7191278338432312,
      "learning_rate": 0.0005254323860554804,
      "loss": 2.4019,
      "step": 52845
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7645668387413025,
      "learning_rate": 0.000525429687081751,
      "loss": 2.4411,
      "step": 52846
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.742431104183197,
      "learning_rate": 0.0005254269880661099,
      "loss": 2.1677,
      "step": 52847
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7293392419815063,
      "learning_rate": 0.0005254242890085577,
      "loss": 2.216,
      "step": 52848
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.6283887624740601,
      "learning_rate": 0.0005254215899090949,
      "loss": 2.3378,
      "step": 52849
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.6694643497467041,
      "learning_rate": 0.0005254188907677219,
      "loss": 2.1115,
      "step": 52850
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7678865790367126,
      "learning_rate": 0.0005254161915844393,
      "loss": 2.3125,
      "step": 52851
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.69884192943573,
      "learning_rate": 0.0005254134923592475,
      "loss": 2.1499,
      "step": 52852
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7065428495407104,
      "learning_rate": 0.0005254107930921471,
      "loss": 2.308,
      "step": 52853
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7027802467346191,
      "learning_rate": 0.0005254080937831387,
      "loss": 2.1032,
      "step": 52854
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7438106536865234,
      "learning_rate": 0.0005254053944322225,
      "loss": 2.4028,
      "step": 52855
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7475607395172119,
      "learning_rate": 0.0005254026950393994,
      "loss": 2.3957,
      "step": 52856
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7660503387451172,
      "learning_rate": 0.0005253999956046695,
      "loss": 2.1012,
      "step": 52857
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7502204775810242,
      "learning_rate": 0.0005253972961280336,
      "loss": 2.0965,
      "step": 52858
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.6756812334060669,
      "learning_rate": 0.000525394596609492,
      "loss": 2.2074,
      "step": 52859
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7435925602912903,
      "learning_rate": 0.0005253918970490454,
      "loss": 2.5143,
      "step": 52860
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.754349946975708,
      "learning_rate": 0.0005253891974466941,
      "loss": 2.3296,
      "step": 52861
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7006878852844238,
      "learning_rate": 0.0005253864978024387,
      "loss": 2.3312,
      "step": 52862
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7552621364593506,
      "learning_rate": 0.0005253837981162797,
      "loss": 2.2899,
      "step": 52863
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.6947312951087952,
      "learning_rate": 0.0005253810983882176,
      "loss": 2.1766,
      "step": 52864
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.0890004634857178,
      "learning_rate": 0.0005253783986182529,
      "loss": 2.2706,
      "step": 52865
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.8573470115661621,
      "learning_rate": 0.0005253756988063861,
      "loss": 2.4596,
      "step": 52866
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7260292172431946,
      "learning_rate": 0.0005253729989526179,
      "loss": 2.22,
      "step": 52867
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7129157781600952,
      "learning_rate": 0.0005253702990569483,
      "loss": 2.3413,
      "step": 52868
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.8496255874633789,
      "learning_rate": 0.0005253675991193782,
      "loss": 2.389,
      "step": 52869
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7728124260902405,
      "learning_rate": 0.0005253648991399081,
      "loss": 2.3035,
      "step": 52870
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7384170889854431,
      "learning_rate": 0.0005253621991185384,
      "loss": 2.3837,
      "step": 52871
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.6793918013572693,
      "learning_rate": 0.0005253594990552695,
      "loss": 2.4963,
      "step": 52872
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7577146291732788,
      "learning_rate": 0.0005253567989501021,
      "loss": 2.0987,
      "step": 52873
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7508485913276672,
      "learning_rate": 0.0005253540988030365,
      "loss": 2.277,
      "step": 52874
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7431964874267578,
      "learning_rate": 0.0005253513986140735,
      "loss": 2.24,
      "step": 52875
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7587730288505554,
      "learning_rate": 0.0005253486983832134,
      "loss": 2.3298,
      "step": 52876
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7309771776199341,
      "learning_rate": 0.0005253459981104567,
      "loss": 2.2254,
      "step": 52877
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.6835566759109497,
      "learning_rate": 0.0005253432977958038,
      "loss": 2.2021,
      "step": 52878
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7590539455413818,
      "learning_rate": 0.0005253405974392554,
      "loss": 2.3068,
      "step": 52879
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.8063789010047913,
      "learning_rate": 0.0005253378970408119,
      "loss": 2.4111,
      "step": 52880
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.6647066473960876,
      "learning_rate": 0.0005253351966004739,
      "loss": 2.2557,
      "step": 52881
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.8419511914253235,
      "learning_rate": 0.0005253324961182419,
      "loss": 2.2613,
      "step": 52882
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7769677042961121,
      "learning_rate": 0.0005253297955941161,
      "loss": 2.1915,
      "step": 52883
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7314556837081909,
      "learning_rate": 0.0005253270950280975,
      "loss": 2.1084,
      "step": 52884
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7509180903434753,
      "learning_rate": 0.0005253243944201862,
      "loss": 2.207,
      "step": 52885
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.6842003464698792,
      "learning_rate": 0.0005253216937703828,
      "loss": 2.5354,
      "step": 52886
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7227951288223267,
      "learning_rate": 0.000525318993078688,
      "loss": 2.3502,
      "step": 52887
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.8136093020439148,
      "learning_rate": 0.000525316292345102,
      "loss": 2.3982,
      "step": 52888
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7900775074958801,
      "learning_rate": 0.0005253135915696254,
      "loss": 2.3466,
      "step": 52889
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7316352128982544,
      "learning_rate": 0.0005253108907522589,
      "loss": 1.973,
      "step": 52890
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7813391089439392,
      "learning_rate": 0.0005253081898930028,
      "loss": 2.3669,
      "step": 52891
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.6594309210777283,
      "learning_rate": 0.0005253054889918576,
      "loss": 2.3665,
      "step": 52892
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7755857706069946,
      "learning_rate": 0.000525302788048824,
      "loss": 2.6809,
      "step": 52893
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7757400870323181,
      "learning_rate": 0.0005253000870639022,
      "loss": 2.3453,
      "step": 52894
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.8573209047317505,
      "learning_rate": 0.000525297386037093,
      "loss": 2.2168,
      "step": 52895
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7626940011978149,
      "learning_rate": 0.0005252946849683967,
      "loss": 2.2117,
      "step": 52896
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7797595262527466,
      "learning_rate": 0.0005252919838578139,
      "loss": 2.1077,
      "step": 52897
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.0197774171829224,
      "learning_rate": 0.000525289282705345,
      "loss": 2.1111,
      "step": 52898
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7316362857818604,
      "learning_rate": 0.0005252865815109905,
      "loss": 2.2775,
      "step": 52899
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7810084223747253,
      "learning_rate": 0.0005252838802747512,
      "loss": 2.0896,
      "step": 52900
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.8848867416381836,
      "learning_rate": 0.0005252811789966271,
      "loss": 2.5041,
      "step": 52901
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7476794719696045,
      "learning_rate": 0.0005252784776766192,
      "loss": 2.2112,
      "step": 52902
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7774168848991394,
      "learning_rate": 0.0005252757763147277,
      "loss": 2.2343,
      "step": 52903
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.8465585112571716,
      "learning_rate": 0.0005252730749109532,
      "loss": 1.9727,
      "step": 52904
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7345570921897888,
      "learning_rate": 0.0005252703734652961,
      "loss": 2.1699,
      "step": 52905
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7701481580734253,
      "learning_rate": 0.0005252676719777571,
      "loss": 2.4442,
      "step": 52906
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7396281361579895,
      "learning_rate": 0.0005252649704483365,
      "loss": 2.3039,
      "step": 52907
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7025395631790161,
      "learning_rate": 0.000525262268877035,
      "loss": 2.3674,
      "step": 52908
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7085279226303101,
      "learning_rate": 0.0005252595672638529,
      "loss": 2.1043,
      "step": 52909
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.8255722522735596,
      "learning_rate": 0.0005252568656087908,
      "loss": 2.173,
      "step": 52910
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.6954447627067566,
      "learning_rate": 0.0005252541639118492,
      "loss": 2.2896,
      "step": 52911
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7848203182220459,
      "learning_rate": 0.0005252514621730286,
      "loss": 2.3197,
      "step": 52912
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.8240116834640503,
      "learning_rate": 0.0005252487603923295,
      "loss": 2.0779,
      "step": 52913
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.8078428506851196,
      "learning_rate": 0.0005252460585697524,
      "loss": 2.243,
      "step": 52914
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.9368119835853577,
      "learning_rate": 0.0005252433567052978,
      "loss": 2.3206,
      "step": 52915
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.0221519470214844,
      "learning_rate": 0.0005252406547989662,
      "loss": 2.3495,
      "step": 52916
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7456132769584656,
      "learning_rate": 0.0005252379528507582,
      "loss": 2.3032,
      "step": 52917
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7496376633644104,
      "learning_rate": 0.0005252352508606741,
      "loss": 2.0579,
      "step": 52918
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7370669841766357,
      "learning_rate": 0.0005252325488287146,
      "loss": 2.3616,
      "step": 52919
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.689182460308075,
      "learning_rate": 0.0005252298467548801,
      "loss": 1.8225,
      "step": 52920
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7460395097732544,
      "learning_rate": 0.000525227144639171,
      "loss": 2.2263,
      "step": 52921
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.6843467950820923,
      "learning_rate": 0.0005252244424815882,
      "loss": 2.0633,
      "step": 52922
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7373204231262207,
      "learning_rate": 0.0005252217402821317,
      "loss": 2.1205,
      "step": 52923
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.8288913369178772,
      "learning_rate": 0.0005252190380408022,
      "loss": 2.2976,
      "step": 52924
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.6926995515823364,
      "learning_rate": 0.0005252163357576003,
      "loss": 2.3062,
      "step": 52925
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7420101761817932,
      "learning_rate": 0.0005252136334325264,
      "loss": 2.3319,
      "step": 52926
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.719826340675354,
      "learning_rate": 0.0005252109310655812,
      "loss": 2.5127,
      "step": 52927
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7420312166213989,
      "learning_rate": 0.0005252082286567648,
      "loss": 2.0897,
      "step": 52928
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.6911506652832031,
      "learning_rate": 0.0005252055262060781,
      "loss": 2.4156,
      "step": 52929
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7260996699333191,
      "learning_rate": 0.0005252028237135214,
      "loss": 2.2521,
      "step": 52930
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.777310848236084,
      "learning_rate": 0.0005252001211790953,
      "loss": 2.2444,
      "step": 52931
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.8096635341644287,
      "learning_rate": 0.0005251974186028,
      "loss": 2.5468,
      "step": 52932
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7272881865501404,
      "learning_rate": 0.0005251947159846366,
      "loss": 2.4001,
      "step": 52933
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.6989075541496277,
      "learning_rate": 0.000525192013324605,
      "loss": 2.3887,
      "step": 52934
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7621259093284607,
      "learning_rate": 0.0005251893106227061,
      "loss": 2.3138,
      "step": 52935
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7463015913963318,
      "learning_rate": 0.0005251866078789402,
      "loss": 2.2033,
      "step": 52936
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.9117019772529602,
      "learning_rate": 0.0005251839050933078,
      "loss": 2.1948,
      "step": 52937
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7451017498970032,
      "learning_rate": 0.0005251812022658095,
      "loss": 2.2512,
      "step": 52938
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7326165437698364,
      "learning_rate": 0.0005251784993964458,
      "loss": 2.1304,
      "step": 52939
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.8537441492080688,
      "learning_rate": 0.0005251757964852172,
      "loss": 2.1488,
      "step": 52940
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.8492515683174133,
      "learning_rate": 0.0005251730935321241,
      "loss": 2.1588,
      "step": 52941
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.863232433795929,
      "learning_rate": 0.000525170390537167,
      "loss": 2.4725,
      "step": 52942
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.9696118831634521,
      "learning_rate": 0.0005251676875003466,
      "loss": 2.3399,
      "step": 52943
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7717235684394836,
      "learning_rate": 0.0005251649844216632,
      "loss": 2.2528,
      "step": 52944
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7347639799118042,
      "learning_rate": 0.0005251622813011174,
      "loss": 2.1149,
      "step": 52945
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7370567917823792,
      "learning_rate": 0.0005251595781387097,
      "loss": 2.2791,
      "step": 52946
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.794487714767456,
      "learning_rate": 0.0005251568749344406,
      "loss": 2.1237,
      "step": 52947
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7343308329582214,
      "learning_rate": 0.0005251541716883105,
      "loss": 1.9493,
      "step": 52948
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7947710156440735,
      "learning_rate": 0.0005251514684003201,
      "loss": 2.5254,
      "step": 52949
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.6962451934814453,
      "learning_rate": 0.0005251487650704697,
      "loss": 2.2981,
      "step": 52950
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.817243754863739,
      "learning_rate": 0.0005251460616987599,
      "loss": 2.1147,
      "step": 52951
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7627050280570984,
      "learning_rate": 0.0005251433582851914,
      "loss": 2.0948,
      "step": 52952
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7973641157150269,
      "learning_rate": 0.0005251406548297642,
      "loss": 2.2595,
      "step": 52953
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7405979633331299,
      "learning_rate": 0.0005251379513324792,
      "loss": 2.2761,
      "step": 52954
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7652212977409363,
      "learning_rate": 0.0005251352477933369,
      "loss": 2.3467,
      "step": 52955
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.8898009061813354,
      "learning_rate": 0.0005251325442123377,
      "loss": 2.0832,
      "step": 52956
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7458170652389526,
      "learning_rate": 0.0005251298405894821,
      "loss": 2.108,
      "step": 52957
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7562448978424072,
      "learning_rate": 0.0005251271369247705,
      "loss": 2.337,
      "step": 52958
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7280628085136414,
      "learning_rate": 0.0005251244332182036,
      "loss": 2.1605,
      "step": 52959
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7419095635414124,
      "learning_rate": 0.0005251217294697818,
      "loss": 2.2836,
      "step": 52960
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7524694800376892,
      "learning_rate": 0.0005251190256795056,
      "loss": 2.379,
      "step": 52961
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7174977660179138,
      "learning_rate": 0.0005251163218473756,
      "loss": 2.1122,
      "step": 52962
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7318335771560669,
      "learning_rate": 0.0005251136179733922,
      "loss": 2.4616,
      "step": 52963
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7839263081550598,
      "learning_rate": 0.0005251109140575558,
      "loss": 2.2707,
      "step": 52964
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7633399963378906,
      "learning_rate": 0.0005251082100998672,
      "loss": 2.3063,
      "step": 52965
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7113935351371765,
      "learning_rate": 0.0005251055061003267,
      "loss": 2.3289,
      "step": 52966
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7514048218727112,
      "learning_rate": 0.0005251028020589348,
      "loss": 2.2401,
      "step": 52967
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7523139119148254,
      "learning_rate": 0.000525100097975692,
      "loss": 2.2753,
      "step": 52968
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7855060696601868,
      "learning_rate": 0.0005250973938505989,
      "loss": 2.3837,
      "step": 52969
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.9664868116378784,
      "learning_rate": 0.0005250946896836559,
      "loss": 2.2093,
      "step": 52970
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7342337369918823,
      "learning_rate": 0.0005250919854748637,
      "loss": 2.4837,
      "step": 52971
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.6975504159927368,
      "learning_rate": 0.0005250892812242225,
      "loss": 2.3325,
      "step": 52972
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.6541864275932312,
      "learning_rate": 0.000525086576931733,
      "loss": 2.0817,
      "step": 52973
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.8733676671981812,
      "learning_rate": 0.0005250838725973957,
      "loss": 2.354,
      "step": 52974
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7382341623306274,
      "learning_rate": 0.0005250811682212109,
      "loss": 2.0226,
      "step": 52975
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7060670256614685,
      "learning_rate": 0.0005250784638031794,
      "loss": 1.9763,
      "step": 52976
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7107109427452087,
      "learning_rate": 0.0005250757593433016,
      "loss": 2.0853,
      "step": 52977
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.6697419881820679,
      "learning_rate": 0.0005250730548415778,
      "loss": 2.0213,
      "step": 52978
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7086216807365417,
      "learning_rate": 0.0005250703502980089,
      "loss": 2.1405,
      "step": 52979
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7342783808708191,
      "learning_rate": 0.000525067645712595,
      "loss": 2.2082,
      "step": 52980
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7195183038711548,
      "learning_rate": 0.0005250649410853368,
      "loss": 2.1864,
      "step": 52981
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7445026636123657,
      "learning_rate": 0.0005250622364162349,
      "loss": 2.1276,
      "step": 52982
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.6809418797492981,
      "learning_rate": 0.0005250595317052895,
      "loss": 2.2924,
      "step": 52983
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.6677490472793579,
      "learning_rate": 0.0005250568269525014,
      "loss": 2.0937,
      "step": 52984
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.674415647983551,
      "learning_rate": 0.0005250541221578709,
      "loss": 2.0646,
      "step": 52985
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.8910202980041504,
      "learning_rate": 0.0005250514173213987,
      "loss": 2.1854,
      "step": 52986
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7333081960678101,
      "learning_rate": 0.0005250487124430852,
      "loss": 2.3876,
      "step": 52987
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7049798965454102,
      "learning_rate": 0.0005250460075229308,
      "loss": 2.0216,
      "step": 52988
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.6521880626678467,
      "learning_rate": 0.0005250433025609362,
      "loss": 2.2618,
      "step": 52989
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7739489078521729,
      "learning_rate": 0.0005250405975571016,
      "loss": 2.0378,
      "step": 52990
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7580196857452393,
      "learning_rate": 0.0005250378925114279,
      "loss": 2.3874,
      "step": 52991
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7333364486694336,
      "learning_rate": 0.0005250351874239153,
      "loss": 2.2655,
      "step": 52992
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7288447618484497,
      "learning_rate": 0.0005250324822945645,
      "loss": 2.3874,
      "step": 52993
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7043346166610718,
      "learning_rate": 0.0005250297771233758,
      "loss": 2.0789,
      "step": 52994
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.756733238697052,
      "learning_rate": 0.0005250270719103499,
      "loss": 2.2869,
      "step": 52995
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7803277373313904,
      "learning_rate": 0.0005250243666554872,
      "loss": 2.5268,
      "step": 52996
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7374722361564636,
      "learning_rate": 0.0005250216613587883,
      "loss": 2.2609,
      "step": 52997
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.6642968058586121,
      "learning_rate": 0.0005250189560202536,
      "loss": 1.9936,
      "step": 52998
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7105655074119568,
      "learning_rate": 0.0005250162506398835,
      "loss": 2.2872,
      "step": 52999
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7227416634559631,
      "learning_rate": 0.0005250135452176788,
      "loss": 2.0362,
      "step": 53000
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7992425560951233,
      "learning_rate": 0.0005250108397536397,
      "loss": 2.1033,
      "step": 53001
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.701080322265625,
      "learning_rate": 0.0005250081342477669,
      "loss": 2.264,
      "step": 53002
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.6858928799629211,
      "learning_rate": 0.0005250054287000608,
      "loss": 2.1483,
      "step": 53003
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7411994338035583,
      "learning_rate": 0.000525002723110522,
      "loss": 2.2312,
      "step": 53004
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7305497527122498,
      "learning_rate": 0.0005250000174791509,
      "loss": 2.4171,
      "step": 53005
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7592578530311584,
      "learning_rate": 0.0005249973118059482,
      "loss": 2.2693,
      "step": 53006
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.745620846748352,
      "learning_rate": 0.0005249946060909141,
      "loss": 2.3893,
      "step": 53007
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7181009650230408,
      "learning_rate": 0.0005249919003340492,
      "loss": 2.4218,
      "step": 53008
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7082529067993164,
      "learning_rate": 0.0005249891945353542,
      "loss": 2.3525,
      "step": 53009
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7178075909614563,
      "learning_rate": 0.0005249864886948294,
      "loss": 2.221,
      "step": 53010
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7156828045845032,
      "learning_rate": 0.0005249837828124753,
      "loss": 2.4212,
      "step": 53011
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.729520320892334,
      "learning_rate": 0.0005249810768882926,
      "loss": 2.1338,
      "step": 53012
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.6960145831108093,
      "learning_rate": 0.0005249783709222816,
      "loss": 2.3791,
      "step": 53013
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.6980034708976746,
      "learning_rate": 0.0005249756649144428,
      "loss": 2.1996,
      "step": 53014
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.6994931697845459,
      "learning_rate": 0.0005249729588647769,
      "loss": 2.5596,
      "step": 53015
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7173706293106079,
      "learning_rate": 0.0005249702527732842,
      "loss": 2.3165,
      "step": 53016
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7651863694190979,
      "learning_rate": 0.0005249675466399654,
      "loss": 2.1745,
      "step": 53017
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.6775535345077515,
      "learning_rate": 0.0005249648404648208,
      "loss": 1.8869,
      "step": 53018
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.8075860738754272,
      "learning_rate": 0.0005249621342478509,
      "loss": 2.1831,
      "step": 53019
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.6721439361572266,
      "learning_rate": 0.0005249594279890564,
      "loss": 2.2352,
      "step": 53020
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.6954010128974915,
      "learning_rate": 0.0005249567216884378,
      "loss": 2.1528,
      "step": 53021
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7775312066078186,
      "learning_rate": 0.0005249540153459953,
      "loss": 1.9296,
      "step": 53022
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7131569385528564,
      "learning_rate": 0.0005249513089617297,
      "loss": 2.0416,
      "step": 53023
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7821464538574219,
      "learning_rate": 0.0005249486025356414,
      "loss": 2.3127,
      "step": 53024
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.8279306888580322,
      "learning_rate": 0.0005249458960677308,
      "loss": 2.1716,
      "step": 53025
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7230243682861328,
      "learning_rate": 0.0005249431895579987,
      "loss": 2.3956,
      "step": 53026
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.676738977432251,
      "learning_rate": 0.0005249404830064454,
      "loss": 2.1006,
      "step": 53027
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.6859630346298218,
      "learning_rate": 0.0005249377764130713,
      "loss": 2.2213,
      "step": 53028
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7599141597747803,
      "learning_rate": 0.000524935069777877,
      "loss": 2.2411,
      "step": 53029
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7502431273460388,
      "learning_rate": 0.0005249323631008632,
      "loss": 2.1997,
      "step": 53030
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7415715456008911,
      "learning_rate": 0.00052492965638203,
      "loss": 2.3,
      "step": 53031
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7553786039352417,
      "learning_rate": 0.0005249269496213784,
      "loss": 2.3005,
      "step": 53032
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.8619409799575806,
      "learning_rate": 0.0005249242428189084,
      "loss": 2.1839,
      "step": 53033
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7655346989631653,
      "learning_rate": 0.0005249215359746208,
      "loss": 2.3602,
      "step": 53034
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7131880521774292,
      "learning_rate": 0.000524918829088516,
      "loss": 2.3187,
      "step": 53035
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.6972103714942932,
      "learning_rate": 0.0005249161221605946,
      "loss": 2.2939,
      "step": 53036
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7396664023399353,
      "learning_rate": 0.0005249134151908571,
      "loss": 2.1284,
      "step": 53037
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.9817243814468384,
      "learning_rate": 0.0005249107081793039,
      "loss": 2.1442,
      "step": 53038
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7721903324127197,
      "learning_rate": 0.0005249080011259355,
      "loss": 2.1297,
      "step": 53039
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.934786319732666,
      "learning_rate": 0.0005249052940307525,
      "loss": 2.3413,
      "step": 53040
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7276613116264343,
      "learning_rate": 0.0005249025868937554,
      "loss": 2.3783,
      "step": 53041
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7977721691131592,
      "learning_rate": 0.0005248998797149447,
      "loss": 2.4114,
      "step": 53042
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.8586551547050476,
      "learning_rate": 0.0005248971724943208,
      "loss": 2.41,
      "step": 53043
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.8885432481765747,
      "learning_rate": 0.0005248944652318843,
      "loss": 2.1581,
      "step": 53044
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7693125009536743,
      "learning_rate": 0.0005248917579276357,
      "loss": 2.3953,
      "step": 53045
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7730953693389893,
      "learning_rate": 0.0005248890505815753,
      "loss": 2.1157,
      "step": 53046
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7635737061500549,
      "learning_rate": 0.0005248863431937039,
      "loss": 2.1689,
      "step": 53047
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.8051022887229919,
      "learning_rate": 0.0005248836357640219,
      "loss": 2.2992,
      "step": 53048
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7454519867897034,
      "learning_rate": 0.0005248809282925298,
      "loss": 2.1484,
      "step": 53049
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7475942969322205,
      "learning_rate": 0.000524878220779228,
      "loss": 2.1105,
      "step": 53050
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.6808380484580994,
      "learning_rate": 0.0005248755132241172,
      "loss": 2.0332,
      "step": 53051
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7503635287284851,
      "learning_rate": 0.0005248728056271978,
      "loss": 2.098,
      "step": 53052
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7259237170219421,
      "learning_rate": 0.0005248700979884703,
      "loss": 2.0892,
      "step": 53053
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7003740668296814,
      "learning_rate": 0.0005248673903079351,
      "loss": 2.1699,
      "step": 53054
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.8137595057487488,
      "learning_rate": 0.0005248646825855929,
      "loss": 2.2196,
      "step": 53055
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7859123349189758,
      "learning_rate": 0.0005248619748214441,
      "loss": 2.2152,
      "step": 53056
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7053965330123901,
      "learning_rate": 0.0005248592670154892,
      "loss": 2.0935,
      "step": 53057
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.780561089515686,
      "learning_rate": 0.0005248565591677287,
      "loss": 2.2901,
      "step": 53058
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7392942309379578,
      "learning_rate": 0.0005248538512781631,
      "loss": 2.4623,
      "step": 53059
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7988934516906738,
      "learning_rate": 0.000524851143346793,
      "loss": 2.4449,
      "step": 53060
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7725606560707092,
      "learning_rate": 0.0005248484353736187,
      "loss": 2.3958,
      "step": 53061
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.6905155777931213,
      "learning_rate": 0.000524845727358641,
      "loss": 2.3109,
      "step": 53062
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7374210357666016,
      "learning_rate": 0.0005248430193018601,
      "loss": 2.2789,
      "step": 53063
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.6860591769218445,
      "learning_rate": 0.0005248403112032767,
      "loss": 2.2985,
      "step": 53064
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7058443427085876,
      "learning_rate": 0.0005248376030628913,
      "loss": 2.0806,
      "step": 53065
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.673442542552948,
      "learning_rate": 0.0005248348948807042,
      "loss": 2.2301,
      "step": 53066
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7860431671142578,
      "learning_rate": 0.0005248321866567162,
      "loss": 2.1387,
      "step": 53067
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.8449832201004028,
      "learning_rate": 0.0005248294783909276,
      "loss": 1.9951,
      "step": 53068
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7648740410804749,
      "learning_rate": 0.0005248267700833388,
      "loss": 2.2538,
      "step": 53069
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7814543843269348,
      "learning_rate": 0.0005248240617339506,
      "loss": 2.2334,
      "step": 53070
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.8372539281845093,
      "learning_rate": 0.0005248213533427634,
      "loss": 2.2595,
      "step": 53071
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.8284928202629089,
      "learning_rate": 0.0005248186449097776,
      "loss": 2.3448,
      "step": 53072
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7204927206039429,
      "learning_rate": 0.0005248159364349939,
      "loss": 2.3269,
      "step": 53073
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.8418305516242981,
      "learning_rate": 0.0005248132279184125,
      "loss": 2.3088,
      "step": 53074
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.8321384191513062,
      "learning_rate": 0.0005248105193600343,
      "loss": 2.2118,
      "step": 53075
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7570304870605469,
      "learning_rate": 0.0005248078107598593,
      "loss": 2.2482,
      "step": 53076
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.8328523635864258,
      "learning_rate": 0.0005248051021178884,
      "loss": 2.1841,
      "step": 53077
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7482519745826721,
      "learning_rate": 0.0005248023934341221,
      "loss": 2.0705,
      "step": 53078
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7722960710525513,
      "learning_rate": 0.0005247996847085606,
      "loss": 2.3274,
      "step": 53079
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7234923243522644,
      "learning_rate": 0.0005247969759412047,
      "loss": 2.2102,
      "step": 53080
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7741420865058899,
      "learning_rate": 0.0005247942671320548,
      "loss": 2.5271,
      "step": 53081
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7945736050605774,
      "learning_rate": 0.0005247915582811114,
      "loss": 2.1652,
      "step": 53082
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7289879322052002,
      "learning_rate": 0.000524788849388375,
      "loss": 2.2497,
      "step": 53083
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.6932478547096252,
      "learning_rate": 0.000524786140453846,
      "loss": 2.2481,
      "step": 53084
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.8137841820716858,
      "learning_rate": 0.0005247834314775252,
      "loss": 2.2503,
      "step": 53085
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7829216718673706,
      "learning_rate": 0.0005247807224594128,
      "loss": 2.1647,
      "step": 53086
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.6848461627960205,
      "learning_rate": 0.0005247780133995094,
      "loss": 2.3779,
      "step": 53087
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7125048637390137,
      "learning_rate": 0.0005247753042978157,
      "loss": 2.1878,
      "step": 53088
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7268174290657043,
      "learning_rate": 0.0005247725951543319,
      "loss": 2.388,
      "step": 53089
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.6963949203491211,
      "learning_rate": 0.0005247698859690586,
      "loss": 2.3937,
      "step": 53090
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7525096535682678,
      "learning_rate": 0.0005247671767419963,
      "loss": 2.0147,
      "step": 53091
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7008267641067505,
      "learning_rate": 0.0005247644674731456,
      "loss": 2.223,
      "step": 53092
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.6668316125869751,
      "learning_rate": 0.0005247617581625071,
      "loss": 2.2912,
      "step": 53093
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.8406965732574463,
      "learning_rate": 0.0005247590488100809,
      "loss": 2.0883,
      "step": 53094
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7129928469657898,
      "learning_rate": 0.0005247563394158679,
      "loss": 2.0465,
      "step": 53095
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7015597224235535,
      "learning_rate": 0.0005247536299798684,
      "loss": 2.2778,
      "step": 53096
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.8305061459541321,
      "learning_rate": 0.000524750920502083,
      "loss": 2.1977,
      "step": 53097
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.6923161745071411,
      "learning_rate": 0.0005247482109825121,
      "loss": 2.3543,
      "step": 53098
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7980679273605347,
      "learning_rate": 0.0005247455014211563,
      "loss": 2.1413,
      "step": 53099
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.6947888135910034,
      "learning_rate": 0.0005247427918180161,
      "loss": 2.3552,
      "step": 53100
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7074053883552551,
      "learning_rate": 0.0005247400821730919,
      "loss": 2.2221,
      "step": 53101
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7595749497413635,
      "learning_rate": 0.0005247373724863843,
      "loss": 2.0543,
      "step": 53102
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7298111319541931,
      "learning_rate": 0.0005247346627578938,
      "loss": 2.2338,
      "step": 53103
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7172733545303345,
      "learning_rate": 0.0005247319529876209,
      "loss": 2.332,
      "step": 53104
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.8239200711250305,
      "learning_rate": 0.000524729243175566,
      "loss": 2.2331,
      "step": 53105
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.6959168314933777,
      "learning_rate": 0.0005247265333217298,
      "loss": 2.2136,
      "step": 53106
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.8453618288040161,
      "learning_rate": 0.0005247238234261126,
      "loss": 2.2531,
      "step": 53107
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.772366464138031,
      "learning_rate": 0.0005247211134887151,
      "loss": 2.4666,
      "step": 53108
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7815894484519958,
      "learning_rate": 0.0005247184035095376,
      "loss": 2.3684,
      "step": 53109
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7601664662361145,
      "learning_rate": 0.0005247156934885808,
      "loss": 2.1097,
      "step": 53110
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.8424062728881836,
      "learning_rate": 0.0005247129834258451,
      "loss": 2.2182,
      "step": 53111
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7478581070899963,
      "learning_rate": 0.0005247102733213309,
      "loss": 2.2138,
      "step": 53112
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7776625752449036,
      "learning_rate": 0.000524707563175039,
      "loss": 2.3252,
      "step": 53113
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.678504467010498,
      "learning_rate": 0.0005247048529869697,
      "loss": 2.373,
      "step": 53114
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.6464405655860901,
      "learning_rate": 0.0005247021427571233,
      "loss": 2.1742,
      "step": 53115
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7246859669685364,
      "learning_rate": 0.0005246994324855008,
      "loss": 2.3358,
      "step": 53116
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7577242255210876,
      "learning_rate": 0.0005246967221721022,
      "loss": 2.1453,
      "step": 53117
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7321481108665466,
      "learning_rate": 0.0005246940118169283,
      "loss": 2.4815,
      "step": 53118
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.6867318749427795,
      "learning_rate": 0.0005246913014199797,
      "loss": 2.1114,
      "step": 53119
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7224921584129333,
      "learning_rate": 0.0005246885909812566,
      "loss": 2.292,
      "step": 53120
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7740961313247681,
      "learning_rate": 0.0005246858805007596,
      "loss": 2.248,
      "step": 53121
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.672177791595459,
      "learning_rate": 0.0005246831699784895,
      "loss": 2.3556,
      "step": 53122
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.699378252029419,
      "learning_rate": 0.0005246804594144463,
      "loss": 2.247,
      "step": 53123
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.702587902545929,
      "learning_rate": 0.0005246777488086309,
      "loss": 2.3102,
      "step": 53124
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7409902811050415,
      "learning_rate": 0.0005246750381610436,
      "loss": 2.4036,
      "step": 53125
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.6450337767601013,
      "learning_rate": 0.000524672327471685,
      "loss": 2.1267,
      "step": 53126
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.6800070405006409,
      "learning_rate": 0.0005246696167405555,
      "loss": 2.1794,
      "step": 53127
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.6919887065887451,
      "learning_rate": 0.0005246669059676558,
      "loss": 2.2952,
      "step": 53128
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.6884225010871887,
      "learning_rate": 0.0005246641951529862,
      "loss": 2.3992,
      "step": 53129
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.6837263107299805,
      "learning_rate": 0.0005246614842965474,
      "loss": 2.2218,
      "step": 53130
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.6605894565582275,
      "learning_rate": 0.0005246587733983396,
      "loss": 2.1618,
      "step": 53131
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7006316184997559,
      "learning_rate": 0.0005246560624583636,
      "loss": 2.2615,
      "step": 53132
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7182183861732483,
      "learning_rate": 0.0005246533514766199,
      "loss": 2.1254,
      "step": 53133
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7725619077682495,
      "learning_rate": 0.0005246506404531087,
      "loss": 2.4344,
      "step": 53134
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.6942382454872131,
      "learning_rate": 0.0005246479293878308,
      "loss": 2.4081,
      "step": 53135
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.6479051113128662,
      "learning_rate": 0.0005246452182807866,
      "loss": 2.1371,
      "step": 53136
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7659614086151123,
      "learning_rate": 0.0005246425071319767,
      "loss": 2.4205,
      "step": 53137
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.706375241279602,
      "learning_rate": 0.0005246397959414014,
      "loss": 2.2119,
      "step": 53138
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7145728468894958,
      "learning_rate": 0.0005246370847090614,
      "loss": 2.1725,
      "step": 53139
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7659305334091187,
      "learning_rate": 0.0005246343734349571,
      "loss": 2.1486,
      "step": 53140
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.6964420676231384,
      "learning_rate": 0.000524631662119089,
      "loss": 2.3404,
      "step": 53141
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.720318078994751,
      "learning_rate": 0.0005246289507614576,
      "loss": 2.3162,
      "step": 53142
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.791665256023407,
      "learning_rate": 0.0005246262393620635,
      "loss": 2.2131,
      "step": 53143
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7975289225578308,
      "learning_rate": 0.0005246235279209072,
      "loss": 2.2875,
      "step": 53144
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7138415575027466,
      "learning_rate": 0.000524620816437989,
      "loss": 2.3341,
      "step": 53145
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.724147617816925,
      "learning_rate": 0.0005246181049133096,
      "loss": 2.4543,
      "step": 53146
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.9789077639579773,
      "learning_rate": 0.0005246153933468695,
      "loss": 2.1357,
      "step": 53147
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.6680880784988403,
      "learning_rate": 0.0005246126817386691,
      "loss": 2.1651,
      "step": 53148
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7165113687515259,
      "learning_rate": 0.000524609970088709,
      "loss": 2.3699,
      "step": 53149
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7072425484657288,
      "learning_rate": 0.0005246072583969897,
      "loss": 2.2111,
      "step": 53150
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7794761061668396,
      "learning_rate": 0.0005246045466635117,
      "loss": 2.1298,
      "step": 53151
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7089092135429382,
      "learning_rate": 0.0005246018348882753,
      "loss": 2.2791,
      "step": 53152
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.818061888217926,
      "learning_rate": 0.0005245991230712814,
      "loss": 2.4368,
      "step": 53153
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7440520524978638,
      "learning_rate": 0.0005245964112125302,
      "loss": 2.3957,
      "step": 53154
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.9281657338142395,
      "learning_rate": 0.0005245936993120222,
      "loss": 2.3303,
      "step": 53155
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.6570509076118469,
      "learning_rate": 0.000524590987369758,
      "loss": 2.1997,
      "step": 53156
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.8489066362380981,
      "learning_rate": 0.0005245882753857382,
      "loss": 2.2695,
      "step": 53157
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.8185364603996277,
      "learning_rate": 0.0005245855633599632,
      "loss": 2.2136,
      "step": 53158
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.8255304098129272,
      "learning_rate": 0.0005245828512924335,
      "loss": 2.3995,
      "step": 53159
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7041655778884888,
      "learning_rate": 0.0005245801391831495,
      "loss": 2.0191,
      "step": 53160
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.1767313480377197,
      "learning_rate": 0.000524577427032112,
      "loss": 2.157,
      "step": 53161
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.8314656019210815,
      "learning_rate": 0.0005245747148393211,
      "loss": 2.3395,
      "step": 53162
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.6712369918823242,
      "learning_rate": 0.0005245720026047776,
      "loss": 2.1536,
      "step": 53163
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.8168014287948608,
      "learning_rate": 0.000524569290328482,
      "loss": 2.388,
      "step": 53164
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7486995458602905,
      "learning_rate": 0.0005245665780104346,
      "loss": 2.3589,
      "step": 53165
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.6743320226669312,
      "learning_rate": 0.0005245638656506362,
      "loss": 2.1245,
      "step": 53166
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7727891802787781,
      "learning_rate": 0.0005245611532490869,
      "loss": 2.2362,
      "step": 53167
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7604758143424988,
      "learning_rate": 0.0005245584408057875,
      "loss": 2.1976,
      "step": 53168
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.9137499928474426,
      "learning_rate": 0.0005245557283207386,
      "loss": 2.2734,
      "step": 53169
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7246878743171692,
      "learning_rate": 0.0005245530157939405,
      "loss": 2.1107,
      "step": 53170
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.6885110139846802,
      "learning_rate": 0.0005245503032253938,
      "loss": 2.3462,
      "step": 53171
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.8036070466041565,
      "learning_rate": 0.0005245475906150989,
      "loss": 2.0292,
      "step": 53172
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.8161662817001343,
      "learning_rate": 0.0005245448779630564,
      "loss": 2.2816,
      "step": 53173
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7387425899505615,
      "learning_rate": 0.0005245421652692667,
      "loss": 2.1127,
      "step": 53174
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.9053654670715332,
      "learning_rate": 0.0005245394525337304,
      "loss": 2.0743,
      "step": 53175
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.8581644296646118,
      "learning_rate": 0.000524536739756448,
      "loss": 2.5251,
      "step": 53176
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.6650668978691101,
      "learning_rate": 0.00052453402693742,
      "loss": 2.0344,
      "step": 53177
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7301226258277893,
      "learning_rate": 0.0005245313140766468,
      "loss": 2.1292,
      "step": 53178
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7335883378982544,
      "learning_rate": 0.000524528601174129,
      "loss": 2.2393,
      "step": 53179
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.6758471131324768,
      "learning_rate": 0.0005245258882298671,
      "loss": 2.1806,
      "step": 53180
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.8091484308242798,
      "learning_rate": 0.0005245231752438617,
      "loss": 2.4865,
      "step": 53181
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.6979675889015198,
      "learning_rate": 0.0005245204622161131,
      "loss": 2.3185,
      "step": 53182
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.781862199306488,
      "learning_rate": 0.000524517749146622,
      "loss": 2.27,
      "step": 53183
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7137585282325745,
      "learning_rate": 0.0005245150360353888,
      "loss": 2.2402,
      "step": 53184
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7635043859481812,
      "learning_rate": 0.0005245123228824139,
      "loss": 2.3352,
      "step": 53185
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7628070712089539,
      "learning_rate": 0.000524509609687698,
      "loss": 2.1292,
      "step": 53186
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7947495579719543,
      "learning_rate": 0.0005245068964512416,
      "loss": 2.3041,
      "step": 53187
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7525191307067871,
      "learning_rate": 0.000524504183173045,
      "loss": 2.1057,
      "step": 53188
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7366121411323547,
      "learning_rate": 0.0005245014698531088,
      "loss": 2.2154,
      "step": 53189
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7249626517295837,
      "learning_rate": 0.0005244987564914337,
      "loss": 2.0996,
      "step": 53190
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7248734831809998,
      "learning_rate": 0.00052449604308802,
      "loss": 2.2609,
      "step": 53191
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7341025471687317,
      "learning_rate": 0.0005244933296428681,
      "loss": 2.2585,
      "step": 53192
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7549604773521423,
      "learning_rate": 0.0005244906161559788,
      "loss": 2.4146,
      "step": 53193
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7134951949119568,
      "learning_rate": 0.0005244879026273525,
      "loss": 2.2053,
      "step": 53194
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7727160453796387,
      "learning_rate": 0.0005244851890569896,
      "loss": 2.2412,
      "step": 53195
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.6906319856643677,
      "learning_rate": 0.0005244824754448906,
      "loss": 2.2873,
      "step": 53196
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7263959646224976,
      "learning_rate": 0.0005244797617910561,
      "loss": 2.2499,
      "step": 53197
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7838168144226074,
      "learning_rate": 0.0005244770480954867,
      "loss": 2.5186,
      "step": 53198
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.8190536499023438,
      "learning_rate": 0.0005244743343581826,
      "loss": 2.4177,
      "step": 53199
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.832568347454071,
      "learning_rate": 0.0005244716205791446,
      "loss": 2.3345,
      "step": 53200
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7215354442596436,
      "learning_rate": 0.0005244689067583731,
      "loss": 1.978,
      "step": 53201
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7282932996749878,
      "learning_rate": 0.0005244661928958686,
      "loss": 2.1858,
      "step": 53202
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7287269830703735,
      "learning_rate": 0.0005244634789916315,
      "loss": 2.2682,
      "step": 53203
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7154975533485413,
      "learning_rate": 0.0005244607650456625,
      "loss": 2.4974,
      "step": 53204
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7898954749107361,
      "learning_rate": 0.0005244580510579619,
      "loss": 2.0208,
      "step": 53205
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7831714153289795,
      "learning_rate": 0.0005244553370285303,
      "loss": 2.3641,
      "step": 53206
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7424333095550537,
      "learning_rate": 0.0005244526229573683,
      "loss": 2.3319,
      "step": 53207
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.6980673670768738,
      "learning_rate": 0.0005244499088444763,
      "loss": 2.182,
      "step": 53208
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7024136185646057,
      "learning_rate": 0.0005244471946898549,
      "loss": 2.151,
      "step": 53209
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.8458977937698364,
      "learning_rate": 0.0005244444804935044,
      "loss": 2.2289,
      "step": 53210
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7531142234802246,
      "learning_rate": 0.0005244417662554254,
      "loss": 2.3505,
      "step": 53211
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7028509974479675,
      "learning_rate": 0.0005244390519756186,
      "loss": 2.1424,
      "step": 53212
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7826367616653442,
      "learning_rate": 0.0005244363376540842,
      "loss": 2.3623,
      "step": 53213
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.9546810388565063,
      "learning_rate": 0.000524433623290823,
      "loss": 2.3002,
      "step": 53214
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7593774795532227,
      "learning_rate": 0.0005244309088858352,
      "loss": 2.395,
      "step": 53215
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7384085655212402,
      "learning_rate": 0.0005244281944391216,
      "loss": 2.1834,
      "step": 53216
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.6857548356056213,
      "learning_rate": 0.0005244254799506824,
      "loss": 2.3008,
      "step": 53217
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7877776622772217,
      "learning_rate": 0.0005244227654205184,
      "loss": 2.2409,
      "step": 53218
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.6851486563682556,
      "learning_rate": 0.0005244200508486299,
      "loss": 2.2903,
      "step": 53219
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.6658962368965149,
      "learning_rate": 0.0005244173362350176,
      "loss": 2.1325,
      "step": 53220
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.8484315872192383,
      "learning_rate": 0.0005244146215796817,
      "loss": 2.405,
      "step": 53221
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7664951086044312,
      "learning_rate": 0.0005244119068826229,
      "loss": 2.4276,
      "step": 53222
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7667683362960815,
      "learning_rate": 0.0005244091921438418,
      "loss": 2.2405,
      "step": 53223
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7471618056297302,
      "learning_rate": 0.0005244064773633388,
      "loss": 2.1536,
      "step": 53224
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7548688650131226,
      "learning_rate": 0.0005244037625411145,
      "loss": 2.4891,
      "step": 53225
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.9090720415115356,
      "learning_rate": 0.000524401047677169,
      "loss": 2.5424,
      "step": 53226
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7707997560501099,
      "learning_rate": 0.0005243983327715033,
      "loss": 2.3964,
      "step": 53227
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.8399964570999146,
      "learning_rate": 0.0005243956178241177,
      "loss": 2.3507,
      "step": 53228
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7407701015472412,
      "learning_rate": 0.0005243929028350127,
      "loss": 2.2717,
      "step": 53229
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.6973389387130737,
      "learning_rate": 0.0005243901878041888,
      "loss": 2.3126,
      "step": 53230
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.6940128207206726,
      "learning_rate": 0.0005243874727316465,
      "loss": 2.13,
      "step": 53231
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.8493485450744629,
      "learning_rate": 0.0005243847576173865,
      "loss": 2.3906,
      "step": 53232
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.792640209197998,
      "learning_rate": 0.0005243820424614091,
      "loss": 2.3917,
      "step": 53233
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.9622728228569031,
      "learning_rate": 0.0005243793272637147,
      "loss": 2.4226,
      "step": 53234
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7217191457748413,
      "learning_rate": 0.0005243766120243041,
      "loss": 2.5045,
      "step": 53235
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7299078702926636,
      "learning_rate": 0.0005243738967431774,
      "loss": 2.3053,
      "step": 53236
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7754483819007874,
      "learning_rate": 0.0005243711814203356,
      "loss": 2.2016,
      "step": 53237
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.6967896223068237,
      "learning_rate": 0.0005243684660557788,
      "loss": 2.1779,
      "step": 53238
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.851586639881134,
      "learning_rate": 0.0005243657506495078,
      "loss": 2.4272,
      "step": 53239
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.8684244155883789,
      "learning_rate": 0.0005243630352015229,
      "loss": 2.326,
      "step": 53240
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7696250677108765,
      "learning_rate": 0.0005243603197118248,
      "loss": 2.3666,
      "step": 53241
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.6881199479103088,
      "learning_rate": 0.0005243576041804137,
      "loss": 2.119,
      "step": 53242
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7659745216369629,
      "learning_rate": 0.0005243548886072904,
      "loss": 2.2403,
      "step": 53243
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.8139200806617737,
      "learning_rate": 0.0005243521729924553,
      "loss": 2.1283,
      "step": 53244
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.6700193881988525,
      "learning_rate": 0.0005243494573359089,
      "loss": 2.4284,
      "step": 53245
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7496914863586426,
      "learning_rate": 0.0005243467416376517,
      "loss": 2.2405,
      "step": 53246
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.6594406962394714,
      "learning_rate": 0.000524344025897684,
      "loss": 2.4046,
      "step": 53247
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7535737156867981,
      "learning_rate": 0.0005243413101160068,
      "loss": 2.3669,
      "step": 53248
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7217803001403809,
      "learning_rate": 0.0005243385942926202,
      "loss": 2.3925,
      "step": 53249
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.6930627822875977,
      "learning_rate": 0.0005243358784275247,
      "loss": 2.2273,
      "step": 53250
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7150321006774902,
      "learning_rate": 0.0005243331625207211,
      "loss": 2.2348,
      "step": 53251
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7763755321502686,
      "learning_rate": 0.0005243304465722097,
      "loss": 2.4532,
      "step": 53252
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7174080610275269,
      "learning_rate": 0.000524327730581991,
      "loss": 2.1571,
      "step": 53253
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7545889616012573,
      "learning_rate": 0.0005243250145500655,
      "loss": 2.1837,
      "step": 53254
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.8779486417770386,
      "learning_rate": 0.0005243222984764339,
      "loss": 2.3338,
      "step": 53255
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.734436571598053,
      "learning_rate": 0.0005243195823610964,
      "loss": 2.3618,
      "step": 53256
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7184382081031799,
      "learning_rate": 0.0005243168662040537,
      "loss": 2.4329,
      "step": 53257
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7677358984947205,
      "learning_rate": 0.0005243141500053063,
      "loss": 2.1923,
      "step": 53258
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7123333811759949,
      "learning_rate": 0.0005243114337648545,
      "loss": 2.2094,
      "step": 53259
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7320652604103088,
      "learning_rate": 0.0005243087174826991,
      "loss": 2.3633,
      "step": 53260
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7083515524864197,
      "learning_rate": 0.0005243060011588406,
      "loss": 2.4172,
      "step": 53261
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.9133310914039612,
      "learning_rate": 0.0005243032847932792,
      "loss": 2.2275,
      "step": 53262
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7388064861297607,
      "learning_rate": 0.0005243005683860157,
      "loss": 2.4581,
      "step": 53263
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.8301113247871399,
      "learning_rate": 0.0005242978519370503,
      "loss": 2.1238,
      "step": 53264
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.6470894813537598,
      "learning_rate": 0.0005242951354463839,
      "loss": 2.3935,
      "step": 53265
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.6279408931732178,
      "learning_rate": 0.0005242924189140166,
      "loss": 2.0816,
      "step": 53266
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7318940162658691,
      "learning_rate": 0.0005242897023399493,
      "loss": 2.1874,
      "step": 53267
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.6433468461036682,
      "learning_rate": 0.0005242869857241822,
      "loss": 2.2738,
      "step": 53268
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.6893389821052551,
      "learning_rate": 0.0005242842690667159,
      "loss": 2.183,
      "step": 53269
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.6908114552497864,
      "learning_rate": 0.000524281552367551,
      "loss": 2.3575,
      "step": 53270
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7546626329421997,
      "learning_rate": 0.000524278835626688,
      "loss": 2.1094,
      "step": 53271
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7320757508277893,
      "learning_rate": 0.0005242761188441272,
      "loss": 2.2402,
      "step": 53272
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7343905568122864,
      "learning_rate": 0.0005242734020198692,
      "loss": 2.4038,
      "step": 53273
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7543586492538452,
      "learning_rate": 0.0005242706851539147,
      "loss": 2.1604,
      "step": 53274
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7689289450645447,
      "learning_rate": 0.0005242679682462639,
      "loss": 2.3213,
      "step": 53275
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.6997729539871216,
      "learning_rate": 0.0005242652512969175,
      "loss": 2.2736,
      "step": 53276
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.6872576475143433,
      "learning_rate": 0.000524262534305876,
      "loss": 1.9552,
      "step": 53277
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.6678394675254822,
      "learning_rate": 0.0005242598172731397,
      "loss": 2.2103,
      "step": 53278
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7089847326278687,
      "learning_rate": 0.0005242571001987094,
      "loss": 2.1336,
      "step": 53279
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.6668392419815063,
      "learning_rate": 0.0005242543830825855,
      "loss": 2.1993,
      "step": 53280
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7188286781311035,
      "learning_rate": 0.0005242516659247686,
      "loss": 2.008,
      "step": 53281
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7124472856521606,
      "learning_rate": 0.0005242489487252588,
      "loss": 2.4102,
      "step": 53282
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7028362154960632,
      "learning_rate": 0.000524246231484057,
      "loss": 2.3449,
      "step": 53283
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7531923651695251,
      "learning_rate": 0.0005242435142011636,
      "loss": 2.3068,
      "step": 53284
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7704173922538757,
      "learning_rate": 0.0005242407968765791,
      "loss": 2.377,
      "step": 53285
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.6868159174919128,
      "learning_rate": 0.0005242380795103039,
      "loss": 2.3545,
      "step": 53286
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7100377082824707,
      "learning_rate": 0.0005242353621023387,
      "loss": 2.2601,
      "step": 53287
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7399786114692688,
      "learning_rate": 0.000524232644652684,
      "loss": 2.1562,
      "step": 53288
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7874155044555664,
      "learning_rate": 0.0005242299271613402,
      "loss": 2.3247,
      "step": 53289
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.6952585577964783,
      "learning_rate": 0.0005242272096283077,
      "loss": 2.389,
      "step": 53290
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7531477808952332,
      "learning_rate": 0.0005242244920535872,
      "loss": 2.3593,
      "step": 53291
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.6723951101303101,
      "learning_rate": 0.0005242217744371792,
      "loss": 2.1853,
      "step": 53292
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.6901235580444336,
      "learning_rate": 0.000524219056779084,
      "loss": 2.1457,
      "step": 53293
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.6876944899559021,
      "learning_rate": 0.0005242163390793024,
      "loss": 2.2211,
      "step": 53294
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7639942169189453,
      "learning_rate": 0.0005242136213378346,
      "loss": 2.0503,
      "step": 53295
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7139038443565369,
      "learning_rate": 0.0005242109035546813,
      "loss": 2.2906,
      "step": 53296
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7898904085159302,
      "learning_rate": 0.000524208185729843,
      "loss": 2.1906,
      "step": 53297
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.6858775615692139,
      "learning_rate": 0.0005242054678633202,
      "loss": 2.3796,
      "step": 53298
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7585856318473816,
      "learning_rate": 0.0005242027499551133,
      "loss": 2.2825,
      "step": 53299
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7484927773475647,
      "learning_rate": 0.0005242000320052229,
      "loss": 2.3164,
      "step": 53300
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.6507616639137268,
      "learning_rate": 0.0005241973140136494,
      "loss": 2.2675,
      "step": 53301
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.6766840219497681,
      "learning_rate": 0.0005241945959803935,
      "loss": 2.349,
      "step": 53302
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.8309949636459351,
      "learning_rate": 0.0005241918779054555,
      "loss": 2.2891,
      "step": 53303
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7353389859199524,
      "learning_rate": 0.000524189159788836,
      "loss": 2.3253,
      "step": 53304
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7239059805870056,
      "learning_rate": 0.0005241864416305357,
      "loss": 2.2652,
      "step": 53305
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7502617835998535,
      "learning_rate": 0.0005241837234305547,
      "loss": 1.9952,
      "step": 53306
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7538092732429504,
      "learning_rate": 0.0005241810051888938,
      "loss": 2.4402,
      "step": 53307
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7350296378135681,
      "learning_rate": 0.0005241782869055534,
      "loss": 2.3012,
      "step": 53308
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.6464038491249084,
      "learning_rate": 0.000524175568580534,
      "loss": 2.1802,
      "step": 53309
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.6913802027702332,
      "learning_rate": 0.0005241728502138361,
      "loss": 2.0447,
      "step": 53310
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.8235549926757812,
      "learning_rate": 0.0005241701318054603,
      "loss": 2.5017,
      "step": 53311
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7319413423538208,
      "learning_rate": 0.0005241674133554071,
      "loss": 2.0946,
      "step": 53312
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.6712439656257629,
      "learning_rate": 0.0005241646948636768,
      "loss": 2.1533,
      "step": 53313
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7164729833602905,
      "learning_rate": 0.0005241619763302702,
      "loss": 1.9276,
      "step": 53314
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.8360509276390076,
      "learning_rate": 0.0005241592577551876,
      "loss": 2.2118,
      "step": 53315
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.6636343002319336,
      "learning_rate": 0.0005241565391384296,
      "loss": 2.1616,
      "step": 53316
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.6878376007080078,
      "learning_rate": 0.0005241538204799965,
      "loss": 2.451,
      "step": 53317
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7805812358856201,
      "learning_rate": 0.0005241511017798891,
      "loss": 2.3281,
      "step": 53318
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.719103991985321,
      "learning_rate": 0.0005241483830381077,
      "loss": 2.3495,
      "step": 53319
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7259904146194458,
      "learning_rate": 0.0005241456642546531,
      "loss": 2.2543,
      "step": 53320
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7707989811897278,
      "learning_rate": 0.0005241429454295254,
      "loss": 2.2104,
      "step": 53321
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7585318088531494,
      "learning_rate": 0.0005241402265627253,
      "loss": 2.1205,
      "step": 53322
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.670982301235199,
      "learning_rate": 0.0005241375076542533,
      "loss": 2.314,
      "step": 53323
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7643867135047913,
      "learning_rate": 0.00052413478870411,
      "loss": 2.3794,
      "step": 53324
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.6561654806137085,
      "learning_rate": 0.0005241320697122957,
      "loss": 2.0397,
      "step": 53325
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7209551334381104,
      "learning_rate": 0.0005241293506788112,
      "loss": 2.2915,
      "step": 53326
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7833914756774902,
      "learning_rate": 0.0005241266316036566,
      "loss": 2.2405,
      "step": 53327
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.777120053768158,
      "learning_rate": 0.0005241239124868327,
      "loss": 2.1446,
      "step": 53328
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7959508299827576,
      "learning_rate": 0.00052412119332834,
      "loss": 2.127,
      "step": 53329
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7239875793457031,
      "learning_rate": 0.0005241184741281788,
      "loss": 2.2952,
      "step": 53330
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7810123562812805,
      "learning_rate": 0.00052411575488635,
      "loss": 2.6919,
      "step": 53331
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7468369007110596,
      "learning_rate": 0.0005241130356028536,
      "loss": 2.2969,
      "step": 53332
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7068464159965515,
      "learning_rate": 0.0005241103162776905,
      "loss": 2.0847,
      "step": 53333
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7190394997596741,
      "learning_rate": 0.000524107596910861,
      "loss": 2.3569,
      "step": 53334
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7292727828025818,
      "learning_rate": 0.0005241048775023656,
      "loss": 2.1261,
      "step": 53335
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.8423408269882202,
      "learning_rate": 0.000524102158052205,
      "loss": 2.1182,
      "step": 53336
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7722931504249573,
      "learning_rate": 0.0005240994385603796,
      "loss": 2.0461,
      "step": 53337
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7219992280006409,
      "learning_rate": 0.0005240967190268898,
      "loss": 2.0297,
      "step": 53338
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7906603217124939,
      "learning_rate": 0.0005240939994517362,
      "loss": 2.3777,
      "step": 53339
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.8787949085235596,
      "learning_rate": 0.0005240912798349193,
      "loss": 2.4021,
      "step": 53340
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7182278037071228,
      "learning_rate": 0.0005240885601764397,
      "loss": 2.2153,
      "step": 53341
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7723338603973389,
      "learning_rate": 0.0005240858404762976,
      "loss": 2.4476,
      "step": 53342
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.781023383140564,
      "learning_rate": 0.0005240831207344939,
      "loss": 2.228,
      "step": 53343
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7714534997940063,
      "learning_rate": 0.0005240804009510288,
      "loss": 2.346,
      "step": 53344
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.6578890085220337,
      "learning_rate": 0.0005240776811259029,
      "loss": 1.9615,
      "step": 53345
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7253119349479675,
      "learning_rate": 0.0005240749612591168,
      "loss": 2.2923,
      "step": 53346
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7929012775421143,
      "learning_rate": 0.0005240722413506709,
      "loss": 2.1799,
      "step": 53347
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.6945451498031616,
      "learning_rate": 0.0005240695214005658,
      "loss": 2.187,
      "step": 53348
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7717111110687256,
      "learning_rate": 0.000524066801408802,
      "loss": 2.3548,
      "step": 53349
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.6928923726081848,
      "learning_rate": 0.0005240640813753798,
      "loss": 2.054,
      "step": 53350
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.682614803314209,
      "learning_rate": 0.0005240613613002999,
      "loss": 2.1987,
      "step": 53351
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.8107877969741821,
      "learning_rate": 0.0005240586411835628,
      "loss": 2.1997,
      "step": 53352
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.0250152349472046,
      "learning_rate": 0.0005240559210251689,
      "loss": 2.0722,
      "step": 53353
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7412725687026978,
      "learning_rate": 0.0005240532008251189,
      "loss": 2.2399,
      "step": 53354
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.9032357931137085,
      "learning_rate": 0.0005240504805834131,
      "loss": 2.3153,
      "step": 53355
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.1208075284957886,
      "learning_rate": 0.0005240477603000521,
      "loss": 2.2592,
      "step": 53356
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.9021468758583069,
      "learning_rate": 0.0005240450399750364,
      "loss": 2.039,
      "step": 53357
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7747407555580139,
      "learning_rate": 0.0005240423196083665,
      "loss": 2.247,
      "step": 53358
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.6902209520339966,
      "learning_rate": 0.0005240395992000428,
      "loss": 2.1032,
      "step": 53359
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.0905342102050781,
      "learning_rate": 0.0005240368787500661,
      "loss": 2.1623,
      "step": 53360
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.8865067958831787,
      "learning_rate": 0.0005240341582584367,
      "loss": 2.3068,
      "step": 53361
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7956147193908691,
      "learning_rate": 0.000524031437725155,
      "loss": 2.1388,
      "step": 53362
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7783205509185791,
      "learning_rate": 0.0005240287171502215,
      "loss": 2.3545,
      "step": 53363
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7033345103263855,
      "learning_rate": 0.0005240259965336371,
      "loss": 2.2517,
      "step": 53364
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.9434952735900879,
      "learning_rate": 0.0005240232758754019,
      "loss": 2.1042,
      "step": 53365
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7641230225563049,
      "learning_rate": 0.0005240205551755166,
      "loss": 2.1196,
      "step": 53366
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.9552933573722839,
      "learning_rate": 0.0005240178344339816,
      "loss": 2.4353,
      "step": 53367
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7472471594810486,
      "learning_rate": 0.0005240151136507975,
      "loss": 2.2823,
      "step": 53368
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7231735587120056,
      "learning_rate": 0.0005240123928259648,
      "loss": 2.2974,
      "step": 53369
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.6934873461723328,
      "learning_rate": 0.0005240096719594839,
      "loss": 2.1929,
      "step": 53370
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.86440509557724,
      "learning_rate": 0.0005240069510513553,
      "loss": 2.2094,
      "step": 53371
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7168779373168945,
      "learning_rate": 0.0005240042301015797,
      "loss": 2.1882,
      "step": 53372
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7345311641693115,
      "learning_rate": 0.0005240015091101575,
      "loss": 2.3822,
      "step": 53373
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7434737682342529,
      "learning_rate": 0.0005239987880770891,
      "loss": 2.2761,
      "step": 53374
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7482578158378601,
      "learning_rate": 0.0005239960670023751,
      "loss": 2.1941,
      "step": 53375
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7289304137229919,
      "learning_rate": 0.0005239933458860162,
      "loss": 2.2887,
      "step": 53376
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7229799032211304,
      "learning_rate": 0.0005239906247280125,
      "loss": 2.2805,
      "step": 53377
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7154006958007812,
      "learning_rate": 0.0005239879035283648,
      "loss": 2.1675,
      "step": 53378
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7593151926994324,
      "learning_rate": 0.0005239851822870734,
      "loss": 2.1233,
      "step": 53379
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7745317816734314,
      "learning_rate": 0.000523982461004139,
      "loss": 2.5353,
      "step": 53380
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.706230640411377,
      "learning_rate": 0.0005239797396795621,
      "loss": 2.2506,
      "step": 53381
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7328310608863831,
      "learning_rate": 0.000523977018313343,
      "loss": 2.1672,
      "step": 53382
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.6706772446632385,
      "learning_rate": 0.0005239742969054825,
      "loss": 2.0478,
      "step": 53383
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.8834144473075867,
      "learning_rate": 0.0005239715754559809,
      "loss": 2.048,
      "step": 53384
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.6262277364730835,
      "learning_rate": 0.0005239688539648387,
      "loss": 2.0985,
      "step": 53385
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.696736216545105,
      "learning_rate": 0.0005239661324320565,
      "loss": 2.5308,
      "step": 53386
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.8427422046661377,
      "learning_rate": 0.0005239634108576348,
      "loss": 2.1565,
      "step": 53387
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.9489619731903076,
      "learning_rate": 0.000523960689241574,
      "loss": 2.3789,
      "step": 53388
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7121726870536804,
      "learning_rate": 0.0005239579675838748,
      "loss": 2.229,
      "step": 53389
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.745897114276886,
      "learning_rate": 0.0005239552458845376,
      "loss": 2.4392,
      "step": 53390
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.9218851923942566,
      "learning_rate": 0.0005239525241435627,
      "loss": 2.3956,
      "step": 53391
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7009562849998474,
      "learning_rate": 0.0005239498023609509,
      "loss": 2.0478,
      "step": 53392
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7132939100265503,
      "learning_rate": 0.0005239470805367026,
      "loss": 2.1817,
      "step": 53393
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7981712222099304,
      "learning_rate": 0.0005239443586708182,
      "loss": 2.5,
      "step": 53394
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7390071749687195,
      "learning_rate": 0.0005239416367632985,
      "loss": 2.2565,
      "step": 53395
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7571144700050354,
      "learning_rate": 0.0005239389148141438,
      "loss": 2.2064,
      "step": 53396
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.6872497797012329,
      "learning_rate": 0.0005239361928233545,
      "loss": 2.1189,
      "step": 53397
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7789784669876099,
      "learning_rate": 0.0005239334707909313,
      "loss": 2.1295,
      "step": 53398
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.6927944421768188,
      "learning_rate": 0.0005239307487168748,
      "loss": 2.1502,
      "step": 53399
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.6837619543075562,
      "learning_rate": 0.0005239280266011851,
      "loss": 2.3129,
      "step": 53400
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7612565159797668,
      "learning_rate": 0.0005239253044438631,
      "loss": 2.2947,
      "step": 53401
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.6857729554176331,
      "learning_rate": 0.0005239225822449091,
      "loss": 2.1551,
      "step": 53402
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7141554951667786,
      "learning_rate": 0.0005239198600043237,
      "loss": 2.2561,
      "step": 53403
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7538033723831177,
      "learning_rate": 0.0005239171377221074,
      "loss": 2.0945,
      "step": 53404
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.8594836592674255,
      "learning_rate": 0.0005239144153982606,
      "loss": 2.0131,
      "step": 53405
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7439397573471069,
      "learning_rate": 0.0005239116930327839,
      "loss": 2.6175,
      "step": 53406
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.823815107345581,
      "learning_rate": 0.0005239089706256777,
      "loss": 2.3561,
      "step": 53407
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.8904247283935547,
      "learning_rate": 0.0005239062481769427,
      "loss": 2.2464,
      "step": 53408
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7731122374534607,
      "learning_rate": 0.0005239035256865794,
      "loss": 2.2339,
      "step": 53409
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7881969809532166,
      "learning_rate": 0.0005239008031545881,
      "loss": 2.3365,
      "step": 53410
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.0691241025924683,
      "learning_rate": 0.0005238980805809694,
      "loss": 2.3161,
      "step": 53411
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.74486243724823,
      "learning_rate": 0.0005238953579657238,
      "loss": 2.2366,
      "step": 53412
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.6923487782478333,
      "learning_rate": 0.0005238926353088519,
      "loss": 2.2174,
      "step": 53413
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7354894876480103,
      "learning_rate": 0.0005238899126103541,
      "loss": 2.4107,
      "step": 53414
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.9030090570449829,
      "learning_rate": 0.000523887189870231,
      "loss": 2.1472,
      "step": 53415
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7106367945671082,
      "learning_rate": 0.000523884467088483,
      "loss": 2.2683,
      "step": 53416
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7484785914421082,
      "learning_rate": 0.0005238817442651106,
      "loss": 2.1954,
      "step": 53417
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7471296787261963,
      "learning_rate": 0.0005238790214001144,
      "loss": 2.1728,
      "step": 53418
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7625128030776978,
      "learning_rate": 0.0005238762984934948,
      "loss": 2.4961,
      "step": 53419
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.811618983745575,
      "learning_rate": 0.0005238735755452525,
      "loss": 2.1467,
      "step": 53420
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7791439294815063,
      "learning_rate": 0.0005238708525553878,
      "loss": 2.4193,
      "step": 53421
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.6939653158187866,
      "learning_rate": 0.0005238681295239013,
      "loss": 2.2594,
      "step": 53422
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.0472962856292725,
      "learning_rate": 0.0005238654064507934,
      "loss": 2.3202,
      "step": 53423
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7858850955963135,
      "learning_rate": 0.0005238626833360648,
      "loss": 2.0587,
      "step": 53424
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7659399509429932,
      "learning_rate": 0.0005238599601797159,
      "loss": 2.1794,
      "step": 53425
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.9910415410995483,
      "learning_rate": 0.0005238572369817472,
      "loss": 2.1658,
      "step": 53426
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.8012010455131531,
      "learning_rate": 0.0005238545137421591,
      "loss": 2.1731,
      "step": 53427
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.6854488849639893,
      "learning_rate": 0.0005238517904609524,
      "loss": 2.3885,
      "step": 53428
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.845036506652832,
      "learning_rate": 0.0005238490671381274,
      "loss": 2.126,
      "step": 53429
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.0062315464019775,
      "learning_rate": 0.0005238463437736846,
      "loss": 2.4465,
      "step": 53430
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.8794227838516235,
      "learning_rate": 0.0005238436203676244,
      "loss": 2.3221,
      "step": 53431
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7020211219787598,
      "learning_rate": 0.0005238408969199476,
      "loss": 2.0214,
      "step": 53432
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.8819947838783264,
      "learning_rate": 0.0005238381734306545,
      "loss": 2.2935,
      "step": 53433
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.9934878945350647,
      "learning_rate": 0.0005238354498997457,
      "loss": 2.17,
      "step": 53434
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7426034808158875,
      "learning_rate": 0.0005238327263272216,
      "loss": 2.1473,
      "step": 53435
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.8834328651428223,
      "learning_rate": 0.0005238300027130829,
      "loss": 2.3938,
      "step": 53436
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.0024545192718506,
      "learning_rate": 0.0005238272790573298,
      "loss": 1.9799,
      "step": 53437
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7460244297981262,
      "learning_rate": 0.0005238245553599631,
      "loss": 2.1766,
      "step": 53438
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7334990501403809,
      "learning_rate": 0.0005238218316209831,
      "loss": 2.1795,
      "step": 53439
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7511106729507446,
      "learning_rate": 0.0005238191078403904,
      "loss": 2.2947,
      "step": 53440
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7377594113349915,
      "learning_rate": 0.0005238163840181856,
      "loss": 2.1957,
      "step": 53441
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7420209050178528,
      "learning_rate": 0.000523813660154369,
      "loss": 2.3528,
      "step": 53442
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7456640005111694,
      "learning_rate": 0.0005238109362489413,
      "loss": 2.2578,
      "step": 53443
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7946483492851257,
      "learning_rate": 0.0005238082123019029,
      "loss": 2.3591,
      "step": 53444
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7403627634048462,
      "learning_rate": 0.0005238054883132543,
      "loss": 2.1866,
      "step": 53445
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7432719469070435,
      "learning_rate": 0.0005238027642829962,
      "loss": 2.1796,
      "step": 53446
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.869909942150116,
      "learning_rate": 0.0005238000402111287,
      "loss": 2.0767,
      "step": 53447
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7273648381233215,
      "learning_rate": 0.0005237973160976527,
      "loss": 1.9709,
      "step": 53448
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7411403059959412,
      "learning_rate": 0.0005237945919425686,
      "loss": 2.3611,
      "step": 53449
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.8215027451515198,
      "learning_rate": 0.0005237918677458767,
      "loss": 2.1213,
      "step": 53450
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7489351630210876,
      "learning_rate": 0.0005237891435075777,
      "loss": 2.2446,
      "step": 53451
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.697124183177948,
      "learning_rate": 0.0005237864192276722,
      "loss": 2.0207,
      "step": 53452
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7660968899726868,
      "learning_rate": 0.0005237836949061605,
      "loss": 2.4666,
      "step": 53453
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7455653548240662,
      "learning_rate": 0.0005237809705430431,
      "loss": 2.2612,
      "step": 53454
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.75799560546875,
      "learning_rate": 0.0005237782461383207,
      "loss": 2.1777,
      "step": 53455
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.6686487197875977,
      "learning_rate": 0.0005237755216919936,
      "loss": 2.3517,
      "step": 53456
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7010961771011353,
      "learning_rate": 0.0005237727972040626,
      "loss": 2.291,
      "step": 53457
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7491225004196167,
      "learning_rate": 0.0005237700726745279,
      "loss": 2.3427,
      "step": 53458
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7283812165260315,
      "learning_rate": 0.00052376734810339,
      "loss": 2.4831,
      "step": 53459
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7277742028236389,
      "learning_rate": 0.0005237646234906498,
      "loss": 2.2898,
      "step": 53460
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7506743669509888,
      "learning_rate": 0.0005237618988363073,
      "loss": 2.4385,
      "step": 53461
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7956585884094238,
      "learning_rate": 0.0005237591741403633,
      "loss": 2.3521,
      "step": 53462
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.715465784072876,
      "learning_rate": 0.0005237564494028182,
      "loss": 2.2461,
      "step": 53463
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7658122181892395,
      "learning_rate": 0.0005237537246236727,
      "loss": 2.269,
      "step": 53464
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.754250168800354,
      "learning_rate": 0.000523750999802927,
      "loss": 2.2171,
      "step": 53465
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7322385311126709,
      "learning_rate": 0.0005237482749405819,
      "loss": 2.4288,
      "step": 53466
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.6811990141868591,
      "learning_rate": 0.0005237455500366377,
      "loss": 2.2642,
      "step": 53467
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7216571569442749,
      "learning_rate": 0.000523742825091095,
      "loss": 2.3035,
      "step": 53468
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7201475501060486,
      "learning_rate": 0.0005237401001039542,
      "loss": 2.1854,
      "step": 53469
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.86494380235672,
      "learning_rate": 0.000523737375075216,
      "loss": 2.3793,
      "step": 53470
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7043038010597229,
      "learning_rate": 0.0005237346500048807,
      "loss": 2.2228,
      "step": 53471
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7199200987815857,
      "learning_rate": 0.0005237319248929491,
      "loss": 2.2002,
      "step": 53472
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.8217135667800903,
      "learning_rate": 0.0005237291997394213,
      "loss": 2.2347,
      "step": 53473
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7572300434112549,
      "learning_rate": 0.0005237264745442981,
      "loss": 2.1805,
      "step": 53474
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7744251489639282,
      "learning_rate": 0.00052372374930758,
      "loss": 2.1877,
      "step": 53475
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.6936486959457397,
      "learning_rate": 0.0005237210240292672,
      "loss": 2.0692,
      "step": 53476
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.8023493885993958,
      "learning_rate": 0.0005237182987093607,
      "loss": 2.1759,
      "step": 53477
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.8174073696136475,
      "learning_rate": 0.0005237155733478606,
      "loss": 2.1679,
      "step": 53478
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.8002102971076965,
      "learning_rate": 0.0005237128479447675,
      "loss": 2.2422,
      "step": 53479
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7193833589553833,
      "learning_rate": 0.0005237101225000821,
      "loss": 2.1401,
      "step": 53480
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.8672506213188171,
      "learning_rate": 0.0005237073970138047,
      "loss": 1.9115,
      "step": 53481
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7648743987083435,
      "learning_rate": 0.0005237046714859359,
      "loss": 2.274,
      "step": 53482
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.8051277995109558,
      "learning_rate": 0.0005237019459164762,
      "loss": 2.2169,
      "step": 53483
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7930187582969666,
      "learning_rate": 0.000523699220305426,
      "loss": 2.107,
      "step": 53484
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7910524606704712,
      "learning_rate": 0.000523696494652786,
      "loss": 2.3508,
      "step": 53485
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7376556396484375,
      "learning_rate": 0.0005236937689585564,
      "loss": 2.3988,
      "step": 53486
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7490383982658386,
      "learning_rate": 0.0005236910432227382,
      "loss": 2.4601,
      "step": 53487
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.803984522819519,
      "learning_rate": 0.0005236883174453313,
      "loss": 2.0246,
      "step": 53488
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.6697700023651123,
      "learning_rate": 0.0005236855916263368,
      "loss": 2.1879,
      "step": 53489
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7454961538314819,
      "learning_rate": 0.0005236828657657548,
      "loss": 2.1663,
      "step": 53490
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.8169015645980835,
      "learning_rate": 0.0005236801398635859,
      "loss": 2.1889,
      "step": 53491
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7020577788352966,
      "learning_rate": 0.0005236774139198308,
      "loss": 2.2509,
      "step": 53492
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.9264461398124695,
      "learning_rate": 0.0005236746879344897,
      "loss": 2.3303,
      "step": 53493
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7229068875312805,
      "learning_rate": 0.0005236719619075632,
      "loss": 2.3183,
      "step": 53494
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.8221144676208496,
      "learning_rate": 0.0005236692358390519,
      "loss": 2.3409,
      "step": 53495
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.0330711603164673,
      "learning_rate": 0.0005236665097289564,
      "loss": 2.2441,
      "step": 53496
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.66831374168396,
      "learning_rate": 0.0005236637835772769,
      "loss": 2.1125,
      "step": 53497
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.724646806716919,
      "learning_rate": 0.0005236610573840142,
      "loss": 2.2699,
      "step": 53498
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.9639710783958435,
      "learning_rate": 0.0005236583311491686,
      "loss": 2.0909,
      "step": 53499
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.9363076686859131,
      "learning_rate": 0.0005236556048727408,
      "loss": 2.2565,
      "step": 53500
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.6853094100952148,
      "learning_rate": 0.0005236528785547311,
      "loss": 2.356,
      "step": 53501
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.8827908039093018,
      "learning_rate": 0.0005236501521951401,
      "loss": 2.1323,
      "step": 53502
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.8863691687583923,
      "learning_rate": 0.0005236474257939685,
      "loss": 2.2615,
      "step": 53503
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.6689690351486206,
      "learning_rate": 0.0005236446993512165,
      "loss": 2.1196,
      "step": 53504
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7533401250839233,
      "learning_rate": 0.0005236419728668846,
      "loss": 2.0281,
      "step": 53505
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.8423270583152771,
      "learning_rate": 0.0005236392463409736,
      "loss": 2.235,
      "step": 53506
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7770270705223083,
      "learning_rate": 0.0005236365197734838,
      "loss": 2.2724,
      "step": 53507
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7864441275596619,
      "learning_rate": 0.0005236337931644157,
      "loss": 2.4498,
      "step": 53508
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.8744335174560547,
      "learning_rate": 0.00052363106651377,
      "loss": 2.3255,
      "step": 53509
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.8656099438667297,
      "learning_rate": 0.0005236283398215469,
      "loss": 2.3056,
      "step": 53510
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7607248425483704,
      "learning_rate": 0.0005236256130877471,
      "loss": 2.0957,
      "step": 53511
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.8012515306472778,
      "learning_rate": 0.0005236228863123711,
      "loss": 2.2759,
      "step": 53512
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.857712984085083,
      "learning_rate": 0.0005236201594954194,
      "loss": 2.0858,
      "step": 53513
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7067596316337585,
      "learning_rate": 0.0005236174326368925,
      "loss": 2.2849,
      "step": 53514
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7794643044471741,
      "learning_rate": 0.000523614705736791,
      "loss": 2.3389,
      "step": 53515
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.6945508718490601,
      "learning_rate": 0.0005236119787951152,
      "loss": 2.1259,
      "step": 53516
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7502777576446533,
      "learning_rate": 0.0005236092518118656,
      "loss": 2.2244,
      "step": 53517
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7557913064956665,
      "learning_rate": 0.000523606524787043,
      "loss": 2.4272,
      "step": 53518
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7319906949996948,
      "learning_rate": 0.0005236037977206475,
      "loss": 2.2545,
      "step": 53519
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.8627819418907166,
      "learning_rate": 0.00052360107061268,
      "loss": 2.3147,
      "step": 53520
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.698512613773346,
      "learning_rate": 0.0005235983434631409,
      "loss": 2.0708,
      "step": 53521
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.8377130031585693,
      "learning_rate": 0.0005235956162720305,
      "loss": 2.3993,
      "step": 53522
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.6908275485038757,
      "learning_rate": 0.0005235928890393495,
      "loss": 2.3506,
      "step": 53523
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7825390100479126,
      "learning_rate": 0.0005235901617650984,
      "loss": 2.4046,
      "step": 53524
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.1026581525802612,
      "learning_rate": 0.0005235874344492776,
      "loss": 2.373,
      "step": 53525
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7510084509849548,
      "learning_rate": 0.0005235847070918879,
      "loss": 2.173,
      "step": 53526
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.6607579588890076,
      "learning_rate": 0.0005235819796929293,
      "loss": 2.1179,
      "step": 53527
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.8364010453224182,
      "learning_rate": 0.0005235792522524027,
      "loss": 2.4162,
      "step": 53528
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7939903736114502,
      "learning_rate": 0.0005235765247703085,
      "loss": 2.2042,
      "step": 53529
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.8488361239433289,
      "learning_rate": 0.0005235737972466471,
      "loss": 2.2488,
      "step": 53530
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7235526442527771,
      "learning_rate": 0.0005235710696814193,
      "loss": 2.2576,
      "step": 53531
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7586182951927185,
      "learning_rate": 0.0005235683420746253,
      "loss": 2.0641,
      "step": 53532
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7957825660705566,
      "learning_rate": 0.0005235656144262658,
      "loss": 2.1068,
      "step": 53533
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7696945071220398,
      "learning_rate": 0.0005235628867363411,
      "loss": 2.4017,
      "step": 53534
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.769945502281189,
      "learning_rate": 0.000523560159004852,
      "loss": 2.2508,
      "step": 53535
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.807108998298645,
      "learning_rate": 0.0005235574312317988,
      "loss": 2.2434,
      "step": 53536
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7041144371032715,
      "learning_rate": 0.000523554703417182,
      "loss": 2.045,
      "step": 53537
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7499496340751648,
      "learning_rate": 0.0005235519755610021,
      "loss": 1.9766,
      "step": 53538
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7851864695549011,
      "learning_rate": 0.0005235492476632598,
      "loss": 2.2237,
      "step": 53539
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7878944873809814,
      "learning_rate": 0.0005235465197239554,
      "loss": 2.1641,
      "step": 53540
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.729866623878479,
      "learning_rate": 0.0005235437917430895,
      "loss": 2.1833,
      "step": 53541
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.6999486088752747,
      "learning_rate": 0.0005235410637206626,
      "loss": 2.2646,
      "step": 53542
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7609847784042358,
      "learning_rate": 0.0005235383356566752,
      "loss": 2.4888,
      "step": 53543
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.8198727369308472,
      "learning_rate": 0.0005235356075511279,
      "loss": 2.1833,
      "step": 53544
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.6706586480140686,
      "learning_rate": 0.000523532879404021,
      "loss": 2.3145,
      "step": 53545
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.6629168391227722,
      "learning_rate": 0.000523530151215355,
      "loss": 2.2865,
      "step": 53546
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7380210161209106,
      "learning_rate": 0.0005235274229851307,
      "loss": 2.094,
      "step": 53547
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7155972719192505,
      "learning_rate": 0.0005235246947133484,
      "loss": 2.3406,
      "step": 53548
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7103670239448547,
      "learning_rate": 0.0005235219664000087,
      "loss": 2.272,
      "step": 53549
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7496001720428467,
      "learning_rate": 0.000523519238045112,
      "loss": 1.8681,
      "step": 53550
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7724013924598694,
      "learning_rate": 0.0005235165096486588,
      "loss": 2.1778,
      "step": 53551
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7616997361183167,
      "learning_rate": 0.0005235137812106497,
      "loss": 2.1779,
      "step": 53552
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7162988781929016,
      "learning_rate": 0.0005235110527310852,
      "loss": 2.3621,
      "step": 53553
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.8679670095443726,
      "learning_rate": 0.0005235083242099657,
      "loss": 2.2093,
      "step": 53554
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.6855819821357727,
      "learning_rate": 0.0005235055956472918,
      "loss": 2.3026,
      "step": 53555
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7320259213447571,
      "learning_rate": 0.000523502867043064,
      "loss": 2.3085,
      "step": 53556
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7346767783164978,
      "learning_rate": 0.0005235001383972828,
      "loss": 2.2726,
      "step": 53557
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7451096177101135,
      "learning_rate": 0.0005234974097099487,
      "loss": 2.1289,
      "step": 53558
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.6848495006561279,
      "learning_rate": 0.0005234946809810623,
      "loss": 2.2929,
      "step": 53559
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.8654614090919495,
      "learning_rate": 0.0005234919522106239,
      "loss": 2.4422,
      "step": 53560
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7587233781814575,
      "learning_rate": 0.0005234892233986341,
      "loss": 2.3446,
      "step": 53561
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7956334948539734,
      "learning_rate": 0.0005234864945450936,
      "loss": 2.498,
      "step": 53562
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.8195297718048096,
      "learning_rate": 0.0005234837656500026,
      "loss": 2.1691,
      "step": 53563
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.809890866279602,
      "learning_rate": 0.0005234810367133619,
      "loss": 2.2521,
      "step": 53564
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7822308540344238,
      "learning_rate": 0.0005234783077351717,
      "loss": 2.3962,
      "step": 53565
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7091029286384583,
      "learning_rate": 0.0005234755787154327,
      "loss": 2.174,
      "step": 53566
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7344364523887634,
      "learning_rate": 0.0005234728496541454,
      "loss": 2.2154,
      "step": 53567
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7329960465431213,
      "learning_rate": 0.0005234701205513103,
      "loss": 2.1629,
      "step": 53568
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.9408736228942871,
      "learning_rate": 0.0005234673914069278,
      "loss": 2.0643,
      "step": 53569
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7420775890350342,
      "learning_rate": 0.0005234646622209986,
      "loss": 2.4606,
      "step": 53570
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.8227288126945496,
      "learning_rate": 0.000523461932993523,
      "loss": 2.0686,
      "step": 53571
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7480866312980652,
      "learning_rate": 0.0005234592037245017,
      "loss": 2.2324,
      "step": 53572
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.846059262752533,
      "learning_rate": 0.0005234564744139351,
      "loss": 2.1821,
      "step": 53573
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.745697021484375,
      "learning_rate": 0.0005234537450618237,
      "loss": 2.2822,
      "step": 53574
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7404388189315796,
      "learning_rate": 0.0005234510156681679,
      "loss": 2.2668,
      "step": 53575
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7194554805755615,
      "learning_rate": 0.0005234482862329686,
      "loss": 2.4642,
      "step": 53576
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.6737251877784729,
      "learning_rate": 0.0005234455567562259,
      "loss": 1.9706,
      "step": 53577
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7890584468841553,
      "learning_rate": 0.0005234428272379404,
      "loss": 2.1501,
      "step": 53578
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7895931005477905,
      "learning_rate": 0.0005234400976781127,
      "loss": 2.1341,
      "step": 53579
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.753663957118988,
      "learning_rate": 0.0005234373680767433,
      "loss": 2.1229,
      "step": 53580
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7712792754173279,
      "learning_rate": 0.0005234346384338327,
      "loss": 2.3216,
      "step": 53581
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7676025032997131,
      "learning_rate": 0.0005234319087493814,
      "loss": 2.1335,
      "step": 53582
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7353826761245728,
      "learning_rate": 0.0005234291790233898,
      "loss": 2.4951,
      "step": 53583
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7190621495246887,
      "learning_rate": 0.0005234264492558585,
      "loss": 2.4113,
      "step": 53584
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7099952101707458,
      "learning_rate": 0.0005234237194467881,
      "loss": 2.1392,
      "step": 53585
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7030792832374573,
      "learning_rate": 0.000523420989596179,
      "loss": 2.2031,
      "step": 53586
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7376285791397095,
      "learning_rate": 0.0005234182597040316,
      "loss": 2.2793,
      "step": 53587
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.6842334866523743,
      "learning_rate": 0.0005234155297703466,
      "loss": 2.3696,
      "step": 53588
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7959226369857788,
      "learning_rate": 0.0005234127997951245,
      "loss": 2.2446,
      "step": 53589
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.6903530955314636,
      "learning_rate": 0.0005234100697783657,
      "loss": 2.1171,
      "step": 53590
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7153552770614624,
      "learning_rate": 0.0005234073397200707,
      "loss": 2.2274,
      "step": 53591
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.9144067168235779,
      "learning_rate": 0.0005234046096202401,
      "loss": 2.3145,
      "step": 53592
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7359910607337952,
      "learning_rate": 0.0005234018794788744,
      "loss": 2.249,
      "step": 53593
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7558603882789612,
      "learning_rate": 0.000523399149295974,
      "loss": 2.4236,
      "step": 53594
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7922936677932739,
      "learning_rate": 0.0005233964190715396,
      "loss": 2.3692,
      "step": 53595
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7616099119186401,
      "learning_rate": 0.0005233936888055714,
      "loss": 2.3498,
      "step": 53596
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7149261832237244,
      "learning_rate": 0.0005233909584980703,
      "loss": 2.1888,
      "step": 53597
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.8179548382759094,
      "learning_rate": 0.0005233882281490364,
      "loss": 2.1841,
      "step": 53598
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.6950511336326599,
      "learning_rate": 0.0005233854977584706,
      "loss": 2.0233,
      "step": 53599
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.787432849407196,
      "learning_rate": 0.0005233827673263731,
      "loss": 2.3266,
      "step": 53600
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.6713427901268005,
      "learning_rate": 0.0005233800368527445,
      "loss": 2.1983,
      "step": 53601
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7694499492645264,
      "learning_rate": 0.0005233773063375855,
      "loss": 2.1639,
      "step": 53602
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.719875156879425,
      "learning_rate": 0.0005233745757808964,
      "loss": 2.1224,
      "step": 53603
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.6725170016288757,
      "learning_rate": 0.0005233718451826777,
      "loss": 2.103,
      "step": 53604
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7503424286842346,
      "learning_rate": 0.0005233691145429299,
      "loss": 2.5236,
      "step": 53605
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7615271806716919,
      "learning_rate": 0.0005233663838616536,
      "loss": 2.2065,
      "step": 53606
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7214093804359436,
      "learning_rate": 0.0005233636531388493,
      "loss": 2.3176,
      "step": 53607
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7312120795249939,
      "learning_rate": 0.0005233609223745174,
      "loss": 2.0239,
      "step": 53608
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.8209894299507141,
      "learning_rate": 0.0005233581915686586,
      "loss": 2.2363,
      "step": 53609
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.6926879286766052,
      "learning_rate": 0.0005233554607212733,
      "loss": 2.1985,
      "step": 53610
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7161132097244263,
      "learning_rate": 0.000523352729832362,
      "loss": 2.3234,
      "step": 53611
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7363694310188293,
      "learning_rate": 0.0005233499989019251,
      "loss": 2.1809,
      "step": 53612
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.6977131366729736,
      "learning_rate": 0.0005233472679299632,
      "loss": 2.1403,
      "step": 53613
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7528301477432251,
      "learning_rate": 0.000523344536916477,
      "loss": 2.1951,
      "step": 53614
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7622473835945129,
      "learning_rate": 0.0005233418058614667,
      "loss": 2.1346,
      "step": 53615
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7359782457351685,
      "learning_rate": 0.000523339074764933,
      "loss": 2.1532,
      "step": 53616
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.769191324710846,
      "learning_rate": 0.0005233363436268764,
      "loss": 2.2638,
      "step": 53617
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.8114593029022217,
      "learning_rate": 0.0005233336124472973,
      "loss": 1.9758,
      "step": 53618
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7362695336341858,
      "learning_rate": 0.0005233308812261962,
      "loss": 2.3675,
      "step": 53619
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7451592087745667,
      "learning_rate": 0.0005233281499635738,
      "loss": 2.5348,
      "step": 53620
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7495726943016052,
      "learning_rate": 0.0005233254186594304,
      "loss": 2.1373,
      "step": 53621
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7902098894119263,
      "learning_rate": 0.0005233226873137666,
      "loss": 2.2185,
      "step": 53622
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.6782227158546448,
      "learning_rate": 0.0005233199559265829,
      "loss": 2.2579,
      "step": 53623
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.8202827572822571,
      "learning_rate": 0.0005233172244978798,
      "loss": 2.1242,
      "step": 53624
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7348642349243164,
      "learning_rate": 0.0005233144930276579,
      "loss": 2.3369,
      "step": 53625
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7279490828514099,
      "learning_rate": 0.0005233117615159174,
      "loss": 2.2959,
      "step": 53626
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.8252779841423035,
      "learning_rate": 0.0005233090299626592,
      "loss": 2.1815,
      "step": 53627
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.6559805274009705,
      "learning_rate": 0.0005233062983678836,
      "loss": 2.251,
      "step": 53628
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7131004929542542,
      "learning_rate": 0.0005233035667315913,
      "loss": 2.0855,
      "step": 53629
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7455781102180481,
      "learning_rate": 0.0005233008350537824,
      "loss": 1.9751,
      "step": 53630
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7086198925971985,
      "learning_rate": 0.0005232981033344578,
      "loss": 2.1272,
      "step": 53631
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.6712733507156372,
      "learning_rate": 0.0005232953715736178,
      "loss": 2.0794,
      "step": 53632
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7690053582191467,
      "learning_rate": 0.000523292639771263,
      "loss": 2.481,
      "step": 53633
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7416684627532959,
      "learning_rate": 0.0005232899079273939,
      "loss": 2.0532,
      "step": 53634
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7155364751815796,
      "learning_rate": 0.000523287176042011,
      "loss": 2.0035,
      "step": 53635
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.8155367374420166,
      "learning_rate": 0.0005232844441151147,
      "loss": 2.0829,
      "step": 53636
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.6645462512969971,
      "learning_rate": 0.0005232817121467058,
      "loss": 2.3924,
      "step": 53637
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.683861494064331,
      "learning_rate": 0.0005232789801367845,
      "loss": 2.1826,
      "step": 53638
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7110449075698853,
      "learning_rate": 0.0005232762480853513,
      "loss": 2.1363,
      "step": 53639
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.6963984966278076,
      "learning_rate": 0.000523273515992407,
      "loss": 2.2485,
      "step": 53640
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7757997512817383,
      "learning_rate": 0.0005232707838579519,
      "loss": 2.2452,
      "step": 53641
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7580887079238892,
      "learning_rate": 0.0005232680516819865,
      "loss": 2.1749,
      "step": 53642
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.702701985836029,
      "learning_rate": 0.0005232653194645115,
      "loss": 2.2692,
      "step": 53643
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7185567021369934,
      "learning_rate": 0.0005232625872055272,
      "loss": 2.1911,
      "step": 53644
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7185554504394531,
      "learning_rate": 0.000523259854905034,
      "loss": 2.3678,
      "step": 53645
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.6632964611053467,
      "learning_rate": 0.0005232571225630328,
      "loss": 2.1091,
      "step": 53646
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7063965797424316,
      "learning_rate": 0.0005232543901795237,
      "loss": 2.2391,
      "step": 53647
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7417425513267517,
      "learning_rate": 0.0005232516577545075,
      "loss": 2.4555,
      "step": 53648
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.6807821989059448,
      "learning_rate": 0.0005232489252879846,
      "loss": 2.0962,
      "step": 53649
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7065984606742859,
      "learning_rate": 0.0005232461927799555,
      "loss": 2.2668,
      "step": 53650
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.6568809151649475,
      "learning_rate": 0.0005232434602304208,
      "loss": 2.3199,
      "step": 53651
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.6788375973701477,
      "learning_rate": 0.0005232407276393807,
      "loss": 2.0131,
      "step": 53652
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7196096181869507,
      "learning_rate": 0.0005232379950068362,
      "loss": 2.1765,
      "step": 53653
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7679676413536072,
      "learning_rate": 0.0005232352623327874,
      "loss": 2.2172,
      "step": 53654
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.8026814460754395,
      "learning_rate": 0.0005232325296172349,
      "loss": 2.2852,
      "step": 53655
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.813676655292511,
      "learning_rate": 0.0005232297968601794,
      "loss": 2.274,
      "step": 53656
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7017669081687927,
      "learning_rate": 0.0005232270640616212,
      "loss": 2.0345,
      "step": 53657
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.6799237728118896,
      "learning_rate": 0.0005232243312215609,
      "loss": 2.1781,
      "step": 53658
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.725952684879303,
      "learning_rate": 0.0005232215983399989,
      "loss": 2.1132,
      "step": 53659
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7488835453987122,
      "learning_rate": 0.0005232188654169359,
      "loss": 2.5093,
      "step": 53660
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7212927937507629,
      "learning_rate": 0.0005232161324523723,
      "loss": 2.2358,
      "step": 53661
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7189372181892395,
      "learning_rate": 0.0005232133994463085,
      "loss": 2.2247,
      "step": 53662
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.881743848323822,
      "learning_rate": 0.0005232106663987452,
      "loss": 2.1944,
      "step": 53663
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7103175520896912,
      "learning_rate": 0.0005232079333096828,
      "loss": 2.1685,
      "step": 53664
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7560240626335144,
      "learning_rate": 0.0005232052001791218,
      "loss": 2.2009,
      "step": 53665
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.6577341556549072,
      "learning_rate": 0.0005232024670070628,
      "loss": 2.1801,
      "step": 53666
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7020641565322876,
      "learning_rate": 0.0005231997337935061,
      "loss": 2.2821,
      "step": 53667
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7685359120368958,
      "learning_rate": 0.0005231970005384525,
      "loss": 2.2285,
      "step": 53668
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7087677717208862,
      "learning_rate": 0.0005231942672419024,
      "loss": 2.3345,
      "step": 53669
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.8342175483703613,
      "learning_rate": 0.0005231915339038562,
      "loss": 2.4586,
      "step": 53670
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.771966814994812,
      "learning_rate": 0.0005231888005243145,
      "loss": 2.1831,
      "step": 53671
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7218235731124878,
      "learning_rate": 0.0005231860671032778,
      "loss": 2.0603,
      "step": 53672
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.736731767654419,
      "learning_rate": 0.0005231833336407465,
      "loss": 2.1778,
      "step": 53673
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.774179220199585,
      "learning_rate": 0.0005231806001367213,
      "loss": 2.0805,
      "step": 53674
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7452197670936584,
      "learning_rate": 0.0005231778665912027,
      "loss": 2.2636,
      "step": 53675
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.780580997467041,
      "learning_rate": 0.000523175133004191,
      "loss": 2.3125,
      "step": 53676
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.6883906126022339,
      "learning_rate": 0.0005231723993756867,
      "loss": 2.2397,
      "step": 53677
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.6538023948669434,
      "learning_rate": 0.0005231696657056906,
      "loss": 2.3148,
      "step": 53678
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.720703661441803,
      "learning_rate": 0.0005231669319942031,
      "loss": 2.3506,
      "step": 53679
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.6564393639564514,
      "learning_rate": 0.0005231641982412246,
      "loss": 2.1599,
      "step": 53680
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.6918699741363525,
      "learning_rate": 0.0005231614644467556,
      "loss": 2.2126,
      "step": 53681
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.6752535104751587,
      "learning_rate": 0.0005231587306107968,
      "loss": 2.3039,
      "step": 53682
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.752040445804596,
      "learning_rate": 0.0005231559967333485,
      "loss": 2.2149,
      "step": 53683
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7694702744483948,
      "learning_rate": 0.0005231532628144112,
      "loss": 2.3829,
      "step": 53684
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.6752876043319702,
      "learning_rate": 0.0005231505288539857,
      "loss": 2.127,
      "step": 53685
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7208533883094788,
      "learning_rate": 0.0005231477948520721,
      "loss": 2.2253,
      "step": 53686
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7480719685554504,
      "learning_rate": 0.0005231450608086712,
      "loss": 2.1156,
      "step": 53687
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7664262056350708,
      "learning_rate": 0.0005231423267237834,
      "loss": 2.0729,
      "step": 53688
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7034708857536316,
      "learning_rate": 0.0005231395925974093,
      "loss": 2.3065,
      "step": 53689
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7954971194267273,
      "learning_rate": 0.0005231368584295493,
      "loss": 2.3006,
      "step": 53690
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7091979384422302,
      "learning_rate": 0.000523134124220204,
      "loss": 2.2281,
      "step": 53691
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7969685196876526,
      "learning_rate": 0.0005231313899693737,
      "loss": 2.1906,
      "step": 53692
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7480906248092651,
      "learning_rate": 0.0005231286556770592,
      "loss": 2.1048,
      "step": 53693
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7323541641235352,
      "learning_rate": 0.0005231259213432608,
      "loss": 2.2155,
      "step": 53694
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7320178747177124,
      "learning_rate": 0.0005231231869679792,
      "loss": 2.1355,
      "step": 53695
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7378529906272888,
      "learning_rate": 0.0005231204525512146,
      "loss": 2.2141,
      "step": 53696
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7799782156944275,
      "learning_rate": 0.0005231177180929677,
      "loss": 2.2111,
      "step": 53697
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.6827625036239624,
      "learning_rate": 0.0005231149835932392,
      "loss": 2.1603,
      "step": 53698
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.6815869808197021,
      "learning_rate": 0.0005231122490520293,
      "loss": 2.0329,
      "step": 53699
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7666332721710205,
      "learning_rate": 0.0005231095144693386,
      "loss": 2.2044,
      "step": 53700
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7255550622940063,
      "learning_rate": 0.0005231067798451677,
      "loss": 2.3076,
      "step": 53701
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7321793437004089,
      "learning_rate": 0.000523104045179517,
      "loss": 2.1683,
      "step": 53702
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7200782895088196,
      "learning_rate": 0.000523101310472387,
      "loss": 2.3197,
      "step": 53703
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7775757312774658,
      "learning_rate": 0.0005230985757237783,
      "loss": 2.1279,
      "step": 53704
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.6994748115539551,
      "learning_rate": 0.0005230958409336914,
      "loss": 2.2858,
      "step": 53705
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.71819007396698,
      "learning_rate": 0.0005230931061021266,
      "loss": 2.1336,
      "step": 53706
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7126657366752625,
      "learning_rate": 0.0005230903712290847,
      "loss": 2.3114,
      "step": 53707
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7401052117347717,
      "learning_rate": 0.0005230876363145661,
      "loss": 2.2391,
      "step": 53708
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.856486976146698,
      "learning_rate": 0.0005230849013585714,
      "loss": 2.0756,
      "step": 53709
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7569789886474609,
      "learning_rate": 0.0005230821663611009,
      "loss": 2.1644,
      "step": 53710
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.8083128333091736,
      "learning_rate": 0.0005230794313221553,
      "loss": 2.3178,
      "step": 53711
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7198279500007629,
      "learning_rate": 0.0005230766962417348,
      "loss": 2.1872,
      "step": 53712
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7212551236152649,
      "learning_rate": 0.0005230739611198403,
      "loss": 2.5329,
      "step": 53713
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7694882154464722,
      "learning_rate": 0.0005230712259564721,
      "loss": 2.2792,
      "step": 53714
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.8618872761726379,
      "learning_rate": 0.0005230684907516308,
      "loss": 2.047,
      "step": 53715
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.6876933574676514,
      "learning_rate": 0.0005230657555053168,
      "loss": 2.1842,
      "step": 53716
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7839429378509521,
      "learning_rate": 0.0005230630202175306,
      "loss": 2.0822,
      "step": 53717
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.9562323093414307,
      "learning_rate": 0.0005230602848882729,
      "loss": 2.2876,
      "step": 53718
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.6633107662200928,
      "learning_rate": 0.000523057549517544,
      "loss": 2.0498,
      "step": 53719
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7144308686256409,
      "learning_rate": 0.0005230548141053446,
      "loss": 2.1681,
      "step": 53720
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.739168107509613,
      "learning_rate": 0.000523052078651675,
      "loss": 2.4063,
      "step": 53721
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.794111430644989,
      "learning_rate": 0.0005230493431565359,
      "loss": 2.1701,
      "step": 53722
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7399880886077881,
      "learning_rate": 0.0005230466076199277,
      "loss": 2.18,
      "step": 53723
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.789028525352478,
      "learning_rate": 0.0005230438720418508,
      "loss": 2.3722,
      "step": 53724
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.6894370913505554,
      "learning_rate": 0.0005230411364223059,
      "loss": 2.3298,
      "step": 53725
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.6695870161056519,
      "learning_rate": 0.0005230384007612935,
      "loss": 2.2788,
      "step": 53726
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.6272968053817749,
      "learning_rate": 0.0005230356650588139,
      "loss": 2.0692,
      "step": 53727
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.6714510321617126,
      "learning_rate": 0.000523032929314868,
      "loss": 2.168,
      "step": 53728
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.8151652216911316,
      "learning_rate": 0.0005230301935294559,
      "loss": 2.3959,
      "step": 53729
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.8083317875862122,
      "learning_rate": 0.0005230274577025783,
      "loss": 2.2509,
      "step": 53730
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7490395307540894,
      "learning_rate": 0.0005230247218342357,
      "loss": 2.2546,
      "step": 53731
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7272126078605652,
      "learning_rate": 0.0005230219859244285,
      "loss": 2.3106,
      "step": 53732
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7563396096229553,
      "learning_rate": 0.0005230192499731575,
      "loss": 2.2338,
      "step": 53733
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.6982638835906982,
      "learning_rate": 0.000523016513980423,
      "loss": 2.2882,
      "step": 53734
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.737324595451355,
      "learning_rate": 0.0005230137779462252,
      "loss": 2.103,
      "step": 53735
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.6312668323516846,
      "learning_rate": 0.0005230110418705653,
      "loss": 2.3357,
      "step": 53736
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7087686061859131,
      "learning_rate": 0.0005230083057534432,
      "loss": 2.0308,
      "step": 53737
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7281817197799683,
      "learning_rate": 0.0005230055695948598,
      "loss": 2.2203,
      "step": 53738
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7505258321762085,
      "learning_rate": 0.0005230028333948153,
      "loss": 2.1319,
      "step": 53739
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.6896322965621948,
      "learning_rate": 0.0005230000971533105,
      "loss": 2.4004,
      "step": 53740
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.6525969505310059,
      "learning_rate": 0.0005229973608703457,
      "loss": 2.0591,
      "step": 53741
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.8845634460449219,
      "learning_rate": 0.0005229946245459215,
      "loss": 2.2785,
      "step": 53742
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7128080129623413,
      "learning_rate": 0.0005229918881800385,
      "loss": 2.0933,
      "step": 53743
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7132266163825989,
      "learning_rate": 0.000522989151772697,
      "loss": 2.1726,
      "step": 53744
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7405337691307068,
      "learning_rate": 0.0005229864153238976,
      "loss": 2.1881,
      "step": 53745
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7204283475875854,
      "learning_rate": 0.0005229836788336409,
      "loss": 2.1059,
      "step": 53746
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7505826950073242,
      "learning_rate": 0.0005229809423019272,
      "loss": 2.2576,
      "step": 53747
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7149648070335388,
      "learning_rate": 0.0005229782057287572,
      "loss": 2.1528,
      "step": 53748
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.6856601238250732,
      "learning_rate": 0.0005229754691141313,
      "loss": 2.1839,
      "step": 53749
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7709128856658936,
      "learning_rate": 0.0005229727324580501,
      "loss": 2.418,
      "step": 53750
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7200853228569031,
      "learning_rate": 0.0005229699957605141,
      "loss": 2.2106,
      "step": 53751
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.6422001123428345,
      "learning_rate": 0.0005229672590215236,
      "loss": 2.1097,
      "step": 53752
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7474754452705383,
      "learning_rate": 0.0005229645222410794,
      "loss": 2.0601,
      "step": 53753
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7602471709251404,
      "learning_rate": 0.0005229617854191819,
      "loss": 2.113,
      "step": 53754
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7439098954200745,
      "learning_rate": 0.0005229590485558315,
      "loss": 2.3801,
      "step": 53755
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.68113774061203,
      "learning_rate": 0.0005229563116510289,
      "loss": 2.1189,
      "step": 53756
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.6358055472373962,
      "learning_rate": 0.0005229535747047745,
      "loss": 1.9156,
      "step": 53757
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7398125529289246,
      "learning_rate": 0.0005229508377170688,
      "loss": 2.0244,
      "step": 53758
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.6784760355949402,
      "learning_rate": 0.0005229481006879123,
      "loss": 2.2344,
      "step": 53759
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.6413047313690186,
      "learning_rate": 0.0005229453636173056,
      "loss": 2.1622,
      "step": 53760
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.725984513759613,
      "learning_rate": 0.000522942626505249,
      "loss": 2.1557,
      "step": 53761
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7505736947059631,
      "learning_rate": 0.0005229398893517433,
      "loss": 2.3188,
      "step": 53762
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.6890159845352173,
      "learning_rate": 0.0005229371521567889,
      "loss": 2.1501,
      "step": 53763
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7430195212364197,
      "learning_rate": 0.0005229344149203862,
      "loss": 2.1357,
      "step": 53764
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7177950143814087,
      "learning_rate": 0.0005229316776425359,
      "loss": 2.3658,
      "step": 53765
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.6759741902351379,
      "learning_rate": 0.0005229289403232382,
      "loss": 2.1584,
      "step": 53766
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.6842458844184875,
      "learning_rate": 0.000522926202962494,
      "loss": 2.1911,
      "step": 53767
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7229107618331909,
      "learning_rate": 0.0005229234655603034,
      "loss": 2.1089,
      "step": 53768
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7283863425254822,
      "learning_rate": 0.0005229207281166673,
      "loss": 2.365,
      "step": 53769
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.8338948488235474,
      "learning_rate": 0.0005229179906315859,
      "loss": 2.3413,
      "step": 53770
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.6837784051895142,
      "learning_rate": 0.00052291525310506,
      "loss": 2.1121,
      "step": 53771
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.6826441287994385,
      "learning_rate": 0.0005229125155370898,
      "loss": 2.1434,
      "step": 53772
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7089798450469971,
      "learning_rate": 0.000522909777927676,
      "loss": 2.2572,
      "step": 53773
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7736393809318542,
      "learning_rate": 0.0005229070402768191,
      "loss": 2.2497,
      "step": 53774
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7283148169517517,
      "learning_rate": 0.0005229043025845195,
      "loss": 2.2877,
      "step": 53775
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7058507800102234,
      "learning_rate": 0.0005229015648507778,
      "loss": 2.2165,
      "step": 53776
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.8113280534744263,
      "learning_rate": 0.0005228988270755945,
      "loss": 2.5171,
      "step": 53777
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.849687397480011,
      "learning_rate": 0.0005228960892589701,
      "loss": 2.189,
      "step": 53778
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.8048753142356873,
      "learning_rate": 0.0005228933514009052,
      "loss": 2.2301,
      "step": 53779
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7230825424194336,
      "learning_rate": 0.0005228906135014001,
      "loss": 2.1349,
      "step": 53780
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7522440552711487,
      "learning_rate": 0.0005228878755604554,
      "loss": 2.2687,
      "step": 53781
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7274650931358337,
      "learning_rate": 0.0005228851375780717,
      "loss": 2.0934,
      "step": 53782
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7520765066146851,
      "learning_rate": 0.0005228823995542495,
      "loss": 2.2846,
      "step": 53783
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7396059632301331,
      "learning_rate": 0.0005228796614889892,
      "loss": 2.1244,
      "step": 53784
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7292670011520386,
      "learning_rate": 0.0005228769233822913,
      "loss": 2.1492,
      "step": 53785
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.8732753992080688,
      "learning_rate": 0.0005228741852341566,
      "loss": 2.207,
      "step": 53786
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7296327352523804,
      "learning_rate": 0.0005228714470445852,
      "loss": 2.282,
      "step": 53787
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7000917792320251,
      "learning_rate": 0.0005228687088135778,
      "loss": 2.3924,
      "step": 53788
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.8362575769424438,
      "learning_rate": 0.0005228659705411348,
      "loss": 2.2681,
      "step": 53789
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.6656442284584045,
      "learning_rate": 0.000522863232227257,
      "loss": 2.2021,
      "step": 53790
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.6665861010551453,
      "learning_rate": 0.0005228604938719446,
      "loss": 2.1044,
      "step": 53791
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7302647829055786,
      "learning_rate": 0.0005228577554751983,
      "loss": 2.1465,
      "step": 53792
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7381079792976379,
      "learning_rate": 0.0005228550170370185,
      "loss": 2.3594,
      "step": 53793
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.8859325647354126,
      "learning_rate": 0.0005228522785574058,
      "loss": 2.4943,
      "step": 53794
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7481381297111511,
      "learning_rate": 0.0005228495400363605,
      "loss": 2.4258,
      "step": 53795
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7171328067779541,
      "learning_rate": 0.0005228468014738835,
      "loss": 2.3936,
      "step": 53796
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.9207150936126709,
      "learning_rate": 0.0005228440628699749,
      "loss": 2.2114,
      "step": 53797
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.8564466834068298,
      "learning_rate": 0.0005228413242246354,
      "loss": 2.2042,
      "step": 53798
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.6736257672309875,
      "learning_rate": 0.0005228385855378656,
      "loss": 2.057,
      "step": 53799
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.024064064025879,
      "learning_rate": 0.0005228358468096657,
      "loss": 2.0297,
      "step": 53800
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.8243236541748047,
      "learning_rate": 0.0005228331080400366,
      "loss": 2.2264,
      "step": 53801
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.707580029964447,
      "learning_rate": 0.0005228303692289785,
      "loss": 2.1473,
      "step": 53802
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.8521878719329834,
      "learning_rate": 0.000522827630376492,
      "loss": 2.4758,
      "step": 53803
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.8718506693840027,
      "learning_rate": 0.0005228248914825778,
      "loss": 2.3061,
      "step": 53804
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7005062103271484,
      "learning_rate": 0.0005228221525472362,
      "loss": 2.1053,
      "step": 53805
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7514469027519226,
      "learning_rate": 0.0005228194135704677,
      "loss": 2.1927,
      "step": 53806
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.8705446720123291,
      "learning_rate": 0.0005228166745522728,
      "loss": 2.3865,
      "step": 53807
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.6946460604667664,
      "learning_rate": 0.0005228139354926522,
      "loss": 2.1512,
      "step": 53808
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.8089213967323303,
      "learning_rate": 0.0005228111963916063,
      "loss": 2.2187,
      "step": 53809
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.9181716442108154,
      "learning_rate": 0.0005228084572491355,
      "loss": 2.2238,
      "step": 53810
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7717932462692261,
      "learning_rate": 0.0005228057180652404,
      "loss": 2.1639,
      "step": 53811
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7028100490570068,
      "learning_rate": 0.0005228029788399216,
      "loss": 2.2507,
      "step": 53812
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.8969067335128784,
      "learning_rate": 0.0005228002395731795,
      "loss": 2.214,
      "step": 53813
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7975372672080994,
      "learning_rate": 0.0005227975002650145,
      "loss": 2.4402,
      "step": 53814
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.6761237978935242,
      "learning_rate": 0.0005227947609154274,
      "loss": 2.192,
      "step": 53815
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7060595154762268,
      "learning_rate": 0.0005227920215244184,
      "loss": 2.5101,
      "step": 53816
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7707923054695129,
      "learning_rate": 0.0005227892820919883,
      "loss": 2.1916,
      "step": 53817
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.0637098550796509,
      "learning_rate": 0.0005227865426181373,
      "loss": 2.2292,
      "step": 53818
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7349221706390381,
      "learning_rate": 0.0005227838031028663,
      "loss": 2.28,
      "step": 53819
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.8312833905220032,
      "learning_rate": 0.0005227810635461754,
      "loss": 2.3062,
      "step": 53820
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.9715821743011475,
      "learning_rate": 0.0005227783239480653,
      "loss": 2.2032,
      "step": 53821
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.8234443664550781,
      "learning_rate": 0.0005227755843085366,
      "loss": 2.3355,
      "step": 53822
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.017724871635437,
      "learning_rate": 0.0005227728446275898,
      "loss": 2.3798,
      "step": 53823
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.8227952718734741,
      "learning_rate": 0.0005227701049052251,
      "loss": 2.4804,
      "step": 53824
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.9596690535545349,
      "learning_rate": 0.0005227673651414433,
      "loss": 2.3688,
      "step": 53825
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7804048657417297,
      "learning_rate": 0.0005227646253362449,
      "loss": 2.2616,
      "step": 53826
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7685815691947937,
      "learning_rate": 0.0005227618854896304,
      "loss": 2.0403,
      "step": 53827
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7639455199241638,
      "learning_rate": 0.0005227591456016001,
      "loss": 2.2762,
      "step": 53828
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.844868540763855,
      "learning_rate": 0.0005227564056721548,
      "loss": 2.1085,
      "step": 53829
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.0093071460723877,
      "learning_rate": 0.0005227536657012948,
      "loss": 2.409,
      "step": 53830
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7448941469192505,
      "learning_rate": 0.0005227509256890208,
      "loss": 2.4646,
      "step": 53831
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.9910117983818054,
      "learning_rate": 0.0005227481856353331,
      "loss": 2.1885,
      "step": 53832
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.8587188720703125,
      "learning_rate": 0.0005227454455402323,
      "loss": 2.2603,
      "step": 53833
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7288858890533447,
      "learning_rate": 0.000522742705403719,
      "loss": 2.0789,
      "step": 53834
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.759109616279602,
      "learning_rate": 0.0005227399652257935,
      "loss": 2.2758,
      "step": 53835
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.8964167833328247,
      "learning_rate": 0.0005227372250064566,
      "loss": 2.2773,
      "step": 53836
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7313401699066162,
      "learning_rate": 0.0005227344847457085,
      "loss": 2.2871,
      "step": 53837
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.6578068137168884,
      "learning_rate": 0.0005227317444435499,
      "loss": 2.2026,
      "step": 53838
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7480047345161438,
      "learning_rate": 0.0005227290040999813,
      "loss": 2.3144,
      "step": 53839
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.8400100469589233,
      "learning_rate": 0.0005227262637150031,
      "loss": 2.1389,
      "step": 53840
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.8283032178878784,
      "learning_rate": 0.000522723523288616,
      "loss": 2.3521,
      "step": 53841
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7536356449127197,
      "learning_rate": 0.0005227207828208203,
      "loss": 2.2471,
      "step": 53842
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7168075442314148,
      "learning_rate": 0.0005227180423116166,
      "loss": 2.2629,
      "step": 53843
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.8697391748428345,
      "learning_rate": 0.0005227153017610054,
      "loss": 2.2139,
      "step": 53844
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7456724047660828,
      "learning_rate": 0.0005227125611689873,
      "loss": 2.0368,
      "step": 53845
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.8672590851783752,
      "learning_rate": 0.0005227098205355627,
      "loss": 2.1859,
      "step": 53846
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7480844855308533,
      "learning_rate": 0.0005227070798607321,
      "loss": 2.268,
      "step": 53847
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7155134677886963,
      "learning_rate": 0.0005227043391444962,
      "loss": 2.4517,
      "step": 53848
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7745475172996521,
      "learning_rate": 0.0005227015983868552,
      "loss": 2.4725,
      "step": 53849
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7089173793792725,
      "learning_rate": 0.0005226988575878097,
      "loss": 2.3892,
      "step": 53850
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.6753892302513123,
      "learning_rate": 0.0005226961167473605,
      "loss": 2.0973,
      "step": 53851
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7983111143112183,
      "learning_rate": 0.0005226933758655077,
      "loss": 2.2538,
      "step": 53852
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.9977187514305115,
      "learning_rate": 0.0005226906349422521,
      "loss": 2.3908,
      "step": 53853
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.6662846207618713,
      "learning_rate": 0.0005226878939775941,
      "loss": 2.4711,
      "step": 53854
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7126690149307251,
      "learning_rate": 0.0005226851529715343,
      "loss": 2.2666,
      "step": 53855
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7374589443206787,
      "learning_rate": 0.000522682411924073,
      "loss": 2.2916,
      "step": 53856
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.8219023942947388,
      "learning_rate": 0.0005226796708352108,
      "loss": 2.2302,
      "step": 53857
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7254002094268799,
      "learning_rate": 0.0005226769297049482,
      "loss": 2.299,
      "step": 53858
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.820833683013916,
      "learning_rate": 0.000522674188533286,
      "loss": 2.3756,
      "step": 53859
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.668755054473877,
      "learning_rate": 0.0005226714473202243,
      "loss": 2.0553,
      "step": 53860
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.6828051805496216,
      "learning_rate": 0.0005226687060657638,
      "loss": 2.2979,
      "step": 53861
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.6920807361602783,
      "learning_rate": 0.000522665964769905,
      "loss": 2.4284,
      "step": 53862
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.6508916616439819,
      "learning_rate": 0.0005226632234326484,
      "loss": 2.2639,
      "step": 53863
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.8214648365974426,
      "learning_rate": 0.0005226604820539945,
      "loss": 2.2449,
      "step": 53864
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7291935682296753,
      "learning_rate": 0.0005226577406339438,
      "loss": 2.5037,
      "step": 53865
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.8391199707984924,
      "learning_rate": 0.0005226549991724968,
      "loss": 2.1832,
      "step": 53866
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.8593655824661255,
      "learning_rate": 0.000522652257669654,
      "loss": 2.2214,
      "step": 53867
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7372633218765259,
      "learning_rate": 0.000522649516125416,
      "loss": 2.1927,
      "step": 53868
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.850127637386322,
      "learning_rate": 0.0005226467745397832,
      "loss": 2.2359,
      "step": 53869
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7718978524208069,
      "learning_rate": 0.0005226440329127563,
      "loss": 2.4115,
      "step": 53870
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7082802653312683,
      "learning_rate": 0.0005226412912443355,
      "loss": 2.1087,
      "step": 53871
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7433556914329529,
      "learning_rate": 0.0005226385495345216,
      "loss": 2.2357,
      "step": 53872
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7083479166030884,
      "learning_rate": 0.000522635807783315,
      "loss": 2.2237,
      "step": 53873
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7568978071212769,
      "learning_rate": 0.0005226330659907161,
      "loss": 2.4763,
      "step": 53874
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.8548679947853088,
      "learning_rate": 0.0005226303241567256,
      "loss": 2.1922,
      "step": 53875
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7433656454086304,
      "learning_rate": 0.0005226275822813439,
      "loss": 2.3497,
      "step": 53876
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7673223614692688,
      "learning_rate": 0.0005226248403645715,
      "loss": 2.283,
      "step": 53877
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7502595782279968,
      "learning_rate": 0.000522622098406409,
      "loss": 2.3613,
      "step": 53878
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.731025218963623,
      "learning_rate": 0.0005226193564068568,
      "loss": 2.2479,
      "step": 53879
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.6485984325408936,
      "learning_rate": 0.0005226166143659154,
      "loss": 2.232,
      "step": 53880
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7177149653434753,
      "learning_rate": 0.0005226138722835854,
      "loss": 2.2074,
      "step": 53881
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7698052525520325,
      "learning_rate": 0.0005226111301598673,
      "loss": 2.1647,
      "step": 53882
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.6951949596405029,
      "learning_rate": 0.0005226083879947617,
      "loss": 2.1987,
      "step": 53883
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.6997413635253906,
      "learning_rate": 0.0005226056457882689,
      "loss": 2.1492,
      "step": 53884
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7390608191490173,
      "learning_rate": 0.0005226029035403896,
      "loss": 2.2333,
      "step": 53885
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.8140829801559448,
      "learning_rate": 0.000522600161251124,
      "loss": 2.4783,
      "step": 53886
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7114002108573914,
      "learning_rate": 0.0005225974189204729,
      "loss": 2.1144,
      "step": 53887
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7283802628517151,
      "learning_rate": 0.0005225946765484369,
      "loss": 2.234,
      "step": 53888
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.6741172671318054,
      "learning_rate": 0.0005225919341350163,
      "loss": 2.2345,
      "step": 53889
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7075223922729492,
      "learning_rate": 0.0005225891916802115,
      "loss": 2.2484,
      "step": 53890
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.6478592753410339,
      "learning_rate": 0.0005225864491840233,
      "loss": 2.0515,
      "step": 53891
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.742164671421051,
      "learning_rate": 0.000522583706646452,
      "loss": 2.2884,
      "step": 53892
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.6872116923332214,
      "learning_rate": 0.0005225809640674983,
      "loss": 2.084,
      "step": 53893
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7382367253303528,
      "learning_rate": 0.0005225782214471625,
      "loss": 2.2903,
      "step": 53894
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.6856489181518555,
      "learning_rate": 0.0005225754787854452,
      "loss": 2.25,
      "step": 53895
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7099543213844299,
      "learning_rate": 0.000522572736082347,
      "loss": 2.0934,
      "step": 53896
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.6518270373344421,
      "learning_rate": 0.0005225699933378682,
      "loss": 2.1725,
      "step": 53897
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7460765838623047,
      "learning_rate": 0.0005225672505520095,
      "loss": 2.1381,
      "step": 53898
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.9190383553504944,
      "learning_rate": 0.0005225645077247714,
      "loss": 2.5492,
      "step": 53899
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7184381484985352,
      "learning_rate": 0.0005225617648561544,
      "loss": 2.2337,
      "step": 53900
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.6843004822731018,
      "learning_rate": 0.0005225590219461589,
      "loss": 2.2943,
      "step": 53901
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7793099284172058,
      "learning_rate": 0.0005225562789947854,
      "loss": 2.0615,
      "step": 53902
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7213342189788818,
      "learning_rate": 0.0005225535360020346,
      "loss": 2.4017,
      "step": 53903
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.794211745262146,
      "learning_rate": 0.0005225507929679068,
      "loss": 2.3423,
      "step": 53904
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7898224592208862,
      "learning_rate": 0.0005225480498924027,
      "loss": 2.226,
      "step": 53905
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.8267809748649597,
      "learning_rate": 0.0005225453067755226,
      "loss": 2.2248,
      "step": 53906
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.738163948059082,
      "learning_rate": 0.0005225425636172672,
      "loss": 2.2998,
      "step": 53907
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.8677815794944763,
      "learning_rate": 0.000522539820417637,
      "loss": 2.1342,
      "step": 53908
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7996432185173035,
      "learning_rate": 0.0005225370771766324,
      "loss": 2.0673,
      "step": 53909
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7442208528518677,
      "learning_rate": 0.0005225343338942539,
      "loss": 2.1268,
      "step": 53910
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.6687600016593933,
      "learning_rate": 0.000522531590570502,
      "loss": 2.0874,
      "step": 53911
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7552194595336914,
      "learning_rate": 0.0005225288472053774,
      "loss": 2.2271,
      "step": 53912
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7329967021942139,
      "learning_rate": 0.0005225261037988806,
      "loss": 2.0497,
      "step": 53913
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.892126202583313,
      "learning_rate": 0.0005225233603510118,
      "loss": 2.2387,
      "step": 53914
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7239581942558289,
      "learning_rate": 0.0005225206168617718,
      "loss": 2.36,
      "step": 53915
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7243675589561462,
      "learning_rate": 0.000522517873331161,
      "loss": 2.5577,
      "step": 53916
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7169361710548401,
      "learning_rate": 0.0005225151297591798,
      "loss": 2.1362,
      "step": 53917
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7308349609375,
      "learning_rate": 0.000522512386145829,
      "loss": 2.1569,
      "step": 53918
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7531789541244507,
      "learning_rate": 0.0005225096424911089,
      "loss": 2.5136,
      "step": 53919
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7790015339851379,
      "learning_rate": 0.0005225068987950201,
      "loss": 2.1709,
      "step": 53920
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7067820429801941,
      "learning_rate": 0.000522504155057563,
      "loss": 1.9964,
      "step": 53921
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7335420250892639,
      "learning_rate": 0.0005225014112787383,
      "loss": 2.3844,
      "step": 53922
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7763164043426514,
      "learning_rate": 0.0005224986674585462,
      "loss": 2.5249,
      "step": 53923
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7105581760406494,
      "learning_rate": 0.0005224959235969875,
      "loss": 2.0485,
      "step": 53924
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7234135866165161,
      "learning_rate": 0.0005224931796940625,
      "loss": 2.1593,
      "step": 53925
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7458698153495789,
      "learning_rate": 0.000522490435749772,
      "loss": 2.2906,
      "step": 53926
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7321901917457581,
      "learning_rate": 0.0005224876917641162,
      "loss": 2.2156,
      "step": 53927
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7064032554626465,
      "learning_rate": 0.0005224849477370958,
      "loss": 2.0048,
      "step": 53928
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7108909487724304,
      "learning_rate": 0.0005224822036687112,
      "loss": 2.3952,
      "step": 53929
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7702979445457458,
      "learning_rate": 0.000522479459558963,
      "loss": 2.211,
      "step": 53930
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7443571090698242,
      "learning_rate": 0.0005224767154078517,
      "loss": 2.48,
      "step": 53931
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.8673529624938965,
      "learning_rate": 0.0005224739712153776,
      "loss": 2.1363,
      "step": 53932
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.6885393261909485,
      "learning_rate": 0.0005224712269815416,
      "loss": 2.132,
      "step": 53933
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7064104080200195,
      "learning_rate": 0.0005224684827063439,
      "loss": 2.125,
      "step": 53934
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7454588413238525,
      "learning_rate": 0.0005224657383897851,
      "loss": 2.3292,
      "step": 53935
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.817186176776886,
      "learning_rate": 0.0005224629940318658,
      "loss": 2.2447,
      "step": 53936
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7366847395896912,
      "learning_rate": 0.0005224602496325864,
      "loss": 2.0987,
      "step": 53937
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.6739513278007507,
      "learning_rate": 0.0005224575051919475,
      "loss": 2.4999,
      "step": 53938
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.6732490062713623,
      "learning_rate": 0.0005224547607099493,
      "loss": 2.1334,
      "step": 53939
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.753677248954773,
      "learning_rate": 0.0005224520161865928,
      "loss": 2.1155,
      "step": 53940
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.6934820413589478,
      "learning_rate": 0.0005224492716218782,
      "loss": 2.4635,
      "step": 53941
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7100560665130615,
      "learning_rate": 0.0005224465270158061,
      "loss": 2.3121,
      "step": 53942
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.8077433109283447,
      "learning_rate": 0.000522443782368377,
      "loss": 2.1362,
      "step": 53943
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7546438574790955,
      "learning_rate": 0.0005224410376795913,
      "loss": 2.2374,
      "step": 53944
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.6682920455932617,
      "learning_rate": 0.0005224382929494497,
      "loss": 2.2503,
      "step": 53945
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7171115875244141,
      "learning_rate": 0.0005224355481779527,
      "loss": 2.2077,
      "step": 53946
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.8265445828437805,
      "learning_rate": 0.0005224328033651007,
      "loss": 2.4069,
      "step": 53947
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7036904096603394,
      "learning_rate": 0.0005224300585108941,
      "loss": 2.319,
      "step": 53948
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7587193250656128,
      "learning_rate": 0.0005224273136153337,
      "loss": 2.0902,
      "step": 53949
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.898566484451294,
      "learning_rate": 0.0005224245686784197,
      "loss": 2.1394,
      "step": 53950
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7615629434585571,
      "learning_rate": 0.0005224218237001529,
      "loss": 2.3501,
      "step": 53951
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.6778848171234131,
      "learning_rate": 0.0005224190786805338,
      "loss": 1.9969,
      "step": 53952
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7902858853340149,
      "learning_rate": 0.0005224163336195626,
      "loss": 2.3039,
      "step": 53953
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.8546710014343262,
      "learning_rate": 0.00052241358851724,
      "loss": 2.3788,
      "step": 53954
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7747666835784912,
      "learning_rate": 0.0005224108433735666,
      "loss": 2.2686,
      "step": 53955
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.8580658435821533,
      "learning_rate": 0.0005224080981885428,
      "loss": 2.2234,
      "step": 53956
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.9879949688911438,
      "learning_rate": 0.0005224053529621691,
      "loss": 2.3996,
      "step": 53957
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7614700198173523,
      "learning_rate": 0.0005224026076944461,
      "loss": 2.1351,
      "step": 53958
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7554782629013062,
      "learning_rate": 0.0005223998623853743,
      "loss": 2.4019,
      "step": 53959
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7833408117294312,
      "learning_rate": 0.0005223971170349541,
      "loss": 2.2875,
      "step": 53960
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.8673781752586365,
      "learning_rate": 0.0005223943716431862,
      "loss": 2.3475,
      "step": 53961
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.757599413394928,
      "learning_rate": 0.0005223916262100708,
      "loss": 2.2862,
      "step": 53962
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7058622241020203,
      "learning_rate": 0.0005223888807356087,
      "loss": 2.1789,
      "step": 53963
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.8433109521865845,
      "learning_rate": 0.0005223861352198003,
      "loss": 2.2264,
      "step": 53964
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.8107309937477112,
      "learning_rate": 0.000522383389662646,
      "loss": 2.2556,
      "step": 53965
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.8333513140678406,
      "learning_rate": 0.0005223806440641465,
      "loss": 2.421,
      "step": 53966
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.9252126216888428,
      "learning_rate": 0.0005223778984243024,
      "loss": 2.1066,
      "step": 53967
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.8498703837394714,
      "learning_rate": 0.0005223751527431139,
      "loss": 2.3554,
      "step": 53968
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.706487774848938,
      "learning_rate": 0.0005223724070205817,
      "loss": 2.22,
      "step": 53969
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.8636574149131775,
      "learning_rate": 0.0005223696612567063,
      "loss": 1.9473,
      "step": 53970
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.8055922389030457,
      "learning_rate": 0.0005223669154514881,
      "loss": 2.2558,
      "step": 53971
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.6911839246749878,
      "learning_rate": 0.0005223641696049277,
      "loss": 2.1572,
      "step": 53972
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.9122378826141357,
      "learning_rate": 0.0005223614237170257,
      "loss": 2.2478,
      "step": 53973
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7692090272903442,
      "learning_rate": 0.0005223586777877824,
      "loss": 2.1912,
      "step": 53974
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7005496025085449,
      "learning_rate": 0.0005223559318171986,
      "loss": 2.4095,
      "step": 53975
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.8387946486473083,
      "learning_rate": 0.0005223531858052744,
      "loss": 2.2325,
      "step": 53976
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7432107925415039,
      "learning_rate": 0.0005223504397520107,
      "loss": 2.4461,
      "step": 53977
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.6685603857040405,
      "learning_rate": 0.0005223476936574078,
      "loss": 2.2377,
      "step": 53978
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.751908540725708,
      "learning_rate": 0.0005223449475214663,
      "loss": 2.2336,
      "step": 53979
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.6953058838844299,
      "learning_rate": 0.0005223422013441866,
      "loss": 2.1123,
      "step": 53980
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7405242323875427,
      "learning_rate": 0.0005223394551255694,
      "loss": 2.067,
      "step": 53981
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7042535543441772,
      "learning_rate": 0.0005223367088656151,
      "loss": 2.5179,
      "step": 53982
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7712662220001221,
      "learning_rate": 0.0005223339625643241,
      "loss": 2.3281,
      "step": 53983
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.760208249092102,
      "learning_rate": 0.000522331216221697,
      "loss": 2.4673,
      "step": 53984
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7348281741142273,
      "learning_rate": 0.0005223284698377344,
      "loss": 2.0694,
      "step": 53985
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7646652460098267,
      "learning_rate": 0.0005223257234124369,
      "loss": 2.1923,
      "step": 53986
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7474448084831238,
      "learning_rate": 0.0005223229769458046,
      "loss": 2.4305,
      "step": 53987
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.713319718837738,
      "learning_rate": 0.0005223202304378384,
      "loss": 2.2004,
      "step": 53988
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.6710377931594849,
      "learning_rate": 0.0005223174838885386,
      "loss": 2.2208,
      "step": 53989
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.720123291015625,
      "learning_rate": 0.0005223147372979058,
      "loss": 2.1872,
      "step": 53990
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.8219625353813171,
      "learning_rate": 0.0005223119906659406,
      "loss": 2.4226,
      "step": 53991
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.6318252086639404,
      "learning_rate": 0.0005223092439926432,
      "loss": 2.2977,
      "step": 53992
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.8805527687072754,
      "learning_rate": 0.0005223064972780145,
      "loss": 2.297,
      "step": 53993
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.725207507610321,
      "learning_rate": 0.0005223037505220548,
      "loss": 2.0731,
      "step": 53994
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7149165272712708,
      "learning_rate": 0.0005223010037247645,
      "loss": 2.0863,
      "step": 53995
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.8040108680725098,
      "learning_rate": 0.0005222982568861443,
      "loss": 2.4144,
      "step": 53996
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7140413522720337,
      "learning_rate": 0.0005222955100061948,
      "loss": 2.1949,
      "step": 53997
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7310904264450073,
      "learning_rate": 0.0005222927630849163,
      "loss": 2.4239,
      "step": 53998
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7303917407989502,
      "learning_rate": 0.0005222900161223092,
      "loss": 2.1637,
      "step": 53999
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.727862536907196,
      "learning_rate": 0.0005222872691183745,
      "loss": 2.3405,
      "step": 54000
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7411434054374695,
      "learning_rate": 0.0005222845220731122,
      "loss": 2.3802,
      "step": 54001
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7275720238685608,
      "learning_rate": 0.000522281774986523,
      "loss": 2.3479,
      "step": 54002
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7813695669174194,
      "learning_rate": 0.0005222790278586074,
      "loss": 2.1663,
      "step": 54003
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7969932556152344,
      "learning_rate": 0.0005222762806893661,
      "loss": 2.2628,
      "step": 54004
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7335529923439026,
      "learning_rate": 0.0005222735334787993,
      "loss": 2.0558,
      "step": 54005
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7082036733627319,
      "learning_rate": 0.0005222707862269077,
      "loss": 2.297,
      "step": 54006
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7111340165138245,
      "learning_rate": 0.0005222680389336918,
      "loss": 2.1231,
      "step": 54007
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.6969478130340576,
      "learning_rate": 0.000522265291599152,
      "loss": 2.4118,
      "step": 54008
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7463346719741821,
      "learning_rate": 0.000522262544223289,
      "loss": 2.2245,
      "step": 54009
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7314621210098267,
      "learning_rate": 0.000522259796806103,
      "loss": 2.1183,
      "step": 54010
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.675264835357666,
      "learning_rate": 0.000522257049347595,
      "loss": 2.2012,
      "step": 54011
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.8191226124763489,
      "learning_rate": 0.000522254301847765,
      "loss": 2.2483,
      "step": 54012
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.6805740594863892,
      "learning_rate": 0.0005222515543066138,
      "loss": 2.076,
      "step": 54013
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7941483855247498,
      "learning_rate": 0.0005222488067241417,
      "loss": 2.3741,
      "step": 54014
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.8101470470428467,
      "learning_rate": 0.0005222460591003495,
      "loss": 2.1736,
      "step": 54015
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7654387950897217,
      "learning_rate": 0.0005222433114352374,
      "loss": 2.3708,
      "step": 54016
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7931587100028992,
      "learning_rate": 0.0005222405637288062,
      "loss": 2.3314,
      "step": 54017
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7650257349014282,
      "learning_rate": 0.0005222378159810564,
      "loss": 2.1483,
      "step": 54018
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.6890956163406372,
      "learning_rate": 0.0005222350681919883,
      "loss": 2.2853,
      "step": 54019
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7039543986320496,
      "learning_rate": 0.0005222323203616026,
      "loss": 2.4626,
      "step": 54020
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.708518922328949,
      "learning_rate": 0.0005222295724898996,
      "loss": 2.416,
      "step": 54021
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7179737091064453,
      "learning_rate": 0.0005222268245768799,
      "loss": 2.2236,
      "step": 54022
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.6615182161331177,
      "learning_rate": 0.0005222240766225441,
      "loss": 2.2122,
      "step": 54023
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7478688359260559,
      "learning_rate": 0.0005222213286268925,
      "loss": 2.1619,
      "step": 54024
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7170103192329407,
      "learning_rate": 0.000522218580589926,
      "loss": 2.0415,
      "step": 54025
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.689594566822052,
      "learning_rate": 0.0005222158325116447,
      "loss": 2.0861,
      "step": 54026
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7317588329315186,
      "learning_rate": 0.0005222130843920493,
      "loss": 2.1936,
      "step": 54027
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7258597612380981,
      "learning_rate": 0.0005222103362311404,
      "loss": 2.2483,
      "step": 54028
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7593918442726135,
      "learning_rate": 0.0005222075880289183,
      "loss": 2.322,
      "step": 54029
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.746082067489624,
      "learning_rate": 0.0005222048397853837,
      "loss": 2.1582,
      "step": 54030
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7137098908424377,
      "learning_rate": 0.000522202091500537,
      "loss": 2.142,
      "step": 54031
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7679272294044495,
      "learning_rate": 0.0005221993431743787,
      "loss": 1.974,
      "step": 54032
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.8677179217338562,
      "learning_rate": 0.0005221965948069093,
      "loss": 2.1478,
      "step": 54033
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7942569255828857,
      "learning_rate": 0.0005221938463981294,
      "loss": 2.1857,
      "step": 54034
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7630184292793274,
      "learning_rate": 0.0005221910979480395,
      "loss": 2.3711,
      "step": 54035
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.8078757524490356,
      "learning_rate": 0.0005221883494566401,
      "loss": 2.0703,
      "step": 54036
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.6901281476020813,
      "learning_rate": 0.0005221856009239317,
      "loss": 2.4756,
      "step": 54037
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7334939241409302,
      "learning_rate": 0.0005221828523499148,
      "loss": 2.2026,
      "step": 54038
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7269511818885803,
      "learning_rate": 0.0005221801037345898,
      "loss": 2.3732,
      "step": 54039
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7968943119049072,
      "learning_rate": 0.0005221773550779575,
      "loss": 2.2703,
      "step": 54040
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7199792861938477,
      "learning_rate": 0.0005221746063800181,
      "loss": 2.1912,
      "step": 54041
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7838200926780701,
      "learning_rate": 0.0005221718576407722,
      "loss": 2.4995,
      "step": 54042
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7112839818000793,
      "learning_rate": 0.0005221691088602205,
      "loss": 2.0381,
      "step": 54043
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7642671465873718,
      "learning_rate": 0.0005221663600383633,
      "loss": 2.1369,
      "step": 54044
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7275418043136597,
      "learning_rate": 0.0005221636111752012,
      "loss": 2.1911,
      "step": 54045
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.8813015222549438,
      "learning_rate": 0.0005221608622707347,
      "loss": 2.0927,
      "step": 54046
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7267979979515076,
      "learning_rate": 0.0005221581133249643,
      "loss": 2.3876,
      "step": 54047
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7385017275810242,
      "learning_rate": 0.0005221553643378905,
      "loss": 2.1119,
      "step": 54048
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7752826809883118,
      "learning_rate": 0.0005221526153095139,
      "loss": 2.29,
      "step": 54049
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7977232336997986,
      "learning_rate": 0.0005221498662398347,
      "loss": 2.6109,
      "step": 54050
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7785822153091431,
      "learning_rate": 0.0005221471171288539,
      "loss": 2.2763,
      "step": 54051
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.9449517726898193,
      "learning_rate": 0.0005221443679765716,
      "loss": 2.1428,
      "step": 54052
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.6965792775154114,
      "learning_rate": 0.0005221416187829886,
      "loss": 2.2858,
      "step": 54053
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7800037860870361,
      "learning_rate": 0.0005221388695481051,
      "loss": 1.9871,
      "step": 54054
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7629216909408569,
      "learning_rate": 0.000522136120271922,
      "loss": 2.1987,
      "step": 54055
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.8310452103614807,
      "learning_rate": 0.0005221333709544394,
      "loss": 1.9298,
      "step": 54056
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7526716589927673,
      "learning_rate": 0.0005221306215956582,
      "loss": 2.2087,
      "step": 54057
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.9039537906646729,
      "learning_rate": 0.0005221278721955786,
      "loss": 2.3045,
      "step": 54058
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.889894962310791,
      "learning_rate": 0.0005221251227542014,
      "loss": 2.3453,
      "step": 54059
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7295064926147461,
      "learning_rate": 0.0005221223732715267,
      "loss": 2.1268,
      "step": 54060
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7217292189598083,
      "learning_rate": 0.0005221196237475554,
      "loss": 2.1493,
      "step": 54061
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.9556732773780823,
      "learning_rate": 0.0005221168741822879,
      "loss": 2.2701,
      "step": 54062
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7150623202323914,
      "learning_rate": 0.0005221141245757246,
      "loss": 2.1842,
      "step": 54063
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7180435657501221,
      "learning_rate": 0.0005221113749278661,
      "loss": 2.3207,
      "step": 54064
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7705990672111511,
      "learning_rate": 0.0005221086252387129,
      "loss": 2.2386,
      "step": 54065
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.8661937117576599,
      "learning_rate": 0.0005221058755082655,
      "loss": 2.2411,
      "step": 54066
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7975959777832031,
      "learning_rate": 0.0005221031257365246,
      "loss": 2.2939,
      "step": 54067
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7531606554985046,
      "learning_rate": 0.0005221003759234903,
      "loss": 2.0756,
      "step": 54068
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.8541178107261658,
      "learning_rate": 0.0005220976260691634,
      "loss": 2.2065,
      "step": 54069
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.8620455861091614,
      "learning_rate": 0.0005220948761735445,
      "loss": 2.4218,
      "step": 54070
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7008188366889954,
      "learning_rate": 0.0005220921262366339,
      "loss": 2.0785,
      "step": 54071
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.8474029898643494,
      "learning_rate": 0.0005220893762584321,
      "loss": 2.3893,
      "step": 54072
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.8565208315849304,
      "learning_rate": 0.0005220866262389398,
      "loss": 2.27,
      "step": 54073
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.8336681723594666,
      "learning_rate": 0.0005220838761781573,
      "loss": 2.222,
      "step": 54074
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7464041113853455,
      "learning_rate": 0.0005220811260760853,
      "loss": 2.031,
      "step": 54075
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7592294812202454,
      "learning_rate": 0.0005220783759327241,
      "loss": 2.2059,
      "step": 54076
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7628539800643921,
      "learning_rate": 0.0005220756257480745,
      "loss": 2.2052,
      "step": 54077
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7458203434944153,
      "learning_rate": 0.0005220728755221368,
      "loss": 2.1729,
      "step": 54078
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.8313648700714111,
      "learning_rate": 0.0005220701252549116,
      "loss": 2.2872,
      "step": 54079
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7896488904953003,
      "learning_rate": 0.0005220673749463993,
      "loss": 2.4524,
      "step": 54080
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.727184534072876,
      "learning_rate": 0.0005220646245966005,
      "loss": 2.4011,
      "step": 54081
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7113772630691528,
      "learning_rate": 0.0005220618742055157,
      "loss": 2.1697,
      "step": 54082
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.782712996006012,
      "learning_rate": 0.0005220591237731453,
      "loss": 2.2305,
      "step": 54083
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7340368628501892,
      "learning_rate": 0.0005220563732994901,
      "loss": 2.3602,
      "step": 54084
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.8079936504364014,
      "learning_rate": 0.0005220536227845502,
      "loss": 2.3959,
      "step": 54085
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7106720209121704,
      "learning_rate": 0.0005220508722283266,
      "loss": 2.1697,
      "step": 54086
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7511614561080933,
      "learning_rate": 0.0005220481216308195,
      "loss": 2.183,
      "step": 54087
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7519554495811462,
      "learning_rate": 0.0005220453709920294,
      "loss": 2.2909,
      "step": 54088
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7457805275917053,
      "learning_rate": 0.0005220426203119569,
      "loss": 2.2005,
      "step": 54089
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.6700049042701721,
      "learning_rate": 0.0005220398695906025,
      "loss": 2.1894,
      "step": 54090
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.8772124648094177,
      "learning_rate": 0.0005220371188279665,
      "loss": 2.2641,
      "step": 54091
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7348310947418213,
      "learning_rate": 0.0005220343680240499,
      "loss": 2.2745,
      "step": 54092
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7505112290382385,
      "learning_rate": 0.0005220316171788527,
      "loss": 2.3315,
      "step": 54093
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7961764335632324,
      "learning_rate": 0.0005220288662923757,
      "loss": 2.2837,
      "step": 54094
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.9386385083198547,
      "learning_rate": 0.0005220261153646194,
      "loss": 2.3425,
      "step": 54095
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.77093505859375,
      "learning_rate": 0.0005220233643955842,
      "loss": 2.1903,
      "step": 54096
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7218533754348755,
      "learning_rate": 0.0005220206133852707,
      "loss": 2.351,
      "step": 54097
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7645930647850037,
      "learning_rate": 0.0005220178623336793,
      "loss": 2.1161,
      "step": 54098
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.0751398801803589,
      "learning_rate": 0.0005220151112408106,
      "loss": 2.2561,
      "step": 54099
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7377564311027527,
      "learning_rate": 0.0005220123601066653,
      "loss": 2.3122,
      "step": 54100
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.6876461505889893,
      "learning_rate": 0.0005220096089312435,
      "loss": 2.1186,
      "step": 54101
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.8714778423309326,
      "learning_rate": 0.000522006857714546,
      "loss": 2.2736,
      "step": 54102
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.8802450299263,
      "learning_rate": 0.0005220041064565732,
      "loss": 2.3049,
      "step": 54103
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7488008141517639,
      "learning_rate": 0.0005220013551573256,
      "loss": 2.4053,
      "step": 54104
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7185050249099731,
      "learning_rate": 0.0005219986038168038,
      "loss": 2.2708,
      "step": 54105
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7951761484146118,
      "learning_rate": 0.0005219958524350082,
      "loss": 1.9787,
      "step": 54106
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7369033694267273,
      "learning_rate": 0.0005219931010119396,
      "loss": 2.3239,
      "step": 54107
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7440524697303772,
      "learning_rate": 0.0005219903495475982,
      "loss": 2.2675,
      "step": 54108
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7045074701309204,
      "learning_rate": 0.0005219875980419845,
      "loss": 2.2978,
      "step": 54109
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7610002160072327,
      "learning_rate": 0.0005219848464950992,
      "loss": 2.1446,
      "step": 54110
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.8305134177207947,
      "learning_rate": 0.0005219820949069427,
      "loss": 2.0166,
      "step": 54111
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7508703470230103,
      "learning_rate": 0.0005219793432775155,
      "loss": 2.1827,
      "step": 54112
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.657425582408905,
      "learning_rate": 0.0005219765916068182,
      "loss": 2.1748,
      "step": 54113
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.1237223148345947,
      "learning_rate": 0.0005219738398948513,
      "loss": 2.2138,
      "step": 54114
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.9029754996299744,
      "learning_rate": 0.0005219710881416152,
      "loss": 2.2076,
      "step": 54115
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.6508799195289612,
      "learning_rate": 0.0005219683363471105,
      "loss": 2.1145,
      "step": 54116
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.919319212436676,
      "learning_rate": 0.0005219655845113378,
      "loss": 2.1336,
      "step": 54117
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.9501317739486694,
      "learning_rate": 0.0005219628326342974,
      "loss": 2.2493,
      "step": 54118
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.6966441869735718,
      "learning_rate": 0.0005219600807159899,
      "loss": 2.2392,
      "step": 54119
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.8780913352966309,
      "learning_rate": 0.0005219573287564158,
      "loss": 1.9618,
      "step": 54120
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.8398557901382446,
      "learning_rate": 0.0005219545767555757,
      "loss": 2.13,
      "step": 54121
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.739133894443512,
      "learning_rate": 0.0005219518247134701,
      "loss": 2.5701,
      "step": 54122
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7176939845085144,
      "learning_rate": 0.0005219490726300994,
      "loss": 2.1591,
      "step": 54123
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.78049635887146,
      "learning_rate": 0.0005219463205054642,
      "loss": 2.2064,
      "step": 54124
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7940566539764404,
      "learning_rate": 0.000521943568339565,
      "loss": 2.0575,
      "step": 54125
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7124010324478149,
      "learning_rate": 0.0005219408161324024,
      "loss": 2.1271,
      "step": 54126
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7292248010635376,
      "learning_rate": 0.0005219380638839767,
      "loss": 2.0571,
      "step": 54127
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7979731559753418,
      "learning_rate": 0.0005219353115942885,
      "loss": 2.2079,
      "step": 54128
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.8501096963882446,
      "learning_rate": 0.0005219325592633383,
      "loss": 2.2308,
      "step": 54129
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.720241367816925,
      "learning_rate": 0.0005219298068911268,
      "loss": 2.2747,
      "step": 54130
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.8101944327354431,
      "learning_rate": 0.0005219270544776543,
      "loss": 2.2834,
      "step": 54131
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.9187290668487549,
      "learning_rate": 0.0005219243020229212,
      "loss": 2.2377,
      "step": 54132
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.8361434936523438,
      "learning_rate": 0.0005219215495269284,
      "loss": 2.1659,
      "step": 54133
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7029227018356323,
      "learning_rate": 0.0005219187969896761,
      "loss": 2.4413,
      "step": 54134
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.8172813057899475,
      "learning_rate": 0.000521916044411165,
      "loss": 2.1377,
      "step": 54135
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7135600447654724,
      "learning_rate": 0.0005219132917913955,
      "loss": 2.3425,
      "step": 54136
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.6933208107948303,
      "learning_rate": 0.0005219105391303681,
      "loss": 2.1638,
      "step": 54137
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.6946204304695129,
      "learning_rate": 0.0005219077864280833,
      "loss": 2.1188,
      "step": 54138
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7663401961326599,
      "learning_rate": 0.0005219050336845416,
      "loss": 2.0985,
      "step": 54139
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7479015588760376,
      "learning_rate": 0.0005219022808997437,
      "loss": 2.3888,
      "step": 54140
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.855968713760376,
      "learning_rate": 0.0005218995280736898,
      "loss": 2.439,
      "step": 54141
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7201042175292969,
      "learning_rate": 0.0005218967752063809,
      "loss": 2.1266,
      "step": 54142
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.6925946474075317,
      "learning_rate": 0.0005218940222978169,
      "loss": 2.117,
      "step": 54143
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7342609763145447,
      "learning_rate": 0.0005218912693479988,
      "loss": 2.2019,
      "step": 54144
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7146081328392029,
      "learning_rate": 0.0005218885163569268,
      "loss": 2.1457,
      "step": 54145
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.6914196014404297,
      "learning_rate": 0.0005218857633246016,
      "loss": 2.3764,
      "step": 54146
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7188904285430908,
      "learning_rate": 0.0005218830102510237,
      "loss": 2.042,
      "step": 54147
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7781381607055664,
      "learning_rate": 0.0005218802571361935,
      "loss": 2.2218,
      "step": 54148
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7587954998016357,
      "learning_rate": 0.0005218775039801116,
      "loss": 2.1581,
      "step": 54149
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.8676753044128418,
      "learning_rate": 0.0005218747507827784,
      "loss": 2.5271,
      "step": 54150
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.680590033531189,
      "learning_rate": 0.0005218719975441944,
      "loss": 2.3696,
      "step": 54151
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7254934906959534,
      "learning_rate": 0.0005218692442643605,
      "loss": 2.3518,
      "step": 54152
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.6886137127876282,
      "learning_rate": 0.0005218664909432767,
      "loss": 2.3794,
      "step": 54153
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7199181914329529,
      "learning_rate": 0.0005218637375809438,
      "loss": 2.2655,
      "step": 54154
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7792772650718689,
      "learning_rate": 0.0005218609841773622,
      "loss": 2.4481,
      "step": 54155
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7024405002593994,
      "learning_rate": 0.0005218582307325325,
      "loss": 2.3214,
      "step": 54156
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7034214735031128,
      "learning_rate": 0.0005218554772464552,
      "loss": 2.2409,
      "step": 54157
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7042518258094788,
      "learning_rate": 0.0005218527237191307,
      "loss": 2.1155,
      "step": 54158
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.6995348334312439,
      "learning_rate": 0.0005218499701505597,
      "loss": 1.9476,
      "step": 54159
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7956987023353577,
      "learning_rate": 0.0005218472165407426,
      "loss": 2.3797,
      "step": 54160
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7092994451522827,
      "learning_rate": 0.0005218444628896798,
      "loss": 2.0819,
      "step": 54161
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.738759458065033,
      "learning_rate": 0.0005218417091973719,
      "loss": 2.3439,
      "step": 54162
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.707512378692627,
      "learning_rate": 0.0005218389554638196,
      "loss": 2.6459,
      "step": 54163
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7165923118591309,
      "learning_rate": 0.0005218362016890231,
      "loss": 2.0963,
      "step": 54164
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.6896770000457764,
      "learning_rate": 0.0005218334478729833,
      "loss": 2.2831,
      "step": 54165
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7221508026123047,
      "learning_rate": 0.0005218306940157002,
      "loss": 2.1957,
      "step": 54166
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7850754261016846,
      "learning_rate": 0.0005218279401171747,
      "loss": 2.3574,
      "step": 54167
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.736142098903656,
      "learning_rate": 0.0005218251861774071,
      "loss": 2.0434,
      "step": 54168
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.9566333889961243,
      "learning_rate": 0.0005218224321963982,
      "loss": 2.2949,
      "step": 54169
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7316128611564636,
      "learning_rate": 0.0005218196781741482,
      "loss": 2.332,
      "step": 54170
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.6942916512489319,
      "learning_rate": 0.0005218169241106577,
      "loss": 2.0826,
      "step": 54171
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.8344993591308594,
      "learning_rate": 0.0005218141700059273,
      "loss": 2.3885,
      "step": 54172
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.8688109517097473,
      "learning_rate": 0.0005218114158599575,
      "loss": 2.1997,
      "step": 54173
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7207426428794861,
      "learning_rate": 0.0005218086616727488,
      "loss": 2.4245,
      "step": 54174
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7478707432746887,
      "learning_rate": 0.0005218059074443016,
      "loss": 2.0893,
      "step": 54175
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7734919786453247,
      "learning_rate": 0.0005218031531746165,
      "loss": 2.2457,
      "step": 54176
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.751431405544281,
      "learning_rate": 0.000521800398863694,
      "loss": 2.165,
      "step": 54177
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.6811875700950623,
      "learning_rate": 0.0005217976445115347,
      "loss": 2.3371,
      "step": 54178
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.8302507400512695,
      "learning_rate": 0.0005217948901181389,
      "loss": 2.358,
      "step": 54179
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7789986729621887,
      "learning_rate": 0.0005217921356835074,
      "loss": 2.2843,
      "step": 54180
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7200241088867188,
      "learning_rate": 0.0005217893812076405,
      "loss": 2.0704,
      "step": 54181
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7168125510215759,
      "learning_rate": 0.0005217866266905388,
      "loss": 2.207,
      "step": 54182
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7354270219802856,
      "learning_rate": 0.0005217838721322027,
      "loss": 2.3279,
      "step": 54183
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7653445601463318,
      "learning_rate": 0.0005217811175326327,
      "loss": 2.5843,
      "step": 54184
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7915899157524109,
      "learning_rate": 0.0005217783628918295,
      "loss": 2.1804,
      "step": 54185
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.716083824634552,
      "learning_rate": 0.0005217756082097937,
      "loss": 2.4938,
      "step": 54186
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7451352477073669,
      "learning_rate": 0.0005217728534865253,
      "loss": 2.3155,
      "step": 54187
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7385850548744202,
      "learning_rate": 0.0005217700987220253,
      "loss": 2.2317,
      "step": 54188
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.8032233715057373,
      "learning_rate": 0.0005217673439162941,
      "loss": 2.3297,
      "step": 54189
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.6819891929626465,
      "learning_rate": 0.000521764589069332,
      "loss": 2.2154,
      "step": 54190
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.8271902799606323,
      "learning_rate": 0.0005217618341811399,
      "loss": 2.3956,
      "step": 54191
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7408470511436462,
      "learning_rate": 0.0005217590792517179,
      "loss": 2.227,
      "step": 54192
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7186654210090637,
      "learning_rate": 0.0005217563242810668,
      "loss": 2.3921,
      "step": 54193
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.8280455470085144,
      "learning_rate": 0.000521753569269187,
      "loss": 2.2839,
      "step": 54194
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7748572826385498,
      "learning_rate": 0.0005217508142160791,
      "loss": 2.2837,
      "step": 54195
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.828632116317749,
      "learning_rate": 0.0005217480591217434,
      "loss": 2.2635,
      "step": 54196
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.8089713454246521,
      "learning_rate": 0.0005217453039861806,
      "loss": 2.1621,
      "step": 54197
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7117592096328735,
      "learning_rate": 0.0005217425488093912,
      "loss": 2.364,
      "step": 54198
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7096526026725769,
      "learning_rate": 0.0005217397935913756,
      "loss": 2.1109,
      "step": 54199
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7610203623771667,
      "learning_rate": 0.0005217370383321344,
      "loss": 2.4383,
      "step": 54200
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.8256657719612122,
      "learning_rate": 0.0005217342830316681,
      "loss": 1.9261,
      "step": 54201
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7461605072021484,
      "learning_rate": 0.0005217315276899773,
      "loss": 2.346,
      "step": 54202
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7484429478645325,
      "learning_rate": 0.0005217287723070624,
      "loss": 1.9864,
      "step": 54203
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7665582299232483,
      "learning_rate": 0.0005217260168829239,
      "loss": 2.3016,
      "step": 54204
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.8488081693649292,
      "learning_rate": 0.0005217232614175624,
      "loss": 2.206,
      "step": 54205
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.6887288093566895,
      "learning_rate": 0.0005217205059109782,
      "loss": 2.2278,
      "step": 54206
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7340449094772339,
      "learning_rate": 0.0005217177503631722,
      "loss": 2.2094,
      "step": 54207
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.8266698122024536,
      "learning_rate": 0.0005217149947741445,
      "loss": 2.0532,
      "step": 54208
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7555558681488037,
      "learning_rate": 0.0005217122391438959,
      "loss": 2.3366,
      "step": 54209
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7824366688728333,
      "learning_rate": 0.0005217094834724268,
      "loss": 2.2849,
      "step": 54210
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7229017615318298,
      "learning_rate": 0.0005217067277597376,
      "loss": 2.3675,
      "step": 54211
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.746522843837738,
      "learning_rate": 0.0005217039720058291,
      "loss": 2.2937,
      "step": 54212
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.77057945728302,
      "learning_rate": 0.0005217012162107015,
      "loss": 2.2537,
      "step": 54213
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.6836368441581726,
      "learning_rate": 0.0005216984603743556,
      "loss": 2.1002,
      "step": 54214
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7887193560600281,
      "learning_rate": 0.0005216957044967917,
      "loss": 2.2636,
      "step": 54215
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7469869256019592,
      "learning_rate": 0.0005216929485780104,
      "loss": 2.3342,
      "step": 54216
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.6589294075965881,
      "learning_rate": 0.0005216901926180121,
      "loss": 2.0191,
      "step": 54217
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.8838593363761902,
      "learning_rate": 0.0005216874366167976,
      "loss": 2.2186,
      "step": 54218
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.692244291305542,
      "learning_rate": 0.0005216846805743671,
      "loss": 2.2061,
      "step": 54219
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7415516376495361,
      "learning_rate": 0.0005216819244907212,
      "loss": 2.3281,
      "step": 54220
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7991957664489746,
      "learning_rate": 0.0005216791683658605,
      "loss": 2.471,
      "step": 54221
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.8145337104797363,
      "learning_rate": 0.0005216764121997855,
      "loss": 2.3276,
      "step": 54222
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.6901805996894836,
      "learning_rate": 0.0005216736559924966,
      "loss": 2.1557,
      "step": 54223
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.8491294384002686,
      "learning_rate": 0.0005216708997439944,
      "loss": 2.3236,
      "step": 54224
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.8042666912078857,
      "learning_rate": 0.0005216681434542795,
      "loss": 2.5236,
      "step": 54225
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7748658657073975,
      "learning_rate": 0.0005216653871233521,
      "loss": 1.9861,
      "step": 54226
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.720143735408783,
      "learning_rate": 0.0005216626307512131,
      "loss": 1.9811,
      "step": 54227
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.6441356539726257,
      "learning_rate": 0.0005216598743378627,
      "loss": 2.0069,
      "step": 54228
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7067667245864868,
      "learning_rate": 0.0005216571178833017,
      "loss": 2.2445,
      "step": 54229
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.8141927123069763,
      "learning_rate": 0.0005216543613875303,
      "loss": 2.0991,
      "step": 54230
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.749311625957489,
      "learning_rate": 0.0005216516048505492,
      "loss": 2.2583,
      "step": 54231
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.6953661441802979,
      "learning_rate": 0.000521648848272359,
      "loss": 2.2599,
      "step": 54232
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7915058732032776,
      "learning_rate": 0.00052164609165296,
      "loss": 2.2535,
      "step": 54233
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7956143617630005,
      "learning_rate": 0.0005216433349923528,
      "loss": 2.3579,
      "step": 54234
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.8253930807113647,
      "learning_rate": 0.0005216405782905379,
      "loss": 2.34,
      "step": 54235
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7665529847145081,
      "learning_rate": 0.0005216378215475159,
      "loss": 2.2048,
      "step": 54236
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7547323703765869,
      "learning_rate": 0.0005216350647632872,
      "loss": 2.3262,
      "step": 54237
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7187122106552124,
      "learning_rate": 0.0005216323079378525,
      "loss": 2.2861,
      "step": 54238
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.8039004802703857,
      "learning_rate": 0.0005216295510712119,
      "loss": 2.345,
      "step": 54239
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7647477984428406,
      "learning_rate": 0.0005216267941633663,
      "loss": 2.4606,
      "step": 54240
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.67626953125,
      "learning_rate": 0.0005216240372143162,
      "loss": 2.0615,
      "step": 54241
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7770243883132935,
      "learning_rate": 0.0005216212802240618,
      "loss": 2.1451,
      "step": 54242
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.8113439083099365,
      "learning_rate": 0.0005216185231926041,
      "loss": 2.3024,
      "step": 54243
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.6985671520233154,
      "learning_rate": 0.0005216157661199432,
      "loss": 2.3343,
      "step": 54244
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.8038685917854309,
      "learning_rate": 0.0005216130090060797,
      "loss": 2.1829,
      "step": 54245
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.8652512431144714,
      "learning_rate": 0.0005216102518510142,
      "loss": 2.3498,
      "step": 54246
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7307891845703125,
      "learning_rate": 0.0005216074946547472,
      "loss": 2.2232,
      "step": 54247
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7407203912734985,
      "learning_rate": 0.0005216047374172791,
      "loss": 2.2176,
      "step": 54248
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7182011604309082,
      "learning_rate": 0.0005216019801386106,
      "loss": 2.2553,
      "step": 54249
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7892264723777771,
      "learning_rate": 0.0005215992228187421,
      "loss": 2.0464,
      "step": 54250
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7459366917610168,
      "learning_rate": 0.000521596465457674,
      "loss": 1.9909,
      "step": 54251
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.6798633337020874,
      "learning_rate": 0.0005215937080554069,
      "loss": 2.266,
      "step": 54252
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7200232148170471,
      "learning_rate": 0.0005215909506119415,
      "loss": 2.1721,
      "step": 54253
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.8763487935066223,
      "learning_rate": 0.0005215881931272782,
      "loss": 2.1585,
      "step": 54254
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7081810235977173,
      "learning_rate": 0.0005215854356014173,
      "loss": 2.2057,
      "step": 54255
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7843149304389954,
      "learning_rate": 0.0005215826780343597,
      "loss": 2.0798,
      "step": 54256
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7652055621147156,
      "learning_rate": 0.0005215799204261055,
      "loss": 2.1618,
      "step": 54257
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7400187849998474,
      "learning_rate": 0.0005215771627766555,
      "loss": 2.3637,
      "step": 54258
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.8883578777313232,
      "learning_rate": 0.0005215744050860101,
      "loss": 2.2704,
      "step": 54259
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7898539900779724,
      "learning_rate": 0.0005215716473541698,
      "loss": 2.2419,
      "step": 54260
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7442578077316284,
      "learning_rate": 0.0005215688895811351,
      "loss": 2.3151,
      "step": 54261
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.726061224937439,
      "learning_rate": 0.0005215661317669067,
      "loss": 2.2015,
      "step": 54262
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7412214875221252,
      "learning_rate": 0.000521563373911485,
      "loss": 2.215,
      "step": 54263
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.6928642392158508,
      "learning_rate": 0.0005215606160148704,
      "loss": 2.0114,
      "step": 54264
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.869539737701416,
      "learning_rate": 0.0005215578580770635,
      "loss": 2.361,
      "step": 54265
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7061970829963684,
      "learning_rate": 0.0005215551000980648,
      "loss": 2.1582,
      "step": 54266
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.718990683555603,
      "learning_rate": 0.0005215523420778748,
      "loss": 2.2031,
      "step": 54267
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.8591445684432983,
      "learning_rate": 0.0005215495840164941,
      "loss": 2.2043,
      "step": 54268
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7043043971061707,
      "learning_rate": 0.0005215468259139231,
      "loss": 2.0507,
      "step": 54269
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.6882919669151306,
      "learning_rate": 0.0005215440677701624,
      "loss": 1.9709,
      "step": 54270
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7600913047790527,
      "learning_rate": 0.0005215413095852124,
      "loss": 2.4562,
      "step": 54271
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.813157320022583,
      "learning_rate": 0.0005215385513590738,
      "loss": 2.1824,
      "step": 54272
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.732541024684906,
      "learning_rate": 0.000521535793091747,
      "loss": 2.198,
      "step": 54273
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7389468550682068,
      "learning_rate": 0.0005215330347832324,
      "loss": 2.2005,
      "step": 54274
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7552217841148376,
      "learning_rate": 0.0005215302764335308,
      "loss": 2.1993,
      "step": 54275
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7277265191078186,
      "learning_rate": 0.0005215275180426425,
      "loss": 2.1373,
      "step": 54276
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7224703431129456,
      "learning_rate": 0.0005215247596105679,
      "loss": 2.3016,
      "step": 54277
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7376236319541931,
      "learning_rate": 0.0005215220011373078,
      "loss": 2.17,
      "step": 54278
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7433890700340271,
      "learning_rate": 0.0005215192426228626,
      "loss": 2.3511,
      "step": 54279
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7792121767997742,
      "learning_rate": 0.0005215164840672326,
      "loss": 2.3713,
      "step": 54280
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7082409262657166,
      "learning_rate": 0.0005215137254704186,
      "loss": 2.3217,
      "step": 54281
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.8137054443359375,
      "learning_rate": 0.0005215109668324211,
      "loss": 2.301,
      "step": 54282
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.77133709192276,
      "learning_rate": 0.0005215082081532405,
      "loss": 2.0799,
      "step": 54283
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7093616724014282,
      "learning_rate": 0.0005215054494328773,
      "loss": 2.3463,
      "step": 54284
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.9689075946807861,
      "learning_rate": 0.000521502690671332,
      "loss": 2.2043,
      "step": 54285
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7577118277549744,
      "learning_rate": 0.0005214999318686053,
      "loss": 2.2067,
      "step": 54286
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7123112082481384,
      "learning_rate": 0.0005214971730246975,
      "loss": 2.2322,
      "step": 54287
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7322902083396912,
      "learning_rate": 0.0005214944141396093,
      "loss": 2.2879,
      "step": 54288
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7224233746528625,
      "learning_rate": 0.000521491655213341,
      "loss": 2.2107,
      "step": 54289
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.8976727724075317,
      "learning_rate": 0.0005214888962458932,
      "loss": 2.1924,
      "step": 54290
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7946513891220093,
      "learning_rate": 0.0005214861372372665,
      "loss": 2.4043,
      "step": 54291
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.8853245377540588,
      "learning_rate": 0.0005214833781874613,
      "loss": 2.1404,
      "step": 54292
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7088368535041809,
      "learning_rate": 0.0005214806190964782,
      "loss": 2.2157,
      "step": 54293
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.6759628057479858,
      "learning_rate": 0.0005214778599643176,
      "loss": 2.1541,
      "step": 54294
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7654184699058533,
      "learning_rate": 0.0005214751007909802,
      "loss": 2.3549,
      "step": 54295
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.8642089366912842,
      "learning_rate": 0.0005214723415764664,
      "loss": 2.1797,
      "step": 54296
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.8188953399658203,
      "learning_rate": 0.0005214695823207767,
      "loss": 2.2902,
      "step": 54297
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.6911213397979736,
      "learning_rate": 0.0005214668230239116,
      "loss": 2.2498,
      "step": 54298
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.8223845958709717,
      "learning_rate": 0.0005214640636858716,
      "loss": 2.5637,
      "step": 54299
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7630418539047241,
      "learning_rate": 0.0005214613043066572,
      "loss": 2.2223,
      "step": 54300
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7402311563491821,
      "learning_rate": 0.0005214585448862691,
      "loss": 2.2092,
      "step": 54301
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.6998122930526733,
      "learning_rate": 0.0005214557854247076,
      "loss": 2.3837,
      "step": 54302
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.8106791377067566,
      "learning_rate": 0.0005214530259219733,
      "loss": 2.182,
      "step": 54303
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7172992825508118,
      "learning_rate": 0.0005214502663780666,
      "loss": 2.2083,
      "step": 54304
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7496500611305237,
      "learning_rate": 0.0005214475067929882,
      "loss": 2.4509,
      "step": 54305
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7699806094169617,
      "learning_rate": 0.0005214447471667386,
      "loss": 2.3039,
      "step": 54306
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7403958439826965,
      "learning_rate": 0.0005214419874993181,
      "loss": 2.1823,
      "step": 54307
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7769070863723755,
      "learning_rate": 0.0005214392277907275,
      "loss": 2.2331,
      "step": 54308
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7367726564407349,
      "learning_rate": 0.0005214364680409671,
      "loss": 2.202,
      "step": 54309
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7401686906814575,
      "learning_rate": 0.0005214337082500375,
      "loss": 2.1671,
      "step": 54310
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.8216292858123779,
      "learning_rate": 0.0005214309484179392,
      "loss": 2.4354,
      "step": 54311
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.6936500072479248,
      "learning_rate": 0.0005214281885446727,
      "loss": 2.2348,
      "step": 54312
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.8194226622581482,
      "learning_rate": 0.0005214254286302385,
      "loss": 2.2717,
      "step": 54313
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7115585207939148,
      "learning_rate": 0.000521422668674637,
      "loss": 2.323,
      "step": 54314
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.8351518511772156,
      "learning_rate": 0.0005214199086778691,
      "loss": 2.2284,
      "step": 54315
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.8075935244560242,
      "learning_rate": 0.0005214171486399349,
      "loss": 2.4666,
      "step": 54316
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.9041328430175781,
      "learning_rate": 0.0005214143885608352,
      "loss": 2.2672,
      "step": 54317
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.8071194291114807,
      "learning_rate": 0.0005214116284405703,
      "loss": 2.2569,
      "step": 54318
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7013082504272461,
      "learning_rate": 0.0005214088682791408,
      "loss": 1.9687,
      "step": 54319
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.8641291856765747,
      "learning_rate": 0.0005214061080765471,
      "loss": 2.3539,
      "step": 54320
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7787050008773804,
      "learning_rate": 0.0005214033478327901,
      "loss": 2.1416,
      "step": 54321
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.6658904552459717,
      "learning_rate": 0.0005214005875478698,
      "loss": 2.2423,
      "step": 54322
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7654808163642883,
      "learning_rate": 0.000521397827221787,
      "loss": 2.0631,
      "step": 54323
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.6747563481330872,
      "learning_rate": 0.0005213950668545422,
      "loss": 2.1564,
      "step": 54324
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.820325493812561,
      "learning_rate": 0.0005213923064461358,
      "loss": 2.0871,
      "step": 54325
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7188587784767151,
      "learning_rate": 0.0005213895459965685,
      "loss": 2.2667,
      "step": 54326
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.9723049402236938,
      "learning_rate": 0.0005213867855058406,
      "loss": 2.0925,
      "step": 54327
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7610418796539307,
      "learning_rate": 0.0005213840249739529,
      "loss": 2.4466,
      "step": 54328
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7634602189064026,
      "learning_rate": 0.0005213812644009056,
      "loss": 2.3098,
      "step": 54329
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7156186699867249,
      "learning_rate": 0.0005213785037866994,
      "loss": 2.2962,
      "step": 54330
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.8704503774642944,
      "learning_rate": 0.0005213757431313347,
      "loss": 2.3146,
      "step": 54331
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7098198533058167,
      "learning_rate": 0.0005213729824348121,
      "loss": 2.2882,
      "step": 54332
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.6897925138473511,
      "learning_rate": 0.000521370221697132,
      "loss": 2.2881,
      "step": 54333
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.6570772528648376,
      "learning_rate": 0.0005213674609182951,
      "loss": 2.1434,
      "step": 54334
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.6889390349388123,
      "learning_rate": 0.0005213647000983017,
      "loss": 2.4448,
      "step": 54335
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.6785518527030945,
      "learning_rate": 0.0005213619392371526,
      "loss": 2.0868,
      "step": 54336
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.886166512966156,
      "learning_rate": 0.0005213591783348481,
      "loss": 2.3212,
      "step": 54337
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7019819021224976,
      "learning_rate": 0.0005213564173913885,
      "loss": 2.1988,
      "step": 54338
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7021514773368835,
      "learning_rate": 0.0005213536564067749,
      "loss": 2.1451,
      "step": 54339
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.8107554912567139,
      "learning_rate": 0.0005213508953810073,
      "loss": 2.3522,
      "step": 54340
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7650540471076965,
      "learning_rate": 0.0005213481343140864,
      "loss": 2.2773,
      "step": 54341
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7377827763557434,
      "learning_rate": 0.0005213453732060127,
      "loss": 2.3538,
      "step": 54342
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.8194620609283447,
      "learning_rate": 0.0005213426120567867,
      "loss": 2.155,
      "step": 54343
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.67340487241745,
      "learning_rate": 0.000521339850866409,
      "loss": 2.2571,
      "step": 54344
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7289897799491882,
      "learning_rate": 0.00052133708963488,
      "loss": 2.2554,
      "step": 54345
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.8469868898391724,
      "learning_rate": 0.0005213343283622003,
      "loss": 2.3702,
      "step": 54346
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7459768056869507,
      "learning_rate": 0.0005213315670483703,
      "loss": 2.4867,
      "step": 54347
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7338458895683289,
      "learning_rate": 0.0005213288056933906,
      "loss": 2.3934,
      "step": 54348
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7372655868530273,
      "learning_rate": 0.0005213260442972618,
      "loss": 1.997,
      "step": 54349
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7701814770698547,
      "learning_rate": 0.0005213232828599842,
      "loss": 2.023,
      "step": 54350
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7385143041610718,
      "learning_rate": 0.0005213205213815585,
      "loss": 2.4376,
      "step": 54351
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.6878027319908142,
      "learning_rate": 0.0005213177598619851,
      "loss": 2.1258,
      "step": 54352
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.8066202998161316,
      "learning_rate": 0.0005213149983012645,
      "loss": 2.2868,
      "step": 54353
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7212937474250793,
      "learning_rate": 0.0005213122366993973,
      "loss": 2.1896,
      "step": 54354
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.76236891746521,
      "learning_rate": 0.000521309475056384,
      "loss": 2.2017,
      "step": 54355
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.710931658744812,
      "learning_rate": 0.0005213067133722251,
      "loss": 2.0268,
      "step": 54356
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7376721501350403,
      "learning_rate": 0.0005213039516469211,
      "loss": 2.0361,
      "step": 54357
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7423847913742065,
      "learning_rate": 0.0005213011898804725,
      "loss": 2.2573,
      "step": 54358
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.6847017407417297,
      "learning_rate": 0.0005212984280728798,
      "loss": 2.1104,
      "step": 54359
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7608408331871033,
      "learning_rate": 0.0005212956662241436,
      "loss": 2.2228,
      "step": 54360
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.6689764857292175,
      "learning_rate": 0.0005212929043342643,
      "loss": 1.9782,
      "step": 54361
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7859686613082886,
      "learning_rate": 0.0005212901424032425,
      "loss": 2.3736,
      "step": 54362
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7646007537841797,
      "learning_rate": 0.0005212873804310787,
      "loss": 2.3238,
      "step": 54363
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.8776983022689819,
      "learning_rate": 0.0005212846184177734,
      "loss": 2.2876,
      "step": 54364
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7206758856773376,
      "learning_rate": 0.0005212818563633271,
      "loss": 2.1124,
      "step": 54365
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7336001992225647,
      "learning_rate": 0.0005212790942677403,
      "loss": 2.2176,
      "step": 54366
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7840051651000977,
      "learning_rate": 0.0005212763321310135,
      "loss": 2.2434,
      "step": 54367
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.8623478412628174,
      "learning_rate": 0.0005212735699531474,
      "loss": 2.2491,
      "step": 54368
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.816332995891571,
      "learning_rate": 0.0005212708077341423,
      "loss": 2.2059,
      "step": 54369
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.6366440057754517,
      "learning_rate": 0.0005212680454739987,
      "loss": 2.1994,
      "step": 54370
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7060950994491577,
      "learning_rate": 0.0005212652831727173,
      "loss": 2.1279,
      "step": 54371
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7050430178642273,
      "learning_rate": 0.0005212625208302985,
      "loss": 2.262,
      "step": 54372
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.8587271571159363,
      "learning_rate": 0.0005212597584467427,
      "loss": 2.1637,
      "step": 54373
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7720015048980713,
      "learning_rate": 0.0005212569960220506,
      "loss": 2.2306,
      "step": 54374
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7727522253990173,
      "learning_rate": 0.0005212542335562227,
      "loss": 2.3809,
      "step": 54375
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.8270203471183777,
      "learning_rate": 0.0005212514710492595,
      "loss": 2.331,
      "step": 54376
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.8312802314758301,
      "learning_rate": 0.0005212487085011614,
      "loss": 2.2323,
      "step": 54377
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.8179589509963989,
      "learning_rate": 0.000521245945911929,
      "loss": 2.2441,
      "step": 54378
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7442824244499207,
      "learning_rate": 0.0005212431832815628,
      "loss": 2.1849,
      "step": 54379
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7172560095787048,
      "learning_rate": 0.0005212404206100632,
      "loss": 2.2156,
      "step": 54380
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7876036167144775,
      "learning_rate": 0.000521237657897431,
      "loss": 2.381,
      "step": 54381
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7442650198936462,
      "learning_rate": 0.0005212348951436665,
      "loss": 2.3523,
      "step": 54382
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.8155369758605957,
      "learning_rate": 0.0005212321323487702,
      "loss": 2.3101,
      "step": 54383
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.65459805727005,
      "learning_rate": 0.0005212293695127428,
      "loss": 2.3049,
      "step": 54384
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.6992391347885132,
      "learning_rate": 0.0005212266066355846,
      "loss": 2.1839,
      "step": 54385
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7561028599739075,
      "learning_rate": 0.0005212238437172961,
      "loss": 2.3197,
      "step": 54386
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.795912504196167,
      "learning_rate": 0.0005212210807578781,
      "loss": 2.4832,
      "step": 54387
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7559448480606079,
      "learning_rate": 0.0005212183177573308,
      "loss": 2.3781,
      "step": 54388
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7425365447998047,
      "learning_rate": 0.0005212155547156549,
      "loss": 2.2342,
      "step": 54389
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7660807371139526,
      "learning_rate": 0.0005212127916328508,
      "loss": 2.3188,
      "step": 54390
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7449027895927429,
      "learning_rate": 0.0005212100285089191,
      "loss": 2.1857,
      "step": 54391
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7478027939796448,
      "learning_rate": 0.0005212072653438603,
      "loss": 2.2197,
      "step": 54392
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7726713418960571,
      "learning_rate": 0.0005212045021376749,
      "loss": 2.0255,
      "step": 54393
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.8068966269493103,
      "learning_rate": 0.0005212017388903634,
      "loss": 2.0669,
      "step": 54394
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.709393322467804,
      "learning_rate": 0.0005211989756019262,
      "loss": 2.1591,
      "step": 54395
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.6751577258110046,
      "learning_rate": 0.000521196212272364,
      "loss": 2.1679,
      "step": 54396
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.8361626267433167,
      "learning_rate": 0.0005211934489016773,
      "loss": 2.0698,
      "step": 54397
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7799957990646362,
      "learning_rate": 0.0005211906854898666,
      "loss": 2.2554,
      "step": 54398
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7247890830039978,
      "learning_rate": 0.0005211879220369323,
      "loss": 2.1842,
      "step": 54399
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.8538451790809631,
      "learning_rate": 0.000521185158542875,
      "loss": 2.278,
      "step": 54400
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.8124895095825195,
      "learning_rate": 0.0005211823950076952,
      "loss": 2.2504,
      "step": 54401
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.0504807233810425,
      "learning_rate": 0.0005211796314313934,
      "loss": 2.2275,
      "step": 54402
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.8852537870407104,
      "learning_rate": 0.0005211768678139702,
      "loss": 2.1513,
      "step": 54403
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7231834530830383,
      "learning_rate": 0.0005211741041554259,
      "loss": 2.3774,
      "step": 54404
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.8072357773780823,
      "learning_rate": 0.0005211713404557614,
      "loss": 2.2487,
      "step": 54405
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7517792582511902,
      "learning_rate": 0.0005211685767149768,
      "loss": 2.3668,
      "step": 54406
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7429013848304749,
      "learning_rate": 0.0005211658129330728,
      "loss": 2.0974,
      "step": 54407
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7240154147148132,
      "learning_rate": 0.0005211630491100498,
      "loss": 2.1901,
      "step": 54408
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7208142876625061,
      "learning_rate": 0.0005211602852459087,
      "loss": 2.3107,
      "step": 54409
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.8228326439857483,
      "learning_rate": 0.0005211575213406495,
      "loss": 2.2694,
      "step": 54410
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.8440611362457275,
      "learning_rate": 0.0005211547573942731,
      "loss": 2.3547,
      "step": 54411
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7443536520004272,
      "learning_rate": 0.0005211519934067797,
      "loss": 2.352,
      "step": 54412
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7743465304374695,
      "learning_rate": 0.00052114922937817,
      "loss": 2.1212,
      "step": 54413
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7805312871932983,
      "learning_rate": 0.0005211464653084446,
      "loss": 2.1428,
      "step": 54414
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.718280553817749,
      "learning_rate": 0.0005211437011976037,
      "loss": 2.2278,
      "step": 54415
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.8409574031829834,
      "learning_rate": 0.0005211409370456482,
      "loss": 2.2386,
      "step": 54416
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.8816573023796082,
      "learning_rate": 0.0005211381728525784,
      "loss": 2.2436,
      "step": 54417
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7184101343154907,
      "learning_rate": 0.0005211354086183947,
      "loss": 2.3011,
      "step": 54418
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7319833636283875,
      "learning_rate": 0.0005211326443430979,
      "loss": 2.2313,
      "step": 54419
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.685269832611084,
      "learning_rate": 0.0005211298800266883,
      "loss": 2.1788,
      "step": 54420
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7229471206665039,
      "learning_rate": 0.0005211271156691665,
      "loss": 2.2781,
      "step": 54421
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.8950734734535217,
      "learning_rate": 0.000521124351270533,
      "loss": 2.2338,
      "step": 54422
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.6536182761192322,
      "learning_rate": 0.0005211215868307883,
      "loss": 2.0622,
      "step": 54423
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7533774375915527,
      "learning_rate": 0.0005211188223499329,
      "loss": 2.1233,
      "step": 54424
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7778816223144531,
      "learning_rate": 0.0005211160578279674,
      "loss": 2.0426,
      "step": 54425
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7500450015068054,
      "learning_rate": 0.0005211132932648922,
      "loss": 2.2048,
      "step": 54426
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7230959534645081,
      "learning_rate": 0.0005211105286607077,
      "loss": 2.1821,
      "step": 54427
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.8954291939735413,
      "learning_rate": 0.0005211077640154149,
      "loss": 2.3521,
      "step": 54428
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7323986291885376,
      "learning_rate": 0.0005211049993290138,
      "loss": 2.0974,
      "step": 54429
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.6966182589530945,
      "learning_rate": 0.0005211022346015052,
      "loss": 2.4302,
      "step": 54430
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7559080123901367,
      "learning_rate": 0.0005210994698328895,
      "loss": 2.0033,
      "step": 54431
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7928826808929443,
      "learning_rate": 0.0005210967050231671,
      "loss": 2.1812,
      "step": 54432
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.8203707337379456,
      "learning_rate": 0.0005210939401723389,
      "loss": 2.2459,
      "step": 54433
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7109631299972534,
      "learning_rate": 0.0005210911752804049,
      "loss": 2.1849,
      "step": 54434
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7806278467178345,
      "learning_rate": 0.0005210884103473661,
      "loss": 2.3404,
      "step": 54435
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.231417179107666,
      "learning_rate": 0.0005210856453732227,
      "loss": 2.1129,
      "step": 54436
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7241883277893066,
      "learning_rate": 0.0005210828803579752,
      "loss": 2.1352,
      "step": 54437
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.8480950593948364,
      "learning_rate": 0.0005210801153016244,
      "loss": 2.3144,
      "step": 54438
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.0708024501800537,
      "learning_rate": 0.0005210773502041705,
      "loss": 2.3534,
      "step": 54439
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.6991602778434753,
      "learning_rate": 0.0005210745850656142,
      "loss": 2.1616,
      "step": 54440
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.8085442781448364,
      "learning_rate": 0.000521071819885956,
      "loss": 2.4236,
      "step": 54441
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7370045185089111,
      "learning_rate": 0.0005210690546651963,
      "loss": 2.2914,
      "step": 54442
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7562001943588257,
      "learning_rate": 0.0005210662894033358,
      "loss": 2.3652,
      "step": 54443
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7255187630653381,
      "learning_rate": 0.0005210635241003747,
      "loss": 2.3571,
      "step": 54444
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.8546103239059448,
      "learning_rate": 0.0005210607587563139,
      "loss": 2.3315,
      "step": 54445
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.8018285632133484,
      "learning_rate": 0.0005210579933711537,
      "loss": 2.2076,
      "step": 54446
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7212913632392883,
      "learning_rate": 0.0005210552279448945,
      "loss": 2.1311,
      "step": 54447
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.6983672976493835,
      "learning_rate": 0.0005210524624775371,
      "loss": 2.2912,
      "step": 54448
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.8374878764152527,
      "learning_rate": 0.0005210496969690819,
      "loss": 2.2149,
      "step": 54449
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7416533827781677,
      "learning_rate": 0.0005210469314195293,
      "loss": 2.2901,
      "step": 54450
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7093379497528076,
      "learning_rate": 0.0005210441658288799,
      "loss": 2.4447,
      "step": 54451
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.6997559666633606,
      "learning_rate": 0.0005210414001971344,
      "loss": 2.0189,
      "step": 54452
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7444897294044495,
      "learning_rate": 0.000521038634524293,
      "loss": 2.0762,
      "step": 54453
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7250150442123413,
      "learning_rate": 0.0005210358688103563,
      "loss": 2.4089,
      "step": 54454
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7967159152030945,
      "learning_rate": 0.0005210331030553249,
      "loss": 2.1686,
      "step": 54455
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7199596762657166,
      "learning_rate": 0.0005210303372591993,
      "loss": 2.3036,
      "step": 54456
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.717450737953186,
      "learning_rate": 0.0005210275714219799,
      "loss": 2.1483,
      "step": 54457
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7150754332542419,
      "learning_rate": 0.0005210248055436674,
      "loss": 2.1461,
      "step": 54458
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.8859190344810486,
      "learning_rate": 0.0005210220396242621,
      "loss": 2.0898,
      "step": 54459
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.8711094260215759,
      "learning_rate": 0.0005210192736637648,
      "loss": 2.1877,
      "step": 54460
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.9241825938224792,
      "learning_rate": 0.0005210165076621758,
      "loss": 2.156,
      "step": 54461
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.773837149143219,
      "learning_rate": 0.0005210137416194955,
      "loss": 2.3421,
      "step": 54462
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.9112409949302673,
      "learning_rate": 0.0005210109755357247,
      "loss": 2.3144,
      "step": 54463
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.0548661947250366,
      "learning_rate": 0.0005210082094108638,
      "loss": 2.2473,
      "step": 54464
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7454598546028137,
      "learning_rate": 0.0005210054432449134,
      "loss": 2.2937,
      "step": 54465
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.713395893573761,
      "learning_rate": 0.0005210026770378737,
      "loss": 2.3848,
      "step": 54466
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.833512008190155,
      "learning_rate": 0.0005209999107897455,
      "loss": 2.2765,
      "step": 54467
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.757142961025238,
      "learning_rate": 0.0005209971445005292,
      "loss": 2.2675,
      "step": 54468
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7072998881340027,
      "learning_rate": 0.0005209943781702255,
      "loss": 2.4597,
      "step": 54469
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7711222767829895,
      "learning_rate": 0.0005209916117988347,
      "loss": 2.2998,
      "step": 54470
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7093541622161865,
      "learning_rate": 0.0005209888453863574,
      "loss": 2.1591,
      "step": 54471
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7198854088783264,
      "learning_rate": 0.0005209860789327941,
      "loss": 2.2027,
      "step": 54472
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7410308122634888,
      "learning_rate": 0.0005209833124381453,
      "loss": 2.3149,
      "step": 54473
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7897410988807678,
      "learning_rate": 0.0005209805459024116,
      "loss": 2.3241,
      "step": 54474
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.6780868768692017,
      "learning_rate": 0.0005209777793255934,
      "loss": 2.2887,
      "step": 54475
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.6916192173957825,
      "learning_rate": 0.0005209750127076911,
      "loss": 2.1307,
      "step": 54476
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.8290380835533142,
      "learning_rate": 0.0005209722460487056,
      "loss": 2.2261,
      "step": 54477
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7514812350273132,
      "learning_rate": 0.0005209694793486371,
      "loss": 2.4426,
      "step": 54478
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.6832162141799927,
      "learning_rate": 0.0005209667126074862,
      "loss": 2.2646,
      "step": 54479
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7023631930351257,
      "learning_rate": 0.0005209639458252535,
      "loss": 2.2675,
      "step": 54480
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.6784519553184509,
      "learning_rate": 0.0005209611790019393,
      "loss": 2.1156,
      "step": 54481
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.755664050579071,
      "learning_rate": 0.0005209584121375443,
      "loss": 2.3341,
      "step": 54482
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.6708166003227234,
      "learning_rate": 0.000520955645232069,
      "loss": 2.2284,
      "step": 54483
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7793132662773132,
      "learning_rate": 0.0005209528782855139,
      "loss": 2.5357,
      "step": 54484
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.674548327922821,
      "learning_rate": 0.0005209501112978794,
      "loss": 2.4121,
      "step": 54485
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7118746042251587,
      "learning_rate": 0.0005209473442691661,
      "loss": 2.3238,
      "step": 54486
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7099692225456238,
      "learning_rate": 0.0005209445771993745,
      "loss": 2.1657,
      "step": 54487
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7078058123588562,
      "learning_rate": 0.0005209418100885052,
      "loss": 2.3395,
      "step": 54488
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7712225317955017,
      "learning_rate": 0.0005209390429365587,
      "loss": 2.2181,
      "step": 54489
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7250187993049622,
      "learning_rate": 0.0005209362757435353,
      "loss": 2.2436,
      "step": 54490
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.6971189379692078,
      "learning_rate": 0.0005209335085094358,
      "loss": 2.1993,
      "step": 54491
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.6860039234161377,
      "learning_rate": 0.0005209307412342606,
      "loss": 2.141,
      "step": 54492
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.6823015213012695,
      "learning_rate": 0.0005209279739180102,
      "loss": 2.0277,
      "step": 54493
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7478351593017578,
      "learning_rate": 0.000520925206560685,
      "loss": 2.2437,
      "step": 54494
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7111197710037231,
      "learning_rate": 0.0005209224391622858,
      "loss": 2.1438,
      "step": 54495
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7178050875663757,
      "learning_rate": 0.0005209196717228128,
      "loss": 2.0503,
      "step": 54496
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.695444643497467,
      "learning_rate": 0.0005209169042422667,
      "loss": 2.198,
      "step": 54497
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.6982404589653015,
      "learning_rate": 0.000520914136720648,
      "loss": 2.3443,
      "step": 54498
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.6917518377304077,
      "learning_rate": 0.0005209113691579572,
      "loss": 2.3403,
      "step": 54499
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7026565074920654,
      "learning_rate": 0.0005209086015541947,
      "loss": 2.2625,
      "step": 54500
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7549282908439636,
      "learning_rate": 0.0005209058339093614,
      "loss": 2.3014,
      "step": 54501
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7416165471076965,
      "learning_rate": 0.0005209030662234572,
      "loss": 2.2305,
      "step": 54502
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7092714309692383,
      "learning_rate": 0.0005209002984964832,
      "loss": 2.4703,
      "step": 54503
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7630362510681152,
      "learning_rate": 0.0005208975307284396,
      "loss": 2.1588,
      "step": 54504
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.6719489097595215,
      "learning_rate": 0.0005208947629193269,
      "loss": 2.1371,
      "step": 54505
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.747704267501831,
      "learning_rate": 0.0005208919950691458,
      "loss": 2.2062,
      "step": 54506
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.8086187243461609,
      "learning_rate": 0.0005208892271778966,
      "loss": 2.263,
      "step": 54507
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7018625736236572,
      "learning_rate": 0.00052088645924558,
      "loss": 2.5654,
      "step": 54508
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7155201435089111,
      "learning_rate": 0.0005208836912721965,
      "loss": 2.3605,
      "step": 54509
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.6813996434211731,
      "learning_rate": 0.0005208809232577464,
      "loss": 2.1169,
      "step": 54510
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7850770950317383,
      "learning_rate": 0.0005208781552022304,
      "loss": 2.2284,
      "step": 54511
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7609333992004395,
      "learning_rate": 0.000520875387105649,
      "loss": 2.3625,
      "step": 54512
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7353646755218506,
      "learning_rate": 0.0005208726189680027,
      "loss": 2.2841,
      "step": 54513
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.8566979169845581,
      "learning_rate": 0.0005208698507892921,
      "loss": 2.4005,
      "step": 54514
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.8387618064880371,
      "learning_rate": 0.0005208670825695176,
      "loss": 2.2003,
      "step": 54515
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7194172739982605,
      "learning_rate": 0.0005208643143086797,
      "loss": 2.1483,
      "step": 54516
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7093796730041504,
      "learning_rate": 0.0005208615460067789,
      "loss": 2.3031,
      "step": 54517
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.8494135737419128,
      "learning_rate": 0.0005208587776638157,
      "loss": 2.0688,
      "step": 54518
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.0528134107589722,
      "learning_rate": 0.0005208560092797909,
      "loss": 1.8954,
      "step": 54519
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7129786610603333,
      "learning_rate": 0.0005208532408547047,
      "loss": 2.4509,
      "step": 54520
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.8114566802978516,
      "learning_rate": 0.0005208504723885576,
      "loss": 2.5167,
      "step": 54521
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.9965749979019165,
      "learning_rate": 0.0005208477038813505,
      "loss": 2.2273,
      "step": 54522
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7386618256568909,
      "learning_rate": 0.0005208449353330835,
      "loss": 2.0182,
      "step": 54523
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.6532712578773499,
      "learning_rate": 0.0005208421667437571,
      "loss": 2.1792,
      "step": 54524
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.9626495838165283,
      "learning_rate": 0.0005208393981133722,
      "loss": 2.1168,
      "step": 54525
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.8975042104721069,
      "learning_rate": 0.0005208366294419289,
      "loss": 2.1608,
      "step": 54526
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7367938160896301,
      "learning_rate": 0.0005208338607294281,
      "loss": 2.2257,
      "step": 54527
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.8085121512413025,
      "learning_rate": 0.00052083109197587,
      "loss": 2.2441,
      "step": 54528
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.1153390407562256,
      "learning_rate": 0.0005208283231812552,
      "loss": 2.3101,
      "step": 54529
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7277460694313049,
      "learning_rate": 0.0005208255543455843,
      "loss": 2.4433,
      "step": 54530
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.769978940486908,
      "learning_rate": 0.0005208227854688578,
      "loss": 2.3754,
      "step": 54531
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.9411984086036682,
      "learning_rate": 0.0005208200165510761,
      "loss": 2.2369,
      "step": 54532
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7243909239768982,
      "learning_rate": 0.0005208172475922398,
      "loss": 2.504,
      "step": 54533
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7167497873306274,
      "learning_rate": 0.0005208144785923494,
      "loss": 2.2787,
      "step": 54534
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.8063714504241943,
      "learning_rate": 0.0005208117095514055,
      "loss": 2.3051,
      "step": 54535
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7251594066619873,
      "learning_rate": 0.0005208089404694085,
      "loss": 2.2117,
      "step": 54536
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.8692400455474854,
      "learning_rate": 0.0005208061713463589,
      "loss": 2.3131,
      "step": 54537
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7114392518997192,
      "learning_rate": 0.0005208034021822574,
      "loss": 2.3046,
      "step": 54538
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7492955327033997,
      "learning_rate": 0.0005208006329771042,
      "loss": 2.2825,
      "step": 54539
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.8778336644172668,
      "learning_rate": 0.0005207978637309001,
      "loss": 2.1579,
      "step": 54540
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7202278971672058,
      "learning_rate": 0.0005207950944436455,
      "loss": 2.2151,
      "step": 54541
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.6973732113838196,
      "learning_rate": 0.0005207923251153409,
      "loss": 2.433,
      "step": 54542
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.8047664165496826,
      "learning_rate": 0.0005207895557459868,
      "loss": 2.0857,
      "step": 54543
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.9721575975418091,
      "learning_rate": 0.0005207867863355837,
      "loss": 2.0988,
      "step": 54544
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7782270312309265,
      "learning_rate": 0.0005207840168841324,
      "loss": 2.2007,
      "step": 54545
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.9634609818458557,
      "learning_rate": 0.000520781247391633,
      "loss": 2.1468,
      "step": 54546
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7740581631660461,
      "learning_rate": 0.0005207784778580863,
      "loss": 2.0822,
      "step": 54547
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7215479016304016,
      "learning_rate": 0.0005207757082834926,
      "loss": 2.1435,
      "step": 54548
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7166793942451477,
      "learning_rate": 0.0005207729386678526,
      "loss": 2.1542,
      "step": 54549
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.9949648380279541,
      "learning_rate": 0.0005207701690111668,
      "loss": 2.2174,
      "step": 54550
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7755703926086426,
      "learning_rate": 0.0005207673993134356,
      "loss": 2.2459,
      "step": 54551
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7217272520065308,
      "learning_rate": 0.0005207646295746595,
      "loss": 2.1054,
      "step": 54552
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7331854701042175,
      "learning_rate": 0.0005207618597948392,
      "loss": 2.4129,
      "step": 54553
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7873021960258484,
      "learning_rate": 0.000520759089973975,
      "loss": 2.454,
      "step": 54554
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.762852132320404,
      "learning_rate": 0.0005207563201120676,
      "loss": 2.4639,
      "step": 54555
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7453267574310303,
      "learning_rate": 0.0005207535502091174,
      "loss": 2.218,
      "step": 54556
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7426041960716248,
      "learning_rate": 0.000520750780265125,
      "loss": 2.334,
      "step": 54557
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.8077446818351746,
      "learning_rate": 0.0005207480102800908,
      "loss": 2.4092,
      "step": 54558
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7754030227661133,
      "learning_rate": 0.0005207452402540154,
      "loss": 2.3442,
      "step": 54559
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7145488262176514,
      "learning_rate": 0.0005207424701868993,
      "loss": 2.4046,
      "step": 54560
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7771093845367432,
      "learning_rate": 0.000520739700078743,
      "loss": 2.1851,
      "step": 54561
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.8341463208198547,
      "learning_rate": 0.0005207369299295469,
      "loss": 2.3852,
      "step": 54562
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.8359986543655396,
      "learning_rate": 0.0005207341597393119,
      "loss": 2.0646,
      "step": 54563
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.6992471814155579,
      "learning_rate": 0.0005207313895080381,
      "loss": 2.4601,
      "step": 54564
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7528351545333862,
      "learning_rate": 0.0005207286192357261,
      "loss": 2.2554,
      "step": 54565
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7321884632110596,
      "learning_rate": 0.0005207258489223766,
      "loss": 2.2089,
      "step": 54566
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7980674505233765,
      "learning_rate": 0.0005207230785679901,
      "loss": 1.9296,
      "step": 54567
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7256154417991638,
      "learning_rate": 0.0005207203081725668,
      "loss": 2.1732,
      "step": 54568
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7280336618423462,
      "learning_rate": 0.0005207175377361075,
      "loss": 2.5503,
      "step": 54569
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.1142743825912476,
      "learning_rate": 0.0005207147672586127,
      "loss": 2.1411,
      "step": 54570
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.9323050379753113,
      "learning_rate": 0.0005207119967400827,
      "loss": 2.4153,
      "step": 54571
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.9090996980667114,
      "learning_rate": 0.0005207092261805183,
      "loss": 2.3757,
      "step": 54572
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.9738038778305054,
      "learning_rate": 0.0005207064555799198,
      "loss": 2.1148,
      "step": 54573
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.0101356506347656,
      "learning_rate": 0.0005207036849382879,
      "loss": 2.3675,
      "step": 54574
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7159430384635925,
      "learning_rate": 0.000520700914255623,
      "loss": 2.045,
      "step": 54575
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.8279927968978882,
      "learning_rate": 0.0005206981435319257,
      "loss": 2.2538,
      "step": 54576
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.1454178094863892,
      "learning_rate": 0.0005206953727671963,
      "loss": 2.3904,
      "step": 54577
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.8031274676322937,
      "learning_rate": 0.0005206926019614356,
      "loss": 2.2816,
      "step": 54578
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.6761521697044373,
      "learning_rate": 0.0005206898311146439,
      "loss": 2.219,
      "step": 54579
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7844289541244507,
      "learning_rate": 0.0005206870602268217,
      "loss": 2.0299,
      "step": 54580
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.2603678703308105,
      "learning_rate": 0.0005206842892979698,
      "loss": 2.2168,
      "step": 54581
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.8338084816932678,
      "learning_rate": 0.0005206815183280885,
      "loss": 2.1978,
      "step": 54582
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.6980464458465576,
      "learning_rate": 0.0005206787473171782,
      "loss": 2.2015,
      "step": 54583
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.1185047626495361,
      "learning_rate": 0.0005206759762652397,
      "loss": 2.257,
      "step": 54584
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.018863320350647,
      "learning_rate": 0.0005206732051722733,
      "loss": 2.2166,
      "step": 54585
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7863925695419312,
      "learning_rate": 0.0005206704340382796,
      "loss": 2.3073,
      "step": 54586
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7498152256011963,
      "learning_rate": 0.0005206676628632591,
      "loss": 2.3068,
      "step": 54587
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.2440972328186035,
      "learning_rate": 0.0005206648916472122,
      "loss": 2.1335,
      "step": 54588
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.0297189950942993,
      "learning_rate": 0.0005206621203901398,
      "loss": 2.2096,
      "step": 54589
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.8636597394943237,
      "learning_rate": 0.0005206593490920419,
      "loss": 2.1732,
      "step": 54590
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7853901982307434,
      "learning_rate": 0.0005206565777529194,
      "loss": 2.3475,
      "step": 54591
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.8379283547401428,
      "learning_rate": 0.0005206538063727725,
      "loss": 2.4766,
      "step": 54592
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.9274119734764099,
      "learning_rate": 0.0005206510349516022,
      "loss": 2.3745,
      "step": 54593
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.9924719333648682,
      "learning_rate": 0.0005206482634894084,
      "loss": 2.2267,
      "step": 54594
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.8056202530860901,
      "learning_rate": 0.0005206454919861922,
      "loss": 2.4665,
      "step": 54595
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.909379780292511,
      "learning_rate": 0.0005206427204419536,
      "loss": 2.4554,
      "step": 54596
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.8730809688568115,
      "learning_rate": 0.0005206399488566935,
      "loss": 2.2271,
      "step": 54597
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.0539851188659668,
      "learning_rate": 0.0005206371772304122,
      "loss": 2.0364,
      "step": 54598
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7025947570800781,
      "learning_rate": 0.0005206344055631103,
      "loss": 2.3272,
      "step": 54599
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.6706569194793701,
      "learning_rate": 0.0005206316338547882,
      "loss": 2.1163,
      "step": 54600
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.3230654001235962,
      "learning_rate": 0.0005206288621054467,
      "loss": 2.1726,
      "step": 54601
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.0205150842666626,
      "learning_rate": 0.000520626090315086,
      "loss": 2.19,
      "step": 54602
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7984985709190369,
      "learning_rate": 0.0005206233184837068,
      "loss": 2.2665,
      "step": 54603
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7426741123199463,
      "learning_rate": 0.0005206205466113096,
      "loss": 2.2161,
      "step": 54604
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.8945054411888123,
      "learning_rate": 0.0005206177746978948,
      "loss": 2.1838,
      "step": 54605
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.0298875570297241,
      "learning_rate": 0.0005206150027434629,
      "loss": 2.2374,
      "step": 54606
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7253843545913696,
      "learning_rate": 0.0005206122307480147,
      "loss": 1.8522,
      "step": 54607
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7401514053344727,
      "learning_rate": 0.0005206094587115504,
      "loss": 2.2383,
      "step": 54608
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.925269603729248,
      "learning_rate": 0.0005206066866340706,
      "loss": 2.144,
      "step": 54609
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.8996704816818237,
      "learning_rate": 0.000520603914515576,
      "loss": 2.2014,
      "step": 54610
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7939762473106384,
      "learning_rate": 0.000520601142356067,
      "loss": 2.2323,
      "step": 54611
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7741219401359558,
      "learning_rate": 0.000520598370155544,
      "loss": 2.2252,
      "step": 54612
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.929097592830658,
      "learning_rate": 0.0005205955979140075,
      "loss": 2.2084,
      "step": 54613
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7258690595626831,
      "learning_rate": 0.0005205928256314582,
      "loss": 2.2103,
      "step": 54614
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7368597984313965,
      "learning_rate": 0.0005205900533078964,
      "loss": 2.2275,
      "step": 54615
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.722571849822998,
      "learning_rate": 0.000520587280943323,
      "loss": 2.2259,
      "step": 54616
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7529920935630798,
      "learning_rate": 0.0005205845085377381,
      "loss": 2.1464,
      "step": 54617
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7894924283027649,
      "learning_rate": 0.0005205817360911425,
      "loss": 2.4429,
      "step": 54618
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.6982550621032715,
      "learning_rate": 0.0005205789636035364,
      "loss": 2.1439,
      "step": 54619
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7458515167236328,
      "learning_rate": 0.0005205761910749207,
      "loss": 2.141,
      "step": 54620
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7801596522331238,
      "learning_rate": 0.0005205734185052955,
      "loss": 2.2853,
      "step": 54621
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7386156916618347,
      "learning_rate": 0.0005205706458946617,
      "loss": 2.3464,
      "step": 54622
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.6690382957458496,
      "learning_rate": 0.0005205678732430197,
      "loss": 2.3303,
      "step": 54623
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.76626056432724,
      "learning_rate": 0.0005205651005503699,
      "loss": 2.3862,
      "step": 54624
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7941516041755676,
      "learning_rate": 0.0005205623278167128,
      "loss": 2.5057,
      "step": 54625
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7523376941680908,
      "learning_rate": 0.0005205595550420491,
      "loss": 2.3586,
      "step": 54626
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.6777973771095276,
      "learning_rate": 0.0005205567822263792,
      "loss": 2.2268,
      "step": 54627
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7898643016815186,
      "learning_rate": 0.0005205540093697036,
      "loss": 2.4753,
      "step": 54628
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.8061432242393494,
      "learning_rate": 0.0005205512364720229,
      "loss": 2.3293,
      "step": 54629
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.79026859998703,
      "learning_rate": 0.0005205484635333375,
      "loss": 2.2663,
      "step": 54630
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7302126884460449,
      "learning_rate": 0.000520545690553648,
      "loss": 2.0639,
      "step": 54631
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7703254818916321,
      "learning_rate": 0.0005205429175329549,
      "loss": 2.3997,
      "step": 54632
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7336721420288086,
      "learning_rate": 0.0005205401444712587,
      "loss": 2.2707,
      "step": 54633
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.8227524161338806,
      "learning_rate": 0.0005205373713685599,
      "loss": 2.2517,
      "step": 54634
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7720363736152649,
      "learning_rate": 0.0005205345982248591,
      "loss": 2.0937,
      "step": 54635
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.68877774477005,
      "learning_rate": 0.0005205318250401567,
      "loss": 2.4627,
      "step": 54636
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.795950174331665,
      "learning_rate": 0.0005205290518144533,
      "loss": 2.2851,
      "step": 54637
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7457318305969238,
      "learning_rate": 0.0005205262785477493,
      "loss": 2.3669,
      "step": 54638
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7194553017616272,
      "learning_rate": 0.0005205235052400455,
      "loss": 1.9636,
      "step": 54639
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.9587432742118835,
      "learning_rate": 0.0005205207318913421,
      "loss": 2.3601,
      "step": 54640
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7892850637435913,
      "learning_rate": 0.0005205179585016397,
      "loss": 2.2621,
      "step": 54641
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.6990891695022583,
      "learning_rate": 0.0005205151850709389,
      "loss": 2.4306,
      "step": 54642
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7410947680473328,
      "learning_rate": 0.0005205124115992401,
      "loss": 2.1204,
      "step": 54643
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7423993945121765,
      "learning_rate": 0.0005205096380865439,
      "loss": 2.3017,
      "step": 54644
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7770807147026062,
      "learning_rate": 0.0005205068645328509,
      "loss": 2.2371,
      "step": 54645
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7125272750854492,
      "learning_rate": 0.0005205040909381615,
      "loss": 2.2075,
      "step": 54646
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7681265473365784,
      "learning_rate": 0.0005205013173024761,
      "loss": 2.2836,
      "step": 54647
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7090849876403809,
      "learning_rate": 0.0005204985436257953,
      "loss": 2.4482,
      "step": 54648
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.8465694785118103,
      "learning_rate": 0.0005204957699081198,
      "loss": 2.3369,
      "step": 54649
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.6931545734405518,
      "learning_rate": 0.00052049299614945,
      "loss": 2.2834,
      "step": 54650
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7251039743423462,
      "learning_rate": 0.0005204902223497863,
      "loss": 2.2957,
      "step": 54651
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7630541920661926,
      "learning_rate": 0.0005204874485091294,
      "loss": 2.1949,
      "step": 54652
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7380825877189636,
      "learning_rate": 0.0005204846746274796,
      "loss": 2.3222,
      "step": 54653
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7098573446273804,
      "learning_rate": 0.0005204819007048376,
      "loss": 2.2741,
      "step": 54654
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7326322793960571,
      "learning_rate": 0.0005204791267412038,
      "loss": 2.2482,
      "step": 54655
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.685227632522583,
      "learning_rate": 0.000520476352736579,
      "loss": 2.2519,
      "step": 54656
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7337247133255005,
      "learning_rate": 0.0005204735786909631,
      "loss": 2.2538,
      "step": 54657
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7621970176696777,
      "learning_rate": 0.0005204708046043573,
      "loss": 2.2099,
      "step": 54658
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.8492852449417114,
      "learning_rate": 0.0005204680304767617,
      "loss": 2.2744,
      "step": 54659
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7364476919174194,
      "learning_rate": 0.000520465256308177,
      "loss": 2.2868,
      "step": 54660
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7930787801742554,
      "learning_rate": 0.0005204624820986035,
      "loss": 2.3191,
      "step": 54661
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7858138680458069,
      "learning_rate": 0.0005204597078480419,
      "loss": 2.1918,
      "step": 54662
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7996617555618286,
      "learning_rate": 0.0005204569335564928,
      "loss": 2.0191,
      "step": 54663
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7019906044006348,
      "learning_rate": 0.0005204541592239566,
      "loss": 2.0497,
      "step": 54664
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.6914121508598328,
      "learning_rate": 0.0005204513848504336,
      "loss": 2.2497,
      "step": 54665
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.8553176522254944,
      "learning_rate": 0.0005204486104359248,
      "loss": 2.1268,
      "step": 54666
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.6895803809165955,
      "learning_rate": 0.0005204458359804302,
      "loss": 2.2977,
      "step": 54667
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7833347916603088,
      "learning_rate": 0.0005204430614839507,
      "loss": 2.1715,
      "step": 54668
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7547076344490051,
      "learning_rate": 0.0005204402869464866,
      "loss": 2.1844,
      "step": 54669
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.8784601092338562,
      "learning_rate": 0.0005204375123680385,
      "loss": 2.0131,
      "step": 54670
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7167667746543884,
      "learning_rate": 0.0005204347377486071,
      "loss": 2.2692,
      "step": 54671
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7473158240318298,
      "learning_rate": 0.0005204319630881925,
      "loss": 2.2526,
      "step": 54672
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7204058766365051,
      "learning_rate": 0.0005204291883867955,
      "loss": 2.1681,
      "step": 54673
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.6786584854125977,
      "learning_rate": 0.0005204264136444165,
      "loss": 2.0135,
      "step": 54674
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7792249917984009,
      "learning_rate": 0.0005204236388610562,
      "loss": 2.0748,
      "step": 54675
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7367514967918396,
      "learning_rate": 0.0005204208640367149,
      "loss": 2.1986,
      "step": 54676
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.6960121393203735,
      "learning_rate": 0.0005204180891713931,
      "loss": 2.3609,
      "step": 54677
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.9066020250320435,
      "learning_rate": 0.0005204153142650916,
      "loss": 2.264,
      "step": 54678
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.6970788240432739,
      "learning_rate": 0.0005204125393178105,
      "loss": 2.1314,
      "step": 54679
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7436447739601135,
      "learning_rate": 0.0005204097643295508,
      "loss": 2.4258,
      "step": 54680
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7122329473495483,
      "learning_rate": 0.0005204069893003126,
      "loss": 2.446,
      "step": 54681
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.6921300292015076,
      "learning_rate": 0.0005204042142300967,
      "loss": 2.16,
      "step": 54682
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.088585615158081,
      "learning_rate": 0.0005204014391189034,
      "loss": 2.128,
      "step": 54683
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7006627321243286,
      "learning_rate": 0.0005203986639667334,
      "loss": 2.0442,
      "step": 54684
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7272399067878723,
      "learning_rate": 0.0005203958887735871,
      "loss": 2.2353,
      "step": 54685
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7339345812797546,
      "learning_rate": 0.000520393113539465,
      "loss": 2.2305,
      "step": 54686
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.74074786901474,
      "learning_rate": 0.0005203903382643677,
      "loss": 2.3557,
      "step": 54687
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.8745547533035278,
      "learning_rate": 0.0005203875629482956,
      "loss": 2.1178,
      "step": 54688
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.6832783222198486,
      "learning_rate": 0.0005203847875912494,
      "loss": 2.2283,
      "step": 54689
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7105496525764465,
      "learning_rate": 0.0005203820121932294,
      "loss": 2.5275,
      "step": 54690
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.681718111038208,
      "learning_rate": 0.0005203792367542363,
      "loss": 2.1318,
      "step": 54691
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.6534314155578613,
      "learning_rate": 0.0005203764612742705,
      "loss": 2.2415,
      "step": 54692
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7025477290153503,
      "learning_rate": 0.0005203736857533326,
      "loss": 2.2087,
      "step": 54693
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.6453639268875122,
      "learning_rate": 0.000520370910191423,
      "loss": 2.4597,
      "step": 54694
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7180684208869934,
      "learning_rate": 0.0005203681345885425,
      "loss": 2.3138,
      "step": 54695
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7344843745231628,
      "learning_rate": 0.0005203653589446911,
      "loss": 2.1605,
      "step": 54696
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7428773045539856,
      "learning_rate": 0.0005203625832598698,
      "loss": 1.9841,
      "step": 54697
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7161601781845093,
      "learning_rate": 0.0005203598075340789,
      "loss": 2.4263,
      "step": 54698
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.8198583722114563,
      "learning_rate": 0.000520357031767319,
      "loss": 2.2423,
      "step": 54699
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.6909875869750977,
      "learning_rate": 0.0005203542559595904,
      "loss": 2.2325,
      "step": 54700
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.8492083549499512,
      "learning_rate": 0.000520351480110894,
      "loss": 2.3632,
      "step": 54701
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7220548391342163,
      "learning_rate": 0.00052034870422123,
      "loss": 2.4903,
      "step": 54702
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7564824223518372,
      "learning_rate": 0.000520345928290599,
      "loss": 2.3324,
      "step": 54703
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.8836866021156311,
      "learning_rate": 0.0005203431523190015,
      "loss": 2.4148,
      "step": 54704
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7189057469367981,
      "learning_rate": 0.000520340376306438,
      "loss": 2.2085,
      "step": 54705
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.6933208703994751,
      "learning_rate": 0.0005203376002529093,
      "loss": 1.9243,
      "step": 54706
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.793334424495697,
      "learning_rate": 0.0005203348241584156,
      "loss": 2.1724,
      "step": 54707
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.8537690043449402,
      "learning_rate": 0.0005203320480229573,
      "loss": 2.4699,
      "step": 54708
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7218706011772156,
      "learning_rate": 0.0005203292718465353,
      "loss": 2.2879,
      "step": 54709
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7702301740646362,
      "learning_rate": 0.00052032649562915,
      "loss": 2.1456,
      "step": 54710
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7835968732833862,
      "learning_rate": 0.0005203237193708017,
      "loss": 2.3161,
      "step": 54711
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7347925305366516,
      "learning_rate": 0.000520320943071491,
      "loss": 2.1624,
      "step": 54712
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7101624608039856,
      "learning_rate": 0.0005203181667312186,
      "loss": 2.228,
      "step": 54713
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.876585841178894,
      "learning_rate": 0.0005203153903499849,
      "loss": 2.3955,
      "step": 54714
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.6753503084182739,
      "learning_rate": 0.0005203126139277903,
      "loss": 2.3275,
      "step": 54715
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7509071230888367,
      "learning_rate": 0.0005203098374646356,
      "loss": 2.4094,
      "step": 54716
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7343906164169312,
      "learning_rate": 0.000520307060960521,
      "loss": 2.2178,
      "step": 54717
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7304323315620422,
      "learning_rate": 0.0005203042844154471,
      "loss": 2.3938,
      "step": 54718
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7498247027397156,
      "learning_rate": 0.0005203015078294146,
      "loss": 2.1183,
      "step": 54719
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7879650592803955,
      "learning_rate": 0.0005202987312024239,
      "loss": 2.3927,
      "step": 54720
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7711196541786194,
      "learning_rate": 0.0005202959545344755,
      "loss": 2.0272,
      "step": 54721
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7724257111549377,
      "learning_rate": 0.0005202931778255699,
      "loss": 2.2576,
      "step": 54722
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.6723586916923523,
      "learning_rate": 0.0005202904010757077,
      "loss": 2.1763,
      "step": 54723
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7294783592224121,
      "learning_rate": 0.0005202876242848892,
      "loss": 2.3759,
      "step": 54724
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7345781922340393,
      "learning_rate": 0.0005202848474531153,
      "loss": 2.2912,
      "step": 54725
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7615470290184021,
      "learning_rate": 0.000520282070580386,
      "loss": 2.4607,
      "step": 54726
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7124562859535217,
      "learning_rate": 0.0005202792936667024,
      "loss": 2.2299,
      "step": 54727
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.6627992391586304,
      "learning_rate": 0.0005202765167120645,
      "loss": 2.0945,
      "step": 54728
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7789299488067627,
      "learning_rate": 0.0005202737397164732,
      "loss": 2.3141,
      "step": 54729
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7454447150230408,
      "learning_rate": 0.0005202709626799287,
      "loss": 2.2799,
      "step": 54730
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7368373274803162,
      "learning_rate": 0.0005202681856024318,
      "loss": 2.2249,
      "step": 54731
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7300331592559814,
      "learning_rate": 0.0005202654084839828,
      "loss": 2.1076,
      "step": 54732
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7257432341575623,
      "learning_rate": 0.0005202626313245823,
      "loss": 2.1926,
      "step": 54733
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7284923791885376,
      "learning_rate": 0.0005202598541242309,
      "loss": 2.3881,
      "step": 54734
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7168121933937073,
      "learning_rate": 0.0005202570768829289,
      "loss": 2.1948,
      "step": 54735
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.6853665709495544,
      "learning_rate": 0.0005202542996006771,
      "loss": 2.2451,
      "step": 54736
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7181962132453918,
      "learning_rate": 0.0005202515222774758,
      "loss": 2.3268,
      "step": 54737
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7180435061454773,
      "learning_rate": 0.0005202487449133255,
      "loss": 2.2633,
      "step": 54738
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7377475500106812,
      "learning_rate": 0.000520245967508227,
      "loss": 2.1165,
      "step": 54739
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7430475354194641,
      "learning_rate": 0.0005202431900621805,
      "loss": 2.3299,
      "step": 54740
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7694647312164307,
      "learning_rate": 0.0005202404125751866,
      "loss": 2.2263,
      "step": 54741
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.8008551597595215,
      "learning_rate": 0.0005202376350472459,
      "loss": 2.3696,
      "step": 54742
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.8266755938529968,
      "learning_rate": 0.0005202348574783589,
      "loss": 2.1489,
      "step": 54743
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.772191047668457,
      "learning_rate": 0.000520232079868526,
      "loss": 2.2771,
      "step": 54744
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.6955043077468872,
      "learning_rate": 0.0005202293022177479,
      "loss": 2.0785,
      "step": 54745
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.9319526553153992,
      "learning_rate": 0.000520226524526025,
      "loss": 2.2846,
      "step": 54746
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.8645871877670288,
      "learning_rate": 0.0005202237467933577,
      "loss": 2.2338,
      "step": 54747
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.721666693687439,
      "learning_rate": 0.0005202209690197468,
      "loss": 2.254,
      "step": 54748
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.8067103624343872,
      "learning_rate": 0.0005202181912051925,
      "loss": 2.0531,
      "step": 54749
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.8226155042648315,
      "learning_rate": 0.0005202154133496957,
      "loss": 2.4197,
      "step": 54750
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7125126123428345,
      "learning_rate": 0.0005202126354532566,
      "loss": 2.2052,
      "step": 54751
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.697627604007721,
      "learning_rate": 0.0005202098575158757,
      "loss": 2.0971,
      "step": 54752
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7621757388114929,
      "learning_rate": 0.0005202070795375537,
      "loss": 2.2175,
      "step": 54753
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7856597900390625,
      "learning_rate": 0.0005202043015182912,
      "loss": 2.4054,
      "step": 54754
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.701697826385498,
      "learning_rate": 0.0005202015234580884,
      "loss": 2.2316,
      "step": 54755
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7274788618087769,
      "learning_rate": 0.000520198745356946,
      "loss": 2.2124,
      "step": 54756
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7381240725517273,
      "learning_rate": 0.0005201959672148645,
      "loss": 2.2078,
      "step": 54757
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7464816570281982,
      "learning_rate": 0.0005201931890318444,
      "loss": 2.1766,
      "step": 54758
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.6825745105743408,
      "learning_rate": 0.0005201904108078862,
      "loss": 2.0671,
      "step": 54759
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7271268367767334,
      "learning_rate": 0.0005201876325429906,
      "loss": 2.0758,
      "step": 54760
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.724185049533844,
      "learning_rate": 0.0005201848542371578,
      "loss": 2.2371,
      "step": 54761
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.8566240668296814,
      "learning_rate": 0.0005201820758903886,
      "loss": 2.2312,
      "step": 54762
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.8196306228637695,
      "learning_rate": 0.0005201792975026832,
      "loss": 2.3203,
      "step": 54763
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7172354459762573,
      "learning_rate": 0.0005201765190740427,
      "loss": 2.2234,
      "step": 54764
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7855591773986816,
      "learning_rate": 0.0005201737406044668,
      "loss": 1.9177,
      "step": 54765
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7186427712440491,
      "learning_rate": 0.0005201709620939566,
      "loss": 2.2693,
      "step": 54766
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7477789521217346,
      "learning_rate": 0.0005201681835425126,
      "loss": 2.3399,
      "step": 54767
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7235129475593567,
      "learning_rate": 0.000520165404950135,
      "loss": 2.235,
      "step": 54768
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.8203492760658264,
      "learning_rate": 0.0005201626263168246,
      "loss": 2.3569,
      "step": 54769
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7262988090515137,
      "learning_rate": 0.0005201598476425818,
      "loss": 2.1884,
      "step": 54770
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7303964495658875,
      "learning_rate": 0.000520157068927407,
      "loss": 2.2214,
      "step": 54771
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7289920449256897,
      "learning_rate": 0.000520154290171301,
      "loss": 2.2121,
      "step": 54772
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7128375172615051,
      "learning_rate": 0.0005201515113742642,
      "loss": 2.222,
      "step": 54773
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.8392212986946106,
      "learning_rate": 0.0005201487325362969,
      "loss": 2.3764,
      "step": 54774
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7087881565093994,
      "learning_rate": 0.0005201459536573999,
      "loss": 1.9943,
      "step": 54775
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7251717448234558,
      "learning_rate": 0.0005201431747375736,
      "loss": 2.0057,
      "step": 54776
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7404573559761047,
      "learning_rate": 0.0005201403957768185,
      "loss": 2.3043,
      "step": 54777
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7666503190994263,
      "learning_rate": 0.0005201376167751351,
      "loss": 2.1013,
      "step": 54778
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.8019543290138245,
      "learning_rate": 0.0005201348377325241,
      "loss": 2.0476,
      "step": 54779
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7675442099571228,
      "learning_rate": 0.0005201320586489858,
      "loss": 2.417,
      "step": 54780
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.8063409328460693,
      "learning_rate": 0.0005201292795245209,
      "loss": 2.3211,
      "step": 54781
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.720052182674408,
      "learning_rate": 0.0005201265003591297,
      "loss": 2.2069,
      "step": 54782
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.744720995426178,
      "learning_rate": 0.0005201237211528128,
      "loss": 1.9863,
      "step": 54783
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.76146399974823,
      "learning_rate": 0.0005201209419055708,
      "loss": 2.2172,
      "step": 54784
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.8316112756729126,
      "learning_rate": 0.0005201181626174041,
      "loss": 2.3305,
      "step": 54785
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.6842141151428223,
      "learning_rate": 0.0005201153832883134,
      "loss": 2.211,
      "step": 54786
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.717353105545044,
      "learning_rate": 0.000520112603918299,
      "loss": 2.3554,
      "step": 54787
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7702008485794067,
      "learning_rate": 0.0005201098245073615,
      "loss": 2.2843,
      "step": 54788
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.6723544597625732,
      "learning_rate": 0.0005201070450555014,
      "loss": 2.281,
      "step": 54789
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7202059626579285,
      "learning_rate": 0.0005201042655627193,
      "loss": 2.2915,
      "step": 54790
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7241100668907166,
      "learning_rate": 0.0005201014860290158,
      "loss": 2.1827,
      "step": 54791
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7219105958938599,
      "learning_rate": 0.0005200987064543912,
      "loss": 2.2402,
      "step": 54792
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7213544249534607,
      "learning_rate": 0.000520095926838846,
      "loss": 2.0823,
      "step": 54793
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7182695865631104,
      "learning_rate": 0.0005200931471823809,
      "loss": 2.243,
      "step": 54794
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.6795761585235596,
      "learning_rate": 0.0005200903674849963,
      "loss": 2.4404,
      "step": 54795
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.8777565360069275,
      "learning_rate": 0.0005200875877466927,
      "loss": 2.1737,
      "step": 54796
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7871388792991638,
      "learning_rate": 0.0005200848079674707,
      "loss": 2.2835,
      "step": 54797
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7309432625770569,
      "learning_rate": 0.0005200820281473308,
      "loss": 2.4019,
      "step": 54798
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7060543894767761,
      "learning_rate": 0.0005200792482862734,
      "loss": 2.3438,
      "step": 54799
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7584033012390137,
      "learning_rate": 0.0005200764683842992,
      "loss": 2.2167,
      "step": 54800
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7228608727455139,
      "learning_rate": 0.0005200736884414086,
      "loss": 2.3494,
      "step": 54801
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7101322412490845,
      "learning_rate": 0.0005200709084576022,
      "loss": 2.4428,
      "step": 54802
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7283852696418762,
      "learning_rate": 0.0005200681284328805,
      "loss": 2.4241,
      "step": 54803
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7465552687644958,
      "learning_rate": 0.0005200653483672438,
      "loss": 1.9946,
      "step": 54804
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7170266509056091,
      "learning_rate": 0.0005200625682606929,
      "loss": 2.1009,
      "step": 54805
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7287813425064087,
      "learning_rate": 0.0005200597881132282,
      "loss": 2.4255,
      "step": 54806
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7588443756103516,
      "learning_rate": 0.0005200570079248503,
      "loss": 2.1483,
      "step": 54807
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7087242007255554,
      "learning_rate": 0.0005200542276955597,
      "loss": 2.2945,
      "step": 54808
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7926048040390015,
      "learning_rate": 0.0005200514474253566,
      "loss": 2.2214,
      "step": 54809
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7785360217094421,
      "learning_rate": 0.0005200486671142421,
      "loss": 2.2881,
      "step": 54810
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.780574381351471,
      "learning_rate": 0.0005200458867622162,
      "loss": 2.2695,
      "step": 54811
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7902163863182068,
      "learning_rate": 0.0005200431063692797,
      "loss": 2.2131,
      "step": 54812
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.6717960238456726,
      "learning_rate": 0.000520040325935433,
      "loss": 2.2785,
      "step": 54813
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.8399602770805359,
      "learning_rate": 0.0005200375454606766,
      "loss": 2.238,
      "step": 54814
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7140203714370728,
      "learning_rate": 0.0005200347649450111,
      "loss": 2.0928,
      "step": 54815
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7491010427474976,
      "learning_rate": 0.000520031984388437,
      "loss": 2.3098,
      "step": 54816
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7224212884902954,
      "learning_rate": 0.0005200292037909548,
      "loss": 2.2753,
      "step": 54817
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.6779347062110901,
      "learning_rate": 0.000520026423152565,
      "loss": 2.0671,
      "step": 54818
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.6302884221076965,
      "learning_rate": 0.0005200236424732682,
      "loss": 2.097,
      "step": 54819
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.6871930956840515,
      "learning_rate": 0.0005200208617530648,
      "loss": 2.3936,
      "step": 54820
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.6993392705917358,
      "learning_rate": 0.0005200180809919554,
      "loss": 2.3638,
      "step": 54821
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7848026156425476,
      "learning_rate": 0.0005200153001899405,
      "loss": 2.1706,
      "step": 54822
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7314448952674866,
      "learning_rate": 0.0005200125193470205,
      "loss": 2.1872,
      "step": 54823
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.780580461025238,
      "learning_rate": 0.0005200097384631961,
      "loss": 2.2258,
      "step": 54824
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7166613340377808,
      "learning_rate": 0.0005200069575384676,
      "loss": 2.4296,
      "step": 54825
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.6901936531066895,
      "learning_rate": 0.0005200041765728358,
      "loss": 2.089,
      "step": 54826
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7526552081108093,
      "learning_rate": 0.0005200013955663011,
      "loss": 2.3324,
      "step": 54827
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.752391517162323,
      "learning_rate": 0.0005199986145188639,
      "loss": 2.0587,
      "step": 54828
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.8128655552864075,
      "learning_rate": 0.0005199958334305248,
      "loss": 2.3281,
      "step": 54829
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.6963410377502441,
      "learning_rate": 0.0005199930523012845,
      "loss": 2.2777,
      "step": 54830
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.6744351983070374,
      "learning_rate": 0.000519990271131143,
      "loss": 2.2666,
      "step": 54831
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.6922268271446228,
      "learning_rate": 0.0005199874899201014,
      "loss": 2.1643,
      "step": 54832
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7803450226783752,
      "learning_rate": 0.00051998470866816,
      "loss": 2.1053,
      "step": 54833
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7688170075416565,
      "learning_rate": 0.0005199819273753192,
      "loss": 2.3456,
      "step": 54834
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.8070250153541565,
      "learning_rate": 0.0005199791460415796,
      "loss": 2.2546,
      "step": 54835
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.689120888710022,
      "learning_rate": 0.0005199763646669417,
      "loss": 2.0763,
      "step": 54836
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.8778627514839172,
      "learning_rate": 0.0005199735832514061,
      "loss": 2.1727,
      "step": 54837
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7360043525695801,
      "learning_rate": 0.0005199708017949732,
      "loss": 2.1208,
      "step": 54838
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.8342877626419067,
      "learning_rate": 0.0005199680202976438,
      "loss": 2.3234,
      "step": 54839
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.6866102814674377,
      "learning_rate": 0.0005199652387594179,
      "loss": 2.1951,
      "step": 54840
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7758791446685791,
      "learning_rate": 0.0005199624571802964,
      "loss": 2.1615,
      "step": 54841
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.8004281520843506,
      "learning_rate": 0.0005199596755602798,
      "loss": 2.3584,
      "step": 54842
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7324379682540894,
      "learning_rate": 0.0005199568938993685,
      "loss": 2.3244,
      "step": 54843
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7546104192733765,
      "learning_rate": 0.0005199541121975631,
      "loss": 2.1183,
      "step": 54844
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7687305808067322,
      "learning_rate": 0.0005199513304548639,
      "loss": 2.3858,
      "step": 54845
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.8059172630310059,
      "learning_rate": 0.0005199485486712718,
      "loss": 2.2424,
      "step": 54846
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7014998197555542,
      "learning_rate": 0.000519945766846787,
      "loss": 2.5282,
      "step": 54847
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7254423499107361,
      "learning_rate": 0.0005199429849814101,
      "loss": 2.0947,
      "step": 54848
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.8471259474754333,
      "learning_rate": 0.0005199402030751419,
      "loss": 2.1335,
      "step": 54849
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7350267171859741,
      "learning_rate": 0.0005199374211279825,
      "loss": 2.2234,
      "step": 54850
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.6865625381469727,
      "learning_rate": 0.0005199346391399325,
      "loss": 2.1182,
      "step": 54851
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7759149074554443,
      "learning_rate": 0.0005199318571109925,
      "loss": 2.1941,
      "step": 54852
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.738741934299469,
      "learning_rate": 0.0005199290750411631,
      "loss": 2.185,
      "step": 54853
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7964637875556946,
      "learning_rate": 0.0005199262929304447,
      "loss": 2.1578,
      "step": 54854
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.6517860293388367,
      "learning_rate": 0.0005199235107788379,
      "loss": 2.212,
      "step": 54855
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7191950082778931,
      "learning_rate": 0.0005199207285863431,
      "loss": 2.2443,
      "step": 54856
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.8712280988693237,
      "learning_rate": 0.000519917946352961,
      "loss": 2.2521,
      "step": 54857
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7097913026809692,
      "learning_rate": 0.000519915164078692,
      "loss": 2.2592,
      "step": 54858
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.8203076720237732,
      "learning_rate": 0.0005199123817635365,
      "loss": 2.2821,
      "step": 54859
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7378079295158386,
      "learning_rate": 0.000519909599407495,
      "loss": 2.5589,
      "step": 54860
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7150458097457886,
      "learning_rate": 0.0005199068170105685,
      "loss": 2.4011,
      "step": 54861
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7266948819160461,
      "learning_rate": 0.0005199040345727569,
      "loss": 2.2786,
      "step": 54862
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.6752908229827881,
      "learning_rate": 0.000519901252094061,
      "loss": 2.1754,
      "step": 54863
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.6658666729927063,
      "learning_rate": 0.0005198984695744815,
      "loss": 2.2629,
      "step": 54864
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7694081664085388,
      "learning_rate": 0.0005198956870140187,
      "loss": 1.9851,
      "step": 54865
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.721923828125,
      "learning_rate": 0.000519892904412673,
      "loss": 2.4019,
      "step": 54866
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.792446494102478,
      "learning_rate": 0.0005198901217704451,
      "loss": 2.0251,
      "step": 54867
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7757455110549927,
      "learning_rate": 0.0005198873390873355,
      "loss": 2.3005,
      "step": 54868
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.8143860697746277,
      "learning_rate": 0.0005198845563633447,
      "loss": 2.3638,
      "step": 54869
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7222692966461182,
      "learning_rate": 0.0005198817735984733,
      "loss": 2.3045,
      "step": 54870
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.8470541834831238,
      "learning_rate": 0.0005198789907927216,
      "loss": 2.153,
      "step": 54871
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.6896193027496338,
      "learning_rate": 0.0005198762079460902,
      "loss": 2.2903,
      "step": 54872
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7383233308792114,
      "learning_rate": 0.0005198734250585797,
      "loss": 2.2127,
      "step": 54873
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7290623784065247,
      "learning_rate": 0.0005198706421301907,
      "loss": 2.3335,
      "step": 54874
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7347460389137268,
      "learning_rate": 0.0005198678591609235,
      "loss": 2.3428,
      "step": 54875
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7576740980148315,
      "learning_rate": 0.0005198650761507788,
      "loss": 2.1248,
      "step": 54876
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.801401674747467,
      "learning_rate": 0.0005198622930997569,
      "loss": 2.1549,
      "step": 54877
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7463572025299072,
      "learning_rate": 0.0005198595100078584,
      "loss": 2.2119,
      "step": 54878
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7643774151802063,
      "learning_rate": 0.0005198567268750841,
      "loss": 2.2756,
      "step": 54879
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.750983715057373,
      "learning_rate": 0.0005198539437014342,
      "loss": 2.2481,
      "step": 54880
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7983998656272888,
      "learning_rate": 0.0005198511604869091,
      "loss": 2.1932,
      "step": 54881
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7462118268013,
      "learning_rate": 0.0005198483772315098,
      "loss": 2.1685,
      "step": 54882
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.8820186257362366,
      "learning_rate": 0.0005198455939352364,
      "loss": 2.2936,
      "step": 54883
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.754721999168396,
      "learning_rate": 0.0005198428105980896,
      "loss": 2.0837,
      "step": 54884
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7878516912460327,
      "learning_rate": 0.0005198400272200698,
      "loss": 2.4143,
      "step": 54885
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.769136905670166,
      "learning_rate": 0.0005198372438011776,
      "loss": 2.1719,
      "step": 54886
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7315909266471863,
      "learning_rate": 0.0005198344603414136,
      "loss": 2.1092,
      "step": 54887
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7208291292190552,
      "learning_rate": 0.0005198316768407782,
      "loss": 2.2264,
      "step": 54888
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.9133073091506958,
      "learning_rate": 0.0005198288932992719,
      "loss": 2.1101,
      "step": 54889
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.9118603467941284,
      "learning_rate": 0.0005198261097168952,
      "loss": 2.3728,
      "step": 54890
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7213817834854126,
      "learning_rate": 0.0005198233260936488,
      "loss": 2.1072,
      "step": 54891
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.8459524512290955,
      "learning_rate": 0.000519820542429533,
      "loss": 2.2064,
      "step": 54892
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.8551100492477417,
      "learning_rate": 0.0005198177587245485,
      "loss": 2.0833,
      "step": 54893
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.8026752471923828,
      "learning_rate": 0.0005198149749786956,
      "loss": 2.0701,
      "step": 54894
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7259284853935242,
      "learning_rate": 0.000519812191191975,
      "loss": 2.0424,
      "step": 54895
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.284990906715393,
      "learning_rate": 0.0005198094073643872,
      "loss": 2.35,
      "step": 54896
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.1256308555603027,
      "learning_rate": 0.0005198066234959328,
      "loss": 2.1949,
      "step": 54897
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7775357961654663,
      "learning_rate": 0.0005198038395866119,
      "loss": 2.3416,
      "step": 54898
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.407042384147644,
      "learning_rate": 0.0005198010556364256,
      "loss": 2.245,
      "step": 54899
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.177674412727356,
      "learning_rate": 0.000519798271645374,
      "loss": 2.2766,
      "step": 54900
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7547638416290283,
      "learning_rate": 0.0005197954876134577,
      "loss": 2.1943,
      "step": 54901
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.841096818447113,
      "learning_rate": 0.0005197927035406774,
      "loss": 1.9974,
      "step": 54902
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.9263397455215454,
      "learning_rate": 0.0005197899194270334,
      "loss": 2.0113,
      "step": 54903
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.8151969909667969,
      "learning_rate": 0.0005197871352725263,
      "loss": 2.487,
      "step": 54904
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7044544219970703,
      "learning_rate": 0.0005197843510771567,
      "loss": 2.4059,
      "step": 54905
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7972066402435303,
      "learning_rate": 0.000519781566840925,
      "loss": 2.2193,
      "step": 54906
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7875263094902039,
      "learning_rate": 0.0005197787825638318,
      "loss": 2.3329,
      "step": 54907
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7955491542816162,
      "learning_rate": 0.0005197759982458776,
      "loss": 2.4448,
      "step": 54908
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7940651178359985,
      "learning_rate": 0.0005197732138870628,
      "loss": 2.3112,
      "step": 54909
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.869333028793335,
      "learning_rate": 0.000519770429487388,
      "loss": 2.1673,
      "step": 54910
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7552674412727356,
      "learning_rate": 0.0005197676450468537,
      "loss": 2.0606,
      "step": 54911
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7447060942649841,
      "learning_rate": 0.0005197648605654605,
      "loss": 2.3105,
      "step": 54912
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.6786442995071411,
      "learning_rate": 0.000519762076043209,
      "loss": 2.1879,
      "step": 54913
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.8169559836387634,
      "learning_rate": 0.0005197592914800993,
      "loss": 2.2214,
      "step": 54914
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7819408178329468,
      "learning_rate": 0.0005197565068761324,
      "loss": 2.2432,
      "step": 54915
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7802920341491699,
      "learning_rate": 0.0005197537222313087,
      "loss": 2.268,
      "step": 54916
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.9593026041984558,
      "learning_rate": 0.0005197509375456285,
      "loss": 2.1916,
      "step": 54917
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7917444705963135,
      "learning_rate": 0.0005197481528190923,
      "loss": 2.5322,
      "step": 54918
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7953441143035889,
      "learning_rate": 0.000519745368051701,
      "loss": 2.2257,
      "step": 54919
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7329858541488647,
      "learning_rate": 0.0005197425832434548,
      "loss": 2.4797,
      "step": 54920
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7450851202011108,
      "learning_rate": 0.0005197397983943544,
      "loss": 2.3696,
      "step": 54921
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.691224992275238,
      "learning_rate": 0.0005197370135044001,
      "loss": 2.1642,
      "step": 54922
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.758300244808197,
      "learning_rate": 0.0005197342285735927,
      "loss": 2.205,
      "step": 54923
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7842313051223755,
      "learning_rate": 0.0005197314436019324,
      "loss": 2.4343,
      "step": 54924
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7689148187637329,
      "learning_rate": 0.0005197286585894199,
      "loss": 2.0496,
      "step": 54925
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.764106273651123,
      "learning_rate": 0.0005197258735360557,
      "loss": 2.247,
      "step": 54926
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7001471519470215,
      "learning_rate": 0.0005197230884418404,
      "loss": 2.3365,
      "step": 54927
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.758427619934082,
      "learning_rate": 0.0005197203033067743,
      "loss": 2.2458,
      "step": 54928
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7704175710678101,
      "learning_rate": 0.0005197175181308581,
      "loss": 2.2005,
      "step": 54929
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7152750492095947,
      "learning_rate": 0.0005197147329140923,
      "loss": 2.3284,
      "step": 54930
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.8494814038276672,
      "learning_rate": 0.0005197119476564773,
      "loss": 2.3297,
      "step": 54931
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.6592726111412048,
      "learning_rate": 0.0005197091623580138,
      "loss": 2.2814,
      "step": 54932
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7591126561164856,
      "learning_rate": 0.0005197063770187023,
      "loss": 2.1247,
      "step": 54933
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.8561354875564575,
      "learning_rate": 0.000519703591638543,
      "loss": 2.3908,
      "step": 54934
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7352066040039062,
      "learning_rate": 0.0005197008062175369,
      "loss": 2.1185,
      "step": 54935
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.6877031326293945,
      "learning_rate": 0.0005196980207556841,
      "loss": 2.0943,
      "step": 54936
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.8044583201408386,
      "learning_rate": 0.0005196952352529853,
      "loss": 2.1087,
      "step": 54937
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.72808438539505,
      "learning_rate": 0.0005196924497094411,
      "loss": 2.4453,
      "step": 54938
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.6284052729606628,
      "learning_rate": 0.0005196896641250519,
      "loss": 2.3044,
      "step": 54939
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7093517184257507,
      "learning_rate": 0.0005196868784998181,
      "loss": 2.0414,
      "step": 54940
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.6951805353164673,
      "learning_rate": 0.0005196840928337406,
      "loss": 2.1538,
      "step": 54941
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.6989983320236206,
      "learning_rate": 0.0005196813071268196,
      "loss": 2.2956,
      "step": 54942
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7076942920684814,
      "learning_rate": 0.0005196785213790556,
      "loss": 2.2092,
      "step": 54943
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7206646800041199,
      "learning_rate": 0.0005196757355904493,
      "loss": 2.1867,
      "step": 54944
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.8302278518676758,
      "learning_rate": 0.000519672949761001,
      "loss": 2.4045,
      "step": 54945
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7250057458877563,
      "learning_rate": 0.0005196701638907115,
      "loss": 2.2213,
      "step": 54946
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7570334076881409,
      "learning_rate": 0.0005196673779795811,
      "loss": 2.0174,
      "step": 54947
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7093132734298706,
      "learning_rate": 0.0005196645920276104,
      "loss": 2.1733,
      "step": 54948
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.725223183631897,
      "learning_rate": 0.0005196618060348,
      "loss": 2.2037,
      "step": 54949
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7039871215820312,
      "learning_rate": 0.0005196590200011502,
      "loss": 2.1575,
      "step": 54950
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7630718946456909,
      "learning_rate": 0.0005196562339266617,
      "loss": 2.4029,
      "step": 54951
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.8154664635658264,
      "learning_rate": 0.000519653447811335,
      "loss": 2.2322,
      "step": 54952
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.6985647678375244,
      "learning_rate": 0.0005196506616551705,
      "loss": 2.2649,
      "step": 54953
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7439324259757996,
      "learning_rate": 0.0005196478754581688,
      "loss": 2.5419,
      "step": 54954
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.8484424352645874,
      "learning_rate": 0.0005196450892203305,
      "loss": 2.4946,
      "step": 54955
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7516342997550964,
      "learning_rate": 0.0005196423029416559,
      "loss": 2.0998,
      "step": 54956
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.6763609647750854,
      "learning_rate": 0.0005196395166221457,
      "loss": 2.3486,
      "step": 54957
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.6878328323364258,
      "learning_rate": 0.0005196367302618004,
      "loss": 1.9749,
      "step": 54958
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.8390675783157349,
      "learning_rate": 0.0005196339438606206,
      "loss": 2.1447,
      "step": 54959
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7529588341712952,
      "learning_rate": 0.0005196311574186065,
      "loss": 2.2953,
      "step": 54960
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7460564970970154,
      "learning_rate": 0.000519628370935759,
      "loss": 2.3771,
      "step": 54961
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7788348197937012,
      "learning_rate": 0.0005196255844120782,
      "loss": 2.2157,
      "step": 54962
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.770504891872406,
      "learning_rate": 0.000519622797847565,
      "loss": 2.3545,
      "step": 54963
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7482771277427673,
      "learning_rate": 0.0005196200112422199,
      "loss": 2.2384,
      "step": 54964
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7551009654998779,
      "learning_rate": 0.000519617224596043,
      "loss": 2.2506,
      "step": 54965
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.6856396794319153,
      "learning_rate": 0.0005196144379090354,
      "loss": 2.3845,
      "step": 54966
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.6750068068504333,
      "learning_rate": 0.0005196116511811972,
      "loss": 2.2476,
      "step": 54967
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7541694045066833,
      "learning_rate": 0.000519608864412529,
      "loss": 2.3593,
      "step": 54968
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.8246185183525085,
      "learning_rate": 0.0005196060776030316,
      "loss": 2.2447,
      "step": 54969
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7832112908363342,
      "learning_rate": 0.000519603290752705,
      "loss": 2.3296,
      "step": 54970
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7276589274406433,
      "learning_rate": 0.0005196005038615501,
      "loss": 2.1479,
      "step": 54971
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.6957823634147644,
      "learning_rate": 0.0005195977169295673,
      "loss": 2.1447,
      "step": 54972
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7159120440483093,
      "learning_rate": 0.0005195949299567572,
      "loss": 2.2459,
      "step": 54973
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.70698481798172,
      "learning_rate": 0.0005195921429431203,
      "loss": 2.1052,
      "step": 54974
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.8030856847763062,
      "learning_rate": 0.000519589355888657,
      "loss": 2.342,
      "step": 54975
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7145678997039795,
      "learning_rate": 0.0005195865687933678,
      "loss": 2.2719,
      "step": 54976
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.6745787858963013,
      "learning_rate": 0.0005195837816572535,
      "loss": 2.2427,
      "step": 54977
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7318700551986694,
      "learning_rate": 0.0005195809944803144,
      "loss": 2.1349,
      "step": 54978
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.6686956286430359,
      "learning_rate": 0.0005195782072625509,
      "loss": 2.1626,
      "step": 54979
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.700884997844696,
      "learning_rate": 0.0005195754200039638,
      "loss": 2.3022,
      "step": 54980
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7372944355010986,
      "learning_rate": 0.0005195726327045534,
      "loss": 2.3455,
      "step": 54981
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.6799297332763672,
      "learning_rate": 0.0005195698453643203,
      "loss": 2.2276,
      "step": 54982
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7842737436294556,
      "learning_rate": 0.0005195670579832651,
      "loss": 2.3242,
      "step": 54983
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.730164110660553,
      "learning_rate": 0.0005195642705613883,
      "loss": 2.1707,
      "step": 54984
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.6727378964424133,
      "learning_rate": 0.0005195614830986902,
      "loss": 2.0988,
      "step": 54985
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7146205306053162,
      "learning_rate": 0.0005195586955951716,
      "loss": 2.0787,
      "step": 54986
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7178872227668762,
      "learning_rate": 0.0005195559080508328,
      "loss": 2.0202,
      "step": 54987
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.6910398602485657,
      "learning_rate": 0.0005195531204656744,
      "loss": 2.3783,
      "step": 54988
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7724307179450989,
      "learning_rate": 0.0005195503328396969,
      "loss": 2.2688,
      "step": 54989
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7852352261543274,
      "learning_rate": 0.000519547545172901,
      "loss": 2.2842,
      "step": 54990
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7301807999610901,
      "learning_rate": 0.0005195447574652869,
      "loss": 2.2151,
      "step": 54991
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7285151481628418,
      "learning_rate": 0.0005195419697168554,
      "loss": 2.2847,
      "step": 54992
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.695135772228241,
      "learning_rate": 0.0005195391819276069,
      "loss": 2.1294,
      "step": 54993
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7039612531661987,
      "learning_rate": 0.000519536394097542,
      "loss": 2.1865,
      "step": 54994
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7599328756332397,
      "learning_rate": 0.000519533606226661,
      "loss": 2.2228,
      "step": 54995
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.6603272557258606,
      "learning_rate": 0.0005195308183149645,
      "loss": 2.2993,
      "step": 54996
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7444250583648682,
      "learning_rate": 0.0005195280303624533,
      "loss": 2.2916,
      "step": 54997
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.6891621351242065,
      "learning_rate": 0.0005195252423691275,
      "loss": 2.0971,
      "step": 54998
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7585523128509521,
      "learning_rate": 0.0005195224543349879,
      "loss": 2.0349,
      "step": 54999
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7074142098426819,
      "learning_rate": 0.0005195196662600349,
      "loss": 2.1804,
      "step": 55000
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7075270414352417,
      "learning_rate": 0.0005195168781442691,
      "loss": 2.1511,
      "step": 55001
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.761803388595581,
      "learning_rate": 0.0005195140899876909,
      "loss": 2.3994,
      "step": 55002
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7328999638557434,
      "learning_rate": 0.000519511301790301,
      "loss": 2.2154,
      "step": 55003
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7214812636375427,
      "learning_rate": 0.0005195085135520997,
      "loss": 2.2641,
      "step": 55004
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7350773215293884,
      "learning_rate": 0.0005195057252730876,
      "loss": 2.2107,
      "step": 55005
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.6871393918991089,
      "learning_rate": 0.0005195029369532653,
      "loss": 2.1456,
      "step": 55006
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7438825368881226,
      "learning_rate": 0.0005195001485926333,
      "loss": 2.1494,
      "step": 55007
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7634891271591187,
      "learning_rate": 0.000519497360191192,
      "loss": 2.3069,
      "step": 55008
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7780221104621887,
      "learning_rate": 0.000519494571748942,
      "loss": 2.0555,
      "step": 55009
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7162334322929382,
      "learning_rate": 0.000519491783265884,
      "loss": 2.2787,
      "step": 55010
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7784967422485352,
      "learning_rate": 0.0005194889947420181,
      "loss": 2.3925,
      "step": 55011
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7210661172866821,
      "learning_rate": 0.0005194862061773452,
      "loss": 2.3923,
      "step": 55012
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.6802520751953125,
      "learning_rate": 0.0005194834175718655,
      "loss": 2.2333,
      "step": 55013
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7308463454246521,
      "learning_rate": 0.0005194806289255798,
      "loss": 2.1378,
      "step": 55014
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.726000189781189,
      "learning_rate": 0.0005194778402384886,
      "loss": 2.2925,
      "step": 55015
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7501394748687744,
      "learning_rate": 0.0005194750515105922,
      "loss": 2.3138,
      "step": 55016
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.6993831396102905,
      "learning_rate": 0.0005194722627418913,
      "loss": 2.2827,
      "step": 55017
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7219120264053345,
      "learning_rate": 0.0005194694739323863,
      "loss": 2.4297,
      "step": 55018
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.6688621044158936,
      "learning_rate": 0.0005194666850820778,
      "loss": 2.2273,
      "step": 55019
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.6406918168067932,
      "learning_rate": 0.0005194638961909663,
      "loss": 1.9714,
      "step": 55020
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.6576544642448425,
      "learning_rate": 0.0005194611072590524,
      "loss": 2.49,
      "step": 55021
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.926594078540802,
      "learning_rate": 0.0005194583182863365,
      "loss": 2.3504,
      "step": 55022
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.6950075626373291,
      "learning_rate": 0.0005194555292728191,
      "loss": 2.0534,
      "step": 55023
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7025631666183472,
      "learning_rate": 0.0005194527402185009,
      "loss": 2.1767,
      "step": 55024
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.8042119741439819,
      "learning_rate": 0.0005194499511233822,
      "loss": 2.409,
      "step": 55025
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7930639982223511,
      "learning_rate": 0.0005194471619874635,
      "loss": 2.1127,
      "step": 55026
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7347415089607239,
      "learning_rate": 0.0005194443728107456,
      "loss": 2.3229,
      "step": 55027
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7601650953292847,
      "learning_rate": 0.0005194415835932288,
      "loss": 2.3789,
      "step": 55028
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7625089883804321,
      "learning_rate": 0.0005194387943349136,
      "loss": 2.2311,
      "step": 55029
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7605098485946655,
      "learning_rate": 0.0005194360050358007,
      "loss": 2.353,
      "step": 55030
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.810371458530426,
      "learning_rate": 0.0005194332156958904,
      "loss": 2.0032,
      "step": 55031
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7164668440818787,
      "learning_rate": 0.0005194304263151834,
      "loss": 2.2494,
      "step": 55032
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.6968504786491394,
      "learning_rate": 0.0005194276368936802,
      "loss": 2.1036,
      "step": 55033
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7345666885375977,
      "learning_rate": 0.0005194248474313811,
      "loss": 2.253,
      "step": 55034
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.712691605091095,
      "learning_rate": 0.0005194220579282869,
      "loss": 2.2098,
      "step": 55035
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7925660610198975,
      "learning_rate": 0.000519419268384398,
      "loss": 2.447,
      "step": 55036
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7589151263237,
      "learning_rate": 0.0005194164787997148,
      "loss": 2.1969,
      "step": 55037
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.6962139010429382,
      "learning_rate": 0.0005194136891742379,
      "loss": 2.1653,
      "step": 55038
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7323964238166809,
      "learning_rate": 0.000519410899507968,
      "loss": 2.0522,
      "step": 55039
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7309183478355408,
      "learning_rate": 0.0005194081098009054,
      "loss": 2.1326,
      "step": 55040
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7087041139602661,
      "learning_rate": 0.0005194053200530508,
      "loss": 2.3939,
      "step": 55041
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7982372641563416,
      "learning_rate": 0.0005194025302644045,
      "loss": 2.2979,
      "step": 55042
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7406044602394104,
      "learning_rate": 0.0005193997404349671,
      "loss": 2.157,
      "step": 55043
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.695020318031311,
      "learning_rate": 0.0005193969505647393,
      "loss": 2.2831,
      "step": 55044
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.9486863613128662,
      "learning_rate": 0.0005193941606537213,
      "loss": 2.2659,
      "step": 55045
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7221465706825256,
      "learning_rate": 0.0005193913707019138,
      "loss": 2.2592,
      "step": 55046
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7005113363265991,
      "learning_rate": 0.0005193885807093174,
      "loss": 2.3412,
      "step": 55047
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.8212511539459229,
      "learning_rate": 0.0005193857906759325,
      "loss": 2.3029,
      "step": 55048
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.6662595868110657,
      "learning_rate": 0.0005193830006017596,
      "loss": 2.2474,
      "step": 55049
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7190871238708496,
      "learning_rate": 0.0005193802104867994,
      "loss": 2.291,
      "step": 55050
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.6925999522209167,
      "learning_rate": 0.0005193774203310521,
      "loss": 2.0384,
      "step": 55051
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7855832576751709,
      "learning_rate": 0.0005193746301345184,
      "loss": 2.0662,
      "step": 55052
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.8185771107673645,
      "learning_rate": 0.0005193718398971989,
      "loss": 2.2243,
      "step": 55053
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7853540182113647,
      "learning_rate": 0.000519369049619094,
      "loss": 2.1899,
      "step": 55054
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.8500884771347046,
      "learning_rate": 0.0005193662593002042,
      "loss": 2.168,
      "step": 55055
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7047272324562073,
      "learning_rate": 0.0005193634689405302,
      "loss": 2.264,
      "step": 55056
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7949672341346741,
      "learning_rate": 0.0005193606785400723,
      "loss": 2.0962,
      "step": 55057
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7498898506164551,
      "learning_rate": 0.0005193578880988312,
      "loss": 2.0991,
      "step": 55058
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.8631690144538879,
      "learning_rate": 0.0005193550976168072,
      "loss": 2.2421,
      "step": 55059
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7201798558235168,
      "learning_rate": 0.000519352307094001,
      "loss": 2.2642,
      "step": 55060
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7539017200469971,
      "learning_rate": 0.0005193495165304131,
      "loss": 2.3378,
      "step": 55061
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7074759602546692,
      "learning_rate": 0.0005193467259260438,
      "loss": 2.2275,
      "step": 55062
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7512304782867432,
      "learning_rate": 0.000519343935280894,
      "loss": 2.0774,
      "step": 55063
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.863584578037262,
      "learning_rate": 0.000519341144594964,
      "loss": 2.2489,
      "step": 55064
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7260735630989075,
      "learning_rate": 0.0005193383538682543,
      "loss": 2.0787,
      "step": 55065
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.001358151435852,
      "learning_rate": 0.0005193355631007655,
      "loss": 2.3859,
      "step": 55066
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7144906520843506,
      "learning_rate": 0.000519332772292498,
      "loss": 2.174,
      "step": 55067
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.6860518455505371,
      "learning_rate": 0.0005193299814434525,
      "loss": 2.219,
      "step": 55068
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.8319110870361328,
      "learning_rate": 0.0005193271905536294,
      "loss": 2.2437,
      "step": 55069
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7634477615356445,
      "learning_rate": 0.0005193243996230292,
      "loss": 2.295,
      "step": 55070
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7048162817955017,
      "learning_rate": 0.0005193216086516525,
      "loss": 2.3702,
      "step": 55071
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.9521545767784119,
      "learning_rate": 0.0005193188176394996,
      "loss": 2.293,
      "step": 55072
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.717173159122467,
      "learning_rate": 0.0005193160265865714,
      "loss": 2.2128,
      "step": 55073
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7616709470748901,
      "learning_rate": 0.000519313235492868,
      "loss": 2.3766,
      "step": 55074
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7517651319503784,
      "learning_rate": 0.0005193104443583903,
      "loss": 2.418,
      "step": 55075
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.8807771801948547,
      "learning_rate": 0.0005193076531831386,
      "loss": 2.1927,
      "step": 55076
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7465932369232178,
      "learning_rate": 0.0005193048619671135,
      "loss": 2.2272,
      "step": 55077
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.6976784467697144,
      "learning_rate": 0.0005193020707103156,
      "loss": 2.1993,
      "step": 55078
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.0027934312820435,
      "learning_rate": 0.0005192992794127451,
      "loss": 2.4281,
      "step": 55079
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7177576422691345,
      "learning_rate": 0.0005192964880744028,
      "loss": 2.2017,
      "step": 55080
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.738241970539093,
      "learning_rate": 0.000519293696695289,
      "loss": 2.3705,
      "step": 55081
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.763611912727356,
      "learning_rate": 0.0005192909052754046,
      "loss": 2.1883,
      "step": 55082
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.8787568807601929,
      "learning_rate": 0.0005192881138147497,
      "loss": 2.1372,
      "step": 55083
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7459545135498047,
      "learning_rate": 0.0005192853223133252,
      "loss": 2.2699,
      "step": 55084
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.9231914281845093,
      "learning_rate": 0.0005192825307711313,
      "loss": 2.5371,
      "step": 55085
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.806206226348877,
      "learning_rate": 0.0005192797391881686,
      "loss": 2.0986,
      "step": 55086
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.768384575843811,
      "learning_rate": 0.0005192769475644377,
      "loss": 2.3798,
      "step": 55087
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.722976803779602,
      "learning_rate": 0.0005192741558999391,
      "loss": 2.3291,
      "step": 55088
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.8595715165138245,
      "learning_rate": 0.0005192713641946732,
      "loss": 2.0572,
      "step": 55089
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7311512231826782,
      "learning_rate": 0.0005192685724486407,
      "loss": 2.2923,
      "step": 55090
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7985476851463318,
      "learning_rate": 0.000519265780661842,
      "loss": 2.1471,
      "step": 55091
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.8774861097335815,
      "learning_rate": 0.0005192629888342777,
      "loss": 2.3351,
      "step": 55092
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7611034512519836,
      "learning_rate": 0.0005192601969659482,
      "loss": 2.1649,
      "step": 55093
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7837921977043152,
      "learning_rate": 0.0005192574050568541,
      "loss": 2.3089,
      "step": 55094
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.6960686445236206,
      "learning_rate": 0.000519254613106996,
      "loss": 2.1569,
      "step": 55095
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.6652238368988037,
      "learning_rate": 0.0005192518211163742,
      "loss": 2.0793,
      "step": 55096
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.6716203689575195,
      "learning_rate": 0.0005192490290849895,
      "loss": 2.243,
      "step": 55097
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7547289729118347,
      "learning_rate": 0.0005192462370128422,
      "loss": 2.0811,
      "step": 55098
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7381510734558105,
      "learning_rate": 0.0005192434448999328,
      "loss": 2.1504,
      "step": 55099
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7287532091140747,
      "learning_rate": 0.000519240652746262,
      "loss": 2.2921,
      "step": 55100
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.831133246421814,
      "learning_rate": 0.0005192378605518302,
      "loss": 2.4529,
      "step": 55101
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.8460767269134521,
      "learning_rate": 0.000519235068316638,
      "loss": 2.284,
      "step": 55102
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.6980317831039429,
      "learning_rate": 0.0005192322760406858,
      "loss": 1.8846,
      "step": 55103
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7757150530815125,
      "learning_rate": 0.0005192294837239741,
      "loss": 2.3005,
      "step": 55104
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7006738185882568,
      "learning_rate": 0.0005192266913665036,
      "loss": 2.3382,
      "step": 55105
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.8050062656402588,
      "learning_rate": 0.0005192238989682747,
      "loss": 2.4459,
      "step": 55106
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7617893815040588,
      "learning_rate": 0.0005192211065292879,
      "loss": 2.3979,
      "step": 55107
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7731707096099854,
      "learning_rate": 0.0005192183140495439,
      "loss": 2.0224,
      "step": 55108
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.741580069065094,
      "learning_rate": 0.000519215521529043,
      "loss": 2.1682,
      "step": 55109
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7705330848693848,
      "learning_rate": 0.0005192127289677858,
      "loss": 2.279,
      "step": 55110
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.854748547077179,
      "learning_rate": 0.0005192099363657728,
      "loss": 2.2557,
      "step": 55111
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7377313375473022,
      "learning_rate": 0.0005192071437230045,
      "loss": 2.1483,
      "step": 55112
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7602927088737488,
      "learning_rate": 0.0005192043510394814,
      "loss": 2.292,
      "step": 55113
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7057240009307861,
      "learning_rate": 0.0005192015583152042,
      "loss": 2.3474,
      "step": 55114
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7219403386116028,
      "learning_rate": 0.0005191987655501733,
      "loss": 2.1311,
      "step": 55115
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7981395125389099,
      "learning_rate": 0.000519195972744389,
      "loss": 2.3479,
      "step": 55116
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.8500915765762329,
      "learning_rate": 0.0005191931798978523,
      "loss": 2.1872,
      "step": 55117
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7386999130249023,
      "learning_rate": 0.0005191903870105633,
      "loss": 2.0498,
      "step": 55118
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7375204563140869,
      "learning_rate": 0.0005191875940825227,
      "loss": 2.2921,
      "step": 55119
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7710079550743103,
      "learning_rate": 0.000519184801113731,
      "loss": 2.4641,
      "step": 55120
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7419416904449463,
      "learning_rate": 0.0005191820081041887,
      "loss": 2.2246,
      "step": 55121
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.8160260319709778,
      "learning_rate": 0.0005191792150538964,
      "loss": 2.3007,
      "step": 55122
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7784391045570374,
      "learning_rate": 0.0005191764219628544,
      "loss": 2.3218,
      "step": 55123
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7084072232246399,
      "learning_rate": 0.0005191736288310636,
      "loss": 2.1068,
      "step": 55124
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7544299960136414,
      "learning_rate": 0.0005191708356585239,
      "loss": 2.1893,
      "step": 55125
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7651153802871704,
      "learning_rate": 0.0005191680424452365,
      "loss": 2.257,
      "step": 55126
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.8156110048294067,
      "learning_rate": 0.0005191652491912017,
      "loss": 1.9723,
      "step": 55127
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7093722224235535,
      "learning_rate": 0.0005191624558964197,
      "loss": 2.3849,
      "step": 55128
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.6698318719863892,
      "learning_rate": 0.0005191596625608914,
      "loss": 2.5527,
      "step": 55129
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.6843541860580444,
      "learning_rate": 0.0005191568691846171,
      "loss": 2.3116,
      "step": 55130
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.6975905895233154,
      "learning_rate": 0.0005191540757675975,
      "loss": 2.1823,
      "step": 55131
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.74223792552948,
      "learning_rate": 0.000519151282309833,
      "loss": 2.1538,
      "step": 55132
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7292811870574951,
      "learning_rate": 0.0005191484888113241,
      "loss": 1.8744,
      "step": 55133
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7057968378067017,
      "learning_rate": 0.0005191456952720713,
      "loss": 2.4216,
      "step": 55134
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.8468438386917114,
      "learning_rate": 0.0005191429016920752,
      "loss": 2.1519,
      "step": 55135
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.8754515051841736,
      "learning_rate": 0.0005191401080713364,
      "loss": 2.2919,
      "step": 55136
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.8909938931465149,
      "learning_rate": 0.0005191373144098553,
      "loss": 2.2406,
      "step": 55137
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7173052430152893,
      "learning_rate": 0.0005191345207076324,
      "loss": 2.2834,
      "step": 55138
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.6683081984519958,
      "learning_rate": 0.0005191317269646682,
      "loss": 2.3456,
      "step": 55139
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.6781718730926514,
      "learning_rate": 0.0005191289331809634,
      "loss": 2.2625,
      "step": 55140
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7263286709785461,
      "learning_rate": 0.0005191261393565183,
      "loss": 2.3079,
      "step": 55141
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7274942398071289,
      "learning_rate": 0.0005191233454913335,
      "loss": 2.2228,
      "step": 55142
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7228373289108276,
      "learning_rate": 0.0005191205515854096,
      "loss": 2.4226,
      "step": 55143
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7041746973991394,
      "learning_rate": 0.000519117757638747,
      "loss": 2.1134,
      "step": 55144
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7791133522987366,
      "learning_rate": 0.0005191149636513463,
      "loss": 2.1342,
      "step": 55145
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7190633416175842,
      "learning_rate": 0.000519112169623208,
      "loss": 2.1853,
      "step": 55146
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7474341988563538,
      "learning_rate": 0.0005191093755543326,
      "loss": 2.186,
      "step": 55147
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7375440001487732,
      "learning_rate": 0.0005191065814447206,
      "loss": 2.1699,
      "step": 55148
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.6627721190452576,
      "learning_rate": 0.0005191037872943726,
      "loss": 2.1527,
      "step": 55149
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7315503358840942,
      "learning_rate": 0.000519100993103289,
      "loss": 2.3331,
      "step": 55150
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.8323343396186829,
      "learning_rate": 0.0005190981988714706,
      "loss": 2.3551,
      "step": 55151
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7059486508369446,
      "learning_rate": 0.0005190954045989174,
      "loss": 2.1789,
      "step": 55152
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.6816363334655762,
      "learning_rate": 0.0005190926102856305,
      "loss": 2.2738,
      "step": 55153
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7355485558509827,
      "learning_rate": 0.00051908981593161,
      "loss": 2.2824,
      "step": 55154
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.6380686163902283,
      "learning_rate": 0.0005190870215368565,
      "loss": 2.2613,
      "step": 55155
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.696863055229187,
      "learning_rate": 0.0005190842271013708,
      "loss": 2.3419,
      "step": 55156
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.6941540241241455,
      "learning_rate": 0.000519081432625153,
      "loss": 2.1209,
      "step": 55157
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7547069787979126,
      "learning_rate": 0.0005190786381082039,
      "loss": 2.3147,
      "step": 55158
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7522847652435303,
      "learning_rate": 0.0005190758435505241,
      "loss": 2.3785,
      "step": 55159
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7490749359130859,
      "learning_rate": 0.0005190730489521137,
      "loss": 2.322,
      "step": 55160
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7640385627746582,
      "learning_rate": 0.0005190702543129737,
      "loss": 2.3341,
      "step": 55161
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.8092783689498901,
      "learning_rate": 0.0005190674596331043,
      "loss": 2.2415,
      "step": 55162
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7231324315071106,
      "learning_rate": 0.0005190646649125061,
      "loss": 2.3722,
      "step": 55163
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.6955395936965942,
      "learning_rate": 0.0005190618701511798,
      "loss": 2.4076,
      "step": 55164
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.8340198397636414,
      "learning_rate": 0.0005190590753491256,
      "loss": 2.2371,
      "step": 55165
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7398189902305603,
      "learning_rate": 0.0005190562805063442,
      "loss": 2.4103,
      "step": 55166
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7677909135818481,
      "learning_rate": 0.0005190534856228361,
      "loss": 2.2606,
      "step": 55167
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.9229421019554138,
      "learning_rate": 0.0005190506906986019,
      "loss": 2.1847,
      "step": 55168
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.8503461480140686,
      "learning_rate": 0.0005190478957336419,
      "loss": 2.2038,
      "step": 55169
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7242085337638855,
      "learning_rate": 0.0005190451007279569,
      "loss": 2.1771,
      "step": 55170
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.6894997954368591,
      "learning_rate": 0.0005190423056815472,
      "loss": 2.382,
      "step": 55171
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.8965461254119873,
      "learning_rate": 0.0005190395105944135,
      "loss": 2.314,
      "step": 55172
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7906724810600281,
      "learning_rate": 0.0005190367154665562,
      "loss": 2.2167,
      "step": 55173
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7033403515815735,
      "learning_rate": 0.0005190339202979758,
      "loss": 2.1736,
      "step": 55174
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.8100410103797913,
      "learning_rate": 0.0005190311250886728,
      "loss": 2.1361,
      "step": 55175
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.9204890131950378,
      "learning_rate": 0.0005190283298386479,
      "loss": 2.1765,
      "step": 55176
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.716213047504425,
      "learning_rate": 0.0005190255345479014,
      "loss": 2.4512,
      "step": 55177
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7166861891746521,
      "learning_rate": 0.000519022739216434,
      "loss": 2.1786,
      "step": 55178
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7644202709197998,
      "learning_rate": 0.000519019943844246,
      "loss": 2.2391,
      "step": 55179
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.6557935476303101,
      "learning_rate": 0.0005190171484313382,
      "loss": 2.0589,
      "step": 55180
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7134360671043396,
      "learning_rate": 0.000519014352977711,
      "loss": 2.3101,
      "step": 55181
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7050355076789856,
      "learning_rate": 0.0005190115574833648,
      "loss": 2.2943,
      "step": 55182
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.718039333820343,
      "learning_rate": 0.0005190087619483002,
      "loss": 2.21,
      "step": 55183
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.8326789140701294,
      "learning_rate": 0.0005190059663725178,
      "loss": 2.1826,
      "step": 55184
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.701744556427002,
      "learning_rate": 0.000519003170756018,
      "loss": 2.5505,
      "step": 55185
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7872990369796753,
      "learning_rate": 0.0005190003750988015,
      "loss": 2.1525,
      "step": 55186
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7866539359092712,
      "learning_rate": 0.0005189975794008685,
      "loss": 2.2434,
      "step": 55187
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7271935343742371,
      "learning_rate": 0.0005189947836622199,
      "loss": 2.2152,
      "step": 55188
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7190361022949219,
      "learning_rate": 0.0005189919878828559,
      "loss": 2.4313,
      "step": 55189
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7704853415489197,
      "learning_rate": 0.0005189891920627773,
      "loss": 2.3346,
      "step": 55190
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.880908727645874,
      "learning_rate": 0.0005189863962019845,
      "loss": 2.2169,
      "step": 55191
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7904066443443298,
      "learning_rate": 0.0005189836003004777,
      "loss": 2.2012,
      "step": 55192
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.758610188961029,
      "learning_rate": 0.000518980804358258,
      "loss": 2.1741,
      "step": 55193
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.8703799247741699,
      "learning_rate": 0.0005189780083753255,
      "loss": 2.3752,
      "step": 55194
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.953091025352478,
      "learning_rate": 0.0005189752123516809,
      "loss": 2.0251,
      "step": 55195
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7716920971870422,
      "learning_rate": 0.0005189724162873246,
      "loss": 2.2626,
      "step": 55196
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.1184406280517578,
      "learning_rate": 0.0005189696201822573,
      "loss": 2.2879,
      "step": 55197
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7932067513465881,
      "learning_rate": 0.0005189668240364794,
      "loss": 2.2008,
      "step": 55198
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.8350074291229248,
      "learning_rate": 0.0005189640278499914,
      "loss": 2.2039,
      "step": 55199
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7623172998428345,
      "learning_rate": 0.0005189612316227939,
      "loss": 2.2775,
      "step": 55200
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.9979351162910461,
      "learning_rate": 0.0005189584353548874,
      "loss": 2.1327,
      "step": 55201
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7858731746673584,
      "learning_rate": 0.0005189556390462723,
      "loss": 2.3077,
      "step": 55202
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7686871290206909,
      "learning_rate": 0.0005189528426969492,
      "loss": 2.4479,
      "step": 55203
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.9786691665649414,
      "learning_rate": 0.0005189500463069187,
      "loss": 2.3245,
      "step": 55204
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.9363145232200623,
      "learning_rate": 0.0005189472498761812,
      "loss": 2.4151,
      "step": 55205
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.74642413854599,
      "learning_rate": 0.0005189444534047374,
      "loss": 2.1518,
      "step": 55206
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.6539953947067261,
      "learning_rate": 0.0005189416568925875,
      "loss": 2.3084,
      "step": 55207
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.8896461725234985,
      "learning_rate": 0.0005189388603397323,
      "loss": 2.2409,
      "step": 55208
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.847284734249115,
      "learning_rate": 0.0005189360637461723,
      "loss": 2.099,
      "step": 55209
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7825366258621216,
      "learning_rate": 0.0005189332671119079,
      "loss": 2.0115,
      "step": 55210
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7488070130348206,
      "learning_rate": 0.0005189304704369396,
      "loss": 2.271,
      "step": 55211
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.711312472820282,
      "learning_rate": 0.0005189276737212681,
      "loss": 2.2097,
      "step": 55212
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7736822962760925,
      "learning_rate": 0.0005189248769648939,
      "loss": 2.2683,
      "step": 55213
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.735291600227356,
      "learning_rate": 0.0005189220801678173,
      "loss": 2.2604,
      "step": 55214
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.9605217576026917,
      "learning_rate": 0.0005189192833300389,
      "loss": 2.1813,
      "step": 55215
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7782993912696838,
      "learning_rate": 0.0005189164864515594,
      "loss": 2.1887,
      "step": 55216
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.9156280159950256,
      "learning_rate": 0.0005189136895323791,
      "loss": 2.2205,
      "step": 55217
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7656070590019226,
      "learning_rate": 0.0005189108925724986,
      "loss": 2.1642,
      "step": 55218
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7258223295211792,
      "learning_rate": 0.0005189080955719185,
      "loss": 2.125,
      "step": 55219
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.6860572099685669,
      "learning_rate": 0.0005189052985306393,
      "loss": 2.1933,
      "step": 55220
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.8186572194099426,
      "learning_rate": 0.0005189025014486613,
      "loss": 2.1053,
      "step": 55221
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.149364948272705,
      "learning_rate": 0.0005188997043259853,
      "loss": 2.3692,
      "step": 55222
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7687588334083557,
      "learning_rate": 0.0005188969071626118,
      "loss": 2.0661,
      "step": 55223
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.826526403427124,
      "learning_rate": 0.0005188941099585411,
      "loss": 2.4194,
      "step": 55224
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.8428943157196045,
      "learning_rate": 0.0005188913127137739,
      "loss": 2.3231,
      "step": 55225
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7673919796943665,
      "learning_rate": 0.0005188885154283106,
      "loss": 2.0349,
      "step": 55226
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.6893208622932434,
      "learning_rate": 0.0005188857181021519,
      "loss": 2.1498,
      "step": 55227
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7880518436431885,
      "learning_rate": 0.0005188829207352981,
      "loss": 2.2521,
      "step": 55228
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7967565059661865,
      "learning_rate": 0.00051888012332775,
      "loss": 2.3645,
      "step": 55229
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7243624925613403,
      "learning_rate": 0.0005188773258795079,
      "loss": 2.2357,
      "step": 55230
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.774987518787384,
      "learning_rate": 0.0005188745283905723,
      "loss": 2.3478,
      "step": 55231
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.8352875113487244,
      "learning_rate": 0.0005188717308609438,
      "loss": 2.0807,
      "step": 55232
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.8398117423057556,
      "learning_rate": 0.000518868933290623,
      "loss": 2.3014,
      "step": 55233
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.702393114566803,
      "learning_rate": 0.0005188661356796102,
      "loss": 2.2708,
      "step": 55234
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.746921718120575,
      "learning_rate": 0.0005188633380279061,
      "loss": 2.2096,
      "step": 55235
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7189213633537292,
      "learning_rate": 0.0005188605403355113,
      "loss": 2.3561,
      "step": 55236
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7761676907539368,
      "learning_rate": 0.000518857742602426,
      "loss": 2.3086,
      "step": 55237
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7663728594779968,
      "learning_rate": 0.000518854944828651,
      "loss": 2.0904,
      "step": 55238
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7865197658538818,
      "learning_rate": 0.0005188521470141867,
      "loss": 2.1663,
      "step": 55239
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7885833382606506,
      "learning_rate": 0.0005188493491590338,
      "loss": 2.2403,
      "step": 55240
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7806154489517212,
      "learning_rate": 0.0005188465512631925,
      "loss": 2.0469,
      "step": 55241
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.6517578959465027,
      "learning_rate": 0.0005188437533266636,
      "loss": 1.9618,
      "step": 55242
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.6693878173828125,
      "learning_rate": 0.0005188409553494476,
      "loss": 2.2527,
      "step": 55243
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.8102237582206726,
      "learning_rate": 0.0005188381573315447,
      "loss": 2.4757,
      "step": 55244
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7141343355178833,
      "learning_rate": 0.0005188353592729558,
      "loss": 2.1223,
      "step": 55245
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.8294488191604614,
      "learning_rate": 0.0005188325611736813,
      "loss": 2.3517,
      "step": 55246
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7605201601982117,
      "learning_rate": 0.0005188297630337216,
      "loss": 2.194,
      "step": 55247
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7318192720413208,
      "learning_rate": 0.0005188269648530773,
      "loss": 2.294,
      "step": 55248
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.6939212679862976,
      "learning_rate": 0.0005188241666317491,
      "loss": 2.2911,
      "step": 55249
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7612340450286865,
      "learning_rate": 0.0005188213683697372,
      "loss": 2.3837,
      "step": 55250
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.8381772637367249,
      "learning_rate": 0.0005188185700670424,
      "loss": 2.2918,
      "step": 55251
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.810840904712677,
      "learning_rate": 0.0005188157717236651,
      "loss": 2.2678,
      "step": 55252
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7498651146888733,
      "learning_rate": 0.0005188129733396056,
      "loss": 2.2282,
      "step": 55253
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.8053131699562073,
      "learning_rate": 0.0005188101749148648,
      "loss": 2.0587,
      "step": 55254
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7557618021965027,
      "learning_rate": 0.000518807376449443,
      "loss": 2.3752,
      "step": 55255
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7131075859069824,
      "learning_rate": 0.000518804577943341,
      "loss": 2.1172,
      "step": 55256
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7295359373092651,
      "learning_rate": 0.0005188017793965588,
      "loss": 2.518,
      "step": 55257
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7820019125938416,
      "learning_rate": 0.0005187989808090974,
      "loss": 2.2644,
      "step": 55258
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.6777012348175049,
      "learning_rate": 0.000518796182180957,
      "loss": 1.9611,
      "step": 55259
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7246367931365967,
      "learning_rate": 0.0005187933835121384,
      "loss": 2.376,
      "step": 55260
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7145184278488159,
      "learning_rate": 0.000518790584802642,
      "loss": 2.3471,
      "step": 55261
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.8571640253067017,
      "learning_rate": 0.0005187877860524682,
      "loss": 2.2005,
      "step": 55262
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7734391093254089,
      "learning_rate": 0.0005187849872616176,
      "loss": 1.9953,
      "step": 55263
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7425837516784668,
      "learning_rate": 0.0005187821884300908,
      "loss": 2.2791,
      "step": 55264
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7128655314445496,
      "learning_rate": 0.0005187793895578881,
      "loss": 2.3937,
      "step": 55265
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7685071229934692,
      "learning_rate": 0.0005187765906450104,
      "loss": 2.2505,
      "step": 55266
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7743420004844666,
      "learning_rate": 0.0005187737916914579,
      "loss": 2.1761,
      "step": 55267
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7507856488227844,
      "learning_rate": 0.0005187709926972313,
      "loss": 2.3883,
      "step": 55268
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7454745769500732,
      "learning_rate": 0.000518768193662331,
      "loss": 2.5271,
      "step": 55269
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.6531113386154175,
      "learning_rate": 0.0005187653945867575,
      "loss": 2.1349,
      "step": 55270
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.708686888217926,
      "learning_rate": 0.0005187625954705115,
      "loss": 2.4412,
      "step": 55271
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.8055379986763,
      "learning_rate": 0.0005187597963135933,
      "loss": 2.0327,
      "step": 55272
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7428187727928162,
      "learning_rate": 0.0005187569971160036,
      "loss": 2.0925,
      "step": 55273
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7370011210441589,
      "learning_rate": 0.0005187541978777428,
      "loss": 2.3777,
      "step": 55274
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.9410090446472168,
      "learning_rate": 0.0005187513985988114,
      "loss": 2.0549,
      "step": 55275
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7008568644523621,
      "learning_rate": 0.0005187485992792101,
      "loss": 2.2414,
      "step": 55276
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7538625001907349,
      "learning_rate": 0.0005187457999189394,
      "loss": 2.4273,
      "step": 55277
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.0599836111068726,
      "learning_rate": 0.0005187430005179995,
      "loss": 2.2155,
      "step": 55278
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7868702411651611,
      "learning_rate": 0.0005187402010763913,
      "loss": 2.1536,
      "step": 55279
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7129567265510559,
      "learning_rate": 0.0005187374015941151,
      "loss": 1.9982,
      "step": 55280
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.969528317451477,
      "learning_rate": 0.0005187346020711715,
      "loss": 2.0021,
      "step": 55281
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.6883689761161804,
      "learning_rate": 0.000518731802507561,
      "loss": 2.4596,
      "step": 55282
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7337866425514221,
      "learning_rate": 0.0005187290029032842,
      "loss": 2.0672,
      "step": 55283
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7860162854194641,
      "learning_rate": 0.0005187262032583414,
      "loss": 2.1072,
      "step": 55284
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.8170375823974609,
      "learning_rate": 0.0005187234035727334,
      "loss": 2.2245,
      "step": 55285
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.8591945767402649,
      "learning_rate": 0.0005187206038464605,
      "loss": 2.2441,
      "step": 55286
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7223475575447083,
      "learning_rate": 0.0005187178040795235,
      "loss": 2.5084,
      "step": 55287
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.1832685470581055,
      "learning_rate": 0.0005187150042719225,
      "loss": 2.4444,
      "step": 55288
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7716315388679504,
      "learning_rate": 0.0005187122044236583,
      "loss": 2.3255,
      "step": 55289
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7495847344398499,
      "learning_rate": 0.0005187094045347315,
      "loss": 2.3405,
      "step": 55290
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7220519781112671,
      "learning_rate": 0.0005187066046051424,
      "loss": 2.1821,
      "step": 55291
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.6577768325805664,
      "learning_rate": 0.0005187038046348916,
      "loss": 2.0487,
      "step": 55292
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.846979022026062,
      "learning_rate": 0.0005187010046239797,
      "loss": 2.2862,
      "step": 55293
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7657657861709595,
      "learning_rate": 0.000518698204572407,
      "loss": 2.1131,
      "step": 55294
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7422025799751282,
      "learning_rate": 0.0005186954044801744,
      "loss": 2.3306,
      "step": 55295
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7236512899398804,
      "learning_rate": 0.000518692604347282,
      "loss": 2.0624,
      "step": 55296
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7324831485748291,
      "learning_rate": 0.0005186898041737306,
      "loss": 2.2699,
      "step": 55297
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7031439542770386,
      "learning_rate": 0.0005186870039595206,
      "loss": 2.2054,
      "step": 55298
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7576075196266174,
      "learning_rate": 0.0005186842037046526,
      "loss": 2.442,
      "step": 55299
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7214532494544983,
      "learning_rate": 0.0005186814034091272,
      "loss": 2.291,
      "step": 55300
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7007068395614624,
      "learning_rate": 0.0005186786030729446,
      "loss": 2.2033,
      "step": 55301
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.774010956287384,
      "learning_rate": 0.0005186758026961056,
      "loss": 2.2017,
      "step": 55302
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7370197176933289,
      "learning_rate": 0.0005186730022786106,
      "loss": 2.3222,
      "step": 55303
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7140681147575378,
      "learning_rate": 0.0005186702018204602,
      "loss": 2.1566,
      "step": 55304
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7320739030838013,
      "learning_rate": 0.0005186674013216548,
      "loss": 2.346,
      "step": 55305
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7373257875442505,
      "learning_rate": 0.0005186646007821952,
      "loss": 2.0899,
      "step": 55306
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7351813316345215,
      "learning_rate": 0.0005186618002020816,
      "loss": 2.1226,
      "step": 55307
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7602266669273376,
      "learning_rate": 0.0005186589995813146,
      "loss": 2.4287,
      "step": 55308
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7501922249794006,
      "learning_rate": 0.0005186561989198948,
      "loss": 2.1385,
      "step": 55309
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.6932605504989624,
      "learning_rate": 0.0005186533982178228,
      "loss": 2.1273,
      "step": 55310
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7481704354286194,
      "learning_rate": 0.0005186505974750987,
      "loss": 2.2941,
      "step": 55311
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7135204672813416,
      "learning_rate": 0.0005186477966917237,
      "loss": 2.2277,
      "step": 55312
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.8078516721725464,
      "learning_rate": 0.0005186449958676977,
      "loss": 2.348,
      "step": 55313
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7879610061645508,
      "learning_rate": 0.0005186421950030216,
      "loss": 2.2618,
      "step": 55314
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7233788371086121,
      "learning_rate": 0.0005186393940976957,
      "loss": 2.3719,
      "step": 55315
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7652350068092346,
      "learning_rate": 0.0005186365931517206,
      "loss": 2.14,
      "step": 55316
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.6924310922622681,
      "learning_rate": 0.000518633792165097,
      "loss": 2.1795,
      "step": 55317
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7886202931404114,
      "learning_rate": 0.0005186309911378251,
      "loss": 2.2865,
      "step": 55318
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.6419289708137512,
      "learning_rate": 0.0005186281900699056,
      "loss": 1.9102,
      "step": 55319
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7240726351737976,
      "learning_rate": 0.0005186253889613389,
      "loss": 2.1876,
      "step": 55320
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7220482230186462,
      "learning_rate": 0.0005186225878121257,
      "loss": 2.444,
      "step": 55321
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.6974877119064331,
      "learning_rate": 0.0005186197866222665,
      "loss": 2.1816,
      "step": 55322
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7216779589653015,
      "learning_rate": 0.0005186169853917616,
      "loss": 2.3074,
      "step": 55323
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7025949954986572,
      "learning_rate": 0.0005186141841206118,
      "loss": 2.2458,
      "step": 55324
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.9475229382514954,
      "learning_rate": 0.0005186113828088175,
      "loss": 2.2094,
      "step": 55325
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.723746120929718,
      "learning_rate": 0.0005186085814563792,
      "loss": 2.091,
      "step": 55326
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.8651598691940308,
      "learning_rate": 0.0005186057800632974,
      "loss": 2.3027,
      "step": 55327
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.8063519597053528,
      "learning_rate": 0.0005186029786295726,
      "loss": 2.045,
      "step": 55328
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.9587134718894958,
      "learning_rate": 0.0005186001771552055,
      "loss": 2.3177,
      "step": 55329
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7484343647956848,
      "learning_rate": 0.0005185973756401964,
      "loss": 2.1761,
      "step": 55330
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7717494368553162,
      "learning_rate": 0.000518594574084546,
      "loss": 2.3013,
      "step": 55331
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.754544734954834,
      "learning_rate": 0.0005185917724882548,
      "loss": 2.1321,
      "step": 55332
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.777996838092804,
      "learning_rate": 0.0005185889708513231,
      "loss": 2.4728,
      "step": 55333
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.8034788966178894,
      "learning_rate": 0.0005185861691737517,
      "loss": 2.3095,
      "step": 55334
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.705413818359375,
      "learning_rate": 0.0005185833674555409,
      "loss": 2.4383,
      "step": 55335
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.72295743227005,
      "learning_rate": 0.0005185805656966915,
      "loss": 2.4106,
      "step": 55336
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.6964191198348999,
      "learning_rate": 0.0005185777638972037,
      "loss": 2.3188,
      "step": 55337
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.6361182332038879,
      "learning_rate": 0.0005185749620570783,
      "loss": 1.8528,
      "step": 55338
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.754483163356781,
      "learning_rate": 0.0005185721601763156,
      "loss": 2.2104,
      "step": 55339
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7164149880409241,
      "learning_rate": 0.0005185693582549162,
      "loss": 2.2251,
      "step": 55340
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7721928358078003,
      "learning_rate": 0.0005185665562928806,
      "loss": 2.3577,
      "step": 55341
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.8007415533065796,
      "learning_rate": 0.0005185637542902094,
      "loss": 2.155,
      "step": 55342
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.6647063493728638,
      "learning_rate": 0.000518560952246903,
      "loss": 2.1049,
      "step": 55343
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7506711483001709,
      "learning_rate": 0.000518558150162962,
      "loss": 2.1951,
      "step": 55344
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7805341482162476,
      "learning_rate": 0.000518555348038387,
      "loss": 2.2662,
      "step": 55345
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7038800716400146,
      "learning_rate": 0.0005185525458731785,
      "loss": 2.3195,
      "step": 55346
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7666370272636414,
      "learning_rate": 0.0005185497436673368,
      "loss": 2.2449,
      "step": 55347
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7473605275154114,
      "learning_rate": 0.0005185469414208628,
      "loss": 2.3907,
      "step": 55348
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.809177041053772,
      "learning_rate": 0.0005185441391337565,
      "loss": 2.3218,
      "step": 55349
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7130239605903625,
      "learning_rate": 0.0005185413368060189,
      "loss": 2.1982,
      "step": 55350
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7516310214996338,
      "learning_rate": 0.0005185385344376503,
      "loss": 2.2355,
      "step": 55351
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7875370979309082,
      "learning_rate": 0.0005185357320286512,
      "loss": 2.3384,
      "step": 55352
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7443514466285706,
      "learning_rate": 0.0005185329295790223,
      "loss": 2.1886,
      "step": 55353
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7285133600234985,
      "learning_rate": 0.000518530127088764,
      "loss": 2.2262,
      "step": 55354
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.8076280951499939,
      "learning_rate": 0.0005185273245578767,
      "loss": 2.1229,
      "step": 55355
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7590464949607849,
      "learning_rate": 0.0005185245219863611,
      "loss": 2.3479,
      "step": 55356
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7354553937911987,
      "learning_rate": 0.0005185217193742176,
      "loss": 2.0258,
      "step": 55357
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.8471258878707886,
      "learning_rate": 0.0005185189167214471,
      "loss": 2.3873,
      "step": 55358
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7960079908370972,
      "learning_rate": 0.0005185161140280495,
      "loss": 2.2944,
      "step": 55359
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7974193692207336,
      "learning_rate": 0.0005185133112940257,
      "loss": 2.3098,
      "step": 55360
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7395751476287842,
      "learning_rate": 0.0005185105085193762,
      "loss": 2.2885,
      "step": 55361
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7178335785865784,
      "learning_rate": 0.0005185077057041014,
      "loss": 2.2634,
      "step": 55362
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.066524863243103,
      "learning_rate": 0.0005185049028482019,
      "loss": 2.2946,
      "step": 55363
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7331182956695557,
      "learning_rate": 0.0005185020999516782,
      "loss": 2.0804,
      "step": 55364
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7592811584472656,
      "learning_rate": 0.0005184992970145309,
      "loss": 2.0517,
      "step": 55365
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.9713371396064758,
      "learning_rate": 0.0005184964940367604,
      "loss": 2.2387,
      "step": 55366
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7095356583595276,
      "learning_rate": 0.0005184936910183673,
      "loss": 2.2446,
      "step": 55367
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7295043468475342,
      "learning_rate": 0.0005184908879593521,
      "loss": 2.2591,
      "step": 55368
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7475628852844238,
      "learning_rate": 0.0005184880848597153,
      "loss": 2.2989,
      "step": 55369
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.798110842704773,
      "learning_rate": 0.0005184852817194575,
      "loss": 2.2895,
      "step": 55370
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.8321564793586731,
      "learning_rate": 0.0005184824785385792,
      "loss": 2.3448,
      "step": 55371
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7209010124206543,
      "learning_rate": 0.0005184796753170808,
      "loss": 2.2128,
      "step": 55372
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7228266596794128,
      "learning_rate": 0.0005184768720549629,
      "loss": 2.2689,
      "step": 55373
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.694806694984436,
      "learning_rate": 0.000518474068752226,
      "loss": 2.1405,
      "step": 55374
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.6441392302513123,
      "learning_rate": 0.0005184712654088707,
      "loss": 2.0521,
      "step": 55375
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7116801142692566,
      "learning_rate": 0.0005184684620248975,
      "loss": 2.2562,
      "step": 55376
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.8407931327819824,
      "learning_rate": 0.0005184656586003068,
      "loss": 2.44,
      "step": 55377
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7929914593696594,
      "learning_rate": 0.0005184628551350993,
      "loss": 2.3612,
      "step": 55378
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7875832915306091,
      "learning_rate": 0.0005184600516292753,
      "loss": 2.4312,
      "step": 55379
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7109509110450745,
      "learning_rate": 0.0005184572480828356,
      "loss": 2.5273,
      "step": 55380
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.9161102175712585,
      "learning_rate": 0.0005184544444957805,
      "loss": 2.3842,
      "step": 55381
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7906438112258911,
      "learning_rate": 0.0005184516408681106,
      "loss": 2.2604,
      "step": 55382
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7778420448303223,
      "learning_rate": 0.0005184488371998264,
      "loss": 2.18,
      "step": 55383
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7888975739479065,
      "learning_rate": 0.0005184460334909284,
      "loss": 2.3335,
      "step": 55384
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7228596806526184,
      "learning_rate": 0.0005184432297414172,
      "loss": 2.3267,
      "step": 55385
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7228099703788757,
      "learning_rate": 0.0005184404259512933,
      "loss": 2.2606,
      "step": 55386
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7316207885742188,
      "learning_rate": 0.0005184376221205573,
      "loss": 2.1123,
      "step": 55387
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7637287378311157,
      "learning_rate": 0.0005184348182492095,
      "loss": 2.0479,
      "step": 55388
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.6880776882171631,
      "learning_rate": 0.0005184320143372506,
      "loss": 2.4272,
      "step": 55389
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.722708523273468,
      "learning_rate": 0.000518429210384681,
      "loss": 2.1559,
      "step": 55390
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7232299447059631,
      "learning_rate": 0.0005184264063915013,
      "loss": 2.0737,
      "step": 55391
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.8632892370223999,
      "learning_rate": 0.0005184236023577121,
      "loss": 2.1662,
      "step": 55392
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.6854923367500305,
      "learning_rate": 0.0005184207982833137,
      "loss": 2.475,
      "step": 55393
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7210283279418945,
      "learning_rate": 0.0005184179941683068,
      "loss": 2.1549,
      "step": 55394
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.686445415019989,
      "learning_rate": 0.0005184151900126919,
      "loss": 2.4004,
      "step": 55395
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7650344967842102,
      "learning_rate": 0.0005184123858164694,
      "loss": 2.4398,
      "step": 55396
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.6783440709114075,
      "learning_rate": 0.00051840958157964,
      "loss": 2.1541,
      "step": 55397
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.6710197329521179,
      "learning_rate": 0.0005184067773022042,
      "loss": 2.2584,
      "step": 55398
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7457711696624756,
      "learning_rate": 0.0005184039729841624,
      "loss": 2.2036,
      "step": 55399
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.690868616104126,
      "learning_rate": 0.0005184011686255151,
      "loss": 2.1675,
      "step": 55400
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7410666346549988,
      "learning_rate": 0.000518398364226263,
      "loss": 2.2395,
      "step": 55401
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.6957713961601257,
      "learning_rate": 0.0005183955597864064,
      "loss": 2.4059,
      "step": 55402
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7586862444877625,
      "learning_rate": 0.000518392755305946,
      "loss": 2.3501,
      "step": 55403
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7201209664344788,
      "learning_rate": 0.0005183899507848823,
      "loss": 2.2144,
      "step": 55404
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.1158348321914673,
      "learning_rate": 0.0005183871462232159,
      "loss": 2.0598,
      "step": 55405
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.8870580196380615,
      "learning_rate": 0.000518384341620947,
      "loss": 2.4752,
      "step": 55406
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.6541267037391663,
      "learning_rate": 0.0005183815369780764,
      "loss": 2.2044,
      "step": 55407
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7002577781677246,
      "learning_rate": 0.0005183787322946045,
      "loss": 2.4045,
      "step": 55408
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7334044575691223,
      "learning_rate": 0.000518375927570532,
      "loss": 2.4863,
      "step": 55409
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7318649291992188,
      "learning_rate": 0.0005183731228058592,
      "loss": 2.1752,
      "step": 55410
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7359327673912048,
      "learning_rate": 0.0005183703180005868,
      "loss": 2.3092,
      "step": 55411
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.8772323131561279,
      "learning_rate": 0.0005183675131547152,
      "loss": 2.4106,
      "step": 55412
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7304777503013611,
      "learning_rate": 0.0005183647082682449,
      "loss": 2.3246,
      "step": 55413
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7295218706130981,
      "learning_rate": 0.0005183619033411764,
      "loss": 2.1066,
      "step": 55414
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.8735884428024292,
      "learning_rate": 0.0005183590983735104,
      "loss": 1.9905,
      "step": 55415
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.732616126537323,
      "learning_rate": 0.0005183562933652474,
      "loss": 2.3566,
      "step": 55416
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.6887636184692383,
      "learning_rate": 0.0005183534883163878,
      "loss": 2.4926,
      "step": 55417
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7672696113586426,
      "learning_rate": 0.0005183506832269322,
      "loss": 2.3466,
      "step": 55418
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7422870993614197,
      "learning_rate": 0.000518347878096881,
      "loss": 2.227,
      "step": 55419
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7296718955039978,
      "learning_rate": 0.0005183450729262348,
      "loss": 2.2039,
      "step": 55420
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7172022461891174,
      "learning_rate": 0.0005183422677149942,
      "loss": 2.176,
      "step": 55421
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7154203653335571,
      "learning_rate": 0.0005183394624631595,
      "loss": 2.2779,
      "step": 55422
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.8229022026062012,
      "learning_rate": 0.0005183366571707316,
      "loss": 2.1877,
      "step": 55423
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7208904027938843,
      "learning_rate": 0.0005183338518377107,
      "loss": 2.0277,
      "step": 55424
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.6960113048553467,
      "learning_rate": 0.0005183310464640973,
      "loss": 2.3257,
      "step": 55425
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7529470920562744,
      "learning_rate": 0.0005183282410498922,
      "loss": 2.3098,
      "step": 55426
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.8440660238265991,
      "learning_rate": 0.0005183254355950956,
      "loss": 2.1346,
      "step": 55427
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7752150893211365,
      "learning_rate": 0.0005183226300997083,
      "loss": 2.3206,
      "step": 55428
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.8967970013618469,
      "learning_rate": 0.0005183198245637307,
      "loss": 2.4252,
      "step": 55429
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7397697567939758,
      "learning_rate": 0.0005183170189871633,
      "loss": 2.3474,
      "step": 55430
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.6537266969680786,
      "learning_rate": 0.0005183142133700066,
      "loss": 2.2299,
      "step": 55431
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7532586455345154,
      "learning_rate": 0.0005183114077122612,
      "loss": 2.1717,
      "step": 55432
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7520478367805481,
      "learning_rate": 0.0005183086020139276,
      "loss": 2.1019,
      "step": 55433
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.76242995262146,
      "learning_rate": 0.0005183057962750064,
      "loss": 1.9562,
      "step": 55434
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7189421057701111,
      "learning_rate": 0.0005183029904954979,
      "loss": 2.2731,
      "step": 55435
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7070180773735046,
      "learning_rate": 0.0005183001846754028,
      "loss": 2.1366,
      "step": 55436
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.8241018652915955,
      "learning_rate": 0.0005182973788147215,
      "loss": 2.1006,
      "step": 55437
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7295582294464111,
      "learning_rate": 0.0005182945729134546,
      "loss": 2.4425,
      "step": 55438
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7261331677436829,
      "learning_rate": 0.0005182917669716028,
      "loss": 2.2166,
      "step": 55439
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.9019696116447449,
      "learning_rate": 0.0005182889609891663,
      "loss": 2.3199,
      "step": 55440
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.772678792476654,
      "learning_rate": 0.0005182861549661459,
      "loss": 2.4592,
      "step": 55441
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.899380087852478,
      "learning_rate": 0.0005182833489025418,
      "loss": 2.3362,
      "step": 55442
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.0137510299682617,
      "learning_rate": 0.0005182805427983547,
      "loss": 2.1617,
      "step": 55443
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7142717838287354,
      "learning_rate": 0.0005182777366535852,
      "loss": 2.1462,
      "step": 55444
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.8028250336647034,
      "learning_rate": 0.0005182749304682338,
      "loss": 2.1506,
      "step": 55445
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.0670944452285767,
      "learning_rate": 0.0005182721242423009,
      "loss": 2.3645,
      "step": 55446
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7779586911201477,
      "learning_rate": 0.0005182693179757871,
      "loss": 2.1189,
      "step": 55447
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7999882698059082,
      "learning_rate": 0.000518266511668693,
      "loss": 2.2328,
      "step": 55448
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7995112538337708,
      "learning_rate": 0.0005182637053210189,
      "loss": 2.0949,
      "step": 55449
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.9889602065086365,
      "learning_rate": 0.0005182608989327656,
      "loss": 2.2375,
      "step": 55450
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.743416965007782,
      "learning_rate": 0.0005182580925039333,
      "loss": 2.2086,
      "step": 55451
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.9927856922149658,
      "learning_rate": 0.0005182552860345228,
      "loss": 2.2527,
      "step": 55452
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.0355942249298096,
      "learning_rate": 0.0005182524795245345,
      "loss": 2.0816,
      "step": 55453
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.753931999206543,
      "learning_rate": 0.000518249672973969,
      "loss": 2.3675,
      "step": 55454
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7929202914237976,
      "learning_rate": 0.0005182468663828268,
      "loss": 2.3864,
      "step": 55455
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.9452279210090637,
      "learning_rate": 0.0005182440597511082,
      "loss": 2.3209,
      "step": 55456
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.9039356112480164,
      "learning_rate": 0.0005182412530788141,
      "loss": 2.1539,
      "step": 55457
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.6968629360198975,
      "learning_rate": 0.0005182384463659447,
      "loss": 2.3434,
      "step": 55458
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7898803949356079,
      "learning_rate": 0.0005182356396125007,
      "loss": 2.1858,
      "step": 55459
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.2496883869171143,
      "learning_rate": 0.0005182328328184826,
      "loss": 2.2938,
      "step": 55460
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.9836259484291077,
      "learning_rate": 0.0005182300259838909,
      "loss": 2.2465,
      "step": 55461
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7259624600410461,
      "learning_rate": 0.0005182272191087261,
      "loss": 2.5488,
      "step": 55462
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.8647617101669312,
      "learning_rate": 0.0005182244121929887,
      "loss": 2.2325,
      "step": 55463
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.8483923077583313,
      "learning_rate": 0.0005182216052366794,
      "loss": 2.1914,
      "step": 55464
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7847205996513367,
      "learning_rate": 0.0005182187982397985,
      "loss": 2.1456,
      "step": 55465
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.6575606465339661,
      "learning_rate": 0.0005182159912023465,
      "loss": 2.1744,
      "step": 55466
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7712724804878235,
      "learning_rate": 0.0005182131841243242,
      "loss": 2.2462,
      "step": 55467
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.8204551339149475,
      "learning_rate": 0.0005182103770057318,
      "loss": 1.9525,
      "step": 55468
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.8470836877822876,
      "learning_rate": 0.00051820756984657,
      "loss": 2.1136,
      "step": 55469
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.6931322813034058,
      "learning_rate": 0.0005182047626468394,
      "loss": 2.3237,
      "step": 55470
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7332943081855774,
      "learning_rate": 0.0005182019554065404,
      "loss": 2.1804,
      "step": 55471
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.8694064021110535,
      "learning_rate": 0.0005181991481256735,
      "loss": 2.3035,
      "step": 55472
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.6829938888549805,
      "learning_rate": 0.0005181963408042392,
      "loss": 2.1577,
      "step": 55473
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.8230358958244324,
      "learning_rate": 0.0005181935334422381,
      "loss": 2.3885,
      "step": 55474
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.657585084438324,
      "learning_rate": 0.0005181907260396707,
      "loss": 2.3329,
      "step": 55475
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.8357337713241577,
      "learning_rate": 0.0005181879185965376,
      "loss": 2.3231,
      "step": 55476
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7090486884117126,
      "learning_rate": 0.0005181851111128393,
      "loss": 2.0447,
      "step": 55477
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7181804180145264,
      "learning_rate": 0.0005181823035885761,
      "loss": 2.1072,
      "step": 55478
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7930432558059692,
      "learning_rate": 0.0005181794960237488,
      "loss": 2.4466,
      "step": 55479
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7097144722938538,
      "learning_rate": 0.0005181766884183578,
      "loss": 2.1765,
      "step": 55480
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.6877281665802002,
      "learning_rate": 0.0005181738807724035,
      "loss": 2.2055,
      "step": 55481
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.6949396729469299,
      "learning_rate": 0.0005181710730858868,
      "loss": 2.0974,
      "step": 55482
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7048989534378052,
      "learning_rate": 0.0005181682653588078,
      "loss": 2.1188,
      "step": 55483
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.662822425365448,
      "learning_rate": 0.0005181654575911672,
      "loss": 2.0848,
      "step": 55484
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.8232846856117249,
      "learning_rate": 0.0005181626497829656,
      "loss": 2.2678,
      "step": 55485
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7919673323631287,
      "learning_rate": 0.0005181598419342035,
      "loss": 2.327,
      "step": 55486
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7396233677864075,
      "learning_rate": 0.0005181570340448812,
      "loss": 2.3504,
      "step": 55487
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7802349925041199,
      "learning_rate": 0.0005181542261149995,
      "loss": 2.5218,
      "step": 55488
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.6978361010551453,
      "learning_rate": 0.0005181514181445588,
      "loss": 2.3619,
      "step": 55489
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7450432181358337,
      "learning_rate": 0.0005181486101335596,
      "loss": 2.2384,
      "step": 55490
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.759750247001648,
      "learning_rate": 0.0005181458020820025,
      "loss": 2.2535,
      "step": 55491
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.6938983201980591,
      "learning_rate": 0.000518142993989888,
      "loss": 2.3154,
      "step": 55492
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.8164542317390442,
      "learning_rate": 0.0005181401858572165,
      "loss": 2.1888,
      "step": 55493
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7033483386039734,
      "learning_rate": 0.0005181373776839888,
      "loss": 2.1665,
      "step": 55494
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.6964625120162964,
      "learning_rate": 0.000518134569470205,
      "loss": 1.973,
      "step": 55495
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7706451416015625,
      "learning_rate": 0.0005181317612158661,
      "loss": 2.3493,
      "step": 55496
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7278476357460022,
      "learning_rate": 0.0005181289529209722,
      "loss": 2.4066,
      "step": 55497
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7578877806663513,
      "learning_rate": 0.0005181261445855242,
      "loss": 2.2278,
      "step": 55498
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7700433731079102,
      "learning_rate": 0.0005181233362095222,
      "loss": 2.187,
      "step": 55499
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7480932474136353,
      "learning_rate": 0.0005181205277929672,
      "loss": 2.0919,
      "step": 55500
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7705210447311401,
      "learning_rate": 0.0005181177193358592,
      "loss": 2.3685,
      "step": 55501
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.669885516166687,
      "learning_rate": 0.0005181149108381992,
      "loss": 2.3637,
      "step": 55502
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.724289059638977,
      "learning_rate": 0.0005181121022999874,
      "loss": 2.0998,
      "step": 55503
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.8352439999580383,
      "learning_rate": 0.0005181092937212245,
      "loss": 2.3177,
      "step": 55504
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7284672856330872,
      "learning_rate": 0.0005181064851019111,
      "loss": 2.1419,
      "step": 55505
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7787251472473145,
      "learning_rate": 0.0005181036764420474,
      "loss": 2.4973,
      "step": 55506
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.6981839537620544,
      "learning_rate": 0.0005181008677416342,
      "loss": 2.2859,
      "step": 55507
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7348374128341675,
      "learning_rate": 0.0005180980590006719,
      "loss": 2.345,
      "step": 55508
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7848836183547974,
      "learning_rate": 0.000518095250219161,
      "loss": 2.2489,
      "step": 55509
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7215482592582703,
      "learning_rate": 0.0005180924413971022,
      "loss": 2.1837,
      "step": 55510
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.713463306427002,
      "learning_rate": 0.0005180896325344958,
      "loss": 2.133,
      "step": 55511
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7720184326171875,
      "learning_rate": 0.0005180868236313424,
      "loss": 2.399,
      "step": 55512
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7303605079650879,
      "learning_rate": 0.0005180840146876427,
      "loss": 2.178,
      "step": 55513
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.698260486125946,
      "learning_rate": 0.000518081205703397,
      "loss": 2.1074,
      "step": 55514
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.864824652671814,
      "learning_rate": 0.0005180783966786058,
      "loss": 2.1942,
      "step": 55515
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7636120319366455,
      "learning_rate": 0.0005180755876132698,
      "loss": 2.248,
      "step": 55516
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7205214500427246,
      "learning_rate": 0.0005180727785073894,
      "loss": 2.2211,
      "step": 55517
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7414657473564148,
      "learning_rate": 0.0005180699693609651,
      "loss": 2.0914,
      "step": 55518
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7362595200538635,
      "learning_rate": 0.0005180671601739975,
      "loss": 1.9923,
      "step": 55519
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.6627606749534607,
      "learning_rate": 0.0005180643509464872,
      "loss": 2.163,
      "step": 55520
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7297462224960327,
      "learning_rate": 0.0005180615416784345,
      "loss": 1.9733,
      "step": 55521
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7726840972900391,
      "learning_rate": 0.0005180587323698401,
      "loss": 2.0908,
      "step": 55522
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7223387360572815,
      "learning_rate": 0.0005180559230207044,
      "loss": 2.2335,
      "step": 55523
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7304900288581848,
      "learning_rate": 0.0005180531136310281,
      "loss": 2.343,
      "step": 55524
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7529780864715576,
      "learning_rate": 0.0005180503042008115,
      "loss": 2.1238,
      "step": 55525
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7679207921028137,
      "learning_rate": 0.0005180474947300553,
      "loss": 2.4098,
      "step": 55526
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7696711421012878,
      "learning_rate": 0.0005180446852187599,
      "loss": 2.3048,
      "step": 55527
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.722052812576294,
      "learning_rate": 0.0005180418756669259,
      "loss": 2.2676,
      "step": 55528
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7021798491477966,
      "learning_rate": 0.0005180390660745539,
      "loss": 2.2613,
      "step": 55529
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7846922278404236,
      "learning_rate": 0.0005180362564416442,
      "loss": 2.2912,
      "step": 55530
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.752907395362854,
      "learning_rate": 0.0005180334467681975,
      "loss": 2.3642,
      "step": 55531
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.8453022241592407,
      "learning_rate": 0.0005180306370542141,
      "loss": 2.1907,
      "step": 55532
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.668996274471283,
      "learning_rate": 0.000518027827299695,
      "loss": 2.2165,
      "step": 55533
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.72950279712677,
      "learning_rate": 0.0005180250175046402,
      "loss": 2.4456,
      "step": 55534
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7705941200256348,
      "learning_rate": 0.0005180222076690504,
      "loss": 1.9786,
      "step": 55535
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.6665571331977844,
      "learning_rate": 0.0005180193977929263,
      "loss": 2.4409,
      "step": 55536
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7536192536354065,
      "learning_rate": 0.0005180165878762682,
      "loss": 2.3239,
      "step": 55537
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7129619121551514,
      "learning_rate": 0.0005180137779190768,
      "loss": 2.2584,
      "step": 55538
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7044633626937866,
      "learning_rate": 0.0005180109679213524,
      "loss": 2.4452,
      "step": 55539
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7626052498817444,
      "learning_rate": 0.0005180081578830956,
      "loss": 2.2133,
      "step": 55540
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.6543909311294556,
      "learning_rate": 0.000518005347804307,
      "loss": 2.2735,
      "step": 55541
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7742694020271301,
      "learning_rate": 0.0005180025376849873,
      "loss": 2.3013,
      "step": 55542
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.8332974314689636,
      "learning_rate": 0.0005179997275251366,
      "loss": 2.5103,
      "step": 55543
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7158398628234863,
      "learning_rate": 0.0005179969173247556,
      "loss": 2.3604,
      "step": 55544
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7687630653381348,
      "learning_rate": 0.000517994107083845,
      "loss": 2.2162,
      "step": 55545
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7501387596130371,
      "learning_rate": 0.0005179912968024052,
      "loss": 2.1319,
      "step": 55546
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7293558120727539,
      "learning_rate": 0.0005179884864804366,
      "loss": 2.3211,
      "step": 55547
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7368017435073853,
      "learning_rate": 0.0005179856761179398,
      "loss": 2.1968,
      "step": 55548
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.818504810333252,
      "learning_rate": 0.0005179828657149155,
      "loss": 2.4023,
      "step": 55549
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.747207760810852,
      "learning_rate": 0.0005179800552713639,
      "loss": 2.3877,
      "step": 55550
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7738780975341797,
      "learning_rate": 0.0005179772447872858,
      "loss": 2.3256,
      "step": 55551
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.9831733107566833,
      "learning_rate": 0.0005179744342626815,
      "loss": 2.1735,
      "step": 55552
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7426365613937378,
      "learning_rate": 0.0005179716236975518,
      "loss": 2.0892,
      "step": 55553
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7465500831604004,
      "learning_rate": 0.000517968813091897,
      "loss": 2.2337,
      "step": 55554
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7752686738967896,
      "learning_rate": 0.0005179660024457177,
      "loss": 2.0962,
      "step": 55555
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7532029151916504,
      "learning_rate": 0.0005179631917590143,
      "loss": 2.365,
      "step": 55556
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7036157846450806,
      "learning_rate": 0.0005179603810317875,
      "loss": 2.0938,
      "step": 55557
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7135667204856873,
      "learning_rate": 0.0005179575702640378,
      "loss": 2.3265,
      "step": 55558
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.6963984370231628,
      "learning_rate": 0.0005179547594557657,
      "loss": 2.3323,
      "step": 55559
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7153477668762207,
      "learning_rate": 0.0005179519486069716,
      "loss": 2.2928,
      "step": 55560
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7490202784538269,
      "learning_rate": 0.000517949137717656,
      "loss": 2.1919,
      "step": 55561
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.8494240045547485,
      "learning_rate": 0.0005179463267878198,
      "loss": 1.9031,
      "step": 55562
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7998002171516418,
      "learning_rate": 0.000517943515817463,
      "loss": 2.2543,
      "step": 55563
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7320318818092346,
      "learning_rate": 0.0005179407048065866,
      "loss": 2.3278,
      "step": 55564
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7180943489074707,
      "learning_rate": 0.0005179378937551908,
      "loss": 2.2345,
      "step": 55565
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7679635882377625,
      "learning_rate": 0.0005179350826632764,
      "loss": 2.4827,
      "step": 55566
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7515372633934021,
      "learning_rate": 0.0005179322715308435,
      "loss": 2.1214,
      "step": 55567
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.6927077174186707,
      "learning_rate": 0.0005179294603578931,
      "loss": 2.1839,
      "step": 55568
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7354919910430908,
      "learning_rate": 0.0005179266491444254,
      "loss": 2.1755,
      "step": 55569
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.6518471240997314,
      "learning_rate": 0.0005179238378904409,
      "loss": 2.2097,
      "step": 55570
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7107179760932922,
      "learning_rate": 0.0005179210265959405,
      "loss": 2.3217,
      "step": 55571
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7183928489685059,
      "learning_rate": 0.0005179182152609243,
      "loss": 2.0869,
      "step": 55572
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.8234042525291443,
      "learning_rate": 0.000517915403885393,
      "loss": 2.1391,
      "step": 55573
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7218043804168701,
      "learning_rate": 0.0005179125924693472,
      "loss": 2.378,
      "step": 55574
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7378962635993958,
      "learning_rate": 0.0005179097810127872,
      "loss": 2.0902,
      "step": 55575
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7566026449203491,
      "learning_rate": 0.0005179069695157139,
      "loss": 2.3735,
      "step": 55576
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.8035424947738647,
      "learning_rate": 0.0005179041579781273,
      "loss": 2.1609,
      "step": 55577
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7195580005645752,
      "learning_rate": 0.0005179013464000283,
      "loss": 2.31,
      "step": 55578
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.6892977356910706,
      "learning_rate": 0.0005178985347814173,
      "loss": 2.4012,
      "step": 55579
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7470816969871521,
      "learning_rate": 0.000517895723122295,
      "loss": 2.4453,
      "step": 55580
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.739848256111145,
      "learning_rate": 0.0005178929114226617,
      "loss": 2.2504,
      "step": 55581
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.711580216884613,
      "learning_rate": 0.0005178900996825179,
      "loss": 2.3508,
      "step": 55582
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.6922894716262817,
      "learning_rate": 0.0005178872879018643,
      "loss": 2.4863,
      "step": 55583
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7193270921707153,
      "learning_rate": 0.0005178844760807014,
      "loss": 2.199,
      "step": 55584
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.716037929058075,
      "learning_rate": 0.0005178816642190295,
      "loss": 2.2498,
      "step": 55585
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.663442075252533,
      "learning_rate": 0.0005178788523168493,
      "loss": 2.3048,
      "step": 55586
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.8010544776916504,
      "learning_rate": 0.0005178760403741615,
      "loss": 2.2227,
      "step": 55587
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7900527715682983,
      "learning_rate": 0.0005178732283909663,
      "loss": 2.1187,
      "step": 55588
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7373901009559631,
      "learning_rate": 0.0005178704163672643,
      "loss": 2.1669,
      "step": 55589
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.6787323355674744,
      "learning_rate": 0.0005178676043030561,
      "loss": 2.343,
      "step": 55590
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7549370527267456,
      "learning_rate": 0.0005178647921983423,
      "loss": 2.2675,
      "step": 55591
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.8366743922233582,
      "learning_rate": 0.0005178619800531233,
      "loss": 1.9969,
      "step": 55592
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.752615749835968,
      "learning_rate": 0.0005178591678673996,
      "loss": 2.3461,
      "step": 55593
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7390809059143066,
      "learning_rate": 0.0005178563556411718,
      "loss": 2.395,
      "step": 55594
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.6858318448066711,
      "learning_rate": 0.0005178535433744403,
      "loss": 2.4209,
      "step": 55595
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.8042978048324585,
      "learning_rate": 0.0005178507310672058,
      "loss": 2.2043,
      "step": 55596
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7283608317375183,
      "learning_rate": 0.0005178479187194687,
      "loss": 2.1355,
      "step": 55597
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.6931050419807434,
      "learning_rate": 0.0005178451063312296,
      "loss": 2.1847,
      "step": 55598
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.8623964786529541,
      "learning_rate": 0.000517842293902489,
      "loss": 2.2071,
      "step": 55599
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.8068014979362488,
      "learning_rate": 0.0005178394814332473,
      "loss": 2.2085,
      "step": 55600
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7775441408157349,
      "learning_rate": 0.0005178366689235052,
      "loss": 2.4412,
      "step": 55601
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.9211675524711609,
      "learning_rate": 0.0005178338563732631,
      "loss": 2.1131,
      "step": 55602
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7538178563117981,
      "learning_rate": 0.0005178310437825216,
      "loss": 2.5682,
      "step": 55603
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.6752071976661682,
      "learning_rate": 0.0005178282311512813,
      "loss": 2.2382,
      "step": 55604
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.6666321158409119,
      "learning_rate": 0.0005178254184795425,
      "loss": 2.2584,
      "step": 55605
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.8405789136886597,
      "learning_rate": 0.0005178226057673059,
      "loss": 2.2022,
      "step": 55606
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7061046957969666,
      "learning_rate": 0.000517819793014572,
      "loss": 2.2805,
      "step": 55607
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.6906201839447021,
      "learning_rate": 0.0005178169802213412,
      "loss": 2.4536,
      "step": 55608
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7380641102790833,
      "learning_rate": 0.0005178141673876141,
      "loss": 2.247,
      "step": 55609
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7507467269897461,
      "learning_rate": 0.0005178113545133914,
      "loss": 1.9615,
      "step": 55610
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7393759489059448,
      "learning_rate": 0.0005178085415986733,
      "loss": 2.1084,
      "step": 55611
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.6806057691574097,
      "learning_rate": 0.0005178057286434604,
      "loss": 2.1001,
      "step": 55612
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.787205159664154,
      "learning_rate": 0.0005178029156477534,
      "loss": 2.1022,
      "step": 55613
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7278212904930115,
      "learning_rate": 0.0005178001026115528,
      "loss": 2.2724,
      "step": 55614
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7344293594360352,
      "learning_rate": 0.000517797289534859,
      "loss": 2.2416,
      "step": 55615
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7424655556678772,
      "learning_rate": 0.0005177944764176726,
      "loss": 2.2319,
      "step": 55616
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.6735600233078003,
      "learning_rate": 0.000517791663259994,
      "loss": 2.1743,
      "step": 55617
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7369029521942139,
      "learning_rate": 0.000517788850061824,
      "loss": 1.9917,
      "step": 55618
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7373586297035217,
      "learning_rate": 0.0005177860368231628,
      "loss": 2.3885,
      "step": 55619
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7839221358299255,
      "learning_rate": 0.0005177832235440111,
      "loss": 2.1854,
      "step": 55620
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7167964577674866,
      "learning_rate": 0.0005177804102243694,
      "loss": 2.2024,
      "step": 55621
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.6863805055618286,
      "learning_rate": 0.0005177775968642381,
      "loss": 2.3076,
      "step": 55622
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.979480504989624,
      "learning_rate": 0.0005177747834636179,
      "loss": 2.0734,
      "step": 55623
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7154159545898438,
      "learning_rate": 0.0005177719700225094,
      "loss": 2.1798,
      "step": 55624
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7129747271537781,
      "learning_rate": 0.0005177691565409128,
      "loss": 2.2412,
      "step": 55625
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7203055024147034,
      "learning_rate": 0.0005177663430188289,
      "loss": 2.0456,
      "step": 55626
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7445189356803894,
      "learning_rate": 0.0005177635294562581,
      "loss": 1.8522,
      "step": 55627
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7531322240829468,
      "learning_rate": 0.000517760715853201,
      "loss": 2.2918,
      "step": 55628
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.0122932195663452,
      "learning_rate": 0.0005177579022096579,
      "loss": 2.2671,
      "step": 55629
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.904717206954956,
      "learning_rate": 0.0005177550885256297,
      "loss": 2.071,
      "step": 55630
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7062270045280457,
      "learning_rate": 0.0005177522748011166,
      "loss": 2.3247,
      "step": 55631
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7382931113243103,
      "learning_rate": 0.0005177494610361192,
      "loss": 2.3438,
      "step": 55632
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7513936161994934,
      "learning_rate": 0.0005177466472306381,
      "loss": 2.452,
      "step": 55633
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7006000876426697,
      "learning_rate": 0.0005177438333846739,
      "loss": 2.2889,
      "step": 55634
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.8345584869384766,
      "learning_rate": 0.0005177410194982269,
      "loss": 2.2207,
      "step": 55635
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.9182789325714111,
      "learning_rate": 0.0005177382055712978,
      "loss": 2.2982,
      "step": 55636
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.6986154913902283,
      "learning_rate": 0.000517735391603887,
      "loss": 2.2219,
      "step": 55637
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.6719643473625183,
      "learning_rate": 0.000517732577595995,
      "loss": 2.2141,
      "step": 55638
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.6863247156143188,
      "learning_rate": 0.0005177297635476226,
      "loss": 2.2796,
      "step": 55639
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7622097134590149,
      "learning_rate": 0.00051772694945877,
      "loss": 2.2041,
      "step": 55640
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7562733292579651,
      "learning_rate": 0.0005177241353294379,
      "loss": 2.1803,
      "step": 55641
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.728561520576477,
      "learning_rate": 0.0005177213211596268,
      "loss": 2.3014,
      "step": 55642
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7481913566589355,
      "learning_rate": 0.0005177185069493371,
      "loss": 2.0759,
      "step": 55643
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7892385125160217,
      "learning_rate": 0.0005177156926985695,
      "loss": 2.1053,
      "step": 55644
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7338718771934509,
      "learning_rate": 0.0005177128784073245,
      "loss": 2.3555,
      "step": 55645
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7454615831375122,
      "learning_rate": 0.0005177100640756025,
      "loss": 2.034,
      "step": 55646
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7515079975128174,
      "learning_rate": 0.0005177072497034041,
      "loss": 2.135,
      "step": 55647
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7018864750862122,
      "learning_rate": 0.0005177044352907297,
      "loss": 2.1318,
      "step": 55648
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.8335319757461548,
      "learning_rate": 0.00051770162083758,
      "loss": 2.2952,
      "step": 55649
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7585796117782593,
      "learning_rate": 0.0005176988063439555,
      "loss": 2.0595,
      "step": 55650
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.8090148568153381,
      "learning_rate": 0.0005176959918098567,
      "loss": 2.3971,
      "step": 55651
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7137030959129333,
      "learning_rate": 0.000517693177235284,
      "loss": 2.3179,
      "step": 55652
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7785164713859558,
      "learning_rate": 0.0005176903626202381,
      "loss": 1.9935,
      "step": 55653
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7231332063674927,
      "learning_rate": 0.0005176875479647195,
      "loss": 2.1911,
      "step": 55654
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7625353336334229,
      "learning_rate": 0.0005176847332687286,
      "loss": 2.3508,
      "step": 55655
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.8679313659667969,
      "learning_rate": 0.0005176819185322661,
      "loss": 2.4562,
      "step": 55656
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.6769853830337524,
      "learning_rate": 0.0005176791037553323,
      "loss": 2.489,
      "step": 55657
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7352830171585083,
      "learning_rate": 0.0005176762889379278,
      "loss": 2.1116,
      "step": 55658
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7025669813156128,
      "learning_rate": 0.0005176734740800533,
      "loss": 2.0616,
      "step": 55659
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.6974838376045227,
      "learning_rate": 0.0005176706591817092,
      "loss": 2.1356,
      "step": 55660
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7049886584281921,
      "learning_rate": 0.000517667844242896,
      "loss": 2.096,
      "step": 55661
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7015830874443054,
      "learning_rate": 0.0005176650292636142,
      "loss": 2.3997,
      "step": 55662
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.8117110729217529,
      "learning_rate": 0.0005176622142438644,
      "loss": 2.0014,
      "step": 55663
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7424699068069458,
      "learning_rate": 0.000517659399183647,
      "loss": 2.2045,
      "step": 55664
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7257530689239502,
      "learning_rate": 0.0005176565840829627,
      "loss": 2.1148,
      "step": 55665
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7260540723800659,
      "learning_rate": 0.0005176537689418119,
      "loss": 2.4629,
      "step": 55666
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.8349752426147461,
      "learning_rate": 0.0005176509537601951,
      "loss": 2.1449,
      "step": 55667
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.8331542611122131,
      "learning_rate": 0.000517648138538113,
      "loss": 2.2002,
      "step": 55668
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7451001405715942,
      "learning_rate": 0.0005176453232755659,
      "loss": 2.1798,
      "step": 55669
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7930333018302917,
      "learning_rate": 0.0005176425079725544,
      "loss": 2.108,
      "step": 55670
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7975085377693176,
      "learning_rate": 0.0005176396926290792,
      "loss": 2.17,
      "step": 55671
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.6533166766166687,
      "learning_rate": 0.0005176368772451406,
      "loss": 2.2869,
      "step": 55672
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7525798678398132,
      "learning_rate": 0.0005176340618207392,
      "loss": 2.2385,
      "step": 55673
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7233511209487915,
      "learning_rate": 0.0005176312463558754,
      "loss": 2.3782,
      "step": 55674
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7243472933769226,
      "learning_rate": 0.00051762843085055,
      "loss": 2.2904,
      "step": 55675
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.6988710165023804,
      "learning_rate": 0.0005176256153047633,
      "loss": 2.1919,
      "step": 55676
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7110689878463745,
      "learning_rate": 0.0005176227997185159,
      "loss": 2.4084,
      "step": 55677
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7385907769203186,
      "learning_rate": 0.0005176199840918082,
      "loss": 2.0783,
      "step": 55678
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.9389095306396484,
      "learning_rate": 0.0005176171684246412,
      "loss": 2.402,
      "step": 55679
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.6800438165664673,
      "learning_rate": 0.0005176143527170147,
      "loss": 2.1498,
      "step": 55680
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.8476948738098145,
      "learning_rate": 0.0005176115369689297,
      "loss": 2.3064,
      "step": 55681
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.9514417052268982,
      "learning_rate": 0.0005176087211803866,
      "loss": 2.2999,
      "step": 55682
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.775216281414032,
      "learning_rate": 0.000517605905351386,
      "loss": 2.2556,
      "step": 55683
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.853644609451294,
      "learning_rate": 0.0005176030894819282,
      "loss": 2.1471,
      "step": 55684
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.8659172058105469,
      "learning_rate": 0.0005176002735720139,
      "loss": 2.2092,
      "step": 55685
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.8650778532028198,
      "learning_rate": 0.0005175974576216438,
      "loss": 2.196,
      "step": 55686
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7300186157226562,
      "learning_rate": 0.000517594641630818,
      "loss": 2.0618,
      "step": 55687
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7637701034545898,
      "learning_rate": 0.0005175918255995375,
      "loss": 2.0815,
      "step": 55688
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.9492164254188538,
      "learning_rate": 0.0005175890095278023,
      "loss": 2.326,
      "step": 55689
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7652392983436584,
      "learning_rate": 0.0005175861934156134,
      "loss": 1.8982,
      "step": 55690
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7658538222312927,
      "learning_rate": 0.000517583377262971,
      "loss": 2.1498,
      "step": 55691
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.9005677700042725,
      "learning_rate": 0.0005175805610698757,
      "loss": 2.2008,
      "step": 55692
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.8122503757476807,
      "learning_rate": 0.0005175777448363282,
      "loss": 2.2488,
      "step": 55693
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7731727361679077,
      "learning_rate": 0.0005175749285623288,
      "loss": 2.1258,
      "step": 55694
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7647645473480225,
      "learning_rate": 0.0005175721122478781,
      "loss": 2.2675,
      "step": 55695
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7155066728591919,
      "learning_rate": 0.0005175692958929767,
      "loss": 2.3251,
      "step": 55696
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.8862811923027039,
      "learning_rate": 0.0005175664794976251,
      "loss": 2.1432,
      "step": 55697
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7340697050094604,
      "learning_rate": 0.0005175636630618237,
      "loss": 2.1626,
      "step": 55698
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7719773054122925,
      "learning_rate": 0.000517560846585573,
      "loss": 2.266,
      "step": 55699
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.9892227649688721,
      "learning_rate": 0.0005175580300688739,
      "loss": 2.4095,
      "step": 55700
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7695756554603577,
      "learning_rate": 0.0005175552135117265,
      "loss": 2.3462,
      "step": 55701
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.8184307217597961,
      "learning_rate": 0.0005175523969141315,
      "loss": 2.0875,
      "step": 55702
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.8463687300682068,
      "learning_rate": 0.0005175495802760894,
      "loss": 2.1142,
      "step": 55703
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7474346160888672,
      "learning_rate": 0.0005175467635976008,
      "loss": 2.3798,
      "step": 55704
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7123333811759949,
      "learning_rate": 0.000517543946878666,
      "loss": 2.2231,
      "step": 55705
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7641062140464783,
      "learning_rate": 0.0005175411301192858,
      "loss": 2.2764,
      "step": 55706
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7326369881629944,
      "learning_rate": 0.0005175383133194605,
      "loss": 2.1331,
      "step": 55707
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7797797918319702,
      "learning_rate": 0.0005175354964791907,
      "loss": 2.2593,
      "step": 55708
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.680069088935852,
      "learning_rate": 0.0005175326795984771,
      "loss": 2.0923,
      "step": 55709
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7291194796562195,
      "learning_rate": 0.0005175298626773201,
      "loss": 2.0796,
      "step": 55710
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.6858925819396973,
      "learning_rate": 0.00051752704571572,
      "loss": 2.3747,
      "step": 55711
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7781758308410645,
      "learning_rate": 0.0005175242287136776,
      "loss": 2.3924,
      "step": 55712
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7421343326568604,
      "learning_rate": 0.0005175214116711933,
      "loss": 2.3184,
      "step": 55713
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7911255359649658,
      "learning_rate": 0.0005175185945882677,
      "loss": 2.1908,
      "step": 55714
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.736093282699585,
      "learning_rate": 0.0005175157774649013,
      "loss": 2.4639,
      "step": 55715
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7121817469596863,
      "learning_rate": 0.0005175129603010945,
      "loss": 2.2032,
      "step": 55716
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.6868986487388611,
      "learning_rate": 0.0005175101430968482,
      "loss": 2.1792,
      "step": 55717
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7266416549682617,
      "learning_rate": 0.0005175073258521625,
      "loss": 2.2169,
      "step": 55718
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7976199388504028,
      "learning_rate": 0.000517504508567038,
      "loss": 2.3454,
      "step": 55719
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.9126660823822021,
      "learning_rate": 0.0005175016912414754,
      "loss": 2.3337,
      "step": 55720
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7404323220252991,
      "learning_rate": 0.0005174988738754752,
      "loss": 2.2642,
      "step": 55721
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.6959590911865234,
      "learning_rate": 0.0005174960564690377,
      "loss": 2.1066,
      "step": 55722
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7084687948226929,
      "learning_rate": 0.0005174932390221636,
      "loss": 2.2092,
      "step": 55723
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7515656352043152,
      "learning_rate": 0.0005174904215348535,
      "loss": 2.4111,
      "step": 55724
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.8293628692626953,
      "learning_rate": 0.0005174876040071077,
      "loss": 2.2786,
      "step": 55725
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7378895282745361,
      "learning_rate": 0.0005174847864389269,
      "loss": 2.3724,
      "step": 55726
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.6998612284660339,
      "learning_rate": 0.0005174819688303116,
      "loss": 2.2132,
      "step": 55727
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.8338648080825806,
      "learning_rate": 0.0005174791511812623,
      "loss": 2.1036,
      "step": 55728
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7508676052093506,
      "learning_rate": 0.0005174763334917794,
      "loss": 2.3165,
      "step": 55729
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.8691783547401428,
      "learning_rate": 0.0005174735157618638,
      "loss": 2.094,
      "step": 55730
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7577354311943054,
      "learning_rate": 0.0005174706979915156,
      "loss": 2.1684,
      "step": 55731
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.8035088777542114,
      "learning_rate": 0.0005174678801807356,
      "loss": 2.1681,
      "step": 55732
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7041084170341492,
      "learning_rate": 0.0005174650623295241,
      "loss": 1.9341,
      "step": 55733
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7226577997207642,
      "learning_rate": 0.0005174622444378817,
      "loss": 2.1583,
      "step": 55734
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7391203045845032,
      "learning_rate": 0.000517459426505809,
      "loss": 2.0524,
      "step": 55735
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.718969464302063,
      "learning_rate": 0.0005174566085333066,
      "loss": 2.2355,
      "step": 55736
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.6581474542617798,
      "learning_rate": 0.0005174537905203747,
      "loss": 2.0445,
      "step": 55737
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7379631400108337,
      "learning_rate": 0.0005174509724670143,
      "loss": 2.2036,
      "step": 55738
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.6964210271835327,
      "learning_rate": 0.0005174481543732255,
      "loss": 2.3297,
      "step": 55739
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7413152456283569,
      "learning_rate": 0.000517445336239009,
      "loss": 2.2901,
      "step": 55740
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.6891880631446838,
      "learning_rate": 0.0005174425180643653,
      "loss": 2.0759,
      "step": 55741
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7764509320259094,
      "learning_rate": 0.0005174396998492949,
      "loss": 2.5034,
      "step": 55742
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.742621660232544,
      "learning_rate": 0.0005174368815937984,
      "loss": 2.2637,
      "step": 55743
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7922094464302063,
      "learning_rate": 0.0005174340632978764,
      "loss": 2.211,
      "step": 55744
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7356433272361755,
      "learning_rate": 0.0005174312449615291,
      "loss": 2.2871,
      "step": 55745
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.6831933259963989,
      "learning_rate": 0.0005174284265847573,
      "loss": 2.2626,
      "step": 55746
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.8640285134315491,
      "learning_rate": 0.0005174256081675615,
      "loss": 2.2018,
      "step": 55747
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7215244770050049,
      "learning_rate": 0.0005174227897099421,
      "loss": 2.1715,
      "step": 55748
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7343459129333496,
      "learning_rate": 0.0005174199712118997,
      "loss": 2.3707,
      "step": 55749
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.9317077398300171,
      "learning_rate": 0.0005174171526734349,
      "loss": 2.0938,
      "step": 55750
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7480846047401428,
      "learning_rate": 0.0005174143340945481,
      "loss": 2.2999,
      "step": 55751
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7417328357696533,
      "learning_rate": 0.0005174115154752399,
      "loss": 2.3736,
      "step": 55752
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7664569020271301,
      "learning_rate": 0.0005174086968155108,
      "loss": 2.1233,
      "step": 55753
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.8220815658569336,
      "learning_rate": 0.0005174058781153612,
      "loss": 2.4275,
      "step": 55754
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.6743699312210083,
      "learning_rate": 0.0005174030593747919,
      "loss": 2.2781,
      "step": 55755
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.8587762117385864,
      "learning_rate": 0.0005174002405938032,
      "loss": 2.2477,
      "step": 55756
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7161445617675781,
      "learning_rate": 0.0005173974217723957,
      "loss": 2.3046,
      "step": 55757
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.721377432346344,
      "learning_rate": 0.00051739460291057,
      "loss": 2.0852,
      "step": 55758
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.8215591907501221,
      "learning_rate": 0.0005173917840083264,
      "loss": 2.2984,
      "step": 55759
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.8004938364028931,
      "learning_rate": 0.0005173889650656655,
      "loss": 2.2031,
      "step": 55760
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7016977667808533,
      "learning_rate": 0.0005173861460825881,
      "loss": 2.2684,
      "step": 55761
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.683881402015686,
      "learning_rate": 0.0005173833270590945,
      "loss": 2.1015,
      "step": 55762
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7880972027778625,
      "learning_rate": 0.0005173805079951851,
      "loss": 2.2304,
      "step": 55763
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7514058351516724,
      "learning_rate": 0.0005173776888908606,
      "loss": 2.2379,
      "step": 55764
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.8991413116455078,
      "learning_rate": 0.0005173748697461215,
      "loss": 2.2162,
      "step": 55765
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.8282750844955444,
      "learning_rate": 0.0005173720505609683,
      "loss": 2.2815,
      "step": 55766
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.6849814653396606,
      "learning_rate": 0.0005173692313354015,
      "loss": 2.1803,
      "step": 55767
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7933224439620972,
      "learning_rate": 0.0005173664120694217,
      "loss": 2.2882,
      "step": 55768
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.9149799346923828,
      "learning_rate": 0.0005173635927630294,
      "loss": 2.459,
      "step": 55769
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.6997538208961487,
      "learning_rate": 0.0005173607734162251,
      "loss": 2.2176,
      "step": 55770
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.8392587304115295,
      "learning_rate": 0.0005173579540290093,
      "loss": 2.1953,
      "step": 55771
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7499605417251587,
      "learning_rate": 0.0005173551346013825,
      "loss": 2.3407,
      "step": 55772
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7254117727279663,
      "learning_rate": 0.0005173523151333453,
      "loss": 2.2249,
      "step": 55773
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7236061096191406,
      "learning_rate": 0.0005173494956248984,
      "loss": 2.2984,
      "step": 55774
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.8816030025482178,
      "learning_rate": 0.000517346676076042,
      "loss": 1.9981,
      "step": 55775
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.8793239593505859,
      "learning_rate": 0.0005173438564867767,
      "loss": 2.1436,
      "step": 55776
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7281639575958252,
      "learning_rate": 0.000517341036857103,
      "loss": 2.1884,
      "step": 55777
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.8179965615272522,
      "learning_rate": 0.0005173382171870216,
      "loss": 2.159,
      "step": 55778
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7698412537574768,
      "learning_rate": 0.000517335397476533,
      "loss": 2.5361,
      "step": 55779
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.9753709435462952,
      "learning_rate": 0.0005173325777256375,
      "loss": 2.3719,
      "step": 55780
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7352707982063293,
      "learning_rate": 0.0005173297579343358,
      "loss": 2.4218,
      "step": 55781
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.8422396183013916,
      "learning_rate": 0.0005173269381026284,
      "loss": 2.2179,
      "step": 55782
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.8128455281257629,
      "learning_rate": 0.000517324118230516,
      "loss": 2.2243,
      "step": 55783
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.807363748550415,
      "learning_rate": 0.0005173212983179987,
      "loss": 2.1921,
      "step": 55784
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.6832032799720764,
      "learning_rate": 0.0005173184783650774,
      "loss": 2.0379,
      "step": 55785
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7814875841140747,
      "learning_rate": 0.0005173156583717524,
      "loss": 2.2501,
      "step": 55786
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7247174978256226,
      "learning_rate": 0.0005173128383380243,
      "loss": 2.2295,
      "step": 55787
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.8081461787223816,
      "learning_rate": 0.0005173100182638938,
      "loss": 2.1054,
      "step": 55788
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7524368166923523,
      "learning_rate": 0.0005173071981493611,
      "loss": 2.2646,
      "step": 55789
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.6765588521957397,
      "learning_rate": 0.000517304377994427,
      "loss": 2.2017,
      "step": 55790
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7242224216461182,
      "learning_rate": 0.0005173015577990919,
      "loss": 2.2229,
      "step": 55791
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.696439266204834,
      "learning_rate": 0.0005172987375633563,
      "loss": 2.0199,
      "step": 55792
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.8209106922149658,
      "learning_rate": 0.0005172959172872209,
      "loss": 2.1007,
      "step": 55793
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.8461571931838989,
      "learning_rate": 0.0005172930969706859,
      "loss": 2.4701,
      "step": 55794
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7764027714729309,
      "learning_rate": 0.0005172902766137521,
      "loss": 2.1978,
      "step": 55795
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.6979119181632996,
      "learning_rate": 0.0005172874562164198,
      "loss": 2.291,
      "step": 55796
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7442794442176819,
      "learning_rate": 0.0005172846357786899,
      "loss": 2.4268,
      "step": 55797
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.8320220112800598,
      "learning_rate": 0.0005172818153005625,
      "loss": 2.296,
      "step": 55798
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7679381370544434,
      "learning_rate": 0.0005172789947820383,
      "loss": 2.2283,
      "step": 55799
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7208007574081421,
      "learning_rate": 0.000517276174223118,
      "loss": 2.2017,
      "step": 55800
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.8030654191970825,
      "learning_rate": 0.0005172733536238019,
      "loss": 2.1615,
      "step": 55801
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.680575966835022,
      "learning_rate": 0.0005172705329840906,
      "loss": 2.4728,
      "step": 55802
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7508136630058289,
      "learning_rate": 0.0005172677123039844,
      "loss": 2.1049,
      "step": 55803
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.8971908092498779,
      "learning_rate": 0.0005172648915834843,
      "loss": 2.3382,
      "step": 55804
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.9623945355415344,
      "learning_rate": 0.0005172620708225903,
      "loss": 2.404,
      "step": 55805
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.6832913160324097,
      "learning_rate": 0.0005172592500213034,
      "loss": 2.2041,
      "step": 55806
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.1654694080352783,
      "learning_rate": 0.0005172564291796238,
      "loss": 2.362,
      "step": 55807
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7811882495880127,
      "learning_rate": 0.0005172536082975522,
      "loss": 2.1624,
      "step": 55808
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.8465876579284668,
      "learning_rate": 0.000517250787375089,
      "loss": 2.3452,
      "step": 55809
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.8281552791595459,
      "learning_rate": 0.0005172479664122347,
      "loss": 2.2514,
      "step": 55810
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.8479554653167725,
      "learning_rate": 0.0005172451454089901,
      "loss": 2.5349,
      "step": 55811
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7546538710594177,
      "learning_rate": 0.0005172423243653553,
      "loss": 2.3662,
      "step": 55812
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7113215327262878,
      "learning_rate": 0.0005172395032813312,
      "loss": 2.1385,
      "step": 55813
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.721443235874176,
      "learning_rate": 0.0005172366821569181,
      "loss": 2.1617,
      "step": 55814
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7816221714019775,
      "learning_rate": 0.0005172338609921167,
      "loss": 2.2706,
      "step": 55815
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7106003761291504,
      "learning_rate": 0.0005172310397869273,
      "loss": 2.3186,
      "step": 55816
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7715883851051331,
      "learning_rate": 0.0005172282185413507,
      "loss": 2.3822,
      "step": 55817
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7476950287818909,
      "learning_rate": 0.0005172253972553872,
      "loss": 2.2711,
      "step": 55818
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.674475908279419,
      "learning_rate": 0.0005172225759290373,
      "loss": 2.1582,
      "step": 55819
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7650400996208191,
      "learning_rate": 0.0005172197545623017,
      "loss": 2.1715,
      "step": 55820
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7521375417709351,
      "learning_rate": 0.0005172169331551809,
      "loss": 2.2828,
      "step": 55821
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.3039393424987793,
      "learning_rate": 0.0005172141117076753,
      "loss": 2.2622,
      "step": 55822
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.8949105739593506,
      "learning_rate": 0.0005172112902197855,
      "loss": 2.1724,
      "step": 55823
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.076054334640503,
      "learning_rate": 0.000517208468691512,
      "loss": 2.1179,
      "step": 55824
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.8577147722244263,
      "learning_rate": 0.0005172056471228555,
      "loss": 2.2878,
      "step": 55825
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.157019853591919,
      "learning_rate": 0.0005172028255138163,
      "loss": 2.0538,
      "step": 55826
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.8041359782218933,
      "learning_rate": 0.0005172000038643949,
      "loss": 2.1905,
      "step": 55827
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.6947027444839478,
      "learning_rate": 0.000517197182174592,
      "loss": 2.1698,
      "step": 55828
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.947046160697937,
      "learning_rate": 0.0005171943604444079,
      "loss": 2.4167,
      "step": 55829
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.9194814562797546,
      "learning_rate": 0.0005171915386738434,
      "loss": 2.374,
      "step": 55830
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7508223056793213,
      "learning_rate": 0.0005171887168628989,
      "loss": 2.2271,
      "step": 55831
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.6991484761238098,
      "learning_rate": 0.0005171858950115749,
      "loss": 2.1607,
      "step": 55832
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7824166417121887,
      "learning_rate": 0.000517183073119872,
      "loss": 2.2467,
      "step": 55833
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.8753973841667175,
      "learning_rate": 0.0005171802511877906,
      "loss": 2.3669,
      "step": 55834
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7368787527084351,
      "learning_rate": 0.0005171774292153313,
      "loss": 2.1266,
      "step": 55835
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7696147561073303,
      "learning_rate": 0.0005171746072024946,
      "loss": 2.1338,
      "step": 55836
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.8283247947692871,
      "learning_rate": 0.0005171717851492811,
      "loss": 2.1423,
      "step": 55837
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7720528244972229,
      "learning_rate": 0.0005171689630556911,
      "loss": 2.1685,
      "step": 55838
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7230263948440552,
      "learning_rate": 0.0005171661409217255,
      "loss": 1.9831,
      "step": 55839
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.9727035164833069,
      "learning_rate": 0.0005171633187473846,
      "loss": 2.3445,
      "step": 55840
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.8215016722679138,
      "learning_rate": 0.0005171604965326687,
      "loss": 2.2124,
      "step": 55841
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.6734904646873474,
      "learning_rate": 0.0005171576742775787,
      "loss": 2.2635,
      "step": 55842
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7602818012237549,
      "learning_rate": 0.0005171548519821151,
      "loss": 2.5005,
      "step": 55843
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7826076149940491,
      "learning_rate": 0.0005171520296462782,
      "loss": 2.1464,
      "step": 55844
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7727939486503601,
      "learning_rate": 0.0005171492072700687,
      "loss": 2.3356,
      "step": 55845
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7065264582633972,
      "learning_rate": 0.000517146384853487,
      "loss": 2.1721,
      "step": 55846
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.8265382051467896,
      "learning_rate": 0.0005171435623965338,
      "loss": 2.3309,
      "step": 55847
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7106539607048035,
      "learning_rate": 0.0005171407398992094,
      "loss": 2.3412,
      "step": 55848
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7450469732284546,
      "learning_rate": 0.0005171379173615145,
      "loss": 2.273,
      "step": 55849
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.6820920705795288,
      "learning_rate": 0.0005171350947834495,
      "loss": 2.1161,
      "step": 55850
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7026157379150391,
      "learning_rate": 0.000517132272165015,
      "loss": 2.2036,
      "step": 55851
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7488951683044434,
      "learning_rate": 0.0005171294495062115,
      "loss": 2.2368,
      "step": 55852
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.8242388963699341,
      "learning_rate": 0.0005171266268070397,
      "loss": 2.0907,
      "step": 55853
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7136266231536865,
      "learning_rate": 0.0005171238040674997,
      "loss": 2.3152,
      "step": 55854
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.8170974254608154,
      "learning_rate": 0.0005171209812875925,
      "loss": 2.1696,
      "step": 55855
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.9541250467300415,
      "learning_rate": 0.0005171181584673184,
      "loss": 2.3932,
      "step": 55856
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7502099275588989,
      "learning_rate": 0.0005171153356066778,
      "loss": 2.1016,
      "step": 55857
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.8771690130233765,
      "learning_rate": 0.0005171125127056714,
      "loss": 2.2988,
      "step": 55858
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7291749119758606,
      "learning_rate": 0.0005171096897642997,
      "loss": 2.1029,
      "step": 55859
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7444608807563782,
      "learning_rate": 0.0005171068667825633,
      "loss": 2.4242,
      "step": 55860
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7202362418174744,
      "learning_rate": 0.0005171040437604625,
      "loss": 2.4838,
      "step": 55861
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.8150175213813782,
      "learning_rate": 0.000517101220697998,
      "loss": 2.3749,
      "step": 55862
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.8138047456741333,
      "learning_rate": 0.0005170983975951703,
      "loss": 2.1277,
      "step": 55863
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7683231830596924,
      "learning_rate": 0.00051709557445198,
      "loss": 2.0755,
      "step": 55864
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.9157580137252808,
      "learning_rate": 0.0005170927512684274,
      "loss": 2.2816,
      "step": 55865
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.8951493501663208,
      "learning_rate": 0.0005170899280445132,
      "loss": 2.1521,
      "step": 55866
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7007973790168762,
      "learning_rate": 0.0005170871047802379,
      "loss": 2.4864,
      "step": 55867
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7509549856185913,
      "learning_rate": 0.000517084281475602,
      "loss": 2.4284,
      "step": 55868
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.6837238073348999,
      "learning_rate": 0.000517081458130606,
      "loss": 2.1203,
      "step": 55869
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7182214856147766,
      "learning_rate": 0.0005170786347452505,
      "loss": 2.1536,
      "step": 55870
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7533532381057739,
      "learning_rate": 0.000517075811319536,
      "loss": 2.3645,
      "step": 55871
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.6337265968322754,
      "learning_rate": 0.000517072987853463,
      "loss": 2.3848,
      "step": 55872
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.8029245138168335,
      "learning_rate": 0.000517070164347032,
      "loss": 2.2462,
      "step": 55873
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7132333517074585,
      "learning_rate": 0.0005170673408002436,
      "loss": 2.2582,
      "step": 55874
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.6727209091186523,
      "learning_rate": 0.0005170645172130983,
      "loss": 2.2577,
      "step": 55875
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7534787058830261,
      "learning_rate": 0.0005170616935855965,
      "loss": 2.1556,
      "step": 55876
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7304762005805969,
      "learning_rate": 0.0005170588699177389,
      "loss": 2.4196,
      "step": 55877
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7471519708633423,
      "learning_rate": 0.0005170560462095259,
      "loss": 2.3295,
      "step": 55878
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.6538556814193726,
      "learning_rate": 0.0005170532224609581,
      "loss": 2.0991,
      "step": 55879
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7160401940345764,
      "learning_rate": 0.0005170503986720362,
      "loss": 1.9991,
      "step": 55880
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.8635545969009399,
      "learning_rate": 0.0005170475748427604,
      "loss": 2.2998,
      "step": 55881
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7860987186431885,
      "learning_rate": 0.0005170447509731314,
      "loss": 2.2689,
      "step": 55882
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7860624194145203,
      "learning_rate": 0.0005170419270631495,
      "loss": 2.4192,
      "step": 55883
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.6593971848487854,
      "learning_rate": 0.0005170391031128157,
      "loss": 2.2319,
      "step": 55884
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7176237106323242,
      "learning_rate": 0.0005170362791221299,
      "loss": 2.1067,
      "step": 55885
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7792068719863892,
      "learning_rate": 0.0005170334550910932,
      "loss": 2.2837,
      "step": 55886
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7570390701293945,
      "learning_rate": 0.0005170306310197058,
      "loss": 2.0991,
      "step": 55887
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7058287262916565,
      "learning_rate": 0.0005170278069079684,
      "loss": 2.0309,
      "step": 55888
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7342108488082886,
      "learning_rate": 0.0005170249827558814,
      "loss": 2.3513,
      "step": 55889
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7820255756378174,
      "learning_rate": 0.0005170221585634453,
      "loss": 2.1469,
      "step": 55890
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.674529492855072,
      "learning_rate": 0.0005170193343306607,
      "loss": 2.0622,
      "step": 55891
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.9454464912414551,
      "learning_rate": 0.0005170165100575281,
      "loss": 2.2406,
      "step": 55892
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.6700136661529541,
      "learning_rate": 0.0005170136857440481,
      "loss": 2.2451,
      "step": 55893
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.76043301820755,
      "learning_rate": 0.000517010861390221,
      "loss": 2.342,
      "step": 55894
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.719131588935852,
      "learning_rate": 0.0005170080369960476,
      "loss": 2.3115,
      "step": 55895
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7141609191894531,
      "learning_rate": 0.0005170052125615283,
      "loss": 2.3865,
      "step": 55896
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.6833979487419128,
      "learning_rate": 0.0005170023880866636,
      "loss": 2.3304,
      "step": 55897
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7386751770973206,
      "learning_rate": 0.0005169995635714542,
      "loss": 2.2527,
      "step": 55898
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7322105765342712,
      "learning_rate": 0.0005169967390159004,
      "loss": 2.2652,
      "step": 55899
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7750241756439209,
      "learning_rate": 0.0005169939144200028,
      "loss": 2.1484,
      "step": 55900
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.8044643402099609,
      "learning_rate": 0.0005169910897837619,
      "loss": 2.4474,
      "step": 55901
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.691657543182373,
      "learning_rate": 0.0005169882651071783,
      "loss": 2.1003,
      "step": 55902
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.8052617907524109,
      "learning_rate": 0.0005169854403902524,
      "loss": 2.2438,
      "step": 55903
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.8517671823501587,
      "learning_rate": 0.0005169826156329849,
      "loss": 2.4593,
      "step": 55904
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.727247953414917,
      "learning_rate": 0.0005169797908353763,
      "loss": 2.4724,
      "step": 55905
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7070173621177673,
      "learning_rate": 0.000516976965997427,
      "loss": 2.1522,
      "step": 55906
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7942100763320923,
      "learning_rate": 0.0005169741411191375,
      "loss": 2.409,
      "step": 55907
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7892495393753052,
      "learning_rate": 0.0005169713162005085,
      "loss": 2.2867,
      "step": 55908
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7150850892066956,
      "learning_rate": 0.0005169684912415404,
      "loss": 2.2356,
      "step": 55909
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.8006916642189026,
      "learning_rate": 0.0005169656662422338,
      "loss": 2.2029,
      "step": 55910
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.841031014919281,
      "learning_rate": 0.0005169628412025891,
      "loss": 2.1793,
      "step": 55911
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7358137369155884,
      "learning_rate": 0.0005169600161226071,
      "loss": 2.244,
      "step": 55912
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7784011960029602,
      "learning_rate": 0.000516957191002288,
      "loss": 2.2613,
      "step": 55913
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.9517529606819153,
      "learning_rate": 0.0005169543658416324,
      "loss": 2.1671,
      "step": 55914
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7408159375190735,
      "learning_rate": 0.000516951540640641,
      "loss": 2.2074,
      "step": 55915
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7217338681221008,
      "learning_rate": 0.0005169487153993142,
      "loss": 2.384,
      "step": 55916
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.8296310901641846,
      "learning_rate": 0.0005169458901176524,
      "loss": 2.0751,
      "step": 55917
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.6808353066444397,
      "learning_rate": 0.0005169430647956564,
      "loss": 2.2556,
      "step": 55918
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.8044347763061523,
      "learning_rate": 0.0005169402394333266,
      "loss": 2.1353,
      "step": 55919
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7276826500892639,
      "learning_rate": 0.0005169374140306635,
      "loss": 2.2976,
      "step": 55920
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.8745144605636597,
      "learning_rate": 0.0005169345885876676,
      "loss": 2.1539,
      "step": 55921
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.8159646391868591,
      "learning_rate": 0.0005169317631043395,
      "loss": 2.2691,
      "step": 55922
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.6707168817520142,
      "learning_rate": 0.0005169289375806797,
      "loss": 1.956,
      "step": 55923
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.9841313362121582,
      "learning_rate": 0.0005169261120166887,
      "loss": 2.2509,
      "step": 55924
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.874349057674408,
      "learning_rate": 0.000516923286412367,
      "loss": 2.3315,
      "step": 55925
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7991213202476501,
      "learning_rate": 0.0005169204607677152,
      "loss": 2.1388,
      "step": 55926
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.8594274520874023,
      "learning_rate": 0.0005169176350827338,
      "loss": 2.3308,
      "step": 55927
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.9605277180671692,
      "learning_rate": 0.0005169148093574234,
      "loss": 2.322,
      "step": 55928
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.8331260681152344,
      "learning_rate": 0.0005169119835917843,
      "loss": 2.1715,
      "step": 55929
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7087947130203247,
      "learning_rate": 0.0005169091577858171,
      "loss": 2.4272,
      "step": 55930
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.8486288189888,
      "learning_rate": 0.0005169063319395227,
      "loss": 2.3611,
      "step": 55931
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.777643620967865,
      "learning_rate": 0.0005169035060529011,
      "loss": 2.2391,
      "step": 55932
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.9591707587242126,
      "learning_rate": 0.0005169006801259531,
      "loss": 2.5281,
      "step": 55933
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.633053719997406,
      "learning_rate": 0.000516897854158679,
      "loss": 2.0207,
      "step": 55934
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.8536723852157593,
      "learning_rate": 0.0005168950281510797,
      "loss": 2.0665,
      "step": 55935
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7855989336967468,
      "learning_rate": 0.0005168922021031555,
      "loss": 2.4319,
      "step": 55936
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.8928079605102539,
      "learning_rate": 0.0005168893760149069,
      "loss": 2.145,
      "step": 55937
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7642560005187988,
      "learning_rate": 0.0005168865498863344,
      "loss": 2.1849,
      "step": 55938
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7323353886604309,
      "learning_rate": 0.0005168837237174387,
      "loss": 2.116,
      "step": 55939
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7835333943367004,
      "learning_rate": 0.0005168808975082201,
      "loss": 2.3508,
      "step": 55940
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7194568514823914,
      "learning_rate": 0.0005168780712586794,
      "loss": 2.3018,
      "step": 55941
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7712763547897339,
      "learning_rate": 0.000516875244968817,
      "loss": 2.098,
      "step": 55942
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7569112181663513,
      "learning_rate": 0.0005168724186386333,
      "loss": 2.2854,
      "step": 55943
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.9170687198638916,
      "learning_rate": 0.000516869592268129,
      "loss": 2.18,
      "step": 55944
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7447990775108337,
      "learning_rate": 0.0005168667658573045,
      "loss": 2.0194,
      "step": 55945
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7433081865310669,
      "learning_rate": 0.0005168639394061603,
      "loss": 2.0642,
      "step": 55946
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7280236482620239,
      "learning_rate": 0.0005168611129146971,
      "loss": 2.3188,
      "step": 55947
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7843690514564514,
      "learning_rate": 0.0005168582863829152,
      "loss": 2.2298,
      "step": 55948
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.747969925403595,
      "learning_rate": 0.0005168554598108154,
      "loss": 2.4856,
      "step": 55949
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7802742123603821,
      "learning_rate": 0.000516852633198398,
      "loss": 2.0251,
      "step": 55950
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7680453658103943,
      "learning_rate": 0.0005168498065456636,
      "loss": 2.1192,
      "step": 55951
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7682265043258667,
      "learning_rate": 0.0005168469798526127,
      "loss": 2.2961,
      "step": 55952
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.768824577331543,
      "learning_rate": 0.0005168441531192459,
      "loss": 2.5273,
      "step": 55953
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.703090488910675,
      "learning_rate": 0.0005168413263455637,
      "loss": 2.2941,
      "step": 55954
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7407529950141907,
      "learning_rate": 0.0005168384995315665,
      "loss": 2.2305,
      "step": 55955
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7394081950187683,
      "learning_rate": 0.0005168356726772551,
      "loss": 2.4422,
      "step": 55956
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7617404460906982,
      "learning_rate": 0.0005168328457826298,
      "loss": 2.2639,
      "step": 55957
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7414925694465637,
      "learning_rate": 0.0005168300188476911,
      "loss": 2.2909,
      "step": 55958
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7479530572891235,
      "learning_rate": 0.0005168271918724397,
      "loss": 2.1732,
      "step": 55959
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7181108593940735,
      "learning_rate": 0.000516824364856876,
      "loss": 2.4828,
      "step": 55960
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7006152868270874,
      "learning_rate": 0.0005168215378010004,
      "loss": 2.1615,
      "step": 55961
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7204808592796326,
      "learning_rate": 0.0005168187107048138,
      "loss": 2.2169,
      "step": 55962
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7183957099914551,
      "learning_rate": 0.0005168158835683165,
      "loss": 2.0861,
      "step": 55963
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.6946883201599121,
      "learning_rate": 0.000516813056391509,
      "loss": 2.0978,
      "step": 55964
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7965971231460571,
      "learning_rate": 0.0005168102291743917,
      "loss": 2.2225,
      "step": 55965
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.714163601398468,
      "learning_rate": 0.0005168074019169655,
      "loss": 2.2911,
      "step": 55966
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7227502465248108,
      "learning_rate": 0.0005168045746192305,
      "loss": 2.2972,
      "step": 55967
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7228808999061584,
      "learning_rate": 0.0005168017472811877,
      "loss": 2.0635,
      "step": 55968
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.717432975769043,
      "learning_rate": 0.0005167989199028371,
      "loss": 2.2337,
      "step": 55969
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.8122594356536865,
      "learning_rate": 0.0005167960924841796,
      "loss": 2.2632,
      "step": 55970
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7000107765197754,
      "learning_rate": 0.0005167932650252157,
      "loss": 2.2856,
      "step": 55971
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.8573316335678101,
      "learning_rate": 0.0005167904375259457,
      "loss": 2.3178,
      "step": 55972
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7580409049987793,
      "learning_rate": 0.0005167876099863704,
      "loss": 2.2436,
      "step": 55973
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7476994395256042,
      "learning_rate": 0.0005167847824064901,
      "loss": 2.4744,
      "step": 55974
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.6826526522636414,
      "learning_rate": 0.0005167819547863055,
      "loss": 2.1355,
      "step": 55975
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7563579678535461,
      "learning_rate": 0.000516779127125817,
      "loss": 2.3105,
      "step": 55976
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.8237115740776062,
      "learning_rate": 0.0005167762994250252,
      "loss": 2.0699,
      "step": 55977
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.747151792049408,
      "learning_rate": 0.0005167734716839305,
      "loss": 2.1242,
      "step": 55978
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.6999931931495667,
      "learning_rate": 0.0005167706439025336,
      "loss": 2.3922,
      "step": 55979
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7176636457443237,
      "learning_rate": 0.0005167678160808349,
      "loss": 2.1402,
      "step": 55980
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.6769154667854309,
      "learning_rate": 0.000516764988218835,
      "loss": 2.3197,
      "step": 55981
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.8695760369300842,
      "learning_rate": 0.0005167621603165343,
      "loss": 2.2244,
      "step": 55982
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7283278107643127,
      "learning_rate": 0.0005167593323739335,
      "loss": 2.0845,
      "step": 55983
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7265849113464355,
      "learning_rate": 0.000516756504391033,
      "loss": 2.3447,
      "step": 55984
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7472515106201172,
      "learning_rate": 0.0005167536763678335,
      "loss": 2.284,
      "step": 55985
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7398166656494141,
      "learning_rate": 0.0005167508483043353,
      "loss": 2.2795,
      "step": 55986
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7538979649543762,
      "learning_rate": 0.0005167480202005391,
      "loss": 2.2012,
      "step": 55987
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.828657865524292,
      "learning_rate": 0.0005167451920564453,
      "loss": 2.1765,
      "step": 55988
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7466695308685303,
      "learning_rate": 0.0005167423638720544,
      "loss": 2.0848,
      "step": 55989
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.6728412508964539,
      "learning_rate": 0.0005167395356473672,
      "loss": 2.1239,
      "step": 55990
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7163621783256531,
      "learning_rate": 0.0005167367073823839,
      "loss": 2.1259,
      "step": 55991
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7671423554420471,
      "learning_rate": 0.0005167338790771052,
      "loss": 2.3217,
      "step": 55992
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7576530575752258,
      "learning_rate": 0.0005167310507315315,
      "loss": 2.1298,
      "step": 55993
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7496033906936646,
      "learning_rate": 0.0005167282223456635,
      "loss": 2.2774,
      "step": 55994
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7349827289581299,
      "learning_rate": 0.0005167253939195015,
      "loss": 2.099,
      "step": 55995
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.9086723327636719,
      "learning_rate": 0.0005167225654530462,
      "loss": 2.2016,
      "step": 55996
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7315061688423157,
      "learning_rate": 0.0005167197369462982,
      "loss": 2.2632,
      "step": 55997
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7133125066757202,
      "learning_rate": 0.0005167169083992579,
      "loss": 2.1717,
      "step": 55998
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.8036257028579712,
      "learning_rate": 0.0005167140798119257,
      "loss": 2.1485,
      "step": 55999
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.6737613081932068,
      "learning_rate": 0.0005167112511843023,
      "loss": 2.1463,
      "step": 56000
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.6689741611480713,
      "learning_rate": 0.0005167084225163883,
      "loss": 2.0976,
      "step": 56001
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.6866194009780884,
      "learning_rate": 0.000516705593808184,
      "loss": 2.1549,
      "step": 56002
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7056878209114075,
      "learning_rate": 0.00051670276505969,
      "loss": 2.3589,
      "step": 56003
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7693122029304504,
      "learning_rate": 0.000516699936270907,
      "loss": 2.1044,
      "step": 56004
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.669247567653656,
      "learning_rate": 0.0005166971074418354,
      "loss": 2.3279,
      "step": 56005
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.823350727558136,
      "learning_rate": 0.0005166942785724756,
      "loss": 2.4201,
      "step": 56006
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7495254874229431,
      "learning_rate": 0.0005166914496628284,
      "loss": 2.1172,
      "step": 56007
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.9353380799293518,
      "learning_rate": 0.0005166886207128939,
      "loss": 2.0316,
      "step": 56008
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7268973588943481,
      "learning_rate": 0.0005166857917226731,
      "loss": 2.1614,
      "step": 56009
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.8466342687606812,
      "learning_rate": 0.0005166829626921663,
      "loss": 2.3807,
      "step": 56010
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7885466814041138,
      "learning_rate": 0.000516680133621374,
      "loss": 2.2841,
      "step": 56011
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.6724600791931152,
      "learning_rate": 0.0005166773045102968,
      "loss": 2.2202,
      "step": 56012
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7292288541793823,
      "learning_rate": 0.0005166744753589352,
      "loss": 2.0609,
      "step": 56013
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.736041784286499,
      "learning_rate": 0.0005166716461672897,
      "loss": 2.2981,
      "step": 56014
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7193325161933899,
      "learning_rate": 0.0005166688169353609,
      "loss": 2.3605,
      "step": 56015
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.704557478427887,
      "learning_rate": 0.0005166659876631492,
      "loss": 2.3176,
      "step": 56016
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7725350856781006,
      "learning_rate": 0.0005166631583506554,
      "loss": 2.1027,
      "step": 56017
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7336488366127014,
      "learning_rate": 0.0005166603289978796,
      "loss": 2.2543,
      "step": 56018
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7049180865287781,
      "learning_rate": 0.0005166574996048226,
      "loss": 2.0915,
      "step": 56019
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.6638621687889099,
      "learning_rate": 0.0005166546701714849,
      "loss": 1.9568,
      "step": 56020
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7088142037391663,
      "learning_rate": 0.000516651840697867,
      "loss": 2.3218,
      "step": 56021
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7599406242370605,
      "learning_rate": 0.0005166490111839694,
      "loss": 2.3756,
      "step": 56022
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.8188700675964355,
      "learning_rate": 0.0005166461816297929,
      "loss": 2.4277,
      "step": 56023
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7063630819320679,
      "learning_rate": 0.0005166433520353376,
      "loss": 2.4574,
      "step": 56024
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.8510469198226929,
      "learning_rate": 0.0005166405224006042,
      "loss": 2.3112,
      "step": 56025
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7292670607566833,
      "learning_rate": 0.0005166376927255932,
      "loss": 2.1858,
      "step": 56026
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7316609621047974,
      "learning_rate": 0.0005166348630103052,
      "loss": 2.2236,
      "step": 56027
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7735403180122375,
      "learning_rate": 0.0005166320332547407,
      "loss": 2.551,
      "step": 56028
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7001203298568726,
      "learning_rate": 0.0005166292034589002,
      "loss": 2.3777,
      "step": 56029
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7277909517288208,
      "learning_rate": 0.0005166263736227844,
      "loss": 2.299,
      "step": 56030
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7899102568626404,
      "learning_rate": 0.0005166235437463935,
      "loss": 2.2864,
      "step": 56031
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7111578583717346,
      "learning_rate": 0.0005166207138297282,
      "loss": 2.2226,
      "step": 56032
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7102737426757812,
      "learning_rate": 0.000516617883872789,
      "loss": 2.378,
      "step": 56033
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7000994086265564,
      "learning_rate": 0.0005166150538755765,
      "loss": 2.0512,
      "step": 56034
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7022071480751038,
      "learning_rate": 0.0005166122238380912,
      "loss": 2.3081,
      "step": 56035
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7255638837814331,
      "learning_rate": 0.0005166093937603336,
      "loss": 2.1913,
      "step": 56036
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7061509490013123,
      "learning_rate": 0.0005166065636423041,
      "loss": 2.4286,
      "step": 56037
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7184524536132812,
      "learning_rate": 0.0005166037334840035,
      "loss": 2.2933,
      "step": 56038
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.6787780523300171,
      "learning_rate": 0.0005166009032854321,
      "loss": 2.2092,
      "step": 56039
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7364927530288696,
      "learning_rate": 0.0005165980730465905,
      "loss": 2.1711,
      "step": 56040
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.676683247089386,
      "learning_rate": 0.0005165952427674793,
      "loss": 2.3505,
      "step": 56041
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.6733198761940002,
      "learning_rate": 0.0005165924124480989,
      "loss": 2.4013,
      "step": 56042
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7245945930480957,
      "learning_rate": 0.00051658958208845,
      "loss": 2.4022,
      "step": 56043
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7103171348571777,
      "learning_rate": 0.0005165867516885328,
      "loss": 2.2318,
      "step": 56044
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7213318347930908,
      "learning_rate": 0.0005165839212483482,
      "loss": 2.3243,
      "step": 56045
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7228844165802002,
      "learning_rate": 0.0005165810907678965,
      "loss": 2.1437,
      "step": 56046
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.766984224319458,
      "learning_rate": 0.0005165782602471783,
      "loss": 2.155,
      "step": 56047
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.6606363654136658,
      "learning_rate": 0.000516575429686194,
      "loss": 2.1899,
      "step": 56048
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7226064205169678,
      "learning_rate": 0.0005165725990849444,
      "loss": 2.3165,
      "step": 56049
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7582960724830627,
      "learning_rate": 0.0005165697684434297,
      "loss": 2.4315,
      "step": 56050
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7104665637016296,
      "learning_rate": 0.0005165669377616508,
      "loss": 2.2801,
      "step": 56051
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.8100976943969727,
      "learning_rate": 0.0005165641070396078,
      "loss": 2.3518,
      "step": 56052
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.6469844579696655,
      "learning_rate": 0.0005165612762773015,
      "loss": 2.2565,
      "step": 56053
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.747559130191803,
      "learning_rate": 0.0005165584454747325,
      "loss": 2.3158,
      "step": 56054
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.6977425813674927,
      "learning_rate": 0.0005165556146319011,
      "loss": 2.3259,
      "step": 56055
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7718155980110168,
      "learning_rate": 0.0005165527837488079,
      "loss": 2.3849,
      "step": 56056
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.728819727897644,
      "learning_rate": 0.0005165499528254535,
      "loss": 2.3494,
      "step": 56057
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7006382346153259,
      "learning_rate": 0.0005165471218618385,
      "loss": 2.3082,
      "step": 56058
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.6935305595397949,
      "learning_rate": 0.0005165442908579631,
      "loss": 2.3354,
      "step": 56059
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.8780423402786255,
      "learning_rate": 0.000516541459813828,
      "loss": 2.2552,
      "step": 56060
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.8144106268882751,
      "learning_rate": 0.000516538628729434,
      "loss": 2.2209,
      "step": 56061
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.6488326787948608,
      "learning_rate": 0.0005165357976047812,
      "loss": 2.2733,
      "step": 56062
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7286982536315918,
      "learning_rate": 0.0005165329664398703,
      "loss": 2.4772,
      "step": 56063
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.9185144901275635,
      "learning_rate": 0.0005165301352347019,
      "loss": 2.1811,
      "step": 56064
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7870407700538635,
      "learning_rate": 0.0005165273039892764,
      "loss": 2.3804,
      "step": 56065
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7104479074478149,
      "learning_rate": 0.0005165244727035945,
      "loss": 2.1544,
      "step": 56066
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.8708077073097229,
      "learning_rate": 0.0005165216413776565,
      "loss": 2.3478,
      "step": 56067
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7686930894851685,
      "learning_rate": 0.0005165188100114631,
      "loss": 2.2188,
      "step": 56068
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.8558363318443298,
      "learning_rate": 0.0005165159786050147,
      "loss": 2.587,
      "step": 56069
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.6943694353103638,
      "learning_rate": 0.000516513147158312,
      "loss": 2.3296,
      "step": 56070
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.8084048628807068,
      "learning_rate": 0.0005165103156713553,
      "loss": 2.3172,
      "step": 56071
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.9525957703590393,
      "learning_rate": 0.0005165074841441453,
      "loss": 2.3022,
      "step": 56072
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7757288813591003,
      "learning_rate": 0.0005165046525766826,
      "loss": 2.3908,
      "step": 56073
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.741692066192627,
      "learning_rate": 0.0005165018209689673,
      "loss": 2.1407,
      "step": 56074
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.748694658279419,
      "learning_rate": 0.0005164989893210004,
      "loss": 2.0687,
      "step": 56075
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7895118594169617,
      "learning_rate": 0.0005164961576327823,
      "loss": 2.3255,
      "step": 56076
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.8652681112289429,
      "learning_rate": 0.0005164933259043134,
      "loss": 2.395,
      "step": 56077
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7729722261428833,
      "learning_rate": 0.0005164904941355943,
      "loss": 2.3248,
      "step": 56078
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7937924861907959,
      "learning_rate": 0.0005164876623266257,
      "loss": 2.1465,
      "step": 56079
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.9528750777244568,
      "learning_rate": 0.0005164848304774078,
      "loss": 2.2766,
      "step": 56080
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7877777218818665,
      "learning_rate": 0.0005164819985879412,
      "loss": 2.1575,
      "step": 56081
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.8030683398246765,
      "learning_rate": 0.0005164791666582266,
      "loss": 2.3541,
      "step": 56082
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.011191964149475,
      "learning_rate": 0.0005164763346882645,
      "loss": 2.3671,
      "step": 56083
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.8144419193267822,
      "learning_rate": 0.0005164735026780554,
      "loss": 2.1048,
      "step": 56084
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7461673021316528,
      "learning_rate": 0.0005164706706275996,
      "loss": 2.1933,
      "step": 56085
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.9191212058067322,
      "learning_rate": 0.000516467838536898,
      "loss": 2.2677,
      "step": 56086
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.8472144603729248,
      "learning_rate": 0.0005164650064059509,
      "loss": 2.1458,
      "step": 56087
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7414368987083435,
      "learning_rate": 0.0005164621742347588,
      "loss": 2.2055,
      "step": 56088
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7717123627662659,
      "learning_rate": 0.0005164593420233222,
      "loss": 2.2632,
      "step": 56089
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.021442174911499,
      "learning_rate": 0.0005164565097716419,
      "loss": 1.9757,
      "step": 56090
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.8344970345497131,
      "learning_rate": 0.0005164536774797182,
      "loss": 2.1919,
      "step": 56091
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.9202265739440918,
      "learning_rate": 0.0005164508451475517,
      "loss": 2.1928,
      "step": 56092
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.9517687559127808,
      "learning_rate": 0.0005164480127751428,
      "loss": 2.3515,
      "step": 56093
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7651447057723999,
      "learning_rate": 0.0005164451803624922,
      "loss": 2.3621,
      "step": 56094
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7831568717956543,
      "learning_rate": 0.0005164423479096004,
      "loss": 2.2053,
      "step": 56095
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.720750093460083,
      "learning_rate": 0.0005164395154164679,
      "loss": 2.2427,
      "step": 56096
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7665800452232361,
      "learning_rate": 0.0005164366828830951,
      "loss": 2.4541,
      "step": 56097
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7350435853004456,
      "learning_rate": 0.0005164338503094826,
      "loss": 2.2114,
      "step": 56098
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7339499592781067,
      "learning_rate": 0.0005164310176956311,
      "loss": 2.3089,
      "step": 56099
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7533501386642456,
      "learning_rate": 0.0005164281850415409,
      "loss": 2.1954,
      "step": 56100
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7138163447380066,
      "learning_rate": 0.0005164253523472126,
      "loss": 2.363,
      "step": 56101
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7174683213233948,
      "learning_rate": 0.0005164225196126469,
      "loss": 2.2311,
      "step": 56102
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.9277250170707703,
      "learning_rate": 0.000516419686837844,
      "loss": 2.382,
      "step": 56103
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.764320433139801,
      "learning_rate": 0.0005164168540228046,
      "loss": 2.2945,
      "step": 56104
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.6685890555381775,
      "learning_rate": 0.0005164140211675294,
      "loss": 2.3773,
      "step": 56105
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7299677133560181,
      "learning_rate": 0.0005164111882720186,
      "loss": 2.3453,
      "step": 56106
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7547982931137085,
      "learning_rate": 0.0005164083553362729,
      "loss": 2.3856,
      "step": 56107
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.809033989906311,
      "learning_rate": 0.0005164055223602928,
      "loss": 2.2515,
      "step": 56108
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7518426179885864,
      "learning_rate": 0.0005164026893440788,
      "loss": 2.2085,
      "step": 56109
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.8256763815879822,
      "learning_rate": 0.0005163998562876315,
      "loss": 2.2418,
      "step": 56110
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7774752378463745,
      "learning_rate": 0.0005163970231909514,
      "loss": 2.34,
      "step": 56111
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.77230304479599,
      "learning_rate": 0.000516394190054039,
      "loss": 2.5281,
      "step": 56112
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.659408450126648,
      "learning_rate": 0.0005163913568768947,
      "loss": 2.3587,
      "step": 56113
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.6889581680297852,
      "learning_rate": 0.0005163885236595193,
      "loss": 2.1369,
      "step": 56114
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7216750383377075,
      "learning_rate": 0.0005163856904019132,
      "loss": 2.3457,
      "step": 56115
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7552973031997681,
      "learning_rate": 0.0005163828571040769,
      "loss": 2.4154,
      "step": 56116
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.718521773815155,
      "learning_rate": 0.0005163800237660108,
      "loss": 2.0917,
      "step": 56117
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7676582336425781,
      "learning_rate": 0.0005163771903877157,
      "loss": 2.2103,
      "step": 56118
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7734397649765015,
      "learning_rate": 0.000516374356969192,
      "loss": 2.4039,
      "step": 56119
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7377347350120544,
      "learning_rate": 0.0005163715235104402,
      "loss": 2.2256,
      "step": 56120
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.6522183418273926,
      "learning_rate": 0.000516368690011461,
      "loss": 2.3141,
      "step": 56121
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7262499332427979,
      "learning_rate": 0.0005163658564722545,
      "loss": 2.0883,
      "step": 56122
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7765480279922485,
      "learning_rate": 0.0005163630228928216,
      "loss": 2.3189,
      "step": 56123
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7295452356338501,
      "learning_rate": 0.0005163601892731629,
      "loss": 2.1652,
      "step": 56124
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7794108390808105,
      "learning_rate": 0.0005163573556132785,
      "loss": 2.3403,
      "step": 56125
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.6348231434822083,
      "learning_rate": 0.0005163545219131694,
      "loss": 2.0781,
      "step": 56126
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.6854232549667358,
      "learning_rate": 0.0005163516881728357,
      "loss": 2.3435,
      "step": 56127
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7266963124275208,
      "learning_rate": 0.0005163488543922783,
      "loss": 2.2021,
      "step": 56128
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7169600129127502,
      "learning_rate": 0.0005163460205714976,
      "loss": 2.1692,
      "step": 56129
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.691321074962616,
      "learning_rate": 0.0005163431867104939,
      "loss": 2.4922,
      "step": 56130
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7377481460571289,
      "learning_rate": 0.000516340352809268,
      "loss": 2.1449,
      "step": 56131
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.6913119554519653,
      "learning_rate": 0.0005163375188678204,
      "loss": 2.2262,
      "step": 56132
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.8778943419456482,
      "learning_rate": 0.0005163346848861516,
      "loss": 2.1085,
      "step": 56133
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7183356285095215,
      "learning_rate": 0.000516331850864262,
      "loss": 2.2552,
      "step": 56134
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7338972091674805,
      "learning_rate": 0.0005163290168021523,
      "loss": 2.3083,
      "step": 56135
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7078492641448975,
      "learning_rate": 0.0005163261826998229,
      "loss": 2.3729,
      "step": 56136
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.6924417018890381,
      "learning_rate": 0.0005163233485572744,
      "loss": 2.2341,
      "step": 56137
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7610892057418823,
      "learning_rate": 0.0005163205143745072,
      "loss": 2.3866,
      "step": 56138
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7153083682060242,
      "learning_rate": 0.0005163176801515221,
      "loss": 2.0413,
      "step": 56139
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7067834138870239,
      "learning_rate": 0.0005163148458883194,
      "loss": 2.2961,
      "step": 56140
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7035712599754333,
      "learning_rate": 0.0005163120115848997,
      "loss": 2.1337,
      "step": 56141
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7463911175727844,
      "learning_rate": 0.0005163091772412634,
      "loss": 2.4898,
      "step": 56142
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7131596207618713,
      "learning_rate": 0.0005163063428574113,
      "loss": 2.2985,
      "step": 56143
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.8704897165298462,
      "learning_rate": 0.0005163035084333437,
      "loss": 2.3815,
      "step": 56144
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.787533700466156,
      "learning_rate": 0.0005163006739690613,
      "loss": 2.2215,
      "step": 56145
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7224329710006714,
      "learning_rate": 0.0005162978394645644,
      "loss": 2.2545,
      "step": 56146
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7136040925979614,
      "learning_rate": 0.0005162950049198536,
      "loss": 2.0119,
      "step": 56147
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7441986203193665,
      "learning_rate": 0.0005162921703349295,
      "loss": 2.1446,
      "step": 56148
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7153060436248779,
      "learning_rate": 0.0005162893357097927,
      "loss": 2.3876,
      "step": 56149
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7642286419868469,
      "learning_rate": 0.0005162865010444436,
      "loss": 2.3654,
      "step": 56150
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.6655853986740112,
      "learning_rate": 0.0005162836663388827,
      "loss": 2.0619,
      "step": 56151
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7094190716743469,
      "learning_rate": 0.0005162808315931106,
      "loss": 2.3035,
      "step": 56152
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7689042091369629,
      "learning_rate": 0.0005162779968071278,
      "loss": 2.229,
      "step": 56153
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7143840789794922,
      "learning_rate": 0.0005162751619809349,
      "loss": 2.0914,
      "step": 56154
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7808451652526855,
      "learning_rate": 0.0005162723271145323,
      "loss": 2.2548,
      "step": 56155
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7402009963989258,
      "learning_rate": 0.0005162694922079207,
      "loss": 2.184,
      "step": 56156
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.6696956753730774,
      "learning_rate": 0.0005162666572611004,
      "loss": 2.097,
      "step": 56157
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.708541750907898,
      "learning_rate": 0.0005162638222740721,
      "loss": 2.1906,
      "step": 56158
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7073109745979309,
      "learning_rate": 0.0005162609872468363,
      "loss": 2.0755,
      "step": 56159
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7792864441871643,
      "learning_rate": 0.0005162581521793935,
      "loss": 2.3662,
      "step": 56160
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.6596792340278625,
      "learning_rate": 0.0005162553170717442,
      "loss": 2.0219,
      "step": 56161
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.6996081471443176,
      "learning_rate": 0.0005162524819238888,
      "loss": 2.1275,
      "step": 56162
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7338992357254028,
      "learning_rate": 0.0005162496467358282,
      "loss": 2.2724,
      "step": 56163
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.6971904635429382,
      "learning_rate": 0.0005162468115075627,
      "loss": 2.1978,
      "step": 56164
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7186470031738281,
      "learning_rate": 0.0005162439762390927,
      "loss": 2.183,
      "step": 56165
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7200242280960083,
      "learning_rate": 0.0005162411409304189,
      "loss": 2.2475,
      "step": 56166
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.693661630153656,
      "learning_rate": 0.0005162383055815419,
      "loss": 2.2271,
      "step": 56167
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7192345261573792,
      "learning_rate": 0.000516235470192462,
      "loss": 2.3632,
      "step": 56168
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7415523529052734,
      "learning_rate": 0.0005162326347631799,
      "loss": 2.2312,
      "step": 56169
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.6644232273101807,
      "learning_rate": 0.000516229799293696,
      "loss": 2.2367,
      "step": 56170
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7828753590583801,
      "learning_rate": 0.000516226963784011,
      "loss": 2.1573,
      "step": 56171
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7634472846984863,
      "learning_rate": 0.0005162241282341253,
      "loss": 2.3836,
      "step": 56172
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7200953960418701,
      "learning_rate": 0.0005162212926440394,
      "loss": 2.5895,
      "step": 56173
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7764202356338501,
      "learning_rate": 0.0005162184570137539,
      "loss": 2.1829,
      "step": 56174
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7378478646278381,
      "learning_rate": 0.0005162156213432694,
      "loss": 2.0797,
      "step": 56175
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7310878038406372,
      "learning_rate": 0.0005162127856325862,
      "loss": 2.4105,
      "step": 56176
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7260293364524841,
      "learning_rate": 0.000516209949881705,
      "loss": 2.1212,
      "step": 56177
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7120311856269836,
      "learning_rate": 0.0005162071140906264,
      "loss": 2.1543,
      "step": 56178
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7270975708961487,
      "learning_rate": 0.0005162042782593506,
      "loss": 2.1686,
      "step": 56179
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7911149859428406,
      "learning_rate": 0.0005162014423878785,
      "loss": 2.3153,
      "step": 56180
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.9806917905807495,
      "learning_rate": 0.0005161986064762104,
      "loss": 2.159,
      "step": 56181
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.8432357311248779,
      "learning_rate": 0.0005161957705243469,
      "loss": 2.359,
      "step": 56182
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.089451551437378,
      "learning_rate": 0.0005161929345322886,
      "loss": 2.3502,
      "step": 56183
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.8460371494293213,
      "learning_rate": 0.000516190098500036,
      "loss": 2.2657,
      "step": 56184
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.8343007564544678,
      "learning_rate": 0.0005161872624275894,
      "loss": 2.2004,
      "step": 56185
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7431554794311523,
      "learning_rate": 0.0005161844263149496,
      "loss": 2.3461,
      "step": 56186
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7213062644004822,
      "learning_rate": 0.0005161815901621171,
      "loss": 2.1677,
      "step": 56187
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7433888912200928,
      "learning_rate": 0.0005161787539690922,
      "loss": 2.1397,
      "step": 56188
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7233916521072388,
      "learning_rate": 0.0005161759177358757,
      "loss": 2.218,
      "step": 56189
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7530608773231506,
      "learning_rate": 0.000516173081462468,
      "loss": 2.0616,
      "step": 56190
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7140586972236633,
      "learning_rate": 0.0005161702451488697,
      "loss": 2.3673,
      "step": 56191
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7471347451210022,
      "learning_rate": 0.0005161674087950811,
      "loss": 2.2103,
      "step": 56192
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7600665092468262,
      "learning_rate": 0.000516164572401103,
      "loss": 2.1549,
      "step": 56193
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7710918188095093,
      "learning_rate": 0.000516161735966936,
      "loss": 2.2929,
      "step": 56194
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.6709163188934326,
      "learning_rate": 0.0005161588994925803,
      "loss": 2.2669,
      "step": 56195
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7819299697875977,
      "learning_rate": 0.0005161560629780365,
      "loss": 2.4322,
      "step": 56196
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7062172293663025,
      "learning_rate": 0.0005161532264233052,
      "loss": 2.1777,
      "step": 56197
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7007883787155151,
      "learning_rate": 0.0005161503898283871,
      "loss": 2.0035,
      "step": 56198
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7068034410476685,
      "learning_rate": 0.0005161475531932825,
      "loss": 2.2187,
      "step": 56199
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.767745316028595,
      "learning_rate": 0.0005161447165179919,
      "loss": 2.2667,
      "step": 56200
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7818726897239685,
      "learning_rate": 0.000516141879802516,
      "loss": 2.3423,
      "step": 56201
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.6811167001724243,
      "learning_rate": 0.0005161390430468552,
      "loss": 2.1865,
      "step": 56202
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7197813391685486,
      "learning_rate": 0.0005161362062510101,
      "loss": 2.0828,
      "step": 56203
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.6907418966293335,
      "learning_rate": 0.0005161333694149812,
      "loss": 2.0229,
      "step": 56204
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.752197802066803,
      "learning_rate": 0.000516130532538769,
      "loss": 2.3393,
      "step": 56205
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.732581377029419,
      "learning_rate": 0.0005161276956223742,
      "loss": 2.3185,
      "step": 56206
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.8004167675971985,
      "learning_rate": 0.0005161248586657971,
      "loss": 2.2437,
      "step": 56207
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.6879624724388123,
      "learning_rate": 0.0005161220216690381,
      "loss": 2.1622,
      "step": 56208
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7322536706924438,
      "learning_rate": 0.0005161191846320982,
      "loss": 2.005,
      "step": 56209
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7558805346488953,
      "learning_rate": 0.0005161163475549776,
      "loss": 2.2406,
      "step": 56210
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.685158371925354,
      "learning_rate": 0.0005161135104376768,
      "loss": 2.2684,
      "step": 56211
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.6884473562240601,
      "learning_rate": 0.0005161106732801964,
      "loss": 2.1432,
      "step": 56212
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7585619688034058,
      "learning_rate": 0.0005161078360825371,
      "loss": 2.1523,
      "step": 56213
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.6582887172698975,
      "learning_rate": 0.0005161049988446992,
      "loss": 2.1334,
      "step": 56214
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7646410465240479,
      "learning_rate": 0.0005161021615666834,
      "loss": 2.1916,
      "step": 56215
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.8065043091773987,
      "learning_rate": 0.0005160993242484899,
      "loss": 2.1548,
      "step": 56216
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7608950138092041,
      "learning_rate": 0.0005160964868901197,
      "loss": 2.3304,
      "step": 56217
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.86150062084198,
      "learning_rate": 0.000516093649491573,
      "loss": 2.3603,
      "step": 56218
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7200864553451538,
      "learning_rate": 0.0005160908120528504,
      "loss": 2.1119,
      "step": 56219
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.6797599196434021,
      "learning_rate": 0.0005160879745739524,
      "loss": 2.2185,
      "step": 56220
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.8946090936660767,
      "learning_rate": 0.0005160851370548796,
      "loss": 2.3422,
      "step": 56221
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.734114408493042,
      "learning_rate": 0.0005160822994956325,
      "loss": 2.3816,
      "step": 56222
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7063311338424683,
      "learning_rate": 0.0005160794618962115,
      "loss": 2.2594,
      "step": 56223
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.8021860718727112,
      "learning_rate": 0.0005160766242566174,
      "loss": 2.1447,
      "step": 56224
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7757103443145752,
      "learning_rate": 0.0005160737865768506,
      "loss": 1.9457,
      "step": 56225
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.768314003944397,
      "learning_rate": 0.0005160709488569114,
      "loss": 2.3314,
      "step": 56226
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.6925808191299438,
      "learning_rate": 0.0005160681110968008,
      "loss": 2.1822,
      "step": 56227
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.8181688785552979,
      "learning_rate": 0.0005160652732965189,
      "loss": 2.2902,
      "step": 56228
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7206706404685974,
      "learning_rate": 0.0005160624354560664,
      "loss": 2.2433,
      "step": 56229
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7256649732589722,
      "learning_rate": 0.0005160595975754438,
      "loss": 2.189,
      "step": 56230
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7125453352928162,
      "learning_rate": 0.0005160567596546517,
      "loss": 2.0985,
      "step": 56231
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7307979464530945,
      "learning_rate": 0.0005160539216936905,
      "loss": 2.2333,
      "step": 56232
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7668030858039856,
      "learning_rate": 0.0005160510836925608,
      "loss": 2.0619,
      "step": 56233
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.946066677570343,
      "learning_rate": 0.0005160482456512632,
      "loss": 2.0359,
      "step": 56234
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7301697731018066,
      "learning_rate": 0.0005160454075697981,
      "loss": 2.365,
      "step": 56235
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7588502764701843,
      "learning_rate": 0.0005160425694481661,
      "loss": 2.2096,
      "step": 56236
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.8037853837013245,
      "learning_rate": 0.0005160397312863678,
      "loss": 2.1461,
      "step": 56237
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7482208013534546,
      "learning_rate": 0.0005160368930844035,
      "loss": 2.21,
      "step": 56238
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.6642420887947083,
      "learning_rate": 0.0005160340548422739,
      "loss": 2.1285,
      "step": 56239
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.0235844850540161,
      "learning_rate": 0.0005160312165599795,
      "loss": 2.2485,
      "step": 56240
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7591104507446289,
      "learning_rate": 0.0005160283782375208,
      "loss": 2.2812,
      "step": 56241
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7870721220970154,
      "learning_rate": 0.0005160255398748984,
      "loss": 2.3306,
      "step": 56242
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.6640816926956177,
      "learning_rate": 0.0005160227014721127,
      "loss": 2.2944,
      "step": 56243
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.6765477061271667,
      "learning_rate": 0.0005160198630291643,
      "loss": 2.2035,
      "step": 56244
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7407061457633972,
      "learning_rate": 0.0005160170245460539,
      "loss": 2.2435,
      "step": 56245
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7233608365058899,
      "learning_rate": 0.0005160141860227816,
      "loss": 2.4267,
      "step": 56246
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7520793080329895,
      "learning_rate": 0.0005160113474593483,
      "loss": 2.1752,
      "step": 56247
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.752475380897522,
      "learning_rate": 0.0005160085088557545,
      "loss": 2.2862,
      "step": 56248
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.8615933060646057,
      "learning_rate": 0.0005160056702120005,
      "loss": 2.3778,
      "step": 56249
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.9238002300262451,
      "learning_rate": 0.000516002831528087,
      "loss": 2.1748,
      "step": 56250
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7869982123374939,
      "learning_rate": 0.0005159999928040146,
      "loss": 2.2787,
      "step": 56251
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7986640930175781,
      "learning_rate": 0.0005159971540397837,
      "loss": 2.2873,
      "step": 56252
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7269471287727356,
      "learning_rate": 0.0005159943152353947,
      "loss": 1.976,
      "step": 56253
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7537637948989868,
      "learning_rate": 0.0005159914763908485,
      "loss": 2.3227,
      "step": 56254
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7300620675086975,
      "learning_rate": 0.0005159886375061453,
      "loss": 2.3266,
      "step": 56255
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7375168800354004,
      "learning_rate": 0.0005159857985812857,
      "loss": 2.5771,
      "step": 56256
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.8543307781219482,
      "learning_rate": 0.0005159829596162702,
      "loss": 2.1826,
      "step": 56257
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7954326868057251,
      "learning_rate": 0.0005159801206110995,
      "loss": 2.1597,
      "step": 56258
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.6682620644569397,
      "learning_rate": 0.000515977281565774,
      "loss": 2.2644,
      "step": 56259
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.723961591720581,
      "learning_rate": 0.0005159744424802943,
      "loss": 2.2184,
      "step": 56260
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7186155915260315,
      "learning_rate": 0.0005159716033546608,
      "loss": 2.2422,
      "step": 56261
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7726778388023376,
      "learning_rate": 0.0005159687641888741,
      "loss": 2.4756,
      "step": 56262
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.6711247563362122,
      "learning_rate": 0.0005159659249829348,
      "loss": 2.2452,
      "step": 56263
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.9501100778579712,
      "learning_rate": 0.0005159630857368432,
      "loss": 2.5271,
      "step": 56264
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7930868864059448,
      "learning_rate": 0.0005159602464506003,
      "loss": 2.5163,
      "step": 56265
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.801067590713501,
      "learning_rate": 0.000515957407124206,
      "loss": 2.0464,
      "step": 56266
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7757721543312073,
      "learning_rate": 0.0005159545677576613,
      "loss": 2.389,
      "step": 56267
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7663279175758362,
      "learning_rate": 0.0005159517283509665,
      "loss": 2.346,
      "step": 56268
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7268264889717102,
      "learning_rate": 0.0005159488889041222,
      "loss": 2.2419,
      "step": 56269
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7541102170944214,
      "learning_rate": 0.000515946049417129,
      "loss": 2.208,
      "step": 56270
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.0212186574935913,
      "learning_rate": 0.0005159432098899873,
      "loss": 2.3131,
      "step": 56271
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7632474899291992,
      "learning_rate": 0.0005159403703226976,
      "loss": 2.186,
      "step": 56272
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7364647388458252,
      "learning_rate": 0.0005159375307152605,
      "loss": 2.1274,
      "step": 56273
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.00271475315094,
      "learning_rate": 0.0005159346910676767,
      "loss": 2.2891,
      "step": 56274
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.8249071836471558,
      "learning_rate": 0.0005159318513799465,
      "loss": 2.2817,
      "step": 56275
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7410593628883362,
      "learning_rate": 0.0005159290116520704,
      "loss": 2.3833,
      "step": 56276
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7390294075012207,
      "learning_rate": 0.0005159261718840492,
      "loss": 2.1144,
      "step": 56277
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.9959478378295898,
      "learning_rate": 0.0005159233320758831,
      "loss": 2.3392,
      "step": 56278
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.714387834072113,
      "learning_rate": 0.0005159204922275728,
      "loss": 2.2703,
      "step": 56279
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.8031436204910278,
      "learning_rate": 0.0005159176523391189,
      "loss": 2.2747,
      "step": 56280
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7447693943977356,
      "learning_rate": 0.0005159148124105218,
      "loss": 2.2122,
      "step": 56281
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.8370769619941711,
      "learning_rate": 0.000515911972441782,
      "loss": 2.1156,
      "step": 56282
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7804773449897766,
      "learning_rate": 0.0005159091324329001,
      "loss": 1.9696,
      "step": 56283
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7743837237358093,
      "learning_rate": 0.0005159062923838767,
      "loss": 2.1904,
      "step": 56284
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.9559649229049683,
      "learning_rate": 0.0005159034522947122,
      "loss": 1.84,
      "step": 56285
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7737144827842712,
      "learning_rate": 0.000515900612165407,
      "loss": 2.4679,
      "step": 56286
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7521674633026123,
      "learning_rate": 0.0005158977719959619,
      "loss": 2.3634,
      "step": 56287
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.8818966746330261,
      "learning_rate": 0.0005158949317863775,
      "loss": 2.5078,
      "step": 56288
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.8290024399757385,
      "learning_rate": 0.000515892091536654,
      "loss": 2.294,
      "step": 56289
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.8287376761436462,
      "learning_rate": 0.0005158892512467921,
      "loss": 2.3304,
      "step": 56290
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.8345792889595032,
      "learning_rate": 0.0005158864109167924,
      "loss": 2.2236,
      "step": 56291
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.0917108058929443,
      "learning_rate": 0.0005158835705466554,
      "loss": 2.161,
      "step": 56292
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.8581992983818054,
      "learning_rate": 0.0005158807301363813,
      "loss": 2.0586,
      "step": 56293
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7574808597564697,
      "learning_rate": 0.000515877889685971,
      "loss": 2.326,
      "step": 56294
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.981640100479126,
      "learning_rate": 0.000515875049195425,
      "loss": 2.1197,
      "step": 56295
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.8816571235656738,
      "learning_rate": 0.0005158722086647437,
      "loss": 2.4896,
      "step": 56296
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7434976100921631,
      "learning_rate": 0.0005158693680939277,
      "loss": 2.3305,
      "step": 56297
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7922999858856201,
      "learning_rate": 0.0005158665274829776,
      "loss": 2.2437,
      "step": 56298
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.9247074723243713,
      "learning_rate": 0.0005158636868318936,
      "loss": 2.2562,
      "step": 56299
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7397540807723999,
      "learning_rate": 0.0005158608461406768,
      "loss": 2.4351,
      "step": 56300
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.6916784644126892,
      "learning_rate": 0.0005158580054093272,
      "loss": 2.1685,
      "step": 56301
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.8379308581352234,
      "learning_rate": 0.0005158551646378455,
      "loss": 2.2517,
      "step": 56302
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7504104375839233,
      "learning_rate": 0.0005158523238262322,
      "loss": 2.0266,
      "step": 56303
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7679620385169983,
      "learning_rate": 0.000515849482974488,
      "loss": 2.3987,
      "step": 56304
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.6743475794792175,
      "learning_rate": 0.0005158466420826132,
      "loss": 1.9507,
      "step": 56305
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7821703553199768,
      "learning_rate": 0.0005158438011506086,
      "loss": 1.9417,
      "step": 56306
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.9075911045074463,
      "learning_rate": 0.0005158409601784743,
      "loss": 2.1738,
      "step": 56307
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7172383069992065,
      "learning_rate": 0.0005158381191662114,
      "loss": 2.1768,
      "step": 56308
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.8156107068061829,
      "learning_rate": 0.0005158352781138199,
      "loss": 2.27,
      "step": 56309
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.800399661064148,
      "learning_rate": 0.0005158324370213007,
      "loss": 1.9736,
      "step": 56310
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7295552492141724,
      "learning_rate": 0.0005158295958886542,
      "loss": 2.3021,
      "step": 56311
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.745048999786377,
      "learning_rate": 0.0005158267547158807,
      "loss": 2.1088,
      "step": 56312
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.9996744990348816,
      "learning_rate": 0.0005158239135029809,
      "loss": 2.369,
      "step": 56313
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7161283493041992,
      "learning_rate": 0.0005158210722499556,
      "loss": 2.3734,
      "step": 56314
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.746820330619812,
      "learning_rate": 0.0005158182309568049,
      "loss": 2.1684,
      "step": 56315
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7620349526405334,
      "learning_rate": 0.0005158153896235297,
      "loss": 2.2219,
      "step": 56316
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.8353061676025391,
      "learning_rate": 0.0005158125482501302,
      "loss": 2.4825,
      "step": 56317
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7974761724472046,
      "learning_rate": 0.0005158097068366071,
      "loss": 2.3642,
      "step": 56318
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7530809044837952,
      "learning_rate": 0.000515806865382961,
      "loss": 2.4312,
      "step": 56319
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.721116840839386,
      "learning_rate": 0.0005158040238891922,
      "loss": 2.1618,
      "step": 56320
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7192631959915161,
      "learning_rate": 0.0005158011823553014,
      "loss": 2.4048,
      "step": 56321
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.882498562335968,
      "learning_rate": 0.0005157983407812891,
      "loss": 2.089,
      "step": 56322
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7148746252059937,
      "learning_rate": 0.0005157954991671558,
      "loss": 2.1832,
      "step": 56323
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7319662570953369,
      "learning_rate": 0.0005157926575129021,
      "loss": 2.0702,
      "step": 56324
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7574344873428345,
      "learning_rate": 0.0005157898158185284,
      "loss": 2.2265,
      "step": 56325
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.8691664934158325,
      "learning_rate": 0.0005157869740840353,
      "loss": 2.1365,
      "step": 56326
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.9020133018493652,
      "learning_rate": 0.0005157841323094233,
      "loss": 2.1947,
      "step": 56327
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7417045831680298,
      "learning_rate": 0.0005157812904946931,
      "loss": 2.2035,
      "step": 56328
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.8788861036300659,
      "learning_rate": 0.0005157784486398449,
      "loss": 2.0442,
      "step": 56329
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7328726649284363,
      "learning_rate": 0.0005157756067448795,
      "loss": 2.3296,
      "step": 56330
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7022736072540283,
      "learning_rate": 0.0005157727648097974,
      "loss": 2.2889,
      "step": 56331
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.916236400604248,
      "learning_rate": 0.0005157699228345991,
      "loss": 2.2924,
      "step": 56332
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7686416506767273,
      "learning_rate": 0.000515767080819285,
      "loss": 2.1301,
      "step": 56333
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.6759086847305298,
      "learning_rate": 0.0005157642387638557,
      "loss": 2.1747,
      "step": 56334
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.8293079137802124,
      "learning_rate": 0.0005157613966683118,
      "loss": 2.3099,
      "step": 56335
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.8019601702690125,
      "learning_rate": 0.0005157585545326538,
      "loss": 2.4924,
      "step": 56336
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7584922909736633,
      "learning_rate": 0.0005157557123568822,
      "loss": 2.2907,
      "step": 56337
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7008852362632751,
      "learning_rate": 0.0005157528701409975,
      "loss": 2.192,
      "step": 56338
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.255517840385437,
      "learning_rate": 0.0005157500278850002,
      "loss": 2.197,
      "step": 56339
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7436196208000183,
      "learning_rate": 0.0005157471855888911,
      "loss": 2.0782,
      "step": 56340
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7305009961128235,
      "learning_rate": 0.0005157443432526705,
      "loss": 2.2035,
      "step": 56341
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.8403096795082092,
      "learning_rate": 0.0005157415008763389,
      "loss": 2.2521,
      "step": 56342
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.742699384689331,
      "learning_rate": 0.0005157386584598968,
      "loss": 2.276,
      "step": 56343
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7534282803535461,
      "learning_rate": 0.0005157358160033449,
      "loss": 2.2547,
      "step": 56344
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7139395475387573,
      "learning_rate": 0.0005157329735066836,
      "loss": 2.3425,
      "step": 56345
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.8476735949516296,
      "learning_rate": 0.0005157301309699135,
      "loss": 2.2137,
      "step": 56346
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.6904986500740051,
      "learning_rate": 0.000515727288393035,
      "loss": 2.0712,
      "step": 56347
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7654688954353333,
      "learning_rate": 0.0005157244457760488,
      "loss": 2.3355,
      "step": 56348
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7371604442596436,
      "learning_rate": 0.0005157216031189554,
      "loss": 2.3284,
      "step": 56349
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.796524703502655,
      "learning_rate": 0.0005157187604217553,
      "loss": 2.4413,
      "step": 56350
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7675008177757263,
      "learning_rate": 0.0005157159176844489,
      "loss": 1.9695,
      "step": 56351
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.8726879358291626,
      "learning_rate": 0.000515713074907037,
      "loss": 2.347,
      "step": 56352
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.9238430857658386,
      "learning_rate": 0.0005157102320895198,
      "loss": 2.4158,
      "step": 56353
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7477434873580933,
      "learning_rate": 0.000515707389231898,
      "loss": 2.3472,
      "step": 56354
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.6024700403213501,
      "learning_rate": 0.0005157045463341722,
      "loss": 2.3445,
      "step": 56355
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.082561731338501,
      "learning_rate": 0.0005157017033963428,
      "loss": 2.2267,
      "step": 56356
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7170639038085938,
      "learning_rate": 0.0005156988604184104,
      "loss": 2.1668,
      "step": 56357
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7143734097480774,
      "learning_rate": 0.0005156960174003756,
      "loss": 1.9976,
      "step": 56358
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.8426660299301147,
      "learning_rate": 0.0005156931743422387,
      "loss": 2.0559,
      "step": 56359
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.0424504280090332,
      "learning_rate": 0.0005156903312440005,
      "loss": 2.2271,
      "step": 56360
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.758974015712738,
      "learning_rate": 0.0005156874881056613,
      "loss": 2.1695,
      "step": 56361
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7685898542404175,
      "learning_rate": 0.0005156846449272217,
      "loss": 2.2088,
      "step": 56362
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.8470433354377747,
      "learning_rate": 0.0005156818017086823,
      "loss": 2.4286,
      "step": 56363
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.9006826281547546,
      "learning_rate": 0.0005156789584500436,
      "loss": 2.2022,
      "step": 56364
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7454227805137634,
      "learning_rate": 0.0005156761151513061,
      "loss": 2.1199,
      "step": 56365
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.9651607275009155,
      "learning_rate": 0.0005156732718124702,
      "loss": 2.3022,
      "step": 56366
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.8862303495407104,
      "learning_rate": 0.0005156704284335368,
      "loss": 2.3002,
      "step": 56367
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.6588283181190491,
      "learning_rate": 0.0005156675850145062,
      "loss": 2.0369,
      "step": 56368
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.9828407764434814,
      "learning_rate": 0.0005156647415553787,
      "loss": 2.2633,
      "step": 56369
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.8413828611373901,
      "learning_rate": 0.0005156618980561552,
      "loss": 2.4357,
      "step": 56370
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.9732651710510254,
      "learning_rate": 0.000515659054516836,
      "loss": 2.1462,
      "step": 56371
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.9049479961395264,
      "learning_rate": 0.0005156562109374218,
      "loss": 2.1836,
      "step": 56372
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.8642070293426514,
      "learning_rate": 0.000515653367317913,
      "loss": 2.2926,
      "step": 56373
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.9673815369606018,
      "learning_rate": 0.0005156505236583102,
      "loss": 2.2148,
      "step": 56374
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7087658643722534,
      "learning_rate": 0.0005156476799586139,
      "loss": 2.2489,
      "step": 56375
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.9185181856155396,
      "learning_rate": 0.0005156448362188247,
      "loss": 2.5444,
      "step": 56376
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.132635474205017,
      "learning_rate": 0.0005156419924389428,
      "loss": 2.0163,
      "step": 56377
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7857977747917175,
      "learning_rate": 0.0005156391486189692,
      "loss": 2.1549,
      "step": 56378
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7075275778770447,
      "learning_rate": 0.0005156363047589042,
      "loss": 2.2429,
      "step": 56379
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.8361416459083557,
      "learning_rate": 0.0005156334608587482,
      "loss": 2.4139,
      "step": 56380
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.8783215880393982,
      "learning_rate": 0.000515630616918502,
      "loss": 2.4357,
      "step": 56381
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7285533547401428,
      "learning_rate": 0.000515627772938166,
      "loss": 2.2571,
      "step": 56382
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.82136070728302,
      "learning_rate": 0.0005156249289177406,
      "loss": 2.433,
      "step": 56383
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.9143562912940979,
      "learning_rate": 0.0005156220848572265,
      "loss": 2.1325,
      "step": 56384
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7398934364318848,
      "learning_rate": 0.0005156192407566243,
      "loss": 2.3523,
      "step": 56385
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.9124652743339539,
      "learning_rate": 0.0005156163966159342,
      "loss": 2.2411,
      "step": 56386
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.8004078269004822,
      "learning_rate": 0.0005156135524351572,
      "loss": 2.2426,
      "step": 56387
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7422825694084167,
      "learning_rate": 0.0005156107082142934,
      "loss": 2.3347,
      "step": 56388
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7607647776603699,
      "learning_rate": 0.0005156078639533436,
      "loss": 2.141,
      "step": 56389
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7936897873878479,
      "learning_rate": 0.0005156050196523081,
      "loss": 2.2275,
      "step": 56390
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7708426713943481,
      "learning_rate": 0.0005156021753111878,
      "loss": 2.1402,
      "step": 56391
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.682715892791748,
      "learning_rate": 0.0005155993309299827,
      "loss": 2.3327,
      "step": 56392
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.8462152481079102,
      "learning_rate": 0.0005155964865086938,
      "loss": 2.1373,
      "step": 56393
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.9033911824226379,
      "learning_rate": 0.0005155936420473213,
      "loss": 2.3126,
      "step": 56394
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.699505627155304,
      "learning_rate": 0.000515590797545866,
      "loss": 2.2279,
      "step": 56395
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7653932571411133,
      "learning_rate": 0.0005155879530043282,
      "loss": 2.1973,
      "step": 56396
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.8114623427391052,
      "learning_rate": 0.0005155851084227087,
      "loss": 2.2513,
      "step": 56397
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.6738584041595459,
      "learning_rate": 0.0005155822638010076,
      "loss": 2.0916,
      "step": 56398
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7661226987838745,
      "learning_rate": 0.0005155794191392259,
      "loss": 2.2516,
      "step": 56399
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.909416675567627,
      "learning_rate": 0.0005155765744373638,
      "loss": 2.2627,
      "step": 56400
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7678772807121277,
      "learning_rate": 0.000515573729695422,
      "loss": 2.4238,
      "step": 56401
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7736102342605591,
      "learning_rate": 0.0005155708849134009,
      "loss": 2.1422,
      "step": 56402
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.8683308362960815,
      "learning_rate": 0.0005155680400913011,
      "loss": 2.1281,
      "step": 56403
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.701153039932251,
      "learning_rate": 0.0005155651952291233,
      "loss": 2.2247,
      "step": 56404
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7281872034072876,
      "learning_rate": 0.0005155623503268677,
      "loss": 2.1487,
      "step": 56405
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7409166693687439,
      "learning_rate": 0.0005155595053845351,
      "loss": 2.2295,
      "step": 56406
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7783638834953308,
      "learning_rate": 0.0005155566604021258,
      "loss": 2.2504,
      "step": 56407
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7510379552841187,
      "learning_rate": 0.0005155538153796406,
      "loss": 2.4998,
      "step": 56408
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.8968744874000549,
      "learning_rate": 0.0005155509703170798,
      "loss": 2.1566,
      "step": 56409
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.8622848987579346,
      "learning_rate": 0.000515548125214444,
      "loss": 2.2373,
      "step": 56410
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.6967481970787048,
      "learning_rate": 0.0005155452800717337,
      "loss": 2.1419,
      "step": 56411
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7601476311683655,
      "learning_rate": 0.0005155424348889494,
      "loss": 2.2836,
      "step": 56412
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7602613568305969,
      "learning_rate": 0.0005155395896660919,
      "loss": 2.3366,
      "step": 56413
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.672031044960022,
      "learning_rate": 0.0005155367444031614,
      "loss": 2.1669,
      "step": 56414
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7462900280952454,
      "learning_rate": 0.0005155338991001586,
      "loss": 2.2347,
      "step": 56415
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.757041871547699,
      "learning_rate": 0.0005155310537570838,
      "loss": 2.2733,
      "step": 56416
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7315559387207031,
      "learning_rate": 0.0005155282083739379,
      "loss": 2.149,
      "step": 56417
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7271373271942139,
      "learning_rate": 0.0005155253629507212,
      "loss": 2.2619,
      "step": 56418
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.752449095249176,
      "learning_rate": 0.0005155225174874343,
      "loss": 2.1365,
      "step": 56419
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7244299650192261,
      "learning_rate": 0.0005155196719840776,
      "loss": 2.1069,
      "step": 56420
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.761269748210907,
      "learning_rate": 0.0005155168264406519,
      "loss": 2.2523,
      "step": 56421
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.8021301627159119,
      "learning_rate": 0.0005155139808571574,
      "loss": 2.1749,
      "step": 56422
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.8050729632377625,
      "learning_rate": 0.0005155111352335947,
      "loss": 2.2119,
      "step": 56423
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7721042633056641,
      "learning_rate": 0.0005155082895699645,
      "loss": 2.2845,
      "step": 56424
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7036551237106323,
      "learning_rate": 0.0005155054438662673,
      "loss": 2.0399,
      "step": 56425
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7593303918838501,
      "learning_rate": 0.0005155025981225036,
      "loss": 2.1639,
      "step": 56426
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.8240087032318115,
      "learning_rate": 0.0005154997523386738,
      "loss": 2.1166,
      "step": 56427
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7395954728126526,
      "learning_rate": 0.0005154969065147787,
      "loss": 2.1632,
      "step": 56428
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7018991708755493,
      "learning_rate": 0.0005154940606508184,
      "loss": 2.2405,
      "step": 56429
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.8535617589950562,
      "learning_rate": 0.0005154912147467938,
      "loss": 2.286,
      "step": 56430
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7044080495834351,
      "learning_rate": 0.0005154883688027053,
      "loss": 2.3853,
      "step": 56431
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7416076064109802,
      "learning_rate": 0.0005154855228185535,
      "loss": 1.9733,
      "step": 56432
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7279612421989441,
      "learning_rate": 0.0005154826767943388,
      "loss": 2.0887,
      "step": 56433
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.8654171228408813,
      "learning_rate": 0.0005154798307300619,
      "loss": 2.1449,
      "step": 56434
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7041286826133728,
      "learning_rate": 0.0005154769846257231,
      "loss": 2.2235,
      "step": 56435
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.8428298234939575,
      "learning_rate": 0.0005154741384813231,
      "loss": 2.2742,
      "step": 56436
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.8388693332672119,
      "learning_rate": 0.0005154712922968625,
      "loss": 2.2403,
      "step": 56437
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7839720845222473,
      "learning_rate": 0.0005154684460723415,
      "loss": 2.2885,
      "step": 56438
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7043125629425049,
      "learning_rate": 0.0005154655998077612,
      "loss": 2.3701,
      "step": 56439
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7314854860305786,
      "learning_rate": 0.0005154627535031215,
      "loss": 2.0324,
      "step": 56440
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7405462265014648,
      "learning_rate": 0.0005154599071584233,
      "loss": 2.4276,
      "step": 56441
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7314215302467346,
      "learning_rate": 0.000515457060773667,
      "loss": 2.4198,
      "step": 56442
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7344151735305786,
      "learning_rate": 0.0005154542143488532,
      "loss": 2.2481,
      "step": 56443
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7167731523513794,
      "learning_rate": 0.0005154513678839824,
      "loss": 2.4027,
      "step": 56444
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7842950224876404,
      "learning_rate": 0.000515448521379055,
      "loss": 2.4658,
      "step": 56445
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7151220440864563,
      "learning_rate": 0.0005154456748340718,
      "loss": 2.2188,
      "step": 56446
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7364509701728821,
      "learning_rate": 0.0005154428282490331,
      "loss": 2.1323,
      "step": 56447
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7849239706993103,
      "learning_rate": 0.0005154399816239395,
      "loss": 2.0974,
      "step": 56448
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7319357395172119,
      "learning_rate": 0.0005154371349587916,
      "loss": 2.3872,
      "step": 56449
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.6880014538764954,
      "learning_rate": 0.0005154342882535899,
      "loss": 2.2602,
      "step": 56450
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7806527018547058,
      "learning_rate": 0.0005154314415083349,
      "loss": 2.2603,
      "step": 56451
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.0917938947677612,
      "learning_rate": 0.000515428594723027,
      "loss": 2.1224,
      "step": 56452
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.6847823262214661,
      "learning_rate": 0.000515425747897667,
      "loss": 2.3614,
      "step": 56453
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.8023833632469177,
      "learning_rate": 0.0005154229010322552,
      "loss": 2.2616,
      "step": 56454
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.6547929644584656,
      "learning_rate": 0.0005154200541267923,
      "loss": 2.3077,
      "step": 56455
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7275779843330383,
      "learning_rate": 0.0005154172071812785,
      "loss": 2.226,
      "step": 56456
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7100690007209778,
      "learning_rate": 0.0005154143601957149,
      "loss": 2.2228,
      "step": 56457
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7391397356987,
      "learning_rate": 0.0005154115131701016,
      "loss": 2.2367,
      "step": 56458
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.793321430683136,
      "learning_rate": 0.0005154086661044392,
      "loss": 2.179,
      "step": 56459
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7246817350387573,
      "learning_rate": 0.0005154058189987282,
      "loss": 2.2906,
      "step": 56460
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7139421701431274,
      "learning_rate": 0.0005154029718529693,
      "loss": 2.2821,
      "step": 56461
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7559739947319031,
      "learning_rate": 0.0005154001246671628,
      "loss": 2.2043,
      "step": 56462
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7441691756248474,
      "learning_rate": 0.0005153972774413094,
      "loss": 2.5518,
      "step": 56463
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.6897080540657043,
      "learning_rate": 0.0005153944301754097,
      "loss": 2.1879,
      "step": 56464
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.81052166223526,
      "learning_rate": 0.0005153915828694641,
      "loss": 2.3224,
      "step": 56465
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7599292397499084,
      "learning_rate": 0.0005153887355234729,
      "loss": 2.2642,
      "step": 56466
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.687701404094696,
      "learning_rate": 0.000515385888137437,
      "loss": 2.12,
      "step": 56467
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7640250325202942,
      "learning_rate": 0.0005153830407113569,
      "loss": 2.2139,
      "step": 56468
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7029498815536499,
      "learning_rate": 0.0005153801932452329,
      "loss": 2.4949,
      "step": 56469
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7404844164848328,
      "learning_rate": 0.0005153773457390658,
      "loss": 2.3303,
      "step": 56470
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.8465098142623901,
      "learning_rate": 0.0005153744981928559,
      "loss": 2.0851,
      "step": 56471
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7749857306480408,
      "learning_rate": 0.0005153716506066038,
      "loss": 2.1766,
      "step": 56472
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7587692141532898,
      "learning_rate": 0.00051536880298031,
      "loss": 2.2557,
      "step": 56473
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.8485618829727173,
      "learning_rate": 0.0005153659553139751,
      "loss": 2.2446,
      "step": 56474
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7321220636367798,
      "learning_rate": 0.0005153631076075997,
      "loss": 2.1344,
      "step": 56475
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.783653974533081,
      "learning_rate": 0.0005153602598611841,
      "loss": 2.1695,
      "step": 56476
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.8242883682250977,
      "learning_rate": 0.0005153574120747291,
      "loss": 2.2844,
      "step": 56477
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7480528354644775,
      "learning_rate": 0.0005153545642482349,
      "loss": 2.0771,
      "step": 56478
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7690389752388,
      "learning_rate": 0.0005153517163817024,
      "loss": 2.1844,
      "step": 56479
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.8260897397994995,
      "learning_rate": 0.000515348868475132,
      "loss": 2.1155,
      "step": 56480
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.9413706064224243,
      "learning_rate": 0.000515346020528524,
      "loss": 2.0262,
      "step": 56481
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7184965014457703,
      "learning_rate": 0.0005153431725418791,
      "loss": 2.2514,
      "step": 56482
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7821534276008606,
      "learning_rate": 0.000515340324515198,
      "loss": 2.0992,
      "step": 56483
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.9317594170570374,
      "learning_rate": 0.000515337476448481,
      "loss": 2.2034,
      "step": 56484
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7933982610702515,
      "learning_rate": 0.0005153346283417287,
      "loss": 2.3822,
      "step": 56485
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.8688056468963623,
      "learning_rate": 0.0005153317801949416,
      "loss": 2.099,
      "step": 56486
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.9223629832267761,
      "learning_rate": 0.0005153289320081203,
      "loss": 2.1203,
      "step": 56487
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7456651329994202,
      "learning_rate": 0.0005153260837812652,
      "loss": 2.3984,
      "step": 56488
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.748771071434021,
      "learning_rate": 0.000515323235514377,
      "loss": 2.3815,
      "step": 56489
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7790266871452332,
      "learning_rate": 0.0005153203872074561,
      "loss": 2.2159,
      "step": 56490
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.9493669867515564,
      "learning_rate": 0.000515317538860503,
      "loss": 2.2898,
      "step": 56491
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.740362823009491,
      "learning_rate": 0.0005153146904735185,
      "loss": 2.1749,
      "step": 56492
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7355989217758179,
      "learning_rate": 0.0005153118420465028,
      "loss": 2.4853,
      "step": 56493
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.8682115077972412,
      "learning_rate": 0.0005153089935794567,
      "loss": 2.1575,
      "step": 56494
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.8198225498199463,
      "learning_rate": 0.0005153061450723804,
      "loss": 2.2093,
      "step": 56495
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.6966723203659058,
      "learning_rate": 0.0005153032965252748,
      "loss": 2.3275,
      "step": 56496
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.780907154083252,
      "learning_rate": 0.0005153004479381401,
      "loss": 2.4613,
      "step": 56497
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.791386067867279,
      "learning_rate": 0.0005152975993109771,
      "loss": 2.1509,
      "step": 56498
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7164132595062256,
      "learning_rate": 0.0005152947506437861,
      "loss": 2.2481,
      "step": 56499
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.8505508899688721,
      "learning_rate": 0.0005152919019365678,
      "loss": 2.2832,
      "step": 56500
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7336428761482239,
      "learning_rate": 0.0005152890531893227,
      "loss": 2.086,
      "step": 56501
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7904344201087952,
      "learning_rate": 0.0005152862044020513,
      "loss": 2.0388,
      "step": 56502
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.0938875675201416,
      "learning_rate": 0.0005152833555747541,
      "loss": 2.1043,
      "step": 56503
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.722948431968689,
      "learning_rate": 0.0005152805067074317,
      "loss": 2.3407,
      "step": 56504
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7202146649360657,
      "learning_rate": 0.0005152776578000846,
      "loss": 2.1711,
      "step": 56505
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.8400896191596985,
      "learning_rate": 0.0005152748088527132,
      "loss": 2.3663,
      "step": 56506
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.8642799258232117,
      "learning_rate": 0.0005152719598653183,
      "loss": 2.2442,
      "step": 56507
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.8384456038475037,
      "learning_rate": 0.0005152691108379001,
      "loss": 2.2196,
      "step": 56508
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7625008821487427,
      "learning_rate": 0.0005152662617704595,
      "loss": 2.409,
      "step": 56509
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.1161928176879883,
      "learning_rate": 0.0005152634126629967,
      "loss": 2.0499,
      "step": 56510
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.9425967335700989,
      "learning_rate": 0.0005152605635155125,
      "loss": 2.2948,
      "step": 56511
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7470546960830688,
      "learning_rate": 0.0005152577143280072,
      "loss": 2.2321,
      "step": 56512
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.8447557091712952,
      "learning_rate": 0.0005152548651004814,
      "loss": 2.5501,
      "step": 56513
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.1099382638931274,
      "learning_rate": 0.0005152520158329358,
      "loss": 2.2127,
      "step": 56514
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7163118720054626,
      "learning_rate": 0.0005152491665253706,
      "loss": 2.3938,
      "step": 56515
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.8924980163574219,
      "learning_rate": 0.0005152463171777867,
      "loss": 2.3206,
      "step": 56516
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7465822696685791,
      "learning_rate": 0.0005152434677901843,
      "loss": 2.1753,
      "step": 56517
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.6703731417655945,
      "learning_rate": 0.0005152406183625642,
      "loss": 2.3099,
      "step": 56518
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7219419479370117,
      "learning_rate": 0.0005152377688949268,
      "loss": 2.0366,
      "step": 56519
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7163859605789185,
      "learning_rate": 0.0005152349193872724,
      "loss": 2.2731,
      "step": 56520
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.715546727180481,
      "learning_rate": 0.000515232069839602,
      "loss": 2.1435,
      "step": 56521
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7351075410842896,
      "learning_rate": 0.0005152292202519159,
      "loss": 2.2682,
      "step": 56522
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.704529881477356,
      "learning_rate": 0.0005152263706242145,
      "loss": 2.3738,
      "step": 56523
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7839866876602173,
      "learning_rate": 0.0005152235209564985,
      "loss": 2.4823,
      "step": 56524
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7256673574447632,
      "learning_rate": 0.0005152206712487684,
      "loss": 2.2744,
      "step": 56525
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7942736744880676,
      "learning_rate": 0.0005152178215010248,
      "loss": 2.2874,
      "step": 56526
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.6677142977714539,
      "learning_rate": 0.0005152149717132681,
      "loss": 2.274,
      "step": 56527
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.6422356367111206,
      "learning_rate": 0.0005152121218854988,
      "loss": 2.211,
      "step": 56528
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.6603028178215027,
      "learning_rate": 0.0005152092720177175,
      "loss": 2.3114,
      "step": 56529
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7180246114730835,
      "learning_rate": 0.0005152064221099247,
      "loss": 2.0038,
      "step": 56530
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7144436836242676,
      "learning_rate": 0.0005152035721621212,
      "loss": 2.1697,
      "step": 56531
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.6921046376228333,
      "learning_rate": 0.000515200722174307,
      "loss": 2.2267,
      "step": 56532
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7500315308570862,
      "learning_rate": 0.0005151978721464831,
      "loss": 2.2019,
      "step": 56533
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7765280604362488,
      "learning_rate": 0.0005151950220786498,
      "loss": 2.1288,
      "step": 56534
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7065517902374268,
      "learning_rate": 0.0005151921719708076,
      "loss": 2.258,
      "step": 56535
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.6891224980354309,
      "learning_rate": 0.0005151893218229573,
      "loss": 2.3312,
      "step": 56536
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7177744507789612,
      "learning_rate": 0.0005151864716350991,
      "loss": 2.1987,
      "step": 56537
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.6955761909484863,
      "learning_rate": 0.0005151836214072338,
      "loss": 2.0747,
      "step": 56538
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7769338488578796,
      "learning_rate": 0.0005151807711393616,
      "loss": 2.4161,
      "step": 56539
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.6583772301673889,
      "learning_rate": 0.0005151779208314835,
      "loss": 2.4607,
      "step": 56540
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.6906964778900146,
      "learning_rate": 0.0005151750704835995,
      "loss": 2.1323,
      "step": 56541
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.6731166243553162,
      "learning_rate": 0.0005151722200957105,
      "loss": 2.1122,
      "step": 56542
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.6743695139884949,
      "learning_rate": 0.000515169369667817,
      "loss": 2.0817,
      "step": 56543
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.6920817494392395,
      "learning_rate": 0.0005151665191999193,
      "loss": 2.096,
      "step": 56544
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7294353246688843,
      "learning_rate": 0.0005151636686920182,
      "loss": 2.4346,
      "step": 56545
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7594826221466064,
      "learning_rate": 0.0005151608181441141,
      "loss": 2.137,
      "step": 56546
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7019851803779602,
      "learning_rate": 0.0005151579675562075,
      "loss": 2.3159,
      "step": 56547
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7443049550056458,
      "learning_rate": 0.0005151551169282991,
      "loss": 2.3629,
      "step": 56548
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7514275908470154,
      "learning_rate": 0.0005151522662603891,
      "loss": 1.942,
      "step": 56549
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7684703469276428,
      "learning_rate": 0.0005151494155524783,
      "loss": 2.3305,
      "step": 56550
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7195431590080261,
      "learning_rate": 0.0005151465648045672,
      "loss": 2.2014,
      "step": 56551
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7823132276535034,
      "learning_rate": 0.0005151437140166564,
      "loss": 2.1972,
      "step": 56552
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.654643714427948,
      "learning_rate": 0.0005151408631887462,
      "loss": 2.0741,
      "step": 56553
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7014924883842468,
      "learning_rate": 0.0005151380123208372,
      "loss": 2.1639,
      "step": 56554
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7192065119743347,
      "learning_rate": 0.00051513516141293,
      "loss": 2.472,
      "step": 56555
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7489738464355469,
      "learning_rate": 0.0005151323104650252,
      "loss": 2.4211,
      "step": 56556
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7673882246017456,
      "learning_rate": 0.0005151294594771231,
      "loss": 2.2873,
      "step": 56557
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.763823926448822,
      "learning_rate": 0.0005151266084492246,
      "loss": 2.1544,
      "step": 56558
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.6964287757873535,
      "learning_rate": 0.0005151237573813299,
      "loss": 2.1644,
      "step": 56559
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.6986036896705627,
      "learning_rate": 0.0005151209062734396,
      "loss": 2.2752,
      "step": 56560
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7268717885017395,
      "learning_rate": 0.0005151180551255543,
      "loss": 2.2061,
      "step": 56561
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.6380611062049866,
      "learning_rate": 0.0005151152039376744,
      "loss": 2.2855,
      "step": 56562
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.6235135197639465,
      "learning_rate": 0.0005151123527098006,
      "loss": 2.1866,
      "step": 56563
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7541178464889526,
      "learning_rate": 0.0005151095014419334,
      "loss": 2.3205,
      "step": 56564
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7134823203086853,
      "learning_rate": 0.0005151066501340731,
      "loss": 2.2438,
      "step": 56565
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.736649751663208,
      "learning_rate": 0.0005151037987862206,
      "loss": 2.4021,
      "step": 56566
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.8493576049804688,
      "learning_rate": 0.0005151009473983761,
      "loss": 2.2746,
      "step": 56567
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.6920475959777832,
      "learning_rate": 0.0005150980959705404,
      "loss": 2.2316,
      "step": 56568
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.6812515258789062,
      "learning_rate": 0.0005150952445027139,
      "loss": 2.1791,
      "step": 56569
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.0121431350708008,
      "learning_rate": 0.0005150923929948971,
      "loss": 2.5204,
      "step": 56570
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.6914825439453125,
      "learning_rate": 0.0005150895414470905,
      "loss": 2.1374,
      "step": 56571
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7380592226982117,
      "learning_rate": 0.0005150866898592948,
      "loss": 2.3424,
      "step": 56572
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7974764704704285,
      "learning_rate": 0.0005150838382315103,
      "loss": 2.1002,
      "step": 56573
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.835188090801239,
      "learning_rate": 0.0005150809865637377,
      "loss": 2.0103,
      "step": 56574
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.6829383969306946,
      "learning_rate": 0.0005150781348559776,
      "loss": 2.1086,
      "step": 56575
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.816346287727356,
      "learning_rate": 0.0005150752831082303,
      "loss": 2.174,
      "step": 56576
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.836850106716156,
      "learning_rate": 0.0005150724313204965,
      "loss": 2.2924,
      "step": 56577
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.746634304523468,
      "learning_rate": 0.0005150695794927766,
      "loss": 2.437,
      "step": 56578
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.6632240414619446,
      "learning_rate": 0.0005150667276250713,
      "loss": 2.3299,
      "step": 56579
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.8701983094215393,
      "learning_rate": 0.0005150638757173808,
      "loss": 2.4449,
      "step": 56580
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7161768674850464,
      "learning_rate": 0.0005150610237697062,
      "loss": 2.2794,
      "step": 56581
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.6797935962677002,
      "learning_rate": 0.0005150581717820474,
      "loss": 2.0505,
      "step": 56582
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.8717221021652222,
      "learning_rate": 0.0005150553197544055,
      "loss": 2.0231,
      "step": 56583
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7898204922676086,
      "learning_rate": 0.0005150524676867806,
      "loss": 2.3732,
      "step": 56584
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7190919518470764,
      "learning_rate": 0.0005150496155791734,
      "loss": 2.2211,
      "step": 56585
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7273305058479309,
      "learning_rate": 0.0005150467634315845,
      "loss": 2.1839,
      "step": 56586
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7773005962371826,
      "learning_rate": 0.0005150439112440142,
      "loss": 2.3232,
      "step": 56587
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7841979265213013,
      "learning_rate": 0.0005150410590164633,
      "loss": 2.3652,
      "step": 56588
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7611631155014038,
      "learning_rate": 0.0005150382067489322,
      "loss": 2.4085,
      "step": 56589
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7547680735588074,
      "learning_rate": 0.0005150353544414214,
      "loss": 2.0939,
      "step": 56590
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7648046612739563,
      "learning_rate": 0.0005150325020939314,
      "loss": 2.1941,
      "step": 56591
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.0031076669692993,
      "learning_rate": 0.000515029649706463,
      "loss": 2.2004,
      "step": 56592
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.8139131665229797,
      "learning_rate": 0.0005150267972790163,
      "loss": 2.2643,
      "step": 56593
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7440571784973145,
      "learning_rate": 0.0005150239448115922,
      "loss": 2.1496,
      "step": 56594
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.8919203877449036,
      "learning_rate": 0.000515021092304191,
      "loss": 2.347,
      "step": 56595
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7237376570701599,
      "learning_rate": 0.0005150182397568135,
      "loss": 2.315,
      "step": 56596
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.9548007249832153,
      "learning_rate": 0.0005150153871694599,
      "loss": 2.1837,
      "step": 56597
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.743992805480957,
      "learning_rate": 0.0005150125345421308,
      "loss": 2.3172,
      "step": 56598
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7585505247116089,
      "learning_rate": 0.0005150096818748269,
      "loss": 2.2081,
      "step": 56599
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7600864171981812,
      "learning_rate": 0.0005150068291675487,
      "loss": 2.0966,
      "step": 56600
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.8361397385597229,
      "learning_rate": 0.0005150039764202966,
      "loss": 2.344,
      "step": 56601
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7014527916908264,
      "learning_rate": 0.0005150011236330713,
      "loss": 2.2673,
      "step": 56602
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.6887580752372742,
      "learning_rate": 0.0005149982708058732,
      "loss": 2.2385,
      "step": 56603
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.8462655544281006,
      "learning_rate": 0.0005149954179387028,
      "loss": 2.1192,
      "step": 56604
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.9143741726875305,
      "learning_rate": 0.0005149925650315606,
      "loss": 2.1382,
      "step": 56605
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7256045937538147,
      "learning_rate": 0.0005149897120844475,
      "loss": 2.1246,
      "step": 56606
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.784108579158783,
      "learning_rate": 0.0005149868590973635,
      "loss": 2.2589,
      "step": 56607
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.074246883392334,
      "learning_rate": 0.0005149840060703095,
      "loss": 2.2804,
      "step": 56608
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7636383771896362,
      "learning_rate": 0.0005149811530032859,
      "loss": 2.2635,
      "step": 56609
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.8106160759925842,
      "learning_rate": 0.0005149782998962933,
      "loss": 2.109,
      "step": 56610
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.8514223694801331,
      "learning_rate": 0.0005149754467493321,
      "loss": 2.4113,
      "step": 56611
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7346892356872559,
      "learning_rate": 0.0005149725935624029,
      "loss": 2.2097,
      "step": 56612
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7283356785774231,
      "learning_rate": 0.0005149697403355063,
      "loss": 2.3791,
      "step": 56613
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7581029534339905,
      "learning_rate": 0.0005149668870686427,
      "loss": 2.3095,
      "step": 56614
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.6596775054931641,
      "learning_rate": 0.0005149640337618127,
      "loss": 2.3196,
      "step": 56615
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7328746318817139,
      "learning_rate": 0.0005149611804150169,
      "loss": 2.0613,
      "step": 56616
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7952883839607239,
      "learning_rate": 0.0005149583270282556,
      "loss": 2.2359,
      "step": 56617
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.8002102375030518,
      "learning_rate": 0.0005149554736015297,
      "loss": 2.4444,
      "step": 56618
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.8136364221572876,
      "learning_rate": 0.0005149526201348395,
      "loss": 2.4004,
      "step": 56619
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7633311748504639,
      "learning_rate": 0.0005149497666281854,
      "loss": 2.4191,
      "step": 56620
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7602258324623108,
      "learning_rate": 0.0005149469130815681,
      "loss": 2.318,
      "step": 56621
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7533027529716492,
      "learning_rate": 0.0005149440594949882,
      "loss": 2.4052,
      "step": 56622
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.8048750758171082,
      "learning_rate": 0.0005149412058684462,
      "loss": 2.3825,
      "step": 56623
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7112104296684265,
      "learning_rate": 0.0005149383522019424,
      "loss": 2.2392,
      "step": 56624
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7058923840522766,
      "learning_rate": 0.0005149354984954776,
      "loss": 2.3099,
      "step": 56625
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7374492287635803,
      "learning_rate": 0.0005149326447490523,
      "loss": 1.9909,
      "step": 56626
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.8045225739479065,
      "learning_rate": 0.0005149297909626668,
      "loss": 2.0623,
      "step": 56627
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7168170213699341,
      "learning_rate": 0.000514926937136322,
      "loss": 2.2692,
      "step": 56628
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7197460532188416,
      "learning_rate": 0.000514924083270018,
      "loss": 2.2782,
      "step": 56629
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7869307398796082,
      "learning_rate": 0.0005149212293637556,
      "loss": 2.0981,
      "step": 56630
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.6658439636230469,
      "learning_rate": 0.0005149183754175353,
      "loss": 2.1522,
      "step": 56631
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.775550127029419,
      "learning_rate": 0.0005149155214313577,
      "loss": 2.1547,
      "step": 56632
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.6997226476669312,
      "learning_rate": 0.0005149126674052233,
      "loss": 2.0442,
      "step": 56633
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7834094166755676,
      "learning_rate": 0.0005149098133391324,
      "loss": 2.2698,
      "step": 56634
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7515313625335693,
      "learning_rate": 0.0005149069592330858,
      "loss": 2.2392,
      "step": 56635
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.6912222504615784,
      "learning_rate": 0.000514904105087084,
      "loss": 2.3579,
      "step": 56636
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7268518209457397,
      "learning_rate": 0.0005149012509011274,
      "loss": 2.141,
      "step": 56637
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7728869318962097,
      "learning_rate": 0.0005148983966752165,
      "loss": 2.0313,
      "step": 56638
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.8294538259506226,
      "learning_rate": 0.000514895542409352,
      "loss": 2.548,
      "step": 56639
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7961023449897766,
      "learning_rate": 0.0005148926881035345,
      "loss": 2.2542,
      "step": 56640
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7053316831588745,
      "learning_rate": 0.0005148898337577643,
      "loss": 2.3754,
      "step": 56641
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.8050388097763062,
      "learning_rate": 0.000514886979372042,
      "loss": 2.146,
      "step": 56642
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.753131628036499,
      "learning_rate": 0.0005148841249463682,
      "loss": 2.4201,
      "step": 56643
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7725518941879272,
      "learning_rate": 0.0005148812704807434,
      "loss": 2.1921,
      "step": 56644
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.8433237075805664,
      "learning_rate": 0.000514878415975168,
      "loss": 2.2715,
      "step": 56645
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.8503816723823547,
      "learning_rate": 0.0005148755614296427,
      "loss": 2.5494,
      "step": 56646
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.6878555417060852,
      "learning_rate": 0.000514872706844168,
      "loss": 2.1741,
      "step": 56647
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.8636438846588135,
      "learning_rate": 0.0005148698522187444,
      "loss": 2.0512,
      "step": 56648
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7572481632232666,
      "learning_rate": 0.0005148669975533725,
      "loss": 2.2274,
      "step": 56649
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7430803775787354,
      "learning_rate": 0.0005148641428480527,
      "loss": 2.2556,
      "step": 56650
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.89152991771698,
      "learning_rate": 0.0005148612881027856,
      "loss": 2.2204,
      "step": 56651
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7118996381759644,
      "learning_rate": 0.0005148584333175718,
      "loss": 2.2663,
      "step": 56652
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7055169343948364,
      "learning_rate": 0.0005148555784924117,
      "loss": 2.2088,
      "step": 56653
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.9106763601303101,
      "learning_rate": 0.0005148527236273059,
      "loss": 2.1812,
      "step": 56654
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7476341724395752,
      "learning_rate": 0.0005148498687222549,
      "loss": 2.1128,
      "step": 56655
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.8089323043823242,
      "learning_rate": 0.0005148470137772593,
      "loss": 2.3246,
      "step": 56656
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.9154690504074097,
      "learning_rate": 0.0005148441587923194,
      "loss": 1.9389,
      "step": 56657
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.9262722730636597,
      "learning_rate": 0.0005148413037674362,
      "loss": 2.4069,
      "step": 56658
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.6992879509925842,
      "learning_rate": 0.0005148384487026098,
      "loss": 2.2737,
      "step": 56659
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7173580527305603,
      "learning_rate": 0.0005148355935978409,
      "loss": 2.0456,
      "step": 56660
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.8121770620346069,
      "learning_rate": 0.0005148327384531299,
      "loss": 2.3757,
      "step": 56661
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7459225058555603,
      "learning_rate": 0.0005148298832684776,
      "loss": 2.3334,
      "step": 56662
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7479652166366577,
      "learning_rate": 0.0005148270280438843,
      "loss": 2.3338,
      "step": 56663
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.6680721640586853,
      "learning_rate": 0.0005148241727793507,
      "loss": 2.2251,
      "step": 56664
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7396623492240906,
      "learning_rate": 0.0005148213174748771,
      "loss": 2.2095,
      "step": 56665
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7703610062599182,
      "learning_rate": 0.0005148184621304641,
      "loss": 2.1931,
      "step": 56666
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.6236504912376404,
      "learning_rate": 0.0005148156067461125,
      "loss": 2.072,
      "step": 56667
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.8216573596000671,
      "learning_rate": 0.0005148127513218225,
      "loss": 2.491,
      "step": 56668
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.8256518244743347,
      "learning_rate": 0.0005148098958575948,
      "loss": 2.2009,
      "step": 56669
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.800662100315094,
      "learning_rate": 0.0005148070403534298,
      "loss": 2.4852,
      "step": 56670
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7199990749359131,
      "learning_rate": 0.0005148041848093282,
      "loss": 2.3282,
      "step": 56671
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.875681459903717,
      "learning_rate": 0.0005148013292252904,
      "loss": 2.0103,
      "step": 56672
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7200422883033752,
      "learning_rate": 0.000514798473601317,
      "loss": 2.3713,
      "step": 56673
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7766270041465759,
      "learning_rate": 0.0005147956179374086,
      "loss": 2.3169,
      "step": 56674
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7501373291015625,
      "learning_rate": 0.0005147927622335655,
      "loss": 2.4419,
      "step": 56675
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7012357115745544,
      "learning_rate": 0.0005147899064897885,
      "loss": 2.033,
      "step": 56676
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7645871639251709,
      "learning_rate": 0.0005147870507060778,
      "loss": 2.1911,
      "step": 56677
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7395753264427185,
      "learning_rate": 0.0005147841948824343,
      "loss": 2.2461,
      "step": 56678
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7440798878669739,
      "learning_rate": 0.0005147813390188583,
      "loss": 2.3491,
      "step": 56679
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.745171070098877,
      "learning_rate": 0.0005147784831153504,
      "loss": 2.3297,
      "step": 56680
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7425584197044373,
      "learning_rate": 0.0005147756271719112,
      "loss": 2.2819,
      "step": 56681
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7393872737884521,
      "learning_rate": 0.0005147727711885411,
      "loss": 2.1303,
      "step": 56682
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.6950517296791077,
      "learning_rate": 0.0005147699151652407,
      "loss": 2.2068,
      "step": 56683
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7363806962966919,
      "learning_rate": 0.0005147670591020105,
      "loss": 2.1446,
      "step": 56684
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7409403324127197,
      "learning_rate": 0.000514764202998851,
      "loss": 2.2322,
      "step": 56685
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.8434657454490662,
      "learning_rate": 0.0005147613468557628,
      "loss": 2.2499,
      "step": 56686
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7665768265724182,
      "learning_rate": 0.0005147584906727466,
      "loss": 2.2019,
      "step": 56687
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7358781695365906,
      "learning_rate": 0.0005147556344498026,
      "loss": 2.3736,
      "step": 56688
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7001875042915344,
      "learning_rate": 0.0005147527781869314,
      "loss": 2.0024,
      "step": 56689
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7564303874969482,
      "learning_rate": 0.0005147499218841336,
      "loss": 2.2503,
      "step": 56690
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.6770954728126526,
      "learning_rate": 0.0005147470655414098,
      "loss": 2.2432,
      "step": 56691
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7260598540306091,
      "learning_rate": 0.0005147442091587604,
      "loss": 2.1516,
      "step": 56692
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7263907194137573,
      "learning_rate": 0.0005147413527361862,
      "loss": 2.178,
      "step": 56693
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7607594132423401,
      "learning_rate": 0.0005147384962736874,
      "loss": 2.1388,
      "step": 56694
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.6724967360496521,
      "learning_rate": 0.0005147356397712646,
      "loss": 2.2479,
      "step": 56695
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7079769372940063,
      "learning_rate": 0.0005147327832289185,
      "loss": 2.128,
      "step": 56696
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.6868159770965576,
      "learning_rate": 0.0005147299266466495,
      "loss": 2.3163,
      "step": 56697
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.6747609972953796,
      "learning_rate": 0.0005147270700244581,
      "loss": 2.3274,
      "step": 56698
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7127411961555481,
      "learning_rate": 0.0005147242133623449,
      "loss": 2.2367,
      "step": 56699
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7453534603118896,
      "learning_rate": 0.0005147213566603103,
      "loss": 2.5681,
      "step": 56700
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.719606339931488,
      "learning_rate": 0.0005147184999183552,
      "loss": 2.3107,
      "step": 56701
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7217041254043579,
      "learning_rate": 0.0005147156431364798,
      "loss": 2.213,
      "step": 56702
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7266810536384583,
      "learning_rate": 0.0005147127863146845,
      "loss": 2.2124,
      "step": 56703
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.8342143893241882,
      "learning_rate": 0.0005147099294529702,
      "loss": 2.2055,
      "step": 56704
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7184691429138184,
      "learning_rate": 0.0005147070725513374,
      "loss": 2.3384,
      "step": 56705
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7705760598182678,
      "learning_rate": 0.0005147042156097864,
      "loss": 2.213,
      "step": 56706
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7508679628372192,
      "learning_rate": 0.0005147013586283177,
      "loss": 2.0539,
      "step": 56707
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7206370234489441,
      "learning_rate": 0.0005146985016069322,
      "loss": 2.2546,
      "step": 56708
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.8024886250495911,
      "learning_rate": 0.00051469564454563,
      "loss": 2.3116,
      "step": 56709
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7752678394317627,
      "learning_rate": 0.000514692787444412,
      "loss": 2.2893,
      "step": 56710
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7449086308479309,
      "learning_rate": 0.0005146899303032783,
      "loss": 2.2023,
      "step": 56711
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.6784533262252808,
      "learning_rate": 0.0005146870731222298,
      "loss": 2.3258,
      "step": 56712
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7874771952629089,
      "learning_rate": 0.0005146842159012671,
      "loss": 2.0892,
      "step": 56713
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7758650779724121,
      "learning_rate": 0.0005146813586403904,
      "loss": 2.3227,
      "step": 56714
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7745761871337891,
      "learning_rate": 0.0005146785013396003,
      "loss": 2.2995,
      "step": 56715
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7500249743461609,
      "learning_rate": 0.0005146756439988975,
      "loss": 2.4108,
      "step": 56716
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.6570736765861511,
      "learning_rate": 0.0005146727866182825,
      "loss": 2.2169,
      "step": 56717
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7464078664779663,
      "learning_rate": 0.0005146699291977558,
      "loss": 2.5227,
      "step": 56718
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.747075080871582,
      "learning_rate": 0.0005146670717373178,
      "loss": 2.3969,
      "step": 56719
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7934724688529968,
      "learning_rate": 0.0005146642142369692,
      "loss": 2.3422,
      "step": 56720
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.6970674991607666,
      "learning_rate": 0.0005146613566967104,
      "loss": 2.4775,
      "step": 56721
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.8439986109733582,
      "learning_rate": 0.000514658499116542,
      "loss": 2.1166,
      "step": 56722
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7357389330863953,
      "learning_rate": 0.0005146556414964646,
      "loss": 2.3288,
      "step": 56723
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.9339353442192078,
      "learning_rate": 0.0005146527838364787,
      "loss": 2.3868,
      "step": 56724
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7342865467071533,
      "learning_rate": 0.0005146499261365847,
      "loss": 2.232,
      "step": 56725
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.8165082931518555,
      "learning_rate": 0.0005146470683967833,
      "loss": 2.0506,
      "step": 56726
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7682172656059265,
      "learning_rate": 0.0005146442106170749,
      "loss": 2.2434,
      "step": 56727
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7647821307182312,
      "learning_rate": 0.0005146413527974601,
      "loss": 2.3263,
      "step": 56728
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7516542077064514,
      "learning_rate": 0.0005146384949379395,
      "loss": 2.1998,
      "step": 56729
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.719840407371521,
      "learning_rate": 0.0005146356370385135,
      "loss": 2.2563,
      "step": 56730
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.8203533291816711,
      "learning_rate": 0.0005146327790991826,
      "loss": 2.3841,
      "step": 56731
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7815080881118774,
      "learning_rate": 0.0005146299211199476,
      "loss": 2.4073,
      "step": 56732
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.676585853099823,
      "learning_rate": 0.0005146270631008087,
      "loss": 2.3486,
      "step": 56733
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7806116342544556,
      "learning_rate": 0.0005146242050417665,
      "loss": 2.3742,
      "step": 56734
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.783420741558075,
      "learning_rate": 0.0005146213469428218,
      "loss": 2.1004,
      "step": 56735
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.9903677105903625,
      "learning_rate": 0.0005146184888039749,
      "loss": 2.3347,
      "step": 56736
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7273252010345459,
      "learning_rate": 0.0005146156306252263,
      "loss": 2.1606,
      "step": 56737
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.6955158710479736,
      "learning_rate": 0.0005146127724065766,
      "loss": 2.1693,
      "step": 56738
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7298281192779541,
      "learning_rate": 0.0005146099141480264,
      "loss": 2.224,
      "step": 56739
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7635104656219482,
      "learning_rate": 0.0005146070558495762,
      "loss": 2.0763,
      "step": 56740
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7298822402954102,
      "learning_rate": 0.0005146041975112263,
      "loss": 2.1418,
      "step": 56741
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7700098752975464,
      "learning_rate": 0.0005146013391329777,
      "loss": 2.4506,
      "step": 56742
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.8258326053619385,
      "learning_rate": 0.0005145984807148305,
      "loss": 2.1649,
      "step": 56743
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.8102141618728638,
      "learning_rate": 0.0005145956222567854,
      "loss": 2.1646,
      "step": 56744
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.794653594493866,
      "learning_rate": 0.000514592763758843,
      "loss": 2.367,
      "step": 56745
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7319318652153015,
      "learning_rate": 0.0005145899052210036,
      "loss": 2.2966,
      "step": 56746
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.8343286514282227,
      "learning_rate": 0.000514587046643268,
      "loss": 2.3377,
      "step": 56747
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7763465642929077,
      "learning_rate": 0.0005145841880256366,
      "loss": 2.1363,
      "step": 56748
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.734036386013031,
      "learning_rate": 0.00051458132936811,
      "loss": 2.1092,
      "step": 56749
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.759985089302063,
      "learning_rate": 0.0005145784706706885,
      "loss": 2.135,
      "step": 56750
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.8006840944290161,
      "learning_rate": 0.0005145756119333731,
      "loss": 2.44,
      "step": 56751
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.8112212419509888,
      "learning_rate": 0.0005145727531561638,
      "loss": 2.2656,
      "step": 56752
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7905628681182861,
      "learning_rate": 0.0005145698943390615,
      "loss": 2.3739,
      "step": 56753
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7345477938652039,
      "learning_rate": 0.0005145670354820666,
      "loss": 2.3102,
      "step": 56754
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7047467827796936,
      "learning_rate": 0.0005145641765851795,
      "loss": 2.3841,
      "step": 56755
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.6967777013778687,
      "learning_rate": 0.0005145613176484009,
      "loss": 2.1665,
      "step": 56756
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.8041479587554932,
      "learning_rate": 0.0005145584586717315,
      "loss": 2.1939,
      "step": 56757
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.868548572063446,
      "learning_rate": 0.0005145555996551714,
      "loss": 2.3692,
      "step": 56758
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.8230968117713928,
      "learning_rate": 0.0005145527405987215,
      "loss": 2.1631,
      "step": 56759
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.708899736404419,
      "learning_rate": 0.0005145498815023822,
      "loss": 2.1653,
      "step": 56760
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.8401138186454773,
      "learning_rate": 0.0005145470223661539,
      "loss": 2.2073,
      "step": 56761
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7464633584022522,
      "learning_rate": 0.0005145441631900375,
      "loss": 2.5085,
      "step": 56762
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7214970588684082,
      "learning_rate": 0.0005145413039740331,
      "loss": 2.6418,
      "step": 56763
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.8059448003768921,
      "learning_rate": 0.0005145384447181414,
      "loss": 2.3004,
      "step": 56764
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.8182259798049927,
      "learning_rate": 0.000514535585422363,
      "loss": 2.2602,
      "step": 56765
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7442341446876526,
      "learning_rate": 0.0005145327260866985,
      "loss": 2.2113,
      "step": 56766
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.6919885873794556,
      "learning_rate": 0.0005145298667111482,
      "loss": 2.234,
      "step": 56767
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.8223006129264832,
      "learning_rate": 0.0005145270072957127,
      "loss": 2.0344,
      "step": 56768
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7518969774246216,
      "learning_rate": 0.0005145241478403926,
      "loss": 2.2869,
      "step": 56769
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7709943056106567,
      "learning_rate": 0.0005145212883451885,
      "loss": 2.3951,
      "step": 56770
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.9489776492118835,
      "learning_rate": 0.0005145184288101008,
      "loss": 2.1992,
      "step": 56771
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.8456485867500305,
      "learning_rate": 0.0005145155692351301,
      "loss": 2.1594,
      "step": 56772
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.727384090423584,
      "learning_rate": 0.0005145127096202769,
      "loss": 2.2032,
      "step": 56773
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7969773411750793,
      "learning_rate": 0.0005145098499655417,
      "loss": 2.2247,
      "step": 56774
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.8301055431365967,
      "learning_rate": 0.0005145069902709251,
      "loss": 2.3201,
      "step": 56775
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7380759119987488,
      "learning_rate": 0.0005145041305364276,
      "loss": 2.2563,
      "step": 56776
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.6948487758636475,
      "learning_rate": 0.0005145012707620497,
      "loss": 2.2823,
      "step": 56777
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.8540332317352295,
      "learning_rate": 0.0005144984109477919,
      "loss": 2.3604,
      "step": 56778
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7852255702018738,
      "learning_rate": 0.000514495551093655,
      "loss": 2.3326,
      "step": 56779
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.8423315286636353,
      "learning_rate": 0.0005144926911996392,
      "loss": 2.1138,
      "step": 56780
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7516928315162659,
      "learning_rate": 0.0005144898312657452,
      "loss": 2.3328,
      "step": 56781
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.0683097839355469,
      "learning_rate": 0.0005144869712919734,
      "loss": 2.3805,
      "step": 56782
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7344973683357239,
      "learning_rate": 0.0005144841112783245,
      "loss": 2.141,
      "step": 56783
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.9186230301856995,
      "learning_rate": 0.0005144812512247989,
      "loss": 2.3267,
      "step": 56784
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7648234963417053,
      "learning_rate": 0.0005144783911313972,
      "loss": 2.4388,
      "step": 56785
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7153318524360657,
      "learning_rate": 0.0005144755309981201,
      "loss": 2.1512,
      "step": 56786
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7623102068901062,
      "learning_rate": 0.0005144726708249677,
      "loss": 2.4279,
      "step": 56787
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7468252778053284,
      "learning_rate": 0.0005144698106119409,
      "loss": 2.2731,
      "step": 56788
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7387685775756836,
      "learning_rate": 0.00051446695035904,
      "loss": 2.1094,
      "step": 56789
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.8434762358665466,
      "learning_rate": 0.0005144640900662658,
      "loss": 2.221,
      "step": 56790
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.677888810634613,
      "learning_rate": 0.0005144612297336185,
      "loss": 2.1571,
      "step": 56791
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7538890838623047,
      "learning_rate": 0.0005144583693610991,
      "loss": 2.3467,
      "step": 56792
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.767418384552002,
      "learning_rate": 0.0005144555089487076,
      "loss": 2.4599,
      "step": 56793
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7021453976631165,
      "learning_rate": 0.0005144526484964448,
      "loss": 2.3733,
      "step": 56794
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7490975260734558,
      "learning_rate": 0.0005144497880043113,
      "loss": 1.8976,
      "step": 56795
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.8931450247764587,
      "learning_rate": 0.0005144469274723074,
      "loss": 2.3895,
      "step": 56796
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7773585915565491,
      "learning_rate": 0.000514444066900434,
      "loss": 2.5785,
      "step": 56797
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7331593632698059,
      "learning_rate": 0.0005144412062886911,
      "loss": 2.3042,
      "step": 56798
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.8473067879676819,
      "learning_rate": 0.0005144383456370797,
      "loss": 2.4539,
      "step": 56799
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7259617447853088,
      "learning_rate": 0.0005144354849456002,
      "loss": 2.1488,
      "step": 56800
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.746623158454895,
      "learning_rate": 0.000514432624214253,
      "loss": 2.2626,
      "step": 56801
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7241357564926147,
      "learning_rate": 0.0005144297634430388,
      "loss": 2.2464,
      "step": 56802
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.8370757102966309,
      "learning_rate": 0.000514426902631958,
      "loss": 2.1066,
      "step": 56803
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.9300305843353271,
      "learning_rate": 0.0005144240417810112,
      "loss": 2.423,
      "step": 56804
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7504752278327942,
      "learning_rate": 0.0005144211808901989,
      "loss": 2.3634,
      "step": 56805
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.8232890963554382,
      "learning_rate": 0.0005144183199595218,
      "loss": 2.1069,
      "step": 56806
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.986371636390686,
      "learning_rate": 0.0005144154589889801,
      "loss": 2.4065,
      "step": 56807
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7696058750152588,
      "learning_rate": 0.0005144125979785745,
      "loss": 2.1461,
      "step": 56808
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.6624923944473267,
      "learning_rate": 0.0005144097369283058,
      "loss": 2.1519,
      "step": 56809
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7808576822280884,
      "learning_rate": 0.000514406875838174,
      "loss": 2.2473,
      "step": 56810
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7496458292007446,
      "learning_rate": 0.0005144040147081801,
      "loss": 2.1043,
      "step": 56811
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7493090033531189,
      "learning_rate": 0.0005144011535383244,
      "loss": 2.1622,
      "step": 56812
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.738423228263855,
      "learning_rate": 0.0005143982923286075,
      "loss": 2.2558,
      "step": 56813
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7779639959335327,
      "learning_rate": 0.0005143954310790299,
      "loss": 2.2196,
      "step": 56814
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7680874466896057,
      "learning_rate": 0.0005143925697895922,
      "loss": 2.1981,
      "step": 56815
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7116042375564575,
      "learning_rate": 0.0005143897084602946,
      "loss": 2.2707,
      "step": 56816
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7755042314529419,
      "learning_rate": 0.0005143868470911382,
      "loss": 2.3021,
      "step": 56817
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7296085953712463,
      "learning_rate": 0.0005143839856821233,
      "loss": 2.217,
      "step": 56818
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.8269388675689697,
      "learning_rate": 0.00051438112423325,
      "loss": 2.4767,
      "step": 56819
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.670630693435669,
      "learning_rate": 0.0005143782627445195,
      "loss": 2.1337,
      "step": 56820
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7950296998023987,
      "learning_rate": 0.000514375401215932,
      "loss": 2.2219,
      "step": 56821
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7626286149024963,
      "learning_rate": 0.000514372539647488,
      "loss": 2.2488,
      "step": 56822
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.8148542642593384,
      "learning_rate": 0.000514369678039188,
      "loss": 2.3019,
      "step": 56823
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7198392748832703,
      "learning_rate": 0.0005143668163910328,
      "loss": 2.2636,
      "step": 56824
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7056032419204712,
      "learning_rate": 0.0005143639547030227,
      "loss": 2.3488,
      "step": 56825
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.6819156408309937,
      "learning_rate": 0.0005143610929751584,
      "loss": 2.0049,
      "step": 56826
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.693697452545166,
      "learning_rate": 0.0005143582312074402,
      "loss": 2.3889,
      "step": 56827
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.8042472004890442,
      "learning_rate": 0.0005143553693998687,
      "loss": 2.174,
      "step": 56828
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7285299301147461,
      "learning_rate": 0.0005143525075524447,
      "loss": 2.32,
      "step": 56829
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7129883766174316,
      "learning_rate": 0.0005143496456651683,
      "loss": 2.186,
      "step": 56830
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7401607036590576,
      "learning_rate": 0.0005143467837380403,
      "loss": 2.0111,
      "step": 56831
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7239598035812378,
      "learning_rate": 0.0005143439217710613,
      "loss": 1.9055,
      "step": 56832
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.8460875153541565,
      "learning_rate": 0.0005143410597642316,
      "loss": 2.3685,
      "step": 56833
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.6903665065765381,
      "learning_rate": 0.000514338197717552,
      "loss": 2.1926,
      "step": 56834
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7352928519248962,
      "learning_rate": 0.0005143353356310227,
      "loss": 2.2759,
      "step": 56835
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.8074942827224731,
      "learning_rate": 0.0005143324735046446,
      "loss": 2.2582,
      "step": 56836
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7616849541664124,
      "learning_rate": 0.0005143296113384179,
      "loss": 2.2284,
      "step": 56837
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7338439226150513,
      "learning_rate": 0.0005143267491323434,
      "loss": 2.2761,
      "step": 56838
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.8749496340751648,
      "learning_rate": 0.0005143238868864214,
      "loss": 2.2447,
      "step": 56839
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.8697088360786438,
      "learning_rate": 0.0005143210246006527,
      "loss": 2.2506,
      "step": 56840
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.6946241855621338,
      "learning_rate": 0.0005143181622750374,
      "loss": 2.3096,
      "step": 56841
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.8241671323776245,
      "learning_rate": 0.0005143152999095766,
      "loss": 2.1201,
      "step": 56842
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.8062452673912048,
      "learning_rate": 0.0005143124375042704,
      "loss": 2.3554,
      "step": 56843
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7584013342857361,
      "learning_rate": 0.0005143095750591195,
      "loss": 2.1914,
      "step": 56844
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.6960092186927795,
      "learning_rate": 0.0005143067125741243,
      "loss": 2.266,
      "step": 56845
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7523418664932251,
      "learning_rate": 0.0005143038500492855,
      "loss": 2.1855,
      "step": 56846
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7812480926513672,
      "learning_rate": 0.0005143009874846037,
      "loss": 2.2471,
      "step": 56847
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7421939969062805,
      "learning_rate": 0.0005142981248800791,
      "loss": 2.2942,
      "step": 56848
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7652735114097595,
      "learning_rate": 0.0005142952622357126,
      "loss": 2.5194,
      "step": 56849
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7466161847114563,
      "learning_rate": 0.0005142923995515045,
      "loss": 2.229,
      "step": 56850
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.76719731092453,
      "learning_rate": 0.0005142895368274554,
      "loss": 2.245,
      "step": 56851
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.8124356865882874,
      "learning_rate": 0.0005142866740635658,
      "loss": 2.2305,
      "step": 56852
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.692459225654602,
      "learning_rate": 0.0005142838112598364,
      "loss": 2.2669,
      "step": 56853
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7714557647705078,
      "learning_rate": 0.0005142809484162674,
      "loss": 2.286,
      "step": 56854
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.740925133228302,
      "learning_rate": 0.0005142780855328596,
      "loss": 2.4298,
      "step": 56855
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7266820669174194,
      "learning_rate": 0.0005142752226096135,
      "loss": 2.1717,
      "step": 56856
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7388259768486023,
      "learning_rate": 0.0005142723596465295,
      "loss": 2.2407,
      "step": 56857
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.6708469390869141,
      "learning_rate": 0.0005142694966436083,
      "loss": 2.451,
      "step": 56858
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.777389645576477,
      "learning_rate": 0.0005142666336008504,
      "loss": 2.3039,
      "step": 56859
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7464303374290466,
      "learning_rate": 0.0005142637705182562,
      "loss": 2.1675,
      "step": 56860
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.6980476379394531,
      "learning_rate": 0.0005142609073958263,
      "loss": 2.1823,
      "step": 56861
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7920401096343994,
      "learning_rate": 0.0005142580442335613,
      "loss": 2.2862,
      "step": 56862
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7320666313171387,
      "learning_rate": 0.0005142551810314618,
      "loss": 2.1617,
      "step": 56863
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7069551944732666,
      "learning_rate": 0.0005142523177895281,
      "loss": 2.3933,
      "step": 56864
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.8064910769462585,
      "learning_rate": 0.0005142494545077608,
      "loss": 2.2216,
      "step": 56865
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7258368730545044,
      "learning_rate": 0.0005142465911861606,
      "loss": 2.1467,
      "step": 56866
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7445983290672302,
      "learning_rate": 0.0005142437278247277,
      "loss": 2.0764,
      "step": 56867
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7254481911659241,
      "learning_rate": 0.000514240864423463,
      "loss": 2.1962,
      "step": 56868
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7186630964279175,
      "learning_rate": 0.0005142380009823669,
      "loss": 2.1715,
      "step": 56869
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7686870098114014,
      "learning_rate": 0.00051423513750144,
      "loss": 2.4297,
      "step": 56870
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.767097532749176,
      "learning_rate": 0.0005142322739806825,
      "loss": 2.3453,
      "step": 56871
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7620824575424194,
      "learning_rate": 0.0005142294104200954,
      "loss": 2.2201,
      "step": 56872
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.809249222278595,
      "learning_rate": 0.0005142265468196788,
      "loss": 2.3326,
      "step": 56873
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.8471565842628479,
      "learning_rate": 0.0005142236831794337,
      "loss": 2.2792,
      "step": 56874
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.6970226168632507,
      "learning_rate": 0.0005142208194993602,
      "loss": 2.2991,
      "step": 56875
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7936578989028931,
      "learning_rate": 0.000514217955779459,
      "loss": 2.4098,
      "step": 56876
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.8610555529594421,
      "learning_rate": 0.0005142150920197306,
      "loss": 2.1609,
      "step": 56877
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7809798121452332,
      "learning_rate": 0.0005142122282201757,
      "loss": 2.1704,
      "step": 56878
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7207661271095276,
      "learning_rate": 0.0005142093643807947,
      "loss": 2.1346,
      "step": 56879
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.8441792726516724,
      "learning_rate": 0.0005142065005015881,
      "loss": 2.442,
      "step": 56880
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.805446445941925,
      "learning_rate": 0.0005142036365825564,
      "loss": 2.0552,
      "step": 56881
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.6838821172714233,
      "learning_rate": 0.0005142007726237003,
      "loss": 2.3095,
      "step": 56882
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.6986653804779053,
      "learning_rate": 0.0005141979086250203,
      "loss": 2.0721,
      "step": 56883
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7265920639038086,
      "learning_rate": 0.0005141950445865166,
      "loss": 2.2615,
      "step": 56884
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7277514934539795,
      "learning_rate": 0.0005141921805081901,
      "loss": 2.2789,
      "step": 56885
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7067639827728271,
      "learning_rate": 0.0005141893163900414,
      "loss": 2.2001,
      "step": 56886
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.739121675491333,
      "learning_rate": 0.0005141864522320707,
      "loss": 2.3421,
      "step": 56887
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7043637037277222,
      "learning_rate": 0.0005141835880342788,
      "loss": 2.3267,
      "step": 56888
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7360632419586182,
      "learning_rate": 0.000514180723796666,
      "loss": 2.0569,
      "step": 56889
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.896144449710846,
      "learning_rate": 0.000514177859519233,
      "loss": 2.1646,
      "step": 56890
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7786649465560913,
      "learning_rate": 0.0005141749952019804,
      "loss": 2.2292,
      "step": 56891
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7261350750923157,
      "learning_rate": 0.0005141721308449085,
      "loss": 2.1553,
      "step": 56892
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.9594128131866455,
      "learning_rate": 0.000514169266448018,
      "loss": 2.1613,
      "step": 56893
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7746391892433167,
      "learning_rate": 0.0005141664020113095,
      "loss": 2.2343,
      "step": 56894
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.8256729245185852,
      "learning_rate": 0.0005141635375347833,
      "loss": 2.2029,
      "step": 56895
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.8179919719696045,
      "learning_rate": 0.00051416067301844,
      "loss": 2.0366,
      "step": 56896
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.657970666885376,
      "learning_rate": 0.0005141578084622802,
      "loss": 2.0087,
      "step": 56897
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7686059474945068,
      "learning_rate": 0.0005141549438663045,
      "loss": 2.2454,
      "step": 56898
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7425439953804016,
      "learning_rate": 0.0005141520792305134,
      "loss": 2.4394,
      "step": 56899
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7634178400039673,
      "learning_rate": 0.0005141492145549074,
      "loss": 2.2661,
      "step": 56900
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7234202027320862,
      "learning_rate": 0.0005141463498394869,
      "loss": 1.9755,
      "step": 56901
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7763373851776123,
      "learning_rate": 0.0005141434850842525,
      "loss": 2.3167,
      "step": 56902
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7444620132446289,
      "learning_rate": 0.0005141406202892049,
      "loss": 2.2928,
      "step": 56903
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.8552548289299011,
      "learning_rate": 0.0005141377554543445,
      "loss": 2.3757,
      "step": 56904
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.702839732170105,
      "learning_rate": 0.0005141348905796718,
      "loss": 2.3177,
      "step": 56905
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7162693738937378,
      "learning_rate": 0.0005141320256651875,
      "loss": 2.2845,
      "step": 56906
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.742582380771637,
      "learning_rate": 0.000514129160710892,
      "loss": 2.2086,
      "step": 56907
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7529931664466858,
      "learning_rate": 0.0005141262957167858,
      "loss": 2.2421,
      "step": 56908
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7760730385780334,
      "learning_rate": 0.0005141234306828694,
      "loss": 2.3463,
      "step": 56909
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.666729211807251,
      "learning_rate": 0.0005141205656091435,
      "loss": 2.2126,
      "step": 56910
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.6865829229354858,
      "learning_rate": 0.0005141177004956085,
      "loss": 2.4488,
      "step": 56911
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.8149089217185974,
      "learning_rate": 0.000514114835342265,
      "loss": 2.2857,
      "step": 56912
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7247909903526306,
      "learning_rate": 0.0005141119701491134,
      "loss": 2.2413,
      "step": 56913
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7497259378433228,
      "learning_rate": 0.0005141091049161545,
      "loss": 2.582,
      "step": 56914
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7700268626213074,
      "learning_rate": 0.0005141062396433886,
      "loss": 2.3605,
      "step": 56915
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.794678807258606,
      "learning_rate": 0.0005141033743308164,
      "loss": 2.2331,
      "step": 56916
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7484913468360901,
      "learning_rate": 0.0005141005089784383,
      "loss": 2.3556,
      "step": 56917
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.768913984298706,
      "learning_rate": 0.0005140976435862549,
      "loss": 2.2242,
      "step": 56918
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.705118715763092,
      "learning_rate": 0.0005140947781542666,
      "loss": 2.1852,
      "step": 56919
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7078880667686462,
      "learning_rate": 0.0005140919126824741,
      "loss": 2.1056,
      "step": 56920
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7194433808326721,
      "learning_rate": 0.0005140890471708779,
      "loss": 2.3864,
      "step": 56921
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.757102370262146,
      "learning_rate": 0.0005140861816194784,
      "loss": 2.255,
      "step": 56922
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7440195679664612,
      "learning_rate": 0.0005140833160282763,
      "loss": 2.2528,
      "step": 56923
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.816501259803772,
      "learning_rate": 0.0005140804503972722,
      "loss": 2.1313,
      "step": 56924
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7029184699058533,
      "learning_rate": 0.0005140775847264663,
      "loss": 2.1905,
      "step": 56925
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7857922315597534,
      "learning_rate": 0.0005140747190158594,
      "loss": 2.6012,
      "step": 56926
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7337294816970825,
      "learning_rate": 0.0005140718532654521,
      "loss": 2.3573,
      "step": 56927
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7081987857818604,
      "learning_rate": 0.0005140689874752447,
      "loss": 2.1388,
      "step": 56928
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7139338850975037,
      "learning_rate": 0.0005140661216452377,
      "loss": 2.3157,
      "step": 56929
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.792507529258728,
      "learning_rate": 0.000514063255775432,
      "loss": 2.1309,
      "step": 56930
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.6966355443000793,
      "learning_rate": 0.0005140603898658278,
      "loss": 2.2177,
      "step": 56931
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.6570457220077515,
      "learning_rate": 0.0005140575239164256,
      "loss": 2.2883,
      "step": 56932
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7119642496109009,
      "learning_rate": 0.0005140546579272263,
      "loss": 2.3639,
      "step": 56933
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7971500754356384,
      "learning_rate": 0.0005140517918982301,
      "loss": 2.0273,
      "step": 56934
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7339476943016052,
      "learning_rate": 0.0005140489258294376,
      "loss": 2.2428,
      "step": 56935
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7678413391113281,
      "learning_rate": 0.0005140460597208494,
      "loss": 2.298,
      "step": 56936
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7447457909584045,
      "learning_rate": 0.0005140431935724659,
      "loss": 2.1094,
      "step": 56937
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.6954735517501831,
      "learning_rate": 0.0005140403273842879,
      "loss": 2.0422,
      "step": 56938
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7377680540084839,
      "learning_rate": 0.0005140374611563156,
      "loss": 1.9648,
      "step": 56939
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7974073886871338,
      "learning_rate": 0.0005140345948885498,
      "loss": 2.1511,
      "step": 56940
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7623920440673828,
      "learning_rate": 0.0005140317285809909,
      "loss": 2.247,
      "step": 56941
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7415142059326172,
      "learning_rate": 0.0005140288622336394,
      "loss": 2.1918,
      "step": 56942
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.8544723987579346,
      "learning_rate": 0.0005140259958464961,
      "loss": 2.3083,
      "step": 56943
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.712692141532898,
      "learning_rate": 0.0005140231294195612,
      "loss": 2.3447,
      "step": 56944
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7299003005027771,
      "learning_rate": 0.0005140202629528353,
      "loss": 2.4168,
      "step": 56945
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7373772263526917,
      "learning_rate": 0.0005140173964463191,
      "loss": 2.3932,
      "step": 56946
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7423146367073059,
      "learning_rate": 0.0005140145299000129,
      "loss": 2.1957,
      "step": 56947
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.8199437856674194,
      "learning_rate": 0.0005140116633139174,
      "loss": 2.2343,
      "step": 56948
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.6939660906791687,
      "learning_rate": 0.0005140087966880332,
      "loss": 2.2739,
      "step": 56949
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7070705890655518,
      "learning_rate": 0.0005140059300223606,
      "loss": 2.3549,
      "step": 56950
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7283248901367188,
      "learning_rate": 0.0005140030633169002,
      "loss": 2.2127,
      "step": 56951
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7369626760482788,
      "learning_rate": 0.0005140001965716528,
      "loss": 2.3252,
      "step": 56952
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7661924958229065,
      "learning_rate": 0.0005139973297866186,
      "loss": 2.3448,
      "step": 56953
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7090915441513062,
      "learning_rate": 0.0005139944629617983,
      "loss": 2.2448,
      "step": 56954
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7073804140090942,
      "learning_rate": 0.0005139915960971923,
      "loss": 2.1854,
      "step": 56955
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.6978369951248169,
      "learning_rate": 0.0005139887291928012,
      "loss": 2.2537,
      "step": 56956
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.8011669516563416,
      "learning_rate": 0.0005139858622486257,
      "loss": 2.2202,
      "step": 56957
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7394532561302185,
      "learning_rate": 0.0005139829952646661,
      "loss": 2.0245,
      "step": 56958
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.6912075281143188,
      "learning_rate": 0.0005139801282409231,
      "loss": 2.1253,
      "step": 56959
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.708189070224762,
      "learning_rate": 0.0005139772611773971,
      "loss": 2.3728,
      "step": 56960
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.8910727500915527,
      "learning_rate": 0.0005139743940740887,
      "loss": 2.1638,
      "step": 56961
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7342208623886108,
      "learning_rate": 0.0005139715269309984,
      "loss": 2.2996,
      "step": 56962
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7740029096603394,
      "learning_rate": 0.0005139686597481267,
      "loss": 2.3737,
      "step": 56963
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.8222871422767639,
      "learning_rate": 0.0005139657925254743,
      "loss": 2.2901,
      "step": 56964
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7397763133049011,
      "learning_rate": 0.0005139629252630416,
      "loss": 2.3349,
      "step": 56965
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7646554708480835,
      "learning_rate": 0.0005139600579608292,
      "loss": 1.971,
      "step": 56966
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7749019265174866,
      "learning_rate": 0.0005139571906188374,
      "loss": 2.3813,
      "step": 56967
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.8072549700737,
      "learning_rate": 0.0005139543232370671,
      "loss": 2.3347,
      "step": 56968
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7083317041397095,
      "learning_rate": 0.0005139514558155186,
      "loss": 2.3334,
      "step": 56969
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.6555625200271606,
      "learning_rate": 0.0005139485883541925,
      "loss": 2.3498,
      "step": 56970
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7304050326347351,
      "learning_rate": 0.0005139457208530893,
      "loss": 2.0025,
      "step": 56971
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.9193352460861206,
      "learning_rate": 0.0005139428533122095,
      "loss": 2.1418,
      "step": 56972
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7364318370819092,
      "learning_rate": 0.0005139399857315537,
      "loss": 2.2407,
      "step": 56973
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7217422127723694,
      "learning_rate": 0.0005139371181111225,
      "loss": 2.1276,
      "step": 56974
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.8497163653373718,
      "learning_rate": 0.0005139342504509163,
      "loss": 2.2904,
      "step": 56975
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.6765872240066528,
      "learning_rate": 0.0005139313827509357,
      "loss": 1.9417,
      "step": 56976
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7215206623077393,
      "learning_rate": 0.0005139285150111811,
      "loss": 2.2949,
      "step": 56977
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7639249563217163,
      "learning_rate": 0.0005139256472316533,
      "loss": 2.2722,
      "step": 56978
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7706352472305298,
      "learning_rate": 0.0005139227794123527,
      "loss": 2.0707,
      "step": 56979
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7370932698249817,
      "learning_rate": 0.0005139199115532796,
      "loss": 2.1819,
      "step": 56980
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7609000205993652,
      "learning_rate": 0.0005139170436544349,
      "loss": 2.0345,
      "step": 56981
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7445647716522217,
      "learning_rate": 0.000513914175715819,
      "loss": 2.4113,
      "step": 56982
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.9944449663162231,
      "learning_rate": 0.0005139113077374325,
      "loss": 2.1875,
      "step": 56983
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.8028838038444519,
      "learning_rate": 0.0005139084397192757,
      "loss": 2.0623,
      "step": 56984
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7990934252738953,
      "learning_rate": 0.0005139055716613494,
      "loss": 2.309,
      "step": 56985
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.779109001159668,
      "learning_rate": 0.0005139027035636538,
      "loss": 2.4246,
      "step": 56986
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.855816125869751,
      "learning_rate": 0.00051389983542619,
      "loss": 2.307,
      "step": 56987
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.762627363204956,
      "learning_rate": 0.0005138969672489579,
      "loss": 2.2879,
      "step": 56988
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.2228052616119385,
      "learning_rate": 0.0005138940990319584,
      "loss": 2.36,
      "step": 56989
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7256196737289429,
      "learning_rate": 0.000513891230775192,
      "loss": 2.1082,
      "step": 56990
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7634908556938171,
      "learning_rate": 0.0005138883624786592,
      "loss": 2.3274,
      "step": 56991
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.8027658462524414,
      "learning_rate": 0.0005138854941423603,
      "loss": 2.2785,
      "step": 56992
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.807622492313385,
      "learning_rate": 0.0005138826257662962,
      "loss": 2.1593,
      "step": 56993
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.9686126112937927,
      "learning_rate": 0.0005138797573504674,
      "loss": 2.435,
      "step": 56994
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7619448304176331,
      "learning_rate": 0.0005138768888948741,
      "loss": 2.2515,
      "step": 56995
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.9591139554977417,
      "learning_rate": 0.0005138740203995171,
      "loss": 2.2617,
      "step": 56996
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.1260020732879639,
      "learning_rate": 0.000513871151864397,
      "loss": 2.2885,
      "step": 56997
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7954820394515991,
      "learning_rate": 0.0005138682832895141,
      "loss": 2.3841,
      "step": 56998
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.9860255718231201,
      "learning_rate": 0.0005138654146748692,
      "loss": 2.3841,
      "step": 56999
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7906455993652344,
      "learning_rate": 0.0005138625460204625,
      "loss": 2.0405,
      "step": 57000
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7255746126174927,
      "learning_rate": 0.0005138596773262947,
      "loss": 2.2496,
      "step": 57001
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7983579635620117,
      "learning_rate": 0.0005138568085923665,
      "loss": 2.2092,
      "step": 57002
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.8224618434906006,
      "learning_rate": 0.0005138539398186783,
      "loss": 2.335,
      "step": 57003
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.6994317770004272,
      "learning_rate": 0.0005138510710052305,
      "loss": 2.2366,
      "step": 57004
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7685932517051697,
      "learning_rate": 0.0005138482021520239,
      "loss": 2.2925,
      "step": 57005
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.6562694907188416,
      "learning_rate": 0.0005138453332590587,
      "loss": 2.3006,
      "step": 57006
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7827444672584534,
      "learning_rate": 0.0005138424643263357,
      "loss": 2.2196,
      "step": 57007
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.8087174296379089,
      "learning_rate": 0.0005138395953538553,
      "loss": 2.4623,
      "step": 57008
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7074517011642456,
      "learning_rate": 0.0005138367263416181,
      "loss": 2.2123,
      "step": 57009
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7227318286895752,
      "learning_rate": 0.0005138338572896246,
      "loss": 2.2374,
      "step": 57010
    },
    {
      "epoch": 0.74,
      "grad_norm": 2.4155712127685547,
      "learning_rate": 0.0005138309881978755,
      "loss": 2.1015,
      "step": 57011
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.8147858381271362,
      "learning_rate": 0.0005138281190663709,
      "loss": 2.1727,
      "step": 57012
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7351046800613403,
      "learning_rate": 0.0005138252498951119,
      "loss": 2.0849,
      "step": 57013
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7328407168388367,
      "learning_rate": 0.0005138223806840987,
      "loss": 2.1414,
      "step": 57014
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.9105939865112305,
      "learning_rate": 0.0005138195114333317,
      "loss": 2.348,
      "step": 57015
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.6677981019020081,
      "learning_rate": 0.0005138166421428117,
      "loss": 2.2719,
      "step": 57016
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7922807931900024,
      "learning_rate": 0.0005138137728125392,
      "loss": 2.3735,
      "step": 57017
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.697532594203949,
      "learning_rate": 0.0005138109034425148,
      "loss": 2.2954,
      "step": 57018
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.6876577138900757,
      "learning_rate": 0.0005138080340327387,
      "loss": 2.3772,
      "step": 57019
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7264766693115234,
      "learning_rate": 0.0005138051645832117,
      "loss": 2.1966,
      "step": 57020
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.6460511684417725,
      "learning_rate": 0.0005138022950939344,
      "loss": 2.3942,
      "step": 57021
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7145769596099854,
      "learning_rate": 0.0005137994255649071,
      "loss": 2.3815,
      "step": 57022
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7467991709709167,
      "learning_rate": 0.0005137965559961304,
      "loss": 2.3209,
      "step": 57023
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7908705472946167,
      "learning_rate": 0.0005137936863876051,
      "loss": 2.3195,
      "step": 57024
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.6781173348426819,
      "learning_rate": 0.0005137908167393313,
      "loss": 2.2732,
      "step": 57025
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7273429036140442,
      "learning_rate": 0.0005137879470513098,
      "loss": 2.4185,
      "step": 57026
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.8017300367355347,
      "learning_rate": 0.0005137850773235412,
      "loss": 2.201,
      "step": 57027
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.6929352283477783,
      "learning_rate": 0.0005137822075560258,
      "loss": 2.3343,
      "step": 57028
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.783097505569458,
      "learning_rate": 0.0005137793377487643,
      "loss": 2.1538,
      "step": 57029
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7428818345069885,
      "learning_rate": 0.0005137764679017571,
      "loss": 2.4849,
      "step": 57030
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7244680523872375,
      "learning_rate": 0.000513773598015005,
      "loss": 2.3334,
      "step": 57031
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7300014495849609,
      "learning_rate": 0.0005137707280885083,
      "loss": 2.2206,
      "step": 57032
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7738208174705505,
      "learning_rate": 0.0005137678581222675,
      "loss": 2.2755,
      "step": 57033
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.6535670161247253,
      "learning_rate": 0.0005137649881162832,
      "loss": 2.2017,
      "step": 57034
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7092850804328918,
      "learning_rate": 0.0005137621180705561,
      "loss": 2.1807,
      "step": 57035
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7170151472091675,
      "learning_rate": 0.0005137592479850865,
      "loss": 2.2342,
      "step": 57036
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7179011702537537,
      "learning_rate": 0.000513756377859875,
      "loss": 2.3311,
      "step": 57037
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.794117271900177,
      "learning_rate": 0.0005137535076949222,
      "loss": 2.3426,
      "step": 57038
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7411352396011353,
      "learning_rate": 0.0005137506374902286,
      "loss": 2.1442,
      "step": 57039
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7504546642303467,
      "learning_rate": 0.0005137477672457946,
      "loss": 2.2606,
      "step": 57040
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7432098388671875,
      "learning_rate": 0.000513744896961621,
      "loss": 2.2802,
      "step": 57041
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7825791835784912,
      "learning_rate": 0.0005137420266377081,
      "loss": 2.071,
      "step": 57042
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.6852337121963501,
      "learning_rate": 0.0005137391562740567,
      "loss": 2.2454,
      "step": 57043
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7667875289916992,
      "learning_rate": 0.0005137362858706669,
      "loss": 2.2717,
      "step": 57044
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.6929007768630981,
      "learning_rate": 0.0005137334154275395,
      "loss": 2.1237,
      "step": 57045
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.6747187376022339,
      "learning_rate": 0.0005137305449446752,
      "loss": 2.2985,
      "step": 57046
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7584214806556702,
      "learning_rate": 0.0005137276744220742,
      "loss": 2.2855,
      "step": 57047
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.691088855266571,
      "learning_rate": 0.0005137248038597374,
      "loss": 2.4567,
      "step": 57048
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.6834338307380676,
      "learning_rate": 0.0005137219332576649,
      "loss": 2.3894,
      "step": 57049
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.8922113180160522,
      "learning_rate": 0.0005137190626158575,
      "loss": 2.1975,
      "step": 57050
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7876408100128174,
      "learning_rate": 0.0005137161919343159,
      "loss": 2.2926,
      "step": 57051
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7267473340034485,
      "learning_rate": 0.0005137133212130403,
      "loss": 2.331,
      "step": 57052
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.6854500770568848,
      "learning_rate": 0.0005137104504520312,
      "loss": 2.3615,
      "step": 57053
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.9018833041191101,
      "learning_rate": 0.0005137075796512895,
      "loss": 2.311,
      "step": 57054
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7015475630760193,
      "learning_rate": 0.0005137047088108154,
      "loss": 2.1011,
      "step": 57055
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7113285064697266,
      "learning_rate": 0.0005137018379306096,
      "loss": 2.0572,
      "step": 57056
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7236529588699341,
      "learning_rate": 0.0005136989670106726,
      "loss": 2.3087,
      "step": 57057
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7205333709716797,
      "learning_rate": 0.0005136960960510049,
      "loss": 2.3563,
      "step": 57058
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.6687728762626648,
      "learning_rate": 0.0005136932250516072,
      "loss": 2.0589,
      "step": 57059
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.745296061038971,
      "learning_rate": 0.0005136903540124798,
      "loss": 2.2894,
      "step": 57060
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7349367737770081,
      "learning_rate": 0.0005136874829336233,
      "loss": 2.1181,
      "step": 57061
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7544941902160645,
      "learning_rate": 0.0005136846118150382,
      "loss": 2.0814,
      "step": 57062
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.8480503559112549,
      "learning_rate": 0.0005136817406567253,
      "loss": 2.4017,
      "step": 57063
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7497159242630005,
      "learning_rate": 0.0005136788694586848,
      "loss": 2.2757,
      "step": 57064
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.773810863494873,
      "learning_rate": 0.0005136759982209174,
      "loss": 2.4004,
      "step": 57065
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7072787284851074,
      "learning_rate": 0.0005136731269434235,
      "loss": 2.2345,
      "step": 57066
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7286262512207031,
      "learning_rate": 0.0005136702556262038,
      "loss": 2.3239,
      "step": 57067
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7289312481880188,
      "learning_rate": 0.0005136673842692588,
      "loss": 2.1827,
      "step": 57068
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.8065093755722046,
      "learning_rate": 0.000513664512872589,
      "loss": 2.3363,
      "step": 57069
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7247793674468994,
      "learning_rate": 0.000513661641436195,
      "loss": 2.3955,
      "step": 57070
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.804675817489624,
      "learning_rate": 0.000513658769960077,
      "loss": 2.0582,
      "step": 57071
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7534400224685669,
      "learning_rate": 0.000513655898444236,
      "loss": 2.4223,
      "step": 57072
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7869423627853394,
      "learning_rate": 0.0005136530268886723,
      "loss": 2.3735,
      "step": 57073
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7436155676841736,
      "learning_rate": 0.0005136501552933865,
      "loss": 2.4537,
      "step": 57074
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7119061946868896,
      "learning_rate": 0.0005136472836583791,
      "loss": 2.128,
      "step": 57075
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7527254819869995,
      "learning_rate": 0.0005136444119836506,
      "loss": 2.2692,
      "step": 57076
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7767964601516724,
      "learning_rate": 0.0005136415402692017,
      "loss": 2.0121,
      "step": 57077
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7821857929229736,
      "learning_rate": 0.0005136386685150327,
      "loss": 2.1455,
      "step": 57078
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7467520833015442,
      "learning_rate": 0.0005136357967211442,
      "loss": 2.1848,
      "step": 57079
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.8176884651184082,
      "learning_rate": 0.0005136329248875368,
      "loss": 2.2636,
      "step": 57080
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.8415560722351074,
      "learning_rate": 0.0005136300530142109,
      "loss": 2.3412,
      "step": 57081
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.742897093296051,
      "learning_rate": 0.0005136271811011673,
      "loss": 2.1836,
      "step": 57082
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7551568150520325,
      "learning_rate": 0.0005136243091484064,
      "loss": 2.238,
      "step": 57083
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7731587290763855,
      "learning_rate": 0.0005136214371559287,
      "loss": 2.6152,
      "step": 57084
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.8905003666877747,
      "learning_rate": 0.0005136185651237346,
      "loss": 2.2449,
      "step": 57085
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.810402512550354,
      "learning_rate": 0.0005136156930518248,
      "loss": 2.2688,
      "step": 57086
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7742096185684204,
      "learning_rate": 0.0005136128209401999,
      "loss": 2.3379,
      "step": 57087
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.9870290756225586,
      "learning_rate": 0.0005136099487888604,
      "loss": 2.2263,
      "step": 57088
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.9653881788253784,
      "learning_rate": 0.0005136070765978067,
      "loss": 2.1947,
      "step": 57089
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7507514953613281,
      "learning_rate": 0.0005136042043670393,
      "loss": 2.2125,
      "step": 57090
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.8279395699501038,
      "learning_rate": 0.000513601332096559,
      "loss": 2.4103,
      "step": 57091
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.9299216270446777,
      "learning_rate": 0.0005135984597863662,
      "loss": 2.0773,
      "step": 57092
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.6968375444412231,
      "learning_rate": 0.0005135955874364613,
      "loss": 2.3217,
      "step": 57093
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.8291040062904358,
      "learning_rate": 0.000513592715046845,
      "loss": 2.1571,
      "step": 57094
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.9080004692077637,
      "learning_rate": 0.0005135898426175177,
      "loss": 2.4381,
      "step": 57095
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7557083368301392,
      "learning_rate": 0.0005135869701484801,
      "loss": 2.3549,
      "step": 57096
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7672808766365051,
      "learning_rate": 0.0005135840976397327,
      "loss": 2.1913,
      "step": 57097
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7714720368385315,
      "learning_rate": 0.000513581225091276,
      "loss": 2.2052,
      "step": 57098
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7876478433609009,
      "learning_rate": 0.0005135783525031104,
      "loss": 2.2707,
      "step": 57099
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7794026136398315,
      "learning_rate": 0.0005135754798752366,
      "loss": 1.9368,
      "step": 57100
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.6737972497940063,
      "learning_rate": 0.000513572607207655,
      "loss": 2.2436,
      "step": 57101
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7764225602149963,
      "learning_rate": 0.0005135697345003663,
      "loss": 2.0353,
      "step": 57102
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7096525430679321,
      "learning_rate": 0.0005135668617533709,
      "loss": 2.46,
      "step": 57103
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7659053802490234,
      "learning_rate": 0.0005135639889666695,
      "loss": 2.1004,
      "step": 57104
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.6977232098579407,
      "learning_rate": 0.0005135611161402625,
      "loss": 2.4618,
      "step": 57105
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7661656141281128,
      "learning_rate": 0.0005135582432741504,
      "loss": 2.208,
      "step": 57106
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7058485746383667,
      "learning_rate": 0.0005135553703683338,
      "loss": 2.0944,
      "step": 57107
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7106456756591797,
      "learning_rate": 0.0005135524974228134,
      "loss": 2.3236,
      "step": 57108
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7428218126296997,
      "learning_rate": 0.0005135496244375893,
      "loss": 2.1524,
      "step": 57109
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.741174578666687,
      "learning_rate": 0.0005135467514126624,
      "loss": 2.5374,
      "step": 57110
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.8955667614936829,
      "learning_rate": 0.0005135438783480332,
      "loss": 2.3815,
      "step": 57111
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.799501895904541,
      "learning_rate": 0.000513541005243702,
      "loss": 2.3195,
      "step": 57112
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7427352666854858,
      "learning_rate": 0.0005135381320996697,
      "loss": 2.3645,
      "step": 57113
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7677843570709229,
      "learning_rate": 0.0005135352589159364,
      "loss": 2.5016,
      "step": 57114
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7614054679870605,
      "learning_rate": 0.000513532385692503,
      "loss": 2.2364,
      "step": 57115
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7791430354118347,
      "learning_rate": 0.00051352951242937,
      "loss": 2.1225,
      "step": 57116
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7413150072097778,
      "learning_rate": 0.0005135266391265377,
      "loss": 2.1524,
      "step": 57117
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7670696973800659,
      "learning_rate": 0.0005135237657840067,
      "loss": 2.3242,
      "step": 57118
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.6988265514373779,
      "learning_rate": 0.0005135208924017777,
      "loss": 2.5729,
      "step": 57119
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.888201892375946,
      "learning_rate": 0.0005135180189798511,
      "loss": 2.1153,
      "step": 57120
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.8540945649147034,
      "learning_rate": 0.0005135151455182274,
      "loss": 2.2413,
      "step": 57121
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7591899037361145,
      "learning_rate": 0.0005135122720169074,
      "loss": 2.1282,
      "step": 57122
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.9158990383148193,
      "learning_rate": 0.0005135093984758913,
      "loss": 2.3812,
      "step": 57123
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7196890711784363,
      "learning_rate": 0.0005135065248951798,
      "loss": 2.1354,
      "step": 57124
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.8678288459777832,
      "learning_rate": 0.0005135036512747734,
      "loss": 2.1664,
      "step": 57125
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.9058196544647217,
      "learning_rate": 0.0005135007776146727,
      "loss": 2.1905,
      "step": 57126
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7942430973052979,
      "learning_rate": 0.0005134979039148782,
      "loss": 2.3236,
      "step": 57127
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.8642890453338623,
      "learning_rate": 0.0005134950301753903,
      "loss": 2.3641,
      "step": 57128
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.8977015018463135,
      "learning_rate": 0.0005134921563962096,
      "loss": 2.1458,
      "step": 57129
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7933945059776306,
      "learning_rate": 0.0005134892825773369,
      "loss": 2.3403,
      "step": 57130
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7680255174636841,
      "learning_rate": 0.0005134864087187722,
      "loss": 2.2728,
      "step": 57131
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.8504226207733154,
      "learning_rate": 0.0005134835348205166,
      "loss": 2.2777,
      "step": 57132
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.8484748005867004,
      "learning_rate": 0.0005134806608825703,
      "loss": 2.1183,
      "step": 57133
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7873201966285706,
      "learning_rate": 0.0005134777869049339,
      "loss": 2.2107,
      "step": 57134
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.06135892868042,
      "learning_rate": 0.0005134749128876079,
      "loss": 2.089,
      "step": 57135
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.729061484336853,
      "learning_rate": 0.000513472038830593,
      "loss": 2.3755,
      "step": 57136
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7101874351501465,
      "learning_rate": 0.0005134691647338895,
      "loss": 2.2065,
      "step": 57137
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7393773794174194,
      "learning_rate": 0.0005134662905974982,
      "loss": 2.2933,
      "step": 57138
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.1540974378585815,
      "learning_rate": 0.0005134634164214194,
      "loss": 2.0055,
      "step": 57139
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7681710124015808,
      "learning_rate": 0.0005134605422056536,
      "loss": 2.1323,
      "step": 57140
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.6846396327018738,
      "learning_rate": 0.0005134576679502017,
      "loss": 2.1628,
      "step": 57141
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.011947751045227,
      "learning_rate": 0.0005134547936550639,
      "loss": 2.3237,
      "step": 57142
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7611804008483887,
      "learning_rate": 0.0005134519193202406,
      "loss": 1.9838,
      "step": 57143
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7590283155441284,
      "learning_rate": 0.0005134490449457329,
      "loss": 2.3463,
      "step": 57144
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7414206862449646,
      "learning_rate": 0.0005134461705315408,
      "loss": 2.1878,
      "step": 57145
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.9149472117424011,
      "learning_rate": 0.0005134432960776651,
      "loss": 2.0405,
      "step": 57146
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.8162096738815308,
      "learning_rate": 0.0005134404215841063,
      "loss": 2.4022,
      "step": 57147
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.67730313539505,
      "learning_rate": 0.0005134375470508648,
      "loss": 2.2946,
      "step": 57148
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.73796546459198,
      "learning_rate": 0.0005134346724779413,
      "loss": 2.2103,
      "step": 57149
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7855197787284851,
      "learning_rate": 0.0005134317978653361,
      "loss": 2.3392,
      "step": 57150
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.839622974395752,
      "learning_rate": 0.0005134289232130502,
      "loss": 2.3749,
      "step": 57151
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7222370505332947,
      "learning_rate": 0.0005134260485210836,
      "loss": 2.1157,
      "step": 57152
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7091136574745178,
      "learning_rate": 0.0005134231737894372,
      "loss": 2.1855,
      "step": 57153
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.8162370920181274,
      "learning_rate": 0.0005134202990181113,
      "loss": 2.2135,
      "step": 57154
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.8534477353096008,
      "learning_rate": 0.0005134174242071066,
      "loss": 2.2403,
      "step": 57155
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7323683500289917,
      "learning_rate": 0.0005134145493564235,
      "loss": 2.1634,
      "step": 57156
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.9863606691360474,
      "learning_rate": 0.0005134116744660626,
      "loss": 1.9019,
      "step": 57157
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.8415795564651489,
      "learning_rate": 0.0005134087995360245,
      "loss": 2.3666,
      "step": 57158
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7498364448547363,
      "learning_rate": 0.0005134059245663097,
      "loss": 2.3828,
      "step": 57159
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7409819960594177,
      "learning_rate": 0.0005134030495569187,
      "loss": 2.216,
      "step": 57160
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7917945384979248,
      "learning_rate": 0.000513400174507852,
      "loss": 2.2777,
      "step": 57161
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.213165283203125,
      "learning_rate": 0.0005133972994191102,
      "loss": 2.0884,
      "step": 57162
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7503060698509216,
      "learning_rate": 0.0005133944242906939,
      "loss": 2.4047,
      "step": 57163
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.8447554111480713,
      "learning_rate": 0.0005133915491226034,
      "loss": 2.2743,
      "step": 57164
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.8667718172073364,
      "learning_rate": 0.0005133886739148394,
      "loss": 2.3555,
      "step": 57165
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7433778643608093,
      "learning_rate": 0.0005133857986674025,
      "loss": 2.1949,
      "step": 57166
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7054034471511841,
      "learning_rate": 0.0005133829233802932,
      "loss": 2.404,
      "step": 57167
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.9465728998184204,
      "learning_rate": 0.0005133800480535118,
      "loss": 2.2048,
      "step": 57168
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.777729868888855,
      "learning_rate": 0.0005133771726870592,
      "loss": 2.2482,
      "step": 57169
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.771519124507904,
      "learning_rate": 0.0005133742972809357,
      "loss": 2.3084,
      "step": 57170
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.6991193294525146,
      "learning_rate": 0.0005133714218351419,
      "loss": 2.1172,
      "step": 57171
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7873746752738953,
      "learning_rate": 0.0005133685463496782,
      "loss": 2.3281,
      "step": 57172
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.8885587453842163,
      "learning_rate": 0.0005133656708245454,
      "loss": 2.2175,
      "step": 57173
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7638522386550903,
      "learning_rate": 0.0005133627952597439,
      "loss": 2.1762,
      "step": 57174
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7823852896690369,
      "learning_rate": 0.0005133599196552742,
      "loss": 2.3192,
      "step": 57175
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.9149937629699707,
      "learning_rate": 0.0005133570440111368,
      "loss": 2.4819,
      "step": 57176
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7006551027297974,
      "learning_rate": 0.0005133541683273324,
      "loss": 2.2963,
      "step": 57177
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7924540042877197,
      "learning_rate": 0.0005133512926038613,
      "loss": 2.1815,
      "step": 57178
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7889112830162048,
      "learning_rate": 0.0005133484168407243,
      "loss": 2.3207,
      "step": 57179
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.8960026502609253,
      "learning_rate": 0.0005133455410379217,
      "loss": 2.4073,
      "step": 57180
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.9186556339263916,
      "learning_rate": 0.0005133426651954542,
      "loss": 2.6452,
      "step": 57181
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.778998076915741,
      "learning_rate": 0.0005133397893133223,
      "loss": 2.1992,
      "step": 57182
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7615366578102112,
      "learning_rate": 0.0005133369133915265,
      "loss": 2.4104,
      "step": 57183
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7565593719482422,
      "learning_rate": 0.0005133340374300671,
      "loss": 2.3447,
      "step": 57184
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.84315425157547,
      "learning_rate": 0.0005133311614289452,
      "loss": 2.2369,
      "step": 57185
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.6955538392066956,
      "learning_rate": 0.000513328285388161,
      "loss": 2.4302,
      "step": 57186
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7434996962547302,
      "learning_rate": 0.0005133254093077149,
      "loss": 2.3482,
      "step": 57187
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.6836192607879639,
      "learning_rate": 0.0005133225331876075,
      "loss": 2.0897,
      "step": 57188
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7673633694648743,
      "learning_rate": 0.0005133196570278396,
      "loss": 2.1392,
      "step": 57189
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7059138417243958,
      "learning_rate": 0.0005133167808284114,
      "loss": 2.3077,
      "step": 57190
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7242527008056641,
      "learning_rate": 0.0005133139045893237,
      "loss": 2.2432,
      "step": 57191
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7657002210617065,
      "learning_rate": 0.0005133110283105769,
      "loss": 2.1164,
      "step": 57192
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.723896324634552,
      "learning_rate": 0.0005133081519921715,
      "loss": 2.3973,
      "step": 57193
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7447243928909302,
      "learning_rate": 0.0005133052756341081,
      "loss": 2.1434,
      "step": 57194
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.8254084587097168,
      "learning_rate": 0.0005133023992363873,
      "loss": 2.5523,
      "step": 57195
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.76847243309021,
      "learning_rate": 0.0005132995227990094,
      "loss": 2.2499,
      "step": 57196
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.6945028901100159,
      "learning_rate": 0.0005132966463219753,
      "loss": 2.1935,
      "step": 57197
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.9102271795272827,
      "learning_rate": 0.0005132937698052852,
      "loss": 2.1982,
      "step": 57198
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.774260938167572,
      "learning_rate": 0.0005132908932489398,
      "loss": 2.4486,
      "step": 57199
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7416828870773315,
      "learning_rate": 0.0005132880166529395,
      "loss": 2.1773,
      "step": 57200
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.72303307056427,
      "learning_rate": 0.000513285140017285,
      "loss": 2.0952,
      "step": 57201
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.6915405988693237,
      "learning_rate": 0.0005132822633419768,
      "loss": 2.0929,
      "step": 57202
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7588738203048706,
      "learning_rate": 0.0005132793866270154,
      "loss": 2.0533,
      "step": 57203
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.822860836982727,
      "learning_rate": 0.0005132765098724014,
      "loss": 2.3696,
      "step": 57204
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.701922595500946,
      "learning_rate": 0.0005132736330781351,
      "loss": 2.217,
      "step": 57205
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7360316514968872,
      "learning_rate": 0.0005132707562442172,
      "loss": 2.2863,
      "step": 57206
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7214932441711426,
      "learning_rate": 0.0005132678793706484,
      "loss": 2.2023,
      "step": 57207
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.6825383901596069,
      "learning_rate": 0.0005132650024574291,
      "loss": 2.053,
      "step": 57208
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.731641411781311,
      "learning_rate": 0.0005132621255045597,
      "loss": 2.2732,
      "step": 57209
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.71320641040802,
      "learning_rate": 0.0005132592485120408,
      "loss": 2.1521,
      "step": 57210
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.6878558397293091,
      "learning_rate": 0.000513256371479873,
      "loss": 2.1475,
      "step": 57211
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7192807793617249,
      "learning_rate": 0.0005132534944080569,
      "loss": 2.077,
      "step": 57212
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.718773365020752,
      "learning_rate": 0.0005132506172965929,
      "loss": 2.2906,
      "step": 57213
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7794111371040344,
      "learning_rate": 0.0005132477401454816,
      "loss": 2.2992,
      "step": 57214
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7258119583129883,
      "learning_rate": 0.0005132448629547235,
      "loss": 2.282,
      "step": 57215
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7068984508514404,
      "learning_rate": 0.0005132419857243191,
      "loss": 2.0773,
      "step": 57216
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7338576912879944,
      "learning_rate": 0.0005132391084542691,
      "loss": 2.2037,
      "step": 57217
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7887305617332458,
      "learning_rate": 0.0005132362311445739,
      "loss": 2.4331,
      "step": 57218
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.683119535446167,
      "learning_rate": 0.0005132333537952341,
      "loss": 2.1379,
      "step": 57219
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.6937960386276245,
      "learning_rate": 0.00051323047640625,
      "loss": 2.2844,
      "step": 57220
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7155951261520386,
      "learning_rate": 0.0005132275989776225,
      "loss": 2.2378,
      "step": 57221
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.8553153872489929,
      "learning_rate": 0.000513224721509352,
      "loss": 2.2246,
      "step": 57222
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7098952531814575,
      "learning_rate": 0.000513221844001439,
      "loss": 2.2337,
      "step": 57223
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7670085430145264,
      "learning_rate": 0.0005132189664538839,
      "loss": 2.3417,
      "step": 57224
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.8154516816139221,
      "learning_rate": 0.0005132160888666874,
      "loss": 2.142,
      "step": 57225
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.6951839923858643,
      "learning_rate": 0.00051321321123985,
      "loss": 2.3931,
      "step": 57226
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7329668998718262,
      "learning_rate": 0.0005132103335733723,
      "loss": 2.2863,
      "step": 57227
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.9723184704780579,
      "learning_rate": 0.0005132074558672548,
      "loss": 2.3959,
      "step": 57228
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7738330960273743,
      "learning_rate": 0.0005132045781214979,
      "loss": 2.4542,
      "step": 57229
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7312203049659729,
      "learning_rate": 0.0005132017003361023,
      "loss": 2.2035,
      "step": 57230
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7259504795074463,
      "learning_rate": 0.0005131988225110685,
      "loss": 2.1534,
      "step": 57231
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.8820453882217407,
      "learning_rate": 0.000513195944646397,
      "loss": 2.2847,
      "step": 57232
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.6911636590957642,
      "learning_rate": 0.0005131930667420883,
      "loss": 2.1062,
      "step": 57233
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.9411184191703796,
      "learning_rate": 0.0005131901887981431,
      "loss": 2.1782,
      "step": 57234
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.791572630405426,
      "learning_rate": 0.0005131873108145617,
      "loss": 2.1551,
      "step": 57235
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7453122735023499,
      "learning_rate": 0.0005131844327913448,
      "loss": 2.1963,
      "step": 57236
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7241057753562927,
      "learning_rate": 0.000513181554728493,
      "loss": 2.0851,
      "step": 57237
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.6814951300621033,
      "learning_rate": 0.0005131786766260065,
      "loss": 2.1814,
      "step": 57238
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7679057717323303,
      "learning_rate": 0.0005131757984838863,
      "loss": 2.2017,
      "step": 57239
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.73598313331604,
      "learning_rate": 0.0005131729203021325,
      "loss": 2.2845,
      "step": 57240
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7414726614952087,
      "learning_rate": 0.0005131700420807459,
      "loss": 2.3691,
      "step": 57241
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7248622179031372,
      "learning_rate": 0.0005131671638197269,
      "loss": 2.1314,
      "step": 57242
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.6509455442428589,
      "learning_rate": 0.0005131642855190762,
      "loss": 2.189,
      "step": 57243
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7140708565711975,
      "learning_rate": 0.0005131614071787942,
      "loss": 2.1989,
      "step": 57244
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7323300242424011,
      "learning_rate": 0.0005131585287988815,
      "loss": 2.1603,
      "step": 57245
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.0989229679107666,
      "learning_rate": 0.0005131556503793385,
      "loss": 2.1425,
      "step": 57246
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.737582266330719,
      "learning_rate": 0.000513152771920166,
      "loss": 2.3237,
      "step": 57247
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7356635332107544,
      "learning_rate": 0.0005131498934213643,
      "loss": 2.2014,
      "step": 57248
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.6841050982475281,
      "learning_rate": 0.000513147014882934,
      "loss": 2.1194,
      "step": 57249
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.702743649482727,
      "learning_rate": 0.0005131441363048757,
      "loss": 2.2186,
      "step": 57250
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7779889702796936,
      "learning_rate": 0.0005131412576871898,
      "loss": 2.2995,
      "step": 57251
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.8090589046478271,
      "learning_rate": 0.000513138379029877,
      "loss": 2.3363,
      "step": 57252
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.836392879486084,
      "learning_rate": 0.0005131355003329377,
      "loss": 2.315,
      "step": 57253
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.739439845085144,
      "learning_rate": 0.0005131326215963725,
      "loss": 2.1076,
      "step": 57254
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7112599611282349,
      "learning_rate": 0.0005131297428201819,
      "loss": 2.0406,
      "step": 57255
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.9113592505455017,
      "learning_rate": 0.0005131268640043665,
      "loss": 2.3064,
      "step": 57256
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.6794267296791077,
      "learning_rate": 0.0005131239851489267,
      "loss": 2.3035,
      "step": 57257
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.6478453278541565,
      "learning_rate": 0.0005131211062538632,
      "loss": 2.0227,
      "step": 57258
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7971965670585632,
      "learning_rate": 0.0005131182273191765,
      "loss": 2.5518,
      "step": 57259
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7086435556411743,
      "learning_rate": 0.000513115348344867,
      "loss": 2.356,
      "step": 57260
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7309253215789795,
      "learning_rate": 0.0005131124693309355,
      "loss": 2.2446,
      "step": 57261
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.6998724341392517,
      "learning_rate": 0.0005131095902773822,
      "loss": 2.436,
      "step": 57262
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.6851970553398132,
      "learning_rate": 0.0005131067111842078,
      "loss": 1.9608,
      "step": 57263
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.8492065668106079,
      "learning_rate": 0.0005131038320514129,
      "loss": 2.5349,
      "step": 57264
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7279972434043884,
      "learning_rate": 0.000513100952878998,
      "loss": 2.3727,
      "step": 57265
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.6942802667617798,
      "learning_rate": 0.0005130980736669636,
      "loss": 2.2368,
      "step": 57266
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.6985692381858826,
      "learning_rate": 0.0005130951944153103,
      "loss": 2.295,
      "step": 57267
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.6889856457710266,
      "learning_rate": 0.0005130923151240384,
      "loss": 2.2014,
      "step": 57268
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7492750287055969,
      "learning_rate": 0.0005130894357931487,
      "loss": 2.1122,
      "step": 57269
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7144154906272888,
      "learning_rate": 0.0005130865564226418,
      "loss": 2.541,
      "step": 57270
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7106326222419739,
      "learning_rate": 0.000513083677012518,
      "loss": 2.307,
      "step": 57271
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.731678307056427,
      "learning_rate": 0.0005130807975627779,
      "loss": 2.3071,
      "step": 57272
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7063686847686768,
      "learning_rate": 0.000513077918073422,
      "loss": 2.3123,
      "step": 57273
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7583559155464172,
      "learning_rate": 0.0005130750385444509,
      "loss": 2.4545,
      "step": 57274
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7365925908088684,
      "learning_rate": 0.0005130721589758652,
      "loss": 2.302,
      "step": 57275
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.767371654510498,
      "learning_rate": 0.0005130692793676654,
      "loss": 2.103,
      "step": 57276
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7278469800949097,
      "learning_rate": 0.0005130663997198519,
      "loss": 2.1931,
      "step": 57277
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7449104189872742,
      "learning_rate": 0.0005130635200324253,
      "loss": 2.241,
      "step": 57278
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7757241129875183,
      "learning_rate": 0.0005130606403053863,
      "loss": 2.2414,
      "step": 57279
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.6952191591262817,
      "learning_rate": 0.0005130577605387352,
      "loss": 2.253,
      "step": 57280
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7103709578514099,
      "learning_rate": 0.0005130548807324727,
      "loss": 2.3707,
      "step": 57281
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.6721940040588379,
      "learning_rate": 0.0005130520008865993,
      "loss": 2.2779,
      "step": 57282
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7393138408660889,
      "learning_rate": 0.0005130491210011155,
      "loss": 2.3507,
      "step": 57283
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.722586452960968,
      "learning_rate": 0.0005130462410760217,
      "loss": 2.1358,
      "step": 57284
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7696468234062195,
      "learning_rate": 0.0005130433611113188,
      "loss": 2.311,
      "step": 57285
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7180210947990417,
      "learning_rate": 0.000513040481107007,
      "loss": 2.336,
      "step": 57286
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7219910025596619,
      "learning_rate": 0.0005130376010630869,
      "loss": 2.2039,
      "step": 57287
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7771499752998352,
      "learning_rate": 0.0005130347209795593,
      "loss": 2.2223,
      "step": 57288
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7120013236999512,
      "learning_rate": 0.0005130318408564243,
      "loss": 2.2466,
      "step": 57289
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7630441188812256,
      "learning_rate": 0.0005130289606936827,
      "loss": 2.5228,
      "step": 57290
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7442563772201538,
      "learning_rate": 0.000513026080491335,
      "loss": 2.4034,
      "step": 57291
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7158564329147339,
      "learning_rate": 0.0005130232002493818,
      "loss": 2.0525,
      "step": 57292
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7650177478790283,
      "learning_rate": 0.0005130203199678236,
      "loss": 2.3006,
      "step": 57293
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.8001524806022644,
      "learning_rate": 0.0005130174396466608,
      "loss": 2.2258,
      "step": 57294
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7280421853065491,
      "learning_rate": 0.0005130145592858941,
      "loss": 2.3626,
      "step": 57295
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7176761627197266,
      "learning_rate": 0.0005130116788855239,
      "loss": 2.1837,
      "step": 57296
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7267851233482361,
      "learning_rate": 0.0005130087984455509,
      "loss": 2.1814,
      "step": 57297
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7167279124259949,
      "learning_rate": 0.0005130059179659754,
      "loss": 2.2486,
      "step": 57298
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7295029759407043,
      "learning_rate": 0.0005130030374467983,
      "loss": 2.1999,
      "step": 57299
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.6666862964630127,
      "learning_rate": 0.0005130001568880198,
      "loss": 2.1077,
      "step": 57300
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.6961488723754883,
      "learning_rate": 0.0005129972762896405,
      "loss": 2.3457,
      "step": 57301
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7644962668418884,
      "learning_rate": 0.0005129943956516611,
      "loss": 2.2091,
      "step": 57302
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7214509844779968,
      "learning_rate": 0.0005129915149740819,
      "loss": 2.39,
      "step": 57303
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7497581839561462,
      "learning_rate": 0.0005129886342569035,
      "loss": 2.4651,
      "step": 57304
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7036178708076477,
      "learning_rate": 0.0005129857535001266,
      "loss": 2.312,
      "step": 57305
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.739614725112915,
      "learning_rate": 0.0005129828727037516,
      "loss": 2.3356,
      "step": 57306
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7956587076187134,
      "learning_rate": 0.0005129799918677792,
      "loss": 2.4434,
      "step": 57307
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7310498356819153,
      "learning_rate": 0.0005129771109922096,
      "loss": 2.383,
      "step": 57308
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.722977340221405,
      "learning_rate": 0.0005129742300770437,
      "loss": 2.1783,
      "step": 57309
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7840824127197266,
      "learning_rate": 0.0005129713491222817,
      "loss": 2.1025,
      "step": 57310
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7436405420303345,
      "learning_rate": 0.0005129684681279245,
      "loss": 2.2791,
      "step": 57311
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7085357904434204,
      "learning_rate": 0.0005129655870939724,
      "loss": 2.337,
      "step": 57312
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.75132817029953,
      "learning_rate": 0.0005129627060204258,
      "loss": 2.1899,
      "step": 57313
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7530611157417297,
      "learning_rate": 0.0005129598249072856,
      "loss": 2.3399,
      "step": 57314
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.730665385723114,
      "learning_rate": 0.000512956943754552,
      "loss": 2.1488,
      "step": 57315
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.6502081751823425,
      "learning_rate": 0.0005129540625622258,
      "loss": 2.2835,
      "step": 57316
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.72637540102005,
      "learning_rate": 0.0005129511813303074,
      "loss": 1.9598,
      "step": 57317
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.8049295544624329,
      "learning_rate": 0.0005129483000587973,
      "loss": 2.1024,
      "step": 57318
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7576642036437988,
      "learning_rate": 0.0005129454187476962,
      "loss": 2.1815,
      "step": 57319
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.6908551454544067,
      "learning_rate": 0.0005129425373970045,
      "loss": 2.3223,
      "step": 57320
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7954142689704895,
      "learning_rate": 0.0005129396560067226,
      "loss": 2.3099,
      "step": 57321
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.8389614224433899,
      "learning_rate": 0.0005129367745768514,
      "loss": 2.3383,
      "step": 57322
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7499744892120361,
      "learning_rate": 0.0005129338931073911,
      "loss": 2.1937,
      "step": 57323
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7078742980957031,
      "learning_rate": 0.0005129310115983424,
      "loss": 2.3907,
      "step": 57324
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7535592913627625,
      "learning_rate": 0.000512928130049706,
      "loss": 2.0613,
      "step": 57325
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7093121409416199,
      "learning_rate": 0.000512925248461482,
      "loss": 2.0201,
      "step": 57326
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7252146005630493,
      "learning_rate": 0.0005129223668336712,
      "loss": 2.1371,
      "step": 57327
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.8639933466911316,
      "learning_rate": 0.0005129194851662742,
      "loss": 2.0478,
      "step": 57328
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7215770483016968,
      "learning_rate": 0.0005129166034592914,
      "loss": 2.2473,
      "step": 57329
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7868651747703552,
      "learning_rate": 0.0005129137217127234,
      "loss": 2.1112,
      "step": 57330
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7724897265434265,
      "learning_rate": 0.0005129108399265707,
      "loss": 2.313,
      "step": 57331
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7786335349082947,
      "learning_rate": 0.0005129079581008339,
      "loss": 2.3106,
      "step": 57332
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7384218573570251,
      "learning_rate": 0.0005129050762355135,
      "loss": 2.3501,
      "step": 57333
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7937842607498169,
      "learning_rate": 0.00051290219433061,
      "loss": 2.3134,
      "step": 57334
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.8369253277778625,
      "learning_rate": 0.0005128993123861238,
      "loss": 2.3905,
      "step": 57335
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7142077684402466,
      "learning_rate": 0.0005128964304020558,
      "loss": 2.2458,
      "step": 57336
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7732022404670715,
      "learning_rate": 0.0005128935483784063,
      "loss": 1.9926,
      "step": 57337
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7577959299087524,
      "learning_rate": 0.0005128906663151759,
      "loss": 2.3437,
      "step": 57338
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7661722302436829,
      "learning_rate": 0.000512887784212365,
      "loss": 2.0583,
      "step": 57339
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.6621905565261841,
      "learning_rate": 0.0005128849020699743,
      "loss": 2.0968,
      "step": 57340
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.8574507832527161,
      "learning_rate": 0.0005128820198880043,
      "loss": 2.4,
      "step": 57341
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7223890423774719,
      "learning_rate": 0.0005128791376664555,
      "loss": 2.1452,
      "step": 57342
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7412081956863403,
      "learning_rate": 0.0005128762554053284,
      "loss": 2.2247,
      "step": 57343
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.8585839867591858,
      "learning_rate": 0.0005128733731046236,
      "loss": 1.9509,
      "step": 57344
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7339772582054138,
      "learning_rate": 0.0005128704907643417,
      "loss": 2.2398,
      "step": 57345
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.8376784920692444,
      "learning_rate": 0.0005128676083844832,
      "loss": 2.14,
      "step": 57346
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7445911765098572,
      "learning_rate": 0.0005128647259650484,
      "loss": 2.3139,
      "step": 57347
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7600821256637573,
      "learning_rate": 0.0005128618435060382,
      "loss": 2.2839,
      "step": 57348
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.8234855532646179,
      "learning_rate": 0.0005128589610074529,
      "loss": 2.1772,
      "step": 57349
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7400074601173401,
      "learning_rate": 0.000512856078469293,
      "loss": 2.2997,
      "step": 57350
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7380180358886719,
      "learning_rate": 0.0005128531958915593,
      "loss": 2.4155,
      "step": 57351
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7011021375656128,
      "learning_rate": 0.0005128503132742522,
      "loss": 2.208,
      "step": 57352
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7372098565101624,
      "learning_rate": 0.000512847430617372,
      "loss": 2.1386,
      "step": 57353
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7307161092758179,
      "learning_rate": 0.0005128445479209195,
      "loss": 2.1238,
      "step": 57354
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7303133010864258,
      "learning_rate": 0.0005128416651848954,
      "loss": 2.4479,
      "step": 57355
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.742512583732605,
      "learning_rate": 0.0005128387824092998,
      "loss": 2.3116,
      "step": 57356
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7716838717460632,
      "learning_rate": 0.0005128358995941336,
      "loss": 2.0511,
      "step": 57357
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7367931604385376,
      "learning_rate": 0.000512833016739397,
      "loss": 2.1991,
      "step": 57358
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.6797036528587341,
      "learning_rate": 0.0005128301338450909,
      "loss": 2.2782,
      "step": 57359
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7308839559555054,
      "learning_rate": 0.0005128272509112155,
      "loss": 2.2811,
      "step": 57360
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7141330242156982,
      "learning_rate": 0.0005128243679377717,
      "loss": 2.0981,
      "step": 57361
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7343946099281311,
      "learning_rate": 0.0005128214849247596,
      "loss": 2.3613,
      "step": 57362
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.710383951663971,
      "learning_rate": 0.0005128186018721802,
      "loss": 2.4236,
      "step": 57363
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.8257753849029541,
      "learning_rate": 0.0005128157187800337,
      "loss": 2.0766,
      "step": 57364
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.762538731098175,
      "learning_rate": 0.0005128128356483207,
      "loss": 2.3321,
      "step": 57365
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7827543020248413,
      "learning_rate": 0.0005128099524770418,
      "loss": 2.0243,
      "step": 57366
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.8152911067008972,
      "learning_rate": 0.0005128070692661975,
      "loss": 2.2682,
      "step": 57367
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7854901552200317,
      "learning_rate": 0.0005128041860157883,
      "loss": 2.1066,
      "step": 57368
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7957748770713806,
      "learning_rate": 0.0005128013027258149,
      "loss": 2.4538,
      "step": 57369
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7152109742164612,
      "learning_rate": 0.0005127984193962777,
      "loss": 2.1292,
      "step": 57370
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7415753602981567,
      "learning_rate": 0.0005127955360271772,
      "loss": 2.2409,
      "step": 57371
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7638168931007385,
      "learning_rate": 0.0005127926526185141,
      "loss": 2.1252,
      "step": 57372
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7986651062965393,
      "learning_rate": 0.0005127897691702887,
      "loss": 2.2856,
      "step": 57373
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.838289201259613,
      "learning_rate": 0.0005127868856825017,
      "loss": 2.3023,
      "step": 57374
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7551513314247131,
      "learning_rate": 0.0005127840021551536,
      "loss": 2.2204,
      "step": 57375
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7830761075019836,
      "learning_rate": 0.000512781118588245,
      "loss": 2.4743,
      "step": 57376
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.8708736300468445,
      "learning_rate": 0.0005127782349817763,
      "loss": 2.2495,
      "step": 57377
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7524048686027527,
      "learning_rate": 0.0005127753513357482,
      "loss": 2.2309,
      "step": 57378
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7396292686462402,
      "learning_rate": 0.0005127724676501611,
      "loss": 2.0508,
      "step": 57379
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.8039893507957458,
      "learning_rate": 0.0005127695839250155,
      "loss": 2.4534,
      "step": 57380
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.968784749507904,
      "learning_rate": 0.0005127667001603121,
      "loss": 2.227,
      "step": 57381
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7242251038551331,
      "learning_rate": 0.0005127638163560514,
      "loss": 2.4332,
      "step": 57382
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.8570968508720398,
      "learning_rate": 0.0005127609325122338,
      "loss": 2.2737,
      "step": 57383
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7034861445426941,
      "learning_rate": 0.0005127580486288599,
      "loss": 2.1552,
      "step": 57384
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7382546067237854,
      "learning_rate": 0.0005127551647059303,
      "loss": 2.2959,
      "step": 57385
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.710747241973877,
      "learning_rate": 0.0005127522807434456,
      "loss": 2.361,
      "step": 57386
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.701227068901062,
      "learning_rate": 0.0005127493967414062,
      "loss": 2.2916,
      "step": 57387
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.8812029957771301,
      "learning_rate": 0.0005127465126998125,
      "loss": 2.1171,
      "step": 57388
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7886424660682678,
      "learning_rate": 0.0005127436286186653,
      "loss": 2.2324,
      "step": 57389
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.858707070350647,
      "learning_rate": 0.0005127407444979651,
      "loss": 2.2377,
      "step": 57390
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.8637458682060242,
      "learning_rate": 0.0005127378603377124,
      "loss": 2.3964,
      "step": 57391
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7562791109085083,
      "learning_rate": 0.0005127349761379077,
      "loss": 2.1916,
      "step": 57392
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7672299742698669,
      "learning_rate": 0.0005127320918985515,
      "loss": 2.4154,
      "step": 57393
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.9035245180130005,
      "learning_rate": 0.0005127292076196444,
      "loss": 2.3527,
      "step": 57394
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.6950680613517761,
      "learning_rate": 0.000512726323301187,
      "loss": 2.1084,
      "step": 57395
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.6675093770027161,
      "learning_rate": 0.0005127234389431796,
      "loss": 2.0843,
      "step": 57396
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.9899516701698303,
      "learning_rate": 0.0005127205545456231,
      "loss": 2.143,
      "step": 57397
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7846965789794922,
      "learning_rate": 0.0005127176701085177,
      "loss": 2.1037,
      "step": 57398
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.8246315717697144,
      "learning_rate": 0.0005127147856318641,
      "loss": 2.3598,
      "step": 57399
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.8186725974082947,
      "learning_rate": 0.0005127119011156628,
      "loss": 2.1801,
      "step": 57400
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7654643058776855,
      "learning_rate": 0.0005127090165599143,
      "loss": 2.2502,
      "step": 57401
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7798849940299988,
      "learning_rate": 0.0005127061319646193,
      "loss": 2.3121,
      "step": 57402
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.6396766901016235,
      "learning_rate": 0.0005127032473297782,
      "loss": 1.9057,
      "step": 57403
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.769188642501831,
      "learning_rate": 0.0005127003626553914,
      "loss": 2.2873,
      "step": 57404
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.8206847906112671,
      "learning_rate": 0.0005126974779414597,
      "loss": 2.0925,
      "step": 57405
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.770525336265564,
      "learning_rate": 0.0005126945931879835,
      "loss": 2.3256,
      "step": 57406
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.8436427116394043,
      "learning_rate": 0.0005126917083949634,
      "loss": 2.0758,
      "step": 57407
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7975592613220215,
      "learning_rate": 0.0005126888235623998,
      "loss": 2.294,
      "step": 57408
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7189339399337769,
      "learning_rate": 0.0005126859386902934,
      "loss": 2.2496,
      "step": 57409
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.890493631362915,
      "learning_rate": 0.0005126830537786447,
      "loss": 2.2747,
      "step": 57410
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7206544876098633,
      "learning_rate": 0.0005126801688274541,
      "loss": 2.2478,
      "step": 57411
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7379610538482666,
      "learning_rate": 0.0005126772838367224,
      "loss": 2.3536,
      "step": 57412
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.8379034399986267,
      "learning_rate": 0.0005126743988064499,
      "loss": 2.0591,
      "step": 57413
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.6794205904006958,
      "learning_rate": 0.0005126715137366371,
      "loss": 2.1718,
      "step": 57414
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.8335151076316833,
      "learning_rate": 0.0005126686286272848,
      "loss": 2.1727,
      "step": 57415
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.733348548412323,
      "learning_rate": 0.0005126657434783934,
      "loss": 2.1827,
      "step": 57416
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.758097231388092,
      "learning_rate": 0.0005126628582899634,
      "loss": 2.3266,
      "step": 57417
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7740750908851624,
      "learning_rate": 0.0005126599730619952,
      "loss": 2.3303,
      "step": 57418
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7655273079872131,
      "learning_rate": 0.0005126570877944897,
      "loss": 2.3259,
      "step": 57419
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7350115776062012,
      "learning_rate": 0.0005126542024874471,
      "loss": 2.2093,
      "step": 57420
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.6880449652671814,
      "learning_rate": 0.0005126513171408682,
      "loss": 2.0172,
      "step": 57421
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7092545628547668,
      "learning_rate": 0.0005126484317547534,
      "loss": 2.1239,
      "step": 57422
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7639402151107788,
      "learning_rate": 0.000512645546329103,
      "loss": 2.3054,
      "step": 57423
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7897802591323853,
      "learning_rate": 0.000512642660863918,
      "loss": 2.2914,
      "step": 57424
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.8145961761474609,
      "learning_rate": 0.0005126397753591987,
      "loss": 2.2868,
      "step": 57425
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7808898091316223,
      "learning_rate": 0.0005126368898149456,
      "loss": 2.1963,
      "step": 57426
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7307876348495483,
      "learning_rate": 0.0005126340042311594,
      "loss": 2.1924,
      "step": 57427
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.6577631235122681,
      "learning_rate": 0.0005126311186078404,
      "loss": 1.9687,
      "step": 57428
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.8341469764709473,
      "learning_rate": 0.0005126282329449893,
      "loss": 2.0744,
      "step": 57429
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7484385371208191,
      "learning_rate": 0.0005126253472426066,
      "loss": 2.1693,
      "step": 57430
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7308757901191711,
      "learning_rate": 0.0005126224615006927,
      "loss": 2.062,
      "step": 57431
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7496144771575928,
      "learning_rate": 0.0005126195757192485,
      "loss": 2.3072,
      "step": 57432
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.8143832087516785,
      "learning_rate": 0.0005126166898982742,
      "loss": 2.0654,
      "step": 57433
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7266696095466614,
      "learning_rate": 0.0005126138040377705,
      "loss": 2.2051,
      "step": 57434
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.8105713129043579,
      "learning_rate": 0.0005126109181377378,
      "loss": 2.2778,
      "step": 57435
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.9198398590087891,
      "learning_rate": 0.0005126080321981768,
      "loss": 2.0668,
      "step": 57436
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7305607199668884,
      "learning_rate": 0.0005126051462190878,
      "loss": 2.0892,
      "step": 57437
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7863562107086182,
      "learning_rate": 0.0005126022602004715,
      "loss": 2.1533,
      "step": 57438
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.9085183143615723,
      "learning_rate": 0.0005125993741423285,
      "loss": 1.9832,
      "step": 57439
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7343303561210632,
      "learning_rate": 0.0005125964880446593,
      "loss": 2.1066,
      "step": 57440
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.795161247253418,
      "learning_rate": 0.0005125936019074645,
      "loss": 2.3074,
      "step": 57441
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.785182535648346,
      "learning_rate": 0.0005125907157307444,
      "loss": 2.2392,
      "step": 57442
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.8215213418006897,
      "learning_rate": 0.0005125878295144996,
      "loss": 2.2036,
      "step": 57443
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7669591903686523,
      "learning_rate": 0.0005125849432587308,
      "loss": 2.3021,
      "step": 57444
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.8149333000183105,
      "learning_rate": 0.0005125820569634386,
      "loss": 2.2203,
      "step": 57445
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7236067056655884,
      "learning_rate": 0.0005125791706286231,
      "loss": 2.1496,
      "step": 57446
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.6554974317550659,
      "learning_rate": 0.0005125762842542853,
      "loss": 2.0924,
      "step": 57447
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.6635884642601013,
      "learning_rate": 0.0005125733978404255,
      "loss": 2.1007,
      "step": 57448
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7395187616348267,
      "learning_rate": 0.0005125705113870442,
      "loss": 2.2229,
      "step": 57449
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7730267643928528,
      "learning_rate": 0.0005125676248941422,
      "loss": 2.1665,
      "step": 57450
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.8326658010482788,
      "learning_rate": 0.0005125647383617198,
      "loss": 2.2757,
      "step": 57451
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7568616271018982,
      "learning_rate": 0.0005125618517897776,
      "loss": 2.2511,
      "step": 57452
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.6962758302688599,
      "learning_rate": 0.0005125589651783162,
      "loss": 2.2365,
      "step": 57453
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7717397212982178,
      "learning_rate": 0.000512556078527336,
      "loss": 2.2711,
      "step": 57454
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7715966701507568,
      "learning_rate": 0.0005125531918368375,
      "loss": 2.1934,
      "step": 57455
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.692284882068634,
      "learning_rate": 0.0005125503051068215,
      "loss": 2.3129,
      "step": 57456
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7222816348075867,
      "learning_rate": 0.0005125474183372884,
      "loss": 2.3857,
      "step": 57457
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7398000955581665,
      "learning_rate": 0.0005125445315282386,
      "loss": 2.2942,
      "step": 57458
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7549343109130859,
      "learning_rate": 0.000512541644679673,
      "loss": 2.2959,
      "step": 57459
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.8447619080543518,
      "learning_rate": 0.0005125387577915916,
      "loss": 2.2054,
      "step": 57460
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7101439833641052,
      "learning_rate": 0.0005125358708639954,
      "loss": 2.0821,
      "step": 57461
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.6975158452987671,
      "learning_rate": 0.0005125329838968848,
      "loss": 2.0499,
      "step": 57462
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.6959660649299622,
      "learning_rate": 0.0005125300968902602,
      "loss": 2.1847,
      "step": 57463
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.866012692451477,
      "learning_rate": 0.0005125272098441223,
      "loss": 2.2921,
      "step": 57464
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.8554948568344116,
      "learning_rate": 0.0005125243227584715,
      "loss": 2.3214,
      "step": 57465
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7052695155143738,
      "learning_rate": 0.0005125214356333084,
      "loss": 2.1254,
      "step": 57466
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7767065167427063,
      "learning_rate": 0.0005125185484686338,
      "loss": 2.1063,
      "step": 57467
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.8067076206207275,
      "learning_rate": 0.0005125156612644479,
      "loss": 2.4276,
      "step": 57468
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7393479347229004,
      "learning_rate": 0.0005125127740207512,
      "loss": 2.1942,
      "step": 57469
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.6902715563774109,
      "learning_rate": 0.0005125098867375444,
      "loss": 2.2934,
      "step": 57470
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.6753080487251282,
      "learning_rate": 0.000512506999414828,
      "loss": 2.2172,
      "step": 57471
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7036249041557312,
      "learning_rate": 0.0005125041120526025,
      "loss": 2.3482,
      "step": 57472
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7922009825706482,
      "learning_rate": 0.0005125012246508686,
      "loss": 2.2212,
      "step": 57473
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7850347757339478,
      "learning_rate": 0.0005124983372096267,
      "loss": 2.1338,
      "step": 57474
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.775768518447876,
      "learning_rate": 0.0005124954497288772,
      "loss": 2.1741,
      "step": 57475
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.6888805627822876,
      "learning_rate": 0.0005124925622086209,
      "loss": 2.1868,
      "step": 57476
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7131955623626709,
      "learning_rate": 0.0005124896746488581,
      "loss": 2.2329,
      "step": 57477
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7660353183746338,
      "learning_rate": 0.0005124867870495897,
      "loss": 2.341,
      "step": 57478
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.666892945766449,
      "learning_rate": 0.0005124838994108159,
      "loss": 2.2033,
      "step": 57479
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7296616435050964,
      "learning_rate": 0.0005124810117325372,
      "loss": 2.271,
      "step": 57480
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.837912917137146,
      "learning_rate": 0.0005124781240147543,
      "loss": 2.2601,
      "step": 57481
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7704423666000366,
      "learning_rate": 0.0005124752362574678,
      "loss": 2.0898,
      "step": 57482
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.6704502105712891,
      "learning_rate": 0.0005124723484606781,
      "loss": 2.3822,
      "step": 57483
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.784246027469635,
      "learning_rate": 0.0005124694606243857,
      "loss": 2.1671,
      "step": 57484
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7533826231956482,
      "learning_rate": 0.0005124665727485914,
      "loss": 2.2135,
      "step": 57485
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7517475485801697,
      "learning_rate": 0.0005124636848332954,
      "loss": 2.1854,
      "step": 57486
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.948699951171875,
      "learning_rate": 0.0005124607968784985,
      "loss": 2.2151,
      "step": 57487
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.8123813271522522,
      "learning_rate": 0.0005124579088842011,
      "loss": 2.1393,
      "step": 57488
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.1102720499038696,
      "learning_rate": 0.0005124550208504036,
      "loss": 2.0534,
      "step": 57489
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.705018162727356,
      "learning_rate": 0.0005124521327771068,
      "loss": 2.0895,
      "step": 57490
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7360351085662842,
      "learning_rate": 0.0005124492446643113,
      "loss": 2.3175,
      "step": 57491
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.8509641289710999,
      "learning_rate": 0.0005124463565120173,
      "loss": 2.1336,
      "step": 57492
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7619470357894897,
      "learning_rate": 0.0005124434683202256,
      "loss": 2.2882,
      "step": 57493
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7310189008712769,
      "learning_rate": 0.0005124405800889366,
      "loss": 2.4028,
      "step": 57494
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7223681807518005,
      "learning_rate": 0.0005124376918181509,
      "loss": 2.1259,
      "step": 57495
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7639400959014893,
      "learning_rate": 0.000512434803507869,
      "loss": 1.9634,
      "step": 57496
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.8565266728401184,
      "learning_rate": 0.0005124319151580915,
      "loss": 2.2837,
      "step": 57497
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.6997881531715393,
      "learning_rate": 0.0005124290267688189,
      "loss": 2.4315,
      "step": 57498
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.8578769564628601,
      "learning_rate": 0.0005124261383400516,
      "loss": 2.0295,
      "step": 57499
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.6986681818962097,
      "learning_rate": 0.0005124232498717905,
      "loss": 2.2601,
      "step": 57500
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7156826853752136,
      "learning_rate": 0.0005124203613640358,
      "loss": 2.2561,
      "step": 57501
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7098984122276306,
      "learning_rate": 0.0005124174728167881,
      "loss": 2.1682,
      "step": 57502
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.736807107925415,
      "learning_rate": 0.000512414584230048,
      "loss": 2.2395,
      "step": 57503
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7370976209640503,
      "learning_rate": 0.000512411695603816,
      "loss": 2.3184,
      "step": 57504
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.736335277557373,
      "learning_rate": 0.0005124088069380926,
      "loss": 2.0269,
      "step": 57505
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7221484780311584,
      "learning_rate": 0.0005124059182328786,
      "loss": 2.3541,
      "step": 57506
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7013067007064819,
      "learning_rate": 0.0005124030294881742,
      "loss": 2.2442,
      "step": 57507
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7140914797782898,
      "learning_rate": 0.00051240014070398,
      "loss": 2.3386,
      "step": 57508
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7303128242492676,
      "learning_rate": 0.0005123972518802968,
      "loss": 2.2649,
      "step": 57509
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.6792864799499512,
      "learning_rate": 0.0005123943630171248,
      "loss": 2.2249,
      "step": 57510
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7327374219894409,
      "learning_rate": 0.0005123914741144647,
      "loss": 2.1532,
      "step": 57511
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7815988659858704,
      "learning_rate": 0.0005123885851723171,
      "loss": 2.3581,
      "step": 57512
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.8209322690963745,
      "learning_rate": 0.0005123856961906823,
      "loss": 2.2388,
      "step": 57513
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.8149336576461792,
      "learning_rate": 0.000512382807169561,
      "loss": 2.2236,
      "step": 57514
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7059080600738525,
      "learning_rate": 0.0005123799181089539,
      "loss": 1.9984,
      "step": 57515
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7485611438751221,
      "learning_rate": 0.0005123770290088611,
      "loss": 2.207,
      "step": 57516
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7502778768539429,
      "learning_rate": 0.0005123741398692837,
      "loss": 2.2499,
      "step": 57517
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7208376526832581,
      "learning_rate": 0.0005123712506902217,
      "loss": 2.0245,
      "step": 57518
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7716284394264221,
      "learning_rate": 0.000512368361471676,
      "loss": 2.142,
      "step": 57519
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.6732848882675171,
      "learning_rate": 0.0005123654722136471,
      "loss": 2.1712,
      "step": 57520
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.9434000253677368,
      "learning_rate": 0.0005123625829161352,
      "loss": 2.3416,
      "step": 57521
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7548136115074158,
      "learning_rate": 0.0005123596935791413,
      "loss": 2.3148,
      "step": 57522
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.6584278345108032,
      "learning_rate": 0.0005123568042026657,
      "loss": 2.0966,
      "step": 57523
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7322126030921936,
      "learning_rate": 0.0005123539147867088,
      "loss": 2.3155,
      "step": 57524
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.6196541786193848,
      "learning_rate": 0.0005123510253312715,
      "loss": 2.3327,
      "step": 57525
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7893585562705994,
      "learning_rate": 0.0005123481358363539,
      "loss": 2.0592,
      "step": 57526
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7153546810150146,
      "learning_rate": 0.000512345246301957,
      "loss": 2.265,
      "step": 57527
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7607592940330505,
      "learning_rate": 0.000512342356728081,
      "loss": 2.2303,
      "step": 57528
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.8153131008148193,
      "learning_rate": 0.0005123394671147267,
      "loss": 2.3074,
      "step": 57529
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.708626925945282,
      "learning_rate": 0.0005123365774618943,
      "loss": 2.1525,
      "step": 57530
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7816267609596252,
      "learning_rate": 0.0005123336877695846,
      "loss": 2.33,
      "step": 57531
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.8048499822616577,
      "learning_rate": 0.000512330798037798,
      "loss": 2.3811,
      "step": 57532
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.6966151595115662,
      "learning_rate": 0.0005123279082665351,
      "loss": 2.3612,
      "step": 57533
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7546470761299133,
      "learning_rate": 0.0005123250184557965,
      "loss": 2.2644,
      "step": 57534
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.786775529384613,
      "learning_rate": 0.0005123221286055826,
      "loss": 2.4157,
      "step": 57535
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7292836308479309,
      "learning_rate": 0.0005123192387158941,
      "loss": 2.182,
      "step": 57536
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7305805087089539,
      "learning_rate": 0.0005123163487867313,
      "loss": 1.9691,
      "step": 57537
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7102907299995422,
      "learning_rate": 0.0005123134588180951,
      "loss": 2.3398,
      "step": 57538
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.730353057384491,
      "learning_rate": 0.0005123105688099856,
      "loss": 2.405,
      "step": 57539
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.6924254298210144,
      "learning_rate": 0.0005123076787624036,
      "loss": 2.3978,
      "step": 57540
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.743180513381958,
      "learning_rate": 0.0005123047886753497,
      "loss": 2.3173,
      "step": 57541
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7306568026542664,
      "learning_rate": 0.0005123018985488243,
      "loss": 2.1911,
      "step": 57542
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7923144102096558,
      "learning_rate": 0.0005122990083828281,
      "loss": 2.2369,
      "step": 57543
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7539434432983398,
      "learning_rate": 0.0005122961181773612,
      "loss": 2.2729,
      "step": 57544
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7190057635307312,
      "learning_rate": 0.0005122932279324246,
      "loss": 2.4046,
      "step": 57545
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.8161634206771851,
      "learning_rate": 0.0005122903376480188,
      "loss": 2.1792,
      "step": 57546
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.8203911781311035,
      "learning_rate": 0.000512287447324144,
      "loss": 2.4008,
      "step": 57547
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.8122639656066895,
      "learning_rate": 0.0005122845569608012,
      "loss": 2.2982,
      "step": 57548
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.8243295550346375,
      "learning_rate": 0.0005122816665579904,
      "loss": 2.3503,
      "step": 57549
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7831243276596069,
      "learning_rate": 0.0005122787761157126,
      "loss": 1.9942,
      "step": 57550
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7320858240127563,
      "learning_rate": 0.0005122758856339683,
      "loss": 2.2622,
      "step": 57551
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.752251923084259,
      "learning_rate": 0.0005122729951127576,
      "loss": 2.2404,
      "step": 57552
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7320964336395264,
      "learning_rate": 0.0005122701045520816,
      "loss": 2.3075,
      "step": 57553
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7476978302001953,
      "learning_rate": 0.0005122672139519405,
      "loss": 2.3911,
      "step": 57554
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7327641844749451,
      "learning_rate": 0.0005122643233123348,
      "loss": 2.2358,
      "step": 57555
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7173476219177246,
      "learning_rate": 0.0005122614326332653,
      "loss": 2.2732,
      "step": 57556
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7934712767601013,
      "learning_rate": 0.0005122585419147324,
      "loss": 2.2048,
      "step": 57557
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7730851769447327,
      "learning_rate": 0.0005122556511567365,
      "loss": 2.2351,
      "step": 57558
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7268297076225281,
      "learning_rate": 0.0005122527603592784,
      "loss": 2.2988,
      "step": 57559
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7905787825584412,
      "learning_rate": 0.0005122498695223585,
      "loss": 2.2591,
      "step": 57560
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7444949150085449,
      "learning_rate": 0.0005122469786459772,
      "loss": 2.1689,
      "step": 57561
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7279576063156128,
      "learning_rate": 0.0005122440877301354,
      "loss": 2.1314,
      "step": 57562
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7197876572608948,
      "learning_rate": 0.0005122411967748332,
      "loss": 2.1543,
      "step": 57563
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7541165947914124,
      "learning_rate": 0.0005122383057800714,
      "loss": 2.036,
      "step": 57564
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7066665887832642,
      "learning_rate": 0.0005122354147458506,
      "loss": 2.0829,
      "step": 57565
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7121049165725708,
      "learning_rate": 0.0005122325236721712,
      "loss": 2.4043,
      "step": 57566
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7243447303771973,
      "learning_rate": 0.0005122296325590337,
      "loss": 2.0501,
      "step": 57567
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.6961730122566223,
      "learning_rate": 0.0005122267414064387,
      "loss": 2.1932,
      "step": 57568
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.661605179309845,
      "learning_rate": 0.0005122238502143868,
      "loss": 2.2743,
      "step": 57569
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7470163702964783,
      "learning_rate": 0.0005122209589828784,
      "loss": 2.2441,
      "step": 57570
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7380092144012451,
      "learning_rate": 0.0005122180677119142,
      "loss": 2.0726,
      "step": 57571
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.8120572566986084,
      "learning_rate": 0.0005122151764014948,
      "loss": 2.3355,
      "step": 57572
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.6766675710678101,
      "learning_rate": 0.0005122122850516204,
      "loss": 2.2171,
      "step": 57573
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.8163347840309143,
      "learning_rate": 0.0005122093936622917,
      "loss": 2.0931,
      "step": 57574
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7124413847923279,
      "learning_rate": 0.0005122065022335094,
      "loss": 2.2067,
      "step": 57575
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.8014805316925049,
      "learning_rate": 0.0005122036107652738,
      "loss": 2.1546,
      "step": 57576
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.8692640662193298,
      "learning_rate": 0.0005122007192575855,
      "loss": 2.1864,
      "step": 57577
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7360012531280518,
      "learning_rate": 0.0005121978277104453,
      "loss": 2.5566,
      "step": 57578
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7694494724273682,
      "learning_rate": 0.0005121949361238533,
      "loss": 2.4699,
      "step": 57579
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.9210821986198425,
      "learning_rate": 0.0005121920444978103,
      "loss": 2.3274,
      "step": 57580
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.8463746905326843,
      "learning_rate": 0.0005121891528323169,
      "loss": 2.271,
      "step": 57581
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7167409658432007,
      "learning_rate": 0.0005121862611273735,
      "loss": 2.2809,
      "step": 57582
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.8205655813217163,
      "learning_rate": 0.0005121833693829806,
      "loss": 2.2106,
      "step": 57583
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7267608642578125,
      "learning_rate": 0.0005121804775991388,
      "loss": 2.2369,
      "step": 57584
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.988827109336853,
      "learning_rate": 0.0005121775857758487,
      "loss": 2.3033,
      "step": 57585
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7383270859718323,
      "learning_rate": 0.0005121746939131108,
      "loss": 2.286,
      "step": 57586
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.0633578300476074,
      "learning_rate": 0.0005121718020109255,
      "loss": 2.3131,
      "step": 57587
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.8502783179283142,
      "learning_rate": 0.0005121689100692936,
      "loss": 2.2645,
      "step": 57588
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7246098518371582,
      "learning_rate": 0.0005121660180882154,
      "loss": 2.2047,
      "step": 57589
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.8733277916908264,
      "learning_rate": 0.0005121631260676915,
      "loss": 2.1714,
      "step": 57590
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.9499695301055908,
      "learning_rate": 0.0005121602340077226,
      "loss": 2.4708,
      "step": 57591
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7735140323638916,
      "learning_rate": 0.000512157341908309,
      "loss": 2.1516,
      "step": 57592
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.9127707481384277,
      "learning_rate": 0.0005121544497694514,
      "loss": 2.4305,
      "step": 57593
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.8398476243019104,
      "learning_rate": 0.0005121515575911503,
      "loss": 2.2628,
      "step": 57594
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7911099195480347,
      "learning_rate": 0.0005121486653734063,
      "loss": 2.3174,
      "step": 57595
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7388902902603149,
      "learning_rate": 0.0005121457731162197,
      "loss": 2.3251,
      "step": 57596
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.9127730131149292,
      "learning_rate": 0.0005121428808195911,
      "loss": 2.2667,
      "step": 57597
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7688856720924377,
      "learning_rate": 0.0005121399884835214,
      "loss": 2.1886,
      "step": 57598
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.6756736636161804,
      "learning_rate": 0.0005121370961080107,
      "loss": 1.9805,
      "step": 57599
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.8476815223693848,
      "learning_rate": 0.0005121342036930597,
      "loss": 2.2698,
      "step": 57600
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.719774067401886,
      "learning_rate": 0.000512131311238669,
      "loss": 2.2446,
      "step": 57601
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.732941746711731,
      "learning_rate": 0.0005121284187448392,
      "loss": 2.1567,
      "step": 57602
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7755223512649536,
      "learning_rate": 0.0005121255262115706,
      "loss": 2.3335,
      "step": 57603
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7293158769607544,
      "learning_rate": 0.0005121226336388639,
      "loss": 2.2924,
      "step": 57604
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.8041389584541321,
      "learning_rate": 0.0005121197410267196,
      "loss": 2.1091,
      "step": 57605
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7182314395904541,
      "learning_rate": 0.0005121168483751381,
      "loss": 2.177,
      "step": 57606
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7086929678916931,
      "learning_rate": 0.0005121139556841202,
      "loss": 2.2929,
      "step": 57607
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7825092077255249,
      "learning_rate": 0.0005121110629536664,
      "loss": 2.0772,
      "step": 57608
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7443191409111023,
      "learning_rate": 0.000512108170183777,
      "loss": 2.0768,
      "step": 57609
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.8497028350830078,
      "learning_rate": 0.0005121052773744527,
      "loss": 2.3546,
      "step": 57610
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7257918119430542,
      "learning_rate": 0.0005121023845256941,
      "loss": 2.2692,
      "step": 57611
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7422024011611938,
      "learning_rate": 0.0005120994916375016,
      "loss": 2.2147,
      "step": 57612
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7209085822105408,
      "learning_rate": 0.0005120965987098758,
      "loss": 2.0926,
      "step": 57613
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.695868194103241,
      "learning_rate": 0.0005120937057428173,
      "loss": 2.2723,
      "step": 57614
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.8345441222190857,
      "learning_rate": 0.0005120908127363265,
      "loss": 2.1646,
      "step": 57615
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.8703977465629578,
      "learning_rate": 0.0005120879196904041,
      "loss": 2.0611,
      "step": 57616
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.745954155921936,
      "learning_rate": 0.0005120850266050505,
      "loss": 2.4721,
      "step": 57617
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7479116916656494,
      "learning_rate": 0.0005120821334802662,
      "loss": 2.1634,
      "step": 57618
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.8621604442596436,
      "learning_rate": 0.0005120792403160519,
      "loss": 2.0461,
      "step": 57619
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7370023727416992,
      "learning_rate": 0.0005120763471124081,
      "loss": 2.3422,
      "step": 57620
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.9111654758453369,
      "learning_rate": 0.0005120734538693354,
      "loss": 2.0629,
      "step": 57621
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7347348928451538,
      "learning_rate": 0.0005120705605868341,
      "loss": 2.2053,
      "step": 57622
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7622241973876953,
      "learning_rate": 0.0005120676672649048,
      "loss": 2.2668,
      "step": 57623
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.6893152594566345,
      "learning_rate": 0.0005120647739035484,
      "loss": 2.106,
      "step": 57624
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7498680353164673,
      "learning_rate": 0.0005120618805027648,
      "loss": 2.2999,
      "step": 57625
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.8019413352012634,
      "learning_rate": 0.0005120589870625552,
      "loss": 2.3045,
      "step": 57626
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.8131526112556458,
      "learning_rate": 0.0005120560935829197,
      "loss": 2.2411,
      "step": 57627
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.6864691972732544,
      "learning_rate": 0.000512053200063859,
      "loss": 2.1339,
      "step": 57628
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7342687845230103,
      "learning_rate": 0.0005120503065053736,
      "loss": 2.1687,
      "step": 57629
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.8094562888145447,
      "learning_rate": 0.000512047412907464,
      "loss": 2.1518,
      "step": 57630
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.8116528987884521,
      "learning_rate": 0.0005120445192701308,
      "loss": 2.1951,
      "step": 57631
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.6677849292755127,
      "learning_rate": 0.0005120416255933746,
      "loss": 2.1125,
      "step": 57632
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7126095294952393,
      "learning_rate": 0.0005120387318771958,
      "loss": 2.1674,
      "step": 57633
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.8369095325469971,
      "learning_rate": 0.0005120358381215949,
      "loss": 2.2363,
      "step": 57634
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.9021503925323486,
      "learning_rate": 0.0005120329443265727,
      "loss": 2.3768,
      "step": 57635
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7173869609832764,
      "learning_rate": 0.0005120300504921295,
      "loss": 2.0785,
      "step": 57636
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7559323310852051,
      "learning_rate": 0.000512027156618266,
      "loss": 2.106,
      "step": 57637
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7157321572303772,
      "learning_rate": 0.0005120242627049825,
      "loss": 2.1679,
      "step": 57638
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.6992685794830322,
      "learning_rate": 0.0005120213687522797,
      "loss": 2.1815,
      "step": 57639
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7306539416313171,
      "learning_rate": 0.0005120184747601582,
      "loss": 2.3567,
      "step": 57640
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7322952151298523,
      "learning_rate": 0.0005120155807286184,
      "loss": 2.3287,
      "step": 57641
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7110650539398193,
      "learning_rate": 0.000512012686657661,
      "loss": 2.1175,
      "step": 57642
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.715107798576355,
      "learning_rate": 0.0005120097925472863,
      "loss": 2.0193,
      "step": 57643
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7528976202011108,
      "learning_rate": 0.0005120068983974951,
      "loss": 2.4025,
      "step": 57644
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7648575305938721,
      "learning_rate": 0.0005120040042082878,
      "loss": 2.2259,
      "step": 57645
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7402747273445129,
      "learning_rate": 0.0005120011099796649,
      "loss": 2.2653,
      "step": 57646
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7149749994277954,
      "learning_rate": 0.000511998215711627,
      "loss": 2.0567,
      "step": 57647
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7328322529792786,
      "learning_rate": 0.0005119953214041747,
      "loss": 2.4556,
      "step": 57648
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7606983184814453,
      "learning_rate": 0.0005119924270573084,
      "loss": 1.9457,
      "step": 57649
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.8449741005897522,
      "learning_rate": 0.0005119895326710287,
      "loss": 2.2746,
      "step": 57650
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7364846467971802,
      "learning_rate": 0.0005119866382453361,
      "loss": 2.0327,
      "step": 57651
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7008443474769592,
      "learning_rate": 0.0005119837437802313,
      "loss": 2.2825,
      "step": 57652
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.6967771053314209,
      "learning_rate": 0.0005119808492757147,
      "loss": 2.1463,
      "step": 57653
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7545195817947388,
      "learning_rate": 0.0005119779547317868,
      "loss": 2.3106,
      "step": 57654
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.6870602965354919,
      "learning_rate": 0.0005119750601484482,
      "loss": 2.2491,
      "step": 57655
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7353931069374084,
      "learning_rate": 0.0005119721655256996,
      "loss": 2.2107,
      "step": 57656
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7655907273292542,
      "learning_rate": 0.0005119692708635412,
      "loss": 2.2335,
      "step": 57657
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7638797163963318,
      "learning_rate": 0.0005119663761619736,
      "loss": 2.1794,
      "step": 57658
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7505386471748352,
      "learning_rate": 0.0005119634814209977,
      "loss": 2.3878,
      "step": 57659
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.8019058704376221,
      "learning_rate": 0.0005119605866406137,
      "loss": 2.266,
      "step": 57660
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.6929694414138794,
      "learning_rate": 0.0005119576918208222,
      "loss": 2.2633,
      "step": 57661
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.8519489765167236,
      "learning_rate": 0.0005119547969616238,
      "loss": 2.2474,
      "step": 57662
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7565611600875854,
      "learning_rate": 0.000511951902063019,
      "loss": 2.2168,
      "step": 57663
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7070388793945312,
      "learning_rate": 0.0005119490071250082,
      "loss": 2.3519,
      "step": 57664
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7311012744903564,
      "learning_rate": 0.0005119461121475922,
      "loss": 2.3067,
      "step": 57665
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7221364974975586,
      "learning_rate": 0.0005119432171307714,
      "loss": 2.5593,
      "step": 57666
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7948901653289795,
      "learning_rate": 0.0005119403220745464,
      "loss": 2.2338,
      "step": 57667
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7208914160728455,
      "learning_rate": 0.0005119374269789177,
      "loss": 2.1757,
      "step": 57668
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7066132426261902,
      "learning_rate": 0.0005119345318438858,
      "loss": 2.1135,
      "step": 57669
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.6980770230293274,
      "learning_rate": 0.0005119316366694512,
      "loss": 2.1042,
      "step": 57670
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.751917839050293,
      "learning_rate": 0.0005119287414556146,
      "loss": 2.2062,
      "step": 57671
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7051166296005249,
      "learning_rate": 0.0005119258462023764,
      "loss": 2.2472,
      "step": 57672
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.764589250087738,
      "learning_rate": 0.0005119229509097372,
      "loss": 2.1429,
      "step": 57673
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.6807299256324768,
      "learning_rate": 0.0005119200555776976,
      "loss": 2.1957,
      "step": 57674
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7263857126235962,
      "learning_rate": 0.000511917160206258,
      "loss": 2.3058,
      "step": 57675
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7447918057441711,
      "learning_rate": 0.0005119142647954189,
      "loss": 2.3358,
      "step": 57676
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.6630443334579468,
      "learning_rate": 0.000511911369345181,
      "loss": 2.3112,
      "step": 57677
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7627851366996765,
      "learning_rate": 0.0005119084738555448,
      "loss": 2.0063,
      "step": 57678
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7753608226776123,
      "learning_rate": 0.0005119055783265108,
      "loss": 2.2562,
      "step": 57679
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.6825602054595947,
      "learning_rate": 0.0005119026827580796,
      "loss": 1.997,
      "step": 57680
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7184303998947144,
      "learning_rate": 0.0005118997871502517,
      "loss": 2.2126,
      "step": 57681
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.8576689958572388,
      "learning_rate": 0.0005118968915030276,
      "loss": 2.3214,
      "step": 57682
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.8527907729148865,
      "learning_rate": 0.0005118939958164078,
      "loss": 2.4659,
      "step": 57683
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7514949440956116,
      "learning_rate": 0.000511891100090393,
      "loss": 2.1672,
      "step": 57684
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7642015218734741,
      "learning_rate": 0.0005118882043249836,
      "loss": 2.3204,
      "step": 57685
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7361096739768982,
      "learning_rate": 0.0005118853085201802,
      "loss": 2.2264,
      "step": 57686
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.784933865070343,
      "learning_rate": 0.0005118824126759833,
      "loss": 2.1743,
      "step": 57687
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7053355574607849,
      "learning_rate": 0.0005118795167923935,
      "loss": 2.1156,
      "step": 57688
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7048704028129578,
      "learning_rate": 0.0005118766208694112,
      "loss": 2.2641,
      "step": 57689
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.750760555267334,
      "learning_rate": 0.0005118737249070371,
      "loss": 2.386,
      "step": 57690
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7744970917701721,
      "learning_rate": 0.0005118708289052717,
      "loss": 2.015,
      "step": 57691
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7469630837440491,
      "learning_rate": 0.0005118679328641154,
      "loss": 2.0638,
      "step": 57692
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.6961755752563477,
      "learning_rate": 0.000511865036783569,
      "loss": 2.0744,
      "step": 57693
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7898419499397278,
      "learning_rate": 0.0005118621406636328,
      "loss": 2.245,
      "step": 57694
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.72721266746521,
      "learning_rate": 0.0005118592445043076,
      "loss": 2.411,
      "step": 57695
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7204676866531372,
      "learning_rate": 0.0005118563483055936,
      "loss": 2.3221,
      "step": 57696
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.776515007019043,
      "learning_rate": 0.0005118534520674915,
      "loss": 2.2618,
      "step": 57697
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.722213089466095,
      "learning_rate": 0.0005118505557900019,
      "loss": 2.454,
      "step": 57698
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7341090440750122,
      "learning_rate": 0.0005118476594731253,
      "loss": 2.2515,
      "step": 57699
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.776950478553772,
      "learning_rate": 0.0005118447631168621,
      "loss": 2.2104,
      "step": 57700
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7061725854873657,
      "learning_rate": 0.000511841866721213,
      "loss": 2.2748,
      "step": 57701
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.682175874710083,
      "learning_rate": 0.0005118389702861787,
      "loss": 2.0459,
      "step": 57702
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.6772274971008301,
      "learning_rate": 0.0005118360738117593,
      "loss": 2.1389,
      "step": 57703
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7514939308166504,
      "learning_rate": 0.0005118331772979557,
      "loss": 2.2774,
      "step": 57704
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.6814284324645996,
      "learning_rate": 0.0005118302807447683,
      "loss": 2.1138,
      "step": 57705
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7867279052734375,
      "learning_rate": 0.0005118273841521975,
      "loss": 2.1664,
      "step": 57706
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.8311213254928589,
      "learning_rate": 0.0005118244875202441,
      "loss": 2.2814,
      "step": 57707
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.6737615466117859,
      "learning_rate": 0.0005118215908489086,
      "loss": 2.2233,
      "step": 57708
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7950533628463745,
      "learning_rate": 0.0005118186941381914,
      "loss": 2.2139,
      "step": 57709
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7298182249069214,
      "learning_rate": 0.000511815797388093,
      "loss": 2.2269,
      "step": 57710
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7862453460693359,
      "learning_rate": 0.0005118129005986143,
      "loss": 2.2816,
      "step": 57711
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.6923983097076416,
      "learning_rate": 0.0005118100037697555,
      "loss": 2.1198,
      "step": 57712
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7314286828041077,
      "learning_rate": 0.0005118071069015171,
      "loss": 2.0407,
      "step": 57713
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.8250262141227722,
      "learning_rate": 0.0005118042099938998,
      "loss": 2.2881,
      "step": 57714
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.9131328463554382,
      "learning_rate": 0.0005118013130469041,
      "loss": 2.0737,
      "step": 57715
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7923629879951477,
      "learning_rate": 0.0005117984160605306,
      "loss": 2.2063,
      "step": 57716
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.1119033098220825,
      "learning_rate": 0.0005117955190347798,
      "loss": 2.3273,
      "step": 57717
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7966126203536987,
      "learning_rate": 0.0005117926219696521,
      "loss": 2.3061,
      "step": 57718
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7559913992881775,
      "learning_rate": 0.0005117897248651482,
      "loss": 2.2055,
      "step": 57719
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.783558189868927,
      "learning_rate": 0.0005117868277212687,
      "loss": 2.1309,
      "step": 57720
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.6742413640022278,
      "learning_rate": 0.0005117839305380139,
      "loss": 2.3618,
      "step": 57721
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7375082969665527,
      "learning_rate": 0.0005117810333153845,
      "loss": 2.2659,
      "step": 57722
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7807518243789673,
      "learning_rate": 0.0005117781360533811,
      "loss": 2.3503,
      "step": 57723
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7270951867103577,
      "learning_rate": 0.000511775238752004,
      "loss": 2.118,
      "step": 57724
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.6784505248069763,
      "learning_rate": 0.0005117723414112538,
      "loss": 2.1106,
      "step": 57725
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7577596306800842,
      "learning_rate": 0.0005117694440311313,
      "loss": 2.2288,
      "step": 57726
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7809835076332092,
      "learning_rate": 0.0005117665466116369,
      "loss": 2.2285,
      "step": 57727
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7945088148117065,
      "learning_rate": 0.000511763649152771,
      "loss": 2.2676,
      "step": 57728
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.6950054168701172,
      "learning_rate": 0.0005117607516545343,
      "loss": 2.2436,
      "step": 57729
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7964103817939758,
      "learning_rate": 0.0005117578541169272,
      "loss": 2.3635,
      "step": 57730
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.6628919243812561,
      "learning_rate": 0.0005117549565399503,
      "loss": 2.2616,
      "step": 57731
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7626356482505798,
      "learning_rate": 0.0005117520589236043,
      "loss": 1.8871,
      "step": 57732
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.6896299719810486,
      "learning_rate": 0.0005117491612678895,
      "loss": 2.3623,
      "step": 57733
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7118727564811707,
      "learning_rate": 0.0005117462635728067,
      "loss": 2.0604,
      "step": 57734
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.8126210570335388,
      "learning_rate": 0.0005117433658383561,
      "loss": 2.1216,
      "step": 57735
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7497033476829529,
      "learning_rate": 0.0005117404680645383,
      "loss": 2.2009,
      "step": 57736
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7520118355751038,
      "learning_rate": 0.0005117375702513541,
      "loss": 2.4504,
      "step": 57737
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7873390316963196,
      "learning_rate": 0.000511734672398804,
      "loss": 2.2483,
      "step": 57738
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.6982487440109253,
      "learning_rate": 0.0005117317745068883,
      "loss": 2.1272,
      "step": 57739
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7201433777809143,
      "learning_rate": 0.0005117288765756076,
      "loss": 2.2066,
      "step": 57740
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.6747909784317017,
      "learning_rate": 0.0005117259786049626,
      "loss": 2.1635,
      "step": 57741
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.6923441290855408,
      "learning_rate": 0.0005117230805949537,
      "loss": 2.1864,
      "step": 57742
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.8711110353469849,
      "learning_rate": 0.0005117201825455814,
      "loss": 2.0468,
      "step": 57743
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7475144267082214,
      "learning_rate": 0.0005117172844568465,
      "loss": 2.5119,
      "step": 57744
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7279165983200073,
      "learning_rate": 0.0005117143863287493,
      "loss": 2.4265,
      "step": 57745
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.6964941620826721,
      "learning_rate": 0.0005117114881612903,
      "loss": 1.9596,
      "step": 57746
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.6975383758544922,
      "learning_rate": 0.0005117085899544703,
      "loss": 2.3554,
      "step": 57747
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.8377482891082764,
      "learning_rate": 0.0005117056917082896,
      "loss": 2.1959,
      "step": 57748
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.6917614340782166,
      "learning_rate": 0.0005117027934227488,
      "loss": 2.1728,
      "step": 57749
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.6920607089996338,
      "learning_rate": 0.0005116998950978485,
      "loss": 2.131,
      "step": 57750
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.8577035069465637,
      "learning_rate": 0.0005116969967335892,
      "loss": 2.0882,
      "step": 57751
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.757266640663147,
      "learning_rate": 0.0005116940983299714,
      "loss": 2.1173,
      "step": 57752
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7101345062255859,
      "learning_rate": 0.0005116911998869956,
      "loss": 2.1864,
      "step": 57753
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7207369208335876,
      "learning_rate": 0.0005116883014046625,
      "loss": 2.4701,
      "step": 57754
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7254704236984253,
      "learning_rate": 0.0005116854028829725,
      "loss": 2.4688,
      "step": 57755
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.704362154006958,
      "learning_rate": 0.0005116825043219262,
      "loss": 2.0618,
      "step": 57756
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7011570334434509,
      "learning_rate": 0.0005116796057215242,
      "loss": 2.2917,
      "step": 57757
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.6891698837280273,
      "learning_rate": 0.0005116767070817669,
      "loss": 2.2269,
      "step": 57758
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.8715806007385254,
      "learning_rate": 0.0005116738084026549,
      "loss": 2.0737,
      "step": 57759
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7835506796836853,
      "learning_rate": 0.0005116709096841887,
      "loss": 1.9731,
      "step": 57760
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7425649762153625,
      "learning_rate": 0.0005116680109263689,
      "loss": 2.0198,
      "step": 57761
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7996712327003479,
      "learning_rate": 0.000511665112129196,
      "loss": 2.1944,
      "step": 57762
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7903391122817993,
      "learning_rate": 0.0005116622132926707,
      "loss": 2.1711,
      "step": 57763
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7311297059059143,
      "learning_rate": 0.0005116593144167933,
      "loss": 2.0295,
      "step": 57764
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7429946660995483,
      "learning_rate": 0.0005116564155015645,
      "loss": 2.2609,
      "step": 57765
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7788742184638977,
      "learning_rate": 0.0005116535165469848,
      "loss": 2.1235,
      "step": 57766
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7986965775489807,
      "learning_rate": 0.0005116506175530546,
      "loss": 2.1124,
      "step": 57767
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.8284637331962585,
      "learning_rate": 0.0005116477185197746,
      "loss": 2.2403,
      "step": 57768
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.8073305487632751,
      "learning_rate": 0.0005116448194471453,
      "loss": 2.2129,
      "step": 57769
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7362361550331116,
      "learning_rate": 0.0005116419203351672,
      "loss": 2.3002,
      "step": 57770
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7645334005355835,
      "learning_rate": 0.0005116390211838409,
      "loss": 2.1635,
      "step": 57771
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7075009942054749,
      "learning_rate": 0.000511636121993167,
      "loss": 2.1885,
      "step": 57772
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.8826961517333984,
      "learning_rate": 0.0005116332227631458,
      "loss": 2.4613,
      "step": 57773
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.6765221357345581,
      "learning_rate": 0.0005116303234937781,
      "loss": 2.3661,
      "step": 57774
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.745193600654602,
      "learning_rate": 0.0005116274241850642,
      "loss": 2.098,
      "step": 57775
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7768890857696533,
      "learning_rate": 0.0005116245248370048,
      "loss": 2.1111,
      "step": 57776
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7366443872451782,
      "learning_rate": 0.0005116216254496005,
      "loss": 2.1732,
      "step": 57777
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.708556592464447,
      "learning_rate": 0.0005116187260228517,
      "loss": 2.2467,
      "step": 57778
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7507292032241821,
      "learning_rate": 0.000511615826556759,
      "loss": 2.3735,
      "step": 57779
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7051146626472473,
      "learning_rate": 0.0005116129270513229,
      "loss": 2.0778,
      "step": 57780
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.8124725818634033,
      "learning_rate": 0.0005116100275065439,
      "loss": 2.2108,
      "step": 57781
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7858258485794067,
      "learning_rate": 0.0005116071279224226,
      "loss": 2.1605,
      "step": 57782
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.8395915031433105,
      "learning_rate": 0.0005116042282989597,
      "loss": 2.2714,
      "step": 57783
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7060158252716064,
      "learning_rate": 0.0005116013286361555,
      "loss": 1.8868,
      "step": 57784
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.8428030014038086,
      "learning_rate": 0.0005115984289340106,
      "loss": 2.4451,
      "step": 57785
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7698498964309692,
      "learning_rate": 0.0005115955291925255,
      "loss": 2.1839,
      "step": 57786
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.8259207606315613,
      "learning_rate": 0.000511592629411701,
      "loss": 2.2944,
      "step": 57787
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.78872150182724,
      "learning_rate": 0.0005115897295915373,
      "loss": 2.0379,
      "step": 57788
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7158737778663635,
      "learning_rate": 0.000511586829732035,
      "loss": 2.1214,
      "step": 57789
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7698379755020142,
      "learning_rate": 0.0005115839298331948,
      "loss": 2.2355,
      "step": 57790
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.77974534034729,
      "learning_rate": 0.0005115810298950171,
      "loss": 2.2803,
      "step": 57791
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.765767514705658,
      "learning_rate": 0.0005115781299175026,
      "loss": 2.0101,
      "step": 57792
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.68445885181427,
      "learning_rate": 0.0005115752299006517,
      "loss": 2.0942,
      "step": 57793
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7601078152656555,
      "learning_rate": 0.0005115723298444649,
      "loss": 1.9635,
      "step": 57794
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.8962042331695557,
      "learning_rate": 0.0005115694297489428,
      "loss": 2.2452,
      "step": 57795
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.8261599540710449,
      "learning_rate": 0.0005115665296140861,
      "loss": 2.2843,
      "step": 57796
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.8927745223045349,
      "learning_rate": 0.000511563629439895,
      "loss": 2.1676,
      "step": 57797
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.744296669960022,
      "learning_rate": 0.0005115607292263704,
      "loss": 2.1937,
      "step": 57798
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.8058292865753174,
      "learning_rate": 0.0005115578289735125,
      "loss": 2.2441,
      "step": 57799
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7943195700645447,
      "learning_rate": 0.0005115549286813221,
      "loss": 2.3361,
      "step": 57800
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7242876887321472,
      "learning_rate": 0.0005115520283497996,
      "loss": 2.0742,
      "step": 57801
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7102958559989929,
      "learning_rate": 0.0005115491279789456,
      "loss": 2.2788,
      "step": 57802
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.8075018525123596,
      "learning_rate": 0.0005115462275687607,
      "loss": 2.2685,
      "step": 57803
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.746826708316803,
      "learning_rate": 0.0005115433271192453,
      "loss": 2.298,
      "step": 57804
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7194612622261047,
      "learning_rate": 0.0005115404266304,
      "loss": 2.247,
      "step": 57805
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.6965299844741821,
      "learning_rate": 0.0005115375261022253,
      "loss": 2.4865,
      "step": 57806
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7538798451423645,
      "learning_rate": 0.0005115346255347218,
      "loss": 2.5455,
      "step": 57807
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.8116873502731323,
      "learning_rate": 0.0005115317249278901,
      "loss": 2.1889,
      "step": 57808
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7430474162101746,
      "learning_rate": 0.0005115288242817306,
      "loss": 2.1106,
      "step": 57809
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7805390954017639,
      "learning_rate": 0.0005115259235962438,
      "loss": 2.2037,
      "step": 57810
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7723032236099243,
      "learning_rate": 0.0005115230228714305,
      "loss": 2.4109,
      "step": 57811
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7049354314804077,
      "learning_rate": 0.0005115201221072911,
      "loss": 2.1451,
      "step": 57812
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7060181498527527,
      "learning_rate": 0.0005115172213038259,
      "loss": 2.2919,
      "step": 57813
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.9979307055473328,
      "learning_rate": 0.0005115143204610359,
      "loss": 2.0141,
      "step": 57814
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7845569849014282,
      "learning_rate": 0.0005115114195789213,
      "loss": 2.5042,
      "step": 57815
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7001528739929199,
      "learning_rate": 0.0005115085186574827,
      "loss": 2.4084,
      "step": 57816
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7156215906143188,
      "learning_rate": 0.0005115056176967207,
      "loss": 2.1697,
      "step": 57817
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.718442440032959,
      "learning_rate": 0.0005115027166966358,
      "loss": 2.2752,
      "step": 57818
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7806732654571533,
      "learning_rate": 0.0005114998156572285,
      "loss": 2.1027,
      "step": 57819
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7334596514701843,
      "learning_rate": 0.0005114969145784995,
      "loss": 2.1754,
      "step": 57820
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7870234251022339,
      "learning_rate": 0.0005114940134604492,
      "loss": 2.1031,
      "step": 57821
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7074207663536072,
      "learning_rate": 0.0005114911123030781,
      "loss": 2.0416,
      "step": 57822
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7983989715576172,
      "learning_rate": 0.0005114882111063869,
      "loss": 2.3866,
      "step": 57823
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.6890310049057007,
      "learning_rate": 0.000511485309870376,
      "loss": 2.3839,
      "step": 57824
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7091423869132996,
      "learning_rate": 0.000511482408595046,
      "loss": 2.2489,
      "step": 57825
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7720927596092224,
      "learning_rate": 0.0005114795072803974,
      "loss": 2.4196,
      "step": 57826
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7451316118240356,
      "learning_rate": 0.0005114766059264308,
      "loss": 2.2626,
      "step": 57827
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.707515299320221,
      "learning_rate": 0.0005114737045331467,
      "loss": 2.0792,
      "step": 57828
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.6815851926803589,
      "learning_rate": 0.0005114708031005457,
      "loss": 2.275,
      "step": 57829
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7388870120048523,
      "learning_rate": 0.0005114679016286281,
      "loss": 2.4091,
      "step": 57830
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.791820764541626,
      "learning_rate": 0.0005114650001173949,
      "loss": 2.3209,
      "step": 57831
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7247000932693481,
      "learning_rate": 0.0005114620985668462,
      "loss": 2.2495,
      "step": 57832
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.77837735414505,
      "learning_rate": 0.0005114591969769828,
      "loss": 2.2529,
      "step": 57833
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.687687873840332,
      "learning_rate": 0.000511456295347805,
      "loss": 2.2114,
      "step": 57834
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7109763622283936,
      "learning_rate": 0.0005114533936793136,
      "loss": 2.3513,
      "step": 57835
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7375330328941345,
      "learning_rate": 0.000511450491971509,
      "loss": 2.2347,
      "step": 57836
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.8055820465087891,
      "learning_rate": 0.0005114475902243917,
      "loss": 2.4237,
      "step": 57837
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.8113582134246826,
      "learning_rate": 0.0005114446884379623,
      "loss": 2.0924,
      "step": 57838
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7553385496139526,
      "learning_rate": 0.0005114417866122215,
      "loss": 2.5269,
      "step": 57839
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.743126630783081,
      "learning_rate": 0.0005114388847471695,
      "loss": 2.2512,
      "step": 57840
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.777977705001831,
      "learning_rate": 0.0005114359828428071,
      "loss": 2.2055,
      "step": 57841
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.8329878449440002,
      "learning_rate": 0.0005114330808991348,
      "loss": 2.1936,
      "step": 57842
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7441913485527039,
      "learning_rate": 0.000511430178916153,
      "loss": 2.2694,
      "step": 57843
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.784745454788208,
      "learning_rate": 0.0005114272768938624,
      "loss": 2.4823,
      "step": 57844
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.825243353843689,
      "learning_rate": 0.0005114243748322634,
      "loss": 2.074,
      "step": 57845
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.817915141582489,
      "learning_rate": 0.0005114214727313567,
      "loss": 2.0556,
      "step": 57846
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7821858525276184,
      "learning_rate": 0.0005114185705911427,
      "loss": 2.3628,
      "step": 57847
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7090701460838318,
      "learning_rate": 0.0005114156684116221,
      "loss": 2.4773,
      "step": 57848
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7603366374969482,
      "learning_rate": 0.0005114127661927952,
      "loss": 2.2183,
      "step": 57849
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7319178581237793,
      "learning_rate": 0.0005114098639346627,
      "loss": 2.4692,
      "step": 57850
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.6843101382255554,
      "learning_rate": 0.0005114069616372253,
      "loss": 2.1756,
      "step": 57851
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.8116160035133362,
      "learning_rate": 0.0005114040593004832,
      "loss": 2.0761,
      "step": 57852
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.8110765814781189,
      "learning_rate": 0.0005114011569244371,
      "loss": 2.309,
      "step": 57853
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7984944581985474,
      "learning_rate": 0.0005113982545090876,
      "loss": 2.1763,
      "step": 57854
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.8897969126701355,
      "learning_rate": 0.0005113953520544351,
      "loss": 2.3226,
      "step": 57855
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.6827236413955688,
      "learning_rate": 0.0005113924495604803,
      "loss": 2.1587,
      "step": 57856
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7205884456634521,
      "learning_rate": 0.0005113895470272235,
      "loss": 2.2994,
      "step": 57857
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.8418291807174683,
      "learning_rate": 0.0005113866444546655,
      "loss": 2.2655,
      "step": 57858
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.6664842367172241,
      "learning_rate": 0.0005113837418428067,
      "loss": 2.3167,
      "step": 57859
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.6398746967315674,
      "learning_rate": 0.0005113808391916477,
      "loss": 2.042,
      "step": 57860
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7635601162910461,
      "learning_rate": 0.0005113779365011889,
      "loss": 2.1499,
      "step": 57861
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.8823133707046509,
      "learning_rate": 0.0005113750337714311,
      "loss": 2.3052,
      "step": 57862
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.6764118671417236,
      "learning_rate": 0.0005113721310023746,
      "loss": 2.1682,
      "step": 57863
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7025126814842224,
      "learning_rate": 0.0005113692281940201,
      "loss": 2.1792,
      "step": 57864
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7199522852897644,
      "learning_rate": 0.0005113663253463681,
      "loss": 2.3374,
      "step": 57865
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.8404206037521362,
      "learning_rate": 0.000511363422459419,
      "loss": 2.1288,
      "step": 57866
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7801717519760132,
      "learning_rate": 0.0005113605195331735,
      "loss": 2.3392,
      "step": 57867
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.905617356300354,
      "learning_rate": 0.0005113576165676322,
      "loss": 2.2117,
      "step": 57868
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.687065839767456,
      "learning_rate": 0.0005113547135627953,
      "loss": 2.0727,
      "step": 57869
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7986461520195007,
      "learning_rate": 0.0005113518105186637,
      "loss": 2.2533,
      "step": 57870
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.806336522102356,
      "learning_rate": 0.0005113489074352377,
      "loss": 2.308,
      "step": 57871
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.8104532361030579,
      "learning_rate": 0.0005113460043125182,
      "loss": 2.103,
      "step": 57872
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7594460844993591,
      "learning_rate": 0.0005113431011505053,
      "loss": 2.4706,
      "step": 57873
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7896266579627991,
      "learning_rate": 0.0005113401979491998,
      "loss": 2.2469,
      "step": 57874
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7699024081230164,
      "learning_rate": 0.0005113372947086022,
      "loss": 2.2566,
      "step": 57875
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7116979360580444,
      "learning_rate": 0.0005113343914287129,
      "loss": 2.2504,
      "step": 57876
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.9425905346870422,
      "learning_rate": 0.0005113314881095326,
      "loss": 2.2997,
      "step": 57877
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7596055865287781,
      "learning_rate": 0.0005113285847510617,
      "loss": 2.3039,
      "step": 57878
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7348288297653198,
      "learning_rate": 0.0005113256813533009,
      "loss": 2.3668,
      "step": 57879
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.6841092705726624,
      "learning_rate": 0.0005113227779162507,
      "loss": 2.195,
      "step": 57880
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.8063613176345825,
      "learning_rate": 0.0005113198744399115,
      "loss": 2.4761,
      "step": 57881
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.8215617537498474,
      "learning_rate": 0.0005113169709242841,
      "loss": 2.3036,
      "step": 57882
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.781934916973114,
      "learning_rate": 0.0005113140673693688,
      "loss": 2.2757,
      "step": 57883
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.8682935237884521,
      "learning_rate": 0.0005113111637751663,
      "loss": 2.246,
      "step": 57884
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.8989427089691162,
      "learning_rate": 0.0005113082601416769,
      "loss": 2.2112,
      "step": 57885
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7488553524017334,
      "learning_rate": 0.0005113053564689015,
      "loss": 2.2258,
      "step": 57886
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.826458215713501,
      "learning_rate": 0.0005113024527568403,
      "loss": 2.4066,
      "step": 57887
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.8862086534500122,
      "learning_rate": 0.000511299549005494,
      "loss": 2.2191,
      "step": 57888
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7173421382904053,
      "learning_rate": 0.0005112966452148632,
      "loss": 2.1674,
      "step": 57889
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.8600257635116577,
      "learning_rate": 0.0005112937413849483,
      "loss": 2.3489,
      "step": 57890
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7505260705947876,
      "learning_rate": 0.00051129083751575,
      "loss": 2.397,
      "step": 57891
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7963030934333801,
      "learning_rate": 0.0005112879336072686,
      "loss": 2.1194,
      "step": 57892
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7596186399459839,
      "learning_rate": 0.0005112850296595049,
      "loss": 2.3035,
      "step": 57893
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7437127828598022,
      "learning_rate": 0.0005112821256724593,
      "loss": 2.2207,
      "step": 57894
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7307325601577759,
      "learning_rate": 0.0005112792216461323,
      "loss": 2.6471,
      "step": 57895
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.9159920811653137,
      "learning_rate": 0.0005112763175805246,
      "loss": 2.3861,
      "step": 57896
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.6642516851425171,
      "learning_rate": 0.0005112734134756366,
      "loss": 2.1434,
      "step": 57897
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.8087356090545654,
      "learning_rate": 0.0005112705093314688,
      "loss": 2.2878,
      "step": 57898
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7914381623268127,
      "learning_rate": 0.0005112676051480219,
      "loss": 2.2046,
      "step": 57899
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7712395191192627,
      "learning_rate": 0.0005112647009252964,
      "loss": 2.185,
      "step": 57900
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.8853926658630371,
      "learning_rate": 0.0005112617966632928,
      "loss": 2.3453,
      "step": 57901
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.639741837978363,
      "learning_rate": 0.0005112588923620117,
      "loss": 2.2684,
      "step": 57902
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7449925541877747,
      "learning_rate": 0.0005112559880214535,
      "loss": 2.0289,
      "step": 57903
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7444836497306824,
      "learning_rate": 0.0005112530836416188,
      "loss": 2.0371,
      "step": 57904
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.6990830898284912,
      "learning_rate": 0.0005112501792225081,
      "loss": 2.2041,
      "step": 57905
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7824432253837585,
      "learning_rate": 0.0005112472747641222,
      "loss": 2.2235,
      "step": 57906
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.6966951489448547,
      "learning_rate": 0.0005112443702664613,
      "loss": 2.1125,
      "step": 57907
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7508643865585327,
      "learning_rate": 0.0005112414657295261,
      "loss": 2.1952,
      "step": 57908
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7668547630310059,
      "learning_rate": 0.0005112385611533172,
      "loss": 2.0783,
      "step": 57909
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7241323590278625,
      "learning_rate": 0.000511235656537835,
      "loss": 2.4653,
      "step": 57910
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.6874990463256836,
      "learning_rate": 0.0005112327518830801,
      "loss": 2.3079,
      "step": 57911
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7413828372955322,
      "learning_rate": 0.000511229847189053,
      "loss": 2.0634,
      "step": 57912
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7283298969268799,
      "learning_rate": 0.0005112269424557544,
      "loss": 2.3778,
      "step": 57913
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7146501541137695,
      "learning_rate": 0.0005112240376831846,
      "loss": 2.1496,
      "step": 57914
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.6964085102081299,
      "learning_rate": 0.0005112211328713444,
      "loss": 2.2628,
      "step": 57915
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.6806009411811829,
      "learning_rate": 0.0005112182280202341,
      "loss": 2.0288,
      "step": 57916
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7635653018951416,
      "learning_rate": 0.0005112153231298544,
      "loss": 2.2834,
      "step": 57917
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.6539117693901062,
      "learning_rate": 0.0005112124182002057,
      "loss": 2.19,
      "step": 57918
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7603663802146912,
      "learning_rate": 0.0005112095132312887,
      "loss": 2.1043,
      "step": 57919
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.6898345947265625,
      "learning_rate": 0.0005112066082231039,
      "loss": 2.1312,
      "step": 57920
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7183433771133423,
      "learning_rate": 0.0005112037031756518,
      "loss": 2.3033,
      "step": 57921
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7507703304290771,
      "learning_rate": 0.0005112007980889329,
      "loss": 2.178,
      "step": 57922
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7571590542793274,
      "learning_rate": 0.0005111978929629477,
      "loss": 2.3196,
      "step": 57923
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7534642219543457,
      "learning_rate": 0.000511194987797697,
      "loss": 2.3136,
      "step": 57924
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7276656031608582,
      "learning_rate": 0.000511192082593181,
      "loss": 1.9945,
      "step": 57925
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.8105528950691223,
      "learning_rate": 0.0005111891773494006,
      "loss": 2.2418,
      "step": 57926
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7652497291564941,
      "learning_rate": 0.000511186272066356,
      "loss": 2.2166,
      "step": 57927
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.8070257902145386,
      "learning_rate": 0.0005111833667440479,
      "loss": 2.2893,
      "step": 57928
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7012683749198914,
      "learning_rate": 0.0005111804613824768,
      "loss": 2.0849,
      "step": 57929
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.8220164179801941,
      "learning_rate": 0.0005111775559816433,
      "loss": 2.3872,
      "step": 57930
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7377141118049622,
      "learning_rate": 0.0005111746505415479,
      "loss": 2.1882,
      "step": 57931
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.764359176158905,
      "learning_rate": 0.0005111717450621912,
      "loss": 2.1632,
      "step": 57932
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7737970352172852,
      "learning_rate": 0.0005111688395435737,
      "loss": 2.242,
      "step": 57933
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7396594882011414,
      "learning_rate": 0.0005111659339856959,
      "loss": 2.4102,
      "step": 57934
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.8796466588973999,
      "learning_rate": 0.0005111630283885582,
      "loss": 2.392,
      "step": 57935
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.757114052772522,
      "learning_rate": 0.0005111601227521615,
      "loss": 2.3709,
      "step": 57936
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.8257396817207336,
      "learning_rate": 0.0005111572170765061,
      "loss": 2.2371,
      "step": 57937
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7020543217658997,
      "learning_rate": 0.0005111543113615926,
      "loss": 2.4692,
      "step": 57938
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7676928043365479,
      "learning_rate": 0.0005111514056074215,
      "loss": 2.2749,
      "step": 57939
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.67784184217453,
      "learning_rate": 0.0005111484998139934,
      "loss": 2.0579,
      "step": 57940
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.6764508485794067,
      "learning_rate": 0.0005111455939813088,
      "loss": 1.998,
      "step": 57941
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7215511798858643,
      "learning_rate": 0.0005111426881093682,
      "loss": 2.1785,
      "step": 57942
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.733415961265564,
      "learning_rate": 0.0005111397821981722,
      "loss": 2.0094,
      "step": 57943
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7351756691932678,
      "learning_rate": 0.0005111368762477213,
      "loss": 2.2917,
      "step": 57944
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7326687574386597,
      "learning_rate": 0.0005111339702580163,
      "loss": 2.0562,
      "step": 57945
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7776450514793396,
      "learning_rate": 0.0005111310642290572,
      "loss": 2.1414,
      "step": 57946
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.6820467114448547,
      "learning_rate": 0.000511128158160845,
      "loss": 2.0928,
      "step": 57947
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7667244672775269,
      "learning_rate": 0.0005111252520533801,
      "loss": 2.1015,
      "step": 57948
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7694720029830933,
      "learning_rate": 0.000511122345906663,
      "loss": 2.2805,
      "step": 57949
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7802996635437012,
      "learning_rate": 0.0005111194397206943,
      "loss": 2.3105,
      "step": 57950
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7438024878501892,
      "learning_rate": 0.0005111165334954744,
      "loss": 2.3615,
      "step": 57951
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7447928786277771,
      "learning_rate": 0.000511113627231004,
      "loss": 1.9861,
      "step": 57952
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.8459942936897278,
      "learning_rate": 0.0005111107209272836,
      "loss": 2.1106,
      "step": 57953
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7289441227912903,
      "learning_rate": 0.0005111078145843137,
      "loss": 2.2896,
      "step": 57954
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7377627491950989,
      "learning_rate": 0.0005111049082020951,
      "loss": 2.4101,
      "step": 57955
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7443806529045105,
      "learning_rate": 0.000511102001780628,
      "loss": 2.3139,
      "step": 57956
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.849032998085022,
      "learning_rate": 0.0005110990953199129,
      "loss": 2.3377,
      "step": 57957
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7569312453269958,
      "learning_rate": 0.0005110961888199506,
      "loss": 2.2995,
      "step": 57958
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.8421024680137634,
      "learning_rate": 0.0005110932822807415,
      "loss": 2.2825,
      "step": 57959
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.8596370816230774,
      "learning_rate": 0.0005110903757022862,
      "loss": 2.481,
      "step": 57960
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.0633752346038818,
      "learning_rate": 0.0005110874690845852,
      "loss": 2.1435,
      "step": 57961
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7527288198471069,
      "learning_rate": 0.000511084562427639,
      "loss": 2.0969,
      "step": 57962
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.1732778549194336,
      "learning_rate": 0.0005110816557314483,
      "loss": 2.2919,
      "step": 57963
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.9151092767715454,
      "learning_rate": 0.0005110787489960135,
      "loss": 2.2648,
      "step": 57964
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7942610383033752,
      "learning_rate": 0.0005110758422213351,
      "loss": 2.1686,
      "step": 57965
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.8088043928146362,
      "learning_rate": 0.0005110729354074139,
      "loss": 2.3369,
      "step": 57966
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.781497597694397,
      "learning_rate": 0.0005110700285542501,
      "loss": 2.3106,
      "step": 57967
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.903724193572998,
      "learning_rate": 0.0005110671216618446,
      "loss": 2.2465,
      "step": 57968
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.9166911244392395,
      "learning_rate": 0.0005110642147301975,
      "loss": 2.1764,
      "step": 57969
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.8833695650100708,
      "learning_rate": 0.0005110613077593096,
      "loss": 2.1372,
      "step": 57970
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.9456117749214172,
      "learning_rate": 0.0005110584007491816,
      "loss": 2.1808,
      "step": 57971
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7459753155708313,
      "learning_rate": 0.0005110554936998137,
      "loss": 2.0038,
      "step": 57972
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7937206029891968,
      "learning_rate": 0.0005110525866112067,
      "loss": 2.2439,
      "step": 57973
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.1473338603973389,
      "learning_rate": 0.0005110496794833608,
      "loss": 1.915,
      "step": 57974
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7127835750579834,
      "learning_rate": 0.000511046772316277,
      "loss": 2.3005,
      "step": 57975
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.862628161907196,
      "learning_rate": 0.0005110438651099555,
      "loss": 2.1534,
      "step": 57976
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7948517203330994,
      "learning_rate": 0.0005110409578643971,
      "loss": 2.2031,
      "step": 57977
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7929841876029968,
      "learning_rate": 0.0005110380505796021,
      "loss": 2.2392,
      "step": 57978
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7613598704338074,
      "learning_rate": 0.0005110351432555712,
      "loss": 2.3707,
      "step": 57979
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7749707698822021,
      "learning_rate": 0.000511032235892305,
      "loss": 2.0814,
      "step": 57980
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7167643904685974,
      "learning_rate": 0.0005110293284898037,
      "loss": 2.2034,
      "step": 57981
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.9127212762832642,
      "learning_rate": 0.0005110264210480682,
      "loss": 2.199,
      "step": 57982
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.8945252299308777,
      "learning_rate": 0.0005110235135670989,
      "loss": 2.3342,
      "step": 57983
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.877755880355835,
      "learning_rate": 0.0005110206060468963,
      "loss": 2.1673,
      "step": 57984
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.823468029499054,
      "learning_rate": 0.0005110176984874609,
      "loss": 2.1764,
      "step": 57985
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.8055535554885864,
      "learning_rate": 0.0005110147908887935,
      "loss": 2.2321,
      "step": 57986
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7559781074523926,
      "learning_rate": 0.0005110118832508944,
      "loss": 2.258,
      "step": 57987
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.8376436829566956,
      "learning_rate": 0.0005110089755737642,
      "loss": 2.1798,
      "step": 57988
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.8173884749412537,
      "learning_rate": 0.0005110060678574034,
      "loss": 2.3129,
      "step": 57989
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7249855399131775,
      "learning_rate": 0.0005110031601018127,
      "loss": 2.334,
      "step": 57990
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.8320547938346863,
      "learning_rate": 0.0005110002523069924,
      "loss": 2.3726,
      "step": 57991
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7519962191581726,
      "learning_rate": 0.0005109973444729432,
      "loss": 2.3899,
      "step": 57992
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7571046352386475,
      "learning_rate": 0.0005109944365996657,
      "loss": 2.2663,
      "step": 57993
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7480888962745667,
      "learning_rate": 0.0005109915286871603,
      "loss": 2.3109,
      "step": 57994
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7487547993659973,
      "learning_rate": 0.0005109886207354276,
      "loss": 2.3435,
      "step": 57995
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7659046053886414,
      "learning_rate": 0.0005109857127444681,
      "loss": 2.0862,
      "step": 57996
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7273465991020203,
      "learning_rate": 0.0005109828047142824,
      "loss": 2.1626,
      "step": 57997
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7129753828048706,
      "learning_rate": 0.000510979896644871,
      "loss": 2.4722,
      "step": 57998
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7078148722648621,
      "learning_rate": 0.0005109769885362344,
      "loss": 2.3771,
      "step": 57999
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7817956805229187,
      "learning_rate": 0.0005109740803883732,
      "loss": 2.3827,
      "step": 58000
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7160720229148865,
      "learning_rate": 0.0005109711722012881,
      "loss": 2.0882,
      "step": 58001
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.6689626574516296,
      "learning_rate": 0.0005109682639749793,
      "loss": 2.2843,
      "step": 58002
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7577555775642395,
      "learning_rate": 0.0005109653557094477,
      "loss": 2.422,
      "step": 58003
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7852551937103271,
      "learning_rate": 0.0005109624474046935,
      "loss": 2.3241,
      "step": 58004
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.6814209222793579,
      "learning_rate": 0.0005109595390607175,
      "loss": 2.2417,
      "step": 58005
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7037058472633362,
      "learning_rate": 0.0005109566306775201,
      "loss": 2.0222,
      "step": 58006
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7977197170257568,
      "learning_rate": 0.0005109537222551018,
      "loss": 2.0446,
      "step": 58007
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7571263909339905,
      "learning_rate": 0.0005109508137934632,
      "loss": 2.1525,
      "step": 58008
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7552587985992432,
      "learning_rate": 0.0005109479052926051,
      "loss": 2.1649,
      "step": 58009
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.9419842958450317,
      "learning_rate": 0.0005109449967525276,
      "loss": 2.345,
      "step": 58010
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.8071591258049011,
      "learning_rate": 0.0005109420881732314,
      "loss": 2.3648,
      "step": 58011
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7432840466499329,
      "learning_rate": 0.0005109391795547173,
      "loss": 1.9638,
      "step": 58012
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7041019201278687,
      "learning_rate": 0.0005109362708969854,
      "loss": 2.0803,
      "step": 58013
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7499566078186035,
      "learning_rate": 0.0005109333622000367,
      "loss": 2.3337,
      "step": 58014
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7012366652488708,
      "learning_rate": 0.0005109304534638713,
      "loss": 2.139,
      "step": 58015
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.6840686798095703,
      "learning_rate": 0.0005109275446884899,
      "loss": 2.2497,
      "step": 58016
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7868279218673706,
      "learning_rate": 0.0005109246358738933,
      "loss": 2.1904,
      "step": 58017
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7603369951248169,
      "learning_rate": 0.0005109217270200817,
      "loss": 2.328,
      "step": 58018
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7331311702728271,
      "learning_rate": 0.0005109188181270559,
      "loss": 2.1338,
      "step": 58019
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.8016307353973389,
      "learning_rate": 0.0005109159091948161,
      "loss": 2.0723,
      "step": 58020
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7960140109062195,
      "learning_rate": 0.0005109130002233632,
      "loss": 2.2196,
      "step": 58021
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.6926124691963196,
      "learning_rate": 0.0005109100912126976,
      "loss": 2.0049,
      "step": 58022
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.6872571706771851,
      "learning_rate": 0.0005109071821628198,
      "loss": 2.2304,
      "step": 58023
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.770596444606781,
      "learning_rate": 0.0005109042730737304,
      "loss": 2.2261,
      "step": 58024
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7564749121665955,
      "learning_rate": 0.0005109013639454298,
      "loss": 2.2927,
      "step": 58025
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7524045705795288,
      "learning_rate": 0.0005108984547779187,
      "loss": 2.1259,
      "step": 58026
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.9301421046257019,
      "learning_rate": 0.0005108955455711977,
      "loss": 2.1328,
      "step": 58027
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7322529554367065,
      "learning_rate": 0.0005108926363252671,
      "loss": 2.1828,
      "step": 58028
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7860919237136841,
      "learning_rate": 0.0005108897270401277,
      "loss": 2.1147,
      "step": 58029
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7457679510116577,
      "learning_rate": 0.0005108868177157799,
      "loss": 2.1718,
      "step": 58030
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7691558599472046,
      "learning_rate": 0.0005108839083522242,
      "loss": 2.1719,
      "step": 58031
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7429739832878113,
      "learning_rate": 0.0005108809989494611,
      "loss": 2.3449,
      "step": 58032
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7016228437423706,
      "learning_rate": 0.0005108780895074915,
      "loss": 2.4034,
      "step": 58033
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.6995283365249634,
      "learning_rate": 0.0005108751800263156,
      "loss": 2.0768,
      "step": 58034
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7911772131919861,
      "learning_rate": 0.000510872270505934,
      "loss": 2.3481,
      "step": 58035
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.728714644908905,
      "learning_rate": 0.0005108693609463472,
      "loss": 2.0443,
      "step": 58036
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.793061375617981,
      "learning_rate": 0.0005108664513475558,
      "loss": 2.2927,
      "step": 58037
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7115577459335327,
      "learning_rate": 0.0005108635417095604,
      "loss": 2.1425,
      "step": 58038
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.813616156578064,
      "learning_rate": 0.0005108606320323615,
      "loss": 1.9462,
      "step": 58039
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7435017228126526,
      "learning_rate": 0.0005108577223159596,
      "loss": 2.2305,
      "step": 58040
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7502511143684387,
      "learning_rate": 0.0005108548125603553,
      "loss": 2.0327,
      "step": 58041
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7285827398300171,
      "learning_rate": 0.000510851902765549,
      "loss": 2.2785,
      "step": 58042
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7717424631118774,
      "learning_rate": 0.0005108489929315415,
      "loss": 2.2802,
      "step": 58043
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.8202091455459595,
      "learning_rate": 0.0005108460830583331,
      "loss": 2.0451,
      "step": 58044
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7368162274360657,
      "learning_rate": 0.0005108431731459244,
      "loss": 2.2275,
      "step": 58045
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7748457789421082,
      "learning_rate": 0.0005108402631943161,
      "loss": 1.9697,
      "step": 58046
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7619972825050354,
      "learning_rate": 0.0005108373532035085,
      "loss": 2.3177,
      "step": 58047
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.924360454082489,
      "learning_rate": 0.0005108344431735023,
      "loss": 2.2077,
      "step": 58048
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7483142614364624,
      "learning_rate": 0.000510831533104298,
      "loss": 2.1503,
      "step": 58049
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.8025392293930054,
      "learning_rate": 0.0005108286229958961,
      "loss": 2.6002,
      "step": 58050
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7675383687019348,
      "learning_rate": 0.0005108257128482972,
      "loss": 2.5649,
      "step": 58051
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7415937781333923,
      "learning_rate": 0.000510822802661502,
      "loss": 2.1374,
      "step": 58052
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7374006509780884,
      "learning_rate": 0.0005108198924355107,
      "loss": 2.2945,
      "step": 58053
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.8213503956794739,
      "learning_rate": 0.000510816982170324,
      "loss": 2.3066,
      "step": 58054
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.6858391761779785,
      "learning_rate": 0.0005108140718659424,
      "loss": 1.9786,
      "step": 58055
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7080220580101013,
      "learning_rate": 0.0005108111615223666,
      "loss": 2.2782,
      "step": 58056
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7051026821136475,
      "learning_rate": 0.0005108082511395969,
      "loss": 2.2106,
      "step": 58057
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7173992991447449,
      "learning_rate": 0.000510805340717634,
      "loss": 2.3916,
      "step": 58058
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7134960293769836,
      "learning_rate": 0.0005108024302564785,
      "loss": 2.3221,
      "step": 58059
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.8593881130218506,
      "learning_rate": 0.0005107995197561308,
      "loss": 2.4313,
      "step": 58060
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7920823097229004,
      "learning_rate": 0.0005107966092165914,
      "loss": 2.2855,
      "step": 58061
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7203542590141296,
      "learning_rate": 0.000510793698637861,
      "loss": 1.9929,
      "step": 58062
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7963685989379883,
      "learning_rate": 0.0005107907880199401,
      "loss": 2.0859,
      "step": 58063
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.6497257351875305,
      "learning_rate": 0.0005107878773628293,
      "loss": 2.0987,
      "step": 58064
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.6918863654136658,
      "learning_rate": 0.0005107849666665289,
      "loss": 2.1192,
      "step": 58065
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.9987881183624268,
      "learning_rate": 0.0005107820559310396,
      "loss": 2.131,
      "step": 58066
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7481147646903992,
      "learning_rate": 0.000510779145156362,
      "loss": 2.3279,
      "step": 58067
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7801063060760498,
      "learning_rate": 0.0005107762343424965,
      "loss": 2.1986,
      "step": 58068
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.8112673759460449,
      "learning_rate": 0.0005107733234894437,
      "loss": 2.2984,
      "step": 58069
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7928225994110107,
      "learning_rate": 0.0005107704125972044,
      "loss": 2.2192,
      "step": 58070
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.749205470085144,
      "learning_rate": 0.0005107675016657787,
      "loss": 2.2744,
      "step": 58071
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.9384695887565613,
      "learning_rate": 0.0005107645906951673,
      "loss": 2.3102,
      "step": 58072
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.0395982265472412,
      "learning_rate": 0.0005107616796853709,
      "loss": 2.2719,
      "step": 58073
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7214241623878479,
      "learning_rate": 0.0005107587686363899,
      "loss": 2.3362,
      "step": 58074
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.8108453154563904,
      "learning_rate": 0.0005107558575482248,
      "loss": 2.0341,
      "step": 58075
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.9468686580657959,
      "learning_rate": 0.0005107529464208763,
      "loss": 2.3581,
      "step": 58076
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7977173328399658,
      "learning_rate": 0.0005107500352543448,
      "loss": 2.3772,
      "step": 58077
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7698655724525452,
      "learning_rate": 0.0005107471240486309,
      "loss": 2.418,
      "step": 58078
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.8787714242935181,
      "learning_rate": 0.0005107442128037352,
      "loss": 2.0104,
      "step": 58079
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7776414752006531,
      "learning_rate": 0.0005107413015196581,
      "loss": 2.1312,
      "step": 58080
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7329766750335693,
      "learning_rate": 0.0005107383901964002,
      "loss": 1.9882,
      "step": 58081
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7239628434181213,
      "learning_rate": 0.0005107354788339621,
      "loss": 2.3295,
      "step": 58082
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.1169325113296509,
      "learning_rate": 0.0005107325674323442,
      "loss": 2.254,
      "step": 58083
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7497085332870483,
      "learning_rate": 0.0005107296559915472,
      "loss": 2.2633,
      "step": 58084
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7062620520591736,
      "learning_rate": 0.0005107267445115716,
      "loss": 2.4697,
      "step": 58085
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.8272967338562012,
      "learning_rate": 0.0005107238329924179,
      "loss": 2.279,
      "step": 58086
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.9833429455757141,
      "learning_rate": 0.0005107209214340867,
      "loss": 2.5016,
      "step": 58087
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7436720728874207,
      "learning_rate": 0.0005107180098365786,
      "loss": 2.3903,
      "step": 58088
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.9554699659347534,
      "learning_rate": 0.0005107150981998939,
      "loss": 2.1439,
      "step": 58089
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.8666054606437683,
      "learning_rate": 0.0005107121865240333,
      "loss": 2.2903,
      "step": 58090
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7951745390892029,
      "learning_rate": 0.0005107092748089973,
      "loss": 2.3229,
      "step": 58091
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.815002977848053,
      "learning_rate": 0.0005107063630547865,
      "loss": 2.1908,
      "step": 58092
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7448982000350952,
      "learning_rate": 0.0005107034512614014,
      "loss": 2.4128,
      "step": 58093
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7194229364395142,
      "learning_rate": 0.0005107005394288425,
      "loss": 2.2718,
      "step": 58094
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.0973767042160034,
      "learning_rate": 0.0005106976275571105,
      "loss": 2.2903,
      "step": 58095
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.8505250215530396,
      "learning_rate": 0.0005106947156462058,
      "loss": 2.2645,
      "step": 58096
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.6952090859413147,
      "learning_rate": 0.0005106918036961289,
      "loss": 2.2381,
      "step": 58097
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7487373352050781,
      "learning_rate": 0.0005106888917068806,
      "loss": 2.257,
      "step": 58098
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7549459338188171,
      "learning_rate": 0.0005106859796784611,
      "loss": 2.0811,
      "step": 58099
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7486991286277771,
      "learning_rate": 0.0005106830676108711,
      "loss": 2.2086,
      "step": 58100
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7701624631881714,
      "learning_rate": 0.0005106801555041112,
      "loss": 2.4368,
      "step": 58101
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7699337005615234,
      "learning_rate": 0.0005106772433581818,
      "loss": 2.4545,
      "step": 58102
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7945565581321716,
      "learning_rate": 0.0005106743311730836,
      "loss": 2.4367,
      "step": 58103
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.8511881828308105,
      "learning_rate": 0.0005106714189488171,
      "loss": 2.1198,
      "step": 58104
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.8777135610580444,
      "learning_rate": 0.0005106685066853827,
      "loss": 2.3393,
      "step": 58105
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7212179899215698,
      "learning_rate": 0.000510665594382781,
      "loss": 2.1974,
      "step": 58106
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.762208104133606,
      "learning_rate": 0.0005106626820410128,
      "loss": 2.0266,
      "step": 58107
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.8097792863845825,
      "learning_rate": 0.0005106597696600783,
      "loss": 2.0526,
      "step": 58108
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7123718857765198,
      "learning_rate": 0.0005106568572399783,
      "loss": 2.3713,
      "step": 58109
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.8186546564102173,
      "learning_rate": 0.000510653944780713,
      "loss": 2.4044,
      "step": 58110
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7918218374252319,
      "learning_rate": 0.0005106510322822833,
      "loss": 2.2453,
      "step": 58111
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7234947085380554,
      "learning_rate": 0.0005106481197446896,
      "loss": 2.1299,
      "step": 58112
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.0994956493377686,
      "learning_rate": 0.0005106452071679323,
      "loss": 2.2076,
      "step": 58113
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7452132701873779,
      "learning_rate": 0.0005106422945520122,
      "loss": 2.1362,
      "step": 58114
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.6761025786399841,
      "learning_rate": 0.0005106393818969298,
      "loss": 2.1015,
      "step": 58115
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7553483843803406,
      "learning_rate": 0.0005106364692026855,
      "loss": 2.3117,
      "step": 58116
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.1224161386489868,
      "learning_rate": 0.0005106335564692798,
      "loss": 2.1536,
      "step": 58117
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7229277491569519,
      "learning_rate": 0.0005106306436967134,
      "loss": 2.1644,
      "step": 58118
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.9558548331260681,
      "learning_rate": 0.0005106277308849868,
      "loss": 2.4484,
      "step": 58119
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.8648178577423096,
      "learning_rate": 0.0005106248180341005,
      "loss": 2.2323,
      "step": 58120
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.741474986076355,
      "learning_rate": 0.0005106219051440551,
      "loss": 2.1224,
      "step": 58121
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7398846745491028,
      "learning_rate": 0.0005106189922148511,
      "loss": 2.425,
      "step": 58122
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.9037163257598877,
      "learning_rate": 0.0005106160792464891,
      "loss": 2.0247,
      "step": 58123
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7982473373413086,
      "learning_rate": 0.0005106131662389696,
      "loss": 2.1207,
      "step": 58124
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.643905520439148,
      "learning_rate": 0.0005106102531922931,
      "loss": 2.1149,
      "step": 58125
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.883547306060791,
      "learning_rate": 0.0005106073401064602,
      "loss": 2.2407,
      "step": 58126
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.8644400835037231,
      "learning_rate": 0.0005106044269814713,
      "loss": 2.2343,
      "step": 58127
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7771515846252441,
      "learning_rate": 0.0005106015138173272,
      "loss": 2.1901,
      "step": 58128
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7679228782653809,
      "learning_rate": 0.0005105986006140283,
      "loss": 2.3403,
      "step": 58129
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.9732360243797302,
      "learning_rate": 0.000510595687371575,
      "loss": 2.1703,
      "step": 58130
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7458009719848633,
      "learning_rate": 0.0005105927740899682,
      "loss": 2.1067,
      "step": 58131
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7226445078849792,
      "learning_rate": 0.0005105898607692081,
      "loss": 2.2004,
      "step": 58132
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.8220588564872742,
      "learning_rate": 0.0005105869474092954,
      "loss": 2.3065,
      "step": 58133
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.9280169606208801,
      "learning_rate": 0.0005105840340102306,
      "loss": 2.2328,
      "step": 58134
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.9632351398468018,
      "learning_rate": 0.0005105811205720141,
      "loss": 2.2126,
      "step": 58135
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7222471833229065,
      "learning_rate": 0.0005105782070946468,
      "loss": 2.3818,
      "step": 58136
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.0571485757827759,
      "learning_rate": 0.000510575293578129,
      "loss": 2.2518,
      "step": 58137
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.8179940581321716,
      "learning_rate": 0.0005105723800224612,
      "loss": 2.3464,
      "step": 58138
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7214499115943909,
      "learning_rate": 0.0005105694664276442,
      "loss": 1.9885,
      "step": 58139
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.8783512115478516,
      "learning_rate": 0.0005105665527936783,
      "loss": 2.1333,
      "step": 58140
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.798579752445221,
      "learning_rate": 0.000510563639120564,
      "loss": 2.1439,
      "step": 58141
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7909329533576965,
      "learning_rate": 0.000510560725408302,
      "loss": 2.1943,
      "step": 58142
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7383168935775757,
      "learning_rate": 0.0005105578116568928,
      "loss": 2.0774,
      "step": 58143
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.8265332579612732,
      "learning_rate": 0.0005105548978663368,
      "loss": 2.224,
      "step": 58144
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.8176242113113403,
      "learning_rate": 0.0005105519840366349,
      "loss": 2.1963,
      "step": 58145
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7294818758964539,
      "learning_rate": 0.0005105490701677873,
      "loss": 2.1693,
      "step": 58146
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7769736647605896,
      "learning_rate": 0.0005105461562597946,
      "loss": 2.3787,
      "step": 58147
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7517663240432739,
      "learning_rate": 0.0005105432423126574,
      "loss": 2.2173,
      "step": 58148
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7044336795806885,
      "learning_rate": 0.0005105403283263763,
      "loss": 2.1744,
      "step": 58149
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7566648125648499,
      "learning_rate": 0.0005105374143009518,
      "loss": 2.1094,
      "step": 58150
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.813198983669281,
      "learning_rate": 0.0005105345002363843,
      "loss": 2.1001,
      "step": 58151
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7474052906036377,
      "learning_rate": 0.0005105315861326746,
      "loss": 2.1345,
      "step": 58152
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7030518651008606,
      "learning_rate": 0.000510528671989823,
      "loss": 2.2137,
      "step": 58153
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.8146479725837708,
      "learning_rate": 0.0005105257578078302,
      "loss": 2.0019,
      "step": 58154
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7553377151489258,
      "learning_rate": 0.0005105228435866966,
      "loss": 2.379,
      "step": 58155
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7541473507881165,
      "learning_rate": 0.000510519929326423,
      "loss": 2.0974,
      "step": 58156
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7710327506065369,
      "learning_rate": 0.0005105170150270095,
      "loss": 2.1074,
      "step": 58157
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.8089218735694885,
      "learning_rate": 0.0005105141006884572,
      "loss": 2.6161,
      "step": 58158
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.8243548274040222,
      "learning_rate": 0.0005105111863107662,
      "loss": 2.331,
      "step": 58159
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7676319479942322,
      "learning_rate": 0.0005105082718939373,
      "loss": 2.2605,
      "step": 58160
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.8797826170921326,
      "learning_rate": 0.0005105053574379708,
      "loss": 2.3258,
      "step": 58161
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.9398801326751709,
      "learning_rate": 0.0005105024429428675,
      "loss": 2.4238,
      "step": 58162
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7866727113723755,
      "learning_rate": 0.0005104995284086278,
      "loss": 2.4382,
      "step": 58163
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.869117021560669,
      "learning_rate": 0.0005104966138352522,
      "loss": 2.3195,
      "step": 58164
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.8919100761413574,
      "learning_rate": 0.0005104936992227414,
      "loss": 2.2169,
      "step": 58165
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7160272598266602,
      "learning_rate": 0.0005104907845710958,
      "loss": 2.123,
      "step": 58166
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.798344612121582,
      "learning_rate": 0.0005104878698803159,
      "loss": 2.19,
      "step": 58167
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.8437972664833069,
      "learning_rate": 0.0005104849551504025,
      "loss": 2.2592,
      "step": 58168
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.6903804540634155,
      "learning_rate": 0.0005104820403813558,
      "loss": 2.3453,
      "step": 58169
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.6849964261054993,
      "learning_rate": 0.0005104791255731766,
      "loss": 2.4134,
      "step": 58170
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.8314611315727234,
      "learning_rate": 0.0005104762107258655,
      "loss": 2.2044,
      "step": 58171
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7344548106193542,
      "learning_rate": 0.0005104732958394227,
      "loss": 2.3156,
      "step": 58172
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.8827991485595703,
      "learning_rate": 0.000510470380913849,
      "loss": 2.4391,
      "step": 58173
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7669401168823242,
      "learning_rate": 0.000510467465949145,
      "loss": 2.4122,
      "step": 58174
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.9830787777900696,
      "learning_rate": 0.0005104645509453109,
      "loss": 2.2018,
      "step": 58175
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7808507680892944,
      "learning_rate": 0.0005104616359023476,
      "loss": 2.1459,
      "step": 58176
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.0397664308547974,
      "learning_rate": 0.0005104587208202556,
      "loss": 2.2373,
      "step": 58177
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7964736819267273,
      "learning_rate": 0.0005104558056990351,
      "loss": 2.2525,
      "step": 58178
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7085660696029663,
      "learning_rate": 0.0005104528905386871,
      "loss": 2.4135,
      "step": 58179
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.8629915714263916,
      "learning_rate": 0.0005104499753392119,
      "loss": 2.3071,
      "step": 58180
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7897219061851501,
      "learning_rate": 0.00051044706010061,
      "loss": 2.3938,
      "step": 58181
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.78667813539505,
      "learning_rate": 0.0005104441448228821,
      "loss": 2.1392,
      "step": 58182
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7288371324539185,
      "learning_rate": 0.0005104412295060286,
      "loss": 1.9859,
      "step": 58183
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.6883336305618286,
      "learning_rate": 0.0005104383141500502,
      "loss": 2.1607,
      "step": 58184
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7399846315383911,
      "learning_rate": 0.0005104353987549472,
      "loss": 2.3071,
      "step": 58185
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7238065004348755,
      "learning_rate": 0.0005104324833207204,
      "loss": 2.4015,
      "step": 58186
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7406042814254761,
      "learning_rate": 0.0005104295678473702,
      "loss": 2.1757,
      "step": 58187
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7772801518440247,
      "learning_rate": 0.0005104266523348971,
      "loss": 2.1721,
      "step": 58188
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7355698347091675,
      "learning_rate": 0.0005104237367833018,
      "loss": 2.4433,
      "step": 58189
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7943463325500488,
      "learning_rate": 0.0005104208211925847,
      "loss": 2.3068,
      "step": 58190
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7626619935035706,
      "learning_rate": 0.0005104179055627464,
      "loss": 2.2257,
      "step": 58191
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7219700217247009,
      "learning_rate": 0.0005104149898937875,
      "loss": 2.3242,
      "step": 58192
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7140577435493469,
      "learning_rate": 0.0005104120741857085,
      "loss": 2.2122,
      "step": 58193
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7530537843704224,
      "learning_rate": 0.0005104091584385098,
      "loss": 2.0992,
      "step": 58194
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7727510929107666,
      "learning_rate": 0.0005104062426521922,
      "loss": 2.2489,
      "step": 58195
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.6950194239616394,
      "learning_rate": 0.0005104033268267561,
      "loss": 2.2094,
      "step": 58196
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7268913984298706,
      "learning_rate": 0.0005104004109622019,
      "loss": 2.3463,
      "step": 58197
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7089601755142212,
      "learning_rate": 0.0005103974950585304,
      "loss": 2.1902,
      "step": 58198
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.8059235215187073,
      "learning_rate": 0.0005103945791157421,
      "loss": 2.3107,
      "step": 58199
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7757359743118286,
      "learning_rate": 0.0005103916631338374,
      "loss": 2.2046,
      "step": 58200
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.8435977101325989,
      "learning_rate": 0.0005103887471128169,
      "loss": 2.2982,
      "step": 58201
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7079887986183167,
      "learning_rate": 0.0005103858310526812,
      "loss": 2.1419,
      "step": 58202
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.779233992099762,
      "learning_rate": 0.0005103829149534307,
      "loss": 2.3881,
      "step": 58203
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7076713442802429,
      "learning_rate": 0.0005103799988150662,
      "loss": 2.133,
      "step": 58204
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7281826138496399,
      "learning_rate": 0.000510377082637588,
      "loss": 2.2628,
      "step": 58205
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7376587390899658,
      "learning_rate": 0.0005103741664209967,
      "loss": 2.2727,
      "step": 58206
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.9409946203231812,
      "learning_rate": 0.0005103712501652929,
      "loss": 2.124,
      "step": 58207
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7361637949943542,
      "learning_rate": 0.0005103683338704771,
      "loss": 2.2425,
      "step": 58208
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.720657229423523,
      "learning_rate": 0.00051036541753655,
      "loss": 2.2963,
      "step": 58209
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7367405891418457,
      "learning_rate": 0.0005103625011635118,
      "loss": 2.2889,
      "step": 58210
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7560753226280212,
      "learning_rate": 0.0005103595847513634,
      "loss": 2.189,
      "step": 58211
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.6517524123191833,
      "learning_rate": 0.000510356668300105,
      "loss": 2.3352,
      "step": 58212
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7348375916481018,
      "learning_rate": 0.0005103537518097375,
      "loss": 2.1121,
      "step": 58213
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.8748464584350586,
      "learning_rate": 0.0005103508352802612,
      "loss": 2.314,
      "step": 58214
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7531591653823853,
      "learning_rate": 0.0005103479187116767,
      "loss": 2.2728,
      "step": 58215
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7925957441329956,
      "learning_rate": 0.0005103450021039846,
      "loss": 2.2923,
      "step": 58216
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7290949821472168,
      "learning_rate": 0.0005103420854571853,
      "loss": 2.1789,
      "step": 58217
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7759530544281006,
      "learning_rate": 0.0005103391687712795,
      "loss": 2.3461,
      "step": 58218
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.6959683299064636,
      "learning_rate": 0.0005103362520462676,
      "loss": 2.2467,
      "step": 58219
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.753864586353302,
      "learning_rate": 0.0005103333352821503,
      "loss": 2.3034,
      "step": 58220
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7819767594337463,
      "learning_rate": 0.0005103304184789281,
      "loss": 1.8136,
      "step": 58221
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7968635559082031,
      "learning_rate": 0.0005103275016366014,
      "loss": 2.4197,
      "step": 58222
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.6965712904930115,
      "learning_rate": 0.0005103245847551709,
      "loss": 2.2386,
      "step": 58223
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7791588306427002,
      "learning_rate": 0.0005103216678346371,
      "loss": 2.185,
      "step": 58224
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7561222910881042,
      "learning_rate": 0.0005103187508750006,
      "loss": 2.3584,
      "step": 58225
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7615325450897217,
      "learning_rate": 0.0005103158338762617,
      "loss": 2.1219,
      "step": 58226
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.9197375178337097,
      "learning_rate": 0.0005103129168384211,
      "loss": 2.3424,
      "step": 58227
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.8844398260116577,
      "learning_rate": 0.0005103099997614796,
      "loss": 2.2052,
      "step": 58228
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7927761673927307,
      "learning_rate": 0.0005103070826454374,
      "loss": 2.155,
      "step": 58229
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.8230445981025696,
      "learning_rate": 0.000510304165490295,
      "loss": 2.1192,
      "step": 58230
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7897368669509888,
      "learning_rate": 0.0005103012482960532,
      "loss": 2.3361,
      "step": 58231
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.9980022311210632,
      "learning_rate": 0.0005102983310627125,
      "loss": 2.2546,
      "step": 58232
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7355700731277466,
      "learning_rate": 0.0005102954137902732,
      "loss": 2.3033,
      "step": 58233
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.9401339888572693,
      "learning_rate": 0.0005102924964787361,
      "loss": 2.1714,
      "step": 58234
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7387373447418213,
      "learning_rate": 0.0005102895791281017,
      "loss": 2.2885,
      "step": 58235
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7427680492401123,
      "learning_rate": 0.0005102866617383705,
      "loss": 2.1037,
      "step": 58236
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7560957670211792,
      "learning_rate": 0.000510283744309543,
      "loss": 2.2938,
      "step": 58237
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.8645772337913513,
      "learning_rate": 0.0005102808268416198,
      "loss": 2.1048,
      "step": 58238
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.9100891947746277,
      "learning_rate": 0.0005102779093346015,
      "loss": 2.3045,
      "step": 58239
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.6890718936920166,
      "learning_rate": 0.0005102749917884884,
      "loss": 2.3687,
      "step": 58240
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.791540265083313,
      "learning_rate": 0.0005102720742032813,
      "loss": 2.0196,
      "step": 58241
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.8066848516464233,
      "learning_rate": 0.0005102691565789806,
      "loss": 2.0428,
      "step": 58242
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.8628429174423218,
      "learning_rate": 0.000510266238915587,
      "loss": 2.1278,
      "step": 58243
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7490845918655396,
      "learning_rate": 0.0005102633212131009,
      "loss": 1.9702,
      "step": 58244
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.8945480585098267,
      "learning_rate": 0.0005102604034715228,
      "loss": 2.3832,
      "step": 58245
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7789487242698669,
      "learning_rate": 0.0005102574856908534,
      "loss": 2.1663,
      "step": 58246
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.8140097856521606,
      "learning_rate": 0.0005102545678710931,
      "loss": 2.2421,
      "step": 58247
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7870472073554993,
      "learning_rate": 0.0005102516500122425,
      "loss": 2.3812,
      "step": 58248
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7767789363861084,
      "learning_rate": 0.0005102487321143022,
      "loss": 2.0531,
      "step": 58249
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7944338917732239,
      "learning_rate": 0.0005102458141772727,
      "loss": 1.8982,
      "step": 58250
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7179468870162964,
      "learning_rate": 0.0005102428962011546,
      "loss": 2.1504,
      "step": 58251
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7555552124977112,
      "learning_rate": 0.0005102399781859483,
      "loss": 2.5314,
      "step": 58252
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.8691136240959167,
      "learning_rate": 0.0005102370601316544,
      "loss": 2.2513,
      "step": 58253
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.6651274561882019,
      "learning_rate": 0.0005102341420382734,
      "loss": 2.0334,
      "step": 58254
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7310775518417358,
      "learning_rate": 0.000510231223905806,
      "loss": 2.36,
      "step": 58255
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7698832154273987,
      "learning_rate": 0.0005102283057342526,
      "loss": 2.1354,
      "step": 58256
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7762096524238586,
      "learning_rate": 0.0005102253875236139,
      "loss": 1.9439,
      "step": 58257
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7307546138763428,
      "learning_rate": 0.0005102224692738902,
      "loss": 2.0725,
      "step": 58258
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7155932188034058,
      "learning_rate": 0.0005102195509850822,
      "loss": 2.2091,
      "step": 58259
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7097261548042297,
      "learning_rate": 0.0005102166326571905,
      "loss": 2.2285,
      "step": 58260
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.8966488838195801,
      "learning_rate": 0.0005102137142902154,
      "loss": 2.3434,
      "step": 58261
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.790386438369751,
      "learning_rate": 0.0005102107958841578,
      "loss": 2.2265,
      "step": 58262
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.6778937578201294,
      "learning_rate": 0.000510207877439018,
      "loss": 2.0881,
      "step": 58263
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.8163291215896606,
      "learning_rate": 0.0005102049589547965,
      "loss": 2.5222,
      "step": 58264
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.8421716690063477,
      "learning_rate": 0.000510202040431494,
      "loss": 2.2031,
      "step": 58265
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7141109704971313,
      "learning_rate": 0.000510199121869111,
      "loss": 2.0499,
      "step": 58266
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7411100268363953,
      "learning_rate": 0.0005101962032676479,
      "loss": 2.5107,
      "step": 58267
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.791666567325592,
      "learning_rate": 0.0005101932846271055,
      "loss": 2.0915,
      "step": 58268
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7487382292747498,
      "learning_rate": 0.000510190365947484,
      "loss": 2.1973,
      "step": 58269
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.8027696013450623,
      "learning_rate": 0.0005101874472287842,
      "loss": 2.246,
      "step": 58270
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7176696062088013,
      "learning_rate": 0.0005101845284710067,
      "loss": 2.2462,
      "step": 58271
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7187358140945435,
      "learning_rate": 0.0005101816096741518,
      "loss": 2.6395,
      "step": 58272
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7805934548377991,
      "learning_rate": 0.0005101786908382203,
      "loss": 2.4734,
      "step": 58273
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.8808989524841309,
      "learning_rate": 0.0005101757719632126,
      "loss": 2.2244,
      "step": 58274
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.6594669818878174,
      "learning_rate": 0.0005101728530491291,
      "loss": 2.0263,
      "step": 58275
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.6639205813407898,
      "learning_rate": 0.0005101699340959707,
      "loss": 1.9338,
      "step": 58276
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7394615411758423,
      "learning_rate": 0.0005101670151037377,
      "loss": 2.1606,
      "step": 58277
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.6819271445274353,
      "learning_rate": 0.0005101640960724305,
      "loss": 2.2525,
      "step": 58278
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.743138313293457,
      "learning_rate": 0.00051016117700205,
      "loss": 1.9122,
      "step": 58279
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7017894983291626,
      "learning_rate": 0.0005101582578925966,
      "loss": 2.2487,
      "step": 58280
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.8063741326332092,
      "learning_rate": 0.0005101553387440708,
      "loss": 2.2849,
      "step": 58281
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7200849056243896,
      "learning_rate": 0.000510152419556473,
      "loss": 2.2461,
      "step": 58282
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.6577445864677429,
      "learning_rate": 0.0005101495003298041,
      "loss": 2.2747,
      "step": 58283
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.8324016332626343,
      "learning_rate": 0.0005101465810640643,
      "loss": 1.9745,
      "step": 58284
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.737636148929596,
      "learning_rate": 0.0005101436617592544,
      "loss": 2.0694,
      "step": 58285
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7979936599731445,
      "learning_rate": 0.0005101407424153747,
      "loss": 2.3468,
      "step": 58286
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7028464078903198,
      "learning_rate": 0.0005101378230324259,
      "loss": 2.1848,
      "step": 58287
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.6853156089782715,
      "learning_rate": 0.0005101349036104086,
      "loss": 2.2916,
      "step": 58288
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.8549462556838989,
      "learning_rate": 0.0005101319841493232,
      "loss": 2.3769,
      "step": 58289
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7374961972236633,
      "learning_rate": 0.0005101290646491702,
      "loss": 2.5443,
      "step": 58290
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7747486233711243,
      "learning_rate": 0.0005101261451099504,
      "loss": 2.3677,
      "step": 58291
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.781463623046875,
      "learning_rate": 0.0005101232255316641,
      "loss": 2.2674,
      "step": 58292
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7120843529701233,
      "learning_rate": 0.000510120305914312,
      "loss": 2.3857,
      "step": 58293
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.6793375015258789,
      "learning_rate": 0.0005101173862578944,
      "loss": 2.0753,
      "step": 58294
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.8000748157501221,
      "learning_rate": 0.0005101144665624122,
      "loss": 2.3184,
      "step": 58295
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7056047320365906,
      "learning_rate": 0.0005101115468278657,
      "loss": 2.184,
      "step": 58296
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.6779049038887024,
      "learning_rate": 0.0005101086270542554,
      "loss": 2.1432,
      "step": 58297
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7193610072135925,
      "learning_rate": 0.000510105707241582,
      "loss": 2.209,
      "step": 58298
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7182174921035767,
      "learning_rate": 0.000510102787389846,
      "loss": 2.3655,
      "step": 58299
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7370383739471436,
      "learning_rate": 0.0005100998674990479,
      "loss": 2.2022,
      "step": 58300
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7366383075714111,
      "learning_rate": 0.0005100969475691884,
      "loss": 2.1173,
      "step": 58301
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.763533890247345,
      "learning_rate": 0.0005100940276002678,
      "loss": 2.1858,
      "step": 58302
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7522585391998291,
      "learning_rate": 0.0005100911075922867,
      "loss": 2.238,
      "step": 58303
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.739722728729248,
      "learning_rate": 0.0005100881875452458,
      "loss": 2.2089,
      "step": 58304
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7353901267051697,
      "learning_rate": 0.0005100852674591455,
      "loss": 2.1698,
      "step": 58305
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.672604501247406,
      "learning_rate": 0.0005100823473339864,
      "loss": 2.3696,
      "step": 58306
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7169225215911865,
      "learning_rate": 0.0005100794271697689,
      "loss": 2.1668,
      "step": 58307
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.772624135017395,
      "learning_rate": 0.0005100765069664938,
      "loss": 2.1663,
      "step": 58308
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7024177312850952,
      "learning_rate": 0.0005100735867241614,
      "loss": 2.411,
      "step": 58309
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7616267204284668,
      "learning_rate": 0.0005100706664427726,
      "loss": 2.2734,
      "step": 58310
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.8146899342536926,
      "learning_rate": 0.0005100677461223274,
      "loss": 2.1088,
      "step": 58311
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7315331697463989,
      "learning_rate": 0.0005100648257628268,
      "loss": 2.0126,
      "step": 58312
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7588338255882263,
      "learning_rate": 0.0005100619053642712,
      "loss": 2.2571,
      "step": 58313
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7540271878242493,
      "learning_rate": 0.000510058984926661,
      "loss": 2.1263,
      "step": 58314
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.8386154174804688,
      "learning_rate": 0.000510056064449997,
      "loss": 2.1985,
      "step": 58315
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7762590050697327,
      "learning_rate": 0.0005100531439342795,
      "loss": 2.0497,
      "step": 58316
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7262539863586426,
      "learning_rate": 0.0005100502233795094,
      "loss": 2.1887,
      "step": 58317
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.6945331692695618,
      "learning_rate": 0.0005100473027856867,
      "loss": 2.0639,
      "step": 58318
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7063003778457642,
      "learning_rate": 0.0005100443821528124,
      "loss": 1.9595,
      "step": 58319
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7839488387107849,
      "learning_rate": 0.0005100414614808868,
      "loss": 2.1851,
      "step": 58320
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7367879152297974,
      "learning_rate": 0.0005100385407699106,
      "loss": 2.4186,
      "step": 58321
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7819890379905701,
      "learning_rate": 0.0005100356200198842,
      "loss": 2.5753,
      "step": 58322
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7319107055664062,
      "learning_rate": 0.0005100326992308085,
      "loss": 2.08,
      "step": 58323
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.728056788444519,
      "learning_rate": 0.0005100297784026834,
      "loss": 2.2605,
      "step": 58324
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7448930740356445,
      "learning_rate": 0.0005100268575355098,
      "loss": 2.286,
      "step": 58325
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.6736279726028442,
      "learning_rate": 0.0005100239366292885,
      "loss": 2.2122,
      "step": 58326
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7299233078956604,
      "learning_rate": 0.0005100210156840196,
      "loss": 2.258,
      "step": 58327
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.679776132106781,
      "learning_rate": 0.0005100180946997039,
      "loss": 2.0469,
      "step": 58328
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7529170513153076,
      "learning_rate": 0.0005100151736763418,
      "loss": 2.1152,
      "step": 58329
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7347592115402222,
      "learning_rate": 0.000510012252613934,
      "loss": 2.1113,
      "step": 58330
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7448505759239197,
      "learning_rate": 0.0005100093315124808,
      "loss": 2.2761,
      "step": 58331
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7673617005348206,
      "learning_rate": 0.0005100064103719832,
      "loss": 2.2003,
      "step": 58332
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7415688037872314,
      "learning_rate": 0.0005100034891924412,
      "loss": 2.4101,
      "step": 58333
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.8508048057556152,
      "learning_rate": 0.0005100005679738556,
      "loss": 2.3936,
      "step": 58334
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7097424268722534,
      "learning_rate": 0.0005099976467162271,
      "loss": 2.222,
      "step": 58335
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7511777281761169,
      "learning_rate": 0.0005099947254195559,
      "loss": 2.4418,
      "step": 58336
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7195419669151306,
      "learning_rate": 0.0005099918040838427,
      "loss": 2.3381,
      "step": 58337
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7884106636047363,
      "learning_rate": 0.0005099888827090882,
      "loss": 2.5214,
      "step": 58338
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.8553080558776855,
      "learning_rate": 0.0005099859612952927,
      "loss": 2.47,
      "step": 58339
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7927150130271912,
      "learning_rate": 0.0005099830398424569,
      "loss": 2.2792,
      "step": 58340
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7722392678260803,
      "learning_rate": 0.0005099801183505812,
      "loss": 2.4459,
      "step": 58341
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7724276185035706,
      "learning_rate": 0.0005099771968196665,
      "loss": 2.2765,
      "step": 58342
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7204127907752991,
      "learning_rate": 0.0005099742752497127,
      "loss": 2.1004,
      "step": 58343
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7111047506332397,
      "learning_rate": 0.0005099713536407209,
      "loss": 2.075,
      "step": 58344
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7682664394378662,
      "learning_rate": 0.0005099684319926916,
      "loss": 2.2556,
      "step": 58345
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.8031096458435059,
      "learning_rate": 0.000509965510305625,
      "loss": 1.9964,
      "step": 58346
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.6946704983711243,
      "learning_rate": 0.0005099625885795219,
      "loss": 2.1318,
      "step": 58347
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7360793352127075,
      "learning_rate": 0.0005099596668143828,
      "loss": 2.206,
      "step": 58348
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7082380056381226,
      "learning_rate": 0.0005099567450102081,
      "loss": 2.532,
      "step": 58349
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.733012318611145,
      "learning_rate": 0.0005099538231669987,
      "loss": 2.084,
      "step": 58350
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7486359477043152,
      "learning_rate": 0.0005099509012847548,
      "loss": 2.2766,
      "step": 58351
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.6961595416069031,
      "learning_rate": 0.0005099479793634772,
      "loss": 2.29,
      "step": 58352
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7741259336471558,
      "learning_rate": 0.0005099450574031661,
      "loss": 2.5284,
      "step": 58353
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7148547768592834,
      "learning_rate": 0.0005099421354038225,
      "loss": 2.4314,
      "step": 58354
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7131136655807495,
      "learning_rate": 0.0005099392133654466,
      "loss": 2.1702,
      "step": 58355
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.6428680419921875,
      "learning_rate": 0.0005099362912880388,
      "loss": 2.11,
      "step": 58356
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7484833002090454,
      "learning_rate": 0.0005099333691716001,
      "loss": 2.0606,
      "step": 58357
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7538639307022095,
      "learning_rate": 0.0005099304470161308,
      "loss": 2.2827,
      "step": 58358
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7775601148605347,
      "learning_rate": 0.0005099275248216314,
      "loss": 2.2372,
      "step": 58359
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7009207606315613,
      "learning_rate": 0.0005099246025881027,
      "loss": 1.9669,
      "step": 58360
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7007752656936646,
      "learning_rate": 0.0005099216803155449,
      "loss": 2.3755,
      "step": 58361
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.6924706101417542,
      "learning_rate": 0.0005099187580039587,
      "loss": 2.4275,
      "step": 58362
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.6937757134437561,
      "learning_rate": 0.0005099158356533446,
      "loss": 2.3774,
      "step": 58363
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7138862013816833,
      "learning_rate": 0.0005099129132637033,
      "loss": 2.1932,
      "step": 58364
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7943659424781799,
      "learning_rate": 0.0005099099908350351,
      "loss": 2.0617,
      "step": 58365
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.6935725212097168,
      "learning_rate": 0.0005099070683673409,
      "loss": 2.2824,
      "step": 58366
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.8438848853111267,
      "learning_rate": 0.0005099041458606208,
      "loss": 2.2627,
      "step": 58367
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.8406177759170532,
      "learning_rate": 0.0005099012233148756,
      "loss": 2.4094,
      "step": 58368
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.6706897020339966,
      "learning_rate": 0.0005098983007301058,
      "loss": 2.1509,
      "step": 58369
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7749183177947998,
      "learning_rate": 0.000509895378106312,
      "loss": 2.1966,
      "step": 58370
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7390196919441223,
      "learning_rate": 0.0005098924554434946,
      "loss": 2.3438,
      "step": 58371
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.782937228679657,
      "learning_rate": 0.0005098895327416543,
      "loss": 2.1263,
      "step": 58372
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.8064415454864502,
      "learning_rate": 0.0005098866100007915,
      "loss": 2.3257,
      "step": 58373
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.8316690921783447,
      "learning_rate": 0.0005098836872209068,
      "loss": 2.4146,
      "step": 58374
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7625516057014465,
      "learning_rate": 0.0005098807644020009,
      "loss": 2.2306,
      "step": 58375
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7036862969398499,
      "learning_rate": 0.0005098778415440742,
      "loss": 2.1091,
      "step": 58376
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7955202460289001,
      "learning_rate": 0.0005098749186471271,
      "loss": 2.352,
      "step": 58377
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7582215666770935,
      "learning_rate": 0.0005098719957111604,
      "loss": 1.9358,
      "step": 58378
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7556919455528259,
      "learning_rate": 0.0005098690727361745,
      "loss": 2.189,
      "step": 58379
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.9011619687080383,
      "learning_rate": 0.00050986614972217,
      "loss": 2.2165,
      "step": 58380
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7780916094779968,
      "learning_rate": 0.0005098632266691475,
      "loss": 2.148,
      "step": 58381
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7064821720123291,
      "learning_rate": 0.0005098603035771074,
      "loss": 2.2998,
      "step": 58382
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.005838394165039,
      "learning_rate": 0.0005098573804460502,
      "loss": 2.1713,
      "step": 58383
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7152225375175476,
      "learning_rate": 0.0005098544572759766,
      "loss": 2.3608,
      "step": 58384
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7711271047592163,
      "learning_rate": 0.0005098515340668871,
      "loss": 2.2198,
      "step": 58385
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.8865649104118347,
      "learning_rate": 0.0005098486108187823,
      "loss": 2.0637,
      "step": 58386
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.720656156539917,
      "learning_rate": 0.0005098456875316627,
      "loss": 2.0896,
      "step": 58387
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.8333992958068848,
      "learning_rate": 0.0005098427642055287,
      "loss": 2.0883,
      "step": 58388
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.8172165155410767,
      "learning_rate": 0.000509839840840381,
      "loss": 2.2919,
      "step": 58389
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7015265822410583,
      "learning_rate": 0.0005098369174362201,
      "loss": 2.1157,
      "step": 58390
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.8458849191665649,
      "learning_rate": 0.0005098339939930467,
      "loss": 2.1304,
      "step": 58391
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7870825529098511,
      "learning_rate": 0.0005098310705108611,
      "loss": 2.1435,
      "step": 58392
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7178145051002502,
      "learning_rate": 0.000509828146989664,
      "loss": 2.2314,
      "step": 58393
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7279054522514343,
      "learning_rate": 0.0005098252234294558,
      "loss": 2.3462,
      "step": 58394
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7565272450447083,
      "learning_rate": 0.0005098222998302372,
      "loss": 2.1895,
      "step": 58395
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7449072003364563,
      "learning_rate": 0.0005098193761920086,
      "loss": 2.5247,
      "step": 58396
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.6984835863113403,
      "learning_rate": 0.0005098164525147706,
      "loss": 2.2131,
      "step": 58397
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.9445284008979797,
      "learning_rate": 0.0005098135287985239,
      "loss": 2.119,
      "step": 58398
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7053080201148987,
      "learning_rate": 0.0005098106050432688,
      "loss": 2.1852,
      "step": 58399
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7924255728721619,
      "learning_rate": 0.0005098076812490059,
      "loss": 2.243,
      "step": 58400
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7625885605812073,
      "learning_rate": 0.0005098047574157359,
      "loss": 2.129,
      "step": 58401
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.6581708192825317,
      "learning_rate": 0.0005098018335434592,
      "loss": 2.1254,
      "step": 58402
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7261607646942139,
      "learning_rate": 0.0005097989096321764,
      "loss": 2.2892,
      "step": 58403
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.747542142868042,
      "learning_rate": 0.000509795985681888,
      "loss": 2.3068,
      "step": 58404
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7960101366043091,
      "learning_rate": 0.0005097930616925946,
      "loss": 2.21,
      "step": 58405
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.6819114685058594,
      "learning_rate": 0.0005097901376642966,
      "loss": 2.334,
      "step": 58406
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.6994616389274597,
      "learning_rate": 0.0005097872135969947,
      "loss": 2.2085,
      "step": 58407
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7603402733802795,
      "learning_rate": 0.0005097842894906894,
      "loss": 2.0318,
      "step": 58408
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.773933470249176,
      "learning_rate": 0.0005097813653453814,
      "loss": 2.4972,
      "step": 58409
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7765783071517944,
      "learning_rate": 0.0005097784411610709,
      "loss": 2.0624,
      "step": 58410
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7254027128219604,
      "learning_rate": 0.0005097755169377587,
      "loss": 2.2597,
      "step": 58411
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7756271958351135,
      "learning_rate": 0.0005097725926754452,
      "loss": 2.1734,
      "step": 58412
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7961752414703369,
      "learning_rate": 0.0005097696683741311,
      "loss": 2.2528,
      "step": 58413
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7577844858169556,
      "learning_rate": 0.0005097667440338169,
      "loss": 2.2516,
      "step": 58414
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7361635565757751,
      "learning_rate": 0.000509763819654503,
      "loss": 2.3078,
      "step": 58415
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.8365862965583801,
      "learning_rate": 0.00050976089523619,
      "loss": 2.2604,
      "step": 58416
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.6637933254241943,
      "learning_rate": 0.0005097579707788786,
      "loss": 2.2453,
      "step": 58417
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.8277449607849121,
      "learning_rate": 0.0005097550462825692,
      "loss": 2.2734,
      "step": 58418
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.769596517086029,
      "learning_rate": 0.0005097521217472624,
      "loss": 2.185,
      "step": 58419
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.8065441250801086,
      "learning_rate": 0.0005097491971729586,
      "loss": 2.0588,
      "step": 58420
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7663593292236328,
      "learning_rate": 0.0005097462725596587,
      "loss": 2.2338,
      "step": 58421
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7082911133766174,
      "learning_rate": 0.0005097433479073629,
      "loss": 2.0538,
      "step": 58422
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.683413028717041,
      "learning_rate": 0.0005097404232160717,
      "loss": 2.3427,
      "step": 58423
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.8374927639961243,
      "learning_rate": 0.000509737498485786,
      "loss": 2.3846,
      "step": 58424
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.8002291321754456,
      "learning_rate": 0.000509734573716506,
      "loss": 2.2821,
      "step": 58425
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7307795882225037,
      "learning_rate": 0.0005097316489082324,
      "loss": 2.1447,
      "step": 58426
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.791331946849823,
      "learning_rate": 0.0005097287240609658,
      "loss": 2.3025,
      "step": 58427
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7472568154335022,
      "learning_rate": 0.0005097257991747066,
      "loss": 2.2603,
      "step": 58428
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7157481908798218,
      "learning_rate": 0.0005097228742494555,
      "loss": 2.2829,
      "step": 58429
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.0690081119537354,
      "learning_rate": 0.0005097199492852129,
      "loss": 2.3738,
      "step": 58430
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7315780520439148,
      "learning_rate": 0.0005097170242819794,
      "loss": 2.1654,
      "step": 58431
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7217597365379333,
      "learning_rate": 0.0005097140992397556,
      "loss": 2.024,
      "step": 58432
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.801569402217865,
      "learning_rate": 0.000509711174158542,
      "loss": 2.0351,
      "step": 58433
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7848520874977112,
      "learning_rate": 0.0005097082490383391,
      "loss": 2.4055,
      "step": 58434
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.8616014122962952,
      "learning_rate": 0.0005097053238791474,
      "loss": 2.2286,
      "step": 58435
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7010206580162048,
      "learning_rate": 0.0005097023986809676,
      "loss": 2.0164,
      "step": 58436
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7857681512832642,
      "learning_rate": 0.0005096994734438002,
      "loss": 2.4575,
      "step": 58437
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.0455514192581177,
      "learning_rate": 0.0005096965481676455,
      "loss": 2.2411,
      "step": 58438
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7053537964820862,
      "learning_rate": 0.0005096936228525045,
      "loss": 2.0343,
      "step": 58439
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7635599970817566,
      "learning_rate": 0.0005096906974983774,
      "loss": 2.017,
      "step": 58440
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.8344900608062744,
      "learning_rate": 0.0005096877721052649,
      "loss": 2.0923,
      "step": 58441
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7213981747627258,
      "learning_rate": 0.0005096848466731675,
      "loss": 2.2053,
      "step": 58442
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7787887454032898,
      "learning_rate": 0.0005096819212020856,
      "loss": 2.2367,
      "step": 58443
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7332843542098999,
      "learning_rate": 0.0005096789956920199,
      "loss": 2.2218,
      "step": 58444
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.8266660571098328,
      "learning_rate": 0.000509676070142971,
      "loss": 2.2907,
      "step": 58445
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7407537698745728,
      "learning_rate": 0.0005096731445549394,
      "loss": 2.1817,
      "step": 58446
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.6889812350273132,
      "learning_rate": 0.0005096702189279254,
      "loss": 2.1915,
      "step": 58447
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.8106186985969543,
      "learning_rate": 0.0005096672932619299,
      "loss": 1.9356,
      "step": 58448
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.6616166830062866,
      "learning_rate": 0.0005096643675569532,
      "loss": 2.2597,
      "step": 58449
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.6900670528411865,
      "learning_rate": 0.0005096614418129961,
      "loss": 2.1933,
      "step": 58450
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.8618471026420593,
      "learning_rate": 0.0005096585160300588,
      "loss": 2.3563,
      "step": 58451
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.8174021244049072,
      "learning_rate": 0.0005096555902081422,
      "loss": 2.0768,
      "step": 58452
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.783669650554657,
      "learning_rate": 0.0005096526643472466,
      "loss": 2.1569,
      "step": 58453
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.8502557277679443,
      "learning_rate": 0.0005096497384473725,
      "loss": 2.555,
      "step": 58454
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7185290455818176,
      "learning_rate": 0.0005096468125085206,
      "loss": 2.1946,
      "step": 58455
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7613083124160767,
      "learning_rate": 0.0005096438865306915,
      "loss": 2.2041,
      "step": 58456
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.761473536491394,
      "learning_rate": 0.0005096409605138856,
      "loss": 2.1418,
      "step": 58457
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.800413966178894,
      "learning_rate": 0.0005096380344581034,
      "loss": 2.2781,
      "step": 58458
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7609380483627319,
      "learning_rate": 0.0005096351083633457,
      "loss": 2.0261,
      "step": 58459
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7547569274902344,
      "learning_rate": 0.0005096321822296127,
      "loss": 2.2505,
      "step": 58460
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7402843236923218,
      "learning_rate": 0.0005096292560569052,
      "loss": 2.2204,
      "step": 58461
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.744964599609375,
      "learning_rate": 0.0005096263298452237,
      "loss": 2.3247,
      "step": 58462
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7297467589378357,
      "learning_rate": 0.0005096234035945687,
      "loss": 2.1093,
      "step": 58463
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7660169005393982,
      "learning_rate": 0.0005096204773049407,
      "loss": 2.2453,
      "step": 58464
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.809810221195221,
      "learning_rate": 0.0005096175509763404,
      "loss": 2.2018,
      "step": 58465
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7725179195404053,
      "learning_rate": 0.0005096146246087682,
      "loss": 2.364,
      "step": 58466
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7214215993881226,
      "learning_rate": 0.0005096116982022245,
      "loss": 2.2364,
      "step": 58467
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.769670307636261,
      "learning_rate": 0.0005096087717567102,
      "loss": 2.0625,
      "step": 58468
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7108182907104492,
      "learning_rate": 0.0005096058452722257,
      "loss": 2.0541,
      "step": 58469
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7895933389663696,
      "learning_rate": 0.0005096029187487715,
      "loss": 2.5203,
      "step": 58470
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7118865251541138,
      "learning_rate": 0.0005095999921863481,
      "loss": 2.2689,
      "step": 58471
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.2881861925125122,
      "learning_rate": 0.0005095970655849562,
      "loss": 2.2075,
      "step": 58472
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.8066494464874268,
      "learning_rate": 0.0005095941389445961,
      "loss": 2.3901,
      "step": 58473
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.6942356824874878,
      "learning_rate": 0.0005095912122652685,
      "loss": 2.1686,
      "step": 58474
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7467913627624512,
      "learning_rate": 0.0005095882855469741,
      "loss": 2.1936,
      "step": 58475
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.6844888925552368,
      "learning_rate": 0.0005095853587897132,
      "loss": 2.2455,
      "step": 58476
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7250213623046875,
      "learning_rate": 0.0005095824319934864,
      "loss": 2.2671,
      "step": 58477
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7744989395141602,
      "learning_rate": 0.0005095795051582943,
      "loss": 2.3338,
      "step": 58478
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7044691443443298,
      "learning_rate": 0.0005095765782841375,
      "loss": 2.2084,
      "step": 58479
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7714431285858154,
      "learning_rate": 0.0005095736513710162,
      "loss": 2.2449,
      "step": 58480
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7250229120254517,
      "learning_rate": 0.0005095707244189314,
      "loss": 2.4016,
      "step": 58481
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.6700470447540283,
      "learning_rate": 0.0005095677974278834,
      "loss": 2.1994,
      "step": 58482
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.950171172618866,
      "learning_rate": 0.0005095648703978728,
      "loss": 2.3215,
      "step": 58483
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.6839796304702759,
      "learning_rate": 0.0005095619433289001,
      "loss": 2.1276,
      "step": 58484
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.9242676496505737,
      "learning_rate": 0.000509559016220966,
      "loss": 2.1147,
      "step": 58485
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7454855442047119,
      "learning_rate": 0.0005095560890740708,
      "loss": 2.3949,
      "step": 58486
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.752091646194458,
      "learning_rate": 0.0005095531618882152,
      "loss": 2.2907,
      "step": 58487
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.846173882484436,
      "learning_rate": 0.0005095502346633998,
      "loss": 2.1986,
      "step": 58488
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7196739912033081,
      "learning_rate": 0.0005095473073996249,
      "loss": 2.3078,
      "step": 58489
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.702909529209137,
      "learning_rate": 0.0005095443800968913,
      "loss": 2.1385,
      "step": 58490
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7548989653587341,
      "learning_rate": 0.0005095414527551994,
      "loss": 2.0547,
      "step": 58491
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.870819091796875,
      "learning_rate": 0.0005095385253745498,
      "loss": 2.2615,
      "step": 58492
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7807162404060364,
      "learning_rate": 0.000509535597954943,
      "loss": 2.3092,
      "step": 58493
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7288114428520203,
      "learning_rate": 0.0005095326704963796,
      "loss": 2.2086,
      "step": 58494
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.9372386932373047,
      "learning_rate": 0.0005095297429988601,
      "loss": 2.1427,
      "step": 58495
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.8095829486846924,
      "learning_rate": 0.0005095268154623852,
      "loss": 2.1969,
      "step": 58496
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7768053412437439,
      "learning_rate": 0.0005095238878869552,
      "loss": 2.3167,
      "step": 58497
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7546250820159912,
      "learning_rate": 0.0005095209602725707,
      "loss": 2.2319,
      "step": 58498
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7691709995269775,
      "learning_rate": 0.0005095180326192324,
      "loss": 2.3026,
      "step": 58499
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.787183403968811,
      "learning_rate": 0.0005095151049269407,
      "loss": 2.1928,
      "step": 58500
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.6501472592353821,
      "learning_rate": 0.0005095121771956962,
      "loss": 1.9614,
      "step": 58501
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7535653114318848,
      "learning_rate": 0.0005095092494254995,
      "loss": 2.3755,
      "step": 58502
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7561513781547546,
      "learning_rate": 0.000509506321616351,
      "loss": 2.0502,
      "step": 58503
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7180114984512329,
      "learning_rate": 0.0005095033937682513,
      "loss": 2.1344,
      "step": 58504
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.708515465259552,
      "learning_rate": 0.0005095004658812009,
      "loss": 2.0889,
      "step": 58505
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7058992981910706,
      "learning_rate": 0.0005094975379552006,
      "loss": 2.3062,
      "step": 58506
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.8219034671783447,
      "learning_rate": 0.0005094946099902506,
      "loss": 2.5248,
      "step": 58507
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7289397120475769,
      "learning_rate": 0.0005094916819863516,
      "loss": 2.2977,
      "step": 58508
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7335706353187561,
      "learning_rate": 0.0005094887539435042,
      "loss": 2.312,
      "step": 58509
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7639700174331665,
      "learning_rate": 0.0005094858258617088,
      "loss": 2.2731,
      "step": 58510
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.6889108419418335,
      "learning_rate": 0.0005094828977409662,
      "loss": 2.2415,
      "step": 58511
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7343301177024841,
      "learning_rate": 0.0005094799695812767,
      "loss": 2.4001,
      "step": 58512
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7140148878097534,
      "learning_rate": 0.0005094770413826408,
      "loss": 2.4169,
      "step": 58513
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.771236777305603,
      "learning_rate": 0.0005094741131450594,
      "loss": 2.1183,
      "step": 58514
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.710639238357544,
      "learning_rate": 0.0005094711848685326,
      "loss": 2.4602,
      "step": 58515
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.8296058177947998,
      "learning_rate": 0.0005094682565530611,
      "loss": 2.3123,
      "step": 58516
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7814151644706726,
      "learning_rate": 0.0005094653281986457,
      "loss": 1.9766,
      "step": 58517
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.6957808136940002,
      "learning_rate": 0.0005094623998052867,
      "loss": 2.0787,
      "step": 58518
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.8302090167999268,
      "learning_rate": 0.0005094594713729847,
      "loss": 2.0621,
      "step": 58519
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.9204534888267517,
      "learning_rate": 0.0005094565429017401,
      "loss": 2.316,
      "step": 58520
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.706304132938385,
      "learning_rate": 0.0005094536143915537,
      "loss": 2.0805,
      "step": 58521
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7402727007865906,
      "learning_rate": 0.0005094506858424259,
      "loss": 2.2185,
      "step": 58522
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.8653865456581116,
      "learning_rate": 0.0005094477572543572,
      "loss": 2.0717,
      "step": 58523
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.8298028111457825,
      "learning_rate": 0.0005094448286273482,
      "loss": 2.1793,
      "step": 58524
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7848743200302124,
      "learning_rate": 0.0005094418999613995,
      "loss": 2.1794,
      "step": 58525
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.9114692807197571,
      "learning_rate": 0.0005094389712565116,
      "loss": 2.2906,
      "step": 58526
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.6835264563560486,
      "learning_rate": 0.000509436042512685,
      "loss": 2.3041,
      "step": 58527
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7973566651344299,
      "learning_rate": 0.0005094331137299203,
      "loss": 2.2265,
      "step": 58528
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.819968581199646,
      "learning_rate": 0.000509430184908218,
      "loss": 2.323,
      "step": 58529
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.8099339604377747,
      "learning_rate": 0.0005094272560475789,
      "loss": 2.4382,
      "step": 58530
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.6942521929740906,
      "learning_rate": 0.0005094243271480031,
      "loss": 2.3346,
      "step": 58531
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7715256214141846,
      "learning_rate": 0.0005094213982094913,
      "loss": 2.1986,
      "step": 58532
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7831740379333496,
      "learning_rate": 0.0005094184692320444,
      "loss": 2.1221,
      "step": 58533
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7684116363525391,
      "learning_rate": 0.0005094155402156624,
      "loss": 2.3719,
      "step": 58534
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7530739307403564,
      "learning_rate": 0.0005094126111603462,
      "loss": 2.3195,
      "step": 58535
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.8016360998153687,
      "learning_rate": 0.0005094096820660961,
      "loss": 2.3834,
      "step": 58536
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7868679761886597,
      "learning_rate": 0.000509406752932913,
      "loss": 2.0465,
      "step": 58537
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7480388879776001,
      "learning_rate": 0.0005094038237607972,
      "loss": 2.0095,
      "step": 58538
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7853989601135254,
      "learning_rate": 0.0005094008945497492,
      "loss": 2.1785,
      "step": 58539
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7144656777381897,
      "learning_rate": 0.0005093979652997696,
      "loss": 2.2918,
      "step": 58540
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7294363975524902,
      "learning_rate": 0.000509395036010859,
      "loss": 2.2516,
      "step": 58541
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7752012014389038,
      "learning_rate": 0.0005093921066830179,
      "loss": 1.9422,
      "step": 58542
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7496294975280762,
      "learning_rate": 0.0005093891773162468,
      "loss": 2.4246,
      "step": 58543
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7082540392875671,
      "learning_rate": 0.0005093862479105464,
      "loss": 2.3396,
      "step": 58544
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7175833582878113,
      "learning_rate": 0.000509383318465917,
      "loss": 2.2151,
      "step": 58545
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7790651917457581,
      "learning_rate": 0.0005093803889823595,
      "loss": 2.1997,
      "step": 58546
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.682013988494873,
      "learning_rate": 0.0005093774594598741,
      "loss": 2.3323,
      "step": 58547
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.8377148509025574,
      "learning_rate": 0.0005093745298984615,
      "loss": 2.175,
      "step": 58548
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7279303669929504,
      "learning_rate": 0.0005093716002981223,
      "loss": 2.2914,
      "step": 58549
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7476195096969604,
      "learning_rate": 0.0005093686706588569,
      "loss": 2.2574,
      "step": 58550
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.8257572054862976,
      "learning_rate": 0.0005093657409806658,
      "loss": 2.1861,
      "step": 58551
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.768778920173645,
      "learning_rate": 0.0005093628112635498,
      "loss": 2.2994,
      "step": 58552
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7200341820716858,
      "learning_rate": 0.0005093598815075093,
      "loss": 1.9917,
      "step": 58553
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.759576678276062,
      "learning_rate": 0.0005093569517125448,
      "loss": 2.261,
      "step": 58554
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7008370161056519,
      "learning_rate": 0.0005093540218786569,
      "loss": 2.3244,
      "step": 58555
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.8546415567398071,
      "learning_rate": 0.0005093510920058462,
      "loss": 2.1993,
      "step": 58556
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7613314986228943,
      "learning_rate": 0.000509348162094113,
      "loss": 2.2434,
      "step": 58557
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.8056701421737671,
      "learning_rate": 0.0005093452321434581,
      "loss": 2.3779,
      "step": 58558
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.975372314453125,
      "learning_rate": 0.0005093423021538822,
      "loss": 2.2065,
      "step": 58559
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.8980352878570557,
      "learning_rate": 0.0005093393721253853,
      "loss": 2.4825,
      "step": 58560
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7532752752304077,
      "learning_rate": 0.0005093364420579685,
      "loss": 2.3034,
      "step": 58561
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.9302133321762085,
      "learning_rate": 0.000509333511951632,
      "loss": 2.0081,
      "step": 58562
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.8888399004936218,
      "learning_rate": 0.0005093305818063763,
      "loss": 2.1589,
      "step": 58563
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7186502814292908,
      "learning_rate": 0.0005093276516222023,
      "loss": 1.9416,
      "step": 58564
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7173252701759338,
      "learning_rate": 0.0005093247213991103,
      "loss": 2.3394,
      "step": 58565
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7552083730697632,
      "learning_rate": 0.0005093217911371008,
      "loss": 2.1321,
      "step": 58566
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.8130278587341309,
      "learning_rate": 0.0005093188608361745,
      "loss": 2.3495,
      "step": 58567
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7052369117736816,
      "learning_rate": 0.0005093159304963319,
      "loss": 2.3657,
      "step": 58568
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.8499144911766052,
      "learning_rate": 0.0005093130001175734,
      "loss": 2.4599,
      "step": 58569
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7712970972061157,
      "learning_rate": 0.0005093100696998997,
      "loss": 2.2717,
      "step": 58570
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7724106311798096,
      "learning_rate": 0.0005093071392433114,
      "loss": 2.1952,
      "step": 58571
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.8102099895477295,
      "learning_rate": 0.0005093042087478089,
      "loss": 2.2926,
      "step": 58572
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7722018361091614,
      "learning_rate": 0.0005093012782133927,
      "loss": 2.2248,
      "step": 58573
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.6859356760978699,
      "learning_rate": 0.0005092983476400636,
      "loss": 2.2233,
      "step": 58574
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.8312362432479858,
      "learning_rate": 0.0005092954170278219,
      "loss": 2.457,
      "step": 58575
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7280153036117554,
      "learning_rate": 0.0005092924863766683,
      "loss": 2.232,
      "step": 58576
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7398442029953003,
      "learning_rate": 0.0005092895556866032,
      "loss": 2.3331,
      "step": 58577
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.9073887467384338,
      "learning_rate": 0.0005092866249576273,
      "loss": 2.1351,
      "step": 58578
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.6970571279525757,
      "learning_rate": 0.000509283694189741,
      "loss": 2.2637,
      "step": 58579
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.6494988203048706,
      "learning_rate": 0.0005092807633829448,
      "loss": 2.3236,
      "step": 58580
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7281655073165894,
      "learning_rate": 0.0005092778325372395,
      "loss": 2.2327,
      "step": 58581
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.8370325565338135,
      "learning_rate": 0.0005092749016526255,
      "loss": 2.4102,
      "step": 58582
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7768383026123047,
      "learning_rate": 0.0005092719707291034,
      "loss": 1.9898,
      "step": 58583
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.8186244964599609,
      "learning_rate": 0.0005092690397666735,
      "loss": 2.197,
      "step": 58584
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.9427661299705505,
      "learning_rate": 0.0005092661087653368,
      "loss": 2.3647,
      "step": 58585
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.6521085500717163,
      "learning_rate": 0.0005092631777250932,
      "loss": 2.242,
      "step": 58586
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.8583616614341736,
      "learning_rate": 0.000509260246645944,
      "loss": 2.3012,
      "step": 58587
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.6950228810310364,
      "learning_rate": 0.0005092573155278892,
      "loss": 2.0704,
      "step": 58588
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7490106225013733,
      "learning_rate": 0.0005092543843709295,
      "loss": 2.2618,
      "step": 58589
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7601893544197083,
      "learning_rate": 0.0005092514531750655,
      "loss": 2.2024,
      "step": 58590
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7515648007392883,
      "learning_rate": 0.0005092485219402977,
      "loss": 2.2037,
      "step": 58591
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.6531433463096619,
      "learning_rate": 0.0005092455906666267,
      "loss": 1.9572,
      "step": 58592
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.8021819591522217,
      "learning_rate": 0.0005092426593540529,
      "loss": 2.2247,
      "step": 58593
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7414857745170593,
      "learning_rate": 0.000509239728002577,
      "loss": 2.2894,
      "step": 58594
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.8306512832641602,
      "learning_rate": 0.0005092367966121994,
      "loss": 2.2953,
      "step": 58595
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.8001199960708618,
      "learning_rate": 0.0005092338651829209,
      "loss": 2.0554,
      "step": 58596
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7247896790504456,
      "learning_rate": 0.0005092309337147419,
      "loss": 2.2762,
      "step": 58597
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7662127614021301,
      "learning_rate": 0.0005092280022076627,
      "loss": 2.3641,
      "step": 58598
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.8623729348182678,
      "learning_rate": 0.0005092250706616841,
      "loss": 2.2964,
      "step": 58599
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7285361886024475,
      "learning_rate": 0.0005092221390768068,
      "loss": 2.1891,
      "step": 58600
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.9654768705368042,
      "learning_rate": 0.000509219207453031,
      "loss": 2.1757,
      "step": 58601
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.6778009533882141,
      "learning_rate": 0.0005092162757903574,
      "loss": 2.0633,
      "step": 58602
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.9437763690948486,
      "learning_rate": 0.0005092133440887865,
      "loss": 2.2742,
      "step": 58603
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7764180302619934,
      "learning_rate": 0.000509210412348319,
      "loss": 1.9646,
      "step": 58604
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7235668897628784,
      "learning_rate": 0.0005092074805689553,
      "loss": 2.3729,
      "step": 58605
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7688712477684021,
      "learning_rate": 0.000509204548750696,
      "loss": 2.3759,
      "step": 58606
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7652215361595154,
      "learning_rate": 0.0005092016168935416,
      "loss": 2.061,
      "step": 58607
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7739300727844238,
      "learning_rate": 0.0005091986849974926,
      "loss": 2.1324,
      "step": 58608
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.8133871555328369,
      "learning_rate": 0.0005091957530625498,
      "loss": 2.2431,
      "step": 58609
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.9136308431625366,
      "learning_rate": 0.0005091928210887134,
      "loss": 2.089,
      "step": 58610
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.8101409077644348,
      "learning_rate": 0.0005091898890759842,
      "loss": 2.3692,
      "step": 58611
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7493852376937866,
      "learning_rate": 0.0005091869570243625,
      "loss": 2.1271,
      "step": 58612
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7025917768478394,
      "learning_rate": 0.0005091840249338491,
      "loss": 2.2088,
      "step": 58613
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.8078809976577759,
      "learning_rate": 0.0005091810928044445,
      "loss": 2.3198,
      "step": 58614
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.6987242698669434,
      "learning_rate": 0.0005091781606361491,
      "loss": 2.2679,
      "step": 58615
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7831531763076782,
      "learning_rate": 0.0005091752284289636,
      "loss": 2.1482,
      "step": 58616
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.8048117756843567,
      "learning_rate": 0.0005091722961828884,
      "loss": 2.2207,
      "step": 58617
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7302945256233215,
      "learning_rate": 0.0005091693638979242,
      "loss": 2.0325,
      "step": 58618
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.72967129945755,
      "learning_rate": 0.0005091664315740714,
      "loss": 2.2927,
      "step": 58619
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7518161535263062,
      "learning_rate": 0.0005091634992113307,
      "loss": 2.3049,
      "step": 58620
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7116807699203491,
      "learning_rate": 0.0005091605668097025,
      "loss": 2.014,
      "step": 58621
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.6958324909210205,
      "learning_rate": 0.0005091576343691874,
      "loss": 2.2895,
      "step": 58622
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.8107318878173828,
      "learning_rate": 0.0005091547018897859,
      "loss": 1.9505,
      "step": 58623
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7870209217071533,
      "learning_rate": 0.0005091517693714986,
      "loss": 2.2865,
      "step": 58624
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.6853539943695068,
      "learning_rate": 0.0005091488368143261,
      "loss": 2.2852,
      "step": 58625
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7229764461517334,
      "learning_rate": 0.0005091459042182689,
      "loss": 2.1183,
      "step": 58626
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7612860202789307,
      "learning_rate": 0.0005091429715833274,
      "loss": 2.3557,
      "step": 58627
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.8085497617721558,
      "learning_rate": 0.0005091400389095024,
      "loss": 2.219,
      "step": 58628
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7145102024078369,
      "learning_rate": 0.0005091371061967942,
      "loss": 2.315,
      "step": 58629
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7071848511695862,
      "learning_rate": 0.0005091341734452036,
      "loss": 2.3257,
      "step": 58630
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.6818504929542542,
      "learning_rate": 0.000509131240654731,
      "loss": 2.2918,
      "step": 58631
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7623614072799683,
      "learning_rate": 0.0005091283078253769,
      "loss": 2.1239,
      "step": 58632
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7442917227745056,
      "learning_rate": 0.0005091253749571419,
      "loss": 2.2602,
      "step": 58633
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7461099624633789,
      "learning_rate": 0.0005091224420500266,
      "loss": 2.1665,
      "step": 58634
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7087357640266418,
      "learning_rate": 0.0005091195091040314,
      "loss": 2.3133,
      "step": 58635
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7204567193984985,
      "learning_rate": 0.000509116576119157,
      "loss": 2.3649,
      "step": 58636
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.6990029215812683,
      "learning_rate": 0.0005091136430954038,
      "loss": 2.2147,
      "step": 58637
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7497076988220215,
      "learning_rate": 0.0005091107100327726,
      "loss": 2.2587,
      "step": 58638
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.9506104588508606,
      "learning_rate": 0.0005091077769312636,
      "loss": 2.3132,
      "step": 58639
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.6658037304878235,
      "learning_rate": 0.0005091048437908776,
      "loss": 2.0783,
      "step": 58640
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.6884639859199524,
      "learning_rate": 0.0005091019106116151,
      "loss": 2.0984,
      "step": 58641
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7834532260894775,
      "learning_rate": 0.0005090989773934764,
      "loss": 2.2765,
      "step": 58642
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.757268488407135,
      "learning_rate": 0.0005090960441364624,
      "loss": 2.37,
      "step": 58643
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.688444197177887,
      "learning_rate": 0.0005090931108405735,
      "loss": 2.1587,
      "step": 58644
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7533361911773682,
      "learning_rate": 0.0005090901775058103,
      "loss": 2.3148,
      "step": 58645
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.698418140411377,
      "learning_rate": 0.0005090872441321733,
      "loss": 2.1744,
      "step": 58646
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7468228340148926,
      "learning_rate": 0.0005090843107196629,
      "loss": 2.2509,
      "step": 58647
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7239041924476624,
      "learning_rate": 0.0005090813772682798,
      "loss": 2.0856,
      "step": 58648
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7538676857948303,
      "learning_rate": 0.0005090784437780247,
      "loss": 2.2136,
      "step": 58649
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7812027931213379,
      "learning_rate": 0.0005090755102488977,
      "loss": 2.1914,
      "step": 58650
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.6977427005767822,
      "learning_rate": 0.0005090725766808998,
      "loss": 2.1549,
      "step": 58651
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7494135499000549,
      "learning_rate": 0.0005090696430740314,
      "loss": 2.2833,
      "step": 58652
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.652812123298645,
      "learning_rate": 0.0005090667094282929,
      "loss": 2.2122,
      "step": 58653
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.949769914150238,
      "learning_rate": 0.0005090637757436849,
      "loss": 2.2473,
      "step": 58654
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7415622472763062,
      "learning_rate": 0.0005090608420202081,
      "loss": 2.3686,
      "step": 58655
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7517051696777344,
      "learning_rate": 0.000509057908257863,
      "loss": 2.3005,
      "step": 58656
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7282642126083374,
      "learning_rate": 0.00050905497445665,
      "loss": 2.3826,
      "step": 58657
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.8081064224243164,
      "learning_rate": 0.0005090520406165698,
      "loss": 2.3984,
      "step": 58658
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.682428240776062,
      "learning_rate": 0.0005090491067376227,
      "loss": 2.15,
      "step": 58659
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7280694246292114,
      "learning_rate": 0.0005090461728198095,
      "loss": 2.1098,
      "step": 58660
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7424120903015137,
      "learning_rate": 0.0005090432388631307,
      "loss": 2.3457,
      "step": 58661
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.6871633529663086,
      "learning_rate": 0.0005090403048675869,
      "loss": 2.3653,
      "step": 58662
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.914940357208252,
      "learning_rate": 0.0005090373708331786,
      "loss": 2.0064,
      "step": 58663
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.8562894463539124,
      "learning_rate": 0.0005090344367599061,
      "loss": 2.4456,
      "step": 58664
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7591568827629089,
      "learning_rate": 0.0005090315026477703,
      "loss": 2.4088,
      "step": 58665
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.8432257771492004,
      "learning_rate": 0.0005090285684967715,
      "loss": 2.0343,
      "step": 58666
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7673584818840027,
      "learning_rate": 0.0005090256343069105,
      "loss": 2.4603,
      "step": 58667
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.6594617366790771,
      "learning_rate": 0.0005090227000781875,
      "loss": 2.4535,
      "step": 58668
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7088986039161682,
      "learning_rate": 0.0005090197658106033,
      "loss": 2.3443,
      "step": 58669
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.8243334293365479,
      "learning_rate": 0.0005090168315041584,
      "loss": 2.1689,
      "step": 58670
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7824035286903381,
      "learning_rate": 0.0005090138971588533,
      "loss": 2.3323,
      "step": 58671
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.797758162021637,
      "learning_rate": 0.0005090109627746886,
      "loss": 2.2645,
      "step": 58672
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.694348156452179,
      "learning_rate": 0.0005090080283516649,
      "loss": 2.1035,
      "step": 58673
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7510031461715698,
      "learning_rate": 0.0005090050938897825,
      "loss": 2.0109,
      "step": 58674
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7005009651184082,
      "learning_rate": 0.0005090021593890422,
      "loss": 2.2768,
      "step": 58675
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.8528991341590881,
      "learning_rate": 0.0005089992248494444,
      "loss": 1.9221,
      "step": 58676
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.1063374280929565,
      "learning_rate": 0.0005089962902709897,
      "loss": 2.1702,
      "step": 58677
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7305768132209778,
      "learning_rate": 0.0005089933556536787,
      "loss": 2.3065,
      "step": 58678
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.9600463509559631,
      "learning_rate": 0.0005089904209975119,
      "loss": 2.2749,
      "step": 58679
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7856385707855225,
      "learning_rate": 0.0005089874863024897,
      "loss": 2.2067,
      "step": 58680
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.705215573310852,
      "learning_rate": 0.0005089845515686129,
      "loss": 2.1228,
      "step": 58681
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.9913445115089417,
      "learning_rate": 0.000508981616795882,
      "loss": 2.4239,
      "step": 58682
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7768709659576416,
      "learning_rate": 0.0005089786819842974,
      "loss": 2.2056,
      "step": 58683
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.714210569858551,
      "learning_rate": 0.0005089757471338597,
      "loss": 2.2021,
      "step": 58684
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7603204846382141,
      "learning_rate": 0.0005089728122445695,
      "loss": 2.0518,
      "step": 58685
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.8125444650650024,
      "learning_rate": 0.0005089698773164273,
      "loss": 2.0457,
      "step": 58686
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.8332834839820862,
      "learning_rate": 0.0005089669423494335,
      "loss": 2.1508,
      "step": 58687
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.8068933486938477,
      "learning_rate": 0.0005089640073435889,
      "loss": 2.1938,
      "step": 58688
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7408482432365417,
      "learning_rate": 0.000508961072298894,
      "loss": 2.1345,
      "step": 58689
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.8342018127441406,
      "learning_rate": 0.0005089581372153493,
      "loss": 2.1904,
      "step": 58690
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.722406804561615,
      "learning_rate": 0.0005089552020929553,
      "loss": 2.1547,
      "step": 58691
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.8360464572906494,
      "learning_rate": 0.0005089522669317126,
      "loss": 2.0324,
      "step": 58692
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.8348581194877625,
      "learning_rate": 0.0005089493317316217,
      "loss": 2.181,
      "step": 58693
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.6820447444915771,
      "learning_rate": 0.0005089463964926832,
      "loss": 2.1188,
      "step": 58694
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.9447801113128662,
      "learning_rate": 0.0005089434612148976,
      "loss": 2.2762,
      "step": 58695
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.8713384866714478,
      "learning_rate": 0.0005089405258982656,
      "loss": 2.3039,
      "step": 58696
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7799121737480164,
      "learning_rate": 0.0005089375905427874,
      "loss": 2.2253,
      "step": 58697
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.8668200373649597,
      "learning_rate": 0.000508934655148464,
      "loss": 2.147,
      "step": 58698
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.9240849614143372,
      "learning_rate": 0.0005089317197152955,
      "loss": 2.3052,
      "step": 58699
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.696656346321106,
      "learning_rate": 0.0005089287842432826,
      "loss": 2.2223,
      "step": 58700
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.9285154342651367,
      "learning_rate": 0.000508925848732426,
      "loss": 2.2822,
      "step": 58701
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.876264750957489,
      "learning_rate": 0.0005089229131827262,
      "loss": 2.1882,
      "step": 58702
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.6983354091644287,
      "learning_rate": 0.0005089199775941837,
      "loss": 2.3553,
      "step": 58703
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.786307156085968,
      "learning_rate": 0.000508917041966799,
      "loss": 2.0949,
      "step": 58704
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7701312899589539,
      "learning_rate": 0.0005089141063005726,
      "loss": 2.3111,
      "step": 58705
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.9284817576408386,
      "learning_rate": 0.0005089111705955051,
      "loss": 2.3166,
      "step": 58706
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.8065298795700073,
      "learning_rate": 0.0005089082348515971,
      "loss": 2.3083,
      "step": 58707
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.8766371607780457,
      "learning_rate": 0.0005089052990688491,
      "loss": 2.218,
      "step": 58708
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.823954701423645,
      "learning_rate": 0.0005089023632472618,
      "loss": 2.0835,
      "step": 58709
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.8319088816642761,
      "learning_rate": 0.0005088994273868355,
      "loss": 2.2001,
      "step": 58710
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7739641070365906,
      "learning_rate": 0.0005088964914875709,
      "loss": 2.1111,
      "step": 58711
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.0493783950805664,
      "learning_rate": 0.0005088935555494685,
      "loss": 2.2129,
      "step": 58712
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7410413026809692,
      "learning_rate": 0.0005088906195725288,
      "loss": 2.1659,
      "step": 58713
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7651513814926147,
      "learning_rate": 0.0005088876835567524,
      "loss": 2.2973,
      "step": 58714
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.9255501627922058,
      "learning_rate": 0.0005088847475021399,
      "loss": 2.3979,
      "step": 58715
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.9195132255554199,
      "learning_rate": 0.0005088818114086917,
      "loss": 2.2492,
      "step": 58716
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7006834149360657,
      "learning_rate": 0.0005088788752764084,
      "loss": 2.1469,
      "step": 58717
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7232257723808289,
      "learning_rate": 0.0005088759391052906,
      "loss": 2.1357,
      "step": 58718
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.054367184638977,
      "learning_rate": 0.0005088730028953389,
      "loss": 2.3417,
      "step": 58719
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.111828088760376,
      "learning_rate": 0.0005088700666465536,
      "loss": 2.1829,
      "step": 58720
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7817121148109436,
      "learning_rate": 0.0005088671303589356,
      "loss": 2.455,
      "step": 58721
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.4183571338653564,
      "learning_rate": 0.0005088641940324852,
      "loss": 2.1167,
      "step": 58722
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.9445133209228516,
      "learning_rate": 0.000508861257667203,
      "loss": 2.3961,
      "step": 58723
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7502378821372986,
      "learning_rate": 0.0005088583212630895,
      "loss": 2.3016,
      "step": 58724
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.8382307291030884,
      "learning_rate": 0.0005088553848201455,
      "loss": 2.2948,
      "step": 58725
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.3392837047576904,
      "learning_rate": 0.000508852448338371,
      "loss": 2.2001,
      "step": 58726
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.083197832107544,
      "learning_rate": 0.0005088495118177671,
      "loss": 2.142,
      "step": 58727
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.716612696647644,
      "learning_rate": 0.0005088465752583341,
      "loss": 2.3001,
      "step": 58728
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.025647521018982,
      "learning_rate": 0.0005088436386600726,
      "loss": 2.1688,
      "step": 58729
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.2888882160186768,
      "learning_rate": 0.0005088407020229831,
      "loss": 2.3477,
      "step": 58730
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.9711860418319702,
      "learning_rate": 0.0005088377653470663,
      "loss": 2.0712,
      "step": 58731
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.9103339910507202,
      "learning_rate": 0.0005088348286323224,
      "loss": 2.2895,
      "step": 58732
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7892279624938965,
      "learning_rate": 0.0005088318918787522,
      "loss": 2.2616,
      "step": 58733
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.1824802160263062,
      "learning_rate": 0.0005088289550863564,
      "loss": 2.1874,
      "step": 58734
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.6954492330551147,
      "learning_rate": 0.0005088260182551351,
      "loss": 2.324,
      "step": 58735
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7615936398506165,
      "learning_rate": 0.0005088230813850893,
      "loss": 2.3305,
      "step": 58736
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.9742361903190613,
      "learning_rate": 0.0005088201444762192,
      "loss": 2.2548,
      "step": 58737
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.8446083068847656,
      "learning_rate": 0.0005088172075285256,
      "loss": 2.1801,
      "step": 58738
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.6832330226898193,
      "learning_rate": 0.0005088142705420088,
      "loss": 2.3649,
      "step": 58739
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.9011044502258301,
      "learning_rate": 0.0005088113335166696,
      "loss": 2.3059,
      "step": 58740
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.8954125642776489,
      "learning_rate": 0.0005088083964525084,
      "loss": 2.3416,
      "step": 58741
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.8000876903533936,
      "learning_rate": 0.0005088054593495258,
      "loss": 2.184,
      "step": 58742
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7769106030464172,
      "learning_rate": 0.0005088025222077224,
      "loss": 2.3392,
      "step": 58743
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.0686644315719604,
      "learning_rate": 0.0005087995850270985,
      "loss": 2.2847,
      "step": 58744
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7191597819328308,
      "learning_rate": 0.000508796647807655,
      "loss": 2.253,
      "step": 58745
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7604290843009949,
      "learning_rate": 0.0005087937105493921,
      "loss": 2.2649,
      "step": 58746
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7376850843429565,
      "learning_rate": 0.0005087907732523106,
      "loss": 2.2318,
      "step": 58747
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.8196538686752319,
      "learning_rate": 0.000508787835916411,
      "loss": 2.1234,
      "step": 58748
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7957730889320374,
      "learning_rate": 0.0005087848985416936,
      "loss": 2.0503,
      "step": 58749
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.6597731113433838,
      "learning_rate": 0.0005087819611281593,
      "loss": 2.2081,
      "step": 58750
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7681751251220703,
      "learning_rate": 0.0005087790236758085,
      "loss": 2.1768,
      "step": 58751
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7719635963439941,
      "learning_rate": 0.0005087760861846416,
      "loss": 2.3064,
      "step": 58752
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7235195636749268,
      "learning_rate": 0.0005087731486546595,
      "loss": 2.1423,
      "step": 58753
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7131325602531433,
      "learning_rate": 0.0005087702110858623,
      "loss": 2.126,
      "step": 58754
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.9249720573425293,
      "learning_rate": 0.0005087672734782509,
      "loss": 2.4673,
      "step": 58755
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7394824028015137,
      "learning_rate": 0.0005087643358318258,
      "loss": 2.218,
      "step": 58756
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7770876288414001,
      "learning_rate": 0.0005087613981465874,
      "loss": 2.367,
      "step": 58757
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7205691337585449,
      "learning_rate": 0.0005087584604225362,
      "loss": 2.1757,
      "step": 58758
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.8712612390518188,
      "learning_rate": 0.000508755522659673,
      "loss": 2.3865,
      "step": 58759
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.8245609998703003,
      "learning_rate": 0.0005087525848579982,
      "loss": 2.2192,
      "step": 58760
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7977557182312012,
      "learning_rate": 0.0005087496470175123,
      "loss": 2.2557,
      "step": 58761
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.8073087334632874,
      "learning_rate": 0.000508746709138216,
      "loss": 2.4892,
      "step": 58762
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7037141919136047,
      "learning_rate": 0.0005087437712201096,
      "loss": 2.1716,
      "step": 58763
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.8246905207633972,
      "learning_rate": 0.0005087408332631938,
      "loss": 2.2862,
      "step": 58764
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.663902223110199,
      "learning_rate": 0.0005087378952674693,
      "loss": 2.4466,
      "step": 58765
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7892654538154602,
      "learning_rate": 0.0005087349572329363,
      "loss": 2.2645,
      "step": 58766
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.6627905368804932,
      "learning_rate": 0.0005087320191595957,
      "loss": 2.1576,
      "step": 58767
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7427956461906433,
      "learning_rate": 0.0005087290810474478,
      "loss": 2.1413,
      "step": 58768
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7315015196800232,
      "learning_rate": 0.0005087261428964932,
      "loss": 2.1767,
      "step": 58769
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7862303853034973,
      "learning_rate": 0.0005087232047067326,
      "loss": 2.3328,
      "step": 58770
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7178751826286316,
      "learning_rate": 0.0005087202664781662,
      "loss": 2.235,
      "step": 58771
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7557379603385925,
      "learning_rate": 0.0005087173282107949,
      "loss": 2.0873,
      "step": 58772
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.730060875415802,
      "learning_rate": 0.0005087143899046191,
      "loss": 2.3164,
      "step": 58773
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.6882561445236206,
      "learning_rate": 0.0005087114515596394,
      "loss": 2.403,
      "step": 58774
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7483329772949219,
      "learning_rate": 0.0005087085131758562,
      "loss": 2.2688,
      "step": 58775
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7132648229598999,
      "learning_rate": 0.0005087055747532701,
      "loss": 2.2183,
      "step": 58776
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7625402212142944,
      "learning_rate": 0.0005087026362918818,
      "loss": 2.3149,
      "step": 58777
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7684158682823181,
      "learning_rate": 0.0005086996977916918,
      "loss": 2.1008,
      "step": 58778
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7372185587882996,
      "learning_rate": 0.0005086967592527004,
      "loss": 2.3235,
      "step": 58779
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7236969470977783,
      "learning_rate": 0.0005086938206749085,
      "loss": 2.055,
      "step": 58780
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.8119413256645203,
      "learning_rate": 0.0005086908820583164,
      "loss": 2.172,
      "step": 58781
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.6983885765075684,
      "learning_rate": 0.0005086879434029248,
      "loss": 2.0166,
      "step": 58782
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7945489883422852,
      "learning_rate": 0.0005086850047087341,
      "loss": 2.3925,
      "step": 58783
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7707653045654297,
      "learning_rate": 0.0005086820659757449,
      "loss": 2.1162,
      "step": 58784
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7185069918632507,
      "learning_rate": 0.0005086791272039578,
      "loss": 2.0603,
      "step": 58785
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.732883095741272,
      "learning_rate": 0.0005086761883933733,
      "loss": 2.043,
      "step": 58786
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7378641366958618,
      "learning_rate": 0.000508673249543992,
      "loss": 2.3146,
      "step": 58787
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7995617985725403,
      "learning_rate": 0.0005086703106558144,
      "loss": 2.2316,
      "step": 58788
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7767847776412964,
      "learning_rate": 0.0005086673717288409,
      "loss": 2.4577,
      "step": 58789
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.790751576423645,
      "learning_rate": 0.0005086644327630724,
      "loss": 2.2576,
      "step": 58790
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.9035463929176331,
      "learning_rate": 0.0005086614937585091,
      "loss": 2.2069,
      "step": 58791
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.705299437046051,
      "learning_rate": 0.0005086585547151518,
      "loss": 2.264,
      "step": 58792
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7338820099830627,
      "learning_rate": 0.0005086556156330009,
      "loss": 1.9944,
      "step": 58793
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7735067009925842,
      "learning_rate": 0.0005086526765120568,
      "loss": 2.0352,
      "step": 58794
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7617096900939941,
      "learning_rate": 0.0005086497373523205,
      "loss": 2.2651,
      "step": 58795
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.8060019016265869,
      "learning_rate": 0.0005086467981537922,
      "loss": 2.2789,
      "step": 58796
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7994581460952759,
      "learning_rate": 0.0005086438589164725,
      "loss": 2.1229,
      "step": 58797
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.8713759183883667,
      "learning_rate": 0.000508640919640362,
      "loss": 2.3958,
      "step": 58798
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.8288728594779968,
      "learning_rate": 0.000508637980325461,
      "loss": 2.2415,
      "step": 58799
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7380298972129822,
      "learning_rate": 0.0005086350409717706,
      "loss": 2.1397,
      "step": 58800
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7308512330055237,
      "learning_rate": 0.0005086321015792908,
      "loss": 2.3882,
      "step": 58801
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7618637681007385,
      "learning_rate": 0.0005086291621480225,
      "loss": 2.175,
      "step": 58802
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.6803237199783325,
      "learning_rate": 0.0005086262226779659,
      "loss": 2.3933,
      "step": 58803
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.750930905342102,
      "learning_rate": 0.0005086232831691219,
      "loss": 2.2086,
      "step": 58804
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.6886219382286072,
      "learning_rate": 0.0005086203436214907,
      "loss": 2.3309,
      "step": 58805
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7228663563728333,
      "learning_rate": 0.0005086174040350732,
      "loss": 2.2683,
      "step": 58806
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7588908672332764,
      "learning_rate": 0.0005086144644098698,
      "loss": 2.2443,
      "step": 58807
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7567117810249329,
      "learning_rate": 0.000508611524745881,
      "loss": 2.2348,
      "step": 58808
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.6422680020332336,
      "learning_rate": 0.0005086085850431074,
      "loss": 2.0755,
      "step": 58809
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.8214638233184814,
      "learning_rate": 0.0005086056453015496,
      "loss": 2.2336,
      "step": 58810
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7147426009178162,
      "learning_rate": 0.0005086027055212079,
      "loss": 2.1887,
      "step": 58811
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7450968027114868,
      "learning_rate": 0.0005085997657020831,
      "loss": 2.2028,
      "step": 58812
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7316017746925354,
      "learning_rate": 0.0005085968258441758,
      "loss": 2.2054,
      "step": 58813
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7057927846908569,
      "learning_rate": 0.0005085938859474862,
      "loss": 2.3004,
      "step": 58814
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7651185393333435,
      "learning_rate": 0.0005085909460120151,
      "loss": 2.1285,
      "step": 58815
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.6802278757095337,
      "learning_rate": 0.0005085880060377631,
      "loss": 2.1709,
      "step": 58816
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.727746307849884,
      "learning_rate": 0.0005085850660247306,
      "loss": 2.1088,
      "step": 58817
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.8289254903793335,
      "learning_rate": 0.0005085821259729181,
      "loss": 2.6018,
      "step": 58818
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7545258402824402,
      "learning_rate": 0.0005085791858823263,
      "loss": 2.0347,
      "step": 58819
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.766076922416687,
      "learning_rate": 0.0005085762457529558,
      "loss": 2.2166,
      "step": 58820
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.6885440349578857,
      "learning_rate": 0.0005085733055848069,
      "loss": 2.195,
      "step": 58821
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7173672914505005,
      "learning_rate": 0.0005085703653778804,
      "loss": 2.2592,
      "step": 58822
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.6641541719436646,
      "learning_rate": 0.0005085674251321768,
      "loss": 2.4142,
      "step": 58823
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7148734331130981,
      "learning_rate": 0.0005085644848476964,
      "loss": 2.2195,
      "step": 58824
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7044141888618469,
      "learning_rate": 0.00050856154452444,
      "loss": 2.2051,
      "step": 58825
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.752794086933136,
      "learning_rate": 0.000508558604162408,
      "loss": 2.3206,
      "step": 58826
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7367383241653442,
      "learning_rate": 0.0005085556637616012,
      "loss": 2.274,
      "step": 58827
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.781291127204895,
      "learning_rate": 0.0005085527233220197,
      "loss": 2.3778,
      "step": 58828
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7031127214431763,
      "learning_rate": 0.0005085497828436645,
      "loss": 2.1706,
      "step": 58829
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7744466066360474,
      "learning_rate": 0.0005085468423265359,
      "loss": 2.6262,
      "step": 58830
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.737097442150116,
      "learning_rate": 0.0005085439017706346,
      "loss": 2.374,
      "step": 58831
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7294697761535645,
      "learning_rate": 0.000508540961175961,
      "loss": 2.1889,
      "step": 58832
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.6757153272628784,
      "learning_rate": 0.0005085380205425156,
      "loss": 2.088,
      "step": 58833
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7872212529182434,
      "learning_rate": 0.0005085350798702991,
      "loss": 2.3656,
      "step": 58834
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7652127146720886,
      "learning_rate": 0.000508532139159312,
      "loss": 2.2625,
      "step": 58835
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7210112810134888,
      "learning_rate": 0.0005085291984095549,
      "loss": 2.4622,
      "step": 58836
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.741515576839447,
      "learning_rate": 0.0005085262576210282,
      "loss": 1.9348,
      "step": 58837
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.6675577759742737,
      "learning_rate": 0.0005085233167937326,
      "loss": 2.1427,
      "step": 58838
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7065921425819397,
      "learning_rate": 0.0005085203759276685,
      "loss": 2.4992,
      "step": 58839
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7007772326469421,
      "learning_rate": 0.0005085174350228365,
      "loss": 1.9507,
      "step": 58840
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7559874057769775,
      "learning_rate": 0.0005085144940792374,
      "loss": 2.1909,
      "step": 58841
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.8050852417945862,
      "learning_rate": 0.0005085115530968714,
      "loss": 2.2839,
      "step": 58842
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7121101021766663,
      "learning_rate": 0.000508508612075739,
      "loss": 2.1333,
      "step": 58843
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7799914479255676,
      "learning_rate": 0.0005085056710158411,
      "loss": 2.1273,
      "step": 58844
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.9131612181663513,
      "learning_rate": 0.000508502729917178,
      "loss": 2.385,
      "step": 58845
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.8583283424377441,
      "learning_rate": 0.0005084997887797503,
      "loss": 2.1702,
      "step": 58846
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7551153302192688,
      "learning_rate": 0.0005084968476035585,
      "loss": 2.2569,
      "step": 58847
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7013691067695618,
      "learning_rate": 0.0005084939063886033,
      "loss": 2.104,
      "step": 58848
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7854438424110413,
      "learning_rate": 0.0005084909651348852,
      "loss": 2.4049,
      "step": 58849
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7478964924812317,
      "learning_rate": 0.0005084880238424046,
      "loss": 2.1175,
      "step": 58850
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7352229356765747,
      "learning_rate": 0.0005084850825111621,
      "loss": 2.0016,
      "step": 58851
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7580346465110779,
      "learning_rate": 0.0005084821411411583,
      "loss": 2.1817,
      "step": 58852
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.729246199131012,
      "learning_rate": 0.0005084791997323939,
      "loss": 2.4742,
      "step": 58853
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.6876649856567383,
      "learning_rate": 0.0005084762582848691,
      "loss": 2.4079,
      "step": 58854
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.8409432768821716,
      "learning_rate": 0.0005084733167985846,
      "loss": 2.0891,
      "step": 58855
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7795807123184204,
      "learning_rate": 0.0005084703752735411,
      "loss": 2.1592,
      "step": 58856
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.6996034979820251,
      "learning_rate": 0.000508467433709739,
      "loss": 2.0437,
      "step": 58857
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.827586829662323,
      "learning_rate": 0.0005084644921071788,
      "loss": 2.1656,
      "step": 58858
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7096595764160156,
      "learning_rate": 0.0005084615504658612,
      "loss": 2.2399,
      "step": 58859
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7751711010932922,
      "learning_rate": 0.0005084586087857866,
      "loss": 2.1917,
      "step": 58860
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.8911451697349548,
      "learning_rate": 0.0005084556670669557,
      "loss": 2.1566,
      "step": 58861
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.9176905155181885,
      "learning_rate": 0.0005084527253093688,
      "loss": 2.5082,
      "step": 58862
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7118179798126221,
      "learning_rate": 0.0005084497835130267,
      "loss": 2.253,
      "step": 58863
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.846802830696106,
      "learning_rate": 0.0005084468416779298,
      "loss": 2.2703,
      "step": 58864
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.8412085175514221,
      "learning_rate": 0.0005084438998040788,
      "loss": 2.3811,
      "step": 58865
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7243956923484802,
      "learning_rate": 0.0005084409578914741,
      "loss": 2.1946,
      "step": 58866
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.816268265247345,
      "learning_rate": 0.0005084380159401164,
      "loss": 2.3009,
      "step": 58867
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.2058924436569214,
      "learning_rate": 0.000508435073950006,
      "loss": 2.3116,
      "step": 58868
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7551348805427551,
      "learning_rate": 0.0005084321319211436,
      "loss": 2.2767,
      "step": 58869
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7982301115989685,
      "learning_rate": 0.0005084291898535296,
      "loss": 2.2456,
      "step": 58870
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.8663001656532288,
      "learning_rate": 0.0005084262477471649,
      "loss": 2.2549,
      "step": 58871
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7240513563156128,
      "learning_rate": 0.0005084233056020498,
      "loss": 2.2895,
      "step": 58872
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7186712026596069,
      "learning_rate": 0.0005084203634181847,
      "loss": 2.2733,
      "step": 58873
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.8212109804153442,
      "learning_rate": 0.0005084174211955704,
      "loss": 2.1699,
      "step": 58874
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.8062954545021057,
      "learning_rate": 0.0005084144789342075,
      "loss": 2.2613,
      "step": 58875
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.6488468647003174,
      "learning_rate": 0.0005084115366340962,
      "loss": 2.1769,
      "step": 58876
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7459312081336975,
      "learning_rate": 0.0005084085942952375,
      "loss": 2.4587,
      "step": 58877
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7765868306159973,
      "learning_rate": 0.0005084056519176316,
      "loss": 2.3544,
      "step": 58878
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.6823466420173645,
      "learning_rate": 0.000508402709501279,
      "loss": 2.3016,
      "step": 58879
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7552608847618103,
      "learning_rate": 0.0005083997670461806,
      "loss": 2.1936,
      "step": 58880
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7533353567123413,
      "learning_rate": 0.0005083968245523367,
      "loss": 2.2862,
      "step": 58881
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.725002646446228,
      "learning_rate": 0.0005083938820197479,
      "loss": 2.3161,
      "step": 58882
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7612528800964355,
      "learning_rate": 0.0005083909394484147,
      "loss": 2.4007,
      "step": 58883
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7778780460357666,
      "learning_rate": 0.0005083879968383377,
      "loss": 2.1456,
      "step": 58884
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7340990900993347,
      "learning_rate": 0.0005083850541895175,
      "loss": 2.1045,
      "step": 58885
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.8200554251670837,
      "learning_rate": 0.0005083821115019545,
      "loss": 2.3658,
      "step": 58886
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7524300217628479,
      "learning_rate": 0.0005083791687756494,
      "loss": 2.397,
      "step": 58887
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7014597058296204,
      "learning_rate": 0.0005083762260106026,
      "loss": 2.105,
      "step": 58888
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7153162956237793,
      "learning_rate": 0.0005083732832068148,
      "loss": 2.375,
      "step": 58889
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.6960328817367554,
      "learning_rate": 0.0005083703403642865,
      "loss": 2.2798,
      "step": 58890
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7738122344017029,
      "learning_rate": 0.0005083673974830182,
      "loss": 2.3493,
      "step": 58891
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7653378844261169,
      "learning_rate": 0.0005083644545630104,
      "loss": 2.2855,
      "step": 58892
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.721152126789093,
      "learning_rate": 0.0005083615116042638,
      "loss": 2.2936,
      "step": 58893
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7211809158325195,
      "learning_rate": 0.0005083585686067787,
      "loss": 2.185,
      "step": 58894
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7350467443466187,
      "learning_rate": 0.000508355625570556,
      "loss": 2.1675,
      "step": 58895
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.8087567687034607,
      "learning_rate": 0.0005083526824955961,
      "loss": 2.1442,
      "step": 58896
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7181373834609985,
      "learning_rate": 0.0005083497393818993,
      "loss": 2.2691,
      "step": 58897
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7619695067405701,
      "learning_rate": 0.0005083467962294665,
      "loss": 2.4631,
      "step": 58898
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.754814088344574,
      "learning_rate": 0.000508343853038298,
      "loss": 2.3007,
      "step": 58899
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.6882510185241699,
      "learning_rate": 0.0005083409098083944,
      "loss": 2.4207,
      "step": 58900
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7532076835632324,
      "learning_rate": 0.0005083379665397564,
      "loss": 2.3644,
      "step": 58901
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7151617407798767,
      "learning_rate": 0.0005083350232323844,
      "loss": 2.1972,
      "step": 58902
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7889677882194519,
      "learning_rate": 0.000508332079886279,
      "loss": 2.1257,
      "step": 58903
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7925182580947876,
      "learning_rate": 0.0005083291365014408,
      "loss": 2.2373,
      "step": 58904
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.756264865398407,
      "learning_rate": 0.0005083261930778702,
      "loss": 2.2814,
      "step": 58905
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7375267744064331,
      "learning_rate": 0.0005083232496155678,
      "loss": 2.1707,
      "step": 58906
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7578607201576233,
      "learning_rate": 0.0005083203061145343,
      "loss": 2.3568,
      "step": 58907
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7041258811950684,
      "learning_rate": 0.00050831736257477,
      "loss": 2.2826,
      "step": 58908
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.6775592565536499,
      "learning_rate": 0.0005083144189962756,
      "loss": 2.308,
      "step": 58909
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7625934481620789,
      "learning_rate": 0.0005083114753790516,
      "loss": 2.1788,
      "step": 58910
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7911788821220398,
      "learning_rate": 0.0005083085317230986,
      "loss": 2.3247,
      "step": 58911
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7341009974479675,
      "learning_rate": 0.0005083055880284171,
      "loss": 2.2497,
      "step": 58912
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7683049440383911,
      "learning_rate": 0.0005083026442950077,
      "loss": 2.3586,
      "step": 58913
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7462707757949829,
      "learning_rate": 0.0005082997005228709,
      "loss": 2.2707,
      "step": 58914
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.6696059107780457,
      "learning_rate": 0.0005082967567120073,
      "loss": 2.0739,
      "step": 58915
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7916483283042908,
      "learning_rate": 0.0005082938128624172,
      "loss": 2.367,
      "step": 58916
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.909282922744751,
      "learning_rate": 0.0005082908689741014,
      "loss": 2.2558,
      "step": 58917
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7296378016471863,
      "learning_rate": 0.0005082879250470605,
      "loss": 1.9711,
      "step": 58918
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.73218834400177,
      "learning_rate": 0.000508284981081295,
      "loss": 2.1972,
      "step": 58919
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.8195323348045349,
      "learning_rate": 0.0005082820370768053,
      "loss": 2.4105,
      "step": 58920
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.746373176574707,
      "learning_rate": 0.000508279093033592,
      "loss": 2.1958,
      "step": 58921
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7237393856048584,
      "learning_rate": 0.0005082761489516557,
      "loss": 2.4803,
      "step": 58922
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.754034698009491,
      "learning_rate": 0.000508273204830997,
      "loss": 2.1133,
      "step": 58923
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7304043173789978,
      "learning_rate": 0.0005082702606716162,
      "loss": 2.4393,
      "step": 58924
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.8496737480163574,
      "learning_rate": 0.0005082673164735144,
      "loss": 2.3112,
      "step": 58925
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.8207365274429321,
      "learning_rate": 0.0005082643722366915,
      "loss": 2.3715,
      "step": 58926
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.706416130065918,
      "learning_rate": 0.0005082614279611484,
      "loss": 2.0589,
      "step": 58927
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7045131325721741,
      "learning_rate": 0.0005082584836468854,
      "loss": 2.1187,
      "step": 58928
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.717199981212616,
      "learning_rate": 0.0005082555392939034,
      "loss": 2.3059,
      "step": 58929
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7398940324783325,
      "learning_rate": 0.0005082525949022027,
      "loss": 2.0881,
      "step": 58930
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.8029861450195312,
      "learning_rate": 0.0005082496504717839,
      "loss": 2.0537,
      "step": 58931
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.923431932926178,
      "learning_rate": 0.0005082467060026476,
      "loss": 2.3101,
      "step": 58932
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7983598709106445,
      "learning_rate": 0.0005082437614947942,
      "loss": 2.262,
      "step": 58933
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.8107402920722961,
      "learning_rate": 0.0005082408169482244,
      "loss": 2.1447,
      "step": 58934
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.6884527206420898,
      "learning_rate": 0.0005082378723629389,
      "loss": 2.0334,
      "step": 58935
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.8741264343261719,
      "learning_rate": 0.0005082349277389379,
      "loss": 2.2844,
      "step": 58936
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.1796891689300537,
      "learning_rate": 0.000508231983076222,
      "loss": 2.1778,
      "step": 58937
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7982633709907532,
      "learning_rate": 0.000508229038374792,
      "loss": 2.6587,
      "step": 58938
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.8099743127822876,
      "learning_rate": 0.0005082260936346482,
      "loss": 2.3747,
      "step": 58939
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7824429869651794,
      "learning_rate": 0.0005082231488557912,
      "loss": 2.2371,
      "step": 58940
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7671394348144531,
      "learning_rate": 0.0005082202040382217,
      "loss": 2.3847,
      "step": 58941
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.9270867109298706,
      "learning_rate": 0.0005082172591819401,
      "loss": 2.3875,
      "step": 58942
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7653167247772217,
      "learning_rate": 0.0005082143142869469,
      "loss": 2.2778,
      "step": 58943
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7029701471328735,
      "learning_rate": 0.0005082113693532428,
      "loss": 2.4083,
      "step": 58944
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.012208104133606,
      "learning_rate": 0.0005082084243808283,
      "loss": 2.2253,
      "step": 58945
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.6788871884346008,
      "learning_rate": 0.0005082054793697039,
      "loss": 2.4095,
      "step": 58946
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7181406021118164,
      "learning_rate": 0.0005082025343198702,
      "loss": 2.2783,
      "step": 58947
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.6922894716262817,
      "learning_rate": 0.0005081995892313277,
      "loss": 2.1348,
      "step": 58948
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7195996046066284,
      "learning_rate": 0.0005081966441040769,
      "loss": 2.3362,
      "step": 58949
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7138350009918213,
      "learning_rate": 0.0005081936989381185,
      "loss": 2.217,
      "step": 58950
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.6788986921310425,
      "learning_rate": 0.0005081907537334529,
      "loss": 2.0225,
      "step": 58951
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7776631116867065,
      "learning_rate": 0.0005081878084900807,
      "loss": 2.2254,
      "step": 58952
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7677388191223145,
      "learning_rate": 0.0005081848632080024,
      "loss": 2.4294,
      "step": 58953
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7756702899932861,
      "learning_rate": 0.0005081819178872188,
      "loss": 2.2758,
      "step": 58954
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.8067654967308044,
      "learning_rate": 0.0005081789725277301,
      "loss": 2.3707,
      "step": 58955
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.776949405670166,
      "learning_rate": 0.0005081760271295371,
      "loss": 2.4866,
      "step": 58956
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.850820779800415,
      "learning_rate": 0.0005081730816926402,
      "loss": 2.105,
      "step": 58957
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7612977027893066,
      "learning_rate": 0.0005081701362170399,
      "loss": 2.292,
      "step": 58958
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7541216015815735,
      "learning_rate": 0.0005081671907027369,
      "loss": 2.1796,
      "step": 58959
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7393824458122253,
      "learning_rate": 0.0005081642451497319,
      "loss": 2.0627,
      "step": 58960
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7910395264625549,
      "learning_rate": 0.0005081612995580249,
      "loss": 2.2781,
      "step": 58961
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7801690697669983,
      "learning_rate": 0.0005081583539276169,
      "loss": 2.3249,
      "step": 58962
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.8134718537330627,
      "learning_rate": 0.0005081554082585084,
      "loss": 2.0198,
      "step": 58963
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7094494104385376,
      "learning_rate": 0.0005081524625506999,
      "loss": 2.3102,
      "step": 58964
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.8127090930938721,
      "learning_rate": 0.0005081495168041918,
      "loss": 2.1893,
      "step": 58965
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.8112829327583313,
      "learning_rate": 0.0005081465710189849,
      "loss": 2.0459,
      "step": 58966
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.8167968392372131,
      "learning_rate": 0.0005081436251950796,
      "loss": 2.3846,
      "step": 58967
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.8321467638015747,
      "learning_rate": 0.0005081406793324765,
      "loss": 2.2358,
      "step": 58968
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7335651516914368,
      "learning_rate": 0.000508137733431176,
      "loss": 2.0713,
      "step": 58969
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.8124408721923828,
      "learning_rate": 0.0005081347874911789,
      "loss": 2.3352,
      "step": 58970
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7812764644622803,
      "learning_rate": 0.0005081318415124854,
      "loss": 2.1653,
      "step": 58971
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7073390483856201,
      "learning_rate": 0.0005081288954950965,
      "loss": 2.1581,
      "step": 58972
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7741813063621521,
      "learning_rate": 0.0005081259494390124,
      "loss": 2.1977,
      "step": 58973
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7987591624259949,
      "learning_rate": 0.0005081230033442338,
      "loss": 2.0004,
      "step": 58974
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.722805380821228,
      "learning_rate": 0.0005081200572107611,
      "loss": 2.0221,
      "step": 58975
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7632238268852234,
      "learning_rate": 0.0005081171110385951,
      "loss": 2.1493,
      "step": 58976
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7219365835189819,
      "learning_rate": 0.0005081141648277361,
      "loss": 2.2949,
      "step": 58977
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.8023959398269653,
      "learning_rate": 0.0005081112185781848,
      "loss": 2.1988,
      "step": 58978
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.74831223487854,
      "learning_rate": 0.0005081082722899416,
      "loss": 2.2575,
      "step": 58979
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7678298950195312,
      "learning_rate": 0.0005081053259630074,
      "loss": 2.0815,
      "step": 58980
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7530778646469116,
      "learning_rate": 0.0005081023795973822,
      "loss": 2.0176,
      "step": 58981
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.6533644199371338,
      "learning_rate": 0.000508099433193067,
      "loss": 1.9453,
      "step": 58982
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.83073890209198,
      "learning_rate": 0.0005080964867500621,
      "loss": 2.218,
      "step": 58983
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.0120691061019897,
      "learning_rate": 0.0005080935402683682,
      "loss": 2.1422,
      "step": 58984
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7687348127365112,
      "learning_rate": 0.0005080905937479858,
      "loss": 2.3078,
      "step": 58985
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.8491095900535583,
      "learning_rate": 0.0005080876471889154,
      "loss": 2.2478,
      "step": 58986
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7629051804542542,
      "learning_rate": 0.0005080847005911576,
      "loss": 2.2431,
      "step": 58987
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7019540667533875,
      "learning_rate": 0.0005080817539547129,
      "loss": 2.1385,
      "step": 58988
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.732904851436615,
      "learning_rate": 0.0005080788072795817,
      "loss": 2.0141,
      "step": 58989
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.8249638676643372,
      "learning_rate": 0.000508075860565765,
      "loss": 2.4289,
      "step": 58990
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7430707216262817,
      "learning_rate": 0.0005080729138132629,
      "loss": 2.2609,
      "step": 58991
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7145871520042419,
      "learning_rate": 0.0005080699670220762,
      "loss": 2.2164,
      "step": 58992
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7887634038925171,
      "learning_rate": 0.0005080670201922054,
      "loss": 2.1368,
      "step": 58993
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.8713659048080444,
      "learning_rate": 0.0005080640733236509,
      "loss": 2.4618,
      "step": 58994
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.0914416313171387,
      "learning_rate": 0.0005080611264164134,
      "loss": 2.347,
      "step": 58995
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.732932984828949,
      "learning_rate": 0.0005080581794704935,
      "loss": 2.3469,
      "step": 58996
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7226306796073914,
      "learning_rate": 0.0005080552324858915,
      "loss": 2.2166,
      "step": 58997
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7439442873001099,
      "learning_rate": 0.0005080522854626082,
      "loss": 2.1045,
      "step": 58998
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7402875423431396,
      "learning_rate": 0.0005080493384006441,
      "loss": 2.1379,
      "step": 58999
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.6810082793235779,
      "learning_rate": 0.0005080463912999996,
      "loss": 2.3589,
      "step": 59000
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.6711625456809998,
      "learning_rate": 0.0005080434441606754,
      "loss": 2.0534,
      "step": 59001
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.8292039036750793,
      "learning_rate": 0.000508040496982672,
      "loss": 2.3247,
      "step": 59002
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7648800611495972,
      "learning_rate": 0.00050803754976599,
      "loss": 2.1161,
      "step": 59003
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7584372162818909,
      "learning_rate": 0.0005080346025106298,
      "loss": 2.4326,
      "step": 59004
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.8141154646873474,
      "learning_rate": 0.0005080316552165921,
      "loss": 2.116,
      "step": 59005
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7788195610046387,
      "learning_rate": 0.0005080287078838774,
      "loss": 2.2033,
      "step": 59006
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7291262149810791,
      "learning_rate": 0.0005080257605124862,
      "loss": 2.2376,
      "step": 59007
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7666473388671875,
      "learning_rate": 0.000508022813102419,
      "loss": 2.2608,
      "step": 59008
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.8236591219902039,
      "learning_rate": 0.0005080198656536765,
      "loss": 2.2481,
      "step": 59009
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.6876682639122009,
      "learning_rate": 0.0005080169181662592,
      "loss": 2.149,
      "step": 59010
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.8273071050643921,
      "learning_rate": 0.0005080139706401676,
      "loss": 2.107,
      "step": 59011
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7187548279762268,
      "learning_rate": 0.0005080110230754023,
      "loss": 2.3339,
      "step": 59012
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7831263542175293,
      "learning_rate": 0.0005080080754719638,
      "loss": 2.3214,
      "step": 59013
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7049665451049805,
      "learning_rate": 0.0005080051278298527,
      "loss": 2.1996,
      "step": 59014
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.8253951668739319,
      "learning_rate": 0.0005080021801490695,
      "loss": 2.0633,
      "step": 59015
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7585835456848145,
      "learning_rate": 0.0005079992324296148,
      "loss": 2.4676,
      "step": 59016
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7618991136550903,
      "learning_rate": 0.0005079962846714891,
      "loss": 2.2474,
      "step": 59017
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7837296724319458,
      "learning_rate": 0.000507993336874693,
      "loss": 2.4494,
      "step": 59018
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7094671130180359,
      "learning_rate": 0.0005079903890392269,
      "loss": 2.283,
      "step": 59019
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7800607681274414,
      "learning_rate": 0.0005079874411650915,
      "loss": 2.1086,
      "step": 59020
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.6622763276100159,
      "learning_rate": 0.0005079844932522874,
      "loss": 2.3624,
      "step": 59021
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.6787272691726685,
      "learning_rate": 0.0005079815453008149,
      "loss": 2.3985,
      "step": 59022
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.8130649328231812,
      "learning_rate": 0.0005079785973106748,
      "loss": 2.1317,
      "step": 59023
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7939289212226868,
      "learning_rate": 0.0005079756492818677,
      "loss": 2.4033,
      "step": 59024
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.6792850494384766,
      "learning_rate": 0.0005079727012143938,
      "loss": 2.2356,
      "step": 59025
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7035729885101318,
      "learning_rate": 0.000507969753108254,
      "loss": 2.1549,
      "step": 59026
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7626556158065796,
      "learning_rate": 0.0005079668049634485,
      "loss": 2.1153,
      "step": 59027
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7602910995483398,
      "learning_rate": 0.0005079638567799782,
      "loss": 2.339,
      "step": 59028
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7277542948722839,
      "learning_rate": 0.0005079609085578433,
      "loss": 2.204,
      "step": 59029
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.740246057510376,
      "learning_rate": 0.0005079579602970448,
      "loss": 2.3105,
      "step": 59030
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7184795141220093,
      "learning_rate": 0.0005079550119975829,
      "loss": 2.1794,
      "step": 59031
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.751796543598175,
      "learning_rate": 0.0005079520636594581,
      "loss": 2.0292,
      "step": 59032
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7510724067687988,
      "learning_rate": 0.0005079491152826713,
      "loss": 2.2513,
      "step": 59033
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7358691096305847,
      "learning_rate": 0.0005079461668672226,
      "loss": 2.1456,
      "step": 59034
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7960256934165955,
      "learning_rate": 0.000507943218413113,
      "loss": 2.1558,
      "step": 59035
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7779023051261902,
      "learning_rate": 0.0005079402699203427,
      "loss": 2.4796,
      "step": 59036
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7056927680969238,
      "learning_rate": 0.0005079373213889124,
      "loss": 2.0752,
      "step": 59037
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7310534119606018,
      "learning_rate": 0.0005079343728188227,
      "loss": 2.1057,
      "step": 59038
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7349454164505005,
      "learning_rate": 0.0005079314242100741,
      "loss": 2.065,
      "step": 59039
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.794204831123352,
      "learning_rate": 0.000507928475562667,
      "loss": 2.4377,
      "step": 59040
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7444190979003906,
      "learning_rate": 0.0005079255268766021,
      "loss": 2.1531,
      "step": 59041
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7047468423843384,
      "learning_rate": 0.00050792257815188,
      "loss": 2.1557,
      "step": 59042
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.765148937702179,
      "learning_rate": 0.000507919629388501,
      "loss": 2.3424,
      "step": 59043
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7095658183097839,
      "learning_rate": 0.000507916680586466,
      "loss": 2.32,
      "step": 59044
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7932830452919006,
      "learning_rate": 0.0005079137317457752,
      "loss": 2.2491,
      "step": 59045
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7272497415542603,
      "learning_rate": 0.0005079107828664294,
      "loss": 2.2193,
      "step": 59046
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.8413326740264893,
      "learning_rate": 0.000507907833948429,
      "loss": 2.2456,
      "step": 59047
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7215062379837036,
      "learning_rate": 0.0005079048849917748,
      "loss": 2.3704,
      "step": 59048
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7518735527992249,
      "learning_rate": 0.000507901935996467,
      "loss": 2.3026,
      "step": 59049
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7478090524673462,
      "learning_rate": 0.0005078989869625063,
      "loss": 2.3753,
      "step": 59050
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.738618016242981,
      "learning_rate": 0.0005078960378898933,
      "loss": 2.0546,
      "step": 59051
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7144178152084351,
      "learning_rate": 0.0005078930887786285,
      "loss": 2.4276,
      "step": 59052
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7059963941574097,
      "learning_rate": 0.0005078901396287123,
      "loss": 2.3316,
      "step": 59053
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7699936628341675,
      "learning_rate": 0.0005078871904401457,
      "loss": 2.3535,
      "step": 59054
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7004154324531555,
      "learning_rate": 0.0005078842412129288,
      "loss": 2.4009,
      "step": 59055
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7325984239578247,
      "learning_rate": 0.0005078812919470623,
      "loss": 2.1321,
      "step": 59056
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7108476758003235,
      "learning_rate": 0.0005078783426425467,
      "loss": 2.3429,
      "step": 59057
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.6881093382835388,
      "learning_rate": 0.0005078753932993826,
      "loss": 2.1511,
      "step": 59058
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7525532245635986,
      "learning_rate": 0.0005078724439175705,
      "loss": 2.2693,
      "step": 59059
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7672956585884094,
      "learning_rate": 0.000507869494497111,
      "loss": 2.0819,
      "step": 59060
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7806435227394104,
      "learning_rate": 0.0005078665450380048,
      "loss": 2.1155,
      "step": 59061
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.8104906678199768,
      "learning_rate": 0.0005078635955402521,
      "loss": 2.1144,
      "step": 59062
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.8097268342971802,
      "learning_rate": 0.0005078606460038538,
      "loss": 2.2542,
      "step": 59063
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7917258143424988,
      "learning_rate": 0.00050785769642881,
      "loss": 2.4324,
      "step": 59064
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.6699506044387817,
      "learning_rate": 0.0005078547468151219,
      "loss": 2.0615,
      "step": 59065
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7670429348945618,
      "learning_rate": 0.0005078517971627895,
      "loss": 2.4419,
      "step": 59066
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.8347891569137573,
      "learning_rate": 0.0005078488474718134,
      "loss": 2.2262,
      "step": 59067
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.6967252492904663,
      "learning_rate": 0.0005078458977421944,
      "loss": 2.2446,
      "step": 59068
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7081407904624939,
      "learning_rate": 0.000507842947973933,
      "loss": 2.3671,
      "step": 59069
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.8576191663742065,
      "learning_rate": 0.0005078399981670296,
      "loss": 2.168,
      "step": 59070
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7237273454666138,
      "learning_rate": 0.0005078370483214847,
      "loss": 2.1042,
      "step": 59071
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.8631111979484558,
      "learning_rate": 0.0005078340984372992,
      "loss": 2.3289,
      "step": 59072
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.8107332587242126,
      "learning_rate": 0.0005078311485144732,
      "loss": 2.2464,
      "step": 59073
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7935788631439209,
      "learning_rate": 0.0005078281985530076,
      "loss": 2.2409,
      "step": 59074
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7527799010276794,
      "learning_rate": 0.0005078252485529028,
      "loss": 2.4211,
      "step": 59075
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7876417636871338,
      "learning_rate": 0.0005078222985141594,
      "loss": 2.2709,
      "step": 59076
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.9027937054634094,
      "learning_rate": 0.0005078193484367778,
      "loss": 2.1777,
      "step": 59077
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.9064510464668274,
      "learning_rate": 0.0005078163983207588,
      "loss": 2.0708,
      "step": 59078
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.745392382144928,
      "learning_rate": 0.0005078134481661026,
      "loss": 2.316,
      "step": 59079
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.9691567420959473,
      "learning_rate": 0.0005078104979728102,
      "loss": 2.2937,
      "step": 59080
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.046181082725525,
      "learning_rate": 0.0005078075477408817,
      "loss": 2.4076,
      "step": 59081
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.8138407468795776,
      "learning_rate": 0.000507804597470318,
      "loss": 2.1267,
      "step": 59082
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.904628574848175,
      "learning_rate": 0.0005078016471611194,
      "loss": 2.3309,
      "step": 59083
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.8927025198936462,
      "learning_rate": 0.0005077986968132866,
      "loss": 2.22,
      "step": 59084
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.6849740743637085,
      "learning_rate": 0.0005077957464268201,
      "loss": 2.1453,
      "step": 59085
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.6905501484870911,
      "learning_rate": 0.0005077927960017204,
      "loss": 2.3575,
      "step": 59086
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.8408644199371338,
      "learning_rate": 0.0005077898455379882,
      "loss": 2.3051,
      "step": 59087
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.8789063692092896,
      "learning_rate": 0.0005077868950356239,
      "loss": 2.2928,
      "step": 59088
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7113562226295471,
      "learning_rate": 0.0005077839444946281,
      "loss": 2.2068,
      "step": 59089
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7222071886062622,
      "learning_rate": 0.0005077809939150012,
      "loss": 2.2512,
      "step": 59090
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.8289010524749756,
      "learning_rate": 0.0005077780432967441,
      "loss": 2.3665,
      "step": 59091
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.8531068563461304,
      "learning_rate": 0.0005077750926398571,
      "loss": 2.2779,
      "step": 59092
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.696019172668457,
      "learning_rate": 0.0005077721419443406,
      "loss": 2.1418,
      "step": 59093
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.8163362145423889,
      "learning_rate": 0.0005077691912101956,
      "loss": 2.3893,
      "step": 59094
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.8470813632011414,
      "learning_rate": 0.0005077662404374222,
      "loss": 2.1033,
      "step": 59095
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.8719649910926819,
      "learning_rate": 0.0005077632896260211,
      "loss": 2.0833,
      "step": 59096
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7596974968910217,
      "learning_rate": 0.0005077603387759932,
      "loss": 2.1542,
      "step": 59097
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7886125445365906,
      "learning_rate": 0.0005077573878873384,
      "loss": 2.3776,
      "step": 59098
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7669509649276733,
      "learning_rate": 0.0005077544369600577,
      "loss": 2.2486,
      "step": 59099
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.727316677570343,
      "learning_rate": 0.0005077514859941515,
      "loss": 2.0858,
      "step": 59100
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7694956064224243,
      "learning_rate": 0.0005077485349896205,
      "loss": 2.3891,
      "step": 59101
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.8027976751327515,
      "learning_rate": 0.000507745583946465,
      "loss": 2.2253,
      "step": 59102
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7326863408088684,
      "learning_rate": 0.0005077426328646856,
      "loss": 2.3183,
      "step": 59103
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7601032257080078,
      "learning_rate": 0.0005077396817442831,
      "loss": 2.2824,
      "step": 59104
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.863503634929657,
      "learning_rate": 0.0005077367305852577,
      "loss": 2.2498,
      "step": 59105
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7090941667556763,
      "learning_rate": 0.0005077337793876102,
      "loss": 2.1714,
      "step": 59106
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7268595099449158,
      "learning_rate": 0.0005077308281513412,
      "loss": 2.2524,
      "step": 59107
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7776985168457031,
      "learning_rate": 0.0005077278768764509,
      "loss": 2.1593,
      "step": 59108
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.774558424949646,
      "learning_rate": 0.0005077249255629403,
      "loss": 2.1968,
      "step": 59109
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7115839719772339,
      "learning_rate": 0.0005077219742108095,
      "loss": 2.2681,
      "step": 59110
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.856924295425415,
      "learning_rate": 0.0005077190228200593,
      "loss": 2.4016,
      "step": 59111
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7147937417030334,
      "learning_rate": 0.0005077160713906903,
      "loss": 2.1947,
      "step": 59112
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7251065969467163,
      "learning_rate": 0.0005077131199227029,
      "loss": 2.1321,
      "step": 59113
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.8362284302711487,
      "learning_rate": 0.0005077101684160978,
      "loss": 2.2628,
      "step": 59114
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7036928534507751,
      "learning_rate": 0.0005077072168708754,
      "loss": 2.5119,
      "step": 59115
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7167004346847534,
      "learning_rate": 0.0005077042652870363,
      "loss": 2.1901,
      "step": 59116
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7272714972496033,
      "learning_rate": 0.000507701313664581,
      "loss": 2.2285,
      "step": 59117
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.6959860324859619,
      "learning_rate": 0.0005076983620035102,
      "loss": 2.301,
      "step": 59118
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7461203336715698,
      "learning_rate": 0.0005076954103038243,
      "loss": 2.3505,
      "step": 59119
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7533276677131653,
      "learning_rate": 0.0005076924585655239,
      "loss": 2.3599,
      "step": 59120
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7665638327598572,
      "learning_rate": 0.0005076895067886096,
      "loss": 2.2704,
      "step": 59121
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7120107412338257,
      "learning_rate": 0.0005076865549730818,
      "loss": 2.2657,
      "step": 59122
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7137147784233093,
      "learning_rate": 0.0005076836031189412,
      "loss": 2.351,
      "step": 59123
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.906244695186615,
      "learning_rate": 0.0005076806512261885,
      "loss": 2.4022,
      "step": 59124
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7530925273895264,
      "learning_rate": 0.0005076776992948238,
      "loss": 2.1844,
      "step": 59125
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7666881084442139,
      "learning_rate": 0.0005076747473248478,
      "loss": 2.4656,
      "step": 59126
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7722002267837524,
      "learning_rate": 0.0005076717953162614,
      "loss": 2.2535,
      "step": 59127
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7027596831321716,
      "learning_rate": 0.0005076688432690648,
      "loss": 2.2701,
      "step": 59128
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.8839195370674133,
      "learning_rate": 0.0005076658911832586,
      "loss": 2.1182,
      "step": 59129
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.6966499090194702,
      "learning_rate": 0.0005076629390588435,
      "loss": 2.1839,
      "step": 59130
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.6998372077941895,
      "learning_rate": 0.0005076599868958199,
      "loss": 2.4275,
      "step": 59131
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7196102142333984,
      "learning_rate": 0.0005076570346941883,
      "loss": 2.054,
      "step": 59132
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.738853931427002,
      "learning_rate": 0.0005076540824539493,
      "loss": 2.1899,
      "step": 59133
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.701330840587616,
      "learning_rate": 0.0005076511301751037,
      "loss": 2.2717,
      "step": 59134
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7147952914237976,
      "learning_rate": 0.0005076481778576517,
      "loss": 2.3096,
      "step": 59135
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.8146345019340515,
      "learning_rate": 0.000507645225501594,
      "loss": 2.238,
      "step": 59136
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7299972772598267,
      "learning_rate": 0.000507642273106931,
      "loss": 2.0304,
      "step": 59137
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.6556597352027893,
      "learning_rate": 0.0005076393206736637,
      "loss": 2.1537,
      "step": 59138
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7481740117073059,
      "learning_rate": 0.0005076363682017921,
      "loss": 2.2956,
      "step": 59139
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.768915057182312,
      "learning_rate": 0.000507633415691317,
      "loss": 2.1028,
      "step": 59140
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.763673722743988,
      "learning_rate": 0.0005076304631422389,
      "loss": 2.2012,
      "step": 59141
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7408887147903442,
      "learning_rate": 0.0005076275105545585,
      "loss": 2.2673,
      "step": 59142
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.8826411366462708,
      "learning_rate": 0.0005076245579282761,
      "loss": 2.2785,
      "step": 59143
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7563340663909912,
      "learning_rate": 0.0005076216052633924,
      "loss": 2.2684,
      "step": 59144
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7908234000205994,
      "learning_rate": 0.000507618652559908,
      "loss": 2.266,
      "step": 59145
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.8988197445869446,
      "learning_rate": 0.0005076156998178234,
      "loss": 2.0579,
      "step": 59146
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7365298867225647,
      "learning_rate": 0.0005076127470371389,
      "loss": 2.2819,
      "step": 59147
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7757478356361389,
      "learning_rate": 0.0005076097942178556,
      "loss": 2.1494,
      "step": 59148
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7416989803314209,
      "learning_rate": 0.0005076068413599735,
      "loss": 2.2671,
      "step": 59149
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.0319709777832031,
      "learning_rate": 0.0005076038884634933,
      "loss": 2.328,
      "step": 59150
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7682603597640991,
      "learning_rate": 0.0005076009355284158,
      "loss": 2.3244,
      "step": 59151
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.8882542848587036,
      "learning_rate": 0.0005075979825547413,
      "loss": 2.3848,
      "step": 59152
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.8276675343513489,
      "learning_rate": 0.0005075950295424705,
      "loss": 2.1163,
      "step": 59153
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7422166466712952,
      "learning_rate": 0.0005075920764916037,
      "loss": 2.0398,
      "step": 59154
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.6900002360343933,
      "learning_rate": 0.0005075891234021416,
      "loss": 2.2548,
      "step": 59155
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7489467859268188,
      "learning_rate": 0.0005075861702740848,
      "loss": 2.1618,
      "step": 59156
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.0041908025741577,
      "learning_rate": 0.000507583217107434,
      "loss": 2.0764,
      "step": 59157
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7550743222236633,
      "learning_rate": 0.0005075802639021895,
      "loss": 2.3821,
      "step": 59158
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7294694781303406,
      "learning_rate": 0.0005075773106583517,
      "loss": 2.3379,
      "step": 59159
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.8317638039588928,
      "learning_rate": 0.0005075743573759215,
      "loss": 2.236,
      "step": 59160
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.8883976340293884,
      "learning_rate": 0.0005075714040548993,
      "loss": 2.3316,
      "step": 59161
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7042052745819092,
      "learning_rate": 0.0005075684506952857,
      "loss": 2.0072,
      "step": 59162
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7226459383964539,
      "learning_rate": 0.0005075654972970812,
      "loss": 2.0742,
      "step": 59163
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.8474454283714294,
      "learning_rate": 0.0005075625438602864,
      "loss": 2.2545,
      "step": 59164
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7439887523651123,
      "learning_rate": 0.0005075595903849016,
      "loss": 2.2586,
      "step": 59165
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7177255749702454,
      "learning_rate": 0.0005075566368709277,
      "loss": 2.2445,
      "step": 59166
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7978270649909973,
      "learning_rate": 0.0005075536833183652,
      "loss": 2.4427,
      "step": 59167
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.8346077799797058,
      "learning_rate": 0.0005075507297272144,
      "loss": 2.3896,
      "step": 59168
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7794673442840576,
      "learning_rate": 0.0005075477760974761,
      "loss": 2.1302,
      "step": 59169
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7186768651008606,
      "learning_rate": 0.0005075448224291507,
      "loss": 2.0162,
      "step": 59170
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.9642221927642822,
      "learning_rate": 0.0005075418687222388,
      "loss": 2.2957,
      "step": 59171
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7367708683013916,
      "learning_rate": 0.0005075389149767409,
      "loss": 2.1893,
      "step": 59172
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7261995077133179,
      "learning_rate": 0.0005075359611926577,
      "loss": 1.8864,
      "step": 59173
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.8771736025810242,
      "learning_rate": 0.0005075330073699895,
      "loss": 2.0945,
      "step": 59174
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.8605373501777649,
      "learning_rate": 0.0005075300535087371,
      "loss": 2.1983,
      "step": 59175
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.6856203675270081,
      "learning_rate": 0.000507527099608901,
      "loss": 2.3847,
      "step": 59176
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.67294842004776,
      "learning_rate": 0.0005075241456704816,
      "loss": 2.1482,
      "step": 59177
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7454347610473633,
      "learning_rate": 0.0005075211916934795,
      "loss": 2.1101,
      "step": 59178
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.6637908816337585,
      "learning_rate": 0.0005075182376778955,
      "loss": 2.2052,
      "step": 59179
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7530834078788757,
      "learning_rate": 0.0005075152836237298,
      "loss": 2.372,
      "step": 59180
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.790057897567749,
      "learning_rate": 0.000507512329530983,
      "loss": 2.2671,
      "step": 59181
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7643399834632874,
      "learning_rate": 0.0005075093753996559,
      "loss": 2.2565,
      "step": 59182
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7243995070457458,
      "learning_rate": 0.0005075064212297489,
      "loss": 2.2686,
      "step": 59183
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.8396928310394287,
      "learning_rate": 0.0005075034670212623,
      "loss": 2.3294,
      "step": 59184
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.759566068649292,
      "learning_rate": 0.0005075005127741972,
      "loss": 2.3294,
      "step": 59185
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7471746802330017,
      "learning_rate": 0.0005074975584885537,
      "loss": 2.3748,
      "step": 59186
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7717498540878296,
      "learning_rate": 0.0005074946041643324,
      "loss": 2.4051,
      "step": 59187
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7765714526176453,
      "learning_rate": 0.0005074916498015341,
      "loss": 2.0533,
      "step": 59188
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.6642553210258484,
      "learning_rate": 0.0005074886954001591,
      "loss": 2.0297,
      "step": 59189
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.8472619652748108,
      "learning_rate": 0.000507485740960208,
      "loss": 2.3181,
      "step": 59190
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.6956397891044617,
      "learning_rate": 0.0005074827864816814,
      "loss": 2.2585,
      "step": 59191
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7380732297897339,
      "learning_rate": 0.0005074798319645799,
      "loss": 2.3623,
      "step": 59192
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7614744305610657,
      "learning_rate": 0.000507476877408904,
      "loss": 2.376,
      "step": 59193
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.6889785528182983,
      "learning_rate": 0.0005074739228146542,
      "loss": 2.3186,
      "step": 59194
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7553213238716125,
      "learning_rate": 0.0005074709681818309,
      "loss": 2.2991,
      "step": 59195
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.6720112562179565,
      "learning_rate": 0.000507468013510435,
      "loss": 2.1638,
      "step": 59196
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.6865705847740173,
      "learning_rate": 0.0005074650588004669,
      "loss": 2.2791,
      "step": 59197
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.9100114703178406,
      "learning_rate": 0.0005074621040519271,
      "loss": 2.2982,
      "step": 59198
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.8191500306129456,
      "learning_rate": 0.0005074591492648161,
      "loss": 2.2075,
      "step": 59199
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.8167081475257874,
      "learning_rate": 0.0005074561944391347,
      "loss": 2.3301,
      "step": 59200
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.8098505735397339,
      "learning_rate": 0.0005074532395748831,
      "loss": 2.0179,
      "step": 59201
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7439296245574951,
      "learning_rate": 0.0005074502846720622,
      "loss": 2.2752,
      "step": 59202
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7442936897277832,
      "learning_rate": 0.0005074473297306722,
      "loss": 2.0518,
      "step": 59203
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.2122260332107544,
      "learning_rate": 0.000507444374750714,
      "loss": 2.3284,
      "step": 59204
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7739018201828003,
      "learning_rate": 0.0005074414197321879,
      "loss": 2.4854,
      "step": 59205
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.8224854469299316,
      "learning_rate": 0.0005074384646750944,
      "loss": 2.3669,
      "step": 59206
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7178931832313538,
      "learning_rate": 0.0005074355095794343,
      "loss": 2.1642,
      "step": 59207
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7343538403511047,
      "learning_rate": 0.000507432554445208,
      "loss": 2.3477,
      "step": 59208
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.6715402603149414,
      "learning_rate": 0.0005074295992724161,
      "loss": 2.2176,
      "step": 59209
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.0053216218948364,
      "learning_rate": 0.0005074266440610591,
      "loss": 2.2698,
      "step": 59210
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.6667061448097229,
      "learning_rate": 0.0005074236888111375,
      "loss": 2.1682,
      "step": 59211
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.8115702867507935,
      "learning_rate": 0.0005074207335226521,
      "loss": 2.1879,
      "step": 59212
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.8113356232643127,
      "learning_rate": 0.0005074177781956031,
      "loss": 2.1692,
      "step": 59213
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7671934366226196,
      "learning_rate": 0.0005074148228299913,
      "loss": 2.1909,
      "step": 59214
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7584230303764343,
      "learning_rate": 0.0005074118674258172,
      "loss": 2.0927,
      "step": 59215
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7946557998657227,
      "learning_rate": 0.0005074089119830812,
      "loss": 2.013,
      "step": 59216
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7149002552032471,
      "learning_rate": 0.0005074059565017841,
      "loss": 2.1705,
      "step": 59217
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7263511419296265,
      "learning_rate": 0.0005074030009819261,
      "loss": 2.3286,
      "step": 59218
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7124128937721252,
      "learning_rate": 0.0005074000454235081,
      "loss": 2.2087,
      "step": 59219
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7185232043266296,
      "learning_rate": 0.0005073970898265306,
      "loss": 2.1,
      "step": 59220
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7488425374031067,
      "learning_rate": 0.000507394134190994,
      "loss": 2.0714,
      "step": 59221
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7631255388259888,
      "learning_rate": 0.000507391178516899,
      "loss": 2.1637,
      "step": 59222
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.0822516679763794,
      "learning_rate": 0.000507388222804246,
      "loss": 2.2675,
      "step": 59223
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7777271270751953,
      "learning_rate": 0.0005073852670530356,
      "loss": 2.0445,
      "step": 59224
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7343385219573975,
      "learning_rate": 0.0005073823112632683,
      "loss": 2.1897,
      "step": 59225
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.6915590167045593,
      "learning_rate": 0.0005073793554349448,
      "loss": 2.2462,
      "step": 59226
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.6719648241996765,
      "learning_rate": 0.0005073763995680655,
      "loss": 2.4897,
      "step": 59227
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7554236054420471,
      "learning_rate": 0.0005073734436626311,
      "loss": 2.3097,
      "step": 59228
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7713631987571716,
      "learning_rate": 0.000507370487718642,
      "loss": 2.3443,
      "step": 59229
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7020137906074524,
      "learning_rate": 0.0005073675317360988,
      "loss": 2.1569,
      "step": 59230
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.987626850605011,
      "learning_rate": 0.0005073645757150021,
      "loss": 2.4176,
      "step": 59231
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7404839396476746,
      "learning_rate": 0.0005073616196553524,
      "loss": 2.161,
      "step": 59232
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.8827366828918457,
      "learning_rate": 0.0005073586635571503,
      "loss": 2.2789,
      "step": 59233
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.8495067954063416,
      "learning_rate": 0.0005073557074203963,
      "loss": 2.0989,
      "step": 59234
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.8409038186073303,
      "learning_rate": 0.000507352751245091,
      "loss": 2.1543,
      "step": 59235
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7652245759963989,
      "learning_rate": 0.0005073497950312347,
      "loss": 2.1995,
      "step": 59236
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.9257112145423889,
      "learning_rate": 0.0005073468387788283,
      "loss": 2.2105,
      "step": 59237
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.8179872632026672,
      "learning_rate": 0.0005073438824878722,
      "loss": 2.0645,
      "step": 59238
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7112656831741333,
      "learning_rate": 0.000507340926158367,
      "loss": 2.3383,
      "step": 59239
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.8367331624031067,
      "learning_rate": 0.0005073379697903131,
      "loss": 2.392,
      "step": 59240
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.810580313205719,
      "learning_rate": 0.0005073350133837113,
      "loss": 2.3785,
      "step": 59241
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7509381175041199,
      "learning_rate": 0.0005073320569385619,
      "loss": 2.4483,
      "step": 59242
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.6476300954818726,
      "learning_rate": 0.0005073291004548656,
      "loss": 2.0521,
      "step": 59243
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7211341261863708,
      "learning_rate": 0.000507326143932623,
      "loss": 2.277,
      "step": 59244
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.8222432732582092,
      "learning_rate": 0.0005073231873718344,
      "loss": 2.2414,
      "step": 59245
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7229059338569641,
      "learning_rate": 0.0005073202307725005,
      "loss": 2.1069,
      "step": 59246
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7474443316459656,
      "learning_rate": 0.000507317274134622,
      "loss": 2.058,
      "step": 59247
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7460484504699707,
      "learning_rate": 0.0005073143174581992,
      "loss": 2.1225,
      "step": 59248
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.6899083852767944,
      "learning_rate": 0.0005073113607432328,
      "loss": 2.0482,
      "step": 59249
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7794091701507568,
      "learning_rate": 0.0005073084039897232,
      "loss": 2.1142,
      "step": 59250
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.8455892205238342,
      "learning_rate": 0.0005073054471976711,
      "loss": 2.2116,
      "step": 59251
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.6948536038398743,
      "learning_rate": 0.0005073024903670771,
      "loss": 2.2912,
      "step": 59252
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7515817880630493,
      "learning_rate": 0.0005072995334979416,
      "loss": 2.3759,
      "step": 59253
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7823745012283325,
      "learning_rate": 0.0005072965765902652,
      "loss": 2.2356,
      "step": 59254
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.776218593120575,
      "learning_rate": 0.0005072936196440484,
      "loss": 2.4722,
      "step": 59255
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7383750081062317,
      "learning_rate": 0.000507290662659292,
      "loss": 2.1815,
      "step": 59256
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7711272835731506,
      "learning_rate": 0.0005072877056359961,
      "loss": 2.1232,
      "step": 59257
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7086948156356812,
      "learning_rate": 0.0005072847485741616,
      "loss": 2.3674,
      "step": 59258
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.8919075727462769,
      "learning_rate": 0.0005072817914737891,
      "loss": 2.365,
      "step": 59259
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7083972096443176,
      "learning_rate": 0.0005072788343348789,
      "loss": 2.0932,
      "step": 59260
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.6697725057601929,
      "learning_rate": 0.0005072758771574315,
      "loss": 2.2777,
      "step": 59261
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.779373824596405,
      "learning_rate": 0.0005072729199414477,
      "loss": 2.1483,
      "step": 59262
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7563121318817139,
      "learning_rate": 0.000507269962686928,
      "loss": 2.2089,
      "step": 59263
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7204303741455078,
      "learning_rate": 0.0005072670053938729,
      "loss": 2.116,
      "step": 59264
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7281767725944519,
      "learning_rate": 0.0005072640480622829,
      "loss": 2.2755,
      "step": 59265
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7620317935943604,
      "learning_rate": 0.0005072610906921586,
      "loss": 2.1027,
      "step": 59266
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.753339946269989,
      "learning_rate": 0.0005072581332835007,
      "loss": 2.2182,
      "step": 59267
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7650668621063232,
      "learning_rate": 0.0005072551758363093,
      "loss": 2.2891,
      "step": 59268
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.8483889102935791,
      "learning_rate": 0.0005072522183505856,
      "loss": 2.2986,
      "step": 59269
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.6887796521186829,
      "learning_rate": 0.0005072492608263296,
      "loss": 2.3757,
      "step": 59270
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.6880514025688171,
      "learning_rate": 0.0005072463032635421,
      "loss": 2.1315,
      "step": 59271
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.9886004328727722,
      "learning_rate": 0.0005072433456622236,
      "loss": 2.1482,
      "step": 59272
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7396072745323181,
      "learning_rate": 0.0005072403880223746,
      "loss": 2.0702,
      "step": 59273
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7486482858657837,
      "learning_rate": 0.0005072374303439957,
      "loss": 2.3407,
      "step": 59274
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.8064783215522766,
      "learning_rate": 0.0005072344726270875,
      "loss": 2.2183,
      "step": 59275
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.8428583741188049,
      "learning_rate": 0.0005072315148716504,
      "loss": 2.1279,
      "step": 59276
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7545809745788574,
      "learning_rate": 0.0005072285570776852,
      "loss": 2.157,
      "step": 59277
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7204498052597046,
      "learning_rate": 0.0005072255992451922,
      "loss": 2.2913,
      "step": 59278
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7513315081596375,
      "learning_rate": 0.0005072226413741721,
      "loss": 2.1607,
      "step": 59279
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7561950087547302,
      "learning_rate": 0.0005072196834646253,
      "loss": 2.2613,
      "step": 59280
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7474087476730347,
      "learning_rate": 0.0005072167255165525,
      "loss": 2.1282,
      "step": 59281
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.8766868114471436,
      "learning_rate": 0.0005072137675299544,
      "loss": 1.9893,
      "step": 59282
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7898634672164917,
      "learning_rate": 0.0005072108095048311,
      "loss": 2.175,
      "step": 59283
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7647507786750793,
      "learning_rate": 0.0005072078514411835,
      "loss": 2.2916,
      "step": 59284
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.8371597528457642,
      "learning_rate": 0.000507204893339012,
      "loss": 2.4501,
      "step": 59285
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7341600656509399,
      "learning_rate": 0.0005072019351983172,
      "loss": 2.3025,
      "step": 59286
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7781243324279785,
      "learning_rate": 0.0005071989770190997,
      "loss": 2.2116,
      "step": 59287
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.777154266834259,
      "learning_rate": 0.00050719601880136,
      "loss": 2.191,
      "step": 59288
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.9133731126785278,
      "learning_rate": 0.0005071930605450985,
      "loss": 2.1381,
      "step": 59289
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7494586706161499,
      "learning_rate": 0.0005071901022503161,
      "loss": 2.0452,
      "step": 59290
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.772777259349823,
      "learning_rate": 0.000507187143917013,
      "loss": 2.3884,
      "step": 59291
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.8501629829406738,
      "learning_rate": 0.00050718418554519,
      "loss": 2.1855,
      "step": 59292
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7246792316436768,
      "learning_rate": 0.0005071812271348475,
      "loss": 2.0862,
      "step": 59293
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7437120676040649,
      "learning_rate": 0.0005071782686859861,
      "loss": 2.1251,
      "step": 59294
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7819870114326477,
      "learning_rate": 0.0005071753101986064,
      "loss": 2.444,
      "step": 59295
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7291032671928406,
      "learning_rate": 0.0005071723516727088,
      "loss": 2.2113,
      "step": 59296
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7060962319374084,
      "learning_rate": 0.000507169393108294,
      "loss": 2.2876,
      "step": 59297
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7914505004882812,
      "learning_rate": 0.0005071664345053625,
      "loss": 2.1994,
      "step": 59298
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7918877005577087,
      "learning_rate": 0.0005071634758639149,
      "loss": 2.0366,
      "step": 59299
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.8320480585098267,
      "learning_rate": 0.0005071605171839517,
      "loss": 2.1092,
      "step": 59300
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.8092661499977112,
      "learning_rate": 0.0005071575584654735,
      "loss": 2.3724,
      "step": 59301
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.205008625984192,
      "learning_rate": 0.0005071545997084806,
      "loss": 2.1209,
      "step": 59302
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7089862823486328,
      "learning_rate": 0.0005071516409129739,
      "loss": 2.2856,
      "step": 59303
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7620477080345154,
      "learning_rate": 0.0005071486820789538,
      "loss": 2.0299,
      "step": 59304
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.8208346962928772,
      "learning_rate": 0.0005071457232064207,
      "loss": 2.3441,
      "step": 59305
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.6989115476608276,
      "learning_rate": 0.0005071427642953755,
      "loss": 2.2969,
      "step": 59306
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7241862416267395,
      "learning_rate": 0.0005071398053458184,
      "loss": 2.1096,
      "step": 59307
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.709075391292572,
      "learning_rate": 0.0005071368463577502,
      "loss": 1.9573,
      "step": 59308
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.9970172047615051,
      "learning_rate": 0.0005071338873311712,
      "loss": 2.1639,
      "step": 59309
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.1309534311294556,
      "learning_rate": 0.0005071309282660822,
      "loss": 2.4588,
      "step": 59310
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.8971500992774963,
      "learning_rate": 0.0005071279691624837,
      "loss": 2.3009,
      "step": 59311
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.059588074684143,
      "learning_rate": 0.0005071250100203762,
      "loss": 2.2753,
      "step": 59312
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7937030792236328,
      "learning_rate": 0.0005071220508397602,
      "loss": 2.215,
      "step": 59313
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.8463091850280762,
      "learning_rate": 0.0005071190916206363,
      "loss": 2.2628,
      "step": 59314
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.8269762396812439,
      "learning_rate": 0.000507116132363005,
      "loss": 2.3627,
      "step": 59315
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7487319111824036,
      "learning_rate": 0.000507113173066867,
      "loss": 2.131,
      "step": 59316
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.6919419765472412,
      "learning_rate": 0.0005071102137322227,
      "loss": 2.2117,
      "step": 59317
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7477757334709167,
      "learning_rate": 0.0005071072543590727,
      "loss": 2.3662,
      "step": 59318
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.6633363962173462,
      "learning_rate": 0.0005071042949474176,
      "loss": 2.4196,
      "step": 59319
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.756733775138855,
      "learning_rate": 0.0005071013354972579,
      "loss": 2.1916,
      "step": 59320
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7269042730331421,
      "learning_rate": 0.0005070983760085941,
      "loss": 2.147,
      "step": 59321
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.777262270450592,
      "learning_rate": 0.0005070954164814269,
      "loss": 2.278,
      "step": 59322
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7548041939735413,
      "learning_rate": 0.0005070924569157566,
      "loss": 2.3564,
      "step": 59323
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.9621169567108154,
      "learning_rate": 0.000507089497311584,
      "loss": 2.3896,
      "step": 59324
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.760940670967102,
      "learning_rate": 0.0005070865376689095,
      "loss": 2.1354,
      "step": 59325
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.795346736907959,
      "learning_rate": 0.0005070835779877337,
      "loss": 2.4635,
      "step": 59326
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7950569987297058,
      "learning_rate": 0.0005070806182680573,
      "loss": 2.252,
      "step": 59327
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7875692844390869,
      "learning_rate": 0.0005070776585098806,
      "loss": 2.1525,
      "step": 59328
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7511917948722839,
      "learning_rate": 0.0005070746987132042,
      "loss": 2.5234,
      "step": 59329
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.8980684280395508,
      "learning_rate": 0.0005070717388780287,
      "loss": 2.2387,
      "step": 59330
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.8912327885627747,
      "learning_rate": 0.0005070687790043548,
      "loss": 2.2547,
      "step": 59331
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.9226536750793457,
      "learning_rate": 0.0005070658190921829,
      "loss": 2.1685,
      "step": 59332
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.8058007955551147,
      "learning_rate": 0.0005070628591415133,
      "loss": 2.4677,
      "step": 59333
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.0746721029281616,
      "learning_rate": 0.000507059899152347,
      "loss": 2.3954,
      "step": 59334
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.8662227988243103,
      "learning_rate": 0.0005070569391246844,
      "loss": 2.4699,
      "step": 59335
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.8271474242210388,
      "learning_rate": 0.0005070539790585259,
      "loss": 2.072,
      "step": 59336
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7023751139640808,
      "learning_rate": 0.0005070510189538722,
      "loss": 2.2466,
      "step": 59337
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.9333142638206482,
      "learning_rate": 0.0005070480588107238,
      "loss": 2.3385,
      "step": 59338
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.8848084807395935,
      "learning_rate": 0.0005070450986290813,
      "loss": 2.1046,
      "step": 59339
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7796899676322937,
      "learning_rate": 0.0005070421384089452,
      "loss": 2.4117,
      "step": 59340
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.716568648815155,
      "learning_rate": 0.000507039178150316,
      "loss": 2.2414,
      "step": 59341
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7948315739631653,
      "learning_rate": 0.0005070362178531943,
      "loss": 2.3268,
      "step": 59342
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7982944250106812,
      "learning_rate": 0.0005070332575175808,
      "loss": 2.3132,
      "step": 59343
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7788143157958984,
      "learning_rate": 0.0005070302971434757,
      "loss": 2.3857,
      "step": 59344
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7253732085227966,
      "learning_rate": 0.0005070273367308799,
      "loss": 2.4983,
      "step": 59345
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7822176814079285,
      "learning_rate": 0.0005070243762797938,
      "loss": 2.2622,
      "step": 59346
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7182753086090088,
      "learning_rate": 0.000507021415790218,
      "loss": 2.2638,
      "step": 59347
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7547135353088379,
      "learning_rate": 0.0005070184552621529,
      "loss": 2.1478,
      "step": 59348
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7137260437011719,
      "learning_rate": 0.0005070154946955991,
      "loss": 2.0082,
      "step": 59349
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7373384237289429,
      "learning_rate": 0.0005070125340905574,
      "loss": 2.3685,
      "step": 59350
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.771791934967041,
      "learning_rate": 0.000507009573447028,
      "loss": 2.0867,
      "step": 59351
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7458954453468323,
      "learning_rate": 0.0005070066127650117,
      "loss": 2.4141,
      "step": 59352
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.6830669641494751,
      "learning_rate": 0.000507003652044509,
      "loss": 2.0582,
      "step": 59353
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7868877649307251,
      "learning_rate": 0.0005070006912855203,
      "loss": 2.0962,
      "step": 59354
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7722805738449097,
      "learning_rate": 0.0005069977304880464,
      "loss": 2.4001,
      "step": 59355
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7669320106506348,
      "learning_rate": 0.0005069947696520877,
      "loss": 2.3424,
      "step": 59356
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.6872856020927429,
      "learning_rate": 0.0005069918087776447,
      "loss": 2.5362,
      "step": 59357
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.729253888130188,
      "learning_rate": 0.000506988847864718,
      "loss": 2.1195,
      "step": 59358
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7371688485145569,
      "learning_rate": 0.0005069858869133083,
      "loss": 1.9989,
      "step": 59359
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.752362847328186,
      "learning_rate": 0.0005069829259234159,
      "loss": 2.3043,
      "step": 59360
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7629756331443787,
      "learning_rate": 0.0005069799648950414,
      "loss": 2.3795,
      "step": 59361
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7503603100776672,
      "learning_rate": 0.0005069770038281855,
      "loss": 2.2685,
      "step": 59362
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7687634229660034,
      "learning_rate": 0.0005069740427228486,
      "loss": 1.921,
      "step": 59363
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.704846978187561,
      "learning_rate": 0.0005069710815790314,
      "loss": 2.142,
      "step": 59364
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7979058027267456,
      "learning_rate": 0.0005069681203967345,
      "loss": 2.1533,
      "step": 59365
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.743749737739563,
      "learning_rate": 0.0005069651591759581,
      "loss": 2.2279,
      "step": 59366
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.6907907128334045,
      "learning_rate": 0.000506962197916703,
      "loss": 2.008,
      "step": 59367
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.6926491260528564,
      "learning_rate": 0.0005069592366189697,
      "loss": 2.2969,
      "step": 59368
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7044446468353271,
      "learning_rate": 0.0005069562752827589,
      "loss": 2.2749,
      "step": 59369
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.76746666431427,
      "learning_rate": 0.0005069533139080709,
      "loss": 2.1894,
      "step": 59370
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7553350925445557,
      "learning_rate": 0.0005069503524949065,
      "loss": 2.3181,
      "step": 59371
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7199181318283081,
      "learning_rate": 0.000506947391043266,
      "loss": 2.06,
      "step": 59372
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7472323179244995,
      "learning_rate": 0.0005069444295531502,
      "loss": 2.3142,
      "step": 59373
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.9162932634353638,
      "learning_rate": 0.0005069414680245594,
      "loss": 1.9951,
      "step": 59374
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7564607262611389,
      "learning_rate": 0.0005069385064574943,
      "loss": 2.3076,
      "step": 59375
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.678276002407074,
      "learning_rate": 0.0005069355448519555,
      "loss": 2.2537,
      "step": 59376
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.8889809846878052,
      "learning_rate": 0.0005069325832079434,
      "loss": 2.3653,
      "step": 59377
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7509058117866516,
      "learning_rate": 0.0005069296215254586,
      "loss": 2.3978,
      "step": 59378
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.682302713394165,
      "learning_rate": 0.0005069266598045018,
      "loss": 2.4497,
      "step": 59379
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.717792272567749,
      "learning_rate": 0.0005069236980450733,
      "loss": 2.3013,
      "step": 59380
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7282955050468445,
      "learning_rate": 0.0005069207362471738,
      "loss": 2.3148,
      "step": 59381
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7113979458808899,
      "learning_rate": 0.000506917774410804,
      "loss": 2.1734,
      "step": 59382
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.8209419250488281,
      "learning_rate": 0.0005069148125359641,
      "loss": 2.139,
      "step": 59383
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7205862402915955,
      "learning_rate": 0.0005069118506226549,
      "loss": 2.3173,
      "step": 59384
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7314899563789368,
      "learning_rate": 0.0005069088886708767,
      "loss": 2.3853,
      "step": 59385
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7797661423683167,
      "learning_rate": 0.0005069059266806304,
      "loss": 2.176,
      "step": 59386
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7505258321762085,
      "learning_rate": 0.0005069029646519164,
      "loss": 1.8252,
      "step": 59387
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7512519359588623,
      "learning_rate": 0.0005069000025847353,
      "loss": 2.2739,
      "step": 59388
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7417458891868591,
      "learning_rate": 0.0005068970404790875,
      "loss": 2.1896,
      "step": 59389
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7092138528823853,
      "learning_rate": 0.0005068940783349736,
      "loss": 2.2751,
      "step": 59390
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7411709427833557,
      "learning_rate": 0.0005068911161523942,
      "loss": 2.351,
      "step": 59391
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7734699249267578,
      "learning_rate": 0.0005068881539313499,
      "loss": 2.2173,
      "step": 59392
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7916930317878723,
      "learning_rate": 0.000506885191671841,
      "loss": 2.2426,
      "step": 59393
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.740334153175354,
      "learning_rate": 0.0005068822293738684,
      "loss": 2.2422,
      "step": 59394
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.8025649189949036,
      "learning_rate": 0.0005068792670374325,
      "loss": 2.2551,
      "step": 59395
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7637199759483337,
      "learning_rate": 0.0005068763046625338,
      "loss": 2.3063,
      "step": 59396
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7373513579368591,
      "learning_rate": 0.0005068733422491729,
      "loss": 1.984,
      "step": 59397
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7409480214118958,
      "learning_rate": 0.0005068703797973503,
      "loss": 2.3323,
      "step": 59398
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7585614919662476,
      "learning_rate": 0.0005068674173070667,
      "loss": 2.3084,
      "step": 59399
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.743321418762207,
      "learning_rate": 0.0005068644547783224,
      "loss": 2.1312,
      "step": 59400
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7192073464393616,
      "learning_rate": 0.0005068614922111182,
      "loss": 1.9748,
      "step": 59401
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7383701801300049,
      "learning_rate": 0.0005068585296054545,
      "loss": 2.4882,
      "step": 59402
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7422230839729309,
      "learning_rate": 0.0005068555669613318,
      "loss": 2.3498,
      "step": 59403
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.8228800892829895,
      "learning_rate": 0.000506852604278751,
      "loss": 2.0662,
      "step": 59404
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7885796427726746,
      "learning_rate": 0.0005068496415577121,
      "loss": 2.2019,
      "step": 59405
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7707921266555786,
      "learning_rate": 0.000506846678798216,
      "loss": 2.1847,
      "step": 59406
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7557206749916077,
      "learning_rate": 0.0005068437160002634,
      "loss": 2.0273,
      "step": 59407
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7000008821487427,
      "learning_rate": 0.0005068407531638545,
      "loss": 2.3841,
      "step": 59408
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.8851820826530457,
      "learning_rate": 0.0005068377902889901,
      "loss": 2.2678,
      "step": 59409
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7338812947273254,
      "learning_rate": 0.0005068348273756705,
      "loss": 2.0713,
      "step": 59410
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.883722722530365,
      "learning_rate": 0.0005068318644238965,
      "loss": 2.3645,
      "step": 59411
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.8455027341842651,
      "learning_rate": 0.0005068289014336686,
      "loss": 2.4758,
      "step": 59412
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.8846647143363953,
      "learning_rate": 0.0005068259384049872,
      "loss": 2.3438,
      "step": 59413
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.6813270449638367,
      "learning_rate": 0.0005068229753378529,
      "loss": 2.3125,
      "step": 59414
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7043503522872925,
      "learning_rate": 0.0005068200122322665,
      "loss": 2.3075,
      "step": 59415
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.0505809783935547,
      "learning_rate": 0.0005068170490882282,
      "loss": 2.1026,
      "step": 59416
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7104737758636475,
      "learning_rate": 0.0005068140859057388,
      "loss": 2.2421,
      "step": 59417
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7543931007385254,
      "learning_rate": 0.0005068111226847987,
      "loss": 2.2808,
      "step": 59418
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7802250385284424,
      "learning_rate": 0.0005068081594254085,
      "loss": 2.0933,
      "step": 59419
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.9350693821907043,
      "learning_rate": 0.0005068051961275688,
      "loss": 2.2039,
      "step": 59420
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7644781470298767,
      "learning_rate": 0.00050680223279128,
      "loss": 2.0642,
      "step": 59421
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.801594614982605,
      "learning_rate": 0.000506799269416543,
      "loss": 2.3236,
      "step": 59422
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.9453772902488708,
      "learning_rate": 0.0005067963060033577,
      "loss": 2.2282,
      "step": 59423
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.8197276592254639,
      "learning_rate": 0.0005067933425517254,
      "loss": 2.2689,
      "step": 59424
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.796760618686676,
      "learning_rate": 0.0005067903790616463,
      "loss": 2.1937,
      "step": 59425
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.9777621626853943,
      "learning_rate": 0.0005067874155331209,
      "loss": 2.1667,
      "step": 59426
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.8077026009559631,
      "learning_rate": 0.0005067844519661499,
      "loss": 2.3996,
      "step": 59427
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7288865447044373,
      "learning_rate": 0.0005067814883607336,
      "loss": 2.0653,
      "step": 59428
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.8652193546295166,
      "learning_rate": 0.0005067785247168729,
      "loss": 2.516,
      "step": 59429
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.9304341077804565,
      "learning_rate": 0.000506775561034568,
      "loss": 2.1213,
      "step": 59430
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.836384117603302,
      "learning_rate": 0.0005067725973138197,
      "loss": 2.1469,
      "step": 59431
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.9590659141540527,
      "learning_rate": 0.0005067696335546285,
      "loss": 2.0474,
      "step": 59432
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7700666785240173,
      "learning_rate": 0.0005067666697569948,
      "loss": 2.4006,
      "step": 59433
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7149546146392822,
      "learning_rate": 0.0005067637059209194,
      "loss": 2.1788,
      "step": 59434
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7555532455444336,
      "learning_rate": 0.0005067607420464027,
      "loss": 2.1661,
      "step": 59435
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.8521906137466431,
      "learning_rate": 0.0005067577781334452,
      "loss": 2.3004,
      "step": 59436
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.9206696152687073,
      "learning_rate": 0.0005067548141820476,
      "loss": 2.2985,
      "step": 59437
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7152122855186462,
      "learning_rate": 0.0005067518501922104,
      "loss": 2.2448,
      "step": 59438
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7461928725242615,
      "learning_rate": 0.0005067488861639341,
      "loss": 2.3246,
      "step": 59439
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7953658103942871,
      "learning_rate": 0.0005067459220972193,
      "loss": 2.2081,
      "step": 59440
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.0731303691864014,
      "learning_rate": 0.0005067429579920664,
      "loss": 2.2487,
      "step": 59441
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7509758472442627,
      "learning_rate": 0.0005067399938484763,
      "loss": 2.3367,
      "step": 59442
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7623202204704285,
      "learning_rate": 0.0005067370296664491,
      "loss": 2.1952,
      "step": 59443
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7921054363250732,
      "learning_rate": 0.0005067340654459858,
      "loss": 2.1154,
      "step": 59444
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.6847208142280579,
      "learning_rate": 0.0005067311011870867,
      "loss": 2.1856,
      "step": 59445
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7396894693374634,
      "learning_rate": 0.0005067281368897523,
      "loss": 2.2077,
      "step": 59446
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.8964649438858032,
      "learning_rate": 0.0005067251725539832,
      "loss": 2.2678,
      "step": 59447
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7848243117332458,
      "learning_rate": 0.0005067222081797801,
      "loss": 2.1139,
      "step": 59448
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7616859674453735,
      "learning_rate": 0.0005067192437671433,
      "loss": 2.1169,
      "step": 59449
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.1259169578552246,
      "learning_rate": 0.0005067162793160736,
      "loss": 2.1334,
      "step": 59450
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.6877639889717102,
      "learning_rate": 0.0005067133148265714,
      "loss": 2.0953,
      "step": 59451
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.8710583448410034,
      "learning_rate": 0.0005067103502986374,
      "loss": 2.2819,
      "step": 59452
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.8028798699378967,
      "learning_rate": 0.0005067073857322719,
      "loss": 2.136,
      "step": 59453
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7839353084564209,
      "learning_rate": 0.0005067044211274757,
      "loss": 2.368,
      "step": 59454
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7929736971855164,
      "learning_rate": 0.0005067014564842491,
      "loss": 2.1638,
      "step": 59455
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.0087389945983887,
      "learning_rate": 0.000506698491802593,
      "loss": 2.2079,
      "step": 59456
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.824385404586792,
      "learning_rate": 0.0005066955270825076,
      "loss": 2.2324,
      "step": 59457
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.6999306082725525,
      "learning_rate": 0.0005066925623239936,
      "loss": 2.1706,
      "step": 59458
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.9103536605834961,
      "learning_rate": 0.0005066895975270517,
      "loss": 2.2181,
      "step": 59459
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.978425145149231,
      "learning_rate": 0.0005066866326916821,
      "loss": 2.3873,
      "step": 59460
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7255504727363586,
      "learning_rate": 0.0005066836678178857,
      "loss": 2.0536,
      "step": 59461
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.993696928024292,
      "learning_rate": 0.0005066807029056628,
      "loss": 2.2274,
      "step": 59462
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.9983060359954834,
      "learning_rate": 0.0005066777379550141,
      "loss": 2.2342,
      "step": 59463
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.8212227821350098,
      "learning_rate": 0.0005066747729659401,
      "loss": 2.3242,
      "step": 59464
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.9452906250953674,
      "learning_rate": 0.0005066718079384414,
      "loss": 2.151,
      "step": 59465
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7934491038322449,
      "learning_rate": 0.0005066688428725185,
      "loss": 2.2955,
      "step": 59466
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.8936280012130737,
      "learning_rate": 0.0005066658777681718,
      "loss": 2.0867,
      "step": 59467
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7688677310943604,
      "learning_rate": 0.0005066629126254023,
      "loss": 2.272,
      "step": 59468
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.8096693754196167,
      "learning_rate": 0.0005066599474442101,
      "loss": 2.2612,
      "step": 59469
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.9813547730445862,
      "learning_rate": 0.0005066569822245959,
      "loss": 2.3787,
      "step": 59470
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7866988778114319,
      "learning_rate": 0.0005066540169665604,
      "loss": 2.0086,
      "step": 59471
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7111550569534302,
      "learning_rate": 0.0005066510516701038,
      "loss": 2.2325,
      "step": 59472
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.8093625903129578,
      "learning_rate": 0.000506648086335227,
      "loss": 2.1243,
      "step": 59473
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.6855602860450745,
      "learning_rate": 0.0005066451209619305,
      "loss": 2.1997,
      "step": 59474
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.8446696400642395,
      "learning_rate": 0.0005066421555502146,
      "loss": 2.2763,
      "step": 59475
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.6744139194488525,
      "learning_rate": 0.0005066391901000801,
      "loss": 2.0876,
      "step": 59476
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7955180406570435,
      "learning_rate": 0.0005066362246115275,
      "loss": 2.3213,
      "step": 59477
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7511008381843567,
      "learning_rate": 0.0005066332590845573,
      "loss": 2.3785,
      "step": 59478
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.0071827173233032,
      "learning_rate": 0.0005066302935191701,
      "loss": 2.1207,
      "step": 59479
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.6814709901809692,
      "learning_rate": 0.0005066273279153665,
      "loss": 2.1589,
      "step": 59480
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7188758850097656,
      "learning_rate": 0.0005066243622731468,
      "loss": 2.2876,
      "step": 59481
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.9022852182388306,
      "learning_rate": 0.0005066213965925117,
      "loss": 2.1368,
      "step": 59482
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.6625223755836487,
      "learning_rate": 0.000506618430873462,
      "loss": 1.9833,
      "step": 59483
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7798316478729248,
      "learning_rate": 0.0005066154651159979,
      "loss": 2.1991,
      "step": 59484
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7661476731300354,
      "learning_rate": 0.0005066124993201201,
      "loss": 2.3182,
      "step": 59485
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7750895619392395,
      "learning_rate": 0.0005066095334858292,
      "loss": 2.3096,
      "step": 59486
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7132236957550049,
      "learning_rate": 0.0005066065676131256,
      "loss": 2.1904,
      "step": 59487
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.683864176273346,
      "learning_rate": 0.00050660360170201,
      "loss": 2.3696,
      "step": 59488
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.8831775784492493,
      "learning_rate": 0.000506600635752483,
      "loss": 2.137,
      "step": 59489
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7478861212730408,
      "learning_rate": 0.0005065976697645448,
      "loss": 2.298,
      "step": 59490
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.8232306241989136,
      "learning_rate": 0.0005065947037381964,
      "loss": 2.2555,
      "step": 59491
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.678176760673523,
      "learning_rate": 0.0005065917376734379,
      "loss": 2.118,
      "step": 59492
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.8073033690452576,
      "learning_rate": 0.0005065887715702703,
      "loss": 2.2296,
      "step": 59493
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.6741050481796265,
      "learning_rate": 0.0005065858054286939,
      "loss": 2.0272,
      "step": 59494
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7656802535057068,
      "learning_rate": 0.0005065828392487093,
      "loss": 2.2836,
      "step": 59495
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.8675370216369629,
      "learning_rate": 0.000506579873030317,
      "loss": 2.1621,
      "step": 59496
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7444648146629333,
      "learning_rate": 0.0005065769067735177,
      "loss": 2.1873,
      "step": 59497
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7313437461853027,
      "learning_rate": 0.0005065739404783117,
      "loss": 2.3647,
      "step": 59498
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7348647117614746,
      "learning_rate": 0.0005065709741446999,
      "loss": 2.0566,
      "step": 59499
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.728288471698761,
      "learning_rate": 0.0005065680077726826,
      "loss": 2.0319,
      "step": 59500
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7088640928268433,
      "learning_rate": 0.0005065650413622603,
      "loss": 2.1997,
      "step": 59501
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7829229235649109,
      "learning_rate": 0.0005065620749134338,
      "loss": 2.1915,
      "step": 59502
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7333372831344604,
      "learning_rate": 0.0005065591084262034,
      "loss": 2.0655,
      "step": 59503
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7622244954109192,
      "learning_rate": 0.0005065561419005698,
      "loss": 2.3002,
      "step": 59504
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7460996508598328,
      "learning_rate": 0.0005065531753365335,
      "loss": 2.3796,
      "step": 59505
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.772497296333313,
      "learning_rate": 0.0005065502087340951,
      "loss": 2.3547,
      "step": 59506
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7253440618515015,
      "learning_rate": 0.000506547242093255,
      "loss": 2.2788,
      "step": 59507
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7544572353363037,
      "learning_rate": 0.000506544275414014,
      "loss": 2.2328,
      "step": 59508
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7442982196807861,
      "learning_rate": 0.0005065413086963725,
      "loss": 2.1256,
      "step": 59509
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7357577681541443,
      "learning_rate": 0.0005065383419403311,
      "loss": 2.1063,
      "step": 59510
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7617117762565613,
      "learning_rate": 0.0005065353751458903,
      "loss": 2.3565,
      "step": 59511
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7559747099876404,
      "learning_rate": 0.0005065324083130507,
      "loss": 2.3004,
      "step": 59512
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.0024335384368896,
      "learning_rate": 0.0005065294414418127,
      "loss": 2.2018,
      "step": 59513
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7934392094612122,
      "learning_rate": 0.0005065264745321771,
      "loss": 2.4368,
      "step": 59514
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7545801997184753,
      "learning_rate": 0.0005065235075841442,
      "loss": 2.1432,
      "step": 59515
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7653486132621765,
      "learning_rate": 0.0005065205405977148,
      "loss": 2.1706,
      "step": 59516
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7607478499412537,
      "learning_rate": 0.0005065175735728894,
      "loss": 2.1843,
      "step": 59517
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7830225229263306,
      "learning_rate": 0.0005065146065096683,
      "loss": 2.4725,
      "step": 59518
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.8389259576797485,
      "learning_rate": 0.0005065116394080524,
      "loss": 2.2929,
      "step": 59519
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7003015875816345,
      "learning_rate": 0.000506508672268042,
      "loss": 2.2485,
      "step": 59520
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7445274591445923,
      "learning_rate": 0.0005065057050896378,
      "loss": 2.0893,
      "step": 59521
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.6859397888183594,
      "learning_rate": 0.0005065027378728401,
      "loss": 2.2762,
      "step": 59522
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.8630304336547852,
      "learning_rate": 0.00050649977061765,
      "loss": 2.3479,
      "step": 59523
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7160609364509583,
      "learning_rate": 0.0005064968033240674,
      "loss": 2.3233,
      "step": 59524
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7430220246315002,
      "learning_rate": 0.0005064938359920933,
      "loss": 2.1492,
      "step": 59525
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.750738799571991,
      "learning_rate": 0.0005064908686217279,
      "loss": 2.27,
      "step": 59526
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.6785684823989868,
      "learning_rate": 0.0005064879012129721,
      "loss": 2.2542,
      "step": 59527
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7042213678359985,
      "learning_rate": 0.0005064849337658262,
      "loss": 2.2735,
      "step": 59528
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7376145720481873,
      "learning_rate": 0.0005064819662802912,
      "loss": 2.211,
      "step": 59529
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7396616339683533,
      "learning_rate": 0.0005064789987563669,
      "loss": 2.2686,
      "step": 59530
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7698713541030884,
      "learning_rate": 0.0005064760311940543,
      "loss": 2.2532,
      "step": 59531
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7299940586090088,
      "learning_rate": 0.0005064730635933541,
      "loss": 2.1835,
      "step": 59532
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7451047301292419,
      "learning_rate": 0.0005064700959542666,
      "loss": 2.3585,
      "step": 59533
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7482616901397705,
      "learning_rate": 0.0005064671282767923,
      "loss": 2.2746,
      "step": 59534
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.8129664659500122,
      "learning_rate": 0.000506464160560932,
      "loss": 2.1292,
      "step": 59535
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.6684423685073853,
      "learning_rate": 0.0005064611928066861,
      "loss": 2.1616,
      "step": 59536
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7374320030212402,
      "learning_rate": 0.000506458225014055,
      "loss": 2.0506,
      "step": 59537
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.782063364982605,
      "learning_rate": 0.0005064552571830397,
      "loss": 2.2232,
      "step": 59538
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.976306676864624,
      "learning_rate": 0.0005064522893136402,
      "loss": 2.5163,
      "step": 59539
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7806882262229919,
      "learning_rate": 0.0005064493214058574,
      "loss": 2.338,
      "step": 59540
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7418171167373657,
      "learning_rate": 0.0005064463534596919,
      "loss": 2.4827,
      "step": 59541
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7517709732055664,
      "learning_rate": 0.000506443385475144,
      "loss": 1.8931,
      "step": 59542
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7796469330787659,
      "learning_rate": 0.0005064404174522144,
      "loss": 2.1171,
      "step": 59543
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7033025622367859,
      "learning_rate": 0.0005064374493909037,
      "loss": 2.2985,
      "step": 59544
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7705659866333008,
      "learning_rate": 0.0005064344812912123,
      "loss": 2.1021,
      "step": 59545
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.8031863570213318,
      "learning_rate": 0.0005064315131531409,
      "loss": 2.2136,
      "step": 59546
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.9528904557228088,
      "learning_rate": 0.0005064285449766898,
      "loss": 2.4099,
      "step": 59547
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.6927776336669922,
      "learning_rate": 0.0005064255767618599,
      "loss": 2.2729,
      "step": 59548
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7443686127662659,
      "learning_rate": 0.0005064226085086516,
      "loss": 2.3944,
      "step": 59549
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.8934754133224487,
      "learning_rate": 0.0005064196402170654,
      "loss": 2.252,
      "step": 59550
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7787064909934998,
      "learning_rate": 0.0005064166718871019,
      "loss": 2.1329,
      "step": 59551
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.6994777917861938,
      "learning_rate": 0.0005064137035187616,
      "loss": 2.4218,
      "step": 59552
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.8085877299308777,
      "learning_rate": 0.0005064107351120451,
      "loss": 2.0828,
      "step": 59553
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.9311633706092834,
      "learning_rate": 0.0005064077666669531,
      "loss": 2.2379,
      "step": 59554
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.711969256401062,
      "learning_rate": 0.0005064047981834859,
      "loss": 2.5376,
      "step": 59555
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7124645709991455,
      "learning_rate": 0.000506401829661644,
      "loss": 1.9951,
      "step": 59556
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.8480883836746216,
      "learning_rate": 0.0005063988611014284,
      "loss": 2.2137,
      "step": 59557
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.8380100131034851,
      "learning_rate": 0.0005063958925028392,
      "loss": 2.5561,
      "step": 59558
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7250903248786926,
      "learning_rate": 0.0005063929238658771,
      "loss": 2.1637,
      "step": 59559
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.9377170205116272,
      "learning_rate": 0.0005063899551905426,
      "loss": 2.0116,
      "step": 59560
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7489689588546753,
      "learning_rate": 0.0005063869864768364,
      "loss": 2.5004,
      "step": 59561
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.8250718712806702,
      "learning_rate": 0.0005063840177247589,
      "loss": 2.2596,
      "step": 59562
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.8526027798652649,
      "learning_rate": 0.0005063810489343108,
      "loss": 2.1977,
      "step": 59563
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7606178522109985,
      "learning_rate": 0.0005063780801054925,
      "loss": 2.2016,
      "step": 59564
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7631013989448547,
      "learning_rate": 0.0005063751112383047,
      "loss": 2.1347,
      "step": 59565
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.679669201374054,
      "learning_rate": 0.0005063721423327479,
      "loss": 2.2743,
      "step": 59566
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7324252724647522,
      "learning_rate": 0.0005063691733888224,
      "loss": 2.1859,
      "step": 59567
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.750268816947937,
      "learning_rate": 0.0005063662044065291,
      "loss": 2.2739,
      "step": 59568
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7166984677314758,
      "learning_rate": 0.0005063632353858685,
      "loss": 2.0624,
      "step": 59569
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.752868115901947,
      "learning_rate": 0.000506360266326841,
      "loss": 2.2577,
      "step": 59570
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7630786895751953,
      "learning_rate": 0.0005063572972294471,
      "loss": 2.3827,
      "step": 59571
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7347151041030884,
      "learning_rate": 0.0005063543280936878,
      "loss": 2.2488,
      "step": 59572
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7786006331443787,
      "learning_rate": 0.0005063513589195631,
      "loss": 2.1087,
      "step": 59573
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.6846345067024231,
      "learning_rate": 0.0005063483897070739,
      "loss": 2.2934,
      "step": 59574
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7624136209487915,
      "learning_rate": 0.0005063454204562206,
      "loss": 2.1955,
      "step": 59575
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.79265296459198,
      "learning_rate": 0.0005063424511670038,
      "loss": 2.0096,
      "step": 59576
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7986440062522888,
      "learning_rate": 0.000506339481839424,
      "loss": 2.292,
      "step": 59577
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7091949582099915,
      "learning_rate": 0.0005063365124734819,
      "loss": 2.2554,
      "step": 59578
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.8521773815155029,
      "learning_rate": 0.0005063335430691778,
      "loss": 2.1244,
      "step": 59579
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.739862859249115,
      "learning_rate": 0.0005063305736265124,
      "loss": 2.291,
      "step": 59580
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7536493539810181,
      "learning_rate": 0.0005063276041454863,
      "loss": 2.1973,
      "step": 59581
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7781334519386292,
      "learning_rate": 0.0005063246346261001,
      "loss": 2.1558,
      "step": 59582
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.8430210947990417,
      "learning_rate": 0.0005063216650683541,
      "loss": 2.3363,
      "step": 59583
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7856285572052002,
      "learning_rate": 0.0005063186954722492,
      "loss": 2.1559,
      "step": 59584
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7810474038124084,
      "learning_rate": 0.0005063157258377856,
      "loss": 2.1654,
      "step": 59585
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.09901762008667,
      "learning_rate": 0.0005063127561649641,
      "loss": 2.1902,
      "step": 59586
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.6965338587760925,
      "learning_rate": 0.000506309786453785,
      "loss": 1.8767,
      "step": 59587
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.9252684712409973,
      "learning_rate": 0.0005063068167042492,
      "loss": 2.3859,
      "step": 59588
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7956523299217224,
      "learning_rate": 0.0005063038469163571,
      "loss": 2.2306,
      "step": 59589
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7312651872634888,
      "learning_rate": 0.0005063008770901091,
      "loss": 2.3452,
      "step": 59590
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7789507508277893,
      "learning_rate": 0.0005062979072255059,
      "loss": 2.5575,
      "step": 59591
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.9150442481040955,
      "learning_rate": 0.000506294937322548,
      "loss": 1.9292,
      "step": 59592
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7315290570259094,
      "learning_rate": 0.000506291967381236,
      "loss": 2.3437,
      "step": 59593
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7720601558685303,
      "learning_rate": 0.0005062889974015704,
      "loss": 2.3047,
      "step": 59594
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.8332085609436035,
      "learning_rate": 0.0005062860273835519,
      "loss": 2.1638,
      "step": 59595
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7821359038352966,
      "learning_rate": 0.0005062830573271808,
      "loss": 1.9839,
      "step": 59596
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7943609356880188,
      "learning_rate": 0.0005062800872324579,
      "loss": 2.3141,
      "step": 59597
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7331408262252808,
      "learning_rate": 0.0005062771170993835,
      "loss": 2.2386,
      "step": 59598
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.8221544027328491,
      "learning_rate": 0.0005062741469279584,
      "loss": 2.3077,
      "step": 59599
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7542053461074829,
      "learning_rate": 0.000506271176718183,
      "loss": 2.2633,
      "step": 59600
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7609733939170837,
      "learning_rate": 0.0005062682064700579,
      "loss": 2.3754,
      "step": 59601
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.8680840134620667,
      "learning_rate": 0.0005062652361835836,
      "loss": 2.2397,
      "step": 59602
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.9068056344985962,
      "learning_rate": 0.0005062622658587607,
      "loss": 2.1099,
      "step": 59603
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7303435802459717,
      "learning_rate": 0.0005062592954955899,
      "loss": 2.5139,
      "step": 59604
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.0264403820037842,
      "learning_rate": 0.0005062563250940715,
      "loss": 2.296,
      "step": 59605
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.8079366683959961,
      "learning_rate": 0.0005062533546542062,
      "loss": 2.4685,
      "step": 59606
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.8466072082519531,
      "learning_rate": 0.0005062503841759945,
      "loss": 2.1828,
      "step": 59607
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7027785181999207,
      "learning_rate": 0.0005062474136594369,
      "loss": 2.0134,
      "step": 59608
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.0493501424789429,
      "learning_rate": 0.000506244443104534,
      "loss": 2.2016,
      "step": 59609
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.9767589569091797,
      "learning_rate": 0.0005062414725112864,
      "loss": 2.4808,
      "step": 59610
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7293300032615662,
      "learning_rate": 0.0005062385018796946,
      "loss": 2.1291,
      "step": 59611
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.8699185848236084,
      "learning_rate": 0.0005062355312097592,
      "loss": 2.139,
      "step": 59612
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.9367833137512207,
      "learning_rate": 0.0005062325605014806,
      "loss": 2.1131,
      "step": 59613
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7588633894920349,
      "learning_rate": 0.0005062295897548597,
      "loss": 2.1981,
      "step": 59614
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7464383244514465,
      "learning_rate": 0.0005062266189698967,
      "loss": 2.1546,
      "step": 59615
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.8227822780609131,
      "learning_rate": 0.0005062236481465923,
      "loss": 2.3042,
      "step": 59616
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.9790906310081482,
      "learning_rate": 0.0005062206772849471,
      "loss": 2.0943,
      "step": 59617
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7601984143257141,
      "learning_rate": 0.0005062177063849615,
      "loss": 2.1668,
      "step": 59618
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.8406466245651245,
      "learning_rate": 0.0005062147354466362,
      "loss": 2.158,
      "step": 59619
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.8093639016151428,
      "learning_rate": 0.0005062117644699714,
      "loss": 2.3103,
      "step": 59620
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.9263235926628113,
      "learning_rate": 0.0005062087934549682,
      "loss": 2.2088,
      "step": 59621
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.8207579851150513,
      "learning_rate": 0.000506205822401627,
      "loss": 2.242,
      "step": 59622
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7894731163978577,
      "learning_rate": 0.0005062028513099481,
      "loss": 2.316,
      "step": 59623
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.8057302236557007,
      "learning_rate": 0.0005061998801799321,
      "loss": 2.4153,
      "step": 59624
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.0070276260375977,
      "learning_rate": 0.0005061969090115798,
      "loss": 2.2759,
      "step": 59625
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7503215074539185,
      "learning_rate": 0.0005061939378048915,
      "loss": 2.2163,
      "step": 59626
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.8042718768119812,
      "learning_rate": 0.000506190966559868,
      "loss": 2.262,
      "step": 59627
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.802369236946106,
      "learning_rate": 0.0005061879952765096,
      "loss": 2.2322,
      "step": 59628
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.035535216331482,
      "learning_rate": 0.000506185023954817,
      "loss": 2.0915,
      "step": 59629
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.6885936260223389,
      "learning_rate": 0.0005061820525947907,
      "loss": 2.3239,
      "step": 59630
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.8853822946548462,
      "learning_rate": 0.0005061790811964312,
      "loss": 2.3277,
      "step": 59631
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.9394687414169312,
      "learning_rate": 0.0005061761097597392,
      "loss": 2.326,
      "step": 59632
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.8033081889152527,
      "learning_rate": 0.0005061731382847151,
      "loss": 2.4926,
      "step": 59633
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.8213415741920471,
      "learning_rate": 0.0005061701667713596,
      "loss": 2.4423,
      "step": 59634
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.8619277477264404,
      "learning_rate": 0.0005061671952196731,
      "loss": 2.2602,
      "step": 59635
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7226829528808594,
      "learning_rate": 0.0005061642236296563,
      "loss": 2.0284,
      "step": 59636
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7848958373069763,
      "learning_rate": 0.0005061612520013096,
      "loss": 2.0594,
      "step": 59637
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.8052133917808533,
      "learning_rate": 0.0005061582803346337,
      "loss": 2.3276,
      "step": 59638
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.6787373423576355,
      "learning_rate": 0.000506155308629629,
      "loss": 2.2718,
      "step": 59639
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.669738233089447,
      "learning_rate": 0.0005061523368862963,
      "loss": 2.1639,
      "step": 59640
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7433966994285583,
      "learning_rate": 0.0005061493651046358,
      "loss": 2.3372,
      "step": 59641
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7289500832557678,
      "learning_rate": 0.0005061463932846484,
      "loss": 2.1697,
      "step": 59642
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7510916590690613,
      "learning_rate": 0.0005061434214263344,
      "loss": 2.3077,
      "step": 59643
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7893095016479492,
      "learning_rate": 0.0005061404495296945,
      "loss": 2.1843,
      "step": 59644
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7078975439071655,
      "learning_rate": 0.0005061374775947291,
      "loss": 2.2107,
      "step": 59645
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.844224214553833,
      "learning_rate": 0.0005061345056214389,
      "loss": 2.2033,
      "step": 59646
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7748661637306213,
      "learning_rate": 0.0005061315336098245,
      "loss": 2.3055,
      "step": 59647
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.8012181520462036,
      "learning_rate": 0.0005061285615598863,
      "loss": 2.1187,
      "step": 59648
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7301304340362549,
      "learning_rate": 0.0005061255894716248,
      "loss": 2.2737,
      "step": 59649
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.8227277398109436,
      "learning_rate": 0.0005061226173450408,
      "loss": 2.5516,
      "step": 59650
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.8052531480789185,
      "learning_rate": 0.0005061196451801347,
      "loss": 2.3336,
      "step": 59651
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.8716902732849121,
      "learning_rate": 0.000506116672976907,
      "loss": 2.1667,
      "step": 59652
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7818335890769958,
      "learning_rate": 0.0005061137007353583,
      "loss": 2.2712,
      "step": 59653
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.8988943099975586,
      "learning_rate": 0.0005061107284554892,
      "loss": 2.2314,
      "step": 59654
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.8799469470977783,
      "learning_rate": 0.0005061077561373003,
      "loss": 2.1242,
      "step": 59655
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.8176565170288086,
      "learning_rate": 0.000506104783780792,
      "loss": 2.1472,
      "step": 59656
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7959691882133484,
      "learning_rate": 0.000506101811385965,
      "loss": 2.42,
      "step": 59657
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.8078070878982544,
      "learning_rate": 0.0005060988389528197,
      "loss": 2.375,
      "step": 59658
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.1738661527633667,
      "learning_rate": 0.0005060958664813569,
      "loss": 2.1008,
      "step": 59659
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7415478229522705,
      "learning_rate": 0.0005060928939715768,
      "loss": 2.529,
      "step": 59660
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.8260372877120972,
      "learning_rate": 0.0005060899214234802,
      "loss": 2.2638,
      "step": 59661
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7712257504463196,
      "learning_rate": 0.0005060869488370675,
      "loss": 2.3884,
      "step": 59662
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.0563702583312988,
      "learning_rate": 0.0005060839762123395,
      "loss": 2.3851,
      "step": 59663
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7166857123374939,
      "learning_rate": 0.0005060810035492965,
      "loss": 2.2992,
      "step": 59664
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7784066796302795,
      "learning_rate": 0.0005060780308479392,
      "loss": 2.1793,
      "step": 59665
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.8753145933151245,
      "learning_rate": 0.0005060750581082682,
      "loss": 2.2909,
      "step": 59666
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.8143512606620789,
      "learning_rate": 0.0005060720853302837,
      "loss": 2.2289,
      "step": 59667
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7171977162361145,
      "learning_rate": 0.0005060691125139868,
      "loss": 2.2749,
      "step": 59668
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.8204098343849182,
      "learning_rate": 0.0005060661396593776,
      "loss": 2.3587,
      "step": 59669
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.0202858448028564,
      "learning_rate": 0.0005060631667664568,
      "loss": 2.2583,
      "step": 59670
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.6981387734413147,
      "learning_rate": 0.000506060193835225,
      "loss": 2.3008,
      "step": 59671
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.6959696412086487,
      "learning_rate": 0.0005060572208656828,
      "loss": 2.3477,
      "step": 59672
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.8940105438232422,
      "learning_rate": 0.0005060542478578305,
      "loss": 2.1767,
      "step": 59673
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7426359057426453,
      "learning_rate": 0.000506051274811669,
      "loss": 2.2927,
      "step": 59674
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.6969394087791443,
      "learning_rate": 0.0005060483017271986,
      "loss": 2.1037,
      "step": 59675
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7316416501998901,
      "learning_rate": 0.0005060453286044199,
      "loss": 2.347,
      "step": 59676
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7950436472892761,
      "learning_rate": 0.0005060423554433336,
      "loss": 2.0285,
      "step": 59677
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7854124307632446,
      "learning_rate": 0.00050603938224394,
      "loss": 2.3503,
      "step": 59678
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7600737810134888,
      "learning_rate": 0.0005060364090062399,
      "loss": 2.2947,
      "step": 59679
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7800009250640869,
      "learning_rate": 0.0005060334357302336,
      "loss": 2.135,
      "step": 59680
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.8127260208129883,
      "learning_rate": 0.0005060304624159219,
      "loss": 2.2615,
      "step": 59681
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7814686298370361,
      "learning_rate": 0.0005060274890633053,
      "loss": 2.0358,
      "step": 59682
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7928890585899353,
      "learning_rate": 0.0005060245156723842,
      "loss": 2.3291,
      "step": 59683
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7637761235237122,
      "learning_rate": 0.0005060215422431593,
      "loss": 1.9833,
      "step": 59684
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.6822978258132935,
      "learning_rate": 0.000506018568775631,
      "loss": 2.0058,
      "step": 59685
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.8771734833717346,
      "learning_rate": 0.0005060155952698001,
      "loss": 2.2666,
      "step": 59686
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7522649765014648,
      "learning_rate": 0.000506012621725667,
      "loss": 2.3269,
      "step": 59687
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.9129785299301147,
      "learning_rate": 0.0005060096481432323,
      "loss": 2.3738,
      "step": 59688
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.9223535656929016,
      "learning_rate": 0.0005060066745224964,
      "loss": 2.1297,
      "step": 59689
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7493225336074829,
      "learning_rate": 0.00050600370086346,
      "loss": 2.0158,
      "step": 59690
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7745795249938965,
      "learning_rate": 0.0005060007271661236,
      "loss": 2.3336,
      "step": 59691
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7973461151123047,
      "learning_rate": 0.0005059977534304878,
      "loss": 2.3157,
      "step": 59692
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7609004974365234,
      "learning_rate": 0.0005059947796565532,
      "loss": 2.2645,
      "step": 59693
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.646454393863678,
      "learning_rate": 0.0005059918058443202,
      "loss": 2.0035,
      "step": 59694
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7882074117660522,
      "learning_rate": 0.0005059888319937895,
      "loss": 2.153,
      "step": 59695
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7407649159431458,
      "learning_rate": 0.0005059858581049615,
      "loss": 2.3065,
      "step": 59696
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7163068056106567,
      "learning_rate": 0.0005059828841778369,
      "loss": 2.3256,
      "step": 59697
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7313042283058167,
      "learning_rate": 0.0005059799102124162,
      "loss": 2.1427,
      "step": 59698
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7725716829299927,
      "learning_rate": 0.0005059769362087,
      "loss": 2.2393,
      "step": 59699
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.686165452003479,
      "learning_rate": 0.0005059739621666887,
      "loss": 2.2978,
      "step": 59700
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7029699087142944,
      "learning_rate": 0.000505970988086383,
      "loss": 2.3022,
      "step": 59701
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7648154497146606,
      "learning_rate": 0.0005059680139677833,
      "loss": 2.2357,
      "step": 59702
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.769541323184967,
      "learning_rate": 0.0005059650398108904,
      "loss": 2.2403,
      "step": 59703
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7790800929069519,
      "learning_rate": 0.0005059620656157047,
      "loss": 2.3393,
      "step": 59704
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7632744312286377,
      "learning_rate": 0.0005059590913822266,
      "loss": 2.4995,
      "step": 59705
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7183757424354553,
      "learning_rate": 0.0005059561171104571,
      "loss": 2.1907,
      "step": 59706
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.6769615411758423,
      "learning_rate": 0.0005059531428003961,
      "loss": 2.093,
      "step": 59707
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.6990810632705688,
      "learning_rate": 0.0005059501684520449,
      "loss": 2.073,
      "step": 59708
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7348026633262634,
      "learning_rate": 0.0005059471940654035,
      "loss": 2.3297,
      "step": 59709
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7083126902580261,
      "learning_rate": 0.0005059442196404725,
      "loss": 2.0897,
      "step": 59710
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7318748235702515,
      "learning_rate": 0.0005059412451772528,
      "loss": 2.2079,
      "step": 59711
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.8272051215171814,
      "learning_rate": 0.0005059382706757447,
      "loss": 2.2636,
      "step": 59712
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7008837461471558,
      "learning_rate": 0.0005059352961359488,
      "loss": 2.2599,
      "step": 59713
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7335686683654785,
      "learning_rate": 0.0005059323215578656,
      "loss": 2.0764,
      "step": 59714
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.6444486975669861,
      "learning_rate": 0.0005059293469414956,
      "loss": 2.2525,
      "step": 59715
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.6912781596183777,
      "learning_rate": 0.0005059263722868396,
      "loss": 2.1946,
      "step": 59716
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.6917454600334167,
      "learning_rate": 0.0005059233975938978,
      "loss": 2.327,
      "step": 59717
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.6933855414390564,
      "learning_rate": 0.0005059204228626712,
      "loss": 2.2899,
      "step": 59718
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.6766535639762878,
      "learning_rate": 0.00050591744809316,
      "loss": 2.2676,
      "step": 59719
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.6667323708534241,
      "learning_rate": 0.0005059144732853648,
      "loss": 2.2879,
      "step": 59720
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7346656918525696,
      "learning_rate": 0.0005059114984392862,
      "loss": 2.3146,
      "step": 59721
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.6802854537963867,
      "learning_rate": 0.0005059085235549249,
      "loss": 2.309,
      "step": 59722
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7041165828704834,
      "learning_rate": 0.0005059055486322812,
      "loss": 2.3995,
      "step": 59723
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7335678935050964,
      "learning_rate": 0.0005059025736713558,
      "loss": 2.2709,
      "step": 59724
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.6820918321609497,
      "learning_rate": 0.0005058995986721493,
      "loss": 2.1806,
      "step": 59725
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7039234042167664,
      "learning_rate": 0.0005058966236346621,
      "loss": 2.3009,
      "step": 59726
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7309751510620117,
      "learning_rate": 0.0005058936485588948,
      "loss": 2.4263,
      "step": 59727
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7234973311424255,
      "learning_rate": 0.000505890673444848,
      "loss": 2.3546,
      "step": 59728
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.663459300994873,
      "learning_rate": 0.0005058876982925223,
      "loss": 2.1748,
      "step": 59729
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7250646948814392,
      "learning_rate": 0.0005058847231019181,
      "loss": 2.4879,
      "step": 59730
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7132020592689514,
      "learning_rate": 0.0005058817478730361,
      "loss": 2.273,
      "step": 59731
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.9269931316375732,
      "learning_rate": 0.0005058787726058768,
      "loss": 2.3323,
      "step": 59732
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7793102860450745,
      "learning_rate": 0.0005058757973004408,
      "loss": 2.3792,
      "step": 59733
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7805464267730713,
      "learning_rate": 0.0005058728219567286,
      "loss": 2.2577,
      "step": 59734
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.6770358681678772,
      "learning_rate": 0.0005058698465747406,
      "loss": 2.4536,
      "step": 59735
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7636991739273071,
      "learning_rate": 0.0005058668711544778,
      "loss": 2.152,
      "step": 59736
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7781863808631897,
      "learning_rate": 0.0005058638956959402,
      "loss": 2.4561,
      "step": 59737
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7618839740753174,
      "learning_rate": 0.0005058609201991286,
      "loss": 2.3818,
      "step": 59738
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7931901812553406,
      "learning_rate": 0.0005058579446640437,
      "loss": 2.1544,
      "step": 59739
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.714588463306427,
      "learning_rate": 0.0005058549690906859,
      "loss": 2.3312,
      "step": 59740
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7322375774383545,
      "learning_rate": 0.0005058519934790557,
      "loss": 2.2596,
      "step": 59741
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7331376075744629,
      "learning_rate": 0.0005058490178291539,
      "loss": 2.2639,
      "step": 59742
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7665753960609436,
      "learning_rate": 0.0005058460421409807,
      "loss": 2.1859,
      "step": 59743
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7571130990982056,
      "learning_rate": 0.0005058430664145369,
      "loss": 2.1373,
      "step": 59744
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.9287351965904236,
      "learning_rate": 0.0005058400906498231,
      "loss": 2.2225,
      "step": 59745
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7890344858169556,
      "learning_rate": 0.0005058371148468396,
      "loss": 2.3899,
      "step": 59746
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.742084801197052,
      "learning_rate": 0.000505834139005587,
      "loss": 2.0722,
      "step": 59747
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.8823027014732361,
      "learning_rate": 0.0005058311631260662,
      "loss": 2.2049,
      "step": 59748
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7280551195144653,
      "learning_rate": 0.0005058281872082774,
      "loss": 2.2191,
      "step": 59749
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7098997235298157,
      "learning_rate": 0.0005058252112522212,
      "loss": 2.0093,
      "step": 59750
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.8435181975364685,
      "learning_rate": 0.0005058222352578983,
      "loss": 2.3524,
      "step": 59751
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7480038404464722,
      "learning_rate": 0.0005058192592253091,
      "loss": 2.2856,
      "step": 59752
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.239661455154419,
      "learning_rate": 0.0005058162831544542,
      "loss": 2.1692,
      "step": 59753
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7478736042976379,
      "learning_rate": 0.0005058133070453341,
      "loss": 2.1929,
      "step": 59754
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.8034705519676208,
      "learning_rate": 0.0005058103308979495,
      "loss": 2.1227,
      "step": 59755
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7966196537017822,
      "learning_rate": 0.000505807354712301,
      "loss": 2.1742,
      "step": 59756
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7104167938232422,
      "learning_rate": 0.0005058043784883888,
      "loss": 2.1402,
      "step": 59757
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.8482301831245422,
      "learning_rate": 0.0005058014022262139,
      "loss": 2.3786,
      "step": 59758
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7444067001342773,
      "learning_rate": 0.0005057984259257765,
      "loss": 2.3256,
      "step": 59759
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7858318090438843,
      "learning_rate": 0.0005057954495870774,
      "loss": 2.0897,
      "step": 59760
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7188701629638672,
      "learning_rate": 0.000505792473210117,
      "loss": 1.977,
      "step": 59761
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7609461545944214,
      "learning_rate": 0.0005057894967948958,
      "loss": 1.9942,
      "step": 59762
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7420675754547119,
      "learning_rate": 0.0005057865203414144,
      "loss": 2.2405,
      "step": 59763
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7639345526695251,
      "learning_rate": 0.0005057835438496735,
      "loss": 2.3011,
      "step": 59764
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7416098713874817,
      "learning_rate": 0.0005057805673196736,
      "loss": 2.1035,
      "step": 59765
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7849240899085999,
      "learning_rate": 0.0005057775907514151,
      "loss": 2.1534,
      "step": 59766
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7233533263206482,
      "learning_rate": 0.0005057746141448988,
      "loss": 2.3483,
      "step": 59767
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7578275203704834,
      "learning_rate": 0.0005057716375001251,
      "loss": 2.2617,
      "step": 59768
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.706770658493042,
      "learning_rate": 0.0005057686608170945,
      "loss": 2.1766,
      "step": 59769
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7436904907226562,
      "learning_rate": 0.0005057656840958076,
      "loss": 2.2319,
      "step": 59770
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.6746898293495178,
      "learning_rate": 0.000505762707336265,
      "loss": 2.4322,
      "step": 59771
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.6975591778755188,
      "learning_rate": 0.0005057597305384672,
      "loss": 2.3517,
      "step": 59772
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.9505566358566284,
      "learning_rate": 0.0005057567537024148,
      "loss": 1.9444,
      "step": 59773
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.799839973449707,
      "learning_rate": 0.0005057537768281084,
      "loss": 2.3617,
      "step": 59774
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.6967321634292603,
      "learning_rate": 0.0005057507999155484,
      "loss": 2.0606,
      "step": 59775
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7455937266349792,
      "learning_rate": 0.0005057478229647355,
      "loss": 2.2523,
      "step": 59776
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7375020980834961,
      "learning_rate": 0.0005057448459756702,
      "loss": 2.2855,
      "step": 59777
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7409855723381042,
      "learning_rate": 0.000505741868948353,
      "loss": 2.133,
      "step": 59778
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.78065425157547,
      "learning_rate": 0.0005057388918827846,
      "loss": 2.4623,
      "step": 59779
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.8932858109474182,
      "learning_rate": 0.0005057359147789652,
      "loss": 2.2861,
      "step": 59780
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7726431488990784,
      "learning_rate": 0.0005057329376368959,
      "loss": 2.1691,
      "step": 59781
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.8046289086341858,
      "learning_rate": 0.0005057299604565769,
      "loss": 2.155,
      "step": 59782
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.8098927736282349,
      "learning_rate": 0.0005057269832380088,
      "loss": 2.0434,
      "step": 59783
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.6816763877868652,
      "learning_rate": 0.0005057240059811921,
      "loss": 2.4022,
      "step": 59784
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.684727668762207,
      "learning_rate": 0.0005057210286861276,
      "loss": 2.195,
      "step": 59785
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7482021450996399,
      "learning_rate": 0.0005057180513528155,
      "loss": 2.2328,
      "step": 59786
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7165468335151672,
      "learning_rate": 0.0005057150739812565,
      "loss": 2.2442,
      "step": 59787
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7596302032470703,
      "learning_rate": 0.0005057120965714513,
      "loss": 2.2499,
      "step": 59788
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7681477665901184,
      "learning_rate": 0.0005057091191234004,
      "loss": 2.0922,
      "step": 59789
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.9004974365234375,
      "learning_rate": 0.0005057061416371041,
      "loss": 2.2104,
      "step": 59790
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7180619239807129,
      "learning_rate": 0.0005057031641125632,
      "loss": 2.2645,
      "step": 59791
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7077035903930664,
      "learning_rate": 0.0005057001865497783,
      "loss": 2.265,
      "step": 59792
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.6862160563468933,
      "learning_rate": 0.0005056972089487497,
      "loss": 2.1064,
      "step": 59793
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7811279892921448,
      "learning_rate": 0.0005056942313094783,
      "loss": 2.178,
      "step": 59794
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7548374533653259,
      "learning_rate": 0.0005056912536319643,
      "loss": 2.0879,
      "step": 59795
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7279786467552185,
      "learning_rate": 0.0005056882759162085,
      "loss": 2.2397,
      "step": 59796
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7955643534660339,
      "learning_rate": 0.0005056852981622114,
      "loss": 2.1573,
      "step": 59797
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7095530033111572,
      "learning_rate": 0.0005056823203699734,
      "loss": 2.2555,
      "step": 59798
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7658142447471619,
      "learning_rate": 0.0005056793425394952,
      "loss": 2.3787,
      "step": 59799
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.6951039433479309,
      "learning_rate": 0.0005056763646707774,
      "loss": 2.2364,
      "step": 59800
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.8647535443305969,
      "learning_rate": 0.0005056733867638205,
      "loss": 2.3662,
      "step": 59801
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.8106948733329773,
      "learning_rate": 0.000505670408818625,
      "loss": 2.1861,
      "step": 59802
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7532556056976318,
      "learning_rate": 0.0005056674308351914,
      "loss": 2.1047,
      "step": 59803
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.6890982985496521,
      "learning_rate": 0.0005056644528135204,
      "loss": 2.2208,
      "step": 59804
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7378585934638977,
      "learning_rate": 0.0005056614747536125,
      "loss": 2.3924,
      "step": 59805
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7428688406944275,
      "learning_rate": 0.0005056584966554682,
      "loss": 2.2108,
      "step": 59806
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7299559116363525,
      "learning_rate": 0.0005056555185190883,
      "loss": 2.2589,
      "step": 59807
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7234799265861511,
      "learning_rate": 0.0005056525403444729,
      "loss": 2.1464,
      "step": 59808
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.768329381942749,
      "learning_rate": 0.000505649562131623,
      "loss": 2.1968,
      "step": 59809
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.77620929479599,
      "learning_rate": 0.0005056465838805389,
      "loss": 2.1063,
      "step": 59810
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7326063513755798,
      "learning_rate": 0.0005056436055912211,
      "loss": 2.1632,
      "step": 59811
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.856708824634552,
      "learning_rate": 0.0005056406272636704,
      "loss": 2.2197,
      "step": 59812
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.8070288300514221,
      "learning_rate": 0.0005056376488978872,
      "loss": 2.3579,
      "step": 59813
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7389081120491028,
      "learning_rate": 0.000505634670493872,
      "loss": 2.1,
      "step": 59814
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.8832451701164246,
      "learning_rate": 0.0005056316920516256,
      "loss": 2.2704,
      "step": 59815
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7810119390487671,
      "learning_rate": 0.0005056287135711483,
      "loss": 2.1254,
      "step": 59816
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7712242603302002,
      "learning_rate": 0.0005056257350524408,
      "loss": 2.2421,
      "step": 59817
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.9110493063926697,
      "learning_rate": 0.0005056227564955036,
      "loss": 2.0903,
      "step": 59818
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7918296456336975,
      "learning_rate": 0.0005056197779003371,
      "loss": 2.1439,
      "step": 59819
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.8074376583099365,
      "learning_rate": 0.0005056167992669421,
      "loss": 2.2313,
      "step": 59820
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7398796081542969,
      "learning_rate": 0.000505613820595319,
      "loss": 2.1528,
      "step": 59821
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7000581622123718,
      "learning_rate": 0.0005056108418854685,
      "loss": 2.1597,
      "step": 59822
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7211049795150757,
      "learning_rate": 0.000505607863137391,
      "loss": 2.3963,
      "step": 59823
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7449787259101868,
      "learning_rate": 0.0005056048843510871,
      "loss": 2.3081,
      "step": 59824
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7800760269165039,
      "learning_rate": 0.0005056019055265574,
      "loss": 2.2989,
      "step": 59825
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.704491913318634,
      "learning_rate": 0.0005055989266638024,
      "loss": 2.2862,
      "step": 59826
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7144933938980103,
      "learning_rate": 0.0005055959477628228,
      "loss": 2.1987,
      "step": 59827
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7389988899230957,
      "learning_rate": 0.000505592968823619,
      "loss": 2.1359,
      "step": 59828
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7403419613838196,
      "learning_rate": 0.0005055899898461914,
      "loss": 2.1395,
      "step": 59829
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7619433999061584,
      "learning_rate": 0.0005055870108305408,
      "loss": 2.4653,
      "step": 59830
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.9607600569725037,
      "learning_rate": 0.000505584031776668,
      "loss": 2.2162,
      "step": 59831
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7722778916358948,
      "learning_rate": 0.0005055810526845729,
      "loss": 1.9966,
      "step": 59832
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.8725671768188477,
      "learning_rate": 0.0005055780735542565,
      "loss": 2.2906,
      "step": 59833
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7662615180015564,
      "learning_rate": 0.0005055750943857193,
      "loss": 2.2931,
      "step": 59834
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.6930530667304993,
      "learning_rate": 0.0005055721151789618,
      "loss": 2.3411,
      "step": 59835
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.767518162727356,
      "learning_rate": 0.0005055691359339845,
      "loss": 2.0304,
      "step": 59836
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7615424990653992,
      "learning_rate": 0.0005055661566507881,
      "loss": 2.4514,
      "step": 59837
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7939203977584839,
      "learning_rate": 0.0005055631773293731,
      "loss": 2.1628,
      "step": 59838
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.8384526968002319,
      "learning_rate": 0.0005055601979697398,
      "loss": 2.2826,
      "step": 59839
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7568363547325134,
      "learning_rate": 0.0005055572185718892,
      "loss": 2.2048,
      "step": 59840
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7939437627792358,
      "learning_rate": 0.0005055542391358216,
      "loss": 2.3326,
      "step": 59841
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.8122406601905823,
      "learning_rate": 0.0005055512596615375,
      "loss": 2.3756,
      "step": 59842
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.700907289981842,
      "learning_rate": 0.0005055482801490376,
      "loss": 1.894,
      "step": 59843
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7784979343414307,
      "learning_rate": 0.0005055453005983223,
      "loss": 2.3812,
      "step": 59844
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7612932324409485,
      "learning_rate": 0.0005055423210093923,
      "loss": 2.4313,
      "step": 59845
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.6536112427711487,
      "learning_rate": 0.0005055393413822483,
      "loss": 2.157,
      "step": 59846
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.6802507638931274,
      "learning_rate": 0.0005055363617168905,
      "loss": 2.1542,
      "step": 59847
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7747275829315186,
      "learning_rate": 0.0005055333820133196,
      "loss": 2.1165,
      "step": 59848
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7416362762451172,
      "learning_rate": 0.0005055304022715361,
      "loss": 2.2481,
      "step": 59849
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7400160431861877,
      "learning_rate": 0.0005055274224915407,
      "loss": 2.241,
      "step": 59850
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7655430436134338,
      "learning_rate": 0.0005055244426733339,
      "loss": 2.3165,
      "step": 59851
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7934607863426208,
      "learning_rate": 0.0005055214628169162,
      "loss": 2.3125,
      "step": 59852
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.731807291507721,
      "learning_rate": 0.0005055184829222881,
      "loss": 2.0929,
      "step": 59853
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.6798417568206787,
      "learning_rate": 0.0005055155029894503,
      "loss": 2.0106,
      "step": 59854
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7795639038085938,
      "learning_rate": 0.0005055125230184033,
      "loss": 2.4132,
      "step": 59855
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.8225361704826355,
      "learning_rate": 0.0005055095430091476,
      "loss": 2.285,
      "step": 59856
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7446048259735107,
      "learning_rate": 0.0005055065629616839,
      "loss": 2.3254,
      "step": 59857
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.9973723292350769,
      "learning_rate": 0.0005055035828760126,
      "loss": 2.2298,
      "step": 59858
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.8148465156555176,
      "learning_rate": 0.0005055006027521343,
      "loss": 2.2957,
      "step": 59859
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.6979833841323853,
      "learning_rate": 0.0005054976225900494,
      "loss": 2.2231,
      "step": 59860
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.008968710899353,
      "learning_rate": 0.0005054946423897589,
      "loss": 2.1099,
      "step": 59861
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.6862527132034302,
      "learning_rate": 0.0005054916621512629,
      "loss": 2.334,
      "step": 59862
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7111582159996033,
      "learning_rate": 0.0005054886818745621,
      "loss": 2.1332,
      "step": 59863
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7182471752166748,
      "learning_rate": 0.0005054857015596572,
      "loss": 2.4998,
      "step": 59864
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7580856680870056,
      "learning_rate": 0.0005054827212065485,
      "loss": 2.1981,
      "step": 59865
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.769752562046051,
      "learning_rate": 0.0005054797408152367,
      "loss": 2.1667,
      "step": 59866
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7295357584953308,
      "learning_rate": 0.0005054767603857224,
      "loss": 2.0735,
      "step": 59867
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.6739621162414551,
      "learning_rate": 0.0005054737799180061,
      "loss": 2.0609,
      "step": 59868
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7473781108856201,
      "learning_rate": 0.0005054707994120883,
      "loss": 2.3042,
      "step": 59869
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7754681706428528,
      "learning_rate": 0.0005054678188679696,
      "loss": 2.3281,
      "step": 59870
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7188075184822083,
      "learning_rate": 0.0005054648382856505,
      "loss": 2.1285,
      "step": 59871
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.6819027662277222,
      "learning_rate": 0.0005054618576651317,
      "loss": 2.2737,
      "step": 59872
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7461089491844177,
      "learning_rate": 0.0005054588770064136,
      "loss": 2.159,
      "step": 59873
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7466406226158142,
      "learning_rate": 0.0005054558963094969,
      "loss": 2.2004,
      "step": 59874
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.6808207631111145,
      "learning_rate": 0.0005054529155743819,
      "loss": 2.1202,
      "step": 59875
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7553831934928894,
      "learning_rate": 0.0005054499348010696,
      "loss": 2.2351,
      "step": 59876
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.759673535823822,
      "learning_rate": 0.0005054469539895601,
      "loss": 2.0738,
      "step": 59877
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7298771739006042,
      "learning_rate": 0.0005054439731398541,
      "loss": 2.2911,
      "step": 59878
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.799505352973938,
      "learning_rate": 0.0005054409922519523,
      "loss": 2.3642,
      "step": 59879
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7019193172454834,
      "learning_rate": 0.0005054380113258551,
      "loss": 2.2817,
      "step": 59880
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7017278671264648,
      "learning_rate": 0.000505435030361563,
      "loss": 2.1737,
      "step": 59881
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.749150276184082,
      "learning_rate": 0.0005054320493590768,
      "loss": 2.3208,
      "step": 59882
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.8533284664154053,
      "learning_rate": 0.0005054290683183969,
      "loss": 2.2614,
      "step": 59883
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7725145816802979,
      "learning_rate": 0.0005054260872395237,
      "loss": 2.2021,
      "step": 59884
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7102662324905396,
      "learning_rate": 0.000505423106122458,
      "loss": 2.1335,
      "step": 59885
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7887212038040161,
      "learning_rate": 0.0005054201249672003,
      "loss": 2.224,
      "step": 59886
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.8805699348449707,
      "learning_rate": 0.000505417143773751,
      "loss": 1.9023,
      "step": 59887
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7315428256988525,
      "learning_rate": 0.0005054141625421109,
      "loss": 2.27,
      "step": 59888
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7578727602958679,
      "learning_rate": 0.0005054111812722805,
      "loss": 2.212,
      "step": 59889
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.9309459328651428,
      "learning_rate": 0.0005054081999642602,
      "loss": 2.2304,
      "step": 59890
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7095273733139038,
      "learning_rate": 0.0005054052186180505,
      "loss": 2.3112,
      "step": 59891
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7385409474372864,
      "learning_rate": 0.0005054022372336523,
      "loss": 2.2788,
      "step": 59892
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.8363327980041504,
      "learning_rate": 0.0005053992558110658,
      "loss": 2.2285,
      "step": 59893
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7670785188674927,
      "learning_rate": 0.0005053962743502918,
      "loss": 2.0456,
      "step": 59894
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7114222645759583,
      "learning_rate": 0.0005053932928513307,
      "loss": 2.0314,
      "step": 59895
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7599952220916748,
      "learning_rate": 0.0005053903113141832,
      "loss": 2.1084,
      "step": 59896
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.8010884523391724,
      "learning_rate": 0.0005053873297388496,
      "loss": 2.2314,
      "step": 59897
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7064078450202942,
      "learning_rate": 0.0005053843481253308,
      "loss": 2.1672,
      "step": 59898
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.6983398199081421,
      "learning_rate": 0.000505381366473627,
      "loss": 2.0507,
      "step": 59899
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.0305263996124268,
      "learning_rate": 0.000505378384783739,
      "loss": 2.2397,
      "step": 59900
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7372314929962158,
      "learning_rate": 0.0005053754030556674,
      "loss": 2.0942,
      "step": 59901
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7559490203857422,
      "learning_rate": 0.0005053724212894125,
      "loss": 2.1319,
      "step": 59902
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7754329442977905,
      "learning_rate": 0.000505369439484975,
      "loss": 2.1252,
      "step": 59903
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.8719305396080017,
      "learning_rate": 0.0005053664576423555,
      "loss": 2.1763,
      "step": 59904
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7785783410072327,
      "learning_rate": 0.0005053634757615544,
      "loss": 2.4162,
      "step": 59905
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.8160496950149536,
      "learning_rate": 0.0005053604938425725,
      "loss": 1.9852,
      "step": 59906
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.8016095757484436,
      "learning_rate": 0.00050535751188541,
      "loss": 2.0339,
      "step": 59907
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7118951678276062,
      "learning_rate": 0.0005053545298900678,
      "loss": 2.2191,
      "step": 59908
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7585505247116089,
      "learning_rate": 0.0005053515478565463,
      "loss": 2.1448,
      "step": 59909
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7713939547538757,
      "learning_rate": 0.000505348565784846,
      "loss": 2.1301,
      "step": 59910
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7813879251480103,
      "learning_rate": 0.0005053455836749678,
      "loss": 2.4637,
      "step": 59911
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.8231266140937805,
      "learning_rate": 0.0005053426015269116,
      "loss": 2.4186,
      "step": 59912
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.8564013242721558,
      "learning_rate": 0.0005053396193406786,
      "loss": 2.4315,
      "step": 59913
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7053755521774292,
      "learning_rate": 0.0005053366371162689,
      "loss": 2.2723,
      "step": 59914
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.825018584728241,
      "learning_rate": 0.0005053336548536834,
      "loss": 2.3769,
      "step": 59915
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.6495863199234009,
      "learning_rate": 0.0005053306725529224,
      "loss": 2.1949,
      "step": 59916
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7510326504707336,
      "learning_rate": 0.0005053276902139866,
      "loss": 2.3678,
      "step": 59917
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7412399053573608,
      "learning_rate": 0.0005053247078368765,
      "loss": 2.2311,
      "step": 59918
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7112846374511719,
      "learning_rate": 0.0005053217254215927,
      "loss": 2.0636,
      "step": 59919
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7498552799224854,
      "learning_rate": 0.0005053187429681356,
      "loss": 2.2942,
      "step": 59920
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.8159520030021667,
      "learning_rate": 0.000505315760476506,
      "loss": 2.3271,
      "step": 59921
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7560117840766907,
      "learning_rate": 0.0005053127779467043,
      "loss": 2.3495,
      "step": 59922
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.6831013560295105,
      "learning_rate": 0.0005053097953787311,
      "loss": 2.285,
      "step": 59923
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.699669599533081,
      "learning_rate": 0.0005053068127725868,
      "loss": 2.3644,
      "step": 59924
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7174608707427979,
      "learning_rate": 0.0005053038301282722,
      "loss": 2.1722,
      "step": 59925
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7453898191452026,
      "learning_rate": 0.0005053008474457877,
      "loss": 2.327,
      "step": 59926
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7415821552276611,
      "learning_rate": 0.0005052978647251339,
      "loss": 2.2428,
      "step": 59927
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7588968873023987,
      "learning_rate": 0.0005052948819663114,
      "loss": 2.3511,
      "step": 59928
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.73577481508255,
      "learning_rate": 0.0005052918991693206,
      "loss": 2.0834,
      "step": 59929
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7242279052734375,
      "learning_rate": 0.0005052889163341623,
      "loss": 2.2686,
      "step": 59930
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7410224080085754,
      "learning_rate": 0.0005052859334608368,
      "loss": 2.3182,
      "step": 59931
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7072109580039978,
      "learning_rate": 0.0005052829505493449,
      "loss": 2.3532,
      "step": 59932
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.8725917339324951,
      "learning_rate": 0.0005052799675996869,
      "loss": 2.2485,
      "step": 59933
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7816115617752075,
      "learning_rate": 0.0005052769846118634,
      "loss": 2.3699,
      "step": 59934
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7027175426483154,
      "learning_rate": 0.0005052740015858752,
      "loss": 2.1597,
      "step": 59935
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.6689492464065552,
      "learning_rate": 0.0005052710185217226,
      "loss": 2.1587,
      "step": 59936
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.754392147064209,
      "learning_rate": 0.0005052680354194064,
      "loss": 2.1655,
      "step": 59937
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.689135730266571,
      "learning_rate": 0.0005052650522789268,
      "loss": 2.147,
      "step": 59938
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.8023639917373657,
      "learning_rate": 0.0005052620691002847,
      "loss": 1.9941,
      "step": 59939
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7459883093833923,
      "learning_rate": 0.0005052590858834804,
      "loss": 2.1918,
      "step": 59940
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7407529950141907,
      "learning_rate": 0.0005052561026285146,
      "loss": 2.414,
      "step": 59941
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7590593695640564,
      "learning_rate": 0.0005052531193353878,
      "loss": 2.234,
      "step": 59942
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.8044271469116211,
      "learning_rate": 0.0005052501360041006,
      "loss": 1.9579,
      "step": 59943
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7551223635673523,
      "learning_rate": 0.0005052471526346535,
      "loss": 2.1655,
      "step": 59944
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7469688057899475,
      "learning_rate": 0.000505244169227047,
      "loss": 2.2132,
      "step": 59945
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7590472102165222,
      "learning_rate": 0.000505241185781282,
      "loss": 2.2518,
      "step": 59946
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7012293934822083,
      "learning_rate": 0.0005052382022973586,
      "loss": 2.4588,
      "step": 59947
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.9575127959251404,
      "learning_rate": 0.0005052352187752776,
      "loss": 2.2395,
      "step": 59948
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7108148336410522,
      "learning_rate": 0.0005052322352150394,
      "loss": 2.1648,
      "step": 59949
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.727825939655304,
      "learning_rate": 0.0005052292516166447,
      "loss": 2.2408,
      "step": 59950
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.75352543592453,
      "learning_rate": 0.0005052262679800941,
      "loss": 2.2204,
      "step": 59951
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.8262774348258972,
      "learning_rate": 0.0005052232843053879,
      "loss": 1.9717,
      "step": 59952
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7642423510551453,
      "learning_rate": 0.0005052203005925269,
      "loss": 2.3203,
      "step": 59953
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.6847927570343018,
      "learning_rate": 0.0005052173168415115,
      "loss": 2.1228,
      "step": 59954
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.8373335003852844,
      "learning_rate": 0.0005052143330523425,
      "loss": 2.2687,
      "step": 59955
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.9548159241676331,
      "learning_rate": 0.0005052113492250202,
      "loss": 2.2603,
      "step": 59956
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7087518572807312,
      "learning_rate": 0.0005052083653595452,
      "loss": 2.189,
      "step": 59957
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7259297966957092,
      "learning_rate": 0.0005052053814559181,
      "loss": 2.3902,
      "step": 59958
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.9750885963439941,
      "learning_rate": 0.0005052023975141394,
      "loss": 2.1794,
      "step": 59959
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7353540062904358,
      "learning_rate": 0.0005051994135342098,
      "loss": 2.3228,
      "step": 59960
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.1016899347305298,
      "learning_rate": 0.0005051964295161297,
      "loss": 2.2223,
      "step": 59961
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7397720217704773,
      "learning_rate": 0.0005051934454598997,
      "loss": 2.3032,
      "step": 59962
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7416239380836487,
      "learning_rate": 0.0005051904613655204,
      "loss": 2.4365,
      "step": 59963
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.9577208161354065,
      "learning_rate": 0.0005051874772329922,
      "loss": 2.1551,
      "step": 59964
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.8339650630950928,
      "learning_rate": 0.000505184493062316,
      "loss": 2.1362,
      "step": 59965
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7814753651618958,
      "learning_rate": 0.0005051815088534921,
      "loss": 2.1944,
      "step": 59966
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.732312798500061,
      "learning_rate": 0.0005051785246065209,
      "loss": 2.2661,
      "step": 59967
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7631354928016663,
      "learning_rate": 0.0005051755403214032,
      "loss": 2.2437,
      "step": 59968
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.8320764899253845,
      "learning_rate": 0.0005051725559981396,
      "loss": 2.2359,
      "step": 59969
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7144268155097961,
      "learning_rate": 0.0005051695716367304,
      "loss": 2.3452,
      "step": 59970
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7056975364685059,
      "learning_rate": 0.0005051665872371763,
      "loss": 2.3369,
      "step": 59971
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.8858258724212646,
      "learning_rate": 0.0005051636027994781,
      "loss": 2.1922,
      "step": 59972
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.8890020847320557,
      "learning_rate": 0.0005051606183236359,
      "loss": 2.1924,
      "step": 59973
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7935871481895447,
      "learning_rate": 0.0005051576338096506,
      "loss": 2.1096,
      "step": 59974
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.8042927384376526,
      "learning_rate": 0.0005051546492575225,
      "loss": 2.2544,
      "step": 59975
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7700948119163513,
      "learning_rate": 0.0005051516646672522,
      "loss": 2.216,
      "step": 59976
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7975555062294006,
      "learning_rate": 0.0005051486800388405,
      "loss": 2.2442,
      "step": 59977
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.8781709671020508,
      "learning_rate": 0.0005051456953722878,
      "loss": 2.3158,
      "step": 59978
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.8493776917457581,
      "learning_rate": 0.0005051427106675946,
      "loss": 2.2378,
      "step": 59979
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.6729547381401062,
      "learning_rate": 0.0005051397259247615,
      "loss": 2.0718,
      "step": 59980
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.8265799283981323,
      "learning_rate": 0.000505136741143789,
      "loss": 2.308,
      "step": 59981
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.901771605014801,
      "learning_rate": 0.0005051337563246777,
      "loss": 2.4321,
      "step": 59982
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.8467116951942444,
      "learning_rate": 0.0005051307714674283,
      "loss": 2.3499,
      "step": 59983
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.8126558065414429,
      "learning_rate": 0.0005051277865720411,
      "loss": 2.4653,
      "step": 59984
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7807708382606506,
      "learning_rate": 0.0005051248016385168,
      "loss": 2.217,
      "step": 59985
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7662394046783447,
      "learning_rate": 0.000505121816666856,
      "loss": 2.1401,
      "step": 59986
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.8264995813369751,
      "learning_rate": 0.0005051188316570591,
      "loss": 2.2621,
      "step": 59987
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.8035779595375061,
      "learning_rate": 0.0005051158466091267,
      "loss": 2.2582,
      "step": 59988
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7369210124015808,
      "learning_rate": 0.0005051128615230594,
      "loss": 2.271,
      "step": 59989
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.6912011504173279,
      "learning_rate": 0.0005051098763988578,
      "loss": 2.427,
      "step": 59990
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7957264184951782,
      "learning_rate": 0.0005051068912365224,
      "loss": 2.1551,
      "step": 59991
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.814167320728302,
      "learning_rate": 0.0005051039060360537,
      "loss": 2.1585,
      "step": 59992
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7713273167610168,
      "learning_rate": 0.0005051009207974524,
      "loss": 2.3414,
      "step": 59993
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7265331745147705,
      "learning_rate": 0.0005050979355207189,
      "loss": 2.1019,
      "step": 59994
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.774681568145752,
      "learning_rate": 0.0005050949502058539,
      "loss": 2.2469,
      "step": 59995
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7917425632476807,
      "learning_rate": 0.0005050919648528578,
      "loss": 2.1809,
      "step": 59996
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.8113093376159668,
      "learning_rate": 0.0005050889794617312,
      "loss": 2.2224,
      "step": 59997
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7319151163101196,
      "learning_rate": 0.0005050859940324748,
      "loss": 2.2702,
      "step": 59998
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7338868975639343,
      "learning_rate": 0.000505083008565089,
      "loss": 2.3023,
      "step": 59999
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.6994634866714478,
      "learning_rate": 0.0005050800230595743,
      "loss": 2.23,
      "step": 60000
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7291880249977112,
      "learning_rate": 0.0005050770375159315,
      "loss": 2.4693,
      "step": 60001
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.6925298571586609,
      "learning_rate": 0.000505074051934161,
      "loss": 1.9772,
      "step": 60002
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7288798093795776,
      "learning_rate": 0.0005050710663142633,
      "loss": 2.3328,
      "step": 60003
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7165830731391907,
      "learning_rate": 0.000505068080656239,
      "loss": 2.2442,
      "step": 60004
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7584094405174255,
      "learning_rate": 0.0005050650949600887,
      "loss": 2.2296,
      "step": 60005
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.9098765254020691,
      "learning_rate": 0.0005050621092258129,
      "loss": 2.1217,
      "step": 60006
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7447428703308105,
      "learning_rate": 0.0005050591234534122,
      "loss": 2.3371,
      "step": 60007
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.6999605894088745,
      "learning_rate": 0.000505056137642887,
      "loss": 2.2054,
      "step": 60008
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7369033098220825,
      "learning_rate": 0.0005050531517942382,
      "loss": 2.4995,
      "step": 60009
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.6980552077293396,
      "learning_rate": 0.0005050501659074661,
      "loss": 2.0354,
      "step": 60010
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7599602937698364,
      "learning_rate": 0.0005050471799825713,
      "loss": 2.1333,
      "step": 60011
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.8086260557174683,
      "learning_rate": 0.0005050441940195543,
      "loss": 2.0847,
      "step": 60012
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.8166865110397339,
      "learning_rate": 0.0005050412080184159,
      "loss": 2.2434,
      "step": 60013
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.8151228427886963,
      "learning_rate": 0.0005050382219791563,
      "loss": 2.4973,
      "step": 60014
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7493941783905029,
      "learning_rate": 0.0005050352359017762,
      "loss": 2.3207,
      "step": 60015
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7794542908668518,
      "learning_rate": 0.0005050322497862763,
      "loss": 2.1274,
      "step": 60016
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.8872785568237305,
      "learning_rate": 0.0005050292636326569,
      "loss": 2.3441,
      "step": 60017
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7351748943328857,
      "learning_rate": 0.0005050262774409188,
      "loss": 2.3259,
      "step": 60018
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.8409713506698608,
      "learning_rate": 0.0005050232912110624,
      "loss": 2.1895,
      "step": 60019
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.8027809858322144,
      "learning_rate": 0.0005050203049430883,
      "loss": 1.9849,
      "step": 60020
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7305948734283447,
      "learning_rate": 0.0005050173186369971,
      "loss": 2.2412,
      "step": 60021
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.8516787886619568,
      "learning_rate": 0.0005050143322927892,
      "loss": 2.3303,
      "step": 60022
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7583057880401611,
      "learning_rate": 0.0005050113459104654,
      "loss": 2.2595,
      "step": 60023
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.6664135456085205,
      "learning_rate": 0.000505008359490026,
      "loss": 2.3364,
      "step": 60024
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7701478600502014,
      "learning_rate": 0.0005050053730314718,
      "loss": 2.2455,
      "step": 60025
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7100972533226013,
      "learning_rate": 0.0005050023865348031,
      "loss": 2.2889,
      "step": 60026
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7523159384727478,
      "learning_rate": 0.0005049994000000206,
      "loss": 2.3569,
      "step": 60027
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.725550651550293,
      "learning_rate": 0.000504996413427125,
      "loss": 2.1925,
      "step": 60028
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.9635343551635742,
      "learning_rate": 0.0005049934268161165,
      "loss": 2.095,
      "step": 60029
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7331517934799194,
      "learning_rate": 0.000504990440166996,
      "loss": 1.9845,
      "step": 60030
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.6714689135551453,
      "learning_rate": 0.0005049874534797639,
      "loss": 2.2274,
      "step": 60031
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.8433477282524109,
      "learning_rate": 0.0005049844667544206,
      "loss": 2.2037,
      "step": 60032
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7208775281906128,
      "learning_rate": 0.000504981479990967,
      "loss": 2.4461,
      "step": 60033
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7189382910728455,
      "learning_rate": 0.0005049784931894034,
      "loss": 2.2282,
      "step": 60034
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.870764434337616,
      "learning_rate": 0.0005049755063497304,
      "loss": 1.963,
      "step": 60035
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7025665044784546,
      "learning_rate": 0.0005049725194719487,
      "loss": 2.3486,
      "step": 60036
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7215902209281921,
      "learning_rate": 0.0005049695325560586,
      "loss": 2.2995,
      "step": 60037
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.793417751789093,
      "learning_rate": 0.0005049665456020608,
      "loss": 2.3022,
      "step": 60038
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7278326749801636,
      "learning_rate": 0.0005049635586099559,
      "loss": 2.4307,
      "step": 60039
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.8122528791427612,
      "learning_rate": 0.0005049605715797444,
      "loss": 2.3722,
      "step": 60040
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7015160918235779,
      "learning_rate": 0.0005049575845114269,
      "loss": 2.146,
      "step": 60041
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7640625834465027,
      "learning_rate": 0.0005049545974050039,
      "loss": 2.0535,
      "step": 60042
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7268089056015015,
      "learning_rate": 0.000504951610260476,
      "loss": 2.2194,
      "step": 60043
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7389414310455322,
      "learning_rate": 0.0005049486230778436,
      "loss": 1.9151,
      "step": 60044
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7872282266616821,
      "learning_rate": 0.0005049456358571076,
      "loss": 2.1899,
      "step": 60045
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.8062377572059631,
      "learning_rate": 0.0005049426485982681,
      "loss": 2.3757,
      "step": 60046
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7148653864860535,
      "learning_rate": 0.000504939661301326,
      "loss": 2.1821,
      "step": 60047
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7708777189254761,
      "learning_rate": 0.0005049366739662818,
      "loss": 2.2016,
      "step": 60048
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7175078988075256,
      "learning_rate": 0.0005049336865931359,
      "loss": 2.3171,
      "step": 60049
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7411367297172546,
      "learning_rate": 0.000504930699181889,
      "loss": 2.0519,
      "step": 60050
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.6416572332382202,
      "learning_rate": 0.0005049277117325417,
      "loss": 2.015,
      "step": 60051
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7124398350715637,
      "learning_rate": 0.0005049247242450943,
      "loss": 2.3208,
      "step": 60052
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.8184228539466858,
      "learning_rate": 0.0005049217367195476,
      "loss": 2.1213,
      "step": 60053
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.8058767914772034,
      "learning_rate": 0.0005049187491559022,
      "loss": 2.0663,
      "step": 60054
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.9057220220565796,
      "learning_rate": 0.0005049157615541584,
      "loss": 2.3619,
      "step": 60055
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.6911911368370056,
      "learning_rate": 0.0005049127739143169,
      "loss": 2.1922,
      "step": 60056
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7556922435760498,
      "learning_rate": 0.0005049097862363782,
      "loss": 2.2167,
      "step": 60057
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.8153749108314514,
      "learning_rate": 0.000504906798520343,
      "loss": 2.44,
      "step": 60058
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.771588146686554,
      "learning_rate": 0.0005049038107662118,
      "loss": 2.2897,
      "step": 60059
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.830301821231842,
      "learning_rate": 0.000504900822973985,
      "loss": 2.3759,
      "step": 60060
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7972649931907654,
      "learning_rate": 0.0005048978351436633,
      "loss": 2.3852,
      "step": 60061
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.6898874044418335,
      "learning_rate": 0.0005048948472752473,
      "loss": 2.4984,
      "step": 60062
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7865362167358398,
      "learning_rate": 0.0005048918593687375,
      "loss": 2.2123,
      "step": 60063
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.8362469673156738,
      "learning_rate": 0.0005048888714241342,
      "loss": 2.0878,
      "step": 60064
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7279800772666931,
      "learning_rate": 0.0005048858834414384,
      "loss": 2.3044,
      "step": 60065
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.754360020160675,
      "learning_rate": 0.0005048828954206504,
      "loss": 2.2354,
      "step": 60066
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.8691293001174927,
      "learning_rate": 0.0005048799073617709,
      "loss": 2.0615,
      "step": 60067
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7484150528907776,
      "learning_rate": 0.0005048769192648002,
      "loss": 2.2727,
      "step": 60068
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7016712427139282,
      "learning_rate": 0.000504873931129739,
      "loss": 2.1522,
      "step": 60069
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.8004761934280396,
      "learning_rate": 0.000504870942956588,
      "loss": 2.1308,
      "step": 60070
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7539529204368591,
      "learning_rate": 0.0005048679547453475,
      "loss": 2.2186,
      "step": 60071
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.6968610882759094,
      "learning_rate": 0.0005048649664960183,
      "loss": 2.2484,
      "step": 60072
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7659531831741333,
      "learning_rate": 0.0005048619782086007,
      "loss": 1.9886,
      "step": 60073
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.818994402885437,
      "learning_rate": 0.0005048589898830955,
      "loss": 2.2819,
      "step": 60074
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.8189625144004822,
      "learning_rate": 0.0005048560015195031,
      "loss": 2.2976,
      "step": 60075
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7822574377059937,
      "learning_rate": 0.0005048530131178242,
      "loss": 2.2747,
      "step": 60076
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.6891299486160278,
      "learning_rate": 0.0005048500246780591,
      "loss": 2.4003,
      "step": 60077
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.725246250629425,
      "learning_rate": 0.0005048470362002086,
      "loss": 2.3028,
      "step": 60078
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.8212173581123352,
      "learning_rate": 0.0005048440476842731,
      "loss": 2.1699,
      "step": 60079
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7290099859237671,
      "learning_rate": 0.0005048410591302534,
      "loss": 2.3585,
      "step": 60080
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7598164081573486,
      "learning_rate": 0.0005048380705381497,
      "loss": 2.2579,
      "step": 60081
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7136473655700684,
      "learning_rate": 0.0005048350819079629,
      "loss": 2.2131,
      "step": 60082
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.8630189299583435,
      "learning_rate": 0.0005048320932396932,
      "loss": 2.3339,
      "step": 60083
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.699260950088501,
      "learning_rate": 0.0005048291045333414,
      "loss": 2.1671,
      "step": 60084
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.83165043592453,
      "learning_rate": 0.0005048261157889081,
      "loss": 2.5272,
      "step": 60085
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.8137182593345642,
      "learning_rate": 0.0005048231270063937,
      "loss": 2.3929,
      "step": 60086
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7665737271308899,
      "learning_rate": 0.0005048201381857989,
      "loss": 2.2344,
      "step": 60087
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.6932809948921204,
      "learning_rate": 0.000504817149327124,
      "loss": 2.0629,
      "step": 60088
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.9817632436752319,
      "learning_rate": 0.0005048141604303697,
      "loss": 2.1898,
      "step": 60089
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.8698620200157166,
      "learning_rate": 0.0005048111714955368,
      "loss": 2.3539,
      "step": 60090
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7529534697532654,
      "learning_rate": 0.0005048081825226255,
      "loss": 2.1996,
      "step": 60091
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.890271008014679,
      "learning_rate": 0.0005048051935116365,
      "loss": 2.0958,
      "step": 60092
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7768502235412598,
      "learning_rate": 0.0005048022044625704,
      "loss": 2.1657,
      "step": 60093
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7457029819488525,
      "learning_rate": 0.0005047992153754277,
      "loss": 2.2101,
      "step": 60094
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.0319079160690308,
      "learning_rate": 0.000504796226250209,
      "loss": 2.4859,
      "step": 60095
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.818264365196228,
      "learning_rate": 0.0005047932370869147,
      "loss": 2.3122,
      "step": 60096
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7384822368621826,
      "learning_rate": 0.0005047902478855455,
      "loss": 1.8507,
      "step": 60097
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7461954951286316,
      "learning_rate": 0.000504787258646102,
      "loss": 2.352,
      "step": 60098
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.9024356007575989,
      "learning_rate": 0.0005047842693685846,
      "loss": 2.2655,
      "step": 60099
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7802455425262451,
      "learning_rate": 0.000504781280052994,
      "loss": 2.2469,
      "step": 60100
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.6997843980789185,
      "learning_rate": 0.0005047782906993306,
      "loss": 2.2684,
      "step": 60101
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.6903069019317627,
      "learning_rate": 0.0005047753013075951,
      "loss": 2.3116,
      "step": 60102
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.8792914748191833,
      "learning_rate": 0.0005047723118777881,
      "loss": 2.1434,
      "step": 60103
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.8514339327812195,
      "learning_rate": 0.0005047693224099099,
      "loss": 2.213,
      "step": 60104
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.8826243877410889,
      "learning_rate": 0.0005047663329039615,
      "loss": 2.2726,
      "step": 60105
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.876469612121582,
      "learning_rate": 0.0005047633433599428,
      "loss": 2.2264,
      "step": 60106
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7359649538993835,
      "learning_rate": 0.0005047603537778549,
      "loss": 1.9822,
      "step": 60107
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.8519049882888794,
      "learning_rate": 0.0005047573641576983,
      "loss": 2.2804,
      "step": 60108
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.803722620010376,
      "learning_rate": 0.0005047543744994733,
      "loss": 2.3028,
      "step": 60109
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7582478523254395,
      "learning_rate": 0.0005047513848031805,
      "loss": 2.1547,
      "step": 60110
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7816511988639832,
      "learning_rate": 0.0005047483950688208,
      "loss": 2.1815,
      "step": 60111
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7186675071716309,
      "learning_rate": 0.0005047454052963944,
      "loss": 2.2278,
      "step": 60112
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7101467847824097,
      "learning_rate": 0.0005047424154859018,
      "loss": 2.1478,
      "step": 60113
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7537092566490173,
      "learning_rate": 0.000504739425637344,
      "loss": 2.0785,
      "step": 60114
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7605268359184265,
      "learning_rate": 0.0005047364357507211,
      "loss": 2.1876,
      "step": 60115
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.6597821116447449,
      "learning_rate": 0.000504733445826034,
      "loss": 2.1592,
      "step": 60116
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7575587630271912,
      "learning_rate": 0.0005047304558632829,
      "loss": 2.2154,
      "step": 60117
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.789873480796814,
      "learning_rate": 0.0005047274658624687,
      "loss": 2.1655,
      "step": 60118
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.8339256048202515,
      "learning_rate": 0.0005047244758235916,
      "loss": 2.2523,
      "step": 60119
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.790062665939331,
      "learning_rate": 0.0005047214857466525,
      "loss": 2.0642,
      "step": 60120
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.8237751722335815,
      "learning_rate": 0.0005047184956316517,
      "loss": 2.2676,
      "step": 60121
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.76703280210495,
      "learning_rate": 0.00050471550547859,
      "loss": 2.1834,
      "step": 60122
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7113003134727478,
      "learning_rate": 0.0005047125152874678,
      "loss": 2.1211,
      "step": 60123
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.6853923797607422,
      "learning_rate": 0.0005047095250582856,
      "loss": 2.1581,
      "step": 60124
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7897247076034546,
      "learning_rate": 0.0005047065347910441,
      "loss": 2.1624,
      "step": 60125
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7496705651283264,
      "learning_rate": 0.0005047035444857437,
      "loss": 2.2461,
      "step": 60126
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7766784429550171,
      "learning_rate": 0.0005047005541423853,
      "loss": 2.1678,
      "step": 60127
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7120177149772644,
      "learning_rate": 0.0005046975637609689,
      "loss": 2.1665,
      "step": 60128
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.694329023361206,
      "learning_rate": 0.0005046945733414955,
      "loss": 2.2642,
      "step": 60129
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7428519129753113,
      "learning_rate": 0.0005046915828839655,
      "loss": 2.3283,
      "step": 60130
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7679710984230042,
      "learning_rate": 0.0005046885923883794,
      "loss": 2.0398,
      "step": 60131
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7566273808479309,
      "learning_rate": 0.000504685601854738,
      "loss": 2.1736,
      "step": 60132
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7253199815750122,
      "learning_rate": 0.0005046826112830415,
      "loss": 2.1853,
      "step": 60133
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7669705748558044,
      "learning_rate": 0.0005046796206732908,
      "loss": 2.4043,
      "step": 60134
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7048848271369934,
      "learning_rate": 0.0005046766300254862,
      "loss": 2.2136,
      "step": 60135
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7691848874092102,
      "learning_rate": 0.0005046736393396283,
      "loss": 2.0245,
      "step": 60136
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7491909861564636,
      "learning_rate": 0.0005046706486157178,
      "loss": 2.1645,
      "step": 60137
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.8033586740493774,
      "learning_rate": 0.0005046676578537552,
      "loss": 2.3522,
      "step": 60138
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7185468673706055,
      "learning_rate": 0.000504664667053741,
      "loss": 2.1226,
      "step": 60139
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7678651213645935,
      "learning_rate": 0.0005046616762156757,
      "loss": 2.2997,
      "step": 60140
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7916732430458069,
      "learning_rate": 0.0005046586853395599,
      "loss": 2.1011,
      "step": 60141
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.757634699344635,
      "learning_rate": 0.0005046556944253943,
      "loss": 2.2194,
      "step": 60142
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7951326966285706,
      "learning_rate": 0.0005046527034731792,
      "loss": 2.2278,
      "step": 60143
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.8282544612884521,
      "learning_rate": 0.0005046497124829155,
      "loss": 1.9473,
      "step": 60144
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7416975498199463,
      "learning_rate": 0.0005046467214546034,
      "loss": 2.025,
      "step": 60145
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7527082562446594,
      "learning_rate": 0.0005046437303882436,
      "loss": 2.1693,
      "step": 60146
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.8706346750259399,
      "learning_rate": 0.0005046407392838367,
      "loss": 2.4651,
      "step": 60147
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7042939066886902,
      "learning_rate": 0.0005046377481413833,
      "loss": 2.1805,
      "step": 60148
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7058677673339844,
      "learning_rate": 0.0005046347569608838,
      "loss": 2.4472,
      "step": 60149
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.8293730020523071,
      "learning_rate": 0.0005046317657423389,
      "loss": 2.294,
      "step": 60150
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7885721921920776,
      "learning_rate": 0.0005046287744857491,
      "loss": 2.3041,
      "step": 60151
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7400579452514648,
      "learning_rate": 0.0005046257831911149,
      "loss": 2.2908,
      "step": 60152
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.8183066248893738,
      "learning_rate": 0.0005046227918584368,
      "loss": 2.3168,
      "step": 60153
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7280240654945374,
      "learning_rate": 0.0005046198004877156,
      "loss": 2.3528,
      "step": 60154
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.8286888003349304,
      "learning_rate": 0.0005046168090789518,
      "loss": 2.2279,
      "step": 60155
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7363200783729553,
      "learning_rate": 0.0005046138176321456,
      "loss": 2.2458,
      "step": 60156
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.741276204586029,
      "learning_rate": 0.0005046108261472981,
      "loss": 2.3413,
      "step": 60157
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7344716787338257,
      "learning_rate": 0.0005046078346244094,
      "loss": 2.3886,
      "step": 60158
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7475210428237915,
      "learning_rate": 0.0005046048430634803,
      "loss": 2.1888,
      "step": 60159
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7412373423576355,
      "learning_rate": 0.0005046018514645113,
      "loss": 2.0229,
      "step": 60160
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7555757164955139,
      "learning_rate": 0.000504598859827503,
      "loss": 2.0606,
      "step": 60161
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7908351421356201,
      "learning_rate": 0.0005045958681524558,
      "loss": 2.3146,
      "step": 60162
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.9045370221138,
      "learning_rate": 0.0005045928764393705,
      "loss": 2.3288,
      "step": 60163
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7290937304496765,
      "learning_rate": 0.0005045898846882474,
      "loss": 2.0848,
      "step": 60164
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.848808228969574,
      "learning_rate": 0.0005045868928990874,
      "loss": 2.2178,
      "step": 60165
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.896716833114624,
      "learning_rate": 0.0005045839010718906,
      "loss": 2.4254,
      "step": 60166
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.8190870881080627,
      "learning_rate": 0.0005045809092066579,
      "loss": 2.0507,
      "step": 60167
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7208108901977539,
      "learning_rate": 0.0005045779173033897,
      "loss": 1.9468,
      "step": 60168
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.3478965759277344,
      "learning_rate": 0.0005045749253620866,
      "loss": 2.1587,
      "step": 60169
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.088792085647583,
      "learning_rate": 0.0005045719333827491,
      "loss": 2.1017,
      "step": 60170
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.6843361258506775,
      "learning_rate": 0.000504568941365378,
      "loss": 2.124,
      "step": 60171
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.338186264038086,
      "learning_rate": 0.0005045659493099735,
      "loss": 2.2899,
      "step": 60172
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.1776927709579468,
      "learning_rate": 0.0005045629572165364,
      "loss": 2.0682,
      "step": 60173
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.0096791982650757,
      "learning_rate": 0.0005045599650850672,
      "loss": 2.4076,
      "step": 60174
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.8140316605567932,
      "learning_rate": 0.0005045569729155664,
      "loss": 2.293,
      "step": 60175
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.908367931842804,
      "learning_rate": 0.0005045539807080346,
      "loss": 2.2548,
      "step": 60176
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.074146032333374,
      "learning_rate": 0.0005045509884624724,
      "loss": 2.2346,
      "step": 60177
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.70293790102005,
      "learning_rate": 0.0005045479961788803,
      "loss": 2.1196,
      "step": 60178
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7327956557273865,
      "learning_rate": 0.0005045450038572588,
      "loss": 2.1483,
      "step": 60179
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.9678930640220642,
      "learning_rate": 0.0005045420114976086,
      "loss": 2.1625,
      "step": 60180
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.9552390575408936,
      "learning_rate": 0.0005045390190999301,
      "loss": 2.2185,
      "step": 60181
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.6532427668571472,
      "learning_rate": 0.000504536026664224,
      "loss": 2.1713,
      "step": 60182
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.8555005192756653,
      "learning_rate": 0.0005045330341904908,
      "loss": 2.3571,
      "step": 60183
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.8066641688346863,
      "learning_rate": 0.000504530041678731,
      "loss": 2.3415,
      "step": 60184
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.8659153580665588,
      "learning_rate": 0.0005045270491289452,
      "loss": 2.3021,
      "step": 60185
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7528385519981384,
      "learning_rate": 0.000504524056541134,
      "loss": 2.2153,
      "step": 60186
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7982156276702881,
      "learning_rate": 0.0005045210639152979,
      "loss": 2.4119,
      "step": 60187
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.937334418296814,
      "learning_rate": 0.0005045180712514375,
      "loss": 2.1174,
      "step": 60188
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.8402534127235413,
      "learning_rate": 0.0005045150785495533,
      "loss": 2.1202,
      "step": 60189
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.808466374874115,
      "learning_rate": 0.000504512085809646,
      "loss": 2.3746,
      "step": 60190
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7328329086303711,
      "learning_rate": 0.0005045090930317159,
      "loss": 1.8683,
      "step": 60191
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.9129737019538879,
      "learning_rate": 0.0005045061002157637,
      "loss": 2.2367,
      "step": 60192
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.9077315330505371,
      "learning_rate": 0.00050450310736179,
      "loss": 2.3424,
      "step": 60193
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7625443339347839,
      "learning_rate": 0.0005045001144697954,
      "loss": 2.1467,
      "step": 60194
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.9917307496070862,
      "learning_rate": 0.0005044971215397802,
      "loss": 2.2994,
      "step": 60195
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.0559252500534058,
      "learning_rate": 0.0005044941285717451,
      "loss": 2.213,
      "step": 60196
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.707352876663208,
      "learning_rate": 0.0005044911355656909,
      "loss": 2.2215,
      "step": 60197
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.015204668045044,
      "learning_rate": 0.0005044881425216178,
      "loss": 2.0717,
      "step": 60198
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.74587482213974,
      "learning_rate": 0.0005044851494395264,
      "loss": 2.3486,
      "step": 60199
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.9135667681694031,
      "learning_rate": 0.0005044821563194175,
      "loss": 2.3669,
      "step": 60200
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7376023530960083,
      "learning_rate": 0.0005044791631612915,
      "loss": 2.2127,
      "step": 60201
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.8405235409736633,
      "learning_rate": 0.000504476169965149,
      "loss": 2.2883,
      "step": 60202
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.6895833015441895,
      "learning_rate": 0.0005044731767309904,
      "loss": 2.3047,
      "step": 60203
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.6901544332504272,
      "learning_rate": 0.0005044701834588164,
      "loss": 2.1516,
      "step": 60204
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7405864000320435,
      "learning_rate": 0.0005044671901486276,
      "loss": 2.3439,
      "step": 60205
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7751243114471436,
      "learning_rate": 0.0005044641968004245,
      "loss": 2.1987,
      "step": 60206
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7344838380813599,
      "learning_rate": 0.0005044612034142076,
      "loss": 2.2943,
      "step": 60207
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7285210490226746,
      "learning_rate": 0.0005044582099899775,
      "loss": 2.2766,
      "step": 60208
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.813054084777832,
      "learning_rate": 0.0005044552165277348,
      "loss": 2.2969,
      "step": 60209
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7455704808235168,
      "learning_rate": 0.0005044522230274798,
      "loss": 2.1718,
      "step": 60210
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.0312501192092896,
      "learning_rate": 0.0005044492294892135,
      "loss": 2.2157,
      "step": 60211
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7068539261817932,
      "learning_rate": 0.0005044462359129363,
      "loss": 2.1597,
      "step": 60212
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.811806857585907,
      "learning_rate": 0.0005044432422986484,
      "loss": 2.2078,
      "step": 60213
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.2578587532043457,
      "learning_rate": 0.0005044402486463509,
      "loss": 2.2538,
      "step": 60214
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7252683043479919,
      "learning_rate": 0.0005044372549560441,
      "loss": 2.1587,
      "step": 60215
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7512328624725342,
      "learning_rate": 0.0005044342612277284,
      "loss": 2.2015,
      "step": 60216
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7977925539016724,
      "learning_rate": 0.0005044312674614045,
      "loss": 2.516,
      "step": 60217
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.058483600616455,
      "learning_rate": 0.000504428273657073,
      "loss": 1.9381,
      "step": 60218
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.8177313208580017,
      "learning_rate": 0.0005044252798147346,
      "loss": 2.0989,
      "step": 60219
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7430013418197632,
      "learning_rate": 0.0005044222859343895,
      "loss": 2.2782,
      "step": 60220
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.9369660019874573,
      "learning_rate": 0.0005044192920160385,
      "loss": 2.4606,
      "step": 60221
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7251133918762207,
      "learning_rate": 0.000504416298059682,
      "loss": 2.214,
      "step": 60222
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7220275402069092,
      "learning_rate": 0.0005044133040653209,
      "loss": 2.1217,
      "step": 60223
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.8690023422241211,
      "learning_rate": 0.0005044103100329552,
      "loss": 2.1907,
      "step": 60224
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7029692530632019,
      "learning_rate": 0.000504407315962586,
      "loss": 2.2164,
      "step": 60225
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.794458270072937,
      "learning_rate": 0.0005044043218542134,
      "loss": 2.3747,
      "step": 60226
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7903701066970825,
      "learning_rate": 0.0005044013277078383,
      "loss": 2.2157,
      "step": 60227
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.8463130593299866,
      "learning_rate": 0.0005043983335234612,
      "loss": 2.2582,
      "step": 60228
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.8106371164321899,
      "learning_rate": 0.0005043953393010825,
      "loss": 2.2106,
      "step": 60229
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7311849594116211,
      "learning_rate": 0.0005043923450407029,
      "loss": 2.0372,
      "step": 60230
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.8876814246177673,
      "learning_rate": 0.0005043893507423229,
      "loss": 2.1696,
      "step": 60231
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.8342352509498596,
      "learning_rate": 0.0005043863564059431,
      "loss": 2.3006,
      "step": 60232
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7632337808609009,
      "learning_rate": 0.0005043833620315639,
      "loss": 2.2252,
      "step": 60233
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.9382500648498535,
      "learning_rate": 0.0005043803676191861,
      "loss": 2.3727,
      "step": 60234
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.220055103302002,
      "learning_rate": 0.00050437737316881,
      "loss": 2.4726,
      "step": 60235
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.6772236227989197,
      "learning_rate": 0.0005043743786804364,
      "loss": 2.0747,
      "step": 60236
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7648808360099792,
      "learning_rate": 0.0005043713841540656,
      "loss": 2.2642,
      "step": 60237
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.8635786771774292,
      "learning_rate": 0.0005043683895896985,
      "loss": 2.2538,
      "step": 60238
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.820921778678894,
      "learning_rate": 0.0005043653949873354,
      "loss": 2.1195,
      "step": 60239
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.6764994859695435,
      "learning_rate": 0.0005043624003469769,
      "loss": 1.9135,
      "step": 60240
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.8700030446052551,
      "learning_rate": 0.0005043594056686235,
      "loss": 2.3049,
      "step": 60241
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7963197827339172,
      "learning_rate": 0.0005043564109522759,
      "loss": 2.2782,
      "step": 60242
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.9286930561065674,
      "learning_rate": 0.0005043534161979345,
      "loss": 2.2111,
      "step": 60243
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7002922892570496,
      "learning_rate": 0.0005043504214056001,
      "loss": 2.0602,
      "step": 60244
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7598168253898621,
      "learning_rate": 0.000504347426575273,
      "loss": 2.3115,
      "step": 60245
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.8408503532409668,
      "learning_rate": 0.0005043444317069538,
      "loss": 2.2336,
      "step": 60246
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.8745171427726746,
      "learning_rate": 0.0005043414368006432,
      "loss": 2.3205,
      "step": 60247
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7037904858589172,
      "learning_rate": 0.0005043384418563417,
      "loss": 2.4171,
      "step": 60248
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.832414984703064,
      "learning_rate": 0.0005043354468740497,
      "loss": 2.5755,
      "step": 60249
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.0341060161590576,
      "learning_rate": 0.000504332451853768,
      "loss": 2.2655,
      "step": 60250
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.9407615065574646,
      "learning_rate": 0.000504329456795497,
      "loss": 2.2633,
      "step": 60251
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7625523805618286,
      "learning_rate": 0.0005043264616992373,
      "loss": 2.2696,
      "step": 60252
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.9779901504516602,
      "learning_rate": 0.0005043234665649894,
      "loss": 2.4808,
      "step": 60253
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.9175058007240295,
      "learning_rate": 0.000504320471392754,
      "loss": 2.1899,
      "step": 60254
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7800936102867126,
      "learning_rate": 0.0005043174761825315,
      "loss": 2.281,
      "step": 60255
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.729249119758606,
      "learning_rate": 0.0005043144809343226,
      "loss": 2.3224,
      "step": 60256
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.106797218322754,
      "learning_rate": 0.0005043114856481277,
      "loss": 2.474,
      "step": 60257
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7678402066230774,
      "learning_rate": 0.0005043084903239474,
      "loss": 2.4872,
      "step": 60258
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.8111324310302734,
      "learning_rate": 0.0005043054949617824,
      "loss": 2.3518,
      "step": 60259
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.6881185173988342,
      "learning_rate": 0.0005043024995616331,
      "loss": 2.246,
      "step": 60260
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.8932197093963623,
      "learning_rate": 0.0005042995041235002,
      "loss": 2.093,
      "step": 60261
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.6858906745910645,
      "learning_rate": 0.000504296508647384,
      "loss": 2.2008,
      "step": 60262
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7356575727462769,
      "learning_rate": 0.0005042935131332854,
      "loss": 2.1857,
      "step": 60263
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7157773375511169,
      "learning_rate": 0.0005042905175812047,
      "loss": 2.1752,
      "step": 60264
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7697618007659912,
      "learning_rate": 0.0005042875219911424,
      "loss": 2.356,
      "step": 60265
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.8385177254676819,
      "learning_rate": 0.0005042845263630994,
      "loss": 2.4768,
      "step": 60266
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7834020853042603,
      "learning_rate": 0.000504281530697076,
      "loss": 2.2002,
      "step": 60267
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.732155442237854,
      "learning_rate": 0.0005042785349930728,
      "loss": 2.342,
      "step": 60268
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7625150680541992,
      "learning_rate": 0.0005042755392510903,
      "loss": 2.223,
      "step": 60269
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.8699870109558105,
      "learning_rate": 0.0005042725434711293,
      "loss": 2.4607,
      "step": 60270
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7570605278015137,
      "learning_rate": 0.00050426954765319,
      "loss": 2.0515,
      "step": 60271
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7771651744842529,
      "learning_rate": 0.0005042665517972731,
      "loss": 2.3342,
      "step": 60272
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.8345829844474792,
      "learning_rate": 0.0005042635559033793,
      "loss": 2.4344,
      "step": 60273
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7343312501907349,
      "learning_rate": 0.000504260559971509,
      "loss": 2.3687,
      "step": 60274
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.6956950426101685,
      "learning_rate": 0.0005042575640016629,
      "loss": 2.1472,
      "step": 60275
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.8291492462158203,
      "learning_rate": 0.0005042545679938414,
      "loss": 1.9412,
      "step": 60276
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7795833349227905,
      "learning_rate": 0.000504251571948045,
      "loss": 2.4193,
      "step": 60277
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.6879683136940002,
      "learning_rate": 0.0005042485758642746,
      "loss": 2.1396,
      "step": 60278
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.9017142653465271,
      "learning_rate": 0.0005042455797425304,
      "loss": 2.1821,
      "step": 60279
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7074548006057739,
      "learning_rate": 0.0005042425835828131,
      "loss": 2.3075,
      "step": 60280
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7245827913284302,
      "learning_rate": 0.0005042395873851233,
      "loss": 2.2238,
      "step": 60281
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7913349270820618,
      "learning_rate": 0.0005042365911494614,
      "loss": 2.3365,
      "step": 60282
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.8576739430427551,
      "learning_rate": 0.0005042335948758282,
      "loss": 2.2199,
      "step": 60283
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.689331591129303,
      "learning_rate": 0.0005042305985642239,
      "loss": 2.0793,
      "step": 60284
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.8409201502799988,
      "learning_rate": 0.0005042276022146495,
      "loss": 2.1364,
      "step": 60285
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.6778084635734558,
      "learning_rate": 0.0005042246058271053,
      "loss": 2.2706,
      "step": 60286
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.792098879814148,
      "learning_rate": 0.0005042216094015918,
      "loss": 2.3299,
      "step": 60287
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.8142789006233215,
      "learning_rate": 0.0005042186129381097,
      "loss": 2.223,
      "step": 60288
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.8683226704597473,
      "learning_rate": 0.0005042156164366596,
      "loss": 2.2681,
      "step": 60289
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7694287896156311,
      "learning_rate": 0.0005042126198972417,
      "loss": 2.0756,
      "step": 60290
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.9175705909729004,
      "learning_rate": 0.0005042096233198571,
      "loss": 2.399,
      "step": 60291
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7189516425132751,
      "learning_rate": 0.0005042066267045059,
      "loss": 2.2136,
      "step": 60292
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.677424430847168,
      "learning_rate": 0.0005042036300511889,
      "loss": 2.1632,
      "step": 60293
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7710112929344177,
      "learning_rate": 0.0005042006333599066,
      "loss": 2.4434,
      "step": 60294
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7720710039138794,
      "learning_rate": 0.0005041976366306595,
      "loss": 2.1896,
      "step": 60295
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.715286374092102,
      "learning_rate": 0.0005041946398634483,
      "loss": 2.2156,
      "step": 60296
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7483775019645691,
      "learning_rate": 0.0005041916430582734,
      "loss": 2.4724,
      "step": 60297
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7559149861335754,
      "learning_rate": 0.0005041886462151355,
      "loss": 2.2452,
      "step": 60298
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7271378636360168,
      "learning_rate": 0.0005041856493340349,
      "loss": 2.2514,
      "step": 60299
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7740716934204102,
      "learning_rate": 0.0005041826524149725,
      "loss": 2.086,
      "step": 60300
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7292460203170776,
      "learning_rate": 0.0005041796554579485,
      "loss": 2.1439,
      "step": 60301
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7168441414833069,
      "learning_rate": 0.0005041766584629638,
      "loss": 2.4014,
      "step": 60302
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7121114730834961,
      "learning_rate": 0.0005041736614300187,
      "loss": 2.2608,
      "step": 60303
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.6897507905960083,
      "learning_rate": 0.000504170664359114,
      "loss": 2.233,
      "step": 60304
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7200046181678772,
      "learning_rate": 0.00050416766725025,
      "loss": 2.2744,
      "step": 60305
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7771454453468323,
      "learning_rate": 0.0005041646701034275,
      "loss": 2.2634,
      "step": 60306
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.751464307308197,
      "learning_rate": 0.0005041616729186468,
      "loss": 2.2034,
      "step": 60307
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.6876568794250488,
      "learning_rate": 0.0005041586756959086,
      "loss": 2.2819,
      "step": 60308
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7735568881034851,
      "learning_rate": 0.0005041556784352137,
      "loss": 2.286,
      "step": 60309
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.8364555239677429,
      "learning_rate": 0.0005041526811365622,
      "loss": 2.1893,
      "step": 60310
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7721189856529236,
      "learning_rate": 0.0005041496837999548,
      "loss": 2.4041,
      "step": 60311
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7528923749923706,
      "learning_rate": 0.0005041466864253921,
      "loss": 2.1397,
      "step": 60312
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.8786023855209351,
      "learning_rate": 0.0005041436890128748,
      "loss": 2.2569,
      "step": 60313
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7786679267883301,
      "learning_rate": 0.0005041406915624033,
      "loss": 2.357,
      "step": 60314
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7784863710403442,
      "learning_rate": 0.0005041376940739783,
      "loss": 1.995,
      "step": 60315
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.8017241954803467,
      "learning_rate": 0.0005041346965476,
      "loss": 2.1013,
      "step": 60316
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7247289419174194,
      "learning_rate": 0.0005041316989832695,
      "loss": 2.4172,
      "step": 60317
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.8108651638031006,
      "learning_rate": 0.0005041287013809869,
      "loss": 2.3726,
      "step": 60318
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7564283609390259,
      "learning_rate": 0.0005041257037407529,
      "loss": 2.4723,
      "step": 60319
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.6812683343887329,
      "learning_rate": 0.0005041227060625682,
      "loss": 2.1797,
      "step": 60320
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7842326760292053,
      "learning_rate": 0.0005041197083464331,
      "loss": 2.174,
      "step": 60321
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.6903457045555115,
      "learning_rate": 0.0005041167105923483,
      "loss": 2.2947,
      "step": 60322
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.6978969573974609,
      "learning_rate": 0.0005041137128003144,
      "loss": 2.1695,
      "step": 60323
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7104021310806274,
      "learning_rate": 0.000504110714970332,
      "loss": 2.3253,
      "step": 60324
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7114619612693787,
      "learning_rate": 0.0005041077171024014,
      "loss": 2.1812,
      "step": 60325
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7368997931480408,
      "learning_rate": 0.0005041047191965233,
      "loss": 2.3856,
      "step": 60326
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7040546536445618,
      "learning_rate": 0.0005041017212526985,
      "loss": 2.1172,
      "step": 60327
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.709990382194519,
      "learning_rate": 0.0005040987232709271,
      "loss": 2.4415,
      "step": 60328
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7538233399391174,
      "learning_rate": 0.0005040957252512101,
      "loss": 2.2774,
      "step": 60329
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.6734686493873596,
      "learning_rate": 0.0005040927271935477,
      "loss": 2.2437,
      "step": 60330
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.8158912658691406,
      "learning_rate": 0.0005040897290979406,
      "loss": 2.2262,
      "step": 60331
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7758285403251648,
      "learning_rate": 0.0005040867309643894,
      "loss": 2.4438,
      "step": 60332
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7784394025802612,
      "learning_rate": 0.0005040837327928947,
      "loss": 2.231,
      "step": 60333
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.6820936799049377,
      "learning_rate": 0.000504080734583457,
      "loss": 2.2751,
      "step": 60334
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.6938294768333435,
      "learning_rate": 0.0005040777363360768,
      "loss": 1.8787,
      "step": 60335
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7316604256629944,
      "learning_rate": 0.0005040747380507547,
      "loss": 2.1014,
      "step": 60336
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.8918526768684387,
      "learning_rate": 0.0005040717397274912,
      "loss": 2.2572,
      "step": 60337
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7682321071624756,
      "learning_rate": 0.0005040687413662869,
      "loss": 2.269,
      "step": 60338
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7678156495094299,
      "learning_rate": 0.0005040657429671424,
      "loss": 2.1535,
      "step": 60339
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.8194181323051453,
      "learning_rate": 0.0005040627445300583,
      "loss": 2.0788,
      "step": 60340
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7205206751823425,
      "learning_rate": 0.000504059746055035,
      "loss": 2.1675,
      "step": 60341
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.8774766325950623,
      "learning_rate": 0.0005040567475420732,
      "loss": 2.3944,
      "step": 60342
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.9274694323539734,
      "learning_rate": 0.0005040537489911733,
      "loss": 2.527,
      "step": 60343
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.8274710774421692,
      "learning_rate": 0.000504050750402336,
      "loss": 2.2976,
      "step": 60344
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.9126286506652832,
      "learning_rate": 0.0005040477517755618,
      "loss": 2.3457,
      "step": 60345
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7165940403938293,
      "learning_rate": 0.0005040447531108513,
      "loss": 2.4131,
      "step": 60346
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.758424699306488,
      "learning_rate": 0.0005040417544082051,
      "loss": 2.2018,
      "step": 60347
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.794508159160614,
      "learning_rate": 0.0005040387556676236,
      "loss": 2.282,
      "step": 60348
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7624340057373047,
      "learning_rate": 0.0005040357568891074,
      "loss": 2.1717,
      "step": 60349
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7958792448043823,
      "learning_rate": 0.0005040327580726572,
      "loss": 2.3677,
      "step": 60350
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.8286382555961609,
      "learning_rate": 0.0005040297592182733,
      "loss": 2.369,
      "step": 60351
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7530362606048584,
      "learning_rate": 0.0005040267603259567,
      "loss": 2.1912,
      "step": 60352
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.766191303730011,
      "learning_rate": 0.0005040237613957075,
      "loss": 2.14,
      "step": 60353
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7158088684082031,
      "learning_rate": 0.0005040207624275264,
      "loss": 2.5543,
      "step": 60354
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.8008003234863281,
      "learning_rate": 0.000504017763421414,
      "loss": 2.4312,
      "step": 60355
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.6777324676513672,
      "learning_rate": 0.0005040147643773709,
      "loss": 2.1733,
      "step": 60356
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.9106451272964478,
      "learning_rate": 0.0005040117652953975,
      "loss": 2.2457,
      "step": 60357
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.8111782073974609,
      "learning_rate": 0.0005040087661754946,
      "loss": 2.2777,
      "step": 60358
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.8050601482391357,
      "learning_rate": 0.0005040057670176626,
      "loss": 2.2302,
      "step": 60359
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7718480825424194,
      "learning_rate": 0.0005040027678219019,
      "loss": 2.0851,
      "step": 60360
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7451716065406799,
      "learning_rate": 0.0005039997685882135,
      "loss": 2.1584,
      "step": 60361
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7382791638374329,
      "learning_rate": 0.0005039967693165976,
      "loss": 2.1818,
      "step": 60362
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.8313929438591003,
      "learning_rate": 0.0005039937700070548,
      "loss": 2.5173,
      "step": 60363
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7775654196739197,
      "learning_rate": 0.0005039907706595857,
      "loss": 2.0409,
      "step": 60364
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7729576826095581,
      "learning_rate": 0.0005039877712741908,
      "loss": 2.1649,
      "step": 60365
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7661405205726624,
      "learning_rate": 0.0005039847718508708,
      "loss": 2.4847,
      "step": 60366
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7582408785820007,
      "learning_rate": 0.0005039817723896262,
      "loss": 2.3335,
      "step": 60367
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.8344789743423462,
      "learning_rate": 0.0005039787728904577,
      "loss": 2.255,
      "step": 60368
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7570599913597107,
      "learning_rate": 0.0005039757733533654,
      "loss": 2.2554,
      "step": 60369
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7633283138275146,
      "learning_rate": 0.0005039727737783503,
      "loss": 2.2172,
      "step": 60370
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.8280720710754395,
      "learning_rate": 0.0005039697741654128,
      "loss": 2.1994,
      "step": 60371
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.8217046856880188,
      "learning_rate": 0.0005039667745145535,
      "loss": 2.53,
      "step": 60372
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7452706694602966,
      "learning_rate": 0.0005039637748257728,
      "loss": 2.1464,
      "step": 60373
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7294507622718811,
      "learning_rate": 0.0005039607750990714,
      "loss": 2.392,
      "step": 60374
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7862450480461121,
      "learning_rate": 0.0005039577753344499,
      "loss": 2.3762,
      "step": 60375
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.8445736765861511,
      "learning_rate": 0.0005039547755319089,
      "loss": 2.1985,
      "step": 60376
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7072353363037109,
      "learning_rate": 0.0005039517756914488,
      "loss": 2.1654,
      "step": 60377
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7931312322616577,
      "learning_rate": 0.0005039487758130701,
      "loss": 2.1776,
      "step": 60378
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7715612649917603,
      "learning_rate": 0.0005039457758967737,
      "loss": 2.1363,
      "step": 60379
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.743737518787384,
      "learning_rate": 0.0005039427759425597,
      "loss": 2.2318,
      "step": 60380
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7685654759407043,
      "learning_rate": 0.000503939775950429,
      "loss": 2.4807,
      "step": 60381
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7425923347473145,
      "learning_rate": 0.0005039367759203821,
      "loss": 2.0827,
      "step": 60382
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.8029729127883911,
      "learning_rate": 0.0005039337758524193,
      "loss": 2.2588,
      "step": 60383
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7905066013336182,
      "learning_rate": 0.0005039307757465415,
      "loss": 2.2058,
      "step": 60384
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7050751447677612,
      "learning_rate": 0.0005039277756027492,
      "loss": 2.1708,
      "step": 60385
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7939905524253845,
      "learning_rate": 0.0005039247754210428,
      "loss": 2.2101,
      "step": 60386
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.6938579082489014,
      "learning_rate": 0.0005039217752014229,
      "loss": 2.4181,
      "step": 60387
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.839806854724884,
      "learning_rate": 0.00050391877494389,
      "loss": 2.3097,
      "step": 60388
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7009445428848267,
      "learning_rate": 0.0005039157746484449,
      "loss": 2.2635,
      "step": 60389
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.6973268985748291,
      "learning_rate": 0.000503912774315088,
      "loss": 2.2011,
      "step": 60390
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.9700899720191956,
      "learning_rate": 0.0005039097739438198,
      "loss": 2.1471,
      "step": 60391
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7410499453544617,
      "learning_rate": 0.0005039067735346409,
      "loss": 2.4792,
      "step": 60392
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7637932300567627,
      "learning_rate": 0.000503903773087552,
      "loss": 2.1695,
      "step": 60393
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7516385912895203,
      "learning_rate": 0.0005039007726025534,
      "loss": 2.2459,
      "step": 60394
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7033494710922241,
      "learning_rate": 0.000503897772079646,
      "loss": 2.222,
      "step": 60395
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7475801110267639,
      "learning_rate": 0.00050389477151883,
      "loss": 2.1013,
      "step": 60396
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.66690993309021,
      "learning_rate": 0.0005038917709201061,
      "loss": 2.5446,
      "step": 60397
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.6985200047492981,
      "learning_rate": 0.0005038887702834749,
      "loss": 2.0929,
      "step": 60398
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.6806355714797974,
      "learning_rate": 0.0005038857696089369,
      "loss": 2.1892,
      "step": 60399
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7559393644332886,
      "learning_rate": 0.0005038827688964928,
      "loss": 2.0342,
      "step": 60400
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7523563504219055,
      "learning_rate": 0.000503879768146143,
      "loss": 2.3315,
      "step": 60401
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7614699602127075,
      "learning_rate": 0.0005038767673578881,
      "loss": 2.3392,
      "step": 60402
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.832988977432251,
      "learning_rate": 0.0005038737665317286,
      "loss": 2.4389,
      "step": 60403
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7578952312469482,
      "learning_rate": 0.0005038707656676651,
      "loss": 2.3354,
      "step": 60404
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7925645709037781,
      "learning_rate": 0.0005038677647656983,
      "loss": 2.3223,
      "step": 60405
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.9257552027702332,
      "learning_rate": 0.0005038647638258286,
      "loss": 2.1587,
      "step": 60406
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7443031668663025,
      "learning_rate": 0.0005038617628480565,
      "loss": 2.3348,
      "step": 60407
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.8408371806144714,
      "learning_rate": 0.0005038587618323827,
      "loss": 1.9951,
      "step": 60408
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7141461968421936,
      "learning_rate": 0.0005038557607788078,
      "loss": 2.2668,
      "step": 60409
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7616802453994751,
      "learning_rate": 0.0005038527596873321,
      "loss": 2.2555,
      "step": 60410
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7041999697685242,
      "learning_rate": 0.0005038497585579564,
      "loss": 2.1925,
      "step": 60411
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.8406308889389038,
      "learning_rate": 0.0005038467573906813,
      "loss": 2.1098,
      "step": 60412
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7783221006393433,
      "learning_rate": 0.000503843756185507,
      "loss": 2.167,
      "step": 60413
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.8065539002418518,
      "learning_rate": 0.0005038407549424345,
      "loss": 2.3541,
      "step": 60414
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7382047772407532,
      "learning_rate": 0.000503837753661464,
      "loss": 2.1584,
      "step": 60415
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7665029168128967,
      "learning_rate": 0.0005038347523425963,
      "loss": 2.2192,
      "step": 60416
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.774615466594696,
      "learning_rate": 0.0005038317509858318,
      "loss": 2.2436,
      "step": 60417
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7251414060592651,
      "learning_rate": 0.0005038287495911714,
      "loss": 2.3531,
      "step": 60418
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7378202676773071,
      "learning_rate": 0.0005038257481586151,
      "loss": 2.2024,
      "step": 60419
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.688206672668457,
      "learning_rate": 0.0005038227466881639,
      "loss": 2.1216,
      "step": 60420
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.6999295949935913,
      "learning_rate": 0.000503819745179818,
      "loss": 1.9438,
      "step": 60421
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.0448362827301025,
      "learning_rate": 0.0005038167436335784,
      "loss": 2.0927,
      "step": 60422
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7572025060653687,
      "learning_rate": 0.0005038137420494453,
      "loss": 2.3854,
      "step": 60423
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7315405607223511,
      "learning_rate": 0.0005038107404274193,
      "loss": 2.2252,
      "step": 60424
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7928988933563232,
      "learning_rate": 0.0005038077387675013,
      "loss": 2.4446,
      "step": 60425
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7523275017738342,
      "learning_rate": 0.0005038047370696914,
      "loss": 2.2138,
      "step": 60426
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7132092118263245,
      "learning_rate": 0.0005038017353339902,
      "loss": 2.1894,
      "step": 60427
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7461308240890503,
      "learning_rate": 0.0005037987335603987,
      "loss": 2.09,
      "step": 60428
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7696218490600586,
      "learning_rate": 0.000503795731748917,
      "loss": 2.1882,
      "step": 60429
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.8484192490577698,
      "learning_rate": 0.0005037927298995459,
      "loss": 2.3669,
      "step": 60430
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7666448950767517,
      "learning_rate": 0.0005037897280122859,
      "loss": 2.2067,
      "step": 60431
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7805593609809875,
      "learning_rate": 0.0005037867260871376,
      "loss": 2.1619,
      "step": 60432
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7068013548851013,
      "learning_rate": 0.0005037837241241013,
      "loss": 2.3124,
      "step": 60433
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.6938048005104065,
      "learning_rate": 0.000503780722123178,
      "loss": 2.256,
      "step": 60434
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.8079611659049988,
      "learning_rate": 0.0005037777200843679,
      "loss": 2.2698,
      "step": 60435
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.6634791493415833,
      "learning_rate": 0.0005037747180076716,
      "loss": 2.2703,
      "step": 60436
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7232869267463684,
      "learning_rate": 0.0005037717158930898,
      "loss": 2.3032,
      "step": 60437
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7098690867424011,
      "learning_rate": 0.000503768713740623,
      "loss": 2.1994,
      "step": 60438
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7230364084243774,
      "learning_rate": 0.0005037657115502717,
      "loss": 2.2085,
      "step": 60439
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.6809927821159363,
      "learning_rate": 0.0005037627093220366,
      "loss": 2.1605,
      "step": 60440
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7992904782295227,
      "learning_rate": 0.000503759707055918,
      "loss": 2.2642,
      "step": 60441
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.756830096244812,
      "learning_rate": 0.0005037567047519168,
      "loss": 2.2125,
      "step": 60442
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7148587107658386,
      "learning_rate": 0.0005037537024100334,
      "loss": 2.2699,
      "step": 60443
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7675567865371704,
      "learning_rate": 0.0005037507000302681,
      "loss": 2.0889,
      "step": 60444
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.73394376039505,
      "learning_rate": 0.0005037476976126219,
      "loss": 2.0608,
      "step": 60445
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.8014842867851257,
      "learning_rate": 0.0005037446951570952,
      "loss": 2.1379,
      "step": 60446
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7232958674430847,
      "learning_rate": 0.0005037416926636884,
      "loss": 2.2793,
      "step": 60447
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7396416664123535,
      "learning_rate": 0.0005037386901324022,
      "loss": 2.1094,
      "step": 60448
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7164846658706665,
      "learning_rate": 0.0005037356875632372,
      "loss": 2.0974,
      "step": 60449
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7581759691238403,
      "learning_rate": 0.0005037326849561938,
      "loss": 2.1969,
      "step": 60450
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.8007728457450867,
      "learning_rate": 0.0005037296823112727,
      "loss": 2.5619,
      "step": 60451
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7970504760742188,
      "learning_rate": 0.0005037266796284743,
      "loss": 2.1914,
      "step": 60452
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7101399302482605,
      "learning_rate": 0.0005037236769077993,
      "loss": 2.2759,
      "step": 60453
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7980901598930359,
      "learning_rate": 0.0005037206741492483,
      "loss": 2.2938,
      "step": 60454
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7232425808906555,
      "learning_rate": 0.0005037176713528217,
      "loss": 2.2841,
      "step": 60455
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.6738337278366089,
      "learning_rate": 0.0005037146685185203,
      "loss": 2.1507,
      "step": 60456
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7765339016914368,
      "learning_rate": 0.0005037116656463443,
      "loss": 2.3389,
      "step": 60457
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.2630157470703125,
      "learning_rate": 0.0005037086627362946,
      "loss": 2.3113,
      "step": 60458
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7498470544815063,
      "learning_rate": 0.0005037056597883715,
      "loss": 2.2332,
      "step": 60459
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.6825416684150696,
      "learning_rate": 0.0005037026568025757,
      "loss": 2.4321,
      "step": 60460
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7182799577713013,
      "learning_rate": 0.0005036996537789079,
      "loss": 2.2372,
      "step": 60461
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7367653846740723,
      "learning_rate": 0.0005036966507173683,
      "loss": 2.2466,
      "step": 60462
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7159755825996399,
      "learning_rate": 0.0005036936476179577,
      "loss": 2.3008,
      "step": 60463
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7081071734428406,
      "learning_rate": 0.0005036906444806766,
      "loss": 2.2259,
      "step": 60464
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7032891511917114,
      "learning_rate": 0.0005036876413055256,
      "loss": 2.1872,
      "step": 60465
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.6971189379692078,
      "learning_rate": 0.0005036846380925052,
      "loss": 2.2602,
      "step": 60466
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7930548191070557,
      "learning_rate": 0.0005036816348416159,
      "loss": 2.2558,
      "step": 60467
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7487338185310364,
      "learning_rate": 0.0005036786315528584,
      "loss": 2.107,
      "step": 60468
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.822424590587616,
      "learning_rate": 0.0005036756282262332,
      "loss": 2.3143,
      "step": 60469
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.726982831954956,
      "learning_rate": 0.0005036726248617408,
      "loss": 2.4743,
      "step": 60470
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7451338768005371,
      "learning_rate": 0.0005036696214593819,
      "loss": 2.2791,
      "step": 60471
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.8872257471084595,
      "learning_rate": 0.000503666618019157,
      "loss": 2.3578,
      "step": 60472
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7638629674911499,
      "learning_rate": 0.0005036636145410666,
      "loss": 2.249,
      "step": 60473
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.8494879007339478,
      "learning_rate": 0.0005036606110251113,
      "loss": 2.2517,
      "step": 60474
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7419546246528625,
      "learning_rate": 0.0005036576074712916,
      "loss": 2.1576,
      "step": 60475
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7472876310348511,
      "learning_rate": 0.0005036546038796081,
      "loss": 2.4403,
      "step": 60476
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7348650097846985,
      "learning_rate": 0.0005036516002500614,
      "loss": 2.1929,
      "step": 60477
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7273138761520386,
      "learning_rate": 0.0005036485965826521,
      "loss": 2.1998,
      "step": 60478
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7888773083686829,
      "learning_rate": 0.0005036455928773805,
      "loss": 2.284,
      "step": 60479
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.699489414691925,
      "learning_rate": 0.0005036425891342475,
      "loss": 2.1787,
      "step": 60480
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7309343218803406,
      "learning_rate": 0.0005036395853532533,
      "loss": 2.2197,
      "step": 60481
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7564560174942017,
      "learning_rate": 0.0005036365815343989,
      "loss": 2.3845,
      "step": 60482
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7410135865211487,
      "learning_rate": 0.0005036335776776845,
      "loss": 2.1517,
      "step": 60483
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7022218108177185,
      "learning_rate": 0.0005036305737831108,
      "loss": 2.2476,
      "step": 60484
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7458201050758362,
      "learning_rate": 0.0005036275698506782,
      "loss": 2.2017,
      "step": 60485
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7529792189598083,
      "learning_rate": 0.0005036245658803875,
      "loss": 2.1886,
      "step": 60486
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7320179343223572,
      "learning_rate": 0.0005036215618722392,
      "loss": 2.3585,
      "step": 60487
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7259552478790283,
      "learning_rate": 0.0005036185578262337,
      "loss": 2.3327,
      "step": 60488
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.8860436081886292,
      "learning_rate": 0.0005036155537423718,
      "loss": 2.3629,
      "step": 60489
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7130149602890015,
      "learning_rate": 0.0005036125496206539,
      "loss": 2.1087,
      "step": 60490
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7722346782684326,
      "learning_rate": 0.0005036095454610805,
      "loss": 2.3275,
      "step": 60491
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7375025749206543,
      "learning_rate": 0.0005036065412636523,
      "loss": 2.3736,
      "step": 60492
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7862988114356995,
      "learning_rate": 0.0005036035370283697,
      "loss": 2.2188,
      "step": 60493
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7083020210266113,
      "learning_rate": 0.0005036005327552335,
      "loss": 2.1423,
      "step": 60494
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7529485821723938,
      "learning_rate": 0.000503597528444244,
      "loss": 2.132,
      "step": 60495
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7307862043380737,
      "learning_rate": 0.0005035945240954021,
      "loss": 2.2226,
      "step": 60496
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7768157720565796,
      "learning_rate": 0.0005035915197087078,
      "loss": 2.2716,
      "step": 60497
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7569321990013123,
      "learning_rate": 0.0005035885152841623,
      "loss": 1.9986,
      "step": 60498
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7415684461593628,
      "learning_rate": 0.0005035855108217656,
      "loss": 2.209,
      "step": 60499
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7489643692970276,
      "learning_rate": 0.0005035825063215186,
      "loss": 2.062,
      "step": 60500
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7432904243469238,
      "learning_rate": 0.0005035795017834219,
      "loss": 2.3529,
      "step": 60501
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7311227321624756,
      "learning_rate": 0.0005035764972074758,
      "loss": 2.2261,
      "step": 60502
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7391035556793213,
      "learning_rate": 0.0005035734925936809,
      "loss": 2.3458,
      "step": 60503
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.8402184844017029,
      "learning_rate": 0.0005035704879420381,
      "loss": 2.0175,
      "step": 60504
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7786545753479004,
      "learning_rate": 0.0005035674832525475,
      "loss": 2.2648,
      "step": 60505
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7584580779075623,
      "learning_rate": 0.00050356447852521,
      "loss": 2.1996,
      "step": 60506
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.8748514652252197,
      "learning_rate": 0.0005035614737600259,
      "loss": 2.1538,
      "step": 60507
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.8025579452514648,
      "learning_rate": 0.0005035584689569958,
      "loss": 2.2205,
      "step": 60508
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7348516583442688,
      "learning_rate": 0.0005035554641161205,
      "loss": 2.4022,
      "step": 60509
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.8066927194595337,
      "learning_rate": 0.0005035524592374004,
      "loss": 2.0744,
      "step": 60510
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7783830165863037,
      "learning_rate": 0.000503549454320836,
      "loss": 2.2457,
      "step": 60511
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.741016149520874,
      "learning_rate": 0.0005035464493664279,
      "loss": 2.2534,
      "step": 60512
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7853065729141235,
      "learning_rate": 0.0005035434443741767,
      "loss": 2.3047,
      "step": 60513
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7700362801551819,
      "learning_rate": 0.0005035404393440829,
      "loss": 2.2558,
      "step": 60514
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.8349291682243347,
      "learning_rate": 0.0005035374342761471,
      "loss": 2.1806,
      "step": 60515
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.8537567853927612,
      "learning_rate": 0.00050353442917037,
      "loss": 2.2837,
      "step": 60516
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7907723784446716,
      "learning_rate": 0.0005035314240267518,
      "loss": 2.1581,
      "step": 60517
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7686614990234375,
      "learning_rate": 0.0005035284188452933,
      "loss": 2.0803,
      "step": 60518
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7807943820953369,
      "learning_rate": 0.0005035254136259951,
      "loss": 2.3146,
      "step": 60519
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.6964395642280579,
      "learning_rate": 0.0005035224083688576,
      "loss": 2.1582,
      "step": 60520
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7050447463989258,
      "learning_rate": 0.0005035194030738815,
      "loss": 2.1337,
      "step": 60521
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.6707814335823059,
      "learning_rate": 0.0005035163977410673,
      "loss": 2.3021,
      "step": 60522
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.8135336637496948,
      "learning_rate": 0.0005035133923704156,
      "loss": 2.0985,
      "step": 60523
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.9617533683776855,
      "learning_rate": 0.0005035103869619268,
      "loss": 2.1941,
      "step": 60524
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7421008944511414,
      "learning_rate": 0.0005035073815156017,
      "loss": 2.2093,
      "step": 60525
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.853802502155304,
      "learning_rate": 0.0005035043760314407,
      "loss": 2.3751,
      "step": 60526
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.8521459698677063,
      "learning_rate": 0.0005035013705094444,
      "loss": 1.9892,
      "step": 60527
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.9840027689933777,
      "learning_rate": 0.0005034983649496133,
      "loss": 2.2325,
      "step": 60528
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.8363036513328552,
      "learning_rate": 0.000503495359351948,
      "loss": 2.0785,
      "step": 60529
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.8493204116821289,
      "learning_rate": 0.0005034923537164492,
      "loss": 2.0596,
      "step": 60530
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.8137425780296326,
      "learning_rate": 0.0005034893480431172,
      "loss": 2.2838,
      "step": 60531
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7901808619499207,
      "learning_rate": 0.0005034863423319528,
      "loss": 2.2324,
      "step": 60532
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.8876690864562988,
      "learning_rate": 0.0005034833365829563,
      "loss": 1.9477,
      "step": 60533
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.9864960312843323,
      "learning_rate": 0.0005034803307961285,
      "loss": 2.4544,
      "step": 60534
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7865484356880188,
      "learning_rate": 0.0005034773249714699,
      "loss": 2.2049,
      "step": 60535
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7529088854789734,
      "learning_rate": 0.000503474319108981,
      "loss": 2.1858,
      "step": 60536
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.8603059649467468,
      "learning_rate": 0.0005034713132086624,
      "loss": 2.2092,
      "step": 60537
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.843629777431488,
      "learning_rate": 0.0005034683072705146,
      "loss": 2.158,
      "step": 60538
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7290239334106445,
      "learning_rate": 0.0005034653012945381,
      "loss": 2.2924,
      "step": 60539
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7353906035423279,
      "learning_rate": 0.0005034622952807338,
      "loss": 2.2096,
      "step": 60540
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.928928792476654,
      "learning_rate": 0.0005034592892291018,
      "loss": 2.1556,
      "step": 60541
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.779557466506958,
      "learning_rate": 0.0005034562831396429,
      "loss": 2.2775,
      "step": 60542
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.8233475685119629,
      "learning_rate": 0.0005034532770123578,
      "loss": 2.2443,
      "step": 60543
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7372433543205261,
      "learning_rate": 0.0005034502708472468,
      "loss": 2.1608,
      "step": 60544
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7236634492874146,
      "learning_rate": 0.0005034472646443106,
      "loss": 2.2228,
      "step": 60545
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.8227894306182861,
      "learning_rate": 0.0005034442584035496,
      "loss": 2.288,
      "step": 60546
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7040551900863647,
      "learning_rate": 0.0005034412521249645,
      "loss": 2.2032,
      "step": 60547
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.862900972366333,
      "learning_rate": 0.0005034382458085558,
      "loss": 2.2365,
      "step": 60548
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7350838780403137,
      "learning_rate": 0.0005034352394543243,
      "loss": 2.3299,
      "step": 60549
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7129327058792114,
      "learning_rate": 0.0005034322330622702,
      "loss": 2.0817,
      "step": 60550
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7441055774688721,
      "learning_rate": 0.0005034292266323942,
      "loss": 2.3198,
      "step": 60551
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.6840837597846985,
      "learning_rate": 0.0005034262201646969,
      "loss": 2.1358,
      "step": 60552
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.8248922228813171,
      "learning_rate": 0.0005034232136591787,
      "loss": 2.0125,
      "step": 60553
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7258790731430054,
      "learning_rate": 0.0005034202071158404,
      "loss": 2.2453,
      "step": 60554
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7228822708129883,
      "learning_rate": 0.0005034172005346824,
      "loss": 2.2227,
      "step": 60555
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7590820789337158,
      "learning_rate": 0.0005034141939157054,
      "loss": 2.338,
      "step": 60556
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.8010618090629578,
      "learning_rate": 0.0005034111872589098,
      "loss": 2.1657,
      "step": 60557
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.737808346748352,
      "learning_rate": 0.0005034081805642962,
      "loss": 1.9825,
      "step": 60558
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.6995528936386108,
      "learning_rate": 0.0005034051738318652,
      "loss": 2.2055,
      "step": 60559
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7941458225250244,
      "learning_rate": 0.0005034021670616173,
      "loss": 2.3786,
      "step": 60560
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.9104408025741577,
      "learning_rate": 0.0005033991602535531,
      "loss": 2.2627,
      "step": 60561
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7056277990341187,
      "learning_rate": 0.0005033961534076733,
      "loss": 2.2048,
      "step": 60562
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7292957305908203,
      "learning_rate": 0.0005033931465239782,
      "loss": 2.0565,
      "step": 60563
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7191159725189209,
      "learning_rate": 0.0005033901396024684,
      "loss": 2.0882,
      "step": 60564
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7551178932189941,
      "learning_rate": 0.0005033871326431446,
      "loss": 2.2735,
      "step": 60565
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7560067772865295,
      "learning_rate": 0.0005033841256460073,
      "loss": 2.2305,
      "step": 60566
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.8063347935676575,
      "learning_rate": 0.0005033811186110571,
      "loss": 2.1728,
      "step": 60567
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.6895936131477356,
      "learning_rate": 0.0005033781115382945,
      "loss": 2.1521,
      "step": 60568
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.88607257604599,
      "learning_rate": 0.0005033751044277201,
      "loss": 2.3113,
      "step": 60569
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7802063822746277,
      "learning_rate": 0.0005033720972793343,
      "loss": 2.3444,
      "step": 60570
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.77500981092453,
      "learning_rate": 0.0005033690900931378,
      "loss": 2.6074,
      "step": 60571
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7416046261787415,
      "learning_rate": 0.0005033660828691313,
      "loss": 2.2589,
      "step": 60572
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.9262378811836243,
      "learning_rate": 0.0005033630756073151,
      "loss": 2.3693,
      "step": 60573
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7215166091918945,
      "learning_rate": 0.0005033600683076898,
      "loss": 2.2881,
      "step": 60574
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7218848466873169,
      "learning_rate": 0.0005033570609702561,
      "loss": 2.3383,
      "step": 60575
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.8287162184715271,
      "learning_rate": 0.0005033540535950145,
      "loss": 2.3728,
      "step": 60576
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.8154347538948059,
      "learning_rate": 0.0005033510461819656,
      "loss": 2.3527,
      "step": 60577
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7433775663375854,
      "learning_rate": 0.0005033480387311098,
      "loss": 2.1129,
      "step": 60578
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7595840096473694,
      "learning_rate": 0.0005033450312424479,
      "loss": 2.2166,
      "step": 60579
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7383812069892883,
      "learning_rate": 0.0005033420237159802,
      "loss": 2.1994,
      "step": 60580
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.6766597628593445,
      "learning_rate": 0.0005033390161517074,
      "loss": 2.0585,
      "step": 60581
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7138388156890869,
      "learning_rate": 0.0005033360085496299,
      "loss": 2.2646,
      "step": 60582
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7641809582710266,
      "learning_rate": 0.0005033330009097485,
      "loss": 2.1541,
      "step": 60583
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7283870577812195,
      "learning_rate": 0.0005033299932320637,
      "loss": 2.5299,
      "step": 60584
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7076907753944397,
      "learning_rate": 0.000503326985516576,
      "loss": 2.1541,
      "step": 60585
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.730372965335846,
      "learning_rate": 0.000503323977763286,
      "loss": 2.3381,
      "step": 60586
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.714387834072113,
      "learning_rate": 0.0005033209699721943,
      "loss": 2.1615,
      "step": 60587
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7514910697937012,
      "learning_rate": 0.0005033179621433012,
      "loss": 2.3289,
      "step": 60588
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7157270312309265,
      "learning_rate": 0.0005033149542766076,
      "loss": 2.1102,
      "step": 60589
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7350253462791443,
      "learning_rate": 0.0005033119463721139,
      "loss": 2.0553,
      "step": 60590
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7490023374557495,
      "learning_rate": 0.0005033089384298206,
      "loss": 2.0089,
      "step": 60591
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.8697872757911682,
      "learning_rate": 0.0005033059304497283,
      "loss": 2.142,
      "step": 60592
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7407432794570923,
      "learning_rate": 0.0005033029224318377,
      "loss": 2.2503,
      "step": 60593
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7281883955001831,
      "learning_rate": 0.0005032999143761493,
      "loss": 2.3182,
      "step": 60594
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.6900981664657593,
      "learning_rate": 0.0005032969062826635,
      "loss": 2.212,
      "step": 60595
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7152593731880188,
      "learning_rate": 0.0005032938981513809,
      "loss": 2.262,
      "step": 60596
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.746278703212738,
      "learning_rate": 0.0005032908899823023,
      "loss": 2.3891,
      "step": 60597
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7431265711784363,
      "learning_rate": 0.000503287881775428,
      "loss": 2.3086,
      "step": 60598
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.8654566407203674,
      "learning_rate": 0.0005032848735307587,
      "loss": 2.1352,
      "step": 60599
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.6967837810516357,
      "learning_rate": 0.0005032818652482947,
      "loss": 2.1378,
      "step": 60600
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.790398120880127,
      "learning_rate": 0.000503278856928037,
      "loss": 2.2086,
      "step": 60601
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7535644173622131,
      "learning_rate": 0.0005032758485699858,
      "loss": 2.4332,
      "step": 60602
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.8466601967811584,
      "learning_rate": 0.0005032728401741419,
      "loss": 2.2673,
      "step": 60603
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7443082332611084,
      "learning_rate": 0.0005032698317405057,
      "loss": 2.1897,
      "step": 60604
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7451475262641907,
      "learning_rate": 0.0005032668232690776,
      "loss": 2.2816,
      "step": 60605
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.78668612241745,
      "learning_rate": 0.0005032638147598585,
      "loss": 2.2892,
      "step": 60606
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7137070298194885,
      "learning_rate": 0.0005032608062128489,
      "loss": 2.2491,
      "step": 60607
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.8086944818496704,
      "learning_rate": 0.0005032577976280492,
      "loss": 1.9796,
      "step": 60608
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7652646899223328,
      "learning_rate": 0.0005032547890054601,
      "loss": 2.499,
      "step": 60609
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7540960907936096,
      "learning_rate": 0.0005032517803450821,
      "loss": 2.3987,
      "step": 60610
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.9366492033004761,
      "learning_rate": 0.0005032487716469156,
      "loss": 2.0602,
      "step": 60611
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.6738434433937073,
      "learning_rate": 0.0005032457629109614,
      "loss": 2.355,
      "step": 60612
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.810994565486908,
      "learning_rate": 0.0005032427541372201,
      "loss": 2.2391,
      "step": 60613
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.9548063278198242,
      "learning_rate": 0.0005032397453256921,
      "loss": 2.0306,
      "step": 60614
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7200819849967957,
      "learning_rate": 0.0005032367364763778,
      "loss": 2.2681,
      "step": 60615
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.8170416951179504,
      "learning_rate": 0.0005032337275892782,
      "loss": 2.2674,
      "step": 60616
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.9533237218856812,
      "learning_rate": 0.0005032307186643935,
      "loss": 2.3869,
      "step": 60617
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7260143160820007,
      "learning_rate": 0.0005032277097017243,
      "loss": 2.2507,
      "step": 60618
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7066555619239807,
      "learning_rate": 0.0005032247007012713,
      "loss": 2.3593,
      "step": 60619
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.8451189994812012,
      "learning_rate": 0.0005032216916630351,
      "loss": 2.3412,
      "step": 60620
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.6734654307365417,
      "learning_rate": 0.000503218682587016,
      "loss": 2.2028,
      "step": 60621
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7201346158981323,
      "learning_rate": 0.0005032156734732147,
      "loss": 2.3602,
      "step": 60622
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7660171389579773,
      "learning_rate": 0.0005032126643216319,
      "loss": 2.5258,
      "step": 60623
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.8371836543083191,
      "learning_rate": 0.000503209655132268,
      "loss": 2.1066,
      "step": 60624
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.6979026794433594,
      "learning_rate": 0.0005032066459051235,
      "loss": 2.2639,
      "step": 60625
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7872993350028992,
      "learning_rate": 0.0005032036366401992,
      "loss": 2.0573,
      "step": 60626
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.6596991419792175,
      "learning_rate": 0.0005032006273374954,
      "loss": 2.1702,
      "step": 60627
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7389376163482666,
      "learning_rate": 0.0005031976179970127,
      "loss": 2.0652,
      "step": 60628
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.8125264644622803,
      "learning_rate": 0.0005031946086187519,
      "loss": 2.244,
      "step": 60629
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7709144353866577,
      "learning_rate": 0.0005031915992027132,
      "loss": 2.3989,
      "step": 60630
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.8602899312973022,
      "learning_rate": 0.0005031885897488975,
      "loss": 2.3278,
      "step": 60631
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7275745272636414,
      "learning_rate": 0.0005031855802573052,
      "loss": 2.2141,
      "step": 60632
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7543091773986816,
      "learning_rate": 0.0005031825707279367,
      "loss": 2.4664,
      "step": 60633
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7539887428283691,
      "learning_rate": 0.0005031795611607928,
      "loss": 2.2875,
      "step": 60634
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7339572310447693,
      "learning_rate": 0.0005031765515558741,
      "loss": 2.0631,
      "step": 60635
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.6932860016822815,
      "learning_rate": 0.000503173541913181,
      "loss": 2.2203,
      "step": 60636
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.6958244442939758,
      "learning_rate": 0.0005031705322327141,
      "loss": 2.2922,
      "step": 60637
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.8187342882156372,
      "learning_rate": 0.0005031675225144739,
      "loss": 2.1578,
      "step": 60638
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7201381325721741,
      "learning_rate": 0.000503164512758461,
      "loss": 2.3106,
      "step": 60639
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.6935731172561646,
      "learning_rate": 0.000503161502964676,
      "loss": 2.1572,
      "step": 60640
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.6998162865638733,
      "learning_rate": 0.0005031584931331196,
      "loss": 2.0597,
      "step": 60641
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7056161165237427,
      "learning_rate": 0.000503155483263792,
      "loss": 2.1597,
      "step": 60642
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7371389269828796,
      "learning_rate": 0.000503152473356694,
      "loss": 2.1209,
      "step": 60643
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7301734685897827,
      "learning_rate": 0.0005031494634118263,
      "loss": 2.0847,
      "step": 60644
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7692705392837524,
      "learning_rate": 0.000503146453429189,
      "loss": 1.9547,
      "step": 60645
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.6452979445457458,
      "learning_rate": 0.0005031434434087832,
      "loss": 2.3243,
      "step": 60646
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7414889335632324,
      "learning_rate": 0.000503140433350609,
      "loss": 2.3295,
      "step": 60647
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7473855018615723,
      "learning_rate": 0.0005031374232546673,
      "loss": 2.3289,
      "step": 60648
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.6726541519165039,
      "learning_rate": 0.0005031344131209585,
      "loss": 2.584,
      "step": 60649
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.822158694267273,
      "learning_rate": 0.000503131402949483,
      "loss": 2.3744,
      "step": 60650
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7352964282035828,
      "learning_rate": 0.0005031283927402417,
      "loss": 2.1991,
      "step": 60651
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7297500371932983,
      "learning_rate": 0.000503125382493235,
      "loss": 2.4147,
      "step": 60652
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7344054579734802,
      "learning_rate": 0.0005031223722084633,
      "loss": 2.2225,
      "step": 60653
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7340133786201477,
      "learning_rate": 0.0005031193618859276,
      "loss": 2.1959,
      "step": 60654
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7585939764976501,
      "learning_rate": 0.0005031163515256279,
      "loss": 2.0628,
      "step": 60655
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.8500773906707764,
      "learning_rate": 0.0005031133411275652,
      "loss": 2.3638,
      "step": 60656
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.6995658278465271,
      "learning_rate": 0.0005031103306917399,
      "loss": 2.1424,
      "step": 60657
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7123481035232544,
      "learning_rate": 0.0005031073202181525,
      "loss": 2.4758,
      "step": 60658
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7666981220245361,
      "learning_rate": 0.0005031043097068036,
      "loss": 2.1975,
      "step": 60659
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7557364106178284,
      "learning_rate": 0.0005031012991576938,
      "loss": 2.3094,
      "step": 60660
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.8009799122810364,
      "learning_rate": 0.0005030982885708236,
      "loss": 2.0374,
      "step": 60661
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7799421548843384,
      "learning_rate": 0.0005030952779461937,
      "loss": 2.1953,
      "step": 60662
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7771035432815552,
      "learning_rate": 0.0005030922672838045,
      "loss": 2.5511,
      "step": 60663
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.836253821849823,
      "learning_rate": 0.0005030892565836565,
      "loss": 2.0965,
      "step": 60664
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.8604199886322021,
      "learning_rate": 0.0005030862458457505,
      "loss": 1.9241,
      "step": 60665
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7712870240211487,
      "learning_rate": 0.0005030832350700869,
      "loss": 2.3522,
      "step": 60666
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7447506785392761,
      "learning_rate": 0.0005030802242566662,
      "loss": 2.1706,
      "step": 60667
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7898306846618652,
      "learning_rate": 0.0005030772134054892,
      "loss": 2.1394,
      "step": 60668
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7167097330093384,
      "learning_rate": 0.0005030742025165562,
      "loss": 2.2358,
      "step": 60669
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7488348484039307,
      "learning_rate": 0.000503071191589868,
      "loss": 2.2498,
      "step": 60670
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.6698982119560242,
      "learning_rate": 0.0005030681806254251,
      "loss": 2.1752,
      "step": 60671
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7991189360618591,
      "learning_rate": 0.0005030651696232277,
      "loss": 2.2381,
      "step": 60672
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.8091967105865479,
      "learning_rate": 0.0005030621585832768,
      "loss": 2.2007,
      "step": 60673
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7690125703811646,
      "learning_rate": 0.0005030591475055729,
      "loss": 2.1759,
      "step": 60674
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.8103950619697571,
      "learning_rate": 0.0005030561363901163,
      "loss": 2.093,
      "step": 60675
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.8159937858581543,
      "learning_rate": 0.0005030531252369078,
      "loss": 2.3697,
      "step": 60676
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7418184280395508,
      "learning_rate": 0.000503050114045948,
      "loss": 2.0845,
      "step": 60677
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7571078538894653,
      "learning_rate": 0.0005030471028172373,
      "loss": 2.5263,
      "step": 60678
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.6842180490493774,
      "learning_rate": 0.0005030440915507762,
      "loss": 2.3041,
      "step": 60679
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7967110872268677,
      "learning_rate": 0.0005030410802465655,
      "loss": 2.2805,
      "step": 60680
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7075179219245911,
      "learning_rate": 0.0005030380689046056,
      "loss": 2.2121,
      "step": 60681
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.797710120677948,
      "learning_rate": 0.000503035057524897,
      "loss": 2.0776,
      "step": 60682
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7992382049560547,
      "learning_rate": 0.0005030320461074405,
      "loss": 2.2703,
      "step": 60683
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7994899153709412,
      "learning_rate": 0.0005030290346522364,
      "loss": 2.0816,
      "step": 60684
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.8591665029525757,
      "learning_rate": 0.0005030260231592854,
      "loss": 2.0514,
      "step": 60685
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.8350295424461365,
      "learning_rate": 0.0005030230116285881,
      "loss": 2.2936,
      "step": 60686
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.9706845283508301,
      "learning_rate": 0.0005030200000601448,
      "loss": 2.6425,
      "step": 60687
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.052719235420227,
      "learning_rate": 0.0005030169884539564,
      "loss": 2.2792,
      "step": 60688
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.8243412375450134,
      "learning_rate": 0.0005030139768100232,
      "loss": 2.161,
      "step": 60689
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7322753071784973,
      "learning_rate": 0.000503010965128346,
      "loss": 2.1553,
      "step": 60690
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.217215657234192,
      "learning_rate": 0.0005030079534089251,
      "loss": 1.8293,
      "step": 60691
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7147760987281799,
      "learning_rate": 0.0005030049416517614,
      "loss": 2.1045,
      "step": 60692
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.6827196478843689,
      "learning_rate": 0.0005030019298568551,
      "loss": 2.1059,
      "step": 60693
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.8145153522491455,
      "learning_rate": 0.0005029989180242068,
      "loss": 2.0747,
      "step": 60694
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.945349931716919,
      "learning_rate": 0.0005029959061538174,
      "loss": 2.0953,
      "step": 60695
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7460905313491821,
      "learning_rate": 0.0005029928942456871,
      "loss": 2.1779,
      "step": 60696
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.793666660785675,
      "learning_rate": 0.0005029898822998166,
      "loss": 2.5547,
      "step": 60697
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7786411046981812,
      "learning_rate": 0.0005029868703162064,
      "loss": 2.1437,
      "step": 60698
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.8342140316963196,
      "learning_rate": 0.0005029838582948572,
      "loss": 2.3598,
      "step": 60699
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7434448003768921,
      "learning_rate": 0.0005029808462357695,
      "loss": 2.4403,
      "step": 60700
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.8077792525291443,
      "learning_rate": 0.0005029778341389437,
      "loss": 2.3022,
      "step": 60701
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.8248921036720276,
      "learning_rate": 0.0005029748220043806,
      "loss": 2.0708,
      "step": 60702
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7205451726913452,
      "learning_rate": 0.0005029718098320806,
      "loss": 2.0991,
      "step": 60703
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.734797477722168,
      "learning_rate": 0.0005029687976220443,
      "loss": 2.2386,
      "step": 60704
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.8750219345092773,
      "learning_rate": 0.0005029657853742723,
      "loss": 2.2444,
      "step": 60705
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.8368189930915833,
      "learning_rate": 0.0005029627730887651,
      "loss": 2.1207,
      "step": 60706
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7582284212112427,
      "learning_rate": 0.0005029597607655233,
      "loss": 2.3124,
      "step": 60707
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.6555830240249634,
      "learning_rate": 0.0005029567484045475,
      "loss": 2.3785,
      "step": 60708
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.8744670152664185,
      "learning_rate": 0.0005029537360058381,
      "loss": 2.0451,
      "step": 60709
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.8360421061515808,
      "learning_rate": 0.0005029507235693958,
      "loss": 2.1651,
      "step": 60710
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7149564623832703,
      "learning_rate": 0.0005029477110952212,
      "loss": 2.2503,
      "step": 60711
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.8799405694007874,
      "learning_rate": 0.0005029446985833147,
      "loss": 2.185,
      "step": 60712
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.6944518089294434,
      "learning_rate": 0.000502941686033677,
      "loss": 2.2775,
      "step": 60713
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7097737193107605,
      "learning_rate": 0.0005029386734463087,
      "loss": 2.0831,
      "step": 60714
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.8127036690711975,
      "learning_rate": 0.0005029356608212101,
      "loss": 2.3727,
      "step": 60715
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.8138083219528198,
      "learning_rate": 0.000502932648158382,
      "loss": 2.7758,
      "step": 60716
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.710197389125824,
      "learning_rate": 0.000502929635457825,
      "loss": 2.3774,
      "step": 60717
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7641895413398743,
      "learning_rate": 0.0005029266227195395,
      "loss": 2.3508,
      "step": 60718
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7637879252433777,
      "learning_rate": 0.000502923609943526,
      "loss": 2.2834,
      "step": 60719
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7557597756385803,
      "learning_rate": 0.0005029205971297852,
      "loss": 2.2009,
      "step": 60720
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7568300366401672,
      "learning_rate": 0.0005029175842783176,
      "loss": 2.2937,
      "step": 60721
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7535940408706665,
      "learning_rate": 0.0005029145713891239,
      "loss": 2.3137,
      "step": 60722
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7992995381355286,
      "learning_rate": 0.0005029115584622045,
      "loss": 2.4009,
      "step": 60723
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.6995789408683777,
      "learning_rate": 0.0005029085454975599,
      "loss": 2.3876,
      "step": 60724
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7806625962257385,
      "learning_rate": 0.000502905532495191,
      "loss": 2.5957,
      "step": 60725
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.8653724789619446,
      "learning_rate": 0.000502902519455098,
      "loss": 2.0669,
      "step": 60726
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7185409665107727,
      "learning_rate": 0.0005028995063772816,
      "loss": 2.0896,
      "step": 60727
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7444472312927246,
      "learning_rate": 0.0005028964932617424,
      "loss": 2.1314,
      "step": 60728
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7867222428321838,
      "learning_rate": 0.0005028934801084809,
      "loss": 2.2558,
      "step": 60729
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7241924405097961,
      "learning_rate": 0.0005028904669174976,
      "loss": 2.2146,
      "step": 60730
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.6982911229133606,
      "learning_rate": 0.0005028874536887931,
      "loss": 2.4124,
      "step": 60731
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7665290832519531,
      "learning_rate": 0.0005028844404223682,
      "loss": 2.2326,
      "step": 60732
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7560562491416931,
      "learning_rate": 0.000502881427118223,
      "loss": 2.1982,
      "step": 60733
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7600435018539429,
      "learning_rate": 0.0005028784137763585,
      "loss": 2.1301,
      "step": 60734
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.8845651149749756,
      "learning_rate": 0.000502875400396775,
      "loss": 2.2088,
      "step": 60735
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7399719953536987,
      "learning_rate": 0.0005028723869794731,
      "loss": 2.1136,
      "step": 60736
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.0177022218704224,
      "learning_rate": 0.0005028693735244534,
      "loss": 2.191,
      "step": 60737
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7137740254402161,
      "learning_rate": 0.0005028663600317165,
      "loss": 1.9896,
      "step": 60738
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7233615517616272,
      "learning_rate": 0.000502863346501263,
      "loss": 2.1189,
      "step": 60739
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7879759073257446,
      "learning_rate": 0.0005028603329330933,
      "loss": 2.1571,
      "step": 60740
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.6890738010406494,
      "learning_rate": 0.000502857319327208,
      "loss": 2.4018,
      "step": 60741
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7302653193473816,
      "learning_rate": 0.0005028543056836076,
      "loss": 2.2397,
      "step": 60742
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.6844570636749268,
      "learning_rate": 0.0005028512920022929,
      "loss": 2.2068,
      "step": 60743
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7416165471076965,
      "learning_rate": 0.0005028482782832642,
      "loss": 2.1869,
      "step": 60744
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.8416244387626648,
      "learning_rate": 0.0005028452645265224,
      "loss": 2.1827,
      "step": 60745
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.6505783796310425,
      "learning_rate": 0.0005028422507320677,
      "loss": 2.0598,
      "step": 60746
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7663841843605042,
      "learning_rate": 0.0005028392368999008,
      "loss": 2.2198,
      "step": 60747
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7521601319313049,
      "learning_rate": 0.0005028362230300222,
      "loss": 2.1043,
      "step": 60748
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7376208305358887,
      "learning_rate": 0.0005028332091224325,
      "loss": 2.1542,
      "step": 60749
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.8491144180297852,
      "learning_rate": 0.0005028301951771324,
      "loss": 2.2723,
      "step": 60750
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.8783860206604004,
      "learning_rate": 0.0005028271811941222,
      "loss": 2.3092,
      "step": 60751
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7519707083702087,
      "learning_rate": 0.0005028241671734026,
      "loss": 2.0794,
      "step": 60752
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7722150087356567,
      "learning_rate": 0.0005028211531149743,
      "loss": 2.1525,
      "step": 60753
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.8239350318908691,
      "learning_rate": 0.0005028181390188377,
      "loss": 2.4671,
      "step": 60754
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.8021019697189331,
      "learning_rate": 0.0005028151248849933,
      "loss": 2.3529,
      "step": 60755
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.756963849067688,
      "learning_rate": 0.0005028121107134417,
      "loss": 2.5231,
      "step": 60756
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.8535743355751038,
      "learning_rate": 0.0005028090965041837,
      "loss": 2.171,
      "step": 60757
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7325296401977539,
      "learning_rate": 0.0005028060822572194,
      "loss": 2.2615,
      "step": 60758
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.8254557847976685,
      "learning_rate": 0.0005028030679725498,
      "loss": 2.2989,
      "step": 60759
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.8449225425720215,
      "learning_rate": 0.0005028000536501753,
      "loss": 2.3016,
      "step": 60760
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.8755479454994202,
      "learning_rate": 0.0005027970392900964,
      "loss": 2.1848,
      "step": 60761
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.773075520992279,
      "learning_rate": 0.0005027940248923137,
      "loss": 2.1568,
      "step": 60762
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.968885064125061,
      "learning_rate": 0.0005027910104568277,
      "loss": 2.2881,
      "step": 60763
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7795275449752808,
      "learning_rate": 0.0005027879959836391,
      "loss": 2.2691,
      "step": 60764
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7780358195304871,
      "learning_rate": 0.0005027849814727483,
      "loss": 2.1885,
      "step": 60765
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.9603115916252136,
      "learning_rate": 0.000502781966924156,
      "loss": 2.3199,
      "step": 60766
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.844406247138977,
      "learning_rate": 0.0005027789523378627,
      "loss": 2.3793,
      "step": 60767
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.6998540759086609,
      "learning_rate": 0.0005027759377138689,
      "loss": 2.1747,
      "step": 60768
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.8192086219787598,
      "learning_rate": 0.0005027729230521753,
      "loss": 2.3495,
      "step": 60769
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.026049256324768,
      "learning_rate": 0.0005027699083527823,
      "loss": 2.3099,
      "step": 60770
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.6692365407943726,
      "learning_rate": 0.0005027668936156906,
      "loss": 2.3875,
      "step": 60771
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.8922802805900574,
      "learning_rate": 0.0005027638788409009,
      "loss": 2.1991,
      "step": 60772
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7822314500808716,
      "learning_rate": 0.0005027608640284134,
      "loss": 2.4714,
      "step": 60773
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7536793947219849,
      "learning_rate": 0.0005027578491782288,
      "loss": 2.028,
      "step": 60774
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.6910231709480286,
      "learning_rate": 0.0005027548342903477,
      "loss": 2.027,
      "step": 60775
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.8296204209327698,
      "learning_rate": 0.0005027518193647707,
      "loss": 2.299,
      "step": 60776
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.8865516781806946,
      "learning_rate": 0.0005027488044014982,
      "loss": 1.9755,
      "step": 60777
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7342043519020081,
      "learning_rate": 0.000502745789400531,
      "loss": 2.1781,
      "step": 60778
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.9397910237312317,
      "learning_rate": 0.0005027427743618695,
      "loss": 2.4118,
      "step": 60779
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.9665610790252686,
      "learning_rate": 0.0005027397592855142,
      "loss": 2.3481,
      "step": 60780
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.715993344783783,
      "learning_rate": 0.0005027367441714658,
      "loss": 2.3719,
      "step": 60781
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.724733293056488,
      "learning_rate": 0.0005027337290197248,
      "loss": 2.2326,
      "step": 60782
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.9284487366676331,
      "learning_rate": 0.0005027307138302918,
      "loss": 2.3526,
      "step": 60783
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.8644145727157593,
      "learning_rate": 0.0005027276986031674,
      "loss": 2.2574,
      "step": 60784
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.8351554274559021,
      "learning_rate": 0.000502724683338352,
      "loss": 2.212,
      "step": 60785
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7338648438453674,
      "learning_rate": 0.0005027216680358463,
      "loss": 2.1206,
      "step": 60786
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.8302344083786011,
      "learning_rate": 0.0005027186526956508,
      "loss": 2.3505,
      "step": 60787
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.752358615398407,
      "learning_rate": 0.0005027156373177661,
      "loss": 2.4347,
      "step": 60788
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7262886762619019,
      "learning_rate": 0.0005027126219021926,
      "loss": 2.1321,
      "step": 60789
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7711077928543091,
      "learning_rate": 0.0005027096064489312,
      "loss": 2.5144,
      "step": 60790
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.311070203781128,
      "learning_rate": 0.0005027065909579821,
      "loss": 2.3142,
      "step": 60791
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.946895956993103,
      "learning_rate": 0.0005027035754293461,
      "loss": 2.2833,
      "step": 60792
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7788614630699158,
      "learning_rate": 0.0005027005598630237,
      "loss": 2.2868,
      "step": 60793
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.010144591331482,
      "learning_rate": 0.0005026975442590153,
      "loss": 2.2196,
      "step": 60794
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.878957986831665,
      "learning_rate": 0.0005026945286173217,
      "loss": 1.8652,
      "step": 60795
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7258073687553406,
      "learning_rate": 0.0005026915129379434,
      "loss": 2.1015,
      "step": 60796
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.8794150948524475,
      "learning_rate": 0.000502688497220881,
      "loss": 2.3026,
      "step": 60797
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7738751173019409,
      "learning_rate": 0.0005026854814661348,
      "loss": 2.1219,
      "step": 60798
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7095285654067993,
      "learning_rate": 0.0005026824656737056,
      "loss": 2.2777,
      "step": 60799
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.8056702613830566,
      "learning_rate": 0.0005026794498435938,
      "loss": 2.2273,
      "step": 60800
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.771925687789917,
      "learning_rate": 0.0005026764339758003,
      "loss": 2.3866,
      "step": 60801
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7879939079284668,
      "learning_rate": 0.0005026734180703252,
      "loss": 2.2216,
      "step": 60802
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7694017291069031,
      "learning_rate": 0.0005026704021271694,
      "loss": 2.1172,
      "step": 60803
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7163974642753601,
      "learning_rate": 0.0005026673861463334,
      "loss": 2.315,
      "step": 60804
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7392718195915222,
      "learning_rate": 0.0005026643701278176,
      "loss": 2.2376,
      "step": 60805
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7700018286705017,
      "learning_rate": 0.0005026613540716227,
      "loss": 2.2036,
      "step": 60806
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7224761843681335,
      "learning_rate": 0.0005026583379777492,
      "loss": 1.9886,
      "step": 60807
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.686297595500946,
      "learning_rate": 0.0005026553218461978,
      "loss": 2.3855,
      "step": 60808
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7119961977005005,
      "learning_rate": 0.0005026523056769689,
      "loss": 2.3911,
      "step": 60809
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7334182858467102,
      "learning_rate": 0.000502649289470063,
      "loss": 2.3585,
      "step": 60810
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.6605615615844727,
      "learning_rate": 0.000502646273225481,
      "loss": 2.1471,
      "step": 60811
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7090738415718079,
      "learning_rate": 0.000502643256943223,
      "loss": 2.2893,
      "step": 60812
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.928594172000885,
      "learning_rate": 0.0005026402406232899,
      "loss": 2.1436,
      "step": 60813
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.9674005508422852,
      "learning_rate": 0.0005026372242656822,
      "loss": 2.1739,
      "step": 60814
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7045906186103821,
      "learning_rate": 0.0005026342078704003,
      "loss": 2.3745,
      "step": 60815
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7589284777641296,
      "learning_rate": 0.0005026311914374449,
      "loss": 2.3767,
      "step": 60816
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7516489028930664,
      "learning_rate": 0.0005026281749668166,
      "loss": 2.3633,
      "step": 60817
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7857214212417603,
      "learning_rate": 0.0005026251584585159,
      "loss": 2.2947,
      "step": 60818
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7588110566139221,
      "learning_rate": 0.0005026221419125433,
      "loss": 2.2659,
      "step": 60819
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7400729060173035,
      "learning_rate": 0.0005026191253288994,
      "loss": 2.2238,
      "step": 60820
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7120582461357117,
      "learning_rate": 0.0005026161087075849,
      "loss": 2.2718,
      "step": 60821
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7634811997413635,
      "learning_rate": 0.0005026130920486001,
      "loss": 2.3643,
      "step": 60822
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7076317071914673,
      "learning_rate": 0.0005026100753519457,
      "loss": 2.2564,
      "step": 60823
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.6824151277542114,
      "learning_rate": 0.0005026070586176224,
      "loss": 2.1471,
      "step": 60824
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.8138980865478516,
      "learning_rate": 0.0005026040418456306,
      "loss": 2.3351,
      "step": 60825
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7728006839752197,
      "learning_rate": 0.0005026010250359708,
      "loss": 2.0911,
      "step": 60826
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7121103405952454,
      "learning_rate": 0.0005025980081886437,
      "loss": 2.399,
      "step": 60827
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7699364423751831,
      "learning_rate": 0.0005025949913036498,
      "loss": 2.4343,
      "step": 60828
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.822340190410614,
      "learning_rate": 0.0005025919743809896,
      "loss": 2.3014,
      "step": 60829
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7416843771934509,
      "learning_rate": 0.0005025889574206638,
      "loss": 2.3346,
      "step": 60830
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.8025763630867004,
      "learning_rate": 0.0005025859404226728,
      "loss": 2.245,
      "step": 60831
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7368929386138916,
      "learning_rate": 0.0005025829233870175,
      "loss": 2.0903,
      "step": 60832
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7248227000236511,
      "learning_rate": 0.0005025799063136979,
      "loss": 1.9133,
      "step": 60833
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.730850100517273,
      "learning_rate": 0.0005025768892027151,
      "loss": 2.205,
      "step": 60834
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7139419317245483,
      "learning_rate": 0.0005025738720540692,
      "loss": 2.1142,
      "step": 60835
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.6934426426887512,
      "learning_rate": 0.0005025708548677613,
      "loss": 2.2168,
      "step": 60836
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7565924525260925,
      "learning_rate": 0.0005025678376437913,
      "loss": 2.2466,
      "step": 60837
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.771205484867096,
      "learning_rate": 0.0005025648203821604,
      "loss": 2.2081,
      "step": 60838
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7478210926055908,
      "learning_rate": 0.0005025618030828688,
      "loss": 2.4945,
      "step": 60839
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.6626296639442444,
      "learning_rate": 0.000502558785745917,
      "loss": 2.168,
      "step": 60840
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.714903712272644,
      "learning_rate": 0.0005025557683713058,
      "loss": 1.9788,
      "step": 60841
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.6842871904373169,
      "learning_rate": 0.0005025527509590357,
      "loss": 2.1934,
      "step": 60842
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7646826505661011,
      "learning_rate": 0.0005025497335091071,
      "loss": 2.2462,
      "step": 60843
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.687445878982544,
      "learning_rate": 0.0005025467160215208,
      "loss": 2.1477,
      "step": 60844
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.6951044201850891,
      "learning_rate": 0.0005025436984962772,
      "loss": 2.2295,
      "step": 60845
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7498841881752014,
      "learning_rate": 0.000502540680933377,
      "loss": 2.2858,
      "step": 60846
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.8253962993621826,
      "learning_rate": 0.0005025376633328206,
      "loss": 2.2692,
      "step": 60847
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7666617631912231,
      "learning_rate": 0.0005025346456946086,
      "loss": 2.1196,
      "step": 60848
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7487517595291138,
      "learning_rate": 0.0005025316280187415,
      "loss": 2.3117,
      "step": 60849
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7178939580917358,
      "learning_rate": 0.00050252861030522,
      "loss": 2.0811,
      "step": 60850
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.8212037086486816,
      "learning_rate": 0.0005025255925540448,
      "loss": 1.9934,
      "step": 60851
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.8141077756881714,
      "learning_rate": 0.000502522574765216,
      "loss": 2.4087,
      "step": 60852
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7628560662269592,
      "learning_rate": 0.0005025195569387346,
      "loss": 2.1087,
      "step": 60853
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7455674409866333,
      "learning_rate": 0.000502516539074601,
      "loss": 2.5465,
      "step": 60854
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7751398682594299,
      "learning_rate": 0.0005025135211728156,
      "loss": 2.36,
      "step": 60855
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.8514566421508789,
      "learning_rate": 0.0005025105032333792,
      "loss": 2.1531,
      "step": 60856
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7931700348854065,
      "learning_rate": 0.0005025074852562923,
      "loss": 2.3047,
      "step": 60857
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7358691692352295,
      "learning_rate": 0.0005025044672415554,
      "loss": 1.9808,
      "step": 60858
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7797377109527588,
      "learning_rate": 0.0005025014491891691,
      "loss": 2.181,
      "step": 60859
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.8340030908584595,
      "learning_rate": 0.000502498431099134,
      "loss": 2.3048,
      "step": 60860
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7041780352592468,
      "learning_rate": 0.0005024954129714506,
      "loss": 2.0679,
      "step": 60861
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.6818124651908875,
      "learning_rate": 0.0005024923948061193,
      "loss": 2.1155,
      "step": 60862
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7904914021492004,
      "learning_rate": 0.0005024893766031412,
      "loss": 2.4313,
      "step": 60863
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.738179624080658,
      "learning_rate": 0.0005024863583625163,
      "loss": 2.077,
      "step": 60864
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.6869656443595886,
      "learning_rate": 0.0005024833400842453,
      "loss": 2.0063,
      "step": 60865
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7942671775817871,
      "learning_rate": 0.0005024803217683289,
      "loss": 2.4461,
      "step": 60866
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7569061517715454,
      "learning_rate": 0.0005024773034147676,
      "loss": 2.0071,
      "step": 60867
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.758378803730011,
      "learning_rate": 0.000502474285023562,
      "loss": 2.2714,
      "step": 60868
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.8122900724411011,
      "learning_rate": 0.0005024712665947126,
      "loss": 2.0456,
      "step": 60869
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.8937532901763916,
      "learning_rate": 0.0005024682481282198,
      "loss": 2.4031,
      "step": 60870
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.8477234840393066,
      "learning_rate": 0.0005024652296240845,
      "loss": 2.1982,
      "step": 60871
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7122983932495117,
      "learning_rate": 0.000502462211082307,
      "loss": 2.1255,
      "step": 60872
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7531716823577881,
      "learning_rate": 0.0005024591925028879,
      "loss": 2.1458,
      "step": 60873
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.8516097664833069,
      "learning_rate": 0.0005024561738858279,
      "loss": 2.0175,
      "step": 60874
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7267100811004639,
      "learning_rate": 0.0005024531552311276,
      "loss": 2.1299,
      "step": 60875
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7943752408027649,
      "learning_rate": 0.0005024501365387873,
      "loss": 2.3796,
      "step": 60876
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7113392353057861,
      "learning_rate": 0.0005024471178088077,
      "loss": 2.3212,
      "step": 60877
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.8309319019317627,
      "learning_rate": 0.0005024440990411893,
      "loss": 2.1599,
      "step": 60878
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.837228000164032,
      "learning_rate": 0.0005024410802359328,
      "loss": 2.4259,
      "step": 60879
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7829262614250183,
      "learning_rate": 0.0005024380613930387,
      "loss": 2.2317,
      "step": 60880
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7775447964668274,
      "learning_rate": 0.0005024350425125075,
      "loss": 2.3024,
      "step": 60881
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7514564394950867,
      "learning_rate": 0.0005024320235943398,
      "loss": 2.5541,
      "step": 60882
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7643457651138306,
      "learning_rate": 0.0005024290046385362,
      "loss": 2.2474,
      "step": 60883
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7242224812507629,
      "learning_rate": 0.0005024259856450973,
      "loss": 2.309,
      "step": 60884
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7529359459877014,
      "learning_rate": 0.0005024229666140234,
      "loss": 2.1641,
      "step": 60885
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.8313527703285217,
      "learning_rate": 0.0005024199475453154,
      "loss": 2.0943,
      "step": 60886
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7607598304748535,
      "learning_rate": 0.0005024169284389737,
      "loss": 2.4388,
      "step": 60887
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7580700516700745,
      "learning_rate": 0.0005024139092949987,
      "loss": 2.3057,
      "step": 60888
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.787987470626831,
      "learning_rate": 0.0005024108901133914,
      "loss": 2.0605,
      "step": 60889
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7025986313819885,
      "learning_rate": 0.0005024078708941519,
      "loss": 2.1335,
      "step": 60890
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7786177396774292,
      "learning_rate": 0.000502404851637281,
      "loss": 2.2925,
      "step": 60891
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.8250836133956909,
      "learning_rate": 0.0005024018323427793,
      "loss": 2.4856,
      "step": 60892
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.6858810782432556,
      "learning_rate": 0.0005023988130106471,
      "loss": 2.0352,
      "step": 60893
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7203291654586792,
      "learning_rate": 0.0005023957936408853,
      "loss": 2.1949,
      "step": 60894
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7267332673072815,
      "learning_rate": 0.0005023927742334942,
      "loss": 2.1864,
      "step": 60895
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7457839846611023,
      "learning_rate": 0.0005023897547884747,
      "loss": 2.1438,
      "step": 60896
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7232739925384521,
      "learning_rate": 0.0005023867353058268,
      "loss": 2.2506,
      "step": 60897
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.732620120048523,
      "learning_rate": 0.0005023837157855516,
      "loss": 2.263,
      "step": 60898
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7257995009422302,
      "learning_rate": 0.0005023806962276493,
      "loss": 2.046,
      "step": 60899
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.13041090965271,
      "learning_rate": 0.0005023776766321208,
      "loss": 2.3893,
      "step": 60900
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7167688608169556,
      "learning_rate": 0.0005023746569989663,
      "loss": 2.0827,
      "step": 60901
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7330068349838257,
      "learning_rate": 0.0005023716373281866,
      "loss": 2.1235,
      "step": 60902
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7286560535430908,
      "learning_rate": 0.0005023686176197821,
      "loss": 2.1529,
      "step": 60903
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.6730085611343384,
      "learning_rate": 0.0005023655978737536,
      "loss": 2.1265,
      "step": 60904
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7827398777008057,
      "learning_rate": 0.0005023625780901014,
      "loss": 2.1383,
      "step": 60905
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.8423800468444824,
      "learning_rate": 0.0005023595582688263,
      "loss": 2.182,
      "step": 60906
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7984448671340942,
      "learning_rate": 0.0005023565384099286,
      "loss": 2.2094,
      "step": 60907
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7926986217498779,
      "learning_rate": 0.0005023535185134091,
      "loss": 2.1183,
      "step": 60908
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.8326136469841003,
      "learning_rate": 0.0005023504985792683,
      "loss": 2.1832,
      "step": 60909
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7330127358436584,
      "learning_rate": 0.0005023474786075065,
      "loss": 2.2555,
      "step": 60910
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7491462230682373,
      "learning_rate": 0.0005023444585981246,
      "loss": 2.2435,
      "step": 60911
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.8028565645217896,
      "learning_rate": 0.0005023414385511232,
      "loss": 2.1781,
      "step": 60912
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7015014290809631,
      "learning_rate": 0.0005023384184665025,
      "loss": 2.0267,
      "step": 60913
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.236572265625,
      "learning_rate": 0.0005023353983442633,
      "loss": 2.4507,
      "step": 60914
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7066750526428223,
      "learning_rate": 0.0005023323781844063,
      "loss": 2.0869,
      "step": 60915
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7441772818565369,
      "learning_rate": 0.0005023293579869318,
      "loss": 2.2135,
      "step": 60916
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7792558073997498,
      "learning_rate": 0.0005023263377518404,
      "loss": 2.3907,
      "step": 60917
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7249305844306946,
      "learning_rate": 0.0005023233174791327,
      "loss": 2.1552,
      "step": 60918
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7764432430267334,
      "learning_rate": 0.0005023202971688093,
      "loss": 2.2476,
      "step": 60919
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.669510006904602,
      "learning_rate": 0.0005023172768208708,
      "loss": 2.2353,
      "step": 60920
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.9676660895347595,
      "learning_rate": 0.0005023142564353177,
      "loss": 2.3296,
      "step": 60921
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7164914011955261,
      "learning_rate": 0.0005023112360121504,
      "loss": 2.1868,
      "step": 60922
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.8828497529029846,
      "learning_rate": 0.0005023082155513698,
      "loss": 1.9993,
      "step": 60923
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7555551528930664,
      "learning_rate": 0.0005023051950529762,
      "loss": 2.4928,
      "step": 60924
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.688969612121582,
      "learning_rate": 0.0005023021745169703,
      "loss": 2.2084,
      "step": 60925
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7183823585510254,
      "learning_rate": 0.0005022991539433525,
      "loss": 2.1698,
      "step": 60926
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7280964255332947,
      "learning_rate": 0.0005022961333321236,
      "loss": 2.2687,
      "step": 60927
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7185966372489929,
      "learning_rate": 0.0005022931126832839,
      "loss": 2.3117,
      "step": 60928
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7338190674781799,
      "learning_rate": 0.0005022900919968342,
      "loss": 2.037,
      "step": 60929
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7835745811462402,
      "learning_rate": 0.0005022870712727749,
      "loss": 2.5442,
      "step": 60930
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7228259444236755,
      "learning_rate": 0.0005022840505111067,
      "loss": 2.2649,
      "step": 60931
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7446269989013672,
      "learning_rate": 0.0005022810297118299,
      "loss": 2.2732,
      "step": 60932
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7449617385864258,
      "learning_rate": 0.0005022780088749455,
      "loss": 2.0094,
      "step": 60933
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.745582640171051,
      "learning_rate": 0.0005022749880004535,
      "loss": 2.1724,
      "step": 60934
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.698362410068512,
      "learning_rate": 0.0005022719670883549,
      "loss": 2.0511,
      "step": 60935
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7169455289840698,
      "learning_rate": 0.0005022689461386502,
      "loss": 2.3788,
      "step": 60936
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7295153141021729,
      "learning_rate": 0.0005022659251513398,
      "loss": 2.4282,
      "step": 60937
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.6735559701919556,
      "learning_rate": 0.0005022629041264243,
      "loss": 2.2474,
      "step": 60938
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.8088440299034119,
      "learning_rate": 0.0005022598830639045,
      "loss": 2.1789,
      "step": 60939
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.8105921745300293,
      "learning_rate": 0.0005022568619637806,
      "loss": 2.2193,
      "step": 60940
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7157739996910095,
      "learning_rate": 0.0005022538408260533,
      "loss": 2.2373,
      "step": 60941
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.798021137714386,
      "learning_rate": 0.0005022508196507233,
      "loss": 1.9501,
      "step": 60942
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.6695622801780701,
      "learning_rate": 0.0005022477984377911,
      "loss": 2.046,
      "step": 60943
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7166706323623657,
      "learning_rate": 0.000502244777187257,
      "loss": 2.0919,
      "step": 60944
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7493702173233032,
      "learning_rate": 0.000502241755899122,
      "loss": 2.2057,
      "step": 60945
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.8093373775482178,
      "learning_rate": 0.0005022387345733863,
      "loss": 2.3978,
      "step": 60946
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7189367413520813,
      "learning_rate": 0.0005022357132100507,
      "loss": 2.1873,
      "step": 60947
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.8401360511779785,
      "learning_rate": 0.0005022326918091156,
      "loss": 2.213,
      "step": 60948
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.6909942030906677,
      "learning_rate": 0.0005022296703705817,
      "loss": 2.3513,
      "step": 60949
    },
    {
      "epoch": 0.79,
      "grad_norm": 3.2004427909851074,
      "learning_rate": 0.0005022266488944494,
      "loss": 2.3915,
      "step": 60950
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7232251763343811,
      "learning_rate": 0.0005022236273807194,
      "loss": 2.2907,
      "step": 60951
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7373948097229004,
      "learning_rate": 0.0005022206058293922,
      "loss": 2.166,
      "step": 60952
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7756175398826599,
      "learning_rate": 0.0005022175842404683,
      "loss": 2.2841,
      "step": 60953
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7780812382698059,
      "learning_rate": 0.0005022145626139484,
      "loss": 2.1325,
      "step": 60954
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7959698438644409,
      "learning_rate": 0.000502211540949833,
      "loss": 2.2641,
      "step": 60955
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.6880207061767578,
      "learning_rate": 0.0005022085192481227,
      "loss": 1.9385,
      "step": 60956
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7605907320976257,
      "learning_rate": 0.0005022054975088179,
      "loss": 2.2759,
      "step": 60957
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.753359317779541,
      "learning_rate": 0.0005022024757319194,
      "loss": 2.4561,
      "step": 60958
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.76142418384552,
      "learning_rate": 0.0005021994539174276,
      "loss": 2.3772,
      "step": 60959
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7772701382637024,
      "learning_rate": 0.000502196432065343,
      "loss": 2.1846,
      "step": 60960
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.8145804405212402,
      "learning_rate": 0.0005021934101756665,
      "loss": 1.9987,
      "step": 60961
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7656687498092651,
      "learning_rate": 0.0005021903882483982,
      "loss": 2.2759,
      "step": 60962
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.809873640537262,
      "learning_rate": 0.000502187366283539,
      "loss": 2.4293,
      "step": 60963
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.8861989974975586,
      "learning_rate": 0.0005021843442810893,
      "loss": 2.1407,
      "step": 60964
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.9635348916053772,
      "learning_rate": 0.0005021813222410498,
      "loss": 2.3485,
      "step": 60965
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7213444113731384,
      "learning_rate": 0.0005021783001634209,
      "loss": 1.9197,
      "step": 60966
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.6716784238815308,
      "learning_rate": 0.0005021752780482033,
      "loss": 2.234,
      "step": 60967
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7497633099555969,
      "learning_rate": 0.0005021722558953974,
      "loss": 2.1931,
      "step": 60968
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7121713757514954,
      "learning_rate": 0.0005021692337050039,
      "loss": 2.0507,
      "step": 60969
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7390319108963013,
      "learning_rate": 0.0005021662114770234,
      "loss": 2.3075,
      "step": 60970
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.8552751541137695,
      "learning_rate": 0.0005021631892114563,
      "loss": 2.1825,
      "step": 60971
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.68421870470047,
      "learning_rate": 0.0005021601669083032,
      "loss": 2.3196,
      "step": 60972
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.8295865654945374,
      "learning_rate": 0.0005021571445675649,
      "loss": 2.2415,
      "step": 60973
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7927833199501038,
      "learning_rate": 0.0005021541221892417,
      "loss": 2.169,
      "step": 60974
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.6761652827262878,
      "learning_rate": 0.0005021510997733342,
      "loss": 2.2801,
      "step": 60975
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7611945271492004,
      "learning_rate": 0.0005021480773198429,
      "loss": 2.3861,
      "step": 60976
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7673823237419128,
      "learning_rate": 0.0005021450548287685,
      "loss": 2.2906,
      "step": 60977
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7412928938865662,
      "learning_rate": 0.0005021420323001116,
      "loss": 2.2591,
      "step": 60978
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7911316156387329,
      "learning_rate": 0.0005021390097338727,
      "loss": 2.3289,
      "step": 60979
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7384447455406189,
      "learning_rate": 0.0005021359871300523,
      "loss": 2.2327,
      "step": 60980
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.8214898705482483,
      "learning_rate": 0.0005021329644886509,
      "loss": 2.2957,
      "step": 60981
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.8416452407836914,
      "learning_rate": 0.0005021299418096693,
      "loss": 2.245,
      "step": 60982
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.8498300313949585,
      "learning_rate": 0.0005021269190931079,
      "loss": 2.4405,
      "step": 60983
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.836532711982727,
      "learning_rate": 0.0005021238963389672,
      "loss": 2.269,
      "step": 60984
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.0161670446395874,
      "learning_rate": 0.000502120873547248,
      "loss": 2.1829,
      "step": 60985
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.778836190700531,
      "learning_rate": 0.0005021178507179506,
      "loss": 2.3849,
      "step": 60986
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7512369751930237,
      "learning_rate": 0.0005021148278510757,
      "loss": 2.2267,
      "step": 60987
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.8649515509605408,
      "learning_rate": 0.0005021118049466239,
      "loss": 2.2914,
      "step": 60988
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7869265675544739,
      "learning_rate": 0.0005021087820045957,
      "loss": 2.2828,
      "step": 60989
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7373871803283691,
      "learning_rate": 0.0005021057590249915,
      "loss": 2.1214,
      "step": 60990
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7970401644706726,
      "learning_rate": 0.0005021027360078121,
      "loss": 2.1431,
      "step": 60991
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.91096031665802,
      "learning_rate": 0.0005020997129530581,
      "loss": 2.1244,
      "step": 60992
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.9229754209518433,
      "learning_rate": 0.0005020966898607299,
      "loss": 2.2185,
      "step": 60993
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.8078180551528931,
      "learning_rate": 0.000502093666730828,
      "loss": 2.197,
      "step": 60994
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.0092887878417969,
      "learning_rate": 0.0005020906435633532,
      "loss": 1.9802,
      "step": 60995
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.0075048208236694,
      "learning_rate": 0.0005020876203583059,
      "loss": 2.1886,
      "step": 60996
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7977004051208496,
      "learning_rate": 0.0005020845971156867,
      "loss": 2.2143,
      "step": 60997
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.8112073540687561,
      "learning_rate": 0.0005020815738354962,
      "loss": 2.1055,
      "step": 60998
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.9000392556190491,
      "learning_rate": 0.0005020785505177348,
      "loss": 2.2942,
      "step": 60999
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7273192405700684,
      "learning_rate": 0.0005020755271624033,
      "loss": 2.1508,
      "step": 61000
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7820574641227722,
      "learning_rate": 0.000502072503769502,
      "loss": 2.3058,
      "step": 61001
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.9093548059463501,
      "learning_rate": 0.0005020694803390317,
      "loss": 2.1523,
      "step": 61002
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.791854739189148,
      "learning_rate": 0.000502066456870993,
      "loss": 2.1888,
      "step": 61003
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7644037008285522,
      "learning_rate": 0.0005020634333653862,
      "loss": 2.1221,
      "step": 61004
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.9543978571891785,
      "learning_rate": 0.000502060409822212,
      "loss": 2.2784,
      "step": 61005
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.9452674984931946,
      "learning_rate": 0.0005020573862414709,
      "loss": 2.1174,
      "step": 61006
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7574752569198608,
      "learning_rate": 0.0005020543626231636,
      "loss": 2.1959,
      "step": 61007
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.8665948510169983,
      "learning_rate": 0.0005020513389672905,
      "loss": 2.3366,
      "step": 61008
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.9580288529396057,
      "learning_rate": 0.0005020483152738523,
      "loss": 2.1045,
      "step": 61009
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7223016023635864,
      "learning_rate": 0.0005020452915428495,
      "loss": 2.1144,
      "step": 61010
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.8129125237464905,
      "learning_rate": 0.0005020422677742827,
      "loss": 2.1311,
      "step": 61011
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7481846809387207,
      "learning_rate": 0.0005020392439681525,
      "loss": 2.4653,
      "step": 61012
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.702978789806366,
      "learning_rate": 0.0005020362201244591,
      "loss": 2.1463,
      "step": 61013
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7132262587547302,
      "learning_rate": 0.0005020331962432036,
      "loss": 2.2296,
      "step": 61014
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.8778378367424011,
      "learning_rate": 0.0005020301723243862,
      "loss": 2.1094,
      "step": 61015
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7410650849342346,
      "learning_rate": 0.0005020271483680077,
      "loss": 2.0142,
      "step": 61016
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.79669588804245,
      "learning_rate": 0.0005020241243740684,
      "loss": 2.2443,
      "step": 61017
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.8609344363212585,
      "learning_rate": 0.0005020211003425692,
      "loss": 2.2666,
      "step": 61018
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.8048940300941467,
      "learning_rate": 0.0005020180762735102,
      "loss": 2.3798,
      "step": 61019
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.6773211359977722,
      "learning_rate": 0.0005020150521668924,
      "loss": 2.2373,
      "step": 61020
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7131290435791016,
      "learning_rate": 0.0005020120280227161,
      "loss": 2.2127,
      "step": 61021
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7979329228401184,
      "learning_rate": 0.0005020090038409821,
      "loss": 2.3789,
      "step": 61022
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.9702876210212708,
      "learning_rate": 0.0005020059796216907,
      "loss": 2.1253,
      "step": 61023
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7159841656684875,
      "learning_rate": 0.0005020029553648426,
      "loss": 2.0265,
      "step": 61024
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.737108588218689,
      "learning_rate": 0.0005019999310704383,
      "loss": 2.3656,
      "step": 61025
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7272344827651978,
      "learning_rate": 0.0005019969067384784,
      "loss": 2.3359,
      "step": 61026
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.8280616402626038,
      "learning_rate": 0.0005019938823689634,
      "loss": 2.3731,
      "step": 61027
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7314090728759766,
      "learning_rate": 0.0005019908579618941,
      "loss": 2.0157,
      "step": 61028
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.8301998376846313,
      "learning_rate": 0.0005019878335172708,
      "loss": 2.2191,
      "step": 61029
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7406201958656311,
      "learning_rate": 0.0005019848090350942,
      "loss": 2.396,
      "step": 61030
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7082767486572266,
      "learning_rate": 0.0005019817845153648,
      "loss": 2.2356,
      "step": 61031
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.8623444437980652,
      "learning_rate": 0.000501978759958083,
      "loss": 2.5263,
      "step": 61032
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7556468844413757,
      "learning_rate": 0.0005019757353632497,
      "loss": 2.1864,
      "step": 61033
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.6864467859268188,
      "learning_rate": 0.0005019727107308653,
      "loss": 2.0466,
      "step": 61034
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7630401849746704,
      "learning_rate": 0.0005019696860609303,
      "loss": 2.2453,
      "step": 61035
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7587522268295288,
      "learning_rate": 0.0005019666613534454,
      "loss": 2.0282,
      "step": 61036
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.6954054236412048,
      "learning_rate": 0.0005019636366084111,
      "loss": 2.222,
      "step": 61037
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7313873171806335,
      "learning_rate": 0.0005019606118258277,
      "loss": 2.1124,
      "step": 61038
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.732302725315094,
      "learning_rate": 0.0005019575870056963,
      "loss": 2.3046,
      "step": 61039
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7257407903671265,
      "learning_rate": 0.000501954562148017,
      "loss": 2.3038,
      "step": 61040
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.8059754371643066,
      "learning_rate": 0.0005019515372527906,
      "loss": 2.2264,
      "step": 61041
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.8780079483985901,
      "learning_rate": 0.0005019485123200177,
      "loss": 2.185,
      "step": 61042
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7724797129631042,
      "learning_rate": 0.0005019454873496986,
      "loss": 2.304,
      "step": 61043
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7353728413581848,
      "learning_rate": 0.0005019424623418341,
      "loss": 2.0728,
      "step": 61044
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7592545747756958,
      "learning_rate": 0.0005019394372964246,
      "loss": 2.1958,
      "step": 61045
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.9196463227272034,
      "learning_rate": 0.0005019364122134708,
      "loss": 2.2926,
      "step": 61046
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7399817109107971,
      "learning_rate": 0.0005019333870929732,
      "loss": 2.1258,
      "step": 61047
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.8318672776222229,
      "learning_rate": 0.0005019303619349325,
      "loss": 2.3824,
      "step": 61048
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.6986578106880188,
      "learning_rate": 0.000501927336739349,
      "loss": 2.1224,
      "step": 61049
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7924485802650452,
      "learning_rate": 0.0005019243115062233,
      "loss": 2.0384,
      "step": 61050
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7056272029876709,
      "learning_rate": 0.0005019212862355563,
      "loss": 2.2263,
      "step": 61051
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7563941478729248,
      "learning_rate": 0.0005019182609273482,
      "loss": 2.2059,
      "step": 61052
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7155900001525879,
      "learning_rate": 0.0005019152355815996,
      "loss": 2.288,
      "step": 61053
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.6849721670150757,
      "learning_rate": 0.0005019122101983113,
      "loss": 2.2286,
      "step": 61054
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7250590324401855,
      "learning_rate": 0.0005019091847774836,
      "loss": 2.1619,
      "step": 61055
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7611734867095947,
      "learning_rate": 0.0005019061593191172,
      "loss": 2.547,
      "step": 61056
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.8362810015678406,
      "learning_rate": 0.0005019031338232127,
      "loss": 2.107,
      "step": 61057
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7968720197677612,
      "learning_rate": 0.0005019001082897705,
      "loss": 2.2478,
      "step": 61058
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.729204535484314,
      "learning_rate": 0.0005018970827187913,
      "loss": 2.2269,
      "step": 61059
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.6685501337051392,
      "learning_rate": 0.0005018940571102756,
      "loss": 2.1717,
      "step": 61060
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.740472137928009,
      "learning_rate": 0.0005018910314642241,
      "loss": 2.3222,
      "step": 61061
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7777075171470642,
      "learning_rate": 0.0005018880057806371,
      "loss": 2.1033,
      "step": 61062
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.794852077960968,
      "learning_rate": 0.0005018849800595153,
      "loss": 2.3364,
      "step": 61063
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7857944965362549,
      "learning_rate": 0.0005018819543008594,
      "loss": 2.4069,
      "step": 61064
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7419490814208984,
      "learning_rate": 0.0005018789285046697,
      "loss": 2.1649,
      "step": 61065
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.781741201877594,
      "learning_rate": 0.000501875902670947,
      "loss": 2.3781,
      "step": 61066
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7517474293708801,
      "learning_rate": 0.0005018728767996917,
      "loss": 2.2378,
      "step": 61067
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.6933040022850037,
      "learning_rate": 0.0005018698508909045,
      "loss": 2.3956,
      "step": 61068
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7286804914474487,
      "learning_rate": 0.0005018668249445857,
      "loss": 2.2053,
      "step": 61069
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.6436952948570251,
      "learning_rate": 0.0005018637989607362,
      "loss": 2.2548,
      "step": 61070
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7582740187644958,
      "learning_rate": 0.0005018607729393564,
      "loss": 2.221,
      "step": 61071
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7717229723930359,
      "learning_rate": 0.0005018577468804469,
      "loss": 2.5351,
      "step": 61072
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7561790943145752,
      "learning_rate": 0.0005018547207840081,
      "loss": 2.0608,
      "step": 61073
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7043563723564148,
      "learning_rate": 0.0005018516946500408,
      "loss": 2.1595,
      "step": 61074
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.6830381751060486,
      "learning_rate": 0.0005018486684785455,
      "loss": 2.0979,
      "step": 61075
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.8137121200561523,
      "learning_rate": 0.0005018456422695226,
      "loss": 2.2928,
      "step": 61076
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7457997798919678,
      "learning_rate": 0.0005018426160229729,
      "loss": 2.2742,
      "step": 61077
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7926440834999084,
      "learning_rate": 0.0005018395897388968,
      "loss": 2.4603,
      "step": 61078
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.8380687832832336,
      "learning_rate": 0.0005018365634172949,
      "loss": 2.244,
      "step": 61079
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7775833606719971,
      "learning_rate": 0.0005018335370581678,
      "loss": 2.2493,
      "step": 61080
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7552513480186462,
      "learning_rate": 0.0005018305106615161,
      "loss": 2.3548,
      "step": 61081
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7335495948791504,
      "learning_rate": 0.0005018274842273401,
      "loss": 2.0835,
      "step": 61082
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.902748167514801,
      "learning_rate": 0.0005018244577556407,
      "loss": 2.1389,
      "step": 61083
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7566139101982117,
      "learning_rate": 0.0005018214312464183,
      "loss": 2.4447,
      "step": 61084
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7569339871406555,
      "learning_rate": 0.0005018184046996735,
      "loss": 2.2421,
      "step": 61085
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7313897609710693,
      "learning_rate": 0.0005018153781154069,
      "loss": 2.0225,
      "step": 61086
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7662634253501892,
      "learning_rate": 0.000501812351493619,
      "loss": 2.3803,
      "step": 61087
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7252377867698669,
      "learning_rate": 0.0005018093248343101,
      "loss": 2.1917,
      "step": 61088
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.9474356770515442,
      "learning_rate": 0.0005018062981374814,
      "loss": 2.2415,
      "step": 61089
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.0064175128936768,
      "learning_rate": 0.0005018032714031329,
      "loss": 2.4159,
      "step": 61090
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7727377414703369,
      "learning_rate": 0.0005018002446312654,
      "loss": 2.3039,
      "step": 61091
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7801794409751892,
      "learning_rate": 0.0005017972178218796,
      "loss": 2.2701,
      "step": 61092
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.891548752784729,
      "learning_rate": 0.0005017941909749756,
      "loss": 2.1256,
      "step": 61093
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.780154287815094,
      "learning_rate": 0.0005017911640905544,
      "loss": 2.0491,
      "step": 61094
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.8571950793266296,
      "learning_rate": 0.0005017881371686164,
      "loss": 2.3102,
      "step": 61095
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7602981925010681,
      "learning_rate": 0.0005017851102091623,
      "loss": 2.1528,
      "step": 61096
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7754215002059937,
      "learning_rate": 0.0005017820832121923,
      "loss": 2.1592,
      "step": 61097
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.8110923171043396,
      "learning_rate": 0.0005017790561777073,
      "loss": 2.1426,
      "step": 61098
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.8428376317024231,
      "learning_rate": 0.0005017760291057078,
      "loss": 2.3214,
      "step": 61099
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.8583562970161438,
      "learning_rate": 0.0005017730019961944,
      "loss": 2.2136,
      "step": 61100
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.813553512096405,
      "learning_rate": 0.0005017699748491674,
      "loss": 2.3674,
      "step": 61101
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.735724151134491,
      "learning_rate": 0.0005017669476646276,
      "loss": 2.1426,
      "step": 61102
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.0159496068954468,
      "learning_rate": 0.0005017639204425756,
      "loss": 2.3762,
      "step": 61103
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.8385043144226074,
      "learning_rate": 0.0005017608931830119,
      "loss": 2.2245,
      "step": 61104
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7157199382781982,
      "learning_rate": 0.0005017578658859369,
      "loss": 1.9995,
      "step": 61105
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7702231407165527,
      "learning_rate": 0.0005017548385513514,
      "loss": 2.1596,
      "step": 61106
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.9274448156356812,
      "learning_rate": 0.000501751811179256,
      "loss": 2.177,
      "step": 61107
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7083961367607117,
      "learning_rate": 0.0005017487837696508,
      "loss": 2.4293,
      "step": 61108
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7216614484786987,
      "learning_rate": 0.000501745756322537,
      "loss": 2.5642,
      "step": 61109
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.8005333542823792,
      "learning_rate": 0.0005017427288379147,
      "loss": 2.0461,
      "step": 61110
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.8096534609794617,
      "learning_rate": 0.0005017397013157846,
      "loss": 2.2826,
      "step": 61111
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7472108602523804,
      "learning_rate": 0.0005017366737561474,
      "loss": 2.2523,
      "step": 61112
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.9106616377830505,
      "learning_rate": 0.0005017336461590034,
      "loss": 2.3475,
      "step": 61113
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.4232735633850098,
      "learning_rate": 0.0005017306185243534,
      "loss": 2.0434,
      "step": 61114
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.8215453624725342,
      "learning_rate": 0.0005017275908521979,
      "loss": 2.2621,
      "step": 61115
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7293372750282288,
      "learning_rate": 0.0005017245631425374,
      "loss": 2.1928,
      "step": 61116
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.8565333485603333,
      "learning_rate": 0.0005017215353953725,
      "loss": 2.3632,
      "step": 61117
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.8129885792732239,
      "learning_rate": 0.0005017185076107038,
      "loss": 2.2106,
      "step": 61118
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7419282793998718,
      "learning_rate": 0.0005017154797885317,
      "loss": 2.0501,
      "step": 61119
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.701137900352478,
      "learning_rate": 0.0005017124519288569,
      "loss": 2.3225,
      "step": 61120
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7935361862182617,
      "learning_rate": 0.0005017094240316801,
      "loss": 2.3407,
      "step": 61121
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7800328731536865,
      "learning_rate": 0.0005017063960970016,
      "loss": 2.3654,
      "step": 61122
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.6889153718948364,
      "learning_rate": 0.0005017033681248221,
      "loss": 2.3289,
      "step": 61123
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7406284213066101,
      "learning_rate": 0.000501700340115142,
      "loss": 2.1058,
      "step": 61124
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7790869474411011,
      "learning_rate": 0.0005016973120679622,
      "loss": 2.3783,
      "step": 61125
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7062835693359375,
      "learning_rate": 0.000501694283983283,
      "loss": 2.0835,
      "step": 61126
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7283883094787598,
      "learning_rate": 0.0005016912558611049,
      "loss": 2.0616,
      "step": 61127
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.6900755763053894,
      "learning_rate": 0.0005016882277014288,
      "loss": 2.2165,
      "step": 61128
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7545228600502014,
      "learning_rate": 0.0005016851995042549,
      "loss": 2.3715,
      "step": 61129
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7105728983879089,
      "learning_rate": 0.000501682171269584,
      "loss": 2.5282,
      "step": 61130
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7296081185340881,
      "learning_rate": 0.0005016791429974165,
      "loss": 2.2246,
      "step": 61131
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7853602766990662,
      "learning_rate": 0.0005016761146877532,
      "loss": 2.281,
      "step": 61132
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7702562212944031,
      "learning_rate": 0.0005016730863405943,
      "loss": 2.244,
      "step": 61133
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.8716509342193604,
      "learning_rate": 0.0005016700579559407,
      "loss": 2.1158,
      "step": 61134
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7719129920005798,
      "learning_rate": 0.0005016670295337927,
      "loss": 2.0473,
      "step": 61135
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7776042819023132,
      "learning_rate": 0.000501664001074151,
      "loss": 2.591,
      "step": 61136
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.9013854265213013,
      "learning_rate": 0.0005016609725770163,
      "loss": 2.1167,
      "step": 61137
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7245091795921326,
      "learning_rate": 0.0005016579440423889,
      "loss": 2.312,
      "step": 61138
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.6854574680328369,
      "learning_rate": 0.0005016549154702695,
      "loss": 2.3455,
      "step": 61139
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7658974528312683,
      "learning_rate": 0.0005016518868606587,
      "loss": 2.4165,
      "step": 61140
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.9063041806221008,
      "learning_rate": 0.0005016488582135571,
      "loss": 2.1194,
      "step": 61141
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.8273272514343262,
      "learning_rate": 0.000501645829528965,
      "loss": 2.2046,
      "step": 61142
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7357357144355774,
      "learning_rate": 0.0005016428008068832,
      "loss": 2.3498,
      "step": 61143
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.9171888828277588,
      "learning_rate": 0.0005016397720473122,
      "loss": 2.3024,
      "step": 61144
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.8897716999053955,
      "learning_rate": 0.0005016367432502525,
      "loss": 2.2049,
      "step": 61145
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7471210360527039,
      "learning_rate": 0.0005016337144157048,
      "loss": 2.2734,
      "step": 61146
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7369418740272522,
      "learning_rate": 0.0005016306855436694,
      "loss": 2.0929,
      "step": 61147
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.9539903998374939,
      "learning_rate": 0.0005016276566341473,
      "loss": 2.2605,
      "step": 61148
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.037039875984192,
      "learning_rate": 0.0005016246276871387,
      "loss": 2.305,
      "step": 61149
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.6890963315963745,
      "learning_rate": 0.0005016215987026443,
      "loss": 2.2267,
      "step": 61150
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7431696057319641,
      "learning_rate": 0.0005016185696806646,
      "loss": 2.0412,
      "step": 61151
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.152626872062683,
      "learning_rate": 0.0005016155406212003,
      "loss": 2.2959,
      "step": 61152
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7234959006309509,
      "learning_rate": 0.0005016125115242517,
      "loss": 2.0887,
      "step": 61153
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.8109006285667419,
      "learning_rate": 0.0005016094823898196,
      "loss": 2.3172,
      "step": 61154
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.8749929070472717,
      "learning_rate": 0.0005016064532179045,
      "loss": 2.2668,
      "step": 61155
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7342535257339478,
      "learning_rate": 0.0005016034240085071,
      "loss": 2.491,
      "step": 61156
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.8018450140953064,
      "learning_rate": 0.0005016003947616276,
      "loss": 2.346,
      "step": 61157
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7244539856910706,
      "learning_rate": 0.0005015973654772668,
      "loss": 2.3983,
      "step": 61158
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7311224341392517,
      "learning_rate": 0.0005015943361554255,
      "loss": 2.1173,
      "step": 61159
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7855284214019775,
      "learning_rate": 0.0005015913067961037,
      "loss": 2.512,
      "step": 61160
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.9181697368621826,
      "learning_rate": 0.0005015882773993024,
      "loss": 2.2391,
      "step": 61161
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.8114069700241089,
      "learning_rate": 0.000501585247965022,
      "loss": 2.1378,
      "step": 61162
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7241708040237427,
      "learning_rate": 0.0005015822184932632,
      "loss": 2.3958,
      "step": 61163
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.8265880942344666,
      "learning_rate": 0.0005015791889840263,
      "loss": 2.2277,
      "step": 61164
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.8015475869178772,
      "learning_rate": 0.0005015761594373121,
      "loss": 2.5097,
      "step": 61165
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.8013456463813782,
      "learning_rate": 0.0005015731298531212,
      "loss": 2.0629,
      "step": 61166
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.8458061814308167,
      "learning_rate": 0.0005015701002314539,
      "loss": 2.3474,
      "step": 61167
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7191750407218933,
      "learning_rate": 0.000501567070572311,
      "loss": 2.1905,
      "step": 61168
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.8294870257377625,
      "learning_rate": 0.0005015640408756928,
      "loss": 2.2636,
      "step": 61169
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7429248690605164,
      "learning_rate": 0.0005015610111416002,
      "loss": 2.1216,
      "step": 61170
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.789600133895874,
      "learning_rate": 0.0005015579813700337,
      "loss": 2.0682,
      "step": 61171
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7508416771888733,
      "learning_rate": 0.0005015549515609936,
      "loss": 2.1923,
      "step": 61172
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7130371928215027,
      "learning_rate": 0.0005015519217144806,
      "loss": 2.4387,
      "step": 61173
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7219863533973694,
      "learning_rate": 0.0005015488918304954,
      "loss": 2.1656,
      "step": 61174
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7876666188240051,
      "learning_rate": 0.0005015458619090384,
      "loss": 1.9792,
      "step": 61175
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7440696954727173,
      "learning_rate": 0.0005015428319501103,
      "loss": 2.1099,
      "step": 61176
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7050330638885498,
      "learning_rate": 0.0005015398019537115,
      "loss": 2.1771,
      "step": 61177
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7373352646827698,
      "learning_rate": 0.0005015367719198427,
      "loss": 2.0788,
      "step": 61178
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7153481841087341,
      "learning_rate": 0.0005015337418485044,
      "loss": 2.3885,
      "step": 61179
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7634363770484924,
      "learning_rate": 0.0005015307117396972,
      "loss": 2.1165,
      "step": 61180
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7512783408164978,
      "learning_rate": 0.0005015276815934214,
      "loss": 2.1663,
      "step": 61181
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7191692590713501,
      "learning_rate": 0.0005015246514096781,
      "loss": 2.1882,
      "step": 61182
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7171897888183594,
      "learning_rate": 0.0005015216211884674,
      "loss": 2.4278,
      "step": 61183
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.716378927230835,
      "learning_rate": 0.0005015185909297901,
      "loss": 2.1735,
      "step": 61184
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7273446917533875,
      "learning_rate": 0.0005015155606336467,
      "loss": 2.1278,
      "step": 61185
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7431498169898987,
      "learning_rate": 0.0005015125303000377,
      "loss": 2.2872,
      "step": 61186
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7135434150695801,
      "learning_rate": 0.0005015094999289638,
      "loss": 2.1697,
      "step": 61187
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7984022498130798,
      "learning_rate": 0.0005015064695204253,
      "loss": 2.2029,
      "step": 61188
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.9244358539581299,
      "learning_rate": 0.0005015034390744231,
      "loss": 2.0987,
      "step": 61189
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.672422468662262,
      "learning_rate": 0.0005015004085909576,
      "loss": 2.1758,
      "step": 61190
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.753270149230957,
      "learning_rate": 0.0005014973780700293,
      "loss": 2.2098,
      "step": 61191
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7446466088294983,
      "learning_rate": 0.0005014943475116389,
      "loss": 2.3018,
      "step": 61192
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7278715372085571,
      "learning_rate": 0.0005014913169157869,
      "loss": 2.1883,
      "step": 61193
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.738036572933197,
      "learning_rate": 0.0005014882862824738,
      "loss": 2.2553,
      "step": 61194
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7857374548912048,
      "learning_rate": 0.0005014852556117003,
      "loss": 2.1174,
      "step": 61195
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7933785319328308,
      "learning_rate": 0.0005014822249034668,
      "loss": 2.2852,
      "step": 61196
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7653189897537231,
      "learning_rate": 0.000501479194157774,
      "loss": 2.2195,
      "step": 61197
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7459888458251953,
      "learning_rate": 0.0005014761633746225,
      "loss": 2.254,
      "step": 61198
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.8069417476654053,
      "learning_rate": 0.0005014731325540126,
      "loss": 2.1164,
      "step": 61199
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7109478116035461,
      "learning_rate": 0.0005014701016959451,
      "loss": 2.1038,
      "step": 61200
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7295677661895752,
      "learning_rate": 0.0005014670708004206,
      "loss": 2.3673,
      "step": 61201
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7981896996498108,
      "learning_rate": 0.0005014640398674395,
      "loss": 2.1182,
      "step": 61202
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7183071374893188,
      "learning_rate": 0.0005014610088970023,
      "loss": 2.0806,
      "step": 61203
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7364529371261597,
      "learning_rate": 0.0005014579778891099,
      "loss": 2.3001,
      "step": 61204
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7545143961906433,
      "learning_rate": 0.0005014549468437626,
      "loss": 2.4236,
      "step": 61205
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.8178772330284119,
      "learning_rate": 0.0005014519157609611,
      "loss": 2.1824,
      "step": 61206
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.850391685962677,
      "learning_rate": 0.0005014488846407058,
      "loss": 2.1778,
      "step": 61207
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7003644704818726,
      "learning_rate": 0.0005014458534829974,
      "loss": 2.2441,
      "step": 61208
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7336621284484863,
      "learning_rate": 0.0005014428222878363,
      "loss": 2.3242,
      "step": 61209
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.8322747349739075,
      "learning_rate": 0.0005014397910552232,
      "loss": 2.2878,
      "step": 61210
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7393831014633179,
      "learning_rate": 0.0005014367597851587,
      "loss": 2.2329,
      "step": 61211
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.9798124432563782,
      "learning_rate": 0.0005014337284776432,
      "loss": 2.1232,
      "step": 61212
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7998467683792114,
      "learning_rate": 0.0005014306971326775,
      "loss": 2.2415,
      "step": 61213
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7643362283706665,
      "learning_rate": 0.000501427665750262,
      "loss": 2.1325,
      "step": 61214
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.776603639125824,
      "learning_rate": 0.0005014246343303973,
      "loss": 2.3048,
      "step": 61215
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.8378903269767761,
      "learning_rate": 0.000501421602873084,
      "loss": 2.0303,
      "step": 61216
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7406781315803528,
      "learning_rate": 0.0005014185713783225,
      "loss": 2.3442,
      "step": 61217
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.9780874252319336,
      "learning_rate": 0.0005014155398461135,
      "loss": 2.0933,
      "step": 61218
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7454758882522583,
      "learning_rate": 0.0005014125082764577,
      "loss": 2.114,
      "step": 61219
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7468244433403015,
      "learning_rate": 0.0005014094766693553,
      "loss": 2.2475,
      "step": 61220
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.8920978903770447,
      "learning_rate": 0.0005014064450248072,
      "loss": 2.2914,
      "step": 61221
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.8119810819625854,
      "learning_rate": 0.0005014034133428137,
      "loss": 2.2805,
      "step": 61222
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7344960570335388,
      "learning_rate": 0.0005014003816233757,
      "loss": 2.485,
      "step": 61223
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7209741473197937,
      "learning_rate": 0.0005013973498664935,
      "loss": 2.3161,
      "step": 61224
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7723249793052673,
      "learning_rate": 0.0005013943180721678,
      "loss": 2.3178,
      "step": 61225
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.8605237603187561,
      "learning_rate": 0.0005013912862403989,
      "loss": 2.2998,
      "step": 61226
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7577270269393921,
      "learning_rate": 0.0005013882543711878,
      "loss": 2.2396,
      "step": 61227
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7852746248245239,
      "learning_rate": 0.0005013852224645347,
      "loss": 2.2191,
      "step": 61228
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.8933497071266174,
      "learning_rate": 0.0005013821905204402,
      "loss": 2.3163,
      "step": 61229
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7166987061500549,
      "learning_rate": 0.0005013791585389051,
      "loss": 2.3133,
      "step": 61230
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7105615735054016,
      "learning_rate": 0.0005013761265199297,
      "loss": 2.0448,
      "step": 61231
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.8403124809265137,
      "learning_rate": 0.0005013730944635148,
      "loss": 2.3254,
      "step": 61232
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.6925244927406311,
      "learning_rate": 0.0005013700623696608,
      "loss": 2.0054,
      "step": 61233
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7479897141456604,
      "learning_rate": 0.0005013670302383682,
      "loss": 2.2496,
      "step": 61234
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7599844932556152,
      "learning_rate": 0.0005013639980696377,
      "loss": 2.235,
      "step": 61235
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7237377166748047,
      "learning_rate": 0.0005013609658634699,
      "loss": 2.125,
      "step": 61236
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7491463422775269,
      "learning_rate": 0.0005013579336198652,
      "loss": 2.2373,
      "step": 61237
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7100363373756409,
      "learning_rate": 0.0005013549013388244,
      "loss": 2.1216,
      "step": 61238
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7258540987968445,
      "learning_rate": 0.0005013518690203478,
      "loss": 2.2165,
      "step": 61239
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.758848249912262,
      "learning_rate": 0.0005013488366644362,
      "loss": 2.2921,
      "step": 61240
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.916466236114502,
      "learning_rate": 0.0005013458042710899,
      "loss": 2.3214,
      "step": 61241
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7464656829833984,
      "learning_rate": 0.0005013427718403098,
      "loss": 2.269,
      "step": 61242
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7492056488990784,
      "learning_rate": 0.000501339739372096,
      "loss": 2.3942,
      "step": 61243
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.8595852255821228,
      "learning_rate": 0.0005013367068664496,
      "loss": 2.1449,
      "step": 61244
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7616532444953918,
      "learning_rate": 0.0005013336743233709,
      "loss": 2.1621,
      "step": 61245
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.8129562139511108,
      "learning_rate": 0.0005013306417428603,
      "loss": 2.1625,
      "step": 61246
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7498963475227356,
      "learning_rate": 0.0005013276091249187,
      "loss": 2.2635,
      "step": 61247
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.702575147151947,
      "learning_rate": 0.0005013245764695464,
      "loss": 2.2415,
      "step": 61248
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.8744482398033142,
      "learning_rate": 0.0005013215437767441,
      "loss": 2.191,
      "step": 61249
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7939313650131226,
      "learning_rate": 0.0005013185110465122,
      "loss": 2.3435,
      "step": 61250
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7522059679031372,
      "learning_rate": 0.0005013154782788516,
      "loss": 2.3154,
      "step": 61251
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.785983681678772,
      "learning_rate": 0.0005013124454737625,
      "loss": 2.1668,
      "step": 61252
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7738004922866821,
      "learning_rate": 0.0005013094126312457,
      "loss": 2.0431,
      "step": 61253
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.8161983489990234,
      "learning_rate": 0.0005013063797513016,
      "loss": 2.2677,
      "step": 61254
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7927491664886475,
      "learning_rate": 0.0005013033468339309,
      "loss": 2.1406,
      "step": 61255
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7311463356018066,
      "learning_rate": 0.0005013003138791342,
      "loss": 1.9855,
      "step": 61256
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7672934532165527,
      "learning_rate": 0.0005012972808869118,
      "loss": 2.159,
      "step": 61257
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.750762939453125,
      "learning_rate": 0.0005012942478572644,
      "loss": 2.2415,
      "step": 61258
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7868903279304504,
      "learning_rate": 0.0005012912147901929,
      "loss": 2.3415,
      "step": 61259
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7725332379341125,
      "learning_rate": 0.0005012881816856972,
      "loss": 2.1823,
      "step": 61260
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7042673230171204,
      "learning_rate": 0.0005012851485437784,
      "loss": 2.2768,
      "step": 61261
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.8836125731468201,
      "learning_rate": 0.000501282115364437,
      "loss": 2.3461,
      "step": 61262
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7566598057746887,
      "learning_rate": 0.0005012790821476733,
      "loss": 2.333,
      "step": 61263
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7747426629066467,
      "learning_rate": 0.000501276048893488,
      "loss": 2.2443,
      "step": 61264
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.6822537183761597,
      "learning_rate": 0.0005012730156018818,
      "loss": 2.1507,
      "step": 61265
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7148813009262085,
      "learning_rate": 0.0005012699822728551,
      "loss": 2.2912,
      "step": 61266
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.8318043351173401,
      "learning_rate": 0.0005012669489064084,
      "loss": 2.1317,
      "step": 61267
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7678782939910889,
      "learning_rate": 0.0005012639155025425,
      "loss": 2.2646,
      "step": 61268
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7683834433555603,
      "learning_rate": 0.0005012608820612579,
      "loss": 2.2773,
      "step": 61269
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7741907238960266,
      "learning_rate": 0.000501257848582555,
      "loss": 2.3099,
      "step": 61270
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7339733242988586,
      "learning_rate": 0.0005012548150664345,
      "loss": 2.371,
      "step": 61271
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7212355732917786,
      "learning_rate": 0.0005012517815128969,
      "loss": 2.0716,
      "step": 61272
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7640672326087952,
      "learning_rate": 0.0005012487479219428,
      "loss": 2.3349,
      "step": 61273
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.6967657804489136,
      "learning_rate": 0.0005012457142935727,
      "loss": 2.0489,
      "step": 61274
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.8300361633300781,
      "learning_rate": 0.0005012426806277874,
      "loss": 2.2282,
      "step": 61275
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7663607001304626,
      "learning_rate": 0.0005012396469245871,
      "loss": 2.2831,
      "step": 61276
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.8423701524734497,
      "learning_rate": 0.0005012366131839726,
      "loss": 2.3617,
      "step": 61277
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7530449032783508,
      "learning_rate": 0.0005012335794059446,
      "loss": 2.422,
      "step": 61278
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.8445799350738525,
      "learning_rate": 0.0005012305455905033,
      "loss": 2.4109,
      "step": 61279
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7122852206230164,
      "learning_rate": 0.0005012275117376495,
      "loss": 2.2571,
      "step": 61280
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7214052677154541,
      "learning_rate": 0.0005012244778473836,
      "loss": 2.2232,
      "step": 61281
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7342347502708435,
      "learning_rate": 0.0005012214439197064,
      "loss": 2.2197,
      "step": 61282
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.6884045004844666,
      "learning_rate": 0.0005012184099546183,
      "loss": 2.0212,
      "step": 61283
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.8676577806472778,
      "learning_rate": 0.0005012153759521199,
      "loss": 2.1894,
      "step": 61284
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7335355877876282,
      "learning_rate": 0.0005012123419122116,
      "loss": 2.1699,
      "step": 61285
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7139450907707214,
      "learning_rate": 0.0005012093078348944,
      "loss": 2.407,
      "step": 61286
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7507641911506653,
      "learning_rate": 0.0005012062737201683,
      "loss": 1.9577,
      "step": 61287
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.8617536425590515,
      "learning_rate": 0.0005012032395680344,
      "loss": 2.2596,
      "step": 61288
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7998720407485962,
      "learning_rate": 0.0005012002053784929,
      "loss": 2.074,
      "step": 61289
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.8675350546836853,
      "learning_rate": 0.0005011971711515445,
      "loss": 2.2241,
      "step": 61290
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7931755781173706,
      "learning_rate": 0.0005011941368871899,
      "loss": 2.2931,
      "step": 61291
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.750581681728363,
      "learning_rate": 0.0005011911025854293,
      "loss": 2.2359,
      "step": 61292
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.8937063813209534,
      "learning_rate": 0.0005011880682462636,
      "loss": 2.1073,
      "step": 61293
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.9617228507995605,
      "learning_rate": 0.0005011850338696932,
      "loss": 2.1325,
      "step": 61294
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7550077438354492,
      "learning_rate": 0.0005011819994557187,
      "loss": 2.2503,
      "step": 61295
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.6750480532646179,
      "learning_rate": 0.0005011789650043407,
      "loss": 2.1522,
      "step": 61296
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.0497421026229858,
      "learning_rate": 0.0005011759305155597,
      "loss": 2.2524,
      "step": 61297
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7701600193977356,
      "learning_rate": 0.0005011728959893763,
      "loss": 2.2679,
      "step": 61298
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.6864549517631531,
      "learning_rate": 0.000501169861425791,
      "loss": 2.0575,
      "step": 61299
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.8186736702919006,
      "learning_rate": 0.0005011668268248046,
      "loss": 2.1852,
      "step": 61300
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.8638421893119812,
      "learning_rate": 0.0005011637921864174,
      "loss": 1.874,
      "step": 61301
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.006050705909729,
      "learning_rate": 0.00050116075751063,
      "loss": 2.2137,
      "step": 61302
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7123441696166992,
      "learning_rate": 0.000501157722797443,
      "loss": 2.218,
      "step": 61303
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.9080200791358948,
      "learning_rate": 0.0005011546880468571,
      "loss": 2.1482,
      "step": 61304
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.8645465970039368,
      "learning_rate": 0.0005011516532588726,
      "loss": 2.1474,
      "step": 61305
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7201589345932007,
      "learning_rate": 0.0005011486184334903,
      "loss": 2.1848,
      "step": 61306
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.8164429068565369,
      "learning_rate": 0.0005011455835707107,
      "loss": 2.2174,
      "step": 61307
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.8367486596107483,
      "learning_rate": 0.0005011425486705342,
      "loss": 2.1884,
      "step": 61308
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7339040040969849,
      "learning_rate": 0.0005011395137329617,
      "loss": 2.3132,
      "step": 61309
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.6926211714744568,
      "learning_rate": 0.0005011364787579934,
      "loss": 2.2966,
      "step": 61310
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.6690940856933594,
      "learning_rate": 0.0005011334437456302,
      "loss": 2.3074,
      "step": 61311
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7968543767929077,
      "learning_rate": 0.0005011304086958723,
      "loss": 2.168,
      "step": 61312
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7832287549972534,
      "learning_rate": 0.0005011273736087205,
      "loss": 2.2296,
      "step": 61313
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7979906797409058,
      "learning_rate": 0.0005011243384841756,
      "loss": 2.1537,
      "step": 61314
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.8423689007759094,
      "learning_rate": 0.0005011213033222376,
      "loss": 2.2312,
      "step": 61315
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.8723219633102417,
      "learning_rate": 0.0005011182681229074,
      "loss": 2.4692,
      "step": 61316
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.8493943810462952,
      "learning_rate": 0.0005011152328861856,
      "loss": 2.2834,
      "step": 61317
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7165966033935547,
      "learning_rate": 0.0005011121976120726,
      "loss": 2.1258,
      "step": 61318
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.9114097356796265,
      "learning_rate": 0.000501109162300569,
      "loss": 2.3249,
      "step": 61319
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7761756181716919,
      "learning_rate": 0.0005011061269516755,
      "loss": 2.092,
      "step": 61320
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.6885972023010254,
      "learning_rate": 0.0005011030915653927,
      "loss": 2.2537,
      "step": 61321
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7807157635688782,
      "learning_rate": 0.0005011000561417208,
      "loss": 2.2076,
      "step": 61322
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.734090268611908,
      "learning_rate": 0.0005010970206806608,
      "loss": 2.3044,
      "step": 61323
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.824953556060791,
      "learning_rate": 0.0005010939851822129,
      "loss": 2.5026,
      "step": 61324
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7360444664955139,
      "learning_rate": 0.000501090949646378,
      "loss": 1.9369,
      "step": 61325
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7865224480628967,
      "learning_rate": 0.0005010879140731564,
      "loss": 2.2196,
      "step": 61326
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7341462969779968,
      "learning_rate": 0.0005010848784625486,
      "loss": 2.3117,
      "step": 61327
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7133013010025024,
      "learning_rate": 0.0005010818428145556,
      "loss": 2.0407,
      "step": 61328
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7796359658241272,
      "learning_rate": 0.0005010788071291776,
      "loss": 2.2249,
      "step": 61329
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.8763667941093445,
      "learning_rate": 0.0005010757714064152,
      "loss": 2.303,
      "step": 61330
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.6965779662132263,
      "learning_rate": 0.000501072735646269,
      "loss": 2.2213,
      "step": 61331
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7823144793510437,
      "learning_rate": 0.0005010696998487397,
      "loss": 2.1761,
      "step": 61332
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7322795987129211,
      "learning_rate": 0.0005010666640138276,
      "loss": 2.2504,
      "step": 61333
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7571940422058105,
      "learning_rate": 0.0005010636281415335,
      "loss": 2.4057,
      "step": 61334
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7562241554260254,
      "learning_rate": 0.0005010605922318579,
      "loss": 2.2382,
      "step": 61335
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7449294924736023,
      "learning_rate": 0.0005010575562848014,
      "loss": 2.2145,
      "step": 61336
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7607966661453247,
      "learning_rate": 0.0005010545203003643,
      "loss": 2.1266,
      "step": 61337
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.6905224919319153,
      "learning_rate": 0.0005010514842785476,
      "loss": 2.3232,
      "step": 61338
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.744306743144989,
      "learning_rate": 0.0005010484482193514,
      "loss": 2.3372,
      "step": 61339
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.873551070690155,
      "learning_rate": 0.0005010454121227766,
      "loss": 2.0933,
      "step": 61340
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7728536128997803,
      "learning_rate": 0.0005010423759888238,
      "loss": 2.2878,
      "step": 61341
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.831812858581543,
      "learning_rate": 0.0005010393398174932,
      "loss": 2.2908,
      "step": 61342
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.8559504747390747,
      "learning_rate": 0.0005010363036087856,
      "loss": 2.1852,
      "step": 61343
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7270733118057251,
      "learning_rate": 0.0005010332673627018,
      "loss": 2.1682,
      "step": 61344
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7066840529441833,
      "learning_rate": 0.000501030231079242,
      "loss": 2.0321,
      "step": 61345
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.8402688503265381,
      "learning_rate": 0.0005010271947584068,
      "loss": 2.0745,
      "step": 61346
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7234818935394287,
      "learning_rate": 0.0005010241584001969,
      "loss": 2.2785,
      "step": 61347
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7233043909072876,
      "learning_rate": 0.0005010211220046128,
      "loss": 2.3998,
      "step": 61348
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7199920415878296,
      "learning_rate": 0.0005010180855716551,
      "loss": 2.4012,
      "step": 61349
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.6983907222747803,
      "learning_rate": 0.0005010150491013244,
      "loss": 2.3335,
      "step": 61350
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7972341179847717,
      "learning_rate": 0.0005010120125936211,
      "loss": 2.3534,
      "step": 61351
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7462424039840698,
      "learning_rate": 0.000501008976048546,
      "loss": 2.3445,
      "step": 61352
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7210143804550171,
      "learning_rate": 0.0005010059394660994,
      "loss": 2.2455,
      "step": 61353
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7121041417121887,
      "learning_rate": 0.0005010029028462821,
      "loss": 2.1621,
      "step": 61354
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7640905976295471,
      "learning_rate": 0.0005009998661890945,
      "loss": 2.3304,
      "step": 61355
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.726887047290802,
      "learning_rate": 0.0005009968294945373,
      "loss": 2.4149,
      "step": 61356
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7413915395736694,
      "learning_rate": 0.0005009937927626109,
      "loss": 2.1813,
      "step": 61357
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7230724096298218,
      "learning_rate": 0.0005009907559933161,
      "loss": 2.2721,
      "step": 61358
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7822175621986389,
      "learning_rate": 0.0005009877191866531,
      "loss": 2.078,
      "step": 61359
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.72804856300354,
      "learning_rate": 0.0005009846823426229,
      "loss": 2.2852,
      "step": 61360
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7514206767082214,
      "learning_rate": 0.0005009816454612258,
      "loss": 2.2095,
      "step": 61361
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.9020030498504639,
      "learning_rate": 0.0005009786085424623,
      "loss": 2.2646,
      "step": 61362
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.8308159112930298,
      "learning_rate": 0.0005009755715863334,
      "loss": 2.0893,
      "step": 61363
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.6695312261581421,
      "learning_rate": 0.000500972534592839,
      "loss": 2.0881,
      "step": 61364
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7966017723083496,
      "learning_rate": 0.0005009694975619801,
      "loss": 2.2138,
      "step": 61365
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.8053319454193115,
      "learning_rate": 0.0005009664604937573,
      "loss": 2.1661,
      "step": 61366
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.8174833655357361,
      "learning_rate": 0.000500963423388171,
      "loss": 2.2234,
      "step": 61367
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7608221769332886,
      "learning_rate": 0.0005009603862452217,
      "loss": 2.2288,
      "step": 61368
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7248880863189697,
      "learning_rate": 0.0005009573490649101,
      "loss": 2.1485,
      "step": 61369
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.9053850173950195,
      "learning_rate": 0.0005009543118472369,
      "loss": 2.3043,
      "step": 61370
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7839622497558594,
      "learning_rate": 0.0005009512745922023,
      "loss": 2.237,
      "step": 61371
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7597814202308655,
      "learning_rate": 0.0005009482372998072,
      "loss": 2.1702,
      "step": 61372
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.6593656539916992,
      "learning_rate": 0.000500945199970052,
      "loss": 2.2198,
      "step": 61373
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.8980944752693176,
      "learning_rate": 0.0005009421626029372,
      "loss": 2.1713,
      "step": 61374
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7473605275154114,
      "learning_rate": 0.0005009391251984635,
      "loss": 2.2023,
      "step": 61375
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.8281975984573364,
      "learning_rate": 0.0005009360877566315,
      "loss": 2.1317,
      "step": 61376
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.8193907141685486,
      "learning_rate": 0.0005009330502774416,
      "loss": 2.3327,
      "step": 61377
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.9086251258850098,
      "learning_rate": 0.0005009300127608945,
      "loss": 2.2517,
      "step": 61378
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7258973717689514,
      "learning_rate": 0.0005009269752069906,
      "loss": 2.0564,
      "step": 61379
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.8130362033843994,
      "learning_rate": 0.0005009239376157307,
      "loss": 2.093,
      "step": 61380
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.8333181142807007,
      "learning_rate": 0.0005009208999871153,
      "loss": 2.131,
      "step": 61381
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7349312901496887,
      "learning_rate": 0.0005009178623211449,
      "loss": 2.4498,
      "step": 61382
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7583034634590149,
      "learning_rate": 0.00050091482461782,
      "loss": 2.2617,
      "step": 61383
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.0024304389953613,
      "learning_rate": 0.0005009117868771413,
      "loss": 2.1085,
      "step": 61384
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.8229855298995972,
      "learning_rate": 0.0005009087490991092,
      "loss": 2.3223,
      "step": 61385
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7144412398338318,
      "learning_rate": 0.0005009057112837245,
      "loss": 2.1735,
      "step": 61386
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.8065631985664368,
      "learning_rate": 0.0005009026734309876,
      "loss": 2.3612,
      "step": 61387
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.9516305327415466,
      "learning_rate": 0.0005008996355408991,
      "loss": 2.4839,
      "step": 61388
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7268801927566528,
      "learning_rate": 0.0005008965976134594,
      "loss": 2.326,
      "step": 61389
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7368062734603882,
      "learning_rate": 0.0005008935596486694,
      "loss": 2.3634,
      "step": 61390
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.8195306062698364,
      "learning_rate": 0.0005008905216465295,
      "loss": 2.2873,
      "step": 61391
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.0774062871932983,
      "learning_rate": 0.0005008874836070403,
      "loss": 2.1965,
      "step": 61392
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.8048248291015625,
      "learning_rate": 0.0005008844455302022,
      "loss": 2.3611,
      "step": 61393
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7962604761123657,
      "learning_rate": 0.000500881407416016,
      "loss": 2.2812,
      "step": 61394
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.0730154514312744,
      "learning_rate": 0.000500878369264482,
      "loss": 2.2799,
      "step": 61395
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.137580156326294,
      "learning_rate": 0.0005008753310756011,
      "loss": 2.4644,
      "step": 61396
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.8114743232727051,
      "learning_rate": 0.0005008722928493736,
      "loss": 2.2122,
      "step": 61397
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.9982045292854309,
      "learning_rate": 0.0005008692545858001,
      "loss": 2.0369,
      "step": 61398
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.8595483303070068,
      "learning_rate": 0.0005008662162848814,
      "loss": 2.4696,
      "step": 61399
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.8052760362625122,
      "learning_rate": 0.0005008631779466177,
      "loss": 2.1586,
      "step": 61400
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7345128655433655,
      "learning_rate": 0.0005008601395710097,
      "loss": 2.4215,
      "step": 61401
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.0594769716262817,
      "learning_rate": 0.0005008571011580581,
      "loss": 2.0043,
      "step": 61402
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7338704466819763,
      "learning_rate": 0.0005008540627077635,
      "loss": 2.2381,
      "step": 61403
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7390030026435852,
      "learning_rate": 0.0005008510242201261,
      "loss": 2.1803,
      "step": 61404
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7674877643585205,
      "learning_rate": 0.0005008479856951468,
      "loss": 2.4679,
      "step": 61405
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.8603705167770386,
      "learning_rate": 0.0005008449471328261,
      "loss": 2.2368,
      "step": 61406
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.8329586982727051,
      "learning_rate": 0.0005008419085331645,
      "loss": 2.2838,
      "step": 61407
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7440899014472961,
      "learning_rate": 0.0005008388698961627,
      "loss": 2.1037,
      "step": 61408
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.8802127838134766,
      "learning_rate": 0.000500835831221821,
      "loss": 2.2656,
      "step": 61409
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7579362392425537,
      "learning_rate": 0.0005008327925101402,
      "loss": 2.3288,
      "step": 61410
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7129632234573364,
      "learning_rate": 0.0005008297537611208,
      "loss": 2.2765,
      "step": 61411
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7603059411048889,
      "learning_rate": 0.0005008267149747633,
      "loss": 2.3124,
      "step": 61412
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.8262071013450623,
      "learning_rate": 0.0005008236761510685,
      "loss": 2.1374,
      "step": 61413
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.9452314972877502,
      "learning_rate": 0.0005008206372900365,
      "loss": 1.9897,
      "step": 61414
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.6984689831733704,
      "learning_rate": 0.0005008175983916684,
      "loss": 2.1466,
      "step": 61415
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.8016387224197388,
      "learning_rate": 0.0005008145594559644,
      "loss": 2.2046,
      "step": 61416
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.8459173440933228,
      "learning_rate": 0.0005008115204829253,
      "loss": 2.2632,
      "step": 61417
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7454399466514587,
      "learning_rate": 0.0005008084814725514,
      "loss": 2.2811,
      "step": 61418
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.751436710357666,
      "learning_rate": 0.0005008054424248434,
      "loss": 2.0346,
      "step": 61419
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7199814319610596,
      "learning_rate": 0.000500802403339802,
      "loss": 2.1335,
      "step": 61420
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7791460156440735,
      "learning_rate": 0.0005007993642174276,
      "loss": 2.0888,
      "step": 61421
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.8093631863594055,
      "learning_rate": 0.0005007963250577208,
      "loss": 2.3185,
      "step": 61422
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7281295657157898,
      "learning_rate": 0.0005007932858606822,
      "loss": 2.2747,
      "step": 61423
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7406405210494995,
      "learning_rate": 0.0005007902466263123,
      "loss": 2.1072,
      "step": 61424
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7349360585212708,
      "learning_rate": 0.0005007872073546117,
      "loss": 2.198,
      "step": 61425
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.8254271745681763,
      "learning_rate": 0.000500784168045581,
      "loss": 2.0467,
      "step": 61426
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.6818459630012512,
      "learning_rate": 0.0005007811286992207,
      "loss": 2.2231,
      "step": 61427
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.8325144648551941,
      "learning_rate": 0.0005007780893155314,
      "loss": 2.2796,
      "step": 61428
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.8052466511726379,
      "learning_rate": 0.0005007750498945137,
      "loss": 2.3278,
      "step": 61429
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.8155681490898132,
      "learning_rate": 0.0005007720104361681,
      "loss": 2.0073,
      "step": 61430
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7056958079338074,
      "learning_rate": 0.0005007689709404952,
      "loss": 2.3254,
      "step": 61431
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7612179517745972,
      "learning_rate": 0.0005007659314074956,
      "loss": 2.1983,
      "step": 61432
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.8414683938026428,
      "learning_rate": 0.0005007628918371698,
      "loss": 2.2373,
      "step": 61433
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.875959575176239,
      "learning_rate": 0.0005007598522295183,
      "loss": 2.3128,
      "step": 61434
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7107648253440857,
      "learning_rate": 0.0005007568125845419,
      "loss": 2.3321,
      "step": 61435
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.723192036151886,
      "learning_rate": 0.000500753772902241,
      "loss": 2.2248,
      "step": 61436
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.6950470209121704,
      "learning_rate": 0.0005007507331826161,
      "loss": 2.3458,
      "step": 61437
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7295180559158325,
      "learning_rate": 0.0005007476934256679,
      "loss": 2.1423,
      "step": 61438
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7016606330871582,
      "learning_rate": 0.0005007446536313969,
      "loss": 2.2895,
      "step": 61439
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7698294520378113,
      "learning_rate": 0.0005007416137998036,
      "loss": 2.3828,
      "step": 61440
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7559678554534912,
      "learning_rate": 0.0005007385739308887,
      "loss": 2.2307,
      "step": 61441
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.8071471452713013,
      "learning_rate": 0.0005007355340246528,
      "loss": 2.3357,
      "step": 61442
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.8540372252464294,
      "learning_rate": 0.0005007324940810963,
      "loss": 2.2468,
      "step": 61443
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.8001536130905151,
      "learning_rate": 0.0005007294541002199,
      "loss": 2.1515,
      "step": 61444
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.8100914359092712,
      "learning_rate": 0.000500726414082024,
      "loss": 2.0969,
      "step": 61445
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.8638571500778198,
      "learning_rate": 0.0005007233740265093,
      "loss": 2.2032,
      "step": 61446
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.8272844552993774,
      "learning_rate": 0.0005007203339336764,
      "loss": 2.3665,
      "step": 61447
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7154070734977722,
      "learning_rate": 0.0005007172938035257,
      "loss": 1.9605,
      "step": 61448
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7763770818710327,
      "learning_rate": 0.000500714253636058,
      "loss": 2.4175,
      "step": 61449
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7082287669181824,
      "learning_rate": 0.0005007112134312736,
      "loss": 1.9064,
      "step": 61450
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7634516358375549,
      "learning_rate": 0.0005007081731891732,
      "loss": 2.1953,
      "step": 61451
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7669995427131653,
      "learning_rate": 0.0005007051329097575,
      "loss": 2.1714,
      "step": 61452
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7748323082923889,
      "learning_rate": 0.0005007020925930268,
      "loss": 2.2322,
      "step": 61453
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7828220725059509,
      "learning_rate": 0.0005006990522389818,
      "loss": 2.1825,
      "step": 61454
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7425314784049988,
      "learning_rate": 0.000500696011847623,
      "loss": 2.0361,
      "step": 61455
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.6966699361801147,
      "learning_rate": 0.0005006929714189512,
      "loss": 2.3133,
      "step": 61456
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7332531809806824,
      "learning_rate": 0.0005006899309529668,
      "loss": 2.2473,
      "step": 61457
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7447639107704163,
      "learning_rate": 0.0005006868904496701,
      "loss": 2.1002,
      "step": 61458
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7859523892402649,
      "learning_rate": 0.0005006838499090622,
      "loss": 2.0758,
      "step": 61459
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7817811965942383,
      "learning_rate": 0.0005006808093311432,
      "loss": 2.2247,
      "step": 61460
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7224692702293396,
      "learning_rate": 0.000500677768715914,
      "loss": 2.1272,
      "step": 61461
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7978901863098145,
      "learning_rate": 0.0005006747280633749,
      "loss": 2.2977,
      "step": 61462
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7646669745445251,
      "learning_rate": 0.0005006716873735265,
      "loss": 2.135,
      "step": 61463
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.9103114008903503,
      "learning_rate": 0.0005006686466463695,
      "loss": 2.2311,
      "step": 61464
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.8015130162239075,
      "learning_rate": 0.0005006656058819045,
      "loss": 2.3507,
      "step": 61465
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7036210894584656,
      "learning_rate": 0.000500662565080132,
      "loss": 2.0784,
      "step": 61466
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.845458984375,
      "learning_rate": 0.0005006595242410523,
      "loss": 2.262,
      "step": 61467
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7890347242355347,
      "learning_rate": 0.0005006564833646665,
      "loss": 2.1315,
      "step": 61468
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.8440313935279846,
      "learning_rate": 0.0005006534424509748,
      "loss": 2.2435,
      "step": 61469
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7767564058303833,
      "learning_rate": 0.0005006504014999777,
      "loss": 2.2459,
      "step": 61470
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.8486835360527039,
      "learning_rate": 0.0005006473605116761,
      "loss": 2.0842,
      "step": 61471
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7162039279937744,
      "learning_rate": 0.0005006443194860703,
      "loss": 2.1028,
      "step": 61472
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7545272707939148,
      "learning_rate": 0.0005006412784231608,
      "loss": 2.3525,
      "step": 61473
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7109259366989136,
      "learning_rate": 0.0005006382373229485,
      "loss": 2.2676,
      "step": 61474
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.8711203336715698,
      "learning_rate": 0.0005006351961854336,
      "loss": 2.28,
      "step": 61475
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7576411366462708,
      "learning_rate": 0.0005006321550106169,
      "loss": 2.1313,
      "step": 61476
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.6772199273109436,
      "learning_rate": 0.0005006291137984989,
      "loss": 2.1876,
      "step": 61477
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.746957540512085,
      "learning_rate": 0.0005006260725490801,
      "loss": 2.4785,
      "step": 61478
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7429138422012329,
      "learning_rate": 0.0005006230312623613,
      "loss": 2.1189,
      "step": 61479
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7617230415344238,
      "learning_rate": 0.0005006199899383426,
      "loss": 2.254,
      "step": 61480
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.6512446403503418,
      "learning_rate": 0.000500616948577025,
      "loss": 2.2476,
      "step": 61481
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.6928117871284485,
      "learning_rate": 0.000500613907178409,
      "loss": 2.0828,
      "step": 61482
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7315976023674011,
      "learning_rate": 0.000500610865742495,
      "loss": 2.4315,
      "step": 61483
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.8384137153625488,
      "learning_rate": 0.0005006078242692837,
      "loss": 2.2969,
      "step": 61484
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7463939785957336,
      "learning_rate": 0.0005006047827587756,
      "loss": 2.381,
      "step": 61485
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7536566257476807,
      "learning_rate": 0.0005006017412109712,
      "loss": 2.2339,
      "step": 61486
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.791694164276123,
      "learning_rate": 0.0005005986996258713,
      "loss": 2.2476,
      "step": 61487
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7396467328071594,
      "learning_rate": 0.0005005956580034763,
      "loss": 2.096,
      "step": 61488
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7124923467636108,
      "learning_rate": 0.0005005926163437868,
      "loss": 2.2281,
      "step": 61489
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7676749229431152,
      "learning_rate": 0.0005005895746468032,
      "loss": 2.1824,
      "step": 61490
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7188290357589722,
      "learning_rate": 0.0005005865329125263,
      "loss": 2.2353,
      "step": 61491
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7412760257720947,
      "learning_rate": 0.0005005834911409565,
      "loss": 2.2874,
      "step": 61492
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7377235889434814,
      "learning_rate": 0.0005005804493320946,
      "loss": 2.081,
      "step": 61493
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7010208368301392,
      "learning_rate": 0.0005005774074859409,
      "loss": 2.3653,
      "step": 61494
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.8870305418968201,
      "learning_rate": 0.0005005743656024961,
      "loss": 2.2131,
      "step": 61495
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7691829800605774,
      "learning_rate": 0.0005005713236817606,
      "loss": 2.2883,
      "step": 61496
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7515386343002319,
      "learning_rate": 0.0005005682817237353,
      "loss": 2.1248,
      "step": 61497
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7486869692802429,
      "learning_rate": 0.0005005652397284205,
      "loss": 2.0726,
      "step": 61498
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7522867918014526,
      "learning_rate": 0.0005005621976958168,
      "loss": 2.2954,
      "step": 61499
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7226362228393555,
      "learning_rate": 0.0005005591556259249,
      "loss": 2.3135,
      "step": 61500
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7290734052658081,
      "learning_rate": 0.0005005561135187452,
      "loss": 2.1925,
      "step": 61501
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.6924824118614197,
      "learning_rate": 0.0005005530713742784,
      "loss": 2.1299,
      "step": 61502
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7443302869796753,
      "learning_rate": 0.0005005500291925249,
      "loss": 2.3772,
      "step": 61503
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.802473783493042,
      "learning_rate": 0.0005005469869734854,
      "loss": 2.1205,
      "step": 61504
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7380962371826172,
      "learning_rate": 0.0005005439447171604,
      "loss": 2.1603,
      "step": 61505
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.8601890206336975,
      "learning_rate": 0.0005005409024235505,
      "loss": 2.1976,
      "step": 61506
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7635752558708191,
      "learning_rate": 0.0005005378600926563,
      "loss": 2.0145,
      "step": 61507
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.8320919871330261,
      "learning_rate": 0.0005005348177244784,
      "loss": 1.9639,
      "step": 61508
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7860751152038574,
      "learning_rate": 0.0005005317753190171,
      "loss": 2.2594,
      "step": 61509
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.9569960236549377,
      "learning_rate": 0.0005005287328762733,
      "loss": 2.196,
      "step": 61510
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.8155364394187927,
      "learning_rate": 0.0005005256903962474,
      "loss": 2.2997,
      "step": 61511
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.858673632144928,
      "learning_rate": 0.00050052264787894,
      "loss": 2.1458,
      "step": 61512
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.8632903695106506,
      "learning_rate": 0.0005005196053243517,
      "loss": 2.1461,
      "step": 61513
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.710810661315918,
      "learning_rate": 0.0005005165627324828,
      "loss": 2.3121,
      "step": 61514
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7590214014053345,
      "learning_rate": 0.0005005135201033344,
      "loss": 2.2232,
      "step": 61515
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.8423430323600769,
      "learning_rate": 0.0005005104774369065,
      "loss": 2.2329,
      "step": 61516
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.753760039806366,
      "learning_rate": 0.0005005074347332,
      "loss": 2.2855,
      "step": 61517
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.6910055875778198,
      "learning_rate": 0.0005005043919922154,
      "loss": 2.1145,
      "step": 61518
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.0930160284042358,
      "learning_rate": 0.0005005013492139532,
      "loss": 2.213,
      "step": 61519
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7772252559661865,
      "learning_rate": 0.0005004983063984141,
      "loss": 2.2049,
      "step": 61520
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7662094235420227,
      "learning_rate": 0.0005004952635455985,
      "loss": 2.0804,
      "step": 61521
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7652877569198608,
      "learning_rate": 0.0005004922206555072,
      "loss": 2.3623,
      "step": 61522
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7034059166908264,
      "learning_rate": 0.0005004891777281404,
      "loss": 2.0812,
      "step": 61523
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7863583564758301,
      "learning_rate": 0.000500486134763499,
      "loss": 2.2298,
      "step": 61524
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7267645001411438,
      "learning_rate": 0.0005004830917615834,
      "loss": 2.0796,
      "step": 61525
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7243594527244568,
      "learning_rate": 0.0005004800487223943,
      "loss": 2.202,
      "step": 61526
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.821556031703949,
      "learning_rate": 0.0005004770056459321,
      "loss": 2.3172,
      "step": 61527
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7263797521591187,
      "learning_rate": 0.0005004739625321974,
      "loss": 2.217,
      "step": 61528
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.8314042091369629,
      "learning_rate": 0.0005004709193811908,
      "loss": 1.9221,
      "step": 61529
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.67604660987854,
      "learning_rate": 0.0005004678761929129,
      "loss": 2.5506,
      "step": 61530
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7730615139007568,
      "learning_rate": 0.0005004648329673643,
      "loss": 2.1421,
      "step": 61531
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7238191962242126,
      "learning_rate": 0.0005004617897045455,
      "loss": 2.2414,
      "step": 61532
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.6991216540336609,
      "learning_rate": 0.000500458746404457,
      "loss": 2.1654,
      "step": 61533
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7317184209823608,
      "learning_rate": 0.0005004557030670995,
      "loss": 2.1269,
      "step": 61534
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7993534803390503,
      "learning_rate": 0.0005004526596924734,
      "loss": 2.4339,
      "step": 61535
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7501373887062073,
      "learning_rate": 0.0005004496162805794,
      "loss": 2.2559,
      "step": 61536
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7109098434448242,
      "learning_rate": 0.000500446572831418,
      "loss": 2.4887,
      "step": 61537
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7339810132980347,
      "learning_rate": 0.0005004435293449899,
      "loss": 2.0341,
      "step": 61538
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7192164659500122,
      "learning_rate": 0.0005004404858212955,
      "loss": 2.4105,
      "step": 61539
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7371025085449219,
      "learning_rate": 0.0005004374422603355,
      "loss": 2.2534,
      "step": 61540
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7174563407897949,
      "learning_rate": 0.0005004343986621104,
      "loss": 2.1791,
      "step": 61541
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7206818461418152,
      "learning_rate": 0.0005004313550266206,
      "loss": 2.2861,
      "step": 61542
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7185729146003723,
      "learning_rate": 0.000500428311353867,
      "loss": 2.1826,
      "step": 61543
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7168657779693604,
      "learning_rate": 0.0005004252676438499,
      "loss": 2.3339,
      "step": 61544
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7945020198822021,
      "learning_rate": 0.0005004222238965699,
      "loss": 2.2736,
      "step": 61545
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7160560488700867,
      "learning_rate": 0.0005004191801120278,
      "loss": 2.0896,
      "step": 61546
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.8030914664268494,
      "learning_rate": 0.0005004161362902238,
      "loss": 2.0544,
      "step": 61547
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7814416289329529,
      "learning_rate": 0.000500413092431159,
      "loss": 2.185,
      "step": 61548
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7524251341819763,
      "learning_rate": 0.0005004100485348333,
      "loss": 2.3187,
      "step": 61549
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.9332450032234192,
      "learning_rate": 0.0005004070046012477,
      "loss": 2.0762,
      "step": 61550
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.8359866738319397,
      "learning_rate": 0.0005004039606304025,
      "loss": 2.2868,
      "step": 61551
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7187563180923462,
      "learning_rate": 0.0005004009166222986,
      "loss": 2.3801,
      "step": 61552
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.817645788192749,
      "learning_rate": 0.0005003978725769364,
      "loss": 2.2073,
      "step": 61553
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.8404321074485779,
      "learning_rate": 0.0005003948284943164,
      "loss": 2.3566,
      "step": 61554
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.769087553024292,
      "learning_rate": 0.0005003917843744392,
      "loss": 2.2581,
      "step": 61555
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.6763242483139038,
      "learning_rate": 0.0005003887402173054,
      "loss": 2.2421,
      "step": 61556
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.793057382106781,
      "learning_rate": 0.0005003856960229155,
      "loss": 2.4108,
      "step": 61557
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.751577615737915,
      "learning_rate": 0.0005003826517912702,
      "loss": 2.1486,
      "step": 61558
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7870711088180542,
      "learning_rate": 0.0005003796075223699,
      "loss": 2.2361,
      "step": 61559
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7250864505767822,
      "learning_rate": 0.0005003765632162153,
      "loss": 2.2329,
      "step": 61560
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.6845440864562988,
      "learning_rate": 0.0005003735188728069,
      "loss": 2.2701,
      "step": 61561
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7714915871620178,
      "learning_rate": 0.0005003704744921453,
      "loss": 2.1466,
      "step": 61562
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.72394859790802,
      "learning_rate": 0.000500367430074231,
      "loss": 2.4482,
      "step": 61563
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7248430252075195,
      "learning_rate": 0.0005003643856190647,
      "loss": 2.3504,
      "step": 61564
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.9605768322944641,
      "learning_rate": 0.0005003613411266468,
      "loss": 2.3217,
      "step": 61565
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7427581548690796,
      "learning_rate": 0.000500358296596978,
      "loss": 2.1338,
      "step": 61566
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7464157342910767,
      "learning_rate": 0.0005003552520300588,
      "loss": 2.3768,
      "step": 61567
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.6507402062416077,
      "learning_rate": 0.0005003522074258897,
      "loss": 2.0168,
      "step": 61568
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.6495832204818726,
      "learning_rate": 0.0005003491627844714,
      "loss": 2.4413,
      "step": 61569
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7954673767089844,
      "learning_rate": 0.0005003461181058045,
      "loss": 2.1209,
      "step": 61570
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7359073758125305,
      "learning_rate": 0.0005003430733898893,
      "loss": 2.4049,
      "step": 61571
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.6882510781288147,
      "learning_rate": 0.0005003400286367266,
      "loss": 1.8976,
      "step": 61572
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.8684346675872803,
      "learning_rate": 0.000500336983846317,
      "loss": 2.0756,
      "step": 61573
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.9290491342544556,
      "learning_rate": 0.0005003339390186608,
      "loss": 2.2108,
      "step": 61574
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.728381335735321,
      "learning_rate": 0.000500330894153759,
      "loss": 2.1962,
      "step": 61575
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.8460855484008789,
      "learning_rate": 0.0005003278492516117,
      "loss": 2.1245,
      "step": 61576
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7671363949775696,
      "learning_rate": 0.0005003248043122197,
      "loss": 2.0405,
      "step": 61577
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.6929860711097717,
      "learning_rate": 0.0005003217593355835,
      "loss": 2.0501,
      "step": 61578
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7871312499046326,
      "learning_rate": 0.0005003187143217038,
      "loss": 2.2075,
      "step": 61579
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7987385392189026,
      "learning_rate": 0.0005003156692705811,
      "loss": 2.1931,
      "step": 61580
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.9308366775512695,
      "learning_rate": 0.0005003126241822158,
      "loss": 2.1043,
      "step": 61581
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7595345377922058,
      "learning_rate": 0.0005003095790566086,
      "loss": 2.3353,
      "step": 61582
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.8512170314788818,
      "learning_rate": 0.0005003065338937602,
      "loss": 2.4587,
      "step": 61583
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.8494242429733276,
      "learning_rate": 0.0005003034886936709,
      "loss": 2.4154,
      "step": 61584
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.832147479057312,
      "learning_rate": 0.0005003004434563415,
      "loss": 2.2651,
      "step": 61585
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.801160454750061,
      "learning_rate": 0.0005002973981817725,
      "loss": 2.3156,
      "step": 61586
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.736774206161499,
      "learning_rate": 0.0005002943528699643,
      "loss": 2.1339,
      "step": 61587
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7104600667953491,
      "learning_rate": 0.0005002913075209177,
      "loss": 2.18,
      "step": 61588
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.8513482213020325,
      "learning_rate": 0.0005002882621346331,
      "loss": 2.3531,
      "step": 61589
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7987783551216125,
      "learning_rate": 0.0005002852167111112,
      "loss": 2.2838,
      "step": 61590
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.778805673122406,
      "learning_rate": 0.0005002821712503524,
      "loss": 2.0358,
      "step": 61591
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7960929870605469,
      "learning_rate": 0.0005002791257523575,
      "loss": 2.2969,
      "step": 61592
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.9004468321800232,
      "learning_rate": 0.0005002760802171269,
      "loss": 2.2574,
      "step": 61593
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7101446390151978,
      "learning_rate": 0.0005002730346446611,
      "loss": 2.2443,
      "step": 61594
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7174807190895081,
      "learning_rate": 0.0005002699890349608,
      "loss": 2.1744,
      "step": 61595
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7090750932693481,
      "learning_rate": 0.0005002669433880265,
      "loss": 2.054,
      "step": 61596
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.9947105050086975,
      "learning_rate": 0.000500263897703859,
      "loss": 2.1647,
      "step": 61597
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.8101933598518372,
      "learning_rate": 0.0005002608519824584,
      "loss": 2.4655,
      "step": 61598
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7708874940872192,
      "learning_rate": 0.0005002578062238256,
      "loss": 1.991,
      "step": 61599
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.8057712316513062,
      "learning_rate": 0.0005002547604279612,
      "loss": 2.4704,
      "step": 61600
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.84781414270401,
      "learning_rate": 0.0005002517145948655,
      "loss": 2.3812,
      "step": 61601
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7865815758705139,
      "learning_rate": 0.0005002486687245393,
      "loss": 2.4079,
      "step": 61602
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7797834873199463,
      "learning_rate": 0.0005002456228169832,
      "loss": 2.4498,
      "step": 61603
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.8161443471908569,
      "learning_rate": 0.0005002425768721975,
      "loss": 2.2136,
      "step": 61604
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.8455069065093994,
      "learning_rate": 0.000500239530890183,
      "loss": 2.0186,
      "step": 61605
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7403829097747803,
      "learning_rate": 0.0005002364848709402,
      "loss": 2.3809,
      "step": 61606
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7521491646766663,
      "learning_rate": 0.0005002334388144697,
      "loss": 2.161,
      "step": 61607
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7944079041481018,
      "learning_rate": 0.0005002303927207719,
      "loss": 2.2535,
      "step": 61608
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.9682583808898926,
      "learning_rate": 0.0005002273465898475,
      "loss": 2.3659,
      "step": 61609
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.6888670921325684,
      "learning_rate": 0.0005002243004216971,
      "loss": 2.1334,
      "step": 61610
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.8153354525566101,
      "learning_rate": 0.0005002212542163212,
      "loss": 2.2577,
      "step": 61611
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.9775651693344116,
      "learning_rate": 0.0005002182079737205,
      "loss": 2.5137,
      "step": 61612
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.8393731713294983,
      "learning_rate": 0.0005002151616938953,
      "loss": 2.5319,
      "step": 61613
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.8856573104858398,
      "learning_rate": 0.0005002121153768464,
      "loss": 2.3521,
      "step": 61614
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7763786911964417,
      "learning_rate": 0.0005002090690225744,
      "loss": 2.0306,
      "step": 61615
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.9361152648925781,
      "learning_rate": 0.0005002060226310796,
      "loss": 2.1585,
      "step": 61616
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.0650080442428589,
      "learning_rate": 0.0005002029762023627,
      "loss": 2.0805,
      "step": 61617
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.743220329284668,
      "learning_rate": 0.0005001999297364244,
      "loss": 2.1011,
      "step": 61618
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.8918324708938599,
      "learning_rate": 0.000500196883233265,
      "loss": 2.2107,
      "step": 61619
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.0518919229507446,
      "learning_rate": 0.0005001938366928853,
      "loss": 2.2773,
      "step": 61620
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.8111603856086731,
      "learning_rate": 0.0005001907901152859,
      "loss": 2.2436,
      "step": 61621
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7908235788345337,
      "learning_rate": 0.0005001877435004672,
      "loss": 2.251,
      "step": 61622
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7774280309677124,
      "learning_rate": 0.0005001846968484297,
      "loss": 2.3638,
      "step": 61623
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7660521864891052,
      "learning_rate": 0.0005001816501591742,
      "loss": 1.9885,
      "step": 61624
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.8352908492088318,
      "learning_rate": 0.0005001786034327012,
      "loss": 2.2776,
      "step": 61625
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.743064284324646,
      "learning_rate": 0.0005001755566690111,
      "loss": 2.3219,
      "step": 61626
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.8312636613845825,
      "learning_rate": 0.0005001725098681046,
      "loss": 2.3199,
      "step": 61627
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.8075958490371704,
      "learning_rate": 0.0005001694630299823,
      "loss": 2.383,
      "step": 61628
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.6887689232826233,
      "learning_rate": 0.0005001664161546447,
      "loss": 2.0018,
      "step": 61629
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7631295323371887,
      "learning_rate": 0.0005001633692420923,
      "loss": 2.2392,
      "step": 61630
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7094374299049377,
      "learning_rate": 0.0005001603222923259,
      "loss": 2.3361,
      "step": 61631
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.6989970207214355,
      "learning_rate": 0.0005001572753053459,
      "loss": 2.1219,
      "step": 61632
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.8966554403305054,
      "learning_rate": 0.0005001542282811528,
      "loss": 2.2883,
      "step": 61633
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7428609728813171,
      "learning_rate": 0.0005001511812197472,
      "loss": 1.9566,
      "step": 61634
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.80303955078125,
      "learning_rate": 0.0005001481341211298,
      "loss": 2.2219,
      "step": 61635
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7675424218177795,
      "learning_rate": 0.0005001450869853011,
      "loss": 2.1411,
      "step": 61636
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.70815509557724,
      "learning_rate": 0.0005001420398122616,
      "loss": 2.1793,
      "step": 61637
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7691194415092468,
      "learning_rate": 0.0005001389926020119,
      "loss": 2.4371,
      "step": 61638
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.8712441921234131,
      "learning_rate": 0.0005001359453545527,
      "loss": 2.3886,
      "step": 61639
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.8661999106407166,
      "learning_rate": 0.0005001328980698844,
      "loss": 2.068,
      "step": 61640
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7955812215805054,
      "learning_rate": 0.0005001298507480076,
      "loss": 2.0227,
      "step": 61641
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.9464364051818848,
      "learning_rate": 0.0005001268033889228,
      "loss": 2.1638,
      "step": 61642
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7338139414787292,
      "learning_rate": 0.0005001237559926308,
      "loss": 2.2689,
      "step": 61643
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.736545741558075,
      "learning_rate": 0.0005001207085591319,
      "loss": 2.2443,
      "step": 61644
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.8625474572181702,
      "learning_rate": 0.0005001176610884268,
      "loss": 2.2596,
      "step": 61645
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7475496530532837,
      "learning_rate": 0.0005001146135805162,
      "loss": 2.313,
      "step": 61646
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.736725389957428,
      "learning_rate": 0.0005001115660354004,
      "loss": 2.3869,
      "step": 61647
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7733645439147949,
      "learning_rate": 0.00050010851845308,
      "loss": 2.1221,
      "step": 61648
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7033483982086182,
      "learning_rate": 0.0005001054708335558,
      "loss": 2.4658,
      "step": 61649
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.8061572313308716,
      "learning_rate": 0.0005001024231768282,
      "loss": 2.4488,
      "step": 61650
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.749443531036377,
      "learning_rate": 0.0005000993754828976,
      "loss": 2.1923,
      "step": 61651
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7641410231590271,
      "learning_rate": 0.000500096327751765,
      "loss": 2.277,
      "step": 61652
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7233956456184387,
      "learning_rate": 0.0005000932799834306,
      "loss": 2.2109,
      "step": 61653
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7224538326263428,
      "learning_rate": 0.000500090232177895,
      "loss": 2.1091,
      "step": 61654
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7799975275993347,
      "learning_rate": 0.000500087184335159,
      "loss": 2.0096,
      "step": 61655
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.6835149526596069,
      "learning_rate": 0.0005000841364552229,
      "loss": 2.2437,
      "step": 61656
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7483443021774292,
      "learning_rate": 0.0005000810885380873,
      "loss": 2.1713,
      "step": 61657
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7399123311042786,
      "learning_rate": 0.0005000780405837531,
      "loss": 2.242,
      "step": 61658
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.9259029030799866,
      "learning_rate": 0.0005000749925922204,
      "loss": 2.2652,
      "step": 61659
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7447238564491272,
      "learning_rate": 0.0005000719445634901,
      "loss": 2.2573,
      "step": 61660
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7491870522499084,
      "learning_rate": 0.0005000688964975627,
      "loss": 2.3668,
      "step": 61661
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7520139813423157,
      "learning_rate": 0.0005000658483944385,
      "loss": 2.145,
      "step": 61662
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.8143545389175415,
      "learning_rate": 0.0005000628002541185,
      "loss": 2.1181,
      "step": 61663
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.6971229910850525,
      "learning_rate": 0.000500059752076603,
      "loss": 2.1046,
      "step": 61664
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7523697018623352,
      "learning_rate": 0.0005000567038618925,
      "loss": 2.5545,
      "step": 61665
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7105748057365417,
      "learning_rate": 0.0005000536556099877,
      "loss": 2.2338,
      "step": 61666
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.6559414863586426,
      "learning_rate": 0.0005000506073208893,
      "loss": 2.2177,
      "step": 61667
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7506941556930542,
      "learning_rate": 0.0005000475589945977,
      "loss": 2.4189,
      "step": 61668
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.712139368057251,
      "learning_rate": 0.0005000445106311133,
      "loss": 2.1098,
      "step": 61669
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7163428664207458,
      "learning_rate": 0.000500041462230437,
      "loss": 2.2697,
      "step": 61670
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7529951333999634,
      "learning_rate": 0.0005000384137925692,
      "loss": 2.2172,
      "step": 61671
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.6889492273330688,
      "learning_rate": 0.0005000353653175104,
      "loss": 2.3653,
      "step": 61672
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.8096295595169067,
      "learning_rate": 0.0005000323168052614,
      "loss": 2.2707,
      "step": 61673
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.8693189024925232,
      "learning_rate": 0.0005000292682558224,
      "loss": 2.4232,
      "step": 61674
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7230931520462036,
      "learning_rate": 0.0005000262196691943,
      "loss": 2.1206,
      "step": 61675
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7301388382911682,
      "learning_rate": 0.0005000231710453776,
      "loss": 2.1903,
      "step": 61676
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7162712216377258,
      "learning_rate": 0.0005000201223843727,
      "loss": 2.3976,
      "step": 61677
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7642106413841248,
      "learning_rate": 0.0005000170736861804,
      "loss": 2.3695,
      "step": 61678
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7885375618934631,
      "learning_rate": 0.0005000140249508012,
      "loss": 2.3492,
      "step": 61679
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7634968161582947,
      "learning_rate": 0.0005000109761782354,
      "loss": 2.0506,
      "step": 61680
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.722237229347229,
      "learning_rate": 0.000500007927368484,
      "loss": 2.1063,
      "step": 61681
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.6972879767417908,
      "learning_rate": 0.0005000048785215471,
      "loss": 2.0615,
      "step": 61682
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.8347892165184021,
      "learning_rate": 0.0005000018296374257,
      "loss": 2.4063,
      "step": 61683
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.8872930407524109,
      "learning_rate": 0.0004999987807161202,
      "loss": 2.2999,
      "step": 61684
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.8730583190917969,
      "learning_rate": 0.0004999957317576312,
      "loss": 2.1151,
      "step": 61685
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7295475602149963,
      "learning_rate": 0.000499992682761959,
      "loss": 2.2308,
      "step": 61686
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7497214078903198,
      "learning_rate": 0.0004999896337291046,
      "loss": 2.1428,
      "step": 61687
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7255629301071167,
      "learning_rate": 0.0004999865846590682,
      "loss": 2.1856,
      "step": 61688
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7513954639434814,
      "learning_rate": 0.0004999835355518507,
      "loss": 2.2327,
      "step": 61689
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7618724703788757,
      "learning_rate": 0.0004999804864074523,
      "loss": 2.1849,
      "step": 61690
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.9615852236747742,
      "learning_rate": 0.0004999774372258738,
      "loss": 2.0946,
      "step": 61691
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7719635963439941,
      "learning_rate": 0.0004999743880071157,
      "loss": 2.1188,
      "step": 61692
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.8549370169639587,
      "learning_rate": 0.0004999713387511787,
      "loss": 2.255,
      "step": 61693
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7519204020500183,
      "learning_rate": 0.0004999682894580633,
      "loss": 2.3695,
      "step": 61694
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.8343110084533691,
      "learning_rate": 0.0004999652401277698,
      "loss": 2.2107,
      "step": 61695
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7674896121025085,
      "learning_rate": 0.0004999621907602991,
      "loss": 2.1309,
      "step": 61696
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7273114323616028,
      "learning_rate": 0.0004999591413556516,
      "loss": 2.1667,
      "step": 61697
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7674232721328735,
      "learning_rate": 0.0004999560919138281,
      "loss": 2.0225,
      "step": 61698
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7198535203933716,
      "learning_rate": 0.0004999530424348287,
      "loss": 2.1226,
      "step": 61699
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7106611132621765,
      "learning_rate": 0.0004999499929186545,
      "loss": 2.2275,
      "step": 61700
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7370256781578064,
      "learning_rate": 0.0004999469433653057,
      "loss": 2.372,
      "step": 61701
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7015067934989929,
      "learning_rate": 0.000499943893774783,
      "loss": 2.2376,
      "step": 61702
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7560429573059082,
      "learning_rate": 0.0004999408441470869,
      "loss": 2.4218,
      "step": 61703
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7382708191871643,
      "learning_rate": 0.0004999377944822182,
      "loss": 2.1291,
      "step": 61704
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7591733336448669,
      "learning_rate": 0.0004999347447801773,
      "loss": 2.0789,
      "step": 61705
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.8242702484130859,
      "learning_rate": 0.0004999316950409645,
      "loss": 2.2376,
      "step": 61706
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7432550191879272,
      "learning_rate": 0.0004999286452645808,
      "loss": 2.206,
      "step": 61707
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.8496176600456238,
      "learning_rate": 0.0004999255954510266,
      "loss": 2.2462,
      "step": 61708
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7491233944892883,
      "learning_rate": 0.0004999225456003023,
      "loss": 2.2053,
      "step": 61709
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7365108728408813,
      "learning_rate": 0.0004999194957124089,
      "loss": 2.2359,
      "step": 61710
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.8795672655105591,
      "learning_rate": 0.0004999164457873465,
      "loss": 2.0901,
      "step": 61711
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7143835425376892,
      "learning_rate": 0.0004999133958251159,
      "loss": 2.1737,
      "step": 61712
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.8382895588874817,
      "learning_rate": 0.0004999103458257177,
      "loss": 2.3623,
      "step": 61713
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.6946443915367126,
      "learning_rate": 0.0004999072957891522,
      "loss": 2.3412,
      "step": 61714
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.8225871324539185,
      "learning_rate": 0.0004999042457154202,
      "loss": 2.3834,
      "step": 61715
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7666540145874023,
      "learning_rate": 0.0004999011956045223,
      "loss": 2.0014,
      "step": 61716
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7877503037452698,
      "learning_rate": 0.0004998981454564591,
      "loss": 2.0051,
      "step": 61717
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7570362091064453,
      "learning_rate": 0.0004998950952712309,
      "loss": 2.1744,
      "step": 61718
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7415800094604492,
      "learning_rate": 0.0004998920450488385,
      "loss": 2.0896,
      "step": 61719
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.894727885723114,
      "learning_rate": 0.0004998889947892824,
      "loss": 2.29,
      "step": 61720
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7550927400588989,
      "learning_rate": 0.0004998859444925631,
      "loss": 2.1226,
      "step": 61721
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7260395884513855,
      "learning_rate": 0.0004998828941586813,
      "loss": 2.4128,
      "step": 61722
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.8387855291366577,
      "learning_rate": 0.0004998798437876375,
      "loss": 1.9739,
      "step": 61723
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.8061056137084961,
      "learning_rate": 0.0004998767933794321,
      "loss": 2.2274,
      "step": 61724
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7693748474121094,
      "learning_rate": 0.0004998737429340661,
      "loss": 2.2143,
      "step": 61725
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7392320036888123,
      "learning_rate": 0.0004998706924515396,
      "loss": 2.1966,
      "step": 61726
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7385390996932983,
      "learning_rate": 0.0004998676419318535,
      "loss": 2.1985,
      "step": 61727
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7823624014854431,
      "learning_rate": 0.0004998645913750081,
      "loss": 2.3966,
      "step": 61728
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7775529026985168,
      "learning_rate": 0.0004998615407810042,
      "loss": 2.1183,
      "step": 61729
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7199271321296692,
      "learning_rate": 0.0004998584901498422,
      "loss": 2.2537,
      "step": 61730
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7599937915802002,
      "learning_rate": 0.0004998554394815228,
      "loss": 2.1981,
      "step": 61731
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.746938169002533,
      "learning_rate": 0.0004998523887760465,
      "loss": 2.142,
      "step": 61732
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.697527289390564,
      "learning_rate": 0.0004998493380334138,
      "loss": 2.0464,
      "step": 61733
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.6821772456169128,
      "learning_rate": 0.0004998462872536254,
      "loss": 2.0234,
      "step": 61734
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.6810240745544434,
      "learning_rate": 0.0004998432364366817,
      "loss": 2.3387,
      "step": 61735
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7595574855804443,
      "learning_rate": 0.0004998401855825835,
      "loss": 2.4278,
      "step": 61736
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7361769676208496,
      "learning_rate": 0.0004998371346913312,
      "loss": 2.0969,
      "step": 61737
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.6799720525741577,
      "learning_rate": 0.0004998340837629253,
      "loss": 2.3563,
      "step": 61738
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.8217187523841858,
      "learning_rate": 0.0004998310327973666,
      "loss": 2.2058,
      "step": 61739
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7354795932769775,
      "learning_rate": 0.0004998279817946554,
      "loss": 2.0573,
      "step": 61740
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.8735877871513367,
      "learning_rate": 0.0004998249307547925,
      "loss": 2.3698,
      "step": 61741
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.8283721208572388,
      "learning_rate": 0.0004998218796777784,
      "loss": 2.4563,
      "step": 61742
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.8043554425239563,
      "learning_rate": 0.0004998188285636136,
      "loss": 2.5069,
      "step": 61743
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.765557050704956,
      "learning_rate": 0.0004998157774122987,
      "loss": 2.3809,
      "step": 61744
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7598105669021606,
      "learning_rate": 0.0004998127262238343,
      "loss": 2.1887,
      "step": 61745
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.8129604458808899,
      "learning_rate": 0.0004998096749982211,
      "loss": 2.142,
      "step": 61746
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7998821139335632,
      "learning_rate": 0.0004998066237354592,
      "loss": 2.2398,
      "step": 61747
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7212477922439575,
      "learning_rate": 0.0004998035724355498,
      "loss": 1.9493,
      "step": 61748
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7921688556671143,
      "learning_rate": 0.0004998005210984929,
      "loss": 2.329,
      "step": 61749
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7525529861450195,
      "learning_rate": 0.0004997974697242894,
      "loss": 2.2602,
      "step": 61750
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7833595871925354,
      "learning_rate": 0.0004997944183129396,
      "loss": 2.1715,
      "step": 61751
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.8297454714775085,
      "learning_rate": 0.0004997913668644446,
      "loss": 2.2394,
      "step": 61752
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.9806205034255981,
      "learning_rate": 0.0004997883153788044,
      "loss": 2.4834,
      "step": 61753
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.8143766522407532,
      "learning_rate": 0.0004997852638560197,
      "loss": 2.3043,
      "step": 61754
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7561353445053101,
      "learning_rate": 0.0004997822122960912,
      "loss": 2.2863,
      "step": 61755
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7836651802062988,
      "learning_rate": 0.0004997791606990195,
      "loss": 2.0932,
      "step": 61756
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7179943323135376,
      "learning_rate": 0.0004997761090648051,
      "loss": 2.2467,
      "step": 61757
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7426449060440063,
      "learning_rate": 0.0004997730573934484,
      "loss": 2.3216,
      "step": 61758
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7800806164741516,
      "learning_rate": 0.0004997700056849503,
      "loss": 2.265,
      "step": 61759
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7865830659866333,
      "learning_rate": 0.0004997669539393111,
      "loss": 2.2805,
      "step": 61760
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7345762848854065,
      "learning_rate": 0.0004997639021565314,
      "loss": 2.2819,
      "step": 61761
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.856164813041687,
      "learning_rate": 0.0004997608503366119,
      "loss": 2.067,
      "step": 61762
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.8319942951202393,
      "learning_rate": 0.0004997577984795531,
      "loss": 2.3738,
      "step": 61763
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7017799615859985,
      "learning_rate": 0.0004997547465853554,
      "loss": 2.0585,
      "step": 61764
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7061437368392944,
      "learning_rate": 0.0004997516946540197,
      "loss": 2.257,
      "step": 61765
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7873625159263611,
      "learning_rate": 0.0004997486426855463,
      "loss": 2.2423,
      "step": 61766
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7141280770301819,
      "learning_rate": 0.000499745590679936,
      "loss": 2.3313,
      "step": 61767
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.8163261413574219,
      "learning_rate": 0.0004997425386371891,
      "loss": 2.1337,
      "step": 61768
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7989437580108643,
      "learning_rate": 0.0004997394865573063,
      "loss": 2.2228,
      "step": 61769
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7571811079978943,
      "learning_rate": 0.0004997364344402881,
      "loss": 2.1178,
      "step": 61770
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7658700942993164,
      "learning_rate": 0.0004997333822861352,
      "loss": 1.9939,
      "step": 61771
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.79180508852005,
      "learning_rate": 0.000499730330094848,
      "loss": 2.2322,
      "step": 61772
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7648565173149109,
      "learning_rate": 0.0004997272778664273,
      "loss": 2.1159,
      "step": 61773
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7279692888259888,
      "learning_rate": 0.0004997242256008734,
      "loss": 2.2089,
      "step": 61774
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7054584622383118,
      "learning_rate": 0.0004997211732981871,
      "loss": 2.1321,
      "step": 61775
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7203956246376038,
      "learning_rate": 0.0004997181209583689,
      "loss": 2.2502,
      "step": 61776
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7766432762145996,
      "learning_rate": 0.0004997150685814193,
      "loss": 2.1975,
      "step": 61777
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7339189052581787,
      "learning_rate": 0.0004997120161673388,
      "loss": 2.3854,
      "step": 61778
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.8317434787750244,
      "learning_rate": 0.0004997089637161281,
      "loss": 2.1992,
      "step": 61779
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.8187125325202942,
      "learning_rate": 0.0004997059112277877,
      "loss": 2.1263,
      "step": 61780
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.8072799444198608,
      "learning_rate": 0.0004997028587023183,
      "loss": 2.1009,
      "step": 61781
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.8993245363235474,
      "learning_rate": 0.0004996998061397203,
      "loss": 2.4107,
      "step": 61782
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.1285781860351562,
      "learning_rate": 0.0004996967535399945,
      "loss": 1.9991,
      "step": 61783
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7711650729179382,
      "learning_rate": 0.000499693700903141,
      "loss": 2.3851,
      "step": 61784
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.8641341924667358,
      "learning_rate": 0.0004996906482291609,
      "loss": 2.4079,
      "step": 61785
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.8848050832748413,
      "learning_rate": 0.0004996875955180545,
      "loss": 2.2802,
      "step": 61786
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.018331527709961,
      "learning_rate": 0.0004996845427698224,
      "loss": 2.3787,
      "step": 61787
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.6783443689346313,
      "learning_rate": 0.0004996814899844651,
      "loss": 2.0953,
      "step": 61788
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7718465328216553,
      "learning_rate": 0.0004996784371619833,
      "loss": 2.304,
      "step": 61789
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.9104056358337402,
      "learning_rate": 0.0004996753843023776,
      "loss": 2.3238,
      "step": 61790
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.0217283964157104,
      "learning_rate": 0.0004996723314056483,
      "loss": 2.3205,
      "step": 61791
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7786943316459656,
      "learning_rate": 0.0004996692784717963,
      "loss": 2.2063,
      "step": 61792
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.8291968107223511,
      "learning_rate": 0.000499666225500822,
      "loss": 2.2587,
      "step": 61793
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.957102358341217,
      "learning_rate": 0.0004996631724927258,
      "loss": 2.3335,
      "step": 61794
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7657005786895752,
      "learning_rate": 0.0004996601194475086,
      "loss": 2.0963,
      "step": 61795
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.9475246667861938,
      "learning_rate": 0.0004996570663651708,
      "loss": 2.2179,
      "step": 61796
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.9281960129737854,
      "learning_rate": 0.000499654013245713,
      "loss": 2.0567,
      "step": 61797
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.6840786337852478,
      "learning_rate": 0.0004996509600891357,
      "loss": 2.3288,
      "step": 61798
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.8752380609512329,
      "learning_rate": 0.0004996479068954396,
      "loss": 2.3663,
      "step": 61799
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.74659663438797,
      "learning_rate": 0.000499644853664625,
      "loss": 2.0415,
      "step": 61800
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.8923241496086121,
      "learning_rate": 0.0004996418003966929,
      "loss": 2.2625,
      "step": 61801
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.8213740587234497,
      "learning_rate": 0.0004996387470916434,
      "loss": 2.187,
      "step": 61802
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7152905464172363,
      "learning_rate": 0.0004996356937494774,
      "loss": 2.0323,
      "step": 61803
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.8241956233978271,
      "learning_rate": 0.0004996326403701954,
      "loss": 2.219,
      "step": 61804
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7944925427436829,
      "learning_rate": 0.0004996295869537979,
      "loss": 2.1264,
      "step": 61805
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7331198453903198,
      "learning_rate": 0.0004996265335002854,
      "loss": 2.2736,
      "step": 61806
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7885918021202087,
      "learning_rate": 0.0004996234800096586,
      "loss": 2.1249,
      "step": 61807
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.8097595572471619,
      "learning_rate": 0.000499620426481918,
      "loss": 2.289,
      "step": 61808
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7940861582756042,
      "learning_rate": 0.0004996173729170643,
      "loss": 2.3292,
      "step": 61809
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.9388346672058105,
      "learning_rate": 0.0004996143193150979,
      "loss": 1.9273,
      "step": 61810
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.8328653573989868,
      "learning_rate": 0.0004996112656760193,
      "loss": 2.2605,
      "step": 61811
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7164129614830017,
      "learning_rate": 0.0004996082119998294,
      "loss": 2.2141,
      "step": 61812
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7487974166870117,
      "learning_rate": 0.0004996051582865285,
      "loss": 2.1738,
      "step": 61813
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7630928754806519,
      "learning_rate": 0.0004996021045361172,
      "loss": 2.2045,
      "step": 61814
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7913886904716492,
      "learning_rate": 0.0004995990507485962,
      "loss": 2.2184,
      "step": 61815
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.787598192691803,
      "learning_rate": 0.0004995959969239658,
      "loss": 2.2911,
      "step": 61816
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.1122530698776245,
      "learning_rate": 0.0004995929430622269,
      "loss": 2.2109,
      "step": 61817
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.9304298162460327,
      "learning_rate": 0.0004995898891633798,
      "loss": 2.2442,
      "step": 61818
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.9531804919242859,
      "learning_rate": 0.0004995868352274252,
      "loss": 2.1216,
      "step": 61819
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7739139199256897,
      "learning_rate": 0.0004995837812543637,
      "loss": 1.9531,
      "step": 61820
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.8706159591674805,
      "learning_rate": 0.0004995807272441958,
      "loss": 2.3493,
      "step": 61821
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.765449047088623,
      "learning_rate": 0.000499577673196922,
      "loss": 2.2762,
      "step": 61822
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.8014941811561584,
      "learning_rate": 0.0004995746191125429,
      "loss": 2.3409,
      "step": 61823
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.9204971194267273,
      "learning_rate": 0.0004995715649910592,
      "loss": 2.1153,
      "step": 61824
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.9549369812011719,
      "learning_rate": 0.0004995685108324714,
      "loss": 2.2848,
      "step": 61825
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7856733798980713,
      "learning_rate": 0.00049956545663678,
      "loss": 2.4055,
      "step": 61826
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7167093753814697,
      "learning_rate": 0.0004995624024039856,
      "loss": 2.1797,
      "step": 61827
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7742049694061279,
      "learning_rate": 0.0004995593481340888,
      "loss": 2.1443,
      "step": 61828
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.772415280342102,
      "learning_rate": 0.0004995562938270902,
      "loss": 2.2143,
      "step": 61829
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7131999135017395,
      "learning_rate": 0.0004995532394829903,
      "loss": 2.2106,
      "step": 61830
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7365773916244507,
      "learning_rate": 0.0004995501851017897,
      "loss": 2.2004,
      "step": 61831
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.6887362003326416,
      "learning_rate": 0.0004995471306834888,
      "loss": 2.1377,
      "step": 61832
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.77065509557724,
      "learning_rate": 0.0004995440762280885,
      "loss": 2.4515,
      "step": 61833
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.779559850692749,
      "learning_rate": 0.0004995410217355891,
      "loss": 2.4149,
      "step": 61834
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7161223292350769,
      "learning_rate": 0.0004995379672059914,
      "loss": 2.2116,
      "step": 61835
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7072329521179199,
      "learning_rate": 0.0004995349126392957,
      "loss": 2.1829,
      "step": 61836
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.8427890539169312,
      "learning_rate": 0.0004995318580355026,
      "loss": 2.4253,
      "step": 61837
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7784795165061951,
      "learning_rate": 0.0004995288033946129,
      "loss": 2.205,
      "step": 61838
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7232081294059753,
      "learning_rate": 0.000499525748716627,
      "loss": 1.8795,
      "step": 61839
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7953815460205078,
      "learning_rate": 0.0004995226940015454,
      "loss": 2.3841,
      "step": 61840
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7011368274688721,
      "learning_rate": 0.000499519639249369,
      "loss": 2.3628,
      "step": 61841
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7187842726707458,
      "learning_rate": 0.0004995165844600979,
      "loss": 2.5609,
      "step": 61842
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.6967053413391113,
      "learning_rate": 0.000499513529633733,
      "loss": 2.0829,
      "step": 61843
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.702475368976593,
      "learning_rate": 0.0004995104747702748,
      "loss": 2.2303,
      "step": 61844
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.8805931210517883,
      "learning_rate": 0.0004995074198697238,
      "loss": 2.2005,
      "step": 61845
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7539797425270081,
      "learning_rate": 0.0004995043649320805,
      "loss": 2.1938,
      "step": 61846
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7519095540046692,
      "learning_rate": 0.0004995013099573458,
      "loss": 2.385,
      "step": 61847
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7773141264915466,
      "learning_rate": 0.0004994982549455199,
      "loss": 2.1374,
      "step": 61848
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7800607681274414,
      "learning_rate": 0.0004994951998966035,
      "loss": 2.0692,
      "step": 61849
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.757199227809906,
      "learning_rate": 0.0004994921448105971,
      "loss": 2.3432,
      "step": 61850
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7428739666938782,
      "learning_rate": 0.0004994890896875016,
      "loss": 2.2194,
      "step": 61851
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7405728101730347,
      "learning_rate": 0.000499486034527317,
      "loss": 2.1222,
      "step": 61852
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7340267300605774,
      "learning_rate": 0.0004994829793300443,
      "loss": 2.2351,
      "step": 61853
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.8174325823783875,
      "learning_rate": 0.000499479924095684,
      "loss": 2.2061,
      "step": 61854
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7211512327194214,
      "learning_rate": 0.0004994768688242366,
      "loss": 2.1543,
      "step": 61855
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.6985680460929871,
      "learning_rate": 0.0004994738135157027,
      "loss": 2.1878,
      "step": 61856
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.8407074213027954,
      "learning_rate": 0.0004994707581700829,
      "loss": 2.3835,
      "step": 61857
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7155560255050659,
      "learning_rate": 0.0004994677027873776,
      "loss": 2.188,
      "step": 61858
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.8009531497955322,
      "learning_rate": 0.0004994646473675876,
      "loss": 2.2877,
      "step": 61859
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7521462440490723,
      "learning_rate": 0.0004994615919107132,
      "loss": 2.3678,
      "step": 61860
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.6767048835754395,
      "learning_rate": 0.0004994585364167553,
      "loss": 2.3971,
      "step": 61861
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.697479248046875,
      "learning_rate": 0.0004994554808857143,
      "loss": 2.4311,
      "step": 61862
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7079437375068665,
      "learning_rate": 0.0004994524253175907,
      "loss": 2.2955,
      "step": 61863
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.804219663143158,
      "learning_rate": 0.000499449369712385,
      "loss": 2.2777,
      "step": 61864
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.6950557827949524,
      "learning_rate": 0.0004994463140700981,
      "loss": 2.274,
      "step": 61865
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7819643020629883,
      "learning_rate": 0.0004994432583907303,
      "loss": 2.2393,
      "step": 61866
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7829186916351318,
      "learning_rate": 0.0004994402026742822,
      "loss": 2.2047,
      "step": 61867
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.8576105833053589,
      "learning_rate": 0.0004994371469207544,
      "loss": 2.0376,
      "step": 61868
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7224791049957275,
      "learning_rate": 0.0004994340911301475,
      "loss": 2.105,
      "step": 61869
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7550792098045349,
      "learning_rate": 0.000499431035302462,
      "loss": 2.3641,
      "step": 61870
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7628911137580872,
      "learning_rate": 0.0004994279794376985,
      "loss": 2.2489,
      "step": 61871
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7201350331306458,
      "learning_rate": 0.0004994249235358575,
      "loss": 2.2787,
      "step": 61872
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.6974254250526428,
      "learning_rate": 0.0004994218675969398,
      "loss": 2.2016,
      "step": 61873
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7717193961143494,
      "learning_rate": 0.0004994188116209458,
      "loss": 2.0916,
      "step": 61874
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7262875437736511,
      "learning_rate": 0.000499415755607876,
      "loss": 2.1986,
      "step": 61875
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.742729663848877,
      "learning_rate": 0.000499412699557731,
      "loss": 2.3338,
      "step": 61876
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.9002100229263306,
      "learning_rate": 0.0004994096434705116,
      "loss": 2.4827,
      "step": 61877
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7921664118766785,
      "learning_rate": 0.0004994065873462181,
      "loss": 2.302,
      "step": 61878
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7441388368606567,
      "learning_rate": 0.000499403531184851,
      "loss": 2.1664,
      "step": 61879
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7559005618095398,
      "learning_rate": 0.0004994004749864112,
      "loss": 2.4676,
      "step": 61880
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7602498531341553,
      "learning_rate": 0.000499397418750899,
      "loss": 2.0277,
      "step": 61881
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7211630344390869,
      "learning_rate": 0.0004993943624783152,
      "loss": 2.1354,
      "step": 61882
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.874727189540863,
      "learning_rate": 0.00049939130616866,
      "loss": 2.1992,
      "step": 61883
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7233538627624512,
      "learning_rate": 0.0004993882498219343,
      "loss": 2.2982,
      "step": 61884
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7525073885917664,
      "learning_rate": 0.0004993851934381386,
      "loss": 2.2545,
      "step": 61885
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7438093423843384,
      "learning_rate": 0.0004993821370172734,
      "loss": 2.3048,
      "step": 61886
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.9438441395759583,
      "learning_rate": 0.0004993790805593394,
      "loss": 2.3151,
      "step": 61887
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.8116849660873413,
      "learning_rate": 0.0004993760240643368,
      "loss": 2.3008,
      "step": 61888
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.8068570494651794,
      "learning_rate": 0.0004993729675322666,
      "loss": 2.1421,
      "step": 61889
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7819264531135559,
      "learning_rate": 0.0004993699109631292,
      "loss": 2.4485,
      "step": 61890
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7601709365844727,
      "learning_rate": 0.0004993668543569253,
      "loss": 2.1676,
      "step": 61891
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.6861545443534851,
      "learning_rate": 0.0004993637977136552,
      "loss": 2.2538,
      "step": 61892
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7497936487197876,
      "learning_rate": 0.0004993607410333195,
      "loss": 2.2398,
      "step": 61893
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7529180645942688,
      "learning_rate": 0.000499357684315919,
      "loss": 2.3179,
      "step": 61894
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7292474508285522,
      "learning_rate": 0.0004993546275614541,
      "loss": 2.2085,
      "step": 61895
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.8147150874137878,
      "learning_rate": 0.0004993515707699254,
      "loss": 2.1262,
      "step": 61896
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.8817468881607056,
      "learning_rate": 0.0004993485139413335,
      "loss": 2.2728,
      "step": 61897
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.754561722278595,
      "learning_rate": 0.0004993454570756791,
      "loss": 2.1288,
      "step": 61898
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7404142618179321,
      "learning_rate": 0.0004993424001729624,
      "loss": 2.2871,
      "step": 61899
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7328099608421326,
      "learning_rate": 0.0004993393432331843,
      "loss": 2.171,
      "step": 61900
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.8188390135765076,
      "learning_rate": 0.0004993362862563452,
      "loss": 2.3933,
      "step": 61901
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.8595072031021118,
      "learning_rate": 0.0004993332292424458,
      "loss": 2.0977,
      "step": 61902
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.8059709668159485,
      "learning_rate": 0.0004993301721914864,
      "loss": 2.0403,
      "step": 61903
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7743973731994629,
      "learning_rate": 0.000499327115103468,
      "loss": 2.2899,
      "step": 61904
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.8632641434669495,
      "learning_rate": 0.0004993240579783907,
      "loss": 2.0346,
      "step": 61905
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7789983749389648,
      "learning_rate": 0.0004993210008162554,
      "loss": 2.2551,
      "step": 61906
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7290905117988586,
      "learning_rate": 0.0004993179436170626,
      "loss": 2.1994,
      "step": 61907
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7359890341758728,
      "learning_rate": 0.0004993148863808129,
      "loss": 1.9587,
      "step": 61908
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.9209384322166443,
      "learning_rate": 0.0004993118291075067,
      "loss": 2.2399,
      "step": 61909
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7551867961883545,
      "learning_rate": 0.0004993087717971446,
      "loss": 2.3196,
      "step": 61910
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7276327013969421,
      "learning_rate": 0.0004993057144497274,
      "loss": 2.1322,
      "step": 61911
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.8093967437744141,
      "learning_rate": 0.0004993026570652554,
      "loss": 2.1974,
      "step": 61912
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7190030813217163,
      "learning_rate": 0.0004992995996437293,
      "loss": 2.3545,
      "step": 61913
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7124583721160889,
      "learning_rate": 0.0004992965421851496,
      "loss": 2.1499,
      "step": 61914
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.824906587600708,
      "learning_rate": 0.000499293484689517,
      "loss": 2.3511,
      "step": 61915
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7681547403335571,
      "learning_rate": 0.0004992904271568319,
      "loss": 2.3672,
      "step": 61916
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7950125336647034,
      "learning_rate": 0.0004992873695870949,
      "loss": 2.1225,
      "step": 61917
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7987675070762634,
      "learning_rate": 0.0004992843119803068,
      "loss": 2.0895,
      "step": 61918
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7689024209976196,
      "learning_rate": 0.0004992812543364679,
      "loss": 2.2539,
      "step": 61919
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.724109947681427,
      "learning_rate": 0.0004992781966555788,
      "loss": 2.3232,
      "step": 61920
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7426146864891052,
      "learning_rate": 0.0004992751389376401,
      "loss": 2.1038,
      "step": 61921
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.8242400288581848,
      "learning_rate": 0.0004992720811826524,
      "loss": 2.3108,
      "step": 61922
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7846962213516235,
      "learning_rate": 0.0004992690233906164,
      "loss": 2.1067,
      "step": 61923
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7485396862030029,
      "learning_rate": 0.0004992659655615324,
      "loss": 2.2979,
      "step": 61924
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.8066285848617554,
      "learning_rate": 0.0004992629076954011,
      "loss": 2.0314,
      "step": 61925
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7447248697280884,
      "learning_rate": 0.0004992598497922231,
      "loss": 2.2622,
      "step": 61926
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.6982967853546143,
      "learning_rate": 0.0004992567918519989,
      "loss": 2.1323,
      "step": 61927
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.6886654496192932,
      "learning_rate": 0.0004992537338747291,
      "loss": 2.0846,
      "step": 61928
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7567288875579834,
      "learning_rate": 0.0004992506758604143,
      "loss": 2.2593,
      "step": 61929
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.8367499709129333,
      "learning_rate": 0.000499247617809055,
      "loss": 2.117,
      "step": 61930
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.836105465888977,
      "learning_rate": 0.0004992445597206518,
      "loss": 2.2383,
      "step": 61931
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7828174233436584,
      "learning_rate": 0.0004992415015952054,
      "loss": 2.1489,
      "step": 61932
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7289767861366272,
      "learning_rate": 0.0004992384434327162,
      "loss": 2.2927,
      "step": 61933
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.762854814529419,
      "learning_rate": 0.0004992353852331847,
      "loss": 2.1203,
      "step": 61934
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7341986298561096,
      "learning_rate": 0.0004992323269966116,
      "loss": 2.1564,
      "step": 61935
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7482935786247253,
      "learning_rate": 0.0004992292687229975,
      "loss": 2.179,
      "step": 61936
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7910060286521912,
      "learning_rate": 0.0004992262104123429,
      "loss": 2.1575,
      "step": 61937
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7109115123748779,
      "learning_rate": 0.0004992231520646484,
      "loss": 1.976,
      "step": 61938
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7214987874031067,
      "learning_rate": 0.0004992200936799146,
      "loss": 2.1682,
      "step": 61939
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.8548484444618225,
      "learning_rate": 0.0004992170352581419,
      "loss": 1.9441,
      "step": 61940
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.695923924446106,
      "learning_rate": 0.0004992139767993311,
      "loss": 2.0943,
      "step": 61941
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7510749101638794,
      "learning_rate": 0.0004992109183034827,
      "loss": 1.9971,
      "step": 61942
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7821972370147705,
      "learning_rate": 0.0004992078597705972,
      "loss": 2.3146,
      "step": 61943
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.8233329057693481,
      "learning_rate": 0.0004992048012006751,
      "loss": 2.4218,
      "step": 61944
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.710318922996521,
      "learning_rate": 0.0004992017425937171,
      "loss": 1.9219,
      "step": 61945
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.8935375213623047,
      "learning_rate": 0.0004991986839497237,
      "loss": 2.2073,
      "step": 61946
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7648695111274719,
      "learning_rate": 0.0004991956252686957,
      "loss": 2.4054,
      "step": 61947
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.8211808204650879,
      "learning_rate": 0.0004991925665506333,
      "loss": 2.2333,
      "step": 61948
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.789702296257019,
      "learning_rate": 0.0004991895077955373,
      "loss": 2.2201,
      "step": 61949
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.8747704029083252,
      "learning_rate": 0.0004991864490034082,
      "loss": 2.1905,
      "step": 61950
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.800065279006958,
      "learning_rate": 0.0004991833901742466,
      "loss": 2.4152,
      "step": 61951
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.8160359859466553,
      "learning_rate": 0.000499180331308053,
      "loss": 2.2802,
      "step": 61952
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.8085055351257324,
      "learning_rate": 0.000499177272404828,
      "loss": 2.4432,
      "step": 61953
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.949359118938446,
      "learning_rate": 0.0004991742134645722,
      "loss": 2.2012,
      "step": 61954
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7379594445228577,
      "learning_rate": 0.0004991711544872862,
      "loss": 2.2963,
      "step": 61955
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.899524450302124,
      "learning_rate": 0.0004991680954729705,
      "loss": 2.1757,
      "step": 61956
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.9335824251174927,
      "learning_rate": 0.0004991650364216258,
      "loss": 2.5886,
      "step": 61957
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7622286677360535,
      "learning_rate": 0.0004991619773332523,
      "loss": 2.442,
      "step": 61958
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.8726192712783813,
      "learning_rate": 0.000499158918207851,
      "loss": 2.2911,
      "step": 61959
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.747652530670166,
      "learning_rate": 0.0004991558590454224,
      "loss": 2.3123,
      "step": 61960
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7831047773361206,
      "learning_rate": 0.0004991527998459668,
      "loss": 2.2576,
      "step": 61961
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.8067629933357239,
      "learning_rate": 0.0004991497406094849,
      "loss": 2.3499,
      "step": 61962
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7959408760070801,
      "learning_rate": 0.0004991466813359774,
      "loss": 2.178,
      "step": 61963
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7298864722251892,
      "learning_rate": 0.0004991436220254448,
      "loss": 2.4444,
      "step": 61964
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7456293106079102,
      "learning_rate": 0.0004991405626778876,
      "loss": 2.0871,
      "step": 61965
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.6806922554969788,
      "learning_rate": 0.0004991375032933064,
      "loss": 2.0469,
      "step": 61966
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.745296061038971,
      "learning_rate": 0.0004991344438717018,
      "loss": 2.3134,
      "step": 61967
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7393279671669006,
      "learning_rate": 0.0004991313844130743,
      "loss": 2.0122,
      "step": 61968
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.8036041855812073,
      "learning_rate": 0.0004991283249174246,
      "loss": 2.1202,
      "step": 61969
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7718999981880188,
      "learning_rate": 0.0004991252653847531,
      "loss": 2.2259,
      "step": 61970
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.880021870136261,
      "learning_rate": 0.0004991222058150605,
      "loss": 2.1898,
      "step": 61971
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.9279442429542542,
      "learning_rate": 0.0004991191462083473,
      "loss": 2.1162,
      "step": 61972
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7770102620124817,
      "learning_rate": 0.0004991160865646142,
      "loss": 2.2909,
      "step": 61973
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.8203635811805725,
      "learning_rate": 0.0004991130268838615,
      "loss": 2.2545,
      "step": 61974
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.6858813762664795,
      "learning_rate": 0.00049910996716609,
      "loss": 2.1421,
      "step": 61975
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7153992056846619,
      "learning_rate": 0.0004991069074113001,
      "loss": 2.2282,
      "step": 61976
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7241084575653076,
      "learning_rate": 0.0004991038476194926,
      "loss": 2.3029,
      "step": 61977
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.8586245775222778,
      "learning_rate": 0.000499100787790668,
      "loss": 2.2072,
      "step": 61978
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7805635929107666,
      "learning_rate": 0.0004990977279248266,
      "loss": 2.3968,
      "step": 61979
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7665368914604187,
      "learning_rate": 0.0004990946680219693,
      "loss": 2.3876,
      "step": 61980
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.927916407585144,
      "learning_rate": 0.0004990916080820965,
      "loss": 2.1796,
      "step": 61981
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.754168689250946,
      "learning_rate": 0.0004990885481052089,
      "loss": 2.4785,
      "step": 61982
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.8602439761161804,
      "learning_rate": 0.0004990854880913069,
      "loss": 2.2164,
      "step": 61983
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.9589048027992249,
      "learning_rate": 0.0004990824280403912,
      "loss": 2.3345,
      "step": 61984
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.8442791700363159,
      "learning_rate": 0.0004990793679524623,
      "loss": 2.3231,
      "step": 61985
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.8127744197845459,
      "learning_rate": 0.0004990763078275207,
      "loss": 2.1154,
      "step": 61986
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.8351303339004517,
      "learning_rate": 0.0004990732476655671,
      "loss": 2.1227,
      "step": 61987
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7739003896713257,
      "learning_rate": 0.0004990701874666021,
      "loss": 2.2623,
      "step": 61988
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.8080572485923767,
      "learning_rate": 0.0004990671272306262,
      "loss": 2.1211,
      "step": 61989
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.8372527360916138,
      "learning_rate": 0.0004990640669576399,
      "loss": 2.3004,
      "step": 61990
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.92116779088974,
      "learning_rate": 0.0004990610066476438,
      "loss": 2.3146,
      "step": 61991
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7666406631469727,
      "learning_rate": 0.0004990579463006386,
      "loss": 2.255,
      "step": 61992
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7107515931129456,
      "learning_rate": 0.0004990548859166248,
      "loss": 2.2325,
      "step": 61993
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.6723107099533081,
      "learning_rate": 0.0004990518254956028,
      "loss": 2.2065,
      "step": 61994
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7629802823066711,
      "learning_rate": 0.0004990487650375733,
      "loss": 2.4322,
      "step": 61995
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.8051654100418091,
      "learning_rate": 0.000499045704542537,
      "loss": 2.232,
      "step": 61996
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.863527774810791,
      "learning_rate": 0.0004990426440104942,
      "loss": 2.1477,
      "step": 61997
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7796334028244019,
      "learning_rate": 0.0004990395834414457,
      "loss": 2.273,
      "step": 61998
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.9000649452209473,
      "learning_rate": 0.0004990365228353921,
      "loss": 2.3249,
      "step": 61999
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7185139060020447,
      "learning_rate": 0.0004990334621923336,
      "loss": 2.3628,
      "step": 62000
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.8235902786254883,
      "learning_rate": 0.0004990304015122712,
      "loss": 2.3435,
      "step": 62001
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.8383402824401855,
      "learning_rate": 0.0004990273407952053,
      "loss": 1.9724,
      "step": 62002
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7178806662559509,
      "learning_rate": 0.0004990242800411364,
      "loss": 2.3472,
      "step": 62003
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7261984944343567,
      "learning_rate": 0.000499021219250065,
      "loss": 2.0622,
      "step": 62004
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.8051788210868835,
      "learning_rate": 0.000499018158421992,
      "loss": 2.3219,
      "step": 62005
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.8965407609939575,
      "learning_rate": 0.0004990150975569177,
      "loss": 2.3508,
      "step": 62006
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.8939719796180725,
      "learning_rate": 0.0004990120366548427,
      "loss": 2.3058,
      "step": 62007
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7960549592971802,
      "learning_rate": 0.0004990089757157678,
      "loss": 2.1799,
      "step": 62008
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7595023512840271,
      "learning_rate": 0.0004990059147396932,
      "loss": 2.2596,
      "step": 62009
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.8555631041526794,
      "learning_rate": 0.0004990028537266197,
      "loss": 2.2903,
      "step": 62010
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.8302862048149109,
      "learning_rate": 0.0004989997926765477,
      "loss": 2.1355,
      "step": 62011
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7756310701370239,
      "learning_rate": 0.000498996731589478,
      "loss": 2.1911,
      "step": 62012
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.8928811550140381,
      "learning_rate": 0.000498993670465411,
      "loss": 2.393,
      "step": 62013
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.8164350390434265,
      "learning_rate": 0.0004989906093043474,
      "loss": 2.1556,
      "step": 62014
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7677760124206543,
      "learning_rate": 0.0004989875481062877,
      "loss": 2.5213,
      "step": 62015
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.9440333247184753,
      "learning_rate": 0.0004989844868712323,
      "loss": 2.4277,
      "step": 62016
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7982796430587769,
      "learning_rate": 0.0004989814255991821,
      "loss": 2.0911,
      "step": 62017
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7018404603004456,
      "learning_rate": 0.0004989783642901373,
      "loss": 2.1781,
      "step": 62018
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7170716524124146,
      "learning_rate": 0.0004989753029440988,
      "loss": 2.4722,
      "step": 62019
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.8336488604545593,
      "learning_rate": 0.000498972241561067,
      "loss": 2.2526,
      "step": 62020
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.783268928527832,
      "learning_rate": 0.0004989691801410425,
      "loss": 2.3148,
      "step": 62021
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7546848058700562,
      "learning_rate": 0.0004989661186840259,
      "loss": 2.1249,
      "step": 62022
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.766276478767395,
      "learning_rate": 0.0004989630571900177,
      "loss": 2.1733,
      "step": 62023
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7782894372940063,
      "learning_rate": 0.0004989599956590185,
      "loss": 2.1174,
      "step": 62024
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7349033355712891,
      "learning_rate": 0.000498956934091029,
      "loss": 2.1827,
      "step": 62025
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7519583702087402,
      "learning_rate": 0.0004989538724860496,
      "loss": 2.1414,
      "step": 62026
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.0005842447280884,
      "learning_rate": 0.0004989508108440809,
      "loss": 2.3316,
      "step": 62027
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7467033267021179,
      "learning_rate": 0.0004989477491651235,
      "loss": 2.2865,
      "step": 62028
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.8260908722877502,
      "learning_rate": 0.000498944687449178,
      "loss": 2.4566,
      "step": 62029
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.2534071207046509,
      "learning_rate": 0.0004989416256962447,
      "loss": 2.2456,
      "step": 62030
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7759416699409485,
      "learning_rate": 0.0004989385639063246,
      "loss": 2.3257,
      "step": 62031
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7300283312797546,
      "learning_rate": 0.000498935502079418,
      "loss": 2.3682,
      "step": 62032
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.8697259426116943,
      "learning_rate": 0.0004989324402155256,
      "loss": 2.3011,
      "step": 62033
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7431106567382812,
      "learning_rate": 0.0004989293783146478,
      "loss": 2.2057,
      "step": 62034
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7180284261703491,
      "learning_rate": 0.0004989263163767855,
      "loss": 2.3854,
      "step": 62035
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7564924359321594,
      "learning_rate": 0.0004989232544019388,
      "loss": 2.1897,
      "step": 62036
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7117322683334351,
      "learning_rate": 0.0004989201923901086,
      "loss": 2.3343,
      "step": 62037
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7767226696014404,
      "learning_rate": 0.0004989171303412954,
      "loss": 2.1456,
      "step": 62038
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7166174650192261,
      "learning_rate": 0.0004989140682554998,
      "loss": 2.0579,
      "step": 62039
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.8867841958999634,
      "learning_rate": 0.0004989110061327221,
      "loss": 2.3366,
      "step": 62040
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.8003503680229187,
      "learning_rate": 0.0004989079439729633,
      "loss": 2.175,
      "step": 62041
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.779637336730957,
      "learning_rate": 0.0004989048817762236,
      "loss": 2.1415,
      "step": 62042
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.8608853220939636,
      "learning_rate": 0.0004989018195425039,
      "loss": 2.4211,
      "step": 62043
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.834958016872406,
      "learning_rate": 0.0004988987572718045,
      "loss": 2.1556,
      "step": 62044
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7449740767478943,
      "learning_rate": 0.0004988956949641261,
      "loss": 2.3025,
      "step": 62045
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7669281959533691,
      "learning_rate": 0.0004988926326194693,
      "loss": 2.3119,
      "step": 62046
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.9183769822120667,
      "learning_rate": 0.0004988895702378345,
      "loss": 2.2022,
      "step": 62047
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7797600030899048,
      "learning_rate": 0.0004988865078192225,
      "loss": 2.1934,
      "step": 62048
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7313165664672852,
      "learning_rate": 0.0004988834453636335,
      "loss": 2.3624,
      "step": 62049
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7684172987937927,
      "learning_rate": 0.0004988803828710684,
      "loss": 2.2783,
      "step": 62050
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7526746392250061,
      "learning_rate": 0.0004988773203415278,
      "loss": 2.2598,
      "step": 62051
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7118330597877502,
      "learning_rate": 0.0004988742577750121,
      "loss": 2.1717,
      "step": 62052
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.8926660418510437,
      "learning_rate": 0.000498871195171522,
      "loss": 2.3031,
      "step": 62053
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7779133319854736,
      "learning_rate": 0.0004988681325310579,
      "loss": 2.2613,
      "step": 62054
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.6677210330963135,
      "learning_rate": 0.0004988650698536205,
      "loss": 2.0191,
      "step": 62055
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7447162866592407,
      "learning_rate": 0.0004988620071392103,
      "loss": 2.342,
      "step": 62056
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7535511255264282,
      "learning_rate": 0.000498858944387828,
      "loss": 2.2502,
      "step": 62057
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7973951101303101,
      "learning_rate": 0.0004988558815994739,
      "loss": 2.3062,
      "step": 62058
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7400233149528503,
      "learning_rate": 0.0004988528187741488,
      "loss": 2.2004,
      "step": 62059
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7476639151573181,
      "learning_rate": 0.0004988497559118532,
      "loss": 2.1183,
      "step": 62060
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7723373770713806,
      "learning_rate": 0.0004988466930125877,
      "loss": 2.4348,
      "step": 62061
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7728269696235657,
      "learning_rate": 0.0004988436300763528,
      "loss": 2.3558,
      "step": 62062
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7003057599067688,
      "learning_rate": 0.0004988405671031491,
      "loss": 2.4483,
      "step": 62063
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7702774405479431,
      "learning_rate": 0.0004988375040929772,
      "loss": 2.1314,
      "step": 62064
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.8079028725624084,
      "learning_rate": 0.0004988344410458377,
      "loss": 2.1729,
      "step": 62065
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.793659508228302,
      "learning_rate": 0.0004988313779617312,
      "loss": 2.2399,
      "step": 62066
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7834798097610474,
      "learning_rate": 0.0004988283148406581,
      "loss": 2.1891,
      "step": 62067
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7604409456253052,
      "learning_rate": 0.0004988252516826191,
      "loss": 2.2505,
      "step": 62068
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.8002880215644836,
      "learning_rate": 0.0004988221884876146,
      "loss": 2.2246,
      "step": 62069
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.8288161754608154,
      "learning_rate": 0.0004988191252556454,
      "loss": 2.181,
      "step": 62070
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.8384721875190735,
      "learning_rate": 0.000498816061986712,
      "loss": 2.2997,
      "step": 62071
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.6623949408531189,
      "learning_rate": 0.000498812998680815,
      "loss": 2.2041,
      "step": 62072
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.8193616271018982,
      "learning_rate": 0.0004988099353379548,
      "loss": 2.3613,
      "step": 62073
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.6851428747177124,
      "learning_rate": 0.0004988068719581321,
      "loss": 2.3875,
      "step": 62074
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.8186739683151245,
      "learning_rate": 0.0004988038085413475,
      "loss": 2.1765,
      "step": 62075
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7367925643920898,
      "learning_rate": 0.0004988007450876014,
      "loss": 2.3552,
      "step": 62076
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7906466126441956,
      "learning_rate": 0.0004987976815968946,
      "loss": 2.2146,
      "step": 62077
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7873948216438293,
      "learning_rate": 0.0004987946180692275,
      "loss": 2.2866,
      "step": 62078
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.8157164454460144,
      "learning_rate": 0.0004987915545046007,
      "loss": 2.1968,
      "step": 62079
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7709864974021912,
      "learning_rate": 0.0004987884909030149,
      "loss": 2.3206,
      "step": 62080
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7473379969596863,
      "learning_rate": 0.0004987854272644705,
      "loss": 2.2716,
      "step": 62081
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7256063222885132,
      "learning_rate": 0.0004987823635889682,
      "loss": 2.1516,
      "step": 62082
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.8124331831932068,
      "learning_rate": 0.0004987792998765085,
      "loss": 2.3934,
      "step": 62083
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7624443173408508,
      "learning_rate": 0.0004987762361270919,
      "loss": 2.2183,
      "step": 62084
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7211522459983826,
      "learning_rate": 0.000498773172340719,
      "loss": 2.2436,
      "step": 62085
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.8664517998695374,
      "learning_rate": 0.0004987701085173905,
      "loss": 2.3937,
      "step": 62086
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7741408348083496,
      "learning_rate": 0.000498767044657107,
      "loss": 2.2981,
      "step": 62087
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.6850669980049133,
      "learning_rate": 0.0004987639807598689,
      "loss": 2.1888,
      "step": 62088
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7429377436637878,
      "learning_rate": 0.0004987609168256766,
      "loss": 2.3179,
      "step": 62089
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7632113695144653,
      "learning_rate": 0.0004987578528545311,
      "loss": 2.3571,
      "step": 62090
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.8142903447151184,
      "learning_rate": 0.0004987547888464327,
      "loss": 2.1697,
      "step": 62091
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.865804135799408,
      "learning_rate": 0.0004987517248013821,
      "loss": 2.2748,
      "step": 62092
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.6734141111373901,
      "learning_rate": 0.0004987486607193797,
      "loss": 2.1021,
      "step": 62093
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7675301432609558,
      "learning_rate": 0.0004987455966004263,
      "loss": 2.0784,
      "step": 62094
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.8218046426773071,
      "learning_rate": 0.0004987425324445222,
      "loss": 2.2909,
      "step": 62095
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.6745778918266296,
      "learning_rate": 0.0004987394682516682,
      "loss": 2.1607,
      "step": 62096
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7329961657524109,
      "learning_rate": 0.0004987364040218647,
      "loss": 2.4222,
      "step": 62097
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7366057634353638,
      "learning_rate": 0.0004987333397551124,
      "loss": 2.0083,
      "step": 62098
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.8390712738037109,
      "learning_rate": 0.000498730275451412,
      "loss": 2.1,
      "step": 62099
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.796409547328949,
      "learning_rate": 0.0004987272111107637,
      "loss": 2.2262,
      "step": 62100
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7862803339958191,
      "learning_rate": 0.0004987241467331682,
      "loss": 2.2057,
      "step": 62101
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.802859902381897,
      "learning_rate": 0.0004987210823186262,
      "loss": 2.3366,
      "step": 62102
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7323150038719177,
      "learning_rate": 0.0004987180178671383,
      "loss": 2.0781,
      "step": 62103
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7490919232368469,
      "learning_rate": 0.0004987149533787049,
      "loss": 1.9728,
      "step": 62104
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7162908911705017,
      "learning_rate": 0.0004987118888533265,
      "loss": 2.5187,
      "step": 62105
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7135735154151917,
      "learning_rate": 0.000498708824291004,
      "loss": 2.2045,
      "step": 62106
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7071937918663025,
      "learning_rate": 0.0004987057596917377,
      "loss": 2.4379,
      "step": 62107
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.6843428015708923,
      "learning_rate": 0.0004987026950555283,
      "loss": 2.2144,
      "step": 62108
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7573741674423218,
      "learning_rate": 0.0004986996303823763,
      "loss": 2.2963,
      "step": 62109
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7228197455406189,
      "learning_rate": 0.0004986965656722822,
      "loss": 2.1408,
      "step": 62110
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.6922299861907959,
      "learning_rate": 0.0004986935009252467,
      "loss": 2.2718,
      "step": 62111
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7494048476219177,
      "learning_rate": 0.0004986904361412704,
      "loss": 2.2038,
      "step": 62112
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7890909910202026,
      "learning_rate": 0.0004986873713203537,
      "loss": 2.1766,
      "step": 62113
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7441186308860779,
      "learning_rate": 0.0004986843064624973,
      "loss": 2.1641,
      "step": 62114
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.725717306137085,
      "learning_rate": 0.0004986812415677018,
      "loss": 2.348,
      "step": 62115
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7726121544837952,
      "learning_rate": 0.0004986781766359676,
      "loss": 2.0143,
      "step": 62116
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.6963005661964417,
      "learning_rate": 0.0004986751116672954,
      "loss": 2.4789,
      "step": 62117
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7305765748023987,
      "learning_rate": 0.0004986720466616858,
      "loss": 2.1083,
      "step": 62118
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7236781120300293,
      "learning_rate": 0.0004986689816191393,
      "loss": 2.0823,
      "step": 62119
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7079936861991882,
      "learning_rate": 0.0004986659165396565,
      "loss": 2.1475,
      "step": 62120
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.908051073551178,
      "learning_rate": 0.000498662851423238,
      "loss": 2.2603,
      "step": 62121
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7194229960441589,
      "learning_rate": 0.0004986597862698842,
      "loss": 2.5144,
      "step": 62122
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7773524522781372,
      "learning_rate": 0.000498656721079596,
      "loss": 2.3366,
      "step": 62123
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7534148693084717,
      "learning_rate": 0.0004986536558523735,
      "loss": 2.0554,
      "step": 62124
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.8263247609138489,
      "learning_rate": 0.0004986505905882177,
      "loss": 2.455,
      "step": 62125
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7863152623176575,
      "learning_rate": 0.000498647525287129,
      "loss": 2.3329,
      "step": 62126
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7508148550987244,
      "learning_rate": 0.000498644459949108,
      "loss": 2.2825,
      "step": 62127
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7532596588134766,
      "learning_rate": 0.0004986413945741551,
      "loss": 2.2977,
      "step": 62128
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7876192927360535,
      "learning_rate": 0.000498638329162271,
      "loss": 2.4615,
      "step": 62129
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7603645324707031,
      "learning_rate": 0.0004986352637134565,
      "loss": 2.105,
      "step": 62130
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.6954923868179321,
      "learning_rate": 0.0004986321982277119,
      "loss": 2.0812,
      "step": 62131
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.9135246872901917,
      "learning_rate": 0.0004986291327050377,
      "loss": 2.1064,
      "step": 62132
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7478326559066772,
      "learning_rate": 0.0004986260671454347,
      "loss": 2.1167,
      "step": 62133
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7550222277641296,
      "learning_rate": 0.0004986230015489033,
      "loss": 2.1128,
      "step": 62134
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7046573758125305,
      "learning_rate": 0.0004986199359154442,
      "loss": 2.0898,
      "step": 62135
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7574611306190491,
      "learning_rate": 0.0004986168702450578,
      "loss": 2.2438,
      "step": 62136
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7484973669052124,
      "learning_rate": 0.0004986138045377449,
      "loss": 2.149,
      "step": 62137
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.6801048517227173,
      "learning_rate": 0.0004986107387935059,
      "loss": 2.0954,
      "step": 62138
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.8560289740562439,
      "learning_rate": 0.0004986076730123414,
      "loss": 2.4095,
      "step": 62139
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.923799455165863,
      "learning_rate": 0.000498604607194252,
      "loss": 2.1678,
      "step": 62140
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.8162537813186646,
      "learning_rate": 0.0004986015413392382,
      "loss": 1.8788,
      "step": 62141
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7530941963195801,
      "learning_rate": 0.0004985984754473006,
      "loss": 2.2412,
      "step": 62142
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.8562085628509521,
      "learning_rate": 0.0004985954095184399,
      "loss": 2.2979,
      "step": 62143
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7081500291824341,
      "learning_rate": 0.0004985923435526564,
      "loss": 2.1551,
      "step": 62144
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7888687252998352,
      "learning_rate": 0.000498589277549951,
      "loss": 2.2146,
      "step": 62145
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.8019643425941467,
      "learning_rate": 0.0004985862115103242,
      "loss": 2.2114,
      "step": 62146
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.9618145227432251,
      "learning_rate": 0.0004985831454337763,
      "loss": 2.1535,
      "step": 62147
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.6948022246360779,
      "learning_rate": 0.0004985800793203081,
      "loss": 2.2011,
      "step": 62148
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7184498906135559,
      "learning_rate": 0.0004985770131699201,
      "loss": 2.0304,
      "step": 62149
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7853310704231262,
      "learning_rate": 0.0004985739469826129,
      "loss": 2.0449,
      "step": 62150
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7561332583427429,
      "learning_rate": 0.0004985708807583871,
      "loss": 2.3736,
      "step": 62151
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7096980810165405,
      "learning_rate": 0.0004985678144972431,
      "loss": 2.2201,
      "step": 62152
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.685457706451416,
      "learning_rate": 0.0004985647481991816,
      "loss": 2.1681,
      "step": 62153
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7087190747261047,
      "learning_rate": 0.0004985616818642031,
      "loss": 2.1849,
      "step": 62154
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7577489614486694,
      "learning_rate": 0.0004985586154923086,
      "loss": 2.1535,
      "step": 62155
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7989669442176819,
      "learning_rate": 0.0004985555490834979,
      "loss": 2.0645,
      "step": 62156
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7708840370178223,
      "learning_rate": 0.0004985524826377722,
      "loss": 2.1866,
      "step": 62157
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7064999938011169,
      "learning_rate": 0.0004985494161551318,
      "loss": 2.0951,
      "step": 62158
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7612797021865845,
      "learning_rate": 0.0004985463496355773,
      "loss": 2.3081,
      "step": 62159
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7014559507369995,
      "learning_rate": 0.0004985432830791091,
      "loss": 2.2376,
      "step": 62160
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7289307117462158,
      "learning_rate": 0.0004985402164857282,
      "loss": 2.2857,
      "step": 62161
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.8427466154098511,
      "learning_rate": 0.0004985371498554349,
      "loss": 2.02,
      "step": 62162
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7413867712020874,
      "learning_rate": 0.0004985340831882298,
      "loss": 2.1958,
      "step": 62163
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7407271862030029,
      "learning_rate": 0.0004985310164841134,
      "loss": 2.1534,
      "step": 62164
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7364572286605835,
      "learning_rate": 0.0004985279497430864,
      "loss": 2.3031,
      "step": 62165
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7487402558326721,
      "learning_rate": 0.0004985248829651492,
      "loss": 2.4047,
      "step": 62166
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.743800938129425,
      "learning_rate": 0.0004985218161503026,
      "loss": 2.2482,
      "step": 62167
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7532355189323425,
      "learning_rate": 0.000498518749298547,
      "loss": 2.2959,
      "step": 62168
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7525659203529358,
      "learning_rate": 0.0004985156824098831,
      "loss": 2.2798,
      "step": 62169
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.8033823370933533,
      "learning_rate": 0.0004985126154843113,
      "loss": 2.1103,
      "step": 62170
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7301170825958252,
      "learning_rate": 0.0004985095485218322,
      "loss": 2.4818,
      "step": 62171
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7060425281524658,
      "learning_rate": 0.0004985064815224464,
      "loss": 2.448,
      "step": 62172
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7158012986183167,
      "learning_rate": 0.0004985034144861547,
      "loss": 2.4008,
      "step": 62173
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.8545445799827576,
      "learning_rate": 0.0004985003474129573,
      "loss": 2.37,
      "step": 62174
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7218827605247498,
      "learning_rate": 0.000498497280302855,
      "loss": 2.179,
      "step": 62175
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7477537989616394,
      "learning_rate": 0.0004984942131558484,
      "loss": 2.657,
      "step": 62176
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7432785034179688,
      "learning_rate": 0.0004984911459719378,
      "loss": 2.1698,
      "step": 62177
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.768447756767273,
      "learning_rate": 0.000498488078751124,
      "loss": 2.0543,
      "step": 62178
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7575252056121826,
      "learning_rate": 0.0004984850114934076,
      "loss": 2.2534,
      "step": 62179
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.9246416091918945,
      "learning_rate": 0.000498481944198789,
      "loss": 2.3416,
      "step": 62180
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7741051912307739,
      "learning_rate": 0.0004984788768672689,
      "loss": 2.3944,
      "step": 62181
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.770453155040741,
      "learning_rate": 0.0004984758094988478,
      "loss": 2.1525,
      "step": 62182
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.8585771918296814,
      "learning_rate": 0.0004984727420935263,
      "loss": 2.3517,
      "step": 62183
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7823264002799988,
      "learning_rate": 0.000498469674651305,
      "loss": 2.0899,
      "step": 62184
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.8802469372749329,
      "learning_rate": 0.0004984666071721845,
      "loss": 2.2376,
      "step": 62185
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.8831878900527954,
      "learning_rate": 0.0004984635396561652,
      "loss": 2.1005,
      "step": 62186
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7384616732597351,
      "learning_rate": 0.0004984604721032478,
      "loss": 2.2952,
      "step": 62187
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7944687604904175,
      "learning_rate": 0.000498457404513433,
      "loss": 2.1644,
      "step": 62188
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.8212593197822571,
      "learning_rate": 0.0004984543368867211,
      "loss": 2.2768,
      "step": 62189
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7667408585548401,
      "learning_rate": 0.0004984512692231128,
      "loss": 2.2153,
      "step": 62190
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7384552359580994,
      "learning_rate": 0.0004984482015226086,
      "loss": 2.3466,
      "step": 62191
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.8533931970596313,
      "learning_rate": 0.0004984451337852092,
      "loss": 2.1566,
      "step": 62192
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7050650715827942,
      "learning_rate": 0.0004984420660109151,
      "loss": 2.1809,
      "step": 62193
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7143837809562683,
      "learning_rate": 0.0004984389981997269,
      "loss": 2.0125,
      "step": 62194
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7322474718093872,
      "learning_rate": 0.0004984359303516451,
      "loss": 2.4105,
      "step": 62195
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7392427921295166,
      "learning_rate": 0.0004984328624666703,
      "loss": 2.1584,
      "step": 62196
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.73949134349823,
      "learning_rate": 0.0004984297945448032,
      "loss": 2.2193,
      "step": 62197
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7158046364784241,
      "learning_rate": 0.0004984267265860441,
      "loss": 2.2884,
      "step": 62198
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7161751389503479,
      "learning_rate": 0.0004984236585903938,
      "loss": 2.2651,
      "step": 62199
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.6698550581932068,
      "learning_rate": 0.0004984205905578528,
      "loss": 2.037,
      "step": 62200
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7524623274803162,
      "learning_rate": 0.0004984175224884217,
      "loss": 2.398,
      "step": 62201
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7086425423622131,
      "learning_rate": 0.0004984144543821009,
      "loss": 2.2586,
      "step": 62202
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7378370761871338,
      "learning_rate": 0.0004984113862388913,
      "loss": 2.3984,
      "step": 62203
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.786443829536438,
      "learning_rate": 0.0004984083180587932,
      "loss": 2.1974,
      "step": 62204
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.6911365985870361,
      "learning_rate": 0.0004984052498418071,
      "loss": 2.3491,
      "step": 62205
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7099218368530273,
      "learning_rate": 0.0004984021815879339,
      "loss": 2.142,
      "step": 62206
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.6839674115180969,
      "learning_rate": 0.000498399113297174,
      "loss": 2.2703,
      "step": 62207
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7568658590316772,
      "learning_rate": 0.0004983960449695279,
      "loss": 1.9856,
      "step": 62208
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7338373064994812,
      "learning_rate": 0.0004983929766049963,
      "loss": 2.1645,
      "step": 62209
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7060792446136475,
      "learning_rate": 0.0004983899082035795,
      "loss": 2.1593,
      "step": 62210
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7659592032432556,
      "learning_rate": 0.0004983868397652784,
      "loss": 2.3607,
      "step": 62211
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7887570858001709,
      "learning_rate": 0.0004983837712900935,
      "loss": 2.1203,
      "step": 62212
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.732811689376831,
      "learning_rate": 0.0004983807027780252,
      "loss": 2.2649,
      "step": 62213
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.794793426990509,
      "learning_rate": 0.0004983776342290742,
      "loss": 1.9741,
      "step": 62214
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.767597496509552,
      "learning_rate": 0.0004983745656432411,
      "loss": 2.245,
      "step": 62215
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.8151042461395264,
      "learning_rate": 0.0004983714970205265,
      "loss": 2.3005,
      "step": 62216
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.8054852485656738,
      "learning_rate": 0.0004983684283609307,
      "loss": 2.506,
      "step": 62217
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7160977125167847,
      "learning_rate": 0.0004983653596644546,
      "loss": 2.1304,
      "step": 62218
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7892366647720337,
      "learning_rate": 0.0004983622909310987,
      "loss": 2.0943,
      "step": 62219
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.910517692565918,
      "learning_rate": 0.0004983592221608634,
      "loss": 2.2294,
      "step": 62220
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.8580701947212219,
      "learning_rate": 0.0004983561533537493,
      "loss": 2.3414,
      "step": 62221
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7647243142127991,
      "learning_rate": 0.0004983530845097573,
      "loss": 2.2269,
      "step": 62222
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.812007486820221,
      "learning_rate": 0.0004983500156288875,
      "loss": 2.1417,
      "step": 62223
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.8189089298248291,
      "learning_rate": 0.0004983469467111407,
      "loss": 2.1644,
      "step": 62224
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.774249255657196,
      "learning_rate": 0.0004983438777565175,
      "loss": 1.9984,
      "step": 62225
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7830643057823181,
      "learning_rate": 0.0004983408087650185,
      "loss": 1.9187,
      "step": 62226
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7330114245414734,
      "learning_rate": 0.000498337739736644,
      "loss": 2.1593,
      "step": 62227
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.8214816451072693,
      "learning_rate": 0.000498334670671395,
      "loss": 2.1409,
      "step": 62228
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7353253960609436,
      "learning_rate": 0.0004983316015692717,
      "loss": 2.1533,
      "step": 62229
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.8312187790870667,
      "learning_rate": 0.0004983285324302748,
      "loss": 2.2408,
      "step": 62230
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7454996705055237,
      "learning_rate": 0.000498325463254405,
      "loss": 2.2276,
      "step": 62231
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7296329736709595,
      "learning_rate": 0.0004983223940416627,
      "loss": 2.4265,
      "step": 62232
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7749930024147034,
      "learning_rate": 0.0004983193247920485,
      "loss": 2.3465,
      "step": 62233
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.8411842584609985,
      "learning_rate": 0.0004983162555055631,
      "loss": 2.3656,
      "step": 62234
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7526919841766357,
      "learning_rate": 0.0004983131861822068,
      "loss": 2.3497,
      "step": 62235
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7111424803733826,
      "learning_rate": 0.0004983101168219804,
      "loss": 2.0212,
      "step": 62236
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7822645306587219,
      "learning_rate": 0.0004983070474248845,
      "loss": 2.4288,
      "step": 62237
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.9022179245948792,
      "learning_rate": 0.0004983039779909194,
      "loss": 2.3978,
      "step": 62238
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.8309987187385559,
      "learning_rate": 0.0004983009085200859,
      "loss": 2.3274,
      "step": 62239
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7796730995178223,
      "learning_rate": 0.0004982978390123846,
      "loss": 2.2102,
      "step": 62240
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.8535053133964539,
      "learning_rate": 0.0004982947694678159,
      "loss": 2.1737,
      "step": 62241
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7993288040161133,
      "learning_rate": 0.0004982916998863805,
      "loss": 2.5851,
      "step": 62242
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.8085610866546631,
      "learning_rate": 0.000498288630268079,
      "loss": 2.3714,
      "step": 62243
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.2895451784133911,
      "learning_rate": 0.0004982855606129118,
      "loss": 2.0388,
      "step": 62244
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.9789405465126038,
      "learning_rate": 0.0004982824909208796,
      "loss": 2.4085,
      "step": 62245
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7198681831359863,
      "learning_rate": 0.0004982794211919829,
      "loss": 2.2599,
      "step": 62246
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7511332035064697,
      "learning_rate": 0.0004982763514262224,
      "loss": 2.0268,
      "step": 62247
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7855101823806763,
      "learning_rate": 0.0004982732816235985,
      "loss": 2.3606,
      "step": 62248
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.8537614941596985,
      "learning_rate": 0.0004982702117841119,
      "loss": 2.1473,
      "step": 62249
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7354891896247864,
      "learning_rate": 0.0004982671419077631,
      "loss": 2.3473,
      "step": 62250
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7958183288574219,
      "learning_rate": 0.0004982640719945527,
      "loss": 2.0543,
      "step": 62251
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.8363153338432312,
      "learning_rate": 0.0004982610020444814,
      "loss": 2.145,
      "step": 62252
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7481881976127625,
      "learning_rate": 0.0004982579320575495,
      "loss": 2.1945,
      "step": 62253
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.8542085289955139,
      "learning_rate": 0.0004982548620337577,
      "loss": 2.1587,
      "step": 62254
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.8030229210853577,
      "learning_rate": 0.0004982517919731066,
      "loss": 2.3654,
      "step": 62255
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.80910724401474,
      "learning_rate": 0.0004982487218755968,
      "loss": 2.0777,
      "step": 62256
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7695891261100769,
      "learning_rate": 0.0004982456517412287,
      "loss": 2.3715,
      "step": 62257
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.8695881962776184,
      "learning_rate": 0.0004982425815700031,
      "loss": 2.2125,
      "step": 62258
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7101269364356995,
      "learning_rate": 0.0004982395113619204,
      "loss": 2.099,
      "step": 62259
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.8662653565406799,
      "learning_rate": 0.0004982364411169813,
      "loss": 2.2887,
      "step": 62260
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.663033664226532,
      "learning_rate": 0.0004982333708351863,
      "loss": 1.9865,
      "step": 62261
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7969525456428528,
      "learning_rate": 0.0004982303005165359,
      "loss": 2.3021,
      "step": 62262
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.919849157333374,
      "learning_rate": 0.0004982272301610308,
      "loss": 2.2051,
      "step": 62263
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.6718593239784241,
      "learning_rate": 0.0004982241597686715,
      "loss": 2.2648,
      "step": 62264
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7627784013748169,
      "learning_rate": 0.0004982210893394587,
      "loss": 2.2883,
      "step": 62265
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7815598249435425,
      "learning_rate": 0.0004982180188733926,
      "loss": 2.2168,
      "step": 62266
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.8688751459121704,
      "learning_rate": 0.0004982149483704742,
      "loss": 2.2041,
      "step": 62267
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7974397540092468,
      "learning_rate": 0.0004982118778307039,
      "loss": 2.2504,
      "step": 62268
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.9960431456565857,
      "learning_rate": 0.0004982088072540822,
      "loss": 2.2684,
      "step": 62269
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.8551209568977356,
      "learning_rate": 0.0004982057366406099,
      "loss": 2.1656,
      "step": 62270
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7617550492286682,
      "learning_rate": 0.0004982026659902872,
      "loss": 2.3544,
      "step": 62271
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7294229865074158,
      "learning_rate": 0.000498199595303115,
      "loss": 2.1809,
      "step": 62272
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.8162636160850525,
      "learning_rate": 0.0004981965245790937,
      "loss": 2.1781,
      "step": 62273
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7305495738983154,
      "learning_rate": 0.000498193453818224,
      "loss": 2.1132,
      "step": 62274
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.8422435522079468,
      "learning_rate": 0.0004981903830205063,
      "loss": 2.2089,
      "step": 62275
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.8307128548622131,
      "learning_rate": 0.0004981873121859414,
      "loss": 2.109,
      "step": 62276
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7425643801689148,
      "learning_rate": 0.0004981842413145296,
      "loss": 2.2673,
      "step": 62277
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7490310072898865,
      "learning_rate": 0.0004981811704062717,
      "loss": 2.1975,
      "step": 62278
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7251225709915161,
      "learning_rate": 0.0004981780994611682,
      "loss": 2.1645,
      "step": 62279
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.8040821552276611,
      "learning_rate": 0.0004981750284792196,
      "loss": 2.2864,
      "step": 62280
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.777073085308075,
      "learning_rate": 0.0004981719574604264,
      "loss": 2.1981,
      "step": 62281
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.8599653840065002,
      "learning_rate": 0.0004981688864047894,
      "loss": 2.5758,
      "step": 62282
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.9836894273757935,
      "learning_rate": 0.0004981658153123092,
      "loss": 2.377,
      "step": 62283
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.6933096647262573,
      "learning_rate": 0.000498162744182986,
      "loss": 2.0128,
      "step": 62284
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.8787983059883118,
      "learning_rate": 0.0004981596730168207,
      "loss": 2.1396,
      "step": 62285
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.8728542327880859,
      "learning_rate": 0.000498156601813814,
      "loss": 2.4113,
      "step": 62286
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7532218098640442,
      "learning_rate": 0.0004981535305739659,
      "loss": 2.1738,
      "step": 62287
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.8817452192306519,
      "learning_rate": 0.0004981504592972774,
      "loss": 2.168,
      "step": 62288
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.876410961151123,
      "learning_rate": 0.0004981473879837492,
      "loss": 2.0754,
      "step": 62289
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7856593728065491,
      "learning_rate": 0.0004981443166333814,
      "loss": 2.1878,
      "step": 62290
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7915080189704895,
      "learning_rate": 0.0004981412452461749,
      "loss": 2.1531,
      "step": 62291
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.9158337116241455,
      "learning_rate": 0.0004981381738221304,
      "loss": 2.1027,
      "step": 62292
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.8411141633987427,
      "learning_rate": 0.000498135102361248,
      "loss": 2.0758,
      "step": 62293
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.880896806716919,
      "learning_rate": 0.0004981320308635286,
      "loss": 2.3409,
      "step": 62294
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7577937245368958,
      "learning_rate": 0.0004981289593289728,
      "loss": 2.1928,
      "step": 62295
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7235469222068787,
      "learning_rate": 0.0004981258877575811,
      "loss": 2.3601,
      "step": 62296
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.6749518513679504,
      "learning_rate": 0.000498122816149354,
      "loss": 2.3479,
      "step": 62297
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.8370696902275085,
      "learning_rate": 0.0004981197445042922,
      "loss": 2.2043,
      "step": 62298
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7795836925506592,
      "learning_rate": 0.000498116672822396,
      "loss": 2.3481,
      "step": 62299
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7898898124694824,
      "learning_rate": 0.0004981136011036664,
      "loss": 2.1801,
      "step": 62300
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7319932579994202,
      "learning_rate": 0.0004981105293481037,
      "loss": 2.1548,
      "step": 62301
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.778789758682251,
      "learning_rate": 0.0004981074575557084,
      "loss": 2.1927,
      "step": 62302
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7740177512168884,
      "learning_rate": 0.0004981043857264813,
      "loss": 2.1838,
      "step": 62303
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.8303054571151733,
      "learning_rate": 0.0004981013138604228,
      "loss": 2.1333,
      "step": 62304
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.65840744972229,
      "learning_rate": 0.0004980982419575335,
      "loss": 2.0075,
      "step": 62305
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7959200739860535,
      "learning_rate": 0.000498095170017814,
      "loss": 2.5143,
      "step": 62306
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.706900954246521,
      "learning_rate": 0.000498092098041265,
      "loss": 2.2088,
      "step": 62307
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.9777035713195801,
      "learning_rate": 0.0004980890260278868,
      "loss": 2.1632,
      "step": 62308
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7567307353019714,
      "learning_rate": 0.0004980859539776802,
      "loss": 2.1225,
      "step": 62309
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7434558868408203,
      "learning_rate": 0.0004980828818906456,
      "loss": 2.1964,
      "step": 62310
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7761713862419128,
      "learning_rate": 0.0004980798097667837,
      "loss": 2.0949,
      "step": 62311
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7289303541183472,
      "learning_rate": 0.000498076737606095,
      "loss": 2.3224,
      "step": 62312
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7883101105690002,
      "learning_rate": 0.0004980736654085802,
      "loss": 2.1129,
      "step": 62313
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.830098032951355,
      "learning_rate": 0.0004980705931742396,
      "loss": 2.2387,
      "step": 62314
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.75706946849823,
      "learning_rate": 0.0004980675209030741,
      "loss": 2.3499,
      "step": 62315
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7914532423019409,
      "learning_rate": 0.0004980644485950839,
      "loss": 2.3511,
      "step": 62316
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.8255780339241028,
      "learning_rate": 0.0004980613762502699,
      "loss": 2.3867,
      "step": 62317
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.9241002798080444,
      "learning_rate": 0.0004980583038686326,
      "loss": 2.1212,
      "step": 62318
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7556757926940918,
      "learning_rate": 0.0004980552314501725,
      "loss": 2.1746,
      "step": 62319
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.9659226536750793,
      "learning_rate": 0.0004980521589948901,
      "loss": 2.0771,
      "step": 62320
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.9145845770835876,
      "learning_rate": 0.0004980490865027861,
      "loss": 2.2334,
      "step": 62321
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7969675660133362,
      "learning_rate": 0.0004980460139738612,
      "loss": 2.1313,
      "step": 62322
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.6871483325958252,
      "learning_rate": 0.0004980429414081156,
      "loss": 2.2713,
      "step": 62323
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7891595363616943,
      "learning_rate": 0.0004980398688055501,
      "loss": 2.2543,
      "step": 62324
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.8337429761886597,
      "learning_rate": 0.0004980367961661654,
      "loss": 2.1582,
      "step": 62325
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.6634337306022644,
      "learning_rate": 0.0004980337234899619,
      "loss": 2.1791,
      "step": 62326
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7263674139976501,
      "learning_rate": 0.00049803065077694,
      "loss": 2.321,
      "step": 62327
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.8063749670982361,
      "learning_rate": 0.0004980275780271006,
      "loss": 2.3206,
      "step": 62328
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7949038147926331,
      "learning_rate": 0.0004980245052404442,
      "loss": 2.0539,
      "step": 62329
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7913969159126282,
      "learning_rate": 0.0004980214324169711,
      "loss": 2.1741,
      "step": 62330
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7114205360412598,
      "learning_rate": 0.0004980183595566823,
      "loss": 2.2083,
      "step": 62331
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.8094393014907837,
      "learning_rate": 0.000498015286659578,
      "loss": 2.3211,
      "step": 62332
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7802622318267822,
      "learning_rate": 0.000498012213725659,
      "loss": 2.3984,
      "step": 62333
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7951688766479492,
      "learning_rate": 0.0004980091407549257,
      "loss": 2.2258,
      "step": 62334
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7796964049339294,
      "learning_rate": 0.0004980060677473789,
      "loss": 1.9607,
      "step": 62335
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7443965673446655,
      "learning_rate": 0.0004980029947030191,
      "loss": 2.1009,
      "step": 62336
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.9041691422462463,
      "learning_rate": 0.0004979999216218466,
      "loss": 2.2883,
      "step": 62337
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7561075687408447,
      "learning_rate": 0.0004979968485038622,
      "loss": 2.1552,
      "step": 62338
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.8231015205383301,
      "learning_rate": 0.0004979937753490665,
      "loss": 1.9168,
      "step": 62339
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.9157041311264038,
      "learning_rate": 0.00049799070215746,
      "loss": 2.3477,
      "step": 62340
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7464309930801392,
      "learning_rate": 0.0004979876289290433,
      "loss": 1.8523,
      "step": 62341
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.789057195186615,
      "learning_rate": 0.000497984555663817,
      "loss": 2.2255,
      "step": 62342
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.927041232585907,
      "learning_rate": 0.0004979814823617816,
      "loss": 2.3246,
      "step": 62343
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.8528566360473633,
      "learning_rate": 0.0004979784090229377,
      "loss": 2.1616,
      "step": 62344
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.6907870769500732,
      "learning_rate": 0.0004979753356472859,
      "loss": 2.1369,
      "step": 62345
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.9358710646629333,
      "learning_rate": 0.0004979722622348267,
      "loss": 2.3687,
      "step": 62346
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.8903513550758362,
      "learning_rate": 0.0004979691887855608,
      "loss": 2.306,
      "step": 62347
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7355121970176697,
      "learning_rate": 0.0004979661152994886,
      "loss": 2.383,
      "step": 62348
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.8420038223266602,
      "learning_rate": 0.0004979630417766108,
      "loss": 2.5714,
      "step": 62349
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7337455749511719,
      "learning_rate": 0.0004979599682169279,
      "loss": 2.0974,
      "step": 62350
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7548593878746033,
      "learning_rate": 0.0004979568946204406,
      "loss": 2.2908,
      "step": 62351
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7157658934593201,
      "learning_rate": 0.0004979538209871493,
      "loss": 2.3781,
      "step": 62352
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.0122202634811401,
      "learning_rate": 0.0004979507473170547,
      "loss": 2.3525,
      "step": 62353
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7878209352493286,
      "learning_rate": 0.0004979476736101573,
      "loss": 2.3476,
      "step": 62354
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.9108692407608032,
      "learning_rate": 0.0004979445998664576,
      "loss": 2.1267,
      "step": 62355
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7216756939888,
      "learning_rate": 0.0004979415260859564,
      "loss": 2.3462,
      "step": 62356
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.8787583708763123,
      "learning_rate": 0.000497938452268654,
      "loss": 2.0811,
      "step": 62357
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7944505214691162,
      "learning_rate": 0.0004979353784145512,
      "loss": 2.1824,
      "step": 62358
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7216855883598328,
      "learning_rate": 0.0004979323045236486,
      "loss": 2.2327,
      "step": 62359
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7526002526283264,
      "learning_rate": 0.0004979292305959465,
      "loss": 2.4516,
      "step": 62360
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.788517951965332,
      "learning_rate": 0.0004979261566314454,
      "loss": 2.4656,
      "step": 62361
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7457001209259033,
      "learning_rate": 0.0004979230826301465,
      "loss": 2.0535,
      "step": 62362
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7061965465545654,
      "learning_rate": 0.0004979200085920497,
      "loss": 2.5301,
      "step": 62363
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7349310517311096,
      "learning_rate": 0.0004979169345171559,
      "loss": 2.0278,
      "step": 62364
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.8057346940040588,
      "learning_rate": 0.0004979138604054656,
      "loss": 2.3899,
      "step": 62365
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7605145573616028,
      "learning_rate": 0.0004979107862569795,
      "loss": 2.2324,
      "step": 62366
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.3634098768234253,
      "learning_rate": 0.000497907712071698,
      "loss": 2.3164,
      "step": 62367
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7229921817779541,
      "learning_rate": 0.0004979046378496215,
      "loss": 2.264,
      "step": 62368
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.8107482194900513,
      "learning_rate": 0.000497901563590751,
      "loss": 2.1162,
      "step": 62369
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7712026834487915,
      "learning_rate": 0.0004978984892950869,
      "loss": 2.1219,
      "step": 62370
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7151191830635071,
      "learning_rate": 0.0004978954149626296,
      "loss": 2.1549,
      "step": 62371
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7482664585113525,
      "learning_rate": 0.0004978923405933799,
      "loss": 2.2808,
      "step": 62372
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.6846510171890259,
      "learning_rate": 0.0004978892661873383,
      "loss": 2.1468,
      "step": 62373
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7358240485191345,
      "learning_rate": 0.0004978861917445053,
      "loss": 1.9745,
      "step": 62374
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7399693727493286,
      "learning_rate": 0.0004978831172648815,
      "loss": 2.5549,
      "step": 62375
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7598334550857544,
      "learning_rate": 0.0004978800427484674,
      "loss": 2.0134,
      "step": 62376
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7387147545814514,
      "learning_rate": 0.0004978769681952638,
      "loss": 2.2158,
      "step": 62377
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.8892830610275269,
      "learning_rate": 0.0004978738936052711,
      "loss": 2.094,
      "step": 62378
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7425503134727478,
      "learning_rate": 0.0004978708189784899,
      "loss": 2.2542,
      "step": 62379
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7336648106575012,
      "learning_rate": 0.0004978677443149208,
      "loss": 2.1548,
      "step": 62380
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.8490574359893799,
      "learning_rate": 0.0004978646696145644,
      "loss": 2.2309,
      "step": 62381
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.8101826906204224,
      "learning_rate": 0.0004978615948774211,
      "loss": 2.3755,
      "step": 62382
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7046977877616882,
      "learning_rate": 0.0004978585201034917,
      "loss": 2.2901,
      "step": 62383
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.8108331561088562,
      "learning_rate": 0.0004978554452927767,
      "loss": 2.3513,
      "step": 62384
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7983055710792542,
      "learning_rate": 0.0004978523704452766,
      "loss": 2.2265,
      "step": 62385
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7149854302406311,
      "learning_rate": 0.0004978492955609919,
      "loss": 2.2416,
      "step": 62386
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.8664268851280212,
      "learning_rate": 0.0004978462206399234,
      "loss": 2.1813,
      "step": 62387
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.908755898475647,
      "learning_rate": 0.0004978431456820715,
      "loss": 2.2334,
      "step": 62388
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.8738929629325867,
      "learning_rate": 0.0004978400706874369,
      "loss": 2.301,
      "step": 62389
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7847867608070374,
      "learning_rate": 0.0004978369956560199,
      "loss": 2.2282,
      "step": 62390
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.8014864921569824,
      "learning_rate": 0.0004978339205878214,
      "loss": 2.2521,
      "step": 62391
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.9227964282035828,
      "learning_rate": 0.0004978308454828419,
      "loss": 2.1271,
      "step": 62392
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7347700595855713,
      "learning_rate": 0.0004978277703410818,
      "loss": 2.2019,
      "step": 62393
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7347928881645203,
      "learning_rate": 0.0004978246951625418,
      "loss": 1.9978,
      "step": 62394
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7119839787483215,
      "learning_rate": 0.0004978216199472225,
      "loss": 2.2293,
      "step": 62395
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7807419896125793,
      "learning_rate": 0.0004978185446951244,
      "loss": 2.1431,
      "step": 62396
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7326532602310181,
      "learning_rate": 0.0004978154694062481,
      "loss": 2.3063,
      "step": 62397
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.8196972608566284,
      "learning_rate": 0.0004978123940805943,
      "loss": 2.2216,
      "step": 62398
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.766137421131134,
      "learning_rate": 0.0004978093187181633,
      "loss": 2.0907,
      "step": 62399
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7294051051139832,
      "learning_rate": 0.0004978062433189559,
      "loss": 2.2399,
      "step": 62400
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.9863725900650024,
      "learning_rate": 0.0004978031678829725,
      "loss": 2.1348,
      "step": 62401
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7328945994377136,
      "learning_rate": 0.0004978000924102138,
      "loss": 2.2064,
      "step": 62402
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.8170346021652222,
      "learning_rate": 0.0004977970169006803,
      "loss": 2.1155,
      "step": 62403
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.88303142786026,
      "learning_rate": 0.0004977939413543727,
      "loss": 2.2045,
      "step": 62404
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.8496938347816467,
      "learning_rate": 0.0004977908657712913,
      "loss": 2.2253,
      "step": 62405
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.8161496520042419,
      "learning_rate": 0.0004977877901514369,
      "loss": 2.5103,
      "step": 62406
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.8157566785812378,
      "learning_rate": 0.00049778471449481,
      "loss": 2.1552,
      "step": 62407
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7801796793937683,
      "learning_rate": 0.0004977816388014113,
      "loss": 2.1278,
      "step": 62408
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7363506555557251,
      "learning_rate": 0.0004977785630712412,
      "loss": 2.3769,
      "step": 62409
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.8790420293807983,
      "learning_rate": 0.0004977754873043003,
      "loss": 2.2475,
      "step": 62410
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.8318389058113098,
      "learning_rate": 0.0004977724115005893,
      "loss": 2.2274,
      "step": 62411
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.6912009119987488,
      "learning_rate": 0.0004977693356601086,
      "loss": 2.0154,
      "step": 62412
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.8018290400505066,
      "learning_rate": 0.0004977662597828589,
      "loss": 2.303,
      "step": 62413
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7987655997276306,
      "learning_rate": 0.0004977631838688406,
      "loss": 2.1051,
      "step": 62414
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7899880409240723,
      "learning_rate": 0.0004977601079180545,
      "loss": 2.4529,
      "step": 62415
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.8636599183082581,
      "learning_rate": 0.000497757031930501,
      "loss": 2.6528,
      "step": 62416
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.8679744601249695,
      "learning_rate": 0.0004977539559061807,
      "loss": 2.1529,
      "step": 62417
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7329338788986206,
      "learning_rate": 0.0004977508798450944,
      "loss": 2.2262,
      "step": 62418
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.8235815763473511,
      "learning_rate": 0.0004977478037472423,
      "loss": 2.2783,
      "step": 62419
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.8406043648719788,
      "learning_rate": 0.0004977447276126251,
      "loss": 2.4445,
      "step": 62420
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7160636782646179,
      "learning_rate": 0.0004977416514412436,
      "loss": 2.2582,
      "step": 62421
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7168960571289062,
      "learning_rate": 0.0004977385752330981,
      "loss": 2.2441,
      "step": 62422
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.8102700710296631,
      "learning_rate": 0.0004977354989881893,
      "loss": 2.1003,
      "step": 62423
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.8882349729537964,
      "learning_rate": 0.0004977324227065177,
      "loss": 2.2153,
      "step": 62424
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7690083384513855,
      "learning_rate": 0.0004977293463880839,
      "loss": 2.1855,
      "step": 62425
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7588900327682495,
      "learning_rate": 0.0004977262700328885,
      "loss": 2.2972,
      "step": 62426
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.8967985510826111,
      "learning_rate": 0.0004977231936409321,
      "loss": 2.3319,
      "step": 62427
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7088025212287903,
      "learning_rate": 0.0004977201172122152,
      "loss": 2.0863,
      "step": 62428
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7068694829940796,
      "learning_rate": 0.0004977170407467384,
      "loss": 1.9706,
      "step": 62429
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.6761420965194702,
      "learning_rate": 0.0004977139642445023,
      "loss": 2.1,
      "step": 62430
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7596718072891235,
      "learning_rate": 0.0004977108877055074,
      "loss": 2.0701,
      "step": 62431
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7589666247367859,
      "learning_rate": 0.0004977078111297542,
      "loss": 2.2084,
      "step": 62432
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7398800849914551,
      "learning_rate": 0.0004977047345172435,
      "loss": 2.2345,
      "step": 62433
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7579885125160217,
      "learning_rate": 0.0004977016578679758,
      "loss": 2.28,
      "step": 62434
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7835178971290588,
      "learning_rate": 0.0004976985811819516,
      "loss": 2.3022,
      "step": 62435
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7093747854232788,
      "learning_rate": 0.0004976955044591715,
      "loss": 2.1757,
      "step": 62436
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7420706152915955,
      "learning_rate": 0.0004976924276996361,
      "loss": 2.1149,
      "step": 62437
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7206434011459351,
      "learning_rate": 0.0004976893509033458,
      "loss": 2.1264,
      "step": 62438
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7356576323509216,
      "learning_rate": 0.0004976862740703015,
      "loss": 2.1616,
      "step": 62439
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7272855639457703,
      "learning_rate": 0.0004976831972005035,
      "loss": 2.0042,
      "step": 62440
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7418926954269409,
      "learning_rate": 0.0004976801202939526,
      "loss": 2.244,
      "step": 62441
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7133069038391113,
      "learning_rate": 0.000497677043350649,
      "loss": 2.1656,
      "step": 62442
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.6951408386230469,
      "learning_rate": 0.0004976739663705937,
      "loss": 2.0201,
      "step": 62443
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7455477714538574,
      "learning_rate": 0.000497670889353787,
      "loss": 2.0775,
      "step": 62444
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7470638751983643,
      "learning_rate": 0.0004976678123002295,
      "loss": 2.4003,
      "step": 62445
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7385239601135254,
      "learning_rate": 0.0004976647352099218,
      "loss": 2.114,
      "step": 62446
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7602105736732483,
      "learning_rate": 0.0004976616580828646,
      "loss": 2.3513,
      "step": 62447
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.8563241362571716,
      "learning_rate": 0.0004976585809190583,
      "loss": 2.2804,
      "step": 62448
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7407323718070984,
      "learning_rate": 0.0004976555037185036,
      "loss": 2.3159,
      "step": 62449
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7115494608879089,
      "learning_rate": 0.000497652426481201,
      "loss": 2.2908,
      "step": 62450
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.848479151725769,
      "learning_rate": 0.0004976493492071511,
      "loss": 2.0567,
      "step": 62451
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7805595993995667,
      "learning_rate": 0.0004976462718963544,
      "loss": 2.2129,
      "step": 62452
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7576457858085632,
      "learning_rate": 0.0004976431945488116,
      "loss": 2.1159,
      "step": 62453
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7901715040206909,
      "learning_rate": 0.000497640117164523,
      "loss": 2.1948,
      "step": 62454
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7977126836776733,
      "learning_rate": 0.0004976370397434896,
      "loss": 2.2441,
      "step": 62455
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.8038540482521057,
      "learning_rate": 0.0004976339622857117,
      "loss": 2.162,
      "step": 62456
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7954739928245544,
      "learning_rate": 0.0004976308847911898,
      "loss": 2.2168,
      "step": 62457
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7805820107460022,
      "learning_rate": 0.0004976278072599248,
      "loss": 2.2689,
      "step": 62458
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7481989860534668,
      "learning_rate": 0.0004976247296919169,
      "loss": 2.1682,
      "step": 62459
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7707715034484863,
      "learning_rate": 0.0004976216520871669,
      "loss": 2.2663,
      "step": 62460
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7433347105979919,
      "learning_rate": 0.0004976185744456752,
      "loss": 2.2728,
      "step": 62461
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7589489221572876,
      "learning_rate": 0.0004976154967674425,
      "loss": 2.4214,
      "step": 62462
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7969602942466736,
      "learning_rate": 0.0004976124190524694,
      "loss": 2.2833,
      "step": 62463
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7408697009086609,
      "learning_rate": 0.0004976093413007564,
      "loss": 2.2525,
      "step": 62464
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7130370736122131,
      "learning_rate": 0.0004976062635123041,
      "loss": 2.2645,
      "step": 62465
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7331039905548096,
      "learning_rate": 0.000497603185687113,
      "loss": 2.2346,
      "step": 62466
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7630540132522583,
      "learning_rate": 0.0004976001078251838,
      "loss": 2.1283,
      "step": 62467
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.6943524479866028,
      "learning_rate": 0.0004975970299265171,
      "loss": 2.096,
      "step": 62468
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.9030941128730774,
      "learning_rate": 0.0004975939519911134,
      "loss": 2.1265,
      "step": 62469
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7775512933731079,
      "learning_rate": 0.0004975908740189731,
      "loss": 2.381,
      "step": 62470
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7098486423492432,
      "learning_rate": 0.000497587796010097,
      "loss": 2.1398,
      "step": 62471
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7181364297866821,
      "learning_rate": 0.0004975847179644855,
      "loss": 2.1999,
      "step": 62472
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7509859204292297,
      "learning_rate": 0.0004975816398821393,
      "loss": 2.1864,
      "step": 62473
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7455103993415833,
      "learning_rate": 0.0004975785617630591,
      "loss": 1.9813,
      "step": 62474
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7418569326400757,
      "learning_rate": 0.0004975754836072452,
      "loss": 2.2145,
      "step": 62475
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7543975114822388,
      "learning_rate": 0.0004975724054146983,
      "loss": 2.0934,
      "step": 62476
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7820721864700317,
      "learning_rate": 0.000497569327185419,
      "loss": 2.3743,
      "step": 62477
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7422387599945068,
      "learning_rate": 0.0004975662489194077,
      "loss": 2.0648,
      "step": 62478
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7245340347290039,
      "learning_rate": 0.0004975631706166652,
      "loss": 2.2688,
      "step": 62479
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.8482484221458435,
      "learning_rate": 0.000497560092277192,
      "loss": 2.2507,
      "step": 62480
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.725935697555542,
      "learning_rate": 0.0004975570139009887,
      "loss": 2.2044,
      "step": 62481
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.8476809859275818,
      "learning_rate": 0.0004975539354880556,
      "loss": 2.2635,
      "step": 62482
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.9032484292984009,
      "learning_rate": 0.0004975508570383937,
      "loss": 2.2363,
      "step": 62483
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.9123479127883911,
      "learning_rate": 0.0004975477785520033,
      "loss": 2.3517,
      "step": 62484
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.8638049960136414,
      "learning_rate": 0.0004975447000288851,
      "loss": 2.3021,
      "step": 62485
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7513119578361511,
      "learning_rate": 0.0004975416214690395,
      "loss": 2.2479,
      "step": 62486
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7873213887214661,
      "learning_rate": 0.0004975385428724673,
      "loss": 2.0863,
      "step": 62487
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7032732963562012,
      "learning_rate": 0.0004975354642391688,
      "loss": 2.1295,
      "step": 62488
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7188178896903992,
      "learning_rate": 0.0004975323855691448,
      "loss": 2.3923,
      "step": 62489
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.77675461769104,
      "learning_rate": 0.0004975293068623959,
      "loss": 2.2446,
      "step": 62490
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7427288889884949,
      "learning_rate": 0.0004975262281189225,
      "loss": 2.098,
      "step": 62491
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7957399487495422,
      "learning_rate": 0.0004975231493387252,
      "loss": 2.3293,
      "step": 62492
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.6542432904243469,
      "learning_rate": 0.0004975200705218048,
      "loss": 2.3855,
      "step": 62493
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7200129628181458,
      "learning_rate": 0.0004975169916681614,
      "loss": 2.2463,
      "step": 62494
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7604036331176758,
      "learning_rate": 0.000497513912777796,
      "loss": 2.1205,
      "step": 62495
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7558084726333618,
      "learning_rate": 0.0004975108338507091,
      "loss": 2.3126,
      "step": 62496
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.8293923735618591,
      "learning_rate": 0.0004975077548869011,
      "loss": 2.2669,
      "step": 62497
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7239786982536316,
      "learning_rate": 0.0004975046758863729,
      "loss": 2.2571,
      "step": 62498
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.6946085095405579,
      "learning_rate": 0.0004975015968491246,
      "loss": 2.1415,
      "step": 62499
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.6965973377227783,
      "learning_rate": 0.000497498517775157,
      "loss": 2.2111,
      "step": 62500
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7189903855323792,
      "learning_rate": 0.0004974954386644709,
      "loss": 2.207,
      "step": 62501
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.6817333102226257,
      "learning_rate": 0.0004974923595170666,
      "loss": 2.0891,
      "step": 62502
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7396648526191711,
      "learning_rate": 0.0004974892803329447,
      "loss": 2.4189,
      "step": 62503
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7275152802467346,
      "learning_rate": 0.0004974862011121059,
      "loss": 2.2615,
      "step": 62504
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7236860990524292,
      "learning_rate": 0.0004974831218545506,
      "loss": 2.0284,
      "step": 62505
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7795505523681641,
      "learning_rate": 0.0004974800425602795,
      "loss": 2.3746,
      "step": 62506
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.9501764178276062,
      "learning_rate": 0.000497476963229293,
      "loss": 2.3792,
      "step": 62507
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.722213089466095,
      "learning_rate": 0.0004974738838615921,
      "loss": 2.1833,
      "step": 62508
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7702919840812683,
      "learning_rate": 0.0004974708044571768,
      "loss": 2.1873,
      "step": 62509
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.6719517707824707,
      "learning_rate": 0.0004974677250160481,
      "loss": 2.2891,
      "step": 62510
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.8710412383079529,
      "learning_rate": 0.0004974646455382064,
      "loss": 2.1351,
      "step": 62511
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.6467289924621582,
      "learning_rate": 0.0004974615660236523,
      "loss": 2.2089,
      "step": 62512
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7817180156707764,
      "learning_rate": 0.0004974584864723862,
      "loss": 2.0543,
      "step": 62513
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7169554233551025,
      "learning_rate": 0.000497455406884409,
      "loss": 2.3441,
      "step": 62514
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7449678778648376,
      "learning_rate": 0.0004974523272597212,
      "loss": 2.1477,
      "step": 62515
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7364388704299927,
      "learning_rate": 0.000497449247598323,
      "loss": 2.3654,
      "step": 62516
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7152236104011536,
      "learning_rate": 0.0004974461679002156,
      "loss": 2.2676,
      "step": 62517
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7872355580329895,
      "learning_rate": 0.0004974430881653989,
      "loss": 2.284,
      "step": 62518
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7913883328437805,
      "learning_rate": 0.0004974400083938741,
      "loss": 2.2647,
      "step": 62519
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7682323455810547,
      "learning_rate": 0.0004974369285856413,
      "loss": 2.158,
      "step": 62520
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7542040944099426,
      "learning_rate": 0.0004974338487407012,
      "loss": 2.3523,
      "step": 62521
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7036900520324707,
      "learning_rate": 0.0004974307688590545,
      "loss": 2.2338,
      "step": 62522
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7801982760429382,
      "learning_rate": 0.0004974276889407018,
      "loss": 1.9999,
      "step": 62523
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7170746922492981,
      "learning_rate": 0.0004974246089856435,
      "loss": 2.2173,
      "step": 62524
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.729972779750824,
      "learning_rate": 0.0004974215289938801,
      "loss": 2.1442,
      "step": 62525
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7003674507141113,
      "learning_rate": 0.0004974184489654125,
      "loss": 2.0726,
      "step": 62526
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.6839017271995544,
      "learning_rate": 0.0004974153689002409,
      "loss": 2.2802,
      "step": 62527
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7289875149726868,
      "learning_rate": 0.0004974122887983662,
      "loss": 2.1323,
      "step": 62528
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.0037140846252441,
      "learning_rate": 0.0004974092086597889,
      "loss": 2.3578,
      "step": 62529
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7192815542221069,
      "learning_rate": 0.0004974061284845093,
      "loss": 2.1997,
      "step": 62530
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7760661840438843,
      "learning_rate": 0.0004974030482725282,
      "loss": 2.1278,
      "step": 62531
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7986673712730408,
      "learning_rate": 0.0004973999680238463,
      "loss": 2.0945,
      "step": 62532
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.840017557144165,
      "learning_rate": 0.0004973968877384639,
      "loss": 2.3835,
      "step": 62533
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7144535183906555,
      "learning_rate": 0.0004973938074163817,
      "loss": 2.3414,
      "step": 62534
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.80156010389328,
      "learning_rate": 0.0004973907270576004,
      "loss": 2.2712,
      "step": 62535
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7245239615440369,
      "learning_rate": 0.0004973876466621202,
      "loss": 2.3706,
      "step": 62536
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7352173328399658,
      "learning_rate": 0.0004973845662299421,
      "loss": 2.2567,
      "step": 62537
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7529153227806091,
      "learning_rate": 0.0004973814857610664,
      "loss": 2.4435,
      "step": 62538
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7537885904312134,
      "learning_rate": 0.0004973784052554938,
      "loss": 2.3801,
      "step": 62539
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7396341562271118,
      "learning_rate": 0.0004973753247132247,
      "loss": 2.1433,
      "step": 62540
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.722648561000824,
      "learning_rate": 0.0004973722441342601,
      "loss": 2.3853,
      "step": 62541
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7266688346862793,
      "learning_rate": 0.0004973691635186,
      "loss": 2.3134,
      "step": 62542
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.6781589984893799,
      "learning_rate": 0.0004973660828662453,
      "loss": 2.3272,
      "step": 62543
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7255714535713196,
      "learning_rate": 0.0004973630021771966,
      "loss": 2.2071,
      "step": 62544
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7708664536476135,
      "learning_rate": 0.0004973599214514543,
      "loss": 1.9916,
      "step": 62545
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.8006556630134583,
      "learning_rate": 0.0004973568406890191,
      "loss": 2.2873,
      "step": 62546
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7305904030799866,
      "learning_rate": 0.0004973537598898915,
      "loss": 2.2077,
      "step": 62547
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7662975192070007,
      "learning_rate": 0.0004973506790540722,
      "loss": 2.1592,
      "step": 62548
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.734287440776825,
      "learning_rate": 0.0004973475981815616,
      "loss": 2.2505,
      "step": 62549
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7233109474182129,
      "learning_rate": 0.0004973445172723604,
      "loss": 2.3581,
      "step": 62550
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.8164492845535278,
      "learning_rate": 0.0004973414363264692,
      "loss": 2.2772,
      "step": 62551
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.8011487126350403,
      "learning_rate": 0.0004973383553438884,
      "loss": 2.2769,
      "step": 62552
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.8211161494255066,
      "learning_rate": 0.0004973352743246187,
      "loss": 2.2344,
      "step": 62553
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.814367413520813,
      "learning_rate": 0.0004973321932686606,
      "loss": 2.0972,
      "step": 62554
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.8061786890029907,
      "learning_rate": 0.0004973291121760147,
      "loss": 2.2213,
      "step": 62555
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.9230412244796753,
      "learning_rate": 0.0004973260310466817,
      "loss": 2.5743,
      "step": 62556
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.6992638111114502,
      "learning_rate": 0.0004973229498806621,
      "loss": 2.0756,
      "step": 62557
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7095440626144409,
      "learning_rate": 0.0004973198686779563,
      "loss": 2.2205,
      "step": 62558
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.6805518269538879,
      "learning_rate": 0.000497316787438565,
      "loss": 2.1889,
      "step": 62559
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7424430847167969,
      "learning_rate": 0.0004973137061624889,
      "loss": 2.1641,
      "step": 62560
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.8136159181594849,
      "learning_rate": 0.0004973106248497285,
      "loss": 2.3497,
      "step": 62561
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7048465609550476,
      "learning_rate": 0.0004973075435002843,
      "loss": 2.1244,
      "step": 62562
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7542029023170471,
      "learning_rate": 0.0004973044621141568,
      "loss": 2.145,
      "step": 62563
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7281081676483154,
      "learning_rate": 0.0004973013806913466,
      "loss": 2.3348,
      "step": 62564
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7250291705131531,
      "learning_rate": 0.0004972982992318545,
      "loss": 2.1471,
      "step": 62565
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7760154604911804,
      "learning_rate": 0.000497295217735681,
      "loss": 2.2885,
      "step": 62566
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.8113784193992615,
      "learning_rate": 0.0004972921362028265,
      "loss": 2.4093,
      "step": 62567
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.8389453291893005,
      "learning_rate": 0.0004972890546332916,
      "loss": 2.4391,
      "step": 62568
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7269315719604492,
      "learning_rate": 0.0004972859730270771,
      "loss": 2.0628,
      "step": 62569
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7934104800224304,
      "learning_rate": 0.0004972828913841832,
      "loss": 2.2323,
      "step": 62570
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7823108434677124,
      "learning_rate": 0.0004972798097046108,
      "loss": 2.341,
      "step": 62571
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7919982671737671,
      "learning_rate": 0.0004972767279883604,
      "loss": 2.2673,
      "step": 62572
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7331648468971252,
      "learning_rate": 0.0004972736462354325,
      "loss": 2.287,
      "step": 62573
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7128674983978271,
      "learning_rate": 0.0004972705644458277,
      "loss": 2.2084,
      "step": 62574
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.6995456218719482,
      "learning_rate": 0.0004972674826195466,
      "loss": 2.3387,
      "step": 62575
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.753526508808136,
      "learning_rate": 0.0004972644007565897,
      "loss": 2.309,
      "step": 62576
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.8323063254356384,
      "learning_rate": 0.0004972613188569576,
      "loss": 2.4136,
      "step": 62577
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7189711332321167,
      "learning_rate": 0.0004972582369206511,
      "loss": 2.0058,
      "step": 62578
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7249379754066467,
      "learning_rate": 0.0004972551549476703,
      "loss": 2.2751,
      "step": 62579
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7509508728981018,
      "learning_rate": 0.0004972520729380161,
      "loss": 2.2043,
      "step": 62580
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.8174450397491455,
      "learning_rate": 0.0004972489908916891,
      "loss": 2.1383,
      "step": 62581
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7469733953475952,
      "learning_rate": 0.0004972459088086898,
      "loss": 2.1661,
      "step": 62582
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.775874674320221,
      "learning_rate": 0.0004972428266890186,
      "loss": 2.3111,
      "step": 62583
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.8616536259651184,
      "learning_rate": 0.0004972397445326765,
      "loss": 2.1785,
      "step": 62584
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7758477330207825,
      "learning_rate": 0.0004972366623396637,
      "loss": 2.2365,
      "step": 62585
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7936393022537231,
      "learning_rate": 0.0004972335801099808,
      "loss": 2.2929,
      "step": 62586
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.8525498509407043,
      "learning_rate": 0.0004972304978436285,
      "loss": 2.2538,
      "step": 62587
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.0005323886871338,
      "learning_rate": 0.0004972274155406074,
      "loss": 2.2467,
      "step": 62588
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.8490943908691406,
      "learning_rate": 0.0004972243332009179,
      "loss": 2.0143,
      "step": 62589
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7562013864517212,
      "learning_rate": 0.0004972212508245607,
      "loss": 2.4733,
      "step": 62590
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7581154108047485,
      "learning_rate": 0.0004972181684115364,
      "loss": 2.1679,
      "step": 62591
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7462066411972046,
      "learning_rate": 0.0004972150859618455,
      "loss": 2.3046,
      "step": 62592
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7739698886871338,
      "learning_rate": 0.0004972120034754886,
      "loss": 2.425,
      "step": 62593
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7636221051216125,
      "learning_rate": 0.0004972089209524662,
      "loss": 2.2866,
      "step": 62594
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7850785851478577,
      "learning_rate": 0.0004972058383927789,
      "loss": 2.1398,
      "step": 62595
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7314202785491943,
      "learning_rate": 0.0004972027557964274,
      "loss": 2.1021,
      "step": 62596
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7410023212432861,
      "learning_rate": 0.0004971996731634122,
      "loss": 2.2398,
      "step": 62597
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7715020775794983,
      "learning_rate": 0.0004971965904937339,
      "loss": 2.206,
      "step": 62598
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7597949504852295,
      "learning_rate": 0.0004971935077873928,
      "loss": 2.2541,
      "step": 62599
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7252067923545837,
      "learning_rate": 0.00049719042504439,
      "loss": 2.05,
      "step": 62600
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7827526330947876,
      "learning_rate": 0.0004971873422647256,
      "loss": 2.2811,
      "step": 62601
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7530033588409424,
      "learning_rate": 0.0004971842594484004,
      "loss": 2.2195,
      "step": 62602
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7298858165740967,
      "learning_rate": 0.0004971811765954149,
      "loss": 2.3094,
      "step": 62603
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.6923043727874756,
      "learning_rate": 0.0004971780937057697,
      "loss": 2.172,
      "step": 62604
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7257842421531677,
      "learning_rate": 0.0004971750107794654,
      "loss": 2.059,
      "step": 62605
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.6655833125114441,
      "learning_rate": 0.0004971719278165025,
      "loss": 1.9546,
      "step": 62606
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7320409417152405,
      "learning_rate": 0.0004971688448168817,
      "loss": 2.1612,
      "step": 62607
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7879209518432617,
      "learning_rate": 0.0004971657617806035,
      "loss": 2.3105,
      "step": 62608
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7526300549507141,
      "learning_rate": 0.0004971626787076683,
      "loss": 2.2494,
      "step": 62609
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.8007073998451233,
      "learning_rate": 0.000497159595598077,
      "loss": 2.1944,
      "step": 62610
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.6644284129142761,
      "learning_rate": 0.00049715651245183,
      "loss": 2.3969,
      "step": 62611
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7301495671272278,
      "learning_rate": 0.0004971534292689278,
      "loss": 2.3665,
      "step": 62612
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7492043972015381,
      "learning_rate": 0.000497150346049371,
      "loss": 2.1395,
      "step": 62613
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.8332974314689636,
      "learning_rate": 0.0004971472627931603,
      "loss": 2.1048,
      "step": 62614
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.8477942943572998,
      "learning_rate": 0.0004971441795002962,
      "loss": 2.305,
      "step": 62615
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7195228934288025,
      "learning_rate": 0.0004971410961707793,
      "loss": 2.3483,
      "step": 62616
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7479894757270813,
      "learning_rate": 0.0004971380128046102,
      "loss": 2.139,
      "step": 62617
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7812436819076538,
      "learning_rate": 0.0004971349294017893,
      "loss": 2.1485,
      "step": 62618
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.8119542598724365,
      "learning_rate": 0.0004971318459623174,
      "loss": 2.4457,
      "step": 62619
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.9938051104545593,
      "learning_rate": 0.000497128762486195,
      "loss": 2.0218,
      "step": 62620
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.8352999091148376,
      "learning_rate": 0.0004971256789734226,
      "loss": 2.3867,
      "step": 62621
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.8142129182815552,
      "learning_rate": 0.0004971225954240007,
      "loss": 2.174,
      "step": 62622
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7854841947555542,
      "learning_rate": 0.00049711951183793,
      "loss": 2.006,
      "step": 62623
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7632418274879456,
      "learning_rate": 0.0004971164282152113,
      "loss": 2.1493,
      "step": 62624
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.738947868347168,
      "learning_rate": 0.0004971133445558447,
      "loss": 2.1992,
      "step": 62625
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.731112539768219,
      "learning_rate": 0.000497110260859831,
      "loss": 2.194,
      "step": 62626
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7611022591590881,
      "learning_rate": 0.0004971071771271709,
      "loss": 2.2527,
      "step": 62627
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7419068813323975,
      "learning_rate": 0.0004971040933578648,
      "loss": 2.1994,
      "step": 62628
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7751741409301758,
      "learning_rate": 0.0004971010095519134,
      "loss": 2.1144,
      "step": 62629
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7295520901679993,
      "learning_rate": 0.0004970979257093172,
      "loss": 2.1118,
      "step": 62630
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7058482766151428,
      "learning_rate": 0.0004970948418300767,
      "loss": 2.36,
      "step": 62631
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.8056440353393555,
      "learning_rate": 0.0004970917579141925,
      "loss": 2.2551,
      "step": 62632
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.8244807124137878,
      "learning_rate": 0.0004970886739616653,
      "loss": 2.0729,
      "step": 62633
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.732522189617157,
      "learning_rate": 0.0004970855899724955,
      "loss": 2.1834,
      "step": 62634
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.8103262186050415,
      "learning_rate": 0.0004970825059466839,
      "loss": 2.2923,
      "step": 62635
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7679431438446045,
      "learning_rate": 0.000497079421884231,
      "loss": 2.1707,
      "step": 62636
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7421510815620422,
      "learning_rate": 0.0004970763377851372,
      "loss": 2.4354,
      "step": 62637
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7649151086807251,
      "learning_rate": 0.0004970732536494031,
      "loss": 2.1123,
      "step": 62638
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.6992658972740173,
      "learning_rate": 0.0004970701694770295,
      "loss": 2.165,
      "step": 62639
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7442328333854675,
      "learning_rate": 0.0004970670852680168,
      "loss": 2.0807,
      "step": 62640
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7320978045463562,
      "learning_rate": 0.0004970640010223655,
      "loss": 2.3161,
      "step": 62641
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.8102184534072876,
      "learning_rate": 0.0004970609167400764,
      "loss": 2.1855,
      "step": 62642
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7413044571876526,
      "learning_rate": 0.00049705783242115,
      "loss": 2.3237,
      "step": 62643
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7546623349189758,
      "learning_rate": 0.0004970547480655867,
      "loss": 2.0527,
      "step": 62644
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7062863707542419,
      "learning_rate": 0.0004970516636733874,
      "loss": 2.2811,
      "step": 62645
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.790132999420166,
      "learning_rate": 0.0004970485792445523,
      "loss": 2.4454,
      "step": 62646
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7995415925979614,
      "learning_rate": 0.0004970454947790823,
      "loss": 2.597,
      "step": 62647
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7111900448799133,
      "learning_rate": 0.0004970424102769776,
      "loss": 2.1623,
      "step": 62648
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7655317783355713,
      "learning_rate": 0.0004970393257382391,
      "loss": 2.1668,
      "step": 62649
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7335500121116638,
      "learning_rate": 0.0004970362411628673,
      "loss": 1.9956,
      "step": 62650
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7844672203063965,
      "learning_rate": 0.0004970331565508628,
      "loss": 2.2815,
      "step": 62651
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.715212881565094,
      "learning_rate": 0.0004970300719022261,
      "loss": 2.2426,
      "step": 62652
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.6772384643554688,
      "learning_rate": 0.0004970269872169576,
      "loss": 2.1636,
      "step": 62653
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7664480209350586,
      "learning_rate": 0.0004970239024950583,
      "loss": 2.208,
      "step": 62654
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.947513222694397,
      "learning_rate": 0.0004970208177365285,
      "loss": 2.3801,
      "step": 62655
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7722091674804688,
      "learning_rate": 0.0004970177329413687,
      "loss": 2.2461,
      "step": 62656
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7578223347663879,
      "learning_rate": 0.0004970146481095797,
      "loss": 2.1809,
      "step": 62657
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.8207657933235168,
      "learning_rate": 0.0004970115632411618,
      "loss": 2.3143,
      "step": 62658
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.9392061233520508,
      "learning_rate": 0.0004970084783361158,
      "loss": 1.8412,
      "step": 62659
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7165973782539368,
      "learning_rate": 0.0004970053933944423,
      "loss": 2.2918,
      "step": 62660
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.8142472505569458,
      "learning_rate": 0.0004970023084161417,
      "loss": 2.4574,
      "step": 62661
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7762913107872009,
      "learning_rate": 0.0004969992234012147,
      "loss": 2.3271,
      "step": 62662
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7304692268371582,
      "learning_rate": 0.0004969961383496618,
      "loss": 2.2241,
      "step": 62663
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.8402805924415588,
      "learning_rate": 0.0004969930532614836,
      "loss": 2.0208,
      "step": 62664
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.711663544178009,
      "learning_rate": 0.0004969899681366806,
      "loss": 2.0358,
      "step": 62665
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.9170937538146973,
      "learning_rate": 0.0004969868829752535,
      "loss": 2.1546,
      "step": 62666
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.8019665479660034,
      "learning_rate": 0.0004969837977772028,
      "loss": 2.1155,
      "step": 62667
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7886214852333069,
      "learning_rate": 0.0004969807125425292,
      "loss": 2.2351,
      "step": 62668
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.815893292427063,
      "learning_rate": 0.000496977627271233,
      "loss": 2.0871,
      "step": 62669
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.8758271336555481,
      "learning_rate": 0.0004969745419633151,
      "loss": 2.1538,
      "step": 62670
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7545803189277649,
      "learning_rate": 0.0004969714566187758,
      "loss": 2.2816,
      "step": 62671
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.0479681491851807,
      "learning_rate": 0.0004969683712376159,
      "loss": 1.9718,
      "step": 62672
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7155282497406006,
      "learning_rate": 0.0004969652858198358,
      "loss": 2.351,
      "step": 62673
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.722774088382721,
      "learning_rate": 0.0004969622003654361,
      "loss": 2.3157,
      "step": 62674
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.9825499057769775,
      "learning_rate": 0.0004969591148744174,
      "loss": 2.3453,
      "step": 62675
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7200750708580017,
      "learning_rate": 0.0004969560293467803,
      "loss": 2.4683,
      "step": 62676
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7446075081825256,
      "learning_rate": 0.0004969529437825254,
      "loss": 2.3339,
      "step": 62677
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.742256760597229,
      "learning_rate": 0.0004969498581816532,
      "loss": 2.1228,
      "step": 62678
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.0048201084136963,
      "learning_rate": 0.0004969467725441644,
      "loss": 2.1682,
      "step": 62679
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7904289960861206,
      "learning_rate": 0.0004969436868700593,
      "loss": 2.4588,
      "step": 62680
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.8239766955375671,
      "learning_rate": 0.0004969406011593389,
      "loss": 2.0876,
      "step": 62681
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.0792922973632812,
      "learning_rate": 0.0004969375154120033,
      "loss": 2.1858,
      "step": 62682
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7431401610374451,
      "learning_rate": 0.0004969344296280534,
      "loss": 2.3417,
      "step": 62683
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7778372168540955,
      "learning_rate": 0.0004969313438074896,
      "loss": 2.4574,
      "step": 62684
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.8998996615409851,
      "learning_rate": 0.0004969282579503126,
      "loss": 2.2084,
      "step": 62685
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.8061059713363647,
      "learning_rate": 0.0004969251720565227,
      "loss": 2.0528,
      "step": 62686
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7094681859016418,
      "learning_rate": 0.000496922086126121,
      "loss": 2.398,
      "step": 62687
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7351257801055908,
      "learning_rate": 0.0004969190001591077,
      "loss": 2.3632,
      "step": 62688
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.8881663680076599,
      "learning_rate": 0.0004969159141554834,
      "loss": 2.4587,
      "step": 62689
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.8019678592681885,
      "learning_rate": 0.0004969128281152488,
      "loss": 2.3138,
      "step": 62690
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7075108289718628,
      "learning_rate": 0.0004969097420384043,
      "loss": 2.2883,
      "step": 62691
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.349945068359375,
      "learning_rate": 0.0004969066559249505,
      "loss": 2.188,
      "step": 62692
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7822769284248352,
      "learning_rate": 0.0004969035697748881,
      "loss": 2.3856,
      "step": 62693
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7236354351043701,
      "learning_rate": 0.0004969004835882176,
      "loss": 2.3182,
      "step": 62694
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7408686876296997,
      "learning_rate": 0.0004968973973649396,
      "loss": 2.2591,
      "step": 62695
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7512913346290588,
      "learning_rate": 0.0004968943111050546,
      "loss": 2.0774,
      "step": 62696
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7796218395233154,
      "learning_rate": 0.0004968912248085633,
      "loss": 2.1663,
      "step": 62697
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7366107702255249,
      "learning_rate": 0.0004968881384754662,
      "loss": 2.1128,
      "step": 62698
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7306442856788635,
      "learning_rate": 0.0004968850521057639,
      "loss": 2.2857,
      "step": 62699
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7501250505447388,
      "learning_rate": 0.0004968819656994568,
      "loss": 2.3122,
      "step": 62700
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.8014074563980103,
      "learning_rate": 0.0004968788792565458,
      "loss": 2.1195,
      "step": 62701
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7198458313941956,
      "learning_rate": 0.0004968757927770312,
      "loss": 2.2195,
      "step": 62702
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7931658625602722,
      "learning_rate": 0.0004968727062609138,
      "loss": 2.314,
      "step": 62703
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7030132412910461,
      "learning_rate": 0.000496869619708194,
      "loss": 2.0385,
      "step": 62704
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7258219122886658,
      "learning_rate": 0.0004968665331188723,
      "loss": 2.2202,
      "step": 62705
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7470211386680603,
      "learning_rate": 0.0004968634464929495,
      "loss": 2.2577,
      "step": 62706
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7032393217086792,
      "learning_rate": 0.0004968603598304261,
      "loss": 1.9071,
      "step": 62707
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7245721220970154,
      "learning_rate": 0.0004968572731313026,
      "loss": 2.141,
      "step": 62708
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7867379188537598,
      "learning_rate": 0.0004968541863955796,
      "loss": 2.4543,
      "step": 62709
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.6826561689376831,
      "learning_rate": 0.0004968510996232578,
      "loss": 1.9163,
      "step": 62710
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.8624684810638428,
      "learning_rate": 0.0004968480128143374,
      "loss": 2.2694,
      "step": 62711
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7934043407440186,
      "learning_rate": 0.0004968449259688195,
      "loss": 2.3272,
      "step": 62712
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7497163414955139,
      "learning_rate": 0.0004968418390867042,
      "loss": 2.2691,
      "step": 62713
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.8997151851654053,
      "learning_rate": 0.0004968387521679925,
      "loss": 2.3461,
      "step": 62714
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7489261031150818,
      "learning_rate": 0.0004968356652126847,
      "loss": 2.1306,
      "step": 62715
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7829073667526245,
      "learning_rate": 0.0004968325782207814,
      "loss": 2.44,
      "step": 62716
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7683312296867371,
      "learning_rate": 0.0004968294911922832,
      "loss": 2.395,
      "step": 62717
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7943747043609619,
      "learning_rate": 0.0004968264041271907,
      "loss": 2.3088,
      "step": 62718
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.677015483379364,
      "learning_rate": 0.0004968233170255045,
      "loss": 2.2166,
      "step": 62719
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7948041558265686,
      "learning_rate": 0.000496820229887225,
      "loss": 2.0335,
      "step": 62720
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7813255190849304,
      "learning_rate": 0.000496817142712353,
      "loss": 2.1972,
      "step": 62721
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7880308032035828,
      "learning_rate": 0.0004968140555008889,
      "loss": 2.3917,
      "step": 62722
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7572182416915894,
      "learning_rate": 0.0004968109682528335,
      "loss": 2.3082,
      "step": 62723
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.8772750496864319,
      "learning_rate": 0.0004968078809681871,
      "loss": 2.2559,
      "step": 62724
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.8917573094367981,
      "learning_rate": 0.0004968047936469503,
      "loss": 2.1936,
      "step": 62725
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.6523637771606445,
      "learning_rate": 0.000496801706289124,
      "loss": 2.0789,
      "step": 62726
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.8920820355415344,
      "learning_rate": 0.0004967986188947085,
      "loss": 2.2998,
      "step": 62727
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.868585467338562,
      "learning_rate": 0.0004967955314637044,
      "loss": 2.0891,
      "step": 62728
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.9514058232307434,
      "learning_rate": 0.0004967924439961123,
      "loss": 2.3659,
      "step": 62729
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.6935691237449646,
      "learning_rate": 0.0004967893564919327,
      "loss": 2.3272,
      "step": 62730
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.8857954740524292,
      "learning_rate": 0.0004967862689511663,
      "loss": 2.2466,
      "step": 62731
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.5202885866165161,
      "learning_rate": 0.0004967831813738136,
      "loss": 2.3475,
      "step": 62732
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7597060203552246,
      "learning_rate": 0.0004967800937598752,
      "loss": 2.1797,
      "step": 62733
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.800789475440979,
      "learning_rate": 0.0004967770061093516,
      "loss": 2.258,
      "step": 62734
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.9125881195068359,
      "learning_rate": 0.0004967739184222436,
      "loss": 2.1254,
      "step": 62735
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7550137042999268,
      "learning_rate": 0.0004967708306985515,
      "loss": 2.0434,
      "step": 62736
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7678083181381226,
      "learning_rate": 0.000496767742938276,
      "loss": 2.1689,
      "step": 62737
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7461506128311157,
      "learning_rate": 0.0004967646551414176,
      "loss": 2.2299,
      "step": 62738
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.692345380783081,
      "learning_rate": 0.000496761567307977,
      "loss": 2.1375,
      "step": 62739
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7570926547050476,
      "learning_rate": 0.0004967584794379548,
      "loss": 2.1095,
      "step": 62740
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7672553062438965,
      "learning_rate": 0.0004967553915313513,
      "loss": 2.4035,
      "step": 62741
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7463895082473755,
      "learning_rate": 0.0004967523035881675,
      "loss": 2.0617,
      "step": 62742
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.6682782769203186,
      "learning_rate": 0.0004967492156084034,
      "loss": 2.2055,
      "step": 62743
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7489051222801208,
      "learning_rate": 0.0004967461275920601,
      "loss": 2.139,
      "step": 62744
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7894511818885803,
      "learning_rate": 0.0004967430395391381,
      "loss": 1.9695,
      "step": 62745
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7047226428985596,
      "learning_rate": 0.0004967399514496378,
      "loss": 2.2208,
      "step": 62746
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7688776254653931,
      "learning_rate": 0.0004967368633235597,
      "loss": 2.2063,
      "step": 62747
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7640584111213684,
      "learning_rate": 0.0004967337751609046,
      "loss": 2.1134,
      "step": 62748
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7534985542297363,
      "learning_rate": 0.0004967306869616729,
      "loss": 2.2366,
      "step": 62749
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.8312948346138,
      "learning_rate": 0.0004967275987258653,
      "loss": 2.3747,
      "step": 62750
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.766339123249054,
      "learning_rate": 0.0004967245104534824,
      "loss": 2.1221,
      "step": 62751
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7659011483192444,
      "learning_rate": 0.0004967214221445245,
      "loss": 2.1353,
      "step": 62752
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7307906150817871,
      "learning_rate": 0.0004967183337989925,
      "loss": 2.1879,
      "step": 62753
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7756588459014893,
      "learning_rate": 0.0004967152454168869,
      "loss": 2.2874,
      "step": 62754
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7220055460929871,
      "learning_rate": 0.0004967121569982081,
      "loss": 2.2642,
      "step": 62755
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7312942147254944,
      "learning_rate": 0.0004967090685429569,
      "loss": 2.3387,
      "step": 62756
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.738258957862854,
      "learning_rate": 0.0004967059800511337,
      "loss": 2.1617,
      "step": 62757
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7639858722686768,
      "learning_rate": 0.0004967028915227392,
      "loss": 2.2442,
      "step": 62758
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.6967755556106567,
      "learning_rate": 0.0004966998029577739,
      "loss": 2.3495,
      "step": 62759
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.6862170696258545,
      "learning_rate": 0.0004966967143562384,
      "loss": 2.3178,
      "step": 62760
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.769761323928833,
      "learning_rate": 0.0004966936257181333,
      "loss": 2.334,
      "step": 62761
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.8322130441665649,
      "learning_rate": 0.0004966905370434591,
      "loss": 2.124,
      "step": 62762
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.8045266270637512,
      "learning_rate": 0.0004966874483322164,
      "loss": 2.5042,
      "step": 62763
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7384481430053711,
      "learning_rate": 0.0004966843595844059,
      "loss": 2.1096,
      "step": 62764
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7124475240707397,
      "learning_rate": 0.0004966812708000279,
      "loss": 2.2317,
      "step": 62765
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.8038445115089417,
      "learning_rate": 0.0004966781819790832,
      "loss": 2.1628,
      "step": 62766
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.668034017086029,
      "learning_rate": 0.0004966750931215723,
      "loss": 2.2535,
      "step": 62767
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7840437889099121,
      "learning_rate": 0.0004966720042274959,
      "loss": 2.0745,
      "step": 62768
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.695652961730957,
      "learning_rate": 0.0004966689152968544,
      "loss": 2.1862,
      "step": 62769
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.740278422832489,
      "learning_rate": 0.0004966658263296483,
      "loss": 2.0771,
      "step": 62770
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.8381204605102539,
      "learning_rate": 0.0004966627373258786,
      "loss": 2.3546,
      "step": 62771
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7138247489929199,
      "learning_rate": 0.0004966596482855452,
      "loss": 2.3374,
      "step": 62772
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7367235422134399,
      "learning_rate": 0.0004966565592086494,
      "loss": 2.1232,
      "step": 62773
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7348439693450928,
      "learning_rate": 0.0004966534700951913,
      "loss": 2.19,
      "step": 62774
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.8777356147766113,
      "learning_rate": 0.0004966503809451715,
      "loss": 2.3234,
      "step": 62775
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7476847767829895,
      "learning_rate": 0.0004966472917585909,
      "loss": 2.1723,
      "step": 62776
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7548457980155945,
      "learning_rate": 0.0004966442025354498,
      "loss": 1.9889,
      "step": 62777
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.6978687644004822,
      "learning_rate": 0.0004966411132757486,
      "loss": 2.2421,
      "step": 62778
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7661916017532349,
      "learning_rate": 0.0004966380239794884,
      "loss": 2.2433,
      "step": 62779
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7320266366004944,
      "learning_rate": 0.0004966349346466693,
      "loss": 1.9797,
      "step": 62780
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7453503012657166,
      "learning_rate": 0.0004966318452772922,
      "loss": 2.179,
      "step": 62781
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7794222235679626,
      "learning_rate": 0.0004966287558713574,
      "loss": 2.1496,
      "step": 62782
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7819259166717529,
      "learning_rate": 0.0004966256664288656,
      "loss": 2.4118,
      "step": 62783
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7319028377532959,
      "learning_rate": 0.0004966225769498175,
      "loss": 2.1277,
      "step": 62784
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7257999777793884,
      "learning_rate": 0.0004966194874342134,
      "loss": 2.1281,
      "step": 62785
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7924557328224182,
      "learning_rate": 0.0004966163978820542,
      "loss": 2.2888,
      "step": 62786
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.320360541343689,
      "learning_rate": 0.0004966133082933402,
      "loss": 2.1322,
      "step": 62787
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7021993398666382,
      "learning_rate": 0.000496610218668072,
      "loss": 2.044,
      "step": 62788
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7963626384735107,
      "learning_rate": 0.0004966071290062504,
      "loss": 1.9445,
      "step": 62789
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.8704817295074463,
      "learning_rate": 0.0004966040393078757,
      "loss": 2.1494,
      "step": 62790
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.9444937109947205,
      "learning_rate": 0.0004966009495729487,
      "loss": 2.3083,
      "step": 62791
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7767535448074341,
      "learning_rate": 0.0004965978598014698,
      "loss": 2.2225,
      "step": 62792
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.8742477893829346,
      "learning_rate": 0.0004965947699934396,
      "loss": 2.1835,
      "step": 62793
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.0199967622756958,
      "learning_rate": 0.0004965916801488589,
      "loss": 2.0429,
      "step": 62794
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.8057453632354736,
      "learning_rate": 0.000496588590267728,
      "loss": 2.3776,
      "step": 62795
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.6983590722084045,
      "learning_rate": 0.0004965855003500474,
      "loss": 2.1083,
      "step": 62796
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.0378426313400269,
      "learning_rate": 0.0004965824103958182,
      "loss": 2.2563,
      "step": 62797
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7822529673576355,
      "learning_rate": 0.0004965793204050403,
      "loss": 2.2047,
      "step": 62798
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7602049112319946,
      "learning_rate": 0.0004965762303777147,
      "loss": 2.2417,
      "step": 62799
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7970085740089417,
      "learning_rate": 0.000496573140313842,
      "loss": 2.4421,
      "step": 62800
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7906491756439209,
      "learning_rate": 0.0004965700502134224,
      "loss": 2.0883,
      "step": 62801
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7718112468719482,
      "learning_rate": 0.0004965669600764569,
      "loss": 2.0934,
      "step": 62802
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7163899540901184,
      "learning_rate": 0.0004965638699029459,
      "loss": 2.2322,
      "step": 62803
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.801103413105011,
      "learning_rate": 0.0004965607796928898,
      "loss": 2.249,
      "step": 62804
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.8405841588973999,
      "learning_rate": 0.0004965576894462895,
      "loss": 2.0125,
      "step": 62805
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7225763201713562,
      "learning_rate": 0.0004965545991631454,
      "loss": 2.1891,
      "step": 62806
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7031627893447876,
      "learning_rate": 0.000496551508843458,
      "loss": 2.4243,
      "step": 62807
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.9835386276245117,
      "learning_rate": 0.0004965484184872279,
      "loss": 2.0958,
      "step": 62808
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7401463389396667,
      "learning_rate": 0.0004965453280944559,
      "loss": 2.2446,
      "step": 62809
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7852917909622192,
      "learning_rate": 0.0004965422376651424,
      "loss": 2.0935,
      "step": 62810
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7330784201622009,
      "learning_rate": 0.0004965391471992879,
      "loss": 2.4244,
      "step": 62811
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.719276487827301,
      "learning_rate": 0.0004965360566968932,
      "loss": 2.1739,
      "step": 62812
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7409050464630127,
      "learning_rate": 0.0004965329661579586,
      "loss": 2.3638,
      "step": 62813
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.8067817091941833,
      "learning_rate": 0.0004965298755824849,
      "loss": 2.254,
      "step": 62814
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.6955133080482483,
      "learning_rate": 0.0004965267849704726,
      "loss": 2.3638,
      "step": 62815
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7430004477500916,
      "learning_rate": 0.0004965236943219221,
      "loss": 2.3707,
      "step": 62816
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.8535480499267578,
      "learning_rate": 0.0004965206036368343,
      "loss": 2.1556,
      "step": 62817
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.775484561920166,
      "learning_rate": 0.0004965175129152095,
      "loss": 2.3279,
      "step": 62818
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7523549795150757,
      "learning_rate": 0.0004965144221570485,
      "loss": 2.3829,
      "step": 62819
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.0139847993850708,
      "learning_rate": 0.0004965113313623517,
      "loss": 2.1818,
      "step": 62820
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7320840954780579,
      "learning_rate": 0.0004965082405311196,
      "loss": 2.1869,
      "step": 62821
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7150114178657532,
      "learning_rate": 0.0004965051496633531,
      "loss": 2.0816,
      "step": 62822
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.9111218452453613,
      "learning_rate": 0.0004965020587590525,
      "loss": 2.2222,
      "step": 62823
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.12968111038208,
      "learning_rate": 0.0004964989678182185,
      "loss": 2.1884,
      "step": 62824
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7451751828193665,
      "learning_rate": 0.0004964958768408517,
      "loss": 2.0027,
      "step": 62825
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.879905641078949,
      "learning_rate": 0.0004964927858269524,
      "loss": 2.2455,
      "step": 62826
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.9606308341026306,
      "learning_rate": 0.0004964896947765215,
      "loss": 2.1193,
      "step": 62827
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.8016196489334106,
      "learning_rate": 0.0004964866036895595,
      "loss": 2.1429,
      "step": 62828
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.9567839503288269,
      "learning_rate": 0.0004964835125660669,
      "loss": 2.1218,
      "step": 62829
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7464814186096191,
      "learning_rate": 0.0004964804214060443,
      "loss": 2.08,
      "step": 62830
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7382274270057678,
      "learning_rate": 0.0004964773302094923,
      "loss": 2.2612,
      "step": 62831
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7566052079200745,
      "learning_rate": 0.0004964742389764114,
      "loss": 2.2299,
      "step": 62832
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.9336966872215271,
      "learning_rate": 0.0004964711477068023,
      "loss": 2.1597,
      "step": 62833
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7476241588592529,
      "learning_rate": 0.0004964680564006654,
      "loss": 2.4193,
      "step": 62834
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7272956371307373,
      "learning_rate": 0.0004964649650580015,
      "loss": 2.2294,
      "step": 62835
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7957665324211121,
      "learning_rate": 0.000496461873678811,
      "loss": 2.287,
      "step": 62836
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7542520761489868,
      "learning_rate": 0.0004964587822630945,
      "loss": 1.9647,
      "step": 62837
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7073386907577515,
      "learning_rate": 0.0004964556908108527,
      "loss": 2.0486,
      "step": 62838
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.8379859924316406,
      "learning_rate": 0.000496452599322086,
      "loss": 2.4813,
      "step": 62839
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7867185473442078,
      "learning_rate": 0.000496449507796795,
      "loss": 2.2952,
      "step": 62840
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.851016104221344,
      "learning_rate": 0.0004964464162349804,
      "loss": 2.3043,
      "step": 62841
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.8219321370124817,
      "learning_rate": 0.0004964433246366427,
      "loss": 2.3292,
      "step": 62842
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.8017271161079407,
      "learning_rate": 0.0004964402330017826,
      "loss": 2.2965,
      "step": 62843
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7494914531707764,
      "learning_rate": 0.0004964371413304004,
      "loss": 2.2718,
      "step": 62844
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7650207281112671,
      "learning_rate": 0.0004964340496224968,
      "loss": 2.335,
      "step": 62845
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7785515785217285,
      "learning_rate": 0.0004964309578780725,
      "loss": 2.3323,
      "step": 62846
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.8075865507125854,
      "learning_rate": 0.0004964278660971278,
      "loss": 2.1963,
      "step": 62847
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7873066067695618,
      "learning_rate": 0.0004964247742796637,
      "loss": 2.3966,
      "step": 62848
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7098736763000488,
      "learning_rate": 0.0004964216824256805,
      "loss": 2.1395,
      "step": 62849
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.0616521835327148,
      "learning_rate": 0.0004964185905351787,
      "loss": 2.1222,
      "step": 62850
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7109954953193665,
      "learning_rate": 0.0004964154986081589,
      "loss": 2.215,
      "step": 62851
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.8120941519737244,
      "learning_rate": 0.000496412406644622,
      "loss": 2.1121,
      "step": 62852
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.8678808212280273,
      "learning_rate": 0.0004964093146445681,
      "loss": 1.9961,
      "step": 62853
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7568151950836182,
      "learning_rate": 0.000496406222607998,
      "loss": 2.1487,
      "step": 62854
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.8078007102012634,
      "learning_rate": 0.0004964031305349124,
      "loss": 2.1927,
      "step": 62855
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7571190595626831,
      "learning_rate": 0.0004964000384253117,
      "loss": 2.3048,
      "step": 62856
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.6721296906471252,
      "learning_rate": 0.0004963969462791965,
      "loss": 2.1569,
      "step": 62857
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.9258145093917847,
      "learning_rate": 0.0004963938540965675,
      "loss": 1.97,
      "step": 62858
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7754983305931091,
      "learning_rate": 0.000496390761877425,
      "loss": 2.4217,
      "step": 62859
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.8647390007972717,
      "learning_rate": 0.0004963876696217699,
      "loss": 2.1744,
      "step": 62860
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.8448449969291687,
      "learning_rate": 0.0004963845773296026,
      "loss": 2.0843,
      "step": 62861
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7773630023002625,
      "learning_rate": 0.0004963814850009236,
      "loss": 2.2664,
      "step": 62862
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.8421831727027893,
      "learning_rate": 0.0004963783926357337,
      "loss": 2.1078,
      "step": 62863
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.9251908659934998,
      "learning_rate": 0.0004963753002340332,
      "loss": 2.2835,
      "step": 62864
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7040569186210632,
      "learning_rate": 0.0004963722077958229,
      "loss": 2.2398,
      "step": 62865
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.8214537501335144,
      "learning_rate": 0.0004963691153211034,
      "loss": 2.3334,
      "step": 62866
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.8060728311538696,
      "learning_rate": 0.0004963660228098751,
      "loss": 2.1047,
      "step": 62867
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7434194087982178,
      "learning_rate": 0.0004963629302621387,
      "loss": 2.1584,
      "step": 62868
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7897951602935791,
      "learning_rate": 0.0004963598376778946,
      "loss": 2.3776,
      "step": 62869
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.8549643754959106,
      "learning_rate": 0.0004963567450571436,
      "loss": 2.1669,
      "step": 62870
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.8410260677337646,
      "learning_rate": 0.0004963536523998861,
      "loss": 2.1924,
      "step": 62871
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.6788957118988037,
      "learning_rate": 0.0004963505597061228,
      "loss": 2.2984,
      "step": 62872
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.8003370761871338,
      "learning_rate": 0.0004963474669758542,
      "loss": 2.3311,
      "step": 62873
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.8208824992179871,
      "learning_rate": 0.000496344374209081,
      "loss": 2.32,
      "step": 62874
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7533534169197083,
      "learning_rate": 0.0004963412814058036,
      "loss": 2.2653,
      "step": 62875
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.8041107654571533,
      "learning_rate": 0.0004963381885660226,
      "loss": 2.2473,
      "step": 62876
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.1826459169387817,
      "learning_rate": 0.0004963350956897387,
      "loss": 2.3316,
      "step": 62877
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.6739930510520935,
      "learning_rate": 0.0004963320027769523,
      "loss": 2.3112,
      "step": 62878
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.8517642617225647,
      "learning_rate": 0.0004963289098276643,
      "loss": 2.3971,
      "step": 62879
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.8121204972267151,
      "learning_rate": 0.0004963258168418748,
      "loss": 2.4488,
      "step": 62880
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7935538291931152,
      "learning_rate": 0.0004963227238195848,
      "loss": 2.3594,
      "step": 62881
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7542569041252136,
      "learning_rate": 0.0004963196307607946,
      "loss": 2.145,
      "step": 62882
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7883999943733215,
      "learning_rate": 0.0004963165376655048,
      "loss": 2.2548,
      "step": 62883
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7929112911224365,
      "learning_rate": 0.0004963134445337161,
      "loss": 2.3379,
      "step": 62884
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7978501915931702,
      "learning_rate": 0.0004963103513654292,
      "loss": 2.3559,
      "step": 62885
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7461015582084656,
      "learning_rate": 0.0004963072581606443,
      "loss": 2.2388,
      "step": 62886
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7844163775444031,
      "learning_rate": 0.0004963041649193623,
      "loss": 2.0238,
      "step": 62887
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7508969306945801,
      "learning_rate": 0.0004963010716415835,
      "loss": 2.2441,
      "step": 62888
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7715144753456116,
      "learning_rate": 0.0004962979783273088,
      "loss": 2.3501,
      "step": 62889
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.87990802526474,
      "learning_rate": 0.0004962948849765384,
      "loss": 2.2765,
      "step": 62890
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7082195281982422,
      "learning_rate": 0.0004962917915892732,
      "loss": 2.1443,
      "step": 62891
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7232640981674194,
      "learning_rate": 0.0004962886981655137,
      "loss": 2.2157,
      "step": 62892
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7333753108978271,
      "learning_rate": 0.0004962856047052603,
      "loss": 2.4013,
      "step": 62893
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.9073670506477356,
      "learning_rate": 0.0004962825112085139,
      "loss": 2.1685,
      "step": 62894
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.9506516456604004,
      "learning_rate": 0.0004962794176752747,
      "loss": 2.1361,
      "step": 62895
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7826189398765564,
      "learning_rate": 0.0004962763241055435,
      "loss": 2.301,
      "step": 62896
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7663720846176147,
      "learning_rate": 0.0004962732304993209,
      "loss": 2.3147,
      "step": 62897
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.9104383587837219,
      "learning_rate": 0.0004962701368566073,
      "loss": 2.4035,
      "step": 62898
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7191469073295593,
      "learning_rate": 0.0004962670431774034,
      "loss": 2.2308,
      "step": 62899
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7846565842628479,
      "learning_rate": 0.0004962639494617097,
      "loss": 2.1823,
      "step": 62900
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.867733359336853,
      "learning_rate": 0.0004962608557095269,
      "loss": 2.2987,
      "step": 62901
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7094568014144897,
      "learning_rate": 0.0004962577619208556,
      "loss": 2.2965,
      "step": 62902
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7936276793479919,
      "learning_rate": 0.000496254668095696,
      "loss": 2.4287,
      "step": 62903
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.6971285343170166,
      "learning_rate": 0.0004962515742340492,
      "loss": 2.2267,
      "step": 62904
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7236742973327637,
      "learning_rate": 0.0004962484803359154,
      "loss": 2.1659,
      "step": 62905
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.72335284948349,
      "learning_rate": 0.0004962453864012953,
      "loss": 2.3074,
      "step": 62906
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7409959435462952,
      "learning_rate": 0.0004962422924301896,
      "loss": 2.4072,
      "step": 62907
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.6887266039848328,
      "learning_rate": 0.0004962391984225987,
      "loss": 2.2826,
      "step": 62908
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.6948390007019043,
      "learning_rate": 0.0004962361043785231,
      "loss": 2.1979,
      "step": 62909
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.8457764983177185,
      "learning_rate": 0.0004962330102979637,
      "loss": 2.3799,
      "step": 62910
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7833055257797241,
      "learning_rate": 0.0004962299161809207,
      "loss": 2.2367,
      "step": 62911
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.8834757208824158,
      "learning_rate": 0.0004962268220273949,
      "loss": 2.0867,
      "step": 62912
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7154335379600525,
      "learning_rate": 0.000496223727837387,
      "loss": 2.1232,
      "step": 62913
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.773765504360199,
      "learning_rate": 0.0004962206336108973,
      "loss": 2.4037,
      "step": 62914
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.894169807434082,
      "learning_rate": 0.0004962175393479263,
      "loss": 2.0412,
      "step": 62915
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7094521522521973,
      "learning_rate": 0.000496214445048475,
      "loss": 2.3818,
      "step": 62916
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.8256745934486389,
      "learning_rate": 0.0004962113507125436,
      "loss": 2.3677,
      "step": 62917
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.838955283164978,
      "learning_rate": 0.0004962082563401329,
      "loss": 2.3154,
      "step": 62918
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7343435287475586,
      "learning_rate": 0.0004962051619312433,
      "loss": 2.2253,
      "step": 62919
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7317309379577637,
      "learning_rate": 0.0004962020674858755,
      "loss": 2.3586,
      "step": 62920
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.796726644039154,
      "learning_rate": 0.0004961989730040301,
      "loss": 2.0544,
      "step": 62921
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7925207018852234,
      "learning_rate": 0.0004961958784857075,
      "loss": 2.1376,
      "step": 62922
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.8139008283615112,
      "learning_rate": 0.0004961927839309084,
      "loss": 2.2658,
      "step": 62923
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.777741551399231,
      "learning_rate": 0.0004961896893396333,
      "loss": 2.1493,
      "step": 62924
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7541370987892151,
      "learning_rate": 0.000496186594711883,
      "loss": 2.1654,
      "step": 62925
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7068921327590942,
      "learning_rate": 0.0004961835000476578,
      "loss": 2.2439,
      "step": 62926
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7484166622161865,
      "learning_rate": 0.0004961804053469583,
      "loss": 2.1153,
      "step": 62927
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7831894755363464,
      "learning_rate": 0.0004961773106097853,
      "loss": 2.348,
      "step": 62928
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.8235662579536438,
      "learning_rate": 0.0004961742158361392,
      "loss": 2.2882,
      "step": 62929
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7937467694282532,
      "learning_rate": 0.0004961711210260206,
      "loss": 2.1701,
      "step": 62930
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7462519407272339,
      "learning_rate": 0.0004961680261794301,
      "loss": 2.2304,
      "step": 62931
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7446431517601013,
      "learning_rate": 0.0004961649312963683,
      "loss": 1.853,
      "step": 62932
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.6909084916114807,
      "learning_rate": 0.0004961618363768356,
      "loss": 2.1241,
      "step": 62933
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7698324918746948,
      "learning_rate": 0.0004961587414208327,
      "loss": 2.1113,
      "step": 62934
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7526314854621887,
      "learning_rate": 0.0004961556464283604,
      "loss": 2.2883,
      "step": 62935
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.8013395071029663,
      "learning_rate": 0.000496152551399419,
      "loss": 2.2762,
      "step": 62936
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.6781602501869202,
      "learning_rate": 0.0004961494563340091,
      "loss": 2.1447,
      "step": 62937
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.8196645975112915,
      "learning_rate": 0.0004961463612321313,
      "loss": 2.4163,
      "step": 62938
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.8293994665145874,
      "learning_rate": 0.000496143266093786,
      "loss": 2.4522,
      "step": 62939
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7247685194015503,
      "learning_rate": 0.0004961401709189742,
      "loss": 2.493,
      "step": 62940
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.726358950138092,
      "learning_rate": 0.0004961370757076962,
      "loss": 2.2867,
      "step": 62941
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.8533841967582703,
      "learning_rate": 0.0004961339804599525,
      "loss": 2.2656,
      "step": 62942
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7171679735183716,
      "learning_rate": 0.000496130885175744,
      "loss": 2.2253,
      "step": 62943
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.6983482241630554,
      "learning_rate": 0.000496127789855071,
      "loss": 2.2196,
      "step": 62944
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7698519825935364,
      "learning_rate": 0.000496124694497934,
      "loss": 2.3058,
      "step": 62945
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.721972644329071,
      "learning_rate": 0.0004961215991043339,
      "loss": 2.2763,
      "step": 62946
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.8126868009567261,
      "learning_rate": 0.0004961185036742711,
      "loss": 2.1258,
      "step": 62947
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.8056548237800598,
      "learning_rate": 0.000496115408207746,
      "loss": 2.1192,
      "step": 62948
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7782623767852783,
      "learning_rate": 0.0004961123127047595,
      "loss": 2.4341,
      "step": 62949
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.6929810047149658,
      "learning_rate": 0.0004961092171653119,
      "loss": 2.2654,
      "step": 62950
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.8037365078926086,
      "learning_rate": 0.0004961061215894039,
      "loss": 2.1755,
      "step": 62951
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.952831506729126,
      "learning_rate": 0.0004961030259770361,
      "loss": 2.3115,
      "step": 62952
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.8413158059120178,
      "learning_rate": 0.0004960999303282091,
      "loss": 2.3038,
      "step": 62953
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.8790777921676636,
      "learning_rate": 0.0004960968346429234,
      "loss": 2.296,
      "step": 62954
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.6847735047340393,
      "learning_rate": 0.0004960937389211794,
      "loss": 2.1559,
      "step": 62955
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7166706919670105,
      "learning_rate": 0.0004960906431629782,
      "loss": 2.3184,
      "step": 62956
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.755896806716919,
      "learning_rate": 0.0004960875473683198,
      "loss": 2.4268,
      "step": 62957
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.6838526129722595,
      "learning_rate": 0.0004960844515372052,
      "loss": 2.1232,
      "step": 62958
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7759792804718018,
      "learning_rate": 0.0004960813556696346,
      "loss": 2.1294,
      "step": 62959
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7161139845848083,
      "learning_rate": 0.0004960782597656089,
      "loss": 2.0273,
      "step": 62960
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7479603886604309,
      "learning_rate": 0.0004960751638251286,
      "loss": 2.3374,
      "step": 62961
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7476177215576172,
      "learning_rate": 0.0004960720678481941,
      "loss": 2.1277,
      "step": 62962
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.8126901984214783,
      "learning_rate": 0.0004960689718348062,
      "loss": 2.3787,
      "step": 62963
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7050094604492188,
      "learning_rate": 0.0004960658757849653,
      "loss": 2.1465,
      "step": 62964
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.8054439425468445,
      "learning_rate": 0.0004960627796986722,
      "loss": 2.3034,
      "step": 62965
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.790415346622467,
      "learning_rate": 0.0004960596835759272,
      "loss": 1.9695,
      "step": 62966
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7686182856559753,
      "learning_rate": 0.0004960565874167311,
      "loss": 2.4495,
      "step": 62967
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7408972382545471,
      "learning_rate": 0.0004960534912210842,
      "loss": 2.5132,
      "step": 62968
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7751203179359436,
      "learning_rate": 0.0004960503949889874,
      "loss": 2.1875,
      "step": 62969
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7703478932380676,
      "learning_rate": 0.0004960472987204411,
      "loss": 2.2423,
      "step": 62970
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7664105296134949,
      "learning_rate": 0.0004960442024154459,
      "loss": 2.088,
      "step": 62971
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.8458626866340637,
      "learning_rate": 0.0004960411060740024,
      "loss": 1.9583,
      "step": 62972
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7145316004753113,
      "learning_rate": 0.0004960380096961111,
      "loss": 2.2831,
      "step": 62973
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7334277629852295,
      "learning_rate": 0.0004960349132817727,
      "loss": 2.2423,
      "step": 62974
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7154569625854492,
      "learning_rate": 0.0004960318168309877,
      "loss": 2.3773,
      "step": 62975
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7240117788314819,
      "learning_rate": 0.0004960287203437567,
      "loss": 2.111,
      "step": 62976
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.778160035610199,
      "learning_rate": 0.0004960256238200802,
      "loss": 2.3657,
      "step": 62977
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7371357083320618,
      "learning_rate": 0.0004960225272599588,
      "loss": 2.2882,
      "step": 62978
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7164478898048401,
      "learning_rate": 0.0004960194306633932,
      "loss": 2.3129,
      "step": 62979
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.771373987197876,
      "learning_rate": 0.0004960163340303839,
      "loss": 2.206,
      "step": 62980
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7296808958053589,
      "learning_rate": 0.0004960132373609314,
      "loss": 2.0896,
      "step": 62981
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7867135405540466,
      "learning_rate": 0.0004960101406550364,
      "loss": 2.3746,
      "step": 62982
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7869069576263428,
      "learning_rate": 0.0004960070439126993,
      "loss": 2.1447,
      "step": 62983
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7160292267799377,
      "learning_rate": 0.000496003947133921,
      "loss": 2.3248,
      "step": 62984
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7868867516517639,
      "learning_rate": 0.0004960008503187016,
      "loss": 2.2077,
      "step": 62985
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.688401460647583,
      "learning_rate": 0.0004959977534670422,
      "loss": 2.2002,
      "step": 62986
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7484841346740723,
      "learning_rate": 0.0004959946565789429,
      "loss": 2.0065,
      "step": 62987
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7002922892570496,
      "learning_rate": 0.0004959915596544046,
      "loss": 2.2434,
      "step": 62988
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.6779016852378845,
      "learning_rate": 0.0004959884626934279,
      "loss": 2.3985,
      "step": 62989
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7187592387199402,
      "learning_rate": 0.0004959853656960131,
      "loss": 2.0537,
      "step": 62990
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7490631341934204,
      "learning_rate": 0.0004959822686621608,
      "loss": 2.1228,
      "step": 62991
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7589311599731445,
      "learning_rate": 0.0004959791715918719,
      "loss": 2.256,
      "step": 62992
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7895569801330566,
      "learning_rate": 0.0004959760744851466,
      "loss": 2.126,
      "step": 62993
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7266225218772888,
      "learning_rate": 0.0004959729773419858,
      "loss": 2.1694,
      "step": 62994
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7052329778671265,
      "learning_rate": 0.0004959698801623899,
      "loss": 2.395,
      "step": 62995
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.6852329969406128,
      "learning_rate": 0.0004959667829463595,
      "loss": 1.9038,
      "step": 62996
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7318711876869202,
      "learning_rate": 0.0004959636856938951,
      "loss": 2.3598,
      "step": 62997
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7578347325325012,
      "learning_rate": 0.0004959605884049974,
      "loss": 2.2009,
      "step": 62998
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.8419778347015381,
      "learning_rate": 0.0004959574910796669,
      "loss": 2.3692,
      "step": 62999
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7850630283355713,
      "learning_rate": 0.0004959543937179043,
      "loss": 2.3792,
      "step": 63000
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7586725950241089,
      "learning_rate": 0.00049595129631971,
      "loss": 2.3148,
      "step": 63001
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7789387702941895,
      "learning_rate": 0.0004959481988850847,
      "loss": 2.3358,
      "step": 63002
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.8038011789321899,
      "learning_rate": 0.0004959451014140288,
      "loss": 2.2015,
      "step": 63003
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.8507336974143982,
      "learning_rate": 0.0004959420039065431,
      "loss": 2.1975,
      "step": 63004
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7280917167663574,
      "learning_rate": 0.0004959389063626281,
      "loss": 2.2548,
      "step": 63005
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.001737117767334,
      "learning_rate": 0.0004959358087822844,
      "loss": 2.3522,
      "step": 63006
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7754384279251099,
      "learning_rate": 0.0004959327111655125,
      "loss": 2.2032,
      "step": 63007
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.9147494435310364,
      "learning_rate": 0.0004959296135123129,
      "loss": 2.229,
      "step": 63008
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.8197715282440186,
      "learning_rate": 0.0004959265158226864,
      "loss": 2.3123,
      "step": 63009
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7324252724647522,
      "learning_rate": 0.0004959234180966334,
      "loss": 2.1503,
      "step": 63010
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.6897069215774536,
      "learning_rate": 0.0004959203203341547,
      "loss": 2.1103,
      "step": 63011
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.8086406588554382,
      "learning_rate": 0.0004959172225352504,
      "loss": 2.2891,
      "step": 63012
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7880179286003113,
      "learning_rate": 0.0004959141246999217,
      "loss": 2.3524,
      "step": 63013
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7096036672592163,
      "learning_rate": 0.0004959110268281687,
      "loss": 2.2311,
      "step": 63014
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.8007892370223999,
      "learning_rate": 0.0004959079289199922,
      "loss": 2.1063,
      "step": 63015
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7076429128646851,
      "learning_rate": 0.0004959048309753928,
      "loss": 1.9022,
      "step": 63016
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7426642775535583,
      "learning_rate": 0.0004959017329943708,
      "loss": 2.207,
      "step": 63017
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.6953311562538147,
      "learning_rate": 0.0004958986349769271,
      "loss": 2.2202,
      "step": 63018
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7416672110557556,
      "learning_rate": 0.0004958955369230621,
      "loss": 2.2699,
      "step": 63019
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7944173812866211,
      "learning_rate": 0.0004958924388327764,
      "loss": 2.2567,
      "step": 63020
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.8979582190513611,
      "learning_rate": 0.0004958893407060707,
      "loss": 2.1982,
      "step": 63021
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7364902496337891,
      "learning_rate": 0.0004958862425429454,
      "loss": 2.1632,
      "step": 63022
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.730505645275116,
      "learning_rate": 0.0004958831443434012,
      "loss": 1.9591,
      "step": 63023
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.8216589093208313,
      "learning_rate": 0.0004958800461074385,
      "loss": 1.9657,
      "step": 63024
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7733389139175415,
      "learning_rate": 0.0004958769478350581,
      "loss": 2.3261,
      "step": 63025
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7681528925895691,
      "learning_rate": 0.0004958738495262605,
      "loss": 2.2268,
      "step": 63026
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.6984412670135498,
      "learning_rate": 0.0004958707511810462,
      "loss": 2.2179,
      "step": 63027
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.6961542367935181,
      "learning_rate": 0.0004958676527994159,
      "loss": 2.1161,
      "step": 63028
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7018778920173645,
      "learning_rate": 0.00049586455438137,
      "loss": 2.1976,
      "step": 63029
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7859118580818176,
      "learning_rate": 0.0004958614559269093,
      "loss": 2.2203,
      "step": 63030
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7536674737930298,
      "learning_rate": 0.0004958583574360342,
      "loss": 2.3957,
      "step": 63031
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7971374988555908,
      "learning_rate": 0.0004958552589087453,
      "loss": 2.2969,
      "step": 63032
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.9381697773933411,
      "learning_rate": 0.0004958521603450432,
      "loss": 2.1285,
      "step": 63033
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.8221212029457092,
      "learning_rate": 0.0004958490617449285,
      "loss": 2.2722,
      "step": 63034
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.8935364484786987,
      "learning_rate": 0.0004958459631084018,
      "loss": 2.3739,
      "step": 63035
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.830894410610199,
      "learning_rate": 0.0004958428644354636,
      "loss": 2.3362,
      "step": 63036
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7446445226669312,
      "learning_rate": 0.0004958397657261147,
      "loss": 2.201,
      "step": 63037
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.670921266078949,
      "learning_rate": 0.0004958366669803552,
      "loss": 2.0679,
      "step": 63038
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.8401482105255127,
      "learning_rate": 0.0004958335681981862,
      "loss": 2.232,
      "step": 63039
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.1996959447860718,
      "learning_rate": 0.0004958304693796079,
      "loss": 2.1106,
      "step": 63040
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.754828155040741,
      "learning_rate": 0.000495827370524621,
      "loss": 2.3206,
      "step": 63041
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.9773190021514893,
      "learning_rate": 0.0004958242716332262,
      "loss": 2.6298,
      "step": 63042
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.0840599536895752,
      "learning_rate": 0.000495821172705424,
      "loss": 1.9784,
      "step": 63043
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7547228336334229,
      "learning_rate": 0.0004958180737412148,
      "loss": 2.1811,
      "step": 63044
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.8067383766174316,
      "learning_rate": 0.0004958149747405993,
      "loss": 2.364,
      "step": 63045
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7677457928657532,
      "learning_rate": 0.0004958118757035782,
      "loss": 2.267,
      "step": 63046
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.8877757787704468,
      "learning_rate": 0.000495808776630152,
      "loss": 2.2742,
      "step": 63047
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.8263599276542664,
      "learning_rate": 0.0004958056775203213,
      "loss": 1.8394,
      "step": 63048
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.8231484889984131,
      "learning_rate": 0.0004958025783740865,
      "loss": 2.1346,
      "step": 63049
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.1372461318969727,
      "learning_rate": 0.0004957994791914483,
      "loss": 2.2953,
      "step": 63050
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.9331520199775696,
      "learning_rate": 0.0004957963799724074,
      "loss": 2.2072,
      "step": 63051
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.9103841781616211,
      "learning_rate": 0.0004957932807169642,
      "loss": 2.0792,
      "step": 63052
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.2583458423614502,
      "learning_rate": 0.0004957901814251194,
      "loss": 2.2822,
      "step": 63053
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.8460850119590759,
      "learning_rate": 0.0004957870820968734,
      "loss": 2.2775,
      "step": 63054
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7427467107772827,
      "learning_rate": 0.0004957839827322269,
      "loss": 2.1865,
      "step": 63055
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.8237013220787048,
      "learning_rate": 0.0004957808833311806,
      "loss": 2.184,
      "step": 63056
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.8949776887893677,
      "learning_rate": 0.0004957777838937347,
      "loss": 2.2008,
      "step": 63057
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7959478497505188,
      "learning_rate": 0.0004957746844198902,
      "loss": 2.199,
      "step": 63058
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7597651481628418,
      "learning_rate": 0.0004957715849096474,
      "loss": 2.3631,
      "step": 63059
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7628335356712341,
      "learning_rate": 0.000495768485363007,
      "loss": 2.2927,
      "step": 63060
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7749745845794678,
      "learning_rate": 0.0004957653857799696,
      "loss": 2.1928,
      "step": 63061
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7667270302772522,
      "learning_rate": 0.0004957622861605357,
      "loss": 2.3687,
      "step": 63062
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7212292551994324,
      "learning_rate": 0.0004957591865047058,
      "loss": 2.1344,
      "step": 63063
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7058950066566467,
      "learning_rate": 0.0004957560868124806,
      "loss": 2.0869,
      "step": 63064
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7857851982116699,
      "learning_rate": 0.0004957529870838607,
      "loss": 2.4593,
      "step": 63065
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.8022578954696655,
      "learning_rate": 0.0004957498873188466,
      "loss": 2.2706,
      "step": 63066
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.774364173412323,
      "learning_rate": 0.0004957467875174389,
      "loss": 2.4209,
      "step": 63067
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.6638291478157043,
      "learning_rate": 0.0004957436876796382,
      "loss": 2.1427,
      "step": 63068
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7799988985061646,
      "learning_rate": 0.000495740587805445,
      "loss": 2.0349,
      "step": 63069
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.763979434967041,
      "learning_rate": 0.0004957374878948599,
      "loss": 2.2708,
      "step": 63070
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7438645362854004,
      "learning_rate": 0.0004957343879478836,
      "loss": 2.1524,
      "step": 63071
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7372168302536011,
      "learning_rate": 0.0004957312879645165,
      "loss": 2.1109,
      "step": 63072
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7672694325447083,
      "learning_rate": 0.0004957281879447594,
      "loss": 2.2304,
      "step": 63073
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7508949041366577,
      "learning_rate": 0.0004957250878886126,
      "loss": 2.2665,
      "step": 63074
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.8280650973320007,
      "learning_rate": 0.0004957219877960769,
      "loss": 2.2781,
      "step": 63075
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7017994523048401,
      "learning_rate": 0.0004957188876671527,
      "loss": 2.3288,
      "step": 63076
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.6966469883918762,
      "learning_rate": 0.0004957157875018407,
      "loss": 2.345,
      "step": 63077
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.8465394377708435,
      "learning_rate": 0.0004957126873001415,
      "loss": 2.2234,
      "step": 63078
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.750724196434021,
      "learning_rate": 0.0004957095870620555,
      "loss": 2.1879,
      "step": 63079
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.8015019297599792,
      "learning_rate": 0.0004957064867875836,
      "loss": 2.4098,
      "step": 63080
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7092445492744446,
      "learning_rate": 0.000495703386476726,
      "loss": 2.1571,
      "step": 63081
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7008846998214722,
      "learning_rate": 0.0004957002861294835,
      "loss": 2.3114,
      "step": 63082
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.809903621673584,
      "learning_rate": 0.0004956971857458565,
      "loss": 2.2362,
      "step": 63083
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.6361804008483887,
      "learning_rate": 0.0004956940853258459,
      "loss": 2.1045,
      "step": 63084
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.766194760799408,
      "learning_rate": 0.000495690984869452,
      "loss": 2.149,
      "step": 63085
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7406768798828125,
      "learning_rate": 0.0004956878843766754,
      "loss": 2.2909,
      "step": 63086
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7769048810005188,
      "learning_rate": 0.0004956847838475167,
      "loss": 2.243,
      "step": 63087
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7548275589942932,
      "learning_rate": 0.0004956816832819766,
      "loss": 2.1369,
      "step": 63088
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7040160298347473,
      "learning_rate": 0.0004956785826800555,
      "loss": 2.1775,
      "step": 63089
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.8134738802909851,
      "learning_rate": 0.0004956754820417541,
      "loss": 2.2753,
      "step": 63090
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.8201074600219727,
      "learning_rate": 0.000495672381367073,
      "loss": 2.2568,
      "step": 63091
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7226707935333252,
      "learning_rate": 0.0004956692806560127,
      "loss": 2.127,
      "step": 63092
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7234291434288025,
      "learning_rate": 0.0004956661799085736,
      "loss": 1.949,
      "step": 63093
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7271515727043152,
      "learning_rate": 0.0004956630791247566,
      "loss": 2.2294,
      "step": 63094
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.9435917139053345,
      "learning_rate": 0.0004956599783045621,
      "loss": 2.3733,
      "step": 63095
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.8078430891036987,
      "learning_rate": 0.0004956568774479908,
      "loss": 2.1022,
      "step": 63096
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.800646960735321,
      "learning_rate": 0.0004956537765550431,
      "loss": 2.2681,
      "step": 63097
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.8025297522544861,
      "learning_rate": 0.0004956506756257197,
      "loss": 2.2676,
      "step": 63098
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.8970867395401001,
      "learning_rate": 0.0004956475746600212,
      "loss": 2.2845,
      "step": 63099
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7723516821861267,
      "learning_rate": 0.0004956444736579481,
      "loss": 2.0225,
      "step": 63100
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.76215660572052,
      "learning_rate": 0.000495641372619501,
      "loss": 2.1891,
      "step": 63101
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7897449135780334,
      "learning_rate": 0.0004956382715446805,
      "loss": 2.2322,
      "step": 63102
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.8175381422042847,
      "learning_rate": 0.000495635170433487,
      "loss": 2.3499,
      "step": 63103
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.8859596252441406,
      "learning_rate": 0.0004956320692859214,
      "loss": 2.0585,
      "step": 63104
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7629234790802002,
      "learning_rate": 0.000495628968101984,
      "loss": 2.1852,
      "step": 63105
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7320358753204346,
      "learning_rate": 0.0004956258668816756,
      "loss": 1.9989,
      "step": 63106
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7689608335494995,
      "learning_rate": 0.0004956227656249965,
      "loss": 2.4836,
      "step": 63107
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7765281796455383,
      "learning_rate": 0.0004956196643319476,
      "loss": 2.3059,
      "step": 63108
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.8321325182914734,
      "learning_rate": 0.0004956165630025292,
      "loss": 2.0118,
      "step": 63109
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.803693950176239,
      "learning_rate": 0.000495613461636742,
      "loss": 2.1833,
      "step": 63110
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7416863441467285,
      "learning_rate": 0.0004956103602345866,
      "loss": 2.0384,
      "step": 63111
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.8821856379508972,
      "learning_rate": 0.0004956072587960636,
      "loss": 2.3214,
      "step": 63112
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7463772296905518,
      "learning_rate": 0.0004956041573211733,
      "loss": 2.0956,
      "step": 63113
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.740249752998352,
      "learning_rate": 0.0004956010558099167,
      "loss": 2.4186,
      "step": 63114
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.8918431997299194,
      "learning_rate": 0.0004955979542622941,
      "loss": 2.2342,
      "step": 63115
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.8039065003395081,
      "learning_rate": 0.0004955948526783061,
      "loss": 2.4492,
      "step": 63116
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.8073383569717407,
      "learning_rate": 0.0004955917510579535,
      "loss": 2.3058,
      "step": 63117
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.8157657980918884,
      "learning_rate": 0.0004955886494012366,
      "loss": 2.1677,
      "step": 63118
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.8541283011436462,
      "learning_rate": 0.0004955855477081561,
      "loss": 2.0793,
      "step": 63119
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7647554278373718,
      "learning_rate": 0.0004955824459787125,
      "loss": 2.0612,
      "step": 63120
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7758627533912659,
      "learning_rate": 0.0004955793442129064,
      "loss": 2.081,
      "step": 63121
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7033376693725586,
      "learning_rate": 0.0004955762424107385,
      "loss": 2.25,
      "step": 63122
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.9082615971565247,
      "learning_rate": 0.0004955731405722093,
      "loss": 2.126,
      "step": 63123
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.8016021847724915,
      "learning_rate": 0.0004955700386973192,
      "loss": 2.2646,
      "step": 63124
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7079716920852661,
      "learning_rate": 0.000495566936786069,
      "loss": 2.0223,
      "step": 63125
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.9468523859977722,
      "learning_rate": 0.0004955638348384593,
      "loss": 2.1844,
      "step": 63126
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.9277777671813965,
      "learning_rate": 0.0004955607328544906,
      "loss": 2.117,
      "step": 63127
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7869836091995239,
      "learning_rate": 0.0004955576308341633,
      "loss": 2.1477,
      "step": 63128
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.8583282828330994,
      "learning_rate": 0.0004955545287774783,
      "loss": 2.1404,
      "step": 63129
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.1841284036636353,
      "learning_rate": 0.000495551426684436,
      "loss": 2.3188,
      "step": 63130
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7994307279586792,
      "learning_rate": 0.000495548324555037,
      "loss": 2.1991,
      "step": 63131
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7995167970657349,
      "learning_rate": 0.0004955452223892818,
      "loss": 2.2094,
      "step": 63132
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.9970074892044067,
      "learning_rate": 0.0004955421201871712,
      "loss": 2.3022,
      "step": 63133
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.9894726872444153,
      "learning_rate": 0.0004955390179487055,
      "loss": 2.0771,
      "step": 63134
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.737744152545929,
      "learning_rate": 0.0004955359156738855,
      "loss": 2.2037,
      "step": 63135
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.8645488023757935,
      "learning_rate": 0.0004955328133627115,
      "loss": 2.2976,
      "step": 63136
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.0557770729064941,
      "learning_rate": 0.0004955297110151844,
      "loss": 2.4173,
      "step": 63137
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.8171746134757996,
      "learning_rate": 0.0004955266086313047,
      "loss": 2.0188,
      "step": 63138
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7259312868118286,
      "learning_rate": 0.0004955235062110728,
      "loss": 2.1074,
      "step": 63139
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.9174471497535706,
      "learning_rate": 0.0004955204037544894,
      "loss": 2.0859,
      "step": 63140
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.9324105978012085,
      "learning_rate": 0.0004955173012615552,
      "loss": 2.2609,
      "step": 63141
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.8452011346817017,
      "learning_rate": 0.0004955141987322705,
      "loss": 2.2314,
      "step": 63142
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7472535371780396,
      "learning_rate": 0.000495511096166636,
      "loss": 2.0141,
      "step": 63143
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.9345763325691223,
      "learning_rate": 0.0004955079935646524,
      "loss": 2.2711,
      "step": 63144
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.0384591817855835,
      "learning_rate": 0.00049550489092632,
      "loss": 2.3744,
      "step": 63145
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.702226996421814,
      "learning_rate": 0.0004955017882516397,
      "loss": 2.1273,
      "step": 63146
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7750551104545593,
      "learning_rate": 0.000495498685540612,
      "loss": 2.3765,
      "step": 63147
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.8796821236610413,
      "learning_rate": 0.0004954955827932373,
      "loss": 2.1502,
      "step": 63148
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.9031717777252197,
      "learning_rate": 0.0004954924800095162,
      "loss": 2.2409,
      "step": 63149
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7304689288139343,
      "learning_rate": 0.0004954893771894495,
      "loss": 2.1393,
      "step": 63150
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.861067533493042,
      "learning_rate": 0.0004954862743330376,
      "loss": 2.4082,
      "step": 63151
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.8952597379684448,
      "learning_rate": 0.000495483171440281,
      "loss": 2.2633,
      "step": 63152
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7197098135948181,
      "learning_rate": 0.0004954800685111806,
      "loss": 2.2018,
      "step": 63153
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7503072023391724,
      "learning_rate": 0.0004954769655457366,
      "loss": 2.0379,
      "step": 63154
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7139660120010376,
      "learning_rate": 0.0004954738625439497,
      "loss": 2.1546,
      "step": 63155
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7604289054870605,
      "learning_rate": 0.0004954707595058207,
      "loss": 2.0843,
      "step": 63156
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7763597965240479,
      "learning_rate": 0.0004954676564313499,
      "loss": 2.1773,
      "step": 63157
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7174656391143799,
      "learning_rate": 0.000495464553320538,
      "loss": 2.3442,
      "step": 63158
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7755647301673889,
      "learning_rate": 0.0004954614501733854,
      "loss": 2.3015,
      "step": 63159
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7152485251426697,
      "learning_rate": 0.000495458346989893,
      "loss": 2.0713,
      "step": 63160
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7734143733978271,
      "learning_rate": 0.0004954552437700612,
      "loss": 2.3107,
      "step": 63161
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7705901861190796,
      "learning_rate": 0.0004954521405138905,
      "loss": 2.3017,
      "step": 63162
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7967904806137085,
      "learning_rate": 0.0004954490372213817,
      "loss": 2.1426,
      "step": 63163
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7404935359954834,
      "learning_rate": 0.000495445933892535,
      "loss": 2.2574,
      "step": 63164
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7232121825218201,
      "learning_rate": 0.0004954428305273513,
      "loss": 2.2319,
      "step": 63165
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.0129526853561401,
      "learning_rate": 0.0004954397271258312,
      "loss": 2.1482,
      "step": 63166
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7362951636314392,
      "learning_rate": 0.0004954366236879751,
      "loss": 2.3624,
      "step": 63167
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.9123727083206177,
      "learning_rate": 0.0004954335202137836,
      "loss": 2.2242,
      "step": 63168
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.9880649447441101,
      "learning_rate": 0.0004954304167032573,
      "loss": 2.3108,
      "step": 63169
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7843152284622192,
      "learning_rate": 0.0004954273131563969,
      "loss": 2.2145,
      "step": 63170
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7380250096321106,
      "learning_rate": 0.0004954242095732028,
      "loss": 2.1889,
      "step": 63171
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.8495502471923828,
      "learning_rate": 0.0004954211059536757,
      "loss": 2.3917,
      "step": 63172
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.9115986824035645,
      "learning_rate": 0.0004954180022978161,
      "loss": 2.2946,
      "step": 63173
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7150274515151978,
      "learning_rate": 0.0004954148986056246,
      "loss": 2.2319,
      "step": 63174
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7940348982810974,
      "learning_rate": 0.0004954117948771018,
      "loss": 2.1581,
      "step": 63175
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7778117060661316,
      "learning_rate": 0.0004954086911122483,
      "loss": 2.1887,
      "step": 63176
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.941494345664978,
      "learning_rate": 0.0004954055873110645,
      "loss": 2.2009,
      "step": 63177
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7009589672088623,
      "learning_rate": 0.0004954024834735513,
      "loss": 2.1203,
      "step": 63178
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7828337550163269,
      "learning_rate": 0.0004953993795997089,
      "loss": 2.0619,
      "step": 63179
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.9513159394264221,
      "learning_rate": 0.0004953962756895381,
      "loss": 2.2847,
      "step": 63180
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.759240984916687,
      "learning_rate": 0.0004953931717430396,
      "loss": 2.0467,
      "step": 63181
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7298582792282104,
      "learning_rate": 0.0004953900677602136,
      "loss": 2.2372,
      "step": 63182
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.824064314365387,
      "learning_rate": 0.0004953869637410611,
      "loss": 2.1507,
      "step": 63183
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7644199132919312,
      "learning_rate": 0.0004953838596855823,
      "loss": 2.2841,
      "step": 63184
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.8441781997680664,
      "learning_rate": 0.0004953807555937781,
      "loss": 2.2324,
      "step": 63185
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7414883971214294,
      "learning_rate": 0.0004953776514656489,
      "loss": 2.1807,
      "step": 63186
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.9336175918579102,
      "learning_rate": 0.0004953745473011953,
      "loss": 2.2283,
      "step": 63187
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7286728024482727,
      "learning_rate": 0.0004953714431004178,
      "loss": 2.0048,
      "step": 63188
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7369021773338318,
      "learning_rate": 0.0004953683388633171,
      "loss": 2.193,
      "step": 63189
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7756455540657043,
      "learning_rate": 0.0004953652345898938,
      "loss": 2.3027,
      "step": 63190
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.8290222883224487,
      "learning_rate": 0.0004953621302801485,
      "loss": 2.4046,
      "step": 63191
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.8479223847389221,
      "learning_rate": 0.0004953590259340814,
      "loss": 2.424,
      "step": 63192
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7702842354774475,
      "learning_rate": 0.0004953559215516936,
      "loss": 2.1739,
      "step": 63193
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.68068927526474,
      "learning_rate": 0.0004953528171329854,
      "loss": 2.1239,
      "step": 63194
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.8341148495674133,
      "learning_rate": 0.0004953497126779574,
      "loss": 2.4157,
      "step": 63195
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.8866562843322754,
      "learning_rate": 0.0004953466081866103,
      "loss": 2.48,
      "step": 63196
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.9850080013275146,
      "learning_rate": 0.0004953435036589444,
      "loss": 2.2264,
      "step": 63197
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.725357174873352,
      "learning_rate": 0.0004953403990949605,
      "loss": 2.0775,
      "step": 63198
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.8646631240844727,
      "learning_rate": 0.0004953372944946592,
      "loss": 2.0745,
      "step": 63199
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.8885714411735535,
      "learning_rate": 0.000495334189858041,
      "loss": 2.2254,
      "step": 63200
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.8115989565849304,
      "learning_rate": 0.0004953310851851064,
      "loss": 2.3952,
      "step": 63201
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7611769437789917,
      "learning_rate": 0.0004953279804758562,
      "loss": 2.1896,
      "step": 63202
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.868818461894989,
      "learning_rate": 0.0004953248757302907,
      "loss": 2.2961,
      "step": 63203
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.9420356750488281,
      "learning_rate": 0.0004953217709484108,
      "loss": 2.2449,
      "step": 63204
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.8240721225738525,
      "learning_rate": 0.0004953186661302168,
      "loss": 2.0535,
      "step": 63205
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.8029991388320923,
      "learning_rate": 0.0004953155612757094,
      "loss": 2.2784,
      "step": 63206
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.8744217157363892,
      "learning_rate": 0.000495312456384889,
      "loss": 2.3003,
      "step": 63207
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7922682166099548,
      "learning_rate": 0.0004953093514577565,
      "loss": 2.1142,
      "step": 63208
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7314420342445374,
      "learning_rate": 0.0004953062464943122,
      "loss": 2.1454,
      "step": 63209
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.8509334325790405,
      "learning_rate": 0.0004953031414945568,
      "loss": 2.0655,
      "step": 63210
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.9592381119728088,
      "learning_rate": 0.000495300036458491,
      "loss": 2.138,
      "step": 63211
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.8486562967300415,
      "learning_rate": 0.000495296931386115,
      "loss": 2.0957,
      "step": 63212
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7283416986465454,
      "learning_rate": 0.0004952938262774298,
      "loss": 2.3565,
      "step": 63213
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.0224189758300781,
      "learning_rate": 0.0004952907211324357,
      "loss": 2.3971,
      "step": 63214
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7468257546424866,
      "learning_rate": 0.0004952876159511335,
      "loss": 2.1321,
      "step": 63215
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7637191414833069,
      "learning_rate": 0.0004952845107335235,
      "loss": 2.3546,
      "step": 63216
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7845752239227295,
      "learning_rate": 0.0004952814054796065,
      "loss": 2.1338,
      "step": 63217
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.8834673762321472,
      "learning_rate": 0.0004952783001893829,
      "loss": 2.4132,
      "step": 63218
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7376897931098938,
      "learning_rate": 0.0004952751948628535,
      "loss": 2.1699,
      "step": 63219
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.8156286478042603,
      "learning_rate": 0.0004952720895000187,
      "loss": 2.1562,
      "step": 63220
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7521290183067322,
      "learning_rate": 0.0004952689841008791,
      "loss": 2.0402,
      "step": 63221
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.8236032724380493,
      "learning_rate": 0.0004952658786654354,
      "loss": 2.2592,
      "step": 63222
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7971669435501099,
      "learning_rate": 0.0004952627731936879,
      "loss": 2.3586,
      "step": 63223
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7164273858070374,
      "learning_rate": 0.0004952596676856376,
      "loss": 2.1699,
      "step": 63224
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7614372968673706,
      "learning_rate": 0.0004952565621412847,
      "loss": 2.4212,
      "step": 63225
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.769858181476593,
      "learning_rate": 0.00049525345656063,
      "loss": 2.1086,
      "step": 63226
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7138637900352478,
      "learning_rate": 0.000495250350943674,
      "loss": 2.2743,
      "step": 63227
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.8885826468467712,
      "learning_rate": 0.0004952472452904171,
      "loss": 2.3013,
      "step": 63228
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7014029026031494,
      "learning_rate": 0.0004952441396008602,
      "loss": 1.9868,
      "step": 63229
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7020676732063293,
      "learning_rate": 0.0004952410338750038,
      "loss": 2.1501,
      "step": 63230
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.8751996159553528,
      "learning_rate": 0.0004952379281128482,
      "loss": 2.2544,
      "step": 63231
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.8180898427963257,
      "learning_rate": 0.0004952348223143943,
      "loss": 2.3682,
      "step": 63232
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.8068768382072449,
      "learning_rate": 0.0004952317164796426,
      "loss": 2.2443,
      "step": 63233
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7966864705085754,
      "learning_rate": 0.0004952286106085935,
      "loss": 2.2563,
      "step": 63234
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.0473456382751465,
      "learning_rate": 0.0004952255047012479,
      "loss": 2.274,
      "step": 63235
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7722816467285156,
      "learning_rate": 0.0004952223987576062,
      "loss": 1.9067,
      "step": 63236
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.6949511766433716,
      "learning_rate": 0.0004952192927776688,
      "loss": 2.1796,
      "step": 63237
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.8907098770141602,
      "learning_rate": 0.0004952161867614364,
      "loss": 2.1163,
      "step": 63238
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7683972120285034,
      "learning_rate": 0.0004952130807089099,
      "loss": 2.1484,
      "step": 63239
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7450905442237854,
      "learning_rate": 0.0004952099746200894,
      "loss": 2.3244,
      "step": 63240
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7205768823623657,
      "learning_rate": 0.0004952068684949758,
      "loss": 2.0904,
      "step": 63241
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7804552912712097,
      "learning_rate": 0.0004952037623335695,
      "loss": 2.1468,
      "step": 63242
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7409148812294006,
      "learning_rate": 0.0004952006561358711,
      "loss": 2.241,
      "step": 63243
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7130556702613831,
      "learning_rate": 0.0004951975499018813,
      "loss": 2.2799,
      "step": 63244
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7322367429733276,
      "learning_rate": 0.0004951944436316005,
      "loss": 2.2849,
      "step": 63245
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7960293292999268,
      "learning_rate": 0.0004951913373250293,
      "loss": 2.2784,
      "step": 63246
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7302271127700806,
      "learning_rate": 0.0004951882309821685,
      "loss": 2.1344,
      "step": 63247
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.6751869916915894,
      "learning_rate": 0.0004951851246030185,
      "loss": 2.0555,
      "step": 63248
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7344872355461121,
      "learning_rate": 0.0004951820181875798,
      "loss": 2.2219,
      "step": 63249
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7610346674919128,
      "learning_rate": 0.0004951789117358532,
      "loss": 2.2177,
      "step": 63250
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.8951168060302734,
      "learning_rate": 0.0004951758052478392,
      "loss": 2.1812,
      "step": 63251
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.8656967282295227,
      "learning_rate": 0.0004951726987235382,
      "loss": 2.2084,
      "step": 63252
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.8459174633026123,
      "learning_rate": 0.0004951695921629509,
      "loss": 2.2722,
      "step": 63253
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.0738226175308228,
      "learning_rate": 0.0004951664855660779,
      "loss": 2.088,
      "step": 63254
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.8028287291526794,
      "learning_rate": 0.0004951633789329199,
      "loss": 2.403,
      "step": 63255
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.0511075258255005,
      "learning_rate": 0.0004951602722634772,
      "loss": 2.2247,
      "step": 63256
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.924311101436615,
      "learning_rate": 0.0004951571655577506,
      "loss": 2.2914,
      "step": 63257
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.8201566934585571,
      "learning_rate": 0.0004951540588157405,
      "loss": 2.3011,
      "step": 63258
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.8550922274589539,
      "learning_rate": 0.0004951509520374478,
      "loss": 2.1431,
      "step": 63259
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.8499307036399841,
      "learning_rate": 0.0004951478452228726,
      "loss": 2.2013,
      "step": 63260
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7870831489562988,
      "learning_rate": 0.0004951447383720159,
      "loss": 2.4881,
      "step": 63261
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.8607111573219299,
      "learning_rate": 0.000495141631484878,
      "loss": 2.2622,
      "step": 63262
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.8150700330734253,
      "learning_rate": 0.0004951385245614596,
      "loss": 2.3901,
      "step": 63263
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.8516589999198914,
      "learning_rate": 0.0004951354176017613,
      "loss": 2.2212,
      "step": 63264
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7896721363067627,
      "learning_rate": 0.0004951323106057837,
      "loss": 2.2513,
      "step": 63265
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7687722444534302,
      "learning_rate": 0.0004951292035735273,
      "loss": 2.0656,
      "step": 63266
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7308490872383118,
      "learning_rate": 0.0004951260965049926,
      "loss": 2.2895,
      "step": 63267
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.8323275446891785,
      "learning_rate": 0.0004951229894001803,
      "loss": 1.9863,
      "step": 63268
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7297296524047852,
      "learning_rate": 0.000495119882259091,
      "loss": 2.072,
      "step": 63269
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7868074774742126,
      "learning_rate": 0.0004951167750817252,
      "loss": 2.2192,
      "step": 63270
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.8032678365707397,
      "learning_rate": 0.0004951136678680835,
      "loss": 2.2943,
      "step": 63271
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7793741822242737,
      "learning_rate": 0.0004951105606181666,
      "loss": 2.1759,
      "step": 63272
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7590970396995544,
      "learning_rate": 0.0004951074533319749,
      "loss": 2.1735,
      "step": 63273
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.9051471948623657,
      "learning_rate": 0.000495104346009509,
      "loss": 2.4508,
      "step": 63274
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.0029511451721191,
      "learning_rate": 0.0004951012386507695,
      "loss": 2.1189,
      "step": 63275
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7456891536712646,
      "learning_rate": 0.0004950981312557569,
      "loss": 2.4368,
      "step": 63276
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.8484898209571838,
      "learning_rate": 0.0004950950238244721,
      "loss": 2.3117,
      "step": 63277
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.9637928605079651,
      "learning_rate": 0.0004950919163569154,
      "loss": 2.2583,
      "step": 63278
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7972371578216553,
      "learning_rate": 0.0004950888088530875,
      "loss": 2.2335,
      "step": 63279
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.8301893472671509,
      "learning_rate": 0.0004950857013129886,
      "loss": 2.1383,
      "step": 63280
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7717640995979309,
      "learning_rate": 0.0004950825937366199,
      "loss": 2.2786,
      "step": 63281
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7575429081916809,
      "learning_rate": 0.0004950794861239815,
      "loss": 2.2688,
      "step": 63282
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7506011128425598,
      "learning_rate": 0.0004950763784750742,
      "loss": 2.2111,
      "step": 63283
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7803911566734314,
      "learning_rate": 0.0004950732707898985,
      "loss": 2.0772,
      "step": 63284
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.741193950176239,
      "learning_rate": 0.000495070163068455,
      "loss": 2.1512,
      "step": 63285
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7701887488365173,
      "learning_rate": 0.0004950670553107442,
      "loss": 2.1836,
      "step": 63286
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.8499392867088318,
      "learning_rate": 0.0004950639475167669,
      "loss": 2.0998,
      "step": 63287
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.8396535515785217,
      "learning_rate": 0.0004950608396865234,
      "loss": 2.1636,
      "step": 63288
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7287980318069458,
      "learning_rate": 0.0004950577318200144,
      "loss": 2.0498,
      "step": 63289
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7227617502212524,
      "learning_rate": 0.0004950546239172407,
      "loss": 2.1576,
      "step": 63290
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7603517174720764,
      "learning_rate": 0.0004950515159782025,
      "loss": 2.3385,
      "step": 63291
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7053901553153992,
      "learning_rate": 0.0004950484080029006,
      "loss": 2.3892,
      "step": 63292
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7749807238578796,
      "learning_rate": 0.0004950452999913354,
      "loss": 2.0125,
      "step": 63293
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7166591882705688,
      "learning_rate": 0.0004950421919435076,
      "loss": 2.3259,
      "step": 63294
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7596654295921326,
      "learning_rate": 0.000495039083859418,
      "loss": 2.1148,
      "step": 63295
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7243828773498535,
      "learning_rate": 0.0004950359757390668,
      "loss": 2.3046,
      "step": 63296
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7446784973144531,
      "learning_rate": 0.0004950328675824547,
      "loss": 2.1591,
      "step": 63297
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.8158199191093445,
      "learning_rate": 0.0004950297593895823,
      "loss": 2.051,
      "step": 63298
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7818405032157898,
      "learning_rate": 0.0004950266511604502,
      "loss": 2.2696,
      "step": 63299
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.895890474319458,
      "learning_rate": 0.000495023542895059,
      "loss": 2.2755,
      "step": 63300
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7374659180641174,
      "learning_rate": 0.0004950204345934092,
      "loss": 2.0591,
      "step": 63301
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.9024693965911865,
      "learning_rate": 0.0004950173262555015,
      "loss": 2.1182,
      "step": 63302
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.8614178895950317,
      "learning_rate": 0.0004950142178813363,
      "loss": 2.1146,
      "step": 63303
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7239753603935242,
      "learning_rate": 0.0004950111094709144,
      "loss": 2.3008,
      "step": 63304
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.779670774936676,
      "learning_rate": 0.0004950080010242362,
      "loss": 2.2932,
      "step": 63305
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7188035249710083,
      "learning_rate": 0.0004950048925413023,
      "loss": 2.1996,
      "step": 63306
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.8072920441627502,
      "learning_rate": 0.0004950017840221133,
      "loss": 2.2166,
      "step": 63307
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.773296594619751,
      "learning_rate": 0.0004949986754666698,
      "loss": 2.1983,
      "step": 63308
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.8387427926063538,
      "learning_rate": 0.0004949955668749724,
      "loss": 2.3871,
      "step": 63309
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.9011991620063782,
      "learning_rate": 0.0004949924582470216,
      "loss": 2.1617,
      "step": 63310
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.8472950458526611,
      "learning_rate": 0.000494989349582818,
      "loss": 2.3057,
      "step": 63311
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7513204216957092,
      "learning_rate": 0.0004949862408823623,
      "loss": 2.3087,
      "step": 63312
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7000357508659363,
      "learning_rate": 0.0004949831321456549,
      "loss": 2.2532,
      "step": 63313
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.744094967842102,
      "learning_rate": 0.0004949800233726965,
      "loss": 2.1659,
      "step": 63314
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.733931303024292,
      "learning_rate": 0.0004949769145634875,
      "loss": 2.3994,
      "step": 63315
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7522155046463013,
      "learning_rate": 0.0004949738057180288,
      "loss": 2.0826,
      "step": 63316
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.693956732749939,
      "learning_rate": 0.0004949706968363206,
      "loss": 2.0676,
      "step": 63317
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7655739188194275,
      "learning_rate": 0.0004949675879183639,
      "loss": 2.372,
      "step": 63318
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7217229008674622,
      "learning_rate": 0.0004949644789641588,
      "loss": 2.0837,
      "step": 63319
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.8441243767738342,
      "learning_rate": 0.0004949613699737063,
      "loss": 2.4868,
      "step": 63320
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7478524446487427,
      "learning_rate": 0.0004949582609470067,
      "loss": 2.1376,
      "step": 63321
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7785310745239258,
      "learning_rate": 0.0004949551518840607,
      "loss": 2.2949,
      "step": 63322
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7550055384635925,
      "learning_rate": 0.0004949520427848689,
      "loss": 2.1864,
      "step": 63323
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.6941819787025452,
      "learning_rate": 0.0004949489336494319,
      "loss": 2.2307,
      "step": 63324
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7617946267127991,
      "learning_rate": 0.0004949458244777501,
      "loss": 2.2497,
      "step": 63325
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7532244324684143,
      "learning_rate": 0.0004949427152698241,
      "loss": 1.9693,
      "step": 63326
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7194573879241943,
      "learning_rate": 0.0004949396060256548,
      "loss": 2.1719,
      "step": 63327
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7473447322845459,
      "learning_rate": 0.0004949364967452423,
      "loss": 2.2931,
      "step": 63328
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7365733981132507,
      "learning_rate": 0.0004949333874285876,
      "loss": 2.152,
      "step": 63329
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7546910047531128,
      "learning_rate": 0.0004949302780756911,
      "loss": 2.1428,
      "step": 63330
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7711866497993469,
      "learning_rate": 0.0004949271686865534,
      "loss": 2.0543,
      "step": 63331
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7515560388565063,
      "learning_rate": 0.000494924059261175,
      "loss": 2.3311,
      "step": 63332
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7203084826469421,
      "learning_rate": 0.0004949209497995565,
      "loss": 2.3883,
      "step": 63333
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.6766180992126465,
      "learning_rate": 0.0004949178403016985,
      "loss": 2.1774,
      "step": 63334
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7458553910255432,
      "learning_rate": 0.0004949147307676016,
      "loss": 2.3023,
      "step": 63335
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7395573854446411,
      "learning_rate": 0.0004949116211972666,
      "loss": 2.2927,
      "step": 63336
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7842131853103638,
      "learning_rate": 0.0004949085115906936,
      "loss": 2.138,
      "step": 63337
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.9056710600852966,
      "learning_rate": 0.0004949054019478835,
      "loss": 2.1721,
      "step": 63338
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7195139527320862,
      "learning_rate": 0.0004949022922688368,
      "loss": 2.1778,
      "step": 63339
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7651901841163635,
      "learning_rate": 0.0004948991825535541,
      "loss": 2.0928,
      "step": 63340
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7839454412460327,
      "learning_rate": 0.0004948960728020359,
      "loss": 2.1093,
      "step": 63341
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7462080121040344,
      "learning_rate": 0.000494892963014283,
      "loss": 2.128,
      "step": 63342
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7400456666946411,
      "learning_rate": 0.0004948898531902956,
      "loss": 2.3237,
      "step": 63343
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.6784083843231201,
      "learning_rate": 0.0004948867433300746,
      "loss": 1.9272,
      "step": 63344
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.8052452802658081,
      "learning_rate": 0.0004948836334336205,
      "loss": 2.2862,
      "step": 63345
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7922935485839844,
      "learning_rate": 0.000494880523500934,
      "loss": 2.1055,
      "step": 63346
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7106903195381165,
      "learning_rate": 0.0004948774135320153,
      "loss": 2.0998,
      "step": 63347
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7778761386871338,
      "learning_rate": 0.0004948743035268653,
      "loss": 2.2575,
      "step": 63348
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7232834100723267,
      "learning_rate": 0.0004948711934854845,
      "loss": 2.3213,
      "step": 63349
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.827186644077301,
      "learning_rate": 0.0004948680834078734,
      "loss": 2.2818,
      "step": 63350
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.8094179630279541,
      "learning_rate": 0.0004948649732940326,
      "loss": 2.1701,
      "step": 63351
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.8854137063026428,
      "learning_rate": 0.000494861863143963,
      "loss": 2.1768,
      "step": 63352
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7932008504867554,
      "learning_rate": 0.0004948587529576646,
      "loss": 2.108,
      "step": 63353
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.752485454082489,
      "learning_rate": 0.0004948556427351383,
      "loss": 2.1551,
      "step": 63354
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7519184350967407,
      "learning_rate": 0.0004948525324763849,
      "loss": 2.211,
      "step": 63355
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.8140621185302734,
      "learning_rate": 0.0004948494221814045,
      "loss": 2.1873,
      "step": 63356
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.9171683192253113,
      "learning_rate": 0.0004948463118501981,
      "loss": 2.0952,
      "step": 63357
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.0957529544830322,
      "learning_rate": 0.0004948432014827659,
      "loss": 2.1393,
      "step": 63358
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7363902926445007,
      "learning_rate": 0.0004948400910791087,
      "loss": 2.0014,
      "step": 63359
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.9891685247421265,
      "learning_rate": 0.0004948369806392272,
      "loss": 2.1753,
      "step": 63360
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.9075044393539429,
      "learning_rate": 0.0004948338701631218,
      "loss": 2.3419,
      "step": 63361
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7519476413726807,
      "learning_rate": 0.0004948307596507929,
      "loss": 2.3351,
      "step": 63362
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7477948665618896,
      "learning_rate": 0.0004948276491022415,
      "loss": 2.1056,
      "step": 63363
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.9171126484870911,
      "learning_rate": 0.000494824538517468,
      "loss": 2.3366,
      "step": 63364
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.8666170835494995,
      "learning_rate": 0.0004948214278964727,
      "loss": 2.3911,
      "step": 63365
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7758846282958984,
      "learning_rate": 0.0004948183172392565,
      "loss": 1.9483,
      "step": 63366
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.8950388431549072,
      "learning_rate": 0.00049481520654582,
      "loss": 2.3246,
      "step": 63367
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.8578326106071472,
      "learning_rate": 0.0004948120958161637,
      "loss": 2.226,
      "step": 63368
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.810673713684082,
      "learning_rate": 0.000494808985050288,
      "loss": 1.9834,
      "step": 63369
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.0798300504684448,
      "learning_rate": 0.0004948058742481937,
      "loss": 2.4335,
      "step": 63370
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.8335269093513489,
      "learning_rate": 0.0004948027634098813,
      "loss": 2.4888,
      "step": 63371
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7326350808143616,
      "learning_rate": 0.0004947996525353515,
      "loss": 2.2766,
      "step": 63372
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.8192501664161682,
      "learning_rate": 0.0004947965416246046,
      "loss": 2.1704,
      "step": 63373
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.9576520919799805,
      "learning_rate": 0.0004947934306776414,
      "loss": 2.1643,
      "step": 63374
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7295178771018982,
      "learning_rate": 0.0004947903196944625,
      "loss": 2.2331,
      "step": 63375
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7921493053436279,
      "learning_rate": 0.0004947872086750683,
      "loss": 2.3569,
      "step": 63376
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.8955027461051941,
      "learning_rate": 0.0004947840976194596,
      "loss": 2.2201,
      "step": 63377
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.8145337700843811,
      "learning_rate": 0.0004947809865276367,
      "loss": 2.0981,
      "step": 63378
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.8046559691429138,
      "learning_rate": 0.0004947778753996005,
      "loss": 2.0708,
      "step": 63379
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.8639483451843262,
      "learning_rate": 0.0004947747642353513,
      "loss": 2.0912,
      "step": 63380
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7412113547325134,
      "learning_rate": 0.0004947716530348898,
      "loss": 2.4641,
      "step": 63381
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7596668601036072,
      "learning_rate": 0.0004947685417982167,
      "loss": 2.3492,
      "step": 63382
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.8388153314590454,
      "learning_rate": 0.0004947654305253323,
      "loss": 2.1225,
      "step": 63383
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.694413959980011,
      "learning_rate": 0.0004947623192162374,
      "loss": 2.3008,
      "step": 63384
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7891260981559753,
      "learning_rate": 0.0004947592078709326,
      "loss": 2.0456,
      "step": 63385
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.8431167602539062,
      "learning_rate": 0.0004947560964894183,
      "loss": 2.1855,
      "step": 63386
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7440826892852783,
      "learning_rate": 0.000494752985071695,
      "loss": 2.0678,
      "step": 63387
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7449564337730408,
      "learning_rate": 0.0004947498736177638,
      "loss": 2.0466,
      "step": 63388
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.8778873682022095,
      "learning_rate": 0.0004947467621276246,
      "loss": 2.351,
      "step": 63389
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.8909894824028015,
      "learning_rate": 0.0004947436506012784,
      "loss": 2.0939,
      "step": 63390
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.8080518245697021,
      "learning_rate": 0.0004947405390387258,
      "loss": 2.078,
      "step": 63391
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7361974716186523,
      "learning_rate": 0.0004947374274399671,
      "loss": 2.2857,
      "step": 63392
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.049565076828003,
      "learning_rate": 0.0004947343158050031,
      "loss": 2.0907,
      "step": 63393
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.0037286281585693,
      "learning_rate": 0.0004947312041338343,
      "loss": 2.0224,
      "step": 63394
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.8768172860145569,
      "learning_rate": 0.0004947280924264613,
      "loss": 2.1641,
      "step": 63395
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.9773741960525513,
      "learning_rate": 0.0004947249806828846,
      "loss": 2.1158,
      "step": 63396
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7840185165405273,
      "learning_rate": 0.000494721868903105,
      "loss": 2.1508,
      "step": 63397
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7171210050582886,
      "learning_rate": 0.0004947187570871228,
      "loss": 2.2508,
      "step": 63398
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.8081461787223816,
      "learning_rate": 0.0004947156452349388,
      "loss": 2.4084,
      "step": 63399
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7106524109840393,
      "learning_rate": 0.0004947125333465535,
      "loss": 2.3614,
      "step": 63400
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.6954934000968933,
      "learning_rate": 0.0004947094214219674,
      "loss": 2.1455,
      "step": 63401
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.8101313710212708,
      "learning_rate": 0.0004947063094611811,
      "loss": 2.2978,
      "step": 63402
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.6906216740608215,
      "learning_rate": 0.0004947031974641953,
      "loss": 2.5116,
      "step": 63403
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7758182287216187,
      "learning_rate": 0.0004947000854310105,
      "loss": 2.2295,
      "step": 63404
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7271901369094849,
      "learning_rate": 0.0004946969733616273,
      "loss": 2.3309,
      "step": 63405
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7471354007720947,
      "learning_rate": 0.0004946938612560461,
      "loss": 2.2453,
      "step": 63406
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7015705704689026,
      "learning_rate": 0.0004946907491142678,
      "loss": 2.2021,
      "step": 63407
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7005531191825867,
      "learning_rate": 0.0004946876369362927,
      "loss": 2.2945,
      "step": 63408
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.8140375018119812,
      "learning_rate": 0.0004946845247221217,
      "loss": 2.0998,
      "step": 63409
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7337239980697632,
      "learning_rate": 0.000494681412471755,
      "loss": 2.097,
      "step": 63410
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7580767869949341,
      "learning_rate": 0.0004946783001851933,
      "loss": 2.1635,
      "step": 63411
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.8094986081123352,
      "learning_rate": 0.0004946751878624374,
      "loss": 2.4267,
      "step": 63412
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.8041481971740723,
      "learning_rate": 0.0004946720755034875,
      "loss": 2.1349,
      "step": 63413
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7871552109718323,
      "learning_rate": 0.0004946689631083446,
      "loss": 2.2408,
      "step": 63414
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.700259804725647,
      "learning_rate": 0.000494665850677009,
      "loss": 2.3155,
      "step": 63415
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7741629481315613,
      "learning_rate": 0.0004946627382094813,
      "loss": 2.2716,
      "step": 63416
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.9086700081825256,
      "learning_rate": 0.000494659625705762,
      "loss": 2.4715,
      "step": 63417
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7723459601402283,
      "learning_rate": 0.000494656513165852,
      "loss": 2.2868,
      "step": 63418
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.6708052158355713,
      "learning_rate": 0.0004946534005897516,
      "loss": 2.02,
      "step": 63419
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7572329640388489,
      "learning_rate": 0.0004946502879774616,
      "loss": 2.0791,
      "step": 63420
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7447801232337952,
      "learning_rate": 0.0004946471753289822,
      "loss": 2.1746,
      "step": 63421
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.74856036901474,
      "learning_rate": 0.0004946440626443144,
      "loss": 2.199,
      "step": 63422
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7280017733573914,
      "learning_rate": 0.0004946409499234584,
      "loss": 2.15,
      "step": 63423
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7292962074279785,
      "learning_rate": 0.0004946378371664151,
      "loss": 2.3506,
      "step": 63424
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.8737072348594666,
      "learning_rate": 0.000494634724373185,
      "loss": 2.1994,
      "step": 63425
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7745139002799988,
      "learning_rate": 0.0004946316115437686,
      "loss": 2.2588,
      "step": 63426
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7978408336639404,
      "learning_rate": 0.0004946284986781664,
      "loss": 2.2913,
      "step": 63427
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.758148729801178,
      "learning_rate": 0.0004946253857763792,
      "loss": 2.3009,
      "step": 63428
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.8497017025947571,
      "learning_rate": 0.0004946222728384075,
      "loss": 2.221,
      "step": 63429
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7896623015403748,
      "learning_rate": 0.0004946191598642518,
      "loss": 2.038,
      "step": 63430
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7270284295082092,
      "learning_rate": 0.0004946160468539127,
      "loss": 2.1673,
      "step": 63431
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.885384202003479,
      "learning_rate": 0.0004946129338073907,
      "loss": 2.3573,
      "step": 63432
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7516132593154907,
      "learning_rate": 0.0004946098207246867,
      "loss": 2.23,
      "step": 63433
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7215785980224609,
      "learning_rate": 0.0004946067076058009,
      "loss": 2.327,
      "step": 63434
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.804206907749176,
      "learning_rate": 0.0004946035944507341,
      "loss": 2.3089,
      "step": 63435
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7765359878540039,
      "learning_rate": 0.0004946004812594868,
      "loss": 2.239,
      "step": 63436
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.8052076101303101,
      "learning_rate": 0.0004945973680320597,
      "loss": 2.1156,
      "step": 63437
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7071282863616943,
      "learning_rate": 0.0004945942547684532,
      "loss": 1.9882,
      "step": 63438
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.8366701006889343,
      "learning_rate": 0.0004945911414686679,
      "loss": 2.3719,
      "step": 63439
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.74981689453125,
      "learning_rate": 0.0004945880281327045,
      "loss": 2.2603,
      "step": 63440
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7634946703910828,
      "learning_rate": 0.0004945849147605635,
      "loss": 2.1128,
      "step": 63441
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7948757410049438,
      "learning_rate": 0.0004945818013522454,
      "loss": 2.1613,
      "step": 63442
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.6966650485992432,
      "learning_rate": 0.000494578687907751,
      "loss": 2.0974,
      "step": 63443
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.8362727761268616,
      "learning_rate": 0.0004945755744270807,
      "loss": 2.3612,
      "step": 63444
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.8773509860038757,
      "learning_rate": 0.0004945724609102351,
      "loss": 2.476,
      "step": 63445
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.8535646796226501,
      "learning_rate": 0.0004945693473572149,
      "loss": 2.0126,
      "step": 63446
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.0952905416488647,
      "learning_rate": 0.0004945662337680205,
      "loss": 2.1837,
      "step": 63447
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7942946553230286,
      "learning_rate": 0.0004945631201426526,
      "loss": 2.2394,
      "step": 63448
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.8028442859649658,
      "learning_rate": 0.0004945600064811117,
      "loss": 2.2498,
      "step": 63449
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.791602611541748,
      "learning_rate": 0.0004945568927833985,
      "loss": 2.2594,
      "step": 63450
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7537646889686584,
      "learning_rate": 0.0004945537790495134,
      "loss": 2.3871,
      "step": 63451
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7650905847549438,
      "learning_rate": 0.0004945506652794571,
      "loss": 2.1995,
      "step": 63452
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.9630759954452515,
      "learning_rate": 0.0004945475514732302,
      "loss": 2.3968,
      "step": 63453
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.8011728525161743,
      "learning_rate": 0.0004945444376308331,
      "loss": 2.4076,
      "step": 63454
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.6815988421440125,
      "learning_rate": 0.0004945413237522668,
      "loss": 2.2106,
      "step": 63455
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.9718557000160217,
      "learning_rate": 0.0004945382098375315,
      "loss": 2.2335,
      "step": 63456
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.9096270203590393,
      "learning_rate": 0.0004945350958866277,
      "loss": 2.346,
      "step": 63457
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7909678220748901,
      "learning_rate": 0.0004945319818995563,
      "loss": 2.151,
      "step": 63458
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.8161602020263672,
      "learning_rate": 0.0004945288678763177,
      "loss": 2.173,
      "step": 63459
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.1344945430755615,
      "learning_rate": 0.0004945257538169125,
      "loss": 2.3595,
      "step": 63460
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7302626967430115,
      "learning_rate": 0.0004945226397213414,
      "loss": 2.2149,
      "step": 63461
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.8651162385940552,
      "learning_rate": 0.0004945195255896047,
      "loss": 2.3013,
      "step": 63462
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.8592950105667114,
      "learning_rate": 0.0004945164114217032,
      "loss": 2.1458,
      "step": 63463
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.6564608216285706,
      "learning_rate": 0.0004945132972176375,
      "loss": 2.2055,
      "step": 63464
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.8052700161933899,
      "learning_rate": 0.000494510182977408,
      "loss": 2.11,
      "step": 63465
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7266915440559387,
      "learning_rate": 0.0004945070687010154,
      "loss": 2.4272,
      "step": 63466
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7412470579147339,
      "learning_rate": 0.0004945039543884604,
      "loss": 2.1852,
      "step": 63467
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7863991260528564,
      "learning_rate": 0.0004945008400397433,
      "loss": 2.1508,
      "step": 63468
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.8173812031745911,
      "learning_rate": 0.0004944977256548649,
      "loss": 2.2637,
      "step": 63469
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7460832595825195,
      "learning_rate": 0.0004944946112338257,
      "loss": 2.1951,
      "step": 63470
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7672885060310364,
      "learning_rate": 0.0004944914967766261,
      "loss": 2.4328,
      "step": 63471
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7134829759597778,
      "learning_rate": 0.0004944883822832671,
      "loss": 2.0785,
      "step": 63472
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7981805205345154,
      "learning_rate": 0.000494485267753749,
      "loss": 2.3482,
      "step": 63473
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7640036344528198,
      "learning_rate": 0.0004944821531880723,
      "loss": 2.2238,
      "step": 63474
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.760991096496582,
      "learning_rate": 0.0004944790385862378,
      "loss": 2.0598,
      "step": 63475
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7526598572731018,
      "learning_rate": 0.0004944759239482459,
      "loss": 2.205,
      "step": 63476
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7724250555038452,
      "learning_rate": 0.0004944728092740973,
      "loss": 2.1387,
      "step": 63477
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7189925312995911,
      "learning_rate": 0.0004944696945637925,
      "loss": 2.2802,
      "step": 63478
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7737142443656921,
      "learning_rate": 0.0004944665798173322,
      "loss": 2.3294,
      "step": 63479
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7213658690452576,
      "learning_rate": 0.0004944634650347169,
      "loss": 2.3124,
      "step": 63480
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7123737335205078,
      "learning_rate": 0.000494460350215947,
      "loss": 2.1952,
      "step": 63481
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7433260679244995,
      "learning_rate": 0.0004944572353610234,
      "loss": 2.2238,
      "step": 63482
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7122544646263123,
      "learning_rate": 0.0004944541204699465,
      "loss": 2.2438,
      "step": 63483
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.6630758047103882,
      "learning_rate": 0.0004944510055427169,
      "loss": 2.1592,
      "step": 63484
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7695346474647522,
      "learning_rate": 0.0004944478905793351,
      "loss": 2.16,
      "step": 63485
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7068188786506653,
      "learning_rate": 0.0004944447755798019,
      "loss": 2.2987,
      "step": 63486
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7083594799041748,
      "learning_rate": 0.0004944416605441176,
      "loss": 2.2362,
      "step": 63487
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7036094665527344,
      "learning_rate": 0.000494438545472283,
      "loss": 2.1345,
      "step": 63488
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7760294079780579,
      "learning_rate": 0.0004944354303642987,
      "loss": 2.4279,
      "step": 63489
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7854889631271362,
      "learning_rate": 0.0004944323152201651,
      "loss": 2.4561,
      "step": 63490
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7710257172584534,
      "learning_rate": 0.0004944292000398829,
      "loss": 2.2991,
      "step": 63491
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7444315552711487,
      "learning_rate": 0.0004944260848234526,
      "loss": 2.3088,
      "step": 63492
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7681970596313477,
      "learning_rate": 0.0004944229695708748,
      "loss": 2.2681,
      "step": 63493
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7230709195137024,
      "learning_rate": 0.0004944198542821501,
      "loss": 2.2793,
      "step": 63494
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7626158595085144,
      "learning_rate": 0.0004944167389572792,
      "loss": 2.2804,
      "step": 63495
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.8176487684249878,
      "learning_rate": 0.0004944136235962624,
      "loss": 2.1056,
      "step": 63496
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7223041653633118,
      "learning_rate": 0.0004944105081991005,
      "loss": 2.2796,
      "step": 63497
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.8791496157646179,
      "learning_rate": 0.000494407392765794,
      "loss": 2.1677,
      "step": 63498
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7282547354698181,
      "learning_rate": 0.0004944042772963435,
      "loss": 2.2771,
      "step": 63499
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.8523054718971252,
      "learning_rate": 0.0004944011617907496,
      "loss": 2.2493,
      "step": 63500
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7398489117622375,
      "learning_rate": 0.0004943980462490128,
      "loss": 2.179,
      "step": 63501
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.8205881118774414,
      "learning_rate": 0.0004943949306711338,
      "loss": 2.5637,
      "step": 63502
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.0227560997009277,
      "learning_rate": 0.0004943918150571131,
      "loss": 2.1469,
      "step": 63503
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.712216854095459,
      "learning_rate": 0.0004943886994069513,
      "loss": 2.4074,
      "step": 63504
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7715297937393188,
      "learning_rate": 0.0004943855837206489,
      "loss": 2.1019,
      "step": 63505
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.8510885834693909,
      "learning_rate": 0.0004943824679982066,
      "loss": 2.5307,
      "step": 63506
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.8752304911613464,
      "learning_rate": 0.000494379352239625,
      "loss": 2.4715,
      "step": 63507
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7353054285049438,
      "learning_rate": 0.0004943762364449046,
      "loss": 2.2466,
      "step": 63508
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.680841326713562,
      "learning_rate": 0.0004943731206140459,
      "loss": 2.1995,
      "step": 63509
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.8813406229019165,
      "learning_rate": 0.0004943700047470496,
      "loss": 2.2476,
      "step": 63510
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.8442094326019287,
      "learning_rate": 0.0004943668888439162,
      "loss": 2.3278,
      "step": 63511
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.769428551197052,
      "learning_rate": 0.0004943637729046465,
      "loss": 2.1372,
      "step": 63512
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.994008481502533,
      "learning_rate": 0.0004943606569292407,
      "loss": 2.4003,
      "step": 63513
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.8557240962982178,
      "learning_rate": 0.0004943575409176997,
      "loss": 2.3256,
      "step": 63514
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.745805561542511,
      "learning_rate": 0.000494354424870024,
      "loss": 2.1995,
      "step": 63515
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.724884569644928,
      "learning_rate": 0.000494351308786214,
      "loss": 2.1603,
      "step": 63516
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.9676904082298279,
      "learning_rate": 0.0004943481926662704,
      "loss": 1.9516,
      "step": 63517
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.795831024646759,
      "learning_rate": 0.000494345076510194,
      "loss": 2.1289,
      "step": 63518
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.790585994720459,
      "learning_rate": 0.0004943419603179851,
      "loss": 2.3334,
      "step": 63519
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7560411691665649,
      "learning_rate": 0.0004943388440896443,
      "loss": 2.2929,
      "step": 63520
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.8074831962585449,
      "learning_rate": 0.0004943357278251722,
      "loss": 2.4837,
      "step": 63521
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.8775436878204346,
      "learning_rate": 0.0004943326115245696,
      "loss": 2.3122,
      "step": 63522
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.8605203032493591,
      "learning_rate": 0.0004943294951878368,
      "loss": 2.1235,
      "step": 63523
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7392141222953796,
      "learning_rate": 0.0004943263788149745,
      "loss": 2.3282,
      "step": 63524
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7573409676551819,
      "learning_rate": 0.0004943232624059832,
      "loss": 2.3346,
      "step": 63525
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7130513787269592,
      "learning_rate": 0.0004943201459608636,
      "loss": 2.1248,
      "step": 63526
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.9500682950019836,
      "learning_rate": 0.0004943170294796162,
      "loss": 2.1487,
      "step": 63527
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.8219395279884338,
      "learning_rate": 0.0004943139129622416,
      "loss": 2.3935,
      "step": 63528
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.8005971312522888,
      "learning_rate": 0.0004943107964087404,
      "loss": 2.1353,
      "step": 63529
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7459281086921692,
      "learning_rate": 0.000494307679819113,
      "loss": 2.2036,
      "step": 63530
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7994155287742615,
      "learning_rate": 0.0004943045631933603,
      "loss": 2.1352,
      "step": 63531
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.6580578088760376,
      "learning_rate": 0.0004943014465314826,
      "loss": 2.2725,
      "step": 63532
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7719026207923889,
      "learning_rate": 0.0004942983298334807,
      "loss": 2.3332,
      "step": 63533
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.732315719127655,
      "learning_rate": 0.0004942952130993551,
      "loss": 2.1806,
      "step": 63534
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7737657427787781,
      "learning_rate": 0.0004942920963291063,
      "loss": 2.3001,
      "step": 63535
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7271219491958618,
      "learning_rate": 0.0004942889795227349,
      "loss": 2.2281,
      "step": 63536
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7393068671226501,
      "learning_rate": 0.0004942858626802416,
      "loss": 2.2464,
      "step": 63537
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.6895623207092285,
      "learning_rate": 0.0004942827458016267,
      "loss": 2.2545,
      "step": 63538
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7430773377418518,
      "learning_rate": 0.000494279628886891,
      "loss": 2.1983,
      "step": 63539
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7323305010795593,
      "learning_rate": 0.0004942765119360353,
      "loss": 1.9242,
      "step": 63540
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7634925842285156,
      "learning_rate": 0.0004942733949490597,
      "loss": 1.8937,
      "step": 63541
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7279618978500366,
      "learning_rate": 0.0004942702779259651,
      "loss": 2.3533,
      "step": 63542
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7571519017219543,
      "learning_rate": 0.0004942671608667519,
      "loss": 2.0748,
      "step": 63543
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7900247573852539,
      "learning_rate": 0.000494264043771421,
      "loss": 2.2387,
      "step": 63544
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7633599042892456,
      "learning_rate": 0.0004942609266399725,
      "loss": 2.1217,
      "step": 63545
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.9043583869934082,
      "learning_rate": 0.0004942578094724073,
      "loss": 2.3999,
      "step": 63546
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7425290942192078,
      "learning_rate": 0.0004942546922687259,
      "loss": 2.3285,
      "step": 63547
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7659754157066345,
      "learning_rate": 0.0004942515750289289,
      "loss": 2.6101,
      "step": 63548
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7051941752433777,
      "learning_rate": 0.0004942484577530169,
      "loss": 2.231,
      "step": 63549
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.784950315952301,
      "learning_rate": 0.0004942453404409904,
      "loss": 2.0926,
      "step": 63550
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7518481612205505,
      "learning_rate": 0.00049424222309285,
      "loss": 2.1234,
      "step": 63551
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.6962295174598694,
      "learning_rate": 0.0004942391057085964,
      "loss": 2.4656,
      "step": 63552
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.8225199580192566,
      "learning_rate": 0.00049423598828823,
      "loss": 2.3228,
      "step": 63553
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7395585775375366,
      "learning_rate": 0.0004942328708317515,
      "loss": 2.2308,
      "step": 63554
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.8230952620506287,
      "learning_rate": 0.0004942297533391613,
      "loss": 2.0367,
      "step": 63555
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7162958979606628,
      "learning_rate": 0.0004942266358104603,
      "loss": 2.2482,
      "step": 63556
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7452098727226257,
      "learning_rate": 0.0004942235182456489,
      "loss": 2.1833,
      "step": 63557
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7155946493148804,
      "learning_rate": 0.0004942204006447276,
      "loss": 2.1574,
      "step": 63558
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7431094646453857,
      "learning_rate": 0.0004942172830076972,
      "loss": 2.1997,
      "step": 63559
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7485486268997192,
      "learning_rate": 0.000494214165334558,
      "loss": 2.2386,
      "step": 63560
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7809941172599792,
      "learning_rate": 0.0004942110476253107,
      "loss": 2.1522,
      "step": 63561
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.8104918599128723,
      "learning_rate": 0.000494207929879956,
      "loss": 2.1899,
      "step": 63562
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.77989262342453,
      "learning_rate": 0.0004942048120984943,
      "loss": 2.2205,
      "step": 63563
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7675466537475586,
      "learning_rate": 0.0004942016942809263,
      "loss": 2.217,
      "step": 63564
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7401816844940186,
      "learning_rate": 0.0004941985764272526,
      "loss": 2.1962,
      "step": 63565
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.8087782263755798,
      "learning_rate": 0.0004941954585374736,
      "loss": 2.3094,
      "step": 63566
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.9716504216194153,
      "learning_rate": 0.00049419234061159,
      "loss": 2.1732,
      "step": 63567
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.8022623062133789,
      "learning_rate": 0.0004941892226496026,
      "loss": 2.2033,
      "step": 63568
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.8079102635383606,
      "learning_rate": 0.0004941861046515115,
      "loss": 2.1724,
      "step": 63569
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.730357825756073,
      "learning_rate": 0.0004941829866173176,
      "loss": 2.5251,
      "step": 63570
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7806104421615601,
      "learning_rate": 0.0004941798685470215,
      "loss": 2.2806,
      "step": 63571
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.8390270471572876,
      "learning_rate": 0.0004941767504406236,
      "loss": 2.2146,
      "step": 63572
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7893356084823608,
      "learning_rate": 0.0004941736322981247,
      "loss": 2.1645,
      "step": 63573
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7256879806518555,
      "learning_rate": 0.0004941705141195251,
      "loss": 2.3408,
      "step": 63574
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.759806215763092,
      "learning_rate": 0.0004941673959048256,
      "loss": 2.279,
      "step": 63575
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7183592319488525,
      "learning_rate": 0.0004941642776540268,
      "loss": 2.2772,
      "step": 63576
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7826083302497864,
      "learning_rate": 0.0004941611593671291,
      "loss": 2.2251,
      "step": 63577
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.823733389377594,
      "learning_rate": 0.0004941580410441331,
      "loss": 2.4325,
      "step": 63578
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7708413004875183,
      "learning_rate": 0.0004941549226850396,
      "loss": 2.1392,
      "step": 63579
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.8471686244010925,
      "learning_rate": 0.000494151804289849,
      "loss": 2.2303,
      "step": 63580
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7606525421142578,
      "learning_rate": 0.0004941486858585619,
      "loss": 2.1576,
      "step": 63581
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.6921285390853882,
      "learning_rate": 0.0004941455673911789,
      "loss": 2.0679,
      "step": 63582
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7680691480636597,
      "learning_rate": 0.0004941424488877005,
      "loss": 2.0712,
      "step": 63583
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7294739484786987,
      "learning_rate": 0.0004941393303481274,
      "loss": 2.0809,
      "step": 63584
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7357841730117798,
      "learning_rate": 0.0004941362117724602,
      "loss": 2.2587,
      "step": 63585
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7575820088386536,
      "learning_rate": 0.0004941330931606994,
      "loss": 2.1006,
      "step": 63586
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.8231324553489685,
      "learning_rate": 0.0004941299745128456,
      "loss": 2.2543,
      "step": 63587
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7796342372894287,
      "learning_rate": 0.0004941268558288993,
      "loss": 2.285,
      "step": 63588
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7069295048713684,
      "learning_rate": 0.0004941237371088612,
      "loss": 2.2843,
      "step": 63589
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7944238781929016,
      "learning_rate": 0.0004941206183527319,
      "loss": 2.2871,
      "step": 63590
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7341016530990601,
      "learning_rate": 0.0004941174995605117,
      "loss": 2.2509,
      "step": 63591
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7249823808670044,
      "learning_rate": 0.0004941143807322017,
      "loss": 2.1459,
      "step": 63592
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.873952329158783,
      "learning_rate": 0.000494111261867802,
      "loss": 2.1151,
      "step": 63593
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7679181098937988,
      "learning_rate": 0.0004941081429673133,
      "loss": 2.2677,
      "step": 63594
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7582430839538574,
      "learning_rate": 0.0004941050240307363,
      "loss": 2.1667,
      "step": 63595
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7383562922477722,
      "learning_rate": 0.0004941019050580715,
      "loss": 2.232,
      "step": 63596
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7533888816833496,
      "learning_rate": 0.0004940987860493194,
      "loss": 2.1946,
      "step": 63597
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7272518277168274,
      "learning_rate": 0.0004940956670044809,
      "loss": 2.2682,
      "step": 63598
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7225375175476074,
      "learning_rate": 0.0004940925479235561,
      "loss": 2.1936,
      "step": 63599
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7265726923942566,
      "learning_rate": 0.000494089428806546,
      "loss": 2.2119,
      "step": 63600
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.8963348269462585,
      "learning_rate": 0.0004940863096534509,
      "loss": 2.1401,
      "step": 63601
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.8803433775901794,
      "learning_rate": 0.0004940831904642716,
      "loss": 2.1062,
      "step": 63602
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.9404058456420898,
      "learning_rate": 0.0004940800712390084,
      "loss": 2.1547,
      "step": 63603
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.6665403246879578,
      "learning_rate": 0.0004940769519776623,
      "loss": 2.1111,
      "step": 63604
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.8363701105117798,
      "learning_rate": 0.0004940738326802335,
      "loss": 2.1694,
      "step": 63605
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.8019984364509583,
      "learning_rate": 0.0004940707133467227,
      "loss": 2.3857,
      "step": 63606
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.8594236969947815,
      "learning_rate": 0.0004940675939771304,
      "loss": 2.346,
      "step": 63607
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7171316742897034,
      "learning_rate": 0.0004940644745714575,
      "loss": 2.3477,
      "step": 63608
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.8226612210273743,
      "learning_rate": 0.0004940613551297042,
      "loss": 2.3657,
      "step": 63609
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.087816596031189,
      "learning_rate": 0.0004940582356518713,
      "loss": 2.3305,
      "step": 63610
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.8592138886451721,
      "learning_rate": 0.0004940551161379592,
      "loss": 2.2777,
      "step": 63611
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.8662639260292053,
      "learning_rate": 0.0004940519965879686,
      "loss": 2.1164,
      "step": 63612
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.8635871410369873,
      "learning_rate": 0.0004940488770019002,
      "loss": 2.2562,
      "step": 63613
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.6969161033630371,
      "learning_rate": 0.0004940457573797544,
      "loss": 2.1591,
      "step": 63614
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7570513486862183,
      "learning_rate": 0.0004940426377215318,
      "loss": 2.3358,
      "step": 63615
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.747449517250061,
      "learning_rate": 0.000494039518027233,
      "loss": 2.173,
      "step": 63616
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.8281515836715698,
      "learning_rate": 0.0004940363982968587,
      "loss": 2.2832,
      "step": 63617
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.8307708501815796,
      "learning_rate": 0.0004940332785304093,
      "loss": 2.1736,
      "step": 63618
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.6991141438484192,
      "learning_rate": 0.0004940301587278853,
      "loss": 1.896,
      "step": 63619
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.8440985679626465,
      "learning_rate": 0.0004940270388892877,
      "loss": 2.3776,
      "step": 63620
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7598373293876648,
      "learning_rate": 0.0004940239190146166,
      "loss": 2.4087,
      "step": 63621
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7837886810302734,
      "learning_rate": 0.0004940207991038729,
      "loss": 2.2308,
      "step": 63622
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7262952327728271,
      "learning_rate": 0.000494017679157057,
      "loss": 2.0891,
      "step": 63623
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7350500822067261,
      "learning_rate": 0.0004940145591741696,
      "loss": 2.1286,
      "step": 63624
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.772672176361084,
      "learning_rate": 0.0004940114391552113,
      "loss": 2.269,
      "step": 63625
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7262282967567444,
      "learning_rate": 0.0004940083191001825,
      "loss": 2.0959,
      "step": 63626
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7233361601829529,
      "learning_rate": 0.000494005199009084,
      "loss": 2.3343,
      "step": 63627
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7858791351318359,
      "learning_rate": 0.0004940020788819161,
      "loss": 2.1575,
      "step": 63628
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7463383078575134,
      "learning_rate": 0.0004939989587186797,
      "loss": 2.1114,
      "step": 63629
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.794489324092865,
      "learning_rate": 0.0004939958385193751,
      "loss": 2.1011,
      "step": 63630
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7351644039154053,
      "learning_rate": 0.0004939927182840031,
      "loss": 2.222,
      "step": 63631
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.8859174251556396,
      "learning_rate": 0.0004939895980125643,
      "loss": 2.1276,
      "step": 63632
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.8171528577804565,
      "learning_rate": 0.000493986477705059,
      "loss": 2.0304,
      "step": 63633
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7547059059143066,
      "learning_rate": 0.000493983357361488,
      "loss": 2.2144,
      "step": 63634
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.6847636699676514,
      "learning_rate": 0.0004939802369818518,
      "loss": 1.9653,
      "step": 63635
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7624940276145935,
      "learning_rate": 0.000493977116566151,
      "loss": 2.1325,
      "step": 63636
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.8528097867965698,
      "learning_rate": 0.0004939739961143863,
      "loss": 2.2647,
      "step": 63637
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7946023344993591,
      "learning_rate": 0.0004939708756265581,
      "loss": 2.4146,
      "step": 63638
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7455548048019409,
      "learning_rate": 0.000493967755102667,
      "loss": 2.0887,
      "step": 63639
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7624172568321228,
      "learning_rate": 0.0004939646345427138,
      "loss": 2.4381,
      "step": 63640
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.758062481880188,
      "learning_rate": 0.0004939615139466988,
      "loss": 2.1764,
      "step": 63641
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7641748785972595,
      "learning_rate": 0.0004939583933146227,
      "loss": 2.0679,
      "step": 63642
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.8200767636299133,
      "learning_rate": 0.000493955272646486,
      "loss": 2.49,
      "step": 63643
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7005994915962219,
      "learning_rate": 0.0004939521519422896,
      "loss": 2.2612,
      "step": 63644
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7548659443855286,
      "learning_rate": 0.0004939490312020336,
      "loss": 2.3661,
      "step": 63645
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.6977386474609375,
      "learning_rate": 0.0004939459104257188,
      "loss": 2.3493,
      "step": 63646
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7852392196655273,
      "learning_rate": 0.000493942789613346,
      "loss": 2.2529,
      "step": 63647
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7486379742622375,
      "learning_rate": 0.0004939396687649154,
      "loss": 2.1227,
      "step": 63648
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.742538332939148,
      "learning_rate": 0.0004939365478804278,
      "loss": 2.4012,
      "step": 63649
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7085237503051758,
      "learning_rate": 0.0004939334269598837,
      "loss": 2.182,
      "step": 63650
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7769843935966492,
      "learning_rate": 0.0004939303060032838,
      "loss": 2.3445,
      "step": 63651
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7397487163543701,
      "learning_rate": 0.0004939271850106285,
      "loss": 2.0891,
      "step": 63652
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7534552216529846,
      "learning_rate": 0.0004939240639819186,
      "loss": 2.3555,
      "step": 63653
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.8354413509368896,
      "learning_rate": 0.0004939209429171544,
      "loss": 2.3618,
      "step": 63654
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.6931816339492798,
      "learning_rate": 0.0004939178218163367,
      "loss": 2.1177,
      "step": 63655
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7930264472961426,
      "learning_rate": 0.0004939147006794661,
      "loss": 2.267,
      "step": 63656
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7692988514900208,
      "learning_rate": 0.000493911579506543,
      "loss": 2.1312,
      "step": 63657
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7698186635971069,
      "learning_rate": 0.0004939084582975681,
      "loss": 2.4173,
      "step": 63658
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.714410126209259,
      "learning_rate": 0.000493905337052542,
      "loss": 2.1248,
      "step": 63659
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7327349781990051,
      "learning_rate": 0.0004939022157714652,
      "loss": 2.0728,
      "step": 63660
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7433609962463379,
      "learning_rate": 0.0004938990944543383,
      "loss": 2.3649,
      "step": 63661
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.6787797808647156,
      "learning_rate": 0.000493895973101162,
      "loss": 2.2267,
      "step": 63662
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.0146831274032593,
      "learning_rate": 0.0004938928517119367,
      "loss": 2.1104,
      "step": 63663
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.8162472248077393,
      "learning_rate": 0.000493889730286663,
      "loss": 2.3952,
      "step": 63664
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.8132292628288269,
      "learning_rate": 0.0004938866088253416,
      "loss": 2.2715,
      "step": 63665
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7567691206932068,
      "learning_rate": 0.0004938834873279731,
      "loss": 2.1755,
      "step": 63666
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.8262996673583984,
      "learning_rate": 0.0004938803657945579,
      "loss": 2.0556,
      "step": 63667
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7136178016662598,
      "learning_rate": 0.0004938772442250966,
      "loss": 1.8963,
      "step": 63668
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.8315795063972473,
      "learning_rate": 0.00049387412261959,
      "loss": 2.2281,
      "step": 63669
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7857686877250671,
      "learning_rate": 0.0004938710009780385,
      "loss": 2.2364,
      "step": 63670
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.9289091229438782,
      "learning_rate": 0.0004938678793004427,
      "loss": 2.2034,
      "step": 63671
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.6938059329986572,
      "learning_rate": 0.0004938647575868032,
      "loss": 2.1875,
      "step": 63672
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.8046945929527283,
      "learning_rate": 0.0004938616358371206,
      "loss": 2.0325,
      "step": 63673
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.9284258484840393,
      "learning_rate": 0.0004938585140513955,
      "loss": 2.303,
      "step": 63674
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7994731068611145,
      "learning_rate": 0.0004938553922296284,
      "loss": 2.2954,
      "step": 63675
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7812797427177429,
      "learning_rate": 0.0004938522703718199,
      "loss": 2.3858,
      "step": 63676
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7606263160705566,
      "learning_rate": 0.0004938491484779707,
      "loss": 2.3769,
      "step": 63677
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7147071361541748,
      "learning_rate": 0.0004938460265480812,
      "loss": 2.2352,
      "step": 63678
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7351016402244568,
      "learning_rate": 0.000493842904582152,
      "loss": 2.0968,
      "step": 63679
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.703957736492157,
      "learning_rate": 0.0004938397825801839,
      "loss": 2.3052,
      "step": 63680
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7702630758285522,
      "learning_rate": 0.0004938366605421773,
      "loss": 2.2646,
      "step": 63681
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7965903282165527,
      "learning_rate": 0.0004938335384681327,
      "loss": 2.1265,
      "step": 63682
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7105894088745117,
      "learning_rate": 0.0004938304163580509,
      "loss": 2.188,
      "step": 63683
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7595103979110718,
      "learning_rate": 0.0004938272942119322,
      "loss": 2.2709,
      "step": 63684
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7040263414382935,
      "learning_rate": 0.0004938241720297776,
      "loss": 2.1509,
      "step": 63685
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7493970990180969,
      "learning_rate": 0.0004938210498115872,
      "loss": 2.2245,
      "step": 63686
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7141203880310059,
      "learning_rate": 0.0004938179275573619,
      "loss": 2.2507,
      "step": 63687
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.9466498494148254,
      "learning_rate": 0.0004938148052671022,
      "loss": 2.4984,
      "step": 63688
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.700893759727478,
      "learning_rate": 0.0004938116829408088,
      "loss": 2.2968,
      "step": 63689
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.8394129872322083,
      "learning_rate": 0.0004938085605784818,
      "loss": 2.2533,
      "step": 63690
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.8433142304420471,
      "learning_rate": 0.0004938054381801224,
      "loss": 2.2934,
      "step": 63691
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.8031313419342041,
      "learning_rate": 0.0004938023157457309,
      "loss": 2.2869,
      "step": 63692
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.774308979511261,
      "learning_rate": 0.0004937991932753078,
      "loss": 2.3389,
      "step": 63693
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.8534994125366211,
      "learning_rate": 0.0004937960707688538,
      "loss": 2.4529,
      "step": 63694
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7765490412712097,
      "learning_rate": 0.0004937929482263694,
      "loss": 2.2957,
      "step": 63695
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7713184952735901,
      "learning_rate": 0.0004937898256478554,
      "loss": 2.0869,
      "step": 63696
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7454174160957336,
      "learning_rate": 0.0004937867030333121,
      "loss": 2.1397,
      "step": 63697
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7060843706130981,
      "learning_rate": 0.0004937835803827402,
      "loss": 1.9183,
      "step": 63698
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7552905082702637,
      "learning_rate": 0.0004937804576961403,
      "loss": 2.2396,
      "step": 63699
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.8319664597511292,
      "learning_rate": 0.0004937773349735128,
      "loss": 2.3075,
      "step": 63700
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7412837147712708,
      "learning_rate": 0.0004937742122148586,
      "loss": 2.5873,
      "step": 63701
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7548253536224365,
      "learning_rate": 0.0004937710894201781,
      "loss": 2.2338,
      "step": 63702
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.8158779144287109,
      "learning_rate": 0.0004937679665894719,
      "loss": 2.1658,
      "step": 63703
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7911279797554016,
      "learning_rate": 0.0004937648437227406,
      "loss": 2.3068,
      "step": 63704
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.724899172782898,
      "learning_rate": 0.0004937617208199848,
      "loss": 1.9651,
      "step": 63705
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.9478419423103333,
      "learning_rate": 0.0004937585978812048,
      "loss": 2.1578,
      "step": 63706
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7247552275657654,
      "learning_rate": 0.0004937554749064015,
      "loss": 2.3776,
      "step": 63707
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7038317918777466,
      "learning_rate": 0.0004937523518955756,
      "loss": 2.1315,
      "step": 63708
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.1144211292266846,
      "learning_rate": 0.0004937492288487273,
      "loss": 2.149,
      "step": 63709
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.708420991897583,
      "learning_rate": 0.0004937461057658574,
      "loss": 2.2684,
      "step": 63710
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7729132175445557,
      "learning_rate": 0.0004937429826469664,
      "loss": 2.4324,
      "step": 63711
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.8659835457801819,
      "learning_rate": 0.0004937398594920551,
      "loss": 2.0984,
      "step": 63712
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.8723471164703369,
      "learning_rate": 0.0004937367363011237,
      "loss": 2.2749,
      "step": 63713
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7822950482368469,
      "learning_rate": 0.000493733613074173,
      "loss": 2.0697,
      "step": 63714
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7898529767990112,
      "learning_rate": 0.0004937304898112037,
      "loss": 2.25,
      "step": 63715
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.8478884100914001,
      "learning_rate": 0.0004937273665122162,
      "loss": 2.0563,
      "step": 63716
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7695105075836182,
      "learning_rate": 0.0004937242431772111,
      "loss": 2.2036,
      "step": 63717
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.738064706325531,
      "learning_rate": 0.000493721119806189,
      "loss": 2.5796,
      "step": 63718
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7753939032554626,
      "learning_rate": 0.0004937179963991504,
      "loss": 2.0771,
      "step": 63719
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.838338315486908,
      "learning_rate": 0.0004937148729560961,
      "loss": 2.2442,
      "step": 63720
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.8021968603134155,
      "learning_rate": 0.0004937117494770265,
      "loss": 1.973,
      "step": 63721
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.6878162622451782,
      "learning_rate": 0.0004937086259619423,
      "loss": 2.2224,
      "step": 63722
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.9643805623054504,
      "learning_rate": 0.0004937055024108439,
      "loss": 2.0812,
      "step": 63723
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.8229042291641235,
      "learning_rate": 0.0004937023788237321,
      "loss": 2.6074,
      "step": 63724
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.6904222369194031,
      "learning_rate": 0.0004936992552006073,
      "loss": 2.1785,
      "step": 63725
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7435920834541321,
      "learning_rate": 0.0004936961315414702,
      "loss": 2.2175,
      "step": 63726
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7930627465248108,
      "learning_rate": 0.0004936930078463213,
      "loss": 2.2357,
      "step": 63727
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7701269388198853,
      "learning_rate": 0.0004936898841151611,
      "loss": 2.3514,
      "step": 63728
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.9005278944969177,
      "learning_rate": 0.0004936867603479904,
      "loss": 2.4781,
      "step": 63729
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7701116800308228,
      "learning_rate": 0.0004936836365448098,
      "loss": 2.0541,
      "step": 63730
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.820648193359375,
      "learning_rate": 0.0004936805127056195,
      "loss": 2.1474,
      "step": 63731
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7778252363204956,
      "learning_rate": 0.0004936773888304205,
      "loss": 2.3511,
      "step": 63732
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.8514323234558105,
      "learning_rate": 0.0004936742649192132,
      "loss": 2.2268,
      "step": 63733
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7172902822494507,
      "learning_rate": 0.0004936711409719982,
      "loss": 2.2068,
      "step": 63734
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7734692096710205,
      "learning_rate": 0.0004936680169887761,
      "loss": 2.0548,
      "step": 63735
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7913393378257751,
      "learning_rate": 0.0004936648929695474,
      "loss": 2.1027,
      "step": 63736
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.8931702375411987,
      "learning_rate": 0.0004936617689143127,
      "loss": 2.3652,
      "step": 63737
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7261145114898682,
      "learning_rate": 0.0004936586448230728,
      "loss": 2.3005,
      "step": 63738
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.8410975933074951,
      "learning_rate": 0.000493655520695828,
      "loss": 2.1809,
      "step": 63739
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7815532684326172,
      "learning_rate": 0.0004936523965325788,
      "loss": 2.2059,
      "step": 63740
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7724355459213257,
      "learning_rate": 0.0004936492723333263,
      "loss": 2.0863,
      "step": 63741
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.3078285455703735,
      "learning_rate": 0.0004936461480980705,
      "loss": 2.2154,
      "step": 63742
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.8500512838363647,
      "learning_rate": 0.0004936430238268123,
      "loss": 2.3756,
      "step": 63743
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7021983861923218,
      "learning_rate": 0.0004936398995195523,
      "loss": 1.9862,
      "step": 63744
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.845080554485321,
      "learning_rate": 0.0004936367751762908,
      "loss": 2.2029,
      "step": 63745
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.835211455821991,
      "learning_rate": 0.0004936336507970287,
      "loss": 1.9323,
      "step": 63746
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.8610734343528748,
      "learning_rate": 0.0004936305263817664,
      "loss": 2.0946,
      "step": 63747
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7724281549453735,
      "learning_rate": 0.0004936274019305046,
      "loss": 2.1234,
      "step": 63748
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7436719536781311,
      "learning_rate": 0.0004936242774432436,
      "loss": 2.1531,
      "step": 63749
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7109465003013611,
      "learning_rate": 0.0004936211529199845,
      "loss": 2.2084,
      "step": 63750
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.8906320333480835,
      "learning_rate": 0.0004936180283607274,
      "loss": 2.2153,
      "step": 63751
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.8156107068061829,
      "learning_rate": 0.000493614903765473,
      "loss": 2.1513,
      "step": 63752
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.6797831654548645,
      "learning_rate": 0.0004936117791342221,
      "loss": 2.133,
      "step": 63753
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.788213312625885,
      "learning_rate": 0.0004936086544669751,
      "loss": 2.0557,
      "step": 63754
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.8061349391937256,
      "learning_rate": 0.0004936055297637325,
      "loss": 2.2092,
      "step": 63755
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.8999096155166626,
      "learning_rate": 0.000493602405024495,
      "loss": 2.3135,
      "step": 63756
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.8305551409721375,
      "learning_rate": 0.0004935992802492631,
      "loss": 2.2554,
      "step": 63757
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.9319154620170593,
      "learning_rate": 0.0004935961554380375,
      "loss": 2.2112,
      "step": 63758
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.751013457775116,
      "learning_rate": 0.0004935930305908187,
      "loss": 2.1291,
      "step": 63759
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7284916639328003,
      "learning_rate": 0.0004935899057076073,
      "loss": 2.2237,
      "step": 63760
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.9187503457069397,
      "learning_rate": 0.0004935867807884039,
      "loss": 2.2398,
      "step": 63761
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.0490036010742188,
      "learning_rate": 0.000493583655833209,
      "loss": 2.0683,
      "step": 63762
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.731597363948822,
      "learning_rate": 0.0004935805308420234,
      "loss": 2.316,
      "step": 63763
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.8820974230766296,
      "learning_rate": 0.0004935774058148474,
      "loss": 2.263,
      "step": 63764
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.9968740344047546,
      "learning_rate": 0.0004935742807516818,
      "loss": 2.2383,
      "step": 63765
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7701035737991333,
      "learning_rate": 0.000493571155652527,
      "loss": 2.2637,
      "step": 63766
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.8635095953941345,
      "learning_rate": 0.0004935680305173837,
      "loss": 2.0771,
      "step": 63767
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.0402206182479858,
      "learning_rate": 0.0004935649053462524,
      "loss": 2.2048,
      "step": 63768
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.954775333404541,
      "learning_rate": 0.0004935617801391338,
      "loss": 2.3621,
      "step": 63769
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.6933115124702454,
      "learning_rate": 0.0004935586548960284,
      "loss": 2.1682,
      "step": 63770
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.8020985126495361,
      "learning_rate": 0.0004935555296169368,
      "loss": 2.2465,
      "step": 63771
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7643812298774719,
      "learning_rate": 0.0004935524043018595,
      "loss": 2.0309,
      "step": 63772
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.8559379577636719,
      "learning_rate": 0.0004935492789507973,
      "loss": 2.2624,
      "step": 63773
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7764820456504822,
      "learning_rate": 0.0004935461535637506,
      "loss": 2.249,
      "step": 63774
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.8212389349937439,
      "learning_rate": 0.0004935430281407198,
      "loss": 2.2293,
      "step": 63775
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7863717675209045,
      "learning_rate": 0.000493539902681706,
      "loss": 2.1901,
      "step": 63776
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.763836681842804,
      "learning_rate": 0.0004935367771867092,
      "loss": 2.1245,
      "step": 63777
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7255064249038696,
      "learning_rate": 0.0004935336516557305,
      "loss": 1.9458,
      "step": 63778
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.8079855442047119,
      "learning_rate": 0.0004935305260887701,
      "loss": 2.2158,
      "step": 63779
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.8076452612876892,
      "learning_rate": 0.0004935274004858288,
      "loss": 2.1041,
      "step": 63780
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7247182726860046,
      "learning_rate": 0.0004935242748469071,
      "loss": 2.0611,
      "step": 63781
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7639736533164978,
      "learning_rate": 0.0004935211491720055,
      "loss": 2.2174,
      "step": 63782
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7068425416946411,
      "learning_rate": 0.0004935180234611246,
      "loss": 2.2106,
      "step": 63783
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7523107528686523,
      "learning_rate": 0.0004935148977142652,
      "loss": 2.248,
      "step": 63784
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.748625636100769,
      "learning_rate": 0.0004935117719314277,
      "loss": 2.2054,
      "step": 63785
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7955629229545593,
      "learning_rate": 0.0004935086461126128,
      "loss": 2.2036,
      "step": 63786
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7874383926391602,
      "learning_rate": 0.0004935055202578209,
      "loss": 2.0781,
      "step": 63787
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7636781930923462,
      "learning_rate": 0.0004935023943670526,
      "loss": 2.1018,
      "step": 63788
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.8320621252059937,
      "learning_rate": 0.0004934992684403086,
      "loss": 2.3052,
      "step": 63789
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.8290020823478699,
      "learning_rate": 0.0004934961424775894,
      "loss": 2.2571,
      "step": 63790
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7568521499633789,
      "learning_rate": 0.0004934930164788957,
      "loss": 2.3728,
      "step": 63791
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.6938093304634094,
      "learning_rate": 0.000493489890444228,
      "loss": 2.1174,
      "step": 63792
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7303239107131958,
      "learning_rate": 0.0004934867643735868,
      "loss": 2.1736,
      "step": 63793
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7299805283546448,
      "learning_rate": 0.0004934836382669728,
      "loss": 2.3729,
      "step": 63794
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7787694334983826,
      "learning_rate": 0.0004934805121243864,
      "loss": 2.167,
      "step": 63795
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.6911339163780212,
      "learning_rate": 0.0004934773859458286,
      "loss": 2.0849,
      "step": 63796
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.6851789355278015,
      "learning_rate": 0.0004934742597312994,
      "loss": 2.3133,
      "step": 63797
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.8354483842849731,
      "learning_rate": 0.0004934711334807999,
      "loss": 2.1197,
      "step": 63798
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.737902045249939,
      "learning_rate": 0.0004934680071943305,
      "loss": 2.1962,
      "step": 63799
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7252547144889832,
      "learning_rate": 0.0004934648808718916,
      "loss": 2.0697,
      "step": 63800
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.8177649974822998,
      "learning_rate": 0.000493461754513484,
      "loss": 2.3257,
      "step": 63801
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.6868706345558167,
      "learning_rate": 0.000493458628119108,
      "loss": 2.2746,
      "step": 63802
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7138316035270691,
      "learning_rate": 0.0004934555016887646,
      "loss": 2.467,
      "step": 63803
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7204694747924805,
      "learning_rate": 0.0004934523752224543,
      "loss": 2.0929,
      "step": 63804
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7500928640365601,
      "learning_rate": 0.0004934492487201772,
      "loss": 2.4844,
      "step": 63805
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7880465984344482,
      "learning_rate": 0.0004934461221819344,
      "loss": 2.2954,
      "step": 63806
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7930500507354736,
      "learning_rate": 0.0004934429956077263,
      "loss": 2.3712,
      "step": 63807
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.6644544005393982,
      "learning_rate": 0.0004934398689975536,
      "loss": 1.8735,
      "step": 63808
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7137426137924194,
      "learning_rate": 0.0004934367423514168,
      "loss": 2.3667,
      "step": 63809
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.809817373752594,
      "learning_rate": 0.0004934336156693164,
      "loss": 2.2794,
      "step": 63810
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7556562423706055,
      "learning_rate": 0.000493430488951253,
      "loss": 1.9698,
      "step": 63811
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.8269071578979492,
      "learning_rate": 0.0004934273621972271,
      "loss": 2.34,
      "step": 63812
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7557011842727661,
      "learning_rate": 0.0004934242354072396,
      "loss": 2.4717,
      "step": 63813
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7689253687858582,
      "learning_rate": 0.0004934211085812908,
      "loss": 2.2417,
      "step": 63814
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7488733530044556,
      "learning_rate": 0.0004934179817193814,
      "loss": 2.2335,
      "step": 63815
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7450089454650879,
      "learning_rate": 0.0004934148548215119,
      "loss": 2.3024,
      "step": 63816
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.6966192126274109,
      "learning_rate": 0.000493411727887683,
      "loss": 2.2842,
      "step": 63817
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.9944416284561157,
      "learning_rate": 0.0004934086009178951,
      "loss": 2.2967,
      "step": 63818
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.8346247673034668,
      "learning_rate": 0.000493405473912149,
      "loss": 2.3701,
      "step": 63819
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7656973004341125,
      "learning_rate": 0.0004934023468704451,
      "loss": 2.3284,
      "step": 63820
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7463918328285217,
      "learning_rate": 0.0004933992197927841,
      "loss": 1.9854,
      "step": 63821
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7545908689498901,
      "learning_rate": 0.0004933960926791664,
      "loss": 2.2421,
      "step": 63822
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7409785985946655,
      "learning_rate": 0.0004933929655295929,
      "loss": 2.1425,
      "step": 63823
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7518628239631653,
      "learning_rate": 0.000493389838344064,
      "loss": 2.2321,
      "step": 63824
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.8745301961898804,
      "learning_rate": 0.0004933867111225802,
      "loss": 2.0036,
      "step": 63825
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7069438099861145,
      "learning_rate": 0.0004933835838651421,
      "loss": 2.0146,
      "step": 63826
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7633357048034668,
      "learning_rate": 0.0004933804565717504,
      "loss": 2.1864,
      "step": 63827
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.9719558358192444,
      "learning_rate": 0.0004933773292424056,
      "loss": 2.4913,
      "step": 63828
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.6847003698348999,
      "learning_rate": 0.0004933742018771083,
      "loss": 2.2177,
      "step": 63829
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7737547159194946,
      "learning_rate": 0.0004933710744758591,
      "loss": 2.2937,
      "step": 63830
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7905339598655701,
      "learning_rate": 0.0004933679470386586,
      "loss": 2.3759,
      "step": 63831
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7705670595169067,
      "learning_rate": 0.0004933648195655073,
      "loss": 2.3242,
      "step": 63832
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.8257127404212952,
      "learning_rate": 0.0004933616920564059,
      "loss": 2.327,
      "step": 63833
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7799992561340332,
      "learning_rate": 0.0004933585645113549,
      "loss": 2.3048,
      "step": 63834
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7777552008628845,
      "learning_rate": 0.0004933554369303549,
      "loss": 2.1988,
      "step": 63835
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7629773020744324,
      "learning_rate": 0.0004933523093134064,
      "loss": 2.1008,
      "step": 63836
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7682941555976868,
      "learning_rate": 0.0004933491816605101,
      "loss": 2.157,
      "step": 63837
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.8258607387542725,
      "learning_rate": 0.0004933460539716665,
      "loss": 2.4024,
      "step": 63838
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7341057658195496,
      "learning_rate": 0.0004933429262468763,
      "loss": 2.3588,
      "step": 63839
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7129825949668884,
      "learning_rate": 0.0004933397984861399,
      "loss": 2.1121,
      "step": 63840
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7198662757873535,
      "learning_rate": 0.0004933366706894581,
      "loss": 2.2252,
      "step": 63841
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7298587560653687,
      "learning_rate": 0.0004933335428568312,
      "loss": 2.0344,
      "step": 63842
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.806437611579895,
      "learning_rate": 0.0004933304149882601,
      "loss": 2.3867,
      "step": 63843
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.6537025570869446,
      "learning_rate": 0.0004933272870837451,
      "loss": 2.0678,
      "step": 63844
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7434987425804138,
      "learning_rate": 0.000493324159143287,
      "loss": 2.1893,
      "step": 63845
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7679725885391235,
      "learning_rate": 0.0004933210311668863,
      "loss": 2.4116,
      "step": 63846
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.9522997140884399,
      "learning_rate": 0.0004933179031545435,
      "loss": 2.1906,
      "step": 63847
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7682219743728638,
      "learning_rate": 0.0004933147751062594,
      "loss": 2.224,
      "step": 63848
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.8265196681022644,
      "learning_rate": 0.0004933116470220342,
      "loss": 2.2353,
      "step": 63849
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.8461392521858215,
      "learning_rate": 0.0004933085189018689,
      "loss": 1.9291,
      "step": 63850
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7220643758773804,
      "learning_rate": 0.0004933053907457639,
      "loss": 2.1443,
      "step": 63851
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.8554041385650635,
      "learning_rate": 0.0004933022625537196,
      "loss": 2.373,
      "step": 63852
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7836366891860962,
      "learning_rate": 0.0004932991343257369,
      "loss": 2.0116,
      "step": 63853
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7827697992324829,
      "learning_rate": 0.0004932960060618163,
      "loss": 2.1552,
      "step": 63854
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.8547815680503845,
      "learning_rate": 0.0004932928777619582,
      "loss": 2.1329,
      "step": 63855
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7495091557502747,
      "learning_rate": 0.0004932897494261633,
      "loss": 2.3042,
      "step": 63856
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7070053219795227,
      "learning_rate": 0.0004932866210544323,
      "loss": 2.174,
      "step": 63857
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.8042032718658447,
      "learning_rate": 0.0004932834926467656,
      "loss": 2.3691,
      "step": 63858
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7098113894462585,
      "learning_rate": 0.0004932803642031637,
      "loss": 2.1996,
      "step": 63859
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.727510392665863,
      "learning_rate": 0.0004932772357236275,
      "loss": 2.4412,
      "step": 63860
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7148349285125732,
      "learning_rate": 0.0004932741072081575,
      "loss": 2.265,
      "step": 63861
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7378888726234436,
      "learning_rate": 0.0004932709786567541,
      "loss": 2.1555,
      "step": 63862
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.8613434433937073,
      "learning_rate": 0.000493267850069418,
      "loss": 2.1191,
      "step": 63863
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7748847007751465,
      "learning_rate": 0.0004932647214461497,
      "loss": 2.273,
      "step": 63864
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7336291670799255,
      "learning_rate": 0.0004932615927869498,
      "loss": 2.4231,
      "step": 63865
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.9391078352928162,
      "learning_rate": 0.0004932584640918191,
      "loss": 2.4798,
      "step": 63866
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7434762120246887,
      "learning_rate": 0.0004932553353607578,
      "loss": 2.1693,
      "step": 63867
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7494570016860962,
      "learning_rate": 0.0004932522065937668,
      "loss": 2.064,
      "step": 63868
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.8732916116714478,
      "learning_rate": 0.0004932490777908465,
      "loss": 2.2386,
      "step": 63869
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.792527973651886,
      "learning_rate": 0.0004932459489519977,
      "loss": 2.3377,
      "step": 63870
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7961409091949463,
      "learning_rate": 0.0004932428200772207,
      "loss": 2.1833,
      "step": 63871
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7872522473335266,
      "learning_rate": 0.0004932396911665162,
      "loss": 2.1767,
      "step": 63872
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.9068117141723633,
      "learning_rate": 0.0004932365622198849,
      "loss": 2.2692,
      "step": 63873
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.6939501166343689,
      "learning_rate": 0.0004932334332373271,
      "loss": 1.9417,
      "step": 63874
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7730509638786316,
      "learning_rate": 0.0004932303042188437,
      "loss": 2.2503,
      "step": 63875
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.731768012046814,
      "learning_rate": 0.0004932271751644351,
      "loss": 2.3372,
      "step": 63876
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.8641507029533386,
      "learning_rate": 0.0004932240460741019,
      "loss": 2.1075,
      "step": 63877
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.81004798412323,
      "learning_rate": 0.0004932209169478447,
      "loss": 2.2874,
      "step": 63878
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7348923087120056,
      "learning_rate": 0.0004932177877856642,
      "loss": 2.191,
      "step": 63879
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7341179251670837,
      "learning_rate": 0.0004932146585875607,
      "loss": 2.2037,
      "step": 63880
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7552443742752075,
      "learning_rate": 0.0004932115293535351,
      "loss": 2.2551,
      "step": 63881
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7317947149276733,
      "learning_rate": 0.0004932084000835878,
      "loss": 2.1953,
      "step": 63882
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.795150637626648,
      "learning_rate": 0.0004932052707777193,
      "loss": 2.3154,
      "step": 63883
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.8082413077354431,
      "learning_rate": 0.0004932021414359303,
      "loss": 2.2007,
      "step": 63884
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7417435050010681,
      "learning_rate": 0.0004931990120582214,
      "loss": 2.1687,
      "step": 63885
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.8204495906829834,
      "learning_rate": 0.0004931958826445932,
      "loss": 2.2782,
      "step": 63886
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.697790265083313,
      "learning_rate": 0.0004931927531950463,
      "loss": 2.4548,
      "step": 63887
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.805283784866333,
      "learning_rate": 0.0004931896237095812,
      "loss": 2.3084,
      "step": 63888
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7306476831436157,
      "learning_rate": 0.0004931864941881985,
      "loss": 1.8666,
      "step": 63889
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.8032324314117432,
      "learning_rate": 0.0004931833646308985,
      "loss": 2.245,
      "step": 63890
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7024685740470886,
      "learning_rate": 0.0004931802350376824,
      "loss": 2.2634,
      "step": 63891
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7542319893836975,
      "learning_rate": 0.0004931771054085504,
      "loss": 2.2956,
      "step": 63892
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7698180675506592,
      "learning_rate": 0.000493173975743503,
      "loss": 2.1087,
      "step": 63893
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7998723983764648,
      "learning_rate": 0.0004931708460425411,
      "loss": 2.3293,
      "step": 63894
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7775143980979919,
      "learning_rate": 0.0004931677163056649,
      "loss": 2.1029,
      "step": 63895
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.8284394145011902,
      "learning_rate": 0.0004931645865328753,
      "loss": 2.3399,
      "step": 63896
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.777874767780304,
      "learning_rate": 0.0004931614567241727,
      "loss": 2.1795,
      "step": 63897
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.8190913200378418,
      "learning_rate": 0.0004931583268795577,
      "loss": 2.1832,
      "step": 63898
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7490309476852417,
      "learning_rate": 0.0004931551969990309,
      "loss": 2.1464,
      "step": 63899
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.8461402654647827,
      "learning_rate": 0.0004931520670825931,
      "loss": 2.4149,
      "step": 63900
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.8179447650909424,
      "learning_rate": 0.0004931489371302445,
      "loss": 2.2972,
      "step": 63901
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7115375399589539,
      "learning_rate": 0.0004931458071419859,
      "loss": 2.2275,
      "step": 63902
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.8436136841773987,
      "learning_rate": 0.000493142677117818,
      "loss": 2.3417,
      "step": 63903
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7870243787765503,
      "learning_rate": 0.0004931395470577411,
      "loss": 2.1303,
      "step": 63904
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.742033064365387,
      "learning_rate": 0.0004931364169617558,
      "loss": 2.2747,
      "step": 63905
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.782090425491333,
      "learning_rate": 0.000493133286829863,
      "loss": 2.3175,
      "step": 63906
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7086372375488281,
      "learning_rate": 0.000493130156662063,
      "loss": 2.4034,
      "step": 63907
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.0467592477798462,
      "learning_rate": 0.0004931270264583563,
      "loss": 2.3339,
      "step": 63908
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7339884638786316,
      "learning_rate": 0.0004931238962187439,
      "loss": 2.1488,
      "step": 63909
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7241058945655823,
      "learning_rate": 0.000493120765943226,
      "loss": 2.264,
      "step": 63910
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.805942177772522,
      "learning_rate": 0.0004931176356318033,
      "loss": 2.1688,
      "step": 63911
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.0459121465682983,
      "learning_rate": 0.0004931145052844765,
      "loss": 2.3838,
      "step": 63912
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7654440402984619,
      "learning_rate": 0.0004931113749012459,
      "loss": 2.2515,
      "step": 63913
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.8431317806243896,
      "learning_rate": 0.0004931082444821123,
      "loss": 2.1018,
      "step": 63914
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.8801003098487854,
      "learning_rate": 0.0004931051140270763,
      "loss": 2.0097,
      "step": 63915
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7937914133071899,
      "learning_rate": 0.0004931019835361383,
      "loss": 2.0658,
      "step": 63916
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.9821906685829163,
      "learning_rate": 0.000493098853009299,
      "loss": 2.1052,
      "step": 63917
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.0856366157531738,
      "learning_rate": 0.0004930957224465592,
      "loss": 1.9325,
      "step": 63918
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.8060733079910278,
      "learning_rate": 0.0004930925918479191,
      "loss": 2.2801,
      "step": 63919
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.73934406042099,
      "learning_rate": 0.0004930894612133794,
      "loss": 2.3424,
      "step": 63920
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.8729049563407898,
      "learning_rate": 0.0004930863305429408,
      "loss": 2.0408,
      "step": 63921
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.8510132431983948,
      "learning_rate": 0.0004930831998366037,
      "loss": 2.1797,
      "step": 63922
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7108857035636902,
      "learning_rate": 0.0004930800690943689,
      "loss": 2.2948,
      "step": 63923
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.8030478954315186,
      "learning_rate": 0.0004930769383162368,
      "loss": 2.2175,
      "step": 63924
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7338007092475891,
      "learning_rate": 0.0004930738075022081,
      "loss": 1.958,
      "step": 63925
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.9328942894935608,
      "learning_rate": 0.0004930706766522833,
      "loss": 2.0157,
      "step": 63926
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7962899208068848,
      "learning_rate": 0.0004930675457664631,
      "loss": 2.1126,
      "step": 63927
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.914194643497467,
      "learning_rate": 0.0004930644148447479,
      "loss": 2.239,
      "step": 63928
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7339593172073364,
      "learning_rate": 0.0004930612838871384,
      "loss": 2.3129,
      "step": 63929
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7768736481666565,
      "learning_rate": 0.0004930581528936352,
      "loss": 2.0063,
      "step": 63930
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7601368427276611,
      "learning_rate": 0.0004930550218642388,
      "loss": 2.2048,
      "step": 63931
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.8831256628036499,
      "learning_rate": 0.0004930518907989499,
      "loss": 2.2595,
      "step": 63932
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7613606452941895,
      "learning_rate": 0.0004930487596977689,
      "loss": 2.2948,
      "step": 63933
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7546573281288147,
      "learning_rate": 0.0004930456285606966,
      "loss": 2.3529,
      "step": 63934
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.794921875,
      "learning_rate": 0.0004930424973877334,
      "loss": 2.1,
      "step": 63935
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7175829410552979,
      "learning_rate": 0.00049303936617888,
      "loss": 2.447,
      "step": 63936
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7903882265090942,
      "learning_rate": 0.0004930362349341369,
      "loss": 2.457,
      "step": 63937
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.9282110333442688,
      "learning_rate": 0.0004930331036535048,
      "loss": 2.2538,
      "step": 63938
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7924855947494507,
      "learning_rate": 0.0004930299723369841,
      "loss": 2.2787,
      "step": 63939
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.8700920343399048,
      "learning_rate": 0.0004930268409845755,
      "loss": 2.5578,
      "step": 63940
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7636959552764893,
      "learning_rate": 0.0004930237095962796,
      "loss": 2.2396,
      "step": 63941
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.8201128244400024,
      "learning_rate": 0.000493020578172097,
      "loss": 2.3081,
      "step": 63942
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7957512140274048,
      "learning_rate": 0.0004930174467120282,
      "loss": 2.3865,
      "step": 63943
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7593031525611877,
      "learning_rate": 0.0004930143152160737,
      "loss": 2.2873,
      "step": 63944
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.733920693397522,
      "learning_rate": 0.0004930111836842343,
      "loss": 2.2488,
      "step": 63945
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7278293371200562,
      "learning_rate": 0.0004930080521165104,
      "loss": 2.0496,
      "step": 63946
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7357608079910278,
      "learning_rate": 0.0004930049205129028,
      "loss": 2.1678,
      "step": 63947
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.8047199249267578,
      "learning_rate": 0.0004930017888734119,
      "loss": 2.0064,
      "step": 63948
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7463318705558777,
      "learning_rate": 0.0004929986571980383,
      "loss": 2.1572,
      "step": 63949
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7211882472038269,
      "learning_rate": 0.0004929955254867824,
      "loss": 2.1988,
      "step": 63950
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7398983836174011,
      "learning_rate": 0.0004929923937396453,
      "loss": 2.5318,
      "step": 63951
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.6944879293441772,
      "learning_rate": 0.0004929892619566272,
      "loss": 1.9161,
      "step": 63952
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7580966353416443,
      "learning_rate": 0.0004929861301377287,
      "loss": 2.2088,
      "step": 63953
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7698538303375244,
      "learning_rate": 0.0004929829982829504,
      "loss": 2.2141,
      "step": 63954
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.6660405397415161,
      "learning_rate": 0.000492979866392293,
      "loss": 2.0165,
      "step": 63955
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.6855056285858154,
      "learning_rate": 0.0004929767344657569,
      "loss": 2.2799,
      "step": 63956
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.827269434928894,
      "learning_rate": 0.0004929736025033427,
      "loss": 2.2855,
      "step": 63957
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7557321190834045,
      "learning_rate": 0.0004929704705050514,
      "loss": 2.3303,
      "step": 63958
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7117820382118225,
      "learning_rate": 0.000492967338470883,
      "loss": 2.0889,
      "step": 63959
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7423656582832336,
      "learning_rate": 0.0004929642064008383,
      "loss": 2.199,
      "step": 63960
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7480998039245605,
      "learning_rate": 0.0004929610742949181,
      "loss": 2.4013,
      "step": 63961
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.8090824484825134,
      "learning_rate": 0.0004929579421531227,
      "loss": 2.3134,
      "step": 63962
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.8051567673683167,
      "learning_rate": 0.0004929548099754527,
      "loss": 2.2492,
      "step": 63963
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.8480427861213684,
      "learning_rate": 0.0004929516777619088,
      "loss": 2.3141,
      "step": 63964
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.8221399784088135,
      "learning_rate": 0.0004929485455124916,
      "loss": 2.2777,
      "step": 63965
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7417426705360413,
      "learning_rate": 0.0004929454132272015,
      "loss": 2.178,
      "step": 63966
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7866678237915039,
      "learning_rate": 0.0004929422809060394,
      "loss": 2.4228,
      "step": 63967
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.8149293661117554,
      "learning_rate": 0.0004929391485490055,
      "loss": 2.3417,
      "step": 63968
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.706290602684021,
      "learning_rate": 0.0004929360161561007,
      "loss": 2.2088,
      "step": 63969
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7624117136001587,
      "learning_rate": 0.0004929328837273254,
      "loss": 2.5126,
      "step": 63970
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.8295303583145142,
      "learning_rate": 0.0004929297512626802,
      "loss": 2.3704,
      "step": 63971
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.8318392634391785,
      "learning_rate": 0.0004929266187621657,
      "loss": 2.4018,
      "step": 63972
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7289863228797913,
      "learning_rate": 0.0004929234862257827,
      "loss": 2.2667,
      "step": 63973
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.770749032497406,
      "learning_rate": 0.0004929203536535315,
      "loss": 2.1879,
      "step": 63974
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7555138468742371,
      "learning_rate": 0.0004929172210454126,
      "loss": 2.3323,
      "step": 63975
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7112290859222412,
      "learning_rate": 0.0004929140884014268,
      "loss": 2.1084,
      "step": 63976
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7628563046455383,
      "learning_rate": 0.0004929109557215747,
      "loss": 2.067,
      "step": 63977
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.8137826919555664,
      "learning_rate": 0.0004929078230058568,
      "loss": 2.2032,
      "step": 63978
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.013769268989563,
      "learning_rate": 0.0004929046902542736,
      "loss": 2.3086,
      "step": 63979
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7439821362495422,
      "learning_rate": 0.000492901557466826,
      "loss": 2.2709,
      "step": 63980
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.8423646092414856,
      "learning_rate": 0.0004928984246435141,
      "loss": 2.259,
      "step": 63981
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.9638954997062683,
      "learning_rate": 0.0004928952917843389,
      "loss": 2.2934,
      "step": 63982
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.8998372554779053,
      "learning_rate": 0.0004928921588893008,
      "loss": 2.2202,
      "step": 63983
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.9747347831726074,
      "learning_rate": 0.0004928890259584004,
      "loss": 2.2871,
      "step": 63984
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.0491862297058105,
      "learning_rate": 0.0004928858929916383,
      "loss": 2.1895,
      "step": 63985
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7445324659347534,
      "learning_rate": 0.000492882759989015,
      "loss": 2.1338,
      "step": 63986
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7711876630783081,
      "learning_rate": 0.0004928796269505312,
      "loss": 2.36,
      "step": 63987
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.9488563537597656,
      "learning_rate": 0.0004928764938761875,
      "loss": 2.1475,
      "step": 63988
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.8320630788803101,
      "learning_rate": 0.0004928733607659844,
      "loss": 2.2817,
      "step": 63989
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7569864392280579,
      "learning_rate": 0.0004928702276199223,
      "loss": 2.4093,
      "step": 63990
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7802261114120483,
      "learning_rate": 0.0004928670944380023,
      "loss": 2.2393,
      "step": 63991
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7069077491760254,
      "learning_rate": 0.0004928639612202245,
      "loss": 2.1863,
      "step": 63992
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.8604497313499451,
      "learning_rate": 0.0004928608279665896,
      "loss": 2.3754,
      "step": 63993
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.8120136260986328,
      "learning_rate": 0.0004928576946770984,
      "loss": 2.2621,
      "step": 63994
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7163331508636475,
      "learning_rate": 0.000492854561351751,
      "loss": 2.2166,
      "step": 63995
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.887575089931488,
      "learning_rate": 0.0004928514279905486,
      "loss": 2.5065,
      "step": 63996
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7331997752189636,
      "learning_rate": 0.0004928482945934915,
      "loss": 2.393,
      "step": 63997
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.9193043112754822,
      "learning_rate": 0.00049284516116058,
      "loss": 2.0799,
      "step": 63998
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.8102099895477295,
      "learning_rate": 0.0004928420276918152,
      "loss": 2.3284,
      "step": 63999
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7802667021751404,
      "learning_rate": 0.0004928388941871973,
      "loss": 2.4573,
      "step": 64000
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7931579351425171,
      "learning_rate": 0.000492835760646727,
      "loss": 2.0593,
      "step": 64001
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7244865298271179,
      "learning_rate": 0.000492832627070405,
      "loss": 2.2173,
      "step": 64002
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.8691722750663757,
      "learning_rate": 0.0004928294934582316,
      "loss": 2.1731,
      "step": 64003
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7772843241691589,
      "learning_rate": 0.0004928263598102077,
      "loss": 2.3081,
      "step": 64004
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7161458730697632,
      "learning_rate": 0.0004928232261263337,
      "loss": 2.2044,
      "step": 64005
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7407235503196716,
      "learning_rate": 0.0004928200924066102,
      "loss": 2.266,
      "step": 64006
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.8637632131576538,
      "learning_rate": 0.0004928169586510378,
      "loss": 2.3021,
      "step": 64007
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.0384589433670044,
      "learning_rate": 0.0004928138248596171,
      "loss": 1.969,
      "step": 64008
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.8223977088928223,
      "learning_rate": 0.0004928106910323486,
      "loss": 2.423,
      "step": 64009
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.8982580900192261,
      "learning_rate": 0.0004928075571692331,
      "loss": 2.1315,
      "step": 64010
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7475641965866089,
      "learning_rate": 0.0004928044232702709,
      "loss": 2.1843,
      "step": 64011
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.8431978225708008,
      "learning_rate": 0.0004928012893354629,
      "loss": 2.3282,
      "step": 64012
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7394151091575623,
      "learning_rate": 0.0004927981553648094,
      "loss": 2.0635,
      "step": 64013
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.8102729320526123,
      "learning_rate": 0.000492795021358311,
      "loss": 2.0735,
      "step": 64014
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.8815808892250061,
      "learning_rate": 0.0004927918873159684,
      "loss": 2.4521,
      "step": 64015
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7191551327705383,
      "learning_rate": 0.0004927887532377822,
      "loss": 2.2601,
      "step": 64016
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7443042397499084,
      "learning_rate": 0.0004927856191237528,
      "loss": 2.262,
      "step": 64017
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.9555925130844116,
      "learning_rate": 0.0004927824849738811,
      "loss": 2.1468,
      "step": 64018
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7262600660324097,
      "learning_rate": 0.0004927793507881674,
      "loss": 2.1554,
      "step": 64019
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7593900561332703,
      "learning_rate": 0.0004927762165666124,
      "loss": 2.1565,
      "step": 64020
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.926943302154541,
      "learning_rate": 0.0004927730823092168,
      "loss": 2.2676,
      "step": 64021
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7191219329833984,
      "learning_rate": 0.0004927699480159808,
      "loss": 2.0278,
      "step": 64022
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.8671196699142456,
      "learning_rate": 0.0004927668136869055,
      "loss": 2.2177,
      "step": 64023
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7628868222236633,
      "learning_rate": 0.0004927636793219909,
      "loss": 2.1254,
      "step": 64024
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7410822510719299,
      "learning_rate": 0.0004927605449212381,
      "loss": 2.1873,
      "step": 64025
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7423151135444641,
      "learning_rate": 0.0004927574104846474,
      "loss": 2.1806,
      "step": 64026
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7170273661613464,
      "learning_rate": 0.0004927542760122195,
      "loss": 2.0298,
      "step": 64027
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7627831697463989,
      "learning_rate": 0.000492751141503955,
      "loss": 2.3802,
      "step": 64028
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.800582766532898,
      "learning_rate": 0.0004927480069598542,
      "loss": 2.1544,
      "step": 64029
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7827809453010559,
      "learning_rate": 0.0004927448723799181,
      "loss": 2.2981,
      "step": 64030
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.8765440583229065,
      "learning_rate": 0.0004927417377641471,
      "loss": 2.1468,
      "step": 64031
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7779393196105957,
      "learning_rate": 0.0004927386031125416,
      "loss": 2.0221,
      "step": 64032
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.6994529962539673,
      "learning_rate": 0.0004927354684251024,
      "loss": 2.2022,
      "step": 64033
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7151305675506592,
      "learning_rate": 0.0004927323337018302,
      "loss": 2.2524,
      "step": 64034
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7662041187286377,
      "learning_rate": 0.0004927291989427254,
      "loss": 2.0421,
      "step": 64035
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7487261891365051,
      "learning_rate": 0.0004927260641477884,
      "loss": 2.2578,
      "step": 64036
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7878528833389282,
      "learning_rate": 0.0004927229293170202,
      "loss": 2.4693,
      "step": 64037
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7477196455001831,
      "learning_rate": 0.0004927197944504211,
      "loss": 2.1284,
      "step": 64038
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.8238905668258667,
      "learning_rate": 0.0004927166595479917,
      "loss": 2.2339,
      "step": 64039
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.6501357555389404,
      "learning_rate": 0.0004927135246097327,
      "loss": 2.136,
      "step": 64040
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7092369794845581,
      "learning_rate": 0.0004927103896356446,
      "loss": 2.1427,
      "step": 64041
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.8496602773666382,
      "learning_rate": 0.0004927072546257279,
      "loss": 2.4902,
      "step": 64042
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.796215295791626,
      "learning_rate": 0.0004927041195799835,
      "loss": 2.0513,
      "step": 64043
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7735930681228638,
      "learning_rate": 0.0004927009844984116,
      "loss": 2.4008,
      "step": 64044
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7856945395469666,
      "learning_rate": 0.0004926978493810131,
      "loss": 2.303,
      "step": 64045
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.9316926002502441,
      "learning_rate": 0.0004926947142277882,
      "loss": 2.3327,
      "step": 64046
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7723339796066284,
      "learning_rate": 0.0004926915790387379,
      "loss": 2.497,
      "step": 64047
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.8499628901481628,
      "learning_rate": 0.0004926884438138625,
      "loss": 2.0515,
      "step": 64048
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.8227878212928772,
      "learning_rate": 0.0004926853085531628,
      "loss": 2.2069,
      "step": 64049
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.8159475326538086,
      "learning_rate": 0.0004926821732566391,
      "loss": 2.3558,
      "step": 64050
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.9658529758453369,
      "learning_rate": 0.0004926790379242923,
      "loss": 2.0702,
      "step": 64051
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.8456230759620667,
      "learning_rate": 0.0004926759025561228,
      "loss": 2.2163,
      "step": 64052
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.8400245904922485,
      "learning_rate": 0.0004926727671521312,
      "loss": 1.9161,
      "step": 64053
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7180588841438293,
      "learning_rate": 0.0004926696317123179,
      "loss": 2.1983,
      "step": 64054
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7627869248390198,
      "learning_rate": 0.000492666496236684,
      "loss": 2.1076,
      "step": 64055
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.9357471466064453,
      "learning_rate": 0.0004926633607252295,
      "loss": 2.0534,
      "step": 64056
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7224283218383789,
      "learning_rate": 0.0004926602251779554,
      "loss": 2.1768,
      "step": 64057
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.8413326144218445,
      "learning_rate": 0.0004926570895948621,
      "loss": 1.9902,
      "step": 64058
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.8251721262931824,
      "learning_rate": 0.0004926539539759502,
      "loss": 2.3696,
      "step": 64059
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7551277875900269,
      "learning_rate": 0.0004926508183212202,
      "loss": 2.2599,
      "step": 64060
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7023018002510071,
      "learning_rate": 0.0004926476826306729,
      "loss": 2.1488,
      "step": 64061
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7694528698921204,
      "learning_rate": 0.0004926445469043087,
      "loss": 2.2493,
      "step": 64062
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.9694600701332092,
      "learning_rate": 0.0004926414111421283,
      "loss": 2.2772,
      "step": 64063
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7688608169555664,
      "learning_rate": 0.0004926382753441323,
      "loss": 2.2174,
      "step": 64064
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.8443186283111572,
      "learning_rate": 0.000492635139510321,
      "loss": 2.1071,
      "step": 64065
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.9082740545272827,
      "learning_rate": 0.0004926320036406953,
      "loss": 2.2205,
      "step": 64066
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7836640477180481,
      "learning_rate": 0.0004926288677352557,
      "loss": 2.2002,
      "step": 64067
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7507496476173401,
      "learning_rate": 0.0004926257317940027,
      "loss": 2.1152,
      "step": 64068
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7178013920783997,
      "learning_rate": 0.000492622595816937,
      "loss": 2.2793,
      "step": 64069
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7575529217720032,
      "learning_rate": 0.000492619459804059,
      "loss": 2.2783,
      "step": 64070
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7687427401542664,
      "learning_rate": 0.0004926163237553695,
      "loss": 2.2749,
      "step": 64071
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7455455660820007,
      "learning_rate": 0.000492613187670869,
      "loss": 2.131,
      "step": 64072
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7458333373069763,
      "learning_rate": 0.0004926100515505581,
      "loss": 2.4169,
      "step": 64073
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7249710559844971,
      "learning_rate": 0.0004926069153944373,
      "loss": 2.1414,
      "step": 64074
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7851453423500061,
      "learning_rate": 0.0004926037792025071,
      "loss": 2.4899,
      "step": 64075
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7944808006286621,
      "learning_rate": 0.0004926006429747684,
      "loss": 2.3478,
      "step": 64076
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.8318772912025452,
      "learning_rate": 0.0004925975067112216,
      "loss": 2.3021,
      "step": 64077
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7523617744445801,
      "learning_rate": 0.0004925943704118673,
      "loss": 2.3221,
      "step": 64078
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.9870625138282776,
      "learning_rate": 0.0004925912340767061,
      "loss": 2.3788,
      "step": 64079
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7283472418785095,
      "learning_rate": 0.0004925880977057384,
      "loss": 2.2654,
      "step": 64080
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.77984219789505,
      "learning_rate": 0.000492584961298965,
      "loss": 2.3645,
      "step": 64081
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.8755427002906799,
      "learning_rate": 0.0004925818248563865,
      "loss": 2.4728,
      "step": 64082
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7465599775314331,
      "learning_rate": 0.0004925786883780033,
      "loss": 2.2123,
      "step": 64083
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7461267113685608,
      "learning_rate": 0.0004925755518638161,
      "loss": 2.1533,
      "step": 64084
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7551336884498596,
      "learning_rate": 0.0004925724153138256,
      "loss": 2.3808,
      "step": 64085
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7870410680770874,
      "learning_rate": 0.0004925692787280321,
      "loss": 2.011,
      "step": 64086
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7188824415206909,
      "learning_rate": 0.0004925661421064364,
      "loss": 2.2931,
      "step": 64087
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7157349586486816,
      "learning_rate": 0.000492563005449039,
      "loss": 2.3095,
      "step": 64088
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7505990862846375,
      "learning_rate": 0.0004925598687558405,
      "loss": 2.1725,
      "step": 64089
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.8434901833534241,
      "learning_rate": 0.0004925567320268415,
      "loss": 2.3218,
      "step": 64090
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.8043883442878723,
      "learning_rate": 0.0004925535952620426,
      "loss": 2.113,
      "step": 64091
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7849628925323486,
      "learning_rate": 0.0004925504584614444,
      "loss": 2.362,
      "step": 64092
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7622702717781067,
      "learning_rate": 0.0004925473216250472,
      "loss": 2.415,
      "step": 64093
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7162644863128662,
      "learning_rate": 0.000492544184752852,
      "loss": 2.359,
      "step": 64094
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7042678594589233,
      "learning_rate": 0.0004925410478448592,
      "loss": 2.3762,
      "step": 64095
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.9355239272117615,
      "learning_rate": 0.0004925379109010694,
      "loss": 2.3168,
      "step": 64096
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7317101955413818,
      "learning_rate": 0.0004925347739214831,
      "loss": 2.2732,
      "step": 64097
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7354322671890259,
      "learning_rate": 0.000492531636906101,
      "loss": 2.2635,
      "step": 64098
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7424109578132629,
      "learning_rate": 0.0004925284998549236,
      "loss": 2.2561,
      "step": 64099
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7700704336166382,
      "learning_rate": 0.0004925253627679515,
      "loss": 2.0641,
      "step": 64100
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.697517991065979,
      "learning_rate": 0.0004925222256451854,
      "loss": 2.1433,
      "step": 64101
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7142366170883179,
      "learning_rate": 0.0004925190884866257,
      "loss": 2.0035,
      "step": 64102
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.8423822522163391,
      "learning_rate": 0.000492515951292273,
      "loss": 2.2437,
      "step": 64103
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7854540348052979,
      "learning_rate": 0.0004925128140621281,
      "loss": 2.2994,
      "step": 64104
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7478461265563965,
      "learning_rate": 0.0004925096767961914,
      "loss": 2.1429,
      "step": 64105
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.8903225064277649,
      "learning_rate": 0.0004925065394944635,
      "loss": 2.0737,
      "step": 64106
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.974134624004364,
      "learning_rate": 0.0004925034021569449,
      "loss": 2.2505,
      "step": 64107
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.8646144270896912,
      "learning_rate": 0.0004925002647836364,
      "loss": 2.3096,
      "step": 64108
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7526114583015442,
      "learning_rate": 0.0004924971273745385,
      "loss": 2.0978,
      "step": 64109
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.8784375190734863,
      "learning_rate": 0.0004924939899296516,
      "loss": 2.3753,
      "step": 64110
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7112976312637329,
      "learning_rate": 0.0004924908524489766,
      "loss": 2.3123,
      "step": 64111
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.8652694225311279,
      "learning_rate": 0.0004924877149325138,
      "loss": 2.2847,
      "step": 64112
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7921314239501953,
      "learning_rate": 0.000492484577380264,
      "loss": 2.0636,
      "step": 64113
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.8562432527542114,
      "learning_rate": 0.0004924814397922275,
      "loss": 2.1599,
      "step": 64114
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7464696764945984,
      "learning_rate": 0.0004924783021684053,
      "loss": 2.2932,
      "step": 64115
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7398325204849243,
      "learning_rate": 0.0004924751645087976,
      "loss": 2.043,
      "step": 64116
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7947884202003479,
      "learning_rate": 0.0004924720268134051,
      "loss": 2.4281,
      "step": 64117
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.8836368918418884,
      "learning_rate": 0.0004924688890822285,
      "loss": 2.3274,
      "step": 64118
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.8037503361701965,
      "learning_rate": 0.0004924657513152683,
      "loss": 2.1493,
      "step": 64119
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7458690404891968,
      "learning_rate": 0.0004924626135125251,
      "loss": 1.9999,
      "step": 64120
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7461347579956055,
      "learning_rate": 0.0004924594756739994,
      "loss": 2.0829,
      "step": 64121
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.8913269639015198,
      "learning_rate": 0.000492456337799692,
      "loss": 2.2378,
      "step": 64122
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.0923030376434326,
      "learning_rate": 0.0004924531998896031,
      "loss": 2.2716,
      "step": 64123
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7835288643836975,
      "learning_rate": 0.0004924500619437337,
      "loss": 2.0851,
      "step": 64124
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.0110902786254883,
      "learning_rate": 0.0004924469239620842,
      "loss": 2.3933,
      "step": 64125
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.8925023078918457,
      "learning_rate": 0.000492443785944655,
      "loss": 2.3157,
      "step": 64126
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7772309184074402,
      "learning_rate": 0.0004924406478914471,
      "loss": 2.0637,
      "step": 64127
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7619701027870178,
      "learning_rate": 0.0004924375098024607,
      "loss": 2.2471,
      "step": 64128
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.785830020904541,
      "learning_rate": 0.0004924343716776967,
      "loss": 2.113,
      "step": 64129
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.873399555683136,
      "learning_rate": 0.0004924312335171553,
      "loss": 2.2712,
      "step": 64130
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7940680384635925,
      "learning_rate": 0.0004924280953208375,
      "loss": 2.3833,
      "step": 64131
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7109310626983643,
      "learning_rate": 0.0004924249570887435,
      "loss": 1.9906,
      "step": 64132
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7673417329788208,
      "learning_rate": 0.0004924218188208741,
      "loss": 2.1116,
      "step": 64133
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7416755557060242,
      "learning_rate": 0.0004924186805172299,
      "loss": 2.1657,
      "step": 64134
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.8030861616134644,
      "learning_rate": 0.0004924155421778115,
      "loss": 2.2832,
      "step": 64135
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7487660646438599,
      "learning_rate": 0.0004924124038026194,
      "loss": 2.2625,
      "step": 64136
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.898266077041626,
      "learning_rate": 0.0004924092653916541,
      "loss": 2.3832,
      "step": 64137
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7930773496627808,
      "learning_rate": 0.0004924061269449164,
      "loss": 2.265,
      "step": 64138
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.8005812764167786,
      "learning_rate": 0.0004924029884624067,
      "loss": 2.1455,
      "step": 64139
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.8594101071357727,
      "learning_rate": 0.0004923998499441257,
      "loss": 2.4122,
      "step": 64140
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7235948443412781,
      "learning_rate": 0.0004923967113900739,
      "loss": 2.1535,
      "step": 64141
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7397327423095703,
      "learning_rate": 0.0004923935728002518,
      "loss": 2.2849,
      "step": 64142
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7846605777740479,
      "learning_rate": 0.0004923904341746604,
      "loss": 2.267,
      "step": 64143
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7549317479133606,
      "learning_rate": 0.0004923872955132996,
      "loss": 2.0706,
      "step": 64144
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7207232713699341,
      "learning_rate": 0.0004923841568161707,
      "loss": 2.3259,
      "step": 64145
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7288864254951477,
      "learning_rate": 0.0004923810180832739,
      "loss": 2.3738,
      "step": 64146
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.8139934539794922,
      "learning_rate": 0.0004923778793146097,
      "loss": 2.1674,
      "step": 64147
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.6783424019813538,
      "learning_rate": 0.0004923747405101789,
      "loss": 2.1832,
      "step": 64148
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7514712810516357,
      "learning_rate": 0.000492371601669982,
      "loss": 2.1915,
      "step": 64149
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7613034248352051,
      "learning_rate": 0.0004923684627940197,
      "loss": 2.5,
      "step": 64150
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.730836808681488,
      "learning_rate": 0.0004923653238822922,
      "loss": 2.177,
      "step": 64151
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.9115957021713257,
      "learning_rate": 0.0004923621849348005,
      "loss": 2.2236,
      "step": 64152
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7586566805839539,
      "learning_rate": 0.0004923590459515451,
      "loss": 2.0196,
      "step": 64153
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.9991268515586853,
      "learning_rate": 0.0004923559069325265,
      "loss": 2.2947,
      "step": 64154
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7689141631126404,
      "learning_rate": 0.0004923527678777453,
      "loss": 2.0851,
      "step": 64155
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7959607243537903,
      "learning_rate": 0.0004923496287872021,
      "loss": 2.0879,
      "step": 64156
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.789792537689209,
      "learning_rate": 0.0004923464896608974,
      "loss": 2.1446,
      "step": 64157
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.873389720916748,
      "learning_rate": 0.0004923433504988319,
      "loss": 2.3505,
      "step": 64158
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7155224680900574,
      "learning_rate": 0.0004923402113010061,
      "loss": 2.2987,
      "step": 64159
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7096866965293884,
      "learning_rate": 0.0004923370720674206,
      "loss": 2.2968,
      "step": 64160
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7759868502616882,
      "learning_rate": 0.0004923339327980762,
      "loss": 2.2987,
      "step": 64161
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.8133051991462708,
      "learning_rate": 0.0004923307934929731,
      "loss": 2.3113,
      "step": 64162
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7300593852996826,
      "learning_rate": 0.0004923276541521121,
      "loss": 2.194,
      "step": 64163
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7718963623046875,
      "learning_rate": 0.0004923245147754939,
      "loss": 2.1956,
      "step": 64164
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.8775724172592163,
      "learning_rate": 0.0004923213753631188,
      "loss": 2.3067,
      "step": 64165
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.73819899559021,
      "learning_rate": 0.0004923182359149876,
      "loss": 2.1071,
      "step": 64166
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7152120471000671,
      "learning_rate": 0.0004923150964311007,
      "loss": 2.1353,
      "step": 64167
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.818693995475769,
      "learning_rate": 0.0004923119569114589,
      "loss": 1.955,
      "step": 64168
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7361446619033813,
      "learning_rate": 0.0004923088173560626,
      "loss": 2.202,
      "step": 64169
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7192591428756714,
      "learning_rate": 0.0004923056777649125,
      "loss": 2.2048,
      "step": 64170
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.8426610231399536,
      "learning_rate": 0.0004923025381380092,
      "loss": 2.2756,
      "step": 64171
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7706782221794128,
      "learning_rate": 0.0004922993984753532,
      "loss": 2.1893,
      "step": 64172
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7308065891265869,
      "learning_rate": 0.000492296258776945,
      "loss": 2.1786,
      "step": 64173
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.8485689759254456,
      "learning_rate": 0.0004922931190427855,
      "loss": 2.2666,
      "step": 64174
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.8200873136520386,
      "learning_rate": 0.000492289979272875,
      "loss": 2.2186,
      "step": 64175
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.8058199882507324,
      "learning_rate": 0.000492286839467214,
      "loss": 2.2223,
      "step": 64176
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7690165638923645,
      "learning_rate": 0.0004922836996258034,
      "loss": 2.1919,
      "step": 64177
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.8058084845542908,
      "learning_rate": 0.0004922805597486437,
      "loss": 2.306,
      "step": 64178
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.8368927240371704,
      "learning_rate": 0.0004922774198357352,
      "loss": 2.2705,
      "step": 64179
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.8785995244979858,
      "learning_rate": 0.0004922742798870789,
      "loss": 2.1237,
      "step": 64180
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7829355597496033,
      "learning_rate": 0.0004922711399026751,
      "loss": 2.3517,
      "step": 64181
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.850096583366394,
      "learning_rate": 0.0004922679998825244,
      "loss": 2.2878,
      "step": 64182
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.802371621131897,
      "learning_rate": 0.0004922648598266276,
      "loss": 2.019,
      "step": 64183
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.6929516792297363,
      "learning_rate": 0.0004922617197349851,
      "loss": 2.1383,
      "step": 64184
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.8765885829925537,
      "learning_rate": 0.0004922585796075974,
      "loss": 2.1841,
      "step": 64185
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.8211326599121094,
      "learning_rate": 0.0004922554394444653,
      "loss": 2.3041,
      "step": 64186
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7294678092002869,
      "learning_rate": 0.0004922522992455891,
      "loss": 2.306,
      "step": 64187
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7540248036384583,
      "learning_rate": 0.0004922491590109698,
      "loss": 2.1337,
      "step": 64188
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7628793120384216,
      "learning_rate": 0.0004922460187406077,
      "loss": 2.3744,
      "step": 64189
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.8048520088195801,
      "learning_rate": 0.0004922428784345034,
      "loss": 2.0555,
      "step": 64190
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.807305097579956,
      "learning_rate": 0.0004922397380926575,
      "loss": 2.1123,
      "step": 64191
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7376980185508728,
      "learning_rate": 0.0004922365977150708,
      "loss": 2.2961,
      "step": 64192
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.8104622960090637,
      "learning_rate": 0.0004922334573017434,
      "loss": 2.1823,
      "step": 64193
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.8089362978935242,
      "learning_rate": 0.0004922303168526762,
      "loss": 2.1657,
      "step": 64194
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7749837636947632,
      "learning_rate": 0.0004922271763678699,
      "loss": 2.2443,
      "step": 64195
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7631412744522095,
      "learning_rate": 0.0004922240358473249,
      "loss": 2.261,
      "step": 64196
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.774287223815918,
      "learning_rate": 0.0004922208952910419,
      "loss": 2.1633,
      "step": 64197
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7111934423446655,
      "learning_rate": 0.0004922177546990212,
      "loss": 2.2752,
      "step": 64198
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.8058509826660156,
      "learning_rate": 0.0004922146140712637,
      "loss": 2.1872,
      "step": 64199
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7176871299743652,
      "learning_rate": 0.00049221147340777,
      "loss": 2.1679,
      "step": 64200
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.6706728339195251,
      "learning_rate": 0.0004922083327085404,
      "loss": 2.1706,
      "step": 64201
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.106610894203186,
      "learning_rate": 0.0004922051919735756,
      "loss": 2.4651,
      "step": 64202
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7022353410720825,
      "learning_rate": 0.0004922020512028763,
      "loss": 2.1285,
      "step": 64203
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7017303109169006,
      "learning_rate": 0.0004921989103964431,
      "loss": 2.1304,
      "step": 64204
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7190198302268982,
      "learning_rate": 0.0004921957695542764,
      "loss": 2.1874,
      "step": 64205
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.6983214020729065,
      "learning_rate": 0.0004921926286763769,
      "loss": 2.0252,
      "step": 64206
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.869301974773407,
      "learning_rate": 0.0004921894877627452,
      "loss": 2.3279,
      "step": 64207
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7810932993888855,
      "learning_rate": 0.0004921863468133818,
      "loss": 2.1439,
      "step": 64208
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7170438766479492,
      "learning_rate": 0.0004921832058282872,
      "loss": 2.0757,
      "step": 64209
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7906355261802673,
      "learning_rate": 0.0004921800648074623,
      "loss": 2.4922,
      "step": 64210
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.8577958345413208,
      "learning_rate": 0.0004921769237509075,
      "loss": 2.2947,
      "step": 64211
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.8389056324958801,
      "learning_rate": 0.0004921737826586233,
      "loss": 2.297,
      "step": 64212
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.9882349371910095,
      "learning_rate": 0.0004921706415306104,
      "loss": 2.2724,
      "step": 64213
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.876003623008728,
      "learning_rate": 0.0004921675003668693,
      "loss": 2.058,
      "step": 64214
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7398799061775208,
      "learning_rate": 0.0004921643591674008,
      "loss": 2.1055,
      "step": 64215
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7420791387557983,
      "learning_rate": 0.0004921612179322052,
      "loss": 2.2905,
      "step": 64216
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.8415470719337463,
      "learning_rate": 0.0004921580766612831,
      "loss": 2.3782,
      "step": 64217
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.8623197078704834,
      "learning_rate": 0.0004921549353546353,
      "loss": 2.0692,
      "step": 64218
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7247308492660522,
      "learning_rate": 0.0004921517940122622,
      "loss": 2.145,
      "step": 64219
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7553339600563049,
      "learning_rate": 0.0004921486526341646,
      "loss": 2.2334,
      "step": 64220
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.9018061757087708,
      "learning_rate": 0.0004921455112203429,
      "loss": 2.3903,
      "step": 64221
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.8483530282974243,
      "learning_rate": 0.0004921423697707976,
      "loss": 2.2162,
      "step": 64222
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7136201858520508,
      "learning_rate": 0.0004921392282855295,
      "loss": 2.142,
      "step": 64223
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.9217559695243835,
      "learning_rate": 0.000492136086764539,
      "loss": 2.3027,
      "step": 64224
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.9358050227165222,
      "learning_rate": 0.0004921329452078269,
      "loss": 2.2077,
      "step": 64225
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.8463321924209595,
      "learning_rate": 0.0004921298036153936,
      "loss": 2.3135,
      "step": 64226
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.8722249865531921,
      "learning_rate": 0.0004921266619872397,
      "loss": 2.2292,
      "step": 64227
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.892285943031311,
      "learning_rate": 0.0004921235203233659,
      "loss": 2.2841,
      "step": 64228
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.9427103996276855,
      "learning_rate": 0.0004921203786237727,
      "loss": 2.1617,
      "step": 64229
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7478315830230713,
      "learning_rate": 0.0004921172368884607,
      "loss": 2.3651,
      "step": 64230
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.2233463525772095,
      "learning_rate": 0.0004921140951174304,
      "loss": 2.3162,
      "step": 64231
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.9011171460151672,
      "learning_rate": 0.0004921109533106826,
      "loss": 2.0596,
      "step": 64232
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7617959976196289,
      "learning_rate": 0.0004921078114682176,
      "loss": 2.26,
      "step": 64233
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7698639631271362,
      "learning_rate": 0.0004921046695900362,
      "loss": 2.292,
      "step": 64234
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.8661626577377319,
      "learning_rate": 0.0004921015276761389,
      "loss": 2.2966,
      "step": 64235
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.9627324342727661,
      "learning_rate": 0.0004920983857265264,
      "loss": 2.4015,
      "step": 64236
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7729851007461548,
      "learning_rate": 0.000492095243741199,
      "loss": 2.4868,
      "step": 64237
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.011289119720459,
      "learning_rate": 0.0004920921017201577,
      "loss": 2.1658,
      "step": 64238
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.8996719121932983,
      "learning_rate": 0.0004920889596634027,
      "loss": 2.0693,
      "step": 64239
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.8662487864494324,
      "learning_rate": 0.0004920858175709347,
      "loss": 2.1884,
      "step": 64240
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7680901288986206,
      "learning_rate": 0.0004920826754427544,
      "loss": 2.1954,
      "step": 64241
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.8170753717422485,
      "learning_rate": 0.0004920795332788622,
      "loss": 2.1684,
      "step": 64242
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.9680840969085693,
      "learning_rate": 0.000492076391079259,
      "loss": 2.1081,
      "step": 64243
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7106532454490662,
      "learning_rate": 0.000492073248843945,
      "loss": 2.3643,
      "step": 64244
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.8624396324157715,
      "learning_rate": 0.0004920701065729209,
      "loss": 2.2766,
      "step": 64245
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7350865602493286,
      "learning_rate": 0.0004920669642661875,
      "loss": 2.3118,
      "step": 64246
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7295675873756409,
      "learning_rate": 0.0004920638219237453,
      "loss": 2.2016,
      "step": 64247
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7712650299072266,
      "learning_rate": 0.0004920606795455946,
      "loss": 2.0968,
      "step": 64248
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7644430994987488,
      "learning_rate": 0.0004920575371317363,
      "loss": 2.1836,
      "step": 64249
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7670650482177734,
      "learning_rate": 0.0004920543946821708,
      "loss": 2.1966,
      "step": 64250
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7109470367431641,
      "learning_rate": 0.0004920512521968987,
      "loss": 2.114,
      "step": 64251
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.77046138048172,
      "learning_rate": 0.0004920481096759208,
      "loss": 2.284,
      "step": 64252
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7715879082679749,
      "learning_rate": 0.0004920449671192375,
      "loss": 2.3673,
      "step": 64253
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7722247242927551,
      "learning_rate": 0.0004920418245268495,
      "loss": 2.2772,
      "step": 64254
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.8152747750282288,
      "learning_rate": 0.0004920386818987572,
      "loss": 2.132,
      "step": 64255
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7414395809173584,
      "learning_rate": 0.0004920355392349612,
      "loss": 2.2382,
      "step": 64256
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.818849503993988,
      "learning_rate": 0.0004920323965354622,
      "loss": 2.4451,
      "step": 64257
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.758919358253479,
      "learning_rate": 0.0004920292538002608,
      "loss": 2.2424,
      "step": 64258
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7291790843009949,
      "learning_rate": 0.0004920261110293577,
      "loss": 2.1579,
      "step": 64259
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7183610200881958,
      "learning_rate": 0.0004920229682227531,
      "loss": 2.4973,
      "step": 64260
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7581678628921509,
      "learning_rate": 0.0004920198253804478,
      "loss": 2.041,
      "step": 64261
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.6843672394752502,
      "learning_rate": 0.0004920166825024425,
      "loss": 2.355,
      "step": 64262
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.8351753950119019,
      "learning_rate": 0.0004920135395887378,
      "loss": 2.3767,
      "step": 64263
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.774432897567749,
      "learning_rate": 0.0004920103966393339,
      "loss": 2.1156,
      "step": 64264
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7388282418251038,
      "learning_rate": 0.0004920072536542318,
      "loss": 2.0734,
      "step": 64265
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7450414896011353,
      "learning_rate": 0.0004920041106334318,
      "loss": 2.3147,
      "step": 64266
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.6667225956916809,
      "learning_rate": 0.0004920009675769347,
      "loss": 2.2975,
      "step": 64267
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.8496227860450745,
      "learning_rate": 0.000491997824484741,
      "loss": 2.0863,
      "step": 64268
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.6984001994132996,
      "learning_rate": 0.0004919946813568512,
      "loss": 2.1686,
      "step": 64269
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7009586691856384,
      "learning_rate": 0.000491991538193266,
      "loss": 2.1908,
      "step": 64270
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7775569558143616,
      "learning_rate": 0.000491988394993986,
      "loss": 2.1895,
      "step": 64271
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7196261286735535,
      "learning_rate": 0.0004919852517590117,
      "loss": 2.2065,
      "step": 64272
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.6586230993270874,
      "learning_rate": 0.0004919821084883437,
      "loss": 2.0693,
      "step": 64273
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7522403597831726,
      "learning_rate": 0.0004919789651819827,
      "loss": 2.0489,
      "step": 64274
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7324560284614563,
      "learning_rate": 0.0004919758218399291,
      "loss": 2.2022,
      "step": 64275
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7400434613227844,
      "learning_rate": 0.0004919726784621836,
      "loss": 1.9808,
      "step": 64276
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7608600854873657,
      "learning_rate": 0.0004919695350487467,
      "loss": 2.2922,
      "step": 64277
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7264289259910583,
      "learning_rate": 0.000491966391599619,
      "loss": 2.2554,
      "step": 64278
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7614948749542236,
      "learning_rate": 0.0004919632481148013,
      "loss": 2.4816,
      "step": 64279
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7011774182319641,
      "learning_rate": 0.0004919601045942939,
      "loss": 2.0009,
      "step": 64280
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7618158459663391,
      "learning_rate": 0.0004919569610380975,
      "loss": 2.222,
      "step": 64281
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7855576872825623,
      "learning_rate": 0.0004919538174462127,
      "loss": 2.1997,
      "step": 64282
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.794909656047821,
      "learning_rate": 0.00049195067381864,
      "loss": 2.2995,
      "step": 64283
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.799253523349762,
      "learning_rate": 0.0004919475301553801,
      "loss": 2.0491,
      "step": 64284
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.0586411952972412,
      "learning_rate": 0.0004919443864564335,
      "loss": 2.1895,
      "step": 64285
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.748923122882843,
      "learning_rate": 0.0004919412427218007,
      "loss": 2.4831,
      "step": 64286
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7913820147514343,
      "learning_rate": 0.0004919380989514826,
      "loss": 2.1512,
      "step": 64287
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.8496671319007874,
      "learning_rate": 0.0004919349551454795,
      "loss": 2.1074,
      "step": 64288
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.788722038269043,
      "learning_rate": 0.0004919318113037922,
      "loss": 2.2858,
      "step": 64289
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.715177059173584,
      "learning_rate": 0.000491928667426421,
      "loss": 1.9558,
      "step": 64290
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.8997483849525452,
      "learning_rate": 0.0004919255235133667,
      "loss": 2.2521,
      "step": 64291
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.8830012083053589,
      "learning_rate": 0.0004919223795646297,
      "loss": 2.3172,
      "step": 64292
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7432876229286194,
      "learning_rate": 0.0004919192355802108,
      "loss": 2.3653,
      "step": 64293
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7438012361526489,
      "learning_rate": 0.0004919160915601105,
      "loss": 2.3059,
      "step": 64294
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7269113063812256,
      "learning_rate": 0.0004919129475043294,
      "loss": 2.1981,
      "step": 64295
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7128226161003113,
      "learning_rate": 0.000491909803412868,
      "loss": 2.0864,
      "step": 64296
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7353612184524536,
      "learning_rate": 0.000491906659285727,
      "loss": 2.1329,
      "step": 64297
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.6767690181732178,
      "learning_rate": 0.0004919035151229069,
      "loss": 2.2676,
      "step": 64298
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7695059776306152,
      "learning_rate": 0.0004919003709244084,
      "loss": 2.3458,
      "step": 64299
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7569847106933594,
      "learning_rate": 0.0004918972266902319,
      "loss": 2.2354,
      "step": 64300
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7177153825759888,
      "learning_rate": 0.0004918940824203781,
      "loss": 2.0885,
      "step": 64301
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.8585447669029236,
      "learning_rate": 0.0004918909381148476,
      "loss": 2.2998,
      "step": 64302
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7443433403968811,
      "learning_rate": 0.0004918877937736409,
      "loss": 2.252,
      "step": 64303
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.891211211681366,
      "learning_rate": 0.0004918846493967586,
      "loss": 2.1973,
      "step": 64304
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.8003371953964233,
      "learning_rate": 0.0004918815049842014,
      "loss": 2.0387,
      "step": 64305
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7796606421470642,
      "learning_rate": 0.0004918783605359698,
      "loss": 2.1396,
      "step": 64306
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7371402382850647,
      "learning_rate": 0.0004918752160520644,
      "loss": 2.1278,
      "step": 64307
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7600173354148865,
      "learning_rate": 0.0004918720715324857,
      "loss": 2.1735,
      "step": 64308
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7856727242469788,
      "learning_rate": 0.0004918689269772345,
      "loss": 2.4639,
      "step": 64309
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7588033080101013,
      "learning_rate": 0.0004918657823863111,
      "loss": 2.0397,
      "step": 64310
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.8118525147438049,
      "learning_rate": 0.0004918626377597163,
      "loss": 2.249,
      "step": 64311
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7016116976737976,
      "learning_rate": 0.0004918594930974507,
      "loss": 2.3117,
      "step": 64312
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7637353539466858,
      "learning_rate": 0.0004918563483995146,
      "loss": 2.1067,
      "step": 64313
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.703993558883667,
      "learning_rate": 0.0004918532036659089,
      "loss": 2.3001,
      "step": 64314
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7689394354820251,
      "learning_rate": 0.000491850058896634,
      "loss": 2.2195,
      "step": 64315
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7784748673439026,
      "learning_rate": 0.0004918469140916907,
      "loss": 2.2722,
      "step": 64316
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7342969179153442,
      "learning_rate": 0.0004918437692510792,
      "loss": 2.1778,
      "step": 64317
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.85175621509552,
      "learning_rate": 0.0004918406243748006,
      "loss": 2.4769,
      "step": 64318
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7860023975372314,
      "learning_rate": 0.000491837479462855,
      "loss": 2.1347,
      "step": 64319
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7392813563346863,
      "learning_rate": 0.0004918343345152433,
      "loss": 2.2254,
      "step": 64320
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.8191544413566589,
      "learning_rate": 0.0004918311895319659,
      "loss": 1.8986,
      "step": 64321
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7815722823143005,
      "learning_rate": 0.0004918280445130235,
      "loss": 2.2691,
      "step": 64322
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.8368569612503052,
      "learning_rate": 0.0004918248994584167,
      "loss": 2.0895,
      "step": 64323
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7708002924919128,
      "learning_rate": 0.000491821754368146,
      "loss": 2.3625,
      "step": 64324
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7802383899688721,
      "learning_rate": 0.0004918186092422119,
      "loss": 2.2741,
      "step": 64325
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7090885043144226,
      "learning_rate": 0.0004918154640806152,
      "loss": 2.2287,
      "step": 64326
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.747270405292511,
      "learning_rate": 0.0004918123188833564,
      "loss": 2.2348,
      "step": 64327
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.8480516076087952,
      "learning_rate": 0.000491809173650436,
      "loss": 2.3069,
      "step": 64328
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7368758320808411,
      "learning_rate": 0.0004918060283818547,
      "loss": 2.4425,
      "step": 64329
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7390061616897583,
      "learning_rate": 0.000491802883077613,
      "loss": 2.2311,
      "step": 64330
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7120254635810852,
      "learning_rate": 0.0004917997377377116,
      "loss": 2.3965,
      "step": 64331
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7515490651130676,
      "learning_rate": 0.0004917965923621509,
      "loss": 2.3589,
      "step": 64332
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7238510251045227,
      "learning_rate": 0.0004917934469509317,
      "loss": 1.9773,
      "step": 64333
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7996695041656494,
      "learning_rate": 0.0004917903015040544,
      "loss": 2.284,
      "step": 64334
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7156237959861755,
      "learning_rate": 0.0004917871560215197,
      "loss": 2.063,
      "step": 64335
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7858920097351074,
      "learning_rate": 0.0004917840105033282,
      "loss": 2.2609,
      "step": 64336
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7815393805503845,
      "learning_rate": 0.0004917808649494804,
      "loss": 2.0914,
      "step": 64337
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.8616796731948853,
      "learning_rate": 0.000491777719359977,
      "loss": 2.3304,
      "step": 64338
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.798317015171051,
      "learning_rate": 0.0004917745737348183,
      "loss": 2.5,
      "step": 64339
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.93495112657547,
      "learning_rate": 0.0004917714280740052,
      "loss": 2.4377,
      "step": 64340
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.9469130039215088,
      "learning_rate": 0.0004917682823775382,
      "loss": 2.3147,
      "step": 64341
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7241028547286987,
      "learning_rate": 0.0004917651366454177,
      "loss": 2.2062,
      "step": 64342
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.0843504667282104,
      "learning_rate": 0.0004917619908776446,
      "loss": 2.1269,
      "step": 64343
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.760431170463562,
      "learning_rate": 0.0004917588450742193,
      "loss": 2.2719,
      "step": 64344
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.6799136400222778,
      "learning_rate": 0.0004917556992351424,
      "loss": 2.0858,
      "step": 64345
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7528640031814575,
      "learning_rate": 0.0004917525533604145,
      "loss": 2.061,
      "step": 64346
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.8499454855918884,
      "learning_rate": 0.0004917494074500362,
      "loss": 2.1932,
      "step": 64347
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7666319608688354,
      "learning_rate": 0.000491746261504008,
      "loss": 2.0462,
      "step": 64348
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.8149219751358032,
      "learning_rate": 0.0004917431155223305,
      "loss": 2.2196,
      "step": 64349
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.8413605093955994,
      "learning_rate": 0.0004917399695050045,
      "loss": 2.5068,
      "step": 64350
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7360923886299133,
      "learning_rate": 0.0004917368234520303,
      "loss": 2.0835,
      "step": 64351
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7626315355300903,
      "learning_rate": 0.0004917336773634087,
      "loss": 2.2402,
      "step": 64352
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.8231499195098877,
      "learning_rate": 0.0004917305312391401,
      "loss": 2.2662,
      "step": 64353
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.6934120655059814,
      "learning_rate": 0.0004917273850792252,
      "loss": 2.2072,
      "step": 64354
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7352192401885986,
      "learning_rate": 0.0004917242388836646,
      "loss": 2.3345,
      "step": 64355
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.704332172870636,
      "learning_rate": 0.0004917210926524588,
      "loss": 2.2286,
      "step": 64356
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.8303017616271973,
      "learning_rate": 0.0004917179463856085,
      "loss": 2.2434,
      "step": 64357
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7360751032829285,
      "learning_rate": 0.0004917148000831141,
      "loss": 2.2782,
      "step": 64358
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7447822093963623,
      "learning_rate": 0.0004917116537449764,
      "loss": 2.1079,
      "step": 64359
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7638381123542786,
      "learning_rate": 0.0004917085073711958,
      "loss": 2.1263,
      "step": 64360
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7530330419540405,
      "learning_rate": 0.000491705360961773,
      "loss": 2.3302,
      "step": 64361
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7367590665817261,
      "learning_rate": 0.0004917022145167085,
      "loss": 2.3822,
      "step": 64362
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7268530130386353,
      "learning_rate": 0.0004916990680360031,
      "loss": 2.143,
      "step": 64363
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7077602744102478,
      "learning_rate": 0.0004916959215196571,
      "loss": 2.2638,
      "step": 64364
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.711798369884491,
      "learning_rate": 0.0004916927749676713,
      "loss": 2.1994,
      "step": 64365
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7585235834121704,
      "learning_rate": 0.000491689628380046,
      "loss": 2.2339,
      "step": 64366
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7762454748153687,
      "learning_rate": 0.0004916864817567822,
      "loss": 1.942,
      "step": 64367
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.8651689291000366,
      "learning_rate": 0.0004916833350978801,
      "loss": 2.1691,
      "step": 64368
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7721830010414124,
      "learning_rate": 0.0004916801884033407,
      "loss": 2.5033,
      "step": 64369
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.9349632263183594,
      "learning_rate": 0.0004916770416731641,
      "loss": 2.1378,
      "step": 64370
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7696642279624939,
      "learning_rate": 0.0004916738949073511,
      "loss": 2.447,
      "step": 64371
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7506290078163147,
      "learning_rate": 0.0004916707481059026,
      "loss": 2.1768,
      "step": 64372
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.8653721213340759,
      "learning_rate": 0.0004916676012688186,
      "loss": 2.2682,
      "step": 64373
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.9311633110046387,
      "learning_rate": 0.0004916644543961,
      "loss": 2.2975,
      "step": 64374
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7496954202651978,
      "learning_rate": 0.0004916613074877474,
      "loss": 2.208,
      "step": 64375
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7290092706680298,
      "learning_rate": 0.0004916581605437614,
      "loss": 2.2912,
      "step": 64376
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.132092833518982,
      "learning_rate": 0.0004916550135641425,
      "loss": 2.1902,
      "step": 64377
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.006260871887207,
      "learning_rate": 0.0004916518665488913,
      "loss": 2.0103,
      "step": 64378
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7218693494796753,
      "learning_rate": 0.0004916487194980084,
      "loss": 2.1443,
      "step": 64379
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.9127150177955627,
      "learning_rate": 0.0004916455724114944,
      "loss": 2.1689,
      "step": 64380
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.8960707187652588,
      "learning_rate": 0.0004916424252893499,
      "loss": 1.9409,
      "step": 64381
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.0825541019439697,
      "learning_rate": 0.0004916392781315754,
      "loss": 2.0124,
      "step": 64382
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.9399324059486389,
      "learning_rate": 0.0004916361309381715,
      "loss": 2.3014,
      "step": 64383
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.8526614308357239,
      "learning_rate": 0.000491632983709139,
      "loss": 2.0605,
      "step": 64384
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.9836748838424683,
      "learning_rate": 0.0004916298364444781,
      "loss": 2.2545,
      "step": 64385
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.8412644863128662,
      "learning_rate": 0.0004916266891441896,
      "loss": 2.326,
      "step": 64386
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.1018562316894531,
      "learning_rate": 0.0004916235418082743,
      "loss": 2.0268,
      "step": 64387
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.8575145602226257,
      "learning_rate": 0.0004916203944367324,
      "loss": 2.3144,
      "step": 64388
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7404701113700867,
      "learning_rate": 0.0004916172470295647,
      "loss": 2.3644,
      "step": 64389
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.855757474899292,
      "learning_rate": 0.0004916140995867718,
      "loss": 2.329,
      "step": 64390
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7743454575538635,
      "learning_rate": 0.0004916109521083541,
      "loss": 2.4357,
      "step": 64391
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7921332716941833,
      "learning_rate": 0.0004916078045943123,
      "loss": 2.34,
      "step": 64392
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7436286211013794,
      "learning_rate": 0.0004916046570446472,
      "loss": 2.2243,
      "step": 64393
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.826473593711853,
      "learning_rate": 0.0004916015094593589,
      "loss": 2.3336,
      "step": 64394
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.811531126499176,
      "learning_rate": 0.0004915983618384484,
      "loss": 2.2109,
      "step": 64395
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.8111613392829895,
      "learning_rate": 0.0004915952141819162,
      "loss": 2.4257,
      "step": 64396
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7243918776512146,
      "learning_rate": 0.0004915920664897627,
      "loss": 2.2031,
      "step": 64397
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.8517763614654541,
      "learning_rate": 0.0004915889187619887,
      "loss": 2.0881,
      "step": 64398
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7344075441360474,
      "learning_rate": 0.0004915857709985947,
      "loss": 2.2653,
      "step": 64399
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.8040122985839844,
      "learning_rate": 0.0004915826231995813,
      "loss": 2.4887,
      "step": 64400
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7376818060874939,
      "learning_rate": 0.000491579475364949,
      "loss": 2.1021,
      "step": 64401
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7967382669448853,
      "learning_rate": 0.0004915763274946985,
      "loss": 2.1685,
      "step": 64402
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7822404503822327,
      "learning_rate": 0.0004915731795888304,
      "loss": 2.2615,
      "step": 64403
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.8007011413574219,
      "learning_rate": 0.0004915700316473452,
      "loss": 2.4459,
      "step": 64404
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.8944317698478699,
      "learning_rate": 0.0004915668836702436,
      "loss": 2.3002,
      "step": 64405
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7764566540718079,
      "learning_rate": 0.0004915637356575259,
      "loss": 2.3078,
      "step": 64406
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7674484848976135,
      "learning_rate": 0.0004915605876091929,
      "loss": 2.2865,
      "step": 64407
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7261084914207458,
      "learning_rate": 0.0004915574395252454,
      "loss": 2.2271,
      "step": 64408
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7607158422470093,
      "learning_rate": 0.0004915542914056836,
      "loss": 2.1621,
      "step": 64409
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7753452658653259,
      "learning_rate": 0.0004915511432505083,
      "loss": 2.2523,
      "step": 64410
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.810744047164917,
      "learning_rate": 0.0004915479950597199,
      "loss": 2.257,
      "step": 64411
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7450436949729919,
      "learning_rate": 0.0004915448468333192,
      "loss": 2.4676,
      "step": 64412
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7876586318016052,
      "learning_rate": 0.0004915416985713067,
      "loss": 2.0723,
      "step": 64413
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7801662683486938,
      "learning_rate": 0.0004915385502736829,
      "loss": 2.2478,
      "step": 64414
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7425239086151123,
      "learning_rate": 0.0004915354019404486,
      "loss": 2.2171,
      "step": 64415
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7203046083450317,
      "learning_rate": 0.0004915322535716043,
      "loss": 2.2329,
      "step": 64416
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7732734680175781,
      "learning_rate": 0.0004915291051671503,
      "loss": 2.1775,
      "step": 64417
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7842648029327393,
      "learning_rate": 0.0004915259567270877,
      "loss": 2.2783,
      "step": 64418
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.781358003616333,
      "learning_rate": 0.0004915228082514165,
      "loss": 2.0051,
      "step": 64419
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7448058128356934,
      "learning_rate": 0.0004915196597401378,
      "loss": 2.3822,
      "step": 64420
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7791942954063416,
      "learning_rate": 0.0004915165111932519,
      "loss": 2.3146,
      "step": 64421
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7387650012969971,
      "learning_rate": 0.0004915133626107595,
      "loss": 2.2624,
      "step": 64422
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7042103409767151,
      "learning_rate": 0.0004915102139926613,
      "loss": 2.6415,
      "step": 64423
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7821498513221741,
      "learning_rate": 0.0004915070653389575,
      "loss": 2.3144,
      "step": 64424
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.6738406419754028,
      "learning_rate": 0.0004915039166496491,
      "loss": 2.1132,
      "step": 64425
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7534893751144409,
      "learning_rate": 0.0004915007679247364,
      "loss": 2.3032,
      "step": 64426
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7931748032569885,
      "learning_rate": 0.00049149761916422,
      "loss": 2.2463,
      "step": 64427
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7996885180473328,
      "learning_rate": 0.0004914944703681008,
      "loss": 2.12,
      "step": 64428
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7722697257995605,
      "learning_rate": 0.0004914913215363789,
      "loss": 2.4587,
      "step": 64429
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.747556209564209,
      "learning_rate": 0.0004914881726690554,
      "loss": 2.3302,
      "step": 64430
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.8400218486785889,
      "learning_rate": 0.0004914850237661305,
      "loss": 2.1245,
      "step": 64431
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7656227946281433,
      "learning_rate": 0.000491481874827605,
      "loss": 2.3829,
      "step": 64432
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.746864914894104,
      "learning_rate": 0.0004914787258534794,
      "loss": 2.1343,
      "step": 64433
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7238600254058838,
      "learning_rate": 0.0004914755768437543,
      "loss": 2.176,
      "step": 64434
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7348778247833252,
      "learning_rate": 0.0004914724277984301,
      "loss": 2.2071,
      "step": 64435
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.727111279964447,
      "learning_rate": 0.0004914692787175078,
      "loss": 2.1077,
      "step": 64436
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7606085538864136,
      "learning_rate": 0.0004914661296009877,
      "loss": 2.4226,
      "step": 64437
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7984577417373657,
      "learning_rate": 0.0004914629804488703,
      "loss": 2.2604,
      "step": 64438
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7821401357650757,
      "learning_rate": 0.0004914598312611563,
      "loss": 2.1323,
      "step": 64439
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.9921833872795105,
      "learning_rate": 0.0004914566820378465,
      "loss": 2.213,
      "step": 64440
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7464735507965088,
      "learning_rate": 0.0004914535327789411,
      "loss": 2.1335,
      "step": 64441
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7193393111228943,
      "learning_rate": 0.000491450383484441,
      "loss": 2.1542,
      "step": 64442
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7931236624717712,
      "learning_rate": 0.0004914472341543466,
      "loss": 2.234,
      "step": 64443
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7724589109420776,
      "learning_rate": 0.0004914440847886586,
      "loss": 2.2766,
      "step": 64444
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.669073760509491,
      "learning_rate": 0.0004914409353873775,
      "loss": 2.0142,
      "step": 64445
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7723297476768494,
      "learning_rate": 0.000491437785950504,
      "loss": 2.2979,
      "step": 64446
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7584187388420105,
      "learning_rate": 0.0004914346364780384,
      "loss": 2.1877,
      "step": 64447
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.8954366445541382,
      "learning_rate": 0.0004914314869699816,
      "loss": 2.361,
      "step": 64448
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.8760128021240234,
      "learning_rate": 0.0004914283374263341,
      "loss": 2.1121,
      "step": 64449
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.8295631408691406,
      "learning_rate": 0.0004914251878470964,
      "loss": 2.2345,
      "step": 64450
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7403667569160461,
      "learning_rate": 0.0004914220382322692,
      "loss": 1.9919,
      "step": 64451
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7165793180465698,
      "learning_rate": 0.0004914188885818529,
      "loss": 2.3115,
      "step": 64452
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7898992896080017,
      "learning_rate": 0.0004914157388958484,
      "loss": 2.4224,
      "step": 64453
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.8003596067428589,
      "learning_rate": 0.000491412589174256,
      "loss": 2.2601,
      "step": 64454
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.860974907875061,
      "learning_rate": 0.0004914094394170765,
      "loss": 2.2122,
      "step": 64455
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7656151652336121,
      "learning_rate": 0.0004914062896243102,
      "loss": 2.2768,
      "step": 64456
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.8947401642799377,
      "learning_rate": 0.000491403139795958,
      "loss": 2.3345,
      "step": 64457
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.8210756778717041,
      "learning_rate": 0.0004913999899320203,
      "loss": 1.9739,
      "step": 64458
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.8453981876373291,
      "learning_rate": 0.0004913968400324976,
      "loss": 2.2167,
      "step": 64459
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7530944347381592,
      "learning_rate": 0.0004913936900973907,
      "loss": 2.0815,
      "step": 64460
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7752512693405151,
      "learning_rate": 0.0004913905401267002,
      "loss": 2.2307,
      "step": 64461
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7710752487182617,
      "learning_rate": 0.0004913873901204264,
      "loss": 2.3302,
      "step": 64462
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.9130232930183411,
      "learning_rate": 0.0004913842400785702,
      "loss": 2.2094,
      "step": 64463
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7699986100196838,
      "learning_rate": 0.0004913810900011321,
      "loss": 2.1751,
      "step": 64464
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7570995688438416,
      "learning_rate": 0.0004913779398881126,
      "loss": 2.3121,
      "step": 64465
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.72955721616745,
      "learning_rate": 0.0004913747897395122,
      "loss": 2.0499,
      "step": 64466
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7226343750953674,
      "learning_rate": 0.0004913716395553318,
      "loss": 2.2055,
      "step": 64467
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7404862642288208,
      "learning_rate": 0.0004913684893355717,
      "loss": 2.2911,
      "step": 64468
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7760653495788574,
      "learning_rate": 0.0004913653390802325,
      "loss": 2.4051,
      "step": 64469
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7595347166061401,
      "learning_rate": 0.0004913621887893151,
      "loss": 2.1715,
      "step": 64470
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7014979124069214,
      "learning_rate": 0.0004913590384628196,
      "loss": 2.2759,
      "step": 64471
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7475852370262146,
      "learning_rate": 0.0004913558881007471,
      "loss": 2.2994,
      "step": 64472
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7429772615432739,
      "learning_rate": 0.0004913527377030978,
      "loss": 2.2824,
      "step": 64473
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.6880685687065125,
      "learning_rate": 0.0004913495872698724,
      "loss": 2.2101,
      "step": 64474
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.8044321537017822,
      "learning_rate": 0.0004913464368010716,
      "loss": 2.1233,
      "step": 64475
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7549019455909729,
      "learning_rate": 0.0004913432862966958,
      "loss": 2.2696,
      "step": 64476
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.699230968952179,
      "learning_rate": 0.0004913401357567456,
      "loss": 2.3567,
      "step": 64477
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7789635062217712,
      "learning_rate": 0.0004913369851812217,
      "loss": 2.3499,
      "step": 64478
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.8536040782928467,
      "learning_rate": 0.0004913338345701248,
      "loss": 2.1343,
      "step": 64479
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7600874900817871,
      "learning_rate": 0.0004913306839234551,
      "loss": 2.2182,
      "step": 64480
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.808853268623352,
      "learning_rate": 0.0004913275332412136,
      "loss": 2.439,
      "step": 64481
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7356621623039246,
      "learning_rate": 0.0004913243825234006,
      "loss": 2.1988,
      "step": 64482
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7861109972000122,
      "learning_rate": 0.0004913212317700168,
      "loss": 2.2274,
      "step": 64483
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.6958956122398376,
      "learning_rate": 0.0004913180809810628,
      "loss": 2.305,
      "step": 64484
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7980065941810608,
      "learning_rate": 0.0004913149301565391,
      "loss": 2.128,
      "step": 64485
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7347015142440796,
      "learning_rate": 0.0004913117792964463,
      "loss": 2.1684,
      "step": 64486
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7671777606010437,
      "learning_rate": 0.0004913086284007852,
      "loss": 2.0669,
      "step": 64487
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.8843520283699036,
      "learning_rate": 0.0004913054774695561,
      "loss": 2.2023,
      "step": 64488
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7703608274459839,
      "learning_rate": 0.0004913023265027597,
      "loss": 2.2941,
      "step": 64489
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.8179362416267395,
      "learning_rate": 0.0004912991755003966,
      "loss": 2.2422,
      "step": 64490
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.8074969053268433,
      "learning_rate": 0.0004912960244624674,
      "loss": 2.5468,
      "step": 64491
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7505525946617126,
      "learning_rate": 0.0004912928733889726,
      "loss": 2.2722,
      "step": 64492
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7333809733390808,
      "learning_rate": 0.0004912897222799129,
      "loss": 2.2237,
      "step": 64493
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7707019448280334,
      "learning_rate": 0.0004912865711352888,
      "loss": 2.141,
      "step": 64494
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7551590204238892,
      "learning_rate": 0.0004912834199551009,
      "loss": 2.0712,
      "step": 64495
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.8143525719642639,
      "learning_rate": 0.0004912802687393499,
      "loss": 2.0222,
      "step": 64496
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7691647410392761,
      "learning_rate": 0.0004912771174880362,
      "loss": 2.2306,
      "step": 64497
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7115282416343689,
      "learning_rate": 0.0004912739662011604,
      "loss": 2.3825,
      "step": 64498
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.847825288772583,
      "learning_rate": 0.0004912708148787233,
      "loss": 2.1269,
      "step": 64499
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7363318800926208,
      "learning_rate": 0.0004912676635207253,
      "loss": 2.2309,
      "step": 64500
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7625648975372314,
      "learning_rate": 0.000491264512127167,
      "loss": 2.2218,
      "step": 64501
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7438958287239075,
      "learning_rate": 0.000491261360698049,
      "loss": 2.2022,
      "step": 64502
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.6996598839759827,
      "learning_rate": 0.000491258209233372,
      "loss": 2.2223,
      "step": 64503
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.6791876554489136,
      "learning_rate": 0.0004912550577331364,
      "loss": 2.1043,
      "step": 64504
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7715825438499451,
      "learning_rate": 0.0004912519061973429,
      "loss": 2.2681,
      "step": 64505
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.715095579624176,
      "learning_rate": 0.0004912487546259921,
      "loss": 2.0746,
      "step": 64506
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.6960804462432861,
      "learning_rate": 0.0004912456030190845,
      "loss": 2.2296,
      "step": 64507
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.8050299286842346,
      "learning_rate": 0.0004912424513766208,
      "loss": 2.2873,
      "step": 64508
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.6950493454933167,
      "learning_rate": 0.0004912392996986015,
      "loss": 2.2967,
      "step": 64509
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7343803644180298,
      "learning_rate": 0.000491236147985027,
      "loss": 2.2805,
      "step": 64510
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.8186055421829224,
      "learning_rate": 0.0004912329962358983,
      "loss": 2.2246,
      "step": 64511
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7466335296630859,
      "learning_rate": 0.0004912298444512158,
      "loss": 2.0336,
      "step": 64512
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7529709935188293,
      "learning_rate": 0.00049122669263098,
      "loss": 2.2734,
      "step": 64513
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7751848101615906,
      "learning_rate": 0.0004912235407751915,
      "loss": 2.3056,
      "step": 64514
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7909331321716309,
      "learning_rate": 0.000491220388883851,
      "loss": 2.1841,
      "step": 64515
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7978523373603821,
      "learning_rate": 0.000491217236956959,
      "loss": 2.0455,
      "step": 64516
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.6741541624069214,
      "learning_rate": 0.0004912140849945161,
      "loss": 2.2292,
      "step": 64517
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7142999768257141,
      "learning_rate": 0.0004912109329965229,
      "loss": 2.1608,
      "step": 64518
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.8692218065261841,
      "learning_rate": 0.0004912077809629801,
      "loss": 2.254,
      "step": 64519
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7521561980247498,
      "learning_rate": 0.0004912046288938879,
      "loss": 2.0858,
      "step": 64520
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.853951096534729,
      "learning_rate": 0.0004912014767892473,
      "loss": 2.184,
      "step": 64521
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.8085993528366089,
      "learning_rate": 0.0004911983246490588,
      "loss": 2.3257,
      "step": 64522
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.764933168888092,
      "learning_rate": 0.0004911951724733228,
      "loss": 2.2562,
      "step": 64523
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.8205408453941345,
      "learning_rate": 0.0004911920202620401,
      "loss": 2.2706,
      "step": 64524
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7468656301498413,
      "learning_rate": 0.0004911888680152111,
      "loss": 2.0926,
      "step": 64525
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7335783839225769,
      "learning_rate": 0.0004911857157328365,
      "loss": 2.1943,
      "step": 64526
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7866978645324707,
      "learning_rate": 0.000491182563414917,
      "loss": 2.302,
      "step": 64527
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7572053074836731,
      "learning_rate": 0.0004911794110614529,
      "loss": 2.352,
      "step": 64528
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7514106035232544,
      "learning_rate": 0.000491176258672445,
      "loss": 2.1803,
      "step": 64529
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.737436056137085,
      "learning_rate": 0.0004911731062478939,
      "loss": 2.1158,
      "step": 64530
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.8111647963523865,
      "learning_rate": 0.0004911699537878,
      "loss": 2.2458,
      "step": 64531
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.8368048667907715,
      "learning_rate": 0.0004911668012921639,
      "loss": 2.2428,
      "step": 64532
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.8226460814476013,
      "learning_rate": 0.0004911636487609865,
      "loss": 1.9302,
      "step": 64533
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7728598117828369,
      "learning_rate": 0.000491160496194268,
      "loss": 2.1132,
      "step": 64534
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.829526424407959,
      "learning_rate": 0.0004911573435920092,
      "loss": 2.3146,
      "step": 64535
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7553761601448059,
      "learning_rate": 0.0004911541909542107,
      "loss": 2.2965,
      "step": 64536
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7615548968315125,
      "learning_rate": 0.000491151038280873,
      "loss": 2.059,
      "step": 64537
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.8238176703453064,
      "learning_rate": 0.0004911478855719967,
      "loss": 2.2417,
      "step": 64538
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.817482590675354,
      "learning_rate": 0.0004911447328275825,
      "loss": 2.2802,
      "step": 64539
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7635550498962402,
      "learning_rate": 0.0004911415800476308,
      "loss": 2.0438,
      "step": 64540
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.8940389156341553,
      "learning_rate": 0.0004911384272321423,
      "loss": 2.2093,
      "step": 64541
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7351552248001099,
      "learning_rate": 0.0004911352743811175,
      "loss": 2.2974,
      "step": 64542
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7553096413612366,
      "learning_rate": 0.000491132121494557,
      "loss": 2.1483,
      "step": 64543
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.8006670475006104,
      "learning_rate": 0.0004911289685724616,
      "loss": 2.2425,
      "step": 64544
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7382933497428894,
      "learning_rate": 0.0004911258156148316,
      "loss": 2.0585,
      "step": 64545
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7622928023338318,
      "learning_rate": 0.0004911226626216677,
      "loss": 2.4256,
      "step": 64546
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.8723600506782532,
      "learning_rate": 0.0004911195095929704,
      "loss": 2.0302,
      "step": 64547
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.906557023525238,
      "learning_rate": 0.0004911163565287406,
      "loss": 2.1395,
      "step": 64548
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7142691612243652,
      "learning_rate": 0.0004911132034289785,
      "loss": 2.3592,
      "step": 64549
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7238280177116394,
      "learning_rate": 0.0004911100502936849,
      "loss": 2.3093,
      "step": 64550
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.8456715941429138,
      "learning_rate": 0.0004911068971228604,
      "loss": 2.3134,
      "step": 64551
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7998652458190918,
      "learning_rate": 0.0004911037439165053,
      "loss": 2.3597,
      "step": 64552
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.6838268041610718,
      "learning_rate": 0.0004911005906746206,
      "loss": 1.9855,
      "step": 64553
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7114728689193726,
      "learning_rate": 0.0004910974373972067,
      "loss": 2.4736,
      "step": 64554
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.0212478637695312,
      "learning_rate": 0.0004910942840842641,
      "loss": 2.0697,
      "step": 64555
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7872769236564636,
      "learning_rate": 0.0004910911307357935,
      "loss": 2.2753,
      "step": 64556
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7625014185905457,
      "learning_rate": 0.0004910879773517955,
      "loss": 2.2526,
      "step": 64557
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7782856225967407,
      "learning_rate": 0.0004910848239322705,
      "loss": 2.5303,
      "step": 64558
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.0297011137008667,
      "learning_rate": 0.0004910816704772194,
      "loss": 2.4064,
      "step": 64559
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7925699353218079,
      "learning_rate": 0.0004910785169866426,
      "loss": 2.5198,
      "step": 64560
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7587132453918457,
      "learning_rate": 0.0004910753634605405,
      "loss": 2.399,
      "step": 64561
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.6774784326553345,
      "learning_rate": 0.0004910722098989141,
      "loss": 2.2686,
      "step": 64562
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7882328033447266,
      "learning_rate": 0.0004910690563017636,
      "loss": 2.2577,
      "step": 64563
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.870449960231781,
      "learning_rate": 0.0004910659026690899,
      "loss": 2.3705,
      "step": 64564
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7329701781272888,
      "learning_rate": 0.0004910627490008933,
      "loss": 2.1756,
      "step": 64565
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7905530333518982,
      "learning_rate": 0.0004910595952971746,
      "loss": 2.2479,
      "step": 64566
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.8397776484489441,
      "learning_rate": 0.0004910564415579343,
      "loss": 2.0353,
      "step": 64567
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7401064038276672,
      "learning_rate": 0.000491053287783173,
      "loss": 2.3023,
      "step": 64568
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.8323842287063599,
      "learning_rate": 0.0004910501339728913,
      "loss": 2.171,
      "step": 64569
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.8586673140525818,
      "learning_rate": 0.0004910469801270898,
      "loss": 2.219,
      "step": 64570
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.6967823505401611,
      "learning_rate": 0.000491043826245769,
      "loss": 2.0546,
      "step": 64571
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7193940281867981,
      "learning_rate": 0.0004910406723289296,
      "loss": 2.3907,
      "step": 64572
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.8357440233230591,
      "learning_rate": 0.0004910375183765722,
      "loss": 2.1941,
      "step": 64573
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.73862224817276,
      "learning_rate": 0.0004910343643886972,
      "loss": 2.3293,
      "step": 64574
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7098276615142822,
      "learning_rate": 0.0004910312103653053,
      "loss": 1.9857,
      "step": 64575
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.799428403377533,
      "learning_rate": 0.0004910280563063972,
      "loss": 2.3784,
      "step": 64576
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.6979106068611145,
      "learning_rate": 0.0004910249022119733,
      "loss": 2.3459,
      "step": 64577
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7169642448425293,
      "learning_rate": 0.0004910217480820342,
      "loss": 2.331,
      "step": 64578
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.9425381422042847,
      "learning_rate": 0.0004910185939165807,
      "loss": 2.3428,
      "step": 64579
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7434072494506836,
      "learning_rate": 0.000491015439715613,
      "loss": 2.2312,
      "step": 64580
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.6911755204200745,
      "learning_rate": 0.0004910122854791321,
      "loss": 1.8391,
      "step": 64581
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7152320146560669,
      "learning_rate": 0.0004910091312071385,
      "loss": 2.2735,
      "step": 64582
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7270616888999939,
      "learning_rate": 0.0004910059768996326,
      "loss": 2.2594,
      "step": 64583
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7974346876144409,
      "learning_rate": 0.0004910028225566151,
      "loss": 2.1906,
      "step": 64584
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.9043161273002625,
      "learning_rate": 0.0004909996681780865,
      "loss": 2.1139,
      "step": 64585
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.8819581866264343,
      "learning_rate": 0.0004909965137640475,
      "loss": 2.1345,
      "step": 64586
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7992643713951111,
      "learning_rate": 0.0004909933593144986,
      "loss": 2.2044,
      "step": 64587
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.8788698315620422,
      "learning_rate": 0.0004909902048294405,
      "loss": 2.1357,
      "step": 64588
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.9061776399612427,
      "learning_rate": 0.0004909870503088737,
      "loss": 2.2632,
      "step": 64589
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.8020954728126526,
      "learning_rate": 0.0004909838957527988,
      "loss": 2.2234,
      "step": 64590
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7813624143600464,
      "learning_rate": 0.0004909807411612164,
      "loss": 2.2068,
      "step": 64591
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.9223446249961853,
      "learning_rate": 0.0004909775865341271,
      "loss": 2.3149,
      "step": 64592
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.8029438257217407,
      "learning_rate": 0.0004909744318715313,
      "loss": 2.3611,
      "step": 64593
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.8195558190345764,
      "learning_rate": 0.0004909712771734299,
      "loss": 2.2002,
      "step": 64594
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.9510142803192139,
      "learning_rate": 0.0004909681224398234,
      "loss": 2.3149,
      "step": 64595
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.9670472145080566,
      "learning_rate": 0.0004909649676707121,
      "loss": 2.1803,
      "step": 64596
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.9667044281959534,
      "learning_rate": 0.000490961812866097,
      "loss": 2.0726,
      "step": 64597
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.0159342288970947,
      "learning_rate": 0.0004909586580259784,
      "loss": 2.1901,
      "step": 64598
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7913203239440918,
      "learning_rate": 0.000490955503150357,
      "loss": 2.3348,
      "step": 64599
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7334729433059692,
      "learning_rate": 0.0004909523482392334,
      "loss": 2.2668,
      "step": 64600
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7118069529533386,
      "learning_rate": 0.0004909491932926081,
      "loss": 2.3576,
      "step": 64601
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.8760920166969299,
      "learning_rate": 0.0004909460383104818,
      "loss": 2.4065,
      "step": 64602
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7732147574424744,
      "learning_rate": 0.0004909428832928549,
      "loss": 2.4196,
      "step": 64603
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7780721187591553,
      "learning_rate": 0.0004909397282397282,
      "loss": 2.0434,
      "step": 64604
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7975409626960754,
      "learning_rate": 0.0004909365731511023,
      "loss": 2.1192,
      "step": 64605
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7206987142562866,
      "learning_rate": 0.0004909334180269777,
      "loss": 2.4235,
      "step": 64606
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7692193388938904,
      "learning_rate": 0.0004909302628673547,
      "loss": 2.2561,
      "step": 64607
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.852414071559906,
      "learning_rate": 0.0004909271076722344,
      "loss": 2.1741,
      "step": 64608
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7061490416526794,
      "learning_rate": 0.000490923952441617,
      "loss": 2.2726,
      "step": 64609
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.8306413888931274,
      "learning_rate": 0.0004909207971755033,
      "loss": 2.3038,
      "step": 64610
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7568352818489075,
      "learning_rate": 0.0004909176418738939,
      "loss": 2.3712,
      "step": 64611
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7748658657073975,
      "learning_rate": 0.0004909144865367892,
      "loss": 2.1026,
      "step": 64612
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7666674256324768,
      "learning_rate": 0.0004909113311641899,
      "loss": 2.2237,
      "step": 64613
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7148270010948181,
      "learning_rate": 0.0004909081757560966,
      "loss": 2.0724,
      "step": 64614
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7823917269706726,
      "learning_rate": 0.00049090502031251,
      "loss": 2.1321,
      "step": 64615
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7736852765083313,
      "learning_rate": 0.0004909018648334303,
      "loss": 2.2947,
      "step": 64616
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.8096457123756409,
      "learning_rate": 0.0004908987093188586,
      "loss": 2.0833,
      "step": 64617
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7342663407325745,
      "learning_rate": 0.000490895553768795,
      "loss": 2.1612,
      "step": 64618
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7455158233642578,
      "learning_rate": 0.0004908923981832404,
      "loss": 2.3359,
      "step": 64619
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.3226059675216675,
      "learning_rate": 0.0004908892425621953,
      "loss": 2.1703,
      "step": 64620
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.0172498226165771,
      "learning_rate": 0.0004908860869056604,
      "loss": 2.2641,
      "step": 64621
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7573636174201965,
      "learning_rate": 0.000490882931213636,
      "loss": 2.1255,
      "step": 64622
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7565974593162537,
      "learning_rate": 0.0004908797754861229,
      "loss": 2.2131,
      "step": 64623
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.2232812643051147,
      "learning_rate": 0.0004908766197231217,
      "loss": 2.2431,
      "step": 64624
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7876430749893188,
      "learning_rate": 0.0004908734639246329,
      "loss": 2.2799,
      "step": 64625
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.8161941766738892,
      "learning_rate": 0.0004908703080906572,
      "loss": 2.3753,
      "step": 64626
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.9681396484375,
      "learning_rate": 0.0004908671522211949,
      "loss": 2.3621,
      "step": 64627
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7751269340515137,
      "learning_rate": 0.000490863996316247,
      "loss": 2.2141,
      "step": 64628
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.9371489882469177,
      "learning_rate": 0.0004908608403758138,
      "loss": 2.3648,
      "step": 64629
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7799476385116577,
      "learning_rate": 0.000490857684399896,
      "loss": 2.0634,
      "step": 64630
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.8047109246253967,
      "learning_rate": 0.0004908545283884941,
      "loss": 2.3731,
      "step": 64631
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7574213147163391,
      "learning_rate": 0.0004908513723416088,
      "loss": 2.3117,
      "step": 64632
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.8015221953392029,
      "learning_rate": 0.0004908482162592406,
      "loss": 2.2588,
      "step": 64633
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.735656201839447,
      "learning_rate": 0.0004908450601413901,
      "loss": 2.2786,
      "step": 64634
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.8370814919471741,
      "learning_rate": 0.0004908419039880579,
      "loss": 2.3595,
      "step": 64635
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.729928195476532,
      "learning_rate": 0.0004908387477992446,
      "loss": 2.2993,
      "step": 64636
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7892988920211792,
      "learning_rate": 0.0004908355915749508,
      "loss": 2.4646,
      "step": 64637
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.6785964369773865,
      "learning_rate": 0.000490832435315177,
      "loss": 2.0463,
      "step": 64638
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7948586344718933,
      "learning_rate": 0.0004908292790199238,
      "loss": 2.4026,
      "step": 64639
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.8398773074150085,
      "learning_rate": 0.000490826122689192,
      "loss": 2.115,
      "step": 64640
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7337456941604614,
      "learning_rate": 0.0004908229663229819,
      "loss": 2.2326,
      "step": 64641
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7983689308166504,
      "learning_rate": 0.0004908198099212943,
      "loss": 2.1303,
      "step": 64642
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.8046389222145081,
      "learning_rate": 0.0004908166534841296,
      "loss": 2.0942,
      "step": 64643
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7341111898422241,
      "learning_rate": 0.0004908134970114886,
      "loss": 2.3006,
      "step": 64644
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7362526059150696,
      "learning_rate": 0.0004908103405033716,
      "loss": 2.2876,
      "step": 64645
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7405117154121399,
      "learning_rate": 0.0004908071839597795,
      "loss": 2.1509,
      "step": 64646
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.8014276027679443,
      "learning_rate": 0.0004908040273807127,
      "loss": 2.2125,
      "step": 64647
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7916624546051025,
      "learning_rate": 0.0004908008707661718,
      "loss": 2.3644,
      "step": 64648
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7505239248275757,
      "learning_rate": 0.0004907977141161574,
      "loss": 2.2146,
      "step": 64649
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7806364893913269,
      "learning_rate": 0.0004907945574306702,
      "loss": 2.2385,
      "step": 64650
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.6860321164131165,
      "learning_rate": 0.0004907914007097105,
      "loss": 2.2509,
      "step": 64651
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7856206297874451,
      "learning_rate": 0.0004907882439532794,
      "loss": 2.0097,
      "step": 64652
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7705258131027222,
      "learning_rate": 0.0004907850871613768,
      "loss": 2.2661,
      "step": 64653
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7051074504852295,
      "learning_rate": 0.0004907819303340039,
      "loss": 2.3663,
      "step": 64654
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7071624398231506,
      "learning_rate": 0.000490778773471161,
      "loss": 2.0672,
      "step": 64655
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7859556674957275,
      "learning_rate": 0.0004907756165728487,
      "loss": 2.5059,
      "step": 64656
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7739690542221069,
      "learning_rate": 0.0004907724596390676,
      "loss": 2.4016,
      "step": 64657
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.6828001141548157,
      "learning_rate": 0.0004907693026698183,
      "loss": 2.1385,
      "step": 64658
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.082092523574829,
      "learning_rate": 0.0004907661456651013,
      "loss": 2.2543,
      "step": 64659
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.817111611366272,
      "learning_rate": 0.0004907629886249174,
      "loss": 2.2092,
      "step": 64660
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.98216712474823,
      "learning_rate": 0.0004907598315492671,
      "loss": 2.091,
      "step": 64661
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.0223731994628906,
      "learning_rate": 0.0004907566744381508,
      "loss": 2.4517,
      "step": 64662
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.6968642473220825,
      "learning_rate": 0.0004907535172915693,
      "loss": 2.3504,
      "step": 64663
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.9786227941513062,
      "learning_rate": 0.000490750360109523,
      "loss": 2.0983,
      "step": 64664
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.0045721530914307,
      "learning_rate": 0.0004907472028920128,
      "loss": 2.3924,
      "step": 64665
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.724767804145813,
      "learning_rate": 0.000490744045639039,
      "loss": 2.2683,
      "step": 64666
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7215383052825928,
      "learning_rate": 0.0004907408883506023,
      "loss": 2.1151,
      "step": 64667
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.0408506393432617,
      "learning_rate": 0.0004907377310267032,
      "loss": 2.1309,
      "step": 64668
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7671647071838379,
      "learning_rate": 0.0004907345736673425,
      "loss": 2.1261,
      "step": 64669
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.6597037315368652,
      "learning_rate": 0.0004907314162725206,
      "loss": 2.0898,
      "step": 64670
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.8130272626876831,
      "learning_rate": 0.0004907282588422381,
      "loss": 2.2641,
      "step": 64671
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.8520277738571167,
      "learning_rate": 0.0004907251013764955,
      "loss": 2.261,
      "step": 64672
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.8192871809005737,
      "learning_rate": 0.0004907219438752936,
      "loss": 2.0644,
      "step": 64673
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7427529096603394,
      "learning_rate": 0.000490718786338633,
      "loss": 1.9135,
      "step": 64674
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.752686619758606,
      "learning_rate": 0.000490715628766514,
      "loss": 2.2377,
      "step": 64675
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.836212694644928,
      "learning_rate": 0.0004907124711589375,
      "loss": 2.2315,
      "step": 64676
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.8065704107284546,
      "learning_rate": 0.000490709313515904,
      "loss": 2.405,
      "step": 64677
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.8416355848312378,
      "learning_rate": 0.0004907061558374139,
      "loss": 2.2945,
      "step": 64678
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.8241569399833679,
      "learning_rate": 0.0004907029981234679,
      "loss": 2.1405,
      "step": 64679
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7201299667358398,
      "learning_rate": 0.0004906998403740667,
      "loss": 2.0455,
      "step": 64680
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.8030179738998413,
      "learning_rate": 0.0004906966825892107,
      "loss": 2.3616,
      "step": 64681
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7332613468170166,
      "learning_rate": 0.0004906935247689007,
      "loss": 2.2254,
      "step": 64682
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7598405480384827,
      "learning_rate": 0.0004906903669131371,
      "loss": 2.2328,
      "step": 64683
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.6812275648117065,
      "learning_rate": 0.0004906872090219206,
      "loss": 2.1185,
      "step": 64684
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7942724227905273,
      "learning_rate": 0.0004906840510952517,
      "loss": 2.3153,
      "step": 64685
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7702493667602539,
      "learning_rate": 0.0004906808931331312,
      "loss": 2.1597,
      "step": 64686
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7366200089454651,
      "learning_rate": 0.0004906777351355593,
      "loss": 2.1304,
      "step": 64687
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.8792915344238281,
      "learning_rate": 0.0004906745771025371,
      "loss": 2.4717,
      "step": 64688
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.8079315423965454,
      "learning_rate": 0.0004906714190340646,
      "loss": 2.052,
      "step": 64689
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7591418027877808,
      "learning_rate": 0.0004906682609301428,
      "loss": 2.2148,
      "step": 64690
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.726161003112793,
      "learning_rate": 0.0004906651027907723,
      "loss": 2.3696,
      "step": 64691
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7693800330162048,
      "learning_rate": 0.0004906619446159533,
      "loss": 2.1357,
      "step": 64692
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.8588165044784546,
      "learning_rate": 0.0004906587864056867,
      "loss": 2.0725,
      "step": 64693
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.749806821346283,
      "learning_rate": 0.0004906556281599731,
      "loss": 2.0844,
      "step": 64694
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7899920344352722,
      "learning_rate": 0.0004906524698788131,
      "loss": 2.4092,
      "step": 64695
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7815895080566406,
      "learning_rate": 0.0004906493115622071,
      "loss": 2.1598,
      "step": 64696
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.6839218139648438,
      "learning_rate": 0.0004906461532101558,
      "loss": 2.1066,
      "step": 64697
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.8166754841804504,
      "learning_rate": 0.0004906429948226599,
      "loss": 2.3191,
      "step": 64698
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.8303605914115906,
      "learning_rate": 0.0004906398363997197,
      "loss": 2.1658,
      "step": 64699
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7056317925453186,
      "learning_rate": 0.000490636677941336,
      "loss": 1.8915,
      "step": 64700
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.6983886957168579,
      "learning_rate": 0.0004906335194475094,
      "loss": 2.0874,
      "step": 64701
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7238476276397705,
      "learning_rate": 0.0004906303609182404,
      "loss": 2.357,
      "step": 64702
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7171868681907654,
      "learning_rate": 0.0004906272023535295,
      "loss": 2.193,
      "step": 64703
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7418792247772217,
      "learning_rate": 0.0004906240437533776,
      "loss": 2.2462,
      "step": 64704
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.8062610626220703,
      "learning_rate": 0.000490620885117785,
      "loss": 2.4515,
      "step": 64705
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.6966450810432434,
      "learning_rate": 0.0004906177264467525,
      "loss": 2.0233,
      "step": 64706
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7811427712440491,
      "learning_rate": 0.0004906145677402805,
      "loss": 2.2579,
      "step": 64707
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.8769593834877014,
      "learning_rate": 0.0004906114089983695,
      "loss": 2.2937,
      "step": 64708
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7552400827407837,
      "learning_rate": 0.0004906082502210204,
      "loss": 2.4111,
      "step": 64709
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7458386421203613,
      "learning_rate": 0.0004906050914082336,
      "loss": 2.2995,
      "step": 64710
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7177781462669373,
      "learning_rate": 0.0004906019325600096,
      "loss": 2.1437,
      "step": 64711
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.8123043775558472,
      "learning_rate": 0.0004905987736763493,
      "loss": 2.2783,
      "step": 64712
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.8146939277648926,
      "learning_rate": 0.0004905956147572529,
      "loss": 2.2571,
      "step": 64713
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7104843854904175,
      "learning_rate": 0.0004905924558027213,
      "loss": 2.4185,
      "step": 64714
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7197859883308411,
      "learning_rate": 0.000490589296812755,
      "loss": 2.3124,
      "step": 64715
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.8766235709190369,
      "learning_rate": 0.0004905861377873545,
      "loss": 2.1639,
      "step": 64716
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.6808230876922607,
      "learning_rate": 0.0004905829787265204,
      "loss": 2.1782,
      "step": 64717
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.8703722953796387,
      "learning_rate": 0.0004905798196302533,
      "loss": 2.082,
      "step": 64718
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.8054998517036438,
      "learning_rate": 0.000490576660498554,
      "loss": 2.2755,
      "step": 64719
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.9999683499336243,
      "learning_rate": 0.0004905735013314227,
      "loss": 2.0573,
      "step": 64720
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7798671722412109,
      "learning_rate": 0.0004905703421288603,
      "loss": 2.1733,
      "step": 64721
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.8021388053894043,
      "learning_rate": 0.0004905671828908672,
      "loss": 2.3191,
      "step": 64722
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7927938103675842,
      "learning_rate": 0.000490564023617444,
      "loss": 2.1587,
      "step": 64723
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7146517634391785,
      "learning_rate": 0.0004905608643085915,
      "loss": 2.2136,
      "step": 64724
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7778415083885193,
      "learning_rate": 0.0004905577049643101,
      "loss": 2.1318,
      "step": 64725
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7079577445983887,
      "learning_rate": 0.0004905545455846004,
      "loss": 2.2846,
      "step": 64726
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.740884006023407,
      "learning_rate": 0.000490551386169463,
      "loss": 2.2171,
      "step": 64727
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.8571507930755615,
      "learning_rate": 0.0004905482267188985,
      "loss": 2.3931,
      "step": 64728
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.6733075976371765,
      "learning_rate": 0.0004905450672329076,
      "loss": 2.261,
      "step": 64729
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.8362041711807251,
      "learning_rate": 0.0004905419077114907,
      "loss": 2.3646,
      "step": 64730
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7578575611114502,
      "learning_rate": 0.0004905387481546484,
      "loss": 2.2882,
      "step": 64731
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.6631078720092773,
      "learning_rate": 0.0004905355885623814,
      "loss": 2.2354,
      "step": 64732
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.8377830386161804,
      "learning_rate": 0.0004905324289346903,
      "loss": 2.0688,
      "step": 64733
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.8424639105796814,
      "learning_rate": 0.0004905292692715755,
      "loss": 2.3164,
      "step": 64734
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.743857204914093,
      "learning_rate": 0.0004905261095730377,
      "loss": 2.1532,
      "step": 64735
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.8368749618530273,
      "learning_rate": 0.0004905229498390777,
      "loss": 2.2011,
      "step": 64736
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7237143516540527,
      "learning_rate": 0.0004905197900696958,
      "loss": 2.0686,
      "step": 64737
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.8027551770210266,
      "learning_rate": 0.0004905166302648926,
      "loss": 2.1145,
      "step": 64738
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7177592515945435,
      "learning_rate": 0.0004905134704246688,
      "loss": 2.2646,
      "step": 64739
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.8803791999816895,
      "learning_rate": 0.000490510310549025,
      "loss": 2.216,
      "step": 64740
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7056519985198975,
      "learning_rate": 0.0004905071506379617,
      "loss": 2.3437,
      "step": 64741
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7639529705047607,
      "learning_rate": 0.0004905039906914795,
      "loss": 2.2046,
      "step": 64742
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.9391871094703674,
      "learning_rate": 0.000490500830709579,
      "loss": 2.2321,
      "step": 64743
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7453048825263977,
      "learning_rate": 0.0004904976706922608,
      "loss": 2.3165,
      "step": 64744
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.833893895149231,
      "learning_rate": 0.0004904945106395254,
      "loss": 2.1544,
      "step": 64745
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.8216849565505981,
      "learning_rate": 0.0004904913505513736,
      "loss": 2.3108,
      "step": 64746
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7363691926002502,
      "learning_rate": 0.0004904881904278059,
      "loss": 2.2726,
      "step": 64747
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.8080283999443054,
      "learning_rate": 0.0004904850302688227,
      "loss": 2.2603,
      "step": 64748
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7413451671600342,
      "learning_rate": 0.0004904818700744249,
      "loss": 2.3134,
      "step": 64749
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7913427352905273,
      "learning_rate": 0.0004904787098446128,
      "loss": 2.3933,
      "step": 64750
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.9190649390220642,
      "learning_rate": 0.0004904755495793872,
      "loss": 2.6585,
      "step": 64751
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.8283683061599731,
      "learning_rate": 0.0004904723892787487,
      "loss": 2.3672,
      "step": 64752
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7201704978942871,
      "learning_rate": 0.0004904692289426976,
      "loss": 2.2181,
      "step": 64753
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7139789462089539,
      "learning_rate": 0.0004904660685712347,
      "loss": 2.2651,
      "step": 64754
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7555100321769714,
      "learning_rate": 0.0004904629081643607,
      "loss": 2.2696,
      "step": 64755
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7702030539512634,
      "learning_rate": 0.0004904597477220758,
      "loss": 2.1833,
      "step": 64756
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.743320882320404,
      "learning_rate": 0.000490456587244381,
      "loss": 1.9884,
      "step": 64757
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7315993309020996,
      "learning_rate": 0.0004904534267312767,
      "loss": 2.3798,
      "step": 64758
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.9006924629211426,
      "learning_rate": 0.0004904502661827636,
      "loss": 2.3938,
      "step": 64759
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.9803106188774109,
      "learning_rate": 0.000490447105598842,
      "loss": 2.261,
      "step": 64760
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.674522876739502,
      "learning_rate": 0.0004904439449795128,
      "loss": 1.8755,
      "step": 64761
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.9344869256019592,
      "learning_rate": 0.0004904407843247766,
      "loss": 2.2865,
      "step": 64762
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.0503113269805908,
      "learning_rate": 0.0004904376236346337,
      "loss": 2.4405,
      "step": 64763
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7331517338752747,
      "learning_rate": 0.000490434462909085,
      "loss": 2.4083,
      "step": 64764
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7729535102844238,
      "learning_rate": 0.0004904313021481308,
      "loss": 2.3011,
      "step": 64765
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.2885136604309082,
      "learning_rate": 0.0004904281413517719,
      "loss": 2.121,
      "step": 64766
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7946347594261169,
      "learning_rate": 0.0004904249805200088,
      "loss": 2.1911,
      "step": 64767
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7029721140861511,
      "learning_rate": 0.000490421819652842,
      "loss": 2.2892,
      "step": 64768
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.057464361190796,
      "learning_rate": 0.0004904186587502723,
      "loss": 2.0947,
      "step": 64769
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.8330770134925842,
      "learning_rate": 0.0004904154978123001,
      "loss": 1.9231,
      "step": 64770
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7214277982711792,
      "learning_rate": 0.0004904123368389262,
      "loss": 2.1865,
      "step": 64771
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.9159361124038696,
      "learning_rate": 0.0004904091758301509,
      "loss": 2.1897,
      "step": 64772
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.8809812068939209,
      "learning_rate": 0.0004904060147859749,
      "loss": 2.2685,
      "step": 64773
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.6962738633155823,
      "learning_rate": 0.000490402853706399,
      "loss": 2.1894,
      "step": 64774
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7050828337669373,
      "learning_rate": 0.0004903996925914235,
      "loss": 2.1204,
      "step": 64775
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.8418706059455872,
      "learning_rate": 0.0004903965314410492,
      "loss": 2.1284,
      "step": 64776
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.8073427677154541,
      "learning_rate": 0.0004903933702552765,
      "loss": 2.1935,
      "step": 64777
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7062572836875916,
      "learning_rate": 0.0004903902090341062,
      "loss": 2.0672,
      "step": 64778
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.8475175499916077,
      "learning_rate": 0.0004903870477775387,
      "loss": 2.2883,
      "step": 64779
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.6948290467262268,
      "learning_rate": 0.0004903838864855746,
      "loss": 2.3941,
      "step": 64780
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.8162907958030701,
      "learning_rate": 0.0004903807251582146,
      "loss": 2.171,
      "step": 64781
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7310075163841248,
      "learning_rate": 0.0004903775637954592,
      "loss": 2.2141,
      "step": 64782
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.757770299911499,
      "learning_rate": 0.000490374402397309,
      "loss": 2.1983,
      "step": 64783
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.703679621219635,
      "learning_rate": 0.0004903712409637646,
      "loss": 2.3083,
      "step": 64784
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.8462255597114563,
      "learning_rate": 0.0004903680794948267,
      "loss": 2.2877,
      "step": 64785
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7370457053184509,
      "learning_rate": 0.0004903649179904957,
      "loss": 2.2504,
      "step": 64786
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.8005958199501038,
      "learning_rate": 0.0004903617564507722,
      "loss": 2.4003,
      "step": 64787
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.6694747805595398,
      "learning_rate": 0.0004903585948756569,
      "loss": 2.0504,
      "step": 64788
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7021158337593079,
      "learning_rate": 0.0004903554332651505,
      "loss": 2.4733,
      "step": 64789
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7596986293792725,
      "learning_rate": 0.0004903522716192532,
      "loss": 2.3205,
      "step": 64790
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7222381234169006,
      "learning_rate": 0.0004903491099379659,
      "loss": 2.2059,
      "step": 64791
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.6708552241325378,
      "learning_rate": 0.0004903459482212891,
      "loss": 2.2076,
      "step": 64792
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7261527180671692,
      "learning_rate": 0.0004903427864692236,
      "loss": 2.1261,
      "step": 64793
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.6794202923774719,
      "learning_rate": 0.0004903396246817695,
      "loss": 2.331,
      "step": 64794
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.6912302374839783,
      "learning_rate": 0.0004903364628589279,
      "loss": 2.1022,
      "step": 64795
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.6995736956596375,
      "learning_rate": 0.0004903333010006989,
      "loss": 2.2485,
      "step": 64796
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.8074461221694946,
      "learning_rate": 0.0004903301391070836,
      "loss": 2.1959,
      "step": 64797
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7844504714012146,
      "learning_rate": 0.0004903269771780822,
      "loss": 2.3289,
      "step": 64798
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7011060118675232,
      "learning_rate": 0.0004903238152136955,
      "loss": 2.4948,
      "step": 64799
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.8061003088951111,
      "learning_rate": 0.0004903206532139239,
      "loss": 1.9253,
      "step": 64800
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.05477774143219,
      "learning_rate": 0.0004903174911787681,
      "loss": 2.218,
      "step": 64801
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7834187150001526,
      "learning_rate": 0.0004903143291082288,
      "loss": 2.331,
      "step": 64802
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7037604451179504,
      "learning_rate": 0.0004903111670023064,
      "loss": 2.3246,
      "step": 64803
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.767132580280304,
      "learning_rate": 0.0004903080048610015,
      "loss": 2.2552,
      "step": 64804
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7098579406738281,
      "learning_rate": 0.0004903048426843148,
      "loss": 2.1674,
      "step": 64805
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7312057614326477,
      "learning_rate": 0.0004903016804722469,
      "loss": 2.295,
      "step": 64806
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7633805871009827,
      "learning_rate": 0.0004902985182247983,
      "loss": 2.2369,
      "step": 64807
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7035729289054871,
      "learning_rate": 0.0004902953559419696,
      "loss": 2.2019,
      "step": 64808
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7600232362747192,
      "learning_rate": 0.0004902921936237614,
      "loss": 2.2913,
      "step": 64809
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.8125172257423401,
      "learning_rate": 0.0004902890312701743,
      "loss": 2.0605,
      "step": 64810
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.766329824924469,
      "learning_rate": 0.0004902858688812088,
      "loss": 2.2464,
      "step": 64811
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.761957585811615,
      "learning_rate": 0.0004902827064568656,
      "loss": 2.1896,
      "step": 64812
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.712354838848114,
      "learning_rate": 0.0004902795439971453,
      "loss": 2.3222,
      "step": 64813
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.8948017954826355,
      "learning_rate": 0.0004902763815020485,
      "loss": 2.4559,
      "step": 64814
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7278416156768799,
      "learning_rate": 0.0004902732189715756,
      "loss": 2.1807,
      "step": 64815
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.8100456595420837,
      "learning_rate": 0.0004902700564057274,
      "loss": 2.1972,
      "step": 64816
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.78070467710495,
      "learning_rate": 0.0004902668938045044,
      "loss": 2.2091,
      "step": 64817
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.8336331844329834,
      "learning_rate": 0.000490263731167907,
      "loss": 2.1946,
      "step": 64818
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.763935923576355,
      "learning_rate": 0.0004902605684959363,
      "loss": 2.1727,
      "step": 64819
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.8797396421432495,
      "learning_rate": 0.0004902574057885923,
      "loss": 2.1625,
      "step": 64820
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.6961546540260315,
      "learning_rate": 0.0004902542430458761,
      "loss": 2.1056,
      "step": 64821
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.744837760925293,
      "learning_rate": 0.0004902510802677879,
      "loss": 2.2199,
      "step": 64822
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.8255090117454529,
      "learning_rate": 0.0004902479174543285,
      "loss": 2.2892,
      "step": 64823
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.6975311636924744,
      "learning_rate": 0.0004902447546054983,
      "loss": 2.2565,
      "step": 64824
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7861153483390808,
      "learning_rate": 0.0004902415917212981,
      "loss": 2.1007,
      "step": 64825
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7585186958312988,
      "learning_rate": 0.0004902384288017283,
      "loss": 2.2567,
      "step": 64826
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7444173097610474,
      "learning_rate": 0.0004902352658467898,
      "loss": 2.1648,
      "step": 64827
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7613142728805542,
      "learning_rate": 0.0004902321028564828,
      "loss": 2.2205,
      "step": 64828
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.8049099445343018,
      "learning_rate": 0.0004902289398308081,
      "loss": 2.3474,
      "step": 64829
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.790468692779541,
      "learning_rate": 0.0004902257767697662,
      "loss": 2.2219,
      "step": 64830
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.8281797170639038,
      "learning_rate": 0.0004902226136733578,
      "loss": 2.6223,
      "step": 64831
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.8339315056800842,
      "learning_rate": 0.0004902194505415834,
      "loss": 2.3049,
      "step": 64832
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7722198367118835,
      "learning_rate": 0.0004902162873744436,
      "loss": 2.2973,
      "step": 64833
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.819743812084198,
      "learning_rate": 0.000490213124171939,
      "loss": 2.1513,
      "step": 64834
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.909296989440918,
      "learning_rate": 0.0004902099609340703,
      "loss": 2.1912,
      "step": 64835
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7403753399848938,
      "learning_rate": 0.0004902067976608379,
      "loss": 2.3235,
      "step": 64836
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.892181396484375,
      "learning_rate": 0.0004902036343522424,
      "loss": 2.2019,
      "step": 64837
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7384845018386841,
      "learning_rate": 0.0004902004710082846,
      "loss": 2.0595,
      "step": 64838
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7851549386978149,
      "learning_rate": 0.0004901973076289648,
      "loss": 2.0611,
      "step": 64839
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7204136252403259,
      "learning_rate": 0.0004901941442142837,
      "loss": 2.3561,
      "step": 64840
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.8433884978294373,
      "learning_rate": 0.0004901909807642421,
      "loss": 1.9806,
      "step": 64841
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7134346961975098,
      "learning_rate": 0.0004901878172788403,
      "loss": 2.0667,
      "step": 64842
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.6973276138305664,
      "learning_rate": 0.000490184653758079,
      "loss": 2.2438,
      "step": 64843
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.6725288033485413,
      "learning_rate": 0.0004901814902019588,
      "loss": 2.1001,
      "step": 64844
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7481786608695984,
      "learning_rate": 0.0004901783266104803,
      "loss": 2.2426,
      "step": 64845
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.731360137462616,
      "learning_rate": 0.000490175162983644,
      "loss": 2.052,
      "step": 64846
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7255911827087402,
      "learning_rate": 0.0004901719993214505,
      "loss": 2.4172,
      "step": 64847
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.8446094393730164,
      "learning_rate": 0.0004901688356239006,
      "loss": 2.2897,
      "step": 64848
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7906606197357178,
      "learning_rate": 0.0004901656718909945,
      "loss": 2.4774,
      "step": 64849
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7855886816978455,
      "learning_rate": 0.0004901625081227333,
      "loss": 2.2921,
      "step": 64850
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.8592752814292908,
      "learning_rate": 0.000490159344319117,
      "loss": 2.3442,
      "step": 64851
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7500815391540527,
      "learning_rate": 0.0004901561804801466,
      "loss": 2.4777,
      "step": 64852
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7630661129951477,
      "learning_rate": 0.0004901530166058227,
      "loss": 2.2349,
      "step": 64853
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7087792754173279,
      "learning_rate": 0.0004901498526961457,
      "loss": 2.264,
      "step": 64854
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7535885572433472,
      "learning_rate": 0.0004901466887511162,
      "loss": 2.2196,
      "step": 64855
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7470241785049438,
      "learning_rate": 0.0004901435247707349,
      "loss": 2.3337,
      "step": 64856
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7261781096458435,
      "learning_rate": 0.0004901403607550022,
      "loss": 2.0004,
      "step": 64857
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.8106966614723206,
      "learning_rate": 0.000490137196703919,
      "loss": 2.253,
      "step": 64858
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.8212130665779114,
      "learning_rate": 0.0004901340326174856,
      "loss": 2.4172,
      "step": 64859
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7280015349388123,
      "learning_rate": 0.0004901308684957027,
      "loss": 2.1486,
      "step": 64860
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7961941361427307,
      "learning_rate": 0.0004901277043385708,
      "loss": 2.0566,
      "step": 64861
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.8562465906143188,
      "learning_rate": 0.0004901245401460907,
      "loss": 2.188,
      "step": 64862
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7147522568702698,
      "learning_rate": 0.0004901213759182628,
      "loss": 2.214,
      "step": 64863
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.9617698788642883,
      "learning_rate": 0.0004901182116550878,
      "loss": 2.1942,
      "step": 64864
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.8298777937889099,
      "learning_rate": 0.0004901150473565662,
      "loss": 2.1096,
      "step": 64865
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.802225649356842,
      "learning_rate": 0.0004901118830226986,
      "loss": 2.3644,
      "step": 64866
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7730602622032166,
      "learning_rate": 0.0004901087186534856,
      "loss": 2.4182,
      "step": 64867
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7971688508987427,
      "learning_rate": 0.0004901055542489279,
      "loss": 2.2941,
      "step": 64868
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.0457305908203125,
      "learning_rate": 0.000490102389809026,
      "loss": 2.2857,
      "step": 64869
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.76102215051651,
      "learning_rate": 0.0004900992253337803,
      "loss": 2.1126,
      "step": 64870
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7854657769203186,
      "learning_rate": 0.0004900960608231917,
      "loss": 2.0002,
      "step": 64871
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.8486220836639404,
      "learning_rate": 0.0004900928962772607,
      "loss": 2.0639,
      "step": 64872
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.9485107660293579,
      "learning_rate": 0.0004900897316959876,
      "loss": 2.3393,
      "step": 64873
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7855933308601379,
      "learning_rate": 0.0004900865670793735,
      "loss": 2.1889,
      "step": 64874
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.9604092240333557,
      "learning_rate": 0.0004900834024274186,
      "loss": 2.1632,
      "step": 64875
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.2491121292114258,
      "learning_rate": 0.0004900802377401236,
      "loss": 2.2091,
      "step": 64876
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7541153430938721,
      "learning_rate": 0.0004900770730174892,
      "loss": 2.1432,
      "step": 64877
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7700174450874329,
      "learning_rate": 0.0004900739082595157,
      "loss": 2.1997,
      "step": 64878
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.8600162267684937,
      "learning_rate": 0.0004900707434662041,
      "loss": 2.3086,
      "step": 64879
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.9080355763435364,
      "learning_rate": 0.0004900675786375546,
      "loss": 2.2572,
      "step": 64880
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7825908064842224,
      "learning_rate": 0.0004900644137735679,
      "loss": 2.5145,
      "step": 64881
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7428173422813416,
      "learning_rate": 0.0004900612488742447,
      "loss": 2.534,
      "step": 64882
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.8319343328475952,
      "learning_rate": 0.0004900580839395857,
      "loss": 2.2387,
      "step": 64883
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.0329792499542236,
      "learning_rate": 0.0004900549189695911,
      "loss": 2.3049,
      "step": 64884
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.8186681270599365,
      "learning_rate": 0.0004900517539642618,
      "loss": 2.1928,
      "step": 64885
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7722106575965881,
      "learning_rate": 0.0004900485889235982,
      "loss": 2.4019,
      "step": 64886
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.1078603267669678,
      "learning_rate": 0.0004900454238476012,
      "loss": 2.1967,
      "step": 64887
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.8159162998199463,
      "learning_rate": 0.0004900422587362709,
      "loss": 2.4086,
      "step": 64888
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7529067397117615,
      "learning_rate": 0.0004900390935896082,
      "loss": 1.9266,
      "step": 64889
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.8798949122428894,
      "learning_rate": 0.0004900359284076138,
      "loss": 2.3225,
      "step": 64890
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.8883445262908936,
      "learning_rate": 0.000490032763190288,
      "loss": 2.3307,
      "step": 64891
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.808262288570404,
      "learning_rate": 0.0004900295979376316,
      "loss": 2.3766,
      "step": 64892
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7786534428596497,
      "learning_rate": 0.0004900264326496451,
      "loss": 2.2953,
      "step": 64893
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.8587168455123901,
      "learning_rate": 0.0004900232673263291,
      "loss": 2.2246,
      "step": 64894
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.8865452408790588,
      "learning_rate": 0.0004900201019676842,
      "loss": 2.2549,
      "step": 64895
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.8546381592750549,
      "learning_rate": 0.0004900169365737109,
      "loss": 2.2891,
      "step": 64896
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7533694505691528,
      "learning_rate": 0.0004900137711444098,
      "loss": 2.2048,
      "step": 64897
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.9398859143257141,
      "learning_rate": 0.0004900106056797816,
      "loss": 2.2113,
      "step": 64898
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7579869627952576,
      "learning_rate": 0.000490007440179827,
      "loss": 2.2855,
      "step": 64899
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7301608920097351,
      "learning_rate": 0.0004900042746445463,
      "loss": 2.3134,
      "step": 64900
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7728081941604614,
      "learning_rate": 0.0004900011090739402,
      "loss": 2.1537,
      "step": 64901
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7718649506568909,
      "learning_rate": 0.0004899979434680095,
      "loss": 2.5337,
      "step": 64902
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.8613393902778625,
      "learning_rate": 0.0004899947778267544,
      "loss": 2.0504,
      "step": 64903
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.6903605461120605,
      "learning_rate": 0.0004899916121501758,
      "loss": 1.9026,
      "step": 64904
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.8199750781059265,
      "learning_rate": 0.0004899884464382741,
      "loss": 2.2994,
      "step": 64905
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7552294135093689,
      "learning_rate": 0.0004899852806910499,
      "loss": 2.2305,
      "step": 64906
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7546705603599548,
      "learning_rate": 0.000489982114908504,
      "loss": 2.3368,
      "step": 64907
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.73948734998703,
      "learning_rate": 0.0004899789490906367,
      "loss": 2.2599,
      "step": 64908
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.8698349595069885,
      "learning_rate": 0.0004899757832374488,
      "loss": 2.1347,
      "step": 64909
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7253865599632263,
      "learning_rate": 0.0004899726173489408,
      "loss": 2.2364,
      "step": 64910
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.8369554281234741,
      "learning_rate": 0.0004899694514251133,
      "loss": 2.1671,
      "step": 64911
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7787025570869446,
      "learning_rate": 0.0004899662854659668,
      "loss": 2.2177,
      "step": 64912
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7469533681869507,
      "learning_rate": 0.0004899631194715022,
      "loss": 2.1441,
      "step": 64913
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7956902980804443,
      "learning_rate": 0.0004899599534417198,
      "loss": 2.1194,
      "step": 64914
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7395001649856567,
      "learning_rate": 0.0004899567873766201,
      "loss": 2.1452,
      "step": 64915
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7996668815612793,
      "learning_rate": 0.000489953621276204,
      "loss": 2.1724,
      "step": 64916
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.6851925849914551,
      "learning_rate": 0.0004899504551404718,
      "loss": 2.1253,
      "step": 64917
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7614766955375671,
      "learning_rate": 0.0004899472889694244,
      "loss": 2.3203,
      "step": 64918
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.8207597136497498,
      "learning_rate": 0.0004899441227630621,
      "loss": 2.3295,
      "step": 64919
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.8395949006080627,
      "learning_rate": 0.0004899409565213858,
      "loss": 2.355,
      "step": 64920
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7450070977210999,
      "learning_rate": 0.0004899377902443956,
      "loss": 2.1779,
      "step": 64921
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7225109934806824,
      "learning_rate": 0.0004899346239320925,
      "loss": 2.1493,
      "step": 64922
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.811915397644043,
      "learning_rate": 0.000489931457584477,
      "loss": 2.0588,
      "step": 64923
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7194372415542603,
      "learning_rate": 0.0004899282912015496,
      "loss": 2.1782,
      "step": 64924
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.8194886445999146,
      "learning_rate": 0.000489925124783311,
      "loss": 2.4553,
      "step": 64925
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7361140847206116,
      "learning_rate": 0.0004899219583297617,
      "loss": 2.3929,
      "step": 64926
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.743017852306366,
      "learning_rate": 0.0004899187918409024,
      "loss": 2.1918,
      "step": 64927
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7531678080558777,
      "learning_rate": 0.0004899156253167336,
      "loss": 2.2611,
      "step": 64928
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7407927513122559,
      "learning_rate": 0.0004899124587572557,
      "loss": 2.1382,
      "step": 64929
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.90810626745224,
      "learning_rate": 0.0004899092921624697,
      "loss": 2.089,
      "step": 64930
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.9385786056518555,
      "learning_rate": 0.0004899061255323759,
      "loss": 2.2258,
      "step": 64931
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.6954055428504944,
      "learning_rate": 0.000489902958866975,
      "loss": 2.2903,
      "step": 64932
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.0285369157791138,
      "learning_rate": 0.0004898997921662675,
      "loss": 2.3197,
      "step": 64933
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.9442790746688843,
      "learning_rate": 0.000489896625430254,
      "loss": 2.2975,
      "step": 64934
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7511733770370483,
      "learning_rate": 0.0004898934586589353,
      "loss": 2.2091,
      "step": 64935
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.9154863357543945,
      "learning_rate": 0.0004898902918523117,
      "loss": 2.1084,
      "step": 64936
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7895340323448181,
      "learning_rate": 0.0004898871250103839,
      "loss": 2.0208,
      "step": 64937
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.8263393640518188,
      "learning_rate": 0.0004898839581331527,
      "loss": 2.3225,
      "step": 64938
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7547125816345215,
      "learning_rate": 0.0004898807912206183,
      "loss": 2.2451,
      "step": 64939
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7500715851783752,
      "learning_rate": 0.0004898776242727814,
      "loss": 2.3487,
      "step": 64940
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.0173451900482178,
      "learning_rate": 0.0004898744572896428,
      "loss": 2.2095,
      "step": 64941
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7008169889450073,
      "learning_rate": 0.000489871290271203,
      "loss": 2.3277,
      "step": 64942
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.6902700066566467,
      "learning_rate": 0.0004898681232174623,
      "loss": 2.3114,
      "step": 64943
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.8827340602874756,
      "learning_rate": 0.0004898649561284218,
      "loss": 2.2694,
      "step": 64944
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7685458660125732,
      "learning_rate": 0.0004898617890040818,
      "loss": 2.2263,
      "step": 64945
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7858980894088745,
      "learning_rate": 0.0004898586218444427,
      "loss": 2.2355,
      "step": 64946
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7628154754638672,
      "learning_rate": 0.0004898554546495055,
      "loss": 2.297,
      "step": 64947
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7282248139381409,
      "learning_rate": 0.0004898522874192706,
      "loss": 2.411,
      "step": 64948
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7520604133605957,
      "learning_rate": 0.0004898491201537384,
      "loss": 2.3193,
      "step": 64949
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7492568492889404,
      "learning_rate": 0.0004898459528529098,
      "loss": 2.4562,
      "step": 64950
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7168416976928711,
      "learning_rate": 0.0004898427855167852,
      "loss": 2.2013,
      "step": 64951
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7793372869491577,
      "learning_rate": 0.0004898396181453652,
      "loss": 1.9386,
      "step": 64952
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7257289290428162,
      "learning_rate": 0.0004898364507386505,
      "loss": 2.1829,
      "step": 64953
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7899519205093384,
      "learning_rate": 0.0004898332832966416,
      "loss": 2.0965,
      "step": 64954
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7534818053245544,
      "learning_rate": 0.0004898301158193391,
      "loss": 2.2738,
      "step": 64955
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7698501348495483,
      "learning_rate": 0.0004898269483067436,
      "loss": 2.2311,
      "step": 64956
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.6953043937683105,
      "learning_rate": 0.0004898237807588557,
      "loss": 2.2109,
      "step": 64957
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7284233570098877,
      "learning_rate": 0.000489820613175676,
      "loss": 2.1486,
      "step": 64958
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7589457035064697,
      "learning_rate": 0.0004898174455572051,
      "loss": 2.1061,
      "step": 64959
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7416922450065613,
      "learning_rate": 0.0004898142779034434,
      "loss": 2.2458,
      "step": 64960
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7867914438247681,
      "learning_rate": 0.0004898111102143918,
      "loss": 2.2527,
      "step": 64961
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7996050715446472,
      "learning_rate": 0.0004898079424900507,
      "loss": 2.2833,
      "step": 64962
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7721647620201111,
      "learning_rate": 0.0004898047747304206,
      "loss": 2.3024,
      "step": 64963
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.9402679800987244,
      "learning_rate": 0.0004898016069355024,
      "loss": 2.3015,
      "step": 64964
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7088270783424377,
      "learning_rate": 0.0004897984391052963,
      "loss": 2.0985,
      "step": 64965
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7708314061164856,
      "learning_rate": 0.0004897952712398032,
      "loss": 2.3569,
      "step": 64966
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.8645972609519958,
      "learning_rate": 0.0004897921033390235,
      "loss": 2.3319,
      "step": 64967
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7238401174545288,
      "learning_rate": 0.000489788935402958,
      "loss": 2.2198,
      "step": 64968
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7334926128387451,
      "learning_rate": 0.000489785767431607,
      "loss": 2.338,
      "step": 64969
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.9896153211593628,
      "learning_rate": 0.0004897825994249714,
      "loss": 2.3145,
      "step": 64970
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7909224033355713,
      "learning_rate": 0.0004897794313830515,
      "loss": 2.2117,
      "step": 64971
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.8405562043190002,
      "learning_rate": 0.0004897762633058482,
      "loss": 2.1891,
      "step": 64972
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.8033122420310974,
      "learning_rate": 0.0004897730951933618,
      "loss": 2.1553,
      "step": 64973
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.8279314041137695,
      "learning_rate": 0.0004897699270455929,
      "loss": 2.4948,
      "step": 64974
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.6820397973060608,
      "learning_rate": 0.0004897667588625422,
      "loss": 2.1729,
      "step": 64975
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.8392356038093567,
      "learning_rate": 0.0004897635906442105,
      "loss": 2.1411,
      "step": 64976
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.767526388168335,
      "learning_rate": 0.000489760422390598,
      "loss": 2.0493,
      "step": 64977
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7376514077186584,
      "learning_rate": 0.0004897572541017054,
      "loss": 2.3897,
      "step": 64978
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.6998338103294373,
      "learning_rate": 0.0004897540857775335,
      "loss": 2.2344,
      "step": 64979
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7203473448753357,
      "learning_rate": 0.0004897509174180826,
      "loss": 2.0038,
      "step": 64980
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7544698119163513,
      "learning_rate": 0.0004897477490233534,
      "loss": 1.9676,
      "step": 64981
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7383793592453003,
      "learning_rate": 0.0004897445805933467,
      "loss": 2.1194,
      "step": 64982
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7534334063529968,
      "learning_rate": 0.0004897414121280627,
      "loss": 2.1607,
      "step": 64983
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7969197034835815,
      "learning_rate": 0.0004897382436275024,
      "loss": 2.15,
      "step": 64984
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7768880128860474,
      "learning_rate": 0.000489735075091666,
      "loss": 2.1991,
      "step": 64985
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7218515872955322,
      "learning_rate": 0.0004897319065205542,
      "loss": 2.2593,
      "step": 64986
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.8134980201721191,
      "learning_rate": 0.000489728737914168,
      "loss": 2.4483,
      "step": 64987
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7770019769668579,
      "learning_rate": 0.0004897255692725074,
      "loss": 2.0349,
      "step": 64988
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7937904000282288,
      "learning_rate": 0.0004897224005955733,
      "loss": 2.3004,
      "step": 64989
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7693652510643005,
      "learning_rate": 0.000489719231883366,
      "loss": 2.0308,
      "step": 64990
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7005826830863953,
      "learning_rate": 0.0004897160631358866,
      "loss": 2.0112,
      "step": 64991
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7113214731216431,
      "learning_rate": 0.0004897128943531353,
      "loss": 2.2135,
      "step": 64992
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7679905295372009,
      "learning_rate": 0.0004897097255351128,
      "loss": 2.1053,
      "step": 64993
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.9586140513420105,
      "learning_rate": 0.0004897065566818197,
      "loss": 2.2282,
      "step": 64994
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7191078066825867,
      "learning_rate": 0.0004897033877932565,
      "loss": 2.2262,
      "step": 64995
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.8064554929733276,
      "learning_rate": 0.0004897002188694239,
      "loss": 2.3053,
      "step": 64996
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.8212897777557373,
      "learning_rate": 0.0004896970499103225,
      "loss": 2.2676,
      "step": 64997
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.8009526133537292,
      "learning_rate": 0.0004896938809159527,
      "loss": 2.3193,
      "step": 64998
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7544169425964355,
      "learning_rate": 0.0004896907118863154,
      "loss": 2.3389,
      "step": 64999
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.8853412866592407,
      "learning_rate": 0.0004896875428214109,
      "loss": 2.3184,
      "step": 65000
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.8451301455497742,
      "learning_rate": 0.0004896843737212399,
      "loss": 2.2643,
      "step": 65001
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7680943608283997,
      "learning_rate": 0.0004896812045858031,
      "loss": 2.2116,
      "step": 65002
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7276393175125122,
      "learning_rate": 0.000489678035415101,
      "loss": 2.0661,
      "step": 65003
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.346987009048462,
      "learning_rate": 0.000489674866209134,
      "loss": 2.2658,
      "step": 65004
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.714992344379425,
      "learning_rate": 0.000489671696967903,
      "loss": 2.1283,
      "step": 65005
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7376956939697266,
      "learning_rate": 0.0004896685276914083,
      "loss": 2.2294,
      "step": 65006
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7096270322799683,
      "learning_rate": 0.0004896653583796507,
      "loss": 2.2102,
      "step": 65007
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.763740599155426,
      "learning_rate": 0.0004896621890326309,
      "loss": 2.2922,
      "step": 65008
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.867518961429596,
      "learning_rate": 0.0004896590196503492,
      "loss": 2.2118,
      "step": 65009
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7520563006401062,
      "learning_rate": 0.0004896558502328062,
      "loss": 2.1202,
      "step": 65010
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7048587203025818,
      "learning_rate": 0.0004896526807800028,
      "loss": 2.2287,
      "step": 65011
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7725991010665894,
      "learning_rate": 0.0004896495112919392,
      "loss": 2.3278,
      "step": 65012
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.8173186779022217,
      "learning_rate": 0.0004896463417686164,
      "loss": 2.2237,
      "step": 65013
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7753564715385437,
      "learning_rate": 0.0004896431722100345,
      "loss": 2.2147,
      "step": 65014
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.6529360413551331,
      "learning_rate": 0.0004896400026161945,
      "loss": 2.1351,
      "step": 65015
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7482077479362488,
      "learning_rate": 0.0004896368329870968,
      "loss": 2.2748,
      "step": 65016
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7681512832641602,
      "learning_rate": 0.0004896336633227422,
      "loss": 2.1813,
      "step": 65017
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.8242051601409912,
      "learning_rate": 0.000489630493623131,
      "loss": 2.2874,
      "step": 65018
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7243790030479431,
      "learning_rate": 0.0004896273238882638,
      "loss": 2.4081,
      "step": 65019
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7394943237304688,
      "learning_rate": 0.0004896241541181415,
      "loss": 2.0992,
      "step": 65020
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7161399722099304,
      "learning_rate": 0.0004896209843127643,
      "loss": 2.4162,
      "step": 65021
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7073793411254883,
      "learning_rate": 0.000489617814472133,
      "loss": 2.2579,
      "step": 65022
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7698615789413452,
      "learning_rate": 0.0004896146445962483,
      "loss": 2.2632,
      "step": 65023
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7401276230812073,
      "learning_rate": 0.0004896114746851106,
      "loss": 2.1596,
      "step": 65024
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7322378754615784,
      "learning_rate": 0.0004896083047387205,
      "loss": 2.2108,
      "step": 65025
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7723903059959412,
      "learning_rate": 0.0004896051347570787,
      "loss": 2.4856,
      "step": 65026
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.6924782395362854,
      "learning_rate": 0.0004896019647401857,
      "loss": 2.1153,
      "step": 65027
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7034749388694763,
      "learning_rate": 0.0004895987946880421,
      "loss": 2.1662,
      "step": 65028
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.653872013092041,
      "learning_rate": 0.0004895956246006484,
      "loss": 2.0375,
      "step": 65029
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.8181066513061523,
      "learning_rate": 0.0004895924544780055,
      "loss": 2.5131,
      "step": 65030
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7495590448379517,
      "learning_rate": 0.0004895892843201138,
      "loss": 2.213,
      "step": 65031
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7294031381607056,
      "learning_rate": 0.0004895861141269737,
      "loss": 2.2949,
      "step": 65032
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7166375517845154,
      "learning_rate": 0.0004895829438985862,
      "loss": 1.959,
      "step": 65033
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.728754460811615,
      "learning_rate": 0.0004895797736349513,
      "loss": 2.1182,
      "step": 65034
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7177516222000122,
      "learning_rate": 0.0004895766033360702,
      "loss": 2.1649,
      "step": 65035
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.6765025854110718,
      "learning_rate": 0.0004895734330019432,
      "loss": 2.0916,
      "step": 65036
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7775958776473999,
      "learning_rate": 0.0004895702626325708,
      "loss": 2.2836,
      "step": 65037
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.682769238948822,
      "learning_rate": 0.0004895670922279538,
      "loss": 2.2463,
      "step": 65038
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.6681132316589355,
      "learning_rate": 0.0004895639217880926,
      "loss": 2.1589,
      "step": 65039
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.835109531879425,
      "learning_rate": 0.0004895607513129881,
      "loss": 2.4765,
      "step": 65040
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7403043508529663,
      "learning_rate": 0.0004895575808026405,
      "loss": 2.3117,
      "step": 65041
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7777316570281982,
      "learning_rate": 0.0004895544102570507,
      "loss": 2.1298,
      "step": 65042
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7608276009559631,
      "learning_rate": 0.000489551239676219,
      "loss": 2.2835,
      "step": 65043
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.8843399286270142,
      "learning_rate": 0.0004895480690601463,
      "loss": 2.1253,
      "step": 65044
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7601510286331177,
      "learning_rate": 0.000489544898408833,
      "loss": 2.1568,
      "step": 65045
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7204518914222717,
      "learning_rate": 0.0004895417277222796,
      "loss": 2.1488,
      "step": 65046
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.8083756566047668,
      "learning_rate": 0.0004895385570004869,
      "loss": 2.3483,
      "step": 65047
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7845486998558044,
      "learning_rate": 0.0004895353862434554,
      "loss": 2.2829,
      "step": 65048
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7619678378105164,
      "learning_rate": 0.0004895322154511857,
      "loss": 2.0315,
      "step": 65049
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7696269154548645,
      "learning_rate": 0.0004895290446236784,
      "loss": 2.1598,
      "step": 65050
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7882634997367859,
      "learning_rate": 0.0004895258737609341,
      "loss": 2.2604,
      "step": 65051
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7328293323516846,
      "learning_rate": 0.0004895227028629533,
      "loss": 2.0564,
      "step": 65052
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.9464696645736694,
      "learning_rate": 0.0004895195319297367,
      "loss": 2.2471,
      "step": 65053
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.8547278046607971,
      "learning_rate": 0.0004895163609612848,
      "loss": 2.3037,
      "step": 65054
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7702802419662476,
      "learning_rate": 0.0004895131899575983,
      "loss": 2.4371,
      "step": 65055
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7056005597114563,
      "learning_rate": 0.0004895100189186777,
      "loss": 2.1159,
      "step": 65056
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.8951748013496399,
      "learning_rate": 0.0004895068478445236,
      "loss": 2.2776,
      "step": 65057
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7034212946891785,
      "learning_rate": 0.0004895036767351365,
      "loss": 2.3307,
      "step": 65058
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.9132809042930603,
      "learning_rate": 0.0004895005055905171,
      "loss": 2.3432,
      "step": 65059
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.8657461404800415,
      "learning_rate": 0.0004894973344106661,
      "loss": 2.0956,
      "step": 65060
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.8219035863876343,
      "learning_rate": 0.0004894941631955839,
      "loss": 2.2033,
      "step": 65061
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7476688027381897,
      "learning_rate": 0.0004894909919452712,
      "loss": 2.3825,
      "step": 65062
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.0152614116668701,
      "learning_rate": 0.0004894878206597285,
      "loss": 2.0264,
      "step": 65063
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.8904306292533875,
      "learning_rate": 0.0004894846493389564,
      "loss": 2.2447,
      "step": 65064
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7753829956054688,
      "learning_rate": 0.0004894814779829557,
      "loss": 2.3172,
      "step": 65065
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7817555665969849,
      "learning_rate": 0.0004894783065917266,
      "loss": 2.1022,
      "step": 65066
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.002889633178711,
      "learning_rate": 0.00048947513516527,
      "loss": 2.1658,
      "step": 65067
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.8963099718093872,
      "learning_rate": 0.0004894719637035864,
      "loss": 2.3465,
      "step": 65068
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.834571123123169,
      "learning_rate": 0.0004894687922066764,
      "loss": 2.2764,
      "step": 65069
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.8093063235282898,
      "learning_rate": 0.0004894656206745406,
      "loss": 2.0058,
      "step": 65070
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.0173248052597046,
      "learning_rate": 0.0004894624491071795,
      "loss": 2.49,
      "step": 65071
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7550143599510193,
      "learning_rate": 0.0004894592775045939,
      "loss": 2.2612,
      "step": 65072
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7916308641433716,
      "learning_rate": 0.000489456105866784,
      "loss": 2.1816,
      "step": 65073
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.0415174961090088,
      "learning_rate": 0.0004894529341937508,
      "loss": 2.1775,
      "step": 65074
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7558811902999878,
      "learning_rate": 0.0004894497624854946,
      "loss": 2.3494,
      "step": 65075
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7645986080169678,
      "learning_rate": 0.0004894465907420163,
      "loss": 2.0641,
      "step": 65076
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7207140922546387,
      "learning_rate": 0.0004894434189633162,
      "loss": 2.2672,
      "step": 65077
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7262853980064392,
      "learning_rate": 0.000489440247149395,
      "loss": 2.4137,
      "step": 65078
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.752303421497345,
      "learning_rate": 0.0004894370753002532,
      "loss": 2.4054,
      "step": 65079
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7552024722099304,
      "learning_rate": 0.0004894339034158915,
      "loss": 2.4547,
      "step": 65080
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7450678944587708,
      "learning_rate": 0.0004894307314963106,
      "loss": 2.4529,
      "step": 65081
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7704826593399048,
      "learning_rate": 0.0004894275595415109,
      "loss": 2.1635,
      "step": 65082
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.8394269943237305,
      "learning_rate": 0.000489424387551493,
      "loss": 2.2318,
      "step": 65083
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.8063734769821167,
      "learning_rate": 0.0004894212155262574,
      "loss": 2.1889,
      "step": 65084
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7788170576095581,
      "learning_rate": 0.000489418043465805,
      "loss": 2.3071,
      "step": 65085
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7771084904670715,
      "learning_rate": 0.0004894148713701362,
      "loss": 2.0111,
      "step": 65086
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.8415326476097107,
      "learning_rate": 0.0004894116992392515,
      "loss": 2.1718,
      "step": 65087
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.8320112824440002,
      "learning_rate": 0.0004894085270731517,
      "loss": 2.1993,
      "step": 65088
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7229335904121399,
      "learning_rate": 0.0004894053548718372,
      "loss": 2.2378,
      "step": 65089
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7889816164970398,
      "learning_rate": 0.0004894021826353087,
      "loss": 2.1542,
      "step": 65090
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.8850029706954956,
      "learning_rate": 0.0004893990103635667,
      "loss": 2.2994,
      "step": 65091
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.8217755556106567,
      "learning_rate": 0.000489395838056612,
      "loss": 2.3477,
      "step": 65092
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.0411837100982666,
      "learning_rate": 0.0004893926657144449,
      "loss": 2.3568,
      "step": 65093
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.9715757369995117,
      "learning_rate": 0.0004893894933370662,
      "loss": 2.1589,
      "step": 65094
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.8412352204322815,
      "learning_rate": 0.0004893863209244764,
      "loss": 2.2477,
      "step": 65095
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7501810789108276,
      "learning_rate": 0.0004893831484766761,
      "loss": 1.9112,
      "step": 65096
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.9175946116447449,
      "learning_rate": 0.000489379975993666,
      "loss": 2.0619,
      "step": 65097
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7928262948989868,
      "learning_rate": 0.0004893768034754462,
      "loss": 2.1162,
      "step": 65098
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.8044095635414124,
      "learning_rate": 0.0004893736309220181,
      "loss": 2.2426,
      "step": 65099
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.8905045390129089,
      "learning_rate": 0.0004893704583333817,
      "loss": 2.2761,
      "step": 65100
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7753925323486328,
      "learning_rate": 0.0004893672857095378,
      "loss": 2.2176,
      "step": 65101
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7886735200881958,
      "learning_rate": 0.000489364113050487,
      "loss": 2.2383,
      "step": 65102
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.711819589138031,
      "learning_rate": 0.0004893609403562297,
      "loss": 2.2589,
      "step": 65103
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.8165667057037354,
      "learning_rate": 0.0004893577676267667,
      "loss": 2.3027,
      "step": 65104
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.9293525815010071,
      "learning_rate": 0.0004893545948620985,
      "loss": 2.1461,
      "step": 65105
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.8093376755714417,
      "learning_rate": 0.0004893514220622258,
      "loss": 2.27,
      "step": 65106
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.708439826965332,
      "learning_rate": 0.000489348249227149,
      "loss": 2.1629,
      "step": 65107
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.0019346475601196,
      "learning_rate": 0.0004893450763568687,
      "loss": 2.1049,
      "step": 65108
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.9993443489074707,
      "learning_rate": 0.0004893419034513858,
      "loss": 2.2572,
      "step": 65109
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7116377353668213,
      "learning_rate": 0.0004893387305107005,
      "loss": 2.3245,
      "step": 65110
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.6916345953941345,
      "learning_rate": 0.0004893355575348136,
      "loss": 2.0161,
      "step": 65111
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.9970307946205139,
      "learning_rate": 0.0004893323845237257,
      "loss": 2.3431,
      "step": 65112
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.9266861081123352,
      "learning_rate": 0.0004893292114774373,
      "loss": 2.2079,
      "step": 65113
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.8165210485458374,
      "learning_rate": 0.000489326038395949,
      "loss": 2.3854,
      "step": 65114
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.867039144039154,
      "learning_rate": 0.0004893228652792614,
      "loss": 2.118,
      "step": 65115
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.8189476132392883,
      "learning_rate": 0.000489319692127375,
      "loss": 2.3191,
      "step": 65116
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.8599869012832642,
      "learning_rate": 0.0004893165189402908,
      "loss": 2.4405,
      "step": 65117
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7253897786140442,
      "learning_rate": 0.0004893133457180089,
      "loss": 2.0754,
      "step": 65118
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7405178546905518,
      "learning_rate": 0.0004893101724605301,
      "loss": 2.1255,
      "step": 65119
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.8278772234916687,
      "learning_rate": 0.0004893069991678548,
      "loss": 2.2195,
      "step": 65120
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7521316409111023,
      "learning_rate": 0.000489303825839984,
      "loss": 2.2567,
      "step": 65121
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7256733775138855,
      "learning_rate": 0.000489300652476918,
      "loss": 2.2434,
      "step": 65122
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.8170486688613892,
      "learning_rate": 0.0004892974790786573,
      "loss": 2.1207,
      "step": 65123
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7963859438896179,
      "learning_rate": 0.0004892943056452028,
      "loss": 2.3141,
      "step": 65124
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.771551251411438,
      "learning_rate": 0.0004892911321765547,
      "loss": 2.231,
      "step": 65125
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7720476388931274,
      "learning_rate": 0.000489287958672714,
      "loss": 2.3404,
      "step": 65126
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.6949098110198975,
      "learning_rate": 0.000489284785133681,
      "loss": 2.1568,
      "step": 65127
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7834470272064209,
      "learning_rate": 0.0004892816115594564,
      "loss": 2.4935,
      "step": 65128
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7301411032676697,
      "learning_rate": 0.0004892784379500408,
      "loss": 2.1301,
      "step": 65129
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.9003135561943054,
      "learning_rate": 0.0004892752643054348,
      "loss": 2.3686,
      "step": 65130
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.8631339073181152,
      "learning_rate": 0.000489272090625639,
      "loss": 2.3984,
      "step": 65131
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.8222081661224365,
      "learning_rate": 0.0004892689169106538,
      "loss": 2.3028,
      "step": 65132
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.9688971638679504,
      "learning_rate": 0.00048926574316048,
      "loss": 2.1379,
      "step": 65133
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.798335075378418,
      "learning_rate": 0.000489262569375118,
      "loss": 2.1253,
      "step": 65134
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7776582837104797,
      "learning_rate": 0.0004892593955545687,
      "loss": 2.0326,
      "step": 65135
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.8176950812339783,
      "learning_rate": 0.0004892562216988324,
      "loss": 2.2455,
      "step": 65136
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.9969401955604553,
      "learning_rate": 0.0004892530478079099,
      "loss": 1.9654,
      "step": 65137
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7571834325790405,
      "learning_rate": 0.0004892498738818015,
      "loss": 2.0891,
      "step": 65138
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7536554932594299,
      "learning_rate": 0.0004892466999205082,
      "loss": 2.2866,
      "step": 65139
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.8734561204910278,
      "learning_rate": 0.0004892435259240302,
      "loss": 2.3202,
      "step": 65140
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.6847761273384094,
      "learning_rate": 0.0004892403518923684,
      "loss": 2.2425,
      "step": 65141
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7184231281280518,
      "learning_rate": 0.0004892371778255231,
      "loss": 2.0577,
      "step": 65142
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7471514344215393,
      "learning_rate": 0.0004892340037234951,
      "loss": 2.1953,
      "step": 65143
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7326741814613342,
      "learning_rate": 0.000489230829586285,
      "loss": 2.3608,
      "step": 65144
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.755569338798523,
      "learning_rate": 0.0004892276554138933,
      "loss": 2.2648,
      "step": 65145
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7856767773628235,
      "learning_rate": 0.0004892244812063205,
      "loss": 2.339,
      "step": 65146
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7842791676521301,
      "learning_rate": 0.0004892213069635673,
      "loss": 2.1036,
      "step": 65147
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.8149091005325317,
      "learning_rate": 0.0004892181326856343,
      "loss": 2.367,
      "step": 65148
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7732712030410767,
      "learning_rate": 0.0004892149583725222,
      "loss": 2.1324,
      "step": 65149
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7412878274917603,
      "learning_rate": 0.0004892117840242314,
      "loss": 2.3402,
      "step": 65150
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7611308097839355,
      "learning_rate": 0.0004892086096407624,
      "loss": 2.1752,
      "step": 65151
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7784425020217896,
      "learning_rate": 0.0004892054352221162,
      "loss": 2.1445,
      "step": 65152
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7633475065231323,
      "learning_rate": 0.000489202260768293,
      "loss": 2.1894,
      "step": 65153
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.0062799453735352,
      "learning_rate": 0.0004891990862792936,
      "loss": 2.1524,
      "step": 65154
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7761231660842896,
      "learning_rate": 0.0004891959117551184,
      "loss": 2.4702,
      "step": 65155
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.8300521373748779,
      "learning_rate": 0.0004891927371957682,
      "loss": 2.3147,
      "step": 65156
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.8885089159011841,
      "learning_rate": 0.0004891895626012436,
      "loss": 2.3791,
      "step": 65157
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.8152177333831787,
      "learning_rate": 0.0004891863879715448,
      "loss": 2.2954,
      "step": 65158
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7185888886451721,
      "learning_rate": 0.0004891832133066729,
      "loss": 1.9739,
      "step": 65159
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.775359034538269,
      "learning_rate": 0.0004891800386066282,
      "loss": 2.3055,
      "step": 65160
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.8238202333450317,
      "learning_rate": 0.0004891768638714114,
      "loss": 1.9757,
      "step": 65161
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7681960463523865,
      "learning_rate": 0.0004891736891010231,
      "loss": 2.1654,
      "step": 65162
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7160412669181824,
      "learning_rate": 0.0004891705142954637,
      "loss": 2.094,
      "step": 65163
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.9394149780273438,
      "learning_rate": 0.0004891673394547341,
      "loss": 2.4012,
      "step": 65164
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.959775984287262,
      "learning_rate": 0.0004891641645788344,
      "loss": 2.1391,
      "step": 65165
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7671535015106201,
      "learning_rate": 0.0004891609896677658,
      "loss": 2.1955,
      "step": 65166
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.6857892870903015,
      "learning_rate": 0.0004891578147215286,
      "loss": 2.2201,
      "step": 65167
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7457496523857117,
      "learning_rate": 0.0004891546397401233,
      "loss": 2.481,
      "step": 65168
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.9536371827125549,
      "learning_rate": 0.0004891514647235506,
      "loss": 2.1991,
      "step": 65169
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7576504945755005,
      "learning_rate": 0.0004891482896718112,
      "loss": 2.135,
      "step": 65170
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.8502799272537231,
      "learning_rate": 0.0004891451145849053,
      "loss": 2.35,
      "step": 65171
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7869746685028076,
      "learning_rate": 0.0004891419394628339,
      "loss": 2.1321,
      "step": 65172
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.8422857522964478,
      "learning_rate": 0.0004891387643055975,
      "loss": 2.1915,
      "step": 65173
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7452463507652283,
      "learning_rate": 0.0004891355891131965,
      "loss": 2.1744,
      "step": 65174
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.9151031970977783,
      "learning_rate": 0.0004891324138856318,
      "loss": 2.0332,
      "step": 65175
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.8308212161064148,
      "learning_rate": 0.0004891292386229037,
      "loss": 2.4004,
      "step": 65176
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7305150032043457,
      "learning_rate": 0.0004891260633250128,
      "loss": 2.3126,
      "step": 65177
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.8501158356666565,
      "learning_rate": 0.00048912288799196,
      "loss": 2.2531,
      "step": 65178
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.9298223257064819,
      "learning_rate": 0.0004891197126237455,
      "loss": 2.308,
      "step": 65179
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7424796223640442,
      "learning_rate": 0.0004891165372203702,
      "loss": 2.2508,
      "step": 65180
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.8191269636154175,
      "learning_rate": 0.0004891133617818345,
      "loss": 2.1517,
      "step": 65181
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.8936263918876648,
      "learning_rate": 0.0004891101863081391,
      "loss": 2.125,
      "step": 65182
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.9297631978988647,
      "learning_rate": 0.0004891070107992846,
      "loss": 2.2421,
      "step": 65183
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.895404577255249,
      "learning_rate": 0.0004891038352552714,
      "loss": 2.2886,
      "step": 65184
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.1169414520263672,
      "learning_rate": 0.0004891006596761003,
      "loss": 2.2596,
      "step": 65185
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.8455082178115845,
      "learning_rate": 0.0004890974840617719,
      "loss": 2.4001,
      "step": 65186
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7881773114204407,
      "learning_rate": 0.0004890943084122865,
      "loss": 2.3606,
      "step": 65187
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.0378073453903198,
      "learning_rate": 0.0004890911327276452,
      "loss": 2.2485,
      "step": 65188
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7651519179344177,
      "learning_rate": 0.000489087957007848,
      "loss": 2.4173,
      "step": 65189
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7115671038627625,
      "learning_rate": 0.0004890847812528959,
      "loss": 2.0478,
      "step": 65190
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7122981548309326,
      "learning_rate": 0.0004890816054627894,
      "loss": 2.3077,
      "step": 65191
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.8153901696205139,
      "learning_rate": 0.0004890784296375291,
      "loss": 2.0598,
      "step": 65192
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.8053113222122192,
      "learning_rate": 0.0004890752537771154,
      "loss": 2.3432,
      "step": 65193
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7169587016105652,
      "learning_rate": 0.0004890720778815491,
      "loss": 2.3219,
      "step": 65194
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7405133247375488,
      "learning_rate": 0.0004890689019508308,
      "loss": 2.2297,
      "step": 65195
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.8061590194702148,
      "learning_rate": 0.0004890657259849608,
      "loss": 2.3802,
      "step": 65196
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.9147644639015198,
      "learning_rate": 0.0004890625499839402,
      "loss": 2.3246,
      "step": 65197
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7763270139694214,
      "learning_rate": 0.0004890593739477691,
      "loss": 2.3942,
      "step": 65198
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.9422774910926819,
      "learning_rate": 0.0004890561978764485,
      "loss": 2.2728,
      "step": 65199
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.0051913261413574,
      "learning_rate": 0.0004890530217699787,
      "loss": 2.0519,
      "step": 65200
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7565645575523376,
      "learning_rate": 0.0004890498456283603,
      "loss": 2.1843,
      "step": 65201
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.8417689204216003,
      "learning_rate": 0.000489046669451594,
      "loss": 2.1383,
      "step": 65202
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7358524799346924,
      "learning_rate": 0.0004890434932396804,
      "loss": 1.94,
      "step": 65203
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7253144979476929,
      "learning_rate": 0.0004890403169926199,
      "loss": 2.4059,
      "step": 65204
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.8498010039329529,
      "learning_rate": 0.0004890371407104134,
      "loss": 2.2453,
      "step": 65205
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7172322869300842,
      "learning_rate": 0.0004890339643930613,
      "loss": 2.2708,
      "step": 65206
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7233296036720276,
      "learning_rate": 0.0004890307880405641,
      "loss": 2.1119,
      "step": 65207
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.8135407567024231,
      "learning_rate": 0.0004890276116529226,
      "loss": 2.2305,
      "step": 65208
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7827592492103577,
      "learning_rate": 0.0004890244352301373,
      "loss": 2.2913,
      "step": 65209
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7694294452667236,
      "learning_rate": 0.0004890212587722088,
      "loss": 2.2928,
      "step": 65210
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.8003301620483398,
      "learning_rate": 0.0004890180822791376,
      "loss": 2.1913,
      "step": 65211
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7101588845252991,
      "learning_rate": 0.0004890149057509245,
      "loss": 2.1363,
      "step": 65212
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7240502834320068,
      "learning_rate": 0.0004890117291875699,
      "loss": 2.267,
      "step": 65213
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7420694231987,
      "learning_rate": 0.0004890085525890745,
      "loss": 2.1194,
      "step": 65214
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.8462226986885071,
      "learning_rate": 0.0004890053759554387,
      "loss": 1.9796,
      "step": 65215
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7497376799583435,
      "learning_rate": 0.0004890021992866633,
      "loss": 2.2468,
      "step": 65216
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7684024572372437,
      "learning_rate": 0.0004889990225827489,
      "loss": 2.1973,
      "step": 65217
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.9662711024284363,
      "learning_rate": 0.000488995845843696,
      "loss": 2.1427,
      "step": 65218
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7654399275779724,
      "learning_rate": 0.0004889926690695051,
      "loss": 2.2132,
      "step": 65219
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.8860137462615967,
      "learning_rate": 0.000488989492260177,
      "loss": 2.1918,
      "step": 65220
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7703394293785095,
      "learning_rate": 0.0004889863154157122,
      "loss": 2.3411,
      "step": 65221
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7834303975105286,
      "learning_rate": 0.0004889831385361111,
      "loss": 2.1637,
      "step": 65222
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.873859703540802,
      "learning_rate": 0.0004889799616213746,
      "loss": 2.3354,
      "step": 65223
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.8217769265174866,
      "learning_rate": 0.0004889767846715031,
      "loss": 2.0971,
      "step": 65224
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.8649242520332336,
      "learning_rate": 0.0004889736076864974,
      "loss": 2.1693,
      "step": 65225
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.8920350670814514,
      "learning_rate": 0.0004889704306663577,
      "loss": 2.2553,
      "step": 65226
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.1693633794784546,
      "learning_rate": 0.000488967253611085,
      "loss": 2.3177,
      "step": 65227
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.8708314299583435,
      "learning_rate": 0.0004889640765206797,
      "loss": 2.2234,
      "step": 65228
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.9425538182258606,
      "learning_rate": 0.0004889608993951424,
      "loss": 2.323,
      "step": 65229
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.1394191980361938,
      "learning_rate": 0.0004889577222344738,
      "loss": 2.4785,
      "step": 65230
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.99444180727005,
      "learning_rate": 0.0004889545450386742,
      "loss": 2.2577,
      "step": 65231
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7570157051086426,
      "learning_rate": 0.0004889513678077445,
      "loss": 2.2839,
      "step": 65232
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.6992756724357605,
      "learning_rate": 0.0004889481905416851,
      "loss": 1.9958,
      "step": 65233
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.9120266437530518,
      "learning_rate": 0.0004889450132404968,
      "loss": 2.2812,
      "step": 65234
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7684415578842163,
      "learning_rate": 0.00048894183590418,
      "loss": 2.1777,
      "step": 65235
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7302283048629761,
      "learning_rate": 0.0004889386585327352,
      "loss": 2.0541,
      "step": 65236
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.747979998588562,
      "learning_rate": 0.0004889354811261633,
      "loss": 2.2032,
      "step": 65237
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7618381381034851,
      "learning_rate": 0.0004889323036844647,
      "loss": 2.144,
      "step": 65238
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.8478513360023499,
      "learning_rate": 0.00048892912620764,
      "loss": 2.5404,
      "step": 65239
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7044802904129028,
      "learning_rate": 0.0004889259486956898,
      "loss": 2.2976,
      "step": 65240
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7328451871871948,
      "learning_rate": 0.0004889227711486146,
      "loss": 2.1062,
      "step": 65241
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7858418822288513,
      "learning_rate": 0.0004889195935664153,
      "loss": 2.1228,
      "step": 65242
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.6937085390090942,
      "learning_rate": 0.0004889164159490922,
      "loss": 2.3279,
      "step": 65243
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7366965413093567,
      "learning_rate": 0.0004889132382966459,
      "loss": 2.1881,
      "step": 65244
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.790851891040802,
      "learning_rate": 0.0004889100606090771,
      "loss": 2.2439,
      "step": 65245
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7215853333473206,
      "learning_rate": 0.0004889068828863864,
      "loss": 2.1265,
      "step": 65246
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.8662115335464478,
      "learning_rate": 0.0004889037051285743,
      "loss": 2.4769,
      "step": 65247
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.735274076461792,
      "learning_rate": 0.0004889005273356414,
      "loss": 2.2217,
      "step": 65248
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7325066924095154,
      "learning_rate": 0.0004888973495075884,
      "loss": 2.1383,
      "step": 65249
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7319266200065613,
      "learning_rate": 0.0004888941716444158,
      "loss": 2.2718,
      "step": 65250
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7024779915809631,
      "learning_rate": 0.0004888909937461242,
      "loss": 2.0977,
      "step": 65251
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7880330681800842,
      "learning_rate": 0.0004888878158127143,
      "loss": 2.2556,
      "step": 65252
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.8248717188835144,
      "learning_rate": 0.0004888846378441864,
      "loss": 2.2285,
      "step": 65253
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7130141854286194,
      "learning_rate": 0.0004888814598405413,
      "loss": 2.152,
      "step": 65254
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.6755853295326233,
      "learning_rate": 0.0004888782818017797,
      "loss": 2.1818,
      "step": 65255
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.9142712950706482,
      "learning_rate": 0.0004888751037279021,
      "loss": 2.2142,
      "step": 65256
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.818511962890625,
      "learning_rate": 0.0004888719256189089,
      "loss": 2.2832,
      "step": 65257
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.8853560090065002,
      "learning_rate": 0.0004888687474748009,
      "loss": 2.0574,
      "step": 65258
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7712913751602173,
      "learning_rate": 0.0004888655692955786,
      "loss": 2.1469,
      "step": 65259
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.6465560793876648,
      "learning_rate": 0.0004888623910812426,
      "loss": 2.0695,
      "step": 65260
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7532790303230286,
      "learning_rate": 0.0004888592128317935,
      "loss": 2.2301,
      "step": 65261
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7280661463737488,
      "learning_rate": 0.000488856034547232,
      "loss": 2.4579,
      "step": 65262
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7001765966415405,
      "learning_rate": 0.0004888528562275586,
      "loss": 2.2766,
      "step": 65263
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7302223443984985,
      "learning_rate": 0.0004888496778727739,
      "loss": 2.1087,
      "step": 65264
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7655285596847534,
      "learning_rate": 0.0004888464994828785,
      "loss": 2.2897,
      "step": 65265
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7050333023071289,
      "learning_rate": 0.0004888433210578729,
      "loss": 2.2288,
      "step": 65266
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7604945302009583,
      "learning_rate": 0.0004888401425977579,
      "loss": 2.2648,
      "step": 65267
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.6765736937522888,
      "learning_rate": 0.0004888369641025337,
      "loss": 2.2025,
      "step": 65268
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7510561943054199,
      "learning_rate": 0.0004888337855722013,
      "loss": 2.2217,
      "step": 65269
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7323125004768372,
      "learning_rate": 0.0004888306070067611,
      "loss": 2.2515,
      "step": 65270
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7555961608886719,
      "learning_rate": 0.0004888274284062137,
      "loss": 2.3793,
      "step": 65271
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.8559510707855225,
      "learning_rate": 0.0004888242497705598,
      "loss": 2.1681,
      "step": 65272
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7195924520492554,
      "learning_rate": 0.0004888210710997998,
      "loss": 2.2128,
      "step": 65273
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7374744415283203,
      "learning_rate": 0.0004888178923939346,
      "loss": 1.9762,
      "step": 65274
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.8083618879318237,
      "learning_rate": 0.0004888147136529644,
      "loss": 2.3548,
      "step": 65275
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7277138233184814,
      "learning_rate": 0.00048881153487689,
      "loss": 2.2426,
      "step": 65276
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7735035419464111,
      "learning_rate": 0.0004888083560657119,
      "loss": 2.0091,
      "step": 65277
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7992178201675415,
      "learning_rate": 0.000488805177219431,
      "loss": 2.2326,
      "step": 65278
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.8434140086174011,
      "learning_rate": 0.0004888019983380475,
      "loss": 2.4249,
      "step": 65279
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7244062423706055,
      "learning_rate": 0.0004887988194215621,
      "loss": 2.1976,
      "step": 65280
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.8439875841140747,
      "learning_rate": 0.0004887956404699755,
      "loss": 2.0502,
      "step": 65281
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.719119668006897,
      "learning_rate": 0.0004887924614832883,
      "loss": 2.172,
      "step": 65282
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7280697226524353,
      "learning_rate": 0.0004887892824615009,
      "loss": 2.3304,
      "step": 65283
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.8187242150306702,
      "learning_rate": 0.000488786103404614,
      "loss": 2.2516,
      "step": 65284
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7452712655067444,
      "learning_rate": 0.0004887829243126284,
      "loss": 2.126,
      "step": 65285
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7239654660224915,
      "learning_rate": 0.0004887797451855443,
      "loss": 2.2289,
      "step": 65286
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.726502537727356,
      "learning_rate": 0.0004887765660233625,
      "loss": 2.1889,
      "step": 65287
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.8663672208786011,
      "learning_rate": 0.0004887733868260836,
      "loss": 2.2844,
      "step": 65288
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.8257318735122681,
      "learning_rate": 0.0004887702075937081,
      "loss": 2.4533,
      "step": 65289
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.8200247883796692,
      "learning_rate": 0.0004887670283262367,
      "loss": 2.24,
      "step": 65290
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7681185603141785,
      "learning_rate": 0.00048876384902367,
      "loss": 2.3765,
      "step": 65291
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.004687786102295,
      "learning_rate": 0.0004887606696860084,
      "loss": 2.1694,
      "step": 65292
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7944586277008057,
      "learning_rate": 0.0004887574903132528,
      "loss": 2.2361,
      "step": 65293
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7708859443664551,
      "learning_rate": 0.0004887543109054036,
      "loss": 2.0062,
      "step": 65294
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.8646442890167236,
      "learning_rate": 0.0004887511314624613,
      "loss": 2.199,
      "step": 65295
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7297769784927368,
      "learning_rate": 0.0004887479519844267,
      "loss": 2.0621,
      "step": 65296
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.8093650341033936,
      "learning_rate": 0.0004887447724713002,
      "loss": 2.131,
      "step": 65297
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7365899085998535,
      "learning_rate": 0.0004887415929230826,
      "loss": 2.2837,
      "step": 65298
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7497914433479309,
      "learning_rate": 0.0004887384133397743,
      "loss": 2.2289,
      "step": 65299
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.790905773639679,
      "learning_rate": 0.0004887352337213759,
      "loss": 2.4701,
      "step": 65300
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.8641846179962158,
      "learning_rate": 0.0004887320540678882,
      "loss": 2.3942,
      "step": 65301
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.9334662556648254,
      "learning_rate": 0.0004887288743793116,
      "loss": 2.0691,
      "step": 65302
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7209283709526062,
      "learning_rate": 0.0004887256946556468,
      "loss": 2.2209,
      "step": 65303
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.8209680914878845,
      "learning_rate": 0.0004887225148968941,
      "loss": 2.3842,
      "step": 65304
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7105980515480042,
      "learning_rate": 0.0004887193351030546,
      "loss": 2.1802,
      "step": 65305
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7551665306091309,
      "learning_rate": 0.0004887161552741284,
      "loss": 2.2903,
      "step": 65306
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7098672389984131,
      "learning_rate": 0.0004887129754101165,
      "loss": 2.0934,
      "step": 65307
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7185237407684326,
      "learning_rate": 0.0004887097955110192,
      "loss": 2.2337,
      "step": 65308
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7850534915924072,
      "learning_rate": 0.0004887066155768371,
      "loss": 1.9859,
      "step": 65309
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7548317909240723,
      "learning_rate": 0.0004887034356075709,
      "loss": 2.2676,
      "step": 65310
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7597931623458862,
      "learning_rate": 0.0004887002556032212,
      "loss": 2.042,
      "step": 65311
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7836908102035522,
      "learning_rate": 0.0004886970755637886,
      "loss": 2.4289,
      "step": 65312
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7754237055778503,
      "learning_rate": 0.0004886938954892737,
      "loss": 2.1652,
      "step": 65313
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7325823903083801,
      "learning_rate": 0.000488690715379677,
      "loss": 2.2459,
      "step": 65314
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7964046597480774,
      "learning_rate": 0.000488687535234999,
      "loss": 2.3194,
      "step": 65315
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.858557939529419,
      "learning_rate": 0.0004886843550552407,
      "loss": 2.2005,
      "step": 65316
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.8035776019096375,
      "learning_rate": 0.0004886811748404022,
      "loss": 2.2718,
      "step": 65317
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.8820279240608215,
      "learning_rate": 0.0004886779945904843,
      "loss": 2.1089,
      "step": 65318
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7973668575286865,
      "learning_rate": 0.0004886748143054878,
      "loss": 2.4889,
      "step": 65319
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.9011660218238831,
      "learning_rate": 0.0004886716339854129,
      "loss": 2.0978,
      "step": 65320
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7991254329681396,
      "learning_rate": 0.0004886684536302604,
      "loss": 2.2806,
      "step": 65321
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.802554190158844,
      "learning_rate": 0.000488665273240031,
      "loss": 2.1554,
      "step": 65322
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.8975942134857178,
      "learning_rate": 0.0004886620928147249,
      "loss": 2.3323,
      "step": 65323
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7761215567588806,
      "learning_rate": 0.0004886589123543432,
      "loss": 2.1847,
      "step": 65324
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7597522735595703,
      "learning_rate": 0.0004886557318588863,
      "loss": 2.0646,
      "step": 65325
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7436767220497131,
      "learning_rate": 0.0004886525513283545,
      "loss": 2.1296,
      "step": 65326
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7908897995948792,
      "learning_rate": 0.0004886493707627487,
      "loss": 2.4164,
      "step": 65327
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.831432580947876,
      "learning_rate": 0.0004886461901620696,
      "loss": 2.0098,
      "step": 65328
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7777420878410339,
      "learning_rate": 0.0004886430095263174,
      "loss": 2.4552,
      "step": 65329
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7346542477607727,
      "learning_rate": 0.0004886398288554929,
      "loss": 2.082,
      "step": 65330
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7732387781143188,
      "learning_rate": 0.0004886366481495969,
      "loss": 2.3995,
      "step": 65331
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.8266642689704895,
      "learning_rate": 0.0004886334674086296,
      "loss": 2.4602,
      "step": 65332
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7359257936477661,
      "learning_rate": 0.0004886302866325918,
      "loss": 2.0402,
      "step": 65333
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7881151437759399,
      "learning_rate": 0.0004886271058214841,
      "loss": 2.1527,
      "step": 65334
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.8994447588920593,
      "learning_rate": 0.0004886239249753072,
      "loss": 2.2356,
      "step": 65335
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.815966784954071,
      "learning_rate": 0.0004886207440940613,
      "loss": 2.489,
      "step": 65336
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7599233984947205,
      "learning_rate": 0.0004886175631777475,
      "loss": 2.1122,
      "step": 65337
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.8351233005523682,
      "learning_rate": 0.0004886143822263659,
      "loss": 2.199,
      "step": 65338
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.9932863116264343,
      "learning_rate": 0.0004886112012399174,
      "loss": 2.431,
      "step": 65339
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7614584565162659,
      "learning_rate": 0.0004886080202184025,
      "loss": 2.4998,
      "step": 65340
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7942287921905518,
      "learning_rate": 0.0004886048391618219,
      "loss": 2.3059,
      "step": 65341
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.9504888653755188,
      "learning_rate": 0.000488601658070176,
      "loss": 2.157,
      "step": 65342
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7854145765304565,
      "learning_rate": 0.0004885984769434657,
      "loss": 2.418,
      "step": 65343
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7612941861152649,
      "learning_rate": 0.0004885952957816911,
      "loss": 2.1881,
      "step": 65344
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.9895309805870056,
      "learning_rate": 0.0004885921145848533,
      "loss": 2.1832,
      "step": 65345
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7781394124031067,
      "learning_rate": 0.0004885889333529525,
      "loss": 2.2164,
      "step": 65346
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7163580060005188,
      "learning_rate": 0.0004885857520859894,
      "loss": 2.4348,
      "step": 65347
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7263369560241699,
      "learning_rate": 0.0004885825707839649,
      "loss": 2.2915,
      "step": 65348
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7315653562545776,
      "learning_rate": 0.0004885793894468792,
      "loss": 2.1287,
      "step": 65349
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.8146139979362488,
      "learning_rate": 0.000488576208074733,
      "loss": 2.2868,
      "step": 65350
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7764118313789368,
      "learning_rate": 0.0004885730266675271,
      "loss": 2.2697,
      "step": 65351
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.6915837526321411,
      "learning_rate": 0.0004885698452252616,
      "loss": 1.8873,
      "step": 65352
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.8288974165916443,
      "learning_rate": 0.0004885666637479376,
      "loss": 2.291,
      "step": 65353
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.8102906942367554,
      "learning_rate": 0.0004885634822355555,
      "loss": 2.3203,
      "step": 65354
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7759542465209961,
      "learning_rate": 0.000488560300688116,
      "loss": 2.0947,
      "step": 65355
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.8499199151992798,
      "learning_rate": 0.0004885571191056194,
      "loss": 2.2935,
      "step": 65356
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.9872127175331116,
      "learning_rate": 0.0004885539374880665,
      "loss": 2.0252,
      "step": 65357
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7870935201644897,
      "learning_rate": 0.0004885507558354578,
      "loss": 2.3849,
      "step": 65358
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.8447356820106506,
      "learning_rate": 0.0004885475741477942,
      "loss": 2.3076,
      "step": 65359
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.8863713145256042,
      "learning_rate": 0.0004885443924250759,
      "loss": 2.3714,
      "step": 65360
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.0136152505874634,
      "learning_rate": 0.0004885412106673036,
      "loss": 2.2067,
      "step": 65361
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7700210213661194,
      "learning_rate": 0.000488538028874478,
      "loss": 2.3789,
      "step": 65362
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.0723872184753418,
      "learning_rate": 0.0004885348470465994,
      "loss": 2.1906,
      "step": 65363
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7973451614379883,
      "learning_rate": 0.0004885316651836688,
      "loss": 2.2687,
      "step": 65364
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7353290915489197,
      "learning_rate": 0.0004885284832856867,
      "loss": 2.34,
      "step": 65365
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7141881585121155,
      "learning_rate": 0.0004885253013526535,
      "loss": 2.3174,
      "step": 65366
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.8290736079216003,
      "learning_rate": 0.0004885221193845699,
      "loss": 1.9419,
      "step": 65367
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.8248958587646484,
      "learning_rate": 0.0004885189373814364,
      "loss": 2.2106,
      "step": 65368
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.8019511103630066,
      "learning_rate": 0.0004885157553432538,
      "loss": 2.6512,
      "step": 65369
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7862874865531921,
      "learning_rate": 0.0004885125732700225,
      "loss": 2.0507,
      "step": 65370
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.9739153981208801,
      "learning_rate": 0.0004885093911617432,
      "loss": 2.2398,
      "step": 65371
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7840731739997864,
      "learning_rate": 0.0004885062090184163,
      "loss": 2.1003,
      "step": 65372
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7477855682373047,
      "learning_rate": 0.0004885030268400426,
      "loss": 2.3635,
      "step": 65373
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.018454670906067,
      "learning_rate": 0.0004884998446266228,
      "loss": 2.2798,
      "step": 65374
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7675075531005859,
      "learning_rate": 0.0004884966623781571,
      "loss": 2.5103,
      "step": 65375
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7527024149894714,
      "learning_rate": 0.0004884934800946463,
      "loss": 2.4048,
      "step": 65376
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7302677035331726,
      "learning_rate": 0.0004884902977760913,
      "loss": 2.2312,
      "step": 65377
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.093390703201294,
      "learning_rate": 0.0004884871154224921,
      "loss": 2.2179,
      "step": 65378
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7798289060592651,
      "learning_rate": 0.0004884839330338497,
      "loss": 2.18,
      "step": 65379
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.8622243404388428,
      "learning_rate": 0.0004884807506101646,
      "loss": 2.3038,
      "step": 65380
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.9405795335769653,
      "learning_rate": 0.0004884775681514373,
      "loss": 2.0773,
      "step": 65381
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7657008171081543,
      "learning_rate": 0.0004884743856576686,
      "loss": 2.0846,
      "step": 65382
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7217671275138855,
      "learning_rate": 0.0004884712031288588,
      "loss": 2.2712,
      "step": 65383
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.8030598759651184,
      "learning_rate": 0.0004884680205650087,
      "loss": 2.3485,
      "step": 65384
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.8040837645530701,
      "learning_rate": 0.0004884648379661188,
      "loss": 2.1047,
      "step": 65385
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.8086891174316406,
      "learning_rate": 0.0004884616553321898,
      "loss": 2.1779,
      "step": 65386
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7674778699874878,
      "learning_rate": 0.0004884584726632221,
      "loss": 2.4384,
      "step": 65387
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.735930860042572,
      "learning_rate": 0.0004884552899592165,
      "loss": 2.124,
      "step": 65388
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.8303931355476379,
      "learning_rate": 0.0004884521072201736,
      "loss": 1.9614,
      "step": 65389
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7583054304122925,
      "learning_rate": 0.0004884489244460937,
      "loss": 2.2118,
      "step": 65390
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7763271331787109,
      "learning_rate": 0.0004884457416369777,
      "loss": 2.3711,
      "step": 65391
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.829976499080658,
      "learning_rate": 0.0004884425587928261,
      "loss": 2.3328,
      "step": 65392
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.8447715640068054,
      "learning_rate": 0.0004884393759136394,
      "loss": 2.5445,
      "step": 65393
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.888882040977478,
      "learning_rate": 0.0004884361929994183,
      "loss": 2.2808,
      "step": 65394
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.8601523041725159,
      "learning_rate": 0.0004884330100501634,
      "loss": 2.2314,
      "step": 65395
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7615736126899719,
      "learning_rate": 0.0004884298270658752,
      "loss": 2.1641,
      "step": 65396
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.722339391708374,
      "learning_rate": 0.0004884266440465543,
      "loss": 2.1793,
      "step": 65397
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.8925102353096008,
      "learning_rate": 0.0004884234609922013,
      "loss": 2.1288,
      "step": 65398
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7376114726066589,
      "learning_rate": 0.0004884202779028169,
      "loss": 2.2623,
      "step": 65399
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.783187747001648,
      "learning_rate": 0.0004884170947784016,
      "loss": 2.3204,
      "step": 65400
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.8890585899353027,
      "learning_rate": 0.000488413911618956,
      "loss": 2.3447,
      "step": 65401
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.9083835482597351,
      "learning_rate": 0.0004884107284244808,
      "loss": 2.4133,
      "step": 65402
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7174956798553467,
      "learning_rate": 0.0004884075451949764,
      "loss": 2.1608,
      "step": 65403
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.8371433615684509,
      "learning_rate": 0.0004884043619304435,
      "loss": 2.0899,
      "step": 65404
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7545642256736755,
      "learning_rate": 0.0004884011786308826,
      "loss": 2.1559,
      "step": 65405
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.8685469627380371,
      "learning_rate": 0.0004883979952962944,
      "loss": 2.3213,
      "step": 65406
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7169358134269714,
      "learning_rate": 0.0004883948119266795,
      "loss": 2.0038,
      "step": 65407
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.8047723770141602,
      "learning_rate": 0.0004883916285220384,
      "loss": 2.24,
      "step": 65408
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7768892049789429,
      "learning_rate": 0.0004883884450823718,
      "loss": 2.1283,
      "step": 65409
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.8079730272293091,
      "learning_rate": 0.0004883852616076801,
      "loss": 2.1331,
      "step": 65410
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.8496909141540527,
      "learning_rate": 0.0004883820780979641,
      "loss": 2.2562,
      "step": 65411
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.9214516282081604,
      "learning_rate": 0.0004883788945532244,
      "loss": 2.1438,
      "step": 65412
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.8215911984443665,
      "learning_rate": 0.0004883757109734614,
      "loss": 2.3171,
      "step": 65413
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7320662140846252,
      "learning_rate": 0.0004883725273586759,
      "loss": 2.1843,
      "step": 65414
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.8337923288345337,
      "learning_rate": 0.0004883693437088682,
      "loss": 2.0973,
      "step": 65415
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7745478749275208,
      "learning_rate": 0.0004883661600240392,
      "loss": 2.3716,
      "step": 65416
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.6904152035713196,
      "learning_rate": 0.0004883629763041895,
      "loss": 2.2861,
      "step": 65417
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.856454074382782,
      "learning_rate": 0.0004883597925493193,
      "loss": 2.1216,
      "step": 65418
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7894916534423828,
      "learning_rate": 0.0004883566087594298,
      "loss": 2.101,
      "step": 65419
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.8104332089424133,
      "learning_rate": 0.000488353424934521,
      "loss": 2.1114,
      "step": 65420
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.8352302312850952,
      "learning_rate": 0.0004883502410745938,
      "loss": 2.1398,
      "step": 65421
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.781075656414032,
      "learning_rate": 0.0004883470571796488,
      "loss": 2.297,
      "step": 65422
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.804962158203125,
      "learning_rate": 0.0004883438732496865,
      "loss": 2.1827,
      "step": 65423
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7947297692298889,
      "learning_rate": 0.0004883406892847074,
      "loss": 2.3884,
      "step": 65424
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7274585962295532,
      "learning_rate": 0.0004883375052847122,
      "loss": 2.1268,
      "step": 65425
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.9884893298149109,
      "learning_rate": 0.0004883343212497017,
      "loss": 2.0638,
      "step": 65426
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7712981700897217,
      "learning_rate": 0.0004883311371796762,
      "loss": 2.3638,
      "step": 65427
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.8027414679527283,
      "learning_rate": 0.0004883279530746363,
      "loss": 2.1015,
      "step": 65428
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7641007900238037,
      "learning_rate": 0.0004883247689345828,
      "loss": 2.2688,
      "step": 65429
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7369482517242432,
      "learning_rate": 0.0004883215847595162,
      "loss": 2.0428,
      "step": 65430
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.687146782875061,
      "learning_rate": 0.0004883184005494369,
      "loss": 2.2689,
      "step": 65431
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7164077162742615,
      "learning_rate": 0.0004883152163043458,
      "loss": 2.2709,
      "step": 65432
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7499405741691589,
      "learning_rate": 0.0004883120320242433,
      "loss": 2.4043,
      "step": 65433
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.8608490228652954,
      "learning_rate": 0.00048830884770913,
      "loss": 2.2741,
      "step": 65434
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.8281062245368958,
      "learning_rate": 0.0004883056633590066,
      "loss": 2.0009,
      "step": 65435
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.796002984046936,
      "learning_rate": 0.0004883024789738735,
      "loss": 2.0979,
      "step": 65436
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7452920079231262,
      "learning_rate": 0.0004882992945537315,
      "loss": 2.3479,
      "step": 65437
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7483240962028503,
      "learning_rate": 0.0004882961100985811,
      "loss": 2.2922,
      "step": 65438
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.8264606595039368,
      "learning_rate": 0.0004882929256084229,
      "loss": 2.2237,
      "step": 65439
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.6790310740470886,
      "learning_rate": 0.0004882897410832575,
      "loss": 2.2464,
      "step": 65440
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7738495469093323,
      "learning_rate": 0.0004882865565230855,
      "loss": 2.1075,
      "step": 65441
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.855298638343811,
      "learning_rate": 0.0004882833719279075,
      "loss": 2.3576,
      "step": 65442
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7393179535865784,
      "learning_rate": 0.000488280187297724,
      "loss": 2.4586,
      "step": 65443
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7231028079986572,
      "learning_rate": 0.00048827700263253564,
      "loss": 2.4219,
      "step": 65444
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7284873127937317,
      "learning_rate": 0.00048827381793234306,
      "loss": 2.2761,
      "step": 65445
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.6996386647224426,
      "learning_rate": 0.0004882706331971469,
      "loss": 2.3188,
      "step": 65446
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7259454727172852,
      "learning_rate": 0.0004882674484269476,
      "loss": 2.2938,
      "step": 65447
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7674811482429504,
      "learning_rate": 0.0004882642636217458,
      "loss": 2.069,
      "step": 65448
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.8899829387664795,
      "learning_rate": 0.0004882610787815422,
      "loss": 2.3599,
      "step": 65449
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.8124983906745911,
      "learning_rate": 0.00048825789390633727,
      "loss": 2.2461,
      "step": 65450
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.8416747450828552,
      "learning_rate": 0.00048825470899613164,
      "loss": 2.1762,
      "step": 65451
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7149934768676758,
      "learning_rate": 0.00048825152405092585,
      "loss": 2.1603,
      "step": 65452
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.785832405090332,
      "learning_rate": 0.0004882483390707206,
      "loss": 2.1528,
      "step": 65453
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7445181608200073,
      "learning_rate": 0.0004882451540555165,
      "loss": 2.3125,
      "step": 65454
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7092226147651672,
      "learning_rate": 0.000488241969005314,
      "loss": 2.0876,
      "step": 65455
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7450448274612427,
      "learning_rate": 0.00048823878392011377,
      "loss": 2.4149,
      "step": 65456
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7731013894081116,
      "learning_rate": 0.00048823559879991645,
      "loss": 2.2238,
      "step": 65457
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.8345930576324463,
      "learning_rate": 0.00048823241364472243,
      "loss": 2.2655,
      "step": 65458
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.9868433475494385,
      "learning_rate": 0.00048822922845453265,
      "loss": 2.049,
      "step": 65459
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7382763624191284,
      "learning_rate": 0.00048822604322934737,
      "loss": 2.3276,
      "step": 65460
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7598885893821716,
      "learning_rate": 0.00048822285796916735,
      "loss": 2.3438,
      "step": 65461
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.9771155118942261,
      "learning_rate": 0.0004882196726739932,
      "loss": 2.1599,
      "step": 65462
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7731888294219971,
      "learning_rate": 0.00048821648734382537,
      "loss": 2.2404,
      "step": 65463
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7097954154014587,
      "learning_rate": 0.00048821330197866455,
      "loss": 2.4119,
      "step": 65464
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.8257102966308594,
      "learning_rate": 0.0004882101165785113,
      "loss": 2.3707,
      "step": 65465
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.8392729759216309,
      "learning_rate": 0.00048820693114336633,
      "loss": 2.3343,
      "step": 65466
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7571464776992798,
      "learning_rate": 0.00048820374567323013,
      "loss": 2.2475,
      "step": 65467
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.8547573685646057,
      "learning_rate": 0.00048820056016810325,
      "loss": 2.2826,
      "step": 65468
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.9289993643760681,
      "learning_rate": 0.00048819737462798635,
      "loss": 2.1555,
      "step": 65469
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7412267923355103,
      "learning_rate": 0.00048819418905287997,
      "loss": 2.3088,
      "step": 65470
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7440453767776489,
      "learning_rate": 0.0004881910034427848,
      "loss": 2.2576,
      "step": 65471
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.968326985836029,
      "learning_rate": 0.00048818781779770136,
      "loss": 2.4696,
      "step": 65472
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.8065727949142456,
      "learning_rate": 0.0004881846321176303,
      "loss": 2.2693,
      "step": 65473
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.6704303622245789,
      "learning_rate": 0.00048818144640257205,
      "loss": 2.2336,
      "step": 65474
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.8271335959434509,
      "learning_rate": 0.00048817826065252737,
      "loss": 2.0377,
      "step": 65475
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.8470661044120789,
      "learning_rate": 0.0004881750748674968,
      "loss": 2.3022,
      "step": 65476
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7504131197929382,
      "learning_rate": 0.00048817188904748095,
      "loss": 2.2926,
      "step": 65477
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.8348487019538879,
      "learning_rate": 0.0004881687031924803,
      "loss": 2.2447,
      "step": 65478
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7600193619728088,
      "learning_rate": 0.0004881655173024957,
      "loss": 2.4522,
      "step": 65479
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7118062376976013,
      "learning_rate": 0.00048816233137752746,
      "loss": 2.2146,
      "step": 65480
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7607899308204651,
      "learning_rate": 0.0004881591454175763,
      "loss": 2.2992,
      "step": 65481
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.750558078289032,
      "learning_rate": 0.00048815595942264283,
      "loss": 2.2559,
      "step": 65482
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.8651338815689087,
      "learning_rate": 0.0004881527733927276,
      "loss": 2.2731,
      "step": 65483
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.8106259703636169,
      "learning_rate": 0.00048814958732783123,
      "loss": 2.1271,
      "step": 65484
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7381409406661987,
      "learning_rate": 0.0004881464012279543,
      "loss": 1.971,
      "step": 65485
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.6893198490142822,
      "learning_rate": 0.0004881432150930974,
      "loss": 2.177,
      "step": 65486
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.9286487698554993,
      "learning_rate": 0.00048814002892326116,
      "loss": 2.3434,
      "step": 65487
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.8571054935455322,
      "learning_rate": 0.0004881368427184461,
      "loss": 2.124,
      "step": 65488
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.6995698809623718,
      "learning_rate": 0.0004881336564786529,
      "loss": 2.2057,
      "step": 65489
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.796362042427063,
      "learning_rate": 0.000488130470203882,
      "loss": 2.2308,
      "step": 65490
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.762104868888855,
      "learning_rate": 0.00048812728389413415,
      "loss": 2.2492,
      "step": 65491
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7866469025611877,
      "learning_rate": 0.0004881240975494098,
      "loss": 2.0673,
      "step": 65492
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7712622880935669,
      "learning_rate": 0.00048812091116970976,
      "loss": 2.1358,
      "step": 65493
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7745331525802612,
      "learning_rate": 0.0004881177247550344,
      "loss": 2.2392,
      "step": 65494
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.9024894833564758,
      "learning_rate": 0.00048811453830538453,
      "loss": 2.2881,
      "step": 65495
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7595905065536499,
      "learning_rate": 0.00048811135182076046,
      "loss": 2.2296,
      "step": 65496
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7826541662216187,
      "learning_rate": 0.000488108165301163,
      "loss": 2.2937,
      "step": 65497
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.8807822465896606,
      "learning_rate": 0.0004881049787465928,
      "loss": 2.237,
      "step": 65498
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.9234534502029419,
      "learning_rate": 0.0004881017921570502,
      "loss": 2.1483,
      "step": 65499
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.6346187591552734,
      "learning_rate": 0.0004880986055325359,
      "loss": 2.0922,
      "step": 65500
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.012134075164795,
      "learning_rate": 0.0004880954188730506,
      "loss": 2.2063,
      "step": 65501
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.890727698802948,
      "learning_rate": 0.00048809223217859485,
      "loss": 2.3334,
      "step": 65502
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7672082781791687,
      "learning_rate": 0.00048808904544916904,
      "loss": 2.241,
      "step": 65503
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7993566989898682,
      "learning_rate": 0.00048808585868477403,
      "loss": 2.3374,
      "step": 65504
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7282058596611023,
      "learning_rate": 0.00048808267188541035,
      "loss": 1.9887,
      "step": 65505
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.8598458170890808,
      "learning_rate": 0.0004880794850510785,
      "loss": 2.2481,
      "step": 65506
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7433176636695862,
      "learning_rate": 0.0004880762981817791,
      "loss": 2.126,
      "step": 65507
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7854006290435791,
      "learning_rate": 0.00048807311127751276,
      "loss": 2.2899,
      "step": 65508
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7634682655334473,
      "learning_rate": 0.0004880699243382801,
      "loss": 2.0177,
      "step": 65509
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7395622730255127,
      "learning_rate": 0.00048806673736408174,
      "loss": 2.0352,
      "step": 65510
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7895839214324951,
      "learning_rate": 0.0004880635503549182,
      "loss": 2.1969,
      "step": 65511
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7367753386497498,
      "learning_rate": 0.0004880603633107901,
      "loss": 2.1618,
      "step": 65512
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7927395105361938,
      "learning_rate": 0.000488057176231698,
      "loss": 2.309,
      "step": 65513
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.794232964515686,
      "learning_rate": 0.00048805398911764257,
      "loss": 2.2642,
      "step": 65514
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7961112260818481,
      "learning_rate": 0.0004880508019686243,
      "loss": 2.2015,
      "step": 65515
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7505789995193481,
      "learning_rate": 0.0004880476147846438,
      "loss": 2.1061,
      "step": 65516
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7272190451622009,
      "learning_rate": 0.00048804442756570184,
      "loss": 2.3507,
      "step": 65517
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7459356188774109,
      "learning_rate": 0.00048804124031179877,
      "loss": 2.1963,
      "step": 65518
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7664790749549866,
      "learning_rate": 0.00048803805302293526,
      "loss": 2.1976,
      "step": 65519
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.699871301651001,
      "learning_rate": 0.000488034865699112,
      "loss": 2.2038,
      "step": 65520
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7828841805458069,
      "learning_rate": 0.0004880316783403295,
      "loss": 2.1706,
      "step": 65521
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.8434458374977112,
      "learning_rate": 0.0004880284909465883,
      "loss": 2.1972,
      "step": 65522
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.6894726753234863,
      "learning_rate": 0.000488025303517889,
      "loss": 2.1656,
      "step": 65523
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.8216612935066223,
      "learning_rate": 0.0004880221160542324,
      "loss": 2.3074,
      "step": 65524
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.8429512977600098,
      "learning_rate": 0.0004880189285556188,
      "loss": 2.1028,
      "step": 65525
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7555306553840637,
      "learning_rate": 0.00048801574102204907,
      "loss": 2.4409,
      "step": 65526
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7517753839492798,
      "learning_rate": 0.0004880125534535236,
      "loss": 2.2738,
      "step": 65527
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.8592848181724548,
      "learning_rate": 0.00048800936585004305,
      "loss": 2.0873,
      "step": 65528
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.709201455116272,
      "learning_rate": 0.00048800617821160794,
      "loss": 2.4083,
      "step": 65529
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.8663290143013,
      "learning_rate": 0.000488002990538219,
      "loss": 2.0263,
      "step": 65530
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7347660660743713,
      "learning_rate": 0.0004879998028298768,
      "loss": 2.2066,
      "step": 65531
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7278194427490234,
      "learning_rate": 0.0004879966150865818,
      "loss": 2.2999,
      "step": 65532
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7436323761940002,
      "learning_rate": 0.00048799342730833475,
      "loss": 2.3007,
      "step": 65533
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7484349608421326,
      "learning_rate": 0.0004879902394951362,
      "loss": 2.219,
      "step": 65534
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7215303778648376,
      "learning_rate": 0.00048798705164698666,
      "loss": 2.1787,
      "step": 65535
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7248905301094055,
      "learning_rate": 0.00048798386376388674,
      "loss": 2.2338,
      "step": 65536
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7951688170433044,
      "learning_rate": 0.0004879806758458371,
      "loss": 2.183,
      "step": 65537
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.809580385684967,
      "learning_rate": 0.00048797748789283833,
      "loss": 2.2783,
      "step": 65538
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7113946080207825,
      "learning_rate": 0.000487974299904891,
      "loss": 2.3971,
      "step": 65539
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7440187931060791,
      "learning_rate": 0.00048797111188199566,
      "loss": 2.0753,
      "step": 65540
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7019940614700317,
      "learning_rate": 0.000487967923824153,
      "loss": 2.1434,
      "step": 65541
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7217992544174194,
      "learning_rate": 0.0004879647357313635,
      "loss": 2.0748,
      "step": 65542
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7290558218955994,
      "learning_rate": 0.00048796154760362783,
      "loss": 2.2431,
      "step": 65543
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.8215296268463135,
      "learning_rate": 0.00048795835944094655,
      "loss": 2.0866,
      "step": 65544
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7486039400100708,
      "learning_rate": 0.0004879551712433203,
      "loss": 2.039,
      "step": 65545
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.6761118769645691,
      "learning_rate": 0.0004879519830107496,
      "loss": 2.1637,
      "step": 65546
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7301153540611267,
      "learning_rate": 0.0004879487947432351,
      "loss": 2.2788,
      "step": 65547
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7038268446922302,
      "learning_rate": 0.00048794560644077725,
      "loss": 2.1538,
      "step": 65548
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7796300053596497,
      "learning_rate": 0.0004879424181033769,
      "loss": 2.2573,
      "step": 65549
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.8028501868247986,
      "learning_rate": 0.0004879392297310345,
      "loss": 2.1098,
      "step": 65550
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7708708643913269,
      "learning_rate": 0.00048793604132375064,
      "loss": 2.5392,
      "step": 65551
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.8428710103034973,
      "learning_rate": 0.00048793285288152593,
      "loss": 2.158,
      "step": 65552
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.8684422373771667,
      "learning_rate": 0.00048792966440436097,
      "loss": 2.2207,
      "step": 65553
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7233068943023682,
      "learning_rate": 0.00048792647589225623,
      "loss": 2.3598,
      "step": 65554
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7321509122848511,
      "learning_rate": 0.0004879232873452125,
      "loss": 2.064,
      "step": 65555
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.8428732752799988,
      "learning_rate": 0.0004879200987632303,
      "loss": 2.0335,
      "step": 65556
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.9059145450592041,
      "learning_rate": 0.0004879169101463101,
      "loss": 2.4285,
      "step": 65557
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.748816192150116,
      "learning_rate": 0.0004879137214944527,
      "loss": 2.0248,
      "step": 65558
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.8109869956970215,
      "learning_rate": 0.0004879105328076585,
      "loss": 2.1075,
      "step": 65559
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.8287616968154907,
      "learning_rate": 0.0004879073440859283,
      "loss": 2.2422,
      "step": 65560
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7486351132392883,
      "learning_rate": 0.0004879041553292625,
      "loss": 2.1591,
      "step": 65561
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.8349263668060303,
      "learning_rate": 0.0004879009665376618,
      "loss": 1.9858,
      "step": 65562
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.8401462435722351,
      "learning_rate": 0.0004878977777111268,
      "loss": 2.069,
      "step": 65563
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7167554497718811,
      "learning_rate": 0.000487894588849658,
      "loss": 2.3865,
      "step": 65564
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.6913900375366211,
      "learning_rate": 0.000487891399953256,
      "loss": 2.2165,
      "step": 65565
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.8694576025009155,
      "learning_rate": 0.0004878882110219216,
      "loss": 2.1535,
      "step": 65566
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7372793555259705,
      "learning_rate": 0.0004878850220556551,
      "loss": 2.1934,
      "step": 65567
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.6938832998275757,
      "learning_rate": 0.0004878818330544573,
      "loss": 2.274,
      "step": 65568
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.8129391074180603,
      "learning_rate": 0.0004878786440183287,
      "loss": 2.207,
      "step": 65569
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7725723385810852,
      "learning_rate": 0.0004878754549472699,
      "loss": 2.0438,
      "step": 65570
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.8504824042320251,
      "learning_rate": 0.0004878722658412815,
      "loss": 2.2987,
      "step": 65571
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7561123967170715,
      "learning_rate": 0.00048786907670036404,
      "loss": 2.4434,
      "step": 65572
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7550054788589478,
      "learning_rate": 0.0004878658875245183,
      "loss": 2.3573,
      "step": 65573
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7410888671875,
      "learning_rate": 0.00048786269831374465,
      "loss": 2.0576,
      "step": 65574
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7734822630882263,
      "learning_rate": 0.00048785950906804384,
      "loss": 2.1359,
      "step": 65575
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.9240953326225281,
      "learning_rate": 0.0004878563197874164,
      "loss": 2.2684,
      "step": 65576
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7713436484336853,
      "learning_rate": 0.0004878531304718629,
      "loss": 2.2686,
      "step": 65577
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.815570592880249,
      "learning_rate": 0.00048784994112138395,
      "loss": 2.0141,
      "step": 65578
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7696236968040466,
      "learning_rate": 0.0004878467517359801,
      "loss": 2.2997,
      "step": 65579
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7529116868972778,
      "learning_rate": 0.0004878435623156521,
      "loss": 1.9809,
      "step": 65580
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7054443955421448,
      "learning_rate": 0.00048784037286040034,
      "loss": 2.0584,
      "step": 65581
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.838635265827179,
      "learning_rate": 0.00048783718337022555,
      "loss": 2.3655,
      "step": 65582
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7621883153915405,
      "learning_rate": 0.00048783399384512834,
      "loss": 2.184,
      "step": 65583
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7637276649475098,
      "learning_rate": 0.0004878308042851092,
      "loss": 2.3262,
      "step": 65584
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.8126168251037598,
      "learning_rate": 0.00048782761469016875,
      "loss": 2.0842,
      "step": 65585
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7388768196105957,
      "learning_rate": 0.0004878244250603076,
      "loss": 2.1001,
      "step": 65586
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.8354209065437317,
      "learning_rate": 0.00048782123539552636,
      "loss": 2.2411,
      "step": 65587
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7439534068107605,
      "learning_rate": 0.00048781804569582554,
      "loss": 2.26,
      "step": 65588
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7669044137001038,
      "learning_rate": 0.00048781485596120593,
      "loss": 2.3419,
      "step": 65589
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7468212842941284,
      "learning_rate": 0.00048781166619166796,
      "loss": 2.2714,
      "step": 65590
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7541022300720215,
      "learning_rate": 0.0004878084763872122,
      "loss": 2.3435,
      "step": 65591
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.8292965888977051,
      "learning_rate": 0.0004878052865478393,
      "loss": 2.0889,
      "step": 65592
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7918586134910583,
      "learning_rate": 0.0004878020966735499,
      "loss": 2.3537,
      "step": 65593
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7777567505836487,
      "learning_rate": 0.00048779890676434444,
      "loss": 2.05,
      "step": 65594
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.8917641043663025,
      "learning_rate": 0.00048779571682022374,
      "loss": 2.1763,
      "step": 65595
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.8564223051071167,
      "learning_rate": 0.00048779252684118826,
      "loss": 2.1779,
      "step": 65596
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7449842691421509,
      "learning_rate": 0.00048778933682723855,
      "loss": 2.135,
      "step": 65597
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.8828331232070923,
      "learning_rate": 0.0004877861467783753,
      "loss": 2.3539,
      "step": 65598
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7442587018013,
      "learning_rate": 0.00048778295669459906,
      "loss": 2.2127,
      "step": 65599
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.8813806772232056,
      "learning_rate": 0.0004877797665759104,
      "loss": 2.1518,
      "step": 65600
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.8325899839401245,
      "learning_rate": 0.00048777657642230994,
      "loss": 2.1004,
      "step": 65601
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.8161018490791321,
      "learning_rate": 0.0004877733862337983,
      "loss": 2.1619,
      "step": 65602
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.8823895454406738,
      "learning_rate": 0.0004877701960103759,
      "loss": 2.3442,
      "step": 65603
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.8108558058738708,
      "learning_rate": 0.0004877670057520436,
      "loss": 2.1986,
      "step": 65604
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7879632711410522,
      "learning_rate": 0.0004877638154588019,
      "loss": 2.3116,
      "step": 65605
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.9796862602233887,
      "learning_rate": 0.0004877606251306513,
      "loss": 2.2989,
      "step": 65606
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.9623726606369019,
      "learning_rate": 0.0004877574347675925,
      "loss": 2.1644,
      "step": 65607
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7414332628250122,
      "learning_rate": 0.000487754244369626,
      "loss": 2.2199,
      "step": 65608
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7888386249542236,
      "learning_rate": 0.00048775105393675255,
      "loss": 2.0979,
      "step": 65609
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.9217237830162048,
      "learning_rate": 0.00048774786346897245,
      "loss": 2.1685,
      "step": 65610
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7943374514579773,
      "learning_rate": 0.00048774467296628667,
      "loss": 1.942,
      "step": 65611
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.8195043206214905,
      "learning_rate": 0.00048774148242869544,
      "loss": 2.3026,
      "step": 65612
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7691687345504761,
      "learning_rate": 0.0004877382918561997,
      "loss": 2.3613,
      "step": 65613
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.85187828540802,
      "learning_rate": 0.0004877351012487997,
      "loss": 2.0153,
      "step": 65614
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7898656129837036,
      "learning_rate": 0.0004877319106064963,
      "loss": 2.2472,
      "step": 65615
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7356268763542175,
      "learning_rate": 0.0004877287199292899,
      "loss": 2.0891,
      "step": 65616
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7421901226043701,
      "learning_rate": 0.0004877255292171813,
      "loss": 2.4014,
      "step": 65617
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7745688557624817,
      "learning_rate": 0.0004877223384701709,
      "loss": 2.058,
      "step": 65618
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7998470664024353,
      "learning_rate": 0.0004877191476882594,
      "loss": 2.0597,
      "step": 65619
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.814670979976654,
      "learning_rate": 0.0004877159568714474,
      "loss": 2.0876,
      "step": 65620
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7963978052139282,
      "learning_rate": 0.00048771276601973544,
      "loss": 2.068,
      "step": 65621
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.8628315925598145,
      "learning_rate": 0.00048770957513312414,
      "loss": 2.2892,
      "step": 65622
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.9612768292427063,
      "learning_rate": 0.0004877063842116141,
      "loss": 2.1764,
      "step": 65623
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7406188249588013,
      "learning_rate": 0.00048770319325520587,
      "loss": 2.2569,
      "step": 65624
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.8005919456481934,
      "learning_rate": 0.0004877000022639001,
      "loss": 2.4326,
      "step": 65625
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.8213056325912476,
      "learning_rate": 0.0004876968112376973,
      "loss": 2.1286,
      "step": 65626
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.087226390838623,
      "learning_rate": 0.00048769362017659815,
      "loss": 2.0401,
      "step": 65627
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.9041296243667603,
      "learning_rate": 0.0004876904290806032,
      "loss": 2.1456,
      "step": 65628
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.2537363767623901,
      "learning_rate": 0.0004876872379497131,
      "loss": 2.233,
      "step": 65629
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.9472646713256836,
      "learning_rate": 0.0004876840467839284,
      "loss": 2.1037,
      "step": 65630
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.951724648475647,
      "learning_rate": 0.00048768085558324964,
      "loss": 1.9961,
      "step": 65631
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7973999977111816,
      "learning_rate": 0.00048767766434767746,
      "loss": 2.1991,
      "step": 65632
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.0185502767562866,
      "learning_rate": 0.0004876744730772125,
      "loss": 2.3459,
      "step": 65633
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.9715756177902222,
      "learning_rate": 0.0004876712817718553,
      "loss": 2.0341,
      "step": 65634
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.8565947413444519,
      "learning_rate": 0.00048766809043160645,
      "loss": 2.1973,
      "step": 65635
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.8749676942825317,
      "learning_rate": 0.0004876648990564667,
      "loss": 2.1558,
      "step": 65636
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.0806623697280884,
      "learning_rate": 0.0004876617076464363,
      "loss": 2.3696,
      "step": 65637
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.9119154214859009,
      "learning_rate": 0.00048765851620151613,
      "loss": 2.2181,
      "step": 65638
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7083272933959961,
      "learning_rate": 0.00048765532472170675,
      "loss": 2.2049,
      "step": 65639
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.8671049475669861,
      "learning_rate": 0.00048765213320700867,
      "loss": 2.2765,
      "step": 65640
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.8907038569450378,
      "learning_rate": 0.0004876489416574224,
      "loss": 2.229,
      "step": 65641
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.8222396373748779,
      "learning_rate": 0.0004876457500729488,
      "loss": 2.2477,
      "step": 65642
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.0913891792297363,
      "learning_rate": 0.00048764255845358827,
      "loss": 2.5456,
      "step": 65643
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.848877489566803,
      "learning_rate": 0.00048763936679934144,
      "loss": 2.2791,
      "step": 65644
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.8798558115959167,
      "learning_rate": 0.0004876361751102089,
      "loss": 2.4424,
      "step": 65645
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.8246667981147766,
      "learning_rate": 0.0004876329833861913,
      "loss": 2.5647,
      "step": 65646
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7581637501716614,
      "learning_rate": 0.0004876297916272891,
      "loss": 2.2748,
      "step": 65647
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.8537386059761047,
      "learning_rate": 0.00048762659983350314,
      "loss": 2.29,
      "step": 65648
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.8274981379508972,
      "learning_rate": 0.00048762340800483366,
      "loss": 2.48,
      "step": 65649
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7625014185905457,
      "learning_rate": 0.0004876202161412815,
      "loss": 2.1843,
      "step": 65650
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7409610748291016,
      "learning_rate": 0.00048761702424284724,
      "loss": 2.411,
      "step": 65651
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7600525617599487,
      "learning_rate": 0.0004876138323095315,
      "loss": 2.4109,
      "step": 65652
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.8342705368995667,
      "learning_rate": 0.0004876106403413346,
      "loss": 2.1569,
      "step": 65653
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7714040875434875,
      "learning_rate": 0.00048760744833825756,
      "loss": 2.5104,
      "step": 65654
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.9016863107681274,
      "learning_rate": 0.0004876042563003007,
      "loss": 2.2631,
      "step": 65655
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7818988561630249,
      "learning_rate": 0.00048760106422746454,
      "loss": 2.0702,
      "step": 65656
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.920616865158081,
      "learning_rate": 0.00048759787211974994,
      "loss": 2.2277,
      "step": 65657
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7442760467529297,
      "learning_rate": 0.0004875946799771573,
      "loss": 2.4505,
      "step": 65658
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7688987255096436,
      "learning_rate": 0.00048759148779968725,
      "loss": 2.1591,
      "step": 65659
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7832061052322388,
      "learning_rate": 0.0004875882955873405,
      "loss": 2.3355,
      "step": 65660
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7687676548957825,
      "learning_rate": 0.00048758510334011746,
      "loss": 2.0476,
      "step": 65661
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.8770294189453125,
      "learning_rate": 0.0004875819110580188,
      "loss": 2.1907,
      "step": 65662
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7949879169464111,
      "learning_rate": 0.00048757871874104515,
      "loss": 2.3946,
      "step": 65663
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.097477912902832,
      "learning_rate": 0.000487575526389197,
      "loss": 2.1878,
      "step": 65664
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7675949335098267,
      "learning_rate": 0.0004875723340024751,
      "loss": 2.1968,
      "step": 65665
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.8382118940353394,
      "learning_rate": 0.00048756914158088,
      "loss": 2.0796,
      "step": 65666
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.9519418478012085,
      "learning_rate": 0.0004875659491244122,
      "loss": 2.1189,
      "step": 65667
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7877206206321716,
      "learning_rate": 0.0004875627566330723,
      "loss": 2.2284,
      "step": 65668
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.697748601436615,
      "learning_rate": 0.00048755956410686103,
      "loss": 2.3303,
      "step": 65669
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.8015878200531006,
      "learning_rate": 0.0004875563715457789,
      "loss": 2.1611,
      "step": 65670
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.8077278137207031,
      "learning_rate": 0.0004875531789498264,
      "loss": 2.609,
      "step": 65671
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.789616584777832,
      "learning_rate": 0.0004875499863190043,
      "loss": 2.2646,
      "step": 65672
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7185479402542114,
      "learning_rate": 0.00048754679365331304,
      "loss": 2.0589,
      "step": 65673
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.759689211845398,
      "learning_rate": 0.0004875436009527534,
      "loss": 2.2977,
      "step": 65674
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.8250513076782227,
      "learning_rate": 0.0004875404082173259,
      "loss": 2.2168,
      "step": 65675
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.8261075019836426,
      "learning_rate": 0.00048753721544703095,
      "loss": 2.1566,
      "step": 65676
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7544288039207458,
      "learning_rate": 0.00048753402264186936,
      "loss": 2.1959,
      "step": 65677
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.731770396232605,
      "learning_rate": 0.0004875308298018417,
      "loss": 2.1036,
      "step": 65678
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7839875221252441,
      "learning_rate": 0.00048752763692694844,
      "loss": 1.8982,
      "step": 65679
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.8778178691864014,
      "learning_rate": 0.0004875244440171903,
      "loss": 2.2069,
      "step": 65680
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.8265948295593262,
      "learning_rate": 0.0004875212510725678,
      "loss": 2.2468,
      "step": 65681
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7586690783500671,
      "learning_rate": 0.00048751805809308155,
      "loss": 2.3314,
      "step": 65682
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.8456761837005615,
      "learning_rate": 0.00048751486507873215,
      "loss": 2.4406,
      "step": 65683
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.8510322570800781,
      "learning_rate": 0.00048751167202952035,
      "loss": 2.0501,
      "step": 65684
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.8026219606399536,
      "learning_rate": 0.0004875084789454464,
      "loss": 2.3822,
      "step": 65685
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.9005615711212158,
      "learning_rate": 0.0004875052858265111,
      "loss": 1.9018,
      "step": 65686
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.8654912710189819,
      "learning_rate": 0.00048750209267271505,
      "loss": 2.2679,
      "step": 65687
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.740039587020874,
      "learning_rate": 0.0004874988994840589,
      "loss": 2.2358,
      "step": 65688
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.790076494216919,
      "learning_rate": 0.00048749570626054306,
      "loss": 2.5414,
      "step": 65689
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.705190896987915,
      "learning_rate": 0.0004874925130021684,
      "loss": 2.078,
      "step": 65690
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7549322843551636,
      "learning_rate": 0.00048748931970893513,
      "loss": 2.3012,
      "step": 65691
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.9583513736724854,
      "learning_rate": 0.0004874861263808442,
      "loss": 2.315,
      "step": 65692
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.8450031280517578,
      "learning_rate": 0.000487482933017896,
      "loss": 2.1121,
      "step": 65693
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.826653003692627,
      "learning_rate": 0.0004874797396200911,
      "loss": 2.1907,
      "step": 65694
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.8046769499778748,
      "learning_rate": 0.0004874765461874303,
      "loss": 2.1831,
      "step": 65695
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.8913255333900452,
      "learning_rate": 0.0004874733527199141,
      "loss": 2.1533,
      "step": 65696
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.752021312713623,
      "learning_rate": 0.00048747015921754295,
      "loss": 2.2409,
      "step": 65697
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7194352149963379,
      "learning_rate": 0.00048746696568031765,
      "loss": 2.1678,
      "step": 65698
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.8883070945739746,
      "learning_rate": 0.00048746377210823866,
      "loss": 2.345,
      "step": 65699
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.8488043546676636,
      "learning_rate": 0.0004874605785013067,
      "loss": 2.1695,
      "step": 65700
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.8577503561973572,
      "learning_rate": 0.00048745738485952213,
      "loss": 1.9603,
      "step": 65701
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.8019777536392212,
      "learning_rate": 0.0004874541911828857,
      "loss": 2.0543,
      "step": 65702
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.8670641183853149,
      "learning_rate": 0.0004874509974713981,
      "loss": 2.1242,
      "step": 65703
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.8438167572021484,
      "learning_rate": 0.0004874478037250598,
      "loss": 2.3623,
      "step": 65704
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7375988960266113,
      "learning_rate": 0.0004874446099438714,
      "loss": 2.2286,
      "step": 65705
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7939298748970032,
      "learning_rate": 0.0004874414161278335,
      "loss": 2.3021,
      "step": 65706
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7370960712432861,
      "learning_rate": 0.0004874382222769467,
      "loss": 2.0466,
      "step": 65707
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.6725056767463684,
      "learning_rate": 0.00048743502839121157,
      "loss": 1.9438,
      "step": 65708
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.791620135307312,
      "learning_rate": 0.00048743183447062886,
      "loss": 2.422,
      "step": 65709
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.8082186579704285,
      "learning_rate": 0.0004874286405151989,
      "loss": 2.1866,
      "step": 65710
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.8029666543006897,
      "learning_rate": 0.0004874254465249225,
      "loss": 2.2859,
      "step": 65711
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.704078733921051,
      "learning_rate": 0.0004874222524998001,
      "loss": 2.2716,
      "step": 65712
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7220430970191956,
      "learning_rate": 0.0004874190584398324,
      "loss": 2.047,
      "step": 65713
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7260389924049377,
      "learning_rate": 0.00048741586434501987,
      "loss": 2.2467,
      "step": 65714
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.8062517046928406,
      "learning_rate": 0.0004874126702153634,
      "loss": 2.1348,
      "step": 65715
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.8103724122047424,
      "learning_rate": 0.00048740947605086323,
      "loss": 2.2957,
      "step": 65716
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7585554122924805,
      "learning_rate": 0.0004874062818515201,
      "loss": 2.2337,
      "step": 65717
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.750214695930481,
      "learning_rate": 0.0004874030876173347,
      "loss": 2.4268,
      "step": 65718
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7852432131767273,
      "learning_rate": 0.0004873998933483074,
      "loss": 2.1466,
      "step": 65719
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7101444602012634,
      "learning_rate": 0.00048739669904443895,
      "loss": 2.4018,
      "step": 65720
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7630298137664795,
      "learning_rate": 0.00048739350470573,
      "loss": 2.2249,
      "step": 65721
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7353984713554382,
      "learning_rate": 0.000487390310332181,
      "loss": 2.2379,
      "step": 65722
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.6887065172195435,
      "learning_rate": 0.0004873871159237926,
      "loss": 2.3477,
      "step": 65723
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7309170961380005,
      "learning_rate": 0.00048738392148056544,
      "loss": 2.3049,
      "step": 65724
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.6786728501319885,
      "learning_rate": 0.0004873807270025,
      "loss": 2.2981,
      "step": 65725
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7710758447647095,
      "learning_rate": 0.00048737753248959706,
      "loss": 2.1671,
      "step": 65726
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7740009427070618,
      "learning_rate": 0.00048737433794185695,
      "loss": 2.275,
      "step": 65727
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.8267326354980469,
      "learning_rate": 0.0004873711433592805,
      "loss": 2.3121,
      "step": 65728
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7479855418205261,
      "learning_rate": 0.0004873679487418682,
      "loss": 2.1358,
      "step": 65729
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7251996397972107,
      "learning_rate": 0.0004873647540896207,
      "loss": 2.2133,
      "step": 65730
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.850633442401886,
      "learning_rate": 0.0004873615594025385,
      "loss": 2.1669,
      "step": 65731
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.891042947769165,
      "learning_rate": 0.0004873583646806223,
      "loss": 2.3209,
      "step": 65732
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7670798301696777,
      "learning_rate": 0.00048735516992387263,
      "loss": 2.2702,
      "step": 65733
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.859470784664154,
      "learning_rate": 0.0004873519751322901,
      "loss": 1.983,
      "step": 65734
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.9048071503639221,
      "learning_rate": 0.0004873487803058753,
      "loss": 2.2269,
      "step": 65735
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7701780796051025,
      "learning_rate": 0.0004873455854446288,
      "loss": 2.264,
      "step": 65736
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7687620520591736,
      "learning_rate": 0.0004873423905485512,
      "loss": 2.103,
      "step": 65737
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.8554542660713196,
      "learning_rate": 0.0004873391956176431,
      "loss": 2.4283,
      "step": 65738
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.6819402575492859,
      "learning_rate": 0.0004873360006519053,
      "loss": 1.9779,
      "step": 65739
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7552914619445801,
      "learning_rate": 0.00048733280565133795,
      "loss": 2.2405,
      "step": 65740
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.763020396232605,
      "learning_rate": 0.000487329610615942,
      "loss": 2.1599,
      "step": 65741
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.8397610783576965,
      "learning_rate": 0.000487326415545718,
      "loss": 2.3299,
      "step": 65742
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7836984992027283,
      "learning_rate": 0.0004873232204406664,
      "loss": 2.1092,
      "step": 65743
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7995543479919434,
      "learning_rate": 0.00048732002530078785,
      "loss": 2.259,
      "step": 65744
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7441925406455994,
      "learning_rate": 0.000487316830126083,
      "loss": 2.3108,
      "step": 65745
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7650380730628967,
      "learning_rate": 0.00048731363491655245,
      "loss": 2.2534,
      "step": 65746
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.762954592704773,
      "learning_rate": 0.0004873104396721968,
      "loss": 2.2011,
      "step": 65747
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7322445511817932,
      "learning_rate": 0.00048730724439301654,
      "loss": 2.2397,
      "step": 65748
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.6974003911018372,
      "learning_rate": 0.0004873040490790123,
      "loss": 2.0953,
      "step": 65749
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7761193513870239,
      "learning_rate": 0.00048730085373018477,
      "loss": 2.367,
      "step": 65750
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.8617779612541199,
      "learning_rate": 0.00048729765834653444,
      "loss": 2.3839,
      "step": 65751
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7098469138145447,
      "learning_rate": 0.000487294462928062,
      "loss": 2.1849,
      "step": 65752
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7135477066040039,
      "learning_rate": 0.00048729126747476786,
      "loss": 2.3516,
      "step": 65753
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7440060377120972,
      "learning_rate": 0.0004872880719866528,
      "loss": 1.9455,
      "step": 65754
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.8379747867584229,
      "learning_rate": 0.00048728487646371735,
      "loss": 2.4504,
      "step": 65755
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7472037076950073,
      "learning_rate": 0.00048728168090596215,
      "loss": 2.444,
      "step": 65756
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.8553862571716309,
      "learning_rate": 0.00048727848531338774,
      "loss": 2.4092,
      "step": 65757
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.733701765537262,
      "learning_rate": 0.0004872752896859946,
      "loss": 2.2409,
      "step": 65758
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.8249698281288147,
      "learning_rate": 0.00048727209402378357,
      "loss": 2.13,
      "step": 65759
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7912076711654663,
      "learning_rate": 0.00048726889832675506,
      "loss": 2.1381,
      "step": 65760
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.8282365798950195,
      "learning_rate": 0.0004872657025949098,
      "loss": 2.174,
      "step": 65761
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7054274678230286,
      "learning_rate": 0.0004872625068282483,
      "loss": 2.0854,
      "step": 65762
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7186214327812195,
      "learning_rate": 0.0004872593110267711,
      "loss": 2.1632,
      "step": 65763
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7911028861999512,
      "learning_rate": 0.000487256115190479,
      "loss": 2.2424,
      "step": 65764
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.8794183135032654,
      "learning_rate": 0.0004872529193193723,
      "loss": 2.0183,
      "step": 65765
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7294415235519409,
      "learning_rate": 0.00048724972341345174,
      "loss": 2.4806,
      "step": 65766
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.737790584564209,
      "learning_rate": 0.00048724652747271805,
      "loss": 2.2472,
      "step": 65767
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.8796054124832153,
      "learning_rate": 0.00048724333149717156,
      "loss": 2.1324,
      "step": 65768
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.8212388157844543,
      "learning_rate": 0.00048724013548681307,
      "loss": 2.2352,
      "step": 65769
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.6942290663719177,
      "learning_rate": 0.0004872369394416432,
      "loss": 2.2173,
      "step": 65770
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.941963791847229,
      "learning_rate": 0.0004872337433616623,
      "loss": 2.1356,
      "step": 65771
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.8303411602973938,
      "learning_rate": 0.00048723054724687115,
      "loss": 2.2642,
      "step": 65772
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7352915406227112,
      "learning_rate": 0.00048722735109727036,
      "loss": 1.8225,
      "step": 65773
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7883532643318176,
      "learning_rate": 0.0004872241549128604,
      "loss": 2.2442,
      "step": 65774
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.9193233847618103,
      "learning_rate": 0.0004872209586936419,
      "loss": 2.0699,
      "step": 65775
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.0228919982910156,
      "learning_rate": 0.0004872177624396157,
      "loss": 2.2461,
      "step": 65776
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.781395435333252,
      "learning_rate": 0.000487214566150782,
      "loss": 2.2465,
      "step": 65777
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.929370105266571,
      "learning_rate": 0.00048721136982714166,
      "loss": 2.1418,
      "step": 65778
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7850310802459717,
      "learning_rate": 0.0004872081734686951,
      "loss": 2.2208,
      "step": 65779
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.8258935213088989,
      "learning_rate": 0.0004872049770754431,
      "loss": 2.1495,
      "step": 65780
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7953270077705383,
      "learning_rate": 0.0004872017806473861,
      "loss": 2.3154,
      "step": 65781
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.8534274697303772,
      "learning_rate": 0.0004871985841845248,
      "loss": 2.1891,
      "step": 65782
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.168097734451294,
      "learning_rate": 0.0004871953876868597,
      "loss": 2.0103,
      "step": 65783
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.6937680244445801,
      "learning_rate": 0.0004871921911543915,
      "loss": 2.1936,
      "step": 65784
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.9731001257896423,
      "learning_rate": 0.0004871889945871207,
      "loss": 2.3629,
      "step": 65785
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.9548837542533875,
      "learning_rate": 0.00048718579798504795,
      "loss": 2.302,
      "step": 65786
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7649466395378113,
      "learning_rate": 0.0004871826013481738,
      "loss": 2.3919,
      "step": 65787
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7075262069702148,
      "learning_rate": 0.0004871794046764989,
      "loss": 2.408,
      "step": 65788
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.8896785378456116,
      "learning_rate": 0.0004871762079700238,
      "loss": 2.3748,
      "step": 65789
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7439350485801697,
      "learning_rate": 0.0004871730112287491,
      "loss": 2.1098,
      "step": 65790
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7078741192817688,
      "learning_rate": 0.0004871698144526754,
      "loss": 2.3273,
      "step": 65791
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.8788136839866638,
      "learning_rate": 0.00048716661764180335,
      "loss": 2.2063,
      "step": 65792
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.8631988763809204,
      "learning_rate": 0.00048716342079613344,
      "loss": 2.2938,
      "step": 65793
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7363533973693848,
      "learning_rate": 0.00048716022391566633,
      "loss": 2.1412,
      "step": 65794
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7993907332420349,
      "learning_rate": 0.00048715702700040256,
      "loss": 2.1843,
      "step": 65795
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7402511835098267,
      "learning_rate": 0.0004871538300503429,
      "loss": 2.191,
      "step": 65796
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.9014644026756287,
      "learning_rate": 0.00048715063306548767,
      "loss": 2.5655,
      "step": 65797
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.785288393497467,
      "learning_rate": 0.0004871474360458377,
      "loss": 1.9736,
      "step": 65798
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7510053515434265,
      "learning_rate": 0.0004871442389913934,
      "loss": 2.0013,
      "step": 65799
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7775483131408691,
      "learning_rate": 0.0004871410419021555,
      "loss": 2.2523,
      "step": 65800
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.8361464142799377,
      "learning_rate": 0.0004871378447781246,
      "loss": 2.3277,
      "step": 65801
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7934619188308716,
      "learning_rate": 0.00048713464761930114,
      "loss": 2.2707,
      "step": 65802
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7452647686004639,
      "learning_rate": 0.00048713145042568587,
      "loss": 2.0935,
      "step": 65803
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7738807797431946,
      "learning_rate": 0.0004871282531972793,
      "loss": 2.2239,
      "step": 65804
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7713228464126587,
      "learning_rate": 0.00048712505593408206,
      "loss": 2.3018,
      "step": 65805
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7575558423995972,
      "learning_rate": 0.0004871218586360948,
      "loss": 2.3546,
      "step": 65806
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7243227958679199,
      "learning_rate": 0.0004871186613033181,
      "loss": 2.3857,
      "step": 65807
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7796456217765808,
      "learning_rate": 0.0004871154639357523,
      "loss": 2.1705,
      "step": 65808
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7436792254447937,
      "learning_rate": 0.00048711226653339843,
      "loss": 2.1859,
      "step": 65809
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.890079915523529,
      "learning_rate": 0.00048710906909625677,
      "loss": 2.1804,
      "step": 65810
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7030432224273682,
      "learning_rate": 0.00048710587162432794,
      "loss": 2.036,
      "step": 65811
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7637006044387817,
      "learning_rate": 0.0004871026741176127,
      "loss": 2.3113,
      "step": 65812
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.8693444728851318,
      "learning_rate": 0.0004870994765761114,
      "loss": 2.3764,
      "step": 65813
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7865258455276489,
      "learning_rate": 0.0004870962789998249,
      "loss": 2.2703,
      "step": 65814
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7286983132362366,
      "learning_rate": 0.00048709308138875367,
      "loss": 2.0905,
      "step": 65815
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.8118640780448914,
      "learning_rate": 0.00048708988374289826,
      "loss": 2.2629,
      "step": 65816
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7811224460601807,
      "learning_rate": 0.0004870866860622593,
      "loss": 2.1933,
      "step": 65817
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7249118089675903,
      "learning_rate": 0.0004870834883468374,
      "loss": 2.1516,
      "step": 65818
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.782809317111969,
      "learning_rate": 0.00048708029059663316,
      "loss": 2.0745,
      "step": 65819
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7356269955635071,
      "learning_rate": 0.00048707709281164713,
      "loss": 2.0471,
      "step": 65820
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.8466070890426636,
      "learning_rate": 0.00048707389499188,
      "loss": 2.227,
      "step": 65821
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.8105471134185791,
      "learning_rate": 0.0004870706971373323,
      "loss": 2.271,
      "step": 65822
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.6969590187072754,
      "learning_rate": 0.0004870674992480046,
      "loss": 2.271,
      "step": 65823
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7077383995056152,
      "learning_rate": 0.00048706430132389747,
      "loss": 2.0612,
      "step": 65824
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7504881024360657,
      "learning_rate": 0.00048706110336501165,
      "loss": 2.2442,
      "step": 65825
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7006335258483887,
      "learning_rate": 0.00048705790537134755,
      "loss": 2.3663,
      "step": 65826
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7164832353591919,
      "learning_rate": 0.0004870547073429059,
      "loss": 2.2861,
      "step": 65827
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7077950239181519,
      "learning_rate": 0.0004870515092796872,
      "loss": 2.2362,
      "step": 65828
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.8467288017272949,
      "learning_rate": 0.0004870483111816922,
      "loss": 2.3427,
      "step": 65829
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7439658045768738,
      "learning_rate": 0.00048704511304892127,
      "loss": 2.2669,
      "step": 65830
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7237943410873413,
      "learning_rate": 0.0004870419148813752,
      "loss": 2.2325,
      "step": 65831
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7480741143226624,
      "learning_rate": 0.0004870387166790545,
      "loss": 2.0877,
      "step": 65832
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7786546945571899,
      "learning_rate": 0.00048703551844195975,
      "loss": 2.3125,
      "step": 65833
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7168574333190918,
      "learning_rate": 0.0004870323201700916,
      "loss": 2.0943,
      "step": 65834
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.6726166009902954,
      "learning_rate": 0.0004870291218634506,
      "loss": 2.5093,
      "step": 65835
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.8009937405586243,
      "learning_rate": 0.0004870259235220373,
      "loss": 2.2349,
      "step": 65836
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7466238141059875,
      "learning_rate": 0.00048702272514585237,
      "loss": 2.1907,
      "step": 65837
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.6898496150970459,
      "learning_rate": 0.0004870195267348964,
      "loss": 2.1957,
      "step": 65838
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.813771665096283,
      "learning_rate": 0.00048701632828917,
      "loss": 2.1373,
      "step": 65839
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.785724937915802,
      "learning_rate": 0.0004870131298086737,
      "loss": 2.1331,
      "step": 65840
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7058380246162415,
      "learning_rate": 0.0004870099312934081,
      "loss": 2.21,
      "step": 65841
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7811790704727173,
      "learning_rate": 0.0004870067327433739,
      "loss": 2.2145,
      "step": 65842
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7279791831970215,
      "learning_rate": 0.0004870035341585716,
      "loss": 2.196,
      "step": 65843
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.8424235582351685,
      "learning_rate": 0.0004870003355390018,
      "loss": 2.2364,
      "step": 65844
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7561702728271484,
      "learning_rate": 0.0004869971368846651,
      "loss": 2.3975,
      "step": 65845
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.782014787197113,
      "learning_rate": 0.00048699393819556216,
      "loss": 2.1826,
      "step": 65846
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7595255374908447,
      "learning_rate": 0.00048699073947169343,
      "loss": 2.3427,
      "step": 65847
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7910444140434265,
      "learning_rate": 0.0004869875407130596,
      "loss": 2.0854,
      "step": 65848
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7424771189689636,
      "learning_rate": 0.0004869843419196613,
      "loss": 2.3237,
      "step": 65849
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7533845901489258,
      "learning_rate": 0.0004869811430914991,
      "loss": 2.3751,
      "step": 65850
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7874385118484497,
      "learning_rate": 0.00048697794422857346,
      "loss": 2.2411,
      "step": 65851
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.8944454789161682,
      "learning_rate": 0.00048697474533088516,
      "loss": 2.2412,
      "step": 65852
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7713539004325867,
      "learning_rate": 0.00048697154639843474,
      "loss": 2.0075,
      "step": 65853
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.8162487745285034,
      "learning_rate": 0.0004869683474312228,
      "loss": 2.1773,
      "step": 65854
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7542466521263123,
      "learning_rate": 0.00048696514842924986,
      "loss": 2.3257,
      "step": 65855
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7678845524787903,
      "learning_rate": 0.00048696194939251664,
      "loss": 2.2405,
      "step": 65856
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7857107520103455,
      "learning_rate": 0.0004869587503210236,
      "loss": 2.3838,
      "step": 65857
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7464695572853088,
      "learning_rate": 0.00048695555121477143,
      "loss": 2.0597,
      "step": 65858
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.0147639513015747,
      "learning_rate": 0.00048695235207376063,
      "loss": 2.1486,
      "step": 65859
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.9090452790260315,
      "learning_rate": 0.00048694915289799193,
      "loss": 1.9943,
      "step": 65860
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.74868243932724,
      "learning_rate": 0.0004869459536874658,
      "loss": 2.3258,
      "step": 65861
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.9988166093826294,
      "learning_rate": 0.000486942754442183,
      "loss": 2.3449,
      "step": 65862
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.9025688171386719,
      "learning_rate": 0.00048693955516214394,
      "loss": 2.2921,
      "step": 65863
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.704380452632904,
      "learning_rate": 0.0004869363558473493,
      "loss": 2.2166,
      "step": 65864
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.0657087564468384,
      "learning_rate": 0.00048693315649779963,
      "loss": 2.4337,
      "step": 65865
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.086118459701538,
      "learning_rate": 0.0004869299571134956,
      "loss": 2.2373,
      "step": 65866
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.8418554067611694,
      "learning_rate": 0.0004869267576944378,
      "loss": 2.3726,
      "step": 65867
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.9099916815757751,
      "learning_rate": 0.0004869235582406267,
      "loss": 2.0519,
      "step": 65868
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.5650078058242798,
      "learning_rate": 0.00048692035875206295,
      "loss": 2.3959,
      "step": 65869
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7366889119148254,
      "learning_rate": 0.0004869171592287473,
      "loss": 2.1069,
      "step": 65870
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.8280970454216003,
      "learning_rate": 0.0004869139596706801,
      "loss": 2.0403,
      "step": 65871
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.9491407871246338,
      "learning_rate": 0.00048691076007786215,
      "loss": 2.2726,
      "step": 65872
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.777587890625,
      "learning_rate": 0.000486907560450294,
      "loss": 2.0364,
      "step": 65873
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.8340244889259338,
      "learning_rate": 0.00048690436078797607,
      "loss": 2.0706,
      "step": 65874
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.840771496295929,
      "learning_rate": 0.0004869011610909092,
      "loss": 2.1077,
      "step": 65875
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.6826299428939819,
      "learning_rate": 0.0004868979613590938,
      "loss": 2.081,
      "step": 65876
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.711945652961731,
      "learning_rate": 0.0004868947615925306,
      "loss": 2.2538,
      "step": 65877
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.855384886264801,
      "learning_rate": 0.0004868915617912201,
      "loss": 2.4682,
      "step": 65878
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.735637366771698,
      "learning_rate": 0.000486888361955163,
      "loss": 2.3647,
      "step": 65879
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7337685823440552,
      "learning_rate": 0.0004868851620843597,
      "loss": 2.3167,
      "step": 65880
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.9148088693618774,
      "learning_rate": 0.00048688196217881106,
      "loss": 2.0802,
      "step": 65881
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7296618819236755,
      "learning_rate": 0.00048687876223851747,
      "loss": 2.2036,
      "step": 65882
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.8110126852989197,
      "learning_rate": 0.0004868755622634795,
      "loss": 2.262,
      "step": 65883
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.8223816156387329,
      "learning_rate": 0.00048687236225369797,
      "loss": 2.2397,
      "step": 65884
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7666366100311279,
      "learning_rate": 0.0004868691622091733,
      "loss": 2.1251,
      "step": 65885
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7600442171096802,
      "learning_rate": 0.0004868659621299061,
      "loss": 2.2412,
      "step": 65886
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.792872428894043,
      "learning_rate": 0.000486862762015897,
      "loss": 2.2815,
      "step": 65887
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7743217945098877,
      "learning_rate": 0.0004868595618671467,
      "loss": 2.2242,
      "step": 65888
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7545608282089233,
      "learning_rate": 0.0004868563616836555,
      "loss": 2.3698,
      "step": 65889
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7279228568077087,
      "learning_rate": 0.00048685316146542424,
      "loss": 2.2359,
      "step": 65890
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.9712733626365662,
      "learning_rate": 0.00048684996121245346,
      "loss": 2.1731,
      "step": 65891
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.088118076324463,
      "learning_rate": 0.0004868467609247438,
      "loss": 2.2024,
      "step": 65892
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7617495656013489,
      "learning_rate": 0.0004868435606022957,
      "loss": 2.2043,
      "step": 65893
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.8614217042922974,
      "learning_rate": 0.0004868403602451099,
      "loss": 2.2957,
      "step": 65894
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.9725929498672485,
      "learning_rate": 0.00048683715985318695,
      "loss": 2.2841,
      "step": 65895
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7461194396018982,
      "learning_rate": 0.0004868339594265275,
      "loss": 2.267,
      "step": 65896
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7112942337989807,
      "learning_rate": 0.00048683075896513195,
      "loss": 2.2009,
      "step": 65897
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7274386882781982,
      "learning_rate": 0.0004868275584690012,
      "loss": 2.2196,
      "step": 65898
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.9333754777908325,
      "learning_rate": 0.0004868243579381356,
      "loss": 2.0741,
      "step": 65899
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.8727854490280151,
      "learning_rate": 0.00048682115737253587,
      "loss": 2.0366,
      "step": 65900
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.8326034545898438,
      "learning_rate": 0.0004868179567722025,
      "loss": 2.0325,
      "step": 65901
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.0044673681259155,
      "learning_rate": 0.0004868147561371361,
      "loss": 2.3682,
      "step": 65902
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.6981108784675598,
      "learning_rate": 0.0004868115554673374,
      "loss": 2.1016,
      "step": 65903
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.8810930848121643,
      "learning_rate": 0.00048680835476280695,
      "loss": 2.195,
      "step": 65904
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7011376619338989,
      "learning_rate": 0.0004868051540235452,
      "loss": 2.1401,
      "step": 65905
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.8150160908699036,
      "learning_rate": 0.0004868019532495528,
      "loss": 2.2535,
      "step": 65906
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.8457236289978027,
      "learning_rate": 0.00048679875244083055,
      "loss": 2.1098,
      "step": 65907
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.9111199975013733,
      "learning_rate": 0.00048679555159737877,
      "loss": 2.3435,
      "step": 65908
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.6959871649742126,
      "learning_rate": 0.0004867923507191982,
      "loss": 2.1573,
      "step": 65909
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7722300887107849,
      "learning_rate": 0.00048678914980628945,
      "loss": 2.365,
      "step": 65910
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7349993586540222,
      "learning_rate": 0.000486785948858653,
      "loss": 2.2652,
      "step": 65911
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7704023718833923,
      "learning_rate": 0.00048678274787628955,
      "loss": 2.0567,
      "step": 65912
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.8022553324699402,
      "learning_rate": 0.0004867795468591997,
      "loss": 2.1843,
      "step": 65913
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.6965587735176086,
      "learning_rate": 0.00048677634580738394,
      "loss": 2.1984,
      "step": 65914
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7874884605407715,
      "learning_rate": 0.0004867731447208429,
      "loss": 2.1099,
      "step": 65915
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.777126133441925,
      "learning_rate": 0.0004867699435995774,
      "loss": 2.1313,
      "step": 65916
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.8004518151283264,
      "learning_rate": 0.00048676674244358764,
      "loss": 2.1934,
      "step": 65917
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.8072488903999329,
      "learning_rate": 0.00048676354125287447,
      "loss": 2.1603,
      "step": 65918
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7936784029006958,
      "learning_rate": 0.0004867603400274384,
      "loss": 2.1789,
      "step": 65919
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7758207321166992,
      "learning_rate": 0.00048675713876728015,
      "loss": 2.3521,
      "step": 65920
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.872379720211029,
      "learning_rate": 0.0004867539374724001,
      "loss": 2.0314,
      "step": 65921
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.8141086101531982,
      "learning_rate": 0.0004867507361427991,
      "loss": 2.1804,
      "step": 65922
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.8982181549072266,
      "learning_rate": 0.00048674753477847757,
      "loss": 2.4358,
      "step": 65923
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.8087480664253235,
      "learning_rate": 0.00048674433337943615,
      "loss": 2.1594,
      "step": 65924
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7589192986488342,
      "learning_rate": 0.00048674113194567537,
      "loss": 2.3181,
      "step": 65925
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.726265549659729,
      "learning_rate": 0.000486737930477196,
      "loss": 2.1732,
      "step": 65926
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7576211094856262,
      "learning_rate": 0.00048673472897399836,
      "loss": 2.2358,
      "step": 65927
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7729759812355042,
      "learning_rate": 0.0004867315274360833,
      "loss": 2.3621,
      "step": 65928
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7618862986564636,
      "learning_rate": 0.00048672832586345137,
      "loss": 2.363,
      "step": 65929
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.0067739486694336,
      "learning_rate": 0.0004867251242561031,
      "loss": 2.177,
      "step": 65930
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7536956667900085,
      "learning_rate": 0.00048672192261403906,
      "loss": 2.1102,
      "step": 65931
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7603738307952881,
      "learning_rate": 0.0004867187209372599,
      "loss": 2.2697,
      "step": 65932
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.0043269395828247,
      "learning_rate": 0.00048671551922576626,
      "loss": 2.0843,
      "step": 65933
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.9061493277549744,
      "learning_rate": 0.00048671231747955863,
      "loss": 2.2708,
      "step": 65934
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.9043453931808472,
      "learning_rate": 0.00048670911569863773,
      "loss": 2.4237,
      "step": 65935
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.811877429485321,
      "learning_rate": 0.00048670591388300403,
      "loss": 2.1089,
      "step": 65936
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.9178934097290039,
      "learning_rate": 0.0004867027120326581,
      "loss": 2.3102,
      "step": 65937
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.9887132048606873,
      "learning_rate": 0.00048669951014760075,
      "loss": 2.0212,
      "step": 65938
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7989515066146851,
      "learning_rate": 0.00048669630822783237,
      "loss": 1.9736,
      "step": 65939
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7793037295341492,
      "learning_rate": 0.00048669310627335357,
      "loss": 2.2788,
      "step": 65940
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.8880059123039246,
      "learning_rate": 0.0004866899042841651,
      "loss": 2.1074,
      "step": 65941
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.8022071719169617,
      "learning_rate": 0.00048668670226026736,
      "loss": 2.3779,
      "step": 65942
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.8368610143661499,
      "learning_rate": 0.0004866835002016611,
      "loss": 2.1735,
      "step": 65943
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.783311128616333,
      "learning_rate": 0.0004866802981083468,
      "loss": 2.4554,
      "step": 65944
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7268558740615845,
      "learning_rate": 0.0004866770959803251,
      "loss": 2.0975,
      "step": 65945
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.9344990253448486,
      "learning_rate": 0.00048667389381759667,
      "loss": 2.388,
      "step": 65946
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.93437260389328,
      "learning_rate": 0.000486670691620162,
      "loss": 2.2762,
      "step": 65947
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7415173649787903,
      "learning_rate": 0.00048666748938802173,
      "loss": 2.075,
      "step": 65948
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.0252031087875366,
      "learning_rate": 0.0004866642871211765,
      "loss": 2.2189,
      "step": 65949
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7912513613700867,
      "learning_rate": 0.0004866610848196268,
      "loss": 2.3009,
      "step": 65950
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7762351036071777,
      "learning_rate": 0.0004866578824833732,
      "loss": 2.4239,
      "step": 65951
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7740069627761841,
      "learning_rate": 0.0004866546801124165,
      "loss": 2.1448,
      "step": 65952
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7794298529624939,
      "learning_rate": 0.00048665147770675724,
      "loss": 2.0903,
      "step": 65953
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.8851615786552429,
      "learning_rate": 0.0004866482752663958,
      "loss": 2.1137,
      "step": 65954
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7318558096885681,
      "learning_rate": 0.000486645072791333,
      "loss": 2.4436,
      "step": 65955
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7582915425300598,
      "learning_rate": 0.00048664187028156925,
      "loss": 2.1384,
      "step": 65956
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.8033658862113953,
      "learning_rate": 0.00048663866773710537,
      "loss": 2.3528,
      "step": 65957
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7161094546318054,
      "learning_rate": 0.0004866354651579418,
      "loss": 2.1634,
      "step": 65958
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7046196460723877,
      "learning_rate": 0.00048663226254407917,
      "loss": 2.1272,
      "step": 65959
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.8101916313171387,
      "learning_rate": 0.00048662905989551805,
      "loss": 2.3835,
      "step": 65960
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7568742036819458,
      "learning_rate": 0.00048662585721225915,
      "loss": 2.2791,
      "step": 65961
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7377332448959351,
      "learning_rate": 0.0004866226544943029,
      "loss": 2.145,
      "step": 65962
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7847214341163635,
      "learning_rate": 0.00048661945174165,
      "loss": 2.4254,
      "step": 65963
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7343407869338989,
      "learning_rate": 0.000486616248954301,
      "loss": 2.1925,
      "step": 65964
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7645509839057922,
      "learning_rate": 0.00048661304613225654,
      "loss": 2.3367,
      "step": 65965
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7521565556526184,
      "learning_rate": 0.0004866098432755172,
      "loss": 1.9489,
      "step": 65966
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.705531895160675,
      "learning_rate": 0.0004866066403840836,
      "loss": 2.2103,
      "step": 65967
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.8266826272010803,
      "learning_rate": 0.0004866034374579562,
      "loss": 2.4461,
      "step": 65968
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.6939268112182617,
      "learning_rate": 0.00048660023449713575,
      "loss": 2.2269,
      "step": 65969
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.6761490702629089,
      "learning_rate": 0.00048659703150162283,
      "loss": 2.1181,
      "step": 65970
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.6917221546173096,
      "learning_rate": 0.000486593828471418,
      "loss": 2.1509,
      "step": 65971
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7150692343711853,
      "learning_rate": 0.00048659062540652183,
      "loss": 2.0034,
      "step": 65972
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7526944279670715,
      "learning_rate": 0.00048658742230693494,
      "loss": 2.2024,
      "step": 65973
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.901107132434845,
      "learning_rate": 0.0004865842191726579,
      "loss": 2.3821,
      "step": 65974
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7780025601387024,
      "learning_rate": 0.00048658101600369136,
      "loss": 2.2673,
      "step": 65975
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.8091381192207336,
      "learning_rate": 0.0004865778128000359,
      "loss": 2.396,
      "step": 65976
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.8375236392021179,
      "learning_rate": 0.00048657460956169206,
      "loss": 2.4568,
      "step": 65977
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.906393826007843,
      "learning_rate": 0.0004865714062886605,
      "loss": 2.1508,
      "step": 65978
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7773898243904114,
      "learning_rate": 0.00048656820298094184,
      "loss": 2.3051,
      "step": 65979
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7179341912269592,
      "learning_rate": 0.0004865649996385366,
      "loss": 2.0159,
      "step": 65980
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7505097985267639,
      "learning_rate": 0.0004865617962614454,
      "loss": 2.3155,
      "step": 65981
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7389591932296753,
      "learning_rate": 0.0004865585928496688,
      "loss": 2.208,
      "step": 65982
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7281208634376526,
      "learning_rate": 0.0004865553894032075,
      "loss": 2.2454,
      "step": 65983
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7637880444526672,
      "learning_rate": 0.00048655218592206205,
      "loss": 2.0614,
      "step": 65984
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.8075168132781982,
      "learning_rate": 0.00048654898240623287,
      "loss": 2.2449,
      "step": 65985
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7959195971488953,
      "learning_rate": 0.00048654577885572087,
      "loss": 2.2796,
      "step": 65986
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7112605571746826,
      "learning_rate": 0.0004865425752705265,
      "loss": 2.0184,
      "step": 65987
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.8103986382484436,
      "learning_rate": 0.00048653937165065017,
      "loss": 2.3061,
      "step": 65988
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7029966115951538,
      "learning_rate": 0.0004865361679960928,
      "loss": 2.2362,
      "step": 65989
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7282195687294006,
      "learning_rate": 0.00048653296430685477,
      "loss": 2.3203,
      "step": 65990
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.9469476938247681,
      "learning_rate": 0.00048652976058293683,
      "loss": 2.1093,
      "step": 65991
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7342991828918457,
      "learning_rate": 0.0004865265568243394,
      "loss": 2.0568,
      "step": 65992
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7983590364456177,
      "learning_rate": 0.0004865233530310632,
      "loss": 2.1693,
      "step": 65993
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7783660888671875,
      "learning_rate": 0.00048652014920310874,
      "loss": 2.082,
      "step": 65994
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.8576685786247253,
      "learning_rate": 0.00048651694534047673,
      "loss": 2.2084,
      "step": 65995
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7548394203186035,
      "learning_rate": 0.0004865137414431677,
      "loss": 2.0912,
      "step": 65996
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7453027367591858,
      "learning_rate": 0.0004865105375111822,
      "loss": 2.129,
      "step": 65997
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7109788060188293,
      "learning_rate": 0.0004865073335445209,
      "loss": 2.2743,
      "step": 65998
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.745347797870636,
      "learning_rate": 0.00048650412954318436,
      "loss": 2.1863,
      "step": 65999
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7838717699050903,
      "learning_rate": 0.0004865009255071731,
      "loss": 2.3271,
      "step": 66000
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7794375419616699,
      "learning_rate": 0.00048649772143648793,
      "loss": 2.3666,
      "step": 66001
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.8871501684188843,
      "learning_rate": 0.0004864945173311293,
      "loss": 2.3363,
      "step": 66002
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.8531023263931274,
      "learning_rate": 0.0004864913131910977,
      "loss": 2.5251,
      "step": 66003
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7853518724441528,
      "learning_rate": 0.000486488109016394,
      "loss": 2.1986,
      "step": 66004
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7674315571784973,
      "learning_rate": 0.00048648490480701845,
      "loss": 2.1588,
      "step": 66005
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.805169939994812,
      "learning_rate": 0.000486481700562972,
      "loss": 2.2906,
      "step": 66006
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7108801603317261,
      "learning_rate": 0.0004864784962842551,
      "loss": 2.0839,
      "step": 66007
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7433289289474487,
      "learning_rate": 0.0004864752919708682,
      "loss": 2.2902,
      "step": 66008
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7178884148597717,
      "learning_rate": 0.000486472087622812,
      "loss": 2.3894,
      "step": 66009
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.8269380927085876,
      "learning_rate": 0.00048646888324008723,
      "loss": 2.4006,
      "step": 66010
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7925464510917664,
      "learning_rate": 0.00048646567882269434,
      "loss": 2.3509,
      "step": 66011
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.723993718624115,
      "learning_rate": 0.0004864624743706339,
      "loss": 2.3173,
      "step": 66012
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7959614396095276,
      "learning_rate": 0.0004864592698839067,
      "loss": 2.1638,
      "step": 66013
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7912447452545166,
      "learning_rate": 0.0004864560653625131,
      "loss": 2.1771,
      "step": 66014
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7493796944618225,
      "learning_rate": 0.00048645286080645385,
      "loss": 2.2863,
      "step": 66015
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7674622535705566,
      "learning_rate": 0.0004864496562157295,
      "loss": 2.1849,
      "step": 66016
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.8601896166801453,
      "learning_rate": 0.00048644645159034056,
      "loss": 2.045,
      "step": 66017
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.839281439781189,
      "learning_rate": 0.00048644324693028777,
      "loss": 2.2569,
      "step": 66018
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7888830900192261,
      "learning_rate": 0.00048644004223557163,
      "loss": 2.1874,
      "step": 66019
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7817237973213196,
      "learning_rate": 0.0004864368375061928,
      "loss": 2.1743,
      "step": 66020
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7358703017234802,
      "learning_rate": 0.0004864336327421518,
      "loss": 2.3647,
      "step": 66021
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7789011597633362,
      "learning_rate": 0.00048643042794344933,
      "loss": 2.1897,
      "step": 66022
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.8126747012138367,
      "learning_rate": 0.00048642722311008585,
      "loss": 2.4341,
      "step": 66023
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7163922190666199,
      "learning_rate": 0.0004864240182420621,
      "loss": 2.2756,
      "step": 66024
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7994298338890076,
      "learning_rate": 0.0004864208133393786,
      "loss": 2.1727,
      "step": 66025
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.8034640550613403,
      "learning_rate": 0.0004864176084020359,
      "loss": 2.0428,
      "step": 66026
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7542587518692017,
      "learning_rate": 0.0004864144034300346,
      "loss": 2.188,
      "step": 66027
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.6979730129241943,
      "learning_rate": 0.00048641119842337546,
      "loss": 2.0461,
      "step": 66028
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7459173202514648,
      "learning_rate": 0.0004864079933820589,
      "loss": 2.2474,
      "step": 66029
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7201327681541443,
      "learning_rate": 0.00048640478830608557,
      "loss": 2.3362,
      "step": 66030
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.752444863319397,
      "learning_rate": 0.00048640158319545614,
      "loss": 2.1966,
      "step": 66031
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7838016748428345,
      "learning_rate": 0.0004863983780501711,
      "loss": 2.3829,
      "step": 66032
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.8757978081703186,
      "learning_rate": 0.00048639517287023113,
      "loss": 2.1444,
      "step": 66033
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7369980812072754,
      "learning_rate": 0.00048639196765563663,
      "loss": 2.2187,
      "step": 66034
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.058377981185913,
      "learning_rate": 0.0004863887624063885,
      "loss": 2.1766,
      "step": 66035
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.8306912779808044,
      "learning_rate": 0.00048638555712248707,
      "loss": 2.1515,
      "step": 66036
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7942389845848083,
      "learning_rate": 0.0004863823518039331,
      "loss": 2.4151,
      "step": 66037
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7883551716804504,
      "learning_rate": 0.00048637914645072714,
      "loss": 2.3244,
      "step": 66038
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.8365297913551331,
      "learning_rate": 0.00048637594106286977,
      "loss": 2.1938,
      "step": 66039
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.824275016784668,
      "learning_rate": 0.0004863727356403616,
      "loss": 2.333,
      "step": 66040
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7239817976951599,
      "learning_rate": 0.0004863695301832032,
      "loss": 2.1642,
      "step": 66041
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7549415826797485,
      "learning_rate": 0.0004863663246913952,
      "loss": 2.2249,
      "step": 66042
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7287503480911255,
      "learning_rate": 0.00048636311916493814,
      "loss": 2.2234,
      "step": 66043
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7583742737770081,
      "learning_rate": 0.00048635991360383273,
      "loss": 2.4907,
      "step": 66044
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7829383015632629,
      "learning_rate": 0.00048635670800807943,
      "loss": 2.2266,
      "step": 66045
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.8207685351371765,
      "learning_rate": 0.000486353502377679,
      "loss": 2.1171,
      "step": 66046
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7111896276473999,
      "learning_rate": 0.0004863502967126318,
      "loss": 2.2457,
      "step": 66047
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.6962405443191528,
      "learning_rate": 0.00048634709101293865,
      "loss": 1.9399,
      "step": 66048
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.8817258477210999,
      "learning_rate": 0.0004863438852786,
      "loss": 2.1808,
      "step": 66049
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7798528671264648,
      "learning_rate": 0.0004863406795096166,
      "loss": 2.2423,
      "step": 66050
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7329267263412476,
      "learning_rate": 0.0004863374737059889,
      "loss": 2.3152,
      "step": 66051
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7962097525596619,
      "learning_rate": 0.00048633426786771753,
      "loss": 2.2066,
      "step": 66052
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.8104537129402161,
      "learning_rate": 0.0004863310619948031,
      "loss": 2.1061,
      "step": 66053
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7489593029022217,
      "learning_rate": 0.0004863278560872462,
      "loss": 2.2403,
      "step": 66054
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.8281257748603821,
      "learning_rate": 0.00048632465014504745,
      "loss": 2.3085,
      "step": 66055
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7648028135299683,
      "learning_rate": 0.00048632144416820743,
      "loss": 2.0713,
      "step": 66056
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7322072982788086,
      "learning_rate": 0.00048631823815672683,
      "loss": 2.3142,
      "step": 66057
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.8163362145423889,
      "learning_rate": 0.00048631503211060597,
      "loss": 2.3471,
      "step": 66058
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.737602949142456,
      "learning_rate": 0.0004863118260298458,
      "loss": 2.3439,
      "step": 66059
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.9360785484313965,
      "learning_rate": 0.00048630861991444656,
      "loss": 2.1862,
      "step": 66060
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.8654859662055969,
      "learning_rate": 0.0004863054137644092,
      "loss": 2.3976,
      "step": 66061
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7462168335914612,
      "learning_rate": 0.00048630220757973403,
      "loss": 2.2579,
      "step": 66062
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.9689273238182068,
      "learning_rate": 0.00048629900136042185,
      "loss": 2.4415,
      "step": 66063
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.016109585762024,
      "learning_rate": 0.0004862957951064731,
      "loss": 2.2893,
      "step": 66064
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7292582988739014,
      "learning_rate": 0.0004862925888178885,
      "loss": 2.3525,
      "step": 66065
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.8495808243751526,
      "learning_rate": 0.00048628938249466863,
      "loss": 2.201,
      "step": 66066
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.8259532451629639,
      "learning_rate": 0.00048628617613681395,
      "loss": 2.3203,
      "step": 66067
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7990220785140991,
      "learning_rate": 0.0004862829697443252,
      "loss": 2.236,
      "step": 66068
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7747520208358765,
      "learning_rate": 0.00048627976331720285,
      "loss": 2.0911,
      "step": 66069
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.8132026791572571,
      "learning_rate": 0.00048627655685544766,
      "loss": 2.2329,
      "step": 66070
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7258013486862183,
      "learning_rate": 0.0004862733503590602,
      "loss": 2.1625,
      "step": 66071
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7333254814147949,
      "learning_rate": 0.0004862701438280409,
      "loss": 2.227,
      "step": 66072
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.8942261338233948,
      "learning_rate": 0.0004862669372623905,
      "loss": 2.1795,
      "step": 66073
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7250292897224426,
      "learning_rate": 0.0004862637306621096,
      "loss": 2.1305,
      "step": 66074
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.845309317111969,
      "learning_rate": 0.00048626052402719866,
      "loss": 2.2184,
      "step": 66075
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.8754007816314697,
      "learning_rate": 0.0004862573173576585,
      "loss": 2.2218,
      "step": 66076
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7938337922096252,
      "learning_rate": 0.0004862541106534895,
      "loss": 2.046,
      "step": 66077
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.845607578754425,
      "learning_rate": 0.0004862509039146923,
      "loss": 2.3016,
      "step": 66078
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.8680567741394043,
      "learning_rate": 0.00048624769714126766,
      "loss": 2.4523,
      "step": 66079
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.8765818476676941,
      "learning_rate": 0.000486244490333216,
      "loss": 2.3443,
      "step": 66080
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7076078057289124,
      "learning_rate": 0.00048624128349053795,
      "loss": 2.264,
      "step": 66081
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.720020592212677,
      "learning_rate": 0.00048623807661323415,
      "loss": 2.1591,
      "step": 66082
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.77520751953125,
      "learning_rate": 0.0004862348697013052,
      "loss": 2.1372,
      "step": 66083
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7324134707450867,
      "learning_rate": 0.00048623166275475163,
      "loss": 2.2832,
      "step": 66084
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.8782988786697388,
      "learning_rate": 0.0004862284557735741,
      "loss": 2.5377,
      "step": 66085
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.708701491355896,
      "learning_rate": 0.00048622524875777324,
      "loss": 2.1236,
      "step": 66086
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7422024011611938,
      "learning_rate": 0.00048622204170734956,
      "loss": 2.1472,
      "step": 66087
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.8492929935455322,
      "learning_rate": 0.00048621883462230363,
      "loss": 2.2583,
      "step": 66088
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7322603464126587,
      "learning_rate": 0.0004862156275026362,
      "loss": 2.231,
      "step": 66089
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7746382355690002,
      "learning_rate": 0.00048621242034834763,
      "loss": 1.8199,
      "step": 66090
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7929380536079407,
      "learning_rate": 0.0004862092131594388,
      "loss": 2.2858,
      "step": 66091
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.837422251701355,
      "learning_rate": 0.00048620600593591006,
      "loss": 2.3575,
      "step": 66092
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7429512739181519,
      "learning_rate": 0.0004862027986777622,
      "loss": 2.1425,
      "step": 66093
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.764806866645813,
      "learning_rate": 0.0004861995913849956,
      "loss": 2.2342,
      "step": 66094
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.8155049681663513,
      "learning_rate": 0.00048619638405761106,
      "loss": 2.3421,
      "step": 66095
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7971125841140747,
      "learning_rate": 0.00048619317669560915,
      "loss": 2.0785,
      "step": 66096
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.6868628263473511,
      "learning_rate": 0.0004861899692989903,
      "loss": 2.2415,
      "step": 66097
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.757429838180542,
      "learning_rate": 0.0004861867618677553,
      "loss": 2.297,
      "step": 66098
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7691226601600647,
      "learning_rate": 0.00048618355440190465,
      "loss": 2.4158,
      "step": 66099
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7447051405906677,
      "learning_rate": 0.0004861803469014389,
      "loss": 2.4095,
      "step": 66100
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.791800320148468,
      "learning_rate": 0.0004861771393663588,
      "loss": 2.1832,
      "step": 66101
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7825118899345398,
      "learning_rate": 0.0004861739317966648,
      "loss": 2.387,
      "step": 66102
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7178428769111633,
      "learning_rate": 0.0004861707241923576,
      "loss": 2.1819,
      "step": 66103
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7901391983032227,
      "learning_rate": 0.00048616751655343763,
      "loss": 2.0124,
      "step": 66104
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.8971300721168518,
      "learning_rate": 0.0004861643088799058,
      "loss": 2.0342,
      "step": 66105
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.8318144679069519,
      "learning_rate": 0.00048616110117176236,
      "loss": 2.3856,
      "step": 66106
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.6912404298782349,
      "learning_rate": 0.00048615789342900805,
      "loss": 2.2234,
      "step": 66107
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7427268028259277,
      "learning_rate": 0.0004861546856516435,
      "loss": 2.2607,
      "step": 66108
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7610200643539429,
      "learning_rate": 0.00048615147783966935,
      "loss": 2.325,
      "step": 66109
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.826178789138794,
      "learning_rate": 0.00048614826999308604,
      "loss": 2.1183,
      "step": 66110
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.8170717358589172,
      "learning_rate": 0.00048614506211189435,
      "loss": 2.371,
      "step": 66111
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.8401796221733093,
      "learning_rate": 0.0004861418541960947,
      "loss": 2.2754,
      "step": 66112
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.8492593765258789,
      "learning_rate": 0.00048613864624568777,
      "loss": 2.477,
      "step": 66113
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7792153358459473,
      "learning_rate": 0.00048613543826067423,
      "loss": 2.155,
      "step": 66114
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7572984099388123,
      "learning_rate": 0.00048613223024105453,
      "loss": 2.3237,
      "step": 66115
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.8297052383422852,
      "learning_rate": 0.0004861290221868293,
      "loss": 2.318,
      "step": 66116
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7013244032859802,
      "learning_rate": 0.0004861258140979993,
      "loss": 1.9839,
      "step": 66117
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7299823760986328,
      "learning_rate": 0.00048612260597456487,
      "loss": 2.1872,
      "step": 66118
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7412209510803223,
      "learning_rate": 0.00048611939781652676,
      "loss": 2.388,
      "step": 66119
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.811964213848114,
      "learning_rate": 0.00048611618962388564,
      "loss": 2.2475,
      "step": 66120
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.6658905148506165,
      "learning_rate": 0.0004861129813966419,
      "loss": 1.9317,
      "step": 66121
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7487893104553223,
      "learning_rate": 0.00048610977313479623,
      "loss": 2.0009,
      "step": 66122
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7984074950218201,
      "learning_rate": 0.0004861065648383493,
      "loss": 2.0864,
      "step": 66123
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7809644937515259,
      "learning_rate": 0.00048610335650730154,
      "loss": 2.2742,
      "step": 66124
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7713179588317871,
      "learning_rate": 0.0004861001481416537,
      "loss": 2.48,
      "step": 66125
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7725412845611572,
      "learning_rate": 0.00048609693974140653,
      "loss": 2.1239,
      "step": 66126
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.8805826306343079,
      "learning_rate": 0.00048609373130656017,
      "loss": 2.0577,
      "step": 66127
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7394745945930481,
      "learning_rate": 0.0004860905228371155,
      "loss": 2.2005,
      "step": 66128
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7838651537895203,
      "learning_rate": 0.00048608731433307325,
      "loss": 2.2361,
      "step": 66129
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7068682909011841,
      "learning_rate": 0.0004860841057944338,
      "loss": 2.5438,
      "step": 66130
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.8267816305160522,
      "learning_rate": 0.00048608089722119774,
      "loss": 2.0772,
      "step": 66131
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.711854875087738,
      "learning_rate": 0.0004860776886133658,
      "loss": 2.3809,
      "step": 66132
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.8630120158195496,
      "learning_rate": 0.0004860744799709384,
      "loss": 2.2955,
      "step": 66133
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.711078941822052,
      "learning_rate": 0.00048607127129391623,
      "loss": 2.2967,
      "step": 66134
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7704018354415894,
      "learning_rate": 0.00048606806258230006,
      "loss": 1.9761,
      "step": 66135
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7695721983909607,
      "learning_rate": 0.0004860648538360902,
      "loss": 2.1518,
      "step": 66136
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7954633831977844,
      "learning_rate": 0.0004860616450552874,
      "loss": 2.2562,
      "step": 66137
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7510513663291931,
      "learning_rate": 0.00048605843623989223,
      "loss": 2.1723,
      "step": 66138
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7586975693702698,
      "learning_rate": 0.00048605522738990524,
      "loss": 2.198,
      "step": 66139
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.9003919959068298,
      "learning_rate": 0.00048605201850532715,
      "loss": 2.2109,
      "step": 66140
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.785656213760376,
      "learning_rate": 0.0004860488095861584,
      "loss": 2.2471,
      "step": 66141
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.8061627149581909,
      "learning_rate": 0.00048604560063239977,
      "loss": 2.0172,
      "step": 66142
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.892493486404419,
      "learning_rate": 0.00048604239164405167,
      "loss": 2.2268,
      "step": 66143
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.8157609701156616,
      "learning_rate": 0.0004860391826211148,
      "loss": 2.2397,
      "step": 66144
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7730137705802917,
      "learning_rate": 0.0004860359735635897,
      "loss": 2.4921,
      "step": 66145
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7197399735450745,
      "learning_rate": 0.00048603276447147707,
      "loss": 2.4598,
      "step": 66146
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.9353861212730408,
      "learning_rate": 0.0004860295553447774,
      "loss": 2.2848,
      "step": 66147
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7814486622810364,
      "learning_rate": 0.0004860263461834913,
      "loss": 2.1681,
      "step": 66148
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.6827496886253357,
      "learning_rate": 0.0004860231369876194,
      "loss": 2.0721,
      "step": 66149
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7222484946250916,
      "learning_rate": 0.00048601992775716235,
      "loss": 2.2282,
      "step": 66150
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.9315997958183289,
      "learning_rate": 0.00048601671849212054,
      "loss": 2.2814,
      "step": 66151
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7590912580490112,
      "learning_rate": 0.0004860135091924949,
      "loss": 2.1791,
      "step": 66152
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.8518334031105042,
      "learning_rate": 0.0004860102998582857,
      "loss": 2.3603,
      "step": 66153
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.9797300100326538,
      "learning_rate": 0.0004860070904894937,
      "loss": 2.1144,
      "step": 66154
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.6974063515663147,
      "learning_rate": 0.0004860038810861195,
      "loss": 2.3082,
      "step": 66155
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.748723566532135,
      "learning_rate": 0.00048600067164816373,
      "loss": 2.4443,
      "step": 66156
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7046918272972107,
      "learning_rate": 0.0004859974621756268,
      "loss": 2.1814,
      "step": 66157
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.8986148834228516,
      "learning_rate": 0.00048599425266850944,
      "loss": 2.5333,
      "step": 66158
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7306280732154846,
      "learning_rate": 0.0004859910431268123,
      "loss": 2.0928,
      "step": 66159
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7756922245025635,
      "learning_rate": 0.0004859878335505359,
      "loss": 2.0735,
      "step": 66160
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7201540470123291,
      "learning_rate": 0.00048598462393968086,
      "loss": 2.1732,
      "step": 66161
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.747323751449585,
      "learning_rate": 0.0004859814142942477,
      "loss": 2.3729,
      "step": 66162
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7219546437263489,
      "learning_rate": 0.0004859782046142372,
      "loss": 2.348,
      "step": 66163
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7408134937286377,
      "learning_rate": 0.00048597499489964977,
      "loss": 2.0685,
      "step": 66164
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.6883370876312256,
      "learning_rate": 0.00048597178515048605,
      "loss": 2.151,
      "step": 66165
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7412692904472351,
      "learning_rate": 0.0004859685753667467,
      "loss": 1.9744,
      "step": 66166
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.804760217666626,
      "learning_rate": 0.0004859653655484323,
      "loss": 2.4113,
      "step": 66167
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.6911606192588806,
      "learning_rate": 0.0004859621556955434,
      "loss": 2.2258,
      "step": 66168
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.9470815658569336,
      "learning_rate": 0.00048595894580808066,
      "loss": 2.3788,
      "step": 66169
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7075979113578796,
      "learning_rate": 0.0004859557358860446,
      "loss": 1.9861,
      "step": 66170
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.9368545413017273,
      "learning_rate": 0.0004859525259294358,
      "loss": 2.2751,
      "step": 66171
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.9759756326675415,
      "learning_rate": 0.00048594931593825506,
      "loss": 2.3675,
      "step": 66172
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.8134487271308899,
      "learning_rate": 0.0004859461059125028,
      "loss": 2.4355,
      "step": 66173
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.6782652735710144,
      "learning_rate": 0.0004859428958521795,
      "loss": 2.2471,
      "step": 66174
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7391072511672974,
      "learning_rate": 0.00048593968575728614,
      "loss": 2.1945,
      "step": 66175
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.8038713932037354,
      "learning_rate": 0.0004859364756278229,
      "loss": 2.2369,
      "step": 66176
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7547476887702942,
      "learning_rate": 0.0004859332654637906,
      "loss": 2.0791,
      "step": 66177
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7587615847587585,
      "learning_rate": 0.0004859300552651898,
      "loss": 2.1478,
      "step": 66178
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7574378848075867,
      "learning_rate": 0.0004859268450320212,
      "loss": 2.0793,
      "step": 66179
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.8238459229469299,
      "learning_rate": 0.00048592363476428516,
      "loss": 2.1097,
      "step": 66180
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7799581289291382,
      "learning_rate": 0.00048592042446198247,
      "loss": 2.261,
      "step": 66181
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7275053262710571,
      "learning_rate": 0.0004859172141251136,
      "loss": 2.1793,
      "step": 66182
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7553882002830505,
      "learning_rate": 0.00048591400375367917,
      "loss": 2.3527,
      "step": 66183
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.8022063374519348,
      "learning_rate": 0.00048591079334768,
      "loss": 2.0392,
      "step": 66184
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7636703848838806,
      "learning_rate": 0.00048590758290711634,
      "loss": 2.0637,
      "step": 66185
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7755008935928345,
      "learning_rate": 0.000485904372431989,
      "loss": 2.3256,
      "step": 66186
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7435330152511597,
      "learning_rate": 0.00048590116192229866,
      "loss": 2.211,
      "step": 66187
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7293890714645386,
      "learning_rate": 0.00048589795137804557,
      "loss": 2.1656,
      "step": 66188
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.756281316280365,
      "learning_rate": 0.00048589474079923067,
      "loss": 2.0734,
      "step": 66189
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7608759999275208,
      "learning_rate": 0.0004858915301858544,
      "loss": 2.1776,
      "step": 66190
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7751045227050781,
      "learning_rate": 0.00048588831953791736,
      "loss": 2.1972,
      "step": 66191
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7663058042526245,
      "learning_rate": 0.0004858851088554202,
      "loss": 2.1695,
      "step": 66192
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7179272770881653,
      "learning_rate": 0.00048588189813836345,
      "loss": 2.2832,
      "step": 66193
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7582870125770569,
      "learning_rate": 0.00048587868738674775,
      "loss": 2.1098,
      "step": 66194
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7431384921073914,
      "learning_rate": 0.00048587547660057376,
      "loss": 2.2156,
      "step": 66195
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7538190484046936,
      "learning_rate": 0.00048587226577984197,
      "loss": 2.3677,
      "step": 66196
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.767652690410614,
      "learning_rate": 0.00048586905492455296,
      "loss": 2.0397,
      "step": 66197
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7049272060394287,
      "learning_rate": 0.0004858658440347075,
      "loss": 2.5278,
      "step": 66198
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7163634896278381,
      "learning_rate": 0.0004858626331103061,
      "loss": 2.0137,
      "step": 66199
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7946177124977112,
      "learning_rate": 0.00048585942215134914,
      "loss": 2.1592,
      "step": 66200
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.8013297915458679,
      "learning_rate": 0.0004858562111578375,
      "loss": 2.2755,
      "step": 66201
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7647626399993896,
      "learning_rate": 0.00048585300012977175,
      "loss": 2.3958,
      "step": 66202
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7230315804481506,
      "learning_rate": 0.00048584978906715233,
      "loss": 2.1672,
      "step": 66203
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.8288612365722656,
      "learning_rate": 0.00048584657796997994,
      "loss": 2.1714,
      "step": 66204
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7740159630775452,
      "learning_rate": 0.0004858433668382552,
      "loss": 2.2289,
      "step": 66205
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7816391587257385,
      "learning_rate": 0.00048584015567197864,
      "loss": 1.9895,
      "step": 66206
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7660821080207825,
      "learning_rate": 0.0004858369444711509,
      "loss": 2.142,
      "step": 66207
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.8094955086708069,
      "learning_rate": 0.0004858337332357726,
      "loss": 2.1686,
      "step": 66208
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7210847735404968,
      "learning_rate": 0.0004858305219658442,
      "loss": 2.2795,
      "step": 66209
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.8302671313285828,
      "learning_rate": 0.0004858273106613664,
      "loss": 2.1279,
      "step": 66210
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.8835864663124084,
      "learning_rate": 0.00048582409932233995,
      "loss": 2.128,
      "step": 66211
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7215436100959778,
      "learning_rate": 0.00048582088794876517,
      "loss": 2.1948,
      "step": 66212
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.9355220794677734,
      "learning_rate": 0.0004858176765406428,
      "loss": 1.9945,
      "step": 66213
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.1259604692459106,
      "learning_rate": 0.0004858144650979734,
      "loss": 2.1551,
      "step": 66214
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.8032268285751343,
      "learning_rate": 0.0004858112536207576,
      "loss": 2.2867,
      "step": 66215
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.8077797889709473,
      "learning_rate": 0.00048580804210899593,
      "loss": 2.3273,
      "step": 66216
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.9100147485733032,
      "learning_rate": 0.0004858048305626892,
      "loss": 2.2516,
      "step": 66217
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.9456181526184082,
      "learning_rate": 0.0004858016189818377,
      "loss": 2.0205,
      "step": 66218
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7970708012580872,
      "learning_rate": 0.0004857984073664422,
      "loss": 2.2118,
      "step": 66219
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.9517572522163391,
      "learning_rate": 0.00048579519571650325,
      "loss": 2.4583,
      "step": 66220
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.8016906380653381,
      "learning_rate": 0.00048579198403202157,
      "loss": 2.176,
      "step": 66221
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.8650104999542236,
      "learning_rate": 0.0004857887723129975,
      "loss": 2.2397,
      "step": 66222
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7821219563484192,
      "learning_rate": 0.0004857855605594319,
      "loss": 2.148,
      "step": 66223
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.8210583925247192,
      "learning_rate": 0.00048578234877132517,
      "loss": 2.2396,
      "step": 66224
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.8791893124580383,
      "learning_rate": 0.0004857791369486781,
      "loss": 2.13,
      "step": 66225
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.8079162836074829,
      "learning_rate": 0.0004857759250914911,
      "loss": 2.1834,
      "step": 66226
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.8214564919471741,
      "learning_rate": 0.00048577271319976496,
      "loss": 2.3161,
      "step": 66227
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.9179489612579346,
      "learning_rate": 0.00048576950127349997,
      "loss": 2.1944,
      "step": 66228
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.0365149974822998,
      "learning_rate": 0.00048576628931269706,
      "loss": 2.1626,
      "step": 66229
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.773485541343689,
      "learning_rate": 0.0004857630773173567,
      "loss": 2.3641,
      "step": 66230
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.9332787394523621,
      "learning_rate": 0.0004857598652874794,
      "loss": 1.962,
      "step": 66231
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.8094213008880615,
      "learning_rate": 0.0004857566532230659,
      "loss": 2.1674,
      "step": 66232
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7603576183319092,
      "learning_rate": 0.00048575344112411674,
      "loss": 2.1058,
      "step": 66233
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7634079456329346,
      "learning_rate": 0.00048575022899063246,
      "loss": 2.3087,
      "step": 66234
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.6946003437042236,
      "learning_rate": 0.00048574701682261375,
      "loss": 2.1823,
      "step": 66235
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.884148120880127,
      "learning_rate": 0.00048574380462006113,
      "loss": 2.323,
      "step": 66236
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.9000335335731506,
      "learning_rate": 0.0004857405923829752,
      "loss": 2.1556,
      "step": 66237
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.730221152305603,
      "learning_rate": 0.00048573738011135664,
      "loss": 2.2551,
      "step": 66238
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.8399050831794739,
      "learning_rate": 0.00048573416780520596,
      "loss": 2.4715,
      "step": 66239
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7708495259284973,
      "learning_rate": 0.0004857309554645239,
      "loss": 2.217,
      "step": 66240
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7591255307197571,
      "learning_rate": 0.0004857277430893108,
      "loss": 2.2458,
      "step": 66241
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.8265445828437805,
      "learning_rate": 0.00048572453067956755,
      "loss": 2.2926,
      "step": 66242
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7486060857772827,
      "learning_rate": 0.0004857213182352944,
      "loss": 2.3393,
      "step": 66243
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.8166477084159851,
      "learning_rate": 0.0004857181057564924,
      "loss": 2.2174,
      "step": 66244
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7035667300224304,
      "learning_rate": 0.00048571489324316167,
      "loss": 2.2399,
      "step": 66245
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7251585721969604,
      "learning_rate": 0.00048571168069530313,
      "loss": 2.1702,
      "step": 66246
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.6917318105697632,
      "learning_rate": 0.0004857084681129173,
      "loss": 2.1227,
      "step": 66247
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.8296527862548828,
      "learning_rate": 0.0004857052554960048,
      "loss": 2.3134,
      "step": 66248
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7513467073440552,
      "learning_rate": 0.00048570204284456613,
      "loss": 2.1212,
      "step": 66249
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7230857014656067,
      "learning_rate": 0.0004856988301586019,
      "loss": 2.2658,
      "step": 66250
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7487186789512634,
      "learning_rate": 0.0004856956174381129,
      "loss": 1.8799,
      "step": 66251
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.776531994342804,
      "learning_rate": 0.00048569240468309945,
      "loss": 2.1115,
      "step": 66252
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.823228657245636,
      "learning_rate": 0.0004856891918935623,
      "loss": 2.2371,
      "step": 66253
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7887976169586182,
      "learning_rate": 0.00048568597906950204,
      "loss": 2.1537,
      "step": 66254
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.8923062086105347,
      "learning_rate": 0.0004856827662109193,
      "loss": 2.1233,
      "step": 66255
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7436323165893555,
      "learning_rate": 0.0004856795533178145,
      "loss": 2.1199,
      "step": 66256
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7759292125701904,
      "learning_rate": 0.0004856763403901885,
      "loss": 2.2922,
      "step": 66257
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7215451598167419,
      "learning_rate": 0.0004856731274280417,
      "loss": 1.9314,
      "step": 66258
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7411835193634033,
      "learning_rate": 0.0004856699144313747,
      "loss": 2.0954,
      "step": 66259
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7711043357849121,
      "learning_rate": 0.0004856667014001883,
      "loss": 2.3898,
      "step": 66260
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7595282196998596,
      "learning_rate": 0.0004856634883344829,
      "loss": 2.183,
      "step": 66261
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.776597797870636,
      "learning_rate": 0.00048566027523425917,
      "loss": 2.1825,
      "step": 66262
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7199121117591858,
      "learning_rate": 0.00048565706209951766,
      "loss": 2.3518,
      "step": 66263
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7962088584899902,
      "learning_rate": 0.00048565384893025894,
      "loss": 1.9998,
      "step": 66264
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.8115184307098389,
      "learning_rate": 0.0004856506357264837,
      "loss": 2.2852,
      "step": 66265
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.8340899348258972,
      "learning_rate": 0.0004856474224881925,
      "loss": 2.3789,
      "step": 66266
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7664659023284912,
      "learning_rate": 0.000485644209215386,
      "loss": 2.1871,
      "step": 66267
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.9196848273277283,
      "learning_rate": 0.0004856409959080647,
      "loss": 2.5081,
      "step": 66268
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.6998118162155151,
      "learning_rate": 0.0004856377825662293,
      "loss": 2.0133,
      "step": 66269
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7589303851127625,
      "learning_rate": 0.00048563456918988016,
      "loss": 2.1418,
      "step": 66270
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7900044918060303,
      "learning_rate": 0.0004856313557790182,
      "loss": 2.2022,
      "step": 66271
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7739080786705017,
      "learning_rate": 0.00048562814233364383,
      "loss": 2.1025,
      "step": 66272
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7970724105834961,
      "learning_rate": 0.0004856249288537577,
      "loss": 2.313,
      "step": 66273
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.8050148487091064,
      "learning_rate": 0.00048562171533936033,
      "loss": 2.2179,
      "step": 66274
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7646691203117371,
      "learning_rate": 0.0004856185017904524,
      "loss": 2.3452,
      "step": 66275
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7902188897132874,
      "learning_rate": 0.0004856152882070345,
      "loss": 2.1405,
      "step": 66276
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.696808934211731,
      "learning_rate": 0.0004856120745891072,
      "loss": 2.121,
      "step": 66277
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7840909361839294,
      "learning_rate": 0.00048560886093667113,
      "loss": 2.2186,
      "step": 66278
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7788689136505127,
      "learning_rate": 0.00048560564724972686,
      "loss": 2.0363,
      "step": 66279
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7564026117324829,
      "learning_rate": 0.00048560243352827504,
      "loss": 2.2192,
      "step": 66280
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7910560369491577,
      "learning_rate": 0.00048559921977231615,
      "loss": 2.0768,
      "step": 66281
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7667300701141357,
      "learning_rate": 0.0004855960059818509,
      "loss": 2.1688,
      "step": 66282
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7548820376396179,
      "learning_rate": 0.0004855927921568799,
      "loss": 2.1513,
      "step": 66283
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7721831202507019,
      "learning_rate": 0.00048558957829740365,
      "loss": 2.1638,
      "step": 66284
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.8244031071662903,
      "learning_rate": 0.00048558636440342273,
      "loss": 2.1146,
      "step": 66285
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.8030242323875427,
      "learning_rate": 0.0004855831504749379,
      "loss": 2.2023,
      "step": 66286
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7750188112258911,
      "learning_rate": 0.0004855799365119496,
      "loss": 2.3265,
      "step": 66287
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.8066402077674866,
      "learning_rate": 0.0004855767225144585,
      "loss": 2.1292,
      "step": 66288
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7763462662696838,
      "learning_rate": 0.00048557350848246517,
      "loss": 2.2886,
      "step": 66289
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7294549345970154,
      "learning_rate": 0.00048557029441597023,
      "loss": 2.3135,
      "step": 66290
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7477514743804932,
      "learning_rate": 0.00048556708031497435,
      "loss": 2.2399,
      "step": 66291
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7337666749954224,
      "learning_rate": 0.0004855638661794779,
      "loss": 2.1457,
      "step": 66292
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.8469517230987549,
      "learning_rate": 0.00048556065200948173,
      "loss": 2.3864,
      "step": 66293
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7853269577026367,
      "learning_rate": 0.00048555743780498634,
      "loss": 2.2353,
      "step": 66294
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7603389024734497,
      "learning_rate": 0.00048555422356599224,
      "loss": 2.0882,
      "step": 66295
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.8092876076698303,
      "learning_rate": 0.0004855510092925002,
      "loss": 2.1645,
      "step": 66296
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.726080596446991,
      "learning_rate": 0.0004855477949845107,
      "loss": 2.1724,
      "step": 66297
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7761636972427368,
      "learning_rate": 0.0004855445806420243,
      "loss": 2.276,
      "step": 66298
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7471542954444885,
      "learning_rate": 0.00048554136626504166,
      "loss": 2.1781,
      "step": 66299
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.8469858169555664,
      "learning_rate": 0.0004855381518535635,
      "loss": 2.4002,
      "step": 66300
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7392856478691101,
      "learning_rate": 0.00048553493740759017,
      "loss": 2.1087,
      "step": 66301
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.8883039951324463,
      "learning_rate": 0.0004855317229271224,
      "loss": 2.0301,
      "step": 66302
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7953271269798279,
      "learning_rate": 0.0004855285084121609,
      "loss": 2.2295,
      "step": 66303
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.8689523339271545,
      "learning_rate": 0.000485525293862706,
      "loss": 2.1752,
      "step": 66304
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7707480788230896,
      "learning_rate": 0.00048552207927875853,
      "loss": 2.2141,
      "step": 66305
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7300030589103699,
      "learning_rate": 0.000485518864660319,
      "loss": 2.191,
      "step": 66306
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7214043736457825,
      "learning_rate": 0.000485515650007388,
      "loss": 2.0728,
      "step": 66307
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7756252884864807,
      "learning_rate": 0.00048551243531996606,
      "loss": 2.275,
      "step": 66308
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.754155158996582,
      "learning_rate": 0.000485509220598054,
      "loss": 2.2632,
      "step": 66309
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7733917832374573,
      "learning_rate": 0.00048550600584165216,
      "loss": 2.2173,
      "step": 66310
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7907509207725525,
      "learning_rate": 0.00048550279105076137,
      "loss": 2.2396,
      "step": 66311
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7375085353851318,
      "learning_rate": 0.000485499576225382,
      "loss": 2.308,
      "step": 66312
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.70614093542099,
      "learning_rate": 0.0004854963613655148,
      "loss": 2.1486,
      "step": 66313
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7865310311317444,
      "learning_rate": 0.00048549314647116027,
      "loss": 2.3973,
      "step": 66314
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.819014310836792,
      "learning_rate": 0.0004854899315423191,
      "loss": 2.1663,
      "step": 66315
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7169440984725952,
      "learning_rate": 0.00048548671657899187,
      "loss": 2.1081,
      "step": 66316
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7268792390823364,
      "learning_rate": 0.0004854835015811792,
      "loss": 2.2579,
      "step": 66317
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7662935853004456,
      "learning_rate": 0.0004854802865488816,
      "loss": 2.184,
      "step": 66318
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.8157879710197449,
      "learning_rate": 0.00048547707148209975,
      "loss": 2.4052,
      "step": 66319
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.781470000743866,
      "learning_rate": 0.00048547385638083413,
      "loss": 2.1022,
      "step": 66320
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7768602967262268,
      "learning_rate": 0.0004854706412450854,
      "loss": 2.1129,
      "step": 66321
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7850051522254944,
      "learning_rate": 0.0004854674260748543,
      "loss": 2.2368,
      "step": 66322
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7058348655700684,
      "learning_rate": 0.00048546421087014113,
      "loss": 2.2782,
      "step": 66323
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7791459560394287,
      "learning_rate": 0.0004854609956309468,
      "loss": 2.3254,
      "step": 66324
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7315725684165955,
      "learning_rate": 0.00048545778035727176,
      "loss": 2.2686,
      "step": 66325
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7945675849914551,
      "learning_rate": 0.00048545456504911656,
      "loss": 2.3922,
      "step": 66326
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.6759589314460754,
      "learning_rate": 0.00048545134970648195,
      "loss": 2.0489,
      "step": 66327
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7698301076889038,
      "learning_rate": 0.0004854481343293684,
      "loss": 2.1554,
      "step": 66328
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7311129570007324,
      "learning_rate": 0.0004854449189177765,
      "loss": 2.1128,
      "step": 66329
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.8536285161972046,
      "learning_rate": 0.00048544170347170697,
      "loss": 2.1652,
      "step": 66330
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7255625128746033,
      "learning_rate": 0.00048543848799116016,
      "loss": 2.0744,
      "step": 66331
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7985953092575073,
      "learning_rate": 0.000485435272476137,
      "loss": 2.3866,
      "step": 66332
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7351520657539368,
      "learning_rate": 0.0004854320569266379,
      "loss": 2.2201,
      "step": 66333
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.8269766569137573,
      "learning_rate": 0.0004854288413426634,
      "loss": 2.217,
      "step": 66334
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7614437937736511,
      "learning_rate": 0.0004854256257242142,
      "loss": 2.425,
      "step": 66335
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7525830864906311,
      "learning_rate": 0.00048542241007129094,
      "loss": 2.0309,
      "step": 66336
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.8048357963562012,
      "learning_rate": 0.0004854191943838941,
      "loss": 2.2344,
      "step": 66337
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7570009827613831,
      "learning_rate": 0.0004854159786620243,
      "loss": 2.3096,
      "step": 66338
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7391970753669739,
      "learning_rate": 0.0004854127629056822,
      "loss": 2.2846,
      "step": 66339
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7562300562858582,
      "learning_rate": 0.00048540954711486845,
      "loss": 2.2836,
      "step": 66340
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7507316470146179,
      "learning_rate": 0.00048540633128958346,
      "loss": 2.1801,
      "step": 66341
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7599100470542908,
      "learning_rate": 0.00048540311542982805,
      "loss": 2.3222,
      "step": 66342
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7488611340522766,
      "learning_rate": 0.0004853998995356026,
      "loss": 2.2671,
      "step": 66343
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.6925413012504578,
      "learning_rate": 0.00048539668360690776,
      "loss": 2.1266,
      "step": 66344
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.8309589624404907,
      "learning_rate": 0.0004853934676437443,
      "loss": 2.0592,
      "step": 66345
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7640048265457153,
      "learning_rate": 0.0004853902516461126,
      "loss": 2.3911,
      "step": 66346
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7733217477798462,
      "learning_rate": 0.00048538703561401336,
      "loss": 2.1578,
      "step": 66347
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.792189359664917,
      "learning_rate": 0.00048538381954744725,
      "loss": 2.4947,
      "step": 66348
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7883984446525574,
      "learning_rate": 0.00048538060344641474,
      "loss": 1.9717,
      "step": 66349
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7112308144569397,
      "learning_rate": 0.00048537738731091646,
      "loss": 2.3099,
      "step": 66350
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.9325655698776245,
      "learning_rate": 0.00048537417114095303,
      "loss": 1.9887,
      "step": 66351
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7684295177459717,
      "learning_rate": 0.00048537095493652503,
      "loss": 2.0657,
      "step": 66352
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7994836568832397,
      "learning_rate": 0.0004853677386976332,
      "loss": 2.2089,
      "step": 66353
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.8577742576599121,
      "learning_rate": 0.00048536452242427784,
      "loss": 2.2801,
      "step": 66354
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7192705869674683,
      "learning_rate": 0.0004853613061164598,
      "loss": 2.221,
      "step": 66355
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7781251072883606,
      "learning_rate": 0.00048535808977417954,
      "loss": 2.1496,
      "step": 66356
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.8155192732810974,
      "learning_rate": 0.0004853548733974378,
      "loss": 2.1311,
      "step": 66357
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.6924461722373962,
      "learning_rate": 0.000485351656986235,
      "loss": 2.233,
      "step": 66358
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7174047231674194,
      "learning_rate": 0.0004853484405405719,
      "loss": 2.4798,
      "step": 66359
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7664986252784729,
      "learning_rate": 0.00048534522406044894,
      "loss": 2.2559,
      "step": 66360
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7963367104530334,
      "learning_rate": 0.00048534200754586686,
      "loss": 2.2123,
      "step": 66361
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7972704172134399,
      "learning_rate": 0.00048533879099682623,
      "loss": 2.3149,
      "step": 66362
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7267950773239136,
      "learning_rate": 0.0004853355744133276,
      "loss": 1.9742,
      "step": 66363
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7466939687728882,
      "learning_rate": 0.0004853323577953716,
      "loss": 2.0958,
      "step": 66364
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7024623155593872,
      "learning_rate": 0.00048532914114295876,
      "loss": 2.2957,
      "step": 66365
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7344403266906738,
      "learning_rate": 0.0004853259244560898,
      "loss": 2.2717,
      "step": 66366
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.8579126000404358,
      "learning_rate": 0.00048532270773476516,
      "loss": 2.2223,
      "step": 66367
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7152814269065857,
      "learning_rate": 0.0004853194909789856,
      "loss": 2.4236,
      "step": 66368
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7656660079956055,
      "learning_rate": 0.00048531627418875163,
      "loss": 2.1169,
      "step": 66369
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7239264249801636,
      "learning_rate": 0.00048531305736406393,
      "loss": 2.348,
      "step": 66370
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.9482998847961426,
      "learning_rate": 0.00048530984050492294,
      "loss": 2.2838,
      "step": 66371
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7865714430809021,
      "learning_rate": 0.00048530662361132937,
      "loss": 2.3683,
      "step": 66372
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.760180652141571,
      "learning_rate": 0.0004853034066832838,
      "loss": 2.2202,
      "step": 66373
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7668936848640442,
      "learning_rate": 0.00048530018972078686,
      "loss": 2.2585,
      "step": 66374
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.8606613278388977,
      "learning_rate": 0.0004852969727238391,
      "loss": 2.1272,
      "step": 66375
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7769345045089722,
      "learning_rate": 0.00048529375569244116,
      "loss": 2.2637,
      "step": 66376
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7747011184692383,
      "learning_rate": 0.0004852905386265936,
      "loss": 2.1156,
      "step": 66377
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7720719575881958,
      "learning_rate": 0.00048528732152629706,
      "loss": 2.1898,
      "step": 66378
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7681739330291748,
      "learning_rate": 0.00048528410439155204,
      "loss": 2.3917,
      "step": 66379
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7969093322753906,
      "learning_rate": 0.00048528088722235926,
      "loss": 2.3513,
      "step": 66380
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7602638006210327,
      "learning_rate": 0.0004852776700187192,
      "loss": 2.2758,
      "step": 66381
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7403832674026489,
      "learning_rate": 0.0004852744527806326,
      "loss": 2.4059,
      "step": 66382
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.8801780939102173,
      "learning_rate": 0.00048527123550809996,
      "loss": 2.0636,
      "step": 66383
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7640669941902161,
      "learning_rate": 0.0004852680182011219,
      "loss": 2.231,
      "step": 66384
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.8740807771682739,
      "learning_rate": 0.00048526480085969904,
      "loss": 2.3377,
      "step": 66385
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.8145154118537903,
      "learning_rate": 0.00048526158348383185,
      "loss": 2.2921,
      "step": 66386
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.9413387179374695,
      "learning_rate": 0.00048525836607352113,
      "loss": 2.1644,
      "step": 66387
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.764565646648407,
      "learning_rate": 0.00048525514862876747,
      "loss": 2.3464,
      "step": 66388
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7261546850204468,
      "learning_rate": 0.0004852519311495712,
      "loss": 2.2179,
      "step": 66389
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7957680225372314,
      "learning_rate": 0.00048524871363593314,
      "loss": 2.5074,
      "step": 66390
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.8141797184944153,
      "learning_rate": 0.000485245496087854,
      "loss": 2.2202,
      "step": 66391
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.767257571220398,
      "learning_rate": 0.00048524227850533407,
      "loss": 2.1585,
      "step": 66392
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7850729823112488,
      "learning_rate": 0.00048523906088837414,
      "loss": 2.1984,
      "step": 66393
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7852678894996643,
      "learning_rate": 0.0004852358432369748,
      "loss": 2.1996,
      "step": 66394
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7566859126091003,
      "learning_rate": 0.0004852326255511366,
      "loss": 2.0425,
      "step": 66395
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7567339539527893,
      "learning_rate": 0.00048522940783086006,
      "loss": 2.5295,
      "step": 66396
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.8892748951911926,
      "learning_rate": 0.000485226190076146,
      "loss": 2.4496,
      "step": 66397
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.803617537021637,
      "learning_rate": 0.0004852229722869949,
      "loss": 2.3458,
      "step": 66398
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.9093769192695618,
      "learning_rate": 0.0004852197544634073,
      "loss": 2.08,
      "step": 66399
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.8436951637268066,
      "learning_rate": 0.0004852165366053839,
      "loss": 2.2585,
      "step": 66400
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.8243981003761292,
      "learning_rate": 0.0004852133187129252,
      "loss": 2.1217,
      "step": 66401
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7486024498939514,
      "learning_rate": 0.0004852101007860319,
      "loss": 2.2162,
      "step": 66402
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7642883658409119,
      "learning_rate": 0.0004852068828247045,
      "loss": 2.159,
      "step": 66403
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.8002336025238037,
      "learning_rate": 0.00048520366482894366,
      "loss": 2.1559,
      "step": 66404
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.9328659176826477,
      "learning_rate": 0.00048520044679874994,
      "loss": 2.2186,
      "step": 66405
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7514661550521851,
      "learning_rate": 0.00048519722873412396,
      "loss": 2.1865,
      "step": 66406
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.6664922833442688,
      "learning_rate": 0.00048519401063506633,
      "loss": 2.2658,
      "step": 66407
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.9801446795463562,
      "learning_rate": 0.0004851907925015777,
      "loss": 2.311,
      "step": 66408
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.6992533206939697,
      "learning_rate": 0.0004851875743336586,
      "loss": 2.4667,
      "step": 66409
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7742949724197388,
      "learning_rate": 0.00048518435613130955,
      "loss": 2.144,
      "step": 66410
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7526636123657227,
      "learning_rate": 0.0004851811378945312,
      "loss": 2.1114,
      "step": 66411
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.8039351105690002,
      "learning_rate": 0.00048517791962332437,
      "loss": 2.286,
      "step": 66412
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.8269882798194885,
      "learning_rate": 0.0004851747013176893,
      "loss": 2.1326,
      "step": 66413
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.8078527450561523,
      "learning_rate": 0.00048517148297762676,
      "loss": 2.3431,
      "step": 66414
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.8107579946517944,
      "learning_rate": 0.0004851682646031374,
      "loss": 2.2843,
      "step": 66415
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.6942615509033203,
      "learning_rate": 0.00048516504619422175,
      "loss": 2.2802,
      "step": 66416
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.0051130056381226,
      "learning_rate": 0.00048516182775088044,
      "loss": 2.3229,
      "step": 66417
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7959322929382324,
      "learning_rate": 0.0004851586092731141,
      "loss": 2.2624,
      "step": 66418
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.8735755085945129,
      "learning_rate": 0.0004851553907609232,
      "loss": 2.34,
      "step": 66419
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7395957708358765,
      "learning_rate": 0.00048515217221430844,
      "loss": 2.2182,
      "step": 66420
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7830047607421875,
      "learning_rate": 0.0004851489536332704,
      "loss": 2.2986,
      "step": 66421
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7538124918937683,
      "learning_rate": 0.00048514573501780967,
      "loss": 2.1635,
      "step": 66422
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.6708759069442749,
      "learning_rate": 0.00048514251636792686,
      "loss": 2.2388,
      "step": 66423
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7392512559890747,
      "learning_rate": 0.0004851392976836226,
      "loss": 2.3551,
      "step": 66424
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.9042981863021851,
      "learning_rate": 0.00048513607896489734,
      "loss": 2.1506,
      "step": 66425
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7994608283042908,
      "learning_rate": 0.00048513286021175186,
      "loss": 2.3011,
      "step": 66426
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7344202399253845,
      "learning_rate": 0.00048512964142418666,
      "loss": 2.2352,
      "step": 66427
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7445887327194214,
      "learning_rate": 0.0004851264226022024,
      "loss": 2.2834,
      "step": 66428
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7576372027397156,
      "learning_rate": 0.00048512320374579965,
      "loss": 2.1842,
      "step": 66429
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.8195573091506958,
      "learning_rate": 0.000485119984854979,
      "loss": 2.1733,
      "step": 66430
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.748121976852417,
      "learning_rate": 0.000485116765929741,
      "loss": 2.2833,
      "step": 66431
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7743488550186157,
      "learning_rate": 0.0004851135469700863,
      "loss": 2.2828,
      "step": 66432
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7728455662727356,
      "learning_rate": 0.00048511032797601556,
      "loss": 2.1612,
      "step": 66433
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.8257834315299988,
      "learning_rate": 0.00048510710894752925,
      "loss": 2.2657,
      "step": 66434
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7859975695610046,
      "learning_rate": 0.0004851038898846281,
      "loss": 2.2684,
      "step": 66435
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.0585224628448486,
      "learning_rate": 0.0004851006707873126,
      "loss": 2.2357,
      "step": 66436
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7788741588592529,
      "learning_rate": 0.0004850974516555834,
      "loss": 2.3547,
      "step": 66437
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.891523540019989,
      "learning_rate": 0.0004850942324894411,
      "loss": 2.1464,
      "step": 66438
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7989355325698853,
      "learning_rate": 0.00048509101328888626,
      "loss": 2.2864,
      "step": 66439
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7254055142402649,
      "learning_rate": 0.00048508779405391955,
      "loss": 2.4237,
      "step": 66440
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.9217643737792969,
      "learning_rate": 0.00048508457478454157,
      "loss": 2.4055,
      "step": 66441
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7445432543754578,
      "learning_rate": 0.00048508135548075274,
      "loss": 2.2205,
      "step": 66442
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7650392055511475,
      "learning_rate": 0.00048507813614255383,
      "loss": 2.1794,
      "step": 66443
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7351316809654236,
      "learning_rate": 0.0004850749167699455,
      "loss": 2.1771,
      "step": 66444
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7127470970153809,
      "learning_rate": 0.0004850716973629281,
      "loss": 2.2174,
      "step": 66445
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7939407229423523,
      "learning_rate": 0.0004850684779215025,
      "loss": 2.1851,
      "step": 66446
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.8197878003120422,
      "learning_rate": 0.0004850652584456691,
      "loss": 2.2308,
      "step": 66447
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7627041935920715,
      "learning_rate": 0.0004850620389354286,
      "loss": 2.3248,
      "step": 66448
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.773065984249115,
      "learning_rate": 0.00048505881939078164,
      "loss": 2.0801,
      "step": 66449
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7662324905395508,
      "learning_rate": 0.0004850555998117287,
      "loss": 2.2298,
      "step": 66450
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7311631441116333,
      "learning_rate": 0.00048505238019827036,
      "loss": 2.2476,
      "step": 66451
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7976625561714172,
      "learning_rate": 0.00048504916055040737,
      "loss": 2.3074,
      "step": 66452
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7887244820594788,
      "learning_rate": 0.00048504594086814026,
      "loss": 2.078,
      "step": 66453
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7989478707313538,
      "learning_rate": 0.0004850427211514696,
      "loss": 2.4396,
      "step": 66454
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7529649138450623,
      "learning_rate": 0.00048503950140039596,
      "loss": 2.3705,
      "step": 66455
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7656015157699585,
      "learning_rate": 0.00048503628161491996,
      "loss": 2.2113,
      "step": 66456
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.8991966247558594,
      "learning_rate": 0.00048503306179504234,
      "loss": 2.1506,
      "step": 66457
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.766319990158081,
      "learning_rate": 0.0004850298419407635,
      "loss": 2.3665,
      "step": 66458
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.8424845337867737,
      "learning_rate": 0.0004850266220520841,
      "loss": 2.2814,
      "step": 66459
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7269392013549805,
      "learning_rate": 0.00048502340212900485,
      "loss": 2.3078,
      "step": 66460
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7311177849769592,
      "learning_rate": 0.00048502018217152625,
      "loss": 2.1662,
      "step": 66461
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7522881627082825,
      "learning_rate": 0.00048501696217964884,
      "loss": 2.2226,
      "step": 66462
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7827586531639099,
      "learning_rate": 0.0004850137421533733,
      "loss": 2.3478,
      "step": 66463
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.8097971677780151,
      "learning_rate": 0.00048501052209270015,
      "loss": 2.267,
      "step": 66464
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.9999359250068665,
      "learning_rate": 0.00048500730199763016,
      "loss": 2.3539,
      "step": 66465
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7935714721679688,
      "learning_rate": 0.00048500408186816376,
      "loss": 2.3285,
      "step": 66466
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7095814347267151,
      "learning_rate": 0.0004850008617043017,
      "loss": 2.186,
      "step": 66467
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.0401232242584229,
      "learning_rate": 0.0004849976415060444,
      "loss": 2.0075,
      "step": 66468
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7380495071411133,
      "learning_rate": 0.0004849944212733925,
      "loss": 2.2845,
      "step": 66469
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.8254613876342773,
      "learning_rate": 0.0004849912010063468,
      "loss": 2.236,
      "step": 66470
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.8966312408447266,
      "learning_rate": 0.00048498798070490757,
      "loss": 2.5704,
      "step": 66471
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.8052279949188232,
      "learning_rate": 0.0004849847603690757,
      "loss": 2.0166,
      "step": 66472
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.697903573513031,
      "learning_rate": 0.0004849815399988516,
      "loss": 2.1495,
      "step": 66473
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.8572008609771729,
      "learning_rate": 0.00048497831959423595,
      "loss": 2.3282,
      "step": 66474
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7987306714057922,
      "learning_rate": 0.00048497509915522945,
      "loss": 2.1563,
      "step": 66475
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7714893817901611,
      "learning_rate": 0.00048497187868183254,
      "loss": 1.9377,
      "step": 66476
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7761422991752625,
      "learning_rate": 0.00048496865817404576,
      "loss": 2.2575,
      "step": 66477
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.762066662311554,
      "learning_rate": 0.0004849654376318699,
      "loss": 2.2519,
      "step": 66478
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7502613663673401,
      "learning_rate": 0.00048496221705530545,
      "loss": 2.4467,
      "step": 66479
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7965133786201477,
      "learning_rate": 0.000484958996444353,
      "loss": 2.0726,
      "step": 66480
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7895549535751343,
      "learning_rate": 0.0004849557757990132,
      "loss": 2.234,
      "step": 66481
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.0356526374816895,
      "learning_rate": 0.0004849525551192867,
      "loss": 2.2166,
      "step": 66482
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.6840718388557434,
      "learning_rate": 0.000484949334405174,
      "loss": 2.232,
      "step": 66483
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7954529523849487,
      "learning_rate": 0.0004849461136566757,
      "loss": 2.1719,
      "step": 66484
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.0809930562973022,
      "learning_rate": 0.0004849428928737925,
      "loss": 2.2037,
      "step": 66485
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7698615193367004,
      "learning_rate": 0.0004849396720565248,
      "loss": 2.2708,
      "step": 66486
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.8695812821388245,
      "learning_rate": 0.00048493645120487337,
      "loss": 2.3014,
      "step": 66487
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7317085266113281,
      "learning_rate": 0.0004849332303188388,
      "loss": 2.1552,
      "step": 66488
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.0066351890563965,
      "learning_rate": 0.0004849300093984216,
      "loss": 2.2143,
      "step": 66489
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.785952627658844,
      "learning_rate": 0.0004849267884436224,
      "loss": 2.4794,
      "step": 66490
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.1026822328567505,
      "learning_rate": 0.00048492356745444195,
      "loss": 2.1494,
      "step": 66491
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7726693749427795,
      "learning_rate": 0.00048492034643088063,
      "loss": 2.0652,
      "step": 66492
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.745491087436676,
      "learning_rate": 0.00048491712537293905,
      "loss": 2.273,
      "step": 66493
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7162989377975464,
      "learning_rate": 0.000484913904280618,
      "loss": 2.261,
      "step": 66494
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.8930965662002563,
      "learning_rate": 0.00048491068315391795,
      "loss": 2.3181,
      "step": 66495
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7838555574417114,
      "learning_rate": 0.00048490746199283946,
      "loss": 2.3657,
      "step": 66496
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.8000275492668152,
      "learning_rate": 0.00048490424079738326,
      "loss": 2.2796,
      "step": 66497
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7849395275115967,
      "learning_rate": 0.0004849010195675498,
      "loss": 2.3006,
      "step": 66498
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7171133756637573,
      "learning_rate": 0.0004848977983033398,
      "loss": 2.291,
      "step": 66499
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.686095118522644,
      "learning_rate": 0.0004848945770047538,
      "loss": 2.1049,
      "step": 66500
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.8510206937789917,
      "learning_rate": 0.00048489135567179235,
      "loss": 2.1157,
      "step": 66501
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7520235776901245,
      "learning_rate": 0.00048488813430445617,
      "loss": 2.1735,
      "step": 66502
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7368006110191345,
      "learning_rate": 0.0004848849129027458,
      "loss": 2.2899,
      "step": 66503
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7150377631187439,
      "learning_rate": 0.00048488169146666176,
      "loss": 2.1478,
      "step": 66504
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7585704922676086,
      "learning_rate": 0.0004848784699962049,
      "loss": 2.4022,
      "step": 66505
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7457928657531738,
      "learning_rate": 0.00048487524849137543,
      "loss": 2.348,
      "step": 66506
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7222400307655334,
      "learning_rate": 0.0004848720269521743,
      "loss": 2.5198,
      "step": 66507
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7508853077888489,
      "learning_rate": 0.00048486880537860187,
      "loss": 2.1133,
      "step": 66508
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7689347267150879,
      "learning_rate": 0.0004848655837706589,
      "loss": 2.2252,
      "step": 66509
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.8168250918388367,
      "learning_rate": 0.0004848623621283459,
      "loss": 2.3388,
      "step": 66510
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7111196517944336,
      "learning_rate": 0.00048485914045166346,
      "loss": 2.3045,
      "step": 66511
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7594112753868103,
      "learning_rate": 0.00048485591874061236,
      "loss": 2.1132,
      "step": 66512
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.0204662084579468,
      "learning_rate": 0.000484852696995193,
      "loss": 2.2611,
      "step": 66513
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7413608431816101,
      "learning_rate": 0.00048484947521540596,
      "loss": 2.0259,
      "step": 66514
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.9269527792930603,
      "learning_rate": 0.00048484625340125184,
      "loss": 2.1873,
      "step": 66515
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.8108546137809753,
      "learning_rate": 0.0004848430315527315,
      "loss": 2.1732,
      "step": 66516
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7691090106964111,
      "learning_rate": 0.00048483980966984526,
      "loss": 2.2162,
      "step": 66517
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.8396901488304138,
      "learning_rate": 0.0004848365877525939,
      "loss": 2.288,
      "step": 66518
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.8161930441856384,
      "learning_rate": 0.00048483336580097783,
      "loss": 2.1383,
      "step": 66519
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.6883924007415771,
      "learning_rate": 0.0004848301438149978,
      "loss": 2.2228,
      "step": 66520
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.749200701713562,
      "learning_rate": 0.0004848269217946543,
      "loss": 2.1813,
      "step": 66521
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7949803471565247,
      "learning_rate": 0.00048482369973994806,
      "loss": 2.1288,
      "step": 66522
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.8565440773963928,
      "learning_rate": 0.0004848204776508795,
      "loss": 2.1734,
      "step": 66523
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7137824892997742,
      "learning_rate": 0.0004848172555274494,
      "loss": 2.2103,
      "step": 66524
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.9366944432258606,
      "learning_rate": 0.0004848140333696583,
      "loss": 2.1083,
      "step": 66525
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.0569565296173096,
      "learning_rate": 0.00048481081117750675,
      "loss": 2.1983,
      "step": 66526
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7010660171508789,
      "learning_rate": 0.00048480758895099536,
      "loss": 2.2505,
      "step": 66527
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.8732314109802246,
      "learning_rate": 0.0004848043666901248,
      "loss": 2.3441,
      "step": 66528
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7824134230613708,
      "learning_rate": 0.00048480114439489565,
      "loss": 2.1201,
      "step": 66529
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7409153580665588,
      "learning_rate": 0.00048479792206530836,
      "loss": 2.2688,
      "step": 66530
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.9166150689125061,
      "learning_rate": 0.0004847946997013637,
      "loss": 2.0866,
      "step": 66531
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7188517451286316,
      "learning_rate": 0.00048479147730306235,
      "loss": 2.2575,
      "step": 66532
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7380461692810059,
      "learning_rate": 0.00048478825487040456,
      "loss": 2.3293,
      "step": 66533
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.8633520007133484,
      "learning_rate": 0.00048478503240339133,
      "loss": 2.0994,
      "step": 66534
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7539222836494446,
      "learning_rate": 0.00048478180990202293,
      "loss": 2.1387,
      "step": 66535
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.679338812828064,
      "learning_rate": 0.0004847785873663002,
      "loss": 2.3153,
      "step": 66536
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7689788341522217,
      "learning_rate": 0.00048477536479622356,
      "loss": 2.2039,
      "step": 66537
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.8879700899124146,
      "learning_rate": 0.0004847721421917937,
      "loss": 2.2499,
      "step": 66538
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7342545986175537,
      "learning_rate": 0.00048476891955301127,
      "loss": 2.1949,
      "step": 66539
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7822123765945435,
      "learning_rate": 0.00048476569687987684,
      "loss": 2.1101,
      "step": 66540
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7903386354446411,
      "learning_rate": 0.00048476247417239093,
      "loss": 2.3144,
      "step": 66541
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7234642505645752,
      "learning_rate": 0.00048475925143055414,
      "loss": 2.1641,
      "step": 66542
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7966828346252441,
      "learning_rate": 0.00048475602865436716,
      "loss": 2.1055,
      "step": 66543
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7169928550720215,
      "learning_rate": 0.00048475280584383054,
      "loss": 2.1209,
      "step": 66544
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7313978672027588,
      "learning_rate": 0.0004847495829989449,
      "loss": 2.2989,
      "step": 66545
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7465676069259644,
      "learning_rate": 0.0004847463601197109,
      "loss": 2.1678,
      "step": 66546
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.761375367641449,
      "learning_rate": 0.0004847431372061289,
      "loss": 2.2655,
      "step": 66547
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7787147760391235,
      "learning_rate": 0.0004847399142581998,
      "loss": 2.2552,
      "step": 66548
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.746724545955658,
      "learning_rate": 0.000484736691275924,
      "loss": 2.175,
      "step": 66549
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7704795002937317,
      "learning_rate": 0.00048473346825930215,
      "loss": 2.2113,
      "step": 66550
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7812293767929077,
      "learning_rate": 0.0004847302452083349,
      "loss": 2.5123,
      "step": 66551
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7364702820777893,
      "learning_rate": 0.00048472702212302277,
      "loss": 2.2439,
      "step": 66552
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.9243839383125305,
      "learning_rate": 0.0004847237990033664,
      "loss": 2.3307,
      "step": 66553
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.8399602174758911,
      "learning_rate": 0.00048472057584936646,
      "loss": 2.0596,
      "step": 66554
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.8045220971107483,
      "learning_rate": 0.00048471735266102345,
      "loss": 1.9587,
      "step": 66555
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.6801058650016785,
      "learning_rate": 0.00048471412943833795,
      "loss": 2.321,
      "step": 66556
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.742206871509552,
      "learning_rate": 0.0004847109061813106,
      "loss": 2.1428,
      "step": 66557
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7773568630218506,
      "learning_rate": 0.0004847076828899421,
      "loss": 2.2797,
      "step": 66558
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.78692626953125,
      "learning_rate": 0.00048470445956423283,
      "loss": 2.4209,
      "step": 66559
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7366071939468384,
      "learning_rate": 0.00048470123620418363,
      "loss": 2.1648,
      "step": 66560
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7539046406745911,
      "learning_rate": 0.0004846980128097949,
      "loss": 2.401,
      "step": 66561
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.709556519985199,
      "learning_rate": 0.0004846947893810674,
      "loss": 2.0089,
      "step": 66562
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7726112604141235,
      "learning_rate": 0.0004846915659180016,
      "loss": 2.2328,
      "step": 66563
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7729793787002563,
      "learning_rate": 0.0004846883424205981,
      "loss": 2.2591,
      "step": 66564
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7128164768218994,
      "learning_rate": 0.00048468511888885763,
      "loss": 2.1991,
      "step": 66565
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7470458745956421,
      "learning_rate": 0.00048468189532278075,
      "loss": 2.3221,
      "step": 66566
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7550538778305054,
      "learning_rate": 0.00048467867172236794,
      "loss": 2.2676,
      "step": 66567
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7631002068519592,
      "learning_rate": 0.0004846754480876199,
      "loss": 1.9843,
      "step": 66568
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.8660888671875,
      "learning_rate": 0.00048467222441853715,
      "loss": 2.3661,
      "step": 66569
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.866361677646637,
      "learning_rate": 0.0004846690007151204,
      "loss": 2.2545,
      "step": 66570
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7538547515869141,
      "learning_rate": 0.00048466577697737026,
      "loss": 2.2652,
      "step": 66571
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7196264863014221,
      "learning_rate": 0.0004846625532052872,
      "loss": 1.9907,
      "step": 66572
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.0469008684158325,
      "learning_rate": 0.0004846593293988719,
      "loss": 2.1278,
      "step": 66573
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.6949573159217834,
      "learning_rate": 0.00048465610555812493,
      "loss": 2.1666,
      "step": 66574
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.8469276428222656,
      "learning_rate": 0.0004846528816830469,
      "loss": 2.1445,
      "step": 66575
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.3652524948120117,
      "learning_rate": 0.00048464965777363843,
      "loss": 1.8656,
      "step": 66576
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7607539892196655,
      "learning_rate": 0.0004846464338299002,
      "loss": 2.1516,
      "step": 66577
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7860150337219238,
      "learning_rate": 0.0004846432098518325,
      "loss": 2.1864,
      "step": 66578
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.0007261037826538,
      "learning_rate": 0.00048463998583943627,
      "loss": 2.0591,
      "step": 66579
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7952940464019775,
      "learning_rate": 0.00048463676179271196,
      "loss": 1.9779,
      "step": 66580
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7308682799339294,
      "learning_rate": 0.00048463353771166024,
      "loss": 2.2816,
      "step": 66581
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7810869812965393,
      "learning_rate": 0.00048463031359628154,
      "loss": 2.2826,
      "step": 66582
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7225002646446228,
      "learning_rate": 0.0004846270894465767,
      "loss": 2.3971,
      "step": 66583
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7575283646583557,
      "learning_rate": 0.0004846238652625462,
      "loss": 2.3079,
      "step": 66584
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7628201246261597,
      "learning_rate": 0.00048462064104419055,
      "loss": 2.2497,
      "step": 66585
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7748435139656067,
      "learning_rate": 0.0004846174167915105,
      "loss": 2.149,
      "step": 66586
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7422330379486084,
      "learning_rate": 0.00048461419250450655,
      "loss": 2.237,
      "step": 66587
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7921363115310669,
      "learning_rate": 0.00048461096818317936,
      "loss": 2.0972,
      "step": 66588
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.8045016527175903,
      "learning_rate": 0.0004846077438275295,
      "loss": 2.297,
      "step": 66589
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.802202045917511,
      "learning_rate": 0.0004846045194375576,
      "loss": 2.3364,
      "step": 66590
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7754592895507812,
      "learning_rate": 0.00048460129501326423,
      "loss": 2.2945,
      "step": 66591
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7700011134147644,
      "learning_rate": 0.00048459807055465,
      "loss": 1.9514,
      "step": 66592
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7195154428482056,
      "learning_rate": 0.00048459484606171545,
      "loss": 2.2296,
      "step": 66593
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7409312129020691,
      "learning_rate": 0.00048459162153446125,
      "loss": 2.0305,
      "step": 66594
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7114846110343933,
      "learning_rate": 0.000484588396972888,
      "loss": 2.1335,
      "step": 66595
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7478323578834534,
      "learning_rate": 0.00048458517237699625,
      "loss": 2.1268,
      "step": 66596
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7933059334754944,
      "learning_rate": 0.0004845819477467867,
      "loss": 2.1956,
      "step": 66597
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.8053691387176514,
      "learning_rate": 0.0004845787230822598,
      "loss": 2.0609,
      "step": 66598
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.8476147651672363,
      "learning_rate": 0.00048457549838341624,
      "loss": 2.2014,
      "step": 66599
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7601375579833984,
      "learning_rate": 0.0004845722736502567,
      "loss": 2.0543,
      "step": 66600
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7636531591415405,
      "learning_rate": 0.0004845690488827816,
      "loss": 2.0596,
      "step": 66601
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.838455319404602,
      "learning_rate": 0.0004845658240809917,
      "loss": 2.203,
      "step": 66602
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7426788210868835,
      "learning_rate": 0.00048456259924488745,
      "loss": 2.4511,
      "step": 66603
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.0291173458099365,
      "learning_rate": 0.00048455937437446965,
      "loss": 2.0423,
      "step": 66604
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.8248998522758484,
      "learning_rate": 0.0004845561494697387,
      "loss": 2.3363,
      "step": 66605
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7142850160598755,
      "learning_rate": 0.0004845529245306953,
      "loss": 2.0968,
      "step": 66606
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.782337486743927,
      "learning_rate": 0.00048454969955734,
      "loss": 2.2847,
      "step": 66607
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7972157001495361,
      "learning_rate": 0.0004845464745496734,
      "loss": 2.3089,
      "step": 66608
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.8128120303153992,
      "learning_rate": 0.0004845432495076961,
      "loss": 2.2945,
      "step": 66609
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7686552405357361,
      "learning_rate": 0.00048454002443140893,
      "loss": 1.8744,
      "step": 66610
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7062019109725952,
      "learning_rate": 0.00048453679932081205,
      "loss": 2.1295,
      "step": 66611
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7737067937850952,
      "learning_rate": 0.0004845335741759064,
      "loss": 2.4423,
      "step": 66612
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7793692946434021,
      "learning_rate": 0.0004845303489966925,
      "loss": 2.3168,
      "step": 66613
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7296956777572632,
      "learning_rate": 0.00048452712378317084,
      "loss": 2.3341,
      "step": 66614
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.8319305181503296,
      "learning_rate": 0.00048452389853534216,
      "loss": 2.4756,
      "step": 66615
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7935264110565186,
      "learning_rate": 0.0004845206732532071,
      "loss": 2.082,
      "step": 66616
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.8699133992195129,
      "learning_rate": 0.00048451744793676603,
      "loss": 2.1102,
      "step": 66617
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7563970685005188,
      "learning_rate": 0.0004845142225860197,
      "loss": 2.1187,
      "step": 66618
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7695271372795105,
      "learning_rate": 0.00048451099720096867,
      "loss": 2.3787,
      "step": 66619
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.8013922572135925,
      "learning_rate": 0.0004845077717816136,
      "loss": 2.2556,
      "step": 66620
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.8014932870864868,
      "learning_rate": 0.00048450454632795507,
      "loss": 2.3469,
      "step": 66621
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.8637850880622864,
      "learning_rate": 0.00048450132083999365,
      "loss": 2.2147,
      "step": 66622
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7988260984420776,
      "learning_rate": 0.00048449809531773,
      "loss": 2.2172,
      "step": 66623
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7208066582679749,
      "learning_rate": 0.00048449486976116454,
      "loss": 2.2372,
      "step": 66624
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.790809690952301,
      "learning_rate": 0.00048449164417029814,
      "loss": 2.1899,
      "step": 66625
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7826312780380249,
      "learning_rate": 0.00048448841854513116,
      "loss": 2.1462,
      "step": 66626
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.8463528752326965,
      "learning_rate": 0.00048448519288566436,
      "loss": 2.08,
      "step": 66627
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7156538963317871,
      "learning_rate": 0.0004844819671918982,
      "loss": 2.0495,
      "step": 66628
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.8022211790084839,
      "learning_rate": 0.0004844787414638335,
      "loss": 2.2513,
      "step": 66629
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7712193131446838,
      "learning_rate": 0.00048447551570147057,
      "loss": 2.2232,
      "step": 66630
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7609055042266846,
      "learning_rate": 0.0004844722899048103,
      "loss": 2.0532,
      "step": 66631
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7333406209945679,
      "learning_rate": 0.0004844690640738531,
      "loss": 2.3428,
      "step": 66632
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7365673780441284,
      "learning_rate": 0.0004844658382085996,
      "loss": 2.1592,
      "step": 66633
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.823188841342926,
      "learning_rate": 0.00048446261230905037,
      "loss": 2.3964,
      "step": 66634
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7933433055877686,
      "learning_rate": 0.0004844593863752061,
      "loss": 2.3061,
      "step": 66635
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7868130207061768,
      "learning_rate": 0.0004844561604070674,
      "loss": 2.4085,
      "step": 66636
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.797597348690033,
      "learning_rate": 0.0004844529344046347,
      "loss": 2.1473,
      "step": 66637
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7868232727050781,
      "learning_rate": 0.00048444970836790884,
      "loss": 2.3586,
      "step": 66638
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.8200173377990723,
      "learning_rate": 0.0004844464822968902,
      "loss": 2.0299,
      "step": 66639
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7134116291999817,
      "learning_rate": 0.0004844432561915796,
      "loss": 2.2354,
      "step": 66640
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.6983765363693237,
      "learning_rate": 0.0004844400300519774,
      "loss": 2.1194,
      "step": 66641
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.6918426752090454,
      "learning_rate": 0.00048443680387808436,
      "loss": 2.2868,
      "step": 66642
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.6906249523162842,
      "learning_rate": 0.00048443357766990105,
      "loss": 2.2038,
      "step": 66643
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7477974891662598,
      "learning_rate": 0.00048443035142742806,
      "loss": 2.2601,
      "step": 66644
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.785403847694397,
      "learning_rate": 0.00048442712515066593,
      "loss": 1.9805,
      "step": 66645
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7400965690612793,
      "learning_rate": 0.0004844238988396153,
      "loss": 2.1627,
      "step": 66646
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.737829327583313,
      "learning_rate": 0.00048442067249427695,
      "loss": 2.1591,
      "step": 66647
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7642929553985596,
      "learning_rate": 0.00048441744611465114,
      "loss": 2.0799,
      "step": 66648
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7270846366882324,
      "learning_rate": 0.00048441421970073874,
      "loss": 2.2767,
      "step": 66649
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7737936973571777,
      "learning_rate": 0.0004844109932525402,
      "loss": 2.231,
      "step": 66650
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7755049467086792,
      "learning_rate": 0.0004844077667700562,
      "loss": 2.1089,
      "step": 66651
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7411619424819946,
      "learning_rate": 0.0004844045402532874,
      "loss": 2.1577,
      "step": 66652
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.768293023109436,
      "learning_rate": 0.00048440131370223425,
      "loss": 2.3514,
      "step": 66653
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.840164065361023,
      "learning_rate": 0.00048439808711689733,
      "loss": 2.0994,
      "step": 66654
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7377912998199463,
      "learning_rate": 0.00048439486049727735,
      "loss": 2.2114,
      "step": 66655
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7354313731193542,
      "learning_rate": 0.00048439163384337495,
      "loss": 2.0101,
      "step": 66656
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7602818012237549,
      "learning_rate": 0.0004843884071551906,
      "loss": 1.9501,
      "step": 66657
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7218285202980042,
      "learning_rate": 0.0004843851804327251,
      "loss": 2.2653,
      "step": 66658
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.707129716873169,
      "learning_rate": 0.0004843819536759788,
      "loss": 2.3322,
      "step": 66659
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7420199513435364,
      "learning_rate": 0.0004843787268849524,
      "loss": 2.1621,
      "step": 66660
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7275144457817078,
      "learning_rate": 0.0004843755000596466,
      "loss": 2.211,
      "step": 66661
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.8423999547958374,
      "learning_rate": 0.0004843722732000619,
      "loss": 1.9338,
      "step": 66662
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7958509922027588,
      "learning_rate": 0.00048436904630619885,
      "loss": 2.1287,
      "step": 66663
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.8259373903274536,
      "learning_rate": 0.0004843658193780581,
      "loss": 2.4536,
      "step": 66664
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.6883259415626526,
      "learning_rate": 0.0004843625924156403,
      "loss": 2.4339,
      "step": 66665
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7090427875518799,
      "learning_rate": 0.00048435936541894604,
      "loss": 2.0664,
      "step": 66666
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.853952169418335,
      "learning_rate": 0.0004843561383879759,
      "loss": 2.2119,
      "step": 66667
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7619284987449646,
      "learning_rate": 0.00048435291132273045,
      "loss": 2.2115,
      "step": 66668
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7284162640571594,
      "learning_rate": 0.00048434968422321034,
      "loss": 2.2896,
      "step": 66669
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.769569993019104,
      "learning_rate": 0.000484346457089416,
      "loss": 2.2562,
      "step": 66670
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7412376999855042,
      "learning_rate": 0.0004843432299213483,
      "loss": 2.3082,
      "step": 66671
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7274236679077148,
      "learning_rate": 0.00048434000271900773,
      "loss": 2.1686,
      "step": 66672
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7134795188903809,
      "learning_rate": 0.00048433677548239486,
      "loss": 1.9812,
      "step": 66673
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7678437829017639,
      "learning_rate": 0.0004843335482115103,
      "loss": 2.0841,
      "step": 66674
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7985376119613647,
      "learning_rate": 0.00048433032090635465,
      "loss": 2.3079,
      "step": 66675
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7506991624832153,
      "learning_rate": 0.0004843270935669285,
      "loss": 2.1412,
      "step": 66676
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7765495181083679,
      "learning_rate": 0.0004843238661932325,
      "loss": 2.1181,
      "step": 66677
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.777337372303009,
      "learning_rate": 0.00048432063878526714,
      "loss": 2.2614,
      "step": 66678
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7182474136352539,
      "learning_rate": 0.0004843174113430331,
      "loss": 2.5613,
      "step": 66679
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.8378973007202148,
      "learning_rate": 0.00048431418386653106,
      "loss": 2.2194,
      "step": 66680
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7900293469429016,
      "learning_rate": 0.0004843109563557615,
      "loss": 1.9791,
      "step": 66681
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7381693720817566,
      "learning_rate": 0.00048430772881072505,
      "loss": 1.8849,
      "step": 66682
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7182626724243164,
      "learning_rate": 0.0004843045012314223,
      "loss": 2.2715,
      "step": 66683
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7682068347930908,
      "learning_rate": 0.0004843012736178539,
      "loss": 2.1603,
      "step": 66684
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7065502405166626,
      "learning_rate": 0.00048429804597002036,
      "loss": 2.0605,
      "step": 66685
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.8520026803016663,
      "learning_rate": 0.00048429481828792236,
      "loss": 2.2977,
      "step": 66686
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7593370079994202,
      "learning_rate": 0.00048429159057156043,
      "loss": 2.2338,
      "step": 66687
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7842921018600464,
      "learning_rate": 0.0004842883628209353,
      "loss": 2.2045,
      "step": 66688
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.8088041543960571,
      "learning_rate": 0.00048428513503604744,
      "loss": 2.4165,
      "step": 66689
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7362887859344482,
      "learning_rate": 0.00048428190721689745,
      "loss": 2.037,
      "step": 66690
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.810367226600647,
      "learning_rate": 0.000484278679363486,
      "loss": 2.1927,
      "step": 66691
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.8564998507499695,
      "learning_rate": 0.00048427545147581376,
      "loss": 2.2942,
      "step": 66692
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7909547090530396,
      "learning_rate": 0.00048427222355388114,
      "loss": 2.0345,
      "step": 66693
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7812850475311279,
      "learning_rate": 0.0004842689955976888,
      "loss": 2.1726,
      "step": 66694
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.777332067489624,
      "learning_rate": 0.00048426576760723743,
      "loss": 2.2069,
      "step": 66695
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.821959912776947,
      "learning_rate": 0.00048426253958252765,
      "loss": 2.0862,
      "step": 66696
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7110419273376465,
      "learning_rate": 0.00048425931152355995,
      "loss": 2.3377,
      "step": 66697
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.8272507786750793,
      "learning_rate": 0.00048425608343033487,
      "loss": 2.1705,
      "step": 66698
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7679585218429565,
      "learning_rate": 0.0004842528553028532,
      "loss": 2.4284,
      "step": 66699
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.6603538393974304,
      "learning_rate": 0.0004842496271411154,
      "loss": 2.135,
      "step": 66700
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.745400071144104,
      "learning_rate": 0.0004842463989451221,
      "loss": 2.2299,
      "step": 66701
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7479777336120605,
      "learning_rate": 0.00048424317071487396,
      "loss": 2.2409,
      "step": 66702
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.8135178685188293,
      "learning_rate": 0.0004842399424503715,
      "loss": 2.1185,
      "step": 66703
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7289515137672424,
      "learning_rate": 0.0004842367141516153,
      "loss": 2.0084,
      "step": 66704
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7925931811332703,
      "learning_rate": 0.0004842334858186062,
      "loss": 2.2152,
      "step": 66705
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7795125246047974,
      "learning_rate": 0.00048423025745134447,
      "loss": 2.2882,
      "step": 66706
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7010983824729919,
      "learning_rate": 0.0004842270290498309,
      "loss": 2.189,
      "step": 66707
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.9022027850151062,
      "learning_rate": 0.000484223800614066,
      "loss": 2.217,
      "step": 66708
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7691541910171509,
      "learning_rate": 0.0004842205721440506,
      "loss": 2.158,
      "step": 66709
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7894856333732605,
      "learning_rate": 0.0004842173436397849,
      "loss": 2.0955,
      "step": 66710
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.8564199209213257,
      "learning_rate": 0.0004842141151012699,
      "loss": 2.1387,
      "step": 66711
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.1889004707336426,
      "learning_rate": 0.0004842108865285059,
      "loss": 2.2458,
      "step": 66712
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.9247046709060669,
      "learning_rate": 0.0004842076579214936,
      "loss": 2.1917,
      "step": 66713
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7430123686790466,
      "learning_rate": 0.0004842044292802337,
      "loss": 2.3304,
      "step": 66714
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.8060213327407837,
      "learning_rate": 0.00048420120060472665,
      "loss": 2.1493,
      "step": 66715
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.8978748321533203,
      "learning_rate": 0.00048419797189497315,
      "loss": 2.4521,
      "step": 66716
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.8214575052261353,
      "learning_rate": 0.0004841947431509738,
      "loss": 2.2092,
      "step": 66717
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.8240448236465454,
      "learning_rate": 0.00048419151437272906,
      "loss": 2.1828,
      "step": 66718
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.8200860023498535,
      "learning_rate": 0.0004841882855602397,
      "loss": 2.3533,
      "step": 66719
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.954043984413147,
      "learning_rate": 0.00048418505671350633,
      "loss": 2.0866,
      "step": 66720
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.8006298542022705,
      "learning_rate": 0.00048418182783252945,
      "loss": 2.2365,
      "step": 66721
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.0953519344329834,
      "learning_rate": 0.00048417859891730967,
      "loss": 2.1833,
      "step": 66722
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.9596356749534607,
      "learning_rate": 0.0004841753699678476,
      "loss": 2.1994,
      "step": 66723
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.8110076785087585,
      "learning_rate": 0.00048417214098414385,
      "loss": 2.2375,
      "step": 66724
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7686590552330017,
      "learning_rate": 0.00048416891196619905,
      "loss": 2.1719,
      "step": 66725
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.8302915692329407,
      "learning_rate": 0.00048416568291401375,
      "loss": 2.04,
      "step": 66726
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.8053625226020813,
      "learning_rate": 0.0004841624538275886,
      "loss": 2.288,
      "step": 66727
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.8972323536872864,
      "learning_rate": 0.0004841592247069242,
      "loss": 2.2743,
      "step": 66728
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.8190818428993225,
      "learning_rate": 0.00048415599555202103,
      "loss": 2.0247,
      "step": 66729
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.844952404499054,
      "learning_rate": 0.00048415276636287985,
      "loss": 2.3569,
      "step": 66730
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.9321897625923157,
      "learning_rate": 0.00048414953713950113,
      "loss": 2.2966,
      "step": 66731
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.8007972836494446,
      "learning_rate": 0.0004841463078818857,
      "loss": 2.1015,
      "step": 66732
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.8032857775688171,
      "learning_rate": 0.0004841430785900338,
      "loss": 2.4107,
      "step": 66733
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.749433159828186,
      "learning_rate": 0.0004841398492639463,
      "loss": 2.2703,
      "step": 66734
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7919644713401794,
      "learning_rate": 0.0004841366199036238,
      "loss": 2.3886,
      "step": 66735
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.9201961159706116,
      "learning_rate": 0.0004841333905090667,
      "loss": 2.5402,
      "step": 66736
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7300511598587036,
      "learning_rate": 0.0004841301610802758,
      "loss": 2.2995,
      "step": 66737
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.8309223055839539,
      "learning_rate": 0.00048412693161725164,
      "loss": 2.5031,
      "step": 66738
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7967931032180786,
      "learning_rate": 0.0004841237021199948,
      "loss": 2.3227,
      "step": 66739
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.8498629331588745,
      "learning_rate": 0.00048412047258850587,
      "loss": 2.5146,
      "step": 66740
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.8762251734733582,
      "learning_rate": 0.0004841172430227854,
      "loss": 2.2272,
      "step": 66741
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.8684508204460144,
      "learning_rate": 0.0004841140134228342,
      "loss": 2.4238,
      "step": 66742
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.765251874923706,
      "learning_rate": 0.00048411078378865263,
      "loss": 2.241,
      "step": 66743
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.9966456294059753,
      "learning_rate": 0.0004841075541202414,
      "loss": 2.2188,
      "step": 66744
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.700632631778717,
      "learning_rate": 0.00048410432441760114,
      "loss": 2.0713,
      "step": 66745
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.0016026496887207,
      "learning_rate": 0.0004841010946807324,
      "loss": 2.249,
      "step": 66746
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.8347336649894714,
      "learning_rate": 0.00048409786490963584,
      "loss": 1.9624,
      "step": 66747
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7481525540351868,
      "learning_rate": 0.0004840946351043118,
      "loss": 2.1509,
      "step": 66748
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.8347729444503784,
      "learning_rate": 0.0004840914052647613,
      "loss": 2.1128,
      "step": 66749
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.9123613834381104,
      "learning_rate": 0.0004840881753909847,
      "loss": 2.282,
      "step": 66750
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7287871837615967,
      "learning_rate": 0.0004840849454829826,
      "loss": 2.2591,
      "step": 66751
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7932249307632446,
      "learning_rate": 0.00048408171554075564,
      "loss": 2.2434,
      "step": 66752
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7770662903785706,
      "learning_rate": 0.0004840784855643044,
      "loss": 2.2211,
      "step": 66753
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.755233645439148,
      "learning_rate": 0.00048407525555362957,
      "loss": 2.1444,
      "step": 66754
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7982033491134644,
      "learning_rate": 0.0004840720255087315,
      "loss": 2.4673,
      "step": 66755
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.717008650302887,
      "learning_rate": 0.0004840687954296112,
      "loss": 1.937,
      "step": 66756
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.6924856305122375,
      "learning_rate": 0.0004840655653162689,
      "loss": 2.2345,
      "step": 66757
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7177797555923462,
      "learning_rate": 0.00048406233516870535,
      "loss": 2.1586,
      "step": 66758
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7450863718986511,
      "learning_rate": 0.0004840591049869211,
      "loss": 2.3116,
      "step": 66759
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7503740191459656,
      "learning_rate": 0.0004840558747709169,
      "loss": 2.2153,
      "step": 66760
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7955161929130554,
      "learning_rate": 0.0004840526445206931,
      "loss": 2.1506,
      "step": 66761
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7206764817237854,
      "learning_rate": 0.0004840494142362505,
      "loss": 2.2621,
      "step": 66762
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7965302467346191,
      "learning_rate": 0.00048404618391758973,
      "loss": 2.3848,
      "step": 66763
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.8695478439331055,
      "learning_rate": 0.0004840429535647112,
      "loss": 2.3343,
      "step": 66764
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7052538394927979,
      "learning_rate": 0.0004840397231776156,
      "loss": 2.5422,
      "step": 66765
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7335253953933716,
      "learning_rate": 0.0004840364927563036,
      "loss": 2.1909,
      "step": 66766
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7490602135658264,
      "learning_rate": 0.00048403326230077575,
      "loss": 2.2452,
      "step": 66767
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7606136798858643,
      "learning_rate": 0.0004840300318110325,
      "loss": 2.4531,
      "step": 66768
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7457542419433594,
      "learning_rate": 0.0004840268012870748,
      "loss": 2.1462,
      "step": 66769
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7886890769004822,
      "learning_rate": 0.0004840235707289029,
      "loss": 2.1106,
      "step": 66770
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.736038863658905,
      "learning_rate": 0.0004840203401365176,
      "loss": 2.3254,
      "step": 66771
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7202576994895935,
      "learning_rate": 0.00048401710950991947,
      "loss": 2.2606,
      "step": 66772
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7307723760604858,
      "learning_rate": 0.000484013878849109,
      "loss": 2.1439,
      "step": 66773
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7939351201057434,
      "learning_rate": 0.00048401064815408693,
      "loss": 2.1137,
      "step": 66774
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7961139678955078,
      "learning_rate": 0.0004840074174248538,
      "loss": 2.1613,
      "step": 66775
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.8939075469970703,
      "learning_rate": 0.00048400418666141024,
      "loss": 2.1453,
      "step": 66776
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.9014990925788879,
      "learning_rate": 0.00048400095586375675,
      "loss": 2.2328,
      "step": 66777
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7623984217643738,
      "learning_rate": 0.0004839977250318942,
      "loss": 2.1243,
      "step": 66778
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.8865535259246826,
      "learning_rate": 0.0004839944941658228,
      "loss": 2.3289,
      "step": 66779
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.8204598426818848,
      "learning_rate": 0.0004839912632655434,
      "loss": 2.1548,
      "step": 66780
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7109872102737427,
      "learning_rate": 0.00048398803233105664,
      "loss": 2.2091,
      "step": 66781
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.794863224029541,
      "learning_rate": 0.0004839848013623629,
      "loss": 2.2369,
      "step": 66782
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7136666774749756,
      "learning_rate": 0.0004839815703594629,
      "loss": 2.0772,
      "step": 66783
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7790993452072144,
      "learning_rate": 0.0004839783393223574,
      "loss": 2.3752,
      "step": 66784
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7503196597099304,
      "learning_rate": 0.0004839751082510468,
      "loss": 2.0975,
      "step": 66785
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7922395467758179,
      "learning_rate": 0.0004839718771455317,
      "loss": 2.2656,
      "step": 66786
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7417314648628235,
      "learning_rate": 0.00048396864600581285,
      "loss": 2.0868,
      "step": 66787
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.782145619392395,
      "learning_rate": 0.00048396541483189067,
      "loss": 2.1793,
      "step": 66788
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.6915330290794373,
      "learning_rate": 0.0004839621836237659,
      "loss": 1.8966,
      "step": 66789
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.9748258590698242,
      "learning_rate": 0.00048395895238143906,
      "loss": 2.0486,
      "step": 66790
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7273184657096863,
      "learning_rate": 0.00048395572110491075,
      "loss": 2.1952,
      "step": 66791
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7640300393104553,
      "learning_rate": 0.0004839524897941816,
      "loss": 2.278,
      "step": 66792
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7431037425994873,
      "learning_rate": 0.0004839492584492523,
      "loss": 2.2749,
      "step": 66793
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7681423425674438,
      "learning_rate": 0.0004839460270701233,
      "loss": 2.2175,
      "step": 66794
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.725461483001709,
      "learning_rate": 0.00048394279565679523,
      "loss": 2.288,
      "step": 66795
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7704618573188782,
      "learning_rate": 0.00048393956420926873,
      "loss": 2.0411,
      "step": 66796
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7411687970161438,
      "learning_rate": 0.00048393633272754443,
      "loss": 2.033,
      "step": 66797
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.8596869707107544,
      "learning_rate": 0.0004839331012116229,
      "loss": 2.4313,
      "step": 66798
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.8599493503570557,
      "learning_rate": 0.00048392986966150467,
      "loss": 2.1526,
      "step": 66799
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.8651026487350464,
      "learning_rate": 0.00048392663807719045,
      "loss": 2.2522,
      "step": 66800
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7621732354164124,
      "learning_rate": 0.0004839234064586808,
      "loss": 2.2568,
      "step": 66801
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7525411248207092,
      "learning_rate": 0.0004839201748059763,
      "loss": 2.0504,
      "step": 66802
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.0145741701126099,
      "learning_rate": 0.00048391694311907755,
      "loss": 2.3607,
      "step": 66803
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.79044109582901,
      "learning_rate": 0.0004839137113979852,
      "loss": 2.4743,
      "step": 66804
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7647911310195923,
      "learning_rate": 0.00048391047964269985,
      "loss": 1.9968,
      "step": 66805
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.8313052654266357,
      "learning_rate": 0.0004839072478532221,
      "loss": 2.1996,
      "step": 66806
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.8293166756629944,
      "learning_rate": 0.0004839040160295524,
      "loss": 2.3805,
      "step": 66807
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.905828058719635,
      "learning_rate": 0.00048390078417169156,
      "loss": 2.1369,
      "step": 66808
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.4038022756576538,
      "learning_rate": 0.0004838975522796401,
      "loss": 2.0473,
      "step": 66809
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.776709794998169,
      "learning_rate": 0.0004838943203533985,
      "loss": 2.2242,
      "step": 66810
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.8793426752090454,
      "learning_rate": 0.00048389108839296755,
      "loss": 2.157,
      "step": 66811
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.8144680857658386,
      "learning_rate": 0.00048388785639834776,
      "loss": 2.2601,
      "step": 66812
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7691790461540222,
      "learning_rate": 0.0004838846243695398,
      "loss": 2.2721,
      "step": 66813
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.8988201022148132,
      "learning_rate": 0.00048388139230654415,
      "loss": 2.3441,
      "step": 66814
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7932606339454651,
      "learning_rate": 0.00048387816020936154,
      "loss": 2.2201,
      "step": 66815
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7892478704452515,
      "learning_rate": 0.0004838749280779925,
      "loss": 2.4634,
      "step": 66816
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.8529158234596252,
      "learning_rate": 0.00048387169591243757,
      "loss": 2.1172,
      "step": 66817
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7202576994895935,
      "learning_rate": 0.00048386846371269745,
      "loss": 2.2497,
      "step": 66818
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.8888037204742432,
      "learning_rate": 0.0004838652314787727,
      "loss": 2.1388,
      "step": 66819
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7540076971054077,
      "learning_rate": 0.00048386199921066406,
      "loss": 2.367,
      "step": 66820
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.6718700528144836,
      "learning_rate": 0.00048385876690837186,
      "loss": 2.2187,
      "step": 66821
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7717775702476501,
      "learning_rate": 0.00048385553457189694,
      "loss": 2.1043,
      "step": 66822
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7681344747543335,
      "learning_rate": 0.00048385230220123973,
      "loss": 2.2137,
      "step": 66823
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.9611677527427673,
      "learning_rate": 0.0004838490697964009,
      "loss": 2.1814,
      "step": 66824
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7906678318977356,
      "learning_rate": 0.0004838458373573811,
      "loss": 2.1168,
      "step": 66825
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7881866097450256,
      "learning_rate": 0.00048384260488418084,
      "loss": 1.9106,
      "step": 66826
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7709908485412598,
      "learning_rate": 0.0004838393723768009,
      "loss": 2.2165,
      "step": 66827
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.8576134443283081,
      "learning_rate": 0.0004838361398352416,
      "loss": 2.1991,
      "step": 66828
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.737423837184906,
      "learning_rate": 0.00048383290725950376,
      "loss": 2.1771,
      "step": 66829
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.862598180770874,
      "learning_rate": 0.00048382967464958795,
      "loss": 2.2361,
      "step": 66830
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.784841775894165,
      "learning_rate": 0.0004838264420054946,
      "loss": 2.0296,
      "step": 66831
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.862113356590271,
      "learning_rate": 0.0004838232093272246,
      "loss": 2.2145,
      "step": 66832
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.8029114603996277,
      "learning_rate": 0.00048381997661477833,
      "loss": 2.4327,
      "step": 66833
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7074652910232544,
      "learning_rate": 0.0004838167438681565,
      "loss": 2.0921,
      "step": 66834
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7891766428947449,
      "learning_rate": 0.00048381351108735954,
      "loss": 2.0387,
      "step": 66835
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7724156975746155,
      "learning_rate": 0.0004838102782723883,
      "loss": 2.3743,
      "step": 66836
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7561011910438538,
      "learning_rate": 0.0004838070454232432,
      "loss": 2.5266,
      "step": 66837
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7792650461196899,
      "learning_rate": 0.0004838038125399249,
      "loss": 2.1448,
      "step": 66838
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7889323234558105,
      "learning_rate": 0.00048380057962243403,
      "loss": 2.1866,
      "step": 66839
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7296691536903381,
      "learning_rate": 0.0004837973466707712,
      "loss": 2.0312,
      "step": 66840
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7500773072242737,
      "learning_rate": 0.00048379411368493685,
      "loss": 2.1868,
      "step": 66841
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7114553451538086,
      "learning_rate": 0.0004837908806649318,
      "loss": 2.1613,
      "step": 66842
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7765116095542908,
      "learning_rate": 0.0004837876476107566,
      "loss": 2.1673,
      "step": 66843
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.9187632203102112,
      "learning_rate": 0.00048378441452241175,
      "loss": 2.1416,
      "step": 66844
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.8625193238258362,
      "learning_rate": 0.0004837811813998979,
      "loss": 2.2671,
      "step": 66845
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7976981997489929,
      "learning_rate": 0.0004837779482432156,
      "loss": 2.2137,
      "step": 66846
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.8590934872627258,
      "learning_rate": 0.0004837747150523656,
      "loss": 2.165,
      "step": 66847
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.9867100119590759,
      "learning_rate": 0.0004837714818273484,
      "loss": 2.2454,
      "step": 66848
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.822723925113678,
      "learning_rate": 0.0004837682485681646,
      "loss": 2.322,
      "step": 66849
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7530022859573364,
      "learning_rate": 0.00048376501527481477,
      "loss": 2.2756,
      "step": 66850
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.9399475455284119,
      "learning_rate": 0.0004837617819472997,
      "loss": 2.2865,
      "step": 66851
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.027032732963562,
      "learning_rate": 0.0004837585485856197,
      "loss": 2.1705,
      "step": 66852
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.6984407305717468,
      "learning_rate": 0.00048375531518977553,
      "loss": 2.089,
      "step": 66853
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.8246746063232422,
      "learning_rate": 0.0004837520817597678,
      "loss": 2.0042,
      "step": 66854
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7735342383384705,
      "learning_rate": 0.0004837488482955971,
      "loss": 2.3914,
      "step": 66855
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7475823760032654,
      "learning_rate": 0.00048374561479726404,
      "loss": 2.2468,
      "step": 66856
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7159331440925598,
      "learning_rate": 0.00048374238126476916,
      "loss": 2.0911,
      "step": 66857
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.6773695349693298,
      "learning_rate": 0.0004837391476981131,
      "loss": 2.1032,
      "step": 66858
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.8048368096351624,
      "learning_rate": 0.00048373591409729645,
      "loss": 2.1505,
      "step": 66859
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.761187732219696,
      "learning_rate": 0.00048373268046231986,
      "loss": 2.2489,
      "step": 66860
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7063602805137634,
      "learning_rate": 0.00048372944679318396,
      "loss": 2.052,
      "step": 66861
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7876632213592529,
      "learning_rate": 0.00048372621308988924,
      "loss": 2.1527,
      "step": 66862
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.762927770614624,
      "learning_rate": 0.00048372297935243635,
      "loss": 2.2486,
      "step": 66863
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7926895022392273,
      "learning_rate": 0.00048371974558082583,
      "loss": 2.3428,
      "step": 66864
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.8947861790657043,
      "learning_rate": 0.00048371651177505834,
      "loss": 2.3263,
      "step": 66865
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.8404145836830139,
      "learning_rate": 0.0004837132779351346,
      "loss": 2.3498,
      "step": 66866
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7141275405883789,
      "learning_rate": 0.000483710044061055,
      "loss": 2.3533,
      "step": 66867
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.8131996393203735,
      "learning_rate": 0.0004837068101528202,
      "loss": 2.4055,
      "step": 66868
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7277369499206543,
      "learning_rate": 0.00048370357621043097,
      "loss": 2.1571,
      "step": 66869
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7895522713661194,
      "learning_rate": 0.00048370034223388765,
      "loss": 2.3365,
      "step": 66870
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7726221680641174,
      "learning_rate": 0.00048369710822319104,
      "loss": 2.1498,
      "step": 66871
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7917836308479309,
      "learning_rate": 0.00048369387417834163,
      "loss": 2.173,
      "step": 66872
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7818361520767212,
      "learning_rate": 0.0004836906400993401,
      "loss": 2.3349,
      "step": 66873
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7056397199630737,
      "learning_rate": 0.00048368740598618695,
      "loss": 2.0159,
      "step": 66874
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7693666219711304,
      "learning_rate": 0.000483684171838883,
      "loss": 2.3243,
      "step": 66875
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7894846200942993,
      "learning_rate": 0.0004836809376574285,
      "loss": 2.2147,
      "step": 66876
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.8157211542129517,
      "learning_rate": 0.00048367770344182433,
      "loss": 2.3801,
      "step": 66877
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7469246983528137,
      "learning_rate": 0.000483674469192071,
      "loss": 2.3657,
      "step": 66878
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7490823864936829,
      "learning_rate": 0.0004836712349081691,
      "loss": 2.2574,
      "step": 66879
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.74793940782547,
      "learning_rate": 0.0004836680005901193,
      "loss": 2.1061,
      "step": 66880
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7789164781570435,
      "learning_rate": 0.0004836647662379222,
      "loss": 2.1937,
      "step": 66881
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7970839738845825,
      "learning_rate": 0.0004836615318515782,
      "loss": 2.1795,
      "step": 66882
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7897595167160034,
      "learning_rate": 0.00048365829743108815,
      "loss": 2.2239,
      "step": 66883
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7861960530281067,
      "learning_rate": 0.00048365506297645255,
      "loss": 2.407,
      "step": 66884
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7109039425849915,
      "learning_rate": 0.000483651828487672,
      "loss": 2.2939,
      "step": 66885
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7630193829536438,
      "learning_rate": 0.0004836485939647471,
      "loss": 2.3321,
      "step": 66886
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7337875366210938,
      "learning_rate": 0.0004836453594076785,
      "loss": 2.1239,
      "step": 66887
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7513931393623352,
      "learning_rate": 0.00048364212481646673,
      "loss": 2.2541,
      "step": 66888
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7636544108390808,
      "learning_rate": 0.0004836388901911124,
      "loss": 2.0592,
      "step": 66889
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7546977996826172,
      "learning_rate": 0.00048363565553161613,
      "loss": 2.1468,
      "step": 66890
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.8790063261985779,
      "learning_rate": 0.0004836324208379785,
      "loss": 2.2174,
      "step": 66891
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7449779510498047,
      "learning_rate": 0.00048362918611020024,
      "loss": 2.1461,
      "step": 66892
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.746086835861206,
      "learning_rate": 0.0004836259513482818,
      "loss": 2.2377,
      "step": 66893
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.8411172032356262,
      "learning_rate": 0.0004836227165522238,
      "loss": 2.1119,
      "step": 66894
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.0215303897857666,
      "learning_rate": 0.0004836194817220269,
      "loss": 2.3215,
      "step": 66895
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.8978782296180725,
      "learning_rate": 0.00048361624685769165,
      "loss": 2.1465,
      "step": 66896
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7550108432769775,
      "learning_rate": 0.00048361301195921875,
      "loss": 2.4081,
      "step": 66897
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7666019797325134,
      "learning_rate": 0.00048360977702660866,
      "loss": 2.0578,
      "step": 66898
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7213390469551086,
      "learning_rate": 0.00048360654205986203,
      "loss": 2.1571,
      "step": 66899
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7004711031913757,
      "learning_rate": 0.00048360330705897946,
      "loss": 2.2808,
      "step": 66900
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.8098042011260986,
      "learning_rate": 0.00048360007202396166,
      "loss": 2.2122,
      "step": 66901
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7481071352958679,
      "learning_rate": 0.00048359683695480904,
      "loss": 2.0322,
      "step": 66902
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7820387482643127,
      "learning_rate": 0.00048359360185152244,
      "loss": 2.2145,
      "step": 66903
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.745148241519928,
      "learning_rate": 0.0004835903667141022,
      "loss": 2.2535,
      "step": 66904
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.867953360080719,
      "learning_rate": 0.00048358713154254905,
      "loss": 2.1504,
      "step": 66905
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.6998430490493774,
      "learning_rate": 0.00048358389633686367,
      "loss": 2.2182,
      "step": 66906
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.8161929845809937,
      "learning_rate": 0.0004835806610970464,
      "loss": 2.1657,
      "step": 66907
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7764623165130615,
      "learning_rate": 0.0004835774258230982,
      "loss": 2.3943,
      "step": 66908
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7114356756210327,
      "learning_rate": 0.00048357419051501947,
      "loss": 2.0625,
      "step": 66909
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7536186575889587,
      "learning_rate": 0.00048357095517281075,
      "loss": 2.2957,
      "step": 66910
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.8045496940612793,
      "learning_rate": 0.0004835677197964728,
      "loss": 2.4074,
      "step": 66911
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7495805025100708,
      "learning_rate": 0.0004835644843860061,
      "loss": 2.0045,
      "step": 66912
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7786208987236023,
      "learning_rate": 0.00048356124894141135,
      "loss": 2.2503,
      "step": 66913
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7754896283149719,
      "learning_rate": 0.00048355801346268906,
      "loss": 2.0725,
      "step": 66914
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7528694272041321,
      "learning_rate": 0.0004835547779498399,
      "loss": 2.1856,
      "step": 66915
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7400290966033936,
      "learning_rate": 0.0004835515424028644,
      "loss": 2.1656,
      "step": 66916
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7659862041473389,
      "learning_rate": 0.00048354830682176324,
      "loss": 2.2236,
      "step": 66917
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.8059578537940979,
      "learning_rate": 0.00048354507120653697,
      "loss": 2.1574,
      "step": 66918
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7175180315971375,
      "learning_rate": 0.0004835418355571862,
      "loss": 2.1027,
      "step": 66919
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7411078214645386,
      "learning_rate": 0.00048353859987371146,
      "loss": 2.3594,
      "step": 66920
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.8745980858802795,
      "learning_rate": 0.00048353536415611363,
      "loss": 2.0495,
      "step": 66921
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7252039313316345,
      "learning_rate": 0.0004835321284043929,
      "loss": 2.196,
      "step": 66922
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7457798719406128,
      "learning_rate": 0.0004835288926185502,
      "loss": 2.2954,
      "step": 66923
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7511622905731201,
      "learning_rate": 0.000483525656798586,
      "loss": 2.1927,
      "step": 66924
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7375520467758179,
      "learning_rate": 0.0004835224209445009,
      "loss": 2.2643,
      "step": 66925
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7002493739128113,
      "learning_rate": 0.0004835191850562955,
      "loss": 2.2779,
      "step": 66926
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7707048058509827,
      "learning_rate": 0.00048351594913397043,
      "loss": 2.3578,
      "step": 66927
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7479676008224487,
      "learning_rate": 0.00048351271317752633,
      "loss": 2.3343,
      "step": 66928
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7520739436149597,
      "learning_rate": 0.00048350947718696374,
      "loss": 2.2975,
      "step": 66929
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7617012858390808,
      "learning_rate": 0.00048350624116228326,
      "loss": 2.089,
      "step": 66930
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7983035445213318,
      "learning_rate": 0.00048350300510348554,
      "loss": 2.0255,
      "step": 66931
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7959721684455872,
      "learning_rate": 0.000483499769010571,
      "loss": 2.2407,
      "step": 66932
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.735997200012207,
      "learning_rate": 0.0004834965328835406,
      "loss": 2.1608,
      "step": 66933
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7304949164390564,
      "learning_rate": 0.00048349329672239463,
      "loss": 2.0751,
      "step": 66934
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.755780816078186,
      "learning_rate": 0.0004834900605271337,
      "loss": 2.2051,
      "step": 66935
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7426193356513977,
      "learning_rate": 0.0004834868242977586,
      "loss": 2.0154,
      "step": 66936
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.8773377537727356,
      "learning_rate": 0.0004834835880342698,
      "loss": 2.2613,
      "step": 66937
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7048577666282654,
      "learning_rate": 0.000483480351736668,
      "loss": 2.1517,
      "step": 66938
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7540620565414429,
      "learning_rate": 0.0004834771154049537,
      "loss": 2.2087,
      "step": 66939
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7432090044021606,
      "learning_rate": 0.00048347387903912756,
      "loss": 2.2022,
      "step": 66940
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7089577913284302,
      "learning_rate": 0.00048347064263919013,
      "loss": 2.1629,
      "step": 66941
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.723724365234375,
      "learning_rate": 0.0004834674062051421,
      "loss": 2.2892,
      "step": 66942
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7654197812080383,
      "learning_rate": 0.000483464169736984,
      "loss": 2.2805,
      "step": 66943
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.8504801392555237,
      "learning_rate": 0.0004834609332347163,
      "loss": 2.2959,
      "step": 66944
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.8010228276252747,
      "learning_rate": 0.00048345769669833993,
      "loss": 2.1366,
      "step": 66945
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.8278170824050903,
      "learning_rate": 0.0004834544601278553,
      "loss": 2.3352,
      "step": 66946
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7802865505218506,
      "learning_rate": 0.0004834512235232629,
      "loss": 2.3112,
      "step": 66947
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7141077518463135,
      "learning_rate": 0.00048344798688456357,
      "loss": 2.2166,
      "step": 66948
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7805044651031494,
      "learning_rate": 0.00048344475021175776,
      "loss": 2.1888,
      "step": 66949
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7321459650993347,
      "learning_rate": 0.0004834415135048461,
      "loss": 2.2911,
      "step": 66950
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.9035871624946594,
      "learning_rate": 0.0004834382767638292,
      "loss": 2.3213,
      "step": 66951
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7391541004180908,
      "learning_rate": 0.00048343503998870765,
      "loss": 2.2321,
      "step": 66952
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7850450873374939,
      "learning_rate": 0.00048343180317948206,
      "loss": 2.1756,
      "step": 66953
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7995269894599915,
      "learning_rate": 0.0004834285663361531,
      "loss": 2.3257,
      "step": 66954
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7429932355880737,
      "learning_rate": 0.00048342532945872126,
      "loss": 2.0597,
      "step": 66955
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.8837010860443115,
      "learning_rate": 0.00048342209254718715,
      "loss": 2.1847,
      "step": 66956
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7524463534355164,
      "learning_rate": 0.00048341885560155147,
      "loss": 2.1516,
      "step": 66957
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7852060198783875,
      "learning_rate": 0.0004834156186218148,
      "loss": 2.2799,
      "step": 66958
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.9567903280258179,
      "learning_rate": 0.0004834123816079776,
      "loss": 2.092,
      "step": 66959
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.9430578351020813,
      "learning_rate": 0.0004834091445600406,
      "loss": 2.2453,
      "step": 66960
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.9616695642471313,
      "learning_rate": 0.00048340590747800443,
      "loss": 2.1394,
      "step": 66961
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.8172184228897095,
      "learning_rate": 0.00048340267036186966,
      "loss": 2.3762,
      "step": 66962
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7511464953422546,
      "learning_rate": 0.0004833994332116368,
      "loss": 2.1924,
      "step": 66963
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7832494974136353,
      "learning_rate": 0.00048339619602730656,
      "loss": 2.2942,
      "step": 66964
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7478673458099365,
      "learning_rate": 0.00048339295880887943,
      "loss": 2.1449,
      "step": 66965
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.8044715523719788,
      "learning_rate": 0.0004833897215563562,
      "loss": 2.2181,
      "step": 66966
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.8364522457122803,
      "learning_rate": 0.0004833864842697373,
      "loss": 2.4319,
      "step": 66967
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.8835774660110474,
      "learning_rate": 0.0004833832469490234,
      "loss": 2.1138,
      "step": 66968
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7597094774246216,
      "learning_rate": 0.0004833800095942151,
      "loss": 2.3198,
      "step": 66969
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.8353225588798523,
      "learning_rate": 0.00048337677220531303,
      "loss": 2.1517,
      "step": 66970
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.6820680499076843,
      "learning_rate": 0.0004833735347823177,
      "loss": 2.171,
      "step": 66971
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.8671383857727051,
      "learning_rate": 0.0004833702973252297,
      "loss": 2.2059,
      "step": 66972
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7407081127166748,
      "learning_rate": 0.0004833670598340499,
      "loss": 2.2597,
      "step": 66973
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7842488288879395,
      "learning_rate": 0.0004833638223087786,
      "loss": 2.1125,
      "step": 66974
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.8502532839775085,
      "learning_rate": 0.0004833605847494164,
      "loss": 2.3677,
      "step": 66975
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7322883605957031,
      "learning_rate": 0.0004833573471559641,
      "loss": 2.025,
      "step": 66976
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.8201274871826172,
      "learning_rate": 0.0004833541095284222,
      "loss": 2.2893,
      "step": 66977
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.8759310245513916,
      "learning_rate": 0.0004833508718667913,
      "loss": 2.2689,
      "step": 66978
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.768155574798584,
      "learning_rate": 0.0004833476341710721,
      "loss": 2.1678,
      "step": 66979
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7736309766769409,
      "learning_rate": 0.00048334439644126504,
      "loss": 2.2447,
      "step": 66980
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.6658922433853149,
      "learning_rate": 0.00048334115867737076,
      "loss": 2.1176,
      "step": 66981
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7687377333641052,
      "learning_rate": 0.00048333792087938997,
      "loss": 2.1915,
      "step": 66982
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7865360975265503,
      "learning_rate": 0.00048333468304732316,
      "loss": 2.1707,
      "step": 66983
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7113168835639954,
      "learning_rate": 0.00048333144518117094,
      "loss": 2.15,
      "step": 66984
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.754806399345398,
      "learning_rate": 0.00048332820728093394,
      "loss": 2.3606,
      "step": 66985
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.6586290597915649,
      "learning_rate": 0.0004833249693466129,
      "loss": 2.0423,
      "step": 66986
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7350353598594666,
      "learning_rate": 0.00048332173137820815,
      "loss": 2.3019,
      "step": 66987
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7660309672355652,
      "learning_rate": 0.00048331849337572043,
      "loss": 2.2782,
      "step": 66988
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.8003128170967102,
      "learning_rate": 0.0004833152553391505,
      "loss": 2.2814,
      "step": 66989
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7101322412490845,
      "learning_rate": 0.0004833120172684986,
      "loss": 2.2187,
      "step": 66990
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.8049843907356262,
      "learning_rate": 0.0004833087791637656,
      "loss": 2.5165,
      "step": 66991
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7749287486076355,
      "learning_rate": 0.0004833055410249521,
      "loss": 2.2136,
      "step": 66992
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.8744447827339172,
      "learning_rate": 0.0004833023028520586,
      "loss": 2.3911,
      "step": 66993
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7667762637138367,
      "learning_rate": 0.0004832990646450857,
      "loss": 2.1838,
      "step": 66994
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7598569393157959,
      "learning_rate": 0.00048329582640403414,
      "loss": 2.0672,
      "step": 66995
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7763519883155823,
      "learning_rate": 0.0004832925881289043,
      "loss": 2.0503,
      "step": 66996
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7321534752845764,
      "learning_rate": 0.000483289349819697,
      "loss": 1.988,
      "step": 66997
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7811832427978516,
      "learning_rate": 0.00048328611147641277,
      "loss": 2.0639,
      "step": 66998
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.6810539364814758,
      "learning_rate": 0.0004832828730990522,
      "loss": 2.2328,
      "step": 66999
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7689448595046997,
      "learning_rate": 0.0004832796346876158,
      "loss": 2.4299,
      "step": 67000
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.819603681564331,
      "learning_rate": 0.00048327639624210427,
      "loss": 2.2024,
      "step": 67001
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7500593066215515,
      "learning_rate": 0.0004832731577625182,
      "loss": 2.2799,
      "step": 67002
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7461298704147339,
      "learning_rate": 0.0004832699192488582,
      "loss": 2.3752,
      "step": 67003
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.8465506434440613,
      "learning_rate": 0.00048326668070112485,
      "loss": 2.2027,
      "step": 67004
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.6992202997207642,
      "learning_rate": 0.0004832634421193188,
      "loss": 2.1475,
      "step": 67005
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.8072642087936401,
      "learning_rate": 0.00048326020350344067,
      "loss": 2.3573,
      "step": 67006
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7068399786949158,
      "learning_rate": 0.000483256964853491,
      "loss": 2.4181,
      "step": 67007
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.794560432434082,
      "learning_rate": 0.00048325372616947026,
      "loss": 2.1482,
      "step": 67008
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7789525389671326,
      "learning_rate": 0.00048325048745137935,
      "loss": 2.1336,
      "step": 67009
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7330965995788574,
      "learning_rate": 0.0004832472486992186,
      "loss": 2.3146,
      "step": 67010
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7710064649581909,
      "learning_rate": 0.00048324400991298875,
      "loss": 2.0696,
      "step": 67011
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7045128345489502,
      "learning_rate": 0.0004832407710926904,
      "loss": 2.1451,
      "step": 67012
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.810165286064148,
      "learning_rate": 0.00048323753223832413,
      "loss": 2.3162,
      "step": 67013
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7818471789360046,
      "learning_rate": 0.0004832342933498905,
      "loss": 2.295,
      "step": 67014
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7139650583267212,
      "learning_rate": 0.0004832310544273903,
      "loss": 2.1357,
      "step": 67015
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7562706470489502,
      "learning_rate": 0.00048322781547082387,
      "loss": 2.5144,
      "step": 67016
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7750190496444702,
      "learning_rate": 0.0004832245764801919,
      "loss": 2.1162,
      "step": 67017
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7803413271903992,
      "learning_rate": 0.0004832213374554951,
      "loss": 2.1915,
      "step": 67018
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7284529805183411,
      "learning_rate": 0.00048321809839673395,
      "loss": 2.0376,
      "step": 67019
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.901035726070404,
      "learning_rate": 0.0004832148593039091,
      "loss": 2.4837,
      "step": 67020
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7250381112098694,
      "learning_rate": 0.00048321162017702117,
      "loss": 2.247,
      "step": 67021
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.8461683392524719,
      "learning_rate": 0.0004832083810160708,
      "loss": 2.0505,
      "step": 67022
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7818578481674194,
      "learning_rate": 0.00048320514182105843,
      "loss": 2.2201,
      "step": 67023
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7598116993904114,
      "learning_rate": 0.00048320190259198477,
      "loss": 2.2274,
      "step": 67024
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.8286886811256409,
      "learning_rate": 0.0004831986633288505,
      "loss": 1.9653,
      "step": 67025
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7558720111846924,
      "learning_rate": 0.00048319542403165604,
      "loss": 2.0635,
      "step": 67026
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7151795029640198,
      "learning_rate": 0.0004831921847004021,
      "loss": 2.0691,
      "step": 67027
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7525515556335449,
      "learning_rate": 0.0004831889453350894,
      "loss": 2.1714,
      "step": 67028
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7448166608810425,
      "learning_rate": 0.00048318570593571837,
      "loss": 2.4342,
      "step": 67029
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7557084560394287,
      "learning_rate": 0.00048318246650228957,
      "loss": 2.4161,
      "step": 67030
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.8602585792541504,
      "learning_rate": 0.0004831792270348038,
      "loss": 2.3263,
      "step": 67031
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7409717440605164,
      "learning_rate": 0.0004831759875332615,
      "loss": 1.976,
      "step": 67032
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.8210235238075256,
      "learning_rate": 0.00048317274799766324,
      "loss": 2.3004,
      "step": 67033
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7485267519950867,
      "learning_rate": 0.00048316950842800985,
      "loss": 2.0802,
      "step": 67034
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.8845465779304504,
      "learning_rate": 0.0004831662688243018,
      "loss": 2.1572,
      "step": 67035
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7836126685142517,
      "learning_rate": 0.0004831630291865396,
      "loss": 2.2826,
      "step": 67036
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.8198681473731995,
      "learning_rate": 0.0004831597895147239,
      "loss": 2.3576,
      "step": 67037
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7886285185813904,
      "learning_rate": 0.0004831565498088554,
      "loss": 2.2563,
      "step": 67038
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7913841605186462,
      "learning_rate": 0.00048315331006893467,
      "loss": 2.1917,
      "step": 67039
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7427304983139038,
      "learning_rate": 0.00048315007029496224,
      "loss": 2.1023,
      "step": 67040
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.8808189630508423,
      "learning_rate": 0.00048314683048693877,
      "loss": 2.1507,
      "step": 67041
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7627881765365601,
      "learning_rate": 0.0004831435906448648,
      "loss": 2.2685,
      "step": 67042
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7885980010032654,
      "learning_rate": 0.00048314035076874105,
      "loss": 2.0932,
      "step": 67043
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7579176425933838,
      "learning_rate": 0.000483137110858568,
      "loss": 2.062,
      "step": 67044
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7814542055130005,
      "learning_rate": 0.00048313387091434635,
      "loss": 2.132,
      "step": 67045
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.783096969127655,
      "learning_rate": 0.0004831306309360766,
      "loss": 2.0407,
      "step": 67046
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7331926822662354,
      "learning_rate": 0.0004831273909237594,
      "loss": 2.2974,
      "step": 67047
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7586142420768738,
      "learning_rate": 0.00048312415087739546,
      "loss": 2.0535,
      "step": 67048
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.710445761680603,
      "learning_rate": 0.0004831209107969852,
      "loss": 2.0655,
      "step": 67049
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.9391577839851379,
      "learning_rate": 0.0004831176706825294,
      "loss": 2.301,
      "step": 67050
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7371084690093994,
      "learning_rate": 0.00048311443053402844,
      "loss": 2.1313,
      "step": 67051
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.789848804473877,
      "learning_rate": 0.00048311119035148315,
      "loss": 2.173,
      "step": 67052
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.8859080672264099,
      "learning_rate": 0.000483107950134894,
      "loss": 2.3312,
      "step": 67053
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.8647474050521851,
      "learning_rate": 0.00048310470988426155,
      "loss": 2.2833,
      "step": 67054
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.8847882747650146,
      "learning_rate": 0.0004831014695995866,
      "loss": 2.1935,
      "step": 67055
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7819051742553711,
      "learning_rate": 0.0004830982292808695,
      "loss": 2.3201,
      "step": 67056
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7905208468437195,
      "learning_rate": 0.00048309498892811116,
      "loss": 2.2403,
      "step": 67057
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.8319293856620789,
      "learning_rate": 0.00048309174854131183,
      "loss": 2.3872,
      "step": 67058
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7840563654899597,
      "learning_rate": 0.0004830885081204724,
      "loss": 2.1731,
      "step": 67059
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7723115086555481,
      "learning_rate": 0.0004830852676655933,
      "loss": 2.3112,
      "step": 67060
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7357391715049744,
      "learning_rate": 0.0004830820271766752,
      "loss": 2.2296,
      "step": 67061
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.8527591824531555,
      "learning_rate": 0.00048307878665371874,
      "loss": 2.0709,
      "step": 67062
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7518466711044312,
      "learning_rate": 0.00048307554609672445,
      "loss": 2.1434,
      "step": 67063
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7341555953025818,
      "learning_rate": 0.000483072305505693,
      "loss": 2.4202,
      "step": 67064
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.778639554977417,
      "learning_rate": 0.00048306906488062483,
      "loss": 2.2525,
      "step": 67065
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7604085206985474,
      "learning_rate": 0.00048306582422152076,
      "loss": 2.2798,
      "step": 67066
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7305007576942444,
      "learning_rate": 0.00048306258352838135,
      "loss": 2.0757,
      "step": 67067
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7030308246612549,
      "learning_rate": 0.00048305934280120706,
      "loss": 2.3363,
      "step": 67068
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.8247957229614258,
      "learning_rate": 0.0004830561020399986,
      "loss": 2.2463,
      "step": 67069
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7168915271759033,
      "learning_rate": 0.0004830528612447566,
      "loss": 2.0545,
      "step": 67070
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.766355037689209,
      "learning_rate": 0.00048304962041548157,
      "loss": 2.3098,
      "step": 67071
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7569512724876404,
      "learning_rate": 0.0004830463795521742,
      "loss": 2.4252,
      "step": 67072
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.8374341130256653,
      "learning_rate": 0.000483043138654835,
      "loss": 2.0571,
      "step": 67073
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7311239838600159,
      "learning_rate": 0.0004830398977234647,
      "loss": 2.1013,
      "step": 67074
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.8637510538101196,
      "learning_rate": 0.0004830366567580637,
      "loss": 2.4014,
      "step": 67075
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.8117820620536804,
      "learning_rate": 0.0004830334157586329,
      "loss": 2.2153,
      "step": 67076
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7524279356002808,
      "learning_rate": 0.0004830301747251726,
      "loss": 2.1324,
      "step": 67077
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.8312994241714478,
      "learning_rate": 0.0004830269336576835,
      "loss": 2.1238,
      "step": 67078
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7479496002197266,
      "learning_rate": 0.00048302369255616635,
      "loss": 2.3198,
      "step": 67079
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.810778021812439,
      "learning_rate": 0.0004830204514206216,
      "loss": 2.4592,
      "step": 67080
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.9018799662590027,
      "learning_rate": 0.00048301721025104994,
      "loss": 2.3995,
      "step": 67081
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7845872640609741,
      "learning_rate": 0.0004830139690474519,
      "loss": 2.3288,
      "step": 67082
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7563658952713013,
      "learning_rate": 0.000483010727809828,
      "loss": 2.2202,
      "step": 67083
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7818989753723145,
      "learning_rate": 0.00048300748653817903,
      "loss": 2.2011,
      "step": 67084
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.749325692653656,
      "learning_rate": 0.00048300424523250557,
      "loss": 2.4192,
      "step": 67085
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7900692224502563,
      "learning_rate": 0.00048300100389280816,
      "loss": 2.3077,
      "step": 67086
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.740608811378479,
      "learning_rate": 0.0004829977625190873,
      "loss": 2.3165,
      "step": 67087
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.755402684211731,
      "learning_rate": 0.00048299452111134385,
      "loss": 2.4721,
      "step": 67088
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7365575432777405,
      "learning_rate": 0.0004829912796695782,
      "loss": 2.0966,
      "step": 67089
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7690215110778809,
      "learning_rate": 0.0004829880381937909,
      "loss": 2.2344,
      "step": 67090
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7333117723464966,
      "learning_rate": 0.0004829847966839829,
      "loss": 2.1415,
      "step": 67091
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7870737910270691,
      "learning_rate": 0.0004829815551401544,
      "loss": 2.0819,
      "step": 67092
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.6760720014572144,
      "learning_rate": 0.0004829783135623062,
      "loss": 2.0911,
      "step": 67093
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.6966121792793274,
      "learning_rate": 0.00048297507195043884,
      "loss": 2.1949,
      "step": 67094
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.8033069372177124,
      "learning_rate": 0.0004829718303045531,
      "loss": 2.3308,
      "step": 67095
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7283859252929688,
      "learning_rate": 0.0004829685886246493,
      "loss": 2.1235,
      "step": 67096
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7755278944969177,
      "learning_rate": 0.00048296534691072825,
      "loss": 2.1712,
      "step": 67097
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.6688113212585449,
      "learning_rate": 0.00048296210516279056,
      "loss": 2.2164,
      "step": 67098
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.8168056607246399,
      "learning_rate": 0.0004829588633808366,
      "loss": 2.2917,
      "step": 67099
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7345271706581116,
      "learning_rate": 0.0004829556215648672,
      "loss": 2.4105,
      "step": 67100
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.6881526112556458,
      "learning_rate": 0.000482952379714883,
      "loss": 2.1531,
      "step": 67101
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7464468479156494,
      "learning_rate": 0.00048294913783088444,
      "loss": 2.2166,
      "step": 67102
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.9237109422683716,
      "learning_rate": 0.00048294589591287213,
      "loss": 1.9216,
      "step": 67103
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.0100544691085815,
      "learning_rate": 0.0004829426539608467,
      "loss": 2.0664,
      "step": 67104
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.8531559705734253,
      "learning_rate": 0.00048293941197480886,
      "loss": 2.033,
      "step": 67105
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.8364384770393372,
      "learning_rate": 0.0004829361699547591,
      "loss": 2.183,
      "step": 67106
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7509014010429382,
      "learning_rate": 0.0004829329279006981,
      "loss": 2.2775,
      "step": 67107
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7032427787780762,
      "learning_rate": 0.0004829296858126263,
      "loss": 1.9977,
      "step": 67108
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7207264304161072,
      "learning_rate": 0.00048292644369054454,
      "loss": 2.2425,
      "step": 67109
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7342104911804199,
      "learning_rate": 0.0004829232015344532,
      "loss": 2.1105,
      "step": 67110
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7212637662887573,
      "learning_rate": 0.0004829199593443531,
      "loss": 2.1706,
      "step": 67111
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7909072041511536,
      "learning_rate": 0.00048291671712024454,
      "loss": 2.1513,
      "step": 67112
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7395380735397339,
      "learning_rate": 0.0004829134748621285,
      "loss": 2.2572,
      "step": 67113
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.8004741072654724,
      "learning_rate": 0.0004829102325700053,
      "loss": 2.4317,
      "step": 67114
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7498859763145447,
      "learning_rate": 0.0004829069902438757,
      "loss": 2.073,
      "step": 67115
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7368036508560181,
      "learning_rate": 0.0004829037478837402,
      "loss": 2.0333,
      "step": 67116
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7906705737113953,
      "learning_rate": 0.0004829005054895994,
      "loss": 2.307,
      "step": 67117
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7681898474693298,
      "learning_rate": 0.000482897263061454,
      "loss": 2.3027,
      "step": 67118
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7063586711883545,
      "learning_rate": 0.0004828940205993045,
      "loss": 2.3329,
      "step": 67119
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.1306978464126587,
      "learning_rate": 0.00048289077810315155,
      "loss": 1.9373,
      "step": 67120
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7632648348808289,
      "learning_rate": 0.0004828875355729958,
      "loss": 2.2959,
      "step": 67121
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.9649910926818848,
      "learning_rate": 0.00048288429300883784,
      "loss": 2.1171,
      "step": 67122
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.8341831564903259,
      "learning_rate": 0.00048288105041067813,
      "loss": 2.2621,
      "step": 67123
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.8122684359550476,
      "learning_rate": 0.00048287780777851746,
      "loss": 2.3441,
      "step": 67124
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.8249856233596802,
      "learning_rate": 0.00048287456511235627,
      "loss": 2.3015,
      "step": 67125
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.8786277174949646,
      "learning_rate": 0.0004828713224121953,
      "loss": 2.305,
      "step": 67126
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.8350828289985657,
      "learning_rate": 0.00048286807967803507,
      "loss": 2.3871,
      "step": 67127
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.8272780776023865,
      "learning_rate": 0.0004828648369098763,
      "loss": 2.4258,
      "step": 67128
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.9949491620063782,
      "learning_rate": 0.00048286159410771936,
      "loss": 2.3082,
      "step": 67129
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7821881175041199,
      "learning_rate": 0.0004828583512715651,
      "loss": 2.285,
      "step": 67130
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.6952712535858154,
      "learning_rate": 0.000482855108401414,
      "loss": 2.0835,
      "step": 67131
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.928943395614624,
      "learning_rate": 0.00048285186549726664,
      "loss": 2.2373,
      "step": 67132
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7497629523277283,
      "learning_rate": 0.0004828486225591238,
      "loss": 2.1903,
      "step": 67133
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7989386916160583,
      "learning_rate": 0.0004828453795869859,
      "loss": 2.115,
      "step": 67134
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.748302698135376,
      "learning_rate": 0.0004828421365808535,
      "loss": 2.2227,
      "step": 67135
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7829426527023315,
      "learning_rate": 0.00048283889354072733,
      "loss": 2.1305,
      "step": 67136
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.9309709072113037,
      "learning_rate": 0.000482835650466608,
      "loss": 2.1585,
      "step": 67137
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.8423357009887695,
      "learning_rate": 0.00048283240735849604,
      "loss": 2.2831,
      "step": 67138
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.8316192626953125,
      "learning_rate": 0.0004828291642163921,
      "loss": 2.3444,
      "step": 67139
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.1307573318481445,
      "learning_rate": 0.0004828259210402968,
      "loss": 1.9895,
      "step": 67140
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7408581376075745,
      "learning_rate": 0.00048282267783021063,
      "loss": 2.2333,
      "step": 67141
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7995800971984863,
      "learning_rate": 0.00048281943458613433,
      "loss": 2.2666,
      "step": 67142
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.8220217227935791,
      "learning_rate": 0.0004828161913080685,
      "loss": 2.2247,
      "step": 67143
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.9798441529273987,
      "learning_rate": 0.00048281294799601356,
      "loss": 2.3085,
      "step": 67144
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.6862928867340088,
      "learning_rate": 0.0004828097046499702,
      "loss": 2.2983,
      "step": 67145
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.9057489633560181,
      "learning_rate": 0.0004828064612699393,
      "loss": 2.3288,
      "step": 67146
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.0029903650283813,
      "learning_rate": 0.00048280321785592104,
      "loss": 2.2822,
      "step": 67147
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.9092136025428772,
      "learning_rate": 0.0004827999744079163,
      "loss": 2.0373,
      "step": 67148
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7201122641563416,
      "learning_rate": 0.00048279673092592554,
      "loss": 2.3269,
      "step": 67149
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.6024210453033447,
      "learning_rate": 0.0004827934874099495,
      "loss": 2.1948,
      "step": 67150
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.0098633766174316,
      "learning_rate": 0.0004827902438599886,
      "loss": 2.0254,
      "step": 67151
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7623711824417114,
      "learning_rate": 0.0004827870002760436,
      "loss": 2.3332,
      "step": 67152
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.134033203125,
      "learning_rate": 0.000482783756658115,
      "loss": 2.1573,
      "step": 67153
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.2066396474838257,
      "learning_rate": 0.00048278051300620347,
      "loss": 2.1848,
      "step": 67154
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.8270810842514038,
      "learning_rate": 0.0004827772693203096,
      "loss": 2.1526,
      "step": 67155
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.8210980892181396,
      "learning_rate": 0.00048277402560043393,
      "loss": 2.3713,
      "step": 67156
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.8821669816970825,
      "learning_rate": 0.0004827707818465772,
      "loss": 2.2502,
      "step": 67157
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.3546223640441895,
      "learning_rate": 0.00048276753805873997,
      "loss": 2.2736,
      "step": 67158
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.8736252784729004,
      "learning_rate": 0.00048276429423692264,
      "loss": 2.0307,
      "step": 67159
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.8236656188964844,
      "learning_rate": 0.00048276105038112615,
      "loss": 2.2922,
      "step": 67160
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.028725266456604,
      "learning_rate": 0.0004827578064913508,
      "loss": 2.1388,
      "step": 67161
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.9267048239707947,
      "learning_rate": 0.0004827545625675975,
      "loss": 2.3278,
      "step": 67162
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.8282703161239624,
      "learning_rate": 0.00048275131860986653,
      "loss": 2.0196,
      "step": 67163
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7836036086082458,
      "learning_rate": 0.00048274807461815857,
      "loss": 2.1352,
      "step": 67164
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.884253978729248,
      "learning_rate": 0.0004827448305924744,
      "loss": 2.4184,
      "step": 67165
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.8036272525787354,
      "learning_rate": 0.0004827415865328145,
      "loss": 2.4376,
      "step": 67166
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7024098634719849,
      "learning_rate": 0.0004827383424391795,
      "loss": 2.0481,
      "step": 67167
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7668225765228271,
      "learning_rate": 0.00048273509831157,
      "loss": 2.1165,
      "step": 67168
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7714211344718933,
      "learning_rate": 0.0004827318541499866,
      "loss": 2.1203,
      "step": 67169
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.8355425596237183,
      "learning_rate": 0.0004827286099544299,
      "loss": 2.1247,
      "step": 67170
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7644510269165039,
      "learning_rate": 0.0004827253657249005,
      "loss": 2.166,
      "step": 67171
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7123111486434937,
      "learning_rate": 0.000482722121461399,
      "loss": 2.1579,
      "step": 67172
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.9103779792785645,
      "learning_rate": 0.00048271887716392595,
      "loss": 2.1603,
      "step": 67173
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.747086226940155,
      "learning_rate": 0.00048271563283248204,
      "loss": 2.0612,
      "step": 67174
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7102636694908142,
      "learning_rate": 0.00048271238846706786,
      "loss": 2.2748,
      "step": 67175
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7420223951339722,
      "learning_rate": 0.00048270914406768405,
      "loss": 2.0394,
      "step": 67176
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7179997563362122,
      "learning_rate": 0.00048270589963433116,
      "loss": 2.2397,
      "step": 67177
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7060865163803101,
      "learning_rate": 0.00048270265516700973,
      "loss": 2.0636,
      "step": 67178
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7605196237564087,
      "learning_rate": 0.00048269941066572046,
      "loss": 2.3491,
      "step": 67179
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7696667909622192,
      "learning_rate": 0.0004826961661304639,
      "loss": 2.1925,
      "step": 67180
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.8678318858146667,
      "learning_rate": 0.0004826929215612407,
      "loss": 2.2153,
      "step": 67181
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7497202157974243,
      "learning_rate": 0.00048268967695805137,
      "loss": 2.4322,
      "step": 67182
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7889158725738525,
      "learning_rate": 0.00048268643232089666,
      "loss": 2.293,
      "step": 67183
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.78354412317276,
      "learning_rate": 0.0004826831876497771,
      "loss": 2.3426,
      "step": 67184
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.8107206225395203,
      "learning_rate": 0.0004826799429446932,
      "loss": 2.3237,
      "step": 67185
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7072023749351501,
      "learning_rate": 0.0004826766982056458,
      "loss": 2.0913,
      "step": 67186
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.8342472910881042,
      "learning_rate": 0.0004826734534326352,
      "loss": 2.4963,
      "step": 67187
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.6702640652656555,
      "learning_rate": 0.00048267020862566225,
      "loss": 2.1984,
      "step": 67188
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.8263192772865295,
      "learning_rate": 0.00048266696378472735,
      "loss": 2.2341,
      "step": 67189
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.6881287693977356,
      "learning_rate": 0.00048266371890983133,
      "loss": 2.1582,
      "step": 67190
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7672205567359924,
      "learning_rate": 0.00048266047400097464,
      "loss": 2.2653,
      "step": 67191
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.6733397245407104,
      "learning_rate": 0.0004826572290581579,
      "loss": 2.1426,
      "step": 67192
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7511453628540039,
      "learning_rate": 0.00048265398408138175,
      "loss": 1.9685,
      "step": 67193
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7724021673202515,
      "learning_rate": 0.0004826507390706467,
      "loss": 2.1854,
      "step": 67194
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.8016911745071411,
      "learning_rate": 0.0004826474940259536,
      "loss": 2.1669,
      "step": 67195
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7783172726631165,
      "learning_rate": 0.0004826442489473028,
      "loss": 2.0759,
      "step": 67196
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.751923680305481,
      "learning_rate": 0.00048264100383469495,
      "loss": 2.4064,
      "step": 67197
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7833729982376099,
      "learning_rate": 0.0004826377586881307,
      "loss": 2.2636,
      "step": 67198
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7100954055786133,
      "learning_rate": 0.00048263451350761063,
      "loss": 2.3684,
      "step": 67199
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.6887449026107788,
      "learning_rate": 0.0004826312682931354,
      "loss": 2.0144,
      "step": 67200
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.8125099539756775,
      "learning_rate": 0.0004826280230447056,
      "loss": 2.3923,
      "step": 67201
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7489335536956787,
      "learning_rate": 0.00048262477776232174,
      "loss": 2.2723,
      "step": 67202
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7509725093841553,
      "learning_rate": 0.00048262153244598446,
      "loss": 2.3012,
      "step": 67203
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.8330546617507935,
      "learning_rate": 0.0004826182870956945,
      "loss": 2.1528,
      "step": 67204
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.9695520997047424,
      "learning_rate": 0.0004826150417114522,
      "loss": 2.2873,
      "step": 67205
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.8832222819328308,
      "learning_rate": 0.0004826117962932584,
      "loss": 2.3644,
      "step": 67206
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7798430919647217,
      "learning_rate": 0.00048260855084111365,
      "loss": 2.0659,
      "step": 67207
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.1926699876785278,
      "learning_rate": 0.00048260530535501844,
      "loss": 2.1652,
      "step": 67208
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.0469316244125366,
      "learning_rate": 0.00048260205983497356,
      "loss": 2.387,
      "step": 67209
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7585325837135315,
      "learning_rate": 0.00048259881428097944,
      "loss": 2.302,
      "step": 67210
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.8900478482246399,
      "learning_rate": 0.0004825955686930368,
      "loss": 2.2992,
      "step": 67211
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.505210280418396,
      "learning_rate": 0.0004825923230711462,
      "loss": 2.2691,
      "step": 67212
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.100913166999817,
      "learning_rate": 0.00048258907741530817,
      "loss": 2.3033,
      "step": 67213
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.8259833455085754,
      "learning_rate": 0.00048258583172552336,
      "loss": 2.27,
      "step": 67214
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.9342623949050903,
      "learning_rate": 0.00048258258600179256,
      "loss": 1.9694,
      "step": 67215
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.9486158490180969,
      "learning_rate": 0.0004825793402441161,
      "loss": 2.1023,
      "step": 67216
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7467307448387146,
      "learning_rate": 0.0004825760944524947,
      "loss": 2.1916,
      "step": 67217
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.882151186466217,
      "learning_rate": 0.0004825728486269289,
      "loss": 2.1814,
      "step": 67218
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.1348481178283691,
      "learning_rate": 0.0004825696027674195,
      "loss": 2.4563,
      "step": 67219
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.8238082528114319,
      "learning_rate": 0.00048256635687396693,
      "loss": 2.3869,
      "step": 67220
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7803383469581604,
      "learning_rate": 0.0004825631109465718,
      "loss": 2.235,
      "step": 67221
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.8247397541999817,
      "learning_rate": 0.0004825598649852347,
      "loss": 2.0948,
      "step": 67222
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.9040051102638245,
      "learning_rate": 0.0004825566189899563,
      "loss": 2.3886,
      "step": 67223
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.932329535484314,
      "learning_rate": 0.00048255337296073715,
      "loss": 2.3081,
      "step": 67224
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.879929780960083,
      "learning_rate": 0.00048255012689757794,
      "loss": 2.2341,
      "step": 67225
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7647460103034973,
      "learning_rate": 0.00048254688080047925,
      "loss": 2.1275,
      "step": 67226
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.146047830581665,
      "learning_rate": 0.0004825436346694415,
      "loss": 2.1141,
      "step": 67227
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.790958821773529,
      "learning_rate": 0.0004825403885044655,
      "loss": 2.2576,
      "step": 67228
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.783281147480011,
      "learning_rate": 0.0004825371423055519,
      "loss": 2.2432,
      "step": 67229
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.0656434297561646,
      "learning_rate": 0.00048253389607270117,
      "loss": 2.2039,
      "step": 67230
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.963129997253418,
      "learning_rate": 0.00048253064980591383,
      "loss": 2.1475,
      "step": 67231
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7556990385055542,
      "learning_rate": 0.00048252740350519075,
      "loss": 2.3295,
      "step": 67232
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.8976767063140869,
      "learning_rate": 0.00048252415717053225,
      "loss": 2.0432,
      "step": 67233
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.1679129600524902,
      "learning_rate": 0.0004825209108019391,
      "loss": 2.1135,
      "step": 67234
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.9282369017601013,
      "learning_rate": 0.0004825176643994119,
      "loss": 2.3266,
      "step": 67235
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7726927399635315,
      "learning_rate": 0.00048251441796295116,
      "loss": 1.9961,
      "step": 67236
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.061589002609253,
      "learning_rate": 0.0004825111714925576,
      "loss": 2.3499,
      "step": 67237
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.0277297496795654,
      "learning_rate": 0.00048250792498823176,
      "loss": 2.216,
      "step": 67238
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7752165198326111,
      "learning_rate": 0.0004825046784499743,
      "loss": 2.4701,
      "step": 67239
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7983217239379883,
      "learning_rate": 0.00048250143187778563,
      "loss": 2.1596,
      "step": 67240
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.011773943901062,
      "learning_rate": 0.00048249818527166665,
      "loss": 2.1456,
      "step": 67241
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.0454933643341064,
      "learning_rate": 0.0004824949386316178,
      "loss": 2.4581,
      "step": 67242
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.8677204847335815,
      "learning_rate": 0.00048249169195763956,
      "loss": 2.1348,
      "step": 67243
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.9036698341369629,
      "learning_rate": 0.00048248844524973285,
      "loss": 2.2399,
      "step": 67244
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.201655387878418,
      "learning_rate": 0.000482485198507898,
      "loss": 2.1268,
      "step": 67245
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.8315677046775818,
      "learning_rate": 0.00048248195173213564,
      "loss": 2.0164,
      "step": 67246
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7372230887413025,
      "learning_rate": 0.00048247870492244653,
      "loss": 2.2001,
      "step": 67247
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.8974090218544006,
      "learning_rate": 0.0004824754580788312,
      "loss": 2.2315,
      "step": 67248
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.9097076654434204,
      "learning_rate": 0.00048247221120129014,
      "loss": 2.1725,
      "step": 67249
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7770434617996216,
      "learning_rate": 0.00048246896428982415,
      "loss": 2.2209,
      "step": 67250
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7572658061981201,
      "learning_rate": 0.00048246571734443376,
      "loss": 1.9935,
      "step": 67251
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.9124259948730469,
      "learning_rate": 0.0004824624703651194,
      "loss": 2.1847,
      "step": 67252
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.9478597044944763,
      "learning_rate": 0.0004824592233518819,
      "loss": 2.3173,
      "step": 67253
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.8349993824958801,
      "learning_rate": 0.0004824559763047218,
      "loss": 2.3919,
      "step": 67254
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.8417953252792358,
      "learning_rate": 0.00048245272922363967,
      "loss": 2.0052,
      "step": 67255
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.8975391983985901,
      "learning_rate": 0.0004824494821086362,
      "loss": 2.2943,
      "step": 67256
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.8143271803855896,
      "learning_rate": 0.0004824462349597118,
      "loss": 2.0522,
      "step": 67257
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7497325539588928,
      "learning_rate": 0.0004824429877768674,
      "loss": 2.0931,
      "step": 67258
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.8839133381843567,
      "learning_rate": 0.00048243974056010324,
      "loss": 2.3483,
      "step": 67259
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7263467311859131,
      "learning_rate": 0.00048243649330942005,
      "loss": 2.1612,
      "step": 67260
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7484598159790039,
      "learning_rate": 0.00048243324602481857,
      "loss": 2.122,
      "step": 67261
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.0622708797454834,
      "learning_rate": 0.0004824299987062993,
      "loss": 2.1601,
      "step": 67262
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7381064891815186,
      "learning_rate": 0.0004824267513538629,
      "loss": 2.1042,
      "step": 67263
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.6904958486557007,
      "learning_rate": 0.0004824235039675098,
      "loss": 2.0385,
      "step": 67264
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.8584508895874023,
      "learning_rate": 0.0004824202565472408,
      "loss": 2.1839,
      "step": 67265
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.8505396842956543,
      "learning_rate": 0.0004824170090930564,
      "loss": 2.2971,
      "step": 67266
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7219299077987671,
      "learning_rate": 0.0004824137616049573,
      "loss": 2.1012,
      "step": 67267
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7992521524429321,
      "learning_rate": 0.000482410514082944,
      "loss": 2.0949,
      "step": 67268
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7703721523284912,
      "learning_rate": 0.0004824072665270171,
      "loss": 2.4333,
      "step": 67269
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7897453308105469,
      "learning_rate": 0.0004824040189371773,
      "loss": 2.3846,
      "step": 67270
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7923294305801392,
      "learning_rate": 0.0004824007713134251,
      "loss": 2.3023,
      "step": 67271
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7679263353347778,
      "learning_rate": 0.0004823975236557612,
      "loss": 2.2615,
      "step": 67272
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.8407266139984131,
      "learning_rate": 0.0004823942759641861,
      "loss": 2.2516,
      "step": 67273
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.819393515586853,
      "learning_rate": 0.0004823910282387006,
      "loss": 2.3582,
      "step": 67274
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.746965765953064,
      "learning_rate": 0.000482387780479305,
      "loss": 2.3905,
      "step": 67275
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.863612711429596,
      "learning_rate": 0.0004823845326860001,
      "loss": 2.1087,
      "step": 67276
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7580662965774536,
      "learning_rate": 0.00048238128485878656,
      "loss": 2.1137,
      "step": 67277
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.8581541776657104,
      "learning_rate": 0.00048237803699766486,
      "loss": 2.2753,
      "step": 67278
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7554876208305359,
      "learning_rate": 0.0004823747891026356,
      "loss": 2.1425,
      "step": 67279
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.6945355534553528,
      "learning_rate": 0.00048237154117369954,
      "loss": 2.1451,
      "step": 67280
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.8348299860954285,
      "learning_rate": 0.00048236829321085696,
      "loss": 2.2874,
      "step": 67281
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.8375431895256042,
      "learning_rate": 0.00048236504521410883,
      "loss": 2.356,
      "step": 67282
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.6974390149116516,
      "learning_rate": 0.0004823617971834556,
      "loss": 2.066,
      "step": 67283
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7680099010467529,
      "learning_rate": 0.0004823585491188978,
      "loss": 2.1389,
      "step": 67284
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.6842314004898071,
      "learning_rate": 0.00048235530102043613,
      "loss": 2.1881,
      "step": 67285
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.9205206632614136,
      "learning_rate": 0.0004823520528880712,
      "loss": 2.3155,
      "step": 67286
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.763709545135498,
      "learning_rate": 0.0004823488047218035,
      "loss": 1.9486,
      "step": 67287
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.829609215259552,
      "learning_rate": 0.00048234555652163375,
      "loss": 2.434,
      "step": 67288
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7487383484840393,
      "learning_rate": 0.00048234230828756255,
      "loss": 2.1281,
      "step": 67289
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.8039624691009521,
      "learning_rate": 0.00048233906001959046,
      "loss": 2.1206,
      "step": 67290
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7454922795295715,
      "learning_rate": 0.0004823358117177181,
      "loss": 2.2039,
      "step": 67291
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7529416680335999,
      "learning_rate": 0.00048233256338194594,
      "loss": 2.145,
      "step": 67292
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.8411181569099426,
      "learning_rate": 0.00048232931501227493,
      "loss": 2.1996,
      "step": 67293
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7609456181526184,
      "learning_rate": 0.0004823260666087054,
      "loss": 2.056,
      "step": 67294
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.8432849645614624,
      "learning_rate": 0.0004823228181712379,
      "loss": 2.0687,
      "step": 67295
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.807699978351593,
      "learning_rate": 0.00048231956969987326,
      "loss": 2.4844,
      "step": 67296
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.9660053849220276,
      "learning_rate": 0.0004823163211946119,
      "loss": 2.156,
      "step": 67297
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.6907378435134888,
      "learning_rate": 0.0004823130726554545,
      "loss": 2.4255,
      "step": 67298
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.0293803215026855,
      "learning_rate": 0.0004823098240824017,
      "loss": 2.0751,
      "step": 67299
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.8994472026824951,
      "learning_rate": 0.00048230657547545413,
      "loss": 2.2626,
      "step": 67300
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7279170751571655,
      "learning_rate": 0.00048230332683461217,
      "loss": 2.0827,
      "step": 67301
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7621623277664185,
      "learning_rate": 0.0004823000781598767,
      "loss": 2.3791,
      "step": 67302
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.9148478507995605,
      "learning_rate": 0.00048229682945124814,
      "loss": 2.3373,
      "step": 67303
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7251411080360413,
      "learning_rate": 0.0004822935807087272,
      "loss": 2.3684,
      "step": 67304
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7753713726997375,
      "learning_rate": 0.0004822903319323144,
      "loss": 2.2656,
      "step": 67305
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.748552143573761,
      "learning_rate": 0.0004822870831220104,
      "loss": 2.2753,
      "step": 67306
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7517543435096741,
      "learning_rate": 0.0004822838342778158,
      "loss": 2.4127,
      "step": 67307
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7660584449768066,
      "learning_rate": 0.0004822805853997312,
      "loss": 2.3072,
      "step": 67308
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7657461762428284,
      "learning_rate": 0.0004822773364877572,
      "loss": 2.0873,
      "step": 67309
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.8152744174003601,
      "learning_rate": 0.00048227408754189434,
      "loss": 2.1158,
      "step": 67310
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7602244019508362,
      "learning_rate": 0.00048227083856214336,
      "loss": 2.4139,
      "step": 67311
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7745736241340637,
      "learning_rate": 0.0004822675895485047,
      "loss": 2.1755,
      "step": 67312
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.86318039894104,
      "learning_rate": 0.0004822643405009792,
      "loss": 2.2294,
      "step": 67313
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.762846052646637,
      "learning_rate": 0.0004822610914195673,
      "loss": 2.2695,
      "step": 67314
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7563065886497498,
      "learning_rate": 0.0004822578423042695,
      "loss": 2.3683,
      "step": 67315
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7175301909446716,
      "learning_rate": 0.0004822545931550866,
      "loss": 2.3041,
      "step": 67316
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7239639759063721,
      "learning_rate": 0.00048225134397201905,
      "loss": 2.3142,
      "step": 67317
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7294282913208008,
      "learning_rate": 0.00048224809475506773,
      "loss": 1.9767,
      "step": 67318
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.8229610919952393,
      "learning_rate": 0.00048224484550423287,
      "loss": 2.3489,
      "step": 67319
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.8121946454048157,
      "learning_rate": 0.0004822415962195154,
      "loss": 2.0902,
      "step": 67320
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7235229015350342,
      "learning_rate": 0.00048223834690091556,
      "loss": 2.2787,
      "step": 67321
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7701721787452698,
      "learning_rate": 0.0004822350975484343,
      "loss": 2.0576,
      "step": 67322
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7848174571990967,
      "learning_rate": 0.0004822318481620721,
      "loss": 2.1329,
      "step": 67323
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.8170595765113831,
      "learning_rate": 0.00048222859874182957,
      "loss": 2.1525,
      "step": 67324
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7203156352043152,
      "learning_rate": 0.0004822253492877073,
      "loss": 2.5337,
      "step": 67325
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7334590554237366,
      "learning_rate": 0.00048222209979970586,
      "loss": 2.1031,
      "step": 67326
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7466294765472412,
      "learning_rate": 0.00048221885027782593,
      "loss": 2.2562,
      "step": 67327
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.806820809841156,
      "learning_rate": 0.00048221560072206804,
      "loss": 2.3605,
      "step": 67328
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7634927034378052,
      "learning_rate": 0.0004822123511324329,
      "loss": 1.9989,
      "step": 67329
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.767784833908081,
      "learning_rate": 0.00048220910150892094,
      "loss": 2.1812,
      "step": 67330
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7251574993133545,
      "learning_rate": 0.000482205851851533,
      "loss": 2.0648,
      "step": 67331
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7816469073295593,
      "learning_rate": 0.00048220260216026944,
      "loss": 2.34,
      "step": 67332
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7265837788581848,
      "learning_rate": 0.000482199352435131,
      "loss": 2.0303,
      "step": 67333
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.826701283454895,
      "learning_rate": 0.00048219610267611826,
      "loss": 2.1455,
      "step": 67334
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.9489480257034302,
      "learning_rate": 0.00048219285288323186,
      "loss": 2.0952,
      "step": 67335
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.6602166891098022,
      "learning_rate": 0.0004821896030564724,
      "loss": 2.1482,
      "step": 67336
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7795100808143616,
      "learning_rate": 0.00048218635319584037,
      "loss": 2.053,
      "step": 67337
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7224118113517761,
      "learning_rate": 0.00048218310330133647,
      "loss": 2.1006,
      "step": 67338
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7893846035003662,
      "learning_rate": 0.0004821798533729614,
      "loss": 2.1682,
      "step": 67339
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7507139444351196,
      "learning_rate": 0.0004821766034107155,
      "loss": 2.0,
      "step": 67340
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7451329231262207,
      "learning_rate": 0.00048217335341459963,
      "loss": 2.1936,
      "step": 67341
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7019199728965759,
      "learning_rate": 0.00048217010338461426,
      "loss": 2.2977,
      "step": 67342
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.8615555167198181,
      "learning_rate": 0.00048216685332076,
      "loss": 2.1835,
      "step": 67343
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7340042591094971,
      "learning_rate": 0.00048216360322303757,
      "loss": 2.1579,
      "step": 67344
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7539275884628296,
      "learning_rate": 0.00048216035309144744,
      "loss": 2.206,
      "step": 67345
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.9551499485969543,
      "learning_rate": 0.00048215710292599025,
      "loss": 2.0824,
      "step": 67346
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.8004611730575562,
      "learning_rate": 0.0004821538527266666,
      "loss": 2.1988,
      "step": 67347
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.9781984090805054,
      "learning_rate": 0.00048215060249347713,
      "loss": 2.3525,
      "step": 67348
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.8721414804458618,
      "learning_rate": 0.00048214735222642234,
      "loss": 1.9575,
      "step": 67349
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7403779029846191,
      "learning_rate": 0.00048214410192550314,
      "loss": 2.1551,
      "step": 67350
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.8669809103012085,
      "learning_rate": 0.0004821408515907197,
      "loss": 2.1744,
      "step": 67351
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.9263132214546204,
      "learning_rate": 0.00048213760122207294,
      "loss": 2.1604,
      "step": 67352
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7955093383789062,
      "learning_rate": 0.0004821343508195634,
      "loss": 2.2879,
      "step": 67353
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7424130439758301,
      "learning_rate": 0.00048213110038319157,
      "loss": 2.1118,
      "step": 67354
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.743381679058075,
      "learning_rate": 0.0004821278499129581,
      "loss": 2.0866,
      "step": 67355
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7427864670753479,
      "learning_rate": 0.00048212459940886364,
      "loss": 2.2465,
      "step": 67356
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7366917133331299,
      "learning_rate": 0.00048212134887090885,
      "loss": 2.1228,
      "step": 67357
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7018290758132935,
      "learning_rate": 0.0004821180982990942,
      "loss": 2.2141,
      "step": 67358
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.8317696452140808,
      "learning_rate": 0.00048211484769342046,
      "loss": 2.1905,
      "step": 67359
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.718970000743866,
      "learning_rate": 0.000482111597053888,
      "loss": 2.1459,
      "step": 67360
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.8419908881187439,
      "learning_rate": 0.00048210834638049755,
      "loss": 2.2378,
      "step": 67361
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7809467315673828,
      "learning_rate": 0.0004821050956732498,
      "loss": 2.1865,
      "step": 67362
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.8792561292648315,
      "learning_rate": 0.0004821018449321453,
      "loss": 2.3043,
      "step": 67363
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.8101649284362793,
      "learning_rate": 0.0004820985941571846,
      "loss": 2.349,
      "step": 67364
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.0269323587417603,
      "learning_rate": 0.0004820953433483683,
      "loss": 2.1241,
      "step": 67365
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.735049843788147,
      "learning_rate": 0.00048209209250569705,
      "loss": 2.3137,
      "step": 67366
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7950050830841064,
      "learning_rate": 0.0004820888416291715,
      "loss": 2.3171,
      "step": 67367
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7751955986022949,
      "learning_rate": 0.00048208559071879207,
      "loss": 2.4277,
      "step": 67368
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.8435413837432861,
      "learning_rate": 0.0004820823397745596,
      "loss": 2.2573,
      "step": 67369
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7546937465667725,
      "learning_rate": 0.00048207908879647457,
      "loss": 2.3207,
      "step": 67370
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.877517580986023,
      "learning_rate": 0.0004820758377845376,
      "loss": 1.9035,
      "step": 67371
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7392427921295166,
      "learning_rate": 0.0004820725867387493,
      "loss": 2.1665,
      "step": 67372
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7666763067245483,
      "learning_rate": 0.0004820693356591102,
      "loss": 2.4442,
      "step": 67373
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7845157384872437,
      "learning_rate": 0.000482066084545621,
      "loss": 2.2198,
      "step": 67374
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7361398339271545,
      "learning_rate": 0.00048206283339828236,
      "loss": 2.3297,
      "step": 67375
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.8558056354522705,
      "learning_rate": 0.0004820595822170947,
      "loss": 2.1927,
      "step": 67376
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.735936164855957,
      "learning_rate": 0.00048205633100205885,
      "loss": 2.3119,
      "step": 67377
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.877149760723114,
      "learning_rate": 0.0004820530797531752,
      "loss": 2.3764,
      "step": 67378
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.8284832835197449,
      "learning_rate": 0.00048204982847044445,
      "loss": 2.2001,
      "step": 67379
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.82164466381073,
      "learning_rate": 0.00048204657715386715,
      "loss": 2.3457,
      "step": 67380
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7423539757728577,
      "learning_rate": 0.00048204332580344404,
      "loss": 2.2432,
      "step": 67381
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.6883662939071655,
      "learning_rate": 0.00048204007441917567,
      "loss": 2.0631,
      "step": 67382
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.8317310810089111,
      "learning_rate": 0.0004820368230010625,
      "loss": 2.2006,
      "step": 67383
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.8315973281860352,
      "learning_rate": 0.00048203357154910534,
      "loss": 2.1821,
      "step": 67384
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7725449800491333,
      "learning_rate": 0.0004820303200633046,
      "loss": 2.3482,
      "step": 67385
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7923803329467773,
      "learning_rate": 0.0004820270685436611,
      "loss": 2.1874,
      "step": 67386
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.8304226398468018,
      "learning_rate": 0.0004820238169901753,
      "loss": 2.1837,
      "step": 67387
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.8038710951805115,
      "learning_rate": 0.00048202056540284786,
      "loss": 2.3084,
      "step": 67388
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7560194134712219,
      "learning_rate": 0.0004820173137816793,
      "loss": 2.1845,
      "step": 67389
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7339435815811157,
      "learning_rate": 0.00048201406212667035,
      "loss": 2.2236,
      "step": 67390
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7481176853179932,
      "learning_rate": 0.0004820108104378215,
      "loss": 2.1146,
      "step": 67391
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7900475859642029,
      "learning_rate": 0.00048200755871513336,
      "loss": 2.2669,
      "step": 67392
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7106317281723022,
      "learning_rate": 0.00048200430695860666,
      "loss": 2.1291,
      "step": 67393
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7209854125976562,
      "learning_rate": 0.0004820010551682419,
      "loss": 2.1692,
      "step": 67394
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.797126054763794,
      "learning_rate": 0.0004819978033440397,
      "loss": 2.1425,
      "step": 67395
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7322772741317749,
      "learning_rate": 0.0004819945514860007,
      "loss": 2.2628,
      "step": 67396
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.823937177658081,
      "learning_rate": 0.0004819912995941254,
      "loss": 2.3531,
      "step": 67397
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.8289585709571838,
      "learning_rate": 0.0004819880476684145,
      "loss": 2.3218,
      "step": 67398
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.8008434176445007,
      "learning_rate": 0.0004819847957088687,
      "loss": 2.2218,
      "step": 67399
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.710742175579071,
      "learning_rate": 0.0004819815437154884,
      "loss": 2.1338,
      "step": 67400
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7426375150680542,
      "learning_rate": 0.0004819782916882744,
      "loss": 2.0727,
      "step": 67401
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7665113210678101,
      "learning_rate": 0.0004819750396272271,
      "loss": 2.1232,
      "step": 67402
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7848499417304993,
      "learning_rate": 0.00048197178753234715,
      "loss": 2.166,
      "step": 67403
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7311826944351196,
      "learning_rate": 0.00048196853540363526,
      "loss": 2.2167,
      "step": 67404
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.8248296976089478,
      "learning_rate": 0.00048196528324109204,
      "loss": 2.373,
      "step": 67405
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7718608975410461,
      "learning_rate": 0.00048196203104471807,
      "loss": 2.2035,
      "step": 67406
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7925191521644592,
      "learning_rate": 0.00048195877881451374,
      "loss": 2.1331,
      "step": 67407
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7681962251663208,
      "learning_rate": 0.00048195552655048003,
      "loss": 2.2953,
      "step": 67408
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.8075483441352844,
      "learning_rate": 0.00048195227425261726,
      "loss": 2.159,
      "step": 67409
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7488425970077515,
      "learning_rate": 0.0004819490219209261,
      "loss": 2.2409,
      "step": 67410
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.8282750844955444,
      "learning_rate": 0.00048194576955540725,
      "loss": 2.6128,
      "step": 67411
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7924769520759583,
      "learning_rate": 0.0004819425171560612,
      "loss": 2.2697,
      "step": 67412
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7432215809822083,
      "learning_rate": 0.00048193926472288865,
      "loss": 2.1747,
      "step": 67413
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7326067686080933,
      "learning_rate": 0.0004819360122558902,
      "loss": 2.2077,
      "step": 67414
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7476804256439209,
      "learning_rate": 0.00048193275975506633,
      "loss": 2.093,
      "step": 67415
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.8372213840484619,
      "learning_rate": 0.00048192950722041766,
      "loss": 2.2964,
      "step": 67416
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7104945182800293,
      "learning_rate": 0.000481926254651945,
      "loss": 2.2532,
      "step": 67417
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.8085311055183411,
      "learning_rate": 0.0004819230020496488,
      "loss": 2.2035,
      "step": 67418
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7801998853683472,
      "learning_rate": 0.00048191974941352953,
      "loss": 2.2899,
      "step": 67419
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7641534209251404,
      "learning_rate": 0.00048191649674358814,
      "loss": 2.2903,
      "step": 67420
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.791404128074646,
      "learning_rate": 0.00048191324403982485,
      "loss": 2.3658,
      "step": 67421
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.8364405035972595,
      "learning_rate": 0.0004819099913022406,
      "loss": 2.0329,
      "step": 67422
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7024243474006653,
      "learning_rate": 0.00048190673853083576,
      "loss": 2.2146,
      "step": 67423
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.8837987184524536,
      "learning_rate": 0.00048190348572561115,
      "loss": 2.1529,
      "step": 67424
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7595704793930054,
      "learning_rate": 0.00048190023288656713,
      "loss": 2.3066,
      "step": 67425
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.971132755279541,
      "learning_rate": 0.00048189698001370446,
      "loss": 2.143,
      "step": 67426
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.8066801428794861,
      "learning_rate": 0.0004818937271070236,
      "loss": 2.2312,
      "step": 67427
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.8378674983978271,
      "learning_rate": 0.0004818904741665255,
      "loss": 2.3219,
      "step": 67428
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7572953701019287,
      "learning_rate": 0.0004818872211922103,
      "loss": 2.3766,
      "step": 67429
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.8415932655334473,
      "learning_rate": 0.000481883968184079,
      "loss": 2.1451,
      "step": 67430
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7496857643127441,
      "learning_rate": 0.00048188071514213194,
      "loss": 2.245,
      "step": 67431
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7284197211265564,
      "learning_rate": 0.00048187746206636983,
      "loss": 1.9271,
      "step": 67432
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7522780299186707,
      "learning_rate": 0.00048187420895679325,
      "loss": 2.253,
      "step": 67433
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7037391066551208,
      "learning_rate": 0.0004818709558134029,
      "loss": 2.2854,
      "step": 67434
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7693600654602051,
      "learning_rate": 0.0004818677026361992,
      "loss": 2.169,
      "step": 67435
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7316022515296936,
      "learning_rate": 0.0004818644494251829,
      "loss": 2.1202,
      "step": 67436
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7495138645172119,
      "learning_rate": 0.0004818611961803546,
      "loss": 2.293,
      "step": 67437
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7498175501823425,
      "learning_rate": 0.00048185794290171485,
      "loss": 2.1523,
      "step": 67438
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.8226870894432068,
      "learning_rate": 0.00048185468958926436,
      "loss": 2.1618,
      "step": 67439
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7288411259651184,
      "learning_rate": 0.00048185143624300357,
      "loss": 2.1832,
      "step": 67440
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.6866003274917603,
      "learning_rate": 0.0004818481828629332,
      "loss": 2.1524,
      "step": 67441
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.72789067029953,
      "learning_rate": 0.00048184492944905374,
      "loss": 2.1534,
      "step": 67442
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.90460604429245,
      "learning_rate": 0.0004818416760013659,
      "loss": 2.2479,
      "step": 67443
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7800241112709045,
      "learning_rate": 0.0004818384225198703,
      "loss": 2.3235,
      "step": 67444
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.8215741515159607,
      "learning_rate": 0.0004818351690045675,
      "loss": 2.1116,
      "step": 67445
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.759058952331543,
      "learning_rate": 0.0004818319154554581,
      "loss": 2.029,
      "step": 67446
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.764767050743103,
      "learning_rate": 0.0004818286618725427,
      "loss": 2.3713,
      "step": 67447
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7219122052192688,
      "learning_rate": 0.0004818254082558219,
      "loss": 2.1496,
      "step": 67448
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.8934426307678223,
      "learning_rate": 0.00048182215460529636,
      "loss": 2.1077,
      "step": 67449
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.8054167032241821,
      "learning_rate": 0.00048181890092096664,
      "loss": 2.1119,
      "step": 67450
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.8332302570343018,
      "learning_rate": 0.0004818156472028334,
      "loss": 2.3491,
      "step": 67451
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.8241938948631287,
      "learning_rate": 0.0004818123934508971,
      "loss": 2.2796,
      "step": 67452
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7001253962516785,
      "learning_rate": 0.0004818091396651585,
      "loss": 2.195,
      "step": 67453
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.711474597454071,
      "learning_rate": 0.0004818058858456182,
      "loss": 2.1612,
      "step": 67454
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7753554582595825,
      "learning_rate": 0.0004818026319922767,
      "loss": 2.1307,
      "step": 67455
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.8238754272460938,
      "learning_rate": 0.00048179937810513464,
      "loss": 2.0572,
      "step": 67456
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7628582119941711,
      "learning_rate": 0.0004817961241841927,
      "loss": 2.1179,
      "step": 67457
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.6622436046600342,
      "learning_rate": 0.0004817928702294514,
      "loss": 2.1056,
      "step": 67458
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.8440585136413574,
      "learning_rate": 0.00048178961624091137,
      "loss": 2.2634,
      "step": 67459
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.9177443981170654,
      "learning_rate": 0.0004817863622185732,
      "loss": 2.3098,
      "step": 67460
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.8946168422698975,
      "learning_rate": 0.00048178310816243756,
      "loss": 2.2024,
      "step": 67461
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7670770883560181,
      "learning_rate": 0.000481779854072505,
      "loss": 2.2028,
      "step": 67462
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7952684164047241,
      "learning_rate": 0.00048177659994877615,
      "loss": 2.2618,
      "step": 67463
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7831988334655762,
      "learning_rate": 0.0004817733457912515,
      "loss": 2.1107,
      "step": 67464
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7708569765090942,
      "learning_rate": 0.00048177009159993176,
      "loss": 2.373,
      "step": 67465
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.795115053653717,
      "learning_rate": 0.0004817668373748177,
      "loss": 2.2963,
      "step": 67466
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7536752223968506,
      "learning_rate": 0.00048176358311590963,
      "loss": 2.3241,
      "step": 67467
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7766674160957336,
      "learning_rate": 0.00048176032882320816,
      "loss": 2.3965,
      "step": 67468
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7832762598991394,
      "learning_rate": 0.00048175707449671433,
      "loss": 2.3053,
      "step": 67469
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7545444965362549,
      "learning_rate": 0.00048175382013642814,
      "loss": 2.3297,
      "step": 67470
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.772921085357666,
      "learning_rate": 0.00048175056574235056,
      "loss": 2.1347,
      "step": 67471
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.8892498016357422,
      "learning_rate": 0.00048174731131448215,
      "loss": 2.1702,
      "step": 67472
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7187473773956299,
      "learning_rate": 0.0004817440568528235,
      "loss": 2.012,
      "step": 67473
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7235631942749023,
      "learning_rate": 0.0004817408023573751,
      "loss": 2.21,
      "step": 67474
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.8551329970359802,
      "learning_rate": 0.0004817375478281378,
      "loss": 2.1262,
      "step": 67475
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.8308204412460327,
      "learning_rate": 0.000481734293265112,
      "loss": 2.4194,
      "step": 67476
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.797927737236023,
      "learning_rate": 0.0004817310386682984,
      "loss": 2.5656,
      "step": 67477
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7450308203697205,
      "learning_rate": 0.00048172778403769755,
      "loss": 2.401,
      "step": 67478
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.8787452578544617,
      "learning_rate": 0.00048172452937330994,
      "loss": 2.1678,
      "step": 67479
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7703081965446472,
      "learning_rate": 0.00048172127467513643,
      "loss": 2.4629,
      "step": 67480
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7530520558357239,
      "learning_rate": 0.00048171801994317755,
      "loss": 2.2482,
      "step": 67481
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.8886854648590088,
      "learning_rate": 0.00048171476517743384,
      "loss": 2.2507,
      "step": 67482
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.86049884557724,
      "learning_rate": 0.0004817115103779058,
      "loss": 2.4448,
      "step": 67483
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7747399806976318,
      "learning_rate": 0.0004817082555445943,
      "loss": 1.9546,
      "step": 67484
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.757952868938446,
      "learning_rate": 0.00048170500067749975,
      "loss": 2.2296,
      "step": 67485
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.9085976481437683,
      "learning_rate": 0.0004817017457766228,
      "loss": 2.0571,
      "step": 67486
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7805067896842957,
      "learning_rate": 0.0004816984908419641,
      "loss": 2.095,
      "step": 67487
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.832145631313324,
      "learning_rate": 0.0004816952358735242,
      "loss": 2.2016,
      "step": 67488
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.8037509918212891,
      "learning_rate": 0.00048169198087130374,
      "loss": 2.3841,
      "step": 67489
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7575249671936035,
      "learning_rate": 0.0004816887258353033,
      "loss": 2.437,
      "step": 67490
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.6890841722488403,
      "learning_rate": 0.00048168547076552344,
      "loss": 2.2003,
      "step": 67491
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7139618396759033,
      "learning_rate": 0.0004816822156619649,
      "loss": 2.2895,
      "step": 67492
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7014123201370239,
      "learning_rate": 0.00048167896052462813,
      "loss": 1.951,
      "step": 67493
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.9422043561935425,
      "learning_rate": 0.00048167570535351397,
      "loss": 2.399,
      "step": 67494
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7246080040931702,
      "learning_rate": 0.00048167245014862274,
      "loss": 2.355,
      "step": 67495
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7835918068885803,
      "learning_rate": 0.00048166919490995515,
      "loss": 2.2724,
      "step": 67496
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7587388753890991,
      "learning_rate": 0.0004816659396375119,
      "loss": 2.0021,
      "step": 67497
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7188851237297058,
      "learning_rate": 0.00048166268433129345,
      "loss": 2.0853,
      "step": 67498
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7440348267555237,
      "learning_rate": 0.00048165942899130055,
      "loss": 2.2196,
      "step": 67499
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.8378003835678101,
      "learning_rate": 0.0004816561736175337,
      "loss": 2.4087,
      "step": 67500
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7204294204711914,
      "learning_rate": 0.0004816529182099935,
      "loss": 2.3289,
      "step": 67501
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7281321883201599,
      "learning_rate": 0.0004816496627686807,
      "loss": 2.1907,
      "step": 67502
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.8404684066772461,
      "learning_rate": 0.0004816464072935958,
      "loss": 2.1035,
      "step": 67503
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7610151171684265,
      "learning_rate": 0.00048164315178473926,
      "loss": 2.1328,
      "step": 67504
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7578087449073792,
      "learning_rate": 0.0004816398962421119,
      "loss": 2.3059,
      "step": 67505
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7394981980323792,
      "learning_rate": 0.00048163664066571435,
      "loss": 2.0437,
      "step": 67506
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7467379570007324,
      "learning_rate": 0.00048163338505554694,
      "loss": 2.0648,
      "step": 67507
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7883866429328918,
      "learning_rate": 0.0004816301294116106,
      "loss": 2.1624,
      "step": 67508
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7862486243247986,
      "learning_rate": 0.00048162687373390573,
      "loss": 1.9868,
      "step": 67509
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.8397005796432495,
      "learning_rate": 0.00048162361802243307,
      "loss": 2.2203,
      "step": 67510
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.8326914310455322,
      "learning_rate": 0.000481620362277193,
      "loss": 2.1185,
      "step": 67511
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.8096774816513062,
      "learning_rate": 0.00048161710649818646,
      "loss": 2.4218,
      "step": 67512
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7942575812339783,
      "learning_rate": 0.00048161385068541375,
      "loss": 2.0002,
      "step": 67513
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7117404937744141,
      "learning_rate": 0.0004816105948388756,
      "loss": 2.1813,
      "step": 67514
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7464461922645569,
      "learning_rate": 0.0004816073389585727,
      "loss": 2.4049,
      "step": 67515
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7302122116088867,
      "learning_rate": 0.00048160408304450536,
      "loss": 2.1499,
      "step": 67516
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.8551588654518127,
      "learning_rate": 0.00048160082709667457,
      "loss": 2.2474,
      "step": 67517
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.738025426864624,
      "learning_rate": 0.00048159757111508074,
      "loss": 2.3163,
      "step": 67518
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.6929049491882324,
      "learning_rate": 0.0004815943150997245,
      "loss": 2.0756,
      "step": 67519
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.754175066947937,
      "learning_rate": 0.00048159105905060644,
      "loss": 2.3185,
      "step": 67520
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7833830118179321,
      "learning_rate": 0.0004815878029677271,
      "loss": 2.2348,
      "step": 67521
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.8017891645431519,
      "learning_rate": 0.00048158454685108727,
      "loss": 2.3328,
      "step": 67522
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7235301733016968,
      "learning_rate": 0.00048158129070068734,
      "loss": 2.0399,
      "step": 67523
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.8786382675170898,
      "learning_rate": 0.0004815780345165281,
      "loss": 2.4187,
      "step": 67524
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.8206352591514587,
      "learning_rate": 0.00048157477829861,
      "loss": 2.2569,
      "step": 67525
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.8220681548118591,
      "learning_rate": 0.0004815715220469338,
      "loss": 2.1863,
      "step": 67526
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.8849300146102905,
      "learning_rate": 0.00048156826576150003,
      "loss": 2.2854,
      "step": 67527
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7878568768501282,
      "learning_rate": 0.0004815650094423092,
      "loss": 2.0344,
      "step": 67528
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7353805899620056,
      "learning_rate": 0.00048156175308936206,
      "loss": 2.1486,
      "step": 67529
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7766516804695129,
      "learning_rate": 0.0004815584967026592,
      "loss": 2.1584,
      "step": 67530
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7453651428222656,
      "learning_rate": 0.0004815552402822012,
      "loss": 2.2678,
      "step": 67531
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.8772819638252258,
      "learning_rate": 0.00048155198382798857,
      "loss": 2.1037,
      "step": 67532
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7977944612503052,
      "learning_rate": 0.0004815487273400221,
      "loss": 2.3783,
      "step": 67533
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.8418499231338501,
      "learning_rate": 0.0004815454708183022,
      "loss": 2.1768,
      "step": 67534
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7046923637390137,
      "learning_rate": 0.0004815422142628296,
      "loss": 2.1113,
      "step": 67535
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7770276069641113,
      "learning_rate": 0.0004815389576736049,
      "loss": 1.8629,
      "step": 67536
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.741770327091217,
      "learning_rate": 0.0004815357010506287,
      "loss": 2.1607,
      "step": 67537
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.9406720399856567,
      "learning_rate": 0.00048153244439390146,
      "loss": 2.1232,
      "step": 67538
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.789261519908905,
      "learning_rate": 0.000481529187703424,
      "loss": 2.2949,
      "step": 67539
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.8853893876075745,
      "learning_rate": 0.0004815259309791969,
      "loss": 2.2714,
      "step": 67540
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.8111243844032288,
      "learning_rate": 0.00048152267422122054,
      "loss": 2.209,
      "step": 67541
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.840354859828949,
      "learning_rate": 0.00048151941742949583,
      "loss": 2.4901,
      "step": 67542
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7626011371612549,
      "learning_rate": 0.0004815161606040232,
      "loss": 2.0749,
      "step": 67543
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7851340174674988,
      "learning_rate": 0.0004815129037448033,
      "loss": 2.2209,
      "step": 67544
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7614040374755859,
      "learning_rate": 0.00048150964685183677,
      "loss": 2.2877,
      "step": 67545
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.8981332182884216,
      "learning_rate": 0.00048150638992512407,
      "loss": 2.3345,
      "step": 67546
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7763856649398804,
      "learning_rate": 0.0004815031329646659,
      "loss": 2.2229,
      "step": 67547
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7935956716537476,
      "learning_rate": 0.000481499875970463,
      "loss": 2.5368,
      "step": 67548
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7542468309402466,
      "learning_rate": 0.0004814966189425157,
      "loss": 2.2323,
      "step": 67549
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7856955528259277,
      "learning_rate": 0.00048149336188082483,
      "loss": 2.1714,
      "step": 67550
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.8120522499084473,
      "learning_rate": 0.0004814901047853909,
      "loss": 2.3579,
      "step": 67551
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7693755626678467,
      "learning_rate": 0.0004814868476562145,
      "loss": 2.1964,
      "step": 67552
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.9083817005157471,
      "learning_rate": 0.00048148359049329635,
      "loss": 2.451,
      "step": 67553
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.8001750111579895,
      "learning_rate": 0.00048148033329663693,
      "loss": 2.1906,
      "step": 67554
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.8099073171615601,
      "learning_rate": 0.00048147707606623687,
      "loss": 2.159,
      "step": 67555
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.9062535762786865,
      "learning_rate": 0.0004814738188020968,
      "loss": 1.9897,
      "step": 67556
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.8466832041740417,
      "learning_rate": 0.0004814705615042174,
      "loss": 2.0717,
      "step": 67557
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.8016937375068665,
      "learning_rate": 0.0004814673041725991,
      "loss": 2.0147,
      "step": 67558
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7747564315795898,
      "learning_rate": 0.0004814640468072426,
      "loss": 2.2127,
      "step": 67559
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7529509663581848,
      "learning_rate": 0.00048146078940814856,
      "loss": 2.2986,
      "step": 67560
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7827427983283997,
      "learning_rate": 0.0004814575319753176,
      "loss": 2.2605,
      "step": 67561
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7568643689155579,
      "learning_rate": 0.0004814542745087501,
      "loss": 2.2337,
      "step": 67562
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7106471657752991,
      "learning_rate": 0.00048145101700844693,
      "loss": 2.3506,
      "step": 67563
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7365410923957825,
      "learning_rate": 0.0004814477594744086,
      "loss": 1.9682,
      "step": 67564
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7746480107307434,
      "learning_rate": 0.00048144450190663564,
      "loss": 2.26,
      "step": 67565
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7214232087135315,
      "learning_rate": 0.00048144124430512874,
      "loss": 2.0609,
      "step": 67566
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.8282462954521179,
      "learning_rate": 0.0004814379866698885,
      "loss": 2.0993,
      "step": 67567
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7793415784835815,
      "learning_rate": 0.00048143472900091555,
      "loss": 2.507,
      "step": 67568
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7627957463264465,
      "learning_rate": 0.00048143147129821036,
      "loss": 2.1714,
      "step": 67569
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.8463664054870605,
      "learning_rate": 0.00048142821356177373,
      "loss": 2.2538,
      "step": 67570
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.8135808706283569,
      "learning_rate": 0.0004814249557916061,
      "loss": 2.4107,
      "step": 67571
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7450971603393555,
      "learning_rate": 0.00048142169798770826,
      "loss": 2.1579,
      "step": 67572
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7100313901901245,
      "learning_rate": 0.0004814184401500806,
      "loss": 1.9874,
      "step": 67573
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.8223276734352112,
      "learning_rate": 0.00048141518227872386,
      "loss": 2.0884,
      "step": 67574
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.8249185681343079,
      "learning_rate": 0.0004814119243736386,
      "loss": 2.2803,
      "step": 67575
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.675945520401001,
      "learning_rate": 0.00048140866643482547,
      "loss": 2.0158,
      "step": 67576
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7433452010154724,
      "learning_rate": 0.000481405408462285,
      "loss": 2.2407,
      "step": 67577
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.8313812613487244,
      "learning_rate": 0.00048140215045601784,
      "loss": 2.2217,
      "step": 67578
    },
    {
      "epoch": 0.88,
      "grad_norm": 2.1808993816375732,
      "learning_rate": 0.0004813988924160246,
      "loss": 2.1365,
      "step": 67579
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.8221882581710815,
      "learning_rate": 0.00048139563434230597,
      "loss": 2.1997,
      "step": 67580
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.824790358543396,
      "learning_rate": 0.00048139237623486236,
      "loss": 2.2568,
      "step": 67581
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.9147642850875854,
      "learning_rate": 0.0004813891180936946,
      "loss": 2.3859,
      "step": 67582
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7963331937789917,
      "learning_rate": 0.000481385859918803,
      "loss": 2.1532,
      "step": 67583
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7692711353302002,
      "learning_rate": 0.00048138260171018844,
      "loss": 2.0633,
      "step": 67584
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.8065856695175171,
      "learning_rate": 0.0004813793434678515,
      "loss": 2.057,
      "step": 67585
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7649160623550415,
      "learning_rate": 0.00048137608519179263,
      "loss": 2.2359,
      "step": 67586
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7777224183082581,
      "learning_rate": 0.0004813728268820125,
      "loss": 2.2404,
      "step": 67587
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7911548018455505,
      "learning_rate": 0.00048136956853851183,
      "loss": 2.3309,
      "step": 67588
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.8084859251976013,
      "learning_rate": 0.00048136631016129107,
      "loss": 2.2237,
      "step": 67589
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7778350114822388,
      "learning_rate": 0.00048136305175035086,
      "loss": 2.1881,
      "step": 67590
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.781196653842926,
      "learning_rate": 0.0004813597933056919,
      "loss": 2.2512,
      "step": 67591
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.8450536727905273,
      "learning_rate": 0.0004813565348273147,
      "loss": 2.3134,
      "step": 67592
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7486741542816162,
      "learning_rate": 0.0004813532763152199,
      "loss": 2.1157,
      "step": 67593
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7060490846633911,
      "learning_rate": 0.0004813500177694081,
      "loss": 2.0558,
      "step": 67594
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7682048678398132,
      "learning_rate": 0.0004813467591898799,
      "loss": 2.0622,
      "step": 67595
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.8181040287017822,
      "learning_rate": 0.000481343500576636,
      "loss": 2.1726,
      "step": 67596
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.745587170124054,
      "learning_rate": 0.00048134024192967684,
      "loss": 2.2081,
      "step": 67597
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.9172337651252747,
      "learning_rate": 0.0004813369832490031,
      "loss": 2.31,
      "step": 67598
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7527541518211365,
      "learning_rate": 0.00048133372453461543,
      "loss": 2.2567,
      "step": 67599
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7650070786476135,
      "learning_rate": 0.0004813304657865144,
      "loss": 2.0499,
      "step": 67600
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7873722314834595,
      "learning_rate": 0.0004813272070047006,
      "loss": 1.9921,
      "step": 67601
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7275636792182922,
      "learning_rate": 0.00048132394818917456,
      "loss": 2.3691,
      "step": 67602
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7471764087677002,
      "learning_rate": 0.0004813206893399371,
      "loss": 2.1374,
      "step": 67603
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7652050852775574,
      "learning_rate": 0.00048131743045698867,
      "loss": 2.1902,
      "step": 67604
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7935252785682678,
      "learning_rate": 0.0004813141715403299,
      "loss": 2.2811,
      "step": 67605
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7863772511482239,
      "learning_rate": 0.0004813109125899614,
      "loss": 2.0685,
      "step": 67606
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7930091619491577,
      "learning_rate": 0.0004813076536058837,
      "loss": 2.358,
      "step": 67607
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.8008758425712585,
      "learning_rate": 0.0004813043945880976,
      "loss": 2.1276,
      "step": 67608
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7909917235374451,
      "learning_rate": 0.00048130113553660354,
      "loss": 2.1573,
      "step": 67609
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7383347153663635,
      "learning_rate": 0.00048129787645140223,
      "loss": 2.1422,
      "step": 67610
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7894229888916016,
      "learning_rate": 0.0004812946173324941,
      "loss": 2.054,
      "step": 67611
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.802094578742981,
      "learning_rate": 0.00048129135817988,
      "loss": 2.1827,
      "step": 67612
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7807101011276245,
      "learning_rate": 0.0004812880989935604,
      "loss": 2.2832,
      "step": 67613
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7472065091133118,
      "learning_rate": 0.00048128483977353593,
      "loss": 2.0022,
      "step": 67614
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7816953659057617,
      "learning_rate": 0.00048128158051980715,
      "loss": 2.223,
      "step": 67615
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7196886539459229,
      "learning_rate": 0.00048127832123237465,
      "loss": 2.3053,
      "step": 67616
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.853696346282959,
      "learning_rate": 0.00048127506191123913,
      "loss": 2.1281,
      "step": 67617
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7594988346099854,
      "learning_rate": 0.0004812718025564012,
      "loss": 2.1416,
      "step": 67618
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7529904246330261,
      "learning_rate": 0.0004812685431678614,
      "loss": 2.3752,
      "step": 67619
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.796718180179596,
      "learning_rate": 0.0004812652837456203,
      "loss": 2.3409,
      "step": 67620
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7215931415557861,
      "learning_rate": 0.0004812620242896786,
      "loss": 2.0571,
      "step": 67621
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7065727710723877,
      "learning_rate": 0.00048125876480003686,
      "loss": 2.3955,
      "step": 67622
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7210286259651184,
      "learning_rate": 0.0004812555052766957,
      "loss": 2.1382,
      "step": 67623
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7475113272666931,
      "learning_rate": 0.00048125224571965577,
      "loss": 2.2737,
      "step": 67624
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7103195190429688,
      "learning_rate": 0.0004812489861289175,
      "loss": 2.5132,
      "step": 67625
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7766111493110657,
      "learning_rate": 0.00048124572650448173,
      "loss": 2.134,
      "step": 67626
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.738314688205719,
      "learning_rate": 0.0004812424668463489,
      "loss": 2.2705,
      "step": 67627
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7080544233322144,
      "learning_rate": 0.0004812392071545198,
      "loss": 2.2173,
      "step": 67628
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.6982748508453369,
      "learning_rate": 0.00048123594742899474,
      "loss": 2.25,
      "step": 67629
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.8223143815994263,
      "learning_rate": 0.00048123268766977453,
      "loss": 2.1802,
      "step": 67630
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.9078844785690308,
      "learning_rate": 0.0004812294278768599,
      "loss": 2.33,
      "step": 67631
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7755273580551147,
      "learning_rate": 0.0004812261680502511,
      "loss": 2.2144,
      "step": 67632
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7372268438339233,
      "learning_rate": 0.000481222908189949,
      "loss": 2.2238,
      "step": 67633
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.8013880252838135,
      "learning_rate": 0.0004812196482959542,
      "loss": 2.2461,
      "step": 67634
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7742553949356079,
      "learning_rate": 0.00048121638836826726,
      "loss": 2.2551,
      "step": 67635
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7498345375061035,
      "learning_rate": 0.00048121312840688867,
      "loss": 2.2646,
      "step": 67636
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7191005349159241,
      "learning_rate": 0.00048120986841181915,
      "loss": 2.3072,
      "step": 67637
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.8419685363769531,
      "learning_rate": 0.00048120660838305934,
      "loss": 2.3067,
      "step": 67638
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7573583126068115,
      "learning_rate": 0.0004812033483206098,
      "loss": 2.0062,
      "step": 67639
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.9681615829467773,
      "learning_rate": 0.00048120008822447117,
      "loss": 2.2486,
      "step": 67640
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7237126231193542,
      "learning_rate": 0.00048119682809464394,
      "loss": 2.0895,
      "step": 67641
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.8851133584976196,
      "learning_rate": 0.00048119356793112887,
      "loss": 2.0337,
      "step": 67642
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7846494317054749,
      "learning_rate": 0.00048119030773392644,
      "loss": 2.1319,
      "step": 67643
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7548224329948425,
      "learning_rate": 0.0004811870475030373,
      "loss": 2.2182,
      "step": 67644
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7698423862457275,
      "learning_rate": 0.0004811837872384622,
      "loss": 2.1808,
      "step": 67645
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.9358806610107422,
      "learning_rate": 0.0004811805269402015,
      "loss": 2.3585,
      "step": 67646
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7924357652664185,
      "learning_rate": 0.00048117726660825593,
      "loss": 2.2196,
      "step": 67647
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7671747803688049,
      "learning_rate": 0.0004811740062426261,
      "loss": 2.1637,
      "step": 67648
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7673481106758118,
      "learning_rate": 0.0004811707458433126,
      "loss": 2.0271,
      "step": 67649
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7230339050292969,
      "learning_rate": 0.00048116748541031596,
      "loss": 2.1692,
      "step": 67650
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.8060752153396606,
      "learning_rate": 0.0004811642249436369,
      "loss": 2.52,
      "step": 67651
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7439325451850891,
      "learning_rate": 0.0004811609644432761,
      "loss": 2.1603,
      "step": 67652
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7473366856575012,
      "learning_rate": 0.00048115770390923403,
      "loss": 2.3593,
      "step": 67653
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7752019166946411,
      "learning_rate": 0.00048115444334151126,
      "loss": 2.1845,
      "step": 67654
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7078080773353577,
      "learning_rate": 0.00048115118274010845,
      "loss": 2.2015,
      "step": 67655
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7252870202064514,
      "learning_rate": 0.00048114792210502625,
      "loss": 2.321,
      "step": 67656
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7701039910316467,
      "learning_rate": 0.00048114466143626527,
      "loss": 2.1454,
      "step": 67657
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.8893454670906067,
      "learning_rate": 0.000481141400733826,
      "loss": 2.3931,
      "step": 67658
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7595271468162537,
      "learning_rate": 0.00048113813999770916,
      "loss": 2.3136,
      "step": 67659
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.9267526865005493,
      "learning_rate": 0.0004811348792279154,
      "loss": 2.2312,
      "step": 67660
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7579396367073059,
      "learning_rate": 0.0004811316184244451,
      "loss": 2.2098,
      "step": 67661
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7806925177574158,
      "learning_rate": 0.00048112835758729906,
      "loss": 2.2762,
      "step": 67662
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7661358714103699,
      "learning_rate": 0.00048112509671647787,
      "loss": 2.115,
      "step": 67663
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7295591831207275,
      "learning_rate": 0.00048112183581198217,
      "loss": 2.2699,
      "step": 67664
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7533268928527832,
      "learning_rate": 0.0004811185748738124,
      "loss": 2.1133,
      "step": 67665
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.6609579920768738,
      "learning_rate": 0.00048111531390196925,
      "loss": 2.0096,
      "step": 67666
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.2352516651153564,
      "learning_rate": 0.00048111205289645345,
      "loss": 2.2516,
      "step": 67667
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.841863214969635,
      "learning_rate": 0.0004811087918572654,
      "loss": 2.3133,
      "step": 67668
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.8236646056175232,
      "learning_rate": 0.0004811055307844058,
      "loss": 2.2399,
      "step": 67669
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.8465625643730164,
      "learning_rate": 0.00048110226967787537,
      "loss": 2.1371,
      "step": 67670
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7175601720809937,
      "learning_rate": 0.00048109900853767455,
      "loss": 2.0084,
      "step": 67671
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7943918704986572,
      "learning_rate": 0.000481095747363804,
      "loss": 2.3649,
      "step": 67672
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.8396710753440857,
      "learning_rate": 0.0004810924861562644,
      "loss": 2.284,
      "step": 67673
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7240161299705505,
      "learning_rate": 0.00048108922491505617,
      "loss": 2.1575,
      "step": 67674
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.8286395072937012,
      "learning_rate": 0.0004810859636401801,
      "loss": 2.1808,
      "step": 67675
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.8119491934776306,
      "learning_rate": 0.0004810827023316367,
      "loss": 2.5447,
      "step": 67676
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.75199294090271,
      "learning_rate": 0.00048107944098942665,
      "loss": 2.3517,
      "step": 67677
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7713935375213623,
      "learning_rate": 0.0004810761796135505,
      "loss": 2.3771,
      "step": 67678
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.8790249824523926,
      "learning_rate": 0.0004810729182040089,
      "loss": 2.2656,
      "step": 67679
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7746114730834961,
      "learning_rate": 0.00048106965676080236,
      "loss": 2.1307,
      "step": 67680
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7903823256492615,
      "learning_rate": 0.0004810663952839315,
      "loss": 2.1409,
      "step": 67681
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7513973116874695,
      "learning_rate": 0.0004810631337733972,
      "loss": 2.0453,
      "step": 67682
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7386381030082703,
      "learning_rate": 0.00048105987222919966,
      "loss": 2.3649,
      "step": 67683
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.8134201765060425,
      "learning_rate": 0.00048105661065133966,
      "loss": 2.3804,
      "step": 67684
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.75849449634552,
      "learning_rate": 0.00048105334903981794,
      "loss": 1.993,
      "step": 67685
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7946389317512512,
      "learning_rate": 0.00048105008739463494,
      "loss": 2.2437,
      "step": 67686
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.8055689334869385,
      "learning_rate": 0.00048104682571579126,
      "loss": 1.9683,
      "step": 67687
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.6366305947303772,
      "learning_rate": 0.00048104356400328755,
      "loss": 2.0577,
      "step": 67688
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.8570786714553833,
      "learning_rate": 0.0004810403022571245,
      "loss": 2.2652,
      "step": 67689
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.8030480146408081,
      "learning_rate": 0.0004810370404773026,
      "loss": 2.0557,
      "step": 67690
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.8334009051322937,
      "learning_rate": 0.0004810337786638225,
      "loss": 2.1165,
      "step": 67691
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.8789821267127991,
      "learning_rate": 0.00048103051681668485,
      "loss": 2.2524,
      "step": 67692
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.4128049612045288,
      "learning_rate": 0.00048102725493589013,
      "loss": 2.2973,
      "step": 67693
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7724784016609192,
      "learning_rate": 0.000481023993021439,
      "loss": 2.354,
      "step": 67694
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7266797423362732,
      "learning_rate": 0.0004810207310733322,
      "loss": 1.8409,
      "step": 67695
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.8975969552993774,
      "learning_rate": 0.00048101746909157025,
      "loss": 2.2187,
      "step": 67696
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.8141018152236938,
      "learning_rate": 0.0004810142070761536,
      "loss": 2.2817,
      "step": 67697
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7496604919433594,
      "learning_rate": 0.00048101094502708304,
      "loss": 2.2867,
      "step": 67698
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7351111769676208,
      "learning_rate": 0.0004810076829443592,
      "loss": 2.107,
      "step": 67699
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7958410382270813,
      "learning_rate": 0.0004810044208279825,
      "loss": 2.0422,
      "step": 67700
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.8376737236976624,
      "learning_rate": 0.00048100115867795375,
      "loss": 2.1404,
      "step": 67701
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7445804476737976,
      "learning_rate": 0.00048099789649427345,
      "loss": 2.1026,
      "step": 67702
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.8464241623878479,
      "learning_rate": 0.00048099463427694227,
      "loss": 2.3299,
      "step": 67703
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.8257797360420227,
      "learning_rate": 0.00048099137202596066,
      "loss": 2.2513,
      "step": 67704
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7408672571182251,
      "learning_rate": 0.0004809881097413294,
      "loss": 2.3928,
      "step": 67705
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.0161701440811157,
      "learning_rate": 0.00048098484742304905,
      "loss": 2.0572,
      "step": 67706
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7715833783149719,
      "learning_rate": 0.00048098158507112025,
      "loss": 2.2021,
      "step": 67707
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7795800566673279,
      "learning_rate": 0.0004809783226855434,
      "loss": 2.2063,
      "step": 67708
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.8146698474884033,
      "learning_rate": 0.00048097506026631936,
      "loss": 2.4015,
      "step": 67709
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7435253858566284,
      "learning_rate": 0.0004809717978134487,
      "loss": 2.1694,
      "step": 67710
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7513852715492249,
      "learning_rate": 0.00048096853532693184,
      "loss": 2.281,
      "step": 67711
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.9360318183898926,
      "learning_rate": 0.00048096527280676955,
      "loss": 2.0679,
      "step": 67712
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.8341954350471497,
      "learning_rate": 0.0004809620102529625,
      "loss": 2.1138,
      "step": 67713
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.8310441374778748,
      "learning_rate": 0.0004809587476655111,
      "loss": 2.1213,
      "step": 67714
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7837573289871216,
      "learning_rate": 0.000480955485044416,
      "loss": 2.2526,
      "step": 67715
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.8304706811904907,
      "learning_rate": 0.00048095222238967795,
      "loss": 2.3,
      "step": 67716
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.8293466567993164,
      "learning_rate": 0.0004809489597012974,
      "loss": 2.096,
      "step": 67717
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.8504772186279297,
      "learning_rate": 0.00048094569697927517,
      "loss": 2.0198,
      "step": 67718
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7402859330177307,
      "learning_rate": 0.00048094243422361165,
      "loss": 2.5294,
      "step": 67719
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.9193617701530457,
      "learning_rate": 0.00048093917143430745,
      "loss": 2.3265,
      "step": 67720
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7989915609359741,
      "learning_rate": 0.0004809359086113632,
      "loss": 2.255,
      "step": 67721
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7640171051025391,
      "learning_rate": 0.0004809326457547797,
      "loss": 2.2356,
      "step": 67722
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7353593707084656,
      "learning_rate": 0.00048092938286455733,
      "loss": 2.1601,
      "step": 67723
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.758532702922821,
      "learning_rate": 0.00048092611994069676,
      "loss": 2.3681,
      "step": 67724
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7821378111839294,
      "learning_rate": 0.00048092285698319865,
      "loss": 2.4209,
      "step": 67725
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7472708821296692,
      "learning_rate": 0.00048091959399206355,
      "loss": 2.2048,
      "step": 67726
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.8464715480804443,
      "learning_rate": 0.00048091633096729205,
      "loss": 2.1215,
      "step": 67727
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7368746399879456,
      "learning_rate": 0.0004809130679088848,
      "loss": 2.2708,
      "step": 67728
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.0057309865951538,
      "learning_rate": 0.00048090980481684247,
      "loss": 2.2309,
      "step": 67729
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.8303714990615845,
      "learning_rate": 0.00048090654169116547,
      "loss": 2.1739,
      "step": 67730
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.8332535624504089,
      "learning_rate": 0.0004809032785318546,
      "loss": 2.3972,
      "step": 67731
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.0585472583770752,
      "learning_rate": 0.00048090001533891035,
      "loss": 2.2363,
      "step": 67732
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7744922637939453,
      "learning_rate": 0.00048089675211233343,
      "loss": 2.3506,
      "step": 67733
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.762833297252655,
      "learning_rate": 0.00048089348885212434,
      "loss": 2.3539,
      "step": 67734
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.8524764180183411,
      "learning_rate": 0.0004808902255582838,
      "loss": 2.4213,
      "step": 67735
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7835204005241394,
      "learning_rate": 0.0004808869622308123,
      "loss": 2.113,
      "step": 67736
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7446216344833374,
      "learning_rate": 0.00048088369886971057,
      "loss": 2.0376,
      "step": 67737
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.6827059984207153,
      "learning_rate": 0.00048088043547497904,
      "loss": 2.0702,
      "step": 67738
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.8178641200065613,
      "learning_rate": 0.00048087717204661843,
      "loss": 2.3113,
      "step": 67739
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7916970252990723,
      "learning_rate": 0.0004808739085846294,
      "loss": 2.283,
      "step": 67740
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7822158932685852,
      "learning_rate": 0.0004808706450890125,
      "loss": 1.9786,
      "step": 67741
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.8332691788673401,
      "learning_rate": 0.00048086738155976827,
      "loss": 2.1469,
      "step": 67742
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.8026912212371826,
      "learning_rate": 0.0004808641179968974,
      "loss": 2.1665,
      "step": 67743
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7790041565895081,
      "learning_rate": 0.00048086085440040053,
      "loss": 2.0839,
      "step": 67744
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7827446460723877,
      "learning_rate": 0.0004808575907702782,
      "loss": 2.3146,
      "step": 67745
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7642008662223816,
      "learning_rate": 0.00048085432710653106,
      "loss": 2.2339,
      "step": 67746
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7485726475715637,
      "learning_rate": 0.00048085106340915956,
      "loss": 2.251,
      "step": 67747
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.9321109652519226,
      "learning_rate": 0.00048084779967816447,
      "loss": 2.3321,
      "step": 67748
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.8489705920219421,
      "learning_rate": 0.00048084453591354647,
      "loss": 2.1884,
      "step": 67749
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.85859614610672,
      "learning_rate": 0.0004808412721153059,
      "loss": 2.0199,
      "step": 67750
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.833387017250061,
      "learning_rate": 0.0004808380082834437,
      "loss": 2.1985,
      "step": 67751
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.8698165416717529,
      "learning_rate": 0.0004808347444179602,
      "loss": 2.278,
      "step": 67752
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.8392424583435059,
      "learning_rate": 0.00048083148051885613,
      "loss": 2.1227,
      "step": 67753
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.8148357272148132,
      "learning_rate": 0.00048082821658613204,
      "loss": 2.3323,
      "step": 67754
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.8044811487197876,
      "learning_rate": 0.0004808249526197886,
      "loss": 2.149,
      "step": 67755
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.8337361216545105,
      "learning_rate": 0.0004808216886198264,
      "loss": 2.1377,
      "step": 67756
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7740441560745239,
      "learning_rate": 0.000480818424586246,
      "loss": 2.2308,
      "step": 67757
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7782135009765625,
      "learning_rate": 0.00048081516051904803,
      "loss": 2.2474,
      "step": 67758
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7533371448516846,
      "learning_rate": 0.0004808118964182332,
      "loss": 2.1655,
      "step": 67759
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7324205040931702,
      "learning_rate": 0.0004808086322838019,
      "loss": 2.1086,
      "step": 67760
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7075577974319458,
      "learning_rate": 0.0004808053681157549,
      "loss": 2.2093,
      "step": 67761
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7613150477409363,
      "learning_rate": 0.00048080210391409283,
      "loss": 2.0935,
      "step": 67762
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7806139588356018,
      "learning_rate": 0.0004807988396788162,
      "loss": 2.2573,
      "step": 67763
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7477442026138306,
      "learning_rate": 0.00048079557540992566,
      "loss": 2.3271,
      "step": 67764
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7495553493499756,
      "learning_rate": 0.0004807923111074218,
      "loss": 2.0633,
      "step": 67765
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.8482139110565186,
      "learning_rate": 0.00048078904677130517,
      "loss": 2.2319,
      "step": 67766
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7662745714187622,
      "learning_rate": 0.00048078578240157654,
      "loss": 2.061,
      "step": 67767
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7407185435295105,
      "learning_rate": 0.0004807825179982365,
      "loss": 2.185,
      "step": 67768
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.761554479598999,
      "learning_rate": 0.0004807792535612854,
      "loss": 2.1168,
      "step": 67769
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7859838604927063,
      "learning_rate": 0.0004807759890907241,
      "loss": 2.4435,
      "step": 67770
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7467780113220215,
      "learning_rate": 0.00048077272458655316,
      "loss": 2.3824,
      "step": 67771
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7885226607322693,
      "learning_rate": 0.0004807694600487731,
      "loss": 2.2013,
      "step": 67772
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7871661186218262,
      "learning_rate": 0.00048076619547738464,
      "loss": 2.1089,
      "step": 67773
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.80488520860672,
      "learning_rate": 0.00048076293087238826,
      "loss": 1.9911,
      "step": 67774
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.8485549688339233,
      "learning_rate": 0.0004807596662337847,
      "loss": 2.2074,
      "step": 67775
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.8036102652549744,
      "learning_rate": 0.00048075640156157445,
      "loss": 2.2137,
      "step": 67776
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7921437621116638,
      "learning_rate": 0.0004807531368557582,
      "loss": 2.25,
      "step": 67777
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.906808614730835,
      "learning_rate": 0.0004807498721163366,
      "loss": 1.9511,
      "step": 67778
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7440335154533386,
      "learning_rate": 0.00048074660734331013,
      "loss": 2.0666,
      "step": 67779
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7363717555999756,
      "learning_rate": 0.00048074334253667947,
      "loss": 2.2329,
      "step": 67780
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7392978668212891,
      "learning_rate": 0.0004807400776964452,
      "loss": 2.2341,
      "step": 67781
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.8244208693504333,
      "learning_rate": 0.0004807368128226079,
      "loss": 2.1639,
      "step": 67782
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.8441076874732971,
      "learning_rate": 0.00048073354791516816,
      "loss": 2.1543,
      "step": 67783
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7448850274085999,
      "learning_rate": 0.0004807302829741268,
      "loss": 2.2124,
      "step": 67784
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7660137414932251,
      "learning_rate": 0.00048072701799948414,
      "loss": 2.2402,
      "step": 67785
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.8367798924446106,
      "learning_rate": 0.00048072375299124104,
      "loss": 2.2812,
      "step": 67786
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.8306022882461548,
      "learning_rate": 0.00048072048794939787,
      "loss": 2.1967,
      "step": 67787
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7456564903259277,
      "learning_rate": 0.0004807172228739554,
      "loss": 2.1816,
      "step": 67788
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7218129634857178,
      "learning_rate": 0.00048071395776491416,
      "loss": 2.4916,
      "step": 67789
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.8387436270713806,
      "learning_rate": 0.0004807106926222748,
      "loss": 2.3293,
      "step": 67790
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.8051859736442566,
      "learning_rate": 0.0004807074274460379,
      "loss": 2.1395,
      "step": 67791
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7931331992149353,
      "learning_rate": 0.00048070416223620416,
      "loss": 2.2994,
      "step": 67792
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.8347803950309753,
      "learning_rate": 0.000480700896992774,
      "loss": 2.3528,
      "step": 67793
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.905465304851532,
      "learning_rate": 0.0004806976317157481,
      "loss": 2.23,
      "step": 67794
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7516095042228699,
      "learning_rate": 0.00048069436640512726,
      "loss": 2.5599,
      "step": 67795
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.0236948728561401,
      "learning_rate": 0.00048069110106091175,
      "loss": 2.2568,
      "step": 67796
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.779097318649292,
      "learning_rate": 0.00048068783568310246,
      "loss": 2.4571,
      "step": 67797
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.726066529750824,
      "learning_rate": 0.0004806845702716999,
      "loss": 2.0626,
      "step": 67798
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7919312715530396,
      "learning_rate": 0.00048068130482670457,
      "loss": 2.2778,
      "step": 67799
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.0134927034378052,
      "learning_rate": 0.00048067803934811733,
      "loss": 2.3348,
      "step": 67800
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.863354504108429,
      "learning_rate": 0.0004806747738359385,
      "loss": 2.367,
      "step": 67801
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.8443279266357422,
      "learning_rate": 0.00048067150829016886,
      "loss": 2.1847,
      "step": 67802
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.8248094916343689,
      "learning_rate": 0.00048066824271080904,
      "loss": 2.2364,
      "step": 67803
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7681052684783936,
      "learning_rate": 0.0004806649770978595,
      "loss": 2.2207,
      "step": 67804
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.78757643699646,
      "learning_rate": 0.000480661711451321,
      "loss": 2.2713,
      "step": 67805
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.8054988384246826,
      "learning_rate": 0.000480658445771194,
      "loss": 2.3234,
      "step": 67806
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.1014546155929565,
      "learning_rate": 0.0004806551800574792,
      "loss": 2.4352,
      "step": 67807
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.8400236964225769,
      "learning_rate": 0.00048065191431017723,
      "loss": 2.4443,
      "step": 67808
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7487632036209106,
      "learning_rate": 0.0004806486485292886,
      "loss": 2.1808,
      "step": 67809
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.9326361417770386,
      "learning_rate": 0.000480645382714814,
      "loss": 2.1435,
      "step": 67810
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7545958757400513,
      "learning_rate": 0.00048064211686675405,
      "loss": 2.1941,
      "step": 67811
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7995709180831909,
      "learning_rate": 0.0004806388509851093,
      "loss": 2.2328,
      "step": 67812
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7210538387298584,
      "learning_rate": 0.00048063558506988044,
      "loss": 2.1849,
      "step": 67813
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.6671228408813477,
      "learning_rate": 0.0004806323191210679,
      "loss": 2.1739,
      "step": 67814
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7629424333572388,
      "learning_rate": 0.00048062905313867244,
      "loss": 2.1741,
      "step": 67815
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.768369197845459,
      "learning_rate": 0.0004806257871226947,
      "loss": 2.0726,
      "step": 67816
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7340085506439209,
      "learning_rate": 0.0004806225210731351,
      "loss": 2.1866,
      "step": 67817
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7563604116439819,
      "learning_rate": 0.00048061925498999434,
      "loss": 2.3676,
      "step": 67818
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7717064023017883,
      "learning_rate": 0.0004806159888732733,
      "loss": 2.1037,
      "step": 67819
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.806489109992981,
      "learning_rate": 0.0004806127227229721,
      "loss": 2.2568,
      "step": 67820
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.759811520576477,
      "learning_rate": 0.00048060945653909164,
      "loss": 2.2969,
      "step": 67821
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7693963646888733,
      "learning_rate": 0.00048060619032163254,
      "loss": 2.3038,
      "step": 67822
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.8105546832084656,
      "learning_rate": 0.0004806029240705952,
      "loss": 2.4078,
      "step": 67823
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7964242100715637,
      "learning_rate": 0.00048059965778598053,
      "loss": 2.2796,
      "step": 67824
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.72176194190979,
      "learning_rate": 0.00048059639146778887,
      "loss": 2.5583,
      "step": 67825
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7648013830184937,
      "learning_rate": 0.00048059312511602103,
      "loss": 2.2187,
      "step": 67826
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.8085736632347107,
      "learning_rate": 0.00048058985873067735,
      "loss": 2.2322,
      "step": 67827
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7431217432022095,
      "learning_rate": 0.0004805865923117587,
      "loss": 2.231,
      "step": 67828
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7103920578956604,
      "learning_rate": 0.00048058332585926565,
      "loss": 2.1478,
      "step": 67829
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7833465933799744,
      "learning_rate": 0.00048058005937319867,
      "loss": 2.2466,
      "step": 67830
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7310747504234314,
      "learning_rate": 0.00048057679285355844,
      "loss": 2.0487,
      "step": 67831
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7646681666374207,
      "learning_rate": 0.00048057352630034573,
      "loss": 2.3726,
      "step": 67832
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7501605153083801,
      "learning_rate": 0.00048057025971356087,
      "loss": 2.2291,
      "step": 67833
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.8081168532371521,
      "learning_rate": 0.00048056699309320455,
      "loss": 2.3866,
      "step": 67834
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.6929547190666199,
      "learning_rate": 0.0004805637264392775,
      "loss": 1.8904,
      "step": 67835
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7775982022285461,
      "learning_rate": 0.00048056045975178015,
      "loss": 2.2151,
      "step": 67836
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7309526801109314,
      "learning_rate": 0.0004805571930307133,
      "loss": 1.9114,
      "step": 67837
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7394514679908752,
      "learning_rate": 0.0004805539262760774,
      "loss": 2.3862,
      "step": 67838
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.8562138676643372,
      "learning_rate": 0.0004805506594878731,
      "loss": 2.3421,
      "step": 67839
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.75879967212677,
      "learning_rate": 0.0004805473926661011,
      "loss": 2.0928,
      "step": 67840
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7816740274429321,
      "learning_rate": 0.0004805441258107619,
      "loss": 2.2875,
      "step": 67841
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7158424854278564,
      "learning_rate": 0.0004805408589218561,
      "loss": 2.1999,
      "step": 67842
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.822790801525116,
      "learning_rate": 0.0004805375919993844,
      "loss": 2.3884,
      "step": 67843
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.6965141892433167,
      "learning_rate": 0.0004805343250433474,
      "loss": 2.3289,
      "step": 67844
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7531143426895142,
      "learning_rate": 0.00048053105805374554,
      "loss": 2.3632,
      "step": 67845
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.8820031881332397,
      "learning_rate": 0.0004805277910305796,
      "loss": 2.2394,
      "step": 67846
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.862501323223114,
      "learning_rate": 0.00048052452397385017,
      "loss": 2.3621,
      "step": 67847
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7744649648666382,
      "learning_rate": 0.0004805212568835578,
      "loss": 1.9869,
      "step": 67848
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7196617126464844,
      "learning_rate": 0.0004805179897597031,
      "loss": 2.323,
      "step": 67849
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.9159161448478699,
      "learning_rate": 0.0004805147226022867,
      "loss": 2.3904,
      "step": 67850
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7987574338912964,
      "learning_rate": 0.0004805114554113092,
      "loss": 2.47,
      "step": 67851
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.8215039372444153,
      "learning_rate": 0.0004805081881867712,
      "loss": 2.3608,
      "step": 67852
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7652302980422974,
      "learning_rate": 0.0004805049209286734,
      "loss": 2.3393,
      "step": 67853
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.738479733467102,
      "learning_rate": 0.0004805016536370163,
      "loss": 2.0456,
      "step": 67854
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7589467763900757,
      "learning_rate": 0.00048049838631180047,
      "loss": 2.4286,
      "step": 67855
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.0497198104858398,
      "learning_rate": 0.0004804951189530267,
      "loss": 2.2132,
      "step": 67856
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7422195672988892,
      "learning_rate": 0.00048049185156069543,
      "loss": 2.0614,
      "step": 67857
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.783433735370636,
      "learning_rate": 0.00048048858413480725,
      "loss": 2.451,
      "step": 67858
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7993407845497131,
      "learning_rate": 0.00048048531667536297,
      "loss": 2.2583,
      "step": 67859
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.001562476158142,
      "learning_rate": 0.00048048204918236295,
      "loss": 2.1778,
      "step": 67860
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7094374299049377,
      "learning_rate": 0.00048047878165580785,
      "loss": 2.3793,
      "step": 67861
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7580661773681641,
      "learning_rate": 0.0004804755140956984,
      "loss": 2.3691,
      "step": 67862
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.8093872666358948,
      "learning_rate": 0.0004804722465020352,
      "loss": 2.3198,
      "step": 67863
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.857025146484375,
      "learning_rate": 0.0004804689788748188,
      "loss": 2.2395,
      "step": 67864
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.8216773271560669,
      "learning_rate": 0.00048046571121404975,
      "loss": 2.1112,
      "step": 67865
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7597872614860535,
      "learning_rate": 0.0004804624435197288,
      "loss": 1.9552,
      "step": 67866
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.8116429448127747,
      "learning_rate": 0.0004804591757918564,
      "loss": 2.2791,
      "step": 67867
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.8156945705413818,
      "learning_rate": 0.00048045590803043325,
      "loss": 2.1926,
      "step": 67868
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.8151847124099731,
      "learning_rate": 0.00048045264023546,
      "loss": 2.4067,
      "step": 67869
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7159424424171448,
      "learning_rate": 0.0004804493724069371,
      "loss": 2.0695,
      "step": 67870
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.771163284778595,
      "learning_rate": 0.00048044610454486527,
      "loss": 2.0134,
      "step": 67871
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.9206110835075378,
      "learning_rate": 0.00048044283664924514,
      "loss": 2.0998,
      "step": 67872
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.900242805480957,
      "learning_rate": 0.0004804395687200773,
      "loss": 2.2293,
      "step": 67873
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.800096333026886,
      "learning_rate": 0.00048043630075736226,
      "loss": 2.1835,
      "step": 67874
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.0681259632110596,
      "learning_rate": 0.0004804330327611008,
      "loss": 2.2241,
      "step": 67875
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7203999757766724,
      "learning_rate": 0.00048042976473129336,
      "loss": 2.4276,
      "step": 67876
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7102364897727966,
      "learning_rate": 0.00048042649666794066,
      "loss": 2.2525,
      "step": 67877
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.9549354910850525,
      "learning_rate": 0.00048042322857104323,
      "loss": 2.1601,
      "step": 67878
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.8987049460411072,
      "learning_rate": 0.0004804199604406018,
      "loss": 2.1388,
      "step": 67879
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7883325815200806,
      "learning_rate": 0.0004804166922766169,
      "loss": 2.3927,
      "step": 67880
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7960825562477112,
      "learning_rate": 0.000480413424079089,
      "loss": 2.2359,
      "step": 67881
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7379144430160522,
      "learning_rate": 0.0004804101558480189,
      "loss": 2.1846,
      "step": 67882
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.9327352643013,
      "learning_rate": 0.00048040688758340717,
      "loss": 2.334,
      "step": 67883
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7052170038223267,
      "learning_rate": 0.0004804036192852544,
      "loss": 2.1164,
      "step": 67884
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7724363207817078,
      "learning_rate": 0.0004804003509535611,
      "loss": 2.5086,
      "step": 67885
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.8354884386062622,
      "learning_rate": 0.0004803970825883281,
      "loss": 2.2048,
      "step": 67886
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.749360203742981,
      "learning_rate": 0.0004803938141895558,
      "loss": 2.3506,
      "step": 67887
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7529410719871521,
      "learning_rate": 0.00048039054575724487,
      "loss": 2.3171,
      "step": 67888
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7477529048919678,
      "learning_rate": 0.00048038727729139604,
      "loss": 2.0664,
      "step": 67889
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7469513416290283,
      "learning_rate": 0.00048038400879200975,
      "loss": 2.1023,
      "step": 67890
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.9193992018699646,
      "learning_rate": 0.00048038074025908665,
      "loss": 2.2314,
      "step": 67891
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7172934412956238,
      "learning_rate": 0.0004803774716926273,
      "loss": 2.2736,
      "step": 67892
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.8408069014549255,
      "learning_rate": 0.0004803742030926325,
      "loss": 2.2373,
      "step": 67893
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.8246894478797913,
      "learning_rate": 0.00048037093445910264,
      "loss": 2.1438,
      "step": 67894
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7153156399726868,
      "learning_rate": 0.0004803676657920384,
      "loss": 2.2322,
      "step": 67895
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.9462683200836182,
      "learning_rate": 0.0004803643970914405,
      "loss": 2.4034,
      "step": 67896
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7895532846450806,
      "learning_rate": 0.0004803611283573094,
      "loss": 2.399,
      "step": 67897
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.8019081354141235,
      "learning_rate": 0.00048035785958964577,
      "loss": 2.2039,
      "step": 67898
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.8319385647773743,
      "learning_rate": 0.0004803545907884503,
      "loss": 2.2322,
      "step": 67899
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7738222479820251,
      "learning_rate": 0.00048035132195372335,
      "loss": 2.2768,
      "step": 67900
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.8047749996185303,
      "learning_rate": 0.0004803480530854658,
      "loss": 2.1592,
      "step": 67901
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.8253551721572876,
      "learning_rate": 0.0004803447841836781,
      "loss": 2.0548,
      "step": 67902
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.8079259991645813,
      "learning_rate": 0.00048034151524836086,
      "loss": 2.1144,
      "step": 67903
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.8445935249328613,
      "learning_rate": 0.0004803382462795147,
      "loss": 2.3545,
      "step": 67904
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.8406998515129089,
      "learning_rate": 0.00048033497727714046,
      "loss": 2.4805,
      "step": 67905
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7544258832931519,
      "learning_rate": 0.00048033170824123835,
      "loss": 2.2851,
      "step": 67906
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7701297998428345,
      "learning_rate": 0.0004803284391718092,
      "loss": 2.291,
      "step": 67907
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7256007194519043,
      "learning_rate": 0.0004803251700688537,
      "loss": 2.2148,
      "step": 67908
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7981542944908142,
      "learning_rate": 0.0004803219009323722,
      "loss": 2.2302,
      "step": 67909
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.8771587014198303,
      "learning_rate": 0.0004803186317623655,
      "loss": 2.1162,
      "step": 67910
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7488712668418884,
      "learning_rate": 0.00048031536255883423,
      "loss": 2.3777,
      "step": 67911
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.9333078861236572,
      "learning_rate": 0.00048031209332177884,
      "loss": 2.0717,
      "step": 67912
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7927396893501282,
      "learning_rate": 0.0004803088240512001,
      "loss": 2.3139,
      "step": 67913
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.6764751076698303,
      "learning_rate": 0.0004803055547470985,
      "loss": 2.0244,
      "step": 67914
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.8342719078063965,
      "learning_rate": 0.0004803022854094747,
      "loss": 2.0677,
      "step": 67915
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.8113405704498291,
      "learning_rate": 0.00048029901603832925,
      "loss": 2.1077,
      "step": 67916
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7668499946594238,
      "learning_rate": 0.00048029574663366293,
      "loss": 2.4757,
      "step": 67917
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7484986186027527,
      "learning_rate": 0.00048029247719547617,
      "loss": 2.3538,
      "step": 67918
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.693824052810669,
      "learning_rate": 0.00048028920772376967,
      "loss": 2.1799,
      "step": 67919
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7711295485496521,
      "learning_rate": 0.0004802859382185439,
      "loss": 1.9355,
      "step": 67920
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7156383991241455,
      "learning_rate": 0.00048028266867979964,
      "loss": 1.9917,
      "step": 67921
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7682040333747864,
      "learning_rate": 0.00048027939910753745,
      "loss": 2.0376,
      "step": 67922
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.8356922268867493,
      "learning_rate": 0.0004802761295017579,
      "loss": 2.1898,
      "step": 67923
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7661848664283752,
      "learning_rate": 0.0004802728598624615,
      "loss": 2.158,
      "step": 67924
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7836142182350159,
      "learning_rate": 0.0004802695901896492,
      "loss": 2.2487,
      "step": 67925
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.781608521938324,
      "learning_rate": 0.0004802663204833212,
      "loss": 2.1812,
      "step": 67926
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7918423414230347,
      "learning_rate": 0.00048026305074347835,
      "loss": 2.2073,
      "step": 67927
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7775049209594727,
      "learning_rate": 0.00048025978097012117,
      "loss": 2.3946,
      "step": 67928
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.9004958868026733,
      "learning_rate": 0.0004802565111632504,
      "loss": 2.2354,
      "step": 67929
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7241283655166626,
      "learning_rate": 0.0004802532413228664,
      "loss": 2.1148,
      "step": 67930
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7947759628295898,
      "learning_rate": 0.00048024997144896996,
      "loss": 2.2099,
      "step": 67931
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7874326705932617,
      "learning_rate": 0.0004802467015415617,
      "loss": 2.1196,
      "step": 67932
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.8412101864814758,
      "learning_rate": 0.00048024343160064215,
      "loss": 2.2349,
      "step": 67933
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7078837752342224,
      "learning_rate": 0.0004802401616262119,
      "loss": 2.047,
      "step": 67934
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7736688852310181,
      "learning_rate": 0.00048023689161827166,
      "loss": 2.0221,
      "step": 67935
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.836383044719696,
      "learning_rate": 0.0004802336215768219,
      "loss": 2.4416,
      "step": 67936
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.8578323125839233,
      "learning_rate": 0.00048023035150186337,
      "loss": 2.1136,
      "step": 67937
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.6760575175285339,
      "learning_rate": 0.0004802270813933967,
      "loss": 2.2219,
      "step": 67938
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.8156428337097168,
      "learning_rate": 0.00048022381125142227,
      "loss": 2.097,
      "step": 67939
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.9426660537719727,
      "learning_rate": 0.00048022054107594095,
      "loss": 2.0283,
      "step": 67940
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.8248448967933655,
      "learning_rate": 0.00048021727086695323,
      "loss": 2.0546,
      "step": 67941
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.882081925868988,
      "learning_rate": 0.0004802140006244596,
      "loss": 2.2541,
      "step": 67942
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.8692105412483215,
      "learning_rate": 0.0004802107303484608,
      "loss": 2.046,
      "step": 67943
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.8891956210136414,
      "learning_rate": 0.00048020746003895747,
      "loss": 2.2356,
      "step": 67944
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.762100338935852,
      "learning_rate": 0.00048020418969595016,
      "loss": 2.3346,
      "step": 67945
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.8720406293869019,
      "learning_rate": 0.0004802009193194395,
      "loss": 2.1494,
      "step": 67946
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.8087834119796753,
      "learning_rate": 0.00048019764890942616,
      "loss": 2.0125,
      "step": 67947
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.8073320984840393,
      "learning_rate": 0.00048019437846591055,
      "loss": 2.338,
      "step": 67948
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7584097981452942,
      "learning_rate": 0.0004801911079888934,
      "loss": 2.1838,
      "step": 67949
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.8733383417129517,
      "learning_rate": 0.0004801878374783754,
      "loss": 2.09,
      "step": 67950
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7767871022224426,
      "learning_rate": 0.000480184566934357,
      "loss": 2.1264,
      "step": 67951
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7936584949493408,
      "learning_rate": 0.00048018129635683896,
      "loss": 2.353,
      "step": 67952
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7642418742179871,
      "learning_rate": 0.0004801780257458218,
      "loss": 2.266,
      "step": 67953
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.8341968059539795,
      "learning_rate": 0.0004801747551013061,
      "loss": 2.1014,
      "step": 67954
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.8659088015556335,
      "learning_rate": 0.0004801714844232926,
      "loss": 2.1944,
      "step": 67955
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7564100623130798,
      "learning_rate": 0.0004801682137117818,
      "loss": 2.2063,
      "step": 67956
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.8169867992401123,
      "learning_rate": 0.00048016494296677426,
      "loss": 2.2892,
      "step": 67957
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.8595455884933472,
      "learning_rate": 0.0004801616721882707,
      "loss": 2.1466,
      "step": 67958
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.9016736745834351,
      "learning_rate": 0.00048015840137627164,
      "loss": 2.4851,
      "step": 67959
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7883842587471008,
      "learning_rate": 0.0004801551305307779,
      "loss": 2.086,
      "step": 67960
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.9055325388908386,
      "learning_rate": 0.00048015185965178964,
      "loss": 2.3848,
      "step": 67961
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7108038663864136,
      "learning_rate": 0.00048014858873930797,
      "loss": 2.1895,
      "step": 67962
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7114497423171997,
      "learning_rate": 0.00048014531779333317,
      "loss": 2.4342,
      "step": 67963
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.845681369304657,
      "learning_rate": 0.000480142046813866,
      "loss": 2.2304,
      "step": 67964
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.9078760147094727,
      "learning_rate": 0.0004801387758009071,
      "loss": 2.0448,
      "step": 67965
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7159550786018372,
      "learning_rate": 0.00048013550475445684,
      "loss": 2.2741,
      "step": 67966
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7379512786865234,
      "learning_rate": 0.0004801322336745161,
      "loss": 2.1145,
      "step": 67967
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.8617726564407349,
      "learning_rate": 0.00048012896256108533,
      "loss": 2.0984,
      "step": 67968
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.8177044987678528,
      "learning_rate": 0.0004801256914141652,
      "loss": 2.1767,
      "step": 67969
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.8375417590141296,
      "learning_rate": 0.0004801224202337563,
      "loss": 2.2353,
      "step": 67970
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.8138329982757568,
      "learning_rate": 0.0004801191490198592,
      "loss": 2.2476,
      "step": 67971
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7916852235794067,
      "learning_rate": 0.00048011587777247464,
      "loss": 2.3602,
      "step": 67972
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.8229463696479797,
      "learning_rate": 0.00048011260649160313,
      "loss": 2.0574,
      "step": 67973
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7685678601264954,
      "learning_rate": 0.00048010933517724516,
      "loss": 2.1109,
      "step": 67974
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.9073395133018494,
      "learning_rate": 0.0004801060638294016,
      "loss": 2.5207,
      "step": 67975
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.8250929117202759,
      "learning_rate": 0.00048010279244807287,
      "loss": 2.2417,
      "step": 67976
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.6858149170875549,
      "learning_rate": 0.0004800995210332596,
      "loss": 2.1628,
      "step": 67977
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7209813594818115,
      "learning_rate": 0.0004800962495849625,
      "loss": 2.0986,
      "step": 67978
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7343339323997498,
      "learning_rate": 0.0004800929781031821,
      "loss": 2.2678,
      "step": 67979
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7616012692451477,
      "learning_rate": 0.000480089706587919,
      "loss": 2.0077,
      "step": 67980
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.8255729079246521,
      "learning_rate": 0.00048008643503917386,
      "loss": 2.3302,
      "step": 67981
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.8633270859718323,
      "learning_rate": 0.00048008316345694716,
      "loss": 2.0659,
      "step": 67982
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.8997029662132263,
      "learning_rate": 0.00048007989184123965,
      "loss": 2.2592,
      "step": 67983
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.8286288380622864,
      "learning_rate": 0.0004800766201920519,
      "loss": 2.1468,
      "step": 67984
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.9707013368606567,
      "learning_rate": 0.00048007334850938446,
      "loss": 2.0619,
      "step": 67985
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.9156398773193359,
      "learning_rate": 0.00048007007679323803,
      "loss": 2.2608,
      "step": 67986
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.0089991092681885,
      "learning_rate": 0.0004800668050436132,
      "loss": 2.3537,
      "step": 67987
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.8275713324546814,
      "learning_rate": 0.0004800635332605105,
      "loss": 2.1674,
      "step": 67988
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.8743302226066589,
      "learning_rate": 0.0004800602614439306,
      "loss": 2.1,
      "step": 67989
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.2229863405227661,
      "learning_rate": 0.00048005698959387416,
      "loss": 2.0307,
      "step": 67990
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7942125797271729,
      "learning_rate": 0.0004800537177103416,
      "loss": 2.0967,
      "step": 67991
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7083824276924133,
      "learning_rate": 0.00048005044579333377,
      "loss": 2.2019,
      "step": 67992
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.0065441131591797,
      "learning_rate": 0.0004800471738428511,
      "loss": 2.0637,
      "step": 67993
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7921179533004761,
      "learning_rate": 0.00048004390185889424,
      "loss": 2.1338,
      "step": 67994
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7743226289749146,
      "learning_rate": 0.0004800406298414639,
      "loss": 2.4051,
      "step": 67995
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7823184132575989,
      "learning_rate": 0.00048003735779056063,
      "loss": 2.4069,
      "step": 67996
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.0657435655593872,
      "learning_rate": 0.0004800340857061849,
      "loss": 2.0776,
      "step": 67997
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7647972106933594,
      "learning_rate": 0.0004800308135883375,
      "loss": 2.3749,
      "step": 67998
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7945569753646851,
      "learning_rate": 0.000480027541437019,
      "loss": 2.2364,
      "step": 67999
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7408044338226318,
      "learning_rate": 0.0004800242692522299,
      "loss": 2.0429,
      "step": 68000
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.8014633059501648,
      "learning_rate": 0.00048002099703397096,
      "loss": 2.0545,
      "step": 68001
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7742155194282532,
      "learning_rate": 0.0004800177247822427,
      "loss": 2.606,
      "step": 68002
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7907354831695557,
      "learning_rate": 0.0004800144524970457,
      "loss": 2.6005,
      "step": 68003
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7596675753593445,
      "learning_rate": 0.00048001118017838065,
      "loss": 2.2883,
      "step": 68004
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.710175633430481,
      "learning_rate": 0.0004800079078262482,
      "loss": 2.1167,
      "step": 68005
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7395500540733337,
      "learning_rate": 0.0004800046354406488,
      "loss": 2.4625,
      "step": 68006
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7656994462013245,
      "learning_rate": 0.00048000136302158315,
      "loss": 2.3483,
      "step": 68007
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7624390721321106,
      "learning_rate": 0.0004799980905690518,
      "loss": 2.1005,
      "step": 68008
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7308460474014282,
      "learning_rate": 0.00047999481808305553,
      "loss": 1.9373,
      "step": 68009
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.8119055032730103,
      "learning_rate": 0.00047999154556359477,
      "loss": 2.2296,
      "step": 68010
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7950518131256104,
      "learning_rate": 0.0004799882730106701,
      "loss": 2.1483,
      "step": 68011
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.8360522985458374,
      "learning_rate": 0.0004799850004242823,
      "loss": 2.2573,
      "step": 68012
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.9849295616149902,
      "learning_rate": 0.00047998172780443194,
      "loss": 2.2997,
      "step": 68013
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.184398889541626,
      "learning_rate": 0.0004799784551511194,
      "loss": 2.2333,
      "step": 68014
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.0445530414581299,
      "learning_rate": 0.0004799751824643456,
      "loss": 2.2236,
      "step": 68015
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.8258242011070251,
      "learning_rate": 0.00047997190974411103,
      "loss": 2.2091,
      "step": 68016
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7585628032684326,
      "learning_rate": 0.00047996863699041625,
      "loss": 2.0499,
      "step": 68017
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7234458327293396,
      "learning_rate": 0.0004799653642032619,
      "loss": 2.2047,
      "step": 68018
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.8496501445770264,
      "learning_rate": 0.0004799620913826486,
      "loss": 2.349,
      "step": 68019
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.8729774951934814,
      "learning_rate": 0.00047995881852857695,
      "loss": 2.1797,
      "step": 68020
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7246124148368835,
      "learning_rate": 0.0004799555456410475,
      "loss": 2.4371,
      "step": 68021
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.9430050253868103,
      "learning_rate": 0.000479952272720061,
      "loss": 2.438,
      "step": 68022
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.8888996243476868,
      "learning_rate": 0.00047994899976561796,
      "loss": 1.992,
      "step": 68023
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7376379370689392,
      "learning_rate": 0.00047994572677771897,
      "loss": 2.3324,
      "step": 68024
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.8078404664993286,
      "learning_rate": 0.0004799424537563647,
      "loss": 2.2572,
      "step": 68025
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.828160285949707,
      "learning_rate": 0.0004799391807015558,
      "loss": 2.0226,
      "step": 68026
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.8665119409561157,
      "learning_rate": 0.0004799359076132927,
      "loss": 2.2772,
      "step": 68027
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.8913074135780334,
      "learning_rate": 0.00047993263449157617,
      "loss": 2.112,
      "step": 68028
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.9335451722145081,
      "learning_rate": 0.00047992936133640674,
      "loss": 2.3231,
      "step": 68029
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.1304627656936646,
      "learning_rate": 0.00047992608814778506,
      "loss": 2.1617,
      "step": 68030
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7468105554580688,
      "learning_rate": 0.00047992281492571174,
      "loss": 2.0159,
      "step": 68031
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.901121199131012,
      "learning_rate": 0.0004799195416701873,
      "loss": 2.0689,
      "step": 68032
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.8960300087928772,
      "learning_rate": 0.0004799162683812125,
      "loss": 2.3352,
      "step": 68033
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.8054041266441345,
      "learning_rate": 0.00047991299505878784,
      "loss": 2.1337,
      "step": 68034
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7098730206489563,
      "learning_rate": 0.0004799097217029139,
      "loss": 2.096,
      "step": 68035
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.8203999996185303,
      "learning_rate": 0.0004799064483135915,
      "loss": 2.3019,
      "step": 68036
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.2065261602401733,
      "learning_rate": 0.0004799031748908209,
      "loss": 2.1669,
      "step": 68037
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7785791158676147,
      "learning_rate": 0.000479899901434603,
      "loss": 2.1507,
      "step": 68038
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.960319995880127,
      "learning_rate": 0.0004798966279449384,
      "loss": 2.1997,
      "step": 68039
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.9794970750808716,
      "learning_rate": 0.0004798933544218275,
      "loss": 2.2103,
      "step": 68040
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.723346471786499,
      "learning_rate": 0.0004798900808652711,
      "loss": 2.1015,
      "step": 68041
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.8387688994407654,
      "learning_rate": 0.00047988680727526975,
      "loss": 2.1609,
      "step": 68042
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.9052370190620422,
      "learning_rate": 0.00047988353365182397,
      "loss": 2.1281,
      "step": 68043
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7999789118766785,
      "learning_rate": 0.00047988025999493446,
      "loss": 2.2669,
      "step": 68044
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.8586452007293701,
      "learning_rate": 0.0004798769863046018,
      "loss": 2.1627,
      "step": 68045
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.9654214978218079,
      "learning_rate": 0.0004798737125808267,
      "loss": 2.2392,
      "step": 68046
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7110809087753296,
      "learning_rate": 0.0004798704388236096,
      "loss": 1.9886,
      "step": 68047
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.8220176696777344,
      "learning_rate": 0.0004798671650329512,
      "loss": 1.8593,
      "step": 68048
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7569752931594849,
      "learning_rate": 0.00047986389120885213,
      "loss": 2.2899,
      "step": 68049
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.76492840051651,
      "learning_rate": 0.000479860617351313,
      "loss": 2.3504,
      "step": 68050
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.757014274597168,
      "learning_rate": 0.00047985734346033427,
      "loss": 2.2085,
      "step": 68051
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.852775514125824,
      "learning_rate": 0.0004798540695359168,
      "loss": 2.262,
      "step": 68052
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.8052689433097839,
      "learning_rate": 0.000479850795578061,
      "loss": 2.3162,
      "step": 68053
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.8529755473136902,
      "learning_rate": 0.0004798475215867676,
      "loss": 2.1358,
      "step": 68054
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.8009071350097656,
      "learning_rate": 0.00047984424756203705,
      "loss": 2.299,
      "step": 68055
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.8182416558265686,
      "learning_rate": 0.00047984097350387006,
      "loss": 2.0475,
      "step": 68056
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.763428270816803,
      "learning_rate": 0.0004798376994122673,
      "loss": 2.1849,
      "step": 68057
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.73328697681427,
      "learning_rate": 0.00047983442528722926,
      "loss": 2.1972,
      "step": 68058
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.9960442781448364,
      "learning_rate": 0.00047983115112875665,
      "loss": 2.2922,
      "step": 68059
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7461333870887756,
      "learning_rate": 0.00047982787693685007,
      "loss": 2.3121,
      "step": 68060
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.8307914137840271,
      "learning_rate": 0.00047982460271151,
      "loss": 2.1596,
      "step": 68061
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.9795413613319397,
      "learning_rate": 0.00047982132845273713,
      "loss": 2.2463,
      "step": 68062
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7443976402282715,
      "learning_rate": 0.00047981805416053225,
      "loss": 2.2788,
      "step": 68063
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7454213500022888,
      "learning_rate": 0.00047981477983489567,
      "loss": 2.1943,
      "step": 68064
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7547105550765991,
      "learning_rate": 0.00047981150547582815,
      "loss": 2.0782,
      "step": 68065
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.8620654940605164,
      "learning_rate": 0.0004798082310833303,
      "loss": 2.1879,
      "step": 68066
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7656890749931335,
      "learning_rate": 0.00047980495665740267,
      "loss": 2.2907,
      "step": 68067
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7517374753952026,
      "learning_rate": 0.0004798016821980459,
      "loss": 2.3739,
      "step": 68068
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.8523142337799072,
      "learning_rate": 0.0004797984077052606,
      "loss": 1.8767,
      "step": 68069
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7084076404571533,
      "learning_rate": 0.0004797951331790474,
      "loss": 2.3104,
      "step": 68070
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7580739259719849,
      "learning_rate": 0.00047979185861940694,
      "loss": 2.4435,
      "step": 68071
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.8186161518096924,
      "learning_rate": 0.00047978858402633975,
      "loss": 2.205,
      "step": 68072
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.8015502691268921,
      "learning_rate": 0.00047978530939984644,
      "loss": 2.1784,
      "step": 68073
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.8890773057937622,
      "learning_rate": 0.0004797820347399276,
      "loss": 2.152,
      "step": 68074
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.9699700474739075,
      "learning_rate": 0.000479778760046584,
      "loss": 2.1298,
      "step": 68075
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7837303280830383,
      "learning_rate": 0.00047977548531981607,
      "loss": 2.1335,
      "step": 68076
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.2353076934814453,
      "learning_rate": 0.0004797722105596245,
      "loss": 2.2138,
      "step": 68077
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.9342976212501526,
      "learning_rate": 0.00047976893576600986,
      "loss": 2.2197,
      "step": 68078
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.801361083984375,
      "learning_rate": 0.0004797656609389728,
      "loss": 2.1862,
      "step": 68079
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.8986285924911499,
      "learning_rate": 0.0004797623860785139,
      "loss": 2.3216,
      "step": 68080
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.8579932451248169,
      "learning_rate": 0.00047975911118463383,
      "loss": 2.2672,
      "step": 68081
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7353119850158691,
      "learning_rate": 0.0004797558362573331,
      "loss": 2.3288,
      "step": 68082
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.8411665558815002,
      "learning_rate": 0.00047975256129661234,
      "loss": 2.0311,
      "step": 68083
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7841562628746033,
      "learning_rate": 0.00047974928630247225,
      "loss": 2.0084,
      "step": 68084
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7478211522102356,
      "learning_rate": 0.0004797460112749133,
      "loss": 2.0119,
      "step": 68085
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.8085925579071045,
      "learning_rate": 0.00047974273621393624,
      "loss": 2.2551,
      "step": 68086
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7495442628860474,
      "learning_rate": 0.0004797394611195416,
      "loss": 2.1378,
      "step": 68087
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.75459223985672,
      "learning_rate": 0.00047973618599173,
      "loss": 2.1687,
      "step": 68088
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7788872122764587,
      "learning_rate": 0.000479732910830502,
      "loss": 2.0986,
      "step": 68089
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.729817807674408,
      "learning_rate": 0.00047972963563585827,
      "loss": 2.3271,
      "step": 68090
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7722369432449341,
      "learning_rate": 0.0004797263604077995,
      "loss": 2.1197,
      "step": 68091
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7851327657699585,
      "learning_rate": 0.0004797230851463262,
      "loss": 1.998,
      "step": 68092
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7778847217559814,
      "learning_rate": 0.00047971980985143887,
      "loss": 2.154,
      "step": 68093
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7858610153198242,
      "learning_rate": 0.00047971653452313825,
      "loss": 2.3412,
      "step": 68094
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.8040043711662292,
      "learning_rate": 0.000479713259161425,
      "loss": 2.284,
      "step": 68095
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.8092272877693176,
      "learning_rate": 0.00047970998376629963,
      "loss": 2.1818,
      "step": 68096
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7543202042579651,
      "learning_rate": 0.0004797067083377628,
      "loss": 2.1791,
      "step": 68097
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7564382553100586,
      "learning_rate": 0.0004797034328758151,
      "loss": 2.1603,
      "step": 68098
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.756923496723175,
      "learning_rate": 0.00047970015738045714,
      "loss": 2.2919,
      "step": 68099
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7048698663711548,
      "learning_rate": 0.00047969688185168954,
      "loss": 1.9898,
      "step": 68100
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.1150156259536743,
      "learning_rate": 0.00047969360628951285,
      "loss": 2.2067,
      "step": 68101
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.8690218329429626,
      "learning_rate": 0.0004796903306939278,
      "loss": 2.0963,
      "step": 68102
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.8010212779045105,
      "learning_rate": 0.0004796870550649349,
      "loss": 2.0849,
      "step": 68103
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.1718209981918335,
      "learning_rate": 0.0004796837794025347,
      "loss": 2.1825,
      "step": 68104
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.79515540599823,
      "learning_rate": 0.00047968050370672793,
      "loss": 2.274,
      "step": 68105
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.9689135551452637,
      "learning_rate": 0.00047967722797751525,
      "loss": 2.1005,
      "step": 68106
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.2665164470672607,
      "learning_rate": 0.0004796739522148971,
      "loss": 2.3989,
      "step": 68107
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.8144617080688477,
      "learning_rate": 0.0004796706764188742,
      "loss": 2.4321,
      "step": 68108
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.8178573250770569,
      "learning_rate": 0.00047966740058944717,
      "loss": 2.4284,
      "step": 68109
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.9406676292419434,
      "learning_rate": 0.00047966412472661655,
      "loss": 2.1304,
      "step": 68110
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.8391678333282471,
      "learning_rate": 0.00047966084883038296,
      "loss": 2.2088,
      "step": 68111
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7267930507659912,
      "learning_rate": 0.000479657572900747,
      "loss": 2.2293,
      "step": 68112
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.77825927734375,
      "learning_rate": 0.0004796542969377093,
      "loss": 2.3349,
      "step": 68113
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7949263453483582,
      "learning_rate": 0.0004796510209412706,
      "loss": 2.3091,
      "step": 68114
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7527937293052673,
      "learning_rate": 0.0004796477449114313,
      "loss": 2.0866,
      "step": 68115
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.8303450345993042,
      "learning_rate": 0.00047964446884819203,
      "loss": 2.3938,
      "step": 68116
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7025100588798523,
      "learning_rate": 0.0004796411927515535,
      "loss": 2.404,
      "step": 68117
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.8302112221717834,
      "learning_rate": 0.0004796379166215163,
      "loss": 2.2625,
      "step": 68118
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7673095464706421,
      "learning_rate": 0.00047963464045808104,
      "loss": 2.2247,
      "step": 68119
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.8235254287719727,
      "learning_rate": 0.0004796313642612483,
      "loss": 2.3388,
      "step": 68120
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7811226844787598,
      "learning_rate": 0.0004796280880310187,
      "loss": 1.9559,
      "step": 68121
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7629153728485107,
      "learning_rate": 0.00047962481176739284,
      "loss": 2.2517,
      "step": 68122
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7055322527885437,
      "learning_rate": 0.0004796215354703713,
      "loss": 2.0786,
      "step": 68123
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7543527483940125,
      "learning_rate": 0.0004796182591399548,
      "loss": 2.2601,
      "step": 68124
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.8032504916191101,
      "learning_rate": 0.0004796149827761438,
      "loss": 2.2495,
      "step": 68125
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7901787757873535,
      "learning_rate": 0.000479611706378939,
      "loss": 2.1609,
      "step": 68126
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.6928238868713379,
      "learning_rate": 0.000479608429948341,
      "loss": 2.0774,
      "step": 68127
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7303774952888489,
      "learning_rate": 0.0004796051534843505,
      "loss": 2.4404,
      "step": 68128
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7576904892921448,
      "learning_rate": 0.0004796018769869679,
      "loss": 2.1594,
      "step": 68129
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7694898843765259,
      "learning_rate": 0.00047959860045619396,
      "loss": 2.2783,
      "step": 68130
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.777107834815979,
      "learning_rate": 0.00047959532389202917,
      "loss": 1.9933,
      "step": 68131
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7212088704109192,
      "learning_rate": 0.0004795920472944743,
      "loss": 2.0522,
      "step": 68132
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7017143964767456,
      "learning_rate": 0.0004795887706635299,
      "loss": 2.3124,
      "step": 68133
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.8243118524551392,
      "learning_rate": 0.00047958549399919657,
      "loss": 2.1512,
      "step": 68134
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7268903851509094,
      "learning_rate": 0.0004795822173014747,
      "loss": 2.1833,
      "step": 68135
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.962689220905304,
      "learning_rate": 0.00047957894057036534,
      "loss": 2.3626,
      "step": 68136
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7155120968818665,
      "learning_rate": 0.0004795756638058688,
      "loss": 2.2688,
      "step": 68137
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.9112232327461243,
      "learning_rate": 0.00047957238700798577,
      "loss": 2.1354,
      "step": 68138
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7857849597930908,
      "learning_rate": 0.00047956911017671676,
      "loss": 2.2473,
      "step": 68139
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7447055578231812,
      "learning_rate": 0.0004795658333120626,
      "loss": 2.1447,
      "step": 68140
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.8632941842079163,
      "learning_rate": 0.00047956255641402363,
      "loss": 2.2483,
      "step": 68141
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.8465297818183899,
      "learning_rate": 0.0004795592794826007,
      "loss": 2.2849,
      "step": 68142
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7947179079055786,
      "learning_rate": 0.0004795560025177942,
      "loss": 2.0677,
      "step": 68143
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.8131875395774841,
      "learning_rate": 0.0004795527255196049,
      "loss": 2.0843,
      "step": 68144
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.8024550080299377,
      "learning_rate": 0.00047954944848803334,
      "loss": 2.2244,
      "step": 68145
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.828750491142273,
      "learning_rate": 0.00047954617142308023,
      "loss": 2.482,
      "step": 68146
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7780129909515381,
      "learning_rate": 0.000479542894324746,
      "loss": 2.1286,
      "step": 68147
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.8122219443321228,
      "learning_rate": 0.0004795396171930314,
      "loss": 2.287,
      "step": 68148
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.2143739461898804,
      "learning_rate": 0.00047953634002793704,
      "loss": 2.0645,
      "step": 68149
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.8596910238265991,
      "learning_rate": 0.00047953306282946334,
      "loss": 2.2883,
      "step": 68150
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7827427983283997,
      "learning_rate": 0.0004795297855976112,
      "loss": 2.0244,
      "step": 68151
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.9264509081840515,
      "learning_rate": 0.00047952650833238105,
      "loss": 2.4953,
      "step": 68152
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.0568065643310547,
      "learning_rate": 0.00047952323103377353,
      "loss": 2.0046,
      "step": 68153
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7665619254112244,
      "learning_rate": 0.00047951995370178923,
      "loss": 2.2008,
      "step": 68154
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.8708148002624512,
      "learning_rate": 0.00047951667633642874,
      "loss": 2.2162,
      "step": 68155
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.110504150390625,
      "learning_rate": 0.0004795133989376928,
      "loss": 2.0562,
      "step": 68156
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7320742011070251,
      "learning_rate": 0.00047951012150558196,
      "loss": 2.185,
      "step": 68157
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7264577150344849,
      "learning_rate": 0.0004795068440400967,
      "loss": 2.356,
      "step": 68158
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.8385607004165649,
      "learning_rate": 0.0004795035665412377,
      "loss": 2.1938,
      "step": 68159
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.8614726066589355,
      "learning_rate": 0.0004795002890090057,
      "loss": 2.315,
      "step": 68160
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7150343060493469,
      "learning_rate": 0.0004794970114434012,
      "loss": 2.2838,
      "step": 68161
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7679562568664551,
      "learning_rate": 0.00047949373384442473,
      "loss": 2.3235,
      "step": 68162
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.9018326997756958,
      "learning_rate": 0.000479490456212077,
      "loss": 2.3001,
      "step": 68163
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7423735857009888,
      "learning_rate": 0.0004794871785463587,
      "loss": 2.0323,
      "step": 68164
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7438433170318604,
      "learning_rate": 0.0004794839008472703,
      "loss": 2.1116,
      "step": 68165
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.0271244049072266,
      "learning_rate": 0.00047948062311481247,
      "loss": 2.288,
      "step": 68166
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.8975719809532166,
      "learning_rate": 0.0004794773453489858,
      "loss": 2.3645,
      "step": 68167
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7690054178237915,
      "learning_rate": 0.0004794740675497908,
      "loss": 2.214,
      "step": 68168
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.9394071698188782,
      "learning_rate": 0.0004794707897172283,
      "loss": 2.0702,
      "step": 68169
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.8392083048820496,
      "learning_rate": 0.00047946751185129884,
      "loss": 2.116,
      "step": 68170
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.8397069573402405,
      "learning_rate": 0.00047946423395200293,
      "loss": 1.9882,
      "step": 68171
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.8387424349784851,
      "learning_rate": 0.00047946095601934113,
      "loss": 2.0901,
      "step": 68172
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.9016107320785522,
      "learning_rate": 0.0004794576780533143,
      "loss": 2.275,
      "step": 68173
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.8291035890579224,
      "learning_rate": 0.0004794544000539228,
      "loss": 2.1222,
      "step": 68174
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.8369641900062561,
      "learning_rate": 0.0004794511220211673,
      "loss": 2.1988,
      "step": 68175
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.8414483666419983,
      "learning_rate": 0.0004794478439550486,
      "loss": 2.2087,
      "step": 68176
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.921960711479187,
      "learning_rate": 0.00047944456585556703,
      "loss": 2.4483,
      "step": 68177
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7530828714370728,
      "learning_rate": 0.00047944128772272335,
      "loss": 2.239,
      "step": 68178
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.8389407992362976,
      "learning_rate": 0.0004794380095565183,
      "loss": 2.2337,
      "step": 68179
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7912919521331787,
      "learning_rate": 0.0004794347313569521,
      "loss": 2.2757,
      "step": 68180
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.8102344870567322,
      "learning_rate": 0.00047943145312402577,
      "loss": 2.3922,
      "step": 68181
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7879418730735779,
      "learning_rate": 0.00047942817485773957,
      "loss": 2.1919,
      "step": 68182
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7019481658935547,
      "learning_rate": 0.0004794248965580945,
      "loss": 2.3155,
      "step": 68183
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.8209248185157776,
      "learning_rate": 0.00047942161822509075,
      "loss": 2.272,
      "step": 68184
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7220303416252136,
      "learning_rate": 0.00047941833985872933,
      "loss": 2.2522,
      "step": 68185
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.8573166131973267,
      "learning_rate": 0.0004794150614590106,
      "loss": 1.966,
      "step": 68186
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7615233063697815,
      "learning_rate": 0.00047941178302593506,
      "loss": 2.153,
      "step": 68187
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.8140076398849487,
      "learning_rate": 0.0004794085045595036,
      "loss": 2.242,
      "step": 68188
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7782556414604187,
      "learning_rate": 0.00047940522605971676,
      "loss": 2.0824,
      "step": 68189
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7898566722869873,
      "learning_rate": 0.00047940194752657506,
      "loss": 2.1385,
      "step": 68190
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.8046284914016724,
      "learning_rate": 0.0004793986689600792,
      "loss": 2.3224,
      "step": 68191
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.0473917722702026,
      "learning_rate": 0.00047939539036022963,
      "loss": 1.9974,
      "step": 68192
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7817632555961609,
      "learning_rate": 0.0004793921117270271,
      "loss": 2.1357,
      "step": 68193
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7101667523384094,
      "learning_rate": 0.0004793888330604723,
      "loss": 2.0699,
      "step": 68194
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.8376296162605286,
      "learning_rate": 0.0004793855543605656,
      "loss": 2.1111,
      "step": 68195
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.8155761957168579,
      "learning_rate": 0.0004793822756273078,
      "loss": 2.2411,
      "step": 68196
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7595974206924438,
      "learning_rate": 0.0004793789968606995,
      "loss": 2.4879,
      "step": 68197
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.687628448009491,
      "learning_rate": 0.00047937571806074117,
      "loss": 2.1958,
      "step": 68198
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.8721043467521667,
      "learning_rate": 0.0004793724392274335,
      "loss": 2.1074,
      "step": 68199
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7984763979911804,
      "learning_rate": 0.00047936916036077717,
      "loss": 1.9257,
      "step": 68200
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7380422949790955,
      "learning_rate": 0.00047936588146077274,
      "loss": 2.2495,
      "step": 68201
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.834781289100647,
      "learning_rate": 0.0004793626025274207,
      "loss": 2.2794,
      "step": 68202
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.8383548259735107,
      "learning_rate": 0.00047935932356072184,
      "loss": 2.1809,
      "step": 68203
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7676329612731934,
      "learning_rate": 0.0004793560445606767,
      "loss": 2.1569,
      "step": 68204
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7064197659492493,
      "learning_rate": 0.0004793527655272858,
      "loss": 2.1963,
      "step": 68205
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.8548043370246887,
      "learning_rate": 0.00047934948646054996,
      "loss": 2.3543,
      "step": 68206
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.9824483394622803,
      "learning_rate": 0.0004793462073604697,
      "loss": 2.1871,
      "step": 68207
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.8673895597457886,
      "learning_rate": 0.0004793429282270454,
      "loss": 2.0396,
      "step": 68208
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.9085458517074585,
      "learning_rate": 0.00047933964906027803,
      "loss": 2.3059,
      "step": 68209
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7744494080543518,
      "learning_rate": 0.000479336369860168,
      "loss": 2.4599,
      "step": 68210
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.8493661284446716,
      "learning_rate": 0.0004793330906267159,
      "loss": 2.2708,
      "step": 68211
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.720750093460083,
      "learning_rate": 0.0004793298113599224,
      "loss": 2.1689,
      "step": 68212
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7771106958389282,
      "learning_rate": 0.0004793265320597882,
      "loss": 2.1214,
      "step": 68213
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.8389657735824585,
      "learning_rate": 0.0004793232527263137,
      "loss": 2.0367,
      "step": 68214
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7840099930763245,
      "learning_rate": 0.00047931997335949974,
      "loss": 2.3143,
      "step": 68215
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7003839612007141,
      "learning_rate": 0.00047931669395934667,
      "loss": 2.1544,
      "step": 68216
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.827759325504303,
      "learning_rate": 0.00047931341452585525,
      "loss": 2.1724,
      "step": 68217
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.9173179268836975,
      "learning_rate": 0.00047931013505902624,
      "loss": 2.1283,
      "step": 68218
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7854644656181335,
      "learning_rate": 0.00047930685555886,
      "loss": 2.1048,
      "step": 68219
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.8748196959495544,
      "learning_rate": 0.0004793035760253572,
      "loss": 2.4918,
      "step": 68220
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.758162796497345,
      "learning_rate": 0.0004793002964585185,
      "loss": 2.0655,
      "step": 68221
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7543267607688904,
      "learning_rate": 0.0004792970168583445,
      "loss": 2.281,
      "step": 68222
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7694367170333862,
      "learning_rate": 0.00047929373722483575,
      "loss": 2.1764,
      "step": 68223
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7505956292152405,
      "learning_rate": 0.00047929045755799296,
      "loss": 2.2084,
      "step": 68224
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.9160436987876892,
      "learning_rate": 0.0004792871778578167,
      "loss": 2.3022,
      "step": 68225
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.728258490562439,
      "learning_rate": 0.00047928389812430756,
      "loss": 2.3,
      "step": 68226
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7552348971366882,
      "learning_rate": 0.0004792806183574661,
      "loss": 2.2616,
      "step": 68227
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.8062137961387634,
      "learning_rate": 0.00047927733855729305,
      "loss": 2.2292,
      "step": 68228
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.809664249420166,
      "learning_rate": 0.0004792740587237889,
      "loss": 2.2207,
      "step": 68229
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.937873899936676,
      "learning_rate": 0.0004792707788569544,
      "loss": 2.066,
      "step": 68230
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7046145796775818,
      "learning_rate": 0.0004792674989567901,
      "loss": 2.1571,
      "step": 68231
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.769512951374054,
      "learning_rate": 0.00047926421902329654,
      "loss": 1.9774,
      "step": 68232
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.8716320395469666,
      "learning_rate": 0.0004792609390564743,
      "loss": 2.2961,
      "step": 68233
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.8380244970321655,
      "learning_rate": 0.0004792576590563242,
      "loss": 2.3778,
      "step": 68234
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.716014564037323,
      "learning_rate": 0.0004792543790228466,
      "loss": 2.2067,
      "step": 68235
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.8273143768310547,
      "learning_rate": 0.0004792510989560423,
      "loss": 2.1909,
      "step": 68236
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.6800544261932373,
      "learning_rate": 0.00047924781885591187,
      "loss": 2.3663,
      "step": 68237
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7566697597503662,
      "learning_rate": 0.0004792445387224558,
      "loss": 2.1797,
      "step": 68238
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7280482649803162,
      "learning_rate": 0.0004792412585556748,
      "loss": 2.048,
      "step": 68239
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7727657556533813,
      "learning_rate": 0.00047923797835556955,
      "loss": 2.1845,
      "step": 68240
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.8476694822311401,
      "learning_rate": 0.00047923469812214043,
      "loss": 2.1938,
      "step": 68241
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7769355773925781,
      "learning_rate": 0.0004792314178553883,
      "loss": 2.0484,
      "step": 68242
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.816374659538269,
      "learning_rate": 0.00047922813755531366,
      "loss": 2.2681,
      "step": 68243
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.9172568917274475,
      "learning_rate": 0.00047922485722191714,
      "loss": 2.113,
      "step": 68244
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7866869568824768,
      "learning_rate": 0.0004792215768551993,
      "loss": 2.288,
      "step": 68245
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7231903076171875,
      "learning_rate": 0.0004792182964551608,
      "loss": 2.2994,
      "step": 68246
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.0025780200958252,
      "learning_rate": 0.00047921501602180217,
      "loss": 2.3846,
      "step": 68247
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.8012787699699402,
      "learning_rate": 0.00047921173555512414,
      "loss": 2.3262,
      "step": 68248
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7693164944648743,
      "learning_rate": 0.0004792084550551273,
      "loss": 2.4094,
      "step": 68249
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.0601496696472168,
      "learning_rate": 0.0004792051745218122,
      "loss": 2.2951,
      "step": 68250
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.8346431851387024,
      "learning_rate": 0.0004792018939551795,
      "loss": 2.2623,
      "step": 68251
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7475645542144775,
      "learning_rate": 0.00047919861335522974,
      "loss": 2.243,
      "step": 68252
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.6563404202461243,
      "learning_rate": 0.0004791953327219636,
      "loss": 2.2512,
      "step": 68253
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.836513102054596,
      "learning_rate": 0.00047919205205538163,
      "loss": 2.1458,
      "step": 68254
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7556683421134949,
      "learning_rate": 0.00047918877135548455,
      "loss": 2.4374,
      "step": 68255
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.8918792605400085,
      "learning_rate": 0.00047918549062227287,
      "loss": 2.2455,
      "step": 68256
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7336540222167969,
      "learning_rate": 0.0004791822098557471,
      "loss": 2.1589,
      "step": 68257
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.6927137970924377,
      "learning_rate": 0.0004791789290559082,
      "loss": 2.2326,
      "step": 68258
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.813521683216095,
      "learning_rate": 0.00047917564822275643,
      "loss": 2.2021,
      "step": 68259
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7675907611846924,
      "learning_rate": 0.00047917236735629245,
      "loss": 2.2206,
      "step": 68260
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7782913446426392,
      "learning_rate": 0.00047916908645651707,
      "loss": 2.2214,
      "step": 68261
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.8754429817199707,
      "learning_rate": 0.0004791658055234307,
      "loss": 2.3711,
      "step": 68262
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7675700783729553,
      "learning_rate": 0.00047916252455703403,
      "loss": 2.2903,
      "step": 68263
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7808210849761963,
      "learning_rate": 0.0004791592435573278,
      "loss": 2.2193,
      "step": 68264
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.8650221824645996,
      "learning_rate": 0.0004791559625243123,
      "loss": 2.2984,
      "step": 68265
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.8598151206970215,
      "learning_rate": 0.0004791526814579884,
      "loss": 2.1596,
      "step": 68266
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7444985508918762,
      "learning_rate": 0.00047914940035835666,
      "loss": 2.2351,
      "step": 68267
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7167554497718811,
      "learning_rate": 0.00047914611922541763,
      "loss": 2.2313,
      "step": 68268
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.8579375743865967,
      "learning_rate": 0.000479142838059172,
      "loss": 2.3581,
      "step": 68269
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.8581370115280151,
      "learning_rate": 0.0004791395568596202,
      "loss": 2.3876,
      "step": 68270
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7946123480796814,
      "learning_rate": 0.00047913627562676313,
      "loss": 2.2876,
      "step": 68271
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7724596261978149,
      "learning_rate": 0.00047913299436060113,
      "loss": 2.3859,
      "step": 68272
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.8271191716194153,
      "learning_rate": 0.000479129713061135,
      "loss": 2.1515,
      "step": 68273
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7038756608963013,
      "learning_rate": 0.0004791264317283653,
      "loss": 2.4293,
      "step": 68274
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7415677309036255,
      "learning_rate": 0.0004791231503622926,
      "loss": 2.3373,
      "step": 68275
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.8079079389572144,
      "learning_rate": 0.00047911986896291743,
      "loss": 2.2285,
      "step": 68276
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.9537838101387024,
      "learning_rate": 0.0004791165875302406,
      "loss": 2.2575,
      "step": 68277
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.6909745335578918,
      "learning_rate": 0.00047911330606426256,
      "loss": 2.193,
      "step": 68278
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.800044596195221,
      "learning_rate": 0.000479110024564984,
      "loss": 2.2396,
      "step": 68279
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.774782121181488,
      "learning_rate": 0.00047910674303240544,
      "loss": 2.2235,
      "step": 68280
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7766528725624084,
      "learning_rate": 0.0004791034614665276,
      "loss": 2.2308,
      "step": 68281
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7372073531150818,
      "learning_rate": 0.0004791001798673511,
      "loss": 2.1423,
      "step": 68282
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.8240416049957275,
      "learning_rate": 0.0004790968982348765,
      "loss": 2.2469,
      "step": 68283
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7055122256278992,
      "learning_rate": 0.00047909361656910435,
      "loss": 2.2814,
      "step": 68284
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7359199523925781,
      "learning_rate": 0.00047909033487003534,
      "loss": 2.0911,
      "step": 68285
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.9141887426376343,
      "learning_rate": 0.00047908705313767,
      "loss": 2.2689,
      "step": 68286
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.881598949432373,
      "learning_rate": 0.000479083771372009,
      "loss": 2.3004,
      "step": 68287
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7462626695632935,
      "learning_rate": 0.00047908048957305305,
      "loss": 2.0126,
      "step": 68288
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.8050752282142639,
      "learning_rate": 0.00047907720774080254,
      "loss": 2.2799,
      "step": 68289
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7994528412818909,
      "learning_rate": 0.00047907392587525824,
      "loss": 2.2502,
      "step": 68290
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7696278095245361,
      "learning_rate": 0.00047907064397642075,
      "loss": 2.3504,
      "step": 68291
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7195547223091125,
      "learning_rate": 0.00047906736204429066,
      "loss": 2.505,
      "step": 68292
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.824988842010498,
      "learning_rate": 0.00047906408007886847,
      "loss": 2.2946,
      "step": 68293
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.8247288465499878,
      "learning_rate": 0.000479060798080155,
      "loss": 2.3469,
      "step": 68294
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7662463188171387,
      "learning_rate": 0.0004790575160481507,
      "loss": 2.321,
      "step": 68295
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.70052170753479,
      "learning_rate": 0.00047905423398285623,
      "loss": 2.0326,
      "step": 68296
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7469080090522766,
      "learning_rate": 0.0004790509518842721,
      "loss": 2.0827,
      "step": 68297
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7477255463600159,
      "learning_rate": 0.0004790476697523991,
      "loss": 2.2736,
      "step": 68298
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7171066403388977,
      "learning_rate": 0.00047904438758723783,
      "loss": 2.3201,
      "step": 68299
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.78398597240448,
      "learning_rate": 0.0004790411053887887,
      "loss": 2.2126,
      "step": 68300
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7351365685462952,
      "learning_rate": 0.0004790378231570526,
      "loss": 2.3639,
      "step": 68301
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7820448279380798,
      "learning_rate": 0.0004790345408920299,
      "loss": 2.2765,
      "step": 68302
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.6923346519470215,
      "learning_rate": 0.0004790312585937213,
      "loss": 2.1069,
      "step": 68303
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7348549365997314,
      "learning_rate": 0.00047902797626212743,
      "loss": 2.3555,
      "step": 68304
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.8024727702140808,
      "learning_rate": 0.00047902469389724884,
      "loss": 2.3602,
      "step": 68305
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7136980891227722,
      "learning_rate": 0.00047902141149908616,
      "loss": 2.1852,
      "step": 68306
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7411040663719177,
      "learning_rate": 0.00047901812906764015,
      "loss": 2.062,
      "step": 68307
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.781907856464386,
      "learning_rate": 0.00047901484660291114,
      "loss": 2.2018,
      "step": 68308
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7012523412704468,
      "learning_rate": 0.00047901156410489994,
      "loss": 2.23,
      "step": 68309
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.741315484046936,
      "learning_rate": 0.00047900828157360716,
      "loss": 2.3475,
      "step": 68310
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7624481916427612,
      "learning_rate": 0.00047900499900903337,
      "loss": 2.2155,
      "step": 68311
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.8100109696388245,
      "learning_rate": 0.0004790017164111791,
      "loss": 2.1718,
      "step": 68312
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7361852526664734,
      "learning_rate": 0.00047899843378004504,
      "loss": 2.0987,
      "step": 68313
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7258586287498474,
      "learning_rate": 0.0004789951511156318,
      "loss": 2.4013,
      "step": 68314
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7949199676513672,
      "learning_rate": 0.00047899186841794,
      "loss": 2.1804,
      "step": 68315
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7768603563308716,
      "learning_rate": 0.0004789885856869703,
      "loss": 2.0386,
      "step": 68316
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.735041081905365,
      "learning_rate": 0.00047898530292272317,
      "loss": 2.2286,
      "step": 68317
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7718848586082458,
      "learning_rate": 0.0004789820201251992,
      "loss": 2.3541,
      "step": 68318
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.9809250831604004,
      "learning_rate": 0.0004789787372943992,
      "loss": 2.12,
      "step": 68319
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7734420299530029,
      "learning_rate": 0.0004789754544303237,
      "loss": 2.4148,
      "step": 68320
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7606316804885864,
      "learning_rate": 0.0004789721715329732,
      "loss": 2.3205,
      "step": 68321
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.8141091465950012,
      "learning_rate": 0.0004789688886023485,
      "loss": 2.2278,
      "step": 68322
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7397567629814148,
      "learning_rate": 0.00047896560563845007,
      "loss": 2.0561,
      "step": 68323
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.677523136138916,
      "learning_rate": 0.0004789623226412785,
      "loss": 2.1157,
      "step": 68324
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.736142098903656,
      "learning_rate": 0.0004789590396108345,
      "loss": 2.1357,
      "step": 68325
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7885141968727112,
      "learning_rate": 0.0004789557565471186,
      "loss": 2.287,
      "step": 68326
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7632418274879456,
      "learning_rate": 0.0004789524734501314,
      "loss": 2.2592,
      "step": 68327
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.8668009042739868,
      "learning_rate": 0.00047894919031987374,
      "loss": 2.3183,
      "step": 68328
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.759579062461853,
      "learning_rate": 0.0004789459071563459,
      "loss": 2.2373,
      "step": 68329
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.8789958357810974,
      "learning_rate": 0.0004789426239595487,
      "loss": 2.1586,
      "step": 68330
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7586717009544373,
      "learning_rate": 0.00047893934072948266,
      "loss": 2.3274,
      "step": 68331
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7942084670066833,
      "learning_rate": 0.00047893605746614844,
      "loss": 2.2062,
      "step": 68332
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7853183746337891,
      "learning_rate": 0.0004789327741695465,
      "loss": 2.0477,
      "step": 68333
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.9615517854690552,
      "learning_rate": 0.0004789294908396778,
      "loss": 2.2807,
      "step": 68334
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.8113537430763245,
      "learning_rate": 0.00047892620747654263,
      "loss": 2.3377,
      "step": 68335
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.8450011610984802,
      "learning_rate": 0.0004789229240801417,
      "loss": 2.2128,
      "step": 68336
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.8133493065834045,
      "learning_rate": 0.00047891964065047555,
      "loss": 2.0166,
      "step": 68337
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.8341578841209412,
      "learning_rate": 0.00047891635718754497,
      "loss": 2.4736,
      "step": 68338
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.9231669306755066,
      "learning_rate": 0.0004789130736913504,
      "loss": 2.0969,
      "step": 68339
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7706089615821838,
      "learning_rate": 0.00047890979016189253,
      "loss": 2.1616,
      "step": 68340
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.6734086275100708,
      "learning_rate": 0.000478906506599172,
      "loss": 2.198,
      "step": 68341
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7718068957328796,
      "learning_rate": 0.00047890322300318924,
      "loss": 2.2145,
      "step": 68342
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7022374272346497,
      "learning_rate": 0.00047889993937394514,
      "loss": 2.0762,
      "step": 68343
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.8209255933761597,
      "learning_rate": 0.0004788966557114401,
      "loss": 2.1569,
      "step": 68344
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7263187170028687,
      "learning_rate": 0.0004788933720156747,
      "loss": 2.1737,
      "step": 68345
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.8245494365692139,
      "learning_rate": 0.0004788900882866499,
      "loss": 2.4297,
      "step": 68346
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.8293246626853943,
      "learning_rate": 0.0004788868045243658,
      "loss": 2.2944,
      "step": 68347
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7830421924591064,
      "learning_rate": 0.0004788835207288234,
      "loss": 2.2419,
      "step": 68348
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7166581749916077,
      "learning_rate": 0.0004788802369000232,
      "loss": 2.0929,
      "step": 68349
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.8088735342025757,
      "learning_rate": 0.0004788769530379657,
      "loss": 2.455,
      "step": 68350
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.788641095161438,
      "learning_rate": 0.0004788736691426516,
      "loss": 2.0442,
      "step": 68351
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.8317372798919678,
      "learning_rate": 0.00047887038521408155,
      "loss": 2.1425,
      "step": 68352
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7989789247512817,
      "learning_rate": 0.00047886710125225604,
      "loss": 2.2519,
      "step": 68353
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7621347308158875,
      "learning_rate": 0.0004788638172571759,
      "loss": 2.423,
      "step": 68354
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7501071095466614,
      "learning_rate": 0.00047886053322884146,
      "loss": 2.2262,
      "step": 68355
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.777739942073822,
      "learning_rate": 0.0004788572491672536,
      "loss": 2.3139,
      "step": 68356
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7958358526229858,
      "learning_rate": 0.0004788539650724127,
      "loss": 2.2177,
      "step": 68357
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.8984612822532654,
      "learning_rate": 0.0004788506809443195,
      "loss": 2.2312,
      "step": 68358
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7103216648101807,
      "learning_rate": 0.00047884739678297466,
      "loss": 2.3126,
      "step": 68359
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.021047592163086,
      "learning_rate": 0.00047884411258837864,
      "loss": 2.2682,
      "step": 68360
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7782225608825684,
      "learning_rate": 0.00047884082836053204,
      "loss": 2.1503,
      "step": 68361
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7628593444824219,
      "learning_rate": 0.00047883754409943567,
      "loss": 2.1725,
      "step": 68362
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7558049559593201,
      "learning_rate": 0.00047883425980509007,
      "loss": 2.1632,
      "step": 68363
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7692236304283142,
      "learning_rate": 0.0004788309754774956,
      "loss": 2.2202,
      "step": 68364
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.787155270576477,
      "learning_rate": 0.00047882769111665325,
      "loss": 2.2208,
      "step": 68365
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.8243914246559143,
      "learning_rate": 0.00047882440672256343,
      "loss": 2.1352,
      "step": 68366
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.8635026812553406,
      "learning_rate": 0.0004788211222952267,
      "loss": 2.1592,
      "step": 68367
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.8054457902908325,
      "learning_rate": 0.0004788178378346439,
      "loss": 2.2453,
      "step": 68368
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.845125138759613,
      "learning_rate": 0.0004788145533408154,
      "loss": 2.0889,
      "step": 68369
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7675489187240601,
      "learning_rate": 0.0004788112688137419,
      "loss": 2.1361,
      "step": 68370
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7820850610733032,
      "learning_rate": 0.000478807984253424,
      "loss": 2.2173,
      "step": 68371
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7779382467269897,
      "learning_rate": 0.0004788046996598623,
      "loss": 2.2536,
      "step": 68372
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7957836389541626,
      "learning_rate": 0.0004788014150330574,
      "loss": 2.133,
      "step": 68373
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7926303744316101,
      "learning_rate": 0.0004787981303730101,
      "loss": 2.3898,
      "step": 68374
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.8513176441192627,
      "learning_rate": 0.00047879484567972075,
      "loss": 2.0055,
      "step": 68375
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.775475025177002,
      "learning_rate": 0.00047879156095318994,
      "loss": 2.1131,
      "step": 68376
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7876160740852356,
      "learning_rate": 0.00047878827619341857,
      "loss": 2.252,
      "step": 68377
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7991092205047607,
      "learning_rate": 0.00047878499140040714,
      "loss": 2.0679,
      "step": 68378
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.8159581422805786,
      "learning_rate": 0.00047878170657415596,
      "loss": 2.0265,
      "step": 68379
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.8549329042434692,
      "learning_rate": 0.0004787784217146661,
      "loss": 2.2686,
      "step": 68380
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7955979108810425,
      "learning_rate": 0.00047877513682193786,
      "loss": 2.2765,
      "step": 68381
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.943233847618103,
      "learning_rate": 0.000478771851895972,
      "loss": 2.3392,
      "step": 68382
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.8736362457275391,
      "learning_rate": 0.0004787685669367691,
      "loss": 2.2284,
      "step": 68383
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.921561062335968,
      "learning_rate": 0.00047876528194432973,
      "loss": 2.0709,
      "step": 68384
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7811518311500549,
      "learning_rate": 0.0004787619969186544,
      "loss": 1.9442,
      "step": 68385
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7810497283935547,
      "learning_rate": 0.0004787587118597441,
      "loss": 2.1812,
      "step": 68386
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7591020464897156,
      "learning_rate": 0.00047875542676759894,
      "loss": 2.1977,
      "step": 68387
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.8069024085998535,
      "learning_rate": 0.0004787521416422198,
      "loss": 2.1636,
      "step": 68388
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7754999399185181,
      "learning_rate": 0.0004787488564836073,
      "loss": 2.2235,
      "step": 68389
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7777966856956482,
      "learning_rate": 0.00047874557129176204,
      "loss": 2.248,
      "step": 68390
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.6536989212036133,
      "learning_rate": 0.0004787422860666846,
      "loss": 1.9411,
      "step": 68391
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7591586112976074,
      "learning_rate": 0.0004787390008083756,
      "loss": 2.0655,
      "step": 68392
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.8063802123069763,
      "learning_rate": 0.0004787357155168356,
      "loss": 2.1381,
      "step": 68393
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7187283039093018,
      "learning_rate": 0.0004787324301920653,
      "loss": 2.273,
      "step": 68394
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.767406165599823,
      "learning_rate": 0.0004787291448340653,
      "loss": 2.2528,
      "step": 68395
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7834304571151733,
      "learning_rate": 0.0004787258594428361,
      "loss": 2.1501,
      "step": 68396
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.8395443558692932,
      "learning_rate": 0.00047872257401837846,
      "loss": 2.4027,
      "step": 68397
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.812670886516571,
      "learning_rate": 0.0004787192885606929,
      "loss": 2.3173,
      "step": 68398
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7917619347572327,
      "learning_rate": 0.00047871600306978,
      "loss": 2.1578,
      "step": 68399
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7756178975105286,
      "learning_rate": 0.0004787127175456404,
      "loss": 2.3124,
      "step": 68400
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.769777238368988,
      "learning_rate": 0.0004787094319882748,
      "loss": 2.2864,
      "step": 68401
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7461439967155457,
      "learning_rate": 0.00047870614639768376,
      "loss": 2.3611,
      "step": 68402
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.8254609107971191,
      "learning_rate": 0.0004787028607738678,
      "loss": 2.1988,
      "step": 68403
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7473642826080322,
      "learning_rate": 0.00047869957511682766,
      "loss": 2.5368,
      "step": 68404
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.8070028424263,
      "learning_rate": 0.0004786962894265639,
      "loss": 2.3676,
      "step": 68405
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7921056151390076,
      "learning_rate": 0.0004786930037030771,
      "loss": 2.1882,
      "step": 68406
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.9173064827919006,
      "learning_rate": 0.00047868971794636785,
      "loss": 2.1114,
      "step": 68407
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7015044093132019,
      "learning_rate": 0.0004786864321564369,
      "loss": 2.2294,
      "step": 68408
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.0618619918823242,
      "learning_rate": 0.0004786831463332848,
      "loss": 2.1616,
      "step": 68409
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7185558676719666,
      "learning_rate": 0.00047867986047691203,
      "loss": 2.1925,
      "step": 68410
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7790132761001587,
      "learning_rate": 0.0004786765745873193,
      "loss": 2.4692,
      "step": 68411
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.8052018284797668,
      "learning_rate": 0.0004786732886645073,
      "loss": 2.2762,
      "step": 68412
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.8294057250022888,
      "learning_rate": 0.00047867000270847656,
      "loss": 2.2002,
      "step": 68413
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.8415151834487915,
      "learning_rate": 0.00047866671671922763,
      "loss": 2.4473,
      "step": 68414
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7659586668014526,
      "learning_rate": 0.0004786634306967612,
      "loss": 2.0944,
      "step": 68415
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.9296188354492188,
      "learning_rate": 0.0004786601446410779,
      "loss": 2.1512,
      "step": 68416
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7996103167533875,
      "learning_rate": 0.00047865685855217824,
      "loss": 2.2068,
      "step": 68417
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7615526914596558,
      "learning_rate": 0.0004786535724300629,
      "loss": 2.3501,
      "step": 68418
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7074335813522339,
      "learning_rate": 0.0004786502862747326,
      "loss": 2.356,
      "step": 68419
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.9009614586830139,
      "learning_rate": 0.0004786470000861878,
      "loss": 2.2812,
      "step": 68420
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.8480323553085327,
      "learning_rate": 0.00047864371386442905,
      "loss": 2.244,
      "step": 68421
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7087565660476685,
      "learning_rate": 0.0004786404276094572,
      "loss": 2.1245,
      "step": 68422
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.8853198289871216,
      "learning_rate": 0.0004786371413212726,
      "loss": 2.033,
      "step": 68423
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.8433536887168884,
      "learning_rate": 0.00047863385499987603,
      "loss": 2.3215,
      "step": 68424
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7305185198783875,
      "learning_rate": 0.00047863056864526815,
      "loss": 2.2041,
      "step": 68425
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.6991954445838928,
      "learning_rate": 0.0004786272822574493,
      "loss": 1.9972,
      "step": 68426
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7317987680435181,
      "learning_rate": 0.00047862399583642034,
      "loss": 2.0548,
      "step": 68427
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.8835195302963257,
      "learning_rate": 0.0004786207093821818,
      "loss": 2.1266,
      "step": 68428
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7372889518737793,
      "learning_rate": 0.0004786174228947344,
      "loss": 2.3278,
      "step": 68429
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.8504889607429504,
      "learning_rate": 0.0004786141363740785,
      "loss": 2.1991,
      "step": 68430
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.8334611058235168,
      "learning_rate": 0.00047861084982021493,
      "loss": 2.3088,
      "step": 68431
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7388622164726257,
      "learning_rate": 0.0004786075632331443,
      "loss": 2.2666,
      "step": 68432
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.8196899890899658,
      "learning_rate": 0.00047860427661286704,
      "loss": 2.2393,
      "step": 68433
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7510912418365479,
      "learning_rate": 0.00047860098995938387,
      "loss": 2.2771,
      "step": 68434
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.8362532258033752,
      "learning_rate": 0.00047859770327269554,
      "loss": 2.2782,
      "step": 68435
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7881152033805847,
      "learning_rate": 0.00047859441655280233,
      "loss": 2.2419,
      "step": 68436
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7146135568618774,
      "learning_rate": 0.0004785911297997052,
      "loss": 2.1712,
      "step": 68437
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.9181478023529053,
      "learning_rate": 0.0004785878430134046,
      "loss": 2.1064,
      "step": 68438
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7342121601104736,
      "learning_rate": 0.0004785845561939011,
      "loss": 2.1077,
      "step": 68439
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7743532061576843,
      "learning_rate": 0.00047858126934119535,
      "loss": 2.1006,
      "step": 68440
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7874565124511719,
      "learning_rate": 0.00047857798245528807,
      "loss": 2.174,
      "step": 68441
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.8262782096862793,
      "learning_rate": 0.0004785746955361796,
      "loss": 2.2244,
      "step": 68442
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.8582175374031067,
      "learning_rate": 0.0004785714085838708,
      "loss": 2.1457,
      "step": 68443
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7577062249183655,
      "learning_rate": 0.00047856812159836226,
      "loss": 2.2949,
      "step": 68444
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.9383233785629272,
      "learning_rate": 0.00047856483457965447,
      "loss": 2.3861,
      "step": 68445
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.9010231494903564,
      "learning_rate": 0.0004785615475277481,
      "loss": 2.1523,
      "step": 68446
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7102277278900146,
      "learning_rate": 0.00047855826044264387,
      "loss": 2.3373,
      "step": 68447
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.8037561178207397,
      "learning_rate": 0.00047855497332434214,
      "loss": 2.1781,
      "step": 68448
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.9811937212944031,
      "learning_rate": 0.0004785516861728438,
      "loss": 2.1044,
      "step": 68449
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7677992582321167,
      "learning_rate": 0.00047854839898814924,
      "loss": 2.3262,
      "step": 68450
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.2703040838241577,
      "learning_rate": 0.0004785451117702592,
      "loss": 2.2549,
      "step": 68451
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7821943163871765,
      "learning_rate": 0.0004785418245191743,
      "loss": 2.3897,
      "step": 68452
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7391655445098877,
      "learning_rate": 0.0004785385372348951,
      "loss": 2.2441,
      "step": 68453
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7855818867683411,
      "learning_rate": 0.0004785352499174221,
      "loss": 2.2102,
      "step": 68454
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.8264662623405457,
      "learning_rate": 0.0004785319625667561,
      "loss": 2.1824,
      "step": 68455
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7938779592514038,
      "learning_rate": 0.00047852867518289764,
      "loss": 2.4708,
      "step": 68456
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7970462441444397,
      "learning_rate": 0.00047852538776584736,
      "loss": 2.201,
      "step": 68457
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.917873740196228,
      "learning_rate": 0.00047852210031560586,
      "loss": 2.281,
      "step": 68458
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.790481448173523,
      "learning_rate": 0.00047851881283217363,
      "loss": 2.0831,
      "step": 68459
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.8358462452888489,
      "learning_rate": 0.0004785155253155514,
      "loss": 2.2155,
      "step": 68460
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.9011142253875732,
      "learning_rate": 0.0004785122377657399,
      "loss": 2.0539,
      "step": 68461
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7944867610931396,
      "learning_rate": 0.0004785089501827395,
      "loss": 2.1448,
      "step": 68462
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.9764322638511658,
      "learning_rate": 0.0004785056625665509,
      "loss": 2.2565,
      "step": 68463
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7055040001869202,
      "learning_rate": 0.0004785023749171747,
      "loss": 2.0885,
      "step": 68464
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.9655933380126953,
      "learning_rate": 0.0004784990872346116,
      "loss": 2.2751,
      "step": 68465
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.853422224521637,
      "learning_rate": 0.00047849579951886215,
      "loss": 2.1741,
      "step": 68466
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.8528819680213928,
      "learning_rate": 0.00047849251176992696,
      "loss": 2.0702,
      "step": 68467
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.9350572824478149,
      "learning_rate": 0.0004784892239878066,
      "loss": 2.2788,
      "step": 68468
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.9207747578620911,
      "learning_rate": 0.0004784859361725018,
      "loss": 2.0935,
      "step": 68469
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.0879535675048828,
      "learning_rate": 0.00047848264832401304,
      "loss": 2.2885,
      "step": 68470
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7489603161811829,
      "learning_rate": 0.000478479360442341,
      "loss": 1.841,
      "step": 68471
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.0721198320388794,
      "learning_rate": 0.00047847607252748627,
      "loss": 1.9401,
      "step": 68472
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.0575722455978394,
      "learning_rate": 0.00047847278457944945,
      "loss": 2.26,
      "step": 68473
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.8086027503013611,
      "learning_rate": 0.0004784694965982312,
      "loss": 2.2474,
      "step": 68474
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7766618132591248,
      "learning_rate": 0.00047846620858383215,
      "loss": 2.0095,
      "step": 68475
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.9116320013999939,
      "learning_rate": 0.0004784629205362527,
      "loss": 2.1984,
      "step": 68476
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.8051678538322449,
      "learning_rate": 0.0004784596324554939,
      "loss": 2.2541,
      "step": 68477
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.8579542636871338,
      "learning_rate": 0.0004784563443415558,
      "loss": 2.4245,
      "step": 68478
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7752426266670227,
      "learning_rate": 0.0004784530561944394,
      "loss": 2.2355,
      "step": 68479
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.9534340500831604,
      "learning_rate": 0.0004784497680141453,
      "loss": 2.1398,
      "step": 68480
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.8345492482185364,
      "learning_rate": 0.0004784464798006739,
      "loss": 2.1868,
      "step": 68481
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.8112359046936035,
      "learning_rate": 0.00047844319155402595,
      "loss": 2.1193,
      "step": 68482
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.921999454498291,
      "learning_rate": 0.0004784399032742021,
      "loss": 2.1336,
      "step": 68483
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.031276822090149,
      "learning_rate": 0.0004784366149612029,
      "loss": 1.9687,
      "step": 68484
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7040321230888367,
      "learning_rate": 0.00047843332661502886,
      "loss": 2.4175,
      "step": 68485
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.8076331615447998,
      "learning_rate": 0.00047843003823568086,
      "loss": 2.0277,
      "step": 68486
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7651723027229309,
      "learning_rate": 0.00047842674982315924,
      "loss": 2.3689,
      "step": 68487
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7638721466064453,
      "learning_rate": 0.00047842346137746467,
      "loss": 2.3807,
      "step": 68488
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.8644598126411438,
      "learning_rate": 0.0004784201728985979,
      "loss": 2.0515,
      "step": 68489
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.8074357509613037,
      "learning_rate": 0.00047841688438655944,
      "loss": 2.3905,
      "step": 68490
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.8328344821929932,
      "learning_rate": 0.00047841359584134996,
      "loss": 2.2085,
      "step": 68491
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.9772103428840637,
      "learning_rate": 0.00047841030726296997,
      "loss": 2.4053,
      "step": 68492
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.8462274074554443,
      "learning_rate": 0.00047840701865142006,
      "loss": 2.2058,
      "step": 68493
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.8573850393295288,
      "learning_rate": 0.000478403730006701,
      "loss": 2.3094,
      "step": 68494
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7838911414146423,
      "learning_rate": 0.0004784004413288133,
      "loss": 2.288,
      "step": 68495
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.786125659942627,
      "learning_rate": 0.00047839715261775766,
      "loss": 2.1118,
      "step": 68496
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.757754385471344,
      "learning_rate": 0.0004783938638735346,
      "loss": 2.2583,
      "step": 68497
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.8550018072128296,
      "learning_rate": 0.0004783905750961447,
      "loss": 2.425,
      "step": 68498
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7513367533683777,
      "learning_rate": 0.00047838728628558865,
      "loss": 2.3045,
      "step": 68499
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.9000009894371033,
      "learning_rate": 0.00047838399744186703,
      "loss": 2.1211,
      "step": 68500
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7617443203926086,
      "learning_rate": 0.00047838070856498044,
      "loss": 2.0501,
      "step": 68501
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.9709819555282593,
      "learning_rate": 0.0004783774196549295,
      "loss": 2.2897,
      "step": 68502
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.9740697145462036,
      "learning_rate": 0.00047837413071171483,
      "loss": 2.1892,
      "step": 68503
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.812656581401825,
      "learning_rate": 0.00047837084173533713,
      "loss": 2.3238,
      "step": 68504
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.738000750541687,
      "learning_rate": 0.0004783675527257968,
      "loss": 2.1436,
      "step": 68505
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.9104195833206177,
      "learning_rate": 0.0004783642636830946,
      "loss": 2.3307,
      "step": 68506
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7822718024253845,
      "learning_rate": 0.00047836097460723123,
      "loss": 2.3181,
      "step": 68507
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.743156373500824,
      "learning_rate": 0.00047835768549820714,
      "loss": 2.211,
      "step": 68508
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.799403965473175,
      "learning_rate": 0.00047835439635602293,
      "loss": 2.2348,
      "step": 68509
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.712365984916687,
      "learning_rate": 0.0004783511071806792,
      "loss": 2.3165,
      "step": 68510
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7841585278511047,
      "learning_rate": 0.00047834781797217687,
      "loss": 2.2927,
      "step": 68511
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.8212823271751404,
      "learning_rate": 0.0004783445287305161,
      "loss": 2.4154,
      "step": 68512
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.724211573600769,
      "learning_rate": 0.00047834123945569786,
      "loss": 2.0928,
      "step": 68513
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.805743396282196,
      "learning_rate": 0.0004783379501477225,
      "loss": 2.4208,
      "step": 68514
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7583330273628235,
      "learning_rate": 0.00047833466080659083,
      "loss": 2.5372,
      "step": 68515
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.6787958145141602,
      "learning_rate": 0.0004783313714323033,
      "loss": 2.2859,
      "step": 68516
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.8422784209251404,
      "learning_rate": 0.00047832808202486073,
      "loss": 2.222,
      "step": 68517
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.725220799446106,
      "learning_rate": 0.0004783247925842635,
      "loss": 2.1465,
      "step": 68518
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.8071536421775818,
      "learning_rate": 0.0004783215031105124,
      "loss": 2.2056,
      "step": 68519
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.8229994773864746,
      "learning_rate": 0.0004783182136036079,
      "loss": 2.4033,
      "step": 68520
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.8268702626228333,
      "learning_rate": 0.0004783149240635507,
      "loss": 2.1573,
      "step": 68521
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.763749361038208,
      "learning_rate": 0.0004783116344903413,
      "loss": 2.2041,
      "step": 68522
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7665212154388428,
      "learning_rate": 0.0004783083448839805,
      "loss": 2.2682,
      "step": 68523
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.794722318649292,
      "learning_rate": 0.0004783050552444688,
      "loss": 2.0948,
      "step": 68524
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.8243274092674255,
      "learning_rate": 0.00047830176557180684,
      "loss": 2.4407,
      "step": 68525
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7144200205802917,
      "learning_rate": 0.00047829847586599526,
      "loss": 2.1138,
      "step": 68526
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7760862708091736,
      "learning_rate": 0.0004782951861270345,
      "loss": 2.3026,
      "step": 68527
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7522116303443909,
      "learning_rate": 0.0004782918963549254,
      "loss": 2.1339,
      "step": 68528
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.809821605682373,
      "learning_rate": 0.00047828860654966845,
      "loss": 2.1816,
      "step": 68529
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7880455255508423,
      "learning_rate": 0.0004782853167112642,
      "loss": 2.1474,
      "step": 68530
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.759710967540741,
      "learning_rate": 0.00047828202683971346,
      "loss": 2.2567,
      "step": 68531
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.817831814289093,
      "learning_rate": 0.0004782787369350167,
      "loss": 2.3814,
      "step": 68532
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7763688564300537,
      "learning_rate": 0.00047827544699717453,
      "loss": 2.312,
      "step": 68533
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7469188570976257,
      "learning_rate": 0.0004782721570261876,
      "loss": 2.4178,
      "step": 68534
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7792336940765381,
      "learning_rate": 0.00047826886702205654,
      "loss": 2.0776,
      "step": 68535
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.8353365063667297,
      "learning_rate": 0.0004782655769847819,
      "loss": 2.0003,
      "step": 68536
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.731656014919281,
      "learning_rate": 0.0004782622869143643,
      "loss": 2.2218,
      "step": 68537
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7345714569091797,
      "learning_rate": 0.0004782589968108044,
      "loss": 2.1983,
      "step": 68538
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.9046345353126526,
      "learning_rate": 0.0004782557066741028,
      "loss": 2.2329,
      "step": 68539
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7686400413513184,
      "learning_rate": 0.0004782524165042601,
      "loss": 2.3714,
      "step": 68540
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.8064398169517517,
      "learning_rate": 0.00047824912630127687,
      "loss": 2.0575,
      "step": 68541
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.9193221926689148,
      "learning_rate": 0.00047824583606515377,
      "loss": 2.2533,
      "step": 68542
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.8147339820861816,
      "learning_rate": 0.0004782425457958915,
      "loss": 2.3641,
      "step": 68543
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7441594004631042,
      "learning_rate": 0.00047823925549349045,
      "loss": 2.2786,
      "step": 68544
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.8733265995979309,
      "learning_rate": 0.0004782359651579514,
      "loss": 2.3572,
      "step": 68545
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7584869861602783,
      "learning_rate": 0.0004782326747892749,
      "loss": 2.3443,
      "step": 68546
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7992066144943237,
      "learning_rate": 0.0004782293843874617,
      "loss": 2.1158,
      "step": 68547
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7843571901321411,
      "learning_rate": 0.0004782260939525121,
      "loss": 2.2589,
      "step": 68548
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7553185224533081,
      "learning_rate": 0.00047822280348442704,
      "loss": 2.1538,
      "step": 68549
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.8088008761405945,
      "learning_rate": 0.000478219512983207,
      "loss": 2.3138,
      "step": 68550
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7722027897834778,
      "learning_rate": 0.0004782162224488526,
      "loss": 2.1967,
      "step": 68551
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7741453051567078,
      "learning_rate": 0.00047821293188136433,
      "loss": 2.2384,
      "step": 68552
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.8142913579940796,
      "learning_rate": 0.00047820964128074303,
      "loss": 2.3197,
      "step": 68553
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.718498170375824,
      "learning_rate": 0.00047820635064698905,
      "loss": 2.2713,
      "step": 68554
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7202954292297363,
      "learning_rate": 0.0004782030599801032,
      "loss": 2.1317,
      "step": 68555
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7729870676994324,
      "learning_rate": 0.0004781997692800862,
      "loss": 2.1211,
      "step": 68556
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7429133653640747,
      "learning_rate": 0.0004781964785469383,
      "loss": 2.0577,
      "step": 68557
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.801853358745575,
      "learning_rate": 0.00047819318778066035,
      "loss": 2.1434,
      "step": 68558
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7109798192977905,
      "learning_rate": 0.00047818989698125296,
      "loss": 2.3125,
      "step": 68559
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7455645203590393,
      "learning_rate": 0.0004781866061487168,
      "loss": 2.1264,
      "step": 68560
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7130734324455261,
      "learning_rate": 0.0004781833152830522,
      "loss": 2.2395,
      "step": 68561
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7543339729309082,
      "learning_rate": 0.00047818002438426,
      "loss": 2.0681,
      "step": 68562
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.8275969624519348,
      "learning_rate": 0.0004781767334523409,
      "loss": 2.193,
      "step": 68563
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7575856447219849,
      "learning_rate": 0.00047817344248729526,
      "loss": 2.1886,
      "step": 68564
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.6767757534980774,
      "learning_rate": 0.00047817015148912386,
      "loss": 2.1862,
      "step": 68565
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.8520119786262512,
      "learning_rate": 0.0004781668604578273,
      "loss": 2.0535,
      "step": 68566
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.814568281173706,
      "learning_rate": 0.0004781635693934061,
      "loss": 2.3566,
      "step": 68567
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7549453973770142,
      "learning_rate": 0.00047816027829586095,
      "loss": 2.3061,
      "step": 68568
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7226384282112122,
      "learning_rate": 0.0004781569871651924,
      "loss": 2.1915,
      "step": 68569
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7271689176559448,
      "learning_rate": 0.00047815369600140114,
      "loss": 2.1387,
      "step": 68570
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7524417042732239,
      "learning_rate": 0.0004781504048044877,
      "loss": 2.0686,
      "step": 68571
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7965783476829529,
      "learning_rate": 0.0004781471135744529,
      "loss": 2.0942,
      "step": 68572
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7177719473838806,
      "learning_rate": 0.000478143822311297,
      "loss": 2.2513,
      "step": 68573
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.745529055595398,
      "learning_rate": 0.0004781405310150208,
      "loss": 2.2402,
      "step": 68574
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.8070530891418457,
      "learning_rate": 0.00047813723968562503,
      "loss": 1.9561,
      "step": 68575
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7614380121231079,
      "learning_rate": 0.0004781339483231102,
      "loss": 2.3525,
      "step": 68576
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.8663358688354492,
      "learning_rate": 0.00047813065692747683,
      "loss": 2.0137,
      "step": 68577
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7960948944091797,
      "learning_rate": 0.0004781273654987256,
      "loss": 2.3031,
      "step": 68578
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7886304259300232,
      "learning_rate": 0.0004781240740368571,
      "loss": 2.2373,
      "step": 68579
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.9792904853820801,
      "learning_rate": 0.000478120782541872,
      "loss": 2.1152,
      "step": 68580
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.796657383441925,
      "learning_rate": 0.00047811749101377095,
      "loss": 2.3335,
      "step": 68581
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.8032203316688538,
      "learning_rate": 0.0004781141994525545,
      "loss": 2.1196,
      "step": 68582
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.8334988355636597,
      "learning_rate": 0.00047811090785822313,
      "loss": 2.2408,
      "step": 68583
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.8546702265739441,
      "learning_rate": 0.0004781076162307777,
      "loss": 2.2704,
      "step": 68584
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7449828386306763,
      "learning_rate": 0.00047810432457021866,
      "loss": 2.5733,
      "step": 68585
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7340637445449829,
      "learning_rate": 0.00047810103287654663,
      "loss": 2.2815,
      "step": 68586
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7213163375854492,
      "learning_rate": 0.00047809774114976236,
      "loss": 2.3701,
      "step": 68587
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.8394018411636353,
      "learning_rate": 0.0004780944493898663,
      "loss": 2.0526,
      "step": 68588
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.9204718470573425,
      "learning_rate": 0.00047809115759685905,
      "loss": 2.1486,
      "step": 68589
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7525904178619385,
      "learning_rate": 0.0004780878657707413,
      "loss": 2.104,
      "step": 68590
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.8662611842155457,
      "learning_rate": 0.00047808457391151373,
      "loss": 2.2547,
      "step": 68591
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.226234793663025,
      "learning_rate": 0.0004780812820191769,
      "loss": 2.2055,
      "step": 68592
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.8105562329292297,
      "learning_rate": 0.0004780779900937314,
      "loss": 2.1629,
      "step": 68593
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.8221744894981384,
      "learning_rate": 0.00047807469813517775,
      "loss": 2.2187,
      "step": 68594
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.0564520359039307,
      "learning_rate": 0.0004780714061435167,
      "loss": 2.1625,
      "step": 68595
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.9491364359855652,
      "learning_rate": 0.00047806811411874885,
      "loss": 2.4145,
      "step": 68596
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7374988198280334,
      "learning_rate": 0.00047806482206087466,
      "loss": 2.2366,
      "step": 68597
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7860091924667358,
      "learning_rate": 0.00047806152996989496,
      "loss": 2.2954,
      "step": 68598
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.2257025241851807,
      "learning_rate": 0.0004780582378458103,
      "loss": 2.3178,
      "step": 68599
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.8926081657409668,
      "learning_rate": 0.0004780549456886211,
      "loss": 2.1892,
      "step": 68600
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.8226163387298584,
      "learning_rate": 0.00047805165349832827,
      "loss": 2.2957,
      "step": 68601
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.8152293562889099,
      "learning_rate": 0.00047804836127493224,
      "loss": 2.2052,
      "step": 68602
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.0264779329299927,
      "learning_rate": 0.0004780450690184336,
      "loss": 2.3272,
      "step": 68603
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.9182763695716858,
      "learning_rate": 0.00047804177672883307,
      "loss": 2.3354,
      "step": 68604
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7887060046195984,
      "learning_rate": 0.0004780384844061312,
      "loss": 2.178,
      "step": 68605
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.9021496176719666,
      "learning_rate": 0.00047803519205032863,
      "loss": 2.4185,
      "step": 68606
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.983099639415741,
      "learning_rate": 0.0004780318996614259,
      "loss": 2.3289,
      "step": 68607
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.9960557222366333,
      "learning_rate": 0.00047802860723942375,
      "loss": 2.1562,
      "step": 68608
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7592593431472778,
      "learning_rate": 0.00047802531478432266,
      "loss": 2.2631,
      "step": 68609
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.9439935684204102,
      "learning_rate": 0.0004780220222961233,
      "loss": 2.3034,
      "step": 68610
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.0726290941238403,
      "learning_rate": 0.0004780187297748264,
      "loss": 2.3884,
      "step": 68611
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7736442685127258,
      "learning_rate": 0.0004780154372204323,
      "loss": 2.3091,
      "step": 68612
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7451945543289185,
      "learning_rate": 0.0004780121446329419,
      "loss": 2.1401,
      "step": 68613
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.8163747191429138,
      "learning_rate": 0.0004780088520123556,
      "loss": 2.1835,
      "step": 68614
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7551363110542297,
      "learning_rate": 0.00047800555935867417,
      "loss": 2.1883,
      "step": 68615
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7453306913375854,
      "learning_rate": 0.00047800226667189805,
      "loss": 2.0749,
      "step": 68616
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7975128293037415,
      "learning_rate": 0.000477998973952028,
      "loss": 2.2002,
      "step": 68617
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.8228522539138794,
      "learning_rate": 0.00047799568119906465,
      "loss": 2.3285,
      "step": 68618
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.8246328830718994,
      "learning_rate": 0.00047799238841300846,
      "loss": 2.2837,
      "step": 68619
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.8019915819168091,
      "learning_rate": 0.00047798909559386015,
      "loss": 2.2017,
      "step": 68620
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7293211817741394,
      "learning_rate": 0.0004779858027416203,
      "loss": 2.1637,
      "step": 68621
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.9598735570907593,
      "learning_rate": 0.00047798250985628946,
      "loss": 2.2436,
      "step": 68622
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7521276473999023,
      "learning_rate": 0.0004779792169378684,
      "loss": 2.2453,
      "step": 68623
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.839948296546936,
      "learning_rate": 0.0004779759239863576,
      "loss": 2.3263,
      "step": 68624
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7253074049949646,
      "learning_rate": 0.0004779726310017578,
      "loss": 2.4233,
      "step": 68625
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7758492231369019,
      "learning_rate": 0.0004779693379840694,
      "loss": 2.157,
      "step": 68626
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7585663199424744,
      "learning_rate": 0.0004779660449332932,
      "loss": 2.361,
      "step": 68627
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.8044930100440979,
      "learning_rate": 0.0004779627518494298,
      "loss": 2.2106,
      "step": 68628
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7448480129241943,
      "learning_rate": 0.00047795945873247965,
      "loss": 2.2537,
      "step": 68629
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7856441736221313,
      "learning_rate": 0.0004779561655824436,
      "loss": 2.3888,
      "step": 68630
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.775776743888855,
      "learning_rate": 0.00047795287239932213,
      "loss": 2.3505,
      "step": 68631
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7612584233283997,
      "learning_rate": 0.00047794957918311583,
      "loss": 2.1572,
      "step": 68632
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.737885057926178,
      "learning_rate": 0.00047794628593382526,
      "loss": 2.2088,
      "step": 68633
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7401518225669861,
      "learning_rate": 0.00047794299265145124,
      "loss": 2.1682,
      "step": 68634
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7879078388214111,
      "learning_rate": 0.00047793969933599426,
      "loss": 2.1947,
      "step": 68635
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7349500060081482,
      "learning_rate": 0.00047793640598745485,
      "loss": 2.205,
      "step": 68636
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7398000955581665,
      "learning_rate": 0.0004779331126058338,
      "loss": 2.3302,
      "step": 68637
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.8157235383987427,
      "learning_rate": 0.00047792981919113147,
      "loss": 2.0663,
      "step": 68638
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.8888394236564636,
      "learning_rate": 0.0004779265257433487,
      "loss": 2.2341,
      "step": 68639
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.733447790145874,
      "learning_rate": 0.0004779232322624861,
      "loss": 2.1246,
      "step": 68640
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.058415412902832,
      "learning_rate": 0.00047791993874854417,
      "loss": 2.26,
      "step": 68641
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7492038607597351,
      "learning_rate": 0.0004779166452015236,
      "loss": 2.0562,
      "step": 68642
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.733608603477478,
      "learning_rate": 0.0004779133516214249,
      "loss": 1.866,
      "step": 68643
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7975144386291504,
      "learning_rate": 0.00047791005800824877,
      "loss": 2.0768,
      "step": 68644
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7904048562049866,
      "learning_rate": 0.00047790676436199585,
      "loss": 2.037,
      "step": 68645
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.780807614326477,
      "learning_rate": 0.0004779034706826666,
      "loss": 2.2284,
      "step": 68646
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.724212646484375,
      "learning_rate": 0.0004779001769702618,
      "loss": 2.1519,
      "step": 68647
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.0396897792816162,
      "learning_rate": 0.00047789688322478205,
      "loss": 2.3022,
      "step": 68648
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7623406648635864,
      "learning_rate": 0.0004778935894462278,
      "loss": 2.2574,
      "step": 68649
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.8285585045814514,
      "learning_rate": 0.00047789029563459983,
      "loss": 2.337,
      "step": 68650
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7656322717666626,
      "learning_rate": 0.00047788700178989877,
      "loss": 2.3599,
      "step": 68651
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.887885570526123,
      "learning_rate": 0.00047788370791212504,
      "loss": 2.0861,
      "step": 68652
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.8288717865943909,
      "learning_rate": 0.00047788041400127936,
      "loss": 2.2805,
      "step": 68653
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.8648611307144165,
      "learning_rate": 0.0004778771200573625,
      "loss": 2.1335,
      "step": 68654
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.8487404584884644,
      "learning_rate": 0.0004778738260803748,
      "loss": 2.0243,
      "step": 68655
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7601878643035889,
      "learning_rate": 0.00047787053207031704,
      "loss": 2.1797,
      "step": 68656
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.745820164680481,
      "learning_rate": 0.00047786723802718985,
      "loss": 2.3788,
      "step": 68657
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.8194056153297424,
      "learning_rate": 0.0004778639439509936,
      "loss": 2.2345,
      "step": 68658
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.849715530872345,
      "learning_rate": 0.00047786064984172924,
      "loss": 2.1954,
      "step": 68659
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7075414061546326,
      "learning_rate": 0.0004778573556993972,
      "loss": 1.9816,
      "step": 68660
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7638113498687744,
      "learning_rate": 0.0004778540615239981,
      "loss": 2.163,
      "step": 68661
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7635592222213745,
      "learning_rate": 0.0004778507673155326,
      "loss": 2.3519,
      "step": 68662
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7758728861808777,
      "learning_rate": 0.0004778474730740013,
      "loss": 2.1437,
      "step": 68663
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7125690579414368,
      "learning_rate": 0.00047784417879940475,
      "loss": 2.0516,
      "step": 68664
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.8035701513290405,
      "learning_rate": 0.0004778408844917436,
      "loss": 2.3762,
      "step": 68665
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7345132827758789,
      "learning_rate": 0.00047783759015101855,
      "loss": 2.3886,
      "step": 68666
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7942543029785156,
      "learning_rate": 0.0004778342957772301,
      "loss": 2.3357,
      "step": 68667
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7998239398002625,
      "learning_rate": 0.00047783100137037887,
      "loss": 2.1721,
      "step": 68668
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7056055665016174,
      "learning_rate": 0.0004778277069304655,
      "loss": 2.213,
      "step": 68669
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.8249467611312866,
      "learning_rate": 0.0004778244124574906,
      "loss": 2.4053,
      "step": 68670
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7194281816482544,
      "learning_rate": 0.00047782111795145483,
      "loss": 2.2369,
      "step": 68671
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7953585386276245,
      "learning_rate": 0.0004778178234123587,
      "loss": 2.1601,
      "step": 68672
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7605799436569214,
      "learning_rate": 0.0004778145288402029,
      "loss": 2.35,
      "step": 68673
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.8038753867149353,
      "learning_rate": 0.00047781123423498804,
      "loss": 2.2729,
      "step": 68674
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7890426516532898,
      "learning_rate": 0.0004778079395967147,
      "loss": 2.222,
      "step": 68675
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7509768605232239,
      "learning_rate": 0.00047780464492538346,
      "loss": 2.1101,
      "step": 68676
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7467167973518372,
      "learning_rate": 0.0004778013502209951,
      "loss": 2.0317,
      "step": 68677
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.743532121181488,
      "learning_rate": 0.00047779805548355003,
      "loss": 2.3357,
      "step": 68678
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.8524971604347229,
      "learning_rate": 0.00047779476071304895,
      "loss": 2.266,
      "step": 68679
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7326281666755676,
      "learning_rate": 0.0004777914659094925,
      "loss": 2.2308,
      "step": 68680
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7780808210372925,
      "learning_rate": 0.0004777881710728812,
      "loss": 2.0481,
      "step": 68681
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7524860501289368,
      "learning_rate": 0.00047778487620321586,
      "loss": 2.1628,
      "step": 68682
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7691356539726257,
      "learning_rate": 0.00047778158130049676,
      "loss": 2.3928,
      "step": 68683
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7289021015167236,
      "learning_rate": 0.00047777828636472484,
      "loss": 2.0954,
      "step": 68684
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.6579389572143555,
      "learning_rate": 0.0004777749913959005,
      "loss": 2.061,
      "step": 68685
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.862309455871582,
      "learning_rate": 0.00047777169639402447,
      "loss": 2.1949,
      "step": 68686
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.8241179585456848,
      "learning_rate": 0.00047776840135909743,
      "loss": 2.1055,
      "step": 68687
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7895469069480896,
      "learning_rate": 0.0004777651062911198,
      "loss": 2.1499,
      "step": 68688
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7726374864578247,
      "learning_rate": 0.00047776181119009225,
      "loss": 2.0861,
      "step": 68689
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.8958973288536072,
      "learning_rate": 0.00047775851605601547,
      "loss": 2.2298,
      "step": 68690
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.775844156742096,
      "learning_rate": 0.00047775522088889,
      "loss": 2.1347,
      "step": 68691
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7671840786933899,
      "learning_rate": 0.00047775192568871644,
      "loss": 2.1427,
      "step": 68692
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.6667783260345459,
      "learning_rate": 0.00047774863045549554,
      "loss": 2.0338,
      "step": 68693
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7454613447189331,
      "learning_rate": 0.00047774533518922773,
      "loss": 2.1704,
      "step": 68694
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.836073637008667,
      "learning_rate": 0.00047774203988991377,
      "loss": 2.1781,
      "step": 68695
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7221630811691284,
      "learning_rate": 0.0004777387445575541,
      "loss": 1.9019,
      "step": 68696
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.8573644161224365,
      "learning_rate": 0.00047773544919214953,
      "loss": 2.338,
      "step": 68697
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.8063821196556091,
      "learning_rate": 0.0004777321537937006,
      "loss": 2.057,
      "step": 68698
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7417356371879578,
      "learning_rate": 0.0004777288583622079,
      "loss": 2.1348,
      "step": 68699
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.8113195896148682,
      "learning_rate": 0.0004777255628976721,
      "loss": 2.4893,
      "step": 68700
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.9547227621078491,
      "learning_rate": 0.00047772226740009364,
      "loss": 2.061,
      "step": 68701
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.8505289554595947,
      "learning_rate": 0.00047771897186947327,
      "loss": 2.2815,
      "step": 68702
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.8496684432029724,
      "learning_rate": 0.0004777156763058117,
      "loss": 2.0653,
      "step": 68703
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7830938100814819,
      "learning_rate": 0.00047771238070910935,
      "loss": 2.3189,
      "step": 68704
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.873136579990387,
      "learning_rate": 0.00047770908507936694,
      "loss": 2.299,
      "step": 68705
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7477142214775085,
      "learning_rate": 0.00047770578941658503,
      "loss": 2.2064,
      "step": 68706
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7130242586135864,
      "learning_rate": 0.00047770249372076424,
      "loss": 2.316,
      "step": 68707
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7923840880393982,
      "learning_rate": 0.00047769919799190525,
      "loss": 2.1159,
      "step": 68708
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.8598295450210571,
      "learning_rate": 0.0004776959022300086,
      "loss": 2.2117,
      "step": 68709
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7905685305595398,
      "learning_rate": 0.00047769260643507495,
      "loss": 2.0664,
      "step": 68710
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7802470922470093,
      "learning_rate": 0.00047768931060710487,
      "loss": 2.1519,
      "step": 68711
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7228879332542419,
      "learning_rate": 0.00047768601474609904,
      "loss": 2.049,
      "step": 68712
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.8714085817337036,
      "learning_rate": 0.00047768271885205796,
      "loss": 2.1083,
      "step": 68713
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.9053436517715454,
      "learning_rate": 0.00047767942292498226,
      "loss": 2.1773,
      "step": 68714
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.8114075064659119,
      "learning_rate": 0.0004776761269648727,
      "loss": 2.3074,
      "step": 68715
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.8246335387229919,
      "learning_rate": 0.00047767283097172974,
      "loss": 2.3009,
      "step": 68716
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7970120906829834,
      "learning_rate": 0.00047766953494555417,
      "loss": 2.3943,
      "step": 68717
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.800922155380249,
      "learning_rate": 0.0004776662388863463,
      "loss": 2.2207,
      "step": 68718
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.733137845993042,
      "learning_rate": 0.00047766294279410706,
      "loss": 2.0132,
      "step": 68719
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7293576598167419,
      "learning_rate": 0.0004776596466688369,
      "loss": 2.0828,
      "step": 68720
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7080650329589844,
      "learning_rate": 0.0004776563505105364,
      "loss": 2.1659,
      "step": 68721
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.9238507151603699,
      "learning_rate": 0.0004776530543192063,
      "loss": 2.2485,
      "step": 68722
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7646899819374084,
      "learning_rate": 0.00047764975809484706,
      "loss": 2.169,
      "step": 68723
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.8007586002349854,
      "learning_rate": 0.0004776464618374594,
      "loss": 2.2234,
      "step": 68724
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.9130878448486328,
      "learning_rate": 0.00047764316554704394,
      "loss": 2.3254,
      "step": 68725
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7607523202896118,
      "learning_rate": 0.0004776398692236012,
      "loss": 2.0009,
      "step": 68726
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.8905215263366699,
      "learning_rate": 0.00047763657286713197,
      "loss": 2.4143,
      "step": 68727
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7887743711471558,
      "learning_rate": 0.0004776332764776367,
      "loss": 2.0101,
      "step": 68728
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7855430841445923,
      "learning_rate": 0.00047762998005511603,
      "loss": 2.4162,
      "step": 68729
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7992402911186218,
      "learning_rate": 0.0004776266835995706,
      "loss": 2.2072,
      "step": 68730
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.8108603954315186,
      "learning_rate": 0.000477623387111001,
      "loss": 2.2479,
      "step": 68731
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7883763909339905,
      "learning_rate": 0.0004776200905894078,
      "loss": 2.2467,
      "step": 68732
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.8355216383934021,
      "learning_rate": 0.00047761679403479185,
      "loss": 2.0575,
      "step": 68733
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.842054545879364,
      "learning_rate": 0.00047761349744715345,
      "loss": 2.4813,
      "step": 68734
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.0430707931518555,
      "learning_rate": 0.00047761020082649335,
      "loss": 2.2276,
      "step": 68735
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.9490113258361816,
      "learning_rate": 0.00047760690417281225,
      "loss": 2.3567,
      "step": 68736
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7597470283508301,
      "learning_rate": 0.0004776036074861106,
      "loss": 2.1858,
      "step": 68737
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.8383627533912659,
      "learning_rate": 0.00047760031076638914,
      "loss": 1.9848,
      "step": 68738
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.0004178285598755,
      "learning_rate": 0.0004775970140136484,
      "loss": 2.2388,
      "step": 68739
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.799210250377655,
      "learning_rate": 0.000477593717227889,
      "loss": 2.3015,
      "step": 68740
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.8310843110084534,
      "learning_rate": 0.00047759042040911156,
      "loss": 2.2695,
      "step": 68741
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.9816194176673889,
      "learning_rate": 0.0004775871235573168,
      "loss": 2.3845,
      "step": 68742
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7928508520126343,
      "learning_rate": 0.00047758382667250515,
      "loss": 2.1049,
      "step": 68743
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.8377786874771118,
      "learning_rate": 0.0004775805297546773,
      "loss": 2.3654,
      "step": 68744
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7958177328109741,
      "learning_rate": 0.000477577232803834,
      "loss": 2.1848,
      "step": 68745
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7875360250473022,
      "learning_rate": 0.0004775739358199757,
      "loss": 2.2516,
      "step": 68746
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7566880583763123,
      "learning_rate": 0.000477570638803103,
      "loss": 2.2419,
      "step": 68747
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7696954607963562,
      "learning_rate": 0.0004775673417532166,
      "loss": 2.3216,
      "step": 68748
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.8206058740615845,
      "learning_rate": 0.00047756404467031707,
      "loss": 2.2663,
      "step": 68749
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.8198421001434326,
      "learning_rate": 0.00047756074755440496,
      "loss": 2.2326,
      "step": 68750
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7163205146789551,
      "learning_rate": 0.00047755745040548107,
      "loss": 2.1274,
      "step": 68751
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.8644382357597351,
      "learning_rate": 0.0004775541532235458,
      "loss": 2.2968,
      "step": 68752
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7564602494239807,
      "learning_rate": 0.0004775508560085999,
      "loss": 2.0537,
      "step": 68753
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.823360025882721,
      "learning_rate": 0.00047754755876064397,
      "loss": 2.4122,
      "step": 68754
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7909398078918457,
      "learning_rate": 0.0004775442614796786,
      "loss": 2.076,
      "step": 68755
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.833685576915741,
      "learning_rate": 0.00047754096416570435,
      "loss": 2.2545,
      "step": 68756
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7759900093078613,
      "learning_rate": 0.0004775376668187219,
      "loss": 2.1733,
      "step": 68757
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7941684126853943,
      "learning_rate": 0.00047753436943873194,
      "loss": 2.1621,
      "step": 68758
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7626033425331116,
      "learning_rate": 0.0004775310720257349,
      "loss": 2.1558,
      "step": 68759
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.8579163551330566,
      "learning_rate": 0.00047752777457973146,
      "loss": 2.2088,
      "step": 68760
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.820974588394165,
      "learning_rate": 0.00047752447710072227,
      "loss": 2.2432,
      "step": 68761
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.700951337814331,
      "learning_rate": 0.000477521179588708,
      "loss": 2.2853,
      "step": 68762
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7689304351806641,
      "learning_rate": 0.0004775178820436891,
      "loss": 2.3527,
      "step": 68763
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7715657353401184,
      "learning_rate": 0.00047751458446566636,
      "loss": 2.1982,
      "step": 68764
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.728349506855011,
      "learning_rate": 0.00047751128685464026,
      "loss": 2.0098,
      "step": 68765
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7153612971305847,
      "learning_rate": 0.0004775079892106115,
      "loss": 2.4922,
      "step": 68766
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7854710221290588,
      "learning_rate": 0.0004775046915335806,
      "loss": 2.1698,
      "step": 68767
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.8504638671875,
      "learning_rate": 0.0004775013938235482,
      "loss": 2.1586,
      "step": 68768
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.023567795753479,
      "learning_rate": 0.000477498096080515,
      "loss": 2.1904,
      "step": 68769
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.9202847480773926,
      "learning_rate": 0.00047749479830448157,
      "loss": 2.0728,
      "step": 68770
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.9046532511711121,
      "learning_rate": 0.0004774915004954485,
      "loss": 2.2903,
      "step": 68771
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.757882833480835,
      "learning_rate": 0.0004774882026534164,
      "loss": 2.2113,
      "step": 68772
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7160758972167969,
      "learning_rate": 0.0004774849047783858,
      "loss": 2.21,
      "step": 68773
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.9330363273620605,
      "learning_rate": 0.0004774816068703575,
      "loss": 2.275,
      "step": 68774
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.958763599395752,
      "learning_rate": 0.00047747830892933197,
      "loss": 2.252,
      "step": 68775
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.8205330967903137,
      "learning_rate": 0.00047747501095530995,
      "loss": 2.4337,
      "step": 68776
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.8275771737098694,
      "learning_rate": 0.00047747171294829197,
      "loss": 2.1789,
      "step": 68777
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.9954178929328918,
      "learning_rate": 0.00047746841490827857,
      "loss": 2.0877,
      "step": 68778
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.8823317289352417,
      "learning_rate": 0.0004774651168352705,
      "loss": 2.2008,
      "step": 68779
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7672368288040161,
      "learning_rate": 0.00047746181872926823,
      "loss": 2.2567,
      "step": 68780
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.9751124978065491,
      "learning_rate": 0.00047745852059027255,
      "loss": 2.2514,
      "step": 68781
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.8524562120437622,
      "learning_rate": 0.00047745522241828404,
      "loss": 2.4407,
      "step": 68782
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.8636329770088196,
      "learning_rate": 0.00047745192421330316,
      "loss": 2.4403,
      "step": 68783
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.8499823212623596,
      "learning_rate": 0.00047744862597533054,
      "loss": 1.9567,
      "step": 68784
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7608966827392578,
      "learning_rate": 0.000477445327704367,
      "loss": 2.1815,
      "step": 68785
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7469852566719055,
      "learning_rate": 0.000477442029400413,
      "loss": 2.2357,
      "step": 68786
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7736268043518066,
      "learning_rate": 0.0004774387310634691,
      "loss": 2.1634,
      "step": 68787
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.8219138979911804,
      "learning_rate": 0.0004774354326935362,
      "loss": 2.0328,
      "step": 68788
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.8208918571472168,
      "learning_rate": 0.00047743213429061446,
      "loss": 2.3612,
      "step": 68789
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.8131241798400879,
      "learning_rate": 0.00047742883585470485,
      "loss": 2.0117,
      "step": 68790
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.737115740776062,
      "learning_rate": 0.0004774255373858079,
      "loss": 2.0731,
      "step": 68791
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.882593035697937,
      "learning_rate": 0.0004774222388839241,
      "loss": 2.2819,
      "step": 68792
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.9414351582527161,
      "learning_rate": 0.00047741894034905423,
      "loss": 1.9963,
      "step": 68793
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.772602379322052,
      "learning_rate": 0.00047741564178119885,
      "loss": 2.5808,
      "step": 68794
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7883850336074829,
      "learning_rate": 0.0004774123431803584,
      "loss": 2.2084,
      "step": 68795
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.9103336334228516,
      "learning_rate": 0.0004774090445465337,
      "loss": 2.0874,
      "step": 68796
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.8518499732017517,
      "learning_rate": 0.00047740574587972547,
      "loss": 2.5013,
      "step": 68797
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7504425048828125,
      "learning_rate": 0.000477402447179934,
      "loss": 2.3696,
      "step": 68798
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.8613386154174805,
      "learning_rate": 0.00047739914844716015,
      "loss": 2.2057,
      "step": 68799
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7672103047370911,
      "learning_rate": 0.0004773958496814044,
      "loss": 2.3624,
      "step": 68800
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.8569425344467163,
      "learning_rate": 0.00047739255088266743,
      "loss": 2.3308,
      "step": 68801
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7077314853668213,
      "learning_rate": 0.0004773892520509498,
      "loss": 2.3894,
      "step": 68802
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7766072750091553,
      "learning_rate": 0.00047738595318625224,
      "loss": 2.123,
      "step": 68803
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.799210786819458,
      "learning_rate": 0.0004773826542885752,
      "loss": 2.2495,
      "step": 68804
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7395730018615723,
      "learning_rate": 0.00047737935535791947,
      "loss": 2.2827,
      "step": 68805
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.8048513531684875,
      "learning_rate": 0.0004773760563942855,
      "loss": 2.1117,
      "step": 68806
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7636476755142212,
      "learning_rate": 0.000477372757397674,
      "loss": 2.1606,
      "step": 68807
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7457551956176758,
      "learning_rate": 0.00047736945836808557,
      "loss": 2.2611,
      "step": 68808
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.6687825322151184,
      "learning_rate": 0.00047736615930552077,
      "loss": 2.18,
      "step": 68809
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.8015771508216858,
      "learning_rate": 0.0004773628602099802,
      "loss": 2.2258,
      "step": 68810
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7529764175415039,
      "learning_rate": 0.0004773595610814646,
      "loss": 2.2798,
      "step": 68811
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.8178549408912659,
      "learning_rate": 0.00047735626191997455,
      "loss": 2.2735,
      "step": 68812
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7646709680557251,
      "learning_rate": 0.0004773529627255106,
      "loss": 2.0106,
      "step": 68813
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7537705898284912,
      "learning_rate": 0.00047734966349807334,
      "loss": 1.9246,
      "step": 68814
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7371792793273926,
      "learning_rate": 0.0004773463642376635,
      "loss": 2.1193,
      "step": 68815
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7473028302192688,
      "learning_rate": 0.0004773430649442816,
      "loss": 2.288,
      "step": 68816
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7324488162994385,
      "learning_rate": 0.0004773397656179282,
      "loss": 2.2281,
      "step": 68817
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7923789024353027,
      "learning_rate": 0.0004773364662586041,
      "loss": 2.1936,
      "step": 68818
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7737924456596375,
      "learning_rate": 0.00047733316686630965,
      "loss": 2.0318,
      "step": 68819
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7496907711029053,
      "learning_rate": 0.0004773298674410457,
      "loss": 1.9415,
      "step": 68820
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.6858909726142883,
      "learning_rate": 0.00047732656798281283,
      "loss": 2.1057,
      "step": 68821
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.8023247718811035,
      "learning_rate": 0.00047732326849161163,
      "loss": 2.3466,
      "step": 68822
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7383477091789246,
      "learning_rate": 0.00047731996896744255,
      "loss": 2.0312,
      "step": 68823
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7253422737121582,
      "learning_rate": 0.0004773166694103065,
      "loss": 2.0058,
      "step": 68824
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7531100511550903,
      "learning_rate": 0.00047731336982020375,
      "loss": 2.4569,
      "step": 68825
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7732794284820557,
      "learning_rate": 0.00047731007019713526,
      "loss": 2.4456,
      "step": 68826
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7079816460609436,
      "learning_rate": 0.00047730677054110145,
      "loss": 2.1152,
      "step": 68827
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7409703731536865,
      "learning_rate": 0.00047730347085210296,
      "loss": 2.1524,
      "step": 68828
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7266480922698975,
      "learning_rate": 0.00047730017113014033,
      "loss": 2.0992,
      "step": 68829
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.8595328330993652,
      "learning_rate": 0.0004772968713752142,
      "loss": 2.2356,
      "step": 68830
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7261162400245667,
      "learning_rate": 0.00047729357158732543,
      "loss": 2.273,
      "step": 68831
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.8163747787475586,
      "learning_rate": 0.00047729027176647435,
      "loss": 2.2966,
      "step": 68832
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.739546537399292,
      "learning_rate": 0.0004772869719126616,
      "loss": 2.0401,
      "step": 68833
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.8022855520248413,
      "learning_rate": 0.00047728367202588796,
      "loss": 2.3073,
      "step": 68834
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.986758291721344,
      "learning_rate": 0.0004772803721061539,
      "loss": 2.3649,
      "step": 68835
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7762566804885864,
      "learning_rate": 0.00047727707215346,
      "loss": 2.2787,
      "step": 68836
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.9033491611480713,
      "learning_rate": 0.00047727377216780713,
      "loss": 2.3457,
      "step": 68837
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7997462749481201,
      "learning_rate": 0.00047727047214919557,
      "loss": 2.5965,
      "step": 68838
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.8431262373924255,
      "learning_rate": 0.0004772671720976261,
      "loss": 2.2207,
      "step": 68839
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7400920391082764,
      "learning_rate": 0.0004772638720130994,
      "loss": 2.0257,
      "step": 68840
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.8520900011062622,
      "learning_rate": 0.0004772605718956158,
      "loss": 2.0866,
      "step": 68841
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.8810238838195801,
      "learning_rate": 0.00047725727174517625,
      "loss": 2.1166,
      "step": 68842
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7716989517211914,
      "learning_rate": 0.00047725397156178124,
      "loss": 2.1957,
      "step": 68843
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.8126449584960938,
      "learning_rate": 0.0004772506713454313,
      "loss": 2.0574,
      "step": 68844
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7073306441307068,
      "learning_rate": 0.0004772473710961272,
      "loss": 2.3438,
      "step": 68845
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7898633480072021,
      "learning_rate": 0.00047724407081386946,
      "loss": 2.083,
      "step": 68846
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.6953237056732178,
      "learning_rate": 0.0004772407704986587,
      "loss": 2.0373,
      "step": 68847
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7315313220024109,
      "learning_rate": 0.00047723747015049545,
      "loss": 2.3737,
      "step": 68848
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.817233145236969,
      "learning_rate": 0.0004772341697693805,
      "loss": 2.2039,
      "step": 68849
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.821712851524353,
      "learning_rate": 0.00047723086935531434,
      "loss": 2.5025,
      "step": 68850
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7489796876907349,
      "learning_rate": 0.00047722756890829763,
      "loss": 2.4413,
      "step": 68851
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.8685358166694641,
      "learning_rate": 0.00047722426842833094,
      "loss": 2.358,
      "step": 68852
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7426019906997681,
      "learning_rate": 0.0004772209679154149,
      "loss": 2.3719,
      "step": 68853
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7665213942527771,
      "learning_rate": 0.00047721766736955013,
      "loss": 2.3244,
      "step": 68854
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.741600751876831,
      "learning_rate": 0.0004772143667907374,
      "loss": 2.119,
      "step": 68855
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7445108294487,
      "learning_rate": 0.0004772110661789771,
      "loss": 2.0601,
      "step": 68856
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7466566562652588,
      "learning_rate": 0.00047720776553426976,
      "loss": 2.1744,
      "step": 68857
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7492733597755432,
      "learning_rate": 0.0004772044648566163,
      "loss": 2.4925,
      "step": 68858
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.8496918082237244,
      "learning_rate": 0.0004772011641460172,
      "loss": 2.3566,
      "step": 68859
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7395817041397095,
      "learning_rate": 0.000477197863402473,
      "loss": 2.1105,
      "step": 68860
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.9070324897766113,
      "learning_rate": 0.0004771945626259844,
      "loss": 2.3234,
      "step": 68861
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7760879993438721,
      "learning_rate": 0.00047719126181655196,
      "loss": 2.1542,
      "step": 68862
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7672494053840637,
      "learning_rate": 0.00047718796097417633,
      "loss": 2.4135,
      "step": 68863
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7720358371734619,
      "learning_rate": 0.0004771846600988581,
      "loss": 2.1093,
      "step": 68864
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.1094485521316528,
      "learning_rate": 0.0004771813591905979,
      "loss": 2.3845,
      "step": 68865
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7785565853118896,
      "learning_rate": 0.0004771780582493964,
      "loss": 2.1708,
      "step": 68866
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.1652820110321045,
      "learning_rate": 0.0004771747572752541,
      "loss": 2.0704,
      "step": 68867
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.9664647579193115,
      "learning_rate": 0.0004771714562681716,
      "loss": 2.3206,
      "step": 68868
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.8198718428611755,
      "learning_rate": 0.0004771681552281497,
      "loss": 2.1622,
      "step": 68869
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.2050344944000244,
      "learning_rate": 0.0004771648541551889,
      "loss": 2.2441,
      "step": 68870
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.8965773582458496,
      "learning_rate": 0.0004771615530492897,
      "loss": 2.5829,
      "step": 68871
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7805607318878174,
      "learning_rate": 0.00047715825191045293,
      "loss": 2.0738,
      "step": 68872
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.8196122646331787,
      "learning_rate": 0.00047715495073867906,
      "loss": 2.2846,
      "step": 68873
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.9950287342071533,
      "learning_rate": 0.0004771516495339687,
      "loss": 2.0444,
      "step": 68874
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7853500843048096,
      "learning_rate": 0.00047714834829632255,
      "loss": 2.2753,
      "step": 68875
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.8817980289459229,
      "learning_rate": 0.00047714504702574116,
      "loss": 2.2986,
      "step": 68876
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7521511316299438,
      "learning_rate": 0.0004771417457222252,
      "loss": 2.1326,
      "step": 68877
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.6671704053878784,
      "learning_rate": 0.00047713844438577514,
      "loss": 2.3353,
      "step": 68878
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7968968152999878,
      "learning_rate": 0.0004771351430163918,
      "loss": 2.1684,
      "step": 68879
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7417294383049011,
      "learning_rate": 0.0004771318416140757,
      "loss": 2.0575,
      "step": 68880
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7617126703262329,
      "learning_rate": 0.00047712854017882734,
      "loss": 2.126,
      "step": 68881
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.8169304132461548,
      "learning_rate": 0.00047712523871064756,
      "loss": 2.1801,
      "step": 68882
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7460412979125977,
      "learning_rate": 0.0004771219372095368,
      "loss": 2.0456,
      "step": 68883
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7965824007987976,
      "learning_rate": 0.00047711863567549576,
      "loss": 2.2798,
      "step": 68884
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7020305395126343,
      "learning_rate": 0.000477115334108525,
      "loss": 2.3225,
      "step": 68885
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7780981063842773,
      "learning_rate": 0.0004771120325086251,
      "loss": 2.2271,
      "step": 68886
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.8283862471580505,
      "learning_rate": 0.0004771087308757968,
      "loss": 2.1514,
      "step": 68887
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7423750758171082,
      "learning_rate": 0.00047710542921004063,
      "loss": 2.2697,
      "step": 68888
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.674369752407074,
      "learning_rate": 0.00047710212751135724,
      "loss": 2.1606,
      "step": 68889
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.782782256603241,
      "learning_rate": 0.0004770988257797471,
      "loss": 2.1838,
      "step": 68890
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.8918067216873169,
      "learning_rate": 0.0004770955240152111,
      "loss": 2.1679,
      "step": 68891
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.813904345035553,
      "learning_rate": 0.00047709222221774966,
      "loss": 2.4571,
      "step": 68892
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.8874533772468567,
      "learning_rate": 0.0004770889203873634,
      "loss": 2.1683,
      "step": 68893
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7548919916152954,
      "learning_rate": 0.00047708561852405297,
      "loss": 2.1182,
      "step": 68894
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.0048364400863647,
      "learning_rate": 0.000477082316627819,
      "loss": 2.2702,
      "step": 68895
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.9030835032463074,
      "learning_rate": 0.00047707901469866215,
      "loss": 2.2574,
      "step": 68896
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7974748611450195,
      "learning_rate": 0.00047707571273658284,
      "loss": 2.4403,
      "step": 68897
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.1803430318832397,
      "learning_rate": 0.00047707241074158185,
      "loss": 2.2652,
      "step": 68898
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.9929415583610535,
      "learning_rate": 0.00047706910871365976,
      "loss": 2.2035,
      "step": 68899
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7538739442825317,
      "learning_rate": 0.00047706580665281723,
      "loss": 2.321,
      "step": 68900
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.762220025062561,
      "learning_rate": 0.0004770625045590548,
      "loss": 2.2758,
      "step": 68901
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7388573288917542,
      "learning_rate": 0.00047705920243237306,
      "loss": 2.1244,
      "step": 68902
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.8920897245407104,
      "learning_rate": 0.0004770559002727727,
      "loss": 2.0105,
      "step": 68903
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.8016600012779236,
      "learning_rate": 0.0004770525980802544,
      "loss": 2.2058,
      "step": 68904
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.9010413289070129,
      "learning_rate": 0.00047704929585481857,
      "loss": 2.0132,
      "step": 68905
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.993024468421936,
      "learning_rate": 0.0004770459935964659,
      "loss": 2.3023,
      "step": 68906
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.926655650138855,
      "learning_rate": 0.0004770426913051971,
      "loss": 2.3681,
      "step": 68907
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7325774431228638,
      "learning_rate": 0.0004770393889810128,
      "loss": 2.3114,
      "step": 68908
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.8427385687828064,
      "learning_rate": 0.0004770360866239134,
      "loss": 2.1446,
      "step": 68909
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7394216060638428,
      "learning_rate": 0.00047703278423389967,
      "loss": 2.1177,
      "step": 68910
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7030991315841675,
      "learning_rate": 0.00047702948181097227,
      "loss": 2.1757,
      "step": 68911
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7076459527015686,
      "learning_rate": 0.0004770261793551317,
      "loss": 2.3435,
      "step": 68912
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7873291969299316,
      "learning_rate": 0.0004770228768663786,
      "loss": 2.4807,
      "step": 68913
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7321699857711792,
      "learning_rate": 0.0004770195743447137,
      "loss": 2.273,
      "step": 68914
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7930075526237488,
      "learning_rate": 0.00047701627179013746,
      "loss": 2.1129,
      "step": 68915
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.8103816509246826,
      "learning_rate": 0.00047701296920265056,
      "loss": 2.4223,
      "step": 68916
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7748136520385742,
      "learning_rate": 0.0004770096665822536,
      "loss": 2.1736,
      "step": 68917
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.8285121321678162,
      "learning_rate": 0.0004770063639289472,
      "loss": 1.9936,
      "step": 68918
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.748623788356781,
      "learning_rate": 0.000477003061242732,
      "loss": 2.1655,
      "step": 68919
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7279627323150635,
      "learning_rate": 0.00047699975852360854,
      "loss": 2.2597,
      "step": 68920
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7768440842628479,
      "learning_rate": 0.0004769964557715776,
      "loss": 1.9622,
      "step": 68921
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7884965538978577,
      "learning_rate": 0.00047699315298663957,
      "loss": 2.1747,
      "step": 68922
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.734215497970581,
      "learning_rate": 0.00047698985016879516,
      "loss": 2.2949,
      "step": 68923
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7430576682090759,
      "learning_rate": 0.000476986547318045,
      "loss": 2.2587,
      "step": 68924
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7537627220153809,
      "learning_rate": 0.0004769832444343897,
      "loss": 2.3421,
      "step": 68925
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7479965686798096,
      "learning_rate": 0.00047697994151782994,
      "loss": 2.3812,
      "step": 68926
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7329843044281006,
      "learning_rate": 0.0004769766385683662,
      "loss": 2.2903,
      "step": 68927
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.8220054507255554,
      "learning_rate": 0.00047697333558599926,
      "loss": 2.1821,
      "step": 68928
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.81868577003479,
      "learning_rate": 0.00047697003257072954,
      "loss": 2.3011,
      "step": 68929
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.8458403944969177,
      "learning_rate": 0.0004769667295225578,
      "loss": 2.2428,
      "step": 68930
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7568482756614685,
      "learning_rate": 0.0004769634264414846,
      "loss": 2.4571,
      "step": 68931
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7709290385246277,
      "learning_rate": 0.00047696012332751054,
      "loss": 2.1652,
      "step": 68932
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.8635187149047852,
      "learning_rate": 0.00047695682018063625,
      "loss": 2.3402,
      "step": 68933
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7554975748062134,
      "learning_rate": 0.0004769535170008623,
      "loss": 2.0491,
      "step": 68934
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.9218799471855164,
      "learning_rate": 0.00047695021378818943,
      "loss": 2.2431,
      "step": 68935
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.8200965523719788,
      "learning_rate": 0.0004769469105426182,
      "loss": 2.2193,
      "step": 68936
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7111980319023132,
      "learning_rate": 0.0004769436072641491,
      "loss": 2.2346,
      "step": 68937
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7218464016914368,
      "learning_rate": 0.0004769403039527829,
      "loss": 2.2018,
      "step": 68938
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7571782469749451,
      "learning_rate": 0.00047693700060852016,
      "loss": 1.9957,
      "step": 68939
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7826334238052368,
      "learning_rate": 0.0004769336972313615,
      "loss": 2.2349,
      "step": 68940
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.8164299130439758,
      "learning_rate": 0.0004769303938213074,
      "loss": 2.199,
      "step": 68941
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7938732504844666,
      "learning_rate": 0.00047692709037835875,
      "loss": 2.2531,
      "step": 68942
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7792803645133972,
      "learning_rate": 0.00047692378690251605,
      "loss": 2.1871,
      "step": 68943
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7519242167472839,
      "learning_rate": 0.00047692048339377974,
      "loss": 2.0723,
      "step": 68944
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.9708088040351868,
      "learning_rate": 0.00047691717985215063,
      "loss": 2.0333,
      "step": 68945
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7929518222808838,
      "learning_rate": 0.00047691387627762927,
      "loss": 2.3801,
      "step": 68946
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7465870380401611,
      "learning_rate": 0.0004769105726702163,
      "loss": 2.1676,
      "step": 68947
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7890515327453613,
      "learning_rate": 0.0004769072690299123,
      "loss": 2.1242,
      "step": 68948
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.8290640711784363,
      "learning_rate": 0.000476903965356718,
      "loss": 2.3334,
      "step": 68949
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.8302655220031738,
      "learning_rate": 0.00047690066165063374,
      "loss": 2.1508,
      "step": 68950
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7714608311653137,
      "learning_rate": 0.0004768973579116603,
      "loss": 2.1802,
      "step": 68951
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.8798049688339233,
      "learning_rate": 0.00047689405413979855,
      "loss": 2.0623,
      "step": 68952
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7500142455101013,
      "learning_rate": 0.00047689075033504865,
      "loss": 2.2558,
      "step": 68953
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.8418676257133484,
      "learning_rate": 0.0004768874464974114,
      "loss": 2.0872,
      "step": 68954
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7510036826133728,
      "learning_rate": 0.0004768841426268876,
      "loss": 2.1604,
      "step": 68955
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.0993082523345947,
      "learning_rate": 0.00047688083872347755,
      "loss": 2.085,
      "step": 68956
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.8624719381332397,
      "learning_rate": 0.00047687753478718205,
      "loss": 2.2498,
      "step": 68957
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.8712978959083557,
      "learning_rate": 0.00047687423081800166,
      "loss": 2.2031,
      "step": 68958
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7511482834815979,
      "learning_rate": 0.00047687092681593705,
      "loss": 2.2449,
      "step": 68959
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.9840708374977112,
      "learning_rate": 0.00047686762278098874,
      "loss": 2.1692,
      "step": 68960
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7481825351715088,
      "learning_rate": 0.00047686431871315745,
      "loss": 2.0979,
      "step": 68961
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.021660566329956,
      "learning_rate": 0.00047686101461244377,
      "loss": 2.404,
      "step": 68962
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7745944857597351,
      "learning_rate": 0.00047685771047884824,
      "loss": 2.2186,
      "step": 68963
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.8100194931030273,
      "learning_rate": 0.0004768544063123715,
      "loss": 2.1183,
      "step": 68964
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7362126708030701,
      "learning_rate": 0.0004768511021130143,
      "loss": 2.3454,
      "step": 68965
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.8326353430747986,
      "learning_rate": 0.00047684779788077704,
      "loss": 2.2202,
      "step": 68966
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7262767553329468,
      "learning_rate": 0.0004768444936156604,
      "loss": 2.2764,
      "step": 68967
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7154409885406494,
      "learning_rate": 0.0004768411893176652,
      "loss": 2.2086,
      "step": 68968
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.726702868938446,
      "learning_rate": 0.0004768378849867918,
      "loss": 2.3045,
      "step": 68969
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.8380650281906128,
      "learning_rate": 0.0004768345806230409,
      "loss": 1.9098,
      "step": 68970
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.8629409074783325,
      "learning_rate": 0.0004768312762264131,
      "loss": 2.3229,
      "step": 68971
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7578385472297668,
      "learning_rate": 0.000476827971796909,
      "loss": 2.1929,
      "step": 68972
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7395908236503601,
      "learning_rate": 0.0004768246673345292,
      "loss": 2.1069,
      "step": 68973
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7842415571212769,
      "learning_rate": 0.0004768213628392744,
      "loss": 2.0362,
      "step": 68974
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7456212043762207,
      "learning_rate": 0.0004768180583111453,
      "loss": 2.1091,
      "step": 68975
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.6871486902236938,
      "learning_rate": 0.0004768147537501422,
      "loss": 2.1954,
      "step": 68976
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.8091968297958374,
      "learning_rate": 0.00047681144915626605,
      "loss": 2.1991,
      "step": 68977
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.869431734085083,
      "learning_rate": 0.0004768081445295172,
      "loss": 2.1222,
      "step": 68978
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7695350646972656,
      "learning_rate": 0.0004768048398698964,
      "loss": 2.2114,
      "step": 68979
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.795268714427948,
      "learning_rate": 0.0004768015351774043,
      "loss": 2.2888,
      "step": 68980
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.8671104311943054,
      "learning_rate": 0.00047679823045204136,
      "loss": 2.1332,
      "step": 68981
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7356788516044617,
      "learning_rate": 0.00047679492569380845,
      "loss": 2.2642,
      "step": 68982
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7531042098999023,
      "learning_rate": 0.0004767916209027059,
      "loss": 2.096,
      "step": 68983
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.9336240291595459,
      "learning_rate": 0.0004767883160787345,
      "loss": 2.3553,
      "step": 68984
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.9184231162071228,
      "learning_rate": 0.0004767850112218947,
      "loss": 2.2445,
      "step": 68985
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7865809798240662,
      "learning_rate": 0.00047678170633218736,
      "loss": 2.2433,
      "step": 68986
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7988584637641907,
      "learning_rate": 0.00047677840140961293,
      "loss": 2.3052,
      "step": 68987
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.776413083076477,
      "learning_rate": 0.00047677509645417204,
      "loss": 2.369,
      "step": 68988
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7861906290054321,
      "learning_rate": 0.0004767717914658654,
      "loss": 2.487,
      "step": 68989
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7698037624359131,
      "learning_rate": 0.00047676848644469347,
      "loss": 2.2807,
      "step": 68990
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7607149481773376,
      "learning_rate": 0.00047676518139065687,
      "loss": 2.5074,
      "step": 68991
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.9194262623786926,
      "learning_rate": 0.0004767618763037564,
      "loss": 2.1304,
      "step": 68992
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7442206144332886,
      "learning_rate": 0.00047675857118399255,
      "loss": 2.1725,
      "step": 68993
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7565991878509521,
      "learning_rate": 0.0004767552660313659,
      "loss": 2.1228,
      "step": 68994
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.8622326850891113,
      "learning_rate": 0.0004767519608458771,
      "loss": 2.0294,
      "step": 68995
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.0490353107452393,
      "learning_rate": 0.0004767486556275269,
      "loss": 2.328,
      "step": 68996
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7497974634170532,
      "learning_rate": 0.0004767453503763157,
      "loss": 2.0562,
      "step": 68997
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.0529630184173584,
      "learning_rate": 0.00047674204509224417,
      "loss": 2.3893,
      "step": 68998
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.9452686309814453,
      "learning_rate": 0.000476738739775313,
      "loss": 2.3121,
      "step": 68999
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7707850933074951,
      "learning_rate": 0.00047673543442552273,
      "loss": 2.3957,
      "step": 69000
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7564367055892944,
      "learning_rate": 0.0004767321290428741,
      "loss": 2.0729,
      "step": 69001
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.0429741144180298,
      "learning_rate": 0.00047672882362736753,
      "loss": 2.3815,
      "step": 69002
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7828831076622009,
      "learning_rate": 0.0004767255181790038,
      "loss": 2.1228,
      "step": 69003
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.8640187382698059,
      "learning_rate": 0.00047672221269778344,
      "loss": 2.2381,
      "step": 69004
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.9372420310974121,
      "learning_rate": 0.000476718907183707,
      "loss": 2.2221,
      "step": 69005
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.0588241815567017,
      "learning_rate": 0.00047671560163677524,
      "loss": 2.2547,
      "step": 69006
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7834380865097046,
      "learning_rate": 0.00047671229605698883,
      "loss": 2.3523,
      "step": 69007
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7534972429275513,
      "learning_rate": 0.00047670899044434807,
      "loss": 2.1724,
      "step": 69008
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.0133241415023804,
      "learning_rate": 0.00047670568479885386,
      "loss": 2.079,
      "step": 69009
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.0839279890060425,
      "learning_rate": 0.0004767023791205068,
      "loss": 2.2408,
      "step": 69010
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7692042589187622,
      "learning_rate": 0.0004766990734093073,
      "loss": 2.3253,
      "step": 69011
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7767491340637207,
      "learning_rate": 0.0004766957676652562,
      "loss": 2.3936,
      "step": 69012
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.9529640674591064,
      "learning_rate": 0.000476692461888354,
      "loss": 2.1673,
      "step": 69013
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.0480117797851562,
      "learning_rate": 0.0004766891560786013,
      "loss": 2.3122,
      "step": 69014
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7991084456443787,
      "learning_rate": 0.0004766858502359988,
      "loss": 2.2883,
      "step": 69015
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7889454960823059,
      "learning_rate": 0.00047668254436054706,
      "loss": 2.1434,
      "step": 69016
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7874859571456909,
      "learning_rate": 0.0004766792384522467,
      "loss": 2.2852,
      "step": 69017
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.0307902097702026,
      "learning_rate": 0.00047667593251109826,
      "loss": 2.2556,
      "step": 69018
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.8200070261955261,
      "learning_rate": 0.0004766726265371025,
      "loss": 2.3132,
      "step": 69019
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7942019701004028,
      "learning_rate": 0.0004766693205302599,
      "loss": 2.3453,
      "step": 69020
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.9664660096168518,
      "learning_rate": 0.0004766660144905713,
      "loss": 2.35,
      "step": 69021
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.8430874943733215,
      "learning_rate": 0.000476662708418037,
      "loss": 2.3567,
      "step": 69022
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.8538908362388611,
      "learning_rate": 0.0004766594023126578,
      "loss": 2.3977,
      "step": 69023
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.785729169845581,
      "learning_rate": 0.00047665609617443435,
      "loss": 2.2002,
      "step": 69024
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7416203618049622,
      "learning_rate": 0.0004766527900033671,
      "loss": 2.1024,
      "step": 69025
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.8535917401313782,
      "learning_rate": 0.00047664948379945676,
      "loss": 2.1721,
      "step": 69026
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.8536182641983032,
      "learning_rate": 0.000476646177562704,
      "loss": 2.1164,
      "step": 69027
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7939031720161438,
      "learning_rate": 0.0004766428712931094,
      "loss": 2.1952,
      "step": 69028
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.8925206661224365,
      "learning_rate": 0.00047663956499067355,
      "loss": 2.3261,
      "step": 69029
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7922170758247375,
      "learning_rate": 0.000476636258655397,
      "loss": 2.3098,
      "step": 69030
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.8446798324584961,
      "learning_rate": 0.00047663295228728055,
      "loss": 2.2746,
      "step": 69031
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7845187187194824,
      "learning_rate": 0.0004766296458863246,
      "loss": 2.103,
      "step": 69032
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7458149790763855,
      "learning_rate": 0.0004766263394525299,
      "loss": 2.2884,
      "step": 69033
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7295623421669006,
      "learning_rate": 0.00047662303298589697,
      "loss": 2.1109,
      "step": 69034
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7560625076293945,
      "learning_rate": 0.0004766197264864266,
      "loss": 2.2541,
      "step": 69035
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7738764882087708,
      "learning_rate": 0.0004766164199541192,
      "loss": 2.2665,
      "step": 69036
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7623389363288879,
      "learning_rate": 0.0004766131133889755,
      "loss": 2.1678,
      "step": 69037
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.9426952600479126,
      "learning_rate": 0.0004766098067909961,
      "loss": 2.0893,
      "step": 69038
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.8263213038444519,
      "learning_rate": 0.0004766065001601816,
      "loss": 2.1615,
      "step": 69039
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7704329490661621,
      "learning_rate": 0.0004766031934965326,
      "loss": 2.1945,
      "step": 69040
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.9580467939376831,
      "learning_rate": 0.00047659988680004976,
      "loss": 2.3188,
      "step": 69041
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.8965041041374207,
      "learning_rate": 0.00047659658007073366,
      "loss": 2.14,
      "step": 69042
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7920128703117371,
      "learning_rate": 0.0004765932733085849,
      "loss": 1.9993,
      "step": 69043
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.9598477482795715,
      "learning_rate": 0.0004765899665136041,
      "loss": 2.1342,
      "step": 69044
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.9909403920173645,
      "learning_rate": 0.000476586659685792,
      "loss": 2.3216,
      "step": 69045
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7195165753364563,
      "learning_rate": 0.00047658335282514897,
      "loss": 2.4021,
      "step": 69046
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.8464258909225464,
      "learning_rate": 0.0004765800459316759,
      "loss": 2.3471,
      "step": 69047
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.9336034655570984,
      "learning_rate": 0.0004765767390053731,
      "loss": 2.1755,
      "step": 69048
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7768262624740601,
      "learning_rate": 0.0004765734320462415,
      "loss": 2.0989,
      "step": 69049
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7186631560325623,
      "learning_rate": 0.0004765701250542815,
      "loss": 2.2421,
      "step": 69050
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.8117210865020752,
      "learning_rate": 0.00047656681802949374,
      "loss": 2.2211,
      "step": 69051
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7890861630439758,
      "learning_rate": 0.0004765635109718789,
      "loss": 2.3111,
      "step": 69052
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7507667541503906,
      "learning_rate": 0.00047656020388143767,
      "loss": 2.1775,
      "step": 69053
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7913408279418945,
      "learning_rate": 0.0004765568967581705,
      "loss": 2.3748,
      "step": 69054
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.6999505758285522,
      "learning_rate": 0.000476553589602078,
      "loss": 2.1396,
      "step": 69055
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7634307742118835,
      "learning_rate": 0.00047655028241316095,
      "loss": 2.5282,
      "step": 69056
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7794861793518066,
      "learning_rate": 0.0004765469751914198,
      "loss": 2.3074,
      "step": 69057
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7411316633224487,
      "learning_rate": 0.0004765436679368553,
      "loss": 2.2323,
      "step": 69058
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7753491997718811,
      "learning_rate": 0.000476540360649468,
      "loss": 2.2096,
      "step": 69059
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.716771125793457,
      "learning_rate": 0.0004765370533292584,
      "loss": 2.3062,
      "step": 69060
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7285212278366089,
      "learning_rate": 0.0004765337459762273,
      "loss": 2.103,
      "step": 69061
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7814971804618835,
      "learning_rate": 0.0004765304385903753,
      "loss": 2.0693,
      "step": 69062
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7881216406822205,
      "learning_rate": 0.0004765271311717029,
      "loss": 2.3226,
      "step": 69063
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7119127511978149,
      "learning_rate": 0.00047652382372021077,
      "loss": 2.2823,
      "step": 69064
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7399626970291138,
      "learning_rate": 0.0004765205162358996,
      "loss": 2.1701,
      "step": 69065
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.8461043238639832,
      "learning_rate": 0.00047651720871876984,
      "loss": 2.3509,
      "step": 69066
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.8285623788833618,
      "learning_rate": 0.0004765139011688222,
      "loss": 2.2333,
      "step": 69067
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7665231823921204,
      "learning_rate": 0.0004765105935860574,
      "loss": 2.108,
      "step": 69068
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7512332201004028,
      "learning_rate": 0.00047650728597047585,
      "loss": 2.2496,
      "step": 69069
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7373296618461609,
      "learning_rate": 0.00047650397832207827,
      "loss": 2.3526,
      "step": 69070
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7103321552276611,
      "learning_rate": 0.00047650067064086536,
      "loss": 2.2353,
      "step": 69071
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.8840531706809998,
      "learning_rate": 0.00047649736292683755,
      "loss": 2.2875,
      "step": 69072
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.741682767868042,
      "learning_rate": 0.00047649405517999555,
      "loss": 2.1213,
      "step": 69073
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7718570232391357,
      "learning_rate": 0.00047649074740034006,
      "loss": 2.1319,
      "step": 69074
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7326745390892029,
      "learning_rate": 0.00047648743958787147,
      "loss": 2.3337,
      "step": 69075
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.8580151796340942,
      "learning_rate": 0.0004764841317425907,
      "loss": 2.1481,
      "step": 69076
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7876178622245789,
      "learning_rate": 0.00047648082386449804,
      "loss": 2.4124,
      "step": 69077
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7054956555366516,
      "learning_rate": 0.00047647751595359435,
      "loss": 2.1619,
      "step": 69078
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.782005250453949,
      "learning_rate": 0.0004764742080098801,
      "loss": 2.078,
      "step": 69079
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.743398904800415,
      "learning_rate": 0.0004764709000333561,
      "loss": 2.1283,
      "step": 69080
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.8393172025680542,
      "learning_rate": 0.0004764675920240227,
      "loss": 2.1664,
      "step": 69081
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7700151801109314,
      "learning_rate": 0.00047646428398188067,
      "loss": 2.176,
      "step": 69082
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.8181211948394775,
      "learning_rate": 0.0004764609759069306,
      "loss": 2.181,
      "step": 69083
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.8533318638801575,
      "learning_rate": 0.0004764576677991731,
      "loss": 2.2234,
      "step": 69084
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7253773212432861,
      "learning_rate": 0.00047645435965860885,
      "loss": 2.2476,
      "step": 69085
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.771262526512146,
      "learning_rate": 0.0004764510514852384,
      "loss": 2.4081,
      "step": 69086
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.8379756808280945,
      "learning_rate": 0.0004764477432790623,
      "loss": 2.0126,
      "step": 69087
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.79892498254776,
      "learning_rate": 0.0004764444350400812,
      "loss": 2.3723,
      "step": 69088
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7403934001922607,
      "learning_rate": 0.00047644112676829587,
      "loss": 2.2752,
      "step": 69089
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.0341490507125854,
      "learning_rate": 0.00047643781846370666,
      "loss": 2.2493,
      "step": 69090
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.8200328350067139,
      "learning_rate": 0.0004764345101263145,
      "loss": 2.1339,
      "step": 69091
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7605202794075012,
      "learning_rate": 0.0004764312017561197,
      "loss": 2.1527,
      "step": 69092
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7492040395736694,
      "learning_rate": 0.00047642789335312306,
      "loss": 1.9951,
      "step": 69093
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.6855273246765137,
      "learning_rate": 0.0004764245849173251,
      "loss": 2.3846,
      "step": 69094
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7561194896697998,
      "learning_rate": 0.00047642127644872647,
      "loss": 2.1932,
      "step": 69095
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7201937437057495,
      "learning_rate": 0.00047641796794732784,
      "loss": 2.0632,
      "step": 69096
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7977197170257568,
      "learning_rate": 0.00047641465941312977,
      "loss": 2.1525,
      "step": 69097
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.8464465737342834,
      "learning_rate": 0.00047641135084613284,
      "loss": 2.2927,
      "step": 69098
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7850370407104492,
      "learning_rate": 0.0004764080422463379,
      "loss": 2.3125,
      "step": 69099
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.8521201610565186,
      "learning_rate": 0.0004764047336137452,
      "loss": 2.2547,
      "step": 69100
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.8197380900382996,
      "learning_rate": 0.0004764014249483556,
      "loss": 2.2515,
      "step": 69101
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.8426285982131958,
      "learning_rate": 0.0004763981162501696,
      "loss": 2.1446,
      "step": 69102
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7501193881034851,
      "learning_rate": 0.00047639480751918785,
      "loss": 2.345,
      "step": 69103
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7322404980659485,
      "learning_rate": 0.00047639149875541105,
      "loss": 1.8369,
      "step": 69104
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.8595519065856934,
      "learning_rate": 0.0004763881899588396,
      "loss": 2.3148,
      "step": 69105
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7842198014259338,
      "learning_rate": 0.0004763848811294744,
      "loss": 2.2902,
      "step": 69106
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.8030200600624084,
      "learning_rate": 0.00047638157226731584,
      "loss": 2.3345,
      "step": 69107
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7655899524688721,
      "learning_rate": 0.0004763782633723647,
      "loss": 2.3933,
      "step": 69108
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7409952282905579,
      "learning_rate": 0.00047637495444462143,
      "loss": 2.2792,
      "step": 69109
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7732639312744141,
      "learning_rate": 0.00047637164548408676,
      "loss": 1.8924,
      "step": 69110
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7746474146842957,
      "learning_rate": 0.00047636833649076127,
      "loss": 2.4833,
      "step": 69111
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.781444251537323,
      "learning_rate": 0.0004763650274646455,
      "loss": 2.3153,
      "step": 69112
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7703086733818054,
      "learning_rate": 0.00047636171840574024,
      "loss": 2.0429,
      "step": 69113
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7301254868507385,
      "learning_rate": 0.00047635840931404595,
      "loss": 2.2774,
      "step": 69114
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.8208679556846619,
      "learning_rate": 0.0004763551001895634,
      "loss": 2.2746,
      "step": 69115
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.808660089969635,
      "learning_rate": 0.0004763517910322931,
      "loss": 2.2113,
      "step": 69116
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.832743227481842,
      "learning_rate": 0.00047634848184223557,
      "loss": 2.1648,
      "step": 69117
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7585814595222473,
      "learning_rate": 0.0004763451726193915,
      "loss": 2.0425,
      "step": 69118
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7564107179641724,
      "learning_rate": 0.0004763418633637617,
      "loss": 2.181,
      "step": 69119
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.796527624130249,
      "learning_rate": 0.0004763385540753465,
      "loss": 2.2137,
      "step": 69120
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.8079319596290588,
      "learning_rate": 0.0004763352447541466,
      "loss": 2.2984,
      "step": 69121
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7874399423599243,
      "learning_rate": 0.00047633193540016274,
      "loss": 2.4425,
      "step": 69122
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7511581182479858,
      "learning_rate": 0.00047632862601339546,
      "loss": 2.3615,
      "step": 69123
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7664686441421509,
      "learning_rate": 0.0004763253165938453,
      "loss": 2.3061,
      "step": 69124
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7394794225692749,
      "learning_rate": 0.00047632200714151286,
      "loss": 1.9563,
      "step": 69125
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7414323091506958,
      "learning_rate": 0.000476318697656399,
      "loss": 2.3401,
      "step": 69126
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7620668411254883,
      "learning_rate": 0.0004763153881385041,
      "loss": 1.9977,
      "step": 69127
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7082740664482117,
      "learning_rate": 0.00047631207858782874,
      "loss": 2.1269,
      "step": 69128
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7802744507789612,
      "learning_rate": 0.00047630876900437385,
      "loss": 2.2651,
      "step": 69129
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7239210605621338,
      "learning_rate": 0.0004763054593881396,
      "loss": 2.3806,
      "step": 69130
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7600970268249512,
      "learning_rate": 0.000476302149739127,
      "loss": 2.3666,
      "step": 69131
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7237907648086548,
      "learning_rate": 0.0004762988400573364,
      "loss": 2.1919,
      "step": 69132
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7553574442863464,
      "learning_rate": 0.00047629553034276857,
      "loss": 2.4894,
      "step": 69133
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7700140476226807,
      "learning_rate": 0.000476292220595424,
      "loss": 2.2029,
      "step": 69134
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7620927095413208,
      "learning_rate": 0.0004762889108153035,
      "loss": 2.2604,
      "step": 69135
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7800168991088867,
      "learning_rate": 0.00047628560100240747,
      "loss": 2.2909,
      "step": 69136
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7791542410850525,
      "learning_rate": 0.0004762822911567367,
      "loss": 2.4827,
      "step": 69137
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7609661817550659,
      "learning_rate": 0.00047627898127829165,
      "loss": 2.2772,
      "step": 69138
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.8442913889884949,
      "learning_rate": 0.0004762756713670731,
      "loss": 2.1487,
      "step": 69139
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.8830441236495972,
      "learning_rate": 0.0004762723614230815,
      "loss": 2.1092,
      "step": 69140
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7797999978065491,
      "learning_rate": 0.0004762690514463175,
      "loss": 2.1918,
      "step": 69141
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7148115038871765,
      "learning_rate": 0.0004762657414367819,
      "loss": 2.3174,
      "step": 69142
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7879059314727783,
      "learning_rate": 0.000476262431394475,
      "loss": 2.3565,
      "step": 69143
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7766703367233276,
      "learning_rate": 0.0004762591213193977,
      "loss": 2.4383,
      "step": 69144
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7955247759819031,
      "learning_rate": 0.0004762558112115505,
      "loss": 2.216,
      "step": 69145
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7445776462554932,
      "learning_rate": 0.00047625250107093396,
      "loss": 2.1319,
      "step": 69146
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7874758243560791,
      "learning_rate": 0.00047624919089754884,
      "loss": 2.2183,
      "step": 69147
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7607524991035461,
      "learning_rate": 0.0004762458806913956,
      "loss": 2.2722,
      "step": 69148
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7658771276473999,
      "learning_rate": 0.00047624257045247485,
      "loss": 2.2884,
      "step": 69149
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7468937635421753,
      "learning_rate": 0.00047623926018078744,
      "loss": 2.0812,
      "step": 69150
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7854958772659302,
      "learning_rate": 0.0004762359498763337,
      "loss": 2.3421,
      "step": 69151
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7670445442199707,
      "learning_rate": 0.0004762326395391144,
      "loss": 2.3499,
      "step": 69152
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7957674264907837,
      "learning_rate": 0.00047622932916913016,
      "loss": 1.9722,
      "step": 69153
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.8316775560379028,
      "learning_rate": 0.0004762260187663816,
      "loss": 2.1473,
      "step": 69154
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7499967217445374,
      "learning_rate": 0.00047622270833086916,
      "loss": 2.1423,
      "step": 69155
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7638242244720459,
      "learning_rate": 0.00047621939786259373,
      "loss": 2.0756,
      "step": 69156
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7591654062271118,
      "learning_rate": 0.00047621608736155577,
      "loss": 2.1021,
      "step": 69157
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.8214125037193298,
      "learning_rate": 0.0004762127768277558,
      "loss": 2.2081,
      "step": 69158
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7529749274253845,
      "learning_rate": 0.00047620946626119465,
      "loss": 2.0698,
      "step": 69159
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.769897997379303,
      "learning_rate": 0.0004762061556618728,
      "loss": 2.1588,
      "step": 69160
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.6807843446731567,
      "learning_rate": 0.00047620284502979094,
      "loss": 1.9693,
      "step": 69161
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7112728357315063,
      "learning_rate": 0.0004761995343649496,
      "loss": 2.3242,
      "step": 69162
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7349686622619629,
      "learning_rate": 0.0004761962236673494,
      "loss": 2.0576,
      "step": 69163
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7034079432487488,
      "learning_rate": 0.0004761929129369911,
      "loss": 2.1281,
      "step": 69164
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7569472789764404,
      "learning_rate": 0.0004761896021738752,
      "loss": 2.2969,
      "step": 69165
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7622421383857727,
      "learning_rate": 0.00047618629137800226,
      "loss": 2.3258,
      "step": 69166
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7266470193862915,
      "learning_rate": 0.00047618298054937294,
      "loss": 2.0959,
      "step": 69167
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.797282338142395,
      "learning_rate": 0.000476179669687988,
      "loss": 2.3471,
      "step": 69168
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7797077298164368,
      "learning_rate": 0.0004761763587938478,
      "loss": 2.2035,
      "step": 69169
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7439626455307007,
      "learning_rate": 0.00047617304786695317,
      "loss": 2.3332,
      "step": 69170
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7345790266990662,
      "learning_rate": 0.0004761697369073046,
      "loss": 2.4375,
      "step": 69171
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.730714738368988,
      "learning_rate": 0.0004761664259149029,
      "loss": 2.4794,
      "step": 69172
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7441417574882507,
      "learning_rate": 0.00047616311488974826,
      "loss": 2.281,
      "step": 69173
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7485173344612122,
      "learning_rate": 0.0004761598038318418,
      "loss": 2.1825,
      "step": 69174
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.732595682144165,
      "learning_rate": 0.00047615649274118383,
      "loss": 2.0989,
      "step": 69175
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7626912593841553,
      "learning_rate": 0.00047615318161777507,
      "loss": 2.3189,
      "step": 69176
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.747122585773468,
      "learning_rate": 0.0004761498704616161,
      "loss": 2.3784,
      "step": 69177
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.71399986743927,
      "learning_rate": 0.0004761465592727075,
      "loss": 1.9891,
      "step": 69178
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.8180071711540222,
      "learning_rate": 0.0004761432480510499,
      "loss": 2.0588,
      "step": 69179
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.8512598872184753,
      "learning_rate": 0.000476139936796644,
      "loss": 2.3552,
      "step": 69180
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7455071210861206,
      "learning_rate": 0.00047613662550949045,
      "loss": 2.085,
      "step": 69181
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7607568502426147,
      "learning_rate": 0.00047613331418958966,
      "loss": 2.3392,
      "step": 69182
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7396353483200073,
      "learning_rate": 0.00047613000283694233,
      "loss": 2.2542,
      "step": 69183
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.8321921229362488,
      "learning_rate": 0.0004761266914515493,
      "loss": 2.4155,
      "step": 69184
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.8692451119422913,
      "learning_rate": 0.0004761233800334108,
      "loss": 2.0166,
      "step": 69185
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7816885113716125,
      "learning_rate": 0.0004761200685825277,
      "loss": 2.1857,
      "step": 69186
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7288305759429932,
      "learning_rate": 0.00047611675709890056,
      "loss": 2.3512,
      "step": 69187
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7245851755142212,
      "learning_rate": 0.00047611344558253,
      "loss": 2.2188,
      "step": 69188
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7724142670631409,
      "learning_rate": 0.0004761101340334166,
      "loss": 2.2067,
      "step": 69189
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.8253039717674255,
      "learning_rate": 0.000476106822451561,
      "loss": 2.2517,
      "step": 69190
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.8330710530281067,
      "learning_rate": 0.00047610351083696384,
      "loss": 2.0563,
      "step": 69191
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7506772875785828,
      "learning_rate": 0.00047610019918962565,
      "loss": 2.1844,
      "step": 69192
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.8094878196716309,
      "learning_rate": 0.00047609688750954726,
      "loss": 2.1344,
      "step": 69193
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.8019325137138367,
      "learning_rate": 0.000476093575796729,
      "loss": 2.4048,
      "step": 69194
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7951676845550537,
      "learning_rate": 0.0004760902640511717,
      "loss": 2.0883,
      "step": 69195
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.9455103874206543,
      "learning_rate": 0.0004760869522728758,
      "loss": 2.2336,
      "step": 69196
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7995693683624268,
      "learning_rate": 0.00047608364046184206,
      "loss": 2.1614,
      "step": 69197
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.8306928873062134,
      "learning_rate": 0.00047608032861807105,
      "loss": 2.2237,
      "step": 69198
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.8579022884368896,
      "learning_rate": 0.0004760770167415634,
      "loss": 2.1833,
      "step": 69199
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.8759551644325256,
      "learning_rate": 0.00047607370483231966,
      "loss": 1.9731,
      "step": 69200
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.8175527453422546,
      "learning_rate": 0.00047607039289034053,
      "loss": 2.4651,
      "step": 69201
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.8132516741752625,
      "learning_rate": 0.0004760670809156267,
      "loss": 2.2949,
      "step": 69202
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.9297003746032715,
      "learning_rate": 0.0004760637689081785,
      "loss": 2.2769,
      "step": 69203
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.9141098260879517,
      "learning_rate": 0.00047606045686799677,
      "loss": 2.1098,
      "step": 69204
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.8183491826057434,
      "learning_rate": 0.0004760571447950821,
      "loss": 2.081,
      "step": 69205
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.0008349418640137,
      "learning_rate": 0.00047605383268943514,
      "loss": 1.8973,
      "step": 69206
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.8512921929359436,
      "learning_rate": 0.00047605052055105636,
      "loss": 2.3834,
      "step": 69207
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7799148559570312,
      "learning_rate": 0.00047604720837994644,
      "loss": 2.2895,
      "step": 69208
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.877852737903595,
      "learning_rate": 0.0004760438961761061,
      "loss": 2.0739,
      "step": 69209
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.9732744097709656,
      "learning_rate": 0.00047604058393953583,
      "loss": 2.2507,
      "step": 69210
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7960482239723206,
      "learning_rate": 0.0004760372716702363,
      "loss": 2.3094,
      "step": 69211
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7946471571922302,
      "learning_rate": 0.0004760339593682082,
      "loss": 2.0799,
      "step": 69212
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.8237022757530212,
      "learning_rate": 0.000476030647033452,
      "loss": 2.3682,
      "step": 69213
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.826888382434845,
      "learning_rate": 0.00047602733466596835,
      "loss": 2.2607,
      "step": 69214
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7822076678276062,
      "learning_rate": 0.00047602402226575794,
      "loss": 2.018,
      "step": 69215
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7471960783004761,
      "learning_rate": 0.0004760207098328213,
      "loss": 2.0749,
      "step": 69216
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7854103446006775,
      "learning_rate": 0.0004760173973671591,
      "loss": 2.1744,
      "step": 69217
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.877281665802002,
      "learning_rate": 0.000476014084868772,
      "loss": 2.2997,
      "step": 69218
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7475138902664185,
      "learning_rate": 0.0004760107723376604,
      "loss": 2.1336,
      "step": 69219
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7593721151351929,
      "learning_rate": 0.00047600745977382527,
      "loss": 2.0499,
      "step": 69220
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.800303041934967,
      "learning_rate": 0.00047600414717726693,
      "loss": 2.0676,
      "step": 69221
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7553446292877197,
      "learning_rate": 0.0004760008345479861,
      "loss": 2.2248,
      "step": 69222
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7854269742965698,
      "learning_rate": 0.0004759975218859835,
      "loss": 2.3404,
      "step": 69223
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7654510736465454,
      "learning_rate": 0.0004759942091912595,
      "loss": 2.3074,
      "step": 69224
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7236751317977905,
      "learning_rate": 0.0004759908964638149,
      "loss": 2.2303,
      "step": 69225
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7981859445571899,
      "learning_rate": 0.0004759875837036503,
      "loss": 2.3941,
      "step": 69226
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7929283380508423,
      "learning_rate": 0.0004759842709107662,
      "loss": 2.3101,
      "step": 69227
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.8238526582717896,
      "learning_rate": 0.0004759809580851634,
      "loss": 2.2504,
      "step": 69228
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.6880503296852112,
      "learning_rate": 0.0004759776452268424,
      "loss": 2.0384,
      "step": 69229
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.9664967656135559,
      "learning_rate": 0.0004759743323358038,
      "loss": 2.0034,
      "step": 69230
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.705108642578125,
      "learning_rate": 0.0004759710194120482,
      "loss": 2.2115,
      "step": 69231
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7879148721694946,
      "learning_rate": 0.0004759677064555763,
      "loss": 2.149,
      "step": 69232
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.0264863967895508,
      "learning_rate": 0.0004759643934663887,
      "loss": 2.1066,
      "step": 69233
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.8308941721916199,
      "learning_rate": 0.00047596108044448604,
      "loss": 2.067,
      "step": 69234
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.8107117414474487,
      "learning_rate": 0.0004759577673898689,
      "loss": 2.3394,
      "step": 69235
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.757719874382019,
      "learning_rate": 0.00047595445430253787,
      "loss": 2.2988,
      "step": 69236
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.9073411226272583,
      "learning_rate": 0.00047595114118249355,
      "loss": 2.2276,
      "step": 69237
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.9266239404678345,
      "learning_rate": 0.0004759478280297366,
      "loss": 2.1845,
      "step": 69238
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7775816321372986,
      "learning_rate": 0.00047594451484426765,
      "loss": 2.255,
      "step": 69239
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7999153733253479,
      "learning_rate": 0.00047594120162608726,
      "loss": 2.2535,
      "step": 69240
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.9166380763053894,
      "learning_rate": 0.00047593788837519606,
      "loss": 2.1538,
      "step": 69241
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.8682429194450378,
      "learning_rate": 0.00047593457509159475,
      "loss": 2.3307,
      "step": 69242
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7605297565460205,
      "learning_rate": 0.0004759312617752838,
      "loss": 2.123,
      "step": 69243
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7915493845939636,
      "learning_rate": 0.000475927948426264,
      "loss": 2.1788,
      "step": 69244
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.0898001194000244,
      "learning_rate": 0.0004759246350445358,
      "loss": 2.2933,
      "step": 69245
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7433757781982422,
      "learning_rate": 0.0004759213216301,
      "loss": 2.2919,
      "step": 69246
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.8545650839805603,
      "learning_rate": 0.000475918008182957,
      "loss": 2.2394,
      "step": 69247
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.8453707098960876,
      "learning_rate": 0.00047591469470310756,
      "loss": 2.3002,
      "step": 69248
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7335337400436401,
      "learning_rate": 0.0004759113811905523,
      "loss": 2.2277,
      "step": 69249
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7844675779342651,
      "learning_rate": 0.00047590806764529175,
      "loss": 2.1892,
      "step": 69250
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.782812774181366,
      "learning_rate": 0.0004759047540673265,
      "loss": 2.1171,
      "step": 69251
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7323024272918701,
      "learning_rate": 0.00047590144045665735,
      "loss": 2.2153,
      "step": 69252
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.8049222230911255,
      "learning_rate": 0.0004758981268132847,
      "loss": 2.2528,
      "step": 69253
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7492417693138123,
      "learning_rate": 0.0004758948131372093,
      "loss": 2.2823,
      "step": 69254
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.804776668548584,
      "learning_rate": 0.0004758914994284318,
      "loss": 2.1765,
      "step": 69255
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.9421930313110352,
      "learning_rate": 0.0004758881856869527,
      "loss": 2.1,
      "step": 69256
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.749131441116333,
      "learning_rate": 0.00047588487191277265,
      "loss": 2.4922,
      "step": 69257
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.732579231262207,
      "learning_rate": 0.0004758815581058923,
      "loss": 2.2573,
      "step": 69258
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7542674541473389,
      "learning_rate": 0.00047587824426631225,
      "loss": 2.2244,
      "step": 69259
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.8678618669509888,
      "learning_rate": 0.00047587493039403313,
      "loss": 2.1984,
      "step": 69260
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7037982940673828,
      "learning_rate": 0.0004758716164890555,
      "loss": 2.0275,
      "step": 69261
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7663923501968384,
      "learning_rate": 0.00047586830255138006,
      "loss": 2.2722,
      "step": 69262
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.9137545824050903,
      "learning_rate": 0.0004758649885810074,
      "loss": 2.2243,
      "step": 69263
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.990896999835968,
      "learning_rate": 0.000475861674577938,
      "loss": 2.1569,
      "step": 69264
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.9044774174690247,
      "learning_rate": 0.0004758583605421727,
      "loss": 1.9148,
      "step": 69265
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.9104448556900024,
      "learning_rate": 0.000475855046473712,
      "loss": 2.2326,
      "step": 69266
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.8485966324806213,
      "learning_rate": 0.0004758517323725565,
      "loss": 2.1399,
      "step": 69267
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.6893925070762634,
      "learning_rate": 0.0004758484182387068,
      "loss": 2.1484,
      "step": 69268
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.8085911870002747,
      "learning_rate": 0.00047584510407216367,
      "loss": 2.102,
      "step": 69269
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7847681641578674,
      "learning_rate": 0.0004758417898729276,
      "loss": 2.3429,
      "step": 69270
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.8208966255187988,
      "learning_rate": 0.00047583847564099916,
      "loss": 2.2102,
      "step": 69271
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7306315898895264,
      "learning_rate": 0.0004758351613763791,
      "loss": 2.4334,
      "step": 69272
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.8473891615867615,
      "learning_rate": 0.00047583184707906793,
      "loss": 2.2568,
      "step": 69273
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.8160462379455566,
      "learning_rate": 0.00047582853274906624,
      "loss": 2.0327,
      "step": 69274
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.8131062388420105,
      "learning_rate": 0.00047582521838637477,
      "loss": 2.4466,
      "step": 69275
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7192859053611755,
      "learning_rate": 0.00047582190399099406,
      "loss": 2.0578,
      "step": 69276
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7009173631668091,
      "learning_rate": 0.0004758185895629247,
      "loss": 2.2726,
      "step": 69277
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7167436480522156,
      "learning_rate": 0.0004758152751021675,
      "loss": 2.0659,
      "step": 69278
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7782840728759766,
      "learning_rate": 0.0004758119606087227,
      "loss": 2.1647,
      "step": 69279
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.8275870084762573,
      "learning_rate": 0.00047580864608259127,
      "loss": 2.2109,
      "step": 69280
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7681248188018799,
      "learning_rate": 0.0004758053315237737,
      "loss": 2.3035,
      "step": 69281
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7680582404136658,
      "learning_rate": 0.0004758020169322706,
      "loss": 2.1325,
      "step": 69282
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.720564067363739,
      "learning_rate": 0.0004757987023080825,
      "loss": 2.2788,
      "step": 69283
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.1001322269439697,
      "learning_rate": 0.0004757953876512102,
      "loss": 2.1503,
      "step": 69284
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7816740274429321,
      "learning_rate": 0.0004757920729616542,
      "loss": 2.2676,
      "step": 69285
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.8648397326469421,
      "learning_rate": 0.00047578875823941504,
      "loss": 2.4207,
      "step": 69286
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7447873950004578,
      "learning_rate": 0.0004757854434844935,
      "loss": 2.354,
      "step": 69287
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7484656572341919,
      "learning_rate": 0.0004757821286968901,
      "loss": 2.1608,
      "step": 69288
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.8052438497543335,
      "learning_rate": 0.00047577881387660554,
      "loss": 2.4433,
      "step": 69289
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.8805530071258545,
      "learning_rate": 0.00047577549902364034,
      "loss": 2.0582,
      "step": 69290
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7127509713172913,
      "learning_rate": 0.0004757721841379952,
      "loss": 2.3266,
      "step": 69291
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7788827419281006,
      "learning_rate": 0.00047576886921967064,
      "loss": 2.3278,
      "step": 69292
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7897784113883972,
      "learning_rate": 0.0004757655542686674,
      "loss": 2.3519,
      "step": 69293
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7746202945709229,
      "learning_rate": 0.00047576223928498594,
      "loss": 2.3197,
      "step": 69294
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.8028851747512817,
      "learning_rate": 0.00047575892426862703,
      "loss": 2.0333,
      "step": 69295
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.8423585891723633,
      "learning_rate": 0.0004757556092195912,
      "loss": 2.374,
      "step": 69296
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7589079141616821,
      "learning_rate": 0.00047575229413787906,
      "loss": 2.1808,
      "step": 69297
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7211049795150757,
      "learning_rate": 0.00047574897902349125,
      "loss": 2.045,
      "step": 69298
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7947054505348206,
      "learning_rate": 0.0004757456638764284,
      "loss": 2.0739,
      "step": 69299
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7245633602142334,
      "learning_rate": 0.0004757423486966911,
      "loss": 2.3208,
      "step": 69300
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7649167776107788,
      "learning_rate": 0.00047573903348428,
      "loss": 2.4571,
      "step": 69301
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7213575839996338,
      "learning_rate": 0.00047573571823919574,
      "loss": 2.2537,
      "step": 69302
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7740203738212585,
      "learning_rate": 0.0004757324029614388,
      "loss": 2.1004,
      "step": 69303
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.8656894564628601,
      "learning_rate": 0.0004757290876510099,
      "loss": 2.1336,
      "step": 69304
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7051681280136108,
      "learning_rate": 0.0004757257723079097,
      "loss": 2.3451,
      "step": 69305
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.8096809387207031,
      "learning_rate": 0.0004757224569321388,
      "loss": 2.1622,
      "step": 69306
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.8215209245681763,
      "learning_rate": 0.0004757191415236977,
      "loss": 2.1651,
      "step": 69307
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.9069448709487915,
      "learning_rate": 0.0004757158260825871,
      "loss": 2.3496,
      "step": 69308
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7864352464675903,
      "learning_rate": 0.00047571251060880767,
      "loss": 2.305,
      "step": 69309
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.8470577597618103,
      "learning_rate": 0.00047570919510235983,
      "loss": 2.1559,
      "step": 69310
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.8317269086837769,
      "learning_rate": 0.0004757058795632445,
      "loss": 2.2705,
      "step": 69311
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7704955339431763,
      "learning_rate": 0.0004757025639914621,
      "loss": 2.3401,
      "step": 69312
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7962915301322937,
      "learning_rate": 0.00047569924838701316,
      "loss": 1.9667,
      "step": 69313
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7793071866035461,
      "learning_rate": 0.00047569593274989847,
      "loss": 2.2153,
      "step": 69314
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7446079254150391,
      "learning_rate": 0.00047569261708011866,
      "loss": 2.3218,
      "step": 69315
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7212225794792175,
      "learning_rate": 0.0004756893013776742,
      "loss": 2.2415,
      "step": 69316
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.700628936290741,
      "learning_rate": 0.00047568598564256577,
      "loss": 1.8059,
      "step": 69317
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7686187624931335,
      "learning_rate": 0.0004756826698747941,
      "loss": 2.1376,
      "step": 69318
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7551847100257874,
      "learning_rate": 0.00047567935407435963,
      "loss": 2.3145,
      "step": 69319
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7538554668426514,
      "learning_rate": 0.000475676038241263,
      "loss": 2.0726,
      "step": 69320
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7582230567932129,
      "learning_rate": 0.000475672722375505,
      "loss": 2.4155,
      "step": 69321
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7588305473327637,
      "learning_rate": 0.00047566940647708606,
      "loss": 2.2856,
      "step": 69322
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7259646058082581,
      "learning_rate": 0.00047566609054600686,
      "loss": 2.2339,
      "step": 69323
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7220086455345154,
      "learning_rate": 0.0004756627745822681,
      "loss": 2.334,
      "step": 69324
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7991723418235779,
      "learning_rate": 0.00047565945858587015,
      "loss": 2.4674,
      "step": 69325
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7447242736816406,
      "learning_rate": 0.0004756561425568139,
      "loss": 2.188,
      "step": 69326
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.8556937575340271,
      "learning_rate": 0.00047565282649509986,
      "loss": 2.2742,
      "step": 69327
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7760512232780457,
      "learning_rate": 0.0004756495104007286,
      "loss": 2.2249,
      "step": 69328
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7819081544876099,
      "learning_rate": 0.00047564619427370084,
      "loss": 2.3058,
      "step": 69329
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.8112565279006958,
      "learning_rate": 0.00047564287811401707,
      "loss": 2.0942,
      "step": 69330
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7378765940666199,
      "learning_rate": 0.000475639561921678,
      "loss": 2.2355,
      "step": 69331
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7437355518341064,
      "learning_rate": 0.00047563624569668423,
      "loss": 2.0926,
      "step": 69332
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7818539142608643,
      "learning_rate": 0.00047563292943903634,
      "loss": 2.2193,
      "step": 69333
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.8970766067504883,
      "learning_rate": 0.000475629613148735,
      "loss": 2.1654,
      "step": 69334
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7838671803474426,
      "learning_rate": 0.00047562629682578074,
      "loss": 2.2567,
      "step": 69335
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.935002863407135,
      "learning_rate": 0.0004756229804701743,
      "loss": 2.2765,
      "step": 69336
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.9050446152687073,
      "learning_rate": 0.00047561966408191624,
      "loss": 2.3575,
      "step": 69337
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.8891139626502991,
      "learning_rate": 0.0004756163476610072,
      "loss": 2.1022,
      "step": 69338
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7714600563049316,
      "learning_rate": 0.0004756130312074477,
      "loss": 2.3393,
      "step": 69339
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.0874030590057373,
      "learning_rate": 0.0004756097147212384,
      "loss": 2.541,
      "step": 69340
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.3894978761672974,
      "learning_rate": 0.00047560639820238,
      "loss": 2.2226,
      "step": 69341
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7505698204040527,
      "learning_rate": 0.000475603081650873,
      "loss": 2.0046,
      "step": 69342
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.1256310939788818,
      "learning_rate": 0.0004755997650667181,
      "loss": 2.077,
      "step": 69343
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.9323859214782715,
      "learning_rate": 0.00047559644844991593,
      "loss": 2.4205,
      "step": 69344
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7404536604881287,
      "learning_rate": 0.00047559313180046704,
      "loss": 1.9248,
      "step": 69345
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.022926926612854,
      "learning_rate": 0.000475589815118372,
      "loss": 1.8793,
      "step": 69346
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.8210722804069519,
      "learning_rate": 0.0004755864984036315,
      "loss": 2.3421,
      "step": 69347
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7070338129997253,
      "learning_rate": 0.0004755831816562463,
      "loss": 2.2407,
      "step": 69348
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.8942877054214478,
      "learning_rate": 0.0004755798648762168,
      "loss": 2.2301,
      "step": 69349
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7386693954467773,
      "learning_rate": 0.0004755765480635437,
      "loss": 2.1071,
      "step": 69350
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.850676953792572,
      "learning_rate": 0.0004755732312182276,
      "loss": 2.2559,
      "step": 69351
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.8656115531921387,
      "learning_rate": 0.0004755699143402691,
      "loss": 2.1223,
      "step": 69352
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.8142945766448975,
      "learning_rate": 0.0004755665974296688,
      "loss": 2.2314,
      "step": 69353
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7808703184127808,
      "learning_rate": 0.00047556328048642735,
      "loss": 2.3633,
      "step": 69354
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.8562629818916321,
      "learning_rate": 0.0004755599635105455,
      "loss": 2.174,
      "step": 69355
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7937551736831665,
      "learning_rate": 0.00047555664650202365,
      "loss": 2.2471,
      "step": 69356
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7542986273765564,
      "learning_rate": 0.0004755533294608625,
      "loss": 2.1933,
      "step": 69357
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7685419321060181,
      "learning_rate": 0.0004755500123870627,
      "loss": 2.141,
      "step": 69358
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.8493829369544983,
      "learning_rate": 0.00047554669528062477,
      "loss": 2.3789,
      "step": 69359
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7592692375183105,
      "learning_rate": 0.0004755433781415495,
      "loss": 2.3507,
      "step": 69360
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7538132071495056,
      "learning_rate": 0.0004755400609698374,
      "loss": 2.4556,
      "step": 69361
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7882342338562012,
      "learning_rate": 0.00047553674376548896,
      "loss": 2.2377,
      "step": 69362
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.749037504196167,
      "learning_rate": 0.000475533426528505,
      "loss": 2.3771,
      "step": 69363
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.74405437707901,
      "learning_rate": 0.00047553010925888606,
      "loss": 2.1252,
      "step": 69364
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7090246677398682,
      "learning_rate": 0.0004755267919566328,
      "loss": 2.319,
      "step": 69365
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7375348210334778,
      "learning_rate": 0.00047552347462174576,
      "loss": 2.4105,
      "step": 69366
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.9363419413566589,
      "learning_rate": 0.00047552015725422565,
      "loss": 2.2728,
      "step": 69367
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7196112871170044,
      "learning_rate": 0.000475516839854073,
      "loss": 2.3666,
      "step": 69368
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7232903838157654,
      "learning_rate": 0.0004755135224212884,
      "loss": 2.0695,
      "step": 69369
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7545177936553955,
      "learning_rate": 0.00047551020495587254,
      "loss": 2.1592,
      "step": 69370
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7925044894218445,
      "learning_rate": 0.00047550688745782603,
      "loss": 2.2817,
      "step": 69371
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.9704651236534119,
      "learning_rate": 0.00047550356992714953,
      "loss": 2.0812,
      "step": 69372
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.8378512859344482,
      "learning_rate": 0.0004755002523638436,
      "loss": 2.2453,
      "step": 69373
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.063904047012329,
      "learning_rate": 0.00047549693476790875,
      "loss": 2.1899,
      "step": 69374
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.8882864117622375,
      "learning_rate": 0.0004754936171393458,
      "loss": 2.0239,
      "step": 69375
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.9294224977493286,
      "learning_rate": 0.00047549029947815524,
      "loss": 2.4266,
      "step": 69376
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.2014628648757935,
      "learning_rate": 0.0004754869817843377,
      "loss": 2.1006,
      "step": 69377
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.8108475804328918,
      "learning_rate": 0.00047548366405789387,
      "loss": 2.176,
      "step": 69378
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.8856566548347473,
      "learning_rate": 0.0004754803462988243,
      "loss": 2.3883,
      "step": 69379
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.796222448348999,
      "learning_rate": 0.00047547702850712963,
      "loss": 2.2248,
      "step": 69380
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7995372414588928,
      "learning_rate": 0.0004754737106828104,
      "loss": 2.2884,
      "step": 69381
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7408008575439453,
      "learning_rate": 0.0004754703928258674,
      "loss": 2.1568,
      "step": 69382
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.8278216123580933,
      "learning_rate": 0.0004754670749363011,
      "loss": 2.1648,
      "step": 69383
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.8731918931007385,
      "learning_rate": 0.0004754637570141121,
      "loss": 2.1722,
      "step": 69384
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7630431652069092,
      "learning_rate": 0.00047546043905930116,
      "loss": 2.5328,
      "step": 69385
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.8862056732177734,
      "learning_rate": 0.0004754571210718687,
      "loss": 2.2546,
      "step": 69386
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.8352327942848206,
      "learning_rate": 0.0004754538030518156,
      "loss": 2.1356,
      "step": 69387
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7546094655990601,
      "learning_rate": 0.0004754504849991422,
      "loss": 2.2974,
      "step": 69388
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.906927227973938,
      "learning_rate": 0.00047544716691384927,
      "loss": 2.1349,
      "step": 69389
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.8314216136932373,
      "learning_rate": 0.00047544384879593743,
      "loss": 2.1052,
      "step": 69390
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7648167014122009,
      "learning_rate": 0.0004754405306454073,
      "loss": 2.172,
      "step": 69391
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7634355425834656,
      "learning_rate": 0.0004754372124622595,
      "loss": 2.2146,
      "step": 69392
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.8271952867507935,
      "learning_rate": 0.00047543389424649444,
      "loss": 2.3993,
      "step": 69393
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7503827810287476,
      "learning_rate": 0.000475430575998113,
      "loss": 2.1846,
      "step": 69394
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.8731980919837952,
      "learning_rate": 0.0004754272577171157,
      "loss": 2.2427,
      "step": 69395
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7678000926971436,
      "learning_rate": 0.00047542393940350317,
      "loss": 2.3154,
      "step": 69396
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7537819147109985,
      "learning_rate": 0.0004754206210572761,
      "loss": 2.057,
      "step": 69397
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7242492437362671,
      "learning_rate": 0.0004754173026784349,
      "loss": 2.2124,
      "step": 69398
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7962122559547424,
      "learning_rate": 0.0004754139842669803,
      "loss": 2.1669,
      "step": 69399
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7651887536048889,
      "learning_rate": 0.00047541066582291303,
      "loss": 2.1786,
      "step": 69400
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7666240930557251,
      "learning_rate": 0.0004754073473462335,
      "loss": 2.1534,
      "step": 69401
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7467783093452454,
      "learning_rate": 0.00047540402883694256,
      "loss": 2.1629,
      "step": 69402
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7597958445549011,
      "learning_rate": 0.0004754007102950406,
      "loss": 2.2412,
      "step": 69403
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7548094391822815,
      "learning_rate": 0.00047539739172052837,
      "loss": 2.2931,
      "step": 69404
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7600250840187073,
      "learning_rate": 0.00047539407311340647,
      "loss": 2.1056,
      "step": 69405
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7824682593345642,
      "learning_rate": 0.0004753907544736755,
      "loss": 2.0083,
      "step": 69406
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.9243265390396118,
      "learning_rate": 0.000475387435801336,
      "loss": 2.2547,
      "step": 69407
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7710067629814148,
      "learning_rate": 0.0004753841170963887,
      "loss": 2.2654,
      "step": 69408
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7748246192932129,
      "learning_rate": 0.00047538079835883436,
      "loss": 2.2398,
      "step": 69409
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.8151630759239197,
      "learning_rate": 0.00047537747958867316,
      "loss": 2.0659,
      "step": 69410
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.9218966364860535,
      "learning_rate": 0.0004753741607859061,
      "loss": 2.082,
      "step": 69411
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7351731657981873,
      "learning_rate": 0.00047537084195053373,
      "loss": 2.1642,
      "step": 69412
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.8958792090415955,
      "learning_rate": 0.00047536752308255647,
      "loss": 2.3057,
      "step": 69413
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7858064770698547,
      "learning_rate": 0.00047536420418197513,
      "loss": 2.2031,
      "step": 69414
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.8194663524627686,
      "learning_rate": 0.0004753608852487903,
      "loss": 2.3859,
      "step": 69415
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7927163243293762,
      "learning_rate": 0.00047535756628300257,
      "loss": 2.118,
      "step": 69416
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7574169039726257,
      "learning_rate": 0.00047535424728461254,
      "loss": 2.4254,
      "step": 69417
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7269008159637451,
      "learning_rate": 0.0004753509282536208,
      "loss": 2.3219,
      "step": 69418
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.953667402267456,
      "learning_rate": 0.0004753476091900281,
      "loss": 2.0789,
      "step": 69419
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7553795576095581,
      "learning_rate": 0.00047534429009383496,
      "loss": 2.2788,
      "step": 69420
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7895997166633606,
      "learning_rate": 0.0004753409709650419,
      "loss": 2.1144,
      "step": 69421
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7476202249526978,
      "learning_rate": 0.0004753376518036498,
      "loss": 2.3429,
      "step": 69422
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7663960456848145,
      "learning_rate": 0.00047533433260965905,
      "loss": 2.3851,
      "step": 69423
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.8192564249038696,
      "learning_rate": 0.00047533101338307027,
      "loss": 2.3982,
      "step": 69424
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7221909761428833,
      "learning_rate": 0.00047532769412388424,
      "loss": 2.0783,
      "step": 69425
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7099025845527649,
      "learning_rate": 0.0004753243748321015,
      "loss": 2.27,
      "step": 69426
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.8225560784339905,
      "learning_rate": 0.0004753210555077225,
      "loss": 2.3232,
      "step": 69427
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7999077439308167,
      "learning_rate": 0.00047531773615074816,
      "loss": 2.1727,
      "step": 69428
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7570069432258606,
      "learning_rate": 0.0004753144167611789,
      "loss": 2.3164,
      "step": 69429
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.8095948696136475,
      "learning_rate": 0.00047531109733901534,
      "loss": 2.3645,
      "step": 69430
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7564269304275513,
      "learning_rate": 0.0004753077778842582,
      "loss": 2.2319,
      "step": 69431
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.8383417129516602,
      "learning_rate": 0.00047530445839690796,
      "loss": 2.5279,
      "step": 69432
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7348740696907043,
      "learning_rate": 0.0004753011388769654,
      "loss": 2.0165,
      "step": 69433
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7938363552093506,
      "learning_rate": 0.00047529781932443096,
      "loss": 2.1804,
      "step": 69434
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.8459193706512451,
      "learning_rate": 0.0004752944997393054,
      "loss": 2.3042,
      "step": 69435
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.8412908315658569,
      "learning_rate": 0.0004752911801215893,
      "loss": 2.234,
      "step": 69436
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.8131548166275024,
      "learning_rate": 0.0004752878604712832,
      "loss": 2.0514,
      "step": 69437
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7772655487060547,
      "learning_rate": 0.0004752845407883878,
      "loss": 2.3533,
      "step": 69438
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7394806146621704,
      "learning_rate": 0.00047528122107290367,
      "loss": 2.3619,
      "step": 69439
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7947137951850891,
      "learning_rate": 0.0004752779013248316,
      "loss": 2.0793,
      "step": 69440
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7737408876419067,
      "learning_rate": 0.00047527458154417195,
      "loss": 2.324,
      "step": 69441
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7181495428085327,
      "learning_rate": 0.0004752712617309254,
      "loss": 2.411,
      "step": 69442
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.6984826922416687,
      "learning_rate": 0.0004752679418850927,
      "loss": 2.1391,
      "step": 69443
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7570358514785767,
      "learning_rate": 0.0004752646220066743,
      "loss": 2.1494,
      "step": 69444
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7765247225761414,
      "learning_rate": 0.00047526130209567095,
      "loss": 2.2711,
      "step": 69445
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.8518055081367493,
      "learning_rate": 0.0004752579821520832,
      "loss": 2.1631,
      "step": 69446
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7301232814788818,
      "learning_rate": 0.0004752546621759117,
      "loss": 2.231,
      "step": 69447
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7035789489746094,
      "learning_rate": 0.00047525134216715707,
      "loss": 2.3234,
      "step": 69448
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.8183330297470093,
      "learning_rate": 0.0004752480221258199,
      "loss": 2.1623,
      "step": 69449
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.8274641633033752,
      "learning_rate": 0.00047524470205190076,
      "loss": 2.1795,
      "step": 69450
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7445815205574036,
      "learning_rate": 0.0004752413819454004,
      "loss": 2.2614,
      "step": 69451
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.8283449411392212,
      "learning_rate": 0.00047523806180631924,
      "loss": 2.0912,
      "step": 69452
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.9810183644294739,
      "learning_rate": 0.0004752347416346581,
      "loss": 2.2574,
      "step": 69453
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7037788033485413,
      "learning_rate": 0.0004752314214304175,
      "loss": 2.2005,
      "step": 69454
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.8219366073608398,
      "learning_rate": 0.00047522810119359805,
      "loss": 2.0495,
      "step": 69455
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.8018357753753662,
      "learning_rate": 0.00047522478092420047,
      "loss": 2.2443,
      "step": 69456
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.8200008273124695,
      "learning_rate": 0.00047522146062222516,
      "loss": 2.2956,
      "step": 69457
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.9236628413200378,
      "learning_rate": 0.00047521814028767304,
      "loss": 2.1505,
      "step": 69458
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7689066529273987,
      "learning_rate": 0.00047521481992054444,
      "loss": 2.2696,
      "step": 69459
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.0079600811004639,
      "learning_rate": 0.00047521149952084007,
      "loss": 2.169,
      "step": 69460
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.8669114112854004,
      "learning_rate": 0.0004752081790885607,
      "loss": 2.1831,
      "step": 69461
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7207195162773132,
      "learning_rate": 0.00047520485862370665,
      "loss": 2.1571,
      "step": 69462
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.6966003775596619,
      "learning_rate": 0.00047520153812627885,
      "loss": 2.1843,
      "step": 69463
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.8960549235343933,
      "learning_rate": 0.0004751982175962778,
      "loss": 2.339,
      "step": 69464
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7544692158699036,
      "learning_rate": 0.00047519489703370406,
      "loss": 2.2488,
      "step": 69465
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7402139902114868,
      "learning_rate": 0.0004751915764385582,
      "loss": 2.2248,
      "step": 69466
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.8220046758651733,
      "learning_rate": 0.00047518825581084097,
      "loss": 2.394,
      "step": 69467
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.8874048590660095,
      "learning_rate": 0.000475184935150553,
      "loss": 2.1629,
      "step": 69468
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7177123427391052,
      "learning_rate": 0.0004751816144576947,
      "loss": 2.1293,
      "step": 69469
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7701827883720398,
      "learning_rate": 0.00047517829373226695,
      "loss": 2.286,
      "step": 69470
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.9695706963539124,
      "learning_rate": 0.00047517497297427014,
      "loss": 2.083,
      "step": 69471
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.796589195728302,
      "learning_rate": 0.0004751716521837051,
      "loss": 2.3008,
      "step": 69472
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7450060844421387,
      "learning_rate": 0.0004751683313605724,
      "loss": 2.2333,
      "step": 69473
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.9732843041419983,
      "learning_rate": 0.00047516501050487247,
      "loss": 2.1917,
      "step": 69474
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.8505808115005493,
      "learning_rate": 0.0004751616896166061,
      "loss": 2.0997,
      "step": 69475
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7876497507095337,
      "learning_rate": 0.0004751583686957739,
      "loss": 2.1945,
      "step": 69476
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.8985998034477234,
      "learning_rate": 0.0004751550477423764,
      "loss": 2.1417,
      "step": 69477
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.8291850090026855,
      "learning_rate": 0.0004751517267564143,
      "loss": 2.2814,
      "step": 69478
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7392218708992004,
      "learning_rate": 0.0004751484057378882,
      "loss": 2.1988,
      "step": 69479
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.8732414841651917,
      "learning_rate": 0.00047514508468679865,
      "loss": 2.1495,
      "step": 69480
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7730008959770203,
      "learning_rate": 0.0004751417636031463,
      "loss": 2.2581,
      "step": 69481
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.829781711101532,
      "learning_rate": 0.0004751384424869319,
      "loss": 2.1362,
      "step": 69482
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7471664547920227,
      "learning_rate": 0.0004751351213381559,
      "loss": 2.1447,
      "step": 69483
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7895286083221436,
      "learning_rate": 0.000475131800156819,
      "loss": 2.0835,
      "step": 69484
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7534011006355286,
      "learning_rate": 0.0004751284789429218,
      "loss": 2.1958,
      "step": 69485
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.9302142262458801,
      "learning_rate": 0.00047512515769646484,
      "loss": 2.2752,
      "step": 69486
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7174326777458191,
      "learning_rate": 0.0004751218364174488,
      "loss": 2.3007,
      "step": 69487
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.8688084483146667,
      "learning_rate": 0.00047511851510587447,
      "loss": 2.3669,
      "step": 69488
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.8521973490715027,
      "learning_rate": 0.00047511519376174213,
      "loss": 2.3407,
      "step": 69489
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7133296728134155,
      "learning_rate": 0.0004751118723850526,
      "loss": 2.1104,
      "step": 69490
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7894084453582764,
      "learning_rate": 0.00047510855097580656,
      "loss": 2.1416,
      "step": 69491
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.8405422568321228,
      "learning_rate": 0.0004751052295340044,
      "loss": 2.2129,
      "step": 69492
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.71757972240448,
      "learning_rate": 0.0004751019080596469,
      "loss": 2.14,
      "step": 69493
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.762925386428833,
      "learning_rate": 0.00047509858655273474,
      "loss": 2.4606,
      "step": 69494
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7517940402030945,
      "learning_rate": 0.0004750952650132684,
      "loss": 2.0716,
      "step": 69495
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.8499823212623596,
      "learning_rate": 0.00047509194344124854,
      "loss": 2.3464,
      "step": 69496
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7150415182113647,
      "learning_rate": 0.0004750886218366758,
      "loss": 1.947,
      "step": 69497
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7178914546966553,
      "learning_rate": 0.0004750853001995508,
      "loss": 2.2449,
      "step": 69498
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7349579930305481,
      "learning_rate": 0.0004750819785298741,
      "loss": 2.2987,
      "step": 69499
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.776289165019989,
      "learning_rate": 0.00047507865682764634,
      "loss": 2.1331,
      "step": 69500
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7871864438056946,
      "learning_rate": 0.0004750753350928682,
      "loss": 2.1644,
      "step": 69501
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.8079332113265991,
      "learning_rate": 0.0004750720133255402,
      "loss": 2.1623,
      "step": 69502
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.8841020464897156,
      "learning_rate": 0.00047506869152566305,
      "loss": 2.175,
      "step": 69503
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.6935661435127258,
      "learning_rate": 0.00047506536969323733,
      "loss": 2.1971,
      "step": 69504
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.8136736154556274,
      "learning_rate": 0.00047506204782826365,
      "loss": 2.2303,
      "step": 69505
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.6756103038787842,
      "learning_rate": 0.0004750587259307426,
      "loss": 2.2866,
      "step": 69506
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7599541544914246,
      "learning_rate": 0.0004750554040006748,
      "loss": 2.132,
      "step": 69507
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7672774195671082,
      "learning_rate": 0.000475052082038061,
      "loss": 2.1587,
      "step": 69508
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7339123487472534,
      "learning_rate": 0.0004750487600429016,
      "loss": 2.0553,
      "step": 69509
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.8089518547058105,
      "learning_rate": 0.00047504543801519736,
      "loss": 2.3974,
      "step": 69510
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7392427921295166,
      "learning_rate": 0.00047504211595494883,
      "loss": 2.2937,
      "step": 69511
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.6850476861000061,
      "learning_rate": 0.0004750387938621568,
      "loss": 2.2987,
      "step": 69512
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7208643555641174,
      "learning_rate": 0.00047503547173682165,
      "loss": 2.206,
      "step": 69513
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.9100558757781982,
      "learning_rate": 0.00047503214957894414,
      "loss": 2.1858,
      "step": 69514
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7768545746803284,
      "learning_rate": 0.00047502882738852483,
      "loss": 2.3672,
      "step": 69515
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7487308382987976,
      "learning_rate": 0.00047502550516556434,
      "loss": 2.1153,
      "step": 69516
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7288552522659302,
      "learning_rate": 0.00047502218291006336,
      "loss": 2.2135,
      "step": 69517
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.8526048064231873,
      "learning_rate": 0.00047501886062202244,
      "loss": 2.5217,
      "step": 69518
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7286602854728699,
      "learning_rate": 0.0004750155383014422,
      "loss": 2.3483,
      "step": 69519
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.8168735504150391,
      "learning_rate": 0.0004750122159483232,
      "loss": 2.2936,
      "step": 69520
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.8562570214271545,
      "learning_rate": 0.0004750088935626662,
      "loss": 2.3596,
      "step": 69521
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7893028855323792,
      "learning_rate": 0.00047500557114447175,
      "loss": 2.1282,
      "step": 69522
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.8230159878730774,
      "learning_rate": 0.00047500224869374044,
      "loss": 2.29,
      "step": 69523
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7829655408859253,
      "learning_rate": 0.0004749989262104729,
      "loss": 2.1749,
      "step": 69524
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.73023921251297,
      "learning_rate": 0.0004749956036946698,
      "loss": 2.2396,
      "step": 69525
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.8410391807556152,
      "learning_rate": 0.0004749922811463316,
      "loss": 2.1611,
      "step": 69526
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.8040904402732849,
      "learning_rate": 0.0004749889585654592,
      "loss": 2.4154,
      "step": 69527
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.6951155066490173,
      "learning_rate": 0.00047498563595205293,
      "loss": 2.1887,
      "step": 69528
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.8484203219413757,
      "learning_rate": 0.0004749823133061136,
      "loss": 2.2058,
      "step": 69529
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.748473048210144,
      "learning_rate": 0.00047497899062764164,
      "loss": 2.3071,
      "step": 69530
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.8744578957557678,
      "learning_rate": 0.0004749756679166379,
      "loss": 2.3909,
      "step": 69531
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.8226385116577148,
      "learning_rate": 0.0004749723451731029,
      "loss": 2.3098,
      "step": 69532
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7526246905326843,
      "learning_rate": 0.0004749690223970371,
      "loss": 2.2903,
      "step": 69533
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7828232645988464,
      "learning_rate": 0.00047496569958844137,
      "loss": 2.3758,
      "step": 69534
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.8552340269088745,
      "learning_rate": 0.0004749623767473162,
      "loss": 2.3175,
      "step": 69535
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.774071455001831,
      "learning_rate": 0.00047495905387366215,
      "loss": 2.187,
      "step": 69536
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7429542541503906,
      "learning_rate": 0.00047495573096748,
      "loss": 2.1805,
      "step": 69537
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7081019282341003,
      "learning_rate": 0.00047495240802877017,
      "loss": 2.2784,
      "step": 69538
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7513382434844971,
      "learning_rate": 0.0004749490850575335,
      "loss": 2.4243,
      "step": 69539
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7858167886734009,
      "learning_rate": 0.0004749457620537704,
      "loss": 2.2561,
      "step": 69540
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.695192813873291,
      "learning_rate": 0.0004749424390174816,
      "loss": 2.0551,
      "step": 69541
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7973486185073853,
      "learning_rate": 0.00047493911594866773,
      "loss": 2.6042,
      "step": 69542
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7984328866004944,
      "learning_rate": 0.00047493579284732935,
      "loss": 2.3992,
      "step": 69543
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7483583688735962,
      "learning_rate": 0.00047493246971346714,
      "loss": 2.1325,
      "step": 69544
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7670192718505859,
      "learning_rate": 0.0004749291465470816,
      "loss": 2.2735,
      "step": 69545
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7958983778953552,
      "learning_rate": 0.00047492582334817356,
      "loss": 2.0814,
      "step": 69546
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.8067667484283447,
      "learning_rate": 0.00047492250011674345,
      "loss": 2.5619,
      "step": 69547
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7358214259147644,
      "learning_rate": 0.0004749191768527919,
      "loss": 2.176,
      "step": 69548
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.735845685005188,
      "learning_rate": 0.0004749158535563196,
      "loss": 2.1973,
      "step": 69549
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7266450524330139,
      "learning_rate": 0.0004749125302273272,
      "loss": 1.9803,
      "step": 69550
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7861617207527161,
      "learning_rate": 0.0004749092068658152,
      "loss": 2.0363,
      "step": 69551
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.8863798379898071,
      "learning_rate": 0.0004749058834717843,
      "loss": 2.2899,
      "step": 69552
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7294643521308899,
      "learning_rate": 0.000474902560045235,
      "loss": 2.1833,
      "step": 69553
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.8060588836669922,
      "learning_rate": 0.0004748992365861681,
      "loss": 1.9741,
      "step": 69554
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7283391952514648,
      "learning_rate": 0.0004748959130945842,
      "loss": 2.1656,
      "step": 69555
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.667892336845398,
      "learning_rate": 0.0004748925895704837,
      "loss": 2.1423,
      "step": 69556
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.751312792301178,
      "learning_rate": 0.0004748892660138674,
      "loss": 2.0103,
      "step": 69557
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7028352618217468,
      "learning_rate": 0.000474885942424736,
      "loss": 2.1299,
      "step": 69558
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7719520330429077,
      "learning_rate": 0.00047488261880308986,
      "loss": 2.453,
      "step": 69559
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7613582611083984,
      "learning_rate": 0.00047487929514892986,
      "loss": 2.1178,
      "step": 69560
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7434173226356506,
      "learning_rate": 0.00047487597146225643,
      "loss": 2.211,
      "step": 69561
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.8396309614181519,
      "learning_rate": 0.0004748726477430702,
      "loss": 2.1173,
      "step": 69562
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7898138761520386,
      "learning_rate": 0.0004748693239913719,
      "loss": 2.0451,
      "step": 69563
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.8554715514183044,
      "learning_rate": 0.00047486600020716214,
      "loss": 2.2856,
      "step": 69564
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.8336923718452454,
      "learning_rate": 0.0004748626763904414,
      "loss": 2.2812,
      "step": 69565
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7601854205131531,
      "learning_rate": 0.0004748593525412105,
      "loss": 2.1618,
      "step": 69566
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7406298518180847,
      "learning_rate": 0.0004748560286594699,
      "loss": 2.3435,
      "step": 69567
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.8383252620697021,
      "learning_rate": 0.0004748527047452203,
      "loss": 2.5205,
      "step": 69568
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7466963529586792,
      "learning_rate": 0.00047484938079846216,
      "loss": 2.1251,
      "step": 69569
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7545397281646729,
      "learning_rate": 0.0004748460568191963,
      "loss": 2.3394,
      "step": 69570
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.9125849008560181,
      "learning_rate": 0.00047484273280742325,
      "loss": 2.3067,
      "step": 69571
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.70885169506073,
      "learning_rate": 0.0004748394087631436,
      "loss": 2.2858,
      "step": 69572
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7370674014091492,
      "learning_rate": 0.00047483608468635804,
      "loss": 2.1125,
      "step": 69573
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7203024625778198,
      "learning_rate": 0.00047483276057706716,
      "loss": 2.3107,
      "step": 69574
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.8271870017051697,
      "learning_rate": 0.0004748294364352715,
      "loss": 2.3002,
      "step": 69575
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7510582208633423,
      "learning_rate": 0.00047482611226097186,
      "loss": 2.2168,
      "step": 69576
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7927923798561096,
      "learning_rate": 0.0004748227880541687,
      "loss": 2.4471,
      "step": 69577
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7722428441047668,
      "learning_rate": 0.0004748194638148627,
      "loss": 2.3484,
      "step": 69578
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7679385542869568,
      "learning_rate": 0.0004748161395430544,
      "loss": 2.2203,
      "step": 69579
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7413835525512695,
      "learning_rate": 0.0004748128152387445,
      "loss": 2.2005,
      "step": 69580
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7567787766456604,
      "learning_rate": 0.0004748094909019336,
      "loss": 2.1946,
      "step": 69581
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7052643895149231,
      "learning_rate": 0.0004748061665326223,
      "loss": 2.2071,
      "step": 69582
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7874898314476013,
      "learning_rate": 0.00047480284213081135,
      "loss": 2.3779,
      "step": 69583
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.8871598243713379,
      "learning_rate": 0.00047479951769650115,
      "loss": 2.2617,
      "step": 69584
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7595643401145935,
      "learning_rate": 0.0004747961932296924,
      "loss": 2.159,
      "step": 69585
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.8813982009887695,
      "learning_rate": 0.0004747928687303858,
      "loss": 2.3452,
      "step": 69586
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.8365644216537476,
      "learning_rate": 0.0004747895441985819,
      "loss": 2.1631,
      "step": 69587
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7907124757766724,
      "learning_rate": 0.00047478621963428124,
      "loss": 2.2547,
      "step": 69588
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7923920750617981,
      "learning_rate": 0.00047478289503748465,
      "loss": 2.1188,
      "step": 69589
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.8418516516685486,
      "learning_rate": 0.00047477957040819256,
      "loss": 2.2887,
      "step": 69590
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.8121969699859619,
      "learning_rate": 0.00047477624574640564,
      "loss": 2.1072,
      "step": 69591
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.772871196269989,
      "learning_rate": 0.00047477292105212453,
      "loss": 2.2793,
      "step": 69592
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7224678993225098,
      "learning_rate": 0.00047476959632534977,
      "loss": 2.261,
      "step": 69593
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.8132685422897339,
      "learning_rate": 0.00047476627156608207,
      "loss": 2.2196,
      "step": 69594
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.8100937008857727,
      "learning_rate": 0.00047476294677432213,
      "loss": 2.0973,
      "step": 69595
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.8554949760437012,
      "learning_rate": 0.00047475962195007033,
      "loss": 2.2703,
      "step": 69596
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.8999274969100952,
      "learning_rate": 0.0004747562970933275,
      "loss": 2.3367,
      "step": 69597
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7823039293289185,
      "learning_rate": 0.00047475297220409414,
      "loss": 2.192,
      "step": 69598
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7298112511634827,
      "learning_rate": 0.0004747496472823709,
      "loss": 2.0679,
      "step": 69599
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.8073738217353821,
      "learning_rate": 0.00047474632232815837,
      "loss": 2.1968,
      "step": 69600
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.8301811814308167,
      "learning_rate": 0.00047474299734145725,
      "loss": 2.1491,
      "step": 69601
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.785467803478241,
      "learning_rate": 0.0004747396723222681,
      "loss": 2.1577,
      "step": 69602
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7074111700057983,
      "learning_rate": 0.00047473634727059153,
      "loss": 2.1207,
      "step": 69603
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7975130081176758,
      "learning_rate": 0.00047473302218642823,
      "loss": 2.2487,
      "step": 69604
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.8316735625267029,
      "learning_rate": 0.00047472969706977865,
      "loss": 2.2358,
      "step": 69605
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7259053587913513,
      "learning_rate": 0.00047472637192064366,
      "loss": 2.1086,
      "step": 69606
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7667003870010376,
      "learning_rate": 0.00047472304673902365,
      "loss": 2.1577,
      "step": 69607
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.8105787038803101,
      "learning_rate": 0.0004747197215249193,
      "loss": 2.2094,
      "step": 69608
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7456119060516357,
      "learning_rate": 0.00047471639627833136,
      "loss": 2.1378,
      "step": 69609
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.8401910662651062,
      "learning_rate": 0.0004747130709992603,
      "loss": 2.5364,
      "step": 69610
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7692955136299133,
      "learning_rate": 0.00047470974568770676,
      "loss": 2.3163,
      "step": 69611
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7535963654518127,
      "learning_rate": 0.0004747064203436714,
      "loss": 2.2168,
      "step": 69612
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.9328357577323914,
      "learning_rate": 0.0004747030949671548,
      "loss": 2.3958,
      "step": 69613
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7976895570755005,
      "learning_rate": 0.0004746997695581576,
      "loss": 2.4387,
      "step": 69614
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7485116720199585,
      "learning_rate": 0.00047469644411668037,
      "loss": 2.3739,
      "step": 69615
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.0097213983535767,
      "learning_rate": 0.0004746931186427239,
      "loss": 2.3718,
      "step": 69616
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.8417488932609558,
      "learning_rate": 0.0004746897931362886,
      "loss": 2.3408,
      "step": 69617
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.8339932560920715,
      "learning_rate": 0.00047468646759737514,
      "loss": 2.1853,
      "step": 69618
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.8736656308174133,
      "learning_rate": 0.00047468314202598423,
      "loss": 2.2662,
      "step": 69619
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7573424577713013,
      "learning_rate": 0.0004746798164221164,
      "loss": 2.3597,
      "step": 69620
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.829163670539856,
      "learning_rate": 0.0004746764907857723,
      "loss": 2.4257,
      "step": 69621
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.8653476238250732,
      "learning_rate": 0.0004746731651169525,
      "loss": 2.1819,
      "step": 69622
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7234298586845398,
      "learning_rate": 0.00047466983941565774,
      "loss": 2.2484,
      "step": 69623
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.912468671798706,
      "learning_rate": 0.00047466651368188853,
      "loss": 2.0996,
      "step": 69624
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7676526308059692,
      "learning_rate": 0.0004746631879156455,
      "loss": 2.2998,
      "step": 69625
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7443005442619324,
      "learning_rate": 0.0004746598621169293,
      "loss": 2.1921,
      "step": 69626
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.772907555103302,
      "learning_rate": 0.0004746565362857405,
      "loss": 2.1796,
      "step": 69627
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.1988880634307861,
      "learning_rate": 0.00047465321042207984,
      "loss": 2.2875,
      "step": 69628
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.8378355503082275,
      "learning_rate": 0.00047464988452594777,
      "loss": 2.1601,
      "step": 69629
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7519919276237488,
      "learning_rate": 0.000474646558597345,
      "loss": 2.1409,
      "step": 69630
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7414110898971558,
      "learning_rate": 0.00047464323263627213,
      "loss": 2.0784,
      "step": 69631
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.9107555747032166,
      "learning_rate": 0.00047463990664272987,
      "loss": 2.1539,
      "step": 69632
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.846351146697998,
      "learning_rate": 0.00047463658061671856,
      "loss": 2.2813,
      "step": 69633
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.8423950672149658,
      "learning_rate": 0.0004746332545582392,
      "loss": 2.3307,
      "step": 69634
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.8577823042869568,
      "learning_rate": 0.0004746299284672922,
      "loss": 2.1145,
      "step": 69635
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7915037274360657,
      "learning_rate": 0.0004746266023438782,
      "loss": 2.3668,
      "step": 69636
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.8280589580535889,
      "learning_rate": 0.00047462327618799774,
      "loss": 2.2462,
      "step": 69637
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.8232893943786621,
      "learning_rate": 0.0004746199499996516,
      "loss": 2.2399,
      "step": 69638
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7247985601425171,
      "learning_rate": 0.0004746166237788402,
      "loss": 2.1037,
      "step": 69639
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.8028908967971802,
      "learning_rate": 0.0004746132975255644,
      "loss": 2.0224,
      "step": 69640
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.8670272827148438,
      "learning_rate": 0.00047460997123982464,
      "loss": 2.3729,
      "step": 69641
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7773727774620056,
      "learning_rate": 0.0004746066449216216,
      "loss": 2.2807,
      "step": 69642
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.728952944278717,
      "learning_rate": 0.00047460331857095584,
      "loss": 2.341,
      "step": 69643
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7484005093574524,
      "learning_rate": 0.0004745999921878281,
      "loss": 2.1152,
      "step": 69644
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7980889678001404,
      "learning_rate": 0.0004745966657722389,
      "loss": 2.1949,
      "step": 69645
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.8191020488739014,
      "learning_rate": 0.0004745933393241889,
      "loss": 2.4244,
      "step": 69646
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7083356976509094,
      "learning_rate": 0.00047459001284367864,
      "loss": 2.0994,
      "step": 69647
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7166804075241089,
      "learning_rate": 0.0004745866863307089,
      "loss": 2.4766,
      "step": 69648
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.8601356148719788,
      "learning_rate": 0.00047458335978528,
      "loss": 2.3251,
      "step": 69649
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7774568796157837,
      "learning_rate": 0.000474580033207393,
      "loss": 2.3555,
      "step": 69650
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7852951884269714,
      "learning_rate": 0.00047457670659704817,
      "loss": 2.3336,
      "step": 69651
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.8002909421920776,
      "learning_rate": 0.0004745733799542462,
      "loss": 2.144,
      "step": 69652
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.8015159368515015,
      "learning_rate": 0.0004745700532789879,
      "loss": 2.1692,
      "step": 69653
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.8561529517173767,
      "learning_rate": 0.0004745667265712735,
      "loss": 2.0947,
      "step": 69654
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7625516653060913,
      "learning_rate": 0.00047456339983110394,
      "loss": 2.1062,
      "step": 69655
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.8875929713249207,
      "learning_rate": 0.00047456007305847985,
      "loss": 2.2447,
      "step": 69656
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7725802063941956,
      "learning_rate": 0.00047455674625340163,
      "loss": 2.4708,
      "step": 69657
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.8340036273002625,
      "learning_rate": 0.00047455341941587004,
      "loss": 2.4258,
      "step": 69658
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7972865700721741,
      "learning_rate": 0.00047455009254588574,
      "loss": 2.5825,
      "step": 69659
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7731030583381653,
      "learning_rate": 0.00047454676564344925,
      "loss": 2.2898,
      "step": 69660
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7895549535751343,
      "learning_rate": 0.00047454343870856114,
      "loss": 2.4395,
      "step": 69661
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.8045088052749634,
      "learning_rate": 0.0004745401117412222,
      "loss": 2.1703,
      "step": 69662
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.8007403016090393,
      "learning_rate": 0.0004745367847414329,
      "loss": 2.3304,
      "step": 69663
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.8942546248435974,
      "learning_rate": 0.000474533457709194,
      "loss": 2.1776,
      "step": 69664
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.8018400073051453,
      "learning_rate": 0.000474530130644506,
      "loss": 2.4173,
      "step": 69665
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.770819902420044,
      "learning_rate": 0.00047452680354736956,
      "loss": 2.2183,
      "step": 69666
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7580431699752808,
      "learning_rate": 0.00047452347641778524,
      "loss": 2.1232,
      "step": 69667
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.6995049715042114,
      "learning_rate": 0.00047452014925575384,
      "loss": 2.2826,
      "step": 69668
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7407681941986084,
      "learning_rate": 0.0004745168220612757,
      "loss": 2.3012,
      "step": 69669
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.8166491985321045,
      "learning_rate": 0.00047451349483435164,
      "loss": 2.1242,
      "step": 69670
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.761369526386261,
      "learning_rate": 0.0004745101675749823,
      "loss": 2.2658,
      "step": 69671
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.8549558520317078,
      "learning_rate": 0.00047450684028316805,
      "loss": 1.9837,
      "step": 69672
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.9433660507202148,
      "learning_rate": 0.0004745035129589098,
      "loss": 1.9867,
      "step": 69673
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.8587285876274109,
      "learning_rate": 0.00047450018560220807,
      "loss": 2.2749,
      "step": 69674
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.885257363319397,
      "learning_rate": 0.00047449685821306347,
      "loss": 2.2068,
      "step": 69675
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.8134204149246216,
      "learning_rate": 0.0004744935307914765,
      "loss": 2.3146,
      "step": 69676
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.8505305051803589,
      "learning_rate": 0.000474490203337448,
      "loss": 2.3486,
      "step": 69677
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.8453558683395386,
      "learning_rate": 0.0004744868758509784,
      "loss": 2.0516,
      "step": 69678
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.8304961323738098,
      "learning_rate": 0.0004744835483320685,
      "loss": 2.2071,
      "step": 69679
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.76222825050354,
      "learning_rate": 0.0004744802207807188,
      "loss": 2.0958,
      "step": 69680
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7740390300750732,
      "learning_rate": 0.00047447689319692976,
      "loss": 2.1698,
      "step": 69681
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.766048789024353,
      "learning_rate": 0.00047447356558070233,
      "loss": 2.1811,
      "step": 69682
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.8670488595962524,
      "learning_rate": 0.000474470237932037,
      "loss": 2.2647,
      "step": 69683
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.8266028761863708,
      "learning_rate": 0.00047446691025093425,
      "loss": 2.3215,
      "step": 69684
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.795758843421936,
      "learning_rate": 0.0004744635825373949,
      "loss": 2.2269,
      "step": 69685
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7233229875564575,
      "learning_rate": 0.00047446025479141936,
      "loss": 2.2273,
      "step": 69686
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.6927379965782166,
      "learning_rate": 0.0004744569270130085,
      "loss": 2.1919,
      "step": 69687
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.8393734693527222,
      "learning_rate": 0.0004744535992021627,
      "loss": 2.1833,
      "step": 69688
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.8017646074295044,
      "learning_rate": 0.00047445027135888275,
      "loss": 2.1698,
      "step": 69689
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.6948125958442688,
      "learning_rate": 0.00047444694348316915,
      "loss": 2.2969,
      "step": 69690
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7752953767776489,
      "learning_rate": 0.00047444361557502266,
      "loss": 2.2655,
      "step": 69691
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.745644748210907,
      "learning_rate": 0.00047444028763444373,
      "loss": 2.3121,
      "step": 69692
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7383374571800232,
      "learning_rate": 0.000474436959661433,
      "loss": 2.1196,
      "step": 69693
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7480687499046326,
      "learning_rate": 0.00047443363165599125,
      "loss": 2.1676,
      "step": 69694
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7193155884742737,
      "learning_rate": 0.000474430303618119,
      "loss": 2.2688,
      "step": 69695
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7979605793952942,
      "learning_rate": 0.00047442697554781686,
      "loss": 2.3044,
      "step": 69696
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7664710879325867,
      "learning_rate": 0.00047442364744508536,
      "loss": 2.2313,
      "step": 69697
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7102023363113403,
      "learning_rate": 0.0004744203193099253,
      "loss": 2.1524,
      "step": 69698
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7440622448921204,
      "learning_rate": 0.00047441699114233717,
      "loss": 2.1454,
      "step": 69699
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.761389970779419,
      "learning_rate": 0.0004744136629423216,
      "loss": 2.2231,
      "step": 69700
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7413387894630432,
      "learning_rate": 0.00047441033470987937,
      "loss": 2.1719,
      "step": 69701
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.8159114718437195,
      "learning_rate": 0.00047440700644501085,
      "loss": 2.2091,
      "step": 69702
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7386995553970337,
      "learning_rate": 0.0004744036781477167,
      "loss": 2.3367,
      "step": 69703
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7436130046844482,
      "learning_rate": 0.0004744003498179978,
      "loss": 2.1758,
      "step": 69704
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7373675107955933,
      "learning_rate": 0.0004743970214558545,
      "loss": 2.0871,
      "step": 69705
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7870786190032959,
      "learning_rate": 0.00047439369306128754,
      "loss": 2.1699,
      "step": 69706
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7843711376190186,
      "learning_rate": 0.0004743903646342974,
      "loss": 2.2496,
      "step": 69707
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.8243816494941711,
      "learning_rate": 0.0004743870361748849,
      "loss": 2.1524,
      "step": 69708
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.88075852394104,
      "learning_rate": 0.0004743837076830505,
      "loss": 2.1633,
      "step": 69709
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7415298819541931,
      "learning_rate": 0.0004743803791587949,
      "loss": 2.3968,
      "step": 69710
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7851458787918091,
      "learning_rate": 0.0004743770506021187,
      "loss": 2.2228,
      "step": 69711
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.8739240169525146,
      "learning_rate": 0.00047437372201302255,
      "loss": 2.0813,
      "step": 69712
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7495054602622986,
      "learning_rate": 0.000474370393391507,
      "loss": 2.1396,
      "step": 69713
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7593271732330322,
      "learning_rate": 0.0004743670647375726,
      "loss": 2.3597,
      "step": 69714
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.2866817712783813,
      "learning_rate": 0.0004743637360512202,
      "loss": 2.3529,
      "step": 69715
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.8650243282318115,
      "learning_rate": 0.0004743604073324502,
      "loss": 2.334,
      "step": 69716
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.8314894437789917,
      "learning_rate": 0.0004743570785812634,
      "loss": 2.13,
      "step": 69717
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.869158148765564,
      "learning_rate": 0.0004743537497976603,
      "loss": 2.1551,
      "step": 69718
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7371383905410767,
      "learning_rate": 0.00047435042098164154,
      "loss": 1.8863,
      "step": 69719
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7517392039299011,
      "learning_rate": 0.00047434709213320776,
      "loss": 2.0786,
      "step": 69720
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.8581276535987854,
      "learning_rate": 0.00047434376325235954,
      "loss": 2.1123,
      "step": 69721
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.8108201026916504,
      "learning_rate": 0.0004743404343390976,
      "loss": 2.1349,
      "step": 69722
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.8032198548316956,
      "learning_rate": 0.00047433710539342236,
      "loss": 2.0329,
      "step": 69723
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7534734010696411,
      "learning_rate": 0.0004743337764153346,
      "loss": 2.1498,
      "step": 69724
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7728316783905029,
      "learning_rate": 0.00047433044740483493,
      "loss": 2.1691,
      "step": 69725
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.8666200041770935,
      "learning_rate": 0.00047432711836192397,
      "loss": 2.1428,
      "step": 69726
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7797046899795532,
      "learning_rate": 0.00047432378928660224,
      "loss": 2.2634,
      "step": 69727
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7449385523796082,
      "learning_rate": 0.0004743204601788704,
      "loss": 2.2797,
      "step": 69728
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7694264054298401,
      "learning_rate": 0.0004743171310387292,
      "loss": 2.4296,
      "step": 69729
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.8161844611167908,
      "learning_rate": 0.0004743138018661791,
      "loss": 2.2502,
      "step": 69730
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7187086343765259,
      "learning_rate": 0.0004743104726612207,
      "loss": 2.0948,
      "step": 69731
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.80647212266922,
      "learning_rate": 0.0004743071434238548,
      "loss": 2.2353,
      "step": 69732
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.8503173589706421,
      "learning_rate": 0.0004743038141540819,
      "loss": 2.0072,
      "step": 69733
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7854766845703125,
      "learning_rate": 0.0004743004848519026,
      "loss": 2.0495,
      "step": 69734
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7052996754646301,
      "learning_rate": 0.00047429715551731755,
      "loss": 2.1699,
      "step": 69735
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.053923487663269,
      "learning_rate": 0.0004742938261503274,
      "loss": 2.3518,
      "step": 69736
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7845821380615234,
      "learning_rate": 0.0004742904967509327,
      "loss": 2.1826,
      "step": 69737
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.9346467852592468,
      "learning_rate": 0.00047428716731913413,
      "loss": 2.3929,
      "step": 69738
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7541909217834473,
      "learning_rate": 0.0004742838378549323,
      "loss": 2.3428,
      "step": 69739
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7917271852493286,
      "learning_rate": 0.00047428050835832776,
      "loss": 2.185,
      "step": 69740
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7951312065124512,
      "learning_rate": 0.0004742771788293212,
      "loss": 2.3841,
      "step": 69741
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.9146613478660583,
      "learning_rate": 0.00047427384926791325,
      "loss": 2.1849,
      "step": 69742
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.8351098299026489,
      "learning_rate": 0.0004742705196741044,
      "loss": 2.1516,
      "step": 69743
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7822495698928833,
      "learning_rate": 0.00047426719004789556,
      "loss": 2.3893,
      "step": 69744
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.8695151209831238,
      "learning_rate": 0.00047426386038928705,
      "loss": 2.114,
      "step": 69745
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7064419388771057,
      "learning_rate": 0.0004742605306982796,
      "loss": 2.1523,
      "step": 69746
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7291088700294495,
      "learning_rate": 0.00047425720097487385,
      "loss": 2.1996,
      "step": 69747
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7526135444641113,
      "learning_rate": 0.0004742538712190704,
      "loss": 1.9727,
      "step": 69748
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7863882780075073,
      "learning_rate": 0.0004742505414308698,
      "loss": 2.3478,
      "step": 69749
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7097315788269043,
      "learning_rate": 0.00047424721161027283,
      "loss": 2.0688,
      "step": 69750
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.925625205039978,
      "learning_rate": 0.00047424388175728,
      "loss": 2.275,
      "step": 69751
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7974206805229187,
      "learning_rate": 0.0004742405518718919,
      "loss": 2.4325,
      "step": 69752
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.702934741973877,
      "learning_rate": 0.00047423722195410925,
      "loss": 1.967,
      "step": 69753
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7087332606315613,
      "learning_rate": 0.0004742338920039325,
      "loss": 2.2178,
      "step": 69754
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7163129448890686,
      "learning_rate": 0.0004742305620213624,
      "loss": 2.1633,
      "step": 69755
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7982038855552673,
      "learning_rate": 0.0004742272320063997,
      "loss": 2.2004,
      "step": 69756
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.749808669090271,
      "learning_rate": 0.00047422390195904474,
      "loss": 2.253,
      "step": 69757
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7424942851066589,
      "learning_rate": 0.0004742205718792983,
      "loss": 2.4288,
      "step": 69758
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7217341065406799,
      "learning_rate": 0.00047421724176716095,
      "loss": 2.4141,
      "step": 69759
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7682034373283386,
      "learning_rate": 0.0004742139116226333,
      "loss": 2.3939,
      "step": 69760
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.8069795966148376,
      "learning_rate": 0.000474210581445716,
      "loss": 2.2934,
      "step": 69761
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7938315868377686,
      "learning_rate": 0.0004742072512364097,
      "loss": 2.3329,
      "step": 69762
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7781485319137573,
      "learning_rate": 0.000474203920994715,
      "loss": 2.2411,
      "step": 69763
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7878028154373169,
      "learning_rate": 0.0004742005907206325,
      "loss": 2.217,
      "step": 69764
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7847458124160767,
      "learning_rate": 0.0004741972604141628,
      "loss": 2.4335,
      "step": 69765
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7305484414100647,
      "learning_rate": 0.00047419393007530657,
      "loss": 2.1206,
      "step": 69766
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7403988838195801,
      "learning_rate": 0.0004741905997040644,
      "loss": 2.2665,
      "step": 69767
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.815966010093689,
      "learning_rate": 0.0004741872693004368,
      "loss": 2.1032,
      "step": 69768
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7265130877494812,
      "learning_rate": 0.00047418393886442466,
      "loss": 2.2152,
      "step": 69769
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.6976717114448547,
      "learning_rate": 0.0004741806083960283,
      "loss": 2.3556,
      "step": 69770
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.8187167644500732,
      "learning_rate": 0.00047417727789524853,
      "loss": 2.0208,
      "step": 69771
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.8519338965415955,
      "learning_rate": 0.000474173947362086,
      "loss": 2.3158,
      "step": 69772
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.6758950352668762,
      "learning_rate": 0.00047417061679654114,
      "loss": 2.0352,
      "step": 69773
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7957034707069397,
      "learning_rate": 0.00047416728619861467,
      "loss": 2.4121,
      "step": 69774
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.8148587346076965,
      "learning_rate": 0.0004741639555683073,
      "loss": 2.2989,
      "step": 69775
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.810228705406189,
      "learning_rate": 0.0004741606249056195,
      "loss": 2.1763,
      "step": 69776
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.765812337398529,
      "learning_rate": 0.000474157294210552,
      "loss": 1.9705,
      "step": 69777
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7626649737358093,
      "learning_rate": 0.00047415396348310536,
      "loss": 2.2932,
      "step": 69778
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.8174556493759155,
      "learning_rate": 0.00047415063272328013,
      "loss": 1.9823,
      "step": 69779
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.839946985244751,
      "learning_rate": 0.0004741473019310771,
      "loss": 2.2291,
      "step": 69780
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.8485181927680969,
      "learning_rate": 0.0004741439711064967,
      "loss": 2.0887,
      "step": 69781
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7178118824958801,
      "learning_rate": 0.00047414064024953976,
      "loss": 2.0982,
      "step": 69782
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7681868672370911,
      "learning_rate": 0.00047413730936020674,
      "loss": 2.1999,
      "step": 69783
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.9225344061851501,
      "learning_rate": 0.00047413397843849825,
      "loss": 2.3771,
      "step": 69784
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.8719766736030579,
      "learning_rate": 0.00047413064748441506,
      "loss": 2.3276,
      "step": 69785
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7676650881767273,
      "learning_rate": 0.00047412731649795775,
      "loss": 2.1529,
      "step": 69786
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7624354362487793,
      "learning_rate": 0.0004741239854791267,
      "loss": 2.096,
      "step": 69787
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7375466823577881,
      "learning_rate": 0.00047412065442792284,
      "loss": 2.3017,
      "step": 69788
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.876070499420166,
      "learning_rate": 0.00047411732334434655,
      "loss": 2.3068,
      "step": 69789
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.8073851466178894,
      "learning_rate": 0.0004741139922283987,
      "loss": 2.2861,
      "step": 69790
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7587121725082397,
      "learning_rate": 0.00047411066108007976,
      "loss": 2.0631,
      "step": 69791
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.8122518062591553,
      "learning_rate": 0.0004741073298993903,
      "loss": 2.2287,
      "step": 69792
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.8842341899871826,
      "learning_rate": 0.0004741039986863311,
      "loss": 2.3068,
      "step": 69793
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.8815290927886963,
      "learning_rate": 0.0004741006674409025,
      "loss": 2.22,
      "step": 69794
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7284902334213257,
      "learning_rate": 0.0004740973361631054,
      "loss": 2.109,
      "step": 69795
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.75377357006073,
      "learning_rate": 0.00047409400485294036,
      "loss": 2.2801,
      "step": 69796
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7476638555526733,
      "learning_rate": 0.00047409067351040785,
      "loss": 2.0161,
      "step": 69797
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.8122673630714417,
      "learning_rate": 0.0004740873421355087,
      "loss": 2.1241,
      "step": 69798
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.6892209649085999,
      "learning_rate": 0.0004740840107282435,
      "loss": 2.1926,
      "step": 69799
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.8174312114715576,
      "learning_rate": 0.00047408067928861265,
      "loss": 2.0397,
      "step": 69800
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7323657870292664,
      "learning_rate": 0.0004740773478166169,
      "loss": 2.1618,
      "step": 69801
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7788311839103699,
      "learning_rate": 0.00047407401631225694,
      "loss": 2.3922,
      "step": 69802
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7043195366859436,
      "learning_rate": 0.00047407068477553336,
      "loss": 2.4369,
      "step": 69803
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7529653310775757,
      "learning_rate": 0.00047406735320644673,
      "loss": 2.2583,
      "step": 69804
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7322495579719543,
      "learning_rate": 0.00047406402160499775,
      "loss": 2.094,
      "step": 69805
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7989083528518677,
      "learning_rate": 0.00047406068997118684,
      "loss": 2.1886,
      "step": 69806
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7723949551582336,
      "learning_rate": 0.0004740573583050148,
      "loss": 2.3302,
      "step": 69807
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.743983268737793,
      "learning_rate": 0.00047405402660648233,
      "loss": 2.0789,
      "step": 69808
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7677164673805237,
      "learning_rate": 0.0004740506948755899,
      "loss": 2.0906,
      "step": 69809
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.8368595838546753,
      "learning_rate": 0.00047404736311233806,
      "loss": 2.1834,
      "step": 69810
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.8570436239242554,
      "learning_rate": 0.0004740440313167276,
      "loss": 2.3385,
      "step": 69811
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.887312114238739,
      "learning_rate": 0.000474040699488759,
      "loss": 2.2044,
      "step": 69812
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7901234030723572,
      "learning_rate": 0.00047403736762843303,
      "loss": 2.2569,
      "step": 69813
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.9507498145103455,
      "learning_rate": 0.0004740340357357503,
      "loss": 2.2248,
      "step": 69814
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.8419762849807739,
      "learning_rate": 0.0004740307038107112,
      "loss": 2.4048,
      "step": 69815
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.9290878772735596,
      "learning_rate": 0.00047402737185331655,
      "loss": 2.2476,
      "step": 69816
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.8326717019081116,
      "learning_rate": 0.000474024039863567,
      "loss": 2.067,
      "step": 69817
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7580491304397583,
      "learning_rate": 0.00047402070784146303,
      "loss": 2.3335,
      "step": 69818
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.8344035744667053,
      "learning_rate": 0.0004740173757870053,
      "loss": 2.1461,
      "step": 69819
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.783048152923584,
      "learning_rate": 0.00047401404370019454,
      "loss": 2.0717,
      "step": 69820
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7585142850875854,
      "learning_rate": 0.0004740107115810312,
      "loss": 2.1963,
      "step": 69821
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.727576732635498,
      "learning_rate": 0.00047400737942951605,
      "loss": 2.2068,
      "step": 69822
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7780190706253052,
      "learning_rate": 0.0004740040472456496,
      "loss": 2.1279,
      "step": 69823
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.867569088935852,
      "learning_rate": 0.0004740007150294325,
      "loss": 2.3334,
      "step": 69824
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7406817078590393,
      "learning_rate": 0.0004739973827808654,
      "loss": 2.1721,
      "step": 69825
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7830644249916077,
      "learning_rate": 0.00047399405049994894,
      "loss": 2.4204,
      "step": 69826
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7404648065567017,
      "learning_rate": 0.00047399071818668366,
      "loss": 2.2749,
      "step": 69827
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7205949425697327,
      "learning_rate": 0.00047398738584107024,
      "loss": 2.2657,
      "step": 69828
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7437793612480164,
      "learning_rate": 0.00047398405346310926,
      "loss": 2.2846,
      "step": 69829
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.8220852017402649,
      "learning_rate": 0.00047398072105280144,
      "loss": 2.2099,
      "step": 69830
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.8031255602836609,
      "learning_rate": 0.0004739773886101472,
      "loss": 2.2583,
      "step": 69831
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.8205573558807373,
      "learning_rate": 0.0004739740561351474,
      "loss": 2.2042,
      "step": 69832
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7908097505569458,
      "learning_rate": 0.00047397072362780247,
      "loss": 2.2908,
      "step": 69833
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.8987963795661926,
      "learning_rate": 0.00047396739108811306,
      "loss": 2.1016,
      "step": 69834
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.9968023896217346,
      "learning_rate": 0.00047396405851607984,
      "loss": 2.2574,
      "step": 69835
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.888811469078064,
      "learning_rate": 0.0004739607259117035,
      "loss": 2.2478,
      "step": 69836
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7778355479240417,
      "learning_rate": 0.0004739573932749845,
      "loss": 2.2136,
      "step": 69837
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.8710489869117737,
      "learning_rate": 0.00047395406060592356,
      "loss": 2.0897,
      "step": 69838
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7647841572761536,
      "learning_rate": 0.0004739507279045213,
      "loss": 2.1762,
      "step": 69839
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7600715160369873,
      "learning_rate": 0.00047394739517077827,
      "loss": 1.9646,
      "step": 69840
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7251219749450684,
      "learning_rate": 0.00047394406240469514,
      "loss": 2.0349,
      "step": 69841
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7717054486274719,
      "learning_rate": 0.0004739407296062726,
      "loss": 2.4742,
      "step": 69842
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.8402735590934753,
      "learning_rate": 0.00047393739677551117,
      "loss": 2.2271,
      "step": 69843
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.823212206363678,
      "learning_rate": 0.00047393406391241146,
      "loss": 2.4237,
      "step": 69844
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7517461776733398,
      "learning_rate": 0.0004739307310169741,
      "loss": 2.0803,
      "step": 69845
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7417370676994324,
      "learning_rate": 0.00047392739808919975,
      "loss": 2.3555,
      "step": 69846
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7416616678237915,
      "learning_rate": 0.00047392406512908905,
      "loss": 1.9907,
      "step": 69847
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.9317192435264587,
      "learning_rate": 0.00047392073213664256,
      "loss": 2.3392,
      "step": 69848
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7470896244049072,
      "learning_rate": 0.00047391739911186093,
      "loss": 2.1882,
      "step": 69849
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7258160710334778,
      "learning_rate": 0.00047391406605474477,
      "loss": 2.1632,
      "step": 69850
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7447501420974731,
      "learning_rate": 0.0004739107329652947,
      "loss": 2.2868,
      "step": 69851
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7684193253517151,
      "learning_rate": 0.00047390739984351136,
      "loss": 2.403,
      "step": 69852
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7569887638092041,
      "learning_rate": 0.0004739040666893953,
      "loss": 2.0763,
      "step": 69853
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.0354959964752197,
      "learning_rate": 0.0004739007335029473,
      "loss": 2.1436,
      "step": 69854
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7658420205116272,
      "learning_rate": 0.00047389740028416777,
      "loss": 2.4735,
      "step": 69855
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.8694774508476257,
      "learning_rate": 0.00047389406703305737,
      "loss": 2.0933,
      "step": 69856
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7554697394371033,
      "learning_rate": 0.0004738907337496169,
      "loss": 2.406,
      "step": 69857
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7828671336174011,
      "learning_rate": 0.00047388740043384687,
      "loss": 2.3642,
      "step": 69858
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.6816598176956177,
      "learning_rate": 0.00047388406708574776,
      "loss": 2.3332,
      "step": 69859
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.8021595478057861,
      "learning_rate": 0.0004738807337053205,
      "loss": 2.1181,
      "step": 69860
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7309696078300476,
      "learning_rate": 0.0004738774002925654,
      "loss": 2.1374,
      "step": 69861
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7482591271400452,
      "learning_rate": 0.0004738740668474832,
      "loss": 2.2247,
      "step": 69862
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.6940858960151672,
      "learning_rate": 0.0004738707333700746,
      "loss": 2.0752,
      "step": 69863
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7069413065910339,
      "learning_rate": 0.00047386739986034015,
      "loss": 2.2151,
      "step": 69864
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7629536986351013,
      "learning_rate": 0.0004738640663182804,
      "loss": 2.2137,
      "step": 69865
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.6855977773666382,
      "learning_rate": 0.0004738607327438961,
      "loss": 2.0389,
      "step": 69866
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7270303964614868,
      "learning_rate": 0.0004738573991371878,
      "loss": 2.2594,
      "step": 69867
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.6914981007575989,
      "learning_rate": 0.00047385406549815606,
      "loss": 2.117,
      "step": 69868
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.800265908241272,
      "learning_rate": 0.00047385073182680163,
      "loss": 2.0998,
      "step": 69869
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7219738364219666,
      "learning_rate": 0.000473847398123125,
      "loss": 2.3525,
      "step": 69870
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.8745033144950867,
      "learning_rate": 0.000473844064387127,
      "loss": 2.2607,
      "step": 69871
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.8765631318092346,
      "learning_rate": 0.000473840730618808,
      "loss": 2.1891,
      "step": 69872
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7230373024940491,
      "learning_rate": 0.0004738373968181688,
      "loss": 2.0959,
      "step": 69873
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7118432521820068,
      "learning_rate": 0.0004738340629852099,
      "loss": 2.292,
      "step": 69874
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.9439979195594788,
      "learning_rate": 0.000473830729119932,
      "loss": 2.169,
      "step": 69875
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.782779335975647,
      "learning_rate": 0.00047382739522233556,
      "loss": 2.2071,
      "step": 69876
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7907453775405884,
      "learning_rate": 0.0004738240612924214,
      "loss": 2.49,
      "step": 69877
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.1684596538543701,
      "learning_rate": 0.0004738207273301901,
      "loss": 2.0521,
      "step": 69878
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.8182936906814575,
      "learning_rate": 0.00047381739333564224,
      "loss": 2.2075,
      "step": 69879
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.076111912727356,
      "learning_rate": 0.0004738140593087784,
      "loss": 2.1623,
      "step": 69880
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.0565704107284546,
      "learning_rate": 0.0004738107252495994,
      "loss": 2.2084,
      "step": 69881
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.9042574167251587,
      "learning_rate": 0.0004738073911581055,
      "loss": 2.2023,
      "step": 69882
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.5121263265609741,
      "learning_rate": 0.00047380405703429757,
      "loss": 2.1548,
      "step": 69883
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7550384998321533,
      "learning_rate": 0.00047380072287817623,
      "loss": 2.3556,
      "step": 69884
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7635873556137085,
      "learning_rate": 0.0004737973886897421,
      "loss": 2.3682,
      "step": 69885
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.9123274087905884,
      "learning_rate": 0.00047379405446899565,
      "loss": 2.0752,
      "step": 69886
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7244894504547119,
      "learning_rate": 0.0004737907202159377,
      "loss": 2.2118,
      "step": 69887
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.8183499574661255,
      "learning_rate": 0.00047378738593056873,
      "loss": 2.379,
      "step": 69888
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7914809584617615,
      "learning_rate": 0.00047378405161288935,
      "loss": 2.3363,
      "step": 69889
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7161004543304443,
      "learning_rate": 0.00047378071726290037,
      "loss": 2.2923,
      "step": 69890
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7648439407348633,
      "learning_rate": 0.00047377738288060216,
      "loss": 2.1903,
      "step": 69891
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7474831342697144,
      "learning_rate": 0.0004737740484659954,
      "loss": 2.3566,
      "step": 69892
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.8254475593566895,
      "learning_rate": 0.000473770714019081,
      "loss": 2.066,
      "step": 69893
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.8422552943229675,
      "learning_rate": 0.0004737673795398591,
      "loss": 2.1038,
      "step": 69894
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7729390263557434,
      "learning_rate": 0.00047376404502833067,
      "loss": 1.9587,
      "step": 69895
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7335048317909241,
      "learning_rate": 0.00047376071048449615,
      "loss": 2.0374,
      "step": 69896
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.9190917015075684,
      "learning_rate": 0.0004737573759083564,
      "loss": 2.2251,
      "step": 69897
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.8205370903015137,
      "learning_rate": 0.00047375404129991174,
      "loss": 2.1958,
      "step": 69898
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7241714000701904,
      "learning_rate": 0.000473750706659163,
      "loss": 2.0859,
      "step": 69899
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7905612587928772,
      "learning_rate": 0.00047374737198611065,
      "loss": 2.3184,
      "step": 69900
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.9026476144790649,
      "learning_rate": 0.0004737440372807554,
      "loss": 2.487,
      "step": 69901
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7534489631652832,
      "learning_rate": 0.00047374070254309786,
      "loss": 1.9741,
      "step": 69902
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.8692417740821838,
      "learning_rate": 0.0004737373677731387,
      "loss": 2.4963,
      "step": 69903
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.8216286897659302,
      "learning_rate": 0.0004737340329708784,
      "loss": 2.1008,
      "step": 69904
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.0496892929077148,
      "learning_rate": 0.0004737306981363177,
      "loss": 2.5472,
      "step": 69905
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.8438763618469238,
      "learning_rate": 0.0004737273632694572,
      "loss": 2.3337,
      "step": 69906
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7685381174087524,
      "learning_rate": 0.0004737240283702975,
      "loss": 2.0999,
      "step": 69907
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7313833236694336,
      "learning_rate": 0.0004737206934388392,
      "loss": 1.9397,
      "step": 69908
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7927945852279663,
      "learning_rate": 0.0004737173584750829,
      "loss": 2.2033,
      "step": 69909
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7557762265205383,
      "learning_rate": 0.00047371402347902935,
      "loss": 2.0783,
      "step": 69910
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7392865419387817,
      "learning_rate": 0.0004737106884506791,
      "loss": 2.1306,
      "step": 69911
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7741430401802063,
      "learning_rate": 0.00047370735339003265,
      "loss": 2.2077,
      "step": 69912
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.8406762480735779,
      "learning_rate": 0.0004737040182970908,
      "loss": 2.2054,
      "step": 69913
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.9239987134933472,
      "learning_rate": 0.000473700683171854,
      "loss": 2.3011,
      "step": 69914
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7922035455703735,
      "learning_rate": 0.00047369734801432306,
      "loss": 2.201,
      "step": 69915
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7718561887741089,
      "learning_rate": 0.0004736940128244985,
      "loss": 2.3383,
      "step": 69916
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.8256416320800781,
      "learning_rate": 0.00047369067760238097,
      "loss": 2.2773,
      "step": 69917
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.9450151324272156,
      "learning_rate": 0.00047368734234797097,
      "loss": 2.1199,
      "step": 69918
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7524948716163635,
      "learning_rate": 0.00047368400706126925,
      "loss": 2.2474,
      "step": 69919
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.9369596242904663,
      "learning_rate": 0.0004736806717422765,
      "loss": 2.1571,
      "step": 69920
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.016663908958435,
      "learning_rate": 0.0004736773363909931,
      "loss": 2.152,
      "step": 69921
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7001869678497314,
      "learning_rate": 0.00047367400100741983,
      "loss": 2.1102,
      "step": 69922
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.9371299147605896,
      "learning_rate": 0.0004736706655915573,
      "loss": 2.1348,
      "step": 69923
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.8419358730316162,
      "learning_rate": 0.00047366733014340615,
      "loss": 2.3434,
      "step": 69924
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.765097975730896,
      "learning_rate": 0.00047366399466296694,
      "loss": 2.383,
      "step": 69925
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7732841968536377,
      "learning_rate": 0.0004736606591502403,
      "loss": 2.1398,
      "step": 69926
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7525808811187744,
      "learning_rate": 0.00047365732360522687,
      "loss": 2.0554,
      "step": 69927
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7986246943473816,
      "learning_rate": 0.0004736539880279273,
      "loss": 2.4438,
      "step": 69928
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7290304899215698,
      "learning_rate": 0.0004736506524183421,
      "loss": 1.9792,
      "step": 69929
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.8518460988998413,
      "learning_rate": 0.0004736473167764721,
      "loss": 2.0681,
      "step": 69930
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.8102447390556335,
      "learning_rate": 0.00047364398110231765,
      "loss": 2.2185,
      "step": 69931
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.783574640750885,
      "learning_rate": 0.00047364064539587953,
      "loss": 2.2648,
      "step": 69932
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7060657739639282,
      "learning_rate": 0.0004736373096571584,
      "loss": 2.2124,
      "step": 69933
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7432608008384705,
      "learning_rate": 0.00047363397388615477,
      "loss": 2.3065,
      "step": 69934
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7961665391921997,
      "learning_rate": 0.0004736306380828693,
      "loss": 2.4289,
      "step": 69935
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.755743145942688,
      "learning_rate": 0.00047362730224730266,
      "loss": 2.2903,
      "step": 69936
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.4018746614456177,
      "learning_rate": 0.00047362396637945536,
      "loss": 2.3162,
      "step": 69937
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.6935610771179199,
      "learning_rate": 0.0004736206304793281,
      "loss": 2.3262,
      "step": 69938
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.8078490495681763,
      "learning_rate": 0.00047361729454692154,
      "loss": 2.0257,
      "step": 69939
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7702012062072754,
      "learning_rate": 0.0004736139585822362,
      "loss": 2.1952,
      "step": 69940
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7025687098503113,
      "learning_rate": 0.00047361062258527274,
      "loss": 2.0971,
      "step": 69941
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.8390244245529175,
      "learning_rate": 0.0004736072865560319,
      "loss": 2.3931,
      "step": 69942
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.8814591765403748,
      "learning_rate": 0.00047360395049451405,
      "loss": 2.1231,
      "step": 69943
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.8210168480873108,
      "learning_rate": 0.00047360061440072,
      "loss": 2.2315,
      "step": 69944
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.846750795841217,
      "learning_rate": 0.0004735972782746504,
      "loss": 2.3447,
      "step": 69945
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.871008038520813,
      "learning_rate": 0.0004735939421163056,
      "loss": 2.1544,
      "step": 69946
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.8919159770011902,
      "learning_rate": 0.0004735906059256865,
      "loss": 2.034,
      "step": 69947
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7491233944892883,
      "learning_rate": 0.00047358726970279377,
      "loss": 2.0367,
      "step": 69948
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.8715978264808655,
      "learning_rate": 0.00047358393344762776,
      "loss": 2.2558,
      "step": 69949
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.9087792038917542,
      "learning_rate": 0.00047358059716018917,
      "loss": 2.379,
      "step": 69950
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7502326369285583,
      "learning_rate": 0.00047357726084047876,
      "loss": 2.2157,
      "step": 69951
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7656316757202148,
      "learning_rate": 0.00047357392448849707,
      "loss": 2.0913,
      "step": 69952
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.8205066919326782,
      "learning_rate": 0.0004735705881042446,
      "loss": 2.2919,
      "step": 69953
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7520931363105774,
      "learning_rate": 0.0004735672516877222,
      "loss": 2.3438,
      "step": 69954
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7342482209205627,
      "learning_rate": 0.0004735639152389303,
      "loss": 2.2403,
      "step": 69955
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.9386430978775024,
      "learning_rate": 0.00047356057875786963,
      "loss": 2.1191,
      "step": 69956
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.8741094470024109,
      "learning_rate": 0.0004735572422445407,
      "loss": 2.0939,
      "step": 69957
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.8007768392562866,
      "learning_rate": 0.00047355390569894427,
      "loss": 2.4012,
      "step": 69958
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7966388463973999,
      "learning_rate": 0.0004735505691210809,
      "loss": 2.2223,
      "step": 69959
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.8315015435218811,
      "learning_rate": 0.00047354723251095125,
      "loss": 2.415,
      "step": 69960
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7688013315200806,
      "learning_rate": 0.0004735438958685558,
      "loss": 2.1706,
      "step": 69961
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.8030455112457275,
      "learning_rate": 0.00047354055919389523,
      "loss": 2.294,
      "step": 69962
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.8252500891685486,
      "learning_rate": 0.0004735372224869703,
      "loss": 2.158,
      "step": 69963
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.8056260347366333,
      "learning_rate": 0.0004735338857477815,
      "loss": 2.3309,
      "step": 69964
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7760405540466309,
      "learning_rate": 0.0004735305489763294,
      "loss": 2.1701,
      "step": 69965
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.9842209219932556,
      "learning_rate": 0.00047352721217261477,
      "loss": 2.3563,
      "step": 69966
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7462141513824463,
      "learning_rate": 0.00047352387533663813,
      "loss": 2.2425,
      "step": 69967
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.8363586068153381,
      "learning_rate": 0.0004735205384684002,
      "loss": 2.2876,
      "step": 69968
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7180966734886169,
      "learning_rate": 0.00047351720156790144,
      "loss": 2.1516,
      "step": 69969
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7345911264419556,
      "learning_rate": 0.0004735138646351425,
      "loss": 2.1575,
      "step": 69970
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.8713774085044861,
      "learning_rate": 0.00047351052767012424,
      "loss": 2.2535,
      "step": 69971
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.8607679605484009,
      "learning_rate": 0.00047350719067284694,
      "loss": 2.2353,
      "step": 69972
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7278862595558167,
      "learning_rate": 0.00047350385364331143,
      "loss": 2.304,
      "step": 69973
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.8897144198417664,
      "learning_rate": 0.00047350051658151825,
      "loss": 2.3687,
      "step": 69974
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.8084754347801208,
      "learning_rate": 0.0004734971794874681,
      "loss": 2.3272,
      "step": 69975
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7054635286331177,
      "learning_rate": 0.00047349384236116163,
      "loss": 2.0948,
      "step": 69976
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7741990089416504,
      "learning_rate": 0.0004734905052025992,
      "loss": 2.1925,
      "step": 69977
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.8259900808334351,
      "learning_rate": 0.00047348716801178165,
      "loss": 2.1413,
      "step": 69978
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7676094770431519,
      "learning_rate": 0.0004734838307887096,
      "loss": 2.1242,
      "step": 69979
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7374207973480225,
      "learning_rate": 0.0004734804935333837,
      "loss": 2.1679,
      "step": 69980
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7238060235977173,
      "learning_rate": 0.00047347715624580436,
      "loss": 2.3319,
      "step": 69981
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.8149900436401367,
      "learning_rate": 0.00047347381892597247,
      "loss": 2.419,
      "step": 69982
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7634815573692322,
      "learning_rate": 0.0004734704815738885,
      "loss": 2.1092,
      "step": 69983
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7363628149032593,
      "learning_rate": 0.00047346714418955297,
      "loss": 2.3361,
      "step": 69984
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.761290967464447,
      "learning_rate": 0.00047346380677296677,
      "loss": 2.1802,
      "step": 69985
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7289736270904541,
      "learning_rate": 0.00047346046932413027,
      "loss": 2.2302,
      "step": 69986
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.6831058859825134,
      "learning_rate": 0.00047345713184304423,
      "loss": 2.1342,
      "step": 69987
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7578179836273193,
      "learning_rate": 0.00047345379432970936,
      "loss": 2.1378,
      "step": 69988
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.8170748353004456,
      "learning_rate": 0.000473450456784126,
      "loss": 2.2614,
      "step": 69989
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.8090552687644958,
      "learning_rate": 0.0004734471192062949,
      "loss": 2.184,
      "step": 69990
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7807788848876953,
      "learning_rate": 0.0004734437815962169,
      "loss": 2.6033,
      "step": 69991
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.766985297203064,
      "learning_rate": 0.0004734404439538922,
      "loss": 2.3052,
      "step": 69992
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7005562782287598,
      "learning_rate": 0.00047343710627932177,
      "loss": 2.0749,
      "step": 69993
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7264320254325867,
      "learning_rate": 0.0004734337685725061,
      "loss": 2.309,
      "step": 69994
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7359350323677063,
      "learning_rate": 0.00047343043083344585,
      "loss": 2.3115,
      "step": 69995
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7341690063476562,
      "learning_rate": 0.0004734270930621415,
      "loss": 2.0782,
      "step": 69996
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7806341648101807,
      "learning_rate": 0.00047342375525859384,
      "loss": 2.1781,
      "step": 69997
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7146846652030945,
      "learning_rate": 0.0004734204174228034,
      "loss": 2.2093,
      "step": 69998
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.8367272615432739,
      "learning_rate": 0.00047341707955477093,
      "loss": 2.2894,
      "step": 69999
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7787534594535828,
      "learning_rate": 0.0004734137416544969,
      "loss": 2.2485,
      "step": 70000
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.715561032295227,
      "learning_rate": 0.00047341040372198196,
      "loss": 2.1026,
      "step": 70001
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7141337394714355,
      "learning_rate": 0.00047340706575722667,
      "loss": 2.1962,
      "step": 70002
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7789064645767212,
      "learning_rate": 0.00047340372776023186,
      "loss": 2.2899,
      "step": 70003
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7411941289901733,
      "learning_rate": 0.000473400389730998,
      "loss": 2.1839,
      "step": 70004
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7374838590621948,
      "learning_rate": 0.0004733970516695258,
      "loss": 2.3144,
      "step": 70005
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.9551958441734314,
      "learning_rate": 0.00047339371357581564,
      "loss": 2.179,
      "step": 70006
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7245554327964783,
      "learning_rate": 0.00047339037544986837,
      "loss": 2.222,
      "step": 70007
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7413139939308167,
      "learning_rate": 0.00047338703729168463,
      "loss": 2.3796,
      "step": 70008
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.8856233954429626,
      "learning_rate": 0.00047338369910126496,
      "loss": 2.2959,
      "step": 70009
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7394217848777771,
      "learning_rate": 0.00047338036087860995,
      "loss": 2.1941,
      "step": 70010
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.8741570711135864,
      "learning_rate": 0.00047337702262372026,
      "loss": 2.2113,
      "step": 70011
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.8172698020935059,
      "learning_rate": 0.0004733736843365966,
      "loss": 2.2984,
      "step": 70012
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7840721607208252,
      "learning_rate": 0.00047337034601723936,
      "loss": 2.1894,
      "step": 70013
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.8932340741157532,
      "learning_rate": 0.00047336700766564935,
      "loss": 2.2342,
      "step": 70014
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.728632390499115,
      "learning_rate": 0.00047336366928182725,
      "loss": 1.8681,
      "step": 70015
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7550404071807861,
      "learning_rate": 0.00047336033086577335,
      "loss": 2.0761,
      "step": 70016
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7046592235565186,
      "learning_rate": 0.0004733569924174886,
      "loss": 2.0239,
      "step": 70017
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7640184760093689,
      "learning_rate": 0.00047335365393697367,
      "loss": 2.2064,
      "step": 70018
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7614951729774475,
      "learning_rate": 0.00047335031542422877,
      "loss": 2.0862,
      "step": 70019
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.8898398280143738,
      "learning_rate": 0.00047334697687925493,
      "loss": 2.3057,
      "step": 70020
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.8422633409500122,
      "learning_rate": 0.00047334363830205253,
      "loss": 2.2685,
      "step": 70021
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7787482738494873,
      "learning_rate": 0.0004733402996926224,
      "loss": 2.2303,
      "step": 70022
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.8197922706604004,
      "learning_rate": 0.0004733369610509648,
      "loss": 2.1751,
      "step": 70023
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.8465510606765747,
      "learning_rate": 0.0004733336223770808,
      "loss": 2.3539,
      "step": 70024
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.6987722516059875,
      "learning_rate": 0.00047333028367097077,
      "loss": 2.1885,
      "step": 70025
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.8155287504196167,
      "learning_rate": 0.00047332694493263527,
      "loss": 2.1732,
      "step": 70026
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7191153168678284,
      "learning_rate": 0.0004733236061620751,
      "loss": 2.2632,
      "step": 70027
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.0066787004470825,
      "learning_rate": 0.0004733202673592909,
      "loss": 2.1125,
      "step": 70028
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7410467267036438,
      "learning_rate": 0.000473316928524283,
      "loss": 2.0305,
      "step": 70029
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.9027765989303589,
      "learning_rate": 0.00047331358965705224,
      "loss": 2.2554,
      "step": 70030
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7184259295463562,
      "learning_rate": 0.0004733102507575993,
      "loss": 2.0286,
      "step": 70031
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7898582220077515,
      "learning_rate": 0.0004733069118259247,
      "loss": 2.4071,
      "step": 70032
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.8171802163124084,
      "learning_rate": 0.0004733035728620291,
      "loss": 2.2087,
      "step": 70033
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.8304087519645691,
      "learning_rate": 0.00047330023386591303,
      "loss": 2.3117,
      "step": 70034
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.8627153038978577,
      "learning_rate": 0.00047329689483757714,
      "loss": 2.0586,
      "step": 70035
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7196007370948792,
      "learning_rate": 0.0004732935557770221,
      "loss": 2.2414,
      "step": 70036
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.8289200067520142,
      "learning_rate": 0.0004732902166842487,
      "loss": 2.4129,
      "step": 70037
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.8382450938224792,
      "learning_rate": 0.00047328687755925715,
      "loss": 2.0778,
      "step": 70038
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7539678812026978,
      "learning_rate": 0.0004732835384020484,
      "loss": 2.2215,
      "step": 70039
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7841105461120605,
      "learning_rate": 0.000473280199212623,
      "loss": 2.0799,
      "step": 70040
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.8298884034156799,
      "learning_rate": 0.00047327685999098144,
      "loss": 2.1378,
      "step": 70041
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.8238049149513245,
      "learning_rate": 0.0004732735207371245,
      "loss": 2.4203,
      "step": 70042
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.6796128749847412,
      "learning_rate": 0.0004732701814510527,
      "loss": 2.1207,
      "step": 70043
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.8436726331710815,
      "learning_rate": 0.0004732668421327668,
      "loss": 2.2073,
      "step": 70044
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7945609092712402,
      "learning_rate": 0.00047326350278226723,
      "loss": 2.2363,
      "step": 70045
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7874115109443665,
      "learning_rate": 0.00047326016339955477,
      "loss": 2.0536,
      "step": 70046
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.8163573741912842,
      "learning_rate": 0.0004732568239846299,
      "loss": 2.2599,
      "step": 70047
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7652566432952881,
      "learning_rate": 0.00047325348453749333,
      "loss": 1.9914,
      "step": 70048
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7865658402442932,
      "learning_rate": 0.00047325014505814576,
      "loss": 2.3297,
      "step": 70049
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7372321486473083,
      "learning_rate": 0.0004732468055465875,
      "loss": 1.9416,
      "step": 70050
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.8408707976341248,
      "learning_rate": 0.0004732434660028196,
      "loss": 2.076,
      "step": 70051
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.1949148178100586,
      "learning_rate": 0.00047324012642684245,
      "loss": 2.2133,
      "step": 70052
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7699349522590637,
      "learning_rate": 0.00047323678681865665,
      "loss": 2.2237,
      "step": 70053
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.8176221251487732,
      "learning_rate": 0.00047323344717826283,
      "loss": 2.0977,
      "step": 70054
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.8633456230163574,
      "learning_rate": 0.0004732301075056617,
      "loss": 2.2175,
      "step": 70055
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7559570670127869,
      "learning_rate": 0.0004732267678008538,
      "loss": 1.9396,
      "step": 70056
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.78756183385849,
      "learning_rate": 0.00047322342806383974,
      "loss": 2.4052,
      "step": 70057
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7454450130462646,
      "learning_rate": 0.0004732200882946202,
      "loss": 2.1323,
      "step": 70058
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7421209812164307,
      "learning_rate": 0.0004732167484931958,
      "loss": 2.0729,
      "step": 70059
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.8409935235977173,
      "learning_rate": 0.00047321340865956707,
      "loss": 2.1427,
      "step": 70060
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7801926732063293,
      "learning_rate": 0.0004732100687937348,
      "loss": 2.1772,
      "step": 70061
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7659808993339539,
      "learning_rate": 0.00047320672889569944,
      "loss": 2.4437,
      "step": 70062
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7161610722541809,
      "learning_rate": 0.0004732033889654617,
      "loss": 2.143,
      "step": 70063
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.6985626816749573,
      "learning_rate": 0.0004732000490030222,
      "loss": 2.1478,
      "step": 70064
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7906396985054016,
      "learning_rate": 0.0004731967090083815,
      "loss": 2.2508,
      "step": 70065
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7487013339996338,
      "learning_rate": 0.00047319336898154026,
      "loss": 2.3516,
      "step": 70066
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7266811728477478,
      "learning_rate": 0.0004731900289224991,
      "loss": 2.3222,
      "step": 70067
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7613460421562195,
      "learning_rate": 0.0004731866888312587,
      "loss": 2.1874,
      "step": 70068
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7589494585990906,
      "learning_rate": 0.0004731833487078195,
      "loss": 2.2686,
      "step": 70069
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7037801742553711,
      "learning_rate": 0.00047318000855218237,
      "loss": 2.2597,
      "step": 70070
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7000300884246826,
      "learning_rate": 0.00047317666836434776,
      "loss": 2.3978,
      "step": 70071
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.9900071024894714,
      "learning_rate": 0.0004731733281443164,
      "loss": 2.1451,
      "step": 70072
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7715857625007629,
      "learning_rate": 0.0004731699878920888,
      "loss": 2.4209,
      "step": 70073
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.8257703185081482,
      "learning_rate": 0.0004731666476076656,
      "loss": 2.2925,
      "step": 70074
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.9023505449295044,
      "learning_rate": 0.0004731633072910475,
      "loss": 2.0787,
      "step": 70075
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7543454766273499,
      "learning_rate": 0.00047315996694223506,
      "loss": 2.2303,
      "step": 70076
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.8322516083717346,
      "learning_rate": 0.0004731566265612289,
      "loss": 2.0451,
      "step": 70077
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.9487784504890442,
      "learning_rate": 0.0004731532861480297,
      "loss": 2.4001,
      "step": 70078
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7626696825027466,
      "learning_rate": 0.000473149945702638,
      "loss": 2.0606,
      "step": 70079
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.8054561614990234,
      "learning_rate": 0.00047314660522505445,
      "loss": 2.1999,
      "step": 70080
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7952931523323059,
      "learning_rate": 0.0004731432647152797,
      "loss": 2.0025,
      "step": 70081
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.77418053150177,
      "learning_rate": 0.00047313992417331434,
      "loss": 2.2246,
      "step": 70082
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.8241018652915955,
      "learning_rate": 0.00047313658359915894,
      "loss": 2.0854,
      "step": 70083
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.8230209946632385,
      "learning_rate": 0.0004731332429928142,
      "loss": 2.1366,
      "step": 70084
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7645183801651001,
      "learning_rate": 0.0004731299023542808,
      "loss": 2.078,
      "step": 70085
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.777899980545044,
      "learning_rate": 0.00047312656168355924,
      "loss": 2.0619,
      "step": 70086
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7904110550880432,
      "learning_rate": 0.00047312322098065016,
      "loss": 2.168,
      "step": 70087
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.829972505569458,
      "learning_rate": 0.00047311988024555427,
      "loss": 2.1988,
      "step": 70088
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7975290417671204,
      "learning_rate": 0.00047311653947827205,
      "loss": 2.1994,
      "step": 70089
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.900647759437561,
      "learning_rate": 0.00047311319867880425,
      "loss": 2.4109,
      "step": 70090
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.8725314736366272,
      "learning_rate": 0.0004731098578471514,
      "loss": 2.2658,
      "step": 70091
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7908841371536255,
      "learning_rate": 0.00047310651698331417,
      "loss": 2.4279,
      "step": 70092
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7111347317695618,
      "learning_rate": 0.0004731031760872932,
      "loss": 1.9596,
      "step": 70093
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7828814387321472,
      "learning_rate": 0.000473099835159089,
      "loss": 2.2579,
      "step": 70094
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7705616354942322,
      "learning_rate": 0.0004730964941987024,
      "loss": 2.1035,
      "step": 70095
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.8346777558326721,
      "learning_rate": 0.0004730931532061338,
      "loss": 2.3309,
      "step": 70096
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7177346348762512,
      "learning_rate": 0.0004730898121813839,
      "loss": 2.2643,
      "step": 70097
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7635866403579712,
      "learning_rate": 0.0004730864711244534,
      "loss": 2.22,
      "step": 70098
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7381247878074646,
      "learning_rate": 0.00047308313003534286,
      "loss": 2.0745,
      "step": 70099
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.8005241751670837,
      "learning_rate": 0.00047307978891405285,
      "loss": 2.27,
      "step": 70100
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7235057353973389,
      "learning_rate": 0.00047307644776058415,
      "loss": 2.13,
      "step": 70101
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.772050678730011,
      "learning_rate": 0.0004730731065749371,
      "loss": 2.1987,
      "step": 70102
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7098063230514526,
      "learning_rate": 0.00047306976535711265,
      "loss": 1.9876,
      "step": 70103
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7400866150856018,
      "learning_rate": 0.00047306642410711115,
      "loss": 2.2459,
      "step": 70104
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.785578191280365,
      "learning_rate": 0.0004730630828249334,
      "loss": 2.3067,
      "step": 70105
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.8104516267776489,
      "learning_rate": 0.00047305974151057995,
      "loss": 2.4375,
      "step": 70106
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.857713520526886,
      "learning_rate": 0.00047305640016405144,
      "loss": 2.4138,
      "step": 70107
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7557674050331116,
      "learning_rate": 0.00047305305878534836,
      "loss": 2.2527,
      "step": 70108
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.858905017375946,
      "learning_rate": 0.0004730497173744716,
      "loss": 2.2424,
      "step": 70109
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.822568416595459,
      "learning_rate": 0.0004730463759314216,
      "loss": 2.374,
      "step": 70110
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7273069024085999,
      "learning_rate": 0.00047304303445619894,
      "loss": 2.3608,
      "step": 70111
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.674588680267334,
      "learning_rate": 0.00047303969294880437,
      "loss": 2.0136,
      "step": 70112
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.8537012338638306,
      "learning_rate": 0.00047303635140923843,
      "loss": 2.526,
      "step": 70113
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.710073709487915,
      "learning_rate": 0.0004730330098375018,
      "loss": 2.2521,
      "step": 70114
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.9291771054267883,
      "learning_rate": 0.000473029668233595,
      "loss": 2.1698,
      "step": 70115
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.8084146976470947,
      "learning_rate": 0.0004730263265975188,
      "loss": 2.0361,
      "step": 70116
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.8033562302589417,
      "learning_rate": 0.0004730229849292737,
      "loss": 1.9226,
      "step": 70117
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.8948477506637573,
      "learning_rate": 0.0004730196432288604,
      "loss": 2.085,
      "step": 70118
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7430166602134705,
      "learning_rate": 0.0004730163014962795,
      "loss": 2.3533,
      "step": 70119
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.807449221611023,
      "learning_rate": 0.0004730129597315315,
      "loss": 2.2186,
      "step": 70120
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.938256025314331,
      "learning_rate": 0.0004730096179346171,
      "loss": 2.2954,
      "step": 70121
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.8073556423187256,
      "learning_rate": 0.0004730062761055371,
      "loss": 2.317,
      "step": 70122
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7536136507987976,
      "learning_rate": 0.0004730029342442919,
      "loss": 2.2183,
      "step": 70123
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.6951804161071777,
      "learning_rate": 0.00047299959235088215,
      "loss": 2.1724,
      "step": 70124
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.8102818727493286,
      "learning_rate": 0.0004729962504253086,
      "loss": 2.4437,
      "step": 70125
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7609350681304932,
      "learning_rate": 0.0004729929084675716,
      "loss": 2.3149,
      "step": 70126
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7500419616699219,
      "learning_rate": 0.0004729895664776721,
      "loss": 2.2685,
      "step": 70127
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.8879930377006531,
      "learning_rate": 0.0004729862244556106,
      "loss": 2.164,
      "step": 70128
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7232918739318848,
      "learning_rate": 0.0004729828824013877,
      "loss": 1.9632,
      "step": 70129
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.8391820788383484,
      "learning_rate": 0.0004729795403150039,
      "loss": 2.1297,
      "step": 70130
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.768513023853302,
      "learning_rate": 0.00047297619819645996,
      "loss": 1.9371,
      "step": 70131
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7245152592658997,
      "learning_rate": 0.0004729728560457566,
      "loss": 2.249,
      "step": 70132
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.8586165308952332,
      "learning_rate": 0.00047296951386289413,
      "loss": 2.0593,
      "step": 70133
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7997061014175415,
      "learning_rate": 0.0004729661716478735,
      "loss": 2.0632,
      "step": 70134
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.8656125068664551,
      "learning_rate": 0.00047296282940069515,
      "loss": 2.2812,
      "step": 70135
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.8242332339286804,
      "learning_rate": 0.00047295948712135975,
      "loss": 2.3492,
      "step": 70136
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.8393843770027161,
      "learning_rate": 0.00047295614480986794,
      "loss": 2.3499,
      "step": 70137
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.8725525736808777,
      "learning_rate": 0.0004729528024662203,
      "loss": 2.2727,
      "step": 70138
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.8470275402069092,
      "learning_rate": 0.0004729494600904175,
      "loss": 2.2547,
      "step": 70139
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7214274406433105,
      "learning_rate": 0.0004729461176824601,
      "loss": 2.1119,
      "step": 70140
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7705059051513672,
      "learning_rate": 0.00047294277524234874,
      "loss": 2.138,
      "step": 70141
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7246375679969788,
      "learning_rate": 0.00047293943277008406,
      "loss": 2.3636,
      "step": 70142
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.8639471530914307,
      "learning_rate": 0.00047293609026566677,
      "loss": 2.1541,
      "step": 70143
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.8240920901298523,
      "learning_rate": 0.0004729327477290972,
      "loss": 2.321,
      "step": 70144
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.8191457986831665,
      "learning_rate": 0.00047292940516037636,
      "loss": 2.0783,
      "step": 70145
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7717273831367493,
      "learning_rate": 0.0004729260625595046,
      "loss": 2.2049,
      "step": 70146
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.8554778099060059,
      "learning_rate": 0.00047292271992648255,
      "loss": 2.4164,
      "step": 70147
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7573451399803162,
      "learning_rate": 0.0004729193772613109,
      "loss": 2.5286,
      "step": 70148
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7524252533912659,
      "learning_rate": 0.0004729160345639905,
      "loss": 2.407,
      "step": 70149
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.8547751903533936,
      "learning_rate": 0.00047291269183452156,
      "loss": 2.1121,
      "step": 70150
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7836843132972717,
      "learning_rate": 0.0004729093490729049,
      "loss": 2.0812,
      "step": 70151
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7893977165222168,
      "learning_rate": 0.0004729060062791412,
      "loss": 2.1266,
      "step": 70152
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.8227593302726746,
      "learning_rate": 0.00047290266345323086,
      "loss": 2.4292,
      "step": 70153
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7762755751609802,
      "learning_rate": 0.00047289932059517474,
      "loss": 2.1282,
      "step": 70154
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7355479598045349,
      "learning_rate": 0.00047289597770497344,
      "loss": 2.6272,
      "step": 70155
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.6957672238349915,
      "learning_rate": 0.00047289263478262743,
      "loss": 2.0322,
      "step": 70156
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7864566445350647,
      "learning_rate": 0.00047288929182813744,
      "loss": 2.4019,
      "step": 70157
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.8387801647186279,
      "learning_rate": 0.0004728859488415041,
      "loss": 2.073,
      "step": 70158
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7446726560592651,
      "learning_rate": 0.00047288260582272797,
      "loss": 2.0626,
      "step": 70159
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.8671659827232361,
      "learning_rate": 0.0004728792627718096,
      "loss": 2.0947,
      "step": 70160
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.827764630317688,
      "learning_rate": 0.0004728759196887499,
      "loss": 2.3283,
      "step": 70161
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.914764940738678,
      "learning_rate": 0.0004728725765735492,
      "loss": 2.1713,
      "step": 70162
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7539780139923096,
      "learning_rate": 0.0004728692334262082,
      "loss": 2.3474,
      "step": 70163
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7429863214492798,
      "learning_rate": 0.00047286589024672757,
      "loss": 2.1544,
      "step": 70164
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.9327199459075928,
      "learning_rate": 0.0004728625470351079,
      "loss": 2.2136,
      "step": 70165
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7002739310264587,
      "learning_rate": 0.0004728592037913499,
      "loss": 2.1614,
      "step": 70166
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.8560709357261658,
      "learning_rate": 0.00047285586051545403,
      "loss": 2.213,
      "step": 70167
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7982165217399597,
      "learning_rate": 0.000472852517207421,
      "loss": 2.159,
      "step": 70168
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.786582887172699,
      "learning_rate": 0.00047284917386725145,
      "loss": 2.246,
      "step": 70169
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.8701234459877014,
      "learning_rate": 0.0004728458304949459,
      "loss": 2.2548,
      "step": 70170
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.8953381776809692,
      "learning_rate": 0.0004728424870905052,
      "loss": 2.1377,
      "step": 70171
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7671186327934265,
      "learning_rate": 0.00047283914365392964,
      "loss": 2.141,
      "step": 70172
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.8092663884162903,
      "learning_rate": 0.0004728358001852202,
      "loss": 2.3627,
      "step": 70173
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7222695350646973,
      "learning_rate": 0.0004728324566843772,
      "loss": 2.1299,
      "step": 70174
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.8144539594650269,
      "learning_rate": 0.00047282911315140136,
      "loss": 2.2353,
      "step": 70175
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.8367838859558105,
      "learning_rate": 0.00047282576958629336,
      "loss": 2.4105,
      "step": 70176
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7929894328117371,
      "learning_rate": 0.0004728224259890539,
      "loss": 2.3541,
      "step": 70177
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.9549142718315125,
      "learning_rate": 0.0004728190823596833,
      "loss": 2.2984,
      "step": 70178
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7572000622749329,
      "learning_rate": 0.0004728157386981825,
      "loss": 2.118,
      "step": 70179
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7995558977127075,
      "learning_rate": 0.00047281239500455193,
      "loss": 2.1203,
      "step": 70180
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.9446442127227783,
      "learning_rate": 0.00047280905127879233,
      "loss": 2.118,
      "step": 70181
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.8489573001861572,
      "learning_rate": 0.0004728057075209043,
      "loss": 2.2965,
      "step": 70182
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.8330613374710083,
      "learning_rate": 0.0004728023637308883,
      "loss": 2.2576,
      "step": 70183
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.8790816068649292,
      "learning_rate": 0.00047279901990874516,
      "loss": 2.347,
      "step": 70184
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.9035241603851318,
      "learning_rate": 0.0004727956760544754,
      "loss": 2.2769,
      "step": 70185
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7285313606262207,
      "learning_rate": 0.00047279233216807965,
      "loss": 2.2081,
      "step": 70186
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.8184444904327393,
      "learning_rate": 0.00047278898824955855,
      "loss": 2.2486,
      "step": 70187
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.8838220834732056,
      "learning_rate": 0.0004727856442989127,
      "loss": 2.0495,
      "step": 70188
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.8358584642410278,
      "learning_rate": 0.0004727823003161428,
      "loss": 2.409,
      "step": 70189
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7777127623558044,
      "learning_rate": 0.0004727789563012494,
      "loss": 2.1815,
      "step": 70190
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.9177470207214355,
      "learning_rate": 0.0004727756122542331,
      "loss": 2.1282,
      "step": 70191
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.0315933227539062,
      "learning_rate": 0.00047277226817509454,
      "loss": 2.0573,
      "step": 70192
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7553253769874573,
      "learning_rate": 0.00047276892406383433,
      "loss": 2.0952,
      "step": 70193
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.0297614336013794,
      "learning_rate": 0.0004727655799204531,
      "loss": 2.3779,
      "step": 70194
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.8290573954582214,
      "learning_rate": 0.0004727622357449516,
      "loss": 2.4077,
      "step": 70195
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7665286660194397,
      "learning_rate": 0.00047275889153733033,
      "loss": 2.174,
      "step": 70196
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.8964084386825562,
      "learning_rate": 0.0004727555472975898,
      "loss": 2.2233,
      "step": 70197
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.9849116802215576,
      "learning_rate": 0.00047275220302573083,
      "loss": 2.3266,
      "step": 70198
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.792316198348999,
      "learning_rate": 0.000472748858721754,
      "loss": 1.9157,
      "step": 70199
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.739371657371521,
      "learning_rate": 0.00047274551438565983,
      "loss": 2.3516,
      "step": 70200
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.0111925601959229,
      "learning_rate": 0.0004727421700174491,
      "loss": 2.0471,
      "step": 70201
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.8970106244087219,
      "learning_rate": 0.00047273882561712225,
      "loss": 2.1341,
      "step": 70202
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.8586493730545044,
      "learning_rate": 0.00047273548118467994,
      "loss": 2.3272,
      "step": 70203
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.2171128988265991,
      "learning_rate": 0.00047273213672012303,
      "loss": 2.3742,
      "step": 70204
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.8178197741508484,
      "learning_rate": 0.00047272879222345183,
      "loss": 2.2451,
      "step": 70205
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.8034761548042297,
      "learning_rate": 0.00047272544769466705,
      "loss": 2.2501,
      "step": 70206
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7881807684898376,
      "learning_rate": 0.0004727221031337694,
      "loss": 2.2411,
      "step": 70207
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.002075433731079,
      "learning_rate": 0.00047271875854075946,
      "loss": 2.2267,
      "step": 70208
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7985948324203491,
      "learning_rate": 0.0004727154139156378,
      "loss": 2.0485,
      "step": 70209
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7805739641189575,
      "learning_rate": 0.00047271206925840513,
      "loss": 2.2022,
      "step": 70210
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.8206040859222412,
      "learning_rate": 0.000472708724569062,
      "loss": 2.4286,
      "step": 70211
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7834111452102661,
      "learning_rate": 0.00047270537984760906,
      "loss": 2.2118,
      "step": 70212
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.8716827034950256,
      "learning_rate": 0.000472702035094047,
      "loss": 2.3833,
      "step": 70213
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7984545230865479,
      "learning_rate": 0.0004726986903083763,
      "loss": 1.8483,
      "step": 70214
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.8445045948028564,
      "learning_rate": 0.00047269534549059764,
      "loss": 2.115,
      "step": 70215
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.768997073173523,
      "learning_rate": 0.0004726920006407117,
      "loss": 2.2921,
      "step": 70216
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7733954191207886,
      "learning_rate": 0.00047268865575871894,
      "loss": 2.085,
      "step": 70217
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.884436309337616,
      "learning_rate": 0.0004726853108446202,
      "loss": 2.1561,
      "step": 70218
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7374313473701477,
      "learning_rate": 0.00047268196589841607,
      "loss": 2.3369,
      "step": 70219
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.778730571269989,
      "learning_rate": 0.00047267862092010703,
      "loss": 2.2307,
      "step": 70220
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.8290085196495056,
      "learning_rate": 0.00047267527590969373,
      "loss": 2.1911,
      "step": 70221
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.782112717628479,
      "learning_rate": 0.0004726719308671769,
      "loss": 2.4694,
      "step": 70222
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.8308127522468567,
      "learning_rate": 0.000472668585792557,
      "loss": 2.1837,
      "step": 70223
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7945356369018555,
      "learning_rate": 0.0004726652406858349,
      "loss": 2.1387,
      "step": 70224
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.8165624737739563,
      "learning_rate": 0.00047266189554701104,
      "loss": 2.3183,
      "step": 70225
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7607318758964539,
      "learning_rate": 0.000472658550376086,
      "loss": 2.1127,
      "step": 70226
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.759627640247345,
      "learning_rate": 0.0004726552051730605,
      "loss": 2.4776,
      "step": 70227
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7272023558616638,
      "learning_rate": 0.0004726518599379352,
      "loss": 2.2837,
      "step": 70228
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7448817491531372,
      "learning_rate": 0.0004726485146707106,
      "loss": 2.3266,
      "step": 70229
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.8247932195663452,
      "learning_rate": 0.00047264516937138736,
      "loss": 2.1559,
      "step": 70230
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.8489683270454407,
      "learning_rate": 0.00047264182403996614,
      "loss": 2.2743,
      "step": 70231
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.8077049255371094,
      "learning_rate": 0.0004726384786764476,
      "loss": 2.2844,
      "step": 70232
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.828702986240387,
      "learning_rate": 0.0004726351332808323,
      "loss": 2.4432,
      "step": 70233
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.842613697052002,
      "learning_rate": 0.0004726317878531208,
      "loss": 2.1455,
      "step": 70234
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.8270326256752014,
      "learning_rate": 0.0004726284423933139,
      "loss": 2.0816,
      "step": 70235
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.6987654566764832,
      "learning_rate": 0.000472625096901412,
      "loss": 2.3356,
      "step": 70236
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7765175700187683,
      "learning_rate": 0.0004726217513774159,
      "loss": 2.3014,
      "step": 70237
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7342493534088135,
      "learning_rate": 0.00047261840582132615,
      "loss": 2.3377,
      "step": 70238
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.8372297286987305,
      "learning_rate": 0.00047261506023314336,
      "loss": 2.5433,
      "step": 70239
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.9765235781669617,
      "learning_rate": 0.0004726117146128681,
      "loss": 2.2839,
      "step": 70240
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.8603883385658264,
      "learning_rate": 0.00047260836896050127,
      "loss": 2.199,
      "step": 70241
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.840719997882843,
      "learning_rate": 0.0004726050232760431,
      "loss": 2.4343,
      "step": 70242
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.8312703371047974,
      "learning_rate": 0.00047260167755949447,
      "loss": 2.2482,
      "step": 70243
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.8444414734840393,
      "learning_rate": 0.000472598331810856,
      "loss": 2.4525,
      "step": 70244
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.785799503326416,
      "learning_rate": 0.0004725949860301281,
      "loss": 2.0663,
      "step": 70245
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7516947984695435,
      "learning_rate": 0.0004725916402173116,
      "loss": 2.1728,
      "step": 70246
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7533640265464783,
      "learning_rate": 0.00047258829437240706,
      "loss": 2.0017,
      "step": 70247
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.8230287432670593,
      "learning_rate": 0.00047258494849541504,
      "loss": 2.3021,
      "step": 70248
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7899177670478821,
      "learning_rate": 0.0004725816025863363,
      "loss": 2.2874,
      "step": 70249
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.8220016956329346,
      "learning_rate": 0.0004725782566451714,
      "loss": 2.2311,
      "step": 70250
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7694267630577087,
      "learning_rate": 0.0004725749106719209,
      "loss": 2.336,
      "step": 70251
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.8430315852165222,
      "learning_rate": 0.00047257156466658543,
      "loss": 2.2773,
      "step": 70252
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7870548963546753,
      "learning_rate": 0.00047256821862916575,
      "loss": 2.1522,
      "step": 70253
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.6834592223167419,
      "learning_rate": 0.0004725648725596623,
      "loss": 2.3272,
      "step": 70254
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.8892271518707275,
      "learning_rate": 0.00047256152645807573,
      "loss": 2.1828,
      "step": 70255
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.8184133768081665,
      "learning_rate": 0.00047255818032440687,
      "loss": 2.2913,
      "step": 70256
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7585616707801819,
      "learning_rate": 0.000472554834158656,
      "loss": 2.2652,
      "step": 70257
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.6782912015914917,
      "learning_rate": 0.0004725514879608241,
      "loss": 2.153,
      "step": 70258
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7448232173919678,
      "learning_rate": 0.00047254814173091156,
      "loss": 2.3892,
      "step": 70259
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.743021547794342,
      "learning_rate": 0.000472544795468919,
      "loss": 2.3032,
      "step": 70260
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.8278744220733643,
      "learning_rate": 0.0004725414491748472,
      "loss": 2.252,
      "step": 70261
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7605255842208862,
      "learning_rate": 0.0004725381028486966,
      "loss": 2.2314,
      "step": 70262
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7542141079902649,
      "learning_rate": 0.00047253475649046796,
      "loss": 2.2211,
      "step": 70263
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7479299306869507,
      "learning_rate": 0.00047253141010016183,
      "loss": 2.2297,
      "step": 70264
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7953758835792542,
      "learning_rate": 0.0004725280636777789,
      "loss": 1.9314,
      "step": 70265
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7207611203193665,
      "learning_rate": 0.0004725247172233197,
      "loss": 2.1753,
      "step": 70266
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.811221718788147,
      "learning_rate": 0.00047252137073678497,
      "loss": 2.2027,
      "step": 70267
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.8045717477798462,
      "learning_rate": 0.0004725180242181752,
      "loss": 2.2505,
      "step": 70268
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7763455510139465,
      "learning_rate": 0.000472514677667491,
      "loss": 2.409,
      "step": 70269
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.8088997006416321,
      "learning_rate": 0.0004725113310847332,
      "loss": 2.3438,
      "step": 70270
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7756061553955078,
      "learning_rate": 0.00047250798446990224,
      "loss": 2.2581,
      "step": 70271
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7049912214279175,
      "learning_rate": 0.0004725046378229988,
      "loss": 2.2005,
      "step": 70272
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7404457330703735,
      "learning_rate": 0.00047250129114402345,
      "loss": 2.2023,
      "step": 70273
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7336068153381348,
      "learning_rate": 0.00047249794443297686,
      "loss": 2.3117,
      "step": 70274
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.776874303817749,
      "learning_rate": 0.0004724945976898597,
      "loss": 1.9772,
      "step": 70275
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.8007516860961914,
      "learning_rate": 0.00047249125091467246,
      "loss": 2.2339,
      "step": 70276
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7783315181732178,
      "learning_rate": 0.00047248790410741595,
      "loss": 2.1665,
      "step": 70277
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.6963658332824707,
      "learning_rate": 0.00047248455726809057,
      "loss": 2.2658,
      "step": 70278
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7565383315086365,
      "learning_rate": 0.0004724812103966971,
      "loss": 2.1727,
      "step": 70279
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.8711389303207397,
      "learning_rate": 0.0004724778634932361,
      "loss": 2.1352,
      "step": 70280
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7828083634376526,
      "learning_rate": 0.00047247451655770825,
      "loss": 2.2036,
      "step": 70281
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7967708706855774,
      "learning_rate": 0.00047247116959011406,
      "loss": 1.9018,
      "step": 70282
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.9720958471298218,
      "learning_rate": 0.00047246782259045437,
      "loss": 1.8079,
      "step": 70283
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.8127384185791016,
      "learning_rate": 0.0004724644755587295,
      "loss": 2.1259,
      "step": 70284
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.9628897309303284,
      "learning_rate": 0.0004724611284949403,
      "loss": 2.0869,
      "step": 70285
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.8656514286994934,
      "learning_rate": 0.0004724577813990873,
      "loss": 2.0149,
      "step": 70286
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.8649395704269409,
      "learning_rate": 0.00047245443427117113,
      "loss": 2.3268,
      "step": 70287
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7805838584899902,
      "learning_rate": 0.0004724510871111925,
      "loss": 2.1175,
      "step": 70288
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.9773650765419006,
      "learning_rate": 0.00047244773991915185,
      "loss": 2.1622,
      "step": 70289
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.1827272176742554,
      "learning_rate": 0.00047244439269505,
      "loss": 1.9186,
      "step": 70290
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.8075300455093384,
      "learning_rate": 0.0004724410454388874,
      "loss": 2.4146,
      "step": 70291
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.9107685685157776,
      "learning_rate": 0.00047243769815066486,
      "loss": 2.2566,
      "step": 70292
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.88050776720047,
      "learning_rate": 0.00047243435083038285,
      "loss": 2.1171,
      "step": 70293
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.9154177904129028,
      "learning_rate": 0.00047243100347804195,
      "loss": 2.281,
      "step": 70294
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7128692865371704,
      "learning_rate": 0.000472427656093643,
      "loss": 2.0465,
      "step": 70295
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.9566665887832642,
      "learning_rate": 0.0004724243086771864,
      "loss": 2.3394,
      "step": 70296
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.0155341625213623,
      "learning_rate": 0.00047242096122867297,
      "loss": 2.2399,
      "step": 70297
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.8520192503929138,
      "learning_rate": 0.00047241761374810316,
      "loss": 2.2379,
      "step": 70298
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.8170853853225708,
      "learning_rate": 0.00047241426623547767,
      "loss": 2.082,
      "step": 70299
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.9351327419281006,
      "learning_rate": 0.00047241091869079703,
      "loss": 2.2242,
      "step": 70300
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7829923033714294,
      "learning_rate": 0.00047240757111406207,
      "loss": 2.2539,
      "step": 70301
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7371866106987,
      "learning_rate": 0.00047240422350527326,
      "loss": 2.2242,
      "step": 70302
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.9127691984176636,
      "learning_rate": 0.0004724008758644312,
      "loss": 2.1404,
      "step": 70303
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7560968399047852,
      "learning_rate": 0.0004723975281915366,
      "loss": 2.1683,
      "step": 70304
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.8043095469474792,
      "learning_rate": 0.00047239418048659003,
      "loss": 2.2554,
      "step": 70305
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.8578006625175476,
      "learning_rate": 0.00047239083274959215,
      "loss": 2.2569,
      "step": 70306
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7444345355033875,
      "learning_rate": 0.0004723874849805435,
      "loss": 2.2551,
      "step": 70307
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7525861859321594,
      "learning_rate": 0.0004723841371794449,
      "loss": 2.3646,
      "step": 70308
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7857362627983093,
      "learning_rate": 0.00047238078934629677,
      "loss": 2.2645,
      "step": 70309
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.9092007279396057,
      "learning_rate": 0.0004723774414810997,
      "loss": 2.248,
      "step": 70310
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7769802808761597,
      "learning_rate": 0.0004723740935838545,
      "loss": 2.2528,
      "step": 70311
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7455687522888184,
      "learning_rate": 0.0004723707456545617,
      "loss": 2.2196,
      "step": 70312
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.8263554573059082,
      "learning_rate": 0.0004723673976932219,
      "loss": 2.3656,
      "step": 70313
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7780152559280396,
      "learning_rate": 0.0004723640496998358,
      "loss": 2.3416,
      "step": 70314
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7379417419433594,
      "learning_rate": 0.0004723607016744039,
      "loss": 2.0448,
      "step": 70315
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7431167960166931,
      "learning_rate": 0.00047235735361692694,
      "loss": 2.114,
      "step": 70316
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7612056732177734,
      "learning_rate": 0.00047235400552740544,
      "loss": 2.2138,
      "step": 70317
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7689367532730103,
      "learning_rate": 0.0004723506574058401,
      "loss": 2.1979,
      "step": 70318
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.8055710792541504,
      "learning_rate": 0.00047234730925223154,
      "loss": 2.2756,
      "step": 70319
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.8119636178016663,
      "learning_rate": 0.0004723439610665804,
      "loss": 2.4481,
      "step": 70320
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7032146453857422,
      "learning_rate": 0.0004723406128488872,
      "loss": 2.2428,
      "step": 70321
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.8450940251350403,
      "learning_rate": 0.0004723372645991526,
      "loss": 2.3673,
      "step": 70322
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7793616056442261,
      "learning_rate": 0.00047233391631737735,
      "loss": 2.1488,
      "step": 70323
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.8869086503982544,
      "learning_rate": 0.0004723305680035619,
      "loss": 2.3525,
      "step": 70324
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7800024151802063,
      "learning_rate": 0.0004723272196577069,
      "loss": 2.2548,
      "step": 70325
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.8544995188713074,
      "learning_rate": 0.00047232387127981313,
      "loss": 2.2221,
      "step": 70326
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.8689990043640137,
      "learning_rate": 0.00047232052286988105,
      "loss": 1.9599,
      "step": 70327
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.817148745059967,
      "learning_rate": 0.00047231717442791134,
      "loss": 2.2884,
      "step": 70328
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.012308955192566,
      "learning_rate": 0.0004723138259539046,
      "loss": 2.1497,
      "step": 70329
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7491101026535034,
      "learning_rate": 0.0004723104774478615,
      "loss": 2.1414,
      "step": 70330
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7472119331359863,
      "learning_rate": 0.0004723071289097825,
      "loss": 2.3377,
      "step": 70331
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.834248960018158,
      "learning_rate": 0.0004723037803396685,
      "loss": 2.216,
      "step": 70332
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.743330717086792,
      "learning_rate": 0.0004723004317375199,
      "loss": 2.1881,
      "step": 70333
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7899813652038574,
      "learning_rate": 0.0004722970831033374,
      "loss": 2.2189,
      "step": 70334
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7794855833053589,
      "learning_rate": 0.0004722937344371218,
      "loss": 2.0506,
      "step": 70335
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.6881862878799438,
      "learning_rate": 0.00047229038573887334,
      "loss": 2.1346,
      "step": 70336
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7626510262489319,
      "learning_rate": 0.0004722870370085928,
      "loss": 2.2805,
      "step": 70337
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7539385557174683,
      "learning_rate": 0.00047228368824628103,
      "loss": 2.2362,
      "step": 70338
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.8244007229804993,
      "learning_rate": 0.00047228033945193834,
      "loss": 2.4901,
      "step": 70339
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7940069437026978,
      "learning_rate": 0.0004722769906255655,
      "loss": 2.1056,
      "step": 70340
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.814424455165863,
      "learning_rate": 0.0004722736417671632,
      "loss": 2.0646,
      "step": 70341
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7546072602272034,
      "learning_rate": 0.00047227029287673193,
      "loss": 2.4782,
      "step": 70342
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7768387198448181,
      "learning_rate": 0.00047226694395427227,
      "loss": 2.1798,
      "step": 70343
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.8443103432655334,
      "learning_rate": 0.0004722635949997851,
      "loss": 2.4085,
      "step": 70344
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.98073810338974,
      "learning_rate": 0.00047226024601327074,
      "loss": 2.2038,
      "step": 70345
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.8094627261161804,
      "learning_rate": 0.00047225689699472996,
      "loss": 2.3155,
      "step": 70346
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7945579290390015,
      "learning_rate": 0.0004722535479441635,
      "loss": 2.2707,
      "step": 70347
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.9010443687438965,
      "learning_rate": 0.0004722501988615717,
      "loss": 2.36,
      "step": 70348
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.760167121887207,
      "learning_rate": 0.0004722468497469554,
      "loss": 2.2139,
      "step": 70349
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.8461949229240417,
      "learning_rate": 0.0004722435006003151,
      "loss": 2.0681,
      "step": 70350
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.8084971308708191,
      "learning_rate": 0.0004722401514216516,
      "loss": 2.1068,
      "step": 70351
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7344006299972534,
      "learning_rate": 0.00047223680221096526,
      "loss": 2.4518,
      "step": 70352
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.8099271059036255,
      "learning_rate": 0.00047223345296825705,
      "loss": 2.3306,
      "step": 70353
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.8162901401519775,
      "learning_rate": 0.0004722301036935272,
      "loss": 2.315,
      "step": 70354
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.8028533458709717,
      "learning_rate": 0.0004722267543867766,
      "loss": 2.1694,
      "step": 70355
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.8010020852088928,
      "learning_rate": 0.0004722234050480058,
      "loss": 2.3307,
      "step": 70356
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.8858641982078552,
      "learning_rate": 0.0004722200556772154,
      "loss": 2.2902,
      "step": 70357
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7865302562713623,
      "learning_rate": 0.00047221670627440606,
      "loss": 2.3919,
      "step": 70358
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.826645016670227,
      "learning_rate": 0.00047221335683957837,
      "loss": 2.3887,
      "step": 70359
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.765166699886322,
      "learning_rate": 0.00047221000737273294,
      "loss": 2.1952,
      "step": 70360
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7459118366241455,
      "learning_rate": 0.0004722066578738704,
      "loss": 2.2609,
      "step": 70361
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7132315635681152,
      "learning_rate": 0.00047220330834299147,
      "loss": 2.0542,
      "step": 70362
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7853586673736572,
      "learning_rate": 0.0004721999587800967,
      "loss": 2.3442,
      "step": 70363
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7743858098983765,
      "learning_rate": 0.0004721966091851866,
      "loss": 2.2891,
      "step": 70364
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7959640622138977,
      "learning_rate": 0.000472193259558262,
      "loss": 2.3796,
      "step": 70365
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7328070402145386,
      "learning_rate": 0.0004721899098993234,
      "loss": 2.5428,
      "step": 70366
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.9318662285804749,
      "learning_rate": 0.00047218656020837137,
      "loss": 2.2074,
      "step": 70367
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7196783423423767,
      "learning_rate": 0.0004721832104854067,
      "loss": 2.3653,
      "step": 70368
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7191541790962219,
      "learning_rate": 0.0004721798607304299,
      "loss": 2.2342,
      "step": 70369
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7467942237854004,
      "learning_rate": 0.0004721765109434416,
      "loss": 2.3321,
      "step": 70370
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7830678820610046,
      "learning_rate": 0.00047217316112444245,
      "loss": 2.2573,
      "step": 70371
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.9217341542243958,
      "learning_rate": 0.0004721698112734331,
      "loss": 2.3598,
      "step": 70372
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.8151479959487915,
      "learning_rate": 0.0004721664613904141,
      "loss": 2.2126,
      "step": 70373
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7926856279373169,
      "learning_rate": 0.00047216311147538605,
      "loss": 2.2986,
      "step": 70374
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.6907990574836731,
      "learning_rate": 0.0004721597615283497,
      "loss": 2.2403,
      "step": 70375
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7334228157997131,
      "learning_rate": 0.0004721564115493056,
      "loss": 1.9762,
      "step": 70376
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7253805994987488,
      "learning_rate": 0.0004721530615382542,
      "loss": 2.1297,
      "step": 70377
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.764266848564148,
      "learning_rate": 0.0004721497114951966,
      "loss": 2.3431,
      "step": 70378
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.8297220468521118,
      "learning_rate": 0.00047214636142013286,
      "loss": 2.2181,
      "step": 70379
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.8392030000686646,
      "learning_rate": 0.00047214301131306405,
      "loss": 2.3162,
      "step": 70380
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.8627558946609497,
      "learning_rate": 0.00047213966117399045,
      "loss": 2.0986,
      "step": 70381
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7942631840705872,
      "learning_rate": 0.00047213631100291294,
      "loss": 2.1627,
      "step": 70382
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.879610538482666,
      "learning_rate": 0.00047213296079983195,
      "loss": 2.3993,
      "step": 70383
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.8110666275024414,
      "learning_rate": 0.00047212961056474834,
      "loss": 2.1648,
      "step": 70384
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.8156863451004028,
      "learning_rate": 0.00047212626029766245,
      "loss": 2.2579,
      "step": 70385
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7471742033958435,
      "learning_rate": 0.0004721229099985751,
      "loss": 2.1751,
      "step": 70386
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.8169884085655212,
      "learning_rate": 0.0004721195596674868,
      "loss": 2.4859,
      "step": 70387
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.9864161610603333,
      "learning_rate": 0.00047211620930439827,
      "loss": 2.1782,
      "step": 70388
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7574982643127441,
      "learning_rate": 0.00047211285890931006,
      "loss": 2.2349,
      "step": 70389
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.8601399660110474,
      "learning_rate": 0.00047210950848222287,
      "loss": 2.2667,
      "step": 70390
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7865161299705505,
      "learning_rate": 0.0004721061580231372,
      "loss": 2.31,
      "step": 70391
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.783394455909729,
      "learning_rate": 0.0004721028075320538,
      "loss": 2.0753,
      "step": 70392
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7157632112503052,
      "learning_rate": 0.0004720994570089732,
      "loss": 2.2941,
      "step": 70393
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7684061527252197,
      "learning_rate": 0.00047209610645389613,
      "loss": 2.122,
      "step": 70394
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7618213891983032,
      "learning_rate": 0.0004720927558668231,
      "loss": 2.0433,
      "step": 70395
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.8157532215118408,
      "learning_rate": 0.0004720894052477548,
      "loss": 2.3191,
      "step": 70396
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7656708359718323,
      "learning_rate": 0.0004720860545966918,
      "loss": 2.1225,
      "step": 70397
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.8160353899002075,
      "learning_rate": 0.00047208270391363476,
      "loss": 2.1592,
      "step": 70398
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7655682563781738,
      "learning_rate": 0.0004720793531985843,
      "loss": 2.2077,
      "step": 70399
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7699452042579651,
      "learning_rate": 0.000472076002451541,
      "loss": 2.2978,
      "step": 70400
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.8271318078041077,
      "learning_rate": 0.00047207265167250554,
      "loss": 2.1833,
      "step": 70401
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7570432424545288,
      "learning_rate": 0.00047206930086147867,
      "loss": 2.0752,
      "step": 70402
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.8466007113456726,
      "learning_rate": 0.0004720659500184607,
      "loss": 2.217,
      "step": 70403
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.8226828575134277,
      "learning_rate": 0.00047206259914345244,
      "loss": 1.9593,
      "step": 70404
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.8653271794319153,
      "learning_rate": 0.0004720592482364546,
      "loss": 2.3202,
      "step": 70405
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.8699228167533875,
      "learning_rate": 0.00047205589729746754,
      "loss": 2.0423,
      "step": 70406
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.9140231013298035,
      "learning_rate": 0.00047205254632649217,
      "loss": 2.288,
      "step": 70407
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.707286536693573,
      "learning_rate": 0.00047204919532352894,
      "loss": 2.1938,
      "step": 70408
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7274122834205627,
      "learning_rate": 0.0004720458442885785,
      "loss": 2.1181,
      "step": 70409
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.9262213706970215,
      "learning_rate": 0.00047204249322164147,
      "loss": 2.0799,
      "step": 70410
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.8010216951370239,
      "learning_rate": 0.0004720391421227186,
      "loss": 2.1281,
      "step": 70411
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.9265974760055542,
      "learning_rate": 0.00047203579099181034,
      "loss": 2.3595,
      "step": 70412
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7849656343460083,
      "learning_rate": 0.00047203243982891737,
      "loss": 2.0694,
      "step": 70413
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7389267683029175,
      "learning_rate": 0.0004720290886340404,
      "loss": 2.0643,
      "step": 70414
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.8203298449516296,
      "learning_rate": 0.0004720257374071798,
      "loss": 2.3778,
      "step": 70415
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7154681086540222,
      "learning_rate": 0.00047202238614833645,
      "loss": 2.1992,
      "step": 70416
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7993866205215454,
      "learning_rate": 0.000472019034857511,
      "loss": 2.4345,
      "step": 70417
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7504928708076477,
      "learning_rate": 0.0004720156835347038,
      "loss": 2.1794,
      "step": 70418
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7529386878013611,
      "learning_rate": 0.00047201233217991574,
      "loss": 2.3538,
      "step": 70419
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7751399874687195,
      "learning_rate": 0.0004720089807931473,
      "loss": 2.1564,
      "step": 70420
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7988380193710327,
      "learning_rate": 0.00047200562937439923,
      "loss": 2.2066,
      "step": 70421
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7314075231552124,
      "learning_rate": 0.00047200227792367194,
      "loss": 2.2768,
      "step": 70422
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7182881236076355,
      "learning_rate": 0.00047199892644096626,
      "loss": 2.2651,
      "step": 70423
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.6940794587135315,
      "learning_rate": 0.00047199557492628275,
      "loss": 2.1966,
      "step": 70424
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7719023823738098,
      "learning_rate": 0.0004719922233796219,
      "loss": 2.1932,
      "step": 70425
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7418211102485657,
      "learning_rate": 0.0004719888718009846,
      "loss": 2.2226,
      "step": 70426
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7664254307746887,
      "learning_rate": 0.0004719855201903712,
      "loss": 2.3439,
      "step": 70427
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.8116010427474976,
      "learning_rate": 0.00047198216854778247,
      "loss": 2.2571,
      "step": 70428
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7587661743164062,
      "learning_rate": 0.000471978816873219,
      "loss": 2.1536,
      "step": 70429
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.8210648894309998,
      "learning_rate": 0.00047197546516668145,
      "loss": 2.1779,
      "step": 70430
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7101874351501465,
      "learning_rate": 0.0004719721134281705,
      "loss": 2.0956,
      "step": 70431
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7372727990150452,
      "learning_rate": 0.00047196876165768654,
      "loss": 2.0977,
      "step": 70432
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7841911911964417,
      "learning_rate": 0.0004719654098552304,
      "loss": 2.2932,
      "step": 70433
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7181555032730103,
      "learning_rate": 0.00047196205802080266,
      "loss": 2.2281,
      "step": 70434
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7358202934265137,
      "learning_rate": 0.00047195870615440396,
      "loss": 2.1452,
      "step": 70435
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.787665843963623,
      "learning_rate": 0.0004719553542560348,
      "loss": 2.2772,
      "step": 70436
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.8188977837562561,
      "learning_rate": 0.00047195200232569593,
      "loss": 2.423,
      "step": 70437
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.8122673034667969,
      "learning_rate": 0.00047194865036338796,
      "loss": 2.3293,
      "step": 70438
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7353386282920837,
      "learning_rate": 0.0004719452983691115,
      "loss": 2.2605,
      "step": 70439
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7852931022644043,
      "learning_rate": 0.00047194194634286716,
      "loss": 2.3344,
      "step": 70440
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7395468354225159,
      "learning_rate": 0.00047193859428465554,
      "loss": 2.4649,
      "step": 70441
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7589579820632935,
      "learning_rate": 0.0004719352421944773,
      "loss": 2.175,
      "step": 70442
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.8520625233650208,
      "learning_rate": 0.00047193189007233304,
      "loss": 1.9705,
      "step": 70443
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7514030933380127,
      "learning_rate": 0.0004719285379182234,
      "loss": 1.9881,
      "step": 70444
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7448879480361938,
      "learning_rate": 0.00047192518573214895,
      "loss": 2.02,
      "step": 70445
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.6934393644332886,
      "learning_rate": 0.00047192183351411047,
      "loss": 2.1217,
      "step": 70446
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7442917227745056,
      "learning_rate": 0.00047191848126410836,
      "loss": 2.1935,
      "step": 70447
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.8236722350120544,
      "learning_rate": 0.0004719151289821435,
      "loss": 2.131,
      "step": 70448
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7723570466041565,
      "learning_rate": 0.0004719117766682163,
      "loss": 2.2026,
      "step": 70449
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.875019371509552,
      "learning_rate": 0.0004719084243223274,
      "loss": 2.4579,
      "step": 70450
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.6684795022010803,
      "learning_rate": 0.00047190507194447756,
      "loss": 2.0801,
      "step": 70451
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.920594334602356,
      "learning_rate": 0.00047190171953466725,
      "loss": 2.0474,
      "step": 70452
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7898706793785095,
      "learning_rate": 0.0004718983670928972,
      "loss": 2.208,
      "step": 70453
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.6676896810531616,
      "learning_rate": 0.0004718950146191681,
      "loss": 2.0459,
      "step": 70454
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.8710930347442627,
      "learning_rate": 0.0004718916621134803,
      "loss": 1.9872,
      "step": 70455
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.9399438500404358,
      "learning_rate": 0.0004718883095758347,
      "loss": 2.3195,
      "step": 70456
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.8015550374984741,
      "learning_rate": 0.00047188495700623185,
      "loss": 2.0714,
      "step": 70457
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7955723404884338,
      "learning_rate": 0.00047188160440467226,
      "loss": 2.2923,
      "step": 70458
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.0788705348968506,
      "learning_rate": 0.0004718782517711566,
      "loss": 2.411,
      "step": 70459
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.8163173198699951,
      "learning_rate": 0.00047187489910568563,
      "loss": 2.3306,
      "step": 70460
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.8242671489715576,
      "learning_rate": 0.0004718715464082598,
      "loss": 2.3758,
      "step": 70461
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.8178099393844604,
      "learning_rate": 0.00047186819367887986,
      "loss": 2.0808,
      "step": 70462
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.642943024635315,
      "learning_rate": 0.0004718648409175464,
      "loss": 2.3739,
      "step": 70463
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.0503592491149902,
      "learning_rate": 0.0004718614881242599,
      "loss": 2.3086,
      "step": 70464
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.8475102186203003,
      "learning_rate": 0.0004718581352990212,
      "loss": 2.3914,
      "step": 70465
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.9954097867012024,
      "learning_rate": 0.00047185478244183085,
      "loss": 1.9976,
      "step": 70466
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.9958770275115967,
      "learning_rate": 0.0004718514295526894,
      "loss": 2.1737,
      "step": 70467
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7917836904525757,
      "learning_rate": 0.00047184807663159747,
      "loss": 2.1324,
      "step": 70468
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.9222559332847595,
      "learning_rate": 0.00047184472367855585,
      "loss": 2.36,
      "step": 70469
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.059768795967102,
      "learning_rate": 0.000471841370693565,
      "loss": 2.0861,
      "step": 70470
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.8186217546463013,
      "learning_rate": 0.00047183801767662555,
      "loss": 2.0069,
      "step": 70471
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.685256838798523,
      "learning_rate": 0.0004718346646277383,
      "loss": 2.1029,
      "step": 70472
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.9175204038619995,
      "learning_rate": 0.00047183131154690365,
      "loss": 2.2384,
      "step": 70473
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.9358441829681396,
      "learning_rate": 0.00047182795843412234,
      "loss": 2.0922,
      "step": 70474
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.8054420948028564,
      "learning_rate": 0.00047182460528939497,
      "loss": 2.3818,
      "step": 70475
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7618263363838196,
      "learning_rate": 0.0004718212521127221,
      "loss": 2.2469,
      "step": 70476
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.8607648611068726,
      "learning_rate": 0.00047181789890410444,
      "loss": 2.1499,
      "step": 70477
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.8708047866821289,
      "learning_rate": 0.00047181454566354265,
      "loss": 2.2459,
      "step": 70478
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7412707209587097,
      "learning_rate": 0.0004718111923910373,
      "loss": 1.9532,
      "step": 70479
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7721543908119202,
      "learning_rate": 0.00047180783908658887,
      "loss": 2.1789,
      "step": 70480
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.9292397499084473,
      "learning_rate": 0.0004718044857501983,
      "loss": 2.2537,
      "step": 70481
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.0555962324142456,
      "learning_rate": 0.000471801132381866,
      "loss": 2.2394,
      "step": 70482
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7426425814628601,
      "learning_rate": 0.0004717977789815925,
      "loss": 2.3026,
      "step": 70483
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.8406957387924194,
      "learning_rate": 0.0004717944255493786,
      "loss": 2.1748,
      "step": 70484
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.8826961517333984,
      "learning_rate": 0.0004717910720852249,
      "loss": 2.1608,
      "step": 70485
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.8666746616363525,
      "learning_rate": 0.000471787718589132,
      "loss": 2.29,
      "step": 70486
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.9453949928283691,
      "learning_rate": 0.00047178436506110054,
      "loss": 2.2789,
      "step": 70487
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.8740825057029724,
      "learning_rate": 0.0004717810115011311,
      "loss": 2.038,
      "step": 70488
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.8439647555351257,
      "learning_rate": 0.0004717776579092243,
      "loss": 2.4222,
      "step": 70489
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.9260640144348145,
      "learning_rate": 0.0004717743042853808,
      "loss": 2.1728,
      "step": 70490
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.8522335886955261,
      "learning_rate": 0.0004717709506296012,
      "loss": 2.1829,
      "step": 70491
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7694554328918457,
      "learning_rate": 0.0004717675969418862,
      "loss": 2.0751,
      "step": 70492
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.0074726343154907,
      "learning_rate": 0.00047176424322223636,
      "loss": 2.1668,
      "step": 70493
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.8285794258117676,
      "learning_rate": 0.0004717608894706523,
      "loss": 2.2723,
      "step": 70494
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7438583970069885,
      "learning_rate": 0.0004717575356871346,
      "loss": 2.3613,
      "step": 70495
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.8016377687454224,
      "learning_rate": 0.0004717541818716839,
      "loss": 2.3993,
      "step": 70496
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.9015644788742065,
      "learning_rate": 0.00047175082802430093,
      "loss": 2.3948,
      "step": 70497
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.8523087501525879,
      "learning_rate": 0.0004717474741449862,
      "loss": 2.3598,
      "step": 70498
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7228159308433533,
      "learning_rate": 0.0004717441202337404,
      "loss": 2.2445,
      "step": 70499
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.9845401644706726,
      "learning_rate": 0.00047174076629056416,
      "loss": 2.0781,
      "step": 70500
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.8191615343093872,
      "learning_rate": 0.000471737412315458,
      "loss": 2.277,
      "step": 70501
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.6945133209228516,
      "learning_rate": 0.0004717340583084226,
      "loss": 2.0291,
      "step": 70502
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.8283433318138123,
      "learning_rate": 0.00047173070426945874,
      "loss": 2.1723,
      "step": 70503
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.1166040897369385,
      "learning_rate": 0.0004717273501985668,
      "loss": 2.214,
      "step": 70504
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7829874753952026,
      "learning_rate": 0.00047172399609574745,
      "loss": 2.2178,
      "step": 70505
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.8679265975952148,
      "learning_rate": 0.0004717206419610014,
      "loss": 1.9474,
      "step": 70506
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.8393001556396484,
      "learning_rate": 0.00047171728779432924,
      "loss": 2.3432,
      "step": 70507
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7762045860290527,
      "learning_rate": 0.00047171393359573166,
      "loss": 2.2662,
      "step": 70508
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7481077313423157,
      "learning_rate": 0.00047171057936520915,
      "loss": 2.1402,
      "step": 70509
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7892791628837585,
      "learning_rate": 0.00047170722510276247,
      "loss": 2.1145,
      "step": 70510
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7728294730186462,
      "learning_rate": 0.00047170387080839205,
      "loss": 2.1389,
      "step": 70511
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.880226194858551,
      "learning_rate": 0.0004717005164820988,
      "loss": 2.3841,
      "step": 70512
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7809193730354309,
      "learning_rate": 0.0004716971621238831,
      "loss": 2.3833,
      "step": 70513
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7611083388328552,
      "learning_rate": 0.0004716938077337457,
      "loss": 2.2601,
      "step": 70514
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.8521739840507507,
      "learning_rate": 0.00047169045331168716,
      "loss": 2.283,
      "step": 70515
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7767464518547058,
      "learning_rate": 0.00047168709885770803,
      "loss": 2.3163,
      "step": 70516
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7604958415031433,
      "learning_rate": 0.0004716837443718091,
      "loss": 2.2701,
      "step": 70517
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7354855537414551,
      "learning_rate": 0.0004716803898539909,
      "loss": 2.2985,
      "step": 70518
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7577228546142578,
      "learning_rate": 0.00047167703530425414,
      "loss": 2.1732,
      "step": 70519
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7972590327262878,
      "learning_rate": 0.00047167368072259926,
      "loss": 2.3199,
      "step": 70520
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.8463490605354309,
      "learning_rate": 0.00047167032610902714,
      "loss": 2.2511,
      "step": 70521
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.8743662238121033,
      "learning_rate": 0.00047166697146353823,
      "loss": 2.2796,
      "step": 70522
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.707319438457489,
      "learning_rate": 0.0004716636167861331,
      "loss": 2.0818,
      "step": 70523
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7192674279212952,
      "learning_rate": 0.0004716602620768126,
      "loss": 2.2779,
      "step": 70524
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7782270312309265,
      "learning_rate": 0.00047165690733557715,
      "loss": 2.4288,
      "step": 70525
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.8020387291908264,
      "learning_rate": 0.0004716535525624274,
      "loss": 2.1061,
      "step": 70526
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.8806670308113098,
      "learning_rate": 0.00047165019775736414,
      "loss": 2.2533,
      "step": 70527
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.9117865562438965,
      "learning_rate": 0.0004716468429203877,
      "loss": 1.9614,
      "step": 70528
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7532132267951965,
      "learning_rate": 0.00047164348805149895,
      "loss": 2.1843,
      "step": 70529
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7668196558952332,
      "learning_rate": 0.0004716401331506984,
      "loss": 2.1353,
      "step": 70530
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.8078686594963074,
      "learning_rate": 0.0004716367782179868,
      "loss": 2.2978,
      "step": 70531
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7280861139297485,
      "learning_rate": 0.00047163342325336454,
      "loss": 2.1527,
      "step": 70532
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7819023132324219,
      "learning_rate": 0.0004716300682568325,
      "loss": 2.3012,
      "step": 70533
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.8078798651695251,
      "learning_rate": 0.00047162671322839117,
      "loss": 2.309,
      "step": 70534
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.8028796911239624,
      "learning_rate": 0.00047162335816804124,
      "loss": 2.0154,
      "step": 70535
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7029082775115967,
      "learning_rate": 0.0004716200030757832,
      "loss": 2.1953,
      "step": 70536
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.8800215125083923,
      "learning_rate": 0.00047161664795161777,
      "loss": 2.1496,
      "step": 70537
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7543812394142151,
      "learning_rate": 0.0004716132927955456,
      "loss": 2.1593,
      "step": 70538
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7979384064674377,
      "learning_rate": 0.0004716099376075674,
      "loss": 2.2258,
      "step": 70539
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.728166937828064,
      "learning_rate": 0.00047160658238768345,
      "loss": 2.2874,
      "step": 70540
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7048458456993103,
      "learning_rate": 0.00047160322713589464,
      "loss": 2.4548,
      "step": 70541
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.8793063163757324,
      "learning_rate": 0.00047159987185220166,
      "loss": 2.3373,
      "step": 70542
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7717456817626953,
      "learning_rate": 0.00047159651653660495,
      "loss": 2.4849,
      "step": 70543
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.9301654696464539,
      "learning_rate": 0.0004715931611891052,
      "loss": 2.3109,
      "step": 70544
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7877170443534851,
      "learning_rate": 0.00047158980580970316,
      "loss": 2.3193,
      "step": 70545
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.787922203540802,
      "learning_rate": 0.0004715864503983992,
      "loss": 2.164,
      "step": 70546
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.8089845776557922,
      "learning_rate": 0.0004715830949551941,
      "loss": 2.324,
      "step": 70547
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7955160737037659,
      "learning_rate": 0.0004715797394800885,
      "loss": 2.2085,
      "step": 70548
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7688660025596619,
      "learning_rate": 0.000471576383973083,
      "loss": 2.1238,
      "step": 70549
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7510293126106262,
      "learning_rate": 0.00047157302843417817,
      "loss": 2.1693,
      "step": 70550
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.747277557849884,
      "learning_rate": 0.00047156967286337474,
      "loss": 2.3031,
      "step": 70551
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7714391946792603,
      "learning_rate": 0.0004715663172606732,
      "loss": 1.9983,
      "step": 70552
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7467005252838135,
      "learning_rate": 0.00047156296162607426,
      "loss": 2.2258,
      "step": 70553
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7640299797058105,
      "learning_rate": 0.00047155960595957855,
      "loss": 2.2158,
      "step": 70554
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.776995837688446,
      "learning_rate": 0.00047155625026118665,
      "loss": 2.2209,
      "step": 70555
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7763193845748901,
      "learning_rate": 0.00047155289453089914,
      "loss": 2.1671,
      "step": 70556
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7013998627662659,
      "learning_rate": 0.0004715495387687169,
      "loss": 1.9674,
      "step": 70557
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7961117029190063,
      "learning_rate": 0.00047154618297464015,
      "loss": 2.1995,
      "step": 70558
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7525473237037659,
      "learning_rate": 0.0004715428271486698,
      "loss": 2.2833,
      "step": 70559
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.765078604221344,
      "learning_rate": 0.0004715394712908065,
      "loss": 1.9235,
      "step": 70560
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.722758948802948,
      "learning_rate": 0.0004715361154010507,
      "loss": 2.4017,
      "step": 70561
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7644104361534119,
      "learning_rate": 0.000471532759479403,
      "loss": 2.1581,
      "step": 70562
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.76163649559021,
      "learning_rate": 0.00047152940352586424,
      "loss": 2.4455,
      "step": 70563
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7693536877632141,
      "learning_rate": 0.00047152604754043497,
      "loss": 2.134,
      "step": 70564
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7845203876495361,
      "learning_rate": 0.0004715226915231156,
      "loss": 2.3896,
      "step": 70565
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7209916114807129,
      "learning_rate": 0.00047151933547390706,
      "loss": 2.1456,
      "step": 70566
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.9376252889633179,
      "learning_rate": 0.0004715159793928098,
      "loss": 2.1631,
      "step": 70567
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.6948378086090088,
      "learning_rate": 0.0004715126232798245,
      "loss": 2.1681,
      "step": 70568
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7598335146903992,
      "learning_rate": 0.0004715092671349517,
      "loss": 2.0788,
      "step": 70569
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7766785025596619,
      "learning_rate": 0.0004715059109581922,
      "loss": 2.29,
      "step": 70570
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.9072774648666382,
      "learning_rate": 0.00047150255474954643,
      "loss": 2.2788,
      "step": 70571
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7273612022399902,
      "learning_rate": 0.00047149919850901506,
      "loss": 2.3023,
      "step": 70572
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7575882077217102,
      "learning_rate": 0.0004714958422365988,
      "loss": 2.1261,
      "step": 70573
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7769255042076111,
      "learning_rate": 0.00047149248593229826,
      "loss": 2.1805,
      "step": 70574
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.6900779008865356,
      "learning_rate": 0.000471489129596114,
      "loss": 2.1213,
      "step": 70575
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7761288285255432,
      "learning_rate": 0.0004714857732280467,
      "loss": 2.0805,
      "step": 70576
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.8990520238876343,
      "learning_rate": 0.00047148241682809695,
      "loss": 2.1277,
      "step": 70577
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7494268417358398,
      "learning_rate": 0.0004714790603962653,
      "loss": 2.1236,
      "step": 70578
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.8186816573143005,
      "learning_rate": 0.0004714757039325526,
      "loss": 2.2272,
      "step": 70579
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.8998276591300964,
      "learning_rate": 0.0004714723474369592,
      "loss": 2.2282,
      "step": 70580
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7984055280685425,
      "learning_rate": 0.000471468990909486,
      "loss": 2.3673,
      "step": 70581
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7886048555374146,
      "learning_rate": 0.0004714656343501333,
      "loss": 1.9903,
      "step": 70582
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7557599544525146,
      "learning_rate": 0.000471462277758902,
      "loss": 1.933,
      "step": 70583
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7021009922027588,
      "learning_rate": 0.0004714589211357926,
      "loss": 2.2786,
      "step": 70584
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.8526948690414429,
      "learning_rate": 0.0004714555644808058,
      "loss": 2.1311,
      "step": 70585
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.8619911670684814,
      "learning_rate": 0.0004714522077939421,
      "loss": 2.1185,
      "step": 70586
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.9510964155197144,
      "learning_rate": 0.00047144885107520225,
      "loss": 2.2061,
      "step": 70587
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7522621750831604,
      "learning_rate": 0.0004714454943245868,
      "loss": 2.0649,
      "step": 70588
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.739719808101654,
      "learning_rate": 0.0004714421375420964,
      "loss": 2.2474,
      "step": 70589
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.8727431893348694,
      "learning_rate": 0.00047143878072773167,
      "loss": 2.2128,
      "step": 70590
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.8315041065216064,
      "learning_rate": 0.00047143542388149325,
      "loss": 2.1298,
      "step": 70591
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.863658607006073,
      "learning_rate": 0.0004714320670033817,
      "loss": 2.1013,
      "step": 70592
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7242951989173889,
      "learning_rate": 0.00047142871009339774,
      "loss": 2.067,
      "step": 70593
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7879037857055664,
      "learning_rate": 0.00047142535315154206,
      "loss": 2.1362,
      "step": 70594
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.9530113935470581,
      "learning_rate": 0.00047142199617781496,
      "loss": 2.3083,
      "step": 70595
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.6871728897094727,
      "learning_rate": 0.00047141863917221736,
      "loss": 2.3026,
      "step": 70596
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7005473971366882,
      "learning_rate": 0.0004714152821347498,
      "loss": 2.2063,
      "step": 70597
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.9295975565910339,
      "learning_rate": 0.00047141192506541287,
      "loss": 2.3629,
      "step": 70598
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.8335031867027283,
      "learning_rate": 0.00047140856796420733,
      "loss": 2.4235,
      "step": 70599
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7554889917373657,
      "learning_rate": 0.00047140521083113363,
      "loss": 2.2138,
      "step": 70600
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.8252784013748169,
      "learning_rate": 0.0004714018536661924,
      "loss": 2.229,
      "step": 70601
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.8495884537696838,
      "learning_rate": 0.00047139849646938436,
      "loss": 2.2887,
      "step": 70602
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.8106619715690613,
      "learning_rate": 0.00047139513924071027,
      "loss": 2.2263,
      "step": 70603
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7513989806175232,
      "learning_rate": 0.00047139178198017044,
      "loss": 2.444,
      "step": 70604
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7460154294967651,
      "learning_rate": 0.0004713884246877657,
      "loss": 2.4136,
      "step": 70605
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7855300307273865,
      "learning_rate": 0.00047138506736349656,
      "loss": 2.2891,
      "step": 70606
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.9084498286247253,
      "learning_rate": 0.0004713817100073637,
      "loss": 2.3713,
      "step": 70607
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7914063334465027,
      "learning_rate": 0.0004713783526193678,
      "loss": 2.1659,
      "step": 70608
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.8903523087501526,
      "learning_rate": 0.00047137499519950943,
      "loss": 2.3448,
      "step": 70609
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.8447909951210022,
      "learning_rate": 0.00047137163774778915,
      "loss": 2.2631,
      "step": 70610
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7490419149398804,
      "learning_rate": 0.0004713682802642076,
      "loss": 2.2613,
      "step": 70611
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.8660712242126465,
      "learning_rate": 0.0004713649227487656,
      "loss": 2.4186,
      "step": 70612
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7768875360488892,
      "learning_rate": 0.0004713615652014635,
      "loss": 2.3215,
      "step": 70613
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.877972424030304,
      "learning_rate": 0.0004713582076223021,
      "loss": 2.2237,
      "step": 70614
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.9303569793701172,
      "learning_rate": 0.000471354850011282,
      "loss": 2.1705,
      "step": 70615
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.8042401075363159,
      "learning_rate": 0.00047135149236840387,
      "loss": 2.45,
      "step": 70616
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.9079487919807434,
      "learning_rate": 0.0004713481346936681,
      "loss": 2.2723,
      "step": 70617
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.9396382570266724,
      "learning_rate": 0.0004713447769870755,
      "loss": 2.384,
      "step": 70618
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7554427981376648,
      "learning_rate": 0.00047134141924862677,
      "loss": 2.0825,
      "step": 70619
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7901491522789001,
      "learning_rate": 0.00047133806147832245,
      "loss": 2.1346,
      "step": 70620
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.770519495010376,
      "learning_rate": 0.00047133470367616303,
      "loss": 2.1994,
      "step": 70621
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7253440022468567,
      "learning_rate": 0.00047133134584214926,
      "loss": 2.3761,
      "step": 70622
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.8289096355438232,
      "learning_rate": 0.0004713279879762819,
      "loss": 2.4535,
      "step": 70623
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.8430460095405579,
      "learning_rate": 0.00047132463007856134,
      "loss": 2.2519,
      "step": 70624
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7480931282043457,
      "learning_rate": 0.0004713212721489883,
      "loss": 2.3192,
      "step": 70625
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7433614134788513,
      "learning_rate": 0.0004713179141875634,
      "loss": 2.2472,
      "step": 70626
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.8117163181304932,
      "learning_rate": 0.00047131455619428726,
      "loss": 2.1775,
      "step": 70627
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.8460196852684021,
      "learning_rate": 0.00047131119816916047,
      "loss": 2.2744,
      "step": 70628
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.8102657794952393,
      "learning_rate": 0.0004713078401121837,
      "loss": 2.2884,
      "step": 70629
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7441737651824951,
      "learning_rate": 0.00047130448202335767,
      "loss": 2.2653,
      "step": 70630
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7736673951148987,
      "learning_rate": 0.00047130112390268285,
      "loss": 2.0138,
      "step": 70631
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7005910873413086,
      "learning_rate": 0.0004712977657501599,
      "loss": 2.0746,
      "step": 70632
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.847591757774353,
      "learning_rate": 0.00047129440756578947,
      "loss": 2.3628,
      "step": 70633
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.790733277797699,
      "learning_rate": 0.00047129104934957226,
      "loss": 2.0811,
      "step": 70634
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7863287925720215,
      "learning_rate": 0.0004712876911015086,
      "loss": 2.2422,
      "step": 70635
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7476980090141296,
      "learning_rate": 0.0004712843328215995,
      "loss": 2.2115,
      "step": 70636
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7286154627799988,
      "learning_rate": 0.00047128097450984535,
      "loss": 2.4544,
      "step": 70637
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.8177059292793274,
      "learning_rate": 0.0004712776161662469,
      "loss": 2.2949,
      "step": 70638
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.8264389634132385,
      "learning_rate": 0.0004712742577908046,
      "loss": 2.0779,
      "step": 70639
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7925137281417847,
      "learning_rate": 0.00047127089938351935,
      "loss": 2.3711,
      "step": 70640
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.8340064883232117,
      "learning_rate": 0.0004712675409443915,
      "loss": 2.1783,
      "step": 70641
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.8288554549217224,
      "learning_rate": 0.0004712641824734217,
      "loss": 2.3961,
      "step": 70642
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.8400275111198425,
      "learning_rate": 0.00047126082397061075,
      "loss": 2.1431,
      "step": 70643
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7517394423484802,
      "learning_rate": 0.0004712574654359591,
      "loss": 2.1468,
      "step": 70644
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.9189222455024719,
      "learning_rate": 0.00047125410686946754,
      "loss": 2.1132,
      "step": 70645
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.73225998878479,
      "learning_rate": 0.0004712507482711366,
      "loss": 2.2518,
      "step": 70646
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7416694164276123,
      "learning_rate": 0.0004712473896409669,
      "loss": 2.21,
      "step": 70647
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.8131746053695679,
      "learning_rate": 0.00047124403097895904,
      "loss": 2.3651,
      "step": 70648
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.847855269908905,
      "learning_rate": 0.0004712406722851138,
      "loss": 2.0505,
      "step": 70649
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.8811198472976685,
      "learning_rate": 0.0004712373135594316,
      "loss": 2.1818,
      "step": 70650
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7491446733474731,
      "learning_rate": 0.0004712339548019131,
      "loss": 2.1593,
      "step": 70651
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.8042553067207336,
      "learning_rate": 0.00047123059601255913,
      "loss": 1.8586,
      "step": 70652
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.777872622013092,
      "learning_rate": 0.00047122723719137,
      "loss": 2.2832,
      "step": 70653
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7572222352027893,
      "learning_rate": 0.00047122387833834655,
      "loss": 2.0669,
      "step": 70654
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7519557476043701,
      "learning_rate": 0.00047122051945348945,
      "loss": 2.2461,
      "step": 70655
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.8352375626564026,
      "learning_rate": 0.00047121716053679904,
      "loss": 2.2167,
      "step": 70656
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.8516282439231873,
      "learning_rate": 0.0004712138015882763,
      "loss": 2.241,
      "step": 70657
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.8075504302978516,
      "learning_rate": 0.0004712104426079216,
      "loss": 2.249,
      "step": 70658
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7238727807998657,
      "learning_rate": 0.0004712070835957356,
      "loss": 2.3139,
      "step": 70659
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7141897678375244,
      "learning_rate": 0.000471203724551719,
      "loss": 2.4623,
      "step": 70660
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.8852913975715637,
      "learning_rate": 0.0004712003654758724,
      "loss": 2.2141,
      "step": 70661
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7326437830924988,
      "learning_rate": 0.00047119700636819647,
      "loss": 2.1187,
      "step": 70662
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7168964147567749,
      "learning_rate": 0.00047119364722869166,
      "loss": 2.338,
      "step": 70663
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.8475450277328491,
      "learning_rate": 0.00047119028805735886,
      "loss": 2.1586,
      "step": 70664
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.796597957611084,
      "learning_rate": 0.00047118692885419854,
      "loss": 2.2967,
      "step": 70665
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.8414088487625122,
      "learning_rate": 0.0004711835696192113,
      "loss": 2.2547,
      "step": 70666
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7983983159065247,
      "learning_rate": 0.0004711802103523978,
      "loss": 2.2051,
      "step": 70667
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.6996751427650452,
      "learning_rate": 0.0004711768510537587,
      "loss": 2.1971,
      "step": 70668
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.8627012372016907,
      "learning_rate": 0.00047117349172329454,
      "loss": 1.9597,
      "step": 70669
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7435051202774048,
      "learning_rate": 0.000471170132361006,
      "loss": 2.3131,
      "step": 70670
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.8673730492591858,
      "learning_rate": 0.0004711667729668937,
      "loss": 2.1509,
      "step": 70671
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.8232434391975403,
      "learning_rate": 0.00047116341354095834,
      "loss": 2.2157,
      "step": 70672
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7449013590812683,
      "learning_rate": 0.00047116005408320036,
      "loss": 2.257,
      "step": 70673
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.8419862985610962,
      "learning_rate": 0.00047115669459362055,
      "loss": 2.1522,
      "step": 70674
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.0976839065551758,
      "learning_rate": 0.00047115333507221946,
      "loss": 2.1194,
      "step": 70675
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7543510794639587,
      "learning_rate": 0.0004711499755189978,
      "loss": 2.1238,
      "step": 70676
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.9982298016548157,
      "learning_rate": 0.00047114661593395605,
      "loss": 2.1971,
      "step": 70677
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.8969643712043762,
      "learning_rate": 0.0004711432563170949,
      "loss": 2.2713,
      "step": 70678
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.8113272786140442,
      "learning_rate": 0.00047113989666841517,
      "loss": 2.3399,
      "step": 70679
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7470662593841553,
      "learning_rate": 0.0004711365369879171,
      "loss": 2.0322,
      "step": 70680
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.9118845462799072,
      "learning_rate": 0.00047113317727560163,
      "loss": 2.3707,
      "step": 70681
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7718797922134399,
      "learning_rate": 0.00047112981753146926,
      "loss": 2.3051,
      "step": 70682
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.8994059562683105,
      "learning_rate": 0.00047112645775552054,
      "loss": 2.431,
      "step": 70683
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.8986789584159851,
      "learning_rate": 0.00047112309794775624,
      "loss": 2.3603,
      "step": 70684
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7774249315261841,
      "learning_rate": 0.000471119738108177,
      "loss": 2.357,
      "step": 70685
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.8644322156906128,
      "learning_rate": 0.00047111637823678325,
      "loss": 2.3677,
      "step": 70686
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.695945680141449,
      "learning_rate": 0.0004711130183335757,
      "loss": 2.2582,
      "step": 70687
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7404331564903259,
      "learning_rate": 0.0004711096583985552,
      "loss": 2.3156,
      "step": 70688
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.9540036916732788,
      "learning_rate": 0.0004711062984317221,
      "loss": 2.4951,
      "step": 70689
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.8039594888687134,
      "learning_rate": 0.000471102938433077,
      "loss": 2.1652,
      "step": 70690
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.8500940203666687,
      "learning_rate": 0.00047109957840262077,
      "loss": 2.2537,
      "step": 70691
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7244113683700562,
      "learning_rate": 0.00047109621834035383,
      "loss": 2.1367,
      "step": 70692
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.808533251285553,
      "learning_rate": 0.00047109285824627683,
      "loss": 2.4217,
      "step": 70693
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.9452435374259949,
      "learning_rate": 0.0004710894981203906,
      "loss": 2.0281,
      "step": 70694
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7261620759963989,
      "learning_rate": 0.0004710861379626955,
      "loss": 2.2374,
      "step": 70695
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.9034472703933716,
      "learning_rate": 0.00047108277777319215,
      "loss": 2.1878,
      "step": 70696
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.8145327568054199,
      "learning_rate": 0.00047107941755188137,
      "loss": 2.263,
      "step": 70697
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7934721112251282,
      "learning_rate": 0.00047107605729876373,
      "loss": 2.1714,
      "step": 70698
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7751345038414001,
      "learning_rate": 0.0004710726970138398,
      "loss": 2.0887,
      "step": 70699
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.8772770762443542,
      "learning_rate": 0.0004710693366971103,
      "loss": 2.1601,
      "step": 70700
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7656235694885254,
      "learning_rate": 0.0004710659763485757,
      "loss": 1.9724,
      "step": 70701
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.8032463788986206,
      "learning_rate": 0.0004710626159682367,
      "loss": 2.363,
      "step": 70702
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7589042782783508,
      "learning_rate": 0.00047105925555609394,
      "loss": 2.4058,
      "step": 70703
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.8840605020523071,
      "learning_rate": 0.00047105589511214815,
      "loss": 2.1826,
      "step": 70704
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7900090217590332,
      "learning_rate": 0.0004710525346363997,
      "loss": 2.1846,
      "step": 70705
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.2012782096862793,
      "learning_rate": 0.00047104917412884933,
      "loss": 2.1738,
      "step": 70706
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.8532795310020447,
      "learning_rate": 0.00047104581358949776,
      "loss": 2.0998,
      "step": 70707
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.8239800930023193,
      "learning_rate": 0.00047104245301834557,
      "loss": 2.1351,
      "step": 70708
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.8605217933654785,
      "learning_rate": 0.0004710390924153933,
      "loss": 2.2644,
      "step": 70709
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.9092983603477478,
      "learning_rate": 0.0004710357317806417,
      "loss": 2.185,
      "step": 70710
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.678844153881073,
      "learning_rate": 0.0004710323711140913,
      "loss": 2.4682,
      "step": 70711
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.8356027007102966,
      "learning_rate": 0.0004710290104157427,
      "loss": 2.3541,
      "step": 70712
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7658651471138,
      "learning_rate": 0.0004710256496855966,
      "loss": 2.1799,
      "step": 70713
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7659592032432556,
      "learning_rate": 0.0004710222889236536,
      "loss": 2.3963,
      "step": 70714
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7488041520118713,
      "learning_rate": 0.0004710189281299144,
      "loss": 2.1409,
      "step": 70715
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7874970436096191,
      "learning_rate": 0.0004710155673043795,
      "loss": 2.2186,
      "step": 70716
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.771257758140564,
      "learning_rate": 0.0004710122064470496,
      "loss": 2.0956,
      "step": 70717
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7611657381057739,
      "learning_rate": 0.00047100884555792525,
      "loss": 2.054,
      "step": 70718
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7404943108558655,
      "learning_rate": 0.0004710054846370072,
      "loss": 2.2754,
      "step": 70719
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7613041400909424,
      "learning_rate": 0.0004710021236842959,
      "loss": 2.3102,
      "step": 70720
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7543968558311462,
      "learning_rate": 0.00047099876269979214,
      "loss": 2.0755,
      "step": 70721
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.8305810689926147,
      "learning_rate": 0.00047099540168349656,
      "loss": 2.2551,
      "step": 70722
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7300847172737122,
      "learning_rate": 0.0004709920406354096,
      "loss": 2.1883,
      "step": 70723
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.8147988319396973,
      "learning_rate": 0.000470988679555532,
      "loss": 2.114,
      "step": 70724
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7463380694389343,
      "learning_rate": 0.0004709853184438644,
      "loss": 2.2881,
      "step": 70725
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7918149828910828,
      "learning_rate": 0.0004709819573004074,
      "loss": 2.1816,
      "step": 70726
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.9424297213554382,
      "learning_rate": 0.00047097859612516165,
      "loss": 2.1216,
      "step": 70727
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7748528718948364,
      "learning_rate": 0.00047097523491812775,
      "loss": 2.2805,
      "step": 70728
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.9284659028053284,
      "learning_rate": 0.00047097187367930635,
      "loss": 2.2158,
      "step": 70729
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.9250520467758179,
      "learning_rate": 0.000470968512408698,
      "loss": 2.05,
      "step": 70730
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.8209595680236816,
      "learning_rate": 0.0004709651511063034,
      "loss": 2.2302,
      "step": 70731
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.0258361101150513,
      "learning_rate": 0.00047096178977212307,
      "loss": 2.199,
      "step": 70732
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.8369582295417786,
      "learning_rate": 0.00047095842840615777,
      "loss": 2.3192,
      "step": 70733
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7227302193641663,
      "learning_rate": 0.00047095506700840816,
      "loss": 2.0095,
      "step": 70734
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.8797166347503662,
      "learning_rate": 0.0004709517055788747,
      "loss": 2.3407,
      "step": 70735
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.8623251914978027,
      "learning_rate": 0.000470948344117558,
      "loss": 2.3496,
      "step": 70736
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7903640270233154,
      "learning_rate": 0.00047094498262445896,
      "loss": 2.4291,
      "step": 70737
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.8005994558334351,
      "learning_rate": 0.0004709416210995779,
      "loss": 2.1473,
      "step": 70738
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.8487106561660767,
      "learning_rate": 0.0004709382595429156,
      "loss": 2.2132,
      "step": 70739
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7840331196784973,
      "learning_rate": 0.00047093489795447266,
      "loss": 2.3778,
      "step": 70740
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.706820011138916,
      "learning_rate": 0.00047093153633424964,
      "loss": 2.0427,
      "step": 70741
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7148332595825195,
      "learning_rate": 0.00047092817468224725,
      "loss": 2.0847,
      "step": 70742
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.8074659705162048,
      "learning_rate": 0.0004709248129984661,
      "loss": 2.1603,
      "step": 70743
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.9577849507331848,
      "learning_rate": 0.00047092145128290684,
      "loss": 2.0536,
      "step": 70744
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.8019911646842957,
      "learning_rate": 0.00047091808953556995,
      "loss": 2.3494,
      "step": 70745
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.8111911416053772,
      "learning_rate": 0.0004709147277564563,
      "loss": 2.1312,
      "step": 70746
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7680861353874207,
      "learning_rate": 0.00047091136594556624,
      "loss": 2.178,
      "step": 70747
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.78666752576828,
      "learning_rate": 0.0004709080041029006,
      "loss": 2.1983,
      "step": 70748
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.8323673605918884,
      "learning_rate": 0.00047090464222845994,
      "loss": 2.3566,
      "step": 70749
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.761220395565033,
      "learning_rate": 0.0004709012803222449,
      "loss": 2.3724,
      "step": 70750
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.8330830335617065,
      "learning_rate": 0.00047089791838425603,
      "loss": 2.3035,
      "step": 70751
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7965453863143921,
      "learning_rate": 0.00047089455641449404,
      "loss": 2.2445,
      "step": 70752
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7764441967010498,
      "learning_rate": 0.00047089119441295945,
      "loss": 2.2555,
      "step": 70753
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7443946003913879,
      "learning_rate": 0.000470887832379653,
      "loss": 2.3158,
      "step": 70754
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.8847893476486206,
      "learning_rate": 0.00047088447031457534,
      "loss": 2.2341,
      "step": 70755
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7776309847831726,
      "learning_rate": 0.000470881108217727,
      "loss": 2.0311,
      "step": 70756
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7250361442565918,
      "learning_rate": 0.00047087774608910865,
      "loss": 2.1379,
      "step": 70757
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7698743939399719,
      "learning_rate": 0.000470874383928721,
      "loss": 1.8083,
      "step": 70758
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7910937070846558,
      "learning_rate": 0.0004708710217365644,
      "loss": 2.0914,
      "step": 70759
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.6984460949897766,
      "learning_rate": 0.0004708676595126397,
      "loss": 2.0385,
      "step": 70760
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.838865339756012,
      "learning_rate": 0.00047086429725694756,
      "loss": 2.3321,
      "step": 70761
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7659803032875061,
      "learning_rate": 0.0004708609349694884,
      "loss": 2.2378,
      "step": 70762
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7170183658599854,
      "learning_rate": 0.00047085757265026296,
      "loss": 2.0214,
      "step": 70763
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7446286678314209,
      "learning_rate": 0.000470854210299272,
      "loss": 2.2948,
      "step": 70764
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.8727638721466064,
      "learning_rate": 0.00047085084791651595,
      "loss": 2.2429,
      "step": 70765
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.689351499080658,
      "learning_rate": 0.00047084748550199547,
      "loss": 2.3168,
      "step": 70766
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7854653000831604,
      "learning_rate": 0.00047084412305571123,
      "loss": 2.0903,
      "step": 70767
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7888056635856628,
      "learning_rate": 0.00047084076057766393,
      "loss": 2.2323,
      "step": 70768
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7688656449317932,
      "learning_rate": 0.000470837398067854,
      "loss": 2.0982,
      "step": 70769
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7632122039794922,
      "learning_rate": 0.0004708340355262822,
      "loss": 2.144,
      "step": 70770
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.8034504652023315,
      "learning_rate": 0.00047083067295294916,
      "loss": 2.3981,
      "step": 70771
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7280858159065247,
      "learning_rate": 0.0004708273103478554,
      "loss": 2.305,
      "step": 70772
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7300976514816284,
      "learning_rate": 0.0004708239477110017,
      "loss": 2.3291,
      "step": 70773
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7605444192886353,
      "learning_rate": 0.0004708205850423886,
      "loss": 2.2571,
      "step": 70774
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7235773801803589,
      "learning_rate": 0.0004708172223420167,
      "loss": 2.1341,
      "step": 70775
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.9337411522865295,
      "learning_rate": 0.00047081385960988663,
      "loss": 2.1659,
      "step": 70776
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7347953915596008,
      "learning_rate": 0.00047081049684599913,
      "loss": 2.2798,
      "step": 70777
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.8021900057792664,
      "learning_rate": 0.00047080713405035465,
      "loss": 2.5122,
      "step": 70778
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7479065656661987,
      "learning_rate": 0.0004708037712229539,
      "loss": 2.0959,
      "step": 70779
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7959755063056946,
      "learning_rate": 0.00047080040836379754,
      "loss": 2.189,
      "step": 70780
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7911352515220642,
      "learning_rate": 0.00047079704547288616,
      "loss": 2.1939,
      "step": 70781
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7734533548355103,
      "learning_rate": 0.00047079368255022034,
      "loss": 2.3268,
      "step": 70782
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7470468282699585,
      "learning_rate": 0.00047079031959580084,
      "loss": 2.2564,
      "step": 70783
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.8024135828018188,
      "learning_rate": 0.0004707869566096281,
      "loss": 2.2176,
      "step": 70784
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7626923322677612,
      "learning_rate": 0.00047078359359170287,
      "loss": 2.0483,
      "step": 70785
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7986029982566833,
      "learning_rate": 0.0004707802305420258,
      "loss": 2.0877,
      "step": 70786
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7859135270118713,
      "learning_rate": 0.0004707768674605974,
      "loss": 2.3415,
      "step": 70787
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.8912109136581421,
      "learning_rate": 0.0004707735043474184,
      "loss": 2.2822,
      "step": 70788
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.8304762840270996,
      "learning_rate": 0.0004707701412024893,
      "loss": 2.3898,
      "step": 70789
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7555253505706787,
      "learning_rate": 0.0004707667780258109,
      "loss": 1.9155,
      "step": 70790
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7740436792373657,
      "learning_rate": 0.00047076341481738364,
      "loss": 2.1624,
      "step": 70791
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.8316128253936768,
      "learning_rate": 0.0004707600515772083,
      "loss": 2.4589,
      "step": 70792
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7517975568771362,
      "learning_rate": 0.00047075668830528545,
      "loss": 2.1491,
      "step": 70793
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7324724197387695,
      "learning_rate": 0.00047075332500161565,
      "loss": 2.2002,
      "step": 70794
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7833729386329651,
      "learning_rate": 0.00047074996166619965,
      "loss": 2.1209,
      "step": 70795
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.8804010152816772,
      "learning_rate": 0.00047074659829903805,
      "loss": 2.3622,
      "step": 70796
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7441614866256714,
      "learning_rate": 0.00047074323490013133,
      "loss": 2.2975,
      "step": 70797
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.8097414374351501,
      "learning_rate": 0.00047073987146948026,
      "loss": 2.2334,
      "step": 70798
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.00812828540802,
      "learning_rate": 0.00047073650800708543,
      "loss": 1.9042,
      "step": 70799
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7600685954093933,
      "learning_rate": 0.0004707331445129475,
      "loss": 2.2432,
      "step": 70800
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7595780491828918,
      "learning_rate": 0.00047072978098706704,
      "loss": 2.4103,
      "step": 70801
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7991542816162109,
      "learning_rate": 0.0004707264174294446,
      "loss": 2.2066,
      "step": 70802
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.8486202359199524,
      "learning_rate": 0.000470723053840081,
      "loss": 2.3036,
      "step": 70803
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.9432953000068665,
      "learning_rate": 0.00047071969021897673,
      "loss": 2.0364,
      "step": 70804
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.0186200141906738,
      "learning_rate": 0.0004707163265661323,
      "loss": 2.2713,
      "step": 70805
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.9495283961296082,
      "learning_rate": 0.00047071296288154867,
      "loss": 2.1883,
      "step": 70806
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.8121451735496521,
      "learning_rate": 0.00047070959916522626,
      "loss": 2.3321,
      "step": 70807
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.1743395328521729,
      "learning_rate": 0.0004707062354171657,
      "loss": 2.2349,
      "step": 70808
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.811336874961853,
      "learning_rate": 0.00047070287163736757,
      "loss": 2.1061,
      "step": 70809
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.8095131516456604,
      "learning_rate": 0.00047069950782583264,
      "loss": 2.2659,
      "step": 70810
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7490153908729553,
      "learning_rate": 0.0004706961439825614,
      "loss": 2.1529,
      "step": 70811
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.6788768768310547,
      "learning_rate": 0.00047069278010755444,
      "loss": 2.3093,
      "step": 70812
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.9080654978752136,
      "learning_rate": 0.00047068941620081263,
      "loss": 1.8272,
      "step": 70813
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.9452289342880249,
      "learning_rate": 0.00047068605226233633,
      "loss": 2.2637,
      "step": 70814
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7731713652610779,
      "learning_rate": 0.0004706826882921263,
      "loss": 2.2983,
      "step": 70815
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.0375456809997559,
      "learning_rate": 0.0004706793242901832,
      "loss": 2.2962,
      "step": 70816
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.985072672367096,
      "learning_rate": 0.0004706759602565075,
      "loss": 2.1903,
      "step": 70817
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7972400188446045,
      "learning_rate": 0.0004706725961910999,
      "loss": 2.2747,
      "step": 70818
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7464048266410828,
      "learning_rate": 0.00047066923209396113,
      "loss": 2.0751,
      "step": 70819
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.8460513353347778,
      "learning_rate": 0.00047066586796509167,
      "loss": 2.2252,
      "step": 70820
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.9149582386016846,
      "learning_rate": 0.0004706625038044922,
      "loss": 2.2346,
      "step": 70821
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.9717518091201782,
      "learning_rate": 0.0004706591396121634,
      "loss": 2.2911,
      "step": 70822
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7953052520751953,
      "learning_rate": 0.0004706557753881058,
      "loss": 2.4146,
      "step": 70823
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.9874263405799866,
      "learning_rate": 0.0004706524111323201,
      "loss": 2.0317,
      "step": 70824
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7905973792076111,
      "learning_rate": 0.00047064904684480696,
      "loss": 2.2406,
      "step": 70825
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7763450145721436,
      "learning_rate": 0.00047064568252556683,
      "loss": 2.1964,
      "step": 70826
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7728863954544067,
      "learning_rate": 0.00047064231817460045,
      "loss": 2.3211,
      "step": 70827
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.8789455890655518,
      "learning_rate": 0.00047063895379190847,
      "loss": 2.1428,
      "step": 70828
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.8582509160041809,
      "learning_rate": 0.00047063558937749154,
      "loss": 2.2582,
      "step": 70829
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7397245168685913,
      "learning_rate": 0.0004706322249313501,
      "loss": 2.1534,
      "step": 70830
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.6727695465087891,
      "learning_rate": 0.000470628860453485,
      "loss": 2.0231,
      "step": 70831
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.849471390247345,
      "learning_rate": 0.0004706254959438968,
      "loss": 2.1002,
      "step": 70832
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.9794468879699707,
      "learning_rate": 0.00047062213140258605,
      "loss": 2.4739,
      "step": 70833
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7660256624221802,
      "learning_rate": 0.0004706187668295534,
      "loss": 2.1636,
      "step": 70834
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7936076521873474,
      "learning_rate": 0.0004706154022247996,
      "loss": 2.2206,
      "step": 70835
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.8287773132324219,
      "learning_rate": 0.0004706120375883251,
      "loss": 2.3019,
      "step": 70836
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.8501930236816406,
      "learning_rate": 0.00047060867292013055,
      "loss": 2.3057,
      "step": 70837
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7843044400215149,
      "learning_rate": 0.0004706053082202168,
      "loss": 1.9876,
      "step": 70838
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7621592283248901,
      "learning_rate": 0.00047060194348858415,
      "loss": 2.2254,
      "step": 70839
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.858933687210083,
      "learning_rate": 0.0004705985787252334,
      "loss": 2.1837,
      "step": 70840
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7999572157859802,
      "learning_rate": 0.0004705952139301652,
      "loss": 2.1573,
      "step": 70841
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.8145816922187805,
      "learning_rate": 0.0004705918491033802,
      "loss": 2.0818,
      "step": 70842
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7504476308822632,
      "learning_rate": 0.00047058848424487876,
      "loss": 2.3125,
      "step": 70843
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7926145195960999,
      "learning_rate": 0.0004705851193546618,
      "loss": 2.4626,
      "step": 70844
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.6968531012535095,
      "learning_rate": 0.00047058175443272986,
      "loss": 2.3038,
      "step": 70845
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.9357869625091553,
      "learning_rate": 0.0004705783894790836,
      "loss": 2.2131,
      "step": 70846
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7723003029823303,
      "learning_rate": 0.0004705750244937235,
      "loss": 2.3971,
      "step": 70847
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7512744069099426,
      "learning_rate": 0.0004705716594766503,
      "loss": 2.1195,
      "step": 70848
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7450428009033203,
      "learning_rate": 0.0004705682944278646,
      "loss": 2.3652,
      "step": 70849
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.8168482184410095,
      "learning_rate": 0.0004705649293473671,
      "loss": 2.1902,
      "step": 70850
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7686374187469482,
      "learning_rate": 0.0004705615642351583,
      "loss": 2.2807,
      "step": 70851
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.6792708039283752,
      "learning_rate": 0.00047055819909123893,
      "loss": 2.3248,
      "step": 70852
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.6922873258590698,
      "learning_rate": 0.00047055483391560956,
      "loss": 2.0992,
      "step": 70853
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7132382988929749,
      "learning_rate": 0.0004705514687082708,
      "loss": 1.8806,
      "step": 70854
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7247219681739807,
      "learning_rate": 0.0004705481034692233,
      "loss": 2.2125,
      "step": 70855
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7945729494094849,
      "learning_rate": 0.00047054473819846783,
      "loss": 2.1767,
      "step": 70856
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7339940667152405,
      "learning_rate": 0.0004705413728960047,
      "loss": 2.2091,
      "step": 70857
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7288957238197327,
      "learning_rate": 0.00047053800756183473,
      "loss": 2.2805,
      "step": 70858
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.730320394039154,
      "learning_rate": 0.0004705346421959586,
      "loss": 2.2026,
      "step": 70859
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.8624706864356995,
      "learning_rate": 0.0004705312767983768,
      "loss": 2.2752,
      "step": 70860
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.9548410773277283,
      "learning_rate": 0.00047052791136909005,
      "loss": 2.1373,
      "step": 70861
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7860878705978394,
      "learning_rate": 0.000470524545908099,
      "loss": 2.1905,
      "step": 70862
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.752894937992096,
      "learning_rate": 0.00047052118041540403,
      "loss": 2.1016,
      "step": 70863
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7308224439620972,
      "learning_rate": 0.00047051781489100607,
      "loss": 2.1566,
      "step": 70864
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7399543523788452,
      "learning_rate": 0.00047051444933490576,
      "loss": 2.1882,
      "step": 70865
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7306422591209412,
      "learning_rate": 0.00047051108374710344,
      "loss": 2.0712,
      "step": 70866
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7946195602416992,
      "learning_rate": 0.00047050771812759987,
      "loss": 2.2261,
      "step": 70867
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.77993243932724,
      "learning_rate": 0.0004705043524763958,
      "loss": 2.1027,
      "step": 70868
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.73850417137146,
      "learning_rate": 0.0004705009867934917,
      "loss": 2.1912,
      "step": 70869
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7539323568344116,
      "learning_rate": 0.00047049762107888814,
      "loss": 2.1484,
      "step": 70870
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.8431571125984192,
      "learning_rate": 0.0004704942553325861,
      "loss": 2.4032,
      "step": 70871
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7341394424438477,
      "learning_rate": 0.00047049088955458576,
      "loss": 2.1884,
      "step": 70872
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7685306668281555,
      "learning_rate": 0.00047048752374488796,
      "loss": 2.3772,
      "step": 70873
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7338254451751709,
      "learning_rate": 0.00047048415790349336,
      "loss": 2.3919,
      "step": 70874
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7699369192123413,
      "learning_rate": 0.0004704807920304026,
      "loss": 2.1171,
      "step": 70875
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7099932432174683,
      "learning_rate": 0.0004704774261256161,
      "loss": 2.3524,
      "step": 70876
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7607786059379578,
      "learning_rate": 0.00047047406018913473,
      "loss": 2.2261,
      "step": 70877
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7798819541931152,
      "learning_rate": 0.00047047069422095894,
      "loss": 2.1887,
      "step": 70878
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.6965916752815247,
      "learning_rate": 0.00047046732822108954,
      "loss": 2.1014,
      "step": 70879
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7393035888671875,
      "learning_rate": 0.00047046396218952696,
      "loss": 2.123,
      "step": 70880
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7594435214996338,
      "learning_rate": 0.0004704605961262719,
      "loss": 2.0787,
      "step": 70881
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7648556232452393,
      "learning_rate": 0.000470457230031325,
      "loss": 2.0822,
      "step": 70882
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.9456337690353394,
      "learning_rate": 0.0004704538639046869,
      "loss": 2.2702,
      "step": 70883
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.795005202293396,
      "learning_rate": 0.0004704504977463583,
      "loss": 2.2937,
      "step": 70884
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7694198489189148,
      "learning_rate": 0.00047044713155633964,
      "loss": 2.3884,
      "step": 70885
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.8388994336128235,
      "learning_rate": 0.0004704437653346317,
      "loss": 2.4219,
      "step": 70886
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7742712497711182,
      "learning_rate": 0.00047044039908123494,
      "loss": 2.2551,
      "step": 70887
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7682762742042542,
      "learning_rate": 0.00047043703279615016,
      "loss": 2.2036,
      "step": 70888
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7246298789978027,
      "learning_rate": 0.00047043366647937795,
      "loss": 2.2725,
      "step": 70889
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7464569807052612,
      "learning_rate": 0.00047043030013091885,
      "loss": 2.2552,
      "step": 70890
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.6889904737472534,
      "learning_rate": 0.00047042693375077356,
      "loss": 2.1523,
      "step": 70891
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7408833503723145,
      "learning_rate": 0.00047042356733894273,
      "loss": 2.1687,
      "step": 70892
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.711283266544342,
      "learning_rate": 0.00047042020089542685,
      "loss": 2.2775,
      "step": 70893
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.6826277375221252,
      "learning_rate": 0.0004704168344202267,
      "loss": 1.9606,
      "step": 70894
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7680367827415466,
      "learning_rate": 0.0004704134679133428,
      "loss": 2.3949,
      "step": 70895
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7720019817352295,
      "learning_rate": 0.00047041010137477586,
      "loss": 2.5125,
      "step": 70896
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.8419980406761169,
      "learning_rate": 0.00047040673480452645,
      "loss": 2.346,
      "step": 70897
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7215747237205505,
      "learning_rate": 0.0004704033682025952,
      "loss": 2.2479,
      "step": 70898
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7493071556091309,
      "learning_rate": 0.0004704000015689828,
      "loss": 2.2926,
      "step": 70899
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7181157469749451,
      "learning_rate": 0.0004703966349036898,
      "loss": 2.0623,
      "step": 70900
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.9071401357650757,
      "learning_rate": 0.00047039326820671676,
      "loss": 2.3972,
      "step": 70901
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.8394542932510376,
      "learning_rate": 0.0004703899014780645,
      "loss": 2.2584,
      "step": 70902
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.8289948105812073,
      "learning_rate": 0.00047038653471773357,
      "loss": 2.1353,
      "step": 70903
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.9395891427993774,
      "learning_rate": 0.00047038316792572443,
      "loss": 2.4378,
      "step": 70904
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7539211511611938,
      "learning_rate": 0.000470379801102038,
      "loss": 2.3231,
      "step": 70905
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.9350486993789673,
      "learning_rate": 0.00047037643424667464,
      "loss": 2.1464,
      "step": 70906
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7408434748649597,
      "learning_rate": 0.0004703730673596351,
      "loss": 2.2453,
      "step": 70907
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.6838836073875427,
      "learning_rate": 0.00047036970044092005,
      "loss": 2.3885,
      "step": 70908
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7850950956344604,
      "learning_rate": 0.00047036633349053,
      "loss": 2.4081,
      "step": 70909
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.8165579438209534,
      "learning_rate": 0.00047036296650846565,
      "loss": 2.2891,
      "step": 70910
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.9492335915565491,
      "learning_rate": 0.00047035959949472757,
      "loss": 1.8848,
      "step": 70911
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7541828751564026,
      "learning_rate": 0.0004703562324493165,
      "loss": 2.3374,
      "step": 70912
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7867885231971741,
      "learning_rate": 0.00047035286537223286,
      "loss": 2.2601,
      "step": 70913
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.8270335793495178,
      "learning_rate": 0.0004703494982634776,
      "loss": 2.2223,
      "step": 70914
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7170290946960449,
      "learning_rate": 0.00047034613112305096,
      "loss": 2.1072,
      "step": 70915
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.8187224268913269,
      "learning_rate": 0.0004703427639509539,
      "loss": 2.3661,
      "step": 70916
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.8046621680259705,
      "learning_rate": 0.00047033939674718687,
      "loss": 2.2805,
      "step": 70917
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.8165075182914734,
      "learning_rate": 0.0004703360295117505,
      "loss": 2.0553,
      "step": 70918
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7370122671127319,
      "learning_rate": 0.00047033266224464547,
      "loss": 2.2026,
      "step": 70919
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.8539823889732361,
      "learning_rate": 0.0004703292949458724,
      "loss": 2.052,
      "step": 70920
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.8519101738929749,
      "learning_rate": 0.0004703259276154318,
      "loss": 2.3421,
      "step": 70921
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.780134379863739,
      "learning_rate": 0.00047032256025332454,
      "loss": 2.3494,
      "step": 70922
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7683749198913574,
      "learning_rate": 0.000470319192859551,
      "loss": 2.2719,
      "step": 70923
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.8453258275985718,
      "learning_rate": 0.00047031582543411197,
      "loss": 2.0447,
      "step": 70924
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.9321858286857605,
      "learning_rate": 0.0004703124579770079,
      "loss": 2.3041,
      "step": 70925
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7693862915039062,
      "learning_rate": 0.0004703090904882396,
      "loss": 2.3004,
      "step": 70926
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.8491277098655701,
      "learning_rate": 0.00047030572296780765,
      "loss": 2.3038,
      "step": 70927
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.9677075743675232,
      "learning_rate": 0.0004703023554157126,
      "loss": 2.1573,
      "step": 70928
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7956902384757996,
      "learning_rate": 0.0004702989878319552,
      "loss": 2.1222,
      "step": 70929
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.8524754047393799,
      "learning_rate": 0.000470295620216536,
      "loss": 2.3902,
      "step": 70930
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.1548479795455933,
      "learning_rate": 0.00047029225256945556,
      "loss": 2.1408,
      "step": 70931
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7373194694519043,
      "learning_rate": 0.0004702888848907146,
      "loss": 2.3064,
      "step": 70932
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7903866767883301,
      "learning_rate": 0.00047028551718031374,
      "loss": 2.1043,
      "step": 70933
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.1538774967193604,
      "learning_rate": 0.0004702821494382536,
      "loss": 2.115,
      "step": 70934
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.847000002861023,
      "learning_rate": 0.00047027878166453483,
      "loss": 2.4163,
      "step": 70935
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7948179244995117,
      "learning_rate": 0.00047027541385915786,
      "loss": 1.9281,
      "step": 70936
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.9350362420082092,
      "learning_rate": 0.00047027204602212365,
      "loss": 2.1698,
      "step": 70937
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7521461844444275,
      "learning_rate": 0.00047026867815343253,
      "loss": 2.2324,
      "step": 70938
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7671036124229431,
      "learning_rate": 0.0004702653102530853,
      "loss": 2.2087,
      "step": 70939
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7659156918525696,
      "learning_rate": 0.0004702619423210825,
      "loss": 2.1287,
      "step": 70940
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.8053762912750244,
      "learning_rate": 0.00047025857435742494,
      "loss": 2.0871,
      "step": 70941
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7601727843284607,
      "learning_rate": 0.00047025520636211293,
      "loss": 2.2134,
      "step": 70942
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7512386441230774,
      "learning_rate": 0.0004702518383351473,
      "loss": 2.5888,
      "step": 70943
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7161017656326294,
      "learning_rate": 0.0004702484702765287,
      "loss": 2.0485,
      "step": 70944
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.8375412225723267,
      "learning_rate": 0.0004702451021862576,
      "loss": 2.3336,
      "step": 70945
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.809471070766449,
      "learning_rate": 0.0004702417340643348,
      "loss": 2.2549,
      "step": 70946
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7638370990753174,
      "learning_rate": 0.0004702383659107609,
      "loss": 2.3865,
      "step": 70947
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.8711905479431152,
      "learning_rate": 0.0004702349977255363,
      "loss": 2.0964,
      "step": 70948
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.8267832398414612,
      "learning_rate": 0.00047023162950866193,
      "loss": 2.1415,
      "step": 70949
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7418146729469299,
      "learning_rate": 0.00047022826126013825,
      "loss": 2.1438,
      "step": 70950
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.9558887481689453,
      "learning_rate": 0.00047022489297996595,
      "loss": 2.0422,
      "step": 70951
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7549802660942078,
      "learning_rate": 0.0004702215246681456,
      "loss": 2.0882,
      "step": 70952
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.806983470916748,
      "learning_rate": 0.00047021815632467787,
      "loss": 2.3302,
      "step": 70953
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7609391212463379,
      "learning_rate": 0.0004702147879495633,
      "loss": 2.3747,
      "step": 70954
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.8049367666244507,
      "learning_rate": 0.0004702114195428027,
      "loss": 2.3888,
      "step": 70955
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.9378583431243896,
      "learning_rate": 0.0004702080511043965,
      "loss": 2.1034,
      "step": 70956
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.717239260673523,
      "learning_rate": 0.0004702046826343455,
      "loss": 2.1797,
      "step": 70957
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.9124734401702881,
      "learning_rate": 0.0004702013141326502,
      "loss": 2.0982,
      "step": 70958
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.9470691084861755,
      "learning_rate": 0.00047019794559931125,
      "loss": 2.3675,
      "step": 70959
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.9419980645179749,
      "learning_rate": 0.00047019457703432923,
      "loss": 1.9697,
      "step": 70960
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.9985544085502625,
      "learning_rate": 0.0004701912084377049,
      "loss": 2.0623,
      "step": 70961
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.8799894452095032,
      "learning_rate": 0.00047018783980943884,
      "loss": 2.196,
      "step": 70962
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.8671882152557373,
      "learning_rate": 0.00047018447114953156,
      "loss": 2.2667,
      "step": 70963
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.8412954807281494,
      "learning_rate": 0.0004701811024579838,
      "loss": 2.474,
      "step": 70964
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.8151686787605286,
      "learning_rate": 0.00047017773373479613,
      "loss": 2.1753,
      "step": 70965
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.8726083040237427,
      "learning_rate": 0.0004701743649799693,
      "loss": 1.9835,
      "step": 70966
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7839812636375427,
      "learning_rate": 0.0004701709961935038,
      "loss": 2.1326,
      "step": 70967
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.813903272151947,
      "learning_rate": 0.00047016762737540023,
      "loss": 2.2848,
      "step": 70968
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.8118608593940735,
      "learning_rate": 0.00047016425852565945,
      "loss": 2.1381,
      "step": 70969
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7219974398612976,
      "learning_rate": 0.0004701608896442817,
      "loss": 2.5233,
      "step": 70970
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.8973559737205505,
      "learning_rate": 0.00047015752073126797,
      "loss": 2.1262,
      "step": 70971
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.8262009620666504,
      "learning_rate": 0.0004701541517866188,
      "loss": 1.9105,
      "step": 70972
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.739855170249939,
      "learning_rate": 0.00047015078281033464,
      "loss": 2.1028,
      "step": 70973
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.1052565574645996,
      "learning_rate": 0.00047014741380241624,
      "loss": 2.1177,
      "step": 70974
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7514440417289734,
      "learning_rate": 0.0004701440447628643,
      "loss": 2.0076,
      "step": 70975
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7749494910240173,
      "learning_rate": 0.00047014067569167933,
      "loss": 2.0389,
      "step": 70976
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.912009596824646,
      "learning_rate": 0.000470137306588862,
      "loss": 2.2424,
      "step": 70977
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.8889318108558655,
      "learning_rate": 0.000470133937454413,
      "loss": 2.4755,
      "step": 70978
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7412186861038208,
      "learning_rate": 0.0004701305682883327,
      "loss": 2.2299,
      "step": 70979
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.9292322993278503,
      "learning_rate": 0.00047012719909062217,
      "loss": 2.2853,
      "step": 70980
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.9144635200500488,
      "learning_rate": 0.0004701238298612816,
      "loss": 2.139,
      "step": 70981
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7997433543205261,
      "learning_rate": 0.0004701204606003119,
      "loss": 2.3196,
      "step": 70982
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7875404953956604,
      "learning_rate": 0.0004701170913077135,
      "loss": 2.2294,
      "step": 70983
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.8113964200019836,
      "learning_rate": 0.00047011372198348724,
      "loss": 2.2309,
      "step": 70984
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7688676714897156,
      "learning_rate": 0.0004701103526276335,
      "loss": 2.0959,
      "step": 70985
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.8348047137260437,
      "learning_rate": 0.00047010698324015306,
      "loss": 2.5031,
      "step": 70986
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7825014591217041,
      "learning_rate": 0.0004701036138210466,
      "loss": 2.1682,
      "step": 70987
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7752015590667725,
      "learning_rate": 0.00047010024437031466,
      "loss": 2.1798,
      "step": 70988
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.0866055488586426,
      "learning_rate": 0.0004700968748879578,
      "loss": 2.2675,
      "step": 70989
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.8134288787841797,
      "learning_rate": 0.00047009350537397676,
      "loss": 2.3396,
      "step": 70990
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7625988125801086,
      "learning_rate": 0.0004700901358283721,
      "loss": 2.1748,
      "step": 70991
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.1929994821548462,
      "learning_rate": 0.00047008676625114444,
      "loss": 1.9316,
      "step": 70992
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7590587735176086,
      "learning_rate": 0.00047008339664229463,
      "loss": 2.2603,
      "step": 70993
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.716169536113739,
      "learning_rate": 0.0004700800270018229,
      "loss": 2.0184,
      "step": 70994
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.9005177617073059,
      "learning_rate": 0.0004700766573297302,
      "loss": 2.1598,
      "step": 70995
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.8741752505302429,
      "learning_rate": 0.00047007328762601697,
      "loss": 2.1255,
      "step": 70996
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7059884667396545,
      "learning_rate": 0.0004700699178906839,
      "loss": 2.1042,
      "step": 70997
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.761971116065979,
      "learning_rate": 0.0004700665481237317,
      "loss": 2.1777,
      "step": 70998
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7463113069534302,
      "learning_rate": 0.00047006317832516085,
      "loss": 2.1454,
      "step": 70999
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.8263766169548035,
      "learning_rate": 0.00047005980849497203,
      "loss": 2.1767,
      "step": 71000
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7738704085350037,
      "learning_rate": 0.0004700564386331659,
      "loss": 2.0629,
      "step": 71001
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.8975642919540405,
      "learning_rate": 0.0004700530687397431,
      "loss": 2.1015,
      "step": 71002
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7634577751159668,
      "learning_rate": 0.00047004969881470424,
      "loss": 2.2337,
      "step": 71003
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.8488197326660156,
      "learning_rate": 0.0004700463288580499,
      "loss": 2.2051,
      "step": 71004
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7405781149864197,
      "learning_rate": 0.0004700429588697807,
      "loss": 2.3618,
      "step": 71005
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7339516878128052,
      "learning_rate": 0.0004700395888498974,
      "loss": 2.3759,
      "step": 71006
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7364751100540161,
      "learning_rate": 0.00047003621879840047,
      "loss": 2.1996,
      "step": 71007
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7890857458114624,
      "learning_rate": 0.0004700328487152906,
      "loss": 2.1372,
      "step": 71008
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.8402655124664307,
      "learning_rate": 0.00047002947860056837,
      "loss": 2.3581,
      "step": 71009
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7176071405410767,
      "learning_rate": 0.00047002610845423443,
      "loss": 2.1971,
      "step": 71010
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.8287169933319092,
      "learning_rate": 0.0004700227382762896,
      "loss": 2.1858,
      "step": 71011
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.8463764190673828,
      "learning_rate": 0.00047001936806673417,
      "loss": 2.2887,
      "step": 71012
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.8363721370697021,
      "learning_rate": 0.000470015997825569,
      "loss": 2.1782,
      "step": 71013
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7451871037483215,
      "learning_rate": 0.0004700126275527946,
      "loss": 2.0815,
      "step": 71014
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7886339426040649,
      "learning_rate": 0.0004700092572484117,
      "loss": 2.0611,
      "step": 71015
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7844886183738708,
      "learning_rate": 0.00047000588691242085,
      "loss": 2.1158,
      "step": 71016
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.8755831122398376,
      "learning_rate": 0.00047000251654482267,
      "loss": 2.3254,
      "step": 71017
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.9091007709503174,
      "learning_rate": 0.0004699991461456179,
      "loss": 2.2796,
      "step": 71018
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.8302921056747437,
      "learning_rate": 0.0004699957757148069,
      "loss": 2.1897,
      "step": 71019
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.8548622131347656,
      "learning_rate": 0.0004699924052523907,
      "loss": 2.3952,
      "step": 71020
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.977820634841919,
      "learning_rate": 0.00046998903475836955,
      "loss": 2.2913,
      "step": 71021
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.9037764668464661,
      "learning_rate": 0.00046998566423274424,
      "loss": 2.1434,
      "step": 71022
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7262268662452698,
      "learning_rate": 0.0004699822936755154,
      "loss": 2.2872,
      "step": 71023
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.9275442361831665,
      "learning_rate": 0.0004699789230866837,
      "loss": 2.308,
      "step": 71024
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.8225081562995911,
      "learning_rate": 0.0004699755524662497,
      "loss": 2.0369,
      "step": 71025
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.8830583095550537,
      "learning_rate": 0.000469972181814214,
      "loss": 1.9871,
      "step": 71026
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7471771836280823,
      "learning_rate": 0.0004699688111305772,
      "loss": 2.4477,
      "step": 71027
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.0399326086044312,
      "learning_rate": 0.00046996544041534013,
      "loss": 2.317,
      "step": 71028
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.958024263381958,
      "learning_rate": 0.0004699620696685032,
      "loss": 2.2604,
      "step": 71029
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7689461708068848,
      "learning_rate": 0.0004699586988900671,
      "loss": 2.3274,
      "step": 71030
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7633888721466064,
      "learning_rate": 0.00046995532808003247,
      "loss": 2.1601,
      "step": 71031
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.9384682774543762,
      "learning_rate": 0.00046995195723839996,
      "loss": 2.4319,
      "step": 71032
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.8369975090026855,
      "learning_rate": 0.00046994858636517016,
      "loss": 2.3669,
      "step": 71033
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7406707406044006,
      "learning_rate": 0.0004699452154603438,
      "loss": 2.1273,
      "step": 71034
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.804926335811615,
      "learning_rate": 0.0004699418445239212,
      "loss": 2.2756,
      "step": 71035
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.8433033227920532,
      "learning_rate": 0.00046993847355590346,
      "loss": 2.0716,
      "step": 71036
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7490946650505066,
      "learning_rate": 0.00046993510255629076,
      "loss": 2.1845,
      "step": 71037
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.9640869498252869,
      "learning_rate": 0.0004699317315250839,
      "loss": 2.3208,
      "step": 71038
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.0382766723632812,
      "learning_rate": 0.0004699283604622837,
      "loss": 2.2975,
      "step": 71039
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.9631460309028625,
      "learning_rate": 0.0004699249893678905,
      "loss": 2.3643,
      "step": 71040
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7581190466880798,
      "learning_rate": 0.00046992161824190503,
      "loss": 2.2832,
      "step": 71041
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.878973126411438,
      "learning_rate": 0.00046991824708432793,
      "loss": 2.2304,
      "step": 71042
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.8583933711051941,
      "learning_rate": 0.0004699148758951599,
      "loss": 2.1573,
      "step": 71043
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7958199381828308,
      "learning_rate": 0.00046991150467440144,
      "loss": 2.3849,
      "step": 71044
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7654781341552734,
      "learning_rate": 0.00046990813342205325,
      "loss": 2.284,
      "step": 71045
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7980263829231262,
      "learning_rate": 0.0004699047621381158,
      "loss": 2.357,
      "step": 71046
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.9532409310340881,
      "learning_rate": 0.00046990139082258995,
      "loss": 2.0135,
      "step": 71047
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7567452192306519,
      "learning_rate": 0.00046989801947547625,
      "loss": 2.2927,
      "step": 71048
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.9275072813034058,
      "learning_rate": 0.00046989464809677525,
      "loss": 2.1878,
      "step": 71049
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.8393030762672424,
      "learning_rate": 0.0004698912766864876,
      "loss": 2.3348,
      "step": 71050
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7919554710388184,
      "learning_rate": 0.000469887905244614,
      "loss": 2.3524,
      "step": 71051
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.8303858041763306,
      "learning_rate": 0.00046988453377115505,
      "loss": 2.2512,
      "step": 71052
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.8831527233123779,
      "learning_rate": 0.0004698811622661113,
      "loss": 2.094,
      "step": 71053
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7867283225059509,
      "learning_rate": 0.0004698777907294835,
      "loss": 2.2258,
      "step": 71054
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7108339071273804,
      "learning_rate": 0.0004698744191612722,
      "loss": 2.1661,
      "step": 71055
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.9555689096450806,
      "learning_rate": 0.000469871047561478,
      "loss": 2.1456,
      "step": 71056
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7115049362182617,
      "learning_rate": 0.00046986767593010156,
      "loss": 2.1859,
      "step": 71057
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7640561461448669,
      "learning_rate": 0.00046986430426714356,
      "loss": 2.5234,
      "step": 71058
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.877060055732727,
      "learning_rate": 0.0004698609325726046,
      "loss": 2.4089,
      "step": 71059
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7519638538360596,
      "learning_rate": 0.0004698575608464852,
      "loss": 2.2127,
      "step": 71060
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7690283060073853,
      "learning_rate": 0.00046985418908878615,
      "loss": 2.2131,
      "step": 71061
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7591676712036133,
      "learning_rate": 0.0004698508172995079,
      "loss": 2.2623,
      "step": 71062
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7411906719207764,
      "learning_rate": 0.00046984744547865127,
      "loss": 2.0418,
      "step": 71063
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.8728219866752625,
      "learning_rate": 0.0004698440736262168,
      "loss": 2.4333,
      "step": 71064
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7516840696334839,
      "learning_rate": 0.00046984070174220505,
      "loss": 2.1586,
      "step": 71065
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.1257493495941162,
      "learning_rate": 0.0004698373298266168,
      "loss": 2.1593,
      "step": 71066
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.8032985329627991,
      "learning_rate": 0.0004698339578794525,
      "loss": 2.1582,
      "step": 71067
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.9346776008605957,
      "learning_rate": 0.00046983058590071284,
      "loss": 2.2541,
      "step": 71068
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.8484782576560974,
      "learning_rate": 0.00046982721389039853,
      "loss": 2.1569,
      "step": 71069
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.9577776789665222,
      "learning_rate": 0.00046982384184851014,
      "loss": 1.9473,
      "step": 71070
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7560974359512329,
      "learning_rate": 0.00046982046977504826,
      "loss": 2.1925,
      "step": 71071
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7228862047195435,
      "learning_rate": 0.00046981709767001355,
      "loss": 2.3814,
      "step": 71072
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.809539258480072,
      "learning_rate": 0.0004698137255334067,
      "loss": 2.2988,
      "step": 71073
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7808197736740112,
      "learning_rate": 0.0004698103533652281,
      "loss": 2.1766,
      "step": 71074
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7052937150001526,
      "learning_rate": 0.00046980698116547875,
      "loss": 2.1332,
      "step": 71075
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7617934346199036,
      "learning_rate": 0.000469803608934159,
      "loss": 2.1774,
      "step": 71076
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.754208505153656,
      "learning_rate": 0.0004698002366712695,
      "loss": 2.3061,
      "step": 71077
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.8477672934532166,
      "learning_rate": 0.000469796864376811,
      "loss": 2.2463,
      "step": 71078
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.8729967474937439,
      "learning_rate": 0.000469793492050784,
      "loss": 2.1608,
      "step": 71079
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.9920881986618042,
      "learning_rate": 0.0004697901196931892,
      "loss": 2.2441,
      "step": 71080
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.8015496134757996,
      "learning_rate": 0.0004697867473040273,
      "loss": 2.1235,
      "step": 71081
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.8996719121932983,
      "learning_rate": 0.00046978337488329876,
      "loss": 2.2405,
      "step": 71082
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.8210344314575195,
      "learning_rate": 0.00046978000243100423,
      "loss": 2.3513,
      "step": 71083
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7367826700210571,
      "learning_rate": 0.00046977662994714456,
      "loss": 2.0901,
      "step": 71084
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7601659893989563,
      "learning_rate": 0.0004697732574317201,
      "loss": 2.2622,
      "step": 71085
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.0511789321899414,
      "learning_rate": 0.0004697698848847316,
      "loss": 2.2249,
      "step": 71086
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.0126477479934692,
      "learning_rate": 0.0004697665123061797,
      "loss": 2.4172,
      "step": 71087
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7543436288833618,
      "learning_rate": 0.00046976313969606493,
      "loss": 2.1381,
      "step": 71088
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.8157713413238525,
      "learning_rate": 0.000469759767054388,
      "loss": 2.0592,
      "step": 71089
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.8026092052459717,
      "learning_rate": 0.00046975639438114964,
      "loss": 2.1082,
      "step": 71090
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.8112847208976746,
      "learning_rate": 0.00046975302167635017,
      "loss": 2.3644,
      "step": 71091
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7760255336761475,
      "learning_rate": 0.0004697496489399905,
      "loss": 2.4399,
      "step": 71092
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7336520552635193,
      "learning_rate": 0.0004697462761720712,
      "loss": 2.004,
      "step": 71093
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.8583621382713318,
      "learning_rate": 0.00046974290337259287,
      "loss": 2.0906,
      "step": 71094
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7148640751838684,
      "learning_rate": 0.00046973953054155605,
      "loss": 2.0279,
      "step": 71095
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7768535017967224,
      "learning_rate": 0.0004697361576789615,
      "loss": 2.0881,
      "step": 71096
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.8086167573928833,
      "learning_rate": 0.0004697327847848098,
      "loss": 2.2005,
      "step": 71097
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7662615180015564,
      "learning_rate": 0.0004697294118591016,
      "loss": 2.2315,
      "step": 71098
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7928082942962646,
      "learning_rate": 0.00046972603890183743,
      "loss": 2.1585,
      "step": 71099
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7832741141319275,
      "learning_rate": 0.00046972266591301803,
      "loss": 2.2333,
      "step": 71100
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.8846216797828674,
      "learning_rate": 0.000469719292892644,
      "loss": 2.0937,
      "step": 71101
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7613566517829895,
      "learning_rate": 0.00046971591984071586,
      "loss": 2.1808,
      "step": 71102
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.8208436369895935,
      "learning_rate": 0.00046971254675723444,
      "loss": 2.1826,
      "step": 71103
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.8524882793426514,
      "learning_rate": 0.0004697091736422002,
      "loss": 2.2516,
      "step": 71104
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.8660019636154175,
      "learning_rate": 0.0004697058004956138,
      "loss": 2.2308,
      "step": 71105
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.8322111964225769,
      "learning_rate": 0.00046970242731747596,
      "loss": 2.0655,
      "step": 71106
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7725794315338135,
      "learning_rate": 0.00046969905410778717,
      "loss": 2.1217,
      "step": 71107
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.9321438670158386,
      "learning_rate": 0.0004696956808665481,
      "loss": 2.0833,
      "step": 71108
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7772123217582703,
      "learning_rate": 0.00046969230759375945,
      "loss": 2.1974,
      "step": 71109
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.8369829654693604,
      "learning_rate": 0.0004696889342894218,
      "loss": 2.2712,
      "step": 71110
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.9699594378471375,
      "learning_rate": 0.00046968556095353575,
      "loss": 2.1382,
      "step": 71111
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7936943769454956,
      "learning_rate": 0.000469682187586102,
      "loss": 2.1614,
      "step": 71112
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.696379542350769,
      "learning_rate": 0.0004696788141871211,
      "loss": 2.0306,
      "step": 71113
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.8185177445411682,
      "learning_rate": 0.0004696754407565936,
      "loss": 2.0217,
      "step": 71114
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.0776050090789795,
      "learning_rate": 0.00046967206729452035,
      "loss": 2.4196,
      "step": 71115
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7564767003059387,
      "learning_rate": 0.0004696686938009019,
      "loss": 2.1214,
      "step": 71116
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.8442845940589905,
      "learning_rate": 0.00046966532027573873,
      "loss": 2.3565,
      "step": 71117
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.924272894859314,
      "learning_rate": 0.00046966194671903163,
      "loss": 2.3413,
      "step": 71118
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7601479291915894,
      "learning_rate": 0.00046965857313078114,
      "loss": 2.3883,
      "step": 71119
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.837956964969635,
      "learning_rate": 0.00046965519951098795,
      "loss": 2.1584,
      "step": 71120
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.8892877101898193,
      "learning_rate": 0.00046965182585965266,
      "loss": 2.0766,
      "step": 71121
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.9132624864578247,
      "learning_rate": 0.00046964845217677587,
      "loss": 2.1285,
      "step": 71122
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7480248212814331,
      "learning_rate": 0.0004696450784623582,
      "loss": 1.9991,
      "step": 71123
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.9032580256462097,
      "learning_rate": 0.00046964170471640044,
      "loss": 2.3661,
      "step": 71124
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.9171069264411926,
      "learning_rate": 0.000469638330938903,
      "loss": 2.0774,
      "step": 71125
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7838513255119324,
      "learning_rate": 0.00046963495712986654,
      "loss": 2.1994,
      "step": 71126
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7875299453735352,
      "learning_rate": 0.0004696315832892918,
      "loss": 2.3541,
      "step": 71127
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.8306875228881836,
      "learning_rate": 0.00046962820941717933,
      "loss": 2.3167,
      "step": 71128
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.8665170073509216,
      "learning_rate": 0.0004696248355135298,
      "loss": 2.141,
      "step": 71129
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.8360390067100525,
      "learning_rate": 0.0004696214615783437,
      "loss": 2.1875,
      "step": 71130
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7090974450111389,
      "learning_rate": 0.0004696180876116219,
      "loss": 2.306,
      "step": 71131
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.700793981552124,
      "learning_rate": 0.00046961471361336483,
      "loss": 2.1421,
      "step": 71132
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7998279929161072,
      "learning_rate": 0.00046961133958357327,
      "loss": 2.5,
      "step": 71133
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.8107281923294067,
      "learning_rate": 0.0004696079655222476,
      "loss": 2.1639,
      "step": 71134
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.752536952495575,
      "learning_rate": 0.00046960459142938874,
      "loss": 1.9233,
      "step": 71135
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7989194989204407,
      "learning_rate": 0.0004696012173049971,
      "loss": 2.1903,
      "step": 71136
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.8117133975028992,
      "learning_rate": 0.0004695978431490735,
      "loss": 2.0324,
      "step": 71137
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7701858282089233,
      "learning_rate": 0.0004695944689616184,
      "loss": 2.2788,
      "step": 71138
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.8056370615959167,
      "learning_rate": 0.00046959109474263246,
      "loss": 2.2961,
      "step": 71139
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.702694833278656,
      "learning_rate": 0.00046958772049211637,
      "loss": 2.4381,
      "step": 71140
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.8382788896560669,
      "learning_rate": 0.0004695843462100707,
      "loss": 2.2888,
      "step": 71141
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.9738512635231018,
      "learning_rate": 0.00046958097189649616,
      "loss": 2.2371,
      "step": 71142
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7538637518882751,
      "learning_rate": 0.00046957759755139323,
      "loss": 2.2798,
      "step": 71143
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.8086113929748535,
      "learning_rate": 0.00046957422317476263,
      "loss": 2.1454,
      "step": 71144
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.8225440979003906,
      "learning_rate": 0.000469570848766605,
      "loss": 2.3678,
      "step": 71145
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.8275408744812012,
      "learning_rate": 0.000469567474326921,
      "loss": 2.0174,
      "step": 71146
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.8049620389938354,
      "learning_rate": 0.0004695640998557111,
      "loss": 2.2943,
      "step": 71147
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7403361201286316,
      "learning_rate": 0.00046956072535297614,
      "loss": 2.2542,
      "step": 71148
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7656792402267456,
      "learning_rate": 0.0004695573508187166,
      "loss": 2.2129,
      "step": 71149
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7597172260284424,
      "learning_rate": 0.00046955397625293314,
      "loss": 1.9977,
      "step": 71150
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7237904667854309,
      "learning_rate": 0.00046955060165562636,
      "loss": 2.5354,
      "step": 71151
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7821540236473083,
      "learning_rate": 0.00046954722702679695,
      "loss": 2.1012,
      "step": 71152
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7703256607055664,
      "learning_rate": 0.0004695438523664455,
      "loss": 2.259,
      "step": 71153
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.6895791888237,
      "learning_rate": 0.00046954047767457276,
      "loss": 2.0366,
      "step": 71154
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7404707670211792,
      "learning_rate": 0.0004695371029511791,
      "loss": 2.3777,
      "step": 71155
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.8145004510879517,
      "learning_rate": 0.0004695337281962654,
      "loss": 2.3075,
      "step": 71156
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7396436929702759,
      "learning_rate": 0.00046953035340983206,
      "loss": 2.0531,
      "step": 71157
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.8127154111862183,
      "learning_rate": 0.0004695269785918799,
      "loss": 2.5715,
      "step": 71158
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7451320290565491,
      "learning_rate": 0.0004695236037424095,
      "loss": 2.2934,
      "step": 71159
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7421842217445374,
      "learning_rate": 0.0004695202288614215,
      "loss": 2.2442,
      "step": 71160
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7845868468284607,
      "learning_rate": 0.00046951685394891636,
      "loss": 1.9096,
      "step": 71161
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.8146725296974182,
      "learning_rate": 0.00046951347900489485,
      "loss": 2.2349,
      "step": 71162
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.8123510479927063,
      "learning_rate": 0.0004695101040293577,
      "loss": 2.432,
      "step": 71163
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.8787915110588074,
      "learning_rate": 0.0004695067290223053,
      "loss": 2.2482,
      "step": 71164
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.812199056148529,
      "learning_rate": 0.0004695033539837384,
      "loss": 2.2542,
      "step": 71165
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.0906075239181519,
      "learning_rate": 0.0004694999789136577,
      "loss": 2.2346,
      "step": 71166
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7190083861351013,
      "learning_rate": 0.0004694966038120638,
      "loss": 2.1577,
      "step": 71167
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7510330677032471,
      "learning_rate": 0.0004694932286789572,
      "loss": 2.194,
      "step": 71168
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7882880568504333,
      "learning_rate": 0.0004694898535143386,
      "loss": 2.2301,
      "step": 71169
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.8554412722587585,
      "learning_rate": 0.0004694864783182087,
      "loss": 2.3099,
      "step": 71170
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.8527287840843201,
      "learning_rate": 0.00046948310309056807,
      "loss": 2.2446,
      "step": 71171
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.8491634726524353,
      "learning_rate": 0.00046947972783141723,
      "loss": 2.2532,
      "step": 71172
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7416828870773315,
      "learning_rate": 0.000469476352540757,
      "loss": 2.1684,
      "step": 71173
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7333333492279053,
      "learning_rate": 0.0004694729772185878,
      "loss": 2.3218,
      "step": 71174
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.8216413855552673,
      "learning_rate": 0.0004694696018649104,
      "loss": 2.5186,
      "step": 71175
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7954718470573425,
      "learning_rate": 0.00046946622647972563,
      "loss": 2.1505,
      "step": 71176
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.8137170672416687,
      "learning_rate": 0.00046946285106303366,
      "loss": 2.357,
      "step": 71177
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7726633548736572,
      "learning_rate": 0.00046945947561483536,
      "loss": 2.2188,
      "step": 71178
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.816368579864502,
      "learning_rate": 0.00046945610013513145,
      "loss": 1.9742,
      "step": 71179
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.791265606880188,
      "learning_rate": 0.00046945272462392237,
      "loss": 2.1281,
      "step": 71180
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7933664321899414,
      "learning_rate": 0.00046944934908120886,
      "loss": 2.3616,
      "step": 71181
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.8065706491470337,
      "learning_rate": 0.0004694459735069915,
      "loss": 2.1792,
      "step": 71182
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7130695581436157,
      "learning_rate": 0.000469442597901271,
      "loss": 2.2745,
      "step": 71183
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.8143773674964905,
      "learning_rate": 0.00046943922226404776,
      "loss": 2.2312,
      "step": 71184
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7988095283508301,
      "learning_rate": 0.00046943584659532285,
      "loss": 2.1678,
      "step": 71185
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7371955513954163,
      "learning_rate": 0.00046943247089509634,
      "loss": 2.1013,
      "step": 71186
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7942442297935486,
      "learning_rate": 0.00046942909516336926,
      "loss": 2.2952,
      "step": 71187
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7991564273834229,
      "learning_rate": 0.0004694257194001421,
      "loss": 2.3053,
      "step": 71188
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7494426369667053,
      "learning_rate": 0.0004694223436054155,
      "loss": 2.2631,
      "step": 71189
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7370967864990234,
      "learning_rate": 0.00046941896777919005,
      "loss": 2.0928,
      "step": 71190
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.8177375197410583,
      "learning_rate": 0.0004694155919214665,
      "loss": 2.2403,
      "step": 71191
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7766931056976318,
      "learning_rate": 0.0004694122160322453,
      "loss": 2.2673,
      "step": 71192
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.785662055015564,
      "learning_rate": 0.0004694088401115272,
      "loss": 2.2583,
      "step": 71193
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.8954848647117615,
      "learning_rate": 0.0004694054641593128,
      "loss": 2.1613,
      "step": 71194
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.786473274230957,
      "learning_rate": 0.0004694020881756028,
      "loss": 2.366,
      "step": 71195
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7723281383514404,
      "learning_rate": 0.00046939871216039765,
      "loss": 2.2217,
      "step": 71196
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7066882848739624,
      "learning_rate": 0.00046939533611369813,
      "loss": 2.1358,
      "step": 71197
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7332935333251953,
      "learning_rate": 0.00046939196003550486,
      "loss": 2.1991,
      "step": 71198
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7861829400062561,
      "learning_rate": 0.0004693885839258184,
      "loss": 2.2654,
      "step": 71199
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.703572154045105,
      "learning_rate": 0.0004693852077846393,
      "loss": 2.1453,
      "step": 71200
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.720975935459137,
      "learning_rate": 0.00046938183161196847,
      "loss": 2.1203,
      "step": 71201
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7498924136161804,
      "learning_rate": 0.0004693784554078062,
      "loss": 2.3268,
      "step": 71202
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7586283683776855,
      "learning_rate": 0.00046937507917215344,
      "loss": 2.3992,
      "step": 71203
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.8182945251464844,
      "learning_rate": 0.0004693717029050105,
      "loss": 2.1871,
      "step": 71204
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.8485960960388184,
      "learning_rate": 0.00046936832660637824,
      "loss": 2.0991,
      "step": 71205
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7042450904846191,
      "learning_rate": 0.0004693649502762572,
      "loss": 2.0879,
      "step": 71206
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7649405598640442,
      "learning_rate": 0.000469361573914648,
      "loss": 2.1596,
      "step": 71207
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.747990608215332,
      "learning_rate": 0.0004693581975215513,
      "loss": 2.4391,
      "step": 71208
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7584351897239685,
      "learning_rate": 0.0004693548210969677,
      "loss": 2.0945,
      "step": 71209
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7609873414039612,
      "learning_rate": 0.00046935144464089784,
      "loss": 2.3503,
      "step": 71210
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7561546564102173,
      "learning_rate": 0.00046934806815334236,
      "loss": 1.8532,
      "step": 71211
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7605714201927185,
      "learning_rate": 0.0004693446916343019,
      "loss": 2.175,
      "step": 71212
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.769930899143219,
      "learning_rate": 0.0004693413150837771,
      "loss": 2.1427,
      "step": 71213
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7941585183143616,
      "learning_rate": 0.0004693379385017685,
      "loss": 2.2545,
      "step": 71214
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7941270470619202,
      "learning_rate": 0.00046933456188827675,
      "loss": 1.924,
      "step": 71215
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7288497090339661,
      "learning_rate": 0.0004693311852433026,
      "loss": 2.4157,
      "step": 71216
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7612256407737732,
      "learning_rate": 0.00046932780856684647,
      "loss": 2.1633,
      "step": 71217
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.764843225479126,
      "learning_rate": 0.0004693244318589092,
      "loss": 2.353,
      "step": 71218
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7296152114868164,
      "learning_rate": 0.00046932105511949136,
      "loss": 2.3901,
      "step": 71219
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.8395226001739502,
      "learning_rate": 0.00046931767834859335,
      "loss": 2.2777,
      "step": 71220
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7868028283119202,
      "learning_rate": 0.0004693143015462162,
      "loss": 2.2393,
      "step": 71221
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7339128851890564,
      "learning_rate": 0.0004693109247123602,
      "loss": 2.1175,
      "step": 71222
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7273526787757874,
      "learning_rate": 0.00046930754784702607,
      "loss": 2.2718,
      "step": 71223
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7804350256919861,
      "learning_rate": 0.00046930417095021463,
      "loss": 2.2044,
      "step": 71224
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.6790159940719604,
      "learning_rate": 0.0004693007940219262,
      "loss": 2.1437,
      "step": 71225
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.697843074798584,
      "learning_rate": 0.00046929741706216153,
      "loss": 2.2611,
      "step": 71226
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7712956070899963,
      "learning_rate": 0.00046929404007092144,
      "loss": 2.2638,
      "step": 71227
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7829869389533997,
      "learning_rate": 0.0004692906630482063,
      "loss": 2.0674,
      "step": 71228
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7604697346687317,
      "learning_rate": 0.0004692872859940168,
      "loss": 2.1508,
      "step": 71229
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.8523040413856506,
      "learning_rate": 0.0004692839089083536,
      "loss": 2.3623,
      "step": 71230
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7453020811080933,
      "learning_rate": 0.00046928053179121743,
      "loss": 2.1649,
      "step": 71231
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7159785032272339,
      "learning_rate": 0.0004692771546426087,
      "loss": 2.1502,
      "step": 71232
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7710793614387512,
      "learning_rate": 0.00046927377746252814,
      "loss": 2.1571,
      "step": 71233
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.817050039768219,
      "learning_rate": 0.00046927040025097655,
      "loss": 2.1253,
      "step": 71234
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7852981090545654,
      "learning_rate": 0.0004692670230079542,
      "loss": 2.2395,
      "step": 71235
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7797883152961731,
      "learning_rate": 0.000469263645733462,
      "loss": 2.1444,
      "step": 71236
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7912221550941467,
      "learning_rate": 0.0004692602684275006,
      "loss": 2.2469,
      "step": 71237
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7220441699028015,
      "learning_rate": 0.00046925689109007036,
      "loss": 2.0908,
      "step": 71238
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7632505893707275,
      "learning_rate": 0.00046925351372117214,
      "loss": 2.1741,
      "step": 71239
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.8476550579071045,
      "learning_rate": 0.00046925013632080647,
      "loss": 2.3209,
      "step": 71240
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7465649843215942,
      "learning_rate": 0.00046924675888897405,
      "loss": 2.369,
      "step": 71241
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7842036485671997,
      "learning_rate": 0.00046924338142567543,
      "loss": 2.2355,
      "step": 71242
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.745313823223114,
      "learning_rate": 0.0004692400039309113,
      "loss": 2.2226,
      "step": 71243
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.83364337682724,
      "learning_rate": 0.0004692366264046822,
      "loss": 2.4414,
      "step": 71244
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.9131232500076294,
      "learning_rate": 0.0004692332488469888,
      "loss": 2.3112,
      "step": 71245
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7688121795654297,
      "learning_rate": 0.0004692298712578318,
      "loss": 2.3204,
      "step": 71246
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7689874768257141,
      "learning_rate": 0.0004692264936372118,
      "loss": 2.4601,
      "step": 71247
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.8105661273002625,
      "learning_rate": 0.0004692231159851294,
      "loss": 2.1826,
      "step": 71248
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.799302339553833,
      "learning_rate": 0.0004692197383015852,
      "loss": 2.2957,
      "step": 71249
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7196795344352722,
      "learning_rate": 0.0004692163605865798,
      "loss": 2.202,
      "step": 71250
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7308626770973206,
      "learning_rate": 0.000469212982840114,
      "loss": 2.0731,
      "step": 71251
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.8542506694793701,
      "learning_rate": 0.00046920960506218823,
      "loss": 2.0898,
      "step": 71252
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7248513102531433,
      "learning_rate": 0.00046920622725280317,
      "loss": 2.3665,
      "step": 71253
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7905145883560181,
      "learning_rate": 0.00046920284941195955,
      "loss": 2.2119,
      "step": 71254
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.811336874961853,
      "learning_rate": 0.00046919947153965797,
      "loss": 2.211,
      "step": 71255
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7853356003761292,
      "learning_rate": 0.00046919609363589885,
      "loss": 2.1454,
      "step": 71256
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.8360818028450012,
      "learning_rate": 0.00046919271570068313,
      "loss": 2.0395,
      "step": 71257
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.8158029317855835,
      "learning_rate": 0.0004691893377340112,
      "loss": 2.3389,
      "step": 71258
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7920272350311279,
      "learning_rate": 0.00046918595973588387,
      "loss": 2.1711,
      "step": 71259
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7946906685829163,
      "learning_rate": 0.00046918258170630157,
      "loss": 2.1997,
      "step": 71260
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7825780510902405,
      "learning_rate": 0.00046917920364526516,
      "loss": 2.1134,
      "step": 71261
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.9510223865509033,
      "learning_rate": 0.000469175825552775,
      "loss": 2.1245,
      "step": 71262
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7738295197486877,
      "learning_rate": 0.0004691724474288319,
      "loss": 2.2175,
      "step": 71263
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7677026391029358,
      "learning_rate": 0.00046916906927343656,
      "loss": 2.1137,
      "step": 71264
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7699869275093079,
      "learning_rate": 0.00046916569108658944,
      "loss": 2.2189,
      "step": 71265
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.8087380528450012,
      "learning_rate": 0.00046916231286829113,
      "loss": 2.1409,
      "step": 71266
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.0047398805618286,
      "learning_rate": 0.0004691589346185425,
      "loss": 2.44,
      "step": 71267
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7937915921211243,
      "learning_rate": 0.00046915555633734394,
      "loss": 2.1285,
      "step": 71268
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.9153128266334534,
      "learning_rate": 0.00046915217802469614,
      "loss": 2.3448,
      "step": 71269
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.4161181449890137,
      "learning_rate": 0.00046914879968059986,
      "loss": 2.0633,
      "step": 71270
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7888526320457458,
      "learning_rate": 0.00046914542130505553,
      "loss": 2.1928,
      "step": 71271
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.8145788311958313,
      "learning_rate": 0.0004691420428980639,
      "loss": 2.3646,
      "step": 71272
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.9042935967445374,
      "learning_rate": 0.00046913866445962563,
      "loss": 2.348,
      "step": 71273
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.9730797410011292,
      "learning_rate": 0.00046913528598974125,
      "loss": 2.1729,
      "step": 71274
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7339223027229309,
      "learning_rate": 0.0004691319074884114,
      "loss": 2.1039,
      "step": 71275
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7198277711868286,
      "learning_rate": 0.0004691285289556368,
      "loss": 2.0943,
      "step": 71276
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7673199772834778,
      "learning_rate": 0.00046912515039141794,
      "loss": 2.2715,
      "step": 71277
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.9901590943336487,
      "learning_rate": 0.0004691217717957556,
      "loss": 2.2592,
      "step": 71278
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7074082493782043,
      "learning_rate": 0.00046911839316865023,
      "loss": 2.3357,
      "step": 71279
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.8100680708885193,
      "learning_rate": 0.00046911501451010263,
      "loss": 2.2603,
      "step": 71280
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.0806406736373901,
      "learning_rate": 0.0004691116358201133,
      "loss": 2.2405,
      "step": 71281
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.8042583465576172,
      "learning_rate": 0.00046910825709868293,
      "loss": 2.1735,
      "step": 71282
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.8942945003509521,
      "learning_rate": 0.00046910487834581223,
      "loss": 2.0667,
      "step": 71283
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.8362292647361755,
      "learning_rate": 0.00046910149956150163,
      "loss": 2.2004,
      "step": 71284
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.8503983020782471,
      "learning_rate": 0.000469098120745752,
      "loss": 2.4476,
      "step": 71285
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7535106539726257,
      "learning_rate": 0.0004690947418985637,
      "loss": 2.3316,
      "step": 71286
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7268832921981812,
      "learning_rate": 0.00046909136301993755,
      "loss": 2.168,
      "step": 71287
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.9761894941329956,
      "learning_rate": 0.00046908798410987413,
      "loss": 2.2248,
      "step": 71288
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.8346443772315979,
      "learning_rate": 0.0004690846051683741,
      "loss": 2.1708,
      "step": 71289
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7248466610908508,
      "learning_rate": 0.00046908122619543797,
      "loss": 2.0841,
      "step": 71290
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.9385316371917725,
      "learning_rate": 0.0004690778471910665,
      "loss": 2.4748,
      "step": 71291
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.9002946615219116,
      "learning_rate": 0.00046907446815526026,
      "loss": 2.33,
      "step": 71292
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7321273684501648,
      "learning_rate": 0.0004690710890880199,
      "loss": 2.2051,
      "step": 71293
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.8481924533843994,
      "learning_rate": 0.000469067709989346,
      "loss": 2.265,
      "step": 71294
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.9467728137969971,
      "learning_rate": 0.0004690643308592392,
      "loss": 2.0637,
      "step": 71295
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.8826081156730652,
      "learning_rate": 0.00046906095169770023,
      "loss": 2.1296,
      "step": 71296
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.8383388519287109,
      "learning_rate": 0.0004690575725047296,
      "loss": 2.1351,
      "step": 71297
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.9534717798233032,
      "learning_rate": 0.00046905419328032794,
      "loss": 2.3426,
      "step": 71298
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.9876376390457153,
      "learning_rate": 0.000469050814024496,
      "loss": 2.2205,
      "step": 71299
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.767997682094574,
      "learning_rate": 0.00046904743473723427,
      "loss": 1.9824,
      "step": 71300
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.8993046283721924,
      "learning_rate": 0.00046904405541854335,
      "loss": 2.3395,
      "step": 71301
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.9590296745300293,
      "learning_rate": 0.00046904067606842406,
      "loss": 2.1749,
      "step": 71302
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7879010438919067,
      "learning_rate": 0.0004690372966868769,
      "loss": 2.2019,
      "step": 71303
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.8573010563850403,
      "learning_rate": 0.0004690339172739025,
      "loss": 2.1573,
      "step": 71304
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7714481353759766,
      "learning_rate": 0.00046903053782950153,
      "loss": 2.2091,
      "step": 71305
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.9654969573020935,
      "learning_rate": 0.0004690271583536745,
      "loss": 2.3212,
      "step": 71306
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.851199209690094,
      "learning_rate": 0.0004690237788464223,
      "loss": 2.0571,
      "step": 71307
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.6824502944946289,
      "learning_rate": 0.00046902039930774524,
      "loss": 2.3634,
      "step": 71308
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7928300499916077,
      "learning_rate": 0.00046901701973764415,
      "loss": 2.2698,
      "step": 71309
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.8603675365447998,
      "learning_rate": 0.00046901364013611957,
      "loss": 2.3113,
      "step": 71310
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7713518738746643,
      "learning_rate": 0.0004690102605031722,
      "loss": 2.2759,
      "step": 71311
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7900580763816833,
      "learning_rate": 0.0004690068808388026,
      "loss": 2.1218,
      "step": 71312
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.8005067110061646,
      "learning_rate": 0.0004690035011430115,
      "loss": 2.3868,
      "step": 71313
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.8653135299682617,
      "learning_rate": 0.0004690001214157994,
      "loss": 2.1294,
      "step": 71314
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.8534291386604309,
      "learning_rate": 0.000468996741657167,
      "loss": 2.1656,
      "step": 71315
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7607982158660889,
      "learning_rate": 0.00046899336186711495,
      "loss": 2.3157,
      "step": 71316
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.8418039679527283,
      "learning_rate": 0.0004689899820456438,
      "loss": 2.2371,
      "step": 71317
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7364827990531921,
      "learning_rate": 0.0004689866021927542,
      "loss": 2.1993,
      "step": 71318
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.795422375202179,
      "learning_rate": 0.00046898322230844684,
      "loss": 2.1897,
      "step": 71319
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7764442563056946,
      "learning_rate": 0.0004689798423927224,
      "loss": 2.122,
      "step": 71320
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.8013537526130676,
      "learning_rate": 0.00046897646244558125,
      "loss": 2.3557,
      "step": 71321
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.6792142987251282,
      "learning_rate": 0.0004689730824670243,
      "loss": 2.1475,
      "step": 71322
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.8179757595062256,
      "learning_rate": 0.000468969702457052,
      "loss": 2.2958,
      "step": 71323
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.8192718625068665,
      "learning_rate": 0.000468966322415665,
      "loss": 2.66,
      "step": 71324
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.677454948425293,
      "learning_rate": 0.0004689629423428642,
      "loss": 2.0399,
      "step": 71325
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7611773610115051,
      "learning_rate": 0.00046895956223864976,
      "loss": 2.3915,
      "step": 71326
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.8716006875038147,
      "learning_rate": 0.0004689561821030226,
      "loss": 2.2418,
      "step": 71327
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7518754601478577,
      "learning_rate": 0.0004689528019359833,
      "loss": 2.1725,
      "step": 71328
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7771539092063904,
      "learning_rate": 0.00046894942173753256,
      "loss": 2.0837,
      "step": 71329
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7599908709526062,
      "learning_rate": 0.0004689460415076708,
      "loss": 1.9577,
      "step": 71330
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7186987400054932,
      "learning_rate": 0.0004689426612463989,
      "loss": 2.1631,
      "step": 71331
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7603714466094971,
      "learning_rate": 0.00046893928095371735,
      "loss": 2.1868,
      "step": 71332
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.8414944410324097,
      "learning_rate": 0.0004689359006296267,
      "loss": 2.4395,
      "step": 71333
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.73281329870224,
      "learning_rate": 0.00046893252027412785,
      "loss": 2.1163,
      "step": 71334
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.8051698803901672,
      "learning_rate": 0.0004689291398872211,
      "loss": 2.2244,
      "step": 71335
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.804395318031311,
      "learning_rate": 0.00046892575946890733,
      "loss": 2.3041,
      "step": 71336
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.6981444358825684,
      "learning_rate": 0.00046892237901918706,
      "loss": 2.0822,
      "step": 71337
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7949661612510681,
      "learning_rate": 0.0004689189985380609,
      "loss": 2.3637,
      "step": 71338
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.8016278743743896,
      "learning_rate": 0.00046891561802552955,
      "loss": 2.0845,
      "step": 71339
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7850341200828552,
      "learning_rate": 0.00046891223748159357,
      "loss": 2.2501,
      "step": 71340
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7415001392364502,
      "learning_rate": 0.00046890885690625356,
      "loss": 2.0627,
      "step": 71341
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7522761225700378,
      "learning_rate": 0.0004689054762995103,
      "loss": 2.1903,
      "step": 71342
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.8982945084571838,
      "learning_rate": 0.00046890209566136435,
      "loss": 2.3219,
      "step": 71343
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7442299127578735,
      "learning_rate": 0.00046889871499181617,
      "loss": 2.0951,
      "step": 71344
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7964350581169128,
      "learning_rate": 0.00046889533429086657,
      "loss": 2.194,
      "step": 71345
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7882542610168457,
      "learning_rate": 0.00046889195355851625,
      "loss": 2.1487,
      "step": 71346
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.8391005396842957,
      "learning_rate": 0.00046888857279476554,
      "loss": 2.1744,
      "step": 71347
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7341365218162537,
      "learning_rate": 0.0004688851919996155,
      "loss": 2.3107,
      "step": 71348
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.8361160159111023,
      "learning_rate": 0.0004688818111730663,
      "loss": 2.2086,
      "step": 71349
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7661839127540588,
      "learning_rate": 0.00046887843031511887,
      "loss": 2.2466,
      "step": 71350
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.8814265727996826,
      "learning_rate": 0.0004688750494257737,
      "loss": 2.3603,
      "step": 71351
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.8440382480621338,
      "learning_rate": 0.00046887166850503154,
      "loss": 2.3836,
      "step": 71352
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.8017206192016602,
      "learning_rate": 0.00046886828755289295,
      "loss": 2.1887,
      "step": 71353
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7046295404434204,
      "learning_rate": 0.0004688649065693586,
      "loss": 2.0619,
      "step": 71354
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7596986293792725,
      "learning_rate": 0.000468861525554429,
      "loss": 2.4607,
      "step": 71355
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7224808931350708,
      "learning_rate": 0.0004688581445081048,
      "loss": 2.2253,
      "step": 71356
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7246058583259583,
      "learning_rate": 0.00046885476343038677,
      "loss": 2.2955,
      "step": 71357
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.8797016143798828,
      "learning_rate": 0.00046885138232127555,
      "loss": 2.2858,
      "step": 71358
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7919893860816956,
      "learning_rate": 0.0004688480011807714,
      "loss": 2.151,
      "step": 71359
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7553334832191467,
      "learning_rate": 0.0004688446200088754,
      "loss": 2.0838,
      "step": 71360
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.8076423406600952,
      "learning_rate": 0.000468841238805588,
      "loss": 2.3607,
      "step": 71361
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.8345438241958618,
      "learning_rate": 0.0004688378575709098,
      "loss": 2.3513,
      "step": 71362
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.793883204460144,
      "learning_rate": 0.0004688344763048414,
      "loss": 2.5609,
      "step": 71363
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7942827939987183,
      "learning_rate": 0.0004688310950073836,
      "loss": 2.4276,
      "step": 71364
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.9728772044181824,
      "learning_rate": 0.00046882771367853684,
      "loss": 2.269,
      "step": 71365
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.8957083821296692,
      "learning_rate": 0.00046882433231830185,
      "loss": 2.4376,
      "step": 71366
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.8077300190925598,
      "learning_rate": 0.00046882095092667914,
      "loss": 2.3866,
      "step": 71367
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7712602615356445,
      "learning_rate": 0.00046881756950366955,
      "loss": 2.0952,
      "step": 71368
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7945038080215454,
      "learning_rate": 0.00046881418804927354,
      "loss": 2.2026,
      "step": 71369
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.6823163628578186,
      "learning_rate": 0.00046881080656349173,
      "loss": 2.169,
      "step": 71370
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.827774703502655,
      "learning_rate": 0.0004688074250463249,
      "loss": 2.3032,
      "step": 71371
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.851613461971283,
      "learning_rate": 0.00046880404349777355,
      "loss": 2.3011,
      "step": 71372
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.861405611038208,
      "learning_rate": 0.0004688006619178383,
      "loss": 2.0632,
      "step": 71373
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7681185007095337,
      "learning_rate": 0.0004687972803065199,
      "loss": 2.095,
      "step": 71374
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7241658568382263,
      "learning_rate": 0.0004687938986638188,
      "loss": 2.1036,
      "step": 71375
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7769985795021057,
      "learning_rate": 0.00046879051698973576,
      "loss": 2.4465,
      "step": 71376
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7802316546440125,
      "learning_rate": 0.00046878713528427144,
      "loss": 2.3248,
      "step": 71377
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.8221256136894226,
      "learning_rate": 0.00046878375354742635,
      "loss": 1.9766,
      "step": 71378
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.8273062109947205,
      "learning_rate": 0.0004687803717792012,
      "loss": 2.2955,
      "step": 71379
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7349770069122314,
      "learning_rate": 0.0004687769899795966,
      "loss": 2.3816,
      "step": 71380
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.9930105209350586,
      "learning_rate": 0.0004687736081486131,
      "loss": 2.0631,
      "step": 71381
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7814010977745056,
      "learning_rate": 0.0004687702262862515,
      "loss": 2.1271,
      "step": 71382
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.8140527606010437,
      "learning_rate": 0.00046876684439251226,
      "loss": 2.3332,
      "step": 71383
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7789846062660217,
      "learning_rate": 0.0004687634624673961,
      "loss": 2.3187,
      "step": 71384
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7441078424453735,
      "learning_rate": 0.0004687600805109036,
      "loss": 2.1981,
      "step": 71385
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7660828232765198,
      "learning_rate": 0.0004687566985230355,
      "loss": 2.1482,
      "step": 71386
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7555420398712158,
      "learning_rate": 0.0004687533165037922,
      "loss": 2.0518,
      "step": 71387
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.9157981872558594,
      "learning_rate": 0.0004687499344531746,
      "loss": 2.278,
      "step": 71388
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.8303346633911133,
      "learning_rate": 0.0004687465523711832,
      "loss": 2.1402,
      "step": 71389
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7721786499023438,
      "learning_rate": 0.00046874317025781857,
      "loss": 2.0232,
      "step": 71390
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7493132948875427,
      "learning_rate": 0.00046873978811308133,
      "loss": 2.2546,
      "step": 71391
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.8414438366889954,
      "learning_rate": 0.0004687364059369724,
      "loss": 2.2139,
      "step": 71392
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7839400172233582,
      "learning_rate": 0.00046873302372949194,
      "loss": 2.3944,
      "step": 71393
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7158721089363098,
      "learning_rate": 0.00046872964149064097,
      "loss": 1.9741,
      "step": 71394
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.8634436726570129,
      "learning_rate": 0.00046872625922042,
      "loss": 2.1025,
      "step": 71395
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7402158379554749,
      "learning_rate": 0.00046872287691882957,
      "loss": 2.2407,
      "step": 71396
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7895950078964233,
      "learning_rate": 0.00046871949458587034,
      "loss": 2.2378,
      "step": 71397
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7402541041374207,
      "learning_rate": 0.0004687161122215431,
      "loss": 2.1512,
      "step": 71398
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.9253246784210205,
      "learning_rate": 0.0004687127298258482,
      "loss": 2.1063,
      "step": 71399
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.8337324857711792,
      "learning_rate": 0.0004687093473987865,
      "loss": 2.3064,
      "step": 71400
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.6973569393157959,
      "learning_rate": 0.00046870596494035855,
      "loss": 2.2177,
      "step": 71401
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.6954430341720581,
      "learning_rate": 0.00046870258245056487,
      "loss": 2.1666,
      "step": 71402
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.0201212167739868,
      "learning_rate": 0.0004686991999294063,
      "loss": 2.4075,
      "step": 71403
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.8143169283866882,
      "learning_rate": 0.0004686958173768834,
      "loss": 2.4224,
      "step": 71404
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7412499189376831,
      "learning_rate": 0.00046869243479299663,
      "loss": 2.4063,
      "step": 71405
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7970673441886902,
      "learning_rate": 0.0004686890521777469,
      "loss": 2.0811,
      "step": 71406
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.8323497772216797,
      "learning_rate": 0.0004686856695311346,
      "loss": 2.1878,
      "step": 71407
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.787340521812439,
      "learning_rate": 0.00046868228685316047,
      "loss": 2.175,
      "step": 71408
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7957021594047546,
      "learning_rate": 0.0004686789041438251,
      "loss": 2.3704,
      "step": 71409
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7597581148147583,
      "learning_rate": 0.0004686755214031292,
      "loss": 2.4532,
      "step": 71410
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7658573389053345,
      "learning_rate": 0.00046867213863107326,
      "loss": 2.184,
      "step": 71411
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.6934779286384583,
      "learning_rate": 0.000468668755827658,
      "loss": 2.0264,
      "step": 71412
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7104917168617249,
      "learning_rate": 0.00046866537299288414,
      "loss": 2.1821,
      "step": 71413
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.747648298740387,
      "learning_rate": 0.0004686619901267521,
      "loss": 2.2737,
      "step": 71414
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.8767180442810059,
      "learning_rate": 0.00046865860722926264,
      "loss": 2.2389,
      "step": 71415
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7235977649688721,
      "learning_rate": 0.0004686552243004163,
      "loss": 2.0372,
      "step": 71416
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.8034447431564331,
      "learning_rate": 0.00046865184134021386,
      "loss": 2.3483,
      "step": 71417
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7632095813751221,
      "learning_rate": 0.0004686484583486558,
      "loss": 2.0779,
      "step": 71418
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.8622053265571594,
      "learning_rate": 0.0004686450753257428,
      "loss": 2.0794,
      "step": 71419
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7800474762916565,
      "learning_rate": 0.0004686416922714755,
      "loss": 2.222,
      "step": 71420
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7437958717346191,
      "learning_rate": 0.00046863830918585455,
      "loss": 1.9186,
      "step": 71421
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.8165695667266846,
      "learning_rate": 0.00046863492606888065,
      "loss": 2.4138,
      "step": 71422
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7945185303688049,
      "learning_rate": 0.00046863154292055417,
      "loss": 2.0428,
      "step": 71423
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7005875706672668,
      "learning_rate": 0.000468628159740876,
      "loss": 2.1233,
      "step": 71424
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7377711534500122,
      "learning_rate": 0.0004686247765298466,
      "loss": 2.2426,
      "step": 71425
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.8818010687828064,
      "learning_rate": 0.0004686213932874668,
      "loss": 2.5081,
      "step": 71426
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7588608860969543,
      "learning_rate": 0.00046861801001373693,
      "loss": 2.1029,
      "step": 71427
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7706587910652161,
      "learning_rate": 0.0004686146267086579,
      "loss": 2.129,
      "step": 71428
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7446700930595398,
      "learning_rate": 0.0004686112433722302,
      "loss": 2.2111,
      "step": 71429
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7872251272201538,
      "learning_rate": 0.0004686078600044545,
      "loss": 2.3226,
      "step": 71430
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.778211772441864,
      "learning_rate": 0.0004686044766053314,
      "loss": 2.0618,
      "step": 71431
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7495943903923035,
      "learning_rate": 0.00046860109317486157,
      "loss": 1.9266,
      "step": 71432
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.02700674533844,
      "learning_rate": 0.00046859770971304556,
      "loss": 2.0791,
      "step": 71433
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.8116741180419922,
      "learning_rate": 0.000468594326219884,
      "loss": 2.2303,
      "step": 71434
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7424188852310181,
      "learning_rate": 0.0004685909426953777,
      "loss": 2.1133,
      "step": 71435
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.8195914030075073,
      "learning_rate": 0.0004685875591395272,
      "loss": 2.0526,
      "step": 71436
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7792947292327881,
      "learning_rate": 0.000468584175552333,
      "loss": 2.2637,
      "step": 71437
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7494149208068848,
      "learning_rate": 0.00046858079193379585,
      "loss": 2.1926,
      "step": 71438
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7611608505249023,
      "learning_rate": 0.0004685774082839163,
      "loss": 2.3223,
      "step": 71439
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7213377952575684,
      "learning_rate": 0.0004685740246026951,
      "loss": 2.1209,
      "step": 71440
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7427468299865723,
      "learning_rate": 0.00046857064089013276,
      "loss": 2.0318,
      "step": 71441
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7083885669708252,
      "learning_rate": 0.00046856725714622996,
      "loss": 2.1566,
      "step": 71442
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7099714875221252,
      "learning_rate": 0.0004685638733709873,
      "loss": 2.1431,
      "step": 71443
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.771088719367981,
      "learning_rate": 0.0004685604895644055,
      "loss": 2.0251,
      "step": 71444
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7183086276054382,
      "learning_rate": 0.00046855710572648506,
      "loss": 2.1336,
      "step": 71445
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7691479921340942,
      "learning_rate": 0.0004685537218572267,
      "loss": 2.3035,
      "step": 71446
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.6801543831825256,
      "learning_rate": 0.0004685503379566311,
      "loss": 2.2052,
      "step": 71447
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7306683659553528,
      "learning_rate": 0.0004685469540246987,
      "loss": 2.1844,
      "step": 71448
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7843918204307556,
      "learning_rate": 0.0004685435700614302,
      "loss": 2.2993,
      "step": 71449
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.8010631799697876,
      "learning_rate": 0.00046854018606682637,
      "loss": 2.1406,
      "step": 71450
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7118043899536133,
      "learning_rate": 0.00046853680204088767,
      "loss": 2.0857,
      "step": 71451
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.8471498489379883,
      "learning_rate": 0.00046853341798361477,
      "loss": 2.212,
      "step": 71452
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7589181065559387,
      "learning_rate": 0.00046853003389500853,
      "loss": 2.0133,
      "step": 71453
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7811664342880249,
      "learning_rate": 0.0004685266497750692,
      "loss": 2.17,
      "step": 71454
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.8391671180725098,
      "learning_rate": 0.0004685232656237976,
      "loss": 2.1279,
      "step": 71455
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.8366714715957642,
      "learning_rate": 0.00046851988144119437,
      "loss": 2.1241,
      "step": 71456
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.8063310384750366,
      "learning_rate": 0.00046851649722726006,
      "loss": 2.1459,
      "step": 71457
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.0043871402740479,
      "learning_rate": 0.0004685131129819954,
      "loss": 2.1871,
      "step": 71458
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.6806452870368958,
      "learning_rate": 0.0004685097287054009,
      "loss": 1.9555,
      "step": 71459
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7435575723648071,
      "learning_rate": 0.00046850634439747743,
      "loss": 2.1499,
      "step": 71460
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.803361713886261,
      "learning_rate": 0.00046850296005822534,
      "loss": 2.1833,
      "step": 71461
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7540663480758667,
      "learning_rate": 0.00046849957568764533,
      "loss": 2.1474,
      "step": 71462
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.6809694766998291,
      "learning_rate": 0.00046849619128573817,
      "loss": 2.0472,
      "step": 71463
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.889470100402832,
      "learning_rate": 0.0004684928068525043,
      "loss": 2.2828,
      "step": 71464
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.772139847278595,
      "learning_rate": 0.0004684894223879445,
      "loss": 2.2435,
      "step": 71465
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7800536155700684,
      "learning_rate": 0.00046848603789205923,
      "loss": 2.0821,
      "step": 71466
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.8354290127754211,
      "learning_rate": 0.00046848265336484936,
      "loss": 2.137,
      "step": 71467
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7047271728515625,
      "learning_rate": 0.0004684792688063153,
      "loss": 1.9872,
      "step": 71468
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.8931406140327454,
      "learning_rate": 0.0004684758842164578,
      "loss": 2.3311,
      "step": 71469
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.8343872427940369,
      "learning_rate": 0.0004684724995952774,
      "loss": 2.3697,
      "step": 71470
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7362826466560364,
      "learning_rate": 0.00046846911494277487,
      "loss": 2.0794,
      "step": 71471
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.0133715867996216,
      "learning_rate": 0.00046846573025895074,
      "loss": 2.2701,
      "step": 71472
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.8312990069389343,
      "learning_rate": 0.0004684623455438055,
      "loss": 2.1656,
      "step": 71473
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7349703311920166,
      "learning_rate": 0.0004684589607973401,
      "loss": 1.9643,
      "step": 71474
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.0318570137023926,
      "learning_rate": 0.000468455576019555,
      "loss": 2.1998,
      "step": 71475
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.8215383291244507,
      "learning_rate": 0.0004684521912104507,
      "loss": 2.2607,
      "step": 71476
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7170220017433167,
      "learning_rate": 0.00046844880637002804,
      "loss": 2.0266,
      "step": 71477
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.1180609464645386,
      "learning_rate": 0.00046844542149828756,
      "loss": 2.3364,
      "step": 71478
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.8637303113937378,
      "learning_rate": 0.0004684420365952299,
      "loss": 2.3229,
      "step": 71479
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7831091284751892,
      "learning_rate": 0.0004684386516608557,
      "loss": 2.2339,
      "step": 71480
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7656325101852417,
      "learning_rate": 0.0004684352666951656,
      "loss": 2.3717,
      "step": 71481
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.9574880003929138,
      "learning_rate": 0.00046843188169816007,
      "loss": 2.2566,
      "step": 71482
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.9262301325798035,
      "learning_rate": 0.00046842849666984003,
      "loss": 2.4393,
      "step": 71483
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.8241358399391174,
      "learning_rate": 0.00046842511161020585,
      "loss": 2.0861,
      "step": 71484
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.9372434616088867,
      "learning_rate": 0.0004684217265192583,
      "loss": 2.1481,
      "step": 71485
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.8265842199325562,
      "learning_rate": 0.00046841834139699803,
      "loss": 2.1803,
      "step": 71486
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7932595014572144,
      "learning_rate": 0.00046841495624342554,
      "loss": 2.1638,
      "step": 71487
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.1276319026947021,
      "learning_rate": 0.0004684115710585415,
      "loss": 2.5195,
      "step": 71488
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.9787246584892273,
      "learning_rate": 0.0004684081858423467,
      "loss": 2.1784,
      "step": 71489
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.047851800918579,
      "learning_rate": 0.0004684048005948415,
      "loss": 2.2194,
      "step": 71490
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.8746970295906067,
      "learning_rate": 0.0004684014153160267,
      "loss": 2.0485,
      "step": 71491
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.8028860092163086,
      "learning_rate": 0.00046839803000590295,
      "loss": 2.3517,
      "step": 71492
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.774776816368103,
      "learning_rate": 0.00046839464466447075,
      "loss": 2.3823,
      "step": 71493
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.8975134491920471,
      "learning_rate": 0.0004683912592917309,
      "loss": 2.1043,
      "step": 71494
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7856522798538208,
      "learning_rate": 0.0004683878738876839,
      "loss": 2.1709,
      "step": 71495
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.8750699758529663,
      "learning_rate": 0.0004683844884523304,
      "loss": 2.4227,
      "step": 71496
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7899414896965027,
      "learning_rate": 0.000468381102985671,
      "loss": 2.2319,
      "step": 71497
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.8413918614387512,
      "learning_rate": 0.0004683777174877064,
      "loss": 2.356,
      "step": 71498
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7573030591011047,
      "learning_rate": 0.00046837433195843727,
      "loss": 2.1922,
      "step": 71499
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.8101093769073486,
      "learning_rate": 0.0004683709463978641,
      "loss": 2.2908,
      "step": 71500
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.8708736896514893,
      "learning_rate": 0.0004683675608059876,
      "loss": 2.372,
      "step": 71501
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.8103799223899841,
      "learning_rate": 0.0004683641751828085,
      "loss": 2.2683,
      "step": 71502
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7290313839912415,
      "learning_rate": 0.0004683607895283272,
      "loss": 2.1809,
      "step": 71503
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7865581512451172,
      "learning_rate": 0.0004683574038425445,
      "loss": 2.1588,
      "step": 71504
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7514193654060364,
      "learning_rate": 0.000468354018125461,
      "loss": 2.3096,
      "step": 71505
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7172655463218689,
      "learning_rate": 0.0004683506323770772,
      "loss": 2.2107,
      "step": 71506
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7946435809135437,
      "learning_rate": 0.00046834724659739396,
      "loss": 2.1358,
      "step": 71507
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7573495507240295,
      "learning_rate": 0.0004683438607864118,
      "loss": 2.2892,
      "step": 71508
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7768111228942871,
      "learning_rate": 0.0004683404749441313,
      "loss": 2.2099,
      "step": 71509
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7725288271903992,
      "learning_rate": 0.0004683370890705531,
      "loss": 2.3181,
      "step": 71510
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.763019859790802,
      "learning_rate": 0.00046833370316567784,
      "loss": 2.2924,
      "step": 71511
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7763168215751648,
      "learning_rate": 0.0004683303172295062,
      "loss": 2.2753,
      "step": 71512
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.8284302353858948,
      "learning_rate": 0.0004683269312620388,
      "loss": 2.1247,
      "step": 71513
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7368448972702026,
      "learning_rate": 0.00046832354526327624,
      "loss": 2.0093,
      "step": 71514
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7873483896255493,
      "learning_rate": 0.0004683201592332191,
      "loss": 2.3771,
      "step": 71515
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7869746088981628,
      "learning_rate": 0.0004683167731718681,
      "loss": 2.361,
      "step": 71516
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.8061820864677429,
      "learning_rate": 0.0004683133870792239,
      "loss": 2.2034,
      "step": 71517
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.799656867980957,
      "learning_rate": 0.00046831000095528695,
      "loss": 2.0914,
      "step": 71518
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.741986095905304,
      "learning_rate": 0.00046830661480005807,
      "loss": 2.2859,
      "step": 71519
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7219934463500977,
      "learning_rate": 0.0004683032286135378,
      "loss": 2.2167,
      "step": 71520
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7884129881858826,
      "learning_rate": 0.0004682998423957267,
      "loss": 2.1894,
      "step": 71521
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7547692060470581,
      "learning_rate": 0.00046829645614662556,
      "loss": 2.3422,
      "step": 71522
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7853090167045593,
      "learning_rate": 0.000468293069866235,
      "loss": 2.3157,
      "step": 71523
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.8178554773330688,
      "learning_rate": 0.0004682896835545555,
      "loss": 2.0447,
      "step": 71524
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7604816555976868,
      "learning_rate": 0.00046828629721158775,
      "loss": 2.1545,
      "step": 71525
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7804399728775024,
      "learning_rate": 0.00046828291083733246,
      "loss": 2.1649,
      "step": 71526
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.8530039191246033,
      "learning_rate": 0.0004682795244317902,
      "loss": 2.1241,
      "step": 71527
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.8382623791694641,
      "learning_rate": 0.0004682761379949615,
      "loss": 2.4915,
      "step": 71528
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7586754560470581,
      "learning_rate": 0.0004682727515268472,
      "loss": 1.9694,
      "step": 71529
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.8450652956962585,
      "learning_rate": 0.0004682693650274477,
      "loss": 2.0717,
      "step": 71530
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7290602922439575,
      "learning_rate": 0.00046826597849676386,
      "loss": 2.2359,
      "step": 71531
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.8332824110984802,
      "learning_rate": 0.0004682625919347962,
      "loss": 2.1261,
      "step": 71532
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.9545650482177734,
      "learning_rate": 0.0004682592053415453,
      "loss": 2.0222,
      "step": 71533
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7165572047233582,
      "learning_rate": 0.0004682558187170118,
      "loss": 2.2811,
      "step": 71534
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.8833076357841492,
      "learning_rate": 0.0004682524320611965,
      "loss": 2.1579,
      "step": 71535
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7567291259765625,
      "learning_rate": 0.00046824904537409983,
      "loss": 2.178,
      "step": 71536
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.8186320066452026,
      "learning_rate": 0.00046824565865572247,
      "loss": 2.4266,
      "step": 71537
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.8209232687950134,
      "learning_rate": 0.0004682422719060651,
      "loss": 2.4364,
      "step": 71538
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7837088108062744,
      "learning_rate": 0.00046823888512512825,
      "loss": 2.336,
      "step": 71539
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.8735043406486511,
      "learning_rate": 0.0004682354983129126,
      "loss": 2.1298,
      "step": 71540
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7200120687484741,
      "learning_rate": 0.00046823211146941894,
      "loss": 2.2184,
      "step": 71541
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.795360267162323,
      "learning_rate": 0.00046822872459464763,
      "loss": 2.1648,
      "step": 71542
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.8133584260940552,
      "learning_rate": 0.0004682253376885995,
      "loss": 2.3788,
      "step": 71543
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7308308482170105,
      "learning_rate": 0.000468221950751275,
      "loss": 2.3645,
      "step": 71544
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7138098478317261,
      "learning_rate": 0.0004682185637826749,
      "loss": 2.2349,
      "step": 71545
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7965709567070007,
      "learning_rate": 0.0004682151767827998,
      "loss": 2.3008,
      "step": 71546
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7210817933082581,
      "learning_rate": 0.00046821178975165037,
      "loss": 2.2875,
      "step": 71547
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7864030599594116,
      "learning_rate": 0.0004682084026892272,
      "loss": 2.1692,
      "step": 71548
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7428321838378906,
      "learning_rate": 0.00046820501559553086,
      "loss": 2.2915,
      "step": 71549
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7022908329963684,
      "learning_rate": 0.000468201628470562,
      "loss": 2.171,
      "step": 71550
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7884138226509094,
      "learning_rate": 0.00046819824131432135,
      "loss": 2.113,
      "step": 71551
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.705548882484436,
      "learning_rate": 0.00046819485412680944,
      "loss": 2.249,
      "step": 71552
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7490715980529785,
      "learning_rate": 0.00046819146690802696,
      "loss": 2.2685,
      "step": 71553
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.8914731740951538,
      "learning_rate": 0.00046818807965797447,
      "loss": 2.2719,
      "step": 71554
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7744921445846558,
      "learning_rate": 0.00046818469237665266,
      "loss": 1.8675,
      "step": 71555
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7384083867073059,
      "learning_rate": 0.00046818130506406225,
      "loss": 2.075,
      "step": 71556
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.750129222869873,
      "learning_rate": 0.0004681779177202036,
      "loss": 2.2447,
      "step": 71557
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7644795179367065,
      "learning_rate": 0.0004681745303450775,
      "loss": 2.1621,
      "step": 71558
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7824902534484863,
      "learning_rate": 0.00046817114293868473,
      "loss": 2.2717,
      "step": 71559
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7579840421676636,
      "learning_rate": 0.00046816775550102564,
      "loss": 2.2105,
      "step": 71560
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.8368246555328369,
      "learning_rate": 0.000468164368032101,
      "loss": 2.2011,
      "step": 71561
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7592470049858093,
      "learning_rate": 0.0004681609805319115,
      "loss": 2.5326,
      "step": 71562
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.830182671546936,
      "learning_rate": 0.00046815759300045763,
      "loss": 2.0445,
      "step": 71563
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.6974921822547913,
      "learning_rate": 0.0004681542054377401,
      "loss": 2.2017,
      "step": 71564
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.1941503286361694,
      "learning_rate": 0.00046815081784375953,
      "loss": 2.1543,
      "step": 71565
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7317429184913635,
      "learning_rate": 0.0004681474302185166,
      "loss": 2.1588,
      "step": 71566
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7571085095405579,
      "learning_rate": 0.0004681440425620119,
      "loss": 2.1745,
      "step": 71567
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7791802883148193,
      "learning_rate": 0.00046814065487424595,
      "loss": 2.1363,
      "step": 71568
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7528131604194641,
      "learning_rate": 0.00046813726715521953,
      "loss": 2.1943,
      "step": 71569
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.8090932369232178,
      "learning_rate": 0.0004681338794049332,
      "loss": 2.1298,
      "step": 71570
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.801400899887085,
      "learning_rate": 0.00046813049162338765,
      "loss": 2.2771,
      "step": 71571
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7780868411064148,
      "learning_rate": 0.0004681271038105835,
      "loss": 2.1455,
      "step": 71572
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.8085888624191284,
      "learning_rate": 0.00046812371596652124,
      "loss": 2.2632,
      "step": 71573
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7499849796295166,
      "learning_rate": 0.00046812032809120165,
      "loss": 2.2229,
      "step": 71574
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7250544428825378,
      "learning_rate": 0.00046811694018462533,
      "loss": 2.2814,
      "step": 71575
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.8140048980712891,
      "learning_rate": 0.0004681135522467929,
      "loss": 2.2883,
      "step": 71576
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.8028789758682251,
      "learning_rate": 0.0004681101642777049,
      "loss": 2.321,
      "step": 71577
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7620598673820496,
      "learning_rate": 0.00046810677627736225,
      "loss": 2.1578,
      "step": 71578
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7834042310714722,
      "learning_rate": 0.0004681033882457653,
      "loss": 2.4309,
      "step": 71579
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7800390124320984,
      "learning_rate": 0.0004681000001829146,
      "loss": 2.3487,
      "step": 71580
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7930076718330383,
      "learning_rate": 0.0004680966120888111,
      "loss": 2.0636,
      "step": 71581
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7574266791343689,
      "learning_rate": 0.00046809322396345514,
      "loss": 2.1967,
      "step": 71582
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7344955205917358,
      "learning_rate": 0.0004680898358068476,
      "loss": 2.0692,
      "step": 71583
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7721828818321228,
      "learning_rate": 0.00046808644761898896,
      "loss": 2.1947,
      "step": 71584
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.8532025814056396,
      "learning_rate": 0.00046808305939987987,
      "loss": 2.3014,
      "step": 71585
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7491832971572876,
      "learning_rate": 0.000468079671149521,
      "loss": 2.2332,
      "step": 71586
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7805386185646057,
      "learning_rate": 0.0004680762828679129,
      "loss": 2.0943,
      "step": 71587
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7485958933830261,
      "learning_rate": 0.0004680728945550562,
      "loss": 2.0956,
      "step": 71588
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.8658948540687561,
      "learning_rate": 0.00046806950621095157,
      "loss": 2.3088,
      "step": 71589
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.8791990280151367,
      "learning_rate": 0.00046806611783559984,
      "loss": 2.1793,
      "step": 71590
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7039190530776978,
      "learning_rate": 0.0004680627294290012,
      "loss": 2.0251,
      "step": 71591
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7993247509002686,
      "learning_rate": 0.0004680593409911566,
      "loss": 2.089,
      "step": 71592
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.806171715259552,
      "learning_rate": 0.00046805595252206664,
      "loss": 2.2917,
      "step": 71593
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.832271933555603,
      "learning_rate": 0.0004680525640217319,
      "loss": 2.5004,
      "step": 71594
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7821388840675354,
      "learning_rate": 0.000468049175490153,
      "loss": 2.3688,
      "step": 71595
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.8073497414588928,
      "learning_rate": 0.00046804578692733064,
      "loss": 2.1226,
      "step": 71596
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.8561311960220337,
      "learning_rate": 0.0004680423983332653,
      "loss": 1.9555,
      "step": 71597
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.8144240975379944,
      "learning_rate": 0.0004680390097079578,
      "loss": 2.3044,
      "step": 71598
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.0959498882293701,
      "learning_rate": 0.00046803562105140865,
      "loss": 1.8966,
      "step": 71599
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.8092470169067383,
      "learning_rate": 0.0004680322323636185,
      "loss": 2.1952,
      "step": 71600
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.8400818705558777,
      "learning_rate": 0.00046802884364458794,
      "loss": 2.2156,
      "step": 71601
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.8647432327270508,
      "learning_rate": 0.0004680254548943177,
      "loss": 2.3144,
      "step": 71602
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.0989516973495483,
      "learning_rate": 0.0004680220661128083,
      "loss": 2.21,
      "step": 71603
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7384896874427795,
      "learning_rate": 0.00046801867730006044,
      "loss": 2.4239,
      "step": 71604
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7864940762519836,
      "learning_rate": 0.00046801528845607485,
      "loss": 2.2807,
      "step": 71605
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.195389986038208,
      "learning_rate": 0.0004680118995808519,
      "loss": 2.2737,
      "step": 71606
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.8387326598167419,
      "learning_rate": 0.0004680085106743924,
      "loss": 2.3314,
      "step": 71607
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.8222270011901855,
      "learning_rate": 0.000468005121736697,
      "loss": 2.2276,
      "step": 71608
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.9076449871063232,
      "learning_rate": 0.00046800173276776623,
      "loss": 2.3314,
      "step": 71609
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.8238511681556702,
      "learning_rate": 0.00046799834376760077,
      "loss": 2.2961,
      "step": 71610
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.9215655326843262,
      "learning_rate": 0.00046799495473620134,
      "loss": 2.0445,
      "step": 71611
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.8306818008422852,
      "learning_rate": 0.0004679915656735684,
      "loss": 2.1503,
      "step": 71612
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.818976104259491,
      "learning_rate": 0.00046798817657970265,
      "loss": 2.0876,
      "step": 71613
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.0748268365859985,
      "learning_rate": 0.00046798478745460473,
      "loss": 2.5721,
      "step": 71614
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.8306731581687927,
      "learning_rate": 0.0004679813982982753,
      "loss": 2.2515,
      "step": 71615
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.9399040937423706,
      "learning_rate": 0.0004679780091107148,
      "loss": 2.2199,
      "step": 71616
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.9342244267463684,
      "learning_rate": 0.00046797461989192427,
      "loss": 2.3217,
      "step": 71617
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7988407611846924,
      "learning_rate": 0.0004679712306419039,
      "loss": 2.167,
      "step": 71618
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7932528257369995,
      "learning_rate": 0.0004679678413606545,
      "loss": 2.3297,
      "step": 71619
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.9065296649932861,
      "learning_rate": 0.0004679644520481769,
      "loss": 2.1518,
      "step": 71620
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.87052983045578,
      "learning_rate": 0.0004679610627044713,
      "loss": 2.311,
      "step": 71621
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7666473388671875,
      "learning_rate": 0.00046795767332953874,
      "loss": 2.2377,
      "step": 71622
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7721459865570068,
      "learning_rate": 0.0004679542839233796,
      "loss": 1.9595,
      "step": 71623
    },
    {
      "epoch": 0.93,
      "grad_norm": 3.34024715423584,
      "learning_rate": 0.0004679508944859947,
      "loss": 2.1993,
      "step": 71624
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.8305349946022034,
      "learning_rate": 0.0004679475050173843,
      "loss": 2.2646,
      "step": 71625
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.8657656311988831,
      "learning_rate": 0.00046794411551754955,
      "loss": 2.0903,
      "step": 71626
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.8058380484580994,
      "learning_rate": 0.00046794072598649073,
      "loss": 2.3016,
      "step": 71627
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.8008561134338379,
      "learning_rate": 0.0004679373364242085,
      "loss": 2.2895,
      "step": 71628
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.9000049233436584,
      "learning_rate": 0.00046793394683070365,
      "loss": 2.2466,
      "step": 71629
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.8256575465202332,
      "learning_rate": 0.0004679305572059766,
      "loss": 2.4317,
      "step": 71630
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.787591278553009,
      "learning_rate": 0.00046792716755002814,
      "loss": 2.2803,
      "step": 71631
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7755110859870911,
      "learning_rate": 0.00046792377786285884,
      "loss": 2.3995,
      "step": 71632
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.8120427131652832,
      "learning_rate": 0.0004679203881444695,
      "loss": 2.2354,
      "step": 71633
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7843742370605469,
      "learning_rate": 0.00046791699839486035,
      "loss": 2.3965,
      "step": 71634
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.886824905872345,
      "learning_rate": 0.0004679136086140323,
      "loss": 2.2354,
      "step": 71635
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.8982212543487549,
      "learning_rate": 0.00046791021880198606,
      "loss": 2.0698,
      "step": 71636
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7462490200996399,
      "learning_rate": 0.0004679068289587221,
      "loss": 2.1471,
      "step": 71637
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7636144757270813,
      "learning_rate": 0.00046790343908424106,
      "loss": 2.2526,
      "step": 71638
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7010171413421631,
      "learning_rate": 0.0004679000491785436,
      "loss": 2.0973,
      "step": 71639
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7819061875343323,
      "learning_rate": 0.0004678966592416304,
      "loss": 2.5583,
      "step": 71640
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.8283064365386963,
      "learning_rate": 0.000467893269273502,
      "loss": 2.2135,
      "step": 71641
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7736629843711853,
      "learning_rate": 0.00046788987927415905,
      "loss": 2.3563,
      "step": 71642
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7887083888053894,
      "learning_rate": 0.0004678864892436022,
      "loss": 2.0957,
      "step": 71643
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.8410782814025879,
      "learning_rate": 0.0004678830991818322,
      "loss": 2.2954,
      "step": 71644
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7545764446258545,
      "learning_rate": 0.00046787970908884947,
      "loss": 2.2075,
      "step": 71645
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7622350454330444,
      "learning_rate": 0.00046787631896465465,
      "loss": 2.1856,
      "step": 71646
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.87237548828125,
      "learning_rate": 0.00046787292880924857,
      "loss": 2.2723,
      "step": 71647
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.828238308429718,
      "learning_rate": 0.00046786953862263173,
      "loss": 2.116,
      "step": 71648
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7745656967163086,
      "learning_rate": 0.0004678661484048047,
      "loss": 2.0535,
      "step": 71649
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.813327968120575,
      "learning_rate": 0.0004678627581557683,
      "loss": 2.3205,
      "step": 71650
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.9118219614028931,
      "learning_rate": 0.000467859367875523,
      "loss": 2.4313,
      "step": 71651
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.845445454120636,
      "learning_rate": 0.00046785597756406945,
      "loss": 2.4917,
      "step": 71652
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7748544812202454,
      "learning_rate": 0.0004678525872214083,
      "loss": 2.4393,
      "step": 71653
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.014771819114685,
      "learning_rate": 0.0004678491968475403,
      "loss": 2.101,
      "step": 71654
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.9163860082626343,
      "learning_rate": 0.00046784580644246575,
      "loss": 2.4655,
      "step": 71655
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7572588324546814,
      "learning_rate": 0.00046784241600618566,
      "loss": 2.2468,
      "step": 71656
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.8148126602172852,
      "learning_rate": 0.0004678390255387005,
      "loss": 2.394,
      "step": 71657
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.944205105304718,
      "learning_rate": 0.0004678356350400108,
      "loss": 2.1216,
      "step": 71658
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.8064517974853516,
      "learning_rate": 0.0004678322445101173,
      "loss": 2.3307,
      "step": 71659
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.8033072352409363,
      "learning_rate": 0.00046782885394902076,
      "loss": 2.1454,
      "step": 71660
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.8126694560050964,
      "learning_rate": 0.00046782546335672154,
      "loss": 2.1161,
      "step": 71661
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7428858876228333,
      "learning_rate": 0.0004678220727332204,
      "loss": 2.3465,
      "step": 71662
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7079795598983765,
      "learning_rate": 0.00046781868207851807,
      "loss": 2.322,
      "step": 71663
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.8038743138313293,
      "learning_rate": 0.00046781529139261496,
      "loss": 2.3014,
      "step": 71664
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.0122286081314087,
      "learning_rate": 0.0004678119006755118,
      "loss": 2.1719,
      "step": 71665
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.70711749792099,
      "learning_rate": 0.0004678085099272093,
      "loss": 2.2369,
      "step": 71666
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7703897356987,
      "learning_rate": 0.00046780511914770804,
      "loss": 2.2057,
      "step": 71667
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.85163813829422,
      "learning_rate": 0.0004678017283370086,
      "loss": 2.2141,
      "step": 71668
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.8397945761680603,
      "learning_rate": 0.00046779833749511173,
      "loss": 2.2572,
      "step": 71669
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.874588131904602,
      "learning_rate": 0.0004677949466220179,
      "loss": 2.2584,
      "step": 71670
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.799948513507843,
      "learning_rate": 0.0004677915557177279,
      "loss": 2.1059,
      "step": 71671
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.8265801668167114,
      "learning_rate": 0.00046778816478224223,
      "loss": 1.914,
      "step": 71672
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7732238173484802,
      "learning_rate": 0.0004677847738155616,
      "loss": 2.1054,
      "step": 71673
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.8298696279525757,
      "learning_rate": 0.0004677813828176866,
      "loss": 2.4353,
      "step": 71674
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7684812545776367,
      "learning_rate": 0.00046777799178861787,
      "loss": 1.969,
      "step": 71675
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.8045904636383057,
      "learning_rate": 0.000467774600728356,
      "loss": 2.2958,
      "step": 71676
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7661129832267761,
      "learning_rate": 0.0004677712096369017,
      "loss": 2.3046,
      "step": 71677
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.9465979933738708,
      "learning_rate": 0.0004677678185142556,
      "loss": 2.3603,
      "step": 71678
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.8213505148887634,
      "learning_rate": 0.0004677644273604183,
      "loss": 2.4656,
      "step": 71679
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7745507955551147,
      "learning_rate": 0.0004677610361753903,
      "loss": 2.1731,
      "step": 71680
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.8660316467285156,
      "learning_rate": 0.00046775764495917245,
      "loss": 2.2317,
      "step": 71681
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.9611712694168091,
      "learning_rate": 0.00046775425371176527,
      "loss": 2.4033,
      "step": 71682
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7902995944023132,
      "learning_rate": 0.0004677508624331695,
      "loss": 2.1475,
      "step": 71683
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7505288124084473,
      "learning_rate": 0.00046774747112338555,
      "loss": 2.2199,
      "step": 71684
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.8802068829536438,
      "learning_rate": 0.0004677440797824143,
      "loss": 2.1573,
      "step": 71685
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7840805053710938,
      "learning_rate": 0.0004677406884102561,
      "loss": 2.395,
      "step": 71686
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7995684742927551,
      "learning_rate": 0.0004677372970069119,
      "loss": 2.2619,
      "step": 71687
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.8140729069709778,
      "learning_rate": 0.0004677339055723821,
      "loss": 2.2739,
      "step": 71688
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.9032279849052429,
      "learning_rate": 0.0004677305141066674,
      "loss": 2.3306,
      "step": 71689
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7558199167251587,
      "learning_rate": 0.0004677271226097684,
      "loss": 2.1715,
      "step": 71690
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7650465965270996,
      "learning_rate": 0.0004677237310816858,
      "loss": 2.3356,
      "step": 71691
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.8562520742416382,
      "learning_rate": 0.0004677203395224202,
      "loss": 2.2604,
      "step": 71692
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7567102909088135,
      "learning_rate": 0.00046771694793197225,
      "loss": 2.0834,
      "step": 71693
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7614030838012695,
      "learning_rate": 0.0004677135563103425,
      "loss": 2.3953,
      "step": 71694
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.8350732326507568,
      "learning_rate": 0.0004677101646575316,
      "loss": 2.2544,
      "step": 71695
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.1285817623138428,
      "learning_rate": 0.0004677067729735403,
      "loss": 2.1661,
      "step": 71696
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7516634464263916,
      "learning_rate": 0.0004677033812583691,
      "loss": 2.3183,
      "step": 71697
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.9831699132919312,
      "learning_rate": 0.0004676999895120187,
      "loss": 2.4618,
      "step": 71698
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7984312176704407,
      "learning_rate": 0.0004676965977344896,
      "loss": 2.2007,
      "step": 71699
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.8523868322372437,
      "learning_rate": 0.00046769320592578265,
      "loss": 2.0357,
      "step": 71700
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.8429837226867676,
      "learning_rate": 0.0004676898140858984,
      "loss": 2.2023,
      "step": 71701
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.8620724678039551,
      "learning_rate": 0.00046768642221483735,
      "loss": 2.2123,
      "step": 71702
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.821111798286438,
      "learning_rate": 0.00046768303031260034,
      "loss": 2.3836,
      "step": 71703
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7518002986907959,
      "learning_rate": 0.00046767963837918776,
      "loss": 2.1785,
      "step": 71704
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7642397284507751,
      "learning_rate": 0.00046767624641460046,
      "loss": 2.1871,
      "step": 71705
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.8128980398178101,
      "learning_rate": 0.000467672854418839,
      "loss": 1.9661,
      "step": 71706
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7408574819564819,
      "learning_rate": 0.0004676694623919039,
      "loss": 2.0643,
      "step": 71707
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7712984085083008,
      "learning_rate": 0.0004676660703337959,
      "loss": 2.1621,
      "step": 71708
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.857663631439209,
      "learning_rate": 0.00046766267824451565,
      "loss": 2.3785,
      "step": 71709
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.8282198309898376,
      "learning_rate": 0.0004676592861240637,
      "loss": 2.0466,
      "step": 71710
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7701511979103088,
      "learning_rate": 0.0004676558939724407,
      "loss": 2.1541,
      "step": 71711
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.8769676089286804,
      "learning_rate": 0.00046765250178964747,
      "loss": 2.0387,
      "step": 71712
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7287076711654663,
      "learning_rate": 0.00046764910957568435,
      "loss": 2.1304,
      "step": 71713
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7194246053695679,
      "learning_rate": 0.0004676457173305521,
      "loss": 2.2067,
      "step": 71714
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7526473999023438,
      "learning_rate": 0.0004676423250542514,
      "loss": 2.4095,
      "step": 71715
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.6853054165840149,
      "learning_rate": 0.00046763893274678274,
      "loss": 2.1416,
      "step": 71716
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7906163930892944,
      "learning_rate": 0.0004676355404081469,
      "loss": 2.1694,
      "step": 71717
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.8438618183135986,
      "learning_rate": 0.00046763214803834444,
      "loss": 2.3185,
      "step": 71718
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.6960172057151794,
      "learning_rate": 0.00046762875563737596,
      "loss": 2.1546,
      "step": 71719
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.6997750401496887,
      "learning_rate": 0.00046762536320524214,
      "loss": 2.1552,
      "step": 71720
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.776139497756958,
      "learning_rate": 0.00046762197074194363,
      "loss": 2.1825,
      "step": 71721
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7799777388572693,
      "learning_rate": 0.00046761857824748103,
      "loss": 2.3355,
      "step": 71722
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7319552302360535,
      "learning_rate": 0.00046761518572185494,
      "loss": 2.1627,
      "step": 71723
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.8777107000350952,
      "learning_rate": 0.00046761179316506606,
      "loss": 2.2875,
      "step": 71724
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7621031999588013,
      "learning_rate": 0.00046760840057711504,
      "loss": 2.1712,
      "step": 71725
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7361437082290649,
      "learning_rate": 0.0004676050079580023,
      "loss": 2.2861,
      "step": 71726
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7993205785751343,
      "learning_rate": 0.0004676016153077287,
      "loss": 2.3132,
      "step": 71727
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7728797197341919,
      "learning_rate": 0.00046759822262629484,
      "loss": 2.0596,
      "step": 71728
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.870116114616394,
      "learning_rate": 0.0004675948299137013,
      "loss": 2.258,
      "step": 71729
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.744687557220459,
      "learning_rate": 0.0004675914371699487,
      "loss": 2.2246,
      "step": 71730
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.8495084047317505,
      "learning_rate": 0.00046758804439503757,
      "loss": 2.1579,
      "step": 71731
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.9045710563659668,
      "learning_rate": 0.0004675846515889688,
      "loss": 1.902,
      "step": 71732
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.8165369033813477,
      "learning_rate": 0.00046758125875174286,
      "loss": 2.3413,
      "step": 71733
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.9634406566619873,
      "learning_rate": 0.0004675778658833604,
      "loss": 2.0802,
      "step": 71734
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.9559938311576843,
      "learning_rate": 0.0004675744729838221,
      "loss": 2.1997,
      "step": 71735
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.0256024599075317,
      "learning_rate": 0.0004675710800531285,
      "loss": 2.2181,
      "step": 71736
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.9670503735542297,
      "learning_rate": 0.0004675676870912802,
      "loss": 2.2243,
      "step": 71737
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7063900232315063,
      "learning_rate": 0.0004675642940982779,
      "loss": 2.3067,
      "step": 71738
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.9327768683433533,
      "learning_rate": 0.0004675609010741224,
      "loss": 2.1723,
      "step": 71739
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.2050113677978516,
      "learning_rate": 0.00046755750801881404,
      "loss": 2.2016,
      "step": 71740
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.8124053478240967,
      "learning_rate": 0.00046755411493235353,
      "loss": 2.0442,
      "step": 71741
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7829471230506897,
      "learning_rate": 0.00046755072181474167,
      "loss": 2.3218,
      "step": 71742
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.9455634951591492,
      "learning_rate": 0.0004675473286659789,
      "loss": 2.1581,
      "step": 71743
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.2418804168701172,
      "learning_rate": 0.00046754393548606584,
      "loss": 2.3537,
      "step": 71744
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.8884868025779724,
      "learning_rate": 0.0004675405422750034,
      "loss": 2.19,
      "step": 71745
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.8061901330947876,
      "learning_rate": 0.0004675371490327918,
      "loss": 2.0844,
      "step": 71746
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.0450043678283691,
      "learning_rate": 0.0004675337557594321,
      "loss": 2.3239,
      "step": 71747
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.9771271347999573,
      "learning_rate": 0.00046753036245492454,
      "loss": 2.1028,
      "step": 71748
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.8170304894447327,
      "learning_rate": 0.00046752696911927,
      "loss": 2.4089,
      "step": 71749
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.8436574339866638,
      "learning_rate": 0.00046752357575246894,
      "loss": 2.2323,
      "step": 71750
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.0272804498672485,
      "learning_rate": 0.00046752018235452224,
      "loss": 2.1939,
      "step": 71751
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.921146810054779,
      "learning_rate": 0.0004675167889254303,
      "loss": 2.3627,
      "step": 71752
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7548428773880005,
      "learning_rate": 0.00046751339546519376,
      "loss": 2.1,
      "step": 71753
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.9754748344421387,
      "learning_rate": 0.00046751000197381345,
      "loss": 2.0525,
      "step": 71754
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.1948332786560059,
      "learning_rate": 0.00046750660845128975,
      "loss": 2.2403,
      "step": 71755
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.8268151879310608,
      "learning_rate": 0.0004675032148976235,
      "loss": 2.2389,
      "step": 71756
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.8064892888069153,
      "learning_rate": 0.0004674998213128152,
      "loss": 2.2727,
      "step": 71757
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.00876784324646,
      "learning_rate": 0.0004674964276968655,
      "loss": 2.0863,
      "step": 71758
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.8172284960746765,
      "learning_rate": 0.00046749303404977497,
      "loss": 2.0097,
      "step": 71759
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7474684119224548,
      "learning_rate": 0.00046748964037154455,
      "loss": 1.9546,
      "step": 71760
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.8479885458946228,
      "learning_rate": 0.0004674862466621745,
      "loss": 2.181,
      "step": 71761
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.8216657638549805,
      "learning_rate": 0.0004674828529216656,
      "loss": 2.3123,
      "step": 71762
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.8988009095191956,
      "learning_rate": 0.0004674794591500185,
      "loss": 2.1775,
      "step": 71763
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.8018584847450256,
      "learning_rate": 0.0004674760653472338,
      "loss": 2.234,
      "step": 71764
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.741158664226532,
      "learning_rate": 0.00046747267151331216,
      "loss": 2.1949,
      "step": 71765
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7904891967773438,
      "learning_rate": 0.0004674692776482541,
      "loss": 2.0141,
      "step": 71766
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.8538476228713989,
      "learning_rate": 0.0004674658837520604,
      "loss": 2.1631,
      "step": 71767
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.8457136154174805,
      "learning_rate": 0.00046746248982473164,
      "loss": 2.2723,
      "step": 71768
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.744679868221283,
      "learning_rate": 0.0004674590958662684,
      "loss": 2.3753,
      "step": 71769
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7874074578285217,
      "learning_rate": 0.00046745570187667145,
      "loss": 2.2337,
      "step": 71770
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7945775389671326,
      "learning_rate": 0.00046745230785594124,
      "loss": 2.1712,
      "step": 71771
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.9694791436195374,
      "learning_rate": 0.0004674489138040785,
      "loss": 2.0362,
      "step": 71772
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7801581025123596,
      "learning_rate": 0.00046744551972108386,
      "loss": 2.1697,
      "step": 71773
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.8373099565505981,
      "learning_rate": 0.00046744212560695793,
      "loss": 2.2617,
      "step": 71774
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.9145022034645081,
      "learning_rate": 0.00046743873146170135,
      "loss": 2.3158,
      "step": 71775
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.8285848498344421,
      "learning_rate": 0.00046743533728531474,
      "loss": 2.2841,
      "step": 71776
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7881010174751282,
      "learning_rate": 0.0004674319430777988,
      "loss": 2.2275,
      "step": 71777
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7563806176185608,
      "learning_rate": 0.00046742854883915396,
      "loss": 2.3045,
      "step": 71778
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.9667901396751404,
      "learning_rate": 0.0004674251545693811,
      "loss": 2.2628,
      "step": 71779
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.8168855905532837,
      "learning_rate": 0.00046742176026848075,
      "loss": 2.4483,
      "step": 71780
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7193483710289001,
      "learning_rate": 0.0004674183659364535,
      "loss": 1.9564,
      "step": 71781
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7558185458183289,
      "learning_rate": 0.0004674149715733001,
      "loss": 2.395,
      "step": 71782
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.88435959815979,
      "learning_rate": 0.000467411577179021,
      "loss": 2.279,
      "step": 71783
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.8826323747634888,
      "learning_rate": 0.00046740818275361687,
      "loss": 1.9659,
      "step": 71784
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7670092582702637,
      "learning_rate": 0.00046740478829708856,
      "loss": 2.2363,
      "step": 71785
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.8503881692886353,
      "learning_rate": 0.00046740139380943644,
      "loss": 2.3958,
      "step": 71786
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7285361886024475,
      "learning_rate": 0.00046739799929066116,
      "loss": 2.3581,
      "step": 71787
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7046100497245789,
      "learning_rate": 0.0004673946047407636,
      "loss": 2.1917,
      "step": 71788
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7530684471130371,
      "learning_rate": 0.0004673912101597441,
      "loss": 2.0034,
      "step": 71789
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7766501307487488,
      "learning_rate": 0.00046738781554760356,
      "loss": 2.1271,
      "step": 71790
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7754218578338623,
      "learning_rate": 0.0004673844209043424,
      "loss": 2.394,
      "step": 71791
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7780370116233826,
      "learning_rate": 0.0004673810262299612,
      "loss": 2.035,
      "step": 71792
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.9602246284484863,
      "learning_rate": 0.0004673776315244608,
      "loss": 2.4085,
      "step": 71793
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.8753728270530701,
      "learning_rate": 0.00046737423678784183,
      "loss": 2.1466,
      "step": 71794
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.8089385628700256,
      "learning_rate": 0.0004673708420201047,
      "loss": 2.1797,
      "step": 71795
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.8253023624420166,
      "learning_rate": 0.0004673674472212501,
      "loss": 2.3903,
      "step": 71796
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.8488733172416687,
      "learning_rate": 0.000467364052391279,
      "loss": 2.1153,
      "step": 71797
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.8109673857688904,
      "learning_rate": 0.00046736065753019153,
      "loss": 2.2077,
      "step": 71798
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7823424339294434,
      "learning_rate": 0.00046735726263798856,
      "loss": 2.4105,
      "step": 71799
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.8135861754417419,
      "learning_rate": 0.0004673538677146709,
      "loss": 2.1583,
      "step": 71800
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.781368613243103,
      "learning_rate": 0.0004673504727602388,
      "loss": 2.223,
      "step": 71801
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7324497699737549,
      "learning_rate": 0.00046734707777469316,
      "loss": 2.1373,
      "step": 71802
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.8434904217720032,
      "learning_rate": 0.00046734368275803457,
      "loss": 2.3372,
      "step": 71803
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7931864857673645,
      "learning_rate": 0.00046734028771026366,
      "loss": 2.2874,
      "step": 71804
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.8255857229232788,
      "learning_rate": 0.000467336892631381,
      "loss": 2.294,
      "step": 71805
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7699365615844727,
      "learning_rate": 0.0004673334975213872,
      "loss": 2.4389,
      "step": 71806
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7319189310073853,
      "learning_rate": 0.000467330102380283,
      "loss": 2.3203,
      "step": 71807
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.8507566452026367,
      "learning_rate": 0.0004673267072080689,
      "loss": 2.215,
      "step": 71808
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.9386863112449646,
      "learning_rate": 0.0004673233120047457,
      "loss": 2.2436,
      "step": 71809
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7032318711280823,
      "learning_rate": 0.0004673199167703139,
      "loss": 2.0557,
      "step": 71810
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.8327575325965881,
      "learning_rate": 0.00046731652150477416,
      "loss": 2.2846,
      "step": 71811
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.8323601484298706,
      "learning_rate": 0.00046731312620812714,
      "loss": 2.1367,
      "step": 71812
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.0582771301269531,
      "learning_rate": 0.0004673097308803734,
      "loss": 2.2872,
      "step": 71813
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7937459945678711,
      "learning_rate": 0.0004673063355215137,
      "loss": 2.1743,
      "step": 71814
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.8203102350234985,
      "learning_rate": 0.0004673029401315485,
      "loss": 2.5014,
      "step": 71815
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.8487750887870789,
      "learning_rate": 0.0004672995447104786,
      "loss": 2.0925,
      "step": 71816
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.8251446485519409,
      "learning_rate": 0.0004672961492583045,
      "loss": 2.2467,
      "step": 71817
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.8026120066642761,
      "learning_rate": 0.000467292753775027,
      "loss": 2.2255,
      "step": 71818
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7779019474983215,
      "learning_rate": 0.00046728935826064654,
      "loss": 2.2121,
      "step": 71819
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.9498162269592285,
      "learning_rate": 0.0004672859627151638,
      "loss": 2.3138,
      "step": 71820
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7780360579490662,
      "learning_rate": 0.0004672825671385795,
      "loss": 2.2274,
      "step": 71821
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7965132594108582,
      "learning_rate": 0.0004672791715308942,
      "loss": 2.2566,
      "step": 71822
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.0000823736190796,
      "learning_rate": 0.00046727577589210844,
      "loss": 2.1096,
      "step": 71823
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7813273072242737,
      "learning_rate": 0.0004672723802222232,
      "loss": 2.0976,
      "step": 71824
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7760330438613892,
      "learning_rate": 0.00046726898452123864,
      "loss": 2.2668,
      "step": 71825
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.9573095440864563,
      "learning_rate": 0.00046726558878915563,
      "loss": 2.3105,
      "step": 71826
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7186667919158936,
      "learning_rate": 0.00046726219302597495,
      "loss": 2.2715,
      "step": 71827
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.8089287281036377,
      "learning_rate": 0.00046725879723169696,
      "loss": 2.072,
      "step": 71828
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.9215585589408875,
      "learning_rate": 0.00046725540140632243,
      "loss": 1.9485,
      "step": 71829
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7979348301887512,
      "learning_rate": 0.0004672520055498519,
      "loss": 2.3812,
      "step": 71830
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7607870697975159,
      "learning_rate": 0.0004672486096622862,
      "loss": 1.9334,
      "step": 71831
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7488767504692078,
      "learning_rate": 0.0004672452137436257,
      "loss": 2.3652,
      "step": 71832
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7636176943778992,
      "learning_rate": 0.00046724181779387116,
      "loss": 1.9825,
      "step": 71833
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7578780055046082,
      "learning_rate": 0.0004672384218130233,
      "loss": 2.1831,
      "step": 71834
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7823296785354614,
      "learning_rate": 0.0004672350258010826,
      "loss": 2.1022,
      "step": 71835
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.8028362989425659,
      "learning_rate": 0.0004672316297580497,
      "loss": 1.9593,
      "step": 71836
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7878254055976868,
      "learning_rate": 0.0004672282336839254,
      "loss": 2.0341,
      "step": 71837
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7985113263130188,
      "learning_rate": 0.0004672248375787102,
      "loss": 1.962,
      "step": 71838
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7317934036254883,
      "learning_rate": 0.0004672214414424047,
      "loss": 1.9953,
      "step": 71839
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7873829007148743,
      "learning_rate": 0.0004672180452750096,
      "loss": 2.2901,
      "step": 71840
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7303838729858398,
      "learning_rate": 0.00046721464907652547,
      "loss": 2.2892,
      "step": 71841
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7252658009529114,
      "learning_rate": 0.000467211252846953,
      "loss": 2.2427,
      "step": 71842
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7569735646247864,
      "learning_rate": 0.0004672078565862928,
      "loss": 2.0744,
      "step": 71843
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.8567796349525452,
      "learning_rate": 0.0004672044602945455,
      "loss": 2.1751,
      "step": 71844
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.8037216067314148,
      "learning_rate": 0.0004672010639717117,
      "loss": 2.0829,
      "step": 71845
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.815464198589325,
      "learning_rate": 0.00046719766761779214,
      "loss": 2.3611,
      "step": 71846
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.8514477610588074,
      "learning_rate": 0.0004671942712327873,
      "loss": 2.2173,
      "step": 71847
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7747989296913147,
      "learning_rate": 0.000467190874816698,
      "loss": 2.121,
      "step": 71848
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.8106781244277954,
      "learning_rate": 0.0004671874783695247,
      "loss": 2.1469,
      "step": 71849
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.8568592667579651,
      "learning_rate": 0.0004671840818912681,
      "loss": 2.2835,
      "step": 71850
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7388262152671814,
      "learning_rate": 0.00046718068538192877,
      "loss": 2.2042,
      "step": 71851
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7879296541213989,
      "learning_rate": 0.0004671772888415075,
      "loss": 2.2729,
      "step": 71852
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.8006449341773987,
      "learning_rate": 0.0004671738922700047,
      "loss": 2.1701,
      "step": 71853
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.8260143399238586,
      "learning_rate": 0.00046717049566742116,
      "loss": 2.2364,
      "step": 71854
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.8137547373771667,
      "learning_rate": 0.0004671670990337576,
      "loss": 2.1931,
      "step": 71855
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.79349684715271,
      "learning_rate": 0.0004671637023690143,
      "loss": 2.2773,
      "step": 71856
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7985597252845764,
      "learning_rate": 0.0004671603056731922,
      "loss": 2.2654,
      "step": 71857
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7945178151130676,
      "learning_rate": 0.00046715690894629186,
      "loss": 2.2187,
      "step": 71858
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.8166669011116028,
      "learning_rate": 0.00046715351218831394,
      "loss": 2.3086,
      "step": 71859
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7983983159065247,
      "learning_rate": 0.00046715011539925894,
      "loss": 2.2624,
      "step": 71860
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7783203721046448,
      "learning_rate": 0.00046714671857912767,
      "loss": 2.2233,
      "step": 71861
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.77888023853302,
      "learning_rate": 0.0004671433217279206,
      "loss": 2.3949,
      "step": 71862
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.6756548285484314,
      "learning_rate": 0.00046713992484563843,
      "loss": 1.9146,
      "step": 71863
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.8115372657775879,
      "learning_rate": 0.0004671365279322818,
      "loss": 2.2617,
      "step": 71864
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7473791241645813,
      "learning_rate": 0.0004671331309878513,
      "loss": 2.2953,
      "step": 71865
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7390280961990356,
      "learning_rate": 0.0004671297340123477,
      "loss": 2.1721,
      "step": 71866
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.8140924572944641,
      "learning_rate": 0.0004671263370057715,
      "loss": 2.1982,
      "step": 71867
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7705750465393066,
      "learning_rate": 0.0004671229399681233,
      "loss": 2.1468,
      "step": 71868
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.6938009262084961,
      "learning_rate": 0.00046711954289940376,
      "loss": 2.1743,
      "step": 71869
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7673405408859253,
      "learning_rate": 0.00046711614579961365,
      "loss": 2.2186,
      "step": 71870
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7637261748313904,
      "learning_rate": 0.0004671127486687534,
      "loss": 1.9651,
      "step": 71871
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.8272339701652527,
      "learning_rate": 0.00046710935150682376,
      "loss": 2.1485,
      "step": 71872
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7610731720924377,
      "learning_rate": 0.0004671059543138254,
      "loss": 2.3362,
      "step": 71873
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7054708003997803,
      "learning_rate": 0.0004671025570897589,
      "loss": 2.5157,
      "step": 71874
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7776033282279968,
      "learning_rate": 0.0004670991598346248,
      "loss": 2.1272,
      "step": 71875
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.9254629015922546,
      "learning_rate": 0.00046709576254842387,
      "loss": 2.3167,
      "step": 71876
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.775637686252594,
      "learning_rate": 0.00046709236523115663,
      "loss": 2.0667,
      "step": 71877
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7754209041595459,
      "learning_rate": 0.00046708896788282374,
      "loss": 2.1398,
      "step": 71878
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.8462409973144531,
      "learning_rate": 0.0004670855705034259,
      "loss": 2.2144,
      "step": 71879
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.8678182363510132,
      "learning_rate": 0.0004670821730929638,
      "loss": 2.1967,
      "step": 71880
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7852115631103516,
      "learning_rate": 0.00046707877565143783,
      "loss": 2.2305,
      "step": 71881
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.8322168588638306,
      "learning_rate": 0.0004670753781788488,
      "loss": 2.35,
      "step": 71882
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7591764330863953,
      "learning_rate": 0.00046707198067519735,
      "loss": 2.3323,
      "step": 71883
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.8333176970481873,
      "learning_rate": 0.000467068583140484,
      "loss": 2.0967,
      "step": 71884
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.9177088141441345,
      "learning_rate": 0.0004670651855747094,
      "loss": 2.261,
      "step": 71885
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.8008513450622559,
      "learning_rate": 0.00046706178797787436,
      "loss": 2.0843,
      "step": 71886
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.8871145844459534,
      "learning_rate": 0.00046705839034997935,
      "loss": 2.2875,
      "step": 71887
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.0146797895431519,
      "learning_rate": 0.00046705499269102503,
      "loss": 2.2562,
      "step": 71888
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7486401200294495,
      "learning_rate": 0.00046705159500101196,
      "loss": 2.3907,
      "step": 71889
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.782493531703949,
      "learning_rate": 0.00046704819727994083,
      "loss": 2.2871,
      "step": 71890
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.8532956838607788,
      "learning_rate": 0.0004670447995278124,
      "loss": 2.2833,
      "step": 71891
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.8684868812561035,
      "learning_rate": 0.0004670414017446271,
      "loss": 2.312,
      "step": 71892
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.81772381067276,
      "learning_rate": 0.00046703800393038566,
      "loss": 2.2395,
      "step": 71893
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7996523976325989,
      "learning_rate": 0.0004670346060850888,
      "loss": 2.4722,
      "step": 71894
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7478366494178772,
      "learning_rate": 0.00046703120820873695,
      "loss": 2.3641,
      "step": 71895
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.8092031478881836,
      "learning_rate": 0.0004670278103013308,
      "loss": 2.4495,
      "step": 71896
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.8241941332817078,
      "learning_rate": 0.00046702441236287114,
      "loss": 2.1676,
      "step": 71897
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.8683944344520569,
      "learning_rate": 0.00046702101439335844,
      "loss": 2.4249,
      "step": 71898
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.2082452774047852,
      "learning_rate": 0.0004670176163927934,
      "loss": 2.1452,
      "step": 71899
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7758563160896301,
      "learning_rate": 0.00046701421836117666,
      "loss": 2.2028,
      "step": 71900
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7584396004676819,
      "learning_rate": 0.0004670108202985088,
      "loss": 2.301,
      "step": 71901
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7694752812385559,
      "learning_rate": 0.00046700742220479047,
      "loss": 2.1128,
      "step": 71902
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.8487582802772522,
      "learning_rate": 0.00046700402408002224,
      "loss": 2.4291,
      "step": 71903
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.70778888463974,
      "learning_rate": 0.00046700062592420496,
      "loss": 2.2088,
      "step": 71904
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.847683310508728,
      "learning_rate": 0.000466997227737339,
      "loss": 2.4821,
      "step": 71905
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.8845902681350708,
      "learning_rate": 0.00046699382951942507,
      "loss": 2.1243,
      "step": 71906
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.9294347763061523,
      "learning_rate": 0.0004669904312704639,
      "loss": 2.2913,
      "step": 71907
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.8440319299697876,
      "learning_rate": 0.0004669870329904561,
      "loss": 2.4405,
      "step": 71908
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.816938579082489,
      "learning_rate": 0.0004669836346794022,
      "loss": 2.2605,
      "step": 71909
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.877535343170166,
      "learning_rate": 0.0004669802363373029,
      "loss": 2.1753,
      "step": 71910
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.8327579498291016,
      "learning_rate": 0.0004669768379641588,
      "loss": 2.2762,
      "step": 71911
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7426567077636719,
      "learning_rate": 0.0004669734395599706,
      "loss": 2.094,
      "step": 71912
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.8657522797584534,
      "learning_rate": 0.00046697004112473885,
      "loss": 2.2407,
      "step": 71913
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7731558680534363,
      "learning_rate": 0.0004669666426584642,
      "loss": 2.2251,
      "step": 71914
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.691034734249115,
      "learning_rate": 0.0004669632441611473,
      "loss": 2.1445,
      "step": 71915
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7883918881416321,
      "learning_rate": 0.0004669598456327888,
      "loss": 2.1895,
      "step": 71916
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.8197320699691772,
      "learning_rate": 0.0004669564470733893,
      "loss": 2.301,
      "step": 71917
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.84846431016922,
      "learning_rate": 0.0004669530484829495,
      "loss": 2.1352,
      "step": 71918
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.8523412346839905,
      "learning_rate": 0.00046694964986147,
      "loss": 2.4928,
      "step": 71919
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7410371899604797,
      "learning_rate": 0.0004669462512089513,
      "loss": 2.1546,
      "step": 71920
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.8605723977088928,
      "learning_rate": 0.00046694285252539417,
      "loss": 2.2938,
      "step": 71921
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7147129774093628,
      "learning_rate": 0.0004669394538107993,
      "loss": 2.3365,
      "step": 71922
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7730159759521484,
      "learning_rate": 0.00046693605506516715,
      "loss": 2.1699,
      "step": 71923
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.8339976072311401,
      "learning_rate": 0.0004669326562884984,
      "loss": 2.2799,
      "step": 71924
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.8636652827262878,
      "learning_rate": 0.00046692925748079384,
      "loss": 2.0856,
      "step": 71925
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7092137932777405,
      "learning_rate": 0.00046692585864205387,
      "loss": 2.0494,
      "step": 71926
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.8600935339927673,
      "learning_rate": 0.00046692245977227926,
      "loss": 2.2936,
      "step": 71927
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.0038597583770752,
      "learning_rate": 0.00046691906087147067,
      "loss": 2.1067,
      "step": 71928
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7402514219284058,
      "learning_rate": 0.00046691566193962864,
      "loss": 2.1809,
      "step": 71929
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.8378032445907593,
      "learning_rate": 0.00046691226297675375,
      "loss": 1.9775,
      "step": 71930
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.8511340022087097,
      "learning_rate": 0.0004669088639828469,
      "loss": 2.4287,
      "step": 71931
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.8209501504898071,
      "learning_rate": 0.00046690546495790843,
      "loss": 2.2209,
      "step": 71932
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7417203783988953,
      "learning_rate": 0.00046690206590193907,
      "loss": 1.924,
      "step": 71933
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7101030349731445,
      "learning_rate": 0.00046689866681493946,
      "loss": 2.1254,
      "step": 71934
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.9897347688674927,
      "learning_rate": 0.0004668952676969103,
      "loss": 2.4272,
      "step": 71935
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.8244406580924988,
      "learning_rate": 0.00046689186854785206,
      "loss": 2.1349,
      "step": 71936
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7709000110626221,
      "learning_rate": 0.0004668884693677656,
      "loss": 2.1999,
      "step": 71937
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.9660876393318176,
      "learning_rate": 0.00046688507015665143,
      "loss": 2.5892,
      "step": 71938
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.8876495361328125,
      "learning_rate": 0.00046688167091451003,
      "loss": 2.1296,
      "step": 71939
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7539030313491821,
      "learning_rate": 0.0004668782716413423,
      "loss": 2.2161,
      "step": 71940
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.8659428358078003,
      "learning_rate": 0.0004668748723371487,
      "loss": 2.1992,
      "step": 71941
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.8160476088523865,
      "learning_rate": 0.00046687147300192994,
      "loss": 2.0524,
      "step": 71942
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.8078938126564026,
      "learning_rate": 0.0004668680736356866,
      "loss": 2.0933,
      "step": 71943
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.811034619808197,
      "learning_rate": 0.00046686467423841927,
      "loss": 2.3663,
      "step": 71944
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7786861062049866,
      "learning_rate": 0.00046686127481012876,
      "loss": 2.0779,
      "step": 71945
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.8186500072479248,
      "learning_rate": 0.00046685787535081553,
      "loss": 2.3515,
      "step": 71946
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.8603577017784119,
      "learning_rate": 0.0004668544758604803,
      "loss": 2.1606,
      "step": 71947
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7146895527839661,
      "learning_rate": 0.0004668510763391236,
      "loss": 2.4588,
      "step": 71948
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7882389426231384,
      "learning_rate": 0.0004668476767867462,
      "loss": 2.3002,
      "step": 71949
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.757770836353302,
      "learning_rate": 0.00046684427720334866,
      "loss": 2.1516,
      "step": 71950
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.809533417224884,
      "learning_rate": 0.00046684087758893164,
      "loss": 1.9225,
      "step": 71951
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7566590905189514,
      "learning_rate": 0.00046683747794349566,
      "loss": 2.4745,
      "step": 71952
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.765859067440033,
      "learning_rate": 0.00046683407826704157,
      "loss": 2.0828,
      "step": 71953
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.8748462200164795,
      "learning_rate": 0.00046683067855956975,
      "loss": 2.2807,
      "step": 71954
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.813930332660675,
      "learning_rate": 0.00046682727882108103,
      "loss": 2.4099,
      "step": 71955
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7784478664398193,
      "learning_rate": 0.00046682387905157605,
      "loss": 2.1882,
      "step": 71956
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.9573509693145752,
      "learning_rate": 0.0004668204792510552,
      "loss": 2.2864,
      "step": 71957
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7397013306617737,
      "learning_rate": 0.0004668170794195194,
      "loss": 2.2531,
      "step": 71958
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.834029495716095,
      "learning_rate": 0.0004668136795569691,
      "loss": 2.3185,
      "step": 71959
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7645977139472961,
      "learning_rate": 0.000466810279663405,
      "loss": 2.2165,
      "step": 71960
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.8544465899467468,
      "learning_rate": 0.00046680687973882774,
      "loss": 2.2143,
      "step": 71961
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.8053680062294006,
      "learning_rate": 0.0004668034797832379,
      "loss": 2.1736,
      "step": 71962
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7834388613700867,
      "learning_rate": 0.00046680007979663606,
      "loss": 2.0985,
      "step": 71963
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.8651960492134094,
      "learning_rate": 0.000466796679779023,
      "loss": 2.3068,
      "step": 71964
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7510945796966553,
      "learning_rate": 0.00046679327973039945,
      "loss": 2.2296,
      "step": 71965
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.9120705723762512,
      "learning_rate": 0.00046678987965076575,
      "loss": 2.637,
      "step": 71966
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.8205751180648804,
      "learning_rate": 0.0004667864795401226,
      "loss": 2.3118,
      "step": 71967
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.8062088489532471,
      "learning_rate": 0.0004667830793984708,
      "loss": 2.2582,
      "step": 71968
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.2076975107192993,
      "learning_rate": 0.00046677967922581075,
      "loss": 2.2782,
      "step": 71969
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7493335008621216,
      "learning_rate": 0.00046677627902214335,
      "loss": 2.1938,
      "step": 71970
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7214088439941406,
      "learning_rate": 0.0004667728787874691,
      "loss": 2.0998,
      "step": 71971
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.9565911293029785,
      "learning_rate": 0.0004667694785217885,
      "loss": 2.3694,
      "step": 71972
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.8428055644035339,
      "learning_rate": 0.00046676607822510236,
      "loss": 2.3568,
      "step": 71973
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7266731858253479,
      "learning_rate": 0.0004667626778974113,
      "loss": 2.2178,
      "step": 71974
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7963375449180603,
      "learning_rate": 0.0004667592775387158,
      "loss": 2.0533,
      "step": 71975
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.9104965329170227,
      "learning_rate": 0.0004667558771490167,
      "loss": 2.1138,
      "step": 71976
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.769568681716919,
      "learning_rate": 0.00046675247672831456,
      "loss": 2.2671,
      "step": 71977
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.8254085779190063,
      "learning_rate": 0.0004667490762766099,
      "loss": 2.1994,
      "step": 71978
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.8369233012199402,
      "learning_rate": 0.00046674567579390344,
      "loss": 1.9182,
      "step": 71979
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7511044144630432,
      "learning_rate": 0.0004667422752801959,
      "loss": 2.237,
      "step": 71980
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.8597446084022522,
      "learning_rate": 0.00046673887473548773,
      "loss": 2.2438,
      "step": 71981
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.8564079999923706,
      "learning_rate": 0.00046673547415977966,
      "loss": 2.0269,
      "step": 71982
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.8260499835014343,
      "learning_rate": 0.0004667320735530724,
      "loss": 2.0332,
      "step": 71983
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7670230865478516,
      "learning_rate": 0.00046672867291536646,
      "loss": 2.4039,
      "step": 71984
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7618542909622192,
      "learning_rate": 0.00046672527224666247,
      "loss": 2.3194,
      "step": 71985
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.8458260893821716,
      "learning_rate": 0.00046672187154696125,
      "loss": 1.9991,
      "step": 71986
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.8294704556465149,
      "learning_rate": 0.00046671847081626317,
      "loss": 2.1184,
      "step": 71987
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7673788666725159,
      "learning_rate": 0.0004667150700545689,
      "loss": 2.3753,
      "step": 71988
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.8225612640380859,
      "learning_rate": 0.0004667116692618793,
      "loss": 1.9987,
      "step": 71989
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7717183828353882,
      "learning_rate": 0.0004667082684381948,
      "loss": 2.2984,
      "step": 71990
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.8337627649307251,
      "learning_rate": 0.00046670486758351607,
      "loss": 2.4463,
      "step": 71991
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.8460944890975952,
      "learning_rate": 0.0004667014666978437,
      "loss": 2.1259,
      "step": 71992
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.811260461807251,
      "learning_rate": 0.0004666980657811785,
      "loss": 2.1121,
      "step": 71993
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7901143431663513,
      "learning_rate": 0.0004666946648335209,
      "loss": 1.9451,
      "step": 71994
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.8040000796318054,
      "learning_rate": 0.0004666912638548717,
      "loss": 2.3509,
      "step": 71995
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7473398447036743,
      "learning_rate": 0.00046668786284523127,
      "loss": 2.0755,
      "step": 71996
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.9420056939125061,
      "learning_rate": 0.0004666844618046006,
      "loss": 2.3239,
      "step": 71997
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7223827838897705,
      "learning_rate": 0.0004666810607329801,
      "loss": 2.0656,
      "step": 71998
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.8315812945365906,
      "learning_rate": 0.00046667765963037043,
      "loss": 2.3208,
      "step": 71999
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.8768757581710815,
      "learning_rate": 0.00046667425849677226,
      "loss": 2.0831,
      "step": 72000
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7833520174026489,
      "learning_rate": 0.0004666708573321862,
      "loss": 2.2822,
      "step": 72001
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.826012909412384,
      "learning_rate": 0.0004666674561366128,
      "loss": 2.152,
      "step": 72002
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.795173704624176,
      "learning_rate": 0.0004666640549100528,
      "loss": 1.99,
      "step": 72003
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.8003137707710266,
      "learning_rate": 0.0004666606536525069,
      "loss": 2.2056,
      "step": 72004
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7791997194290161,
      "learning_rate": 0.00046665725236397557,
      "loss": 2.1954,
      "step": 72005
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.8905639052391052,
      "learning_rate": 0.0004666538510444595,
      "loss": 2.1743,
      "step": 72006
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7517468333244324,
      "learning_rate": 0.0004666504496939594,
      "loss": 2.0942,
      "step": 72007
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.8437157869338989,
      "learning_rate": 0.0004666470483124758,
      "loss": 2.24,
      "step": 72008
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.8037564754486084,
      "learning_rate": 0.0004666436469000093,
      "loss": 2.3123,
      "step": 72009
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.9279863238334656,
      "learning_rate": 0.00046664024545656067,
      "loss": 2.3521,
      "step": 72010
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.8227156400680542,
      "learning_rate": 0.0004666368439821304,
      "loss": 2.2788,
      "step": 72011
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.8364116549491882,
      "learning_rate": 0.00046663344247671926,
      "loss": 2.1997,
      "step": 72012
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.8519566655158997,
      "learning_rate": 0.00046663004094032784,
      "loss": 2.2968,
      "step": 72013
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7876026630401611,
      "learning_rate": 0.00046662663937295675,
      "loss": 2.2145,
      "step": 72014
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.756513237953186,
      "learning_rate": 0.00046662323777460653,
      "loss": 2.1641,
      "step": 72015
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.918544352054596,
      "learning_rate": 0.00046661983614527806,
      "loss": 2.0392,
      "step": 72016
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7562775015830994,
      "learning_rate": 0.0004666164344849717,
      "loss": 2.0159,
      "step": 72017
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7813524007797241,
      "learning_rate": 0.0004666130327936882,
      "loss": 2.0396,
      "step": 72018
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7210078239440918,
      "learning_rate": 0.00046660963107142823,
      "loss": 2.2381,
      "step": 72019
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7822444438934326,
      "learning_rate": 0.00046660622931819236,
      "loss": 2.0421,
      "step": 72020
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.8110262751579285,
      "learning_rate": 0.0004666028275339812,
      "loss": 2.3984,
      "step": 72021
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7356932759284973,
      "learning_rate": 0.00046659942571879555,
      "loss": 2.1481,
      "step": 72022
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.8174280524253845,
      "learning_rate": 0.00046659602387263586,
      "loss": 2.2036,
      "step": 72023
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.825844407081604,
      "learning_rate": 0.0004665926219955028,
      "loss": 2.3241,
      "step": 72024
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7986912131309509,
      "learning_rate": 0.0004665892200873971,
      "loss": 2.2232,
      "step": 72025
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7921238541603088,
      "learning_rate": 0.00046658581814831927,
      "loss": 2.3621,
      "step": 72026
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.8039537668228149,
      "learning_rate": 0.00046658241617827,
      "loss": 2.2927,
      "step": 72027
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7671025991439819,
      "learning_rate": 0.0004665790141772499,
      "loss": 2.286,
      "step": 72028
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.8853602409362793,
      "learning_rate": 0.00046657561214525966,
      "loss": 2.2402,
      "step": 72029
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7262201905250549,
      "learning_rate": 0.00046657221008229976,
      "loss": 2.1485,
      "step": 72030
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.8218325972557068,
      "learning_rate": 0.000466568807988371,
      "loss": 2.3496,
      "step": 72031
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.0676134824752808,
      "learning_rate": 0.000466565405863474,
      "loss": 2.1272,
      "step": 72032
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7744558453559875,
      "learning_rate": 0.0004665620037076093,
      "loss": 2.2358,
      "step": 72033
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.8883844614028931,
      "learning_rate": 0.0004665586015207776,
      "loss": 2.1871,
      "step": 72034
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.8040479421615601,
      "learning_rate": 0.00046655519930297956,
      "loss": 2.2341,
      "step": 72035
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.8095608353614807,
      "learning_rate": 0.00046655179705421565,
      "loss": 2.2082,
      "step": 72036
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.8422357439994812,
      "learning_rate": 0.0004665483947744866,
      "loss": 2.1087,
      "step": 72037
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.810019314289093,
      "learning_rate": 0.00046654499246379317,
      "loss": 2.0901,
      "step": 72038
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7670142650604248,
      "learning_rate": 0.0004665415901221359,
      "loss": 2.2862,
      "step": 72039
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7846511602401733,
      "learning_rate": 0.00046653818774951533,
      "loss": 2.1394,
      "step": 72040
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7313905358314514,
      "learning_rate": 0.0004665347853459322,
      "loss": 2.2832,
      "step": 72041
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7662327885627747,
      "learning_rate": 0.00046653138291138707,
      "loss": 2.3724,
      "step": 72042
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7259662747383118,
      "learning_rate": 0.0004665279804458807,
      "loss": 2.2867,
      "step": 72043
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7806887626647949,
      "learning_rate": 0.00046652457794941354,
      "loss": 2.1877,
      "step": 72044
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.8272044658660889,
      "learning_rate": 0.00046652117542198636,
      "loss": 2.326,
      "step": 72045
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7937692999839783,
      "learning_rate": 0.0004665177728635998,
      "loss": 2.1029,
      "step": 72046
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7340348362922668,
      "learning_rate": 0.00046651437027425436,
      "loss": 2.3599,
      "step": 72047
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.6954585313796997,
      "learning_rate": 0.0004665109676539508,
      "loss": 2.0485,
      "step": 72048
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.8417590856552124,
      "learning_rate": 0.0004665075650026896,
      "loss": 2.1767,
      "step": 72049
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.8122745752334595,
      "learning_rate": 0.00046650416232047164,
      "loss": 2.1165,
      "step": 72050
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.8053229451179504,
      "learning_rate": 0.0004665007596072973,
      "loss": 2.3557,
      "step": 72051
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.8643327355384827,
      "learning_rate": 0.00046649735686316745,
      "loss": 2.3358,
      "step": 72052
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7867370843887329,
      "learning_rate": 0.00046649395408808263,
      "loss": 2.3639,
      "step": 72053
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7197425365447998,
      "learning_rate": 0.0004664905512820433,
      "loss": 2.1481,
      "step": 72054
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7526267766952515,
      "learning_rate": 0.00046648714844505033,
      "loss": 2.2842,
      "step": 72055
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.778908908367157,
      "learning_rate": 0.0004664837455771042,
      "loss": 2.1849,
      "step": 72056
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.8068583607673645,
      "learning_rate": 0.00046648034267820555,
      "loss": 2.3594,
      "step": 72057
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7475740909576416,
      "learning_rate": 0.0004664769397483551,
      "loss": 2.0974,
      "step": 72058
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.87857985496521,
      "learning_rate": 0.00046647353678755355,
      "loss": 2.4065,
      "step": 72059
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.8824877142906189,
      "learning_rate": 0.00046647013379580134,
      "loss": 2.328,
      "step": 72060
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7708914875984192,
      "learning_rate": 0.00046646673077309917,
      "loss": 2.3347,
      "step": 72061
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.8173912763595581,
      "learning_rate": 0.00046646332771944774,
      "loss": 2.1199,
      "step": 72062
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7157697081565857,
      "learning_rate": 0.00046645992463484765,
      "loss": 2.1142,
      "step": 72063
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.742743194103241,
      "learning_rate": 0.0004664565215192994,
      "loss": 2.2098,
      "step": 72064
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.810021162033081,
      "learning_rate": 0.0004664531183728039,
      "loss": 2.2097,
      "step": 72065
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.8342331051826477,
      "learning_rate": 0.00046644971519536145,
      "loss": 2.2987,
      "step": 72066
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7773181200027466,
      "learning_rate": 0.00046644631198697296,
      "loss": 2.3243,
      "step": 72067
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.8514420390129089,
      "learning_rate": 0.000466442908747639,
      "loss": 2.211,
      "step": 72068
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7932611107826233,
      "learning_rate": 0.00046643950547736013,
      "loss": 2.3205,
      "step": 72069
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7199303507804871,
      "learning_rate": 0.00046643610217613694,
      "loss": 2.1929,
      "step": 72070
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.777022123336792,
      "learning_rate": 0.00046643269884397024,
      "loss": 2.1718,
      "step": 72071
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.810283899307251,
      "learning_rate": 0.00046642929548086046,
      "loss": 2.1731,
      "step": 72072
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7726523280143738,
      "learning_rate": 0.00046642589208680836,
      "loss": 2.2096,
      "step": 72073
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7866501808166504,
      "learning_rate": 0.00046642248866181465,
      "loss": 2.116,
      "step": 72074
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.781471312046051,
      "learning_rate": 0.0004664190852058797,
      "loss": 2.354,
      "step": 72075
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7707802057266235,
      "learning_rate": 0.0004664156817190044,
      "loss": 2.2856,
      "step": 72076
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7650855779647827,
      "learning_rate": 0.0004664122782011892,
      "loss": 2.1389,
      "step": 72077
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.6832291483879089,
      "learning_rate": 0.000466408874652435,
      "loss": 2.0611,
      "step": 72078
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7809440493583679,
      "learning_rate": 0.00046640547107274204,
      "loss": 2.3862,
      "step": 72079
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7640592455863953,
      "learning_rate": 0.00046640206746211124,
      "loss": 2.3648,
      "step": 72080
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.8075899481773376,
      "learning_rate": 0.0004663986638205432,
      "loss": 2.0801,
      "step": 72081
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7372570633888245,
      "learning_rate": 0.00046639526014803846,
      "loss": 2.1732,
      "step": 72082
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7830974459648132,
      "learning_rate": 0.0004663918564445977,
      "loss": 2.2423,
      "step": 72083
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.8088527321815491,
      "learning_rate": 0.0004663884527102215,
      "loss": 2.1137,
      "step": 72084
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7339608073234558,
      "learning_rate": 0.0004663850489449107,
      "loss": 2.1212,
      "step": 72085
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7512279748916626,
      "learning_rate": 0.00046638164514866564,
      "loss": 2.1189,
      "step": 72086
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7075319290161133,
      "learning_rate": 0.00046637824132148714,
      "loss": 1.9687,
      "step": 72087
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7326645851135254,
      "learning_rate": 0.00046637483746337575,
      "loss": 2.2829,
      "step": 72088
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7637635469436646,
      "learning_rate": 0.0004663714335743323,
      "loss": 1.9192,
      "step": 72089
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7566279172897339,
      "learning_rate": 0.00046636802965435703,
      "loss": 2.2332,
      "step": 72090
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7933189272880554,
      "learning_rate": 0.0004663646257034509,
      "loss": 1.9726,
      "step": 72091
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7987020611763,
      "learning_rate": 0.0004663612217216144,
      "loss": 1.99,
      "step": 72092
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7710894346237183,
      "learning_rate": 0.0004663578177088483,
      "loss": 2.4701,
      "step": 72093
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.8515082001686096,
      "learning_rate": 0.00046635441366515306,
      "loss": 2.2637,
      "step": 72094
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.756859540939331,
      "learning_rate": 0.00046635100959052954,
      "loss": 2.4763,
      "step": 72095
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7146825790405273,
      "learning_rate": 0.0004663476054849781,
      "loss": 2.2516,
      "step": 72096
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.8334013223648071,
      "learning_rate": 0.00046634420134849944,
      "loss": 2.2632,
      "step": 72097
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7860686182975769,
      "learning_rate": 0.00046634079718109434,
      "loss": 2.1098,
      "step": 72098
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7493628859519958,
      "learning_rate": 0.0004663373929827634,
      "loss": 2.1289,
      "step": 72099
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.8098785281181335,
      "learning_rate": 0.0004663339887535071,
      "loss": 2.1864,
      "step": 72100
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.8304417729377747,
      "learning_rate": 0.00046633058449332623,
      "loss": 2.403,
      "step": 72101
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7172730565071106,
      "learning_rate": 0.00046632718020222136,
      "loss": 2.1513,
      "step": 72102
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7880244851112366,
      "learning_rate": 0.0004663237758801931,
      "loss": 2.1824,
      "step": 72103
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.8002139925956726,
      "learning_rate": 0.00046632037152724207,
      "loss": 2.3316,
      "step": 72104
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.8041154146194458,
      "learning_rate": 0.000466316967143369,
      "loss": 2.4377,
      "step": 72105
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.6981661915779114,
      "learning_rate": 0.0004663135627285745,
      "loss": 2.005,
      "step": 72106
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.763438880443573,
      "learning_rate": 0.00046631015828285904,
      "loss": 2.2317,
      "step": 72107
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.878600537776947,
      "learning_rate": 0.0004663067538062236,
      "loss": 2.1502,
      "step": 72108
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.9123224020004272,
      "learning_rate": 0.00046630334929866844,
      "loss": 2.3331,
      "step": 72109
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.796532928943634,
      "learning_rate": 0.0004662999447601944,
      "loss": 2.1799,
      "step": 72110
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.8584251999855042,
      "learning_rate": 0.000466296540190802,
      "loss": 1.9474,
      "step": 72111
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.9096219539642334,
      "learning_rate": 0.00046629313559049196,
      "loss": 2.4094,
      "step": 72112
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7127777934074402,
      "learning_rate": 0.0004662897309592649,
      "loss": 2.3462,
      "step": 72113
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7463613152503967,
      "learning_rate": 0.0004662863262971215,
      "loss": 2.128,
      "step": 72114
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.8218548893928528,
      "learning_rate": 0.00046628292160406215,
      "loss": 2.2085,
      "step": 72115
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.0912100076675415,
      "learning_rate": 0.00046627951688008783,
      "loss": 2.0948,
      "step": 72116
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.8145657181739807,
      "learning_rate": 0.000466276112125199,
      "loss": 2.2567,
      "step": 72117
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7419745922088623,
      "learning_rate": 0.00046627270733939625,
      "loss": 2.2384,
      "step": 72118
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.758131206035614,
      "learning_rate": 0.00046626930252268025,
      "loss": 2.1576,
      "step": 72119
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.8510081171989441,
      "learning_rate": 0.00046626589767505175,
      "loss": 2.5494,
      "step": 72120
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.737922728061676,
      "learning_rate": 0.0004662624927965112,
      "loss": 2.1299,
      "step": 72121
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7412181496620178,
      "learning_rate": 0.00046625908788705924,
      "loss": 2.168,
      "step": 72122
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.8996970057487488,
      "learning_rate": 0.00046625568294669675,
      "loss": 2.1954,
      "step": 72123
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7154164910316467,
      "learning_rate": 0.00046625227797542415,
      "loss": 2.2812,
      "step": 72124
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.8343884348869324,
      "learning_rate": 0.000466248872973242,
      "loss": 2.0521,
      "step": 72125
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7553341388702393,
      "learning_rate": 0.0004662454679401511,
      "loss": 2.1046,
      "step": 72126
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7967805862426758,
      "learning_rate": 0.00046624206287615205,
      "loss": 2.1896,
      "step": 72127
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7619122266769409,
      "learning_rate": 0.0004662386577812455,
      "loss": 2.0745,
      "step": 72128
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7501295804977417,
      "learning_rate": 0.00046623525265543196,
      "loss": 2.1979,
      "step": 72129
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7462430596351624,
      "learning_rate": 0.00046623184749871216,
      "loss": 2.1241,
      "step": 72130
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7731941938400269,
      "learning_rate": 0.0004662284423110868,
      "loss": 2.1773,
      "step": 72131
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.8763361573219299,
      "learning_rate": 0.0004662250370925564,
      "loss": 2.2912,
      "step": 72132
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7936768531799316,
      "learning_rate": 0.0004662216318431216,
      "loss": 2.1233,
      "step": 72133
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.8273239135742188,
      "learning_rate": 0.0004662182265627831,
      "loss": 2.2526,
      "step": 72134
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.8236898183822632,
      "learning_rate": 0.00046621482125154144,
      "loss": 2.2163,
      "step": 72135
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.8299633264541626,
      "learning_rate": 0.0004662114159093973,
      "loss": 2.1718,
      "step": 72136
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7315662503242493,
      "learning_rate": 0.0004662080105363513,
      "loss": 2.2775,
      "step": 72137
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.8133893013000488,
      "learning_rate": 0.00046620460513240427,
      "loss": 2.371,
      "step": 72138
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7711077928543091,
      "learning_rate": 0.0004662011996975565,
      "loss": 2.2142,
      "step": 72139
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7951064109802246,
      "learning_rate": 0.00046619779423180886,
      "loss": 2.1935,
      "step": 72140
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.8336393237113953,
      "learning_rate": 0.0004661943887351619,
      "loss": 2.2869,
      "step": 72141
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.8205772042274475,
      "learning_rate": 0.00046619098320761624,
      "loss": 2.0648,
      "step": 72142
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.9343152046203613,
      "learning_rate": 0.00046618757764917254,
      "loss": 2.423,
      "step": 72143
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7227523326873779,
      "learning_rate": 0.0004661841720598315,
      "loss": 2.1843,
      "step": 72144
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7699068188667297,
      "learning_rate": 0.00046618076643959355,
      "loss": 2.5079,
      "step": 72145
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.8087255954742432,
      "learning_rate": 0.0004661773607884595,
      "loss": 2.4453,
      "step": 72146
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.8260194063186646,
      "learning_rate": 0.0004661739551064301,
      "loss": 2.1913,
      "step": 72147
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7852334380149841,
      "learning_rate": 0.00046617054939350564,
      "loss": 2.2669,
      "step": 72148
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7206018567085266,
      "learning_rate": 0.00046616714364968706,
      "loss": 2.1225,
      "step": 72149
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.8886626958847046,
      "learning_rate": 0.0004661637378749748,
      "loss": 2.0819,
      "step": 72150
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.8007485270500183,
      "learning_rate": 0.0004661603320693696,
      "loss": 2.1759,
      "step": 72151
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.8196470141410828,
      "learning_rate": 0.00046615692623287204,
      "loss": 2.3051,
      "step": 72152
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.775266706943512,
      "learning_rate": 0.00046615352036548277,
      "loss": 2.1909,
      "step": 72153
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.877930760383606,
      "learning_rate": 0.00046615011446720235,
      "loss": 2.1037,
      "step": 72154
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7715115547180176,
      "learning_rate": 0.0004661467085380316,
      "loss": 2.155,
      "step": 72155
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.8582817316055298,
      "learning_rate": 0.000466143302577971,
      "loss": 2.2641,
      "step": 72156
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.8052730560302734,
      "learning_rate": 0.0004661398965870212,
      "loss": 2.2388,
      "step": 72157
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7237127423286438,
      "learning_rate": 0.0004661364905651829,
      "loss": 2.1424,
      "step": 72158
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7122084498405457,
      "learning_rate": 0.00046613308451245674,
      "loss": 2.3078,
      "step": 72159
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7790579199790955,
      "learning_rate": 0.00046612967842884324,
      "loss": 2.0707,
      "step": 72160
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7284008264541626,
      "learning_rate": 0.00046612627231434303,
      "loss": 2.314,
      "step": 72161
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.8029701709747314,
      "learning_rate": 0.0004661228661689569,
      "loss": 2.2896,
      "step": 72162
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7771305441856384,
      "learning_rate": 0.0004661194599926853,
      "loss": 2.2734,
      "step": 72163
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.8007948994636536,
      "learning_rate": 0.0004661160537855291,
      "loss": 2.3571,
      "step": 72164
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7703953385353088,
      "learning_rate": 0.0004661126475474887,
      "loss": 2.1684,
      "step": 72165
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7023608684539795,
      "learning_rate": 0.0004661092412785648,
      "loss": 2.0844,
      "step": 72166
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7341570854187012,
      "learning_rate": 0.0004661058349787581,
      "loss": 2.0668,
      "step": 72167
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.8226627111434937,
      "learning_rate": 0.00046610242864806916,
      "loss": 2.3,
      "step": 72168
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7865532636642456,
      "learning_rate": 0.0004660990222864987,
      "loss": 2.1881,
      "step": 72169
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.8215011954307556,
      "learning_rate": 0.00046609561589404727,
      "loss": 2.2421,
      "step": 72170
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7313429713249207,
      "learning_rate": 0.0004660922094707155,
      "loss": 1.9834,
      "step": 72171
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.8036991357803345,
      "learning_rate": 0.00046608880301650407,
      "loss": 2.1831,
      "step": 72172
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.8393162488937378,
      "learning_rate": 0.00046608539653141364,
      "loss": 2.3574,
      "step": 72173
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.8219805955886841,
      "learning_rate": 0.00046608199001544476,
      "loss": 2.5218,
      "step": 72174
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.753720223903656,
      "learning_rate": 0.0004660785834685981,
      "loss": 2.1001,
      "step": 72175
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.8540490865707397,
      "learning_rate": 0.0004660751768908742,
      "loss": 2.3069,
      "step": 72176
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.9892740249633789,
      "learning_rate": 0.000466071770282274,
      "loss": 2.0848,
      "step": 72177
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.8107983469963074,
      "learning_rate": 0.0004660683636427978,
      "loss": 2.3099,
      "step": 72178
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7134909629821777,
      "learning_rate": 0.00046606495697244636,
      "loss": 2.3495,
      "step": 72179
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7996415495872498,
      "learning_rate": 0.00046606155027122034,
      "loss": 2.0836,
      "step": 72180
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7769725918769836,
      "learning_rate": 0.00046605814353912036,
      "loss": 2.3314,
      "step": 72181
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.9948790669441223,
      "learning_rate": 0.0004660547367761469,
      "loss": 2.194,
      "step": 72182
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.6529839038848877,
      "learning_rate": 0.0004660513299823008,
      "loss": 2.1852,
      "step": 72183
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.9289087653160095,
      "learning_rate": 0.00046604792315758274,
      "loss": 2.0816,
      "step": 72184
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.8618460297584534,
      "learning_rate": 0.0004660445163019931,
      "loss": 2.2395,
      "step": 72185
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.8574469685554504,
      "learning_rate": 0.0004660411094155327,
      "loss": 2.6072,
      "step": 72186
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7595444321632385,
      "learning_rate": 0.00046603770249820217,
      "loss": 2.3205,
      "step": 72187
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7752867937088013,
      "learning_rate": 0.0004660342955500021,
      "loss": 2.0361,
      "step": 72188
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.8899109363555908,
      "learning_rate": 0.00046603088857093296,
      "loss": 2.4223,
      "step": 72189
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.8904144167900085,
      "learning_rate": 0.0004660274815609957,
      "loss": 1.9878,
      "step": 72190
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.8037683963775635,
      "learning_rate": 0.0004660240745201908,
      "loss": 2.3724,
      "step": 72191
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7433598041534424,
      "learning_rate": 0.0004660206674485188,
      "loss": 2.4813,
      "step": 72192
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.8041735887527466,
      "learning_rate": 0.0004660172603459805,
      "loss": 2.1497,
      "step": 72193
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.8209198713302612,
      "learning_rate": 0.0004660138532125764,
      "loss": 2.3272,
      "step": 72194
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7933212518692017,
      "learning_rate": 0.0004660104460483072,
      "loss": 2.0615,
      "step": 72195
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7707507610321045,
      "learning_rate": 0.0004660070388531736,
      "loss": 2.2574,
      "step": 72196
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7763211727142334,
      "learning_rate": 0.00046600363162717606,
      "loss": 2.3387,
      "step": 72197
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.8544971346855164,
      "learning_rate": 0.00046600022437031535,
      "loss": 2.2502,
      "step": 72198
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7380637526512146,
      "learning_rate": 0.000465996817082592,
      "loss": 1.9345,
      "step": 72199
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.790184736251831,
      "learning_rate": 0.0004659934097640068,
      "loss": 2.1711,
      "step": 72200
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7736315131187439,
      "learning_rate": 0.0004659900024145603,
      "loss": 2.2163,
      "step": 72201
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7438315153121948,
      "learning_rate": 0.0004659865950342531,
      "loss": 2.3457,
      "step": 72202
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7989547848701477,
      "learning_rate": 0.0004659831876230858,
      "loss": 2.1296,
      "step": 72203
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7643016576766968,
      "learning_rate": 0.0004659797801810591,
      "loss": 2.2058,
      "step": 72204
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.9783946871757507,
      "learning_rate": 0.0004659763727081737,
      "loss": 2.0056,
      "step": 72205
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7496238946914673,
      "learning_rate": 0.00046597296520443015,
      "loss": 2.1768,
      "step": 72206
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.9311008453369141,
      "learning_rate": 0.00046596955766982904,
      "loss": 2.1079,
      "step": 72207
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7681293487548828,
      "learning_rate": 0.00046596615010437116,
      "loss": 2.2449,
      "step": 72208
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7941269874572754,
      "learning_rate": 0.00046596274250805693,
      "loss": 2.2402,
      "step": 72209
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.8510377407073975,
      "learning_rate": 0.0004659593348808871,
      "loss": 2.2323,
      "step": 72210
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7142969369888306,
      "learning_rate": 0.00046595592722286234,
      "loss": 2.2681,
      "step": 72211
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.9449607729911804,
      "learning_rate": 0.00046595251953398316,
      "loss": 2.2769,
      "step": 72212
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7713283896446228,
      "learning_rate": 0.0004659491118142504,
      "loss": 2.2514,
      "step": 72213
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.753736674785614,
      "learning_rate": 0.0004659457040636645,
      "loss": 2.3343,
      "step": 72214
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.8222188949584961,
      "learning_rate": 0.0004659422962822262,
      "loss": 2.3572,
      "step": 72215
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7895825505256653,
      "learning_rate": 0.000465938888469936,
      "loss": 2.0904,
      "step": 72216
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.756791353225708,
      "learning_rate": 0.00046593548062679466,
      "loss": 2.2422,
      "step": 72217
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.8203850984573364,
      "learning_rate": 0.00046593207275280287,
      "loss": 2.3155,
      "step": 72218
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7873238921165466,
      "learning_rate": 0.0004659286648479611,
      "loss": 2.1035,
      "step": 72219
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.8461229205131531,
      "learning_rate": 0.0004659252569122701,
      "loss": 2.2064,
      "step": 72220
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.8349444270133972,
      "learning_rate": 0.00046592184894573045,
      "loss": 2.0193,
      "step": 72221
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.9072080850601196,
      "learning_rate": 0.0004659184409483427,
      "loss": 2.3484,
      "step": 72222
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7290167212486267,
      "learning_rate": 0.00046591503292010773,
      "loss": 2.1906,
      "step": 72223
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.8219211101531982,
      "learning_rate": 0.00046591162486102594,
      "loss": 2.4154,
      "step": 72224
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7685564160346985,
      "learning_rate": 0.00046590821677109805,
      "loss": 2.3087,
      "step": 72225
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.8039700388908386,
      "learning_rate": 0.0004659048086503247,
      "loss": 2.1862,
      "step": 72226
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.8009565472602844,
      "learning_rate": 0.00046590140049870653,
      "loss": 2.1488,
      "step": 72227
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.8123990297317505,
      "learning_rate": 0.00046589799231624415,
      "loss": 2.4393,
      "step": 72228
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.881483256816864,
      "learning_rate": 0.0004658945841029382,
      "loss": 1.9578,
      "step": 72229
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7962501049041748,
      "learning_rate": 0.00046589117585878935,
      "loss": 2.3725,
      "step": 72230
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.8047850728034973,
      "learning_rate": 0.0004658877675837981,
      "loss": 2.2309,
      "step": 72231
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.8389862179756165,
      "learning_rate": 0.0004658843592779652,
      "loss": 2.2506,
      "step": 72232
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.8113260865211487,
      "learning_rate": 0.0004658809509412914,
      "loss": 2.0357,
      "step": 72233
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7318357229232788,
      "learning_rate": 0.0004658775425737771,
      "loss": 2.1733,
      "step": 72234
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.8629632592201233,
      "learning_rate": 0.00046587413417542307,
      "loss": 1.9371,
      "step": 72235
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.8051778674125671,
      "learning_rate": 0.00046587072574622994,
      "loss": 2.2172,
      "step": 72236
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.837529182434082,
      "learning_rate": 0.00046586731728619815,
      "loss": 2.0826,
      "step": 72237
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.8407296538352966,
      "learning_rate": 0.0004658639087953286,
      "loss": 2.2813,
      "step": 72238
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.8597979545593262,
      "learning_rate": 0.00046586050027362186,
      "loss": 2.2531,
      "step": 72239
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7645791172981262,
      "learning_rate": 0.0004658570917210785,
      "loss": 1.9862,
      "step": 72240
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7960901856422424,
      "learning_rate": 0.0004658536831376991,
      "loss": 2.1481,
      "step": 72241
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.0208439826965332,
      "learning_rate": 0.0004658502745234845,
      "loss": 2.0713,
      "step": 72242
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7469887733459473,
      "learning_rate": 0.00046584686587843504,
      "loss": 2.0782,
      "step": 72243
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7395553588867188,
      "learning_rate": 0.00046584345720255166,
      "loss": 2.2094,
      "step": 72244
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.8222907185554504,
      "learning_rate": 0.0004658400484958348,
      "loss": 2.3144,
      "step": 72245
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.8511467576026917,
      "learning_rate": 0.00046583663975828516,
      "loss": 2.2398,
      "step": 72246
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7187526226043701,
      "learning_rate": 0.0004658332309899033,
      "loss": 2.1337,
      "step": 72247
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.8510147333145142,
      "learning_rate": 0.00046582982219069,
      "loss": 2.4289,
      "step": 72248
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.9508428573608398,
      "learning_rate": 0.0004658264133606457,
      "loss": 2.269,
      "step": 72249
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7089742422103882,
      "learning_rate": 0.0004658230044997712,
      "loss": 2.1525,
      "step": 72250
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.9624225497245789,
      "learning_rate": 0.00046581959560806706,
      "loss": 2.2132,
      "step": 72251
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7220609784126282,
      "learning_rate": 0.00046581618668553403,
      "loss": 2.1614,
      "step": 72252
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7211862802505493,
      "learning_rate": 0.0004658127777321725,
      "loss": 2.0711,
      "step": 72253
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7667012214660645,
      "learning_rate": 0.00046580936874798335,
      "loss": 2.2904,
      "step": 72254
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.9572588801383972,
      "learning_rate": 0.00046580595973296694,
      "loss": 2.4053,
      "step": 72255
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7466238141059875,
      "learning_rate": 0.0004658025506871243,
      "loss": 2.2307,
      "step": 72256
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7317742705345154,
      "learning_rate": 0.00046579914161045565,
      "loss": 2.3844,
      "step": 72257
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.8835885524749756,
      "learning_rate": 0.00046579573250296195,
      "loss": 2.2133,
      "step": 72258
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7890147566795349,
      "learning_rate": 0.0004657923233646436,
      "loss": 2.2998,
      "step": 72259
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7488754987716675,
      "learning_rate": 0.00046578891419550133,
      "loss": 2.2571,
      "step": 72260
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.945672333240509,
      "learning_rate": 0.00046578550499553576,
      "loss": 1.9846,
      "step": 72261
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.828045666217804,
      "learning_rate": 0.00046578209576474757,
      "loss": 2.294,
      "step": 72262
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7060101628303528,
      "learning_rate": 0.00046577868650313735,
      "loss": 2.2496,
      "step": 72263
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7726141810417175,
      "learning_rate": 0.00046577527721070576,
      "loss": 2.3295,
      "step": 72264
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.2121703624725342,
      "learning_rate": 0.0004657718678874534,
      "loss": 2.2554,
      "step": 72265
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7978119254112244,
      "learning_rate": 0.00046576845853338095,
      "loss": 2.1659,
      "step": 72266
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.8001774549484253,
      "learning_rate": 0.000465765049148489,
      "loss": 2.2268,
      "step": 72267
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.0086259841918945,
      "learning_rate": 0.0004657616397327782,
      "loss": 2.2922,
      "step": 72268
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.8106942176818848,
      "learning_rate": 0.00046575823028624914,
      "loss": 2.1278,
      "step": 72269
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7247692346572876,
      "learning_rate": 0.0004657548208089025,
      "loss": 2.3642,
      "step": 72270
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.8488460779190063,
      "learning_rate": 0.00046575141130073894,
      "loss": 2.063,
      "step": 72271
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7800900936126709,
      "learning_rate": 0.0004657480017617591,
      "loss": 2.3183,
      "step": 72272
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.8595640063285828,
      "learning_rate": 0.0004657445921919635,
      "loss": 2.0557,
      "step": 72273
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.8398005366325378,
      "learning_rate": 0.00046574118259135294,
      "loss": 2.2358,
      "step": 72274
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.1006540060043335,
      "learning_rate": 0.00046573777295992787,
      "loss": 2.1786,
      "step": 72275
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7908003330230713,
      "learning_rate": 0.00046573436329768906,
      "loss": 2.3168,
      "step": 72276
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7760244011878967,
      "learning_rate": 0.0004657309536046371,
      "loss": 2.0995,
      "step": 72277
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.815048336982727,
      "learning_rate": 0.0004657275438807727,
      "loss": 2.1574,
      "step": 72278
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.9249367117881775,
      "learning_rate": 0.00046572413412609634,
      "loss": 2.3745,
      "step": 72279
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.9195756912231445,
      "learning_rate": 0.00046572072434060874,
      "loss": 2.0574,
      "step": 72280
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.8562604188919067,
      "learning_rate": 0.00046571731452431055,
      "loss": 2.298,
      "step": 72281
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7348403930664062,
      "learning_rate": 0.0004657139046772024,
      "loss": 2.2448,
      "step": 72282
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.780463695526123,
      "learning_rate": 0.00046571049479928486,
      "loss": 2.1009,
      "step": 72283
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.8926069736480713,
      "learning_rate": 0.00046570708489055866,
      "loss": 2.2706,
      "step": 72284
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.002221941947937,
      "learning_rate": 0.0004657036749510243,
      "loss": 2.3184,
      "step": 72285
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7772808074951172,
      "learning_rate": 0.00046570026498068257,
      "loss": 2.3978,
      "step": 72286
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.1618415117263794,
      "learning_rate": 0.00046569685497953404,
      "loss": 2.4064,
      "step": 72287
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.8143448233604431,
      "learning_rate": 0.0004656934449475793,
      "loss": 2.3003,
      "step": 72288
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7478396892547607,
      "learning_rate": 0.00046569003488481904,
      "loss": 2.2483,
      "step": 72289
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.8489178419113159,
      "learning_rate": 0.00046568662479125393,
      "loss": 2.2155,
      "step": 72290
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.8632534146308899,
      "learning_rate": 0.00046568321466688447,
      "loss": 2.3472,
      "step": 72291
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.9290992021560669,
      "learning_rate": 0.00046567980451171136,
      "loss": 2.099,
      "step": 72292
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7638344168663025,
      "learning_rate": 0.00046567639432573536,
      "loss": 2.1834,
      "step": 72293
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.8512452244758606,
      "learning_rate": 0.00046567298410895685,
      "loss": 2.1654,
      "step": 72294
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7856426239013672,
      "learning_rate": 0.00046566957386137665,
      "loss": 2.2087,
      "step": 72295
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.6949425339698792,
      "learning_rate": 0.0004656661635829954,
      "loss": 2.1311,
      "step": 72296
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7960790395736694,
      "learning_rate": 0.00046566275327381365,
      "loss": 2.1256,
      "step": 72297
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7681925296783447,
      "learning_rate": 0.00046565934293383204,
      "loss": 2.2064,
      "step": 72298
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.8842224478721619,
      "learning_rate": 0.00046565593256305124,
      "loss": 2.2744,
      "step": 72299
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7897588610649109,
      "learning_rate": 0.00046565252216147193,
      "loss": 2.145,
      "step": 72300
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7313583493232727,
      "learning_rate": 0.00046564911172909467,
      "loss": 2.2061,
      "step": 72301
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.9999793171882629,
      "learning_rate": 0.00046564570126592,
      "loss": 2.1854,
      "step": 72302
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.8644862771034241,
      "learning_rate": 0.0004656422907719488,
      "loss": 2.2924,
      "step": 72303
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7496190667152405,
      "learning_rate": 0.0004656388802471815,
      "loss": 1.8715,
      "step": 72304
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.020348310470581,
      "learning_rate": 0.0004656354696916189,
      "loss": 2.2988,
      "step": 72305
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.8686296939849854,
      "learning_rate": 0.0004656320591052615,
      "loss": 2.3817,
      "step": 72306
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.8745582103729248,
      "learning_rate": 0.0004656286484881099,
      "loss": 2.3296,
      "step": 72307
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.0219905376434326,
      "learning_rate": 0.0004656252378401649,
      "loss": 2.1689,
      "step": 72308
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7614548206329346,
      "learning_rate": 0.0004656218271614269,
      "loss": 2.1401,
      "step": 72309
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7322739362716675,
      "learning_rate": 0.0004656184164518968,
      "loss": 2.0868,
      "step": 72310
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.8128718733787537,
      "learning_rate": 0.00046561500571157515,
      "loss": 2.0898,
      "step": 72311
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.9230467677116394,
      "learning_rate": 0.00046561159494046245,
      "loss": 2.3134,
      "step": 72312
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7573271989822388,
      "learning_rate": 0.0004656081841385595,
      "loss": 2.2418,
      "step": 72313
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.8271130323410034,
      "learning_rate": 0.0004656047733058668,
      "loss": 2.1419,
      "step": 72314
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.8428608775138855,
      "learning_rate": 0.0004656013624423851,
      "loss": 2.1774,
      "step": 72315
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.8178601861000061,
      "learning_rate": 0.00046559795154811493,
      "loss": 2.2026,
      "step": 72316
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.727618932723999,
      "learning_rate": 0.0004655945406230569,
      "loss": 2.133,
      "step": 72317
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7494703531265259,
      "learning_rate": 0.0004655911296672119,
      "loss": 2.1998,
      "step": 72318
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.719489574432373,
      "learning_rate": 0.00046558771868058027,
      "loss": 2.2959,
      "step": 72319
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.8995427489280701,
      "learning_rate": 0.0004655843076631628,
      "loss": 2.2823,
      "step": 72320
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.760210394859314,
      "learning_rate": 0.0004655808966149601,
      "loss": 2.0493,
      "step": 72321
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7891821265220642,
      "learning_rate": 0.0004655774855359727,
      "loss": 2.0024,
      "step": 72322
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7825325131416321,
      "learning_rate": 0.0004655740744262014,
      "loss": 2.2451,
      "step": 72323
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7434820532798767,
      "learning_rate": 0.00046557066328564673,
      "loss": 1.8909,
      "step": 72324
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7538249492645264,
      "learning_rate": 0.0004655672521143093,
      "loss": 2.1898,
      "step": 72325
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.8027829527854919,
      "learning_rate": 0.00046556384091218987,
      "loss": 2.2979,
      "step": 72326
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.8790187239646912,
      "learning_rate": 0.000465560429679289,
      "loss": 2.0558,
      "step": 72327
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.8717056512832642,
      "learning_rate": 0.00046555701841560727,
      "loss": 2.0655,
      "step": 72328
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7779305577278137,
      "learning_rate": 0.00046555360712114536,
      "loss": 1.9999,
      "step": 72329
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7671792507171631,
      "learning_rate": 0.0004655501957959039,
      "loss": 2.1434,
      "step": 72330
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7837486863136292,
      "learning_rate": 0.0004655467844398836,
      "loss": 2.4131,
      "step": 72331
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.8579456210136414,
      "learning_rate": 0.00046554337305308496,
      "loss": 2.2427,
      "step": 72332
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.8265076875686646,
      "learning_rate": 0.0004655399616355088,
      "loss": 2.223,
      "step": 72333
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7673166990280151,
      "learning_rate": 0.0004655365501871555,
      "loss": 2.083,
      "step": 72334
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.859223484992981,
      "learning_rate": 0.0004655331387080259,
      "loss": 2.0338,
      "step": 72335
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7327427268028259,
      "learning_rate": 0.0004655297271981206,
      "loss": 2.1921,
      "step": 72336
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.728009819984436,
      "learning_rate": 0.0004655263156574401,
      "loss": 2.3289,
      "step": 72337
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.8790581226348877,
      "learning_rate": 0.0004655229040859851,
      "loss": 2.1653,
      "step": 72338
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.8141035437583923,
      "learning_rate": 0.0004655194924837564,
      "loss": 2.3461,
      "step": 72339
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.8364704251289368,
      "learning_rate": 0.00046551608085075444,
      "loss": 2.2485,
      "step": 72340
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7636027932167053,
      "learning_rate": 0.0004655126691869799,
      "loss": 2.055,
      "step": 72341
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.793806254863739,
      "learning_rate": 0.0004655092574924335,
      "loss": 2.2696,
      "step": 72342
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7720125913619995,
      "learning_rate": 0.0004655058457671157,
      "loss": 2.2417,
      "step": 72343
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7854337692260742,
      "learning_rate": 0.00046550243401102733,
      "loss": 2.0893,
      "step": 72344
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.8175194263458252,
      "learning_rate": 0.00046549902222416887,
      "loss": 2.1171,
      "step": 72345
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7263529300689697,
      "learning_rate": 0.00046549561040654114,
      "loss": 1.9565,
      "step": 72346
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7188274264335632,
      "learning_rate": 0.0004654921985581445,
      "loss": 1.9921,
      "step": 72347
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7950266003608704,
      "learning_rate": 0.0004654887866789798,
      "loss": 2.2469,
      "step": 72348
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7971691489219666,
      "learning_rate": 0.0004654853747690475,
      "loss": 2.0832,
      "step": 72349
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.8370405435562134,
      "learning_rate": 0.00046548196282834853,
      "loss": 2.3006,
      "step": 72350
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7705392837524414,
      "learning_rate": 0.0004654785508568833,
      "loss": 2.3836,
      "step": 72351
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.8008784651756287,
      "learning_rate": 0.0004654751388546524,
      "loss": 2.2889,
      "step": 72352
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7941259145736694,
      "learning_rate": 0.00046547172682165664,
      "loss": 2.4164,
      "step": 72353
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.000093698501587,
      "learning_rate": 0.00046546831475789653,
      "loss": 2.2114,
      "step": 72354
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.8575361967086792,
      "learning_rate": 0.0004654649026633727,
      "loss": 2.2147,
      "step": 72355
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7995367050170898,
      "learning_rate": 0.0004654614905380858,
      "loss": 2.1686,
      "step": 72356
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.8518210053443909,
      "learning_rate": 0.00046545807838203657,
      "loss": 2.1581,
      "step": 72357
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.837805449962616,
      "learning_rate": 0.00046545466619522547,
      "loss": 2.293,
      "step": 72358
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7449531555175781,
      "learning_rate": 0.0004654512539776533,
      "loss": 2.1713,
      "step": 72359
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.8499477505683899,
      "learning_rate": 0.0004654478417293206,
      "loss": 2.1173,
      "step": 72360
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.775996208190918,
      "learning_rate": 0.00046544442945022806,
      "loss": 2.0312,
      "step": 72361
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.786137580871582,
      "learning_rate": 0.00046544101714037617,
      "loss": 2.2276,
      "step": 72362
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.8788679838180542,
      "learning_rate": 0.0004654376047997658,
      "loss": 2.3841,
      "step": 72363
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.9988377690315247,
      "learning_rate": 0.00046543419242839736,
      "loss": 2.164,
      "step": 72364
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.8147125840187073,
      "learning_rate": 0.0004654307800262716,
      "loss": 2.3393,
      "step": 72365
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7847902178764343,
      "learning_rate": 0.0004654273675933891,
      "loss": 2.0836,
      "step": 72366
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.9203317761421204,
      "learning_rate": 0.00046542395512975065,
      "loss": 2.4994,
      "step": 72367
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7928902506828308,
      "learning_rate": 0.00046542054263535665,
      "loss": 2.297,
      "step": 72368
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.8173489570617676,
      "learning_rate": 0.0004654171301102079,
      "loss": 2.316,
      "step": 72369
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.8273113369941711,
      "learning_rate": 0.000465413717554305,
      "loss": 2.0439,
      "step": 72370
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.8301674723625183,
      "learning_rate": 0.00046541030496764855,
      "loss": 2.4567,
      "step": 72371
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7463619112968445,
      "learning_rate": 0.00046540689235023913,
      "loss": 2.176,
      "step": 72372
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7432560324668884,
      "learning_rate": 0.0004654034797020775,
      "loss": 2.1518,
      "step": 72373
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7945590019226074,
      "learning_rate": 0.0004654000670231643,
      "loss": 2.073,
      "step": 72374
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7590237855911255,
      "learning_rate": 0.0004653966543135,
      "loss": 2.1127,
      "step": 72375
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7221857309341431,
      "learning_rate": 0.00046539324157308546,
      "loss": 2.2338,
      "step": 72376
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.8332304954528809,
      "learning_rate": 0.00046538982880192116,
      "loss": 2.0961,
      "step": 72377
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7998718023300171,
      "learning_rate": 0.0004653864160000077,
      "loss": 2.3264,
      "step": 72378
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7993865013122559,
      "learning_rate": 0.0004653830031673459,
      "loss": 2.2194,
      "step": 72379
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7663809657096863,
      "learning_rate": 0.00046537959030393613,
      "loss": 2.0409,
      "step": 72380
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.810799777507782,
      "learning_rate": 0.00046537617740977923,
      "loss": 2.4157,
      "step": 72381
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7414966225624084,
      "learning_rate": 0.0004653727644848758,
      "loss": 2.1944,
      "step": 72382
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7445254921913147,
      "learning_rate": 0.00046536935152922645,
      "loss": 2.26,
      "step": 72383
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.8622307181358337,
      "learning_rate": 0.00046536593854283175,
      "loss": 2.2414,
      "step": 72384
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.8135966658592224,
      "learning_rate": 0.0004653625255256925,
      "loss": 2.3046,
      "step": 72385
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7531160116195679,
      "learning_rate": 0.00046535911247780917,
      "loss": 2.4214,
      "step": 72386
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.787283718585968,
      "learning_rate": 0.00046535569939918254,
      "loss": 2.125,
      "step": 72387
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.8496403098106384,
      "learning_rate": 0.00046535228628981313,
      "loss": 2.352,
      "step": 72388
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.8009811639785767,
      "learning_rate": 0.00046534887314970155,
      "loss": 2.0944,
      "step": 72389
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.8411204218864441,
      "learning_rate": 0.0004653454599788485,
      "loss": 2.3071,
      "step": 72390
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7907138466835022,
      "learning_rate": 0.00046534204677725464,
      "loss": 2.3205,
      "step": 72391
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.8210259079933167,
      "learning_rate": 0.00046533863354492055,
      "loss": 2.1156,
      "step": 72392
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.818954348564148,
      "learning_rate": 0.00046533522028184684,
      "loss": 2.1547,
      "step": 72393
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.8297279477119446,
      "learning_rate": 0.0004653318069880343,
      "loss": 2.1178,
      "step": 72394
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7645143270492554,
      "learning_rate": 0.00046532839366348347,
      "loss": 2.1056,
      "step": 72395
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.8011391162872314,
      "learning_rate": 0.0004653249803081949,
      "loss": 2.4917,
      "step": 72396
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7230425477027893,
      "learning_rate": 0.00046532156692216935,
      "loss": 2.153,
      "step": 72397
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.802213191986084,
      "learning_rate": 0.0004653181535054073,
      "loss": 2.1463,
      "step": 72398
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.755772590637207,
      "learning_rate": 0.0004653147400579096,
      "loss": 2.2624,
      "step": 72399
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7848238348960876,
      "learning_rate": 0.0004653113265796767,
      "loss": 1.9695,
      "step": 72400
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7958061099052429,
      "learning_rate": 0.00046530791307070925,
      "loss": 2.3027,
      "step": 72401
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.8378555774688721,
      "learning_rate": 0.00046530449953100807,
      "loss": 2.2636,
      "step": 72402
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.8245312571525574,
      "learning_rate": 0.0004653010859605736,
      "loss": 2.1097,
      "step": 72403
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7733345627784729,
      "learning_rate": 0.00046529767235940653,
      "loss": 2.3619,
      "step": 72404
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.8336931467056274,
      "learning_rate": 0.00046529425872750746,
      "loss": 2.0328,
      "step": 72405
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.798011064529419,
      "learning_rate": 0.00046529084506487716,
      "loss": 2.4223,
      "step": 72406
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7879629731178284,
      "learning_rate": 0.0004652874313715161,
      "loss": 1.9851,
      "step": 72407
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7776751518249512,
      "learning_rate": 0.000465284017647425,
      "loss": 2.253,
      "step": 72408
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.9227256774902344,
      "learning_rate": 0.00046528060389260453,
      "loss": 2.2687,
      "step": 72409
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7565123438835144,
      "learning_rate": 0.0004652771901070552,
      "loss": 2.1885,
      "step": 72410
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.78510981798172,
      "learning_rate": 0.00046527377629077774,
      "loss": 2.147,
      "step": 72411
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.742975652217865,
      "learning_rate": 0.0004652703624437728,
      "loss": 2.1933,
      "step": 72412
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7544581890106201,
      "learning_rate": 0.00046526694856604097,
      "loss": 2.1456,
      "step": 72413
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.8261754512786865,
      "learning_rate": 0.0004652635346575829,
      "loss": 1.9872,
      "step": 72414
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.8179534673690796,
      "learning_rate": 0.0004652601207183993,
      "loss": 2.24,
      "step": 72415
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.8089142441749573,
      "learning_rate": 0.00046525670674849056,
      "loss": 2.1906,
      "step": 72416
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.8044564127922058,
      "learning_rate": 0.00046525329274785753,
      "loss": 2.2741,
      "step": 72417
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.3548235893249512,
      "learning_rate": 0.0004652498787165008,
      "loss": 2.2502,
      "step": 72418
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.8948073387145996,
      "learning_rate": 0.00046524646465442106,
      "loss": 2.3119,
      "step": 72419
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7477571368217468,
      "learning_rate": 0.0004652430505616188,
      "loss": 2.3315,
      "step": 72420
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.9007666110992432,
      "learning_rate": 0.0004652396364380948,
      "loss": 2.0204,
      "step": 72421
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.8598966598510742,
      "learning_rate": 0.00046523622228384964,
      "loss": 2.162,
      "step": 72422
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.8949536681175232,
      "learning_rate": 0.0004652328080988839,
      "loss": 2.1697,
      "step": 72423
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.9790593981742859,
      "learning_rate": 0.00046522939388319825,
      "loss": 2.2864,
      "step": 72424
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7442774176597595,
      "learning_rate": 0.0004652259796367934,
      "loss": 2.0362,
      "step": 72425
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7666056156158447,
      "learning_rate": 0.0004652225653596699,
      "loss": 2.3018,
      "step": 72426
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.0489895343780518,
      "learning_rate": 0.0004652191510518284,
      "loss": 2.1567,
      "step": 72427
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.8408093452453613,
      "learning_rate": 0.0004652157367132695,
      "loss": 2.2478,
      "step": 72428
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7974704504013062,
      "learning_rate": 0.00046521232234399395,
      "loss": 2.3752,
      "step": 72429
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.9148162603378296,
      "learning_rate": 0.00046520890794400226,
      "loss": 2.2409,
      "step": 72430
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7684599161148071,
      "learning_rate": 0.00046520549351329507,
      "loss": 2.2517,
      "step": 72431
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7453467845916748,
      "learning_rate": 0.00046520207905187313,
      "loss": 2.1492,
      "step": 72432
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7062523365020752,
      "learning_rate": 0.000465198664559737,
      "loss": 1.9514,
      "step": 72433
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7506616115570068,
      "learning_rate": 0.0004651952500368873,
      "loss": 2.0326,
      "step": 72434
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.810273289680481,
      "learning_rate": 0.0004651918354833248,
      "loss": 2.2179,
      "step": 72435
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.8319383859634399,
      "learning_rate": 0.00046518842089904983,
      "loss": 2.1125,
      "step": 72436
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7813879251480103,
      "learning_rate": 0.00046518500628406334,
      "loss": 2.3627,
      "step": 72437
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.8218521475791931,
      "learning_rate": 0.0004651815916383658,
      "loss": 2.4234,
      "step": 72438
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7883198857307434,
      "learning_rate": 0.00046517817696195785,
      "loss": 2.3024,
      "step": 72439
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.697286069393158,
      "learning_rate": 0.00046517476225484014,
      "loss": 2.1323,
      "step": 72440
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7681443691253662,
      "learning_rate": 0.0004651713475170134,
      "loss": 2.0202,
      "step": 72441
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.8594037890434265,
      "learning_rate": 0.0004651679327484781,
      "loss": 2.3515,
      "step": 72442
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7842336297035217,
      "learning_rate": 0.00046516451794923516,
      "loss": 2.2987,
      "step": 72443
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7568531632423401,
      "learning_rate": 0.0004651611031192848,
      "loss": 2.321,
      "step": 72444
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7717365026473999,
      "learning_rate": 0.00046515768825862794,
      "loss": 2.2003,
      "step": 72445
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7561132311820984,
      "learning_rate": 0.0004651542733672652,
      "loss": 2.3094,
      "step": 72446
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.753944456577301,
      "learning_rate": 0.00046515085844519715,
      "loss": 2.1635,
      "step": 72447
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7872032523155212,
      "learning_rate": 0.0004651474434924244,
      "loss": 2.0288,
      "step": 72448
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.8135714530944824,
      "learning_rate": 0.00046514402850894766,
      "loss": 2.3144,
      "step": 72449
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.8163759708404541,
      "learning_rate": 0.0004651406134947675,
      "loss": 2.1715,
      "step": 72450
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.8156546950340271,
      "learning_rate": 0.0004651371984498845,
      "loss": 2.1218,
      "step": 72451
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7236968278884888,
      "learning_rate": 0.0004651337833742995,
      "loss": 2.0457,
      "step": 72452
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.9720273017883301,
      "learning_rate": 0.00046513036826801305,
      "loss": 2.1639,
      "step": 72453
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.8448486328125,
      "learning_rate": 0.00046512695313102566,
      "loss": 2.0926,
      "step": 72454
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7502511739730835,
      "learning_rate": 0.00046512353796333806,
      "loss": 2.5445,
      "step": 72455
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.8316687345504761,
      "learning_rate": 0.0004651201227649509,
      "loss": 2.0806,
      "step": 72456
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.9243011474609375,
      "learning_rate": 0.00046511670753586476,
      "loss": 2.2128,
      "step": 72457
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7537307143211365,
      "learning_rate": 0.00046511329227608036,
      "loss": 2.0913,
      "step": 72458
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.8573189973831177,
      "learning_rate": 0.00046510987698559825,
      "loss": 2.088,
      "step": 72459
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.8318885564804077,
      "learning_rate": 0.000465106461664419,
      "loss": 2.1845,
      "step": 72460
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7655619382858276,
      "learning_rate": 0.00046510304631254356,
      "loss": 2.2632,
      "step": 72461
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7881672978401184,
      "learning_rate": 0.0004650996309299722,
      "loss": 2.2279,
      "step": 72462
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.8379685878753662,
      "learning_rate": 0.0004650962155167057,
      "loss": 2.2083,
      "step": 72463
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7346916198730469,
      "learning_rate": 0.0004650928000727448,
      "loss": 2.4715,
      "step": 72464
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.8342112898826599,
      "learning_rate": 0.00046508938459808994,
      "loss": 2.3752,
      "step": 72465
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.8954318165779114,
      "learning_rate": 0.00046508596909274183,
      "loss": 2.3686,
      "step": 72466
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7610809206962585,
      "learning_rate": 0.0004650825535567012,
      "loss": 2.2926,
      "step": 72467
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.825958788394928,
      "learning_rate": 0.00046507913798996854,
      "loss": 2.3004,
      "step": 72468
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7436463832855225,
      "learning_rate": 0.00046507572239254456,
      "loss": 2.2408,
      "step": 72469
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.8608673214912415,
      "learning_rate": 0.00046507230676442996,
      "loss": 1.915,
      "step": 72470
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7381160259246826,
      "learning_rate": 0.00046506889110562523,
      "loss": 2.1276,
      "step": 72471
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7717821002006531,
      "learning_rate": 0.00046506547541613107,
      "loss": 2.2072,
      "step": 72472
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7949864864349365,
      "learning_rate": 0.0004650620596959482,
      "loss": 2.2895,
      "step": 72473
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7451722025871277,
      "learning_rate": 0.0004650586439450771,
      "loss": 2.0111,
      "step": 72474
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7515273690223694,
      "learning_rate": 0.0004650552281635185,
      "loss": 2.3488,
      "step": 72475
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7115975022315979,
      "learning_rate": 0.000465051812351273,
      "loss": 2.1717,
      "step": 72476
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.8002029657363892,
      "learning_rate": 0.0004650483965083413,
      "loss": 2.2847,
      "step": 72477
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.8036383986473083,
      "learning_rate": 0.0004650449806347239,
      "loss": 2.2013,
      "step": 72478
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.8352089524269104,
      "learning_rate": 0.00046504156473042163,
      "loss": 2.5142,
      "step": 72479
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.8045647740364075,
      "learning_rate": 0.00046503814879543493,
      "loss": 2.3176,
      "step": 72480
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.8940383195877075,
      "learning_rate": 0.0004650347328297646,
      "loss": 2.3113,
      "step": 72481
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7891346216201782,
      "learning_rate": 0.00046503131683341115,
      "loss": 2.3681,
      "step": 72482
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.9105530977249146,
      "learning_rate": 0.0004650279008063753,
      "loss": 2.2535,
      "step": 72483
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.8076528906822205,
      "learning_rate": 0.00046502448474865764,
      "loss": 2.1012,
      "step": 72484
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.9500630497932434,
      "learning_rate": 0.0004650210686602587,
      "loss": 2.2921,
      "step": 72485
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.8823046088218689,
      "learning_rate": 0.00046501765254117937,
      "loss": 2.1772,
      "step": 72486
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7901224493980408,
      "learning_rate": 0.0004650142363914201,
      "loss": 2.1555,
      "step": 72487
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.8795098662376404,
      "learning_rate": 0.0004650108202109816,
      "loss": 2.4597,
      "step": 72488
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.87039715051651,
      "learning_rate": 0.00046500740399986446,
      "loss": 2.1662,
      "step": 72489
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7569795846939087,
      "learning_rate": 0.00046500398775806926,
      "loss": 2.1255,
      "step": 72490
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.801696240901947,
      "learning_rate": 0.0004650005714855968,
      "loss": 2.2898,
      "step": 72491
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.9920247197151184,
      "learning_rate": 0.0004649971551824476,
      "loss": 2.3407,
      "step": 72492
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7605591416358948,
      "learning_rate": 0.00046499373884862223,
      "loss": 2.2894,
      "step": 72493
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.6989094614982605,
      "learning_rate": 0.0004649903224841215,
      "loss": 2.1834,
      "step": 72494
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.8000325560569763,
      "learning_rate": 0.00046498690608894587,
      "loss": 2.2148,
      "step": 72495
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.3218201398849487,
      "learning_rate": 0.0004649834896630961,
      "loss": 2.2223,
      "step": 72496
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7107232213020325,
      "learning_rate": 0.00046498007320657284,
      "loss": 2.2889,
      "step": 72497
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.9116699695587158,
      "learning_rate": 0.0004649766567193766,
      "loss": 2.3557,
      "step": 72498
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.353716254234314,
      "learning_rate": 0.00046497324020150813,
      "loss": 2.4141,
      "step": 72499
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.8781578540802002,
      "learning_rate": 0.00046496982365296796,
      "loss": 2.2484,
      "step": 72500
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7815955877304077,
      "learning_rate": 0.0004649664070737569,
      "loss": 2.1512,
      "step": 72501
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.8735501170158386,
      "learning_rate": 0.0004649629904638754,
      "loss": 2.0399,
      "step": 72502
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.2868849039077759,
      "learning_rate": 0.00046495957382332407,
      "loss": 2.0361,
      "step": 72503
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.8855425119400024,
      "learning_rate": 0.0004649561571521038,
      "loss": 2.3721,
      "step": 72504
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7770078182220459,
      "learning_rate": 0.000464952740450215,
      "loss": 2.2384,
      "step": 72505
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.8657190203666687,
      "learning_rate": 0.0004649493237176584,
      "loss": 2.0967,
      "step": 72506
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.8376259803771973,
      "learning_rate": 0.00046494590695443454,
      "loss": 2.2753,
      "step": 72507
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.0482401847839355,
      "learning_rate": 0.0004649424901605442,
      "loss": 2.2642,
      "step": 72508
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.9198048710823059,
      "learning_rate": 0.0004649390733359879,
      "loss": 2.1208,
      "step": 72509
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.028361439704895,
      "learning_rate": 0.0004649356564807663,
      "loss": 2.3289,
      "step": 72510
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.6851271986961365,
      "learning_rate": 0.00046493223959488006,
      "loss": 2.0822,
      "step": 72511
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7840589880943298,
      "learning_rate": 0.00046492882267832984,
      "loss": 2.5713,
      "step": 72512
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.8009787797927856,
      "learning_rate": 0.0004649254057311163,
      "loss": 2.2106,
      "step": 72513
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7638780474662781,
      "learning_rate": 0.0004649219887532398,
      "loss": 2.3381,
      "step": 72514
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.083662509918213,
      "learning_rate": 0.00046491857174470126,
      "loss": 2.154,
      "step": 72515
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7867418527603149,
      "learning_rate": 0.0004649151547055014,
      "loss": 2.374,
      "step": 72516
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.8406891822814941,
      "learning_rate": 0.00046491173763564054,
      "loss": 1.935,
      "step": 72517
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.0360403060913086,
      "learning_rate": 0.0004649083205351195,
      "loss": 2.2573,
      "step": 72518
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.859948456287384,
      "learning_rate": 0.00046490490340393893,
      "loss": 2.1697,
      "step": 72519
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.9514926671981812,
      "learning_rate": 0.00046490148624209947,
      "loss": 2.3557,
      "step": 72520
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.8229230046272278,
      "learning_rate": 0.0004648980690496016,
      "loss": 2.3901,
      "step": 72521
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.8738908767700195,
      "learning_rate": 0.00046489465182644616,
      "loss": 2.4546,
      "step": 72522
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7887104749679565,
      "learning_rate": 0.00046489123457263366,
      "loss": 2.3558,
      "step": 72523
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7528956532478333,
      "learning_rate": 0.0004648878172881647,
      "loss": 2.1064,
      "step": 72524
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.8944608569145203,
      "learning_rate": 0.0004648843999730401,
      "loss": 2.3634,
      "step": 72525
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7966116666793823,
      "learning_rate": 0.0004648809826272603,
      "loss": 2.1701,
      "step": 72526
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7848861217498779,
      "learning_rate": 0.00046487756525082606,
      "loss": 2.335,
      "step": 72527
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.9069133996963501,
      "learning_rate": 0.00046487414784373795,
      "loss": 2.326,
      "step": 72528
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.8288862109184265,
      "learning_rate": 0.0004648707304059965,
      "loss": 2.1273,
      "step": 72529
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7046199440956116,
      "learning_rate": 0.0004648673129376026,
      "loss": 2.3544,
      "step": 72530
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7570754885673523,
      "learning_rate": 0.0004648638954385568,
      "loss": 2.0549,
      "step": 72531
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.8148162364959717,
      "learning_rate": 0.0004648604779088596,
      "loss": 2.1883,
      "step": 72532
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.822404146194458,
      "learning_rate": 0.0004648570603485118,
      "loss": 2.0308,
      "step": 72533
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.8213332295417786,
      "learning_rate": 0.00046485364275751387,
      "loss": 2.3321,
      "step": 72534
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7258232831954956,
      "learning_rate": 0.0004648502251358666,
      "loss": 2.295,
      "step": 72535
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7518619298934937,
      "learning_rate": 0.0004648468074835705,
      "loss": 2.2573,
      "step": 72536
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7493920922279358,
      "learning_rate": 0.0004648433898006264,
      "loss": 2.1658,
      "step": 72537
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.8278387784957886,
      "learning_rate": 0.00046483997208703466,
      "loss": 2.3427,
      "step": 72538
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7806196212768555,
      "learning_rate": 0.0004648365543427961,
      "loss": 2.0705,
      "step": 72539
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7530981302261353,
      "learning_rate": 0.00046483313656791133,
      "loss": 2.1244,
      "step": 72540
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.8273419141769409,
      "learning_rate": 0.00046482971876238093,
      "loss": 2.1502,
      "step": 72541
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7528178095817566,
      "learning_rate": 0.0004648263009262056,
      "loss": 2.1628,
      "step": 72542
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.880430281162262,
      "learning_rate": 0.00046482288305938597,
      "loss": 2.305,
      "step": 72543
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.764528751373291,
      "learning_rate": 0.00046481946516192265,
      "loss": 2.229,
      "step": 72544
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.8502234220504761,
      "learning_rate": 0.0004648160472338162,
      "loss": 2.2418,
      "step": 72545
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7833420038223267,
      "learning_rate": 0.0004648126292750675,
      "loss": 2.325,
      "step": 72546
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.827113926410675,
      "learning_rate": 0.0004648092112856769,
      "loss": 2.2573,
      "step": 72547
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.125405192375183,
      "learning_rate": 0.0004648057932656451,
      "loss": 2.1209,
      "step": 72548
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.8159023523330688,
      "learning_rate": 0.0004648023752149729,
      "loss": 2.1435,
      "step": 72549
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.921106219291687,
      "learning_rate": 0.00046479895713366084,
      "loss": 2.3016,
      "step": 72550
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7874011397361755,
      "learning_rate": 0.0004647955390217094,
      "loss": 2.1143,
      "step": 72551
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7467309236526489,
      "learning_rate": 0.00046479212087911954,
      "loss": 2.2328,
      "step": 72552
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7929931879043579,
      "learning_rate": 0.0004647887027058917,
      "loss": 2.0508,
      "step": 72553
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.8740759491920471,
      "learning_rate": 0.00046478528450202643,
      "loss": 1.9898,
      "step": 72554
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7530280947685242,
      "learning_rate": 0.0004647818662675245,
      "loss": 2.258,
      "step": 72555
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.8520378470420837,
      "learning_rate": 0.0004647784480023865,
      "loss": 2.4106,
      "step": 72556
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.8044115304946899,
      "learning_rate": 0.00046477502970661305,
      "loss": 2.0779,
      "step": 72557
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7694820761680603,
      "learning_rate": 0.00046477161138020487,
      "loss": 2.2942,
      "step": 72558
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.0320038795471191,
      "learning_rate": 0.0004647681930231625,
      "loss": 2.1502,
      "step": 72559
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7873669266700745,
      "learning_rate": 0.00046476477463548667,
      "loss": 2.2353,
      "step": 72560
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.8429595828056335,
      "learning_rate": 0.00046476135621717794,
      "loss": 2.521,
      "step": 72561
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7916507124900818,
      "learning_rate": 0.0004647579377682369,
      "loss": 2.0385,
      "step": 72562
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.854496955871582,
      "learning_rate": 0.0004647545192886642,
      "loss": 2.0727,
      "step": 72563
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.693223237991333,
      "learning_rate": 0.0004647511007784607,
      "loss": 2.2771,
      "step": 72564
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7994121313095093,
      "learning_rate": 0.0004647476822376268,
      "loss": 2.3357,
      "step": 72565
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.68867427110672,
      "learning_rate": 0.00046474426366616313,
      "loss": 1.8856,
      "step": 72566
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.940998375415802,
      "learning_rate": 0.00046474084506407037,
      "loss": 2.0891,
      "step": 72567
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7544867396354675,
      "learning_rate": 0.00046473742643134933,
      "loss": 2.2104,
      "step": 72568
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.8871341943740845,
      "learning_rate": 0.0004647340077680004,
      "loss": 2.1185,
      "step": 72569
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.0388875007629395,
      "learning_rate": 0.0004647305890740243,
      "loss": 2.3553,
      "step": 72570
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.8345754146575928,
      "learning_rate": 0.0004647271703494217,
      "loss": 2.1822,
      "step": 72571
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.9405566453933716,
      "learning_rate": 0.0004647237515941932,
      "loss": 2.2354,
      "step": 72572
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.8636261224746704,
      "learning_rate": 0.0004647203328083394,
      "loss": 2.2444,
      "step": 72573
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.8158081769943237,
      "learning_rate": 0.00046471691399186104,
      "loss": 2.229,
      "step": 72574
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7552204728126526,
      "learning_rate": 0.0004647134951447587,
      "loss": 2.1351,
      "step": 72575
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.0484265089035034,
      "learning_rate": 0.00046471007626703296,
      "loss": 2.2267,
      "step": 72576
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.290993094444275,
      "learning_rate": 0.00046470665735868453,
      "loss": 2.0827,
      "step": 72577
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.7566167116165161,
      "learning_rate": 0.00046470323841971405,
      "loss": 2.1441,
      "step": 72578
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.9002784490585327,
      "learning_rate": 0.0004646998194501221,
      "loss": 2.2636,
      "step": 72579
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.9174522757530212,
      "learning_rate": 0.00046469640044990935,
      "loss": 2.0397,
      "step": 72580
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.9370411038398743,
      "learning_rate": 0.00046469298141907643,
      "loss": 2.0768,
      "step": 72581
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.7729990482330322,
      "learning_rate": 0.0004646895623576239,
      "loss": 2.1336,
      "step": 72582
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.7289678454399109,
      "learning_rate": 0.0004646861432655526,
      "loss": 2.464,
      "step": 72583
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.77439284324646,
      "learning_rate": 0.000464682724142863,
      "loss": 2.429,
      "step": 72584
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.7314103245735168,
      "learning_rate": 0.00046467930498955576,
      "loss": 2.0149,
      "step": 72585
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.7359350323677063,
      "learning_rate": 0.00046467588580563154,
      "loss": 2.337,
      "step": 72586
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.7395395636558533,
      "learning_rate": 0.00046467246659109097,
      "loss": 2.0145,
      "step": 72587
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.7540379762649536,
      "learning_rate": 0.00046466904734593465,
      "loss": 2.5981,
      "step": 72588
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.7703223824501038,
      "learning_rate": 0.0004646656280701633,
      "loss": 2.1103,
      "step": 72589
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.7776120901107788,
      "learning_rate": 0.00046466220876377744,
      "loss": 2.2215,
      "step": 72590
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.8209280967712402,
      "learning_rate": 0.00046465878942677775,
      "loss": 2.2596,
      "step": 72591
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.7673326730728149,
      "learning_rate": 0.00046465537005916497,
      "loss": 2.1454,
      "step": 72592
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.7618445754051208,
      "learning_rate": 0.00046465195066093957,
      "loss": 2.0805,
      "step": 72593
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.7399649620056152,
      "learning_rate": 0.0004646485312321023,
      "loss": 2.1625,
      "step": 72594
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.7101507186889648,
      "learning_rate": 0.00046464511177265376,
      "loss": 2.1932,
      "step": 72595
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.8943051695823669,
      "learning_rate": 0.00046464169228259454,
      "loss": 2.22,
      "step": 72596
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.8365468978881836,
      "learning_rate": 0.00046463827276192537,
      "loss": 2.1919,
      "step": 72597
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.8262515068054199,
      "learning_rate": 0.00046463485321064684,
      "loss": 2.3205,
      "step": 72598
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.7644268870353699,
      "learning_rate": 0.0004646314336287596,
      "loss": 2.3176,
      "step": 72599
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.7767800092697144,
      "learning_rate": 0.0004646280140162642,
      "loss": 2.2852,
      "step": 72600
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.7107194662094116,
      "learning_rate": 0.00046462459437316144,
      "loss": 2.4228,
      "step": 72601
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.8881351947784424,
      "learning_rate": 0.0004646211746994518,
      "loss": 2.146,
      "step": 72602
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.7623499035835266,
      "learning_rate": 0.0004646177549951359,
      "loss": 2.3302,
      "step": 72603
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.7626333236694336,
      "learning_rate": 0.00046461433526021463,
      "loss": 2.3964,
      "step": 72604
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.7829813957214355,
      "learning_rate": 0.0004646109154946883,
      "loss": 2.1723,
      "step": 72605
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.7971854209899902,
      "learning_rate": 0.0004646074956985578,
      "loss": 2.2619,
      "step": 72606
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.797482430934906,
      "learning_rate": 0.0004646040758718236,
      "loss": 2.1533,
      "step": 72607
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.7652517557144165,
      "learning_rate": 0.0004646006560144864,
      "loss": 2.2377,
      "step": 72608
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.7899474501609802,
      "learning_rate": 0.0004645972361265468,
      "loss": 2.2642,
      "step": 72609
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.7402288317680359,
      "learning_rate": 0.0004645938162080055,
      "loss": 2.0417,
      "step": 72610
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.7846461534500122,
      "learning_rate": 0.0004645903962588631,
      "loss": 2.2111,
      "step": 72611
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.8287490606307983,
      "learning_rate": 0.0004645869762791202,
      "loss": 2.2387,
      "step": 72612
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.8053181171417236,
      "learning_rate": 0.00046458355626877753,
      "loss": 2.4957,
      "step": 72613
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.7401235103607178,
      "learning_rate": 0.0004645801362278356,
      "loss": 2.29,
      "step": 72614
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.8256405591964722,
      "learning_rate": 0.00046457671615629526,
      "loss": 2.22,
      "step": 72615
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.8594950437545776,
      "learning_rate": 0.0004645732960541569,
      "loss": 2.0845,
      "step": 72616
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.8934280872344971,
      "learning_rate": 0.0004645698759214213,
      "loss": 2.1754,
      "step": 72617
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.7391238808631897,
      "learning_rate": 0.000464566455758089,
      "loss": 1.9664,
      "step": 72618
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.7233579158782959,
      "learning_rate": 0.00046456303556416064,
      "loss": 2.3202,
      "step": 72619
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.8045787811279297,
      "learning_rate": 0.00046455961533963707,
      "loss": 2.2116,
      "step": 72620
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.740595281124115,
      "learning_rate": 0.0004645561950845186,
      "loss": 2.2529,
      "step": 72621
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.7782968878746033,
      "learning_rate": 0.00046455277479880613,
      "loss": 2.2128,
      "step": 72622
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.787749171257019,
      "learning_rate": 0.0004645493544825002,
      "loss": 2.2938,
      "step": 72623
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.7312566637992859,
      "learning_rate": 0.0004645459341356013,
      "loss": 2.1929,
      "step": 72624
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.7225466370582581,
      "learning_rate": 0.00046454251375811035,
      "loss": 2.3626,
      "step": 72625
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.8895851373672485,
      "learning_rate": 0.0004645390933500278,
      "loss": 1.9772,
      "step": 72626
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.903066873550415,
      "learning_rate": 0.0004645356729113543,
      "loss": 2.2451,
      "step": 72627
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.7709351778030396,
      "learning_rate": 0.0004645322524420906,
      "loss": 2.17,
      "step": 72628
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.8930262327194214,
      "learning_rate": 0.00046452883194223713,
      "loss": 2.196,
      "step": 72629
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.7524371147155762,
      "learning_rate": 0.00046452541141179463,
      "loss": 2.4235,
      "step": 72630
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.9380256533622742,
      "learning_rate": 0.0004645219908507638,
      "loss": 2.3209,
      "step": 72631
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.7908179759979248,
      "learning_rate": 0.0004645185702591453,
      "loss": 2.3446,
      "step": 72632
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.8109312057495117,
      "learning_rate": 0.00046451514963693965,
      "loss": 2.0875,
      "step": 72633
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.9876080751419067,
      "learning_rate": 0.0004645117289841475,
      "loss": 2.0874,
      "step": 72634
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.7828947901725769,
      "learning_rate": 0.00046450830830076955,
      "loss": 2.1543,
      "step": 72635
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.8150717616081238,
      "learning_rate": 0.0004645048875868063,
      "loss": 2.1137,
      "step": 72636
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.8260253667831421,
      "learning_rate": 0.00046450146684225857,
      "loss": 2.0896,
      "step": 72637
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.7532151937484741,
      "learning_rate": 0.000464498046067127,
      "loss": 2.2657,
      "step": 72638
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.8047839403152466,
      "learning_rate": 0.000464494625261412,
      "loss": 2.3267,
      "step": 72639
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.7134133577346802,
      "learning_rate": 0.00046449120442511435,
      "loss": 2.1964,
      "step": 72640
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.7724513411521912,
      "learning_rate": 0.00046448778355823476,
      "loss": 2.211,
      "step": 72641
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.7322047352790833,
      "learning_rate": 0.00046448436266077373,
      "loss": 2.0251,
      "step": 72642
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.7012904286384583,
      "learning_rate": 0.00046448094173273197,
      "loss": 2.1115,
      "step": 72643
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.8366860151290894,
      "learning_rate": 0.0004644775207741101,
      "loss": 2.3606,
      "step": 72644
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.762698769569397,
      "learning_rate": 0.00046447409978490873,
      "loss": 2.1968,
      "step": 72645
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.8504198789596558,
      "learning_rate": 0.00046447067876512855,
      "loss": 2.3272,
      "step": 72646
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.6842115521430969,
      "learning_rate": 0.0004644672577147702,
      "loss": 2.1771,
      "step": 72647
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.8100424408912659,
      "learning_rate": 0.0004644638366338342,
      "loss": 2.2735,
      "step": 72648
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.7955703735351562,
      "learning_rate": 0.00046446041552232134,
      "loss": 2.3318,
      "step": 72649
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.989073634147644,
      "learning_rate": 0.0004644569943802321,
      "loss": 2.1721,
      "step": 72650
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.7149428725242615,
      "learning_rate": 0.0004644535732075672,
      "loss": 2.2472,
      "step": 72651
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.8013163805007935,
      "learning_rate": 0.00046445015200432735,
      "loss": 2.1264,
      "step": 72652
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.8737453818321228,
      "learning_rate": 0.0004644467307705132,
      "loss": 2.3105,
      "step": 72653
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.789555549621582,
      "learning_rate": 0.00046444330950612514,
      "loss": 2.2661,
      "step": 72654
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.8377499580383301,
      "learning_rate": 0.00046443988821116394,
      "loss": 2.0924,
      "step": 72655
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.7965318560600281,
      "learning_rate": 0.00046443646688563035,
      "loss": 2.241,
      "step": 72656
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.8200043439865112,
      "learning_rate": 0.00046443304552952486,
      "loss": 2.2592,
      "step": 72657
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.7891167998313904,
      "learning_rate": 0.0004644296241428482,
      "loss": 2.1223,
      "step": 72658
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.9362871646881104,
      "learning_rate": 0.000464426202725601,
      "loss": 2.2407,
      "step": 72659
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.7771075963973999,
      "learning_rate": 0.0004644227812777839,
      "loss": 2.0084,
      "step": 72660
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.7710344195365906,
      "learning_rate": 0.00046441935979939735,
      "loss": 2.3113,
      "step": 72661
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.8334192037582397,
      "learning_rate": 0.00046441593829044223,
      "loss": 2.1894,
      "step": 72662
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.8492308259010315,
      "learning_rate": 0.0004644125167509191,
      "loss": 2.0716,
      "step": 72663
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.7767390012741089,
      "learning_rate": 0.0004644090951808286,
      "loss": 2.2294,
      "step": 72664
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.756864070892334,
      "learning_rate": 0.0004644056735801712,
      "loss": 1.8253,
      "step": 72665
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.7712651491165161,
      "learning_rate": 0.0004644022519489478,
      "loss": 2.2453,
      "step": 72666
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.8567323684692383,
      "learning_rate": 0.00046439883028715894,
      "loss": 2.2436,
      "step": 72667
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.7983084917068481,
      "learning_rate": 0.0004643954085948053,
      "loss": 2.2611,
      "step": 72668
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.8103829026222229,
      "learning_rate": 0.00046439198687188725,
      "loss": 2.3402,
      "step": 72669
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.7483493089675903,
      "learning_rate": 0.0004643885651184057,
      "loss": 2.0222,
      "step": 72670
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.7390536069869995,
      "learning_rate": 0.0004643851433343613,
      "loss": 2.1233,
      "step": 72671
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.7487979531288147,
      "learning_rate": 0.00046438172151975446,
      "loss": 2.3562,
      "step": 72672
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.822964072227478,
      "learning_rate": 0.00046437829967458603,
      "loss": 2.0119,
      "step": 72673
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.7790555357933044,
      "learning_rate": 0.0004643748777988566,
      "loss": 2.3621,
      "step": 72674
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.6835823059082031,
      "learning_rate": 0.00046437145589256674,
      "loss": 2.1945,
      "step": 72675
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.7167946100234985,
      "learning_rate": 0.00046436803395571714,
      "loss": 2.0978,
      "step": 72676
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.7422547936439514,
      "learning_rate": 0.00046436461198830843,
      "loss": 2.5001,
      "step": 72677
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.7507866621017456,
      "learning_rate": 0.00046436118999034116,
      "loss": 2.1909,
      "step": 72678
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.7157936692237854,
      "learning_rate": 0.0004643577679618161,
      "loss": 2.3295,
      "step": 72679
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.7671462297439575,
      "learning_rate": 0.00046435434590273386,
      "loss": 2.302,
      "step": 72680
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.7528343796730042,
      "learning_rate": 0.000464350923813095,
      "loss": 2.204,
      "step": 72681
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.7386232018470764,
      "learning_rate": 0.00046434750169290027,
      "loss": 2.1759,
      "step": 72682
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.7773209810256958,
      "learning_rate": 0.0004643440795421502,
      "loss": 2.1767,
      "step": 72683
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.7518998384475708,
      "learning_rate": 0.0004643406573608454,
      "loss": 2.317,
      "step": 72684
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.721291720867157,
      "learning_rate": 0.00046433723514898665,
      "loss": 1.9841,
      "step": 72685
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.8402929902076721,
      "learning_rate": 0.0004643338129065745,
      "loss": 2.2713,
      "step": 72686
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.7296922206878662,
      "learning_rate": 0.0004643303906336095,
      "loss": 1.9532,
      "step": 72687
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.8402258157730103,
      "learning_rate": 0.0004643269683300924,
      "loss": 2.2735,
      "step": 72688
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.7730361819267273,
      "learning_rate": 0.00046432354599602394,
      "loss": 2.2018,
      "step": 72689
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.7340688109397888,
      "learning_rate": 0.00046432012363140454,
      "loss": 2.1436,
      "step": 72690
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.7435908913612366,
      "learning_rate": 0.0004643167012362349,
      "loss": 2.2417,
      "step": 72691
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.7578758001327515,
      "learning_rate": 0.00046431327881051574,
      "loss": 2.2769,
      "step": 72692
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.7667621970176697,
      "learning_rate": 0.00046430985635424765,
      "loss": 2.1484,
      "step": 72693
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.7085278034210205,
      "learning_rate": 0.0004643064338674312,
      "loss": 2.0611,
      "step": 72694
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.9753749966621399,
      "learning_rate": 0.0004643030113500671,
      "loss": 2.3712,
      "step": 72695
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.751173198223114,
      "learning_rate": 0.00046429958880215595,
      "loss": 2.0945,
      "step": 72696
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.7726941108703613,
      "learning_rate": 0.00046429616622369844,
      "loss": 2.1675,
      "step": 72697
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.7438653111457825,
      "learning_rate": 0.00046429274361469515,
      "loss": 2.2772,
      "step": 72698
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.7784177660942078,
      "learning_rate": 0.0004642893209751468,
      "loss": 2.091,
      "step": 72699
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.7531324028968811,
      "learning_rate": 0.0004642858983050538,
      "loss": 2.0751,
      "step": 72700
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.8505839109420776,
      "learning_rate": 0.0004642824756044171,
      "loss": 2.263,
      "step": 72701
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.7786633372306824,
      "learning_rate": 0.00046427905287323715,
      "loss": 2.0003,
      "step": 72702
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.822845995426178,
      "learning_rate": 0.00046427563011151467,
      "loss": 2.1542,
      "step": 72703
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.8700094223022461,
      "learning_rate": 0.00046427220731925015,
      "loss": 2.2896,
      "step": 72704
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.7369875311851501,
      "learning_rate": 0.00046426878449644436,
      "loss": 2.0931,
      "step": 72705
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.0590132474899292,
      "learning_rate": 0.000464265361643098,
      "loss": 2.2929,
      "step": 72706
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.8074789643287659,
      "learning_rate": 0.00046426193875921147,
      "loss": 2.2246,
      "step": 72707
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.7723020911216736,
      "learning_rate": 0.00046425851584478567,
      "loss": 1.9732,
      "step": 72708
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.9060260057449341,
      "learning_rate": 0.000464255092899821,
      "loss": 2.3773,
      "step": 72709
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.7823256850242615,
      "learning_rate": 0.0004642516699243182,
      "loss": 2.2986,
      "step": 72710
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.8002750873565674,
      "learning_rate": 0.000464248246918278,
      "loss": 2.3779,
      "step": 72711
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.749646008014679,
      "learning_rate": 0.0004642448238817009,
      "loss": 2.1486,
      "step": 72712
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.7343958616256714,
      "learning_rate": 0.00046424140081458763,
      "loss": 2.2095,
      "step": 72713
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.7407013177871704,
      "learning_rate": 0.00046423797771693876,
      "loss": 1.8991,
      "step": 72714
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.7993975877761841,
      "learning_rate": 0.0004642345545887549,
      "loss": 2.2291,
      "step": 72715
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.7138799428939819,
      "learning_rate": 0.0004642311314300368,
      "loss": 2.4449,
      "step": 72716
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.7193382382392883,
      "learning_rate": 0.0004642277082407851,
      "loss": 2.3613,
      "step": 72717
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.8485941290855408,
      "learning_rate": 0.0004642242850210003,
      "loss": 2.2717,
      "step": 72718
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.8665745854377747,
      "learning_rate": 0.000464220861770683,
      "loss": 2.2016,
      "step": 72719
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.7430359125137329,
      "learning_rate": 0.0004642174384898341,
      "loss": 2.2601,
      "step": 72720
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.8223675489425659,
      "learning_rate": 0.000464214015178454,
      "loss": 2.2279,
      "step": 72721
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.7904258966445923,
      "learning_rate": 0.00046421059183654337,
      "loss": 2.184,
      "step": 72722
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.755221962928772,
      "learning_rate": 0.000464207168464103,
      "loss": 2.3235,
      "step": 72723
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.8769258260726929,
      "learning_rate": 0.00046420374506113334,
      "loss": 2.3704,
      "step": 72724
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.8754869103431702,
      "learning_rate": 0.0004642003216276351,
      "loss": 2.1977,
      "step": 72725
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.8135853409767151,
      "learning_rate": 0.00046419689816360897,
      "loss": 2.2254,
      "step": 72726
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.1252025365829468,
      "learning_rate": 0.00046419347466905553,
      "loss": 2.1974,
      "step": 72727
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.762344241142273,
      "learning_rate": 0.0004641900511439754,
      "loss": 2.3648,
      "step": 72728
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.7677159309387207,
      "learning_rate": 0.0004641866275883693,
      "loss": 2.3007,
      "step": 72729
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.7979857325553894,
      "learning_rate": 0.0004641832040022377,
      "loss": 2.1979,
      "step": 72730
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.9033545255661011,
      "learning_rate": 0.0004641797803855814,
      "loss": 2.3526,
      "step": 72731
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.8058859705924988,
      "learning_rate": 0.00046417635673840104,
      "loss": 2.2393,
      "step": 72732
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.768351674079895,
      "learning_rate": 0.00046417293306069707,
      "loss": 2.1849,
      "step": 72733
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.1177016496658325,
      "learning_rate": 0.0004641695093524703,
      "loss": 2.3242,
      "step": 72734
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.7125799655914307,
      "learning_rate": 0.0004641660856137214,
      "loss": 2.0428,
      "step": 72735
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.7194627523422241,
      "learning_rate": 0.00046416266184445083,
      "loss": 2.2895,
      "step": 72736
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.8843749165534973,
      "learning_rate": 0.0004641592380446593,
      "loss": 2.5064,
      "step": 72737
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.8730132579803467,
      "learning_rate": 0.00046415581421434764,
      "loss": 2.3976,
      "step": 72738
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.7302972674369812,
      "learning_rate": 0.0004641523903535161,
      "loss": 2.1951,
      "step": 72739
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.7450282573699951,
      "learning_rate": 0.00046414896646216567,
      "loss": 2.376,
      "step": 72740
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.7629125118255615,
      "learning_rate": 0.0004641455425402969,
      "loss": 2.1917,
      "step": 72741
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.8170020580291748,
      "learning_rate": 0.00046414211858791023,
      "loss": 2.5441,
      "step": 72742
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.0583733320236206,
      "learning_rate": 0.00046413869460500644,
      "loss": 2.2721,
      "step": 72743
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.7535322904586792,
      "learning_rate": 0.0004641352705915863,
      "loss": 2.2125,
      "step": 72744
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.8602720499038696,
      "learning_rate": 0.0004641318465476502,
      "loss": 2.1085,
      "step": 72745
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.9168147444725037,
      "learning_rate": 0.00046412842247319896,
      "loss": 2.1981,
      "step": 72746
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.9379705786705017,
      "learning_rate": 0.00046412499836823314,
      "loss": 2.0056,
      "step": 72747
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.7500388622283936,
      "learning_rate": 0.00046412157423275337,
      "loss": 2.2329,
      "step": 72748
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.1701011657714844,
      "learning_rate": 0.0004641181500667602,
      "loss": 2.0929,
      "step": 72749
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.760877788066864,
      "learning_rate": 0.00046411472587025454,
      "loss": 2.1058,
      "step": 72750
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.8787038326263428,
      "learning_rate": 0.0004641113016432367,
      "loss": 2.0955,
      "step": 72751
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.8165819644927979,
      "learning_rate": 0.0004641078773857076,
      "loss": 2.1121,
      "step": 72752
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.8522832989692688,
      "learning_rate": 0.0004641044530976677,
      "loss": 2.2109,
      "step": 72753
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.873232364654541,
      "learning_rate": 0.00046410102877911773,
      "loss": 2.0311,
      "step": 72754
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.7358068227767944,
      "learning_rate": 0.0004640976044300581,
      "loss": 2.24,
      "step": 72755
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.83333820104599,
      "learning_rate": 0.0004640941800504899,
      "loss": 2.3575,
      "step": 72756
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.8805646300315857,
      "learning_rate": 0.0004640907556404133,
      "loss": 2.1688,
      "step": 72757
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.8026355504989624,
      "learning_rate": 0.0004640873311998292,
      "loss": 2.0258,
      "step": 72758
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.7704986929893494,
      "learning_rate": 0.00046408390672873816,
      "loss": 2.3881,
      "step": 72759
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.9145684242248535,
      "learning_rate": 0.00046408048222714083,
      "loss": 2.2536,
      "step": 72760
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.0724149942398071,
      "learning_rate": 0.0004640770576950378,
      "loss": 2.31,
      "step": 72761
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.7355631589889526,
      "learning_rate": 0.0004640736331324298,
      "loss": 2.2557,
      "step": 72762
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.7617429494857788,
      "learning_rate": 0.00046407020853931735,
      "loss": 2.2313,
      "step": 72763
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.9889493584632874,
      "learning_rate": 0.0004640667839157012,
      "loss": 2.0594,
      "step": 72764
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.8592816591262817,
      "learning_rate": 0.00046406335926158193,
      "loss": 2.3111,
      "step": 72765
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.7786646485328674,
      "learning_rate": 0.0004640599345769602,
      "loss": 2.3191,
      "step": 72766
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.9244505763053894,
      "learning_rate": 0.0004640565098618366,
      "loss": 2.3955,
      "step": 72767
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.8321834206581116,
      "learning_rate": 0.0004640530851162118,
      "loss": 2.2153,
      "step": 72768
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.7666512727737427,
      "learning_rate": 0.00046404966034008655,
      "loss": 2.2409,
      "step": 72769
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.8409931659698486,
      "learning_rate": 0.0004640462355334612,
      "loss": 2.2663,
      "step": 72770
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.8917661905288696,
      "learning_rate": 0.0004640428106963366,
      "loss": 2.2377,
      "step": 72771
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.8250394463539124,
      "learning_rate": 0.00046403938582871345,
      "loss": 2.3277,
      "step": 72772
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.7807167768478394,
      "learning_rate": 0.00046403596093059216,
      "loss": 2.1323,
      "step": 72773
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.845597505569458,
      "learning_rate": 0.00046403253600197356,
      "loss": 2.2224,
      "step": 72774
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.0350006818771362,
      "learning_rate": 0.0004640291110428582,
      "loss": 2.2413,
      "step": 72775
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.7413280010223389,
      "learning_rate": 0.00046402568605324667,
      "loss": 2.2301,
      "step": 72776
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.8736182451248169,
      "learning_rate": 0.00046402226103313973,
      "loss": 2.1405,
      "step": 72777
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.9533293843269348,
      "learning_rate": 0.000464018835982538,
      "loss": 2.214,
      "step": 72778
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.9275262355804443,
      "learning_rate": 0.00046401541090144196,
      "loss": 2.1204,
      "step": 72779
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.9576216340065002,
      "learning_rate": 0.00046401198578985243,
      "loss": 2.1533,
      "step": 72780
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.8495107889175415,
      "learning_rate": 0.00046400856064777004,
      "loss": 2.0494,
      "step": 72781
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.8969660401344299,
      "learning_rate": 0.00046400513547519527,
      "loss": 2.1979,
      "step": 72782
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.7809061408042908,
      "learning_rate": 0.0004640017102721288,
      "loss": 2.323,
      "step": 72783
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.860985517501831,
      "learning_rate": 0.00046399828503857144,
      "loss": 2.0902,
      "step": 72784
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.8187127709388733,
      "learning_rate": 0.0004639948597745236,
      "loss": 2.1817,
      "step": 72785
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.9324709177017212,
      "learning_rate": 0.0004639914344799861,
      "loss": 2.3344,
      "step": 72786
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.8536074757575989,
      "learning_rate": 0.0004639880091549595,
      "loss": 2.0395,
      "step": 72787
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.7851385474205017,
      "learning_rate": 0.0004639845837994443,
      "loss": 2.3453,
      "step": 72788
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.9621000289916992,
      "learning_rate": 0.0004639811584134414,
      "loss": 2.3179,
      "step": 72789
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.8447161316871643,
      "learning_rate": 0.00046397773299695124,
      "loss": 2.1518,
      "step": 72790
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.713284969329834,
      "learning_rate": 0.0004639743075499746,
      "loss": 2.3356,
      "step": 72791
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.8806899785995483,
      "learning_rate": 0.000463970882072512,
      "loss": 2.2466,
      "step": 72792
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.8229818344116211,
      "learning_rate": 0.00046396745656456407,
      "loss": 1.9979,
      "step": 72793
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.7891585230827332,
      "learning_rate": 0.00046396403102613157,
      "loss": 1.9849,
      "step": 72794
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.7609657049179077,
      "learning_rate": 0.00046396060545721504,
      "loss": 2.1763,
      "step": 72795
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.7932345867156982,
      "learning_rate": 0.0004639571798578152,
      "loss": 2.1035,
      "step": 72796
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.8607833981513977,
      "learning_rate": 0.00046395375422793246,
      "loss": 2.1796,
      "step": 72797
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.9055583477020264,
      "learning_rate": 0.00046395032856756777,
      "loss": 2.3369,
      "step": 72798
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.8958671689033508,
      "learning_rate": 0.0004639469028767216,
      "loss": 2.3844,
      "step": 72799
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.0595084428787231,
      "learning_rate": 0.0004639434771553945,
      "loss": 2.1389,
      "step": 72800
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.8062659502029419,
      "learning_rate": 0.00046394005140358726,
      "loss": 2.4572,
      "step": 72801
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.8283886313438416,
      "learning_rate": 0.00046393662562130056,
      "loss": 2.3739,
      "step": 72802
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.9193187952041626,
      "learning_rate": 0.0004639331998085349,
      "loss": 2.4737,
      "step": 72803
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.8363604545593262,
      "learning_rate": 0.00046392977396529086,
      "loss": 2.038,
      "step": 72804
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.7367744445800781,
      "learning_rate": 0.0004639263480915694,
      "loss": 2.2467,
      "step": 72805
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.8444755673408508,
      "learning_rate": 0.0004639229221873707,
      "loss": 2.193,
      "step": 72806
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.9570201635360718,
      "learning_rate": 0.00046391949625269576,
      "loss": 2.3301,
      "step": 72807
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.8429615497589111,
      "learning_rate": 0.00046391607028754513,
      "loss": 2.1691,
      "step": 72808
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.8490508198738098,
      "learning_rate": 0.00046391264429191927,
      "loss": 2.393,
      "step": 72809
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.9394180774688721,
      "learning_rate": 0.000463909218265819,
      "loss": 2.0134,
      "step": 72810
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.9099150896072388,
      "learning_rate": 0.00046390579220924506,
      "loss": 2.181,
      "step": 72811
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.7956773042678833,
      "learning_rate": 0.0004639023661221978,
      "loss": 2.1518,
      "step": 72812
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.7993009090423584,
      "learning_rate": 0.00046389894000467796,
      "loss": 2.3351,
      "step": 72813
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.9862198829650879,
      "learning_rate": 0.00046389551385668634,
      "loss": 2.2459,
      "step": 72814
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.1123698949813843,
      "learning_rate": 0.00046389208767822333,
      "loss": 2.3026,
      "step": 72815
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.8284459114074707,
      "learning_rate": 0.00046388866146928974,
      "loss": 2.2949,
      "step": 72816
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.7760177254676819,
      "learning_rate": 0.00046388523522988617,
      "loss": 2.1539,
      "step": 72817
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.4601606130599976,
      "learning_rate": 0.00046388180896001326,
      "loss": 2.3559,
      "step": 72818
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.0423001050949097,
      "learning_rate": 0.00046387838265967156,
      "loss": 2.2057,
      "step": 72819
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.765290379524231,
      "learning_rate": 0.00046387495632886183,
      "loss": 2.1842,
      "step": 72820
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.7405857443809509,
      "learning_rate": 0.0004638715299675846,
      "loss": 2.0649,
      "step": 72821
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.2861419916152954,
      "learning_rate": 0.00046386810357584065,
      "loss": 2.1158,
      "step": 72822
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.959928035736084,
      "learning_rate": 0.00046386467715363045,
      "loss": 2.2168,
      "step": 72823
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.7603201270103455,
      "learning_rate": 0.0004638612507009547,
      "loss": 2.2016,
      "step": 72824
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.7873603105545044,
      "learning_rate": 0.00046385782421781414,
      "loss": 2.2785,
      "step": 72825
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.0857763290405273,
      "learning_rate": 0.0004638543977042093,
      "loss": 2.1418,
      "step": 72826
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.127394199371338,
      "learning_rate": 0.0004638509711601407,
      "loss": 2.4058,
      "step": 72827
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.0251671075820923,
      "learning_rate": 0.0004638475445856092,
      "loss": 2.1452,
      "step": 72828
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.9128121733665466,
      "learning_rate": 0.0004638441179806154,
      "loss": 2.052,
      "step": 72829
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.8198130130767822,
      "learning_rate": 0.0004638406913451598,
      "loss": 2.2033,
      "step": 72830
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.9847007989883423,
      "learning_rate": 0.0004638372646792432,
      "loss": 2.0929,
      "step": 72831
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.968458354473114,
      "learning_rate": 0.00046383383798286607,
      "loss": 2.357,
      "step": 72832
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.9056761264801025,
      "learning_rate": 0.00046383041125602927,
      "loss": 2.2229,
      "step": 72833
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.2131948471069336,
      "learning_rate": 0.00046382698449873316,
      "loss": 2.242,
      "step": 72834
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.1454821825027466,
      "learning_rate": 0.00046382355771097856,
      "loss": 2.2353,
      "step": 72835
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.118045687675476,
      "learning_rate": 0.0004638201308927661,
      "loss": 2.2129,
      "step": 72836
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.7152284383773804,
      "learning_rate": 0.00046381670404409644,
      "loss": 2.044,
      "step": 72837
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.7304309606552124,
      "learning_rate": 0.00046381327716497005,
      "loss": 2.0082,
      "step": 72838
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.2281434535980225,
      "learning_rate": 0.0004638098502553878,
      "loss": 2.3852,
      "step": 72839
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.2791404724121094,
      "learning_rate": 0.00046380642331535004,
      "loss": 2.1808,
      "step": 72840
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.930195152759552,
      "learning_rate": 0.0004638029963448577,
      "loss": 2.1571,
      "step": 72841
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.8396121263504028,
      "learning_rate": 0.0004637995693439113,
      "loss": 2.04,
      "step": 72842
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.0017797946929932,
      "learning_rate": 0.0004637961423125114,
      "loss": 2.2023,
      "step": 72843
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.9720106720924377,
      "learning_rate": 0.00046379271525065876,
      "loss": 2.0836,
      "step": 72844
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.8726927042007446,
      "learning_rate": 0.00046378928815835394,
      "loss": 2.1148,
      "step": 72845
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.7420133948326111,
      "learning_rate": 0.00046378586103559756,
      "loss": 2.3852,
      "step": 72846
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.328287124633789,
      "learning_rate": 0.00046378243388239034,
      "loss": 2.218,
      "step": 72847
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.8113968372344971,
      "learning_rate": 0.0004637790066987329,
      "loss": 2.0331,
      "step": 72848
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.8058974146842957,
      "learning_rate": 0.00046377557948462585,
      "loss": 2.2517,
      "step": 72849
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.7359395623207092,
      "learning_rate": 0.0004637721522400697,
      "loss": 2.0318,
      "step": 72850
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.7412182688713074,
      "learning_rate": 0.0004637687249650654,
      "loss": 2.2059,
      "step": 72851
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.7811773419380188,
      "learning_rate": 0.0004637652976596133,
      "loss": 2.124,
      "step": 72852
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.815524697303772,
      "learning_rate": 0.00046376187032371406,
      "loss": 1.9874,
      "step": 72853
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.8556891679763794,
      "learning_rate": 0.00046375844295736863,
      "loss": 2.2794,
      "step": 72854
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.8096434473991394,
      "learning_rate": 0.00046375501556057723,
      "loss": 2.0145,
      "step": 72855
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.7809597849845886,
      "learning_rate": 0.0004637515881333407,
      "loss": 2.0029,
      "step": 72856
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.7779842615127563,
      "learning_rate": 0.0004637481606756597,
      "loss": 1.9997,
      "step": 72857
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.7844524383544922,
      "learning_rate": 0.0004637447331875348,
      "loss": 2.3113,
      "step": 72858
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.7517491579055786,
      "learning_rate": 0.00046374130566896675,
      "loss": 2.2001,
      "step": 72859
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.8693498373031616,
      "learning_rate": 0.00046373787811995596,
      "loss": 2.049,
      "step": 72860
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.7282227277755737,
      "learning_rate": 0.00046373445054050333,
      "loss": 2.2728,
      "step": 72861
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.7602939605712891,
      "learning_rate": 0.0004637310229306093,
      "loss": 2.1923,
      "step": 72862
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.8386940956115723,
      "learning_rate": 0.00046372759529027463,
      "loss": 2.2583,
      "step": 72863
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.7683849930763245,
      "learning_rate": 0.0004637241676194998,
      "loss": 2.3359,
      "step": 72864
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.7987968921661377,
      "learning_rate": 0.00046372073991828564,
      "loss": 2.4546,
      "step": 72865
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.8120642304420471,
      "learning_rate": 0.00046371731218663277,
      "loss": 2.3138,
      "step": 72866
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.7682520747184753,
      "learning_rate": 0.00046371388442454167,
      "loss": 2.2306,
      "step": 72867
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.7514742016792297,
      "learning_rate": 0.0004637104566320131,
      "loss": 2.2964,
      "step": 72868
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.7498088479042053,
      "learning_rate": 0.00046370702880904765,
      "loss": 2.2584,
      "step": 72869
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.8044735789299011,
      "learning_rate": 0.00046370360095564597,
      "loss": 2.2727,
      "step": 72870
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.8514264822006226,
      "learning_rate": 0.00046370017307180867,
      "loss": 2.2857,
      "step": 72871
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.7969760298728943,
      "learning_rate": 0.0004636967451575365,
      "loss": 2.1965,
      "step": 72872
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.7704139947891235,
      "learning_rate": 0.00046369331721283,
      "loss": 2.3845,
      "step": 72873
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.8013939261436462,
      "learning_rate": 0.0004636898892376897,
      "loss": 2.2507,
      "step": 72874
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.7986116409301758,
      "learning_rate": 0.0004636864612321165,
      "loss": 2.342,
      "step": 72875
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.7497191429138184,
      "learning_rate": 0.00046368303319611084,
      "loss": 2.3397,
      "step": 72876
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.7270492911338806,
      "learning_rate": 0.00046367960512967346,
      "loss": 2.055,
      "step": 72877
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.75597083568573,
      "learning_rate": 0.00046367617703280497,
      "loss": 2.3362,
      "step": 72878
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.7752695083618164,
      "learning_rate": 0.00046367274890550585,
      "loss": 2.1482,
      "step": 72879
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.8384668827056885,
      "learning_rate": 0.00046366932074777697,
      "loss": 2.357,
      "step": 72880
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.8254760503768921,
      "learning_rate": 0.000463665892559619,
      "loss": 2.2899,
      "step": 72881
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.7798791527748108,
      "learning_rate": 0.0004636624643410322,
      "loss": 2.2725,
      "step": 72882
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.8879164457321167,
      "learning_rate": 0.0004636590360920176,
      "loss": 2.431,
      "step": 72883
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.772811770439148,
      "learning_rate": 0.0004636556078125757,
      "loss": 2.1322,
      "step": 72884
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.8050172924995422,
      "learning_rate": 0.0004636521795027071,
      "loss": 2.3205,
      "step": 72885
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.7461523413658142,
      "learning_rate": 0.0004636487511624124,
      "loss": 2.3128,
      "step": 72886
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.7722629904747009,
      "learning_rate": 0.00046364532279169246,
      "loss": 2.2313,
      "step": 72887
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.7387471199035645,
      "learning_rate": 0.00046364189439054767,
      "loss": 2.1626,
      "step": 72888
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.7440888285636902,
      "learning_rate": 0.00046363846595897875,
      "loss": 2.1922,
      "step": 72889
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.878515362739563,
      "learning_rate": 0.0004636350374969865,
      "loss": 2.363,
      "step": 72890
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.8032267093658447,
      "learning_rate": 0.0004636316090045712,
      "loss": 2.3854,
      "step": 72891
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.7711780667304993,
      "learning_rate": 0.0004636281804817338,
      "loss": 2.2163,
      "step": 72892
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.7727120518684387,
      "learning_rate": 0.0004636247519284748,
      "loss": 2.3107,
      "step": 72893
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.8288542032241821,
      "learning_rate": 0.00046362132334479495,
      "loss": 2.3383,
      "step": 72894
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.8014918565750122,
      "learning_rate": 0.0004636178947306947,
      "loss": 2.1039,
      "step": 72895
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.9882861971855164,
      "learning_rate": 0.00046361446608617486,
      "loss": 2.232,
      "step": 72896
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.7352675795555115,
      "learning_rate": 0.00046361103741123595,
      "loss": 2.2313,
      "step": 72897
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.8168820738792419,
      "learning_rate": 0.0004636076087058787,
      "loss": 2.1326,
      "step": 72898
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.8282576203346252,
      "learning_rate": 0.0004636041799701037,
      "loss": 2.3262,
      "step": 72899
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.7264589667320251,
      "learning_rate": 0.0004636007512039117,
      "loss": 2.1386,
      "step": 72900
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.7502917051315308,
      "learning_rate": 0.000463597322407303,
      "loss": 2.3206,
      "step": 72901
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.9121111035346985,
      "learning_rate": 0.0004635938935802786,
      "loss": 2.3054,
      "step": 72902
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.8365342020988464,
      "learning_rate": 0.0004635904647228391,
      "loss": 2.2048,
      "step": 72903
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.7913736701011658,
      "learning_rate": 0.0004635870358349849,
      "loss": 2.4202,
      "step": 72904
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.7260270118713379,
      "learning_rate": 0.00046358360691671686,
      "loss": 2.1077,
      "step": 72905
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.7764148712158203,
      "learning_rate": 0.0004635801779680355,
      "loss": 2.2625,
      "step": 72906
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.8045164346694946,
      "learning_rate": 0.0004635767489889415,
      "loss": 2.2545,
      "step": 72907
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.8155032992362976,
      "learning_rate": 0.0004635733199794355,
      "loss": 2.3535,
      "step": 72908
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.873754620552063,
      "learning_rate": 0.00046356989093951817,
      "loss": 2.1692,
      "step": 72909
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.7921943068504333,
      "learning_rate": 0.0004635664618691901,
      "loss": 2.1929,
      "step": 72910
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.8018242120742798,
      "learning_rate": 0.00046356303276845186,
      "loss": 2.1253,
      "step": 72911
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.803925096988678,
      "learning_rate": 0.0004635596036373042,
      "loss": 2.1683,
      "step": 72912
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.7919953465461731,
      "learning_rate": 0.00046355617447574777,
      "loss": 2.298,
      "step": 72913
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.8705773949623108,
      "learning_rate": 0.0004635527452837831,
      "loss": 2.2611,
      "step": 72914
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.7905001044273376,
      "learning_rate": 0.0004635493160614109,
      "loss": 2.1543,
      "step": 72915
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.7909024953842163,
      "learning_rate": 0.00046354588680863184,
      "loss": 2.2564,
      "step": 72916
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.7655071020126343,
      "learning_rate": 0.0004635424575254464,
      "loss": 2.4145,
      "step": 72917
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.8115125298500061,
      "learning_rate": 0.0004635390282118555,
      "loss": 2.085,
      "step": 72918
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.8082641959190369,
      "learning_rate": 0.0004635355988678595,
      "loss": 2.2741,
      "step": 72919
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.7744162678718567,
      "learning_rate": 0.00046353216949345915,
      "loss": 2.2197,
      "step": 72920
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.8456600904464722,
      "learning_rate": 0.0004635287400886551,
      "loss": 2.3674,
      "step": 72921
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.75814288854599,
      "learning_rate": 0.000463525310653448,
      "loss": 2.2217,
      "step": 72922
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.7168615460395813,
      "learning_rate": 0.00046352188118783835,
      "loss": 2.0588,
      "step": 72923
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.7750514149665833,
      "learning_rate": 0.00046351845169182706,
      "loss": 2.4608,
      "step": 72924
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.8989003300666809,
      "learning_rate": 0.0004635150221654145,
      "loss": 2.3696,
      "step": 72925
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.8563267588615417,
      "learning_rate": 0.0004635115926086014,
      "loss": 2.1899,
      "step": 72926
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.7892013788223267,
      "learning_rate": 0.00046350816302138844,
      "loss": 2.0796,
      "step": 72927
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.7253478169441223,
      "learning_rate": 0.0004635047334037762,
      "loss": 1.9506,
      "step": 72928
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.804027795791626,
      "learning_rate": 0.0004635013037557654,
      "loss": 2.0869,
      "step": 72929
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.9841980338096619,
      "learning_rate": 0.00046349787407735655,
      "loss": 2.2466,
      "step": 72930
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.8448086977005005,
      "learning_rate": 0.00046349444436855037,
      "loss": 1.9534,
      "step": 72931
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.7763807773590088,
      "learning_rate": 0.00046349101462934755,
      "loss": 2.2511,
      "step": 72932
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.7635148763656616,
      "learning_rate": 0.00046348758485974863,
      "loss": 2.2686,
      "step": 72933
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.7723006010055542,
      "learning_rate": 0.0004634841550597543,
      "loss": 2.0096,
      "step": 72934
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.7823545932769775,
      "learning_rate": 0.00046348072522936517,
      "loss": 2.1107,
      "step": 72935
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.8653333187103271,
      "learning_rate": 0.0004634772953685819,
      "loss": 2.2054,
      "step": 72936
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.7801294922828674,
      "learning_rate": 0.000463473865477405,
      "loss": 1.991,
      "step": 72937
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.8195005059242249,
      "learning_rate": 0.00046347043555583533,
      "loss": 2.405,
      "step": 72938
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.7488869428634644,
      "learning_rate": 0.0004634670056038735,
      "loss": 2.0619,
      "step": 72939
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.8428762555122375,
      "learning_rate": 0.00046346357562151985,
      "loss": 2.3529,
      "step": 72940
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.8253066539764404,
      "learning_rate": 0.0004634601456087754,
      "loss": 2.1589,
      "step": 72941
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.7926336526870728,
      "learning_rate": 0.0004634567155656406,
      "loss": 2.1284,
      "step": 72942
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.696908712387085,
      "learning_rate": 0.0004634532854921161,
      "loss": 2.091,
      "step": 72943
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.6917981505393982,
      "learning_rate": 0.00046344985538820254,
      "loss": 2.0733,
      "step": 72944
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.7177711129188538,
      "learning_rate": 0.0004634464252539006,
      "loss": 2.2242,
      "step": 72945
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.7291672825813293,
      "learning_rate": 0.0004634429950892108,
      "loss": 2.0754,
      "step": 72946
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.7698588371276855,
      "learning_rate": 0.0004634395648941339,
      "loss": 2.2599,
      "step": 72947
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.7595241665840149,
      "learning_rate": 0.00046343613466867054,
      "loss": 2.2196,
      "step": 72948
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.8479901552200317,
      "learning_rate": 0.0004634327044128213,
      "loss": 2.2519,
      "step": 72949
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.345954418182373,
      "learning_rate": 0.0004634292741265868,
      "loss": 2.3683,
      "step": 72950
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.7616199851036072,
      "learning_rate": 0.00046342584380996776,
      "loss": 2.217,
      "step": 72951
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.9437817335128784,
      "learning_rate": 0.0004634224134629648,
      "loss": 2.4794,
      "step": 72952
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.7493017315864563,
      "learning_rate": 0.00046341898308557833,
      "loss": 2.3753,
      "step": 72953
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.7565028667449951,
      "learning_rate": 0.00046341555267780945,
      "loss": 2.0542,
      "step": 72954
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.7733122706413269,
      "learning_rate": 0.0004634121222396584,
      "loss": 2.2038,
      "step": 72955
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.7402402758598328,
      "learning_rate": 0.0004634086917711259,
      "loss": 2.1663,
      "step": 72956
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.8099181652069092,
      "learning_rate": 0.00046340526127221275,
      "loss": 2.1769,
      "step": 72957
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.7615033984184265,
      "learning_rate": 0.0004634018307429195,
      "loss": 2.0136,
      "step": 72958
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.722185492515564,
      "learning_rate": 0.00046339840018324663,
      "loss": 2.1895,
      "step": 72959
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.7230331897735596,
      "learning_rate": 0.00046339496959319496,
      "loss": 2.276,
      "step": 72960
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.7446742653846741,
      "learning_rate": 0.0004633915389727651,
      "loss": 2.1813,
      "step": 72961
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.7927968502044678,
      "learning_rate": 0.00046338810832195765,
      "loss": 2.1273,
      "step": 72962
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.8282654285430908,
      "learning_rate": 0.00046338467764077336,
      "loss": 2.2948,
      "step": 72963
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.7765108942985535,
      "learning_rate": 0.0004633812469292127,
      "loss": 2.0972,
      "step": 72964
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.8073722124099731,
      "learning_rate": 0.00046337781618727635,
      "loss": 2.3476,
      "step": 72965
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.7369052171707153,
      "learning_rate": 0.00046337438541496504,
      "loss": 2.2108,
      "step": 72966
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.7527763843536377,
      "learning_rate": 0.0004633709546122793,
      "loss": 2.4511,
      "step": 72967
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.7565894722938538,
      "learning_rate": 0.0004633675237792199,
      "loss": 2.4425,
      "step": 72968
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.7980767488479614,
      "learning_rate": 0.00046336409291578733,
      "loss": 2.2631,
      "step": 72969
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.861303985118866,
      "learning_rate": 0.0004633606620219823,
      "loss": 2.3007,
      "step": 72970
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.7952149510383606,
      "learning_rate": 0.00046335723109780535,
      "loss": 2.0485,
      "step": 72971
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.9104182720184326,
      "learning_rate": 0.00046335380014325727,
      "loss": 2.1562,
      "step": 72972
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.8516539931297302,
      "learning_rate": 0.00046335036915833873,
      "loss": 2.3547,
      "step": 72973
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.8263817429542542,
      "learning_rate": 0.0004633469381430502,
      "loss": 2.3068,
      "step": 72974
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.7727488279342651,
      "learning_rate": 0.0004633435070973924,
      "loss": 2.6091,
      "step": 72975
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.9235787987709045,
      "learning_rate": 0.000463340076021366,
      "loss": 2.1214,
      "step": 72976
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.8359308838844299,
      "learning_rate": 0.0004633366449149715,
      "loss": 2.2201,
      "step": 72977
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.8447695374488831,
      "learning_rate": 0.00046333321377820967,
      "loss": 2.2317,
      "step": 72978
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.8726341128349304,
      "learning_rate": 0.00046332978261108115,
      "loss": 2.3031,
      "step": 72979
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.8539136648178101,
      "learning_rate": 0.0004633263514135865,
      "loss": 2.2422,
      "step": 72980
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.7661781907081604,
      "learning_rate": 0.0004633229201857264,
      "loss": 2.3352,
      "step": 72981
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.701847493648529,
      "learning_rate": 0.0004633194889275015,
      "loss": 2.1764,
      "step": 72982
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.769188642501831,
      "learning_rate": 0.0004633160576389124,
      "loss": 2.4266,
      "step": 72983
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.7977595329284668,
      "learning_rate": 0.0004633126263199598,
      "loss": 1.9949,
      "step": 72984
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.7929712533950806,
      "learning_rate": 0.0004633091949706443,
      "loss": 1.9647,
      "step": 72985
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.8546260595321655,
      "learning_rate": 0.00046330576359096645,
      "loss": 2.2439,
      "step": 72986
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.8812338709831238,
      "learning_rate": 0.0004633023321809271,
      "loss": 2.2278,
      "step": 72987
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.8328786492347717,
      "learning_rate": 0.0004632989007405267,
      "loss": 2.2678,
      "step": 72988
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.7902733683586121,
      "learning_rate": 0.0004632954692697659,
      "loss": 2.3526,
      "step": 72989
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.8320762515068054,
      "learning_rate": 0.00046329203776864544,
      "loss": 1.9901,
      "step": 72990
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.7947450280189514,
      "learning_rate": 0.000463288606237166,
      "loss": 2.0266,
      "step": 72991
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.8065284490585327,
      "learning_rate": 0.00046328517467532804,
      "loss": 2.2437,
      "step": 72992
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.7380841374397278,
      "learning_rate": 0.00046328174308313226,
      "loss": 2.1934,
      "step": 72993
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.7876881957054138,
      "learning_rate": 0.0004632783114605794,
      "loss": 2.1124,
      "step": 72994
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.756053626537323,
      "learning_rate": 0.00046327487980767,
      "loss": 2.1511,
      "step": 72995
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.8585925698280334,
      "learning_rate": 0.0004632714481244046,
      "loss": 2.142,
      "step": 72996
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.8321137428283691,
      "learning_rate": 0.0004632680164107841,
      "loss": 2.365,
      "step": 72997
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.8958353996276855,
      "learning_rate": 0.00046326458466680896,
      "loss": 2.2535,
      "step": 72998
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.839392900466919,
      "learning_rate": 0.0004632611528924798,
      "loss": 2.1077,
      "step": 72999
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.8741758465766907,
      "learning_rate": 0.00046325772108779734,
      "loss": 2.1144,
      "step": 73000
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.7736489772796631,
      "learning_rate": 0.0004632542892527622,
      "loss": 2.1288,
      "step": 73001
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.8262973427772522,
      "learning_rate": 0.000463250857387375,
      "loss": 2.32,
      "step": 73002
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.8177314400672913,
      "learning_rate": 0.00046324742549163635,
      "loss": 2.1888,
      "step": 73003
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.807451069355011,
      "learning_rate": 0.00046324399356554696,
      "loss": 2.2548,
      "step": 73004
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.8413519263267517,
      "learning_rate": 0.00046324056160910743,
      "loss": 2.3412,
      "step": 73005
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.8070011138916016,
      "learning_rate": 0.00046323712962231844,
      "loss": 2.3388,
      "step": 73006
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.8460001945495605,
      "learning_rate": 0.00046323369760518045,
      "loss": 2.245,
      "step": 73007
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.7889623045921326,
      "learning_rate": 0.00046323026555769435,
      "loss": 2.3911,
      "step": 73008
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.8611578941345215,
      "learning_rate": 0.0004632268334798606,
      "loss": 2.3339,
      "step": 73009
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.8432956337928772,
      "learning_rate": 0.00046322340137168,
      "loss": 2.1978,
      "step": 73010
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.8938066363334656,
      "learning_rate": 0.00046321996923315295,
      "loss": 2.2546,
      "step": 73011
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.8114611506462097,
      "learning_rate": 0.00046321653706428033,
      "loss": 2.4418,
      "step": 73012
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.812545895576477,
      "learning_rate": 0.0004632131048650626,
      "loss": 2.3034,
      "step": 73013
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.7483635544776917,
      "learning_rate": 0.00046320967263550054,
      "loss": 2.473,
      "step": 73014
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.8228039145469666,
      "learning_rate": 0.0004632062403755947,
      "loss": 2.193,
      "step": 73015
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.0257070064544678,
      "learning_rate": 0.0004632028080853458,
      "loss": 2.1309,
      "step": 73016
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.7147210240364075,
      "learning_rate": 0.00046319937576475435,
      "loss": 2.1987,
      "step": 73017
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.7943411469459534,
      "learning_rate": 0.000463195943413821,
      "loss": 1.9326,
      "step": 73018
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.8689981698989868,
      "learning_rate": 0.0004631925110325465,
      "loss": 2.2644,
      "step": 73019
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.8345586061477661,
      "learning_rate": 0.0004631890786209314,
      "loss": 2.2143,
      "step": 73020
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.8139110803604126,
      "learning_rate": 0.00046318564617897645,
      "loss": 2.2729,
      "step": 73021
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.746242344379425,
      "learning_rate": 0.0004631822137066822,
      "loss": 2.0812,
      "step": 73022
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.7507234215736389,
      "learning_rate": 0.0004631787812040492,
      "loss": 2.0945,
      "step": 73023
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.7696079015731812,
      "learning_rate": 0.0004631753486710783,
      "loss": 2.2958,
      "step": 73024
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.8958613276481628,
      "learning_rate": 0.00046317191610776995,
      "loss": 2.171,
      "step": 73025
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.7248997688293457,
      "learning_rate": 0.0004631684835141248,
      "loss": 2.1254,
      "step": 73026
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.7196749448776245,
      "learning_rate": 0.0004631650508901437,
      "loss": 2.2612,
      "step": 73027
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.7831565737724304,
      "learning_rate": 0.000463161618235827,
      "loss": 2.2197,
      "step": 73028
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.7684338688850403,
      "learning_rate": 0.0004631581855511756,
      "loss": 2.1268,
      "step": 73029
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.7827356457710266,
      "learning_rate": 0.00046315475283619,
      "loss": 2.4109,
      "step": 73030
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.786900520324707,
      "learning_rate": 0.0004631513200908707,
      "loss": 2.3134,
      "step": 73031
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.7734280228614807,
      "learning_rate": 0.0004631478873152186,
      "loss": 2.1013,
      "step": 73032
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.8433643579483032,
      "learning_rate": 0.00046314445450923425,
      "loss": 1.9601,
      "step": 73033
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.8007025122642517,
      "learning_rate": 0.00046314102167291826,
      "loss": 2.1903,
      "step": 73034
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.8207885026931763,
      "learning_rate": 0.0004631375888062712,
      "loss": 2.0395,
      "step": 73035
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.7574419379234314,
      "learning_rate": 0.0004631341559092938,
      "loss": 2.0824,
      "step": 73036
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.109904408454895,
      "learning_rate": 0.00046313072298198674,
      "loss": 2.0336,
      "step": 73037
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.746433436870575,
      "learning_rate": 0.00046312729002435054,
      "loss": 2.078,
      "step": 73038
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.8620445132255554,
      "learning_rate": 0.000463123857036386,
      "loss": 2.2659,
      "step": 73039
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.8639559149742126,
      "learning_rate": 0.00046312042401809356,
      "loss": 2.0985,
      "step": 73040
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.7640602588653564,
      "learning_rate": 0.00046311699096947403,
      "loss": 2.2005,
      "step": 73041
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.8334112763404846,
      "learning_rate": 0.00046311355789052785,
      "loss": 2.2036,
      "step": 73042
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.7456256151199341,
      "learning_rate": 0.0004631101247812559,
      "loss": 2.2617,
      "step": 73043
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.7944135665893555,
      "learning_rate": 0.0004631066916416586,
      "loss": 2.2779,
      "step": 73044
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.7529224753379822,
      "learning_rate": 0.00046310325847173677,
      "loss": 2.2617,
      "step": 73045
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.78586745262146,
      "learning_rate": 0.000463099825271491,
      "loss": 2.3546,
      "step": 73046
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.0025843381881714,
      "learning_rate": 0.00046309639204092183,
      "loss": 2.2672,
      "step": 73047
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.7248573899269104,
      "learning_rate": 0.0004630929587800299,
      "loss": 2.1981,
      "step": 73048
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.8297158479690552,
      "learning_rate": 0.00046308952548881595,
      "loss": 2.3623,
      "step": 73049
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.8051862120628357,
      "learning_rate": 0.0004630860921672806,
      "loss": 2.3998,
      "step": 73050
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.7538151144981384,
      "learning_rate": 0.0004630826588154245,
      "loss": 2.2995,
      "step": 73051
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.8080542683601379,
      "learning_rate": 0.00046307922543324827,
      "loss": 2.272,
      "step": 73052
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.7946982979774475,
      "learning_rate": 0.0004630757920207524,
      "loss": 2.1913,
      "step": 73053
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.0053215026855469,
      "learning_rate": 0.0004630723585779378,
      "loss": 2.2107,
      "step": 73054
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.8653287291526794,
      "learning_rate": 0.00046306892510480495,
      "loss": 2.4542,
      "step": 73055
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.8512579202651978,
      "learning_rate": 0.00046306549160135443,
      "loss": 2.2503,
      "step": 73056
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.8654144406318665,
      "learning_rate": 0.000463062058067587,
      "loss": 2.2934,
      "step": 73057
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.7938389778137207,
      "learning_rate": 0.0004630586245035033,
      "loss": 2.0922,
      "step": 73058
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.8544030785560608,
      "learning_rate": 0.0004630551909091038,
      "loss": 2.3295,
      "step": 73059
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.7321667075157166,
      "learning_rate": 0.0004630517572843894,
      "loss": 2.2683,
      "step": 73060
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.805570662021637,
      "learning_rate": 0.00046304832362936054,
      "loss": 2.1878,
      "step": 73061
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.9098771810531616,
      "learning_rate": 0.0004630448899440179,
      "loss": 2.3031,
      "step": 73062
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.7311392426490784,
      "learning_rate": 0.00046304145622836214,
      "loss": 2.1413,
      "step": 73063
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.8603811860084534,
      "learning_rate": 0.0004630380224823939,
      "loss": 2.4315,
      "step": 73064
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.8810141682624817,
      "learning_rate": 0.0004630345887061139,
      "loss": 2.2603,
      "step": 73065
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.8279832601547241,
      "learning_rate": 0.0004630311548995226,
      "loss": 2.053,
      "step": 73066
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.8150740265846252,
      "learning_rate": 0.00046302772106262077,
      "loss": 2.1514,
      "step": 73067
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.727933406829834,
      "learning_rate": 0.0004630242871954089,
      "loss": 2.0977,
      "step": 73068
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.8446912169456482,
      "learning_rate": 0.00046302085329788794,
      "loss": 2.1459,
      "step": 73069
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.778806746006012,
      "learning_rate": 0.00046301741937005824,
      "loss": 2.3848,
      "step": 73070
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.7586821913719177,
      "learning_rate": 0.00046301398541192043,
      "loss": 2.1181,
      "step": 73071
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.7536767721176147,
      "learning_rate": 0.0004630105514234753,
      "loss": 2.4595,
      "step": 73072
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.7330327033996582,
      "learning_rate": 0.0004630071174047235,
      "loss": 2.371,
      "step": 73073
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.8669905662536621,
      "learning_rate": 0.00046300368335566554,
      "loss": 2.3328,
      "step": 73074
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.7628858089447021,
      "learning_rate": 0.0004630002492763021,
      "loss": 2.2336,
      "step": 73075
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.8534088730812073,
      "learning_rate": 0.00046299681516663387,
      "loss": 2.3447,
      "step": 73076
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.8369772434234619,
      "learning_rate": 0.0004629933810266615,
      "loss": 1.9186,
      "step": 73077
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.7461602687835693,
      "learning_rate": 0.0004629899468563855,
      "loss": 2.1301,
      "step": 73078
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.8358290195465088,
      "learning_rate": 0.0004629865126558066,
      "loss": 2.1081,
      "step": 73079
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.7535650134086609,
      "learning_rate": 0.0004629830784249255,
      "loss": 2.0621,
      "step": 73080
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.841776967048645,
      "learning_rate": 0.00046297964416374266,
      "loss": 2.3427,
      "step": 73081
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.6854041218757629,
      "learning_rate": 0.00046297620987225893,
      "loss": 2.0583,
      "step": 73082
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.844803512096405,
      "learning_rate": 0.0004629727755504748,
      "loss": 2.2842,
      "step": 73083
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.8681315779685974,
      "learning_rate": 0.0004629693411983909,
      "loss": 2.2356,
      "step": 73084
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.8406422734260559,
      "learning_rate": 0.00046296590681600805,
      "loss": 2.0128,
      "step": 73085
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.801962673664093,
      "learning_rate": 0.0004629624724033267,
      "loss": 2.266,
      "step": 73086
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.7821887135505676,
      "learning_rate": 0.0004629590379603475,
      "loss": 2.2615,
      "step": 73087
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.9893983006477356,
      "learning_rate": 0.0004629556034870712,
      "loss": 2.3471,
      "step": 73088
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.8153099417686462,
      "learning_rate": 0.0004629521689834984,
      "loss": 2.2614,
      "step": 73089
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.788920521736145,
      "learning_rate": 0.0004629487344496297,
      "loss": 2.3156,
      "step": 73090
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.8086596727371216,
      "learning_rate": 0.00046294529988546567,
      "loss": 2.3681,
      "step": 73091
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.9543344378471375,
      "learning_rate": 0.0004629418652910071,
      "loss": 1.9034,
      "step": 73092
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.7399454712867737,
      "learning_rate": 0.0004629384306662546,
      "loss": 2.0646,
      "step": 73093
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.8039703369140625,
      "learning_rate": 0.00046293499601120865,
      "loss": 2.2638,
      "step": 73094
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.1035679578781128,
      "learning_rate": 0.00046293156132587015,
      "loss": 2.4574,
      "step": 73095
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.8163281083106995,
      "learning_rate": 0.00046292812661023953,
      "loss": 2.0649,
      "step": 73096
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.80074542760849,
      "learning_rate": 0.00046292469186431744,
      "loss": 2.2079,
      "step": 73097
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.8583080172538757,
      "learning_rate": 0.0004629212570881047,
      "loss": 2.2932,
      "step": 73098
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.8314806222915649,
      "learning_rate": 0.0004629178222816017,
      "loss": 2.6921,
      "step": 73099
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.8543515801429749,
      "learning_rate": 0.0004629143874448093,
      "loss": 2.2145,
      "step": 73100
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.8326801061630249,
      "learning_rate": 0.000462910952577728,
      "loss": 2.1876,
      "step": 73101
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.8349201083183289,
      "learning_rate": 0.0004629075176803585,
      "loss": 2.28,
      "step": 73102
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.7760002613067627,
      "learning_rate": 0.0004629040827527014,
      "loss": 2.2579,
      "step": 73103
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.7727575898170471,
      "learning_rate": 0.0004629006477947573,
      "loss": 2.336,
      "step": 73104
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.7900226712226868,
      "learning_rate": 0.00046289721280652694,
      "loss": 2.2444,
      "step": 73105
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.8417677283287048,
      "learning_rate": 0.0004628937777880109,
      "loss": 2.4147,
      "step": 73106
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.7487522959709167,
      "learning_rate": 0.0004628903427392099,
      "loss": 2.2969,
      "step": 73107
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.8127534985542297,
      "learning_rate": 0.00046288690766012444,
      "loss": 2.0875,
      "step": 73108
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.7934370636940002,
      "learning_rate": 0.00046288347255075523,
      "loss": 2.1228,
      "step": 73109
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.8451870083808899,
      "learning_rate": 0.00046288003741110296,
      "loss": 2.2641,
      "step": 73110
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.8166922926902771,
      "learning_rate": 0.0004628766022411681,
      "loss": 2.3117,
      "step": 73111
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.8696026802062988,
      "learning_rate": 0.0004628731670409516,
      "loss": 1.8799,
      "step": 73112
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.7798364162445068,
      "learning_rate": 0.0004628697318104538,
      "loss": 2.1504,
      "step": 73113
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.7921351194381714,
      "learning_rate": 0.0004628662965496754,
      "loss": 2.4496,
      "step": 73114
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.894733190536499,
      "learning_rate": 0.00046286286125861713,
      "loss": 2.1779,
      "step": 73115
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.7476506233215332,
      "learning_rate": 0.00046285942593727957,
      "loss": 2.0461,
      "step": 73116
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.0856969356536865,
      "learning_rate": 0.00046285599058566333,
      "loss": 2.0598,
      "step": 73117
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.9969270825386047,
      "learning_rate": 0.00046285255520376906,
      "loss": 2.137,
      "step": 73118
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.7730711102485657,
      "learning_rate": 0.0004628491197915975,
      "loss": 2.1109,
      "step": 73119
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.9165599942207336,
      "learning_rate": 0.00046284568434914927,
      "loss": 2.147,
      "step": 73120
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.9217093586921692,
      "learning_rate": 0.0004628422488764248,
      "loss": 2.1228,
      "step": 73121
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.0621765851974487,
      "learning_rate": 0.000462838813373425,
      "loss": 2.1719,
      "step": 73122
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.7218603491783142,
      "learning_rate": 0.00046283537784015037,
      "loss": 2.3474,
      "step": 73123
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.1333503723144531,
      "learning_rate": 0.00046283194227660155,
      "loss": 2.1,
      "step": 73124
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.8850768208503723,
      "learning_rate": 0.00046282850668277926,
      "loss": 2.2545,
      "step": 73125
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.735858678817749,
      "learning_rate": 0.000462825071058684,
      "loss": 2.3614,
      "step": 73126
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.8821814060211182,
      "learning_rate": 0.0004628216354043165,
      "loss": 2.159,
      "step": 73127
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.8499712347984314,
      "learning_rate": 0.0004628181997196774,
      "loss": 2.09,
      "step": 73128
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.7342832088470459,
      "learning_rate": 0.0004628147640047673,
      "loss": 2.1945,
      "step": 73129
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.7442233562469482,
      "learning_rate": 0.00046281132825958687,
      "loss": 2.1372,
      "step": 73130
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.8326854705810547,
      "learning_rate": 0.0004628078924841368,
      "loss": 2.2417,
      "step": 73131
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.7690215706825256,
      "learning_rate": 0.0004628044566784176,
      "loss": 2.036,
      "step": 73132
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.7323094010353088,
      "learning_rate": 0.00046280102084242993,
      "loss": 2.3005,
      "step": 73133
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.7232418656349182,
      "learning_rate": 0.00046279758497617463,
      "loss": 2.3732,
      "step": 73134
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.9293174147605896,
      "learning_rate": 0.00046279414907965204,
      "loss": 2.0955,
      "step": 73135
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.7125663757324219,
      "learning_rate": 0.00046279071315286297,
      "loss": 2.2745,
      "step": 73136
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.7694016098976135,
      "learning_rate": 0.0004627872771958081,
      "loss": 2.1375,
      "step": 73137
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.7455679178237915,
      "learning_rate": 0.000462783841208488,
      "loss": 2.3938,
      "step": 73138
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.8822556734085083,
      "learning_rate": 0.0004627804051909032,
      "loss": 2.0554,
      "step": 73139
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.8060464859008789,
      "learning_rate": 0.0004627769691430546,
      "loss": 2.0512,
      "step": 73140
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.8296593427658081,
      "learning_rate": 0.00046277353306494265,
      "loss": 2.2528,
      "step": 73141
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.8119629621505737,
      "learning_rate": 0.0004627700969565679,
      "loss": 1.9509,
      "step": 73142
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.7240305542945862,
      "learning_rate": 0.0004627666608179313,
      "loss": 2.2338,
      "step": 73143
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.9133145809173584,
      "learning_rate": 0.0004627632246490332,
      "loss": 2.1947,
      "step": 73144
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.7427096366882324,
      "learning_rate": 0.0004627597884498743,
      "loss": 2.2941,
      "step": 73145
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.8202338218688965,
      "learning_rate": 0.0004627563522204555,
      "loss": 2.1222,
      "step": 73146
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.8249771595001221,
      "learning_rate": 0.00046275291596077693,
      "loss": 2.204,
      "step": 73147
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.8895915746688843,
      "learning_rate": 0.00046274947967083965,
      "loss": 2.0779,
      "step": 73148
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.802726149559021,
      "learning_rate": 0.0004627460433506442,
      "loss": 2.2564,
      "step": 73149
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.7798300981521606,
      "learning_rate": 0.0004627426070001912,
      "loss": 2.1067,
      "step": 73150
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.8376932144165039,
      "learning_rate": 0.0004627391706194812,
      "loss": 2.3475,
      "step": 73151
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.861838698387146,
      "learning_rate": 0.00046273573420851504,
      "loss": 2.3015,
      "step": 73152
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.782413125038147,
      "learning_rate": 0.0004627322977672931,
      "loss": 2.405,
      "step": 73153
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.7460450530052185,
      "learning_rate": 0.0004627288612958162,
      "loss": 2.327,
      "step": 73154
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.806355357170105,
      "learning_rate": 0.00046272542479408504,
      "loss": 2.1317,
      "step": 73155
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.825135350227356,
      "learning_rate": 0.0004627219882621,
      "loss": 2.1573,
      "step": 73156
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.7644850611686707,
      "learning_rate": 0.00046271855169986194,
      "loss": 2.1382,
      "step": 73157
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.8707535862922668,
      "learning_rate": 0.00046271511510737143,
      "loss": 2.1875,
      "step": 73158
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.8384705185890198,
      "learning_rate": 0.00046271167848462903,
      "loss": 2.2155,
      "step": 73159
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.7117570638656616,
      "learning_rate": 0.00046270824183163554,
      "loss": 2.2711,
      "step": 73160
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.7571904063224792,
      "learning_rate": 0.00046270480514839157,
      "loss": 2.2785,
      "step": 73161
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.7440129518508911,
      "learning_rate": 0.00046270136843489754,
      "loss": 2.1257,
      "step": 73162
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.8527916073799133,
      "learning_rate": 0.0004626979316911544,
      "loss": 2.3725,
      "step": 73163
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.8003756999969482,
      "learning_rate": 0.0004626944949171626,
      "loss": 2.2745,
      "step": 73164
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.7106417417526245,
      "learning_rate": 0.0004626910581129228,
      "loss": 2.2322,
      "step": 73165
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.7558120489120483,
      "learning_rate": 0.00046268762127843564,
      "loss": 2.2116,
      "step": 73166
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.754783034324646,
      "learning_rate": 0.00046268418441370186,
      "loss": 2.2483,
      "step": 73167
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.7330324053764343,
      "learning_rate": 0.0004626807475187219,
      "loss": 2.2938,
      "step": 73168
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.8116888403892517,
      "learning_rate": 0.00046267731059349666,
      "loss": 2.3208,
      "step": 73169
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.7799146771430969,
      "learning_rate": 0.0004626738736380266,
      "loss": 2.2248,
      "step": 73170
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.835232138633728,
      "learning_rate": 0.00046267043665231233,
      "loss": 2.1307,
      "step": 73171
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.703007698059082,
      "learning_rate": 0.0004626669996363546,
      "loss": 2.1406,
      "step": 73172
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.7626424431800842,
      "learning_rate": 0.00046266356259015397,
      "loss": 2.0945,
      "step": 73173
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.780800998210907,
      "learning_rate": 0.0004626601255137112,
      "loss": 2.3558,
      "step": 73174
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.7422414422035217,
      "learning_rate": 0.00046265668840702673,
      "loss": 2.234,
      "step": 73175
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.7915742993354797,
      "learning_rate": 0.0004626532512701014,
      "loss": 2.1103,
      "step": 73176
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.76784747838974,
      "learning_rate": 0.0004626498141029357,
      "loss": 2.2653,
      "step": 73177
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.7064598202705383,
      "learning_rate": 0.0004626463769055303,
      "loss": 2.1488,
      "step": 73178
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.7143500447273254,
      "learning_rate": 0.00046264293967788594,
      "loss": 2.1916,
      "step": 73179
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.7775919437408447,
      "learning_rate": 0.0004626395024200032,
      "loss": 2.0652,
      "step": 73180
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.7456152439117432,
      "learning_rate": 0.0004626360651318827,
      "loss": 2.3332,
      "step": 73181
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.7574406266212463,
      "learning_rate": 0.000462632627813525,
      "loss": 2.329,
      "step": 73182
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.7402242422103882,
      "learning_rate": 0.00046262919046493093,
      "loss": 2.2859,
      "step": 73183
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.7979724407196045,
      "learning_rate": 0.00046262575308610095,
      "loss": 2.0319,
      "step": 73184
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.737185537815094,
      "learning_rate": 0.00046262231567703577,
      "loss": 2.2466,
      "step": 73185
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.7781262397766113,
      "learning_rate": 0.0004626188782377361,
      "loss": 1.9556,
      "step": 73186
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.8023853898048401,
      "learning_rate": 0.00046261544076820247,
      "loss": 1.9974,
      "step": 73187
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.7674457430839539,
      "learning_rate": 0.0004626120032684355,
      "loss": 2.0294,
      "step": 73188
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.880018949508667,
      "learning_rate": 0.000462608565738436,
      "loss": 2.2473,
      "step": 73189
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.7523432374000549,
      "learning_rate": 0.0004626051281782045,
      "loss": 2.0046,
      "step": 73190
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.7561016082763672,
      "learning_rate": 0.0004626016905877415,
      "loss": 2.0916,
      "step": 73191
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.7897225618362427,
      "learning_rate": 0.00046259825296704785,
      "loss": 2.2893,
      "step": 73192
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.7250248193740845,
      "learning_rate": 0.00046259481531612417,
      "loss": 2.2925,
      "step": 73193
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.7822798490524292,
      "learning_rate": 0.000462591377634971,
      "loss": 2.0376,
      "step": 73194
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.8753350973129272,
      "learning_rate": 0.00046258793992358903,
      "loss": 2.4321,
      "step": 73195
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.7393550276756287,
      "learning_rate": 0.0004625845021819789,
      "loss": 2.1914,
      "step": 73196
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.809486985206604,
      "learning_rate": 0.0004625810644101413,
      "loss": 2.3005,
      "step": 73197
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.7909843325614929,
      "learning_rate": 0.00046257762660807666,
      "loss": 2.4234,
      "step": 73198
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.6871389746665955,
      "learning_rate": 0.0004625741887757858,
      "loss": 2.3438,
      "step": 73199
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.8366920948028564,
      "learning_rate": 0.00046257075091326943,
      "loss": 2.1333,
      "step": 73200
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.7455968856811523,
      "learning_rate": 0.00046256731302052805,
      "loss": 2.2106,
      "step": 73201
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.8704331517219543,
      "learning_rate": 0.0004625638750975623,
      "loss": 2.302,
      "step": 73202
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.7286500334739685,
      "learning_rate": 0.0004625604371443729,
      "loss": 2.2936,
      "step": 73203
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.7053749561309814,
      "learning_rate": 0.00046255699916096044,
      "loss": 2.1185,
      "step": 73204
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.8672438859939575,
      "learning_rate": 0.00046255356114732553,
      "loss": 2.0444,
      "step": 73205
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.8016873598098755,
      "learning_rate": 0.0004625501231034689,
      "loss": 2.3219,
      "step": 73206
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.8061704635620117,
      "learning_rate": 0.00046254668502939116,
      "loss": 2.251,
      "step": 73207
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.7012339234352112,
      "learning_rate": 0.00046254324692509283,
      "loss": 2.1842,
      "step": 73208
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.8702475428581238,
      "learning_rate": 0.0004625398087905747,
      "loss": 2.1911,
      "step": 73209
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.7558903694152832,
      "learning_rate": 0.00046253637062583736,
      "loss": 2.1937,
      "step": 73210
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.8321119546890259,
      "learning_rate": 0.00046253293243088135,
      "loss": 2.3652,
      "step": 73211
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.7648402452468872,
      "learning_rate": 0.0004625294942057075,
      "loss": 2.1597,
      "step": 73212
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.7378942370414734,
      "learning_rate": 0.0004625260559503163,
      "loss": 1.9993,
      "step": 73213
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.7485141158103943,
      "learning_rate": 0.00046252261766470843,
      "loss": 2.2143,
      "step": 73214
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.7213489413261414,
      "learning_rate": 0.0004625191793488846,
      "loss": 2.4716,
      "step": 73215
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.8086429238319397,
      "learning_rate": 0.0004625157410028454,
      "loss": 2.0982,
      "step": 73216
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.7510873675346375,
      "learning_rate": 0.0004625123026265913,
      "loss": 2.3152,
      "step": 73217
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.7473721504211426,
      "learning_rate": 0.0004625088642201232,
      "loss": 2.1448,
      "step": 73218
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.8085628747940063,
      "learning_rate": 0.00046250542578344164,
      "loss": 2.1862,
      "step": 73219
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.774105429649353,
      "learning_rate": 0.00046250198731654724,
      "loss": 2.3456,
      "step": 73220
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.7083930373191833,
      "learning_rate": 0.0004624985488194407,
      "loss": 2.3276,
      "step": 73221
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.8845906257629395,
      "learning_rate": 0.0004624951102921225,
      "loss": 2.3217,
      "step": 73222
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.7415181994438171,
      "learning_rate": 0.00046249167173459355,
      "loss": 2.0356,
      "step": 73223
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.7781663537025452,
      "learning_rate": 0.0004624882331468542,
      "loss": 2.2918,
      "step": 73224
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.749614953994751,
      "learning_rate": 0.00046248479452890527,
      "loss": 2.3806,
      "step": 73225
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.7577902674674988,
      "learning_rate": 0.00046248135588074734,
      "loss": 2.3338,
      "step": 73226
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.7484720349311829,
      "learning_rate": 0.00046247791720238103,
      "loss": 2.2765,
      "step": 73227
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.7338306903839111,
      "learning_rate": 0.00046247447849380707,
      "loss": 2.2934,
      "step": 73228
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.7274601459503174,
      "learning_rate": 0.000462471039755026,
      "loss": 2.2875,
      "step": 73229
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.7699628472328186,
      "learning_rate": 0.0004624676009860385,
      "loss": 2.0656,
      "step": 73230
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.769131064414978,
      "learning_rate": 0.0004624641621868452,
      "loss": 2.3087,
      "step": 73231
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.7334089875221252,
      "learning_rate": 0.0004624607233574468,
      "loss": 2.2986,
      "step": 73232
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.7474072575569153,
      "learning_rate": 0.0004624572844978438,
      "loss": 2.061,
      "step": 73233
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.7306651473045349,
      "learning_rate": 0.0004624538456080369,
      "loss": 2.2789,
      "step": 73234
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.7232757210731506,
      "learning_rate": 0.00046245040668802683,
      "loss": 2.029,
      "step": 73235
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.7418293356895447,
      "learning_rate": 0.00046244696773781424,
      "loss": 2.1334,
      "step": 73236
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.7624145746231079,
      "learning_rate": 0.00046244352875739956,
      "loss": 2.0339,
      "step": 73237
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.7827640175819397,
      "learning_rate": 0.0004624400897467837,
      "loss": 1.9688,
      "step": 73238
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.8325309753417969,
      "learning_rate": 0.000462436650705967,
      "loss": 2.1887,
      "step": 73239
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.793793797492981,
      "learning_rate": 0.0004624332116349504,
      "loss": 2.2952,
      "step": 73240
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.8399654626846313,
      "learning_rate": 0.0004624297725337343,
      "loss": 2.0358,
      "step": 73241
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.9152366518974304,
      "learning_rate": 0.00046242633340231947,
      "loss": 2.1594,
      "step": 73242
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.6919643878936768,
      "learning_rate": 0.0004624228942407065,
      "loss": 2.3966,
      "step": 73243
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.8348367810249329,
      "learning_rate": 0.00046241945504889607,
      "loss": 2.2665,
      "step": 73244
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.2198656797409058,
      "learning_rate": 0.0004624160158268888,
      "loss": 2.249,
      "step": 73245
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.7142916321754456,
      "learning_rate": 0.00046241257657468534,
      "loss": 2.4084,
      "step": 73246
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.825300931930542,
      "learning_rate": 0.0004624091372922863,
      "loss": 2.1648,
      "step": 73247
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.0667041540145874,
      "learning_rate": 0.0004624056979796923,
      "loss": 2.1247,
      "step": 73248
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.9001132845878601,
      "learning_rate": 0.0004624022586369041,
      "loss": 2.0589,
      "step": 73249
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.7478424310684204,
      "learning_rate": 0.0004623988192639222,
      "loss": 2.1152,
      "step": 73250
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.3252153396606445,
      "learning_rate": 0.0004623953798607473,
      "loss": 2.1872,
      "step": 73251
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.061913013458252,
      "learning_rate": 0.00046239194042737994,
      "loss": 2.2463,
      "step": 73252
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.7040835618972778,
      "learning_rate": 0.0004623885009638211,
      "loss": 2.1949,
      "step": 73253
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.9766073822975159,
      "learning_rate": 0.000462385061470071,
      "loss": 2.24,
      "step": 73254
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.8295583724975586,
      "learning_rate": 0.0004623816219461304,
      "loss": 2.1249,
      "step": 73255
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.8283981680870056,
      "learning_rate": 0.0004623781823920001,
      "loss": 2.1957,
      "step": 73256
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.7600148320198059,
      "learning_rate": 0.0004623747428076806,
      "loss": 2.5031,
      "step": 73257
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.8718411326408386,
      "learning_rate": 0.00046237130319317255,
      "loss": 2.345,
      "step": 73258
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.9206346869468689,
      "learning_rate": 0.0004623678635484767,
      "loss": 2.1663,
      "step": 73259
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.9275092482566833,
      "learning_rate": 0.00046236442387359344,
      "loss": 1.9969,
      "step": 73260
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.6697195172309875,
      "learning_rate": 0.0004623609841685237,
      "loss": 1.8538,
      "step": 73261
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.1160075664520264,
      "learning_rate": 0.000462357544433268,
      "loss": 2.2193,
      "step": 73262
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.8445332050323486,
      "learning_rate": 0.0004623541046678268,
      "loss": 2.4584,
      "step": 73263
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.8609711527824402,
      "learning_rate": 0.000462350664872201,
      "loss": 2.0935,
      "step": 73264
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.7580488324165344,
      "learning_rate": 0.00046234722504639124,
      "loss": 2.0976,
      "step": 73265
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.1134068965911865,
      "learning_rate": 0.000462343785190398,
      "loss": 2.1831,
      "step": 73266
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.9324073791503906,
      "learning_rate": 0.00046234034530422195,
      "loss": 1.9656,
      "step": 73267
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.7060473561286926,
      "learning_rate": 0.0004623369053878638,
      "loss": 2.1822,
      "step": 73268
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.257421851158142,
      "learning_rate": 0.0004623334654413241,
      "loss": 1.9661,
      "step": 73269
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.8652709722518921,
      "learning_rate": 0.0004623300254646036,
      "loss": 2.186,
      "step": 73270
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.8008180856704712,
      "learning_rate": 0.00046232658545770287,
      "loss": 2.2028,
      "step": 73271
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.8768123388290405,
      "learning_rate": 0.0004623231454206226,
      "loss": 2.1882,
      "step": 73272
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.7861962914466858,
      "learning_rate": 0.0004623197053533633,
      "loss": 2.177,
      "step": 73273
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.7947684526443481,
      "learning_rate": 0.0004623162652559258,
      "loss": 2.2068,
      "step": 73274
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.7376096248626709,
      "learning_rate": 0.0004623128251283106,
      "loss": 2.1607,
      "step": 73275
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.8560680150985718,
      "learning_rate": 0.0004623093849705184,
      "loss": 2.2859,
      "step": 73276
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.7990529537200928,
      "learning_rate": 0.00046230594478254976,
      "loss": 2.0842,
      "step": 73277
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.8415324687957764,
      "learning_rate": 0.0004623025045644055,
      "loss": 2.4634,
      "step": 73278
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.7823217511177063,
      "learning_rate": 0.000462299064316086,
      "loss": 2.3076,
      "step": 73279
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.7763420939445496,
      "learning_rate": 0.00046229562403759215,
      "loss": 2.2295,
      "step": 73280
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.7268032431602478,
      "learning_rate": 0.00046229218372892443,
      "loss": 2.3773,
      "step": 73281
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.7215436697006226,
      "learning_rate": 0.00046228874339008353,
      "loss": 2.1031,
      "step": 73282
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.8162051439285278,
      "learning_rate": 0.0004622853030210701,
      "loss": 1.9618,
      "step": 73283
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.8183767199516296,
      "learning_rate": 0.0004622818626218848,
      "loss": 2.357,
      "step": 73284
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.7315323948860168,
      "learning_rate": 0.0004622784221925282,
      "loss": 2.1309,
      "step": 73285
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.7699094414710999,
      "learning_rate": 0.000462274981733001,
      "loss": 2.4144,
      "step": 73286
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.748573362827301,
      "learning_rate": 0.0004622715412433038,
      "loss": 2.3393,
      "step": 73287
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.7373681664466858,
      "learning_rate": 0.0004622681007234372,
      "loss": 2.1193,
      "step": 73288
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.7085955739021301,
      "learning_rate": 0.00046226466017340205,
      "loss": 2.232,
      "step": 73289
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.8042566776275635,
      "learning_rate": 0.00046226121959319866,
      "loss": 2.1029,
      "step": 73290
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.8177739977836609,
      "learning_rate": 0.0004622577789828279,
      "loss": 2.4473,
      "step": 73291
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.9465925693511963,
      "learning_rate": 0.0004622543383422905,
      "loss": 2.2093,
      "step": 73292
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.7153947949409485,
      "learning_rate": 0.0004622508976715868,
      "loss": 2.4676,
      "step": 73293
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.7531834244728088,
      "learning_rate": 0.0004622474569707176,
      "loss": 2.1436,
      "step": 73294
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.812261700630188,
      "learning_rate": 0.0004622440162396836,
      "loss": 2.2436,
      "step": 73295
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.7368202209472656,
      "learning_rate": 0.00046224057547848533,
      "loss": 2.1291,
      "step": 73296
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.7531517744064331,
      "learning_rate": 0.0004622371346871235,
      "loss": 2.0627,
      "step": 73297
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.7785093784332275,
      "learning_rate": 0.00046223369386559865,
      "loss": 2.2248,
      "step": 73298
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.8365461230278015,
      "learning_rate": 0.00046223025301391163,
      "loss": 2.2761,
      "step": 73299
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.8087366223335266,
      "learning_rate": 0.00046222681213206287,
      "loss": 2.205,
      "step": 73300
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.8246921300888062,
      "learning_rate": 0.00046222337122005307,
      "loss": 2.1896,
      "step": 73301
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.726635217666626,
      "learning_rate": 0.0004622199302778829,
      "loss": 1.9631,
      "step": 73302
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.7962785363197327,
      "learning_rate": 0.000462216489305553,
      "loss": 2.3318,
      "step": 73303
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.7872145771980286,
      "learning_rate": 0.00046221304830306397,
      "loss": 2.2312,
      "step": 73304
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.7726721167564392,
      "learning_rate": 0.0004622096072704165,
      "loss": 2.2261,
      "step": 73305
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.9065912365913391,
      "learning_rate": 0.00046220616620761115,
      "loss": 2.2399,
      "step": 73306
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.8675181865692139,
      "learning_rate": 0.0004622027251146486,
      "loss": 2.2774,
      "step": 73307
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.7734659910202026,
      "learning_rate": 0.00046219928399152956,
      "loss": 2.2113,
      "step": 73308
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.7970399856567383,
      "learning_rate": 0.0004621958428382546,
      "loss": 2.2884,
      "step": 73309
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.7350096702575684,
      "learning_rate": 0.00046219240165482436,
      "loss": 1.9244,
      "step": 73310
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.7344797849655151,
      "learning_rate": 0.00046218896044123955,
      "loss": 2.1428,
      "step": 73311
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.0548210144042969,
      "learning_rate": 0.0004621855191975007,
      "loss": 2.2371,
      "step": 73312
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.7441035509109497,
      "learning_rate": 0.0004621820779236085,
      "loss": 2.1283,
      "step": 73313
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.8470118045806885,
      "learning_rate": 0.00046217863661956354,
      "loss": 2.0359,
      "step": 73314
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.7921443581581116,
      "learning_rate": 0.0004621751952853666,
      "loss": 2.277,
      "step": 73315
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.7913528084754944,
      "learning_rate": 0.00046217175392101816,
      "loss": 2.1446,
      "step": 73316
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.8242061138153076,
      "learning_rate": 0.00046216831252651893,
      "loss": 2.2271,
      "step": 73317
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.7145015597343445,
      "learning_rate": 0.0004621648711018696,
      "loss": 2.0408,
      "step": 73318
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.8845381140708923,
      "learning_rate": 0.0004621614296470707,
      "loss": 2.4052,
      "step": 73319
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.7569249868392944,
      "learning_rate": 0.00046215798816212297,
      "loss": 2.058,
      "step": 73320
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.8097983598709106,
      "learning_rate": 0.000462154546647027,
      "loss": 2.3411,
      "step": 73321
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.8177009224891663,
      "learning_rate": 0.00046215110510178345,
      "loss": 2.4535,
      "step": 73322
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.854503333568573,
      "learning_rate": 0.00046214766352639295,
      "loss": 2.312,
      "step": 73323
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.8403397798538208,
      "learning_rate": 0.0004621442219208562,
      "loss": 2.2579,
      "step": 73324
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.7696805000305176,
      "learning_rate": 0.00046214078028517365,
      "loss": 2.0919,
      "step": 73325
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.8417826294898987,
      "learning_rate": 0.0004621373386193461,
      "loss": 2.3194,
      "step": 73326
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.8819641470909119,
      "learning_rate": 0.0004621338969233742,
      "loss": 2.2646,
      "step": 73327
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.8044407367706299,
      "learning_rate": 0.00046213045519725857,
      "loss": 2.3576,
      "step": 73328
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.783929169178009,
      "learning_rate": 0.0004621270134409998,
      "loss": 2.2855,
      "step": 73329
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.8410902619361877,
      "learning_rate": 0.0004621235716545985,
      "loss": 2.135,
      "step": 73330
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.0676822662353516,
      "learning_rate": 0.0004621201298380554,
      "loss": 2.1525,
      "step": 73331
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.7465527653694153,
      "learning_rate": 0.0004621166879913711,
      "loss": 2.3975,
      "step": 73332
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.7876469492912292,
      "learning_rate": 0.00046211324611454633,
      "loss": 2.2927,
      "step": 73333
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.8539506793022156,
      "learning_rate": 0.0004621098042075816,
      "loss": 2.1448,
      "step": 73334
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.2472596168518066,
      "learning_rate": 0.0004621063622704776,
      "loss": 2.2797,
      "step": 73335
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.757917582988739,
      "learning_rate": 0.0004621029203032349,
      "loss": 2.1464,
      "step": 73336
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.7777174711227417,
      "learning_rate": 0.0004620994783058543,
      "loss": 2.0818,
      "step": 73337
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.9220550656318665,
      "learning_rate": 0.00046209603627833635,
      "loss": 2.2333,
      "step": 73338
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.4520249366760254,
      "learning_rate": 0.00046209259422068166,
      "loss": 2.1064,
      "step": 73339
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.9018977880477905,
      "learning_rate": 0.0004620891521328908,
      "loss": 2.4316,
      "step": 73340
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.958323061466217,
      "learning_rate": 0.0004620857100149646,
      "loss": 2.1281,
      "step": 73341
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.8710836172103882,
      "learning_rate": 0.00046208226786690357,
      "loss": 2.1118,
      "step": 73342
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.9278813004493713,
      "learning_rate": 0.0004620788256887084,
      "loss": 2.3416,
      "step": 73343
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.7588797807693481,
      "learning_rate": 0.00046207538348037977,
      "loss": 2.1739,
      "step": 73344
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.9613377451896667,
      "learning_rate": 0.0004620719412419182,
      "loss": 2.2577,
      "step": 73345
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.8928102254867554,
      "learning_rate": 0.0004620684989733244,
      "loss": 2.1902,
      "step": 73346
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7736641764640808,
      "learning_rate": 0.00046206505667459915,
      "loss": 2.1844,
      "step": 73347
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7992275953292847,
      "learning_rate": 0.0004620616143457428,
      "loss": 2.2422,
      "step": 73348
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.0876384973526,
      "learning_rate": 0.0004620581719867561,
      "loss": 2.2434,
      "step": 73349
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7815670371055603,
      "learning_rate": 0.00046205472959763984,
      "loss": 2.3734,
      "step": 73350
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.8026976585388184,
      "learning_rate": 0.0004620512871783945,
      "loss": 2.1816,
      "step": 73351
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7283380031585693,
      "learning_rate": 0.0004620478447290207,
      "loss": 2.2782,
      "step": 73352
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.8355032801628113,
      "learning_rate": 0.0004620444022495193,
      "loss": 2.4103,
      "step": 73353
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.829400897026062,
      "learning_rate": 0.0004620409597398907,
      "loss": 2.4188,
      "step": 73354
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7456133961677551,
      "learning_rate": 0.0004620375172001356,
      "loss": 2.2351,
      "step": 73355
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.04219651222229,
      "learning_rate": 0.0004620340746302547,
      "loss": 2.0964,
      "step": 73356
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.9252164959907532,
      "learning_rate": 0.00046203063203024855,
      "loss": 2.4597,
      "step": 73357
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.8561058640480042,
      "learning_rate": 0.0004620271894001179,
      "loss": 2.3226,
      "step": 73358
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7611480951309204,
      "learning_rate": 0.0004620237467398634,
      "loss": 2.3546,
      "step": 73359
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.8776301145553589,
      "learning_rate": 0.00046202030404948557,
      "loss": 2.3568,
      "step": 73360
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7978928089141846,
      "learning_rate": 0.000462016861328985,
      "loss": 2.2371,
      "step": 73361
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.9059807658195496,
      "learning_rate": 0.00046201341857836256,
      "loss": 2.164,
      "step": 73362
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.6831046342849731,
      "learning_rate": 0.00046200997579761875,
      "loss": 2.0842,
      "step": 73363
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.8265053629875183,
      "learning_rate": 0.0004620065329867542,
      "loss": 2.2161,
      "step": 73364
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.9387851357460022,
      "learning_rate": 0.00046200309014576965,
      "loss": 2.2301,
      "step": 73365
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7289637923240662,
      "learning_rate": 0.00046199964727466555,
      "loss": 2.1917,
      "step": 73366
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7777484655380249,
      "learning_rate": 0.00046199620437344276,
      "loss": 2.2986,
      "step": 73367
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.8222950100898743,
      "learning_rate": 0.00046199276144210174,
      "loss": 2.2256,
      "step": 73368
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7861115336418152,
      "learning_rate": 0.0004619893184806433,
      "loss": 2.0886,
      "step": 73369
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.8828309178352356,
      "learning_rate": 0.0004619858754890679,
      "loss": 2.1332,
      "step": 73370
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.9228705167770386,
      "learning_rate": 0.0004619824324673763,
      "loss": 2.0096,
      "step": 73371
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.8070431351661682,
      "learning_rate": 0.00046197898941556916,
      "loss": 2.2216,
      "step": 73372
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.8016660809516907,
      "learning_rate": 0.00046197554633364694,
      "loss": 1.9392,
      "step": 73373
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.015015959739685,
      "learning_rate": 0.0004619721032216105,
      "loss": 2.1883,
      "step": 73374
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7746247053146362,
      "learning_rate": 0.00046196866007946043,
      "loss": 2.051,
      "step": 73375
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.9717579483985901,
      "learning_rate": 0.0004619652169071973,
      "loss": 2.1748,
      "step": 73376
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.8460226058959961,
      "learning_rate": 0.00046196177370482166,
      "loss": 2.209,
      "step": 73377
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.805824339389801,
      "learning_rate": 0.0004619583304723344,
      "loss": 2.2281,
      "step": 73378
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7787501811981201,
      "learning_rate": 0.000461954887209736,
      "loss": 2.1737,
      "step": 73379
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.9919118285179138,
      "learning_rate": 0.00046195144391702705,
      "loss": 2.2614,
      "step": 73380
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.8633220791816711,
      "learning_rate": 0.00046194800059420844,
      "loss": 2.1057,
      "step": 73381
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7600938677787781,
      "learning_rate": 0.0004619445572412805,
      "loss": 2.2124,
      "step": 73382
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7653621435165405,
      "learning_rate": 0.000461941113858244,
      "loss": 2.1757,
      "step": 73383
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7971150875091553,
      "learning_rate": 0.0004619376704450997,
      "loss": 2.1612,
      "step": 73384
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.785261332988739,
      "learning_rate": 0.00046193422700184804,
      "loss": 2.3458,
      "step": 73385
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7585919499397278,
      "learning_rate": 0.0004619307835284898,
      "loss": 2.1857,
      "step": 73386
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.8025751709938049,
      "learning_rate": 0.0004619273400250255,
      "loss": 2.381,
      "step": 73387
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7033371329307556,
      "learning_rate": 0.000461923896491456,
      "loss": 2.3853,
      "step": 73388
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.758948802947998,
      "learning_rate": 0.00046192045292778165,
      "loss": 2.3641,
      "step": 73389
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.6779574155807495,
      "learning_rate": 0.00046191700933400325,
      "loss": 2.1189,
      "step": 73390
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7771066427230835,
      "learning_rate": 0.0004619135657101215,
      "loss": 2.1793,
      "step": 73391
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.806504487991333,
      "learning_rate": 0.00046191012205613685,
      "loss": 2.1517,
      "step": 73392
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.8239536881446838,
      "learning_rate": 0.00046190667837205015,
      "loss": 2.306,
      "step": 73393
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.8062826991081238,
      "learning_rate": 0.0004619032346578618,
      "loss": 2.1569,
      "step": 73394
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.84310382604599,
      "learning_rate": 0.00046189979091357274,
      "loss": 2.3872,
      "step": 73395
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.8036121726036072,
      "learning_rate": 0.00046189634713918344,
      "loss": 2.1098,
      "step": 73396
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7620709538459778,
      "learning_rate": 0.0004618929033346944,
      "loss": 2.2694,
      "step": 73397
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7684076428413391,
      "learning_rate": 0.0004618894595001065,
      "loss": 2.2252,
      "step": 73398
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7444130778312683,
      "learning_rate": 0.0004618860156354204,
      "loss": 2.2585,
      "step": 73399
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.8081053495407104,
      "learning_rate": 0.00046188257174063654,
      "loss": 2.3393,
      "step": 73400
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.6977502107620239,
      "learning_rate": 0.0004618791278157557,
      "loss": 2.354,
      "step": 73401
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.78331059217453,
      "learning_rate": 0.00046187568386077844,
      "loss": 2.1046,
      "step": 73402
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.8359408378601074,
      "learning_rate": 0.00046187223987570534,
      "loss": 2.0702,
      "step": 73403
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.8143748641014099,
      "learning_rate": 0.00046186879586053724,
      "loss": 2.1988,
      "step": 73404
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7575626373291016,
      "learning_rate": 0.0004618653518152747,
      "loss": 2.3019,
      "step": 73405
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.8276402950286865,
      "learning_rate": 0.00046186190773991827,
      "loss": 2.1583,
      "step": 73406
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.8719186782836914,
      "learning_rate": 0.0004618584636344687,
      "loss": 2.1288,
      "step": 73407
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7517348527908325,
      "learning_rate": 0.0004618550194989266,
      "loss": 2.1694,
      "step": 73408
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7610933184623718,
      "learning_rate": 0.0004618515753332926,
      "loss": 2.2845,
      "step": 73409
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.8971090912818909,
      "learning_rate": 0.0004618481311375672,
      "loss": 2.231,
      "step": 73410
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.8255674839019775,
      "learning_rate": 0.0004618446869117514,
      "loss": 2.1629,
      "step": 73411
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7485834956169128,
      "learning_rate": 0.0004618412426558455,
      "loss": 2.3169,
      "step": 73412
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.8324899673461914,
      "learning_rate": 0.0004618377983698502,
      "loss": 2.338,
      "step": 73413
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7669681310653687,
      "learning_rate": 0.00046183435405376634,
      "loss": 2.2464,
      "step": 73414
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.0473105907440186,
      "learning_rate": 0.0004618309097075943,
      "loss": 1.9754,
      "step": 73415
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7780943512916565,
      "learning_rate": 0.0004618274653313348,
      "loss": 2.0183,
      "step": 73416
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7646212577819824,
      "learning_rate": 0.0004618240209249887,
      "loss": 2.2221,
      "step": 73417
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.8229844570159912,
      "learning_rate": 0.0004618205764885564,
      "loss": 2.2597,
      "step": 73418
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7901668548583984,
      "learning_rate": 0.0004618171320220385,
      "loss": 2.2495,
      "step": 73419
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.830308735370636,
      "learning_rate": 0.00046181368752543577,
      "loss": 2.0291,
      "step": 73420
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.8429433107376099,
      "learning_rate": 0.00046181024299874883,
      "loss": 2.1993,
      "step": 73421
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7489988207817078,
      "learning_rate": 0.0004618067984419783,
      "loss": 1.909,
      "step": 73422
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7903096675872803,
      "learning_rate": 0.00046180335385512494,
      "loss": 2.46,
      "step": 73423
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.9733963012695312,
      "learning_rate": 0.0004617999092381891,
      "loss": 2.2465,
      "step": 73424
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7815215587615967,
      "learning_rate": 0.0004617964645911718,
      "loss": 2.2987,
      "step": 73425
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7593926787376404,
      "learning_rate": 0.00046179301991407343,
      "loss": 2.1174,
      "step": 73426
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.8176271319389343,
      "learning_rate": 0.0004617895752068946,
      "loss": 2.4287,
      "step": 73427
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.8471730947494507,
      "learning_rate": 0.00046178613046963607,
      "loss": 2.1047,
      "step": 73428
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.588090181350708,
      "learning_rate": 0.0004617826857022984,
      "loss": 2.1965,
      "step": 73429
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7693981528282166,
      "learning_rate": 0.00046177924090488236,
      "loss": 2.1729,
      "step": 73430
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7377070188522339,
      "learning_rate": 0.00046177579607738846,
      "loss": 2.3183,
      "step": 73431
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.9064460396766663,
      "learning_rate": 0.0004617723512198174,
      "loss": 2.2312,
      "step": 73432
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7931095361709595,
      "learning_rate": 0.0004617689063321698,
      "loss": 2.4556,
      "step": 73433
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7735304236412048,
      "learning_rate": 0.0004617654614144463,
      "loss": 2.0365,
      "step": 73434
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.8933286070823669,
      "learning_rate": 0.00046176201646664755,
      "loss": 2.1622,
      "step": 73435
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.806702196598053,
      "learning_rate": 0.0004617585714887742,
      "loss": 2.1496,
      "step": 73436
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7740262746810913,
      "learning_rate": 0.00046175512648082684,
      "loss": 2.1598,
      "step": 73437
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7585229277610779,
      "learning_rate": 0.0004617516814428062,
      "loss": 2.123,
      "step": 73438
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.8133217096328735,
      "learning_rate": 0.00046174823637471293,
      "loss": 2.2948,
      "step": 73439
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7827340960502625,
      "learning_rate": 0.00046174479127654753,
      "loss": 2.2974,
      "step": 73440
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.8135082125663757,
      "learning_rate": 0.00046174134614831064,
      "loss": 2.1607,
      "step": 73441
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.8017542362213135,
      "learning_rate": 0.00046173790099000313,
      "loss": 2.1769,
      "step": 73442
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.9993951916694641,
      "learning_rate": 0.00046173445580162533,
      "loss": 2.3044,
      "step": 73443
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7785212397575378,
      "learning_rate": 0.00046173101058317816,
      "loss": 2.1647,
      "step": 73444
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.8540835380554199,
      "learning_rate": 0.0004617275653346621,
      "loss": 2.1663,
      "step": 73445
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.9789422750473022,
      "learning_rate": 0.0004617241200560779,
      "loss": 2.242,
      "step": 73446
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7310588955879211,
      "learning_rate": 0.00046172067474742605,
      "loss": 2.1508,
      "step": 73447
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7613518834114075,
      "learning_rate": 0.00046171722940870736,
      "loss": 2.1471,
      "step": 73448
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.9604348540306091,
      "learning_rate": 0.0004617137840399223,
      "loss": 2.3523,
      "step": 73449
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.9186525940895081,
      "learning_rate": 0.00046171033864107156,
      "loss": 2.2063,
      "step": 73450
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.8462467193603516,
      "learning_rate": 0.000461706893212156,
      "loss": 2.0622,
      "step": 73451
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.8553269505500793,
      "learning_rate": 0.0004617034477531759,
      "loss": 2.341,
      "step": 73452
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.096108317375183,
      "learning_rate": 0.0004617000022641321,
      "loss": 2.1556,
      "step": 73453
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.8446018099784851,
      "learning_rate": 0.00046169655674502524,
      "loss": 2.2304,
      "step": 73454
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7695085406303406,
      "learning_rate": 0.00046169311119585596,
      "loss": 2.24,
      "step": 73455
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.0395376682281494,
      "learning_rate": 0.0004616896656166248,
      "loss": 2.4788,
      "step": 73456
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.8284326791763306,
      "learning_rate": 0.00046168622000733266,
      "loss": 2.1298,
      "step": 73457
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.864573061466217,
      "learning_rate": 0.0004616827743679799,
      "loss": 2.262,
      "step": 73458
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.8324002623558044,
      "learning_rate": 0.00046167932869856717,
      "loss": 2.1085,
      "step": 73459
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.9476035833358765,
      "learning_rate": 0.00046167588299909527,
      "loss": 2.2729,
      "step": 73460
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.8868224620819092,
      "learning_rate": 0.0004616724372695648,
      "loss": 2.3372,
      "step": 73461
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.830653727054596,
      "learning_rate": 0.0004616689915099763,
      "loss": 2.2317,
      "step": 73462
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.8591489791870117,
      "learning_rate": 0.00046166554572033057,
      "loss": 2.0488,
      "step": 73463
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7661468386650085,
      "learning_rate": 0.0004616620999006282,
      "loss": 2.1519,
      "step": 73464
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7597799301147461,
      "learning_rate": 0.00046165865405086966,
      "loss": 2.1044,
      "step": 73465
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.8736942410469055,
      "learning_rate": 0.00046165520817105586,
      "loss": 2.3451,
      "step": 73466
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.8274970054626465,
      "learning_rate": 0.0004616517622611872,
      "loss": 2.3094,
      "step": 73467
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7718225717544556,
      "learning_rate": 0.00046164831632126437,
      "loss": 2.3356,
      "step": 73468
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.8228753805160522,
      "learning_rate": 0.0004616448703512882,
      "loss": 2.3317,
      "step": 73469
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.8499679565429688,
      "learning_rate": 0.00046164142435125915,
      "loss": 1.9803,
      "step": 73470
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7429923415184021,
      "learning_rate": 0.0004616379783211779,
      "loss": 2.3672,
      "step": 73471
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.8798841238021851,
      "learning_rate": 0.0004616345322610451,
      "loss": 2.1985,
      "step": 73472
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.6874468922615051,
      "learning_rate": 0.0004616310861708614,
      "loss": 2.0944,
      "step": 73473
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.8103340864181519,
      "learning_rate": 0.0004616276400506274,
      "loss": 2.0392,
      "step": 73474
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.8499125838279724,
      "learning_rate": 0.0004616241939003438,
      "loss": 2.471,
      "step": 73475
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7241716980934143,
      "learning_rate": 0.0004616207477200113,
      "loss": 2.2452,
      "step": 73476
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.8011030554771423,
      "learning_rate": 0.0004616173015096303,
      "loss": 2.1855,
      "step": 73477
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.8091766238212585,
      "learning_rate": 0.00046161385526920164,
      "loss": 2.2624,
      "step": 73478
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.8451868295669556,
      "learning_rate": 0.0004616104089987259,
      "loss": 2.0273,
      "step": 73479
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7996892929077148,
      "learning_rate": 0.00046160696269820375,
      "loss": 2.3596,
      "step": 73480
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7381299138069153,
      "learning_rate": 0.00046160351636763587,
      "loss": 2.1791,
      "step": 73481
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.8245052099227905,
      "learning_rate": 0.00046160007000702275,
      "loss": 2.4995,
      "step": 73482
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7297082543373108,
      "learning_rate": 0.0004615966236163652,
      "loss": 2.2312,
      "step": 73483
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.8012659549713135,
      "learning_rate": 0.0004615931771956638,
      "loss": 2.4228,
      "step": 73484
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7597938776016235,
      "learning_rate": 0.00046158973074491904,
      "loss": 2.048,
      "step": 73485
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7346914410591125,
      "learning_rate": 0.0004615862842641318,
      "loss": 2.2081,
      "step": 73486
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7419808506965637,
      "learning_rate": 0.0004615828377533027,
      "loss": 2.2525,
      "step": 73487
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.8054779767990112,
      "learning_rate": 0.0004615793912124322,
      "loss": 2.1873,
      "step": 73488
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7107006311416626,
      "learning_rate": 0.00046157594464152097,
      "loss": 2.121,
      "step": 73489
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7743285298347473,
      "learning_rate": 0.00046157249804056993,
      "loss": 2.328,
      "step": 73490
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7651591300964355,
      "learning_rate": 0.00046156905140957934,
      "loss": 2.3741,
      "step": 73491
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.8172095417976379,
      "learning_rate": 0.0004615656047485501,
      "loss": 2.4646,
      "step": 73492
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7829270362854004,
      "learning_rate": 0.00046156215805748273,
      "loss": 2.2897,
      "step": 73493
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7452484965324402,
      "learning_rate": 0.0004615587113363779,
      "loss": 2.2203,
      "step": 73494
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.1247076988220215,
      "learning_rate": 0.00046155526458523626,
      "loss": 2.341,
      "step": 73495
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.820205807685852,
      "learning_rate": 0.0004615518178040585,
      "loss": 2.2795,
      "step": 73496
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7676188349723816,
      "learning_rate": 0.00046154837099284516,
      "loss": 2.1247,
      "step": 73497
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7834953665733337,
      "learning_rate": 0.0004615449241515969,
      "loss": 2.2281,
      "step": 73498
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7729787230491638,
      "learning_rate": 0.0004615414772803144,
      "loss": 2.2105,
      "step": 73499
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7380210161209106,
      "learning_rate": 0.0004615380303789984,
      "loss": 2.2489,
      "step": 73500
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.79835045337677,
      "learning_rate": 0.00046153458344764923,
      "loss": 2.3844,
      "step": 73501
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7041990160942078,
      "learning_rate": 0.00046153113648626787,
      "loss": 2.1884,
      "step": 73502
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.8177814483642578,
      "learning_rate": 0.0004615276894948548,
      "loss": 2.1904,
      "step": 73503
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7126762270927429,
      "learning_rate": 0.00046152424247341065,
      "loss": 2.159,
      "step": 73504
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7412828207015991,
      "learning_rate": 0.00046152079542193615,
      "loss": 1.9707,
      "step": 73505
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.801050066947937,
      "learning_rate": 0.00046151734834043185,
      "loss": 2.2997,
      "step": 73506
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.8066124320030212,
      "learning_rate": 0.0004615139012288984,
      "loss": 2.4694,
      "step": 73507
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7886311411857605,
      "learning_rate": 0.0004615104540873365,
      "loss": 2.1838,
      "step": 73508
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7055298686027527,
      "learning_rate": 0.00046150700691574684,
      "loss": 2.2771,
      "step": 73509
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7946739196777344,
      "learning_rate": 0.0004615035597141299,
      "loss": 2.4289,
      "step": 73510
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.9521177411079407,
      "learning_rate": 0.00046150011248248627,
      "loss": 2.1915,
      "step": 73511
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.768313467502594,
      "learning_rate": 0.00046149666522081695,
      "loss": 2.0756,
      "step": 73512
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.9049090147018433,
      "learning_rate": 0.00046149321792912217,
      "loss": 2.1195,
      "step": 73513
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.8301687836647034,
      "learning_rate": 0.0004614897706074029,
      "loss": 2.0838,
      "step": 73514
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7647901177406311,
      "learning_rate": 0.0004614863232556596,
      "loss": 1.9826,
      "step": 73515
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.9154335260391235,
      "learning_rate": 0.0004614828758738929,
      "loss": 2.3698,
      "step": 73516
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7644655108451843,
      "learning_rate": 0.0004614794284621034,
      "loss": 2.2486,
      "step": 73517
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7567885518074036,
      "learning_rate": 0.000461475981020292,
      "loss": 2.1447,
      "step": 73518
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.8084813952445984,
      "learning_rate": 0.00046147253354845904,
      "loss": 2.2635,
      "step": 73519
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.8138133883476257,
      "learning_rate": 0.00046146908604660533,
      "loss": 2.2403,
      "step": 73520
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.9271551966667175,
      "learning_rate": 0.0004614656385147315,
      "loss": 2.2547,
      "step": 73521
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.8390458226203918,
      "learning_rate": 0.0004614621909528381,
      "loss": 2.1885,
      "step": 73522
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.8420043587684631,
      "learning_rate": 0.0004614587433609259,
      "loss": 2.2646,
      "step": 73523
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7647895216941833,
      "learning_rate": 0.0004614552957389954,
      "loss": 1.896,
      "step": 73524
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7865559458732605,
      "learning_rate": 0.00046145184808704735,
      "loss": 2.2642,
      "step": 73525
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.8310389518737793,
      "learning_rate": 0.0004614484004050824,
      "loss": 2.2822,
      "step": 73526
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7358577251434326,
      "learning_rate": 0.00046144495269310115,
      "loss": 2.0774,
      "step": 73527
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7757369875907898,
      "learning_rate": 0.0004614415049511041,
      "loss": 2.234,
      "step": 73528
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7524585127830505,
      "learning_rate": 0.00046143805717909213,
      "loss": 2.1977,
      "step": 73529
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.8386107683181763,
      "learning_rate": 0.0004614346093770657,
      "loss": 2.2833,
      "step": 73530
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7992590665817261,
      "learning_rate": 0.0004614311615450256,
      "loss": 2.3903,
      "step": 73531
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.708733856678009,
      "learning_rate": 0.0004614277136829724,
      "loss": 2.3289,
      "step": 73532
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.8442087769508362,
      "learning_rate": 0.0004614242657909068,
      "loss": 2.2328,
      "step": 73533
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7822837233543396,
      "learning_rate": 0.00046142081786882923,
      "loss": 2.2435,
      "step": 73534
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7884345650672913,
      "learning_rate": 0.00046141736991674057,
      "loss": 2.1709,
      "step": 73535
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7914822697639465,
      "learning_rate": 0.0004614139219346414,
      "loss": 2.3612,
      "step": 73536
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.8835031390190125,
      "learning_rate": 0.00046141047392253223,
      "loss": 2.3913,
      "step": 73537
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7639527916908264,
      "learning_rate": 0.0004614070258804139,
      "loss": 2.2164,
      "step": 73538
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7035788297653198,
      "learning_rate": 0.000461403577808287,
      "loss": 2.3096,
      "step": 73539
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.8296096324920654,
      "learning_rate": 0.000461400129706152,
      "loss": 2.0949,
      "step": 73540
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7892787456512451,
      "learning_rate": 0.0004613966815740098,
      "loss": 2.3701,
      "step": 73541
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.890638530254364,
      "learning_rate": 0.00046139323341186085,
      "loss": 2.4263,
      "step": 73542
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.8612558245658875,
      "learning_rate": 0.00046138978521970575,
      "loss": 1.9805,
      "step": 73543
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.8458174467086792,
      "learning_rate": 0.0004613863369975454,
      "loss": 2.2913,
      "step": 73544
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.9062672853469849,
      "learning_rate": 0.0004613828887453802,
      "loss": 2.2352,
      "step": 73545
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.8444557785987854,
      "learning_rate": 0.00046137944046321087,
      "loss": 2.2029,
      "step": 73546
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.8023192882537842,
      "learning_rate": 0.0004613759921510381,
      "loss": 2.2515,
      "step": 73547
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7491990923881531,
      "learning_rate": 0.00046137254380886246,
      "loss": 2.2998,
      "step": 73548
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.8119027614593506,
      "learning_rate": 0.00046136909543668464,
      "loss": 2.1147,
      "step": 73549
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7435985803604126,
      "learning_rate": 0.0004613656470345052,
      "loss": 2.1941,
      "step": 73550
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7760869860649109,
      "learning_rate": 0.0004613621986023249,
      "loss": 2.1461,
      "step": 73551
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7990962266921997,
      "learning_rate": 0.0004613587501401443,
      "loss": 2.3146,
      "step": 73552
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7728710770606995,
      "learning_rate": 0.0004613553016479641,
      "loss": 2.3568,
      "step": 73553
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7867435812950134,
      "learning_rate": 0.00046135185312578495,
      "loss": 2.0443,
      "step": 73554
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7693442702293396,
      "learning_rate": 0.00046134840457360724,
      "loss": 1.9155,
      "step": 73555
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.8366338014602661,
      "learning_rate": 0.00046134495599143194,
      "loss": 2.152,
      "step": 73556
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.8187406063079834,
      "learning_rate": 0.0004613415073792596,
      "loss": 2.0382,
      "step": 73557
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7551908493041992,
      "learning_rate": 0.0004613380587370908,
      "loss": 2.1777,
      "step": 73558
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7318750619888306,
      "learning_rate": 0.0004613346100649262,
      "loss": 2.0397,
      "step": 73559
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.8560308218002319,
      "learning_rate": 0.00046133116136276646,
      "loss": 2.099,
      "step": 73560
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.8047338724136353,
      "learning_rate": 0.0004613277126306122,
      "loss": 1.9639,
      "step": 73561
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7347987294197083,
      "learning_rate": 0.00046132426386846407,
      "loss": 2.5235,
      "step": 73562
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.8766636848449707,
      "learning_rate": 0.00046132081507632277,
      "loss": 2.0172,
      "step": 73563
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7744478583335876,
      "learning_rate": 0.00046131736625418886,
      "loss": 2.3084,
      "step": 73564
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.877452552318573,
      "learning_rate": 0.000461313917402063,
      "loss": 2.3185,
      "step": 73565
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7758459448814392,
      "learning_rate": 0.00046131046851994575,
      "loss": 2.2487,
      "step": 73566
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7454162240028381,
      "learning_rate": 0.0004613070196078379,
      "loss": 2.1404,
      "step": 73567
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7384437918663025,
      "learning_rate": 0.00046130357066574007,
      "loss": 1.9716,
      "step": 73568
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.761343777179718,
      "learning_rate": 0.0004613001216936528,
      "loss": 2.3865,
      "step": 73569
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.8521150350570679,
      "learning_rate": 0.00046129667269157685,
      "loss": 2.251,
      "step": 73570
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7855225205421448,
      "learning_rate": 0.0004612932236595128,
      "loss": 2.1346,
      "step": 73571
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.74165940284729,
      "learning_rate": 0.00046128977459746124,
      "loss": 1.9936,
      "step": 73572
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7640168070793152,
      "learning_rate": 0.0004612863255054229,
      "loss": 2.332,
      "step": 73573
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.043701171875,
      "learning_rate": 0.0004612828763833984,
      "loss": 2.1605,
      "step": 73574
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.8313788771629333,
      "learning_rate": 0.0004612794272313884,
      "loss": 2.413,
      "step": 73575
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7897188663482666,
      "learning_rate": 0.0004612759780493934,
      "loss": 2.2798,
      "step": 73576
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7750877141952515,
      "learning_rate": 0.0004612725288374143,
      "loss": 2.2457,
      "step": 73577
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.8301630616188049,
      "learning_rate": 0.0004612690795954514,
      "loss": 2.0949,
      "step": 73578
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.8380107879638672,
      "learning_rate": 0.0004612656303235057,
      "loss": 2.2583,
      "step": 73579
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7498285174369812,
      "learning_rate": 0.00046126218102157765,
      "loss": 2.1706,
      "step": 73580
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.8676257133483887,
      "learning_rate": 0.0004612587316896678,
      "loss": 2.1366,
      "step": 73581
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.8045236468315125,
      "learning_rate": 0.00046125528232777703,
      "loss": 2.3092,
      "step": 73582
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.723719596862793,
      "learning_rate": 0.00046125183293590583,
      "loss": 2.0535,
      "step": 73583
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.8216602206230164,
      "learning_rate": 0.00046124838351405486,
      "loss": 2.1367,
      "step": 73584
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.9092353582382202,
      "learning_rate": 0.00046124493406222484,
      "loss": 2.1866,
      "step": 73585
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7771474123001099,
      "learning_rate": 0.00046124148458041625,
      "loss": 1.9873,
      "step": 73586
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.8926712870597839,
      "learning_rate": 0.00046123803506862985,
      "loss": 2.3181,
      "step": 73587
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7269355654716492,
      "learning_rate": 0.0004612345855268663,
      "loss": 2.2637,
      "step": 73588
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.9069117903709412,
      "learning_rate": 0.0004612311359551261,
      "loss": 2.4875,
      "step": 73589
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.8886127471923828,
      "learning_rate": 0.00046122768635341007,
      "loss": 2.0979,
      "step": 73590
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.773745596408844,
      "learning_rate": 0.00046122423672171877,
      "loss": 2.0046,
      "step": 73591
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.793437659740448,
      "learning_rate": 0.0004612207870600528,
      "loss": 1.994,
      "step": 73592
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.9476311802864075,
      "learning_rate": 0.00046121733736841287,
      "loss": 2.0733,
      "step": 73593
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.9296644926071167,
      "learning_rate": 0.00046121388764679963,
      "loss": 2.1671,
      "step": 73594
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.8756987452507019,
      "learning_rate": 0.00046121043789521356,
      "loss": 2.2218,
      "step": 73595
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7944265604019165,
      "learning_rate": 0.00046120698811365554,
      "loss": 2.053,
      "step": 73596
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7898776531219482,
      "learning_rate": 0.0004612035383021261,
      "loss": 2.1991,
      "step": 73597
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7942003011703491,
      "learning_rate": 0.0004612000884606258,
      "loss": 2.1779,
      "step": 73598
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7680780291557312,
      "learning_rate": 0.0004611966385891554,
      "loss": 2.417,
      "step": 73599
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7653918266296387,
      "learning_rate": 0.0004611931886877155,
      "loss": 2.1691,
      "step": 73600
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.8255361914634705,
      "learning_rate": 0.00046118973875630676,
      "loss": 2.0833,
      "step": 73601
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.758270263671875,
      "learning_rate": 0.0004611862887949298,
      "loss": 2.0645,
      "step": 73602
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7882360816001892,
      "learning_rate": 0.00046118283880358527,
      "loss": 2.2973,
      "step": 73603
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7574360370635986,
      "learning_rate": 0.0004611793887822738,
      "loss": 2.2173,
      "step": 73604
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.8241010308265686,
      "learning_rate": 0.000461175938730996,
      "loss": 2.2884,
      "step": 73605
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.8474888801574707,
      "learning_rate": 0.0004611724886497526,
      "loss": 2.1684,
      "step": 73606
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.6963117718696594,
      "learning_rate": 0.0004611690385385442,
      "loss": 2.3133,
      "step": 73607
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.8612751960754395,
      "learning_rate": 0.0004611655883973714,
      "loss": 2.1195,
      "step": 73608
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.8624686598777771,
      "learning_rate": 0.0004611621382262349,
      "loss": 1.9782,
      "step": 73609
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.802720844745636,
      "learning_rate": 0.0004611586880251353,
      "loss": 2.2009,
      "step": 73610
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7973902225494385,
      "learning_rate": 0.00046115523779407326,
      "loss": 2.2154,
      "step": 73611
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.8455541729927063,
      "learning_rate": 0.00046115178753304944,
      "loss": 2.3689,
      "step": 73612
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7347835898399353,
      "learning_rate": 0.0004611483372420644,
      "loss": 2.1053,
      "step": 73613
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.8701362609863281,
      "learning_rate": 0.0004611448869211189,
      "loss": 2.2545,
      "step": 73614
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.8512774109840393,
      "learning_rate": 0.0004611414365702135,
      "loss": 2.3994,
      "step": 73615
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7668747305870056,
      "learning_rate": 0.0004611379861893489,
      "loss": 2.1338,
      "step": 73616
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7650607228279114,
      "learning_rate": 0.00046113453577852564,
      "loss": 2.1715,
      "step": 73617
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.8110939264297485,
      "learning_rate": 0.00046113108533774445,
      "loss": 2.1625,
      "step": 73618
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.765034556388855,
      "learning_rate": 0.000461127634867006,
      "loss": 2.2271,
      "step": 73619
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7452099323272705,
      "learning_rate": 0.00046112418436631083,
      "loss": 2.356,
      "step": 73620
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7905543446540833,
      "learning_rate": 0.0004611207338356597,
      "loss": 2.2042,
      "step": 73621
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7382633090019226,
      "learning_rate": 0.0004611172832750531,
      "loss": 2.2365,
      "step": 73622
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.8470227122306824,
      "learning_rate": 0.00046111383268449175,
      "loss": 2.2433,
      "step": 73623
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7777403593063354,
      "learning_rate": 0.0004611103820639764,
      "loss": 2.3142,
      "step": 73624
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.6992872953414917,
      "learning_rate": 0.0004611069314135075,
      "loss": 2.2357,
      "step": 73625
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7801952958106995,
      "learning_rate": 0.0004611034807330858,
      "loss": 2.027,
      "step": 73626
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.8828296065330505,
      "learning_rate": 0.000461100030022712,
      "loss": 2.3809,
      "step": 73627
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.9242720603942871,
      "learning_rate": 0.0004610965792823865,
      "loss": 2.2535,
      "step": 73628
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.8657408356666565,
      "learning_rate": 0.00046109312851211017,
      "loss": 2.3193,
      "step": 73629
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.1183505058288574,
      "learning_rate": 0.00046108967771188366,
      "loss": 2.5172,
      "step": 73630
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.8237625360488892,
      "learning_rate": 0.0004610862268817075,
      "loss": 2.1962,
      "step": 73631
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7318029403686523,
      "learning_rate": 0.0004610827760215824,
      "loss": 2.2274,
      "step": 73632
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.9263594150543213,
      "learning_rate": 0.00046107932513150883,
      "loss": 2.4367,
      "step": 73633
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7810475826263428,
      "learning_rate": 0.0004610758742114878,
      "loss": 2.3171,
      "step": 73634
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7943081259727478,
      "learning_rate": 0.00046107242326151956,
      "loss": 2.0702,
      "step": 73635
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7699013352394104,
      "learning_rate": 0.00046106897228160497,
      "loss": 2.189,
      "step": 73636
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7722260355949402,
      "learning_rate": 0.0004610655212717446,
      "loss": 2.272,
      "step": 73637
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.8898926377296448,
      "learning_rate": 0.00046106207023193906,
      "loss": 2.2418,
      "step": 73638
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7957570552825928,
      "learning_rate": 0.0004610586191621891,
      "loss": 2.3396,
      "step": 73639
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.8597621321678162,
      "learning_rate": 0.00046105516806249527,
      "loss": 2.2718,
      "step": 73640
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.8763969540596008,
      "learning_rate": 0.00046105171693285827,
      "loss": 2.2895,
      "step": 73641
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7998927235603333,
      "learning_rate": 0.0004610482657732787,
      "loss": 2.0366,
      "step": 73642
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.9217004179954529,
      "learning_rate": 0.00046104481458375735,
      "loss": 2.1533,
      "step": 73643
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7873169183731079,
      "learning_rate": 0.0004610413633642946,
      "loss": 2.3455,
      "step": 73644
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7667584419250488,
      "learning_rate": 0.00046103791211489113,
      "loss": 2.2403,
      "step": 73645
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7829684019088745,
      "learning_rate": 0.00046103446083554787,
      "loss": 2.2692,
      "step": 73646
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7966114282608032,
      "learning_rate": 0.0004610310095262651,
      "loss": 2.1973,
      "step": 73647
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7536559700965881,
      "learning_rate": 0.0004610275581870437,
      "loss": 2.0543,
      "step": 73648
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.736465334892273,
      "learning_rate": 0.0004610241068178843,
      "loss": 2.1615,
      "step": 73649
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7825344204902649,
      "learning_rate": 0.0004610206554187874,
      "loss": 2.1618,
      "step": 73650
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.8972117304801941,
      "learning_rate": 0.00046101720398975367,
      "loss": 1.954,
      "step": 73651
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7864046096801758,
      "learning_rate": 0.00046101375253078386,
      "loss": 2.0785,
      "step": 73652
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7702436447143555,
      "learning_rate": 0.00046101030104187863,
      "loss": 2.065,
      "step": 73653
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7700318098068237,
      "learning_rate": 0.0004610068495230384,
      "loss": 2.0879,
      "step": 73654
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.8744717836380005,
      "learning_rate": 0.00046100339797426403,
      "loss": 2.3962,
      "step": 73655
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.9522814750671387,
      "learning_rate": 0.00046099994639555613,
      "loss": 2.4409,
      "step": 73656
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7894487380981445,
      "learning_rate": 0.00046099649478691524,
      "loss": 1.9657,
      "step": 73657
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.8957837820053101,
      "learning_rate": 0.00046099304314834214,
      "loss": 2.4032,
      "step": 73658
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7431167364120483,
      "learning_rate": 0.00046098959147983725,
      "loss": 1.9145,
      "step": 73659
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7818411588668823,
      "learning_rate": 0.0004609861397814015,
      "loss": 2.0916,
      "step": 73660
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.9028448462486267,
      "learning_rate": 0.0004609826880530353,
      "loss": 2.5065,
      "step": 73661
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.799808919429779,
      "learning_rate": 0.00046097923629473945,
      "loss": 2.3473,
      "step": 73662
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.8022598624229431,
      "learning_rate": 0.00046097578450651446,
      "loss": 2.4351,
      "step": 73663
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7773522138595581,
      "learning_rate": 0.00046097233268836105,
      "loss": 2.2665,
      "step": 73664
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.8550673127174377,
      "learning_rate": 0.00046096888084027986,
      "loss": 2.1266,
      "step": 73665
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.8021944761276245,
      "learning_rate": 0.0004609654289622715,
      "loss": 2.2522,
      "step": 73666
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7824355959892273,
      "learning_rate": 0.00046096197705433665,
      "loss": 2.318,
      "step": 73667
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.796496570110321,
      "learning_rate": 0.0004609585251164759,
      "loss": 2.2533,
      "step": 73668
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.8098191618919373,
      "learning_rate": 0.00046095507314868994,
      "loss": 2.2119,
      "step": 73669
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.8287330269813538,
      "learning_rate": 0.0004609516211509794,
      "loss": 2.1505,
      "step": 73670
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7647674679756165,
      "learning_rate": 0.0004609481691233449,
      "loss": 2.0625,
      "step": 73671
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.784548282623291,
      "learning_rate": 0.00046094471706578714,
      "loss": 2.2677,
      "step": 73672
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7384454607963562,
      "learning_rate": 0.00046094126497830663,
      "loss": 2.3567,
      "step": 73673
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.8626431226730347,
      "learning_rate": 0.0004609378128609041,
      "loss": 2.217,
      "step": 73674
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7965729236602783,
      "learning_rate": 0.00046093436071358024,
      "loss": 2.1997,
      "step": 73675
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.8078790903091431,
      "learning_rate": 0.0004609309085363358,
      "loss": 2.3029,
      "step": 73676
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.8327712416648865,
      "learning_rate": 0.00046092745632917104,
      "loss": 2.5091,
      "step": 73677
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7699401378631592,
      "learning_rate": 0.0004609240040920869,
      "loss": 2.0867,
      "step": 73678
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.8335989117622375,
      "learning_rate": 0.00046092055182508397,
      "loss": 2.2627,
      "step": 73679
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.718939483165741,
      "learning_rate": 0.00046091709952816287,
      "loss": 2.1268,
      "step": 73680
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.81132972240448,
      "learning_rate": 0.0004609136472013242,
      "loss": 2.2162,
      "step": 73681
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7591356635093689,
      "learning_rate": 0.0004609101948445687,
      "loss": 2.1222,
      "step": 73682
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7511366009712219,
      "learning_rate": 0.0004609067424578969,
      "loss": 2.4173,
      "step": 73683
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.8081150054931641,
      "learning_rate": 0.0004609032900413095,
      "loss": 2.0984,
      "step": 73684
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7669430375099182,
      "learning_rate": 0.0004608998375948073,
      "loss": 2.2235,
      "step": 73685
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.6940465569496155,
      "learning_rate": 0.0004608963851183906,
      "loss": 2.0973,
      "step": 73686
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7545639276504517,
      "learning_rate": 0.0004608929326120603,
      "loss": 2.121,
      "step": 73687
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7359764575958252,
      "learning_rate": 0.00046088948007581696,
      "loss": 2.1405,
      "step": 73688
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.8478037118911743,
      "learning_rate": 0.0004608860275096612,
      "loss": 2.0775,
      "step": 73689
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.8455211520195007,
      "learning_rate": 0.00046088257491359364,
      "loss": 2.3762,
      "step": 73690
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7724042534828186,
      "learning_rate": 0.0004608791222876151,
      "loss": 2.4713,
      "step": 73691
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7304609417915344,
      "learning_rate": 0.000460875669631726,
      "loss": 2.2744,
      "step": 73692
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7806776762008667,
      "learning_rate": 0.0004608722169459271,
      "loss": 2.2426,
      "step": 73693
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7417515516281128,
      "learning_rate": 0.0004608687642302191,
      "loss": 2.2554,
      "step": 73694
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.8207535147666931,
      "learning_rate": 0.0004608653114846025,
      "loss": 2.2368,
      "step": 73695
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.883474588394165,
      "learning_rate": 0.00046086185870907793,
      "loss": 2.1893,
      "step": 73696
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.8446896076202393,
      "learning_rate": 0.0004608584059036462,
      "loss": 2.258,
      "step": 73697
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7682278752326965,
      "learning_rate": 0.0004608549530683078,
      "loss": 2.1997,
      "step": 73698
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7754082083702087,
      "learning_rate": 0.0004608515002030634,
      "loss": 2.3542,
      "step": 73699
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7955491542816162,
      "learning_rate": 0.00046084804730791375,
      "loss": 2.2051,
      "step": 73700
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7333859205245972,
      "learning_rate": 0.0004608445943828594,
      "loss": 2.1708,
      "step": 73701
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.750683605670929,
      "learning_rate": 0.00046084114142790094,
      "loss": 2.3238,
      "step": 73702
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7437801361083984,
      "learning_rate": 0.00046083768844303914,
      "loss": 2.3768,
      "step": 73703
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.8011853098869324,
      "learning_rate": 0.0004608342354282746,
      "loss": 2.0515,
      "step": 73704
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.8015904426574707,
      "learning_rate": 0.00046083078238360784,
      "loss": 2.2561,
      "step": 73705
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.8410851359367371,
      "learning_rate": 0.00046082732930903964,
      "loss": 2.3255,
      "step": 73706
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7395765781402588,
      "learning_rate": 0.00046082387620457067,
      "loss": 2.2766,
      "step": 73707
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.752019464969635,
      "learning_rate": 0.0004608204230702015,
      "loss": 2.3531,
      "step": 73708
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.8921460509300232,
      "learning_rate": 0.0004608169699059327,
      "loss": 2.0354,
      "step": 73709
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.8106677532196045,
      "learning_rate": 0.00046081351671176503,
      "loss": 2.1391,
      "step": 73710
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.896534264087677,
      "learning_rate": 0.00046081006348769907,
      "loss": 2.1755,
      "step": 73711
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7960725426673889,
      "learning_rate": 0.00046080661023373547,
      "loss": 2.2121,
      "step": 73712
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7933301329612732,
      "learning_rate": 0.00046080315694987494,
      "loss": 2.1057,
      "step": 73713
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.8922322988510132,
      "learning_rate": 0.0004607997036361181,
      "loss": 2.3025,
      "step": 73714
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.8748693466186523,
      "learning_rate": 0.0004607962502924655,
      "loss": 2.2246,
      "step": 73715
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.8704321980476379,
      "learning_rate": 0.00046079279691891786,
      "loss": 2.4116,
      "step": 73716
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.8094360828399658,
      "learning_rate": 0.0004607893435154757,
      "loss": 2.3652,
      "step": 73717
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.8325755000114441,
      "learning_rate": 0.00046078589008213984,
      "loss": 2.513,
      "step": 73718
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.8372556567192078,
      "learning_rate": 0.00046078243661891097,
      "loss": 2.044,
      "step": 73719
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.8433130383491516,
      "learning_rate": 0.00046077898312578945,
      "loss": 2.1087,
      "step": 73720
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7981235384941101,
      "learning_rate": 0.0004607755296027761,
      "loss": 2.0814,
      "step": 73721
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.9053160548210144,
      "learning_rate": 0.00046077207604987167,
      "loss": 2.144,
      "step": 73722
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.8405110239982605,
      "learning_rate": 0.0004607686224670766,
      "loss": 2.1169,
      "step": 73723
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7944075465202332,
      "learning_rate": 0.00046076516885439154,
      "loss": 2.23,
      "step": 73724
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7830591797828674,
      "learning_rate": 0.0004607617152118173,
      "loss": 1.9935,
      "step": 73725
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7547187209129333,
      "learning_rate": 0.0004607582615393543,
      "loss": 2.0359,
      "step": 73726
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7564954161643982,
      "learning_rate": 0.00046075480783700336,
      "loss": 2.1874,
      "step": 73727
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7257443070411682,
      "learning_rate": 0.00046075135410476517,
      "loss": 2.2189,
      "step": 73728
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.822725772857666,
      "learning_rate": 0.0004607479003426402,
      "loss": 2.1317,
      "step": 73729
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7933764457702637,
      "learning_rate": 0.00046074444655062907,
      "loss": 2.1884,
      "step": 73730
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.838555634021759,
      "learning_rate": 0.00046074099272873267,
      "loss": 2.1949,
      "step": 73731
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7596396803855896,
      "learning_rate": 0.0004607375388769513,
      "loss": 2.0681,
      "step": 73732
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7929973006248474,
      "learning_rate": 0.00046073408499528595,
      "loss": 2.2668,
      "step": 73733
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.8502457141876221,
      "learning_rate": 0.00046073063108373704,
      "loss": 2.2824,
      "step": 73734
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.8553695678710938,
      "learning_rate": 0.00046072717714230524,
      "loss": 2.3952,
      "step": 73735
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.6967253088951111,
      "learning_rate": 0.0004607237231709913,
      "loss": 1.9829,
      "step": 73736
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.931787371635437,
      "learning_rate": 0.00046072026916979574,
      "loss": 2.2916,
      "step": 73737
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.8582549691200256,
      "learning_rate": 0.0004607168151387193,
      "loss": 2.1254,
      "step": 73738
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.729168176651001,
      "learning_rate": 0.0004607133610777625,
      "loss": 2.013,
      "step": 73739
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7561053037643433,
      "learning_rate": 0.000460709906986926,
      "loss": 2.2635,
      "step": 73740
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.8879667520523071,
      "learning_rate": 0.0004607064528662107,
      "loss": 2.263,
      "step": 73741
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7379958629608154,
      "learning_rate": 0.0004607029987156169,
      "loss": 2.1915,
      "step": 73742
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.8135001063346863,
      "learning_rate": 0.0004606995445351453,
      "loss": 2.322,
      "step": 73743
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.814765989780426,
      "learning_rate": 0.00046069609032479666,
      "loss": 2.2459,
      "step": 73744
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7904233336448669,
      "learning_rate": 0.00046069263608457175,
      "loss": 2.2008,
      "step": 73745
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.8859415054321289,
      "learning_rate": 0.00046068918181447083,
      "loss": 2.3579,
      "step": 73746
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7880557775497437,
      "learning_rate": 0.00046068572751449484,
      "loss": 2.2647,
      "step": 73747
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7885631918907166,
      "learning_rate": 0.00046068227318464436,
      "loss": 2.1161,
      "step": 73748
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.8499410152435303,
      "learning_rate": 0.00046067881882492004,
      "loss": 2.0952,
      "step": 73749
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7908448576927185,
      "learning_rate": 0.0004606753644353224,
      "loss": 2.3173,
      "step": 73750
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.79867023229599,
      "learning_rate": 0.0004606719100158523,
      "loss": 2.1012,
      "step": 73751
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.8476862907409668,
      "learning_rate": 0.0004606684555665102,
      "loss": 2.1329,
      "step": 73752
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.8494610786437988,
      "learning_rate": 0.0004606650010872968,
      "loss": 2.3225,
      "step": 73753
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.6849764585494995,
      "learning_rate": 0.00046066154657821266,
      "loss": 2.1247,
      "step": 73754
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.819459080696106,
      "learning_rate": 0.0004606580920392587,
      "loss": 1.8761,
      "step": 73755
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7891961932182312,
      "learning_rate": 0.0004606546374704352,
      "loss": 2.1361,
      "step": 73756
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7151030898094177,
      "learning_rate": 0.000460651182871743,
      "loss": 2.2008,
      "step": 73757
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.8335791230201721,
      "learning_rate": 0.0004606477282431828,
      "loss": 2.192,
      "step": 73758
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.8383044600486755,
      "learning_rate": 0.0004606442735847551,
      "loss": 2.2898,
      "step": 73759
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.8454224467277527,
      "learning_rate": 0.00046064081889646047,
      "loss": 2.1248,
      "step": 73760
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.856589674949646,
      "learning_rate": 0.0004606373641782999,
      "loss": 2.2875,
      "step": 73761
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.8994846940040588,
      "learning_rate": 0.0004606339094302737,
      "loss": 2.1644,
      "step": 73762
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.8839719891548157,
      "learning_rate": 0.0004606304546523826,
      "loss": 2.1625,
      "step": 73763
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.8213614225387573,
      "learning_rate": 0.0004606269998446274,
      "loss": 2.3815,
      "step": 73764
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.8443822860717773,
      "learning_rate": 0.00046062354500700845,
      "loss": 2.1733,
      "step": 73765
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7193478345870972,
      "learning_rate": 0.0004606200901395266,
      "loss": 2.0767,
      "step": 73766
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7396339774131775,
      "learning_rate": 0.00046061663524218243,
      "loss": 2.2136,
      "step": 73767
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.8906859755516052,
      "learning_rate": 0.0004606131803149766,
      "loss": 2.3285,
      "step": 73768
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7574141025543213,
      "learning_rate": 0.0004606097253579098,
      "loss": 2.171,
      "step": 73769
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.0098671913146973,
      "learning_rate": 0.00046060627037098256,
      "loss": 2.3395,
      "step": 73770
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7134730815887451,
      "learning_rate": 0.0004606028153541956,
      "loss": 2.3026,
      "step": 73771
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.8637965321540833,
      "learning_rate": 0.00046059936030754965,
      "loss": 2.3577,
      "step": 73772
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.8327733874320984,
      "learning_rate": 0.00046059590523104513,
      "loss": 2.3198,
      "step": 73773
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.9086928963661194,
      "learning_rate": 0.0004605924501246828,
      "loss": 2.3503,
      "step": 73774
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7126340866088867,
      "learning_rate": 0.0004605889949884633,
      "loss": 2.2272,
      "step": 73775
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7869834899902344,
      "learning_rate": 0.0004605855398223874,
      "loss": 2.1967,
      "step": 73776
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.8327314257621765,
      "learning_rate": 0.00046058208462645544,
      "loss": 2.1898,
      "step": 73777
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.794870913028717,
      "learning_rate": 0.00046057862940066836,
      "loss": 2.3028,
      "step": 73778
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.8075698018074036,
      "learning_rate": 0.00046057517414502665,
      "loss": 2.2351,
      "step": 73779
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.8227759599685669,
      "learning_rate": 0.000460571718859531,
      "loss": 2.1743,
      "step": 73780
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7402492165565491,
      "learning_rate": 0.000460568263544182,
      "loss": 2.1507,
      "step": 73781
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.808444619178772,
      "learning_rate": 0.0004605648081989803,
      "loss": 2.2152,
      "step": 73782
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7980706691741943,
      "learning_rate": 0.0004605613528239266,
      "loss": 2.1739,
      "step": 73783
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.8548056483268738,
      "learning_rate": 0.00046055789741902153,
      "loss": 2.1896,
      "step": 73784
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7927192449569702,
      "learning_rate": 0.00046055444198426566,
      "loss": 2.2958,
      "step": 73785
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7693036794662476,
      "learning_rate": 0.00046055098651965975,
      "loss": 2.0882,
      "step": 73786
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.8187465667724609,
      "learning_rate": 0.0004605475310252044,
      "loss": 2.0818,
      "step": 73787
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7981200218200684,
      "learning_rate": 0.0004605440755009002,
      "loss": 2.1717,
      "step": 73788
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.8243316411972046,
      "learning_rate": 0.0004605406199467478,
      "loss": 2.1927,
      "step": 73789
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7720295190811157,
      "learning_rate": 0.0004605371643627479,
      "loss": 2.3594,
      "step": 73790
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.8857343196868896,
      "learning_rate": 0.0004605337087489011,
      "loss": 2.2305,
      "step": 73791
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.8412748575210571,
      "learning_rate": 0.00046053025310520814,
      "loss": 2.248,
      "step": 73792
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7638545632362366,
      "learning_rate": 0.0004605267974316694,
      "loss": 2.0479,
      "step": 73793
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7115042209625244,
      "learning_rate": 0.0004605233417282858,
      "loss": 2.1603,
      "step": 73794
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7514244914054871,
      "learning_rate": 0.000460519885995058,
      "loss": 2.1747,
      "step": 73795
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.864460825920105,
      "learning_rate": 0.0004605164302319863,
      "loss": 2.2875,
      "step": 73796
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7702088356018066,
      "learning_rate": 0.00046051297443907164,
      "loss": 1.9736,
      "step": 73797
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.776325523853302,
      "learning_rate": 0.00046050951861631467,
      "loss": 1.9253,
      "step": 73798
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.8222454190254211,
      "learning_rate": 0.00046050606276371584,
      "loss": 2.439,
      "step": 73799
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.8491889834403992,
      "learning_rate": 0.000460502606881276,
      "loss": 2.198,
      "step": 73800
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.923198401927948,
      "learning_rate": 0.0004604991509689957,
      "loss": 2.0758,
      "step": 73801
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7571311593055725,
      "learning_rate": 0.00046049569502687546,
      "loss": 2.2918,
      "step": 73802
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.8058531880378723,
      "learning_rate": 0.00046049223905491606,
      "loss": 2.212,
      "step": 73803
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.82477205991745,
      "learning_rate": 0.0004604887830531183,
      "loss": 2.2744,
      "step": 73804
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7794298529624939,
      "learning_rate": 0.00046048532702148244,
      "loss": 2.2063,
      "step": 73805
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.8011462688446045,
      "learning_rate": 0.0004604818709600094,
      "loss": 2.2757,
      "step": 73806
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.9045658707618713,
      "learning_rate": 0.0004604784148686998,
      "loss": 2.0968,
      "step": 73807
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.8641617894172668,
      "learning_rate": 0.00046047495874755416,
      "loss": 2.3363,
      "step": 73808
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7847457528114319,
      "learning_rate": 0.00046047150259657325,
      "loss": 2.1663,
      "step": 73809
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.892085075378418,
      "learning_rate": 0.00046046804641575764,
      "loss": 2.1104,
      "step": 73810
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7870705723762512,
      "learning_rate": 0.00046046459020510804,
      "loss": 2.2048,
      "step": 73811
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.8799281120300293,
      "learning_rate": 0.0004604611339646249,
      "loss": 2.17,
      "step": 73812
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.9800320863723755,
      "learning_rate": 0.0004604576776943092,
      "loss": 2.3453,
      "step": 73813
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.8184040188789368,
      "learning_rate": 0.00046045422139416126,
      "loss": 2.2569,
      "step": 73814
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7621217370033264,
      "learning_rate": 0.0004604507650641818,
      "loss": 2.1432,
      "step": 73815
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.8205146789550781,
      "learning_rate": 0.0004604473087043716,
      "loss": 2.1731,
      "step": 73816
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.9049017429351807,
      "learning_rate": 0.0004604438523147312,
      "loss": 2.302,
      "step": 73817
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.8329300284385681,
      "learning_rate": 0.00046044039589526124,
      "loss": 2.246,
      "step": 73818
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.765275239944458,
      "learning_rate": 0.0004604369394459625,
      "loss": 2.11,
      "step": 73819
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.9373565316200256,
      "learning_rate": 0.0004604334829668354,
      "loss": 2.2323,
      "step": 73820
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7462530136108398,
      "learning_rate": 0.00046043002645788065,
      "loss": 2.2133,
      "step": 73821
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.8171672821044922,
      "learning_rate": 0.00046042656991909906,
      "loss": 2.1203,
      "step": 73822
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.9089643955230713,
      "learning_rate": 0.000460423113350491,
      "loss": 2.0564,
      "step": 73823
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.9713876843452454,
      "learning_rate": 0.00046041965675205725,
      "loss": 2.1845,
      "step": 73824
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.8013543486595154,
      "learning_rate": 0.00046041620012379856,
      "loss": 2.2166,
      "step": 73825
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.919514536857605,
      "learning_rate": 0.0004604127434657155,
      "loss": 2.3323,
      "step": 73826
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.1351338624954224,
      "learning_rate": 0.0004604092867778086,
      "loss": 2.0801,
      "step": 73827
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7210693955421448,
      "learning_rate": 0.0004604058300600786,
      "loss": 1.9991,
      "step": 73828
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.8855517506599426,
      "learning_rate": 0.00046040237331252613,
      "loss": 2.106,
      "step": 73829
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.9804957509040833,
      "learning_rate": 0.00046039891653515175,
      "loss": 2.2148,
      "step": 73830
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7457194328308105,
      "learning_rate": 0.00046039545972795633,
      "loss": 2.0062,
      "step": 73831
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.8519515991210938,
      "learning_rate": 0.00046039200289094026,
      "loss": 2.0853,
      "step": 73832
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.8386803865432739,
      "learning_rate": 0.0004603885460241044,
      "loss": 2.2662,
      "step": 73833
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7596379518508911,
      "learning_rate": 0.0004603850891274492,
      "loss": 2.2807,
      "step": 73834
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7541570663452148,
      "learning_rate": 0.0004603816322009753,
      "loss": 2.121,
      "step": 73835
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7023720741271973,
      "learning_rate": 0.0004603781752446836,
      "loss": 2.1912,
      "step": 73836
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.6969888806343079,
      "learning_rate": 0.00046037471825857446,
      "loss": 2.078,
      "step": 73837
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.8338922262191772,
      "learning_rate": 0.00046037126124264866,
      "loss": 2.2086,
      "step": 73838
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.734893262386322,
      "learning_rate": 0.00046036780419690687,
      "loss": 2.3311,
      "step": 73839
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7912707924842834,
      "learning_rate": 0.0004603643471213496,
      "loss": 2.3232,
      "step": 73840
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.8340392708778381,
      "learning_rate": 0.00046036089001597763,
      "loss": 2.0283,
      "step": 73841
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.779145359992981,
      "learning_rate": 0.0004603574328807915,
      "loss": 2.1441,
      "step": 73842
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.8148189783096313,
      "learning_rate": 0.00046035397571579184,
      "loss": 2.0278,
      "step": 73843
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7712956666946411,
      "learning_rate": 0.0004603505185209795,
      "loss": 2.1799,
      "step": 73844
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.813535749912262,
      "learning_rate": 0.0004603470612963549,
      "loss": 2.0737,
      "step": 73845
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7985110878944397,
      "learning_rate": 0.00046034360404191874,
      "loss": 2.2085,
      "step": 73846
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7493026852607727,
      "learning_rate": 0.0004603401467576717,
      "loss": 2.2007,
      "step": 73847
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.8770049214363098,
      "learning_rate": 0.00046033668944361433,
      "loss": 2.1362,
      "step": 73848
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.8211694955825806,
      "learning_rate": 0.0004603332320997474,
      "loss": 2.0377,
      "step": 73849
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.8597853183746338,
      "learning_rate": 0.00046032977472607153,
      "loss": 2.3335,
      "step": 73850
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.8636221885681152,
      "learning_rate": 0.00046032631732258726,
      "loss": 2.4345,
      "step": 73851
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.8217856884002686,
      "learning_rate": 0.00046032285988929534,
      "loss": 2.0695,
      "step": 73852
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7506243586540222,
      "learning_rate": 0.00046031940242619637,
      "loss": 2.2965,
      "step": 73853
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7905173301696777,
      "learning_rate": 0.000460315944933291,
      "loss": 2.2088,
      "step": 73854
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7908408045768738,
      "learning_rate": 0.0004603124874105799,
      "loss": 2.1735,
      "step": 73855
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7825758457183838,
      "learning_rate": 0.0004603090298580636,
      "loss": 2.2949,
      "step": 73856
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7914809584617615,
      "learning_rate": 0.00046030557227574284,
      "loss": 2.3217,
      "step": 73857
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7469648122787476,
      "learning_rate": 0.0004603021146636183,
      "loss": 2.1122,
      "step": 73858
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7148000001907349,
      "learning_rate": 0.0004602986570216906,
      "loss": 2.1344,
      "step": 73859
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7853285074234009,
      "learning_rate": 0.00046029519934996026,
      "loss": 2.2231,
      "step": 73860
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7109513878822327,
      "learning_rate": 0.0004602917416484281,
      "loss": 2.1857,
      "step": 73861
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7563799619674683,
      "learning_rate": 0.00046028828391709464,
      "loss": 2.0828,
      "step": 73862
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.8309172987937927,
      "learning_rate": 0.00046028482615596047,
      "loss": 2.061,
      "step": 73863
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7355760335922241,
      "learning_rate": 0.0004602813683650265,
      "loss": 2.2808,
      "step": 73864
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.8027989268302917,
      "learning_rate": 0.00046027791054429316,
      "loss": 1.9539,
      "step": 73865
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7738279700279236,
      "learning_rate": 0.000460274452693761,
      "loss": 2.0657,
      "step": 73866
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7655680775642395,
      "learning_rate": 0.00046027099481343094,
      "loss": 2.2429,
      "step": 73867
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7637947797775269,
      "learning_rate": 0.00046026753690330344,
      "loss": 2.409,
      "step": 73868
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.778883159160614,
      "learning_rate": 0.00046026407896337907,
      "loss": 2.3448,
      "step": 73869
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.8102290630340576,
      "learning_rate": 0.00046026062099365866,
      "loss": 2.3163,
      "step": 73870
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.761898398399353,
      "learning_rate": 0.00046025716299414286,
      "loss": 2.4269,
      "step": 73871
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.8614596724510193,
      "learning_rate": 0.0004602537049648321,
      "loss": 2.4635,
      "step": 73872
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.8950071930885315,
      "learning_rate": 0.00046025024690572725,
      "loss": 2.2949,
      "step": 73873
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7982754707336426,
      "learning_rate": 0.0004602467888168288,
      "loss": 2.3078,
      "step": 73874
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7709939479827881,
      "learning_rate": 0.0004602433306981375,
      "loss": 2.3598,
      "step": 73875
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7767987847328186,
      "learning_rate": 0.0004602398725496538,
      "loss": 2.1218,
      "step": 73876
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.8126572370529175,
      "learning_rate": 0.00046023641437137867,
      "loss": 2.2922,
      "step": 73877
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.8584775924682617,
      "learning_rate": 0.00046023295616331246,
      "loss": 2.0716,
      "step": 73878
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7673978805541992,
      "learning_rate": 0.0004602294979254559,
      "loss": 2.3701,
      "step": 73879
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.9637221693992615,
      "learning_rate": 0.0004602260396578097,
      "loss": 2.2313,
      "step": 73880
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7600436806678772,
      "learning_rate": 0.00046022258136037443,
      "loss": 2.2585,
      "step": 73881
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.9167807698249817,
      "learning_rate": 0.00046021912303315077,
      "loss": 2.0025,
      "step": 73882
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.8814312815666199,
      "learning_rate": 0.0004602156646761393,
      "loss": 2.1838,
      "step": 73883
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.8496922850608826,
      "learning_rate": 0.0004602122062893408,
      "loss": 2.1986,
      "step": 73884
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.8461676836013794,
      "learning_rate": 0.0004602087478727558,
      "loss": 2.2888,
      "step": 73885
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.9273483157157898,
      "learning_rate": 0.0004602052894263849,
      "loss": 2.103,
      "step": 73886
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7883856892585754,
      "learning_rate": 0.0004602018309502289,
      "loss": 2.3893,
      "step": 73887
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.8357557058334351,
      "learning_rate": 0.00046019837244428834,
      "loss": 2.1626,
      "step": 73888
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.9375985860824585,
      "learning_rate": 0.0004601949139085639,
      "loss": 2.4225,
      "step": 73889
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7753139734268188,
      "learning_rate": 0.00046019145534305616,
      "loss": 2.3022,
      "step": 73890
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7880743145942688,
      "learning_rate": 0.00046018799674776574,
      "loss": 2.306,
      "step": 73891
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.8161657452583313,
      "learning_rate": 0.00046018453812269347,
      "loss": 2.2451,
      "step": 73892
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7598562240600586,
      "learning_rate": 0.0004601810794678398,
      "loss": 2.1825,
      "step": 73893
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7406227588653564,
      "learning_rate": 0.0004601776207832055,
      "loss": 2.153,
      "step": 73894
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7394858598709106,
      "learning_rate": 0.0004601741620687911,
      "loss": 2.0592,
      "step": 73895
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.8296125531196594,
      "learning_rate": 0.0004601707033245973,
      "loss": 2.1202,
      "step": 73896
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7428585290908813,
      "learning_rate": 0.00046016724455062475,
      "loss": 1.9352,
      "step": 73897
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7978946566581726,
      "learning_rate": 0.0004601637857468742,
      "loss": 1.9198,
      "step": 73898
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.8476780652999878,
      "learning_rate": 0.00046016032691334604,
      "loss": 2.3792,
      "step": 73899
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.9607895016670227,
      "learning_rate": 0.00046015686805004106,
      "loss": 2.0856,
      "step": 73900
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7650190591812134,
      "learning_rate": 0.0004601534091569599,
      "loss": 2.1664,
      "step": 73901
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7471888661384583,
      "learning_rate": 0.00046014995023410324,
      "loss": 2.2748,
      "step": 73902
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.8113258481025696,
      "learning_rate": 0.00046014649128147165,
      "loss": 2.2069,
      "step": 73903
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.6987308263778687,
      "learning_rate": 0.00046014303229906575,
      "loss": 2.0346,
      "step": 73904
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7702184915542603,
      "learning_rate": 0.0004601395732868864,
      "loss": 2.1908,
      "step": 73905
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7422174215316772,
      "learning_rate": 0.0004601361142449339,
      "loss": 2.3507,
      "step": 73906
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7775757908821106,
      "learning_rate": 0.0004601326551732091,
      "loss": 2.1624,
      "step": 73907
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.8503140807151794,
      "learning_rate": 0.00046012919607171283,
      "loss": 2.3606,
      "step": 73908
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.8363019824028015,
      "learning_rate": 0.0004601257369404453,
      "loss": 2.1471,
      "step": 73909
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7943437099456787,
      "learning_rate": 0.0004601222777794074,
      "loss": 2.4119,
      "step": 73910
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7052733302116394,
      "learning_rate": 0.0004601188185885999,
      "loss": 2.3444,
      "step": 73911
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7612562775611877,
      "learning_rate": 0.00046011535936802306,
      "loss": 2.1293,
      "step": 73912
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7975199222564697,
      "learning_rate": 0.0004601119001176779,
      "loss": 2.1173,
      "step": 73913
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.8042423725128174,
      "learning_rate": 0.00046010844083756495,
      "loss": 2.1629,
      "step": 73914
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.6335945129394531,
      "learning_rate": 0.0004601049815276847,
      "loss": 2.0919,
      "step": 73915
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7279382944107056,
      "learning_rate": 0.0004601015221880379,
      "loss": 2.0581,
      "step": 73916
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.8262439370155334,
      "learning_rate": 0.00046009806281862534,
      "loss": 2.0767,
      "step": 73917
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.8401778936386108,
      "learning_rate": 0.00046009460341944747,
      "loss": 2.2641,
      "step": 73918
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7314215302467346,
      "learning_rate": 0.000460091143990505,
      "loss": 2.1858,
      "step": 73919
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7551701664924622,
      "learning_rate": 0.0004600876845317984,
      "loss": 2.1662,
      "step": 73920
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.8004679679870605,
      "learning_rate": 0.00046008422504332873,
      "loss": 2.1574,
      "step": 73921
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.8095916509628296,
      "learning_rate": 0.00046008076552509625,
      "loss": 2.2815,
      "step": 73922
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7294262647628784,
      "learning_rate": 0.0004600773059771017,
      "loss": 2.2635,
      "step": 73923
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.8117780685424805,
      "learning_rate": 0.0004600738463993458,
      "loss": 2.0616,
      "step": 73924
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.8518773913383484,
      "learning_rate": 0.00046007038679182923,
      "loss": 2.3118,
      "step": 73925
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7584139108657837,
      "learning_rate": 0.00046006692715455243,
      "loss": 2.1494,
      "step": 73926
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.8672510385513306,
      "learning_rate": 0.00046006346748751623,
      "loss": 2.3201,
      "step": 73927
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.8807649612426758,
      "learning_rate": 0.0004600600077907212,
      "loss": 2.1469,
      "step": 73928
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7708418965339661,
      "learning_rate": 0.0004600565480641679,
      "loss": 2.3054,
      "step": 73929
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.8807337284088135,
      "learning_rate": 0.0004600530883078572,
      "loss": 2.3512,
      "step": 73930
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7267431616783142,
      "learning_rate": 0.0004600496285217896,
      "loss": 2.3739,
      "step": 73931
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.8755843639373779,
      "learning_rate": 0.0004600461687059657,
      "loss": 2.0647,
      "step": 73932
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.8666558265686035,
      "learning_rate": 0.0004600427088603862,
      "loss": 2.0391,
      "step": 73933
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.8324338793754578,
      "learning_rate": 0.00046003924898505185,
      "loss": 2.3409,
      "step": 73934
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7514062523841858,
      "learning_rate": 0.00046003578907996307,
      "loss": 2.2769,
      "step": 73935
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7374660968780518,
      "learning_rate": 0.00046003232914512064,
      "loss": 2.3487,
      "step": 73936
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.9491842985153198,
      "learning_rate": 0.00046002886918052516,
      "loss": 2.2262,
      "step": 73937
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.8114445209503174,
      "learning_rate": 0.0004600254091861773,
      "loss": 2.3237,
      "step": 73938
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7911621928215027,
      "learning_rate": 0.0004600219491620777,
      "loss": 2.2006,
      "step": 73939
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7366012930870056,
      "learning_rate": 0.000460018489108227,
      "loss": 1.9753,
      "step": 73940
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.8736607432365417,
      "learning_rate": 0.00046001502902462583,
      "loss": 2.2607,
      "step": 73941
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7971757650375366,
      "learning_rate": 0.0004600115689112749,
      "loss": 1.9357,
      "step": 73942
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7908602952957153,
      "learning_rate": 0.00046000810876817473,
      "loss": 2.3601,
      "step": 73943
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.8858931064605713,
      "learning_rate": 0.00046000464859532607,
      "loss": 2.0489,
      "step": 73944
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7854649424552917,
      "learning_rate": 0.0004600011883927295,
      "loss": 2.1767,
      "step": 73945
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.8321418166160583,
      "learning_rate": 0.0004599977281603857,
      "loss": 2.2753,
      "step": 73946
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7672718167304993,
      "learning_rate": 0.0004599942678982953,
      "loss": 2.2198,
      "step": 73947
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.8477258086204529,
      "learning_rate": 0.00045999080760645894,
      "loss": 2.4304,
      "step": 73948
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7636576890945435,
      "learning_rate": 0.0004599873472848773,
      "loss": 2.2253,
      "step": 73949
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.8332922458648682,
      "learning_rate": 0.00045998388693355103,
      "loss": 2.0357,
      "step": 73950
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7270333170890808,
      "learning_rate": 0.0004599804265524806,
      "loss": 2.1139,
      "step": 73951
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7671520113945007,
      "learning_rate": 0.00045997696614166685,
      "loss": 2.1481,
      "step": 73952
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7648152709007263,
      "learning_rate": 0.00045997350570111046,
      "loss": 2.1726,
      "step": 73953
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7620077729225159,
      "learning_rate": 0.0004599700452308119,
      "loss": 2.1162,
      "step": 73954
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.8406445980072021,
      "learning_rate": 0.00045996658473077185,
      "loss": 2.3105,
      "step": 73955
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7228270173072815,
      "learning_rate": 0.00045996312420099105,
      "loss": 2.2968,
      "step": 73956
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7371822595596313,
      "learning_rate": 0.00045995966364147003,
      "loss": 2.3556,
      "step": 73957
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.732923150062561,
      "learning_rate": 0.00045995620305220946,
      "loss": 2.1367,
      "step": 73958
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7954069375991821,
      "learning_rate": 0.00045995274243321014,
      "loss": 1.9726,
      "step": 73959
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.8559157252311707,
      "learning_rate": 0.00045994928178447257,
      "loss": 2.3513,
      "step": 73960
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7429178953170776,
      "learning_rate": 0.0004599458211059973,
      "loss": 2.0775,
      "step": 73961
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7617287039756775,
      "learning_rate": 0.00045994236039778515,
      "loss": 2.0022,
      "step": 73962
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7583197355270386,
      "learning_rate": 0.00045993889965983666,
      "loss": 2.1361,
      "step": 73963
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7407722473144531,
      "learning_rate": 0.0004599354388921525,
      "loss": 2.3429,
      "step": 73964
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.8615928292274475,
      "learning_rate": 0.00045993197809473347,
      "loss": 2.2333,
      "step": 73965
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.8153601884841919,
      "learning_rate": 0.00045992851726757997,
      "loss": 2.2563,
      "step": 73966
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.8384479880332947,
      "learning_rate": 0.0004599250564106926,
      "loss": 2.0966,
      "step": 73967
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.8582465052604675,
      "learning_rate": 0.00045992159552407235,
      "loss": 2.3306,
      "step": 73968
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7986895442008972,
      "learning_rate": 0.0004599181346077196,
      "loss": 2.243,
      "step": 73969
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7626840472221375,
      "learning_rate": 0.00045991467366163506,
      "loss": 2.267,
      "step": 73970
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7805086970329285,
      "learning_rate": 0.00045991121268581923,
      "loss": 2.2002,
      "step": 73971
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7689128518104553,
      "learning_rate": 0.00045990775168027305,
      "loss": 2.2184,
      "step": 73972
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7472233176231384,
      "learning_rate": 0.0004599042906449969,
      "loss": 2.3354,
      "step": 73973
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.8677016496658325,
      "learning_rate": 0.00045990082957999157,
      "loss": 2.49,
      "step": 73974
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.8059542775154114,
      "learning_rate": 0.0004598973684852577,
      "loss": 2.2594,
      "step": 73975
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7640796303749084,
      "learning_rate": 0.0004598939073607958,
      "loss": 2.3224,
      "step": 73976
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.8271854519844055,
      "learning_rate": 0.0004598904462066066,
      "loss": 2.3203,
      "step": 73977
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7943880558013916,
      "learning_rate": 0.00045988698502269086,
      "loss": 2.1591,
      "step": 73978
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7044112086296082,
      "learning_rate": 0.00045988352380904905,
      "loss": 1.9669,
      "step": 73979
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7517379522323608,
      "learning_rate": 0.00045988006256568177,
      "loss": 2.2782,
      "step": 73980
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7106034159660339,
      "learning_rate": 0.0004598766012925899,
      "loss": 2.0284,
      "step": 73981
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7787261009216309,
      "learning_rate": 0.0004598731399897739,
      "loss": 2.2979,
      "step": 73982
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.8148598074913025,
      "learning_rate": 0.00045986967865723453,
      "loss": 2.5206,
      "step": 73983
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.796900749206543,
      "learning_rate": 0.00045986621729497234,
      "loss": 2.2925,
      "step": 73984
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7863552570343018,
      "learning_rate": 0.00045986275590298795,
      "loss": 2.2299,
      "step": 73985
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7810335755348206,
      "learning_rate": 0.0004598592944812821,
      "loss": 2.1021,
      "step": 73986
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7781426310539246,
      "learning_rate": 0.0004598558330298554,
      "loss": 2.3199,
      "step": 73987
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7348865270614624,
      "learning_rate": 0.0004598523715487085,
      "loss": 2.0648,
      "step": 73988
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7315459251403809,
      "learning_rate": 0.00045984891003784186,
      "loss": 2.3851,
      "step": 73989
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7728059887886047,
      "learning_rate": 0.0004598454484972565,
      "loss": 2.1409,
      "step": 73990
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7297972440719604,
      "learning_rate": 0.00045984198692695276,
      "loss": 2.0159,
      "step": 73991
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7755027413368225,
      "learning_rate": 0.0004598385253269313,
      "loss": 2.2115,
      "step": 73992
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.8017567992210388,
      "learning_rate": 0.00045983506369719305,
      "loss": 2.1961,
      "step": 73993
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.837499737739563,
      "learning_rate": 0.00045983160203773836,
      "loss": 2.2453,
      "step": 73994
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7513158917427063,
      "learning_rate": 0.00045982814034856785,
      "loss": 2.2,
      "step": 73995
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7841541171073914,
      "learning_rate": 0.0004598246786296824,
      "loss": 2.1536,
      "step": 73996
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7670043110847473,
      "learning_rate": 0.0004598212168810825,
      "loss": 2.0695,
      "step": 73997
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7849106788635254,
      "learning_rate": 0.0004598177551027688,
      "loss": 2.3643,
      "step": 73998
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7880113124847412,
      "learning_rate": 0.0004598142932947419,
      "loss": 2.2495,
      "step": 73999
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7900598645210266,
      "learning_rate": 0.0004598108314570026,
      "loss": 2.4337,
      "step": 74000
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.9527093768119812,
      "learning_rate": 0.00045980736958955136,
      "loss": 2.2615,
      "step": 74001
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7690160870552063,
      "learning_rate": 0.00045980390769238903,
      "loss": 2.0884,
      "step": 74002
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.827881932258606,
      "learning_rate": 0.00045980044576551603,
      "loss": 2.2111,
      "step": 74003
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7691105604171753,
      "learning_rate": 0.0004597969838089332,
      "loss": 2.3309,
      "step": 74004
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7670857310295105,
      "learning_rate": 0.000459793521822641,
      "loss": 2.2524,
      "step": 74005
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7547605633735657,
      "learning_rate": 0.0004597900598066402,
      "loss": 2.1443,
      "step": 74006
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.9209322333335876,
      "learning_rate": 0.0004597865977609314,
      "loss": 2.2864,
      "step": 74007
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.8071604371070862,
      "learning_rate": 0.0004597831356855153,
      "loss": 2.0439,
      "step": 74008
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7412691712379456,
      "learning_rate": 0.0004597796735803925,
      "loss": 2.0297,
      "step": 74009
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.8982515335083008,
      "learning_rate": 0.00045977621144556365,
      "loss": 2.0648,
      "step": 74010
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7289302945137024,
      "learning_rate": 0.00045977274928102926,
      "loss": 2.2643,
      "step": 74011
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.8213066458702087,
      "learning_rate": 0.0004597692870867902,
      "loss": 2.1742,
      "step": 74012
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7189167737960815,
      "learning_rate": 0.0004597658248628471,
      "loss": 2.0731,
      "step": 74013
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7704986333847046,
      "learning_rate": 0.0004597623626092004,
      "loss": 2.055,
      "step": 74014
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.8086460828781128,
      "learning_rate": 0.00045975890032585087,
      "loss": 2.2004,
      "step": 74015
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7496984004974365,
      "learning_rate": 0.00045975543801279916,
      "loss": 2.1629,
      "step": 74016
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7598862648010254,
      "learning_rate": 0.0004597519756700459,
      "loss": 2.1862,
      "step": 74017
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.8465682864189148,
      "learning_rate": 0.0004597485132975917,
      "loss": 2.1427,
      "step": 74018
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7955421805381775,
      "learning_rate": 0.00045974505089543734,
      "loss": 2.1972,
      "step": 74019
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7930959463119507,
      "learning_rate": 0.0004597415884635833,
      "loss": 2.6116,
      "step": 74020
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7681795358657837,
      "learning_rate": 0.0004597381260020302,
      "loss": 2.0989,
      "step": 74021
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7827035188674927,
      "learning_rate": 0.0004597346635107788,
      "loss": 2.3662,
      "step": 74022
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.8209987282752991,
      "learning_rate": 0.0004597312009898299,
      "loss": 2.2608,
      "step": 74023
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7481627464294434,
      "learning_rate": 0.0004597277384391837,
      "loss": 2.3526,
      "step": 74024
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7464035749435425,
      "learning_rate": 0.0004597242758588412,
      "loss": 2.1056,
      "step": 74025
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.8049894571304321,
      "learning_rate": 0.000459720813248803,
      "loss": 2.1209,
      "step": 74026
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7398363947868347,
      "learning_rate": 0.00045971735060906956,
      "loss": 2.0621,
      "step": 74027
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7849779725074768,
      "learning_rate": 0.00045971388793964173,
      "loss": 2.1335,
      "step": 74028
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.2957282066345215,
      "learning_rate": 0.00045971042524052015,
      "loss": 2.4413,
      "step": 74029
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7537211775779724,
      "learning_rate": 0.00045970696251170526,
      "loss": 1.9809,
      "step": 74030
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.790280282497406,
      "learning_rate": 0.00045970349975319786,
      "loss": 2.1158,
      "step": 74031
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7956268191337585,
      "learning_rate": 0.00045970003696499856,
      "loss": 2.0628,
      "step": 74032
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.8712283372879028,
      "learning_rate": 0.00045969657414710806,
      "loss": 2.2597,
      "step": 74033
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.870934009552002,
      "learning_rate": 0.0004596931112995269,
      "loss": 2.2655,
      "step": 74034
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7772731781005859,
      "learning_rate": 0.00045968964842225577,
      "loss": 2.3843,
      "step": 74035
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.8536625504493713,
      "learning_rate": 0.0004596861855152954,
      "loss": 2.1705,
      "step": 74036
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.8952901363372803,
      "learning_rate": 0.00045968272257864627,
      "loss": 2.1088,
      "step": 74037
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7843181490898132,
      "learning_rate": 0.00045967925961230914,
      "loss": 2.2479,
      "step": 74038
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.8353151679039001,
      "learning_rate": 0.00045967579661628465,
      "loss": 2.1555,
      "step": 74039
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.8246362209320068,
      "learning_rate": 0.00045967233359057335,
      "loss": 2.0179,
      "step": 74040
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7748635411262512,
      "learning_rate": 0.00045966887053517595,
      "loss": 2.1589,
      "step": 74041
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7176334261894226,
      "learning_rate": 0.0004596654074500932,
      "loss": 2.1124,
      "step": 74042
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.9377711415290833,
      "learning_rate": 0.00045966194433532553,
      "loss": 2.1981,
      "step": 74043
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.8148777484893799,
      "learning_rate": 0.00045965848119087376,
      "loss": 2.0971,
      "step": 74044
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7844366431236267,
      "learning_rate": 0.00045965501801673844,
      "loss": 2.2437,
      "step": 74045
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.781466007232666,
      "learning_rate": 0.00045965155481292016,
      "loss": 2.1556,
      "step": 74046
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.8175665736198425,
      "learning_rate": 0.0004596480915794197,
      "loss": 2.2731,
      "step": 74047
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.767656147480011,
      "learning_rate": 0.0004596446283162377,
      "loss": 2.3434,
      "step": 74048
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.8584785461425781,
      "learning_rate": 0.0004596411650233747,
      "loss": 2.1217,
      "step": 74049
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7463973164558411,
      "learning_rate": 0.0004596377017008314,
      "loss": 2.0924,
      "step": 74050
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.8624234795570374,
      "learning_rate": 0.00045963423834860853,
      "loss": 2.071,
      "step": 74051
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7866773009300232,
      "learning_rate": 0.00045963077496670646,
      "loss": 2.1344,
      "step": 74052
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7329644560813904,
      "learning_rate": 0.0004596273115551261,
      "loss": 2.293,
      "step": 74053
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.8205497860908508,
      "learning_rate": 0.0004596238481138681,
      "loss": 2.1326,
      "step": 74054
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7271756529808044,
      "learning_rate": 0.0004596203846429328,
      "loss": 2.3125,
      "step": 74055
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7594128251075745,
      "learning_rate": 0.0004596169211423212,
      "loss": 2.1199,
      "step": 74056
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7446795701980591,
      "learning_rate": 0.00045961345761203387,
      "loss": 2.2806,
      "step": 74057
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7650343775749207,
      "learning_rate": 0.0004596099940520712,
      "loss": 2.4068,
      "step": 74058
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.8425419926643372,
      "learning_rate": 0.0004596065304624341,
      "loss": 2.2013,
      "step": 74059
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7565872073173523,
      "learning_rate": 0.00045960306684312326,
      "loss": 2.3829,
      "step": 74060
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7946435213088989,
      "learning_rate": 0.00045959960319413905,
      "loss": 2.1512,
      "step": 74061
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7507909536361694,
      "learning_rate": 0.00045959613951548233,
      "loss": 2.0522,
      "step": 74062
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7571834325790405,
      "learning_rate": 0.0004595926758071536,
      "loss": 2.152,
      "step": 74063
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7939822673797607,
      "learning_rate": 0.0004595892120691537,
      "loss": 2.1624,
      "step": 74064
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7597811222076416,
      "learning_rate": 0.00045958574830148305,
      "loss": 2.1801,
      "step": 74065
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7793367505073547,
      "learning_rate": 0.0004595822845041424,
      "loss": 2.1394,
      "step": 74066
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7408576607704163,
      "learning_rate": 0.00045957882067713246,
      "loss": 2.1806,
      "step": 74067
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7676994800567627,
      "learning_rate": 0.0004595753568204537,
      "loss": 2.0284,
      "step": 74068
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7662912607192993,
      "learning_rate": 0.00045957189293410695,
      "loss": 2.2145,
      "step": 74069
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7520825266838074,
      "learning_rate": 0.0004595684290180928,
      "loss": 1.8618,
      "step": 74070
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7439628839492798,
      "learning_rate": 0.0004595649650724118,
      "loss": 2.2315,
      "step": 74071
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7872671484947205,
      "learning_rate": 0.00045956150109706476,
      "loss": 2.1763,
      "step": 74072
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.833890974521637,
      "learning_rate": 0.0004595580370920521,
      "loss": 2.3235,
      "step": 74073
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7550393342971802,
      "learning_rate": 0.00045955457305737454,
      "loss": 2.1857,
      "step": 74074
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.8579209446907043,
      "learning_rate": 0.0004595511089930329,
      "loss": 2.3414,
      "step": 74075
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7215877175331116,
      "learning_rate": 0.00045954764489902766,
      "loss": 2.2958,
      "step": 74076
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.8362372517585754,
      "learning_rate": 0.00045954418077535957,
      "loss": 2.1947,
      "step": 74077
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.9872560501098633,
      "learning_rate": 0.00045954071662202917,
      "loss": 2.4144,
      "step": 74078
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.8303114175796509,
      "learning_rate": 0.0004595372524390371,
      "loss": 2.2788,
      "step": 74079
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.8296827673912048,
      "learning_rate": 0.00045953378822638404,
      "loss": 2.3963,
      "step": 74080
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.80228191614151,
      "learning_rate": 0.0004595303239840707,
      "loss": 2.1136,
      "step": 74081
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.777433454990387,
      "learning_rate": 0.00045952685971209755,
      "loss": 2.2882,
      "step": 74082
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.0246084928512573,
      "learning_rate": 0.0004595233954104654,
      "loss": 2.0128,
      "step": 74083
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7454330325126648,
      "learning_rate": 0.00045951993107917494,
      "loss": 2.2719,
      "step": 74084
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.809541642665863,
      "learning_rate": 0.00045951646671822655,
      "loss": 2.1283,
      "step": 74085
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.806530237197876,
      "learning_rate": 0.0004595130023276211,
      "loss": 2.2425,
      "step": 74086
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7655007839202881,
      "learning_rate": 0.0004595095379073592,
      "loss": 2.1438,
      "step": 74087
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.72272789478302,
      "learning_rate": 0.0004595060734574415,
      "loss": 2.1547,
      "step": 74088
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.6978434920310974,
      "learning_rate": 0.00045950260897786855,
      "loss": 2.131,
      "step": 74089
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.819107174873352,
      "learning_rate": 0.00045949914446864103,
      "loss": 2.2925,
      "step": 74090
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7307577729225159,
      "learning_rate": 0.00045949567992975966,
      "loss": 2.2761,
      "step": 74091
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7619660496711731,
      "learning_rate": 0.00045949221536122493,
      "loss": 2.3151,
      "step": 74092
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7736756205558777,
      "learning_rate": 0.00045948875076303765,
      "loss": 2.031,
      "step": 74093
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7521518468856812,
      "learning_rate": 0.00045948528613519847,
      "loss": 2.3106,
      "step": 74094
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.8575418591499329,
      "learning_rate": 0.00045948182147770793,
      "loss": 2.1128,
      "step": 74095
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7077165246009827,
      "learning_rate": 0.00045947835679056664,
      "loss": 2.335,
      "step": 74096
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.8043309450149536,
      "learning_rate": 0.00045947489207377534,
      "loss": 2.2906,
      "step": 74097
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.8299197554588318,
      "learning_rate": 0.0004594714273273348,
      "loss": 2.0944,
      "step": 74098
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7496111989021301,
      "learning_rate": 0.0004594679625512453,
      "loss": 2.2489,
      "step": 74099
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7668317556381226,
      "learning_rate": 0.00045946449774550776,
      "loss": 2.136,
      "step": 74100
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7942520976066589,
      "learning_rate": 0.0004594610329101229,
      "loss": 2.1226,
      "step": 74101
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7404314875602722,
      "learning_rate": 0.0004594575680450911,
      "loss": 2.1935,
      "step": 74102
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7968283891677856,
      "learning_rate": 0.000459454103150413,
      "loss": 2.2084,
      "step": 74103
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7512412667274475,
      "learning_rate": 0.0004594506382260895,
      "loss": 2.2104,
      "step": 74104
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.8010262846946716,
      "learning_rate": 0.00045944717327212124,
      "loss": 2.0765,
      "step": 74105
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7194793820381165,
      "learning_rate": 0.0004594437082885085,
      "loss": 2.2255,
      "step": 74106
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.8189378976821899,
      "learning_rate": 0.00045944024327525235,
      "loss": 2.1583,
      "step": 74107
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.0494779348373413,
      "learning_rate": 0.0004594367782323531,
      "loss": 2.3286,
      "step": 74108
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7917590737342834,
      "learning_rate": 0.0004594333131598117,
      "loss": 2.2858,
      "step": 74109
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7440814971923828,
      "learning_rate": 0.0004594298480576286,
      "loss": 1.969,
      "step": 74110
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.8652300834655762,
      "learning_rate": 0.00045942638292580443,
      "loss": 2.1777,
      "step": 74111
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7732957601547241,
      "learning_rate": 0.00045942291776433986,
      "loss": 2.2452,
      "step": 74112
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.7256274819374084,
      "learning_rate": 0.00045941945257323564,
      "loss": 1.9569,
      "step": 74113
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.8925941586494446,
      "learning_rate": 0.00045941598735249226,
      "loss": 2.2614,
      "step": 74114
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.8716416358947754,
      "learning_rate": 0.00045941252210211054,
      "loss": 2.3246,
      "step": 74115
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.8376812934875488,
      "learning_rate": 0.00045940905682209096,
      "loss": 2.0546,
      "step": 74116
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.7462431788444519,
      "learning_rate": 0.00045940559151243416,
      "loss": 2.3368,
      "step": 74117
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.947561502456665,
      "learning_rate": 0.00045940212617314103,
      "loss": 2.3771,
      "step": 74118
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.9745608568191528,
      "learning_rate": 0.0004593986608042119,
      "loss": 2.284,
      "step": 74119
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.010438323020935,
      "learning_rate": 0.0004593951954056475,
      "loss": 2.1694,
      "step": 74120
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.9026908278465271,
      "learning_rate": 0.0004593917299774487,
      "loss": 2.18,
      "step": 74121
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.7426998615264893,
      "learning_rate": 0.00045938826451961577,
      "loss": 2.1262,
      "step": 74122
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.7777371406555176,
      "learning_rate": 0.00045938479903214965,
      "loss": 2.0599,
      "step": 74123
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.7471713423728943,
      "learning_rate": 0.0004593813335150509,
      "loss": 2.1145,
      "step": 74124
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.7585386037826538,
      "learning_rate": 0.0004593778679683201,
      "loss": 2.0277,
      "step": 74125
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.7935360670089722,
      "learning_rate": 0.00045937440239195805,
      "loss": 2.0608,
      "step": 74126
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.7969094514846802,
      "learning_rate": 0.0004593709367859653,
      "loss": 2.3245,
      "step": 74127
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.8589740991592407,
      "learning_rate": 0.0004593674711503424,
      "loss": 2.3469,
      "step": 74128
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.7961390018463135,
      "learning_rate": 0.00045936400548509,
      "loss": 2.0923,
      "step": 74129
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.8842247724533081,
      "learning_rate": 0.0004593605397902089,
      "loss": 2.0501,
      "step": 74130
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.8137680292129517,
      "learning_rate": 0.00045935707406569967,
      "loss": 2.2399,
      "step": 74131
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.7396951913833618,
      "learning_rate": 0.000459353608311563,
      "loss": 2.067,
      "step": 74132
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.7750201225280762,
      "learning_rate": 0.0004593501425277995,
      "loss": 2.1769,
      "step": 74133
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.844560444355011,
      "learning_rate": 0.00045934667671440973,
      "loss": 2.1918,
      "step": 74134
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.7540417313575745,
      "learning_rate": 0.00045934321087139437,
      "loss": 2.2168,
      "step": 74135
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.810903787612915,
      "learning_rate": 0.0004593397449987542,
      "loss": 2.3848,
      "step": 74136
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.0258649587631226,
      "learning_rate": 0.00045933627909648976,
      "loss": 2.218,
      "step": 74137
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.7868130207061768,
      "learning_rate": 0.0004593328131646016,
      "loss": 2.015,
      "step": 74138
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.7979326844215393,
      "learning_rate": 0.0004593293472030906,
      "loss": 2.2151,
      "step": 74139
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.8258324861526489,
      "learning_rate": 0.00045932588121195715,
      "loss": 2.279,
      "step": 74140
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.9239752888679504,
      "learning_rate": 0.00045932241519120197,
      "loss": 2.363,
      "step": 74141
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.7806971073150635,
      "learning_rate": 0.0004593189491408259,
      "loss": 2.2326,
      "step": 74142
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.0351537466049194,
      "learning_rate": 0.0004593154830608293,
      "loss": 1.9557,
      "step": 74143
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.9199458956718445,
      "learning_rate": 0.000459312016951213,
      "loss": 1.9989,
      "step": 74144
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.8975297212600708,
      "learning_rate": 0.00045930855081197764,
      "loss": 2.2223,
      "step": 74145
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.9386550784111023,
      "learning_rate": 0.00045930508464312375,
      "loss": 2.4144,
      "step": 74146
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.831218421459198,
      "learning_rate": 0.00045930161844465206,
      "loss": 2.0233,
      "step": 74147
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.7500979900360107,
      "learning_rate": 0.00045929815221656316,
      "loss": 2.106,
      "step": 74148
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.8364932537078857,
      "learning_rate": 0.00045929468595885765,
      "loss": 2.3216,
      "step": 74149
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.7598150372505188,
      "learning_rate": 0.00045929121967153635,
      "loss": 2.3417,
      "step": 74150
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.7823778986930847,
      "learning_rate": 0.0004592877533545999,
      "loss": 2.2428,
      "step": 74151
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.8957590460777283,
      "learning_rate": 0.0004592842870080487,
      "loss": 2.1239,
      "step": 74152
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.7586519122123718,
      "learning_rate": 0.0004592808206318836,
      "loss": 2.0089,
      "step": 74153
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.0347050428390503,
      "learning_rate": 0.0004592773542261052,
      "loss": 2.0778,
      "step": 74154
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.7479154467582703,
      "learning_rate": 0.0004592738877907141,
      "loss": 2.324,
      "step": 74155
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.0288112163543701,
      "learning_rate": 0.000459270421325711,
      "loss": 2.1053,
      "step": 74156
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.7336037158966064,
      "learning_rate": 0.0004592669548310965,
      "loss": 1.9028,
      "step": 74157
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.7222870588302612,
      "learning_rate": 0.0004592634883068713,
      "loss": 2.14,
      "step": 74158
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.8217942714691162,
      "learning_rate": 0.00045926002175303597,
      "loss": 2.2038,
      "step": 74159
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.7864879965782166,
      "learning_rate": 0.0004592565551695913,
      "loss": 2.109,
      "step": 74160
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.7829383015632629,
      "learning_rate": 0.00045925308855653766,
      "loss": 2.038,
      "step": 74161
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.8367012739181519,
      "learning_rate": 0.000459249621913876,
      "loss": 2.2803,
      "step": 74162
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.773565948009491,
      "learning_rate": 0.0004592461552416068,
      "loss": 2.3109,
      "step": 74163
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.8594460487365723,
      "learning_rate": 0.0004592426885397307,
      "loss": 2.2003,
      "step": 74164
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.7976549863815308,
      "learning_rate": 0.0004592392218082484,
      "loss": 2.4455,
      "step": 74165
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.8163631558418274,
      "learning_rate": 0.0004592357550471605,
      "loss": 2.1055,
      "step": 74166
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.7520624399185181,
      "learning_rate": 0.0004592322882564677,
      "loss": 2.2737,
      "step": 74167
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.8147267699241638,
      "learning_rate": 0.0004592288214361706,
      "loss": 2.1697,
      "step": 74168
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.7308835387229919,
      "learning_rate": 0.00045922535458626986,
      "loss": 2.1554,
      "step": 74169
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.79849773645401,
      "learning_rate": 0.0004592218877067661,
      "loss": 2.1736,
      "step": 74170
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.7776941061019897,
      "learning_rate": 0.00045921842079766007,
      "loss": 2.363,
      "step": 74171
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.7432680130004883,
      "learning_rate": 0.0004592149538589522,
      "loss": 2.2087,
      "step": 74172
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.7408596873283386,
      "learning_rate": 0.00045921148689064337,
      "loss": 2.4342,
      "step": 74173
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.7578748464584351,
      "learning_rate": 0.00045920801989273406,
      "loss": 1.9551,
      "step": 74174
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.8519943356513977,
      "learning_rate": 0.00045920455286522496,
      "loss": 2.1133,
      "step": 74175
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.7646047472953796,
      "learning_rate": 0.00045920108580811675,
      "loss": 2.1548,
      "step": 74176
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.8129486441612244,
      "learning_rate": 0.0004591976187214101,
      "loss": 2.3617,
      "step": 74177
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.8564357757568359,
      "learning_rate": 0.0004591941516051055,
      "loss": 2.1235,
      "step": 74178
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.835506021976471,
      "learning_rate": 0.0004591906844592038,
      "loss": 2.2801,
      "step": 74179
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.8337203860282898,
      "learning_rate": 0.00045918721728370547,
      "loss": 2.1964,
      "step": 74180
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.7663065791130066,
      "learning_rate": 0.0004591837500786113,
      "loss": 2.2108,
      "step": 74181
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.909567654132843,
      "learning_rate": 0.0004591802828439219,
      "loss": 2.1122,
      "step": 74182
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.8611263036727905,
      "learning_rate": 0.00045917681557963784,
      "loss": 2.1539,
      "step": 74183
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.8138033747673035,
      "learning_rate": 0.0004591733482857597,
      "loss": 2.3579,
      "step": 74184
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.7652063965797424,
      "learning_rate": 0.0004591698809622883,
      "loss": 2.2581,
      "step": 74185
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.6972942352294922,
      "learning_rate": 0.00045916641360922424,
      "loss": 2.3628,
      "step": 74186
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.9381478428840637,
      "learning_rate": 0.0004591629462265682,
      "loss": 2.2334,
      "step": 74187
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.7448172569274902,
      "learning_rate": 0.00045915947881432066,
      "loss": 2.1137,
      "step": 74188
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.7456708550453186,
      "learning_rate": 0.0004591560113724824,
      "loss": 2.2905,
      "step": 74189
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.7876035571098328,
      "learning_rate": 0.000459152543901054,
      "loss": 2.2613,
      "step": 74190
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.7487939596176147,
      "learning_rate": 0.00045914907640003626,
      "loss": 1.9119,
      "step": 74191
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.7502227425575256,
      "learning_rate": 0.0004591456088694296,
      "loss": 2.1662,
      "step": 74192
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.7985246777534485,
      "learning_rate": 0.00045914214130923477,
      "loss": 2.0613,
      "step": 74193
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.7591603398323059,
      "learning_rate": 0.00045913867371945245,
      "loss": 2.1814,
      "step": 74194
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.8262767195701599,
      "learning_rate": 0.0004591352061000832,
      "loss": 2.1033,
      "step": 74195
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.8606953620910645,
      "learning_rate": 0.00045913173845112766,
      "loss": 2.2649,
      "step": 74196
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.7990938425064087,
      "learning_rate": 0.0004591282707725867,
      "loss": 2.0342,
      "step": 74197
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.83775395154953,
      "learning_rate": 0.00045912480306446073,
      "loss": 2.0334,
      "step": 74198
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.8874098062515259,
      "learning_rate": 0.0004591213353267503,
      "loss": 2.016,
      "step": 74199
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.8656566143035889,
      "learning_rate": 0.0004591178675594564,
      "loss": 2.3224,
      "step": 74200
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.7251991629600525,
      "learning_rate": 0.0004591143997625794,
      "loss": 2.1826,
      "step": 74201
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.7993119955062866,
      "learning_rate": 0.00045911093193612,
      "loss": 2.2409,
      "step": 74202
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.8059959411621094,
      "learning_rate": 0.000459107464080079,
      "loss": 2.4142,
      "step": 74203
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.9179877638816833,
      "learning_rate": 0.0004591039961944568,
      "loss": 2.1089,
      "step": 74204
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.8445175886154175,
      "learning_rate": 0.0004591005282792542,
      "loss": 2.2617,
      "step": 74205
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.8295363783836365,
      "learning_rate": 0.00045909706033447185,
      "loss": 2.2128,
      "step": 74206
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.7083798050880432,
      "learning_rate": 0.0004590935923601103,
      "loss": 2.1272,
      "step": 74207
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.8289445638656616,
      "learning_rate": 0.00045909012435617027,
      "loss": 1.8853,
      "step": 74208
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.8730139136314392,
      "learning_rate": 0.00045908665632265236,
      "loss": 2.127,
      "step": 74209
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.7968814969062805,
      "learning_rate": 0.0004590831882595572,
      "loss": 2.2032,
      "step": 74210
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.8569329380989075,
      "learning_rate": 0.00045907972016688553,
      "loss": 2.3837,
      "step": 74211
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.7949603199958801,
      "learning_rate": 0.000459076252044638,
      "loss": 2.0288,
      "step": 74212
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.934159517288208,
      "learning_rate": 0.000459072783892815,
      "loss": 2.3801,
      "step": 74213
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.8415063619613647,
      "learning_rate": 0.00045906931571141756,
      "loss": 2.0435,
      "step": 74214
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.8100424408912659,
      "learning_rate": 0.0004590658475004461,
      "loss": 2.2582,
      "step": 74215
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.8541603684425354,
      "learning_rate": 0.00045906237925990125,
      "loss": 2.3312,
      "step": 74216
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.7741979360580444,
      "learning_rate": 0.00045905891098978364,
      "loss": 2.2267,
      "step": 74217
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.8458675146102905,
      "learning_rate": 0.0004590554426900941,
      "loss": 2.3232,
      "step": 74218
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.7827706336975098,
      "learning_rate": 0.00045905197436083304,
      "loss": 2.3192,
      "step": 74219
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.8276810646057129,
      "learning_rate": 0.0004590485060020013,
      "loss": 2.0247,
      "step": 74220
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.811386227607727,
      "learning_rate": 0.0004590450376135995,
      "loss": 2.2504,
      "step": 74221
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.7827578186988831,
      "learning_rate": 0.00045904156919562806,
      "loss": 2.2195,
      "step": 74222
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.8597962260246277,
      "learning_rate": 0.00045903810074808787,
      "loss": 2.3883,
      "step": 74223
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.8150815367698669,
      "learning_rate": 0.0004590346322709796,
      "loss": 2.236,
      "step": 74224
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.7696680426597595,
      "learning_rate": 0.00045903116376430354,
      "loss": 2.1047,
      "step": 74225
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.8671023845672607,
      "learning_rate": 0.00045902769522806076,
      "loss": 2.1059,
      "step": 74226
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.8437431454658508,
      "learning_rate": 0.00045902422666225167,
      "loss": 2.2241,
      "step": 74227
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.759618878364563,
      "learning_rate": 0.00045902075806687705,
      "loss": 2.1301,
      "step": 74228
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.7473219633102417,
      "learning_rate": 0.00045901728944193736,
      "loss": 2.3265,
      "step": 74229
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.7846558690071106,
      "learning_rate": 0.0004590138207874335,
      "loss": 2.1717,
      "step": 74230
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.8551344275474548,
      "learning_rate": 0.0004590103521033658,
      "loss": 2.3755,
      "step": 74231
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.794886589050293,
      "learning_rate": 0.00045900688338973514,
      "loss": 2.2298,
      "step": 74232
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.8998664617538452,
      "learning_rate": 0.0004590034146465421,
      "loss": 2.1552,
      "step": 74233
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.8132026791572571,
      "learning_rate": 0.00045899994587378747,
      "loss": 2.3564,
      "step": 74234
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.7866875529289246,
      "learning_rate": 0.00045899647707147156,
      "loss": 2.1061,
      "step": 74235
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.8259013891220093,
      "learning_rate": 0.00045899300823959525,
      "loss": 2.1509,
      "step": 74236
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.8244385719299316,
      "learning_rate": 0.00045898953937815914,
      "loss": 2.3538,
      "step": 74237
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.7538961172103882,
      "learning_rate": 0.0004589860704871639,
      "loss": 2.1989,
      "step": 74238
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.7602307200431824,
      "learning_rate": 0.0004589826015666101,
      "loss": 2.2246,
      "step": 74239
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.828347384929657,
      "learning_rate": 0.0004589791326164985,
      "loss": 2.1129,
      "step": 74240
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.7801343202590942,
      "learning_rate": 0.0004589756636368296,
      "loss": 2.1381,
      "step": 74241
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.7651849985122681,
      "learning_rate": 0.00045897219462760416,
      "loss": 2.1939,
      "step": 74242
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.7418060302734375,
      "learning_rate": 0.00045896872558882283,
      "loss": 2.2832,
      "step": 74243
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.9167413115501404,
      "learning_rate": 0.0004589652565204862,
      "loss": 2.2135,
      "step": 74244
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.8297713398933411,
      "learning_rate": 0.00045896178742259493,
      "loss": 2.4576,
      "step": 74245
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.8155364394187927,
      "learning_rate": 0.0004589583182951496,
      "loss": 2.1895,
      "step": 74246
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.7482908964157104,
      "learning_rate": 0.000458954849138151,
      "loss": 2.1231,
      "step": 74247
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.8395313620567322,
      "learning_rate": 0.0004589513799515996,
      "loss": 2.3263,
      "step": 74248
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.7743333578109741,
      "learning_rate": 0.00045894791073549616,
      "loss": 2.1394,
      "step": 74249
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.767819881439209,
      "learning_rate": 0.0004589444414898414,
      "loss": 2.0818,
      "step": 74250
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.8518656492233276,
      "learning_rate": 0.00045894097221463575,
      "loss": 2.186,
      "step": 74251
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.7649878859519958,
      "learning_rate": 0.00045893750290988005,
      "loss": 2.0359,
      "step": 74252
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.1611186265945435,
      "learning_rate": 0.00045893403357557486,
      "loss": 2.0532,
      "step": 74253
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.7666608095169067,
      "learning_rate": 0.00045893056421172077,
      "loss": 2.0251,
      "step": 74254
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.8669617176055908,
      "learning_rate": 0.0004589270948183186,
      "loss": 2.316,
      "step": 74255
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.8656491041183472,
      "learning_rate": 0.00045892362539536876,
      "loss": 2.1917,
      "step": 74256
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.7425276041030884,
      "learning_rate": 0.00045892015594287204,
      "loss": 2.3438,
      "step": 74257
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.7624658346176147,
      "learning_rate": 0.0004589166864608291,
      "loss": 2.1911,
      "step": 74258
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.7612719535827637,
      "learning_rate": 0.00045891321694924056,
      "loss": 1.9162,
      "step": 74259
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.7833874225616455,
      "learning_rate": 0.00045890974740810705,
      "loss": 2.2556,
      "step": 74260
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.7416110634803772,
      "learning_rate": 0.0004589062778374293,
      "loss": 2.2722,
      "step": 74261
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.795936644077301,
      "learning_rate": 0.0004589028082372077,
      "loss": 2.2883,
      "step": 74262
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.7521209716796875,
      "learning_rate": 0.0004588993386074431,
      "loss": 2.34,
      "step": 74263
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.7844521403312683,
      "learning_rate": 0.0004588958689481363,
      "loss": 2.3425,
      "step": 74264
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.8016332387924194,
      "learning_rate": 0.0004588923992592875,
      "loss": 2.1073,
      "step": 74265
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.7292292714118958,
      "learning_rate": 0.0004588889295408978,
      "loss": 2.2837,
      "step": 74266
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.7020992636680603,
      "learning_rate": 0.0004588854597929675,
      "loss": 2.0144,
      "step": 74267
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.8759433031082153,
      "learning_rate": 0.0004588819900154975,
      "loss": 2.0802,
      "step": 74268
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.7857193350791931,
      "learning_rate": 0.00045887852020848834,
      "loss": 2.2389,
      "step": 74269
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.7753648161888123,
      "learning_rate": 0.0004588750503719406,
      "loss": 2.2051,
      "step": 74270
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.8403524160385132,
      "learning_rate": 0.000458871580505855,
      "loss": 2.0828,
      "step": 74271
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.8617025017738342,
      "learning_rate": 0.00045886811061023215,
      "loss": 2.3842,
      "step": 74272
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.7972829341888428,
      "learning_rate": 0.00045886464068507283,
      "loss": 2.2395,
      "step": 74273
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.0691455602645874,
      "learning_rate": 0.0004588611707303775,
      "loss": 2.4001,
      "step": 74274
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.7760077118873596,
      "learning_rate": 0.0004588577007461469,
      "loss": 2.2645,
      "step": 74275
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.8112211227416992,
      "learning_rate": 0.00045885423073238163,
      "loss": 2.448,
      "step": 74276
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.8646746277809143,
      "learning_rate": 0.0004588507606890823,
      "loss": 2.2135,
      "step": 74277
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.0597575902938843,
      "learning_rate": 0.00045884729061624967,
      "loss": 2.3272,
      "step": 74278
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.7980979681015015,
      "learning_rate": 0.0004588438205138844,
      "loss": 2.4317,
      "step": 74279
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.899256706237793,
      "learning_rate": 0.00045884035038198695,
      "loss": 2.1269,
      "step": 74280
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.9331725239753723,
      "learning_rate": 0.0004588368802205582,
      "loss": 2.2526,
      "step": 74281
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.7836596369743347,
      "learning_rate": 0.0004588334100295986,
      "loss": 2.4714,
      "step": 74282
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.7673338055610657,
      "learning_rate": 0.0004588299398091089,
      "loss": 2.0139,
      "step": 74283
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.8401187658309937,
      "learning_rate": 0.0004588264695590897,
      "loss": 2.1896,
      "step": 74284
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.1811686754226685,
      "learning_rate": 0.0004588229992795417,
      "loss": 2.1792,
      "step": 74285
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.7715456485748291,
      "learning_rate": 0.0004588195289704654,
      "loss": 2.2033,
      "step": 74286
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.0666853189468384,
      "learning_rate": 0.00045881605863186165,
      "loss": 2.2514,
      "step": 74287
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.9223931431770325,
      "learning_rate": 0.000458812588263731,
      "loss": 2.3045,
      "step": 74288
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.898609459400177,
      "learning_rate": 0.000458809117866074,
      "loss": 2.323,
      "step": 74289
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.8134385943412781,
      "learning_rate": 0.00045880564743889146,
      "loss": 2.1605,
      "step": 74290
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.0024566650390625,
      "learning_rate": 0.00045880217698218397,
      "loss": 2.1551,
      "step": 74291
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.9849187135696411,
      "learning_rate": 0.00045879870649595213,
      "loss": 2.0093,
      "step": 74292
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.792925238609314,
      "learning_rate": 0.0004587952359801966,
      "loss": 2.0289,
      "step": 74293
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.8393859267234802,
      "learning_rate": 0.0004587917654349181,
      "loss": 2.0969,
      "step": 74294
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.7406072020530701,
      "learning_rate": 0.0004587882948601171,
      "loss": 2.1765,
      "step": 74295
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.9426333904266357,
      "learning_rate": 0.00045878482425579443,
      "loss": 2.2891,
      "step": 74296
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.7186757922172546,
      "learning_rate": 0.00045878135362195074,
      "loss": 2.3289,
      "step": 74297
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.8608970046043396,
      "learning_rate": 0.0004587778829585864,
      "loss": 2.2981,
      "step": 74298
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.8487495183944702,
      "learning_rate": 0.0004587744122657024,
      "loss": 2.1417,
      "step": 74299
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.8212424516677856,
      "learning_rate": 0.00045877094154329916,
      "loss": 2.228,
      "step": 74300
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.7884172201156616,
      "learning_rate": 0.0004587674707913775,
      "loss": 2.1235,
      "step": 74301
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.9906315207481384,
      "learning_rate": 0.0004587640000099379,
      "loss": 2.219,
      "step": 74302
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.0338664054870605,
      "learning_rate": 0.0004587605291989811,
      "loss": 2.4433,
      "step": 74303
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.8506498336791992,
      "learning_rate": 0.0004587570583585078,
      "loss": 2.2059,
      "step": 74304
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.8290531635284424,
      "learning_rate": 0.00045875358748851845,
      "loss": 1.9574,
      "step": 74305
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.18927800655365,
      "learning_rate": 0.00045875011658901384,
      "loss": 2.1971,
      "step": 74306
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.7824074029922485,
      "learning_rate": 0.00045874664565999466,
      "loss": 2.2194,
      "step": 74307
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.7378257513046265,
      "learning_rate": 0.00045874317470146137,
      "loss": 2.128,
      "step": 74308
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.41386079788208,
      "learning_rate": 0.00045873970371341474,
      "loss": 2.1217,
      "step": 74309
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.1751874685287476,
      "learning_rate": 0.0004587362326958555,
      "loss": 2.0839,
      "step": 74310
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.8436093330383301,
      "learning_rate": 0.0004587327616487841,
      "loss": 2.2467,
      "step": 74311
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.9954465627670288,
      "learning_rate": 0.0004587292905722013,
      "loss": 2.1001,
      "step": 74312
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.2198129892349243,
      "learning_rate": 0.00045872581946610776,
      "loss": 2.1205,
      "step": 74313
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.8591641783714294,
      "learning_rate": 0.0004587223483305041,
      "loss": 2.0624,
      "step": 74314
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.8255946636199951,
      "learning_rate": 0.0004587188771653909,
      "loss": 2.3086,
      "step": 74315
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.8442099690437317,
      "learning_rate": 0.000458715405970769,
      "loss": 2.159,
      "step": 74316
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.071306824684143,
      "learning_rate": 0.00045871193474663875,
      "loss": 2.2792,
      "step": 74317
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.7537564039230347,
      "learning_rate": 0.00045870846349300103,
      "loss": 2.084,
      "step": 74318
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.9259750843048096,
      "learning_rate": 0.00045870499220985637,
      "loss": 2.1766,
      "step": 74319
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.8994112014770508,
      "learning_rate": 0.0004587015208972055,
      "loss": 2.1779,
      "step": 74320
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.8915225863456726,
      "learning_rate": 0.000458698049555049,
      "loss": 2.0187,
      "step": 74321
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.9306488633155823,
      "learning_rate": 0.0004586945781833876,
      "loss": 2.2067,
      "step": 74322
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.8157747983932495,
      "learning_rate": 0.0004586911067822218,
      "loss": 2.2704,
      "step": 74323
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.8821114301681519,
      "learning_rate": 0.0004586876353515523,
      "loss": 2.1783,
      "step": 74324
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.8015962243080139,
      "learning_rate": 0.0004586841638913799,
      "loss": 2.3684,
      "step": 74325
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.9633042216300964,
      "learning_rate": 0.00045868069240170503,
      "loss": 2.418,
      "step": 74326
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.8421227931976318,
      "learning_rate": 0.00045867722088252847,
      "loss": 2.1856,
      "step": 74327
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.8165438771247864,
      "learning_rate": 0.0004586737493338508,
      "loss": 2.4466,
      "step": 74328
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.7985460162162781,
      "learning_rate": 0.0004586702777556726,
      "loss": 2.1855,
      "step": 74329
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.744922399520874,
      "learning_rate": 0.00045866680614799476,
      "loss": 2.3802,
      "step": 74330
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.800821840763092,
      "learning_rate": 0.0004586633345108177,
      "loss": 2.1588,
      "step": 74331
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.817338764667511,
      "learning_rate": 0.00045865986284414204,
      "loss": 2.154,
      "step": 74332
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.8782045245170593,
      "learning_rate": 0.0004586563911479686,
      "loss": 2.2188,
      "step": 74333
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.7897236943244934,
      "learning_rate": 0.00045865291942229794,
      "loss": 2.2428,
      "step": 74334
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.8480792045593262,
      "learning_rate": 0.0004586494476671307,
      "loss": 2.1394,
      "step": 74335
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.7634388208389282,
      "learning_rate": 0.0004586459758824675,
      "loss": 2.1315,
      "step": 74336
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.8106369376182556,
      "learning_rate": 0.0004586425040683091,
      "loss": 2.2974,
      "step": 74337
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.8333921432495117,
      "learning_rate": 0.000458639032224656,
      "loss": 2.0066,
      "step": 74338
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.7972092628479004,
      "learning_rate": 0.0004586355603515089,
      "loss": 2.4001,
      "step": 74339
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.166350245475769,
      "learning_rate": 0.0004586320884488685,
      "loss": 2.2708,
      "step": 74340
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.8315321207046509,
      "learning_rate": 0.0004586286165167354,
      "loss": 2.3724,
      "step": 74341
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.1725603342056274,
      "learning_rate": 0.00045862514455511017,
      "loss": 2.3054,
      "step": 74342
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.9676968455314636,
      "learning_rate": 0.00045862167256399363,
      "loss": 2.2769,
      "step": 74343
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.9450860619544983,
      "learning_rate": 0.00045861820054338625,
      "loss": 2.0255,
      "step": 74344
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.7620487213134766,
      "learning_rate": 0.0004586147284932888,
      "loss": 2.1439,
      "step": 74345
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.8202475905418396,
      "learning_rate": 0.0004586112564137019,
      "loss": 1.9922,
      "step": 74346
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.7995129227638245,
      "learning_rate": 0.0004586077843046261,
      "loss": 2.0469,
      "step": 74347
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.8554159998893738,
      "learning_rate": 0.0004586043121660622,
      "loss": 2.1862,
      "step": 74348
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.7730799913406372,
      "learning_rate": 0.0004586008399980107,
      "loss": 2.0973,
      "step": 74349
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.8592979907989502,
      "learning_rate": 0.00045859736780047225,
      "loss": 2.2944,
      "step": 74350
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.9735314846038818,
      "learning_rate": 0.0004585938955734477,
      "loss": 2.155,
      "step": 74351
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.8320541381835938,
      "learning_rate": 0.00045859042331693753,
      "loss": 2.2176,
      "step": 74352
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.7744375467300415,
      "learning_rate": 0.00045858695103094234,
      "loss": 2.2254,
      "step": 74353
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.0285135507583618,
      "learning_rate": 0.0004585834787154628,
      "loss": 2.2606,
      "step": 74354
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.8558119535446167,
      "learning_rate": 0.00045858000637049977,
      "loss": 2.1049,
      "step": 74355
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.8382801413536072,
      "learning_rate": 0.0004585765339960536,
      "loss": 2.1053,
      "step": 74356
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.3003287315368652,
      "learning_rate": 0.00045857306159212505,
      "loss": 2.2006,
      "step": 74357
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.931562602519989,
      "learning_rate": 0.00045856958915871474,
      "loss": 2.1824,
      "step": 74358
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.8895631432533264,
      "learning_rate": 0.00045856611669582346,
      "loss": 2.5423,
      "step": 74359
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.7771223783493042,
      "learning_rate": 0.0004585626442034517,
      "loss": 2.3256,
      "step": 74360
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.8970435857772827,
      "learning_rate": 0.0004585591716816001,
      "loss": 2.152,
      "step": 74361
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.8629925847053528,
      "learning_rate": 0.00045855569913026944,
      "loss": 2.2527,
      "step": 74362
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.8275150656700134,
      "learning_rate": 0.00045855222654946027,
      "loss": 2.1643,
      "step": 74363
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.7783620357513428,
      "learning_rate": 0.00045854875393917317,
      "loss": 2.0455,
      "step": 74364
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.8465513586997986,
      "learning_rate": 0.00045854528129940893,
      "loss": 2.2378,
      "step": 74365
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.0131633281707764,
      "learning_rate": 0.0004585418086301681,
      "loss": 2.2795,
      "step": 74366
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.8527767658233643,
      "learning_rate": 0.0004585383359314514,
      "loss": 2.243,
      "step": 74367
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.7436586022377014,
      "learning_rate": 0.0004585348632032594,
      "loss": 2.1263,
      "step": 74368
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.9295563697814941,
      "learning_rate": 0.00045853139044559277,
      "loss": 2.2184,
      "step": 74369
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.7798423767089844,
      "learning_rate": 0.0004585279176584522,
      "loss": 2.2813,
      "step": 74370
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.9112445116043091,
      "learning_rate": 0.00045852444484183825,
      "loss": 2.3471,
      "step": 74371
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.787184476852417,
      "learning_rate": 0.0004585209719957516,
      "loss": 2.189,
      "step": 74372
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.7350564002990723,
      "learning_rate": 0.00045851749912019296,
      "loss": 2.0686,
      "step": 74373
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.9553890824317932,
      "learning_rate": 0.0004585140262151629,
      "loss": 2.0891,
      "step": 74374
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.7281280159950256,
      "learning_rate": 0.0004585105532806621,
      "loss": 2.2734,
      "step": 74375
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.7863522171974182,
      "learning_rate": 0.00045850708031669116,
      "loss": 2.2255,
      "step": 74376
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.8279904723167419,
      "learning_rate": 0.00045850360732325085,
      "loss": 2.3313,
      "step": 74377
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.884345531463623,
      "learning_rate": 0.0004585001343003416,
      "loss": 2.2485,
      "step": 74378
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.6608074307441711,
      "learning_rate": 0.00045849666124796434,
      "loss": 2.0653,
      "step": 74379
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.8401838541030884,
      "learning_rate": 0.0004584931881661195,
      "loss": 2.3378,
      "step": 74380
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.9103103876113892,
      "learning_rate": 0.00045848971505480765,
      "loss": 2.2357,
      "step": 74381
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.770179033279419,
      "learning_rate": 0.00045848624191402974,
      "loss": 2.1435,
      "step": 74382
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.7604293823242188,
      "learning_rate": 0.0004584827687437862,
      "loss": 2.4619,
      "step": 74383
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.8433619141578674,
      "learning_rate": 0.0004584792955440777,
      "loss": 2.3037,
      "step": 74384
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.890049397945404,
      "learning_rate": 0.0004584758223149049,
      "loss": 2.19,
      "step": 74385
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.7765346765518188,
      "learning_rate": 0.00045847234905626845,
      "loss": 2.062,
      "step": 74386
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.7612088322639465,
      "learning_rate": 0.00045846887576816905,
      "loss": 2.0746,
      "step": 74387
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.8769478797912598,
      "learning_rate": 0.00045846540245060726,
      "loss": 2.0882,
      "step": 74388
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.8813724517822266,
      "learning_rate": 0.0004584619291035838,
      "loss": 2.3269,
      "step": 74389
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.8652392625808716,
      "learning_rate": 0.0004584584557270992,
      "loss": 2.2225,
      "step": 74390
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.8372641205787659,
      "learning_rate": 0.00045845498232115423,
      "loss": 2.2789,
      "step": 74391
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.8761958479881287,
      "learning_rate": 0.0004584515088857494,
      "loss": 2.1867,
      "step": 74392
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.1349046230316162,
      "learning_rate": 0.0004584480354208856,
      "loss": 2.1938,
      "step": 74393
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.7806916832923889,
      "learning_rate": 0.0004584445619265632,
      "loss": 2.2022,
      "step": 74394
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.8601844310760498,
      "learning_rate": 0.00045844108840278306,
      "loss": 1.944,
      "step": 74395
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.8070000410079956,
      "learning_rate": 0.00045843761484954564,
      "loss": 2.3015,
      "step": 74396
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.9296589493751526,
      "learning_rate": 0.0004584341412668517,
      "loss": 2.1753,
      "step": 74397
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.7825109362602234,
      "learning_rate": 0.0004584306676547019,
      "loss": 2.1804,
      "step": 74398
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.8258282542228699,
      "learning_rate": 0.0004584271940130968,
      "loss": 2.3135,
      "step": 74399
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.8123411536216736,
      "learning_rate": 0.00045842372034203706,
      "loss": 2.1128,
      "step": 74400
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.8413922190666199,
      "learning_rate": 0.0004584202466415235,
      "loss": 2.2699,
      "step": 74401
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.7832033634185791,
      "learning_rate": 0.00045841677291155654,
      "loss": 2.1278,
      "step": 74402
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.8918664455413818,
      "learning_rate": 0.00045841329915213684,
      "loss": 2.2817,
      "step": 74403
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.9187740683555603,
      "learning_rate": 0.0004584098253632652,
      "loss": 2.3625,
      "step": 74404
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.8233006596565247,
      "learning_rate": 0.0004584063515449422,
      "loss": 2.126,
      "step": 74405
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.7703844904899597,
      "learning_rate": 0.0004584028776971685,
      "loss": 2.1863,
      "step": 74406
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.7697013020515442,
      "learning_rate": 0.0004583994038199446,
      "loss": 2.2003,
      "step": 74407
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.761616051197052,
      "learning_rate": 0.0004583959299132714,
      "loss": 2.0059,
      "step": 74408
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.830549418926239,
      "learning_rate": 0.0004583924559771493,
      "loss": 2.2456,
      "step": 74409
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.8373361229896545,
      "learning_rate": 0.0004583889820115791,
      "loss": 2.3441,
      "step": 74410
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.8106829524040222,
      "learning_rate": 0.00045838550801656134,
      "loss": 2.3699,
      "step": 74411
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.8640817403793335,
      "learning_rate": 0.0004583820339920967,
      "loss": 2.3016,
      "step": 74412
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.9351057410240173,
      "learning_rate": 0.00045837855993818605,
      "loss": 2.3648,
      "step": 74413
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.7666497826576233,
      "learning_rate": 0.0004583750858548296,
      "loss": 2.0491,
      "step": 74414
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.8603149652481079,
      "learning_rate": 0.00045837161174202835,
      "loss": 2.1402,
      "step": 74415
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.0792760848999023,
      "learning_rate": 0.0004583681375997828,
      "loss": 2.2097,
      "step": 74416
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.7929701805114746,
      "learning_rate": 0.00045836466342809366,
      "loss": 2.2551,
      "step": 74417
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.9177027344703674,
      "learning_rate": 0.0004583611892269615,
      "loss": 2.2542,
      "step": 74418
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.8793400526046753,
      "learning_rate": 0.000458357714996387,
      "loss": 2.0836,
      "step": 74419
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.7251711487770081,
      "learning_rate": 0.0004583542407363709,
      "loss": 2.0018,
      "step": 74420
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.7391539812088013,
      "learning_rate": 0.0004583507664469136,
      "loss": 2.1746,
      "step": 74421
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.8482029438018799,
      "learning_rate": 0.00045834729212801607,
      "loss": 2.0807,
      "step": 74422
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.8927356600761414,
      "learning_rate": 0.0004583438177796786,
      "loss": 2.2465,
      "step": 74423
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.7986513376235962,
      "learning_rate": 0.00045834034340190224,
      "loss": 2.0814,
      "step": 74424
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.7761688232421875,
      "learning_rate": 0.0004583368689946873,
      "loss": 2.203,
      "step": 74425
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.7657400369644165,
      "learning_rate": 0.0004583333945580346,
      "loss": 2.0877,
      "step": 74426
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.7268270254135132,
      "learning_rate": 0.0004583299200919447,
      "loss": 2.2255,
      "step": 74427
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.7619820833206177,
      "learning_rate": 0.0004583264455964182,
      "loss": 2.2506,
      "step": 74428
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.787290632724762,
      "learning_rate": 0.0004583229710714559,
      "loss": 2.3262,
      "step": 74429
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.756883442401886,
      "learning_rate": 0.00045831949651705837,
      "loss": 2.1415,
      "step": 74430
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.6993659138679504,
      "learning_rate": 0.0004583160219332263,
      "loss": 2.542,
      "step": 74431
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.8360020518302917,
      "learning_rate": 0.0004583125473199602,
      "loss": 2.1387,
      "step": 74432
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.7124416828155518,
      "learning_rate": 0.0004583090726772609,
      "loss": 2.2995,
      "step": 74433
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.7868742942810059,
      "learning_rate": 0.0004583055980051289,
      "loss": 2.1153,
      "step": 74434
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.7684575319290161,
      "learning_rate": 0.00045830212330356495,
      "loss": 2.2971,
      "step": 74435
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.7572928667068481,
      "learning_rate": 0.0004582986485725696,
      "loss": 2.5785,
      "step": 74436
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.8471524715423584,
      "learning_rate": 0.00045829517381214355,
      "loss": 2.2365,
      "step": 74437
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.6776310205459595,
      "learning_rate": 0.0004582916990222874,
      "loss": 1.9258,
      "step": 74438
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.7436388731002808,
      "learning_rate": 0.0004582882242030019,
      "loss": 2.072,
      "step": 74439
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.786664605140686,
      "learning_rate": 0.00045828474935428754,
      "loss": 2.1804,
      "step": 74440
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.8629237413406372,
      "learning_rate": 0.00045828127447614526,
      "loss": 2.1603,
      "step": 74441
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.732913076877594,
      "learning_rate": 0.0004582777995685753,
      "loss": 2.1414,
      "step": 74442
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.850826621055603,
      "learning_rate": 0.0004582743246315786,
      "loss": 2.1613,
      "step": 74443
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.7561392188072205,
      "learning_rate": 0.00045827084966515577,
      "loss": 2.1891,
      "step": 74444
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.7416782379150391,
      "learning_rate": 0.0004582673746693073,
      "loss": 2.0929,
      "step": 74445
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.9438835978507996,
      "learning_rate": 0.00045826389964403395,
      "loss": 2.1482,
      "step": 74446
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.6898390054702759,
      "learning_rate": 0.00045826042458933634,
      "loss": 2.0935,
      "step": 74447
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.8010023832321167,
      "learning_rate": 0.0004582569495052152,
      "loss": 2.168,
      "step": 74448
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.9612265825271606,
      "learning_rate": 0.00045825347439167107,
      "loss": 2.2014,
      "step": 74449
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.8446063995361328,
      "learning_rate": 0.00045824999924870465,
      "loss": 2.3764,
      "step": 74450
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.8353022933006287,
      "learning_rate": 0.0004582465240763165,
      "loss": 2.3614,
      "step": 74451
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.9051447510719299,
      "learning_rate": 0.00045824304887450736,
      "loss": 2.0714,
      "step": 74452
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.8620356917381287,
      "learning_rate": 0.00045823957364327783,
      "loss": 2.3298,
      "step": 74453
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.7710203528404236,
      "learning_rate": 0.00045823609838262864,
      "loss": 2.1995,
      "step": 74454
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.8108920454978943,
      "learning_rate": 0.0004582326230925604,
      "loss": 2.0928,
      "step": 74455
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.9081425666809082,
      "learning_rate": 0.0004582291477730736,
      "loss": 2.372,
      "step": 74456
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.8131814002990723,
      "learning_rate": 0.0004582256724241691,
      "loss": 2.3634,
      "step": 74457
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.8292785882949829,
      "learning_rate": 0.0004582221970458475,
      "loss": 2.0614,
      "step": 74458
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.7707153558731079,
      "learning_rate": 0.0004582187216381094,
      "loss": 2.247,
      "step": 74459
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.8974789381027222,
      "learning_rate": 0.00045821524620095534,
      "loss": 2.1799,
      "step": 74460
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.8245538473129272,
      "learning_rate": 0.0004582117707343861,
      "loss": 2.1437,
      "step": 74461
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.8310532569885254,
      "learning_rate": 0.00045820829523840245,
      "loss": 2.0972,
      "step": 74462
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.7520933151245117,
      "learning_rate": 0.00045820481971300476,
      "loss": 2.272,
      "step": 74463
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.7157039642333984,
      "learning_rate": 0.0004582013441581938,
      "loss": 2.1679,
      "step": 74464
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.8666172623634338,
      "learning_rate": 0.0004581978685739703,
      "loss": 2.1852,
      "step": 74465
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.894957423210144,
      "learning_rate": 0.00045819439296033483,
      "loss": 2.1317,
      "step": 74466
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.7499314546585083,
      "learning_rate": 0.00045819091731728797,
      "loss": 2.1191,
      "step": 74467
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.9540002346038818,
      "learning_rate": 0.0004581874416448305,
      "loss": 2.093,
      "step": 74468
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.9302952289581299,
      "learning_rate": 0.00045818396594296296,
      "loss": 2.1167,
      "step": 74469
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.7212089896202087,
      "learning_rate": 0.00045818049021168606,
      "loss": 2.1389,
      "step": 74470
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.817743718624115,
      "learning_rate": 0.0004581770144510004,
      "loss": 2.3133,
      "step": 74471
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.9345203042030334,
      "learning_rate": 0.0004581735386609066,
      "loss": 2.2196,
      "step": 74472
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.9459280967712402,
      "learning_rate": 0.0004581700628414054,
      "loss": 1.9886,
      "step": 74473
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.8626216650009155,
      "learning_rate": 0.00045816658699249745,
      "loss": 2.4699,
      "step": 74474
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.8385159373283386,
      "learning_rate": 0.0004581631111141833,
      "loss": 2.2199,
      "step": 74475
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.759312093257904,
      "learning_rate": 0.00045815963520646363,
      "loss": 2.2244,
      "step": 74476
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.8357818126678467,
      "learning_rate": 0.00045815615926933913,
      "loss": 2.1061,
      "step": 74477
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.7557224035263062,
      "learning_rate": 0.0004581526833028104,
      "loss": 2.1796,
      "step": 74478
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.8743109107017517,
      "learning_rate": 0.0004581492073068781,
      "loss": 2.3658,
      "step": 74479
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.8669611215591431,
      "learning_rate": 0.0004581457312815429,
      "loss": 2.1187,
      "step": 74480
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.760403037071228,
      "learning_rate": 0.0004581422552268053,
      "loss": 2.2382,
      "step": 74481
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.7426232695579529,
      "learning_rate": 0.0004581387791426662,
      "loss": 2.05,
      "step": 74482
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.9397055506706238,
      "learning_rate": 0.00045813530302912613,
      "loss": 2.1266,
      "step": 74483
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.7803389430046082,
      "learning_rate": 0.0004581318268861856,
      "loss": 2.2079,
      "step": 74484
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.8491828441619873,
      "learning_rate": 0.00045812835071384547,
      "loss": 2.2732,
      "step": 74485
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.7756583094596863,
      "learning_rate": 0.00045812487451210636,
      "loss": 2.4908,
      "step": 74486
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.815632164478302,
      "learning_rate": 0.0004581213982809687,
      "loss": 2.2002,
      "step": 74487
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.8451787233352661,
      "learning_rate": 0.00045811792202043333,
      "loss": 2.0004,
      "step": 74488
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.7630190849304199,
      "learning_rate": 0.00045811444573050093,
      "loss": 1.9743,
      "step": 74489
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.8021194934844971,
      "learning_rate": 0.000458110969411172,
      "loss": 2.3032,
      "step": 74490
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.9747101068496704,
      "learning_rate": 0.00045810749306244723,
      "loss": 2.2406,
      "step": 74491
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.7685840129852295,
      "learning_rate": 0.0004581040166843274,
      "loss": 2.192,
      "step": 74492
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.8551020622253418,
      "learning_rate": 0.00045810054027681295,
      "loss": 2.383,
      "step": 74493
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.84499591588974,
      "learning_rate": 0.0004580970638399047,
      "loss": 2.3421,
      "step": 74494
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.8158170580863953,
      "learning_rate": 0.00045809358737360317,
      "loss": 2.1709,
      "step": 74495
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.7906907796859741,
      "learning_rate": 0.00045809011087790904,
      "loss": 2.3632,
      "step": 74496
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.7141324877738953,
      "learning_rate": 0.000458086634352823,
      "loss": 2.0355,
      "step": 74497
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.8165304064750671,
      "learning_rate": 0.00045808315779834576,
      "loss": 2.4748,
      "step": 74498
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.843926191329956,
      "learning_rate": 0.00045807968121447775,
      "loss": 2.2362,
      "step": 74499
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.8401919603347778,
      "learning_rate": 0.0004580762046012197,
      "loss": 2.289,
      "step": 74500
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.7332771420478821,
      "learning_rate": 0.00045807272795857245,
      "loss": 2.1725,
      "step": 74501
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.814138650894165,
      "learning_rate": 0.00045806925128653647,
      "loss": 2.2303,
      "step": 74502
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.9372482895851135,
      "learning_rate": 0.00045806577458511237,
      "loss": 2.0562,
      "step": 74503
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.769460141658783,
      "learning_rate": 0.00045806229785430087,
      "loss": 2.3471,
      "step": 74504
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.0325392484664917,
      "learning_rate": 0.00045805882109410255,
      "loss": 2.0615,
      "step": 74505
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.8076837658882141,
      "learning_rate": 0.0004580553443045182,
      "loss": 2.2466,
      "step": 74506
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.8901628255844116,
      "learning_rate": 0.0004580518674855484,
      "loss": 2.2661,
      "step": 74507
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.9376854300498962,
      "learning_rate": 0.0004580483906371937,
      "loss": 2.1729,
      "step": 74508
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.8315156698226929,
      "learning_rate": 0.0004580449137594549,
      "loss": 2.0575,
      "step": 74509
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.8080297708511353,
      "learning_rate": 0.0004580414368523325,
      "loss": 2.2116,
      "step": 74510
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.8645514249801636,
      "learning_rate": 0.0004580379599158272,
      "loss": 2.1853,
      "step": 74511
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.7751082181930542,
      "learning_rate": 0.00045803448294993976,
      "loss": 2.1363,
      "step": 74512
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.9346778392791748,
      "learning_rate": 0.0004580310059546706,
      "loss": 2.2814,
      "step": 74513
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.7864680290222168,
      "learning_rate": 0.00045802752893002056,
      "loss": 2.3848,
      "step": 74514
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.7811646461486816,
      "learning_rate": 0.00045802405187599023,
      "loss": 2.3403,
      "step": 74515
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.0357948541641235,
      "learning_rate": 0.0004580205747925802,
      "loss": 2.3049,
      "step": 74516
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.717775821685791,
      "learning_rate": 0.0004580170976797912,
      "loss": 2.1648,
      "step": 74517
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.7712622284889221,
      "learning_rate": 0.0004580136205376238,
      "loss": 2.2114,
      "step": 74518
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.9129180312156677,
      "learning_rate": 0.0004580101433660787,
      "loss": 2.3104,
      "step": 74519
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.8049378395080566,
      "learning_rate": 0.00045800666616515657,
      "loss": 2.1868,
      "step": 74520
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.8215335011482239,
      "learning_rate": 0.00045800318893485806,
      "loss": 2.3186,
      "step": 74521
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.7879771590232849,
      "learning_rate": 0.0004579997116751836,
      "loss": 2.1589,
      "step": 74522
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.8311188817024231,
      "learning_rate": 0.0004579962343861342,
      "loss": 2.3794,
      "step": 74523
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.7635234594345093,
      "learning_rate": 0.00045799275706771023,
      "loss": 2.206,
      "step": 74524
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.808459460735321,
      "learning_rate": 0.0004579892797199124,
      "loss": 2.3084,
      "step": 74525
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.7715334892272949,
      "learning_rate": 0.0004579858023427415,
      "loss": 2.1514,
      "step": 74526
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.7893439531326294,
      "learning_rate": 0.0004579823249361979,
      "loss": 2.3656,
      "step": 74527
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.7391664385795593,
      "learning_rate": 0.00045797884750028246,
      "loss": 2.2866,
      "step": 74528
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.7747847437858582,
      "learning_rate": 0.0004579753700349958,
      "loss": 2.2189,
      "step": 74529
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.7055772542953491,
      "learning_rate": 0.0004579718925403385,
      "loss": 2.0427,
      "step": 74530
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.7717550992965698,
      "learning_rate": 0.00045796841501631125,
      "loss": 2.0467,
      "step": 74531
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.9522518515586853,
      "learning_rate": 0.00045796493746291475,
      "loss": 2.2922,
      "step": 74532
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.7242735028266907,
      "learning_rate": 0.00045796145988014953,
      "loss": 2.3308,
      "step": 74533
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.8241133689880371,
      "learning_rate": 0.00045795798226801627,
      "loss": 2.2961,
      "step": 74534
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.7836624979972839,
      "learning_rate": 0.00045795450462651576,
      "loss": 2.3385,
      "step": 74535
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.9172700643539429,
      "learning_rate": 0.00045795102695564833,
      "loss": 2.1094,
      "step": 74536
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.7819082736968994,
      "learning_rate": 0.00045794754925541497,
      "loss": 2.3779,
      "step": 74537
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.7326692342758179,
      "learning_rate": 0.0004579440715258161,
      "loss": 2.1667,
      "step": 74538
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.7894703149795532,
      "learning_rate": 0.0004579405937668525,
      "loss": 2.2269,
      "step": 74539
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.7863996624946594,
      "learning_rate": 0.0004579371159785248,
      "loss": 2.1559,
      "step": 74540
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.8056021332740784,
      "learning_rate": 0.00045793363816083347,
      "loss": 2.4179,
      "step": 74541
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.7792989611625671,
      "learning_rate": 0.0004579301603137794,
      "loss": 2.1773,
      "step": 74542
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.7833540439605713,
      "learning_rate": 0.00045792668243736307,
      "loss": 2.194,
      "step": 74543
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.7900243401527405,
      "learning_rate": 0.0004579232045315853,
      "loss": 2.0817,
      "step": 74544
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.7430753111839294,
      "learning_rate": 0.00045791972659644647,
      "loss": 2.207,
      "step": 74545
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.8195350766181946,
      "learning_rate": 0.00045791624863194755,
      "loss": 2.3437,
      "step": 74546
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.8208619356155396,
      "learning_rate": 0.0004579127706380889,
      "loss": 2.3672,
      "step": 74547
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.7647112607955933,
      "learning_rate": 0.00045790929261487135,
      "loss": 2.184,
      "step": 74548
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.7778408527374268,
      "learning_rate": 0.0004579058145622954,
      "loss": 2.2713,
      "step": 74549
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.8171625733375549,
      "learning_rate": 0.00045790233648036185,
      "loss": 2.3612,
      "step": 74550
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.6894910335540771,
      "learning_rate": 0.0004578988583690713,
      "loss": 2.1963,
      "step": 74551
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.8230115175247192,
      "learning_rate": 0.00045789538022842426,
      "loss": 2.2505,
      "step": 74552
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.7836881875991821,
      "learning_rate": 0.0004578919020584216,
      "loss": 2.1749,
      "step": 74553
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.8120554089546204,
      "learning_rate": 0.0004578884238590638,
      "loss": 2.3159,
      "step": 74554
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.8109502792358398,
      "learning_rate": 0.00045788494563035157,
      "loss": 2.3611,
      "step": 74555
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.7658222317695618,
      "learning_rate": 0.00045788146737228556,
      "loss": 2.2948,
      "step": 74556
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.7678593397140503,
      "learning_rate": 0.0004578779890848664,
      "loss": 2.3579,
      "step": 74557
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.7757195830345154,
      "learning_rate": 0.00045787451076809474,
      "loss": 1.9799,
      "step": 74558
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.8102980852127075,
      "learning_rate": 0.0004578710324219712,
      "loss": 2.3745,
      "step": 74559
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.7652153372764587,
      "learning_rate": 0.0004578675540464965,
      "loss": 2.4069,
      "step": 74560
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.8950546979904175,
      "learning_rate": 0.00045786407564167125,
      "loss": 1.9766,
      "step": 74561
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.7346877455711365,
      "learning_rate": 0.000457860597207496,
      "loss": 2.2489,
      "step": 74562
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.9256382584571838,
      "learning_rate": 0.0004578571187439716,
      "loss": 2.1698,
      "step": 74563
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.871082067489624,
      "learning_rate": 0.00045785364025109854,
      "loss": 2.1868,
      "step": 74564
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.794570803642273,
      "learning_rate": 0.00045785016172887746,
      "loss": 2.3178,
      "step": 74565
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.8944856524467468,
      "learning_rate": 0.0004578466831773091,
      "loss": 2.1422,
      "step": 74566
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.7578518390655518,
      "learning_rate": 0.000457843204596394,
      "loss": 2.298,
      "step": 74567
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.7595393657684326,
      "learning_rate": 0.00045783972598613295,
      "loss": 2.1122,
      "step": 74568
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.8323574662208557,
      "learning_rate": 0.00045783624734652655,
      "loss": 2.1451,
      "step": 74569
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.7698861360549927,
      "learning_rate": 0.0004578327686775753,
      "loss": 2.2989,
      "step": 74570
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.7415741086006165,
      "learning_rate": 0.00045782928997928,
      "loss": 2.0906,
      "step": 74571
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.7627551555633545,
      "learning_rate": 0.00045782581125164125,
      "loss": 2.4096,
      "step": 74572
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.8017526268959045,
      "learning_rate": 0.0004578223324946597,
      "loss": 2.2185,
      "step": 74573
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.8333359360694885,
      "learning_rate": 0.00045781885370833604,
      "loss": 2.3257,
      "step": 74574
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.7963542342185974,
      "learning_rate": 0.0004578153748926709,
      "loss": 2.106,
      "step": 74575
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.8623607158660889,
      "learning_rate": 0.00045781189604766486,
      "loss": 2.3296,
      "step": 74576
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.9126401543617249,
      "learning_rate": 0.0004578084171733186,
      "loss": 2.2178,
      "step": 74577
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.839034914970398,
      "learning_rate": 0.0004578049382696328,
      "loss": 2.1046,
      "step": 74578
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.7689846158027649,
      "learning_rate": 0.000457801459336608,
      "loss": 2.2155,
      "step": 74579
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.7883470058441162,
      "learning_rate": 0.000457797980374245,
      "loss": 2.3987,
      "step": 74580
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.8228636980056763,
      "learning_rate": 0.0004577945013825444,
      "loss": 2.2866,
      "step": 74581
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.8336690664291382,
      "learning_rate": 0.0004577910223615068,
      "loss": 2.0688,
      "step": 74582
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.7703977227210999,
      "learning_rate": 0.0004577875433111328,
      "loss": 2.4251,
      "step": 74583
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.7911773920059204,
      "learning_rate": 0.00045778406423142333,
      "loss": 2.2195,
      "step": 74584
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.8105182647705078,
      "learning_rate": 0.0004577805851223786,
      "loss": 2.2192,
      "step": 74585
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.843991219997406,
      "learning_rate": 0.00045777710598399955,
      "loss": 2.0591,
      "step": 74586
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.74408358335495,
      "learning_rate": 0.0004577736268162868,
      "loss": 2.1142,
      "step": 74587
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.803006112575531,
      "learning_rate": 0.0004577701476192409,
      "loss": 2.2699,
      "step": 74588
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.8416121602058411,
      "learning_rate": 0.0004577666683928625,
      "loss": 2.4619,
      "step": 74589
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.7451393604278564,
      "learning_rate": 0.0004577631891371525,
      "loss": 2.1857,
      "step": 74590
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.7402238845825195,
      "learning_rate": 0.00045775970985211114,
      "loss": 2.1666,
      "step": 74591
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.0791113376617432,
      "learning_rate": 0.0004577562305377393,
      "loss": 2.2232,
      "step": 74592
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.7048757076263428,
      "learning_rate": 0.00045775275119403776,
      "loss": 2.0027,
      "step": 74593
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.8845129013061523,
      "learning_rate": 0.00045774927182100686,
      "loss": 2.2705,
      "step": 74594
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.6779967546463013,
      "learning_rate": 0.00045774579241864737,
      "loss": 2.1484,
      "step": 74595
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.921384871006012,
      "learning_rate": 0.00045774231298696005,
      "loss": 2.1701,
      "step": 74596
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.7969222664833069,
      "learning_rate": 0.00045773883352594544,
      "loss": 2.1675,
      "step": 74597
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.7204015851020813,
      "learning_rate": 0.0004577353540356042,
      "loss": 2.1711,
      "step": 74598
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.8369871377944946,
      "learning_rate": 0.000457731874515937,
      "loss": 2.1408,
      "step": 74599
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.7826676964759827,
      "learning_rate": 0.0004577283949669443,
      "loss": 2.0368,
      "step": 74600
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.7015482783317566,
      "learning_rate": 0.0004577249153886271,
      "loss": 2.0911,
      "step": 74601
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.8086870908737183,
      "learning_rate": 0.00045772143578098586,
      "loss": 2.074,
      "step": 74602
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.7624393701553345,
      "learning_rate": 0.0004577179561440211,
      "loss": 2.1666,
      "step": 74603
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.7362934947013855,
      "learning_rate": 0.0004577144764777337,
      "loss": 2.113,
      "step": 74604
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.8041395545005798,
      "learning_rate": 0.00045771099678212413,
      "loss": 2.2236,
      "step": 74605
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.8468707799911499,
      "learning_rate": 0.00045770751705719314,
      "loss": 2.2534,
      "step": 74606
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.7240598797798157,
      "learning_rate": 0.00045770403730294136,
      "loss": 2.1174,
      "step": 74607
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.7317937016487122,
      "learning_rate": 0.0004577005575193695,
      "loss": 2.1354,
      "step": 74608
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.7500512599945068,
      "learning_rate": 0.000457697077706478,
      "loss": 2.226,
      "step": 74609
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.7853767275810242,
      "learning_rate": 0.0004576935978642677,
      "loss": 2.2747,
      "step": 74610
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.8504509329795837,
      "learning_rate": 0.00045769011799273913,
      "loss": 2.2391,
      "step": 74611
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.7574602365493774,
      "learning_rate": 0.0004576866380918931,
      "loss": 2.2301,
      "step": 74612
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.8086495399475098,
      "learning_rate": 0.00045768315816173004,
      "loss": 2.1579,
      "step": 74613
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.7335127592086792,
      "learning_rate": 0.0004576796782022508,
      "loss": 2.2681,
      "step": 74614
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.7545271515846252,
      "learning_rate": 0.0004576761982134558,
      "loss": 2.112,
      "step": 74615
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.766724705696106,
      "learning_rate": 0.00045767271819534586,
      "loss": 2.2947,
      "step": 74616
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.7546126842498779,
      "learning_rate": 0.00045766923814792164,
      "loss": 1.9738,
      "step": 74617
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.7813991904258728,
      "learning_rate": 0.0004576657580711837,
      "loss": 2.2421,
      "step": 74618
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.7390264272689819,
      "learning_rate": 0.0004576622779651327,
      "loss": 2.1684,
      "step": 74619
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.8087549209594727,
      "learning_rate": 0.0004576587978297694,
      "loss": 2.2288,
      "step": 74620
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.7520129084587097,
      "learning_rate": 0.00045765531766509427,
      "loss": 2.1664,
      "step": 74621
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.8306406736373901,
      "learning_rate": 0.00045765183747110803,
      "loss": 2.11,
      "step": 74622
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.7954469323158264,
      "learning_rate": 0.0004576483572478113,
      "loss": 2.1662,
      "step": 74623
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.8265341520309448,
      "learning_rate": 0.0004576448769952049,
      "loss": 2.1001,
      "step": 74624
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.8146767616271973,
      "learning_rate": 0.0004576413967132893,
      "loss": 2.0557,
      "step": 74625
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.7400203943252563,
      "learning_rate": 0.00045763791640206507,
      "loss": 2.2204,
      "step": 74626
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.7376883625984192,
      "learning_rate": 0.0004576344360615332,
      "loss": 2.225,
      "step": 74627
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.7444893717765808,
      "learning_rate": 0.0004576309556916939,
      "loss": 2.2894,
      "step": 74628
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.7526605725288391,
      "learning_rate": 0.0004576274752925481,
      "loss": 1.9766,
      "step": 74629
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.0591745376586914,
      "learning_rate": 0.00045762399486409637,
      "loss": 2.2073,
      "step": 74630
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.8446276783943176,
      "learning_rate": 0.00045762051440633946,
      "loss": 2.1686,
      "step": 74631
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.9740315675735474,
      "learning_rate": 0.00045761703391927777,
      "loss": 2.0421,
      "step": 74632
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.7828416228294373,
      "learning_rate": 0.00045761355340291214,
      "loss": 2.3527,
      "step": 74633
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.958862841129303,
      "learning_rate": 0.00045761007285724314,
      "loss": 2.1292,
      "step": 74634
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.9137848019599915,
      "learning_rate": 0.00045760659228227156,
      "loss": 2.2594,
      "step": 74635
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.842600405216217,
      "learning_rate": 0.0004576031116779979,
      "loss": 2.2532,
      "step": 74636
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.856965184211731,
      "learning_rate": 0.0004575996310444228,
      "loss": 2.2073,
      "step": 74637
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.8391244411468506,
      "learning_rate": 0.0004575961503815471,
      "loss": 2.0144,
      "step": 74638
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.7495052218437195,
      "learning_rate": 0.00045759266968937117,
      "loss": 2.2086,
      "step": 74639
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.7857068777084351,
      "learning_rate": 0.0004575891889678958,
      "loss": 2.1808,
      "step": 74640
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.7757656574249268,
      "learning_rate": 0.00045758570821712155,
      "loss": 2.2317,
      "step": 74641
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.8126053214073181,
      "learning_rate": 0.0004575822274370493,
      "loss": 2.2744,
      "step": 74642
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.6898651123046875,
      "learning_rate": 0.0004575787466276795,
      "loss": 2.1215,
      "step": 74643
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.78501957654953,
      "learning_rate": 0.00045757526578901276,
      "loss": 2.0779,
      "step": 74644
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.8604222536087036,
      "learning_rate": 0.0004575717849210499,
      "loss": 2.2741,
      "step": 74645
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.1534006595611572,
      "learning_rate": 0.0004575683040237914,
      "loss": 2.1848,
      "step": 74646
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.8652051091194153,
      "learning_rate": 0.0004575648230972379,
      "loss": 2.0907,
      "step": 74647
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.8801360726356506,
      "learning_rate": 0.00045756134214139035,
      "loss": 2.2473,
      "step": 74648
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.9995653033256531,
      "learning_rate": 0.000457557861156249,
      "loss": 2.3023,
      "step": 74649
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.8414096236228943,
      "learning_rate": 0.0004575543801418147,
      "loss": 2.3661,
      "step": 74650
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.7153797149658203,
      "learning_rate": 0.0004575508990980881,
      "loss": 2.2169,
      "step": 74651
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.7701061367988586,
      "learning_rate": 0.0004575474180250698,
      "loss": 2.1365,
      "step": 74652
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.7632868885993958,
      "learning_rate": 0.0004575439369227604,
      "loss": 2.2888,
      "step": 74653
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.9715867042541504,
      "learning_rate": 0.00045754045579116077,
      "loss": 2.3333,
      "step": 74654
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.8477426171302795,
      "learning_rate": 0.0004575369746302712,
      "loss": 2.1697,
      "step": 74655
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.8084689974784851,
      "learning_rate": 0.00045753349344009264,
      "loss": 2.1638,
      "step": 74656
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.8243059515953064,
      "learning_rate": 0.00045753001222062565,
      "loss": 2.1768,
      "step": 74657
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.041318416595459,
      "learning_rate": 0.00045752653097187084,
      "loss": 2.1914,
      "step": 74658
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.7672119736671448,
      "learning_rate": 0.0004575230496938288,
      "loss": 2.1138,
      "step": 74659
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.888515830039978,
      "learning_rate": 0.00045751956838650034,
      "loss": 2.2019,
      "step": 74660
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.7237443327903748,
      "learning_rate": 0.000457516087049886,
      "loss": 2.038,
      "step": 74661
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.7807070016860962,
      "learning_rate": 0.0004575126056839864,
      "loss": 2.2113,
      "step": 74662
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.6942935585975647,
      "learning_rate": 0.0004575091242888023,
      "loss": 1.9711,
      "step": 74663
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.7815002202987671,
      "learning_rate": 0.00045750564286433417,
      "loss": 2.3842,
      "step": 74664
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.7241664528846741,
      "learning_rate": 0.00045750216141058286,
      "loss": 2.3042,
      "step": 74665
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.7197334170341492,
      "learning_rate": 0.00045749867992754897,
      "loss": 2.0044,
      "step": 74666
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.7812973856925964,
      "learning_rate": 0.000457495198415233,
      "loss": 2.4353,
      "step": 74667
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.8423702120780945,
      "learning_rate": 0.00045749171687363573,
      "loss": 2.3443,
      "step": 74668
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.7136768102645874,
      "learning_rate": 0.0004574882353027578,
      "loss": 2.1373,
      "step": 74669
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.8125203847885132,
      "learning_rate": 0.0004574847537025998,
      "loss": 2.3858,
      "step": 74670
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.7285664081573486,
      "learning_rate": 0.0004574812720731625,
      "loss": 2.3504,
      "step": 74671
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.8598998188972473,
      "learning_rate": 0.0004574777904144463,
      "loss": 2.1641,
      "step": 74672
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.7741333842277527,
      "learning_rate": 0.00045747430872645215,
      "loss": 2.196,
      "step": 74673
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.7434834241867065,
      "learning_rate": 0.00045747082700918045,
      "loss": 2.1445,
      "step": 74674
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.7980808615684509,
      "learning_rate": 0.00045746734526263205,
      "loss": 2.3483,
      "step": 74675
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.781349778175354,
      "learning_rate": 0.00045746386348680747,
      "loss": 2.1073,
      "step": 74676
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.7577760219573975,
      "learning_rate": 0.00045746038168170737,
      "loss": 2.1337,
      "step": 74677
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.8593142628669739,
      "learning_rate": 0.0004574568998473324,
      "loss": 2.2417,
      "step": 74678
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.7818456888198853,
      "learning_rate": 0.00045745341798368323,
      "loss": 2.1035,
      "step": 74679
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.8488531708717346,
      "learning_rate": 0.0004574499360907605,
      "loss": 2.1704,
      "step": 74680
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.8082473874092102,
      "learning_rate": 0.00045744645416856487,
      "loss": 2.1873,
      "step": 74681
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.8148426413536072,
      "learning_rate": 0.00045744297221709694,
      "loss": 2.2351,
      "step": 74682
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.8341089487075806,
      "learning_rate": 0.0004574394902363574,
      "loss": 2.3831,
      "step": 74683
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.7882538437843323,
      "learning_rate": 0.0004574360082263468,
      "loss": 2.3566,
      "step": 74684
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.7881891131401062,
      "learning_rate": 0.00045743252618706605,
      "loss": 2.1647,
      "step": 74685
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.7790598273277283,
      "learning_rate": 0.0004574290441185155,
      "loss": 2.3895,
      "step": 74686
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.877615213394165,
      "learning_rate": 0.00045742556202069595,
      "loss": 2.281,
      "step": 74687
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.7261937260627747,
      "learning_rate": 0.0004574220798936081,
      "loss": 2.0482,
      "step": 74688
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.7752172350883484,
      "learning_rate": 0.00045741859773725234,
      "loss": 2.2838,
      "step": 74689
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.797710657119751,
      "learning_rate": 0.00045741511555162965,
      "loss": 2.1796,
      "step": 74690
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.8656375408172607,
      "learning_rate": 0.0004574116333367404,
      "loss": 2.3496,
      "step": 74691
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.8887731432914734,
      "learning_rate": 0.00045740815109258535,
      "loss": 2.3804,
      "step": 74692
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.7531155943870544,
      "learning_rate": 0.0004574046688191654,
      "loss": 2.0196,
      "step": 74693
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.7879717946052551,
      "learning_rate": 0.00045740118651648067,
      "loss": 2.2631,
      "step": 74694
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.81807941198349,
      "learning_rate": 0.0004573977041845322,
      "loss": 2.0484,
      "step": 74695
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.8295378088951111,
      "learning_rate": 0.00045739422182332053,
      "loss": 2.4377,
      "step": 74696
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.8601018190383911,
      "learning_rate": 0.00045739073943284623,
      "loss": 2.0519,
      "step": 74697
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.7965062856674194,
      "learning_rate": 0.00045738725701311003,
      "loss": 2.1156,
      "step": 74698
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.7418482899665833,
      "learning_rate": 0.00045738377456411267,
      "loss": 2.0104,
      "step": 74699
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.9040155410766602,
      "learning_rate": 0.00045738029208585465,
      "loss": 2.3088,
      "step": 74700
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.773094654083252,
      "learning_rate": 0.0004573768095783366,
      "loss": 2.349,
      "step": 74701
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.7553842067718506,
      "learning_rate": 0.00045737332704155927,
      "loss": 2.261,
      "step": 74702
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.8989914655685425,
      "learning_rate": 0.0004573698444755233,
      "loss": 1.9053,
      "step": 74703
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.7325055003166199,
      "learning_rate": 0.00045736636188022924,
      "loss": 2.0897,
      "step": 74704
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.7287653088569641,
      "learning_rate": 0.0004573628792556778,
      "loss": 1.961,
      "step": 74705
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.7787123918533325,
      "learning_rate": 0.00045735939660186984,
      "loss": 2.1893,
      "step": 74706
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.7726361751556396,
      "learning_rate": 0.00045735591391880554,
      "loss": 2.231,
      "step": 74707
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.7914659976959229,
      "learning_rate": 0.0004573524312064859,
      "loss": 2.3761,
      "step": 74708
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.7356697916984558,
      "learning_rate": 0.00045734894846491154,
      "loss": 2.1078,
      "step": 74709
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.8277146816253662,
      "learning_rate": 0.00045734546569408286,
      "loss": 2.0162,
      "step": 74710
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.9040343165397644,
      "learning_rate": 0.00045734198289400083,
      "loss": 2.2966,
      "step": 74711
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.8753200769424438,
      "learning_rate": 0.000457338500064666,
      "loss": 2.4025,
      "step": 74712
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.9731337428092957,
      "learning_rate": 0.00045733501720607886,
      "loss": 2.1187,
      "step": 74713
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.9230556488037109,
      "learning_rate": 0.0004573315343182402,
      "loss": 2.1717,
      "step": 74714
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.7633126378059387,
      "learning_rate": 0.0004573280514011506,
      "loss": 2.2396,
      "step": 74715
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.7504720687866211,
      "learning_rate": 0.0004573245684548108,
      "loss": 2.2088,
      "step": 74716
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.869778573513031,
      "learning_rate": 0.0004573210854792214,
      "loss": 2.047,
      "step": 74717
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.8223505616188049,
      "learning_rate": 0.000457317602474383,
      "loss": 2.3419,
      "step": 74718
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.8844043016433716,
      "learning_rate": 0.00045731411944029636,
      "loss": 2.1854,
      "step": 74719
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.763660192489624,
      "learning_rate": 0.0004573106363769619,
      "loss": 2.3526,
      "step": 74720
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.8707869648933411,
      "learning_rate": 0.0004573071532843806,
      "loss": 2.1999,
      "step": 74721
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.0652776956558228,
      "learning_rate": 0.00045730367016255283,
      "loss": 1.9306,
      "step": 74722
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.7478567361831665,
      "learning_rate": 0.00045730018701147944,
      "loss": 2.1156,
      "step": 74723
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.7621809244155884,
      "learning_rate": 0.0004572967038311609,
      "loss": 2.2648,
      "step": 74724
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.1409107446670532,
      "learning_rate": 0.0004572932206215978,
      "loss": 2.2086,
      "step": 74725
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.8640302419662476,
      "learning_rate": 0.00045728973738279116,
      "loss": 2.1379,
      "step": 74726
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.766878068447113,
      "learning_rate": 0.00045728625411474124,
      "loss": 2.2609,
      "step": 74727
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.8123118281364441,
      "learning_rate": 0.0004572827708174489,
      "loss": 2.0746,
      "step": 74728
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.8763864040374756,
      "learning_rate": 0.00045727928749091475,
      "loss": 2.3582,
      "step": 74729
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.7169472575187683,
      "learning_rate": 0.0004572758041351393,
      "loss": 2.0964,
      "step": 74730
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.7009767293930054,
      "learning_rate": 0.00045727232075012345,
      "loss": 2.1244,
      "step": 74731
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.8644737601280212,
      "learning_rate": 0.00045726883733586757,
      "loss": 2.0654,
      "step": 74732
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.8228548169136047,
      "learning_rate": 0.0004572653538923726,
      "loss": 2.2044,
      "step": 74733
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.752683699131012,
      "learning_rate": 0.00045726187041963886,
      "loss": 2.1622,
      "step": 74734
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.870916485786438,
      "learning_rate": 0.00045725838691766733,
      "loss": 2.4436,
      "step": 74735
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.9196515083312988,
      "learning_rate": 0.00045725490338645845,
      "loss": 2.2866,
      "step": 74736
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.7714473605155945,
      "learning_rate": 0.0004572514198260128,
      "loss": 2.2327,
      "step": 74737
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.6567095518112183,
      "learning_rate": 0.0004572479362363313,
      "loss": 2.0848,
      "step": 74738
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.7492710947990417,
      "learning_rate": 0.00045724445261741435,
      "loss": 1.9732,
      "step": 74739
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.9162429571151733,
      "learning_rate": 0.00045724096896926266,
      "loss": 2.1972,
      "step": 74740
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.7729508876800537,
      "learning_rate": 0.000457237485291877,
      "loss": 2.1481,
      "step": 74741
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.8146417737007141,
      "learning_rate": 0.00045723400158525785,
      "loss": 2.2335,
      "step": 74742
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.7960929274559021,
      "learning_rate": 0.0004572305178494059,
      "loss": 2.2688,
      "step": 74743
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.7535642981529236,
      "learning_rate": 0.0004572270340843219,
      "loss": 2.3054,
      "step": 74744
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.7346253991127014,
      "learning_rate": 0.00045722355029000647,
      "loss": 2.0153,
      "step": 74745
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.7026469111442566,
      "learning_rate": 0.00045722006646646015,
      "loss": 2.2383,
      "step": 74746
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.8751468062400818,
      "learning_rate": 0.0004572165826136836,
      "loss": 2.2126,
      "step": 74747
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.7537088990211487,
      "learning_rate": 0.00045721309873167766,
      "loss": 2.0673,
      "step": 74748
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.7584232091903687,
      "learning_rate": 0.0004572096148204427,
      "loss": 2.2622,
      "step": 74749
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.7785065174102783,
      "learning_rate": 0.0004572061308799796,
      "loss": 2.3667,
      "step": 74750
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.7252416014671326,
      "learning_rate": 0.00045720264691028893,
      "loss": 2.368,
      "step": 74751
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.7890951037406921,
      "learning_rate": 0.0004571991629113712,
      "loss": 2.1699,
      "step": 74752
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.8057978749275208,
      "learning_rate": 0.0004571956788832272,
      "loss": 2.3842,
      "step": 74753
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.8109760880470276,
      "learning_rate": 0.00045719219482585763,
      "loss": 2.3623,
      "step": 74754
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.7503268718719482,
      "learning_rate": 0.00045718871073926304,
      "loss": 2.0691,
      "step": 74755
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.7941396832466125,
      "learning_rate": 0.00045718522662344406,
      "loss": 2.2684,
      "step": 74756
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.8556254506111145,
      "learning_rate": 0.0004571817424784015,
      "loss": 2.1808,
      "step": 74757
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.7653657793998718,
      "learning_rate": 0.00045717825830413576,
      "loss": 2.0508,
      "step": 74758
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.7930416464805603,
      "learning_rate": 0.00045717477410064765,
      "loss": 2.4046,
      "step": 74759
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.7499001622200012,
      "learning_rate": 0.00045717128986793784,
      "loss": 2.2584,
      "step": 74760
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.8139638900756836,
      "learning_rate": 0.00045716780560600684,
      "loss": 2.2476,
      "step": 74761
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.8209646940231323,
      "learning_rate": 0.0004571643213148554,
      "loss": 2.2028,
      "step": 74762
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.8858340382575989,
      "learning_rate": 0.00045716083699448416,
      "loss": 2.1675,
      "step": 74763
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.8143507242202759,
      "learning_rate": 0.0004571573526448938,
      "loss": 2.2893,
      "step": 74764
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.7647221684455872,
      "learning_rate": 0.0004571538682660848,
      "loss": 2.2918,
      "step": 74765
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.7149155139923096,
      "learning_rate": 0.0004571503838580579,
      "loss": 2.0605,
      "step": 74766
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.8626286387443542,
      "learning_rate": 0.000457146899420814,
      "loss": 2.426,
      "step": 74767
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.9290565848350525,
      "learning_rate": 0.00045714341495435335,
      "loss": 2.2037,
      "step": 74768
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.7875704765319824,
      "learning_rate": 0.0004571399304586768,
      "loss": 2.2426,
      "step": 74769
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.8747330904006958,
      "learning_rate": 0.00045713644593378504,
      "loss": 2.2126,
      "step": 74770
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.9144232869148254,
      "learning_rate": 0.00045713296137967857,
      "loss": 2.0512,
      "step": 74771
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.8054470419883728,
      "learning_rate": 0.00045712947679635817,
      "loss": 2.0583,
      "step": 74772
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.7445554733276367,
      "learning_rate": 0.0004571259921838244,
      "loss": 2.2661,
      "step": 74773
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.8461171388626099,
      "learning_rate": 0.00045712250754207794,
      "loss": 2.2177,
      "step": 74774
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.8884108662605286,
      "learning_rate": 0.0004571190228711194,
      "loss": 2.3943,
      "step": 74775
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.6987476348876953,
      "learning_rate": 0.0004571155381709496,
      "loss": 2.2891,
      "step": 74776
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.8189932703971863,
      "learning_rate": 0.00045711205344156896,
      "loss": 2.0674,
      "step": 74777
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.8285001516342163,
      "learning_rate": 0.0004571085686829782,
      "loss": 2.1589,
      "step": 74778
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.8562822341918945,
      "learning_rate": 0.0004571050838951781,
      "loss": 2.1126,
      "step": 74779
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.7748062014579773,
      "learning_rate": 0.00045710159907816914,
      "loss": 2.06,
      "step": 74780
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.9912622570991516,
      "learning_rate": 0.000457098114231952,
      "loss": 2.2134,
      "step": 74781
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.0355063676834106,
      "learning_rate": 0.00045709462935652735,
      "loss": 2.2927,
      "step": 74782
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.9877257943153381,
      "learning_rate": 0.00045709114445189583,
      "loss": 2.3355,
      "step": 74783
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.8640091419219971,
      "learning_rate": 0.0004570876595180582,
      "loss": 2.2921,
      "step": 74784
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.7595232725143433,
      "learning_rate": 0.000457084174555015,
      "loss": 2.1128,
      "step": 74785
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.7089712619781494,
      "learning_rate": 0.00045708068956276676,
      "loss": 2.3651,
      "step": 74786
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.7914038896560669,
      "learning_rate": 0.0004570772045413143,
      "loss": 2.2771,
      "step": 74787
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.8080762028694153,
      "learning_rate": 0.0004570737194906583,
      "loss": 2.3124,
      "step": 74788
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.7042227983474731,
      "learning_rate": 0.0004570702344107992,
      "loss": 2.318,
      "step": 74789
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.7032032012939453,
      "learning_rate": 0.0004570667493017379,
      "loss": 2.4431,
      "step": 74790
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.7473818063735962,
      "learning_rate": 0.00045706326416347496,
      "loss": 2.1551,
      "step": 74791
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.7767637968063354,
      "learning_rate": 0.0004570597789960109,
      "loss": 2.1169,
      "step": 74792
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.8286925554275513,
      "learning_rate": 0.0004570562937993464,
      "loss": 2.327,
      "step": 74793
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.7912430167198181,
      "learning_rate": 0.00045705280857348237,
      "loss": 2.553,
      "step": 74794
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.760524332523346,
      "learning_rate": 0.0004570493233184191,
      "loss": 2.2044,
      "step": 74795
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.76579350233078,
      "learning_rate": 0.0004570458380341574,
      "loss": 2.1124,
      "step": 74796
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.8099415302276611,
      "learning_rate": 0.000457042352720698,
      "loss": 2.2452,
      "step": 74797
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.8095232844352722,
      "learning_rate": 0.00045703886737804135,
      "loss": 2.1064,
      "step": 74798
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.7642560005187988,
      "learning_rate": 0.0004570353820061883,
      "loss": 2.1164,
      "step": 74799
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.8295825719833374,
      "learning_rate": 0.0004570318966051394,
      "loss": 2.0951,
      "step": 74800
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.8354520201683044,
      "learning_rate": 0.00045702841117489524,
      "loss": 2.2224,
      "step": 74801
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.8114158511161804,
      "learning_rate": 0.0004570249257154566,
      "loss": 2.2418,
      "step": 74802
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.8151683211326599,
      "learning_rate": 0.00045702144022682407,
      "loss": 2.2743,
      "step": 74803
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.7969086766242981,
      "learning_rate": 0.0004570179547089982,
      "loss": 2.1499,
      "step": 74804
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.8019171357154846,
      "learning_rate": 0.0004570144691619799,
      "loss": 2.2836,
      "step": 74805
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.9330817461013794,
      "learning_rate": 0.0004570109835857694,
      "loss": 2.0246,
      "step": 74806
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.8327160477638245,
      "learning_rate": 0.0004570074979803678,
      "loss": 2.1627,
      "step": 74807
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.824516773223877,
      "learning_rate": 0.0004570040123457755,
      "loss": 2.0389,
      "step": 74808
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.7843321561813354,
      "learning_rate": 0.00045700052668199315,
      "loss": 2.061,
      "step": 74809
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.851986289024353,
      "learning_rate": 0.0004569970409890215,
      "loss": 2.1894,
      "step": 74810
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.7979575395584106,
      "learning_rate": 0.0004569935552668611,
      "loss": 2.2055,
      "step": 74811
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.7862012982368469,
      "learning_rate": 0.0004569900695155126,
      "loss": 2.5158,
      "step": 74812
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.876663863658905,
      "learning_rate": 0.00045698658373497663,
      "loss": 2.2424,
      "step": 74813
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.7779210805892944,
      "learning_rate": 0.000456983097925254,
      "loss": 2.3761,
      "step": 74814
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.8064411878585815,
      "learning_rate": 0.0004569796120863452,
      "loss": 2.1879,
      "step": 74815
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.9065929055213928,
      "learning_rate": 0.000456976126218251,
      "loss": 2.4363,
      "step": 74816
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.7920675277709961,
      "learning_rate": 0.0004569726403209719,
      "loss": 2.4408,
      "step": 74817
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.7994293570518494,
      "learning_rate": 0.0004569691543945087,
      "loss": 2.1914,
      "step": 74818
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.9257487654685974,
      "learning_rate": 0.00045696566843886187,
      "loss": 2.6574,
      "step": 74819
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.8062866926193237,
      "learning_rate": 0.0004569621824540322,
      "loss": 2.1651,
      "step": 74820
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.7544698119163513,
      "learning_rate": 0.0004569586964400203,
      "loss": 2.1712,
      "step": 74821
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.8551318049430847,
      "learning_rate": 0.0004569552103968268,
      "loss": 2.1491,
      "step": 74822
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.9430073499679565,
      "learning_rate": 0.00045695172432445234,
      "loss": 2.474,
      "step": 74823
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.8137345314025879,
      "learning_rate": 0.0004569482382228977,
      "loss": 2.1914,
      "step": 74824
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.8247371315956116,
      "learning_rate": 0.0004569447520921634,
      "loss": 2.5345,
      "step": 74825
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.126023530960083,
      "learning_rate": 0.00045694126593224995,
      "loss": 2.1451,
      "step": 74826
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.0021014213562012,
      "learning_rate": 0.00045693777974315826,
      "loss": 1.8031,
      "step": 74827
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.7705050706863403,
      "learning_rate": 0.0004569342935248889,
      "loss": 2.4045,
      "step": 74828
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.0258238315582275,
      "learning_rate": 0.0004569308072774425,
      "loss": 2.1006,
      "step": 74829
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.9204255938529968,
      "learning_rate": 0.0004569273210008197,
      "loss": 2.049,
      "step": 74830
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.9057199954986572,
      "learning_rate": 0.00045692383469502114,
      "loss": 2.231,
      "step": 74831
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.7869703769683838,
      "learning_rate": 0.0004569203483600474,
      "loss": 2.2633,
      "step": 74832
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.1429985761642456,
      "learning_rate": 0.0004569168619958992,
      "loss": 2.2639,
      "step": 74833
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.9577412605285645,
      "learning_rate": 0.0004569133756025774,
      "loss": 2.422,
      "step": 74834
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.8758426308631897,
      "learning_rate": 0.00045690988918008224,
      "loss": 2.2926,
      "step": 74835
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.755427360534668,
      "learning_rate": 0.0004569064027284146,
      "loss": 2.2458,
      "step": 74836
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.1911553144454956,
      "learning_rate": 0.0004569029162475752,
      "loss": 2.1542,
      "step": 74837
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.8748302459716797,
      "learning_rate": 0.00045689942973756447,
      "loss": 2.2935,
      "step": 74838
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.7697862386703491,
      "learning_rate": 0.00045689594319838324,
      "loss": 2.3244,
      "step": 74839
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.9799453616142273,
      "learning_rate": 0.00045689245663003205,
      "loss": 2.1346,
      "step": 74840
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.8798233866691589,
      "learning_rate": 0.0004568889700325116,
      "loss": 2.3167,
      "step": 74841
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.7547428011894226,
      "learning_rate": 0.00045688548340582245,
      "loss": 2.3193,
      "step": 74842
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.8334596157073975,
      "learning_rate": 0.0004568819967499655,
      "loss": 2.3523,
      "step": 74843
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.9943914413452148,
      "learning_rate": 0.00045687851006494116,
      "loss": 2.4464,
      "step": 74844
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.763014554977417,
      "learning_rate": 0.0004568750233507501,
      "loss": 2.0876,
      "step": 74845
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.7877135276794434,
      "learning_rate": 0.00045687153660739304,
      "loss": 2.3489,
      "step": 74846
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.7497513294219971,
      "learning_rate": 0.00045686804983487064,
      "loss": 2.2143,
      "step": 74847
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.8469206094741821,
      "learning_rate": 0.00045686456303318347,
      "loss": 2.2711,
      "step": 74848
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.7763665318489075,
      "learning_rate": 0.0004568610762023322,
      "loss": 2.0439,
      "step": 74849
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.9237961769104004,
      "learning_rate": 0.0004568575893423175,
      "loss": 2.1882,
      "step": 74850
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.7921302914619446,
      "learning_rate": 0.00045685410245314,
      "loss": 2.4061,
      "step": 74851
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.8808572292327881,
      "learning_rate": 0.0004568506155348004,
      "loss": 2.1886,
      "step": 74852
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.8979509472846985,
      "learning_rate": 0.00045684712858729925,
      "loss": 2.0613,
      "step": 74853
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.8201817274093628,
      "learning_rate": 0.0004568436416106373,
      "loss": 2.1378,
      "step": 74854
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.8175500631332397,
      "learning_rate": 0.00045684015460481525,
      "loss": 1.9665,
      "step": 74855
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.9144274592399597,
      "learning_rate": 0.0004568366675698335,
      "loss": 2.1115,
      "step": 74856
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.8849504590034485,
      "learning_rate": 0.0004568331805056929,
      "loss": 2.2145,
      "step": 74857
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.7091628909111023,
      "learning_rate": 0.0004568296934123941,
      "loss": 2.1146,
      "step": 74858
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.8565212488174438,
      "learning_rate": 0.0004568262062899376,
      "loss": 2.1078,
      "step": 74859
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.9968287348747253,
      "learning_rate": 0.00045682271913832426,
      "loss": 2.0761,
      "step": 74860
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.9972066879272461,
      "learning_rate": 0.00045681923195755464,
      "loss": 2.094,
      "step": 74861
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.7467015981674194,
      "learning_rate": 0.0004568157447476293,
      "loss": 2.3446,
      "step": 74862
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.0319249629974365,
      "learning_rate": 0.00045681225750854884,
      "loss": 2.2743,
      "step": 74863
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.1283316612243652,
      "learning_rate": 0.0004568087702403142,
      "loss": 2.1454,
      "step": 74864
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.8402917981147766,
      "learning_rate": 0.0004568052829429258,
      "loss": 2.1677,
      "step": 74865
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.8113662004470825,
      "learning_rate": 0.00045680179561638425,
      "loss": 2.2877,
      "step": 74866
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.034893274307251,
      "learning_rate": 0.0004567983082606904,
      "loss": 2.1386,
      "step": 74867
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.011181354522705,
      "learning_rate": 0.00045679482087584473,
      "loss": 2.2202,
      "step": 74868
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.8266322612762451,
      "learning_rate": 0.00045679133346184794,
      "loss": 2.4123,
      "step": 74869
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.7793948650360107,
      "learning_rate": 0.00045678784601870074,
      "loss": 2.328,
      "step": 74870
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.7935599684715271,
      "learning_rate": 0.0004567843585464036,
      "loss": 2.0398,
      "step": 74871
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.9543087482452393,
      "learning_rate": 0.00045678087104495735,
      "loss": 2.3083,
      "step": 74872
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.9976864457130432,
      "learning_rate": 0.00045677738351436264,
      "loss": 2.4588,
      "step": 74873
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.7962535619735718,
      "learning_rate": 0.00045677389595462,
      "loss": 2.2735,
      "step": 74874
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.1575545072555542,
      "learning_rate": 0.0004567704083657301,
      "loss": 2.2412,
      "step": 74875
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.9498449563980103,
      "learning_rate": 0.00045676692074769374,
      "loss": 2.2166,
      "step": 74876
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.6925273537635803,
      "learning_rate": 0.0004567634331005113,
      "loss": 1.95,
      "step": 74877
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.9247404336929321,
      "learning_rate": 0.0004567599454241836,
      "loss": 2.0011,
      "step": 74878
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.7761214971542358,
      "learning_rate": 0.0004567564577187114,
      "loss": 2.3175,
      "step": 74879
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.712364912033081,
      "learning_rate": 0.0004567529699840951,
      "loss": 2.2427,
      "step": 74880
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.7817527055740356,
      "learning_rate": 0.0004567494822203355,
      "loss": 2.1947,
      "step": 74881
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.7899331450462341,
      "learning_rate": 0.00045674599442743324,
      "loss": 2.159,
      "step": 74882
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.8330462574958801,
      "learning_rate": 0.0004567425066053889,
      "loss": 2.2692,
      "step": 74883
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.7988288998603821,
      "learning_rate": 0.0004567390187542031,
      "loss": 2.3883,
      "step": 74884
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.9116020202636719,
      "learning_rate": 0.0004567355308738767,
      "loss": 2.4809,
      "step": 74885
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.7337115406990051,
      "learning_rate": 0.00045673204296441014,
      "loss": 2.1542,
      "step": 74886
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.7500531673431396,
      "learning_rate": 0.00045672855502580407,
      "loss": 2.1517,
      "step": 74887
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.8260349035263062,
      "learning_rate": 0.0004567250670580593,
      "loss": 2.4895,
      "step": 74888
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.7622726559638977,
      "learning_rate": 0.0004567215790611764,
      "loss": 2.2839,
      "step": 74889
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.7776961326599121,
      "learning_rate": 0.00045671809103515587,
      "loss": 2.2921,
      "step": 74890
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.7517197132110596,
      "learning_rate": 0.00045671460297999855,
      "loss": 2.1744,
      "step": 74891
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.7367721796035767,
      "learning_rate": 0.00045671111489570504,
      "loss": 2.2357,
      "step": 74892
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.8222343921661377,
      "learning_rate": 0.00045670762678227603,
      "loss": 2.2122,
      "step": 74893
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.7497379183769226,
      "learning_rate": 0.000456704138639712,
      "loss": 1.8911,
      "step": 74894
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.8131728172302246,
      "learning_rate": 0.00045670065046801375,
      "loss": 2.123,
      "step": 74895
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.8147886991500854,
      "learning_rate": 0.00045669716226718195,
      "loss": 2.396,
      "step": 74896
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.9023866653442383,
      "learning_rate": 0.0004566936740372172,
      "loss": 2.2906,
      "step": 74897
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.7627364993095398,
      "learning_rate": 0.00045669018577812004,
      "loss": 2.1688,
      "step": 74898
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.8136943578720093,
      "learning_rate": 0.0004566866974898913,
      "loss": 2.2752,
      "step": 74899
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.8245350122451782,
      "learning_rate": 0.00045668320917253145,
      "loss": 2.1567,
      "step": 74900
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.85377436876297,
      "learning_rate": 0.00045667972082604136,
      "loss": 2.1868,
      "step": 74901
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.7596610188484192,
      "learning_rate": 0.0004566762324504214,
      "loss": 2.1627,
      "step": 74902
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.7944461703300476,
      "learning_rate": 0.0004566727440456726,
      "loss": 2.1435,
      "step": 74903
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.7806872725486755,
      "learning_rate": 0.00045666925561179517,
      "loss": 2.2516,
      "step": 74904
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.747348427772522,
      "learning_rate": 0.00045666576714879,
      "loss": 2.1663,
      "step": 74905
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.7809354662895203,
      "learning_rate": 0.0004566622786566578,
      "loss": 2.1684,
      "step": 74906
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.7565858364105225,
      "learning_rate": 0.00045665879013539914,
      "loss": 1.9806,
      "step": 74907
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.7749305963516235,
      "learning_rate": 0.00045665530158501454,
      "loss": 2.2619,
      "step": 74908
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.775155782699585,
      "learning_rate": 0.00045665181300550474,
      "loss": 2.169,
      "step": 74909
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.9114474654197693,
      "learning_rate": 0.0004566483243968705,
      "loss": 2.0136,
      "step": 74910
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.8309794664382935,
      "learning_rate": 0.0004566448357591124,
      "loss": 2.0996,
      "step": 74911
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.8609195947647095,
      "learning_rate": 0.00045664134709223097,
      "loss": 2.0845,
      "step": 74912
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.802934467792511,
      "learning_rate": 0.00045663785839622706,
      "loss": 2.1342,
      "step": 74913
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.8868705034255981,
      "learning_rate": 0.00045663436967110116,
      "loss": 2.2205,
      "step": 74914
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.8416674137115479,
      "learning_rate": 0.00045663088091685397,
      "loss": 2.1363,
      "step": 74915
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.7145642638206482,
      "learning_rate": 0.00045662739213348614,
      "loss": 2.0281,
      "step": 74916
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.829670250415802,
      "learning_rate": 0.0004566239033209984,
      "loss": 2.2135,
      "step": 74917
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.7520136833190918,
      "learning_rate": 0.0004566204144793912,
      "loss": 2.0187,
      "step": 74918
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.7288049459457397,
      "learning_rate": 0.00045661692560866547,
      "loss": 2.3884,
      "step": 74919
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.7579700350761414,
      "learning_rate": 0.0004566134367088215,
      "loss": 2.3448,
      "step": 74920
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.8224413990974426,
      "learning_rate": 0.0004566099477798603,
      "loss": 2.3816,
      "step": 74921
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.7120500802993774,
      "learning_rate": 0.00045660645882178224,
      "loss": 2.1834,
      "step": 74922
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.7689464688301086,
      "learning_rate": 0.00045660296983458814,
      "loss": 2.1272,
      "step": 74923
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.71124267578125,
      "learning_rate": 0.00045659948081827855,
      "loss": 2.1808,
      "step": 74924
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.8480238318443298,
      "learning_rate": 0.00045659599177285423,
      "loss": 2.0215,
      "step": 74925
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.9295278191566467,
      "learning_rate": 0.00045659250269831567,
      "loss": 2.3368,
      "step": 74926
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.7881435751914978,
      "learning_rate": 0.0004565890135946636,
      "loss": 2.1949,
      "step": 74927
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.8841837048530579,
      "learning_rate": 0.00045658552446189884,
      "loss": 2.1504,
      "step": 74928
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.7162224650382996,
      "learning_rate": 0.0004565820353000218,
      "loss": 2.0547,
      "step": 74929
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.7695367336273193,
      "learning_rate": 0.00045657854610903307,
      "loss": 2.2506,
      "step": 74930
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.7717894315719604,
      "learning_rate": 0.0004565750568889336,
      "loss": 2.2258,
      "step": 74931
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.8176413774490356,
      "learning_rate": 0.0004565715676397238,
      "loss": 2.1607,
      "step": 74932
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.7101575136184692,
      "learning_rate": 0.0004565680783614044,
      "loss": 2.1901,
      "step": 74933
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.8030770421028137,
      "learning_rate": 0.0004565645890539761,
      "loss": 2.2333,
      "step": 74934
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.8753220438957214,
      "learning_rate": 0.00045656109971743935,
      "loss": 2.1312,
      "step": 74935
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.7988557815551758,
      "learning_rate": 0.000456557610351795,
      "loss": 2.1679,
      "step": 74936
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.8014369606971741,
      "learning_rate": 0.0004565541209570437,
      "loss": 2.1542,
      "step": 74937
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.8027953505516052,
      "learning_rate": 0.0004565506315331859,
      "loss": 2.0451,
      "step": 74938
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.7400645613670349,
      "learning_rate": 0.0004565471420802225,
      "loss": 2.3634,
      "step": 74939
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.8133347034454346,
      "learning_rate": 0.00045654365259815397,
      "loss": 2.3204,
      "step": 74940
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.7840728163719177,
      "learning_rate": 0.000456540163086981,
      "loss": 2.2556,
      "step": 74941
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.78788161277771,
      "learning_rate": 0.0004565366735467043,
      "loss": 2.0859,
      "step": 74942
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.769636332988739,
      "learning_rate": 0.00045653318397732455,
      "loss": 2.3037,
      "step": 74943
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.7238425612449646,
      "learning_rate": 0.00045652969437884225,
      "loss": 2.2508,
      "step": 74944
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.7418264746665955,
      "learning_rate": 0.0004565262047512581,
      "loss": 2.2391,
      "step": 74945
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.6978875994682312,
      "learning_rate": 0.0004565227150945728,
      "loss": 2.1446,
      "step": 74946
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.8441889882087708,
      "learning_rate": 0.000456519225408787,
      "loss": 2.1542,
      "step": 74947
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.8465925455093384,
      "learning_rate": 0.00045651573569390126,
      "loss": 2.1282,
      "step": 74948
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.7285498976707458,
      "learning_rate": 0.00045651224594991637,
      "loss": 2.1554,
      "step": 74949
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.8402864336967468,
      "learning_rate": 0.00045650875617683285,
      "loss": 2.3875,
      "step": 74950
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.693697988986969,
      "learning_rate": 0.00045650526637465135,
      "loss": 2.2918,
      "step": 74951
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.7516961693763733,
      "learning_rate": 0.00045650177654337273,
      "loss": 2.2998,
      "step": 74952
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.7391204237937927,
      "learning_rate": 0.0004564982866829973,
      "loss": 1.8829,
      "step": 74953
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.7846606969833374,
      "learning_rate": 0.00045649479679352596,
      "loss": 2.1746,
      "step": 74954
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.8034942150115967,
      "learning_rate": 0.0004564913068749593,
      "loss": 2.2061,
      "step": 74955
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.7447853088378906,
      "learning_rate": 0.0004564878169272979,
      "loss": 2.1174,
      "step": 74956
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.7540237307548523,
      "learning_rate": 0.00045648432695054243,
      "loss": 2.0794,
      "step": 74957
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.7700310349464417,
      "learning_rate": 0.0004564808369446937,
      "loss": 1.9544,
      "step": 74958
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.079201340675354,
      "learning_rate": 0.00045647734690975217,
      "loss": 2.2291,
      "step": 74959
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.7938365340232849,
      "learning_rate": 0.0004564738568457185,
      "loss": 2.2349,
      "step": 74960
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.8214313983917236,
      "learning_rate": 0.00045647036675259346,
      "loss": 2.0829,
      "step": 74961
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.8748869895935059,
      "learning_rate": 0.0004564668766303776,
      "loss": 2.4546,
      "step": 74962
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.7827829718589783,
      "learning_rate": 0.00045646338647907145,
      "loss": 2.1453,
      "step": 74963
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.883596658706665,
      "learning_rate": 0.000456459896298676,
      "loss": 2.1247,
      "step": 74964
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.799839437007904,
      "learning_rate": 0.0004564564060891916,
      "loss": 2.2043,
      "step": 74965
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.7944869995117188,
      "learning_rate": 0.00045645291585061907,
      "loss": 2.0852,
      "step": 74966
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.7511815428733826,
      "learning_rate": 0.0004564494255829589,
      "loss": 2.1278,
      "step": 74967
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.7915090918540955,
      "learning_rate": 0.0004564459352862119,
      "loss": 2.2376,
      "step": 74968
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.1040664911270142,
      "learning_rate": 0.0004564424449603786,
      "loss": 2.1186,
      "step": 74969
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.7435837984085083,
      "learning_rate": 0.00045643895460545977,
      "loss": 2.187,
      "step": 74970
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.7084254026412964,
      "learning_rate": 0.00045643546422145597,
      "loss": 2.0606,
      "step": 74971
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.8048799633979797,
      "learning_rate": 0.0004564319738083678,
      "loss": 2.4236,
      "step": 74972
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.8128685355186462,
      "learning_rate": 0.00045642848336619596,
      "loss": 2.2161,
      "step": 74973
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.8597452640533447,
      "learning_rate": 0.0004564249928949412,
      "loss": 1.9591,
      "step": 74974
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.7589619159698486,
      "learning_rate": 0.00045642150239460404,
      "loss": 1.9352,
      "step": 74975
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.7342110872268677,
      "learning_rate": 0.00045641801186518515,
      "loss": 2.0401,
      "step": 74976
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.0469324588775635,
      "learning_rate": 0.0004564145213066853,
      "loss": 2.1838,
      "step": 74977
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.8001264333724976,
      "learning_rate": 0.00045641103071910493,
      "loss": 2.1844,
      "step": 74978
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.9692471027374268,
      "learning_rate": 0.0004564075401024448,
      "loss": 2.2038,
      "step": 74979
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.7844244837760925,
      "learning_rate": 0.0004564040494567056,
      "loss": 2.1932,
      "step": 74980
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.7786102294921875,
      "learning_rate": 0.0004564005587818879,
      "loss": 2.2176,
      "step": 74981
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.7521893382072449,
      "learning_rate": 0.00045639706807799234,
      "loss": 2.3372,
      "step": 74982
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.9416189193725586,
      "learning_rate": 0.00045639357734501964,
      "loss": 2.273,
      "step": 74983
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.8394873142242432,
      "learning_rate": 0.00045639008658297054,
      "loss": 2.2667,
      "step": 74984
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.7657321691513062,
      "learning_rate": 0.00045638659579184547,
      "loss": 2.0666,
      "step": 74985
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.9374650120735168,
      "learning_rate": 0.0004563831049716452,
      "loss": 2.1644,
      "step": 74986
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.7348337769508362,
      "learning_rate": 0.00045637961412237034,
      "loss": 2.4033,
      "step": 74987
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.948557436466217,
      "learning_rate": 0.0004563761232440216,
      "loss": 2.3299,
      "step": 74988
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.8227942585945129,
      "learning_rate": 0.00045637263233659956,
      "loss": 2.5638,
      "step": 74989
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.8907933831214905,
      "learning_rate": 0.0004563691414001048,
      "loss": 2.2886,
      "step": 74990
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.7848833203315735,
      "learning_rate": 0.00045636565043453823,
      "loss": 2.3765,
      "step": 74991
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.9013253450393677,
      "learning_rate": 0.0004563621594399003,
      "loss": 2.2424,
      "step": 74992
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.7788551449775696,
      "learning_rate": 0.0004563586684161916,
      "loss": 2.0746,
      "step": 74993
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.770492672920227,
      "learning_rate": 0.00045635517736341294,
      "loss": 2.1425,
      "step": 74994
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.8179892301559448,
      "learning_rate": 0.0004563516862815649,
      "loss": 2.2854,
      "step": 74995
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.817794680595398,
      "learning_rate": 0.00045634819517064806,
      "loss": 2.0502,
      "step": 74996
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.9181939959526062,
      "learning_rate": 0.0004563447040306632,
      "loss": 2.2857,
      "step": 74997
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.7598221302032471,
      "learning_rate": 0.00045634121286161096,
      "loss": 2.3465,
      "step": 74998
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.8067951798439026,
      "learning_rate": 0.0004563377216634919,
      "loss": 2.2695,
      "step": 74999
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.8247597217559814,
      "learning_rate": 0.00045633423043630667,
      "loss": 1.9941,
      "step": 75000
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.8180976510047913,
      "learning_rate": 0.00045633073918005594,
      "loss": 2.5077,
      "step": 75001
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.7938418984413147,
      "learning_rate": 0.00045632724789474046,
      "loss": 2.158,
      "step": 75002
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.7556608319282532,
      "learning_rate": 0.0004563237565803607,
      "loss": 2.0344,
      "step": 75003
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.8050435781478882,
      "learning_rate": 0.00045632026523691747,
      "loss": 2.1752,
      "step": 75004
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.8033950924873352,
      "learning_rate": 0.00045631677386441137,
      "loss": 1.921,
      "step": 75005
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.7772133946418762,
      "learning_rate": 0.00045631328246284296,
      "loss": 2.1946,
      "step": 75006
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.8379670977592468,
      "learning_rate": 0.00045630979103221305,
      "loss": 2.2926,
      "step": 75007
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.8760372996330261,
      "learning_rate": 0.0004563062995725221,
      "loss": 2.2844,
      "step": 75008
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.7546728253364563,
      "learning_rate": 0.0004563028080837709,
      "loss": 2.2386,
      "step": 75009
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.7613463401794434,
      "learning_rate": 0.00045629931656596004,
      "loss": 2.1524,
      "step": 75010
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.7819070816040039,
      "learning_rate": 0.0004562958250190902,
      "loss": 2.089,
      "step": 75011
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.8164697885513306,
      "learning_rate": 0.00045629233344316204,
      "loss": 2.1345,
      "step": 75012
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.7380990982055664,
      "learning_rate": 0.00045628884183817616,
      "loss": 2.2116,
      "step": 75013
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.811798095703125,
      "learning_rate": 0.0004562853502041333,
      "loss": 2.144,
      "step": 75014
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.9130100607872009,
      "learning_rate": 0.0004562818585410339,
      "loss": 2.2287,
      "step": 75015
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.8038443922996521,
      "learning_rate": 0.00045627836684887887,
      "loss": 2.214,
      "step": 75016
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.8476836681365967,
      "learning_rate": 0.0004562748751276687,
      "loss": 2.1161,
      "step": 75017
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.7836292386054993,
      "learning_rate": 0.00045627138337740405,
      "loss": 2.1688,
      "step": 75018
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.8216245174407959,
      "learning_rate": 0.00045626789159808567,
      "loss": 2.2434,
      "step": 75019
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.8086738586425781,
      "learning_rate": 0.0004562643997897141,
      "loss": 2.0635,
      "step": 75020
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.7683844566345215,
      "learning_rate": 0.00045626090795228996,
      "loss": 2.231,
      "step": 75021
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.7854528427124023,
      "learning_rate": 0.00045625741608581414,
      "loss": 2.1624,
      "step": 75022
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.8888799548149109,
      "learning_rate": 0.000456253924190287,
      "loss": 2.1192,
      "step": 75023
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.7597007155418396,
      "learning_rate": 0.0004562504322657093,
      "loss": 2.2487,
      "step": 75024
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.8093050718307495,
      "learning_rate": 0.0004562469403120818,
      "loss": 2.1105,
      "step": 75025
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.8678706884384155,
      "learning_rate": 0.0004562434483294049,
      "loss": 2.3003,
      "step": 75026
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.8102552890777588,
      "learning_rate": 0.0004562399563176794,
      "loss": 2.2465,
      "step": 75027
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.7539597749710083,
      "learning_rate": 0.00045623646427690604,
      "loss": 2.2951,
      "step": 75028
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.7952409386634827,
      "learning_rate": 0.00045623297220708534,
      "loss": 2.4745,
      "step": 75029
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.013188123703003,
      "learning_rate": 0.000456229480108218,
      "loss": 2.2356,
      "step": 75030
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.835991382598877,
      "learning_rate": 0.0004562259879803046,
      "loss": 2.4548,
      "step": 75031
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.719491720199585,
      "learning_rate": 0.00045622249582334587,
      "loss": 2.2207,
      "step": 75032
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.791163444519043,
      "learning_rate": 0.0004562190036373424,
      "loss": 2.2485,
      "step": 75033
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.7966721057891846,
      "learning_rate": 0.0004562155114222948,
      "loss": 2.2567,
      "step": 75034
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.7819231748580933,
      "learning_rate": 0.000456212019178204,
      "loss": 2.2019,
      "step": 75035
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.7381287217140198,
      "learning_rate": 0.00045620852690507024,
      "loss": 2.0928,
      "step": 75036
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.8124021291732788,
      "learning_rate": 0.00045620503460289444,
      "loss": 2.2414,
      "step": 75037
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.8753218054771423,
      "learning_rate": 0.00045620154227167726,
      "loss": 2.248,
      "step": 75038
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.7690795660018921,
      "learning_rate": 0.0004561980499114191,
      "loss": 2.1099,
      "step": 75039
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.7909749150276184,
      "learning_rate": 0.0004561945575221209,
      "loss": 2.0172,
      "step": 75040
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.8734102249145508,
      "learning_rate": 0.0004561910651037832,
      "loss": 2.2977,
      "step": 75041
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.7967479825019836,
      "learning_rate": 0.00045618757265640655,
      "loss": 2.2491,
      "step": 75042
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.8966495394706726,
      "learning_rate": 0.0004561840801799917,
      "loss": 2.2292,
      "step": 75043
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.9194934368133545,
      "learning_rate": 0.0004561805876745392,
      "loss": 2.2167,
      "step": 75044
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.8489848971366882,
      "learning_rate": 0.0004561770951400499,
      "loss": 2.1376,
      "step": 75045
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.8244029879570007,
      "learning_rate": 0.0004561736025765243,
      "loss": 2.362,
      "step": 75046
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.8195680379867554,
      "learning_rate": 0.00045617010998396317,
      "loss": 2.1409,
      "step": 75047
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.7686567306518555,
      "learning_rate": 0.0004561666173623669,
      "loss": 2.005,
      "step": 75048
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.7854686975479126,
      "learning_rate": 0.00045616312471173637,
      "loss": 2.1038,
      "step": 75049
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.7774807810783386,
      "learning_rate": 0.0004561596320320722,
      "loss": 2.3279,
      "step": 75050
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.9499270915985107,
      "learning_rate": 0.000456156139323375,
      "loss": 2.2232,
      "step": 75051
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.8297359943389893,
      "learning_rate": 0.0004561526465856453,
      "loss": 2.1663,
      "step": 75052
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.8240877389907837,
      "learning_rate": 0.0004561491538188841,
      "loss": 2.0434,
      "step": 75053
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.8693091869354248,
      "learning_rate": 0.0004561456610230917,
      "loss": 2.09,
      "step": 75054
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.8350887894630432,
      "learning_rate": 0.0004561421681982689,
      "loss": 2.3815,
      "step": 75055
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.7631285786628723,
      "learning_rate": 0.0004561386753444163,
      "loss": 2.3261,
      "step": 75056
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.7484487295150757,
      "learning_rate": 0.0004561351824615346,
      "loss": 2.2375,
      "step": 75057
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.7824994325637817,
      "learning_rate": 0.00045613168954962435,
      "loss": 2.3654,
      "step": 75058
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.7288517355918884,
      "learning_rate": 0.0004561281966086864,
      "loss": 2.2644,
      "step": 75059
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.7615271806716919,
      "learning_rate": 0.00045612470363872115,
      "loss": 2.0693,
      "step": 75060
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.7245367765426636,
      "learning_rate": 0.0004561212106397293,
      "loss": 2.2252,
      "step": 75061
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.7769455909729004,
      "learning_rate": 0.00045611771761171183,
      "loss": 2.1502,
      "step": 75062
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.8664252758026123,
      "learning_rate": 0.000456114224554669,
      "loss": 2.2143,
      "step": 75063
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.8261361718177795,
      "learning_rate": 0.00045611073146860153,
      "loss": 2.3559,
      "step": 75064
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.8273558020591736,
      "learning_rate": 0.0004561072383535102,
      "loss": 2.2197,
      "step": 75065
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.9058014750480652,
      "learning_rate": 0.0004561037452093955,
      "loss": 2.0154,
      "step": 75066
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.7541828155517578,
      "learning_rate": 0.0004561002520362582,
      "loss": 2.2033,
      "step": 75067
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.7457855939865112,
      "learning_rate": 0.00045609675883409893,
      "loss": 1.9987,
      "step": 75068
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.8023107051849365,
      "learning_rate": 0.00045609326560291843,
      "loss": 2.1935,
      "step": 75069
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.8361617922782898,
      "learning_rate": 0.00045608977234271714,
      "loss": 2.0746,
      "step": 75070
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.7716735601425171,
      "learning_rate": 0.0004560862790534958,
      "loss": 2.1265,
      "step": 75071
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.0357820987701416,
      "learning_rate": 0.00045608278573525514,
      "loss": 1.9741,
      "step": 75072
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.7222652435302734,
      "learning_rate": 0.0004560792923879957,
      "loss": 2.0763,
      "step": 75073
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.8440853357315063,
      "learning_rate": 0.0004560757990117182,
      "loss": 2.2218,
      "step": 75074
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.80561363697052,
      "learning_rate": 0.0004560723056064232,
      "loss": 2.1986,
      "step": 75075
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.8108463883399963,
      "learning_rate": 0.00045606881217211153,
      "loss": 2.3142,
      "step": 75076
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.8848555684089661,
      "learning_rate": 0.0004560653187087836,
      "loss": 2.2499,
      "step": 75077
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.865604043006897,
      "learning_rate": 0.0004560618252164402,
      "loss": 2.1765,
      "step": 75078
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.0716891288757324,
      "learning_rate": 0.000456058331695082,
      "loss": 1.8111,
      "step": 75079
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.7997188568115234,
      "learning_rate": 0.0004560548381447096,
      "loss": 2.2639,
      "step": 75080
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.7429832816123962,
      "learning_rate": 0.00045605134456532366,
      "loss": 2.2687,
      "step": 75081
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.0454257726669312,
      "learning_rate": 0.0004560478509569248,
      "loss": 2.2359,
      "step": 75082
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.9175795912742615,
      "learning_rate": 0.00045604435731951383,
      "loss": 2.1768,
      "step": 75083
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.7885056734085083,
      "learning_rate": 0.00045604086365309116,
      "loss": 2.2481,
      "step": 75084
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.9971046447753906,
      "learning_rate": 0.0004560373699576575,
      "loss": 2.2472,
      "step": 75085
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.9727131724357605,
      "learning_rate": 0.00045603387623321365,
      "loss": 2.2322,
      "step": 75086
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.7218990325927734,
      "learning_rate": 0.0004560303824797601,
      "loss": 2.1324,
      "step": 75087
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.9347793459892273,
      "learning_rate": 0.0004560268886972975,
      "loss": 2.2343,
      "step": 75088
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.0374889373779297,
      "learning_rate": 0.0004560233948858267,
      "loss": 2.1798,
      "step": 75089
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.8621639609336853,
      "learning_rate": 0.00045601990104534814,
      "loss": 2.151,
      "step": 75090
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.8600512146949768,
      "learning_rate": 0.0004560164071758624,
      "loss": 2.2756,
      "step": 75091
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.8833786249160767,
      "learning_rate": 0.0004560129132773705,
      "loss": 2.0692,
      "step": 75092
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.8288094997406006,
      "learning_rate": 0.0004560094193498727,
      "loss": 2.229,
      "step": 75093
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.8134384155273438,
      "learning_rate": 0.00045600592539336984,
      "loss": 2.1113,
      "step": 75094
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.7545436024665833,
      "learning_rate": 0.0004560024314078626,
      "loss": 2.1006,
      "step": 75095
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.8534213304519653,
      "learning_rate": 0.0004559989373933515,
      "loss": 2.4003,
      "step": 75096
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.7881028652191162,
      "learning_rate": 0.00045599544334983726,
      "loss": 2.025,
      "step": 75097
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.721508264541626,
      "learning_rate": 0.00045599194927732053,
      "loss": 2.216,
      "step": 75098
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.9750610589981079,
      "learning_rate": 0.00045598845517580194,
      "loss": 2.0678,
      "step": 75099
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.7839951515197754,
      "learning_rate": 0.00045598496104528214,
      "loss": 2.281,
      "step": 75100
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.8465253114700317,
      "learning_rate": 0.00045598146688576177,
      "loss": 2.2688,
      "step": 75101
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.8155646920204163,
      "learning_rate": 0.0004559779726972416,
      "loss": 2.1948,
      "step": 75102
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.8330227136611938,
      "learning_rate": 0.00045597447847972204,
      "loss": 2.0608,
      "step": 75103
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.8451045155525208,
      "learning_rate": 0.000455970984233204,
      "loss": 2.2731,
      "step": 75104
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.8289464712142944,
      "learning_rate": 0.00045596748995768797,
      "loss": 2.1155,
      "step": 75105
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.7331326603889465,
      "learning_rate": 0.00045596399565317464,
      "loss": 2.1007,
      "step": 75106
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.8614036440849304,
      "learning_rate": 0.0004559605013196646,
      "loss": 1.9066,
      "step": 75107
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.7608959674835205,
      "learning_rate": 0.00045595700695715864,
      "loss": 2.1873,
      "step": 75108
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.8602086901664734,
      "learning_rate": 0.00045595351256565733,
      "loss": 1.9681,
      "step": 75109
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.8309274911880493,
      "learning_rate": 0.00045595001814516126,
      "loss": 2.2419,
      "step": 75110
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.8236215114593506,
      "learning_rate": 0.00045594652369567123,
      "loss": 2.3193,
      "step": 75111
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.8982613682746887,
      "learning_rate": 0.00045594302921718775,
      "loss": 2.2707,
      "step": 75112
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.7868062853813171,
      "learning_rate": 0.0004559395347097114,
      "loss": 2.2967,
      "step": 75113
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.8359325528144836,
      "learning_rate": 0.0004559360401732431,
      "loss": 2.0419,
      "step": 75114
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.8627613186836243,
      "learning_rate": 0.0004559325456077833,
      "loss": 2.1481,
      "step": 75115
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.8628137707710266,
      "learning_rate": 0.00045592905101333266,
      "loss": 1.9683,
      "step": 75116
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.9384738802909851,
      "learning_rate": 0.000455925556389892,
      "loss": 2.3849,
      "step": 75117
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.8728629946708679,
      "learning_rate": 0.00045592206173746165,
      "loss": 2.1395,
      "step": 75118
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.7815095782279968,
      "learning_rate": 0.00045591856705604256,
      "loss": 2.2776,
      "step": 75119
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.9144499897956848,
      "learning_rate": 0.0004559150723456353,
      "loss": 2.1578,
      "step": 75120
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.7494299411773682,
      "learning_rate": 0.0004559115776062404,
      "loss": 2.4366,
      "step": 75121
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.7784925103187561,
      "learning_rate": 0.0004559080828378586,
      "loss": 2.2297,
      "step": 75122
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.8511133790016174,
      "learning_rate": 0.0004559045880404906,
      "loss": 2.1478,
      "step": 75123
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.825325608253479,
      "learning_rate": 0.0004559010932141369,
      "loss": 2.2765,
      "step": 75124
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.809880793094635,
      "learning_rate": 0.00045589759835879825,
      "loss": 2.2659,
      "step": 75125
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.8008129596710205,
      "learning_rate": 0.0004558941034744754,
      "loss": 2.2165,
      "step": 75126
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.9343413710594177,
      "learning_rate": 0.0004558906085611688,
      "loss": 2.0656,
      "step": 75127
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.7731980085372925,
      "learning_rate": 0.00045588711361887926,
      "loss": 2.1507,
      "step": 75128
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.7749903202056885,
      "learning_rate": 0.00045588361864760734,
      "loss": 2.2621,
      "step": 75129
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.7707942724227905,
      "learning_rate": 0.0004558801236473537,
      "loss": 2.1273,
      "step": 75130
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.6985184550285339,
      "learning_rate": 0.00045587662861811895,
      "loss": 2.1144,
      "step": 75131
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.7940582036972046,
      "learning_rate": 0.0004558731335599039,
      "loss": 2.1525,
      "step": 75132
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.7580525279045105,
      "learning_rate": 0.000455869638472709,
      "loss": 2.1156,
      "step": 75133
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.9098547697067261,
      "learning_rate": 0.00045586614335653506,
      "loss": 2.4298,
      "step": 75134
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.7846788763999939,
      "learning_rate": 0.0004558626482113827,
      "loss": 2.2868,
      "step": 75135
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.7593176364898682,
      "learning_rate": 0.00045585915303725244,
      "loss": 2.2357,
      "step": 75136
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.0376123189926147,
      "learning_rate": 0.000455855657834145,
      "loss": 2.353,
      "step": 75137
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.7551188468933105,
      "learning_rate": 0.00045585216260206113,
      "loss": 2.1549,
      "step": 75138
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.7468655109405518,
      "learning_rate": 0.0004558486673410014,
      "loss": 2.1445,
      "step": 75139
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.8326389193534851,
      "learning_rate": 0.00045584517205096643,
      "loss": 2.2629,
      "step": 75140
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.8758808374404907,
      "learning_rate": 0.00045584167673195684,
      "loss": 2.1622,
      "step": 75141
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.8502594232559204,
      "learning_rate": 0.0004558381813839734,
      "loss": 2.4107,
      "step": 75142
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.7913022637367249,
      "learning_rate": 0.0004558346860070167,
      "loss": 2.0728,
      "step": 75143
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.8928865194320679,
      "learning_rate": 0.00045583119060108745,
      "loss": 2.2731,
      "step": 75144
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.9644135236740112,
      "learning_rate": 0.00045582769516618613,
      "loss": 2.1436,
      "step": 75145
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.873746395111084,
      "learning_rate": 0.00045582419970231357,
      "loss": 2.1001,
      "step": 75146
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.7684000730514526,
      "learning_rate": 0.0004558207042094703,
      "loss": 2.1204,
      "step": 75147
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.7665640711784363,
      "learning_rate": 0.0004558172086876571,
      "loss": 2.4212,
      "step": 75148
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.8099161386489868,
      "learning_rate": 0.0004558137131368745,
      "loss": 2.2941,
      "step": 75149
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.8127942085266113,
      "learning_rate": 0.00045581021755712316,
      "loss": 2.1034,
      "step": 75150
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.8162489533424377,
      "learning_rate": 0.00045580672194840373,
      "loss": 2.1784,
      "step": 75151
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.8317784667015076,
      "learning_rate": 0.000455803226310717,
      "loss": 2.2218,
      "step": 75152
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.8397468328475952,
      "learning_rate": 0.0004557997306440635,
      "loss": 2.1884,
      "step": 75153
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.8259725570678711,
      "learning_rate": 0.0004557962349484438,
      "loss": 2.1278,
      "step": 75154
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.8795949816703796,
      "learning_rate": 0.00045579273922385867,
      "loss": 2.1015,
      "step": 75155
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.9975742697715759,
      "learning_rate": 0.00045578924347030876,
      "loss": 2.1647,
      "step": 75156
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.7072070240974426,
      "learning_rate": 0.00045578574768779465,
      "loss": 2.1427,
      "step": 75157
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.8073061108589172,
      "learning_rate": 0.000455782251876317,
      "loss": 2.2598,
      "step": 75158
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.8930642604827881,
      "learning_rate": 0.00045577875603587653,
      "loss": 2.2071,
      "step": 75159
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.7692278027534485,
      "learning_rate": 0.0004557752601664739,
      "loss": 2.2562,
      "step": 75160
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.8019400238990784,
      "learning_rate": 0.0004557717642681097,
      "loss": 2.1438,
      "step": 75161
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.8199127316474915,
      "learning_rate": 0.00045576826834078444,
      "loss": 2.2513,
      "step": 75162
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.9457185864448547,
      "learning_rate": 0.000455764772384499,
      "loss": 2.3157,
      "step": 75163
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.7897529006004333,
      "learning_rate": 0.00045576127639925407,
      "loss": 2.2971,
      "step": 75164
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.796284556388855,
      "learning_rate": 0.00045575778038505003,
      "loss": 2.0908,
      "step": 75165
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.7552054524421692,
      "learning_rate": 0.0004557542843418877,
      "loss": 2.0574,
      "step": 75166
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.7105000019073486,
      "learning_rate": 0.0004557507882697678,
      "loss": 2.3368,
      "step": 75167
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.8564769625663757,
      "learning_rate": 0.0004557472921686909,
      "loss": 2.305,
      "step": 75168
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.7856951951980591,
      "learning_rate": 0.00045574379603865756,
      "loss": 2.1665,
      "step": 75169
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.9577603340148926,
      "learning_rate": 0.0004557402998796684,
      "loss": 2.0058,
      "step": 75170
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.8996404409408569,
      "learning_rate": 0.0004557368036917243,
      "loss": 2.1026,
      "step": 75171
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.8020234107971191,
      "learning_rate": 0.0004557333074748258,
      "loss": 2.2608,
      "step": 75172
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.7916266918182373,
      "learning_rate": 0.0004557298112289735,
      "loss": 1.8646,
      "step": 75173
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.7618659138679504,
      "learning_rate": 0.0004557263149541681,
      "loss": 2.3573,
      "step": 75174
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.7534337639808655,
      "learning_rate": 0.00045572281865041025,
      "loss": 2.1097,
      "step": 75175
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.7480960488319397,
      "learning_rate": 0.0004557193223177007,
      "loss": 1.9474,
      "step": 75176
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.7989804148674011,
      "learning_rate": 0.0004557158259560398,
      "loss": 2.1133,
      "step": 75177
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.7186879515647888,
      "learning_rate": 0.0004557123295654285,
      "loss": 2.0497,
      "step": 75178
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.7776159644126892,
      "learning_rate": 0.0004557088331458673,
      "loss": 2.3174,
      "step": 75179
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.8463575839996338,
      "learning_rate": 0.00045570533669735683,
      "loss": 2.0687,
      "step": 75180
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.7538875937461853,
      "learning_rate": 0.0004557018402198979,
      "loss": 2.0302,
      "step": 75181
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.7765353322029114,
      "learning_rate": 0.000455698343713491,
      "loss": 2.1914,
      "step": 75182
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.8199601769447327,
      "learning_rate": 0.00045569484717813686,
      "loss": 2.1465,
      "step": 75183
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.7248743772506714,
      "learning_rate": 0.0004556913506138362,
      "loss": 1.9464,
      "step": 75184
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.7708380222320557,
      "learning_rate": 0.0004556878540205894,
      "loss": 2.2746,
      "step": 75185
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.7776349782943726,
      "learning_rate": 0.00045568435739839737,
      "loss": 2.333,
      "step": 75186
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.7564991116523743,
      "learning_rate": 0.0004556808607472607,
      "loss": 2.1455,
      "step": 75187
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.760263204574585,
      "learning_rate": 0.00045567736406718,
      "loss": 2.2149,
      "step": 75188
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.7749719619750977,
      "learning_rate": 0.0004556738673581559,
      "loss": 2.3862,
      "step": 75189
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.8467317223548889,
      "learning_rate": 0.00045567037062018925,
      "loss": 2.0421,
      "step": 75190
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.7843599319458008,
      "learning_rate": 0.0004556668738532804,
      "loss": 2.2639,
      "step": 75191
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.7518360614776611,
      "learning_rate": 0.0004556633770574301,
      "loss": 2.1433,
      "step": 75192
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.7877721786499023,
      "learning_rate": 0.0004556598802326392,
      "loss": 2.1704,
      "step": 75193
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.7283169627189636,
      "learning_rate": 0.0004556563833789081,
      "loss": 1.8736,
      "step": 75194
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.7262022495269775,
      "learning_rate": 0.00045565288649623744,
      "loss": 2.4715,
      "step": 75195
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.7905296683311462,
      "learning_rate": 0.0004556493895846281,
      "loss": 2.3174,
      "step": 75196
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.8052549362182617,
      "learning_rate": 0.0004556458926440806,
      "loss": 2.1825,
      "step": 75197
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.7974269390106201,
      "learning_rate": 0.0004556423956745955,
      "loss": 2.2239,
      "step": 75198
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.7841665744781494,
      "learning_rate": 0.0004556388986761736,
      "loss": 2.2556,
      "step": 75199
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.8875176906585693,
      "learning_rate": 0.0004556354016488155,
      "loss": 2.1276,
      "step": 75200
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.8374307155609131,
      "learning_rate": 0.0004556319045925218,
      "loss": 2.1418,
      "step": 75201
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.784733772277832,
      "learning_rate": 0.0004556284075072933,
      "loss": 2.2642,
      "step": 75202
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.7512402534484863,
      "learning_rate": 0.00045562491039313036,
      "loss": 2.1543,
      "step": 75203
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.729403555393219,
      "learning_rate": 0.00045562141325003393,
      "loss": 2.0029,
      "step": 75204
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.7693867683410645,
      "learning_rate": 0.0004556179160780046,
      "loss": 1.9781,
      "step": 75205
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.8662765622138977,
      "learning_rate": 0.0004556144188770428,
      "loss": 2.2724,
      "step": 75206
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.796953558921814,
      "learning_rate": 0.00045561092164714945,
      "loss": 2.2441,
      "step": 75207
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.7444989085197449,
      "learning_rate": 0.00045560742438832506,
      "loss": 2.1534,
      "step": 75208
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.8951240181922913,
      "learning_rate": 0.00045560392710057037,
      "loss": 2.1253,
      "step": 75209
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.7793947458267212,
      "learning_rate": 0.0004556004297838858,
      "loss": 2.0827,
      "step": 75210
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.6961417198181152,
      "learning_rate": 0.00045559693243827243,
      "loss": 2.0575,
      "step": 75211
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.8118630051612854,
      "learning_rate": 0.0004555934350637305,
      "loss": 2.1167,
      "step": 75212
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.7470789551734924,
      "learning_rate": 0.00045558993766026075,
      "loss": 2.353,
      "step": 75213
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.7641134858131409,
      "learning_rate": 0.0004555864402278641,
      "loss": 2.2867,
      "step": 75214
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.891682505607605,
      "learning_rate": 0.0004555829427665408,
      "loss": 2.1226,
      "step": 75215
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.7754825949668884,
      "learning_rate": 0.0004555794452762917,
      "loss": 2.2514,
      "step": 75216
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.7785941958427429,
      "learning_rate": 0.0004555759477571176,
      "loss": 2.397,
      "step": 75217
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.7459810376167297,
      "learning_rate": 0.0004555724502090189,
      "loss": 2.2479,
      "step": 75218
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.7279956340789795,
      "learning_rate": 0.0004555689526319963,
      "loss": 2.2747,
      "step": 75219
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.7903537750244141,
      "learning_rate": 0.00045556545502605064,
      "loss": 2.0363,
      "step": 75220
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.7496628165245056,
      "learning_rate": 0.0004555619573911822,
      "loss": 2.1345,
      "step": 75221
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.707908570766449,
      "learning_rate": 0.000455558459727392,
      "loss": 2.2008,
      "step": 75222
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.7075222730636597,
      "learning_rate": 0.00045555496203468054,
      "loss": 2.1379,
      "step": 75223
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.7488125562667847,
      "learning_rate": 0.0004555514643130485,
      "loss": 2.2099,
      "step": 75224
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.9991872310638428,
      "learning_rate": 0.0004555479665624965,
      "loss": 2.3026,
      "step": 75225
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.8484964966773987,
      "learning_rate": 0.00045554446878302515,
      "loss": 2.003,
      "step": 75226
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.8890617489814758,
      "learning_rate": 0.00045554097097463516,
      "loss": 2.1509,
      "step": 75227
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.8238238096237183,
      "learning_rate": 0.00045553747313732717,
      "loss": 2.3662,
      "step": 75228
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.8695634603500366,
      "learning_rate": 0.0004555339752711018,
      "loss": 2.3144,
      "step": 75229
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.8598368167877197,
      "learning_rate": 0.0004555304773759598,
      "loss": 2.298,
      "step": 75230
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.7217005491256714,
      "learning_rate": 0.00045552697945190173,
      "loss": 2.0732,
      "step": 75231
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.774884819984436,
      "learning_rate": 0.00045552348149892824,
      "loss": 2.2029,
      "step": 75232
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.8610120415687561,
      "learning_rate": 0.00045551998351703993,
      "loss": 2.2223,
      "step": 75233
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.8154435753822327,
      "learning_rate": 0.0004555164855062376,
      "loss": 2.1735,
      "step": 75234
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.6951228380203247,
      "learning_rate": 0.00045551298746652184,
      "loss": 2.3251,
      "step": 75235
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.7863672375679016,
      "learning_rate": 0.00045550948939789324,
      "loss": 2.0078,
      "step": 75236
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.7819123268127441,
      "learning_rate": 0.0004555059913003525,
      "loss": 2.2881,
      "step": 75237
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.8265607357025146,
      "learning_rate": 0.0004555024931739002,
      "loss": 2.252,
      "step": 75238
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.7665858864784241,
      "learning_rate": 0.00045549899501853716,
      "loss": 2.1454,
      "step": 75239
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.829887866973877,
      "learning_rate": 0.0004554954968342639,
      "loss": 2.1937,
      "step": 75240
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.8967759013175964,
      "learning_rate": 0.000455491998621081,
      "loss": 1.9358,
      "step": 75241
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.7947109341621399,
      "learning_rate": 0.0004554885003789893,
      "loss": 2.1938,
      "step": 75242
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.779279887676239,
      "learning_rate": 0.0004554850021079893,
      "loss": 2.321,
      "step": 75243
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.7798079252243042,
      "learning_rate": 0.00045548150380808166,
      "loss": 2.1896,
      "step": 75244
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.8472403883934021,
      "learning_rate": 0.0004554780054792672,
      "loss": 2.0236,
      "step": 75245
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.8977144360542297,
      "learning_rate": 0.00045547450712154634,
      "loss": 2.0962,
      "step": 75246
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.7914378643035889,
      "learning_rate": 0.0004554710087349199,
      "loss": 2.4937,
      "step": 75247
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.797978937625885,
      "learning_rate": 0.0004554675103193884,
      "loss": 2.1362,
      "step": 75248
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.7430301904678345,
      "learning_rate": 0.00045546401187495267,
      "loss": 2.1882,
      "step": 75249
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.8140287399291992,
      "learning_rate": 0.00045546051340161314,
      "loss": 2.118,
      "step": 75250
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.808452844619751,
      "learning_rate": 0.00045545701489937057,
      "loss": 2.074,
      "step": 75251
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.7106918692588806,
      "learning_rate": 0.0004554535163682256,
      "loss": 2.0634,
      "step": 75252
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.8054044246673584,
      "learning_rate": 0.00045545001780817896,
      "loss": 2.1569,
      "step": 75253
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.7468189597129822,
      "learning_rate": 0.0004554465192192311,
      "loss": 2.3089,
      "step": 75254
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.8378617167472839,
      "learning_rate": 0.0004554430206013829,
      "loss": 2.289,
      "step": 75255
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.7697364091873169,
      "learning_rate": 0.0004554395219546349,
      "loss": 2.1757,
      "step": 75256
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.7814497947692871,
      "learning_rate": 0.0004554360232789878,
      "loss": 2.3221,
      "step": 75257
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.8062534332275391,
      "learning_rate": 0.0004554325245744421,
      "loss": 2.3696,
      "step": 75258
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.7502455115318298,
      "learning_rate": 0.00045542902584099865,
      "loss": 2.1228,
      "step": 75259
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.8768087029457092,
      "learning_rate": 0.000455425527078658,
      "loss": 2.1403,
      "step": 75260
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.8731072545051575,
      "learning_rate": 0.0004554220282874207,
      "loss": 2.2315,
      "step": 75261
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.890741229057312,
      "learning_rate": 0.00045541852946728764,
      "loss": 1.9683,
      "step": 75262
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.7421852946281433,
      "learning_rate": 0.0004554150306182593,
      "loss": 2.1574,
      "step": 75263
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.7520852088928223,
      "learning_rate": 0.0004554115317403364,
      "loss": 1.9833,
      "step": 75264
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.8148404955863953,
      "learning_rate": 0.00045540803283351946,
      "loss": 2.1879,
      "step": 75265
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.7469830513000488,
      "learning_rate": 0.00045540453389780934,
      "loss": 2.2237,
      "step": 75266
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.8309189081192017,
      "learning_rate": 0.0004554010349332066,
      "loss": 2.2208,
      "step": 75267
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.7908792495727539,
      "learning_rate": 0.00045539753593971184,
      "loss": 2.2609,
      "step": 75268
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.9582479596138,
      "learning_rate": 0.0004553940369173257,
      "loss": 1.9141,
      "step": 75269
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.8464091420173645,
      "learning_rate": 0.00045539053786604886,
      "loss": 2.1434,
      "step": 75270
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.8152508735656738,
      "learning_rate": 0.000455387038785882,
      "loss": 2.1622,
      "step": 75271
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.8398997187614441,
      "learning_rate": 0.00045538353967682587,
      "loss": 2.0949,
      "step": 75272
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.83756023645401,
      "learning_rate": 0.000455380040538881,
      "loss": 1.9159,
      "step": 75273
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.7648220062255859,
      "learning_rate": 0.00045537654137204784,
      "loss": 2.1306,
      "step": 75274
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.8858487010002136,
      "learning_rate": 0.0004553730421763275,
      "loss": 2.3729,
      "step": 75275
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.8466948866844177,
      "learning_rate": 0.0004553695429517202,
      "loss": 2.1537,
      "step": 75276
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.8194929957389832,
      "learning_rate": 0.00045536604369822686,
      "loss": 2.346,
      "step": 75277
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.7260658740997314,
      "learning_rate": 0.00045536254441584797,
      "loss": 2.1619,
      "step": 75278
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.7583386301994324,
      "learning_rate": 0.0004553590451045843,
      "loss": 2.1591,
      "step": 75279
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.7493966221809387,
      "learning_rate": 0.00045535554576443647,
      "loss": 1.9373,
      "step": 75280
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.8483462929725647,
      "learning_rate": 0.0004553520463954051,
      "loss": 2.1842,
      "step": 75281
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.9665166735649109,
      "learning_rate": 0.0004553485469974908,
      "loss": 2.2751,
      "step": 75282
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.8888785243034363,
      "learning_rate": 0.00045534504757069425,
      "loss": 2.1294,
      "step": 75283
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.801663339138031,
      "learning_rate": 0.0004553415481150162,
      "loss": 2.0312,
      "step": 75284
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.7693489789962769,
      "learning_rate": 0.0004553380486304572,
      "loss": 2.4186,
      "step": 75285
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.8843629956245422,
      "learning_rate": 0.0004553345491170179,
      "loss": 2.042,
      "step": 75286
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.8619825839996338,
      "learning_rate": 0.00045533104957469906,
      "loss": 2.1367,
      "step": 75287
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.7847825884819031,
      "learning_rate": 0.00045532755000350124,
      "loss": 2.2612,
      "step": 75288
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.7711136937141418,
      "learning_rate": 0.000455324050403425,
      "loss": 2.113,
      "step": 75289
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.6993469595909119,
      "learning_rate": 0.0004553205507744711,
      "loss": 2.2702,
      "step": 75290
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.7815970778465271,
      "learning_rate": 0.0004553170511166403,
      "loss": 2.1848,
      "step": 75291
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.8125151991844177,
      "learning_rate": 0.0004553135514299329,
      "loss": 2.3123,
      "step": 75292
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.8086699843406677,
      "learning_rate": 0.00045531005171435,
      "loss": 2.3462,
      "step": 75293
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.7530081272125244,
      "learning_rate": 0.00045530655196989193,
      "loss": 2.1076,
      "step": 75294
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.8035334348678589,
      "learning_rate": 0.0004553030521965594,
      "loss": 2.2425,
      "step": 75295
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.8102195858955383,
      "learning_rate": 0.00045529955239435323,
      "loss": 2.3326,
      "step": 75296
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.7884239554405212,
      "learning_rate": 0.0004552960525632739,
      "loss": 2.2561,
      "step": 75297
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.8192732334136963,
      "learning_rate": 0.000455292552703322,
      "loss": 2.2844,
      "step": 75298
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.7869594693183899,
      "learning_rate": 0.00045528905281449843,
      "loss": 1.9543,
      "step": 75299
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.8198930025100708,
      "learning_rate": 0.00045528555289680363,
      "loss": 2.2373,
      "step": 75300
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.7823789715766907,
      "learning_rate": 0.00045528205295023827,
      "loss": 2.2072,
      "step": 75301
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.7967775464057922,
      "learning_rate": 0.00045527855297480303,
      "loss": 2.3534,
      "step": 75302
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.8972768783569336,
      "learning_rate": 0.00045527505297049864,
      "loss": 2.3812,
      "step": 75303
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.7761788964271545,
      "learning_rate": 0.00045527155293732563,
      "loss": 2.1932,
      "step": 75304
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.77849280834198,
      "learning_rate": 0.00045526805287528476,
      "loss": 2.1826,
      "step": 75305
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.8018741011619568,
      "learning_rate": 0.0004552645527843767,
      "loss": 2.1004,
      "step": 75306
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.7795982360839844,
      "learning_rate": 0.00045526105266460194,
      "loss": 2.2074,
      "step": 75307
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.7170912027359009,
      "learning_rate": 0.00045525755251596113,
      "loss": 2.2995,
      "step": 75308
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.7487385869026184,
      "learning_rate": 0.0004552540523384552,
      "loss": 1.9994,
      "step": 75309
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.8295113444328308,
      "learning_rate": 0.00045525055213208444,
      "loss": 2.2103,
      "step": 75310
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.8153802156448364,
      "learning_rate": 0.0004552470518968497,
      "loss": 2.1892,
      "step": 75311
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.8364572525024414,
      "learning_rate": 0.00045524355163275177,
      "loss": 2.0213,
      "step": 75312
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.9236965179443359,
      "learning_rate": 0.000455240051339791,
      "loss": 2.153,
      "step": 75313
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.9026653170585632,
      "learning_rate": 0.00045523655101796815,
      "loss": 2.1303,
      "step": 75314
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.756310760974884,
      "learning_rate": 0.000455233050667284,
      "loss": 2.096,
      "step": 75315
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.8339945673942566,
      "learning_rate": 0.00045522955028773896,
      "loss": 2.1838,
      "step": 75316
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.7940614223480225,
      "learning_rate": 0.0004552260498793339,
      "loss": 2.2705,
      "step": 75317
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.7920001745223999,
      "learning_rate": 0.0004552225494420694,
      "loss": 2.2086,
      "step": 75318
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.7813503742218018,
      "learning_rate": 0.000455219048975946,
      "loss": 2.4288,
      "step": 75319
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.7830258011817932,
      "learning_rate": 0.00045521554848096456,
      "loss": 2.2064,
      "step": 75320
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.7891995906829834,
      "learning_rate": 0.0004552120479571256,
      "loss": 2.347,
      "step": 75321
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.787814199924469,
      "learning_rate": 0.00045520854740442975,
      "loss": 2.2342,
      "step": 75322
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.7777167558670044,
      "learning_rate": 0.00045520504682287764,
      "loss": 2.1574,
      "step": 75323
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.8136785626411438,
      "learning_rate": 0.00045520154621247013,
      "loss": 2.2857,
      "step": 75324
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.8995034098625183,
      "learning_rate": 0.00045519804557320767,
      "loss": 2.2956,
      "step": 75325
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.7671645283699036,
      "learning_rate": 0.0004551945449050909,
      "loss": 2.2454,
      "step": 75326
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.9028406739234924,
      "learning_rate": 0.00045519104420812064,
      "loss": 2.2689,
      "step": 75327
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.8318772315979004,
      "learning_rate": 0.0004551875434822974,
      "loss": 2.1698,
      "step": 75328
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.0694985389709473,
      "learning_rate": 0.0004551840427276218,
      "loss": 2.0331,
      "step": 75329
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.865879476070404,
      "learning_rate": 0.0004551805419440946,
      "loss": 2.3577,
      "step": 75330
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.7984933853149414,
      "learning_rate": 0.0004551770411317165,
      "loss": 2.4384,
      "step": 75331
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.8050299882888794,
      "learning_rate": 0.0004551735402904879,
      "loss": 2.1169,
      "step": 75332
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.8294264674186707,
      "learning_rate": 0.0004551700394204097,
      "loss": 2.1534,
      "step": 75333
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.7884328365325928,
      "learning_rate": 0.0004551665385214824,
      "loss": 2.3582,
      "step": 75334
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.8044179677963257,
      "learning_rate": 0.00045516303759370676,
      "loss": 2.2191,
      "step": 75335
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.7939074635505676,
      "learning_rate": 0.00045515953663708336,
      "loss": 2.2741,
      "step": 75336
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.7566513419151306,
      "learning_rate": 0.0004551560356516129,
      "loss": 1.9976,
      "step": 75337
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.8071627020835876,
      "learning_rate": 0.000455152534637296,
      "loss": 2.0705,
      "step": 75338
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.8235465288162231,
      "learning_rate": 0.00045514903359413324,
      "loss": 2.0615,
      "step": 75339
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.8044562339782715,
      "learning_rate": 0.0004551455325221254,
      "loss": 2.1032,
      "step": 75340
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.7369959354400635,
      "learning_rate": 0.00045514203142127315,
      "loss": 2.2045,
      "step": 75341
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.8509157299995422,
      "learning_rate": 0.000455138530291577,
      "loss": 2.0557,
      "step": 75342
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.7372026443481445,
      "learning_rate": 0.00045513502913303763,
      "loss": 2.2988,
      "step": 75343
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.7429221868515015,
      "learning_rate": 0.00045513152794565584,
      "loss": 2.4169,
      "step": 75344
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.9793159365653992,
      "learning_rate": 0.0004551280267294321,
      "loss": 2.1576,
      "step": 75345
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.8073632121086121,
      "learning_rate": 0.0004551245254843671,
      "loss": 2.2806,
      "step": 75346
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.8783616423606873,
      "learning_rate": 0.00045512102421046155,
      "loss": 2.3917,
      "step": 75347
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.7889284491539001,
      "learning_rate": 0.0004551175229077161,
      "loss": 2.4743,
      "step": 75348
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.8189685940742493,
      "learning_rate": 0.0004551140215761313,
      "loss": 2.1488,
      "step": 75349
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.8409425020217896,
      "learning_rate": 0.00045511052021570797,
      "loss": 2.2496,
      "step": 75350
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.8959014415740967,
      "learning_rate": 0.00045510701882644665,
      "loss": 2.2351,
      "step": 75351
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.7508929967880249,
      "learning_rate": 0.000455103517408348,
      "loss": 2.2306,
      "step": 75352
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.8347926735877991,
      "learning_rate": 0.00045510001596141264,
      "loss": 2.1985,
      "step": 75353
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.8290458917617798,
      "learning_rate": 0.0004550965144856413,
      "loss": 2.1932,
      "step": 75354
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.7568382024765015,
      "learning_rate": 0.00045509301298103454,
      "loss": 2.093,
      "step": 75355
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.7973282337188721,
      "learning_rate": 0.0004550895114475931,
      "loss": 2.2125,
      "step": 75356
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.8754868507385254,
      "learning_rate": 0.0004550860098853176,
      "loss": 2.0489,
      "step": 75357
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.8746297359466553,
      "learning_rate": 0.0004550825082942087,
      "loss": 2.1433,
      "step": 75358
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.8162643313407898,
      "learning_rate": 0.00045507900667426693,
      "loss": 2.2425,
      "step": 75359
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.7808905839920044,
      "learning_rate": 0.0004550755050254932,
      "loss": 2.0733,
      "step": 75360
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.9258636832237244,
      "learning_rate": 0.00045507200334788794,
      "loss": 2.0723,
      "step": 75361
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.781950831413269,
      "learning_rate": 0.00045506850164145184,
      "loss": 2.2375,
      "step": 75362
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.8624984622001648,
      "learning_rate": 0.00045506499990618555,
      "loss": 2.2267,
      "step": 75363
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.3053925037384033,
      "learning_rate": 0.0004550614981420898,
      "loss": 2.3681,
      "step": 75364
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.81484454870224,
      "learning_rate": 0.0004550579963491652,
      "loss": 2.1799,
      "step": 75365
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.7474466562271118,
      "learning_rate": 0.0004550544945274124,
      "loss": 2.3941,
      "step": 75366
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.047062873840332,
      "learning_rate": 0.000455050992676832,
      "loss": 2.2785,
      "step": 75367
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.0156283378601074,
      "learning_rate": 0.0004550474907974247,
      "loss": 2.1171,
      "step": 75368
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.7973455190658569,
      "learning_rate": 0.00045504398888919116,
      "loss": 2.4293,
      "step": 75369
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.0473304986953735,
      "learning_rate": 0.000455040486952132,
      "loss": 2.3422,
      "step": 75370
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.773932695388794,
      "learning_rate": 0.00045503698498624786,
      "loss": 2.3914,
      "step": 75371
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.8302336931228638,
      "learning_rate": 0.00045503348299153943,
      "loss": 2.4313,
      "step": 75372
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.7379457950592041,
      "learning_rate": 0.0004550299809680075,
      "loss": 2.0178,
      "step": 75373
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.7829946875572205,
      "learning_rate": 0.0004550264789156524,
      "loss": 2.097,
      "step": 75374
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.9602113366127014,
      "learning_rate": 0.000455022976834475,
      "loss": 2.1823,
      "step": 75375
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.7807492017745972,
      "learning_rate": 0.0004550194747244759,
      "loss": 2.0816,
      "step": 75376
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.8086178302764893,
      "learning_rate": 0.0004550159725856557,
      "loss": 2.246,
      "step": 75377
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.0985665321350098,
      "learning_rate": 0.00045501247041801517,
      "loss": 2.2816,
      "step": 75378
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.8381310105323792,
      "learning_rate": 0.0004550089682215549,
      "loss": 2.2224,
      "step": 75379
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.8013201355934143,
      "learning_rate": 0.00045500546599627556,
      "loss": 2.195,
      "step": 75380
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.8463398218154907,
      "learning_rate": 0.00045500196374217766,
      "loss": 2.3528,
      "step": 75381
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.0381735563278198,
      "learning_rate": 0.00045499846145926205,
      "loss": 2.426,
      "step": 75382
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.0153950452804565,
      "learning_rate": 0.0004549949591475293,
      "loss": 2.2388,
      "step": 75383
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.8188923001289368,
      "learning_rate": 0.00045499145680698007,
      "loss": 2.3235,
      "step": 75384
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.2598060369491577,
      "learning_rate": 0.00045498795443761505,
      "loss": 2.2603,
      "step": 75385
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.0954325199127197,
      "learning_rate": 0.00045498445203943477,
      "loss": 2.418,
      "step": 75386
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.7686559557914734,
      "learning_rate": 0.00045498094961243996,
      "loss": 2.1926,
      "step": 75387
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.9975945949554443,
      "learning_rate": 0.00045497744715663124,
      "loss": 2.16,
      "step": 75388
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.0224096775054932,
      "learning_rate": 0.0004549739446720093,
      "loss": 2.1177,
      "step": 75389
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.8413086533546448,
      "learning_rate": 0.00045497044215857485,
      "loss": 2.231,
      "step": 75390
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.8558236956596375,
      "learning_rate": 0.00045496693961632844,
      "loss": 2.0592,
      "step": 75391
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.0166548490524292,
      "learning_rate": 0.00045496343704527076,
      "loss": 2.0487,
      "step": 75392
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.870145320892334,
      "learning_rate": 0.00045495993444540236,
      "loss": 2.2686,
      "step": 75393
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.8096702694892883,
      "learning_rate": 0.00045495643181672405,
      "loss": 2.207,
      "step": 75394
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.945588231086731,
      "learning_rate": 0.00045495292915923637,
      "loss": 2.3307,
      "step": 75395
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.936857283115387,
      "learning_rate": 0.00045494942647294004,
      "loss": 2.141,
      "step": 75396
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.7785888314247131,
      "learning_rate": 0.0004549459237578358,
      "loss": 1.9746,
      "step": 75397
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.7513319253921509,
      "learning_rate": 0.0004549424210139241,
      "loss": 2.303,
      "step": 75398
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.9031614661216736,
      "learning_rate": 0.0004549389182412056,
      "loss": 2.1855,
      "step": 75399
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.9143140316009521,
      "learning_rate": 0.0004549354154396811,
      "loss": 2.111,
      "step": 75400
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.7977038621902466,
      "learning_rate": 0.0004549319126093512,
      "loss": 2.2287,
      "step": 75401
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.8493748903274536,
      "learning_rate": 0.0004549284097502165,
      "loss": 2.2125,
      "step": 75402
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.9340327382087708,
      "learning_rate": 0.0004549249068622777,
      "loss": 2.3591,
      "step": 75403
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.7702457308769226,
      "learning_rate": 0.00045492140394553544,
      "loss": 2.2321,
      "step": 75404
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.7732565402984619,
      "learning_rate": 0.0004549179009999903,
      "loss": 2.4293,
      "step": 75405
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.8520650267601013,
      "learning_rate": 0.00045491439802564315,
      "loss": 2.4577,
      "step": 75406
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.1030213832855225,
      "learning_rate": 0.0004549108950224944,
      "loss": 2.3701,
      "step": 75407
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.9099810123443604,
      "learning_rate": 0.00045490739199054473,
      "loss": 2.0528,
      "step": 75408
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.8123637437820435,
      "learning_rate": 0.00045490388892979494,
      "loss": 2.1413,
      "step": 75409
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.9066258072853088,
      "learning_rate": 0.00045490038584024555,
      "loss": 2.1344,
      "step": 75410
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.0896605253219604,
      "learning_rate": 0.00045489688272189725,
      "loss": 2.1698,
      "step": 75411
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.8546644449234009,
      "learning_rate": 0.00045489337957475075,
      "loss": 2.1491,
      "step": 75412
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.7459810376167297,
      "learning_rate": 0.0004548898763988066,
      "loss": 1.881,
      "step": 75413
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.8838063478469849,
      "learning_rate": 0.0004548863731940655,
      "loss": 2.2901,
      "step": 75414
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.8590091466903687,
      "learning_rate": 0.00045488286996052814,
      "loss": 2.3224,
      "step": 75415
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.9121777415275574,
      "learning_rate": 0.0004548793666981951,
      "loss": 2.2863,
      "step": 75416
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.8013924956321716,
      "learning_rate": 0.000454875863407067,
      "loss": 2.2487,
      "step": 75417
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.839636504650116,
      "learning_rate": 0.00045487236008714466,
      "loss": 2.2345,
      "step": 75418
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.9059805870056152,
      "learning_rate": 0.0004548688567384286,
      "loss": 2.2157,
      "step": 75419
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.7438520193099976,
      "learning_rate": 0.00045486535336091945,
      "loss": 2.0235,
      "step": 75420
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.731886625289917,
      "learning_rate": 0.00045486184995461784,
      "loss": 2.249,
      "step": 75421
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.7681344747543335,
      "learning_rate": 0.0004548583465195246,
      "loss": 2.1393,
      "step": 75422
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.000397801399231,
      "learning_rate": 0.0004548548430556403,
      "loss": 2.3961,
      "step": 75423
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.8444449305534363,
      "learning_rate": 0.0004548513395629655,
      "loss": 2.1152,
      "step": 75424
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.7287932634353638,
      "learning_rate": 0.000454847836041501,
      "loss": 1.8921,
      "step": 75425
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.981001079082489,
      "learning_rate": 0.0004548443324912472,
      "loss": 2.1493,
      "step": 75426
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.0657720565795898,
      "learning_rate": 0.00045484082891220496,
      "loss": 1.9661,
      "step": 75427
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.8589581847190857,
      "learning_rate": 0.000454837325304375,
      "loss": 2.2398,
      "step": 75428
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.9729194045066833,
      "learning_rate": 0.00045483382166775773,
      "loss": 2.1957,
      "step": 75429
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.8430814743041992,
      "learning_rate": 0.0004548303180023541,
      "loss": 2.0397,
      "step": 75430
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.7656830549240112,
      "learning_rate": 0.0004548268143081644,
      "loss": 2.371,
      "step": 75431
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.8426516056060791,
      "learning_rate": 0.00045482331058518954,
      "loss": 2.304,
      "step": 75432
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.7276582717895508,
      "learning_rate": 0.00045481980683343017,
      "loss": 2.3757,
      "step": 75433
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.9141827821731567,
      "learning_rate": 0.0004548163030528868,
      "loss": 2.1907,
      "step": 75434
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.763089656829834,
      "learning_rate": 0.0004548127992435602,
      "loss": 2.1789,
      "step": 75435
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.7942328453063965,
      "learning_rate": 0.00045480929540545094,
      "loss": 2.3462,
      "step": 75436
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.7450486421585083,
      "learning_rate": 0.00045480579153855976,
      "loss": 2.3051,
      "step": 75437
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.8857292532920837,
      "learning_rate": 0.00045480228764288724,
      "loss": 2.166,
      "step": 75438
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.8207176923751831,
      "learning_rate": 0.00045479878371843397,
      "loss": 2.0766,
      "step": 75439
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.7238662838935852,
      "learning_rate": 0.0004547952797652008,
      "loss": 2.3051,
      "step": 75440
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.9423372745513916,
      "learning_rate": 0.0004547917757831882,
      "loss": 2.199,
      "step": 75441
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.7488207221031189,
      "learning_rate": 0.0004547882717723969,
      "loss": 2.3014,
      "step": 75442
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.7782965898513794,
      "learning_rate": 0.0004547847677328276,
      "loss": 2.3772,
      "step": 75443
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.7020297050476074,
      "learning_rate": 0.00045478126366448073,
      "loss": 2.2218,
      "step": 75444
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.7958512306213379,
      "learning_rate": 0.0004547777595673572,
      "loss": 2.2859,
      "step": 75445
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.8440797924995422,
      "learning_rate": 0.00045477425544145767,
      "loss": 2.1886,
      "step": 75446
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.7382197976112366,
      "learning_rate": 0.0004547707512867826,
      "loss": 2.2535,
      "step": 75447
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.7797465920448303,
      "learning_rate": 0.0004547672471033326,
      "loss": 2.1101,
      "step": 75448
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.7448374032974243,
      "learning_rate": 0.0004547637428911086,
      "loss": 2.3308,
      "step": 75449
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.8065468072891235,
      "learning_rate": 0.0004547602386501111,
      "loss": 2.3354,
      "step": 75450
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.9068818092346191,
      "learning_rate": 0.00045475673438034067,
      "loss": 2.1606,
      "step": 75451
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.9687321782112122,
      "learning_rate": 0.0004547532300817981,
      "loss": 2.2437,
      "step": 75452
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.8286092281341553,
      "learning_rate": 0.000454749725754484,
      "loss": 2.1968,
      "step": 75453
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.8480024933815002,
      "learning_rate": 0.000454746221398399,
      "loss": 2.2815,
      "step": 75454
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.8080662488937378,
      "learning_rate": 0.0004547427170135437,
      "loss": 2.1014,
      "step": 75455
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.8422929048538208,
      "learning_rate": 0.00045473921259991885,
      "loss": 2.4286,
      "step": 75456
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.8586357831954956,
      "learning_rate": 0.00045473570815752497,
      "loss": 2.372,
      "step": 75457
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.6998485326766968,
      "learning_rate": 0.00045473220368636287,
      "loss": 2.1431,
      "step": 75458
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.795823872089386,
      "learning_rate": 0.0004547286991864332,
      "loss": 2.1814,
      "step": 75459
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.7902646064758301,
      "learning_rate": 0.0004547251946577364,
      "loss": 2.1143,
      "step": 75460
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.7282448410987854,
      "learning_rate": 0.0004547216901002734,
      "loss": 1.9605,
      "step": 75461
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.7558028697967529,
      "learning_rate": 0.0004547181855140446,
      "loss": 2.1412,
      "step": 75462
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.7338884472846985,
      "learning_rate": 0.0004547146808990508,
      "loss": 2.2161,
      "step": 75463
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.7619484066963196,
      "learning_rate": 0.00045471117625529275,
      "loss": 2.2492,
      "step": 75464
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.8046254515647888,
      "learning_rate": 0.0004547076715827709,
      "loss": 2.0615,
      "step": 75465
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.7520448565483093,
      "learning_rate": 0.00045470416688148585,
      "loss": 2.3394,
      "step": 75466
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.8578566312789917,
      "learning_rate": 0.00045470066215143855,
      "loss": 2.2376,
      "step": 75467
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.7644671201705933,
      "learning_rate": 0.00045469715739262937,
      "loss": 2.2453,
      "step": 75468
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.778439998626709,
      "learning_rate": 0.0004546936526050591,
      "loss": 2.3396,
      "step": 75469
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.7114161849021912,
      "learning_rate": 0.00045469014778872836,
      "loss": 1.9902,
      "step": 75470
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.7793073058128357,
      "learning_rate": 0.0004546866429436378,
      "loss": 2.3383,
      "step": 75471
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.9306659698486328,
      "learning_rate": 0.000454683138069788,
      "loss": 2.1804,
      "step": 75472
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.8385830521583557,
      "learning_rate": 0.00045467963316717986,
      "loss": 2.2401,
      "step": 75473
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.7798057794570923,
      "learning_rate": 0.00045467612823581373,
      "loss": 2.2178,
      "step": 75474
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.7953957915306091,
      "learning_rate": 0.0004546726232756904,
      "loss": 2.1665,
      "step": 75475
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.7679852843284607,
      "learning_rate": 0.0004546691182868105,
      "loss": 2.028,
      "step": 75476
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.7060919404029846,
      "learning_rate": 0.00045466561326917474,
      "loss": 2.2086,
      "step": 75477
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.8325225114822388,
      "learning_rate": 0.00045466210822278364,
      "loss": 2.2643,
      "step": 75478
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.8279874920845032,
      "learning_rate": 0.00045465860314763805,
      "loss": 2.3303,
      "step": 75479
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.7737913131713867,
      "learning_rate": 0.00045465509804373843,
      "loss": 2.3187,
      "step": 75480
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.8569996356964111,
      "learning_rate": 0.0004546515929110855,
      "loss": 2.2591,
      "step": 75481
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.876460611820221,
      "learning_rate": 0.00045464808774967996,
      "loss": 2.148,
      "step": 75482
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.7980623841285706,
      "learning_rate": 0.0004546445825595224,
      "loss": 2.3303,
      "step": 75483
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.7993522882461548,
      "learning_rate": 0.0004546410773406135,
      "loss": 2.065,
      "step": 75484
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.8933855295181274,
      "learning_rate": 0.0004546375720929539,
      "loss": 1.9796,
      "step": 75485
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.7786784172058105,
      "learning_rate": 0.00045463406681654425,
      "loss": 2.2859,
      "step": 75486
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.7880057692527771,
      "learning_rate": 0.00045463056151138516,
      "loss": 2.2618,
      "step": 75487
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.7647138237953186,
      "learning_rate": 0.0004546270561774775,
      "loss": 2.1732,
      "step": 75488
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.8803665041923523,
      "learning_rate": 0.0004546235508148216,
      "loss": 2.3717,
      "step": 75489
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.7157773375511169,
      "learning_rate": 0.00045462004542341826,
      "loss": 2.2826,
      "step": 75490
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.8191563487052917,
      "learning_rate": 0.0004546165400032682,
      "loss": 2.0903,
      "step": 75491
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.7508739233016968,
      "learning_rate": 0.000454613034554372,
      "loss": 2.362,
      "step": 75492
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.769397497177124,
      "learning_rate": 0.00045460952907673026,
      "loss": 2.1578,
      "step": 75493
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.8484178185462952,
      "learning_rate": 0.00045460602357034384,
      "loss": 2.0805,
      "step": 75494
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.9073370695114136,
      "learning_rate": 0.00045460251803521306,
      "loss": 2.1049,
      "step": 75495
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.7975021600723267,
      "learning_rate": 0.0004545990124713388,
      "loss": 2.1666,
      "step": 75496
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.8573504686355591,
      "learning_rate": 0.00045459550687872174,
      "loss": 2.1802,
      "step": 75497
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.8524354696273804,
      "learning_rate": 0.0004545920012573624,
      "loss": 2.0082,
      "step": 75498
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.9658421874046326,
      "learning_rate": 0.00045458849560726147,
      "loss": 2.3895,
      "step": 75499
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.9330065846443176,
      "learning_rate": 0.00045458498992841963,
      "loss": 2.0611,
      "step": 75500
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.8820298314094543,
      "learning_rate": 0.00045458148422083754,
      "loss": 2.1864,
      "step": 75501
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.0620890855789185,
      "learning_rate": 0.00045457797848451586,
      "loss": 2.3758,
      "step": 75502
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.9763249158859253,
      "learning_rate": 0.0004545744727194551,
      "loss": 2.1466,
      "step": 75503
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.8608149290084839,
      "learning_rate": 0.00045457096692565625,
      "loss": 2.2099,
      "step": 75504
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.9086683988571167,
      "learning_rate": 0.00045456746110311957,
      "loss": 2.0688,
      "step": 75505
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.8669680953025818,
      "learning_rate": 0.0004545639552518459,
      "loss": 2.1189,
      "step": 75506
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.8482881188392639,
      "learning_rate": 0.0004545604493718359,
      "loss": 2.0946,
      "step": 75507
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.7760517597198486,
      "learning_rate": 0.00045455694346309015,
      "loss": 2.3687,
      "step": 75508
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.9135133028030396,
      "learning_rate": 0.0004545534375256094,
      "loss": 2.1943,
      "step": 75509
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.9560078382492065,
      "learning_rate": 0.0004545499315593943,
      "loss": 2.0135,
      "step": 75510
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.9565097689628601,
      "learning_rate": 0.0004545464255644453,
      "loss": 2.3348,
      "step": 75511
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.9043368101119995,
      "learning_rate": 0.00045454291954076325,
      "loss": 2.1905,
      "step": 75512
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.9263372421264648,
      "learning_rate": 0.00045453941348834887,
      "loss": 2.4145,
      "step": 75513
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.9197166562080383,
      "learning_rate": 0.0004545359074072026,
      "loss": 2.1154,
      "step": 75514
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.8172571063041687,
      "learning_rate": 0.0004545324012973252,
      "loss": 2.2172,
      "step": 75515
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.7591024041175842,
      "learning_rate": 0.00045452889515871733,
      "loss": 2.1771,
      "step": 75516
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.8659564256668091,
      "learning_rate": 0.0004545253889913796,
      "loss": 2.4084,
      "step": 75517
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.8307701945304871,
      "learning_rate": 0.0004545218827953127,
      "loss": 2.4192,
      "step": 75518
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.7075379490852356,
      "learning_rate": 0.00045451837657051723,
      "loss": 2.0614,
      "step": 75519
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.7666940689086914,
      "learning_rate": 0.0004545148703169939,
      "loss": 2.0232,
      "step": 75520
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.999354362487793,
      "learning_rate": 0.0004545113640347434,
      "loss": 2.1769,
      "step": 75521
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.8312125205993652,
      "learning_rate": 0.0004545078577237662,
      "loss": 2.2746,
      "step": 75522
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.7592967748641968,
      "learning_rate": 0.00045450435138406307,
      "loss": 2.3395,
      "step": 75523
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.7453472018241882,
      "learning_rate": 0.0004545008450156347,
      "loss": 2.288,
      "step": 75524
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.8102617859840393,
      "learning_rate": 0.0004544973386184818,
      "loss": 2.1286,
      "step": 75525
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.7249459028244019,
      "learning_rate": 0.00045449383219260485,
      "loss": 2.0986,
      "step": 75526
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.7549824714660645,
      "learning_rate": 0.00045449032573800454,
      "loss": 2.2519,
      "step": 75527
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.7890201807022095,
      "learning_rate": 0.00045448681925468166,
      "loss": 1.9264,
      "step": 75528
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.0811177492141724,
      "learning_rate": 0.00045448331274263674,
      "loss": 2.3457,
      "step": 75529
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.7984861731529236,
      "learning_rate": 0.0004544798062018704,
      "loss": 2.398,
      "step": 75530
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.8314033150672913,
      "learning_rate": 0.0004544762996323834,
      "loss": 2.1651,
      "step": 75531
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.9477945566177368,
      "learning_rate": 0.0004544727930341763,
      "loss": 2.0989,
      "step": 75532
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.9499620795249939,
      "learning_rate": 0.00045446928640724985,
      "loss": 2.2248,
      "step": 75533
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.8293306231498718,
      "learning_rate": 0.0004544657797516046,
      "loss": 2.0482,
      "step": 75534
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.133989930152893,
      "learning_rate": 0.0004544622730672412,
      "loss": 2.2785,
      "step": 75535
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.940582275390625,
      "learning_rate": 0.0004544587663541604,
      "loss": 2.1101,
      "step": 75536
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.7231612801551819,
      "learning_rate": 0.00045445525961236284,
      "loss": 2.162,
      "step": 75537
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.9149842858314514,
      "learning_rate": 0.00045445175284184906,
      "loss": 2.1676,
      "step": 75538
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.8555485010147095,
      "learning_rate": 0.0004544482460426198,
      "loss": 2.2532,
      "step": 75539
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.8276995420455933,
      "learning_rate": 0.0004544447392146758,
      "loss": 2.1185,
      "step": 75540
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.7589731216430664,
      "learning_rate": 0.0004544412323580175,
      "loss": 2.1264,
      "step": 75541
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.7965825796127319,
      "learning_rate": 0.0004544377254726457,
      "loss": 2.2086,
      "step": 75542
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.8579277396202087,
      "learning_rate": 0.00045443421855856093,
      "loss": 2.16,
      "step": 75543
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.8814387321472168,
      "learning_rate": 0.000454430711615764,
      "loss": 2.3194,
      "step": 75544
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.8266739845275879,
      "learning_rate": 0.00045442720464425546,
      "loss": 2.3029,
      "step": 75545
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.738725483417511,
      "learning_rate": 0.00045442369764403604,
      "loss": 2.1942,
      "step": 75546
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.9498921632766724,
      "learning_rate": 0.00045442019061510625,
      "loss": 2.0912,
      "step": 75547
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.8533241748809814,
      "learning_rate": 0.0004544166835574669,
      "loss": 2.1608,
      "step": 75548
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.7698138356208801,
      "learning_rate": 0.00045441317647111856,
      "loss": 2.0724,
      "step": 75549
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.7278416752815247,
      "learning_rate": 0.00045440966935606186,
      "loss": 2.1237,
      "step": 75550
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.9573795795440674,
      "learning_rate": 0.0004544061622122975,
      "loss": 2.3585,
      "step": 75551
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.841026246547699,
      "learning_rate": 0.0004544026550398261,
      "loss": 1.9672,
      "step": 75552
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.820601224899292,
      "learning_rate": 0.0004543991478386484,
      "loss": 2.2964,
      "step": 75553
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.9962354898452759,
      "learning_rate": 0.00045439564060876494,
      "loss": 2.2484,
      "step": 75554
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.0265979766845703,
      "learning_rate": 0.00045439213335017646,
      "loss": 2.2522,
      "step": 75555
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.6789458394050598,
      "learning_rate": 0.0004543886260628835,
      "loss": 2.1423,
      "step": 75556
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.0046274662017822,
      "learning_rate": 0.00045438511874688686,
      "loss": 2.3095,
      "step": 75557
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.9014347195625305,
      "learning_rate": 0.0004543816114021871,
      "loss": 2.2027,
      "step": 75558
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.7966256141662598,
      "learning_rate": 0.0004543781040287848,
      "loss": 2.0314,
      "step": 75559
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.8432251811027527,
      "learning_rate": 0.0004543745966266807,
      "loss": 2.2457,
      "step": 75560
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.8067324161529541,
      "learning_rate": 0.0004543710891958756,
      "loss": 2.2007,
      "step": 75561
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.7364682555198669,
      "learning_rate": 0.00045436758173636986,
      "loss": 2.1098,
      "step": 75562
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.8124528527259827,
      "learning_rate": 0.00045436407424816433,
      "loss": 1.9475,
      "step": 75563
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.8151012659072876,
      "learning_rate": 0.00045436056673125953,
      "loss": 2.1159,
      "step": 75564
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.9463785290718079,
      "learning_rate": 0.00045435705918565623,
      "loss": 2.0261,
      "step": 75565
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.8969033360481262,
      "learning_rate": 0.00045435355161135507,
      "loss": 2.3878,
      "step": 75566
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.7878536581993103,
      "learning_rate": 0.00045435004400835665,
      "loss": 2.0936,
      "step": 75567
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.8782626390457153,
      "learning_rate": 0.0004543465363766617,
      "loss": 2.1047,
      "step": 75568
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.0435512065887451,
      "learning_rate": 0.00045434302871627075,
      "loss": 2.3851,
      "step": 75569
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.021220088005066,
      "learning_rate": 0.00045433952102718445,
      "loss": 2.3335,
      "step": 75570
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.7085371613502502,
      "learning_rate": 0.0004543360133094036,
      "loss": 2.2341,
      "step": 75571
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.9494747519493103,
      "learning_rate": 0.0004543325055629288,
      "loss": 2.0176,
      "step": 75572
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.2529449462890625,
      "learning_rate": 0.0004543289977877606,
      "loss": 2.0541,
      "step": 75573
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.8015288710594177,
      "learning_rate": 0.00045432548998389986,
      "loss": 1.9979,
      "step": 75574
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.7865855693817139,
      "learning_rate": 0.00045432198215134695,
      "loss": 2.2871,
      "step": 75575
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.8125523924827576,
      "learning_rate": 0.00045431847429010274,
      "loss": 2.1143,
      "step": 75576
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.8051283955574036,
      "learning_rate": 0.0004543149664001678,
      "loss": 2.0957,
      "step": 75577
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.2141718864440918,
      "learning_rate": 0.0004543114584815428,
      "loss": 2.123,
      "step": 75578
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.8398020267486572,
      "learning_rate": 0.00045430795053422835,
      "loss": 2.151,
      "step": 75579
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.8282028436660767,
      "learning_rate": 0.0004543044425582252,
      "loss": 2.091,
      "step": 75580
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.03118896484375,
      "learning_rate": 0.00045430093455353394,
      "loss": 2.1842,
      "step": 75581
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.8656408190727234,
      "learning_rate": 0.0004542974265201552,
      "loss": 2.3193,
      "step": 75582
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.8305927515029907,
      "learning_rate": 0.0004542939184580896,
      "loss": 2.1929,
      "step": 75583
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.8724633455276489,
      "learning_rate": 0.0004542904103673379,
      "loss": 2.1405,
      "step": 75584
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.0019645690917969,
      "learning_rate": 0.0004542869022479007,
      "loss": 2.367,
      "step": 75585
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.7973591685295105,
      "learning_rate": 0.00045428339409977873,
      "loss": 2.3189,
      "step": 75586
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.7580308318138123,
      "learning_rate": 0.00045427988592297244,
      "loss": 2.3168,
      "step": 75587
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.8237376809120178,
      "learning_rate": 0.0004542763777174826,
      "loss": 2.0634,
      "step": 75588
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.8862783312797546,
      "learning_rate": 0.00045427286948331,
      "loss": 2.1326,
      "step": 75589
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.8061206340789795,
      "learning_rate": 0.0004542693612204551,
      "loss": 1.9931,
      "step": 75590
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.8209667205810547,
      "learning_rate": 0.00045426585292891855,
      "loss": 2.0487,
      "step": 75591
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.7715332508087158,
      "learning_rate": 0.00045426234460870115,
      "loss": 1.9962,
      "step": 75592
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.7374567985534668,
      "learning_rate": 0.0004542588362598034,
      "loss": 2.2791,
      "step": 75593
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.7493056654930115,
      "learning_rate": 0.000454255327882226,
      "loss": 2.374,
      "step": 75594
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.779617190361023,
      "learning_rate": 0.0004542518194759697,
      "loss": 2.2531,
      "step": 75595
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.7631903290748596,
      "learning_rate": 0.0004542483110410351,
      "loss": 2.2468,
      "step": 75596
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.7939856648445129,
      "learning_rate": 0.00045424480257742273,
      "loss": 2.1512,
      "step": 75597
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.8473299741744995,
      "learning_rate": 0.0004542412940851335,
      "loss": 2.2846,
      "step": 75598
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.7779426574707031,
      "learning_rate": 0.0004542377855641677,
      "loss": 2.2134,
      "step": 75599
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.7396396994590759,
      "learning_rate": 0.00045423427701452626,
      "loss": 2.1155,
      "step": 75600
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.1227755546569824,
      "learning_rate": 0.00045423076843620976,
      "loss": 2.3067,
      "step": 75601
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.7428512573242188,
      "learning_rate": 0.0004542272598292189,
      "loss": 2.0214,
      "step": 75602
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.9567478895187378,
      "learning_rate": 0.0004542237511935542,
      "loss": 2.2506,
      "step": 75603
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.739983081817627,
      "learning_rate": 0.00045422024252921646,
      "loss": 2.0633,
      "step": 75604
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.7760643362998962,
      "learning_rate": 0.00045421673383620623,
      "loss": 2.0948,
      "step": 75605
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.7746805548667908,
      "learning_rate": 0.0004542132251145242,
      "loss": 2.0744,
      "step": 75606
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.796883225440979,
      "learning_rate": 0.000454209716364171,
      "loss": 2.286,
      "step": 75607
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.7813528180122375,
      "learning_rate": 0.00045420620758514734,
      "loss": 2.2358,
      "step": 75608
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.8226494193077087,
      "learning_rate": 0.0004542026987774539,
      "loss": 2.2269,
      "step": 75609
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.770494282245636,
      "learning_rate": 0.00045419918994109114,
      "loss": 2.1316,
      "step": 75610
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.7645393013954163,
      "learning_rate": 0.00045419568107605987,
      "loss": 2.3766,
      "step": 75611
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.7945826053619385,
      "learning_rate": 0.0004541921721823607,
      "loss": 2.229,
      "step": 75612
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.8186100125312805,
      "learning_rate": 0.00045418866325999435,
      "loss": 2.2089,
      "step": 75613
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.7428172826766968,
      "learning_rate": 0.0004541851543089614,
      "loss": 2.194,
      "step": 75614
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.7769761681556702,
      "learning_rate": 0.0004541816453292625,
      "loss": 2.0839,
      "step": 75615
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.7622637748718262,
      "learning_rate": 0.0004541781363208984,
      "loss": 2.1577,
      "step": 75616
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.7551047801971436,
      "learning_rate": 0.00045417462728386955,
      "loss": 2.2298,
      "step": 75617
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.6970888376235962,
      "learning_rate": 0.0004541711182181768,
      "loss": 2.128,
      "step": 75618
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.8469902873039246,
      "learning_rate": 0.0004541676091238208,
      "loss": 2.2586,
      "step": 75619
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.794459342956543,
      "learning_rate": 0.000454164100000802,
      "loss": 2.2452,
      "step": 75620
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.842471182346344,
      "learning_rate": 0.0004541605908491212,
      "loss": 2.3767,
      "step": 75621
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.7996352910995483,
      "learning_rate": 0.00045415708166877906,
      "loss": 2.2585,
      "step": 75622
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.7827382683753967,
      "learning_rate": 0.00045415357245977625,
      "loss": 2.2185,
      "step": 75623
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.7808569669723511,
      "learning_rate": 0.00045415006322211337,
      "loss": 2.3332,
      "step": 75624
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.7465031147003174,
      "learning_rate": 0.00045414655395579107,
      "loss": 2.1327,
      "step": 75625
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.7289624214172363,
      "learning_rate": 0.00045414304466081,
      "loss": 1.9746,
      "step": 75626
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.8325449228286743,
      "learning_rate": 0.00045413953533717075,
      "loss": 2.2553,
      "step": 75627
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.8173056840896606,
      "learning_rate": 0.00045413602598487425,
      "loss": 2.4411,
      "step": 75628
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.7519657015800476,
      "learning_rate": 0.0004541325166039208,
      "loss": 2.1809,
      "step": 75629
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.7760642766952515,
      "learning_rate": 0.00045412900719431123,
      "loss": 2.1846,
      "step": 75630
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.8484119176864624,
      "learning_rate": 0.00045412549775604617,
      "loss": 2.1601,
      "step": 75631
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.7242778539657593,
      "learning_rate": 0.0004541219882891263,
      "loss": 2.1129,
      "step": 75632
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.7096695899963379,
      "learning_rate": 0.0004541184787935522,
      "loss": 2.142,
      "step": 75633
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.808891773223877,
      "learning_rate": 0.0004541149692693246,
      "loss": 2.0787,
      "step": 75634
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.8179898858070374,
      "learning_rate": 0.00045411145971644416,
      "loss": 2.2908,
      "step": 75635
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.8661585450172424,
      "learning_rate": 0.0004541079501349114,
      "loss": 2.2711,
      "step": 75636
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.8160679340362549,
      "learning_rate": 0.0004541044405247271,
      "loss": 2.3767,
      "step": 75637
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.8331442475318909,
      "learning_rate": 0.0004541009308858919,
      "loss": 2.0274,
      "step": 75638
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.9237504601478577,
      "learning_rate": 0.00045409742121840644,
      "loss": 2.1572,
      "step": 75639
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.7736001014709473,
      "learning_rate": 0.00045409391152227127,
      "loss": 2.2891,
      "step": 75640
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.8415489196777344,
      "learning_rate": 0.00045409040179748725,
      "loss": 2.2058,
      "step": 75641
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.9177013635635376,
      "learning_rate": 0.00045408689204405486,
      "loss": 2.2663,
      "step": 75642
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.8716902732849121,
      "learning_rate": 0.0004540833822619748,
      "loss": 2.1327,
      "step": 75643
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.7517408728599548,
      "learning_rate": 0.0004540798724512478,
      "loss": 2.1661,
      "step": 75644
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.7078774571418762,
      "learning_rate": 0.00045407636261187426,
      "loss": 2.2374,
      "step": 75645
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.9835681915283203,
      "learning_rate": 0.0004540728527438551,
      "loss": 2.3758,
      "step": 75646
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.7665276527404785,
      "learning_rate": 0.000454069342847191,
      "loss": 2.1183,
      "step": 75647
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.7693279981613159,
      "learning_rate": 0.0004540658329218825,
      "loss": 2.3069,
      "step": 75648
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.8000364899635315,
      "learning_rate": 0.00045406232296793005,
      "loss": 2.0381,
      "step": 75649
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.85185307264328,
      "learning_rate": 0.0004540588129853347,
      "loss": 1.8936,
      "step": 75650
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.7519306540489197,
      "learning_rate": 0.00045405530297409685,
      "loss": 2.3179,
      "step": 75651
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.8285380601882935,
      "learning_rate": 0.0004540517929342172,
      "loss": 2.3305,
      "step": 75652
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.7584308385848999,
      "learning_rate": 0.00045404828286569644,
      "loss": 2.4236,
      "step": 75653
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.8969349265098572,
      "learning_rate": 0.0004540447727685351,
      "loss": 2.2462,
      "step": 75654
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.8079909086227417,
      "learning_rate": 0.00045404126264273403,
      "loss": 2.1792,
      "step": 75655
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.7632575631141663,
      "learning_rate": 0.00045403775248829386,
      "loss": 2.1373,
      "step": 75656
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.7994547486305237,
      "learning_rate": 0.000454034242305215,
      "loss": 2.2798,
      "step": 75657
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.7716119885444641,
      "learning_rate": 0.00045403073209349825,
      "loss": 2.2488,
      "step": 75658
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.7851172089576721,
      "learning_rate": 0.00045402722185314444,
      "loss": 2.1529,
      "step": 75659
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.7665680646896362,
      "learning_rate": 0.0004540237115841539,
      "loss": 2.3618,
      "step": 75660
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.7893677949905396,
      "learning_rate": 0.00045402020128652756,
      "loss": 2.2358,
      "step": 75661
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.9273534417152405,
      "learning_rate": 0.00045401669096026596,
      "loss": 2.4506,
      "step": 75662
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.9357068538665771,
      "learning_rate": 0.00045401318060536965,
      "loss": 2.0658,
      "step": 75663
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.8714236617088318,
      "learning_rate": 0.00045400967022183946,
      "loss": 2.4917,
      "step": 75664
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.8173056244850159,
      "learning_rate": 0.000454006159809676,
      "loss": 2.1908,
      "step": 75665
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.804503858089447,
      "learning_rate": 0.0004540026493688798,
      "loss": 2.2841,
      "step": 75666
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.8270673751831055,
      "learning_rate": 0.0004539991388994516,
      "loss": 2.0972,
      "step": 75667
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.7289506793022156,
      "learning_rate": 0.0004539956284013921,
      "loss": 2.3115,
      "step": 75668
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.7486105561256409,
      "learning_rate": 0.0004539921178747019,
      "loss": 2.3019,
      "step": 75669
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.9626901149749756,
      "learning_rate": 0.0004539886073193817,
      "loss": 2.2801,
      "step": 75670
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.8250195384025574,
      "learning_rate": 0.000453985096735432,
      "loss": 2.0418,
      "step": 75671
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.8602458238601685,
      "learning_rate": 0.00045398158612285365,
      "loss": 1.9654,
      "step": 75672
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.8508990406990051,
      "learning_rate": 0.0004539780754816472,
      "loss": 2.202,
      "step": 75673
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.9045248627662659,
      "learning_rate": 0.0004539745648118133,
      "loss": 2.1025,
      "step": 75674
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.8655646443367004,
      "learning_rate": 0.00045397105411335264,
      "loss": 2.4379,
      "step": 75675
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.8462854027748108,
      "learning_rate": 0.00045396754338626583,
      "loss": 2.2928,
      "step": 75676
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.9078410267829895,
      "learning_rate": 0.0004539640326305536,
      "loss": 2.196,
      "step": 75677
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.8267627358436584,
      "learning_rate": 0.0004539605218462165,
      "loss": 2.0817,
      "step": 75678
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.7892180681228638,
      "learning_rate": 0.00045395701103325526,
      "loss": 2.3385,
      "step": 75679
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.9246447086334229,
      "learning_rate": 0.00045395350019167047,
      "loss": 2.1774,
      "step": 75680
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.9037793874740601,
      "learning_rate": 0.00045394998932146285,
      "loss": 2.2578,
      "step": 75681
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.8056540489196777,
      "learning_rate": 0.000453946478422633,
      "loss": 2.0546,
      "step": 75682
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.7924754023551941,
      "learning_rate": 0.00045394296749518166,
      "loss": 2.083,
      "step": 75683
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.8092784285545349,
      "learning_rate": 0.00045393945653910935,
      "loss": 2.4458,
      "step": 75684
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.9323078989982605,
      "learning_rate": 0.0004539359455544168,
      "loss": 2.1677,
      "step": 75685
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.006044864654541,
      "learning_rate": 0.0004539324345411046,
      "loss": 2.3485,
      "step": 75686
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.7802701592445374,
      "learning_rate": 0.0004539289234991736,
      "loss": 2.2343,
      "step": 75687
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.966493546962738,
      "learning_rate": 0.00045392541242862417,
      "loss": 1.9531,
      "step": 75688
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.693804144859314,
      "learning_rate": 0.0004539219013294571,
      "loss": 2.2076,
      "step": 75689
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.7876738905906677,
      "learning_rate": 0.00045391839020167315,
      "loss": 2.1154,
      "step": 75690
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.7224921584129333,
      "learning_rate": 0.0004539148790452728,
      "loss": 2.0541,
      "step": 75691
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.8526284694671631,
      "learning_rate": 0.0004539113678602568,
      "loss": 2.0468,
      "step": 75692
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.8462595343589783,
      "learning_rate": 0.0004539078566466257,
      "loss": 2.0099,
      "step": 75693
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.818376362323761,
      "learning_rate": 0.00045390434540438026,
      "loss": 2.3891,
      "step": 75694
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.8116321563720703,
      "learning_rate": 0.0004539008341335211,
      "loss": 2.2448,
      "step": 75695
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.7909775376319885,
      "learning_rate": 0.00045389732283404886,
      "loss": 2.3105,
      "step": 75696
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.8355305194854736,
      "learning_rate": 0.0004538938115059643,
      "loss": 2.1679,
      "step": 75697
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.8490499258041382,
      "learning_rate": 0.00045389030014926783,
      "loss": 2.3125,
      "step": 75698
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.7880921959877014,
      "learning_rate": 0.0004538867887639603,
      "loss": 2.3776,
      "step": 75699
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.8079793453216553,
      "learning_rate": 0.0004538832773500424,
      "loss": 2.1853,
      "step": 75700
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.7813659310340881,
      "learning_rate": 0.0004538797659075145,
      "loss": 2.1771,
      "step": 75701
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.8145244121551514,
      "learning_rate": 0.0004538762544363776,
      "loss": 2.1533,
      "step": 75702
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.7986326217651367,
      "learning_rate": 0.0004538727429366322,
      "loss": 2.2473,
      "step": 75703
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.8233194947242737,
      "learning_rate": 0.00045386923140827885,
      "loss": 2.1856,
      "step": 75704
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.8019412159919739,
      "learning_rate": 0.00045386571985131844,
      "loss": 2.4481,
      "step": 75705
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.8795783519744873,
      "learning_rate": 0.0004538622082657514,
      "loss": 2.2365,
      "step": 75706
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.7117651104927063,
      "learning_rate": 0.00045385869665157846,
      "loss": 2.1746,
      "step": 75707
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.814329206943512,
      "learning_rate": 0.0004538551850088004,
      "loss": 2.0655,
      "step": 75708
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.6923215389251709,
      "learning_rate": 0.00045385167333741763,
      "loss": 2.2255,
      "step": 75709
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.8423208594322205,
      "learning_rate": 0.00045384816163743095,
      "loss": 2.1083,
      "step": 75710
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.8082595467567444,
      "learning_rate": 0.0004538446499088411,
      "loss": 2.1759,
      "step": 75711
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.8540821075439453,
      "learning_rate": 0.00045384113815164845,
      "loss": 2.0788,
      "step": 75712
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.8954212665557861,
      "learning_rate": 0.0004538376263658539,
      "loss": 2.1701,
      "step": 75713
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.7820250988006592,
      "learning_rate": 0.0004538341145514581,
      "loss": 2.1771,
      "step": 75714
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.753244161605835,
      "learning_rate": 0.00045383060270846155,
      "loss": 2.4738,
      "step": 75715
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.7747724056243896,
      "learning_rate": 0.00045382709083686507,
      "loss": 2.1525,
      "step": 75716
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.7228472232818604,
      "learning_rate": 0.00045382357893666923,
      "loss": 2.3544,
      "step": 75717
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.8040812611579895,
      "learning_rate": 0.0004538200670078746,
      "loss": 2.4265,
      "step": 75718
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.7551747560501099,
      "learning_rate": 0.0004538165550504819,
      "loss": 2.2753,
      "step": 75719
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.817913293838501,
      "learning_rate": 0.0004538130430644919,
      "loss": 2.3432,
      "step": 75720
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.8346391320228577,
      "learning_rate": 0.0004538095310499051,
      "loss": 2.0286,
      "step": 75721
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.7922753095626831,
      "learning_rate": 0.00045380601900672215,
      "loss": 2.1134,
      "step": 75722
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.7949962615966797,
      "learning_rate": 0.00045380250693494384,
      "loss": 2.119,
      "step": 75723
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.7673748731613159,
      "learning_rate": 0.0004537989948345708,
      "loss": 2.2185,
      "step": 75724
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.9008060097694397,
      "learning_rate": 0.00045379548270560346,
      "loss": 2.2876,
      "step": 75725
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.8487427830696106,
      "learning_rate": 0.00045379197054804275,
      "loss": 2.1299,
      "step": 75726
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.7885342836380005,
      "learning_rate": 0.0004537884583618892,
      "loss": 2.3602,
      "step": 75727
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.8260474801063538,
      "learning_rate": 0.0004537849461471434,
      "loss": 2.065,
      "step": 75728
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.9454648494720459,
      "learning_rate": 0.00045378143390380617,
      "loss": 2.4686,
      "step": 75729
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.7381471991539001,
      "learning_rate": 0.000453777921631878,
      "loss": 2.1721,
      "step": 75730
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.9022406935691833,
      "learning_rate": 0.00045377440933135966,
      "loss": 2.0577,
      "step": 75731
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.7256866097450256,
      "learning_rate": 0.0004537708970022517,
      "loss": 2.2062,
      "step": 75732
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.0482537746429443,
      "learning_rate": 0.00045376738464455486,
      "loss": 2.1613,
      "step": 75733
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.7619469165802002,
      "learning_rate": 0.0004537638722582697,
      "loss": 2.1997,
      "step": 75734
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.8704280853271484,
      "learning_rate": 0.00045376035984339704,
      "loss": 2.4067,
      "step": 75735
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.8861064910888672,
      "learning_rate": 0.0004537568473999373,
      "loss": 2.3344,
      "step": 75736
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.8104379773139954,
      "learning_rate": 0.00045375333492789146,
      "loss": 2.1464,
      "step": 75737
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.7272995710372925,
      "learning_rate": 0.00045374982242725985,
      "loss": 2.1,
      "step": 75738
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.0299166440963745,
      "learning_rate": 0.0004537463098980432,
      "loss": 2.2483,
      "step": 75739
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.9153732657432556,
      "learning_rate": 0.00045374279734024226,
      "loss": 2.1855,
      "step": 75740
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.8139176368713379,
      "learning_rate": 0.00045373928475385763,
      "loss": 2.1381,
      "step": 75741
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.0454809665679932,
      "learning_rate": 0.00045373577213888997,
      "loss": 2.3217,
      "step": 75742
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.9962133765220642,
      "learning_rate": 0.0004537322594953399,
      "loss": 2.2822,
      "step": 75743
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.0848556756973267,
      "learning_rate": 0.0004537287468232081,
      "loss": 2.1011,
      "step": 75744
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.1058732271194458,
      "learning_rate": 0.00045372523412249525,
      "loss": 2.2164,
      "step": 75745
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.8689478039741516,
      "learning_rate": 0.000453721721393202,
      "loss": 2.4005,
      "step": 75746
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.7598288655281067,
      "learning_rate": 0.00045371820863532894,
      "loss": 2.2629,
      "step": 75747
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.9476047158241272,
      "learning_rate": 0.0004537146958488768,
      "loss": 2.1372,
      "step": 75748
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.8159341812133789,
      "learning_rate": 0.0004537111830338461,
      "loss": 2.1581,
      "step": 75749
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.7354353666305542,
      "learning_rate": 0.0004537076701902377,
      "loss": 2.196,
      "step": 75750
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.9235755205154419,
      "learning_rate": 0.00045370415731805217,
      "loss": 2.0006,
      "step": 75751
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.8845416903495789,
      "learning_rate": 0.00045370064441729,
      "loss": 2.247,
      "step": 75752
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.7824840545654297,
      "learning_rate": 0.00045369713148795207,
      "loss": 2.2627,
      "step": 75753
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.7910237908363342,
      "learning_rate": 0.00045369361853003895,
      "loss": 2.0906,
      "step": 75754
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.7407915592193604,
      "learning_rate": 0.00045369010554355125,
      "loss": 2.0911,
      "step": 75755
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.8773114085197449,
      "learning_rate": 0.00045368659252848967,
      "loss": 2.1104,
      "step": 75756
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.9573953747749329,
      "learning_rate": 0.0004536830794848548,
      "loss": 2.1014,
      "step": 75757
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.8462108373641968,
      "learning_rate": 0.0004536795664126474,
      "loss": 2.1865,
      "step": 75758
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.8101176619529724,
      "learning_rate": 0.0004536760533118681,
      "loss": 2.37,
      "step": 75759
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.9032514691352844,
      "learning_rate": 0.00045367254018251744,
      "loss": 2.2635,
      "step": 75760
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.8172879219055176,
      "learning_rate": 0.0004536690270245962,
      "loss": 2.1617,
      "step": 75761
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.8599136471748352,
      "learning_rate": 0.000453665513838105,
      "loss": 2.0947,
      "step": 75762
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.8724344372749329,
      "learning_rate": 0.00045366200062304444,
      "loss": 2.1809,
      "step": 75763
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.7424240112304688,
      "learning_rate": 0.0004536584873794152,
      "loss": 2.0322,
      "step": 75764
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.8945167064666748,
      "learning_rate": 0.000453654974107218,
      "loss": 2.2291,
      "step": 75765
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.8103078007698059,
      "learning_rate": 0.00045365146080645344,
      "loss": 2.1864,
      "step": 75766
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.8244835734367371,
      "learning_rate": 0.0004536479474771221,
      "loss": 2.3731,
      "step": 75767
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.8140875101089478,
      "learning_rate": 0.00045364443411922484,
      "loss": 2.2424,
      "step": 75768
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.7605607509613037,
      "learning_rate": 0.00045364092073276206,
      "loss": 2.1651,
      "step": 75769
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.7565231919288635,
      "learning_rate": 0.00045363740731773456,
      "loss": 2.0897,
      "step": 75770
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.7222540378570557,
      "learning_rate": 0.00045363389387414294,
      "loss": 2.0361,
      "step": 75771
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.7662003636360168,
      "learning_rate": 0.0004536303804019879,
      "loss": 2.1096,
      "step": 75772
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.7164955735206604,
      "learning_rate": 0.0004536268669012701,
      "loss": 2.1373,
      "step": 75773
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.7879960536956787,
      "learning_rate": 0.0004536233533719902,
      "loss": 2.1819,
      "step": 75774
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.7497960329055786,
      "learning_rate": 0.00045361983981414883,
      "loss": 2.2741,
      "step": 75775
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.8154563903808594,
      "learning_rate": 0.0004536163262277465,
      "loss": 2.2628,
      "step": 75776
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.7932795882225037,
      "learning_rate": 0.000453612812612784,
      "loss": 2.0837,
      "step": 75777
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.7761169672012329,
      "learning_rate": 0.00045360929896926216,
      "loss": 2.162,
      "step": 75778
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.7388694286346436,
      "learning_rate": 0.0004536057852971813,
      "loss": 2.1794,
      "step": 75779
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.7463194131851196,
      "learning_rate": 0.0004536022715965423,
      "loss": 2.2063,
      "step": 75780
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.8403204679489136,
      "learning_rate": 0.00045359875786734564,
      "loss": 2.2215,
      "step": 75781
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.780754566192627,
      "learning_rate": 0.0004535952441095922,
      "loss": 2.4789,
      "step": 75782
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.7808013558387756,
      "learning_rate": 0.0004535917303232824,
      "loss": 2.1488,
      "step": 75783
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.7674810290336609,
      "learning_rate": 0.00045358821650841704,
      "loss": 2.1156,
      "step": 75784
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.9862790107727051,
      "learning_rate": 0.0004535847026649967,
      "loss": 2.1854,
      "step": 75785
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.8071394562721252,
      "learning_rate": 0.0004535811887930221,
      "loss": 2.2857,
      "step": 75786
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.7378564476966858,
      "learning_rate": 0.0004535776748924938,
      "loss": 2.1043,
      "step": 75787
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.9307796359062195,
      "learning_rate": 0.00045357416096341257,
      "loss": 2.0303,
      "step": 75788
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.8419792056083679,
      "learning_rate": 0.0004535706470057791,
      "loss": 2.201,
      "step": 75789
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.8306241631507874,
      "learning_rate": 0.00045356713301959376,
      "loss": 2.2701,
      "step": 75790
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.048157811164856,
      "learning_rate": 0.0004535636190048575,
      "loss": 2.1542,
      "step": 75791
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.7346176505088806,
      "learning_rate": 0.0004535601049615708,
      "loss": 2.1028,
      "step": 75792
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.8733363747596741,
      "learning_rate": 0.0004535565908897345,
      "loss": 2.3287,
      "step": 75793
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.869367778301239,
      "learning_rate": 0.000453553076789349,
      "loss": 2.1149,
      "step": 75794
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.7564988136291504,
      "learning_rate": 0.0004535495626604151,
      "loss": 2.4304,
      "step": 75795
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.8173730969429016,
      "learning_rate": 0.00045354604850293353,
      "loss": 2.4528,
      "step": 75796
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.7419191598892212,
      "learning_rate": 0.00045354253431690477,
      "loss": 2.2401,
      "step": 75797
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.8102799654006958,
      "learning_rate": 0.00045353902010232956,
      "loss": 2.0306,
      "step": 75798
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.8906869292259216,
      "learning_rate": 0.0004535355058592086,
      "loss": 2.0432,
      "step": 75799
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.8127511739730835,
      "learning_rate": 0.00045353199158754246,
      "loss": 1.9767,
      "step": 75800
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.8200116157531738,
      "learning_rate": 0.0004535284772873318,
      "loss": 2.4191,
      "step": 75801
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.7556272745132446,
      "learning_rate": 0.00045352496295857737,
      "loss": 2.0395,
      "step": 75802
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.8002604842185974,
      "learning_rate": 0.0004535214486012797,
      "loss": 2.1043,
      "step": 75803
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.8255971074104309,
      "learning_rate": 0.00045351793421543945,
      "loss": 2.1788,
      "step": 75804
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.8435056209564209,
      "learning_rate": 0.0004535144198010574,
      "loss": 2.3807,
      "step": 75805
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.8234971761703491,
      "learning_rate": 0.0004535109053581341,
      "loss": 2.3421,
      "step": 75806
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.8574817180633545,
      "learning_rate": 0.0004535073908866701,
      "loss": 1.999,
      "step": 75807
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.724036693572998,
      "learning_rate": 0.0004535038763866664,
      "loss": 2.0749,
      "step": 75808
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.8079548478126526,
      "learning_rate": 0.0004535003618581233,
      "loss": 2.0186,
      "step": 75809
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.7839769721031189,
      "learning_rate": 0.00045349684730104157,
      "loss": 2.3672,
      "step": 75810
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.9100713729858398,
      "learning_rate": 0.00045349333271542196,
      "loss": 2.1794,
      "step": 75811
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.7255446910858154,
      "learning_rate": 0.00045348981810126496,
      "loss": 2.3676,
      "step": 75812
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.8117079734802246,
      "learning_rate": 0.00045348630345857133,
      "loss": 2.0392,
      "step": 75813
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.7640823125839233,
      "learning_rate": 0.00045348278878734176,
      "loss": 2.1066,
      "step": 75814
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.8823134899139404,
      "learning_rate": 0.00045347927408757676,
      "loss": 2.1793,
      "step": 75815
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.7254089117050171,
      "learning_rate": 0.0004534757593592771,
      "loss": 2.2683,
      "step": 75816
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.7991896271705627,
      "learning_rate": 0.0004534722446024434,
      "loss": 2.2528,
      "step": 75817
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.8283326625823975,
      "learning_rate": 0.0004534687298170763,
      "loss": 2.1188,
      "step": 75818
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.7718372941017151,
      "learning_rate": 0.00045346521500317646,
      "loss": 2.146,
      "step": 75819
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.7797533273696899,
      "learning_rate": 0.00045346170016074454,
      "loss": 2.167,
      "step": 75820
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.9289876222610474,
      "learning_rate": 0.0004534581852897812,
      "loss": 2.0512,
      "step": 75821
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.7782247066497803,
      "learning_rate": 0.0004534546703902872,
      "loss": 2.2176,
      "step": 75822
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.78232342004776,
      "learning_rate": 0.00045345115546226297,
      "loss": 2.3914,
      "step": 75823
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.8389057517051697,
      "learning_rate": 0.00045344764050570923,
      "loss": 2.143,
      "step": 75824
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.735392153263092,
      "learning_rate": 0.0004534441255206267,
      "loss": 2.1524,
      "step": 75825
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.7713492512702942,
      "learning_rate": 0.00045344061050701614,
      "loss": 2.0352,
      "step": 75826
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.797828197479248,
      "learning_rate": 0.000453437095464878,
      "loss": 2.2362,
      "step": 75827
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.7739273309707642,
      "learning_rate": 0.0004534335803942129,
      "loss": 2.3806,
      "step": 75828
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.7576951384544373,
      "learning_rate": 0.0004534300652950217,
      "loss": 1.9768,
      "step": 75829
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.7985027432441711,
      "learning_rate": 0.00045342655016730493,
      "loss": 2.2324,
      "step": 75830
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.749006986618042,
      "learning_rate": 0.0004534230350110633,
      "loss": 2.2753,
      "step": 75831
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.7773086428642273,
      "learning_rate": 0.0004534195198262974,
      "loss": 2.2258,
      "step": 75832
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.9252335429191589,
      "learning_rate": 0.00045341600461300794,
      "loss": 2.4738,
      "step": 75833
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.7886967658996582,
      "learning_rate": 0.0004534124893711955,
      "loss": 2.1742,
      "step": 75834
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.8594757914543152,
      "learning_rate": 0.00045340897410086077,
      "loss": 2.201,
      "step": 75835
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.7927383184432983,
      "learning_rate": 0.00045340545880200446,
      "loss": 2.0645,
      "step": 75836
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.7086806297302246,
      "learning_rate": 0.00045340194347462723,
      "loss": 2.193,
      "step": 75837
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.9727367758750916,
      "learning_rate": 0.0004533984281187295,
      "loss": 2.1697,
      "step": 75838
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.775704562664032,
      "learning_rate": 0.00045339491273431237,
      "loss": 2.1829,
      "step": 75839
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.7769156098365784,
      "learning_rate": 0.000453391397321376,
      "loss": 2.4016,
      "step": 75840
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.7838506698608398,
      "learning_rate": 0.0004533878818799214,
      "loss": 2.2192,
      "step": 75841
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.8147051930427551,
      "learning_rate": 0.0004533843664099491,
      "loss": 2.3647,
      "step": 75842
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.732705295085907,
      "learning_rate": 0.0004533808509114597,
      "loss": 2.2329,
      "step": 75843
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.764421820640564,
      "learning_rate": 0.0004533773353844539,
      "loss": 2.1162,
      "step": 75844
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.7468469142913818,
      "learning_rate": 0.0004533738198289324,
      "loss": 2.1547,
      "step": 75845
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.7980996370315552,
      "learning_rate": 0.0004533703042448958,
      "loss": 2.2047,
      "step": 75846
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.7644765377044678,
      "learning_rate": 0.0004533667886323447,
      "loss": 2.2656,
      "step": 75847
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.737209677696228,
      "learning_rate": 0.00045336327299128,
      "loss": 2.1506,
      "step": 75848
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.7683011889457703,
      "learning_rate": 0.000453359757321702,
      "loss": 2.2223,
      "step": 75849
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.7596307396888733,
      "learning_rate": 0.00045335624162361155,
      "loss": 2.3449,
      "step": 75850
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.7689234614372253,
      "learning_rate": 0.00045335272589700935,
      "loss": 2.1076,
      "step": 75851
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.8027809858322144,
      "learning_rate": 0.00045334921014189594,
      "loss": 2.113,
      "step": 75852
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.7809366583824158,
      "learning_rate": 0.00045334569435827196,
      "loss": 2.1922,
      "step": 75853
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.8245272040367126,
      "learning_rate": 0.00045334217854613824,
      "loss": 2.1367,
      "step": 75854
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.8084785342216492,
      "learning_rate": 0.00045333866270549526,
      "loss": 2.0221,
      "step": 75855
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.7496296167373657,
      "learning_rate": 0.0004533351468363436,
      "loss": 2.3714,
      "step": 75856
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.747067928314209,
      "learning_rate": 0.0004533316309386843,
      "loss": 2.3308,
      "step": 75857
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.8925220370292664,
      "learning_rate": 0.00045332811501251754,
      "loss": 2.3086,
      "step": 75858
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.8453806638717651,
      "learning_rate": 0.0004533245990578443,
      "loss": 2.2,
      "step": 75859
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.8253716826438904,
      "learning_rate": 0.00045332108307466506,
      "loss": 2.1575,
      "step": 75860
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.8328973054885864,
      "learning_rate": 0.00045331756706298057,
      "loss": 2.2491,
      "step": 75861
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.7419785857200623,
      "learning_rate": 0.0004533140510227914,
      "loss": 2.1148,
      "step": 75862
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.808523952960968,
      "learning_rate": 0.0004533105349540984,
      "loss": 2.1469,
      "step": 75863
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.809367835521698,
      "learning_rate": 0.00045330701885690195,
      "loss": 2.1292,
      "step": 75864
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.841414213180542,
      "learning_rate": 0.0004533035027312028,
      "loss": 2.2904,
      "step": 75865
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.775110125541687,
      "learning_rate": 0.0004532999865770017,
      "loss": 2.4471,
      "step": 75866
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.781208872795105,
      "learning_rate": 0.0004532964703942992,
      "loss": 2.171,
      "step": 75867
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.8377179503440857,
      "learning_rate": 0.0004532929541830961,
      "loss": 2.2454,
      "step": 75868
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.7624197602272034,
      "learning_rate": 0.0004532894379433928,
      "loss": 2.1332,
      "step": 75869
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.7807723879814148,
      "learning_rate": 0.0004532859216751902,
      "loss": 2.209,
      "step": 75870
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.7679661512374878,
      "learning_rate": 0.00045328240537848886,
      "loss": 2.09,
      "step": 75871
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.8032463192939758,
      "learning_rate": 0.00045327888905328937,
      "loss": 2.3464,
      "step": 75872
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.8441805243492126,
      "learning_rate": 0.00045327537269959243,
      "loss": 2.146,
      "step": 75873
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.7552595138549805,
      "learning_rate": 0.0004532718563173988,
      "loss": 2.027,
      "step": 75874
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.7907793521881104,
      "learning_rate": 0.00045326833990670895,
      "loss": 2.2188,
      "step": 75875
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.8055801391601562,
      "learning_rate": 0.00045326482346752365,
      "loss": 1.9233,
      "step": 75876
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.935727596282959,
      "learning_rate": 0.0004532613069998435,
      "loss": 2.3384,
      "step": 75877
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.9534308910369873,
      "learning_rate": 0.0004532577905036692,
      "loss": 2.1678,
      "step": 75878
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.7805917859077454,
      "learning_rate": 0.00045325427397900137,
      "loss": 2.1576,
      "step": 75879
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.9726474285125732,
      "learning_rate": 0.0004532507574258407,
      "loss": 1.9491,
      "step": 75880
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.9065505266189575,
      "learning_rate": 0.00045324724084418785,
      "loss": 2.0367,
      "step": 75881
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.766486406326294,
      "learning_rate": 0.00045324372423404343,
      "loss": 1.9916,
      "step": 75882
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.8248571753501892,
      "learning_rate": 0.0004532402075954081,
      "loss": 2.0993,
      "step": 75883
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.8737235069274902,
      "learning_rate": 0.0004532366909282825,
      "loss": 2.0707,
      "step": 75884
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.7867676019668579,
      "learning_rate": 0.0004532331742326673,
      "loss": 2.1179,
      "step": 75885
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.7507167458534241,
      "learning_rate": 0.00045322965750856315,
      "loss": 2.2534,
      "step": 75886
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.7438637018203735,
      "learning_rate": 0.0004532261407559708,
      "loss": 2.3136,
      "step": 75887
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.7304561734199524,
      "learning_rate": 0.0004532226239748907,
      "loss": 2.1277,
      "step": 75888
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.9179975390434265,
      "learning_rate": 0.0004532191071653237,
      "loss": 2.3425,
      "step": 75889
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.7689133286476135,
      "learning_rate": 0.0004532155903272704,
      "loss": 2.1708,
      "step": 75890
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.7977026104927063,
      "learning_rate": 0.0004532120734607314,
      "loss": 2.1012,
      "step": 75891
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.836641788482666,
      "learning_rate": 0.00045320855656570734,
      "loss": 2.4231,
      "step": 75892
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.752110481262207,
      "learning_rate": 0.000453205039642199,
      "loss": 2.1172,
      "step": 75893
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.7753775715827942,
      "learning_rate": 0.00045320152269020684,
      "loss": 2.043,
      "step": 75894
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.7580287456512451,
      "learning_rate": 0.00045319800570973177,
      "loss": 2.192,
      "step": 75895
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.8146043419837952,
      "learning_rate": 0.0004531944887007742,
      "loss": 2.1393,
      "step": 75896
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.7797629833221436,
      "learning_rate": 0.0004531909716633349,
      "loss": 2.3194,
      "step": 75897
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.7344207763671875,
      "learning_rate": 0.0004531874545974144,
      "loss": 2.082,
      "step": 75898
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.8131853938102722,
      "learning_rate": 0.00045318393750301366,
      "loss": 2.2456,
      "step": 75899
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.7944907546043396,
      "learning_rate": 0.00045318042038013307,
      "loss": 2.3027,
      "step": 75900
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.7457627654075623,
      "learning_rate": 0.0004531769032287732,
      "loss": 2.3071,
      "step": 75901
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.7461745142936707,
      "learning_rate": 0.000453173386048935,
      "loss": 2.1747,
      "step": 75902
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.7684109807014465,
      "learning_rate": 0.000453169868840619,
      "loss": 2.0278,
      "step": 75903
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.7226699590682983,
      "learning_rate": 0.00045316635160382577,
      "loss": 2.0888,
      "step": 75904
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.83021080493927,
      "learning_rate": 0.000453162834338556,
      "loss": 2.2192,
      "step": 75905
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.8384027481079102,
      "learning_rate": 0.0004531593170448104,
      "loss": 2.3117,
      "step": 75906
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.7422871589660645,
      "learning_rate": 0.0004531557997225896,
      "loss": 2.2899,
      "step": 75907
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.8988266587257385,
      "learning_rate": 0.0004531522823718942,
      "loss": 2.3377,
      "step": 75908
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.7483640909194946,
      "learning_rate": 0.00045314876499272504,
      "loss": 2.1023,
      "step": 75909
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.803992748260498,
      "learning_rate": 0.0004531452475850825,
      "loss": 2.1465,
      "step": 75910
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.7817906737327576,
      "learning_rate": 0.0004531417301489674,
      "loss": 2.08,
      "step": 75911
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.9057872295379639,
      "learning_rate": 0.00045313821268438037,
      "loss": 2.2332,
      "step": 75912
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.7410874366760254,
      "learning_rate": 0.000453134695191322,
      "loss": 1.9223,
      "step": 75913
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.7594957947731018,
      "learning_rate": 0.0004531311776697931,
      "loss": 2.0427,
      "step": 75914
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.8915207386016846,
      "learning_rate": 0.00045312766011979417,
      "loss": 2.0267,
      "step": 75915
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.7738199830055237,
      "learning_rate": 0.00045312414254132586,
      "loss": 2.2816,
      "step": 75916
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.8445528149604797,
      "learning_rate": 0.0004531206249343889,
      "loss": 2.1721,
      "step": 75917
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.7823426723480225,
      "learning_rate": 0.0004531171072989841,
      "loss": 2.1714,
      "step": 75918
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.7964328527450562,
      "learning_rate": 0.0004531135896351118,
      "loss": 2.0824,
      "step": 75919
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.7955349683761597,
      "learning_rate": 0.00045311007194277276,
      "loss": 2.3872,
      "step": 75920
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.7420012950897217,
      "learning_rate": 0.0004531065542219677,
      "loss": 2.3377,
      "step": 75921
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.7367303967475891,
      "learning_rate": 0.00045310303647269726,
      "loss": 1.976,
      "step": 75922
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.9347630143165588,
      "learning_rate": 0.0004530995186949621,
      "loss": 2.1002,
      "step": 75923
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.76657634973526,
      "learning_rate": 0.0004530960008887628,
      "loss": 2.2939,
      "step": 75924
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.8713091015815735,
      "learning_rate": 0.0004530924830541,
      "loss": 2.0412,
      "step": 75925
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.9285836815834045,
      "learning_rate": 0.0004530889651909745,
      "loss": 2.1054,
      "step": 75926
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.7753876447677612,
      "learning_rate": 0.0004530854472993869,
      "loss": 2.1737,
      "step": 75927
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.8740327954292297,
      "learning_rate": 0.0004530819293793378,
      "loss": 2.3266,
      "step": 75928
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.8958595991134644,
      "learning_rate": 0.0004530784114308278,
      "loss": 2.407,
      "step": 75929
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.8001911044120789,
      "learning_rate": 0.00045307489345385775,
      "loss": 2.1409,
      "step": 75930
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.8898096084594727,
      "learning_rate": 0.0004530713754484282,
      "loss": 2.3489,
      "step": 75931
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.8242273330688477,
      "learning_rate": 0.0004530678574145396,
      "loss": 2.3105,
      "step": 75932
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.7926715016365051,
      "learning_rate": 0.000453064339352193,
      "loss": 2.389,
      "step": 75933
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.8998869061470032,
      "learning_rate": 0.0004530608212613888,
      "loss": 2.0656,
      "step": 75934
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.78995680809021,
      "learning_rate": 0.00045305730314212765,
      "loss": 1.9943,
      "step": 75935
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.8835586905479431,
      "learning_rate": 0.0004530537849944103,
      "loss": 2.4908,
      "step": 75936
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.745395302772522,
      "learning_rate": 0.00045305026681823735,
      "loss": 2.4443,
      "step": 75937
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.8453027606010437,
      "learning_rate": 0.00045304674861360944,
      "loss": 2.4142,
      "step": 75938
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.8422925472259521,
      "learning_rate": 0.00045304323038052735,
      "loss": 2.1832,
      "step": 75939
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.0290205478668213,
      "learning_rate": 0.0004530397121189915,
      "loss": 2.332,
      "step": 75940
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.7916216850280762,
      "learning_rate": 0.00045303619382900274,
      "loss": 2.1745,
      "step": 75941
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.8723132610321045,
      "learning_rate": 0.00045303267551056173,
      "loss": 2.2305,
      "step": 75942
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.9111899137496948,
      "learning_rate": 0.000453029157163669,
      "loss": 2.3953,
      "step": 75943
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.7426512241363525,
      "learning_rate": 0.0004530256387883252,
      "loss": 2.2842,
      "step": 75944
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.7505838871002197,
      "learning_rate": 0.0004530221203845311,
      "loss": 2.1502,
      "step": 75945
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.0710705518722534,
      "learning_rate": 0.0004530186019522873,
      "loss": 2.1491,
      "step": 75946
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.8394708037376404,
      "learning_rate": 0.0004530150834915944,
      "loss": 2.1271,
      "step": 75947
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.7544143795967102,
      "learning_rate": 0.0004530115650024532,
      "loss": 2.2153,
      "step": 75948
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.8785556554794312,
      "learning_rate": 0.0004530080464848642,
      "loss": 2.3191,
      "step": 75949
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.8390508890151978,
      "learning_rate": 0.0004530045279388282,
      "loss": 2.2059,
      "step": 75950
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.8656548261642456,
      "learning_rate": 0.00045300100936434575,
      "loss": 2.168,
      "step": 75951
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.9485811591148376,
      "learning_rate": 0.00045299749076141736,
      "loss": 2.1306,
      "step": 75952
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.7332674264907837,
      "learning_rate": 0.00045299397213004406,
      "loss": 2.2208,
      "step": 75953
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.8644839525222778,
      "learning_rate": 0.0004529904534702261,
      "loss": 2.1533,
      "step": 75954
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.7689575552940369,
      "learning_rate": 0.0004529869347819645,
      "loss": 2.4014,
      "step": 75955
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.8191704154014587,
      "learning_rate": 0.0004529834160652596,
      "loss": 2.1408,
      "step": 75956
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.8725914359092712,
      "learning_rate": 0.00045297989732011226,
      "loss": 2.2548,
      "step": 75957
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.8785659074783325,
      "learning_rate": 0.0004529763785465231,
      "loss": 2.2626,
      "step": 75958
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.7188734412193298,
      "learning_rate": 0.0004529728597444928,
      "loss": 2.1334,
      "step": 75959
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.8005038499832153,
      "learning_rate": 0.0004529693409140218,
      "loss": 2.3239,
      "step": 75960
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.7417566180229187,
      "learning_rate": 0.000452965822055111,
      "loss": 2.218,
      "step": 75961
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.8734518885612488,
      "learning_rate": 0.0004529623031677609,
      "loss": 2.0843,
      "step": 75962
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.7234074473381042,
      "learning_rate": 0.00045295878425197233,
      "loss": 2.1328,
      "step": 75963
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.8232898712158203,
      "learning_rate": 0.0004529552653077458,
      "loss": 2.0834,
      "step": 75964
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.7275545597076416,
      "learning_rate": 0.0004529517463350819,
      "loss": 2.1195,
      "step": 75965
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.7508488893508911,
      "learning_rate": 0.00045294822733398145,
      "loss": 2.1305,
      "step": 75966
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.8295899033546448,
      "learning_rate": 0.0004529447083044451,
      "loss": 2.0842,
      "step": 75967
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.85164475440979,
      "learning_rate": 0.0004529411892464733,
      "loss": 2.1401,
      "step": 75968
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.7997254133224487,
      "learning_rate": 0.0004529376701600669,
      "loss": 2.1286,
      "step": 75969
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.824793815612793,
      "learning_rate": 0.00045293415104522667,
      "loss": 2.2361,
      "step": 75970
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.9392521977424622,
      "learning_rate": 0.00045293063190195284,
      "loss": 2.2886,
      "step": 75971
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.8161964416503906,
      "learning_rate": 0.00045292711273024647,
      "loss": 2.2784,
      "step": 75972
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.7731184363365173,
      "learning_rate": 0.0004529235935301081,
      "loss": 2.3771,
      "step": 75973
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.9524919390678406,
      "learning_rate": 0.0004529200743015383,
      "loss": 2.2596,
      "step": 75974
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.7396345734596252,
      "learning_rate": 0.00045291655504453766,
      "loss": 2.1553,
      "step": 75975
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.7586054801940918,
      "learning_rate": 0.00045291303575910715,
      "loss": 2.2059,
      "step": 75976
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.0075322389602661,
      "learning_rate": 0.00045290951644524706,
      "loss": 2.2119,
      "step": 75977
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.8879925608634949,
      "learning_rate": 0.0004529059971029582,
      "loss": 2.1053,
      "step": 75978
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.7883530259132385,
      "learning_rate": 0.00045290247773224133,
      "loss": 2.2195,
      "step": 75979
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.862009584903717,
      "learning_rate": 0.00045289895833309693,
      "loss": 2.0309,
      "step": 75980
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.7493033409118652,
      "learning_rate": 0.0004528954389055257,
      "loss": 2.222,
      "step": 75981
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.7910176515579224,
      "learning_rate": 0.0004528919194495285,
      "loss": 2.0516,
      "step": 75982
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.9347794055938721,
      "learning_rate": 0.00045288839996510566,
      "loss": 2.4545,
      "step": 75983
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.8066021203994751,
      "learning_rate": 0.00045288488045225794,
      "loss": 2.3483,
      "step": 75984
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.7280902862548828,
      "learning_rate": 0.00045288136091098613,
      "loss": 2.2941,
      "step": 75985
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.752351701259613,
      "learning_rate": 0.0004528778413412907,
      "loss": 2.1513,
      "step": 75986
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.7897246479988098,
      "learning_rate": 0.0004528743217431725,
      "loss": 2.0536,
      "step": 75987
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.7937746644020081,
      "learning_rate": 0.00045287080211663203,
      "loss": 2.53,
      "step": 75988
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.7311633229255676,
      "learning_rate": 0.00045286728246167,
      "loss": 2.2562,
      "step": 75989
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.8491038680076599,
      "learning_rate": 0.000452863762778287,
      "loss": 2.2571,
      "step": 75990
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.7534672617912292,
      "learning_rate": 0.0004528602430664838,
      "loss": 2.2739,
      "step": 75991
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.8481461405754089,
      "learning_rate": 0.000452856723326261,
      "loss": 2.2961,
      "step": 75992
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.8028753399848938,
      "learning_rate": 0.00045285320355761913,
      "loss": 2.3674,
      "step": 75993
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.7358438968658447,
      "learning_rate": 0.0004528496837605592,
      "loss": 2.1381,
      "step": 75994
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.7726858258247375,
      "learning_rate": 0.00045284616393508146,
      "loss": 2.4382,
      "step": 75995
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.7901561260223389,
      "learning_rate": 0.0004528426440811867,
      "loss": 2.2483,
      "step": 75996
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.7021148204803467,
      "learning_rate": 0.00045283912419887576,
      "loss": 2.2409,
      "step": 75997
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.8015419244766235,
      "learning_rate": 0.000452835604288149,
      "loss": 2.3026,
      "step": 75998
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.7419211268424988,
      "learning_rate": 0.00045283208434900723,
      "loss": 2.167,
      "step": 75999
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.8097683191299438,
      "learning_rate": 0.0004528285643814512,
      "loss": 2.2509,
      "step": 76000
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.7319713830947876,
      "learning_rate": 0.00045282504438548136,
      "loss": 2.2924,
      "step": 76001
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.8233213424682617,
      "learning_rate": 0.00045282152436109844,
      "loss": 1.9917,
      "step": 76002
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.8070684671401978,
      "learning_rate": 0.00045281800430830324,
      "loss": 2.1313,
      "step": 76003
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.7623414993286133,
      "learning_rate": 0.00045281448422709614,
      "loss": 2.219,
      "step": 76004
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.8464698195457458,
      "learning_rate": 0.000452810964117478,
      "loss": 2.1678,
      "step": 76005
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.8223344683647156,
      "learning_rate": 0.0004528074439794495,
      "loss": 2.5179,
      "step": 76006
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.749788224697113,
      "learning_rate": 0.0004528039238130111,
      "loss": 2.1276,
      "step": 76007
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.7411238551139832,
      "learning_rate": 0.00045280040361816363,
      "loss": 2.0868,
      "step": 76008
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.8672145009040833,
      "learning_rate": 0.0004527968833949077,
      "loss": 2.3283,
      "step": 76009
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.8378060460090637,
      "learning_rate": 0.0004527933631432439,
      "loss": 2.117,
      "step": 76010
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.8968138098716736,
      "learning_rate": 0.00045278984286317286,
      "loss": 2.3576,
      "step": 76011
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.7609619498252869,
      "learning_rate": 0.00045278632255469546,
      "loss": 2.0689,
      "step": 76012
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.7767886519432068,
      "learning_rate": 0.00045278280221781213,
      "loss": 1.9992,
      "step": 76013
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.8512711524963379,
      "learning_rate": 0.0004527792818525235,
      "loss": 2.1876,
      "step": 76014
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.8126067519187927,
      "learning_rate": 0.0004527757614588304,
      "loss": 2.2187,
      "step": 76015
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.7950950860977173,
      "learning_rate": 0.0004527722410367334,
      "loss": 2.0222,
      "step": 76016
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.7532350420951843,
      "learning_rate": 0.00045276872058623314,
      "loss": 2.2928,
      "step": 76017
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.8155906796455383,
      "learning_rate": 0.00045276520010733033,
      "loss": 2.2279,
      "step": 76018
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.9140805006027222,
      "learning_rate": 0.00045276167960002554,
      "loss": 2.364,
      "step": 76019
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.7758538722991943,
      "learning_rate": 0.0004527581590643195,
      "loss": 2.2362,
      "step": 76020
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.8147163391113281,
      "learning_rate": 0.00045275463850021286,
      "loss": 2.1601,
      "step": 76021
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.892785906791687,
      "learning_rate": 0.0004527511179077062,
      "loss": 2.1215,
      "step": 76022
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.8680402040481567,
      "learning_rate": 0.0004527475972868002,
      "loss": 2.1371,
      "step": 76023
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.7145501971244812,
      "learning_rate": 0.00045274407663749565,
      "loss": 2.4096,
      "step": 76024
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.9339711666107178,
      "learning_rate": 0.00045274055595979294,
      "loss": 2.4513,
      "step": 76025
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.868888795375824,
      "learning_rate": 0.00045273703525369296,
      "loss": 2.3551,
      "step": 76026
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.7560353875160217,
      "learning_rate": 0.0004527335145191963,
      "loss": 2.1932,
      "step": 76027
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.7673714756965637,
      "learning_rate": 0.00045272999375630356,
      "loss": 2.1355,
      "step": 76028
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.055249810218811,
      "learning_rate": 0.00045272647296501543,
      "loss": 2.1789,
      "step": 76029
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.7769514322280884,
      "learning_rate": 0.00045272295214533265,
      "loss": 2.1567,
      "step": 76030
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.9004716873168945,
      "learning_rate": 0.0004527194312972557,
      "loss": 2.0189,
      "step": 76031
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.8891023993492126,
      "learning_rate": 0.00045271591042078533,
      "loss": 2.3358,
      "step": 76032
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.7967095375061035,
      "learning_rate": 0.0004527123895159222,
      "loss": 2.3905,
      "step": 76033
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.7119958400726318,
      "learning_rate": 0.00045270886858266696,
      "loss": 2.1356,
      "step": 76034
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.8195884823799133,
      "learning_rate": 0.00045270534762102026,
      "loss": 2.239,
      "step": 76035
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.8012812733650208,
      "learning_rate": 0.00045270182663098275,
      "loss": 2.0348,
      "step": 76036
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.8061591982841492,
      "learning_rate": 0.0004526983056125551,
      "loss": 2.179,
      "step": 76037
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.8024730086326599,
      "learning_rate": 0.0004526947845657379,
      "loss": 2.2404,
      "step": 76038
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.8346325755119324,
      "learning_rate": 0.00045269126349053193,
      "loss": 2.2933,
      "step": 76039
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.8779581785202026,
      "learning_rate": 0.00045268774238693774,
      "loss": 2.1092,
      "step": 76040
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.7888007760047913,
      "learning_rate": 0.00045268422125495605,
      "loss": 2.1921,
      "step": 76041
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.8339040279388428,
      "learning_rate": 0.00045268070009458746,
      "loss": 2.1936,
      "step": 76042
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.8380248546600342,
      "learning_rate": 0.0004526771789058326,
      "loss": 2.101,
      "step": 76043
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.778948962688446,
      "learning_rate": 0.0004526736576886922,
      "loss": 2.2723,
      "step": 76044
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.7734962105751038,
      "learning_rate": 0.0004526701364431669,
      "loss": 2.1748,
      "step": 76045
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.8324052095413208,
      "learning_rate": 0.0004526666151692573,
      "loss": 2.3076,
      "step": 76046
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.7723863124847412,
      "learning_rate": 0.0004526630938669642,
      "loss": 2.2244,
      "step": 76047
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.8311149477958679,
      "learning_rate": 0.000452659572536288,
      "loss": 2.1137,
      "step": 76048
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.8005392551422119,
      "learning_rate": 0.00045265605117722957,
      "loss": 2.2969,
      "step": 76049
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.7335454821586609,
      "learning_rate": 0.0004526525297897895,
      "loss": 2.1365,
      "step": 76050
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.7733843326568604,
      "learning_rate": 0.00045264900837396847,
      "loss": 2.0929,
      "step": 76051
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.7576225399971008,
      "learning_rate": 0.00045264548692976707,
      "loss": 2.2534,
      "step": 76052
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.8332183361053467,
      "learning_rate": 0.00045264196545718603,
      "loss": 2.2541,
      "step": 76053
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.7496340870857239,
      "learning_rate": 0.00045263844395622595,
      "loss": 2.2111,
      "step": 76054
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.7781710624694824,
      "learning_rate": 0.0004526349224268875,
      "loss": 2.3107,
      "step": 76055
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.8806115388870239,
      "learning_rate": 0.00045263140086917135,
      "loss": 2.2757,
      "step": 76056
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.8082126975059509,
      "learning_rate": 0.0004526278792830781,
      "loss": 2.2012,
      "step": 76057
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.8938233256340027,
      "learning_rate": 0.0004526243576686085,
      "loss": 2.1018,
      "step": 76058
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.7992718815803528,
      "learning_rate": 0.0004526208360257631,
      "loss": 2.1491,
      "step": 76059
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.0180270671844482,
      "learning_rate": 0.0004526173143545426,
      "loss": 1.9706,
      "step": 76060
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.766677975654602,
      "learning_rate": 0.0004526137926549477,
      "loss": 2.0837,
      "step": 76061
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.7157576680183411,
      "learning_rate": 0.000452610270926979,
      "loss": 2.2404,
      "step": 76062
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.733601450920105,
      "learning_rate": 0.0004526067491706372,
      "loss": 2.2542,
      "step": 76063
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.7862779498100281,
      "learning_rate": 0.0004526032273859229,
      "loss": 2.3071,
      "step": 76064
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.8087329268455505,
      "learning_rate": 0.00045259970557283674,
      "loss": 2.3311,
      "step": 76065
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.8440452218055725,
      "learning_rate": 0.0004525961837313795,
      "loss": 2.3187,
      "step": 76066
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.8479041457176208,
      "learning_rate": 0.0004525926618615517,
      "loss": 2.2796,
      "step": 76067
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.7874919772148132,
      "learning_rate": 0.00045258913996335404,
      "loss": 2.3352,
      "step": 76068
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.8347145915031433,
      "learning_rate": 0.0004525856180367872,
      "loss": 2.2711,
      "step": 76069
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.9378063082695007,
      "learning_rate": 0.0004525820960818518,
      "loss": 2.3138,
      "step": 76070
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.7883778214454651,
      "learning_rate": 0.0004525785740985485,
      "loss": 2.0663,
      "step": 76071
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.7461467981338501,
      "learning_rate": 0.0004525750520868779,
      "loss": 1.9427,
      "step": 76072
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.8617148995399475,
      "learning_rate": 0.0004525715300468408,
      "loss": 2.2114,
      "step": 76073
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.7620468735694885,
      "learning_rate": 0.00045256800797843777,
      "loss": 2.0396,
      "step": 76074
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.8352615833282471,
      "learning_rate": 0.00045256448588166956,
      "loss": 2.0304,
      "step": 76075
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.8593555688858032,
      "learning_rate": 0.00045256096375653656,
      "loss": 2.4017,
      "step": 76076
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.820309579372406,
      "learning_rate": 0.00045255744160303964,
      "loss": 2.2325,
      "step": 76077
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.7973192930221558,
      "learning_rate": 0.0004525539194211795,
      "loss": 2.3372,
      "step": 76078
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.9882568717002869,
      "learning_rate": 0.0004525503972109567,
      "loss": 2.3046,
      "step": 76079
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.8617855906486511,
      "learning_rate": 0.00045254687497237175,
      "loss": 2.3341,
      "step": 76080
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.7487959265708923,
      "learning_rate": 0.00045254335270542554,
      "loss": 2.0967,
      "step": 76081
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.8228897452354431,
      "learning_rate": 0.0004525398304101187,
      "loss": 2.2581,
      "step": 76082
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.7665581107139587,
      "learning_rate": 0.0004525363080864517,
      "loss": 1.9833,
      "step": 76083
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.7731225490570068,
      "learning_rate": 0.0004525327857344254,
      "loss": 2.1218,
      "step": 76084
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.8839011788368225,
      "learning_rate": 0.00045252926335404047,
      "loss": 2.2386,
      "step": 76085
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.9469969272613525,
      "learning_rate": 0.00045252574094529737,
      "loss": 2.2073,
      "step": 76086
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.8203333020210266,
      "learning_rate": 0.00045252221850819685,
      "loss": 2.1581,
      "step": 76087
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.8850593566894531,
      "learning_rate": 0.0004525186960427397,
      "loss": 2.204,
      "step": 76088
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.7117343544960022,
      "learning_rate": 0.0004525151735489262,
      "loss": 2.0037,
      "step": 76089
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.9792497754096985,
      "learning_rate": 0.0004525116510267574,
      "loss": 2.3648,
      "step": 76090
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.8956757187843323,
      "learning_rate": 0.0004525081284762338,
      "loss": 2.2203,
      "step": 76091
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.7591962814331055,
      "learning_rate": 0.000452504605897356,
      "loss": 2.3623,
      "step": 76092
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.9273672699928284,
      "learning_rate": 0.00045250108329012475,
      "loss": 2.4382,
      "step": 76093
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.8368592858314514,
      "learning_rate": 0.0004524975606545407,
      "loss": 2.2233,
      "step": 76094
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.7327065467834473,
      "learning_rate": 0.0004524940379906045,
      "loss": 2.1294,
      "step": 76095
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.7776572704315186,
      "learning_rate": 0.0004524905152983167,
      "loss": 2.3564,
      "step": 76096
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.7406085729598999,
      "learning_rate": 0.00045248699257767813,
      "loss": 2.3142,
      "step": 76097
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.8388044238090515,
      "learning_rate": 0.00045248346982868924,
      "loss": 2.2766,
      "step": 76098
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.788352370262146,
      "learning_rate": 0.0004524799470513509,
      "loss": 2.2803,
      "step": 76099
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.8358339071273804,
      "learning_rate": 0.0004524764242456636,
      "loss": 2.0936,
      "step": 76100
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.8466995358467102,
      "learning_rate": 0.00045247290141162805,
      "loss": 2.2061,
      "step": 76101
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.827019453048706,
      "learning_rate": 0.0004524693785492449,
      "loss": 2.0359,
      "step": 76102
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.7958152294158936,
      "learning_rate": 0.00045246585565851484,
      "loss": 2.1625,
      "step": 76103
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.9070421457290649,
      "learning_rate": 0.0004524623327394385,
      "loss": 2.1676,
      "step": 76104
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.9653340578079224,
      "learning_rate": 0.00045245880979201647,
      "loss": 2.3879,
      "step": 76105
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.8052482604980469,
      "learning_rate": 0.00045245528681624945,
      "loss": 2.3088,
      "step": 76106
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.8372153639793396,
      "learning_rate": 0.0004524517638121383,
      "loss": 2.2841,
      "step": 76107
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.8953987956047058,
      "learning_rate": 0.0004524482407796834,
      "loss": 2.3212,
      "step": 76108
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.880928635597229,
      "learning_rate": 0.0004524447177188854,
      "loss": 2.134,
      "step": 76109
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.7879276275634766,
      "learning_rate": 0.0004524411946297452,
      "loss": 2.347,
      "step": 76110
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.7875191569328308,
      "learning_rate": 0.0004524376715122632,
      "loss": 2.1747,
      "step": 76111
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.8492176532745361,
      "learning_rate": 0.0004524341483664401,
      "loss": 2.2502,
      "step": 76112
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.771800696849823,
      "learning_rate": 0.0004524306251922768,
      "loss": 2.2736,
      "step": 76113
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.765152633190155,
      "learning_rate": 0.00045242710198977365,
      "loss": 2.19,
      "step": 76114
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.7154845595359802,
      "learning_rate": 0.00045242357875893143,
      "loss": 2.2034,
      "step": 76115
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.8338580131530762,
      "learning_rate": 0.0004524200554997508,
      "loss": 2.1922,
      "step": 76116
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.7537640333175659,
      "learning_rate": 0.00045241653221223243,
      "loss": 2.094,
      "step": 76117
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.7255314588546753,
      "learning_rate": 0.0004524130088963769,
      "loss": 2.1915,
      "step": 76118
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.7355672121047974,
      "learning_rate": 0.000452409485552185,
      "loss": 2.0842,
      "step": 76119
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.7714632749557495,
      "learning_rate": 0.0004524059621796572,
      "loss": 2.088,
      "step": 76120
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.8812524676322937,
      "learning_rate": 0.0004524024387787943,
      "loss": 2.1775,
      "step": 76121
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.8157671093940735,
      "learning_rate": 0.0004523989153495969,
      "loss": 2.1013,
      "step": 76122
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.7694061398506165,
      "learning_rate": 0.00045239539189206564,
      "loss": 2.0714,
      "step": 76123
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.8115893602371216,
      "learning_rate": 0.00045239186840620124,
      "loss": 1.9943,
      "step": 76124
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.7893549799919128,
      "learning_rate": 0.00045238834489200435,
      "loss": 2.0898,
      "step": 76125
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.742235541343689,
      "learning_rate": 0.00045238482134947553,
      "loss": 2.4135,
      "step": 76126
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.7815943956375122,
      "learning_rate": 0.0004523812977786155,
      "loss": 2.3966,
      "step": 76127
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.7605671882629395,
      "learning_rate": 0.000452377774179425,
      "loss": 2.2149,
      "step": 76128
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.792046070098877,
      "learning_rate": 0.0004523742505519045,
      "loss": 2.1413,
      "step": 76129
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.7884691953659058,
      "learning_rate": 0.0004523707268960547,
      "loss": 2.2113,
      "step": 76130
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.750666081905365,
      "learning_rate": 0.00045236720321187643,
      "loss": 2.171,
      "step": 76131
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.7757242321968079,
      "learning_rate": 0.00045236367949937013,
      "loss": 2.4603,
      "step": 76132
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.8131185173988342,
      "learning_rate": 0.00045236015575853656,
      "loss": 2.1617,
      "step": 76133
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.7668133974075317,
      "learning_rate": 0.00045235663198937636,
      "loss": 2.3947,
      "step": 76134
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.7724049687385559,
      "learning_rate": 0.0004523531081918903,
      "loss": 2.2621,
      "step": 76135
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.7516063451766968,
      "learning_rate": 0.00045234958436607875,
      "loss": 2.3574,
      "step": 76136
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.8281126022338867,
      "learning_rate": 0.0004523460605119427,
      "loss": 2.308,
      "step": 76137
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.9207812547683716,
      "learning_rate": 0.00045234253662948244,
      "loss": 2.1828,
      "step": 76138
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.8677982687950134,
      "learning_rate": 0.000452339012718699,
      "loss": 2.1745,
      "step": 76139
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.8686894774436951,
      "learning_rate": 0.0004523354887795928,
      "loss": 2.3992,
      "step": 76140
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.7748702764511108,
      "learning_rate": 0.00045233196481216456,
      "loss": 2.2006,
      "step": 76141
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.7407296299934387,
      "learning_rate": 0.000452328440816415,
      "loss": 2.3315,
      "step": 76142
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.7881219387054443,
      "learning_rate": 0.00045232491679234456,
      "loss": 2.2874,
      "step": 76143
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.7137155532836914,
      "learning_rate": 0.00045232139273995416,
      "loss": 2.1999,
      "step": 76144
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.7925815582275391,
      "learning_rate": 0.00045231786865924424,
      "loss": 2.2146,
      "step": 76145
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.7974104881286621,
      "learning_rate": 0.0004523143445502157,
      "loss": 2.1361,
      "step": 76146
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.807300865650177,
      "learning_rate": 0.00045231082041286886,
      "loss": 2.3703,
      "step": 76147
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.7317440509796143,
      "learning_rate": 0.0004523072962472047,
      "loss": 2.2239,
      "step": 76148
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.8232427835464478,
      "learning_rate": 0.00045230377205322374,
      "loss": 2.1834,
      "step": 76149
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.7377837300300598,
      "learning_rate": 0.00045230024783092656,
      "loss": 2.184,
      "step": 76150
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.7913010120391846,
      "learning_rate": 0.000452296723580314,
      "loss": 2.2446,
      "step": 76151
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.7618675827980042,
      "learning_rate": 0.0004522931993013865,
      "loss": 2.1359,
      "step": 76152
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.7749783992767334,
      "learning_rate": 0.00045228967499414487,
      "loss": 2.0909,
      "step": 76153
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.8232830166816711,
      "learning_rate": 0.0004522861506585896,
      "loss": 2.0664,
      "step": 76154
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.8284468650817871,
      "learning_rate": 0.0004522826262947217,
      "loss": 2.4083,
      "step": 76155
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.809328556060791,
      "learning_rate": 0.00045227910190254135,
      "loss": 2.2115,
      "step": 76156
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.7487112879753113,
      "learning_rate": 0.00045227557748204957,
      "loss": 2.3559,
      "step": 76157
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.7779573798179626,
      "learning_rate": 0.00045227205303324694,
      "loss": 2.2262,
      "step": 76158
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.8234038949012756,
      "learning_rate": 0.0004522685285561339,
      "loss": 2.2309,
      "step": 76159
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.7937113642692566,
      "learning_rate": 0.0004522650040507113,
      "loss": 2.1447,
      "step": 76160
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.709092915058136,
      "learning_rate": 0.00045226147951697986,
      "loss": 2.2827,
      "step": 76161
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.7187531590461731,
      "learning_rate": 0.00045225795495494,
      "loss": 2.016,
      "step": 76162
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.7924056649208069,
      "learning_rate": 0.00045225443036459257,
      "loss": 2.2878,
      "step": 76163
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.7516347765922546,
      "learning_rate": 0.0004522509057459383,
      "loss": 2.2261,
      "step": 76164
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.7567318677902222,
      "learning_rate": 0.00045224738109897754,
      "loss": 2.2601,
      "step": 76165
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.7354902029037476,
      "learning_rate": 0.00045224385642371115,
      "loss": 1.9636,
      "step": 76166
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.7999674081802368,
      "learning_rate": 0.0004522403317201398,
      "loss": 1.8673,
      "step": 76167
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.759765088558197,
      "learning_rate": 0.00045223680698826404,
      "loss": 2.2048,
      "step": 76168
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.8652456402778625,
      "learning_rate": 0.0004522332822280846,
      "loss": 2.2188,
      "step": 76169
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.7857709527015686,
      "learning_rate": 0.00045222975743960215,
      "loss": 2.2214,
      "step": 76170
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.8249683976173401,
      "learning_rate": 0.00045222623262281735,
      "loss": 2.0037,
      "step": 76171
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.8043588399887085,
      "learning_rate": 0.00045222270777773075,
      "loss": 2.2725,
      "step": 76172
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.7931102514266968,
      "learning_rate": 0.0004522191829043431,
      "loss": 2.2167,
      "step": 76173
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.8156501054763794,
      "learning_rate": 0.00045221565800265503,
      "loss": 2.4513,
      "step": 76174
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.757631778717041,
      "learning_rate": 0.0004522121330726672,
      "loss": 2.1552,
      "step": 76175
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.8373979926109314,
      "learning_rate": 0.0004522086081143802,
      "loss": 2.0935,
      "step": 76176
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.7555906772613525,
      "learning_rate": 0.0004522050831277949,
      "loss": 2.0239,
      "step": 76177
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.8147291541099548,
      "learning_rate": 0.0004522015581129117,
      "loss": 2.0906,
      "step": 76178
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.7733346223831177,
      "learning_rate": 0.0004521980330697313,
      "loss": 2.16,
      "step": 76179
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.8365415334701538,
      "learning_rate": 0.0004521945079982545,
      "loss": 1.9928,
      "step": 76180
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.7882252335548401,
      "learning_rate": 0.00045219098289848184,
      "loss": 2.3887,
      "step": 76181
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.8171291351318359,
      "learning_rate": 0.000452187457770414,
      "loss": 2.1412,
      "step": 76182
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.9521896839141846,
      "learning_rate": 0.00045218393261405173,
      "loss": 2.2992,
      "step": 76183
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.842419445514679,
      "learning_rate": 0.0004521804074293954,
      "loss": 2.2331,
      "step": 76184
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.7989113330841064,
      "learning_rate": 0.000452176882216446,
      "loss": 2.1369,
      "step": 76185
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.8457010388374329,
      "learning_rate": 0.00045217335697520406,
      "loss": 2.0516,
      "step": 76186
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.8649605512619019,
      "learning_rate": 0.00045216983170567014,
      "loss": 2.26,
      "step": 76187
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.7789553999900818,
      "learning_rate": 0.000452166306407845,
      "loss": 2.214,
      "step": 76188
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.8026784658432007,
      "learning_rate": 0.0004521627810817293,
      "loss": 1.9959,
      "step": 76189
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.8526171445846558,
      "learning_rate": 0.0004521592557273237,
      "loss": 2.4059,
      "step": 76190
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.7640339136123657,
      "learning_rate": 0.0004521557303446287,
      "loss": 2.2211,
      "step": 76191
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.7787795662879944,
      "learning_rate": 0.0004521522049336452,
      "loss": 2.2093,
      "step": 76192
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.8265371918678284,
      "learning_rate": 0.00045214867949437367,
      "loss": 2.1928,
      "step": 76193
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.9254041314125061,
      "learning_rate": 0.0004521451540268148,
      "loss": 2.2327,
      "step": 76194
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.7602233290672302,
      "learning_rate": 0.0004521416285309694,
      "loss": 2.3757,
      "step": 76195
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.9155264496803284,
      "learning_rate": 0.00045213810300683785,
      "loss": 2.2359,
      "step": 76196
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.9054003357887268,
      "learning_rate": 0.000452134577454421,
      "loss": 2.4386,
      "step": 76197
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.6967082023620605,
      "learning_rate": 0.0004521310518737195,
      "loss": 2.1009,
      "step": 76198
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.7794792652130127,
      "learning_rate": 0.00045212752626473397,
      "loss": 2.0961,
      "step": 76199
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.7618675827980042,
      "learning_rate": 0.000452124000627465,
      "loss": 2.2484,
      "step": 76200
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.8719433546066284,
      "learning_rate": 0.0004521204749619134,
      "loss": 2.1112,
      "step": 76201
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.7588107585906982,
      "learning_rate": 0.00045211694926807963,
      "loss": 2.1081,
      "step": 76202
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.7807159423828125,
      "learning_rate": 0.0004521134235459645,
      "loss": 2.1674,
      "step": 76203
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.7553226947784424,
      "learning_rate": 0.0004521098977955686,
      "loss": 2.2728,
      "step": 76204
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.7056287527084351,
      "learning_rate": 0.0004521063720168927,
      "loss": 2.2919,
      "step": 76205
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.6801181435585022,
      "learning_rate": 0.0004521028462099372,
      "loss": 2.1726,
      "step": 76206
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.8211966156959534,
      "learning_rate": 0.00045209932037470297,
      "loss": 2.0416,
      "step": 76207
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.7778806686401367,
      "learning_rate": 0.0004520957945111905,
      "loss": 2.2473,
      "step": 76208
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.8814694881439209,
      "learning_rate": 0.00045209226861940067,
      "loss": 2.1172,
      "step": 76209
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.7817173600196838,
      "learning_rate": 0.00045208874269933407,
      "loss": 2.1732,
      "step": 76210
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.7885029911994934,
      "learning_rate": 0.0004520852167509912,
      "loss": 2.0396,
      "step": 76211
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.8781539797782898,
      "learning_rate": 0.0004520816907743729,
      "loss": 2.3045,
      "step": 76212
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.760007381439209,
      "learning_rate": 0.00045207816476947965,
      "loss": 2.1147,
      "step": 76213
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.8268862962722778,
      "learning_rate": 0.0004520746387363123,
      "loss": 2.1704,
      "step": 76214
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.7787221670150757,
      "learning_rate": 0.0004520711126748713,
      "loss": 2.3335,
      "step": 76215
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.8445895910263062,
      "learning_rate": 0.0004520675865851575,
      "loss": 2.2199,
      "step": 76216
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.7839984893798828,
      "learning_rate": 0.00045206406046717143,
      "loss": 2.1536,
      "step": 76217
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.8078170418739319,
      "learning_rate": 0.0004520605343209137,
      "loss": 2.4175,
      "step": 76218
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.7107303738594055,
      "learning_rate": 0.00045205700814638516,
      "loss": 1.9854,
      "step": 76219
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.8331849575042725,
      "learning_rate": 0.00045205348194358633,
      "loss": 2.0987,
      "step": 76220
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.797221839427948,
      "learning_rate": 0.00045204995571251785,
      "loss": 2.0859,
      "step": 76221
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.7959405183792114,
      "learning_rate": 0.0004520464294531804,
      "loss": 2.0077,
      "step": 76222
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.864520788192749,
      "learning_rate": 0.00045204290316557466,
      "loss": 2.1945,
      "step": 76223
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.8025647401809692,
      "learning_rate": 0.0004520393768497013,
      "loss": 2.1543,
      "step": 76224
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.7842733263969421,
      "learning_rate": 0.0004520358505055609,
      "loss": 2.0194,
      "step": 76225
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.8317868113517761,
      "learning_rate": 0.0004520323241331542,
      "loss": 2.1777,
      "step": 76226
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.7318950891494751,
      "learning_rate": 0.00045202879773248184,
      "loss": 2.1436,
      "step": 76227
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.8485534191131592,
      "learning_rate": 0.00045202527130354446,
      "loss": 2.0326,
      "step": 76228
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.8107895255088806,
      "learning_rate": 0.0004520217448463427,
      "loss": 2.1015,
      "step": 76229
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.7870410680770874,
      "learning_rate": 0.0004520182183608772,
      "loss": 2.0968,
      "step": 76230
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.866291344165802,
      "learning_rate": 0.00045201469184714873,
      "loss": 2.3781,
      "step": 76231
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.9544569849967957,
      "learning_rate": 0.0004520111653051578,
      "loss": 2.3073,
      "step": 76232
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.7440645098686218,
      "learning_rate": 0.000452007638734905,
      "loss": 1.8174,
      "step": 76233
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.8316317200660706,
      "learning_rate": 0.00045200411213639136,
      "loss": 2.0568,
      "step": 76234
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.7654517889022827,
      "learning_rate": 0.0004520005855096171,
      "loss": 2.0683,
      "step": 76235
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.7371489405632019,
      "learning_rate": 0.00045199705885458314,
      "loss": 2.2658,
      "step": 76236
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.8205626606941223,
      "learning_rate": 0.00045199353217129,
      "loss": 2.2725,
      "step": 76237
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.7736210227012634,
      "learning_rate": 0.00045199000545973844,
      "loss": 2.0448,
      "step": 76238
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.8729352355003357,
      "learning_rate": 0.0004519864787199291,
      "loss": 2.185,
      "step": 76239
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.7813028693199158,
      "learning_rate": 0.0004519829519518626,
      "loss": 2.2104,
      "step": 76240
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.8985069394111633,
      "learning_rate": 0.0004519794251555396,
      "loss": 2.0784,
      "step": 76241
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.7495681643486023,
      "learning_rate": 0.00045197589833096064,
      "loss": 2.0182,
      "step": 76242
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.7823406457901001,
      "learning_rate": 0.00045197237147812655,
      "loss": 2.1334,
      "step": 76243
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.8783328533172607,
      "learning_rate": 0.00045196884459703794,
      "loss": 2.153,
      "step": 76244
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.8154186606407166,
      "learning_rate": 0.0004519653176876955,
      "loss": 2.2554,
      "step": 76245
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.7802298069000244,
      "learning_rate": 0.0004519617907500998,
      "loss": 2.1205,
      "step": 76246
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.8085319995880127,
      "learning_rate": 0.0004519582637842516,
      "loss": 2.1665,
      "step": 76247
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.8903442025184631,
      "learning_rate": 0.0004519547367901514,
      "loss": 2.3111,
      "step": 76248
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.7752403616905212,
      "learning_rate": 0.00045195120976779994,
      "loss": 2.2872,
      "step": 76249
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.0493392944335938,
      "learning_rate": 0.00045194768271719795,
      "loss": 2.103,
      "step": 76250
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.8650557994842529,
      "learning_rate": 0.000451944155638346,
      "loss": 2.1944,
      "step": 76251
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.8035440444946289,
      "learning_rate": 0.00045194062853124464,
      "loss": 2.3124,
      "step": 76252
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.8384285569190979,
      "learning_rate": 0.00045193710139589485,
      "loss": 2.2711,
      "step": 76253
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.8726894855499268,
      "learning_rate": 0.00045193357423229703,
      "loss": 2.0947,
      "step": 76254
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.8684025406837463,
      "learning_rate": 0.00045193004704045174,
      "loss": 2.2586,
      "step": 76255
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.0917168855667114,
      "learning_rate": 0.00045192651982036,
      "loss": 2.1281,
      "step": 76256
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.8786236643791199,
      "learning_rate": 0.0004519229925720221,
      "loss": 2.2292,
      "step": 76257
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.8757027387619019,
      "learning_rate": 0.00045191946529543885,
      "loss": 2.2481,
      "step": 76258
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.912372887134552,
      "learning_rate": 0.000451915937990611,
      "loss": 2.3269,
      "step": 76259
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.8100382089614868,
      "learning_rate": 0.00045191241065753907,
      "loss": 2.2852,
      "step": 76260
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.7651466727256775,
      "learning_rate": 0.0004519088832962237,
      "loss": 2.014,
      "step": 76261
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.8302409052848816,
      "learning_rate": 0.00045190535590666574,
      "loss": 2.1788,
      "step": 76262
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.8059689402580261,
      "learning_rate": 0.00045190182848886556,
      "loss": 2.2184,
      "step": 76263
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.7363686561584473,
      "learning_rate": 0.000451898301042824,
      "loss": 2.1939,
      "step": 76264
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.7472363710403442,
      "learning_rate": 0.00045189477356854174,
      "loss": 2.16,
      "step": 76265
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.7734909653663635,
      "learning_rate": 0.00045189124606601934,
      "loss": 2.1549,
      "step": 76266
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.8756899833679199,
      "learning_rate": 0.0004518877185352575,
      "loss": 2.2889,
      "step": 76267
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.7352267503738403,
      "learning_rate": 0.00045188419097625683,
      "loss": 2.2296,
      "step": 76268
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.7394211292266846,
      "learning_rate": 0.000451880663389018,
      "loss": 1.9909,
      "step": 76269
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.8532896637916565,
      "learning_rate": 0.00045187713577354177,
      "loss": 2.1585,
      "step": 76270
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.7702236771583557,
      "learning_rate": 0.00045187360812982865,
      "loss": 2.1452,
      "step": 76271
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.8032936453819275,
      "learning_rate": 0.0004518700804578794,
      "loss": 2.3316,
      "step": 76272
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.9049227833747864,
      "learning_rate": 0.00045186655275769464,
      "loss": 2.2116,
      "step": 76273
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.8314059972763062,
      "learning_rate": 0.000451863025029275,
      "loss": 2.1484,
      "step": 76274
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.7264007925987244,
      "learning_rate": 0.0004518594972726211,
      "loss": 2.2534,
      "step": 76275
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.8190026879310608,
      "learning_rate": 0.00045185596948773367,
      "loss": 2.166,
      "step": 76276
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.7865518927574158,
      "learning_rate": 0.0004518524416746135,
      "loss": 2.0784,
      "step": 76277
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.7171759009361267,
      "learning_rate": 0.0004518489138332609,
      "loss": 2.1244,
      "step": 76278
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.7689493894577026,
      "learning_rate": 0.0004518453859636768,
      "loss": 2.3014,
      "step": 76279
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.7856184840202332,
      "learning_rate": 0.0004518418580658619,
      "loss": 2.0908,
      "step": 76280
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.8206257820129395,
      "learning_rate": 0.0004518383301398165,
      "loss": 2.1702,
      "step": 76281
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.8213170170783997,
      "learning_rate": 0.0004518348021855416,
      "loss": 2.3531,
      "step": 76282
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.8086553812026978,
      "learning_rate": 0.00045183127420303775,
      "loss": 2.2864,
      "step": 76283
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.8071134686470032,
      "learning_rate": 0.0004518277461923056,
      "loss": 2.2433,
      "step": 76284
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.8120651841163635,
      "learning_rate": 0.0004518242181533458,
      "loss": 2.3461,
      "step": 76285
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.8320773839950562,
      "learning_rate": 0.00045182069008615903,
      "loss": 2.1394,
      "step": 76286
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.80149245262146,
      "learning_rate": 0.0004518171619907459,
      "loss": 2.0832,
      "step": 76287
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.757021963596344,
      "learning_rate": 0.00045181363386710704,
      "loss": 2.2602,
      "step": 76288
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.8702031970024109,
      "learning_rate": 0.00045181010571524333,
      "loss": 2.267,
      "step": 76289
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.8459867238998413,
      "learning_rate": 0.0004518065775351551,
      "loss": 2.2201,
      "step": 76290
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.7630782127380371,
      "learning_rate": 0.0004518030493268432,
      "loss": 1.9578,
      "step": 76291
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.7865425944328308,
      "learning_rate": 0.00045179952109030823,
      "loss": 2.4184,
      "step": 76292
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.7560628652572632,
      "learning_rate": 0.0004517959928255509,
      "loss": 2.1421,
      "step": 76293
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.7774987816810608,
      "learning_rate": 0.00045179246453257177,
      "loss": 2.2297,
      "step": 76294
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.772822916507721,
      "learning_rate": 0.0004517889362113717,
      "loss": 2.3552,
      "step": 76295
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.9009048938751221,
      "learning_rate": 0.0004517854078619511,
      "loss": 2.0877,
      "step": 76296
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.8598880767822266,
      "learning_rate": 0.0004517818794843107,
      "loss": 2.3103,
      "step": 76297
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.7373629808425903,
      "learning_rate": 0.00045177835107845126,
      "loss": 2.0777,
      "step": 76298
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.7460266351699829,
      "learning_rate": 0.0004517748226443733,
      "loss": 2.3117,
      "step": 76299
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.8042493462562561,
      "learning_rate": 0.00045177129418207747,
      "loss": 2.0231,
      "step": 76300
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.8481500148773193,
      "learning_rate": 0.0004517677656915646,
      "loss": 2.0495,
      "step": 76301
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.8191068172454834,
      "learning_rate": 0.00045176423717283525,
      "loss": 2.1361,
      "step": 76302
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.8167341351509094,
      "learning_rate": 0.00045176070862589,
      "loss": 2.2883,
      "step": 76303
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.7896071076393127,
      "learning_rate": 0.00045175718005072954,
      "loss": 2.2159,
      "step": 76304
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.7812807559967041,
      "learning_rate": 0.0004517536514473547,
      "loss": 2.2701,
      "step": 76305
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.8529189229011536,
      "learning_rate": 0.00045175012281576586,
      "loss": 2.233,
      "step": 76306
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.7498882412910461,
      "learning_rate": 0.00045174659415596384,
      "loss": 2.2573,
      "step": 76307
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.8090516328811646,
      "learning_rate": 0.0004517430654679493,
      "loss": 2.2942,
      "step": 76308
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.8176491260528564,
      "learning_rate": 0.0004517395367517228,
      "loss": 2.4084,
      "step": 76309
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.8138307929039001,
      "learning_rate": 0.0004517360080072851,
      "loss": 2.2237,
      "step": 76310
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.8181275725364685,
      "learning_rate": 0.0004517324792346368,
      "loss": 2.1592,
      "step": 76311
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.8928412199020386,
      "learning_rate": 0.0004517289504337786,
      "loss": 2.3832,
      "step": 76312
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.7937821745872498,
      "learning_rate": 0.0004517254216047111,
      "loss": 2.1506,
      "step": 76313
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.9915840029716492,
      "learning_rate": 0.0004517218927474349,
      "loss": 2.0049,
      "step": 76314
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.8310959339141846,
      "learning_rate": 0.00045171836386195086,
      "loss": 2.289,
      "step": 76315
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.7881278991699219,
      "learning_rate": 0.0004517148349482595,
      "loss": 2.0897,
      "step": 76316
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.7473865747451782,
      "learning_rate": 0.00045171130600636135,
      "loss": 2.1052,
      "step": 76317
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.7873269319534302,
      "learning_rate": 0.0004517077770362573,
      "loss": 2.2377,
      "step": 76318
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.7981058359146118,
      "learning_rate": 0.00045170424803794797,
      "loss": 2.4052,
      "step": 76319
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.8381962180137634,
      "learning_rate": 0.0004517007190114339,
      "loss": 2.1093,
      "step": 76320
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.8249973654747009,
      "learning_rate": 0.0004516971899567157,
      "loss": 2.3646,
      "step": 76321
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.80393385887146,
      "learning_rate": 0.0004516936608737942,
      "loss": 2.1522,
      "step": 76322
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.9790559411048889,
      "learning_rate": 0.0004516901317626701,
      "loss": 2.0866,
      "step": 76323
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.8673962950706482,
      "learning_rate": 0.00045168660262334383,
      "loss": 2.41,
      "step": 76324
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.8490657210350037,
      "learning_rate": 0.00045168307345581617,
      "loss": 2.1801,
      "step": 76325
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.9466840028762817,
      "learning_rate": 0.0004516795442600878,
      "loss": 1.9263,
      "step": 76326
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.7138875722885132,
      "learning_rate": 0.00045167601503615936,
      "loss": 1.9811,
      "step": 76327
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.8223708868026733,
      "learning_rate": 0.0004516724857840314,
      "loss": 2.1949,
      "step": 76328
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.8414797782897949,
      "learning_rate": 0.00045166895650370476,
      "loss": 2.1559,
      "step": 76329
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.9079170227050781,
      "learning_rate": 0.0004516654271951799,
      "loss": 2.2864,
      "step": 76330
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.7870264649391174,
      "learning_rate": 0.0004516618978584576,
      "loss": 2.2244,
      "step": 76331
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.9490258693695068,
      "learning_rate": 0.0004516583684935386,
      "loss": 2.1135,
      "step": 76332
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.9926561117172241,
      "learning_rate": 0.00045165483910042327,
      "loss": 2.4127,
      "step": 76333
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.8973520994186401,
      "learning_rate": 0.00045165130967911256,
      "loss": 1.9242,
      "step": 76334
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.8409013748168945,
      "learning_rate": 0.000451647780229607,
      "loss": 2.0685,
      "step": 76335
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.101239800453186,
      "learning_rate": 0.00045164425075190715,
      "loss": 2.3844,
      "step": 76336
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.8962370157241821,
      "learning_rate": 0.0004516407212460139,
      "loss": 2.2102,
      "step": 76337
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.7774068713188171,
      "learning_rate": 0.00045163719171192776,
      "loss": 2.4905,
      "step": 76338
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.186979055404663,
      "learning_rate": 0.00045163366214964935,
      "loss": 2.4593,
      "step": 76339
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.984521210193634,
      "learning_rate": 0.0004516301325591794,
      "loss": 2.1165,
      "step": 76340
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.8005222082138062,
      "learning_rate": 0.0004516266029405186,
      "loss": 2.3034,
      "step": 76341
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.9182636737823486,
      "learning_rate": 0.0004516230732936675,
      "loss": 2.325,
      "step": 76342
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.1951911449432373,
      "learning_rate": 0.00045161954361862673,
      "loss": 2.2183,
      "step": 76343
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.9845441579818726,
      "learning_rate": 0.0004516160139153972,
      "loss": 2.1498,
      "step": 76344
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.8919329047203064,
      "learning_rate": 0.0004516124841839793,
      "loss": 2.3927,
      "step": 76345
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.14716374874115,
      "learning_rate": 0.0004516089544243737,
      "loss": 2.3233,
      "step": 76346
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.9981261491775513,
      "learning_rate": 0.00045160542463658127,
      "loss": 1.8025,
      "step": 76347
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.8319882154464722,
      "learning_rate": 0.0004516018948206025,
      "loss": 2.3271,
      "step": 76348
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.1022170782089233,
      "learning_rate": 0.00045159836497643796,
      "loss": 2.308,
      "step": 76349
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.3871769905090332,
      "learning_rate": 0.00045159483510408857,
      "loss": 2.186,
      "step": 76350
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.8642013669013977,
      "learning_rate": 0.00045159130520355475,
      "loss": 2.3525,
      "step": 76351
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.8705689311027527,
      "learning_rate": 0.0004515877752748372,
      "loss": 2.3242,
      "step": 76352
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.859992265701294,
      "learning_rate": 0.0004515842453179368,
      "loss": 2.0805,
      "step": 76353
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.9527873992919922,
      "learning_rate": 0.00045158071533285386,
      "loss": 2.378,
      "step": 76354
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.8612698316574097,
      "learning_rate": 0.00045157718531958925,
      "loss": 2.243,
      "step": 76355
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.8021193146705627,
      "learning_rate": 0.00045157365527814363,
      "loss": 2.2048,
      "step": 76356
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.8872320055961609,
      "learning_rate": 0.00045157012520851753,
      "loss": 2.1263,
      "step": 76357
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.8907502293586731,
      "learning_rate": 0.0004515665951107117,
      "loss": 2.1454,
      "step": 76358
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.7577201724052429,
      "learning_rate": 0.0004515630649847269,
      "loss": 2.3462,
      "step": 76359
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.7927311062812805,
      "learning_rate": 0.0004515595348305635,
      "loss": 2.3543,
      "step": 76360
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.8268691897392273,
      "learning_rate": 0.0004515560046482224,
      "loss": 1.9343,
      "step": 76361
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.0279476642608643,
      "learning_rate": 0.0004515524744377042,
      "loss": 2.1582,
      "step": 76362
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.7651705145835876,
      "learning_rate": 0.00045154894419900947,
      "loss": 2.3866,
      "step": 76363
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.93634432554245,
      "learning_rate": 0.0004515454139321389,
      "loss": 2.3558,
      "step": 76364
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.9413875937461853,
      "learning_rate": 0.0004515418836370933,
      "loss": 2.1777,
      "step": 76365
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.7752526998519897,
      "learning_rate": 0.00045153835331387314,
      "loss": 2.2083,
      "step": 76366
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.7406780123710632,
      "learning_rate": 0.0004515348229624791,
      "loss": 2.097,
      "step": 76367
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.7450960874557495,
      "learning_rate": 0.0004515312925829119,
      "loss": 2.3285,
      "step": 76368
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.8460844159126282,
      "learning_rate": 0.0004515277621751723,
      "loss": 2.1218,
      "step": 76369
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.7693671584129333,
      "learning_rate": 0.00045152423173926067,
      "loss": 2.2764,
      "step": 76370
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.7503822445869446,
      "learning_rate": 0.00045152070127517786,
      "loss": 2.2547,
      "step": 76371
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.8372067809104919,
      "learning_rate": 0.0004515171707829245,
      "loss": 2.1202,
      "step": 76372
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.8188750743865967,
      "learning_rate": 0.00045151364026250123,
      "loss": 2.0639,
      "step": 76373
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.767840564250946,
      "learning_rate": 0.00045151010971390866,
      "loss": 2.0649,
      "step": 76374
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.8245471715927124,
      "learning_rate": 0.0004515065791371477,
      "loss": 2.2193,
      "step": 76375
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.8902442455291748,
      "learning_rate": 0.00045150304853221857,
      "loss": 2.119,
      "step": 76376
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.7592784762382507,
      "learning_rate": 0.0004514995178991223,
      "loss": 2.0957,
      "step": 76377
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.9968781471252441,
      "learning_rate": 0.00045149598723785933,
      "loss": 2.119,
      "step": 76378
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.7722684741020203,
      "learning_rate": 0.00045149245654843047,
      "loss": 2.2348,
      "step": 76379
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.8701424598693848,
      "learning_rate": 0.00045148892583083626,
      "loss": 2.3052,
      "step": 76380
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.8539250493049622,
      "learning_rate": 0.0004514853950850774,
      "loss": 2.135,
      "step": 76381
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.8716400265693665,
      "learning_rate": 0.0004514818643111546,
      "loss": 2.2659,
      "step": 76382
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.8036671280860901,
      "learning_rate": 0.0004514783335090684,
      "loss": 2.3227,
      "step": 76383
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.8085566163063049,
      "learning_rate": 0.00045147480267881956,
      "loss": 2.0443,
      "step": 76384
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.8428043127059937,
      "learning_rate": 0.00045147127182040874,
      "loss": 2.3688,
      "step": 76385
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.7595593333244324,
      "learning_rate": 0.0004514677409338364,
      "loss": 2.2308,
      "step": 76386
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.9061349630355835,
      "learning_rate": 0.00045146421001910345,
      "loss": 2.3934,
      "step": 76387
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.8126955032348633,
      "learning_rate": 0.0004514606790762104,
      "loss": 2.1078,
      "step": 76388
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.8023718595504761,
      "learning_rate": 0.00045145714810515796,
      "loss": 2.5305,
      "step": 76389
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.8072147965431213,
      "learning_rate": 0.0004514536171059469,
      "loss": 2.0846,
      "step": 76390
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.7820766568183899,
      "learning_rate": 0.0004514500860785776,
      "loss": 2.2167,
      "step": 76391
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.7463145852088928,
      "learning_rate": 0.0004514465550230509,
      "loss": 2.2101,
      "step": 76392
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.7754225134849548,
      "learning_rate": 0.0004514430239393675,
      "loss": 1.814,
      "step": 76393
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.8266406059265137,
      "learning_rate": 0.0004514394928275279,
      "loss": 2.2245,
      "step": 76394
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.8200145959854126,
      "learning_rate": 0.00045143596168753286,
      "loss": 2.1551,
      "step": 76395
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.854324996471405,
      "learning_rate": 0.0004514324305193831,
      "loss": 2.1283,
      "step": 76396
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.8656816482543945,
      "learning_rate": 0.0004514288993230791,
      "loss": 2.2683,
      "step": 76397
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.7691468000411987,
      "learning_rate": 0.00045142536809862155,
      "loss": 2.2847,
      "step": 76398
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.7331292033195496,
      "learning_rate": 0.00045142183684601124,
      "loss": 2.2002,
      "step": 76399
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.8080973625183105,
      "learning_rate": 0.00045141830556524876,
      "loss": 2.3529,
      "step": 76400
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.957685649394989,
      "learning_rate": 0.00045141477425633474,
      "loss": 2.1894,
      "step": 76401
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.7777099609375,
      "learning_rate": 0.0004514112429192699,
      "loss": 2.2543,
      "step": 76402
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.7809498310089111,
      "learning_rate": 0.00045140771155405474,
      "loss": 2.2084,
      "step": 76403
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.8566798567771912,
      "learning_rate": 0.0004514041801606902,
      "loss": 2.049,
      "step": 76404
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.7675730586051941,
      "learning_rate": 0.0004514006487391766,
      "loss": 2.3597,
      "step": 76405
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.890960156917572,
      "learning_rate": 0.0004513971172895148,
      "loss": 2.2058,
      "step": 76406
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.7393977046012878,
      "learning_rate": 0.0004513935858117056,
      "loss": 2.2925,
      "step": 76407
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.7167373299598694,
      "learning_rate": 0.0004513900543057492,
      "loss": 2.1895,
      "step": 76408
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.7782175540924072,
      "learning_rate": 0.0004513865227716467,
      "loss": 2.2317,
      "step": 76409
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.7653936147689819,
      "learning_rate": 0.00045138299120939853,
      "loss": 2.194,
      "step": 76410
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.8554660677909851,
      "learning_rate": 0.00045137945961900547,
      "loss": 1.8475,
      "step": 76411
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.7380506992340088,
      "learning_rate": 0.000451375928000468,
      "loss": 2.4219,
      "step": 76412
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.7435377836227417,
      "learning_rate": 0.00045137239635378694,
      "loss": 2.3024,
      "step": 76413
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.7553332448005676,
      "learning_rate": 0.000451368864678963,
      "loss": 2.1257,
      "step": 76414
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.782920777797699,
      "learning_rate": 0.0004513653329759966,
      "loss": 2.3911,
      "step": 76415
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.7561960220336914,
      "learning_rate": 0.00045136180124488853,
      "loss": 2.2343,
      "step": 76416
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.7391951084136963,
      "learning_rate": 0.0004513582694856395,
      "loss": 2.372,
      "step": 76417
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.7602595090866089,
      "learning_rate": 0.00045135473769825014,
      "loss": 2.0153,
      "step": 76418
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.8340243697166443,
      "learning_rate": 0.000451351205882721,
      "loss": 2.18,
      "step": 76419
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.8133131265640259,
      "learning_rate": 0.0004513476740390528,
      "loss": 2.165,
      "step": 76420
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.77001953125,
      "learning_rate": 0.0004513441421672462,
      "loss": 2.3903,
      "step": 76421
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.8278497457504272,
      "learning_rate": 0.00045134061026730197,
      "loss": 2.413,
      "step": 76422
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.7984361052513123,
      "learning_rate": 0.00045133707833922064,
      "loss": 2.211,
      "step": 76423
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.8011370897293091,
      "learning_rate": 0.0004513335463830028,
      "loss": 2.2246,
      "step": 76424
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.7960618138313293,
      "learning_rate": 0.0004513300143986492,
      "loss": 2.1887,
      "step": 76425
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.737930953502655,
      "learning_rate": 0.0004513264823861606,
      "loss": 2.1388,
      "step": 76426
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.7799635529518127,
      "learning_rate": 0.00045132295034553755,
      "loss": 1.9815,
      "step": 76427
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.7916164994239807,
      "learning_rate": 0.0004513194182767806,
      "loss": 2.2592,
      "step": 76428
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.7404372096061707,
      "learning_rate": 0.00045131588617989063,
      "loss": 2.0833,
      "step": 76429
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.7934862375259399,
      "learning_rate": 0.0004513123540548681,
      "loss": 2.1378,
      "step": 76430
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.7976888418197632,
      "learning_rate": 0.00045130882190171377,
      "loss": 2.2814,
      "step": 76431
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.8508879542350769,
      "learning_rate": 0.0004513052897204283,
      "loss": 2.0878,
      "step": 76432
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.8428378701210022,
      "learning_rate": 0.0004513017575110123,
      "loss": 2.0884,
      "step": 76433
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.8194971680641174,
      "learning_rate": 0.00045129822527346637,
      "loss": 2.1977,
      "step": 76434
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.8842753767967224,
      "learning_rate": 0.0004512946930077913,
      "loss": 2.2954,
      "step": 76435
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.8179613947868347,
      "learning_rate": 0.00045129116071398773,
      "loss": 2.2106,
      "step": 76436
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.8277419805526733,
      "learning_rate": 0.00045128762839205625,
      "loss": 2.2283,
      "step": 76437
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.7512231469154358,
      "learning_rate": 0.0004512840960419975,
      "loss": 2.0619,
      "step": 76438
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.7793399691581726,
      "learning_rate": 0.0004512805636638123,
      "loss": 2.1339,
      "step": 76439
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.8304674625396729,
      "learning_rate": 0.0004512770312575011,
      "loss": 2.3501,
      "step": 76440
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.819144606590271,
      "learning_rate": 0.00045127349882306464,
      "loss": 2.2975,
      "step": 76441
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.8956250548362732,
      "learning_rate": 0.00045126996636050363,
      "loss": 2.1925,
      "step": 76442
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.8188141584396362,
      "learning_rate": 0.0004512664338698187,
      "loss": 2.1819,
      "step": 76443
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.7523820996284485,
      "learning_rate": 0.00045126290135101034,
      "loss": 2.3738,
      "step": 76444
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.799468457698822,
      "learning_rate": 0.0004512593688040795,
      "loss": 2.1914,
      "step": 76445
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.8419860601425171,
      "learning_rate": 0.00045125583622902663,
      "loss": 2.2132,
      "step": 76446
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.7606196999549866,
      "learning_rate": 0.00045125230362585246,
      "loss": 2.1703,
      "step": 76447
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.7560197710990906,
      "learning_rate": 0.00045124877099455765,
      "loss": 2.1572,
      "step": 76448
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.7542285919189453,
      "learning_rate": 0.0004512452383351428,
      "loss": 2.2919,
      "step": 76449
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.7583093047142029,
      "learning_rate": 0.00045124170564760855,
      "loss": 2.0182,
      "step": 76450
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.7535253763198853,
      "learning_rate": 0.0004512381729319558,
      "loss": 2.2391,
      "step": 76451
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.7769712209701538,
      "learning_rate": 0.00045123464018818475,
      "loss": 2.2166,
      "step": 76452
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.8286806344985962,
      "learning_rate": 0.00045123110741629647,
      "loss": 2.3984,
      "step": 76453
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.803320586681366,
      "learning_rate": 0.00045122757461629157,
      "loss": 2.2158,
      "step": 76454
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.7554879188537598,
      "learning_rate": 0.0004512240417881705,
      "loss": 2.1882,
      "step": 76455
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.8994541168212891,
      "learning_rate": 0.00045122050893193404,
      "loss": 2.1838,
      "step": 76456
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.800081729888916,
      "learning_rate": 0.0004512169760475828,
      "loss": 2.262,
      "step": 76457
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.8048927187919617,
      "learning_rate": 0.00045121344313511747,
      "loss": 2.0911,
      "step": 76458
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.7721772193908691,
      "learning_rate": 0.00045120991019453877,
      "loss": 2.2533,
      "step": 76459
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.7931172251701355,
      "learning_rate": 0.00045120637722584724,
      "loss": 2.2903,
      "step": 76460
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.7643287181854248,
      "learning_rate": 0.0004512028442290436,
      "loss": 2.314,
      "step": 76461
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.7841673493385315,
      "learning_rate": 0.00045119931120412855,
      "loss": 2.1659,
      "step": 76462
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.7724550366401672,
      "learning_rate": 0.0004511957781511027,
      "loss": 2.0281,
      "step": 76463
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.7341964244842529,
      "learning_rate": 0.00045119224506996665,
      "loss": 2.033,
      "step": 76464
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.7525367140769958,
      "learning_rate": 0.00045118871196072105,
      "loss": 2.2141,
      "step": 76465
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.7765758037567139,
      "learning_rate": 0.00045118517882336665,
      "loss": 1.9593,
      "step": 76466
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.7639602422714233,
      "learning_rate": 0.00045118164565790414,
      "loss": 2.271,
      "step": 76467
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.7815910577774048,
      "learning_rate": 0.00045117811246433407,
      "loss": 2.3269,
      "step": 76468
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.799179196357727,
      "learning_rate": 0.0004511745792426571,
      "loss": 2.2246,
      "step": 76469
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.6867390275001526,
      "learning_rate": 0.00045117104599287395,
      "loss": 2.2479,
      "step": 76470
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.7385300993919373,
      "learning_rate": 0.00045116751271498525,
      "loss": 2.0906,
      "step": 76471
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.1145000457763672,
      "learning_rate": 0.0004511639794089916,
      "loss": 2.2091,
      "step": 76472
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.8024569749832153,
      "learning_rate": 0.0004511604460748938,
      "loss": 2.2435,
      "step": 76473
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.7899343967437744,
      "learning_rate": 0.00045115691271269245,
      "loss": 2.1814,
      "step": 76474
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.778645396232605,
      "learning_rate": 0.0004511533793223881,
      "loss": 2.35,
      "step": 76475
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.7736061811447144,
      "learning_rate": 0.0004511498459039815,
      "loss": 2.3118,
      "step": 76476
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.7693304419517517,
      "learning_rate": 0.00045114631245747323,
      "loss": 2.2307,
      "step": 76477
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.7491952180862427,
      "learning_rate": 0.00045114277898286413,
      "loss": 2.0328,
      "step": 76478
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.7942433953285217,
      "learning_rate": 0.00045113924548015463,
      "loss": 2.3411,
      "step": 76479
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.716672956943512,
      "learning_rate": 0.00045113571194934553,
      "loss": 2.2104,
      "step": 76480
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.7582605481147766,
      "learning_rate": 0.00045113217839043754,
      "loss": 2.2855,
      "step": 76481
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.7673553228378296,
      "learning_rate": 0.00045112864480343105,
      "loss": 2.2296,
      "step": 76482
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.7267617583274841,
      "learning_rate": 0.00045112511118832697,
      "loss": 2.3006,
      "step": 76483
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.7619279623031616,
      "learning_rate": 0.00045112157754512584,
      "loss": 2.2906,
      "step": 76484
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.7489840388298035,
      "learning_rate": 0.0004511180438738285,
      "loss": 2.223,
      "step": 76485
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.7530876398086548,
      "learning_rate": 0.0004511145101744353,
      "loss": 2.3117,
      "step": 76486
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.7348623871803284,
      "learning_rate": 0.0004511109764469471,
      "loss": 2.0525,
      "step": 76487
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.7363554239273071,
      "learning_rate": 0.00045110744269136465,
      "loss": 2.2457,
      "step": 76488
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.7425145506858826,
      "learning_rate": 0.0004511039089076883,
      "loss": 2.2611,
      "step": 76489
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.7883252501487732,
      "learning_rate": 0.00045110037509591897,
      "loss": 2.2323,
      "step": 76490
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.7444888353347778,
      "learning_rate": 0.00045109684125605717,
      "loss": 2.109,
      "step": 76491
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.8069875240325928,
      "learning_rate": 0.0004510933073881037,
      "loss": 2.2742,
      "step": 76492
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.749776303768158,
      "learning_rate": 0.00045108977349205904,
      "loss": 2.1636,
      "step": 76493
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.8717110753059387,
      "learning_rate": 0.000451086239567924,
      "loss": 2.1748,
      "step": 76494
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.739268958568573,
      "learning_rate": 0.00045108270561569906,
      "loss": 1.9247,
      "step": 76495
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.7255337834358215,
      "learning_rate": 0.0004510791716353852,
      "loss": 2.2172,
      "step": 76496
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.8574029803276062,
      "learning_rate": 0.00045107563762698274,
      "loss": 2.2803,
      "step": 76497
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.8762497305870056,
      "learning_rate": 0.0004510721035904924,
      "loss": 2.3856,
      "step": 76498
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.8037955164909363,
      "learning_rate": 0.00045106856952591514,
      "loss": 2.2672,
      "step": 76499
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.7785753011703491,
      "learning_rate": 0.0004510650354332511,
      "loss": 2.1217,
      "step": 76500
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.7276690006256104,
      "learning_rate": 0.0004510615013125014,
      "loss": 2.3086,
      "step": 76501
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.981631875038147,
      "learning_rate": 0.00045105796716366647,
      "loss": 2.0079,
      "step": 76502
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.7939454913139343,
      "learning_rate": 0.00045105443298674704,
      "loss": 2.129,
      "step": 76503
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.7414097189903259,
      "learning_rate": 0.0004510508987817437,
      "loss": 2.133,
      "step": 76504
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.8818597197532654,
      "learning_rate": 0.00045104736454865715,
      "loss": 2.1973,
      "step": 76505
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.8476542234420776,
      "learning_rate": 0.0004510438302874881,
      "loss": 2.0857,
      "step": 76506
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.7725737690925598,
      "learning_rate": 0.00045104029599823703,
      "loss": 2.1583,
      "step": 76507
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.8398645520210266,
      "learning_rate": 0.00045103676168090475,
      "loss": 2.101,
      "step": 76508
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.8948982954025269,
      "learning_rate": 0.00045103322733549195,
      "loss": 2.2799,
      "step": 76509
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.7774834632873535,
      "learning_rate": 0.00045102969296199923,
      "loss": 2.0309,
      "step": 76510
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.8485836386680603,
      "learning_rate": 0.0004510261585604272,
      "loss": 2.4541,
      "step": 76511
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.8477866053581238,
      "learning_rate": 0.0004510226241307765,
      "loss": 2.0013,
      "step": 76512
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.8496161103248596,
      "learning_rate": 0.00045101908967304794,
      "loss": 2.1362,
      "step": 76513
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.7972075343132019,
      "learning_rate": 0.00045101555518724197,
      "loss": 2.2377,
      "step": 76514
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.7273058295249939,
      "learning_rate": 0.00045101202067335955,
      "loss": 2.0587,
      "step": 76515
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.9938647150993347,
      "learning_rate": 0.00045100848613140106,
      "loss": 1.9452,
      "step": 76516
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.8795552849769592,
      "learning_rate": 0.00045100495156136716,
      "loss": 2.3242,
      "step": 76517
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.9492216110229492,
      "learning_rate": 0.00045100141696325865,
      "loss": 2.1907,
      "step": 76518
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.142628788948059,
      "learning_rate": 0.0004509978823370761,
      "loss": 2.3225,
      "step": 76519
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.7580457329750061,
      "learning_rate": 0.0004509943476828202,
      "loss": 2.2577,
      "step": 76520
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.7620386481285095,
      "learning_rate": 0.00045099081300049165,
      "loss": 2.2425,
      "step": 76521
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.86038738489151,
      "learning_rate": 0.00045098727829009103,
      "loss": 2.2107,
      "step": 76522
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.9965804815292358,
      "learning_rate": 0.00045098374355161895,
      "loss": 2.1466,
      "step": 76523
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.7388913631439209,
      "learning_rate": 0.00045098020878507627,
      "loss": 2.1509,
      "step": 76524
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.8071736097335815,
      "learning_rate": 0.0004509766739904634,
      "loss": 2.293,
      "step": 76525
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.9930196404457092,
      "learning_rate": 0.0004509731391677811,
      "loss": 2.2005,
      "step": 76526
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.7840842008590698,
      "learning_rate": 0.0004509696043170302,
      "loss": 2.3488,
      "step": 76527
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.7483385801315308,
      "learning_rate": 0.00045096606943821106,
      "loss": 2.3307,
      "step": 76528
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.9530362486839294,
      "learning_rate": 0.00045096253453132454,
      "loss": 2.4169,
      "step": 76529
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.8413047790527344,
      "learning_rate": 0.0004509589995963712,
      "loss": 2.0717,
      "step": 76530
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.7300122976303101,
      "learning_rate": 0.0004509554646333518,
      "loss": 1.9671,
      "step": 76531
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.8195948004722595,
      "learning_rate": 0.0004509519296422669,
      "loss": 2.09,
      "step": 76532
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.726280927658081,
      "learning_rate": 0.0004509483946231172,
      "loss": 2.2883,
      "step": 76533
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.8109886050224304,
      "learning_rate": 0.0004509448595759033,
      "loss": 2.1306,
      "step": 76534
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.7634851336479187,
      "learning_rate": 0.0004509413245006259,
      "loss": 2.5161,
      "step": 76535
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.7606356143951416,
      "learning_rate": 0.0004509377893972857,
      "loss": 2.3242,
      "step": 76536
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.7475857138633728,
      "learning_rate": 0.0004509342542658833,
      "loss": 2.0779,
      "step": 76537
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.9079414010047913,
      "learning_rate": 0.0004509307191064193,
      "loss": 2.1533,
      "step": 76538
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.8278067708015442,
      "learning_rate": 0.00045092718391889455,
      "loss": 2.181,
      "step": 76539
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.7452874779701233,
      "learning_rate": 0.00045092364870330954,
      "loss": 2.2062,
      "step": 76540
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.7939519882202148,
      "learning_rate": 0.000450920113459665,
      "loss": 2.4656,
      "step": 76541
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.8237813115119934,
      "learning_rate": 0.00045091657818796153,
      "loss": 2.1706,
      "step": 76542
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.8000304102897644,
      "learning_rate": 0.0004509130428881998,
      "loss": 2.1627,
      "step": 76543
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.7587438821792603,
      "learning_rate": 0.0004509095075603805,
      "loss": 2.2669,
      "step": 76544
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.791410505771637,
      "learning_rate": 0.0004509059722045043,
      "loss": 2.0907,
      "step": 76545
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.8185046315193176,
      "learning_rate": 0.00045090243682057183,
      "loss": 2.2338,
      "step": 76546
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.800327479839325,
      "learning_rate": 0.0004508989014085837,
      "loss": 2.0575,
      "step": 76547
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.7339216470718384,
      "learning_rate": 0.0004508953659685407,
      "loss": 2.0074,
      "step": 76548
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.7006804347038269,
      "learning_rate": 0.0004508918305004433,
      "loss": 2.1765,
      "step": 76549
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.9539844989776611,
      "learning_rate": 0.0004508882950042923,
      "loss": 2.2712,
      "step": 76550
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.7968529462814331,
      "learning_rate": 0.0004508847594800884,
      "loss": 2.1828,
      "step": 76551
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.0900253057479858,
      "learning_rate": 0.0004508812239278321,
      "loss": 2.1704,
      "step": 76552
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.9128443002700806,
      "learning_rate": 0.0004508776883475241,
      "loss": 2.049,
      "step": 76553
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.9927265048027039,
      "learning_rate": 0.00045087415273916515,
      "loss": 2.3516,
      "step": 76554
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.993618369102478,
      "learning_rate": 0.00045087061710275575,
      "loss": 2.0249,
      "step": 76555
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.8282512426376343,
      "learning_rate": 0.00045086708143829673,
      "loss": 2.2468,
      "step": 76556
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.8349514007568359,
      "learning_rate": 0.0004508635457457887,
      "loss": 2.1506,
      "step": 76557
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.0850423574447632,
      "learning_rate": 0.00045086001002523225,
      "loss": 2.2703,
      "step": 76558
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.8527348041534424,
      "learning_rate": 0.0004508564742766281,
      "loss": 2.1077,
      "step": 76559
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.8076640963554382,
      "learning_rate": 0.00045085293849997686,
      "loss": 2.1951,
      "step": 76560
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.8385586142539978,
      "learning_rate": 0.00045084940269527927,
      "loss": 1.9351,
      "step": 76561
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.1777535676956177,
      "learning_rate": 0.00045084586686253587,
      "loss": 2.1947,
      "step": 76562
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.8315787315368652,
      "learning_rate": 0.00045084233100174735,
      "loss": 2.241,
      "step": 76563
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.049425482749939,
      "learning_rate": 0.00045083879511291443,
      "loss": 2.1191,
      "step": 76564
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.9411253333091736,
      "learning_rate": 0.00045083525919603775,
      "loss": 2.0665,
      "step": 76565
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.9360970258712769,
      "learning_rate": 0.0004508317232511179,
      "loss": 2.2151,
      "step": 76566
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.7581412196159363,
      "learning_rate": 0.0004508281872781556,
      "loss": 2.1726,
      "step": 76567
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.8041083216667175,
      "learning_rate": 0.00045082465127715157,
      "loss": 2.293,
      "step": 76568
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.8682152032852173,
      "learning_rate": 0.00045082111524810626,
      "loss": 2.3179,
      "step": 76569
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.7820355296134949,
      "learning_rate": 0.0004508175791910206,
      "loss": 2.15,
      "step": 76570
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.8113682866096497,
      "learning_rate": 0.00045081404310589495,
      "loss": 2.2093,
      "step": 76571
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.743645429611206,
      "learning_rate": 0.0004508105069927302,
      "loss": 2.2494,
      "step": 76572
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.8008224368095398,
      "learning_rate": 0.00045080697085152693,
      "loss": 2.09,
      "step": 76573
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.8319821953773499,
      "learning_rate": 0.00045080343468228577,
      "loss": 2.4371,
      "step": 76574
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.7206936478614807,
      "learning_rate": 0.00045079989848500747,
      "loss": 2.1317,
      "step": 76575
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.8263487219810486,
      "learning_rate": 0.00045079636225969257,
      "loss": 2.2622,
      "step": 76576
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.7376995086669922,
      "learning_rate": 0.0004507928260063418,
      "loss": 2.2745,
      "step": 76577
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.7876344919204712,
      "learning_rate": 0.00045078928972495583,
      "loss": 2.0505,
      "step": 76578
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.7966344952583313,
      "learning_rate": 0.0004507857534155352,
      "loss": 2.1659,
      "step": 76579
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.8535372018814087,
      "learning_rate": 0.0004507822170780807,
      "loss": 2.1103,
      "step": 76580
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.7777372598648071,
      "learning_rate": 0.00045077868071259294,
      "loss": 2.097,
      "step": 76581
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.8103060722351074,
      "learning_rate": 0.00045077514431907253,
      "loss": 2.3428,
      "step": 76582
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.8321788907051086,
      "learning_rate": 0.0004507716078975203,
      "loss": 2.3227,
      "step": 76583
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.8289769887924194,
      "learning_rate": 0.0004507680714479367,
      "loss": 2.2238,
      "step": 76584
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.7940512895584106,
      "learning_rate": 0.00045076453497032247,
      "loss": 2.3087,
      "step": 76585
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.7954626679420471,
      "learning_rate": 0.0004507609984646782,
      "loss": 2.1331,
      "step": 76586
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.7777354717254639,
      "learning_rate": 0.00045075746193100477,
      "loss": 1.95,
      "step": 76587
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.8489269614219666,
      "learning_rate": 0.00045075392536930253,
      "loss": 2.2836,
      "step": 76588
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.7876312732696533,
      "learning_rate": 0.00045075038877957237,
      "loss": 2.265,
      "step": 76589
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.828273594379425,
      "learning_rate": 0.0004507468521618148,
      "loss": 2.2966,
      "step": 76590
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.8628455996513367,
      "learning_rate": 0.00045074331551603065,
      "loss": 2.3793,
      "step": 76591
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.7832009792327881,
      "learning_rate": 0.00045073977884222045,
      "loss": 2.1686,
      "step": 76592
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.8205504417419434,
      "learning_rate": 0.0004507362421403848,
      "loss": 2.0485,
      "step": 76593
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.8741188049316406,
      "learning_rate": 0.0004507327054105245,
      "loss": 2.1476,
      "step": 76594
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.8223587870597839,
      "learning_rate": 0.0004507291686526401,
      "loss": 2.3447,
      "step": 76595
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.7606415152549744,
      "learning_rate": 0.0004507256318667323,
      "loss": 2.1418,
      "step": 76596
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.737568736076355,
      "learning_rate": 0.0004507220950528019,
      "loss": 2.3143,
      "step": 76597
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.8409773111343384,
      "learning_rate": 0.00045071855821084925,
      "loss": 2.1193,
      "step": 76598
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.7530984282493591,
      "learning_rate": 0.00045071502134087527,
      "loss": 2.1377,
      "step": 76599
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.752531886100769,
      "learning_rate": 0.0004507114844428805,
      "loss": 2.2765,
      "step": 76600
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.8001223206520081,
      "learning_rate": 0.00045070794751686554,
      "loss": 2.1373,
      "step": 76601
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.8518093228340149,
      "learning_rate": 0.00045070441056283125,
      "loss": 2.0335,
      "step": 76602
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.768846333026886,
      "learning_rate": 0.00045070087358077814,
      "loss": 2.2301,
      "step": 76603
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.8011882901191711,
      "learning_rate": 0.0004506973365707068,
      "loss": 2.3898,
      "step": 76604
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.8108850717544556,
      "learning_rate": 0.00045069379953261807,
      "loss": 2.3138,
      "step": 76605
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.8648889660835266,
      "learning_rate": 0.00045069026246651246,
      "loss": 2.3016,
      "step": 76606
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.8519872426986694,
      "learning_rate": 0.00045068672537239073,
      "loss": 2.192,
      "step": 76607
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.7609586119651794,
      "learning_rate": 0.0004506831882502535,
      "loss": 2.1939,
      "step": 76608
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.791716456413269,
      "learning_rate": 0.00045067965110010146,
      "loss": 2.2295,
      "step": 76609
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.7870810031890869,
      "learning_rate": 0.0004506761139219351,
      "loss": 2.1739,
      "step": 76610
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.9079741835594177,
      "learning_rate": 0.00045067257671575523,
      "loss": 2.2568,
      "step": 76611
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.9284166693687439,
      "learning_rate": 0.0004506690394815626,
      "loss": 2.1313,
      "step": 76612
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.8178247213363647,
      "learning_rate": 0.0004506655022193577,
      "loss": 2.2737,
      "step": 76613
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.910822868347168,
      "learning_rate": 0.00045066196492914113,
      "loss": 2.3547,
      "step": 76614
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.7654383778572083,
      "learning_rate": 0.0004506584276109139,
      "loss": 2.1833,
      "step": 76615
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.7763001322746277,
      "learning_rate": 0.00045065489026467625,
      "loss": 2.3649,
      "step": 76616
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.019843339920044,
      "learning_rate": 0.00045065135289042895,
      "loss": 2.3805,
      "step": 76617
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.1064262390136719,
      "learning_rate": 0.00045064781548817285,
      "loss": 2.2579,
      "step": 76618
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.7624050974845886,
      "learning_rate": 0.0004506442780579084,
      "loss": 2.1064,
      "step": 76619
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.8987152576446533,
      "learning_rate": 0.0004506407405996364,
      "loss": 2.2205,
      "step": 76620
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.9598727226257324,
      "learning_rate": 0.0004506372031133574,
      "loss": 2.2035,
      "step": 76621
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.9477541446685791,
      "learning_rate": 0.0004506336655990721,
      "loss": 1.923,
      "step": 76622
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.8799256086349487,
      "learning_rate": 0.00045063012805678115,
      "loss": 2.3159,
      "step": 76623
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.2310844659805298,
      "learning_rate": 0.00045062659048648526,
      "loss": 2.1268,
      "step": 76624
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.8838638663291931,
      "learning_rate": 0.00045062305288818503,
      "loss": 2.1769,
      "step": 76625
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.7129701375961304,
      "learning_rate": 0.00045061951526188104,
      "loss": 2.3328,
      "step": 76626
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.8538214564323425,
      "learning_rate": 0.00045061597760757423,
      "loss": 2.1945,
      "step": 76627
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.0188217163085938,
      "learning_rate": 0.0004506124399252649,
      "loss": 2.303,
      "step": 76628
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.8141896724700928,
      "learning_rate": 0.0004506089022149539,
      "loss": 2.2542,
      "step": 76629
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.7454819679260254,
      "learning_rate": 0.000450605364476642,
      "loss": 2.04,
      "step": 76630
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.7941980957984924,
      "learning_rate": 0.00045060182671032955,
      "loss": 2.5611,
      "step": 76631
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.3484413623809814,
      "learning_rate": 0.0004505982889160174,
      "loss": 2.1826,
      "step": 76632
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.9323899149894714,
      "learning_rate": 0.0004505947510937063,
      "loss": 2.0933,
      "step": 76633
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.8148136138916016,
      "learning_rate": 0.00045059121324339665,
      "loss": 2.0301,
      "step": 76634
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.9195563793182373,
      "learning_rate": 0.00045058767536508933,
      "loss": 2.1559,
      "step": 76635
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.0981277227401733,
      "learning_rate": 0.0004505841374587849,
      "loss": 2.3141,
      "step": 76636
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.7967213988304138,
      "learning_rate": 0.0004505805995244841,
      "loss": 2.0841,
      "step": 76637
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.8583163022994995,
      "learning_rate": 0.00045057706156218744,
      "loss": 2.1187,
      "step": 76638
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.0133798122406006,
      "learning_rate": 0.0004505735235718957,
      "loss": 2.005,
      "step": 76639
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.7752854824066162,
      "learning_rate": 0.0004505699855536095,
      "loss": 2.2059,
      "step": 76640
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.7834827899932861,
      "learning_rate": 0.0004505664475073294,
      "loss": 2.4965,
      "step": 76641
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.8574358820915222,
      "learning_rate": 0.00045056290943305626,
      "loss": 2.1397,
      "step": 76642
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.7838910222053528,
      "learning_rate": 0.00045055937133079067,
      "loss": 2.1935,
      "step": 76643
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.7723244428634644,
      "learning_rate": 0.00045055583320053314,
      "loss": 2.1,
      "step": 76644
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.745094895362854,
      "learning_rate": 0.00045055229504228445,
      "loss": 2.2429,
      "step": 76645
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.797971785068512,
      "learning_rate": 0.0004505487568560454,
      "loss": 2.2245,
      "step": 76646
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.8011969327926636,
      "learning_rate": 0.0004505452186418163,
      "loss": 2.1746,
      "step": 76647
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.8444758057594299,
      "learning_rate": 0.0004505416803995981,
      "loss": 2.3079,
      "step": 76648
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.6803216338157654,
      "learning_rate": 0.00045053814212939143,
      "loss": 2.0492,
      "step": 76649
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.7662796974182129,
      "learning_rate": 0.0004505346038311968,
      "loss": 2.3422,
      "step": 76650
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.779892086982727,
      "learning_rate": 0.0004505310655050149,
      "loss": 2.1899,
      "step": 76651
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.7535197734832764,
      "learning_rate": 0.0004505275271508465,
      "loss": 2.0376,
      "step": 76652
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.7506579756736755,
      "learning_rate": 0.0004505239887686921,
      "loss": 2.0647,
      "step": 76653
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.7941493988037109,
      "learning_rate": 0.00045052045035855247,
      "loss": 1.9462,
      "step": 76654
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.8029659986495972,
      "learning_rate": 0.00045051691192042837,
      "loss": 2.1289,
      "step": 76655
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.7946928143501282,
      "learning_rate": 0.00045051337345432024,
      "loss": 2.3526,
      "step": 76656
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.8265641927719116,
      "learning_rate": 0.0004505098349602288,
      "loss": 2.4308,
      "step": 76657
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.9061933159828186,
      "learning_rate": 0.0004505062964381549,
      "loss": 2.1301,
      "step": 76658
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.7731772661209106,
      "learning_rate": 0.0004505027578880989,
      "loss": 2.2499,
      "step": 76659
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.881142795085907,
      "learning_rate": 0.00045049921931006163,
      "loss": 2.099,
      "step": 76660
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.7357070446014404,
      "learning_rate": 0.00045049568070404367,
      "loss": 2.2369,
      "step": 76661
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.8687011003494263,
      "learning_rate": 0.0004504921420700457,
      "loss": 2.1413,
      "step": 76662
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.6742978692054749,
      "learning_rate": 0.00045048860340806855,
      "loss": 2.2091,
      "step": 76663
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.805030345916748,
      "learning_rate": 0.0004504850647181125,
      "loss": 2.0831,
      "step": 76664
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.8421409130096436,
      "learning_rate": 0.00045048152600017866,
      "loss": 2.0988,
      "step": 76665
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.8649033308029175,
      "learning_rate": 0.00045047798725426745,
      "loss": 2.2396,
      "step": 76666
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.7845974564552307,
      "learning_rate": 0.0004504744484803794,
      "loss": 2.0368,
      "step": 76667
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.878827691078186,
      "learning_rate": 0.0004504709096785154,
      "loss": 2.2869,
      "step": 76668
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.788392961025238,
      "learning_rate": 0.000450467370848676,
      "loss": 2.2441,
      "step": 76669
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.7776094675064087,
      "learning_rate": 0.00045046383199086185,
      "loss": 2.1887,
      "step": 76670
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.7804005146026611,
      "learning_rate": 0.0004504602931050736,
      "loss": 2.2221,
      "step": 76671
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.8683122992515564,
      "learning_rate": 0.00045045675419131206,
      "loss": 2.1798,
      "step": 76672
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.8559226393699646,
      "learning_rate": 0.0004504532152495777,
      "loss": 2.2028,
      "step": 76673
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.7623304724693298,
      "learning_rate": 0.0004504496762798712,
      "loss": 2.2608,
      "step": 76674
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.738126814365387,
      "learning_rate": 0.0004504461372821933,
      "loss": 1.9903,
      "step": 76675
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.8784909844398499,
      "learning_rate": 0.0004504425982565447,
      "loss": 2.2933,
      "step": 76676
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.8696749210357666,
      "learning_rate": 0.00045043905920292594,
      "loss": 2.1506,
      "step": 76677
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.7659165859222412,
      "learning_rate": 0.0004504355201213376,
      "loss": 2.1713,
      "step": 76678
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.8128592371940613,
      "learning_rate": 0.0004504319810117806,
      "loss": 1.982,
      "step": 76679
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.9968361854553223,
      "learning_rate": 0.0004504284418742554,
      "loss": 2.0681,
      "step": 76680
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.846132755279541,
      "learning_rate": 0.00045042490270876277,
      "loss": 2.0893,
      "step": 76681
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.8902668356895447,
      "learning_rate": 0.00045042136351530326,
      "loss": 2.2893,
      "step": 76682
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.7590675354003906,
      "learning_rate": 0.00045041782429387756,
      "loss": 2.0169,
      "step": 76683
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.9899193048477173,
      "learning_rate": 0.00045041428504448644,
      "loss": 2.0837,
      "step": 76684
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.7833158373832703,
      "learning_rate": 0.00045041074576713044,
      "loss": 2.3151,
      "step": 76685
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.7757771611213684,
      "learning_rate": 0.00045040720646181016,
      "loss": 2.2306,
      "step": 76686
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.1437628269195557,
      "learning_rate": 0.00045040366712852635,
      "loss": 2.1251,
      "step": 76687
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.7817366123199463,
      "learning_rate": 0.0004504001277672797,
      "loss": 2.2204,
      "step": 76688
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.8060305714607239,
      "learning_rate": 0.0004503965883780708,
      "loss": 2.2262,
      "step": 76689
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.8651816844940186,
      "learning_rate": 0.0004503930489609004,
      "loss": 2.1598,
      "step": 76690
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.0252690315246582,
      "learning_rate": 0.000450389509515769,
      "loss": 2.0611,
      "step": 76691
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.7826311588287354,
      "learning_rate": 0.00045038597004267746,
      "loss": 2.2546,
      "step": 76692
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.7874162793159485,
      "learning_rate": 0.0004503824305416262,
      "loss": 2.2394,
      "step": 76693
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.9713791012763977,
      "learning_rate": 0.00045037889101261615,
      "loss": 2.1737,
      "step": 76694
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.0254086256027222,
      "learning_rate": 0.00045037535145564766,
      "loss": 2.063,
      "step": 76695
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.9736395478248596,
      "learning_rate": 0.0004503718118707217,
      "loss": 2.3813,
      "step": 76696
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.0433450937271118,
      "learning_rate": 0.00045036827225783876,
      "loss": 2.2013,
      "step": 76697
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.8551832437515259,
      "learning_rate": 0.00045036473261699943,
      "loss": 2.1576,
      "step": 76698
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.8131473064422607,
      "learning_rate": 0.00045036119294820454,
      "loss": 1.9779,
      "step": 76699
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.873442530632019,
      "learning_rate": 0.0004503576532514547,
      "loss": 2.2296,
      "step": 76700
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.8974905610084534,
      "learning_rate": 0.0004503541135267504,
      "loss": 2.2556,
      "step": 76701
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.76829594373703,
      "learning_rate": 0.0004503505737740925,
      "loss": 2.1144,
      "step": 76702
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.8290014266967773,
      "learning_rate": 0.0004503470339934816,
      "loss": 2.0942,
      "step": 76703
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.7806669473648071,
      "learning_rate": 0.0004503434941849184,
      "loss": 2.2127,
      "step": 76704
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.7915050983428955,
      "learning_rate": 0.00045033995434840343,
      "loss": 2.2114,
      "step": 76705
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.8320766091346741,
      "learning_rate": 0.0004503364144839374,
      "loss": 2.3053,
      "step": 76706
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.7905815839767456,
      "learning_rate": 0.00045033287459152107,
      "loss": 2.12,
      "step": 76707
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.7253206372261047,
      "learning_rate": 0.00045032933467115506,
      "loss": 2.1729,
      "step": 76708
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.7869235873222351,
      "learning_rate": 0.00045032579472283985,
      "loss": 2.0992,
      "step": 76709
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.7778292894363403,
      "learning_rate": 0.0004503222547465763,
      "loss": 2.185,
      "step": 76710
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.863093376159668,
      "learning_rate": 0.00045031871474236507,
      "loss": 2.3578,
      "step": 76711
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.7831282615661621,
      "learning_rate": 0.0004503151747102066,
      "loss": 2.0747,
      "step": 76712
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.6898410320281982,
      "learning_rate": 0.00045031163465010185,
      "loss": 2.1142,
      "step": 76713
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.7471430897712708,
      "learning_rate": 0.00045030809456205124,
      "loss": 2.1144,
      "step": 76714
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.7864482402801514,
      "learning_rate": 0.00045030455444605553,
      "loss": 2.208,
      "step": 76715
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.8492059111595154,
      "learning_rate": 0.00045030101430211544,
      "loss": 2.1637,
      "step": 76716
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.7303531169891357,
      "learning_rate": 0.0004502974741302315,
      "loss": 1.9364,
      "step": 76717
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.7529029250144958,
      "learning_rate": 0.0004502939339304044,
      "loss": 1.9994,
      "step": 76718
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.7639006972312927,
      "learning_rate": 0.0004502903937026349,
      "loss": 2.1946,
      "step": 76719
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.8261415958404541,
      "learning_rate": 0.0004502868534469235,
      "loss": 1.9454,
      "step": 76720
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.8341881632804871,
      "learning_rate": 0.000450283313163271,
      "loss": 1.9815,
      "step": 76721
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.7760084867477417,
      "learning_rate": 0.000450279772851678,
      "loss": 2.2319,
      "step": 76722
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.7210465669631958,
      "learning_rate": 0.00045027623251214506,
      "loss": 2.2032,
      "step": 76723
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.8185999393463135,
      "learning_rate": 0.000450272692144673,
      "loss": 2.059,
      "step": 76724
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.7751058340072632,
      "learning_rate": 0.0004502691517492624,
      "loss": 2.3912,
      "step": 76725
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.8557022213935852,
      "learning_rate": 0.0004502656113259139,
      "loss": 2.1412,
      "step": 76726
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.8348363637924194,
      "learning_rate": 0.00045026207087462826,
      "loss": 2.2751,
      "step": 76727
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.8276919722557068,
      "learning_rate": 0.00045025853039540597,
      "loss": 2.0091,
      "step": 76728
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.758090078830719,
      "learning_rate": 0.0004502549898882478,
      "loss": 2.1399,
      "step": 76729
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.820761501789093,
      "learning_rate": 0.00045025144935315444,
      "loss": 2.2071,
      "step": 76730
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.7498396635055542,
      "learning_rate": 0.00045024790879012646,
      "loss": 1.9624,
      "step": 76731
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.819852352142334,
      "learning_rate": 0.0004502443681991646,
      "loss": 2.39,
      "step": 76732
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.8560655117034912,
      "learning_rate": 0.0004502408275802694,
      "loss": 2.1607,
      "step": 76733
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.7115378975868225,
      "learning_rate": 0.0004502372869334417,
      "loss": 2.1998,
      "step": 76734
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.8744218945503235,
      "learning_rate": 0.000450233746258682,
      "loss": 2.366,
      "step": 76735
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.8575623631477356,
      "learning_rate": 0.00045023020555599104,
      "loss": 2.2881,
      "step": 76736
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.0154173374176025,
      "learning_rate": 0.0004502266648253694,
      "loss": 2.0717,
      "step": 76737
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.8247232437133789,
      "learning_rate": 0.0004502231240668178,
      "loss": 2.3216,
      "step": 76738
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.9149110317230225,
      "learning_rate": 0.0004502195832803369,
      "loss": 2.1809,
      "step": 76739
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.8019573092460632,
      "learning_rate": 0.00045021604246592734,
      "loss": 2.1661,
      "step": 76740
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.0845922231674194,
      "learning_rate": 0.00045021250162358974,
      "loss": 2.2228,
      "step": 76741
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.7977443337440491,
      "learning_rate": 0.00045020896075332483,
      "loss": 2.2991,
      "step": 76742
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.0071011781692505,
      "learning_rate": 0.0004502054198551333,
      "loss": 2.197,
      "step": 76743
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.855505108833313,
      "learning_rate": 0.0004502018789290156,
      "loss": 2.1222,
      "step": 76744
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.8301329016685486,
      "learning_rate": 0.0004501983379749727,
      "loss": 2.0997,
      "step": 76745
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.8119494318962097,
      "learning_rate": 0.000450194796993005,
      "loss": 2.3261,
      "step": 76746
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.8191336989402771,
      "learning_rate": 0.00045019125598311325,
      "loss": 1.9522,
      "step": 76747
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.9431115984916687,
      "learning_rate": 0.00045018771494529817,
      "loss": 2.2527,
      "step": 76748
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.953816294670105,
      "learning_rate": 0.00045018417387956034,
      "loss": 2.402,
      "step": 76749
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.8443711996078491,
      "learning_rate": 0.0004501806327859004,
      "loss": 2.3314,
      "step": 76750
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.0182833671569824,
      "learning_rate": 0.00045017709166431905,
      "loss": 2.004,
      "step": 76751
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.0090625286102295,
      "learning_rate": 0.00045017355051481695,
      "loss": 2.2991,
      "step": 76752
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.8053706288337708,
      "learning_rate": 0.0004501700093373948,
      "loss": 2.1201,
      "step": 76753
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.893899142742157,
      "learning_rate": 0.0004501664681320531,
      "loss": 2.0508,
      "step": 76754
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.830525815486908,
      "learning_rate": 0.00045016292689879274,
      "loss": 2.0888,
      "step": 76755
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.8984461426734924,
      "learning_rate": 0.0004501593856376142,
      "loss": 2.3519,
      "step": 76756
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.8019222021102905,
      "learning_rate": 0.00045015584434851823,
      "loss": 2.4012,
      "step": 76757
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.7908998131752014,
      "learning_rate": 0.0004501523030315054,
      "loss": 1.9847,
      "step": 76758
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.8009045124053955,
      "learning_rate": 0.00045014876168657643,
      "loss": 2.2481,
      "step": 76759
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.9107602834701538,
      "learning_rate": 0.00045014522031373204,
      "loss": 2.319,
      "step": 76760
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.9245600700378418,
      "learning_rate": 0.0004501416789129727,
      "loss": 2.1543,
      "step": 76761
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.8063071966171265,
      "learning_rate": 0.0004501381374842993,
      "loss": 2.2974,
      "step": 76762
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.8848146200180054,
      "learning_rate": 0.0004501345960277123,
      "loss": 2.2132,
      "step": 76763
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.97856205701828,
      "learning_rate": 0.0004501310545432125,
      "loss": 2.2318,
      "step": 76764
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.8248692154884338,
      "learning_rate": 0.0004501275130308005,
      "loss": 2.0211,
      "step": 76765
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.8216351270675659,
      "learning_rate": 0.00045012397149047696,
      "loss": 2.2465,
      "step": 76766
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.8675010800361633,
      "learning_rate": 0.00045012042992224253,
      "loss": 2.2784,
      "step": 76767
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.7452389001846313,
      "learning_rate": 0.0004501168883260979,
      "loss": 2.1742,
      "step": 76768
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.7196610569953918,
      "learning_rate": 0.0004501133467020436,
      "loss": 2.3208,
      "step": 76769
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.7648992538452148,
      "learning_rate": 0.0004501098050500805,
      "loss": 2.2811,
      "step": 76770
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.8053988814353943,
      "learning_rate": 0.00045010626337020924,
      "loss": 2.2272,
      "step": 76771
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.8138481974601746,
      "learning_rate": 0.0004501027216624302,
      "loss": 2.251,
      "step": 76772
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.7784658670425415,
      "learning_rate": 0.0004500991799267443,
      "loss": 2.2327,
      "step": 76773
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.8798481225967407,
      "learning_rate": 0.0004500956381631522,
      "loss": 2.4057,
      "step": 76774
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.7825412154197693,
      "learning_rate": 0.00045009209637165444,
      "loss": 2.1528,
      "step": 76775
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.9377449750900269,
      "learning_rate": 0.00045008855455225174,
      "loss": 2.297,
      "step": 76776
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.7725374102592468,
      "learning_rate": 0.0004500850127049447,
      "loss": 2.2846,
      "step": 76777
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.8090527057647705,
      "learning_rate": 0.0004500814708297341,
      "loss": 2.4316,
      "step": 76778
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.9531480073928833,
      "learning_rate": 0.0004500779289266204,
      "loss": 2.0084,
      "step": 76779
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.7877481579780579,
      "learning_rate": 0.0004500743869956045,
      "loss": 2.1996,
      "step": 76780
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.7982777953147888,
      "learning_rate": 0.0004500708450366869,
      "loss": 2.3512,
      "step": 76781
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.8077756762504578,
      "learning_rate": 0.00045006730304986825,
      "loss": 2.1876,
      "step": 76782
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.8285584449768066,
      "learning_rate": 0.00045006376103514935,
      "loss": 2.3106,
      "step": 76783
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.6983848214149475,
      "learning_rate": 0.00045006021899253066,
      "loss": 2.113,
      "step": 76784
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.7506278157234192,
      "learning_rate": 0.00045005667692201297,
      "loss": 2.4116,
      "step": 76785
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.8038216233253479,
      "learning_rate": 0.000450053134823597,
      "loss": 2.0304,
      "step": 76786
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.778779149055481,
      "learning_rate": 0.0004500495926972832,
      "loss": 2.3777,
      "step": 76787
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.850354790687561,
      "learning_rate": 0.0004500460505430724,
      "loss": 2.2381,
      "step": 76788
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.8206478357315063,
      "learning_rate": 0.0004500425083609653,
      "loss": 2.2041,
      "step": 76789
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.8060787320137024,
      "learning_rate": 0.0004500389661509623,
      "loss": 2.2714,
      "step": 76790
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.8212447762489319,
      "learning_rate": 0.0004500354239130643,
      "loss": 2.0885,
      "step": 76791
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.7560267448425293,
      "learning_rate": 0.0004500318816472719,
      "loss": 2.1003,
      "step": 76792
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.7266668677330017,
      "learning_rate": 0.0004500283393535857,
      "loss": 2.3258,
      "step": 76793
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.7724506855010986,
      "learning_rate": 0.00045002479703200635,
      "loss": 2.1687,
      "step": 76794
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.7926478385925293,
      "learning_rate": 0.00045002125468253477,
      "loss": 2.2524,
      "step": 76795
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.8058357834815979,
      "learning_rate": 0.0004500177123051712,
      "loss": 2.1255,
      "step": 76796
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.7056899070739746,
      "learning_rate": 0.0004500141698999166,
      "loss": 2.1524,
      "step": 76797
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.7840386033058167,
      "learning_rate": 0.0004500106274667715,
      "loss": 2.3575,
      "step": 76798
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.7632517218589783,
      "learning_rate": 0.0004500070850057366,
      "loss": 2.0381,
      "step": 76799
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.8477819561958313,
      "learning_rate": 0.00045000354251681254,
      "loss": 2.3286,
      "step": 76800
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.0828646421432495,
      "learning_rate": 0.00045,
      "loss": 2.2398,
      "step": 76801
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.8175270557403564,
      "learning_rate": 0.0004499964574552996,
      "loss": 1.9858,
      "step": 76802
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.8186551332473755,
      "learning_rate": 0.00044999291488271204,
      "loss": 2.1651,
      "step": 76803
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.7341175079345703,
      "learning_rate": 0.000449989372282238,
      "loss": 2.2545,
      "step": 76804
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.8065849542617798,
      "learning_rate": 0.00044998582965387806,
      "loss": 2.2121,
      "step": 76805
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.7662988901138306,
      "learning_rate": 0.000449982286997633,
      "loss": 2.1795,
      "step": 76806
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.8345136046409607,
      "learning_rate": 0.0004499787443135033,
      "loss": 2.1059,
      "step": 76807
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.7600079774856567,
      "learning_rate": 0.00044997520160148976,
      "loss": 2.0602,
      "step": 76808
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.7628100514411926,
      "learning_rate": 0.00044997165886159307,
      "loss": 2.276,
      "step": 76809
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.8575999140739441,
      "learning_rate": 0.0004499681160938138,
      "loss": 2.0692,
      "step": 76810
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.7540914416313171,
      "learning_rate": 0.00044996457329815257,
      "loss": 2.1829,
      "step": 76811
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.8301327228546143,
      "learning_rate": 0.00044996103047461,
      "loss": 2.1671,
      "step": 76812
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.7865912914276123,
      "learning_rate": 0.00044995748762318696,
      "loss": 2.2396,
      "step": 76813
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.8631354570388794,
      "learning_rate": 0.00044995394474388404,
      "loss": 2.1358,
      "step": 76814
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.8567230701446533,
      "learning_rate": 0.00044995040183670173,
      "loss": 2.2143,
      "step": 76815
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.7621196508407593,
      "learning_rate": 0.00044994685890164095,
      "loss": 2.2471,
      "step": 76816
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.7764261960983276,
      "learning_rate": 0.00044994331593870213,
      "loss": 1.7811,
      "step": 76817
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.8305420875549316,
      "learning_rate": 0.000449939772947886,
      "loss": 2.2088,
      "step": 76818
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.7590813040733337,
      "learning_rate": 0.0004499362299291933,
      "loss": 2.2364,
      "step": 76819
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.8145157694816589,
      "learning_rate": 0.00044993268688262456,
      "loss": 2.1894,
      "step": 76820
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.7466550469398499,
      "learning_rate": 0.00044992914380818054,
      "loss": 2.4351,
      "step": 76821
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.848891019821167,
      "learning_rate": 0.00044992560070586184,
      "loss": 2.1423,
      "step": 76822
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.7668488025665283,
      "learning_rate": 0.0004499220575756692,
      "loss": 2.2417,
      "step": 76823
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.8087143898010254,
      "learning_rate": 0.00044991851441760313,
      "loss": 2.2608,
      "step": 76824
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.7399218082427979,
      "learning_rate": 0.00044991497123166447,
      "loss": 2.1185,
      "step": 76825
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.7659419178962708,
      "learning_rate": 0.00044991142801785374,
      "loss": 2.2006,
      "step": 76826
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.779452383518219,
      "learning_rate": 0.0004499078847761716,
      "loss": 2.0116,
      "step": 76827
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.9007984399795532,
      "learning_rate": 0.0004499043415066188,
      "loss": 2.2141,
      "step": 76828
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.8262582421302795,
      "learning_rate": 0.000449900798209196,
      "loss": 2.3398,
      "step": 76829
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.793443500995636,
      "learning_rate": 0.00044989725488390374,
      "loss": 2.1691,
      "step": 76830
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.7750304937362671,
      "learning_rate": 0.0004498937115307428,
      "loss": 2.3019,
      "step": 76831
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.7602730989456177,
      "learning_rate": 0.0004498901681497138,
      "loss": 2.1583,
      "step": 76832
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.7625091075897217,
      "learning_rate": 0.0004498866247408173,
      "loss": 2.2151,
      "step": 76833
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.8138971924781799,
      "learning_rate": 0.00044988308130405414,
      "loss": 2.1873,
      "step": 76834
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.7445616722106934,
      "learning_rate": 0.00044987953783942483,
      "loss": 2.1234,
      "step": 76835
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.8014116287231445,
      "learning_rate": 0.00044987599434693007,
      "loss": 2.3251,
      "step": 76836
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.7976281046867371,
      "learning_rate": 0.00044987245082657055,
      "loss": 2.176,
      "step": 76837
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.9360517859458923,
      "learning_rate": 0.00044986890727834704,
      "loss": 2.2645,
      "step": 76838
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.7122043967247009,
      "learning_rate": 0.0004498653637022599,
      "loss": 2.5015,
      "step": 76839
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.7840973138809204,
      "learning_rate": 0.00044986182009831,
      "loss": 2.1999,
      "step": 76840
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.75251704454422,
      "learning_rate": 0.000449858276466498,
      "loss": 2.2048,
      "step": 76841
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.7392685413360596,
      "learning_rate": 0.00044985473280682446,
      "loss": 2.138,
      "step": 76842
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.7501930594444275,
      "learning_rate": 0.00044985118911929013,
      "loss": 2.1268,
      "step": 76843
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.7405192255973816,
      "learning_rate": 0.0004498476454038956,
      "loss": 2.0694,
      "step": 76844
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.8553953766822815,
      "learning_rate": 0.0004498441016606416,
      "loss": 2.2203,
      "step": 76845
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.8662563562393188,
      "learning_rate": 0.0004498405578895288,
      "loss": 2.2932,
      "step": 76846
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.7451642751693726,
      "learning_rate": 0.00044983701409055774,
      "loss": 1.9143,
      "step": 76847
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.8365846276283264,
      "learning_rate": 0.0004498334702637292,
      "loss": 2.1227,
      "step": 76848
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.7480329871177673,
      "learning_rate": 0.0004498299264090438,
      "loss": 1.7845,
      "step": 76849
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.7640714049339294,
      "learning_rate": 0.0004498263825265021,
      "loss": 2.2301,
      "step": 76850
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.8090569972991943,
      "learning_rate": 0.0004498228386161049,
      "loss": 2.15,
      "step": 76851
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.829337477684021,
      "learning_rate": 0.0004498192946778529,
      "loss": 2.2636,
      "step": 76852
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.8188166618347168,
      "learning_rate": 0.0004498157507117465,
      "loss": 2.318,
      "step": 76853
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.7886238098144531,
      "learning_rate": 0.00044981220671778656,
      "loss": 2.1372,
      "step": 76854
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.8501866459846497,
      "learning_rate": 0.0004498086626959737,
      "loss": 2.4405,
      "step": 76855
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.7442622780799866,
      "learning_rate": 0.0004498051186463087,
      "loss": 2.3154,
      "step": 76856
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.8065770864486694,
      "learning_rate": 0.000449801574568792,
      "loss": 2.1395,
      "step": 76857
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.070682168006897,
      "learning_rate": 0.00044979803046342443,
      "loss": 2.2907,
      "step": 76858
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.8465968370437622,
      "learning_rate": 0.0004497944863302065,
      "loss": 2.2765,
      "step": 76859
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.8083667159080505,
      "learning_rate": 0.000449790942169139,
      "loss": 2.0778,
      "step": 76860
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.7963557839393616,
      "learning_rate": 0.00044978739798022245,
      "loss": 2.1518,
      "step": 76861
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.8477439880371094,
      "learning_rate": 0.00044978385376345776,
      "loss": 2.1651,
      "step": 76862
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.7890263199806213,
      "learning_rate": 0.0004497803095188453,
      "loss": 2.1097,
      "step": 76863
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.871334969997406,
      "learning_rate": 0.00044977676524638583,
      "loss": 2.3851,
      "step": 76864
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.7898331880569458,
      "learning_rate": 0.0004497732209460801,
      "loss": 2.2886,
      "step": 76865
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.7657337784767151,
      "learning_rate": 0.00044976967661792866,
      "loss": 1.9242,
      "step": 76866
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.9530994892120361,
      "learning_rate": 0.00044976613226193224,
      "loss": 2.145,
      "step": 76867
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.6945809721946716,
      "learning_rate": 0.00044976258787809157,
      "loss": 1.9325,
      "step": 76868
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.8252794146537781,
      "learning_rate": 0.000449759043466407,
      "loss": 2.2326,
      "step": 76869
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.7476653456687927,
      "learning_rate": 0.00044975549902687953,
      "loss": 2.1826,
      "step": 76870
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.791617751121521,
      "learning_rate": 0.00044975195455950964,
      "loss": 2.2011,
      "step": 76871
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.7597211003303528,
      "learning_rate": 0.00044974841006429803,
      "loss": 2.1049,
      "step": 76872
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.7439458966255188,
      "learning_rate": 0.00044974486554124536,
      "loss": 2.1804,
      "step": 76873
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.7839480042457581,
      "learning_rate": 0.0004497413209903524,
      "loss": 2.0523,
      "step": 76874
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.732347309589386,
      "learning_rate": 0.0004497377764116196,
      "loss": 2.0769,
      "step": 76875
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.749323308467865,
      "learning_rate": 0.0004497342318050476,
      "loss": 2.1123,
      "step": 76876
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.7623950242996216,
      "learning_rate": 0.00044973068717063736,
      "loss": 2.137,
      "step": 76877
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.7616875767707825,
      "learning_rate": 0.0004497271425083894,
      "loss": 2.3124,
      "step": 76878
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.8837000727653503,
      "learning_rate": 0.0004497235978183042,
      "loss": 2.3008,
      "step": 76879
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.7644611597061157,
      "learning_rate": 0.0004497200531003826,
      "loss": 2.0393,
      "step": 76880
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.7914358377456665,
      "learning_rate": 0.00044971650835462523,
      "loss": 2.2694,
      "step": 76881
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.8268826007843018,
      "learning_rate": 0.0004497129635810327,
      "loss": 2.39,
      "step": 76882
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.7601100206375122,
      "learning_rate": 0.0004497094187796058,
      "loss": 2.1886,
      "step": 76883
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.7592827081680298,
      "learning_rate": 0.000449705873950345,
      "loss": 2.1437,
      "step": 76884
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.839479386806488,
      "learning_rate": 0.00044970232909325103,
      "loss": 2.2963,
      "step": 76885
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.7711524963378906,
      "learning_rate": 0.00044969878420832464,
      "loss": 2.2017,
      "step": 76886
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.7617907524108887,
      "learning_rate": 0.00044969523929556645,
      "loss": 2.1381,
      "step": 76887
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.7593419551849365,
      "learning_rate": 0.00044969169435497705,
      "loss": 2.3086,
      "step": 76888
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.7788277864456177,
      "learning_rate": 0.0004496881493865571,
      "loss": 2.3287,
      "step": 76889
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.7509558200836182,
      "learning_rate": 0.00044968460439030733,
      "loss": 2.0912,
      "step": 76890
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.822527289390564,
      "learning_rate": 0.00044968105936622836,
      "loss": 2.2756,
      "step": 76891
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.7588285207748413,
      "learning_rate": 0.0004496775143143209,
      "loss": 2.3344,
      "step": 76892
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.945813000202179,
      "learning_rate": 0.0004496739692345855,
      "loss": 2.1841,
      "step": 76893
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.73636394739151,
      "learning_rate": 0.0004496704241270229,
      "loss": 1.9463,
      "step": 76894
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.7499922513961792,
      "learning_rate": 0.0004496668789916338,
      "loss": 2.1848,
      "step": 76895
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.7930821776390076,
      "learning_rate": 0.0004496633338284187,
      "loss": 2.208,
      "step": 76896
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.8746386766433716,
      "learning_rate": 0.00044965978863737837,
      "loss": 2.2349,
      "step": 76897
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.8551818132400513,
      "learning_rate": 0.0004496562434185136,
      "loss": 2.2049,
      "step": 76898
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.8123628497123718,
      "learning_rate": 0.00044965269817182475,
      "loss": 2.2061,
      "step": 76899
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.7725400924682617,
      "learning_rate": 0.0004496491528973127,
      "loss": 1.9122,
      "step": 76900
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.1732131242752075,
      "learning_rate": 0.0004496456075949781,
      "loss": 2.2767,
      "step": 76901
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.7191963195800781,
      "learning_rate": 0.00044964206226482143,
      "loss": 2.1599,
      "step": 76902
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.8914449214935303,
      "learning_rate": 0.0004496385169068436,
      "loss": 2.3987,
      "step": 76903
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.8530666828155518,
      "learning_rate": 0.00044963497152104506,
      "loss": 1.9182,
      "step": 76904
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.8221474885940552,
      "learning_rate": 0.0004496314261074266,
      "loss": 2.0103,
      "step": 76905
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.9144446849822998,
      "learning_rate": 0.00044962788066598883,
      "loss": 2.2596,
      "step": 76906
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.7607340216636658,
      "learning_rate": 0.0004496243351967324,
      "loss": 2.0806,
      "step": 76907
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.7566065788269043,
      "learning_rate": 0.000449620789699658,
      "loss": 2.1551,
      "step": 76908
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.9032139182090759,
      "learning_rate": 0.00044961724417476626,
      "loss": 2.1792,
      "step": 76909
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.7623376846313477,
      "learning_rate": 0.00044961369862205774,
      "loss": 2.1785,
      "step": 76910
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.8835412263870239,
      "learning_rate": 0.00044961015304153336,
      "loss": 2.3806,
      "step": 76911
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.8188939690589905,
      "learning_rate": 0.00044960660743319366,
      "loss": 2.132,
      "step": 76912
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.7600767612457275,
      "learning_rate": 0.00044960306179703914,
      "loss": 2.1766,
      "step": 76913
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.9501831531524658,
      "learning_rate": 0.0004495995161330707,
      "loss": 2.1999,
      "step": 76914
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.770530104637146,
      "learning_rate": 0.00044959597044128875,
      "loss": 2.4379,
      "step": 76915
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.8321277499198914,
      "learning_rate": 0.0004495924247216942,
      "loss": 2.137,
      "step": 76916
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.7194135785102844,
      "learning_rate": 0.00044958887897428757,
      "loss": 2.2572,
      "step": 76917
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.7825336456298828,
      "learning_rate": 0.00044958533319906953,
      "loss": 2.1896,
      "step": 76918
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.8037216663360596,
      "learning_rate": 0.00044958178739604073,
      "loss": 2.2914,
      "step": 76919
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.8871818780899048,
      "learning_rate": 0.0004495782415652019,
      "loss": 2.0856,
      "step": 76920
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.7952587008476257,
      "learning_rate": 0.0004495746957065536,
      "loss": 2.1823,
      "step": 76921
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.8120893836021423,
      "learning_rate": 0.00044957114982009655,
      "loss": 2.2305,
      "step": 76922
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.8271867036819458,
      "learning_rate": 0.0004495676039058315,
      "loss": 2.2344,
      "step": 76923
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.7384669780731201,
      "learning_rate": 0.0004495640579637589,
      "loss": 2.1518,
      "step": 76924
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.8371381759643555,
      "learning_rate": 0.00044956051199387953,
      "loss": 2.2266,
      "step": 76925
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.7763643264770508,
      "learning_rate": 0.0004495569659961941,
      "loss": 2.2203,
      "step": 76926
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.8324849605560303,
      "learning_rate": 0.0004495534199707031,
      "loss": 2.1515,
      "step": 76927
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.8759915232658386,
      "learning_rate": 0.0004495498739174073,
      "loss": 2.216,
      "step": 76928
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.7706494331359863,
      "learning_rate": 0.0004495463278363074,
      "loss": 1.9763,
      "step": 76929
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.8207958936691284,
      "learning_rate": 0.0004495427817274041,
      "loss": 2.3189,
      "step": 76930
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.8438799381256104,
      "learning_rate": 0.0004495392355906979,
      "loss": 2.3598,
      "step": 76931
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.8715928792953491,
      "learning_rate": 0.00044953568942618946,
      "loss": 2.1346,
      "step": 76932
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.7947679162025452,
      "learning_rate": 0.00044953214323387966,
      "loss": 1.9733,
      "step": 76933
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.8258807063102722,
      "learning_rate": 0.00044952859701376886,
      "loss": 2.2602,
      "step": 76934
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.8240293860435486,
      "learning_rate": 0.000449525050765858,
      "loss": 2.2024,
      "step": 76935
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.7803515791893005,
      "learning_rate": 0.00044952150449014755,
      "loss": 1.8913,
      "step": 76936
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.8483638167381287,
      "learning_rate": 0.00044951795818663824,
      "loss": 2.2908,
      "step": 76937
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.8566305637359619,
      "learning_rate": 0.00044951441185533074,
      "loss": 2.2439,
      "step": 76938
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.8090721368789673,
      "learning_rate": 0.00044951086549622566,
      "loss": 2.1464,
      "step": 76939
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.7577635645866394,
      "learning_rate": 0.00044950731910932363,
      "loss": 2.2498,
      "step": 76940
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.8443906307220459,
      "learning_rate": 0.00044950377269462543,
      "loss": 2.2158,
      "step": 76941
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.7555362582206726,
      "learning_rate": 0.0004495002262521317,
      "loss": 2.3578,
      "step": 76942
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.8513668179512024,
      "learning_rate": 0.000449496679781843,
      "loss": 2.0056,
      "step": 76943
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.7578856945037842,
      "learning_rate": 0.0004494931332837601,
      "loss": 2.156,
      "step": 76944
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.9358155131340027,
      "learning_rate": 0.00044948958675788356,
      "loss": 2.1982,
      "step": 76945
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.8366459012031555,
      "learning_rate": 0.000449486040204214,
      "loss": 2.0572,
      "step": 76946
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.7693918347358704,
      "learning_rate": 0.0004494824936227523,
      "loss": 2.2673,
      "step": 76947
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.0657058954238892,
      "learning_rate": 0.00044947894701349893,
      "loss": 2.0405,
      "step": 76948
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.817625880241394,
      "learning_rate": 0.00044947540037645455,
      "loss": 2.0187,
      "step": 76949
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.9592325091362,
      "learning_rate": 0.00044947185371162,
      "loss": 2.2902,
      "step": 76950
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.0684572458267212,
      "learning_rate": 0.00044946830701899574,
      "loss": 2.039,
      "step": 76951
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.874472439289093,
      "learning_rate": 0.00044946476029858245,
      "loss": 2.0762,
      "step": 76952
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.7936912775039673,
      "learning_rate": 0.00044946121355038095,
      "loss": 2.1227,
      "step": 76953
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.2014950513839722,
      "learning_rate": 0.0004494576667743917,
      "loss": 2.07,
      "step": 76954
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.8333819508552551,
      "learning_rate": 0.0004494541199706155,
      "loss": 2.227,
      "step": 76955
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.8401343822479248,
      "learning_rate": 0.000449450573139053,
      "loss": 1.9912,
      "step": 76956
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.9869391918182373,
      "learning_rate": 0.00044944702627970466,
      "loss": 2.265,
      "step": 76957
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.8062217235565186,
      "learning_rate": 0.00044944347939257143,
      "loss": 2.233,
      "step": 76958
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.8338137865066528,
      "learning_rate": 0.0004494399324776538,
      "loss": 2.2579,
      "step": 76959
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.7604697942733765,
      "learning_rate": 0.00044943638553495243,
      "loss": 2.1158,
      "step": 76960
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.8772037625312805,
      "learning_rate": 0.000449432838564468,
      "loss": 2.1504,
      "step": 76961
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.7668171525001526,
      "learning_rate": 0.00044942929156620125,
      "loss": 2.1106,
      "step": 76962
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.8402384519577026,
      "learning_rate": 0.00044942574454015277,
      "loss": 2.0294,
      "step": 76963
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.8415061831474304,
      "learning_rate": 0.0004494221974863232,
      "loss": 2.2705,
      "step": 76964
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.7012847065925598,
      "learning_rate": 0.0004494186504047133,
      "loss": 2.1369,
      "step": 76965
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.7497692704200745,
      "learning_rate": 0.00044941510329532357,
      "loss": 2.2976,
      "step": 76966
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.77872234582901,
      "learning_rate": 0.00044941155615815467,
      "loss": 2.0497,
      "step": 76967
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.9750679135322571,
      "learning_rate": 0.0004494080089932075,
      "loss": 2.3279,
      "step": 76968
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.7879987955093384,
      "learning_rate": 0.0004494044618004825,
      "loss": 2.2359,
      "step": 76969
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.8159500956535339,
      "learning_rate": 0.00044940091457998035,
      "loss": 2.1888,
      "step": 76970
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.8877102732658386,
      "learning_rate": 0.0004493973673317018,
      "loss": 2.2248,
      "step": 76971
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.7225762605667114,
      "learning_rate": 0.0004493938200556475,
      "loss": 2.2772,
      "step": 76972
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.7911847233772278,
      "learning_rate": 0.00044939027275181796,
      "loss": 2.3549,
      "step": 76973
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.7394332885742188,
      "learning_rate": 0.000449386725420214,
      "loss": 2.1508,
      "step": 76974
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.8094253540039062,
      "learning_rate": 0.00044938317806083623,
      "loss": 2.4027,
      "step": 76975
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.7441266775131226,
      "learning_rate": 0.0004493796306736853,
      "loss": 2.1755,
      "step": 76976
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.8405532240867615,
      "learning_rate": 0.0004493760832587619,
      "loss": 2.37,
      "step": 76977
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.816508412361145,
      "learning_rate": 0.00044937253581606667,
      "loss": 2.3049,
      "step": 76978
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.8379045724868774,
      "learning_rate": 0.00044936898834560025,
      "loss": 2.391,
      "step": 76979
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.8399463295936584,
      "learning_rate": 0.0004493654408473633,
      "loss": 2.3082,
      "step": 76980
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.803392767906189,
      "learning_rate": 0.0004493618933213565,
      "loss": 2.0715,
      "step": 76981
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.9789999723434448,
      "learning_rate": 0.00044935834576758055,
      "loss": 2.2507,
      "step": 76982
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.7797888517379761,
      "learning_rate": 0.000449354798186036,
      "loss": 2.248,
      "step": 76983
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.8230208158493042,
      "learning_rate": 0.0004493512505767236,
      "loss": 2.1019,
      "step": 76984
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.8460871577262878,
      "learning_rate": 0.000449347702939644,
      "loss": 2.1685,
      "step": 76985
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.7670221328735352,
      "learning_rate": 0.0004493441552747978,
      "loss": 2.1518,
      "step": 76986
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.7756989002227783,
      "learning_rate": 0.0004493406075821858,
      "loss": 2.25,
      "step": 76987
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.8844255208969116,
      "learning_rate": 0.00044933705986180845,
      "loss": 2.109,
      "step": 76988
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.7534140944480896,
      "learning_rate": 0.0004493335121136665,
      "loss": 1.9624,
      "step": 76989
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.7138897180557251,
      "learning_rate": 0.0004493299643377608,
      "loss": 2.12,
      "step": 76990
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.709586501121521,
      "learning_rate": 0.00044932641653409167,
      "loss": 2.1014,
      "step": 76991
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.7716357707977295,
      "learning_rate": 0.00044932286870266,
      "loss": 2.3191,
      "step": 76992
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.702785074710846,
      "learning_rate": 0.00044931932084346643,
      "loss": 2.0606,
      "step": 76993
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.784271776676178,
      "learning_rate": 0.00044931577295651156,
      "loss": 2.1791,
      "step": 76994
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.7257563471794128,
      "learning_rate": 0.00044931222504179596,
      "loss": 2.0608,
      "step": 76995
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.74834805727005,
      "learning_rate": 0.00044930867709932056,
      "loss": 2.2057,
      "step": 76996
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.8516300916671753,
      "learning_rate": 0.0004493051291290858,
      "loss": 2.1767,
      "step": 76997
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.7918274998664856,
      "learning_rate": 0.00044930158113109234,
      "loss": 2.3702,
      "step": 76998
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.7915287017822266,
      "learning_rate": 0.00044929803310534105,
      "loss": 2.0908,
      "step": 76999
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.7285925149917603,
      "learning_rate": 0.0004492944850518323,
      "loss": 2.2026,
      "step": 77000
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.7565723657608032,
      "learning_rate": 0.0004492909369705669,
      "loss": 2.2428,
      "step": 77001
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.7307555079460144,
      "learning_rate": 0.00044928738886154554,
      "loss": 2.09,
      "step": 77002
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.7954447269439697,
      "learning_rate": 0.0004492838407247688,
      "loss": 2.0852,
      "step": 77003
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.7886577248573303,
      "learning_rate": 0.0004492802925602374,
      "loss": 2.3925,
      "step": 77004
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.8210458755493164,
      "learning_rate": 0.00044927674436795196,
      "loss": 2.2909,
      "step": 77005
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.7544689774513245,
      "learning_rate": 0.00044927319614791316,
      "loss": 2.0747,
      "step": 77006
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.8661720156669617,
      "learning_rate": 0.00044926964790012165,
      "loss": 2.26,
      "step": 77007
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.8277830481529236,
      "learning_rate": 0.00044926609962457815,
      "loss": 2.2214,
      "step": 77008
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.7670496702194214,
      "learning_rate": 0.0004492625513212832,
      "loss": 2.086,
      "step": 77009
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.6918942928314209,
      "learning_rate": 0.0004492590029902375,
      "loss": 1.8795,
      "step": 77010
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.7249923348426819,
      "learning_rate": 0.00044925545463144183,
      "loss": 2.0209,
      "step": 77011
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.8192130327224731,
      "learning_rate": 0.00044925190624489663,
      "loss": 2.1825,
      "step": 77012
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.8049904108047485,
      "learning_rate": 0.0004492483578306028,
      "loss": 2.0254,
      "step": 77013
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.8509118556976318,
      "learning_rate": 0.00044924480938856085,
      "loss": 2.3445,
      "step": 77014
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.8134943842887878,
      "learning_rate": 0.0004492412609187714,
      "loss": 2.2804,
      "step": 77015
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.8255791068077087,
      "learning_rate": 0.00044923771242123524,
      "loss": 2.098,
      "step": 77016
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.8809378147125244,
      "learning_rate": 0.0004492341638959529,
      "loss": 2.3438,
      "step": 77017
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.7872121334075928,
      "learning_rate": 0.0004492306153429253,
      "loss": 2.0196,
      "step": 77018
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.7797731757164001,
      "learning_rate": 0.0004492270667621528,
      "loss": 2.1168,
      "step": 77019
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.851806104183197,
      "learning_rate": 0.0004492235181536361,
      "loss": 2.4482,
      "step": 77020
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.8122544288635254,
      "learning_rate": 0.00044921996951737605,
      "loss": 2.1629,
      "step": 77021
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.778057336807251,
      "learning_rate": 0.0004492164208533731,
      "loss": 2.283,
      "step": 77022
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.7453928589820862,
      "learning_rate": 0.00044921287216162807,
      "loss": 2.2429,
      "step": 77023
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.8601147532463074,
      "learning_rate": 0.00044920932344214145,
      "loss": 2.276,
      "step": 77024
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.7770687937736511,
      "learning_rate": 0.00044920577469491417,
      "loss": 2.155,
      "step": 77025
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.7969553470611572,
      "learning_rate": 0.00044920222591994654,
      "loss": 2.0593,
      "step": 77026
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.8128430247306824,
      "learning_rate": 0.0004491986771172395,
      "loss": 2.2582,
      "step": 77027
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.790648341178894,
      "learning_rate": 0.0004491951282867936,
      "loss": 2.2762,
      "step": 77028
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.8273960947990417,
      "learning_rate": 0.00044919157942860947,
      "loss": 2.2195,
      "step": 77029
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.8111796975135803,
      "learning_rate": 0.00044918803054268784,
      "loss": 2.2251,
      "step": 77030
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.8147391676902771,
      "learning_rate": 0.00044918448162902934,
      "loss": 1.9413,
      "step": 77031
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.8227025866508484,
      "learning_rate": 0.0004491809326876346,
      "loss": 2.1371,
      "step": 77032
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.8763030767440796,
      "learning_rate": 0.00044917738371850433,
      "loss": 1.8663,
      "step": 77033
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.7612796425819397,
      "learning_rate": 0.0004491738347216392,
      "loss": 2.188,
      "step": 77034
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.0219327211380005,
      "learning_rate": 0.0004491702856970398,
      "loss": 2.0218,
      "step": 77035
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.8273671269416809,
      "learning_rate": 0.00044916673664470685,
      "loss": 2.2309,
      "step": 77036
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.8416298031806946,
      "learning_rate": 0.00044916318756464094,
      "loss": 2.2549,
      "step": 77037
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.8584176898002625,
      "learning_rate": 0.0004491596384568428,
      "loss": 2.4666,
      "step": 77038
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.7639999389648438,
      "learning_rate": 0.0004491560893213132,
      "loss": 2.0865,
      "step": 77039
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.7535151243209839,
      "learning_rate": 0.0004491525401580525,
      "loss": 2.2912,
      "step": 77040
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.7773622274398804,
      "learning_rate": 0.00044914899096706157,
      "loss": 2.1354,
      "step": 77041
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.9248298406600952,
      "learning_rate": 0.00044914544174834106,
      "loss": 2.324,
      "step": 77042
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.8407742381095886,
      "learning_rate": 0.0004491418925018915,
      "loss": 1.931,
      "step": 77043
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.6967951655387878,
      "learning_rate": 0.00044913834322771374,
      "loss": 1.9763,
      "step": 77044
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.8484141826629639,
      "learning_rate": 0.00044913479392580836,
      "loss": 2.0799,
      "step": 77045
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.8342034816741943,
      "learning_rate": 0.00044913124459617596,
      "loss": 2.1285,
      "step": 77046
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.9042600393295288,
      "learning_rate": 0.00044912769523881727,
      "loss": 2.2525,
      "step": 77047
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.9818476438522339,
      "learning_rate": 0.0004491241458537329,
      "loss": 2.2065,
      "step": 77048
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.8588499426841736,
      "learning_rate": 0.00044912059644092357,
      "loss": 2.1643,
      "step": 77049
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.7984892129898071,
      "learning_rate": 0.00044911704700038987,
      "loss": 2.1335,
      "step": 77050
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.897479772567749,
      "learning_rate": 0.0004491134975321326,
      "loss": 2.3072,
      "step": 77051
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.7918517589569092,
      "learning_rate": 0.0004491099480361522,
      "loss": 2.3767,
      "step": 77052
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.7552681565284729,
      "learning_rate": 0.0004491063985124494,
      "loss": 2.4356,
      "step": 77053
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.7900367379188538,
      "learning_rate": 0.000449102848961025,
      "loss": 2.2154,
      "step": 77054
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.8720604777336121,
      "learning_rate": 0.0004490992993818796,
      "loss": 2.4908,
      "step": 77055
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.8061754703521729,
      "learning_rate": 0.0004490957497750137,
      "loss": 2.2647,
      "step": 77056
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.7947660684585571,
      "learning_rate": 0.0004490922001404282,
      "loss": 2.0339,
      "step": 77057
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.9264478087425232,
      "learning_rate": 0.0004490886504781236,
      "loss": 2.3508,
      "step": 77058
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.7636450529098511,
      "learning_rate": 0.00044908510078810063,
      "loss": 2.1284,
      "step": 77059
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.823086678981781,
      "learning_rate": 0.00044908155107035996,
      "loss": 2.2317,
      "step": 77060
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.7787341475486755,
      "learning_rate": 0.0004490780013249021,
      "loss": 2.3184,
      "step": 77061
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.7912230491638184,
      "learning_rate": 0.0004490744515517279,
      "loss": 2.2173,
      "step": 77062
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.7921919822692871,
      "learning_rate": 0.00044907090175083794,
      "loss": 2.1152,
      "step": 77063
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.7971464991569519,
      "learning_rate": 0.0004490673519222329,
      "loss": 2.178,
      "step": 77064
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.9284459352493286,
      "learning_rate": 0.0004490638020659134,
      "loss": 2.1762,
      "step": 77065
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.876298189163208,
      "learning_rate": 0.00044906025218188013,
      "loss": 2.2456,
      "step": 77066
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.831254243850708,
      "learning_rate": 0.0004490567022701337,
      "loss": 2.1736,
      "step": 77067
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.9616130590438843,
      "learning_rate": 0.0004490531523306748,
      "loss": 2.2768,
      "step": 77068
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.925679624080658,
      "learning_rate": 0.00044904960236350427,
      "loss": 2.2016,
      "step": 77069
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.7554011344909668,
      "learning_rate": 0.0004490460523686225,
      "loss": 2.4083,
      "step": 77070
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.8433900475502014,
      "learning_rate": 0.0004490425023460302,
      "loss": 2.1636,
      "step": 77071
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.9355069994926453,
      "learning_rate": 0.0004490389522957282,
      "loss": 2.0581,
      "step": 77072
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.8490619659423828,
      "learning_rate": 0.00044903540221771695,
      "loss": 2.3506,
      "step": 77073
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.7105820178985596,
      "learning_rate": 0.00044903185211199725,
      "loss": 2.2316,
      "step": 77074
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.7951627373695374,
      "learning_rate": 0.0004490283019785697,
      "loss": 2.0622,
      "step": 77075
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.8885449767112732,
      "learning_rate": 0.000449024751817435,
      "loss": 2.1083,
      "step": 77076
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.7211670279502869,
      "learning_rate": 0.0004490212016285938,
      "loss": 2.3059,
      "step": 77077
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.8177680969238281,
      "learning_rate": 0.00044901765141204666,
      "loss": 2.1975,
      "step": 77078
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.7854669690132141,
      "learning_rate": 0.0004490141011677944,
      "loss": 1.9639,
      "step": 77079
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.858144223690033,
      "learning_rate": 0.00044901055089583753,
      "loss": 2.1865,
      "step": 77080
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.8628998398780823,
      "learning_rate": 0.0004490070005961769,
      "loss": 2.2735,
      "step": 77081
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.7785743474960327,
      "learning_rate": 0.00044900345026881297,
      "loss": 2.0606,
      "step": 77082
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.7718808650970459,
      "learning_rate": 0.0004489998999137465,
      "loss": 2.1061,
      "step": 77083
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.8062455058097839,
      "learning_rate": 0.0004489963495309782,
      "loss": 2.0031,
      "step": 77084
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.8446924686431885,
      "learning_rate": 0.0004489927991205086,
      "loss": 2.2139,
      "step": 77085
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.8861081600189209,
      "learning_rate": 0.00044898924868233845,
      "loss": 2.135,
      "step": 77086
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.8178513646125793,
      "learning_rate": 0.00044898569821646836,
      "loss": 2.2005,
      "step": 77087
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.8921036720275879,
      "learning_rate": 0.0004489821477228991,
      "loss": 2.0595,
      "step": 77088
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.8741271495819092,
      "learning_rate": 0.00044897859720163113,
      "loss": 2.3718,
      "step": 77089
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.8595057725906372,
      "learning_rate": 0.0004489750466526653,
      "loss": 2.2733,
      "step": 77090
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.9074302911758423,
      "learning_rate": 0.00044897149607600216,
      "loss": 2.2504,
      "step": 77091
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.7730269432067871,
      "learning_rate": 0.0004489679454716424,
      "loss": 2.0533,
      "step": 77092
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.907750129699707,
      "learning_rate": 0.00044896439483958675,
      "loss": 2.1681,
      "step": 77093
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.8473204374313354,
      "learning_rate": 0.0004489608441798357,
      "loss": 2.4041,
      "step": 77094
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.861478865146637,
      "learning_rate": 0.0004489572934923901,
      "loss": 2.1376,
      "step": 77095
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.8312442302703857,
      "learning_rate": 0.00044895374277725055,
      "loss": 2.218,
      "step": 77096
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.8900772929191589,
      "learning_rate": 0.00044895019203441765,
      "loss": 2.1334,
      "step": 77097
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.7386486530303955,
      "learning_rate": 0.000448946641263892,
      "loss": 2.343,
      "step": 77098
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.7959995269775391,
      "learning_rate": 0.0004489430904656745,
      "loss": 1.9785,
      "step": 77099
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.7853266596794128,
      "learning_rate": 0.00044893953963976563,
      "loss": 2.28,
      "step": 77100
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.9074830412864685,
      "learning_rate": 0.00044893598878616603,
      "loss": 2.0684,
      "step": 77101
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.7464927434921265,
      "learning_rate": 0.0004489324379048765,
      "loss": 2.4815,
      "step": 77102
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.9054352045059204,
      "learning_rate": 0.0004489288869958976,
      "loss": 2.1182,
      "step": 77103
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.7529595494270325,
      "learning_rate": 0.00044892533605923,
      "loss": 2.2354,
      "step": 77104
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.7815511226654053,
      "learning_rate": 0.00044892178509487433,
      "loss": 2.1684,
      "step": 77105
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.7272029519081116,
      "learning_rate": 0.0004489182341028314,
      "loss": 2.2636,
      "step": 77106
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.7956930994987488,
      "learning_rate": 0.00044891468308310166,
      "loss": 1.9765,
      "step": 77107
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.7568750977516174,
      "learning_rate": 0.00044891113203568585,
      "loss": 2.132,
      "step": 77108
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.8294996619224548,
      "learning_rate": 0.0004489075809605847,
      "loss": 2.0859,
      "step": 77109
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.7598170042037964,
      "learning_rate": 0.0004489040298577988,
      "loss": 2.2283,
      "step": 77110
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.7965050935745239,
      "learning_rate": 0.00044890047872732884,
      "loss": 1.9507,
      "step": 77111
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.9500582218170166,
      "learning_rate": 0.00044889692756917545,
      "loss": 2.4033,
      "step": 77112
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.8287925124168396,
      "learning_rate": 0.00044889337638333933,
      "loss": 2.422,
      "step": 77113
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.9981009364128113,
      "learning_rate": 0.00044888982516982115,
      "loss": 2.1632,
      "step": 77114
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.8173401951789856,
      "learning_rate": 0.0004488862739286215,
      "loss": 2.0335,
      "step": 77115
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.9289069175720215,
      "learning_rate": 0.0004488827226597411,
      "loss": 2.255,
      "step": 77116
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.7702820301055908,
      "learning_rate": 0.0004488791713631806,
      "loss": 2.1896,
      "step": 77117
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.8121087551116943,
      "learning_rate": 0.00044887562003894066,
      "loss": 2.293,
      "step": 77118
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.767181932926178,
      "learning_rate": 0.00044887206868702184,
      "loss": 2.1437,
      "step": 77119
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.804356575012207,
      "learning_rate": 0.0004488685173074249,
      "loss": 2.2122,
      "step": 77120
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.7900869846343994,
      "learning_rate": 0.00044886496590015065,
      "loss": 2.2086,
      "step": 77121
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.8332599401473999,
      "learning_rate": 0.00044886141446519946,
      "loss": 2.2099,
      "step": 77122
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.7940362095832825,
      "learning_rate": 0.0004488578630025721,
      "loss": 2.2331,
      "step": 77123
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.7886948585510254,
      "learning_rate": 0.00044885431151226937,
      "loss": 2.3393,
      "step": 77124
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.7339354157447815,
      "learning_rate": 0.00044885075999429177,
      "loss": 2.1024,
      "step": 77125
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.8088457584381104,
      "learning_rate": 0.00044884720844863996,
      "loss": 2.116,
      "step": 77126
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.7366175651550293,
      "learning_rate": 0.0004488436568753147,
      "loss": 2.3947,
      "step": 77127
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.8600674867630005,
      "learning_rate": 0.00044884010527431655,
      "loss": 2.2809,
      "step": 77128
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.8463141918182373,
      "learning_rate": 0.00044883655364564624,
      "loss": 2.1905,
      "step": 77129
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.7611374258995056,
      "learning_rate": 0.00044883300198930444,
      "loss": 2.0196,
      "step": 77130
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.7904495596885681,
      "learning_rate": 0.00044882945030529173,
      "loss": 2.4614,
      "step": 77131
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.8156646490097046,
      "learning_rate": 0.0004488258985936088,
      "loss": 2.2872,
      "step": 77132
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.7431836128234863,
      "learning_rate": 0.00044882234685425634,
      "loss": 2.0244,
      "step": 77133
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.7747078537940979,
      "learning_rate": 0.000448818795087235,
      "loss": 2.3175,
      "step": 77134
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.8872860074043274,
      "learning_rate": 0.00044881524329254543,
      "loss": 2.1856,
      "step": 77135
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.7754629850387573,
      "learning_rate": 0.0004488116914701883,
      "loss": 2.0073,
      "step": 77136
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.8042237758636475,
      "learning_rate": 0.0004488081396201643,
      "loss": 2.1961,
      "step": 77137
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.0503560304641724,
      "learning_rate": 0.000448804587742474,
      "loss": 2.0285,
      "step": 77138
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.7505396008491516,
      "learning_rate": 0.00044880103583711824,
      "loss": 1.9721,
      "step": 77139
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.8508594632148743,
      "learning_rate": 0.0004487974839040974,
      "loss": 2.1518,
      "step": 77140
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.8486478328704834,
      "learning_rate": 0.00044879393194341237,
      "loss": 2.2155,
      "step": 77141
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.8374521732330322,
      "learning_rate": 0.0004487903799550638,
      "loss": 2.1846,
      "step": 77142
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.8548639416694641,
      "learning_rate": 0.0004487868279390522,
      "loss": 2.3126,
      "step": 77143
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.7981901168823242,
      "learning_rate": 0.00044878327589537834,
      "loss": 2.0111,
      "step": 77144
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.7856885194778442,
      "learning_rate": 0.0004487797238240429,
      "loss": 2.1352,
      "step": 77145
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.7852011919021606,
      "learning_rate": 0.00044877617172504647,
      "loss": 2.0578,
      "step": 77146
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.8326340913772583,
      "learning_rate": 0.0004487726195983897,
      "loss": 2.2075,
      "step": 77147
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.7786270976066589,
      "learning_rate": 0.00044876906744407334,
      "loss": 2.1258,
      "step": 77148
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.7357915043830872,
      "learning_rate": 0.00044876551526209797,
      "loss": 2.1458,
      "step": 77149
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.7879032492637634,
      "learning_rate": 0.0004487619630524643,
      "loss": 2.3204,
      "step": 77150
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.7793459892272949,
      "learning_rate": 0.000448758410815173,
      "loss": 2.4017,
      "step": 77151
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.8208812475204468,
      "learning_rate": 0.0004487548585502247,
      "loss": 2.2553,
      "step": 77152
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.8127601146697998,
      "learning_rate": 0.00044875130625762,
      "loss": 2.1234,
      "step": 77153
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.7815362811088562,
      "learning_rate": 0.00044874775393735975,
      "loss": 2.2945,
      "step": 77154
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.8285761475563049,
      "learning_rate": 0.0004487442015894444,
      "loss": 1.9722,
      "step": 77155
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.7661488652229309,
      "learning_rate": 0.00044874064921387465,
      "loss": 2.2578,
      "step": 77156
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.8651711940765381,
      "learning_rate": 0.0004487370968106513,
      "loss": 2.2129,
      "step": 77157
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.0136547088623047,
      "learning_rate": 0.00044873354437977486,
      "loss": 2.1897,
      "step": 77158
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.872111439704895,
      "learning_rate": 0.000448729991921246,
      "loss": 2.2009,
      "step": 77159
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.8839287161827087,
      "learning_rate": 0.0004487264394350655,
      "loss": 1.993,
      "step": 77160
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.8653317093849182,
      "learning_rate": 0.00044872288692123397,
      "loss": 2.2248,
      "step": 77161
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.7782688736915588,
      "learning_rate": 0.000448719334379752,
      "loss": 2.1259,
      "step": 77162
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.8511021137237549,
      "learning_rate": 0.0004487157818106204,
      "loss": 2.281,
      "step": 77163
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.7721962928771973,
      "learning_rate": 0.00044871222921383957,
      "loss": 1.9738,
      "step": 77164
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.9074192643165588,
      "learning_rate": 0.0004487086765894104,
      "loss": 2.4801,
      "step": 77165
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.7727360725402832,
      "learning_rate": 0.0004487051239373335,
      "loss": 2.0944,
      "step": 77166
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.7978218793869019,
      "learning_rate": 0.00044870157125760944,
      "loss": 2.3341,
      "step": 77167
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.8949087858200073,
      "learning_rate": 0.000448698018550239,
      "loss": 2.3837,
      "step": 77168
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.811068594455719,
      "learning_rate": 0.0004486944658152228,
      "loss": 2.192,
      "step": 77169
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.7980236411094666,
      "learning_rate": 0.0004486909130525615,
      "loss": 2.2213,
      "step": 77170
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.8520844578742981,
      "learning_rate": 0.00044868736026225576,
      "loss": 2.0683,
      "step": 77171
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.0576438903808594,
      "learning_rate": 0.0004486838074443062,
      "loss": 2.2748,
      "step": 77172
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.7916925549507141,
      "learning_rate": 0.00044868025459871355,
      "loss": 2.2692,
      "step": 77173
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.8279685378074646,
      "learning_rate": 0.00044867670172547836,
      "loss": 2.168,
      "step": 77174
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.9128187894821167,
      "learning_rate": 0.00044867314882460137,
      "loss": 2.2023,
      "step": 77175
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.8323043584823608,
      "learning_rate": 0.0004486695958960833,
      "loss": 2.2416,
      "step": 77176
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.7519555687904358,
      "learning_rate": 0.00044866604293992474,
      "loss": 2.2081,
      "step": 77177
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.842738687992096,
      "learning_rate": 0.00044866248995612635,
      "loss": 2.1164,
      "step": 77178
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.8176180124282837,
      "learning_rate": 0.00044865893694468877,
      "loss": 2.1243,
      "step": 77179
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.7425330877304077,
      "learning_rate": 0.0004486553839056127,
      "loss": 2.1359,
      "step": 77180
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.9040693044662476,
      "learning_rate": 0.0004486518308388988,
      "loss": 2.1364,
      "step": 77181
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.8686391115188599,
      "learning_rate": 0.00044864827774454774,
      "loss": 2.1415,
      "step": 77182
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.7727627158164978,
      "learning_rate": 0.0004486447246225601,
      "loss": 2.1158,
      "step": 77183
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.7775937914848328,
      "learning_rate": 0.00044864117147293666,
      "loss": 2.202,
      "step": 77184
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.8476127982139587,
      "learning_rate": 0.000448637618295678,
      "loss": 2.2835,
      "step": 77185
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.7976827621459961,
      "learning_rate": 0.00044863406509078476,
      "loss": 2.1627,
      "step": 77186
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.7972273230552673,
      "learning_rate": 0.0004486305118582577,
      "loss": 2.3298,
      "step": 77187
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.8024572730064392,
      "learning_rate": 0.00044862695859809737,
      "loss": 2.1026,
      "step": 77188
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.8546749353408813,
      "learning_rate": 0.0004486234053103045,
      "loss": 2.2965,
      "step": 77189
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.7881336808204651,
      "learning_rate": 0.00044861985199487974,
      "loss": 1.8789,
      "step": 77190
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.859346330165863,
      "learning_rate": 0.00044861629865182375,
      "loss": 2.0643,
      "step": 77191
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.7884117960929871,
      "learning_rate": 0.0004486127452811372,
      "loss": 2.301,
      "step": 77192
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.8333044648170471,
      "learning_rate": 0.00044860919188282065,
      "loss": 2.1771,
      "step": 77193
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.8222270607948303,
      "learning_rate": 0.0004486056384568749,
      "loss": 2.0871,
      "step": 77194
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.8039995431900024,
      "learning_rate": 0.00044860208500330066,
      "loss": 2.0161,
      "step": 77195
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.743971049785614,
      "learning_rate": 0.0004485985315220984,
      "loss": 2.1564,
      "step": 77196
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.8589136600494385,
      "learning_rate": 0.00044859497801326875,
      "loss": 2.2934,
      "step": 77197
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.769163191318512,
      "learning_rate": 0.00044859142447681267,
      "loss": 2.055,
      "step": 77198
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.813858151435852,
      "learning_rate": 0.00044858787091273057,
      "loss": 2.2887,
      "step": 77199
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.769978940486908,
      "learning_rate": 0.0004485843173210231,
      "loss": 2.0194,
      "step": 77200
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.7906296253204346,
      "learning_rate": 0.00044858076370169115,
      "loss": 2.2825,
      "step": 77201
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.7873865365982056,
      "learning_rate": 0.0004485772100547352,
      "loss": 2.0767,
      "step": 77202
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.7538589835166931,
      "learning_rate": 0.0004485736563801558,
      "loss": 2.1593,
      "step": 77203
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.7596120238304138,
      "learning_rate": 0.00044857010267795394,
      "loss": 2.1143,
      "step": 77204
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.7936294674873352,
      "learning_rate": 0.00044856654894813,
      "loss": 2.3593,
      "step": 77205
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.7510046362876892,
      "learning_rate": 0.00044856299519068475,
      "loss": 2.2055,
      "step": 77206
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.8153666257858276,
      "learning_rate": 0.0004485594414056188,
      "loss": 2.1673,
      "step": 77207
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.8312719464302063,
      "learning_rate": 0.00044855588759293286,
      "loss": 2.1462,
      "step": 77208
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.8561986088752747,
      "learning_rate": 0.0004485523337526276,
      "loss": 1.9404,
      "step": 77209
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.7395948767662048,
      "learning_rate": 0.00044854877988470364,
      "loss": 2.1943,
      "step": 77210
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.8430536985397339,
      "learning_rate": 0.0004485452259891617,
      "loss": 2.1687,
      "step": 77211
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.9316176772117615,
      "learning_rate": 0.00044854167206600235,
      "loss": 2.2421,
      "step": 77212
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.7729836106300354,
      "learning_rate": 0.0004485381181152263,
      "loss": 2.0846,
      "step": 77213
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.8910800814628601,
      "learning_rate": 0.0004485345641368342,
      "loss": 2.2334,
      "step": 77214
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.8373487591743469,
      "learning_rate": 0.0004485310101308268,
      "loss": 2.0853,
      "step": 77215
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.7328425049781799,
      "learning_rate": 0.0004485274560972046,
      "loss": 2.1271,
      "step": 77216
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.7516352534294128,
      "learning_rate": 0.0004485239020359683,
      "loss": 2.2236,
      "step": 77217
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.7602120637893677,
      "learning_rate": 0.00044852034794711876,
      "loss": 2.3469,
      "step": 77218
    },
    {
      "epoch": 1.01,
      "grad_norm": 1.0860941410064697,
      "learning_rate": 0.00044851679383065636,
      "loss": 1.992,
      "step": 77219
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.7306085228919983,
      "learning_rate": 0.00044851323968658194,
      "loss": 1.942,
      "step": 77220
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.8353833556175232,
      "learning_rate": 0.0004485096855148961,
      "loss": 2.3583,
      "step": 77221
    },
    {
      "epoch": 1.01,
      "grad_norm": 1.025368332862854,
      "learning_rate": 0.0004485061313155995,
      "loss": 1.8785,
      "step": 77222
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.7547022104263306,
      "learning_rate": 0.0004485025770886928,
      "loss": 1.9853,
      "step": 77223
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.9556995630264282,
      "learning_rate": 0.00044849902283417665,
      "loss": 2.154,
      "step": 77224
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.8960601687431335,
      "learning_rate": 0.00044849546855205174,
      "loss": 2.1571,
      "step": 77225
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.7371132969856262,
      "learning_rate": 0.0004484919142423187,
      "loss": 2.2319,
      "step": 77226
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.8232641816139221,
      "learning_rate": 0.00044848835990497825,
      "loss": 2.001,
      "step": 77227
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.8069757223129272,
      "learning_rate": 0.00044848480554003097,
      "loss": 2.141,
      "step": 77228
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.7544840574264526,
      "learning_rate": 0.00044848125114747763,
      "loss": 2.1142,
      "step": 77229
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.7659860849380493,
      "learning_rate": 0.0004484776967273188,
      "loss": 2.2271,
      "step": 77230
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.8783968091011047,
      "learning_rate": 0.0004484741422795551,
      "loss": 2.0526,
      "step": 77231
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.9321781992912292,
      "learning_rate": 0.00044847058780418724,
      "loss": 2.3662,
      "step": 77232
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.7861936688423157,
      "learning_rate": 0.00044846703330121603,
      "loss": 2.1555,
      "step": 77233
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.8365548253059387,
      "learning_rate": 0.00044846347877064197,
      "loss": 2.0206,
      "step": 77234
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.888236403465271,
      "learning_rate": 0.00044845992421246563,
      "loss": 2.3127,
      "step": 77235
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.8186671733856201,
      "learning_rate": 0.00044845636962668784,
      "loss": 1.9774,
      "step": 77236
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.9499418139457703,
      "learning_rate": 0.00044845281501330924,
      "loss": 2.098,
      "step": 77237
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.842738151550293,
      "learning_rate": 0.00044844926037233043,
      "loss": 2.3331,
      "step": 77238
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.8386041522026062,
      "learning_rate": 0.00044844570570375206,
      "loss": 2.3722,
      "step": 77239
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.7883127331733704,
      "learning_rate": 0.000448442151007575,
      "loss": 2.2606,
      "step": 77240
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.7655860185623169,
      "learning_rate": 0.0004484385962837995,
      "loss": 1.9836,
      "step": 77241
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.8413421511650085,
      "learning_rate": 0.00044843504153242665,
      "loss": 2.2519,
      "step": 77242
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.8837109208106995,
      "learning_rate": 0.0004484314867534569,
      "loss": 2.1553,
      "step": 77243
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.8724426627159119,
      "learning_rate": 0.00044842793194689086,
      "loss": 2.0319,
      "step": 77244
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.9604678750038147,
      "learning_rate": 0.0004484243771127293,
      "loss": 2.2708,
      "step": 77245
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.8530508279800415,
      "learning_rate": 0.00044842082225097287,
      "loss": 2.1403,
      "step": 77246
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.8108639121055603,
      "learning_rate": 0.00044841726736162215,
      "loss": 2.2728,
      "step": 77247
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.7794802784919739,
      "learning_rate": 0.0004484137124446779,
      "loss": 2.4766,
      "step": 77248
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.8411117196083069,
      "learning_rate": 0.00044841015750014075,
      "loss": 2.2169,
      "step": 77249
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.7631365656852722,
      "learning_rate": 0.0004484066025280113,
      "loss": 2.123,
      "step": 77250
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.7250197529792786,
      "learning_rate": 0.0004484030475282903,
      "loss": 2.1521,
      "step": 77251
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.8452064990997314,
      "learning_rate": 0.0004483994925009784,
      "loss": 2.2121,
      "step": 77252
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.8058192729949951,
      "learning_rate": 0.0004483959374460762,
      "loss": 2.1505,
      "step": 77253
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.8546979427337646,
      "learning_rate": 0.00044839238236358436,
      "loss": 2.291,
      "step": 77254
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.7494680881500244,
      "learning_rate": 0.00044838882725350363,
      "loss": 2.1118,
      "step": 77255
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.8848692774772644,
      "learning_rate": 0.00044838527211583457,
      "loss": 2.1002,
      "step": 77256
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.8905059099197388,
      "learning_rate": 0.00044838171695057793,
      "loss": 2.0906,
      "step": 77257
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.922257661819458,
      "learning_rate": 0.00044837816175773436,
      "loss": 2.2644,
      "step": 77258
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.8519724011421204,
      "learning_rate": 0.0004483746065373044,
      "loss": 2.3121,
      "step": 77259
    },
    {
      "epoch": 1.01,
      "grad_norm": 1.0053975582122803,
      "learning_rate": 0.0004483710512892888,
      "loss": 2.3245,
      "step": 77260
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.7682397961616516,
      "learning_rate": 0.0004483674960136883,
      "loss": 2.1506,
      "step": 77261
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.9374012351036072,
      "learning_rate": 0.00044836394071050343,
      "loss": 2.3486,
      "step": 77262
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.978915274143219,
      "learning_rate": 0.00044836038537973493,
      "loss": 2.0931,
      "step": 77263
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.7285599112510681,
      "learning_rate": 0.0004483568300213834,
      "loss": 2.286,
      "step": 77264
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.8125535249710083,
      "learning_rate": 0.00044835327463544955,
      "loss": 2.1512,
      "step": 77265
    },
    {
      "epoch": 1.01,
      "grad_norm": 1.0673184394836426,
      "learning_rate": 0.00044834971922193403,
      "loss": 2.0989,
      "step": 77266
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.7584474682807922,
      "learning_rate": 0.00044834616378083755,
      "loss": 2.001,
      "step": 77267
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.835181713104248,
      "learning_rate": 0.0004483426083121606,
      "loss": 2.3972,
      "step": 77268
    },
    {
      "epoch": 1.01,
      "grad_norm": 1.0257974863052368,
      "learning_rate": 0.00044833905281590407,
      "loss": 2.2863,
      "step": 77269
    },
    {
      "epoch": 1.01,
      "grad_norm": 1.0933581590652466,
      "learning_rate": 0.00044833549729206843,
      "loss": 2.0795,
      "step": 77270
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.7720274925231934,
      "learning_rate": 0.0004483319417406545,
      "loss": 2.178,
      "step": 77271
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.8997048735618591,
      "learning_rate": 0.00044832838616166277,
      "loss": 2.2629,
      "step": 77272
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.7685297727584839,
      "learning_rate": 0.00044832483055509405,
      "loss": 2.3256,
      "step": 77273
    },
    {
      "epoch": 1.01,
      "grad_norm": 1.0711947679519653,
      "learning_rate": 0.00044832127492094897,
      "loss": 2.3547,
      "step": 77274
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.766643226146698,
      "learning_rate": 0.000448317719259228,
      "loss": 2.0904,
      "step": 77275
    },
    {
      "epoch": 1.01,
      "grad_norm": 1.1133949756622314,
      "learning_rate": 0.00044831416356993223,
      "loss": 2.3683,
      "step": 77276
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.7395982146263123,
      "learning_rate": 0.0004483106078530619,
      "loss": 2.0083,
      "step": 77277
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.7897670865058899,
      "learning_rate": 0.0004483070521086178,
      "loss": 2.3531,
      "step": 77278
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.7794855237007141,
      "learning_rate": 0.0004483034963366007,
      "loss": 2.371,
      "step": 77279
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.7318292856216431,
      "learning_rate": 0.00044829994053701116,
      "loss": 2.1069,
      "step": 77280
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.8443881273269653,
      "learning_rate": 0.00044829638470984985,
      "loss": 2.3573,
      "step": 77281
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.7696110010147095,
      "learning_rate": 0.0004482928288551174,
      "loss": 2.2669,
      "step": 77282
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.8206218481063843,
      "learning_rate": 0.00044828927297281454,
      "loss": 2.1287,
      "step": 77283
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.8307989239692688,
      "learning_rate": 0.00044828571706294194,
      "loss": 2.5196,
      "step": 77284
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.7950413227081299,
      "learning_rate": 0.0004482821611255002,
      "loss": 2.0193,
      "step": 77285
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.8063830733299255,
      "learning_rate": 0.00044827860516049006,
      "loss": 2.1021,
      "step": 77286
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.8314949870109558,
      "learning_rate": 0.0004482750491679121,
      "loss": 2.2045,
      "step": 77287
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.733130693435669,
      "learning_rate": 0.00044827149314776696,
      "loss": 2.0852,
      "step": 77288
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.7011135220527649,
      "learning_rate": 0.00044826793710005535,
      "loss": 2.2995,
      "step": 77289
    },
    {
      "epoch": 1.01,
      "grad_norm": 1.016750693321228,
      "learning_rate": 0.00044826438102477806,
      "loss": 2.0277,
      "step": 77290
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.8681641817092896,
      "learning_rate": 0.00044826082492193547,
      "loss": 1.9711,
      "step": 77291
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.9287093281745911,
      "learning_rate": 0.00044825726879152845,
      "loss": 2.0706,
      "step": 77292
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.9316509962081909,
      "learning_rate": 0.00044825371263355754,
      "loss": 2.0491,
      "step": 77293
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.8081072568893433,
      "learning_rate": 0.00044825015644802356,
      "loss": 2.1535,
      "step": 77294
    },
    {
      "epoch": 1.01,
      "grad_norm": 1.0202304124832153,
      "learning_rate": 0.0004482466002349271,
      "loss": 2.1958,
      "step": 77295
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.7672572135925293,
      "learning_rate": 0.00044824304399426865,
      "loss": 2.1396,
      "step": 77296
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.8609641194343567,
      "learning_rate": 0.00044823948772604923,
      "loss": 2.1668,
      "step": 77297
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.7774255871772766,
      "learning_rate": 0.0004482359314302692,
      "loss": 2.113,
      "step": 77298
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.9609140157699585,
      "learning_rate": 0.0004482323751069292,
      "loss": 2.1504,
      "step": 77299
    },
    {
      "epoch": 1.01,
      "grad_norm": 1.0412533283233643,
      "learning_rate": 0.0004482288187560301,
      "loss": 2.2759,
      "step": 77300
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.7754999399185181,
      "learning_rate": 0.00044822526237757253,
      "loss": 2.2781,
      "step": 77301
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.939744770526886,
      "learning_rate": 0.000448221705971557,
      "loss": 1.941,
      "step": 77302
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.8243250250816345,
      "learning_rate": 0.0004482181495379842,
      "loss": 2.1678,
      "step": 77303
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.7940868735313416,
      "learning_rate": 0.000448214593076855,
      "loss": 2.144,
      "step": 77304
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.7975550889968872,
      "learning_rate": 0.0004482110365881698,
      "loss": 2.0982,
      "step": 77305
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.7721366286277771,
      "learning_rate": 0.00044820748007192937,
      "loss": 2.1589,
      "step": 77306
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.7759959101676941,
      "learning_rate": 0.0004482039235281345,
      "loss": 2.3535,
      "step": 77307
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.8588951826095581,
      "learning_rate": 0.0004482003669567856,
      "loss": 2.1577,
      "step": 77308
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.7784602642059326,
      "learning_rate": 0.0004481968103578835,
      "loss": 1.9492,
      "step": 77309
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.9304197430610657,
      "learning_rate": 0.00044819325373142877,
      "loss": 2.17,
      "step": 77310
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.7742835879325867,
      "learning_rate": 0.00044818969707742214,
      "loss": 2.1969,
      "step": 77311
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.7747865915298462,
      "learning_rate": 0.0004481861403958642,
      "loss": 2.1463,
      "step": 77312
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.7324763536453247,
      "learning_rate": 0.00044818258368675576,
      "loss": 2.2767,
      "step": 77313
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.7375144362449646,
      "learning_rate": 0.0004481790269500973,
      "loss": 2.1761,
      "step": 77314
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.9086648225784302,
      "learning_rate": 0.0004481754701858896,
      "loss": 2.1648,
      "step": 77315
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.7913933992385864,
      "learning_rate": 0.0004481719133941333,
      "loss": 2.084,
      "step": 77316
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.8689400553703308,
      "learning_rate": 0.000448168356574829,
      "loss": 2.1015,
      "step": 77317
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.9167827367782593,
      "learning_rate": 0.00044816479972797744,
      "loss": 2.1862,
      "step": 77318
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.7907032370567322,
      "learning_rate": 0.0004481612428535793,
      "loss": 2.1317,
      "step": 77319
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.8307474255561829,
      "learning_rate": 0.00044815768595163504,
      "loss": 2.0357,
      "step": 77320
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.7934426069259644,
      "learning_rate": 0.00044815412902214554,
      "loss": 2.2185,
      "step": 77321
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.8163880705833435,
      "learning_rate": 0.00044815057206511145,
      "loss": 2.0646,
      "step": 77322
    },
    {
      "epoch": 1.01,
      "grad_norm": 1.2839481830596924,
      "learning_rate": 0.0004481470150805333,
      "loss": 2.1124,
      "step": 77323
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.846560537815094,
      "learning_rate": 0.00044814345806841184,
      "loss": 2.292,
      "step": 77324
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.7955273389816284,
      "learning_rate": 0.0004481399010287478,
      "loss": 2.4018,
      "step": 77325
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.7794076800346375,
      "learning_rate": 0.00044813634396154167,
      "loss": 2.0585,
      "step": 77326
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.7541332244873047,
      "learning_rate": 0.0004481327868667941,
      "loss": 2.4161,
      "step": 77327
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.8008896112442017,
      "learning_rate": 0.00044812922974450604,
      "loss": 2.4446,
      "step": 77328
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.7836626172065735,
      "learning_rate": 0.0004481256725946778,
      "loss": 2.2168,
      "step": 77329
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.8080199956893921,
      "learning_rate": 0.0004481221154173103,
      "loss": 2.2229,
      "step": 77330
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.8306698799133301,
      "learning_rate": 0.0004481185582124041,
      "loss": 2.1934,
      "step": 77331
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.8325752019882202,
      "learning_rate": 0.00044811500097995983,
      "loss": 2.2186,
      "step": 77332
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.7875065207481384,
      "learning_rate": 0.0004481114437199782,
      "loss": 2.0735,
      "step": 77333
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.7919569611549377,
      "learning_rate": 0.00044810788643245993,
      "loss": 2.2992,
      "step": 77334
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.8578854203224182,
      "learning_rate": 0.0004481043291174055,
      "loss": 2.064,
      "step": 77335
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.7847127318382263,
      "learning_rate": 0.0004481007717748157,
      "loss": 2.2532,
      "step": 77336
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.920692503452301,
      "learning_rate": 0.0004480972144046912,
      "loss": 2.0342,
      "step": 77337
    },
    {
      "epoch": 1.01,
      "grad_norm": 1.0321345329284668,
      "learning_rate": 0.0004480936570070326,
      "loss": 2.2961,
      "step": 77338
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.8509325385093689,
      "learning_rate": 0.0004480900995818406,
      "loss": 2.0925,
      "step": 77339
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.8225324749946594,
      "learning_rate": 0.0004480865421291159,
      "loss": 2.0702,
      "step": 77340
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.8271706104278564,
      "learning_rate": 0.000448082984648859,
      "loss": 1.9749,
      "step": 77341
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.7936313152313232,
      "learning_rate": 0.0004480794271410708,
      "loss": 2.2246,
      "step": 77342
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.8318241834640503,
      "learning_rate": 0.0004480758696057518,
      "loss": 2.1908,
      "step": 77343
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.8142762780189514,
      "learning_rate": 0.0004480723120429027,
      "loss": 2.3255,
      "step": 77344
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.8798806667327881,
      "learning_rate": 0.00044806875445252414,
      "loss": 1.8805,
      "step": 77345
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.7744603157043457,
      "learning_rate": 0.0004480651968346169,
      "loss": 2.1531,
      "step": 77346
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.7332764267921448,
      "learning_rate": 0.00044806163918918143,
      "loss": 2.034,
      "step": 77347
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.88706374168396,
      "learning_rate": 0.0004480580815162185,
      "loss": 2.0939,
      "step": 77348
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.8525891304016113,
      "learning_rate": 0.0004480545238157289,
      "loss": 2.126,
      "step": 77349
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.8249639868736267,
      "learning_rate": 0.00044805096608771305,
      "loss": 2.0852,
      "step": 77350
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.8376197218894958,
      "learning_rate": 0.00044804740833217176,
      "loss": 2.1764,
      "step": 77351
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.8128706216812134,
      "learning_rate": 0.0004480438505491057,
      "loss": 2.2087,
      "step": 77352
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.7377820014953613,
      "learning_rate": 0.0004480402927385155,
      "loss": 2.1201,
      "step": 77353
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.7411031723022461,
      "learning_rate": 0.00044803673490040173,
      "loss": 2.0574,
      "step": 77354
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.9407379031181335,
      "learning_rate": 0.00044803317703476524,
      "loss": 2.2919,
      "step": 77355
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.7387121915817261,
      "learning_rate": 0.0004480296191416065,
      "loss": 2.3027,
      "step": 77356
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.7250602841377258,
      "learning_rate": 0.0004480260612209263,
      "loss": 2.1686,
      "step": 77357
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.9148449897766113,
      "learning_rate": 0.00044802250327272527,
      "loss": 2.4973,
      "step": 77358
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.895995557308197,
      "learning_rate": 0.00044801894529700404,
      "loss": 2.2649,
      "step": 77359
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.7549346089363098,
      "learning_rate": 0.0004480153872937633,
      "loss": 2.3966,
      "step": 77360
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.8107925653457642,
      "learning_rate": 0.0004480118292630037,
      "loss": 2.1671,
      "step": 77361
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.7983518242835999,
      "learning_rate": 0.00044800827120472595,
      "loss": 2.3145,
      "step": 77362
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.7843223214149475,
      "learning_rate": 0.00044800471311893066,
      "loss": 2.0816,
      "step": 77363
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.7124701738357544,
      "learning_rate": 0.0004480011550056185,
      "loss": 2.045,
      "step": 77364
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.8458534479141235,
      "learning_rate": 0.00044799759686479007,
      "loss": 2.1003,
      "step": 77365
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.7720914483070374,
      "learning_rate": 0.0004479940386964461,
      "loss": 2.2658,
      "step": 77366
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.8927174210548401,
      "learning_rate": 0.0004479904805005873,
      "loss": 2.0743,
      "step": 77367
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.8244830965995789,
      "learning_rate": 0.0004479869222772143,
      "loss": 2.3399,
      "step": 77368
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.7324841022491455,
      "learning_rate": 0.00044798336402632763,
      "loss": 2.1208,
      "step": 77369
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.7484297156333923,
      "learning_rate": 0.0004479798057479281,
      "loss": 2.0739,
      "step": 77370
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.8044667840003967,
      "learning_rate": 0.00044797624744201636,
      "loss": 2.0669,
      "step": 77371
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.8462182879447937,
      "learning_rate": 0.00044797268910859304,
      "loss": 2.0981,
      "step": 77372
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.7512006163597107,
      "learning_rate": 0.0004479691307476588,
      "loss": 2.0652,
      "step": 77373
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.7825050354003906,
      "learning_rate": 0.00044796557235921426,
      "loss": 2.0215,
      "step": 77374
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.8483559489250183,
      "learning_rate": 0.0004479620139432602,
      "loss": 2.3054,
      "step": 77375
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.7879372239112854,
      "learning_rate": 0.0004479584554997972,
      "loss": 2.0693,
      "step": 77376
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.7815078496932983,
      "learning_rate": 0.0004479548970288259,
      "loss": 2.0293,
      "step": 77377
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.8119037747383118,
      "learning_rate": 0.000447951338530347,
      "loss": 2.1156,
      "step": 77378
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.9947455525398254,
      "learning_rate": 0.00044794778000436115,
      "loss": 2.0441,
      "step": 77379
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.9276612401008606,
      "learning_rate": 0.000447944221450869,
      "loss": 2.2651,
      "step": 77380
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.716102659702301,
      "learning_rate": 0.00044794066286987116,
      "loss": 2.2013,
      "step": 77381
    },
    {
      "epoch": 1.01,
      "grad_norm": 1.1312077045440674,
      "learning_rate": 0.0004479371042613685,
      "loss": 2.2834,
      "step": 77382
    },
    {
      "epoch": 1.01,
      "grad_norm": 1.0220062732696533,
      "learning_rate": 0.0004479335456253615,
      "loss": 2.2137,
      "step": 77383
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.7767766118049622,
      "learning_rate": 0.0004479299869618508,
      "loss": 2.3011,
      "step": 77384
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.9464924931526184,
      "learning_rate": 0.00044792642827083716,
      "loss": 2.3035,
      "step": 77385
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.9021933674812317,
      "learning_rate": 0.00044792286955232127,
      "loss": 2.0132,
      "step": 77386
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.7787662148475647,
      "learning_rate": 0.0004479193108063037,
      "loss": 1.9921,
      "step": 77387
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.8232616186141968,
      "learning_rate": 0.000447915752032785,
      "loss": 1.8858,
      "step": 77388
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.992779552936554,
      "learning_rate": 0.0004479121932317661,
      "loss": 2.0361,
      "step": 77389
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.8228096961975098,
      "learning_rate": 0.0004479086344032475,
      "loss": 2.2272,
      "step": 77390
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.8221561312675476,
      "learning_rate": 0.0004479050755472299,
      "loss": 2.291,
      "step": 77391
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.8169741630554199,
      "learning_rate": 0.000447901516663714,
      "loss": 2.2954,
      "step": 77392
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.8690973520278931,
      "learning_rate": 0.0004478979577527004,
      "loss": 2.2513,
      "step": 77393
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.7690479159355164,
      "learning_rate": 0.0004478943988141897,
      "loss": 2.1954,
      "step": 77394
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.8141813278198242,
      "learning_rate": 0.00044789083984818275,
      "loss": 2.0726,
      "step": 77395
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.789446234703064,
      "learning_rate": 0.00044788728085468,
      "loss": 2.2851,
      "step": 77396
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.8791523575782776,
      "learning_rate": 0.00044788372183368223,
      "loss": 2.2499,
      "step": 77397
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.7999670505523682,
      "learning_rate": 0.0004478801627851901,
      "loss": 2.0522,
      "step": 77398
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.7970829010009766,
      "learning_rate": 0.00044787660370920426,
      "loss": 2.3245,
      "step": 77399
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.9007185697555542,
      "learning_rate": 0.0004478730446057254,
      "loss": 2.2982,
      "step": 77400
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.8147241473197937,
      "learning_rate": 0.00044786948547475414,
      "loss": 2.3084,
      "step": 77401
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.7848381996154785,
      "learning_rate": 0.0004478659263162911,
      "loss": 2.2481,
      "step": 77402
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.8503088355064392,
      "learning_rate": 0.000447862367130337,
      "loss": 2.323,
      "step": 77403
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.8557820320129395,
      "learning_rate": 0.00044785880791689255,
      "loss": 2.1786,
      "step": 77404
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.7851707339286804,
      "learning_rate": 0.00044785524867595834,
      "loss": 2.3229,
      "step": 77405
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.8206120729446411,
      "learning_rate": 0.000447851689407535,
      "loss": 2.2474,
      "step": 77406
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.9846066832542419,
      "learning_rate": 0.00044784813011162334,
      "loss": 2.1628,
      "step": 77407
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.8176416754722595,
      "learning_rate": 0.00044784457078822385,
      "loss": 2.1019,
      "step": 77408
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.7162631154060364,
      "learning_rate": 0.0004478410114373372,
      "loss": 2.1237,
      "step": 77409
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.8497728109359741,
      "learning_rate": 0.00044783745205896424,
      "loss": 2.3102,
      "step": 77410
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.9751547574996948,
      "learning_rate": 0.0004478338926531055,
      "loss": 2.3299,
      "step": 77411
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.8510832786560059,
      "learning_rate": 0.00044783033321976154,
      "loss": 2.2931,
      "step": 77412
    },
    {
      "epoch": 1.01,
      "grad_norm": 1.0568186044692993,
      "learning_rate": 0.0004478267737589332,
      "loss": 2.2149,
      "step": 77413
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.8042052388191223,
      "learning_rate": 0.0004478232142706211,
      "loss": 2.246,
      "step": 77414
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.7339973449707031,
      "learning_rate": 0.00044781965475482573,
      "loss": 2.3244,
      "step": 77415
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.7085840702056885,
      "learning_rate": 0.0004478160952115481,
      "loss": 2.2182,
      "step": 77416
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.7177793383598328,
      "learning_rate": 0.0004478125356407885,
      "loss": 2.1239,
      "step": 77417
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.7409931421279907,
      "learning_rate": 0.00044780897604254784,
      "loss": 2.0967,
      "step": 77418
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.8003421425819397,
      "learning_rate": 0.0004478054164168267,
      "loss": 2.1893,
      "step": 77419
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.7525426745414734,
      "learning_rate": 0.0004478018567636257,
      "loss": 2.0856,
      "step": 77420
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.7511681914329529,
      "learning_rate": 0.00044779829708294566,
      "loss": 2.1903,
      "step": 77421
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.7604749798774719,
      "learning_rate": 0.00044779473737478705,
      "loss": 1.8734,
      "step": 77422
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.772089421749115,
      "learning_rate": 0.0004477911776391506,
      "loss": 1.9742,
      "step": 77423
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.8396984338760376,
      "learning_rate": 0.000447787617876037,
      "loss": 2.3803,
      "step": 77424
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.7397716045379639,
      "learning_rate": 0.0004477840580854469,
      "loss": 2.1569,
      "step": 77425
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.754743218421936,
      "learning_rate": 0.0004477804982673809,
      "loss": 2.0834,
      "step": 77426
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.8885754942893982,
      "learning_rate": 0.0004477769384218397,
      "loss": 2.159,
      "step": 77427
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.750582218170166,
      "learning_rate": 0.00044777337854882407,
      "loss": 2.3194,
      "step": 77428
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.7761894464492798,
      "learning_rate": 0.00044776981864833455,
      "loss": 2.0074,
      "step": 77429
    },
    {
      "epoch": 1.01,
      "grad_norm": 1.0205597877502441,
      "learning_rate": 0.0004477662587203718,
      "loss": 2.2424,
      "step": 77430
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.7835619449615479,
      "learning_rate": 0.0004477626987649365,
      "loss": 2.3219,
      "step": 77431
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.7768819332122803,
      "learning_rate": 0.00044775913878202943,
      "loss": 2.3383,
      "step": 77432
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.8921135067939758,
      "learning_rate": 0.000447755578771651,
      "loss": 2.0542,
      "step": 77433
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.8698723316192627,
      "learning_rate": 0.00044775201873380215,
      "loss": 2.2229,
      "step": 77434
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.757703959941864,
      "learning_rate": 0.00044774845866848333,
      "loss": 2.2412,
      "step": 77435
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.7532973289489746,
      "learning_rate": 0.00044774489857569527,
      "loss": 1.9349,
      "step": 77436
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.7091661095619202,
      "learning_rate": 0.0004477413384554386,
      "loss": 1.881,
      "step": 77437
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.742948055267334,
      "learning_rate": 0.0004477377783077142,
      "loss": 2.4101,
      "step": 77438
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.7673704624176025,
      "learning_rate": 0.00044773421813252244,
      "loss": 2.1362,
      "step": 77439
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.7395899295806885,
      "learning_rate": 0.00044773065792986404,
      "loss": 2.1741,
      "step": 77440
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.8606601357460022,
      "learning_rate": 0.0004477270976997399,
      "loss": 2.1932,
      "step": 77441
    },
    {
      "epoch": 1.01,
      "grad_norm": 1.0045874118804932,
      "learning_rate": 0.0004477235374421503,
      "loss": 2.0772,
      "step": 77442
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.8963353037834167,
      "learning_rate": 0.00044771997715709625,
      "loss": 2.2969,
      "step": 77443
    },
    {
      "epoch": 1.01,
      "grad_norm": 1.0811492204666138,
      "learning_rate": 0.0004477164168445783,
      "loss": 2.2917,
      "step": 77444
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.9429827332496643,
      "learning_rate": 0.000447712856504597,
      "loss": 2.2439,
      "step": 77445
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.8801696300506592,
      "learning_rate": 0.00044770929613715307,
      "loss": 2.036,
      "step": 77446
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.8288832306861877,
      "learning_rate": 0.00044770573574224725,
      "loss": 2.2006,
      "step": 77447
    },
    {
      "epoch": 1.01,
      "grad_norm": 1.1621018648147583,
      "learning_rate": 0.0004477021753198801,
      "loss": 2.3231,
      "step": 77448
    },
    {
      "epoch": 1.01,
      "grad_norm": 1.0845943689346313,
      "learning_rate": 0.00044769861487005234,
      "loss": 2.0876,
      "step": 77449
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.8252375721931458,
      "learning_rate": 0.00044769505439276467,
      "loss": 2.1767,
      "step": 77450
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.9073213934898376,
      "learning_rate": 0.0004476914938880176,
      "loss": 2.1262,
      "step": 77451
    },
    {
      "epoch": 1.01,
      "grad_norm": 1.2115734815597534,
      "learning_rate": 0.00044768793335581196,
      "loss": 2.3122,
      "step": 77452
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.9168810844421387,
      "learning_rate": 0.0004476843727961484,
      "loss": 2.2123,
      "step": 77453
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.7758169174194336,
      "learning_rate": 0.00044768081220902746,
      "loss": 2.1973,
      "step": 77454
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.9571504592895508,
      "learning_rate": 0.00044767725159444987,
      "loss": 2.0245,
      "step": 77455
    },
    {
      "epoch": 1.01,
      "grad_norm": 1.2718455791473389,
      "learning_rate": 0.0004476736909524163,
      "loss": 2.1391,
      "step": 77456
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.7913890480995178,
      "learning_rate": 0.0004476701302829274,
      "loss": 2.2998,
      "step": 77457
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.9307761192321777,
      "learning_rate": 0.00044766656958598384,
      "loss": 2.2003,
      "step": 77458
    },
    {
      "epoch": 1.01,
      "grad_norm": 1.2760498523712158,
      "learning_rate": 0.0004476630088615863,
      "loss": 2.1899,
      "step": 77459
    },
    {
      "epoch": 1.01,
      "grad_norm": 1.1488220691680908,
      "learning_rate": 0.0004476594481097354,
      "loss": 2.1198,
      "step": 77460
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.837994396686554,
      "learning_rate": 0.00044765588733043177,
      "loss": 2.2629,
      "step": 77461
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.9035682678222656,
      "learning_rate": 0.00044765232652367614,
      "loss": 2.4437,
      "step": 77462
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.9829607605934143,
      "learning_rate": 0.0004476487656894692,
      "loss": 2.0059,
      "step": 77463
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.8629360795021057,
      "learning_rate": 0.00044764520482781155,
      "loss": 2.051,
      "step": 77464
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.8482212424278259,
      "learning_rate": 0.0004476416439387039,
      "loss": 2.3293,
      "step": 77465
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.9054971933364868,
      "learning_rate": 0.0004476380830221468,
      "loss": 1.9673,
      "step": 77466
    },
    {
      "epoch": 1.01,
      "grad_norm": 1.0014907121658325,
      "learning_rate": 0.0004476345220781411,
      "loss": 2.111,
      "step": 77467
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.7586553692817688,
      "learning_rate": 0.00044763096110668726,
      "loss": 2.2941,
      "step": 77468
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.7675370573997498,
      "learning_rate": 0.000447627400107786,
      "loss": 2.0873,
      "step": 77469
    },
    {
      "epoch": 1.01,
      "grad_norm": 1.0505841970443726,
      "learning_rate": 0.0004476238390814382,
      "loss": 2.2473,
      "step": 77470
    },
    {
      "epoch": 1.01,
      "grad_norm": 1.1783279180526733,
      "learning_rate": 0.0004476202780276442,
      "loss": 2.2842,
      "step": 77471
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.8111028671264648,
      "learning_rate": 0.0004476167169464048,
      "loss": 2.3883,
      "step": 77472
    },
    {
      "epoch": 1.01,
      "grad_norm": 1.0316863059997559,
      "learning_rate": 0.0004476131558377208,
      "loss": 2.2233,
      "step": 77473
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.8814354538917542,
      "learning_rate": 0.00044760959470159257,
      "loss": 2.2821,
      "step": 77474
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.7556944489479065,
      "learning_rate": 0.000447606033538021,
      "loss": 2.1282,
      "step": 77475
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.7106242179870605,
      "learning_rate": 0.0004476024723470067,
      "loss": 2.106,
      "step": 77476
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.8103755712509155,
      "learning_rate": 0.00044759891112855027,
      "loss": 2.2549,
      "step": 77477
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.7481840252876282,
      "learning_rate": 0.00044759534988265243,
      "loss": 2.2161,
      "step": 77478
    },
    {
      "epoch": 1.01,
      "grad_norm": 1.019049048423767,
      "learning_rate": 0.00044759178860931387,
      "loss": 2.3849,
      "step": 77479
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.7807308435440063,
      "learning_rate": 0.00044758822730853523,
      "loss": 2.1602,
      "step": 77480
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.7716469764709473,
      "learning_rate": 0.000447584665980317,
      "loss": 2.1889,
      "step": 77481
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.8184802532196045,
      "learning_rate": 0.00044758110462466013,
      "loss": 2.211,
      "step": 77482
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.9209733605384827,
      "learning_rate": 0.0004475775432415651,
      "loss": 2.4362,
      "step": 77483
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.7333733439445496,
      "learning_rate": 0.00044757398183103265,
      "loss": 2.2882,
      "step": 77484
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.8474074602127075,
      "learning_rate": 0.0004475704203930634,
      "loss": 2.0895,
      "step": 77485
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.9441727995872498,
      "learning_rate": 0.00044756685892765805,
      "loss": 2.098,
      "step": 77486
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.9239978790283203,
      "learning_rate": 0.00044756329743481717,
      "loss": 2.177,
      "step": 77487
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.7965266108512878,
      "learning_rate": 0.00044755973591454153,
      "loss": 2.2631,
      "step": 77488
    },
    {
      "epoch": 1.01,
      "grad_norm": 1.342713475227356,
      "learning_rate": 0.00044755617436683173,
      "loss": 2.1381,
      "step": 77489
    },
    {
      "epoch": 1.01,
      "grad_norm": 1.1216477155685425,
      "learning_rate": 0.0004475526127916884,
      "loss": 2.1111,
      "step": 77490
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.7975832223892212,
      "learning_rate": 0.0004475490511891123,
      "loss": 2.0641,
      "step": 77491
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.951228141784668,
      "learning_rate": 0.0004475454895591042,
      "loss": 2.0276,
      "step": 77492
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.7637941837310791,
      "learning_rate": 0.0004475419279016644,
      "loss": 2.2428,
      "step": 77493
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.8733532428741455,
      "learning_rate": 0.0004475383662167938,
      "loss": 2.1162,
      "step": 77494
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.7928242087364197,
      "learning_rate": 0.00044753480450449316,
      "loss": 2.1821,
      "step": 77495
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.8267008066177368,
      "learning_rate": 0.00044753124276476286,
      "loss": 2.1735,
      "step": 77496
    },
    {
      "epoch": 1.01,
      "grad_norm": 1.0802372694015503,
      "learning_rate": 0.00044752768099760377,
      "loss": 2.0967,
      "step": 77497
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.7874983549118042,
      "learning_rate": 0.0004475241192030166,
      "loss": 2.151,
      "step": 77498
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.807611346244812,
      "learning_rate": 0.0004475205573810018,
      "loss": 1.964,
      "step": 77499
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.8506004214286804,
      "learning_rate": 0.00044751699553156015,
      "loss": 2.029,
      "step": 77500
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.9247059226036072,
      "learning_rate": 0.00044751343365469237,
      "loss": 2.0704,
      "step": 77501
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.8475856184959412,
      "learning_rate": 0.000447509871750399,
      "loss": 2.1507,
      "step": 77502
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.7670721411705017,
      "learning_rate": 0.0004475063098186807,
      "loss": 2.3593,
      "step": 77503
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.7695512771606445,
      "learning_rate": 0.00044750274785953825,
      "loss": 2.2223,
      "step": 77504
    },
    {
      "epoch": 1.01,
      "grad_norm": 1.0107135772705078,
      "learning_rate": 0.0004474991858729723,
      "loss": 2.0923,
      "step": 77505
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.7636315822601318,
      "learning_rate": 0.0004474956238589834,
      "loss": 2.3385,
      "step": 77506
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.7579138278961182,
      "learning_rate": 0.0004474920618175723,
      "loss": 2.3377,
      "step": 77507
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.9244365096092224,
      "learning_rate": 0.00044748849974873974,
      "loss": 2.3185,
      "step": 77508
    },
    {
      "epoch": 1.01,
      "grad_norm": 1.0102360248565674,
      "learning_rate": 0.0004474849376524862,
      "loss": 1.8547,
      "step": 77509
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.7658912539482117,
      "learning_rate": 0.00044748137552881237,
      "loss": 2.1199,
      "step": 77510
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.8135449290275574,
      "learning_rate": 0.00044747781337771907,
      "loss": 2.0925,
      "step": 77511
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.814558744430542,
      "learning_rate": 0.00044747425119920684,
      "loss": 2.2853,
      "step": 77512
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.8295556902885437,
      "learning_rate": 0.0004474706889932763,
      "loss": 1.9369,
      "step": 77513
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.7345930337905884,
      "learning_rate": 0.00044746712675992824,
      "loss": 2.1888,
      "step": 77514
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.7871349453926086,
      "learning_rate": 0.00044746356449916323,
      "loss": 2.2396,
      "step": 77515
    },
    {
      "epoch": 1.01,
      "grad_norm": 1.1415661573410034,
      "learning_rate": 0.0004474600022109819,
      "loss": 2.1466,
      "step": 77516
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.7704986333847046,
      "learning_rate": 0.00044745643989538514,
      "loss": 2.1089,
      "step": 77517
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.7675682902336121,
      "learning_rate": 0.0004474528775523733,
      "loss": 2.0562,
      "step": 77518
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.9759286642074585,
      "learning_rate": 0.00044744931518194716,
      "loss": 2.1405,
      "step": 77519
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.779009222984314,
      "learning_rate": 0.00044744575278410755,
      "loss": 2.2118,
      "step": 77520
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.7754679322242737,
      "learning_rate": 0.00044744219035885486,
      "loss": 2.2135,
      "step": 77521
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.9531835913658142,
      "learning_rate": 0.0004474386279061899,
      "loss": 2.1316,
      "step": 77522
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.7902441620826721,
      "learning_rate": 0.0004474350654261134,
      "loss": 2.1417,
      "step": 77523
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.7981431484222412,
      "learning_rate": 0.00044743150291862594,
      "loss": 2.2964,
      "step": 77524
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.8352962136268616,
      "learning_rate": 0.0004474279403837281,
      "loss": 2.4015,
      "step": 77525
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.9780823588371277,
      "learning_rate": 0.00044742437782142067,
      "loss": 2.0973,
      "step": 77526
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.7567596435546875,
      "learning_rate": 0.00044742081523170427,
      "loss": 2.3186,
      "step": 77527
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.779072105884552,
      "learning_rate": 0.00044741725261457946,
      "loss": 2.2244,
      "step": 77528
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.9146429300308228,
      "learning_rate": 0.00044741368997004715,
      "loss": 2.4483,
      "step": 77529
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.8249164819717407,
      "learning_rate": 0.0004474101272981078,
      "loss": 2.3079,
      "step": 77530
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.7553514242172241,
      "learning_rate": 0.000447406564598762,
      "loss": 2.0575,
      "step": 77531
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.8323084115982056,
      "learning_rate": 0.00044740300187201076,
      "loss": 2.1408,
      "step": 77532
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.749077558517456,
      "learning_rate": 0.0004473994391178544,
      "loss": 2.3258,
      "step": 77533
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.844515323638916,
      "learning_rate": 0.0004473958763362937,
      "loss": 2.3566,
      "step": 77534
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.8586936593055725,
      "learning_rate": 0.0004473923135273294,
      "loss": 2.2438,
      "step": 77535
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.9067480564117432,
      "learning_rate": 0.000447388750690962,
      "loss": 1.9395,
      "step": 77536
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.8258357644081116,
      "learning_rate": 0.00044738518782719224,
      "loss": 2.2028,
      "step": 77537
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.723609447479248,
      "learning_rate": 0.0004473816249360208,
      "loss": 2.1383,
      "step": 77538
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.795732855796814,
      "learning_rate": 0.00044737806201744837,
      "loss": 2.0735,
      "step": 77539
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.7706825733184814,
      "learning_rate": 0.0004473744990714755,
      "loss": 2.506,
      "step": 77540
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.8128580451011658,
      "learning_rate": 0.000447370936098103,
      "loss": 2.2125,
      "step": 77541
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.7581309080123901,
      "learning_rate": 0.00044736737309733146,
      "loss": 2.1724,
      "step": 77542
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.8257997632026672,
      "learning_rate": 0.0004473638100691615,
      "loss": 2.109,
      "step": 77543
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.8249915838241577,
      "learning_rate": 0.0004473602470135939,
      "loss": 2.362,
      "step": 77544
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.8291131258010864,
      "learning_rate": 0.00044735668393062914,
      "loss": 2.1634,
      "step": 77545
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.8609110116958618,
      "learning_rate": 0.00044735312082026807,
      "loss": 2.1711,
      "step": 77546
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.8467714786529541,
      "learning_rate": 0.00044734955768251127,
      "loss": 2.1669,
      "step": 77547
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.7426770329475403,
      "learning_rate": 0.0004473459945173594,
      "loss": 2.2446,
      "step": 77548
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.8067000508308411,
      "learning_rate": 0.00044734243132481315,
      "loss": 2.3193,
      "step": 77549
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.7795809507369995,
      "learning_rate": 0.0004473388681048731,
      "loss": 2.0755,
      "step": 77550
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.7641762495040894,
      "learning_rate": 0.00044733530485754,
      "loss": 2.2187,
      "step": 77551
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.9993726015090942,
      "learning_rate": 0.00044733174158281456,
      "loss": 2.2387,
      "step": 77552
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.7482088208198547,
      "learning_rate": 0.00044732817828069726,
      "loss": 2.0059,
      "step": 77553
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.8057750463485718,
      "learning_rate": 0.00044732461495118885,
      "loss": 2.365,
      "step": 77554
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.8232668042182922,
      "learning_rate": 0.0004473210515942902,
      "loss": 2.4134,
      "step": 77555
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.8422945141792297,
      "learning_rate": 0.0004473174882100016,
      "loss": 2.2308,
      "step": 77556
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.8513671159744263,
      "learning_rate": 0.00044731392479832393,
      "loss": 2.1573,
      "step": 77557
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.7545437812805176,
      "learning_rate": 0.0004473103613592579,
      "loss": 2.1296,
      "step": 77558
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.78626948595047,
      "learning_rate": 0.000447306797892804,
      "loss": 2.3432,
      "step": 77559
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.8414831757545471,
      "learning_rate": 0.000447303234398963,
      "loss": 2.1481,
      "step": 77560
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.8226970434188843,
      "learning_rate": 0.00044729967087773554,
      "loss": 1.9627,
      "step": 77561
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.7035917639732361,
      "learning_rate": 0.00044729610732912234,
      "loss": 2.1208,
      "step": 77562
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.766167402267456,
      "learning_rate": 0.0004472925437531241,
      "loss": 1.996,
      "step": 77563
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.8527196049690247,
      "learning_rate": 0.0004472889801497412,
      "loss": 2.5126,
      "step": 77564
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.8248746395111084,
      "learning_rate": 0.00044728541651897465,
      "loss": 2.241,
      "step": 77565
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.954663872718811,
      "learning_rate": 0.0004472818528608249,
      "loss": 2.087,
      "step": 77566
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.7852861881256104,
      "learning_rate": 0.0004472782891752926,
      "loss": 2.1415,
      "step": 77567
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.8736080527305603,
      "learning_rate": 0.00044727472546237865,
      "loss": 2.3343,
      "step": 77568
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.863617479801178,
      "learning_rate": 0.0004472711617220835,
      "loss": 2.3219,
      "step": 77569
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.8159542083740234,
      "learning_rate": 0.0004472675979544078,
      "loss": 2.2792,
      "step": 77570
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.7743284702301025,
      "learning_rate": 0.0004472640341593522,
      "loss": 2.0528,
      "step": 77571
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.8598917126655579,
      "learning_rate": 0.00044726047033691763,
      "loss": 2.2138,
      "step": 77572
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.7645015120506287,
      "learning_rate": 0.0004472569064871044,
      "loss": 2.3974,
      "step": 77573
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.8517484068870544,
      "learning_rate": 0.0004472533426099134,
      "loss": 2.3099,
      "step": 77574
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.7727122902870178,
      "learning_rate": 0.00044724977870534526,
      "loss": 2.1772,
      "step": 77575
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.779009222984314,
      "learning_rate": 0.0004472462147734005,
      "loss": 1.9973,
      "step": 77576
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.7579885125160217,
      "learning_rate": 0.00044724265081407997,
      "loss": 1.9359,
      "step": 77577
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.8564570546150208,
      "learning_rate": 0.00044723908682738425,
      "loss": 2.3546,
      "step": 77578
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.8082691431045532,
      "learning_rate": 0.000447235522813314,
      "loss": 2.2759,
      "step": 77579
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.8982774615287781,
      "learning_rate": 0.0004472319587718698,
      "loss": 2.2467,
      "step": 77580
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.7635768055915833,
      "learning_rate": 0.00044722839470305256,
      "loss": 2.0429,
      "step": 77581
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.771058976650238,
      "learning_rate": 0.00044722483060686267,
      "loss": 2.1798,
      "step": 77582
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.8481311202049255,
      "learning_rate": 0.0004472212664833009,
      "loss": 2.0977,
      "step": 77583
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.8521429300308228,
      "learning_rate": 0.00044721770233236796,
      "loss": 2.2184,
      "step": 77584
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.8481256365776062,
      "learning_rate": 0.0004472141381540645,
      "loss": 2.1731,
      "step": 77585
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.8854119777679443,
      "learning_rate": 0.00044721057394839106,
      "loss": 2.2891,
      "step": 77586
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.7870499491691589,
      "learning_rate": 0.0004472070097153485,
      "loss": 2.2566,
      "step": 77587
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.7816187143325806,
      "learning_rate": 0.00044720344545493733,
      "loss": 2.181,
      "step": 77588
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.6793174743652344,
      "learning_rate": 0.00044719988116715824,
      "loss": 2.0003,
      "step": 77589
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.8352473378181458,
      "learning_rate": 0.00044719631685201193,
      "loss": 2.2317,
      "step": 77590
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.7217522263526917,
      "learning_rate": 0.00044719275250949906,
      "loss": 2.2545,
      "step": 77591
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.7206969857215881,
      "learning_rate": 0.0004471891881396203,
      "loss": 1.9165,
      "step": 77592
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.866636335849762,
      "learning_rate": 0.0004471856237423763,
      "loss": 2.3235,
      "step": 77593
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.7548819780349731,
      "learning_rate": 0.0004471820593177677,
      "loss": 2.1719,
      "step": 77594
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.7852911353111267,
      "learning_rate": 0.00044717849486579505,
      "loss": 2.1145,
      "step": 77595
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.7956947088241577,
      "learning_rate": 0.00044717493038645933,
      "loss": 2.304,
      "step": 77596
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.7935196757316589,
      "learning_rate": 0.00044717136587976085,
      "loss": 2.2136,
      "step": 77597
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.7836941480636597,
      "learning_rate": 0.0004471678013457005,
      "loss": 2.0888,
      "step": 77598
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.8232832551002502,
      "learning_rate": 0.0004471642367842789,
      "loss": 2.1007,
      "step": 77599
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.8459030389785767,
      "learning_rate": 0.0004471606721954967,
      "loss": 2.4146,
      "step": 77600
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.7941240072250366,
      "learning_rate": 0.0004471571075793545,
      "loss": 2.1648,
      "step": 77601
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.7570844292640686,
      "learning_rate": 0.00044715354293585306,
      "loss": 2.1717,
      "step": 77602
    },
    {
      "epoch": 1.01,
      "grad_norm": 1.021181583404541,
      "learning_rate": 0.0004471499782649929,
      "loss": 2.08,
      "step": 77603
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.952242374420166,
      "learning_rate": 0.0004471464135667749,
      "loss": 2.3404,
      "step": 77604
    },
    {
      "epoch": 1.01,
      "grad_norm": 1.0849919319152832,
      "learning_rate": 0.0004471428488411996,
      "loss": 2.0025,
      "step": 77605
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.7979874610900879,
      "learning_rate": 0.00044713928408826766,
      "loss": 2.1689,
      "step": 77606
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.7762320637702942,
      "learning_rate": 0.0004471357193079797,
      "loss": 2.2618,
      "step": 77607
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.8131390810012817,
      "learning_rate": 0.0004471321545003365,
      "loss": 2.2496,
      "step": 77608
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.8150834441184998,
      "learning_rate": 0.00044712858966533856,
      "loss": 2.1346,
      "step": 77609
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.8734650611877441,
      "learning_rate": 0.0004471250248029867,
      "loss": 2.3089,
      "step": 77610
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.8857824802398682,
      "learning_rate": 0.00044712145991328153,
      "loss": 2.4819,
      "step": 77611
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.7834785580635071,
      "learning_rate": 0.00044711789499622364,
      "loss": 2.1134,
      "step": 77612
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.8417541980743408,
      "learning_rate": 0.00044711433005181387,
      "loss": 2.375,
      "step": 77613
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.7759962677955627,
      "learning_rate": 0.0004471107650800527,
      "loss": 2.1472,
      "step": 77614
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.8447902202606201,
      "learning_rate": 0.00044710720008094083,
      "loss": 2.4379,
      "step": 77615
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.8017969131469727,
      "learning_rate": 0.00044710363505447897,
      "loss": 2.4136,
      "step": 77616
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.7928536534309387,
      "learning_rate": 0.00044710007000066783,
      "loss": 2.0607,
      "step": 77617
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.8647276759147644,
      "learning_rate": 0.0004470965049195079,
      "loss": 2.1382,
      "step": 77618
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.7566058039665222,
      "learning_rate": 0.000447092939811,
      "loss": 2.248,
      "step": 77619
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.8278437256813049,
      "learning_rate": 0.0004470893746751449,
      "loss": 2.1333,
      "step": 77620
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.7999809980392456,
      "learning_rate": 0.00044708580951194284,
      "loss": 2.274,
      "step": 77621
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.8048220872879028,
      "learning_rate": 0.0004470822443213949,
      "loss": 2.0774,
      "step": 77622
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.8230763077735901,
      "learning_rate": 0.00044707867910350163,
      "loss": 2.2012,
      "step": 77623
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.7401043772697449,
      "learning_rate": 0.0004470751138582636,
      "loss": 2.1671,
      "step": 77624
    },
    {
      "epoch": 1.01,
      "grad_norm": 1.4459134340286255,
      "learning_rate": 0.0004470715485856815,
      "loss": 2.0467,
      "step": 77625
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.8396400809288025,
      "learning_rate": 0.0004470679832857561,
      "loss": 2.3269,
      "step": 77626
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.9547178149223328,
      "learning_rate": 0.0004470644179584879,
      "loss": 2.1891,
      "step": 77627
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.7770143151283264,
      "learning_rate": 0.00044706085260387774,
      "loss": 2.1002,
      "step": 77628
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.7933021187782288,
      "learning_rate": 0.0004470572872219261,
      "loss": 1.967,
      "step": 77629
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.7626094818115234,
      "learning_rate": 0.00044705372181263374,
      "loss": 2.2363,
      "step": 77630
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.7545062303543091,
      "learning_rate": 0.0004470501563760014,
      "loss": 2.2957,
      "step": 77631
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.7781187891960144,
      "learning_rate": 0.00044704659091202964,
      "loss": 2.1041,
      "step": 77632
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.8446199297904968,
      "learning_rate": 0.00044704302542071913,
      "loss": 2.3248,
      "step": 77633
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.8031569719314575,
      "learning_rate": 0.0004470394599020705,
      "loss": 2.3982,
      "step": 77634
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.8065248727798462,
      "learning_rate": 0.00044703589435608445,
      "loss": 2.4084,
      "step": 77635
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.8286408185958862,
      "learning_rate": 0.0004470323287827618,
      "loss": 2.2937,
      "step": 77636
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.7216399312019348,
      "learning_rate": 0.00044702876318210294,
      "loss": 2.0946,
      "step": 77637
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.792015016078949,
      "learning_rate": 0.0004470251975541086,
      "loss": 2.2234,
      "step": 77638
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.8664920926094055,
      "learning_rate": 0.0004470216318987796,
      "loss": 2.0774,
      "step": 77639
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.8377139568328857,
      "learning_rate": 0.00044701806621611645,
      "loss": 2.144,
      "step": 77640
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.8405691981315613,
      "learning_rate": 0.0004470145005061199,
      "loss": 2.249,
      "step": 77641
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.7926894426345825,
      "learning_rate": 0.0004470109347687906,
      "loss": 2.1928,
      "step": 77642
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.8689056634902954,
      "learning_rate": 0.0004470073690041292,
      "loss": 2.3949,
      "step": 77643
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.7507843971252441,
      "learning_rate": 0.0004470038032121363,
      "loss": 2.2363,
      "step": 77644
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.8476333022117615,
      "learning_rate": 0.0004470002373928126,
      "loss": 2.3491,
      "step": 77645
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.7296361327171326,
      "learning_rate": 0.0004469966715461588,
      "loss": 2.0252,
      "step": 77646
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.8140970468521118,
      "learning_rate": 0.0004469931056721757,
      "loss": 2.1769,
      "step": 77647
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.7827987670898438,
      "learning_rate": 0.00044698953977086354,
      "loss": 2.5253,
      "step": 77648
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.7917929291725159,
      "learning_rate": 0.0004469859738422234,
      "loss": 2.2851,
      "step": 77649
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.7955560684204102,
      "learning_rate": 0.0004469824078862558,
      "loss": 2.2826,
      "step": 77650
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.7669976353645325,
      "learning_rate": 0.00044697884190296136,
      "loss": 2.3398,
      "step": 77651
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.7501378059387207,
      "learning_rate": 0.0004469752758923408,
      "loss": 2.1455,
      "step": 77652
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.8441901206970215,
      "learning_rate": 0.0004469717098543947,
      "loss": 2.5162,
      "step": 77653
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.8592080473899841,
      "learning_rate": 0.0004469681437891239,
      "loss": 2.1456,
      "step": 77654
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.7989119291305542,
      "learning_rate": 0.00044696457769652885,
      "loss": 2.1749,
      "step": 77655
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.8118494153022766,
      "learning_rate": 0.0004469610115766104,
      "loss": 2.2108,
      "step": 77656
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.7837371826171875,
      "learning_rate": 0.00044695744542936906,
      "loss": 2.284,
      "step": 77657
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.9638859033584595,
      "learning_rate": 0.00044695387925480556,
      "loss": 2.3012,
      "step": 77658
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.7070909738540649,
      "learning_rate": 0.00044695031305292055,
      "loss": 1.8952,
      "step": 77659
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.7847621440887451,
      "learning_rate": 0.0004469467468237147,
      "loss": 2.2662,
      "step": 77660
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.8093084692955017,
      "learning_rate": 0.00044694318056718874,
      "loss": 2.1073,
      "step": 77661
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.765008270740509,
      "learning_rate": 0.00044693961428334314,
      "loss": 2.3721,
      "step": 77662
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.7507612705230713,
      "learning_rate": 0.00044693604797217885,
      "loss": 2.0928,
      "step": 77663
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.7253611087799072,
      "learning_rate": 0.0004469324816336963,
      "loss": 2.1559,
      "step": 77664
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.7884332537651062,
      "learning_rate": 0.0004469289152678962,
      "loss": 2.3178,
      "step": 77665
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.7818453907966614,
      "learning_rate": 0.0004469253488747793,
      "loss": 2.2295,
      "step": 77666
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.7314981818199158,
      "learning_rate": 0.00044692178245434613,
      "loss": 2.0119,
      "step": 77667
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.757346510887146,
      "learning_rate": 0.0004469182160065975,
      "loss": 1.9329,
      "step": 77668
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.796644389629364,
      "learning_rate": 0.000446914649531534,
      "loss": 2.2572,
      "step": 77669
    },
    {
      "epoch": 1.01,
      "grad_norm": 1.1256760358810425,
      "learning_rate": 0.0004469110830291563,
      "loss": 2.1491,
      "step": 77670
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.7613105773925781,
      "learning_rate": 0.00044690751649946496,
      "loss": 2.1851,
      "step": 77671
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.9013378024101257,
      "learning_rate": 0.00044690394994246086,
      "loss": 2.3077,
      "step": 77672
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.7771020531654358,
      "learning_rate": 0.00044690038335814446,
      "loss": 2.3209,
      "step": 77673
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.7697900533676147,
      "learning_rate": 0.0004468968167465165,
      "loss": 2.1083,
      "step": 77674
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.7584394216537476,
      "learning_rate": 0.0004468932501075777,
      "loss": 2.2564,
      "step": 77675
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.70545494556427,
      "learning_rate": 0.0004468896834413287,
      "loss": 2.0438,
      "step": 77676
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.7503615021705627,
      "learning_rate": 0.00044688611674777005,
      "loss": 2.3384,
      "step": 77677
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.8695390224456787,
      "learning_rate": 0.00044688255002690256,
      "loss": 2.1407,
      "step": 77678
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.7196843028068542,
      "learning_rate": 0.0004468789832787268,
      "loss": 2.1282,
      "step": 77679
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.8051119446754456,
      "learning_rate": 0.0004468754165032434,
      "loss": 2.4793,
      "step": 77680
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.7789141535758972,
      "learning_rate": 0.0004468718497004532,
      "loss": 2.2849,
      "step": 77681
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.959760844707489,
      "learning_rate": 0.0004468682828703567,
      "loss": 2.1604,
      "step": 77682
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.7634472250938416,
      "learning_rate": 0.0004468647160129546,
      "loss": 2.2091,
      "step": 77683
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.7578443288803101,
      "learning_rate": 0.0004468611491282477,
      "loss": 2.1721,
      "step": 77684
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.7995215058326721,
      "learning_rate": 0.00044685758221623637,
      "loss": 2.4683,
      "step": 77685
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.7642136216163635,
      "learning_rate": 0.00044685401527692155,
      "loss": 2.3065,
      "step": 77686
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.7263584733009338,
      "learning_rate": 0.0004468504483103038,
      "loss": 2.2263,
      "step": 77687
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.9225971698760986,
      "learning_rate": 0.0004468468813163837,
      "loss": 2.0651,
      "step": 77688
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.8623363375663757,
      "learning_rate": 0.00044684331429516194,
      "loss": 2.1633,
      "step": 77689
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.9428474307060242,
      "learning_rate": 0.0004468397472466394,
      "loss": 2.2855,
      "step": 77690
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.812744140625,
      "learning_rate": 0.00044683618017081657,
      "loss": 2.201,
      "step": 77691
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.8307055234909058,
      "learning_rate": 0.000446832613067694,
      "loss": 2.3575,
      "step": 77692
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.9147285223007202,
      "learning_rate": 0.0004468290459372726,
      "loss": 1.9945,
      "step": 77693
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.8058264255523682,
      "learning_rate": 0.00044682547877955277,
      "loss": 2.2754,
      "step": 77694
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.9311991930007935,
      "learning_rate": 0.0004468219115945354,
      "loss": 2.3232,
      "step": 77695
    },
    {
      "epoch": 1.01,
      "grad_norm": 1.0688856840133667,
      "learning_rate": 0.000446818344382221,
      "loss": 2.3314,
      "step": 77696
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.8169265985488892,
      "learning_rate": 0.0004468147771426104,
      "loss": 2.4286,
      "step": 77697
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.8460907340049744,
      "learning_rate": 0.00044681120987570414,
      "loss": 2.2712,
      "step": 77698
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.9796815514564514,
      "learning_rate": 0.0004468076425815029,
      "loss": 2.1157,
      "step": 77699
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.8858323097229004,
      "learning_rate": 0.0004468040752600073,
      "loss": 2.3747,
      "step": 77700
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.790471613407135,
      "learning_rate": 0.00044680050791121804,
      "loss": 2.1431,
      "step": 77701
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.8794527053833008,
      "learning_rate": 0.0004467969405351358,
      "loss": 2.2897,
      "step": 77702
    },
    {
      "epoch": 1.01,
      "grad_norm": 1.349810242652893,
      "learning_rate": 0.0004467933731317613,
      "loss": 2.1712,
      "step": 77703
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.9830493927001953,
      "learning_rate": 0.0004467898057010951,
      "loss": 2.149,
      "step": 77704
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.801164984703064,
      "learning_rate": 0.0004467862382431379,
      "loss": 2.1958,
      "step": 77705
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.8525462746620178,
      "learning_rate": 0.0004467826707578904,
      "loss": 2.443,
      "step": 77706
    },
    {
      "epoch": 1.01,
      "grad_norm": 1.0700764656066895,
      "learning_rate": 0.00044677910324535325,
      "loss": 2.2446,
      "step": 77707
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.846623420715332,
      "learning_rate": 0.00044677553570552703,
      "loss": 2.1496,
      "step": 77708
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.9041938781738281,
      "learning_rate": 0.0004467719681384125,
      "loss": 2.0706,
      "step": 77709
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.9337051510810852,
      "learning_rate": 0.0004467684005440103,
      "loss": 2.1396,
      "step": 77710
    },
    {
      "epoch": 1.01,
      "grad_norm": 1.0546891689300537,
      "learning_rate": 0.000446764832922321,
      "loss": 2.3053,
      "step": 77711
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.7708202004432678,
      "learning_rate": 0.0004467612652733455,
      "loss": 2.3123,
      "step": 77712
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.8017780780792236,
      "learning_rate": 0.00044675769759708415,
      "loss": 2.317,
      "step": 77713
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.9789474606513977,
      "learning_rate": 0.0004467541298935379,
      "loss": 2.3006,
      "step": 77714
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.8482056856155396,
      "learning_rate": 0.0004467505621627072,
      "loss": 2.1634,
      "step": 77715
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.7551398873329163,
      "learning_rate": 0.00044674699440459283,
      "loss": 2.229,
      "step": 77716
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.915213406085968,
      "learning_rate": 0.0004467434266191954,
      "loss": 2.1375,
      "step": 77717
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.8881433010101318,
      "learning_rate": 0.0004467398588065157,
      "loss": 2.2207,
      "step": 77718
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.7858225107192993,
      "learning_rate": 0.0004467362909665542,
      "loss": 2.1187,
      "step": 77719
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.7477415800094604,
      "learning_rate": 0.00044673272309931163,
      "loss": 2.3126,
      "step": 77720
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.8498480916023254,
      "learning_rate": 0.00044672915520478873,
      "loss": 2.2863,
      "step": 77721
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.7536473274230957,
      "learning_rate": 0.0004467255872829861,
      "loss": 2.1885,
      "step": 77722
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.7776135802268982,
      "learning_rate": 0.0004467220193339044,
      "loss": 2.2421,
      "step": 77723
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.7805196642875671,
      "learning_rate": 0.00044671845135754436,
      "loss": 2.2575,
      "step": 77724
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.7749988436698914,
      "learning_rate": 0.0004467148833539066,
      "loss": 2.1427,
      "step": 77725
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.7389546036720276,
      "learning_rate": 0.00044671131532299165,
      "loss": 2.1612,
      "step": 77726
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.816711962223053,
      "learning_rate": 0.00044670774726480035,
      "loss": 2.2224,
      "step": 77727
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.8439189791679382,
      "learning_rate": 0.00044670417917933334,
      "loss": 2.091,
      "step": 77728
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.6984522342681885,
      "learning_rate": 0.0004467006110665913,
      "loss": 2.1926,
      "step": 77729
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.7598192691802979,
      "learning_rate": 0.00044669704292657475,
      "loss": 2.1848,
      "step": 77730
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.9136168956756592,
      "learning_rate": 0.00044669347475928447,
      "loss": 2.1232,
      "step": 77731
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.837020218372345,
      "learning_rate": 0.00044668990656472114,
      "loss": 2.0637,
      "step": 77732
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.8383181691169739,
      "learning_rate": 0.00044668633834288535,
      "loss": 2.2628,
      "step": 77733
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.7372021675109863,
      "learning_rate": 0.0004466827700937778,
      "loss": 1.9928,
      "step": 77734
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.7430248260498047,
      "learning_rate": 0.00044667920181739923,
      "loss": 1.9548,
      "step": 77735
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.7168778777122498,
      "learning_rate": 0.00044667563351375014,
      "loss": 1.9356,
      "step": 77736
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.7958629131317139,
      "learning_rate": 0.00044667206518283125,
      "loss": 2.3628,
      "step": 77737
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.8330893516540527,
      "learning_rate": 0.00044666849682464333,
      "loss": 2.2057,
      "step": 77738
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.8950393199920654,
      "learning_rate": 0.000446664928439187,
      "loss": 2.0953,
      "step": 77739
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.6976038813591003,
      "learning_rate": 0.00044666136002646285,
      "loss": 2.329,
      "step": 77740
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.8772658109664917,
      "learning_rate": 0.0004466577915864716,
      "loss": 2.1534,
      "step": 77741
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.8466660976409912,
      "learning_rate": 0.0004466542231192139,
      "loss": 2.4121,
      "step": 77742
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.7716066837310791,
      "learning_rate": 0.0004466506546246904,
      "loss": 2.1141,
      "step": 77743
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.7433592081069946,
      "learning_rate": 0.0004466470861029017,
      "loss": 2.0089,
      "step": 77744
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.7736724615097046,
      "learning_rate": 0.00044664351755384867,
      "loss": 2.1601,
      "step": 77745
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.7953447103500366,
      "learning_rate": 0.00044663994897753173,
      "loss": 2.1234,
      "step": 77746
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.7608004212379456,
      "learning_rate": 0.0004466363803739517,
      "loss": 2.0309,
      "step": 77747
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.8394717574119568,
      "learning_rate": 0.0004466328117431093,
      "loss": 2.2157,
      "step": 77748
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.8059374094009399,
      "learning_rate": 0.0004466292430850049,
      "loss": 2.3184,
      "step": 77749
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.8855662941932678,
      "learning_rate": 0.0004466256743996395,
      "loss": 2.0901,
      "step": 77750
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.8872720003128052,
      "learning_rate": 0.00044662210568701365,
      "loss": 2.2943,
      "step": 77751
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.7459937334060669,
      "learning_rate": 0.0004466185369471278,
      "loss": 2.1142,
      "step": 77752
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.7735127806663513,
      "learning_rate": 0.00044661496817998287,
      "loss": 2.1491,
      "step": 77753
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.8296061158180237,
      "learning_rate": 0.00044661139938557954,
      "loss": 2.1502,
      "step": 77754
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.827520489692688,
      "learning_rate": 0.00044660783056391833,
      "loss": 2.3184,
      "step": 77755
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.8359377980232239,
      "learning_rate": 0.0004466042617149999,
      "loss": 2.1297,
      "step": 77756
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.8515777587890625,
      "learning_rate": 0.0004466006928388251,
      "loss": 2.2252,
      "step": 77757
    },
    {
      "epoch": 1.01,
      "grad_norm": 1.0634474754333496,
      "learning_rate": 0.0004465971239353944,
      "loss": 2.3284,
      "step": 77758
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.8785507678985596,
      "learning_rate": 0.0004465935550047085,
      "loss": 2.2144,
      "step": 77759
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.8232699632644653,
      "learning_rate": 0.0004465899860467681,
      "loss": 2.19,
      "step": 77760
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.8770831227302551,
      "learning_rate": 0.00044658641706157387,
      "loss": 2.2446,
      "step": 77761
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.9538825750350952,
      "learning_rate": 0.0004465828480491264,
      "loss": 2.1123,
      "step": 77762
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.8288970589637756,
      "learning_rate": 0.0004465792790094265,
      "loss": 2.334,
      "step": 77763
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.783525824546814,
      "learning_rate": 0.0004465757099424747,
      "loss": 2.0194,
      "step": 77764
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.880773663520813,
      "learning_rate": 0.0004465721408482717,
      "loss": 2.259,
      "step": 77765
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.9124559164047241,
      "learning_rate": 0.00044656857172681823,
      "loss": 2.2285,
      "step": 77766
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.7954831123352051,
      "learning_rate": 0.0004465650025781148,
      "loss": 2.0008,
      "step": 77767
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.8450920581817627,
      "learning_rate": 0.00044656143340216224,
      "loss": 2.2643,
      "step": 77768
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.8474077582359314,
      "learning_rate": 0.000446557864198961,
      "loss": 2.1243,
      "step": 77769
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.9084682464599609,
      "learning_rate": 0.0004465542949685121,
      "loss": 2.176,
      "step": 77770
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.8256357312202454,
      "learning_rate": 0.0004465507257108159,
      "loss": 2.3033,
      "step": 77771
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.8035301566123962,
      "learning_rate": 0.00044654715642587317,
      "loss": 2.3749,
      "step": 77772
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.9799274802207947,
      "learning_rate": 0.0004465435871136846,
      "loss": 2.1885,
      "step": 77773
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.7887731194496155,
      "learning_rate": 0.0004465400177742507,
      "loss": 2.0955,
      "step": 77774
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.657865583896637,
      "learning_rate": 0.00044653644840757223,
      "loss": 2.1194,
      "step": 77775
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.7547986507415771,
      "learning_rate": 0.00044653287901365004,
      "loss": 2.3047,
      "step": 77776
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.7801994681358337,
      "learning_rate": 0.00044652930959248444,
      "loss": 2.2348,
      "step": 77777
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.8172123432159424,
      "learning_rate": 0.0004465257401440763,
      "loss": 2.1702,
      "step": 77778
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.951797604560852,
      "learning_rate": 0.0004465221706684264,
      "loss": 2.2506,
      "step": 77779
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.7172883749008179,
      "learning_rate": 0.00044651860116553513,
      "loss": 2.2175,
      "step": 77780
    },
    {
      "epoch": 1.01,
      "grad_norm": 1.016055703163147,
      "learning_rate": 0.00044651503163540337,
      "loss": 2.1125,
      "step": 77781
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.8538762331008911,
      "learning_rate": 0.0004465114620780317,
      "loss": 2.3213,
      "step": 77782
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.7481526136398315,
      "learning_rate": 0.0004465078924934207,
      "loss": 2.1736,
      "step": 77783
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.8369132280349731,
      "learning_rate": 0.0004465043228815711,
      "loss": 2.2322,
      "step": 77784
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.8188430070877075,
      "learning_rate": 0.00044650075324248364,
      "loss": 2.2805,
      "step": 77785
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.7671188116073608,
      "learning_rate": 0.000446497183576159,
      "loss": 2.1014,
      "step": 77786
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.8248372077941895,
      "learning_rate": 0.00044649361388259765,
      "loss": 2.1979,
      "step": 77787
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.8527047634124756,
      "learning_rate": 0.0004464900441618004,
      "loss": 2.0529,
      "step": 77788
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.7076225280761719,
      "learning_rate": 0.00044648647441376793,
      "loss": 2.1602,
      "step": 77789
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.8437214493751526,
      "learning_rate": 0.0004464829046385008,
      "loss": 2.1347,
      "step": 77790
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.7647291421890259,
      "learning_rate": 0.0004464793348359998,
      "loss": 2.2545,
      "step": 77791
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.8022933006286621,
      "learning_rate": 0.00044647576500626547,
      "loss": 2.1555,
      "step": 77792
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.9333431124687195,
      "learning_rate": 0.0004464721951492985,
      "loss": 2.2087,
      "step": 77793
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.9032431840896606,
      "learning_rate": 0.0004464686252650997,
      "loss": 2.1096,
      "step": 77794
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.6884557604789734,
      "learning_rate": 0.00044646505535366957,
      "loss": 2.145,
      "step": 77795
    },
    {
      "epoch": 1.01,
      "grad_norm": 1.021252155303955,
      "learning_rate": 0.0004464614854150088,
      "loss": 2.1439,
      "step": 77796
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.8254850506782532,
      "learning_rate": 0.00044645791544911813,
      "loss": 2.1303,
      "step": 77797
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.8212546110153198,
      "learning_rate": 0.00044645434545599805,
      "loss": 2.3923,
      "step": 77798
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.9340230822563171,
      "learning_rate": 0.00044645077543564943,
      "loss": 2.1096,
      "step": 77799
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.9771890044212341,
      "learning_rate": 0.0004464472053880728,
      "loss": 2.3394,
      "step": 77800
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.7749903202056885,
      "learning_rate": 0.0004464436353132689,
      "loss": 2.0679,
      "step": 77801
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.7174237966537476,
      "learning_rate": 0.00044644006521123836,
      "loss": 2.1738,
      "step": 77802
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.9074690341949463,
      "learning_rate": 0.00044643649508198185,
      "loss": 1.947,
      "step": 77803
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.7884149551391602,
      "learning_rate": 0.0004464329249255001,
      "loss": 2.3416,
      "step": 77804
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.7531910538673401,
      "learning_rate": 0.00044642935474179353,
      "loss": 2.1603,
      "step": 77805
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.8427792191505432,
      "learning_rate": 0.0004464257845308632,
      "loss": 2.1411,
      "step": 77806
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.7850281596183777,
      "learning_rate": 0.0004464222142927094,
      "loss": 2.0966,
      "step": 77807
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.8154980540275574,
      "learning_rate": 0.00044641864402733295,
      "loss": 2.0703,
      "step": 77808
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.8812516331672668,
      "learning_rate": 0.0004464150737347346,
      "loss": 2.3765,
      "step": 77809
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.8558607697486877,
      "learning_rate": 0.00044641150341491486,
      "loss": 2.119,
      "step": 77810
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.8810880780220032,
      "learning_rate": 0.00044640793306787444,
      "loss": 2.4495,
      "step": 77811
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.8135541081428528,
      "learning_rate": 0.00044640436269361404,
      "loss": 2.1727,
      "step": 77812
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.7220208048820496,
      "learning_rate": 0.0004464007922921343,
      "loss": 2.1093,
      "step": 77813
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.9257535338401794,
      "learning_rate": 0.0004463972218634359,
      "loss": 2.2342,
      "step": 77814
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.8999838829040527,
      "learning_rate": 0.00044639365140751946,
      "loss": 2.1281,
      "step": 77815
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.8209730982780457,
      "learning_rate": 0.00044639008092438575,
      "loss": 2.2115,
      "step": 77816
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.774185299873352,
      "learning_rate": 0.0004463865104140353,
      "loss": 2.1058,
      "step": 77817
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.891068696975708,
      "learning_rate": 0.0004463829398764688,
      "loss": 2.2663,
      "step": 77818
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.8698670864105225,
      "learning_rate": 0.00044637936931168703,
      "loss": 2.2555,
      "step": 77819
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.8151272535324097,
      "learning_rate": 0.0004463757987196906,
      "loss": 2.2464,
      "step": 77820
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.9557312726974487,
      "learning_rate": 0.0004463722281004801,
      "loss": 2.0361,
      "step": 77821
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.8238437175750732,
      "learning_rate": 0.00044636865745405614,
      "loss": 2.2387,
      "step": 77822
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.8102536201477051,
      "learning_rate": 0.00044636508678041966,
      "loss": 2.2117,
      "step": 77823
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.7974863052368164,
      "learning_rate": 0.00044636151607957105,
      "loss": 2.1448,
      "step": 77824
    },
    {
      "epoch": 1.01,
      "grad_norm": 1.0975266695022583,
      "learning_rate": 0.00044635794535151103,
      "loss": 2.1714,
      "step": 77825
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.8449552059173584,
      "learning_rate": 0.00044635437459624035,
      "loss": 2.2933,
      "step": 77826
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.8268528580665588,
      "learning_rate": 0.00044635080381375973,
      "loss": 2.0667,
      "step": 77827
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.8091192841529846,
      "learning_rate": 0.0004463472330040696,
      "loss": 2.2268,
      "step": 77828
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.9948039650917053,
      "learning_rate": 0.00044634366216717074,
      "loss": 2.2076,
      "step": 77829
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.8114117383956909,
      "learning_rate": 0.00044634009130306394,
      "loss": 2.1202,
      "step": 77830
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.7941775321960449,
      "learning_rate": 0.0004463365204117497,
      "loss": 2.1025,
      "step": 77831
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.979523241519928,
      "learning_rate": 0.00044633294949322875,
      "loss": 2.15,
      "step": 77832
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.7922772765159607,
      "learning_rate": 0.00044632937854750175,
      "loss": 2.3122,
      "step": 77833
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.7991780638694763,
      "learning_rate": 0.00044632580757456936,
      "loss": 2.3067,
      "step": 77834
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.7055813074111938,
      "learning_rate": 0.00044632223657443216,
      "loss": 2.2258,
      "step": 77835
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.8202432990074158,
      "learning_rate": 0.00044631866554709093,
      "loss": 1.949,
      "step": 77836
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.8529390692710876,
      "learning_rate": 0.0004463150944925464,
      "loss": 2.1357,
      "step": 77837
    },
    {
      "epoch": 1.01,
      "grad_norm": 1.0560895204544067,
      "learning_rate": 0.000446311523410799,
      "loss": 2.2666,
      "step": 77838
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.7812071442604065,
      "learning_rate": 0.0004463079523018496,
      "loss": 2.2752,
      "step": 77839
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.8261979222297668,
      "learning_rate": 0.0004463043811656988,
      "loss": 2.1755,
      "step": 77840
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.7682273387908936,
      "learning_rate": 0.00044630081000234716,
      "loss": 2.1187,
      "step": 77841
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.8611875176429749,
      "learning_rate": 0.0004462972388117955,
      "loss": 2.2816,
      "step": 77842
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.8667953610420227,
      "learning_rate": 0.0004462936675940445,
      "loss": 2.3553,
      "step": 77843
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.8254123330116272,
      "learning_rate": 0.0004462900963490946,
      "loss": 2.1779,
      "step": 77844
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.7775841355323792,
      "learning_rate": 0.00044628652507694666,
      "loss": 2.2249,
      "step": 77845
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.954077959060669,
      "learning_rate": 0.00044628295377760143,
      "loss": 1.9525,
      "step": 77846
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.6974257826805115,
      "learning_rate": 0.00044627938245105927,
      "loss": 2.054,
      "step": 77847
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.7656674385070801,
      "learning_rate": 0.000446275811097321,
      "loss": 2.1599,
      "step": 77848
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.7872822284698486,
      "learning_rate": 0.00044627223971638747,
      "loss": 2.2993,
      "step": 77849
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.7966007590293884,
      "learning_rate": 0.00044626866830825904,
      "loss": 2.2299,
      "step": 77850
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.8912137150764465,
      "learning_rate": 0.0004462650968729365,
      "loss": 2.2517,
      "step": 77851
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.7152033448219299,
      "learning_rate": 0.00044626152541042056,
      "loss": 1.9728,
      "step": 77852
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.8572883009910583,
      "learning_rate": 0.00044625795392071177,
      "loss": 2.4891,
      "step": 77853
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.7761955261230469,
      "learning_rate": 0.00044625438240381097,
      "loss": 2.3164,
      "step": 77854
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.7986831068992615,
      "learning_rate": 0.0004462508108597186,
      "loss": 2.0962,
      "step": 77855
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.9709250926971436,
      "learning_rate": 0.00044624723928843557,
      "loss": 2.0539,
      "step": 77856
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.8677182793617249,
      "learning_rate": 0.0004462436676899623,
      "loss": 2.2316,
      "step": 77857
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.9930226802825928,
      "learning_rate": 0.0004462400960642997,
      "loss": 2.2319,
      "step": 77858
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.8216573596000671,
      "learning_rate": 0.0004462365244114482,
      "loss": 2.1657,
      "step": 77859
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.7579348683357239,
      "learning_rate": 0.00044623295273140854,
      "loss": 2.2951,
      "step": 77860
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.794127881526947,
      "learning_rate": 0.0004462293810241816,
      "loss": 2.1881,
      "step": 77861
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.9264408349990845,
      "learning_rate": 0.00044622580928976765,
      "loss": 2.4713,
      "step": 77862
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.8109481930732727,
      "learning_rate": 0.0004462222375281677,
      "loss": 2.3327,
      "step": 77863
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.8264204859733582,
      "learning_rate": 0.0004462186657393822,
      "loss": 2.1369,
      "step": 77864
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.8709245324134827,
      "learning_rate": 0.00044621509392341193,
      "loss": 2.3411,
      "step": 77865
    },
    {
      "epoch": 1.01,
      "grad_norm": 1.2533191442489624,
      "learning_rate": 0.00044621152208025746,
      "loss": 2.2619,
      "step": 77866
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.8272877931594849,
      "learning_rate": 0.0004462079502099196,
      "loss": 2.3,
      "step": 77867
    },
    {
      "epoch": 1.01,
      "grad_norm": 1.042026400566101,
      "learning_rate": 0.00044620437831239884,
      "loss": 2.117,
      "step": 77868
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.8726567029953003,
      "learning_rate": 0.00044620080638769593,
      "loss": 2.2799,
      "step": 77869
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.7986563444137573,
      "learning_rate": 0.0004461972344358116,
      "loss": 2.2197,
      "step": 77870
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.7899184226989746,
      "learning_rate": 0.0004461936624567463,
      "loss": 2.0675,
      "step": 77871
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.8430776596069336,
      "learning_rate": 0.000446190090450501,
      "loss": 2.3235,
      "step": 77872
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.8330320119857788,
      "learning_rate": 0.00044618651841707614,
      "loss": 2.361,
      "step": 77873
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.8261486291885376,
      "learning_rate": 0.00044618294635647245,
      "loss": 2.2119,
      "step": 77874
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.7183997631072998,
      "learning_rate": 0.00044617937426869053,
      "loss": 2.1269,
      "step": 77875
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.8389331698417664,
      "learning_rate": 0.00044617580215373123,
      "loss": 2.2982,
      "step": 77876
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.8209124803543091,
      "learning_rate": 0.00044617223001159505,
      "loss": 2.1411,
      "step": 77877
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.7673896551132202,
      "learning_rate": 0.00044616865784228264,
      "loss": 1.9449,
      "step": 77878
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.8613869547843933,
      "learning_rate": 0.00044616508564579476,
      "loss": 2.1258,
      "step": 77879
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.8622219562530518,
      "learning_rate": 0.000446161513422132,
      "loss": 2.0224,
      "step": 77880
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.8094030022621155,
      "learning_rate": 0.000446157941171295,
      "loss": 2.3182,
      "step": 77881
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.7647892236709595,
      "learning_rate": 0.00044615436889328466,
      "loss": 2.1493,
      "step": 77882
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.8892121315002441,
      "learning_rate": 0.0004461507965881013,
      "loss": 2.1864,
      "step": 77883
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.810627818107605,
      "learning_rate": 0.00044614722425574574,
      "loss": 2.0259,
      "step": 77884
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.891911506652832,
      "learning_rate": 0.00044614365189621884,
      "loss": 2.1803,
      "step": 77885
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.789402186870575,
      "learning_rate": 0.0004461400795095209,
      "loss": 2.0463,
      "step": 77886
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.8442286849021912,
      "learning_rate": 0.00044613650709565277,
      "loss": 2.3448,
      "step": 77887
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.7555308938026428,
      "learning_rate": 0.00044613293465461517,
      "loss": 2.1951,
      "step": 77888
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.7682392597198486,
      "learning_rate": 0.00044612936218640866,
      "loss": 1.9783,
      "step": 77889
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.7798537015914917,
      "learning_rate": 0.00044612578969103393,
      "loss": 2.3025,
      "step": 77890
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.7849626541137695,
      "learning_rate": 0.0004461222171684917,
      "loss": 2.1108,
      "step": 77891
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.7424437403678894,
      "learning_rate": 0.0004461186446187826,
      "loss": 2.0981,
      "step": 77892
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.8252413272857666,
      "learning_rate": 0.00044611507204190724,
      "loss": 2.3635,
      "step": 77893
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.7950135469436646,
      "learning_rate": 0.0004461114994378664,
      "loss": 2.4137,
      "step": 77894
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.8908587694168091,
      "learning_rate": 0.0004461079268066606,
      "loss": 2.2553,
      "step": 77895
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.7806135416030884,
      "learning_rate": 0.00044610435414829056,
      "loss": 2.0868,
      "step": 77896
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.8953927755355835,
      "learning_rate": 0.0004461007814627571,
      "loss": 2.1733,
      "step": 77897
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.7944023013114929,
      "learning_rate": 0.0004460972087500606,
      "loss": 2.2396,
      "step": 77898
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.8504214286804199,
      "learning_rate": 0.0004460936360102019,
      "loss": 2.1473,
      "step": 77899
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.8035740852355957,
      "learning_rate": 0.0004460900632431817,
      "loss": 2.3249,
      "step": 77900
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.7799754738807678,
      "learning_rate": 0.0004460864904490006,
      "loss": 2.3488,
      "step": 77901
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.7908506989479065,
      "learning_rate": 0.0004460829176276592,
      "loss": 2.3237,
      "step": 77902
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.8377423882484436,
      "learning_rate": 0.00044607934477915824,
      "loss": 2.059,
      "step": 77903
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.7538667321205139,
      "learning_rate": 0.0004460757719034984,
      "loss": 2.253,
      "step": 77904
    },
    {
      "epoch": 1.01,
      "grad_norm": 1.1501950025558472,
      "learning_rate": 0.0004460721990006803,
      "loss": 2.2642,
      "step": 77905
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.7826024889945984,
      "learning_rate": 0.0004460686260707046,
      "loss": 2.0887,
      "step": 77906
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.8203147649765015,
      "learning_rate": 0.00044606505311357196,
      "loss": 2.2184,
      "step": 77907
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.7324986457824707,
      "learning_rate": 0.00044606148012928316,
      "loss": 2.3131,
      "step": 77908
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.8514572978019714,
      "learning_rate": 0.00044605790711783866,
      "loss": 1.962,
      "step": 77909
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.8043680191040039,
      "learning_rate": 0.0004460543340792393,
      "loss": 2.3332,
      "step": 77910
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.9250400066375732,
      "learning_rate": 0.0004460507610134857,
      "loss": 2.1174,
      "step": 77911
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.8179799914360046,
      "learning_rate": 0.00044604718792057855,
      "loss": 2.2874,
      "step": 77912
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.7559690475463867,
      "learning_rate": 0.00044604361480051835,
      "loss": 2.101,
      "step": 77913
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.759881317615509,
      "learning_rate": 0.000446040041653306,
      "loss": 2.2463,
      "step": 77914
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.8542513847351074,
      "learning_rate": 0.000446036468478942,
      "loss": 2.1629,
      "step": 77915
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.7782866954803467,
      "learning_rate": 0.0004460328952774271,
      "loss": 2.2882,
      "step": 77916
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.815980851650238,
      "learning_rate": 0.00044602932204876187,
      "loss": 2.2432,
      "step": 77917
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.8171891570091248,
      "learning_rate": 0.00044602574879294704,
      "loss": 2.2226,
      "step": 77918
    },
    {
      "epoch": 1.01,
      "grad_norm": 1.0895905494689941,
      "learning_rate": 0.0004460221755099833,
      "loss": 2.1119,
      "step": 77919
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.8219428658485413,
      "learning_rate": 0.0004460186021998712,
      "loss": 2.1137,
      "step": 77920
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.8260408043861389,
      "learning_rate": 0.0004460150288626116,
      "loss": 2.0468,
      "step": 77921
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.8714221715927124,
      "learning_rate": 0.00044601145549820503,
      "loss": 2.2122,
      "step": 77922
    },
    {
      "epoch": 1.01,
      "grad_norm": 1.4552663564682007,
      "learning_rate": 0.0004460078821066521,
      "loss": 2.2616,
      "step": 77923
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.7706044912338257,
      "learning_rate": 0.00044600430868795357,
      "loss": 2.2724,
      "step": 77924
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.7955868244171143,
      "learning_rate": 0.00044600073524211014,
      "loss": 2.0654,
      "step": 77925
    },
    {
      "epoch": 1.01,
      "grad_norm": 1.0664072036743164,
      "learning_rate": 0.00044599716176912235,
      "loss": 2.0876,
      "step": 77926
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.9466295838356018,
      "learning_rate": 0.0004459935882689909,
      "loss": 2.3452,
      "step": 77927
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.8006799817085266,
      "learning_rate": 0.00044599001474171664,
      "loss": 2.0149,
      "step": 77928
    },
    {
      "epoch": 1.01,
      "grad_norm": 1.0819449424743652,
      "learning_rate": 0.0004459864411873,
      "loss": 2.224,
      "step": 77929
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.9950194358825684,
      "learning_rate": 0.0004459828676057417,
      "loss": 2.2924,
      "step": 77930
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.8793995380401611,
      "learning_rate": 0.0004459792939970425,
      "loss": 1.9569,
      "step": 77931
    },
    {
      "epoch": 1.01,
      "grad_norm": 1.0054186582565308,
      "learning_rate": 0.0004459757203612029,
      "loss": 2.0002,
      "step": 77932
    },
    {
      "epoch": 1.01,
      "grad_norm": 1.017357349395752,
      "learning_rate": 0.00044597214669822366,
      "loss": 2.1599,
      "step": 77933
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.8328123688697815,
      "learning_rate": 0.0004459685730081056,
      "loss": 2.1733,
      "step": 77934
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.7861084938049316,
      "learning_rate": 0.00044596499929084904,
      "loss": 2.3427,
      "step": 77935
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.8073241114616394,
      "learning_rate": 0.0004459614255464549,
      "loss": 2.1184,
      "step": 77936
    },
    {
      "epoch": 1.01,
      "grad_norm": 1.5012909173965454,
      "learning_rate": 0.0004459578517749238,
      "loss": 2.2391,
      "step": 77937
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.9123454689979553,
      "learning_rate": 0.0004459542779762563,
      "loss": 2.2033,
      "step": 77938
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.8392666578292847,
      "learning_rate": 0.0004459507041504532,
      "loss": 2.4982,
      "step": 77939
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.9666814804077148,
      "learning_rate": 0.00044594713029751516,
      "loss": 2.1113,
      "step": 77940
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.776961624622345,
      "learning_rate": 0.0004459435564174427,
      "loss": 2.0766,
      "step": 77941
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.8495091795921326,
      "learning_rate": 0.00044593998251023665,
      "loss": 2.2868,
      "step": 77942
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.7377758622169495,
      "learning_rate": 0.0004459364085758976,
      "loss": 2.2301,
      "step": 77943
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.7642956376075745,
      "learning_rate": 0.0004459328346144262,
      "loss": 2.0934,
      "step": 77944
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.8673779368400574,
      "learning_rate": 0.0004459292606258231,
      "loss": 2.2289,
      "step": 77945
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.7768963575363159,
      "learning_rate": 0.00044592568661008905,
      "loss": 2.0287,
      "step": 77946
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.7804275751113892,
      "learning_rate": 0.00044592211256722455,
      "loss": 2.3167,
      "step": 77947
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.8449456095695496,
      "learning_rate": 0.0004459185384972305,
      "loss": 2.2268,
      "step": 77948
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.8197237849235535,
      "learning_rate": 0.00044591496440010735,
      "loss": 2.4621,
      "step": 77949
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.8202024698257446,
      "learning_rate": 0.0004459113902758559,
      "loss": 2.248,
      "step": 77950
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.8145365118980408,
      "learning_rate": 0.0004459078161244767,
      "loss": 2.1449,
      "step": 77951
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.7905411124229431,
      "learning_rate": 0.00044590424194597064,
      "loss": 2.319,
      "step": 77952
    },
    {
      "epoch": 1.01,
      "grad_norm": 0.7463365197181702,
      "learning_rate": 0.0004459006677403381,
      "loss": 1.9023,
      "step": 77953
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.8196854591369629,
      "learning_rate": 0.0004458970935075799,
      "loss": 2.001,
      "step": 77954
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.8213902115821838,
      "learning_rate": 0.00044589351924769675,
      "loss": 2.2189,
      "step": 77955
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.7652170062065125,
      "learning_rate": 0.0004458899449606891,
      "loss": 2.0688,
      "step": 77956
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.8051605224609375,
      "learning_rate": 0.0004458863706465578,
      "loss": 2.2072,
      "step": 77957
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.7796052098274231,
      "learning_rate": 0.0004458827963053036,
      "loss": 2.1446,
      "step": 77958
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.7919629812240601,
      "learning_rate": 0.00044587922193692684,
      "loss": 2.1743,
      "step": 77959
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.8928316831588745,
      "learning_rate": 0.00044587564754142845,
      "loss": 2.2429,
      "step": 77960
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.9335440397262573,
      "learning_rate": 0.0004458720731188091,
      "loss": 2.1028,
      "step": 77961
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.8269825577735901,
      "learning_rate": 0.0004458684986690693,
      "loss": 2.2589,
      "step": 77962
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.8421128392219543,
      "learning_rate": 0.0004458649241922098,
      "loss": 2.1828,
      "step": 77963
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.8082072138786316,
      "learning_rate": 0.0004458613496882313,
      "loss": 2.282,
      "step": 77964
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.8165522217750549,
      "learning_rate": 0.0004458577751571344,
      "loss": 2.3052,
      "step": 77965
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.8532772660255432,
      "learning_rate": 0.0004458542005989198,
      "loss": 2.1175,
      "step": 77966
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.8973935842514038,
      "learning_rate": 0.0004458506260135881,
      "loss": 2.236,
      "step": 77967
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.7992055416107178,
      "learning_rate": 0.0004458470514011401,
      "loss": 2.3737,
      "step": 77968
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.7721750140190125,
      "learning_rate": 0.00044584347676157626,
      "loss": 2.1995,
      "step": 77969
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.8559137582778931,
      "learning_rate": 0.00044583990209489744,
      "loss": 2.0756,
      "step": 77970
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.7685227990150452,
      "learning_rate": 0.00044583632740110424,
      "loss": 2.1355,
      "step": 77971
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.8078351020812988,
      "learning_rate": 0.00044583275268019726,
      "loss": 2.2206,
      "step": 77972
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.716632068157196,
      "learning_rate": 0.00044582917793217726,
      "loss": 2.1281,
      "step": 77973
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.7638658881187439,
      "learning_rate": 0.0004458256031570449,
      "loss": 2.1965,
      "step": 77974
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.820273756980896,
      "learning_rate": 0.00044582202835480073,
      "loss": 2.1152,
      "step": 77975
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.8089037537574768,
      "learning_rate": 0.00044581845352544545,
      "loss": 2.3786,
      "step": 77976
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.7534275054931641,
      "learning_rate": 0.00044581487866897997,
      "loss": 1.9066,
      "step": 77977
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.8769180774688721,
      "learning_rate": 0.0004458113037854046,
      "loss": 2.332,
      "step": 77978
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.7844854593276978,
      "learning_rate": 0.0004458077288747201,
      "loss": 2.2356,
      "step": 77979
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.749416708946228,
      "learning_rate": 0.00044580415393692734,
      "loss": 2.0003,
      "step": 77980
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.8331201672554016,
      "learning_rate": 0.00044580057897202673,
      "loss": 2.4951,
      "step": 77981
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.8821511268615723,
      "learning_rate": 0.0004457970039800191,
      "loss": 2.2395,
      "step": 77982
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.7990335822105408,
      "learning_rate": 0.00044579342896090504,
      "loss": 2.1475,
      "step": 77983
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.7484428882598877,
      "learning_rate": 0.0004457898539146852,
      "loss": 2.0792,
      "step": 77984
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.8223550915718079,
      "learning_rate": 0.00044578627884136026,
      "loss": 2.1718,
      "step": 77985
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.7805885076522827,
      "learning_rate": 0.000445782703740931,
      "loss": 1.9799,
      "step": 77986
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.8480431437492371,
      "learning_rate": 0.00044577912861339786,
      "loss": 2.4634,
      "step": 77987
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.8043550848960876,
      "learning_rate": 0.00044577555345876176,
      "loss": 2.2925,
      "step": 77988
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.7901284694671631,
      "learning_rate": 0.00044577197827702316,
      "loss": 2.1827,
      "step": 77989
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.8481377363204956,
      "learning_rate": 0.00044576840306818276,
      "loss": 2.2765,
      "step": 77990
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.8041773438453674,
      "learning_rate": 0.00044576482783224126,
      "loss": 2.2375,
      "step": 77991
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.7798256874084473,
      "learning_rate": 0.00044576125256919936,
      "loss": 2.2035,
      "step": 77992
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.7433537840843201,
      "learning_rate": 0.0004457576772790578,
      "loss": 2.0644,
      "step": 77993
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.817774772644043,
      "learning_rate": 0.000445754101961817,
      "loss": 2.0829,
      "step": 77994
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.866574227809906,
      "learning_rate": 0.0004457505266174777,
      "loss": 2.3064,
      "step": 77995
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.7702414989471436,
      "learning_rate": 0.0004457469512460408,
      "loss": 2.2658,
      "step": 77996
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.744552731513977,
      "learning_rate": 0.0004457433758475067,
      "loss": 2.2336,
      "step": 77997
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.7569363713264465,
      "learning_rate": 0.00044573980042187614,
      "loss": 2.0634,
      "step": 77998
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.7508826851844788,
      "learning_rate": 0.0004457362249691498,
      "loss": 2.1323,
      "step": 77999
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.8334710001945496,
      "learning_rate": 0.00044573264948932843,
      "loss": 2.0843,
      "step": 78000
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.7360894680023193,
      "learning_rate": 0.00044572907398241255,
      "loss": 2.3697,
      "step": 78001
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.866584300994873,
      "learning_rate": 0.00044572549844840284,
      "loss": 2.0934,
      "step": 78002
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.8043105602264404,
      "learning_rate": 0.0004457219228873001,
      "loss": 2.0957,
      "step": 78003
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.7384320497512817,
      "learning_rate": 0.00044571834729910484,
      "loss": 2.092,
      "step": 78004
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.9839629530906677,
      "learning_rate": 0.00044571477168381775,
      "loss": 2.183,
      "step": 78005
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.8262590765953064,
      "learning_rate": 0.0004457111960414396,
      "loss": 2.3109,
      "step": 78006
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.765132486820221,
      "learning_rate": 0.000445707620371971,
      "loss": 2.3009,
      "step": 78007
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.8107720017433167,
      "learning_rate": 0.00044570404467541257,
      "loss": 2.1936,
      "step": 78008
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.7913392782211304,
      "learning_rate": 0.000445700468951765,
      "loss": 2.1886,
      "step": 78009
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.7776126861572266,
      "learning_rate": 0.000445696893201029,
      "loss": 1.9614,
      "step": 78010
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.8721929788589478,
      "learning_rate": 0.00044569331742320517,
      "loss": 2.1462,
      "step": 78011
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.7591230869293213,
      "learning_rate": 0.0004456897416182942,
      "loss": 2.1476,
      "step": 78012
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.832994818687439,
      "learning_rate": 0.00044568616578629684,
      "loss": 2.4427,
      "step": 78013
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.8213256597518921,
      "learning_rate": 0.0004456825899272135,
      "loss": 2.1204,
      "step": 78014
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.9836357831954956,
      "learning_rate": 0.00044567901404104517,
      "loss": 2.219,
      "step": 78015
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.7087617516517639,
      "learning_rate": 0.0004456754381277923,
      "loss": 2.0998,
      "step": 78016
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.8144987225532532,
      "learning_rate": 0.0004456718621874557,
      "loss": 2.1202,
      "step": 78017
    },
    {
      "epoch": 1.02,
      "grad_norm": 1.2274067401885986,
      "learning_rate": 0.00044566828622003577,
      "loss": 2.4169,
      "step": 78018
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.8202109932899475,
      "learning_rate": 0.0004456647102255335,
      "loss": 2.2653,
      "step": 78019
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.7957479953765869,
      "learning_rate": 0.0004456611342039494,
      "loss": 2.3603,
      "step": 78020
    },
    {
      "epoch": 1.02,
      "grad_norm": 1.0241163969039917,
      "learning_rate": 0.00044565755815528406,
      "loss": 2.2581,
      "step": 78021
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.9525859355926514,
      "learning_rate": 0.00044565398207953836,
      "loss": 2.044,
      "step": 78022
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.8089513778686523,
      "learning_rate": 0.00044565040597671277,
      "loss": 2.1788,
      "step": 78023
    },
    {
      "epoch": 1.02,
      "grad_norm": 1.0025750398635864,
      "learning_rate": 0.000445646829846808,
      "loss": 2.1427,
      "step": 78024
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.8759856820106506,
      "learning_rate": 0.00044564325368982475,
      "loss": 2.011,
      "step": 78025
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.8068221211433411,
      "learning_rate": 0.0004456396775057637,
      "loss": 2.1066,
      "step": 78026
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.8359267115592957,
      "learning_rate": 0.00044563610129462537,
      "loss": 2.0423,
      "step": 78027
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.8025457262992859,
      "learning_rate": 0.0004456325250564107,
      "loss": 2.3338,
      "step": 78028
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.9370108842849731,
      "learning_rate": 0.00044562894879112005,
      "loss": 2.3249,
      "step": 78029
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.8073177337646484,
      "learning_rate": 0.00044562537249875434,
      "loss": 2.198,
      "step": 78030
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.8893195390701294,
      "learning_rate": 0.00044562179617931404,
      "loss": 2.4022,
      "step": 78031
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.7890843749046326,
      "learning_rate": 0.0004456182198327999,
      "loss": 2.2466,
      "step": 78032
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.7616434693336487,
      "learning_rate": 0.00044561464345921266,
      "loss": 2.2387,
      "step": 78033
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.7823895215988159,
      "learning_rate": 0.0004456110670585529,
      "loss": 2.2695,
      "step": 78034
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.7766659259796143,
      "learning_rate": 0.0004456074906308213,
      "loss": 2.1539,
      "step": 78035
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.8435434103012085,
      "learning_rate": 0.0004456039141760184,
      "loss": 1.933,
      "step": 78036
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.7733216881752014,
      "learning_rate": 0.0004456003376941451,
      "loss": 2.2109,
      "step": 78037
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.7495347261428833,
      "learning_rate": 0.000445596761185202,
      "loss": 2.2193,
      "step": 78038
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.7905503511428833,
      "learning_rate": 0.00044559318464918953,
      "loss": 2.1265,
      "step": 78039
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.7284918427467346,
      "learning_rate": 0.0004455896080861086,
      "loss": 2.246,
      "step": 78040
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.7740334272384644,
      "learning_rate": 0.0004455860314959599,
      "loss": 2.0958,
      "step": 78041
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.8260069489479065,
      "learning_rate": 0.000445582454878744,
      "loss": 2.365,
      "step": 78042
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.8202111721038818,
      "learning_rate": 0.00044557887823446147,
      "loss": 2.0028,
      "step": 78043
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.8040511012077332,
      "learning_rate": 0.0004455753015631132,
      "loss": 2.2053,
      "step": 78044
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.7791074514389038,
      "learning_rate": 0.00044557172486469964,
      "loss": 2.2909,
      "step": 78045
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.751876711845398,
      "learning_rate": 0.00044556814813922157,
      "loss": 2.1774,
      "step": 78046
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.8012798428535461,
      "learning_rate": 0.0004455645713866797,
      "loss": 2.2367,
      "step": 78047
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.8226023316383362,
      "learning_rate": 0.0004455609946070746,
      "loss": 2.1533,
      "step": 78048
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.730330228805542,
      "learning_rate": 0.0004455574178004069,
      "loss": 2.2131,
      "step": 78049
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.7182220220565796,
      "learning_rate": 0.00044555384096667743,
      "loss": 2.191,
      "step": 78050
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.8791496753692627,
      "learning_rate": 0.0004455502641058867,
      "loss": 2.5023,
      "step": 78051
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.7142228484153748,
      "learning_rate": 0.00044554668721803536,
      "loss": 2.1852,
      "step": 78052
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.7999119162559509,
      "learning_rate": 0.00044554311030312423,
      "loss": 2.1429,
      "step": 78053
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.801129937171936,
      "learning_rate": 0.0004455395333611539,
      "loss": 2.1415,
      "step": 78054
    },
    {
      "epoch": 1.02,
      "grad_norm": 1.0782357454299927,
      "learning_rate": 0.00044553595639212494,
      "loss": 2.2263,
      "step": 78055
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.7004554867744446,
      "learning_rate": 0.0004455323793960382,
      "loss": 2.3085,
      "step": 78056
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.8593107461929321,
      "learning_rate": 0.00044552880237289414,
      "loss": 2.293,
      "step": 78057
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.7628403306007385,
      "learning_rate": 0.0004455252253226936,
      "loss": 2.0968,
      "step": 78058
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.828710675239563,
      "learning_rate": 0.00044552164824543714,
      "loss": 2.0709,
      "step": 78059
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.7533267736434937,
      "learning_rate": 0.0004455180711411255,
      "loss": 2.1477,
      "step": 78060
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.9040451645851135,
      "learning_rate": 0.00044551449400975925,
      "loss": 2.2143,
      "step": 78061
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.7285488247871399,
      "learning_rate": 0.0004455109168513392,
      "loss": 2.1806,
      "step": 78062
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.828940749168396,
      "learning_rate": 0.0004455073396658658,
      "loss": 2.3364,
      "step": 78063
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.8483148813247681,
      "learning_rate": 0.00044550376245333986,
      "loss": 2.3666,
      "step": 78064
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.7881690263748169,
      "learning_rate": 0.0004455001852137621,
      "loss": 2.3934,
      "step": 78065
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.830281674861908,
      "learning_rate": 0.00044549660794713304,
      "loss": 1.8826,
      "step": 78066
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.7791978716850281,
      "learning_rate": 0.00044549303065345345,
      "loss": 2.2449,
      "step": 78067
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.7503972053527832,
      "learning_rate": 0.00044548945333272404,
      "loss": 2.0912,
      "step": 78068
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.7263209819793701,
      "learning_rate": 0.0004454858759849452,
      "loss": 2.0176,
      "step": 78069
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.72696453332901,
      "learning_rate": 0.0004454822986101179,
      "loss": 2.2619,
      "step": 78070
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.7032405734062195,
      "learning_rate": 0.0004454787212082428,
      "loss": 2.2021,
      "step": 78071
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.77887362241745,
      "learning_rate": 0.0004454751437793203,
      "loss": 2.2562,
      "step": 78072
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.8347737193107605,
      "learning_rate": 0.0004454715663233512,
      "loss": 2.1597,
      "step": 78073
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.8843792080879211,
      "learning_rate": 0.00044546798884033634,
      "loss": 2.3431,
      "step": 78074
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.7744360566139221,
      "learning_rate": 0.0004454644113302761,
      "loss": 1.9732,
      "step": 78075
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.8628469705581665,
      "learning_rate": 0.00044546083379317134,
      "loss": 2.2576,
      "step": 78076
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.8928132057189941,
      "learning_rate": 0.0004454572562290227,
      "loss": 2.2151,
      "step": 78077
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.7804067730903625,
      "learning_rate": 0.0004454536786378307,
      "loss": 2.1079,
      "step": 78078
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.8579062819480896,
      "learning_rate": 0.0004454501010195963,
      "loss": 2.3346,
      "step": 78079
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.8448967933654785,
      "learning_rate": 0.0004454465233743198,
      "loss": 2.3382,
      "step": 78080
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.8086347579956055,
      "learning_rate": 0.00044544294570200216,
      "loss": 2.1443,
      "step": 78081
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.8585655689239502,
      "learning_rate": 0.0004454393680026439,
      "loss": 1.9675,
      "step": 78082
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.8561687469482422,
      "learning_rate": 0.0004454357902762457,
      "loss": 2.2334,
      "step": 78083
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.8816055059432983,
      "learning_rate": 0.0004454322125228082,
      "loss": 2.3387,
      "step": 78084
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.7457638382911682,
      "learning_rate": 0.00044542863474233215,
      "loss": 2.277,
      "step": 78085
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.9612133502960205,
      "learning_rate": 0.00044542505693481813,
      "loss": 2.4245,
      "step": 78086
    },
    {
      "epoch": 1.02,
      "grad_norm": 1.1289987564086914,
      "learning_rate": 0.00044542147910026694,
      "loss": 2.065,
      "step": 78087
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.7602832913398743,
      "learning_rate": 0.00044541790123867915,
      "loss": 2.276,
      "step": 78088
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.9153386354446411,
      "learning_rate": 0.0004454143233500554,
      "loss": 2.2102,
      "step": 78089
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.8087641596794128,
      "learning_rate": 0.00044541074543439626,
      "loss": 2.0867,
      "step": 78090
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.8486419916152954,
      "learning_rate": 0.0004454071674917026,
      "loss": 2.1724,
      "step": 78091
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.7763528823852539,
      "learning_rate": 0.00044540358952197505,
      "loss": 2.2308,
      "step": 78092
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.7541797757148743,
      "learning_rate": 0.0004454000115252142,
      "loss": 2.1565,
      "step": 78093
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.9670670628547668,
      "learning_rate": 0.0004453964335014207,
      "loss": 2.3133,
      "step": 78094
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.8566127419471741,
      "learning_rate": 0.00044539285545059536,
      "loss": 2.2016,
      "step": 78095
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.8333330154418945,
      "learning_rate": 0.0004453892773727387,
      "loss": 2.2928,
      "step": 78096
    },
    {
      "epoch": 1.02,
      "grad_norm": 1.0007482767105103,
      "learning_rate": 0.0004453856992678514,
      "loss": 2.1068,
      "step": 78097
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.7596032023429871,
      "learning_rate": 0.00044538212113593415,
      "loss": 2.1099,
      "step": 78098
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.8540906310081482,
      "learning_rate": 0.0004453785429769877,
      "loss": 2.0695,
      "step": 78099
    },
    {
      "epoch": 1.02,
      "grad_norm": 1.255920648574829,
      "learning_rate": 0.0004453749647910125,
      "loss": 2.2102,
      "step": 78100
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.9026926755905151,
      "learning_rate": 0.0004453713865780095,
      "loss": 2.299,
      "step": 78101
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.8458276987075806,
      "learning_rate": 0.0004453678083379791,
      "loss": 2.3235,
      "step": 78102
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.9302418231964111,
      "learning_rate": 0.0004453642300709221,
      "loss": 2.1849,
      "step": 78103
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.9294460415840149,
      "learning_rate": 0.00044536065177683916,
      "loss": 2.1285,
      "step": 78104
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.8582149744033813,
      "learning_rate": 0.00044535707345573093,
      "loss": 2.2501,
      "step": 78105
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.7909330129623413,
      "learning_rate": 0.0004453534951075981,
      "loss": 2.1622,
      "step": 78106
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.8015545010566711,
      "learning_rate": 0.0004453499167324413,
      "loss": 2.297,
      "step": 78107
    },
    {
      "epoch": 1.02,
      "grad_norm": 1.1428704261779785,
      "learning_rate": 0.00044534633833026127,
      "loss": 2.207,
      "step": 78108
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.9652160406112671,
      "learning_rate": 0.0004453427599010584,
      "loss": 2.1687,
      "step": 78109
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.7942095398902893,
      "learning_rate": 0.00044533918144483377,
      "loss": 2.3816,
      "step": 78110
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.9686981439590454,
      "learning_rate": 0.00044533560296158777,
      "loss": 2.3336,
      "step": 78111
    },
    {
      "epoch": 1.02,
      "grad_norm": 1.1369634866714478,
      "learning_rate": 0.0004453320244513212,
      "loss": 2.2516,
      "step": 78112
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.7749650478363037,
      "learning_rate": 0.0004453284459140346,
      "loss": 2.1695,
      "step": 78113
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.7595789432525635,
      "learning_rate": 0.0004453248673497287,
      "loss": 2.2409,
      "step": 78114
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.8552144765853882,
      "learning_rate": 0.0004453212887584042,
      "loss": 2.1225,
      "step": 78115
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.8285489082336426,
      "learning_rate": 0.0004453177101400617,
      "loss": 2.3992,
      "step": 78116
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.8785980939865112,
      "learning_rate": 0.0004453141314947019,
      "loss": 2.106,
      "step": 78117
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.8965138792991638,
      "learning_rate": 0.00044531055282232544,
      "loss": 2.4828,
      "step": 78118
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.832516610622406,
      "learning_rate": 0.00044530697412293295,
      "loss": 1.9957,
      "step": 78119
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.9211006760597229,
      "learning_rate": 0.00044530339539652534,
      "loss": 2.2357,
      "step": 78120
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.8541208505630493,
      "learning_rate": 0.00044529981664310293,
      "loss": 2.3501,
      "step": 78121
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.819526195526123,
      "learning_rate": 0.00044529623786266655,
      "loss": 2.2422,
      "step": 78122
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.8542764186859131,
      "learning_rate": 0.00044529265905521694,
      "loss": 2.3377,
      "step": 78123
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.932079017162323,
      "learning_rate": 0.0004452890802207546,
      "loss": 2.1991,
      "step": 78124
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.8976150751113892,
      "learning_rate": 0.00044528550135928033,
      "loss": 2.0315,
      "step": 78125
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.7726906538009644,
      "learning_rate": 0.00044528192247079475,
      "loss": 2.1024,
      "step": 78126
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.9082221388816833,
      "learning_rate": 0.0004452783435552985,
      "loss": 2.234,
      "step": 78127
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.7962837815284729,
      "learning_rate": 0.0004452747646127922,
      "loss": 2.2998,
      "step": 78128
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.9718211889266968,
      "learning_rate": 0.00044527118564327664,
      "loss": 2.1057,
      "step": 78129
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.9946171641349792,
      "learning_rate": 0.0004452676066467524,
      "loss": 2.161,
      "step": 78130
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.9068312048912048,
      "learning_rate": 0.00044526402762322024,
      "loss": 2.1085,
      "step": 78131
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.7488016486167908,
      "learning_rate": 0.0004452604485726807,
      "loss": 2.2588,
      "step": 78132
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.8976732492446899,
      "learning_rate": 0.00044525686949513447,
      "loss": 2.2916,
      "step": 78133
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.7695606350898743,
      "learning_rate": 0.0004452532903905823,
      "loss": 2.2846,
      "step": 78134
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.7814476490020752,
      "learning_rate": 0.00044524971125902476,
      "loss": 2.3241,
      "step": 78135
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.8775182962417603,
      "learning_rate": 0.00044524613210046257,
      "loss": 2.1843,
      "step": 78136
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.7755380272865295,
      "learning_rate": 0.00044524255291489637,
      "loss": 2.056,
      "step": 78137
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.7881662845611572,
      "learning_rate": 0.0004452389737023269,
      "loss": 2.1068,
      "step": 78138
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.8378065228462219,
      "learning_rate": 0.0004452353944627547,
      "loss": 2.3149,
      "step": 78139
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.8660951256752014,
      "learning_rate": 0.0004452318151961805,
      "loss": 2.2118,
      "step": 78140
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.7909934520721436,
      "learning_rate": 0.00044522823590260505,
      "loss": 2.1949,
      "step": 78141
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.7615509629249573,
      "learning_rate": 0.00044522465658202885,
      "loss": 2.4048,
      "step": 78142
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.7700679898262024,
      "learning_rate": 0.0004452210772344526,
      "loss": 2.3159,
      "step": 78143
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.7758833765983582,
      "learning_rate": 0.0004452174978598771,
      "loss": 2.2641,
      "step": 78144
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.8086461424827576,
      "learning_rate": 0.0004452139184583029,
      "loss": 2.4564,
      "step": 78145
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.7908608913421631,
      "learning_rate": 0.00044521033902973064,
      "loss": 2.3326,
      "step": 78146
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.7724794149398804,
      "learning_rate": 0.0004452067595741612,
      "loss": 2.0112,
      "step": 78147
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.83213871717453,
      "learning_rate": 0.0004452031800915949,
      "loss": 2.1375,
      "step": 78148
    },
    {
      "epoch": 1.02,
      "grad_norm": 1.0093286037445068,
      "learning_rate": 0.0004451996005820326,
      "loss": 2.1083,
      "step": 78149
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.7758976817131042,
      "learning_rate": 0.00044519602104547507,
      "loss": 1.9962,
      "step": 78150
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.792883574962616,
      "learning_rate": 0.0004451924414819228,
      "loss": 2.0558,
      "step": 78151
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.7516169548034668,
      "learning_rate": 0.0004451888618913765,
      "loss": 2.071,
      "step": 78152
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.8815591931343079,
      "learning_rate": 0.0004451852822738368,
      "loss": 2.1897,
      "step": 78153
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.7600426077842712,
      "learning_rate": 0.0004451817026293045,
      "loss": 2.2484,
      "step": 78154
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.7415859699249268,
      "learning_rate": 0.00044517812295778015,
      "loss": 2.1052,
      "step": 78155
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.8179553151130676,
      "learning_rate": 0.0004451745432592645,
      "loss": 2.232,
      "step": 78156
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.8150562047958374,
      "learning_rate": 0.0004451709635337581,
      "loss": 2.2062,
      "step": 78157
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.8620337247848511,
      "learning_rate": 0.0004451673837812617,
      "loss": 2.5712,
      "step": 78158
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.8773792386054993,
      "learning_rate": 0.00044516380400177596,
      "loss": 2.5024,
      "step": 78159
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.7682722210884094,
      "learning_rate": 0.0004451602241953015,
      "loss": 2.2556,
      "step": 78160
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.7767663598060608,
      "learning_rate": 0.00044515664436183894,
      "loss": 2.1837,
      "step": 78161
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.8210179209709167,
      "learning_rate": 0.00044515306450138915,
      "loss": 2.0555,
      "step": 78162
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.8354625701904297,
      "learning_rate": 0.0004451494846139526,
      "loss": 2.4103,
      "step": 78163
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.8997772336006165,
      "learning_rate": 0.00044514590469953,
      "loss": 2.2697,
      "step": 78164
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.8440901041030884,
      "learning_rate": 0.00044514232475812213,
      "loss": 2.1029,
      "step": 78165
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.7646767497062683,
      "learning_rate": 0.0004451387447897295,
      "loss": 2.2665,
      "step": 78166
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.7866397500038147,
      "learning_rate": 0.00044513516479435285,
      "loss": 1.9705,
      "step": 78167
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.7492620944976807,
      "learning_rate": 0.00044513158477199274,
      "loss": 2.0071,
      "step": 78168
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.8190648555755615,
      "learning_rate": 0.0004451280047226501,
      "loss": 2.0851,
      "step": 78169
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.8092839121818542,
      "learning_rate": 0.0004451244246463253,
      "loss": 2.4275,
      "step": 78170
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.8448978066444397,
      "learning_rate": 0.00044512084454301914,
      "loss": 2.2805,
      "step": 78171
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.8726667165756226,
      "learning_rate": 0.0004451172644127323,
      "loss": 2.2281,
      "step": 78172
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.8204910159111023,
      "learning_rate": 0.0004451136842554654,
      "loss": 2.3248,
      "step": 78173
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.9911516904830933,
      "learning_rate": 0.0004451101040712191,
      "loss": 2.0187,
      "step": 78174
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.7504535913467407,
      "learning_rate": 0.0004451065238599942,
      "loss": 2.2699,
      "step": 78175
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.7534414529800415,
      "learning_rate": 0.00044510294362179123,
      "loss": 2.2534,
      "step": 78176
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.9616159796714783,
      "learning_rate": 0.0004450993633566108,
      "loss": 2.2752,
      "step": 78177
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.8083924651145935,
      "learning_rate": 0.00044509578306445373,
      "loss": 2.1137,
      "step": 78178
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.7204369902610779,
      "learning_rate": 0.0004450922027453206,
      "loss": 2.1859,
      "step": 78179
    },
    {
      "epoch": 1.02,
      "grad_norm": 1.1518477201461792,
      "learning_rate": 0.00044508862239921214,
      "loss": 2.1038,
      "step": 78180
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.8125067353248596,
      "learning_rate": 0.00044508504202612885,
      "loss": 2.1206,
      "step": 78181
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.7333747148513794,
      "learning_rate": 0.00044508146162607163,
      "loss": 2.3781,
      "step": 78182
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.8421012759208679,
      "learning_rate": 0.00044507788119904087,
      "loss": 1.9901,
      "step": 78183
    },
    {
      "epoch": 1.02,
      "grad_norm": 1.071846604347229,
      "learning_rate": 0.0004450743007450376,
      "loss": 2.2304,
      "step": 78184
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.7853860259056091,
      "learning_rate": 0.00044507072026406214,
      "loss": 2.1824,
      "step": 78185
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.7670754194259644,
      "learning_rate": 0.0004450671397561153,
      "loss": 2.2603,
      "step": 78186
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.8173693418502808,
      "learning_rate": 0.0004450635592211978,
      "loss": 1.9917,
      "step": 78187
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.763110876083374,
      "learning_rate": 0.0004450599786593102,
      "loss": 2.0401,
      "step": 78188
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.7955100536346436,
      "learning_rate": 0.0004450563980704532,
      "loss": 2.0773,
      "step": 78189
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.7128098607063293,
      "learning_rate": 0.00044505281745462757,
      "loss": 1.949,
      "step": 78190
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.7690690755844116,
      "learning_rate": 0.0004450492368118338,
      "loss": 2.4261,
      "step": 78191
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.793316662311554,
      "learning_rate": 0.0004450456561420727,
      "loss": 2.1636,
      "step": 78192
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.7630197405815125,
      "learning_rate": 0.00044504207544534484,
      "loss": 2.1829,
      "step": 78193
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.824698269367218,
      "learning_rate": 0.0004450384947216509,
      "loss": 2.5318,
      "step": 78194
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.8054556250572205,
      "learning_rate": 0.0004450349139709916,
      "loss": 2.1772,
      "step": 78195
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.7177987098693848,
      "learning_rate": 0.00044503133319336757,
      "loss": 2.2911,
      "step": 78196
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.8726080656051636,
      "learning_rate": 0.00044502775238877945,
      "loss": 2.1533,
      "step": 78197
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.7429004907608032,
      "learning_rate": 0.00044502417155722787,
      "loss": 2.0513,
      "step": 78198
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.8122274279594421,
      "learning_rate": 0.00044502059069871365,
      "loss": 2.2957,
      "step": 78199
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.7949536442756653,
      "learning_rate": 0.0004450170098132374,
      "loss": 2.2403,
      "step": 78200
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.8502854704856873,
      "learning_rate": 0.0004450134289007996,
      "loss": 2.0667,
      "step": 78201
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.8769930601119995,
      "learning_rate": 0.00044500984796140124,
      "loss": 2.3068,
      "step": 78202
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.7387374043464661,
      "learning_rate": 0.00044500626699504273,
      "loss": 2.0812,
      "step": 78203
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.8341890573501587,
      "learning_rate": 0.00044500268600172483,
      "loss": 2.2841,
      "step": 78204
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.8180532455444336,
      "learning_rate": 0.0004449991049814482,
      "loss": 2.123,
      "step": 78205
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.7935935854911804,
      "learning_rate": 0.0004449955239342134,
      "loss": 2.2034,
      "step": 78206
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.8329626321792603,
      "learning_rate": 0.00044499194286002134,
      "loss": 2.1163,
      "step": 78207
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.7572258114814758,
      "learning_rate": 0.0004449883617588725,
      "loss": 2.3457,
      "step": 78208
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.7838630676269531,
      "learning_rate": 0.0004449847806307676,
      "loss": 2.2383,
      "step": 78209
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.8070191144943237,
      "learning_rate": 0.0004449811994757072,
      "loss": 2.1952,
      "step": 78210
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.8795928359031677,
      "learning_rate": 0.0004449776182936921,
      "loss": 2.151,
      "step": 78211
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.7931104302406311,
      "learning_rate": 0.000444974037084723,
      "loss": 2.3589,
      "step": 78212
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.8243759870529175,
      "learning_rate": 0.00044497045584880043,
      "loss": 2.2411,
      "step": 78213
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.8842483162879944,
      "learning_rate": 0.00044496687458592515,
      "loss": 2.3356,
      "step": 78214
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.7358095645904541,
      "learning_rate": 0.00044496329329609774,
      "loss": 2.2785,
      "step": 78215
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.728115975856781,
      "learning_rate": 0.00044495971197931896,
      "loss": 2.0119,
      "step": 78216
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.8106070756912231,
      "learning_rate": 0.00044495613063558936,
      "loss": 2.1157,
      "step": 78217
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.7323631048202515,
      "learning_rate": 0.00044495254926490974,
      "loss": 2.1755,
      "step": 78218
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.7937040328979492,
      "learning_rate": 0.0004449489678672807,
      "loss": 2.1566,
      "step": 78219
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.8312504887580872,
      "learning_rate": 0.0004449453864427029,
      "loss": 2.1853,
      "step": 78220
    },
    {
      "epoch": 1.02,
      "grad_norm": 1.2035778760910034,
      "learning_rate": 0.00044494180499117704,
      "loss": 2.3508,
      "step": 78221
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.7696748971939087,
      "learning_rate": 0.0004449382235127037,
      "loss": 2.15,
      "step": 78222
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.7806006073951721,
      "learning_rate": 0.00044493464200728367,
      "loss": 2.271,
      "step": 78223
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.766173779964447,
      "learning_rate": 0.0004449310604749175,
      "loss": 2.0454,
      "step": 78224
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.8934087753295898,
      "learning_rate": 0.00044492747891560603,
      "loss": 2.1312,
      "step": 78225
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.7512553930282593,
      "learning_rate": 0.0004449238973293498,
      "loss": 2.19,
      "step": 78226
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.7997949123382568,
      "learning_rate": 0.00044492031571614937,
      "loss": 2.1621,
      "step": 78227
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.8520444631576538,
      "learning_rate": 0.0004449167340760055,
      "loss": 2.0369,
      "step": 78228
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.7984672784805298,
      "learning_rate": 0.00044491315240891894,
      "loss": 2.1008,
      "step": 78229
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.8785509467124939,
      "learning_rate": 0.0004449095707148903,
      "loss": 2.1815,
      "step": 78230
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.7339916229248047,
      "learning_rate": 0.0004449059889939202,
      "loss": 2.0245,
      "step": 78231
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.7594373822212219,
      "learning_rate": 0.0004449024072460093,
      "loss": 2.2256,
      "step": 78232
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.9257070422172546,
      "learning_rate": 0.0004448988254711584,
      "loss": 2.1974,
      "step": 78233
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.7713905572891235,
      "learning_rate": 0.000444895243669368,
      "loss": 2.5753,
      "step": 78234
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.7796431183815002,
      "learning_rate": 0.0004448916618406389,
      "loss": 2.1482,
      "step": 78235
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.7729893922805786,
      "learning_rate": 0.0004448880799849717,
      "loss": 2.302,
      "step": 78236
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.7539969086647034,
      "learning_rate": 0.00044488449810236696,
      "loss": 2.11,
      "step": 78237
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.9776703715324402,
      "learning_rate": 0.0004448809161928256,
      "loss": 2.2511,
      "step": 78238
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.7681001424789429,
      "learning_rate": 0.0004448773342563481,
      "loss": 2.194,
      "step": 78239
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.8132648468017578,
      "learning_rate": 0.0004448737522929352,
      "loss": 1.9452,
      "step": 78240
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.783022403717041,
      "learning_rate": 0.00044487017030258745,
      "loss": 2.2236,
      "step": 78241
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.9480072855949402,
      "learning_rate": 0.00044486658828530565,
      "loss": 2.1421,
      "step": 78242
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.7681736350059509,
      "learning_rate": 0.00044486300624109037,
      "loss": 2.2857,
      "step": 78243
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.815420389175415,
      "learning_rate": 0.00044485942416994237,
      "loss": 2.3236,
      "step": 78244
    },
    {
      "epoch": 1.02,
      "grad_norm": 1.1063637733459473,
      "learning_rate": 0.00044485584207186235,
      "loss": 1.9564,
      "step": 78245
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.8427091240882874,
      "learning_rate": 0.0004448522599468507,
      "loss": 2.2593,
      "step": 78246
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.8333205580711365,
      "learning_rate": 0.0004448486777949084,
      "loss": 2.2584,
      "step": 78247
    },
    {
      "epoch": 1.02,
      "grad_norm": 1.2472022771835327,
      "learning_rate": 0.00044484509561603606,
      "loss": 2.0418,
      "step": 78248
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.8019403219223022,
      "learning_rate": 0.0004448415134102341,
      "loss": 2.0777,
      "step": 78249
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.8635172247886658,
      "learning_rate": 0.0004448379311775035,
      "loss": 1.9979,
      "step": 78250
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.8869162797927856,
      "learning_rate": 0.00044483434891784483,
      "loss": 1.9573,
      "step": 78251
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.7588046193122864,
      "learning_rate": 0.0004448307666312586,
      "loss": 2.3052,
      "step": 78252
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.7849225401878357,
      "learning_rate": 0.00044482718431774565,
      "loss": 2.2361,
      "step": 78253
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.8707283139228821,
      "learning_rate": 0.0004448236019773066,
      "loss": 2.4057,
      "step": 78254
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.8657618165016174,
      "learning_rate": 0.00044482001960994216,
      "loss": 2.1478,
      "step": 78255
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.728379487991333,
      "learning_rate": 0.00044481643721565285,
      "loss": 1.82,
      "step": 78256
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.7979217767715454,
      "learning_rate": 0.0004448128547944395,
      "loss": 2.1862,
      "step": 78257
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.9173856973648071,
      "learning_rate": 0.00044480927234630263,
      "loss": 2.0767,
      "step": 78258
    },
    {
      "epoch": 1.02,
      "grad_norm": 1.115583896636963,
      "learning_rate": 0.000444805689871243,
      "loss": 1.9838,
      "step": 78259
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.7529839873313904,
      "learning_rate": 0.00044480210736926127,
      "loss": 2.1001,
      "step": 78260
    },
    {
      "epoch": 1.02,
      "grad_norm": 1.0841220617294312,
      "learning_rate": 0.00044479852484035823,
      "loss": 2.1235,
      "step": 78261
    },
    {
      "epoch": 1.02,
      "grad_norm": 1.089143991470337,
      "learning_rate": 0.00044479494228453424,
      "loss": 2.1981,
      "step": 78262
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.866337776184082,
      "learning_rate": 0.00044479135970179015,
      "loss": 2.3187,
      "step": 78263
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.9703943133354187,
      "learning_rate": 0.0004447877770921267,
      "loss": 2.3741,
      "step": 78264
    },
    {
      "epoch": 1.02,
      "grad_norm": 1.1831358671188354,
      "learning_rate": 0.00044478419445554443,
      "loss": 2.2951,
      "step": 78265
    },
    {
      "epoch": 1.02,
      "grad_norm": 1.1738075017929077,
      "learning_rate": 0.00044478061179204403,
      "loss": 2.0168,
      "step": 78266
    },
    {
      "epoch": 1.02,
      "grad_norm": 1.0359224081039429,
      "learning_rate": 0.00044477702910162615,
      "loss": 2.2065,
      "step": 78267
    },
    {
      "epoch": 1.02,
      "grad_norm": 1.1237069368362427,
      "learning_rate": 0.0004447734463842916,
      "loss": 2.2231,
      "step": 78268
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.8124524354934692,
      "learning_rate": 0.00044476986364004085,
      "loss": 2.1541,
      "step": 78269
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.8515313863754272,
      "learning_rate": 0.00044476628086887463,
      "loss": 2.3172,
      "step": 78270
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.7880082726478577,
      "learning_rate": 0.00044476269807079363,
      "loss": 1.9657,
      "step": 78271
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.8758618235588074,
      "learning_rate": 0.00044475911524579856,
      "loss": 2.2785,
      "step": 78272
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.8270423412322998,
      "learning_rate": 0.00044475553239389,
      "loss": 2.1075,
      "step": 78273
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.8103238940238953,
      "learning_rate": 0.00044475194951506863,
      "loss": 2.1694,
      "step": 78274
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.7820323705673218,
      "learning_rate": 0.0004447483666093353,
      "loss": 2.2479,
      "step": 78275
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.9278969764709473,
      "learning_rate": 0.0004447447836766904,
      "loss": 2.1885,
      "step": 78276
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.8015096187591553,
      "learning_rate": 0.00044474120071713466,
      "loss": 2.2817,
      "step": 78277
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.8245770931243896,
      "learning_rate": 0.00044473761773066887,
      "loss": 2.4481,
      "step": 78278
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.8152544498443604,
      "learning_rate": 0.0004447340347172936,
      "loss": 2.4632,
      "step": 78279
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.8153302669525146,
      "learning_rate": 0.00044473045167700955,
      "loss": 2.0457,
      "step": 78280
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.7302085161209106,
      "learning_rate": 0.00044472686860981737,
      "loss": 2.3376,
      "step": 78281
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.8904193043708801,
      "learning_rate": 0.0004447232855157178,
      "loss": 2.3451,
      "step": 78282
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.8132445812225342,
      "learning_rate": 0.00044471970239471124,
      "loss": 2.0315,
      "step": 78283
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.7876169681549072,
      "learning_rate": 0.00044471611924679874,
      "loss": 2.1737,
      "step": 78284
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.7495191693305969,
      "learning_rate": 0.00044471253607198074,
      "loss": 2.2637,
      "step": 78285
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.9459750056266785,
      "learning_rate": 0.00044470895287025794,
      "loss": 2.27,
      "step": 78286
    },
    {
      "epoch": 1.02,
      "grad_norm": 1.002772331237793,
      "learning_rate": 0.0004447053696416311,
      "loss": 2.0679,
      "step": 78287
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.8104186654090881,
      "learning_rate": 0.00044470178638610064,
      "loss": 1.9358,
      "step": 78288
    },
    {
      "epoch": 1.02,
      "grad_norm": 1.105620265007019,
      "learning_rate": 0.0004446982031036674,
      "loss": 2.4339,
      "step": 78289
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.7814490795135498,
      "learning_rate": 0.00044469461979433216,
      "loss": 2.1996,
      "step": 78290
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.7564435601234436,
      "learning_rate": 0.0004446910364580954,
      "loss": 2.2641,
      "step": 78291
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.8013886213302612,
      "learning_rate": 0.0004446874530949578,
      "loss": 2.2873,
      "step": 78292
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.9222622513771057,
      "learning_rate": 0.00044468386970492014,
      "loss": 2.0824,
      "step": 78293
    },
    {
      "epoch": 1.02,
      "grad_norm": 1.0641518831253052,
      "learning_rate": 0.000444680286287983,
      "loss": 2.0559,
      "step": 78294
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.7518804669380188,
      "learning_rate": 0.000444676702844147,
      "loss": 2.2876,
      "step": 78295
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.9978958964347839,
      "learning_rate": 0.000444673119373413,
      "loss": 2.2225,
      "step": 78296
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.8576219081878662,
      "learning_rate": 0.00044466953587578143,
      "loss": 2.3652,
      "step": 78297
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.9083899855613708,
      "learning_rate": 0.00044466595235125305,
      "loss": 1.9855,
      "step": 78298
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.7114128470420837,
      "learning_rate": 0.00044466236879982864,
      "loss": 2.1792,
      "step": 78299
    },
    {
      "epoch": 1.02,
      "grad_norm": 1.1415733098983765,
      "learning_rate": 0.0004446587852215086,
      "loss": 2.1796,
      "step": 78300
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.9115744233131409,
      "learning_rate": 0.00044465520161629394,
      "loss": 2.2458,
      "step": 78301
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.8019686937332153,
      "learning_rate": 0.00044465161798418505,
      "loss": 2.2634,
      "step": 78302
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.8563382029533386,
      "learning_rate": 0.0004446480343251827,
      "loss": 2.0974,
      "step": 78303
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.891392171382904,
      "learning_rate": 0.00044464445063928756,
      "loss": 2.0602,
      "step": 78304
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.9044723510742188,
      "learning_rate": 0.00044464086692650026,
      "loss": 2.3174,
      "step": 78305
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.7731325626373291,
      "learning_rate": 0.0004446372831868216,
      "loss": 2.3261,
      "step": 78306
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.971393346786499,
      "learning_rate": 0.000444633699420252,
      "loss": 2.3303,
      "step": 78307
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.7852802872657776,
      "learning_rate": 0.0004446301156267924,
      "loss": 2.0145,
      "step": 78308
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.7555347084999084,
      "learning_rate": 0.0004446265318064432,
      "loss": 2.0673,
      "step": 78309
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.8597527742385864,
      "learning_rate": 0.0004446229479592053,
      "loss": 2.0903,
      "step": 78310
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.8974984884262085,
      "learning_rate": 0.0004446193640850792,
      "loss": 2.0381,
      "step": 78311
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.8394842147827148,
      "learning_rate": 0.00044461578018406566,
      "loss": 2.1369,
      "step": 78312
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.7721077799797058,
      "learning_rate": 0.0004446121962561653,
      "loss": 2.0983,
      "step": 78313
    },
    {
      "epoch": 1.02,
      "grad_norm": 1.0495610237121582,
      "learning_rate": 0.00044460861230137874,
      "loss": 2.3091,
      "step": 78314
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.7892523407936096,
      "learning_rate": 0.0004446050283197068,
      "loss": 2.1496,
      "step": 78315
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.8688624501228333,
      "learning_rate": 0.00044460144431115,
      "loss": 2.1088,
      "step": 78316
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.8325863480567932,
      "learning_rate": 0.00044459786027570915,
      "loss": 2.0144,
      "step": 78317
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.8347596526145935,
      "learning_rate": 0.0004445942762133848,
      "loss": 2.0732,
      "step": 78318
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.8410637974739075,
      "learning_rate": 0.0004445906921241776,
      "loss": 2.2416,
      "step": 78319
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.8119339942932129,
      "learning_rate": 0.0004445871080080883,
      "loss": 2.1801,
      "step": 78320
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.7733825445175171,
      "learning_rate": 0.00044458352386511754,
      "loss": 2.0769,
      "step": 78321
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.7579026818275452,
      "learning_rate": 0.0004445799396952659,
      "loss": 2.138,
      "step": 78322
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.749642550945282,
      "learning_rate": 0.00044457635549853417,
      "loss": 2.0201,
      "step": 78323
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.7495559453964233,
      "learning_rate": 0.00044457277127492295,
      "loss": 2.3069,
      "step": 78324
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.7836793661117554,
      "learning_rate": 0.00044456918702443296,
      "loss": 2.2378,
      "step": 78325
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.7452728152275085,
      "learning_rate": 0.0004445656027470648,
      "loss": 2.2588,
      "step": 78326
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.7071903347969055,
      "learning_rate": 0.0004445620184428192,
      "loss": 2.131,
      "step": 78327
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.8245983123779297,
      "learning_rate": 0.00044455843411169674,
      "loss": 2.0201,
      "step": 78328
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.9314661622047424,
      "learning_rate": 0.0004445548497536981,
      "loss": 2.3008,
      "step": 78329
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.7778028249740601,
      "learning_rate": 0.0004445512653688241,
      "loss": 2.2428,
      "step": 78330
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.8021154403686523,
      "learning_rate": 0.0004445476809570752,
      "loss": 2.3503,
      "step": 78331
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.7492367625236511,
      "learning_rate": 0.00044454409651845214,
      "loss": 2.1727,
      "step": 78332
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.7658355832099915,
      "learning_rate": 0.00044454051205295575,
      "loss": 2.191,
      "step": 78333
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.7379337549209595,
      "learning_rate": 0.00044453692756058643,
      "loss": 2.2281,
      "step": 78334
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.8445810079574585,
      "learning_rate": 0.0004445333430413449,
      "loss": 2.355,
      "step": 78335
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.766745924949646,
      "learning_rate": 0.00044452975849523205,
      "loss": 2.3293,
      "step": 78336
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.866919219493866,
      "learning_rate": 0.00044452617392224834,
      "loss": 2.1676,
      "step": 78337
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.7905303239822388,
      "learning_rate": 0.00044452258932239447,
      "loss": 2.061,
      "step": 78338
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.7202526330947876,
      "learning_rate": 0.00044451900469567114,
      "loss": 1.9792,
      "step": 78339
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.8081802725791931,
      "learning_rate": 0.000444515420042079,
      "loss": 2.2084,
      "step": 78340
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.7676619291305542,
      "learning_rate": 0.00044451183536161866,
      "loss": 2.0256,
      "step": 78341
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.9059813022613525,
      "learning_rate": 0.00044450825065429087,
      "loss": 2.1213,
      "step": 78342
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.9820590019226074,
      "learning_rate": 0.00044450466592009627,
      "loss": 2.3075,
      "step": 78343
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.8905106782913208,
      "learning_rate": 0.0004445010811590355,
      "loss": 2.2262,
      "step": 78344
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.7785694003105164,
      "learning_rate": 0.0004444974963711093,
      "loss": 2.2845,
      "step": 78345
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.7989442944526672,
      "learning_rate": 0.0004444939115563183,
      "loss": 2.2056,
      "step": 78346
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.8367443084716797,
      "learning_rate": 0.0004444903267146631,
      "loss": 2.1338,
      "step": 78347
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.7517989873886108,
      "learning_rate": 0.0004444867418461445,
      "loss": 2.2336,
      "step": 78348
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.8073123097419739,
      "learning_rate": 0.00044448315695076295,
      "loss": 2.2257,
      "step": 78349
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.8696640133857727,
      "learning_rate": 0.0004444795720285194,
      "loss": 2.1501,
      "step": 78350
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.8428139090538025,
      "learning_rate": 0.0004444759870794143,
      "loss": 2.2775,
      "step": 78351
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.7939813733100891,
      "learning_rate": 0.0004444724021034483,
      "loss": 2.1401,
      "step": 78352
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.7729503512382507,
      "learning_rate": 0.0004444688171006223,
      "loss": 2.406,
      "step": 78353
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.8589648008346558,
      "learning_rate": 0.0004444652320709368,
      "loss": 2.1644,
      "step": 78354
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.9106554388999939,
      "learning_rate": 0.00044446164701439237,
      "loss": 2.4137,
      "step": 78355
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.8303545713424683,
      "learning_rate": 0.0004444580619309899,
      "loss": 1.9513,
      "step": 78356
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.993041455745697,
      "learning_rate": 0.00044445447682073,
      "loss": 2.1373,
      "step": 78357
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.9527992010116577,
      "learning_rate": 0.00044445089168361317,
      "loss": 2.3684,
      "step": 78358
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.7267777919769287,
      "learning_rate": 0.00044444730651964023,
      "loss": 2.1322,
      "step": 78359
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.8074870705604553,
      "learning_rate": 0.0004444437213288118,
      "loss": 2.1688,
      "step": 78360
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.9416248798370361,
      "learning_rate": 0.00044444013611112855,
      "loss": 2.1672,
      "step": 78361
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.9154845476150513,
      "learning_rate": 0.00044443655086659115,
      "loss": 2.0472,
      "step": 78362
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.7591473460197449,
      "learning_rate": 0.00044443296559520034,
      "loss": 2.3279,
      "step": 78363
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.8768685460090637,
      "learning_rate": 0.00044442938029695667,
      "loss": 2.1732,
      "step": 78364
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.8149554133415222,
      "learning_rate": 0.00044442579497186083,
      "loss": 2.1848,
      "step": 78365
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.822796642780304,
      "learning_rate": 0.00044442220961991353,
      "loss": 2.3599,
      "step": 78366
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.8152110576629639,
      "learning_rate": 0.0004444186242411154,
      "loss": 2.3376,
      "step": 78367
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.8426061868667603,
      "learning_rate": 0.0004444150388354671,
      "loss": 2.1993,
      "step": 78368
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.7902616858482361,
      "learning_rate": 0.00044441145340296944,
      "loss": 2.4159,
      "step": 78369
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.8468390107154846,
      "learning_rate": 0.00044440786794362286,
      "loss": 2.1673,
      "step": 78370
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.8140610456466675,
      "learning_rate": 0.00044440428245742813,
      "loss": 2.1663,
      "step": 78371
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.8766577243804932,
      "learning_rate": 0.0004444006969443859,
      "loss": 2.2079,
      "step": 78372
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.8124142289161682,
      "learning_rate": 0.0004443971114044969,
      "loss": 2.0989,
      "step": 78373
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.8600754737854004,
      "learning_rate": 0.00044439352583776175,
      "loss": 2.2503,
      "step": 78374
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.7542101144790649,
      "learning_rate": 0.0004443899402441811,
      "loss": 2.3094,
      "step": 78375
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.7791130542755127,
      "learning_rate": 0.00044438635462375566,
      "loss": 2.1947,
      "step": 78376
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.8396865725517273,
      "learning_rate": 0.000444382768976486,
      "loss": 2.2143,
      "step": 78377
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.749681830406189,
      "learning_rate": 0.0004443791833023729,
      "loss": 2.0739,
      "step": 78378
    },
    {
      "epoch": 1.02,
      "grad_norm": 1.0670679807662964,
      "learning_rate": 0.0004443755976014171,
      "loss": 2.3133,
      "step": 78379
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.9279245734214783,
      "learning_rate": 0.000444372011873619,
      "loss": 2.1931,
      "step": 78380
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.8216902613639832,
      "learning_rate": 0.00044436842611897946,
      "loss": 2.1022,
      "step": 78381
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.6888372302055359,
      "learning_rate": 0.0004443648403374991,
      "loss": 1.989,
      "step": 78382
    },
    {
      "epoch": 1.02,
      "grad_norm": 1.012793779373169,
      "learning_rate": 0.00044436125452917865,
      "loss": 2.0719,
      "step": 78383
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.8436832427978516,
      "learning_rate": 0.00044435766869401866,
      "loss": 2.2204,
      "step": 78384
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.819885790348053,
      "learning_rate": 0.00044435408283201983,
      "loss": 2.1268,
      "step": 78385
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.8013010025024414,
      "learning_rate": 0.00044435049694318293,
      "loss": 2.3145,
      "step": 78386
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.8353217244148254,
      "learning_rate": 0.00044434691102750845,
      "loss": 2.1444,
      "step": 78387
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.8932396769523621,
      "learning_rate": 0.00044434332508499726,
      "loss": 2.2096,
      "step": 78388
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.8328663110733032,
      "learning_rate": 0.00044433973911564984,
      "loss": 2.2073,
      "step": 78389
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.7364171743392944,
      "learning_rate": 0.000444336153119467,
      "loss": 2.1031,
      "step": 78390
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.8642885684967041,
      "learning_rate": 0.0004443325670964493,
      "loss": 2.3217,
      "step": 78391
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.9199156761169434,
      "learning_rate": 0.00044432898104659744,
      "loss": 2.2028,
      "step": 78392
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.7920138835906982,
      "learning_rate": 0.0004443253949699121,
      "loss": 2.239,
      "step": 78393
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.8535229563713074,
      "learning_rate": 0.0004443218088663941,
      "loss": 2.148,
      "step": 78394
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.9322917461395264,
      "learning_rate": 0.0004443182227360437,
      "loss": 2.381,
      "step": 78395
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.8819958567619324,
      "learning_rate": 0.0004443146365788619,
      "loss": 2.2627,
      "step": 78396
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.8252840638160706,
      "learning_rate": 0.00044431105039484934,
      "loss": 2.2259,
      "step": 78397
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.8216552138328552,
      "learning_rate": 0.0004443074641840067,
      "loss": 2.0944,
      "step": 78398
    },
    {
      "epoch": 1.02,
      "grad_norm": 1.2100270986557007,
      "learning_rate": 0.0004443038779463344,
      "loss": 2.3338,
      "step": 78399
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.8053393959999084,
      "learning_rate": 0.00044430029168183343,
      "loss": 2.2701,
      "step": 78400
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.817089855670929,
      "learning_rate": 0.00044429670539050426,
      "loss": 2.2982,
      "step": 78401
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.7987868785858154,
      "learning_rate": 0.0004442931190723476,
      "loss": 1.9276,
      "step": 78402
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.7622541189193726,
      "learning_rate": 0.0004442895327273641,
      "loss": 2.1795,
      "step": 78403
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.8075871467590332,
      "learning_rate": 0.00044428594635555446,
      "loss": 2.2068,
      "step": 78404
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.7396449446678162,
      "learning_rate": 0.0004442823599569194,
      "loss": 2.0898,
      "step": 78405
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.8665421009063721,
      "learning_rate": 0.0004442787735314595,
      "loss": 2.2242,
      "step": 78406
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.7708008885383606,
      "learning_rate": 0.00044427518707917545,
      "loss": 2.116,
      "step": 78407
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.797664225101471,
      "learning_rate": 0.0004442716006000679,
      "loss": 2.1186,
      "step": 78408
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.7574217915534973,
      "learning_rate": 0.0004442680140941376,
      "loss": 2.138,
      "step": 78409
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.874097466468811,
      "learning_rate": 0.00044426442756138505,
      "loss": 2.1756,
      "step": 78410
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.7979276776313782,
      "learning_rate": 0.00044426084100181103,
      "loss": 2.2796,
      "step": 78411
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.7602013945579529,
      "learning_rate": 0.00044425725441541624,
      "loss": 2.1088,
      "step": 78412
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.7744417786598206,
      "learning_rate": 0.00044425366780220126,
      "loss": 2.1451,
      "step": 78413
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.7616130113601685,
      "learning_rate": 0.00044425008116216687,
      "loss": 2.1747,
      "step": 78414
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.7955095767974854,
      "learning_rate": 0.00044424649449531365,
      "loss": 2.4604,
      "step": 78415
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.826616644859314,
      "learning_rate": 0.00044424290780164225,
      "loss": 1.9904,
      "step": 78416
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.8574562072753906,
      "learning_rate": 0.0004442393210811534,
      "loss": 1.8906,
      "step": 78417
    },
    {
      "epoch": 1.02,
      "grad_norm": 1.0015251636505127,
      "learning_rate": 0.00044423573433384776,
      "loss": 2.0583,
      "step": 78418
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.7552928924560547,
      "learning_rate": 0.0004442321475597259,
      "loss": 2.2013,
      "step": 78419
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.8741669654846191,
      "learning_rate": 0.0004442285607587886,
      "loss": 2.2696,
      "step": 78420
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.8350666165351868,
      "learning_rate": 0.0004442249739310366,
      "loss": 2.1921,
      "step": 78421
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.9147208333015442,
      "learning_rate": 0.0004442213870764703,
      "loss": 2.0898,
      "step": 78422
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.7723994851112366,
      "learning_rate": 0.0004442178001950905,
      "loss": 2.2209,
      "step": 78423
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.8082950711250305,
      "learning_rate": 0.00044421421328689803,
      "loss": 1.9929,
      "step": 78424
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.861436128616333,
      "learning_rate": 0.00044421062635189326,
      "loss": 2.4889,
      "step": 78425
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.8098114132881165,
      "learning_rate": 0.0004442070393900771,
      "loss": 2.0647,
      "step": 78426
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.8545098900794983,
      "learning_rate": 0.0004442034524014502,
      "loss": 2.0933,
      "step": 78427
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.7865607738494873,
      "learning_rate": 0.00044419986538601306,
      "loss": 2.1091,
      "step": 78428
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.8079133629798889,
      "learning_rate": 0.0004441962783437664,
      "loss": 2.1285,
      "step": 78429
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.8278084397315979,
      "learning_rate": 0.0004441926912747111,
      "loss": 2.3328,
      "step": 78430
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.7936764359474182,
      "learning_rate": 0.00044418910417884746,
      "loss": 2.2473,
      "step": 78431
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.7423046827316284,
      "learning_rate": 0.00044418551705617646,
      "loss": 2.3474,
      "step": 78432
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.7293477058410645,
      "learning_rate": 0.0004441819299066986,
      "loss": 2.0115,
      "step": 78433
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.7470955848693848,
      "learning_rate": 0.0004441783427304146,
      "loss": 2.0879,
      "step": 78434
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.8878096342086792,
      "learning_rate": 0.00044417475552732514,
      "loss": 2.2964,
      "step": 78435
    },
    {
      "epoch": 1.02,
      "grad_norm": 1.1004517078399658,
      "learning_rate": 0.0004441711682974309,
      "loss": 2.3419,
      "step": 78436
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.781704843044281,
      "learning_rate": 0.00044416758104073243,
      "loss": 2.1139,
      "step": 78437
    },
    {
      "epoch": 1.02,
      "grad_norm": 1.3135758638381958,
      "learning_rate": 0.0004441639937572306,
      "loss": 2.1923,
      "step": 78438
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.9030342102050781,
      "learning_rate": 0.0004441604064469259,
      "loss": 2.2675,
      "step": 78439
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.7890454530715942,
      "learning_rate": 0.0004441568191098191,
      "loss": 2.2685,
      "step": 78440
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.7491301894187927,
      "learning_rate": 0.00044415323174591085,
      "loss": 2.2278,
      "step": 78441
    },
    {
      "epoch": 1.02,
      "grad_norm": 1.0289373397827148,
      "learning_rate": 0.00044414964435520165,
      "loss": 2.082,
      "step": 78442
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.7128786444664001,
      "learning_rate": 0.0004441460569376924,
      "loss": 2.3358,
      "step": 78443
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.8115358352661133,
      "learning_rate": 0.00044414246949338364,
      "loss": 2.3382,
      "step": 78444
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.7931869029998779,
      "learning_rate": 0.0004441388820222762,
      "loss": 2.2618,
      "step": 78445
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.7698594927787781,
      "learning_rate": 0.0004441352945243705,
      "loss": 2.1841,
      "step": 78446
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.80010986328125,
      "learning_rate": 0.00044413170699966737,
      "loss": 1.98,
      "step": 78447
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.7644833922386169,
      "learning_rate": 0.0004441281194481674,
      "loss": 2.2255,
      "step": 78448
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.7424587607383728,
      "learning_rate": 0.00044412453186987134,
      "loss": 2.2049,
      "step": 78449
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.7768501043319702,
      "learning_rate": 0.00044412094426477975,
      "loss": 2.2768,
      "step": 78450
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.8226280212402344,
      "learning_rate": 0.0004441173566328934,
      "loss": 2.3582,
      "step": 78451
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.8874351978302002,
      "learning_rate": 0.0004441137689742129,
      "loss": 1.9574,
      "step": 78452
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.8457069396972656,
      "learning_rate": 0.00044411018128873896,
      "loss": 2.3327,
      "step": 78453
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.8829298615455627,
      "learning_rate": 0.00044410659357647215,
      "loss": 2.3379,
      "step": 78454
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.8581655621528625,
      "learning_rate": 0.00044410300583741334,
      "loss": 2.2572,
      "step": 78455
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.8180444240570068,
      "learning_rate": 0.00044409941807156293,
      "loss": 2.1642,
      "step": 78456
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.7601826190948486,
      "learning_rate": 0.0004440958302789217,
      "loss": 2.0512,
      "step": 78457
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.7475688457489014,
      "learning_rate": 0.00044409224245949046,
      "loss": 2.1353,
      "step": 78458
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.8641993403434753,
      "learning_rate": 0.00044408865461326966,
      "loss": 2.2691,
      "step": 78459
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.9223002791404724,
      "learning_rate": 0.0004440850667402601,
      "loss": 2.092,
      "step": 78460
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.8134915232658386,
      "learning_rate": 0.0004440814788404624,
      "loss": 2.3213,
      "step": 78461
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.8616688847541809,
      "learning_rate": 0.0004440778909138772,
      "loss": 2.3982,
      "step": 78462
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.8503596782684326,
      "learning_rate": 0.0004440743029605052,
      "loss": 2.1716,
      "step": 78463
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.8551502227783203,
      "learning_rate": 0.00044407071498034714,
      "loss": 2.2596,
      "step": 78464
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.922700047492981,
      "learning_rate": 0.0004440671269734035,
      "loss": 1.9744,
      "step": 78465
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.9272122383117676,
      "learning_rate": 0.00044406353893967513,
      "loss": 2.4157,
      "step": 78466
    },
    {
      "epoch": 1.02,
      "grad_norm": 1.22117280960083,
      "learning_rate": 0.0004440599508791627,
      "loss": 2.2497,
      "step": 78467
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.900530993938446,
      "learning_rate": 0.00044405636279186666,
      "loss": 2.2695,
      "step": 78468
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.7915687561035156,
      "learning_rate": 0.00044405277467778786,
      "loss": 2.3164,
      "step": 78469
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.8789322376251221,
      "learning_rate": 0.00044404918653692704,
      "loss": 2.2473,
      "step": 78470
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.9478167295455933,
      "learning_rate": 0.00044404559836928463,
      "loss": 2.0716,
      "step": 78471
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.8594584465026855,
      "learning_rate": 0.0004440420101748615,
      "loss": 2.1961,
      "step": 78472
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.8331469297409058,
      "learning_rate": 0.0004440384219536582,
      "loss": 2.2348,
      "step": 78473
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.8428839445114136,
      "learning_rate": 0.00044403483370567546,
      "loss": 2.1539,
      "step": 78474
    },
    {
      "epoch": 1.02,
      "grad_norm": 1.0545274019241333,
      "learning_rate": 0.00044403124543091393,
      "loss": 2.2263,
      "step": 78475
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.8893129825592041,
      "learning_rate": 0.0004440276571293743,
      "loss": 2.1684,
      "step": 78476
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.7768742442131042,
      "learning_rate": 0.00044402406880105714,
      "loss": 2.1047,
      "step": 78477
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.9766497611999512,
      "learning_rate": 0.0004440204804459632,
      "loss": 2.3776,
      "step": 78478
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.8409120440483093,
      "learning_rate": 0.0004440168920640932,
      "loss": 2.3431,
      "step": 78479
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.7768291234970093,
      "learning_rate": 0.0004440133036554477,
      "loss": 2.1649,
      "step": 78480
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.8712889552116394,
      "learning_rate": 0.00044400971522002734,
      "loss": 2.056,
      "step": 78481
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.8649199604988098,
      "learning_rate": 0.000444006126757833,
      "loss": 2.0536,
      "step": 78482
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.7868125438690186,
      "learning_rate": 0.0004440025382688651,
      "loss": 1.9801,
      "step": 78483
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.7535959482192993,
      "learning_rate": 0.0004439989497531244,
      "loss": 2.2177,
      "step": 78484
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.920235812664032,
      "learning_rate": 0.0004439953612106117,
      "loss": 2.1896,
      "step": 78485
    },
    {
      "epoch": 1.02,
      "grad_norm": 1.0910013914108276,
      "learning_rate": 0.0004439917726413274,
      "loss": 2.0241,
      "step": 78486
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.8640756607055664,
      "learning_rate": 0.0004439881840452724,
      "loss": 2.0685,
      "step": 78487
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.9509738087654114,
      "learning_rate": 0.0004439845954224472,
      "loss": 2.3996,
      "step": 78488
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.9311442971229553,
      "learning_rate": 0.0004439810067728526,
      "loss": 2.0705,
      "step": 78489
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.8784093260765076,
      "learning_rate": 0.00044397741809648915,
      "loss": 2.1873,
      "step": 78490
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.8397506475448608,
      "learning_rate": 0.0004439738293933577,
      "loss": 2.2707,
      "step": 78491
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.7412571907043457,
      "learning_rate": 0.00044397024066345875,
      "loss": 2.253,
      "step": 78492
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.885892927646637,
      "learning_rate": 0.00044396665190679294,
      "loss": 2.4491,
      "step": 78493
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.7153914570808411,
      "learning_rate": 0.0004439630631233611,
      "loss": 2.0051,
      "step": 78494
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.89528489112854,
      "learning_rate": 0.0004439594743131638,
      "loss": 2.1081,
      "step": 78495
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.7719902992248535,
      "learning_rate": 0.00044395588547620164,
      "loss": 2.191,
      "step": 78496
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.7705762982368469,
      "learning_rate": 0.0004439522966124754,
      "loss": 2.0746,
      "step": 78497
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.9824792742729187,
      "learning_rate": 0.0004439487077219858,
      "loss": 2.3762,
      "step": 78498
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.9546428322792053,
      "learning_rate": 0.00044394511880473326,
      "loss": 2.2338,
      "step": 78499
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.8743169903755188,
      "learning_rate": 0.0004439415298607187,
      "loss": 2.3657,
      "step": 78500
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.7656871676445007,
      "learning_rate": 0.0004439379408899427,
      "loss": 1.7702,
      "step": 78501
    },
    {
      "epoch": 1.02,
      "grad_norm": 1.1090160608291626,
      "learning_rate": 0.0004439343518924059,
      "loss": 2.2455,
      "step": 78502
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.867601752281189,
      "learning_rate": 0.000443930762868109,
      "loss": 2.3613,
      "step": 78503
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.8181085586547852,
      "learning_rate": 0.00044392717381705265,
      "loss": 2.1817,
      "step": 78504
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.8110846877098083,
      "learning_rate": 0.0004439235847392375,
      "loss": 2.2651,
      "step": 78505
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.7291248440742493,
      "learning_rate": 0.0004439199956346642,
      "loss": 2.0672,
      "step": 78506
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.7857943773269653,
      "learning_rate": 0.0004439164065033336,
      "loss": 2.1624,
      "step": 78507
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.8848682045936584,
      "learning_rate": 0.00044391281734524604,
      "loss": 2.1353,
      "step": 78508
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.9256062507629395,
      "learning_rate": 0.00044390922816040244,
      "loss": 2.0396,
      "step": 78509
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.7726258635520935,
      "learning_rate": 0.00044390563894880345,
      "loss": 2.266,
      "step": 78510
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.827254593372345,
      "learning_rate": 0.00044390204971044966,
      "loss": 2.036,
      "step": 78511
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.8834301233291626,
      "learning_rate": 0.00044389846044534176,
      "loss": 2.3049,
      "step": 78512
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.7567273378372192,
      "learning_rate": 0.0004438948711534805,
      "loss": 2.0422,
      "step": 78513
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.8108460903167725,
      "learning_rate": 0.00044389128183486634,
      "loss": 2.1809,
      "step": 78514
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.927079975605011,
      "learning_rate": 0.0004438876924895001,
      "loss": 2.2558,
      "step": 78515
    },
    {
      "epoch": 1.02,
      "grad_norm": 1.041334867477417,
      "learning_rate": 0.0004438841031173825,
      "loss": 2.1934,
      "step": 78516
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.7722075581550598,
      "learning_rate": 0.00044388051371851403,
      "loss": 2.0992,
      "step": 78517
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.8588760495185852,
      "learning_rate": 0.0004438769242928955,
      "loss": 2.1934,
      "step": 78518
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.8810893297195435,
      "learning_rate": 0.00044387333484052755,
      "loss": 2.3982,
      "step": 78519
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.8138153553009033,
      "learning_rate": 0.0004438697453614108,
      "loss": 2.0842,
      "step": 78520
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.8081446886062622,
      "learning_rate": 0.0004438661558555459,
      "loss": 2.091,
      "step": 78521
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.9376379251480103,
      "learning_rate": 0.00044386256632293375,
      "loss": 2.5204,
      "step": 78522
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.8316478133201599,
      "learning_rate": 0.00044385897676357464,
      "loss": 2.1678,
      "step": 78523
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.8263140320777893,
      "learning_rate": 0.0004438553871774695,
      "loss": 2.2054,
      "step": 78524
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.8685610294342041,
      "learning_rate": 0.00044385179756461894,
      "loss": 2.0064,
      "step": 78525
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.7946330904960632,
      "learning_rate": 0.0004438482079250236,
      "loss": 2.1645,
      "step": 78526
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.7838353514671326,
      "learning_rate": 0.0004438446182586841,
      "loss": 2.3046,
      "step": 78527
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.8244288563728333,
      "learning_rate": 0.00044384102856560127,
      "loss": 2.4319,
      "step": 78528
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.8063453435897827,
      "learning_rate": 0.0004438374388457757,
      "loss": 2.2199,
      "step": 78529
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.7478862404823303,
      "learning_rate": 0.0004438338490992079,
      "loss": 2.0487,
      "step": 78530
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.7961472868919373,
      "learning_rate": 0.0004438302593258988,
      "loss": 2.2878,
      "step": 78531
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.9302589297294617,
      "learning_rate": 0.00044382666952584885,
      "loss": 2.1299,
      "step": 78532
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.7216107845306396,
      "learning_rate": 0.00044382307969905886,
      "loss": 2.339,
      "step": 78533
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.7780263423919678,
      "learning_rate": 0.00044381948984552946,
      "loss": 2.1691,
      "step": 78534
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.8265568017959595,
      "learning_rate": 0.00044381589996526124,
      "loss": 2.355,
      "step": 78535
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.85505211353302,
      "learning_rate": 0.000443812310058255,
      "loss": 2.2099,
      "step": 78536
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.788360595703125,
      "learning_rate": 0.0004438087201245113,
      "loss": 2.2384,
      "step": 78537
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.7894700169563293,
      "learning_rate": 0.00044380513016403084,
      "loss": 2.2411,
      "step": 78538
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.843843400478363,
      "learning_rate": 0.00044380154017681426,
      "loss": 2.4855,
      "step": 78539
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.8689147233963013,
      "learning_rate": 0.00044379795016286235,
      "loss": 2.0382,
      "step": 78540
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.8486250042915344,
      "learning_rate": 0.00044379436012217564,
      "loss": 2.1552,
      "step": 78541
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.7613056898117065,
      "learning_rate": 0.00044379077005475483,
      "loss": 2.2034,
      "step": 78542
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.8660030364990234,
      "learning_rate": 0.0004437871799606006,
      "loss": 2.1617,
      "step": 78543
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.77295982837677,
      "learning_rate": 0.0004437835898397136,
      "loss": 2.0508,
      "step": 78544
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.8187785148620605,
      "learning_rate": 0.00044377999969209456,
      "loss": 1.9617,
      "step": 78545
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.7630543112754822,
      "learning_rate": 0.00044377640951774415,
      "loss": 1.9599,
      "step": 78546
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.7517595291137695,
      "learning_rate": 0.00044377281931666285,
      "loss": 2.1578,
      "step": 78547
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.8791439533233643,
      "learning_rate": 0.0004437692290888515,
      "loss": 2.3248,
      "step": 78548
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.8532849550247192,
      "learning_rate": 0.00044376563883431085,
      "loss": 2.4882,
      "step": 78549
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.7672600150108337,
      "learning_rate": 0.00044376204855304147,
      "loss": 2.0763,
      "step": 78550
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.8739878535270691,
      "learning_rate": 0.0004437584582450439,
      "loss": 1.9673,
      "step": 78551
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.6983193159103394,
      "learning_rate": 0.00044375486791031895,
      "loss": 2.085,
      "step": 78552
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.7249811887741089,
      "learning_rate": 0.00044375127754886723,
      "loss": 2.0597,
      "step": 78553
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.8435010313987732,
      "learning_rate": 0.00044374768716068944,
      "loss": 2.2165,
      "step": 78554
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.7835954427719116,
      "learning_rate": 0.00044374409674578633,
      "loss": 2.1786,
      "step": 78555
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.8197384476661682,
      "learning_rate": 0.0004437405063041584,
      "loss": 2.3528,
      "step": 78556
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.8234542012214661,
      "learning_rate": 0.00044373691583580636,
      "loss": 2.1767,
      "step": 78557
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.7876695990562439,
      "learning_rate": 0.000443733325340731,
      "loss": 2.4464,
      "step": 78558
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.7840582728385925,
      "learning_rate": 0.0004437297348189328,
      "loss": 2.1197,
      "step": 78559
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.7942496538162231,
      "learning_rate": 0.00044372614427041265,
      "loss": 2.2393,
      "step": 78560
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.7344622611999512,
      "learning_rate": 0.00044372255369517106,
      "loss": 2.2686,
      "step": 78561
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.871335506439209,
      "learning_rate": 0.0004437189630932087,
      "loss": 2.159,
      "step": 78562
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.8790753483772278,
      "learning_rate": 0.00044371537246452625,
      "loss": 2.1754,
      "step": 78563
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.7596117854118347,
      "learning_rate": 0.0004437117818091245,
      "loss": 2.1926,
      "step": 78564
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.7952696681022644,
      "learning_rate": 0.00044370819112700394,
      "loss": 2.257,
      "step": 78565
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.7967914342880249,
      "learning_rate": 0.00044370460041816533,
      "loss": 2.1071,
      "step": 78566
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.899145245552063,
      "learning_rate": 0.00044370100968260933,
      "loss": 2.0468,
      "step": 78567
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.8794864416122437,
      "learning_rate": 0.0004436974189203366,
      "loss": 2.0794,
      "step": 78568
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.8181644082069397,
      "learning_rate": 0.0004436938281313477,
      "loss": 2.4408,
      "step": 78569
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.8748907446861267,
      "learning_rate": 0.00044369023731564354,
      "loss": 2.099,
      "step": 78570
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.8729477524757385,
      "learning_rate": 0.00044368664647322466,
      "loss": 2.3339,
      "step": 78571
    },
    {
      "epoch": 1.02,
      "grad_norm": 1.051795482635498,
      "learning_rate": 0.0004436830556040916,
      "loss": 2.1992,
      "step": 78572
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.8455396890640259,
      "learning_rate": 0.0004436794647082452,
      "loss": 2.206,
      "step": 78573
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.8634857535362244,
      "learning_rate": 0.00044367587378568615,
      "loss": 2.1611,
      "step": 78574
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.7560642957687378,
      "learning_rate": 0.00044367228283641503,
      "loss": 1.8296,
      "step": 78575
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.9827050566673279,
      "learning_rate": 0.0004436686918604325,
      "loss": 2.2528,
      "step": 78576
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.8199619650840759,
      "learning_rate": 0.0004436651008577392,
      "loss": 1.9891,
      "step": 78577
    },
    {
      "epoch": 1.02,
      "grad_norm": 1.0821670293807983,
      "learning_rate": 0.0004436615098283359,
      "loss": 2.139,
      "step": 78578
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.8249540328979492,
      "learning_rate": 0.00044365791877222315,
      "loss": 2.1864,
      "step": 78579
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.8317500948905945,
      "learning_rate": 0.0004436543276894017,
      "loss": 2.2119,
      "step": 78580
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.8219402432441711,
      "learning_rate": 0.0004436507365798722,
      "loss": 2.315,
      "step": 78581
    },
    {
      "epoch": 1.02,
      "grad_norm": 1.2580082416534424,
      "learning_rate": 0.00044364714544363536,
      "loss": 2.2022,
      "step": 78582
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.9270908832550049,
      "learning_rate": 0.00044364355428069184,
      "loss": 2.2473,
      "step": 78583
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.7481781244277954,
      "learning_rate": 0.00044363996309104215,
      "loss": 2.2343,
      "step": 78584
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.7921122312545776,
      "learning_rate": 0.00044363637187468715,
      "loss": 2.0732,
      "step": 78585
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.8693426251411438,
      "learning_rate": 0.00044363278063162744,
      "loss": 2.1927,
      "step": 78586
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.9002547860145569,
      "learning_rate": 0.00044362918936186366,
      "loss": 2.0738,
      "step": 78587
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.8097209334373474,
      "learning_rate": 0.0004436255980653965,
      "loss": 2.4105,
      "step": 78588
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.79152512550354,
      "learning_rate": 0.00044362200674222664,
      "loss": 2.1962,
      "step": 78589
    },
    {
      "epoch": 1.02,
      "grad_norm": 1.0177251100540161,
      "learning_rate": 0.00044361841539235474,
      "loss": 2.0986,
      "step": 78590
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.8525339365005493,
      "learning_rate": 0.0004436148240157815,
      "loss": 2.2244,
      "step": 78591
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.84540194272995,
      "learning_rate": 0.00044361123261250755,
      "loss": 2.2504,
      "step": 78592
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.8499987125396729,
      "learning_rate": 0.0004436076411825335,
      "loss": 2.0771,
      "step": 78593
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.6923608779907227,
      "learning_rate": 0.0004436040497258601,
      "loss": 2.3475,
      "step": 78594
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.8683924078941345,
      "learning_rate": 0.0004436004582424881,
      "loss": 1.9883,
      "step": 78595
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.908734142780304,
      "learning_rate": 0.0004435968667324179,
      "loss": 2.2182,
      "step": 78596
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.7269027233123779,
      "learning_rate": 0.00044359327519565033,
      "loss": 2.3286,
      "step": 78597
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.7608396410942078,
      "learning_rate": 0.00044358968363218627,
      "loss": 2.2213,
      "step": 78598
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.8126769065856934,
      "learning_rate": 0.00044358609204202596,
      "loss": 2.1054,
      "step": 78599
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.809928297996521,
      "learning_rate": 0.00044358250042517036,
      "loss": 2.1143,
      "step": 78600
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.79627525806427,
      "learning_rate": 0.0004435789087816201,
      "loss": 2.0224,
      "step": 78601
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.8342615365982056,
      "learning_rate": 0.00044357531711137577,
      "loss": 2.3218,
      "step": 78602
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.7799577116966248,
      "learning_rate": 0.00044357172541443807,
      "loss": 2.3063,
      "step": 78603
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.8978183269500732,
      "learning_rate": 0.00044356813369080775,
      "loss": 2.3568,
      "step": 78604
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.8130786418914795,
      "learning_rate": 0.0004435645419404853,
      "loss": 2.2197,
      "step": 78605
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.8257195353507996,
      "learning_rate": 0.0004435609501634715,
      "loss": 2.3083,
      "step": 78606
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.790187656879425,
      "learning_rate": 0.0004435573583597671,
      "loss": 2.1794,
      "step": 78607
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.7704861760139465,
      "learning_rate": 0.0004435537665293726,
      "loss": 2.0384,
      "step": 78608
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.7533073425292969,
      "learning_rate": 0.00044355017467228875,
      "loss": 2.2877,
      "step": 78609
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.8656835556030273,
      "learning_rate": 0.0004435465827885163,
      "loss": 2.2067,
      "step": 78610
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.7759836912155151,
      "learning_rate": 0.00044354299087805577,
      "loss": 2.1024,
      "step": 78611
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.7989208698272705,
      "learning_rate": 0.0004435393989409078,
      "loss": 1.9168,
      "step": 78612
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.7354174256324768,
      "learning_rate": 0.0004435358069770733,
      "loss": 2.0795,
      "step": 78613
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.8057670593261719,
      "learning_rate": 0.0004435322149865527,
      "loss": 2.0952,
      "step": 78614
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.7566723227500916,
      "learning_rate": 0.0004435286229693468,
      "loss": 2.0184,
      "step": 78615
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.7877341508865356,
      "learning_rate": 0.0004435250309254561,
      "loss": 2.1546,
      "step": 78616
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.7955614924430847,
      "learning_rate": 0.0004435214388548815,
      "loss": 1.9934,
      "step": 78617
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.8222023844718933,
      "learning_rate": 0.00044351784675762353,
      "loss": 2.2734,
      "step": 78618
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.7401549816131592,
      "learning_rate": 0.0004435142546336829,
      "loss": 2.0496,
      "step": 78619
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.7663906216621399,
      "learning_rate": 0.0004435106624830602,
      "loss": 2.3355,
      "step": 78620
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.8022374510765076,
      "learning_rate": 0.0004435070703057562,
      "loss": 2.2389,
      "step": 78621
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.7904215455055237,
      "learning_rate": 0.00044350347810177154,
      "loss": 2.2385,
      "step": 78622
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.8303681015968323,
      "learning_rate": 0.00044349988587110686,
      "loss": 1.9827,
      "step": 78623
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.8332076668739319,
      "learning_rate": 0.00044349629361376284,
      "loss": 2.2827,
      "step": 78624
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.7705979347229004,
      "learning_rate": 0.0004434927013297402,
      "loss": 2.0359,
      "step": 78625
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.8201976418495178,
      "learning_rate": 0.0004434891090190395,
      "loss": 2.1528,
      "step": 78626
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.7402136921882629,
      "learning_rate": 0.00044348551668166146,
      "loss": 2.1173,
      "step": 78627
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.8409686088562012,
      "learning_rate": 0.0004434819243176068,
      "loss": 2.0951,
      "step": 78628
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.740065336227417,
      "learning_rate": 0.0004434783319268761,
      "loss": 2.1382,
      "step": 78629
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.7514106035232544,
      "learning_rate": 0.00044347473950947,
      "loss": 2.2077,
      "step": 78630
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.8889366388320923,
      "learning_rate": 0.0004434711470653894,
      "loss": 2.2114,
      "step": 78631
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.8147693872451782,
      "learning_rate": 0.00044346755459463466,
      "loss": 2.1287,
      "step": 78632
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.8826654553413391,
      "learning_rate": 0.00044346396209720676,
      "loss": 2.2199,
      "step": 78633
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.9020521640777588,
      "learning_rate": 0.00044346036957310607,
      "loss": 2.185,
      "step": 78634
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.8318436741828918,
      "learning_rate": 0.0004434567770223334,
      "loss": 2.0262,
      "step": 78635
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.8751415610313416,
      "learning_rate": 0.00044345318444488944,
      "loss": 2.2526,
      "step": 78636
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.8379842638969421,
      "learning_rate": 0.0004434495918407748,
      "loss": 2.2237,
      "step": 78637
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.8349238038063049,
      "learning_rate": 0.0004434459992099902,
      "loss": 2.3792,
      "step": 78638
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.8732275366783142,
      "learning_rate": 0.00044344240655253627,
      "loss": 2.1593,
      "step": 78639
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.7451722025871277,
      "learning_rate": 0.0004434388138684137,
      "loss": 2.1479,
      "step": 78640
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.8269994258880615,
      "learning_rate": 0.00044343522115762314,
      "loss": 2.2795,
      "step": 78641
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.9435796141624451,
      "learning_rate": 0.0004434316284201652,
      "loss": 2.3389,
      "step": 78642
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.936409056186676,
      "learning_rate": 0.00044342803565604067,
      "loss": 2.3145,
      "step": 78643
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.7528108358383179,
      "learning_rate": 0.00044342444286525027,
      "loss": 2.2824,
      "step": 78644
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.8833172917366028,
      "learning_rate": 0.0004434208500477944,
      "loss": 2.2739,
      "step": 78645
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.826726496219635,
      "learning_rate": 0.00044341725720367387,
      "loss": 2.2588,
      "step": 78646
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.7687574028968811,
      "learning_rate": 0.00044341366433288946,
      "loss": 2.1981,
      "step": 78647
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.8182879090309143,
      "learning_rate": 0.0004434100714354417,
      "loss": 2.1747,
      "step": 78648
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.9993631839752197,
      "learning_rate": 0.00044340647851133134,
      "loss": 2.326,
      "step": 78649
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.7754249572753906,
      "learning_rate": 0.00044340288556055904,
      "loss": 2.1972,
      "step": 78650
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.830506443977356,
      "learning_rate": 0.00044339929258312534,
      "loss": 2.1765,
      "step": 78651
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.8483571410179138,
      "learning_rate": 0.000443395699579031,
      "loss": 2.4108,
      "step": 78652
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.7770529985427856,
      "learning_rate": 0.00044339210654827685,
      "loss": 2.2232,
      "step": 78653
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.8096963763237,
      "learning_rate": 0.00044338851349086324,
      "loss": 2.0817,
      "step": 78654
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.8229551911354065,
      "learning_rate": 0.00044338492040679105,
      "loss": 2.1876,
      "step": 78655
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.9230020642280579,
      "learning_rate": 0.0004433813272960609,
      "loss": 2.3408,
      "step": 78656
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.8780824542045593,
      "learning_rate": 0.0004433777341586735,
      "loss": 2.1331,
      "step": 78657
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.7936515808105469,
      "learning_rate": 0.0004433741409946293,
      "loss": 2.219,
      "step": 78658
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.8322664499282837,
      "learning_rate": 0.0004433705478039293,
      "loss": 2.0679,
      "step": 78659
    },
    {
      "epoch": 1.02,
      "grad_norm": 1.190553903579712,
      "learning_rate": 0.000443366954586574,
      "loss": 2.1602,
      "step": 78660
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.8224014639854431,
      "learning_rate": 0.00044336336134256406,
      "loss": 2.0937,
      "step": 78661
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.8276404142379761,
      "learning_rate": 0.0004433597680719002,
      "loss": 2.4367,
      "step": 78662
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.860783576965332,
      "learning_rate": 0.00044335617477458294,
      "loss": 2.1397,
      "step": 78663
    },
    {
      "epoch": 1.02,
      "grad_norm": 1.294371247291565,
      "learning_rate": 0.0004433525814506131,
      "loss": 2.2699,
      "step": 78664
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.9049555063247681,
      "learning_rate": 0.00044334898809999135,
      "loss": 2.3887,
      "step": 78665
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.8495929837226868,
      "learning_rate": 0.0004433453947227183,
      "loss": 2.2242,
      "step": 78666
    },
    {
      "epoch": 1.02,
      "grad_norm": 1.187502145767212,
      "learning_rate": 0.0004433418013187946,
      "loss": 1.8442,
      "step": 78667
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.9149501323699951,
      "learning_rate": 0.00044333820788822106,
      "loss": 2.1827,
      "step": 78668
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.7778299450874329,
      "learning_rate": 0.0004433346144309982,
      "loss": 2.1669,
      "step": 78669
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.9024218916893005,
      "learning_rate": 0.00044333102094712667,
      "loss": 2.3985,
      "step": 78670
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.8517228364944458,
      "learning_rate": 0.00044332742743660725,
      "loss": 2.302,
      "step": 78671
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.8368909955024719,
      "learning_rate": 0.00044332383389944046,
      "loss": 2.299,
      "step": 78672
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.7503969073295593,
      "learning_rate": 0.0004433202403356272,
      "loss": 2.2976,
      "step": 78673
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.78554368019104,
      "learning_rate": 0.00044331664674516797,
      "loss": 2.2134,
      "step": 78674
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.7507612109184265,
      "learning_rate": 0.0004433130531280634,
      "loss": 2.2909,
      "step": 78675
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.8681730031967163,
      "learning_rate": 0.0004433094594843143,
      "loss": 1.9792,
      "step": 78676
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.7167954444885254,
      "learning_rate": 0.0004433058658139212,
      "loss": 2.1922,
      "step": 78677
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.8138352036476135,
      "learning_rate": 0.00044330227211688487,
      "loss": 2.3118,
      "step": 78678
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.8802489042282104,
      "learning_rate": 0.000443298678393206,
      "loss": 2.2127,
      "step": 78679
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.8797309398651123,
      "learning_rate": 0.0004432950846428851,
      "loss": 2.0481,
      "step": 78680
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.990676760673523,
      "learning_rate": 0.000443291490865923,
      "loss": 2.1019,
      "step": 78681
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.8269438147544861,
      "learning_rate": 0.00044328789706232026,
      "loss": 2.1057,
      "step": 78682
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.9670005440711975,
      "learning_rate": 0.00044328430323207764,
      "loss": 2.1713,
      "step": 78683
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.979383111000061,
      "learning_rate": 0.00044328070937519573,
      "loss": 2.1372,
      "step": 78684
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.8490161895751953,
      "learning_rate": 0.0004432771154916753,
      "loss": 2.1545,
      "step": 78685
    },
    {
      "epoch": 1.02,
      "grad_norm": 1.134595513343811,
      "learning_rate": 0.0004432735215815169,
      "loss": 2.3823,
      "step": 78686
    },
    {
      "epoch": 1.02,
      "grad_norm": 1.239117980003357,
      "learning_rate": 0.0004432699276447213,
      "loss": 2.0769,
      "step": 78687
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.8364894986152649,
      "learning_rate": 0.0004432663336812889,
      "loss": 1.9736,
      "step": 78688
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.9516570568084717,
      "learning_rate": 0.00044326273969122086,
      "loss": 2.0828,
      "step": 78689
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.9573150277137756,
      "learning_rate": 0.0004432591456745174,
      "loss": 2.0847,
      "step": 78690
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.8632072806358337,
      "learning_rate": 0.00044325555163117947,
      "loss": 2.3621,
      "step": 78691
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.7369115352630615,
      "learning_rate": 0.00044325195756120766,
      "loss": 2.1918,
      "step": 78692
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.753632664680481,
      "learning_rate": 0.0004432483634646024,
      "loss": 2.202,
      "step": 78693
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.7937310338020325,
      "learning_rate": 0.00044324476934136473,
      "loss": 2.0301,
      "step": 78694
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.8921564817428589,
      "learning_rate": 0.00044324117519149513,
      "loss": 2.1339,
      "step": 78695
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.9799548983573914,
      "learning_rate": 0.0004432375810149943,
      "loss": 2.2902,
      "step": 78696
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.7236500382423401,
      "learning_rate": 0.0004432339868118628,
      "loss": 2.3016,
      "step": 78697
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.936134934425354,
      "learning_rate": 0.00044323039258210145,
      "loss": 2.1529,
      "step": 78698
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.8725200891494751,
      "learning_rate": 0.00044322679832571095,
      "loss": 2.0776,
      "step": 78699
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.8409761190414429,
      "learning_rate": 0.0004432232040426919,
      "loss": 2.2357,
      "step": 78700
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.8350934982299805,
      "learning_rate": 0.0004432196097330448,
      "loss": 2.1616,
      "step": 78701
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.8998572826385498,
      "learning_rate": 0.00044321601539677063,
      "loss": 2.1692,
      "step": 78702
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.7482444643974304,
      "learning_rate": 0.0004432124210338698,
      "loss": 2.1071,
      "step": 78703
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.8934265375137329,
      "learning_rate": 0.000443208826644343,
      "loss": 2.1774,
      "step": 78704
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.9485938549041748,
      "learning_rate": 0.0004432052322281911,
      "loss": 2.2431,
      "step": 78705
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.8030409812927246,
      "learning_rate": 0.00044320163778541465,
      "loss": 2.1683,
      "step": 78706
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.9814192652702332,
      "learning_rate": 0.00044319804331601425,
      "loss": 2.2431,
      "step": 78707
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.7422719597816467,
      "learning_rate": 0.00044319444881999067,
      "loss": 2.0442,
      "step": 78708
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.7094664573669434,
      "learning_rate": 0.00044319085429734444,
      "loss": 2.0988,
      "step": 78709
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.8477348685264587,
      "learning_rate": 0.00044318725974807645,
      "loss": 1.9546,
      "step": 78710
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.7833947539329529,
      "learning_rate": 0.0004431836651721873,
      "loss": 2.3508,
      "step": 78711
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.8190138936042786,
      "learning_rate": 0.0004431800705696775,
      "loss": 2.2605,
      "step": 78712
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.8058279156684875,
      "learning_rate": 0.0004431764759405478,
      "loss": 2.1118,
      "step": 78713
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.7611833810806274,
      "learning_rate": 0.0004431728812847989,
      "loss": 2.0911,
      "step": 78714
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.7952654361724854,
      "learning_rate": 0.0004431692866024315,
      "loss": 2.1368,
      "step": 78715
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.861482560634613,
      "learning_rate": 0.0004431656918934462,
      "loss": 2.242,
      "step": 78716
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.7793318033218384,
      "learning_rate": 0.0004431620971578437,
      "loss": 2.1303,
      "step": 78717
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.7070614695549011,
      "learning_rate": 0.0004431585023956247,
      "loss": 2.0817,
      "step": 78718
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.7650923728942871,
      "learning_rate": 0.0004431549076067898,
      "loss": 2.0179,
      "step": 78719
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.8476481437683105,
      "learning_rate": 0.00044315131279133973,
      "loss": 2.1096,
      "step": 78720
    },
    {
      "epoch": 1.02,
      "grad_norm": 0.8166215419769287,
      "learning_rate": 0.0004431477179492751,
      "loss": 2.1944,
      "step": 78721
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.8840212225914001,
      "learning_rate": 0.0004431441230805966,
      "loss": 2.1963,
      "step": 78722
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.8095898032188416,
      "learning_rate": 0.0004431405281853049,
      "loss": 2.3296,
      "step": 78723
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.7956197261810303,
      "learning_rate": 0.0004431369332634007,
      "loss": 2.2178,
      "step": 78724
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.9137409329414368,
      "learning_rate": 0.00044313333831488464,
      "loss": 2.2198,
      "step": 78725
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.7730966210365295,
      "learning_rate": 0.0004431297433397574,
      "loss": 2.1686,
      "step": 78726
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.9256358742713928,
      "learning_rate": 0.0004431261483380196,
      "loss": 2.1596,
      "step": 78727
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.7950300574302673,
      "learning_rate": 0.00044312255330967196,
      "loss": 2.0396,
      "step": 78728
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.7620077729225159,
      "learning_rate": 0.0004431189582547152,
      "loss": 2.0834,
      "step": 78729
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.801158607006073,
      "learning_rate": 0.00044311536317314987,
      "loss": 2.3006,
      "step": 78730
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.7674355506896973,
      "learning_rate": 0.00044311176806497663,
      "loss": 2.1279,
      "step": 78731
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.9035864472389221,
      "learning_rate": 0.0004431081729301963,
      "loss": 1.978,
      "step": 78732
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.7977733612060547,
      "learning_rate": 0.0004431045777688094,
      "loss": 2.1643,
      "step": 78733
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.8387318253517151,
      "learning_rate": 0.0004431009825808167,
      "loss": 2.3725,
      "step": 78734
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.8671756982803345,
      "learning_rate": 0.00044309738736621884,
      "loss": 2.2468,
      "step": 78735
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.9177421927452087,
      "learning_rate": 0.00044309379212501644,
      "loss": 2.3458,
      "step": 78736
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.7770434021949768,
      "learning_rate": 0.0004430901968572102,
      "loss": 2.3438,
      "step": 78737
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.8285878300666809,
      "learning_rate": 0.00044308660156280085,
      "loss": 2.1589,
      "step": 78738
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.7605685591697693,
      "learning_rate": 0.0004430830062417889,
      "loss": 2.0994,
      "step": 78739
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.8043964505195618,
      "learning_rate": 0.0004430794108941752,
      "loss": 2.2976,
      "step": 78740
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.8200666308403015,
      "learning_rate": 0.00044307581551996034,
      "loss": 2.2222,
      "step": 78741
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.7183419466018677,
      "learning_rate": 0.0004430722201191449,
      "loss": 2.1675,
      "step": 78742
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.845070481300354,
      "learning_rate": 0.00044306862469172974,
      "loss": 2.1344,
      "step": 78743
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.7629730701446533,
      "learning_rate": 0.0004430650292377154,
      "loss": 2.0918,
      "step": 78744
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.7714369893074036,
      "learning_rate": 0.0004430614337571025,
      "loss": 2.1931,
      "step": 78745
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.7670108675956726,
      "learning_rate": 0.0004430578382498918,
      "loss": 2.3806,
      "step": 78746
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.7389385104179382,
      "learning_rate": 0.000443054242716084,
      "loss": 2.1524,
      "step": 78747
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.7729616761207581,
      "learning_rate": 0.0004430506471556797,
      "loss": 2.0575,
      "step": 78748
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.7569680213928223,
      "learning_rate": 0.0004430470515686795,
      "loss": 2.227,
      "step": 78749
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.6737158298492432,
      "learning_rate": 0.0004430434559550843,
      "loss": 2.0661,
      "step": 78750
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.7618774771690369,
      "learning_rate": 0.0004430398603148945,
      "loss": 2.213,
      "step": 78751
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.7666993141174316,
      "learning_rate": 0.00044303626464811094,
      "loss": 2.0713,
      "step": 78752
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.8252415657043457,
      "learning_rate": 0.00044303266895473427,
      "loss": 2.2563,
      "step": 78753
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.7411011457443237,
      "learning_rate": 0.00044302907323476504,
      "loss": 2.003,
      "step": 78754
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.8402172327041626,
      "learning_rate": 0.000443025477488204,
      "loss": 2.1816,
      "step": 78755
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.7946583032608032,
      "learning_rate": 0.000443021881715052,
      "loss": 2.1486,
      "step": 78756
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.7969104051589966,
      "learning_rate": 0.0004430182859153093,
      "loss": 2.2823,
      "step": 78757
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.7853930592536926,
      "learning_rate": 0.00044301469008897694,
      "loss": 2.3277,
      "step": 78758
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.7863012552261353,
      "learning_rate": 0.0004430110942360554,
      "loss": 2.4202,
      "step": 78759
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.7244409918785095,
      "learning_rate": 0.00044300749835654547,
      "loss": 2.2295,
      "step": 78760
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.7661923170089722,
      "learning_rate": 0.0004430039024504477,
      "loss": 2.0919,
      "step": 78761
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.7991419434547424,
      "learning_rate": 0.0004430003065177628,
      "loss": 2.2989,
      "step": 78762
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.8463612198829651,
      "learning_rate": 0.0004429967105584914,
      "loss": 2.2276,
      "step": 78763
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.849826991558075,
      "learning_rate": 0.0004429931145726343,
      "loss": 2.2424,
      "step": 78764
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.7908490896224976,
      "learning_rate": 0.00044298951856019204,
      "loss": 2.0772,
      "step": 78765
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.8128367066383362,
      "learning_rate": 0.00044298592252116527,
      "loss": 2.1427,
      "step": 78766
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.761330783367157,
      "learning_rate": 0.0004429823264555548,
      "loss": 2.2345,
      "step": 78767
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.7923224568367004,
      "learning_rate": 0.00044297873036336117,
      "loss": 2.1738,
      "step": 78768
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.9144835472106934,
      "learning_rate": 0.00044297513424458514,
      "loss": 2.2707,
      "step": 78769
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.7394267916679382,
      "learning_rate": 0.0004429715380992273,
      "loss": 2.1385,
      "step": 78770
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.8770231008529663,
      "learning_rate": 0.00044296794192728834,
      "loss": 2.2453,
      "step": 78771
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.8297748565673828,
      "learning_rate": 0.000442964345728769,
      "loss": 2.1973,
      "step": 78772
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.8388184309005737,
      "learning_rate": 0.0004429607495036698,
      "loss": 2.5166,
      "step": 78773
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.7701875567436218,
      "learning_rate": 0.00044295715325199156,
      "loss": 2.1114,
      "step": 78774
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.7666060924530029,
      "learning_rate": 0.0004429535569737349,
      "loss": 2.1946,
      "step": 78775
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.7231277227401733,
      "learning_rate": 0.00044294996066890043,
      "loss": 2.2254,
      "step": 78776
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.8663849830627441,
      "learning_rate": 0.0004429463643374889,
      "loss": 2.1872,
      "step": 78777
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.7636088728904724,
      "learning_rate": 0.00044294276797950095,
      "loss": 2.1093,
      "step": 78778
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.8119226098060608,
      "learning_rate": 0.0004429391715949372,
      "loss": 2.1633,
      "step": 78779
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.9236435294151306,
      "learning_rate": 0.00044293557518379836,
      "loss": 2.1982,
      "step": 78780
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.8256147503852844,
      "learning_rate": 0.00044293197874608513,
      "loss": 2.1561,
      "step": 78781
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.7673780918121338,
      "learning_rate": 0.00044292838228179816,
      "loss": 1.9679,
      "step": 78782
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.8304007053375244,
      "learning_rate": 0.00044292478579093805,
      "loss": 2.3728,
      "step": 78783
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.8855598568916321,
      "learning_rate": 0.0004429211892735056,
      "loss": 2.2808,
      "step": 78784
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.9832087755203247,
      "learning_rate": 0.00044291759272950136,
      "loss": 2.2368,
      "step": 78785
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.7943662405014038,
      "learning_rate": 0.000442913996158926,
      "loss": 2.2175,
      "step": 78786
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.845527172088623,
      "learning_rate": 0.0004429103995617803,
      "loss": 2.2756,
      "step": 78787
    },
    {
      "epoch": 1.03,
      "grad_norm": 1.1559783220291138,
      "learning_rate": 0.0004429068029380648,
      "loss": 1.9867,
      "step": 78788
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.9695578217506409,
      "learning_rate": 0.00044290320628778023,
      "loss": 2.1571,
      "step": 78789
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.7918870449066162,
      "learning_rate": 0.0004428996096109273,
      "loss": 2.2051,
      "step": 78790
    },
    {
      "epoch": 1.03,
      "grad_norm": 1.0989400148391724,
      "learning_rate": 0.00044289601290750663,
      "loss": 2.1789,
      "step": 78791
    },
    {
      "epoch": 1.03,
      "grad_norm": 1.3228861093521118,
      "learning_rate": 0.0004428924161775189,
      "loss": 2.1513,
      "step": 78792
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.7475229501724243,
      "learning_rate": 0.0004428888194209647,
      "loss": 2.0141,
      "step": 78793
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.8476448655128479,
      "learning_rate": 0.00044288522263784486,
      "loss": 2.3583,
      "step": 78794
    },
    {
      "epoch": 1.03,
      "grad_norm": 1.0272963047027588,
      "learning_rate": 0.0004428816258281599,
      "loss": 1.9881,
      "step": 78795
    },
    {
      "epoch": 1.03,
      "grad_norm": 1.0296446084976196,
      "learning_rate": 0.00044287802899191055,
      "loss": 2.4119,
      "step": 78796
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.7938966751098633,
      "learning_rate": 0.0004428744321290975,
      "loss": 2.2816,
      "step": 78797
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.8715950846672058,
      "learning_rate": 0.0004428708352397214,
      "loss": 2.4899,
      "step": 78798
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.9155330657958984,
      "learning_rate": 0.0004428672383237829,
      "loss": 2.2761,
      "step": 78799
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.9079892635345459,
      "learning_rate": 0.0004428636413812827,
      "loss": 2.2558,
      "step": 78800
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.8491436839103699,
      "learning_rate": 0.00044286004441222146,
      "loss": 2.1427,
      "step": 78801
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.8547040820121765,
      "learning_rate": 0.0004428564474165998,
      "loss": 2.1965,
      "step": 78802
    },
    {
      "epoch": 1.03,
      "grad_norm": 1.2880339622497559,
      "learning_rate": 0.0004428528503944185,
      "loss": 2.0287,
      "step": 78803
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.9373118877410889,
      "learning_rate": 0.00044284925334567803,
      "loss": 1.8445,
      "step": 78804
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.7307907342910767,
      "learning_rate": 0.0004428456562703794,
      "loss": 2.1262,
      "step": 78805
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.8905211091041565,
      "learning_rate": 0.0004428420591685228,
      "loss": 2.1403,
      "step": 78806
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.9160532355308533,
      "learning_rate": 0.0004428384620401094,
      "loss": 2.0857,
      "step": 78807
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.8328091502189636,
      "learning_rate": 0.00044283486488513955,
      "loss": 2.2837,
      "step": 78808
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.7301727533340454,
      "learning_rate": 0.00044283126770361397,
      "loss": 2.2132,
      "step": 78809
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.7754898071289062,
      "learning_rate": 0.00044282767049553336,
      "loss": 2.1483,
      "step": 78810
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.8669469952583313,
      "learning_rate": 0.00044282407326089846,
      "loss": 2.2922,
      "step": 78811
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.754370391368866,
      "learning_rate": 0.0004428204759997098,
      "loss": 2.1955,
      "step": 78812
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.7634820342063904,
      "learning_rate": 0.0004428168787119681,
      "loss": 2.1648,
      "step": 78813
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.7621064782142639,
      "learning_rate": 0.0004428132813976741,
      "loss": 2.1621,
      "step": 78814
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.7590438723564148,
      "learning_rate": 0.0004428096840568284,
      "loss": 2.4459,
      "step": 78815
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.7907596826553345,
      "learning_rate": 0.0004428060866894317,
      "loss": 1.9894,
      "step": 78816
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.8609508275985718,
      "learning_rate": 0.0004428024892954847,
      "loss": 2.3003,
      "step": 78817
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.8410623073577881,
      "learning_rate": 0.0004427988918749879,
      "loss": 1.9613,
      "step": 78818
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.7377291321754456,
      "learning_rate": 0.00044279529442794216,
      "loss": 2.2425,
      "step": 78819
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.8598899245262146,
      "learning_rate": 0.0004427916969543481,
      "loss": 2.3097,
      "step": 78820
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.8771468997001648,
      "learning_rate": 0.0004427880994542063,
      "loss": 2.2185,
      "step": 78821
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.7327508330345154,
      "learning_rate": 0.00044278450192751756,
      "loss": 2.1485,
      "step": 78822
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.80393385887146,
      "learning_rate": 0.0004427809043742825,
      "loss": 2.0974,
      "step": 78823
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.8470876216888428,
      "learning_rate": 0.0004427773067945017,
      "loss": 2.2848,
      "step": 78824
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.9075031876564026,
      "learning_rate": 0.0004427737091881759,
      "loss": 2.4008,
      "step": 78825
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.76808100938797,
      "learning_rate": 0.00044277011155530584,
      "loss": 2.4152,
      "step": 78826
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.7488869428634644,
      "learning_rate": 0.00044276651389589213,
      "loss": 2.1566,
      "step": 78827
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.7893198728561401,
      "learning_rate": 0.00044276291620993537,
      "loss": 2.163,
      "step": 78828
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.8461072444915771,
      "learning_rate": 0.0004427593184974363,
      "loss": 2.4471,
      "step": 78829
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.7814324498176575,
      "learning_rate": 0.00044275572075839554,
      "loss": 2.1681,
      "step": 78830
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.7740560173988342,
      "learning_rate": 0.00044275212299281383,
      "loss": 2.2962,
      "step": 78831
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.7041734457015991,
      "learning_rate": 0.0004427485252006919,
      "loss": 2.1052,
      "step": 78832
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.9509177207946777,
      "learning_rate": 0.00044274492738203023,
      "loss": 2.0775,
      "step": 78833
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.8413024544715881,
      "learning_rate": 0.0004427413295368296,
      "loss": 2.0163,
      "step": 78834
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.9314940571784973,
      "learning_rate": 0.0004427377316650907,
      "loss": 2.2105,
      "step": 78835
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.7814379930496216,
      "learning_rate": 0.0004427341337668141,
      "loss": 2.2422,
      "step": 78836
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.7973542213439941,
      "learning_rate": 0.00044273053584200056,
      "loss": 2.3082,
      "step": 78837
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.7612360119819641,
      "learning_rate": 0.00044272693789065067,
      "loss": 2.1506,
      "step": 78838
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.8009665012359619,
      "learning_rate": 0.00044272333991276526,
      "loss": 2.2187,
      "step": 78839
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.7680224180221558,
      "learning_rate": 0.0004427197419083447,
      "loss": 2.0654,
      "step": 78840
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.7673079371452332,
      "learning_rate": 0.00044271614387739,
      "loss": 2.0319,
      "step": 78841
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.7706837058067322,
      "learning_rate": 0.00044271254581990165,
      "loss": 2.1429,
      "step": 78842
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.7402852773666382,
      "learning_rate": 0.0004427089477358803,
      "loss": 2.0183,
      "step": 78843
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.8368741869926453,
      "learning_rate": 0.0004427053496253267,
      "loss": 2.2428,
      "step": 78844
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.8800942301750183,
      "learning_rate": 0.0004427017514882415,
      "loss": 2.2394,
      "step": 78845
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.8105514049530029,
      "learning_rate": 0.00044269815332462534,
      "loss": 2.3266,
      "step": 78846
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.8525899052619934,
      "learning_rate": 0.00044269455513447885,
      "loss": 2.2686,
      "step": 78847
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.8044233322143555,
      "learning_rate": 0.00044269095691780284,
      "loss": 2.1452,
      "step": 78848
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.9221486449241638,
      "learning_rate": 0.0004426873586745978,
      "loss": 2.2104,
      "step": 78849
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.844653844833374,
      "learning_rate": 0.00044268376040486454,
      "loss": 2.288,
      "step": 78850
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.808709979057312,
      "learning_rate": 0.0004426801621086037,
      "loss": 2.3253,
      "step": 78851
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.7568434476852417,
      "learning_rate": 0.00044267656378581586,
      "loss": 2.4069,
      "step": 78852
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.7439082860946655,
      "learning_rate": 0.0004426729654365017,
      "loss": 1.9891,
      "step": 78853
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.7167071104049683,
      "learning_rate": 0.0004426693670606621,
      "loss": 2.1261,
      "step": 78854
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.8410671949386597,
      "learning_rate": 0.00044266576865829746,
      "loss": 2.1985,
      "step": 78855
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.7563462257385254,
      "learning_rate": 0.00044266217022940856,
      "loss": 2.1557,
      "step": 78856
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.7217128872871399,
      "learning_rate": 0.0004426585717739961,
      "loss": 2.0772,
      "step": 78857
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.8169920444488525,
      "learning_rate": 0.0004426549732920607,
      "loss": 2.1407,
      "step": 78858
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.790863037109375,
      "learning_rate": 0.00044265137478360315,
      "loss": 2.353,
      "step": 78859
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.7228012681007385,
      "learning_rate": 0.0004426477762486239,
      "loss": 2.1772,
      "step": 78860
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.7286965847015381,
      "learning_rate": 0.00044264417768712376,
      "loss": 1.9759,
      "step": 78861
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.9121026992797852,
      "learning_rate": 0.00044264057909910336,
      "loss": 2.068,
      "step": 78862
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.8392531871795654,
      "learning_rate": 0.0004426369804845635,
      "loss": 2.1479,
      "step": 78863
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.8838256597518921,
      "learning_rate": 0.0004426333818435045,
      "loss": 1.982,
      "step": 78864
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.7996924519538879,
      "learning_rate": 0.0004426297831759275,
      "loss": 2.3643,
      "step": 78865
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.9612321257591248,
      "learning_rate": 0.0004426261844818328,
      "loss": 2.0219,
      "step": 78866
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.9018203020095825,
      "learning_rate": 0.00044262258576122126,
      "loss": 2.2115,
      "step": 78867
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.9043044447898865,
      "learning_rate": 0.00044261898701409343,
      "loss": 2.1904,
      "step": 78868
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.8546860218048096,
      "learning_rate": 0.0004426153882404501,
      "loss": 2.1014,
      "step": 78869
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.9246872663497925,
      "learning_rate": 0.00044261178944029187,
      "loss": 2.2813,
      "step": 78870
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.8913024663925171,
      "learning_rate": 0.00044260819061361933,
      "loss": 2.1144,
      "step": 78871
    },
    {
      "epoch": 1.03,
      "grad_norm": 1.1019301414489746,
      "learning_rate": 0.00044260459176043333,
      "loss": 2.2137,
      "step": 78872
    },
    {
      "epoch": 1.03,
      "grad_norm": 1.1479078531265259,
      "learning_rate": 0.0004426009928807344,
      "loss": 2.3325,
      "step": 78873
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.8500048518180847,
      "learning_rate": 0.00044259739397452323,
      "loss": 2.2973,
      "step": 78874
    },
    {
      "epoch": 1.03,
      "grad_norm": 1.027360439300537,
      "learning_rate": 0.0004425937950418006,
      "loss": 1.999,
      "step": 78875
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.8836609721183777,
      "learning_rate": 0.00044259019608256704,
      "loss": 2.3982,
      "step": 78876
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.8338665962219238,
      "learning_rate": 0.0004425865970968232,
      "loss": 2.1486,
      "step": 78877
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.8010473251342773,
      "learning_rate": 0.0004425829980845699,
      "loss": 2.1897,
      "step": 78878
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.8746427297592163,
      "learning_rate": 0.00044257939904580776,
      "loss": 2.1738,
      "step": 78879
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.9847216010093689,
      "learning_rate": 0.00044257579998053734,
      "loss": 2.1323,
      "step": 78880
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.7343930006027222,
      "learning_rate": 0.0004425722008887594,
      "loss": 1.8084,
      "step": 78881
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.7797927856445312,
      "learning_rate": 0.0004425686017704746,
      "loss": 2.0442,
      "step": 78882
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.8386266827583313,
      "learning_rate": 0.0004425650026256837,
      "loss": 2.3224,
      "step": 78883
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.8313499689102173,
      "learning_rate": 0.0004425614034543872,
      "loss": 2.2473,
      "step": 78884
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.8454091548919678,
      "learning_rate": 0.0004425578042565858,
      "loss": 2.1236,
      "step": 78885
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.7770386934280396,
      "learning_rate": 0.0004425542050322803,
      "loss": 2.126,
      "step": 78886
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.7750194072723389,
      "learning_rate": 0.00044255060578147126,
      "loss": 2.1308,
      "step": 78887
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.8146536946296692,
      "learning_rate": 0.0004425470065041593,
      "loss": 2.2531,
      "step": 78888
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.7590404152870178,
      "learning_rate": 0.00044254340720034516,
      "loss": 2.1021,
      "step": 78889
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.7498980164527893,
      "learning_rate": 0.00044253980787002965,
      "loss": 2.2744,
      "step": 78890
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.8530909419059753,
      "learning_rate": 0.0004425362085132132,
      "loss": 2.291,
      "step": 78891
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.7499346733093262,
      "learning_rate": 0.00044253260912989657,
      "loss": 2.1194,
      "step": 78892
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.8296763300895691,
      "learning_rate": 0.0004425290097200805,
      "loss": 2.1909,
      "step": 78893
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.9288927316665649,
      "learning_rate": 0.0004425254102837655,
      "loss": 2.3503,
      "step": 78894
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.8970428109169006,
      "learning_rate": 0.00044252181082095234,
      "loss": 2.2133,
      "step": 78895
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.7670934796333313,
      "learning_rate": 0.00044251821133164184,
      "loss": 2.1259,
      "step": 78896
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.7956094741821289,
      "learning_rate": 0.0004425146118158344,
      "loss": 2.406,
      "step": 78897
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.7761801481246948,
      "learning_rate": 0.0004425110122735308,
      "loss": 2.1742,
      "step": 78898
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.7942386269569397,
      "learning_rate": 0.00044250741270473174,
      "loss": 2.0795,
      "step": 78899
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.8329116106033325,
      "learning_rate": 0.00044250381310943787,
      "loss": 2.2931,
      "step": 78900
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.7810513377189636,
      "learning_rate": 0.00044250021348764984,
      "loss": 2.2402,
      "step": 78901
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.7225229740142822,
      "learning_rate": 0.0004424966138393683,
      "loss": 2.1394,
      "step": 78902
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.8120853304862976,
      "learning_rate": 0.000442493014164594,
      "loss": 2.2664,
      "step": 78903
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.783416211605072,
      "learning_rate": 0.00044248941446332744,
      "loss": 1.9818,
      "step": 78904
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.8352811336517334,
      "learning_rate": 0.0004424858147355695,
      "loss": 2.2116,
      "step": 78905
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.8612643480300903,
      "learning_rate": 0.00044248221498132084,
      "loss": 2.0898,
      "step": 78906
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.8213250041007996,
      "learning_rate": 0.00044247861520058194,
      "loss": 2.1657,
      "step": 78907
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.7765406966209412,
      "learning_rate": 0.00044247501539335354,
      "loss": 2.3512,
      "step": 78908
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.8965455889701843,
      "learning_rate": 0.00044247141555963645,
      "loss": 2.2419,
      "step": 78909
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.7600369453430176,
      "learning_rate": 0.00044246781569943117,
      "loss": 1.959,
      "step": 78910
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.7559424638748169,
      "learning_rate": 0.00044246421581273845,
      "loss": 1.8884,
      "step": 78911
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.8725067377090454,
      "learning_rate": 0.00044246061589955894,
      "loss": 2.0922,
      "step": 78912
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.8185682892799377,
      "learning_rate": 0.0004424570159598933,
      "loss": 2.2483,
      "step": 78913
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.8346060514450073,
      "learning_rate": 0.0004424534159937422,
      "loss": 2.0783,
      "step": 78914
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.8641793727874756,
      "learning_rate": 0.0004424498160011064,
      "loss": 2.2091,
      "step": 78915
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.7897745966911316,
      "learning_rate": 0.00044244621598198646,
      "loss": 2.473,
      "step": 78916
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.7701669335365295,
      "learning_rate": 0.00044244261593638304,
      "loss": 2.2053,
      "step": 78917
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.7801620364189148,
      "learning_rate": 0.00044243901586429695,
      "loss": 2.1657,
      "step": 78918
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.742051899433136,
      "learning_rate": 0.0004424354157657286,
      "loss": 2.1382,
      "step": 78919
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.8179754614830017,
      "learning_rate": 0.0004424318156406789,
      "loss": 2.1398,
      "step": 78920
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.8089069128036499,
      "learning_rate": 0.0004424282154891484,
      "loss": 2.1512,
      "step": 78921
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.7576839327812195,
      "learning_rate": 0.0004424246153111379,
      "loss": 2.1295,
      "step": 78922
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.7960556745529175,
      "learning_rate": 0.0004424210151066479,
      "loss": 2.2314,
      "step": 78923
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.9249151349067688,
      "learning_rate": 0.00044241741487567913,
      "loss": 2.0902,
      "step": 78924
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.7941915988922119,
      "learning_rate": 0.0004424138146182323,
      "loss": 2.0532,
      "step": 78925
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.7746910452842712,
      "learning_rate": 0.000442410214334308,
      "loss": 2.3198,
      "step": 78926
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.76934814453125,
      "learning_rate": 0.0004424066140239071,
      "loss": 2.3382,
      "step": 78927
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.7391594648361206,
      "learning_rate": 0.00044240301368703,
      "loss": 2.3212,
      "step": 78928
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.7774457931518555,
      "learning_rate": 0.00044239941332367753,
      "loss": 2.0092,
      "step": 78929
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.7803553938865662,
      "learning_rate": 0.00044239581293385033,
      "loss": 1.9564,
      "step": 78930
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.7691451907157898,
      "learning_rate": 0.000442392212517549,
      "loss": 2.323,
      "step": 78931
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.787119448184967,
      "learning_rate": 0.0004423886120747743,
      "loss": 2.0593,
      "step": 78932
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.8214001059532166,
      "learning_rate": 0.000442385011605527,
      "loss": 2.322,
      "step": 78933
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.7554973363876343,
      "learning_rate": 0.0004423814111098075,
      "loss": 2.4369,
      "step": 78934
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.790343701839447,
      "learning_rate": 0.00044237781058761654,
      "loss": 2.072,
      "step": 78935
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.8358283042907715,
      "learning_rate": 0.00044237421003895497,
      "loss": 2.3724,
      "step": 78936
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.7643737196922302,
      "learning_rate": 0.00044237060946382336,
      "loss": 2.2803,
      "step": 78937
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.8295261263847351,
      "learning_rate": 0.00044236700886222225,
      "loss": 2.0851,
      "step": 78938
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.9189741015434265,
      "learning_rate": 0.00044236340823415257,
      "loss": 2.2898,
      "step": 78939
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.8192434906959534,
      "learning_rate": 0.00044235980757961475,
      "loss": 2.2504,
      "step": 78940
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.8366116285324097,
      "learning_rate": 0.0004423562068986096,
      "loss": 2.0586,
      "step": 78941
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.9672203063964844,
      "learning_rate": 0.0004423526061911378,
      "loss": 2.1707,
      "step": 78942
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.8830054998397827,
      "learning_rate": 0.0004423490054571998,
      "loss": 2.0009,
      "step": 78943
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.8478730320930481,
      "learning_rate": 0.0004423454046967965,
      "loss": 2.2097,
      "step": 78944
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.7356112003326416,
      "learning_rate": 0.00044234180390992866,
      "loss": 2.1533,
      "step": 78945
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.9961978793144226,
      "learning_rate": 0.0004423382030965966,
      "loss": 2.2986,
      "step": 78946
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.8161225318908691,
      "learning_rate": 0.0004423346022568012,
      "loss": 2.3087,
      "step": 78947
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.806825578212738,
      "learning_rate": 0.00044233100139054324,
      "loss": 2.2767,
      "step": 78948
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.6978856921195984,
      "learning_rate": 0.00044232740049782307,
      "loss": 2.0377,
      "step": 78949
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.908858597278595,
      "learning_rate": 0.00044232379957864174,
      "loss": 2.1386,
      "step": 78950
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.8020942807197571,
      "learning_rate": 0.00044232019863299965,
      "loss": 2.1908,
      "step": 78951
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.8160517811775208,
      "learning_rate": 0.0004423165976608975,
      "loss": 2.0737,
      "step": 78952
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.7916187644004822,
      "learning_rate": 0.00044231299666233604,
      "loss": 2.3381,
      "step": 78953
    },
    {
      "epoch": 1.03,
      "grad_norm": 1.2319347858428955,
      "learning_rate": 0.00044230939563731593,
      "loss": 2.1283,
      "step": 78954
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.8355454802513123,
      "learning_rate": 0.00044230579458583776,
      "loss": 2.3981,
      "step": 78955
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.8898559808731079,
      "learning_rate": 0.00044230219350790235,
      "loss": 2.0267,
      "step": 78956
    },
    {
      "epoch": 1.03,
      "grad_norm": 1.0550562143325806,
      "learning_rate": 0.00044229859240351024,
      "loss": 2.119,
      "step": 78957
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.8721641898155212,
      "learning_rate": 0.0004422949912726621,
      "loss": 2.2042,
      "step": 78958
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.7739684581756592,
      "learning_rate": 0.0004422913901153586,
      "loss": 2.1727,
      "step": 78959
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.9015849232673645,
      "learning_rate": 0.0004422877889316006,
      "loss": 2.0877,
      "step": 78960
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.7633043527603149,
      "learning_rate": 0.0004422841877213885,
      "loss": 2.355,
      "step": 78961
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.803835391998291,
      "learning_rate": 0.00044228058648472306,
      "loss": 2.249,
      "step": 78962
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.7859878540039062,
      "learning_rate": 0.00044227698522160495,
      "loss": 2.1761,
      "step": 78963
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.8339419960975647,
      "learning_rate": 0.000442273383932035,
      "loss": 2.2404,
      "step": 78964
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.7938612103462219,
      "learning_rate": 0.0004422697826160136,
      "loss": 2.3716,
      "step": 78965
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.7487518787384033,
      "learning_rate": 0.0004422661812735416,
      "loss": 2.0431,
      "step": 78966
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.74927818775177,
      "learning_rate": 0.00044226257990461973,
      "loss": 2.2983,
      "step": 78967
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.7237657904624939,
      "learning_rate": 0.0004422589785092484,
      "loss": 2.1908,
      "step": 78968
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.7722688317298889,
      "learning_rate": 0.00044225537708742857,
      "loss": 2.1099,
      "step": 78969
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.7832934260368347,
      "learning_rate": 0.0004422517756391607,
      "loss": 2.1872,
      "step": 78970
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.7355538606643677,
      "learning_rate": 0.0004422481741644457,
      "loss": 2.2488,
      "step": 78971
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.7401593923568726,
      "learning_rate": 0.00044224457266328393,
      "loss": 2.2407,
      "step": 78972
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.9070303440093994,
      "learning_rate": 0.00044224097113567616,
      "loss": 2.1866,
      "step": 78973
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.7496536374092102,
      "learning_rate": 0.00044223736958162323,
      "loss": 2.0333,
      "step": 78974
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.8369591236114502,
      "learning_rate": 0.0004422337680011257,
      "loss": 2.0761,
      "step": 78975
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.7913923859596252,
      "learning_rate": 0.00044223016639418413,
      "loss": 2.3789,
      "step": 78976
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.8171704411506653,
      "learning_rate": 0.00044222656476079926,
      "loss": 2.1913,
      "step": 78977
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.8423900604248047,
      "learning_rate": 0.00044222296310097195,
      "loss": 2.1756,
      "step": 78978
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.7040987014770508,
      "learning_rate": 0.00044221936141470257,
      "loss": 2.1199,
      "step": 78979
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.7515507340431213,
      "learning_rate": 0.00044221575970199194,
      "loss": 2.2755,
      "step": 78980
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.7562342286109924,
      "learning_rate": 0.0004422121579628407,
      "loss": 2.0422,
      "step": 78981
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.8280782699584961,
      "learning_rate": 0.00044220855619724964,
      "loss": 1.9303,
      "step": 78982
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.8268811106681824,
      "learning_rate": 0.0004422049544052192,
      "loss": 2.1781,
      "step": 78983
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.8379507064819336,
      "learning_rate": 0.0004422013525867503,
      "loss": 2.1527,
      "step": 78984
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.8352508544921875,
      "learning_rate": 0.00044219775074184344,
      "loss": 2.25,
      "step": 78985
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.90416419506073,
      "learning_rate": 0.0004421941488704993,
      "loss": 2.2733,
      "step": 78986
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.8179171681404114,
      "learning_rate": 0.00044219054697271855,
      "loss": 2.0576,
      "step": 78987
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.788826584815979,
      "learning_rate": 0.00044218694504850203,
      "loss": 2.1024,
      "step": 78988
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.7960185408592224,
      "learning_rate": 0.00044218334309785017,
      "loss": 2.1731,
      "step": 78989
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.8420130014419556,
      "learning_rate": 0.0004421797411207637,
      "loss": 2.1851,
      "step": 78990
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.8896353840827942,
      "learning_rate": 0.00044217613911724346,
      "loss": 2.2456,
      "step": 78991
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.9266694784164429,
      "learning_rate": 0.0004421725370872899,
      "loss": 2.1232,
      "step": 78992
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.919334352016449,
      "learning_rate": 0.0004421689350309038,
      "loss": 2.1662,
      "step": 78993
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.9007410407066345,
      "learning_rate": 0.00044216533294808583,
      "loss": 2.1237,
      "step": 78994
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.7718385457992554,
      "learning_rate": 0.0004421617308388366,
      "loss": 2.2714,
      "step": 78995
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.7992064952850342,
      "learning_rate": 0.00044215812870315686,
      "loss": 2.2906,
      "step": 78996
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.7876153588294983,
      "learning_rate": 0.0004421545265410473,
      "loss": 2.3351,
      "step": 78997
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.8927603960037231,
      "learning_rate": 0.0004421509243525084,
      "loss": 2.1292,
      "step": 78998
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.7498824000358582,
      "learning_rate": 0.000442147322137541,
      "loss": 2.0747,
      "step": 78999
    },
    {
      "epoch": 1.03,
      "grad_norm": 1.0488224029541016,
      "learning_rate": 0.0004421437198961458,
      "loss": 2.2328,
      "step": 79000
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.835955798625946,
      "learning_rate": 0.0004421401176283233,
      "loss": 2.125,
      "step": 79001
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.7949324250221252,
      "learning_rate": 0.00044213651533407433,
      "loss": 2.1491,
      "step": 79002
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.9693102240562439,
      "learning_rate": 0.0004421329130133995,
      "loss": 2.332,
      "step": 79003
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.8183751702308655,
      "learning_rate": 0.0004421293106662994,
      "loss": 2.1486,
      "step": 79004
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.7159096598625183,
      "learning_rate": 0.0004421257082927749,
      "loss": 2.2455,
      "step": 79005
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.8450494408607483,
      "learning_rate": 0.00044212210589282655,
      "loss": 2.0116,
      "step": 79006
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.7154526114463806,
      "learning_rate": 0.00044211850346645497,
      "loss": 2.1151,
      "step": 79007
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.7934126257896423,
      "learning_rate": 0.0004421149010136608,
      "loss": 2.3213,
      "step": 79008
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.8409054279327393,
      "learning_rate": 0.00044211129853444485,
      "loss": 2.1963,
      "step": 79009
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.822009265422821,
      "learning_rate": 0.00044210769602880774,
      "loss": 2.0962,
      "step": 79010
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.8437609672546387,
      "learning_rate": 0.00044210409349675013,
      "loss": 2.1664,
      "step": 79011
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.8096879720687866,
      "learning_rate": 0.00044210049093827274,
      "loss": 2.0984,
      "step": 79012
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.8313078880310059,
      "learning_rate": 0.00044209688835337605,
      "loss": 1.9298,
      "step": 79013
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.7571488618850708,
      "learning_rate": 0.00044209328574206094,
      "loss": 2.1038,
      "step": 79014
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.8092446327209473,
      "learning_rate": 0.000442089683104328,
      "loss": 2.372,
      "step": 79015
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.7898847460746765,
      "learning_rate": 0.0004420860804401779,
      "loss": 2.0967,
      "step": 79016
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.7949914336204529,
      "learning_rate": 0.0004420824777496113,
      "loss": 2.3584,
      "step": 79017
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.8609142303466797,
      "learning_rate": 0.000442078875032629,
      "loss": 2.425,
      "step": 79018
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.763932466506958,
      "learning_rate": 0.0004420752722892314,
      "loss": 2.0756,
      "step": 79019
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.7114589810371399,
      "learning_rate": 0.00044207166951941933,
      "loss": 2.127,
      "step": 79020
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.7654711604118347,
      "learning_rate": 0.0004420680667231936,
      "loss": 2.1684,
      "step": 79021
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.7367738485336304,
      "learning_rate": 0.0004420644639005546,
      "loss": 2.121,
      "step": 79022
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.7629454731941223,
      "learning_rate": 0.0004420608610515032,
      "loss": 2.2302,
      "step": 79023
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.7492538690567017,
      "learning_rate": 0.00044205725817604005,
      "loss": 2.033,
      "step": 79024
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.8042300343513489,
      "learning_rate": 0.00044205365527416574,
      "loss": 1.996,
      "step": 79025
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.8276073336601257,
      "learning_rate": 0.00044205005234588087,
      "loss": 2.1945,
      "step": 79026
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.8949844241142273,
      "learning_rate": 0.00044204644939118635,
      "loss": 2.4356,
      "step": 79027
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.8357945680618286,
      "learning_rate": 0.0004420428464100826,
      "loss": 2.0806,
      "step": 79028
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.7352533936500549,
      "learning_rate": 0.0004420392434025705,
      "loss": 2.101,
      "step": 79029
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.7234406471252441,
      "learning_rate": 0.0004420356403686506,
      "loss": 2.0823,
      "step": 79030
    },
    {
      "epoch": 1.03,
      "grad_norm": 1.0988264083862305,
      "learning_rate": 0.0004420320373083235,
      "loss": 2.3098,
      "step": 79031
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.7986722588539124,
      "learning_rate": 0.00044202843422159007,
      "loss": 2.068,
      "step": 79032
    },
    {
      "epoch": 1.03,
      "grad_norm": 1.0689712762832642,
      "learning_rate": 0.00044202483110845084,
      "loss": 2.1059,
      "step": 79033
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.8144651055335999,
      "learning_rate": 0.0004420212279689065,
      "loss": 2.1897,
      "step": 79034
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.8131524324417114,
      "learning_rate": 0.00044201762480295775,
      "loss": 2.2092,
      "step": 79035
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.7566618919372559,
      "learning_rate": 0.00044201402161060525,
      "loss": 2.4212,
      "step": 79036
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.8655158281326294,
      "learning_rate": 0.00044201041839184965,
      "loss": 2.1298,
      "step": 79037
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.8399201035499573,
      "learning_rate": 0.0004420068151466916,
      "loss": 2.3941,
      "step": 79038
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.7861366271972656,
      "learning_rate": 0.0004420032118751318,
      "loss": 2.2722,
      "step": 79039
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.7336422204971313,
      "learning_rate": 0.000441999608577171,
      "loss": 2.3167,
      "step": 79040
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.7372648119926453,
      "learning_rate": 0.0004419960052528098,
      "loss": 2.3271,
      "step": 79041
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.8361184597015381,
      "learning_rate": 0.00044199240190204874,
      "loss": 2.0794,
      "step": 79042
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.8053225874900818,
      "learning_rate": 0.0004419887985248887,
      "loss": 2.2152,
      "step": 79043
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.8390663862228394,
      "learning_rate": 0.0004419851951213302,
      "loss": 2.3616,
      "step": 79044
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.7977912425994873,
      "learning_rate": 0.0004419815916913741,
      "loss": 2.1761,
      "step": 79045
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.8189489245414734,
      "learning_rate": 0.0004419779882350209,
      "loss": 2.0502,
      "step": 79046
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.7913929224014282,
      "learning_rate": 0.00044197438475227126,
      "loss": 2.1913,
      "step": 79047
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.8771198391914368,
      "learning_rate": 0.0004419707812431259,
      "loss": 2.2393,
      "step": 79048
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.8243468403816223,
      "learning_rate": 0.00044196717770758555,
      "loss": 2.1075,
      "step": 79049
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.7603517174720764,
      "learning_rate": 0.0004419635741456507,
      "loss": 2.0903,
      "step": 79050
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.8721794486045837,
      "learning_rate": 0.00044195997055732224,
      "loss": 2.233,
      "step": 79051
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.8444475531578064,
      "learning_rate": 0.00044195636694260073,
      "loss": 2.0658,
      "step": 79052
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.7685227394104004,
      "learning_rate": 0.0004419527633014868,
      "loss": 2.1861,
      "step": 79053
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.7587360143661499,
      "learning_rate": 0.0004419491596339812,
      "loss": 2.2889,
      "step": 79054
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.7294251918792725,
      "learning_rate": 0.00044194555594008456,
      "loss": 2.1395,
      "step": 79055
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.7665219902992249,
      "learning_rate": 0.00044194195221979764,
      "loss": 2.1017,
      "step": 79056
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.733584463596344,
      "learning_rate": 0.000441938348473121,
      "loss": 2.1655,
      "step": 79057
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.8898313641548157,
      "learning_rate": 0.0004419347447000553,
      "loss": 2.0113,
      "step": 79058
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.7880719900131226,
      "learning_rate": 0.0004419311409006013,
      "loss": 2.3404,
      "step": 79059
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.8582777380943298,
      "learning_rate": 0.0004419275370747596,
      "loss": 2.1745,
      "step": 79060
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.8021497130393982,
      "learning_rate": 0.00044192393322253087,
      "loss": 2.0064,
      "step": 79061
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.8517976999282837,
      "learning_rate": 0.0004419203293439158,
      "loss": 2.1117,
      "step": 79062
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.7575505375862122,
      "learning_rate": 0.0004419167254389151,
      "loss": 1.9898,
      "step": 79063
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.8016351461410522,
      "learning_rate": 0.0004419131215075294,
      "loss": 2.3382,
      "step": 79064
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.824309229850769,
      "learning_rate": 0.0004419095175497593,
      "loss": 2.315,
      "step": 79065
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.7920992970466614,
      "learning_rate": 0.00044190591356560563,
      "loss": 2.1967,
      "step": 79066
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.8705924153327942,
      "learning_rate": 0.00044190230955506895,
      "loss": 2.0669,
      "step": 79067
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.7930375337600708,
      "learning_rate": 0.00044189870551814995,
      "loss": 2.474,
      "step": 79068
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.7975942492485046,
      "learning_rate": 0.0004418951014548493,
      "loss": 1.88,
      "step": 79069
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.8647525906562805,
      "learning_rate": 0.00044189149736516763,
      "loss": 2.2048,
      "step": 79070
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.7256508469581604,
      "learning_rate": 0.0004418878932491057,
      "loss": 2.1935,
      "step": 79071
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.7733324766159058,
      "learning_rate": 0.0004418842891066641,
      "loss": 2.2791,
      "step": 79072
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.8523275256156921,
      "learning_rate": 0.00044188068493784355,
      "loss": 2.2638,
      "step": 79073
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.8302233815193176,
      "learning_rate": 0.00044187708074264475,
      "loss": 2.2483,
      "step": 79074
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.8462350964546204,
      "learning_rate": 0.0004418734765210682,
      "loss": 2.269,
      "step": 79075
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.9297016859054565,
      "learning_rate": 0.00044186987227311486,
      "loss": 2.3858,
      "step": 79076
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.8079472780227661,
      "learning_rate": 0.0004418662679987851,
      "loss": 2.0766,
      "step": 79077
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.8187372088432312,
      "learning_rate": 0.0004418626636980798,
      "loss": 2.2332,
      "step": 79078
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.7724604606628418,
      "learning_rate": 0.0004418590593709995,
      "loss": 2.2315,
      "step": 79079
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.8041844964027405,
      "learning_rate": 0.0004418554550175449,
      "loss": 2.3772,
      "step": 79080
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.7550681233406067,
      "learning_rate": 0.0004418518506377168,
      "loss": 2.0802,
      "step": 79081
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.7852936387062073,
      "learning_rate": 0.0004418482462315157,
      "loss": 2.3056,
      "step": 79082
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.7737260460853577,
      "learning_rate": 0.0004418446417989423,
      "loss": 2.2484,
      "step": 79083
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.8568987846374512,
      "learning_rate": 0.00044184103733999736,
      "loss": 2.083,
      "step": 79084
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.7819032073020935,
      "learning_rate": 0.00044183743285468145,
      "loss": 2.1649,
      "step": 79085
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.8583467602729797,
      "learning_rate": 0.00044183382834299536,
      "loss": 2.2968,
      "step": 79086
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.8014987111091614,
      "learning_rate": 0.0004418302238049396,
      "loss": 1.9991,
      "step": 79087
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.7803465127944946,
      "learning_rate": 0.00044182661924051497,
      "loss": 2.1485,
      "step": 79088
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.7795513868331909,
      "learning_rate": 0.00044182301464972207,
      "loss": 2.1916,
      "step": 79089
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.8603595495223999,
      "learning_rate": 0.00044181941003256154,
      "loss": 2.3735,
      "step": 79090
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.8203957676887512,
      "learning_rate": 0.00044181580538903424,
      "loss": 1.9727,
      "step": 79091
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.8363445997238159,
      "learning_rate": 0.00044181220071914056,
      "loss": 2.2187,
      "step": 79092
    },
    {
      "epoch": 1.03,
      "grad_norm": 1.1530319452285767,
      "learning_rate": 0.0004418085960228814,
      "loss": 1.9603,
      "step": 79093
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.7915494441986084,
      "learning_rate": 0.0004418049913002574,
      "loss": 2.0801,
      "step": 79094
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.7991171479225159,
      "learning_rate": 0.00044180138655126904,
      "loss": 2.0654,
      "step": 79095
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.7759097814559937,
      "learning_rate": 0.0004417977817759172,
      "loss": 2.1117,
      "step": 79096
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.8181673288345337,
      "learning_rate": 0.0004417941769742025,
      "loss": 2.1771,
      "step": 79097
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.8609347343444824,
      "learning_rate": 0.0004417905721461255,
      "loss": 2.2933,
      "step": 79098
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.9240201115608215,
      "learning_rate": 0.00044178696729168696,
      "loss": 2.1742,
      "step": 79099
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.9351656436920166,
      "learning_rate": 0.0004417833624108877,
      "loss": 2.2681,
      "step": 79100
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.8964804410934448,
      "learning_rate": 0.0004417797575037281,
      "loss": 2.5193,
      "step": 79101
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.8190057277679443,
      "learning_rate": 0.00044177615257020897,
      "loss": 1.9844,
      "step": 79102
    },
    {
      "epoch": 1.03,
      "grad_norm": 1.0032429695129395,
      "learning_rate": 0.0004417725476103311,
      "loss": 2.2425,
      "step": 79103
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.7711754441261292,
      "learning_rate": 0.0004417689426240949,
      "loss": 2.1754,
      "step": 79104
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.8634940981864929,
      "learning_rate": 0.0004417653376115012,
      "loss": 2.1045,
      "step": 79105
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.8633030652999878,
      "learning_rate": 0.0004417617325725506,
      "loss": 2.1773,
      "step": 79106
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.9788082838058472,
      "learning_rate": 0.00044175812750724396,
      "loss": 2.2258,
      "step": 79107
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.8432194590568542,
      "learning_rate": 0.00044175452241558175,
      "loss": 2.2132,
      "step": 79108
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.8826442956924438,
      "learning_rate": 0.00044175091729756465,
      "loss": 2.3835,
      "step": 79109
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.9008092880249023,
      "learning_rate": 0.0004417473121531934,
      "loss": 2.1756,
      "step": 79110
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.816939115524292,
      "learning_rate": 0.00044174370698246864,
      "loss": 2.1518,
      "step": 79111
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.8653441071510315,
      "learning_rate": 0.00044174010178539113,
      "loss": 2.1855,
      "step": 79112
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.9405836462974548,
      "learning_rate": 0.0004417364965619614,
      "loss": 2.0874,
      "step": 79113
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.8381006717681885,
      "learning_rate": 0.00044173289131218017,
      "loss": 2.3697,
      "step": 79114
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.8834881782531738,
      "learning_rate": 0.0004417292860360481,
      "loss": 2.2928,
      "step": 79115
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.7935108542442322,
      "learning_rate": 0.00044172568073356594,
      "loss": 2.1711,
      "step": 79116
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.9820607304573059,
      "learning_rate": 0.0004417220754047343,
      "loss": 2.046,
      "step": 79117
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.8587228059768677,
      "learning_rate": 0.00044171847004955377,
      "loss": 2.5292,
      "step": 79118
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.8243144154548645,
      "learning_rate": 0.00044171486466802523,
      "loss": 1.9745,
      "step": 79119
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.7599542140960693,
      "learning_rate": 0.0004417112592601491,
      "loss": 2.242,
      "step": 79120
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.7744953632354736,
      "learning_rate": 0.0004417076538259262,
      "loss": 2.3265,
      "step": 79121
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.8754376173019409,
      "learning_rate": 0.0004417040483653572,
      "loss": 2.1512,
      "step": 79122
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.7917687296867371,
      "learning_rate": 0.00044170044287844274,
      "loss": 2.1861,
      "step": 79123
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.851751446723938,
      "learning_rate": 0.0004416968373651834,
      "loss": 2.2625,
      "step": 79124
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.8535935282707214,
      "learning_rate": 0.0004416932318255801,
      "loss": 2.374,
      "step": 79125
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.8563113808631897,
      "learning_rate": 0.00044168962625963323,
      "loss": 2.2907,
      "step": 79126
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.9069578051567078,
      "learning_rate": 0.0004416860206673437,
      "loss": 2.1638,
      "step": 79127
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.8562804460525513,
      "learning_rate": 0.000441682415048712,
      "loss": 2.3451,
      "step": 79128
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.928830087184906,
      "learning_rate": 0.00044167880940373884,
      "loss": 2.369,
      "step": 79129
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.8264987468719482,
      "learning_rate": 0.0004416752037324249,
      "loss": 2.089,
      "step": 79130
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.7945113182067871,
      "learning_rate": 0.00044167159803477094,
      "loss": 2.3328,
      "step": 79131
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.7913244962692261,
      "learning_rate": 0.0004416679923107775,
      "loss": 2.1785,
      "step": 79132
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.9506107568740845,
      "learning_rate": 0.0004416643865604453,
      "loss": 2.221,
      "step": 79133
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.8615473508834839,
      "learning_rate": 0.0004416607807837751,
      "loss": 2.2659,
      "step": 79134
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.725368857383728,
      "learning_rate": 0.00044165717498076743,
      "loss": 2.3248,
      "step": 79135
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.838782548904419,
      "learning_rate": 0.00044165356915142295,
      "loss": 2.2413,
      "step": 79136
    },
    {
      "epoch": 1.03,
      "grad_norm": 1.2283082008361816,
      "learning_rate": 0.0004416499632957425,
      "loss": 2.1559,
      "step": 79137
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.73566734790802,
      "learning_rate": 0.0004416463574137266,
      "loss": 2.216,
      "step": 79138
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.7725833654403687,
      "learning_rate": 0.00044164275150537595,
      "loss": 2.2497,
      "step": 79139
    },
    {
      "epoch": 1.03,
      "grad_norm": 1.3064055442810059,
      "learning_rate": 0.0004416391455706913,
      "loss": 2.0448,
      "step": 79140
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.8136278390884399,
      "learning_rate": 0.00044163553960967325,
      "loss": 2.1232,
      "step": 79141
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.8071033358573914,
      "learning_rate": 0.0004416319336223224,
      "loss": 2.024,
      "step": 79142
    },
    {
      "epoch": 1.03,
      "grad_norm": 1.0083836317062378,
      "learning_rate": 0.00044162832760863955,
      "loss": 2.2588,
      "step": 79143
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.8833470940589905,
      "learning_rate": 0.0004416247215686254,
      "loss": 2.2025,
      "step": 79144
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.9444186687469482,
      "learning_rate": 0.00044162111550228043,
      "loss": 2.0963,
      "step": 79145
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.8302100300788879,
      "learning_rate": 0.0004416175094096054,
      "loss": 2.1879,
      "step": 79146
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.8475621342658997,
      "learning_rate": 0.00044161390329060116,
      "loss": 2.1747,
      "step": 79147
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.8082083463668823,
      "learning_rate": 0.0004416102971452681,
      "loss": 2.1689,
      "step": 79148
    },
    {
      "epoch": 1.03,
      "grad_norm": 1.129304051399231,
      "learning_rate": 0.000441606690973607,
      "loss": 2.1131,
      "step": 79149
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.8061252236366272,
      "learning_rate": 0.00044160308477561854,
      "loss": 2.3551,
      "step": 79150
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.8634265065193176,
      "learning_rate": 0.00044159947855130344,
      "loss": 2.081,
      "step": 79151
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.864544153213501,
      "learning_rate": 0.00044159587230066234,
      "loss": 2.4009,
      "step": 79152
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.87649005651474,
      "learning_rate": 0.0004415922660236959,
      "loss": 2.2444,
      "step": 79153
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.8823384046554565,
      "learning_rate": 0.0004415886597204047,
      "loss": 2.1388,
      "step": 79154
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.7645264863967896,
      "learning_rate": 0.0004415850533907896,
      "loss": 2.0112,
      "step": 79155
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.8823555111885071,
      "learning_rate": 0.0004415814470348511,
      "loss": 2.2368,
      "step": 79156
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.8575718998908997,
      "learning_rate": 0.00044157784065258993,
      "loss": 2.524,
      "step": 79157
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.8067934513092041,
      "learning_rate": 0.00044157423424400687,
      "loss": 2.2274,
      "step": 79158
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.7633563280105591,
      "learning_rate": 0.00044157062780910236,
      "loss": 2.2775,
      "step": 79159
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.7763811349868774,
      "learning_rate": 0.0004415670213478772,
      "loss": 2.1878,
      "step": 79160
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.7707641124725342,
      "learning_rate": 0.0004415634148603322,
      "loss": 2.2522,
      "step": 79161
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.8978622555732727,
      "learning_rate": 0.00044155980834646773,
      "loss": 2.4965,
      "step": 79162
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.7830721735954285,
      "learning_rate": 0.0004415562018062847,
      "loss": 2.2547,
      "step": 79163
    },
    {
      "epoch": 1.03,
      "grad_norm": 1.2594324350357056,
      "learning_rate": 0.0004415525952397837,
      "loss": 2.0807,
      "step": 79164
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.7446175813674927,
      "learning_rate": 0.0004415489886469654,
      "loss": 2.2724,
      "step": 79165
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.7797126173973083,
      "learning_rate": 0.00044154538202783043,
      "loss": 2.3688,
      "step": 79166
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.9088096618652344,
      "learning_rate": 0.00044154177538237955,
      "loss": 2.2176,
      "step": 79167
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.800830602645874,
      "learning_rate": 0.0004415381687106133,
      "loss": 2.2229,
      "step": 79168
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.7570172548294067,
      "learning_rate": 0.0004415345620125325,
      "loss": 2.2007,
      "step": 79169
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.780113697052002,
      "learning_rate": 0.0004415309552881377,
      "loss": 2.1501,
      "step": 79170
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.8505576848983765,
      "learning_rate": 0.0004415273485374297,
      "loss": 2.2506,
      "step": 79171
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.835454523563385,
      "learning_rate": 0.0004415237417604091,
      "loss": 2.3823,
      "step": 79172
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.8770018815994263,
      "learning_rate": 0.0004415201349570765,
      "loss": 2.1138,
      "step": 79173
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.7638134956359863,
      "learning_rate": 0.00044151652812743266,
      "loss": 2.169,
      "step": 79174
    },
    {
      "epoch": 1.03,
      "grad_norm": 1.1646276712417603,
      "learning_rate": 0.00044151292127147824,
      "loss": 1.9897,
      "step": 79175
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.8273289203643799,
      "learning_rate": 0.0004415093143892139,
      "loss": 1.8538,
      "step": 79176
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.7452828288078308,
      "learning_rate": 0.0004415057074806403,
      "loss": 2.3019,
      "step": 79177
    },
    {
      "epoch": 1.03,
      "grad_norm": 1.3740743398666382,
      "learning_rate": 0.0004415021005457581,
      "loss": 2.2822,
      "step": 79178
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.8295105695724487,
      "learning_rate": 0.000441498493584568,
      "loss": 2.3501,
      "step": 79179
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.8738281726837158,
      "learning_rate": 0.0004414948865970707,
      "loss": 2.3694,
      "step": 79180
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.7074270248413086,
      "learning_rate": 0.0004414912795832667,
      "loss": 2.3391,
      "step": 79181
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.8400766253471375,
      "learning_rate": 0.0004414876725431569,
      "loss": 2.3182,
      "step": 79182
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.8885472416877747,
      "learning_rate": 0.0004414840654767419,
      "loss": 1.917,
      "step": 79183
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.9935095310211182,
      "learning_rate": 0.0004414804583840223,
      "loss": 2.0624,
      "step": 79184
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.8221617937088013,
      "learning_rate": 0.00044147685126499876,
      "loss": 2.2094,
      "step": 79185
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.9290573596954346,
      "learning_rate": 0.00044147324411967213,
      "loss": 2.1234,
      "step": 79186
    },
    {
      "epoch": 1.03,
      "grad_norm": 1.1980626583099365,
      "learning_rate": 0.00044146963694804286,
      "loss": 2.0788,
      "step": 79187
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.8950650095939636,
      "learning_rate": 0.0004414660297501117,
      "loss": 2.0398,
      "step": 79188
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.8514062166213989,
      "learning_rate": 0.0004414624225258794,
      "loss": 2.2888,
      "step": 79189
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.9377554059028625,
      "learning_rate": 0.0004414588152753465,
      "loss": 2.2238,
      "step": 79190
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.8115143775939941,
      "learning_rate": 0.00044145520799851375,
      "loss": 2.1869,
      "step": 79191
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.9272423386573792,
      "learning_rate": 0.0004414516006953819,
      "loss": 2.1903,
      "step": 79192
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.8295770287513733,
      "learning_rate": 0.0004414479933659514,
      "loss": 2.3808,
      "step": 79193
    },
    {
      "epoch": 1.03,
      "grad_norm": 1.0301182270050049,
      "learning_rate": 0.0004414443860102231,
      "loss": 2.3502,
      "step": 79194
    },
    {
      "epoch": 1.03,
      "grad_norm": 1.1442819833755493,
      "learning_rate": 0.00044144077862819765,
      "loss": 2.2142,
      "step": 79195
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.8702272772789001,
      "learning_rate": 0.0004414371712198756,
      "loss": 2.0821,
      "step": 79196
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.8048452734947205,
      "learning_rate": 0.0004414335637852578,
      "loss": 1.9899,
      "step": 79197
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.8999256491661072,
      "learning_rate": 0.0004414299563243448,
      "loss": 2.536,
      "step": 79198
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.9565010666847229,
      "learning_rate": 0.0004414263488371373,
      "loss": 2.2363,
      "step": 79199
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.732468843460083,
      "learning_rate": 0.000441422741323636,
      "loss": 2.3525,
      "step": 79200
    },
    {
      "epoch": 1.03,
      "grad_norm": 1.0243088006973267,
      "learning_rate": 0.00044141913378384156,
      "loss": 2.3032,
      "step": 79201
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.9372281432151794,
      "learning_rate": 0.0004414155262177545,
      "loss": 2.3601,
      "step": 79202
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.7621018290519714,
      "learning_rate": 0.00044141191862537575,
      "loss": 2.3375,
      "step": 79203
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.8001683354377747,
      "learning_rate": 0.00044140831100670586,
      "loss": 2.2015,
      "step": 79204
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.9879375696182251,
      "learning_rate": 0.00044140470336174537,
      "loss": 2.0996,
      "step": 79205
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.8104962706565857,
      "learning_rate": 0.0004414010956904952,
      "loss": 2.0851,
      "step": 79206
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.8265653848648071,
      "learning_rate": 0.0004413974879929559,
      "loss": 2.1003,
      "step": 79207
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.8813159465789795,
      "learning_rate": 0.00044139388026912804,
      "loss": 2.0761,
      "step": 79208
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.7920615077018738,
      "learning_rate": 0.00044139027251901246,
      "loss": 2.216,
      "step": 79209
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.7064152359962463,
      "learning_rate": 0.0004413866647426098,
      "loss": 2.2612,
      "step": 79210
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.763122022151947,
      "learning_rate": 0.00044138305693992055,
      "loss": 2.23,
      "step": 79211
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.9724963307380676,
      "learning_rate": 0.0004413794491109456,
      "loss": 2.2224,
      "step": 79212
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.8702330589294434,
      "learning_rate": 0.0004413758412556856,
      "loss": 2.1481,
      "step": 79213
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.8417003154754639,
      "learning_rate": 0.0004413722333741411,
      "loss": 2.1257,
      "step": 79214
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.8646700978279114,
      "learning_rate": 0.00044136862546631284,
      "loss": 2.3471,
      "step": 79215
    },
    {
      "epoch": 1.03,
      "grad_norm": 1.0754632949829102,
      "learning_rate": 0.0004413650175322015,
      "loss": 2.0589,
      "step": 79216
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.8962284326553345,
      "learning_rate": 0.0004413614095718077,
      "loss": 2.1731,
      "step": 79217
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.8203389644622803,
      "learning_rate": 0.00044135780158513217,
      "loss": 2.2936,
      "step": 79218
    },
    {
      "epoch": 1.03,
      "grad_norm": 1.004137396812439,
      "learning_rate": 0.00044135419357217557,
      "loss": 2.2422,
      "step": 79219
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.8294067978858948,
      "learning_rate": 0.00044135058553293856,
      "loss": 2.2198,
      "step": 79220
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.8588331341743469,
      "learning_rate": 0.00044134697746742174,
      "loss": 2.2347,
      "step": 79221
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.8631807565689087,
      "learning_rate": 0.00044134336937562597,
      "loss": 2.2175,
      "step": 79222
    },
    {
      "epoch": 1.03,
      "grad_norm": 1.1835992336273193,
      "learning_rate": 0.0004413397612575517,
      "loss": 2.3263,
      "step": 79223
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.7636248469352722,
      "learning_rate": 0.00044133615311319965,
      "loss": 2.1546,
      "step": 79224
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.8038866519927979,
      "learning_rate": 0.00044133254494257073,
      "loss": 2.3134,
      "step": 79225
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.8195881843566895,
      "learning_rate": 0.00044132893674566525,
      "loss": 2.079,
      "step": 79226
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.8211708068847656,
      "learning_rate": 0.0004413253285224841,
      "loss": 2.2631,
      "step": 79227
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.8102672696113586,
      "learning_rate": 0.00044132172027302796,
      "loss": 2.1752,
      "step": 79228
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.7956477999687195,
      "learning_rate": 0.0004413181119972974,
      "loss": 2.2356,
      "step": 79229
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.771430253982544,
      "learning_rate": 0.0004413145036952931,
      "loss": 2.0755,
      "step": 79230
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.7966206669807434,
      "learning_rate": 0.00044131089536701576,
      "loss": 2.3509,
      "step": 79231
    },
    {
      "epoch": 1.03,
      "grad_norm": 1.1054353713989258,
      "learning_rate": 0.00044130728701246614,
      "loss": 1.9575,
      "step": 79232
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.8707669973373413,
      "learning_rate": 0.00044130367863164484,
      "loss": 2.1941,
      "step": 79233
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.7780942916870117,
      "learning_rate": 0.0004413000702245524,
      "loss": 2.2756,
      "step": 79234
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.7662883400917053,
      "learning_rate": 0.00044129646179118964,
      "loss": 2.0553,
      "step": 79235
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.7406877279281616,
      "learning_rate": 0.0004412928533315573,
      "loss": 2.1544,
      "step": 79236
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.8135545253753662,
      "learning_rate": 0.0004412892448456559,
      "loss": 2.1302,
      "step": 79237
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.8166896104812622,
      "learning_rate": 0.00044128563633348615,
      "loss": 2.2955,
      "step": 79238
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.8482689261436462,
      "learning_rate": 0.00044128202779504874,
      "loss": 2.0931,
      "step": 79239
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.8275604248046875,
      "learning_rate": 0.00044127841923034426,
      "loss": 1.9648,
      "step": 79240
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.7858157157897949,
      "learning_rate": 0.00044127481063937357,
      "loss": 2.1566,
      "step": 79241
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.8205106854438782,
      "learning_rate": 0.00044127120202213716,
      "loss": 2.2028,
      "step": 79242
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.8110701441764832,
      "learning_rate": 0.0004412675933786358,
      "loss": 2.4088,
      "step": 79243
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.8889648914337158,
      "learning_rate": 0.00044126398470887006,
      "loss": 2.2569,
      "step": 79244
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.7778943777084351,
      "learning_rate": 0.0004412603760128407,
      "loss": 2.2972,
      "step": 79245
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.8282451033592224,
      "learning_rate": 0.0004412567672905484,
      "loss": 2.1989,
      "step": 79246
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.7147579193115234,
      "learning_rate": 0.0004412531585419939,
      "loss": 2.1432,
      "step": 79247
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.7750597596168518,
      "learning_rate": 0.00044124954976717764,
      "loss": 2.0658,
      "step": 79248
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.8383216261863708,
      "learning_rate": 0.0004412459409661004,
      "loss": 2.2826,
      "step": 79249
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.875063955783844,
      "learning_rate": 0.00044124233213876305,
      "loss": 2.2852,
      "step": 79250
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.7271925210952759,
      "learning_rate": 0.00044123872328516586,
      "loss": 2.1024,
      "step": 79251
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.7625763416290283,
      "learning_rate": 0.0004412351144053099,
      "loss": 2.3483,
      "step": 79252
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.7814648747444153,
      "learning_rate": 0.00044123150549919555,
      "loss": 2.1092,
      "step": 79253
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.7659905552864075,
      "learning_rate": 0.00044122789656682364,
      "loss": 2.0514,
      "step": 79254
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.8303877711296082,
      "learning_rate": 0.0004412242876081948,
      "loss": 2.2489,
      "step": 79255
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.8340625762939453,
      "learning_rate": 0.00044122067862330976,
      "loss": 2.0262,
      "step": 79256
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.8058758974075317,
      "learning_rate": 0.000441217069612169,
      "loss": 2.5311,
      "step": 79257
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.804500937461853,
      "learning_rate": 0.0004412134605747734,
      "loss": 2.113,
      "step": 79258
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.7909157872200012,
      "learning_rate": 0.0004412098515111236,
      "loss": 2.4355,
      "step": 79259
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.8895577192306519,
      "learning_rate": 0.0004412062424212201,
      "loss": 2.1029,
      "step": 79260
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.8506378531455994,
      "learning_rate": 0.00044120263330506374,
      "loss": 2.0132,
      "step": 79261
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.9032843708992004,
      "learning_rate": 0.0004411990241626552,
      "loss": 2.1599,
      "step": 79262
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.7549187541007996,
      "learning_rate": 0.0004411954149939951,
      "loss": 2.0802,
      "step": 79263
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.8804437518119812,
      "learning_rate": 0.00044119180579908405,
      "loss": 2.2349,
      "step": 79264
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.9298784732818604,
      "learning_rate": 0.0004411881965779228,
      "loss": 2.4155,
      "step": 79265
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.8077636957168579,
      "learning_rate": 0.000441184587330512,
      "loss": 2.218,
      "step": 79266
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.8168834447860718,
      "learning_rate": 0.0004411809780568523,
      "loss": 1.9524,
      "step": 79267
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.9494683742523193,
      "learning_rate": 0.0004411773687569444,
      "loss": 2.3556,
      "step": 79268
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.7893847227096558,
      "learning_rate": 0.00044117375943078903,
      "loss": 2.1686,
      "step": 79269
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.8729090690612793,
      "learning_rate": 0.00044117015007838663,
      "loss": 2.3554,
      "step": 79270
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.8332735896110535,
      "learning_rate": 0.00044116654069973824,
      "loss": 2.2228,
      "step": 79271
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.8205195069313049,
      "learning_rate": 0.0004411629312948442,
      "loss": 2.401,
      "step": 79272
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.7462003827095032,
      "learning_rate": 0.0004411593218637053,
      "loss": 2.0383,
      "step": 79273
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.8360676765441895,
      "learning_rate": 0.0004411557124063223,
      "loss": 2.3196,
      "step": 79274
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.9326549768447876,
      "learning_rate": 0.0004411521029226957,
      "loss": 2.4683,
      "step": 79275
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.7848418951034546,
      "learning_rate": 0.00044114849341282626,
      "loss": 2.0005,
      "step": 79276
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.8372923731803894,
      "learning_rate": 0.00044114488387671474,
      "loss": 2.1767,
      "step": 79277
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.8219894170761108,
      "learning_rate": 0.00044114127431436163,
      "loss": 2.3619,
      "step": 79278
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.759894609451294,
      "learning_rate": 0.00044113766472576773,
      "loss": 2.2488,
      "step": 79279
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.7736185193061829,
      "learning_rate": 0.0004411340551109337,
      "loss": 2.0666,
      "step": 79280
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.7871042490005493,
      "learning_rate": 0.00044113044546986014,
      "loss": 2.1575,
      "step": 79281
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.7629780173301697,
      "learning_rate": 0.00044112683580254774,
      "loss": 2.2782,
      "step": 79282
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.7881419062614441,
      "learning_rate": 0.0004411232261089973,
      "loss": 2.2262,
      "step": 79283
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.7572587132453918,
      "learning_rate": 0.00044111961638920933,
      "loss": 2.0996,
      "step": 79284
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.8589150309562683,
      "learning_rate": 0.0004411160066431845,
      "loss": 2.1673,
      "step": 79285
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.8299195170402527,
      "learning_rate": 0.00044111239687092367,
      "loss": 2.3615,
      "step": 79286
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.7769266963005066,
      "learning_rate": 0.0004411087870724273,
      "loss": 2.0848,
      "step": 79287
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.7949736714363098,
      "learning_rate": 0.0004411051772476962,
      "loss": 2.0838,
      "step": 79288
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.8334306478500366,
      "learning_rate": 0.00044110156739673095,
      "loss": 2.0842,
      "step": 79289
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.9229564070701599,
      "learning_rate": 0.0004410979575195322,
      "loss": 2.3288,
      "step": 79290
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.8161500692367554,
      "learning_rate": 0.0004410943476161007,
      "loss": 2.3916,
      "step": 79291
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.8383600115776062,
      "learning_rate": 0.00044109073768643716,
      "loss": 2.2205,
      "step": 79292
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.7911431789398193,
      "learning_rate": 0.0004410871277305421,
      "loss": 2.2218,
      "step": 79293
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.8678605556488037,
      "learning_rate": 0.0004410835177484163,
      "loss": 2.2515,
      "step": 79294
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.8450860977172852,
      "learning_rate": 0.0004410799077400605,
      "loss": 2.1946,
      "step": 79295
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.8291307091712952,
      "learning_rate": 0.00044107629770547517,
      "loss": 2.1381,
      "step": 79296
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.8133359551429749,
      "learning_rate": 0.0004410726876446612,
      "loss": 2.1664,
      "step": 79297
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.7792816758155823,
      "learning_rate": 0.00044106907755761906,
      "loss": 1.9277,
      "step": 79298
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.7285677194595337,
      "learning_rate": 0.00044106546744434954,
      "loss": 2.0958,
      "step": 79299
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.7921255826950073,
      "learning_rate": 0.0004410618573048533,
      "loss": 2.0152,
      "step": 79300
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.8255139589309692,
      "learning_rate": 0.000441058247139131,
      "loss": 2.2781,
      "step": 79301
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.8278146386146545,
      "learning_rate": 0.0004410546369471834,
      "loss": 2.068,
      "step": 79302
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.8536576628684998,
      "learning_rate": 0.0004410510267290109,
      "loss": 2.2751,
      "step": 79303
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.7630842328071594,
      "learning_rate": 0.00044104741648461445,
      "loss": 2.219,
      "step": 79304
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.7587828040122986,
      "learning_rate": 0.00044104380621399466,
      "loss": 2.1129,
      "step": 79305
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.7877808213233948,
      "learning_rate": 0.00044104019591715204,
      "loss": 1.9367,
      "step": 79306
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.8834513425827026,
      "learning_rate": 0.0004410365855940875,
      "loss": 2.0488,
      "step": 79307
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.8067874312400818,
      "learning_rate": 0.00044103297524480153,
      "loss": 2.2335,
      "step": 79308
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.7148118615150452,
      "learning_rate": 0.00044102936486929494,
      "loss": 2.0712,
      "step": 79309
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.8261374831199646,
      "learning_rate": 0.0004410257544675682,
      "loss": 2.2122,
      "step": 79310
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.7984055280685425,
      "learning_rate": 0.00044102214403962225,
      "loss": 2.1697,
      "step": 79311
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.7751327753067017,
      "learning_rate": 0.00044101853358545755,
      "loss": 2.1286,
      "step": 79312
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.8824631571769714,
      "learning_rate": 0.0004410149231050749,
      "loss": 2.2215,
      "step": 79313
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.9564207792282104,
      "learning_rate": 0.0004410113125984749,
      "loss": 2.153,
      "step": 79314
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.9187484383583069,
      "learning_rate": 0.00044100770206565815,
      "loss": 2.042,
      "step": 79315
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.8986249566078186,
      "learning_rate": 0.00044100409150662546,
      "loss": 2.1082,
      "step": 79316
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.7488246560096741,
      "learning_rate": 0.00044100048092137747,
      "loss": 2.0805,
      "step": 79317
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.9044004678726196,
      "learning_rate": 0.00044099687030991477,
      "loss": 2.2629,
      "step": 79318
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.8813203573226929,
      "learning_rate": 0.0004409932596722382,
      "loss": 2.1059,
      "step": 79319
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.8375321626663208,
      "learning_rate": 0.00044098964900834823,
      "loss": 1.97,
      "step": 79320
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.8516660928726196,
      "learning_rate": 0.00044098603831824564,
      "loss": 2.1386,
      "step": 79321
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.8721872568130493,
      "learning_rate": 0.0004409824276019311,
      "loss": 2.0798,
      "step": 79322
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.9085763692855835,
      "learning_rate": 0.00044097881685940517,
      "loss": 2.2448,
      "step": 79323
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.8116710782051086,
      "learning_rate": 0.0004409752060906688,
      "loss": 2.3057,
      "step": 79324
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.8005064129829407,
      "learning_rate": 0.00044097159529572235,
      "loss": 2.1374,
      "step": 79325
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.7580575346946716,
      "learning_rate": 0.0004409679844745666,
      "loss": 2.1103,
      "step": 79326
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.8283680081367493,
      "learning_rate": 0.00044096437362720227,
      "loss": 2.405,
      "step": 79327
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.7961687445640564,
      "learning_rate": 0.00044096076275363007,
      "loss": 2.0813,
      "step": 79328
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.8234811425209045,
      "learning_rate": 0.0004409571518538506,
      "loss": 2.1387,
      "step": 79329
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.808343768119812,
      "learning_rate": 0.0004409535409278644,
      "loss": 2.0389,
      "step": 79330
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.8633535504341125,
      "learning_rate": 0.0004409499299756724,
      "loss": 2.4388,
      "step": 79331
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.8600400686264038,
      "learning_rate": 0.0004409463189972751,
      "loss": 2.3244,
      "step": 79332
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.7889790534973145,
      "learning_rate": 0.0004409427079926732,
      "loss": 2.3615,
      "step": 79333
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.7367764115333557,
      "learning_rate": 0.0004409390969618674,
      "loss": 2.051,
      "step": 79334
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.9133780002593994,
      "learning_rate": 0.0004409354859048585,
      "loss": 2.4816,
      "step": 79335
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.7616884112358093,
      "learning_rate": 0.0004409318748216468,
      "loss": 2.0716,
      "step": 79336
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.7878782749176025,
      "learning_rate": 0.00044092826371223336,
      "loss": 2.1422,
      "step": 79337
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.7870609164237976,
      "learning_rate": 0.00044092465257661865,
      "loss": 2.0766,
      "step": 79338
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.8609892725944519,
      "learning_rate": 0.00044092104141480343,
      "loss": 2.1576,
      "step": 79339
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.8450344204902649,
      "learning_rate": 0.0004409174302267882,
      "loss": 2.2388,
      "step": 79340
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.9094862937927246,
      "learning_rate": 0.00044091381901257393,
      "loss": 2.3563,
      "step": 79341
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.7747776508331299,
      "learning_rate": 0.00044091020777216105,
      "loss": 2.1043,
      "step": 79342
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.7666812539100647,
      "learning_rate": 0.0004409065965055502,
      "loss": 2.1811,
      "step": 79343
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.8249146342277527,
      "learning_rate": 0.00044090298521274235,
      "loss": 1.9629,
      "step": 79344
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.8396989703178406,
      "learning_rate": 0.0004408993738937379,
      "loss": 2.1958,
      "step": 79345
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.8333134055137634,
      "learning_rate": 0.0004408957625485375,
      "loss": 2.1663,
      "step": 79346
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.7485714554786682,
      "learning_rate": 0.00044089215117714206,
      "loss": 2.1467,
      "step": 79347
    },
    {
      "epoch": 1.03,
      "grad_norm": 1.1369097232818604,
      "learning_rate": 0.000440888539779552,
      "loss": 2.1324,
      "step": 79348
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.8326330780982971,
      "learning_rate": 0.00044088492835576813,
      "loss": 2.0254,
      "step": 79349
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.8605303764343262,
      "learning_rate": 0.0004408813169057911,
      "loss": 2.1387,
      "step": 79350
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.9272040128707886,
      "learning_rate": 0.0004408777054296215,
      "loss": 2.2264,
      "step": 79351
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.8448635935783386,
      "learning_rate": 0.00044087409392726024,
      "loss": 2.1992,
      "step": 79352
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.9119443297386169,
      "learning_rate": 0.00044087048239870773,
      "loss": 1.9727,
      "step": 79353
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.8869538307189941,
      "learning_rate": 0.00044086687084396475,
      "loss": 2.1834,
      "step": 79354
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.7924138903617859,
      "learning_rate": 0.0004408632592630319,
      "loss": 2.0437,
      "step": 79355
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.793418288230896,
      "learning_rate": 0.00044085964765591,
      "loss": 2.2299,
      "step": 79356
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.8002640604972839,
      "learning_rate": 0.00044085603602259956,
      "loss": 2.2028,
      "step": 79357
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.7614452838897705,
      "learning_rate": 0.0004408524243631014,
      "loss": 1.9108,
      "step": 79358
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.7626593708992004,
      "learning_rate": 0.0004408488126774161,
      "loss": 2.1455,
      "step": 79359
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.790614664554596,
      "learning_rate": 0.00044084520096554427,
      "loss": 2.3511,
      "step": 79360
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.9150019884109497,
      "learning_rate": 0.0004408415892274867,
      "loss": 2.261,
      "step": 79361
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.9991540312767029,
      "learning_rate": 0.0004408379774632441,
      "loss": 2.2112,
      "step": 79362
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.8472063541412354,
      "learning_rate": 0.000440834365672817,
      "loss": 1.9452,
      "step": 79363
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.8049129247665405,
      "learning_rate": 0.0004408307538562061,
      "loss": 2.2797,
      "step": 79364
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.87335205078125,
      "learning_rate": 0.0004408271420134122,
      "loss": 2.293,
      "step": 79365
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.8381035923957825,
      "learning_rate": 0.00044082353014443575,
      "loss": 2.0684,
      "step": 79366
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.7925013303756714,
      "learning_rate": 0.00044081991824927757,
      "loss": 2.0146,
      "step": 79367
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.8229392170906067,
      "learning_rate": 0.00044081630632793845,
      "loss": 2.1029,
      "step": 79368
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.8019244074821472,
      "learning_rate": 0.00044081269438041873,
      "loss": 2.0982,
      "step": 79369
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.7927300333976746,
      "learning_rate": 0.0004408090824067194,
      "loss": 2.2555,
      "step": 79370
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.7833498120307922,
      "learning_rate": 0.00044080547040684087,
      "loss": 2.379,
      "step": 79371
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.7744497656822205,
      "learning_rate": 0.0004408018583807841,
      "loss": 1.9905,
      "step": 79372
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.8009539246559143,
      "learning_rate": 0.0004407982463285495,
      "loss": 2.2277,
      "step": 79373
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.7622404098510742,
      "learning_rate": 0.00044079463425013786,
      "loss": 2.2158,
      "step": 79374
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.8358532786369324,
      "learning_rate": 0.00044079102214555,
      "loss": 2.038,
      "step": 79375
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.7990978956222534,
      "learning_rate": 0.0004407874100147862,
      "loss": 2.2197,
      "step": 79376
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.7102000713348389,
      "learning_rate": 0.00044078379785784754,
      "loss": 2.0168,
      "step": 79377
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.8181464672088623,
      "learning_rate": 0.0004407801856747344,
      "loss": 2.3867,
      "step": 79378
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.7993168830871582,
      "learning_rate": 0.00044077657346544764,
      "loss": 2.0797,
      "step": 79379
    },
    {
      "epoch": 1.03,
      "grad_norm": 1.0775558948516846,
      "learning_rate": 0.0004407729612299878,
      "loss": 2.1954,
      "step": 79380
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.9297069311141968,
      "learning_rate": 0.00044076934896835566,
      "loss": 2.1995,
      "step": 79381
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.8351582288742065,
      "learning_rate": 0.00044076573668055176,
      "loss": 2.2434,
      "step": 79382
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.8062068223953247,
      "learning_rate": 0.00044076212436657697,
      "loss": 2.2781,
      "step": 79383
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.9797440767288208,
      "learning_rate": 0.00044075851202643175,
      "loss": 2.2023,
      "step": 79384
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.8240264654159546,
      "learning_rate": 0.0004407548996601169,
      "loss": 2.2887,
      "step": 79385
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.8949225544929504,
      "learning_rate": 0.000440751287267633,
      "loss": 2.0913,
      "step": 79386
    },
    {
      "epoch": 1.03,
      "grad_norm": 1.0779675245285034,
      "learning_rate": 0.0004407476748489809,
      "loss": 2.1849,
      "step": 79387
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.9343320727348328,
      "learning_rate": 0.00044074406240416107,
      "loss": 2.159,
      "step": 79388
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.9465702176094055,
      "learning_rate": 0.0004407404499331742,
      "loss": 2.2793,
      "step": 79389
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.7840967774391174,
      "learning_rate": 0.0004407368374360212,
      "loss": 2.1115,
      "step": 79390
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.9515214562416077,
      "learning_rate": 0.00044073322491270246,
      "loss": 2.1921,
      "step": 79391
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.8699567317962646,
      "learning_rate": 0.00044072961236321873,
      "loss": 2.4514,
      "step": 79392
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.8926196694374084,
      "learning_rate": 0.00044072599978757076,
      "loss": 2.3275,
      "step": 79393
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.8991623520851135,
      "learning_rate": 0.0004407223871857591,
      "loss": 2.105,
      "step": 79394
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.8094444870948792,
      "learning_rate": 0.00044071877455778455,
      "loss": 2.2183,
      "step": 79395
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.8634270429611206,
      "learning_rate": 0.00044071516190364777,
      "loss": 2.1724,
      "step": 79396
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.7184420228004456,
      "learning_rate": 0.00044071154922334935,
      "loss": 1.9123,
      "step": 79397
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.7934455871582031,
      "learning_rate": 0.0004407079365168899,
      "loss": 2.067,
      "step": 79398
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.7481910586357117,
      "learning_rate": 0.00044070432378427027,
      "loss": 2.0428,
      "step": 79399
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.8035377860069275,
      "learning_rate": 0.00044070071102549107,
      "loss": 1.9458,
      "step": 79400
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.8726822137832642,
      "learning_rate": 0.00044069709824055294,
      "loss": 2.3927,
      "step": 79401
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.806188702583313,
      "learning_rate": 0.0004406934854294566,
      "loss": 2.1957,
      "step": 79402
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.8896917700767517,
      "learning_rate": 0.0004406898725922026,
      "loss": 2.2198,
      "step": 79403
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.7113597393035889,
      "learning_rate": 0.0004406862597287917,
      "loss": 2.2165,
      "step": 79404
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.8522381782531738,
      "learning_rate": 0.00044068264683922467,
      "loss": 2.1363,
      "step": 79405
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.8447113633155823,
      "learning_rate": 0.00044067903392350195,
      "loss": 2.1602,
      "step": 79406
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.8158107399940491,
      "learning_rate": 0.00044067542098162436,
      "loss": 2.1123,
      "step": 79407
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.7486547231674194,
      "learning_rate": 0.00044067180801359257,
      "loss": 2.1926,
      "step": 79408
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.8602713346481323,
      "learning_rate": 0.0004406681950194073,
      "loss": 2.2823,
      "step": 79409
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.7895419001579285,
      "learning_rate": 0.00044066458199906917,
      "loss": 2.1697,
      "step": 79410
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.7294042110443115,
      "learning_rate": 0.00044066096895257876,
      "loss": 2.3237,
      "step": 79411
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.8757612705230713,
      "learning_rate": 0.0004406573558799369,
      "loss": 2.2265,
      "step": 79412
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.8116385340690613,
      "learning_rate": 0.00044065374278114404,
      "loss": 2.3235,
      "step": 79413
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.8495818972587585,
      "learning_rate": 0.0004406501296562011,
      "loss": 2.0029,
      "step": 79414
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.7992054224014282,
      "learning_rate": 0.00044064651650510864,
      "loss": 2.2313,
      "step": 79415
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.7899221777915955,
      "learning_rate": 0.00044064290332786736,
      "loss": 2.0653,
      "step": 79416
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.804848313331604,
      "learning_rate": 0.0004406392901244778,
      "loss": 2.0936,
      "step": 79417
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.731824517250061,
      "learning_rate": 0.00044063567689494085,
      "loss": 2.0478,
      "step": 79418
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.8380990028381348,
      "learning_rate": 0.00044063206363925705,
      "loss": 2.2354,
      "step": 79419
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.7870455980300903,
      "learning_rate": 0.0004406284503574271,
      "loss": 2.1936,
      "step": 79420
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.9565867781639099,
      "learning_rate": 0.00044062483704945164,
      "loss": 2.058,
      "step": 79421
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.8162545561790466,
      "learning_rate": 0.0004406212237153314,
      "loss": 2.1424,
      "step": 79422
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.8084999918937683,
      "learning_rate": 0.00044061761035506697,
      "loss": 2.0821,
      "step": 79423
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.7339343428611755,
      "learning_rate": 0.0004406139969686591,
      "loss": 2.2786,
      "step": 79424
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.8303924202919006,
      "learning_rate": 0.00044061038355610845,
      "loss": 2.1709,
      "step": 79425
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.8121386766433716,
      "learning_rate": 0.00044060677011741566,
      "loss": 2.0631,
      "step": 79426
    },
    {
      "epoch": 1.03,
      "grad_norm": 1.0446487665176392,
      "learning_rate": 0.00044060315665258145,
      "loss": 2.2571,
      "step": 79427
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.8289212584495544,
      "learning_rate": 0.0004405995431616064,
      "loss": 2.1105,
      "step": 79428
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.8058458566665649,
      "learning_rate": 0.0004405959296444913,
      "loss": 1.9982,
      "step": 79429
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.7583276629447937,
      "learning_rate": 0.0004405923161012367,
      "loss": 2.2738,
      "step": 79430
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.7627438306808472,
      "learning_rate": 0.00044058870253184336,
      "loss": 2.2145,
      "step": 79431
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.772239625453949,
      "learning_rate": 0.000440585088936312,
      "loss": 1.9971,
      "step": 79432
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.7587475180625916,
      "learning_rate": 0.0004405814753146431,
      "loss": 2.1325,
      "step": 79433
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.7975339293479919,
      "learning_rate": 0.00044057786166683754,
      "loss": 2.1915,
      "step": 79434
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.7910429835319519,
      "learning_rate": 0.00044057424799289586,
      "loss": 2.2961,
      "step": 79435
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.7833048701286316,
      "learning_rate": 0.0004405706342928188,
      "loss": 2.1307,
      "step": 79436
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.8243424296379089,
      "learning_rate": 0.00044056702056660696,
      "loss": 2.4522,
      "step": 79437
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.7529969215393066,
      "learning_rate": 0.00044056340681426103,
      "loss": 2.0569,
      "step": 79438
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.7902595400810242,
      "learning_rate": 0.0004405597930357818,
      "loss": 2.3774,
      "step": 79439
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.9846835136413574,
      "learning_rate": 0.0004405561792311698,
      "loss": 2.2055,
      "step": 79440
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.8130331635475159,
      "learning_rate": 0.0004405525654004258,
      "loss": 2.4192,
      "step": 79441
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.7845461368560791,
      "learning_rate": 0.0004405489515435504,
      "loss": 2.1446,
      "step": 79442
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.7844610810279846,
      "learning_rate": 0.0004405453376605443,
      "loss": 2.1561,
      "step": 79443
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.7695479989051819,
      "learning_rate": 0.00044054172375140806,
      "loss": 2.3737,
      "step": 79444
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.828596293926239,
      "learning_rate": 0.00044053810981614265,
      "loss": 2.3286,
      "step": 79445
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.8615438938140869,
      "learning_rate": 0.00044053449585474845,
      "loss": 2.3025,
      "step": 79446
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.7796332836151123,
      "learning_rate": 0.0004405308818672262,
      "loss": 2.1793,
      "step": 79447
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.7903176546096802,
      "learning_rate": 0.0004405272678535767,
      "loss": 2.2192,
      "step": 79448
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.8305240869522095,
      "learning_rate": 0.0004405236538138005,
      "loss": 2.3329,
      "step": 79449
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.8519036769866943,
      "learning_rate": 0.00044052003974789824,
      "loss": 2.3051,
      "step": 79450
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.7955142855644226,
      "learning_rate": 0.00044051642565587075,
      "loss": 2.4651,
      "step": 79451
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.8056617379188538,
      "learning_rate": 0.00044051281153771854,
      "loss": 2.1644,
      "step": 79452
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.8349096775054932,
      "learning_rate": 0.00044050919739344226,
      "loss": 2.2297,
      "step": 79453
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.8525751233100891,
      "learning_rate": 0.00044050558322304283,
      "loss": 2.1173,
      "step": 79454
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.7700703144073486,
      "learning_rate": 0.00044050196902652064,
      "loss": 2.2133,
      "step": 79455
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.8479354381561279,
      "learning_rate": 0.00044049835480387654,
      "loss": 2.1376,
      "step": 79456
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.737761378288269,
      "learning_rate": 0.00044049474055511115,
      "loss": 2.0977,
      "step": 79457
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.8443329334259033,
      "learning_rate": 0.0004404911262802251,
      "loss": 2.2184,
      "step": 79458
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.8569327592849731,
      "learning_rate": 0.00044048751197921915,
      "loss": 2.3873,
      "step": 79459
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.8770474195480347,
      "learning_rate": 0.0004404838976520939,
      "loss": 2.3038,
      "step": 79460
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.7996718287467957,
      "learning_rate": 0.00044048028329884997,
      "loss": 2.1435,
      "step": 79461
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.9000048637390137,
      "learning_rate": 0.0004404766689194882,
      "loss": 1.9226,
      "step": 79462
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.8752184510231018,
      "learning_rate": 0.0004404730545140091,
      "loss": 2.0677,
      "step": 79463
    },
    {
      "epoch": 1.03,
      "grad_norm": 1.0850640535354614,
      "learning_rate": 0.00044046944008241346,
      "loss": 2.102,
      "step": 79464
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.9237703680992126,
      "learning_rate": 0.00044046582562470187,
      "loss": 1.9867,
      "step": 79465
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.9224083423614502,
      "learning_rate": 0.0004404622111408751,
      "loss": 2.1314,
      "step": 79466
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.8739879131317139,
      "learning_rate": 0.00044045859663093364,
      "loss": 2.1524,
      "step": 79467
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.7911018133163452,
      "learning_rate": 0.00044045498209487825,
      "loss": 2.3031,
      "step": 79468
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.8205759525299072,
      "learning_rate": 0.00044045136753270974,
      "loss": 2.0904,
      "step": 79469
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.8083993196487427,
      "learning_rate": 0.00044044775294442865,
      "loss": 2.1126,
      "step": 79470
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.8107585906982422,
      "learning_rate": 0.00044044413833003564,
      "loss": 2.1599,
      "step": 79471
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.8232700824737549,
      "learning_rate": 0.00044044052368953147,
      "loss": 2.2279,
      "step": 79472
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.800598680973053,
      "learning_rate": 0.0004404369090229167,
      "loss": 2.2466,
      "step": 79473
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.7999443411827087,
      "learning_rate": 0.00044043329433019206,
      "loss": 2.1795,
      "step": 79474
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.768787682056427,
      "learning_rate": 0.0004404296796113583,
      "loss": 1.9908,
      "step": 79475
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.8178178668022156,
      "learning_rate": 0.0004404260648664159,
      "loss": 2.1889,
      "step": 79476
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.811652421951294,
      "learning_rate": 0.0004404224500953656,
      "loss": 2.2387,
      "step": 79477
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.7917392253875732,
      "learning_rate": 0.0004404188352982083,
      "loss": 2.2463,
      "step": 79478
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.7354503273963928,
      "learning_rate": 0.00044041522047494434,
      "loss": 2.3643,
      "step": 79479
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.8190224170684814,
      "learning_rate": 0.0004404116056255746,
      "loss": 2.0967,
      "step": 79480
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.9911036491394043,
      "learning_rate": 0.0004404079907500997,
      "loss": 2.2036,
      "step": 79481
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.8202639222145081,
      "learning_rate": 0.0004404043758485203,
      "loss": 2.1544,
      "step": 79482
    },
    {
      "epoch": 1.03,
      "grad_norm": 1.0141983032226562,
      "learning_rate": 0.000440400760920837,
      "loss": 2.2365,
      "step": 79483
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.8078396916389465,
      "learning_rate": 0.00044039714596705065,
      "loss": 2.3305,
      "step": 79484
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.7315192222595215,
      "learning_rate": 0.00044039353098716176,
      "loss": 1.9947,
      "step": 79485
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.7845988273620605,
      "learning_rate": 0.0004403899159811711,
      "loss": 2.0024,
      "step": 79486
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.8413194417953491,
      "learning_rate": 0.0004403863009490793,
      "loss": 2.1342,
      "step": 79487
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.760534942150116,
      "learning_rate": 0.000440382685890887,
      "loss": 2.1111,
      "step": 79488
    },
    {
      "epoch": 1.03,
      "grad_norm": 0.8054656982421875,
      "learning_rate": 0.00044037907080659493,
      "loss": 2.1786,
      "step": 79489
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.8253822922706604,
      "learning_rate": 0.00044037545569620376,
      "loss": 2.349,
      "step": 79490
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.7211458086967468,
      "learning_rate": 0.0004403718405597141,
      "loss": 2.1178,
      "step": 79491
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.7846989631652832,
      "learning_rate": 0.0004403682253971267,
      "loss": 2.2055,
      "step": 79492
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.8961333632469177,
      "learning_rate": 0.0004403646102084422,
      "loss": 2.1091,
      "step": 79493
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.8262784481048584,
      "learning_rate": 0.0004403609949936613,
      "loss": 2.0849,
      "step": 79494
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.8264715671539307,
      "learning_rate": 0.00044035737975278456,
      "loss": 2.4878,
      "step": 79495
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.8140231966972351,
      "learning_rate": 0.00044035376448581275,
      "loss": 2.412,
      "step": 79496
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.7152602672576904,
      "learning_rate": 0.0004403501491927466,
      "loss": 2.2246,
      "step": 79497
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.8839004039764404,
      "learning_rate": 0.00044034653387358666,
      "loss": 2.2178,
      "step": 79498
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.7986621856689453,
      "learning_rate": 0.0004403429185283336,
      "loss": 2.1763,
      "step": 79499
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.7981534600257874,
      "learning_rate": 0.0004403393031569882,
      "loss": 2.2629,
      "step": 79500
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.7765498757362366,
      "learning_rate": 0.0004403356877595511,
      "loss": 2.1594,
      "step": 79501
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.709426999092102,
      "learning_rate": 0.0004403320723360229,
      "loss": 2.1059,
      "step": 79502
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.7931139469146729,
      "learning_rate": 0.00044032845688640434,
      "loss": 2.2249,
      "step": 79503
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.8397909998893738,
      "learning_rate": 0.000440324841410696,
      "loss": 2.2508,
      "step": 79504
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.7773914337158203,
      "learning_rate": 0.0004403212259088987,
      "loss": 2.1904,
      "step": 79505
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.7191546559333801,
      "learning_rate": 0.000440317610381013,
      "loss": 2.1407,
      "step": 79506
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.921373724937439,
      "learning_rate": 0.0004403139948270396,
      "loss": 2.2267,
      "step": 79507
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.7581279873847961,
      "learning_rate": 0.00044031037924697927,
      "loss": 2.3329,
      "step": 79508
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.8388197422027588,
      "learning_rate": 0.0004403067636408325,
      "loss": 2.1691,
      "step": 79509
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.8233943581581116,
      "learning_rate": 0.0004403031480086001,
      "loss": 2.2963,
      "step": 79510
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.7994958162307739,
      "learning_rate": 0.0004402995323502827,
      "loss": 2.3598,
      "step": 79511
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.850974440574646,
      "learning_rate": 0.00044029591666588095,
      "loss": 2.2374,
      "step": 79512
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.8115201592445374,
      "learning_rate": 0.00044029230095539543,
      "loss": 1.9398,
      "step": 79513
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.7885469198226929,
      "learning_rate": 0.00044028868521882704,
      "loss": 2.1137,
      "step": 79514
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.8000528812408447,
      "learning_rate": 0.0004402850694561764,
      "loss": 2.1156,
      "step": 79515
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.7602425217628479,
      "learning_rate": 0.000440281453667444,
      "loss": 2.286,
      "step": 79516
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.7905449867248535,
      "learning_rate": 0.0004402778378526307,
      "loss": 2.3409,
      "step": 79517
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.7891169786453247,
      "learning_rate": 0.00044027422201173706,
      "loss": 2.1711,
      "step": 79518
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.824189305305481,
      "learning_rate": 0.0004402706061447638,
      "loss": 2.1526,
      "step": 79519
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.7744571566581726,
      "learning_rate": 0.00044026699025171155,
      "loss": 1.8791,
      "step": 79520
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.8622518181800842,
      "learning_rate": 0.0004402633743325812,
      "loss": 2.1552,
      "step": 79521
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.7939761877059937,
      "learning_rate": 0.00044025975838737306,
      "loss": 2.2219,
      "step": 79522
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.8962007761001587,
      "learning_rate": 0.000440256142416088,
      "loss": 2.0997,
      "step": 79523
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.7557119727134705,
      "learning_rate": 0.0004402525264187267,
      "loss": 2.2436,
      "step": 79524
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.8281572461128235,
      "learning_rate": 0.0004402489103952898,
      "loss": 2.2134,
      "step": 79525
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.76546311378479,
      "learning_rate": 0.00044024529434577803,
      "loss": 2.1594,
      "step": 79526
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.7750830054283142,
      "learning_rate": 0.000440241678270192,
      "loss": 2.0106,
      "step": 79527
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.8158960342407227,
      "learning_rate": 0.00044023806216853233,
      "loss": 2.3071,
      "step": 79528
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.8055676817893982,
      "learning_rate": 0.00044023444604079985,
      "loss": 2.2707,
      "step": 79529
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.8574807047843933,
      "learning_rate": 0.0004402308298869951,
      "loss": 2.1986,
      "step": 79530
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.889075756072998,
      "learning_rate": 0.00044022721370711875,
      "loss": 2.2998,
      "step": 79531
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.7890416383743286,
      "learning_rate": 0.00044022359750117154,
      "loss": 2.3663,
      "step": 79532
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.8446016907691956,
      "learning_rate": 0.00044021998126915414,
      "loss": 2.3167,
      "step": 79533
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.7858763933181763,
      "learning_rate": 0.00044021636501106716,
      "loss": 2.1387,
      "step": 79534
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.7733578681945801,
      "learning_rate": 0.0004402127487269113,
      "loss": 2.0221,
      "step": 79535
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.8090879917144775,
      "learning_rate": 0.0004402091324166873,
      "loss": 2.1265,
      "step": 79536
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.8205761313438416,
      "learning_rate": 0.00044020551608039575,
      "loss": 2.0178,
      "step": 79537
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.7719104886054993,
      "learning_rate": 0.0004402018997180373,
      "loss": 2.2834,
      "step": 79538
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.8017570376396179,
      "learning_rate": 0.0004401982833296128,
      "loss": 2.2738,
      "step": 79539
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.7239550352096558,
      "learning_rate": 0.00044019466691512266,
      "loss": 2.1785,
      "step": 79540
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.8044537305831909,
      "learning_rate": 0.0004401910504745677,
      "loss": 2.0325,
      "step": 79541
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.8597772717475891,
      "learning_rate": 0.00044018743400794864,
      "loss": 2.1779,
      "step": 79542
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.9443628787994385,
      "learning_rate": 0.00044018381751526606,
      "loss": 2.3535,
      "step": 79543
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.7186810970306396,
      "learning_rate": 0.0004401802009965207,
      "loss": 2.0135,
      "step": 79544
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.8415189385414124,
      "learning_rate": 0.0004401765844517131,
      "loss": 2.4125,
      "step": 79545
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.7662015557289124,
      "learning_rate": 0.0004401729678808441,
      "loss": 2.0947,
      "step": 79546
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.8782827854156494,
      "learning_rate": 0.00044016935128391425,
      "loss": 2.1114,
      "step": 79547
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.8685719966888428,
      "learning_rate": 0.0004401657346609244,
      "loss": 2.097,
      "step": 79548
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.9509384036064148,
      "learning_rate": 0.0004401621180118749,
      "loss": 2.2232,
      "step": 79549
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.9012616276741028,
      "learning_rate": 0.0004401585013367667,
      "loss": 2.0015,
      "step": 79550
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.7777513265609741,
      "learning_rate": 0.0004401548846356005,
      "loss": 2.323,
      "step": 79551
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.9634612202644348,
      "learning_rate": 0.00044015126790837675,
      "loss": 2.108,
      "step": 79552
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.7615570425987244,
      "learning_rate": 0.0004401476511550962,
      "loss": 1.8856,
      "step": 79553
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.8161001205444336,
      "learning_rate": 0.0004401440343757596,
      "loss": 2.3401,
      "step": 79554
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.8531844615936279,
      "learning_rate": 0.0004401404175703676,
      "loss": 2.432,
      "step": 79555
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.8102891445159912,
      "learning_rate": 0.00044013680073892085,
      "loss": 2.3251,
      "step": 79556
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.8938391208648682,
      "learning_rate": 0.00044013318388142,
      "loss": 2.0321,
      "step": 79557
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.9896040558815002,
      "learning_rate": 0.00044012956699786575,
      "loss": 2.1283,
      "step": 79558
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.7728851437568665,
      "learning_rate": 0.0004401259500882588,
      "loss": 2.5286,
      "step": 79559
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.7694491147994995,
      "learning_rate": 0.0004401223331525998,
      "loss": 2.3519,
      "step": 79560
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.770903468132019,
      "learning_rate": 0.00044011871619088934,
      "loss": 2.234,
      "step": 79561
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.7961414456367493,
      "learning_rate": 0.00044011509920312817,
      "loss": 2.288,
      "step": 79562
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.770231306552887,
      "learning_rate": 0.00044011148218931704,
      "loss": 2.169,
      "step": 79563
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.7399477958679199,
      "learning_rate": 0.0004401078651494565,
      "loss": 2.0172,
      "step": 79564
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.8119140267372131,
      "learning_rate": 0.0004401042480835473,
      "loss": 2.276,
      "step": 79565
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.7056649327278137,
      "learning_rate": 0.00044010063099159005,
      "loss": 2.0448,
      "step": 79566
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.809578537940979,
      "learning_rate": 0.00044009701387358536,
      "loss": 2.219,
      "step": 79567
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.8341301083564758,
      "learning_rate": 0.000440093396729534,
      "loss": 2.196,
      "step": 79568
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.8446825742721558,
      "learning_rate": 0.00044008977955943676,
      "loss": 2.0317,
      "step": 79569
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.8117923736572266,
      "learning_rate": 0.0004400861623632941,
      "loss": 2.1057,
      "step": 79570
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.9754396080970764,
      "learning_rate": 0.0004400825451411068,
      "loss": 2.1631,
      "step": 79571
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.784945011138916,
      "learning_rate": 0.00044007892789287543,
      "loss": 2.0841,
      "step": 79572
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.8770259022712708,
      "learning_rate": 0.00044007531061860094,
      "loss": 2.3454,
      "step": 79573
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.8065667152404785,
      "learning_rate": 0.00044007169331828364,
      "loss": 2.1603,
      "step": 79574
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.7542877793312073,
      "learning_rate": 0.0004400680759919244,
      "loss": 2.1928,
      "step": 79575
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.8429005146026611,
      "learning_rate": 0.0004400644586395239,
      "loss": 2.0861,
      "step": 79576
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.807976245880127,
      "learning_rate": 0.0004400608412610827,
      "loss": 2.2057,
      "step": 79577
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.817882776260376,
      "learning_rate": 0.00044005722385660165,
      "loss": 2.3678,
      "step": 79578
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.8993383049964905,
      "learning_rate": 0.0004400536064260813,
      "loss": 2.3758,
      "step": 79579
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.8226792812347412,
      "learning_rate": 0.00044004998896952224,
      "loss": 2.0772,
      "step": 79580
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.7940877676010132,
      "learning_rate": 0.00044004637148692524,
      "loss": 2.3332,
      "step": 79581
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.7968452572822571,
      "learning_rate": 0.0004400427539782911,
      "loss": 2.2145,
      "step": 79582
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.8332690596580505,
      "learning_rate": 0.0004400391364436203,
      "loss": 2.3269,
      "step": 79583
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.816109836101532,
      "learning_rate": 0.0004400355188829136,
      "loss": 2.2377,
      "step": 79584
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.8474830389022827,
      "learning_rate": 0.0004400319012961717,
      "loss": 2.4587,
      "step": 79585
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.7980296611785889,
      "learning_rate": 0.0004400282836833951,
      "loss": 2.2259,
      "step": 79586
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.9019304513931274,
      "learning_rate": 0.0004400246660445847,
      "loss": 2.359,
      "step": 79587
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.7565456032752991,
      "learning_rate": 0.000440021048379741,
      "loss": 2.013,
      "step": 79588
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.8394188284873962,
      "learning_rate": 0.0004400174306888648,
      "loss": 2.2922,
      "step": 79589
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.80605149269104,
      "learning_rate": 0.0004400138129719567,
      "loss": 2.1545,
      "step": 79590
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.822882354259491,
      "learning_rate": 0.0004400101952290174,
      "loss": 2.35,
      "step": 79591
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.8330856561660767,
      "learning_rate": 0.00044000657746004755,
      "loss": 2.2353,
      "step": 79592
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.7501149773597717,
      "learning_rate": 0.00044000295966504783,
      "loss": 1.9644,
      "step": 79593
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.857832670211792,
      "learning_rate": 0.00043999934184401897,
      "loss": 2.3535,
      "step": 79594
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.7791492342948914,
      "learning_rate": 0.0004399957239969615,
      "loss": 1.9987,
      "step": 79595
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.7852245569229126,
      "learning_rate": 0.00043999210612387623,
      "loss": 2.3208,
      "step": 79596
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.7908906936645508,
      "learning_rate": 0.0004399884882247637,
      "loss": 2.1339,
      "step": 79597
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.8358649015426636,
      "learning_rate": 0.0004399848702996248,
      "loss": 2.3348,
      "step": 79598
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.901186466217041,
      "learning_rate": 0.00043998125234846,
      "loss": 2.112,
      "step": 79599
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.740940272808075,
      "learning_rate": 0.0004399776343712701,
      "loss": 2.1181,
      "step": 79600
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.804481029510498,
      "learning_rate": 0.00043997401636805566,
      "loss": 2.2122,
      "step": 79601
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.7633071541786194,
      "learning_rate": 0.0004399703983388174,
      "loss": 2.1381,
      "step": 79602
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.7386183738708496,
      "learning_rate": 0.00043996678028355605,
      "loss": 2.0491,
      "step": 79603
    },
    {
      "epoch": 1.04,
      "grad_norm": 1.0000865459442139,
      "learning_rate": 0.00043996316220227226,
      "loss": 2.2356,
      "step": 79604
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.8514726758003235,
      "learning_rate": 0.00043995954409496653,
      "loss": 2.0946,
      "step": 79605
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.7513469457626343,
      "learning_rate": 0.00043995592596163984,
      "loss": 2.2826,
      "step": 79606
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.8978557586669922,
      "learning_rate": 0.0004399523078022927,
      "loss": 2.0404,
      "step": 79607
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.8523535132408142,
      "learning_rate": 0.00043994868961692564,
      "loss": 2.1017,
      "step": 79608
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.8141266107559204,
      "learning_rate": 0.0004399450714055396,
      "loss": 2.223,
      "step": 79609
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.7736392617225647,
      "learning_rate": 0.000439941453168135,
      "loss": 2.163,
      "step": 79610
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.844107985496521,
      "learning_rate": 0.00043993783490471275,
      "loss": 2.0344,
      "step": 79611
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.8473537564277649,
      "learning_rate": 0.0004399342166152734,
      "loss": 2.1787,
      "step": 79612
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.9539167881011963,
      "learning_rate": 0.0004399305982998176,
      "loss": 1.9928,
      "step": 79613
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.7856525778770447,
      "learning_rate": 0.0004399269799583461,
      "loss": 1.9702,
      "step": 79614
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.844476580619812,
      "learning_rate": 0.0004399233615908596,
      "loss": 2.2387,
      "step": 79615
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.8132606148719788,
      "learning_rate": 0.00043991974319735857,
      "loss": 2.2362,
      "step": 79616
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.8738964200019836,
      "learning_rate": 0.0004399161247778439,
      "loss": 2.124,
      "step": 79617
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.8352709412574768,
      "learning_rate": 0.00043991250633231623,
      "loss": 2.1824,
      "step": 79618
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.8766565918922424,
      "learning_rate": 0.0004399088878607761,
      "loss": 2.2926,
      "step": 79619
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.8500833511352539,
      "learning_rate": 0.00043990526936322425,
      "loss": 2.2936,
      "step": 79620
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.7902220487594604,
      "learning_rate": 0.00043990165083966144,
      "loss": 2.2381,
      "step": 79621
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.8072665333747864,
      "learning_rate": 0.0004398980322900882,
      "loss": 2.1109,
      "step": 79622
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.7966620922088623,
      "learning_rate": 0.0004398944137145053,
      "loss": 2.0581,
      "step": 79623
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.734215259552002,
      "learning_rate": 0.00043989079511291346,
      "loss": 2.2618,
      "step": 79624
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.766882598400116,
      "learning_rate": 0.00043988717648531315,
      "loss": 2.1091,
      "step": 79625
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.759144127368927,
      "learning_rate": 0.00043988355783170527,
      "loss": 2.2871,
      "step": 79626
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.7905460596084595,
      "learning_rate": 0.00043987993915209036,
      "loss": 2.2026,
      "step": 79627
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.8158634901046753,
      "learning_rate": 0.0004398763204464691,
      "loss": 2.0862,
      "step": 79628
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.7947133779525757,
      "learning_rate": 0.00043987270171484227,
      "loss": 2.1372,
      "step": 79629
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.7661787271499634,
      "learning_rate": 0.00043986908295721045,
      "loss": 2.19,
      "step": 79630
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.7822485566139221,
      "learning_rate": 0.0004398654641735743,
      "loss": 2.2724,
      "step": 79631
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.8608330488204956,
      "learning_rate": 0.00043986184536393445,
      "loss": 2.2123,
      "step": 79632
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.8156057000160217,
      "learning_rate": 0.00043985822652829173,
      "loss": 2.1549,
      "step": 79633
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.8110470175743103,
      "learning_rate": 0.0004398546076666467,
      "loss": 2.0647,
      "step": 79634
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.8212031722068787,
      "learning_rate": 0.00043985098877900007,
      "loss": 2.1234,
      "step": 79635
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.824462354183197,
      "learning_rate": 0.00043984736986535253,
      "loss": 2.21,
      "step": 79636
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.7605923414230347,
      "learning_rate": 0.00043984375092570463,
      "loss": 2.3063,
      "step": 79637
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.7927872538566589,
      "learning_rate": 0.00043984013196005724,
      "loss": 2.1969,
      "step": 79638
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.917995035648346,
      "learning_rate": 0.0004398365129684109,
      "loss": 2.1233,
      "step": 79639
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.7088760733604431,
      "learning_rate": 0.0004398328939507663,
      "loss": 2.2318,
      "step": 79640
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.7657005190849304,
      "learning_rate": 0.00043982927490712406,
      "loss": 2.46,
      "step": 79641
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.7696859240531921,
      "learning_rate": 0.00043982565583748504,
      "loss": 2.1731,
      "step": 79642
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.8106712102890015,
      "learning_rate": 0.0004398220367418497,
      "loss": 2.3371,
      "step": 79643
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.8046603202819824,
      "learning_rate": 0.00043981841762021887,
      "loss": 2.3538,
      "step": 79644
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.7465828657150269,
      "learning_rate": 0.00043981479847259307,
      "loss": 2.2404,
      "step": 79645
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.8343204855918884,
      "learning_rate": 0.00043981117929897315,
      "loss": 2.2125,
      "step": 79646
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.8907191753387451,
      "learning_rate": 0.0004398075600993597,
      "loss": 2.0783,
      "step": 79647
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.7986107468605042,
      "learning_rate": 0.00043980394087375326,
      "loss": 2.2669,
      "step": 79648
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.8306220769882202,
      "learning_rate": 0.0004398003216221548,
      "loss": 2.1156,
      "step": 79649
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.8073759078979492,
      "learning_rate": 0.0004397967023445647,
      "loss": 2.323,
      "step": 79650
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.7270660400390625,
      "learning_rate": 0.0004397930830409838,
      "loss": 2.0369,
      "step": 79651
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.8190670609474182,
      "learning_rate": 0.00043978946371141275,
      "loss": 2.0763,
      "step": 79652
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.7235561013221741,
      "learning_rate": 0.0004397858443558521,
      "loss": 2.1823,
      "step": 79653
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.7993044257164001,
      "learning_rate": 0.0004397822249743027,
      "loss": 1.9756,
      "step": 79654
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.8097928166389465,
      "learning_rate": 0.0004397786055667651,
      "loss": 2.0857,
      "step": 79655
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.8143609166145325,
      "learning_rate": 0.0004397749861332401,
      "loss": 2.1758,
      "step": 79656
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.861205518245697,
      "learning_rate": 0.0004397713666737282,
      "loss": 2.2236,
      "step": 79657
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.8132537007331848,
      "learning_rate": 0.00043976774718823027,
      "loss": 2.2465,
      "step": 79658
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.8062249422073364,
      "learning_rate": 0.0004397641276767468,
      "loss": 1.976,
      "step": 79659
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.8159142136573792,
      "learning_rate": 0.00043976050813927857,
      "loss": 2.1146,
      "step": 79660
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.8405742645263672,
      "learning_rate": 0.0004397568885758262,
      "loss": 2.1377,
      "step": 79661
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.8575761914253235,
      "learning_rate": 0.00043975326898639043,
      "loss": 2.1095,
      "step": 79662
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.9153720736503601,
      "learning_rate": 0.00043974964937097184,
      "loss": 2.1987,
      "step": 79663
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.8259270191192627,
      "learning_rate": 0.00043974602972957115,
      "loss": 2.3331,
      "step": 79664
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.8155838847160339,
      "learning_rate": 0.00043974241006218903,
      "loss": 2.3052,
      "step": 79665
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.7913568615913391,
      "learning_rate": 0.0004397387903688262,
      "loss": 2.168,
      "step": 79666
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.8480116724967957,
      "learning_rate": 0.00043973517064948333,
      "loss": 2.1577,
      "step": 79667
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.7928765416145325,
      "learning_rate": 0.00043973155090416095,
      "loss": 2.2433,
      "step": 79668
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.9266163110733032,
      "learning_rate": 0.00043972793113285983,
      "loss": 2.2722,
      "step": 79669
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.7547525763511658,
      "learning_rate": 0.00043972431133558075,
      "loss": 2.3979,
      "step": 79670
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.8138431906700134,
      "learning_rate": 0.00043972069151232423,
      "loss": 2.2598,
      "step": 79671
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.8181677460670471,
      "learning_rate": 0.000439717071663091,
      "loss": 2.1577,
      "step": 79672
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.8514809608459473,
      "learning_rate": 0.00043971345178788173,
      "loss": 2.4317,
      "step": 79673
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.9141581654548645,
      "learning_rate": 0.0004397098318866971,
      "loss": 2.0754,
      "step": 79674
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.8927220106124878,
      "learning_rate": 0.0004397062119595377,
      "loss": 2.2826,
      "step": 79675
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.7886219620704651,
      "learning_rate": 0.00043970259200640433,
      "loss": 2.0897,
      "step": 79676
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.7827708125114441,
      "learning_rate": 0.0004396989720272977,
      "loss": 2.2338,
      "step": 79677
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.8470677733421326,
      "learning_rate": 0.0004396953520222183,
      "loss": 2.1682,
      "step": 79678
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.7505099177360535,
      "learning_rate": 0.00043969173199116693,
      "loss": 1.981,
      "step": 79679
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.9003183841705322,
      "learning_rate": 0.0004396881119341442,
      "loss": 1.9408,
      "step": 79680
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.7993239164352417,
      "learning_rate": 0.00043968449185115087,
      "loss": 2.1123,
      "step": 79681
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.8477693796157837,
      "learning_rate": 0.00043968087174218746,
      "loss": 2.0715,
      "step": 79682
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.8973859548568726,
      "learning_rate": 0.00043967725160725475,
      "loss": 2.2479,
      "step": 79683
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.8477945327758789,
      "learning_rate": 0.0004396736314463535,
      "loss": 2.1761,
      "step": 79684
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.8532857894897461,
      "learning_rate": 0.00043967001125948425,
      "loss": 2.2066,
      "step": 79685
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.8441214561462402,
      "learning_rate": 0.00043966639104664766,
      "loss": 2.2101,
      "step": 79686
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.7674862146377563,
      "learning_rate": 0.00043966277080784453,
      "loss": 2.3527,
      "step": 79687
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.7911025285720825,
      "learning_rate": 0.0004396591505430754,
      "loss": 2.2239,
      "step": 79688
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.801189124584198,
      "learning_rate": 0.00043965553025234096,
      "loss": 2.45,
      "step": 79689
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.8421918749809265,
      "learning_rate": 0.000439651909935642,
      "loss": 2.258,
      "step": 79690
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.7662390470504761,
      "learning_rate": 0.00043964828959297916,
      "loss": 2.2328,
      "step": 79691
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.7832440733909607,
      "learning_rate": 0.00043964466922435296,
      "loss": 2.1176,
      "step": 79692
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.7498390674591064,
      "learning_rate": 0.0004396410488297641,
      "loss": 2.0478,
      "step": 79693
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.8208403587341309,
      "learning_rate": 0.00043963742840921353,
      "loss": 1.8704,
      "step": 79694
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.7689167857170105,
      "learning_rate": 0.0004396338079627016,
      "loss": 2.2667,
      "step": 79695
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.8216749429702759,
      "learning_rate": 0.00043963018749022917,
      "loss": 2.3362,
      "step": 79696
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.8546774387359619,
      "learning_rate": 0.00043962656699179686,
      "loss": 2.1454,
      "step": 79697
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.7228325009346008,
      "learning_rate": 0.0004396229464674053,
      "loss": 2.2422,
      "step": 79698
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.8029301762580872,
      "learning_rate": 0.00043961932591705515,
      "loss": 2.0627,
      "step": 79699
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.911220133304596,
      "learning_rate": 0.0004396157053407472,
      "loss": 2.326,
      "step": 79700
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.8276513814926147,
      "learning_rate": 0.0004396120847384821,
      "loss": 2.0269,
      "step": 79701
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.830319881439209,
      "learning_rate": 0.00043960846411026034,
      "loss": 2.1178,
      "step": 79702
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.783261775970459,
      "learning_rate": 0.0004396048434560828,
      "loss": 2.3307,
      "step": 79703
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.8576444983482361,
      "learning_rate": 0.00043960122277595016,
      "loss": 2.1431,
      "step": 79704
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.8230461478233337,
      "learning_rate": 0.00043959760206986294,
      "loss": 2.3554,
      "step": 79705
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.8417733907699585,
      "learning_rate": 0.0004395939813378219,
      "loss": 2.2934,
      "step": 79706
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.7766379714012146,
      "learning_rate": 0.0004395903605798277,
      "loss": 1.981,
      "step": 79707
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.8512781858444214,
      "learning_rate": 0.000439586739795881,
      "loss": 2.0571,
      "step": 79708
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.7804369926452637,
      "learning_rate": 0.0004395831189859825,
      "loss": 2.1717,
      "step": 79709
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.7328827977180481,
      "learning_rate": 0.0004395794981501329,
      "loss": 2.161,
      "step": 79710
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.8582319021224976,
      "learning_rate": 0.0004395758772883328,
      "loss": 2.3286,
      "step": 79711
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.8012014031410217,
      "learning_rate": 0.0004395722564005829,
      "loss": 2.2016,
      "step": 79712
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.7928342819213867,
      "learning_rate": 0.00043956863548688397,
      "loss": 2.3291,
      "step": 79713
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.8128639459609985,
      "learning_rate": 0.0004395650145472365,
      "loss": 2.268,
      "step": 79714
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.7600680589675903,
      "learning_rate": 0.0004395613935816413,
      "loss": 2.2536,
      "step": 79715
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.7342916131019592,
      "learning_rate": 0.00043955777259009897,
      "loss": 2.2251,
      "step": 79716
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.8524364233016968,
      "learning_rate": 0.00043955415157261024,
      "loss": 2.1367,
      "step": 79717
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.859584629535675,
      "learning_rate": 0.00043955053052917574,
      "loss": 2.2387,
      "step": 79718
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.8325701951980591,
      "learning_rate": 0.00043954690945979623,
      "loss": 2.2452,
      "step": 79719
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.8236909508705139,
      "learning_rate": 0.00043954328836447225,
      "loss": 2.3288,
      "step": 79720
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.8170082569122314,
      "learning_rate": 0.0004395396672432045,
      "loss": 2.2682,
      "step": 79721
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.8290950655937195,
      "learning_rate": 0.00043953604609599386,
      "loss": 2.1608,
      "step": 79722
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.7931323051452637,
      "learning_rate": 0.00043953242492284064,
      "loss": 2.0234,
      "step": 79723
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.8791698217391968,
      "learning_rate": 0.00043952880372374577,
      "loss": 2.2641,
      "step": 79724
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.802455484867096,
      "learning_rate": 0.00043952518249870996,
      "loss": 2.2883,
      "step": 79725
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.7870412468910217,
      "learning_rate": 0.0004395215612477337,
      "loss": 1.9314,
      "step": 79726
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.8169540166854858,
      "learning_rate": 0.00043951793997081776,
      "loss": 2.135,
      "step": 79727
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.8278538584709167,
      "learning_rate": 0.00043951431866796285,
      "loss": 2.3774,
      "step": 79728
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.8475480079650879,
      "learning_rate": 0.0004395106973391695,
      "loss": 2.0913,
      "step": 79729
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.8147134780883789,
      "learning_rate": 0.0004395070759844384,
      "loss": 2.3704,
      "step": 79730
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.7676404118537903,
      "learning_rate": 0.0004395034546037705,
      "loss": 2.1324,
      "step": 79731
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.8582973480224609,
      "learning_rate": 0.0004394998331971662,
      "loss": 2.068,
      "step": 79732
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.8134201169013977,
      "learning_rate": 0.0004394962117646262,
      "loss": 2.2297,
      "step": 79733
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.8191309571266174,
      "learning_rate": 0.0004394925903061513,
      "loss": 2.0734,
      "step": 79734
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.8118851780891418,
      "learning_rate": 0.00043948896882174205,
      "loss": 2.0273,
      "step": 79735
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.7715006470680237,
      "learning_rate": 0.0004394853473113992,
      "loss": 2.1469,
      "step": 79736
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.8260617256164551,
      "learning_rate": 0.0004394817257751234,
      "loss": 2.2842,
      "step": 79737
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.7989558577537537,
      "learning_rate": 0.0004394781042129152,
      "loss": 2.3142,
      "step": 79738
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.8939348459243774,
      "learning_rate": 0.0004394744826247755,
      "loss": 2.3472,
      "step": 79739
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.8990647792816162,
      "learning_rate": 0.00043947086101070486,
      "loss": 2.3223,
      "step": 79740
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.8525282144546509,
      "learning_rate": 0.0004394672393707039,
      "loss": 2.0692,
      "step": 79741
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.8023055791854858,
      "learning_rate": 0.00043946361770477336,
      "loss": 2.1997,
      "step": 79742
    },
    {
      "epoch": 1.04,
      "grad_norm": 1.01334810256958,
      "learning_rate": 0.0004394599960129139,
      "loss": 2.4495,
      "step": 79743
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.8434485197067261,
      "learning_rate": 0.0004394563742951262,
      "loss": 2.364,
      "step": 79744
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.805074155330658,
      "learning_rate": 0.00043945275255141095,
      "loss": 2.2246,
      "step": 79745
    },
    {
      "epoch": 1.04,
      "grad_norm": 1.1855202913284302,
      "learning_rate": 0.0004394491307817688,
      "loss": 2.0969,
      "step": 79746
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.7777953743934631,
      "learning_rate": 0.0004394455089862004,
      "loss": 2.1955,
      "step": 79747
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.8389991521835327,
      "learning_rate": 0.00043944188716470644,
      "loss": 2.6684,
      "step": 79748
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.8208109736442566,
      "learning_rate": 0.0004394382653172877,
      "loss": 2.3572,
      "step": 79749
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.8924978971481323,
      "learning_rate": 0.0004394346434439446,
      "loss": 2.4419,
      "step": 79750
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.7511862516403198,
      "learning_rate": 0.00043943102154467804,
      "loss": 2.2673,
      "step": 79751
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.7716015577316284,
      "learning_rate": 0.0004394273996194887,
      "loss": 2.3122,
      "step": 79752
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.7831665277481079,
      "learning_rate": 0.000439423777668377,
      "loss": 2.1456,
      "step": 79753
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.771135151386261,
      "learning_rate": 0.00043942015569134387,
      "loss": 2.1314,
      "step": 79754
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.8023549318313599,
      "learning_rate": 0.0004394165336883899,
      "loss": 2.2353,
      "step": 79755
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.9427466988563538,
      "learning_rate": 0.0004394129116595158,
      "loss": 2.3388,
      "step": 79756
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.8008907437324524,
      "learning_rate": 0.0004394092896047222,
      "loss": 1.9531,
      "step": 79757
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.8291520476341248,
      "learning_rate": 0.0004394056675240098,
      "loss": 2.021,
      "step": 79758
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.8500378727912903,
      "learning_rate": 0.0004394020454173792,
      "loss": 2.4387,
      "step": 79759
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.7500609755516052,
      "learning_rate": 0.0004393984232848311,
      "loss": 2.0894,
      "step": 79760
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.9138071537017822,
      "learning_rate": 0.00043939480112636627,
      "loss": 2.2385,
      "step": 79761
    },
    {
      "epoch": 1.04,
      "grad_norm": 1.0009196996688843,
      "learning_rate": 0.0004393911789419854,
      "loss": 2.3396,
      "step": 79762
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.8604686856269836,
      "learning_rate": 0.00043938755673168887,
      "loss": 2.2037,
      "step": 79763
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.9008769989013672,
      "learning_rate": 0.0004393839344954777,
      "loss": 2.4246,
      "step": 79764
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.8270280957221985,
      "learning_rate": 0.0004393803122333524,
      "loss": 2.0654,
      "step": 79765
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.8108038306236267,
      "learning_rate": 0.00043937668994531367,
      "loss": 2.0158,
      "step": 79766
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.6995365023612976,
      "learning_rate": 0.0004393730676313621,
      "loss": 2.0073,
      "step": 79767
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.8169406652450562,
      "learning_rate": 0.00043936944529149855,
      "loss": 2.1097,
      "step": 79768
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.8355619311332703,
      "learning_rate": 0.0004393658229257236,
      "loss": 1.9893,
      "step": 79769
    },
    {
      "epoch": 1.04,
      "grad_norm": 1.0410650968551636,
      "learning_rate": 0.00043936220053403783,
      "loss": 2.0853,
      "step": 79770
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.8577740788459778,
      "learning_rate": 0.000439358578116442,
      "loss": 2.2988,
      "step": 79771
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.7746145129203796,
      "learning_rate": 0.0004393549556729368,
      "loss": 2.2614,
      "step": 79772
    },
    {
      "epoch": 1.04,
      "grad_norm": 1.0938957929611206,
      "learning_rate": 0.000439351333203523,
      "loss": 2.3684,
      "step": 79773
    },
    {
      "epoch": 1.04,
      "grad_norm": 1.0214840173721313,
      "learning_rate": 0.000439347710708201,
      "loss": 2.2712,
      "step": 79774
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.7835349440574646,
      "learning_rate": 0.0004393440881869717,
      "loss": 2.0952,
      "step": 79775
    },
    {
      "epoch": 1.04,
      "grad_norm": 1.2260750532150269,
      "learning_rate": 0.0004393404656398357,
      "loss": 2.1017,
      "step": 79776
    },
    {
      "epoch": 1.04,
      "grad_norm": 1.05166494846344,
      "learning_rate": 0.0004393368430667937,
      "loss": 2.0425,
      "step": 79777
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.7511551380157471,
      "learning_rate": 0.00043933322046784627,
      "loss": 2.154,
      "step": 79778
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.8961073160171509,
      "learning_rate": 0.00043932959784299415,
      "loss": 2.4245,
      "step": 79779
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.8779451251029968,
      "learning_rate": 0.0004393259751922382,
      "loss": 2.4448,
      "step": 79780
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.9302272796630859,
      "learning_rate": 0.0004393223525155788,
      "loss": 2.1762,
      "step": 79781
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.8090749382972717,
      "learning_rate": 0.0004393187298130167,
      "loss": 2.1194,
      "step": 79782
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.8004545569419861,
      "learning_rate": 0.0004393151070845527,
      "loss": 1.9968,
      "step": 79783
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.8306881189346313,
      "learning_rate": 0.00043931148433018735,
      "loss": 2.2529,
      "step": 79784
    },
    {
      "epoch": 1.04,
      "grad_norm": 1.1272443532943726,
      "learning_rate": 0.0004393078615499214,
      "loss": 2.0728,
      "step": 79785
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.8317719101905823,
      "learning_rate": 0.0004393042387437555,
      "loss": 2.3632,
      "step": 79786
    },
    {
      "epoch": 1.04,
      "grad_norm": 1.0658258199691772,
      "learning_rate": 0.00043930061591169025,
      "loss": 2.2451,
      "step": 79787
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.9343153238296509,
      "learning_rate": 0.0004392969930537264,
      "loss": 2.3089,
      "step": 79788
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.8159464001655579,
      "learning_rate": 0.0004392933701698647,
      "loss": 2.0428,
      "step": 79789
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.9390801191329956,
      "learning_rate": 0.00043928974726010567,
      "loss": 2.1748,
      "step": 79790
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.8572019934654236,
      "learning_rate": 0.00043928612432445,
      "loss": 2.2643,
      "step": 79791
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.8531039953231812,
      "learning_rate": 0.0004392825013628985,
      "loss": 1.9059,
      "step": 79792
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.9961174726486206,
      "learning_rate": 0.0004392788783754516,
      "loss": 2.1556,
      "step": 79793
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.8208885192871094,
      "learning_rate": 0.0004392752553621103,
      "loss": 2.397,
      "step": 79794
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.7783493995666504,
      "learning_rate": 0.0004392716323228751,
      "loss": 2.3287,
      "step": 79795
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.790544331073761,
      "learning_rate": 0.00043926800925774657,
      "loss": 2.1467,
      "step": 79796
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.7797410488128662,
      "learning_rate": 0.0004392643861667255,
      "loss": 2.3521,
      "step": 79797
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.82753986120224,
      "learning_rate": 0.0004392607630498126,
      "loss": 2.1262,
      "step": 79798
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.805797815322876,
      "learning_rate": 0.0004392571399070084,
      "loss": 2.349,
      "step": 79799
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.9194196462631226,
      "learning_rate": 0.0004392535167383138,
      "loss": 2.1781,
      "step": 79800
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.8009684085845947,
      "learning_rate": 0.0004392498935437293,
      "loss": 2.302,
      "step": 79801
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.7814868092536926,
      "learning_rate": 0.00043924627032325566,
      "loss": 2.2149,
      "step": 79802
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.7819985151290894,
      "learning_rate": 0.0004392426470768934,
      "loss": 2.1708,
      "step": 79803
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.7946255207061768,
      "learning_rate": 0.0004392390238046434,
      "loss": 2.0982,
      "step": 79804
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.7793048620223999,
      "learning_rate": 0.00043923540050650617,
      "loss": 2.0371,
      "step": 79805
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.8194526433944702,
      "learning_rate": 0.0004392317771824824,
      "loss": 2.0775,
      "step": 79806
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.8010666370391846,
      "learning_rate": 0.00043922815383257293,
      "loss": 2.3093,
      "step": 79807
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.8080915212631226,
      "learning_rate": 0.00043922453045677827,
      "loss": 2.1609,
      "step": 79808
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.7073791027069092,
      "learning_rate": 0.0004392209070550991,
      "loss": 2.072,
      "step": 79809
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.8104954957962036,
      "learning_rate": 0.00043921728362753627,
      "loss": 2.2448,
      "step": 79810
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.7936719059944153,
      "learning_rate": 0.00043921366017409024,
      "loss": 2.4187,
      "step": 79811
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.8777433037757874,
      "learning_rate": 0.00043921003669476164,
      "loss": 2.4031,
      "step": 79812
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.7603740692138672,
      "learning_rate": 0.0004392064131895515,
      "loss": 2.1984,
      "step": 79813
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.8958580493927002,
      "learning_rate": 0.00043920278965846007,
      "loss": 2.1532,
      "step": 79814
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.8107107877731323,
      "learning_rate": 0.0004391991661014883,
      "loss": 2.3509,
      "step": 79815
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.8073729872703552,
      "learning_rate": 0.0004391955425186368,
      "loss": 2.036,
      "step": 79816
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.7367323040962219,
      "learning_rate": 0.00043919191890990614,
      "loss": 2.2055,
      "step": 79817
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.749786376953125,
      "learning_rate": 0.0004391882952752971,
      "loss": 2.1261,
      "step": 79818
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.731425404548645,
      "learning_rate": 0.00043918467161481044,
      "loss": 1.852,
      "step": 79819
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.8023139238357544,
      "learning_rate": 0.00043918104792844655,
      "loss": 2.3297,
      "step": 79820
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.7478407025337219,
      "learning_rate": 0.00043917742421620634,
      "loss": 2.1143,
      "step": 79821
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.8173386454582214,
      "learning_rate": 0.0004391738004780905,
      "loss": 2.1262,
      "step": 79822
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.7889660000801086,
      "learning_rate": 0.00043917017671409953,
      "loss": 2.3401,
      "step": 79823
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.8383431434631348,
      "learning_rate": 0.0004391665529242342,
      "loss": 2.2008,
      "step": 79824
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.7827205061912537,
      "learning_rate": 0.00043916292910849527,
      "loss": 2.202,
      "step": 79825
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.7873892188072205,
      "learning_rate": 0.00043915930526688317,
      "loss": 1.9437,
      "step": 79826
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.9335585236549377,
      "learning_rate": 0.0004391556813993988,
      "loss": 2.4018,
      "step": 79827
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.9433748722076416,
      "learning_rate": 0.0004391520575060429,
      "loss": 2.0762,
      "step": 79828
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.8498877882957458,
      "learning_rate": 0.0004391484335868158,
      "loss": 2.244,
      "step": 79829
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.7627983689308167,
      "learning_rate": 0.00043914480964171845,
      "loss": 2.2463,
      "step": 79830
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.7813475728034973,
      "learning_rate": 0.00043914118567075155,
      "loss": 2.2494,
      "step": 79831
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.7791244983673096,
      "learning_rate": 0.0004391375616739156,
      "loss": 2.3126,
      "step": 79832
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.9893968105316162,
      "learning_rate": 0.00043913393765121126,
      "loss": 2.232,
      "step": 79833
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.7937784194946289,
      "learning_rate": 0.0004391303136026395,
      "loss": 2.1793,
      "step": 79834
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.8393492102622986,
      "learning_rate": 0.0004391266895282006,
      "loss": 2.1081,
      "step": 79835
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.7725546956062317,
      "learning_rate": 0.0004391230654278955,
      "loss": 2.0235,
      "step": 79836
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.8637337684631348,
      "learning_rate": 0.0004391194413017248,
      "loss": 2.2583,
      "step": 79837
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.774247407913208,
      "learning_rate": 0.0004391158171496892,
      "loss": 2.02,
      "step": 79838
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.9198245406150818,
      "learning_rate": 0.00043911219297178923,
      "loss": 2.3499,
      "step": 79839
    },
    {
      "epoch": 1.04,
      "grad_norm": 1.0462067127227783,
      "learning_rate": 0.00043910856876802574,
      "loss": 2.2323,
      "step": 79840
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.8052059412002563,
      "learning_rate": 0.0004391049445383994,
      "loss": 2.1902,
      "step": 79841
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.7851402163505554,
      "learning_rate": 0.0004391013202829107,
      "loss": 2.1465,
      "step": 79842
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.8624509572982788,
      "learning_rate": 0.00043909769600156053,
      "loss": 2.1532,
      "step": 79843
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.9552893042564392,
      "learning_rate": 0.0004390940716943495,
      "loss": 2.0259,
      "step": 79844
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.7689236998558044,
      "learning_rate": 0.0004390904473612782,
      "loss": 2.1368,
      "step": 79845
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.7776198387145996,
      "learning_rate": 0.00043908682300234736,
      "loss": 2.3845,
      "step": 79846
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.9424758553504944,
      "learning_rate": 0.00043908319861755773,
      "loss": 2.2373,
      "step": 79847
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.9143288135528564,
      "learning_rate": 0.00043907957420690976,
      "loss": 2.0233,
      "step": 79848
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.8746471405029297,
      "learning_rate": 0.0004390759497704044,
      "loss": 2.4444,
      "step": 79849
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.8693572878837585,
      "learning_rate": 0.00043907232530804216,
      "loss": 2.14,
      "step": 79850
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.8428580164909363,
      "learning_rate": 0.0004390687008198237,
      "loss": 2.421,
      "step": 79851
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.798652172088623,
      "learning_rate": 0.0004390650763057498,
      "loss": 2.2203,
      "step": 79852
    },
    {
      "epoch": 1.04,
      "grad_norm": 1.025473952293396,
      "learning_rate": 0.000439061451765821,
      "loss": 2.3789,
      "step": 79853
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.9021018743515015,
      "learning_rate": 0.00043905782720003814,
      "loss": 2.2506,
      "step": 79854
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.7825297713279724,
      "learning_rate": 0.00043905420260840187,
      "loss": 2.4822,
      "step": 79855
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.8728007674217224,
      "learning_rate": 0.00043905057799091264,
      "loss": 2.1253,
      "step": 79856
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.8329333066940308,
      "learning_rate": 0.00043904695334757136,
      "loss": 2.2656,
      "step": 79857
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.8425706028938293,
      "learning_rate": 0.00043904332867837864,
      "loss": 2.3083,
      "step": 79858
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.8612753748893738,
      "learning_rate": 0.00043903970398333507,
      "loss": 2.3052,
      "step": 79859
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.8976805210113525,
      "learning_rate": 0.0004390360792624414,
      "loss": 2.1403,
      "step": 79860
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.922855794429779,
      "learning_rate": 0.0004390324545156984,
      "loss": 2.2004,
      "step": 79861
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.7298871874809265,
      "learning_rate": 0.0004390288297431066,
      "loss": 2.1993,
      "step": 79862
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.9507595896720886,
      "learning_rate": 0.00043902520494466666,
      "loss": 2.0963,
      "step": 79863
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.9278256297111511,
      "learning_rate": 0.00043902158012037936,
      "loss": 2.2745,
      "step": 79864
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.8128311038017273,
      "learning_rate": 0.0004390179552702453,
      "loss": 2.0633,
      "step": 79865
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.8150454759597778,
      "learning_rate": 0.0004390143303942652,
      "loss": 2.3836,
      "step": 79866
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.9976599812507629,
      "learning_rate": 0.0004390107054924397,
      "loss": 2.1212,
      "step": 79867
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.7506086230278015,
      "learning_rate": 0.00043900708056476953,
      "loss": 2.0704,
      "step": 79868
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.8706132173538208,
      "learning_rate": 0.00043900345561125525,
      "loss": 2.3659,
      "step": 79869
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.9359807968139648,
      "learning_rate": 0.00043899983063189766,
      "loss": 1.9746,
      "step": 79870
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.8244945406913757,
      "learning_rate": 0.00043899620562669735,
      "loss": 2.2397,
      "step": 79871
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.7563738822937012,
      "learning_rate": 0.000438992580595655,
      "loss": 2.1052,
      "step": 79872
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.8626258969306946,
      "learning_rate": 0.0004389889555387713,
      "loss": 2.2387,
      "step": 79873
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.7614569067955017,
      "learning_rate": 0.000438985330456047,
      "loss": 2.1985,
      "step": 79874
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.7832755446434021,
      "learning_rate": 0.0004389817053474827,
      "loss": 2.3086,
      "step": 79875
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.8516536355018616,
      "learning_rate": 0.00043897808021307893,
      "loss": 2.2308,
      "step": 79876
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.7579942345619202,
      "learning_rate": 0.0004389744550528367,
      "loss": 2.2668,
      "step": 79877
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.7868952751159668,
      "learning_rate": 0.0004389708298667564,
      "loss": 2.2231,
      "step": 79878
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.8477567434310913,
      "learning_rate": 0.00043896720465483877,
      "loss": 2.057,
      "step": 79879
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.8754963874816895,
      "learning_rate": 0.0004389635794170846,
      "loss": 2.4504,
      "step": 79880
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.8003530502319336,
      "learning_rate": 0.00043895995415349444,
      "loss": 2.4327,
      "step": 79881
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.8292614817619324,
      "learning_rate": 0.00043895632886406896,
      "loss": 2.1972,
      "step": 79882
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.8612481355667114,
      "learning_rate": 0.00043895270354880894,
      "loss": 2.2568,
      "step": 79883
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.8834969997406006,
      "learning_rate": 0.00043894907820771494,
      "loss": 2.2018,
      "step": 79884
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.9450659155845642,
      "learning_rate": 0.0004389454528407877,
      "loss": 1.9967,
      "step": 79885
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.8320235013961792,
      "learning_rate": 0.00043894182744802794,
      "loss": 2.1861,
      "step": 79886
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.8333799839019775,
      "learning_rate": 0.00043893820202943614,
      "loss": 2.1952,
      "step": 79887
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.9788902997970581,
      "learning_rate": 0.0004389345765850132,
      "loss": 2.2038,
      "step": 79888
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.9040361046791077,
      "learning_rate": 0.00043893095111475973,
      "loss": 2.2131,
      "step": 79889
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.7353512644767761,
      "learning_rate": 0.0004389273256186763,
      "loss": 2.0218,
      "step": 79890
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.8255560994148254,
      "learning_rate": 0.00043892370009676363,
      "loss": 2.2915,
      "step": 79891
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.7844547629356384,
      "learning_rate": 0.00043892007454902253,
      "loss": 2.1652,
      "step": 79892
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.9018248915672302,
      "learning_rate": 0.0004389164489754535,
      "loss": 2.2776,
      "step": 79893
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.8741300702095032,
      "learning_rate": 0.0004389128233760573,
      "loss": 2.1132,
      "step": 79894
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.8703274130821228,
      "learning_rate": 0.00043890919775083456,
      "loss": 2.1117,
      "step": 79895
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.7973549365997314,
      "learning_rate": 0.00043890557209978596,
      "loss": 2.2044,
      "step": 79896
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.7870878577232361,
      "learning_rate": 0.0004389019464229122,
      "loss": 2.3889,
      "step": 79897
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.78094881772995,
      "learning_rate": 0.000438898320720214,
      "loss": 2.0128,
      "step": 79898
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.8926066756248474,
      "learning_rate": 0.00043889469499169197,
      "loss": 2.2219,
      "step": 79899
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.8248680233955383,
      "learning_rate": 0.0004388910692373467,
      "loss": 2.1692,
      "step": 79900
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.7800671458244324,
      "learning_rate": 0.00043888744345717913,
      "loss": 2.2099,
      "step": 79901
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.7967272400856018,
      "learning_rate": 0.00043888381765118964,
      "loss": 2.2883,
      "step": 79902
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.8039930462837219,
      "learning_rate": 0.000438880191819379,
      "loss": 2.005,
      "step": 79903
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.8294024467468262,
      "learning_rate": 0.00043887656596174794,
      "loss": 2.263,
      "step": 79904
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.7480654120445251,
      "learning_rate": 0.0004388729400782972,
      "loss": 2.2024,
      "step": 79905
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.79679936170578,
      "learning_rate": 0.00043886931416902726,
      "loss": 2.0608,
      "step": 79906
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.8395326733589172,
      "learning_rate": 0.00043886568823393895,
      "loss": 2.296,
      "step": 79907
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.900566577911377,
      "learning_rate": 0.00043886206227303285,
      "loss": 2.4112,
      "step": 79908
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.7684208154678345,
      "learning_rate": 0.00043885843628630963,
      "loss": 2.2777,
      "step": 79909
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.7407077550888062,
      "learning_rate": 0.00043885481027377003,
      "loss": 2.327,
      "step": 79910
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.9685988426208496,
      "learning_rate": 0.00043885118423541477,
      "loss": 2.2299,
      "step": 79911
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.8115623593330383,
      "learning_rate": 0.0004388475581712444,
      "loss": 2.0899,
      "step": 79912
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.7670073509216309,
      "learning_rate": 0.0004388439320812596,
      "loss": 2.1337,
      "step": 79913
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.8057729005813599,
      "learning_rate": 0.0004388403059654612,
      "loss": 2.2274,
      "step": 79914
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.8852575421333313,
      "learning_rate": 0.00043883667982384973,
      "loss": 2.1766,
      "step": 79915
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.8129317760467529,
      "learning_rate": 0.00043883305365642584,
      "loss": 2.0599,
      "step": 79916
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.8868014812469482,
      "learning_rate": 0.00043882942746319034,
      "loss": 2.2998,
      "step": 79917
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.9035279154777527,
      "learning_rate": 0.0004388258012441438,
      "loss": 2.1384,
      "step": 79918
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.8515702486038208,
      "learning_rate": 0.00043882217499928693,
      "loss": 2.1113,
      "step": 79919
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.8358152508735657,
      "learning_rate": 0.0004388185487286204,
      "loss": 2.136,
      "step": 79920
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.9761457443237305,
      "learning_rate": 0.00043881492243214487,
      "loss": 2.274,
      "step": 79921
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.9085254669189453,
      "learning_rate": 0.00043881129610986097,
      "loss": 2.1268,
      "step": 79922
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.9171304702758789,
      "learning_rate": 0.0004388076697617696,
      "loss": 2.2234,
      "step": 79923
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.8491182327270508,
      "learning_rate": 0.00043880404338787104,
      "loss": 2.228,
      "step": 79924
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.8369256258010864,
      "learning_rate": 0.0004388004169881663,
      "loss": 2.4131,
      "step": 79925
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.8298279047012329,
      "learning_rate": 0.00043879679056265595,
      "loss": 2.0985,
      "step": 79926
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.7829806208610535,
      "learning_rate": 0.00043879316411134066,
      "loss": 2.126,
      "step": 79927
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.8574934005737305,
      "learning_rate": 0.0004387895376342211,
      "loss": 2.2067,
      "step": 79928
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.7356829047203064,
      "learning_rate": 0.0004387859111312979,
      "loss": 2.3701,
      "step": 79929
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.8597197532653809,
      "learning_rate": 0.00043878228460257184,
      "loss": 2.1832,
      "step": 79930
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.8249083161354065,
      "learning_rate": 0.0004387786580480435,
      "loss": 2.1878,
      "step": 79931
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.8855745196342468,
      "learning_rate": 0.0004387750314677136,
      "loss": 2.2106,
      "step": 79932
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.7628669738769531,
      "learning_rate": 0.0004387714048615828,
      "loss": 2.3546,
      "step": 79933
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.7715505361557007,
      "learning_rate": 0.0004387677782296517,
      "loss": 2.3541,
      "step": 79934
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.7256016135215759,
      "learning_rate": 0.00043876415157192115,
      "loss": 2.0108,
      "step": 79935
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.8818256258964539,
      "learning_rate": 0.0004387605248883917,
      "loss": 2.179,
      "step": 79936
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.8027128577232361,
      "learning_rate": 0.00043875689817906405,
      "loss": 2.1548,
      "step": 79937
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.7769661545753479,
      "learning_rate": 0.0004387532714439389,
      "loss": 2.0784,
      "step": 79938
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.8150255084037781,
      "learning_rate": 0.0004387496446830168,
      "loss": 2.0646,
      "step": 79939
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.7809940576553345,
      "learning_rate": 0.0004387460178962986,
      "loss": 2.2693,
      "step": 79940
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.7990421652793884,
      "learning_rate": 0.0004387423910837849,
      "loss": 2.3243,
      "step": 79941
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.8804425597190857,
      "learning_rate": 0.00043873876424547635,
      "loss": 2.1023,
      "step": 79942
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.7873136401176453,
      "learning_rate": 0.00043873513738137363,
      "loss": 2.0197,
      "step": 79943
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.848477840423584,
      "learning_rate": 0.0004387315104914774,
      "loss": 2.2397,
      "step": 79944
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.7364699840545654,
      "learning_rate": 0.00043872788357578847,
      "loss": 2.0768,
      "step": 79945
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.8323838710784912,
      "learning_rate": 0.0004387242566343074,
      "loss": 2.0744,
      "step": 79946
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.8931518197059631,
      "learning_rate": 0.0004387206296670348,
      "loss": 2.2563,
      "step": 79947
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.7901891469955444,
      "learning_rate": 0.00043871700267397143,
      "loss": 2.026,
      "step": 79948
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.8546571731567383,
      "learning_rate": 0.000438713375655118,
      "loss": 2.1821,
      "step": 79949
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.8444449305534363,
      "learning_rate": 0.000438709748610475,
      "loss": 2.0417,
      "step": 79950
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.8667230010032654,
      "learning_rate": 0.00043870612154004335,
      "loss": 2.1232,
      "step": 79951
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.7922240495681763,
      "learning_rate": 0.0004387024944438236,
      "loss": 2.2179,
      "step": 79952
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.7852328419685364,
      "learning_rate": 0.0004386988673218165,
      "loss": 2.1149,
      "step": 79953
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.7679598331451416,
      "learning_rate": 0.0004386952401740225,
      "loss": 2.2844,
      "step": 79954
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.7864236235618591,
      "learning_rate": 0.00043869161300044257,
      "loss": 2.226,
      "step": 79955
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.737667977809906,
      "learning_rate": 0.0004386879858010772,
      "loss": 2.1354,
      "step": 79956
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.8000385761260986,
      "learning_rate": 0.00043868435857592715,
      "loss": 2.0429,
      "step": 79957
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.8366720080375671,
      "learning_rate": 0.000438680731324993,
      "loss": 2.2843,
      "step": 79958
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.7577499747276306,
      "learning_rate": 0.0004386771040482756,
      "loss": 2.0774,
      "step": 79959
    },
    {
      "epoch": 1.04,
      "grad_norm": 1.0467478036880493,
      "learning_rate": 0.00043867347674577545,
      "loss": 2.3278,
      "step": 79960
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.7428572773933411,
      "learning_rate": 0.0004386698494174932,
      "loss": 2.1094,
      "step": 79961
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.7808212041854858,
      "learning_rate": 0.0004386662220634297,
      "loss": 2.2479,
      "step": 79962
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.8063530325889587,
      "learning_rate": 0.00043866259468358545,
      "loss": 2.2384,
      "step": 79963
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.9368191361427307,
      "learning_rate": 0.0004386589672779613,
      "loss": 2.2322,
      "step": 79964
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.781742513179779,
      "learning_rate": 0.00043865533984655783,
      "loss": 2.2213,
      "step": 79965
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.8642078042030334,
      "learning_rate": 0.00043865171238937565,
      "loss": 2.2686,
      "step": 79966
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.7860738635063171,
      "learning_rate": 0.0004386480849064155,
      "loss": 2.2207,
      "step": 79967
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.8389960527420044,
      "learning_rate": 0.0004386444573976782,
      "loss": 2.208,
      "step": 79968
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.8591889142990112,
      "learning_rate": 0.00043864082986316404,
      "loss": 2.523,
      "step": 79969
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.9855146408081055,
      "learning_rate": 0.0004386372023028741,
      "loss": 2.3326,
      "step": 79970
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.8592756986618042,
      "learning_rate": 0.00043863357471680877,
      "loss": 2.1444,
      "step": 79971
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.7384648323059082,
      "learning_rate": 0.00043862994710496895,
      "loss": 2.3396,
      "step": 79972
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.7680554986000061,
      "learning_rate": 0.00043862631946735514,
      "loss": 2.2806,
      "step": 79973
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.809154212474823,
      "learning_rate": 0.0004386226918039681,
      "loss": 2.241,
      "step": 79974
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.7905088663101196,
      "learning_rate": 0.0004386190641148085,
      "loss": 2.417,
      "step": 79975
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.7243582606315613,
      "learning_rate": 0.000438615436399877,
      "loss": 2.1254,
      "step": 79976
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.775536835193634,
      "learning_rate": 0.00043861180865917423,
      "loss": 2.2016,
      "step": 79977
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.8743962049484253,
      "learning_rate": 0.00043860818089270105,
      "loss": 2.1876,
      "step": 79978
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.8723272085189819,
      "learning_rate": 0.0004386045531004578,
      "loss": 2.1929,
      "step": 79979
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.9009849429130554,
      "learning_rate": 0.0004386009252824455,
      "loss": 2.1808,
      "step": 79980
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.8855020999908447,
      "learning_rate": 0.0004385972974386646,
      "loss": 2.307,
      "step": 79981
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.8319343328475952,
      "learning_rate": 0.00043859366956911584,
      "loss": 2.294,
      "step": 79982
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.7530220150947571,
      "learning_rate": 0.0004385900416737999,
      "loss": 2.0796,
      "step": 79983
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.9213321805000305,
      "learning_rate": 0.0004385864137527175,
      "loss": 2.1386,
      "step": 79984
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.9397045373916626,
      "learning_rate": 0.0004385827858058692,
      "loss": 2.4965,
      "step": 79985
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.8469566106796265,
      "learning_rate": 0.0004385791578332557,
      "loss": 2.0138,
      "step": 79986
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.8183357119560242,
      "learning_rate": 0.00043857552983487787,
      "loss": 2.0213,
      "step": 79987
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.8197003602981567,
      "learning_rate": 0.0004385719018107362,
      "loss": 2.2868,
      "step": 79988
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.8563022017478943,
      "learning_rate": 0.0004385682737608313,
      "loss": 2.1692,
      "step": 79989
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.7897529006004333,
      "learning_rate": 0.00043856464568516404,
      "loss": 2.0674,
      "step": 79990
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.8395076990127563,
      "learning_rate": 0.00043856101758373494,
      "loss": 2.0876,
      "step": 79991
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.7577245235443115,
      "learning_rate": 0.00043855738945654477,
      "loss": 2.2181,
      "step": 79992
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.891352117061615,
      "learning_rate": 0.00043855376130359417,
      "loss": 2.2237,
      "step": 79993
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.7748703360557556,
      "learning_rate": 0.00043855013312488374,
      "loss": 2.1244,
      "step": 79994
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.7734559178352356,
      "learning_rate": 0.0004385465049204143,
      "loss": 2.0772,
      "step": 79995
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.7596852779388428,
      "learning_rate": 0.0004385428766901864,
      "loss": 2.0104,
      "step": 79996
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.8687787055969238,
      "learning_rate": 0.00043853924843420083,
      "loss": 2.3021,
      "step": 79997
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.9243581295013428,
      "learning_rate": 0.00043853562015245813,
      "loss": 2.0968,
      "step": 79998
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.8312778472900391,
      "learning_rate": 0.00043853199184495906,
      "loss": 2.2527,
      "step": 79999
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.8155122995376587,
      "learning_rate": 0.0004385283635117043,
      "loss": 2.4339,
      "step": 80000
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.8652437329292297,
      "learning_rate": 0.00043852473515269445,
      "loss": 2.1266,
      "step": 80001
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.7848244309425354,
      "learning_rate": 0.0004385211067679303,
      "loss": 2.0672,
      "step": 80002
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.8330555558204651,
      "learning_rate": 0.0004385174783574124,
      "loss": 2.4056,
      "step": 80003
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.9629078507423401,
      "learning_rate": 0.00043851384992114157,
      "loss": 2.2654,
      "step": 80004
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.7860971093177795,
      "learning_rate": 0.0004385102214591183,
      "loss": 2.1661,
      "step": 80005
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.8305791020393372,
      "learning_rate": 0.00043850659297134343,
      "loss": 2.0749,
      "step": 80006
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.9389200806617737,
      "learning_rate": 0.0004385029644578176,
      "loss": 2.2275,
      "step": 80007
    },
    {
      "epoch": 1.04,
      "grad_norm": 1.1018174886703491,
      "learning_rate": 0.00043849933591854143,
      "loss": 2.134,
      "step": 80008
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.8059491515159607,
      "learning_rate": 0.00043849570735351556,
      "loss": 2.1082,
      "step": 80009
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.912510871887207,
      "learning_rate": 0.00043849207876274076,
      "loss": 2.4301,
      "step": 80010
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.8426969051361084,
      "learning_rate": 0.0004384884501462177,
      "loss": 1.9809,
      "step": 80011
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.818810224533081,
      "learning_rate": 0.000438484821503947,
      "loss": 2.2843,
      "step": 80012
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.8559480905532837,
      "learning_rate": 0.0004384811928359293,
      "loss": 2.2068,
      "step": 80013
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.8150070905685425,
      "learning_rate": 0.0004384775641421655,
      "loss": 1.9342,
      "step": 80014
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.8268830180168152,
      "learning_rate": 0.00043847393542265593,
      "loss": 2.1874,
      "step": 80015
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.8425835371017456,
      "learning_rate": 0.00043847030667740146,
      "loss": 2.0853,
      "step": 80016
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.7662994861602783,
      "learning_rate": 0.0004384666779064028,
      "loss": 2.0711,
      "step": 80017
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.7166545987129211,
      "learning_rate": 0.0004384630491096606,
      "loss": 1.9488,
      "step": 80018
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.8566431403160095,
      "learning_rate": 0.00043845942028717543,
      "loss": 2.0925,
      "step": 80019
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.7975282669067383,
      "learning_rate": 0.0004384557914389481,
      "loss": 2.1737,
      "step": 80020
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.8962038159370422,
      "learning_rate": 0.0004384521625649792,
      "loss": 2.2931,
      "step": 80021
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.8321472406387329,
      "learning_rate": 0.0004384485336652694,
      "loss": 2.139,
      "step": 80022
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.841239333152771,
      "learning_rate": 0.00043844490473981954,
      "loss": 2.3138,
      "step": 80023
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.7478355169296265,
      "learning_rate": 0.00043844127578862997,
      "loss": 2.0952,
      "step": 80024
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.7633109092712402,
      "learning_rate": 0.0004384376468117016,
      "loss": 2.2584,
      "step": 80025
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.7509982585906982,
      "learning_rate": 0.0004384340178090351,
      "loss": 2.3662,
      "step": 80026
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.8996791243553162,
      "learning_rate": 0.00043843038878063113,
      "loss": 2.196,
      "step": 80027
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.8234269618988037,
      "learning_rate": 0.00043842675972649033,
      "loss": 2.0349,
      "step": 80028
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.8321032524108887,
      "learning_rate": 0.00043842313064661333,
      "loss": 2.2422,
      "step": 80029
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.8094930052757263,
      "learning_rate": 0.0004384195015410009,
      "loss": 2.1774,
      "step": 80030
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.7314737439155579,
      "learning_rate": 0.0004384158724096537,
      "loss": 2.1288,
      "step": 80031
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.7703667879104614,
      "learning_rate": 0.0004384122432525723,
      "loss": 2.2358,
      "step": 80032
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.7385889291763306,
      "learning_rate": 0.0004384086140697574,
      "loss": 2.0937,
      "step": 80033
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.7481921315193176,
      "learning_rate": 0.0004384049848612099,
      "loss": 1.9829,
      "step": 80034
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.8596968054771423,
      "learning_rate": 0.00043840135562693013,
      "loss": 2.1966,
      "step": 80035
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.8156487345695496,
      "learning_rate": 0.000438397726366919,
      "loss": 2.0862,
      "step": 80036
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.8980335593223572,
      "learning_rate": 0.00043839409708117724,
      "loss": 2.2691,
      "step": 80037
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.8839910626411438,
      "learning_rate": 0.0004383904677697053,
      "loss": 2.2486,
      "step": 80038
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.8404234647750854,
      "learning_rate": 0.0004383868384325039,
      "loss": 2.2038,
      "step": 80039
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.7858909964561462,
      "learning_rate": 0.00043838320906957386,
      "loss": 2.5287,
      "step": 80040
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.837504506111145,
      "learning_rate": 0.0004383795796809157,
      "loss": 1.9732,
      "step": 80041
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.8232688903808594,
      "learning_rate": 0.00043837595026653025,
      "loss": 2.3278,
      "step": 80042
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.8810228705406189,
      "learning_rate": 0.00043837232082641803,
      "loss": 1.9603,
      "step": 80043
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.8299957513809204,
      "learning_rate": 0.0004383686913605798,
      "loss": 2.0944,
      "step": 80044
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.9524229764938354,
      "learning_rate": 0.0004383650618690163,
      "loss": 2.2472,
      "step": 80045
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.8707056641578674,
      "learning_rate": 0.000438361432351728,
      "loss": 2.1253,
      "step": 80046
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.8459221720695496,
      "learning_rate": 0.0004383578028087157,
      "loss": 2.108,
      "step": 80047
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.76069575548172,
      "learning_rate": 0.00043835417323998015,
      "loss": 2.3092,
      "step": 80048
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.8324899673461914,
      "learning_rate": 0.0004383505436455219,
      "loss": 2.2704,
      "step": 80049
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.8799229264259338,
      "learning_rate": 0.0004383469140253417,
      "loss": 2.1345,
      "step": 80050
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.9511206746101379,
      "learning_rate": 0.00043834328437944024,
      "loss": 2.1248,
      "step": 80051
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.9244916439056396,
      "learning_rate": 0.0004383396547078181,
      "loss": 2.2146,
      "step": 80052
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.8764528632164001,
      "learning_rate": 0.000438336025010476,
      "loss": 2.2862,
      "step": 80053
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.790184497833252,
      "learning_rate": 0.00043833239528741466,
      "loss": 2.1744,
      "step": 80054
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.9401276707649231,
      "learning_rate": 0.00043832876553863464,
      "loss": 2.1831,
      "step": 80055
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.7962049841880798,
      "learning_rate": 0.00043832513576413674,
      "loss": 2.1599,
      "step": 80056
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.8946812152862549,
      "learning_rate": 0.0004383215059639216,
      "loss": 2.3909,
      "step": 80057
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.8597726225852966,
      "learning_rate": 0.00043831787613798993,
      "loss": 2.1747,
      "step": 80058
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.808488130569458,
      "learning_rate": 0.00043831424628634226,
      "loss": 2.2678,
      "step": 80059
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.7545567154884338,
      "learning_rate": 0.0004383106164089794,
      "loss": 2.1795,
      "step": 80060
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.7536236047744751,
      "learning_rate": 0.000438306986505902,
      "loss": 2.3074,
      "step": 80061
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.7649340629577637,
      "learning_rate": 0.00043830335657711063,
      "loss": 2.0754,
      "step": 80062
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.7998227477073669,
      "learning_rate": 0.00043829972662260617,
      "loss": 2.263,
      "step": 80063
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.7534830570220947,
      "learning_rate": 0.0004382960966423891,
      "loss": 1.9375,
      "step": 80064
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.8720626831054688,
      "learning_rate": 0.0004382924666364602,
      "loss": 2.1093,
      "step": 80065
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.8306477665901184,
      "learning_rate": 0.0004382888366048202,
      "loss": 2.2198,
      "step": 80066
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.8367495536804199,
      "learning_rate": 0.0004382852065474696,
      "loss": 2.2591,
      "step": 80067
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.802535355091095,
      "learning_rate": 0.0004382815764644092,
      "loss": 2.0851,
      "step": 80068
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.8100470304489136,
      "learning_rate": 0.0004382779463556396,
      "loss": 2.2708,
      "step": 80069
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.8662844300270081,
      "learning_rate": 0.0004382743162211616,
      "loss": 2.4124,
      "step": 80070
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.9357035756111145,
      "learning_rate": 0.00043827068606097576,
      "loss": 2.2833,
      "step": 80071
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.8208185434341431,
      "learning_rate": 0.00043826705587508277,
      "loss": 2.2033,
      "step": 80072
    },
    {
      "epoch": 1.04,
      "grad_norm": 1.0086314678192139,
      "learning_rate": 0.00043826342566348334,
      "loss": 2.2671,
      "step": 80073
    },
    {
      "epoch": 1.04,
      "grad_norm": 1.0171416997909546,
      "learning_rate": 0.0004382597954261781,
      "loss": 2.3296,
      "step": 80074
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.833389163017273,
      "learning_rate": 0.00043825616516316785,
      "loss": 2.2931,
      "step": 80075
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.7937884330749512,
      "learning_rate": 0.0004382525348744531,
      "loss": 2.4194,
      "step": 80076
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.7832287549972534,
      "learning_rate": 0.00043824890456003454,
      "loss": 2.4747,
      "step": 80077
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.7718920111656189,
      "learning_rate": 0.000438245274219913,
      "loss": 2.171,
      "step": 80078
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.8213355541229248,
      "learning_rate": 0.000438241643854089,
      "loss": 1.9848,
      "step": 80079
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.8058897256851196,
      "learning_rate": 0.00043823801346256323,
      "loss": 2.2601,
      "step": 80080
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.8005210757255554,
      "learning_rate": 0.00043823438304533646,
      "loss": 2.1625,
      "step": 80081
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.8152462840080261,
      "learning_rate": 0.00043823075260240933,
      "loss": 2.0712,
      "step": 80082
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.8104597330093384,
      "learning_rate": 0.0004382271221337824,
      "loss": 2.1614,
      "step": 80083
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.8072230815887451,
      "learning_rate": 0.0004382234916394566,
      "loss": 2.3766,
      "step": 80084
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.8872296214103699,
      "learning_rate": 0.00043821986111943227,
      "loss": 2.2396,
      "step": 80085
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.8328819274902344,
      "learning_rate": 0.0004382162305737103,
      "loss": 2.3708,
      "step": 80086
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.8392953872680664,
      "learning_rate": 0.00043821260000229145,
      "loss": 2.1841,
      "step": 80087
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.9932992458343506,
      "learning_rate": 0.00043820896940517613,
      "loss": 2.0631,
      "step": 80088
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.7679322361946106,
      "learning_rate": 0.00043820533878236517,
      "loss": 2.096,
      "step": 80089
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.9055071473121643,
      "learning_rate": 0.00043820170813385926,
      "loss": 2.1521,
      "step": 80090
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.9607717394828796,
      "learning_rate": 0.00043819807745965907,
      "loss": 1.9725,
      "step": 80091
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.8747625350952148,
      "learning_rate": 0.0004381944467597652,
      "loss": 2.3067,
      "step": 80092
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.8535438179969788,
      "learning_rate": 0.0004381908160341784,
      "loss": 2.0485,
      "step": 80093
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.7894172072410583,
      "learning_rate": 0.00043818718528289936,
      "loss": 2.1569,
      "step": 80094
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.7454712390899658,
      "learning_rate": 0.0004381835545059286,
      "loss": 2.0916,
      "step": 80095
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.8346810340881348,
      "learning_rate": 0.000438179923703267,
      "loss": 2.243,
      "step": 80096
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.7534779906272888,
      "learning_rate": 0.00043817629287491516,
      "loss": 2.1571,
      "step": 80097
    },
    {
      "epoch": 1.04,
      "grad_norm": 1.0331438779830933,
      "learning_rate": 0.0004381726620208737,
      "loss": 2.2519,
      "step": 80098
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.7954127192497253,
      "learning_rate": 0.00043816903114114335,
      "loss": 2.2065,
      "step": 80099
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.8622703552246094,
      "learning_rate": 0.00043816540023572473,
      "loss": 2.0956,
      "step": 80100
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.990940272808075,
      "learning_rate": 0.00043816176930461855,
      "loss": 2.2806,
      "step": 80101
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.8387935161590576,
      "learning_rate": 0.0004381581383478255,
      "loss": 2.1162,
      "step": 80102
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.7706676721572876,
      "learning_rate": 0.0004381545073653463,
      "loss": 2.1512,
      "step": 80103
    },
    {
      "epoch": 1.04,
      "grad_norm": 1.959143877029419,
      "learning_rate": 0.0004381508763571815,
      "loss": 1.8799,
      "step": 80104
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.8151417374610901,
      "learning_rate": 0.0004381472453233318,
      "loss": 2.2556,
      "step": 80105
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.7692958116531372,
      "learning_rate": 0.000438143614263798,
      "loss": 2.237,
      "step": 80106
    },
    {
      "epoch": 1.04,
      "grad_norm": 1.0121947526931763,
      "learning_rate": 0.00043813998317858075,
      "loss": 2.2837,
      "step": 80107
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.8246358633041382,
      "learning_rate": 0.00043813635206768056,
      "loss": 2.2369,
      "step": 80108
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.8802898526191711,
      "learning_rate": 0.0004381327209310983,
      "loss": 2.2521,
      "step": 80109
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.7847129702568054,
      "learning_rate": 0.0004381290897688345,
      "loss": 2.1223,
      "step": 80110
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.8258669972419739,
      "learning_rate": 0.0004381254585808899,
      "loss": 2.327,
      "step": 80111
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.9334037899971008,
      "learning_rate": 0.0004381218273672652,
      "loss": 2.1861,
      "step": 80112
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.790733277797699,
      "learning_rate": 0.0004381181961279611,
      "loss": 2.4028,
      "step": 80113
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.8802604079246521,
      "learning_rate": 0.0004381145648629781,
      "loss": 1.8965,
      "step": 80114
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.9395473003387451,
      "learning_rate": 0.00043811093357231707,
      "loss": 2.2542,
      "step": 80115
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.7749603390693665,
      "learning_rate": 0.00043810730225597857,
      "loss": 2.4328,
      "step": 80116
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.8202152848243713,
      "learning_rate": 0.0004381036709139633,
      "loss": 2.3797,
      "step": 80117
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.7569835782051086,
      "learning_rate": 0.00043810003954627204,
      "loss": 1.9817,
      "step": 80118
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.8673400282859802,
      "learning_rate": 0.00043809640815290526,
      "loss": 2.0683,
      "step": 80119
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.844186544418335,
      "learning_rate": 0.00043809277673386376,
      "loss": 2.2004,
      "step": 80120
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.7855175733566284,
      "learning_rate": 0.0004380891452891483,
      "loss": 2.3648,
      "step": 80121
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.7614231705665588,
      "learning_rate": 0.00043808551381875946,
      "loss": 2.1132,
      "step": 80122
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.7619749903678894,
      "learning_rate": 0.00043808188232269786,
      "loss": 2.3075,
      "step": 80123
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.8530318140983582,
      "learning_rate": 0.0004380782508009642,
      "loss": 2.163,
      "step": 80124
    },
    {
      "epoch": 1.04,
      "grad_norm": 1.049403429031372,
      "learning_rate": 0.0004380746192535592,
      "loss": 2.2327,
      "step": 80125
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.7413436770439148,
      "learning_rate": 0.0004380709876804836,
      "loss": 2.1612,
      "step": 80126
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.7799111008644104,
      "learning_rate": 0.0004380673560817379,
      "loss": 2.253,
      "step": 80127
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.8678701519966125,
      "learning_rate": 0.0004380637244573229,
      "loss": 2.3192,
      "step": 80128
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.9823087453842163,
      "learning_rate": 0.0004380600928072393,
      "loss": 2.1228,
      "step": 80129
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.7535122036933899,
      "learning_rate": 0.00043805646113148774,
      "loss": 2.0664,
      "step": 80130
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.7822679281234741,
      "learning_rate": 0.00043805282943006876,
      "loss": 2.2326,
      "step": 80131
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.8276597261428833,
      "learning_rate": 0.00043804919770298325,
      "loss": 2.1003,
      "step": 80132
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.9164708256721497,
      "learning_rate": 0.0004380455659502318,
      "loss": 2.088,
      "step": 80133
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.7737702131271362,
      "learning_rate": 0.000438041934171815,
      "loss": 2.1902,
      "step": 80134
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.9430482387542725,
      "learning_rate": 0.0004380383023677336,
      "loss": 2.4108,
      "step": 80135
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.7882263660430908,
      "learning_rate": 0.0004380346705379884,
      "loss": 2.2043,
      "step": 80136
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.7848416566848755,
      "learning_rate": 0.00043803103868257984,
      "loss": 2.1677,
      "step": 80137
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.8330360651016235,
      "learning_rate": 0.00043802740680150867,
      "loss": 2.1446,
      "step": 80138
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.874783992767334,
      "learning_rate": 0.00043802377489477563,
      "loss": 2.0924,
      "step": 80139
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.9945662617683411,
      "learning_rate": 0.0004380201429623814,
      "loss": 2.2069,
      "step": 80140
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.8752434253692627,
      "learning_rate": 0.00043801651100432664,
      "loss": 2.1944,
      "step": 80141
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.8278445601463318,
      "learning_rate": 0.000438012879020612,
      "loss": 2.2979,
      "step": 80142
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.8524801135063171,
      "learning_rate": 0.00043800924701123807,
      "loss": 2.2864,
      "step": 80143
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.9982412457466125,
      "learning_rate": 0.0004380056149762057,
      "loss": 2.0309,
      "step": 80144
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.8047965168952942,
      "learning_rate": 0.00043800198291551546,
      "loss": 2.2303,
      "step": 80145
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.8724899291992188,
      "learning_rate": 0.00043799835082916806,
      "loss": 2.2621,
      "step": 80146
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.7874640822410583,
      "learning_rate": 0.0004379947187171641,
      "loss": 2.1104,
      "step": 80147
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.8992404937744141,
      "learning_rate": 0.0004379910865795044,
      "loss": 2.0951,
      "step": 80148
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.7347730994224548,
      "learning_rate": 0.0004379874544161895,
      "loss": 2.099,
      "step": 80149
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.7379603981971741,
      "learning_rate": 0.00043798382222722015,
      "loss": 1.8914,
      "step": 80150
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.8403026461601257,
      "learning_rate": 0.000437980190012597,
      "loss": 1.795,
      "step": 80151
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.8689004182815552,
      "learning_rate": 0.0004379765577723207,
      "loss": 1.9422,
      "step": 80152
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.8051500916481018,
      "learning_rate": 0.00043797292550639197,
      "loss": 2.4772,
      "step": 80153
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.7907637357711792,
      "learning_rate": 0.0004379692932148115,
      "loss": 2.2621,
      "step": 80154
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.8968068361282349,
      "learning_rate": 0.0004379656608975799,
      "loss": 2.4726,
      "step": 80155
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.7445433735847473,
      "learning_rate": 0.0004379620285546979,
      "loss": 2.28,
      "step": 80156
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.7593511939048767,
      "learning_rate": 0.00043795839618616613,
      "loss": 1.9892,
      "step": 80157
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.8026416897773743,
      "learning_rate": 0.0004379547637919853,
      "loss": 2.1104,
      "step": 80158
    },
    {
      "epoch": 1.04,
      "grad_norm": 1.0278857946395874,
      "learning_rate": 0.0004379511313721561,
      "loss": 2.4754,
      "step": 80159
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.7941251397132874,
      "learning_rate": 0.00043794749892667916,
      "loss": 2.089,
      "step": 80160
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.8254576325416565,
      "learning_rate": 0.00043794386645555514,
      "loss": 2.3799,
      "step": 80161
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.749511182308197,
      "learning_rate": 0.00043794023395878485,
      "loss": 1.9704,
      "step": 80162
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.8406808972358704,
      "learning_rate": 0.0004379366014363688,
      "loss": 2.1647,
      "step": 80163
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.7397257685661316,
      "learning_rate": 0.0004379329688883078,
      "loss": 2.1282,
      "step": 80164
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.7709781527519226,
      "learning_rate": 0.0004379293363146023,
      "loss": 1.9936,
      "step": 80165
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.7435249090194702,
      "learning_rate": 0.00043792570371525326,
      "loss": 2.2984,
      "step": 80166
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.8253958821296692,
      "learning_rate": 0.00043792207109026116,
      "loss": 2.0866,
      "step": 80167
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.8249257206916809,
      "learning_rate": 0.00043791843843962676,
      "loss": 2.0975,
      "step": 80168
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.8388303518295288,
      "learning_rate": 0.00043791480576335076,
      "loss": 2.4189,
      "step": 80169
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.7885351181030273,
      "learning_rate": 0.00043791117306143376,
      "loss": 2.0697,
      "step": 80170
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.8170309662818909,
      "learning_rate": 0.00043790754033387646,
      "loss": 1.9954,
      "step": 80171
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.8531736731529236,
      "learning_rate": 0.0004379039075806795,
      "loss": 2.2423,
      "step": 80172
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.7838762402534485,
      "learning_rate": 0.0004379002748018438,
      "loss": 2.2642,
      "step": 80173
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.7368616461753845,
      "learning_rate": 0.00043789664199736966,
      "loss": 2.3012,
      "step": 80174
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.7647575736045837,
      "learning_rate": 0.00043789300916725795,
      "loss": 1.8602,
      "step": 80175
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.8174054622650146,
      "learning_rate": 0.00043788937631150936,
      "loss": 2.1151,
      "step": 80176
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.746082067489624,
      "learning_rate": 0.00043788574343012454,
      "loss": 1.9622,
      "step": 80177
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.7393977642059326,
      "learning_rate": 0.00043788211052310413,
      "loss": 2.0496,
      "step": 80178
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.9035123586654663,
      "learning_rate": 0.00043787847759044885,
      "loss": 2.1889,
      "step": 80179
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.7753030061721802,
      "learning_rate": 0.00043787484463215934,
      "loss": 2.3075,
      "step": 80180
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.8074520826339722,
      "learning_rate": 0.0004378712116482363,
      "loss": 2.2374,
      "step": 80181
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.8473913073539734,
      "learning_rate": 0.00043786757863868047,
      "loss": 2.2548,
      "step": 80182
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.8750881552696228,
      "learning_rate": 0.0004378639456034923,
      "loss": 2.3076,
      "step": 80183
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.9679217338562012,
      "learning_rate": 0.0004378603125426727,
      "loss": 2.2044,
      "step": 80184
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.8315561413764954,
      "learning_rate": 0.0004378566794562224,
      "loss": 2.1544,
      "step": 80185
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.9205456376075745,
      "learning_rate": 0.0004378530463441417,
      "loss": 2.3154,
      "step": 80186
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.8624889254570007,
      "learning_rate": 0.0004378494132064316,
      "loss": 2.4625,
      "step": 80187
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.8424867987632751,
      "learning_rate": 0.0004378457800430928,
      "loss": 2.2918,
      "step": 80188
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.8621863722801208,
      "learning_rate": 0.0004378421468541258,
      "loss": 2.0654,
      "step": 80189
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.9405360817909241,
      "learning_rate": 0.00043783851363953126,
      "loss": 2.1308,
      "step": 80190
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.7819290161132812,
      "learning_rate": 0.00043783488039931005,
      "loss": 2.1046,
      "step": 80191
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.7597026824951172,
      "learning_rate": 0.00043783124713346274,
      "loss": 2.2039,
      "step": 80192
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.7804234027862549,
      "learning_rate": 0.00043782761384198994,
      "loss": 2.3585,
      "step": 80193
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.8958396911621094,
      "learning_rate": 0.0004378239805248924,
      "loss": 2.0592,
      "step": 80194
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.8292486667633057,
      "learning_rate": 0.00043782034718217077,
      "loss": 2.0216,
      "step": 80195
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.8397920727729797,
      "learning_rate": 0.0004378167138138257,
      "loss": 2.3407,
      "step": 80196
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.8114241361618042,
      "learning_rate": 0.000437813080419858,
      "loss": 2.3818,
      "step": 80197
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.8620941638946533,
      "learning_rate": 0.0004378094470002682,
      "loss": 2.3149,
      "step": 80198
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.8017261624336243,
      "learning_rate": 0.00043780581355505697,
      "loss": 2.2257,
      "step": 80199
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.7720600962638855,
      "learning_rate": 0.00043780218008422514,
      "loss": 2.2909,
      "step": 80200
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.7843122482299805,
      "learning_rate": 0.0004377985465877732,
      "loss": 2.1597,
      "step": 80201
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.7753138542175293,
      "learning_rate": 0.0004377949130657019,
      "loss": 2.0875,
      "step": 80202
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.7848954200744629,
      "learning_rate": 0.00043779127951801207,
      "loss": 2.3462,
      "step": 80203
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.7912889122962952,
      "learning_rate": 0.00043778764594470406,
      "loss": 2.2619,
      "step": 80204
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.7732709050178528,
      "learning_rate": 0.0004377840123457788,
      "loss": 2.1471,
      "step": 80205
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.7236067652702332,
      "learning_rate": 0.0004377803787212369,
      "loss": 2.056,
      "step": 80206
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.9006237983703613,
      "learning_rate": 0.00043777674507107907,
      "loss": 2.3525,
      "step": 80207
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.8036972284317017,
      "learning_rate": 0.0004377731113953058,
      "loss": 2.1511,
      "step": 80208
    },
    {
      "epoch": 1.04,
      "grad_norm": 1.0339267253875732,
      "learning_rate": 0.00043776947769391797,
      "loss": 1.97,
      "step": 80209
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.8409148454666138,
      "learning_rate": 0.00043776584396691627,
      "loss": 2.1161,
      "step": 80210
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.7474493384361267,
      "learning_rate": 0.0004377622102143013,
      "loss": 2.3005,
      "step": 80211
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.8281965851783752,
      "learning_rate": 0.00043775857643607363,
      "loss": 2.1581,
      "step": 80212
    },
    {
      "epoch": 1.04,
      "grad_norm": 1.0099098682403564,
      "learning_rate": 0.0004377549426322341,
      "loss": 2.1338,
      "step": 80213
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.823905348777771,
      "learning_rate": 0.0004377513088027833,
      "loss": 1.9888,
      "step": 80214
    },
    {
      "epoch": 1.04,
      "grad_norm": 1.168634057044983,
      "learning_rate": 0.00043774767494772195,
      "loss": 2.3503,
      "step": 80215
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.9281375408172607,
      "learning_rate": 0.0004377440410670507,
      "loss": 2.3122,
      "step": 80216
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.7944472432136536,
      "learning_rate": 0.0004377404071607702,
      "loss": 2.0647,
      "step": 80217
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.8222276568412781,
      "learning_rate": 0.0004377367732288813,
      "loss": 2.3576,
      "step": 80218
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.7849085927009583,
      "learning_rate": 0.00043773313927138435,
      "loss": 2.2559,
      "step": 80219
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.8702331185340881,
      "learning_rate": 0.00043772950528828033,
      "loss": 2.1921,
      "step": 80220
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.7516043782234192,
      "learning_rate": 0.0004377258712795698,
      "loss": 2.121,
      "step": 80221
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.7979117631912231,
      "learning_rate": 0.0004377222372452533,
      "loss": 2.5442,
      "step": 80222
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.8942142724990845,
      "learning_rate": 0.0004377186031853318,
      "loss": 2.3685,
      "step": 80223
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.7737488150596619,
      "learning_rate": 0.00043771496909980574,
      "loss": 2.0339,
      "step": 80224
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.8172143697738647,
      "learning_rate": 0.00043771133498867577,
      "loss": 2.101,
      "step": 80225
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.6821186542510986,
      "learning_rate": 0.0004377077008519428,
      "loss": 2.1107,
      "step": 80226
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.7958820462226868,
      "learning_rate": 0.00043770406668960736,
      "loss": 2.224,
      "step": 80227
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.8485032916069031,
      "learning_rate": 0.00043770043250167007,
      "loss": 2.0567,
      "step": 80228
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.8664195537567139,
      "learning_rate": 0.0004376967982881317,
      "loss": 2.321,
      "step": 80229
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.8194423317909241,
      "learning_rate": 0.00043769316404899294,
      "loss": 2.4638,
      "step": 80230
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.8048596382141113,
      "learning_rate": 0.0004376895297842544,
      "loss": 2.0828,
      "step": 80231
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.8481811285018921,
      "learning_rate": 0.0004376858954939167,
      "loss": 2.2322,
      "step": 80232
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.8061641454696655,
      "learning_rate": 0.0004376822611779807,
      "loss": 2.1056,
      "step": 80233
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.7507916688919067,
      "learning_rate": 0.00043767862683644684,
      "loss": 2.2823,
      "step": 80234
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.8244577646255493,
      "learning_rate": 0.000437674992469316,
      "loss": 2.2501,
      "step": 80235
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.7852150797843933,
      "learning_rate": 0.0004376713580765888,
      "loss": 2.0896,
      "step": 80236
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.7742204666137695,
      "learning_rate": 0.0004376677236582659,
      "loss": 1.9529,
      "step": 80237
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.7104145288467407,
      "learning_rate": 0.0004376640892143479,
      "loss": 2.0436,
      "step": 80238
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.7455176115036011,
      "learning_rate": 0.0004376604547448356,
      "loss": 2.0941,
      "step": 80239
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.7587140202522278,
      "learning_rate": 0.0004376568202497297,
      "loss": 2.0067,
      "step": 80240
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.8943990468978882,
      "learning_rate": 0.0004376531857290307,
      "loss": 2.1058,
      "step": 80241
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.855019211769104,
      "learning_rate": 0.00043764955118273934,
      "loss": 2.008,
      "step": 80242
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.8317328095436096,
      "learning_rate": 0.00043764591661085646,
      "loss": 2.1273,
      "step": 80243
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.7908402681350708,
      "learning_rate": 0.0004376422820133825,
      "loss": 2.2503,
      "step": 80244
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.7087220549583435,
      "learning_rate": 0.0004376386473903183,
      "loss": 2.3487,
      "step": 80245
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.7842392921447754,
      "learning_rate": 0.0004376350127416644,
      "loss": 2.0504,
      "step": 80246
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.7373623847961426,
      "learning_rate": 0.0004376313780674216,
      "loss": 2.1636,
      "step": 80247
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.7886195182800293,
      "learning_rate": 0.0004376277433675905,
      "loss": 2.3561,
      "step": 80248
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.8094147443771362,
      "learning_rate": 0.0004376241086421719,
      "loss": 2.1459,
      "step": 80249
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.7346927523612976,
      "learning_rate": 0.0004376204738911663,
      "loss": 2.1212,
      "step": 80250
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.8060414791107178,
      "learning_rate": 0.0004376168391145745,
      "loss": 2.366,
      "step": 80251
    },
    {
      "epoch": 1.04,
      "grad_norm": 1.0550605058670044,
      "learning_rate": 0.0004376132043123971,
      "loss": 2.2571,
      "step": 80252
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.7884223461151123,
      "learning_rate": 0.0004376095694846348,
      "loss": 2.2138,
      "step": 80253
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.9837501645088196,
      "learning_rate": 0.00043760593463128825,
      "loss": 2.3275,
      "step": 80254
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.8430070877075195,
      "learning_rate": 0.0004376022997523583,
      "loss": 2.3081,
      "step": 80255
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.8121045827865601,
      "learning_rate": 0.00043759866484784527,
      "loss": 2.172,
      "step": 80256
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.9520584940910339,
      "learning_rate": 0.00043759502991775023,
      "loss": 2.1989,
      "step": 80257
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.845954954624176,
      "learning_rate": 0.00043759139496207366,
      "loss": 2.0928,
      "step": 80258
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.7656960487365723,
      "learning_rate": 0.0004375877599808162,
      "loss": 2.0823,
      "step": 80259
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.8670229911804199,
      "learning_rate": 0.0004375841249739785,
      "loss": 2.2295,
      "step": 80260
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.784080445766449,
      "learning_rate": 0.00043758048994156146,
      "loss": 2.4329,
      "step": 80261
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.8956232070922852,
      "learning_rate": 0.00043757685488356553,
      "loss": 2.2685,
      "step": 80262
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.8046833872795105,
      "learning_rate": 0.0004375732197999915,
      "loss": 2.1283,
      "step": 80263
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.7843583226203918,
      "learning_rate": 0.00043756958469084,
      "loss": 2.3333,
      "step": 80264
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.8910220265388489,
      "learning_rate": 0.0004375659495561117,
      "loss": 2.2412,
      "step": 80265
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.9447489976882935,
      "learning_rate": 0.00043756231439580725,
      "loss": 2.1952,
      "step": 80266
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.7769638895988464,
      "learning_rate": 0.0004375586792099275,
      "loss": 2.2995,
      "step": 80267
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.8632141351699829,
      "learning_rate": 0.0004375550439984729,
      "loss": 2.0626,
      "step": 80268
    },
    {
      "epoch": 1.05,
      "grad_norm": 1.005996823310852,
      "learning_rate": 0.00043755140876144423,
      "loss": 2.1579,
      "step": 80269
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.8706334829330444,
      "learning_rate": 0.0004375477734988422,
      "loss": 2.0849,
      "step": 80270
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.9430604577064514,
      "learning_rate": 0.0004375441382106674,
      "loss": 2.5,
      "step": 80271
    },
    {
      "epoch": 1.05,
      "grad_norm": 1.1127616167068481,
      "learning_rate": 0.0004375405028969206,
      "loss": 2.3511,
      "step": 80272
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.8564465045928955,
      "learning_rate": 0.0004375368675576024,
      "loss": 2.2335,
      "step": 80273
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.8930879235267639,
      "learning_rate": 0.0004375332321927134,
      "loss": 2.1409,
      "step": 80274
    },
    {
      "epoch": 1.05,
      "grad_norm": 1.0152192115783691,
      "learning_rate": 0.00043752959680225446,
      "loss": 2.3691,
      "step": 80275
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.8318428993225098,
      "learning_rate": 0.00043752596138622624,
      "loss": 2.3517,
      "step": 80276
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.8624022006988525,
      "learning_rate": 0.0004375223259446293,
      "loss": 2.1324,
      "step": 80277
    },
    {
      "epoch": 1.05,
      "grad_norm": 1.2051137685775757,
      "learning_rate": 0.0004375186904774643,
      "loss": 2.1325,
      "step": 80278
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.9111275672912598,
      "learning_rate": 0.0004375150549847321,
      "loss": 2.0402,
      "step": 80279
    },
    {
      "epoch": 1.05,
      "grad_norm": 1.0020397901535034,
      "learning_rate": 0.0004375114194664331,
      "loss": 2.3995,
      "step": 80280
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.9825771450996399,
      "learning_rate": 0.00043750778392256825,
      "loss": 2.4521,
      "step": 80281
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.9763473272323608,
      "learning_rate": 0.00043750414835313806,
      "loss": 2.1619,
      "step": 80282
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.8374024033546448,
      "learning_rate": 0.0004375005127581433,
      "loss": 1.9943,
      "step": 80283
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.8820205330848694,
      "learning_rate": 0.00043749687713758446,
      "loss": 2.2828,
      "step": 80284
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.8349747061729431,
      "learning_rate": 0.0004374932414914625,
      "loss": 2.2482,
      "step": 80285
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.9082920551300049,
      "learning_rate": 0.00043748960581977786,
      "loss": 2.0475,
      "step": 80286
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.8097095489501953,
      "learning_rate": 0.0004374859701225314,
      "loss": 2.226,
      "step": 80287
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.8427261114120483,
      "learning_rate": 0.0004374823343997237,
      "loss": 2.0782,
      "step": 80288
    },
    {
      "epoch": 1.05,
      "grad_norm": 1.0080845355987549,
      "learning_rate": 0.00043747869865135534,
      "loss": 2.2297,
      "step": 80289
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.9136160612106323,
      "learning_rate": 0.00043747506287742716,
      "loss": 2.1055,
      "step": 80290
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.8271048665046692,
      "learning_rate": 0.0004374714270779398,
      "loss": 2.3532,
      "step": 80291
    },
    {
      "epoch": 1.05,
      "grad_norm": 1.137735366821289,
      "learning_rate": 0.00043746779125289386,
      "loss": 2.142,
      "step": 80292
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.8156988620758057,
      "learning_rate": 0.00043746415540229003,
      "loss": 2.0929,
      "step": 80293
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.729378342628479,
      "learning_rate": 0.00043746051952612907,
      "loss": 2.1496,
      "step": 80294
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.9277620911598206,
      "learning_rate": 0.0004374568836244116,
      "loss": 2.2363,
      "step": 80295
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.9877418875694275,
      "learning_rate": 0.0004374532476971383,
      "loss": 2.1631,
      "step": 80296
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.8355565071105957,
      "learning_rate": 0.0004374496117443098,
      "loss": 2.2628,
      "step": 80297
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.8862027525901794,
      "learning_rate": 0.00043744597576592684,
      "loss": 2.265,
      "step": 80298
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.8691753149032593,
      "learning_rate": 0.00043744233976199015,
      "loss": 2.2552,
      "step": 80299
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.8991088271141052,
      "learning_rate": 0.0004374387037325003,
      "loss": 2.4583,
      "step": 80300
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.8611574769020081,
      "learning_rate": 0.00043743506767745795,
      "loss": 2.169,
      "step": 80301
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.9152510166168213,
      "learning_rate": 0.00043743143159686394,
      "loss": 2.0215,
      "step": 80302
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.8877467513084412,
      "learning_rate": 0.0004374277954907187,
      "loss": 2.1841,
      "step": 80303
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.8237824440002441,
      "learning_rate": 0.0004374241593590231,
      "loss": 2.5153,
      "step": 80304
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.9995143413543701,
      "learning_rate": 0.00043742052320177774,
      "loss": 2.1238,
      "step": 80305
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.8944666385650635,
      "learning_rate": 0.0004374168870189833,
      "loss": 2.145,
      "step": 80306
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.8564830422401428,
      "learning_rate": 0.0004374132508106405,
      "loss": 2.1311,
      "step": 80307
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.7696419358253479,
      "learning_rate": 0.00043740961457675,
      "loss": 2.4429,
      "step": 80308
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.7969973683357239,
      "learning_rate": 0.0004374059783173124,
      "loss": 2.3115,
      "step": 80309
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.8215392827987671,
      "learning_rate": 0.0004374023420323285,
      "loss": 2.2004,
      "step": 80310
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.7867222428321838,
      "learning_rate": 0.0004373987057217988,
      "loss": 2.0859,
      "step": 80311
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.9117143750190735,
      "learning_rate": 0.00043739506938572414,
      "loss": 2.2678,
      "step": 80312
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.8367763161659241,
      "learning_rate": 0.0004373914330241052,
      "loss": 2.052,
      "step": 80313
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.8860001564025879,
      "learning_rate": 0.0004373877966369425,
      "loss": 2.0444,
      "step": 80314
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.7947543263435364,
      "learning_rate": 0.0004373841602242369,
      "loss": 2.2273,
      "step": 80315
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.8645579218864441,
      "learning_rate": 0.000437380523785989,
      "loss": 2.3947,
      "step": 80316
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.790272057056427,
      "learning_rate": 0.00043737688732219945,
      "loss": 1.9809,
      "step": 80317
    },
    {
      "epoch": 1.05,
      "grad_norm": 1.0023410320281982,
      "learning_rate": 0.0004373732508328689,
      "loss": 2.0205,
      "step": 80318
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.8157296776771545,
      "learning_rate": 0.00043736961431799817,
      "loss": 2.1794,
      "step": 80319
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.825446367263794,
      "learning_rate": 0.00043736597777758774,
      "loss": 2.2442,
      "step": 80320
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.8218371868133545,
      "learning_rate": 0.0004373623412116384,
      "loss": 2.1965,
      "step": 80321
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.7360407114028931,
      "learning_rate": 0.00043735870462015085,
      "loss": 2.2903,
      "step": 80322
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.7804996371269226,
      "learning_rate": 0.00043735506800312565,
      "loss": 2.2535,
      "step": 80323
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.9895757436752319,
      "learning_rate": 0.0004373514313605636,
      "loss": 2.2269,
      "step": 80324
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.7559297680854797,
      "learning_rate": 0.00043734779469246536,
      "loss": 2.0826,
      "step": 80325
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.7822202444076538,
      "learning_rate": 0.00043734415799883146,
      "loss": 2.2742,
      "step": 80326
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.8479766249656677,
      "learning_rate": 0.0004373405212796628,
      "loss": 2.2762,
      "step": 80327
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.8301657438278198,
      "learning_rate": 0.00043733688453495995,
      "loss": 2.1277,
      "step": 80328
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.7534531354904175,
      "learning_rate": 0.00043733324776472354,
      "loss": 1.8852,
      "step": 80329
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.712352991104126,
      "learning_rate": 0.00043732961096895417,
      "loss": 2.1363,
      "step": 80330
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.8076576590538025,
      "learning_rate": 0.0004373259741476528,
      "loss": 2.1144,
      "step": 80331
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.9240577816963196,
      "learning_rate": 0.00043732233730081984,
      "loss": 2.2115,
      "step": 80332
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.8788172006607056,
      "learning_rate": 0.00043731870042845614,
      "loss": 2.0693,
      "step": 80333
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.7888070344924927,
      "learning_rate": 0.0004373150635305623,
      "loss": 2.0833,
      "step": 80334
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.8741433024406433,
      "learning_rate": 0.0004373114266071389,
      "loss": 2.3407,
      "step": 80335
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.7812049984931946,
      "learning_rate": 0.0004373077896581868,
      "loss": 2.1153,
      "step": 80336
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.807981014251709,
      "learning_rate": 0.0004373041526837066,
      "loss": 2.1732,
      "step": 80337
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.8678727746009827,
      "learning_rate": 0.0004373005156836989,
      "loss": 2.341,
      "step": 80338
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.9028183817863464,
      "learning_rate": 0.0004372968786581645,
      "loss": 2.2568,
      "step": 80339
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.8682562708854675,
      "learning_rate": 0.00043729324160710406,
      "loss": 2.3039,
      "step": 80340
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.8340224623680115,
      "learning_rate": 0.00043728960453051814,
      "loss": 2.2963,
      "step": 80341
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.8236239552497864,
      "learning_rate": 0.00043728596742840744,
      "loss": 2.1775,
      "step": 80342
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.9628087878227234,
      "learning_rate": 0.0004372823303007728,
      "loss": 2.0881,
      "step": 80343
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.7699907422065735,
      "learning_rate": 0.0004372786931476147,
      "loss": 2.0592,
      "step": 80344
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.814601480960846,
      "learning_rate": 0.0004372750559689339,
      "loss": 2.2437,
      "step": 80345
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.7603808045387268,
      "learning_rate": 0.00043727141876473115,
      "loss": 2.2104,
      "step": 80346
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.7662056088447571,
      "learning_rate": 0.000437267781535007,
      "loss": 2.0057,
      "step": 80347
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.8686141967773438,
      "learning_rate": 0.0004372641442797622,
      "loss": 2.1874,
      "step": 80348
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.9343196749687195,
      "learning_rate": 0.00043726050699899744,
      "loss": 2.2602,
      "step": 80349
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.8453070521354675,
      "learning_rate": 0.00043725686969271326,
      "loss": 2.2824,
      "step": 80350
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.8136119842529297,
      "learning_rate": 0.0004372532323609105,
      "loss": 2.0853,
      "step": 80351
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.7377775311470032,
      "learning_rate": 0.0004372495950035898,
      "loss": 2.0926,
      "step": 80352
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.8326233625411987,
      "learning_rate": 0.00043724595762075174,
      "loss": 2.372,
      "step": 80353
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.8299431800842285,
      "learning_rate": 0.00043724232021239703,
      "loss": 2.1972,
      "step": 80354
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.7306864857673645,
      "learning_rate": 0.00043723868277852654,
      "loss": 2.128,
      "step": 80355
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.8116869330406189,
      "learning_rate": 0.0004372350453191407,
      "loss": 2.0876,
      "step": 80356
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.8339257836341858,
      "learning_rate": 0.00043723140783424023,
      "loss": 2.0956,
      "step": 80357
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.7461430430412292,
      "learning_rate": 0.00043722777032382593,
      "loss": 2.1795,
      "step": 80358
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.9353334903717041,
      "learning_rate": 0.00043722413278789835,
      "loss": 2.1267,
      "step": 80359
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.8614845275878906,
      "learning_rate": 0.0004372204952264582,
      "loss": 2.1972,
      "step": 80360
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.8712379336357117,
      "learning_rate": 0.0004372168576395062,
      "loss": 2.2119,
      "step": 80361
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.7871886491775513,
      "learning_rate": 0.000437213220027043,
      "loss": 2.0552,
      "step": 80362
    },
    {
      "epoch": 1.05,
      "grad_norm": 1.1134395599365234,
      "learning_rate": 0.00043720958238906924,
      "loss": 2.1492,
      "step": 80363
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.8047786951065063,
      "learning_rate": 0.0004372059447255857,
      "loss": 2.0789,
      "step": 80364
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.8353441953659058,
      "learning_rate": 0.00043720230703659296,
      "loss": 2.1442,
      "step": 80365
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.9688610434532166,
      "learning_rate": 0.00043719866932209164,
      "loss": 2.1549,
      "step": 80366
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.7929905652999878,
      "learning_rate": 0.00043719503158208263,
      "loss": 2.3071,
      "step": 80367
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.7715142965316772,
      "learning_rate": 0.0004371913938165664,
      "loss": 2.3566,
      "step": 80368
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.8066465854644775,
      "learning_rate": 0.00043718775602554367,
      "loss": 2.2153,
      "step": 80369
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.7641918659210205,
      "learning_rate": 0.0004371841182090152,
      "loss": 2.207,
      "step": 80370
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.7895663976669312,
      "learning_rate": 0.00043718048036698156,
      "loss": 2.3305,
      "step": 80371
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.7425734400749207,
      "learning_rate": 0.0004371768424994435,
      "loss": 2.0509,
      "step": 80372
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.7853230834007263,
      "learning_rate": 0.0004371732046064017,
      "loss": 2.0111,
      "step": 80373
    },
    {
      "epoch": 1.05,
      "grad_norm": 1.1042145490646362,
      "learning_rate": 0.0004371695666878568,
      "loss": 2.1537,
      "step": 80374
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.7934410572052002,
      "learning_rate": 0.00043716592874380944,
      "loss": 2.2726,
      "step": 80375
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.870517373085022,
      "learning_rate": 0.0004371622907742604,
      "loss": 2.2976,
      "step": 80376
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.9847955703735352,
      "learning_rate": 0.00043715865277921034,
      "loss": 2.308,
      "step": 80377
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.7785874009132385,
      "learning_rate": 0.00043715501475865984,
      "loss": 1.9906,
      "step": 80378
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.7926568388938904,
      "learning_rate": 0.0004371513767126096,
      "loss": 2.1516,
      "step": 80379
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.8040245771408081,
      "learning_rate": 0.0004371477386410604,
      "loss": 2.1128,
      "step": 80380
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.7651492357254028,
      "learning_rate": 0.0004371441005440128,
      "loss": 1.9773,
      "step": 80381
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.7817143201828003,
      "learning_rate": 0.0004371404624214676,
      "loss": 2.1136,
      "step": 80382
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.7363049387931824,
      "learning_rate": 0.0004371368242734253,
      "loss": 2.5006,
      "step": 80383
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.8294585943222046,
      "learning_rate": 0.0004371331860998867,
      "loss": 1.9829,
      "step": 80384
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.8708057403564453,
      "learning_rate": 0.0004371295479008525,
      "loss": 2.2585,
      "step": 80385
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.8881279826164246,
      "learning_rate": 0.00043712590967632327,
      "loss": 2.1579,
      "step": 80386
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.9055120944976807,
      "learning_rate": 0.00043712227142629975,
      "loss": 2.023,
      "step": 80387
    },
    {
      "epoch": 1.05,
      "grad_norm": 1.077069640159607,
      "learning_rate": 0.00043711863315078263,
      "loss": 2.2435,
      "step": 80388
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.9676259160041809,
      "learning_rate": 0.00043711499484977255,
      "loss": 2.227,
      "step": 80389
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.9718384146690369,
      "learning_rate": 0.0004371113565232702,
      "loss": 2.1174,
      "step": 80390
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.8165732622146606,
      "learning_rate": 0.00043710771817127636,
      "loss": 2.1149,
      "step": 80391
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.8006988763809204,
      "learning_rate": 0.00043710407979379154,
      "loss": 2.1189,
      "step": 80392
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.7367972135543823,
      "learning_rate": 0.0004371004413908164,
      "loss": 2.0162,
      "step": 80393
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.8305193185806274,
      "learning_rate": 0.0004370968029623518,
      "loss": 2.0421,
      "step": 80394
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.9223048686981201,
      "learning_rate": 0.00043709316450839833,
      "loss": 2.3473,
      "step": 80395
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.7518478631973267,
      "learning_rate": 0.00043708952602895655,
      "loss": 2.1202,
      "step": 80396
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.8826646208763123,
      "learning_rate": 0.0004370858875240273,
      "loss": 2.0535,
      "step": 80397
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.8860989212989807,
      "learning_rate": 0.00043708224899361125,
      "loss": 2.059,
      "step": 80398
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.790751039981842,
      "learning_rate": 0.00043707861043770893,
      "loss": 2.2564,
      "step": 80399
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.7212488651275635,
      "learning_rate": 0.0004370749718563212,
      "loss": 2.0871,
      "step": 80400
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.7530355453491211,
      "learning_rate": 0.00043707133324944863,
      "loss": 2.321,
      "step": 80401
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.850423276424408,
      "learning_rate": 0.0004370676946170918,
      "loss": 2.2744,
      "step": 80402
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.8234484195709229,
      "learning_rate": 0.0004370640559592516,
      "loss": 2.231,
      "step": 80403
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.8958954215049744,
      "learning_rate": 0.0004370604172759287,
      "loss": 2.233,
      "step": 80404
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.7792889475822449,
      "learning_rate": 0.0004370567785671235,
      "loss": 2.2585,
      "step": 80405
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.8177200555801392,
      "learning_rate": 0.00043705313983283684,
      "loss": 2.3512,
      "step": 80406
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.8005877137184143,
      "learning_rate": 0.00043704950107306955,
      "loss": 2.343,
      "step": 80407
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.8269446492195129,
      "learning_rate": 0.0004370458622878221,
      "loss": 2.5044,
      "step": 80408
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.8358718752861023,
      "learning_rate": 0.00043704222347709524,
      "loss": 1.9561,
      "step": 80409
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.8411136865615845,
      "learning_rate": 0.00043703858464088974,
      "loss": 2.2786,
      "step": 80410
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.790046215057373,
      "learning_rate": 0.00043703494577920603,
      "loss": 2.2075,
      "step": 80411
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.8466798067092896,
      "learning_rate": 0.000437031306892045,
      "loss": 2.2479,
      "step": 80412
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.827584981918335,
      "learning_rate": 0.0004370276679794073,
      "loss": 2.4694,
      "step": 80413
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.7311226725578308,
      "learning_rate": 0.0004370240290412935,
      "loss": 2.5187,
      "step": 80414
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.7732731103897095,
      "learning_rate": 0.0004370203900777044,
      "loss": 2.1732,
      "step": 80415
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.8065299391746521,
      "learning_rate": 0.00043701675108864064,
      "loss": 2.1641,
      "step": 80416
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.8327314853668213,
      "learning_rate": 0.0004370131120741028,
      "loss": 2.0031,
      "step": 80417
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.9164599180221558,
      "learning_rate": 0.0004370094730340917,
      "loss": 2.1973,
      "step": 80418
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.7619296908378601,
      "learning_rate": 0.000437005833968608,
      "loss": 2.3449,
      "step": 80419
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.7654114365577698,
      "learning_rate": 0.00043700219487765224,
      "loss": 2.1943,
      "step": 80420
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.7940549850463867,
      "learning_rate": 0.00043699855576122515,
      "loss": 2.1432,
      "step": 80421
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.8330698609352112,
      "learning_rate": 0.00043699491661932754,
      "loss": 2.1023,
      "step": 80422
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.7594476938247681,
      "learning_rate": 0.0004369912774519599,
      "loss": 2.1806,
      "step": 80423
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.8254581689834595,
      "learning_rate": 0.0004369876382591231,
      "loss": 2.3014,
      "step": 80424
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.863628625869751,
      "learning_rate": 0.0004369839990408177,
      "loss": 2.1255,
      "step": 80425
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.7588627934455872,
      "learning_rate": 0.00043698035979704427,
      "loss": 2.1377,
      "step": 80426
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.8276925086975098,
      "learning_rate": 0.0004369767205278037,
      "loss": 2.0888,
      "step": 80427
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.7733372449874878,
      "learning_rate": 0.0004369730812330966,
      "loss": 2.3988,
      "step": 80428
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.8157853484153748,
      "learning_rate": 0.00043696944191292357,
      "loss": 2.2244,
      "step": 80429
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.8201404809951782,
      "learning_rate": 0.00043696580256728535,
      "loss": 2.3261,
      "step": 80430
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.7837303280830383,
      "learning_rate": 0.0004369621631961826,
      "loss": 2.0667,
      "step": 80431
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.7580847144126892,
      "learning_rate": 0.0004369585237996159,
      "loss": 2.1303,
      "step": 80432
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.7435108423233032,
      "learning_rate": 0.0004369548843775861,
      "loss": 2.205,
      "step": 80433
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.9152402877807617,
      "learning_rate": 0.00043695124493009393,
      "loss": 2.2649,
      "step": 80434
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.746121883392334,
      "learning_rate": 0.0004369476054571398,
      "loss": 2.212,
      "step": 80435
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.9092490077018738,
      "learning_rate": 0.0004369439659587245,
      "loss": 2.2167,
      "step": 80436
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.8638510704040527,
      "learning_rate": 0.0004369403264348488,
      "loss": 2.4225,
      "step": 80437
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.8952274918556213,
      "learning_rate": 0.00043693668688551333,
      "loss": 2.1679,
      "step": 80438
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.8361468315124512,
      "learning_rate": 0.0004369330473107187,
      "loss": 2.1421,
      "step": 80439
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.8557946085929871,
      "learning_rate": 0.00043692940771046556,
      "loss": 2.2086,
      "step": 80440
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.8489981889724731,
      "learning_rate": 0.0004369257680847548,
      "loss": 2.0956,
      "step": 80441
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.9002597332000732,
      "learning_rate": 0.0004369221284335869,
      "loss": 2.217,
      "step": 80442
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.8126099109649658,
      "learning_rate": 0.00043691848875696254,
      "loss": 2.2484,
      "step": 80443
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.7734184861183167,
      "learning_rate": 0.00043691484905488255,
      "loss": 2.1328,
      "step": 80444
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.8658027052879333,
      "learning_rate": 0.0004369112093273474,
      "loss": 2.1207,
      "step": 80445
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.8156623840332031,
      "learning_rate": 0.00043690756957435794,
      "loss": 2.1425,
      "step": 80446
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.8724274039268494,
      "learning_rate": 0.00043690392979591473,
      "loss": 2.2132,
      "step": 80447
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.8409203886985779,
      "learning_rate": 0.0004369002899920185,
      "loss": 2.0287,
      "step": 80448
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.8812137246131897,
      "learning_rate": 0.00043689665016266994,
      "loss": 2.1909,
      "step": 80449
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.7946633100509644,
      "learning_rate": 0.0004368930103078698,
      "loss": 2.2528,
      "step": 80450
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.817596971988678,
      "learning_rate": 0.0004368893704276185,
      "loss": 2.2604,
      "step": 80451
    },
    {
      "epoch": 1.05,
      "grad_norm": 1.0442343950271606,
      "learning_rate": 0.0004368857305219169,
      "loss": 2.0942,
      "step": 80452
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.8627681136131287,
      "learning_rate": 0.00043688209059076583,
      "loss": 2.1629,
      "step": 80453
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.8889483213424683,
      "learning_rate": 0.00043687845063416564,
      "loss": 2.1955,
      "step": 80454
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.7986074686050415,
      "learning_rate": 0.0004368748106521172,
      "loss": 2.1892,
      "step": 80455
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.9395751357078552,
      "learning_rate": 0.00043687117064462123,
      "loss": 2.0,
      "step": 80456
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.7920783162117004,
      "learning_rate": 0.0004368675306116782,
      "loss": 2.0376,
      "step": 80457
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.8626744747161865,
      "learning_rate": 0.00043686389055328894,
      "loss": 2.4058,
      "step": 80458
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.8826528787612915,
      "learning_rate": 0.00043686025046945417,
      "loss": 2.2152,
      "step": 80459
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.9108327627182007,
      "learning_rate": 0.00043685661036017446,
      "loss": 1.9632,
      "step": 80460
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.8135672807693481,
      "learning_rate": 0.0004368529702254505,
      "loss": 2.1325,
      "step": 80461
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.828688383102417,
      "learning_rate": 0.000436849330065283,
      "loss": 2.1658,
      "step": 80462
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.8532257080078125,
      "learning_rate": 0.00043684568987967264,
      "loss": 2.1532,
      "step": 80463
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.8107950091362,
      "learning_rate": 0.00043684204966862,
      "loss": 2.0361,
      "step": 80464
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.8137624859809875,
      "learning_rate": 0.000436838409432126,
      "loss": 2.039,
      "step": 80465
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.8332447409629822,
      "learning_rate": 0.00043683476917019107,
      "loss": 2.2553,
      "step": 80466
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.9061304330825806,
      "learning_rate": 0.0004368311288828159,
      "loss": 2.2641,
      "step": 80467
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.7613148093223572,
      "learning_rate": 0.0004368274885700014,
      "loss": 2.0928,
      "step": 80468
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.7942404747009277,
      "learning_rate": 0.00043682384823174794,
      "loss": 2.1126,
      "step": 80469
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.8524942398071289,
      "learning_rate": 0.0004368202078680564,
      "loss": 2.132,
      "step": 80470
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.8585212826728821,
      "learning_rate": 0.0004368165674789274,
      "loss": 2.1667,
      "step": 80471
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.7457777857780457,
      "learning_rate": 0.00043681292706436166,
      "loss": 2.0272,
      "step": 80472
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.9148274660110474,
      "learning_rate": 0.00043680928662435974,
      "loss": 2.3649,
      "step": 80473
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.7952683568000793,
      "learning_rate": 0.00043680564615892234,
      "loss": 2.2359,
      "step": 80474
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.7739212512969971,
      "learning_rate": 0.0004368020056680504,
      "loss": 2.1318,
      "step": 80475
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.7739196419715881,
      "learning_rate": 0.0004367983651517442,
      "loss": 2.3683,
      "step": 80476
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.774718165397644,
      "learning_rate": 0.00043679472461000465,
      "loss": 2.291,
      "step": 80477
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.7915916442871094,
      "learning_rate": 0.00043679108404283233,
      "loss": 2.3726,
      "step": 80478
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.9747021198272705,
      "learning_rate": 0.0004367874434502281,
      "loss": 2.0895,
      "step": 80479
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.8090494871139526,
      "learning_rate": 0.00043678380283219243,
      "loss": 2.4367,
      "step": 80480
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.7815842628479004,
      "learning_rate": 0.000436780162188726,
      "loss": 2.3067,
      "step": 80481
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.9091659188270569,
      "learning_rate": 0.00043677652151982965,
      "loss": 2.1628,
      "step": 80482
    },
    {
      "epoch": 1.05,
      "grad_norm": 1.0647858381271362,
      "learning_rate": 0.0004367728808255039,
      "loss": 2.3512,
      "step": 80483
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.8471385836601257,
      "learning_rate": 0.00043676924010574956,
      "loss": 2.3242,
      "step": 80484
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.9002326726913452,
      "learning_rate": 0.0004367655993605672,
      "loss": 2.1776,
      "step": 80485
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.8673189878463745,
      "learning_rate": 0.0004367619585899575,
      "loss": 2.2364,
      "step": 80486
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.8243016004562378,
      "learning_rate": 0.00043675831779392126,
      "loss": 2.1172,
      "step": 80487
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.8786334991455078,
      "learning_rate": 0.00043675467697245897,
      "loss": 2.378,
      "step": 80488
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.8890009522438049,
      "learning_rate": 0.0004367510361255715,
      "loss": 2.32,
      "step": 80489
    },
    {
      "epoch": 1.05,
      "grad_norm": 1.0694328546524048,
      "learning_rate": 0.00043674739525325935,
      "loss": 2.1998,
      "step": 80490
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.8382136225700378,
      "learning_rate": 0.0004367437543555232,
      "loss": 1.927,
      "step": 80491
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.96601402759552,
      "learning_rate": 0.000436740113432364,
      "loss": 2.0475,
      "step": 80492
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.9729775786399841,
      "learning_rate": 0.00043673647248378215,
      "loss": 2.1199,
      "step": 80493
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.7820457220077515,
      "learning_rate": 0.00043673283150977836,
      "loss": 2.209,
      "step": 80494
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.7978071570396423,
      "learning_rate": 0.00043672919051035344,
      "loss": 1.995,
      "step": 80495
    },
    {
      "epoch": 1.05,
      "grad_norm": 1.0819460153579712,
      "learning_rate": 0.0004367255494855079,
      "loss": 2.2278,
      "step": 80496
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.8187598586082458,
      "learning_rate": 0.00043672190843524255,
      "loss": 2.108,
      "step": 80497
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.8136096596717834,
      "learning_rate": 0.00043671826735955805,
      "loss": 2.1787,
      "step": 80498
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.8473708033561707,
      "learning_rate": 0.00043671462625845497,
      "loss": 2.0393,
      "step": 80499
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.9389974474906921,
      "learning_rate": 0.0004367109851319341,
      "loss": 2.1381,
      "step": 80500
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.7707321047782898,
      "learning_rate": 0.000436707343979996,
      "loss": 1.9897,
      "step": 80501
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.8901841640472412,
      "learning_rate": 0.00043670370280264153,
      "loss": 2.3545,
      "step": 80502
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.8880815505981445,
      "learning_rate": 0.0004367000615998712,
      "loss": 2.3995,
      "step": 80503
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.7453529834747314,
      "learning_rate": 0.0004366964203716858,
      "loss": 1.8273,
      "step": 80504
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.8347086906433105,
      "learning_rate": 0.0004366927791180859,
      "loss": 2.4591,
      "step": 80505
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.8512008190155029,
      "learning_rate": 0.0004366891378390722,
      "loss": 2.299,
      "step": 80506
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.8347181081771851,
      "learning_rate": 0.00043668549653464546,
      "loss": 2.1004,
      "step": 80507
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.8268864750862122,
      "learning_rate": 0.00043668185520480635,
      "loss": 2.1812,
      "step": 80508
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.7805219888687134,
      "learning_rate": 0.0004366782138495555,
      "loss": 2.2168,
      "step": 80509
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.8137379288673401,
      "learning_rate": 0.0004366745724688935,
      "loss": 2.1911,
      "step": 80510
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.8244194984436035,
      "learning_rate": 0.0004366709310628212,
      "loss": 2.0469,
      "step": 80511
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.7578700184822083,
      "learning_rate": 0.0004366672896313391,
      "loss": 1.9789,
      "step": 80512
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.8308340311050415,
      "learning_rate": 0.00043666364817444804,
      "loss": 2.0055,
      "step": 80513
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.8474229574203491,
      "learning_rate": 0.0004366600066921487,
      "loss": 2.2146,
      "step": 80514
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.7367538809776306,
      "learning_rate": 0.00043665636518444155,
      "loss": 2.2533,
      "step": 80515
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.7794641852378845,
      "learning_rate": 0.00043665272365132744,
      "loss": 2.1214,
      "step": 80516
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.7575299143791199,
      "learning_rate": 0.0004366490820928071,
      "loss": 2.0,
      "step": 80517
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.7664051651954651,
      "learning_rate": 0.00043664544050888105,
      "loss": 2.2364,
      "step": 80518
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.7441970705986023,
      "learning_rate": 0.00043664179889954996,
      "loss": 2.2387,
      "step": 80519
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.7463885545730591,
      "learning_rate": 0.0004366381572648147,
      "loss": 2.1502,
      "step": 80520
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.8872519135475159,
      "learning_rate": 0.00043663451560467574,
      "loss": 2.167,
      "step": 80521
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.8793171644210815,
      "learning_rate": 0.0004366308739191339,
      "loss": 1.9611,
      "step": 80522
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.9269874095916748,
      "learning_rate": 0.00043662723220818983,
      "loss": 2.274,
      "step": 80523
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.7828740477561951,
      "learning_rate": 0.0004366235904718441,
      "loss": 2.2504,
      "step": 80524
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.8427189588546753,
      "learning_rate": 0.0004366199487100974,
      "loss": 2.291,
      "step": 80525
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.8829267024993896,
      "learning_rate": 0.0004366163069229507,
      "loss": 2.2426,
      "step": 80526
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.8110358119010925,
      "learning_rate": 0.00043661266511040427,
      "loss": 2.3548,
      "step": 80527
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.8170098066329956,
      "learning_rate": 0.000436609023272459,
      "loss": 2.2018,
      "step": 80528
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.7745242118835449,
      "learning_rate": 0.00043660538140911557,
      "loss": 2.2226,
      "step": 80529
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.8456096649169922,
      "learning_rate": 0.0004366017395203746,
      "loss": 2.1721,
      "step": 80530
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.845717191696167,
      "learning_rate": 0.0004365980976062368,
      "loss": 2.1577,
      "step": 80531
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.9305362105369568,
      "learning_rate": 0.00043659445566670283,
      "loss": 2.3101,
      "step": 80532
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.8081367611885071,
      "learning_rate": 0.0004365908137017733,
      "loss": 2.1299,
      "step": 80533
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.8121625781059265,
      "learning_rate": 0.0004365871717114491,
      "loss": 2.2791,
      "step": 80534
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.8177686333656311,
      "learning_rate": 0.0004365835296957307,
      "loss": 2.4796,
      "step": 80535
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.702699601650238,
      "learning_rate": 0.0004365798876546188,
      "loss": 2.2654,
      "step": 80536
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.8886544108390808,
      "learning_rate": 0.0004365762455881141,
      "loss": 2.1385,
      "step": 80537
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.8164927363395691,
      "learning_rate": 0.00043657260349621744,
      "loss": 2.1167,
      "step": 80538
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.9763043522834778,
      "learning_rate": 0.00043656896137892924,
      "loss": 2.1513,
      "step": 80539
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.8467445373535156,
      "learning_rate": 0.0004365653192362503,
      "loss": 2.1941,
      "step": 80540
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.9134406447410583,
      "learning_rate": 0.0004365616770681814,
      "loss": 2.3321,
      "step": 80541
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.8338443636894226,
      "learning_rate": 0.00043655803487472293,
      "loss": 2.3303,
      "step": 80542
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.8116872906684875,
      "learning_rate": 0.0004365543926558758,
      "loss": 2.2574,
      "step": 80543
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.875873327255249,
      "learning_rate": 0.0004365507504116407,
      "loss": 2.0601,
      "step": 80544
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.8349917531013489,
      "learning_rate": 0.00043654710814201826,
      "loss": 2.1059,
      "step": 80545
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.764167070388794,
      "learning_rate": 0.000436543465847009,
      "loss": 2.2224,
      "step": 80546
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.8277297019958496,
      "learning_rate": 0.0004365398235266138,
      "loss": 2.1949,
      "step": 80547
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.812295138835907,
      "learning_rate": 0.0004365361811808333,
      "loss": 2.2191,
      "step": 80548
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.7523643374443054,
      "learning_rate": 0.00043653253880966816,
      "loss": 2.0864,
      "step": 80549
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.8130849599838257,
      "learning_rate": 0.00043652889641311896,
      "loss": 2.4035,
      "step": 80550
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.7601319551467896,
      "learning_rate": 0.0004365252539911865,
      "loss": 2.098,
      "step": 80551
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.8992554545402527,
      "learning_rate": 0.0004365216115438714,
      "loss": 2.1651,
      "step": 80552
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.7740774750709534,
      "learning_rate": 0.00043651796907117447,
      "loss": 2.1145,
      "step": 80553
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.8920599222183228,
      "learning_rate": 0.0004365143265730961,
      "loss": 2.2643,
      "step": 80554
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.8697044253349304,
      "learning_rate": 0.0004365106840496372,
      "loss": 2.2379,
      "step": 80555
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.7887004613876343,
      "learning_rate": 0.0004365070415007985,
      "loss": 2.1408,
      "step": 80556
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.745668351650238,
      "learning_rate": 0.00043650339892658044,
      "loss": 2.2321,
      "step": 80557
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.7822715044021606,
      "learning_rate": 0.0004364997563269838,
      "loss": 2.1761,
      "step": 80558
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.7780089378356934,
      "learning_rate": 0.00043649611370200937,
      "loss": 2.0486,
      "step": 80559
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.828262448310852,
      "learning_rate": 0.00043649247105165777,
      "loss": 2.3345,
      "step": 80560
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.8172544836997986,
      "learning_rate": 0.0004364888283759295,
      "loss": 2.1387,
      "step": 80561
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.8534022569656372,
      "learning_rate": 0.00043648518567482547,
      "loss": 2.2304,
      "step": 80562
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.7994645237922668,
      "learning_rate": 0.0004364815429483463,
      "loss": 2.2086,
      "step": 80563
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.8557689785957336,
      "learning_rate": 0.0004364779001964925,
      "loss": 1.8523,
      "step": 80564
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.7605908513069153,
      "learning_rate": 0.000436474257419265,
      "loss": 2.22,
      "step": 80565
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.848831057548523,
      "learning_rate": 0.00043647061461666426,
      "loss": 2.2928,
      "step": 80566
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.923651933670044,
      "learning_rate": 0.00043646697178869123,
      "loss": 2.2851,
      "step": 80567
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.8485580086708069,
      "learning_rate": 0.00043646332893534624,
      "loss": 2.1671,
      "step": 80568
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.7772002816200256,
      "learning_rate": 0.0004364596860566302,
      "loss": 2.0445,
      "step": 80569
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.9588605165481567,
      "learning_rate": 0.00043645604315254376,
      "loss": 2.1265,
      "step": 80570
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.7846013903617859,
      "learning_rate": 0.0004364524002230875,
      "loss": 2.327,
      "step": 80571
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.8274145126342773,
      "learning_rate": 0.0004364487572682623,
      "loss": 2.0147,
      "step": 80572
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.9393875598907471,
      "learning_rate": 0.0004364451142880685,
      "loss": 1.8193,
      "step": 80573
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.7749828100204468,
      "learning_rate": 0.00043644147128250714,
      "loss": 2.2124,
      "step": 80574
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.8096773028373718,
      "learning_rate": 0.00043643782825157867,
      "loss": 2.2773,
      "step": 80575
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.8179554343223572,
      "learning_rate": 0.00043643418519528383,
      "loss": 2.1612,
      "step": 80576
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.8285561800003052,
      "learning_rate": 0.0004364305421136233,
      "loss": 2.2194,
      "step": 80577
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.7846965789794922,
      "learning_rate": 0.0004364268990065978,
      "loss": 2.0182,
      "step": 80578
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.7978606820106506,
      "learning_rate": 0.0004364232558742079,
      "loss": 2.2378,
      "step": 80579
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.8884396553039551,
      "learning_rate": 0.00043641961271645435,
      "loss": 2.3355,
      "step": 80580
    },
    {
      "epoch": 1.05,
      "grad_norm": 1.010864496231079,
      "learning_rate": 0.0004364159695333379,
      "loss": 2.1004,
      "step": 80581
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.8011276125907898,
      "learning_rate": 0.00043641232632485907,
      "loss": 2.5054,
      "step": 80582
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.8988814353942871,
      "learning_rate": 0.0004364086830910186,
      "loss": 2.0957,
      "step": 80583
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.9270668029785156,
      "learning_rate": 0.00043640503983181727,
      "loss": 2.3111,
      "step": 80584
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.8096912503242493,
      "learning_rate": 0.0004364013965472556,
      "loss": 2.0583,
      "step": 80585
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.8141144514083862,
      "learning_rate": 0.00043639775323733437,
      "loss": 2.3424,
      "step": 80586
    },
    {
      "epoch": 1.05,
      "grad_norm": 1.092591643333435,
      "learning_rate": 0.0004363941099020542,
      "loss": 2.1781,
      "step": 80587
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.7985684871673584,
      "learning_rate": 0.0004363904665414158,
      "loss": 2.2556,
      "step": 80588
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.7847397327423096,
      "learning_rate": 0.00043638682315541973,
      "loss": 2.0122,
      "step": 80589
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.8465451002120972,
      "learning_rate": 0.000436383179744067,
      "loss": 2.3077,
      "step": 80590
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.7998389005661011,
      "learning_rate": 0.0004363795363073578,
      "loss": 2.1447,
      "step": 80591
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.8455547094345093,
      "learning_rate": 0.00043637589284529327,
      "loss": 2.2988,
      "step": 80592
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.7658313512802124,
      "learning_rate": 0.00043637224935787384,
      "loss": 2.4376,
      "step": 80593
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.7950855493545532,
      "learning_rate": 0.0004363686058451002,
      "loss": 2.2688,
      "step": 80594
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.8077980279922485,
      "learning_rate": 0.00043636496230697305,
      "loss": 2.3905,
      "step": 80595
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.8435098528862,
      "learning_rate": 0.00043636131874349315,
      "loss": 2.2817,
      "step": 80596
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.7791774272918701,
      "learning_rate": 0.0004363576751546611,
      "loss": 2.1457,
      "step": 80597
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.8717968463897705,
      "learning_rate": 0.00043635403154047755,
      "loss": 2.3127,
      "step": 80598
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.9013850688934326,
      "learning_rate": 0.00043635038790094315,
      "loss": 2.113,
      "step": 80599
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.8990497589111328,
      "learning_rate": 0.0004363467442360587,
      "loss": 2.0393,
      "step": 80600
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.8361982107162476,
      "learning_rate": 0.0004363431005458248,
      "loss": 2.2618,
      "step": 80601
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.8749777674674988,
      "learning_rate": 0.00043633945683024225,
      "loss": 2.2227,
      "step": 80602
    },
    {
      "epoch": 1.05,
      "grad_norm": 1.0833234786987305,
      "learning_rate": 0.00043633581308931147,
      "loss": 2.3252,
      "step": 80603
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.7697175741195679,
      "learning_rate": 0.00043633216932303335,
      "loss": 2.198,
      "step": 80604
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.8414978981018066,
      "learning_rate": 0.0004363285255314086,
      "loss": 2.1362,
      "step": 80605
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.896791398525238,
      "learning_rate": 0.0004363248817144376,
      "loss": 2.3526,
      "step": 80606
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.8569409251213074,
      "learning_rate": 0.00043632123787212135,
      "loss": 2.185,
      "step": 80607
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.7258808016777039,
      "learning_rate": 0.00043631759400446045,
      "loss": 2.2702,
      "step": 80608
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.986988365650177,
      "learning_rate": 0.00043631395011145546,
      "loss": 2.2622,
      "step": 80609
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.8883551359176636,
      "learning_rate": 0.00043631030619310714,
      "loss": 2.1784,
      "step": 80610
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.7548550963401794,
      "learning_rate": 0.00043630666224941624,
      "loss": 2.0772,
      "step": 80611
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.8882394433021545,
      "learning_rate": 0.0004363030182803833,
      "loss": 2.116,
      "step": 80612
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.7979082465171814,
      "learning_rate": 0.000436299374286009,
      "loss": 2.4813,
      "step": 80613
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.7691401243209839,
      "learning_rate": 0.0004362957302662942,
      "loss": 2.1413,
      "step": 80614
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.7878545522689819,
      "learning_rate": 0.00043629208622123937,
      "loss": 2.1988,
      "step": 80615
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.7672328948974609,
      "learning_rate": 0.0004362884421508452,
      "loss": 2.3864,
      "step": 80616
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.7676973342895508,
      "learning_rate": 0.00043628479805511255,
      "loss": 2.1659,
      "step": 80617
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.7729012370109558,
      "learning_rate": 0.000436281153934042,
      "loss": 2.3916,
      "step": 80618
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.937711238861084,
      "learning_rate": 0.0004362775097876341,
      "loss": 2.3634,
      "step": 80619
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.8274226188659668,
      "learning_rate": 0.0004362738656158897,
      "loss": 2.032,
      "step": 80620
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.8757466673851013,
      "learning_rate": 0.00043627022141880944,
      "loss": 2.1514,
      "step": 80621
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.8237902522087097,
      "learning_rate": 0.0004362665771963939,
      "loss": 2.1553,
      "step": 80622
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.7737563848495483,
      "learning_rate": 0.00043626293294864394,
      "loss": 2.4011,
      "step": 80623
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.8111019134521484,
      "learning_rate": 0.00043625928867556006,
      "loss": 2.2488,
      "step": 80624
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.7666017413139343,
      "learning_rate": 0.00043625564437714293,
      "loss": 2.1962,
      "step": 80625
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.7520456314086914,
      "learning_rate": 0.0004362520000533935,
      "loss": 2.1997,
      "step": 80626
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.7510603070259094,
      "learning_rate": 0.00043624835570431206,
      "loss": 2.095,
      "step": 80627
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.864966630935669,
      "learning_rate": 0.00043624471132989957,
      "loss": 2.0607,
      "step": 80628
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.7981295585632324,
      "learning_rate": 0.00043624106693015665,
      "loss": 2.1653,
      "step": 80629
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.8633807301521301,
      "learning_rate": 0.0004362374225050839,
      "loss": 2.0516,
      "step": 80630
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.755395770072937,
      "learning_rate": 0.000436233778054682,
      "loss": 2.1122,
      "step": 80631
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.7962877750396729,
      "learning_rate": 0.0004362301335789518,
      "loss": 1.9943,
      "step": 80632
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.7471007108688354,
      "learning_rate": 0.0004362264890778937,
      "loss": 2.0005,
      "step": 80633
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.8685730695724487,
      "learning_rate": 0.00043622284455150853,
      "loss": 2.2184,
      "step": 80634
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.9726995825767517,
      "learning_rate": 0.00043621919999979706,
      "loss": 2.2624,
      "step": 80635
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.734944224357605,
      "learning_rate": 0.0004362155554227598,
      "loss": 2.2296,
      "step": 80636
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.8659456372261047,
      "learning_rate": 0.0004362119108203975,
      "loss": 2.0884,
      "step": 80637
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.8173820376396179,
      "learning_rate": 0.0004362082661927109,
      "loss": 2.1401,
      "step": 80638
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.8938267827033997,
      "learning_rate": 0.00043620462153970055,
      "loss": 2.0561,
      "step": 80639
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.9173458218574524,
      "learning_rate": 0.00043620097686136716,
      "loss": 2.1108,
      "step": 80640
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.8468862175941467,
      "learning_rate": 0.0004361973321577115,
      "loss": 2.4519,
      "step": 80641
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.9375051259994507,
      "learning_rate": 0.0004361936874287342,
      "loss": 2.3011,
      "step": 80642
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.807717502117157,
      "learning_rate": 0.00043619004267443585,
      "loss": 2.1464,
      "step": 80643
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.8696248531341553,
      "learning_rate": 0.0004361863978948172,
      "loss": 2.3096,
      "step": 80644
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.878768265247345,
      "learning_rate": 0.000436182753089879,
      "loss": 2.3224,
      "step": 80645
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.7130535244941711,
      "learning_rate": 0.00043617910825962183,
      "loss": 2.1648,
      "step": 80646
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.8229228854179382,
      "learning_rate": 0.0004361754634040464,
      "loss": 2.093,
      "step": 80647
    },
    {
      "epoch": 1.05,
      "grad_norm": 1.0133310556411743,
      "learning_rate": 0.00043617181852315335,
      "loss": 2.254,
      "step": 80648
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.880075216293335,
      "learning_rate": 0.00043616817361694343,
      "loss": 2.2616,
      "step": 80649
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.8478265404701233,
      "learning_rate": 0.00043616452868541723,
      "loss": 2.285,
      "step": 80650
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.8659898042678833,
      "learning_rate": 0.00043616088372857545,
      "loss": 2.2652,
      "step": 80651
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.831018328666687,
      "learning_rate": 0.0004361572387464189,
      "loss": 2.1809,
      "step": 80652
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.8018436431884766,
      "learning_rate": 0.00043615359373894803,
      "loss": 2.0116,
      "step": 80653
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.7384896874427795,
      "learning_rate": 0.0004361499487061637,
      "loss": 2.1096,
      "step": 80654
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.8603835701942444,
      "learning_rate": 0.00043614630364806654,
      "loss": 2.0957,
      "step": 80655
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.9303752183914185,
      "learning_rate": 0.00043614265856465717,
      "loss": 2.203,
      "step": 80656
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.8647971749305725,
      "learning_rate": 0.00043613901345593634,
      "loss": 2.004,
      "step": 80657
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.7750965356826782,
      "learning_rate": 0.0004361353683219047,
      "loss": 2.0289,
      "step": 80658
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.8797058463096619,
      "learning_rate": 0.00043613172316256296,
      "loss": 2.3406,
      "step": 80659
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.7750368118286133,
      "learning_rate": 0.00043612807797791165,
      "loss": 2.0882,
      "step": 80660
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.7908021807670593,
      "learning_rate": 0.0004361244327679516,
      "loss": 2.3745,
      "step": 80661
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.9752767086029053,
      "learning_rate": 0.0004361207875326835,
      "loss": 2.01,
      "step": 80662
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.9403235912322998,
      "learning_rate": 0.00043611714227210797,
      "loss": 2.3973,
      "step": 80663
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.8763086795806885,
      "learning_rate": 0.00043611349698622563,
      "loss": 2.1302,
      "step": 80664
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.9396150708198547,
      "learning_rate": 0.00043610985167503725,
      "loss": 2.047,
      "step": 80665
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.8988196849822998,
      "learning_rate": 0.00043610620633854354,
      "loss": 2.1558,
      "step": 80666
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.8462104201316833,
      "learning_rate": 0.00043610256097674504,
      "loss": 2.3072,
      "step": 80667
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.8457156419754028,
      "learning_rate": 0.0004360989155896425,
      "loss": 2.2205,
      "step": 80668
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.8144652247428894,
      "learning_rate": 0.0004360952701772367,
      "loss": 2.3292,
      "step": 80669
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.8506457209587097,
      "learning_rate": 0.0004360916247395281,
      "loss": 2.184,
      "step": 80670
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.8188347220420837,
      "learning_rate": 0.00043608797927651756,
      "loss": 2.1628,
      "step": 80671
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.778123140335083,
      "learning_rate": 0.00043608433378820573,
      "loss": 2.2936,
      "step": 80672
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.9630226492881775,
      "learning_rate": 0.00043608068827459323,
      "loss": 2.1532,
      "step": 80673
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.8694689273834229,
      "learning_rate": 0.00043607704273568076,
      "loss": 2.0306,
      "step": 80674
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.9592381715774536,
      "learning_rate": 0.000436073397171469,
      "loss": 2.0996,
      "step": 80675
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.9225311279296875,
      "learning_rate": 0.00043606975158195854,
      "loss": 2.3777,
      "step": 80676
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.8453620672225952,
      "learning_rate": 0.00043606610596715024,
      "loss": 1.9666,
      "step": 80677
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.849933385848999,
      "learning_rate": 0.0004360624603270447,
      "loss": 2.0455,
      "step": 80678
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.8914588689804077,
      "learning_rate": 0.0004360588146616426,
      "loss": 2.1508,
      "step": 80679
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.9584825038909912,
      "learning_rate": 0.0004360551689709444,
      "loss": 2.1576,
      "step": 80680
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.8017092943191528,
      "learning_rate": 0.0004360515232549512,
      "loss": 2.2216,
      "step": 80681
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.8565805554389954,
      "learning_rate": 0.00043604787751366336,
      "loss": 2.3625,
      "step": 80682
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.737115740776062,
      "learning_rate": 0.00043604423174708166,
      "loss": 2.2312,
      "step": 80683
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.8165708184242249,
      "learning_rate": 0.0004360405859552068,
      "loss": 2.3083,
      "step": 80684
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.8538922071456909,
      "learning_rate": 0.00043603694013803936,
      "loss": 2.0643,
      "step": 80685
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.8481621146202087,
      "learning_rate": 0.0004360332942955801,
      "loss": 2.2433,
      "step": 80686
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.7786194682121277,
      "learning_rate": 0.0004360296484278297,
      "loss": 1.9735,
      "step": 80687
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.8114707469940186,
      "learning_rate": 0.00043602600253478884,
      "loss": 2.1783,
      "step": 80688
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.7312656044960022,
      "learning_rate": 0.00043602235661645815,
      "loss": 1.9651,
      "step": 80689
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.8701871037483215,
      "learning_rate": 0.0004360187106728384,
      "loss": 2.2313,
      "step": 80690
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.7966622710227966,
      "learning_rate": 0.00043601506470393014,
      "loss": 2.0683,
      "step": 80691
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.9399929642677307,
      "learning_rate": 0.0004360114187097341,
      "loss": 2.1532,
      "step": 80692
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.7503564357757568,
      "learning_rate": 0.00043600777269025106,
      "loss": 2.1589,
      "step": 80693
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.7726930379867554,
      "learning_rate": 0.0004360041266454815,
      "loss": 2.3833,
      "step": 80694
    },
    {
      "epoch": 1.05,
      "grad_norm": 1.0312767028808594,
      "learning_rate": 0.0004360004805754262,
      "loss": 2.295,
      "step": 80695
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.8087409734725952,
      "learning_rate": 0.00043599683448008596,
      "loss": 2.135,
      "step": 80696
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.7569006681442261,
      "learning_rate": 0.00043599318835946125,
      "loss": 2.3998,
      "step": 80697
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.9701641201972961,
      "learning_rate": 0.00043598954221355283,
      "loss": 2.3716,
      "step": 80698
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.8045285940170288,
      "learning_rate": 0.00043598589604236146,
      "loss": 2.0755,
      "step": 80699
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.9184910655021667,
      "learning_rate": 0.0004359822498458877,
      "loss": 2.1274,
      "step": 80700
    },
    {
      "epoch": 1.05,
      "grad_norm": 1.036149501800537,
      "learning_rate": 0.00043597860362413224,
      "loss": 2.2946,
      "step": 80701
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.8350536227226257,
      "learning_rate": 0.0004359749573770959,
      "loss": 2.2459,
      "step": 80702
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.7592102885246277,
      "learning_rate": 0.0004359713111047791,
      "loss": 2.1897,
      "step": 80703
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.8022568225860596,
      "learning_rate": 0.00043596766480718277,
      "loss": 2.416,
      "step": 80704
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.8182395696640015,
      "learning_rate": 0.0004359640184843074,
      "loss": 2.1703,
      "step": 80705
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.8087621331214905,
      "learning_rate": 0.0004359603721361539,
      "loss": 2.1215,
      "step": 80706
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.7686275243759155,
      "learning_rate": 0.0004359567257627227,
      "loss": 1.8719,
      "step": 80707
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.7866872549057007,
      "learning_rate": 0.00043595307936401454,
      "loss": 2.2193,
      "step": 80708
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.8521789908409119,
      "learning_rate": 0.00043594943294003027,
      "loss": 2.1678,
      "step": 80709
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.7838684320449829,
      "learning_rate": 0.0004359457864907703,
      "loss": 2.1002,
      "step": 80710
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.8245274424552917,
      "learning_rate": 0.00043594214001623546,
      "loss": 2.373,
      "step": 80711
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.8752480149269104,
      "learning_rate": 0.0004359384935164265,
      "loss": 2.0909,
      "step": 80712
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.8382777571678162,
      "learning_rate": 0.00043593484699134394,
      "loss": 2.5962,
      "step": 80713
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.9862877130508423,
      "learning_rate": 0.00043593120044098854,
      "loss": 2.1063,
      "step": 80714
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.7984893918037415,
      "learning_rate": 0.000435927553865361,
      "loss": 2.1844,
      "step": 80715
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.8363538980484009,
      "learning_rate": 0.0004359239072644619,
      "loss": 2.0907,
      "step": 80716
    },
    {
      "epoch": 1.05,
      "grad_norm": 1.0099544525146484,
      "learning_rate": 0.00043592026063829194,
      "loss": 1.9931,
      "step": 80717
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.8290232419967651,
      "learning_rate": 0.000435916613986852,
      "loss": 2.1617,
      "step": 80718
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.8555048704147339,
      "learning_rate": 0.00043591296731014246,
      "loss": 2.1286,
      "step": 80719
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.9480733871459961,
      "learning_rate": 0.0004359093206081641,
      "loss": 1.9296,
      "step": 80720
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.7989718317985535,
      "learning_rate": 0.00043590567388091775,
      "loss": 2.2007,
      "step": 80721
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.8269413709640503,
      "learning_rate": 0.0004359020271284039,
      "loss": 2.0834,
      "step": 80722
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.8240319490432739,
      "learning_rate": 0.0004358983803506233,
      "loss": 2.2352,
      "step": 80723
    },
    {
      "epoch": 1.05,
      "grad_norm": 1.0969122648239136,
      "learning_rate": 0.00043589473354757667,
      "loss": 2.2682,
      "step": 80724
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.8110664486885071,
      "learning_rate": 0.00043589108671926463,
      "loss": 2.2287,
      "step": 80725
    },
    {
      "epoch": 1.05,
      "grad_norm": 1.063841462135315,
      "learning_rate": 0.00043588743986568786,
      "loss": 2.3568,
      "step": 80726
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.892440140247345,
      "learning_rate": 0.00043588379298684706,
      "loss": 2.2507,
      "step": 80727
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.8230469822883606,
      "learning_rate": 0.0004358801460827429,
      "loss": 2.1232,
      "step": 80728
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.777187168598175,
      "learning_rate": 0.000435876499153376,
      "loss": 2.1471,
      "step": 80729
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.783500075340271,
      "learning_rate": 0.0004358728521987472,
      "loss": 2.1654,
      "step": 80730
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.7784695029258728,
      "learning_rate": 0.00043586920521885696,
      "loss": 2.1206,
      "step": 80731
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.8103877305984497,
      "learning_rate": 0.0004358655582137061,
      "loss": 2.1419,
      "step": 80732
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.8130342960357666,
      "learning_rate": 0.0004358619111832953,
      "loss": 2.3187,
      "step": 80733
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.8412301540374756,
      "learning_rate": 0.0004358582641276252,
      "loss": 2.0192,
      "step": 80734
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.7982290983200073,
      "learning_rate": 0.0004358546170466965,
      "loss": 2.1738,
      "step": 80735
    },
    {
      "epoch": 1.05,
      "grad_norm": 1.0914667844772339,
      "learning_rate": 0.0004358509699405099,
      "loss": 2.1574,
      "step": 80736
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.8133437633514404,
      "learning_rate": 0.00043584732280906594,
      "loss": 2.3555,
      "step": 80737
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.9733473658561707,
      "learning_rate": 0.0004358436756523654,
      "loss": 2.1199,
      "step": 80738
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.7869798541069031,
      "learning_rate": 0.00043584002847040903,
      "loss": 2.1657,
      "step": 80739
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.909172534942627,
      "learning_rate": 0.0004358363812631974,
      "loss": 2.145,
      "step": 80740
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.8009662628173828,
      "learning_rate": 0.0004358327340307312,
      "loss": 2.2915,
      "step": 80741
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.8689876794815063,
      "learning_rate": 0.0004358290867730112,
      "loss": 2.1136,
      "step": 80742
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.8686897158622742,
      "learning_rate": 0.0004358254394900379,
      "loss": 2.1963,
      "step": 80743
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.8346531391143799,
      "learning_rate": 0.00043582179218181215,
      "loss": 2.56,
      "step": 80744
    },
    {
      "epoch": 1.05,
      "grad_norm": 1.5812571048736572,
      "learning_rate": 0.0004358181448483346,
      "loss": 2.257,
      "step": 80745
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.8021209836006165,
      "learning_rate": 0.0004358144974896058,
      "loss": 2.0691,
      "step": 80746
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.7860500812530518,
      "learning_rate": 0.00043581085010562664,
      "loss": 2.2187,
      "step": 80747
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.7866401076316833,
      "learning_rate": 0.00043580720269639756,
      "loss": 2.1035,
      "step": 80748
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.9455203413963318,
      "learning_rate": 0.0004358035552619194,
      "loss": 2.0105,
      "step": 80749
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.8590476512908936,
      "learning_rate": 0.0004357999078021928,
      "loss": 2.2093,
      "step": 80750
    },
    {
      "epoch": 1.05,
      "grad_norm": 1.0766220092773438,
      "learning_rate": 0.0004357962603172185,
      "loss": 2.0652,
      "step": 80751
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.8589761257171631,
      "learning_rate": 0.00043579261280699705,
      "loss": 2.3005,
      "step": 80752
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.8229376077651978,
      "learning_rate": 0.00043578896527152915,
      "loss": 2.0579,
      "step": 80753
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.7992936968803406,
      "learning_rate": 0.0004357853177108156,
      "loss": 2.0968,
      "step": 80754
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.7128344178199768,
      "learning_rate": 0.00043578167012485697,
      "loss": 2.1479,
      "step": 80755
    },
    {
      "epoch": 1.05,
      "grad_norm": 1.0568089485168457,
      "learning_rate": 0.0004357780225136539,
      "loss": 2.3663,
      "step": 80756
    },
    {
      "epoch": 1.05,
      "grad_norm": 1.030066728591919,
      "learning_rate": 0.00043577437487720723,
      "loss": 2.087,
      "step": 80757
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.8303022980690002,
      "learning_rate": 0.00043577072721551744,
      "loss": 2.396,
      "step": 80758
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.8829760551452637,
      "learning_rate": 0.0004357670795285854,
      "loss": 2.051,
      "step": 80759
    },
    {
      "epoch": 1.05,
      "grad_norm": 1.0919339656829834,
      "learning_rate": 0.0004357634318164117,
      "loss": 2.4832,
      "step": 80760
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.808775007724762,
      "learning_rate": 0.000435759784078997,
      "loss": 2.3561,
      "step": 80761
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.9141871929168701,
      "learning_rate": 0.00043575613631634193,
      "loss": 2.2683,
      "step": 80762
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.8950434923171997,
      "learning_rate": 0.0004357524885284473,
      "loss": 2.1048,
      "step": 80763
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.7348358035087585,
      "learning_rate": 0.0004357488407153137,
      "loss": 2.1915,
      "step": 80764
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.9882128238677979,
      "learning_rate": 0.00043574519287694176,
      "loss": 2.1057,
      "step": 80765
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.8132657408714294,
      "learning_rate": 0.00043574154501333233,
      "loss": 2.1601,
      "step": 80766
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.8086708188056946,
      "learning_rate": 0.0004357378971244859,
      "loss": 2.2066,
      "step": 80767
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.734188973903656,
      "learning_rate": 0.00043573424921040325,
      "loss": 2.4608,
      "step": 80768
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.8519744873046875,
      "learning_rate": 0.00043573060127108517,
      "loss": 2.1124,
      "step": 80769
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.9320570230484009,
      "learning_rate": 0.00043572695330653207,
      "loss": 2.1038,
      "step": 80770
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.849888026714325,
      "learning_rate": 0.00043572330531674476,
      "loss": 2.2067,
      "step": 80771
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.900075376033783,
      "learning_rate": 0.000435719657301724,
      "loss": 2.2373,
      "step": 80772
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.8230928182601929,
      "learning_rate": 0.0004357160092614704,
      "loss": 2.054,
      "step": 80773
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.8612036108970642,
      "learning_rate": 0.0004357123611959846,
      "loss": 2.2538,
      "step": 80774
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.8365136981010437,
      "learning_rate": 0.0004357087131052672,
      "loss": 2.0522,
      "step": 80775
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.8129044771194458,
      "learning_rate": 0.00043570506498931915,
      "loss": 2.1576,
      "step": 80776
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.7773079872131348,
      "learning_rate": 0.00043570141684814095,
      "loss": 2.2955,
      "step": 80777
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.7920240163803101,
      "learning_rate": 0.00043569776868173325,
      "loss": 2.0911,
      "step": 80778
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.8272913098335266,
      "learning_rate": 0.0004356941204900968,
      "loss": 2.1037,
      "step": 80779
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.9799987077713013,
      "learning_rate": 0.0004356904722732322,
      "loss": 2.1564,
      "step": 80780
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.838239848613739,
      "learning_rate": 0.00043568682403114016,
      "loss": 2.3441,
      "step": 80781
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.783577024936676,
      "learning_rate": 0.00043568317576382143,
      "loss": 1.9474,
      "step": 80782
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.8748559951782227,
      "learning_rate": 0.00043567952747127674,
      "loss": 2.1574,
      "step": 80783
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.7760826349258423,
      "learning_rate": 0.0004356758791535065,
      "loss": 2.2758,
      "step": 80784
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.7732468843460083,
      "learning_rate": 0.00043567223081051167,
      "loss": 2.0461,
      "step": 80785
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.7857440114021301,
      "learning_rate": 0.00043566858244229275,
      "loss": 2.1345,
      "step": 80786
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.8261607885360718,
      "learning_rate": 0.0004356649340488504,
      "loss": 2.1667,
      "step": 80787
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.7821928262710571,
      "learning_rate": 0.0004356612856301855,
      "loss": 2.1207,
      "step": 80788
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.7585381865501404,
      "learning_rate": 0.00043565763718629856,
      "loss": 2.0515,
      "step": 80789
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.7853035926818848,
      "learning_rate": 0.0004356539887171903,
      "loss": 2.1316,
      "step": 80790
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.8907716870307922,
      "learning_rate": 0.0004356503402228615,
      "loss": 2.13,
      "step": 80791
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.8141002655029297,
      "learning_rate": 0.0004356466917033126,
      "loss": 2.0754,
      "step": 80792
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.7832050323486328,
      "learning_rate": 0.00043564304315854444,
      "loss": 2.0623,
      "step": 80793
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.7989069223403931,
      "learning_rate": 0.00043563939458855777,
      "loss": 2.2645,
      "step": 80794
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.848020613193512,
      "learning_rate": 0.00043563574599335316,
      "loss": 2.2644,
      "step": 80795
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.7734694480895996,
      "learning_rate": 0.0004356320973729312,
      "loss": 1.9537,
      "step": 80796
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.7865129709243774,
      "learning_rate": 0.0004356284487272928,
      "loss": 2.3631,
      "step": 80797
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.7211293578147888,
      "learning_rate": 0.00043562480005643843,
      "loss": 2.1058,
      "step": 80798
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.8323601484298706,
      "learning_rate": 0.0004356211513603688,
      "loss": 2.2577,
      "step": 80799
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.8148553967475891,
      "learning_rate": 0.0004356175026390848,
      "loss": 2.2301,
      "step": 80800
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.7453306317329407,
      "learning_rate": 0.0004356138538925868,
      "loss": 2.1605,
      "step": 80801
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.8324166536331177,
      "learning_rate": 0.00043561020512087565,
      "loss": 2.2886,
      "step": 80802
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.8543303608894348,
      "learning_rate": 0.0004356065563239521,
      "loss": 2.2666,
      "step": 80803
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.8186496496200562,
      "learning_rate": 0.0004356029075018166,
      "loss": 2.1085,
      "step": 80804
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.9107299447059631,
      "learning_rate": 0.00043559925865446997,
      "loss": 2.1779,
      "step": 80805
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.817068874835968,
      "learning_rate": 0.000435595609781913,
      "loss": 2.2321,
      "step": 80806
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.7771966457366943,
      "learning_rate": 0.00043559196088414616,
      "loss": 2.0606,
      "step": 80807
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.7077457904815674,
      "learning_rate": 0.0004355883119611702,
      "loss": 2.1641,
      "step": 80808
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.7715538144111633,
      "learning_rate": 0.0004355846630129859,
      "loss": 2.0107,
      "step": 80809
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.7760226130485535,
      "learning_rate": 0.0004355810140395938,
      "loss": 2.0966,
      "step": 80810
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.777222752571106,
      "learning_rate": 0.00043557736504099454,
      "loss": 2.1737,
      "step": 80811
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.8191661238670349,
      "learning_rate": 0.00043557371601718903,
      "loss": 2.2194,
      "step": 80812
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.7774333953857422,
      "learning_rate": 0.0004355700669681778,
      "loss": 2.1161,
      "step": 80813
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.7581984996795654,
      "learning_rate": 0.00043556641789396143,
      "loss": 2.304,
      "step": 80814
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.7731263041496277,
      "learning_rate": 0.0004355627687945408,
      "loss": 2.2164,
      "step": 80815
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.7614840865135193,
      "learning_rate": 0.00043555911966991647,
      "loss": 2.211,
      "step": 80816
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.7827906608581543,
      "learning_rate": 0.0004355554705200891,
      "loss": 2.3472,
      "step": 80817
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.7569001913070679,
      "learning_rate": 0.00043555182134505946,
      "loss": 2.0204,
      "step": 80818
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.762519359588623,
      "learning_rate": 0.00043554817214482813,
      "loss": 2.3182,
      "step": 80819
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.7873206734657288,
      "learning_rate": 0.00043554452291939583,
      "loss": 2.4272,
      "step": 80820
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.780098557472229,
      "learning_rate": 0.0004355408736687633,
      "loss": 2.3122,
      "step": 80821
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.7554211020469666,
      "learning_rate": 0.0004355372243929311,
      "loss": 2.2154,
      "step": 80822
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.7565532922744751,
      "learning_rate": 0.00043553357509190004,
      "loss": 2.0175,
      "step": 80823
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.7697260975837708,
      "learning_rate": 0.00043552992576567065,
      "loss": 2.3476,
      "step": 80824
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.8153089880943298,
      "learning_rate": 0.00043552627641424375,
      "loss": 2.3135,
      "step": 80825
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.8300275802612305,
      "learning_rate": 0.00043552262703761994,
      "loss": 2.2461,
      "step": 80826
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.7381420135498047,
      "learning_rate": 0.0004355189776357999,
      "loss": 1.9662,
      "step": 80827
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.7878348231315613,
      "learning_rate": 0.00043551532820878435,
      "loss": 2.1758,
      "step": 80828
    },
    {
      "epoch": 1.05,
      "grad_norm": 1.0899345874786377,
      "learning_rate": 0.0004355116787565739,
      "loss": 2.0767,
      "step": 80829
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.8964205980300903,
      "learning_rate": 0.00043550802927916934,
      "loss": 2.0132,
      "step": 80830
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.9533700346946716,
      "learning_rate": 0.0004355043797765712,
      "loss": 2.1431,
      "step": 80831
    },
    {
      "epoch": 1.05,
      "grad_norm": 1.038256049156189,
      "learning_rate": 0.00043550073024878035,
      "loss": 2.2674,
      "step": 80832
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.8149707913398743,
      "learning_rate": 0.00043549708069579723,
      "loss": 2.3567,
      "step": 80833
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.7758413553237915,
      "learning_rate": 0.00043549343111762273,
      "loss": 2.3563,
      "step": 80834
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.8895781636238098,
      "learning_rate": 0.0004354897815142574,
      "loss": 1.8693,
      "step": 80835
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.8560213446617126,
      "learning_rate": 0.00043548613188570197,
      "loss": 2.1799,
      "step": 80836
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.8818995952606201,
      "learning_rate": 0.00043548248223195717,
      "loss": 2.1366,
      "step": 80837
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.8095658421516418,
      "learning_rate": 0.00043547883255302353,
      "loss": 2.241,
      "step": 80838
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.8939872980117798,
      "learning_rate": 0.0004354751828489019,
      "loss": 2.262,
      "step": 80839
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.7550925612449646,
      "learning_rate": 0.0004354715331195928,
      "loss": 2.15,
      "step": 80840
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.768283486366272,
      "learning_rate": 0.00043546788336509704,
      "loss": 2.3306,
      "step": 80841
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.7162880301475525,
      "learning_rate": 0.00043546423358541516,
      "loss": 2.1999,
      "step": 80842
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.7658731341362,
      "learning_rate": 0.000435460583780548,
      "loss": 2.0503,
      "step": 80843
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.8883666396141052,
      "learning_rate": 0.0004354569339504961,
      "loss": 2.2478,
      "step": 80844
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.8053423762321472,
      "learning_rate": 0.0004354532840952603,
      "loss": 2.2549,
      "step": 80845
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.7424695491790771,
      "learning_rate": 0.00043544963421484115,
      "loss": 2.1261,
      "step": 80846
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.8312833309173584,
      "learning_rate": 0.00043544598430923933,
      "loss": 2.0856,
      "step": 80847
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.8189828395843506,
      "learning_rate": 0.00043544233437845545,
      "loss": 2.0608,
      "step": 80848
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.8224205374717712,
      "learning_rate": 0.00043543868442249043,
      "loss": 2.3552,
      "step": 80849
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.7731678485870361,
      "learning_rate": 0.0004354350344413447,
      "loss": 2.1732,
      "step": 80850
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.8990111947059631,
      "learning_rate": 0.00043543138443501904,
      "loss": 2.1809,
      "step": 80851
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.8737536668777466,
      "learning_rate": 0.0004354277344035143,
      "loss": 2.2506,
      "step": 80852
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.7599825263023376,
      "learning_rate": 0.00043542408434683085,
      "loss": 2.2372,
      "step": 80853
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.7708587050437927,
      "learning_rate": 0.00043542043426496947,
      "loss": 2.3496,
      "step": 80854
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.9496512413024902,
      "learning_rate": 0.0004354167841579309,
      "loss": 2.0416,
      "step": 80855
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.8609330058097839,
      "learning_rate": 0.00043541313402571587,
      "loss": 2.3676,
      "step": 80856
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.7635971903800964,
      "learning_rate": 0.0004354094838683249,
      "loss": 2.4462,
      "step": 80857
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.7852855920791626,
      "learning_rate": 0.0004354058336857588,
      "loss": 2.3515,
      "step": 80858
    },
    {
      "epoch": 1.05,
      "grad_norm": 1.254806399345398,
      "learning_rate": 0.00043540218347801825,
      "loss": 2.0389,
      "step": 80859
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.8102823495864868,
      "learning_rate": 0.00043539853324510376,
      "loss": 2.4536,
      "step": 80860
    },
    {
      "epoch": 1.05,
      "grad_norm": 1.01824152469635,
      "learning_rate": 0.00043539488298701616,
      "loss": 2.2497,
      "step": 80861
    },
    {
      "epoch": 1.05,
      "grad_norm": 1.110460877418518,
      "learning_rate": 0.0004353912327037561,
      "loss": 2.1704,
      "step": 80862
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.9313144683837891,
      "learning_rate": 0.00043538758239532427,
      "loss": 2.2592,
      "step": 80863
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.8819419741630554,
      "learning_rate": 0.0004353839320617214,
      "loss": 2.3478,
      "step": 80864
    },
    {
      "epoch": 1.05,
      "grad_norm": 1.1696604490280151,
      "learning_rate": 0.000435380281702948,
      "loss": 2.0957,
      "step": 80865
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.9587433934211731,
      "learning_rate": 0.00043537663131900485,
      "loss": 2.2081,
      "step": 80866
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.7684641480445862,
      "learning_rate": 0.0004353729809098927,
      "loss": 2.051,
      "step": 80867
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.8355916738510132,
      "learning_rate": 0.0004353693304756121,
      "loss": 2.3159,
      "step": 80868
    },
    {
      "epoch": 1.05,
      "grad_norm": 1.1776398420333862,
      "learning_rate": 0.00043536568001616377,
      "loss": 2.1635,
      "step": 80869
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.7916101813316345,
      "learning_rate": 0.00043536202953154844,
      "loss": 1.9546,
      "step": 80870
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.764815628528595,
      "learning_rate": 0.0004353583790217667,
      "loss": 2.3848,
      "step": 80871
    },
    {
      "epoch": 1.05,
      "grad_norm": 1.0127822160720825,
      "learning_rate": 0.0004353547284868194,
      "loss": 2.2221,
      "step": 80872
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.8491482734680176,
      "learning_rate": 0.0004353510779267071,
      "loss": 2.18,
      "step": 80873
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.7839676737785339,
      "learning_rate": 0.00043534742734143035,
      "loss": 2.3174,
      "step": 80874
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.7759225368499756,
      "learning_rate": 0.00043534377673099,
      "loss": 2.2789,
      "step": 80875
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.9114322066307068,
      "learning_rate": 0.0004353401260953868,
      "loss": 2.2851,
      "step": 80876
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.7383915781974792,
      "learning_rate": 0.0004353364754346212,
      "loss": 2.2435,
      "step": 80877
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.7715353965759277,
      "learning_rate": 0.00043533282474869395,
      "loss": 2.375,
      "step": 80878
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.823226809501648,
      "learning_rate": 0.0004353291740376059,
      "loss": 2.2316,
      "step": 80879
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.8206883072853088,
      "learning_rate": 0.0004353255233013575,
      "loss": 2.2385,
      "step": 80880
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.7594742178916931,
      "learning_rate": 0.0004353218725399496,
      "loss": 2.2396,
      "step": 80881
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.9433055520057678,
      "learning_rate": 0.0004353182217533828,
      "loss": 2.1929,
      "step": 80882
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.7990105152130127,
      "learning_rate": 0.00043531457094165773,
      "loss": 2.0881,
      "step": 80883
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.7550888061523438,
      "learning_rate": 0.0004353109201047752,
      "loss": 2.015,
      "step": 80884
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.8850327134132385,
      "learning_rate": 0.0004353072692427358,
      "loss": 2.1458,
      "step": 80885
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.7980760931968689,
      "learning_rate": 0.00043530361835554017,
      "loss": 1.9595,
      "step": 80886
    },
    {
      "epoch": 1.05,
      "grad_norm": 1.0995407104492188,
      "learning_rate": 0.0004352999674431891,
      "loss": 2.1732,
      "step": 80887
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.7706223726272583,
      "learning_rate": 0.0004352963165056832,
      "loss": 2.4995,
      "step": 80888
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.8461070656776428,
      "learning_rate": 0.00043529266554302314,
      "loss": 2.2592,
      "step": 80889
    },
    {
      "epoch": 1.05,
      "grad_norm": 1.1634132862091064,
      "learning_rate": 0.00043528901455520963,
      "loss": 2.1493,
      "step": 80890
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.7411150336265564,
      "learning_rate": 0.0004352853635422434,
      "loss": 2.0295,
      "step": 80891
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.8862562775611877,
      "learning_rate": 0.000435281712504125,
      "loss": 2.3086,
      "step": 80892
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.9003999829292297,
      "learning_rate": 0.00043527806144085516,
      "loss": 2.22,
      "step": 80893
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.808207094669342,
      "learning_rate": 0.00043527441035243463,
      "loss": 2.2721,
      "step": 80894
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.7800478935241699,
      "learning_rate": 0.00043527075923886397,
      "loss": 2.0721,
      "step": 80895
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.9207016825675964,
      "learning_rate": 0.00043526710810014394,
      "loss": 2.2587,
      "step": 80896
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.821489155292511,
      "learning_rate": 0.0004352634569362753,
      "loss": 2.0715,
      "step": 80897
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.8731721639633179,
      "learning_rate": 0.0004352598057472585,
      "loss": 2.0904,
      "step": 80898
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.7653831839561462,
      "learning_rate": 0.0004352561545330944,
      "loss": 1.9312,
      "step": 80899
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.8556411266326904,
      "learning_rate": 0.00043525250329378367,
      "loss": 2.1537,
      "step": 80900
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.8745763897895813,
      "learning_rate": 0.00043524885202932687,
      "loss": 2.3691,
      "step": 80901
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.9549430012702942,
      "learning_rate": 0.0004352452007397248,
      "loss": 2.2878,
      "step": 80902
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.8617876768112183,
      "learning_rate": 0.0004352415494249781,
      "loss": 2.2722,
      "step": 80903
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.8263710141181946,
      "learning_rate": 0.0004352378980850874,
      "loss": 2.2117,
      "step": 80904
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.7699521780014038,
      "learning_rate": 0.00043523424672005346,
      "loss": 2.1643,
      "step": 80905
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.6901536583900452,
      "learning_rate": 0.00043523059532987686,
      "loss": 1.9989,
      "step": 80906
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.8833721876144409,
      "learning_rate": 0.0004352269439145584,
      "loss": 1.9753,
      "step": 80907
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.799312174320221,
      "learning_rate": 0.0004352232924740987,
      "loss": 2.4192,
      "step": 80908
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.8007022738456726,
      "learning_rate": 0.0004352196410084984,
      "loss": 2.0642,
      "step": 80909
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.8614115118980408,
      "learning_rate": 0.0004352159895177583,
      "loss": 2.2694,
      "step": 80910
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.7471598386764526,
      "learning_rate": 0.00043521233800187896,
      "loss": 2.2195,
      "step": 80911
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.7860097289085388,
      "learning_rate": 0.000435208686460861,
      "loss": 2.191,
      "step": 80912
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.7424548268318176,
      "learning_rate": 0.00043520503489470527,
      "loss": 2.2671,
      "step": 80913
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.8395339250564575,
      "learning_rate": 0.0004352013833034124,
      "loss": 2.1494,
      "step": 80914
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.7302716374397278,
      "learning_rate": 0.000435197731686983,
      "loss": 2.0521,
      "step": 80915
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.8456469774246216,
      "learning_rate": 0.0004351940800454177,
      "loss": 2.013,
      "step": 80916
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.8197437524795532,
      "learning_rate": 0.0004351904283787175,
      "loss": 1.9373,
      "step": 80917
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.7644168734550476,
      "learning_rate": 0.0004351867766868826,
      "loss": 2.344,
      "step": 80918
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.928979218006134,
      "learning_rate": 0.00043518312496991404,
      "loss": 2.3435,
      "step": 80919
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.8383451104164124,
      "learning_rate": 0.00043517947322781246,
      "loss": 2.3534,
      "step": 80920
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.8342673182487488,
      "learning_rate": 0.0004351758214605783,
      "loss": 2.0516,
      "step": 80921
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.7788219451904297,
      "learning_rate": 0.0004351721696682125,
      "loss": 2.321,
      "step": 80922
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.7375712990760803,
      "learning_rate": 0.00043516851785071564,
      "loss": 2.3017,
      "step": 80923
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.8952023983001709,
      "learning_rate": 0.00043516486600808847,
      "loss": 2.5183,
      "step": 80924
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.8456403613090515,
      "learning_rate": 0.0004351612141403315,
      "loss": 1.9765,
      "step": 80925
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.8063743710517883,
      "learning_rate": 0.0004351575622474455,
      "loss": 2.2474,
      "step": 80926
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.9859606623649597,
      "learning_rate": 0.00043515391032943117,
      "loss": 1.9837,
      "step": 80927
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.7950608730316162,
      "learning_rate": 0.00043515025838628923,
      "loss": 2.1809,
      "step": 80928
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.7782432436943054,
      "learning_rate": 0.00043514660641802026,
      "loss": 2.2362,
      "step": 80929
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.8837522864341736,
      "learning_rate": 0.00043514295442462496,
      "loss": 2.2988,
      "step": 80930
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.8558396697044373,
      "learning_rate": 0.00043513930240610414,
      "loss": 2.2145,
      "step": 80931
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.743855357170105,
      "learning_rate": 0.00043513565036245824,
      "loss": 2.2107,
      "step": 80932
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.823804497718811,
      "learning_rate": 0.0004351319982936881,
      "loss": 2.0655,
      "step": 80933
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.7956126928329468,
      "learning_rate": 0.0004351283461997945,
      "loss": 2.2509,
      "step": 80934
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.7786424160003662,
      "learning_rate": 0.0004351246940807779,
      "loss": 2.0901,
      "step": 80935
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.9494152069091797,
      "learning_rate": 0.00043512104193663896,
      "loss": 2.1545,
      "step": 80936
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.7686474323272705,
      "learning_rate": 0.00043511738976737864,
      "loss": 2.1629,
      "step": 80937
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.7738081812858582,
      "learning_rate": 0.00043511373757299735,
      "loss": 2.2177,
      "step": 80938
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.9024365544319153,
      "learning_rate": 0.0004351100853534959,
      "loss": 1.9663,
      "step": 80939
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.7577271461486816,
      "learning_rate": 0.00043510643310887493,
      "loss": 2.2021,
      "step": 80940
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.9520688652992249,
      "learning_rate": 0.0004351027808391352,
      "loss": 2.1247,
      "step": 80941
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.8104764223098755,
      "learning_rate": 0.00043509912854427715,
      "loss": 2.2834,
      "step": 80942
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.8385365009307861,
      "learning_rate": 0.0004350954762243018,
      "loss": 2.0171,
      "step": 80943
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.9212167859077454,
      "learning_rate": 0.00043509182387920946,
      "loss": 1.9982,
      "step": 80944
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.8610765933990479,
      "learning_rate": 0.00043508817150900115,
      "loss": 2.3308,
      "step": 80945
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.8345820307731628,
      "learning_rate": 0.00043508451911367736,
      "loss": 2.0505,
      "step": 80946
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.7924829125404358,
      "learning_rate": 0.0004350808666932388,
      "loss": 2.0339,
      "step": 80947
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.8951506018638611,
      "learning_rate": 0.00043507721424768613,
      "loss": 2.1077,
      "step": 80948
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.7467291355133057,
      "learning_rate": 0.00043507356177702004,
      "loss": 2.0936,
      "step": 80949
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.8167997598648071,
      "learning_rate": 0.0004350699092812413,
      "loss": 2.175,
      "step": 80950
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.8071781396865845,
      "learning_rate": 0.0004350662567603505,
      "loss": 2.0801,
      "step": 80951
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.8229609131813049,
      "learning_rate": 0.0004350626042143483,
      "loss": 2.3632,
      "step": 80952
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.7803617715835571,
      "learning_rate": 0.0004350589516432354,
      "loss": 2.033,
      "step": 80953
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.8373702168464661,
      "learning_rate": 0.0004350552990470125,
      "loss": 2.3511,
      "step": 80954
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.8414955139160156,
      "learning_rate": 0.00043505164642568034,
      "loss": 2.1727,
      "step": 80955
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.8225211501121521,
      "learning_rate": 0.0004350479937792395,
      "loss": 2.2124,
      "step": 80956
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.7636500000953674,
      "learning_rate": 0.0004350443411076906,
      "loss": 2.2384,
      "step": 80957
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.8317660689353943,
      "learning_rate": 0.00043504068841103454,
      "loss": 2.1078,
      "step": 80958
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.7611610293388367,
      "learning_rate": 0.0004350370356892718,
      "loss": 2.2689,
      "step": 80959
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.7913285493850708,
      "learning_rate": 0.0004350333829424031,
      "loss": 1.9473,
      "step": 80960
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.7911643385887146,
      "learning_rate": 0.0004350297301704292,
      "loss": 2.3112,
      "step": 80961
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.9029979109764099,
      "learning_rate": 0.00043502607737335065,
      "loss": 2.1166,
      "step": 80962
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.8210366368293762,
      "learning_rate": 0.0004350224245511682,
      "loss": 2.1299,
      "step": 80963
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.7871065735816956,
      "learning_rate": 0.00043501877170388267,
      "loss": 2.4803,
      "step": 80964
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.8512412309646606,
      "learning_rate": 0.00043501511883149447,
      "loss": 2.1355,
      "step": 80965
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.8389813899993896,
      "learning_rate": 0.0004350114659340044,
      "loss": 2.0566,
      "step": 80966
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.8030772805213928,
      "learning_rate": 0.00043500781301141324,
      "loss": 2.1894,
      "step": 80967
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.7213467955589294,
      "learning_rate": 0.0004350041600637215,
      "loss": 2.0069,
      "step": 80968
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.9071839451789856,
      "learning_rate": 0.00043500050709093,
      "loss": 2.3785,
      "step": 80969
    },
    {
      "epoch": 1.05,
      "grad_norm": 1.1209557056427002,
      "learning_rate": 0.0004349968540930394,
      "loss": 2.2132,
      "step": 80970
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.7925909757614136,
      "learning_rate": 0.00043499320107005026,
      "loss": 1.9449,
      "step": 80971
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.8434035778045654,
      "learning_rate": 0.00043498954802196336,
      "loss": 2.2323,
      "step": 80972
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.9766711592674255,
      "learning_rate": 0.0004349858949487793,
      "loss": 2.3484,
      "step": 80973
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.8010016083717346,
      "learning_rate": 0.0004349822418504989,
      "loss": 2.1163,
      "step": 80974
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.7978221774101257,
      "learning_rate": 0.00043497858872712264,
      "loss": 2.1862,
      "step": 80975
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.7968958616256714,
      "learning_rate": 0.0004349749355786514,
      "loss": 2.2209,
      "step": 80976
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.8532057404518127,
      "learning_rate": 0.0004349712824050858,
      "loss": 2.3058,
      "step": 80977
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.8164072036743164,
      "learning_rate": 0.0004349676292064264,
      "loss": 2.0696,
      "step": 80978
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.7527387142181396,
      "learning_rate": 0.00043496397598267404,
      "loss": 2.2508,
      "step": 80979
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.7556337118148804,
      "learning_rate": 0.0004349603227338293,
      "loss": 2.2331,
      "step": 80980
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.762406051158905,
      "learning_rate": 0.0004349566694598929,
      "loss": 2.2117,
      "step": 80981
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.8215426802635193,
      "learning_rate": 0.0004349530161608655,
      "loss": 2.2351,
      "step": 80982
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.8200104236602783,
      "learning_rate": 0.0004349493628367478,
      "loss": 2.4202,
      "step": 80983
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.8145044445991516,
      "learning_rate": 0.00043494570948754044,
      "loss": 2.2664,
      "step": 80984
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.7445259690284729,
      "learning_rate": 0.00043494205611324416,
      "loss": 2.085,
      "step": 80985
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.7425335645675659,
      "learning_rate": 0.0004349384027138596,
      "loss": 2.2437,
      "step": 80986
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.8060445189476013,
      "learning_rate": 0.0004349347492893875,
      "loss": 2.3291,
      "step": 80987
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.8925763368606567,
      "learning_rate": 0.0004349310958398283,
      "loss": 2.1661,
      "step": 80988
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.7544585466384888,
      "learning_rate": 0.000434927442365183,
      "loss": 2.1614,
      "step": 80989
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.8320084810256958,
      "learning_rate": 0.00043492378886545217,
      "loss": 2.0523,
      "step": 80990
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.785398542881012,
      "learning_rate": 0.0004349201353406363,
      "loss": 2.1672,
      "step": 80991
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.7769333124160767,
      "learning_rate": 0.00043491648179073636,
      "loss": 2.1452,
      "step": 80992
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.8623759150505066,
      "learning_rate": 0.0004349128282157529,
      "loss": 2.1005,
      "step": 80993
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.8433868288993835,
      "learning_rate": 0.0004349091746156865,
      "loss": 2.451,
      "step": 80994
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.9010894894599915,
      "learning_rate": 0.0004349055209905381,
      "loss": 2.2394,
      "step": 80995
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.819233238697052,
      "learning_rate": 0.00043490186734030815,
      "loss": 2.0155,
      "step": 80996
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.78645259141922,
      "learning_rate": 0.0004348982136649973,
      "loss": 2.2117,
      "step": 80997
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.9618176817893982,
      "learning_rate": 0.0004348945599646065,
      "loss": 2.2259,
      "step": 80998
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.8021931052207947,
      "learning_rate": 0.0004348909062391361,
      "loss": 1.9989,
      "step": 80999
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.8371576070785522,
      "learning_rate": 0.00043488725248858696,
      "loss": 1.9783,
      "step": 81000
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.8607171177864075,
      "learning_rate": 0.0004348835987129598,
      "loss": 1.9335,
      "step": 81001
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.8566896319389343,
      "learning_rate": 0.0004348799449122552,
      "loss": 2.2839,
      "step": 81002
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.8415238857269287,
      "learning_rate": 0.0004348762910864739,
      "loss": 2.0915,
      "step": 81003
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.9073525071144104,
      "learning_rate": 0.0004348726372356164,
      "loss": 2.0088,
      "step": 81004
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.760589063167572,
      "learning_rate": 0.00043486898335968376,
      "loss": 2.1717,
      "step": 81005
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.8247671723365784,
      "learning_rate": 0.00043486532945867626,
      "loss": 2.0982,
      "step": 81006
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.8140420317649841,
      "learning_rate": 0.00043486167553259485,
      "loss": 1.9978,
      "step": 81007
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.8730666637420654,
      "learning_rate": 0.00043485802158144013,
      "loss": 2.1751,
      "step": 81008
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.854336142539978,
      "learning_rate": 0.00043485436760521264,
      "loss": 2.1352,
      "step": 81009
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.8653466701507568,
      "learning_rate": 0.0004348507136039132,
      "loss": 2.3109,
      "step": 81010
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.8089661598205566,
      "learning_rate": 0.00043484705957754253,
      "loss": 2.1264,
      "step": 81011
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.7887856960296631,
      "learning_rate": 0.0004348434055261012,
      "loss": 2.1381,
      "step": 81012
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.7963206171989441,
      "learning_rate": 0.0004348397514495899,
      "loss": 2.3733,
      "step": 81013
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.7882453799247742,
      "learning_rate": 0.0004348360973480094,
      "loss": 2.0762,
      "step": 81014
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.7837831974029541,
      "learning_rate": 0.00043483244322136034,
      "loss": 2.3053,
      "step": 81015
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.7489190101623535,
      "learning_rate": 0.0004348287890696434,
      "loss": 2.1395,
      "step": 81016
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.8277745246887207,
      "learning_rate": 0.0004348251348928592,
      "loss": 2.2964,
      "step": 81017
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.8442133069038391,
      "learning_rate": 0.00043482148069100843,
      "loss": 2.2914,
      "step": 81018
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.7595260143280029,
      "learning_rate": 0.00043481782646409193,
      "loss": 2.0583,
      "step": 81019
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.8426488041877747,
      "learning_rate": 0.0004348141722121101,
      "loss": 2.1026,
      "step": 81020
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.8366552591323853,
      "learning_rate": 0.00043481051793506386,
      "loss": 2.1549,
      "step": 81021
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.9945648908615112,
      "learning_rate": 0.00043480686363295376,
      "loss": 2.2926,
      "step": 81022
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.8810449242591858,
      "learning_rate": 0.00043480320930578054,
      "loss": 2.1623,
      "step": 81023
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.8326339721679688,
      "learning_rate": 0.0004347995549535449,
      "loss": 2.3178,
      "step": 81024
    },
    {
      "epoch": 1.05,
      "grad_norm": 0.9134105443954468,
      "learning_rate": 0.00043479590057624745,
      "loss": 1.9561,
      "step": 81025
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.9121882915496826,
      "learning_rate": 0.00043479224617388887,
      "loss": 2.1555,
      "step": 81026
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.8458102941513062,
      "learning_rate": 0.0004347885917464698,
      "loss": 2.2323,
      "step": 81027
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.7773833274841309,
      "learning_rate": 0.00043478493729399114,
      "loss": 2.4146,
      "step": 81028
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.8723393678665161,
      "learning_rate": 0.0004347812828164533,
      "loss": 2.2502,
      "step": 81029
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.8465930819511414,
      "learning_rate": 0.0004347776283138571,
      "loss": 2.4383,
      "step": 81030
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.8309656977653503,
      "learning_rate": 0.00043477397378620326,
      "loss": 2.0779,
      "step": 81031
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.7817084789276123,
      "learning_rate": 0.0004347703192334923,
      "loss": 2.2782,
      "step": 81032
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.7861720323562622,
      "learning_rate": 0.000434766664655725,
      "loss": 2.1044,
      "step": 81033
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.8826223015785217,
      "learning_rate": 0.00043476301005290215,
      "loss": 2.322,
      "step": 81034
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.75914466381073,
      "learning_rate": 0.00043475935542502416,
      "loss": 2.2195,
      "step": 81035
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.8559934496879578,
      "learning_rate": 0.000434755700772092,
      "loss": 2.1603,
      "step": 81036
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.8666302561759949,
      "learning_rate": 0.0004347520460941061,
      "loss": 2.2608,
      "step": 81037
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.780658483505249,
      "learning_rate": 0.00043474839139106733,
      "loss": 2.0973,
      "step": 81038
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.7828530669212341,
      "learning_rate": 0.0004347447366629762,
      "loss": 2.1627,
      "step": 81039
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.8214439153671265,
      "learning_rate": 0.00043474108190983356,
      "loss": 2.1038,
      "step": 81040
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.7619545459747314,
      "learning_rate": 0.00043473742713164,
      "loss": 2.351,
      "step": 81041
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.7969697117805481,
      "learning_rate": 0.00043473377232839614,
      "loss": 2.1469,
      "step": 81042
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.8274211883544922,
      "learning_rate": 0.00043473011750010283,
      "loss": 2.0578,
      "step": 81043
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.7653612494468689,
      "learning_rate": 0.00043472646264676066,
      "loss": 2.3947,
      "step": 81044
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.7881158590316772,
      "learning_rate": 0.00043472280776837017,
      "loss": 2.214,
      "step": 81045
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.7973640561103821,
      "learning_rate": 0.0004347191528649322,
      "loss": 2.1771,
      "step": 81046
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.7481573224067688,
      "learning_rate": 0.0004347154979364475,
      "loss": 2.0816,
      "step": 81047
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.99662184715271,
      "learning_rate": 0.0004347118429829166,
      "loss": 2.3314,
      "step": 81048
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.7688598036766052,
      "learning_rate": 0.0004347081880043402,
      "loss": 2.0818,
      "step": 81049
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.8066086173057556,
      "learning_rate": 0.000434704533000719,
      "loss": 1.9591,
      "step": 81050
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.7891131639480591,
      "learning_rate": 0.0004347008779720537,
      "loss": 2.2122,
      "step": 81051
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.766535758972168,
      "learning_rate": 0.0004346972229183449,
      "loss": 2.2347,
      "step": 81052
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.7874600887298584,
      "learning_rate": 0.0004346935678395935,
      "loss": 2.1419,
      "step": 81053
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.833177924156189,
      "learning_rate": 0.00043468991273579985,
      "loss": 2.1914,
      "step": 81054
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.8335198760032654,
      "learning_rate": 0.00043468625760696494,
      "loss": 2.1771,
      "step": 81055
    },
    {
      "epoch": 1.06,
      "grad_norm": 1.0010759830474854,
      "learning_rate": 0.0004346826024530892,
      "loss": 2.131,
      "step": 81056
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.8666515350341797,
      "learning_rate": 0.00043467894727417346,
      "loss": 2.0383,
      "step": 81057
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.7941606044769287,
      "learning_rate": 0.00043467529207021836,
      "loss": 2.2169,
      "step": 81058
    },
    {
      "epoch": 1.06,
      "grad_norm": 1.1939960718154907,
      "learning_rate": 0.0004346716368412247,
      "loss": 2.5318,
      "step": 81059
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.9456151127815247,
      "learning_rate": 0.0004346679815871929,
      "loss": 2.2314,
      "step": 81060
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.869662880897522,
      "learning_rate": 0.0004346643263081238,
      "loss": 2.1722,
      "step": 81061
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.984095573425293,
      "learning_rate": 0.00043466067100401804,
      "loss": 1.9822,
      "step": 81062
    },
    {
      "epoch": 1.06,
      "grad_norm": 1.1392759084701538,
      "learning_rate": 0.00043465701567487635,
      "loss": 2.1898,
      "step": 81063
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.8345068693161011,
      "learning_rate": 0.0004346533603206994,
      "loss": 1.9928,
      "step": 81064
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.7682278752326965,
      "learning_rate": 0.00043464970494148784,
      "loss": 2.328,
      "step": 81065
    },
    {
      "epoch": 1.06,
      "grad_norm": 1.0056259632110596,
      "learning_rate": 0.0004346460495372423,
      "loss": 2.3115,
      "step": 81066
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.8640689253807068,
      "learning_rate": 0.0004346423941079636,
      "loss": 2.1854,
      "step": 81067
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.7877939343452454,
      "learning_rate": 0.00043463873865365234,
      "loss": 2.0417,
      "step": 81068
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.821191668510437,
      "learning_rate": 0.0004346350831743092,
      "loss": 2.4181,
      "step": 81069
    },
    {
      "epoch": 1.06,
      "grad_norm": 1.0219593048095703,
      "learning_rate": 0.0004346314276699348,
      "loss": 2.1834,
      "step": 81070
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.923911988735199,
      "learning_rate": 0.0004346277721405299,
      "loss": 2.3111,
      "step": 81071
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.8293986916542053,
      "learning_rate": 0.00043462411658609515,
      "loss": 2.3292,
      "step": 81072
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.8769230246543884,
      "learning_rate": 0.0004346204610066312,
      "loss": 2.0342,
      "step": 81073
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.9652426242828369,
      "learning_rate": 0.00043461680540213884,
      "loss": 2.3014,
      "step": 81074
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.7751456499099731,
      "learning_rate": 0.0004346131497726186,
      "loss": 2.0163,
      "step": 81075
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.7975838780403137,
      "learning_rate": 0.00043460949411807125,
      "loss": 2.2388,
      "step": 81076
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.8511800765991211,
      "learning_rate": 0.0004346058384384975,
      "loss": 2.3387,
      "step": 81077
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.9989389777183533,
      "learning_rate": 0.00043460218273389797,
      "loss": 2.3059,
      "step": 81078
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.8211733102798462,
      "learning_rate": 0.00043459852700427335,
      "loss": 2.2515,
      "step": 81079
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.8085325956344604,
      "learning_rate": 0.0004345948712496243,
      "loss": 2.221,
      "step": 81080
    },
    {
      "epoch": 1.06,
      "grad_norm": 1.039842128753662,
      "learning_rate": 0.0004345912154699516,
      "loss": 2.2898,
      "step": 81081
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.9628121256828308,
      "learning_rate": 0.0004345875596652557,
      "loss": 2.0537,
      "step": 81082
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.8682478070259094,
      "learning_rate": 0.00043458390383553755,
      "loss": 2.3272,
      "step": 81083
    },
    {
      "epoch": 1.06,
      "grad_norm": 1.1844788789749146,
      "learning_rate": 0.0004345802479807977,
      "loss": 2.3427,
      "step": 81084
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.7874798774719238,
      "learning_rate": 0.00043457659210103676,
      "loss": 2.2384,
      "step": 81085
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.8885744214057922,
      "learning_rate": 0.00043457293619625565,
      "loss": 2.1105,
      "step": 81086
    },
    {
      "epoch": 1.06,
      "grad_norm": 1.117173194885254,
      "learning_rate": 0.0004345692802664548,
      "loss": 2.2803,
      "step": 81087
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.8109756112098694,
      "learning_rate": 0.00043456562431163487,
      "loss": 2.4336,
      "step": 81088
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.7981996536254883,
      "learning_rate": 0.00043456196833179683,
      "loss": 2.0458,
      "step": 81089
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.872832715511322,
      "learning_rate": 0.0004345583123269411,
      "loss": 2.2711,
      "step": 81090
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.8784627914428711,
      "learning_rate": 0.0004345546562970684,
      "loss": 2.1287,
      "step": 81091
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.9115138053894043,
      "learning_rate": 0.0004345510002421796,
      "loss": 2.2929,
      "step": 81092
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.8211801052093506,
      "learning_rate": 0.00043454734416227504,
      "loss": 2.1079,
      "step": 81093
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.9770424962043762,
      "learning_rate": 0.00043454368805735574,
      "loss": 2.1293,
      "step": 81094
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.8761411905288696,
      "learning_rate": 0.0004345400319274221,
      "loss": 2.1501,
      "step": 81095
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.8074729442596436,
      "learning_rate": 0.000434536375772475,
      "loss": 2.1766,
      "step": 81096
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.9612827897071838,
      "learning_rate": 0.00043453271959251503,
      "loss": 1.9742,
      "step": 81097
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.9989339709281921,
      "learning_rate": 0.00043452906338754295,
      "loss": 2.4072,
      "step": 81098
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.9732127785682678,
      "learning_rate": 0.0004345254071575593,
      "loss": 2.0275,
      "step": 81099
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.9725221991539001,
      "learning_rate": 0.0004345217509025649,
      "loss": 2.0279,
      "step": 81100
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.9220489859580994,
      "learning_rate": 0.00043451809462256033,
      "loss": 2.3281,
      "step": 81101
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.8956411480903625,
      "learning_rate": 0.00043451443831754626,
      "loss": 2.3404,
      "step": 81102
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.7854703068733215,
      "learning_rate": 0.0004345107819875234,
      "loss": 2.0277,
      "step": 81103
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.9312706589698792,
      "learning_rate": 0.0004345071256324926,
      "loss": 2.2344,
      "step": 81104
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.8298673033714294,
      "learning_rate": 0.0004345034692524543,
      "loss": 2.1934,
      "step": 81105
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.9569790959358215,
      "learning_rate": 0.00043449981284740926,
      "loss": 2.1703,
      "step": 81106
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.7980299592018127,
      "learning_rate": 0.0004344961564173581,
      "loss": 2.1575,
      "step": 81107
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.8062732815742493,
      "learning_rate": 0.00043449249996230173,
      "loss": 2.2279,
      "step": 81108
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.857133686542511,
      "learning_rate": 0.0004344888434822405,
      "loss": 2.0973,
      "step": 81109
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.8759581446647644,
      "learning_rate": 0.0004344851869771753,
      "loss": 2.3955,
      "step": 81110
    },
    {
      "epoch": 1.06,
      "grad_norm": 1.0414637327194214,
      "learning_rate": 0.00043448153044710683,
      "loss": 2.3065,
      "step": 81111
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.9966841340065002,
      "learning_rate": 0.0004344778738920357,
      "loss": 2.1921,
      "step": 81112
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.9585130214691162,
      "learning_rate": 0.00043447421731196255,
      "loss": 2.1843,
      "step": 81113
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.7893975377082825,
      "learning_rate": 0.00043447056070688815,
      "loss": 2.255,
      "step": 81114
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.8233380913734436,
      "learning_rate": 0.0004344669040768131,
      "loss": 2.0386,
      "step": 81115
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.7644805312156677,
      "learning_rate": 0.00043446324742173805,
      "loss": 2.1868,
      "step": 81116
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.7880306243896484,
      "learning_rate": 0.00043445959074166386,
      "loss": 2.3978,
      "step": 81117
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.845527708530426,
      "learning_rate": 0.00043445593403659106,
      "loss": 2.0335,
      "step": 81118
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.8231909871101379,
      "learning_rate": 0.0004344522773065203,
      "loss": 2.1203,
      "step": 81119
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.8145391941070557,
      "learning_rate": 0.00043444862055145233,
      "loss": 2.2055,
      "step": 81120
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.7507128715515137,
      "learning_rate": 0.00043444496377138787,
      "loss": 2.1361,
      "step": 81121
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.9371156096458435,
      "learning_rate": 0.00043444130696632756,
      "loss": 2.3108,
      "step": 81122
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.9026821255683899,
      "learning_rate": 0.00043443765013627205,
      "loss": 2.0975,
      "step": 81123
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.745184600353241,
      "learning_rate": 0.0004344339932812221,
      "loss": 2.2166,
      "step": 81124
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.8144562244415283,
      "learning_rate": 0.00043443033640117826,
      "loss": 2.4903,
      "step": 81125
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.7634528279304504,
      "learning_rate": 0.00043442667949614134,
      "loss": 2.3528,
      "step": 81126
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.826434314250946,
      "learning_rate": 0.00043442302256611193,
      "loss": 2.2239,
      "step": 81127
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.7924413681030273,
      "learning_rate": 0.00043441936561109073,
      "loss": 2.0045,
      "step": 81128
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.7797938585281372,
      "learning_rate": 0.00043441570863107847,
      "loss": 2.3132,
      "step": 81129
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.7628639936447144,
      "learning_rate": 0.00043441205162607577,
      "loss": 1.9291,
      "step": 81130
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.7610158920288086,
      "learning_rate": 0.0004344083945960833,
      "loss": 2.1931,
      "step": 81131
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.8333696722984314,
      "learning_rate": 0.0004344047375411018,
      "loss": 2.2883,
      "step": 81132
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.9550337195396423,
      "learning_rate": 0.00043440108046113196,
      "loss": 2.25,
      "step": 81133
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.961499035358429,
      "learning_rate": 0.0004343974233561744,
      "loss": 2.2582,
      "step": 81134
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.7116443514823914,
      "learning_rate": 0.00043439376622622986,
      "loss": 2.021,
      "step": 81135
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.7656426429748535,
      "learning_rate": 0.00043439010907129886,
      "loss": 2.1657,
      "step": 81136
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.8023148775100708,
      "learning_rate": 0.0004343864518913823,
      "loss": 2.0121,
      "step": 81137
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.7523603439331055,
      "learning_rate": 0.0004343827946864807,
      "loss": 2.1471,
      "step": 81138
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.9188691973686218,
      "learning_rate": 0.00043437913745659486,
      "loss": 2.4151,
      "step": 81139
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.8379570841789246,
      "learning_rate": 0.00043437548020172536,
      "loss": 2.2914,
      "step": 81140
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.8313698172569275,
      "learning_rate": 0.000434371822921873,
      "loss": 2.0791,
      "step": 81141
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.7597928643226624,
      "learning_rate": 0.0004343681656170383,
      "loss": 2.154,
      "step": 81142
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.8291480541229248,
      "learning_rate": 0.000434364508287222,
      "loss": 2.115,
      "step": 81143
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.8332286477088928,
      "learning_rate": 0.0004343608509324249,
      "loss": 2.204,
      "step": 81144
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.7942285537719727,
      "learning_rate": 0.0004343571935526475,
      "loss": 2.3019,
      "step": 81145
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.8082539439201355,
      "learning_rate": 0.0004343535361478906,
      "loss": 2.0822,
      "step": 81146
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.9149812459945679,
      "learning_rate": 0.0004343498787181549,
      "loss": 2.5097,
      "step": 81147
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.8427063822746277,
      "learning_rate": 0.0004343462212634409,
      "loss": 2.1952,
      "step": 81148
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.820746898651123,
      "learning_rate": 0.00043434256378374945,
      "loss": 2.0381,
      "step": 81149
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.767638087272644,
      "learning_rate": 0.0004343389062790812,
      "loss": 2.3118,
      "step": 81150
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.6957788467407227,
      "learning_rate": 0.0004343352487494368,
      "loss": 2.2023,
      "step": 81151
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.8702688813209534,
      "learning_rate": 0.00043433159119481694,
      "loss": 2.2725,
      "step": 81152
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.7674291729927063,
      "learning_rate": 0.0004343279336152223,
      "loss": 2.3473,
      "step": 81153
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.749850332736969,
      "learning_rate": 0.0004343242760106536,
      "loss": 2.2557,
      "step": 81154
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.9216074347496033,
      "learning_rate": 0.00043432061838111143,
      "loss": 2.4019,
      "step": 81155
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.8623857498168945,
      "learning_rate": 0.00043431696072659655,
      "loss": 2.4142,
      "step": 81156
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.782272219657898,
      "learning_rate": 0.00043431330304710957,
      "loss": 2.1997,
      "step": 81157
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.8524587750434875,
      "learning_rate": 0.0004343096453426512,
      "loss": 2.0334,
      "step": 81158
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.8192514777183533,
      "learning_rate": 0.00043430598761322217,
      "loss": 2.2718,
      "step": 81159
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.7950227856636047,
      "learning_rate": 0.00043430232985882316,
      "loss": 2.197,
      "step": 81160
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.8257589936256409,
      "learning_rate": 0.00043429867207945476,
      "loss": 2.3315,
      "step": 81161
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.7548574805259705,
      "learning_rate": 0.0004342950142751177,
      "loss": 2.0246,
      "step": 81162
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.7759950757026672,
      "learning_rate": 0.0004342913564458127,
      "loss": 1.9997,
      "step": 81163
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.7766767740249634,
      "learning_rate": 0.00043428769859154035,
      "loss": 2.2661,
      "step": 81164
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.7367409467697144,
      "learning_rate": 0.00043428404071230143,
      "loss": 1.8953,
      "step": 81165
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.782741367816925,
      "learning_rate": 0.0004342803828080966,
      "loss": 2.2149,
      "step": 81166
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.7901224493980408,
      "learning_rate": 0.0004342767248789264,
      "loss": 2.2413,
      "step": 81167
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.9482553005218506,
      "learning_rate": 0.00043427306692479177,
      "loss": 1.9435,
      "step": 81168
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.7485076189041138,
      "learning_rate": 0.00043426940894569316,
      "loss": 2.2292,
      "step": 81169
    },
    {
      "epoch": 1.06,
      "grad_norm": 1.0531721115112305,
      "learning_rate": 0.0004342657509416313,
      "loss": 2.1377,
      "step": 81170
    },
    {
      "epoch": 1.06,
      "grad_norm": 1.0756199359893799,
      "learning_rate": 0.0004342620929126069,
      "loss": 2.0318,
      "step": 81171
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.8031535148620605,
      "learning_rate": 0.00043425843485862066,
      "loss": 2.2149,
      "step": 81172
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.998790979385376,
      "learning_rate": 0.0004342547767796733,
      "loss": 2.1426,
      "step": 81173
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.8302943110466003,
      "learning_rate": 0.00043425111867576537,
      "loss": 2.2922,
      "step": 81174
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.9524674415588379,
      "learning_rate": 0.0004342474605468977,
      "loss": 2.1246,
      "step": 81175
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.8712449073791504,
      "learning_rate": 0.00043424380239307083,
      "loss": 2.3695,
      "step": 81176
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.9345974922180176,
      "learning_rate": 0.0004342401442142855,
      "loss": 2.1235,
      "step": 81177
    },
    {
      "epoch": 1.06,
      "grad_norm": 1.0713462829589844,
      "learning_rate": 0.0004342364860105424,
      "loss": 2.348,
      "step": 81178
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.7693941593170166,
      "learning_rate": 0.00043423282778184226,
      "loss": 2.1563,
      "step": 81179
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.8401147127151489,
      "learning_rate": 0.00043422916952818566,
      "loss": 2.097,
      "step": 81180
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.8853641152381897,
      "learning_rate": 0.0004342255112495733,
      "loss": 2.3587,
      "step": 81181
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.9204699397087097,
      "learning_rate": 0.0004342218529460059,
      "loss": 2.0863,
      "step": 81182
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.7589636445045471,
      "learning_rate": 0.0004342181946174841,
      "loss": 2.0958,
      "step": 81183
    },
    {
      "epoch": 1.06,
      "grad_norm": 1.0008476972579956,
      "learning_rate": 0.00043421453626400866,
      "loss": 2.2518,
      "step": 81184
    },
    {
      "epoch": 1.06,
      "grad_norm": 1.038392424583435,
      "learning_rate": 0.0004342108778855802,
      "loss": 2.1664,
      "step": 81185
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.8159463405609131,
      "learning_rate": 0.0004342072194821994,
      "loss": 2.2781,
      "step": 81186
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.904696524143219,
      "learning_rate": 0.00043420356105386686,
      "loss": 2.1709,
      "step": 81187
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.7830634117126465,
      "learning_rate": 0.00043419990260058343,
      "loss": 2.0821,
      "step": 81188
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.9163115620613098,
      "learning_rate": 0.00043419624412234975,
      "loss": 2.0902,
      "step": 81189
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.8100676536560059,
      "learning_rate": 0.00043419258561916635,
      "loss": 2.449,
      "step": 81190
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.891988217830658,
      "learning_rate": 0.00043418892709103405,
      "loss": 2.4376,
      "step": 81191
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.9689877033233643,
      "learning_rate": 0.00043418526853795356,
      "loss": 2.3512,
      "step": 81192
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.8226370215415955,
      "learning_rate": 0.0004341816099599254,
      "loss": 1.9266,
      "step": 81193
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.8107804656028748,
      "learning_rate": 0.00043417795135695036,
      "loss": 2.0329,
      "step": 81194
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.9363787174224854,
      "learning_rate": 0.00043417429272902907,
      "loss": 2.2297,
      "step": 81195
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.8164222240447998,
      "learning_rate": 0.0004341706340761624,
      "loss": 2.3274,
      "step": 81196
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.7930659055709839,
      "learning_rate": 0.0004341669753983508,
      "loss": 2.1982,
      "step": 81197
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.877728283405304,
      "learning_rate": 0.00043416331669559493,
      "loss": 2.2357,
      "step": 81198
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.8189138770103455,
      "learning_rate": 0.00043415965796789576,
      "loss": 2.2182,
      "step": 81199
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.8038275837898254,
      "learning_rate": 0.0004341559992152536,
      "loss": 1.9572,
      "step": 81200
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.7782403826713562,
      "learning_rate": 0.00043415234043766936,
      "loss": 2.0746,
      "step": 81201
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.8059365749359131,
      "learning_rate": 0.0004341486816351437,
      "loss": 2.041,
      "step": 81202
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.8434480428695679,
      "learning_rate": 0.0004341450228076772,
      "loss": 2.3263,
      "step": 81203
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.8502159714698792,
      "learning_rate": 0.0004341413639552707,
      "loss": 2.2397,
      "step": 81204
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.8344045281410217,
      "learning_rate": 0.0004341377050779248,
      "loss": 2.1617,
      "step": 81205
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.9882696866989136,
      "learning_rate": 0.00043413404617564005,
      "loss": 2.0823,
      "step": 81206
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.8663561344146729,
      "learning_rate": 0.0004341303872484173,
      "loss": 2.2092,
      "step": 81207
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.8461841940879822,
      "learning_rate": 0.00043412672829625724,
      "loss": 2.4576,
      "step": 81208
    },
    {
      "epoch": 1.06,
      "grad_norm": 1.023831844329834,
      "learning_rate": 0.0004341230693191604,
      "loss": 2.007,
      "step": 81209
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.8202247619628906,
      "learning_rate": 0.00043411941031712756,
      "loss": 2.178,
      "step": 81210
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.8196757435798645,
      "learning_rate": 0.0004341157512901595,
      "loss": 2.1907,
      "step": 81211
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.7770540118217468,
      "learning_rate": 0.0004341120922382567,
      "loss": 2.2772,
      "step": 81212
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.8020684123039246,
      "learning_rate": 0.0004341084331614199,
      "loss": 2.1593,
      "step": 81213
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.8269435167312622,
      "learning_rate": 0.0004341047740596499,
      "loss": 2.2943,
      "step": 81214
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.8962435722351074,
      "learning_rate": 0.0004341011149329472,
      "loss": 2.1146,
      "step": 81215
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.8951146006584167,
      "learning_rate": 0.00043409745578131264,
      "loss": 2.4276,
      "step": 81216
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.7732080817222595,
      "learning_rate": 0.0004340937966047468,
      "loss": 2.5061,
      "step": 81217
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.8295674920082092,
      "learning_rate": 0.00043409013740325044,
      "loss": 2.166,
      "step": 81218
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.8448171019554138,
      "learning_rate": 0.00043408647817682407,
      "loss": 2.0956,
      "step": 81219
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.8437909483909607,
      "learning_rate": 0.00043408281892546864,
      "loss": 2.4566,
      "step": 81220
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.7538352608680725,
      "learning_rate": 0.0004340791596491846,
      "loss": 2.1795,
      "step": 81221
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.9333751201629639,
      "learning_rate": 0.00043407550034797276,
      "loss": 2.2187,
      "step": 81222
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.7858245968818665,
      "learning_rate": 0.0004340718410218337,
      "loss": 2.2513,
      "step": 81223
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.9036173224449158,
      "learning_rate": 0.0004340681816707682,
      "loss": 2.3032,
      "step": 81224
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.8218915462493896,
      "learning_rate": 0.00043406452229477683,
      "loss": 2.2216,
      "step": 81225
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.7911876440048218,
      "learning_rate": 0.0004340608628938604,
      "loss": 2.2362,
      "step": 81226
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.8575022220611572,
      "learning_rate": 0.00043405720346801946,
      "loss": 2.3814,
      "step": 81227
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.8178740739822388,
      "learning_rate": 0.0004340535440172548,
      "loss": 1.9769,
      "step": 81228
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.8195286393165588,
      "learning_rate": 0.0004340498845415671,
      "loss": 2.0221,
      "step": 81229
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.8016849756240845,
      "learning_rate": 0.00043404622504095695,
      "loss": 2.1123,
      "step": 81230
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.8212621212005615,
      "learning_rate": 0.000434042565515425,
      "loss": 1.8948,
      "step": 81231
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.803285539150238,
      "learning_rate": 0.00043403890596497213,
      "loss": 2.0229,
      "step": 81232
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.7371401786804199,
      "learning_rate": 0.00043403524638959887,
      "loss": 2.1355,
      "step": 81233
    },
    {
      "epoch": 1.06,
      "grad_norm": 1.0866509675979614,
      "learning_rate": 0.00043403158678930587,
      "loss": 2.238,
      "step": 81234
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.8769524693489075,
      "learning_rate": 0.00043402792716409395,
      "loss": 2.2705,
      "step": 81235
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.7140942811965942,
      "learning_rate": 0.00043402426751396365,
      "loss": 2.1067,
      "step": 81236
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.9163111448287964,
      "learning_rate": 0.0004340206078389157,
      "loss": 2.2317,
      "step": 81237
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.9349671006202698,
      "learning_rate": 0.00043401694813895083,
      "loss": 2.2094,
      "step": 81238
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.8562851548194885,
      "learning_rate": 0.0004340132884140696,
      "loss": 2.2895,
      "step": 81239
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.7584288716316223,
      "learning_rate": 0.0004340096286642729,
      "loss": 2.0886,
      "step": 81240
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.9548024535179138,
      "learning_rate": 0.00043400596888956114,
      "loss": 2.2937,
      "step": 81241
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.7497612237930298,
      "learning_rate": 0.00043400230908993524,
      "loss": 2.1968,
      "step": 81242
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.8482494354248047,
      "learning_rate": 0.00043399864926539573,
      "loss": 2.1867,
      "step": 81243
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.8297255635261536,
      "learning_rate": 0.0004339949894159434,
      "loss": 2.1178,
      "step": 81244
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.8013460636138916,
      "learning_rate": 0.00043399132954157887,
      "loss": 2.159,
      "step": 81245
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.8557264804840088,
      "learning_rate": 0.0004339876696423028,
      "loss": 2.2424,
      "step": 81246
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.8626100420951843,
      "learning_rate": 0.0004339840097181158,
      "loss": 1.9734,
      "step": 81247
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.7819656729698181,
      "learning_rate": 0.0004339803497690188,
      "loss": 2.2165,
      "step": 81248
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.8695771098136902,
      "learning_rate": 0.0004339766897950122,
      "loss": 2.1259,
      "step": 81249
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.8161635994911194,
      "learning_rate": 0.0004339730297960969,
      "loss": 2.0659,
      "step": 81250
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.8185560703277588,
      "learning_rate": 0.0004339693697722734,
      "loss": 2.1917,
      "step": 81251
    },
    {
      "epoch": 1.06,
      "grad_norm": 1.8542778491973877,
      "learning_rate": 0.0004339657097235425,
      "loss": 2.3259,
      "step": 81252
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.8732668161392212,
      "learning_rate": 0.0004339620496499049,
      "loss": 2.2889,
      "step": 81253
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.916267991065979,
      "learning_rate": 0.00043395838955136113,
      "loss": 2.1822,
      "step": 81254
    },
    {
      "epoch": 1.06,
      "grad_norm": 1.1147048473358154,
      "learning_rate": 0.000433954729427912,
      "loss": 1.9898,
      "step": 81255
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.827765166759491,
      "learning_rate": 0.0004339510692795582,
      "loss": 2.0188,
      "step": 81256
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.7845885753631592,
      "learning_rate": 0.00043394740910630033,
      "loss": 2.14,
      "step": 81257
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.7869724035263062,
      "learning_rate": 0.0004339437489081391,
      "loss": 2.1036,
      "step": 81258
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.9004191160202026,
      "learning_rate": 0.0004339400886850752,
      "loss": 2.1244,
      "step": 81259
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.8119967579841614,
      "learning_rate": 0.00043393642843710936,
      "loss": 2.1176,
      "step": 81260
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.9687113761901855,
      "learning_rate": 0.00043393276816424217,
      "loss": 2.3189,
      "step": 81261
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.8283873200416565,
      "learning_rate": 0.0004339291078664743,
      "loss": 2.2442,
      "step": 81262
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.7941448092460632,
      "learning_rate": 0.0004339254475438066,
      "loss": 2.2005,
      "step": 81263
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.7469833493232727,
      "learning_rate": 0.00043392178719623954,
      "loss": 2.3626,
      "step": 81264
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.8066360354423523,
      "learning_rate": 0.00043391812682377396,
      "loss": 2.3539,
      "step": 81265
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.9019662737846375,
      "learning_rate": 0.0004339144664264104,
      "loss": 2.1307,
      "step": 81266
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.7694635391235352,
      "learning_rate": 0.0004339108060041496,
      "loss": 2.3941,
      "step": 81267
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.8617334961891174,
      "learning_rate": 0.0004339071455569923,
      "loss": 2.1421,
      "step": 81268
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.8005865216255188,
      "learning_rate": 0.0004339034850849392,
      "loss": 1.9782,
      "step": 81269
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.7760715484619141,
      "learning_rate": 0.00043389982458799084,
      "loss": 2.0253,
      "step": 81270
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.9382219910621643,
      "learning_rate": 0.000433896164066148,
      "loss": 2.3475,
      "step": 81271
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.7758232951164246,
      "learning_rate": 0.00043389250351941124,
      "loss": 2.0237,
      "step": 81272
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.8200376033782959,
      "learning_rate": 0.00043388884294778146,
      "loss": 2.3901,
      "step": 81273
    },
    {
      "epoch": 1.06,
      "grad_norm": 1.0189710855484009,
      "learning_rate": 0.00043388518235125916,
      "loss": 2.0731,
      "step": 81274
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.8129185438156128,
      "learning_rate": 0.0004338815217298451,
      "loss": 2.246,
      "step": 81275
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.9560312032699585,
      "learning_rate": 0.00043387786108353986,
      "loss": 2.2273,
      "step": 81276
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.829615592956543,
      "learning_rate": 0.00043387420041234427,
      "loss": 2.3928,
      "step": 81277
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.8614052534103394,
      "learning_rate": 0.000433870539716259,
      "loss": 2.0882,
      "step": 81278
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.9028302431106567,
      "learning_rate": 0.0004338668789952845,
      "loss": 2.1922,
      "step": 81279
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.7775391340255737,
      "learning_rate": 0.00043386321824942167,
      "loss": 2.2007,
      "step": 81280
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.8443474173545837,
      "learning_rate": 0.0004338595574786713,
      "loss": 2.1605,
      "step": 81281
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.7382716536521912,
      "learning_rate": 0.0004338558966830338,
      "loss": 2.2148,
      "step": 81282
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.760470986366272,
      "learning_rate": 0.0004338522358625099,
      "loss": 2.0026,
      "step": 81283
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.8226204514503479,
      "learning_rate": 0.0004338485750171005,
      "loss": 2.0671,
      "step": 81284
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.7748962640762329,
      "learning_rate": 0.00043384491414680596,
      "loss": 2.1685,
      "step": 81285
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.9158174395561218,
      "learning_rate": 0.00043384125325162716,
      "loss": 2.4603,
      "step": 81286
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.7107726335525513,
      "learning_rate": 0.00043383759233156483,
      "loss": 2.0779,
      "step": 81287
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.7342050075531006,
      "learning_rate": 0.0004338339313866195,
      "loss": 2.2009,
      "step": 81288
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.7516829967498779,
      "learning_rate": 0.0004338302704167919,
      "loss": 2.0675,
      "step": 81289
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.8102588057518005,
      "learning_rate": 0.00043382660942208284,
      "loss": 2.2943,
      "step": 81290
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.8924450278282166,
      "learning_rate": 0.00043382294840249276,
      "loss": 2.1994,
      "step": 81291
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.7783332467079163,
      "learning_rate": 0.0004338192873580225,
      "loss": 2.1676,
      "step": 81292
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.8217252492904663,
      "learning_rate": 0.00043381562628867277,
      "loss": 2.2216,
      "step": 81293
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.9348435401916504,
      "learning_rate": 0.0004338119651944441,
      "loss": 2.081,
      "step": 81294
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.7897652387619019,
      "learning_rate": 0.0004338083040753372,
      "loss": 2.1135,
      "step": 81295
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.8157250285148621,
      "learning_rate": 0.000433804642931353,
      "loss": 2.2154,
      "step": 81296
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.8323749303817749,
      "learning_rate": 0.0004338009817624919,
      "loss": 2.2157,
      "step": 81297
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.7860942482948303,
      "learning_rate": 0.0004337973205687547,
      "loss": 2.3013,
      "step": 81298
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.8277758359909058,
      "learning_rate": 0.00043379365935014206,
      "loss": 2.3798,
      "step": 81299
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.8097283840179443,
      "learning_rate": 0.0004337899981066546,
      "loss": 2.1673,
      "step": 81300
    },
    {
      "epoch": 1.06,
      "grad_norm": 1.0854978561401367,
      "learning_rate": 0.00043378633683829307,
      "loss": 1.9516,
      "step": 81301
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.8009174466133118,
      "learning_rate": 0.0004337826755450581,
      "loss": 2.3086,
      "step": 81302
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.8766710758209229,
      "learning_rate": 0.0004337790142269504,
      "loss": 2.2948,
      "step": 81303
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.9542959928512573,
      "learning_rate": 0.0004337753528839707,
      "loss": 2.3722,
      "step": 81304
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.7658795118331909,
      "learning_rate": 0.00043377169151611965,
      "loss": 1.9956,
      "step": 81305
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.8379235863685608,
      "learning_rate": 0.000433768030123398,
      "loss": 2.2934,
      "step": 81306
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.8048601150512695,
      "learning_rate": 0.0004337643687058062,
      "loss": 2.2254,
      "step": 81307
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.8868967890739441,
      "learning_rate": 0.0004337607072633451,
      "loss": 2.2702,
      "step": 81308
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.8457099795341492,
      "learning_rate": 0.00043375704579601536,
      "loss": 2.0308,
      "step": 81309
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.8933049440383911,
      "learning_rate": 0.00043375338430381776,
      "loss": 2.0855,
      "step": 81310
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.8062702417373657,
      "learning_rate": 0.0004337497227867528,
      "loss": 2.0597,
      "step": 81311
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.8544110655784607,
      "learning_rate": 0.00043374606124482127,
      "loss": 2.1674,
      "step": 81312
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.8464879989624023,
      "learning_rate": 0.0004337423996780238,
      "loss": 2.25,
      "step": 81313
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.8029214143753052,
      "learning_rate": 0.00043373873808636105,
      "loss": 1.9967,
      "step": 81314
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.7650570869445801,
      "learning_rate": 0.0004337350764698338,
      "loss": 2.1639,
      "step": 81315
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.9095092415809631,
      "learning_rate": 0.00043373141482844267,
      "loss": 2.1965,
      "step": 81316
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.8531824946403503,
      "learning_rate": 0.0004337277531621883,
      "loss": 2.1701,
      "step": 81317
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.8565430045127869,
      "learning_rate": 0.0004337240914710715,
      "loss": 2.214,
      "step": 81318
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.8467795848846436,
      "learning_rate": 0.00043372042975509286,
      "loss": 2.1909,
      "step": 81319
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.8396649360656738,
      "learning_rate": 0.000433716768014253,
      "loss": 2.0706,
      "step": 81320
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.8172285556793213,
      "learning_rate": 0.0004337131062485527,
      "loss": 2.0865,
      "step": 81321
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.8708397746086121,
      "learning_rate": 0.0004337094444579926,
      "loss": 2.3832,
      "step": 81322
    },
    {
      "epoch": 1.06,
      "grad_norm": 1.188918113708496,
      "learning_rate": 0.0004337057826425734,
      "loss": 2.3582,
      "step": 81323
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.7799016237258911,
      "learning_rate": 0.0004337021208022958,
      "loss": 2.017,
      "step": 81324
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.942814826965332,
      "learning_rate": 0.0004336984589371604,
      "loss": 2.0529,
      "step": 81325
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.9803808331489563,
      "learning_rate": 0.000433694797047168,
      "loss": 2.2081,
      "step": 81326
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.8376137018203735,
      "learning_rate": 0.00043369113513231917,
      "loss": 2.2829,
      "step": 81327
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.7512372732162476,
      "learning_rate": 0.00043368747319261463,
      "loss": 2.0425,
      "step": 81328
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.8594152331352234,
      "learning_rate": 0.00043368381122805503,
      "loss": 2.2822,
      "step": 81329
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.8121544718742371,
      "learning_rate": 0.00043368014923864123,
      "loss": 2.2736,
      "step": 81330
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.8828051686286926,
      "learning_rate": 0.00043367648722437356,
      "loss": 2.1442,
      "step": 81331
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.783155083656311,
      "learning_rate": 0.00043367282518525305,
      "loss": 2.1238,
      "step": 81332
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.9671289324760437,
      "learning_rate": 0.00043366916312128023,
      "loss": 2.3044,
      "step": 81333
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.9537542462348938,
      "learning_rate": 0.0004336655010324557,
      "loss": 2.2454,
      "step": 81334
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.8608241081237793,
      "learning_rate": 0.00043366183891878026,
      "loss": 2.0601,
      "step": 81335
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.8429209589958191,
      "learning_rate": 0.0004336581767802547,
      "loss": 2.1739,
      "step": 81336
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.8125250935554504,
      "learning_rate": 0.0004336545146168794,
      "loss": 2.0964,
      "step": 81337
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.8155716061592102,
      "learning_rate": 0.0004336508524286552,
      "loss": 2.1829,
      "step": 81338
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.8565020561218262,
      "learning_rate": 0.000433647190215583,
      "loss": 2.1201,
      "step": 81339
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.8394203782081604,
      "learning_rate": 0.000433643527977663,
      "loss": 2.0451,
      "step": 81340
    },
    {
      "epoch": 1.06,
      "grad_norm": 1.0348374843597412,
      "learning_rate": 0.00043363986571489624,
      "loss": 2.2302,
      "step": 81341
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.7835090160369873,
      "learning_rate": 0.00043363620342728337,
      "loss": 2.3165,
      "step": 81342
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.7589727640151978,
      "learning_rate": 0.0004336325411148249,
      "loss": 2.0179,
      "step": 81343
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.8093440532684326,
      "learning_rate": 0.00043362887877752165,
      "loss": 2.1258,
      "step": 81344
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.8150873184204102,
      "learning_rate": 0.0004336252164153744,
      "loss": 2.2682,
      "step": 81345
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.766576886177063,
      "learning_rate": 0.00043362155402838353,
      "loss": 2.0487,
      "step": 81346
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.8064166307449341,
      "learning_rate": 0.0004336178916165499,
      "loss": 2.1982,
      "step": 81347
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.7956083416938782,
      "learning_rate": 0.00043361422917987427,
      "loss": 2.1745,
      "step": 81348
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.7055965065956116,
      "learning_rate": 0.0004336105667183572,
      "loss": 2.1374,
      "step": 81349
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.7800654172897339,
      "learning_rate": 0.00043360690423199934,
      "loss": 2.2289,
      "step": 81350
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.8109798431396484,
      "learning_rate": 0.00043360324172080154,
      "loss": 2.2952,
      "step": 81351
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.7175944447517395,
      "learning_rate": 0.00043359957918476434,
      "loss": 2.2125,
      "step": 81352
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.7767728567123413,
      "learning_rate": 0.0004335959166238884,
      "loss": 2.041,
      "step": 81353
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.7578009963035583,
      "learning_rate": 0.00043359225403817456,
      "loss": 2.1911,
      "step": 81354
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.7665542364120483,
      "learning_rate": 0.00043358859142762334,
      "loss": 2.2713,
      "step": 81355
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.8408620357513428,
      "learning_rate": 0.00043358492879223544,
      "loss": 2.1564,
      "step": 81356
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.8362100124359131,
      "learning_rate": 0.0004335812661320116,
      "loss": 2.3342,
      "step": 81357
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.7893203496932983,
      "learning_rate": 0.00043357760344695255,
      "loss": 2.3718,
      "step": 81358
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.813585638999939,
      "learning_rate": 0.00043357394073705887,
      "loss": 2.2255,
      "step": 81359
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.803992748260498,
      "learning_rate": 0.0004335702780023312,
      "loss": 2.2116,
      "step": 81360
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.792115330696106,
      "learning_rate": 0.00043356661524277033,
      "loss": 2.3219,
      "step": 81361
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.8051486015319824,
      "learning_rate": 0.00043356295245837695,
      "loss": 2.307,
      "step": 81362
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.8662014007568359,
      "learning_rate": 0.00043355928964915164,
      "loss": 2.407,
      "step": 81363
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.9015880823135376,
      "learning_rate": 0.0004335556268150952,
      "loss": 2.2334,
      "step": 81364
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.8346222043037415,
      "learning_rate": 0.0004335519639562082,
      "loss": 2.1837,
      "step": 81365
    },
    {
      "epoch": 1.06,
      "grad_norm": 1.030042052268982,
      "learning_rate": 0.00043354830107249144,
      "loss": 2.2738,
      "step": 81366
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.8584970831871033,
      "learning_rate": 0.0004335446381639454,
      "loss": 2.3083,
      "step": 81367
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.7510582804679871,
      "learning_rate": 0.00043354097523057094,
      "loss": 2.4406,
      "step": 81368
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.9615317583084106,
      "learning_rate": 0.0004335373122723688,
      "loss": 2.0523,
      "step": 81369
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.8076844811439514,
      "learning_rate": 0.0004335336492893394,
      "loss": 2.1096,
      "step": 81370
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.7602081298828125,
      "learning_rate": 0.00043352998628148366,
      "loss": 2.1693,
      "step": 81371
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.8282708525657654,
      "learning_rate": 0.00043352632324880223,
      "loss": 2.2061,
      "step": 81372
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.7165924310684204,
      "learning_rate": 0.0004335226601912956,
      "loss": 1.9071,
      "step": 81373
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.8423593640327454,
      "learning_rate": 0.0004335189971089646,
      "loss": 2.2194,
      "step": 81374
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.82796710729599,
      "learning_rate": 0.00043351533400181,
      "loss": 2.3066,
      "step": 81375
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.7962450385093689,
      "learning_rate": 0.0004335116708698323,
      "loss": 2.1757,
      "step": 81376
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.82856285572052,
      "learning_rate": 0.0004335080077130323,
      "loss": 2.0974,
      "step": 81377
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.8752725124359131,
      "learning_rate": 0.0004335043445314106,
      "loss": 2.1189,
      "step": 81378
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.7906962633132935,
      "learning_rate": 0.00043350068132496804,
      "loss": 2.1063,
      "step": 81379
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.7803090214729309,
      "learning_rate": 0.0004334970180937051,
      "loss": 2.3148,
      "step": 81380
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.7572823762893677,
      "learning_rate": 0.00043349335483762247,
      "loss": 2.3495,
      "step": 81381
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.8886820077896118,
      "learning_rate": 0.0004334896915567211,
      "loss": 2.134,
      "step": 81382
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.7849677801132202,
      "learning_rate": 0.0004334860282510012,
      "loss": 2.1345,
      "step": 81383
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.8308548927307129,
      "learning_rate": 0.0004334823649204639,
      "loss": 2.0583,
      "step": 81384
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.8380893468856812,
      "learning_rate": 0.00043347870156510977,
      "loss": 2.222,
      "step": 81385
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.8715528845787048,
      "learning_rate": 0.0004334750381849393,
      "loss": 2.3318,
      "step": 81386
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.86598140001297,
      "learning_rate": 0.0004334713747799533,
      "loss": 2.3018,
      "step": 81387
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.7543147802352905,
      "learning_rate": 0.0004334677113501526,
      "loss": 2.2503,
      "step": 81388
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.8292762041091919,
      "learning_rate": 0.00043346404789553763,
      "loss": 2.1655,
      "step": 81389
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.8376050591468811,
      "learning_rate": 0.0004334603844161091,
      "loss": 2.4743,
      "step": 81390
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.9071471095085144,
      "learning_rate": 0.0004334567209118679,
      "loss": 2.2256,
      "step": 81391
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.851020097732544,
      "learning_rate": 0.0004334530573828145,
      "loss": 2.2105,
      "step": 81392
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.7800072431564331,
      "learning_rate": 0.0004334493938289497,
      "loss": 2.2199,
      "step": 81393
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.8576181530952454,
      "learning_rate": 0.0004334457302502741,
      "loss": 2.145,
      "step": 81394
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.8138636350631714,
      "learning_rate": 0.00043344206664678843,
      "loss": 2.1876,
      "step": 81395
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.8211653232574463,
      "learning_rate": 0.0004334384030184933,
      "loss": 2.289,
      "step": 81396
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.8000480532646179,
      "learning_rate": 0.0004334347393653896,
      "loss": 2.1162,
      "step": 81397
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.7640085220336914,
      "learning_rate": 0.00043343107568747773,
      "loss": 2.2148,
      "step": 81398
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.7977412343025208,
      "learning_rate": 0.0004334274119847586,
      "loss": 2.1695,
      "step": 81399
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.8099148273468018,
      "learning_rate": 0.00043342374825723276,
      "loss": 2.1059,
      "step": 81400
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.8398771286010742,
      "learning_rate": 0.0004334200845049009,
      "loss": 2.1222,
      "step": 81401
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.8498582243919373,
      "learning_rate": 0.0004334164207277637,
      "loss": 2.2526,
      "step": 81402
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.8172052502632141,
      "learning_rate": 0.00043341275692582203,
      "loss": 2.2749,
      "step": 81403
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.7531322836875916,
      "learning_rate": 0.00043340909309907624,
      "loss": 2.176,
      "step": 81404
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.8329859375953674,
      "learning_rate": 0.0004334054292475272,
      "loss": 2.0991,
      "step": 81405
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.7873225808143616,
      "learning_rate": 0.0004334017653711757,
      "loss": 2.3452,
      "step": 81406
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.7324122190475464,
      "learning_rate": 0.00043339810147002215,
      "loss": 2.1265,
      "step": 81407
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.8928012251853943,
      "learning_rate": 0.0004333944375440674,
      "loss": 2.2661,
      "step": 81408
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.8060345649719238,
      "learning_rate": 0.0004333907735933122,
      "loss": 2.4383,
      "step": 81409
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.7551370859146118,
      "learning_rate": 0.00043338710961775707,
      "loss": 2.1919,
      "step": 81410
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.7496517896652222,
      "learning_rate": 0.0004333834456174027,
      "loss": 2.3105,
      "step": 81411
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.7435826659202576,
      "learning_rate": 0.00043337978159224997,
      "loss": 2.2857,
      "step": 81412
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.8230379819869995,
      "learning_rate": 0.00043337611754229926,
      "loss": 2.5227,
      "step": 81413
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.8633269667625427,
      "learning_rate": 0.0004333724534675515,
      "loss": 2.2741,
      "step": 81414
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.8530943393707275,
      "learning_rate": 0.0004333687893680073,
      "loss": 2.3769,
      "step": 81415
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.7082541584968567,
      "learning_rate": 0.00043336512524366733,
      "loss": 2.2464,
      "step": 81416
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.8100573420524597,
      "learning_rate": 0.00043336146109453224,
      "loss": 2.1154,
      "step": 81417
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.8204474449157715,
      "learning_rate": 0.00043335779692060274,
      "loss": 2.4111,
      "step": 81418
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.7853328585624695,
      "learning_rate": 0.0004333541327218795,
      "loss": 2.2173,
      "step": 81419
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.7957296371459961,
      "learning_rate": 0.0004333504684983632,
      "loss": 2.3198,
      "step": 81420
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.8834314346313477,
      "learning_rate": 0.0004333468042500545,
      "loss": 2.3253,
      "step": 81421
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.8394108414649963,
      "learning_rate": 0.00043334313997695417,
      "loss": 2.2295,
      "step": 81422
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.8227279186248779,
      "learning_rate": 0.0004333394756790628,
      "loss": 2.3134,
      "step": 81423
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.8578982949256897,
      "learning_rate": 0.0004333358113563812,
      "loss": 2.0529,
      "step": 81424
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.920390248298645,
      "learning_rate": 0.00043333214700890985,
      "loss": 2.0996,
      "step": 81425
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.842423677444458,
      "learning_rate": 0.00043332848263664947,
      "loss": 2.1444,
      "step": 81426
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.7781116962432861,
      "learning_rate": 0.000433324818239601,
      "loss": 2.225,
      "step": 81427
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.7319957613945007,
      "learning_rate": 0.0004333211538177648,
      "loss": 1.8788,
      "step": 81428
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.8042243123054504,
      "learning_rate": 0.0004333174893711417,
      "loss": 2.1132,
      "step": 81429
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.8198614716529846,
      "learning_rate": 0.0004333138248997324,
      "loss": 2.1328,
      "step": 81430
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.779039204120636,
      "learning_rate": 0.0004333101604035375,
      "loss": 2.1106,
      "step": 81431
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.8843331933021545,
      "learning_rate": 0.00043330649588255773,
      "loss": 1.8934,
      "step": 81432
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.7530046701431274,
      "learning_rate": 0.00043330283133679383,
      "loss": 2.0527,
      "step": 81433
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.7741142511367798,
      "learning_rate": 0.0004332991667662463,
      "loss": 2.2444,
      "step": 81434
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.9049648642539978,
      "learning_rate": 0.000433295502170916,
      "loss": 2.1138,
      "step": 81435
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.7567725777626038,
      "learning_rate": 0.00043329183755080365,
      "loss": 2.2219,
      "step": 81436
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.8188947439193726,
      "learning_rate": 0.00043328817290590974,
      "loss": 2.1505,
      "step": 81437
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.8554049730300903,
      "learning_rate": 0.000433284508236235,
      "loss": 2.3658,
      "step": 81438
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.8892785310745239,
      "learning_rate": 0.0004332808435417802,
      "loss": 2.365,
      "step": 81439
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.8498411178588867,
      "learning_rate": 0.00043327717882254594,
      "loss": 2.3755,
      "step": 81440
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.819030225276947,
      "learning_rate": 0.00043327351407853297,
      "loss": 2.1236,
      "step": 81441
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.8297594785690308,
      "learning_rate": 0.00043326984930974185,
      "loss": 2.3318,
      "step": 81442
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.7895355820655823,
      "learning_rate": 0.0004332661845161735,
      "loss": 2.2162,
      "step": 81443
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.7270208597183228,
      "learning_rate": 0.0004332625196978284,
      "loss": 2.2464,
      "step": 81444
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.8239604234695435,
      "learning_rate": 0.0004332588548547072,
      "loss": 2.2433,
      "step": 81445
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.8156800866127014,
      "learning_rate": 0.0004332551899868108,
      "loss": 2.2452,
      "step": 81446
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.7462329268455505,
      "learning_rate": 0.00043325152509413967,
      "loss": 2.3959,
      "step": 81447
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.9479790329933167,
      "learning_rate": 0.0004332478601766945,
      "loss": 2.1763,
      "step": 81448
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.8004940748214722,
      "learning_rate": 0.00043324419523447616,
      "loss": 2.1599,
      "step": 81449
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.7845884561538696,
      "learning_rate": 0.0004332405302674852,
      "loss": 2.3867,
      "step": 81450
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.7765481472015381,
      "learning_rate": 0.0004332368652757222,
      "loss": 1.9583,
      "step": 81451
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.9115322828292847,
      "learning_rate": 0.000433233200259188,
      "loss": 2.1555,
      "step": 81452
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.7154204249382019,
      "learning_rate": 0.0004332295352178833,
      "loss": 2.0871,
      "step": 81453
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.8939758539199829,
      "learning_rate": 0.00043322587015180865,
      "loss": 2.1642,
      "step": 81454
    },
    {
      "epoch": 1.06,
      "grad_norm": 1.0001729726791382,
      "learning_rate": 0.0004332222050609648,
      "loss": 2.1885,
      "step": 81455
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.9154609441757202,
      "learning_rate": 0.00043321853994535245,
      "loss": 2.2762,
      "step": 81456
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.7821491956710815,
      "learning_rate": 0.00043321487480497224,
      "loss": 2.0187,
      "step": 81457
    },
    {
      "epoch": 1.06,
      "grad_norm": 1.1430786848068237,
      "learning_rate": 0.00043321120963982484,
      "loss": 2.1873,
      "step": 81458
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.8297826051712036,
      "learning_rate": 0.000433207544449911,
      "loss": 2.1403,
      "step": 81459
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.7884225249290466,
      "learning_rate": 0.0004332038792352314,
      "loss": 2.3609,
      "step": 81460
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.8416966795921326,
      "learning_rate": 0.00043320021399578665,
      "loss": 2.1288,
      "step": 81461
    },
    {
      "epoch": 1.06,
      "grad_norm": 1.0180630683898926,
      "learning_rate": 0.00043319654873157743,
      "loss": 2.2694,
      "step": 81462
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.8851369619369507,
      "learning_rate": 0.0004331928834426045,
      "loss": 2.1799,
      "step": 81463
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.8049046397209167,
      "learning_rate": 0.0004331892181288685,
      "loss": 2.1545,
      "step": 81464
    },
    {
      "epoch": 1.06,
      "grad_norm": 1.4522452354431152,
      "learning_rate": 0.00043318555279037006,
      "loss": 2.2486,
      "step": 81465
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.9108802676200867,
      "learning_rate": 0.00043318188742710997,
      "loss": 2.3204,
      "step": 81466
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.7935135960578918,
      "learning_rate": 0.0004331782220390889,
      "loss": 2.3397,
      "step": 81467
    },
    {
      "epoch": 1.06,
      "grad_norm": 1.3728721141815186,
      "learning_rate": 0.00043317455662630737,
      "loss": 2.3449,
      "step": 81468
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.8730688095092773,
      "learning_rate": 0.0004331708911887662,
      "loss": 2.2011,
      "step": 81469
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.7707230448722839,
      "learning_rate": 0.0004331672257264661,
      "loss": 2.4437,
      "step": 81470
    },
    {
      "epoch": 1.06,
      "grad_norm": 1.00827956199646,
      "learning_rate": 0.0004331635602394076,
      "loss": 2.1729,
      "step": 81471
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.7994988560676575,
      "learning_rate": 0.00043315989472759157,
      "loss": 2.1142,
      "step": 81472
    },
    {
      "epoch": 1.06,
      "grad_norm": 1.1291215419769287,
      "learning_rate": 0.0004331562291910186,
      "loss": 2.222,
      "step": 81473
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.781505286693573,
      "learning_rate": 0.0004331525636296894,
      "loss": 2.0884,
      "step": 81474
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.7867692708969116,
      "learning_rate": 0.00043314889804360455,
      "loss": 2.1187,
      "step": 81475
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.7502729296684265,
      "learning_rate": 0.0004331452324327649,
      "loss": 2.2309,
      "step": 81476
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.8038434982299805,
      "learning_rate": 0.0004331415667971709,
      "loss": 2.2381,
      "step": 81477
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.8697538375854492,
      "learning_rate": 0.00043313790113682347,
      "loss": 2.1457,
      "step": 81478
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.8078088164329529,
      "learning_rate": 0.0004331342354517232,
      "loss": 2.2506,
      "step": 81479
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.9262149333953857,
      "learning_rate": 0.0004331305697418707,
      "loss": 2.0747,
      "step": 81480
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.7767277359962463,
      "learning_rate": 0.0004331269040072667,
      "loss": 2.1855,
      "step": 81481
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.7636915445327759,
      "learning_rate": 0.00043312323824791206,
      "loss": 2.2514,
      "step": 81482
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.8560079336166382,
      "learning_rate": 0.0004331195724638071,
      "loss": 2.0571,
      "step": 81483
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.7650898694992065,
      "learning_rate": 0.00043311590665495275,
      "loss": 2.4852,
      "step": 81484
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.8673762083053589,
      "learning_rate": 0.0004331122408213497,
      "loss": 2.3098,
      "step": 81485
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.8908253312110901,
      "learning_rate": 0.00043310857496299857,
      "loss": 2.3281,
      "step": 81486
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.7659438848495483,
      "learning_rate": 0.0004331049090798999,
      "loss": 2.1462,
      "step": 81487
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.7903028726577759,
      "learning_rate": 0.00043310124317205473,
      "loss": 2.1752,
      "step": 81488
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.8220983147621155,
      "learning_rate": 0.0004330975772394634,
      "loss": 2.0669,
      "step": 81489
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.9022608995437622,
      "learning_rate": 0.00043309391128212675,
      "loss": 2.2266,
      "step": 81490
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.7708221673965454,
      "learning_rate": 0.00043309024530004543,
      "loss": 2.4284,
      "step": 81491
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.7731911540031433,
      "learning_rate": 0.00043308657929322006,
      "loss": 2.3047,
      "step": 81492
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.7130815386772156,
      "learning_rate": 0.0004330829132616514,
      "loss": 1.9064,
      "step": 81493
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.8388794660568237,
      "learning_rate": 0.00043307924720534024,
      "loss": 2.2165,
      "step": 81494
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.7756951451301575,
      "learning_rate": 0.00043307558112428703,
      "loss": 2.3444,
      "step": 81495
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.8923718333244324,
      "learning_rate": 0.00043307191501849255,
      "loss": 2.0094,
      "step": 81496
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.9233853816986084,
      "learning_rate": 0.00043306824888795755,
      "loss": 2.032,
      "step": 81497
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.7817352414131165,
      "learning_rate": 0.00043306458273268257,
      "loss": 2.3907,
      "step": 81498
    },
    {
      "epoch": 1.06,
      "grad_norm": 1.0644091367721558,
      "learning_rate": 0.0004330609165526684,
      "loss": 2.1372,
      "step": 81499
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.7463264465332031,
      "learning_rate": 0.0004330572503479157,
      "loss": 2.1167,
      "step": 81500
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.8272826671600342,
      "learning_rate": 0.00043305358411842515,
      "loss": 2.074,
      "step": 81501
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.8167449235916138,
      "learning_rate": 0.00043304991786419743,
      "loss": 2.0091,
      "step": 81502
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.8069555163383484,
      "learning_rate": 0.00043304625158523325,
      "loss": 2.2042,
      "step": 81503
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.8182764649391174,
      "learning_rate": 0.0004330425852815332,
      "loss": 2.1539,
      "step": 81504
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.7456874847412109,
      "learning_rate": 0.00043303891895309803,
      "loss": 2.3025,
      "step": 81505
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.7160227298736572,
      "learning_rate": 0.00043303525259992846,
      "loss": 2.1471,
      "step": 81506
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.8306090235710144,
      "learning_rate": 0.0004330315862220251,
      "loss": 2.1896,
      "step": 81507
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.794401228427887,
      "learning_rate": 0.0004330279198193887,
      "loss": 2.1372,
      "step": 81508
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.8326882123947144,
      "learning_rate": 0.0004330242533920198,
      "loss": 2.3484,
      "step": 81509
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.7980014681816101,
      "learning_rate": 0.00043302058693991925,
      "loss": 2.1071,
      "step": 81510
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.8765002489089966,
      "learning_rate": 0.00043301692046308763,
      "loss": 2.2654,
      "step": 81511
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.8929577469825745,
      "learning_rate": 0.0004330132539615257,
      "loss": 2.4166,
      "step": 81512
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.7957909107208252,
      "learning_rate": 0.00043300958743523407,
      "loss": 2.1889,
      "step": 81513
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.9361472725868225,
      "learning_rate": 0.0004330059208842134,
      "loss": 2.2864,
      "step": 81514
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.8068975210189819,
      "learning_rate": 0.00043300225430846446,
      "loss": 2.1285,
      "step": 81515
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.9534019827842712,
      "learning_rate": 0.00043299858770798794,
      "loss": 2.2611,
      "step": 81516
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.7585733532905579,
      "learning_rate": 0.0004329949210827844,
      "loss": 2.2038,
      "step": 81517
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.8018315434455872,
      "learning_rate": 0.0004329912544328546,
      "loss": 2.1467,
      "step": 81518
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.845910370349884,
      "learning_rate": 0.0004329875877581993,
      "loss": 2.2935,
      "step": 81519
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.8128408789634705,
      "learning_rate": 0.00043298392105881896,
      "loss": 2.339,
      "step": 81520
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.7535883784294128,
      "learning_rate": 0.00043298025433471444,
      "loss": 2.2102,
      "step": 81521
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.833082914352417,
      "learning_rate": 0.0004329765875858865,
      "loss": 2.0938,
      "step": 81522
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.7931829690933228,
      "learning_rate": 0.00043297292081233556,
      "loss": 2.4007,
      "step": 81523
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.8040501475334167,
      "learning_rate": 0.0004329692540140625,
      "loss": 2.1816,
      "step": 81524
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.8394084572792053,
      "learning_rate": 0.00043296558719106796,
      "loss": 2.326,
      "step": 81525
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.8090397119522095,
      "learning_rate": 0.00043296192034335257,
      "loss": 2.2757,
      "step": 81526
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.7483900785446167,
      "learning_rate": 0.000432958253470917,
      "loss": 2.1969,
      "step": 81527
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.8690374493598938,
      "learning_rate": 0.00043295458657376215,
      "loss": 2.2067,
      "step": 81528
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.8746992945671082,
      "learning_rate": 0.0004329509196518884,
      "loss": 2.1648,
      "step": 81529
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.9919966459274292,
      "learning_rate": 0.00043294725270529656,
      "loss": 2.1035,
      "step": 81530
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.8896350264549255,
      "learning_rate": 0.00043294358573398744,
      "loss": 2.1081,
      "step": 81531
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.8404288291931152,
      "learning_rate": 0.0004329399187379615,
      "loss": 2.2133,
      "step": 81532
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.8108947277069092,
      "learning_rate": 0.00043293625171721947,
      "loss": 2.1043,
      "step": 81533
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.8976507186889648,
      "learning_rate": 0.0004329325846717622,
      "loss": 2.2298,
      "step": 81534
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.9125518798828125,
      "learning_rate": 0.00043292891760159016,
      "loss": 2.2483,
      "step": 81535
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.8003464341163635,
      "learning_rate": 0.0004329252505067041,
      "loss": 2.4436,
      "step": 81536
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.9531464576721191,
      "learning_rate": 0.0004329215833871048,
      "loss": 2.1643,
      "step": 81537
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.9684455394744873,
      "learning_rate": 0.0004329179162427929,
      "loss": 2.1145,
      "step": 81538
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.8644654154777527,
      "learning_rate": 0.000432914249073769,
      "loss": 2.0962,
      "step": 81539
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.9240405559539795,
      "learning_rate": 0.0004329105818800338,
      "loss": 2.0637,
      "step": 81540
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.971146285533905,
      "learning_rate": 0.000432906914661588,
      "loss": 2.4375,
      "step": 81541
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.8681479096412659,
      "learning_rate": 0.0004329032474184324,
      "loss": 2.2651,
      "step": 81542
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.849888801574707,
      "learning_rate": 0.0004328995801505675,
      "loss": 2.1643,
      "step": 81543
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.7419371008872986,
      "learning_rate": 0.0004328959128579941,
      "loss": 1.9314,
      "step": 81544
    },
    {
      "epoch": 1.06,
      "grad_norm": 1.1019128561019897,
      "learning_rate": 0.0004328922455407128,
      "loss": 2.1757,
      "step": 81545
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.8749786615371704,
      "learning_rate": 0.0004328885781987244,
      "loss": 2.1202,
      "step": 81546
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.8073657751083374,
      "learning_rate": 0.0004328849108320294,
      "loss": 2.2519,
      "step": 81547
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.8529248833656311,
      "learning_rate": 0.00043288124344062864,
      "loss": 2.3732,
      "step": 81548
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.8973259925842285,
      "learning_rate": 0.00043287757602452275,
      "loss": 2.0283,
      "step": 81549
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.9509845972061157,
      "learning_rate": 0.0004328739085837124,
      "loss": 2.0775,
      "step": 81550
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.7574859261512756,
      "learning_rate": 0.00043287024111819824,
      "loss": 2.3784,
      "step": 81551
    },
    {
      "epoch": 1.06,
      "grad_norm": 1.0572075843811035,
      "learning_rate": 0.0004328665736279811,
      "loss": 2.1561,
      "step": 81552
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.9306815266609192,
      "learning_rate": 0.00043286290611306154,
      "loss": 2.0516,
      "step": 81553
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.8441494107246399,
      "learning_rate": 0.0004328592385734402,
      "loss": 2.0326,
      "step": 81554
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.9022666811943054,
      "learning_rate": 0.0004328555710091178,
      "loss": 2.3313,
      "step": 81555
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.8964923620223999,
      "learning_rate": 0.00043285190342009513,
      "loss": 2.2324,
      "step": 81556
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.9423926472663879,
      "learning_rate": 0.0004328482358063727,
      "loss": 2.1836,
      "step": 81557
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.8262656331062317,
      "learning_rate": 0.00043284456816795137,
      "loss": 2.071,
      "step": 81558
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.9281187057495117,
      "learning_rate": 0.0004328409005048316,
      "loss": 2.2449,
      "step": 81559
    },
    {
      "epoch": 1.06,
      "grad_norm": 1.2107820510864258,
      "learning_rate": 0.00043283723281701425,
      "loss": 2.0498,
      "step": 81560
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.8678388595581055,
      "learning_rate": 0.0004328335651045,
      "loss": 1.9934,
      "step": 81561
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.8923333883285522,
      "learning_rate": 0.00043282989736728944,
      "loss": 2.3833,
      "step": 81562
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.8984151482582092,
      "learning_rate": 0.00043282622960538324,
      "loss": 2.1593,
      "step": 81563
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.8475366234779358,
      "learning_rate": 0.0004328225618187823,
      "loss": 1.9617,
      "step": 81564
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.8083937168121338,
      "learning_rate": 0.00043281889400748703,
      "loss": 2.282,
      "step": 81565
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.80479896068573,
      "learning_rate": 0.00043281522617149817,
      "loss": 2.2604,
      "step": 81566
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.8310761451721191,
      "learning_rate": 0.0004328115583108166,
      "loss": 2.1786,
      "step": 81567
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.9258825778961182,
      "learning_rate": 0.0004328078904254427,
      "loss": 2.0589,
      "step": 81568
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.7605021595954895,
      "learning_rate": 0.00043280422251537734,
      "loss": 2.1998,
      "step": 81569
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.8406379818916321,
      "learning_rate": 0.0004328005545806213,
      "loss": 2.3838,
      "step": 81570
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.9241769313812256,
      "learning_rate": 0.00043279688662117496,
      "loss": 2.258,
      "step": 81571
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.8836126923561096,
      "learning_rate": 0.0004327932186370392,
      "loss": 2.0595,
      "step": 81572
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.7798685431480408,
      "learning_rate": 0.00043278955062821473,
      "loss": 1.9724,
      "step": 81573
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.911538302898407,
      "learning_rate": 0.0004327858825947022,
      "loss": 2.3556,
      "step": 81574
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.8430169820785522,
      "learning_rate": 0.00043278221453650216,
      "loss": 2.2346,
      "step": 81575
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.7845273017883301,
      "learning_rate": 0.00043277854645361547,
      "loss": 2.302,
      "step": 81576
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.9277201294898987,
      "learning_rate": 0.0004327748783460428,
      "loss": 2.277,
      "step": 81577
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.7834179997444153,
      "learning_rate": 0.00043277121021378473,
      "loss": 2.0981,
      "step": 81578
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.8579449653625488,
      "learning_rate": 0.00043276754205684193,
      "loss": 2.0906,
      "step": 81579
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.8125571012496948,
      "learning_rate": 0.0004327638738752152,
      "loss": 2.1446,
      "step": 81580
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.8133962750434875,
      "learning_rate": 0.0004327602056689051,
      "loss": 2.3521,
      "step": 81581
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.7783606648445129,
      "learning_rate": 0.0004327565374379124,
      "loss": 2.1561,
      "step": 81582
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.8270311951637268,
      "learning_rate": 0.0004327528691822378,
      "loss": 2.0825,
      "step": 81583
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.7568626999855042,
      "learning_rate": 0.0004327492009018819,
      "loss": 2.1047,
      "step": 81584
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.7893362045288086,
      "learning_rate": 0.0004327455325968454,
      "loss": 2.158,
      "step": 81585
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.7772785425186157,
      "learning_rate": 0.00043274186426712905,
      "loss": 2.1502,
      "step": 81586
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.8713739514350891,
      "learning_rate": 0.0004327381959127334,
      "loss": 2.3293,
      "step": 81587
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.7740610837936401,
      "learning_rate": 0.0004327345275336593,
      "loss": 2.3332,
      "step": 81588
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.8446027636528015,
      "learning_rate": 0.0004327308591299073,
      "loss": 2.2376,
      "step": 81589
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.8073152303695679,
      "learning_rate": 0.00043272719070147815,
      "loss": 2.3223,
      "step": 81590
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.7303893566131592,
      "learning_rate": 0.00043272352224837246,
      "loss": 2.1062,
      "step": 81591
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.8873828053474426,
      "learning_rate": 0.00043271985377059104,
      "loss": 2.2405,
      "step": 81592
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.8228322863578796,
      "learning_rate": 0.0004327161852681344,
      "loss": 2.1087,
      "step": 81593
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.7677196860313416,
      "learning_rate": 0.0004327125167410034,
      "loss": 2.2497,
      "step": 81594
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.8119992613792419,
      "learning_rate": 0.0004327088481891986,
      "loss": 2.2017,
      "step": 81595
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.8161246180534363,
      "learning_rate": 0.0004327051796127207,
      "loss": 1.9801,
      "step": 81596
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.7477669715881348,
      "learning_rate": 0.0004327015110115704,
      "loss": 2.0972,
      "step": 81597
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.8091652989387512,
      "learning_rate": 0.00043269784238574846,
      "loss": 2.0508,
      "step": 81598
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.8264036178588867,
      "learning_rate": 0.00043269417373525547,
      "loss": 2.247,
      "step": 81599
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.9202671647071838,
      "learning_rate": 0.00043269050506009206,
      "loss": 2.2767,
      "step": 81600
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.9246912598609924,
      "learning_rate": 0.00043268683636025904,
      "loss": 2.1017,
      "step": 81601
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.8303109407424927,
      "learning_rate": 0.00043268316763575706,
      "loss": 2.2035,
      "step": 81602
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.9511882662773132,
      "learning_rate": 0.00043267949888658666,
      "loss": 2.1736,
      "step": 81603
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.9446730017662048,
      "learning_rate": 0.0004326758301127487,
      "loss": 2.2122,
      "step": 81604
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.878974199295044,
      "learning_rate": 0.0004326721613142438,
      "loss": 2.415,
      "step": 81605
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.8708125352859497,
      "learning_rate": 0.00043266849249107265,
      "loss": 2.253,
      "step": 81606
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.828597903251648,
      "learning_rate": 0.0004326648236432359,
      "loss": 2.3171,
      "step": 81607
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.8157596588134766,
      "learning_rate": 0.00043266115477073424,
      "loss": 2.1142,
      "step": 81608
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.8094651103019714,
      "learning_rate": 0.00043265748587356846,
      "loss": 2.1014,
      "step": 81609
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.7914835214614868,
      "learning_rate": 0.0004326538169517391,
      "loss": 2.1725,
      "step": 81610
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.844703197479248,
      "learning_rate": 0.0004326501480052468,
      "loss": 2.2937,
      "step": 81611
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.8013208508491516,
      "learning_rate": 0.00043264647903409247,
      "loss": 2.2009,
      "step": 81612
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.7771954536437988,
      "learning_rate": 0.0004326428100382766,
      "loss": 2.1967,
      "step": 81613
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.8599364757537842,
      "learning_rate": 0.0004326391410177999,
      "loss": 2.2164,
      "step": 81614
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.86471027135849,
      "learning_rate": 0.00043263547197266305,
      "loss": 2.3277,
      "step": 81615
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.7625983357429504,
      "learning_rate": 0.00043263180290286686,
      "loss": 2.1441,
      "step": 81616
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.9621495008468628,
      "learning_rate": 0.0004326281338084119,
      "loss": 2.012,
      "step": 81617
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.7926591634750366,
      "learning_rate": 0.0004326244646892988,
      "loss": 2.3669,
      "step": 81618
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.7968742251396179,
      "learning_rate": 0.0004326207955455284,
      "loss": 2.1906,
      "step": 81619
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.8918924331665039,
      "learning_rate": 0.00043261712637710116,
      "loss": 2.0629,
      "step": 81620
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.8209018707275391,
      "learning_rate": 0.00043261345718401795,
      "loss": 2.1426,
      "step": 81621
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.7865509390830994,
      "learning_rate": 0.0004326097879662794,
      "loss": 2.0174,
      "step": 81622
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.8479867577552795,
      "learning_rate": 0.00043260611872388623,
      "loss": 1.9602,
      "step": 81623
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.8865031003952026,
      "learning_rate": 0.000432602449456839,
      "loss": 2.0199,
      "step": 81624
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.8389922380447388,
      "learning_rate": 0.00043259878016513857,
      "loss": 2.1902,
      "step": 81625
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.8558654189109802,
      "learning_rate": 0.0004325951108487855,
      "loss": 2.2755,
      "step": 81626
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.9364673495292664,
      "learning_rate": 0.00043259144150778044,
      "loss": 2.1882,
      "step": 81627
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.7312769889831543,
      "learning_rate": 0.0004325877721421242,
      "loss": 2.1186,
      "step": 81628
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.8096812963485718,
      "learning_rate": 0.0004325841027518172,
      "loss": 2.1611,
      "step": 81629
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.8773483633995056,
      "learning_rate": 0.00043258043333686054,
      "loss": 2.0612,
      "step": 81630
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.7826159000396729,
      "learning_rate": 0.00043257676389725464,
      "loss": 2.1264,
      "step": 81631
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.8435207605361938,
      "learning_rate": 0.0004325730944330001,
      "loss": 2.1692,
      "step": 81632
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.7745461463928223,
      "learning_rate": 0.0004325694249440978,
      "loss": 2.1545,
      "step": 81633
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.7958977818489075,
      "learning_rate": 0.00043256575543054835,
      "loss": 2.0804,
      "step": 81634
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.7680741548538208,
      "learning_rate": 0.0004325620858923524,
      "loss": 2.5874,
      "step": 81635
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.7839905023574829,
      "learning_rate": 0.0004325584163295106,
      "loss": 2.2175,
      "step": 81636
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.7582153677940369,
      "learning_rate": 0.00043255474674202377,
      "loss": 2.2543,
      "step": 81637
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.8236688375473022,
      "learning_rate": 0.0004325510771298926,
      "loss": 2.0874,
      "step": 81638
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.7907158136367798,
      "learning_rate": 0.00043254740749311746,
      "loss": 2.1852,
      "step": 81639
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.9776286482810974,
      "learning_rate": 0.00043254373783169935,
      "loss": 2.4036,
      "step": 81640
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.9672427773475647,
      "learning_rate": 0.00043254006814563896,
      "loss": 2.174,
      "step": 81641
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.9202657341957092,
      "learning_rate": 0.00043253639843493676,
      "loss": 2.2285,
      "step": 81642
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.8200789093971252,
      "learning_rate": 0.00043253272869959357,
      "loss": 2.2154,
      "step": 81643
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.7952446937561035,
      "learning_rate": 0.00043252905893961005,
      "loss": 2.4224,
      "step": 81644
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.8099909424781799,
      "learning_rate": 0.00043252538915498684,
      "loss": 2.1416,
      "step": 81645
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.819233775138855,
      "learning_rate": 0.0004325217193457247,
      "loss": 2.2432,
      "step": 81646
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.9245279431343079,
      "learning_rate": 0.00043251804951182423,
      "loss": 2.2244,
      "step": 81647
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.7995789647102356,
      "learning_rate": 0.0004325143796532862,
      "loss": 2.067,
      "step": 81648
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.8325046300888062,
      "learning_rate": 0.00043251070977011117,
      "loss": 2.3492,
      "step": 81649
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.822556734085083,
      "learning_rate": 0.0004325070398623001,
      "loss": 1.9962,
      "step": 81650
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.872979998588562,
      "learning_rate": 0.0004325033699298533,
      "loss": 2.0997,
      "step": 81651
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.7830531001091003,
      "learning_rate": 0.00043249969997277157,
      "loss": 2.1933,
      "step": 81652
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.8569461703300476,
      "learning_rate": 0.0004324960299910558,
      "loss": 2.1571,
      "step": 81653
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.7833191752433777,
      "learning_rate": 0.0004324923599847064,
      "loss": 2.1085,
      "step": 81654
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.8629951477050781,
      "learning_rate": 0.00043248868995372417,
      "loss": 1.8404,
      "step": 81655
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.7263813614845276,
      "learning_rate": 0.0004324850198981099,
      "loss": 2.4709,
      "step": 81656
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.7749262452125549,
      "learning_rate": 0.00043248134981786404,
      "loss": 2.3052,
      "step": 81657
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.871036171913147,
      "learning_rate": 0.00043247767971298746,
      "loss": 2.158,
      "step": 81658
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.7483974695205688,
      "learning_rate": 0.0004324740095834808,
      "loss": 2.0909,
      "step": 81659
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.8414963483810425,
      "learning_rate": 0.00043247033942934467,
      "loss": 2.1766,
      "step": 81660
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.8171433210372925,
      "learning_rate": 0.00043246666925057976,
      "loss": 2.2383,
      "step": 81661
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.806439995765686,
      "learning_rate": 0.0004324629990471869,
      "loss": 2.3178,
      "step": 81662
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.8080320954322815,
      "learning_rate": 0.00043245932881916656,
      "loss": 2.1168,
      "step": 81663
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.838904857635498,
      "learning_rate": 0.00043245565856651967,
      "loss": 2.0694,
      "step": 81664
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.8813570737838745,
      "learning_rate": 0.0004324519882892467,
      "loss": 2.3482,
      "step": 81665
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.7056735157966614,
      "learning_rate": 0.0004324483179873484,
      "loss": 2.2515,
      "step": 81666
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.81638103723526,
      "learning_rate": 0.00043244464766082545,
      "loss": 2.1685,
      "step": 81667
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.8845722675323486,
      "learning_rate": 0.0004324409773096785,
      "loss": 2.2383,
      "step": 81668
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.8275502324104309,
      "learning_rate": 0.0004324373069339084,
      "loss": 2.2088,
      "step": 81669
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.7863513231277466,
      "learning_rate": 0.0004324336365335156,
      "loss": 2.2165,
      "step": 81670
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.9002556204795837,
      "learning_rate": 0.00043242996610850094,
      "loss": 2.1656,
      "step": 81671
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.8094411492347717,
      "learning_rate": 0.000432426295658865,
      "loss": 2.1188,
      "step": 81672
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.7901530265808105,
      "learning_rate": 0.0004324226251846085,
      "loss": 2.239,
      "step": 81673
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.7267877459526062,
      "learning_rate": 0.0004324189546857322,
      "loss": 2.0542,
      "step": 81674
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.9498149156570435,
      "learning_rate": 0.00043241528416223667,
      "loss": 2.0585,
      "step": 81675
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.7555959820747375,
      "learning_rate": 0.0004324116136141227,
      "loss": 2.1164,
      "step": 81676
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.9099348187446594,
      "learning_rate": 0.00043240794304139085,
      "loss": 2.2362,
      "step": 81677
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.8701882362365723,
      "learning_rate": 0.00043240427244404194,
      "loss": 2.1362,
      "step": 81678
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.7710030674934387,
      "learning_rate": 0.0004324006018220765,
      "loss": 2.3103,
      "step": 81679
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.8397509455680847,
      "learning_rate": 0.00043239693117549526,
      "loss": 2.2843,
      "step": 81680
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.8326753973960876,
      "learning_rate": 0.000432393260504299,
      "loss": 2.1557,
      "step": 81681
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.8076992034912109,
      "learning_rate": 0.0004323895898084883,
      "loss": 2.0911,
      "step": 81682
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.7883366942405701,
      "learning_rate": 0.00043238591908806393,
      "loss": 2.1289,
      "step": 81683
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.8055498003959656,
      "learning_rate": 0.00043238224834302646,
      "loss": 2.0283,
      "step": 81684
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.7345407009124756,
      "learning_rate": 0.00043237857757337663,
      "loss": 2.3326,
      "step": 81685
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.8014696836471558,
      "learning_rate": 0.00043237490677911514,
      "loss": 2.0975,
      "step": 81686
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.9054970145225525,
      "learning_rate": 0.00043237123596024263,
      "loss": 2.4551,
      "step": 81687
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.8376536965370178,
      "learning_rate": 0.0004323675651167599,
      "loss": 2.1305,
      "step": 81688
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.7318776249885559,
      "learning_rate": 0.0004323638942486674,
      "loss": 2.276,
      "step": 81689
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.845025897026062,
      "learning_rate": 0.0004323602233559661,
      "loss": 2.3211,
      "step": 81690
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.7935422658920288,
      "learning_rate": 0.00043235655243865646,
      "loss": 2.0557,
      "step": 81691
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.7660543918609619,
      "learning_rate": 0.00043235288149673926,
      "loss": 2.2156,
      "step": 81692
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.7747331857681274,
      "learning_rate": 0.0004323492105302151,
      "loss": 2.2899,
      "step": 81693
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.7940348982810974,
      "learning_rate": 0.00043234553953908474,
      "loss": 2.0989,
      "step": 81694
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.8551971316337585,
      "learning_rate": 0.00043234186852334893,
      "loss": 2.0984,
      "step": 81695
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.7959882020950317,
      "learning_rate": 0.00043233819748300824,
      "loss": 2.2874,
      "step": 81696
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.7625153660774231,
      "learning_rate": 0.00043233452641806326,
      "loss": 2.2009,
      "step": 81697
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.8041102290153503,
      "learning_rate": 0.00043233085532851497,
      "loss": 2.2594,
      "step": 81698
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.7777401208877563,
      "learning_rate": 0.0004323271842143638,
      "loss": 2.1388,
      "step": 81699
    },
    {
      "epoch": 1.06,
      "grad_norm": 1.1072615385055542,
      "learning_rate": 0.00043232351307561045,
      "loss": 2.1556,
      "step": 81700
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.9050714373588562,
      "learning_rate": 0.0004323198419122558,
      "loss": 2.0685,
      "step": 81701
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.8380647897720337,
      "learning_rate": 0.00043231617072430023,
      "loss": 2.2514,
      "step": 81702
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.7335424423217773,
      "learning_rate": 0.0004323124995117447,
      "loss": 2.05,
      "step": 81703
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.8656161427497864,
      "learning_rate": 0.00043230882827458977,
      "loss": 2.3981,
      "step": 81704
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.761813223361969,
      "learning_rate": 0.00043230515701283607,
      "loss": 2.3214,
      "step": 81705
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.7998489737510681,
      "learning_rate": 0.0004323014857264844,
      "loss": 2.185,
      "step": 81706
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.7878658175468445,
      "learning_rate": 0.0004322978144155353,
      "loss": 2.072,
      "step": 81707
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.7588058710098267,
      "learning_rate": 0.00043229414307998965,
      "loss": 2.2921,
      "step": 81708
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.7304250001907349,
      "learning_rate": 0.00043229047171984797,
      "loss": 2.2369,
      "step": 81709
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.8574177026748657,
      "learning_rate": 0.00043228680033511094,
      "loss": 2.3007,
      "step": 81710
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.7441083788871765,
      "learning_rate": 0.00043228312892577944,
      "loss": 2.1812,
      "step": 81711
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.8091170787811279,
      "learning_rate": 0.0004322794574918539,
      "loss": 2.3447,
      "step": 81712
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.7702744007110596,
      "learning_rate": 0.0004322757860333351,
      "loss": 2.1763,
      "step": 81713
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.753810465335846,
      "learning_rate": 0.0004322721145502238,
      "loss": 2.1057,
      "step": 81714
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.8066391944885254,
      "learning_rate": 0.0004322684430425206,
      "loss": 2.1884,
      "step": 81715
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.7808447480201721,
      "learning_rate": 0.00043226477151022617,
      "loss": 2.163,
      "step": 81716
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.7603339552879333,
      "learning_rate": 0.00043226109995334126,
      "loss": 2.3766,
      "step": 81717
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.7843356132507324,
      "learning_rate": 0.0004322574283718665,
      "loss": 2.1258,
      "step": 81718
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.7362874150276184,
      "learning_rate": 0.00043225375676580254,
      "loss": 2.1649,
      "step": 81719
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.7317103147506714,
      "learning_rate": 0.00043225008513515014,
      "loss": 2.2075,
      "step": 81720
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.878365159034729,
      "learning_rate": 0.00043224641347991,
      "loss": 2.214,
      "step": 81721
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.8117865324020386,
      "learning_rate": 0.0004322427418000828,
      "loss": 2.3871,
      "step": 81722
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.7780164480209351,
      "learning_rate": 0.00043223907009566897,
      "loss": 2.1285,
      "step": 81723
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.797461748123169,
      "learning_rate": 0.00043223539836666956,
      "loss": 2.2294,
      "step": 81724
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.7936904430389404,
      "learning_rate": 0.000432231726613085,
      "loss": 2.1756,
      "step": 81725
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.9969821572303772,
      "learning_rate": 0.00043222805483491617,
      "loss": 2.028,
      "step": 81726
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.8032474517822266,
      "learning_rate": 0.0004322243830321636,
      "loss": 2.4636,
      "step": 81727
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.8722911477088928,
      "learning_rate": 0.000432220711204828,
      "loss": 2.4019,
      "step": 81728
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.8698251843452454,
      "learning_rate": 0.0004322170393529102,
      "loss": 2.2761,
      "step": 81729
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.7894955277442932,
      "learning_rate": 0.00043221336747641066,
      "loss": 2.3416,
      "step": 81730
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.7853080034255981,
      "learning_rate": 0.0004322096955753301,
      "loss": 2.0853,
      "step": 81731
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.7561497688293457,
      "learning_rate": 0.0004322060236496694,
      "loss": 2.1633,
      "step": 81732
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.7711586356163025,
      "learning_rate": 0.000432202351699429,
      "loss": 2.1451,
      "step": 81733
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.8595628142356873,
      "learning_rate": 0.0004321986797246096,
      "loss": 2.1305,
      "step": 81734
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.7797290682792664,
      "learning_rate": 0.00043219500772521214,
      "loss": 2.1996,
      "step": 81735
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.7544559240341187,
      "learning_rate": 0.00043219133570123705,
      "loss": 2.1744,
      "step": 81736
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.8598676323890686,
      "learning_rate": 0.00043218766365268516,
      "loss": 2.1733,
      "step": 81737
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.830085277557373,
      "learning_rate": 0.00043218399157955706,
      "loss": 2.1584,
      "step": 81738
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.809834361076355,
      "learning_rate": 0.0004321803194818534,
      "loss": 2.1354,
      "step": 81739
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.7359780073165894,
      "learning_rate": 0.00043217664735957506,
      "loss": 2.275,
      "step": 81740
    },
    {
      "epoch": 1.06,
      "grad_norm": 1.0202726125717163,
      "learning_rate": 0.00043217297521272246,
      "loss": 2.0031,
      "step": 81741
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.8155772686004639,
      "learning_rate": 0.0004321693030412964,
      "loss": 2.2135,
      "step": 81742
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.9025503396987915,
      "learning_rate": 0.0004321656308452977,
      "loss": 2.1778,
      "step": 81743
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.8373359441757202,
      "learning_rate": 0.0004321619586247269,
      "loss": 2.3524,
      "step": 81744
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.8888542652130127,
      "learning_rate": 0.0004321582863795846,
      "loss": 2.147,
      "step": 81745
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.8392764925956726,
      "learning_rate": 0.0004321546141098715,
      "loss": 2.1651,
      "step": 81746
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.8787331581115723,
      "learning_rate": 0.0004321509418155886,
      "loss": 2.2203,
      "step": 81747
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.7845590114593506,
      "learning_rate": 0.00043214726949673615,
      "loss": 2.1043,
      "step": 81748
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.8874034881591797,
      "learning_rate": 0.0004321435971533151,
      "loss": 2.3196,
      "step": 81749
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.8947757482528687,
      "learning_rate": 0.0004321399247853261,
      "loss": 2.2038,
      "step": 81750
    },
    {
      "epoch": 1.06,
      "grad_norm": 1.2724385261535645,
      "learning_rate": 0.0004321362523927697,
      "loss": 2.0119,
      "step": 81751
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.811183750629425,
      "learning_rate": 0.0004321325799756467,
      "loss": 2.2105,
      "step": 81752
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.8209449648857117,
      "learning_rate": 0.00043212890753395786,
      "loss": 2.1597,
      "step": 81753
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.8077278137207031,
      "learning_rate": 0.0004321252350677037,
      "loss": 2.2587,
      "step": 81754
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.9592207074165344,
      "learning_rate": 0.00043212156257688486,
      "loss": 2.1358,
      "step": 81755
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.7822656631469727,
      "learning_rate": 0.0004321178900615023,
      "loss": 2.127,
      "step": 81756
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.8730166554450989,
      "learning_rate": 0.0004321142175215565,
      "loss": 2.3918,
      "step": 81757
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.7435600161552429,
      "learning_rate": 0.000432110544957048,
      "loss": 1.8795,
      "step": 81758
    },
    {
      "epoch": 1.06,
      "grad_norm": 1.0642130374908447,
      "learning_rate": 0.00043210687236797785,
      "loss": 2.236,
      "step": 81759
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.8217752575874329,
      "learning_rate": 0.0004321031997543465,
      "loss": 2.1842,
      "step": 81760
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.8622695207595825,
      "learning_rate": 0.00043209952711615457,
      "loss": 2.1769,
      "step": 81761
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.9794418811798096,
      "learning_rate": 0.000432095854453403,
      "loss": 2.0615,
      "step": 81762
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.9457009434700012,
      "learning_rate": 0.0004320921817660922,
      "loss": 2.3301,
      "step": 81763
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.8165624141693115,
      "learning_rate": 0.00043208850905422294,
      "loss": 2.091,
      "step": 81764
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.9491126537322998,
      "learning_rate": 0.00043208483631779605,
      "loss": 2.0585,
      "step": 81765
    },
    {
      "epoch": 1.06,
      "grad_norm": 1.0066180229187012,
      "learning_rate": 0.00043208116355681204,
      "loss": 2.116,
      "step": 81766
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.8019520044326782,
      "learning_rate": 0.00043207749077127166,
      "loss": 2.0116,
      "step": 81767
    },
    {
      "epoch": 1.06,
      "grad_norm": 1.0404292345046997,
      "learning_rate": 0.0004320738179611756,
      "loss": 2.2259,
      "step": 81768
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.8696482181549072,
      "learning_rate": 0.00043207014512652446,
      "loss": 2.1627,
      "step": 81769
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.7733590602874756,
      "learning_rate": 0.00043206647226731905,
      "loss": 2.1368,
      "step": 81770
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.7767433524131775,
      "learning_rate": 0.00043206279938356,
      "loss": 2.1887,
      "step": 81771
    },
    {
      "epoch": 1.06,
      "grad_norm": 1.045821189880371,
      "learning_rate": 0.00043205912647524785,
      "loss": 2.0964,
      "step": 81772
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.7813200950622559,
      "learning_rate": 0.00043205545354238354,
      "loss": 2.2278,
      "step": 81773
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.8585524559020996,
      "learning_rate": 0.0004320517805849676,
      "loss": 2.1164,
      "step": 81774
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.7799698710441589,
      "learning_rate": 0.00043204810760300076,
      "loss": 2.0501,
      "step": 81775
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.7792431712150574,
      "learning_rate": 0.0004320444345964836,
      "loss": 2.1739,
      "step": 81776
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.8413525819778442,
      "learning_rate": 0.0004320407615654169,
      "loss": 2.4513,
      "step": 81777
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.7712026238441467,
      "learning_rate": 0.00043203708850980145,
      "loss": 2.1143,
      "step": 81778
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.7647545337677002,
      "learning_rate": 0.0004320334154296378,
      "loss": 2.2756,
      "step": 81779
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.8025570511817932,
      "learning_rate": 0.0004320297423249265,
      "loss": 2.063,
      "step": 81780
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.7757824659347534,
      "learning_rate": 0.0004320260691956686,
      "loss": 2.1715,
      "step": 81781
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.8252231478691101,
      "learning_rate": 0.00043202239604186435,
      "loss": 2.2193,
      "step": 81782
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.961220383644104,
      "learning_rate": 0.0004320187228635147,
      "loss": 2.3083,
      "step": 81783
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.8177346587181091,
      "learning_rate": 0.00043201504966062035,
      "loss": 2.1384,
      "step": 81784
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.9227612018585205,
      "learning_rate": 0.0004320113764331818,
      "loss": 2.2002,
      "step": 81785
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.7142562866210938,
      "learning_rate": 0.00043200770318119994,
      "loss": 2.2491,
      "step": 81786
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.8914186954498291,
      "learning_rate": 0.00043200402990467535,
      "loss": 2.0365,
      "step": 81787
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.8195508122444153,
      "learning_rate": 0.0004320003566036087,
      "loss": 2.32,
      "step": 81788
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.8919023871421814,
      "learning_rate": 0.0004319966832780006,
      "loss": 2.3569,
      "step": 81789
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.8193759918212891,
      "learning_rate": 0.00043199300992785203,
      "loss": 2.0625,
      "step": 81790
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.7991008162498474,
      "learning_rate": 0.0004319893365531633,
      "loss": 2.0275,
      "step": 81791
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.852933406829834,
      "learning_rate": 0.00043198566315393527,
      "loss": 2.3185,
      "step": 81792
    },
    {
      "epoch": 1.06,
      "grad_norm": 0.7960036993026733,
      "learning_rate": 0.0004319819897301687,
      "loss": 2.1306,
      "step": 81793
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.7833894491195679,
      "learning_rate": 0.0004319783162818641,
      "loss": 2.1928,
      "step": 81794
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.9004775881767273,
      "learning_rate": 0.00043197464280902233,
      "loss": 2.1664,
      "step": 81795
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.8355029225349426,
      "learning_rate": 0.0004319709693116439,
      "loss": 2.3198,
      "step": 81796
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.7919629812240601,
      "learning_rate": 0.00043196729578972963,
      "loss": 2.236,
      "step": 81797
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.7604760527610779,
      "learning_rate": 0.00043196362224328007,
      "loss": 1.9788,
      "step": 81798
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.7886080145835876,
      "learning_rate": 0.00043195994867229614,
      "loss": 2.144,
      "step": 81799
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.7859771847724915,
      "learning_rate": 0.0004319562750767782,
      "loss": 2.0388,
      "step": 81800
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.7852991223335266,
      "learning_rate": 0.00043195260145672713,
      "loss": 2.3089,
      "step": 81801
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.8557747602462769,
      "learning_rate": 0.0004319489278121436,
      "loss": 2.3862,
      "step": 81802
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.8039577603340149,
      "learning_rate": 0.00043194525414302833,
      "loss": 2.0601,
      "step": 81803
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.8585014343261719,
      "learning_rate": 0.00043194158044938184,
      "loss": 2.3865,
      "step": 81804
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.7701106071472168,
      "learning_rate": 0.000431937906731205,
      "loss": 2.1693,
      "step": 81805
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.8046770691871643,
      "learning_rate": 0.0004319342329884983,
      "loss": 2.1111,
      "step": 81806
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.8400660753250122,
      "learning_rate": 0.00043193055922126274,
      "loss": 2.2529,
      "step": 81807
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.8022160530090332,
      "learning_rate": 0.00043192688542949867,
      "loss": 2.3851,
      "step": 81808
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.7363412976264954,
      "learning_rate": 0.00043192321161320685,
      "loss": 2.305,
      "step": 81809
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.7434366345405579,
      "learning_rate": 0.00043191953777238815,
      "loss": 2.2763,
      "step": 81810
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.7210142016410828,
      "learning_rate": 0.000431915863907043,
      "loss": 2.1311,
      "step": 81811
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.908682644367218,
      "learning_rate": 0.0004319121900171723,
      "loss": 2.2028,
      "step": 81812
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.7858899235725403,
      "learning_rate": 0.0004319085161027765,
      "loss": 2.1076,
      "step": 81813
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.8081051707267761,
      "learning_rate": 0.00043190484216385655,
      "loss": 2.1149,
      "step": 81814
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.9172825813293457,
      "learning_rate": 0.00043190116820041296,
      "loss": 2.1936,
      "step": 81815
    },
    {
      "epoch": 1.07,
      "grad_norm": 1.0861021280288696,
      "learning_rate": 0.00043189749421244645,
      "loss": 2.1847,
      "step": 81816
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.9211884140968323,
      "learning_rate": 0.00043189382019995774,
      "loss": 2.2888,
      "step": 81817
    },
    {
      "epoch": 1.07,
      "grad_norm": 1.1250826120376587,
      "learning_rate": 0.00043189014616294735,
      "loss": 2.2429,
      "step": 81818
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.8721649050712585,
      "learning_rate": 0.0004318864721014162,
      "loss": 2.4097,
      "step": 81819
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.8783298134803772,
      "learning_rate": 0.0004318827980153649,
      "loss": 2.3588,
      "step": 81820
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.8175653219223022,
      "learning_rate": 0.00043187912390479395,
      "loss": 2.2924,
      "step": 81821
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.8070108294487,
      "learning_rate": 0.00043187544976970436,
      "loss": 2.1323,
      "step": 81822
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.7561386823654175,
      "learning_rate": 0.00043187177561009656,
      "loss": 2.0912,
      "step": 81823
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.7462287545204163,
      "learning_rate": 0.0004318681014259712,
      "loss": 2.0301,
      "step": 81824
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.9599471092224121,
      "learning_rate": 0.00043186442721732923,
      "loss": 2.0706,
      "step": 81825
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.9045935273170471,
      "learning_rate": 0.0004318607529841711,
      "loss": 2.4013,
      "step": 81826
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.7389222979545593,
      "learning_rate": 0.0004318570787264976,
      "loss": 2.3147,
      "step": 81827
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.8383741974830627,
      "learning_rate": 0.00043185340444430937,
      "loss": 2.2715,
      "step": 81828
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.8756577372550964,
      "learning_rate": 0.0004318497301376071,
      "loss": 2.3502,
      "step": 81829
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.81772381067276,
      "learning_rate": 0.00043184605580639146,
      "loss": 2.1783,
      "step": 81830
    },
    {
      "epoch": 1.07,
      "grad_norm": 1.166633129119873,
      "learning_rate": 0.00043184238145066315,
      "loss": 2.1512,
      "step": 81831
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.8917701244354248,
      "learning_rate": 0.0004318387070704229,
      "loss": 2.0796,
      "step": 81832
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.8117952942848206,
      "learning_rate": 0.0004318350326656713,
      "loss": 2.1938,
      "step": 81833
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.9069105982780457,
      "learning_rate": 0.00043183135823640905,
      "loss": 2.1904,
      "step": 81834
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.9812577962875366,
      "learning_rate": 0.00043182768378263693,
      "loss": 2.2781,
      "step": 81835
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.7955308556556702,
      "learning_rate": 0.00043182400930435555,
      "loss": 2.163,
      "step": 81836
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.8421157002449036,
      "learning_rate": 0.0004318203348015655,
      "loss": 2.3289,
      "step": 81837
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.9743717908859253,
      "learning_rate": 0.00043181666027426773,
      "loss": 2.2012,
      "step": 81838
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.8571631908416748,
      "learning_rate": 0.00043181298572246264,
      "loss": 2.2334,
      "step": 81839
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.8177583813667297,
      "learning_rate": 0.000431809311146151,
      "loss": 2.2743,
      "step": 81840
    },
    {
      "epoch": 1.07,
      "grad_norm": 1.010013461112976,
      "learning_rate": 0.00043180563654533354,
      "loss": 2.2926,
      "step": 81841
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.7870632410049438,
      "learning_rate": 0.000431801961920011,
      "loss": 2.1342,
      "step": 81842
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.8899704813957214,
      "learning_rate": 0.00043179828727018393,
      "loss": 2.3027,
      "step": 81843
    },
    {
      "epoch": 1.07,
      "grad_norm": 1.0143598318099976,
      "learning_rate": 0.00043179461259585305,
      "loss": 2.1594,
      "step": 81844
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.9845049381256104,
      "learning_rate": 0.00043179093789701915,
      "loss": 2.4722,
      "step": 81845
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.828569233417511,
      "learning_rate": 0.0004317872631736827,
      "loss": 2.0299,
      "step": 81846
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.8978038430213928,
      "learning_rate": 0.0004317835884258446,
      "loss": 2.2438,
      "step": 81847
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.863862931728363,
      "learning_rate": 0.0004317799136535055,
      "loss": 2.2942,
      "step": 81848
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.8571987152099609,
      "learning_rate": 0.00043177623885666587,
      "loss": 2.2321,
      "step": 81849
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.8191021680831909,
      "learning_rate": 0.0004317725640353266,
      "loss": 2.4399,
      "step": 81850
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.8316926956176758,
      "learning_rate": 0.0004317688891894884,
      "loss": 2.3096,
      "step": 81851
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.8613873720169067,
      "learning_rate": 0.0004317652143191518,
      "loss": 2.38,
      "step": 81852
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.8973730802536011,
      "learning_rate": 0.00043176153942431757,
      "loss": 2.3003,
      "step": 81853
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.8101435899734497,
      "learning_rate": 0.0004317578645049865,
      "loss": 2.2117,
      "step": 81854
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.7766887545585632,
      "learning_rate": 0.000431754189561159,
      "loss": 2.4009,
      "step": 81855
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.84698086977005,
      "learning_rate": 0.0004317505145928359,
      "loss": 2.172,
      "step": 81856
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.7461196184158325,
      "learning_rate": 0.00043174683960001803,
      "loss": 2.2965,
      "step": 81857
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.7986114025115967,
      "learning_rate": 0.00043174316458270587,
      "loss": 2.0053,
      "step": 81858
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.8845654726028442,
      "learning_rate": 0.0004317394895409001,
      "loss": 2.0047,
      "step": 81859
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.7101044654846191,
      "learning_rate": 0.00043173581447460157,
      "loss": 2.0374,
      "step": 81860
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.8799782395362854,
      "learning_rate": 0.0004317321393838108,
      "loss": 2.2384,
      "step": 81861
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.9428746104240417,
      "learning_rate": 0.0004317284642685286,
      "loss": 2.2359,
      "step": 81862
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.8460880517959595,
      "learning_rate": 0.0004317247891287556,
      "loss": 2.1561,
      "step": 81863
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.8360738754272461,
      "learning_rate": 0.00043172111396449245,
      "loss": 2.3147,
      "step": 81864
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.8772745728492737,
      "learning_rate": 0.0004317174387757398,
      "loss": 2.2555,
      "step": 81865
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.8317235708236694,
      "learning_rate": 0.0004317137635624985,
      "loss": 1.9496,
      "step": 81866
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.8078173995018005,
      "learning_rate": 0.000431710088324769,
      "loss": 2.3339,
      "step": 81867
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.7522526383399963,
      "learning_rate": 0.00043170641306255217,
      "loss": 2.1541,
      "step": 81868
    },
    {
      "epoch": 1.07,
      "grad_norm": 1.2120598554611206,
      "learning_rate": 0.00043170273777584867,
      "loss": 2.0799,
      "step": 81869
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.8142341375350952,
      "learning_rate": 0.0004316990624646591,
      "loss": 2.0862,
      "step": 81870
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.824991762638092,
      "learning_rate": 0.00043169538712898415,
      "loss": 2.0323,
      "step": 81871
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.8116843700408936,
      "learning_rate": 0.0004316917117688246,
      "loss": 2.2874,
      "step": 81872
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.8410885334014893,
      "learning_rate": 0.00043168803638418105,
      "loss": 1.9471,
      "step": 81873
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.9455702304840088,
      "learning_rate": 0.0004316843609750542,
      "loss": 2.2454,
      "step": 81874
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.8131135106086731,
      "learning_rate": 0.0004316806855414448,
      "loss": 2.1783,
      "step": 81875
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.9860941171646118,
      "learning_rate": 0.00043167701008335347,
      "loss": 2.2023,
      "step": 81876
    },
    {
      "epoch": 1.07,
      "grad_norm": 1.1677312850952148,
      "learning_rate": 0.0004316733346007808,
      "loss": 2.245,
      "step": 81877
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.8121670484542847,
      "learning_rate": 0.00043166965909372765,
      "loss": 2.2629,
      "step": 81878
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.7424972653388977,
      "learning_rate": 0.00043166598356219466,
      "loss": 2.1194,
      "step": 81879
    },
    {
      "epoch": 1.07,
      "grad_norm": 1.6075220108032227,
      "learning_rate": 0.0004316623080061824,
      "loss": 2.0997,
      "step": 81880
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.8668769598007202,
      "learning_rate": 0.00043165863242569164,
      "loss": 2.3391,
      "step": 81881
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.7969267964363098,
      "learning_rate": 0.0004316549568207231,
      "loss": 2.1868,
      "step": 81882
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.8056877255439758,
      "learning_rate": 0.00043165128119127737,
      "loss": 2.0873,
      "step": 81883
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.9530199766159058,
      "learning_rate": 0.00043164760553735524,
      "loss": 2.179,
      "step": 81884
    },
    {
      "epoch": 1.07,
      "grad_norm": 1.2952649593353271,
      "learning_rate": 0.0004316439298589573,
      "loss": 2.1582,
      "step": 81885
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.8126642107963562,
      "learning_rate": 0.0004316402541560842,
      "loss": 2.3136,
      "step": 81886
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.9317708611488342,
      "learning_rate": 0.0004316365784287367,
      "loss": 2.1556,
      "step": 81887
    },
    {
      "epoch": 1.07,
      "grad_norm": 1.0627652406692505,
      "learning_rate": 0.0004316329026769156,
      "loss": 2.0034,
      "step": 81888
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.87562096118927,
      "learning_rate": 0.0004316292269006214,
      "loss": 2.2819,
      "step": 81889
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.7996150851249695,
      "learning_rate": 0.0004316255510998548,
      "loss": 2.0587,
      "step": 81890
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.7830484509468079,
      "learning_rate": 0.00043162187527461665,
      "loss": 2.0978,
      "step": 81891
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.8596181869506836,
      "learning_rate": 0.00043161819942490735,
      "loss": 2.2061,
      "step": 81892
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.8575705289840698,
      "learning_rate": 0.00043161452355072776,
      "loss": 2.3517,
      "step": 81893
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.8231887817382812,
      "learning_rate": 0.00043161084765207865,
      "loss": 2.1058,
      "step": 81894
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.9052290320396423,
      "learning_rate": 0.00043160717172896056,
      "loss": 2.3419,
      "step": 81895
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.8614843487739563,
      "learning_rate": 0.00043160349578137425,
      "loss": 2.2339,
      "step": 81896
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.8358078002929688,
      "learning_rate": 0.00043159981980932027,
      "loss": 2.1606,
      "step": 81897
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.8575478196144104,
      "learning_rate": 0.0004315961438127994,
      "loss": 2.3117,
      "step": 81898
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.8662757277488708,
      "learning_rate": 0.0004315924677918124,
      "loss": 2.5313,
      "step": 81899
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.7806304693222046,
      "learning_rate": 0.00043158879174635983,
      "loss": 2.1024,
      "step": 81900
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.837037205696106,
      "learning_rate": 0.00043158511567644245,
      "loss": 2.1703,
      "step": 81901
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.825122594833374,
      "learning_rate": 0.0004315814395820609,
      "loss": 2.1155,
      "step": 81902
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.8242734670639038,
      "learning_rate": 0.0004315777634632158,
      "loss": 2.1552,
      "step": 81903
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.8258197903633118,
      "learning_rate": 0.0004315740873199081,
      "loss": 1.9726,
      "step": 81904
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.9595032930374146,
      "learning_rate": 0.00043157041115213815,
      "loss": 2.2246,
      "step": 81905
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.710016667842865,
      "learning_rate": 0.00043156673495990677,
      "loss": 1.9929,
      "step": 81906
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.7838044166564941,
      "learning_rate": 0.0004315630587432147,
      "loss": 2.074,
      "step": 81907
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.8278384208679199,
      "learning_rate": 0.00043155938250206255,
      "loss": 2.0196,
      "step": 81908
    },
    {
      "epoch": 1.07,
      "grad_norm": 1.0260196924209595,
      "learning_rate": 0.000431555706236451,
      "loss": 2.1637,
      "step": 81909
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.8096967339515686,
      "learning_rate": 0.0004315520299463808,
      "loss": 2.1864,
      "step": 81910
    },
    {
      "epoch": 1.07,
      "grad_norm": 1.0092076063156128,
      "learning_rate": 0.0004315483536318526,
      "loss": 2.1859,
      "step": 81911
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.7848634123802185,
      "learning_rate": 0.0004315446772928671,
      "loss": 2.1799,
      "step": 81912
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.9075690507888794,
      "learning_rate": 0.00043154100092942496,
      "loss": 2.085,
      "step": 81913
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.8269091844558716,
      "learning_rate": 0.00043153732454152675,
      "loss": 2.2391,
      "step": 81914
    },
    {
      "epoch": 1.07,
      "grad_norm": 1.725655198097229,
      "learning_rate": 0.00043153364812917344,
      "loss": 2.4479,
      "step": 81915
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.8184647560119629,
      "learning_rate": 0.00043152997169236545,
      "loss": 2.2144,
      "step": 81916
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.8239166736602783,
      "learning_rate": 0.00043152629523110345,
      "loss": 2.1546,
      "step": 81917
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.8487430810928345,
      "learning_rate": 0.0004315226187453884,
      "loss": 2.1756,
      "step": 81918
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.902122437953949,
      "learning_rate": 0.00043151894223522077,
      "loss": 2.1363,
      "step": 81919
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.8543177247047424,
      "learning_rate": 0.0004315152657006012,
      "loss": 2.1645,
      "step": 81920
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.7435388565063477,
      "learning_rate": 0.0004315115891415306,
      "loss": 2.2215,
      "step": 81921
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.7356237769126892,
      "learning_rate": 0.0004315079125580094,
      "loss": 2.0678,
      "step": 81922
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.9291090965270996,
      "learning_rate": 0.00043150423595003846,
      "loss": 2.0449,
      "step": 81923
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.8324939012527466,
      "learning_rate": 0.0004315005593176184,
      "loss": 2.4382,
      "step": 81924
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.7493781447410583,
      "learning_rate": 0.00043149688266074985,
      "loss": 2.1424,
      "step": 81925
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.8981614112854004,
      "learning_rate": 0.0004314932059794336,
      "loss": 2.2356,
      "step": 81926
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.9342609643936157,
      "learning_rate": 0.00043148952927367026,
      "loss": 2.0723,
      "step": 81927
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.836245059967041,
      "learning_rate": 0.00043148585254346053,
      "loss": 1.9237,
      "step": 81928
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.8882701396942139,
      "learning_rate": 0.0004314821757888051,
      "loss": 2.1867,
      "step": 81929
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.8292535543441772,
      "learning_rate": 0.00043147849900970466,
      "loss": 2.1519,
      "step": 81930
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.776945173740387,
      "learning_rate": 0.0004314748222061599,
      "loss": 2.1286,
      "step": 81931
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.8154252767562866,
      "learning_rate": 0.00043147114537817147,
      "loss": 2.0883,
      "step": 81932
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.8459036350250244,
      "learning_rate": 0.00043146746852574006,
      "loss": 2.1779,
      "step": 81933
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.841802179813385,
      "learning_rate": 0.00043146379164886635,
      "loss": 2.4653,
      "step": 81934
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.7863721251487732,
      "learning_rate": 0.0004314601147475511,
      "loss": 2.1881,
      "step": 81935
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.7943509817123413,
      "learning_rate": 0.0004314564378217949,
      "loss": 1.9958,
      "step": 81936
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.7947872877120972,
      "learning_rate": 0.0004314527608715985,
      "loss": 2.4452,
      "step": 81937
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.7184892892837524,
      "learning_rate": 0.00043144908389696244,
      "loss": 2.0684,
      "step": 81938
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.8377586007118225,
      "learning_rate": 0.00043144540689788764,
      "loss": 2.2982,
      "step": 81939
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.8247455358505249,
      "learning_rate": 0.00043144172987437455,
      "loss": 2.1113,
      "step": 81940
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.7634009718894958,
      "learning_rate": 0.00043143805282642403,
      "loss": 2.1318,
      "step": 81941
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.792395830154419,
      "learning_rate": 0.00043143437575403673,
      "loss": 2.3218,
      "step": 81942
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.7485373616218567,
      "learning_rate": 0.0004314306986572132,
      "loss": 2.0112,
      "step": 81943
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.7816972732543945,
      "learning_rate": 0.0004314270215359543,
      "loss": 2.0429,
      "step": 81944
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.816076934337616,
      "learning_rate": 0.0004314233443902606,
      "loss": 2.1703,
      "step": 81945
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.7549573183059692,
      "learning_rate": 0.0004314196672201329,
      "loss": 2.189,
      "step": 81946
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.7410850524902344,
      "learning_rate": 0.00043141599002557166,
      "loss": 2.2443,
      "step": 81947
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.796046793460846,
      "learning_rate": 0.0004314123128065778,
      "loss": 2.2841,
      "step": 81948
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.8628142476081848,
      "learning_rate": 0.0004314086355631519,
      "loss": 2.1896,
      "step": 81949
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.8626904487609863,
      "learning_rate": 0.0004314049582952946,
      "loss": 2.3879,
      "step": 81950
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.8064239621162415,
      "learning_rate": 0.00043140128100300676,
      "loss": 2.1468,
      "step": 81951
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.8603014945983887,
      "learning_rate": 0.0004313976036862888,
      "loss": 2.1109,
      "step": 81952
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.9742667078971863,
      "learning_rate": 0.0004313939263451416,
      "loss": 2.2497,
      "step": 81953
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.8976579308509827,
      "learning_rate": 0.00043139024897956583,
      "loss": 2.379,
      "step": 81954
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.8299369215965271,
      "learning_rate": 0.00043138657158956216,
      "loss": 2.124,
      "step": 81955
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.8279376029968262,
      "learning_rate": 0.0004313828941751311,
      "loss": 2.2325,
      "step": 81956
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.7777693867683411,
      "learning_rate": 0.0004313792167362736,
      "loss": 2.1896,
      "step": 81957
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.8815091252326965,
      "learning_rate": 0.0004313755392729902,
      "loss": 2.1333,
      "step": 81958
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.8177857995033264,
      "learning_rate": 0.00043137186178528154,
      "loss": 2.1597,
      "step": 81959
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.9596148729324341,
      "learning_rate": 0.00043136818427314856,
      "loss": 2.374,
      "step": 81960
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.8440916538238525,
      "learning_rate": 0.00043136450673659154,
      "loss": 2.1892,
      "step": 81961
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.7882018685340881,
      "learning_rate": 0.00043136082917561146,
      "loss": 1.979,
      "step": 81962
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.8171386122703552,
      "learning_rate": 0.0004313571515902089,
      "loss": 2.2923,
      "step": 81963
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.7798525094985962,
      "learning_rate": 0.00043135347398038466,
      "loss": 2.1778,
      "step": 81964
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.8064010143280029,
      "learning_rate": 0.00043134979634613924,
      "loss": 2.2797,
      "step": 81965
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.7452051639556885,
      "learning_rate": 0.00043134611868747346,
      "loss": 2.0075,
      "step": 81966
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.7773962616920471,
      "learning_rate": 0.00043134244100438793,
      "loss": 2.0957,
      "step": 81967
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.761172890663147,
      "learning_rate": 0.0004313387632968833,
      "loss": 1.9623,
      "step": 81968
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.8519788384437561,
      "learning_rate": 0.0004313350855649604,
      "loss": 2.2757,
      "step": 81969
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.781116247177124,
      "learning_rate": 0.00043133140780861974,
      "loss": 2.2492,
      "step": 81970
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.8548094630241394,
      "learning_rate": 0.00043132773002786223,
      "loss": 2.0753,
      "step": 81971
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.7930400967597961,
      "learning_rate": 0.0004313240522226884,
      "loss": 2.3309,
      "step": 81972
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.9336126446723938,
      "learning_rate": 0.00043132037439309884,
      "loss": 2.0488,
      "step": 81973
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.7668192386627197,
      "learning_rate": 0.0004313166965390944,
      "loss": 2.2013,
      "step": 81974
    },
    {
      "epoch": 1.07,
      "grad_norm": 1.000975251197815,
      "learning_rate": 0.00043131301866067566,
      "loss": 2.1875,
      "step": 81975
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.9147614240646362,
      "learning_rate": 0.00043130934075784344,
      "loss": 2.1531,
      "step": 81976
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.7664900422096252,
      "learning_rate": 0.0004313056628305983,
      "loss": 2.184,
      "step": 81977
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.8337723612785339,
      "learning_rate": 0.0004313019848789409,
      "loss": 2.1812,
      "step": 81978
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.8838091492652893,
      "learning_rate": 0.0004312983069028721,
      "loss": 2.3271,
      "step": 81979
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.8411281108856201,
      "learning_rate": 0.00043129462890239233,
      "loss": 2.1739,
      "step": 81980
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.7979980111122131,
      "learning_rate": 0.0004312909508775024,
      "loss": 2.2875,
      "step": 81981
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.7700985074043274,
      "learning_rate": 0.0004312872728282031,
      "loss": 2.1616,
      "step": 81982
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.8017523884773254,
      "learning_rate": 0.000431283594754495,
      "loss": 1.8714,
      "step": 81983
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.7909055352210999,
      "learning_rate": 0.0004312799166563788,
      "loss": 2.1374,
      "step": 81984
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.7241503000259399,
      "learning_rate": 0.0004312762385338552,
      "loss": 2.3235,
      "step": 81985
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.7878481149673462,
      "learning_rate": 0.0004312725603869248,
      "loss": 2.2838,
      "step": 81986
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.7962309122085571,
      "learning_rate": 0.0004312688822155885,
      "loss": 2.166,
      "step": 81987
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.7816600203514099,
      "learning_rate": 0.00043126520401984666,
      "loss": 2.4204,
      "step": 81988
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.7583307027816772,
      "learning_rate": 0.0004312615257997003,
      "loss": 2.0979,
      "step": 81989
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.7874863147735596,
      "learning_rate": 0.00043125784755514986,
      "loss": 2.0483,
      "step": 81990
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.8416613936424255,
      "learning_rate": 0.0004312541692861961,
      "loss": 2.2295,
      "step": 81991
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.9146131277084351,
      "learning_rate": 0.0004312504909928397,
      "loss": 2.1981,
      "step": 81992
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.7842868566513062,
      "learning_rate": 0.00043124681267508135,
      "loss": 2.1093,
      "step": 81993
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.8703349232673645,
      "learning_rate": 0.00043124313433292184,
      "loss": 2.1589,
      "step": 81994
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.7832155823707581,
      "learning_rate": 0.0004312394559663617,
      "loss": 2.1179,
      "step": 81995
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.7542598247528076,
      "learning_rate": 0.00043123577757540166,
      "loss": 2.2592,
      "step": 81996
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.8787348866462708,
      "learning_rate": 0.0004312320991600424,
      "loss": 2.1803,
      "step": 81997
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.7981440424919128,
      "learning_rate": 0.00043122842072028465,
      "loss": 2.0892,
      "step": 81998
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.7717095017433167,
      "learning_rate": 0.00043122474225612903,
      "loss": 2.1044,
      "step": 81999
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.885787844657898,
      "learning_rate": 0.00043122106376757626,
      "loss": 2.1641,
      "step": 82000
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.78452467918396,
      "learning_rate": 0.000431217385254627,
      "loss": 2.1983,
      "step": 82001
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.7856943607330322,
      "learning_rate": 0.0004312137067172819,
      "loss": 2.0625,
      "step": 82002
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.7663036584854126,
      "learning_rate": 0.0004312100281555419,
      "loss": 2.344,
      "step": 82003
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.9408926963806152,
      "learning_rate": 0.0004312063495694073,
      "loss": 2.3059,
      "step": 82004
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.8167259693145752,
      "learning_rate": 0.00043120267095887895,
      "loss": 2.115,
      "step": 82005
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.9071518778800964,
      "learning_rate": 0.00043119899232395764,
      "loss": 2.1876,
      "step": 82006
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.8013420701026917,
      "learning_rate": 0.00043119531366464395,
      "loss": 2.0885,
      "step": 82007
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.7957690358161926,
      "learning_rate": 0.0004311916349809385,
      "loss": 2.2016,
      "step": 82008
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.9815234541893005,
      "learning_rate": 0.00043118795627284214,
      "loss": 2.2304,
      "step": 82009
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.8688364028930664,
      "learning_rate": 0.0004311842775403554,
      "loss": 2.2981,
      "step": 82010
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.829928457736969,
      "learning_rate": 0.000431180598783479,
      "loss": 2.1949,
      "step": 82011
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.8685488700866699,
      "learning_rate": 0.00043117692000221375,
      "loss": 2.1706,
      "step": 82012
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.9921401143074036,
      "learning_rate": 0.0004311732411965602,
      "loss": 2.3729,
      "step": 82013
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.7971184849739075,
      "learning_rate": 0.000431169562366519,
      "loss": 2.3573,
      "step": 82014
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.9079567193984985,
      "learning_rate": 0.00043116588351209097,
      "loss": 2.2523,
      "step": 82015
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.8921775221824646,
      "learning_rate": 0.0004311622046332767,
      "loss": 2.1708,
      "step": 82016
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.8128722906112671,
      "learning_rate": 0.0004311585257300769,
      "loss": 2.199,
      "step": 82017
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.9452365040779114,
      "learning_rate": 0.0004311548468024923,
      "loss": 2.2839,
      "step": 82018
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.8139218688011169,
      "learning_rate": 0.00043115116785052347,
      "loss": 2.2526,
      "step": 82019
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.8656811118125916,
      "learning_rate": 0.0004311474888741712,
      "loss": 2.4141,
      "step": 82020
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.7883248329162598,
      "learning_rate": 0.0004311438098734361,
      "loss": 2.2411,
      "step": 82021
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.7500154972076416,
      "learning_rate": 0.0004311401308483189,
      "loss": 2.1711,
      "step": 82022
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.7834030389785767,
      "learning_rate": 0.0004311364517988203,
      "loss": 2.1309,
      "step": 82023
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.8225591778755188,
      "learning_rate": 0.000431132772724941,
      "loss": 2.4174,
      "step": 82024
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.7473740577697754,
      "learning_rate": 0.0004311290936266816,
      "loss": 2.3241,
      "step": 82025
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.9216569066047668,
      "learning_rate": 0.0004311254145040428,
      "loss": 2.1629,
      "step": 82026
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.7822884917259216,
      "learning_rate": 0.00043112173535702535,
      "loss": 2.2048,
      "step": 82027
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.8373267650604248,
      "learning_rate": 0.00043111805618562985,
      "loss": 2.1904,
      "step": 82028
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.7957550883293152,
      "learning_rate": 0.000431114376989857,
      "loss": 2.1189,
      "step": 82029
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.8774320483207703,
      "learning_rate": 0.0004311106977697076,
      "loss": 2.0341,
      "step": 82030
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.7766706943511963,
      "learning_rate": 0.0004311070185251822,
      "loss": 2.2156,
      "step": 82031
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.8832637071609497,
      "learning_rate": 0.00043110333925628155,
      "loss": 2.1948,
      "step": 82032
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.9530375599861145,
      "learning_rate": 0.0004310996599630063,
      "loss": 2.2538,
      "step": 82033
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.8474305272102356,
      "learning_rate": 0.0004310959806453572,
      "loss": 2.3612,
      "step": 82034
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.8151518106460571,
      "learning_rate": 0.0004310923013033348,
      "loss": 2.1317,
      "step": 82035
    },
    {
      "epoch": 1.07,
      "grad_norm": 1.0790058374404907,
      "learning_rate": 0.0004310886219369399,
      "loss": 2.1356,
      "step": 82036
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.9231900572776794,
      "learning_rate": 0.0004310849425461732,
      "loss": 2.2109,
      "step": 82037
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.9329549074172974,
      "learning_rate": 0.00043108126313103523,
      "loss": 2.3451,
      "step": 82038
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.8811927437782288,
      "learning_rate": 0.0004310775836915268,
      "loss": 2.4467,
      "step": 82039
    },
    {
      "epoch": 1.07,
      "grad_norm": 1.1830326318740845,
      "learning_rate": 0.0004310739042276486,
      "loss": 1.9785,
      "step": 82040
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.8174677491188049,
      "learning_rate": 0.00043107022473940133,
      "loss": 2.4236,
      "step": 82041
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.8496198058128357,
      "learning_rate": 0.0004310665452267856,
      "loss": 2.2217,
      "step": 82042
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.9346959590911865,
      "learning_rate": 0.0004310628656898022,
      "loss": 2.2108,
      "step": 82043
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.8213731050491333,
      "learning_rate": 0.00043105918612845165,
      "loss": 2.0313,
      "step": 82044
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.7277158498764038,
      "learning_rate": 0.00043105550654273464,
      "loss": 2.1622,
      "step": 82045
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.8302536010742188,
      "learning_rate": 0.0004310518269326521,
      "loss": 1.9356,
      "step": 82046
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.9094823598861694,
      "learning_rate": 0.00043104814729820446,
      "loss": 2.5334,
      "step": 82047
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.8328560590744019,
      "learning_rate": 0.0004310444676393925,
      "loss": 1.938,
      "step": 82048
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.7469435930252075,
      "learning_rate": 0.000431040787956217,
      "loss": 2.2785,
      "step": 82049
    },
    {
      "epoch": 1.07,
      "grad_norm": 1.0008519887924194,
      "learning_rate": 0.0004310371082486784,
      "loss": 2.0303,
      "step": 82050
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.791089653968811,
      "learning_rate": 0.0004310334285167776,
      "loss": 2.0455,
      "step": 82051
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.8492722511291504,
      "learning_rate": 0.0004310297487605152,
      "loss": 2.4469,
      "step": 82052
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.881233811378479,
      "learning_rate": 0.0004310260689798919,
      "loss": 2.0945,
      "step": 82053
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.8128768801689148,
      "learning_rate": 0.00043102238917490837,
      "loss": 2.0063,
      "step": 82054
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.7002429366111755,
      "learning_rate": 0.00043101870934556533,
      "loss": 2.036,
      "step": 82055
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.8897330164909363,
      "learning_rate": 0.00043101502949186347,
      "loss": 1.9318,
      "step": 82056
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.7888716459274292,
      "learning_rate": 0.00043101134961380337,
      "loss": 2.1868,
      "step": 82057
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.8053554892539978,
      "learning_rate": 0.0004310076697113858,
      "loss": 2.0784,
      "step": 82058
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.8303235173225403,
      "learning_rate": 0.00043100398978461143,
      "loss": 2.2118,
      "step": 82059
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.7437388896942139,
      "learning_rate": 0.00043100030983348093,
      "loss": 2.1387,
      "step": 82060
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.7537796497344971,
      "learning_rate": 0.00043099662985799514,
      "loss": 2.151,
      "step": 82061
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.7508264183998108,
      "learning_rate": 0.0004309929498581544,
      "loss": 2.3039,
      "step": 82062
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.83004230260849,
      "learning_rate": 0.0004309892698339597,
      "loss": 2.3282,
      "step": 82063
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.7730678915977478,
      "learning_rate": 0.0004309855897854117,
      "loss": 2.1218,
      "step": 82064
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.793882429599762,
      "learning_rate": 0.0004309819097125109,
      "loss": 2.04,
      "step": 82065
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.7777642607688904,
      "learning_rate": 0.00043097822961525806,
      "loss": 2.2318,
      "step": 82066
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.8642932176589966,
      "learning_rate": 0.0004309745494936539,
      "loss": 2.3268,
      "step": 82067
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.8358311653137207,
      "learning_rate": 0.0004309708693476992,
      "loss": 2.1031,
      "step": 82068
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.8168779611587524,
      "learning_rate": 0.00043096718917739443,
      "loss": 2.211,
      "step": 82069
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.8450178503990173,
      "learning_rate": 0.0004309635089827404,
      "loss": 2.4235,
      "step": 82070
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.8474155068397522,
      "learning_rate": 0.00043095982876373794,
      "loss": 2.0103,
      "step": 82071
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.8425430059432983,
      "learning_rate": 0.0004309561485203874,
      "loss": 2.0408,
      "step": 82072
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.7887186408042908,
      "learning_rate": 0.0004309524682526897,
      "loss": 2.2005,
      "step": 82073
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.7985913753509521,
      "learning_rate": 0.0004309487879606455,
      "loss": 2.2409,
      "step": 82074
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.8027364015579224,
      "learning_rate": 0.00043094510764425545,
      "loss": 2.3399,
      "step": 82075
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.8067412376403809,
      "learning_rate": 0.00043094142730352016,
      "loss": 2.2801,
      "step": 82076
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.8155390024185181,
      "learning_rate": 0.0004309377469384404,
      "loss": 2.1393,
      "step": 82077
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.7462285757064819,
      "learning_rate": 0.00043093406654901684,
      "loss": 2.126,
      "step": 82078
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.8400673270225525,
      "learning_rate": 0.00043093038613525023,
      "loss": 2.1136,
      "step": 82079
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.9525688290596008,
      "learning_rate": 0.00043092670569714117,
      "loss": 1.8988,
      "step": 82080
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.8795872926712036,
      "learning_rate": 0.0004309230252346903,
      "loss": 2.2842,
      "step": 82081
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.8155631422996521,
      "learning_rate": 0.0004309193447478984,
      "loss": 2.3691,
      "step": 82082
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.9038892984390259,
      "learning_rate": 0.00043091566423676615,
      "loss": 2.194,
      "step": 82083
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.8219456076622009,
      "learning_rate": 0.00043091198370129416,
      "loss": 2.3233,
      "step": 82084
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.9408005475997925,
      "learning_rate": 0.00043090830314148324,
      "loss": 2.2383,
      "step": 82085
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.8304256200790405,
      "learning_rate": 0.000430904622557334,
      "loss": 2.4798,
      "step": 82086
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.7666450142860413,
      "learning_rate": 0.000430900941948847,
      "loss": 2.0927,
      "step": 82087
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.8094200491905212,
      "learning_rate": 0.00043089726131602315,
      "loss": 2.1543,
      "step": 82088
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.864582359790802,
      "learning_rate": 0.000430893580658863,
      "loss": 2.1021,
      "step": 82089
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.8011994361877441,
      "learning_rate": 0.0004308898999773672,
      "loss": 2.0828,
      "step": 82090
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.8892129063606262,
      "learning_rate": 0.00043088621927153665,
      "loss": 2.196,
      "step": 82091
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.8588801026344299,
      "learning_rate": 0.0004308825385413717,
      "loss": 2.1931,
      "step": 82092
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.8045995831489563,
      "learning_rate": 0.0004308788577868733,
      "loss": 2.3415,
      "step": 82093
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.8483023643493652,
      "learning_rate": 0.0004308751770080421,
      "loss": 2.2009,
      "step": 82094
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.7847486138343811,
      "learning_rate": 0.00043087149620487866,
      "loss": 2.0279,
      "step": 82095
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.9685180187225342,
      "learning_rate": 0.00043086781537738376,
      "loss": 2.0759,
      "step": 82096
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.7466621398925781,
      "learning_rate": 0.00043086413452555805,
      "loss": 1.9316,
      "step": 82097
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.7730399966239929,
      "learning_rate": 0.00043086045364940226,
      "loss": 2.2127,
      "step": 82098
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.8942067623138428,
      "learning_rate": 0.000430856772748917,
      "loss": 2.0791,
      "step": 82099
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.9395214915275574,
      "learning_rate": 0.00043085309182410304,
      "loss": 2.4624,
      "step": 82100
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.7541363835334778,
      "learning_rate": 0.000430849410874961,
      "loss": 1.954,
      "step": 82101
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.9108354449272156,
      "learning_rate": 0.00043084572990149153,
      "loss": 2.2454,
      "step": 82102
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.8266080021858215,
      "learning_rate": 0.00043084204890369546,
      "loss": 2.2354,
      "step": 82103
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.7429473400115967,
      "learning_rate": 0.0004308383678815733,
      "loss": 2.1645,
      "step": 82104
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.7746750712394714,
      "learning_rate": 0.0004308346868351259,
      "loss": 2.2241,
      "step": 82105
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.8387266993522644,
      "learning_rate": 0.0004308310057643538,
      "loss": 1.9244,
      "step": 82106
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.8247583508491516,
      "learning_rate": 0.0004308273246692578,
      "loss": 2.1369,
      "step": 82107
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.7375944256782532,
      "learning_rate": 0.00043082364354983847,
      "loss": 2.3002,
      "step": 82108
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.8628769516944885,
      "learning_rate": 0.00043081996240609664,
      "loss": 2.2561,
      "step": 82109
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.773287296295166,
      "learning_rate": 0.00043081628123803285,
      "loss": 2.0729,
      "step": 82110
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.830381453037262,
      "learning_rate": 0.00043081260004564786,
      "loss": 2.38,
      "step": 82111
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.8472854495048523,
      "learning_rate": 0.00043080891882894227,
      "loss": 2.2936,
      "step": 82112
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.7959766387939453,
      "learning_rate": 0.00043080523758791693,
      "loss": 2.0279,
      "step": 82113
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.7904298305511475,
      "learning_rate": 0.00043080155632257234,
      "loss": 2.2004,
      "step": 82114
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.8289656639099121,
      "learning_rate": 0.0004307978750329093,
      "loss": 2.1551,
      "step": 82115
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.9958509802818298,
      "learning_rate": 0.00043079419371892856,
      "loss": 2.2219,
      "step": 82116
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.7692705392837524,
      "learning_rate": 0.0004307905123806306,
      "loss": 2.1721,
      "step": 82117
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.9267433881759644,
      "learning_rate": 0.00043078683101801623,
      "loss": 1.928,
      "step": 82118
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.9297782182693481,
      "learning_rate": 0.0004307831496310862,
      "loss": 2.4041,
      "step": 82119
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.8531944155693054,
      "learning_rate": 0.00043077946821984105,
      "loss": 2.4312,
      "step": 82120
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.8290066719055176,
      "learning_rate": 0.0004307757867842815,
      "loss": 2.0889,
      "step": 82121
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.792762815952301,
      "learning_rate": 0.00043077210532440827,
      "loss": 2.0911,
      "step": 82122
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.8191179633140564,
      "learning_rate": 0.00043076842384022214,
      "loss": 2.2763,
      "step": 82123
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.8179441094398499,
      "learning_rate": 0.00043076474233172353,
      "loss": 2.1602,
      "step": 82124
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.7862975001335144,
      "learning_rate": 0.0004307610607989135,
      "loss": 2.1857,
      "step": 82125
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.8759863376617432,
      "learning_rate": 0.0004307573792417923,
      "loss": 2.3803,
      "step": 82126
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.7341605424880981,
      "learning_rate": 0.00043075369766036093,
      "loss": 2.2373,
      "step": 82127
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.7965906262397766,
      "learning_rate": 0.00043075001605461996,
      "loss": 2.1628,
      "step": 82128
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.862924337387085,
      "learning_rate": 0.00043074633442457015,
      "loss": 2.0855,
      "step": 82129
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.9647730588912964,
      "learning_rate": 0.00043074265277021203,
      "loss": 2.0329,
      "step": 82130
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.8070489168167114,
      "learning_rate": 0.00043073897109154654,
      "loss": 1.9875,
      "step": 82131
    },
    {
      "epoch": 1.07,
      "grad_norm": 1.0032315254211426,
      "learning_rate": 0.00043073528938857405,
      "loss": 2.2576,
      "step": 82132
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.879409909248352,
      "learning_rate": 0.0004307316076612954,
      "loss": 2.2729,
      "step": 82133
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.8041562438011169,
      "learning_rate": 0.0004307279259097114,
      "loss": 2.0791,
      "step": 82134
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.8160299062728882,
      "learning_rate": 0.00043072424413382254,
      "loss": 2.1548,
      "step": 82135
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.8178431987762451,
      "learning_rate": 0.00043072056233362953,
      "loss": 2.1838,
      "step": 82136
    },
    {
      "epoch": 1.07,
      "grad_norm": 1.015185832977295,
      "learning_rate": 0.0004307168805091332,
      "loss": 2.2793,
      "step": 82137
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.857120931148529,
      "learning_rate": 0.0004307131986603341,
      "loss": 2.0041,
      "step": 82138
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.9163190126419067,
      "learning_rate": 0.0004307095167872329,
      "loss": 2.1178,
      "step": 82139
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.8260666131973267,
      "learning_rate": 0.00043070583488983037,
      "loss": 2.2499,
      "step": 82140
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.7646621465682983,
      "learning_rate": 0.0004307021529681272,
      "loss": 2.215,
      "step": 82141
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.7769560217857361,
      "learning_rate": 0.0004306984710221239,
      "loss": 2.1757,
      "step": 82142
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.7662988901138306,
      "learning_rate": 0.00043069478905182146,
      "loss": 2.113,
      "step": 82143
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.827067494392395,
      "learning_rate": 0.0004306911070572203,
      "loss": 2.0098,
      "step": 82144
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.8402746915817261,
      "learning_rate": 0.00043068742503832124,
      "loss": 2.3388,
      "step": 82145
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.8330632448196411,
      "learning_rate": 0.00043068374299512485,
      "loss": 2.2178,
      "step": 82146
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.761480450630188,
      "learning_rate": 0.0004306800609276319,
      "loss": 2.0377,
      "step": 82147
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.7703100442886353,
      "learning_rate": 0.0004306763788358431,
      "loss": 2.3207,
      "step": 82148
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.7751544117927551,
      "learning_rate": 0.00043067269671975914,
      "loss": 2.04,
      "step": 82149
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.9529295563697815,
      "learning_rate": 0.0004306690145793805,
      "loss": 2.2089,
      "step": 82150
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.7936208248138428,
      "learning_rate": 0.00043066533241470817,
      "loss": 2.1961,
      "step": 82151
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.9503157734870911,
      "learning_rate": 0.0004306616502257426,
      "loss": 2.2749,
      "step": 82152
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.8488442897796631,
      "learning_rate": 0.00043065796801248466,
      "loss": 2.4047,
      "step": 82153
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.8113105297088623,
      "learning_rate": 0.00043065428577493484,
      "loss": 1.9438,
      "step": 82154
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.76326984167099,
      "learning_rate": 0.0004306506035130939,
      "loss": 2.113,
      "step": 82155
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.7936210036277771,
      "learning_rate": 0.00043064692122696263,
      "loss": 2.0384,
      "step": 82156
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.8408489227294922,
      "learning_rate": 0.0004306432389165416,
      "loss": 2.166,
      "step": 82157
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.8925126194953918,
      "learning_rate": 0.00043063955658183153,
      "loss": 2.214,
      "step": 82158
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.8301677107810974,
      "learning_rate": 0.0004306358742228331,
      "loss": 2.2826,
      "step": 82159
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.8242809176445007,
      "learning_rate": 0.00043063219183954703,
      "loss": 2.1212,
      "step": 82160
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.7516636848449707,
      "learning_rate": 0.0004306285094319739,
      "loss": 2.1175,
      "step": 82161
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.8659351468086243,
      "learning_rate": 0.00043062482700011444,
      "loss": 2.1154,
      "step": 82162
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.8661572933197021,
      "learning_rate": 0.0004306211445439695,
      "loss": 2.2059,
      "step": 82163
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.8336755037307739,
      "learning_rate": 0.0004306174620635395,
      "loss": 2.1966,
      "step": 82164
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.8535818457603455,
      "learning_rate": 0.00043061377955882525,
      "loss": 2.1204,
      "step": 82165
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.9487610459327698,
      "learning_rate": 0.0004306100970298275,
      "loss": 2.3825,
      "step": 82166
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.8396641612052917,
      "learning_rate": 0.0004306064144765469,
      "loss": 2.1845,
      "step": 82167
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.8382517099380493,
      "learning_rate": 0.000430602731898984,
      "loss": 2.3776,
      "step": 82168
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.7688278555870056,
      "learning_rate": 0.0004305990492971396,
      "loss": 2.2334,
      "step": 82169
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.7707552909851074,
      "learning_rate": 0.00043059536667101444,
      "loss": 2.0943,
      "step": 82170
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.8037981390953064,
      "learning_rate": 0.0004305916840206091,
      "loss": 2.0452,
      "step": 82171
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.8167664408683777,
      "learning_rate": 0.00043058800134592427,
      "loss": 2.3269,
      "step": 82172
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.8406292796134949,
      "learning_rate": 0.0004305843186469607,
      "loss": 2.27,
      "step": 82173
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.8064607977867126,
      "learning_rate": 0.0004305806359237191,
      "loss": 2.1679,
      "step": 82174
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.8214038014411926,
      "learning_rate": 0.0004305769531762,
      "loss": 2.336,
      "step": 82175
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.797950267791748,
      "learning_rate": 0.0004305732704044041,
      "loss": 2.4071,
      "step": 82176
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.8651111721992493,
      "learning_rate": 0.00043056958760833233,
      "loss": 2.3716,
      "step": 82177
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.7540112137794495,
      "learning_rate": 0.0004305659047879852,
      "loss": 2.3695,
      "step": 82178
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.8539147973060608,
      "learning_rate": 0.00043056222194336327,
      "loss": 2.0007,
      "step": 82179
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.863172173500061,
      "learning_rate": 0.00043055853907446757,
      "loss": 2.1953,
      "step": 82180
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.8523826003074646,
      "learning_rate": 0.00043055485618129837,
      "loss": 2.3271,
      "step": 82181
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.9288581609725952,
      "learning_rate": 0.00043055117326385665,
      "loss": 2.1356,
      "step": 82182
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.9433286786079407,
      "learning_rate": 0.00043054749032214306,
      "loss": 2.0019,
      "step": 82183
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.7543568015098572,
      "learning_rate": 0.00043054380735615803,
      "loss": 2.087,
      "step": 82184
    },
    {
      "epoch": 1.07,
      "grad_norm": 1.5737420320510864,
      "learning_rate": 0.0004305401243659026,
      "loss": 2.0266,
      "step": 82185
    },
    {
      "epoch": 1.07,
      "grad_norm": 1.122917890548706,
      "learning_rate": 0.00043053644135137735,
      "loss": 1.8384,
      "step": 82186
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.8642724752426147,
      "learning_rate": 0.0004305327583125828,
      "loss": 2.0667,
      "step": 82187
    },
    {
      "epoch": 1.07,
      "grad_norm": 1.7246330976486206,
      "learning_rate": 0.0004305290752495197,
      "loss": 2.3879,
      "step": 82188
    },
    {
      "epoch": 1.07,
      "grad_norm": 1.059462547302246,
      "learning_rate": 0.00043052539216218886,
      "loss": 2.1443,
      "step": 82189
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.8479757905006409,
      "learning_rate": 0.00043052170905059087,
      "loss": 2.4225,
      "step": 82190
    },
    {
      "epoch": 1.07,
      "grad_norm": 1.0531628131866455,
      "learning_rate": 0.0004305180259147265,
      "loss": 2.2333,
      "step": 82191
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.9416154623031616,
      "learning_rate": 0.0004305143427545963,
      "loss": 2.1547,
      "step": 82192
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.9888817667961121,
      "learning_rate": 0.000430510659570201,
      "loss": 2.1514,
      "step": 82193
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.756338894367218,
      "learning_rate": 0.0004305069763615413,
      "loss": 2.4086,
      "step": 82194
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.806655764579773,
      "learning_rate": 0.000430503293128618,
      "loss": 2.4514,
      "step": 82195
    },
    {
      "epoch": 1.07,
      "grad_norm": 1.1371822357177734,
      "learning_rate": 0.00043049960987143155,
      "loss": 2.4399,
      "step": 82196
    },
    {
      "epoch": 1.07,
      "grad_norm": 1.2292046546936035,
      "learning_rate": 0.0004304959265899828,
      "loss": 2.2409,
      "step": 82197
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.8326918482780457,
      "learning_rate": 0.0004304922432842724,
      "loss": 2.0156,
      "step": 82198
    },
    {
      "epoch": 1.07,
      "grad_norm": 1.0601643323898315,
      "learning_rate": 0.000430488559954301,
      "loss": 2.0881,
      "step": 82199
    },
    {
      "epoch": 1.07,
      "grad_norm": 1.24580717086792,
      "learning_rate": 0.00043048487660006935,
      "loss": 2.1088,
      "step": 82200
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.8261325359344482,
      "learning_rate": 0.00043048119322157814,
      "loss": 2.2799,
      "step": 82201
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.9546729922294617,
      "learning_rate": 0.0004304775098188279,
      "loss": 2.2686,
      "step": 82202
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.9634767770767212,
      "learning_rate": 0.0004304738263918195,
      "loss": 2.3614,
      "step": 82203
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.9986178278923035,
      "learning_rate": 0.00043047014294055356,
      "loss": 2.0636,
      "step": 82204
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.7947003245353699,
      "learning_rate": 0.0004304664594650307,
      "loss": 1.9017,
      "step": 82205
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.8323647975921631,
      "learning_rate": 0.00043046277596525174,
      "loss": 2.3257,
      "step": 82206
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.9511880278587341,
      "learning_rate": 0.00043045909244121723,
      "loss": 2.1658,
      "step": 82207
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.9248831272125244,
      "learning_rate": 0.0004304554088929279,
      "loss": 2.3343,
      "step": 82208
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.9743853211402893,
      "learning_rate": 0.00043045172532038447,
      "loss": 2.1829,
      "step": 82209
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.8967744708061218,
      "learning_rate": 0.0004304480417235877,
      "loss": 2.3123,
      "step": 82210
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.9398981928825378,
      "learning_rate": 0.000430444358102538,
      "loss": 2.2441,
      "step": 82211
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.786555826663971,
      "learning_rate": 0.0004304406744572363,
      "loss": 1.9661,
      "step": 82212
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.9513705968856812,
      "learning_rate": 0.0004304369907876833,
      "loss": 2.2516,
      "step": 82213
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.7690827250480652,
      "learning_rate": 0.00043043330709387955,
      "loss": 2.2253,
      "step": 82214
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.7563045620918274,
      "learning_rate": 0.00043042962337582575,
      "loss": 2.2263,
      "step": 82215
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.7312842607498169,
      "learning_rate": 0.0004304259396335227,
      "loss": 1.9811,
      "step": 82216
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.815243124961853,
      "learning_rate": 0.00043042225586697086,
      "loss": 2.264,
      "step": 82217
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.9098302721977234,
      "learning_rate": 0.00043041857207617117,
      "loss": 2.2644,
      "step": 82218
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.7138708829879761,
      "learning_rate": 0.0004304148882611242,
      "loss": 2.3505,
      "step": 82219
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.7461739778518677,
      "learning_rate": 0.0004304112044218306,
      "loss": 2.2436,
      "step": 82220
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.8099194169044495,
      "learning_rate": 0.0004304075205582911,
      "loss": 2.1502,
      "step": 82221
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.7685617208480835,
      "learning_rate": 0.00043040383667050647,
      "loss": 2.1263,
      "step": 82222
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.8029200434684753,
      "learning_rate": 0.0004304001527584772,
      "loss": 2.2625,
      "step": 82223
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.8514727354049683,
      "learning_rate": 0.0004303964688222041,
      "loss": 2.0572,
      "step": 82224
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.7861000299453735,
      "learning_rate": 0.00043039278486168794,
      "loss": 2.1173,
      "step": 82225
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.8376467227935791,
      "learning_rate": 0.00043038910087692914,
      "loss": 2.1169,
      "step": 82226
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.874576210975647,
      "learning_rate": 0.0004303854168679286,
      "loss": 2.2055,
      "step": 82227
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.8330773711204529,
      "learning_rate": 0.000430381732834687,
      "loss": 2.047,
      "step": 82228
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.9111883640289307,
      "learning_rate": 0.0004303780487772049,
      "loss": 2.072,
      "step": 82229
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.7329803705215454,
      "learning_rate": 0.0004303743646954831,
      "loss": 2.346,
      "step": 82230
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.8480054140090942,
      "learning_rate": 0.00043037068058952225,
      "loss": 2.3834,
      "step": 82231
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.8511716723442078,
      "learning_rate": 0.00043036699645932305,
      "loss": 2.1308,
      "step": 82232
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.8212804198265076,
      "learning_rate": 0.0004303633123048861,
      "loss": 2.1974,
      "step": 82233
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.9063646197319031,
      "learning_rate": 0.00043035962812621223,
      "loss": 2.2116,
      "step": 82234
    },
    {
      "epoch": 1.07,
      "grad_norm": 1.0133167505264282,
      "learning_rate": 0.00043035594392330204,
      "loss": 2.1248,
      "step": 82235
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.7609572410583496,
      "learning_rate": 0.00043035225969615613,
      "loss": 2.3125,
      "step": 82236
    },
    {
      "epoch": 1.07,
      "grad_norm": 1.0127843618392944,
      "learning_rate": 0.0004303485754447753,
      "loss": 2.1717,
      "step": 82237
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.7468889951705933,
      "learning_rate": 0.0004303448911691603,
      "loss": 2.3118,
      "step": 82238
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.8202242851257324,
      "learning_rate": 0.0004303412068693116,
      "loss": 2.2676,
      "step": 82239
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.7903986573219299,
      "learning_rate": 0.00043033752254523005,
      "loss": 2.2122,
      "step": 82240
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.7824590802192688,
      "learning_rate": 0.00043033383819691636,
      "loss": 1.915,
      "step": 82241
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.7492297291755676,
      "learning_rate": 0.0004303301538243711,
      "loss": 2.3131,
      "step": 82242
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.9526888728141785,
      "learning_rate": 0.00043032646942759497,
      "loss": 2.2258,
      "step": 82243
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.8230302929878235,
      "learning_rate": 0.0004303227850065887,
      "loss": 2.2943,
      "step": 82244
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.9533873200416565,
      "learning_rate": 0.00043031910056135305,
      "loss": 2.3864,
      "step": 82245
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.87875896692276,
      "learning_rate": 0.0004303154160918885,
      "loss": 2.2921,
      "step": 82246
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.7894995212554932,
      "learning_rate": 0.0004303117315981959,
      "loss": 2.062,
      "step": 82247
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.8125870227813721,
      "learning_rate": 0.000430308047080276,
      "loss": 2.4432,
      "step": 82248
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.8923638463020325,
      "learning_rate": 0.0004303043625381292,
      "loss": 2.1531,
      "step": 82249
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.8247655034065247,
      "learning_rate": 0.0004303006779717565,
      "loss": 2.1272,
      "step": 82250
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.8387738466262817,
      "learning_rate": 0.0004302969933811583,
      "loss": 2.2281,
      "step": 82251
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.7627016305923462,
      "learning_rate": 0.0004302933087663355,
      "loss": 1.933,
      "step": 82252
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.9120570421218872,
      "learning_rate": 0.00043028962412728874,
      "loss": 2.3069,
      "step": 82253
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.8415378928184509,
      "learning_rate": 0.0004302859394640186,
      "loss": 2.2003,
      "step": 82254
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.8248581886291504,
      "learning_rate": 0.00043028225477652603,
      "loss": 1.9829,
      "step": 82255
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.9020301699638367,
      "learning_rate": 0.00043027857006481136,
      "loss": 2.1693,
      "step": 82256
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.7589712142944336,
      "learning_rate": 0.0004302748853288755,
      "loss": 2.2273,
      "step": 82257
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.790973961353302,
      "learning_rate": 0.00043027120056871905,
      "loss": 2.208,
      "step": 82258
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.7287124395370483,
      "learning_rate": 0.00043026751578434277,
      "loss": 1.9274,
      "step": 82259
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.9104611277580261,
      "learning_rate": 0.00043026383097574726,
      "loss": 2.2723,
      "step": 82260
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.9371051788330078,
      "learning_rate": 0.0004302601461429332,
      "loss": 2.0115,
      "step": 82261
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.8210272789001465,
      "learning_rate": 0.00043025646128590147,
      "loss": 2.0929,
      "step": 82262
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.9046743512153625,
      "learning_rate": 0.00043025277640465254,
      "loss": 2.2581,
      "step": 82263
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.8566259145736694,
      "learning_rate": 0.0004302490914991871,
      "loss": 2.2254,
      "step": 82264
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.764411211013794,
      "learning_rate": 0.000430245406569506,
      "loss": 1.9847,
      "step": 82265
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.8554683327674866,
      "learning_rate": 0.0004302417216156097,
      "loss": 2.1262,
      "step": 82266
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.8378786444664001,
      "learning_rate": 0.000430238036637499,
      "loss": 2.3577,
      "step": 82267
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.7621562480926514,
      "learning_rate": 0.0004302343516351747,
      "loss": 1.8951,
      "step": 82268
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.8288252949714661,
      "learning_rate": 0.0004302306666086373,
      "loss": 2.2886,
      "step": 82269
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.881405770778656,
      "learning_rate": 0.00043022698155788757,
      "loss": 2.2261,
      "step": 82270
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.947620153427124,
      "learning_rate": 0.00043022329648292624,
      "loss": 2.2505,
      "step": 82271
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.8711454272270203,
      "learning_rate": 0.00043021961138375387,
      "loss": 2.1288,
      "step": 82272
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.7542687654495239,
      "learning_rate": 0.00043021592626037126,
      "loss": 2.0373,
      "step": 82273
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.8347174525260925,
      "learning_rate": 0.000430212241112779,
      "loss": 2.571,
      "step": 82274
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.8804000020027161,
      "learning_rate": 0.00043020855594097794,
      "loss": 2.1152,
      "step": 82275
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.8246292471885681,
      "learning_rate": 0.0004302048707449685,
      "loss": 2.279,
      "step": 82276
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.800061821937561,
      "learning_rate": 0.00043020118552475166,
      "loss": 2.1415,
      "step": 82277
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.8101385235786438,
      "learning_rate": 0.0004301975002803279,
      "loss": 2.2603,
      "step": 82278
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.8236091732978821,
      "learning_rate": 0.0004301938150116979,
      "loss": 2.2474,
      "step": 82279
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.7517805695533752,
      "learning_rate": 0.00043019012971886255,
      "loss": 2.4249,
      "step": 82280
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.9227999448776245,
      "learning_rate": 0.00043018644440182224,
      "loss": 2.1265,
      "step": 82281
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.8078467845916748,
      "learning_rate": 0.0004301827590605779,
      "loss": 2.3098,
      "step": 82282
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.830207347869873,
      "learning_rate": 0.0004301790736951301,
      "loss": 2.2913,
      "step": 82283
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.8191973567008972,
      "learning_rate": 0.0004301753883054796,
      "loss": 2.4798,
      "step": 82284
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.7372347712516785,
      "learning_rate": 0.000430171702891627,
      "loss": 2.1032,
      "step": 82285
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.7476164102554321,
      "learning_rate": 0.0004301680174535731,
      "loss": 1.9149,
      "step": 82286
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.8050780892372131,
      "learning_rate": 0.0004301643319913184,
      "loss": 1.844,
      "step": 82287
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.8002994656562805,
      "learning_rate": 0.0004301606465048637,
      "loss": 2.1343,
      "step": 82288
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.8197303414344788,
      "learning_rate": 0.0004301569609942098,
      "loss": 2.1303,
      "step": 82289
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.8042318224906921,
      "learning_rate": 0.00043015327545935716,
      "loss": 2.3107,
      "step": 82290
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.7288880348205566,
      "learning_rate": 0.00043014958990030656,
      "loss": 2.0884,
      "step": 82291
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.8309482336044312,
      "learning_rate": 0.00043014590431705873,
      "loss": 2.22,
      "step": 82292
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.8605021834373474,
      "learning_rate": 0.00043014221870961433,
      "loss": 2.2184,
      "step": 82293
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.7808752059936523,
      "learning_rate": 0.000430138533077974,
      "loss": 2.1302,
      "step": 82294
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.8080902099609375,
      "learning_rate": 0.0004301348474221386,
      "loss": 2.2113,
      "step": 82295
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.7841665148735046,
      "learning_rate": 0.00043013116174210847,
      "loss": 2.2984,
      "step": 82296
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.7763017416000366,
      "learning_rate": 0.00043012747603788455,
      "loss": 2.2488,
      "step": 82297
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.8003681898117065,
      "learning_rate": 0.0004301237903094676,
      "loss": 2.4034,
      "step": 82298
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.7944095730781555,
      "learning_rate": 0.00043012010455685805,
      "loss": 2.3777,
      "step": 82299
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.7938770055770874,
      "learning_rate": 0.0004301164187800568,
      "loss": 2.3051,
      "step": 82300
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.8558602333068848,
      "learning_rate": 0.0004301127329790644,
      "loss": 2.1604,
      "step": 82301
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.840575635433197,
      "learning_rate": 0.0004301090471538816,
      "loss": 2.1497,
      "step": 82302
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.7837843298912048,
      "learning_rate": 0.00043010536130450906,
      "loss": 2.1907,
      "step": 82303
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.9379994869232178,
      "learning_rate": 0.00043010167543094745,
      "loss": 2.2878,
      "step": 82304
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.8469716906547546,
      "learning_rate": 0.0004300979895331975,
      "loss": 2.1751,
      "step": 82305
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.7523271441459656,
      "learning_rate": 0.0004300943036112599,
      "loss": 2.4646,
      "step": 82306
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.8811891674995422,
      "learning_rate": 0.00043009061766513534,
      "loss": 2.329,
      "step": 82307
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.8840299844741821,
      "learning_rate": 0.00043008693169482447,
      "loss": 2.1159,
      "step": 82308
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.813112199306488,
      "learning_rate": 0.0004300832457003279,
      "loss": 2.5222,
      "step": 82309
    },
    {
      "epoch": 1.07,
      "grad_norm": 1.2181605100631714,
      "learning_rate": 0.0004300795596816465,
      "loss": 1.9717,
      "step": 82310
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.7672202587127686,
      "learning_rate": 0.0004300758736387808,
      "loss": 2.1105,
      "step": 82311
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.8043501973152161,
      "learning_rate": 0.00043007218757173157,
      "loss": 2.1082,
      "step": 82312
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.724976122379303,
      "learning_rate": 0.00043006850148049946,
      "loss": 2.0574,
      "step": 82313
    },
    {
      "epoch": 1.07,
      "grad_norm": 1.2200027704238892,
      "learning_rate": 0.00043006481536508515,
      "loss": 1.9703,
      "step": 82314
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.7972844243049622,
      "learning_rate": 0.00043006112922548933,
      "loss": 2.0309,
      "step": 82315
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.8337573409080505,
      "learning_rate": 0.00043005744306171266,
      "loss": 2.1283,
      "step": 82316
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.7747241854667664,
      "learning_rate": 0.000430053756873756,
      "loss": 2.1851,
      "step": 82317
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.9067915081977844,
      "learning_rate": 0.0004300500706616197,
      "loss": 2.265,
      "step": 82318
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.7749327421188354,
      "learning_rate": 0.00043004638442530475,
      "loss": 2.1036,
      "step": 82319
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.8926702737808228,
      "learning_rate": 0.0004300426981648117,
      "loss": 2.2183,
      "step": 82320
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.9336862564086914,
      "learning_rate": 0.0004300390118801413,
      "loss": 2.3605,
      "step": 82321
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.9486904740333557,
      "learning_rate": 0.00043003532557129404,
      "loss": 2.387,
      "step": 82322
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.8906283974647522,
      "learning_rate": 0.000430031639238271,
      "loss": 2.376,
      "step": 82323
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.8423228859901428,
      "learning_rate": 0.00043002795288107237,
      "loss": 2.167,
      "step": 82324
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.8555125594139099,
      "learning_rate": 0.00043002426649969925,
      "loss": 2.2978,
      "step": 82325
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.9395094513893127,
      "learning_rate": 0.00043002058009415216,
      "loss": 2.4699,
      "step": 82326
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.8475031852722168,
      "learning_rate": 0.0004300168936644317,
      "loss": 2.1744,
      "step": 82327
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.7619027495384216,
      "learning_rate": 0.0004300132072105387,
      "loss": 2.0711,
      "step": 82328
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.977695107460022,
      "learning_rate": 0.0004300095207324738,
      "loss": 2.2212,
      "step": 82329
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.8778484463691711,
      "learning_rate": 0.0004300058342302377,
      "loss": 2.239,
      "step": 82330
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.7435745000839233,
      "learning_rate": 0.00043000214770383096,
      "loss": 2.1326,
      "step": 82331
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.7791659832000732,
      "learning_rate": 0.00042999846115325443,
      "loss": 2.2813,
      "step": 82332
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.8637591600418091,
      "learning_rate": 0.0004299947745785088,
      "loss": 2.1554,
      "step": 82333
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.9922448992729187,
      "learning_rate": 0.0004299910879795946,
      "loss": 2.4796,
      "step": 82334
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.772968590259552,
      "learning_rate": 0.0004299874013565126,
      "loss": 2.2636,
      "step": 82335
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.7418492436408997,
      "learning_rate": 0.00042998371470926355,
      "loss": 2.0445,
      "step": 82336
    },
    {
      "epoch": 1.07,
      "grad_norm": 1.1803690195083618,
      "learning_rate": 0.00042998002803784803,
      "loss": 2.3487,
      "step": 82337
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.8667837381362915,
      "learning_rate": 0.0004299763413422667,
      "loss": 2.2605,
      "step": 82338
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.822040319442749,
      "learning_rate": 0.00042997265462252045,
      "loss": 2.0357,
      "step": 82339
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.8045436143875122,
      "learning_rate": 0.00042996896787860975,
      "loss": 2.2807,
      "step": 82340
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.862057626247406,
      "learning_rate": 0.0004299652811105354,
      "loss": 2.2442,
      "step": 82341
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.8879518508911133,
      "learning_rate": 0.000429961594318298,
      "loss": 2.4542,
      "step": 82342
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.8246685266494751,
      "learning_rate": 0.0004299579075018984,
      "loss": 2.3823,
      "step": 82343
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.7031691074371338,
      "learning_rate": 0.00042995422066133705,
      "loss": 2.1614,
      "step": 82344
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.7928792238235474,
      "learning_rate": 0.0004299505337966147,
      "loss": 2.1526,
      "step": 82345
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.8585694432258606,
      "learning_rate": 0.00042994684690773226,
      "loss": 2.0275,
      "step": 82346
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.8260015845298767,
      "learning_rate": 0.0004299431599946902,
      "loss": 2.2698,
      "step": 82347
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.7666340470314026,
      "learning_rate": 0.00042993947305748924,
      "loss": 2.1121,
      "step": 82348
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.7417380213737488,
      "learning_rate": 0.0004299357860961301,
      "loss": 2.1052,
      "step": 82349
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.9183082580566406,
      "learning_rate": 0.00042993209911061344,
      "loss": 2.355,
      "step": 82350
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.8661207556724548,
      "learning_rate": 0.00042992841210093985,
      "loss": 2.0163,
      "step": 82351
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.790877640247345,
      "learning_rate": 0.00042992472506711023,
      "loss": 2.1307,
      "step": 82352
    },
    {
      "epoch": 1.07,
      "grad_norm": 1.034851312637329,
      "learning_rate": 0.00042992103800912514,
      "loss": 2.1793,
      "step": 82353
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.9129974842071533,
      "learning_rate": 0.0004299173509269852,
      "loss": 2.3857,
      "step": 82354
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.7658569812774658,
      "learning_rate": 0.00042991366382069126,
      "loss": 2.2106,
      "step": 82355
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.8604309558868408,
      "learning_rate": 0.0004299099766902439,
      "loss": 2.2484,
      "step": 82356
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.9832192659378052,
      "learning_rate": 0.0004299062895356438,
      "loss": 2.1295,
      "step": 82357
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.7741734385490417,
      "learning_rate": 0.0004299026023568917,
      "loss": 2.3531,
      "step": 82358
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.9965434074401855,
      "learning_rate": 0.00042989891515398824,
      "loss": 2.2113,
      "step": 82359
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.8884527683258057,
      "learning_rate": 0.00042989522792693407,
      "loss": 2.0023,
      "step": 82360
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.7714280486106873,
      "learning_rate": 0.00042989154067573,
      "loss": 2.1717,
      "step": 82361
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.9343857765197754,
      "learning_rate": 0.00042988785340037657,
      "loss": 2.0512,
      "step": 82362
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.8271569609642029,
      "learning_rate": 0.0004298841661008746,
      "loss": 2.212,
      "step": 82363
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.8621222972869873,
      "learning_rate": 0.0004298804787772246,
      "loss": 2.0882,
      "step": 82364
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.7652009129524231,
      "learning_rate": 0.00042987679142942757,
      "loss": 2.2977,
      "step": 82365
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.8010331988334656,
      "learning_rate": 0.0004298731040574838,
      "loss": 2.2322,
      "step": 82366
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.8336613178253174,
      "learning_rate": 0.0004298694166613943,
      "loss": 2.2732,
      "step": 82367
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.7730253338813782,
      "learning_rate": 0.0004298657292411595,
      "loss": 2.2358,
      "step": 82368
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.8581634759902954,
      "learning_rate": 0.0004298620417967804,
      "loss": 2.2695,
      "step": 82369
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.7712223529815674,
      "learning_rate": 0.0004298583543282573,
      "loss": 2.2133,
      "step": 82370
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.7728896737098694,
      "learning_rate": 0.00042985466683559116,
      "loss": 2.1318,
      "step": 82371
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.7558541893959045,
      "learning_rate": 0.0004298509793187826,
      "loss": 1.9134,
      "step": 82372
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.7746737599372864,
      "learning_rate": 0.0004298472917778323,
      "loss": 2.3424,
      "step": 82373
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.9131388068199158,
      "learning_rate": 0.00042984360421274083,
      "loss": 2.2409,
      "step": 82374
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.8455562591552734,
      "learning_rate": 0.0004298399166235091,
      "loss": 2.2649,
      "step": 82375
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.8471347093582153,
      "learning_rate": 0.00042983622901013763,
      "loss": 2.2418,
      "step": 82376
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.8252827525138855,
      "learning_rate": 0.0004298325413726271,
      "loss": 2.1217,
      "step": 82377
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.7676720023155212,
      "learning_rate": 0.00042982885371097837,
      "loss": 2.2027,
      "step": 82378
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.8573826551437378,
      "learning_rate": 0.00042982516602519197,
      "loss": 2.0603,
      "step": 82379
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.8478882908821106,
      "learning_rate": 0.0004298214783152685,
      "loss": 2.4214,
      "step": 82380
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.8177935481071472,
      "learning_rate": 0.00042981779058120893,
      "loss": 1.8345,
      "step": 82381
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.8046603202819824,
      "learning_rate": 0.0004298141028230137,
      "loss": 2.0879,
      "step": 82382
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.8723780512809753,
      "learning_rate": 0.0004298104150406836,
      "loss": 2.1471,
      "step": 82383
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.9359807968139648,
      "learning_rate": 0.0004298067272342193,
      "loss": 2.2449,
      "step": 82384
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.9937040209770203,
      "learning_rate": 0.00042980303940362146,
      "loss": 2.222,
      "step": 82385
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.8090493679046631,
      "learning_rate": 0.00042979935154889077,
      "loss": 2.0225,
      "step": 82386
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.7897789478302002,
      "learning_rate": 0.000429795663670028,
      "loss": 2.3293,
      "step": 82387
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.8187243938446045,
      "learning_rate": 0.0004297919757670337,
      "loss": 2.394,
      "step": 82388
    },
    {
      "epoch": 1.07,
      "grad_norm": 1.0343166589736938,
      "learning_rate": 0.0004297882878399086,
      "loss": 2.2447,
      "step": 82389
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.8375819325447083,
      "learning_rate": 0.0004297845998886535,
      "loss": 2.1818,
      "step": 82390
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.8314979672431946,
      "learning_rate": 0.00042978091191326896,
      "loss": 2.1297,
      "step": 82391
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.8661279082298279,
      "learning_rate": 0.00042977722391375565,
      "loss": 1.9255,
      "step": 82392
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.8551172614097595,
      "learning_rate": 0.00042977353589011434,
      "loss": 2.3683,
      "step": 82393
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.7863749265670776,
      "learning_rate": 0.00042976984784234563,
      "loss": 2.0774,
      "step": 82394
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.8214616775512695,
      "learning_rate": 0.0004297661597704503,
      "loss": 2.0057,
      "step": 82395
    },
    {
      "epoch": 1.07,
      "grad_norm": 1.0960655212402344,
      "learning_rate": 0.00042976247167442904,
      "loss": 2.2103,
      "step": 82396
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.8179953694343567,
      "learning_rate": 0.0004297587835542824,
      "loss": 2.1762,
      "step": 82397
    },
    {
      "epoch": 1.07,
      "grad_norm": 1.0121756792068481,
      "learning_rate": 0.00042975509541001125,
      "loss": 2.3219,
      "step": 82398
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.9952273368835449,
      "learning_rate": 0.00042975140724161616,
      "loss": 2.1561,
      "step": 82399
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.7912558317184448,
      "learning_rate": 0.0004297477190490977,
      "loss": 2.3678,
      "step": 82400
    },
    {
      "epoch": 1.07,
      "grad_norm": 1.0090916156768799,
      "learning_rate": 0.0004297440308324568,
      "loss": 2.0802,
      "step": 82401
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.9229733943939209,
      "learning_rate": 0.00042974034259169407,
      "loss": 2.3239,
      "step": 82402
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.7860550284385681,
      "learning_rate": 0.0004297366543268101,
      "loss": 2.0796,
      "step": 82403
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.8798532485961914,
      "learning_rate": 0.00042973296603780563,
      "loss": 2.2479,
      "step": 82404
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.8555840253829956,
      "learning_rate": 0.00042972927772468137,
      "loss": 2.0477,
      "step": 82405
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.905585527420044,
      "learning_rate": 0.000429725589387438,
      "loss": 2.1111,
      "step": 82406
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.7647523283958435,
      "learning_rate": 0.00042972190102607617,
      "loss": 1.9955,
      "step": 82407
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.860643744468689,
      "learning_rate": 0.00042971821264059664,
      "loss": 2.052,
      "step": 82408
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.8264535665512085,
      "learning_rate": 0.00042971452423099996,
      "loss": 2.3941,
      "step": 82409
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.912520170211792,
      "learning_rate": 0.000429710835797287,
      "loss": 2.1253,
      "step": 82410
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.7839787006378174,
      "learning_rate": 0.00042970714733945827,
      "loss": 2.058,
      "step": 82411
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.8609030842781067,
      "learning_rate": 0.0004297034588575145,
      "loss": 1.9557,
      "step": 82412
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.9242815375328064,
      "learning_rate": 0.00042969977035145653,
      "loss": 2.339,
      "step": 82413
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.8095183968544006,
      "learning_rate": 0.00042969608182128486,
      "loss": 2.2228,
      "step": 82414
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.7823854088783264,
      "learning_rate": 0.00042969239326700026,
      "loss": 2.0319,
      "step": 82415
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.7449720501899719,
      "learning_rate": 0.0004296887046886033,
      "loss": 2.0396,
      "step": 82416
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.8754721879959106,
      "learning_rate": 0.000429685016086095,
      "loss": 2.4082,
      "step": 82417
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.8991248607635498,
      "learning_rate": 0.00042968132745947556,
      "loss": 2.1369,
      "step": 82418
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.7471627593040466,
      "learning_rate": 0.00042967763880874597,
      "loss": 2.1732,
      "step": 82419
    },
    {
      "epoch": 1.07,
      "grad_norm": 1.1326334476470947,
      "learning_rate": 0.0004296739501339069,
      "loss": 2.3587,
      "step": 82420
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.9237613677978516,
      "learning_rate": 0.000429670261434959,
      "loss": 2.11,
      "step": 82421
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.761132538318634,
      "learning_rate": 0.00042966657271190296,
      "loss": 2.2202,
      "step": 82422
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.8182911276817322,
      "learning_rate": 0.00042966288396473933,
      "loss": 2.2427,
      "step": 82423
    },
    {
      "epoch": 1.07,
      "grad_norm": 1.001408576965332,
      "learning_rate": 0.0004296591951934691,
      "loss": 2.299,
      "step": 82424
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.7781127095222473,
      "learning_rate": 0.0004296555063980927,
      "loss": 2.3889,
      "step": 82425
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.7647132277488708,
      "learning_rate": 0.0004296518175786109,
      "loss": 2.1447,
      "step": 82426
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.9235926270484924,
      "learning_rate": 0.00042964812873502435,
      "loss": 1.9793,
      "step": 82427
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.8367729187011719,
      "learning_rate": 0.0004296444398673339,
      "loss": 2.0876,
      "step": 82428
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.7813940644264221,
      "learning_rate": 0.0004296407509755399,
      "loss": 2.2044,
      "step": 82429
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.7964744567871094,
      "learning_rate": 0.00042963706205964336,
      "loss": 2.0024,
      "step": 82430
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.839901864528656,
      "learning_rate": 0.00042963337311964473,
      "loss": 2.0141,
      "step": 82431
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.8660073280334473,
      "learning_rate": 0.00042962968415554494,
      "loss": 2.0981,
      "step": 82432
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.8517192006111145,
      "learning_rate": 0.0004296259951673445,
      "loss": 2.1869,
      "step": 82433
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.8237354159355164,
      "learning_rate": 0.0004296223061550441,
      "loss": 2.1227,
      "step": 82434
    },
    {
      "epoch": 1.07,
      "grad_norm": 1.0841834545135498,
      "learning_rate": 0.00042961861711864454,
      "loss": 2.3696,
      "step": 82435
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.9432141780853271,
      "learning_rate": 0.0004296149280581464,
      "loss": 2.1366,
      "step": 82436
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.7755632400512695,
      "learning_rate": 0.00042961123897355036,
      "loss": 2.1358,
      "step": 82437
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.8189920783042908,
      "learning_rate": 0.00042960754986485724,
      "loss": 2.1964,
      "step": 82438
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.7772570252418518,
      "learning_rate": 0.0004296038607320675,
      "loss": 2.4605,
      "step": 82439
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.8199267983436584,
      "learning_rate": 0.000429600171575182,
      "loss": 2.2072,
      "step": 82440
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.7783020734786987,
      "learning_rate": 0.00042959648239420143,
      "loss": 2.1649,
      "step": 82441
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.8455641865730286,
      "learning_rate": 0.0004295927931891265,
      "loss": 2.3469,
      "step": 82442
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.7656956315040588,
      "learning_rate": 0.00042958910395995764,
      "loss": 2.1627,
      "step": 82443
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.766334593296051,
      "learning_rate": 0.0004295854147066958,
      "loss": 2.2049,
      "step": 82444
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.8118523359298706,
      "learning_rate": 0.0004295817254293416,
      "loss": 2.2037,
      "step": 82445
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.8354893922805786,
      "learning_rate": 0.0004295780361278957,
      "loss": 2.3099,
      "step": 82446
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.7902933359146118,
      "learning_rate": 0.0004295743468023588,
      "loss": 2.1516,
      "step": 82447
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.8597854971885681,
      "learning_rate": 0.00042957065745273163,
      "loss": 2.2936,
      "step": 82448
    },
    {
      "epoch": 1.07,
      "grad_norm": 1.0090405941009521,
      "learning_rate": 0.00042956696807901483,
      "loss": 1.9717,
      "step": 82449
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.8157731890678406,
      "learning_rate": 0.00042956327868120896,
      "loss": 2.3019,
      "step": 82450
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.8312358260154724,
      "learning_rate": 0.00042955958925931494,
      "loss": 1.9548,
      "step": 82451
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.81082683801651,
      "learning_rate": 0.0004295558998133333,
      "loss": 2.2902,
      "step": 82452
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.780119776725769,
      "learning_rate": 0.00042955221034326476,
      "loss": 2.215,
      "step": 82453
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.814279317855835,
      "learning_rate": 0.0004295485208491101,
      "loss": 2.2307,
      "step": 82454
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.8821766972541809,
      "learning_rate": 0.00042954483133086983,
      "loss": 2.0769,
      "step": 82455
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.7618233561515808,
      "learning_rate": 0.0004295411417885448,
      "loss": 2.0757,
      "step": 82456
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.7846662998199463,
      "learning_rate": 0.0004295374522221356,
      "loss": 2.0322,
      "step": 82457
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.8454465866088867,
      "learning_rate": 0.00042953376263164296,
      "loss": 1.8696,
      "step": 82458
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.8376471996307373,
      "learning_rate": 0.00042953007301706745,
      "loss": 2.0037,
      "step": 82459
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.8654309511184692,
      "learning_rate": 0.00042952638337841,
      "loss": 2.0955,
      "step": 82460
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.8756103515625,
      "learning_rate": 0.00042952269371567106,
      "loss": 2.1245,
      "step": 82461
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.8676391839981079,
      "learning_rate": 0.0004295190040288515,
      "loss": 2.0893,
      "step": 82462
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.7742151021957397,
      "learning_rate": 0.00042951531431795176,
      "loss": 2.1966,
      "step": 82463
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.9392064213752747,
      "learning_rate": 0.0004295116245829728,
      "loss": 2.2334,
      "step": 82464
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.8072288632392883,
      "learning_rate": 0.0004295079348239151,
      "loss": 2.3778,
      "step": 82465
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.7960464954376221,
      "learning_rate": 0.00042950424504077954,
      "loss": 2.046,
      "step": 82466
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.8300544023513794,
      "learning_rate": 0.0004295005552335666,
      "loss": 2.1183,
      "step": 82467
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.727331817150116,
      "learning_rate": 0.0004294968654022771,
      "loss": 2.1854,
      "step": 82468
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.9716978669166565,
      "learning_rate": 0.0004294931755469117,
      "loss": 2.1969,
      "step": 82469
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.7864748239517212,
      "learning_rate": 0.00042948948566747105,
      "loss": 2.102,
      "step": 82470
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.7914537787437439,
      "learning_rate": 0.0004294857957639558,
      "loss": 2.121,
      "step": 82471
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.7486716508865356,
      "learning_rate": 0.0004294821058363668,
      "loss": 2.1535,
      "step": 82472
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.8299735188484192,
      "learning_rate": 0.0004294784158847046,
      "loss": 2.1318,
      "step": 82473
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.8554142117500305,
      "learning_rate": 0.0004294747259089699,
      "loss": 2.5247,
      "step": 82474
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.8062765598297119,
      "learning_rate": 0.00042947103590916346,
      "loss": 2.1023,
      "step": 82475
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.8862629532814026,
      "learning_rate": 0.0004294673458852859,
      "loss": 2.163,
      "step": 82476
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.7739863991737366,
      "learning_rate": 0.0004294636558373378,
      "loss": 2.0165,
      "step": 82477
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.8540568351745605,
      "learning_rate": 0.00042945996576532013,
      "loss": 2.3796,
      "step": 82478
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.7385429739952087,
      "learning_rate": 0.0004294562756692333,
      "loss": 2.0695,
      "step": 82479
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.7430675029754639,
      "learning_rate": 0.00042945258554907806,
      "loss": 2.1143,
      "step": 82480
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.7776839137077332,
      "learning_rate": 0.00042944889540485527,
      "loss": 2.2733,
      "step": 82481
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.905779242515564,
      "learning_rate": 0.00042944520523656546,
      "loss": 2.1546,
      "step": 82482
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.817963182926178,
      "learning_rate": 0.00042944151504420924,
      "loss": 2.3117,
      "step": 82483
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.7901372313499451,
      "learning_rate": 0.0004294378248277875,
      "loss": 2.1349,
      "step": 82484
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.7979587316513062,
      "learning_rate": 0.00042943413458730085,
      "loss": 2.1544,
      "step": 82485
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.9824415445327759,
      "learning_rate": 0.00042943044432274983,
      "loss": 2.1505,
      "step": 82486
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.7960831522941589,
      "learning_rate": 0.0004294267540341354,
      "loss": 2.0473,
      "step": 82487
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.9126778244972229,
      "learning_rate": 0.00042942306372145797,
      "loss": 2.2708,
      "step": 82488
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.8521787524223328,
      "learning_rate": 0.0004294193733847184,
      "loss": 2.2887,
      "step": 82489
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.8000617623329163,
      "learning_rate": 0.0004294156830239173,
      "loss": 2.2292,
      "step": 82490
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.7832358479499817,
      "learning_rate": 0.0004294119926390554,
      "loss": 2.1319,
      "step": 82491
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.8157351613044739,
      "learning_rate": 0.00042940830223013333,
      "loss": 2.0651,
      "step": 82492
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.8616949915885925,
      "learning_rate": 0.0004294046117971519,
      "loss": 2.3867,
      "step": 82493
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.7995226979255676,
      "learning_rate": 0.00042940092134011165,
      "loss": 2.123,
      "step": 82494
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.8445715308189392,
      "learning_rate": 0.0004293972308590133,
      "loss": 2.0751,
      "step": 82495
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.745634913444519,
      "learning_rate": 0.0004293935403538576,
      "loss": 2.0519,
      "step": 82496
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.7595621347427368,
      "learning_rate": 0.0004293898498246452,
      "loss": 2.3385,
      "step": 82497
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.8776386976242065,
      "learning_rate": 0.00042938615927137677,
      "loss": 2.01,
      "step": 82498
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.8496397137641907,
      "learning_rate": 0.00042938246869405304,
      "loss": 2.0576,
      "step": 82499
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.8104150295257568,
      "learning_rate": 0.00042937877809267466,
      "loss": 2.0786,
      "step": 82500
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.7553352117538452,
      "learning_rate": 0.00042937508746724223,
      "loss": 2.2853,
      "step": 82501
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.8120996356010437,
      "learning_rate": 0.0004293713968177567,
      "loss": 2.1361,
      "step": 82502
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.8220798969268799,
      "learning_rate": 0.00042936770614421845,
      "loss": 2.2751,
      "step": 82503
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.7517372965812683,
      "learning_rate": 0.0004293640154466283,
      "loss": 1.9997,
      "step": 82504
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.8039882779121399,
      "learning_rate": 0.000429360324724987,
      "loss": 2.1977,
      "step": 82505
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.8335899710655212,
      "learning_rate": 0.00042935663397929516,
      "loss": 2.0982,
      "step": 82506
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.7999593615531921,
      "learning_rate": 0.00042935294320955346,
      "loss": 2.2341,
      "step": 82507
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.815960705280304,
      "learning_rate": 0.0004293492524157626,
      "loss": 2.1941,
      "step": 82508
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.9543952345848083,
      "learning_rate": 0.0004293455615979233,
      "loss": 2.1479,
      "step": 82509
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.786811351776123,
      "learning_rate": 0.0004293418707560362,
      "loss": 2.3574,
      "step": 82510
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.8039479851722717,
      "learning_rate": 0.00042933817989010203,
      "loss": 2.1722,
      "step": 82511
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.8975935578346252,
      "learning_rate": 0.0004293344890001214,
      "loss": 1.9541,
      "step": 82512
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.8701900243759155,
      "learning_rate": 0.0004293307980860951,
      "loss": 2.2144,
      "step": 82513
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.827223539352417,
      "learning_rate": 0.00042932710714802376,
      "loss": 2.3286,
      "step": 82514
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.8580864667892456,
      "learning_rate": 0.000429323416185908,
      "loss": 2.352,
      "step": 82515
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.7770063877105713,
      "learning_rate": 0.00042931972519974867,
      "loss": 2.1513,
      "step": 82516
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.7882599234580994,
      "learning_rate": 0.00042931603418954633,
      "loss": 2.3173,
      "step": 82517
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.8142984509468079,
      "learning_rate": 0.0004293123431553017,
      "loss": 2.008,
      "step": 82518
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.8142222762107849,
      "learning_rate": 0.00042930865209701543,
      "loss": 2.068,
      "step": 82519
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.8583097457885742,
      "learning_rate": 0.0004293049610146883,
      "loss": 2.2966,
      "step": 82520
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.8212131261825562,
      "learning_rate": 0.0004293012699083209,
      "loss": 2.1028,
      "step": 82521
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.8176040053367615,
      "learning_rate": 0.000429297578777914,
      "loss": 2.0228,
      "step": 82522
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.8463528156280518,
      "learning_rate": 0.00042929388762346813,
      "loss": 2.2977,
      "step": 82523
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.8619163632392883,
      "learning_rate": 0.00042929019644498417,
      "loss": 2.3486,
      "step": 82524
    },
    {
      "epoch": 1.07,
      "grad_norm": 1.3057048320770264,
      "learning_rate": 0.0004292865052424627,
      "loss": 2.0936,
      "step": 82525
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.8483642339706421,
      "learning_rate": 0.0004292828140159044,
      "loss": 2.0353,
      "step": 82526
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.7853929400444031,
      "learning_rate": 0.00042927912276531004,
      "loss": 2.2544,
      "step": 82527
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.7947782874107361,
      "learning_rate": 0.0004292754314906802,
      "loss": 2.108,
      "step": 82528
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.7127081155776978,
      "learning_rate": 0.00042927174019201565,
      "loss": 1.8019,
      "step": 82529
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.87447190284729,
      "learning_rate": 0.0004292680488693171,
      "loss": 2.0778,
      "step": 82530
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.7242128252983093,
      "learning_rate": 0.00042926435752258505,
      "loss": 2.2136,
      "step": 82531
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.8391398191452026,
      "learning_rate": 0.00042926066615182033,
      "loss": 2.1574,
      "step": 82532
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.736217200756073,
      "learning_rate": 0.0004292569747570238,
      "loss": 2.1286,
      "step": 82533
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.8073626756668091,
      "learning_rate": 0.0004292532833381958,
      "loss": 2.3841,
      "step": 82534
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.8018981218338013,
      "learning_rate": 0.0004292495918953372,
      "loss": 2.0716,
      "step": 82535
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.7940728664398193,
      "learning_rate": 0.00042924590042844863,
      "loss": 2.147,
      "step": 82536
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.8485926389694214,
      "learning_rate": 0.0004292422089375309,
      "loss": 2.1724,
      "step": 82537
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.8111258149147034,
      "learning_rate": 0.00042923851742258446,
      "loss": 2.3941,
      "step": 82538
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.7884397506713867,
      "learning_rate": 0.00042923482588361027,
      "loss": 2.0242,
      "step": 82539
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.8622543215751648,
      "learning_rate": 0.0004292311343206088,
      "loss": 2.0223,
      "step": 82540
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.7985016107559204,
      "learning_rate": 0.0004292274427335809,
      "loss": 2.1786,
      "step": 82541
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.7604074478149414,
      "learning_rate": 0.00042922375112252714,
      "loss": 2.0521,
      "step": 82542
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.8741297125816345,
      "learning_rate": 0.00042922005948744824,
      "loss": 2.0194,
      "step": 82543
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.9760916829109192,
      "learning_rate": 0.0004292163678283449,
      "loss": 1.9766,
      "step": 82544
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.7518317103385925,
      "learning_rate": 0.00042921267614521783,
      "loss": 2.0306,
      "step": 82545
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.8975743055343628,
      "learning_rate": 0.0004292089844380676,
      "loss": 2.1742,
      "step": 82546
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.9568137526512146,
      "learning_rate": 0.000429205292706895,
      "loss": 2.1546,
      "step": 82547
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.8740363717079163,
      "learning_rate": 0.0004292016009517008,
      "loss": 2.396,
      "step": 82548
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.7858887314796448,
      "learning_rate": 0.0004291979091724855,
      "loss": 2.2278,
      "step": 82549
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.942958414554596,
      "learning_rate": 0.0004291942173692498,
      "loss": 2.3036,
      "step": 82550
    },
    {
      "epoch": 1.07,
      "grad_norm": 1.0824772119522095,
      "learning_rate": 0.00042919052554199466,
      "loss": 2.0708,
      "step": 82551
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.7924076318740845,
      "learning_rate": 0.0004291868336907204,
      "loss": 2.0995,
      "step": 82552
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.8812323808670044,
      "learning_rate": 0.00042918314181542786,
      "loss": 2.3664,
      "step": 82553
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.8785277009010315,
      "learning_rate": 0.00042917944991611784,
      "loss": 2.2333,
      "step": 82554
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.7951865196228027,
      "learning_rate": 0.00042917575799279083,
      "loss": 2.2788,
      "step": 82555
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.7566342353820801,
      "learning_rate": 0.00042917206604544767,
      "loss": 2.1804,
      "step": 82556
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.8525633215904236,
      "learning_rate": 0.00042916837407408893,
      "loss": 2.3771,
      "step": 82557
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.8680478930473328,
      "learning_rate": 0.0004291646820787154,
      "loss": 2.1174,
      "step": 82558
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.6955278515815735,
      "learning_rate": 0.00042916099005932764,
      "loss": 2.2166,
      "step": 82559
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.8866980671882629,
      "learning_rate": 0.00042915729801592655,
      "loss": 2.1432,
      "step": 82560
    },
    {
      "epoch": 1.07,
      "grad_norm": 0.8175172805786133,
      "learning_rate": 0.0004291536059485126,
      "loss": 2.1915,
      "step": 82561
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.7571945190429688,
      "learning_rate": 0.0004291499138570865,
      "loss": 2.3219,
      "step": 82562
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.7981096506118774,
      "learning_rate": 0.00042914622174164906,
      "loss": 2.1064,
      "step": 82563
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.8186558485031128,
      "learning_rate": 0.0004291425296022009,
      "loss": 2.1035,
      "step": 82564
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.8159207701683044,
      "learning_rate": 0.00042913883743874267,
      "loss": 2.153,
      "step": 82565
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.8408458828926086,
      "learning_rate": 0.0004291351452512752,
      "loss": 1.9211,
      "step": 82566
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.868106484413147,
      "learning_rate": 0.00042913145303979895,
      "loss": 2.1852,
      "step": 82567
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.7907776236534119,
      "learning_rate": 0.00042912776080431474,
      "loss": 2.3073,
      "step": 82568
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.8324974775314331,
      "learning_rate": 0.0004291240685448233,
      "loss": 2.0568,
      "step": 82569
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.8800264596939087,
      "learning_rate": 0.0004291203762613252,
      "loss": 2.1021,
      "step": 82570
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.7751270532608032,
      "learning_rate": 0.0004291166839538212,
      "loss": 2.0768,
      "step": 82571
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.8296101689338684,
      "learning_rate": 0.00042911299162231203,
      "loss": 2.2401,
      "step": 82572
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.7639121413230896,
      "learning_rate": 0.0004291092992667982,
      "loss": 2.0809,
      "step": 82573
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.8780791759490967,
      "learning_rate": 0.0004291056068872806,
      "loss": 2.1316,
      "step": 82574
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.8195126056671143,
      "learning_rate": 0.0004291019144837598,
      "loss": 2.1599,
      "step": 82575
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.9586601853370667,
      "learning_rate": 0.00042909822205623656,
      "loss": 2.2328,
      "step": 82576
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.8106756806373596,
      "learning_rate": 0.0004290945296047115,
      "loss": 2.4404,
      "step": 82577
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.7638047337532043,
      "learning_rate": 0.0004290908371291852,
      "loss": 2.0273,
      "step": 82578
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.7626656293869019,
      "learning_rate": 0.0004290871446296587,
      "loss": 2.1572,
      "step": 82579
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.7795729637145996,
      "learning_rate": 0.00042908345210613234,
      "loss": 2.0743,
      "step": 82580
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.8593224883079529,
      "learning_rate": 0.00042907975955860695,
      "loss": 2.3346,
      "step": 82581
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.8448800444602966,
      "learning_rate": 0.0004290760669870833,
      "loss": 2.1684,
      "step": 82582
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.8885493278503418,
      "learning_rate": 0.00042907237439156177,
      "loss": 2.445,
      "step": 82583
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.8759183883666992,
      "learning_rate": 0.00042906868177204334,
      "loss": 2.2052,
      "step": 82584
    },
    {
      "epoch": 1.08,
      "grad_norm": 1.0097875595092773,
      "learning_rate": 0.00042906498912852865,
      "loss": 2.0429,
      "step": 82585
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.8087126612663269,
      "learning_rate": 0.0004290612964610183,
      "loss": 2.3035,
      "step": 82586
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.7808142900466919,
      "learning_rate": 0.00042905760376951306,
      "loss": 2.1144,
      "step": 82587
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.875740647315979,
      "learning_rate": 0.0004290539110540135,
      "loss": 2.1709,
      "step": 82588
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.8384014368057251,
      "learning_rate": 0.0004290502183145204,
      "loss": 2.1152,
      "step": 82589
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.842920184135437,
      "learning_rate": 0.0004290465255510345,
      "loss": 2.2272,
      "step": 82590
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.8782606720924377,
      "learning_rate": 0.0004290428327635564,
      "loss": 2.3491,
      "step": 82591
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.7617620825767517,
      "learning_rate": 0.00042903913995208674,
      "loss": 2.1398,
      "step": 82592
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.8046529293060303,
      "learning_rate": 0.00042903544711662625,
      "loss": 2.4145,
      "step": 82593
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.7833007574081421,
      "learning_rate": 0.0004290317542571757,
      "loss": 2.0914,
      "step": 82594
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.7934345006942749,
      "learning_rate": 0.00042902806137373576,
      "loss": 2.3926,
      "step": 82595
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.8204193115234375,
      "learning_rate": 0.00042902436846630696,
      "loss": 2.242,
      "step": 82596
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.7930258512496948,
      "learning_rate": 0.0004290206755348902,
      "loss": 2.0554,
      "step": 82597
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.7945374846458435,
      "learning_rate": 0.0004290169825794859,
      "loss": 1.9981,
      "step": 82598
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.8169363141059875,
      "learning_rate": 0.000429013289600095,
      "loss": 2.1618,
      "step": 82599
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.8654765486717224,
      "learning_rate": 0.0004290095965967181,
      "loss": 2.1004,
      "step": 82600
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.8483050465583801,
      "learning_rate": 0.00042900590356935594,
      "loss": 2.179,
      "step": 82601
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.7829052805900574,
      "learning_rate": 0.00042900221051800905,
      "loss": 2.0014,
      "step": 82602
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.7798289656639099,
      "learning_rate": 0.0004289985174426782,
      "loss": 2.2496,
      "step": 82603
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.8394733667373657,
      "learning_rate": 0.00042899482434336417,
      "loss": 2.1868,
      "step": 82604
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.8741812705993652,
      "learning_rate": 0.00042899113122006757,
      "loss": 2.3544,
      "step": 82605
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.7833894491195679,
      "learning_rate": 0.000428987438072789,
      "loss": 2.0887,
      "step": 82606
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.8846497535705566,
      "learning_rate": 0.0004289837449015293,
      "loss": 2.2143,
      "step": 82607
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.803207516670227,
      "learning_rate": 0.0004289800517062891,
      "loss": 2.2417,
      "step": 82608
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.9238481521606445,
      "learning_rate": 0.000428976358487069,
      "loss": 2.2759,
      "step": 82609
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.7733995318412781,
      "learning_rate": 0.0004289726652438698,
      "loss": 2.1951,
      "step": 82610
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.791388213634491,
      "learning_rate": 0.0004289689719766921,
      "loss": 2.3172,
      "step": 82611
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.979065477848053,
      "learning_rate": 0.0004289652786855367,
      "loss": 2.1584,
      "step": 82612
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.8150458335876465,
      "learning_rate": 0.0004289615853704042,
      "loss": 2.3112,
      "step": 82613
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.8022639751434326,
      "learning_rate": 0.0004289578920312953,
      "loss": 2.2227,
      "step": 82614
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.8793064951896667,
      "learning_rate": 0.0004289541986682108,
      "loss": 2.3044,
      "step": 82615
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.8050744533538818,
      "learning_rate": 0.0004289505052811511,
      "loss": 2.2072,
      "step": 82616
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.9703182578086853,
      "learning_rate": 0.0004289468118701172,
      "loss": 2.0781,
      "step": 82617
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.8371450901031494,
      "learning_rate": 0.00042894311843510964,
      "loss": 2.135,
      "step": 82618
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.771324634552002,
      "learning_rate": 0.00042893942497612903,
      "loss": 2.1623,
      "step": 82619
    },
    {
      "epoch": 1.08,
      "grad_norm": 1.2101562023162842,
      "learning_rate": 0.00042893573149317617,
      "loss": 2.0396,
      "step": 82620
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.87175452709198,
      "learning_rate": 0.0004289320379862518,
      "loss": 2.3132,
      "step": 82621
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.9430429339408875,
      "learning_rate": 0.00042892834445535646,
      "loss": 2.2975,
      "step": 82622
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.9059852957725525,
      "learning_rate": 0.0004289246509004909,
      "loss": 2.0494,
      "step": 82623
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.8770322203636169,
      "learning_rate": 0.0004289209573216559,
      "loss": 2.2558,
      "step": 82624
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.8413540124893188,
      "learning_rate": 0.000428917263718852,
      "loss": 2.1601,
      "step": 82625
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.8419857025146484,
      "learning_rate": 0.00042891357009207994,
      "loss": 2.1213,
      "step": 82626
    },
    {
      "epoch": 1.08,
      "grad_norm": 1.0353405475616455,
      "learning_rate": 0.00042890987644134045,
      "loss": 2.157,
      "step": 82627
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.865700364112854,
      "learning_rate": 0.00042890618276663413,
      "loss": 2.3516,
      "step": 82628
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.8551226258277893,
      "learning_rate": 0.00042890248906796174,
      "loss": 2.3326,
      "step": 82629
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.9296488761901855,
      "learning_rate": 0.000428898795345324,
      "loss": 2.3765,
      "step": 82630
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.8483399748802185,
      "learning_rate": 0.0004288951015987215,
      "loss": 2.4082,
      "step": 82631
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.8112564086914062,
      "learning_rate": 0.00042889140782815497,
      "loss": 2.1258,
      "step": 82632
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.787386953830719,
      "learning_rate": 0.0004288877140336251,
      "loss": 2.0793,
      "step": 82633
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.91669762134552,
      "learning_rate": 0.0004288840202151325,
      "loss": 2.3022,
      "step": 82634
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.8420060276985168,
      "learning_rate": 0.000428880326372678,
      "loss": 2.0623,
      "step": 82635
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.9285756349563599,
      "learning_rate": 0.0004288766325062622,
      "loss": 2.3562,
      "step": 82636
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.8332093358039856,
      "learning_rate": 0.00042887293861588576,
      "loss": 2.1275,
      "step": 82637
    },
    {
      "epoch": 1.08,
      "grad_norm": 1.0009288787841797,
      "learning_rate": 0.00042886924470154946,
      "loss": 2.1728,
      "step": 82638
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.8695279359817505,
      "learning_rate": 0.0004288655507632539,
      "loss": 1.962,
      "step": 82639
    },
    {
      "epoch": 1.08,
      "grad_norm": 1.0127451419830322,
      "learning_rate": 0.00042886185680099983,
      "loss": 2.0206,
      "step": 82640
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.9195995926856995,
      "learning_rate": 0.0004288581628147879,
      "loss": 2.2422,
      "step": 82641
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.7439578771591187,
      "learning_rate": 0.00042885446880461877,
      "loss": 2.3487,
      "step": 82642
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.8365310430526733,
      "learning_rate": 0.00042885077477049325,
      "loss": 2.0217,
      "step": 82643
    },
    {
      "epoch": 1.08,
      "grad_norm": 1.028051733970642,
      "learning_rate": 0.0004288470807124119,
      "loss": 2.273,
      "step": 82644
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.8622797131538391,
      "learning_rate": 0.0004288433866303754,
      "loss": 2.2968,
      "step": 82645
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.8499638438224792,
      "learning_rate": 0.00042883969252438455,
      "loss": 2.2342,
      "step": 82646
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.9414075016975403,
      "learning_rate": 0.0004288359983944399,
      "loss": 2.1867,
      "step": 82647
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.7943894863128662,
      "learning_rate": 0.0004288323042405422,
      "loss": 2.2464,
      "step": 82648
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.8147170543670654,
      "learning_rate": 0.0004288286100626922,
      "loss": 2.1583,
      "step": 82649
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.8355873823165894,
      "learning_rate": 0.0004288249158608905,
      "loss": 2.3511,
      "step": 82650
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.911376416683197,
      "learning_rate": 0.0004288212216351379,
      "loss": 2.2062,
      "step": 82651
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.7990962266921997,
      "learning_rate": 0.0004288175273854349,
      "loss": 2.1685,
      "step": 82652
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.8718927502632141,
      "learning_rate": 0.00042881383311178236,
      "loss": 2.1872,
      "step": 82653
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.8906572461128235,
      "learning_rate": 0.0004288101388141808,
      "loss": 2.2641,
      "step": 82654
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.9180262684822083,
      "learning_rate": 0.0004288064444926311,
      "loss": 2.1784,
      "step": 82655
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.7565280199050903,
      "learning_rate": 0.0004288027501471338,
      "loss": 2.0454,
      "step": 82656
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.8186882734298706,
      "learning_rate": 0.00042879905577768965,
      "loss": 1.9106,
      "step": 82657
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.7723197340965271,
      "learning_rate": 0.0004287953613842994,
      "loss": 2.3467,
      "step": 82658
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.8234040141105652,
      "learning_rate": 0.0004287916669669635,
      "loss": 2.2512,
      "step": 82659
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.729972243309021,
      "learning_rate": 0.00042878797252568284,
      "loss": 1.9736,
      "step": 82660
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.8213094472885132,
      "learning_rate": 0.0004287842780604582,
      "loss": 2.0838,
      "step": 82661
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.8919051885604858,
      "learning_rate": 0.00042878058357129,
      "loss": 2.0248,
      "step": 82662
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.9367493987083435,
      "learning_rate": 0.0004287768890581791,
      "loss": 2.2081,
      "step": 82663
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.8153663873672485,
      "learning_rate": 0.0004287731945211261,
      "loss": 2.3539,
      "step": 82664
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.8586822152137756,
      "learning_rate": 0.0004287694999601318,
      "loss": 2.0829,
      "step": 82665
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.8986114859580994,
      "learning_rate": 0.0004287658053751968,
      "loss": 1.9893,
      "step": 82666
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.8695138692855835,
      "learning_rate": 0.00042876211076632185,
      "loss": 2.1336,
      "step": 82667
    },
    {
      "epoch": 1.08,
      "grad_norm": 1.0336363315582275,
      "learning_rate": 0.00042875841613350754,
      "loss": 2.1607,
      "step": 82668
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.7770475745201111,
      "learning_rate": 0.0004287547214767546,
      "loss": 2.2627,
      "step": 82669
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.8632554411888123,
      "learning_rate": 0.00042875102679606374,
      "loss": 2.3216,
      "step": 82670
    },
    {
      "epoch": 1.08,
      "grad_norm": 1.1027618646621704,
      "learning_rate": 0.0004287473320914356,
      "loss": 2.0187,
      "step": 82671
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.8348985910415649,
      "learning_rate": 0.00042874363736287096,
      "loss": 2.1799,
      "step": 82672
    },
    {
      "epoch": 1.08,
      "grad_norm": 1.1413357257843018,
      "learning_rate": 0.00042873994261037043,
      "loss": 2.1566,
      "step": 82673
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.7742188572883606,
      "learning_rate": 0.0004287362478339347,
      "loss": 2.1581,
      "step": 82674
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.826639711856842,
      "learning_rate": 0.00042873255303356445,
      "loss": 2.3696,
      "step": 82675
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.8231716156005859,
      "learning_rate": 0.00042872885820926046,
      "loss": 2.2568,
      "step": 82676
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.8556660413742065,
      "learning_rate": 0.0004287251633610233,
      "loss": 2.1886,
      "step": 82677
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.8125509023666382,
      "learning_rate": 0.0004287214684888537,
      "loss": 2.205,
      "step": 82678
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.7741096019744873,
      "learning_rate": 0.0004287177735927524,
      "loss": 2.2991,
      "step": 82679
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.8448543548583984,
      "learning_rate": 0.00042871407867271994,
      "loss": 1.9567,
      "step": 82680
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.8659362196922302,
      "learning_rate": 0.00042871038372875716,
      "loss": 2.2397,
      "step": 82681
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.8559765219688416,
      "learning_rate": 0.0004287066887608647,
      "loss": 2.3939,
      "step": 82682
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.8066516518592834,
      "learning_rate": 0.0004287029937690432,
      "loss": 2.2733,
      "step": 82683
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.7710385918617249,
      "learning_rate": 0.0004286992987532934,
      "loss": 2.3811,
      "step": 82684
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.8450511693954468,
      "learning_rate": 0.0004286956037136159,
      "loss": 2.2736,
      "step": 82685
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.7945202589035034,
      "learning_rate": 0.0004286919086500116,
      "loss": 2.1451,
      "step": 82686
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.7095014452934265,
      "learning_rate": 0.000428688213562481,
      "loss": 2.0839,
      "step": 82687
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.8902984261512756,
      "learning_rate": 0.00042868451845102476,
      "loss": 2.019,
      "step": 82688
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.7794332504272461,
      "learning_rate": 0.0004286808233156437,
      "loss": 2.1536,
      "step": 82689
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.862712025642395,
      "learning_rate": 0.00042867712815633844,
      "loss": 2.3511,
      "step": 82690
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.8751229643821716,
      "learning_rate": 0.00042867343297310964,
      "loss": 2.2537,
      "step": 82691
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.843241274356842,
      "learning_rate": 0.00042866973776595804,
      "loss": 2.3079,
      "step": 82692
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.8675235509872437,
      "learning_rate": 0.0004286660425348844,
      "loss": 1.9366,
      "step": 82693
    },
    {
      "epoch": 1.08,
      "grad_norm": 1.0015451908111572,
      "learning_rate": 0.00042866234727988915,
      "loss": 2.2841,
      "step": 82694
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.7443937063217163,
      "learning_rate": 0.00042865865200097323,
      "loss": 2.0557,
      "step": 82695
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.7703492045402527,
      "learning_rate": 0.0004286549566981372,
      "loss": 2.2529,
      "step": 82696
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.8988826870918274,
      "learning_rate": 0.00042865126137138184,
      "loss": 2.1975,
      "step": 82697
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.8470406532287598,
      "learning_rate": 0.0004286475660207077,
      "loss": 2.2944,
      "step": 82698
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.929905116558075,
      "learning_rate": 0.0004286438706461157,
      "loss": 2.2204,
      "step": 82699
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.8637348413467407,
      "learning_rate": 0.0004286401752476063,
      "loss": 2.3131,
      "step": 82700
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.8466817736625671,
      "learning_rate": 0.00042863647982518017,
      "loss": 2.2777,
      "step": 82701
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.8597484230995178,
      "learning_rate": 0.00042863278437883817,
      "loss": 2.3642,
      "step": 82702
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.846806526184082,
      "learning_rate": 0.00042862908890858094,
      "loss": 2.1949,
      "step": 82703
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.7217227220535278,
      "learning_rate": 0.0004286253934144091,
      "loss": 2.003,
      "step": 82704
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.7800495624542236,
      "learning_rate": 0.0004286216978963233,
      "loss": 2.1756,
      "step": 82705
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.8352699279785156,
      "learning_rate": 0.00042861800235432444,
      "loss": 2.5276,
      "step": 82706
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.9861782193183899,
      "learning_rate": 0.00042861430678841294,
      "loss": 2.2587,
      "step": 82707
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.789650022983551,
      "learning_rate": 0.0004286106111985897,
      "loss": 2.0875,
      "step": 82708
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.8311426043510437,
      "learning_rate": 0.0004286069155848552,
      "loss": 2.1105,
      "step": 82709
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.7714316248893738,
      "learning_rate": 0.0004286032199472104,
      "loss": 2.0928,
      "step": 82710
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.8156108260154724,
      "learning_rate": 0.00042859952428565574,
      "loss": 2.0932,
      "step": 82711
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.9311373829841614,
      "learning_rate": 0.000428595828600192,
      "loss": 2.1225,
      "step": 82712
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.9983826875686646,
      "learning_rate": 0.00042859213289082,
      "loss": 2.2851,
      "step": 82713
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.841569721698761,
      "learning_rate": 0.0004285884371575401,
      "loss": 2.2655,
      "step": 82714
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.8185243010520935,
      "learning_rate": 0.00042858474140035325,
      "loss": 2.1376,
      "step": 82715
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.8170694708824158,
      "learning_rate": 0.0004285810456192601,
      "loss": 2.2852,
      "step": 82716
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.8064388632774353,
      "learning_rate": 0.00042857734981426135,
      "loss": 1.8858,
      "step": 82717
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.8046771287918091,
      "learning_rate": 0.0004285736539853575,
      "loss": 2.1909,
      "step": 82718
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.8760032057762146,
      "learning_rate": 0.00042856995813254955,
      "loss": 2.0798,
      "step": 82719
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.8915379047393799,
      "learning_rate": 0.0004285662622558379,
      "loss": 2.2226,
      "step": 82720
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.7958616018295288,
      "learning_rate": 0.0004285625663552234,
      "loss": 1.928,
      "step": 82721
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.7664453983306885,
      "learning_rate": 0.0004285588704307067,
      "loss": 2.0292,
      "step": 82722
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.8874483108520508,
      "learning_rate": 0.00042855517448228843,
      "loss": 2.2629,
      "step": 82723
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.8372499942779541,
      "learning_rate": 0.0004285514785099694,
      "loss": 2.0309,
      "step": 82724
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.8063783645629883,
      "learning_rate": 0.0004285477825137502,
      "loss": 2.2756,
      "step": 82725
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.8098617792129517,
      "learning_rate": 0.0004285440864936315,
      "loss": 2.1699,
      "step": 82726
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.8465996980667114,
      "learning_rate": 0.00042854039044961403,
      "loss": 2.279,
      "step": 82727
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.7872287631034851,
      "learning_rate": 0.0004285366943816985,
      "loss": 2.0587,
      "step": 82728
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.7846754789352417,
      "learning_rate": 0.00042853299828988557,
      "loss": 2.0096,
      "step": 82729
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.784034013748169,
      "learning_rate": 0.00042852930217417596,
      "loss": 2.2242,
      "step": 82730
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.8691042065620422,
      "learning_rate": 0.0004285256060345703,
      "loss": 2.5734,
      "step": 82731
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.8142167925834656,
      "learning_rate": 0.0004285219098710693,
      "loss": 2.1224,
      "step": 82732
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.8598754405975342,
      "learning_rate": 0.0004285182136836736,
      "loss": 2.2523,
      "step": 82733
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.8244990110397339,
      "learning_rate": 0.000428514517472384,
      "loss": 2.0895,
      "step": 82734
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.8975067138671875,
      "learning_rate": 0.00042851082123720113,
      "loss": 1.9736,
      "step": 82735
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.7342104911804199,
      "learning_rate": 0.0004285071249781256,
      "loss": 2.1728,
      "step": 82736
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.8608270287513733,
      "learning_rate": 0.00042850342869515826,
      "loss": 2.0155,
      "step": 82737
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.7924336194992065,
      "learning_rate": 0.0004284997323882997,
      "loss": 2.1132,
      "step": 82738
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.8893671035766602,
      "learning_rate": 0.0004284960360575506,
      "loss": 2.3235,
      "step": 82739
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.8507806658744812,
      "learning_rate": 0.0004284923397029117,
      "loss": 2.2945,
      "step": 82740
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.9259881973266602,
      "learning_rate": 0.0004284886433243835,
      "loss": 2.2877,
      "step": 82741
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.7217293381690979,
      "learning_rate": 0.0004284849469219669,
      "loss": 2.2037,
      "step": 82742
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.8592883944511414,
      "learning_rate": 0.0004284812504956626,
      "loss": 2.1362,
      "step": 82743
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.9563609957695007,
      "learning_rate": 0.00042847755404547115,
      "loss": 2.1935,
      "step": 82744
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.9122743606567383,
      "learning_rate": 0.0004284738575713933,
      "loss": 2.2148,
      "step": 82745
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.8812958598136902,
      "learning_rate": 0.0004284701610734298,
      "loss": 2.12,
      "step": 82746
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.971280574798584,
      "learning_rate": 0.0004284664645515812,
      "loss": 2.2012,
      "step": 82747
    },
    {
      "epoch": 1.08,
      "grad_norm": 1.1146984100341797,
      "learning_rate": 0.00042846276800584827,
      "loss": 2.5707,
      "step": 82748
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.7747803926467896,
      "learning_rate": 0.0004284590714362317,
      "loss": 1.9984,
      "step": 82749
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.9042848944664001,
      "learning_rate": 0.00042845537484273216,
      "loss": 1.9852,
      "step": 82750
    },
    {
      "epoch": 1.08,
      "grad_norm": 1.1501131057739258,
      "learning_rate": 0.00042845167822535033,
      "loss": 2.2059,
      "step": 82751
    },
    {
      "epoch": 1.08,
      "grad_norm": 1.1057672500610352,
      "learning_rate": 0.00042844798158408695,
      "loss": 2.3451,
      "step": 82752
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.8655515909194946,
      "learning_rate": 0.00042844428491894265,
      "loss": 2.0779,
      "step": 82753
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.9690262675285339,
      "learning_rate": 0.00042844058822991805,
      "loss": 2.1607,
      "step": 82754
    },
    {
      "epoch": 1.08,
      "grad_norm": 2.144979953765869,
      "learning_rate": 0.00042843689151701405,
      "loss": 2.272,
      "step": 82755
    },
    {
      "epoch": 1.08,
      "grad_norm": 1.0724114179611206,
      "learning_rate": 0.00042843319478023115,
      "loss": 2.1781,
      "step": 82756
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.8540694117546082,
      "learning_rate": 0.00042842949801957006,
      "loss": 2.2515,
      "step": 82757
    },
    {
      "epoch": 1.08,
      "grad_norm": 1.0843931436538696,
      "learning_rate": 0.0004284258012350316,
      "loss": 2.1253,
      "step": 82758
    },
    {
      "epoch": 1.08,
      "grad_norm": 1.2266192436218262,
      "learning_rate": 0.00042842210442661625,
      "loss": 2.031,
      "step": 82759
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.819950520992279,
      "learning_rate": 0.00042841840759432485,
      "loss": 2.1394,
      "step": 82760
    },
    {
      "epoch": 1.08,
      "grad_norm": 1.4772886037826538,
      "learning_rate": 0.000428414710738158,
      "loss": 2.1525,
      "step": 82761
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.8458837866783142,
      "learning_rate": 0.00042841101385811655,
      "loss": 2.104,
      "step": 82762
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.9384921193122864,
      "learning_rate": 0.00042840731695420096,
      "loss": 2.2472,
      "step": 82763
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.9411019086837769,
      "learning_rate": 0.00042840362002641205,
      "loss": 2.1942,
      "step": 82764
    },
    {
      "epoch": 1.08,
      "grad_norm": 1.162329912185669,
      "learning_rate": 0.0004283999230747505,
      "loss": 2.0605,
      "step": 82765
    },
    {
      "epoch": 1.08,
      "grad_norm": 1.2349599599838257,
      "learning_rate": 0.00042839622609921693,
      "loss": 2.3157,
      "step": 82766
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.8186571598052979,
      "learning_rate": 0.0004283925290998122,
      "loss": 2.3471,
      "step": 82767
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.7730208039283752,
      "learning_rate": 0.0004283888320765367,
      "loss": 2.0549,
      "step": 82768
    },
    {
      "epoch": 1.08,
      "grad_norm": 1.204655647277832,
      "learning_rate": 0.00042838513502939135,
      "loss": 2.2951,
      "step": 82769
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.999687671661377,
      "learning_rate": 0.0004283814379583769,
      "loss": 2.4642,
      "step": 82770
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.7990424036979675,
      "learning_rate": 0.00042837774086349384,
      "loss": 2.1413,
      "step": 82771
    },
    {
      "epoch": 1.08,
      "grad_norm": 1.2350969314575195,
      "learning_rate": 0.00042837404374474294,
      "loss": 2.289,
      "step": 82772
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.9095224738121033,
      "learning_rate": 0.0004283703466021248,
      "loss": 2.2398,
      "step": 82773
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.9017031788825989,
      "learning_rate": 0.0004283666494356403,
      "loss": 2.3715,
      "step": 82774
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.7756747603416443,
      "learning_rate": 0.00042836295224529,
      "loss": 2.371,
      "step": 82775
    },
    {
      "epoch": 1.08,
      "grad_norm": 1.3633947372436523,
      "learning_rate": 0.00042835925503107456,
      "loss": 2.0111,
      "step": 82776
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.8428063988685608,
      "learning_rate": 0.00042835555779299463,
      "loss": 1.9099,
      "step": 82777
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.780340313911438,
      "learning_rate": 0.0004283518605310512,
      "loss": 2.2058,
      "step": 82778
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.8127752542495728,
      "learning_rate": 0.0004283481632452446,
      "loss": 2.1064,
      "step": 82779
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.8737683892250061,
      "learning_rate": 0.00042834446593557565,
      "loss": 2.3736,
      "step": 82780
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.8183702230453491,
      "learning_rate": 0.0004283407686020451,
      "loss": 2.2423,
      "step": 82781
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.8458855748176575,
      "learning_rate": 0.0004283370712446535,
      "loss": 2.3244,
      "step": 82782
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.787449061870575,
      "learning_rate": 0.0004283333738634016,
      "loss": 2.0699,
      "step": 82783
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.8125799298286438,
      "learning_rate": 0.0004283296764582902,
      "loss": 2.3722,
      "step": 82784
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.8081597089767456,
      "learning_rate": 0.00042832597902931987,
      "loss": 2.259,
      "step": 82785
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.7990394234657288,
      "learning_rate": 0.0004283222815764913,
      "loss": 2.2311,
      "step": 82786
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.783829391002655,
      "learning_rate": 0.0004283185840998051,
      "loss": 2.0197,
      "step": 82787
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.8140546679496765,
      "learning_rate": 0.0004283148865992623,
      "loss": 2.2287,
      "step": 82788
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.7719601392745972,
      "learning_rate": 0.00042831118907486315,
      "loss": 2.158,
      "step": 82789
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.8102313876152039,
      "learning_rate": 0.0004283074915266085,
      "loss": 2.3017,
      "step": 82790
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.8010250926017761,
      "learning_rate": 0.00042830379395449915,
      "loss": 2.2966,
      "step": 82791
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.8086833953857422,
      "learning_rate": 0.0004283000963585357,
      "loss": 2.0594,
      "step": 82792
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.8146690130233765,
      "learning_rate": 0.0004282963987387188,
      "loss": 2.1884,
      "step": 82793
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.8528698086738586,
      "learning_rate": 0.0004282927010950492,
      "loss": 2.3431,
      "step": 82794
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.8424081206321716,
      "learning_rate": 0.0004282890034275276,
      "loss": 2.3038,
      "step": 82795
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.7812138199806213,
      "learning_rate": 0.0004282853057361546,
      "loss": 2.2851,
      "step": 82796
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.9443435668945312,
      "learning_rate": 0.0004282816080209309,
      "loss": 2.0226,
      "step": 82797
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.8286474943161011,
      "learning_rate": 0.00042827791028185734,
      "loss": 2.0957,
      "step": 82798
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.8505046963691711,
      "learning_rate": 0.00042827421251893444,
      "loss": 2.2319,
      "step": 82799
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.8842052817344666,
      "learning_rate": 0.00042827051473216294,
      "loss": 2.1471,
      "step": 82800
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.9019387364387512,
      "learning_rate": 0.00042826681692154355,
      "loss": 2.2372,
      "step": 82801
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.8375222682952881,
      "learning_rate": 0.0004282631190870769,
      "loss": 2.3291,
      "step": 82802
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.7515714168548584,
      "learning_rate": 0.00042825942122876374,
      "loss": 2.0386,
      "step": 82803
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.8956073522567749,
      "learning_rate": 0.0004282557233466048,
      "loss": 2.0939,
      "step": 82804
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.8599621057510376,
      "learning_rate": 0.0004282520254406006,
      "loss": 2.3381,
      "step": 82805
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.8590460419654846,
      "learning_rate": 0.0004282483275107519,
      "loss": 2.1937,
      "step": 82806
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.8021509051322937,
      "learning_rate": 0.0004282446295570595,
      "loss": 2.221,
      "step": 82807
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.8062871694564819,
      "learning_rate": 0.000428240931579524,
      "loss": 2.1235,
      "step": 82808
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.9200646281242371,
      "learning_rate": 0.000428237233578146,
      "loss": 2.2652,
      "step": 82809
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.8332418203353882,
      "learning_rate": 0.0004282335355529265,
      "loss": 2.0645,
      "step": 82810
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.8094174861907959,
      "learning_rate": 0.0004282298375038657,
      "loss": 1.9735,
      "step": 82811
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.7750905156135559,
      "learning_rate": 0.00042822613943096463,
      "loss": 2.2392,
      "step": 82812
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.7832782864570618,
      "learning_rate": 0.000428222441334224,
      "loss": 2.4297,
      "step": 82813
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.9118656516075134,
      "learning_rate": 0.0004282187432136443,
      "loss": 2.4358,
      "step": 82814
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.9333983659744263,
      "learning_rate": 0.00042821504506922626,
      "loss": 2.4416,
      "step": 82815
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.9119685888290405,
      "learning_rate": 0.0004282113469009708,
      "loss": 2.1776,
      "step": 82816
    },
    {
      "epoch": 1.08,
      "grad_norm": 1.187575101852417,
      "learning_rate": 0.00042820764870887834,
      "loss": 2.114,
      "step": 82817
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.8381453156471252,
      "learning_rate": 0.0004282039504929496,
      "loss": 2.4611,
      "step": 82818
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.9492303729057312,
      "learning_rate": 0.00042820025225318546,
      "loss": 2.3057,
      "step": 82819
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.9745861887931824,
      "learning_rate": 0.00042819655398958635,
      "loss": 2.1682,
      "step": 82820
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.9348161220550537,
      "learning_rate": 0.0004281928557021531,
      "loss": 2.6359,
      "step": 82821
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.7950575947761536,
      "learning_rate": 0.0004281891573908865,
      "loss": 2.2885,
      "step": 82822
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.9595221281051636,
      "learning_rate": 0.00042818545905578696,
      "loss": 2.1283,
      "step": 82823
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.8514853119850159,
      "learning_rate": 0.00042818176069685536,
      "loss": 2.1348,
      "step": 82824
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.7577856779098511,
      "learning_rate": 0.0004281780623140924,
      "loss": 2.2233,
      "step": 82825
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.8332727551460266,
      "learning_rate": 0.00042817436390749867,
      "loss": 2.1765,
      "step": 82826
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.76340252161026,
      "learning_rate": 0.00042817066547707494,
      "loss": 1.9728,
      "step": 82827
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.9426206350326538,
      "learning_rate": 0.0004281669670228219,
      "loss": 2.1388,
      "step": 82828
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.7640552520751953,
      "learning_rate": 0.0004281632685447401,
      "loss": 2.2821,
      "step": 82829
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.8926130533218384,
      "learning_rate": 0.0004281595700428304,
      "loss": 2.1991,
      "step": 82830
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.780342161655426,
      "learning_rate": 0.0004281558715170934,
      "loss": 1.958,
      "step": 82831
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.9172984957695007,
      "learning_rate": 0.00042815217296752984,
      "loss": 2.0063,
      "step": 82832
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.8027618527412415,
      "learning_rate": 0.0004281484743941403,
      "loss": 2.0091,
      "step": 82833
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.7688745856285095,
      "learning_rate": 0.00042814477579692566,
      "loss": 2.3181,
      "step": 82834
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.7827795147895813,
      "learning_rate": 0.00042814107717588633,
      "loss": 1.9994,
      "step": 82835
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.7882429957389832,
      "learning_rate": 0.0004281373785310232,
      "loss": 2.1683,
      "step": 82836
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.86025470495224,
      "learning_rate": 0.00042813367986233703,
      "loss": 2.297,
      "step": 82837
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.8526568412780762,
      "learning_rate": 0.00042812998116982824,
      "loss": 2.2996,
      "step": 82838
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.7871269583702087,
      "learning_rate": 0.0004281262824534977,
      "loss": 2.3543,
      "step": 82839
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.7570662498474121,
      "learning_rate": 0.00042812258371334607,
      "loss": 2.2847,
      "step": 82840
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.9297792911529541,
      "learning_rate": 0.00042811888494937415,
      "loss": 2.2265,
      "step": 82841
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.7549383044242859,
      "learning_rate": 0.00042811518616158237,
      "loss": 2.1701,
      "step": 82842
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.6875090003013611,
      "learning_rate": 0.0004281114873499716,
      "loss": 2.2647,
      "step": 82843
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.9650897979736328,
      "learning_rate": 0.0004281077885145425,
      "loss": 2.2601,
      "step": 82844
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.790690541267395,
      "learning_rate": 0.0004281040896552957,
      "loss": 2.0697,
      "step": 82845
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.7942659258842468,
      "learning_rate": 0.00042810039077223195,
      "loss": 2.2858,
      "step": 82846
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.8679091930389404,
      "learning_rate": 0.00042809669186535205,
      "loss": 2.2671,
      "step": 82847
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.8826347589492798,
      "learning_rate": 0.00042809299293465635,
      "loss": 2.4388,
      "step": 82848
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.7519839406013489,
      "learning_rate": 0.0004280892939801459,
      "loss": 2.2756,
      "step": 82849
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.8767015337944031,
      "learning_rate": 0.0004280855950018212,
      "loss": 2.2278,
      "step": 82850
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.8032746911048889,
      "learning_rate": 0.0004280818959996829,
      "loss": 2.2311,
      "step": 82851
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.7840704321861267,
      "learning_rate": 0.00042807819697373174,
      "loss": 2.2543,
      "step": 82852
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.7696962356567383,
      "learning_rate": 0.0004280744979239685,
      "loss": 2.2994,
      "step": 82853
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.8838754296302795,
      "learning_rate": 0.00042807079885039383,
      "loss": 2.1777,
      "step": 82854
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.7745300531387329,
      "learning_rate": 0.0004280670997530083,
      "loss": 2.4925,
      "step": 82855
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.7943708896636963,
      "learning_rate": 0.00042806340063181273,
      "loss": 1.9911,
      "step": 82856
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.7889518141746521,
      "learning_rate": 0.00042805970148680766,
      "loss": 2.049,
      "step": 82857
    },
    {
      "epoch": 1.08,
      "grad_norm": 1.1424980163574219,
      "learning_rate": 0.000428056002317994,
      "loss": 2.292,
      "step": 82858
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.7686066627502441,
      "learning_rate": 0.00042805230312537224,
      "loss": 2.3596,
      "step": 82859
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.8545920848846436,
      "learning_rate": 0.00042804860390894314,
      "loss": 2.172,
      "step": 82860
    },
    {
      "epoch": 1.08,
      "grad_norm": 1.152158260345459,
      "learning_rate": 0.00042804490466870744,
      "loss": 2.0921,
      "step": 82861
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.8466005325317383,
      "learning_rate": 0.0004280412054046658,
      "loss": 2.1948,
      "step": 82862
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.7572728395462036,
      "learning_rate": 0.0004280375061168188,
      "loss": 1.9772,
      "step": 82863
    },
    {
      "epoch": 1.08,
      "grad_norm": 1.016265869140625,
      "learning_rate": 0.0004280338068051672,
      "loss": 2.2335,
      "step": 82864
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.8133924603462219,
      "learning_rate": 0.0004280301074697117,
      "loss": 2.0524,
      "step": 82865
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.887908399105072,
      "learning_rate": 0.00042802640811045306,
      "loss": 2.2822,
      "step": 82866
    },
    {
      "epoch": 1.08,
      "grad_norm": 1.0185627937316895,
      "learning_rate": 0.00042802270872739184,
      "loss": 1.9572,
      "step": 82867
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.7429115772247314,
      "learning_rate": 0.0004280190093205288,
      "loss": 2.0163,
      "step": 82868
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.7653124332427979,
      "learning_rate": 0.00042801530988986465,
      "loss": 2.0401,
      "step": 82869
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.9561592936515808,
      "learning_rate": 0.00042801161043539993,
      "loss": 2.307,
      "step": 82870
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.8793175220489502,
      "learning_rate": 0.0004280079109571355,
      "loss": 2.2285,
      "step": 82871
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.7988715767860413,
      "learning_rate": 0.000428004211455072,
      "loss": 2.1016,
      "step": 82872
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.8263739347457886,
      "learning_rate": 0.00042800051192921006,
      "loss": 2.3259,
      "step": 82873
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.8810040950775146,
      "learning_rate": 0.00042799681237955044,
      "loss": 2.326,
      "step": 82874
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.8964278101921082,
      "learning_rate": 0.0004279931128060937,
      "loss": 2.2452,
      "step": 82875
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.8214031457901001,
      "learning_rate": 0.00042798941320884077,
      "loss": 2.2236,
      "step": 82876
    },
    {
      "epoch": 1.08,
      "grad_norm": 1.0680060386657715,
      "learning_rate": 0.00042798571358779224,
      "loss": 2.3113,
      "step": 82877
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.8837602734565735,
      "learning_rate": 0.00042798201394294856,
      "loss": 2.2638,
      "step": 82878
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.7644016146659851,
      "learning_rate": 0.0004279783142743107,
      "loss": 2.3405,
      "step": 82879
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.9607621431350708,
      "learning_rate": 0.0004279746145818793,
      "loss": 2.2485,
      "step": 82880
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.7846332788467407,
      "learning_rate": 0.00042797091486565487,
      "loss": 2.3446,
      "step": 82881
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.8008983135223389,
      "learning_rate": 0.00042796721512563835,
      "loss": 2.2303,
      "step": 82882
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.7999501824378967,
      "learning_rate": 0.0004279635153618303,
      "loss": 2.0221,
      "step": 82883
    },
    {
      "epoch": 1.08,
      "grad_norm": 1.0146890878677368,
      "learning_rate": 0.00042795981557423136,
      "loss": 2.3686,
      "step": 82884
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.9100589752197266,
      "learning_rate": 0.0004279561157628423,
      "loss": 2.2826,
      "step": 82885
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.8694826364517212,
      "learning_rate": 0.0004279524159276638,
      "loss": 2.3569,
      "step": 82886
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.8371349573135376,
      "learning_rate": 0.0004279487160686965,
      "loss": 2.0985,
      "step": 82887
    },
    {
      "epoch": 1.08,
      "grad_norm": 1.157728672027588,
      "learning_rate": 0.00042794501618594115,
      "loss": 2.1301,
      "step": 82888
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.8089204430580139,
      "learning_rate": 0.0004279413162793984,
      "loss": 2.1953,
      "step": 82889
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.8758417963981628,
      "learning_rate": 0.0004279376163490689,
      "loss": 2.4779,
      "step": 82890
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.7709676027297974,
      "learning_rate": 0.0004279339163949534,
      "loss": 2.1278,
      "step": 82891
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.7938117980957031,
      "learning_rate": 0.0004279302164170526,
      "loss": 2.1926,
      "step": 82892
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.7373400330543518,
      "learning_rate": 0.0004279265164153672,
      "loss": 2.0911,
      "step": 82893
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.7270998954772949,
      "learning_rate": 0.00042792281638989776,
      "loss": 2.0732,
      "step": 82894
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.8482670783996582,
      "learning_rate": 0.0004279191163406451,
      "loss": 2.1063,
      "step": 82895
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.820506751537323,
      "learning_rate": 0.0004279154162676098,
      "loss": 2.2909,
      "step": 82896
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.8200953006744385,
      "learning_rate": 0.00042791171617079265,
      "loss": 2.2442,
      "step": 82897
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.8690208196640015,
      "learning_rate": 0.00042790801605019433,
      "loss": 2.3652,
      "step": 82898
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.8630041480064392,
      "learning_rate": 0.00042790431590581545,
      "loss": 2.1845,
      "step": 82899
    },
    {
      "epoch": 1.08,
      "grad_norm": 1.0207844972610474,
      "learning_rate": 0.0004279006157376568,
      "loss": 2.154,
      "step": 82900
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.8416759371757507,
      "learning_rate": 0.00042789691554571903,
      "loss": 2.3627,
      "step": 82901
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.8435105681419373,
      "learning_rate": 0.00042789321533000273,
      "loss": 1.9963,
      "step": 82902
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.8610819578170776,
      "learning_rate": 0.00042788951509050864,
      "loss": 2.1869,
      "step": 82903
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.9462425708770752,
      "learning_rate": 0.00042788581482723756,
      "loss": 2.3036,
      "step": 82904
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.7510079145431519,
      "learning_rate": 0.00042788211454019003,
      "loss": 2.203,
      "step": 82905
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.8945187330245972,
      "learning_rate": 0.0004278784142293668,
      "loss": 2.2053,
      "step": 82906
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.8792062997817993,
      "learning_rate": 0.0004278747138947686,
      "loss": 2.2649,
      "step": 82907
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.795576274394989,
      "learning_rate": 0.0004278710135363961,
      "loss": 2.2474,
      "step": 82908
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.8051775097846985,
      "learning_rate": 0.0004278673131542499,
      "loss": 2.0037,
      "step": 82909
    },
    {
      "epoch": 1.08,
      "grad_norm": 1.0213954448699951,
      "learning_rate": 0.00042786361274833076,
      "loss": 2.1003,
      "step": 82910
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.860711932182312,
      "learning_rate": 0.0004278599123186394,
      "loss": 2.1429,
      "step": 82911
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.8427679538726807,
      "learning_rate": 0.00042785621186517645,
      "loss": 2.3282,
      "step": 82912
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.8277234435081482,
      "learning_rate": 0.0004278525113879426,
      "loss": 1.987,
      "step": 82913
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.9689115285873413,
      "learning_rate": 0.00042784881088693864,
      "loss": 2.2154,
      "step": 82914
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.7506341934204102,
      "learning_rate": 0.00042784511036216504,
      "loss": 2.1652,
      "step": 82915
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.9486963748931885,
      "learning_rate": 0.0004278414098136227,
      "loss": 2.0491,
      "step": 82916
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.9706278443336487,
      "learning_rate": 0.00042783770924131225,
      "loss": 2.0828,
      "step": 82917
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.8005170226097107,
      "learning_rate": 0.0004278340086452343,
      "loss": 2.0785,
      "step": 82918
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.7829611301422119,
      "learning_rate": 0.0004278303080253896,
      "loss": 2.1036,
      "step": 82919
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.8312487006187439,
      "learning_rate": 0.0004278266073817789,
      "loss": 2.3103,
      "step": 82920
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.9116177558898926,
      "learning_rate": 0.0004278229067144028,
      "loss": 2.1452,
      "step": 82921
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.8854565024375916,
      "learning_rate": 0.0004278192060232619,
      "loss": 2.3768,
      "step": 82922
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.9815661311149597,
      "learning_rate": 0.0004278155053083571,
      "loss": 1.9565,
      "step": 82923
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.8166897892951965,
      "learning_rate": 0.000427811804569689,
      "loss": 2.2882,
      "step": 82924
    },
    {
      "epoch": 1.08,
      "grad_norm": 1.1747006177902222,
      "learning_rate": 0.0004278081038072582,
      "loss": 2.2887,
      "step": 82925
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.9092726707458496,
      "learning_rate": 0.00042780440302106547,
      "loss": 2.1451,
      "step": 82926
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.9273476004600525,
      "learning_rate": 0.0004278007022111115,
      "loss": 2.0074,
      "step": 82927
    },
    {
      "epoch": 1.08,
      "grad_norm": 1.0452996492385864,
      "learning_rate": 0.000427797001377397,
      "loss": 2.3741,
      "step": 82928
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.7989540100097656,
      "learning_rate": 0.00042779330051992263,
      "loss": 2.1677,
      "step": 82929
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.853829026222229,
      "learning_rate": 0.00042778959963868905,
      "loss": 2.3434,
      "step": 82930
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.8624423742294312,
      "learning_rate": 0.00042778589873369697,
      "loss": 2.2543,
      "step": 82931
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.9854272603988647,
      "learning_rate": 0.0004277821978049471,
      "loss": 2.1892,
      "step": 82932
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.8507792353630066,
      "learning_rate": 0.0004277784968524401,
      "loss": 2.4032,
      "step": 82933
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.8180729150772095,
      "learning_rate": 0.0004277747958761766,
      "loss": 1.9027,
      "step": 82934
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.9197613596916199,
      "learning_rate": 0.0004277710948761574,
      "loss": 2.2013,
      "step": 82935
    },
    {
      "epoch": 1.08,
      "grad_norm": 1.1723203659057617,
      "learning_rate": 0.0004277673938523832,
      "loss": 2.2968,
      "step": 82936
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.8133095502853394,
      "learning_rate": 0.0004277636928048545,
      "loss": 2.2031,
      "step": 82937
    },
    {
      "epoch": 1.08,
      "grad_norm": 1.0325654745101929,
      "learning_rate": 0.00042775999173357227,
      "loss": 1.9846,
      "step": 82938
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.8319414854049683,
      "learning_rate": 0.00042775629063853695,
      "loss": 2.36,
      "step": 82939
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.8383007645606995,
      "learning_rate": 0.0004277525895197493,
      "loss": 2.0716,
      "step": 82940
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.9042065739631653,
      "learning_rate": 0.0004277488883772101,
      "loss": 2.0297,
      "step": 82941
    },
    {
      "epoch": 1.08,
      "grad_norm": 1.0817428827285767,
      "learning_rate": 0.00042774518721092,
      "loss": 1.9953,
      "step": 82942
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.8571682572364807,
      "learning_rate": 0.00042774148602087956,
      "loss": 2.4482,
      "step": 82943
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.8266801238059998,
      "learning_rate": 0.0004277377848070896,
      "loss": 2.0812,
      "step": 82944
    },
    {
      "epoch": 1.08,
      "grad_norm": 1.1778539419174194,
      "learning_rate": 0.00042773408356955084,
      "loss": 2.1884,
      "step": 82945
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.8149247169494629,
      "learning_rate": 0.0004277303823082638,
      "loss": 2.3196,
      "step": 82946
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.7834920287132263,
      "learning_rate": 0.0004277266810232294,
      "loss": 2.1737,
      "step": 82947
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.7811627984046936,
      "learning_rate": 0.0004277229797144481,
      "loss": 2.3962,
      "step": 82948
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.9741843342781067,
      "learning_rate": 0.00042771927838192067,
      "loss": 2.3204,
      "step": 82949
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.8001550436019897,
      "learning_rate": 0.0004277155770256479,
      "loss": 2.2107,
      "step": 82950
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.784813404083252,
      "learning_rate": 0.0004277118756456303,
      "loss": 1.9652,
      "step": 82951
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.7913693785667419,
      "learning_rate": 0.00042770817424186866,
      "loss": 2.3699,
      "step": 82952
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.8130770325660706,
      "learning_rate": 0.0004277044728143637,
      "loss": 2.6223,
      "step": 82953
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.8503124713897705,
      "learning_rate": 0.0004277007713631161,
      "loss": 2.0941,
      "step": 82954
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.9469459056854248,
      "learning_rate": 0.0004276970698881265,
      "loss": 2.1545,
      "step": 82955
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.8745306730270386,
      "learning_rate": 0.00042769336838939547,
      "loss": 2.2385,
      "step": 82956
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.957491397857666,
      "learning_rate": 0.000427689666866924,
      "loss": 2.3318,
      "step": 82957
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.7788407206535339,
      "learning_rate": 0.0004276859653207126,
      "loss": 2.2735,
      "step": 82958
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.7352367639541626,
      "learning_rate": 0.00042768226375076193,
      "loss": 2.1328,
      "step": 82959
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.8102306127548218,
      "learning_rate": 0.00042767856215707265,
      "loss": 2.121,
      "step": 82960
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.9483465552330017,
      "learning_rate": 0.0004276748605396457,
      "loss": 2.2417,
      "step": 82961
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.8066180348396301,
      "learning_rate": 0.0004276711588984814,
      "loss": 2.3454,
      "step": 82962
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.8075473308563232,
      "learning_rate": 0.00042766745723358065,
      "loss": 2.0455,
      "step": 82963
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.8786664605140686,
      "learning_rate": 0.0004276637555449441,
      "loss": 2.0585,
      "step": 82964
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.770511269569397,
      "learning_rate": 0.0004276600538325726,
      "loss": 1.9347,
      "step": 82965
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.6989232897758484,
      "learning_rate": 0.0004276563520964665,
      "loss": 2.2291,
      "step": 82966
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.8511245250701904,
      "learning_rate": 0.0004276526503366267,
      "loss": 2.1891,
      "step": 82967
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.8562635779380798,
      "learning_rate": 0.00042764894855305396,
      "loss": 2.4576,
      "step": 82968
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.8079405426979065,
      "learning_rate": 0.0004276452467457488,
      "loss": 2.0982,
      "step": 82969
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.8402947783470154,
      "learning_rate": 0.0004276415449147121,
      "loss": 2.0098,
      "step": 82970
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.8136887550354004,
      "learning_rate": 0.00042763784305994426,
      "loss": 2.3974,
      "step": 82971
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.8598489165306091,
      "learning_rate": 0.00042763414118144623,
      "loss": 2.2517,
      "step": 82972
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.8362423181533813,
      "learning_rate": 0.00042763043927921855,
      "loss": 1.9908,
      "step": 82973
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.7762818336486816,
      "learning_rate": 0.00042762673735326196,
      "loss": 2.096,
      "step": 82974
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.769699215888977,
      "learning_rate": 0.00042762303540357723,
      "loss": 2.0159,
      "step": 82975
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.8689225316047668,
      "learning_rate": 0.0004276193334301649,
      "loss": 1.9816,
      "step": 82976
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.7776217460632324,
      "learning_rate": 0.0004276156314330257,
      "loss": 2.0547,
      "step": 82977
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.9424702525138855,
      "learning_rate": 0.00042761192941216046,
      "loss": 2.2892,
      "step": 82978
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.867112934589386,
      "learning_rate": 0.00042760822736756965,
      "loss": 2.0517,
      "step": 82979
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.7720557451248169,
      "learning_rate": 0.0004276045252992542,
      "loss": 2.0128,
      "step": 82980
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.7640965580940247,
      "learning_rate": 0.00042760082320721457,
      "loss": 2.2161,
      "step": 82981
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.8396751284599304,
      "learning_rate": 0.0004275971210914515,
      "loss": 2.1475,
      "step": 82982
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.98895263671875,
      "learning_rate": 0.0004275934189519657,
      "loss": 2.3105,
      "step": 82983
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.8898108005523682,
      "learning_rate": 0.00042758971678875793,
      "loss": 2.2614,
      "step": 82984
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.912912905216217,
      "learning_rate": 0.00042758601460182884,
      "loss": 2.2389,
      "step": 82985
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.8630052804946899,
      "learning_rate": 0.00042758231239117903,
      "loss": 2.2505,
      "step": 82986
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.8666058778762817,
      "learning_rate": 0.0004275786101568094,
      "loss": 2.2018,
      "step": 82987
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.7900652289390564,
      "learning_rate": 0.00042757490789872035,
      "loss": 2.1169,
      "step": 82988
    },
    {
      "epoch": 1.08,
      "grad_norm": 1.2880836725234985,
      "learning_rate": 0.00042757120561691273,
      "loss": 2.1574,
      "step": 82989
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.7342014312744141,
      "learning_rate": 0.0004275675033113873,
      "loss": 1.8819,
      "step": 82990
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.8337486386299133,
      "learning_rate": 0.0004275638009821446,
      "loss": 2.2186,
      "step": 82991
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.8022533655166626,
      "learning_rate": 0.0004275600986291854,
      "loss": 2.3627,
      "step": 82992
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.9239524006843567,
      "learning_rate": 0.00042755639625251035,
      "loss": 2.2857,
      "step": 82993
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.8291552066802979,
      "learning_rate": 0.0004275526938521201,
      "loss": 2.2456,
      "step": 82994
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.7347219586372375,
      "learning_rate": 0.0004275489914280155,
      "loss": 1.997,
      "step": 82995
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.815020740032196,
      "learning_rate": 0.0004275452889801972,
      "loss": 2.3201,
      "step": 82996
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.9304067492485046,
      "learning_rate": 0.00042754158650866567,
      "loss": 2.3258,
      "step": 82997
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.9302186965942383,
      "learning_rate": 0.00042753788401342184,
      "loss": 2.4244,
      "step": 82998
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.8375471830368042,
      "learning_rate": 0.00042753418149446624,
      "loss": 2.178,
      "step": 82999
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.9301197528839111,
      "learning_rate": 0.00042753047895179974,
      "loss": 2.2773,
      "step": 83000
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.8810722231864929,
      "learning_rate": 0.0004275267763854228,
      "loss": 2.355,
      "step": 83001
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.825039267539978,
      "learning_rate": 0.0004275230737953363,
      "loss": 2.4455,
      "step": 83002
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.821807324886322,
      "learning_rate": 0.0004275193711815408,
      "loss": 2.0445,
      "step": 83003
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.922161340713501,
      "learning_rate": 0.0004275156685440371,
      "loss": 2.1379,
      "step": 83004
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.823379635810852,
      "learning_rate": 0.0004275119658828258,
      "loss": 2.0262,
      "step": 83005
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.804305911064148,
      "learning_rate": 0.0004275082631979076,
      "loss": 2.2477,
      "step": 83006
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.9176473617553711,
      "learning_rate": 0.0004275045604892832,
      "loss": 2.0788,
      "step": 83007
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.7371008396148682,
      "learning_rate": 0.0004275008577569534,
      "loss": 2.1781,
      "step": 83008
    },
    {
      "epoch": 1.08,
      "grad_norm": 1.0209259986877441,
      "learning_rate": 0.00042749715500091867,
      "loss": 2.3237,
      "step": 83009
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.7793135643005371,
      "learning_rate": 0.0004274934522211798,
      "loss": 2.0482,
      "step": 83010
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.7885080575942993,
      "learning_rate": 0.0004274897494177376,
      "loss": 2.1138,
      "step": 83011
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.8745207190513611,
      "learning_rate": 0.00042748604659059253,
      "loss": 2.0369,
      "step": 83012
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.8380830883979797,
      "learning_rate": 0.00042748234373974546,
      "loss": 1.8274,
      "step": 83013
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.8251351714134216,
      "learning_rate": 0.000427478640865197,
      "loss": 2.0797,
      "step": 83014
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.8433970808982849,
      "learning_rate": 0.0004274749379669478,
      "loss": 2.1578,
      "step": 83015
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.8445072174072266,
      "learning_rate": 0.0004274712350449987,
      "loss": 2.1819,
      "step": 83016
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.8751241564750671,
      "learning_rate": 0.00042746753209935026,
      "loss": 2.2399,
      "step": 83017
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.9035138487815857,
      "learning_rate": 0.00042746382913000326,
      "loss": 2.2879,
      "step": 83018
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.81609708070755,
      "learning_rate": 0.0004274601261369582,
      "loss": 2.0818,
      "step": 83019
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.7919869422912598,
      "learning_rate": 0.000427456423120216,
      "loss": 2.4938,
      "step": 83020
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.8476248979568481,
      "learning_rate": 0.0004274527200797772,
      "loss": 1.9944,
      "step": 83021
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.8557029962539673,
      "learning_rate": 0.00042744901701564247,
      "loss": 2.1129,
      "step": 83022
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.7366660833358765,
      "learning_rate": 0.0004274453139278127,
      "loss": 2.1899,
      "step": 83023
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.8472543358802795,
      "learning_rate": 0.0004274416108162883,
      "loss": 2.1261,
      "step": 83024
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.8392813801765442,
      "learning_rate": 0.0004274379076810702,
      "loss": 2.2967,
      "step": 83025
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.7956491112709045,
      "learning_rate": 0.000427434204522159,
      "loss": 2.0984,
      "step": 83026
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.867580235004425,
      "learning_rate": 0.00042743050133955534,
      "loss": 2.2212,
      "step": 83027
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.7992686629295349,
      "learning_rate": 0.0004274267981332599,
      "loss": 2.2557,
      "step": 83028
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.7914687395095825,
      "learning_rate": 0.0004274230949032735,
      "loss": 2.2161,
      "step": 83029
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.8368154764175415,
      "learning_rate": 0.00042741939164959666,
      "loss": 2.2528,
      "step": 83030
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.8388027548789978,
      "learning_rate": 0.00042741568837223015,
      "loss": 2.3427,
      "step": 83031
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.8036585450172424,
      "learning_rate": 0.00042741198507117474,
      "loss": 2.4403,
      "step": 83032
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.8427576422691345,
      "learning_rate": 0.000427408281746431,
      "loss": 2.0852,
      "step": 83033
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.7431004643440247,
      "learning_rate": 0.00042740457839799957,
      "loss": 2.2653,
      "step": 83034
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.800957202911377,
      "learning_rate": 0.0004274008750258813,
      "loss": 2.2949,
      "step": 83035
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.8393922448158264,
      "learning_rate": 0.00042739717163007684,
      "loss": 2.0254,
      "step": 83036
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.8782541751861572,
      "learning_rate": 0.00042739346821058676,
      "loss": 2.1834,
      "step": 83037
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.767649233341217,
      "learning_rate": 0.00042738976476741186,
      "loss": 2.2139,
      "step": 83038
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.8395994901657104,
      "learning_rate": 0.0004273860613005529,
      "loss": 2.1973,
      "step": 83039
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.8411993980407715,
      "learning_rate": 0.00042738235781001027,
      "loss": 2.228,
      "step": 83040
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.8762807250022888,
      "learning_rate": 0.000427378654295785,
      "loss": 2.2545,
      "step": 83041
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.8253404498100281,
      "learning_rate": 0.00042737495075787765,
      "loss": 2.5446,
      "step": 83042
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.8156188130378723,
      "learning_rate": 0.0004273712471962888,
      "loss": 1.9185,
      "step": 83043
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.870256245136261,
      "learning_rate": 0.0004273675436110192,
      "loss": 2.1514,
      "step": 83044
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.9096505641937256,
      "learning_rate": 0.00042736384000206963,
      "loss": 2.1027,
      "step": 83045
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.8612614870071411,
      "learning_rate": 0.00042736013636944075,
      "loss": 2.28,
      "step": 83046
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.8107912540435791,
      "learning_rate": 0.0004273564327131332,
      "loss": 2.1858,
      "step": 83047
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.8142676949501038,
      "learning_rate": 0.00042735272903314764,
      "loss": 2.0502,
      "step": 83048
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.8900663256645203,
      "learning_rate": 0.0004273490253294848,
      "loss": 2.2471,
      "step": 83049
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.877078652381897,
      "learning_rate": 0.00042734532160214545,
      "loss": 2.3417,
      "step": 83050
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.8032280206680298,
      "learning_rate": 0.0004273416178511302,
      "loss": 2.268,
      "step": 83051
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.8704704642295837,
      "learning_rate": 0.00042733791407643967,
      "loss": 2.1634,
      "step": 83052
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.8675419092178345,
      "learning_rate": 0.0004273342102780747,
      "loss": 2.0325,
      "step": 83053
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.859560489654541,
      "learning_rate": 0.00042733050645603576,
      "loss": 2.1154,
      "step": 83054
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.9353002309799194,
      "learning_rate": 0.00042732680261032375,
      "loss": 2.1072,
      "step": 83055
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.8572162389755249,
      "learning_rate": 0.0004273230987409393,
      "loss": 2.2052,
      "step": 83056
    },
    {
      "epoch": 1.08,
      "grad_norm": 1.1450984477996826,
      "learning_rate": 0.00042731939484788313,
      "loss": 2.1642,
      "step": 83057
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.937583327293396,
      "learning_rate": 0.00042731569093115577,
      "loss": 1.997,
      "step": 83058
    },
    {
      "epoch": 1.08,
      "grad_norm": 1.0670770406723022,
      "learning_rate": 0.00042731198699075805,
      "loss": 2.0333,
      "step": 83059
    },
    {
      "epoch": 1.08,
      "grad_norm": 1.012689471244812,
      "learning_rate": 0.00042730828302669073,
      "loss": 2.083,
      "step": 83060
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.8992218971252441,
      "learning_rate": 0.00042730457903895436,
      "loss": 1.9917,
      "step": 83061
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.9042015075683594,
      "learning_rate": 0.00042730087502754953,
      "loss": 2.2015,
      "step": 83062
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.8327747583389282,
      "learning_rate": 0.0004272971709924772,
      "loss": 2.3586,
      "step": 83063
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.8229473233222961,
      "learning_rate": 0.0004272934669337379,
      "loss": 2.2644,
      "step": 83064
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.8241201043128967,
      "learning_rate": 0.0004272897628513323,
      "loss": 2.2201,
      "step": 83065
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.8506568670272827,
      "learning_rate": 0.00042728605874526123,
      "loss": 2.0961,
      "step": 83066
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.8260433673858643,
      "learning_rate": 0.00042728235461552517,
      "loss": 2.1819,
      "step": 83067
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.8645735383033752,
      "learning_rate": 0.00042727865046212493,
      "loss": 2.3025,
      "step": 83068
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.9430585503578186,
      "learning_rate": 0.00042727494628506127,
      "loss": 2.0886,
      "step": 83069
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.8123787641525269,
      "learning_rate": 0.00042727124208433474,
      "loss": 2.1963,
      "step": 83070
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.8792751431465149,
      "learning_rate": 0.00042726753785994604,
      "loss": 2.1529,
      "step": 83071
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.9233267903327942,
      "learning_rate": 0.000427263833611896,
      "loss": 2.2403,
      "step": 83072
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.8025816082954407,
      "learning_rate": 0.0004272601293401851,
      "loss": 1.9354,
      "step": 83073
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.74285489320755,
      "learning_rate": 0.00042725642504481417,
      "loss": 2.0092,
      "step": 83074
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.9648932218551636,
      "learning_rate": 0.00042725272072578394,
      "loss": 2.2268,
      "step": 83075
    },
    {
      "epoch": 1.08,
      "grad_norm": 1.0285366773605347,
      "learning_rate": 0.00042724901638309496,
      "loss": 2.1328,
      "step": 83076
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.8944585919380188,
      "learning_rate": 0.00042724531201674803,
      "loss": 2.0919,
      "step": 83077
    },
    {
      "epoch": 1.08,
      "grad_norm": 1.1337164640426636,
      "learning_rate": 0.0004272416076267438,
      "loss": 2.2242,
      "step": 83078
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.9081107974052429,
      "learning_rate": 0.00042723790321308285,
      "loss": 2.2524,
      "step": 83079
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.8410394191741943,
      "learning_rate": 0.000427234198775766,
      "loss": 2.3396,
      "step": 83080
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.8263914585113525,
      "learning_rate": 0.000427230494314794,
      "loss": 2.2243,
      "step": 83081
    },
    {
      "epoch": 1.08,
      "grad_norm": 1.0485174655914307,
      "learning_rate": 0.00042722678983016737,
      "loss": 2.2036,
      "step": 83082
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.8244182467460632,
      "learning_rate": 0.000427223085321887,
      "loss": 2.213,
      "step": 83083
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.9682314395904541,
      "learning_rate": 0.0004272193807899534,
      "loss": 2.361,
      "step": 83084
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.8762251138687134,
      "learning_rate": 0.0004272156762343672,
      "loss": 2.2857,
      "step": 83085
    },
    {
      "epoch": 1.08,
      "grad_norm": 1.1328707933425903,
      "learning_rate": 0.0004272119716551293,
      "loss": 2.2777,
      "step": 83086
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.9604665040969849,
      "learning_rate": 0.00042720826705224034,
      "loss": 2.3486,
      "step": 83087
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.7577850818634033,
      "learning_rate": 0.00042720456242570083,
      "loss": 2.3188,
      "step": 83088
    },
    {
      "epoch": 1.08,
      "grad_norm": 1.1327698230743408,
      "learning_rate": 0.0004272008577755117,
      "loss": 2.3504,
      "step": 83089
    },
    {
      "epoch": 1.08,
      "grad_norm": 1.2131099700927734,
      "learning_rate": 0.0004271971531016735,
      "loss": 2.0032,
      "step": 83090
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.8994705080986023,
      "learning_rate": 0.00042719344840418696,
      "loss": 2.1776,
      "step": 83091
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.7802698612213135,
      "learning_rate": 0.00042718974368305266,
      "loss": 1.9557,
      "step": 83092
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.9716203808784485,
      "learning_rate": 0.0004271860389382715,
      "loss": 1.8691,
      "step": 83093
    },
    {
      "epoch": 1.08,
      "grad_norm": 1.1524296998977661,
      "learning_rate": 0.000427182334169844,
      "loss": 2.2035,
      "step": 83094
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.7727179527282715,
      "learning_rate": 0.000427178629377771,
      "loss": 2.0823,
      "step": 83095
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.8003374934196472,
      "learning_rate": 0.000427174924562053,
      "loss": 2.2165,
      "step": 83096
    },
    {
      "epoch": 1.08,
      "grad_norm": 1.080399513244629,
      "learning_rate": 0.0004271712197226908,
      "loss": 2.2094,
      "step": 83097
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.8077199459075928,
      "learning_rate": 0.000427167514859685,
      "loss": 2.0256,
      "step": 83098
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.7691341042518616,
      "learning_rate": 0.0004271638099730365,
      "loss": 2.021,
      "step": 83099
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.8199186325073242,
      "learning_rate": 0.00042716010506274577,
      "loss": 2.2623,
      "step": 83100
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.8436177968978882,
      "learning_rate": 0.00042715640012881353,
      "loss": 2.1367,
      "step": 83101
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.7698572874069214,
      "learning_rate": 0.0004271526951712406,
      "loss": 2.3338,
      "step": 83102
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.7569443583488464,
      "learning_rate": 0.0004271489901900276,
      "loss": 2.1165,
      "step": 83103
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.8891890048980713,
      "learning_rate": 0.00042714528518517504,
      "loss": 2.1312,
      "step": 83104
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.7948278784751892,
      "learning_rate": 0.000427141580156684,
      "loss": 2.0688,
      "step": 83105
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.8327684998512268,
      "learning_rate": 0.00042713787510455475,
      "loss": 2.2832,
      "step": 83106
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.8378968238830566,
      "learning_rate": 0.0004271341700287882,
      "loss": 2.3848,
      "step": 83107
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.9232041239738464,
      "learning_rate": 0.0004271304649293851,
      "loss": 2.1642,
      "step": 83108
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.8474727272987366,
      "learning_rate": 0.000427126759806346,
      "loss": 2.6138,
      "step": 83109
    },
    {
      "epoch": 1.08,
      "grad_norm": 1.0012571811676025,
      "learning_rate": 0.0004271230546596716,
      "loss": 2.2214,
      "step": 83110
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.7947452068328857,
      "learning_rate": 0.00042711934948936264,
      "loss": 2.0189,
      "step": 83111
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.8433468341827393,
      "learning_rate": 0.00042711564429541985,
      "loss": 1.916,
      "step": 83112
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.9836283326148987,
      "learning_rate": 0.0004271119390778438,
      "loss": 2.3202,
      "step": 83113
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.7467065453529358,
      "learning_rate": 0.00042710823383663527,
      "loss": 2.1459,
      "step": 83114
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.8465837240219116,
      "learning_rate": 0.00042710452857179493,
      "loss": 2.4294,
      "step": 83115
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.827774167060852,
      "learning_rate": 0.0004271008232833234,
      "loss": 2.1821,
      "step": 83116
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.9255350232124329,
      "learning_rate": 0.0004270971179712215,
      "loss": 2.2391,
      "step": 83117
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.7851024270057678,
      "learning_rate": 0.0004270934126354898,
      "loss": 2.123,
      "step": 83118
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.7829956412315369,
      "learning_rate": 0.00042708970727612907,
      "loss": 2.1857,
      "step": 83119
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.8107947111129761,
      "learning_rate": 0.0004270860018931399,
      "loss": 1.9097,
      "step": 83120
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.8432010412216187,
      "learning_rate": 0.0004270822964865232,
      "loss": 2.2076,
      "step": 83121
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.7832916378974915,
      "learning_rate": 0.00042707859105627934,
      "loss": 2.1969,
      "step": 83122
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.8179199695587158,
      "learning_rate": 0.00042707488560240917,
      "loss": 2.0602,
      "step": 83123
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.785076379776001,
      "learning_rate": 0.00042707118012491347,
      "loss": 2.3556,
      "step": 83124
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.759033203125,
      "learning_rate": 0.0004270674746237928,
      "loss": 2.1636,
      "step": 83125
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.9708408713340759,
      "learning_rate": 0.0004270637690990479,
      "loss": 2.2881,
      "step": 83126
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.8035533428192139,
      "learning_rate": 0.00042706006355067947,
      "loss": 2.4373,
      "step": 83127
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.7622278928756714,
      "learning_rate": 0.0004270563579786881,
      "loss": 2.2994,
      "step": 83128
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.8429174423217773,
      "learning_rate": 0.0004270526523830745,
      "loss": 2.2309,
      "step": 83129
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.8076223731040955,
      "learning_rate": 0.00042704894676383955,
      "loss": 2.2302,
      "step": 83130
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.7560518383979797,
      "learning_rate": 0.00042704524112098383,
      "loss": 2.0999,
      "step": 83131
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.8473173379898071,
      "learning_rate": 0.0004270415354545079,
      "loss": 2.2891,
      "step": 83132
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.7817150950431824,
      "learning_rate": 0.0004270378297644126,
      "loss": 2.1907,
      "step": 83133
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.8574893474578857,
      "learning_rate": 0.00042703412405069855,
      "loss": 2.2844,
      "step": 83134
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.7733874917030334,
      "learning_rate": 0.00042703041831336643,
      "loss": 2.2465,
      "step": 83135
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.9077336192131042,
      "learning_rate": 0.00042702671255241703,
      "loss": 2.2133,
      "step": 83136
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.8198506236076355,
      "learning_rate": 0.0004270230067678509,
      "loss": 2.1353,
      "step": 83137
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.8803048133850098,
      "learning_rate": 0.0004270193009596688,
      "loss": 2.0872,
      "step": 83138
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.8184368014335632,
      "learning_rate": 0.00042701559512787145,
      "loss": 1.9438,
      "step": 83139
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.9152500629425049,
      "learning_rate": 0.0004270118892724594,
      "loss": 2.0563,
      "step": 83140
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.7991628646850586,
      "learning_rate": 0.00042700818339343357,
      "loss": 2.2355,
      "step": 83141
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.8717115521430969,
      "learning_rate": 0.0004270044774907945,
      "loss": 2.1976,
      "step": 83142
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.8298389911651611,
      "learning_rate": 0.0004270007715645429,
      "loss": 2.1014,
      "step": 83143
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.8256585597991943,
      "learning_rate": 0.0004269970656146794,
      "loss": 2.2937,
      "step": 83144
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.8110708594322205,
      "learning_rate": 0.00042699335964120486,
      "loss": 2.2961,
      "step": 83145
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.8369455933570862,
      "learning_rate": 0.0004269896536441198,
      "loss": 2.0214,
      "step": 83146
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.8316713571548462,
      "learning_rate": 0.0004269859476234249,
      "loss": 2.0714,
      "step": 83147
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.8031242489814758,
      "learning_rate": 0.000426982241579121,
      "loss": 2.1616,
      "step": 83148
    },
    {
      "epoch": 1.08,
      "grad_norm": 1.0353425741195679,
      "learning_rate": 0.0004269785355112087,
      "loss": 2.2632,
      "step": 83149
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.8573499321937561,
      "learning_rate": 0.0004269748294196886,
      "loss": 2.1689,
      "step": 83150
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.7863054871559143,
      "learning_rate": 0.0004269711233045616,
      "loss": 2.0871,
      "step": 83151
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.965709924697876,
      "learning_rate": 0.0004269674171658282,
      "loss": 2.2369,
      "step": 83152
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.93963223695755,
      "learning_rate": 0.00042696371100348914,
      "loss": 2.2369,
      "step": 83153
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.8095775842666626,
      "learning_rate": 0.00042696000481754517,
      "loss": 2.2073,
      "step": 83154
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.8614106774330139,
      "learning_rate": 0.00042695629860799693,
      "loss": 2.0143,
      "step": 83155
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.8575462102890015,
      "learning_rate": 0.00042695259237484517,
      "loss": 2.0788,
      "step": 83156
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.7964070439338684,
      "learning_rate": 0.00042694888611809054,
      "loss": 2.0664,
      "step": 83157
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.77725750207901,
      "learning_rate": 0.0004269451798377336,
      "loss": 2.1123,
      "step": 83158
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.8687844276428223,
      "learning_rate": 0.00042694147353377514,
      "loss": 2.1661,
      "step": 83159
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.9197095036506653,
      "learning_rate": 0.00042693776720621605,
      "loss": 2.23,
      "step": 83160
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.7985687851905823,
      "learning_rate": 0.00042693406085505663,
      "loss": 2.1366,
      "step": 83161
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.7803175449371338,
      "learning_rate": 0.0004269303544802979,
      "loss": 2.1814,
      "step": 83162
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.9066060781478882,
      "learning_rate": 0.0004269266480819404,
      "loss": 2.0262,
      "step": 83163
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.8663623332977295,
      "learning_rate": 0.00042692294165998484,
      "loss": 2.0956,
      "step": 83164
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.8338717818260193,
      "learning_rate": 0.00042691923521443186,
      "loss": 2.2866,
      "step": 83165
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.9439170360565186,
      "learning_rate": 0.0004269155287452823,
      "loss": 2.1064,
      "step": 83166
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.7992067337036133,
      "learning_rate": 0.00042691182225253665,
      "loss": 2.0556,
      "step": 83167
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.7477944493293762,
      "learning_rate": 0.00042690811573619573,
      "loss": 2.0413,
      "step": 83168
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.7729009389877319,
      "learning_rate": 0.0004269044091962602,
      "loss": 2.1584,
      "step": 83169
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.7584773898124695,
      "learning_rate": 0.0004269007026327307,
      "loss": 2.0343,
      "step": 83170
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.8409059643745422,
      "learning_rate": 0.00042689699604560793,
      "loss": 2.2518,
      "step": 83171
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.8505706787109375,
      "learning_rate": 0.0004268932894348928,
      "loss": 2.3831,
      "step": 83172
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.9004558324813843,
      "learning_rate": 0.00042688958280058564,
      "loss": 2.2431,
      "step": 83173
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.7669903635978699,
      "learning_rate": 0.0004268858761426873,
      "loss": 2.2652,
      "step": 83174
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.8854560852050781,
      "learning_rate": 0.0004268821694611986,
      "loss": 2.0652,
      "step": 83175
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.8248500227928162,
      "learning_rate": 0.00042687846275612005,
      "loss": 2.1667,
      "step": 83176
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.74341881275177,
      "learning_rate": 0.00042687475602745244,
      "loss": 2.013,
      "step": 83177
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.8036757111549377,
      "learning_rate": 0.0004268710492751964,
      "loss": 2.011,
      "step": 83178
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.878281831741333,
      "learning_rate": 0.00042686734249935265,
      "loss": 2.2238,
      "step": 83179
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.971149206161499,
      "learning_rate": 0.0004268636356999218,
      "loss": 2.3212,
      "step": 83180
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.8489795327186584,
      "learning_rate": 0.00042685992887690457,
      "loss": 2.2075,
      "step": 83181
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.8597932457923889,
      "learning_rate": 0.00042685622203030184,
      "loss": 2.1929,
      "step": 83182
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.8033298254013062,
      "learning_rate": 0.00042685251516011403,
      "loss": 2.1034,
      "step": 83183
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.8952034711837769,
      "learning_rate": 0.00042684880826634195,
      "loss": 2.0454,
      "step": 83184
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.7873038649559021,
      "learning_rate": 0.00042684510134898636,
      "loss": 2.3861,
      "step": 83185
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.9686381220817566,
      "learning_rate": 0.0004268413944080478,
      "loss": 2.0968,
      "step": 83186
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.7418177723884583,
      "learning_rate": 0.000426837687443527,
      "loss": 2.1107,
      "step": 83187
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.7913339138031006,
      "learning_rate": 0.0004268339804554248,
      "loss": 2.1339,
      "step": 83188
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.7917969226837158,
      "learning_rate": 0.0004268302734437416,
      "loss": 2.1454,
      "step": 83189
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.8037608861923218,
      "learning_rate": 0.0004268265664084784,
      "loss": 2.1713,
      "step": 83190
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.8405866622924805,
      "learning_rate": 0.0004268228593496357,
      "loss": 2.2526,
      "step": 83191
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.7804728746414185,
      "learning_rate": 0.0004268191522672143,
      "loss": 2.127,
      "step": 83192
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.8190274834632874,
      "learning_rate": 0.00042681544516121473,
      "loss": 2.3552,
      "step": 83193
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.8591543436050415,
      "learning_rate": 0.0004268117380316378,
      "loss": 2.137,
      "step": 83194
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.838570237159729,
      "learning_rate": 0.00042680803087848416,
      "loss": 2.1914,
      "step": 83195
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.7816082835197449,
      "learning_rate": 0.00042680432370175456,
      "loss": 2.0749,
      "step": 83196
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.774398684501648,
      "learning_rate": 0.00042680061650144963,
      "loss": 2.282,
      "step": 83197
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.8044715523719788,
      "learning_rate": 0.00042679690927757,
      "loss": 2.0862,
      "step": 83198
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.8238192200660706,
      "learning_rate": 0.0004267932020301165,
      "loss": 2.1257,
      "step": 83199
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.7255069613456726,
      "learning_rate": 0.0004267894947590897,
      "loss": 2.0418,
      "step": 83200
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.8281576037406921,
      "learning_rate": 0.0004267857874644904,
      "loss": 2.1474,
      "step": 83201
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.8537935018539429,
      "learning_rate": 0.0004267820801463193,
      "loss": 2.2475,
      "step": 83202
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.8045050501823425,
      "learning_rate": 0.0004267783728045769,
      "loss": 1.991,
      "step": 83203
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.7976970076560974,
      "learning_rate": 0.00042677466543926407,
      "loss": 2.3479,
      "step": 83204
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.7737144231796265,
      "learning_rate": 0.00042677095805038136,
      "loss": 2.0532,
      "step": 83205
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.9114616513252258,
      "learning_rate": 0.00042676725063792956,
      "loss": 2.0541,
      "step": 83206
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.8219606876373291,
      "learning_rate": 0.0004267635432019093,
      "loss": 2.1907,
      "step": 83207
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.8577962517738342,
      "learning_rate": 0.0004267598357423214,
      "loss": 2.2259,
      "step": 83208
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.9117106795310974,
      "learning_rate": 0.00042675612825916645,
      "loss": 2.0126,
      "step": 83209
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.7620319724082947,
      "learning_rate": 0.0004267524207524451,
      "loss": 2.1459,
      "step": 83210
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.7680783867835999,
      "learning_rate": 0.00042674871322215805,
      "loss": 2.2594,
      "step": 83211
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.9203633666038513,
      "learning_rate": 0.0004267450056683061,
      "loss": 2.3255,
      "step": 83212
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.7858957052230835,
      "learning_rate": 0.0004267412980908898,
      "loss": 1.9971,
      "step": 83213
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.8439932465553284,
      "learning_rate": 0.0004267375904899099,
      "loss": 2.2378,
      "step": 83214
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.8853378891944885,
      "learning_rate": 0.0004267338828653672,
      "loss": 2.2197,
      "step": 83215
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.7896004319190979,
      "learning_rate": 0.0004267301752172621,
      "loss": 2.4437,
      "step": 83216
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.7606070041656494,
      "learning_rate": 0.00042672646754559555,
      "loss": 2.0198,
      "step": 83217
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.7770388722419739,
      "learning_rate": 0.00042672275985036816,
      "loss": 2.0272,
      "step": 83218
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.7954407930374146,
      "learning_rate": 0.0004267190521315806,
      "loss": 2.047,
      "step": 83219
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.8734040260314941,
      "learning_rate": 0.00042671534438923364,
      "loss": 2.0902,
      "step": 83220
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.7838078141212463,
      "learning_rate": 0.00042671163662332787,
      "loss": 2.1777,
      "step": 83221
    },
    {
      "epoch": 1.08,
      "grad_norm": 1.2235491275787354,
      "learning_rate": 0.00042670792883386395,
      "loss": 2.1608,
      "step": 83222
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.8478034734725952,
      "learning_rate": 0.0004267042210208427,
      "loss": 2.3385,
      "step": 83223
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.8201271295547485,
      "learning_rate": 0.00042670051318426474,
      "loss": 2.3538,
      "step": 83224
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.8986556529998779,
      "learning_rate": 0.00042669680532413074,
      "loss": 2.197,
      "step": 83225
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.8761898279190063,
      "learning_rate": 0.0004266930974404415,
      "loss": 2.3501,
      "step": 83226
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.8652514815330505,
      "learning_rate": 0.0004266893895331975,
      "loss": 2.2408,
      "step": 83227
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.8352140784263611,
      "learning_rate": 0.00042668568160239963,
      "loss": 2.0945,
      "step": 83228
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.8401477932929993,
      "learning_rate": 0.00042668197364804846,
      "loss": 2.089,
      "step": 83229
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.8307130932807922,
      "learning_rate": 0.00042667826567014473,
      "loss": 2.013,
      "step": 83230
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.8432080149650574,
      "learning_rate": 0.0004266745576686891,
      "loss": 2.3313,
      "step": 83231
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.8386970162391663,
      "learning_rate": 0.00042667084964368227,
      "loss": 2.2414,
      "step": 83232
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.9263512492179871,
      "learning_rate": 0.000426667141595125,
      "loss": 2.0138,
      "step": 83233
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.8332022428512573,
      "learning_rate": 0.0004266634335230179,
      "loss": 2.066,
      "step": 83234
    },
    {
      "epoch": 1.08,
      "grad_norm": 1.0744088888168335,
      "learning_rate": 0.00042665972542736165,
      "loss": 2.179,
      "step": 83235
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.8992289900779724,
      "learning_rate": 0.000426656017308157,
      "loss": 2.2762,
      "step": 83236
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.8780663013458252,
      "learning_rate": 0.00042665230916540456,
      "loss": 1.9564,
      "step": 83237
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.746310293674469,
      "learning_rate": 0.0004266486009991051,
      "loss": 2.2908,
      "step": 83238
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.8611756563186646,
      "learning_rate": 0.0004266448928092593,
      "loss": 2.2222,
      "step": 83239
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.8381271362304688,
      "learning_rate": 0.00042664118459586774,
      "loss": 2.142,
      "step": 83240
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.7955859899520874,
      "learning_rate": 0.0004266374763589312,
      "loss": 2.2225,
      "step": 83241
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.8136453628540039,
      "learning_rate": 0.0004266337680984504,
      "loss": 2.3084,
      "step": 83242
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.7841350436210632,
      "learning_rate": 0.00042663005981442604,
      "loss": 2.1608,
      "step": 83243
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.8727126717567444,
      "learning_rate": 0.0004266263515068587,
      "loss": 2.0732,
      "step": 83244
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.7781760692596436,
      "learning_rate": 0.00042662264317574913,
      "loss": 2.2598,
      "step": 83245
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.8207905292510986,
      "learning_rate": 0.0004266189348210981,
      "loss": 2.1688,
      "step": 83246
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.9170459508895874,
      "learning_rate": 0.00042661522644290613,
      "loss": 1.9605,
      "step": 83247
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.8023663759231567,
      "learning_rate": 0.00042661151804117406,
      "loss": 2.0498,
      "step": 83248
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.7531132102012634,
      "learning_rate": 0.0004266078096159025,
      "loss": 2.0828,
      "step": 83249
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.7899728417396545,
      "learning_rate": 0.0004266041011670921,
      "loss": 2.1849,
      "step": 83250
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.8690246343612671,
      "learning_rate": 0.00042660039269474365,
      "loss": 2.2307,
      "step": 83251
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.773565948009491,
      "learning_rate": 0.00042659668419885784,
      "loss": 2.0441,
      "step": 83252
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.862879753112793,
      "learning_rate": 0.00042659297567943525,
      "loss": 2.299,
      "step": 83253
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.893818736076355,
      "learning_rate": 0.0004265892671364766,
      "loss": 2.0946,
      "step": 83254
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.7417308688163757,
      "learning_rate": 0.0004265855585699828,
      "loss": 1.9703,
      "step": 83255
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.922814130783081,
      "learning_rate": 0.0004265818499799542,
      "loss": 2.188,
      "step": 83256
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.7882552742958069,
      "learning_rate": 0.0004265781413663917,
      "loss": 1.9567,
      "step": 83257
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.7855504155158997,
      "learning_rate": 0.0004265744327292959,
      "loss": 2.2432,
      "step": 83258
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.7766531109809875,
      "learning_rate": 0.0004265707240686675,
      "loss": 2.0739,
      "step": 83259
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.8575000762939453,
      "learning_rate": 0.0004265670153845072,
      "loss": 2.1315,
      "step": 83260
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.8035487532615662,
      "learning_rate": 0.00042656330667681586,
      "loss": 2.3969,
      "step": 83261
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.7627915740013123,
      "learning_rate": 0.00042655959794559383,
      "loss": 2.204,
      "step": 83262
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.9011284112930298,
      "learning_rate": 0.00042655588919084206,
      "loss": 2.0593,
      "step": 83263
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.8487861752510071,
      "learning_rate": 0.0004265521804125612,
      "loss": 2.0341,
      "step": 83264
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.7536202073097229,
      "learning_rate": 0.00042654847161075187,
      "loss": 2.0072,
      "step": 83265
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.8862277269363403,
      "learning_rate": 0.00042654476278541475,
      "loss": 2.2287,
      "step": 83266
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.8147315382957458,
      "learning_rate": 0.0004265410539365506,
      "loss": 2.0317,
      "step": 83267
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.825425922870636,
      "learning_rate": 0.00042653734506416,
      "loss": 2.2185,
      "step": 83268
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.8815730214118958,
      "learning_rate": 0.00042653363616824383,
      "loss": 2.043,
      "step": 83269
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.8244293332099915,
      "learning_rate": 0.0004265299272488027,
      "loss": 2.2877,
      "step": 83270
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.7576600909233093,
      "learning_rate": 0.0004265262183058372,
      "loss": 1.9399,
      "step": 83271
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.8463140726089478,
      "learning_rate": 0.0004265225093393481,
      "loss": 2.041,
      "step": 83272
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.8672922253608704,
      "learning_rate": 0.0004265188003493361,
      "loss": 2.2145,
      "step": 83273
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.8806546330451965,
      "learning_rate": 0.00042651509133580175,
      "loss": 2.2886,
      "step": 83274
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.8590111136436462,
      "learning_rate": 0.00042651138229874596,
      "loss": 2.2709,
      "step": 83275
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.8490015268325806,
      "learning_rate": 0.00042650767323816933,
      "loss": 2.235,
      "step": 83276
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.7731903791427612,
      "learning_rate": 0.0004265039641540725,
      "loss": 2.2004,
      "step": 83277
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.9421100616455078,
      "learning_rate": 0.0004265002550464562,
      "loss": 2.45,
      "step": 83278
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.8688245415687561,
      "learning_rate": 0.0004264965459153211,
      "loss": 2.3446,
      "step": 83279
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.8761224746704102,
      "learning_rate": 0.0004264928367606679,
      "loss": 2.2125,
      "step": 83280
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.8813487887382507,
      "learning_rate": 0.0004264891275824973,
      "loss": 2.3157,
      "step": 83281
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.794880211353302,
      "learning_rate": 0.00042648541838081,
      "loss": 2.5588,
      "step": 83282
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.8676223158836365,
      "learning_rate": 0.00042648170915560667,
      "loss": 2.1346,
      "step": 83283
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.8049488067626953,
      "learning_rate": 0.00042647799990688795,
      "loss": 2.0353,
      "step": 83284
    },
    {
      "epoch": 1.08,
      "grad_norm": 1.1101880073547363,
      "learning_rate": 0.0004264742906346547,
      "loss": 2.0575,
      "step": 83285
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.8494236469268799,
      "learning_rate": 0.00042647058133890743,
      "loss": 2.1689,
      "step": 83286
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.9485500454902649,
      "learning_rate": 0.0004264668720196469,
      "loss": 2.229,
      "step": 83287
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.958796501159668,
      "learning_rate": 0.00042646316267687373,
      "loss": 2.1517,
      "step": 83288
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.7531982660293579,
      "learning_rate": 0.00042645945331058874,
      "loss": 2.121,
      "step": 83289
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.8993419408798218,
      "learning_rate": 0.00042645574392079254,
      "loss": 1.9049,
      "step": 83290
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.9979603290557861,
      "learning_rate": 0.00042645203450748586,
      "loss": 1.9725,
      "step": 83291
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.8205763101577759,
      "learning_rate": 0.0004264483250706693,
      "loss": 2.2181,
      "step": 83292
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.8390854597091675,
      "learning_rate": 0.00042644461561034357,
      "loss": 2.1555,
      "step": 83293
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.9509071707725525,
      "learning_rate": 0.00042644090612650954,
      "loss": 2.3047,
      "step": 83294
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.8189653158187866,
      "learning_rate": 0.0004264371966191676,
      "loss": 2.3318,
      "step": 83295
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.7235190272331238,
      "learning_rate": 0.0004264334870883187,
      "loss": 2.0991,
      "step": 83296
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.8604058623313904,
      "learning_rate": 0.0004264297775339634,
      "loss": 2.233,
      "step": 83297
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.9946480393409729,
      "learning_rate": 0.00042642606795610243,
      "loss": 2.2291,
      "step": 83298
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.8842012882232666,
      "learning_rate": 0.00042642235835473643,
      "loss": 2.0777,
      "step": 83299
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.7463656663894653,
      "learning_rate": 0.00042641864872986624,
      "loss": 2.1217,
      "step": 83300
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.9397607445716858,
      "learning_rate": 0.00042641493908149227,
      "loss": 2.1645,
      "step": 83301
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.8982421159744263,
      "learning_rate": 0.00042641122940961546,
      "loss": 2.2684,
      "step": 83302
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.8982202410697937,
      "learning_rate": 0.0004264075197142364,
      "loss": 2.2872,
      "step": 83303
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.776573121547699,
      "learning_rate": 0.00042640380999535584,
      "loss": 2.1984,
      "step": 83304
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.8183141350746155,
      "learning_rate": 0.00042640010025297444,
      "loss": 2.2557,
      "step": 83305
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.8278115391731262,
      "learning_rate": 0.0004263963904870928,
      "loss": 2.3106,
      "step": 83306
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.979594349861145,
      "learning_rate": 0.00042639268069771173,
      "loss": 1.9951,
      "step": 83307
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.8093674778938293,
      "learning_rate": 0.0004263889708848318,
      "loss": 2.1736,
      "step": 83308
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.8288705348968506,
      "learning_rate": 0.00042638526104845384,
      "loss": 2.336,
      "step": 83309
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.8695972561836243,
      "learning_rate": 0.0004263815511885785,
      "loss": 2.3215,
      "step": 83310
    },
    {
      "epoch": 1.08,
      "grad_norm": 1.0001181364059448,
      "learning_rate": 0.00042637784130520646,
      "loss": 2.0813,
      "step": 83311
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.8138135075569153,
      "learning_rate": 0.0004263741313983383,
      "loss": 2.1144,
      "step": 83312
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.9076941609382629,
      "learning_rate": 0.0004263704214679748,
      "loss": 2.1817,
      "step": 83313
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.8920762538909912,
      "learning_rate": 0.00042636671151411674,
      "loss": 2.2518,
      "step": 83314
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.8037887811660767,
      "learning_rate": 0.0004263630015367647,
      "loss": 2.1081,
      "step": 83315
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.8305554986000061,
      "learning_rate": 0.0004263592915359194,
      "loss": 2.0662,
      "step": 83316
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.7906289100646973,
      "learning_rate": 0.0004263555815115814,
      "loss": 2.0271,
      "step": 83317
    },
    {
      "epoch": 1.08,
      "grad_norm": 1.0072948932647705,
      "learning_rate": 0.0004263518714637517,
      "loss": 2.2988,
      "step": 83318
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.9021144509315491,
      "learning_rate": 0.00042634816139243073,
      "loss": 2.2194,
      "step": 83319
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.8504799008369446,
      "learning_rate": 0.0004263444512976192,
      "loss": 2.2959,
      "step": 83320
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.8184574246406555,
      "learning_rate": 0.0004263407411793179,
      "loss": 2.3737,
      "step": 83321
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.8116768598556519,
      "learning_rate": 0.0004263370310375275,
      "loss": 2.221,
      "step": 83322
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.9853035807609558,
      "learning_rate": 0.00042633332087224857,
      "loss": 2.1075,
      "step": 83323
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.7715356945991516,
      "learning_rate": 0.00042632961068348195,
      "loss": 2.2967,
      "step": 83324
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.8906883001327515,
      "learning_rate": 0.0004263259004712283,
      "loss": 2.0812,
      "step": 83325
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.8269245028495789,
      "learning_rate": 0.0004263221902354882,
      "loss": 2.189,
      "step": 83326
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.8963453769683838,
      "learning_rate": 0.0004263184799762625,
      "loss": 2.2273,
      "step": 83327
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.773855447769165,
      "learning_rate": 0.00042631476969355174,
      "loss": 2.1134,
      "step": 83328
    },
    {
      "epoch": 1.08,
      "grad_norm": 0.8151702880859375,
      "learning_rate": 0.00042631105938735676,
      "loss": 2.0771,
      "step": 83329
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.8022574186325073,
      "learning_rate": 0.00042630734905767806,
      "loss": 2.1824,
      "step": 83330
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.7999042272567749,
      "learning_rate": 0.0004263036387045165,
      "loss": 2.3482,
      "step": 83331
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.8104643821716309,
      "learning_rate": 0.00042629992832787274,
      "loss": 1.9716,
      "step": 83332
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.8630406260490417,
      "learning_rate": 0.0004262962179277474,
      "loss": 2.1115,
      "step": 83333
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.9155972003936768,
      "learning_rate": 0.0004262925075041413,
      "loss": 2.1847,
      "step": 83334
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.9544374942779541,
      "learning_rate": 0.00042628879705705494,
      "loss": 2.2506,
      "step": 83335
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.864719569683075,
      "learning_rate": 0.0004262850865864891,
      "loss": 2.31,
      "step": 83336
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.907846987247467,
      "learning_rate": 0.00042628137609244455,
      "loss": 2.4142,
      "step": 83337
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.8226677179336548,
      "learning_rate": 0.00042627766557492186,
      "loss": 2.1749,
      "step": 83338
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.845468282699585,
      "learning_rate": 0.00042627395503392175,
      "loss": 2.0154,
      "step": 83339
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.8984400033950806,
      "learning_rate": 0.000426270244469445,
      "loss": 2.0705,
      "step": 83340
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.8562102317810059,
      "learning_rate": 0.00042626653388149214,
      "loss": 2.0924,
      "step": 83341
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.7787016034126282,
      "learning_rate": 0.000426262823270064,
      "loss": 2.0965,
      "step": 83342
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.9621025323867798,
      "learning_rate": 0.00042625911263516125,
      "loss": 1.9683,
      "step": 83343
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.8139026761054993,
      "learning_rate": 0.0004262554019767845,
      "loss": 2.1766,
      "step": 83344
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.7840902805328369,
      "learning_rate": 0.00042625169129493443,
      "loss": 2.2852,
      "step": 83345
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.8669999241828918,
      "learning_rate": 0.0004262479805896119,
      "loss": 2.1927,
      "step": 83346
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.8067463040351868,
      "learning_rate": 0.00042624426986081744,
      "loss": 1.9417,
      "step": 83347
    },
    {
      "epoch": 1.09,
      "grad_norm": 1.2881660461425781,
      "learning_rate": 0.00042624055910855176,
      "loss": 2.245,
      "step": 83348
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.7928033471107483,
      "learning_rate": 0.0004262368483328156,
      "loss": 2.1099,
      "step": 83349
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.72789067029953,
      "learning_rate": 0.00042623313753360965,
      "loss": 2.191,
      "step": 83350
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.9396563768386841,
      "learning_rate": 0.0004262294267109345,
      "loss": 1.9957,
      "step": 83351
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.8316519260406494,
      "learning_rate": 0.000426225715864791,
      "loss": 2.2432,
      "step": 83352
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.7860121130943298,
      "learning_rate": 0.00042622200499517975,
      "loss": 1.8914,
      "step": 83353
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.8906605243682861,
      "learning_rate": 0.00042621829410210136,
      "loss": 1.918,
      "step": 83354
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.8861583471298218,
      "learning_rate": 0.00042621458318555677,
      "loss": 2.4747,
      "step": 83355
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.7827578783035278,
      "learning_rate": 0.00042621087224554635,
      "loss": 2.1509,
      "step": 83356
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.7846828103065491,
      "learning_rate": 0.000426207161282071,
      "loss": 2.0468,
      "step": 83357
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.9057732820510864,
      "learning_rate": 0.00042620345029513137,
      "loss": 2.0263,
      "step": 83358
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.7381384968757629,
      "learning_rate": 0.00042619973928472815,
      "loss": 2.1978,
      "step": 83359
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.8199214935302734,
      "learning_rate": 0.00042619602825086193,
      "loss": 2.0556,
      "step": 83360
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.8261064887046814,
      "learning_rate": 0.00042619231719353363,
      "loss": 2.024,
      "step": 83361
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.913623034954071,
      "learning_rate": 0.0004261886061127437,
      "loss": 2.2626,
      "step": 83362
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.7915872931480408,
      "learning_rate": 0.0004261848950084929,
      "loss": 2.4949,
      "step": 83363
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.834644079208374,
      "learning_rate": 0.00042618118388078205,
      "loss": 2.1783,
      "step": 83364
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.8189061284065247,
      "learning_rate": 0.0004261774727296117,
      "loss": 2.085,
      "step": 83365
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.878976583480835,
      "learning_rate": 0.0004261737615549825,
      "loss": 2.064,
      "step": 83366
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.9002004861831665,
      "learning_rate": 0.0004261700503568953,
      "loss": 2.0513,
      "step": 83367
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.8316457271575928,
      "learning_rate": 0.0004261663391353507,
      "loss": 2.0572,
      "step": 83368
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.8308615684509277,
      "learning_rate": 0.0004261626278903493,
      "loss": 2.5397,
      "step": 83369
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.8602315187454224,
      "learning_rate": 0.00042615891662189204,
      "loss": 2.3928,
      "step": 83370
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.8283051252365112,
      "learning_rate": 0.0004261552053299793,
      "loss": 2.2295,
      "step": 83371
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.836604118347168,
      "learning_rate": 0.00042615149401461203,
      "loss": 2.2665,
      "step": 83372
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.8420352339744568,
      "learning_rate": 0.0004261477826757908,
      "loss": 2.0733,
      "step": 83373
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.7306796312332153,
      "learning_rate": 0.0004261440713135162,
      "loss": 2.2267,
      "step": 83374
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.8646188974380493,
      "learning_rate": 0.00042614035992778916,
      "loss": 1.8875,
      "step": 83375
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.6862082481384277,
      "learning_rate": 0.00042613664851861026,
      "loss": 1.8871,
      "step": 83376
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.7775030136108398,
      "learning_rate": 0.0004261329370859802,
      "loss": 2.3913,
      "step": 83377
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.7616873383522034,
      "learning_rate": 0.00042612922562989956,
      "loss": 2.1281,
      "step": 83378
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.8497294187545776,
      "learning_rate": 0.00042612551415036905,
      "loss": 2.0868,
      "step": 83379
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.8616629838943481,
      "learning_rate": 0.00042612180264738953,
      "loss": 2.2447,
      "step": 83380
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.7103754281997681,
      "learning_rate": 0.00042611809112096165,
      "loss": 2.0153,
      "step": 83381
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.8973167538642883,
      "learning_rate": 0.00042611437957108594,
      "loss": 2.1296,
      "step": 83382
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.8442985415458679,
      "learning_rate": 0.0004261106679977632,
      "loss": 2.0935,
      "step": 83383
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.8249119520187378,
      "learning_rate": 0.0004261069564009941,
      "loss": 2.1823,
      "step": 83384
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.779419481754303,
      "learning_rate": 0.00042610324478077936,
      "loss": 2.4657,
      "step": 83385
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.7387449741363525,
      "learning_rate": 0.0004260995331371196,
      "loss": 2.2688,
      "step": 83386
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.8407860994338989,
      "learning_rate": 0.0004260958214700156,
      "loss": 1.9841,
      "step": 83387
    },
    {
      "epoch": 1.09,
      "grad_norm": 1.027503490447998,
      "learning_rate": 0.000426092109779468,
      "loss": 2.2605,
      "step": 83388
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.7340055108070374,
      "learning_rate": 0.0004260883980654775,
      "loss": 2.0257,
      "step": 83389
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.9378015398979187,
      "learning_rate": 0.00042608468632804477,
      "loss": 2.0841,
      "step": 83390
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.8442812561988831,
      "learning_rate": 0.0004260809745671705,
      "loss": 2.1741,
      "step": 83391
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.789530336856842,
      "learning_rate": 0.00042607726278285545,
      "loss": 2.3077,
      "step": 83392
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.8504749536514282,
      "learning_rate": 0.00042607355097510025,
      "loss": 2.1604,
      "step": 83393
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.8076501488685608,
      "learning_rate": 0.00042606983914390557,
      "loss": 2.2891,
      "step": 83394
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.7731242179870605,
      "learning_rate": 0.0004260661272892722,
      "loss": 2.102,
      "step": 83395
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.8469923138618469,
      "learning_rate": 0.0004260624154112006,
      "loss": 2.3304,
      "step": 83396
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.8003154397010803,
      "learning_rate": 0.0004260587035096917,
      "loss": 2.2967,
      "step": 83397
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.8299270868301392,
      "learning_rate": 0.0004260549915847461,
      "loss": 2.2454,
      "step": 83398
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.8044423460960388,
      "learning_rate": 0.00042605127963636456,
      "loss": 2.073,
      "step": 83399
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.8499447703361511,
      "learning_rate": 0.0004260475676645476,
      "loss": 2.2731,
      "step": 83400
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.8641321659088135,
      "learning_rate": 0.00042604385566929613,
      "loss": 2.1935,
      "step": 83401
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.7629972696304321,
      "learning_rate": 0.0004260401436506106,
      "loss": 2.1732,
      "step": 83402
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.8158810138702393,
      "learning_rate": 0.000426036431608492,
      "loss": 2.0096,
      "step": 83403
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.7789121866226196,
      "learning_rate": 0.0004260327195429407,
      "loss": 2.2205,
      "step": 83404
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.7717127799987793,
      "learning_rate": 0.00042602900745395756,
      "loss": 2.0254,
      "step": 83405
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.8450660109519958,
      "learning_rate": 0.00042602529534154334,
      "loss": 2.3236,
      "step": 83406
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.7771978378295898,
      "learning_rate": 0.00042602158320569853,
      "loss": 2.0918,
      "step": 83407
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.8976418375968933,
      "learning_rate": 0.00042601787104642395,
      "loss": 2.3083,
      "step": 83408
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.7571003437042236,
      "learning_rate": 0.00042601415886372036,
      "loss": 1.9925,
      "step": 83409
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.7782974243164062,
      "learning_rate": 0.00042601044665758824,
      "loss": 2.285,
      "step": 83410
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.7644814252853394,
      "learning_rate": 0.0004260067344280285,
      "loss": 2.2612,
      "step": 83411
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.7749821543693542,
      "learning_rate": 0.00042600302217504164,
      "loss": 2.0611,
      "step": 83412
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.8314816355705261,
      "learning_rate": 0.00042599930989862855,
      "loss": 2.0163,
      "step": 83413
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.86561119556427,
      "learning_rate": 0.0004259955975987897,
      "loss": 2.144,
      "step": 83414
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.8457360863685608,
      "learning_rate": 0.000425991885275526,
      "loss": 2.2674,
      "step": 83415
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.748993992805481,
      "learning_rate": 0.0004259881729288379,
      "loss": 2.1281,
      "step": 83416
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.8888893127441406,
      "learning_rate": 0.0004259844605587263,
      "loss": 2.3438,
      "step": 83417
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.814490795135498,
      "learning_rate": 0.00042598074816519176,
      "loss": 2.1748,
      "step": 83418
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.9014081358909607,
      "learning_rate": 0.0004259770357482351,
      "loss": 2.2478,
      "step": 83419
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.8274964690208435,
      "learning_rate": 0.0004259733233078569,
      "loss": 2.4155,
      "step": 83420
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.9417011141777039,
      "learning_rate": 0.0004259696108440579,
      "loss": 2.3175,
      "step": 83421
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.7573262453079224,
      "learning_rate": 0.00042596589835683867,
      "loss": 2.1161,
      "step": 83422
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.8456779718399048,
      "learning_rate": 0.00042596218584620014,
      "loss": 2.2179,
      "step": 83423
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.8020697832107544,
      "learning_rate": 0.0004259584733121427,
      "loss": 2.0927,
      "step": 83424
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.8698005676269531,
      "learning_rate": 0.00042595476075466736,
      "loss": 2.2016,
      "step": 83425
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.8278550505638123,
      "learning_rate": 0.0004259510481737746,
      "loss": 2.3443,
      "step": 83426
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.7607172131538391,
      "learning_rate": 0.00042594733556946515,
      "loss": 2.078,
      "step": 83427
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.7478788495063782,
      "learning_rate": 0.0004259436229417398,
      "loss": 2.1388,
      "step": 83428
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.9105775356292725,
      "learning_rate": 0.000425939910290599,
      "loss": 1.9975,
      "step": 83429
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.8448607325553894,
      "learning_rate": 0.00042593619761604356,
      "loss": 2.1007,
      "step": 83430
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.9004802107810974,
      "learning_rate": 0.0004259324849180744,
      "loss": 2.1205,
      "step": 83431
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.8801754713058472,
      "learning_rate": 0.0004259287721966919,
      "loss": 2.0046,
      "step": 83432
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.8083924651145935,
      "learning_rate": 0.00042592505945189687,
      "loss": 2.083,
      "step": 83433
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.86182701587677,
      "learning_rate": 0.00042592134668369006,
      "loss": 2.1861,
      "step": 83434
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.8442286849021912,
      "learning_rate": 0.00042591763389207197,
      "loss": 2.0383,
      "step": 83435
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.8710145354270935,
      "learning_rate": 0.00042591392107704345,
      "loss": 2.5562,
      "step": 83436
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.8557652831077576,
      "learning_rate": 0.00042591020823860533,
      "loss": 2.2629,
      "step": 83437
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.9214932322502136,
      "learning_rate": 0.00042590649537675794,
      "loss": 2.0118,
      "step": 83438
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.890703558921814,
      "learning_rate": 0.0004259027824915022,
      "loss": 2.4041,
      "step": 83439
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.9375588893890381,
      "learning_rate": 0.00042589906958283877,
      "loss": 2.0508,
      "step": 83440
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.8611798882484436,
      "learning_rate": 0.0004258953566507683,
      "loss": 2.3474,
      "step": 83441
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.9511393904685974,
      "learning_rate": 0.0004258916436952915,
      "loss": 2.0779,
      "step": 83442
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.9530830383300781,
      "learning_rate": 0.00042588793071640906,
      "loss": 2.2985,
      "step": 83443
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.7991140484809875,
      "learning_rate": 0.00042588421771412176,
      "loss": 2.1209,
      "step": 83444
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.8202939629554749,
      "learning_rate": 0.00042588050468843016,
      "loss": 2.379,
      "step": 83445
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.8512697219848633,
      "learning_rate": 0.0004258767916393349,
      "loss": 2.2086,
      "step": 83446
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.8861930966377258,
      "learning_rate": 0.00042587307856683696,
      "loss": 2.2592,
      "step": 83447
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.7998347878456116,
      "learning_rate": 0.0004258693654709367,
      "loss": 2.0506,
      "step": 83448
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.7960724234580994,
      "learning_rate": 0.000425865652351635,
      "loss": 2.3849,
      "step": 83449
    },
    {
      "epoch": 1.09,
      "grad_norm": 1.0507311820983887,
      "learning_rate": 0.0004258619392089325,
      "loss": 2.2994,
      "step": 83450
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.7856338024139404,
      "learning_rate": 0.00042585822604282984,
      "loss": 2.1826,
      "step": 83451
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.7987677454948425,
      "learning_rate": 0.00042585451285332776,
      "loss": 1.9976,
      "step": 83452
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.7836374640464783,
      "learning_rate": 0.0004258507996404271,
      "loss": 2.1614,
      "step": 83453
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.8491041660308838,
      "learning_rate": 0.00042584708640412824,
      "loss": 2.1471,
      "step": 83454
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.7502838969230652,
      "learning_rate": 0.00042584337314443204,
      "loss": 1.9855,
      "step": 83455
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.7461698651313782,
      "learning_rate": 0.00042583965986133925,
      "loss": 2.1578,
      "step": 83456
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.8380905985832214,
      "learning_rate": 0.0004258359465548505,
      "loss": 2.0085,
      "step": 83457
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.86460942029953,
      "learning_rate": 0.00042583223322496637,
      "loss": 2.2342,
      "step": 83458
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.7708079814910889,
      "learning_rate": 0.0004258285198716878,
      "loss": 2.161,
      "step": 83459
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.8155184984207153,
      "learning_rate": 0.00042582480649501524,
      "loss": 2.1414,
      "step": 83460
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.7735704183578491,
      "learning_rate": 0.00042582109309494945,
      "loss": 1.9943,
      "step": 83461
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.7741259336471558,
      "learning_rate": 0.0004258173796714912,
      "loss": 2.1456,
      "step": 83462
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.7424352765083313,
      "learning_rate": 0.0004258136662246411,
      "loss": 2.1457,
      "step": 83463
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.7531487941741943,
      "learning_rate": 0.0004258099527543998,
      "loss": 2.176,
      "step": 83464
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.7526211738586426,
      "learning_rate": 0.0004258062392607682,
      "loss": 2.195,
      "step": 83465
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.786069393157959,
      "learning_rate": 0.0004258025257437468,
      "loss": 2.3735,
      "step": 83466
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.7958434820175171,
      "learning_rate": 0.0004257988122033362,
      "loss": 2.2171,
      "step": 83467
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.8108161091804504,
      "learning_rate": 0.00042579509863953743,
      "loss": 2.0702,
      "step": 83468
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.7479146122932434,
      "learning_rate": 0.00042579138505235086,
      "loss": 2.1053,
      "step": 83469
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.7938684225082397,
      "learning_rate": 0.00042578767144177724,
      "loss": 2.3636,
      "step": 83470
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.7624810338020325,
      "learning_rate": 0.00042578395780781743,
      "loss": 2.4198,
      "step": 83471
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.8480437994003296,
      "learning_rate": 0.0004257802441504719,
      "loss": 2.381,
      "step": 83472
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.8143423795700073,
      "learning_rate": 0.00042577653046974154,
      "loss": 2.2973,
      "step": 83473
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.7992587089538574,
      "learning_rate": 0.00042577281676562696,
      "loss": 2.1859,
      "step": 83474
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.8130197525024414,
      "learning_rate": 0.00042576910303812886,
      "loss": 2.24,
      "step": 83475
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.8383140563964844,
      "learning_rate": 0.00042576538928724777,
      "loss": 2.0501,
      "step": 83476
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.8079326152801514,
      "learning_rate": 0.00042576167551298467,
      "loss": 2.1539,
      "step": 83477
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.7927972674369812,
      "learning_rate": 0.00042575796171534,
      "loss": 2.3199,
      "step": 83478
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.8721701502799988,
      "learning_rate": 0.0004257542478943146,
      "loss": 2.2441,
      "step": 83479
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.8907123804092407,
      "learning_rate": 0.00042575053404990913,
      "loss": 2.2141,
      "step": 83480
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.772813081741333,
      "learning_rate": 0.0004257468201821242,
      "loss": 2.4308,
      "step": 83481
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.7303819060325623,
      "learning_rate": 0.00042574310629096054,
      "loss": 2.2255,
      "step": 83482
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.8253968954086304,
      "learning_rate": 0.0004257393923764189,
      "loss": 2.3162,
      "step": 83483
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.8638604879379272,
      "learning_rate": 0.0004257356784385,
      "loss": 2.3228,
      "step": 83484
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.8403924703598022,
      "learning_rate": 0.0004257319644772044,
      "loss": 2.2167,
      "step": 83485
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.7937370538711548,
      "learning_rate": 0.00042572825049253286,
      "loss": 2.3688,
      "step": 83486
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.9000864028930664,
      "learning_rate": 0.0004257245364844861,
      "loss": 2.237,
      "step": 83487
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.9630174040794373,
      "learning_rate": 0.0004257208224530647,
      "loss": 1.972,
      "step": 83488
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.8287118673324585,
      "learning_rate": 0.0004257171083982695,
      "loss": 2.2715,
      "step": 83489
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.9049173593521118,
      "learning_rate": 0.000425713394320101,
      "loss": 2.4198,
      "step": 83490
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.8673158884048462,
      "learning_rate": 0.00042570968021856016,
      "loss": 2.1939,
      "step": 83491
    },
    {
      "epoch": 1.09,
      "grad_norm": 1.0594183206558228,
      "learning_rate": 0.0004257059660936475,
      "loss": 1.9648,
      "step": 83492
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.7692622542381287,
      "learning_rate": 0.00042570225194536363,
      "loss": 2.2925,
      "step": 83493
    },
    {
      "epoch": 1.09,
      "grad_norm": 1.044755458831787,
      "learning_rate": 0.00042569853777370937,
      "loss": 1.9773,
      "step": 83494
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.9309423565864563,
      "learning_rate": 0.00042569482357868545,
      "loss": 2.0692,
      "step": 83495
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.8187007308006287,
      "learning_rate": 0.00042569110936029243,
      "loss": 2.3152,
      "step": 83496
    },
    {
      "epoch": 1.09,
      "grad_norm": 1.5143704414367676,
      "learning_rate": 0.00042568739511853105,
      "loss": 2.1957,
      "step": 83497
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.7830873727798462,
      "learning_rate": 0.00042568368085340203,
      "loss": 2.1912,
      "step": 83498
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.8719130754470825,
      "learning_rate": 0.000425679966564906,
      "loss": 2.2491,
      "step": 83499
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.8138114213943481,
      "learning_rate": 0.00042567625225304375,
      "loss": 2.2399,
      "step": 83500
    },
    {
      "epoch": 1.09,
      "grad_norm": 1.0563195943832397,
      "learning_rate": 0.00042567253791781596,
      "loss": 2.2008,
      "step": 83501
    },
    {
      "epoch": 1.09,
      "grad_norm": 1.009321689605713,
      "learning_rate": 0.0004256688235592232,
      "loss": 1.992,
      "step": 83502
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.8278580904006958,
      "learning_rate": 0.00042566510917726617,
      "loss": 2.1996,
      "step": 83503
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.8493701219558716,
      "learning_rate": 0.0004256613947719458,
      "loss": 2.104,
      "step": 83504
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.9791918992996216,
      "learning_rate": 0.00042565768034326243,
      "loss": 2.382,
      "step": 83505
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.9436601400375366,
      "learning_rate": 0.000425653965891217,
      "loss": 2.3138,
      "step": 83506
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.7585821747779846,
      "learning_rate": 0.00042565025141581007,
      "loss": 2.1605,
      "step": 83507
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.8882133364677429,
      "learning_rate": 0.0004256465369170425,
      "loss": 2.1875,
      "step": 83508
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.8733195066452026,
      "learning_rate": 0.00042564282239491476,
      "loss": 2.2111,
      "step": 83509
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.8028760552406311,
      "learning_rate": 0.0004256391078494277,
      "loss": 2.0801,
      "step": 83510
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.7919265627861023,
      "learning_rate": 0.000425635393280582,
      "loss": 2.51,
      "step": 83511
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.8071573376655579,
      "learning_rate": 0.0004256316786883782,
      "loss": 2.2438,
      "step": 83512
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.8167544007301331,
      "learning_rate": 0.0004256279640728172,
      "loss": 2.1822,
      "step": 83513
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.7409310340881348,
      "learning_rate": 0.0004256242494338995,
      "loss": 2.3337,
      "step": 83514
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.6945967078208923,
      "learning_rate": 0.00042562053477162594,
      "loss": 2.1114,
      "step": 83515
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.7820339202880859,
      "learning_rate": 0.00042561682008599713,
      "loss": 2.1876,
      "step": 83516
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.7850202918052673,
      "learning_rate": 0.00042561310537701384,
      "loss": 2.2171,
      "step": 83517
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.7771021723747253,
      "learning_rate": 0.0004256093906446766,
      "loss": 2.2063,
      "step": 83518
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.7797028422355652,
      "learning_rate": 0.0004256056758889863,
      "loss": 2.1181,
      "step": 83519
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.7651598453521729,
      "learning_rate": 0.0004256019611099435,
      "loss": 2.399,
      "step": 83520
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.8027188181877136,
      "learning_rate": 0.0004255982463075489,
      "loss": 2.166,
      "step": 83521
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.7921959161758423,
      "learning_rate": 0.00042559453148180323,
      "loss": 1.8629,
      "step": 83522
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.8640953898429871,
      "learning_rate": 0.00042559081663270725,
      "loss": 2.2199,
      "step": 83523
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.9307940602302551,
      "learning_rate": 0.00042558710176026146,
      "loss": 2.0362,
      "step": 83524
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.9467107057571411,
      "learning_rate": 0.0004255833868644666,
      "loss": 2.0994,
      "step": 83525
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.8279132843017578,
      "learning_rate": 0.0004255796719453236,
      "loss": 2.2928,
      "step": 83526
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.7724599838256836,
      "learning_rate": 0.0004255759570028328,
      "loss": 2.2985,
      "step": 83527
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.8093891739845276,
      "learning_rate": 0.00042557224203699514,
      "loss": 2.0012,
      "step": 83528
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.7643277049064636,
      "learning_rate": 0.00042556852704781124,
      "loss": 2.4223,
      "step": 83529
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.7750470638275146,
      "learning_rate": 0.00042556481203528176,
      "loss": 2.1593,
      "step": 83530
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.8290139436721802,
      "learning_rate": 0.0004255610969994074,
      "loss": 2.3204,
      "step": 83531
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.9064628481864929,
      "learning_rate": 0.0004255573819401889,
      "loss": 2.309,
      "step": 83532
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.7847305536270142,
      "learning_rate": 0.00042555366685762685,
      "loss": 2.2948,
      "step": 83533
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.761520504951477,
      "learning_rate": 0.00042554995175172203,
      "loss": 2.3471,
      "step": 83534
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.7856888175010681,
      "learning_rate": 0.0004255462366224751,
      "loss": 2.2558,
      "step": 83535
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.7945443391799927,
      "learning_rate": 0.0004255425214698868,
      "loss": 2.3967,
      "step": 83536
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.7861089110374451,
      "learning_rate": 0.0004255388062939577,
      "loss": 1.9837,
      "step": 83537
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.7916808724403381,
      "learning_rate": 0.0004255350910946887,
      "loss": 2.1486,
      "step": 83538
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.7922354340553284,
      "learning_rate": 0.0004255313758720802,
      "loss": 2.4057,
      "step": 83539
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.7826058864593506,
      "learning_rate": 0.0004255276606261331,
      "loss": 2.2516,
      "step": 83540
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.9225768446922302,
      "learning_rate": 0.00042552394535684806,
      "loss": 2.3745,
      "step": 83541
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.7837117314338684,
      "learning_rate": 0.00042552023006422573,
      "loss": 2.2276,
      "step": 83542
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.966909646987915,
      "learning_rate": 0.0004255165147482668,
      "loss": 2.3034,
      "step": 83543
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.7902282476425171,
      "learning_rate": 0.0004255127994089721,
      "loss": 2.1176,
      "step": 83544
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.8525835275650024,
      "learning_rate": 0.00042550908404634206,
      "loss": 2.1637,
      "step": 83545
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.9337657690048218,
      "learning_rate": 0.0004255053686603776,
      "loss": 2.1307,
      "step": 83546
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.9875470399856567,
      "learning_rate": 0.0004255016532510793,
      "loss": 2.197,
      "step": 83547
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.7876061797142029,
      "learning_rate": 0.0004254979378184478,
      "loss": 2.2359,
      "step": 83548
    },
    {
      "epoch": 1.09,
      "grad_norm": 1.0578292608261108,
      "learning_rate": 0.0004254942223624839,
      "loss": 2.4169,
      "step": 83549
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.9853307604789734,
      "learning_rate": 0.00042549050688318836,
      "loss": 2.2234,
      "step": 83550
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.9362934231758118,
      "learning_rate": 0.00042548679138056165,
      "loss": 2.2052,
      "step": 83551
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.8442784547805786,
      "learning_rate": 0.0004254830758546046,
      "loss": 2.1682,
      "step": 83552
    },
    {
      "epoch": 1.09,
      "grad_norm": 1.022359848022461,
      "learning_rate": 0.0004254793603053179,
      "loss": 2.1941,
      "step": 83553
    },
    {
      "epoch": 1.09,
      "grad_norm": 1.077015995979309,
      "learning_rate": 0.0004254756447327022,
      "loss": 2.3018,
      "step": 83554
    },
    {
      "epoch": 1.09,
      "grad_norm": 1.0606663227081299,
      "learning_rate": 0.0004254719291367582,
      "loss": 2.1503,
      "step": 83555
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.892873227596283,
      "learning_rate": 0.0004254682135174867,
      "loss": 1.9528,
      "step": 83556
    },
    {
      "epoch": 1.09,
      "grad_norm": 1.0756984949111938,
      "learning_rate": 0.00042546449787488814,
      "loss": 2.2266,
      "step": 83557
    },
    {
      "epoch": 1.09,
      "grad_norm": 1.0291935205459595,
      "learning_rate": 0.00042546078220896346,
      "loss": 2.1924,
      "step": 83558
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.869277834892273,
      "learning_rate": 0.00042545706651971326,
      "loss": 2.2443,
      "step": 83559
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.7719991207122803,
      "learning_rate": 0.0004254533508071382,
      "loss": 2.1763,
      "step": 83560
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.8966699838638306,
      "learning_rate": 0.00042544963507123887,
      "loss": 2.2716,
      "step": 83561
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.9142882227897644,
      "learning_rate": 0.00042544591931201626,
      "loss": 2.253,
      "step": 83562
    },
    {
      "epoch": 1.09,
      "grad_norm": 1.340415120124817,
      "learning_rate": 0.0004254422035294708,
      "loss": 2.2082,
      "step": 83563
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.8134440779685974,
      "learning_rate": 0.00042543848772360325,
      "loss": 2.2599,
      "step": 83564
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.8730207085609436,
      "learning_rate": 0.0004254347718944144,
      "loss": 2.2522,
      "step": 83565
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.951885461807251,
      "learning_rate": 0.00042543105604190477,
      "loss": 2.2399,
      "step": 83566
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.9956233501434326,
      "learning_rate": 0.00042542734016607524,
      "loss": 2.1381,
      "step": 83567
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.8954094052314758,
      "learning_rate": 0.0004254236242669263,
      "loss": 2.1146,
      "step": 83568
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.8038201928138733,
      "learning_rate": 0.0004254199083444588,
      "loss": 2.3031,
      "step": 83569
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.8267388343811035,
      "learning_rate": 0.0004254161923986733,
      "loss": 2.0862,
      "step": 83570
    },
    {
      "epoch": 1.09,
      "grad_norm": 1.0345014333724976,
      "learning_rate": 0.0004254124764295706,
      "loss": 2.1678,
      "step": 83571
    },
    {
      "epoch": 1.09,
      "grad_norm": 1.0014151334762573,
      "learning_rate": 0.00042540876043715146,
      "loss": 1.9769,
      "step": 83572
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.8306522965431213,
      "learning_rate": 0.0004254050444214163,
      "loss": 2.2435,
      "step": 83573
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.7993490695953369,
      "learning_rate": 0.00042540132838236606,
      "loss": 2.4161,
      "step": 83574
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.9992225766181946,
      "learning_rate": 0.0004253976123200013,
      "loss": 2.2132,
      "step": 83575
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.9722571969032288,
      "learning_rate": 0.00042539389623432285,
      "loss": 2.2999,
      "step": 83576
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.7795670032501221,
      "learning_rate": 0.0004253901801253312,
      "loss": 2.0902,
      "step": 83577
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.8321530818939209,
      "learning_rate": 0.0004253864639930272,
      "loss": 2.2688,
      "step": 83578
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.7866131663322449,
      "learning_rate": 0.0004253827478374115,
      "loss": 2.0401,
      "step": 83579
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.8981147408485413,
      "learning_rate": 0.0004253790316584848,
      "loss": 2.0709,
      "step": 83580
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.7823248505592346,
      "learning_rate": 0.0004253753154562477,
      "loss": 2.1895,
      "step": 83581
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.7866173386573792,
      "learning_rate": 0.00042537159923070094,
      "loss": 2.0735,
      "step": 83582
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.7926773428916931,
      "learning_rate": 0.00042536788298184536,
      "loss": 2.1666,
      "step": 83583
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.7748804688453674,
      "learning_rate": 0.00042536416670968145,
      "loss": 2.2465,
      "step": 83584
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.7320629954338074,
      "learning_rate": 0.00042536045041420995,
      "loss": 2.6392,
      "step": 83585
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.8043582439422607,
      "learning_rate": 0.0004253567340954316,
      "loss": 2.1795,
      "step": 83586
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.897946834564209,
      "learning_rate": 0.0004253530177533471,
      "loss": 1.8996,
      "step": 83587
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.755212128162384,
      "learning_rate": 0.00042534930138795707,
      "loss": 1.9836,
      "step": 83588
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.9976218938827515,
      "learning_rate": 0.00042534558499926224,
      "loss": 2.2109,
      "step": 83589
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.769166886806488,
      "learning_rate": 0.00042534186858726334,
      "loss": 2.223,
      "step": 83590
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.7605621218681335,
      "learning_rate": 0.00042533815215196094,
      "loss": 2.1814,
      "step": 83591
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.7870752215385437,
      "learning_rate": 0.0004253344356933559,
      "loss": 2.1608,
      "step": 83592
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.8783025145530701,
      "learning_rate": 0.00042533071921144883,
      "loss": 2.0572,
      "step": 83593
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.9450098276138306,
      "learning_rate": 0.0004253270027062403,
      "loss": 2.3613,
      "step": 83594
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.7352650165557861,
      "learning_rate": 0.00042532328617773116,
      "loss": 2.2452,
      "step": 83595
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.8450143933296204,
      "learning_rate": 0.0004253195696259222,
      "loss": 2.297,
      "step": 83596
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.7892742156982422,
      "learning_rate": 0.00042531585305081375,
      "loss": 2.1027,
      "step": 83597
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.8372528553009033,
      "learning_rate": 0.0004253121364524068,
      "loss": 2.1372,
      "step": 83598
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.803525984287262,
      "learning_rate": 0.000425308419830702,
      "loss": 2.2306,
      "step": 83599
    },
    {
      "epoch": 1.09,
      "grad_norm": 1.0586038827896118,
      "learning_rate": 0.0004253047031857,
      "loss": 2.176,
      "step": 83600
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.9052860140800476,
      "learning_rate": 0.0004253009865174014,
      "loss": 1.9584,
      "step": 83601
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.8461953401565552,
      "learning_rate": 0.0004252972698258071,
      "loss": 2.2822,
      "step": 83602
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.8296229243278503,
      "learning_rate": 0.00042529355311091756,
      "loss": 1.9771,
      "step": 83603
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.8904627561569214,
      "learning_rate": 0.0004252898363727337,
      "loss": 2.2151,
      "step": 83604
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.8233821988105774,
      "learning_rate": 0.00042528611961125605,
      "loss": 2.1864,
      "step": 83605
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.8119893074035645,
      "learning_rate": 0.00042528240282648525,
      "loss": 2.2755,
      "step": 83606
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.7941827774047852,
      "learning_rate": 0.00042527868601842215,
      "loss": 1.9975,
      "step": 83607
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.8870932459831238,
      "learning_rate": 0.00042527496918706747,
      "loss": 2.0346,
      "step": 83608
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.7925960421562195,
      "learning_rate": 0.00042527125233242174,
      "loss": 2.3314,
      "step": 83609
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.835273802280426,
      "learning_rate": 0.0004252675354544856,
      "loss": 2.1509,
      "step": 83610
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.9552025198936462,
      "learning_rate": 0.00042526381855326,
      "loss": 2.2545,
      "step": 83611
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.855217456817627,
      "learning_rate": 0.0004252601016287455,
      "loss": 2.2007,
      "step": 83612
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.9723808169364929,
      "learning_rate": 0.0004252563846809427,
      "loss": 2.3346,
      "step": 83613
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.7783290147781372,
      "learning_rate": 0.00042525266770985246,
      "loss": 2.1878,
      "step": 83614
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.9089902639389038,
      "learning_rate": 0.00042524895071547534,
      "loss": 2.0829,
      "step": 83615
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.7876380681991577,
      "learning_rate": 0.0004252452336978121,
      "loss": 2.1439,
      "step": 83616
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.9004669785499573,
      "learning_rate": 0.00042524151665686337,
      "loss": 2.3016,
      "step": 83617
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.9518863558769226,
      "learning_rate": 0.0004252377995926299,
      "loss": 2.133,
      "step": 83618
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.8430811166763306,
      "learning_rate": 0.0004252340825051123,
      "loss": 2.1014,
      "step": 83619
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.7644710540771484,
      "learning_rate": 0.00042523036539431145,
      "loss": 2.2822,
      "step": 83620
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.8161829113960266,
      "learning_rate": 0.00042522664826022784,
      "loss": 2.1295,
      "step": 83621
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.9246717691421509,
      "learning_rate": 0.0004252229311028622,
      "loss": 2.1812,
      "step": 83622
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.8496538400650024,
      "learning_rate": 0.0004252192139222153,
      "loss": 2.0706,
      "step": 83623
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.7933159470558167,
      "learning_rate": 0.0004252154967182877,
      "loss": 2.2893,
      "step": 83624
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.8896545767784119,
      "learning_rate": 0.00042521177949108024,
      "loss": 2.3684,
      "step": 83625
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.8804619312286377,
      "learning_rate": 0.00042520806224059354,
      "loss": 1.9789,
      "step": 83626
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.8182049989700317,
      "learning_rate": 0.00042520434496682837,
      "loss": 2.359,
      "step": 83627
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.7788448929786682,
      "learning_rate": 0.00042520062766978527,
      "loss": 2.1728,
      "step": 83628
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.7766987085342407,
      "learning_rate": 0.000425196910349465,
      "loss": 2.1766,
      "step": 83629
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.8762685060501099,
      "learning_rate": 0.00042519319300586827,
      "loss": 2.1946,
      "step": 83630
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.7971957921981812,
      "learning_rate": 0.00042518947563899577,
      "loss": 2.1243,
      "step": 83631
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.8406190276145935,
      "learning_rate": 0.0004251857582488482,
      "loss": 2.1864,
      "step": 83632
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.8206669092178345,
      "learning_rate": 0.0004251820408354262,
      "loss": 2.0662,
      "step": 83633
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.8072525262832642,
      "learning_rate": 0.00042517832339873055,
      "loss": 2.3484,
      "step": 83634
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.8399648666381836,
      "learning_rate": 0.0004251746059387619,
      "loss": 2.1108,
      "step": 83635
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.7389960289001465,
      "learning_rate": 0.00042517088845552085,
      "loss": 2.1253,
      "step": 83636
    },
    {
      "epoch": 1.09,
      "grad_norm": 1.2060469388961792,
      "learning_rate": 0.00042516717094900815,
      "loss": 2.4357,
      "step": 83637
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.9682634472846985,
      "learning_rate": 0.00042516345341922463,
      "loss": 2.1585,
      "step": 83638
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.7267316579818726,
      "learning_rate": 0.00042515973586617077,
      "loss": 2.0067,
      "step": 83639
    },
    {
      "epoch": 1.09,
      "grad_norm": 1.0393446683883667,
      "learning_rate": 0.0004251560182898474,
      "loss": 2.3833,
      "step": 83640
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.876627504825592,
      "learning_rate": 0.00042515230069025507,
      "loss": 2.468,
      "step": 83641
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.9909371137619019,
      "learning_rate": 0.00042514858306739474,
      "loss": 2.2729,
      "step": 83642
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.7954438924789429,
      "learning_rate": 0.0004251448654212668,
      "loss": 2.2111,
      "step": 83643
    },
    {
      "epoch": 1.09,
      "grad_norm": 1.013764500617981,
      "learning_rate": 0.00042514114775187207,
      "loss": 1.9313,
      "step": 83644
    },
    {
      "epoch": 1.09,
      "grad_norm": 1.0050344467163086,
      "learning_rate": 0.00042513743005921124,
      "loss": 2.1353,
      "step": 83645
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.8284009099006653,
      "learning_rate": 0.000425133712343285,
      "loss": 1.8639,
      "step": 83646
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.8536670804023743,
      "learning_rate": 0.0004251299946040941,
      "loss": 2.2489,
      "step": 83647
    },
    {
      "epoch": 1.09,
      "grad_norm": 1.2367072105407715,
      "learning_rate": 0.00042512627684163914,
      "loss": 2.3886,
      "step": 83648
    },
    {
      "epoch": 1.09,
      "grad_norm": 1.0174193382263184,
      "learning_rate": 0.0004251225590559208,
      "loss": 2.3236,
      "step": 83649
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.7915294170379639,
      "learning_rate": 0.0004251188412469398,
      "loss": 2.2079,
      "step": 83650
    },
    {
      "epoch": 1.09,
      "grad_norm": 1.0950995683670044,
      "learning_rate": 0.0004251151234146969,
      "loss": 2.2437,
      "step": 83651
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.8326361179351807,
      "learning_rate": 0.0004251114055591927,
      "loss": 2.1942,
      "step": 83652
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.8255053758621216,
      "learning_rate": 0.00042510768768042804,
      "loss": 2.2701,
      "step": 83653
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.8444162607192993,
      "learning_rate": 0.00042510396977840344,
      "loss": 2.0772,
      "step": 83654
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.8408970832824707,
      "learning_rate": 0.0004251002518531196,
      "loss": 2.4832,
      "step": 83655
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.8272619843482971,
      "learning_rate": 0.0004250965339045772,
      "loss": 2.2943,
      "step": 83656
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.8122655153274536,
      "learning_rate": 0.0004250928159327772,
      "loss": 2.3755,
      "step": 83657
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.8081372380256653,
      "learning_rate": 0.0004250890979377199,
      "loss": 2.2246,
      "step": 83658
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.7895283102989197,
      "learning_rate": 0.0004250853799194063,
      "loss": 2.1845,
      "step": 83659
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.798344075679779,
      "learning_rate": 0.00042508166187783694,
      "loss": 2.0669,
      "step": 83660
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.7889718413352966,
      "learning_rate": 0.00042507794381301244,
      "loss": 2.338,
      "step": 83661
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.7122527360916138,
      "learning_rate": 0.0004250742257249337,
      "loss": 2.1508,
      "step": 83662
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.8579971790313721,
      "learning_rate": 0.0004250705076136012,
      "loss": 2.1184,
      "step": 83663
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.9856120347976685,
      "learning_rate": 0.00042506678947901584,
      "loss": 2.1034,
      "step": 83664
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.8293099403381348,
      "learning_rate": 0.00042506307132117807,
      "loss": 2.2484,
      "step": 83665
    },
    {
      "epoch": 1.09,
      "grad_norm": 1.0574969053268433,
      "learning_rate": 0.0004250593531400888,
      "loss": 2.1022,
      "step": 83666
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.8038318157196045,
      "learning_rate": 0.0004250556349357486,
      "loss": 2.0547,
      "step": 83667
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.8464992046356201,
      "learning_rate": 0.0004250519167081583,
      "loss": 2.245,
      "step": 83668
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.759861946105957,
      "learning_rate": 0.0004250481984573184,
      "loss": 2.2449,
      "step": 83669
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.8327757120132446,
      "learning_rate": 0.0004250444801832297,
      "loss": 2.3223,
      "step": 83670
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.867043137550354,
      "learning_rate": 0.0004250407618858928,
      "loss": 2.3921,
      "step": 83671
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.8664206266403198,
      "learning_rate": 0.0004250370435653086,
      "loss": 1.872,
      "step": 83672
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.80934739112854,
      "learning_rate": 0.0004250333252214775,
      "loss": 2.166,
      "step": 83673
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.8456985950469971,
      "learning_rate": 0.00042502960685440044,
      "loss": 2.4044,
      "step": 83674
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.8499696254730225,
      "learning_rate": 0.00042502588846407807,
      "loss": 2.1991,
      "step": 83675
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.9066942930221558,
      "learning_rate": 0.00042502217005051094,
      "loss": 2.1558,
      "step": 83676
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.7758747339248657,
      "learning_rate": 0.0004250184516136998,
      "loss": 2.2796,
      "step": 83677
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.8203797340393066,
      "learning_rate": 0.00042501473315364547,
      "loss": 1.9858,
      "step": 83678
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.8398159742355347,
      "learning_rate": 0.0004250110146703485,
      "loss": 1.8934,
      "step": 83679
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.9277545213699341,
      "learning_rate": 0.0004250072961638096,
      "loss": 2.3672,
      "step": 83680
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.8316710591316223,
      "learning_rate": 0.0004250035776340295,
      "loss": 2.2286,
      "step": 83681
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.7592151165008545,
      "learning_rate": 0.0004249998590810088,
      "loss": 2.1741,
      "step": 83682
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.9879023432731628,
      "learning_rate": 0.0004249961405047483,
      "loss": 2.1872,
      "step": 83683
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.905515730381012,
      "learning_rate": 0.0004249924219052488,
      "loss": 2.1558,
      "step": 83684
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.7491099834442139,
      "learning_rate": 0.00042498870328251074,
      "loss": 1.9285,
      "step": 83685
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.9025073051452637,
      "learning_rate": 0.0004249849846365349,
      "loss": 2.6254,
      "step": 83686
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.8853341341018677,
      "learning_rate": 0.000424981265967322,
      "loss": 2.3689,
      "step": 83687
    },
    {
      "epoch": 1.09,
      "grad_norm": 1.0011022090911865,
      "learning_rate": 0.00042497754727487277,
      "loss": 2.3566,
      "step": 83688
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.7993855476379395,
      "learning_rate": 0.0004249738285591878,
      "loss": 2.2737,
      "step": 83689
    },
    {
      "epoch": 1.09,
      "grad_norm": 1.1017149686813354,
      "learning_rate": 0.00042497010982026794,
      "loss": 2.1362,
      "step": 83690
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.7764410376548767,
      "learning_rate": 0.00042496639105811366,
      "loss": 2.2044,
      "step": 83691
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.7833431959152222,
      "learning_rate": 0.00042496267227272574,
      "loss": 2.1598,
      "step": 83692
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.825342059135437,
      "learning_rate": 0.00042495895346410505,
      "loss": 2.0961,
      "step": 83693
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.7825318574905396,
      "learning_rate": 0.0004249552346322521,
      "loss": 2.0097,
      "step": 83694
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.9056727290153503,
      "learning_rate": 0.0004249515157771674,
      "loss": 2.0116,
      "step": 83695
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.8674629330635071,
      "learning_rate": 0.00042494779689885215,
      "loss": 2.4167,
      "step": 83696
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.7632576823234558,
      "learning_rate": 0.00042494407799730655,
      "loss": 2.0933,
      "step": 83697
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.8506773114204407,
      "learning_rate": 0.0004249403590725316,
      "loss": 2.3059,
      "step": 83698
    },
    {
      "epoch": 1.09,
      "grad_norm": 1.002406120300293,
      "learning_rate": 0.0004249366401245278,
      "loss": 2.2331,
      "step": 83699
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.8852970004081726,
      "learning_rate": 0.00042493292115329594,
      "loss": 2.055,
      "step": 83700
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.8577560186386108,
      "learning_rate": 0.0004249292021588367,
      "loss": 2.5606,
      "step": 83701
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.9309280514717102,
      "learning_rate": 0.00042492548314115077,
      "loss": 2.1237,
      "step": 83702
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.9081020355224609,
      "learning_rate": 0.0004249217641002389,
      "loss": 2.1738,
      "step": 83703
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.7965078353881836,
      "learning_rate": 0.00042491804503610157,
      "loss": 2.1808,
      "step": 83704
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.8243275880813599,
      "learning_rate": 0.00042491432594873976,
      "loss": 2.3053,
      "step": 83705
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.7938734889030457,
      "learning_rate": 0.0004249106068381539,
      "loss": 2.4197,
      "step": 83706
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.7587926387786865,
      "learning_rate": 0.0004249068877043448,
      "loss": 2.1493,
      "step": 83707
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.9450257420539856,
      "learning_rate": 0.0004249031685473133,
      "loss": 2.1777,
      "step": 83708
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.8404183387756348,
      "learning_rate": 0.0004248994493670598,
      "loss": 2.2841,
      "step": 83709
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.7902978658676147,
      "learning_rate": 0.00042489573016358514,
      "loss": 2.1486,
      "step": 83710
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.8962318301200867,
      "learning_rate": 0.0004248920109368901,
      "loss": 2.1963,
      "step": 83711
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.8501601219177246,
      "learning_rate": 0.00042488829168697525,
      "loss": 1.9406,
      "step": 83712
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.8804303407669067,
      "learning_rate": 0.0004248845724138412,
      "loss": 2.0022,
      "step": 83713
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.8099399209022522,
      "learning_rate": 0.00042488085311748885,
      "loss": 2.2622,
      "step": 83714
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.906049370765686,
      "learning_rate": 0.00042487713379791886,
      "loss": 2.3782,
      "step": 83715
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.8686925768852234,
      "learning_rate": 0.0004248734144551317,
      "loss": 2.2042,
      "step": 83716
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.8401151895523071,
      "learning_rate": 0.0004248696950891283,
      "loss": 2.3902,
      "step": 83717
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.7852853536605835,
      "learning_rate": 0.0004248659756999093,
      "loss": 2.3107,
      "step": 83718
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.8863609433174133,
      "learning_rate": 0.0004248622562874753,
      "loss": 2.1961,
      "step": 83719
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.8598996996879578,
      "learning_rate": 0.00042485853685182706,
      "loss": 2.0857,
      "step": 83720
    },
    {
      "epoch": 1.09,
      "grad_norm": 1.1245100498199463,
      "learning_rate": 0.0004248548173929653,
      "loss": 2.0774,
      "step": 83721
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.7164698243141174,
      "learning_rate": 0.0004248510979108906,
      "loss": 2.3523,
      "step": 83722
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.835904598236084,
      "learning_rate": 0.0004248473784056037,
      "loss": 2.1505,
      "step": 83723
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.989658772945404,
      "learning_rate": 0.00042484365887710543,
      "loss": 2.3026,
      "step": 83724
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.8188398480415344,
      "learning_rate": 0.0004248399393253963,
      "loss": 2.2953,
      "step": 83725
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.9875428080558777,
      "learning_rate": 0.00042483621975047703,
      "loss": 2.1732,
      "step": 83726
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.7782143354415894,
      "learning_rate": 0.0004248325001523485,
      "loss": 2.2429,
      "step": 83727
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.8478619456291199,
      "learning_rate": 0.00042482878053101117,
      "loss": 2.1719,
      "step": 83728
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.8671878576278687,
      "learning_rate": 0.0004248250608864657,
      "loss": 2.2926,
      "step": 83729
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.9300147891044617,
      "learning_rate": 0.00042482134121871304,
      "loss": 2.1934,
      "step": 83730
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.8534736037254333,
      "learning_rate": 0.0004248176215277537,
      "loss": 2.1281,
      "step": 83731
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.8149329423904419,
      "learning_rate": 0.0004248139018135884,
      "loss": 2.262,
      "step": 83732
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.9232020974159241,
      "learning_rate": 0.00042481018207621784,
      "loss": 2.1122,
      "step": 83733
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.8748502731323242,
      "learning_rate": 0.0004248064623156427,
      "loss": 2.3349,
      "step": 83734
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.8759885430335999,
      "learning_rate": 0.0004248027425318637,
      "loss": 2.191,
      "step": 83735
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.7750450372695923,
      "learning_rate": 0.0004247990227248815,
      "loss": 2.0939,
      "step": 83736
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.8173167109489441,
      "learning_rate": 0.00042479530289469674,
      "loss": 2.3652,
      "step": 83737
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.849482536315918,
      "learning_rate": 0.00042479158304131027,
      "loss": 2.3881,
      "step": 83738
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.8493480086326599,
      "learning_rate": 0.0004247878631647227,
      "loss": 2.0538,
      "step": 83739
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.7928386926651001,
      "learning_rate": 0.0004247841432649347,
      "loss": 2.2728,
      "step": 83740
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.9121399521827698,
      "learning_rate": 0.00042478042334194693,
      "loss": 2.1453,
      "step": 83741
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.8872672915458679,
      "learning_rate": 0.0004247767033957602,
      "loss": 2.0569,
      "step": 83742
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.794166624546051,
      "learning_rate": 0.000424772983426375,
      "loss": 2.3357,
      "step": 83743
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.8764997124671936,
      "learning_rate": 0.00042476926343379217,
      "loss": 2.0158,
      "step": 83744
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.7574120163917542,
      "learning_rate": 0.0004247655434180125,
      "loss": 2.1421,
      "step": 83745
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.8690786957740784,
      "learning_rate": 0.00042476182337903646,
      "loss": 2.3547,
      "step": 83746
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.7922824025154114,
      "learning_rate": 0.00042475810331686484,
      "loss": 2.0174,
      "step": 83747
    },
    {
      "epoch": 1.09,
      "grad_norm": 1.041014552116394,
      "learning_rate": 0.0004247543832314983,
      "loss": 2.3209,
      "step": 83748
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.8160675764083862,
      "learning_rate": 0.0004247506631229377,
      "loss": 2.2559,
      "step": 83749
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.8406216502189636,
      "learning_rate": 0.0004247469429911835,
      "loss": 2.0267,
      "step": 83750
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.7723656892776489,
      "learning_rate": 0.0004247432228362365,
      "loss": 2.0496,
      "step": 83751
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.7771115303039551,
      "learning_rate": 0.0004247395026580974,
      "loss": 2.2576,
      "step": 83752
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.7157934308052063,
      "learning_rate": 0.0004247357824567669,
      "loss": 2.2823,
      "step": 83753
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.8478019833564758,
      "learning_rate": 0.0004247320622322456,
      "loss": 2.345,
      "step": 83754
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.7723917961120605,
      "learning_rate": 0.0004247283419845342,
      "loss": 2.1771,
      "step": 83755
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.8151699900627136,
      "learning_rate": 0.0004247246217136336,
      "loss": 2.1909,
      "step": 83756
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.8274135589599609,
      "learning_rate": 0.00042472090141954427,
      "loss": 2.434,
      "step": 83757
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.7326134443283081,
      "learning_rate": 0.00042471718110226694,
      "loss": 2.1451,
      "step": 83758
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.7378242611885071,
      "learning_rate": 0.00042471346076180237,
      "loss": 2.2918,
      "step": 83759
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.8428141474723816,
      "learning_rate": 0.0004247097403981512,
      "loss": 2.2515,
      "step": 83760
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.8092107176780701,
      "learning_rate": 0.0004247060200113141,
      "loss": 2.1144,
      "step": 83761
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.8443740606307983,
      "learning_rate": 0.00042470229960129186,
      "loss": 2.3542,
      "step": 83762
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.7699954509735107,
      "learning_rate": 0.0004246985791680851,
      "loss": 2.1083,
      "step": 83763
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.8724813461303711,
      "learning_rate": 0.00042469485871169445,
      "loss": 2.1743,
      "step": 83764
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.8429822325706482,
      "learning_rate": 0.0004246911382321207,
      "loss": 2.2627,
      "step": 83765
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.8137221932411194,
      "learning_rate": 0.00042468741772936464,
      "loss": 2.2511,
      "step": 83766
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.8341773748397827,
      "learning_rate": 0.0004246836972034267,
      "loss": 2.3264,
      "step": 83767
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.7819408178329468,
      "learning_rate": 0.0004246799766543077,
      "loss": 2.1018,
      "step": 83768
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.8048670291900635,
      "learning_rate": 0.00042467625608200843,
      "loss": 2.3209,
      "step": 83769
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.8283469676971436,
      "learning_rate": 0.00042467253548652943,
      "loss": 2.2381,
      "step": 83770
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.7493167519569397,
      "learning_rate": 0.00042466881486787143,
      "loss": 2.0998,
      "step": 83771
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.9032520651817322,
      "learning_rate": 0.00042466509422603525,
      "loss": 2.0737,
      "step": 83772
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.8461766242980957,
      "learning_rate": 0.00042466137356102137,
      "loss": 2.2227,
      "step": 83773
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.8331999182701111,
      "learning_rate": 0.0004246576528728306,
      "loss": 2.1698,
      "step": 83774
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.8005732297897339,
      "learning_rate": 0.0004246539321614636,
      "loss": 2.2087,
      "step": 83775
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.808717668056488,
      "learning_rate": 0.0004246502114269212,
      "loss": 2.2236,
      "step": 83776
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.8586376905441284,
      "learning_rate": 0.0004246464906692039,
      "loss": 2.1295,
      "step": 83777
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.7874640226364136,
      "learning_rate": 0.0004246427698883124,
      "loss": 2.0253,
      "step": 83778
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.803445041179657,
      "learning_rate": 0.00042463904908424756,
      "loss": 2.097,
      "step": 83779
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.8637959957122803,
      "learning_rate": 0.0004246353282570099,
      "loss": 2.2516,
      "step": 83780
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.9634827375411987,
      "learning_rate": 0.0004246316074066002,
      "loss": 2.1378,
      "step": 83781
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.8826581239700317,
      "learning_rate": 0.00042462788653301915,
      "loss": 2.4129,
      "step": 83782
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.902239203453064,
      "learning_rate": 0.00042462416563626736,
      "loss": 2.1845,
      "step": 83783
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.7465793490409851,
      "learning_rate": 0.00042462044471634563,
      "loss": 2.0378,
      "step": 83784
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.8142768144607544,
      "learning_rate": 0.00042461672377325466,
      "loss": 2.056,
      "step": 83785
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.8227904438972473,
      "learning_rate": 0.00042461300280699504,
      "loss": 2.2756,
      "step": 83786
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.7960349321365356,
      "learning_rate": 0.00042460928181756744,
      "loss": 2.2226,
      "step": 83787
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.8196020722389221,
      "learning_rate": 0.0004246055608049727,
      "loss": 2.2025,
      "step": 83788
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.7867741584777832,
      "learning_rate": 0.00042460183976921145,
      "loss": 2.1531,
      "step": 83789
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.7447633147239685,
      "learning_rate": 0.0004245981187102844,
      "loss": 1.9798,
      "step": 83790
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.8344374299049377,
      "learning_rate": 0.00042459439762819207,
      "loss": 2.237,
      "step": 83791
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.7640222907066345,
      "learning_rate": 0.00042459067652293535,
      "loss": 2.261,
      "step": 83792
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.8521015048027039,
      "learning_rate": 0.00042458695539451493,
      "loss": 2.0824,
      "step": 83793
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.9077531099319458,
      "learning_rate": 0.00042458323424293145,
      "loss": 2.3949,
      "step": 83794
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.7912502884864807,
      "learning_rate": 0.0004245795130681855,
      "loss": 2.161,
      "step": 83795
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.8127393126487732,
      "learning_rate": 0.0004245757918702779,
      "loss": 2.3655,
      "step": 83796
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.9553943276405334,
      "learning_rate": 0.0004245720706492093,
      "loss": 2.2698,
      "step": 83797
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.7637428045272827,
      "learning_rate": 0.00042456834940498043,
      "loss": 2.0729,
      "step": 83798
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.8116704821586609,
      "learning_rate": 0.00042456462813759196,
      "loss": 2.1955,
      "step": 83799
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.9163693189620972,
      "learning_rate": 0.0004245609068470446,
      "loss": 2.465,
      "step": 83800
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.8186958432197571,
      "learning_rate": 0.0004245571855333389,
      "loss": 2.2772,
      "step": 83801
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.8737177848815918,
      "learning_rate": 0.0004245534641964758,
      "loss": 2.2753,
      "step": 83802
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.8757508993148804,
      "learning_rate": 0.00042454974283645584,
      "loss": 2.0549,
      "step": 83803
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.7949960827827454,
      "learning_rate": 0.0004245460214532796,
      "loss": 2.1293,
      "step": 83804
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.8075425028800964,
      "learning_rate": 0.0004245423000469481,
      "loss": 2.0666,
      "step": 83805
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.758076012134552,
      "learning_rate": 0.0004245385786174617,
      "loss": 2.281,
      "step": 83806
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.8644261360168457,
      "learning_rate": 0.00042453485716482123,
      "loss": 2.1738,
      "step": 83807
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.9775067567825317,
      "learning_rate": 0.0004245311356890274,
      "loss": 2.1179,
      "step": 83808
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.8482362031936646,
      "learning_rate": 0.0004245274141900809,
      "loss": 2.2217,
      "step": 83809
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.901344358921051,
      "learning_rate": 0.00042452369266798243,
      "loss": 2.391,
      "step": 83810
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.725713849067688,
      "learning_rate": 0.00042451997112273257,
      "loss": 2.2439,
      "step": 83811
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.7363094091415405,
      "learning_rate": 0.0004245162495543322,
      "loss": 1.9928,
      "step": 83812
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.8155255317687988,
      "learning_rate": 0.00042451252796278183,
      "loss": 1.971,
      "step": 83813
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.7305505871772766,
      "learning_rate": 0.00042450880634808225,
      "loss": 2.2729,
      "step": 83814
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.8806734681129456,
      "learning_rate": 0.00042450508471023415,
      "loss": 2.2057,
      "step": 83815
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.8253635764122009,
      "learning_rate": 0.0004245013630492383,
      "loss": 2.1213,
      "step": 83816
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.7696815133094788,
      "learning_rate": 0.0004244976413650951,
      "loss": 2.1095,
      "step": 83817
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.9155309200286865,
      "learning_rate": 0.00042449391965780555,
      "loss": 2.5051,
      "step": 83818
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.9718877077102661,
      "learning_rate": 0.0004244901979273702,
      "loss": 2.2336,
      "step": 83819
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.8566288352012634,
      "learning_rate": 0.00042448647617378976,
      "loss": 2.1544,
      "step": 83820
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.850647509098053,
      "learning_rate": 0.000424482754397065,
      "loss": 2.206,
      "step": 83821
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.9853324294090271,
      "learning_rate": 0.0004244790325971965,
      "loss": 2.2228,
      "step": 83822
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.8100437521934509,
      "learning_rate": 0.000424475310774185,
      "loss": 2.2284,
      "step": 83823
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.8779963254928589,
      "learning_rate": 0.0004244715889280312,
      "loss": 2.1908,
      "step": 83824
    },
    {
      "epoch": 1.09,
      "grad_norm": 1.1292085647583008,
      "learning_rate": 0.0004244678670587358,
      "loss": 2.1237,
      "step": 83825
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.8106388449668884,
      "learning_rate": 0.0004244641451662994,
      "loss": 2.2136,
      "step": 83826
    },
    {
      "epoch": 1.09,
      "grad_norm": 1.0018818378448486,
      "learning_rate": 0.00042446042325072287,
      "loss": 2.2293,
      "step": 83827
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.93922358751297,
      "learning_rate": 0.0004244567013120067,
      "loss": 2.1484,
      "step": 83828
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.7823372483253479,
      "learning_rate": 0.0004244529793501517,
      "loss": 2.1405,
      "step": 83829
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.7431838512420654,
      "learning_rate": 0.00042444925736515856,
      "loss": 2.2261,
      "step": 83830
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.9930073618888855,
      "learning_rate": 0.00042444553535702793,
      "loss": 2.2293,
      "step": 83831
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.8039855360984802,
      "learning_rate": 0.0004244418133257606,
      "loss": 2.1265,
      "step": 83832
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.8304870128631592,
      "learning_rate": 0.0004244380912713571,
      "loss": 2.1085,
      "step": 83833
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.872923731803894,
      "learning_rate": 0.00042443436919381827,
      "loss": 2.0945,
      "step": 83834
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.958932638168335,
      "learning_rate": 0.0004244306470931447,
      "loss": 2.2894,
      "step": 83835
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.9476755857467651,
      "learning_rate": 0.00042442692496933717,
      "loss": 2.1468,
      "step": 83836
    },
    {
      "epoch": 1.09,
      "grad_norm": 1.2232571840286255,
      "learning_rate": 0.0004244232028223963,
      "loss": 2.1199,
      "step": 83837
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.9486807584762573,
      "learning_rate": 0.0004244194806523228,
      "loss": 2.3213,
      "step": 83838
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.8818849325180054,
      "learning_rate": 0.0004244157584591173,
      "loss": 2.1987,
      "step": 83839
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.7996180057525635,
      "learning_rate": 0.0004244120362427807,
      "loss": 2.0217,
      "step": 83840
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.8574355244636536,
      "learning_rate": 0.0004244083140033135,
      "loss": 2.0488,
      "step": 83841
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.8872122168540955,
      "learning_rate": 0.00042440459174071646,
      "loss": 2.1432,
      "step": 83842
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.7785727977752686,
      "learning_rate": 0.00042440086945499017,
      "loss": 2.2096,
      "step": 83843
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.8097733855247498,
      "learning_rate": 0.0004243971471461355,
      "loss": 2.263,
      "step": 83844
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.8964252471923828,
      "learning_rate": 0.00042439342481415305,
      "loss": 2.38,
      "step": 83845
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.7842788100242615,
      "learning_rate": 0.0004243897024590435,
      "loss": 2.1329,
      "step": 83846
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.7387239336967468,
      "learning_rate": 0.0004243859800808075,
      "loss": 2.3116,
      "step": 83847
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.8003196120262146,
      "learning_rate": 0.0004243822576794459,
      "loss": 2.422,
      "step": 83848
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.8905907273292542,
      "learning_rate": 0.0004243785352549593,
      "loss": 2.3451,
      "step": 83849
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.8075355291366577,
      "learning_rate": 0.0004243748128073482,
      "loss": 2.1848,
      "step": 83850
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.7824490070343018,
      "learning_rate": 0.00042437109033661367,
      "loss": 2.2499,
      "step": 83851
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.7811213135719299,
      "learning_rate": 0.0004243673678427561,
      "loss": 2.1429,
      "step": 83852
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.8062782883644104,
      "learning_rate": 0.0004243636453257763,
      "loss": 2.4215,
      "step": 83853
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.8792677521705627,
      "learning_rate": 0.000424359922785675,
      "loss": 2.0933,
      "step": 83854
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.8284580707550049,
      "learning_rate": 0.00042435620022245285,
      "loss": 2.3518,
      "step": 83855
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.7947087287902832,
      "learning_rate": 0.0004243524776361105,
      "loss": 2.0411,
      "step": 83856
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.7865503430366516,
      "learning_rate": 0.0004243487550266486,
      "loss": 2.1296,
      "step": 83857
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.839110255241394,
      "learning_rate": 0.0004243450323940681,
      "loss": 2.2501,
      "step": 83858
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.9219321012496948,
      "learning_rate": 0.00042434130973836935,
      "loss": 2.1478,
      "step": 83859
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.7524553537368774,
      "learning_rate": 0.0004243375870595532,
      "loss": 2.25,
      "step": 83860
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.8905481696128845,
      "learning_rate": 0.0004243338643576205,
      "loss": 2.1215,
      "step": 83861
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.8045872449874878,
      "learning_rate": 0.00042433014163257166,
      "loss": 1.829,
      "step": 83862
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.791265070438385,
      "learning_rate": 0.0004243264188844075,
      "loss": 1.9873,
      "step": 83863
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.8196846842765808,
      "learning_rate": 0.0004243226961131288,
      "loss": 2.1659,
      "step": 83864
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.9195860624313354,
      "learning_rate": 0.0004243189733187361,
      "loss": 2.2599,
      "step": 83865
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.8241574168205261,
      "learning_rate": 0.00042431525050123014,
      "loss": 2.1714,
      "step": 83866
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.8164641261100769,
      "learning_rate": 0.0004243115276606117,
      "loss": 2.3284,
      "step": 83867
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.9181982278823853,
      "learning_rate": 0.00042430780479688133,
      "loss": 2.2372,
      "step": 83868
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.8436444401741028,
      "learning_rate": 0.00042430408191003986,
      "loss": 1.9385,
      "step": 83869
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.8180898427963257,
      "learning_rate": 0.00042430035900008786,
      "loss": 2.0493,
      "step": 83870
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.7840824127197266,
      "learning_rate": 0.0004242966360670261,
      "loss": 2.1563,
      "step": 83871
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.9456338882446289,
      "learning_rate": 0.0004242929131108552,
      "loss": 2.0275,
      "step": 83872
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.8087841272354126,
      "learning_rate": 0.000424289190131576,
      "loss": 2.0333,
      "step": 83873
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.821679413318634,
      "learning_rate": 0.00042428546712918895,
      "loss": 2.3532,
      "step": 83874
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.7429912686347961,
      "learning_rate": 0.000424281744103695,
      "loss": 2.2629,
      "step": 83875
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.838593602180481,
      "learning_rate": 0.0004242780210550947,
      "loss": 2.2226,
      "step": 83876
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.8204081058502197,
      "learning_rate": 0.0004242742979833888,
      "loss": 2.1585,
      "step": 83877
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.816275417804718,
      "learning_rate": 0.00042427057488857785,
      "loss": 2.1909,
      "step": 83878
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.8123052716255188,
      "learning_rate": 0.00042426685177066287,
      "loss": 2.0737,
      "step": 83879
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.8231617212295532,
      "learning_rate": 0.0004242631286296442,
      "loss": 2.1741,
      "step": 83880
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.8861398100852966,
      "learning_rate": 0.0004242594054655226,
      "loss": 2.3506,
      "step": 83881
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.8182310461997986,
      "learning_rate": 0.0004242556822782989,
      "loss": 2.0992,
      "step": 83882
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.8703120946884155,
      "learning_rate": 0.0004242519590679737,
      "loss": 2.325,
      "step": 83883
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.8916143774986267,
      "learning_rate": 0.00042424823583454775,
      "loss": 2.0791,
      "step": 83884
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.8124226927757263,
      "learning_rate": 0.00042424451257802175,
      "loss": 2.2653,
      "step": 83885
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.8165131211280823,
      "learning_rate": 0.0004242407892983962,
      "loss": 2.1641,
      "step": 83886
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.7961296439170837,
      "learning_rate": 0.0004242370659956721,
      "loss": 2.1561,
      "step": 83887
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.7946900129318237,
      "learning_rate": 0.0004242333426698499,
      "loss": 2.0571,
      "step": 83888
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.8223022222518921,
      "learning_rate": 0.00042422961932093046,
      "loss": 2.1987,
      "step": 83889
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.7931625247001648,
      "learning_rate": 0.0004242258959489143,
      "loss": 2.0994,
      "step": 83890
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.8313621878623962,
      "learning_rate": 0.0004242221725538023,
      "loss": 2.3679,
      "step": 83891
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.8107573390007019,
      "learning_rate": 0.0004242184491355949,
      "loss": 2.3622,
      "step": 83892
    },
    {
      "epoch": 1.09,
      "grad_norm": 1.1875841617584229,
      "learning_rate": 0.00042421472569429303,
      "loss": 2.2507,
      "step": 83893
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.8409885168075562,
      "learning_rate": 0.00042421100222989733,
      "loss": 2.1193,
      "step": 83894
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.8080668449401855,
      "learning_rate": 0.0004242072787424084,
      "loss": 2.1425,
      "step": 83895
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.7488918304443359,
      "learning_rate": 0.000424203555231827,
      "loss": 2.2389,
      "step": 83896
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.763806164264679,
      "learning_rate": 0.00042419983169815386,
      "loss": 2.092,
      "step": 83897
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.7293002009391785,
      "learning_rate": 0.00042419610814138956,
      "loss": 2.0947,
      "step": 83898
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.8279670476913452,
      "learning_rate": 0.0004241923845615349,
      "loss": 2.2084,
      "step": 83899
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.8271707892417908,
      "learning_rate": 0.0004241886609585906,
      "loss": 2.1033,
      "step": 83900
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.84909987449646,
      "learning_rate": 0.0004241849373325572,
      "loss": 2.2152,
      "step": 83901
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.8423967957496643,
      "learning_rate": 0.0004241812136834354,
      "loss": 2.3596,
      "step": 83902
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.7687011957168579,
      "learning_rate": 0.0004241774900112261,
      "loss": 2.1379,
      "step": 83903
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.89719158411026,
      "learning_rate": 0.0004241737663159299,
      "loss": 2.0261,
      "step": 83904
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.7644346952438354,
      "learning_rate": 0.0004241700425975473,
      "loss": 2.2304,
      "step": 83905
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.7764192223548889,
      "learning_rate": 0.00042416631885607925,
      "loss": 2.2844,
      "step": 83906
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.8647658824920654,
      "learning_rate": 0.0004241625950915263,
      "loss": 2.174,
      "step": 83907
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.8140792846679688,
      "learning_rate": 0.0004241588713038892,
      "loss": 2.1553,
      "step": 83908
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.7001492977142334,
      "learning_rate": 0.0004241551474931685,
      "loss": 2.0183,
      "step": 83909
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.8177470564842224,
      "learning_rate": 0.0004241514236593652,
      "loss": 2.3762,
      "step": 83910
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.8003684282302856,
      "learning_rate": 0.0004241476998024797,
      "loss": 2.1229,
      "step": 83911
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.7876906991004944,
      "learning_rate": 0.0004241439759225128,
      "loss": 2.0713,
      "step": 83912
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.7938297390937805,
      "learning_rate": 0.0004241402520194653,
      "loss": 2.2095,
      "step": 83913
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.7784068584442139,
      "learning_rate": 0.00042413652809333765,
      "loss": 2.1202,
      "step": 83914
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.790096640586853,
      "learning_rate": 0.0004241328041441307,
      "loss": 2.0199,
      "step": 83915
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.7416049242019653,
      "learning_rate": 0.0004241290801718452,
      "loss": 2.0618,
      "step": 83916
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.8587026596069336,
      "learning_rate": 0.00042412535617648167,
      "loss": 2.2135,
      "step": 83917
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.788128137588501,
      "learning_rate": 0.00042412163215804093,
      "loss": 2.301,
      "step": 83918
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.8178057074546814,
      "learning_rate": 0.0004241179081165237,
      "loss": 2.3373,
      "step": 83919
    },
    {
      "epoch": 1.09,
      "grad_norm": 1.0095806121826172,
      "learning_rate": 0.00042411418405193045,
      "loss": 2.1984,
      "step": 83920
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.77608323097229,
      "learning_rate": 0.00042411045996426216,
      "loss": 2.1497,
      "step": 83921
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.9008397459983826,
      "learning_rate": 0.0004241067358535194,
      "loss": 2.1829,
      "step": 83922
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.7699742317199707,
      "learning_rate": 0.0004241030117197028,
      "loss": 1.8951,
      "step": 83923
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.8623623251914978,
      "learning_rate": 0.0004240992875628131,
      "loss": 2.3493,
      "step": 83924
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.8280127644538879,
      "learning_rate": 0.00042409556338285104,
      "loss": 2.2333,
      "step": 83925
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.812484085559845,
      "learning_rate": 0.00042409183917981725,
      "loss": 2.3449,
      "step": 83926
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.7952091097831726,
      "learning_rate": 0.00042408811495371244,
      "loss": 2.0101,
      "step": 83927
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.7862403988838196,
      "learning_rate": 0.00042408439070453733,
      "loss": 2.0191,
      "step": 83928
    },
    {
      "epoch": 1.09,
      "grad_norm": 1.0044746398925781,
      "learning_rate": 0.00042408066643229256,
      "loss": 2.2896,
      "step": 83929
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.8050235509872437,
      "learning_rate": 0.00042407694213697885,
      "loss": 2.1308,
      "step": 83930
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.7963086366653442,
      "learning_rate": 0.0004240732178185969,
      "loss": 2.1648,
      "step": 83931
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.7651620507240295,
      "learning_rate": 0.00042406949347714745,
      "loss": 2.0512,
      "step": 83932
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.7714815735816956,
      "learning_rate": 0.00042406576911263106,
      "loss": 2.1643,
      "step": 83933
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.8729233741760254,
      "learning_rate": 0.0004240620447250485,
      "loss": 2.3076,
      "step": 83934
    },
    {
      "epoch": 1.09,
      "grad_norm": 1.0201448202133179,
      "learning_rate": 0.0004240583203144005,
      "loss": 2.2835,
      "step": 83935
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.8594640493392944,
      "learning_rate": 0.0004240545958806877,
      "loss": 2.3057,
      "step": 83936
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.7740029096603394,
      "learning_rate": 0.0004240508714239109,
      "loss": 2.1503,
      "step": 83937
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.8840016722679138,
      "learning_rate": 0.00042404714694407056,
      "loss": 1.9465,
      "step": 83938
    },
    {
      "epoch": 1.09,
      "grad_norm": 1.0046828985214233,
      "learning_rate": 0.0004240434224411676,
      "loss": 2.2345,
      "step": 83939
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.8727328777313232,
      "learning_rate": 0.0004240396979152026,
      "loss": 2.2315,
      "step": 83940
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.9241510033607483,
      "learning_rate": 0.0004240359733661763,
      "loss": 2.0546,
      "step": 83941
    },
    {
      "epoch": 1.09,
      "grad_norm": 1.0582531690597534,
      "learning_rate": 0.0004240322487940893,
      "loss": 2.0866,
      "step": 83942
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.7999302744865417,
      "learning_rate": 0.00042402852419894246,
      "loss": 2.261,
      "step": 83943
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.8224033117294312,
      "learning_rate": 0.0004240247995807363,
      "loss": 2.0396,
      "step": 83944
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.8497768640518188,
      "learning_rate": 0.0004240210749394717,
      "loss": 2.1384,
      "step": 83945
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.8198704123497009,
      "learning_rate": 0.00042401735027514917,
      "loss": 2.1665,
      "step": 83946
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.8614813089370728,
      "learning_rate": 0.00042401362558776946,
      "loss": 2.3751,
      "step": 83947
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.8055393099784851,
      "learning_rate": 0.0004240099008773332,
      "loss": 2.1028,
      "step": 83948
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.8001211881637573,
      "learning_rate": 0.0004240061761438413,
      "loss": 2.0929,
      "step": 83949
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.8830098509788513,
      "learning_rate": 0.0004240024513872942,
      "loss": 2.0721,
      "step": 83950
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.757026195526123,
      "learning_rate": 0.0004239987266076928,
      "loss": 2.1109,
      "step": 83951
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.9600433111190796,
      "learning_rate": 0.0004239950018050377,
      "loss": 2.2235,
      "step": 83952
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.7529473304748535,
      "learning_rate": 0.00042399127697932953,
      "loss": 2.2611,
      "step": 83953
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.8007368445396423,
      "learning_rate": 0.000423987552130569,
      "loss": 2.1198,
      "step": 83954
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.7467848062515259,
      "learning_rate": 0.000423983827258757,
      "loss": 2.0476,
      "step": 83955
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.8654048442840576,
      "learning_rate": 0.0004239801023638939,
      "loss": 2.5531,
      "step": 83956
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.7522709965705872,
      "learning_rate": 0.0004239763774459806,
      "loss": 2.1958,
      "step": 83957
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.8314862847328186,
      "learning_rate": 0.00042397265250501784,
      "loss": 2.1828,
      "step": 83958
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.8060296773910522,
      "learning_rate": 0.0004239689275410062,
      "loss": 2.0746,
      "step": 83959
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.858094334602356,
      "learning_rate": 0.00042396520255394627,
      "loss": 2.2141,
      "step": 83960
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.7830691933631897,
      "learning_rate": 0.000423961477543839,
      "loss": 2.2041,
      "step": 83961
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.9419021010398865,
      "learning_rate": 0.0004239577525106848,
      "loss": 2.1279,
      "step": 83962
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.940919816493988,
      "learning_rate": 0.0004239540274544847,
      "loss": 2.4123,
      "step": 83963
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.8180562853813171,
      "learning_rate": 0.00042395030237523914,
      "loss": 2.1304,
      "step": 83964
    },
    {
      "epoch": 1.09,
      "grad_norm": 1.1894856691360474,
      "learning_rate": 0.00042394657727294887,
      "loss": 2.1254,
      "step": 83965
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.9025453329086304,
      "learning_rate": 0.00042394285214761453,
      "loss": 1.87,
      "step": 83966
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.7877814769744873,
      "learning_rate": 0.00042393912699923705,
      "loss": 2.2023,
      "step": 83967
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.9446331858634949,
      "learning_rate": 0.00042393540182781675,
      "loss": 2.3747,
      "step": 83968
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.8694626688957214,
      "learning_rate": 0.0004239316766333546,
      "loss": 2.2348,
      "step": 83969
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.8262307643890381,
      "learning_rate": 0.00042392795141585126,
      "loss": 2.2556,
      "step": 83970
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.7825987935066223,
      "learning_rate": 0.00042392422617530727,
      "loss": 2.0201,
      "step": 83971
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.8619639277458191,
      "learning_rate": 0.0004239205009117235,
      "loss": 2.286,
      "step": 83972
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.9565032720565796,
      "learning_rate": 0.00042391677562510054,
      "loss": 2.3623,
      "step": 83973
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.7863060235977173,
      "learning_rate": 0.00042391305031543914,
      "loss": 2.1394,
      "step": 83974
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.7835385799407959,
      "learning_rate": 0.00042390932498273987,
      "loss": 2.1699,
      "step": 83975
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.8505398631095886,
      "learning_rate": 0.00042390559962700357,
      "loss": 2.2738,
      "step": 83976
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.7894539833068848,
      "learning_rate": 0.000423901874248231,
      "loss": 2.0954,
      "step": 83977
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.808785617351532,
      "learning_rate": 0.0004238981488464226,
      "loss": 2.1416,
      "step": 83978
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.8602547645568848,
      "learning_rate": 0.0004238944234215792,
      "loss": 2.1533,
      "step": 83979
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.8157057166099548,
      "learning_rate": 0.00042389069797370163,
      "loss": 2.1116,
      "step": 83980
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.8945170044898987,
      "learning_rate": 0.00042388697250279027,
      "loss": 2.3055,
      "step": 83981
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.7873409390449524,
      "learning_rate": 0.00042388324700884606,
      "loss": 2.2816,
      "step": 83982
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.7775593400001526,
      "learning_rate": 0.0004238795214918696,
      "loss": 2.3591,
      "step": 83983
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.8523383736610413,
      "learning_rate": 0.00042387579595186154,
      "loss": 2.1433,
      "step": 83984
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.8654635548591614,
      "learning_rate": 0.0004238720703888227,
      "loss": 2.0019,
      "step": 83985
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.8743982911109924,
      "learning_rate": 0.0004238683448027538,
      "loss": 2.1298,
      "step": 83986
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.8959804773330688,
      "learning_rate": 0.0004238646191936552,
      "loss": 2.344,
      "step": 83987
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.8575431704521179,
      "learning_rate": 0.000423860893561528,
      "loss": 2.1038,
      "step": 83988
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.7937473654747009,
      "learning_rate": 0.0004238571679063727,
      "loss": 2.3169,
      "step": 83989
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.9686006903648376,
      "learning_rate": 0.00042385344222819,
      "loss": 2.4059,
      "step": 83990
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.8017719388008118,
      "learning_rate": 0.0004238497165269806,
      "loss": 2.2556,
      "step": 83991
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.809762716293335,
      "learning_rate": 0.00042384599080274525,
      "loss": 2.1302,
      "step": 83992
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.782124400138855,
      "learning_rate": 0.00042384226505548446,
      "loss": 2.3844,
      "step": 83993
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.7764253616333008,
      "learning_rate": 0.0004238385392851992,
      "loss": 2.2314,
      "step": 83994
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.8228678107261658,
      "learning_rate": 0.00042383481349189,
      "loss": 2.2751,
      "step": 83995
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.7716154456138611,
      "learning_rate": 0.00042383108767555746,
      "loss": 2.1861,
      "step": 83996
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.8048495650291443,
      "learning_rate": 0.00042382736183620243,
      "loss": 2.1837,
      "step": 83997
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.8788069486618042,
      "learning_rate": 0.00042382363597382567,
      "loss": 2.2035,
      "step": 83998
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.7336362600326538,
      "learning_rate": 0.0004238199100884276,
      "loss": 2.0865,
      "step": 83999
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.8829684257507324,
      "learning_rate": 0.0004238161841800091,
      "loss": 2.3284,
      "step": 84000
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.8733056783676147,
      "learning_rate": 0.00042381245824857095,
      "loss": 2.1608,
      "step": 84001
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.8538524508476257,
      "learning_rate": 0.00042380873229411366,
      "loss": 2.3731,
      "step": 84002
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.8066157102584839,
      "learning_rate": 0.00042380500631663794,
      "loss": 2.0365,
      "step": 84003
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.9479788541793823,
      "learning_rate": 0.0004238012803161446,
      "loss": 2.3121,
      "step": 84004
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.802275538444519,
      "learning_rate": 0.0004237975542926342,
      "loss": 2.2699,
      "step": 84005
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.9666346907615662,
      "learning_rate": 0.00042379382824610755,
      "loss": 2.0852,
      "step": 84006
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.7668671607971191,
      "learning_rate": 0.00042379010217656534,
      "loss": 2.1038,
      "step": 84007
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.8112884163856506,
      "learning_rate": 0.0004237863760840081,
      "loss": 2.2702,
      "step": 84008
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.9548119306564331,
      "learning_rate": 0.00042378264996843665,
      "loss": 1.9301,
      "step": 84009
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.8082727193832397,
      "learning_rate": 0.00042377892382985174,
      "loss": 2.3214,
      "step": 84010
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.9687052369117737,
      "learning_rate": 0.00042377519766825393,
      "loss": 2.0961,
      "step": 84011
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.8460858464241028,
      "learning_rate": 0.000423771471483644,
      "loss": 2.2045,
      "step": 84012
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.8192710876464844,
      "learning_rate": 0.00042376774527602267,
      "loss": 2.1983,
      "step": 84013
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.7540285587310791,
      "learning_rate": 0.00042376401904539046,
      "loss": 2.3936,
      "step": 84014
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.7629066705703735,
      "learning_rate": 0.0004237602927917482,
      "loss": 1.9163,
      "step": 84015
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.7412903308868408,
      "learning_rate": 0.0004237565665150967,
      "loss": 1.923,
      "step": 84016
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.8847774267196655,
      "learning_rate": 0.0004237528402154364,
      "loss": 1.9825,
      "step": 84017
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.793796718120575,
      "learning_rate": 0.0004237491138927681,
      "loss": 2.1892,
      "step": 84018
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.8184792399406433,
      "learning_rate": 0.00042374538754709256,
      "loss": 2.2924,
      "step": 84019
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.8399141430854797,
      "learning_rate": 0.0004237416611784104,
      "loss": 2.3219,
      "step": 84020
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.8019628524780273,
      "learning_rate": 0.0004237379347867224,
      "loss": 2.1609,
      "step": 84021
    },
    {
      "epoch": 1.09,
      "grad_norm": 1.036953330039978,
      "learning_rate": 0.0004237342083720291,
      "loss": 2.1682,
      "step": 84022
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.7892622947692871,
      "learning_rate": 0.00042373048193433124,
      "loss": 2.0423,
      "step": 84023
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.8577033281326294,
      "learning_rate": 0.00042372675547362965,
      "loss": 2.1157,
      "step": 84024
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.7811818718910217,
      "learning_rate": 0.00042372302898992486,
      "loss": 2.3766,
      "step": 84025
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.7635430097579956,
      "learning_rate": 0.0004237193024832176,
      "loss": 2.1191,
      "step": 84026
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.8604491353034973,
      "learning_rate": 0.0004237155759535086,
      "loss": 2.2857,
      "step": 84027
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.7442436814308167,
      "learning_rate": 0.0004237118494007986,
      "loss": 2.1489,
      "step": 84028
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.8842418789863586,
      "learning_rate": 0.0004237081228250882,
      "loss": 2.2108,
      "step": 84029
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.756182074546814,
      "learning_rate": 0.00042370439622637805,
      "loss": 2.0918,
      "step": 84030
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.7788387537002563,
      "learning_rate": 0.000423700669604669,
      "loss": 2.2031,
      "step": 84031
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.8286453485488892,
      "learning_rate": 0.0004236969429599617,
      "loss": 2.4305,
      "step": 84032
    },
    {
      "epoch": 1.09,
      "grad_norm": 1.0905022621154785,
      "learning_rate": 0.00042369321629225667,
      "loss": 2.081,
      "step": 84033
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.7333146333694458,
      "learning_rate": 0.00042368948960155487,
      "loss": 2.3742,
      "step": 84034
    },
    {
      "epoch": 1.09,
      "grad_norm": 1.1206709146499634,
      "learning_rate": 0.00042368576288785676,
      "loss": 2.1509,
      "step": 84035
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.8552125692367554,
      "learning_rate": 0.0004236820361511632,
      "loss": 2.1227,
      "step": 84036
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.804329514503479,
      "learning_rate": 0.0004236783093914748,
      "loss": 2.1963,
      "step": 84037
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.747826099395752,
      "learning_rate": 0.00042367458260879223,
      "loss": 2.2142,
      "step": 84038
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.8620060086250305,
      "learning_rate": 0.0004236708558031162,
      "loss": 2.4312,
      "step": 84039
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.7754762172698975,
      "learning_rate": 0.00042366712897444757,
      "loss": 2.2105,
      "step": 84040
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.9049426913261414,
      "learning_rate": 0.00042366340212278674,
      "loss": 2.2732,
      "step": 84041
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.8279812335968018,
      "learning_rate": 0.0004236596752481346,
      "loss": 2.1771,
      "step": 84042
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.8531655669212341,
      "learning_rate": 0.00042365594835049177,
      "loss": 2.4436,
      "step": 84043
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.8465142846107483,
      "learning_rate": 0.000423652221429859,
      "loss": 2.1476,
      "step": 84044
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.7280889749526978,
      "learning_rate": 0.0004236484944862369,
      "loss": 2.1865,
      "step": 84045
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.8683210015296936,
      "learning_rate": 0.00042364476751962624,
      "loss": 1.9875,
      "step": 84046
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.8221520781517029,
      "learning_rate": 0.00042364104053002773,
      "loss": 2.1405,
      "step": 84047
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.8269962072372437,
      "learning_rate": 0.0004236373135174419,
      "loss": 2.2642,
      "step": 84048
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.8469903469085693,
      "learning_rate": 0.0004236335864818696,
      "loss": 2.4059,
      "step": 84049
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.7763018608093262,
      "learning_rate": 0.00042362985942331156,
      "loss": 2.0567,
      "step": 84050
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.7500926852226257,
      "learning_rate": 0.0004236261323417684,
      "loss": 2.1261,
      "step": 84051
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.9262427091598511,
      "learning_rate": 0.0004236224052372407,
      "loss": 2.2257,
      "step": 84052
    },
    {
      "epoch": 1.09,
      "grad_norm": 1.0381170511245728,
      "learning_rate": 0.00042361867810972935,
      "loss": 2.4192,
      "step": 84053
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.7923600673675537,
      "learning_rate": 0.0004236149509592349,
      "loss": 2.2715,
      "step": 84054
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.7948529124259949,
      "learning_rate": 0.00042361122378575816,
      "loss": 2.2156,
      "step": 84055
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.8758704662322998,
      "learning_rate": 0.00042360749658929974,
      "loss": 2.1035,
      "step": 84056
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.8029016852378845,
      "learning_rate": 0.0004236037693698603,
      "loss": 2.1436,
      "step": 84057
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.7909260392189026,
      "learning_rate": 0.0004236000421274406,
      "loss": 2.1675,
      "step": 84058
    },
    {
      "epoch": 1.09,
      "grad_norm": 1.4797697067260742,
      "learning_rate": 0.00042359631486204143,
      "loss": 2.0683,
      "step": 84059
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.8004805445671082,
      "learning_rate": 0.0004235925875736632,
      "loss": 2.1551,
      "step": 84060
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.9160372018814087,
      "learning_rate": 0.00042358886026230684,
      "loss": 2.3501,
      "step": 84061
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.7581096887588501,
      "learning_rate": 0.0004235851329279731,
      "loss": 2.2137,
      "step": 84062
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.756912112236023,
      "learning_rate": 0.0004235814055706624,
      "loss": 2.446,
      "step": 84063
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.79502934217453,
      "learning_rate": 0.00042357767819037556,
      "loss": 2.1463,
      "step": 84064
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.7945550680160522,
      "learning_rate": 0.0004235739507871135,
      "loss": 1.9498,
      "step": 84065
    },
    {
      "epoch": 1.09,
      "grad_norm": 1.0559204816818237,
      "learning_rate": 0.00042357022336087646,
      "loss": 2.2582,
      "step": 84066
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.8756598234176636,
      "learning_rate": 0.0004235664959116655,
      "loss": 2.3578,
      "step": 84067
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.8224808573722839,
      "learning_rate": 0.0004235627684394812,
      "loss": 2.0177,
      "step": 84068
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.8525593876838684,
      "learning_rate": 0.00042355904094432427,
      "loss": 2.2078,
      "step": 84069
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.8635746836662292,
      "learning_rate": 0.0004235553134261953,
      "loss": 2.1605,
      "step": 84070
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.8039475679397583,
      "learning_rate": 0.0004235515858850952,
      "loss": 2.2688,
      "step": 84071
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.8156174421310425,
      "learning_rate": 0.0004235478583210245,
      "loss": 2.1281,
      "step": 84072
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.8092705011367798,
      "learning_rate": 0.00042354413073398373,
      "loss": 2.3708,
      "step": 84073
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.7894917726516724,
      "learning_rate": 0.000423540403123974,
      "loss": 2.2503,
      "step": 84074
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.8319767713546753,
      "learning_rate": 0.0004235366754909957,
      "loss": 2.1678,
      "step": 84075
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.747377336025238,
      "learning_rate": 0.00042353294783504955,
      "loss": 2.1576,
      "step": 84076
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.7842516899108887,
      "learning_rate": 0.0004235292201561364,
      "loss": 2.3185,
      "step": 84077
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.8511514067649841,
      "learning_rate": 0.0004235254924542567,
      "loss": 2.1941,
      "step": 84078
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.7743651866912842,
      "learning_rate": 0.00042352176472941136,
      "loss": 2.3092,
      "step": 84079
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.7478960752487183,
      "learning_rate": 0.000423518036981601,
      "loss": 2.3061,
      "step": 84080
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.9113344550132751,
      "learning_rate": 0.00042351430921082627,
      "loss": 2.174,
      "step": 84081
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.7998510003089905,
      "learning_rate": 0.0004235105814170879,
      "loss": 2.2101,
      "step": 84082
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.8475746512413025,
      "learning_rate": 0.00042350685360038664,
      "loss": 2.3017,
      "step": 84083
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.8365733027458191,
      "learning_rate": 0.00042350312576072306,
      "loss": 2.2471,
      "step": 84084
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.8066908121109009,
      "learning_rate": 0.00042349939789809786,
      "loss": 2.0605,
      "step": 84085
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.7975754141807556,
      "learning_rate": 0.0004234956700125119,
      "loss": 2.0699,
      "step": 84086
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.7330639958381653,
      "learning_rate": 0.0004234919421039658,
      "loss": 2.2538,
      "step": 84087
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.8287904858589172,
      "learning_rate": 0.0004234882141724601,
      "loss": 2.1305,
      "step": 84088
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.8826467394828796,
      "learning_rate": 0.00042348448621799573,
      "loss": 2.1586,
      "step": 84089
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.8359854817390442,
      "learning_rate": 0.00042348075824057314,
      "loss": 2.3271,
      "step": 84090
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.7367323637008667,
      "learning_rate": 0.00042347703024019323,
      "loss": 2.037,
      "step": 84091
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.7817562222480774,
      "learning_rate": 0.0004234733022168566,
      "loss": 2.1621,
      "step": 84092
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.7748635411262512,
      "learning_rate": 0.0004234695741705639,
      "loss": 2.056,
      "step": 84093
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.7635265588760376,
      "learning_rate": 0.00042346584610131584,
      "loss": 2.1088,
      "step": 84094
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.90096116065979,
      "learning_rate": 0.00042346211800911327,
      "loss": 2.0664,
      "step": 84095
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.8231229782104492,
      "learning_rate": 0.00042345838989395673,
      "loss": 2.3534,
      "step": 84096
    },
    {
      "epoch": 1.09,
      "grad_norm": 0.8049889802932739,
      "learning_rate": 0.0004234546617558469,
      "loss": 2.1203,
      "step": 84097
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.9296826124191284,
      "learning_rate": 0.00042345093359478456,
      "loss": 2.1898,
      "step": 84098
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.9264127016067505,
      "learning_rate": 0.00042344720541077034,
      "loss": 2.3475,
      "step": 84099
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.8004593253135681,
      "learning_rate": 0.0004234434772038049,
      "loss": 2.1441,
      "step": 84100
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.870148777961731,
      "learning_rate": 0.00042343974897388905,
      "loss": 2.3156,
      "step": 84101
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.8194113969802856,
      "learning_rate": 0.00042343602072102344,
      "loss": 2.1625,
      "step": 84102
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.7649573087692261,
      "learning_rate": 0.0004234322924452087,
      "loss": 2.2448,
      "step": 84103
    },
    {
      "epoch": 1.1,
      "grad_norm": 1.0508028268814087,
      "learning_rate": 0.0004234285641464456,
      "loss": 2.2207,
      "step": 84104
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.8622140884399414,
      "learning_rate": 0.0004234248358247348,
      "loss": 2.192,
      "step": 84105
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.7283963561058044,
      "learning_rate": 0.0004234211074800769,
      "loss": 2.0185,
      "step": 84106
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.8195652961730957,
      "learning_rate": 0.0004234173791124728,
      "loss": 2.1905,
      "step": 84107
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.7762892842292786,
      "learning_rate": 0.000423413650721923,
      "loss": 2.0964,
      "step": 84108
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.8220419883728027,
      "learning_rate": 0.0004234099223084284,
      "loss": 2.1862,
      "step": 84109
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.8762941956520081,
      "learning_rate": 0.0004234061938719894,
      "loss": 2.221,
      "step": 84110
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.7899644374847412,
      "learning_rate": 0.00042340246541260697,
      "loss": 2.1566,
      "step": 84111
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.8561921715736389,
      "learning_rate": 0.0004233987369302816,
      "loss": 2.1813,
      "step": 84112
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.8119867444038391,
      "learning_rate": 0.00042339500842501423,
      "loss": 2.1034,
      "step": 84113
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.8670444488525391,
      "learning_rate": 0.00042339127989680524,
      "loss": 2.155,
      "step": 84114
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.8837101459503174,
      "learning_rate": 0.00042338755134565555,
      "loss": 2.1891,
      "step": 84115
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.7712082266807556,
      "learning_rate": 0.00042338382277156577,
      "loss": 2.1172,
      "step": 84116
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.8674396872520447,
      "learning_rate": 0.00042338009417453666,
      "loss": 2.0668,
      "step": 84117
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.8063734173774719,
      "learning_rate": 0.0004233763655545688,
      "loss": 2.0432,
      "step": 84118
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.9788609147071838,
      "learning_rate": 0.00042337263691166293,
      "loss": 2.1254,
      "step": 84119
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.8617166876792908,
      "learning_rate": 0.00042336890824581984,
      "loss": 2.1589,
      "step": 84120
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.872147262096405,
      "learning_rate": 0.0004233651795570401,
      "loss": 2.1989,
      "step": 84121
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.8417275547981262,
      "learning_rate": 0.00042336145084532437,
      "loss": 2.2862,
      "step": 84122
    },
    {
      "epoch": 1.1,
      "grad_norm": 1.0117363929748535,
      "learning_rate": 0.0004233577221106736,
      "loss": 2.1774,
      "step": 84123
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.8973205089569092,
      "learning_rate": 0.0004233539933530881,
      "loss": 2.1625,
      "step": 84124
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.9162908792495728,
      "learning_rate": 0.0004233502645725689,
      "loss": 2.316,
      "step": 84125
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.7728192806243896,
      "learning_rate": 0.0004233465357691165,
      "loss": 2.3384,
      "step": 84126
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.9680833220481873,
      "learning_rate": 0.00042334280694273166,
      "loss": 2.2907,
      "step": 84127
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.8359102010726929,
      "learning_rate": 0.00042333907809341504,
      "loss": 2.1477,
      "step": 84128
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.829475462436676,
      "learning_rate": 0.00042333534922116744,
      "loss": 2.0783,
      "step": 84129
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.9624640345573425,
      "learning_rate": 0.0004233316203259894,
      "loss": 2.0544,
      "step": 84130
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.9153508543968201,
      "learning_rate": 0.0004233278914078817,
      "loss": 2.0633,
      "step": 84131
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.7783464789390564,
      "learning_rate": 0.00042332416246684503,
      "loss": 2.2451,
      "step": 84132
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.8091954588890076,
      "learning_rate": 0.00042332043350288006,
      "loss": 2.0481,
      "step": 84133
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.8318037986755371,
      "learning_rate": 0.0004233167045159874,
      "loss": 2.048,
      "step": 84134
    },
    {
      "epoch": 1.1,
      "grad_norm": 1.1302071809768677,
      "learning_rate": 0.00042331297550616805,
      "loss": 2.2338,
      "step": 84135
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.6960658431053162,
      "learning_rate": 0.0004233092464734223,
      "loss": 2.2343,
      "step": 84136
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.8445890545845032,
      "learning_rate": 0.0004233055174177511,
      "loss": 2.1454,
      "step": 84137
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.9573836922645569,
      "learning_rate": 0.00042330178833915514,
      "loss": 2.231,
      "step": 84138
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.8448436856269836,
      "learning_rate": 0.0004232980592376349,
      "loss": 2.0619,
      "step": 84139
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.776850700378418,
      "learning_rate": 0.0004232943301131913,
      "loss": 2.0603,
      "step": 84140
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.8200838565826416,
      "learning_rate": 0.000423290600965825,
      "loss": 2.1825,
      "step": 84141
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.732678234577179,
      "learning_rate": 0.00042328687179553657,
      "loss": 2.407,
      "step": 84142
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.8409562110900879,
      "learning_rate": 0.00042328314260232686,
      "loss": 2.2775,
      "step": 84143
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.8259313106536865,
      "learning_rate": 0.0004232794133861965,
      "loss": 2.1662,
      "step": 84144
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.756714403629303,
      "learning_rate": 0.0004232756841471461,
      "loss": 2.1786,
      "step": 84145
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.8007122278213501,
      "learning_rate": 0.00042327195488517643,
      "loss": 2.0998,
      "step": 84146
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.7723026871681213,
      "learning_rate": 0.0004232682256002882,
      "loss": 2.0516,
      "step": 84147
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.7811474204063416,
      "learning_rate": 0.0004232644962924821,
      "loss": 2.0233,
      "step": 84148
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.7890741229057312,
      "learning_rate": 0.00042326076696175876,
      "loss": 2.1355,
      "step": 84149
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.728411078453064,
      "learning_rate": 0.000423257037608119,
      "loss": 2.1402,
      "step": 84150
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.7734509110450745,
      "learning_rate": 0.0004232533082315633,
      "loss": 2.11,
      "step": 84151
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.8068071007728577,
      "learning_rate": 0.00042324957883209255,
      "loss": 2.3744,
      "step": 84152
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.8480450510978699,
      "learning_rate": 0.0004232458494097074,
      "loss": 2.1415,
      "step": 84153
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.8266497850418091,
      "learning_rate": 0.00042324211996440847,
      "loss": 2.1721,
      "step": 84154
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.7703981399536133,
      "learning_rate": 0.00042323839049619657,
      "loss": 2.0859,
      "step": 84155
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.832246720790863,
      "learning_rate": 0.00042323466100507227,
      "loss": 2.3139,
      "step": 84156
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.7478784322738647,
      "learning_rate": 0.00042323093149103635,
      "loss": 2.281,
      "step": 84157
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.8037554025650024,
      "learning_rate": 0.0004232272019540894,
      "loss": 2.1153,
      "step": 84158
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.898987352848053,
      "learning_rate": 0.00042322347239423226,
      "loss": 2.3407,
      "step": 84159
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.9667907953262329,
      "learning_rate": 0.00042321974281146556,
      "loss": 2.101,
      "step": 84160
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.8349364399909973,
      "learning_rate": 0.00042321601320578994,
      "loss": 2.2536,
      "step": 84161
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.755726158618927,
      "learning_rate": 0.00042321228357720616,
      "loss": 2.1788,
      "step": 84162
    },
    {
      "epoch": 1.1,
      "grad_norm": 1.2095859050750732,
      "learning_rate": 0.0004232085539257149,
      "loss": 2.1023,
      "step": 84163
    },
    {
      "epoch": 1.1,
      "grad_norm": 1.0333623886108398,
      "learning_rate": 0.0004232048242513168,
      "loss": 2.171,
      "step": 84164
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.8090579509735107,
      "learning_rate": 0.00042320109455401273,
      "loss": 2.4788,
      "step": 84165
    },
    {
      "epoch": 1.1,
      "grad_norm": 1.222123384475708,
      "learning_rate": 0.00042319736483380314,
      "loss": 1.9277,
      "step": 84166
    },
    {
      "epoch": 1.1,
      "grad_norm": 1.266684651374817,
      "learning_rate": 0.0004231936350906888,
      "loss": 2.1396,
      "step": 84167
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.7076526880264282,
      "learning_rate": 0.00042318990532467057,
      "loss": 2.1811,
      "step": 84168
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.9865896105766296,
      "learning_rate": 0.0004231861755357489,
      "loss": 2.0746,
      "step": 84169
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.9828433990478516,
      "learning_rate": 0.0004231824457239246,
      "loss": 2.2307,
      "step": 84170
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.8540725708007812,
      "learning_rate": 0.00042317871588919836,
      "loss": 2.153,
      "step": 84171
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.8456507325172424,
      "learning_rate": 0.00042317498603157086,
      "loss": 2.2615,
      "step": 84172
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.7504143118858337,
      "learning_rate": 0.00042317125615104283,
      "loss": 2.1148,
      "step": 84173
    },
    {
      "epoch": 1.1,
      "grad_norm": 1.5205200910568237,
      "learning_rate": 0.000423167526247615,
      "loss": 2.2822,
      "step": 84174
    },
    {
      "epoch": 1.1,
      "grad_norm": 1.175592303276062,
      "learning_rate": 0.00042316379632128794,
      "loss": 2.0299,
      "step": 84175
    },
    {
      "epoch": 1.1,
      "grad_norm": 1.0781283378601074,
      "learning_rate": 0.00042316006637206237,
      "loss": 2.4116,
      "step": 84176
    },
    {
      "epoch": 1.1,
      "grad_norm": 1.051085114479065,
      "learning_rate": 0.00042315633639993905,
      "loss": 2.2056,
      "step": 84177
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.8675308227539062,
      "learning_rate": 0.0004231526064049187,
      "loss": 2.2502,
      "step": 84178
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.803625226020813,
      "learning_rate": 0.0004231488763870019,
      "loss": 1.9621,
      "step": 84179
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.8216928243637085,
      "learning_rate": 0.00042314514634618933,
      "loss": 1.8474,
      "step": 84180
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.9336234331130981,
      "learning_rate": 0.0004231414162824819,
      "loss": 2.1809,
      "step": 84181
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.9903097152709961,
      "learning_rate": 0.0004231376861958801,
      "loss": 2.382,
      "step": 84182
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.9274383187294006,
      "learning_rate": 0.0004231339560863846,
      "loss": 2.1091,
      "step": 84183
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.9789345860481262,
      "learning_rate": 0.0004231302259539963,
      "loss": 2.0873,
      "step": 84184
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.8597851991653442,
      "learning_rate": 0.00042312649579871574,
      "loss": 2.2237,
      "step": 84185
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.8332050442695618,
      "learning_rate": 0.0004231227656205435,
      "loss": 2.4038,
      "step": 84186
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.7662329077720642,
      "learning_rate": 0.0004231190354194806,
      "loss": 2.0868,
      "step": 84187
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.9344552755355835,
      "learning_rate": 0.00042311530519552744,
      "loss": 2.0261,
      "step": 84188
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.8054141402244568,
      "learning_rate": 0.0004231115749486848,
      "loss": 2.1473,
      "step": 84189
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.7853482961654663,
      "learning_rate": 0.0004231078446789535,
      "loss": 1.9583,
      "step": 84190
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.8695282936096191,
      "learning_rate": 0.000423104114386334,
      "loss": 2.2699,
      "step": 84191
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.830295205116272,
      "learning_rate": 0.0004231003840708272,
      "loss": 1.9684,
      "step": 84192
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.8816559910774231,
      "learning_rate": 0.00042309665373243367,
      "loss": 2.3369,
      "step": 84193
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.7438182234764099,
      "learning_rate": 0.0004230929233711543,
      "loss": 1.9286,
      "step": 84194
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.7670204043388367,
      "learning_rate": 0.0004230891929869894,
      "loss": 2.2772,
      "step": 84195
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.8003865480422974,
      "learning_rate": 0.00042308546257994003,
      "loss": 2.4122,
      "step": 84196
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.7723062634468079,
      "learning_rate": 0.00042308173215000676,
      "loss": 2.1095,
      "step": 84197
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.7672511339187622,
      "learning_rate": 0.00042307800169719026,
      "loss": 2.1816,
      "step": 84198
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.8752302527427673,
      "learning_rate": 0.00042307427122149124,
      "loss": 2.3524,
      "step": 84199
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.9048312306404114,
      "learning_rate": 0.0004230705407229103,
      "loss": 2.3083,
      "step": 84200
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.814044713973999,
      "learning_rate": 0.00042306681020144835,
      "loss": 2.1213,
      "step": 84201
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.7587125301361084,
      "learning_rate": 0.0004230630796571059,
      "loss": 2.3193,
      "step": 84202
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.8215900659561157,
      "learning_rate": 0.00042305934908988367,
      "loss": 1.8843,
      "step": 84203
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.7880913615226746,
      "learning_rate": 0.0004230556184997824,
      "loss": 2.2053,
      "step": 84204
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.7059615850448608,
      "learning_rate": 0.00042305188788680284,
      "loss": 2.3167,
      "step": 84205
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.9421803951263428,
      "learning_rate": 0.00042304815725094553,
      "loss": 2.0959,
      "step": 84206
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.7488515973091125,
      "learning_rate": 0.0004230444265922112,
      "loss": 2.1313,
      "step": 84207
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.8201250433921814,
      "learning_rate": 0.00042304069591060075,
      "loss": 2.1275,
      "step": 84208
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.9516757726669312,
      "learning_rate": 0.00042303696520611453,
      "loss": 2.1793,
      "step": 84209
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.787350058555603,
      "learning_rate": 0.0004230332344787535,
      "loss": 2.148,
      "step": 84210
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.8608503937721252,
      "learning_rate": 0.00042302950372851827,
      "loss": 2.178,
      "step": 84211
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.8909386992454529,
      "learning_rate": 0.0004230257729554096,
      "loss": 2.1921,
      "step": 84212
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.9059951901435852,
      "learning_rate": 0.00042302204215942797,
      "loss": 2.1713,
      "step": 84213
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.9532684683799744,
      "learning_rate": 0.00042301831134057436,
      "loss": 2.3213,
      "step": 84214
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.7490933537483215,
      "learning_rate": 0.0004230145804988493,
      "loss": 2.2785,
      "step": 84215
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.9889556169509888,
      "learning_rate": 0.00042301084963425345,
      "loss": 2.2784,
      "step": 84216
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.9125022888183594,
      "learning_rate": 0.0004230071187467876,
      "loss": 2.4009,
      "step": 84217
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.8080856204032898,
      "learning_rate": 0.0004230033878364524,
      "loss": 2.0251,
      "step": 84218
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.8271961808204651,
      "learning_rate": 0.00042299965690324845,
      "loss": 2.0805,
      "step": 84219
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.8157861232757568,
      "learning_rate": 0.0004229959259471767,
      "loss": 2.2668,
      "step": 84220
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.844307005405426,
      "learning_rate": 0.00042299219496823765,
      "loss": 2.3921,
      "step": 84221
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.9433531761169434,
      "learning_rate": 0.00042298846396643196,
      "loss": 2.1967,
      "step": 84222
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.823186457157135,
      "learning_rate": 0.00042298473294176045,
      "loss": 2.1476,
      "step": 84223
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.8224675059318542,
      "learning_rate": 0.0004229810018942237,
      "loss": 1.9529,
      "step": 84224
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.8373864889144897,
      "learning_rate": 0.0004229772708238225,
      "loss": 2.242,
      "step": 84225
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.861903965473175,
      "learning_rate": 0.00042297353973055755,
      "loss": 2.3299,
      "step": 84226
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.8816404938697815,
      "learning_rate": 0.0004229698086144294,
      "loss": 1.9444,
      "step": 84227
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.7972382307052612,
      "learning_rate": 0.00042296607747543887,
      "loss": 2.1608,
      "step": 84228
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.7467907071113586,
      "learning_rate": 0.0004229623463135867,
      "loss": 2.0352,
      "step": 84229
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.8546947240829468,
      "learning_rate": 0.00042295861512887343,
      "loss": 2.1921,
      "step": 84230
    },
    {
      "epoch": 1.1,
      "grad_norm": 1.0481138229370117,
      "learning_rate": 0.0004229548839212998,
      "loss": 2.0831,
      "step": 84231
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.8432915210723877,
      "learning_rate": 0.00042295115269086665,
      "loss": 2.2273,
      "step": 84232
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.9985212683677673,
      "learning_rate": 0.0004229474214375745,
      "loss": 2.285,
      "step": 84233
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.8332579731941223,
      "learning_rate": 0.00042294369016142406,
      "loss": 2.2271,
      "step": 84234
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.8547017574310303,
      "learning_rate": 0.0004229399588624161,
      "loss": 2.2519,
      "step": 84235
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.8271975517272949,
      "learning_rate": 0.00042293622754055135,
      "loss": 2.3766,
      "step": 84236
    },
    {
      "epoch": 1.1,
      "grad_norm": 1.0073938369750977,
      "learning_rate": 0.0004229324961958303,
      "loss": 2.3232,
      "step": 84237
    },
    {
      "epoch": 1.1,
      "grad_norm": 1.009936809539795,
      "learning_rate": 0.00042292876482825394,
      "loss": 2.111,
      "step": 84238
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.7929772734642029,
      "learning_rate": 0.0004229250334378227,
      "loss": 2.0239,
      "step": 84239
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.9489707350730896,
      "learning_rate": 0.0004229213020245374,
      "loss": 2.2805,
      "step": 84240
    },
    {
      "epoch": 1.1,
      "grad_norm": 1.3495404720306396,
      "learning_rate": 0.00042291757058839866,
      "loss": 2.182,
      "step": 84241
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.8579087257385254,
      "learning_rate": 0.00042291383912940726,
      "loss": 2.419,
      "step": 84242
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.9888721704483032,
      "learning_rate": 0.00042291010764756385,
      "loss": 2.3743,
      "step": 84243
    },
    {
      "epoch": 1.1,
      "grad_norm": 1.107151746749878,
      "learning_rate": 0.00042290637614286915,
      "loss": 1.9668,
      "step": 84244
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.8582131862640381,
      "learning_rate": 0.00042290264461532386,
      "loss": 2.2074,
      "step": 84245
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.7507959604263306,
      "learning_rate": 0.0004228989130649286,
      "loss": 2.2782,
      "step": 84246
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.8503844738006592,
      "learning_rate": 0.00042289518149168416,
      "loss": 2.1887,
      "step": 84247
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.9570384621620178,
      "learning_rate": 0.00042289144989559116,
      "loss": 2.2654,
      "step": 84248
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.9012434482574463,
      "learning_rate": 0.0004228877182766503,
      "loss": 2.1077,
      "step": 84249
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.8946581482887268,
      "learning_rate": 0.0004228839866348622,
      "loss": 2.4869,
      "step": 84250
    },
    {
      "epoch": 1.1,
      "grad_norm": 1.143076777458191,
      "learning_rate": 0.00042288025497022786,
      "loss": 2.2534,
      "step": 84251
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.8196356892585754,
      "learning_rate": 0.0004228765232827477,
      "loss": 2.0948,
      "step": 84252
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.8134200572967529,
      "learning_rate": 0.0004228727915724224,
      "loss": 2.2974,
      "step": 84253
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.7379936575889587,
      "learning_rate": 0.0004228690598392528,
      "loss": 2.4456,
      "step": 84254
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.8436508774757385,
      "learning_rate": 0.0004228653280832394,
      "loss": 2.1184,
      "step": 84255
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.8771201968193054,
      "learning_rate": 0.0004228615963043831,
      "loss": 2.3184,
      "step": 84256
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.8050954341888428,
      "learning_rate": 0.00042285786450268456,
      "loss": 2.3016,
      "step": 84257
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.7576761841773987,
      "learning_rate": 0.0004228541326781443,
      "loss": 2.114,
      "step": 84258
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.8922234177589417,
      "learning_rate": 0.0004228504008307632,
      "loss": 2.1362,
      "step": 84259
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.7923654317855835,
      "learning_rate": 0.00042284666896054196,
      "loss": 2.088,
      "step": 84260
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.7875233888626099,
      "learning_rate": 0.0004228429370674811,
      "loss": 2.17,
      "step": 84261
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.8252155184745789,
      "learning_rate": 0.0004228392051515815,
      "loss": 2.0127,
      "step": 84262
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.8159292936325073,
      "learning_rate": 0.0004228354732128438,
      "loss": 2.2609,
      "step": 84263
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.9041263461112976,
      "learning_rate": 0.0004228317412512686,
      "loss": 2.3583,
      "step": 84264
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.8400786519050598,
      "learning_rate": 0.00042282800926685663,
      "loss": 2.3108,
      "step": 84265
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.9456192255020142,
      "learning_rate": 0.00042282427725960864,
      "loss": 1.9494,
      "step": 84266
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.8457788228988647,
      "learning_rate": 0.0004228205452295253,
      "loss": 2.0703,
      "step": 84267
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.832080066204071,
      "learning_rate": 0.00042281681317660733,
      "loss": 1.938,
      "step": 84268
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.7507219910621643,
      "learning_rate": 0.00042281308110085543,
      "loss": 2.0234,
      "step": 84269
    },
    {
      "epoch": 1.1,
      "grad_norm": 1.252760410308838,
      "learning_rate": 0.0004228093490022702,
      "loss": 2.068,
      "step": 84270
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.84440016746521,
      "learning_rate": 0.00042280561688085234,
      "loss": 2.4529,
      "step": 84271
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.7266421318054199,
      "learning_rate": 0.00042280188473660266,
      "loss": 2.0366,
      "step": 84272
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.722842276096344,
      "learning_rate": 0.00042279815256952185,
      "loss": 2.2151,
      "step": 84273
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.8810316920280457,
      "learning_rate": 0.00042279442037961045,
      "loss": 2.3171,
      "step": 84274
    },
    {
      "epoch": 1.1,
      "grad_norm": 1.184033751487732,
      "learning_rate": 0.00042279068816686934,
      "loss": 2.2193,
      "step": 84275
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.747546374797821,
      "learning_rate": 0.000422786955931299,
      "loss": 1.9839,
      "step": 84276
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.8526830077171326,
      "learning_rate": 0.00042278322367290033,
      "loss": 2.1208,
      "step": 84277
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.9658946990966797,
      "learning_rate": 0.00042277949139167396,
      "loss": 2.1703,
      "step": 84278
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.9143015146255493,
      "learning_rate": 0.00042277575908762047,
      "loss": 2.1979,
      "step": 84279
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.819085419178009,
      "learning_rate": 0.00042277202676074073,
      "loss": 2.3217,
      "step": 84280
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.8925281167030334,
      "learning_rate": 0.0004227682944110353,
      "loss": 2.1534,
      "step": 84281
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.8156896233558655,
      "learning_rate": 0.000422764562038505,
      "loss": 2.3402,
      "step": 84282
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.7601709365844727,
      "learning_rate": 0.0004227608296431504,
      "loss": 2.2341,
      "step": 84283
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.8909851312637329,
      "learning_rate": 0.0004227570972249722,
      "loss": 2.2927,
      "step": 84284
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.8916462659835815,
      "learning_rate": 0.0004227533647839713,
      "loss": 2.2859,
      "step": 84285
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.7462751269340515,
      "learning_rate": 0.00042274963232014805,
      "loss": 2.0726,
      "step": 84286
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.7370322346687317,
      "learning_rate": 0.0004227458998335034,
      "loss": 2.0116,
      "step": 84287
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.8181920647621155,
      "learning_rate": 0.000422742167324038,
      "loss": 2.3589,
      "step": 84288
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.7754511833190918,
      "learning_rate": 0.0004227384347917525,
      "loss": 2.053,
      "step": 84289
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.7767933011054993,
      "learning_rate": 0.00042273470223664754,
      "loss": 1.8918,
      "step": 84290
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.7580015063285828,
      "learning_rate": 0.00042273096965872393,
      "loss": 1.8642,
      "step": 84291
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.7964754700660706,
      "learning_rate": 0.0004227272370579823,
      "loss": 2.04,
      "step": 84292
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.8862162828445435,
      "learning_rate": 0.0004227235044344235,
      "loss": 2.0497,
      "step": 84293
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.7541818022727966,
      "learning_rate": 0.0004227197717880479,
      "loss": 2.3378,
      "step": 84294
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.9317913055419922,
      "learning_rate": 0.00042271603911885644,
      "loss": 2.2081,
      "step": 84295
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.7691319584846497,
      "learning_rate": 0.0004227123064268498,
      "loss": 2.289,
      "step": 84296
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.8069339990615845,
      "learning_rate": 0.00042270857371202856,
      "loss": 1.9786,
      "step": 84297
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.8226953148841858,
      "learning_rate": 0.00042270484097439346,
      "loss": 2.3413,
      "step": 84298
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.7720955014228821,
      "learning_rate": 0.00042270110821394534,
      "loss": 2.222,
      "step": 84299
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.7754986882209778,
      "learning_rate": 0.0004226973754306846,
      "loss": 2.2911,
      "step": 84300
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.8172694444656372,
      "learning_rate": 0.0004226936426246122,
      "loss": 2.0753,
      "step": 84301
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.8440443873405457,
      "learning_rate": 0.0004226899097957288,
      "loss": 2.1723,
      "step": 84302
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.856685996055603,
      "learning_rate": 0.0004226861769440349,
      "loss": 2.1959,
      "step": 84303
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.7182141542434692,
      "learning_rate": 0.0004226824440695313,
      "loss": 2.2884,
      "step": 84304
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.870839536190033,
      "learning_rate": 0.0004226787111722189,
      "loss": 2.3056,
      "step": 84305
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.8580607771873474,
      "learning_rate": 0.0004226749782520981,
      "loss": 2.2794,
      "step": 84306
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.8180031180381775,
      "learning_rate": 0.00042267124530916963,
      "loss": 2.1866,
      "step": 84307
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.8636745810508728,
      "learning_rate": 0.0004226675123434344,
      "loss": 2.0201,
      "step": 84308
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.7421292662620544,
      "learning_rate": 0.0004226637793548929,
      "loss": 2.3164,
      "step": 84309
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.9842462539672852,
      "learning_rate": 0.00042266004634354586,
      "loss": 2.2413,
      "step": 84310
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.8511056303977966,
      "learning_rate": 0.00042265631330939404,
      "loss": 2.2863,
      "step": 84311
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.9420791268348694,
      "learning_rate": 0.0004226525802524381,
      "loss": 2.2169,
      "step": 84312
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.8349155187606812,
      "learning_rate": 0.0004226488471726787,
      "loss": 2.232,
      "step": 84313
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.7760406136512756,
      "learning_rate": 0.00042264511407011656,
      "loss": 2.1533,
      "step": 84314
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.8714386224746704,
      "learning_rate": 0.00042264138094475244,
      "loss": 2.1307,
      "step": 84315
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.9081994891166687,
      "learning_rate": 0.00042263764779658695,
      "loss": 2.0462,
      "step": 84316
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.7906110882759094,
      "learning_rate": 0.0004226339146256208,
      "loss": 2.1768,
      "step": 84317
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.8643433451652527,
      "learning_rate": 0.00042263018143185474,
      "loss": 2.2747,
      "step": 84318
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.8637568950653076,
      "learning_rate": 0.0004226264482152893,
      "loss": 2.3098,
      "step": 84319
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.8324226140975952,
      "learning_rate": 0.0004226227149759253,
      "loss": 2.2448,
      "step": 84320
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.8075237274169922,
      "learning_rate": 0.00042261898171376354,
      "loss": 2.4324,
      "step": 84321
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.7681461572647095,
      "learning_rate": 0.0004226152484288045,
      "loss": 2.1039,
      "step": 84322
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.7728225588798523,
      "learning_rate": 0.000422611515121049,
      "loss": 2.1877,
      "step": 84323
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.8251071572303772,
      "learning_rate": 0.0004226077817904977,
      "loss": 2.1815,
      "step": 84324
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.7515202760696411,
      "learning_rate": 0.0004226040484371513,
      "loss": 2.1336,
      "step": 84325
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.8087941408157349,
      "learning_rate": 0.00042260031506101046,
      "loss": 2.0891,
      "step": 84326
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.8820185661315918,
      "learning_rate": 0.000422596581662076,
      "loss": 2.2527,
      "step": 84327
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.8363786935806274,
      "learning_rate": 0.0004225928482403484,
      "loss": 2.3279,
      "step": 84328
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.8543521165847778,
      "learning_rate": 0.00042258911479582857,
      "loss": 2.2438,
      "step": 84329
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.8635408282279968,
      "learning_rate": 0.0004225853813285171,
      "loss": 2.1313,
      "step": 84330
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.79938143491745,
      "learning_rate": 0.00042258164783841464,
      "loss": 2.1862,
      "step": 84331
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.7747882604598999,
      "learning_rate": 0.0004225779143255219,
      "loss": 2.1025,
      "step": 84332
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.8993769884109497,
      "learning_rate": 0.0004225741807898398,
      "loss": 2.0185,
      "step": 84333
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.828274667263031,
      "learning_rate": 0.0004225704472313687,
      "loss": 2.1754,
      "step": 84334
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.8058143258094788,
      "learning_rate": 0.0004225667136501094,
      "loss": 2.3028,
      "step": 84335
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.8468239903450012,
      "learning_rate": 0.0004225629800460628,
      "loss": 2.28,
      "step": 84336
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.8119862079620361,
      "learning_rate": 0.0004225592464192293,
      "loss": 2.1611,
      "step": 84337
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.8164494037628174,
      "learning_rate": 0.00042255551276960973,
      "loss": 2.2132,
      "step": 84338
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.7777599692344666,
      "learning_rate": 0.0004225517790972048,
      "loss": 2.0558,
      "step": 84339
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.7603384256362915,
      "learning_rate": 0.0004225480454020152,
      "loss": 2.1735,
      "step": 84340
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.7880939245223999,
      "learning_rate": 0.00042254431168404154,
      "loss": 2.3397,
      "step": 84341
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.7497562170028687,
      "learning_rate": 0.0004225405779432846,
      "loss": 2.0576,
      "step": 84342
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.7644187211990356,
      "learning_rate": 0.0004225368441797451,
      "loss": 2.1474,
      "step": 84343
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.8158823251724243,
      "learning_rate": 0.0004225331103934236,
      "loss": 2.3197,
      "step": 84344
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.7713443040847778,
      "learning_rate": 0.00042252937658432107,
      "loss": 2.0281,
      "step": 84345
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.8597873449325562,
      "learning_rate": 0.00042252564275243785,
      "loss": 2.0868,
      "step": 84346
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.8238347768783569,
      "learning_rate": 0.00042252190889777483,
      "loss": 2.3247,
      "step": 84347
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.7896968722343445,
      "learning_rate": 0.0004225181750203327,
      "loss": 2.0365,
      "step": 84348
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.8374702334403992,
      "learning_rate": 0.00042251444112011204,
      "loss": 2.3238,
      "step": 84349
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.7349129915237427,
      "learning_rate": 0.0004225107071971137,
      "loss": 2.0556,
      "step": 84350
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.8571543097496033,
      "learning_rate": 0.0004225069732513384,
      "loss": 1.9821,
      "step": 84351
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.8319485187530518,
      "learning_rate": 0.0004225032392827866,
      "loss": 2.1464,
      "step": 84352
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.7841405272483826,
      "learning_rate": 0.0004224995052914592,
      "loss": 2.1816,
      "step": 84353
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.775696337223053,
      "learning_rate": 0.00042249577127735687,
      "loss": 1.9243,
      "step": 84354
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.779059886932373,
      "learning_rate": 0.0004224920372404802,
      "loss": 2.2961,
      "step": 84355
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.9428223967552185,
      "learning_rate": 0.0004224883031808299,
      "loss": 2.3345,
      "step": 84356
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.8233159184455872,
      "learning_rate": 0.00042248456909840684,
      "loss": 2.1625,
      "step": 84357
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.7850885391235352,
      "learning_rate": 0.0004224808349932115,
      "loss": 2.3076,
      "step": 84358
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.8471032381057739,
      "learning_rate": 0.0004224771008652447,
      "loss": 2.2774,
      "step": 84359
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.8672785758972168,
      "learning_rate": 0.0004224733667145071,
      "loss": 2.2088,
      "step": 84360
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.8659898638725281,
      "learning_rate": 0.0004224696325409993,
      "loss": 2.1359,
      "step": 84361
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.7650076746940613,
      "learning_rate": 0.0004224658983447222,
      "loss": 1.9803,
      "step": 84362
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.8819826245307922,
      "learning_rate": 0.00042246216412567634,
      "loss": 2.0261,
      "step": 84363
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.8403614163398743,
      "learning_rate": 0.0004224584298838624,
      "loss": 2.4229,
      "step": 84364
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.762283444404602,
      "learning_rate": 0.00042245469561928115,
      "loss": 2.0981,
      "step": 84365
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.7973255515098572,
      "learning_rate": 0.0004224509613319333,
      "loss": 2.1263,
      "step": 84366
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.8053964376449585,
      "learning_rate": 0.0004224472270218195,
      "loss": 2.3762,
      "step": 84367
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.7899459600448608,
      "learning_rate": 0.0004224434926889404,
      "loss": 2.2218,
      "step": 84368
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.7474242448806763,
      "learning_rate": 0.00042243975833329675,
      "loss": 2.1667,
      "step": 84369
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.8313027024269104,
      "learning_rate": 0.00042243602395488933,
      "loss": 2.4991,
      "step": 84370
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.8478609919548035,
      "learning_rate": 0.0004224322895537186,
      "loss": 2.1224,
      "step": 84371
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.8595537543296814,
      "learning_rate": 0.0004224285551297855,
      "loss": 2.124,
      "step": 84372
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.8250415921211243,
      "learning_rate": 0.0004224248206830905,
      "loss": 2.3962,
      "step": 84373
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.7532041668891907,
      "learning_rate": 0.00042242108621363457,
      "loss": 2.2601,
      "step": 84374
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.8286503553390503,
      "learning_rate": 0.0004224173517214182,
      "loss": 2.0145,
      "step": 84375
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.772896945476532,
      "learning_rate": 0.0004224136172064421,
      "loss": 1.975,
      "step": 84376
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.7608585953712463,
      "learning_rate": 0.00042240988266870706,
      "loss": 2.453,
      "step": 84377
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.7945131063461304,
      "learning_rate": 0.0004224061481082136,
      "loss": 2.1785,
      "step": 84378
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.9641555547714233,
      "learning_rate": 0.00042240241352496264,
      "loss": 2.3079,
      "step": 84379
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.743481457233429,
      "learning_rate": 0.00042239867891895463,
      "loss": 1.9796,
      "step": 84380
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.7960324883460999,
      "learning_rate": 0.00042239494429019056,
      "loss": 2.3327,
      "step": 84381
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.830995500087738,
      "learning_rate": 0.00042239120963867083,
      "loss": 2.1538,
      "step": 84382
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.8365708589553833,
      "learning_rate": 0.00042238747496439635,
      "loss": 2.3886,
      "step": 84383
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.7688197493553162,
      "learning_rate": 0.0004223837402673676,
      "loss": 2.2588,
      "step": 84384
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.9181045293807983,
      "learning_rate": 0.00042238000554758553,
      "loss": 2.0671,
      "step": 84385
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.7763035893440247,
      "learning_rate": 0.0004223762708050507,
      "loss": 2.1063,
      "step": 84386
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.759662389755249,
      "learning_rate": 0.00042237253603976375,
      "loss": 2.2373,
      "step": 84387
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.9065221548080444,
      "learning_rate": 0.0004223688012517255,
      "loss": 2.1255,
      "step": 84388
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.788668155670166,
      "learning_rate": 0.00042236506644093653,
      "loss": 2.1755,
      "step": 84389
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.817047655582428,
      "learning_rate": 0.00042236133160739763,
      "loss": 2.288,
      "step": 84390
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.8111841082572937,
      "learning_rate": 0.0004223575967511094,
      "loss": 2.1832,
      "step": 84391
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.8920667171478271,
      "learning_rate": 0.0004223538618720726,
      "loss": 2.1748,
      "step": 84392
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.864591658115387,
      "learning_rate": 0.0004223501269702879,
      "loss": 2.1878,
      "step": 84393
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.8925663232803345,
      "learning_rate": 0.00042234639204575603,
      "loss": 2.1245,
      "step": 84394
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.8017196655273438,
      "learning_rate": 0.00042234265709847766,
      "loss": 2.2134,
      "step": 84395
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.9209843873977661,
      "learning_rate": 0.00042233892212845337,
      "loss": 2.1277,
      "step": 84396
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.803495466709137,
      "learning_rate": 0.00042233518713568415,
      "loss": 2.0826,
      "step": 84397
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.8392477631568909,
      "learning_rate": 0.00042233145212017037,
      "loss": 2.2407,
      "step": 84398
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.7680808305740356,
      "learning_rate": 0.0004223277170819129,
      "loss": 2.1071,
      "step": 84399
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.9173980951309204,
      "learning_rate": 0.0004223239820209124,
      "loss": 2.0454,
      "step": 84400
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.782462477684021,
      "learning_rate": 0.0004223202469371696,
      "loss": 2.1498,
      "step": 84401
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.8169621825218201,
      "learning_rate": 0.0004223165118306851,
      "loss": 2.2431,
      "step": 84402
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.8290235996246338,
      "learning_rate": 0.0004223127767014597,
      "loss": 2.4181,
      "step": 84403
    },
    {
      "epoch": 1.1,
      "grad_norm": 1.086039423942566,
      "learning_rate": 0.00042230904154949404,
      "loss": 2.3825,
      "step": 84404
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.722247838973999,
      "learning_rate": 0.0004223053063747888,
      "loss": 2.2491,
      "step": 84405
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.8173424005508423,
      "learning_rate": 0.0004223015711773447,
      "loss": 2.1371,
      "step": 84406
    },
    {
      "epoch": 1.1,
      "grad_norm": 1.0733593702316284,
      "learning_rate": 0.0004222978359571624,
      "loss": 2.2281,
      "step": 84407
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.9694542288780212,
      "learning_rate": 0.00042229410071424264,
      "loss": 2.1545,
      "step": 84408
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.9017011523246765,
      "learning_rate": 0.0004222903654485862,
      "loss": 2.3776,
      "step": 84409
    },
    {
      "epoch": 1.1,
      "grad_norm": 1.1282494068145752,
      "learning_rate": 0.0004222866301601936,
      "loss": 2.1992,
      "step": 84410
    },
    {
      "epoch": 1.1,
      "grad_norm": 1.1845906972885132,
      "learning_rate": 0.00042228289484906546,
      "loss": 2.3037,
      "step": 84411
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.8478403687477112,
      "learning_rate": 0.00042227915951520287,
      "loss": 2.0604,
      "step": 84412
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.9132630228996277,
      "learning_rate": 0.0004222754241586061,
      "loss": 2.2606,
      "step": 84413
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.9425142407417297,
      "learning_rate": 0.000422271688779276,
      "loss": 2.2636,
      "step": 84414
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.8890820145606995,
      "learning_rate": 0.0004222679533772134,
      "loss": 2.1783,
      "step": 84415
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.8672221302986145,
      "learning_rate": 0.0004222642179524189,
      "loss": 2.249,
      "step": 84416
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.8389315009117126,
      "learning_rate": 0.00042226048250489307,
      "loss": 2.2047,
      "step": 84417
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.9059930443763733,
      "learning_rate": 0.0004222567470346368,
      "loss": 2.2734,
      "step": 84418
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.8249073624610901,
      "learning_rate": 0.0004222530115416507,
      "loss": 2.1941,
      "step": 84419
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.9913344383239746,
      "learning_rate": 0.0004222492760259353,
      "loss": 2.0219,
      "step": 84420
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.9351620674133301,
      "learning_rate": 0.00042224554048749164,
      "loss": 2.0828,
      "step": 84421
    },
    {
      "epoch": 1.1,
      "grad_norm": 1.2566933631896973,
      "learning_rate": 0.0004222418049263201,
      "loss": 2.2588,
      "step": 84422
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.9195428490638733,
      "learning_rate": 0.0004222380693424215,
      "loss": 2.2064,
      "step": 84423
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.8646924495697021,
      "learning_rate": 0.00042223433373579666,
      "loss": 2.1777,
      "step": 84424
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.8147239089012146,
      "learning_rate": 0.00042223059810644604,
      "loss": 2.3327,
      "step": 84425
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.764775276184082,
      "learning_rate": 0.00042222686245437045,
      "loss": 2.2368,
      "step": 84426
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.8239250779151917,
      "learning_rate": 0.0004222231267795707,
      "loss": 2.1011,
      "step": 84427
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.7309030294418335,
      "learning_rate": 0.0004222193910820472,
      "loss": 1.9757,
      "step": 84428
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.8524960875511169,
      "learning_rate": 0.0004222156553618009,
      "loss": 2.2134,
      "step": 84429
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.9204477071762085,
      "learning_rate": 0.0004222119196188324,
      "loss": 2.2918,
      "step": 84430
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.7958473563194275,
      "learning_rate": 0.0004222081838531423,
      "loss": 2.0848,
      "step": 84431
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.710643470287323,
      "learning_rate": 0.0004222044480647315,
      "loss": 2.4056,
      "step": 84432
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.8812975883483887,
      "learning_rate": 0.00042220071225360057,
      "loss": 2.2013,
      "step": 84433
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.9007120132446289,
      "learning_rate": 0.0004221969764197502,
      "loss": 2.1505,
      "step": 84434
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.7381187677383423,
      "learning_rate": 0.0004221932405631811,
      "loss": 2.1294,
      "step": 84435
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.8452659845352173,
      "learning_rate": 0.000422189504683894,
      "loss": 2.2895,
      "step": 84436
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.8793522119522095,
      "learning_rate": 0.00042218576878188955,
      "loss": 2.1576,
      "step": 84437
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.8737512230873108,
      "learning_rate": 0.00042218203285716843,
      "loss": 2.0941,
      "step": 84438
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.9171340465545654,
      "learning_rate": 0.00042217829690973144,
      "loss": 2.3941,
      "step": 84439
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.9598298668861389,
      "learning_rate": 0.0004221745609395791,
      "loss": 2.1033,
      "step": 84440
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.7636851668357849,
      "learning_rate": 0.00042217082494671226,
      "loss": 2.2314,
      "step": 84441
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.9254810214042664,
      "learning_rate": 0.00042216708893113154,
      "loss": 2.3108,
      "step": 84442
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.8723598718643188,
      "learning_rate": 0.00042216335289283765,
      "loss": 2.3169,
      "step": 84443
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.8876456022262573,
      "learning_rate": 0.0004221596168318313,
      "loss": 2.2257,
      "step": 84444
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.9089072942733765,
      "learning_rate": 0.0004221558807481132,
      "loss": 2.2098,
      "step": 84445
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.8437208533287048,
      "learning_rate": 0.00042215214464168396,
      "loss": 2.2656,
      "step": 84446
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.8607953786849976,
      "learning_rate": 0.00042214840851254433,
      "loss": 2.2513,
      "step": 84447
    },
    {
      "epoch": 1.1,
      "grad_norm": 1.1374571323394775,
      "learning_rate": 0.00042214467236069495,
      "loss": 2.1617,
      "step": 84448
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.8632992506027222,
      "learning_rate": 0.0004221409361861367,
      "loss": 2.1897,
      "step": 84449
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.9426654577255249,
      "learning_rate": 0.00042213719998887014,
      "loss": 2.0074,
      "step": 84450
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.8671525716781616,
      "learning_rate": 0.0004221334637688959,
      "loss": 1.9671,
      "step": 84451
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.7775416374206543,
      "learning_rate": 0.00042212972752621477,
      "loss": 2.2887,
      "step": 84452
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.7852565050125122,
      "learning_rate": 0.0004221259912608274,
      "loss": 2.0667,
      "step": 84453
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.8611544966697693,
      "learning_rate": 0.00042212225497273453,
      "loss": 2.3008,
      "step": 84454
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.8837092518806458,
      "learning_rate": 0.00042211851866193683,
      "loss": 2.1748,
      "step": 84455
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.9901701807975769,
      "learning_rate": 0.00042211478232843496,
      "loss": 2.0084,
      "step": 84456
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.7764729857444763,
      "learning_rate": 0.0004221110459722297,
      "loss": 2.1383,
      "step": 84457
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.9784306287765503,
      "learning_rate": 0.00042210730959332164,
      "loss": 2.2471,
      "step": 84458
    },
    {
      "epoch": 1.1,
      "grad_norm": 1.0393308401107788,
      "learning_rate": 0.00042210357319171155,
      "loss": 2.2581,
      "step": 84459
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.7986984252929688,
      "learning_rate": 0.00042209983676740005,
      "loss": 2.0715,
      "step": 84460
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.8204662203788757,
      "learning_rate": 0.00042209610032038796,
      "loss": 2.333,
      "step": 84461
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.8965035080909729,
      "learning_rate": 0.0004220923638506758,
      "loss": 2.1258,
      "step": 84462
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.8973103761672974,
      "learning_rate": 0.00042208862735826445,
      "loss": 1.967,
      "step": 84463
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.790010392665863,
      "learning_rate": 0.0004220848908431545,
      "loss": 2.3521,
      "step": 84464
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.9551151990890503,
      "learning_rate": 0.0004220811543053466,
      "loss": 2.2443,
      "step": 84465
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.9244906306266785,
      "learning_rate": 0.0004220774177448417,
      "loss": 2.2886,
      "step": 84466
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.8131955862045288,
      "learning_rate": 0.0004220736811616401,
      "loss": 2.1993,
      "step": 84467
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.813681423664093,
      "learning_rate": 0.00042206994455574277,
      "loss": 2.2988,
      "step": 84468
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.7825279235839844,
      "learning_rate": 0.0004220662079271504,
      "loss": 2.2874,
      "step": 84469
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.8075242042541504,
      "learning_rate": 0.0004220624712758635,
      "loss": 1.9657,
      "step": 84470
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.7606834173202515,
      "learning_rate": 0.00042205873460188296,
      "loss": 2.1607,
      "step": 84471
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.745478093624115,
      "learning_rate": 0.00042205499790520936,
      "loss": 2.2381,
      "step": 84472
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.9585059881210327,
      "learning_rate": 0.00042205126118584346,
      "loss": 2.1309,
      "step": 84473
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.7990195155143738,
      "learning_rate": 0.00042204752444378586,
      "loss": 2.0253,
      "step": 84474
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.9188464283943176,
      "learning_rate": 0.0004220437876790374,
      "loss": 2.1286,
      "step": 84475
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.8248525261878967,
      "learning_rate": 0.0004220400508915987,
      "loss": 2.1143,
      "step": 84476
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.9723002910614014,
      "learning_rate": 0.0004220363140814704,
      "loss": 2.1561,
      "step": 84477
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.7934396266937256,
      "learning_rate": 0.00042203257724865326,
      "loss": 2.0644,
      "step": 84478
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.8161571025848389,
      "learning_rate": 0.00042202884039314805,
      "loss": 1.9385,
      "step": 84479
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.7944727540016174,
      "learning_rate": 0.0004220251035149552,
      "loss": 2.2685,
      "step": 84480
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.7824983596801758,
      "learning_rate": 0.0004220213666140757,
      "loss": 2.285,
      "step": 84481
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.7584031820297241,
      "learning_rate": 0.00042201762969051014,
      "loss": 2.2096,
      "step": 84482
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.7910079956054688,
      "learning_rate": 0.0004220138927442591,
      "loss": 2.133,
      "step": 84483
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.8384988903999329,
      "learning_rate": 0.0004220101557753235,
      "loss": 2.3393,
      "step": 84484
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.8457014560699463,
      "learning_rate": 0.0004220064187837039,
      "loss": 2.27,
      "step": 84485
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.8805578947067261,
      "learning_rate": 0.00042200268176940085,
      "loss": 2.0733,
      "step": 84486
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.8412607908248901,
      "learning_rate": 0.00042199894473241527,
      "loss": 2.1472,
      "step": 84487
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.8235940337181091,
      "learning_rate": 0.00042199520767274786,
      "loss": 2.1061,
      "step": 84488
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.8289300799369812,
      "learning_rate": 0.00042199147059039915,
      "loss": 2.0183,
      "step": 84489
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.7670934796333313,
      "learning_rate": 0.00042198773348536996,
      "loss": 2.2097,
      "step": 84490
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.7613021731376648,
      "learning_rate": 0.000421983996357661,
      "loss": 2.1838,
      "step": 84491
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.8948604464530945,
      "learning_rate": 0.00042198025920727286,
      "loss": 2.1963,
      "step": 84492
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.7431955933570862,
      "learning_rate": 0.00042197652203420624,
      "loss": 2.3204,
      "step": 84493
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.9049611687660217,
      "learning_rate": 0.00042197278483846197,
      "loss": 2.4398,
      "step": 84494
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.8187990784645081,
      "learning_rate": 0.0004219690476200406,
      "loss": 2.1476,
      "step": 84495
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.8515763878822327,
      "learning_rate": 0.0004219653103789428,
      "loss": 2.2907,
      "step": 84496
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.7763751149177551,
      "learning_rate": 0.0004219615731151696,
      "loss": 2.2528,
      "step": 84497
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.8289538621902466,
      "learning_rate": 0.0004219578358287212,
      "loss": 2.1651,
      "step": 84498
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.8449972867965698,
      "learning_rate": 0.0004219540985195986,
      "loss": 2.2247,
      "step": 84499
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.8925896883010864,
      "learning_rate": 0.0004219503611878025,
      "loss": 2.2779,
      "step": 84500
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.8048609495162964,
      "learning_rate": 0.00042194662383333344,
      "loss": 2.0656,
      "step": 84501
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.8223457336425781,
      "learning_rate": 0.00042194288645619227,
      "loss": 2.0104,
      "step": 84502
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.7989024519920349,
      "learning_rate": 0.0004219391490563796,
      "loss": 2.0659,
      "step": 84503
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.760520339012146,
      "learning_rate": 0.00042193541163389615,
      "loss": 2.1624,
      "step": 84504
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.8217867612838745,
      "learning_rate": 0.0004219316741887425,
      "loss": 2.2424,
      "step": 84505
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.8546847701072693,
      "learning_rate": 0.0004219279367209196,
      "loss": 2.2038,
      "step": 84506
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.8345798850059509,
      "learning_rate": 0.0004219241992304279,
      "loss": 2.0068,
      "step": 84507
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.8499086499214172,
      "learning_rate": 0.00042192046171726815,
      "loss": 2.0887,
      "step": 84508
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.765448272228241,
      "learning_rate": 0.0004219167241814412,
      "loss": 2.0566,
      "step": 84509
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.782959520816803,
      "learning_rate": 0.00042191298662294755,
      "loss": 1.9781,
      "step": 84510
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.8548772931098938,
      "learning_rate": 0.00042190924904178805,
      "loss": 2.1617,
      "step": 84511
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.802214503288269,
      "learning_rate": 0.0004219055114379632,
      "loss": 1.9927,
      "step": 84512
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.8146923184394836,
      "learning_rate": 0.00042190177381147396,
      "loss": 2.0911,
      "step": 84513
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.8596043586730957,
      "learning_rate": 0.0004218980361623208,
      "loss": 2.19,
      "step": 84514
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.8021079897880554,
      "learning_rate": 0.00042189429849050447,
      "loss": 2.1138,
      "step": 84515
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.9584491848945618,
      "learning_rate": 0.00042189056079602573,
      "loss": 2.0468,
      "step": 84516
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.8109138607978821,
      "learning_rate": 0.0004218868230788852,
      "loss": 2.3403,
      "step": 84517
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.8270046710968018,
      "learning_rate": 0.00042188308533908363,
      "loss": 2.3049,
      "step": 84518
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.7869786024093628,
      "learning_rate": 0.00042187934757662173,
      "loss": 2.1718,
      "step": 84519
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.8244909644126892,
      "learning_rate": 0.0004218756097915001,
      "loss": 2.2817,
      "step": 84520
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.843521237373352,
      "learning_rate": 0.0004218718719837195,
      "loss": 2.2689,
      "step": 84521
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.837943434715271,
      "learning_rate": 0.00042186813415328065,
      "loss": 2.2297,
      "step": 84522
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.7312375903129578,
      "learning_rate": 0.0004218643963001842,
      "loss": 2.1462,
      "step": 84523
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.7758699655532837,
      "learning_rate": 0.00042186065842443084,
      "loss": 2.1761,
      "step": 84524
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.8926680088043213,
      "learning_rate": 0.00042185692052602137,
      "loss": 2.255,
      "step": 84525
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.8491897583007812,
      "learning_rate": 0.0004218531826049563,
      "loss": 2.2879,
      "step": 84526
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.8820859789848328,
      "learning_rate": 0.0004218494446612364,
      "loss": 2.483,
      "step": 84527
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.9036117792129517,
      "learning_rate": 0.0004218457066948625,
      "loss": 2.1873,
      "step": 84528
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.7803374528884888,
      "learning_rate": 0.0004218419687058351,
      "loss": 2.1292,
      "step": 84529
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.801032543182373,
      "learning_rate": 0.000421838230694155,
      "loss": 2.1992,
      "step": 84530
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.8148956298828125,
      "learning_rate": 0.0004218344926598229,
      "loss": 2.2582,
      "step": 84531
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.748120129108429,
      "learning_rate": 0.0004218307546028394,
      "loss": 2.1736,
      "step": 84532
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.850272536277771,
      "learning_rate": 0.00042182701652320533,
      "loss": 2.1418,
      "step": 84533
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.8070278763771057,
      "learning_rate": 0.00042182327842092134,
      "loss": 2.0106,
      "step": 84534
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.8067375421524048,
      "learning_rate": 0.000421819540295988,
      "loss": 1.9787,
      "step": 84535
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.7683022022247314,
      "learning_rate": 0.0004218158021484062,
      "loss": 2.1437,
      "step": 84536
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.7867030501365662,
      "learning_rate": 0.0004218120639781765,
      "loss": 2.2361,
      "step": 84537
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.7632793188095093,
      "learning_rate": 0.0004218083257852996,
      "loss": 2.0387,
      "step": 84538
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.8681908845901489,
      "learning_rate": 0.0004218045875697763,
      "loss": 2.461,
      "step": 84539
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.8662911653518677,
      "learning_rate": 0.0004218008493316073,
      "loss": 2.1425,
      "step": 84540
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.7917581796646118,
      "learning_rate": 0.0004217971110707931,
      "loss": 2.4157,
      "step": 84541
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.8478441834449768,
      "learning_rate": 0.0004217933727873345,
      "loss": 2.2711,
      "step": 84542
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.7612791657447815,
      "learning_rate": 0.0004217896344812323,
      "loss": 2.3082,
      "step": 84543
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.8421732187271118,
      "learning_rate": 0.00042178589615248706,
      "loss": 1.9869,
      "step": 84544
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.8819752931594849,
      "learning_rate": 0.0004217821578010995,
      "loss": 2.0035,
      "step": 84545
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.8024057149887085,
      "learning_rate": 0.0004217784194270703,
      "loss": 2.2683,
      "step": 84546
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.994775116443634,
      "learning_rate": 0.0004217746810304003,
      "loss": 2.2621,
      "step": 84547
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.8323565125465393,
      "learning_rate": 0.00042177094261109006,
      "loss": 2.2357,
      "step": 84548
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.7646148800849915,
      "learning_rate": 0.00042176720416914034,
      "loss": 2.3901,
      "step": 84549
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.8427426815032959,
      "learning_rate": 0.00042176346570455177,
      "loss": 2.1621,
      "step": 84550
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.7923035025596619,
      "learning_rate": 0.00042175972721732505,
      "loss": 2.1162,
      "step": 84551
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.829850971698761,
      "learning_rate": 0.0004217559887074608,
      "loss": 2.284,
      "step": 84552
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.8272948265075684,
      "learning_rate": 0.0004217522501749599,
      "loss": 2.19,
      "step": 84553
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.9882650971412659,
      "learning_rate": 0.00042174851161982306,
      "loss": 2.0175,
      "step": 84554
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.7821288704872131,
      "learning_rate": 0.0004217447730420507,
      "loss": 2.1842,
      "step": 84555
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.8170264363288879,
      "learning_rate": 0.0004217410344416438,
      "loss": 2.2572,
      "step": 84556
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.8123465776443481,
      "learning_rate": 0.00042173729581860295,
      "loss": 2.3864,
      "step": 84557
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.7622663974761963,
      "learning_rate": 0.0004217335571729288,
      "loss": 2.3603,
      "step": 84558
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.8733210563659668,
      "learning_rate": 0.00042172981850462207,
      "loss": 2.3712,
      "step": 84559
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.7486581206321716,
      "learning_rate": 0.00042172607981368353,
      "loss": 2.0651,
      "step": 84560
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.8152006268501282,
      "learning_rate": 0.0004217223411001138,
      "loss": 2.0301,
      "step": 84561
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.9321771264076233,
      "learning_rate": 0.00042171860236391354,
      "loss": 2.1958,
      "step": 84562
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.8512092232704163,
      "learning_rate": 0.0004217148636050835,
      "loss": 2.554,
      "step": 84563
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.9025765061378479,
      "learning_rate": 0.00042171112482362443,
      "loss": 2.0913,
      "step": 84564
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.8036964535713196,
      "learning_rate": 0.0004217073860195369,
      "loss": 2.1825,
      "step": 84565
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.8592684864997864,
      "learning_rate": 0.00042170364719282163,
      "loss": 2.3336,
      "step": 84566
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.9435755610466003,
      "learning_rate": 0.0004216999083434795,
      "loss": 2.0493,
      "step": 84567
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.9171153903007507,
      "learning_rate": 0.00042169616947151093,
      "loss": 2.2638,
      "step": 84568
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.9687913060188293,
      "learning_rate": 0.00042169243057691683,
      "loss": 2.1583,
      "step": 84569
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.8780663013458252,
      "learning_rate": 0.0004216886916596978,
      "loss": 2.2685,
      "step": 84570
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.837032675743103,
      "learning_rate": 0.0004216849527198545,
      "loss": 2.1426,
      "step": 84571
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.8127222657203674,
      "learning_rate": 0.0004216812137573877,
      "loss": 2.1719,
      "step": 84572
    },
    {
      "epoch": 1.1,
      "grad_norm": 1.07985520362854,
      "learning_rate": 0.000421677474772298,
      "loss": 2.3051,
      "step": 84573
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.840114414691925,
      "learning_rate": 0.0004216737357645862,
      "loss": 1.995,
      "step": 84574
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.829765796661377,
      "learning_rate": 0.00042166999673425297,
      "loss": 1.9926,
      "step": 84575
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.8259829878807068,
      "learning_rate": 0.000421666257681299,
      "loss": 2.0983,
      "step": 84576
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.8497101068496704,
      "learning_rate": 0.000421662518605725,
      "loss": 2.2295,
      "step": 84577
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.7904677391052246,
      "learning_rate": 0.0004216587795075316,
      "loss": 2.2036,
      "step": 84578
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.8574379086494446,
      "learning_rate": 0.00042165504038671947,
      "loss": 2.2407,
      "step": 84579
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.9152313470840454,
      "learning_rate": 0.00042165130124328953,
      "loss": 2.1059,
      "step": 84580
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.8817061185836792,
      "learning_rate": 0.0004216475620772422,
      "loss": 2.083,
      "step": 84581
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.7800261378288269,
      "learning_rate": 0.0004216438228885783,
      "loss": 2.0825,
      "step": 84582
    },
    {
      "epoch": 1.1,
      "grad_norm": 1.1079128980636597,
      "learning_rate": 0.0004216400836772986,
      "loss": 2.0119,
      "step": 84583
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.7845914363861084,
      "learning_rate": 0.0004216363444434037,
      "loss": 2.1764,
      "step": 84584
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.8714200258255005,
      "learning_rate": 0.00042163260518689416,
      "loss": 2.2963,
      "step": 84585
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.795706033706665,
      "learning_rate": 0.00042162886590777095,
      "loss": 1.832,
      "step": 84586
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.7575036883354187,
      "learning_rate": 0.00042162512660603465,
      "loss": 2.34,
      "step": 84587
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.8660447597503662,
      "learning_rate": 0.00042162138728168584,
      "loss": 2.1903,
      "step": 84588
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.8121603727340698,
      "learning_rate": 0.00042161764793472546,
      "loss": 2.0807,
      "step": 84589
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.8951451182365417,
      "learning_rate": 0.000421613908565154,
      "loss": 2.1282,
      "step": 84590
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.7296906113624573,
      "learning_rate": 0.0004216101691729721,
      "loss": 1.9887,
      "step": 84591
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.833130955696106,
      "learning_rate": 0.0004216064297581808,
      "loss": 1.9917,
      "step": 84592
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.7981790900230408,
      "learning_rate": 0.0004216026903207804,
      "loss": 2.325,
      "step": 84593
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.8176008462905884,
      "learning_rate": 0.0004215989508607718,
      "loss": 1.9321,
      "step": 84594
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.7783126831054688,
      "learning_rate": 0.0004215952113781557,
      "loss": 2.201,
      "step": 84595
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.7744521498680115,
      "learning_rate": 0.0004215914718729327,
      "loss": 2.0786,
      "step": 84596
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.8167920112609863,
      "learning_rate": 0.0004215877323451036,
      "loss": 2.0895,
      "step": 84597
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.7977032661437988,
      "learning_rate": 0.00042158399279466907,
      "loss": 2.3138,
      "step": 84598
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.8561261296272278,
      "learning_rate": 0.0004215802532216297,
      "loss": 2.1445,
      "step": 84599
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.9534586668014526,
      "learning_rate": 0.00042157651362598626,
      "loss": 2.093,
      "step": 84600
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.7253054976463318,
      "learning_rate": 0.00042157277400773957,
      "loss": 2.0415,
      "step": 84601
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.9143662452697754,
      "learning_rate": 0.0004215690343668901,
      "loss": 2.2683,
      "step": 84602
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.8387793302536011,
      "learning_rate": 0.0004215652947034387,
      "loss": 2.3838,
      "step": 84603
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.8537682294845581,
      "learning_rate": 0.00042156155501738593,
      "loss": 2.204,
      "step": 84604
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.9141175150871277,
      "learning_rate": 0.00042155781530873267,
      "loss": 2.1842,
      "step": 84605
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.7489531636238098,
      "learning_rate": 0.0004215540755774795,
      "loss": 2.1604,
      "step": 84606
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.8202418088912964,
      "learning_rate": 0.0004215503358236271,
      "loss": 2.3201,
      "step": 84607
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.8449794054031372,
      "learning_rate": 0.0004215465960471762,
      "loss": 2.2717,
      "step": 84608
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.8203606009483337,
      "learning_rate": 0.00042154285624812753,
      "loss": 2.1164,
      "step": 84609
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.9367407560348511,
      "learning_rate": 0.00042153911642648175,
      "loss": 2.2851,
      "step": 84610
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.8124335408210754,
      "learning_rate": 0.0004215353765822395,
      "loss": 2.2394,
      "step": 84611
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.9333339333534241,
      "learning_rate": 0.0004215316367154016,
      "loss": 2.3412,
      "step": 84612
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.8448570370674133,
      "learning_rate": 0.0004215278968259686,
      "loss": 2.1963,
      "step": 84613
    },
    {
      "epoch": 1.1,
      "grad_norm": 1.112491250038147,
      "learning_rate": 0.0004215241569139413,
      "loss": 2.3163,
      "step": 84614
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.9375879168510437,
      "learning_rate": 0.00042152041697932037,
      "loss": 2.1344,
      "step": 84615
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.9308109283447266,
      "learning_rate": 0.00042151667702210654,
      "loss": 2.0124,
      "step": 84616
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.9872463941574097,
      "learning_rate": 0.0004215129370423004,
      "loss": 1.9318,
      "step": 84617
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.8362721800804138,
      "learning_rate": 0.00042150919703990276,
      "loss": 2.1421,
      "step": 84618
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.7914028763771057,
      "learning_rate": 0.00042150545701491427,
      "loss": 2.1648,
      "step": 84619
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.8213065266609192,
      "learning_rate": 0.00042150171696733556,
      "loss": 2.0166,
      "step": 84620
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.8384805917739868,
      "learning_rate": 0.00042149797689716744,
      "loss": 2.01,
      "step": 84621
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.8367123007774353,
      "learning_rate": 0.00042149423680441057,
      "loss": 2.1133,
      "step": 84622
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.842653214931488,
      "learning_rate": 0.0004214904966890656,
      "loss": 2.1973,
      "step": 84623
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.9373433589935303,
      "learning_rate": 0.00042148675655113315,
      "loss": 1.9708,
      "step": 84624
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.7497914433479309,
      "learning_rate": 0.0004214830163906142,
      "loss": 2.1791,
      "step": 84625
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.7912837862968445,
      "learning_rate": 0.00042147927620750917,
      "loss": 2.1387,
      "step": 84626
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.8664565086364746,
      "learning_rate": 0.0004214755360018188,
      "loss": 2.1217,
      "step": 84627
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.8194950819015503,
      "learning_rate": 0.000421471795773544,
      "loss": 1.9388,
      "step": 84628
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.823020339012146,
      "learning_rate": 0.0004214680555226851,
      "loss": 2.2049,
      "step": 84629
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.8022837042808533,
      "learning_rate": 0.00042146431524924313,
      "loss": 1.9355,
      "step": 84630
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.8162601590156555,
      "learning_rate": 0.0004214605749532187,
      "loss": 1.9825,
      "step": 84631
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.794147789478302,
      "learning_rate": 0.0004214568346346123,
      "loss": 2.0963,
      "step": 84632
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.8555383086204529,
      "learning_rate": 0.0004214530942934248,
      "loss": 2.3487,
      "step": 84633
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.8528575301170349,
      "learning_rate": 0.00042144935392965697,
      "loss": 2.2997,
      "step": 84634
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.8519964814186096,
      "learning_rate": 0.00042144561354330937,
      "loss": 2.3148,
      "step": 84635
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.8290736079216003,
      "learning_rate": 0.0004214418731343827,
      "loss": 2.2731,
      "step": 84636
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.7393144369125366,
      "learning_rate": 0.00042143813270287777,
      "loss": 2.2619,
      "step": 84637
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.8099507689476013,
      "learning_rate": 0.0004214343922487951,
      "loss": 1.9131,
      "step": 84638
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.7838436961174011,
      "learning_rate": 0.00042143065177213565,
      "loss": 2.2417,
      "step": 84639
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.9530293941497803,
      "learning_rate": 0.0004214269112728998,
      "loss": 2.2562,
      "step": 84640
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.787993311882019,
      "learning_rate": 0.0004214231707510884,
      "loss": 2.0761,
      "step": 84641
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.7663672566413879,
      "learning_rate": 0.0004214194302067022,
      "loss": 1.9492,
      "step": 84642
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.7843601107597351,
      "learning_rate": 0.00042141568963974187,
      "loss": 2.256,
      "step": 84643
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.7845779657363892,
      "learning_rate": 0.000421411949050208,
      "loss": 2.1029,
      "step": 84644
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.8472983837127686,
      "learning_rate": 0.0004214082084381013,
      "loss": 2.2573,
      "step": 84645
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.7234196662902832,
      "learning_rate": 0.0004214044678034227,
      "loss": 2.1674,
      "step": 84646
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.7373520135879517,
      "learning_rate": 0.00042140072714617254,
      "loss": 2.0964,
      "step": 84647
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.8103972673416138,
      "learning_rate": 0.00042139698646635174,
      "loss": 2.2865,
      "step": 84648
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.766299307346344,
      "learning_rate": 0.00042139324576396104,
      "loss": 2.1294,
      "step": 84649
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.9472388625144958,
      "learning_rate": 0.000421389505039001,
      "loss": 2.3733,
      "step": 84650
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.9514774680137634,
      "learning_rate": 0.0004213857642914723,
      "loss": 2.0734,
      "step": 84651
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.824316680431366,
      "learning_rate": 0.0004213820235213757,
      "loss": 2.1268,
      "step": 84652
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.9423574209213257,
      "learning_rate": 0.00042137828272871195,
      "loss": 2.3022,
      "step": 84653
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.8586681485176086,
      "learning_rate": 0.0004213745419134817,
      "loss": 2.3627,
      "step": 84654
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.8545203804969788,
      "learning_rate": 0.00042137080107568556,
      "loss": 2.4198,
      "step": 84655
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.8291406631469727,
      "learning_rate": 0.00042136706021532434,
      "loss": 2.2743,
      "step": 84656
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.7994438409805298,
      "learning_rate": 0.00042136331933239864,
      "loss": 2.2195,
      "step": 84657
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.895149827003479,
      "learning_rate": 0.0004213595784269093,
      "loss": 2.0891,
      "step": 84658
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.7572851777076721,
      "learning_rate": 0.0004213558374988569,
      "loss": 2.2867,
      "step": 84659
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.7937840223312378,
      "learning_rate": 0.0004213520965482421,
      "loss": 2.1975,
      "step": 84660
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.8292528986930847,
      "learning_rate": 0.0004213483555750657,
      "loss": 2.1053,
      "step": 84661
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.8482562899589539,
      "learning_rate": 0.0004213446145793284,
      "loss": 2.0333,
      "step": 84662
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.8723231554031372,
      "learning_rate": 0.0004213408735610308,
      "loss": 2.2035,
      "step": 84663
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.777617871761322,
      "learning_rate": 0.0004213371325201735,
      "loss": 2.0708,
      "step": 84664
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.8766306042671204,
      "learning_rate": 0.00042133339145675755,
      "loss": 2.0876,
      "step": 84665
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.8145902752876282,
      "learning_rate": 0.00042132965037078334,
      "loss": 2.3297,
      "step": 84666
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.7819159626960754,
      "learning_rate": 0.0004213259092622516,
      "loss": 2.0966,
      "step": 84667
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.957301139831543,
      "learning_rate": 0.0004213221681311632,
      "loss": 2.2346,
      "step": 84668
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.7972890734672546,
      "learning_rate": 0.00042131842697751867,
      "loss": 2.2447,
      "step": 84669
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.8041067719459534,
      "learning_rate": 0.00042131468580131867,
      "loss": 2.0993,
      "step": 84670
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.8153960704803467,
      "learning_rate": 0.0004213109446025642,
      "loss": 2.0854,
      "step": 84671
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.8193444013595581,
      "learning_rate": 0.0004213072033812555,
      "loss": 2.2867,
      "step": 84672
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.8531634211540222,
      "learning_rate": 0.0004213034621373936,
      "loss": 2.1814,
      "step": 84673
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.8803592920303345,
      "learning_rate": 0.0004212997208709791,
      "loss": 1.9762,
      "step": 84674
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.7465489506721497,
      "learning_rate": 0.00042129597958201275,
      "loss": 2.2633,
      "step": 84675
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.9196060299873352,
      "learning_rate": 0.00042129223827049505,
      "loss": 2.1952,
      "step": 84676
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.9297918081283569,
      "learning_rate": 0.0004212884969364269,
      "loss": 2.1579,
      "step": 84677
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.7335121035575867,
      "learning_rate": 0.00042128475557980897,
      "loss": 2.1861,
      "step": 84678
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.7921279668807983,
      "learning_rate": 0.0004212810142006419,
      "loss": 2.1605,
      "step": 84679
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.8468453288078308,
      "learning_rate": 0.00042127727279892636,
      "loss": 2.2134,
      "step": 84680
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.8141107559204102,
      "learning_rate": 0.0004212735313746631,
      "loss": 2.0404,
      "step": 84681
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.7907003164291382,
      "learning_rate": 0.0004212697899278528,
      "loss": 2.1584,
      "step": 84682
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.7705532908439636,
      "learning_rate": 0.00042126604845849617,
      "loss": 1.9089,
      "step": 84683
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.8125872015953064,
      "learning_rate": 0.0004212623069665939,
      "loss": 2.3804,
      "step": 84684
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.8199456334114075,
      "learning_rate": 0.0004212585654521466,
      "loss": 2.3138,
      "step": 84685
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.9333600401878357,
      "learning_rate": 0.0004212548239151552,
      "loss": 2.1547,
      "step": 84686
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.7993083000183105,
      "learning_rate": 0.0004212510823556201,
      "loss": 2.4409,
      "step": 84687
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.8812504410743713,
      "learning_rate": 0.00042124734077354217,
      "loss": 2.3409,
      "step": 84688
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.8601145148277283,
      "learning_rate": 0.00042124359916892207,
      "loss": 2.2947,
      "step": 84689
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.8204858303070068,
      "learning_rate": 0.00042123985754176054,
      "loss": 2.2291,
      "step": 84690
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.8438476920127869,
      "learning_rate": 0.0004212361158920582,
      "loss": 2.2575,
      "step": 84691
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.8681350946426392,
      "learning_rate": 0.00042123237421981584,
      "loss": 2.0755,
      "step": 84692
    },
    {
      "epoch": 1.1,
      "grad_norm": 1.179760456085205,
      "learning_rate": 0.0004212286325250339,
      "loss": 2.0348,
      "step": 84693
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.7677989602088928,
      "learning_rate": 0.00042122489080771345,
      "loss": 1.9885,
      "step": 84694
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.859209418296814,
      "learning_rate": 0.000421221149067855,
      "loss": 2.2775,
      "step": 84695
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.7804763913154602,
      "learning_rate": 0.00042121740730545915,
      "loss": 2.122,
      "step": 84696
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.7775442600250244,
      "learning_rate": 0.0004212136655205268,
      "loss": 2.1966,
      "step": 84697
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.8282973170280457,
      "learning_rate": 0.0004212099237130585,
      "loss": 2.2092,
      "step": 84698
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.7903985977172852,
      "learning_rate": 0.000421206181883055,
      "loss": 2.2476,
      "step": 84699
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.9635006785392761,
      "learning_rate": 0.00042120244003051693,
      "loss": 2.307,
      "step": 84700
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.8277245163917542,
      "learning_rate": 0.0004211986981554451,
      "loss": 2.204,
      "step": 84701
    },
    {
      "epoch": 1.1,
      "grad_norm": 1.208732008934021,
      "learning_rate": 0.00042119495625784006,
      "loss": 2.259,
      "step": 84702
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.8541164994239807,
      "learning_rate": 0.00042119121433770266,
      "loss": 1.962,
      "step": 84703
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.7379263639450073,
      "learning_rate": 0.0004211874723950335,
      "loss": 2.0095,
      "step": 84704
    },
    {
      "epoch": 1.1,
      "grad_norm": 1.487309455871582,
      "learning_rate": 0.0004211837304298333,
      "loss": 2.148,
      "step": 84705
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.865399956703186,
      "learning_rate": 0.0004211799884421028,
      "loss": 2.2074,
      "step": 84706
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.8293666243553162,
      "learning_rate": 0.00042117624643184265,
      "loss": 1.957,
      "step": 84707
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.8365475535392761,
      "learning_rate": 0.0004211725043990535,
      "loss": 2.3532,
      "step": 84708
    },
    {
      "epoch": 1.1,
      "grad_norm": 1.0654550790786743,
      "learning_rate": 0.0004211687623437361,
      "loss": 2.3175,
      "step": 84709
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.804092288017273,
      "learning_rate": 0.00042116502026589116,
      "loss": 2.105,
      "step": 84710
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.8208259344100952,
      "learning_rate": 0.00042116127816551935,
      "loss": 2.0698,
      "step": 84711
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.84536212682724,
      "learning_rate": 0.00042115753604262134,
      "loss": 2.2427,
      "step": 84712
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.8404999375343323,
      "learning_rate": 0.0004211537938971979,
      "loss": 2.1266,
      "step": 84713
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.8849474787712097,
      "learning_rate": 0.0004211500517292497,
      "loss": 2.2795,
      "step": 84714
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.8080591559410095,
      "learning_rate": 0.0004211463095387774,
      "loss": 2.3959,
      "step": 84715
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.8156775236129761,
      "learning_rate": 0.0004211425673257816,
      "loss": 2.196,
      "step": 84716
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.7546651363372803,
      "learning_rate": 0.0004211388250902633,
      "loss": 2.3268,
      "step": 84717
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.7552717924118042,
      "learning_rate": 0.00042113508283222285,
      "loss": 2.27,
      "step": 84718
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.897909939289093,
      "learning_rate": 0.00042113134055166124,
      "loss": 2.2062,
      "step": 84719
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.7895414233207703,
      "learning_rate": 0.0004211275982485789,
      "loss": 1.9589,
      "step": 84720
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.8590485453605652,
      "learning_rate": 0.00042112385592297673,
      "loss": 1.9468,
      "step": 84721
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.8013895750045776,
      "learning_rate": 0.00042112011357485534,
      "loss": 2.2724,
      "step": 84722
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.7867401242256165,
      "learning_rate": 0.0004211163712042154,
      "loss": 2.0297,
      "step": 84723
    },
    {
      "epoch": 1.1,
      "grad_norm": 1.0238293409347534,
      "learning_rate": 0.0004211126288110577,
      "loss": 2.2951,
      "step": 84724
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.8564858436584473,
      "learning_rate": 0.00042110888639538283,
      "loss": 1.9975,
      "step": 84725
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.8034760355949402,
      "learning_rate": 0.0004211051439571915,
      "loss": 2.2243,
      "step": 84726
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.9726179242134094,
      "learning_rate": 0.00042110140149648456,
      "loss": 2.1638,
      "step": 84727
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.9086870551109314,
      "learning_rate": 0.0004210976590132624,
      "loss": 2.3058,
      "step": 84728
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.8826450705528259,
      "learning_rate": 0.00042109391650752606,
      "loss": 2.0897,
      "step": 84729
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.876242995262146,
      "learning_rate": 0.0004210901739792761,
      "loss": 2.1574,
      "step": 84730
    },
    {
      "epoch": 1.1,
      "grad_norm": 1.1529463529586792,
      "learning_rate": 0.000421086431428513,
      "loss": 2.2565,
      "step": 84731
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.9047800898551941,
      "learning_rate": 0.00042108268885523775,
      "loss": 2.0498,
      "step": 84732
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.9650924205780029,
      "learning_rate": 0.0004210789462594509,
      "loss": 2.385,
      "step": 84733
    },
    {
      "epoch": 1.1,
      "grad_norm": 1.163329839706421,
      "learning_rate": 0.00042107520364115334,
      "loss": 2.1531,
      "step": 84734
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.7829297184944153,
      "learning_rate": 0.00042107146100034545,
      "loss": 2.0987,
      "step": 84735
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.8536416292190552,
      "learning_rate": 0.0004210677183370281,
      "loss": 2.3737,
      "step": 84736
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.9718017578125,
      "learning_rate": 0.0004210639756512021,
      "loss": 2.2299,
      "step": 84737
    },
    {
      "epoch": 1.1,
      "grad_norm": 1.0299465656280518,
      "learning_rate": 0.00042106023294286797,
      "loss": 2.1066,
      "step": 84738
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.7365350127220154,
      "learning_rate": 0.00042105649021202636,
      "loss": 2.1746,
      "step": 84739
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.8367244601249695,
      "learning_rate": 0.00042105274745867817,
      "loss": 2.0013,
      "step": 84740
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.952320396900177,
      "learning_rate": 0.000421049004682824,
      "loss": 2.3042,
      "step": 84741
    },
    {
      "epoch": 1.1,
      "grad_norm": 1.0015156269073486,
      "learning_rate": 0.0004210452618844645,
      "loss": 2.3766,
      "step": 84742
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.7953089475631714,
      "learning_rate": 0.00042104151906360035,
      "loss": 2.2235,
      "step": 84743
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.8779036402702332,
      "learning_rate": 0.0004210377762202324,
      "loss": 2.4618,
      "step": 84744
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.8207556009292603,
      "learning_rate": 0.00042103403335436114,
      "loss": 1.951,
      "step": 84745
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.8298118710517883,
      "learning_rate": 0.00042103029046598744,
      "loss": 2.2683,
      "step": 84746
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.9818238019943237,
      "learning_rate": 0.0004210265475551119,
      "loss": 2.1107,
      "step": 84747
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.8888843059539795,
      "learning_rate": 0.0004210228046217352,
      "loss": 2.2688,
      "step": 84748
    },
    {
      "epoch": 1.1,
      "grad_norm": 1.0591504573822021,
      "learning_rate": 0.0004210190616658581,
      "loss": 2.206,
      "step": 84749
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.8587420582771301,
      "learning_rate": 0.0004210153186874813,
      "loss": 2.2252,
      "step": 84750
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.752740204334259,
      "learning_rate": 0.0004210115756866055,
      "loss": 2.1328,
      "step": 84751
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.8814853429794312,
      "learning_rate": 0.0004210078326632312,
      "loss": 2.3175,
      "step": 84752
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.9480266571044922,
      "learning_rate": 0.00042100408961735943,
      "loss": 2.4299,
      "step": 84753
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.837243914604187,
      "learning_rate": 0.00042100034654899066,
      "loss": 2.0155,
      "step": 84754
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.7846843600273132,
      "learning_rate": 0.0004209966034581257,
      "loss": 2.2648,
      "step": 84755
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.7845278978347778,
      "learning_rate": 0.0004209928603447651,
      "loss": 2.2043,
      "step": 84756
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.8389118313789368,
      "learning_rate": 0.00042098911720890966,
      "loss": 2.1258,
      "step": 84757
    },
    {
      "epoch": 1.1,
      "grad_norm": 1.0721893310546875,
      "learning_rate": 0.0004209853740505601,
      "loss": 2.1429,
      "step": 84758
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.9344087243080139,
      "learning_rate": 0.0004209816308697171,
      "loss": 2.3638,
      "step": 84759
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.9158050417900085,
      "learning_rate": 0.0004209778876663812,
      "loss": 2.2177,
      "step": 84760
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.8423876762390137,
      "learning_rate": 0.00042097414444055333,
      "loss": 2.2232,
      "step": 84761
    },
    {
      "epoch": 1.1,
      "grad_norm": 1.0830721855163574,
      "learning_rate": 0.00042097040119223406,
      "loss": 2.2385,
      "step": 84762
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.9202588200569153,
      "learning_rate": 0.00042096665792142415,
      "loss": 2.0999,
      "step": 84763
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.8031112551689148,
      "learning_rate": 0.00042096291462812415,
      "loss": 2.4798,
      "step": 84764
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.9098652005195618,
      "learning_rate": 0.0004209591713123349,
      "loss": 2.3081,
      "step": 84765
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.8564488887786865,
      "learning_rate": 0.0004209554279740571,
      "loss": 2.4043,
      "step": 84766
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.8000081181526184,
      "learning_rate": 0.00042095168461329143,
      "loss": 2.238,
      "step": 84767
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.7946696877479553,
      "learning_rate": 0.00042094794123003854,
      "loss": 2.2714,
      "step": 84768
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.8539766073226929,
      "learning_rate": 0.00042094419782429907,
      "loss": 2.1911,
      "step": 84769
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.8092125058174133,
      "learning_rate": 0.0004209404543960738,
      "loss": 2.1575,
      "step": 84770
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.8690043687820435,
      "learning_rate": 0.0004209367109453635,
      "loss": 2.1046,
      "step": 84771
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.8306646347045898,
      "learning_rate": 0.0004209329674721688,
      "loss": 2.2525,
      "step": 84772
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.7865267395973206,
      "learning_rate": 0.00042092922397649025,
      "loss": 2.1439,
      "step": 84773
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.8738778233528137,
      "learning_rate": 0.0004209254804583288,
      "loss": 2.0219,
      "step": 84774
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.8515051007270813,
      "learning_rate": 0.0004209217369176849,
      "loss": 2.1212,
      "step": 84775
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.7650913000106812,
      "learning_rate": 0.0004209179933545594,
      "loss": 2.1819,
      "step": 84776
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.8152673840522766,
      "learning_rate": 0.00042091424976895296,
      "loss": 2.2312,
      "step": 84777
    },
    {
      "epoch": 1.1,
      "grad_norm": 1.098138689994812,
      "learning_rate": 0.0004209105061608664,
      "loss": 2.1938,
      "step": 84778
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.7912413477897644,
      "learning_rate": 0.00042090676253030016,
      "loss": 2.1715,
      "step": 84779
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.8138352036476135,
      "learning_rate": 0.0004209030188772551,
      "loss": 2.1771,
      "step": 84780
    },
    {
      "epoch": 1.1,
      "grad_norm": 1.0575002431869507,
      "learning_rate": 0.0004208992752017319,
      "loss": 2.4059,
      "step": 84781
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.9429829120635986,
      "learning_rate": 0.00042089553150373123,
      "loss": 2.1796,
      "step": 84782
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.8819922804832458,
      "learning_rate": 0.0004208917877832538,
      "loss": 2.4148,
      "step": 84783
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.9179839491844177,
      "learning_rate": 0.0004208880440403004,
      "loss": 2.2506,
      "step": 84784
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.9826653003692627,
      "learning_rate": 0.0004208843002748715,
      "loss": 2.1588,
      "step": 84785
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.8474801182746887,
      "learning_rate": 0.00042088055648696794,
      "loss": 2.0646,
      "step": 84786
    },
    {
      "epoch": 1.1,
      "grad_norm": 1.0447978973388672,
      "learning_rate": 0.0004208768126765905,
      "loss": 2.1795,
      "step": 84787
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.834550678730011,
      "learning_rate": 0.0004208730688437397,
      "loss": 2.4064,
      "step": 84788
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.8119718432426453,
      "learning_rate": 0.00042086932498841637,
      "loss": 2.1259,
      "step": 84789
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.7547709941864014,
      "learning_rate": 0.0004208655811106211,
      "loss": 2.3686,
      "step": 84790
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.8401750326156616,
      "learning_rate": 0.0004208618372103546,
      "loss": 2.2671,
      "step": 84791
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.842450737953186,
      "learning_rate": 0.0004208580932876177,
      "loss": 2.1588,
      "step": 84792
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.8272616863250732,
      "learning_rate": 0.0004208543493424109,
      "loss": 2.4521,
      "step": 84793
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.8335622549057007,
      "learning_rate": 0.00042085060537473505,
      "loss": 2.2664,
      "step": 84794
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.7991335988044739,
      "learning_rate": 0.0004208468613845907,
      "loss": 1.9754,
      "step": 84795
    },
    {
      "epoch": 1.1,
      "grad_norm": 1.0317093133926392,
      "learning_rate": 0.00042084311737197884,
      "loss": 2.3481,
      "step": 84796
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.8833746314048767,
      "learning_rate": 0.0004208393733368998,
      "loss": 2.2684,
      "step": 84797
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.8347947597503662,
      "learning_rate": 0.00042083562927935453,
      "loss": 2.1628,
      "step": 84798
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.9855279326438904,
      "learning_rate": 0.0004208318851993436,
      "loss": 2.1529,
      "step": 84799
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.7903975248336792,
      "learning_rate": 0.0004208281410968677,
      "loss": 2.4589,
      "step": 84800
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.7508310675621033,
      "learning_rate": 0.0004208243969719276,
      "loss": 2.3327,
      "step": 84801
    },
    {
      "epoch": 1.1,
      "grad_norm": 2.5288853645324707,
      "learning_rate": 0.000420820652824524,
      "loss": 2.1831,
      "step": 84802
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.9042905569076538,
      "learning_rate": 0.00042081690865465745,
      "loss": 2.463,
      "step": 84803
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.8688248991966248,
      "learning_rate": 0.00042081316446232887,
      "loss": 2.1736,
      "step": 84804
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.8128952980041504,
      "learning_rate": 0.0004208094202475388,
      "loss": 2.2697,
      "step": 84805
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.8262803554534912,
      "learning_rate": 0.000420805676010288,
      "loss": 1.9018,
      "step": 84806
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.9153228998184204,
      "learning_rate": 0.0004208019317505771,
      "loss": 2.1402,
      "step": 84807
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.7933900952339172,
      "learning_rate": 0.0004207981874684069,
      "loss": 2.0728,
      "step": 84808
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.7679408192634583,
      "learning_rate": 0.000420794443163778,
      "loss": 2.1383,
      "step": 84809
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.7453118562698364,
      "learning_rate": 0.0004207906988366911,
      "loss": 2.2269,
      "step": 84810
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.8866788148880005,
      "learning_rate": 0.000420786954487147,
      "loss": 2.1193,
      "step": 84811
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.7985972166061401,
      "learning_rate": 0.0004207832101151463,
      "loss": 2.1936,
      "step": 84812
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.797293484210968,
      "learning_rate": 0.00042077946572068975,
      "loss": 2.1976,
      "step": 84813
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.8778148293495178,
      "learning_rate": 0.0004207757213037779,
      "loss": 2.197,
      "step": 84814
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.7691678404808044,
      "learning_rate": 0.0004207719768644117,
      "loss": 1.9283,
      "step": 84815
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.9395890235900879,
      "learning_rate": 0.0004207682324025916,
      "loss": 2.3982,
      "step": 84816
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.8793709874153137,
      "learning_rate": 0.00042076448791831844,
      "loss": 2.1896,
      "step": 84817
    },
    {
      "epoch": 1.1,
      "grad_norm": 1.0818239450454712,
      "learning_rate": 0.0004207607434115929,
      "loss": 2.2256,
      "step": 84818
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.8374897241592407,
      "learning_rate": 0.0004207569988824157,
      "loss": 2.0955,
      "step": 84819
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.8205899000167847,
      "learning_rate": 0.00042075325433078744,
      "loss": 2.0925,
      "step": 84820
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.8615878820419312,
      "learning_rate": 0.0004207495097567089,
      "loss": 2.3332,
      "step": 84821
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.7788600921630859,
      "learning_rate": 0.0004207457651601807,
      "loss": 2.1644,
      "step": 84822
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.9052727222442627,
      "learning_rate": 0.0004207420205412036,
      "loss": 2.0431,
      "step": 84823
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.7838167548179626,
      "learning_rate": 0.0004207382758997783,
      "loss": 2.01,
      "step": 84824
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.7660532593727112,
      "learning_rate": 0.0004207345312359055,
      "loss": 2.1533,
      "step": 84825
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.8259730339050293,
      "learning_rate": 0.0004207307865495858,
      "loss": 2.1059,
      "step": 84826
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.7960045337677002,
      "learning_rate": 0.00042072704184082006,
      "loss": 2.2683,
      "step": 84827
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.8837535977363586,
      "learning_rate": 0.0004207232971096088,
      "loss": 2.1816,
      "step": 84828
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.8227670192718506,
      "learning_rate": 0.0004207195523559528,
      "loss": 2.11,
      "step": 84829
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.8052159547805786,
      "learning_rate": 0.0004207158075798528,
      "loss": 2.248,
      "step": 84830
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.873817503452301,
      "learning_rate": 0.0004207120627813094,
      "loss": 2.1158,
      "step": 84831
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.8408767580986023,
      "learning_rate": 0.00042070831796032343,
      "loss": 2.3581,
      "step": 84832
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.8053997159004211,
      "learning_rate": 0.0004207045731168955,
      "loss": 2.0068,
      "step": 84833
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.8703901767730713,
      "learning_rate": 0.00042070082825102623,
      "loss": 2.1815,
      "step": 84834
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.9264363646507263,
      "learning_rate": 0.0004206970833627165,
      "loss": 2.2273,
      "step": 84835
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.755709707736969,
      "learning_rate": 0.0004206933384519668,
      "loss": 2.3232,
      "step": 84836
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.914588987827301,
      "learning_rate": 0.000420689593518778,
      "loss": 2.1518,
      "step": 84837
    },
    {
      "epoch": 1.1,
      "grad_norm": 1.0088322162628174,
      "learning_rate": 0.0004206858485631507,
      "loss": 2.1458,
      "step": 84838
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.8214133977890015,
      "learning_rate": 0.0004206821035850856,
      "loss": 2.2393,
      "step": 84839
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.9637120366096497,
      "learning_rate": 0.0004206783585845835,
      "loss": 2.0205,
      "step": 84840
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.7942929863929749,
      "learning_rate": 0.00042067461356164496,
      "loss": 2.0964,
      "step": 84841
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.8802812099456787,
      "learning_rate": 0.0004206708685162707,
      "loss": 2.1302,
      "step": 84842
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.7634038329124451,
      "learning_rate": 0.00042066712344846147,
      "loss": 2.0416,
      "step": 84843
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.7997186779975891,
      "learning_rate": 0.00042066337835821796,
      "loss": 2.3013,
      "step": 84844
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.8861427307128906,
      "learning_rate": 0.00042065963324554084,
      "loss": 2.4938,
      "step": 84845
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.7835862040519714,
      "learning_rate": 0.0004206558881104308,
      "loss": 2.0571,
      "step": 84846
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.8619831800460815,
      "learning_rate": 0.0004206521429528886,
      "loss": 2.1744,
      "step": 84847
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.7891262769699097,
      "learning_rate": 0.00042064839777291495,
      "loss": 2.2152,
      "step": 84848
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.8294413089752197,
      "learning_rate": 0.00042064465257051034,
      "loss": 2.1638,
      "step": 84849
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.8273054957389832,
      "learning_rate": 0.00042064090734567566,
      "loss": 2.5796,
      "step": 84850
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.7877978682518005,
      "learning_rate": 0.0004206371620984116,
      "loss": 2.2566,
      "step": 84851
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.8534640073776245,
      "learning_rate": 0.00042063341682871874,
      "loss": 2.2401,
      "step": 84852
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.7847154140472412,
      "learning_rate": 0.00042062967153659787,
      "loss": 2.3665,
      "step": 84853
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.7664341926574707,
      "learning_rate": 0.00042062592622204973,
      "loss": 2.0271,
      "step": 84854
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.8820683360099792,
      "learning_rate": 0.00042062218088507493,
      "loss": 2.1764,
      "step": 84855
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.8357759714126587,
      "learning_rate": 0.0004206184355256741,
      "loss": 1.9609,
      "step": 84856
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.8873968124389648,
      "learning_rate": 0.00042061469014384814,
      "loss": 2.2509,
      "step": 84857
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.8842947483062744,
      "learning_rate": 0.00042061094473959756,
      "loss": 2.1065,
      "step": 84858
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.8775890469551086,
      "learning_rate": 0.00042060719931292313,
      "loss": 2.319,
      "step": 84859
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.796789288520813,
      "learning_rate": 0.00042060345386382567,
      "loss": 2.1427,
      "step": 84860
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.7724055647850037,
      "learning_rate": 0.0004205997083923056,
      "loss": 2.034,
      "step": 84861
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.9003016352653503,
      "learning_rate": 0.0004205959628983638,
      "loss": 2.1575,
      "step": 84862
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.7806779742240906,
      "learning_rate": 0.00042059221738200103,
      "loss": 2.4013,
      "step": 84863
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.8848811388015747,
      "learning_rate": 0.0004205884718432178,
      "loss": 2.3208,
      "step": 84864
    },
    {
      "epoch": 1.1,
      "grad_norm": 0.7899201512336731,
      "learning_rate": 0.0004205847262820149,
      "loss": 2.2051,
      "step": 84865
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.8213616609573364,
      "learning_rate": 0.0004205809806983931,
      "loss": 2.1004,
      "step": 84866
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.7829725742340088,
      "learning_rate": 0.0004205772350923529,
      "loss": 2.1576,
      "step": 84867
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.7649094462394714,
      "learning_rate": 0.0004205734894638952,
      "loss": 2.1793,
      "step": 84868
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.8491531610488892,
      "learning_rate": 0.0004205697438130206,
      "loss": 2.182,
      "step": 84869
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.8625372052192688,
      "learning_rate": 0.0004205659981397298,
      "loss": 2.1962,
      "step": 84870
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.8042615652084351,
      "learning_rate": 0.00042056225244402345,
      "loss": 2.1114,
      "step": 84871
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.7989990711212158,
      "learning_rate": 0.0004205585067259024,
      "loss": 2.205,
      "step": 84872
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.8256434798240662,
      "learning_rate": 0.0004205547609853672,
      "loss": 2.1517,
      "step": 84873
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.8237212300300598,
      "learning_rate": 0.00042055101522241856,
      "loss": 2.2778,
      "step": 84874
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.8761471509933472,
      "learning_rate": 0.00042054726943705726,
      "loss": 2.3005,
      "step": 84875
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.7603906989097595,
      "learning_rate": 0.000420543523629284,
      "loss": 2.0515,
      "step": 84876
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.7817615270614624,
      "learning_rate": 0.00042053977779909926,
      "loss": 2.1831,
      "step": 84877
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.813136100769043,
      "learning_rate": 0.00042053603194650404,
      "loss": 2.1373,
      "step": 84878
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.8293558359146118,
      "learning_rate": 0.00042053228607149884,
      "loss": 2.1429,
      "step": 84879
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.7987191677093506,
      "learning_rate": 0.0004205285401740844,
      "loss": 2.2149,
      "step": 84880
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.9348205327987671,
      "learning_rate": 0.00042052479425426146,
      "loss": 2.163,
      "step": 84881
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.9157043695449829,
      "learning_rate": 0.0004205210483120307,
      "loss": 2.3075,
      "step": 84882
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.854182779788971,
      "learning_rate": 0.0004205173023473928,
      "loss": 2.077,
      "step": 84883
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.9342398047447205,
      "learning_rate": 0.00042051355636034843,
      "loss": 2.1496,
      "step": 84884
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.9600440263748169,
      "learning_rate": 0.0004205098103508983,
      "loss": 2.1093,
      "step": 84885
    },
    {
      "epoch": 1.11,
      "grad_norm": 1.0216516256332397,
      "learning_rate": 0.00042050606431904316,
      "loss": 2.3694,
      "step": 84886
    },
    {
      "epoch": 1.11,
      "grad_norm": 1.0393787622451782,
      "learning_rate": 0.0004205023182647837,
      "loss": 2.17,
      "step": 84887
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.8570615649223328,
      "learning_rate": 0.0004204985721881205,
      "loss": 2.2752,
      "step": 84888
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.8258914947509766,
      "learning_rate": 0.00042049482608905436,
      "loss": 2.1392,
      "step": 84889
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.8542423844337463,
      "learning_rate": 0.00042049107996758604,
      "loss": 2.0302,
      "step": 84890
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.854679524898529,
      "learning_rate": 0.00042048733382371605,
      "loss": 2.0533,
      "step": 84891
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.8914042115211487,
      "learning_rate": 0.0004204835876574452,
      "loss": 2.1803,
      "step": 84892
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.8338227272033691,
      "learning_rate": 0.0004204798414687743,
      "loss": 2.2965,
      "step": 84893
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.9005901217460632,
      "learning_rate": 0.0004204760952577038,
      "loss": 2.1004,
      "step": 84894
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.8511793613433838,
      "learning_rate": 0.00042047234902423456,
      "loss": 2.2839,
      "step": 84895
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.832515299320221,
      "learning_rate": 0.00042046860276836724,
      "loss": 2.1141,
      "step": 84896
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.8180915713310242,
      "learning_rate": 0.0004204648564901025,
      "loss": 1.9139,
      "step": 84897
    },
    {
      "epoch": 1.11,
      "grad_norm": 1.0300277471542358,
      "learning_rate": 0.0004204611101894411,
      "loss": 2.2157,
      "step": 84898
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.8534438610076904,
      "learning_rate": 0.00042045736386638364,
      "loss": 2.3507,
      "step": 84899
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.8304205536842346,
      "learning_rate": 0.000420453617520931,
      "loss": 2.4444,
      "step": 84900
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.855652391910553,
      "learning_rate": 0.00042044987115308363,
      "loss": 2.1346,
      "step": 84901
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.810832142829895,
      "learning_rate": 0.0004204461247628424,
      "loss": 2.1038,
      "step": 84902
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.8206844925880432,
      "learning_rate": 0.00042044237835020797,
      "loss": 2.1029,
      "step": 84903
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.8005918860435486,
      "learning_rate": 0.00042043863191518107,
      "loss": 2.0959,
      "step": 84904
    },
    {
      "epoch": 1.11,
      "grad_norm": 1.2332274913787842,
      "learning_rate": 0.0004204348854577623,
      "loss": 2.0689,
      "step": 84905
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.9882103800773621,
      "learning_rate": 0.00042043113897795244,
      "loss": 2.1402,
      "step": 84906
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.8861929178237915,
      "learning_rate": 0.00042042739247575217,
      "loss": 2.325,
      "step": 84907
    },
    {
      "epoch": 1.11,
      "grad_norm": 1.0016238689422607,
      "learning_rate": 0.00042042364595116214,
      "loss": 2.071,
      "step": 84908
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.857616126537323,
      "learning_rate": 0.0004204198994041831,
      "loss": 2.1683,
      "step": 84909
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.9326931834220886,
      "learning_rate": 0.0004204161528348157,
      "loss": 2.2921,
      "step": 84910
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.8270063400268555,
      "learning_rate": 0.0004204124062430608,
      "loss": 2.2044,
      "step": 84911
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.816479504108429,
      "learning_rate": 0.00042040865962891875,
      "loss": 2.2685,
      "step": 84912
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.8175671100616455,
      "learning_rate": 0.00042040491299239055,
      "loss": 2.0461,
      "step": 84913
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.8495866656303406,
      "learning_rate": 0.0004204011663334769,
      "loss": 2.073,
      "step": 84914
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.829004168510437,
      "learning_rate": 0.0004203974196521783,
      "loss": 2.3554,
      "step": 84915
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.8735123872756958,
      "learning_rate": 0.0004203936729484955,
      "loss": 2.1486,
      "step": 84916
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.7827958464622498,
      "learning_rate": 0.0004203899262224293,
      "loss": 2.2367,
      "step": 84917
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.8980833888053894,
      "learning_rate": 0.0004203861794739804,
      "loss": 2.2095,
      "step": 84918
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.9280189871788025,
      "learning_rate": 0.00042038243270314937,
      "loss": 2.3888,
      "step": 84919
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.8233128786087036,
      "learning_rate": 0.00042037868590993693,
      "loss": 2.3158,
      "step": 84920
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.7918206453323364,
      "learning_rate": 0.00042037493909434395,
      "loss": 2.4093,
      "step": 84921
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.8329049348831177,
      "learning_rate": 0.00042037119225637093,
      "loss": 2.2841,
      "step": 84922
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.8877322673797607,
      "learning_rate": 0.00042036744539601856,
      "loss": 1.9742,
      "step": 84923
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.889104962348938,
      "learning_rate": 0.0004203636985132877,
      "loss": 2.0022,
      "step": 84924
    },
    {
      "epoch": 1.11,
      "grad_norm": 1.0046874284744263,
      "learning_rate": 0.00042035995160817895,
      "loss": 2.23,
      "step": 84925
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.8981911540031433,
      "learning_rate": 0.00042035620468069295,
      "loss": 2.2685,
      "step": 84926
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.7717472314834595,
      "learning_rate": 0.00042035245773083054,
      "loss": 2.0074,
      "step": 84927
    },
    {
      "epoch": 1.11,
      "grad_norm": 1.0066946744918823,
      "learning_rate": 0.00042034871075859236,
      "loss": 2.2829,
      "step": 84928
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.8604186773300171,
      "learning_rate": 0.00042034496376397895,
      "loss": 2.1345,
      "step": 84929
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.8391816020011902,
      "learning_rate": 0.00042034121674699125,
      "loss": 2.5381,
      "step": 84930
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.8142220973968506,
      "learning_rate": 0.0004203374697076298,
      "loss": 2.1412,
      "step": 84931
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.857437014579773,
      "learning_rate": 0.00042033372264589533,
      "loss": 2.0609,
      "step": 84932
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.771472692489624,
      "learning_rate": 0.00042032997556178857,
      "loss": 2.3005,
      "step": 84933
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.843163251876831,
      "learning_rate": 0.0004203262284553102,
      "loss": 2.2428,
      "step": 84934
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.7676634788513184,
      "learning_rate": 0.0004203224813264609,
      "loss": 2.0985,
      "step": 84935
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.9257640242576599,
      "learning_rate": 0.0004203187341752414,
      "loss": 2.3777,
      "step": 84936
    },
    {
      "epoch": 1.11,
      "grad_norm": 1.00433349609375,
      "learning_rate": 0.0004203149870016524,
      "loss": 2.2671,
      "step": 84937
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.866968035697937,
      "learning_rate": 0.0004203112398056945,
      "loss": 2.1309,
      "step": 84938
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.9435493350028992,
      "learning_rate": 0.0004203074925873686,
      "loss": 2.12,
      "step": 84939
    },
    {
      "epoch": 1.11,
      "grad_norm": 1.0017544031143188,
      "learning_rate": 0.0004203037453466751,
      "loss": 2.2085,
      "step": 84940
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.8917931914329529,
      "learning_rate": 0.0004202999980836149,
      "loss": 2.3151,
      "step": 84941
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.8827200531959534,
      "learning_rate": 0.00042029625079818876,
      "loss": 2.2745,
      "step": 84942
    },
    {
      "epoch": 1.11,
      "grad_norm": 1.3562625646591187,
      "learning_rate": 0.0004202925034903972,
      "loss": 2.1844,
      "step": 84943
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.967205286026001,
      "learning_rate": 0.000420288756160241,
      "loss": 2.3245,
      "step": 84944
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.7803433537483215,
      "learning_rate": 0.0004202850088077209,
      "loss": 2.2667,
      "step": 84945
    },
    {
      "epoch": 1.11,
      "grad_norm": 1.2043070793151855,
      "learning_rate": 0.0004202812614328374,
      "loss": 2.1557,
      "step": 84946
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.9734296202659607,
      "learning_rate": 0.0004202775140355914,
      "loss": 2.0271,
      "step": 84947
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.8483076691627502,
      "learning_rate": 0.00042027376661598367,
      "loss": 2.2514,
      "step": 84948
    },
    {
      "epoch": 1.11,
      "grad_norm": 1.0086432695388794,
      "learning_rate": 0.00042027001917401474,
      "loss": 2.0613,
      "step": 84949
    },
    {
      "epoch": 1.11,
      "grad_norm": 1.0385817289352417,
      "learning_rate": 0.0004202662717096852,
      "loss": 2.0192,
      "step": 84950
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.7977290153503418,
      "learning_rate": 0.0004202625242229961,
      "loss": 2.1579,
      "step": 84951
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.7861172556877136,
      "learning_rate": 0.00042025877671394773,
      "loss": 2.103,
      "step": 84952
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.8572934865951538,
      "learning_rate": 0.00042025502918254106,
      "loss": 2.0281,
      "step": 84953
    },
    {
      "epoch": 1.11,
      "grad_norm": 1.02254056930542,
      "learning_rate": 0.00042025128162877674,
      "loss": 2.1225,
      "step": 84954
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.9207874536514282,
      "learning_rate": 0.0004202475340526554,
      "loss": 2.0985,
      "step": 84955
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.78508460521698,
      "learning_rate": 0.00042024378645417776,
      "loss": 2.0815,
      "step": 84956
    },
    {
      "epoch": 1.11,
      "grad_norm": 1.1413673162460327,
      "learning_rate": 0.0004202400388333446,
      "loss": 2.0263,
      "step": 84957
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.8020426630973816,
      "learning_rate": 0.00042023629119015656,
      "loss": 2.2557,
      "step": 84958
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.8751927018165588,
      "learning_rate": 0.0004202325435246142,
      "loss": 2.355,
      "step": 84959
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.846169650554657,
      "learning_rate": 0.00042022879583671844,
      "loss": 1.8845,
      "step": 84960
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.805004894733429,
      "learning_rate": 0.0004202250481264698,
      "loss": 2.2021,
      "step": 84961
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.920749843120575,
      "learning_rate": 0.0004202213003938691,
      "loss": 2.2366,
      "step": 84962
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.7599351406097412,
      "learning_rate": 0.00042021755263891707,
      "loss": 2.0346,
      "step": 84963
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.8219725489616394,
      "learning_rate": 0.00042021380486161416,
      "loss": 2.2697,
      "step": 84964
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.8558619022369385,
      "learning_rate": 0.00042021005706196135,
      "loss": 2.3002,
      "step": 84965
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.960436224937439,
      "learning_rate": 0.00042020630923995917,
      "loss": 2.1093,
      "step": 84966
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.7541775107383728,
      "learning_rate": 0.0004202025613956084,
      "loss": 2.1751,
      "step": 84967
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.8287792801856995,
      "learning_rate": 0.00042019881352890965,
      "loss": 2.1156,
      "step": 84968
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.9035105109214783,
      "learning_rate": 0.0004201950656398638,
      "loss": 2.0029,
      "step": 84969
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.9050894975662231,
      "learning_rate": 0.00042019131772847134,
      "loss": 2.2246,
      "step": 84970
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.9401998519897461,
      "learning_rate": 0.00042018756979473297,
      "loss": 2.3488,
      "step": 84971
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.7617170810699463,
      "learning_rate": 0.0004201838218386496,
      "loss": 2.1866,
      "step": 84972
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.8903119564056396,
      "learning_rate": 0.00042018007386022174,
      "loss": 2.0802,
      "step": 84973
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.862872838973999,
      "learning_rate": 0.00042017632585945007,
      "loss": 2.324,
      "step": 84974
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.9208704829216003,
      "learning_rate": 0.0004201725778363354,
      "loss": 2.1712,
      "step": 84975
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.899152934551239,
      "learning_rate": 0.0004201688297908784,
      "loss": 2.2759,
      "step": 84976
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.8431330323219299,
      "learning_rate": 0.0004201650817230797,
      "loss": 2.1112,
      "step": 84977
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.770795464515686,
      "learning_rate": 0.00042016133363294007,
      "loss": 2.1351,
      "step": 84978
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.873658299446106,
      "learning_rate": 0.00042015758552046024,
      "loss": 2.1496,
      "step": 84979
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.7835065126419067,
      "learning_rate": 0.0004201538373856407,
      "loss": 2.2203,
      "step": 84980
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.8411352634429932,
      "learning_rate": 0.0004201500892284824,
      "loss": 2.3917,
      "step": 84981
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.8087694644927979,
      "learning_rate": 0.00042014634104898597,
      "loss": 2.4063,
      "step": 84982
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.8054165840148926,
      "learning_rate": 0.000420142592847152,
      "loss": 2.2292,
      "step": 84983
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.8184084892272949,
      "learning_rate": 0.0004201388446229812,
      "loss": 2.0141,
      "step": 84984
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.8167998790740967,
      "learning_rate": 0.00042013509637647443,
      "loss": 2.0911,
      "step": 84985
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.8363161087036133,
      "learning_rate": 0.0004201313481076322,
      "loss": 2.3456,
      "step": 84986
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.9406519532203674,
      "learning_rate": 0.00042012759981645525,
      "loss": 2.1531,
      "step": 84987
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.8589222431182861,
      "learning_rate": 0.00042012385150294444,
      "loss": 2.1669,
      "step": 84988
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.8394202589988708,
      "learning_rate": 0.00042012010316710025,
      "loss": 2.1764,
      "step": 84989
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.8798428177833557,
      "learning_rate": 0.00042011635480892345,
      "loss": 2.3527,
      "step": 84990
    },
    {
      "epoch": 1.11,
      "grad_norm": 1.032791256904602,
      "learning_rate": 0.0004201126064284148,
      "loss": 2.232,
      "step": 84991
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.9281342625617981,
      "learning_rate": 0.00042010885802557494,
      "loss": 2.211,
      "step": 84992
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.7880417704582214,
      "learning_rate": 0.0004201051096004046,
      "loss": 2.1435,
      "step": 84993
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.9603952169418335,
      "learning_rate": 0.00042010136115290444,
      "loss": 2.1772,
      "step": 84994
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.9561823606491089,
      "learning_rate": 0.0004200976126830752,
      "loss": 2.046,
      "step": 84995
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.7489838600158691,
      "learning_rate": 0.00042009386419091747,
      "loss": 2.1715,
      "step": 84996
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.8709090948104858,
      "learning_rate": 0.000420090115676432,
      "loss": 2.2065,
      "step": 84997
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.8166689276695251,
      "learning_rate": 0.0004200863671396196,
      "loss": 2.1989,
      "step": 84998
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.7907410860061646,
      "learning_rate": 0.0004200826185804809,
      "loss": 2.2295,
      "step": 84999
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.7861245274543762,
      "learning_rate": 0.00042007886999901646,
      "loss": 2.2332,
      "step": 85000
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.8235434293746948,
      "learning_rate": 0.00042007512139522716,
      "loss": 2.1321,
      "step": 85001
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.6847262382507324,
      "learning_rate": 0.0004200713727691136,
      "loss": 2.2319,
      "step": 85002
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.889656126499176,
      "learning_rate": 0.00042006762412067657,
      "loss": 2.362,
      "step": 85003
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.8797309398651123,
      "learning_rate": 0.00042006387544991664,
      "loss": 2.2905,
      "step": 85004
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.7600874304771423,
      "learning_rate": 0.00042006012675683456,
      "loss": 2.0692,
      "step": 85005
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.8047547340393066,
      "learning_rate": 0.00042005637804143114,
      "loss": 2.3618,
      "step": 85006
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.919011116027832,
      "learning_rate": 0.0004200526293037069,
      "loss": 2.1747,
      "step": 85007
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.790314257144928,
      "learning_rate": 0.0004200488805436625,
      "loss": 2.1962,
      "step": 85008
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.8470141887664795,
      "learning_rate": 0.0004200451317612989,
      "loss": 2.1602,
      "step": 85009
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.9006487727165222,
      "learning_rate": 0.0004200413829566166,
      "loss": 2.1812,
      "step": 85010
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.7180798053741455,
      "learning_rate": 0.00042003763412961633,
      "loss": 2.2001,
      "step": 85011
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.8195027112960815,
      "learning_rate": 0.0004200338852802988,
      "loss": 2.4391,
      "step": 85012
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.87326580286026,
      "learning_rate": 0.00042003013640866475,
      "loss": 2.2133,
      "step": 85013
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.8350606560707092,
      "learning_rate": 0.00042002638751471475,
      "loss": 1.9675,
      "step": 85014
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.8469626903533936,
      "learning_rate": 0.00042002263859844966,
      "loss": 2.216,
      "step": 85015
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.7574301958084106,
      "learning_rate": 0.0004200188896598701,
      "loss": 2.0471,
      "step": 85016
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.8259179592132568,
      "learning_rate": 0.00042001514069897665,
      "loss": 2.2549,
      "step": 85017
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.7287168502807617,
      "learning_rate": 0.00042001139171577025,
      "loss": 2.2556,
      "step": 85018
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.7718141078948975,
      "learning_rate": 0.0004200076427102514,
      "loss": 2.2244,
      "step": 85019
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.9647694230079651,
      "learning_rate": 0.00042000389368242083,
      "loss": 2.0449,
      "step": 85020
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.7515948414802551,
      "learning_rate": 0.00042000014463227937,
      "loss": 1.898,
      "step": 85021
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.8488521575927734,
      "learning_rate": 0.0004199963955598275,
      "loss": 2.1684,
      "step": 85022
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.8006938099861145,
      "learning_rate": 0.00041999264646506616,
      "loss": 2.0189,
      "step": 85023
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.8895719647407532,
      "learning_rate": 0.0004199888973479959,
      "loss": 1.9687,
      "step": 85024
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.7668548226356506,
      "learning_rate": 0.00041998514820861736,
      "loss": 2.0773,
      "step": 85025
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.776470422744751,
      "learning_rate": 0.0004199813990469313,
      "loss": 2.1543,
      "step": 85026
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.7807663679122925,
      "learning_rate": 0.00041997764986293856,
      "loss": 2.3422,
      "step": 85027
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.8205394148826599,
      "learning_rate": 0.0004199739006566396,
      "loss": 2.1411,
      "step": 85028
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.7441952228546143,
      "learning_rate": 0.00041997015142803524,
      "loss": 2.1204,
      "step": 85029
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.9918986558914185,
      "learning_rate": 0.00041996640217712626,
      "loss": 2.1252,
      "step": 85030
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.7794566750526428,
      "learning_rate": 0.00041996265290391313,
      "loss": 2.3861,
      "step": 85031
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.8298813104629517,
      "learning_rate": 0.0004199589036083967,
      "loss": 2.1881,
      "step": 85032
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.7442427277565002,
      "learning_rate": 0.0004199551542905778,
      "loss": 2.3042,
      "step": 85033
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.8370258808135986,
      "learning_rate": 0.0004199514049504568,
      "loss": 2.3091,
      "step": 85034
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.7907612919807434,
      "learning_rate": 0.00041994765558803465,
      "loss": 2.1791,
      "step": 85035
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.8072589635848999,
      "learning_rate": 0.000419943906203312,
      "loss": 2.2635,
      "step": 85036
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.793314516544342,
      "learning_rate": 0.0004199401567962894,
      "loss": 2.3234,
      "step": 85037
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.8133758306503296,
      "learning_rate": 0.00041993640736696767,
      "loss": 2.0467,
      "step": 85038
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.7716866135597229,
      "learning_rate": 0.00041993265791534756,
      "loss": 2.0705,
      "step": 85039
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.8124152421951294,
      "learning_rate": 0.0004199289084414297,
      "loss": 2.1788,
      "step": 85040
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.9430531859397888,
      "learning_rate": 0.00041992515894521477,
      "loss": 2.0815,
      "step": 85041
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.8366729021072388,
      "learning_rate": 0.0004199214094267035,
      "loss": 2.2647,
      "step": 85042
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.8336071968078613,
      "learning_rate": 0.0004199176598858966,
      "loss": 2.0626,
      "step": 85043
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.7282096147537231,
      "learning_rate": 0.0004199139103227947,
      "loss": 2.2587,
      "step": 85044
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.8382914662361145,
      "learning_rate": 0.00041991016073739855,
      "loss": 2.33,
      "step": 85045
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.8813959956169128,
      "learning_rate": 0.00041990641112970884,
      "loss": 2.3128,
      "step": 85046
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.8583210706710815,
      "learning_rate": 0.0004199026614997263,
      "loss": 2.3264,
      "step": 85047
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.879755437374115,
      "learning_rate": 0.00041989891184745153,
      "loss": 2.4106,
      "step": 85048
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.7206326127052307,
      "learning_rate": 0.00041989516217288535,
      "loss": 2.2806,
      "step": 85049
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.8016628623008728,
      "learning_rate": 0.00041989141247602836,
      "loss": 2.2605,
      "step": 85050
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.7949392199516296,
      "learning_rate": 0.0004198876627568813,
      "loss": 2.1456,
      "step": 85051
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.766424834728241,
      "learning_rate": 0.00041988391301544495,
      "loss": 2.2474,
      "step": 85052
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.8072309494018555,
      "learning_rate": 0.0004198801632517198,
      "loss": 2.1602,
      "step": 85053
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.7762875556945801,
      "learning_rate": 0.0004198764134657067,
      "loss": 1.9072,
      "step": 85054
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.923898458480835,
      "learning_rate": 0.0004198726636574063,
      "loss": 2.3329,
      "step": 85055
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.9078971147537231,
      "learning_rate": 0.0004198689138268193,
      "loss": 2.1649,
      "step": 85056
    },
    {
      "epoch": 1.11,
      "grad_norm": 1.230783224105835,
      "learning_rate": 0.0004198651639739464,
      "loss": 2.2894,
      "step": 85057
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.8516930937767029,
      "learning_rate": 0.00041986141409878835,
      "loss": 2.2489,
      "step": 85058
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.7630598545074463,
      "learning_rate": 0.00041985766420134577,
      "loss": 2.4953,
      "step": 85059
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.8271868228912354,
      "learning_rate": 0.00041985391428161936,
      "loss": 2.3278,
      "step": 85060
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.8393093943595886,
      "learning_rate": 0.00041985016433960994,
      "loss": 1.9695,
      "step": 85061
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.8231688141822815,
      "learning_rate": 0.00041984641437531805,
      "loss": 2.366,
      "step": 85062
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.7748714089393616,
      "learning_rate": 0.0004198426643887444,
      "loss": 2.093,
      "step": 85063
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.875129222869873,
      "learning_rate": 0.00041983891437988985,
      "loss": 2.0986,
      "step": 85064
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.7613081932067871,
      "learning_rate": 0.00041983516434875495,
      "loss": 2.181,
      "step": 85065
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.8515143990516663,
      "learning_rate": 0.00041983141429534034,
      "loss": 2.1665,
      "step": 85066
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.8066927790641785,
      "learning_rate": 0.0004198276642196469,
      "loss": 2.2081,
      "step": 85067
    },
    {
      "epoch": 1.11,
      "grad_norm": 1.176831603050232,
      "learning_rate": 0.0004198239141216752,
      "loss": 2.2609,
      "step": 85068
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.7943501472473145,
      "learning_rate": 0.000419820164001426,
      "loss": 2.1779,
      "step": 85069
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.7880948781967163,
      "learning_rate": 0.00041981641385889994,
      "loss": 1.8887,
      "step": 85070
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.7926640510559082,
      "learning_rate": 0.0004198126636940977,
      "loss": 1.9767,
      "step": 85071
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.7588486671447754,
      "learning_rate": 0.0004198089135070201,
      "loss": 2.0655,
      "step": 85072
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.7996913194656372,
      "learning_rate": 0.0004198051632976678,
      "loss": 2.2429,
      "step": 85073
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.8121513724327087,
      "learning_rate": 0.00041980141306604136,
      "loss": 2.1366,
      "step": 85074
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.7877061367034912,
      "learning_rate": 0.0004197976628121416,
      "loss": 2.0008,
      "step": 85075
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.8992189168930054,
      "learning_rate": 0.0004197939125359692,
      "loss": 2.3408,
      "step": 85076
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.7935132384300232,
      "learning_rate": 0.0004197901622375249,
      "loss": 2.2691,
      "step": 85077
    },
    {
      "epoch": 1.11,
      "grad_norm": 1.0099557638168335,
      "learning_rate": 0.0004197864119168093,
      "loss": 2.2272,
      "step": 85078
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.8430466055870056,
      "learning_rate": 0.0004197826615738231,
      "loss": 2.134,
      "step": 85079
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.778632402420044,
      "learning_rate": 0.00041977891120856714,
      "loss": 2.0651,
      "step": 85080
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.800963282585144,
      "learning_rate": 0.0004197751608210419,
      "loss": 2.171,
      "step": 85081
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.771691083908081,
      "learning_rate": 0.0004197714104112484,
      "loss": 2.1933,
      "step": 85082
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.8459041714668274,
      "learning_rate": 0.0004197676599791869,
      "loss": 1.9871,
      "step": 85083
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.7202295660972595,
      "learning_rate": 0.0004197639095248585,
      "loss": 2.0528,
      "step": 85084
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.8529558777809143,
      "learning_rate": 0.0004197601590482636,
      "loss": 2.2296,
      "step": 85085
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.8121931552886963,
      "learning_rate": 0.0004197564085494031,
      "loss": 2.3851,
      "step": 85086
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.7769201397895813,
      "learning_rate": 0.00041975265802827766,
      "loss": 2.2328,
      "step": 85087
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.7990137338638306,
      "learning_rate": 0.00041974890748488786,
      "loss": 2.2582,
      "step": 85088
    },
    {
      "epoch": 1.11,
      "grad_norm": 1.0224915742874146,
      "learning_rate": 0.00041974515691923446,
      "loss": 2.1976,
      "step": 85089
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.8368625044822693,
      "learning_rate": 0.0004197414063313182,
      "loss": 2.041,
      "step": 85090
    },
    {
      "epoch": 1.11,
      "grad_norm": 1.0068176984786987,
      "learning_rate": 0.00041973765572113987,
      "loss": 2.0561,
      "step": 85091
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.7967127561569214,
      "learning_rate": 0.0004197339050886999,
      "loss": 2.2937,
      "step": 85092
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.815822422504425,
      "learning_rate": 0.0004197301544339992,
      "loss": 2.1709,
      "step": 85093
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.7245954871177673,
      "learning_rate": 0.0004197264037570385,
      "loss": 2.179,
      "step": 85094
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.8693028092384338,
      "learning_rate": 0.0004197226530578182,
      "loss": 2.3598,
      "step": 85095
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.8378757834434509,
      "learning_rate": 0.00041971890233633935,
      "loss": 2.0906,
      "step": 85096
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.8658955693244934,
      "learning_rate": 0.0004197151515926024,
      "loss": 2.2918,
      "step": 85097
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.9194886088371277,
      "learning_rate": 0.0004197114008266082,
      "loss": 2.0845,
      "step": 85098
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.9244770407676697,
      "learning_rate": 0.00041970765003835735,
      "loss": 2.474,
      "step": 85099
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.7685976624488831,
      "learning_rate": 0.0004197038992278507,
      "loss": 2.224,
      "step": 85100
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.8317330479621887,
      "learning_rate": 0.0004197001483950887,
      "loss": 2.4452,
      "step": 85101
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.812792956829071,
      "learning_rate": 0.0004196963975400722,
      "loss": 2.0898,
      "step": 85102
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.7922821640968323,
      "learning_rate": 0.000419692646662802,
      "loss": 2.1081,
      "step": 85103
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.8152615427970886,
      "learning_rate": 0.0004196888957632786,
      "loss": 2.1048,
      "step": 85104
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.7686147093772888,
      "learning_rate": 0.0004196851448415027,
      "loss": 2.1804,
      "step": 85105
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.9539121985435486,
      "learning_rate": 0.0004196813938974752,
      "loss": 2.3416,
      "step": 85106
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.8165491819381714,
      "learning_rate": 0.0004196776429311966,
      "loss": 2.1057,
      "step": 85107
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.8302640914916992,
      "learning_rate": 0.00041967389194266764,
      "loss": 2.2312,
      "step": 85108
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.9543730616569519,
      "learning_rate": 0.00041967014093188907,
      "loss": 1.9835,
      "step": 85109
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.8566051721572876,
      "learning_rate": 0.0004196663898988616,
      "loss": 2.0944,
      "step": 85110
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.7638260126113892,
      "learning_rate": 0.0004196626388435859,
      "loss": 2.2878,
      "step": 85111
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.7911558747291565,
      "learning_rate": 0.00041965888776606254,
      "loss": 2.3093,
      "step": 85112
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.931432843208313,
      "learning_rate": 0.0004196551366662925,
      "loss": 2.0204,
      "step": 85113
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.8573980927467346,
      "learning_rate": 0.00041965138554427614,
      "loss": 2.1277,
      "step": 85114
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.7552816271781921,
      "learning_rate": 0.0004196476344000144,
      "loss": 2.1109,
      "step": 85115
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.8555973172187805,
      "learning_rate": 0.000419643883233508,
      "loss": 2.312,
      "step": 85116
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.793969988822937,
      "learning_rate": 0.00041964013204475744,
      "loss": 2.0454,
      "step": 85117
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.8706700801849365,
      "learning_rate": 0.00041963638083376354,
      "loss": 2.1858,
      "step": 85118
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.8099779486656189,
      "learning_rate": 0.000419632629600527,
      "loss": 2.3102,
      "step": 85119
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.7784695029258728,
      "learning_rate": 0.00041962887834504846,
      "loss": 2.2957,
      "step": 85120
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.7909006476402283,
      "learning_rate": 0.0004196251270673287,
      "loss": 2.4232,
      "step": 85121
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.9222921133041382,
      "learning_rate": 0.00041962137576736835,
      "loss": 1.8973,
      "step": 85122
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.9052188992500305,
      "learning_rate": 0.00041961762444516805,
      "loss": 2.2971,
      "step": 85123
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.8433716893196106,
      "learning_rate": 0.0004196138731007287,
      "loss": 2.2119,
      "step": 85124
    },
    {
      "epoch": 1.11,
      "grad_norm": 1.116113305091858,
      "learning_rate": 0.0004196101217340509,
      "loss": 2.4945,
      "step": 85125
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.8864260315895081,
      "learning_rate": 0.0004196063703451352,
      "loss": 2.2919,
      "step": 85126
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.7759316563606262,
      "learning_rate": 0.0004196026189339824,
      "loss": 2.2538,
      "step": 85127
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.8026745915412903,
      "learning_rate": 0.0004195988675005933,
      "loss": 2.115,
      "step": 85128
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.9752845168113708,
      "learning_rate": 0.00041959511604496854,
      "loss": 1.9724,
      "step": 85129
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.8731217384338379,
      "learning_rate": 0.0004195913645671087,
      "loss": 2.2678,
      "step": 85130
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.8785759210586548,
      "learning_rate": 0.00041958761306701465,
      "loss": 2.2533,
      "step": 85131
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.8570852875709534,
      "learning_rate": 0.00041958386154468695,
      "loss": 2.1677,
      "step": 85132
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.8144147992134094,
      "learning_rate": 0.0004195801100001263,
      "loss": 2.2346,
      "step": 85133
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.7927656769752502,
      "learning_rate": 0.0004195763584333336,
      "loss": 2.2131,
      "step": 85134
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.7635661363601685,
      "learning_rate": 0.0004195726068443093,
      "loss": 2.1585,
      "step": 85135
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.9250140190124512,
      "learning_rate": 0.0004195688552330541,
      "loss": 2.1287,
      "step": 85136
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.85204017162323,
      "learning_rate": 0.000419565103599569,
      "loss": 2.1125,
      "step": 85137
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.7860868573188782,
      "learning_rate": 0.00041956135194385437,
      "loss": 2.1633,
      "step": 85138
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.9852305054664612,
      "learning_rate": 0.000419557600265911,
      "loss": 1.9252,
      "step": 85139
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.8491476774215698,
      "learning_rate": 0.00041955384856573973,
      "loss": 1.9017,
      "step": 85140
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.7818125486373901,
      "learning_rate": 0.00041955009684334105,
      "loss": 2.0875,
      "step": 85141
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.9663756489753723,
      "learning_rate": 0.00041954634509871574,
      "loss": 2.0919,
      "step": 85142
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.991165280342102,
      "learning_rate": 0.00041954259333186456,
      "loss": 2.3726,
      "step": 85143
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.7922221422195435,
      "learning_rate": 0.00041953884154278815,
      "loss": 2.2682,
      "step": 85144
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.7886257171630859,
      "learning_rate": 0.0004195350897314872,
      "loss": 1.9119,
      "step": 85145
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.9015475511550903,
      "learning_rate": 0.0004195313378979625,
      "loss": 2.0472,
      "step": 85146
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.8040358424186707,
      "learning_rate": 0.00041952758604221454,
      "loss": 2.2137,
      "step": 85147
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.9285292625427246,
      "learning_rate": 0.0004195238341642442,
      "loss": 1.9722,
      "step": 85148
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.8414136171340942,
      "learning_rate": 0.0004195200822640522,
      "loss": 2.1582,
      "step": 85149
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.8058829307556152,
      "learning_rate": 0.000419516330341639,
      "loss": 2.3835,
      "step": 85150
    },
    {
      "epoch": 1.11,
      "grad_norm": 1.3190534114837646,
      "learning_rate": 0.0004195125783970055,
      "loss": 2.2203,
      "step": 85151
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.9595047235488892,
      "learning_rate": 0.00041950882643015247,
      "loss": 2.3733,
      "step": 85152
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.8871780037879944,
      "learning_rate": 0.0004195050744410804,
      "loss": 2.2296,
      "step": 85153
    },
    {
      "epoch": 1.11,
      "grad_norm": 1.1038786172866821,
      "learning_rate": 0.0004195013224297901,
      "loss": 2.2219,
      "step": 85154
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.9092011451721191,
      "learning_rate": 0.00041949757039628225,
      "loss": 2.1023,
      "step": 85155
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.7727529406547546,
      "learning_rate": 0.00041949381834055754,
      "loss": 2.0264,
      "step": 85156
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.9660047292709351,
      "learning_rate": 0.00041949006626261666,
      "loss": 2.4424,
      "step": 85157
    },
    {
      "epoch": 1.11,
      "grad_norm": 1.0523452758789062,
      "learning_rate": 0.00041948631416246037,
      "loss": 2.2978,
      "step": 85158
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.873187780380249,
      "learning_rate": 0.00041948256204008927,
      "loss": 2.1222,
      "step": 85159
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.8576874732971191,
      "learning_rate": 0.0004194788098955041,
      "loss": 2.0671,
      "step": 85160
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.869435727596283,
      "learning_rate": 0.00041947505772870565,
      "loss": 2.1057,
      "step": 85161
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.8924543857574463,
      "learning_rate": 0.00041947130553969444,
      "loss": 2.2822,
      "step": 85162
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.8050252199172974,
      "learning_rate": 0.00041946755332847123,
      "loss": 2.2034,
      "step": 85163
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.7863430380821228,
      "learning_rate": 0.0004194638010950368,
      "loss": 2.1508,
      "step": 85164
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.9821972846984863,
      "learning_rate": 0.00041946004883939187,
      "loss": 2.1668,
      "step": 85165
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.963736891746521,
      "learning_rate": 0.00041945629656153696,
      "loss": 2.0637,
      "step": 85166
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.8131264448165894,
      "learning_rate": 0.0004194525442614729,
      "loss": 2.0777,
      "step": 85167
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.8290901184082031,
      "learning_rate": 0.00041944879193920036,
      "loss": 2.0881,
      "step": 85168
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.9634493589401245,
      "learning_rate": 0.00041944503959472,
      "loss": 2.22,
      "step": 85169
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.8554213047027588,
      "learning_rate": 0.0004194412872280326,
      "loss": 2.1464,
      "step": 85170
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.8056537508964539,
      "learning_rate": 0.00041943753483913873,
      "loss": 2.3295,
      "step": 85171
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.8422428965568542,
      "learning_rate": 0.00041943378242803933,
      "loss": 2.2313,
      "step": 85172
    },
    {
      "epoch": 1.11,
      "grad_norm": 1.0737214088439941,
      "learning_rate": 0.00041943002999473476,
      "loss": 2.1699,
      "step": 85173
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.811657190322876,
      "learning_rate": 0.00041942627753922595,
      "loss": 2.2659,
      "step": 85174
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.7967514395713806,
      "learning_rate": 0.0004194225250615136,
      "loss": 2.3424,
      "step": 85175
    },
    {
      "epoch": 1.11,
      "grad_norm": 1.012351155281067,
      "learning_rate": 0.00041941877256159826,
      "loss": 2.241,
      "step": 85176
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.9393290281295776,
      "learning_rate": 0.00041941502003948084,
      "loss": 1.961,
      "step": 85177
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.7981711626052856,
      "learning_rate": 0.0004194112674951618,
      "loss": 2.2402,
      "step": 85178
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.9139937162399292,
      "learning_rate": 0.00041940751492864205,
      "loss": 2.3379,
      "step": 85179
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.8471601605415344,
      "learning_rate": 0.0004194037623399221,
      "loss": 1.9905,
      "step": 85180
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.8256453275680542,
      "learning_rate": 0.00041940000972900276,
      "loss": 2.2474,
      "step": 85181
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.8196026086807251,
      "learning_rate": 0.0004193962570958846,
      "loss": 2.1932,
      "step": 85182
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.8240729570388794,
      "learning_rate": 0.0004193925044405686,
      "loss": 2.2617,
      "step": 85183
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.9144033789634705,
      "learning_rate": 0.0004193887517630552,
      "loss": 2.2617,
      "step": 85184
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.7846344113349915,
      "learning_rate": 0.00041938499906334515,
      "loss": 2.1038,
      "step": 85185
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.7428039312362671,
      "learning_rate": 0.00041938124634143925,
      "loss": 2.0645,
      "step": 85186
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.8554707765579224,
      "learning_rate": 0.0004193774935973381,
      "loss": 2.0661,
      "step": 85187
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.7669625282287598,
      "learning_rate": 0.0004193737408310424,
      "loss": 2.4695,
      "step": 85188
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.8046802282333374,
      "learning_rate": 0.0004193699880425529,
      "loss": 2.2206,
      "step": 85189
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.8559361696243286,
      "learning_rate": 0.00041936623523187023,
      "loss": 2.2379,
      "step": 85190
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.8855218291282654,
      "learning_rate": 0.0004193624823989951,
      "loss": 1.8408,
      "step": 85191
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.7802923321723938,
      "learning_rate": 0.00041935872954392835,
      "loss": 2.2655,
      "step": 85192
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.776549220085144,
      "learning_rate": 0.00041935497666667043,
      "loss": 2.1399,
      "step": 85193
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.825025200843811,
      "learning_rate": 0.0004193512237672222,
      "loss": 2.2274,
      "step": 85194
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.8346222043037415,
      "learning_rate": 0.0004193474708455844,
      "loss": 2.0167,
      "step": 85195
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.8239488005638123,
      "learning_rate": 0.00041934371790175754,
      "loss": 2.3015,
      "step": 85196
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.7780623435974121,
      "learning_rate": 0.0004193399649357425,
      "loss": 2.0714,
      "step": 85197
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.7264289259910583,
      "learning_rate": 0.0004193362119475399,
      "loss": 1.9182,
      "step": 85198
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.8241272568702698,
      "learning_rate": 0.00041933245893715045,
      "loss": 2.2541,
      "step": 85199
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.8189530968666077,
      "learning_rate": 0.00041932870590457476,
      "loss": 2.2996,
      "step": 85200
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.802198588848114,
      "learning_rate": 0.00041932495284981375,
      "loss": 2.1086,
      "step": 85201
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.8530809879302979,
      "learning_rate": 0.00041932119977286785,
      "loss": 2.2014,
      "step": 85202
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.8843404054641724,
      "learning_rate": 0.0004193174466737379,
      "loss": 2.1267,
      "step": 85203
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.8158595561981201,
      "learning_rate": 0.00041931369355242473,
      "loss": 2.315,
      "step": 85204
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.8403838276863098,
      "learning_rate": 0.0004193099404089287,
      "loss": 2.1564,
      "step": 85205
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.755136251449585,
      "learning_rate": 0.00041930618724325084,
      "loss": 2.185,
      "step": 85206
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.8204453587532043,
      "learning_rate": 0.00041930243405539165,
      "loss": 2.2752,
      "step": 85207
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.7961896061897278,
      "learning_rate": 0.0004192986808453519,
      "loss": 2.2748,
      "step": 85208
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.7248779535293579,
      "learning_rate": 0.00041929492761313226,
      "loss": 2.159,
      "step": 85209
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.8993608951568604,
      "learning_rate": 0.0004192911743587335,
      "loss": 2.0476,
      "step": 85210
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.8619869947433472,
      "learning_rate": 0.00041928742108215614,
      "loss": 2.3751,
      "step": 85211
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.8398141264915466,
      "learning_rate": 0.00041928366778340106,
      "loss": 2.0627,
      "step": 85212
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.9698242545127869,
      "learning_rate": 0.00041927991446246896,
      "loss": 2.1414,
      "step": 85213
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.844756007194519,
      "learning_rate": 0.00041927616111936037,
      "loss": 2.1435,
      "step": 85214
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.8494327068328857,
      "learning_rate": 0.00041927240775407616,
      "loss": 2.2465,
      "step": 85215
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.8403744697570801,
      "learning_rate": 0.000419268654366617,
      "loss": 2.1579,
      "step": 85216
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.8319776654243469,
      "learning_rate": 0.0004192649009569834,
      "loss": 2.0457,
      "step": 85217
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.739630937576294,
      "learning_rate": 0.00041926114752517625,
      "loss": 2.0806,
      "step": 85218
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.8020381927490234,
      "learning_rate": 0.0004192573940711963,
      "loss": 1.9387,
      "step": 85219
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.8956850171089172,
      "learning_rate": 0.00041925364059504404,
      "loss": 2.226,
      "step": 85220
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.7871184945106506,
      "learning_rate": 0.0004192498870967203,
      "loss": 1.8775,
      "step": 85221
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.7464621663093567,
      "learning_rate": 0.0004192461335762258,
      "loss": 2.132,
      "step": 85222
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.8036623001098633,
      "learning_rate": 0.0004192423800335611,
      "loss": 2.2901,
      "step": 85223
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.8876157999038696,
      "learning_rate": 0.00041923862646872705,
      "loss": 2.1716,
      "step": 85224
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.8726493716239929,
      "learning_rate": 0.00041923487288172436,
      "loss": 2.3864,
      "step": 85225
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.8213701248168945,
      "learning_rate": 0.00041923111927255353,
      "loss": 2.2067,
      "step": 85226
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.7178576588630676,
      "learning_rate": 0.00041922736564121543,
      "loss": 2.2098,
      "step": 85227
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.8075270056724548,
      "learning_rate": 0.0004192236119877108,
      "loss": 2.1746,
      "step": 85228
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.8417894244194031,
      "learning_rate": 0.0004192198583120401,
      "loss": 2.363,
      "step": 85229
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.7777760624885559,
      "learning_rate": 0.0004192161046142042,
      "loss": 2.0843,
      "step": 85230
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.8629752397537231,
      "learning_rate": 0.0004192123508942039,
      "loss": 2.3818,
      "step": 85231
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.8185417056083679,
      "learning_rate": 0.00041920859715203975,
      "loss": 2.1116,
      "step": 85232
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.7750756740570068,
      "learning_rate": 0.0004192048433877124,
      "loss": 2.3024,
      "step": 85233
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.7914690971374512,
      "learning_rate": 0.00041920108960122267,
      "loss": 2.2322,
      "step": 85234
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.7902302742004395,
      "learning_rate": 0.00041919733579257116,
      "loss": 1.9442,
      "step": 85235
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.8277695178985596,
      "learning_rate": 0.0004191935819617586,
      "loss": 2.2436,
      "step": 85236
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.8329079151153564,
      "learning_rate": 0.00041918982810878574,
      "loss": 2.2566,
      "step": 85237
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.8286166787147522,
      "learning_rate": 0.0004191860742336533,
      "loss": 2.1376,
      "step": 85238
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.8411938548088074,
      "learning_rate": 0.0004191823203363618,
      "loss": 2.2681,
      "step": 85239
    },
    {
      "epoch": 1.11,
      "grad_norm": 1.0215502977371216,
      "learning_rate": 0.0004191785664169122,
      "loss": 2.1265,
      "step": 85240
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.9426751136779785,
      "learning_rate": 0.0004191748124753049,
      "loss": 2.235,
      "step": 85241
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.8004818558692932,
      "learning_rate": 0.0004191710585115408,
      "loss": 2.132,
      "step": 85242
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.8490459322929382,
      "learning_rate": 0.0004191673045256206,
      "loss": 1.9845,
      "step": 85243
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.8096505403518677,
      "learning_rate": 0.0004191635505175449,
      "loss": 2.4003,
      "step": 85244
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.8666747212409973,
      "learning_rate": 0.00041915979648731453,
      "loss": 2.3643,
      "step": 85245
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.887792706489563,
      "learning_rate": 0.00041915604243493,
      "loss": 2.3099,
      "step": 85246
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.8321656584739685,
      "learning_rate": 0.0004191522883603922,
      "loss": 2.1805,
      "step": 85247
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.8791669011116028,
      "learning_rate": 0.00041914853426370163,
      "loss": 2.1272,
      "step": 85248
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.7862319946289062,
      "learning_rate": 0.0004191447801448592,
      "loss": 2.1343,
      "step": 85249
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.7703086137771606,
      "learning_rate": 0.0004191410260038655,
      "loss": 2.1594,
      "step": 85250
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.7890387177467346,
      "learning_rate": 0.00041913727184072114,
      "loss": 2.0124,
      "step": 85251
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.7882133722305298,
      "learning_rate": 0.00041913351765542696,
      "loss": 2.1682,
      "step": 85252
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.7952211499214172,
      "learning_rate": 0.0004191297634479837,
      "loss": 2.1949,
      "step": 85253
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.8100847005844116,
      "learning_rate": 0.00041912600921839186,
      "loss": 2.2082,
      "step": 85254
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.9203805327415466,
      "learning_rate": 0.0004191222549666523,
      "loss": 2.2871,
      "step": 85255
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.8408229947090149,
      "learning_rate": 0.0004191185006927657,
      "loss": 2.1039,
      "step": 85256
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.7786757349967957,
      "learning_rate": 0.0004191147463967326,
      "loss": 2.173,
      "step": 85257
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.853022575378418,
      "learning_rate": 0.0004191109920785539,
      "loss": 2.3084,
      "step": 85258
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.8779212832450867,
      "learning_rate": 0.0004191072377382302,
      "loss": 2.2184,
      "step": 85259
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.8167649507522583,
      "learning_rate": 0.00041910348337576226,
      "loss": 2.224,
      "step": 85260
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.7296329140663147,
      "learning_rate": 0.0004190997289911507,
      "loss": 2.1821,
      "step": 85261
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.8502725958824158,
      "learning_rate": 0.0004190959745843962,
      "loss": 2.1584,
      "step": 85262
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.774241030216217,
      "learning_rate": 0.00041909222015549955,
      "loss": 2.0905,
      "step": 85263
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.9452835917472839,
      "learning_rate": 0.0004190884657044614,
      "loss": 2.185,
      "step": 85264
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.8332834839820862,
      "learning_rate": 0.0004190847112312825,
      "loss": 2.152,
      "step": 85265
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.7688848972320557,
      "learning_rate": 0.0004190809567359634,
      "loss": 2.243,
      "step": 85266
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.8767044544219971,
      "learning_rate": 0.0004190772022185051,
      "loss": 1.9952,
      "step": 85267
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.8661067485809326,
      "learning_rate": 0.00041907344767890795,
      "loss": 2.0755,
      "step": 85268
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.817354142665863,
      "learning_rate": 0.0004190696931171728,
      "loss": 2.3232,
      "step": 85269
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.909479558467865,
      "learning_rate": 0.0004190659385333003,
      "loss": 2.2008,
      "step": 85270
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.8413347005844116,
      "learning_rate": 0.00041906218392729136,
      "loss": 2.1483,
      "step": 85271
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.8280975222587585,
      "learning_rate": 0.0004190584292991464,
      "loss": 2.2099,
      "step": 85272
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.8317326903343201,
      "learning_rate": 0.00041905467464886626,
      "loss": 2.2338,
      "step": 85273
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.8207705616950989,
      "learning_rate": 0.00041905091997645165,
      "loss": 2.0982,
      "step": 85274
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.8911966681480408,
      "learning_rate": 0.0004190471652819031,
      "loss": 2.2871,
      "step": 85275
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.8220389485359192,
      "learning_rate": 0.00041904341056522154,
      "loss": 2.0741,
      "step": 85276
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.7801272869110107,
      "learning_rate": 0.0004190396558264076,
      "loss": 2.1384,
      "step": 85277
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.8300949335098267,
      "learning_rate": 0.00041903590106546183,
      "loss": 2.3903,
      "step": 85278
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.7750497460365295,
      "learning_rate": 0.0004190321462823851,
      "loss": 2.2532,
      "step": 85279
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.8197293281555176,
      "learning_rate": 0.00041902839147717804,
      "loss": 2.3851,
      "step": 85280
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.8501294851303101,
      "learning_rate": 0.0004190246366498413,
      "loss": 2.3763,
      "step": 85281
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.818225085735321,
      "learning_rate": 0.0004190208818003757,
      "loss": 2.221,
      "step": 85282
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.8914546370506287,
      "learning_rate": 0.00041901712692878193,
      "loss": 2.1893,
      "step": 85283
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.842010498046875,
      "learning_rate": 0.0004190133720350605,
      "loss": 2.2731,
      "step": 85284
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.7541237473487854,
      "learning_rate": 0.00041900961711921226,
      "loss": 2.0376,
      "step": 85285
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.9048465490341187,
      "learning_rate": 0.00041900586218123793,
      "loss": 2.1652,
      "step": 85286
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.9039738774299622,
      "learning_rate": 0.0004190021072211382,
      "loss": 2.3683,
      "step": 85287
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.8822052478790283,
      "learning_rate": 0.00041899835223891367,
      "loss": 2.1302,
      "step": 85288
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.8186523914337158,
      "learning_rate": 0.00041899459723456514,
      "loss": 2.3491,
      "step": 85289
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.7667919397354126,
      "learning_rate": 0.0004189908422080932,
      "loss": 2.2871,
      "step": 85290
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.9498318433761597,
      "learning_rate": 0.00041898708715949864,
      "loss": 2.179,
      "step": 85291
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.7467770576477051,
      "learning_rate": 0.0004189833320887822,
      "loss": 2.1535,
      "step": 85292
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.7894250154495239,
      "learning_rate": 0.0004189795769959444,
      "loss": 1.9415,
      "step": 85293
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.9525585770606995,
      "learning_rate": 0.00041897582188098615,
      "loss": 2.1777,
      "step": 85294
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.737418532371521,
      "learning_rate": 0.000418972066743908,
      "loss": 2.3138,
      "step": 85295
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.7336820960044861,
      "learning_rate": 0.00041896831158471076,
      "loss": 2.1336,
      "step": 85296
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.9600473046302795,
      "learning_rate": 0.00041896455640339495,
      "loss": 2.3191,
      "step": 85297
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.8781276941299438,
      "learning_rate": 0.0004189608011999615,
      "loss": 2.013,
      "step": 85298
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.7963392734527588,
      "learning_rate": 0.00041895704597441093,
      "loss": 2.0383,
      "step": 85299
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.9626179933547974,
      "learning_rate": 0.000418953290726744,
      "loss": 2.4301,
      "step": 85300
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.8929916620254517,
      "learning_rate": 0.00041894953545696143,
      "loss": 2.1772,
      "step": 85301
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.832624614238739,
      "learning_rate": 0.00041894578016506384,
      "loss": 2.0006,
      "step": 85302
    },
    {
      "epoch": 1.11,
      "grad_norm": 1.0789543390274048,
      "learning_rate": 0.0004189420248510521,
      "loss": 2.3084,
      "step": 85303
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.908790647983551,
      "learning_rate": 0.0004189382695149267,
      "loss": 2.0091,
      "step": 85304
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.7751842737197876,
      "learning_rate": 0.00041893451415668845,
      "loss": 2.0535,
      "step": 85305
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.7764785885810852,
      "learning_rate": 0.000418930758776338,
      "loss": 1.9809,
      "step": 85306
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.8682713508605957,
      "learning_rate": 0.00041892700337387614,
      "loss": 2.1207,
      "step": 85307
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.9012007117271423,
      "learning_rate": 0.0004189232479493034,
      "loss": 2.2225,
      "step": 85308
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.840954065322876,
      "learning_rate": 0.00041891949250262074,
      "loss": 2.0135,
      "step": 85309
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.8963555097579956,
      "learning_rate": 0.0004189157370338285,
      "loss": 2.3856,
      "step": 85310
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.9612135291099548,
      "learning_rate": 0.0004189119815429278,
      "loss": 2.2679,
      "step": 85311
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.8644856214523315,
      "learning_rate": 0.000418908226029919,
      "loss": 2.3086,
      "step": 85312
    },
    {
      "epoch": 1.11,
      "grad_norm": 1.2572598457336426,
      "learning_rate": 0.00041890447049480295,
      "loss": 2.1595,
      "step": 85313
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.8825107216835022,
      "learning_rate": 0.00041890071493758027,
      "loss": 2.1421,
      "step": 85314
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.7957561612129211,
      "learning_rate": 0.00041889695935825173,
      "loss": 2.2444,
      "step": 85315
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.8301275372505188,
      "learning_rate": 0.00041889320375681794,
      "loss": 2.2853,
      "step": 85316
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.9773224592208862,
      "learning_rate": 0.0004188894481332798,
      "loss": 2.0975,
      "step": 85317
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.8611599802970886,
      "learning_rate": 0.00041888569248763773,
      "loss": 2.1953,
      "step": 85318
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.7427577972412109,
      "learning_rate": 0.0004188819368198927,
      "loss": 2.1065,
      "step": 85319
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.8374611139297485,
      "learning_rate": 0.0004188781811300451,
      "loss": 1.8892,
      "step": 85320
    },
    {
      "epoch": 1.11,
      "grad_norm": 1.0547480583190918,
      "learning_rate": 0.00041887442541809595,
      "loss": 2.3233,
      "step": 85321
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.836751401424408,
      "learning_rate": 0.00041887066968404575,
      "loss": 2.263,
      "step": 85322
    },
    {
      "epoch": 1.11,
      "grad_norm": 1.0200002193450928,
      "learning_rate": 0.0004188669139278953,
      "loss": 2.065,
      "step": 85323
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.7499369382858276,
      "learning_rate": 0.00041886315814964513,
      "loss": 2.1795,
      "step": 85324
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.8004269599914551,
      "learning_rate": 0.0004188594023492961,
      "loss": 2.3053,
      "step": 85325
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.8064867854118347,
      "learning_rate": 0.00041885564652684897,
      "loss": 2.2384,
      "step": 85326
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.8966580033302307,
      "learning_rate": 0.0004188518906823042,
      "loss": 2.275,
      "step": 85327
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.8586137890815735,
      "learning_rate": 0.00041884813481566263,
      "loss": 2.0,
      "step": 85328
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.8276800513267517,
      "learning_rate": 0.00041884437892692497,
      "loss": 2.1801,
      "step": 85329
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.8641799092292786,
      "learning_rate": 0.00041884062301609197,
      "loss": 2.1188,
      "step": 85330
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.8554044365882874,
      "learning_rate": 0.0004188368670831641,
      "loss": 2.0427,
      "step": 85331
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.9141408801078796,
      "learning_rate": 0.00041883311112814235,
      "loss": 2.2226,
      "step": 85332
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.9534728527069092,
      "learning_rate": 0.00041882935515102725,
      "loss": 2.2141,
      "step": 85333
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.9162741303443909,
      "learning_rate": 0.0004188255991518195,
      "loss": 2.1469,
      "step": 85334
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.9945793747901917,
      "learning_rate": 0.0004188218431305199,
      "loss": 2.1705,
      "step": 85335
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.8963852524757385,
      "learning_rate": 0.000418818087087129,
      "loss": 1.9586,
      "step": 85336
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.8457198739051819,
      "learning_rate": 0.0004188143310216476,
      "loss": 2.1745,
      "step": 85337
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.98153156042099,
      "learning_rate": 0.0004188105749340764,
      "loss": 2.4418,
      "step": 85338
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.9831352829933167,
      "learning_rate": 0.00041880681882441596,
      "loss": 2.3785,
      "step": 85339
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.8448916673660278,
      "learning_rate": 0.00041880306269266723,
      "loss": 2.2606,
      "step": 85340
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.8695118427276611,
      "learning_rate": 0.0004187993065388307,
      "loss": 2.2743,
      "step": 85341
    },
    {
      "epoch": 1.11,
      "grad_norm": 1.0444024801254272,
      "learning_rate": 0.0004187955503629072,
      "loss": 2.3428,
      "step": 85342
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.8837022185325623,
      "learning_rate": 0.00041879179416489726,
      "loss": 2.2072,
      "step": 85343
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.8694204092025757,
      "learning_rate": 0.00041878803794480167,
      "loss": 1.9784,
      "step": 85344
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.7817251086235046,
      "learning_rate": 0.00041878428170262125,
      "loss": 2.163,
      "step": 85345
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.8343329429626465,
      "learning_rate": 0.00041878052543835656,
      "loss": 2.4602,
      "step": 85346
    },
    {
      "epoch": 1.11,
      "grad_norm": 1.0719020366668701,
      "learning_rate": 0.00041877676915200824,
      "loss": 2.3839,
      "step": 85347
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.8994320034980774,
      "learning_rate": 0.0004187730128435772,
      "loss": 2.1493,
      "step": 85348
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.7508947253227234,
      "learning_rate": 0.000418769256513064,
      "loss": 2.2205,
      "step": 85349
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.8957261443138123,
      "learning_rate": 0.0004187655001604693,
      "loss": 2.2186,
      "step": 85350
    },
    {
      "epoch": 1.11,
      "grad_norm": 1.0222454071044922,
      "learning_rate": 0.00041876174378579383,
      "loss": 2.1202,
      "step": 85351
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.8767658472061157,
      "learning_rate": 0.0004187579873890384,
      "loss": 2.3479,
      "step": 85352
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.9594323635101318,
      "learning_rate": 0.00041875423097020354,
      "loss": 2.1986,
      "step": 85353
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.8456136584281921,
      "learning_rate": 0.00041875047452929,
      "loss": 2.3273,
      "step": 85354
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.8128107786178589,
      "learning_rate": 0.00041874671806629854,
      "loss": 2.4517,
      "step": 85355
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.8431798219680786,
      "learning_rate": 0.00041874296158122995,
      "loss": 2.1255,
      "step": 85356
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.95396488904953,
      "learning_rate": 0.00041873920507408467,
      "loss": 2.2192,
      "step": 85357
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.8945883512496948,
      "learning_rate": 0.0004187354485448635,
      "loss": 2.2314,
      "step": 85358
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.8354357481002808,
      "learning_rate": 0.00041873169199356725,
      "loss": 2.0596,
      "step": 85359
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.8789061307907104,
      "learning_rate": 0.00041872793542019647,
      "loss": 2.2655,
      "step": 85360
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.9595025181770325,
      "learning_rate": 0.000418724178824752,
      "loss": 2.2508,
      "step": 85361
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.9275349974632263,
      "learning_rate": 0.00041872042220723443,
      "loss": 2.1302,
      "step": 85362
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.9859008193016052,
      "learning_rate": 0.00041871666556764453,
      "loss": 2.3349,
      "step": 85363
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.8762347102165222,
      "learning_rate": 0.0004187129089059829,
      "loss": 2.0881,
      "step": 85364
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.8321022987365723,
      "learning_rate": 0.00041870915222225034,
      "loss": 1.976,
      "step": 85365
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.8201960325241089,
      "learning_rate": 0.0004187053955164475,
      "loss": 2.0634,
      "step": 85366
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.7675942778587341,
      "learning_rate": 0.00041870163878857504,
      "loss": 1.993,
      "step": 85367
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.7567893266677856,
      "learning_rate": 0.0004186978820386338,
      "loss": 2.1876,
      "step": 85368
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.7774102091789246,
      "learning_rate": 0.00041869412526662433,
      "loss": 2.0808,
      "step": 85369
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.8107678890228271,
      "learning_rate": 0.0004186903684725473,
      "loss": 2.2478,
      "step": 85370
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.800277054309845,
      "learning_rate": 0.0004186866116564036,
      "loss": 2.0948,
      "step": 85371
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.8110798597335815,
      "learning_rate": 0.0004186828548181938,
      "loss": 2.1666,
      "step": 85372
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.8348940014839172,
      "learning_rate": 0.0004186790979579186,
      "loss": 2.152,
      "step": 85373
    },
    {
      "epoch": 1.11,
      "grad_norm": 1.0371644496917725,
      "learning_rate": 0.0004186753410755788,
      "loss": 2.3591,
      "step": 85374
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.8619211316108704,
      "learning_rate": 0.00041867158417117483,
      "loss": 2.1949,
      "step": 85375
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.8429746627807617,
      "learning_rate": 0.00041866782724470764,
      "loss": 2.0981,
      "step": 85376
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.8452118635177612,
      "learning_rate": 0.0004186640702961779,
      "loss": 2.096,
      "step": 85377
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.806240975856781,
      "learning_rate": 0.00041866031332558635,
      "loss": 2.2044,
      "step": 85378
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.9649558663368225,
      "learning_rate": 0.0004186565563329335,
      "loss": 2.3875,
      "step": 85379
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.811782717704773,
      "learning_rate": 0.00041865279931822016,
      "loss": 2.2444,
      "step": 85380
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.8100801110267639,
      "learning_rate": 0.0004186490422814471,
      "loss": 2.081,
      "step": 85381
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.8452161550521851,
      "learning_rate": 0.00041864528522261483,
      "loss": 2.248,
      "step": 85382
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.9264952540397644,
      "learning_rate": 0.0004186415281417242,
      "loss": 2.1234,
      "step": 85383
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.7521790266036987,
      "learning_rate": 0.000418637771038776,
      "loss": 2.3029,
      "step": 85384
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.8016849160194397,
      "learning_rate": 0.00041863401391377056,
      "loss": 2.0666,
      "step": 85385
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.8188118934631348,
      "learning_rate": 0.000418630256766709,
      "loss": 2.0522,
      "step": 85386
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.8599408268928528,
      "learning_rate": 0.0004186264995975918,
      "loss": 2.1066,
      "step": 85387
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.7851352691650391,
      "learning_rate": 0.00041862274240641966,
      "loss": 2.2332,
      "step": 85388
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.7573158144950867,
      "learning_rate": 0.0004186189851931933,
      "loss": 2.1514,
      "step": 85389
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.8337279558181763,
      "learning_rate": 0.00041861522795791356,
      "loss": 2.2969,
      "step": 85390
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.8284475207328796,
      "learning_rate": 0.00041861147070058083,
      "loss": 2.0492,
      "step": 85391
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.7946389317512512,
      "learning_rate": 0.0004186077134211961,
      "loss": 2.1722,
      "step": 85392
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.932083785533905,
      "learning_rate": 0.00041860395611975995,
      "loss": 2.1892,
      "step": 85393
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.7449852824211121,
      "learning_rate": 0.000418600198796273,
      "loss": 2.1101,
      "step": 85394
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.7871067523956299,
      "learning_rate": 0.0004185964414507361,
      "loss": 2.0912,
      "step": 85395
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.8757705092430115,
      "learning_rate": 0.0004185926840831499,
      "loss": 2.0745,
      "step": 85396
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.8474907875061035,
      "learning_rate": 0.00041858892669351507,
      "loss": 2.3688,
      "step": 85397
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.8320536613464355,
      "learning_rate": 0.0004185851692818323,
      "loss": 2.0622,
      "step": 85398
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.7635740637779236,
      "learning_rate": 0.0004185814118481024,
      "loss": 2.1891,
      "step": 85399
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.8355728983879089,
      "learning_rate": 0.00041857765439232586,
      "loss": 2.2046,
      "step": 85400
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.8401345014572144,
      "learning_rate": 0.0004185738969145035,
      "loss": 2.3229,
      "step": 85401
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.9618166089057922,
      "learning_rate": 0.0004185701394146361,
      "loss": 2.3218,
      "step": 85402
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.9008444547653198,
      "learning_rate": 0.0004185663818927242,
      "loss": 2.0,
      "step": 85403
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.8026748895645142,
      "learning_rate": 0.0004185626243487685,
      "loss": 2.4543,
      "step": 85404
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.7510485053062439,
      "learning_rate": 0.00041855886678276994,
      "loss": 2.2634,
      "step": 85405
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.8478298187255859,
      "learning_rate": 0.0004185551091947289,
      "loss": 2.1932,
      "step": 85406
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.7528988718986511,
      "learning_rate": 0.00041855135158464635,
      "loss": 2.3384,
      "step": 85407
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.8216106295585632,
      "learning_rate": 0.0004185475939525228,
      "loss": 2.0972,
      "step": 85408
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.8114449381828308,
      "learning_rate": 0.0004185438362983591,
      "loss": 1.9082,
      "step": 85409
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.7255854606628418,
      "learning_rate": 0.0004185400786221557,
      "loss": 2.1628,
      "step": 85410
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.7921852469444275,
      "learning_rate": 0.00041853632092391364,
      "loss": 2.2169,
      "step": 85411
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.7524616718292236,
      "learning_rate": 0.0004185325632036333,
      "loss": 2.11,
      "step": 85412
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.8204556703567505,
      "learning_rate": 0.0004185288054613155,
      "loss": 2.0236,
      "step": 85413
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.9322437644004822,
      "learning_rate": 0.0004185250476969611,
      "loss": 1.9815,
      "step": 85414
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.9110906720161438,
      "learning_rate": 0.0004185212899105706,
      "loss": 2.1425,
      "step": 85415
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.8162392973899841,
      "learning_rate": 0.0004185175321021447,
      "loss": 2.1644,
      "step": 85416
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.9015944004058838,
      "learning_rate": 0.0004185137742716843,
      "loss": 2.25,
      "step": 85417
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.8125282526016235,
      "learning_rate": 0.00041851001641918977,
      "loss": 2.166,
      "step": 85418
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.8007880449295044,
      "learning_rate": 0.0004185062585446621,
      "loss": 1.8386,
      "step": 85419
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.9364385604858398,
      "learning_rate": 0.00041850250064810187,
      "loss": 2.1937,
      "step": 85420
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.8994570970535278,
      "learning_rate": 0.00041849874272950973,
      "loss": 2.1964,
      "step": 85421
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.7797534465789795,
      "learning_rate": 0.00041849498478888645,
      "loss": 2.2423,
      "step": 85422
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.9147762656211853,
      "learning_rate": 0.00041849122682623277,
      "loss": 2.2952,
      "step": 85423
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.7778569459915161,
      "learning_rate": 0.00041848746884154934,
      "loss": 2.257,
      "step": 85424
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.8395735025405884,
      "learning_rate": 0.0004184837108348367,
      "loss": 2.2667,
      "step": 85425
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.8376319408416748,
      "learning_rate": 0.00041847995280609593,
      "loss": 2.2662,
      "step": 85426
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.9075111150741577,
      "learning_rate": 0.00041847619475532735,
      "loss": 2.1717,
      "step": 85427
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.7834432721138,
      "learning_rate": 0.0004184724366825318,
      "loss": 2.0245,
      "step": 85428
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.7932720184326172,
      "learning_rate": 0.00041846867858771013,
      "loss": 2.1357,
      "step": 85429
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.8415870666503906,
      "learning_rate": 0.00041846492047086274,
      "loss": 2.3556,
      "step": 85430
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.9279935359954834,
      "learning_rate": 0.0004184611623319906,
      "loss": 2.1699,
      "step": 85431
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.868629515171051,
      "learning_rate": 0.00041845740417109417,
      "loss": 2.1496,
      "step": 85432
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.7942410707473755,
      "learning_rate": 0.00041845364598817436,
      "loss": 2.2682,
      "step": 85433
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.7830676436424255,
      "learning_rate": 0.00041844988778323174,
      "loss": 2.0453,
      "step": 85434
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.8551202416419983,
      "learning_rate": 0.0004184461295562671,
      "loss": 2.5306,
      "step": 85435
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.8414165377616882,
      "learning_rate": 0.000418442371307281,
      "loss": 2.2032,
      "step": 85436
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.8360356092453003,
      "learning_rate": 0.0004184386130362744,
      "loss": 2.1911,
      "step": 85437
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.8480843901634216,
      "learning_rate": 0.00041843485474324767,
      "loss": 2.1032,
      "step": 85438
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.829937219619751,
      "learning_rate": 0.00041843109642820167,
      "loss": 2.2784,
      "step": 85439
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.9581397771835327,
      "learning_rate": 0.0004184273380911372,
      "loss": 1.9369,
      "step": 85440
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.8661550879478455,
      "learning_rate": 0.00041842357973205475,
      "loss": 2.2496,
      "step": 85441
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.7963901162147522,
      "learning_rate": 0.0004184198213509551,
      "loss": 2.0774,
      "step": 85442
    },
    {
      "epoch": 1.11,
      "grad_norm": 1.0848629474639893,
      "learning_rate": 0.000418416062947839,
      "loss": 2.2003,
      "step": 85443
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.8075013756752014,
      "learning_rate": 0.00041841230452270724,
      "loss": 2.1951,
      "step": 85444
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.8378454446792603,
      "learning_rate": 0.00041840854607556025,
      "loss": 2.0189,
      "step": 85445
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.8258378505706787,
      "learning_rate": 0.0004184047876063989,
      "loss": 2.2317,
      "step": 85446
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.7849140763282776,
      "learning_rate": 0.0004184010291152239,
      "loss": 2.2079,
      "step": 85447
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.8387120366096497,
      "learning_rate": 0.00041839727060203594,
      "loss": 2.2391,
      "step": 85448
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.8490180373191833,
      "learning_rate": 0.0004183935120668357,
      "loss": 2.3045,
      "step": 85449
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.7805072665214539,
      "learning_rate": 0.00041838975350962375,
      "loss": 2.4317,
      "step": 85450
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.8049222230911255,
      "learning_rate": 0.000418385994930401,
      "loss": 1.8895,
      "step": 85451
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.780963659286499,
      "learning_rate": 0.0004183822363291681,
      "loss": 2.1367,
      "step": 85452
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.9420599341392517,
      "learning_rate": 0.0004183784777059256,
      "loss": 2.0159,
      "step": 85453
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.774627685546875,
      "learning_rate": 0.0004183747190606744,
      "loss": 2.1584,
      "step": 85454
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.8214958906173706,
      "learning_rate": 0.00041837096039341507,
      "loss": 2.0866,
      "step": 85455
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.7910686731338501,
      "learning_rate": 0.00041836720170414834,
      "loss": 2.1205,
      "step": 85456
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.8817424178123474,
      "learning_rate": 0.0004183634429928749,
      "loss": 2.524,
      "step": 85457
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.8575522303581238,
      "learning_rate": 0.0004183596842595955,
      "loss": 2.1726,
      "step": 85458
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.8055132031440735,
      "learning_rate": 0.00041835592550431075,
      "loss": 2.1046,
      "step": 85459
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.8310118913650513,
      "learning_rate": 0.00041835216672702144,
      "loss": 2.3072,
      "step": 85460
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.8459472060203552,
      "learning_rate": 0.00041834840792772824,
      "loss": 2.2859,
      "step": 85461
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.8410091996192932,
      "learning_rate": 0.0004183446491064318,
      "loss": 2.207,
      "step": 85462
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.8259422183036804,
      "learning_rate": 0.0004183408902631329,
      "loss": 2.1399,
      "step": 85463
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.8442686796188354,
      "learning_rate": 0.00041833713139783214,
      "loss": 2.4298,
      "step": 85464
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.8677829504013062,
      "learning_rate": 0.0004183333725105303,
      "loss": 2.3682,
      "step": 85465
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.813749372959137,
      "learning_rate": 0.0004183296136012281,
      "loss": 2.1897,
      "step": 85466
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.9148549437522888,
      "learning_rate": 0.0004183258546699261,
      "loss": 2.363,
      "step": 85467
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.8666291236877441,
      "learning_rate": 0.0004183220957166251,
      "loss": 2.0519,
      "step": 85468
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.8335905075073242,
      "learning_rate": 0.00041831833674132584,
      "loss": 2.2147,
      "step": 85469
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.8776965141296387,
      "learning_rate": 0.000418314577744029,
      "loss": 2.1155,
      "step": 85470
    },
    {
      "epoch": 1.11,
      "grad_norm": 1.0182157754898071,
      "learning_rate": 0.0004183108187247351,
      "loss": 2.153,
      "step": 85471
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.7473816275596619,
      "learning_rate": 0.0004183070596834452,
      "loss": 2.3459,
      "step": 85472
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.748689591884613,
      "learning_rate": 0.0004183033006201596,
      "loss": 2.1872,
      "step": 85473
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.9511394500732422,
      "learning_rate": 0.0004182995415348792,
      "loss": 2.4593,
      "step": 85474
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.8550666570663452,
      "learning_rate": 0.0004182957824276048,
      "loss": 2.2582,
      "step": 85475
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.847370982170105,
      "learning_rate": 0.00041829202329833685,
      "loss": 2.5213,
      "step": 85476
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.8576341271400452,
      "learning_rate": 0.00041828826414707623,
      "loss": 2.2178,
      "step": 85477
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.9493613839149475,
      "learning_rate": 0.00041828450497382366,
      "loss": 2.2459,
      "step": 85478
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.8190502524375916,
      "learning_rate": 0.00041828074577857967,
      "loss": 2.2061,
      "step": 85479
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.7700543403625488,
      "learning_rate": 0.00041827698656134513,
      "loss": 1.9868,
      "step": 85480
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.8169849514961243,
      "learning_rate": 0.0004182732273221206,
      "loss": 2.2116,
      "step": 85481
    },
    {
      "epoch": 1.11,
      "grad_norm": 1.52104914188385,
      "learning_rate": 0.0004182694680609069,
      "loss": 2.092,
      "step": 85482
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.8698451519012451,
      "learning_rate": 0.0004182657087777046,
      "loss": 2.2124,
      "step": 85483
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.9438873529434204,
      "learning_rate": 0.0004182619494725146,
      "loss": 2.1811,
      "step": 85484
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.8278913497924805,
      "learning_rate": 0.0004182581901453373,
      "loss": 2.3569,
      "step": 85485
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.8893393278121948,
      "learning_rate": 0.0004182544307961737,
      "loss": 1.9103,
      "step": 85486
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.8363728523254395,
      "learning_rate": 0.00041825067142502436,
      "loss": 2.0952,
      "step": 85487
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.9052688479423523,
      "learning_rate": 0.0004182469120318899,
      "loss": 2.153,
      "step": 85488
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.8670458197593689,
      "learning_rate": 0.00041824315261677116,
      "loss": 2.206,
      "step": 85489
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.7975342869758606,
      "learning_rate": 0.00041823939317966883,
      "loss": 2.0276,
      "step": 85490
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.9261216521263123,
      "learning_rate": 0.00041823563372058353,
      "loss": 2.2823,
      "step": 85491
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.8186578750610352,
      "learning_rate": 0.000418231874239516,
      "loss": 2.2553,
      "step": 85492
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.8773143291473389,
      "learning_rate": 0.0004182281147364669,
      "loss": 2.3078,
      "step": 85493
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.8058599233627319,
      "learning_rate": 0.00041822435521143695,
      "loss": 2.1568,
      "step": 85494
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.7551969289779663,
      "learning_rate": 0.0004182205956644269,
      "loss": 2.2205,
      "step": 85495
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.8198279142379761,
      "learning_rate": 0.00041821683609543747,
      "loss": 2.517,
      "step": 85496
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.8266019225120544,
      "learning_rate": 0.0004182130765044692,
      "loss": 2.3248,
      "step": 85497
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.9115340113639832,
      "learning_rate": 0.00041820931689152296,
      "loss": 2.3714,
      "step": 85498
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.8093506097793579,
      "learning_rate": 0.0004182055572565993,
      "loss": 2.0832,
      "step": 85499
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.8688725233078003,
      "learning_rate": 0.000418201797599699,
      "loss": 2.2281,
      "step": 85500
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.8031324744224548,
      "learning_rate": 0.0004181980379208228,
      "loss": 2.0665,
      "step": 85501
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.7850606441497803,
      "learning_rate": 0.0004181942782199714,
      "loss": 1.9424,
      "step": 85502
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.7904559969902039,
      "learning_rate": 0.00041819051849714537,
      "loss": 2.1048,
      "step": 85503
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.8470747470855713,
      "learning_rate": 0.00041818675875234554,
      "loss": 2.3753,
      "step": 85504
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.8818855881690979,
      "learning_rate": 0.00041818299898557256,
      "loss": 1.989,
      "step": 85505
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.8922564387321472,
      "learning_rate": 0.0004181792391968271,
      "loss": 2.2258,
      "step": 85506
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.8709370493888855,
      "learning_rate": 0.00041817547938610987,
      "loss": 2.2673,
      "step": 85507
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.8682963252067566,
      "learning_rate": 0.00041817171955342157,
      "loss": 2.0514,
      "step": 85508
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.8837659358978271,
      "learning_rate": 0.000418167959698763,
      "loss": 2.1241,
      "step": 85509
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.7894763946533203,
      "learning_rate": 0.00041816419982213464,
      "loss": 2.1079,
      "step": 85510
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.8950421214103699,
      "learning_rate": 0.0004181604399235375,
      "loss": 2.209,
      "step": 85511
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.8799901604652405,
      "learning_rate": 0.00041815668000297205,
      "loss": 2.1773,
      "step": 85512
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.8121140003204346,
      "learning_rate": 0.000418152920060439,
      "loss": 2.1182,
      "step": 85513
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.8645693063735962,
      "learning_rate": 0.00041814916009593904,
      "loss": 2.3665,
      "step": 85514
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.7907187938690186,
      "learning_rate": 0.00041814540010947303,
      "loss": 2.1964,
      "step": 85515
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.8553368449211121,
      "learning_rate": 0.00041814164010104153,
      "loss": 2.2066,
      "step": 85516
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.7966146469116211,
      "learning_rate": 0.0004181378800706452,
      "loss": 2.1269,
      "step": 85517
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.8655533194541931,
      "learning_rate": 0.00041813412001828493,
      "loss": 1.9966,
      "step": 85518
    },
    {
      "epoch": 1.11,
      "grad_norm": 1.0580633878707886,
      "learning_rate": 0.00041813035994396126,
      "loss": 2.3156,
      "step": 85519
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.8761417269706726,
      "learning_rate": 0.00041812659984767484,
      "loss": 2.4394,
      "step": 85520
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.8986151218414307,
      "learning_rate": 0.00041812283972942656,
      "loss": 2.3015,
      "step": 85521
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.9372154474258423,
      "learning_rate": 0.00041811907958921694,
      "loss": 2.172,
      "step": 85522
    },
    {
      "epoch": 1.11,
      "grad_norm": 1.0072355270385742,
      "learning_rate": 0.0004181153194270468,
      "loss": 2.17,
      "step": 85523
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.9137654900550842,
      "learning_rate": 0.0004181115592429167,
      "loss": 2.1597,
      "step": 85524
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.8609007000923157,
      "learning_rate": 0.00041810779903682756,
      "loss": 2.1214,
      "step": 85525
    },
    {
      "epoch": 1.11,
      "grad_norm": 1.0509315729141235,
      "learning_rate": 0.00041810403880877985,
      "loss": 2.1141,
      "step": 85526
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.8895156979560852,
      "learning_rate": 0.0004181002785587744,
      "loss": 2.1365,
      "step": 85527
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.8139506578445435,
      "learning_rate": 0.00041809651828681187,
      "loss": 2.324,
      "step": 85528
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.8286749720573425,
      "learning_rate": 0.00041809275799289307,
      "loss": 1.831,
      "step": 85529
    },
    {
      "epoch": 1.11,
      "grad_norm": 1.0654950141906738,
      "learning_rate": 0.00041808899767701847,
      "loss": 2.2224,
      "step": 85530
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.8048982620239258,
      "learning_rate": 0.00041808523733918894,
      "loss": 2.225,
      "step": 85531
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.7817183136940002,
      "learning_rate": 0.0004180814769794052,
      "loss": 2.2708,
      "step": 85532
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.9048609733581543,
      "learning_rate": 0.0004180777165976678,
      "loss": 2.1688,
      "step": 85533
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.837907075881958,
      "learning_rate": 0.0004180739561939775,
      "loss": 2.2204,
      "step": 85534
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.8531898260116577,
      "learning_rate": 0.00041807019576833507,
      "loss": 2.2746,
      "step": 85535
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.9178576469421387,
      "learning_rate": 0.00041806643532074124,
      "loss": 2.2053,
      "step": 85536
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.7694432139396667,
      "learning_rate": 0.0004180626748511965,
      "loss": 2.2609,
      "step": 85537
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.8447384834289551,
      "learning_rate": 0.0004180589143597017,
      "loss": 2.1708,
      "step": 85538
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.8052316904067993,
      "learning_rate": 0.0004180551538462576,
      "loss": 2.2219,
      "step": 85539
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.8416894674301147,
      "learning_rate": 0.0004180513933108648,
      "loss": 2.3655,
      "step": 85540
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.8385451436042786,
      "learning_rate": 0.0004180476327535239,
      "loss": 2.2276,
      "step": 85541
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.8365650177001953,
      "learning_rate": 0.0004180438721742359,
      "loss": 2.0158,
      "step": 85542
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.8280514478683472,
      "learning_rate": 0.00041804011157300117,
      "loss": 2.2127,
      "step": 85543
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.8779438138008118,
      "learning_rate": 0.0004180363509498206,
      "loss": 2.3287,
      "step": 85544
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.936087429523468,
      "learning_rate": 0.00041803259030469487,
      "loss": 2.4736,
      "step": 85545
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.9378997683525085,
      "learning_rate": 0.0004180288296376247,
      "loss": 2.1077,
      "step": 85546
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.7415765523910522,
      "learning_rate": 0.0004180250689486106,
      "loss": 2.0893,
      "step": 85547
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.9081546068191528,
      "learning_rate": 0.0004180213082376536,
      "loss": 2.015,
      "step": 85548
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.8449047207832336,
      "learning_rate": 0.00041801754750475406,
      "loss": 2.066,
      "step": 85549
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.8299239873886108,
      "learning_rate": 0.0004180137867499128,
      "loss": 2.2809,
      "step": 85550
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.9162977933883667,
      "learning_rate": 0.0004180100259731307,
      "loss": 2.0163,
      "step": 85551
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.8852705359458923,
      "learning_rate": 0.0004180062651744082,
      "loss": 2.2018,
      "step": 85552
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.8221861720085144,
      "learning_rate": 0.00041800250435374616,
      "loss": 2.1557,
      "step": 85553
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.9008235931396484,
      "learning_rate": 0.0004179987435111452,
      "loss": 2.0825,
      "step": 85554
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.8173028826713562,
      "learning_rate": 0.00041799498264660605,
      "loss": 2.2215,
      "step": 85555
    },
    {
      "epoch": 1.11,
      "grad_norm": 1.0005216598510742,
      "learning_rate": 0.00041799122176012935,
      "loss": 2.3375,
      "step": 85556
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.9899296164512634,
      "learning_rate": 0.00041798746085171593,
      "loss": 2.263,
      "step": 85557
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.8222188353538513,
      "learning_rate": 0.0004179836999213664,
      "loss": 2.0567,
      "step": 85558
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.8620901107788086,
      "learning_rate": 0.00041797993896908143,
      "loss": 2.374,
      "step": 85559
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.878526508808136,
      "learning_rate": 0.0004179761779948619,
      "loss": 2.161,
      "step": 85560
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.7633184790611267,
      "learning_rate": 0.0004179724169987082,
      "loss": 2.1333,
      "step": 85561
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.921701967716217,
      "learning_rate": 0.00041796865598062123,
      "loss": 2.0631,
      "step": 85562
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.8253971934318542,
      "learning_rate": 0.0004179648949406018,
      "loss": 2.3356,
      "step": 85563
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.7782731056213379,
      "learning_rate": 0.00041796113387865036,
      "loss": 2.1732,
      "step": 85564
    },
    {
      "epoch": 1.11,
      "grad_norm": 1.0311601161956787,
      "learning_rate": 0.0004179573727947677,
      "loss": 2.2815,
      "step": 85565
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.8627854585647583,
      "learning_rate": 0.00041795361168895463,
      "loss": 2.2571,
      "step": 85566
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.8336894512176514,
      "learning_rate": 0.00041794985056121167,
      "loss": 2.316,
      "step": 85567
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.8108744025230408,
      "learning_rate": 0.0004179460894115396,
      "loss": 2.1689,
      "step": 85568
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.8903906941413879,
      "learning_rate": 0.0004179423282399392,
      "loss": 2.0207,
      "step": 85569
    },
    {
      "epoch": 1.11,
      "grad_norm": 1.0095309019088745,
      "learning_rate": 0.00041793856704641104,
      "loss": 2.2852,
      "step": 85570
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.851548433303833,
      "learning_rate": 0.0004179348058309558,
      "loss": 2.0121,
      "step": 85571
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.9113506078720093,
      "learning_rate": 0.0004179310445935744,
      "loss": 2.3319,
      "step": 85572
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.9346033334732056,
      "learning_rate": 0.0004179272833342673,
      "loss": 2.1231,
      "step": 85573
    },
    {
      "epoch": 1.11,
      "grad_norm": 1.0074299573898315,
      "learning_rate": 0.0004179235220530353,
      "loss": 2.1883,
      "step": 85574
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.8421506881713867,
      "learning_rate": 0.0004179197607498792,
      "loss": 2.2211,
      "step": 85575
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.8665976524353027,
      "learning_rate": 0.00041791599942479947,
      "loss": 2.2409,
      "step": 85576
    },
    {
      "epoch": 1.11,
      "grad_norm": 1.0657328367233276,
      "learning_rate": 0.00041791223807779694,
      "loss": 2.2065,
      "step": 85577
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.8043864965438843,
      "learning_rate": 0.00041790847670887234,
      "loss": 2.1756,
      "step": 85578
    },
    {
      "epoch": 1.11,
      "grad_norm": 1.1077255010604858,
      "learning_rate": 0.00041790471531802636,
      "loss": 1.9724,
      "step": 85579
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.9148780703544617,
      "learning_rate": 0.0004179009539052596,
      "loss": 1.9492,
      "step": 85580
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.8653829097747803,
      "learning_rate": 0.0004178971924705728,
      "loss": 2.2891,
      "step": 85581
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.9015481472015381,
      "learning_rate": 0.0004178934310139668,
      "loss": 2.1813,
      "step": 85582
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.8232989311218262,
      "learning_rate": 0.0004178896695354421,
      "loss": 2.0671,
      "step": 85583
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.8255395293235779,
      "learning_rate": 0.00041788590803499946,
      "loss": 2.0936,
      "step": 85584
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.9397520422935486,
      "learning_rate": 0.00041788214651263965,
      "loss": 2.3502,
      "step": 85585
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.8879362940788269,
      "learning_rate": 0.00041787838496836334,
      "loss": 2.0668,
      "step": 85586
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.9568161964416504,
      "learning_rate": 0.00041787462340217124,
      "loss": 2.1458,
      "step": 85587
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.807585597038269,
      "learning_rate": 0.00041787086181406394,
      "loss": 2.2975,
      "step": 85588
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.8986862301826477,
      "learning_rate": 0.00041786710020404226,
      "loss": 2.0688,
      "step": 85589
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.7787973284721375,
      "learning_rate": 0.00041786333857210685,
      "loss": 2.41,
      "step": 85590
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.8357827663421631,
      "learning_rate": 0.00041785957691825847,
      "loss": 2.1735,
      "step": 85591
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.8868952393531799,
      "learning_rate": 0.0004178558152424977,
      "loss": 2.1968,
      "step": 85592
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.8258017301559448,
      "learning_rate": 0.00041785205354482534,
      "loss": 2.1858,
      "step": 85593
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.8769987225532532,
      "learning_rate": 0.00041784829182524215,
      "loss": 2.3691,
      "step": 85594
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.835382878780365,
      "learning_rate": 0.00041784453008374855,
      "loss": 2.055,
      "step": 85595
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.7905504107475281,
      "learning_rate": 0.0004178407683203455,
      "loss": 2.1418,
      "step": 85596
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.843439519405365,
      "learning_rate": 0.00041783700653503375,
      "loss": 2.0994,
      "step": 85597
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.9303908944129944,
      "learning_rate": 0.00041783324472781375,
      "loss": 1.9347,
      "step": 85598
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.7745744585990906,
      "learning_rate": 0.00041782948289868636,
      "loss": 2.1142,
      "step": 85599
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.8559730648994446,
      "learning_rate": 0.0004178257210476523,
      "loss": 2.3264,
      "step": 85600
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.8672439455986023,
      "learning_rate": 0.0004178219591747121,
      "loss": 2.0909,
      "step": 85601
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.8986899256706238,
      "learning_rate": 0.0004178181972798666,
      "loss": 2.247,
      "step": 85602
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.857319712638855,
      "learning_rate": 0.0004178144353631165,
      "loss": 2.2986,
      "step": 85603
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.8141180872917175,
      "learning_rate": 0.00041781067342446253,
      "loss": 2.2116,
      "step": 85604
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.912879228591919,
      "learning_rate": 0.0004178069114639052,
      "loss": 2.3232,
      "step": 85605
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.9167534112930298,
      "learning_rate": 0.00041780314948144546,
      "loss": 2.1686,
      "step": 85606
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.8137494325637817,
      "learning_rate": 0.00041779938747708384,
      "loss": 2.3795,
      "step": 85607
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.795428991317749,
      "learning_rate": 0.00041779562545082107,
      "loss": 2.2021,
      "step": 85608
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.9059439301490784,
      "learning_rate": 0.0004177918634026579,
      "loss": 2.1226,
      "step": 85609
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.9474866986274719,
      "learning_rate": 0.00041778810133259503,
      "loss": 2.1295,
      "step": 85610
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.8554973006248474,
      "learning_rate": 0.00041778433924063307,
      "loss": 2.1944,
      "step": 85611
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.8609967827796936,
      "learning_rate": 0.0004177805771267728,
      "loss": 2.3192,
      "step": 85612
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.8090553879737854,
      "learning_rate": 0.0004177768149910149,
      "loss": 2.2671,
      "step": 85613
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.8641365766525269,
      "learning_rate": 0.0004177730528333601,
      "loss": 2.2387,
      "step": 85614
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.7754138112068176,
      "learning_rate": 0.000417769290653809,
      "loss": 2.0316,
      "step": 85615
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.8051830530166626,
      "learning_rate": 0.00041776552845236234,
      "loss": 2.4246,
      "step": 85616
    },
    {
      "epoch": 1.11,
      "grad_norm": 1.0212674140930176,
      "learning_rate": 0.00041776176622902093,
      "loss": 2.2118,
      "step": 85617
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.7969825267791748,
      "learning_rate": 0.0004177580039837853,
      "loss": 2.22,
      "step": 85618
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.8996054530143738,
      "learning_rate": 0.00041775424171665635,
      "loss": 2.2122,
      "step": 85619
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.8365018963813782,
      "learning_rate": 0.0004177504794276346,
      "loss": 2.3919,
      "step": 85620
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.9022846817970276,
      "learning_rate": 0.0004177467171167208,
      "loss": 2.1509,
      "step": 85621
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.7502131462097168,
      "learning_rate": 0.00041774295478391575,
      "loss": 2.1417,
      "step": 85622
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.9225109815597534,
      "learning_rate": 0.00041773919242921994,
      "loss": 2.1388,
      "step": 85623
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.8256253600120544,
      "learning_rate": 0.00041773543005263426,
      "loss": 2.1333,
      "step": 85624
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.8156729340553284,
      "learning_rate": 0.0004177316676541593,
      "loss": 2.2006,
      "step": 85625
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.8116127848625183,
      "learning_rate": 0.0004177279052337958,
      "loss": 2.2226,
      "step": 85626
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.8181217312812805,
      "learning_rate": 0.00041772414279154455,
      "loss": 2.3295,
      "step": 85627
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.9060774445533752,
      "learning_rate": 0.0004177203803274061,
      "loss": 2.2473,
      "step": 85628
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.8350023627281189,
      "learning_rate": 0.00041771661784138113,
      "loss": 2.1854,
      "step": 85629
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.773639440536499,
      "learning_rate": 0.0004177128553334706,
      "loss": 2.2532,
      "step": 85630
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.8040686249732971,
      "learning_rate": 0.00041770909280367484,
      "loss": 2.1199,
      "step": 85631
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.7817437052726746,
      "learning_rate": 0.00041770533025199485,
      "loss": 1.9819,
      "step": 85632
    },
    {
      "epoch": 1.11,
      "grad_norm": 0.8839647173881531,
      "learning_rate": 0.00041770156767843126,
      "loss": 2.2659,
      "step": 85633
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.7962695956230164,
      "learning_rate": 0.00041769780508298457,
      "loss": 2.0844,
      "step": 85634
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.8104801177978516,
      "learning_rate": 0.00041769404246565575,
      "loss": 2.0313,
      "step": 85635
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.9047756195068359,
      "learning_rate": 0.0004176902798264454,
      "loss": 2.0534,
      "step": 85636
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.8781098127365112,
      "learning_rate": 0.0004176865171653541,
      "loss": 2.1922,
      "step": 85637
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.7935932874679565,
      "learning_rate": 0.0004176827544823827,
      "loss": 2.0014,
      "step": 85638
    },
    {
      "epoch": 1.12,
      "grad_norm": 1.0878442525863647,
      "learning_rate": 0.00041767899177753195,
      "loss": 2.0748,
      "step": 85639
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.8044023513793945,
      "learning_rate": 0.00041767522905080236,
      "loss": 2.1098,
      "step": 85640
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.9127084612846375,
      "learning_rate": 0.0004176714663021947,
      "loss": 2.0486,
      "step": 85641
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.8891280293464661,
      "learning_rate": 0.0004176677035317098,
      "loss": 2.2942,
      "step": 85642
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.7441263794898987,
      "learning_rate": 0.00041766394073934816,
      "loss": 1.9135,
      "step": 85643
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.7853368520736694,
      "learning_rate": 0.00041766017792511055,
      "loss": 2.0835,
      "step": 85644
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.8927241563796997,
      "learning_rate": 0.00041765641508899776,
      "loss": 2.3701,
      "step": 85645
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.8361490964889526,
      "learning_rate": 0.00041765265223101043,
      "loss": 2.2513,
      "step": 85646
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.7860010266304016,
      "learning_rate": 0.00041764888935114927,
      "loss": 2.3038,
      "step": 85647
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.8861610293388367,
      "learning_rate": 0.0004176451264494148,
      "loss": 2.0425,
      "step": 85648
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.851370632648468,
      "learning_rate": 0.00041764136352580804,
      "loss": 2.1251,
      "step": 85649
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.7828765511512756,
      "learning_rate": 0.0004176376005803295,
      "loss": 2.3808,
      "step": 85650
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.8705084323883057,
      "learning_rate": 0.00041763383761297983,
      "loss": 2.2175,
      "step": 85651
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.8428728580474854,
      "learning_rate": 0.00041763007462375997,
      "loss": 2.1065,
      "step": 85652
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.8288232684135437,
      "learning_rate": 0.0004176263116126703,
      "loss": 2.1838,
      "step": 85653
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.8273183107376099,
      "learning_rate": 0.00041762254857971175,
      "loss": 2.3098,
      "step": 85654
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.8047272562980652,
      "learning_rate": 0.00041761878552488496,
      "loss": 2.3353,
      "step": 85655
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.7976297736167908,
      "learning_rate": 0.0004176150224481906,
      "loss": 2.1475,
      "step": 85656
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.9136161804199219,
      "learning_rate": 0.00041761125934962936,
      "loss": 2.2025,
      "step": 85657
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.8349213004112244,
      "learning_rate": 0.00041760749622920195,
      "loss": 2.2342,
      "step": 85658
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.831679105758667,
      "learning_rate": 0.0004176037330869091,
      "loss": 2.2273,
      "step": 85659
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.784937858581543,
      "learning_rate": 0.0004175999699227516,
      "loss": 2.2732,
      "step": 85660
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.8187023997306824,
      "learning_rate": 0.00041759620673672995,
      "loss": 2.2944,
      "step": 85661
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.8319103121757507,
      "learning_rate": 0.000417592443528845,
      "loss": 2.283,
      "step": 85662
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.7741925716400146,
      "learning_rate": 0.0004175886802990973,
      "loss": 2.0234,
      "step": 85663
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.784140944480896,
      "learning_rate": 0.0004175849170474878,
      "loss": 2.3543,
      "step": 85664
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.7564695477485657,
      "learning_rate": 0.0004175811537740169,
      "loss": 2.1202,
      "step": 85665
    },
    {
      "epoch": 1.12,
      "grad_norm": 1.0793085098266602,
      "learning_rate": 0.00041757739047868544,
      "loss": 2.332,
      "step": 85666
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.798995316028595,
      "learning_rate": 0.00041757362716149425,
      "loss": 2.4259,
      "step": 85667
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.8820427656173706,
      "learning_rate": 0.00041756986382244376,
      "loss": 2.0357,
      "step": 85668
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.9820702075958252,
      "learning_rate": 0.0004175661004615349,
      "loss": 2.1707,
      "step": 85669
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.8653734922409058,
      "learning_rate": 0.00041756233707876834,
      "loss": 2.0275,
      "step": 85670
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.92721027135849,
      "learning_rate": 0.0004175585736741446,
      "loss": 2.2452,
      "step": 85671
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.7752413749694824,
      "learning_rate": 0.00041755481024766453,
      "loss": 1.9701,
      "step": 85672
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.8815609812736511,
      "learning_rate": 0.00041755104679932884,
      "loss": 2.2952,
      "step": 85673
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.8909837007522583,
      "learning_rate": 0.00041754728332913814,
      "loss": 2.24,
      "step": 85674
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.9138062000274658,
      "learning_rate": 0.0004175435198370932,
      "loss": 2.1241,
      "step": 85675
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.9016297459602356,
      "learning_rate": 0.00041753975632319475,
      "loss": 2.3229,
      "step": 85676
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.7860860824584961,
      "learning_rate": 0.00041753599278744337,
      "loss": 2.1328,
      "step": 85677
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.8025091290473938,
      "learning_rate": 0.0004175322292298399,
      "loss": 2.3759,
      "step": 85678
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.8159134387969971,
      "learning_rate": 0.0004175284656503849,
      "loss": 2.2032,
      "step": 85679
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.8434306979179382,
      "learning_rate": 0.0004175247020490792,
      "loss": 2.1432,
      "step": 85680
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.9379210472106934,
      "learning_rate": 0.0004175209384259234,
      "loss": 2.185,
      "step": 85681
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.768427312374115,
      "learning_rate": 0.00041751717478091826,
      "loss": 2.1027,
      "step": 85682
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.8280022144317627,
      "learning_rate": 0.00041751341111406445,
      "loss": 2.0354,
      "step": 85683
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.9001135230064392,
      "learning_rate": 0.00041750964742536266,
      "loss": 2.1673,
      "step": 85684
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.7577400207519531,
      "learning_rate": 0.0004175058837148137,
      "loss": 2.2217,
      "step": 85685
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.8091970086097717,
      "learning_rate": 0.0004175021199824181,
      "loss": 2.302,
      "step": 85686
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.9330868721008301,
      "learning_rate": 0.00041749835622817655,
      "loss": 2.3671,
      "step": 85687
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.8629102110862732,
      "learning_rate": 0.00041749459245209,
      "loss": 2.1671,
      "step": 85688
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.8299854397773743,
      "learning_rate": 0.0004174908286541589,
      "loss": 2.1023,
      "step": 85689
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.921828031539917,
      "learning_rate": 0.00041748706483438403,
      "loss": 1.8824,
      "step": 85690
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.8638662695884705,
      "learning_rate": 0.00041748330099276615,
      "loss": 2.2009,
      "step": 85691
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.8501549363136292,
      "learning_rate": 0.0004174795371293059,
      "loss": 1.9303,
      "step": 85692
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.8210201263427734,
      "learning_rate": 0.0004174757732440039,
      "loss": 2.2987,
      "step": 85693
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.8108570575714111,
      "learning_rate": 0.00041747200933686104,
      "loss": 2.358,
      "step": 85694
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.8505623936653137,
      "learning_rate": 0.0004174682454078779,
      "loss": 2.335,
      "step": 85695
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.8743743300437927,
      "learning_rate": 0.00041746448145705515,
      "loss": 2.2144,
      "step": 85696
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.833895206451416,
      "learning_rate": 0.0004174607174843935,
      "loss": 2.2811,
      "step": 85697
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.7776248455047607,
      "learning_rate": 0.00041745695348989383,
      "loss": 2.1876,
      "step": 85698
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.9815762042999268,
      "learning_rate": 0.0004174531894735566,
      "loss": 2.4686,
      "step": 85699
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.8012171983718872,
      "learning_rate": 0.0004174494254353826,
      "loss": 2.3236,
      "step": 85700
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.7887644171714783,
      "learning_rate": 0.0004174456613753726,
      "loss": 2.1955,
      "step": 85701
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.9062774181365967,
      "learning_rate": 0.0004174418972935272,
      "loss": 2.1017,
      "step": 85702
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.74058598279953,
      "learning_rate": 0.000417438133189847,
      "loss": 2.085,
      "step": 85703
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.8038554191589355,
      "learning_rate": 0.000417434369064333,
      "loss": 2.2047,
      "step": 85704
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.7800835371017456,
      "learning_rate": 0.00041743060491698574,
      "loss": 1.9897,
      "step": 85705
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.8274244070053101,
      "learning_rate": 0.0004174268407478058,
      "loss": 2.0133,
      "step": 85706
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.8490798473358154,
      "learning_rate": 0.000417423076556794,
      "loss": 2.2958,
      "step": 85707
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.7926515340805054,
      "learning_rate": 0.00041741931234395114,
      "loss": 2.2411,
      "step": 85708
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.7925609946250916,
      "learning_rate": 0.0004174155481092778,
      "loss": 2.0867,
      "step": 85709
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.9607038497924805,
      "learning_rate": 0.0004174117838527746,
      "loss": 2.2576,
      "step": 85710
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.788513720035553,
      "learning_rate": 0.00041740801957444235,
      "loss": 2.1635,
      "step": 85711
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.825094997882843,
      "learning_rate": 0.0004174042552742818,
      "loss": 2.046,
      "step": 85712
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.9003598093986511,
      "learning_rate": 0.0004174004909522936,
      "loss": 2.2773,
      "step": 85713
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.8663116097450256,
      "learning_rate": 0.0004173967266084783,
      "loss": 2.2774,
      "step": 85714
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.7284483909606934,
      "learning_rate": 0.00041739296224283683,
      "loss": 2.2426,
      "step": 85715
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.7394618391990662,
      "learning_rate": 0.00041738919785536984,
      "loss": 2.224,
      "step": 85716
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.929868221282959,
      "learning_rate": 0.0004173854334460779,
      "loss": 2.1781,
      "step": 85717
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.8404777646064758,
      "learning_rate": 0.0004173816690149618,
      "loss": 2.0113,
      "step": 85718
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.872950553894043,
      "learning_rate": 0.0004173779045620223,
      "loss": 2.2787,
      "step": 85719
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.9206806421279907,
      "learning_rate": 0.00041737414008725997,
      "loss": 2.2537,
      "step": 85720
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.8883699178695679,
      "learning_rate": 0.0004173703755906755,
      "loss": 2.0031,
      "step": 85721
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.8505404591560364,
      "learning_rate": 0.00041736661107226984,
      "loss": 2.0616,
      "step": 85722
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.9621952772140503,
      "learning_rate": 0.00041736284653204336,
      "loss": 2.1222,
      "step": 85723
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.8295631408691406,
      "learning_rate": 0.00041735908196999694,
      "loss": 2.0464,
      "step": 85724
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.7941656708717346,
      "learning_rate": 0.0004173553173861313,
      "loss": 2.0811,
      "step": 85725
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.9399883151054382,
      "learning_rate": 0.00041735155278044706,
      "loss": 2.1531,
      "step": 85726
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.8769234418869019,
      "learning_rate": 0.00041734778815294497,
      "loss": 2.0335,
      "step": 85727
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.7689481973648071,
      "learning_rate": 0.0004173440235036257,
      "loss": 2.0853,
      "step": 85728
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.8514216542243958,
      "learning_rate": 0.00041734025883248993,
      "loss": 1.9862,
      "step": 85729
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.8426156640052795,
      "learning_rate": 0.0004173364941395384,
      "loss": 2.0368,
      "step": 85730
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.9810048341751099,
      "learning_rate": 0.0004173327294247719,
      "loss": 2.2335,
      "step": 85731
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.7953462600708008,
      "learning_rate": 0.00041732896468819094,
      "loss": 2.1155,
      "step": 85732
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.8950077295303345,
      "learning_rate": 0.0004173251999297963,
      "loss": 2.1588,
      "step": 85733
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.8272391557693481,
      "learning_rate": 0.00041732143514958874,
      "loss": 2.0213,
      "step": 85734
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.8041521906852722,
      "learning_rate": 0.00041731767034756885,
      "loss": 2.3641,
      "step": 85735
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.8306055665016174,
      "learning_rate": 0.0004173139055237374,
      "loss": 2.0994,
      "step": 85736
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.8483037352561951,
      "learning_rate": 0.0004173101406780952,
      "loss": 2.3213,
      "step": 85737
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.9240252375602722,
      "learning_rate": 0.00041730637581064274,
      "loss": 2.2662,
      "step": 85738
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.755656361579895,
      "learning_rate": 0.0004173026109213808,
      "loss": 1.9178,
      "step": 85739
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.9302209615707397,
      "learning_rate": 0.0004172988460103101,
      "loss": 2.2388,
      "step": 85740
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.7901124954223633,
      "learning_rate": 0.0004172950810774313,
      "loss": 2.2127,
      "step": 85741
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.7842974662780762,
      "learning_rate": 0.00041729131612274514,
      "loss": 2.4651,
      "step": 85742
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.8279808759689331,
      "learning_rate": 0.0004172875511462524,
      "loss": 2.171,
      "step": 85743
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.8429467082023621,
      "learning_rate": 0.0004172837861479536,
      "loss": 2.0614,
      "step": 85744
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.8028485178947449,
      "learning_rate": 0.0004172800211278495,
      "loss": 2.3581,
      "step": 85745
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.8210832476615906,
      "learning_rate": 0.0004172762560859409,
      "loss": 1.9234,
      "step": 85746
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.8698844313621521,
      "learning_rate": 0.0004172724910222284,
      "loss": 2.2573,
      "step": 85747
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.8163522481918335,
      "learning_rate": 0.00041726872593671275,
      "loss": 2.1483,
      "step": 85748
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.7797470688819885,
      "learning_rate": 0.00041726496082939465,
      "loss": 2.0187,
      "step": 85749
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.8510779738426208,
      "learning_rate": 0.00041726119570027477,
      "loss": 1.9142,
      "step": 85750
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.8565192222595215,
      "learning_rate": 0.00041725743054935374,
      "loss": 2.2575,
      "step": 85751
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.7832134962081909,
      "learning_rate": 0.00041725366537663245,
      "loss": 2.1579,
      "step": 85752
    },
    {
      "epoch": 1.12,
      "grad_norm": 1.256060242652893,
      "learning_rate": 0.00041724990018211143,
      "loss": 2.2019,
      "step": 85753
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.9359970092773438,
      "learning_rate": 0.00041724613496579144,
      "loss": 2.2356,
      "step": 85754
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.8217275738716125,
      "learning_rate": 0.00041724236972767324,
      "loss": 2.2053,
      "step": 85755
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.7885702848434448,
      "learning_rate": 0.00041723860446775743,
      "loss": 2.0118,
      "step": 85756
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.7505247592926025,
      "learning_rate": 0.0004172348391860447,
      "loss": 2.1055,
      "step": 85757
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.9551699161529541,
      "learning_rate": 0.0004172310738825359,
      "loss": 2.0657,
      "step": 85758
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.8420913815498352,
      "learning_rate": 0.0004172273085572315,
      "loss": 2.2061,
      "step": 85759
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.9141417741775513,
      "learning_rate": 0.0004172235432101324,
      "loss": 1.8555,
      "step": 85760
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.8354183435440063,
      "learning_rate": 0.00041721977784123926,
      "loss": 2.1084,
      "step": 85761
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.8195798397064209,
      "learning_rate": 0.0004172160124505527,
      "loss": 2.2223,
      "step": 85762
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.8693254590034485,
      "learning_rate": 0.0004172122470380735,
      "loss": 2.293,
      "step": 85763
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.911742091178894,
      "learning_rate": 0.0004172084816038023,
      "loss": 2.0732,
      "step": 85764
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.9154816269874573,
      "learning_rate": 0.0004172047161477399,
      "loss": 2.1238,
      "step": 85765
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.9419666528701782,
      "learning_rate": 0.00041720095066988684,
      "loss": 2.0469,
      "step": 85766
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.9116207957267761,
      "learning_rate": 0.00041719718517024406,
      "loss": 2.2498,
      "step": 85767
    },
    {
      "epoch": 1.12,
      "grad_norm": 1.0039856433868408,
      "learning_rate": 0.00041719341964881196,
      "loss": 2.1916,
      "step": 85768
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.7991940975189209,
      "learning_rate": 0.00041718965410559146,
      "loss": 1.9956,
      "step": 85769
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.8420090675354004,
      "learning_rate": 0.0004171858885405832,
      "loss": 2.3783,
      "step": 85770
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.9171953797340393,
      "learning_rate": 0.0004171821229537878,
      "loss": 2.1869,
      "step": 85771
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.9057896733283997,
      "learning_rate": 0.00041717835734520603,
      "loss": 2.1871,
      "step": 85772
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.8072848320007324,
      "learning_rate": 0.0004171745917148387,
      "loss": 2.1906,
      "step": 85773
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.8396236300468445,
      "learning_rate": 0.0004171708260626863,
      "loss": 2.1398,
      "step": 85774
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.8463890552520752,
      "learning_rate": 0.0004171670603887497,
      "loss": 2.1785,
      "step": 85775
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.9977611899375916,
      "learning_rate": 0.00041716329469302944,
      "loss": 2.2616,
      "step": 85776
    },
    {
      "epoch": 1.12,
      "grad_norm": 1.1028019189834595,
      "learning_rate": 0.0004171595289755264,
      "loss": 2.2769,
      "step": 85777
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.7742425799369812,
      "learning_rate": 0.00041715576323624124,
      "loss": 2.2602,
      "step": 85778
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.8472887873649597,
      "learning_rate": 0.00041715199747517444,
      "loss": 2.3164,
      "step": 85779
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.7833738923072815,
      "learning_rate": 0.00041714823169232705,
      "loss": 2.0742,
      "step": 85780
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.8726974129676819,
      "learning_rate": 0.0004171444658876995,
      "loss": 2.2127,
      "step": 85781
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.8079659342765808,
      "learning_rate": 0.0004171407000612925,
      "loss": 1.8074,
      "step": 85782
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.945723831653595,
      "learning_rate": 0.000417136934213107,
      "loss": 2.0793,
      "step": 85783
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.8341003656387329,
      "learning_rate": 0.0004171331683431435,
      "loss": 2.0807,
      "step": 85784
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.8261221051216125,
      "learning_rate": 0.00041712940245140263,
      "loss": 2.1715,
      "step": 85785
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.8588148951530457,
      "learning_rate": 0.0004171256365378853,
      "loss": 2.2543,
      "step": 85786
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.8489080667495728,
      "learning_rate": 0.00041712187060259203,
      "loss": 2.1711,
      "step": 85787
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.7341146469116211,
      "learning_rate": 0.0004171181046455236,
      "loss": 1.9176,
      "step": 85788
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.756942629814148,
      "learning_rate": 0.0004171143386666807,
      "loss": 2.1474,
      "step": 85789
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.7978930473327637,
      "learning_rate": 0.00041711057266606414,
      "loss": 2.0027,
      "step": 85790
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.8050444722175598,
      "learning_rate": 0.0004171068066436744,
      "loss": 2.3939,
      "step": 85791
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.8522365093231201,
      "learning_rate": 0.0004171030405995123,
      "loss": 2.1021,
      "step": 85792
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.826803982257843,
      "learning_rate": 0.00041709927453357864,
      "loss": 2.3613,
      "step": 85793
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.7846895456314087,
      "learning_rate": 0.00041709550844587387,
      "loss": 2.203,
      "step": 85794
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.7892168164253235,
      "learning_rate": 0.00041709174233639897,
      "loss": 2.1965,
      "step": 85795
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.9200645089149475,
      "learning_rate": 0.0004170879762051543,
      "loss": 1.9839,
      "step": 85796
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.8316824436187744,
      "learning_rate": 0.000417084210052141,
      "loss": 2.3336,
      "step": 85797
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.7931361198425293,
      "learning_rate": 0.0004170804438773594,
      "loss": 2.1308,
      "step": 85798
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.9607912302017212,
      "learning_rate": 0.00041707667768081034,
      "loss": 2.1594,
      "step": 85799
    },
    {
      "epoch": 1.12,
      "grad_norm": 1.0818536281585693,
      "learning_rate": 0.00041707291146249456,
      "loss": 2.1456,
      "step": 85800
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.7673908472061157,
      "learning_rate": 0.00041706914522241276,
      "loss": 2.3099,
      "step": 85801
    },
    {
      "epoch": 1.12,
      "grad_norm": 1.0720924139022827,
      "learning_rate": 0.0004170653789605655,
      "loss": 2.2592,
      "step": 85802
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.8675478100776672,
      "learning_rate": 0.0004170616126769536,
      "loss": 2.2422,
      "step": 85803
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.8349890112876892,
      "learning_rate": 0.0004170578463715778,
      "loss": 2.1305,
      "step": 85804
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.819713830947876,
      "learning_rate": 0.0004170540800444386,
      "loss": 2.3363,
      "step": 85805
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.8566808104515076,
      "learning_rate": 0.00041705031369553697,
      "loss": 2.116,
      "step": 85806
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.8144016861915588,
      "learning_rate": 0.0004170465473248735,
      "loss": 2.2215,
      "step": 85807
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.7735687494277954,
      "learning_rate": 0.00041704278093244876,
      "loss": 2.1836,
      "step": 85808
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.8072500228881836,
      "learning_rate": 0.00041703901451826356,
      "loss": 2.2787,
      "step": 85809
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.7944168448448181,
      "learning_rate": 0.0004170352480823187,
      "loss": 2.1074,
      "step": 85810
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.8631988167762756,
      "learning_rate": 0.00041703148162461465,
      "loss": 2.3269,
      "step": 85811
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.8916945457458496,
      "learning_rate": 0.00041702771514515234,
      "loss": 2.2294,
      "step": 85812
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.7421867847442627,
      "learning_rate": 0.0004170239486439324,
      "loss": 2.0017,
      "step": 85813
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.8725194931030273,
      "learning_rate": 0.0004170201821209554,
      "loss": 2.215,
      "step": 85814
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.9350161552429199,
      "learning_rate": 0.0004170164155762221,
      "loss": 2.1708,
      "step": 85815
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.7700936198234558,
      "learning_rate": 0.00041701264900973335,
      "loss": 2.1318,
      "step": 85816
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.868476152420044,
      "learning_rate": 0.0004170088824214897,
      "loss": 2.2576,
      "step": 85817
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.8068209886550903,
      "learning_rate": 0.0004170051158114919,
      "loss": 2.2629,
      "step": 85818
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.7692463994026184,
      "learning_rate": 0.00041700134917974064,
      "loss": 2.168,
      "step": 85819
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.8100389242172241,
      "learning_rate": 0.00041699758252623656,
      "loss": 2.1251,
      "step": 85820
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.7991319894790649,
      "learning_rate": 0.00041699381585098047,
      "loss": 2.127,
      "step": 85821
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.7971677184104919,
      "learning_rate": 0.0004169900491539731,
      "loss": 2.172,
      "step": 85822
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.8285939693450928,
      "learning_rate": 0.00041698628243521497,
      "loss": 2.0747,
      "step": 85823
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.8680174946784973,
      "learning_rate": 0.00041698251569470687,
      "loss": 2.1864,
      "step": 85824
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.8323320746421814,
      "learning_rate": 0.00041697874893244956,
      "loss": 2.1716,
      "step": 85825
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.7765839099884033,
      "learning_rate": 0.00041697498214844365,
      "loss": 2.3228,
      "step": 85826
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.8398610353469849,
      "learning_rate": 0.0004169712153426899,
      "loss": 2.0919,
      "step": 85827
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.850755512714386,
      "learning_rate": 0.00041696744851518904,
      "loss": 2.0467,
      "step": 85828
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.7263346314430237,
      "learning_rate": 0.0004169636816659417,
      "loss": 2.1321,
      "step": 85829
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.8280904293060303,
      "learning_rate": 0.0004169599147949486,
      "loss": 2.1977,
      "step": 85830
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.8490105867385864,
      "learning_rate": 0.00041695614790221045,
      "loss": 2.1656,
      "step": 85831
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.8140764236450195,
      "learning_rate": 0.0004169523809877279,
      "loss": 1.998,
      "step": 85832
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.7993914484977722,
      "learning_rate": 0.0004169486140515017,
      "loss": 2.247,
      "step": 85833
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.7626917362213135,
      "learning_rate": 0.0004169448470935326,
      "loss": 2.1849,
      "step": 85834
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.832263171672821,
      "learning_rate": 0.0004169410801138213,
      "loss": 2.248,
      "step": 85835
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.7882657051086426,
      "learning_rate": 0.00041693731311236826,
      "loss": 2.1957,
      "step": 85836
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.897208571434021,
      "learning_rate": 0.0004169335460891745,
      "loss": 2.2578,
      "step": 85837
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.8760308027267456,
      "learning_rate": 0.0004169297790442406,
      "loss": 2.1049,
      "step": 85838
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.7777121663093567,
      "learning_rate": 0.00041692601197756713,
      "loss": 2.0791,
      "step": 85839
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.802375078201294,
      "learning_rate": 0.00041692224488915503,
      "loss": 2.1387,
      "step": 85840
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.9398114681243896,
      "learning_rate": 0.00041691847777900484,
      "loss": 2.0911,
      "step": 85841
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.8625625967979431,
      "learning_rate": 0.00041691471064711727,
      "loss": 2.3201,
      "step": 85842
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.8059583306312561,
      "learning_rate": 0.000416910943493493,
      "loss": 2.3454,
      "step": 85843
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.9026523232460022,
      "learning_rate": 0.0004169071763181329,
      "loss": 2.1183,
      "step": 85844
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.7755535840988159,
      "learning_rate": 0.0004169034091210375,
      "loss": 2.1235,
      "step": 85845
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.8901609778404236,
      "learning_rate": 0.00041689964190220753,
      "loss": 2.3753,
      "step": 85846
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.7722964286804199,
      "learning_rate": 0.00041689587466164376,
      "loss": 2.1453,
      "step": 85847
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.767109215259552,
      "learning_rate": 0.0004168921073993467,
      "loss": 2.143,
      "step": 85848
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.8518153429031372,
      "learning_rate": 0.00041688834011531734,
      "loss": 2.3775,
      "step": 85849
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.8860372304916382,
      "learning_rate": 0.0004168845728095562,
      "loss": 2.0317,
      "step": 85850
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.7996075749397278,
      "learning_rate": 0.000416880805482064,
      "loss": 2.1776,
      "step": 85851
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.8970286250114441,
      "learning_rate": 0.0004168770381328414,
      "loss": 2.1958,
      "step": 85852
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.7870593667030334,
      "learning_rate": 0.00041687327076188926,
      "loss": 2.3262,
      "step": 85853
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.789921760559082,
      "learning_rate": 0.0004168695033692081,
      "loss": 2.4019,
      "step": 85854
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.8820779919624329,
      "learning_rate": 0.0004168657359547986,
      "loss": 2.2962,
      "step": 85855
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.8568138480186462,
      "learning_rate": 0.00041686196851866173,
      "loss": 2.1366,
      "step": 85856
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.8271114230155945,
      "learning_rate": 0.00041685820106079793,
      "loss": 2.2173,
      "step": 85857
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.8067487478256226,
      "learning_rate": 0.00041685443358120797,
      "loss": 2.2009,
      "step": 85858
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.8230987787246704,
      "learning_rate": 0.00041685066607989265,
      "loss": 2.198,
      "step": 85859
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.9107187986373901,
      "learning_rate": 0.00041684689855685253,
      "loss": 2.32,
      "step": 85860
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.7739903330802917,
      "learning_rate": 0.00041684313101208835,
      "loss": 2.2429,
      "step": 85861
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.8417555093765259,
      "learning_rate": 0.00041683936344560083,
      "loss": 2.183,
      "step": 85862
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.8003227710723877,
      "learning_rate": 0.00041683559585739067,
      "loss": 2.2463,
      "step": 85863
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.8403459787368774,
      "learning_rate": 0.00041683182824745857,
      "loss": 2.1743,
      "step": 85864
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.8234722018241882,
      "learning_rate": 0.00041682806061580524,
      "loss": 2.2135,
      "step": 85865
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.8044763803482056,
      "learning_rate": 0.0004168242929624314,
      "loss": 2.0405,
      "step": 85866
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.8340288996696472,
      "learning_rate": 0.00041682052528733764,
      "loss": 2.0231,
      "step": 85867
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.7602745890617371,
      "learning_rate": 0.00041681675759052485,
      "loss": 2.1659,
      "step": 85868
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.7993623614311218,
      "learning_rate": 0.0004168129898719935,
      "loss": 1.8886,
      "step": 85869
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.8161554336547852,
      "learning_rate": 0.0004168092221317444,
      "loss": 2.1344,
      "step": 85870
    },
    {
      "epoch": 1.12,
      "grad_norm": 1.0064412355422974,
      "learning_rate": 0.00041680545436977837,
      "loss": 2.3688,
      "step": 85871
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.819832444190979,
      "learning_rate": 0.00041680168658609593,
      "loss": 2.1598,
      "step": 85872
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.8787040710449219,
      "learning_rate": 0.00041679791878069785,
      "loss": 2.4202,
      "step": 85873
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.8049090504646301,
      "learning_rate": 0.0004167941509535849,
      "loss": 2.2845,
      "step": 85874
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.8165313005447388,
      "learning_rate": 0.00041679038310475766,
      "loss": 2.0792,
      "step": 85875
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.888366162776947,
      "learning_rate": 0.00041678661523421686,
      "loss": 2.1898,
      "step": 85876
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.8766975402832031,
      "learning_rate": 0.00041678284734196323,
      "loss": 2.0184,
      "step": 85877
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.7748641967773438,
      "learning_rate": 0.00041677907942799753,
      "loss": 2.2453,
      "step": 85878
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.7757228016853333,
      "learning_rate": 0.0004167753114923203,
      "loss": 2.2057,
      "step": 85879
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.7964905500411987,
      "learning_rate": 0.0004167715435349324,
      "loss": 1.9664,
      "step": 85880
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.9444499015808105,
      "learning_rate": 0.0004167677755558344,
      "loss": 2.3293,
      "step": 85881
    },
    {
      "epoch": 1.12,
      "grad_norm": 1.0002199411392212,
      "learning_rate": 0.0004167640075550272,
      "loss": 2.139,
      "step": 85882
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.8004662394523621,
      "learning_rate": 0.00041676023953251124,
      "loss": 2.3526,
      "step": 85883
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.9691365361213684,
      "learning_rate": 0.00041675647148828744,
      "loss": 2.1412,
      "step": 85884
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.9967418909072876,
      "learning_rate": 0.00041675270342235633,
      "loss": 2.1412,
      "step": 85885
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.7907679677009583,
      "learning_rate": 0.00041674893533471874,
      "loss": 2.3087,
      "step": 85886
    },
    {
      "epoch": 1.12,
      "grad_norm": 1.114241600036621,
      "learning_rate": 0.00041674516722537525,
      "loss": 2.2574,
      "step": 85887
    },
    {
      "epoch": 1.12,
      "grad_norm": 1.1020689010620117,
      "learning_rate": 0.0004167413990943267,
      "loss": 2.051,
      "step": 85888
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.9017982482910156,
      "learning_rate": 0.0004167376309415737,
      "loss": 2.1286,
      "step": 85889
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.9972520470619202,
      "learning_rate": 0.000416733862767117,
      "loss": 2.2332,
      "step": 85890
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.9056938886642456,
      "learning_rate": 0.0004167300945709572,
      "loss": 2.1477,
      "step": 85891
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.8420514464378357,
      "learning_rate": 0.0004167263263530952,
      "loss": 2.196,
      "step": 85892
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.9515300393104553,
      "learning_rate": 0.0004167225581135314,
      "loss": 2.027,
      "step": 85893
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.7345487475395203,
      "learning_rate": 0.00041671878985226677,
      "loss": 2.3206,
      "step": 85894
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.8424374461174011,
      "learning_rate": 0.00041671502156930197,
      "loss": 2.1662,
      "step": 85895
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.7961385846138,
      "learning_rate": 0.0004167112532646375,
      "loss": 2.2153,
      "step": 85896
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.9725359082221985,
      "learning_rate": 0.00041670748493827426,
      "loss": 2.3643,
      "step": 85897
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.8767052292823792,
      "learning_rate": 0.00041670371659021293,
      "loss": 2.2627,
      "step": 85898
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.7535034418106079,
      "learning_rate": 0.00041669994822045417,
      "loss": 2.1707,
      "step": 85899
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.8250537514686584,
      "learning_rate": 0.0004166961798289987,
      "loss": 2.1126,
      "step": 85900
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.7756422758102417,
      "learning_rate": 0.0004166924114158472,
      "loss": 2.2493,
      "step": 85901
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.8655022978782654,
      "learning_rate": 0.0004166886429810004,
      "loss": 2.2462,
      "step": 85902
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.796750545501709,
      "learning_rate": 0.00041668487452445894,
      "loss": 2.1981,
      "step": 85903
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.7935773730278015,
      "learning_rate": 0.0004166811060462236,
      "loss": 2.005,
      "step": 85904
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.8886436223983765,
      "learning_rate": 0.0004166773375462949,
      "loss": 2.1787,
      "step": 85905
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.8776932954788208,
      "learning_rate": 0.0004166735690246738,
      "loss": 2.1055,
      "step": 85906
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.831746518611908,
      "learning_rate": 0.0004166698004813609,
      "loss": 2.2972,
      "step": 85907
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.7740324139595032,
      "learning_rate": 0.0004166660319163568,
      "loss": 1.8904,
      "step": 85908
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.8229486346244812,
      "learning_rate": 0.0004166622633296623,
      "loss": 2.2849,
      "step": 85909
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.8180678486824036,
      "learning_rate": 0.0004166584947212781,
      "loss": 2.1655,
      "step": 85910
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.8710052967071533,
      "learning_rate": 0.000416654726091205,
      "loss": 2.1338,
      "step": 85911
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.733607292175293,
      "learning_rate": 0.00041665095743944345,
      "loss": 2.1707,
      "step": 85912
    },
    {
      "epoch": 1.12,
      "grad_norm": 1.0635432004928589,
      "learning_rate": 0.00041664718876599423,
      "loss": 2.2316,
      "step": 85913
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.8172620534896851,
      "learning_rate": 0.00041664342007085827,
      "loss": 2.142,
      "step": 85914
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.8533119559288025,
      "learning_rate": 0.000416639651354036,
      "loss": 2.2758,
      "step": 85915
    },
    {
      "epoch": 1.12,
      "grad_norm": 1.1338061094284058,
      "learning_rate": 0.0004166358826155282,
      "loss": 1.9288,
      "step": 85916
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.7460789084434509,
      "learning_rate": 0.0004166321138553357,
      "loss": 2.0141,
      "step": 85917
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.8327071070671082,
      "learning_rate": 0.00041662834507345895,
      "loss": 2.2816,
      "step": 85918
    },
    {
      "epoch": 1.12,
      "grad_norm": 1.102401852607727,
      "learning_rate": 0.0004166245762698988,
      "loss": 2.1645,
      "step": 85919
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.7849215865135193,
      "learning_rate": 0.00041662080744465604,
      "loss": 2.2922,
      "step": 85920
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.8181428909301758,
      "learning_rate": 0.00041661703859773117,
      "loss": 2.2746,
      "step": 85921
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.9759379625320435,
      "learning_rate": 0.000416613269729125,
      "loss": 2.1244,
      "step": 85922
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.8632584810256958,
      "learning_rate": 0.0004166095008388383,
      "loss": 2.1145,
      "step": 85923
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.8838034868240356,
      "learning_rate": 0.00041660573192687166,
      "loss": 2.1207,
      "step": 85924
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.9140863418579102,
      "learning_rate": 0.00041660196299322575,
      "loss": 2.1382,
      "step": 85925
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.9915216565132141,
      "learning_rate": 0.0004165981940379014,
      "loss": 2.2152,
      "step": 85926
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.7939258813858032,
      "learning_rate": 0.0004165944250608993,
      "loss": 2.1892,
      "step": 85927
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.8278926014900208,
      "learning_rate": 0.00041659065606221995,
      "loss": 2.2603,
      "step": 85928
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.8328264951705933,
      "learning_rate": 0.0004165868870418643,
      "loss": 2.3078,
      "step": 85929
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.8204864859580994,
      "learning_rate": 0.0004165831179998329,
      "loss": 2.1223,
      "step": 85930
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.7658972144126892,
      "learning_rate": 0.0004165793489361265,
      "loss": 1.9989,
      "step": 85931
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.8823407888412476,
      "learning_rate": 0.00041657557985074584,
      "loss": 2.0068,
      "step": 85932
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.9583902359008789,
      "learning_rate": 0.00041657181074369155,
      "loss": 2.1532,
      "step": 85933
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.7682125568389893,
      "learning_rate": 0.0004165680416149644,
      "loss": 2.2138,
      "step": 85934
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.739384651184082,
      "learning_rate": 0.000416564272464565,
      "loss": 2.1587,
      "step": 85935
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.7822492122650146,
      "learning_rate": 0.0004165605032924941,
      "loss": 2.2553,
      "step": 85936
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.7842036485671997,
      "learning_rate": 0.00041655673409875244,
      "loss": 2.1024,
      "step": 85937
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.7807686924934387,
      "learning_rate": 0.0004165529648833407,
      "loss": 2.1233,
      "step": 85938
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.8588107824325562,
      "learning_rate": 0.00041654919564625945,
      "loss": 2.0913,
      "step": 85939
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.8010901808738708,
      "learning_rate": 0.0004165454263875096,
      "loss": 2.1622,
      "step": 85940
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.9500741958618164,
      "learning_rate": 0.00041654165710709176,
      "loss": 2.064,
      "step": 85941
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.7795125842094421,
      "learning_rate": 0.00041653788780500657,
      "loss": 2.3348,
      "step": 85942
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.9231199622154236,
      "learning_rate": 0.00041653411848125484,
      "loss": 2.1587,
      "step": 85943
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.8745703101158142,
      "learning_rate": 0.0004165303491358372,
      "loss": 1.9499,
      "step": 85944
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.8103327751159668,
      "learning_rate": 0.0004165265797687543,
      "loss": 2.3108,
      "step": 85945
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.904106855392456,
      "learning_rate": 0.000416522810380007,
      "loss": 2.2077,
      "step": 85946
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.8475261330604553,
      "learning_rate": 0.0004165190409695959,
      "loss": 2.1609,
      "step": 85947
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.8925936818122864,
      "learning_rate": 0.0004165152715375217,
      "loss": 2.2472,
      "step": 85948
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.8305756449699402,
      "learning_rate": 0.0004165115020837851,
      "loss": 2.1967,
      "step": 85949
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.994161069393158,
      "learning_rate": 0.0004165077326083868,
      "loss": 2.1689,
      "step": 85950
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.7972879409790039,
      "learning_rate": 0.0004165039631113275,
      "loss": 2.271,
      "step": 85951
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.924958348274231,
      "learning_rate": 0.0004165001935926079,
      "loss": 2.1038,
      "step": 85952
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.8337539434432983,
      "learning_rate": 0.0004164964240522288,
      "loss": 2.0608,
      "step": 85953
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.7528795003890991,
      "learning_rate": 0.0004164926544901908,
      "loss": 2.1744,
      "step": 85954
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.8222541809082031,
      "learning_rate": 0.0004164888849064946,
      "loss": 2.3592,
      "step": 85955
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.8032026886940002,
      "learning_rate": 0.00041648511530114094,
      "loss": 2.0689,
      "step": 85956
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.8368369340896606,
      "learning_rate": 0.00041648134567413045,
      "loss": 2.1565,
      "step": 85957
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.8335092067718506,
      "learning_rate": 0.0004164775760254638,
      "loss": 2.2519,
      "step": 85958
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.8120678067207336,
      "learning_rate": 0.00041647380635514194,
      "loss": 2.5482,
      "step": 85959
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.8493525385856628,
      "learning_rate": 0.0004164700366631653,
      "loss": 2.0915,
      "step": 85960
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.8263002634048462,
      "learning_rate": 0.00041646626694953474,
      "loss": 2.2439,
      "step": 85961
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.7936620712280273,
      "learning_rate": 0.00041646249721425077,
      "loss": 2.0565,
      "step": 85962
    },
    {
      "epoch": 1.12,
      "grad_norm": 1.1986671686172485,
      "learning_rate": 0.0004164587274573144,
      "loss": 2.0172,
      "step": 85963
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.8025627136230469,
      "learning_rate": 0.00041645495767872606,
      "loss": 2.3054,
      "step": 85964
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.7580503225326538,
      "learning_rate": 0.00041645118787848656,
      "loss": 2.2597,
      "step": 85965
    },
    {
      "epoch": 1.12,
      "grad_norm": 1.191691517829895,
      "learning_rate": 0.00041644741805659663,
      "loss": 2.0603,
      "step": 85966
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.9713187217712402,
      "learning_rate": 0.00041644364821305686,
      "loss": 2.2048,
      "step": 85967
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.8126248121261597,
      "learning_rate": 0.000416439878347868,
      "loss": 2.2389,
      "step": 85968
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.8317875862121582,
      "learning_rate": 0.0004164361084610308,
      "loss": 2.2052,
      "step": 85969
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.9399236440658569,
      "learning_rate": 0.000416432338552546,
      "loss": 2.2044,
      "step": 85970
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.8384181261062622,
      "learning_rate": 0.0004164285686224141,
      "loss": 1.956,
      "step": 85971
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.8547629117965698,
      "learning_rate": 0.00041642479867063606,
      "loss": 2.2088,
      "step": 85972
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.8647533655166626,
      "learning_rate": 0.00041642102869721235,
      "loss": 2.1707,
      "step": 85973
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.8261281251907349,
      "learning_rate": 0.0004164172587021439,
      "loss": 2.0154,
      "step": 85974
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.8062881827354431,
      "learning_rate": 0.0004164134886854311,
      "loss": 2.0999,
      "step": 85975
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.8848654627799988,
      "learning_rate": 0.00041640971864707497,
      "loss": 2.089,
      "step": 85976
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.9215626120567322,
      "learning_rate": 0.00041640594858707603,
      "loss": 2.0906,
      "step": 85977
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.8529564738273621,
      "learning_rate": 0.00041640217850543506,
      "loss": 2.1563,
      "step": 85978
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.8384774923324585,
      "learning_rate": 0.0004163984084021527,
      "loss": 2.1617,
      "step": 85979
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.8331137299537659,
      "learning_rate": 0.0004163946382772296,
      "loss": 2.081,
      "step": 85980
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.8244777917861938,
      "learning_rate": 0.0004163908681306667,
      "loss": 2.0155,
      "step": 85981
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.8082924485206604,
      "learning_rate": 0.00041638709796246447,
      "loss": 2.0986,
      "step": 85982
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.7412146329879761,
      "learning_rate": 0.00041638332777262365,
      "loss": 2.16,
      "step": 85983
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.8514326214790344,
      "learning_rate": 0.0004163795575611451,
      "loss": 2.0736,
      "step": 85984
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.8781723976135254,
      "learning_rate": 0.0004163757873280292,
      "loss": 2.1159,
      "step": 85985
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.7720231413841248,
      "learning_rate": 0.00041637201707327696,
      "loss": 2.4056,
      "step": 85986
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.8778079748153687,
      "learning_rate": 0.000416368246796889,
      "loss": 2.243,
      "step": 85987
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.8316865563392639,
      "learning_rate": 0.0004163644764988659,
      "loss": 2.1883,
      "step": 85988
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.8217073678970337,
      "learning_rate": 0.00041636070617920843,
      "loss": 2.0826,
      "step": 85989
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.8728166222572327,
      "learning_rate": 0.0004163569358379174,
      "loss": 2.1605,
      "step": 85990
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.8498985171318054,
      "learning_rate": 0.00041635316547499336,
      "loss": 2.4268,
      "step": 85991
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.793576180934906,
      "learning_rate": 0.00041634939509043703,
      "loss": 2.1236,
      "step": 85992
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.9757837653160095,
      "learning_rate": 0.0004163456246842493,
      "loss": 2.122,
      "step": 85993
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.826143205165863,
      "learning_rate": 0.0004163418542564306,
      "loss": 2.2739,
      "step": 85994
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.8560765981674194,
      "learning_rate": 0.00041633808380698174,
      "loss": 2.2482,
      "step": 85995
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.8924028873443604,
      "learning_rate": 0.00041633431333590353,
      "loss": 2.1785,
      "step": 85996
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.7962669730186462,
      "learning_rate": 0.0004163305428431965,
      "loss": 2.1097,
      "step": 85997
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.8486401438713074,
      "learning_rate": 0.00041632677232886147,
      "loss": 2.4495,
      "step": 85998
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.8304189443588257,
      "learning_rate": 0.00041632300179289907,
      "loss": 1.8474,
      "step": 85999
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.8540036678314209,
      "learning_rate": 0.00041631923123531004,
      "loss": 2.4346,
      "step": 86000
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.7987638711929321,
      "learning_rate": 0.00041631546065609503,
      "loss": 2.1756,
      "step": 86001
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.8897412419319153,
      "learning_rate": 0.00041631169005525485,
      "loss": 2.2139,
      "step": 86002
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.8549078106880188,
      "learning_rate": 0.0004163079194327901,
      "loss": 2.0176,
      "step": 86003
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.8199347257614136,
      "learning_rate": 0.00041630414878870146,
      "loss": 2.0372,
      "step": 86004
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.8077635765075684,
      "learning_rate": 0.0004163003781229898,
      "loss": 2.0265,
      "step": 86005
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.8671873807907104,
      "learning_rate": 0.00041629660743565557,
      "loss": 2.1905,
      "step": 86006
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.9592827558517456,
      "learning_rate": 0.00041629283672669966,
      "loss": 2.3696,
      "step": 86007
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.8703542351722717,
      "learning_rate": 0.00041628906599612276,
      "loss": 2.0502,
      "step": 86008
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.7994541525840759,
      "learning_rate": 0.00041628529524392546,
      "loss": 2.1524,
      "step": 86009
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.8385932445526123,
      "learning_rate": 0.00041628152447010857,
      "loss": 2.0865,
      "step": 86010
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.8277751803398132,
      "learning_rate": 0.0004162777536746728,
      "loss": 2.3892,
      "step": 86011
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.7681797742843628,
      "learning_rate": 0.00041627398285761874,
      "loss": 2.2863,
      "step": 86012
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.797536313533783,
      "learning_rate": 0.0004162702120189471,
      "loss": 2.1482,
      "step": 86013
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.7616693377494812,
      "learning_rate": 0.00041626644115865875,
      "loss": 2.1166,
      "step": 86014
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.7910473346710205,
      "learning_rate": 0.0004162626702767542,
      "loss": 2.2581,
      "step": 86015
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.8267085552215576,
      "learning_rate": 0.00041625889937323417,
      "loss": 2.1088,
      "step": 86016
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.8310745358467102,
      "learning_rate": 0.0004162551284480995,
      "loss": 2.1473,
      "step": 86017
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.768944501876831,
      "learning_rate": 0.0004162513575013508,
      "loss": 2.319,
      "step": 86018
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.7942913770675659,
      "learning_rate": 0.0004162475865329887,
      "loss": 1.9096,
      "step": 86019
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.7646503448486328,
      "learning_rate": 0.00041624381554301415,
      "loss": 2.2694,
      "step": 86020
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.8456836938858032,
      "learning_rate": 0.00041624004453142755,
      "loss": 2.2581,
      "step": 86021
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.8253294229507446,
      "learning_rate": 0.0004162362734982297,
      "loss": 2.1859,
      "step": 86022
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.8427711129188538,
      "learning_rate": 0.00041623250244342144,
      "loss": 2.2674,
      "step": 86023
    },
    {
      "epoch": 1.12,
      "grad_norm": 1.402204990386963,
      "learning_rate": 0.0004162287313670033,
      "loss": 2.2959,
      "step": 86024
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.8722660541534424,
      "learning_rate": 0.0004162249602689761,
      "loss": 2.3506,
      "step": 86025
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.8072503805160522,
      "learning_rate": 0.00041622118914934044,
      "loss": 2.2105,
      "step": 86026
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.8861396312713623,
      "learning_rate": 0.0004162174180080971,
      "loss": 2.1386,
      "step": 86027
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.7967190742492676,
      "learning_rate": 0.00041621364684524666,
      "loss": 2.2467,
      "step": 86028
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.7917269468307495,
      "learning_rate": 0.00041620987566079004,
      "loss": 2.0871,
      "step": 86029
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.8176664113998413,
      "learning_rate": 0.00041620610445472766,
      "loss": 2.0162,
      "step": 86030
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.7853496670722961,
      "learning_rate": 0.00041620233322706043,
      "loss": 2.3388,
      "step": 86031
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.8193072080612183,
      "learning_rate": 0.00041619856197778907,
      "loss": 2.1924,
      "step": 86032
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.9449694156646729,
      "learning_rate": 0.0004161947907069141,
      "loss": 2.133,
      "step": 86033
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.8264172077178955,
      "learning_rate": 0.0004161910194144364,
      "loss": 2.1172,
      "step": 86034
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.7538941502571106,
      "learning_rate": 0.0004161872481003566,
      "loss": 2.4012,
      "step": 86035
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.8194326162338257,
      "learning_rate": 0.0004161834767646753,
      "loss": 2.3074,
      "step": 86036
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.824530303478241,
      "learning_rate": 0.0004161797054073934,
      "loss": 2.185,
      "step": 86037
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.8225100040435791,
      "learning_rate": 0.0004161759340285115,
      "loss": 2.1848,
      "step": 86038
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.7989867925643921,
      "learning_rate": 0.00041617216262803026,
      "loss": 2.1368,
      "step": 86039
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.7643502354621887,
      "learning_rate": 0.00041616839120595036,
      "loss": 2.1574,
      "step": 86040
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.826192319393158,
      "learning_rate": 0.0004161646197622727,
      "loss": 2.446,
      "step": 86041
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.8470970392227173,
      "learning_rate": 0.0004161608482969978,
      "loss": 2.1963,
      "step": 86042
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.8593319058418274,
      "learning_rate": 0.00041615707681012635,
      "loss": 2.1629,
      "step": 86043
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.8063570261001587,
      "learning_rate": 0.00041615330530165906,
      "loss": 2.2496,
      "step": 86044
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.8442242741584778,
      "learning_rate": 0.0004161495337715968,
      "loss": 2.2031,
      "step": 86045
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.8210657835006714,
      "learning_rate": 0.0004161457622199401,
      "loss": 2.113,
      "step": 86046
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.8264404535293579,
      "learning_rate": 0.0004161419906466897,
      "loss": 2.1012,
      "step": 86047
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.8360705375671387,
      "learning_rate": 0.0004161382190518464,
      "loss": 2.338,
      "step": 86048
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.9342276453971863,
      "learning_rate": 0.0004161344474354107,
      "loss": 2.1876,
      "step": 86049
    },
    {
      "epoch": 1.12,
      "grad_norm": 1.0523496866226196,
      "learning_rate": 0.0004161306757973834,
      "loss": 2.0408,
      "step": 86050
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.9246320128440857,
      "learning_rate": 0.0004161269041377653,
      "loss": 2.0818,
      "step": 86051
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.7510421872138977,
      "learning_rate": 0.000416123132456557,
      "loss": 2.0948,
      "step": 86052
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.868840217590332,
      "learning_rate": 0.00041611936075375923,
      "loss": 2.0872,
      "step": 86053
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.9335522651672363,
      "learning_rate": 0.0004161155890293726,
      "loss": 2.2065,
      "step": 86054
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.8295429944992065,
      "learning_rate": 0.000416111817283398,
      "loss": 2.0879,
      "step": 86055
    },
    {
      "epoch": 1.12,
      "grad_norm": 1.0310357809066772,
      "learning_rate": 0.00041610804551583594,
      "loss": 2.1785,
      "step": 86056
    },
    {
      "epoch": 1.12,
      "grad_norm": 1.088863730430603,
      "learning_rate": 0.0004161042737266872,
      "loss": 2.2369,
      "step": 86057
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.743990957736969,
      "learning_rate": 0.00041610050191595257,
      "loss": 2.264,
      "step": 86058
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.8005649447441101,
      "learning_rate": 0.00041609673008363257,
      "loss": 2.1049,
      "step": 86059
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.8492656946182251,
      "learning_rate": 0.000416092958229728,
      "loss": 2.2971,
      "step": 86060
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.8441958427429199,
      "learning_rate": 0.0004160891863542396,
      "loss": 2.261,
      "step": 86061
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.8120623230934143,
      "learning_rate": 0.00041608541445716803,
      "loss": 2.3194,
      "step": 86062
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.8369865417480469,
      "learning_rate": 0.000416081642538514,
      "loss": 1.9205,
      "step": 86063
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.8895432353019714,
      "learning_rate": 0.00041607787059827814,
      "loss": 2.2416,
      "step": 86064
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.8308078050613403,
      "learning_rate": 0.0004160740986364613,
      "loss": 2.1027,
      "step": 86065
    },
    {
      "epoch": 1.12,
      "grad_norm": 1.040664792060852,
      "learning_rate": 0.000416070326653064,
      "loss": 2.1684,
      "step": 86066
    },
    {
      "epoch": 1.12,
      "grad_norm": 1.0019268989562988,
      "learning_rate": 0.00041606655464808705,
      "loss": 2.2648,
      "step": 86067
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.873003363609314,
      "learning_rate": 0.00041606278262153114,
      "loss": 2.3821,
      "step": 86068
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.8674821257591248,
      "learning_rate": 0.00041605901057339707,
      "loss": 2.2709,
      "step": 86069
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.8837100863456726,
      "learning_rate": 0.00041605523850368536,
      "loss": 1.8527,
      "step": 86070
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.9751591682434082,
      "learning_rate": 0.00041605146641239673,
      "loss": 2.24,
      "step": 86071
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.914671778678894,
      "learning_rate": 0.0004160476942995321,
      "loss": 2.4158,
      "step": 86072
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.7972708344459534,
      "learning_rate": 0.0004160439221650919,
      "loss": 2.2543,
      "step": 86073
    },
    {
      "epoch": 1.12,
      "grad_norm": 1.0591073036193848,
      "learning_rate": 0.00041604015000907695,
      "loss": 1.9783,
      "step": 86074
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.8382067084312439,
      "learning_rate": 0.000416036377831488,
      "loss": 2.1184,
      "step": 86075
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.9096060395240784,
      "learning_rate": 0.0004160326056323256,
      "loss": 2.3742,
      "step": 86076
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.8446833491325378,
      "learning_rate": 0.00041602883341159064,
      "loss": 2.3111,
      "step": 86077
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.962598443031311,
      "learning_rate": 0.0004160250611692837,
      "loss": 2.2044,
      "step": 86078
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.8657174110412598,
      "learning_rate": 0.0004160212889054055,
      "loss": 2.4248,
      "step": 86079
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.8570507764816284,
      "learning_rate": 0.00041601751661995675,
      "loss": 2.1193,
      "step": 86080
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.9914554953575134,
      "learning_rate": 0.00041601374431293823,
      "loss": 2.1247,
      "step": 86081
    },
    {
      "epoch": 1.12,
      "grad_norm": 1.0947083234786987,
      "learning_rate": 0.00041600997198435044,
      "loss": 2.4995,
      "step": 86082
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.8159812092781067,
      "learning_rate": 0.0004160061996341943,
      "loss": 2.1684,
      "step": 86083
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.8814986944198608,
      "learning_rate": 0.0004160024272624704,
      "loss": 2.2475,
      "step": 86084
    },
    {
      "epoch": 1.12,
      "grad_norm": 1.085557460784912,
      "learning_rate": 0.0004159986548691794,
      "loss": 2.4174,
      "step": 86085
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.9200456738471985,
      "learning_rate": 0.00041599488245432206,
      "loss": 2.0792,
      "step": 86086
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.7677309513092041,
      "learning_rate": 0.0004159911100178992,
      "loss": 2.1706,
      "step": 86087
    },
    {
      "epoch": 1.12,
      "grad_norm": 1.2025678157806396,
      "learning_rate": 0.00041598733755991127,
      "loss": 1.9455,
      "step": 86088
    },
    {
      "epoch": 1.12,
      "grad_norm": 1.0515533685684204,
      "learning_rate": 0.00041598356508035916,
      "loss": 2.3207,
      "step": 86089
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.8663884401321411,
      "learning_rate": 0.0004159797925792435,
      "loss": 2.1006,
      "step": 86090
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.977724015712738,
      "learning_rate": 0.00041597602005656505,
      "loss": 2.0472,
      "step": 86091
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.9286338686943054,
      "learning_rate": 0.0004159722475123244,
      "loss": 2.1975,
      "step": 86092
    },
    {
      "epoch": 1.12,
      "grad_norm": 1.232668161392212,
      "learning_rate": 0.0004159684749465224,
      "loss": 2.1158,
      "step": 86093
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.9024965167045593,
      "learning_rate": 0.00041596470235915966,
      "loss": 2.5931,
      "step": 86094
    },
    {
      "epoch": 1.12,
      "grad_norm": 1.1605921983718872,
      "learning_rate": 0.00041596092975023686,
      "loss": 2.226,
      "step": 86095
    },
    {
      "epoch": 1.12,
      "grad_norm": 1.3952083587646484,
      "learning_rate": 0.00041595715711975476,
      "loss": 1.9807,
      "step": 86096
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.887394368648529,
      "learning_rate": 0.000415953384467714,
      "loss": 2.1835,
      "step": 86097
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.8879501223564148,
      "learning_rate": 0.0004159496117941153,
      "loss": 2.1807,
      "step": 86098
    },
    {
      "epoch": 1.12,
      "grad_norm": 1.3958643674850464,
      "learning_rate": 0.00041594583909895945,
      "loss": 2.271,
      "step": 86099
    },
    {
      "epoch": 1.12,
      "grad_norm": 1.477945327758789,
      "learning_rate": 0.000415942066382247,
      "loss": 2.3158,
      "step": 86100
    },
    {
      "epoch": 1.12,
      "grad_norm": 1.2065513134002686,
      "learning_rate": 0.0004159382936439788,
      "loss": 2.1692,
      "step": 86101
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.9750357866287231,
      "learning_rate": 0.0004159345208841555,
      "loss": 2.1625,
      "step": 86102
    },
    {
      "epoch": 1.12,
      "grad_norm": 1.2310303449630737,
      "learning_rate": 0.0004159307481027777,
      "loss": 2.0902,
      "step": 86103
    },
    {
      "epoch": 1.12,
      "grad_norm": 1.4656773805618286,
      "learning_rate": 0.00041592697529984626,
      "loss": 2.1056,
      "step": 86104
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.8925339579582214,
      "learning_rate": 0.0004159232024753618,
      "loss": 2.3892,
      "step": 86105
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.8019412159919739,
      "learning_rate": 0.00041591942962932495,
      "loss": 2.4888,
      "step": 86106
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.9252456426620483,
      "learning_rate": 0.00041591565676173657,
      "loss": 2.1865,
      "step": 86107
    },
    {
      "epoch": 1.12,
      "grad_norm": 1.081104040145874,
      "learning_rate": 0.0004159118838725972,
      "loss": 2.2504,
      "step": 86108
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.9986298680305481,
      "learning_rate": 0.00041590811096190773,
      "loss": 2.22,
      "step": 86109
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.8651350140571594,
      "learning_rate": 0.0004159043380296687,
      "loss": 2.041,
      "step": 86110
    },
    {
      "epoch": 1.12,
      "grad_norm": 1.0039517879486084,
      "learning_rate": 0.0004159005650758808,
      "loss": 2.1386,
      "step": 86111
    },
    {
      "epoch": 1.12,
      "grad_norm": 1.0678588151931763,
      "learning_rate": 0.0004158967921005449,
      "loss": 2.2142,
      "step": 86112
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.7496021389961243,
      "learning_rate": 0.00041589301910366155,
      "loss": 2.2149,
      "step": 86113
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.7797998785972595,
      "learning_rate": 0.0004158892460852315,
      "loss": 2.0717,
      "step": 86114
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.7742631435394287,
      "learning_rate": 0.00041588547304525553,
      "loss": 2.3499,
      "step": 86115
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.957983672618866,
      "learning_rate": 0.00041588169998373417,
      "loss": 2.149,
      "step": 86116
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.9525269269943237,
      "learning_rate": 0.0004158779269006682,
      "loss": 2.0542,
      "step": 86117
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.8872284293174744,
      "learning_rate": 0.00041587415379605835,
      "loss": 2.3805,
      "step": 86118
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.8863619565963745,
      "learning_rate": 0.0004158703806699054,
      "loss": 2.1776,
      "step": 86119
    },
    {
      "epoch": 1.12,
      "grad_norm": 1.0426656007766724,
      "learning_rate": 0.0004158666075222098,
      "loss": 2.0763,
      "step": 86120
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.8081396222114563,
      "learning_rate": 0.0004158628343529726,
      "loss": 2.1307,
      "step": 86121
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.7577883005142212,
      "learning_rate": 0.0004158590611621942,
      "loss": 2.3445,
      "step": 86122
    },
    {
      "epoch": 1.12,
      "grad_norm": 1.0115458965301514,
      "learning_rate": 0.0004158552879498754,
      "loss": 2.1852,
      "step": 86123
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.8242233991622925,
      "learning_rate": 0.000415851514716017,
      "loss": 2.2569,
      "step": 86124
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.9339253902435303,
      "learning_rate": 0.0004158477414606195,
      "loss": 2.1834,
      "step": 86125
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.8574214577674866,
      "learning_rate": 0.0004158439681836838,
      "loss": 2.1698,
      "step": 86126
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.8377748131752014,
      "learning_rate": 0.0004158401948852105,
      "loss": 2.1312,
      "step": 86127
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.835930585861206,
      "learning_rate": 0.00041583642156520033,
      "loss": 2.0898,
      "step": 86128
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.7991776466369629,
      "learning_rate": 0.00041583264822365394,
      "loss": 2.126,
      "step": 86129
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.8828072547912598,
      "learning_rate": 0.00041582887486057215,
      "loss": 2.2435,
      "step": 86130
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.8373233079910278,
      "learning_rate": 0.00041582510147595555,
      "loss": 2.1281,
      "step": 86131
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.884300947189331,
      "learning_rate": 0.0004158213280698048,
      "loss": 2.183,
      "step": 86132
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.8012047410011292,
      "learning_rate": 0.0004158175546421208,
      "loss": 2.1604,
      "step": 86133
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.9498389363288879,
      "learning_rate": 0.00041581378119290405,
      "loss": 2.0269,
      "step": 86134
    },
    {
      "epoch": 1.12,
      "grad_norm": 1.070382833480835,
      "learning_rate": 0.0004158100077221554,
      "loss": 2.2168,
      "step": 86135
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.7743593454360962,
      "learning_rate": 0.00041580623422987547,
      "loss": 2.3424,
      "step": 86136
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.825996994972229,
      "learning_rate": 0.0004158024607160649,
      "loss": 2.2264,
      "step": 86137
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.8576443791389465,
      "learning_rate": 0.00041579868718072453,
      "loss": 2.3269,
      "step": 86138
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.8384290933609009,
      "learning_rate": 0.000415794913623855,
      "loss": 2.3149,
      "step": 86139
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.7727408409118652,
      "learning_rate": 0.000415791140045457,
      "loss": 2.2654,
      "step": 86140
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.8511572480201721,
      "learning_rate": 0.0004157873664455312,
      "loss": 2.1422,
      "step": 86141
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.9357905387878418,
      "learning_rate": 0.0004157835928240784,
      "loss": 2.2772,
      "step": 86142
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.8587132692337036,
      "learning_rate": 0.0004157798191810992,
      "loss": 2.1608,
      "step": 86143
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.7977409362792969,
      "learning_rate": 0.0004157760455165943,
      "loss": 2.2702,
      "step": 86144
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.8320411443710327,
      "learning_rate": 0.00041577227183056453,
      "loss": 2.3654,
      "step": 86145
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.7956923842430115,
      "learning_rate": 0.00041576849812301057,
      "loss": 2.2312,
      "step": 86146
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.8861798048019409,
      "learning_rate": 0.0004157647243939329,
      "loss": 2.235,
      "step": 86147
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.7435941696166992,
      "learning_rate": 0.00041576095064333254,
      "loss": 2.0764,
      "step": 86148
    },
    {
      "epoch": 1.12,
      "grad_norm": 1.0977139472961426,
      "learning_rate": 0.00041575717687120995,
      "loss": 2.1367,
      "step": 86149
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.7739378213882446,
      "learning_rate": 0.00041575340307756594,
      "loss": 2.1211,
      "step": 86150
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.8244059681892395,
      "learning_rate": 0.00041574962926240117,
      "loss": 2.1762,
      "step": 86151
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.9261874556541443,
      "learning_rate": 0.0004157458554257163,
      "loss": 2.2477,
      "step": 86152
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.9042986035346985,
      "learning_rate": 0.00041574208156751216,
      "loss": 2.236,
      "step": 86153
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.8582239747047424,
      "learning_rate": 0.00041573830768778944,
      "loss": 2.3193,
      "step": 86154
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.8654776215553284,
      "learning_rate": 0.00041573453378654867,
      "loss": 2.0637,
      "step": 86155
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.8583284616470337,
      "learning_rate": 0.0004157307598637907,
      "loss": 2.1704,
      "step": 86156
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.7465090155601501,
      "learning_rate": 0.00041572698591951627,
      "loss": 2.1795,
      "step": 86157
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.800446093082428,
      "learning_rate": 0.00041572321195372595,
      "loss": 2.1543,
      "step": 86158
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.7538399696350098,
      "learning_rate": 0.00041571943796642045,
      "loss": 2.1967,
      "step": 86159
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.86359041929245,
      "learning_rate": 0.0004157156639576006,
      "loss": 2.3414,
      "step": 86160
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.7521259784698486,
      "learning_rate": 0.000415711889927267,
      "loss": 2.128,
      "step": 86161
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.8029762506484985,
      "learning_rate": 0.0004157081158754204,
      "loss": 2.0588,
      "step": 86162
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.9331990480422974,
      "learning_rate": 0.0004157043418020615,
      "loss": 2.3035,
      "step": 86163
    },
    {
      "epoch": 1.12,
      "grad_norm": 1.0197287797927856,
      "learning_rate": 0.00041570056770719085,
      "loss": 2.1458,
      "step": 86164
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.8087978363037109,
      "learning_rate": 0.00041569679359080934,
      "loss": 2.256,
      "step": 86165
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.9559056162834167,
      "learning_rate": 0.0004156930194529177,
      "loss": 2.3563,
      "step": 86166
    },
    {
      "epoch": 1.12,
      "grad_norm": 1.069083333015442,
      "learning_rate": 0.0004156892452935164,
      "loss": 2.3735,
      "step": 86167
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.8175476789474487,
      "learning_rate": 0.0004156854711126064,
      "loss": 2.0935,
      "step": 86168
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.8147615194320679,
      "learning_rate": 0.00041568169691018824,
      "loss": 2.2047,
      "step": 86169
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.8836294412612915,
      "learning_rate": 0.0004156779226862627,
      "loss": 2.1784,
      "step": 86170
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.9286510348320007,
      "learning_rate": 0.0004156741484408304,
      "loss": 2.0967,
      "step": 86171
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.8451040387153625,
      "learning_rate": 0.0004156703741738922,
      "loss": 2.2535,
      "step": 86172
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.8595227003097534,
      "learning_rate": 0.0004156665998854486,
      "loss": 2.2235,
      "step": 86173
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.8941690921783447,
      "learning_rate": 0.0004156628255755004,
      "loss": 2.2422,
      "step": 86174
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.8589820265769958,
      "learning_rate": 0.00041565905124404836,
      "loss": 2.2099,
      "step": 86175
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.8338525891304016,
      "learning_rate": 0.00041565527689109297,
      "loss": 2.0983,
      "step": 86176
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.8240160942077637,
      "learning_rate": 0.0004156515025166352,
      "loss": 2.1882,
      "step": 86177
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.8670088052749634,
      "learning_rate": 0.00041564772812067557,
      "loss": 2.3253,
      "step": 86178
    },
    {
      "epoch": 1.12,
      "grad_norm": 1.0349202156066895,
      "learning_rate": 0.00041564395370321497,
      "loss": 2.2049,
      "step": 86179
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.8772637248039246,
      "learning_rate": 0.0004156401792642539,
      "loss": 2.2725,
      "step": 86180
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.8280784487724304,
      "learning_rate": 0.000415636404803793,
      "loss": 2.1203,
      "step": 86181
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.9319197535514832,
      "learning_rate": 0.00041563263032183334,
      "loss": 2.3469,
      "step": 86182
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.8785219788551331,
      "learning_rate": 0.00041562885581837523,
      "loss": 2.0189,
      "step": 86183
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.8298817873001099,
      "learning_rate": 0.00041562508129341957,
      "loss": 2.1513,
      "step": 86184
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.8033084869384766,
      "learning_rate": 0.0004156213067469671,
      "loss": 2.0914,
      "step": 86185
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.7897724509239197,
      "learning_rate": 0.00041561753217901837,
      "loss": 2.2341,
      "step": 86186
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.9257453680038452,
      "learning_rate": 0.0004156137575895742,
      "loss": 2.1726,
      "step": 86187
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.8921458125114441,
      "learning_rate": 0.00041560998297863523,
      "loss": 2.4795,
      "step": 86188
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.9867387413978577,
      "learning_rate": 0.0004156062083462022,
      "loss": 2.2955,
      "step": 86189
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.8274372816085815,
      "learning_rate": 0.0004156024336922757,
      "loss": 2.1587,
      "step": 86190
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.8801432847976685,
      "learning_rate": 0.0004155986590168567,
      "loss": 2.066,
      "step": 86191
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.8707197904586792,
      "learning_rate": 0.00041559488431994565,
      "loss": 2.3821,
      "step": 86192
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.818587601184845,
      "learning_rate": 0.00041559110960154327,
      "loss": 2.1348,
      "step": 86193
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.7941927313804626,
      "learning_rate": 0.00041558733486165043,
      "loss": 2.1407,
      "step": 86194
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.8859934210777283,
      "learning_rate": 0.0004155835601002676,
      "loss": 2.0594,
      "step": 86195
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.8604840636253357,
      "learning_rate": 0.00041557978531739565,
      "loss": 2.2808,
      "step": 86196
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.8694303035736084,
      "learning_rate": 0.00041557601051303523,
      "loss": 2.2158,
      "step": 86197
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.785959005355835,
      "learning_rate": 0.00041557223568718706,
      "loss": 2.1715,
      "step": 86198
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.8646513223648071,
      "learning_rate": 0.00041556846083985185,
      "loss": 2.3398,
      "step": 86199
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.7535086870193481,
      "learning_rate": 0.0004155646859710303,
      "loss": 2.2057,
      "step": 86200
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.9447054266929626,
      "learning_rate": 0.000415560911080723,
      "loss": 2.2526,
      "step": 86201
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.8070095777511597,
      "learning_rate": 0.0004155571361689308,
      "loss": 1.9757,
      "step": 86202
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.8712990880012512,
      "learning_rate": 0.00041555336123565444,
      "loss": 2.15,
      "step": 86203
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.9243515133857727,
      "learning_rate": 0.0004155495862808944,
      "loss": 2.3311,
      "step": 86204
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.8229740262031555,
      "learning_rate": 0.0004155458113046515,
      "loss": 2.0982,
      "step": 86205
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.8228906989097595,
      "learning_rate": 0.00041554203630692657,
      "loss": 2.3063,
      "step": 86206
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.8060668110847473,
      "learning_rate": 0.00041553826128772005,
      "loss": 2.0058,
      "step": 86207
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.9177583456039429,
      "learning_rate": 0.0004155344862470329,
      "loss": 2.1189,
      "step": 86208
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.849696695804596,
      "learning_rate": 0.0004155307111848657,
      "loss": 2.2223,
      "step": 86209
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.8510748744010925,
      "learning_rate": 0.0004155269361012191,
      "loss": 2.187,
      "step": 86210
    },
    {
      "epoch": 1.12,
      "grad_norm": 1.1292154788970947,
      "learning_rate": 0.00041552316099609384,
      "loss": 2.2475,
      "step": 86211
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.822365939617157,
      "learning_rate": 0.0004155193858694908,
      "loss": 2.1737,
      "step": 86212
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.8573095798492432,
      "learning_rate": 0.0004155156107214104,
      "loss": 2.4398,
      "step": 86213
    },
    {
      "epoch": 1.12,
      "grad_norm": 1.0759246349334717,
      "learning_rate": 0.00041551183555185343,
      "loss": 2.1707,
      "step": 86214
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.8919103145599365,
      "learning_rate": 0.00041550806036082076,
      "loss": 2.2667,
      "step": 86215
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.7720006108283997,
      "learning_rate": 0.00041550428514831286,
      "loss": 2.0882,
      "step": 86216
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.9585869312286377,
      "learning_rate": 0.0004155005099143305,
      "loss": 2.1634,
      "step": 86217
    },
    {
      "epoch": 1.12,
      "grad_norm": 1.107539415359497,
      "learning_rate": 0.00041549673465887454,
      "loss": 2.1814,
      "step": 86218
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.8779703974723816,
      "learning_rate": 0.00041549295938194547,
      "loss": 2.0367,
      "step": 86219
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.9363670349121094,
      "learning_rate": 0.0004154891840835441,
      "loss": 2.2739,
      "step": 86220
    },
    {
      "epoch": 1.12,
      "grad_norm": 1.4813681840896606,
      "learning_rate": 0.00041548540876367113,
      "loss": 2.3698,
      "step": 86221
    },
    {
      "epoch": 1.12,
      "grad_norm": 1.1179300546646118,
      "learning_rate": 0.0004154816334223272,
      "loss": 2.2245,
      "step": 86222
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.8358333706855774,
      "learning_rate": 0.00041547785805951307,
      "loss": 2.0741,
      "step": 86223
    },
    {
      "epoch": 1.12,
      "grad_norm": 1.0401864051818848,
      "learning_rate": 0.00041547408267522944,
      "loss": 2.1307,
      "step": 86224
    },
    {
      "epoch": 1.12,
      "grad_norm": 1.3304287195205688,
      "learning_rate": 0.00041547030726947704,
      "loss": 2.1418,
      "step": 86225
    },
    {
      "epoch": 1.12,
      "grad_norm": 1.0560057163238525,
      "learning_rate": 0.00041546653184225646,
      "loss": 2.4194,
      "step": 86226
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.8204270601272583,
      "learning_rate": 0.00041546275639356856,
      "loss": 2.2575,
      "step": 86227
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.8733688592910767,
      "learning_rate": 0.00041545898092341384,
      "loss": 2.3126,
      "step": 86228
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.9957025647163391,
      "learning_rate": 0.00041545520543179317,
      "loss": 2.3484,
      "step": 86229
    },
    {
      "epoch": 1.12,
      "grad_norm": 1.0035195350646973,
      "learning_rate": 0.00041545142991870714,
      "loss": 2.2108,
      "step": 86230
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.8483208417892456,
      "learning_rate": 0.0004154476543841566,
      "loss": 2.1716,
      "step": 86231
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.9190934300422668,
      "learning_rate": 0.00041544387882814214,
      "loss": 1.9817,
      "step": 86232
    },
    {
      "epoch": 1.12,
      "grad_norm": 1.0107148885726929,
      "learning_rate": 0.00041544010325066447,
      "loss": 2.2909,
      "step": 86233
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.9574170112609863,
      "learning_rate": 0.0004154363276517243,
      "loss": 2.1702,
      "step": 86234
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.8260243535041809,
      "learning_rate": 0.0004154325520313224,
      "loss": 2.1311,
      "step": 86235
    },
    {
      "epoch": 1.12,
      "grad_norm": 1.0005933046340942,
      "learning_rate": 0.0004154287763894593,
      "loss": 2.2646,
      "step": 86236
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.9371345043182373,
      "learning_rate": 0.0004154250007261358,
      "loss": 2.2495,
      "step": 86237
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.827869176864624,
      "learning_rate": 0.00041542122504135276,
      "loss": 2.1632,
      "step": 86238
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.9342823624610901,
      "learning_rate": 0.00041541744933511063,
      "loss": 2.0492,
      "step": 86239
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.9455348253250122,
      "learning_rate": 0.00041541367360741025,
      "loss": 2.2112,
      "step": 86240
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.7608792781829834,
      "learning_rate": 0.00041540989785825226,
      "loss": 2.083,
      "step": 86241
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.7938637137413025,
      "learning_rate": 0.0004154061220876374,
      "loss": 2.0417,
      "step": 86242
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.9178848266601562,
      "learning_rate": 0.00041540234629556643,
      "loss": 2.0539,
      "step": 86243
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.7963862419128418,
      "learning_rate": 0.0004153985704820399,
      "loss": 2.3389,
      "step": 86244
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.888748049736023,
      "learning_rate": 0.0004153947946470586,
      "loss": 2.2145,
      "step": 86245
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.801934540271759,
      "learning_rate": 0.00041539101879062333,
      "loss": 2.2381,
      "step": 86246
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.8855271339416504,
      "learning_rate": 0.00041538724291273455,
      "loss": 2.2386,
      "step": 86247
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.7654328942298889,
      "learning_rate": 0.00041538346701339314,
      "loss": 2.1287,
      "step": 86248
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.8082321286201477,
      "learning_rate": 0.0004153796910925999,
      "loss": 2.2176,
      "step": 86249
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.7813230752944946,
      "learning_rate": 0.00041537591515035524,
      "loss": 2.0501,
      "step": 86250
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.873885989189148,
      "learning_rate": 0.00041537213918666003,
      "loss": 2.1668,
      "step": 86251
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.8772631883621216,
      "learning_rate": 0.0004153683632015151,
      "loss": 2.0878,
      "step": 86252
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.8742188811302185,
      "learning_rate": 0.0004153645871949209,
      "loss": 2.2336,
      "step": 86253
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.7973393797874451,
      "learning_rate": 0.0004153608111668782,
      "loss": 2.1875,
      "step": 86254
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.7454960346221924,
      "learning_rate": 0.0004153570351173879,
      "loss": 2.1044,
      "step": 86255
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.8539864420890808,
      "learning_rate": 0.00041535325904645045,
      "loss": 2.1068,
      "step": 86256
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.7891504168510437,
      "learning_rate": 0.0004153494829540666,
      "loss": 2.4411,
      "step": 86257
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.8037310838699341,
      "learning_rate": 0.00041534570684023724,
      "loss": 2.4127,
      "step": 86258
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.8537278771400452,
      "learning_rate": 0.0004153419307049628,
      "loss": 2.3303,
      "step": 86259
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.8329088091850281,
      "learning_rate": 0.00041533815454824417,
      "loss": 2.0647,
      "step": 86260
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.7826127409934998,
      "learning_rate": 0.00041533437837008207,
      "loss": 2.0331,
      "step": 86261
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.8084722757339478,
      "learning_rate": 0.00041533060217047705,
      "loss": 2.0755,
      "step": 86262
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.8276911377906799,
      "learning_rate": 0.0004153268259494299,
      "loss": 2.0534,
      "step": 86263
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.9512535929679871,
      "learning_rate": 0.00041532304970694136,
      "loss": 2.2032,
      "step": 86264
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.7548596262931824,
      "learning_rate": 0.0004153192734430121,
      "loss": 2.1318,
      "step": 86265
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.7898654341697693,
      "learning_rate": 0.0004153154971576427,
      "loss": 2.1542,
      "step": 86266
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.9058883190155029,
      "learning_rate": 0.00041531172085083407,
      "loss": 2.1179,
      "step": 86267
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.7954457998275757,
      "learning_rate": 0.0004153079445225868,
      "loss": 2.2786,
      "step": 86268
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.8424466848373413,
      "learning_rate": 0.00041530416817290157,
      "loss": 2.237,
      "step": 86269
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.9180063605308533,
      "learning_rate": 0.0004153003918017792,
      "loss": 2.2864,
      "step": 86270
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.8388320803642273,
      "learning_rate": 0.0004152966154092202,
      "loss": 2.4246,
      "step": 86271
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.8474509119987488,
      "learning_rate": 0.00041529283899522545,
      "loss": 2.2228,
      "step": 86272
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.8175588846206665,
      "learning_rate": 0.0004152890625597956,
      "loss": 2.357,
      "step": 86273
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.9230194687843323,
      "learning_rate": 0.0004152852861029313,
      "loss": 2.306,
      "step": 86274
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.8109374642372131,
      "learning_rate": 0.0004152815096246333,
      "loss": 2.2925,
      "step": 86275
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.809255838394165,
      "learning_rate": 0.0004152777331249023,
      "loss": 2.3386,
      "step": 86276
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.7837450504302979,
      "learning_rate": 0.00041527395660373895,
      "loss": 2.0963,
      "step": 86277
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.8298041224479675,
      "learning_rate": 0.00041527018006114403,
      "loss": 2.2495,
      "step": 86278
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.7608813047409058,
      "learning_rate": 0.0004152664034971183,
      "loss": 2.1437,
      "step": 86279
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.8896182775497437,
      "learning_rate": 0.00041526262691166224,
      "loss": 2.3514,
      "step": 86280
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.8186733722686768,
      "learning_rate": 0.0004152588503047767,
      "loss": 2.2234,
      "step": 86281
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.7984327673912048,
      "learning_rate": 0.0004152550736764624,
      "loss": 2.2236,
      "step": 86282
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.8632647395133972,
      "learning_rate": 0.00041525129702672,
      "loss": 2.1061,
      "step": 86283
    },
    {
      "epoch": 1.12,
      "grad_norm": 1.125846028327942,
      "learning_rate": 0.00041524752035555014,
      "loss": 1.9228,
      "step": 86284
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.8636299967765808,
      "learning_rate": 0.0004152437436629537,
      "loss": 2.1023,
      "step": 86285
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.8098549246788025,
      "learning_rate": 0.00041523996694893127,
      "loss": 2.1559,
      "step": 86286
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.9489747285842896,
      "learning_rate": 0.0004152361902134834,
      "loss": 2.2129,
      "step": 86287
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.7267194986343384,
      "learning_rate": 0.0004152324134566111,
      "loss": 2.1274,
      "step": 86288
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.876782238483429,
      "learning_rate": 0.00041522863667831485,
      "loss": 2.1787,
      "step": 86289
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.8925295472145081,
      "learning_rate": 0.00041522485987859547,
      "loss": 2.119,
      "step": 86290
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.850225031375885,
      "learning_rate": 0.00041522108305745356,
      "loss": 2.1878,
      "step": 86291
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.810295581817627,
      "learning_rate": 0.0004152173062148899,
      "loss": 2.2623,
      "step": 86292
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.9607701301574707,
      "learning_rate": 0.0004152135293509052,
      "loss": 2.3144,
      "step": 86293
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.859552264213562,
      "learning_rate": 0.0004152097524655001,
      "loss": 2.1508,
      "step": 86294
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.8363729119300842,
      "learning_rate": 0.0004152059755586753,
      "loss": 2.2468,
      "step": 86295
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.8748681545257568,
      "learning_rate": 0.0004152021986304316,
      "loss": 2.2105,
      "step": 86296
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.872799813747406,
      "learning_rate": 0.0004151984216807696,
      "loss": 2.0198,
      "step": 86297
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.8178871870040894,
      "learning_rate": 0.00041519464470969006,
      "loss": 2.02,
      "step": 86298
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.8003613352775574,
      "learning_rate": 0.00041519086771719363,
      "loss": 2.1103,
      "step": 86299
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.820898711681366,
      "learning_rate": 0.00041518709070328106,
      "loss": 2.0703,
      "step": 86300
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.7981457710266113,
      "learning_rate": 0.000415183313667953,
      "loss": 2.1436,
      "step": 86301
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.867060661315918,
      "learning_rate": 0.00041517953661121017,
      "loss": 2.4155,
      "step": 86302
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.8523908257484436,
      "learning_rate": 0.0004151757595330534,
      "loss": 2.3972,
      "step": 86303
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.781621515750885,
      "learning_rate": 0.00041517198243348324,
      "loss": 2.1615,
      "step": 86304
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.8423941731452942,
      "learning_rate": 0.0004151682053125004,
      "loss": 1.9993,
      "step": 86305
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.8850606679916382,
      "learning_rate": 0.0004151644281701057,
      "loss": 2.1481,
      "step": 86306
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.9382720589637756,
      "learning_rate": 0.0004151606510062997,
      "loss": 2.287,
      "step": 86307
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.8170416355133057,
      "learning_rate": 0.00041515687382108306,
      "loss": 2.1386,
      "step": 86308
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.8913933634757996,
      "learning_rate": 0.00041515309661445675,
      "loss": 2.2158,
      "step": 86309
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.8940561413764954,
      "learning_rate": 0.0004151493193864212,
      "loss": 2.2086,
      "step": 86310
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.8005601763725281,
      "learning_rate": 0.0004151455421369772,
      "loss": 2.1769,
      "step": 86311
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.8340774178504944,
      "learning_rate": 0.0004151417648661255,
      "loss": 2.0679,
      "step": 86312
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.7475112080574036,
      "learning_rate": 0.00041513798757386686,
      "loss": 2.0643,
      "step": 86313
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.8531250357627869,
      "learning_rate": 0.00041513421026020183,
      "loss": 2.2291,
      "step": 86314
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.8163533806800842,
      "learning_rate": 0.0004151304329251312,
      "loss": 2.1776,
      "step": 86315
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.848433792591095,
      "learning_rate": 0.0004151266555686556,
      "loss": 2.2363,
      "step": 86316
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.9820835590362549,
      "learning_rate": 0.00041512287819077577,
      "loss": 2.3581,
      "step": 86317
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.9440320134162903,
      "learning_rate": 0.0004151191007914925,
      "loss": 2.24,
      "step": 86318
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.8125438094139099,
      "learning_rate": 0.0004151153233708064,
      "loss": 2.0094,
      "step": 86319
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.8890895843505859,
      "learning_rate": 0.0004151115459287181,
      "loss": 2.332,
      "step": 86320
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.858238160610199,
      "learning_rate": 0.0004151077684652284,
      "loss": 2.2526,
      "step": 86321
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.8550231456756592,
      "learning_rate": 0.0004151039909803381,
      "loss": 2.2578,
      "step": 86322
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.810133695602417,
      "learning_rate": 0.0004151002134740478,
      "loss": 2.101,
      "step": 86323
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.8843991756439209,
      "learning_rate": 0.0004150964359463581,
      "loss": 2.2389,
      "step": 86324
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.8358837366104126,
      "learning_rate": 0.0004150926583972698,
      "loss": 2.1912,
      "step": 86325
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.8163949251174927,
      "learning_rate": 0.0004150888808267836,
      "loss": 2.3319,
      "step": 86326
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.894787073135376,
      "learning_rate": 0.0004150851032349003,
      "loss": 2.1794,
      "step": 86327
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.8420936465263367,
      "learning_rate": 0.00041508132562162047,
      "loss": 2.2191,
      "step": 86328
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.9613562226295471,
      "learning_rate": 0.0004150775479869448,
      "loss": 2.3758,
      "step": 86329
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.9227580428123474,
      "learning_rate": 0.00041507377033087414,
      "loss": 2.251,
      "step": 86330
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.8932260274887085,
      "learning_rate": 0.000415069992653409,
      "loss": 2.2213,
      "step": 86331
    },
    {
      "epoch": 1.12,
      "grad_norm": 1.0266191959381104,
      "learning_rate": 0.00041506621495455016,
      "loss": 2.1559,
      "step": 86332
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.8266319036483765,
      "learning_rate": 0.0004150624372342984,
      "loss": 2.2925,
      "step": 86333
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.826066255569458,
      "learning_rate": 0.00041505865949265433,
      "loss": 2.2219,
      "step": 86334
    },
    {
      "epoch": 1.12,
      "grad_norm": 1.0084667205810547,
      "learning_rate": 0.00041505488172961867,
      "loss": 2.5221,
      "step": 86335
    },
    {
      "epoch": 1.12,
      "grad_norm": 1.20466148853302,
      "learning_rate": 0.0004150511039451922,
      "loss": 2.2471,
      "step": 86336
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.8699058294296265,
      "learning_rate": 0.0004150473261393756,
      "loss": 2.1756,
      "step": 86337
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.8086903095245361,
      "learning_rate": 0.00041504354831216935,
      "loss": 2.3523,
      "step": 86338
    },
    {
      "epoch": 1.12,
      "grad_norm": 1.2555500268936157,
      "learning_rate": 0.00041503977046357447,
      "loss": 2.4571,
      "step": 86339
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.9566100835800171,
      "learning_rate": 0.0004150359925935915,
      "loss": 2.0254,
      "step": 86340
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.9758743047714233,
      "learning_rate": 0.0004150322147022212,
      "loss": 2.3327,
      "step": 86341
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.9455165863037109,
      "learning_rate": 0.00041502843678946415,
      "loss": 2.3931,
      "step": 86342
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.8482984900474548,
      "learning_rate": 0.0004150246588553212,
      "loss": 2.2595,
      "step": 86343
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.8245747089385986,
      "learning_rate": 0.000415020880899793,
      "loss": 2.4306,
      "step": 86344
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.7864697575569153,
      "learning_rate": 0.0004150171029228801,
      "loss": 2.1075,
      "step": 86345
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.9588726162910461,
      "learning_rate": 0.0004150133249245836,
      "loss": 2.2458,
      "step": 86346
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.7858366966247559,
      "learning_rate": 0.00041500954690490383,
      "loss": 2.0688,
      "step": 86347
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.8120269179344177,
      "learning_rate": 0.0004150057688638416,
      "loss": 2.1234,
      "step": 86348
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.8745869398117065,
      "learning_rate": 0.0004150019908013977,
      "loss": 2.0835,
      "step": 86349
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.8557159304618835,
      "learning_rate": 0.0004149982127175727,
      "loss": 2.0658,
      "step": 86350
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.801937460899353,
      "learning_rate": 0.0004149944346123673,
      "loss": 2.1249,
      "step": 86351
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.7841147184371948,
      "learning_rate": 0.00041499065648578236,
      "loss": 2.109,
      "step": 86352
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.8172738552093506,
      "learning_rate": 0.0004149868783378185,
      "loss": 2.1684,
      "step": 86353
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.8680450320243835,
      "learning_rate": 0.0004149831001684763,
      "loss": 1.8667,
      "step": 86354
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.8550686240196228,
      "learning_rate": 0.0004149793219777567,
      "loss": 2.0191,
      "step": 86355
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.845426082611084,
      "learning_rate": 0.00041497554376566015,
      "loss": 2.2122,
      "step": 86356
    },
    {
      "epoch": 1.12,
      "grad_norm": 1.0706108808517456,
      "learning_rate": 0.00041497176553218754,
      "loss": 2.1703,
      "step": 86357
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.7914355397224426,
      "learning_rate": 0.00041496798727733955,
      "loss": 2.1905,
      "step": 86358
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.8487394452095032,
      "learning_rate": 0.00041496420900111674,
      "loss": 2.1155,
      "step": 86359
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.8588463664054871,
      "learning_rate": 0.00041496043070351997,
      "loss": 2.4068,
      "step": 86360
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.7795570492744446,
      "learning_rate": 0.00041495665238454994,
      "loss": 2.1506,
      "step": 86361
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.8071375489234924,
      "learning_rate": 0.0004149528740442072,
      "loss": 2.0784,
      "step": 86362
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.8261392116546631,
      "learning_rate": 0.00041494909568249254,
      "loss": 2.031,
      "step": 86363
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.8066617846488953,
      "learning_rate": 0.0004149453172994068,
      "loss": 2.4111,
      "step": 86364
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.8304178714752197,
      "learning_rate": 0.0004149415388949505,
      "loss": 2.4,
      "step": 86365
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.7565460801124573,
      "learning_rate": 0.0004149377604691243,
      "loss": 2.1368,
      "step": 86366
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.8784148693084717,
      "learning_rate": 0.00041493398202192914,
      "loss": 2.0798,
      "step": 86367
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.8248911499977112,
      "learning_rate": 0.0004149302035533655,
      "loss": 2.4035,
      "step": 86368
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.9093681573867798,
      "learning_rate": 0.0004149264250634342,
      "loss": 2.2595,
      "step": 86369
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.8766239881515503,
      "learning_rate": 0.00041492264655213596,
      "loss": 2.1763,
      "step": 86370
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.7606671452522278,
      "learning_rate": 0.0004149188680194713,
      "loss": 2.3453,
      "step": 86371
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.9369733929634094,
      "learning_rate": 0.0004149150894654411,
      "loss": 2.2819,
      "step": 86372
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.8952853679656982,
      "learning_rate": 0.0004149113108900461,
      "loss": 2.2493,
      "step": 86373
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.9494975209236145,
      "learning_rate": 0.0004149075322932868,
      "loss": 2.3299,
      "step": 86374
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.7485358715057373,
      "learning_rate": 0.00041490375367516405,
      "loss": 2.238,
      "step": 86375
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.8789442181587219,
      "learning_rate": 0.00041489997503567865,
      "loss": 2.0443,
      "step": 86376
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.9012105464935303,
      "learning_rate": 0.000414896196374831,
      "loss": 2.0575,
      "step": 86377
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.854022204875946,
      "learning_rate": 0.00041489241769262207,
      "loss": 2.1598,
      "step": 86378
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.7842133045196533,
      "learning_rate": 0.0004148886389890524,
      "loss": 2.1872,
      "step": 86379
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.8663541674613953,
      "learning_rate": 0.00041488486026412287,
      "loss": 2.2521,
      "step": 86380
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.8071783185005188,
      "learning_rate": 0.000414881081517834,
      "loss": 2.2387,
      "step": 86381
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.9055576920509338,
      "learning_rate": 0.0004148773027501866,
      "loss": 2.4133,
      "step": 86382
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.7847916483879089,
      "learning_rate": 0.00041487352396118136,
      "loss": 2.0552,
      "step": 86383
    },
    {
      "epoch": 1.12,
      "grad_norm": 1.1267035007476807,
      "learning_rate": 0.0004148697451508189,
      "loss": 2.0247,
      "step": 86384
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.9005297422409058,
      "learning_rate": 0.00041486596631910003,
      "loss": 2.2099,
      "step": 86385
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.854306697845459,
      "learning_rate": 0.0004148621874660254,
      "loss": 2.0238,
      "step": 86386
    },
    {
      "epoch": 1.12,
      "grad_norm": 1.024707555770874,
      "learning_rate": 0.0004148584085915957,
      "loss": 2.0258,
      "step": 86387
    },
    {
      "epoch": 1.12,
      "grad_norm": 1.0715137720108032,
      "learning_rate": 0.0004148546296958117,
      "loss": 2.0465,
      "step": 86388
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.8405140042304993,
      "learning_rate": 0.0004148508507786741,
      "loss": 2.1273,
      "step": 86389
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.9078349471092224,
      "learning_rate": 0.0004148470718401834,
      "loss": 2.1944,
      "step": 86390
    },
    {
      "epoch": 1.12,
      "grad_norm": 1.4576038122177124,
      "learning_rate": 0.00041484329288034056,
      "loss": 2.2769,
      "step": 86391
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.7988701462745667,
      "learning_rate": 0.0004148395138991462,
      "loss": 2.2787,
      "step": 86392
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.766255795955658,
      "learning_rate": 0.000414835734896601,
      "loss": 2.0572,
      "step": 86393
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.9447593092918396,
      "learning_rate": 0.0004148319558727056,
      "loss": 1.9703,
      "step": 86394
    },
    {
      "epoch": 1.12,
      "grad_norm": 1.0480910539627075,
      "learning_rate": 0.00041482817682746083,
      "loss": 2.1391,
      "step": 86395
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.9083760380744934,
      "learning_rate": 0.00041482439776086737,
      "loss": 2.0337,
      "step": 86396
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.864105224609375,
      "learning_rate": 0.0004148206186729258,
      "loss": 2.3434,
      "step": 86397
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.8809439539909363,
      "learning_rate": 0.000414816839563637,
      "loss": 2.0943,
      "step": 86398
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.830741286277771,
      "learning_rate": 0.0004148130604330015,
      "loss": 2.1307,
      "step": 86399
    },
    {
      "epoch": 1.12,
      "grad_norm": 1.1522395610809326,
      "learning_rate": 0.0004148092812810201,
      "loss": 1.8604,
      "step": 86400
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.7922068238258362,
      "learning_rate": 0.00041480550210769355,
      "loss": 1.9761,
      "step": 86401
    },
    {
      "epoch": 1.13,
      "grad_norm": 1.0219863653182983,
      "learning_rate": 0.00041480172291302243,
      "loss": 2.2561,
      "step": 86402
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.9239467978477478,
      "learning_rate": 0.0004147979436970076,
      "loss": 2.3087,
      "step": 86403
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.7772565484046936,
      "learning_rate": 0.0004147941644596496,
      "loss": 2.1934,
      "step": 86404
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.8273065090179443,
      "learning_rate": 0.0004147903852009491,
      "loss": 2.1005,
      "step": 86405
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.8621827363967896,
      "learning_rate": 0.000414786605920907,
      "loss": 2.0729,
      "step": 86406
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.8900803327560425,
      "learning_rate": 0.0004147828266195239,
      "loss": 2.2728,
      "step": 86407
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.9022611379623413,
      "learning_rate": 0.0004147790472968004,
      "loss": 2.1337,
      "step": 86408
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.8144363760948181,
      "learning_rate": 0.00041477526795273746,
      "loss": 2.1704,
      "step": 86409
    },
    {
      "epoch": 1.13,
      "grad_norm": 1.1633411645889282,
      "learning_rate": 0.00041477148858733555,
      "loss": 2.0919,
      "step": 86410
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.9128066301345825,
      "learning_rate": 0.00041476770920059557,
      "loss": 2.2118,
      "step": 86411
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.8764550089836121,
      "learning_rate": 0.000414763929792518,
      "loss": 2.1573,
      "step": 86412
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.8187872171401978,
      "learning_rate": 0.0004147601503631036,
      "loss": 2.1185,
      "step": 86413
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.8044338226318359,
      "learning_rate": 0.0004147563709123532,
      "loss": 2.2382,
      "step": 86414
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.9939023852348328,
      "learning_rate": 0.0004147525914402674,
      "loss": 2.2151,
      "step": 86415
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.8268337845802307,
      "learning_rate": 0.00041474881194684694,
      "loss": 2.2864,
      "step": 86416
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.9695691466331482,
      "learning_rate": 0.0004147450324320925,
      "loss": 2.2308,
      "step": 86417
    },
    {
      "epoch": 1.13,
      "grad_norm": 2.0075623989105225,
      "learning_rate": 0.0004147412528960048,
      "loss": 2.2159,
      "step": 86418
    },
    {
      "epoch": 1.13,
      "grad_norm": 1.043677568435669,
      "learning_rate": 0.00041473747333858457,
      "loss": 2.2924,
      "step": 86419
    },
    {
      "epoch": 1.13,
      "grad_norm": 1.07798433303833,
      "learning_rate": 0.0004147336937598324,
      "loss": 2.239,
      "step": 86420
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.9377069473266602,
      "learning_rate": 0.00041472991415974906,
      "loss": 2.3998,
      "step": 86421
    },
    {
      "epoch": 1.13,
      "grad_norm": 1.0785350799560547,
      "learning_rate": 0.00041472613453833536,
      "loss": 2.0601,
      "step": 86422
    },
    {
      "epoch": 1.13,
      "grad_norm": 1.1737638711929321,
      "learning_rate": 0.00041472235489559183,
      "loss": 2.1381,
      "step": 86423
    },
    {
      "epoch": 1.13,
      "grad_norm": 1.0370630025863647,
      "learning_rate": 0.00041471857523151933,
      "loss": 2.2344,
      "step": 86424
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.7945494651794434,
      "learning_rate": 0.0004147147955461184,
      "loss": 2.4567,
      "step": 86425
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.8853356242179871,
      "learning_rate": 0.00041471101583938987,
      "loss": 2.2207,
      "step": 86426
    },
    {
      "epoch": 1.13,
      "grad_norm": 1.3742330074310303,
      "learning_rate": 0.00041470723611133437,
      "loss": 2.2251,
      "step": 86427
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.7780700922012329,
      "learning_rate": 0.0004147034563619527,
      "loss": 2.1799,
      "step": 86428
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.9510746002197266,
      "learning_rate": 0.0004146996765912454,
      "loss": 2.2124,
      "step": 86429
    },
    {
      "epoch": 1.13,
      "grad_norm": 1.1160385608673096,
      "learning_rate": 0.00041469589679921326,
      "loss": 2.2785,
      "step": 86430
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.7873579859733582,
      "learning_rate": 0.00041469211698585707,
      "loss": 2.2208,
      "step": 86431
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.7988722920417786,
      "learning_rate": 0.00041468833715117744,
      "loss": 2.3197,
      "step": 86432
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.8075559735298157,
      "learning_rate": 0.000414684557295175,
      "loss": 2.1892,
      "step": 86433
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.8130846619606018,
      "learning_rate": 0.00041468077741785064,
      "loss": 2.3584,
      "step": 86434
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.7642278075218201,
      "learning_rate": 0.0004146769975192049,
      "loss": 1.8931,
      "step": 86435
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.7451494932174683,
      "learning_rate": 0.0004146732175992385,
      "loss": 2.0098,
      "step": 86436
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.8923115730285645,
      "learning_rate": 0.00041466943765795225,
      "loss": 2.0895,
      "step": 86437
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.7846377491950989,
      "learning_rate": 0.00041466565769534674,
      "loss": 2.2362,
      "step": 86438
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.9354414343833923,
      "learning_rate": 0.0004146618777114227,
      "loss": 2.2382,
      "step": 86439
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.812565267086029,
      "learning_rate": 0.00041465809770618097,
      "loss": 1.9373,
      "step": 86440
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.8692739605903625,
      "learning_rate": 0.0004146543176796221,
      "loss": 2.0902,
      "step": 86441
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.852085292339325,
      "learning_rate": 0.00041465053763174676,
      "loss": 2.2285,
      "step": 86442
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.8111055493354797,
      "learning_rate": 0.0004146467575625557,
      "loss": 1.9481,
      "step": 86443
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.8516535758972168,
      "learning_rate": 0.00041464297747204976,
      "loss": 2.2804,
      "step": 86444
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.7848683595657349,
      "learning_rate": 0.00041463919736022946,
      "loss": 2.3708,
      "step": 86445
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.8028718829154968,
      "learning_rate": 0.0004146354172270955,
      "loss": 1.9711,
      "step": 86446
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.9111617803573608,
      "learning_rate": 0.00041463163707264886,
      "loss": 2.1757,
      "step": 86447
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.8448842763900757,
      "learning_rate": 0.0004146278568968898,
      "loss": 2.2615,
      "step": 86448
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.9501234292984009,
      "learning_rate": 0.0004146240766998194,
      "loss": 2.1193,
      "step": 86449
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.8995000720024109,
      "learning_rate": 0.0004146202964814382,
      "loss": 1.9898,
      "step": 86450
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.8616811633110046,
      "learning_rate": 0.00041461651624174686,
      "loss": 2.1228,
      "step": 86451
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.7917488813400269,
      "learning_rate": 0.00041461273598074615,
      "loss": 1.9877,
      "step": 86452
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.9350923299789429,
      "learning_rate": 0.00041460895569843685,
      "loss": 2.0652,
      "step": 86453
    },
    {
      "epoch": 1.13,
      "grad_norm": 1.043827772140503,
      "learning_rate": 0.00041460517539481956,
      "loss": 2.1879,
      "step": 86454
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.8085700869560242,
      "learning_rate": 0.00041460139506989494,
      "loss": 2.0252,
      "step": 86455
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.9435589909553528,
      "learning_rate": 0.0004145976147236638,
      "loss": 2.1729,
      "step": 86456
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.8810911178588867,
      "learning_rate": 0.0004145938343561268,
      "loss": 2.212,
      "step": 86457
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.8144645094871521,
      "learning_rate": 0.00041459005396728463,
      "loss": 2.1131,
      "step": 86458
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.8564374446868896,
      "learning_rate": 0.00041458627355713814,
      "loss": 2.2173,
      "step": 86459
    },
    {
      "epoch": 1.13,
      "grad_norm": 1.0053871870040894,
      "learning_rate": 0.00041458249312568767,
      "loss": 2.0243,
      "step": 86460
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.8194332718849182,
      "learning_rate": 0.00041457871267293426,
      "loss": 2.3841,
      "step": 86461
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.839640200138092,
      "learning_rate": 0.0004145749321988786,
      "loss": 2.117,
      "step": 86462
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.819858968257904,
      "learning_rate": 0.00041457115170352114,
      "loss": 2.1417,
      "step": 86463
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.956815242767334,
      "learning_rate": 0.0004145673711868628,
      "loss": 2.1397,
      "step": 86464
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.904449462890625,
      "learning_rate": 0.00041456359064890425,
      "loss": 2.1802,
      "step": 86465
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.8213188052177429,
      "learning_rate": 0.00041455981008964613,
      "loss": 2.0946,
      "step": 86466
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.8622743487358093,
      "learning_rate": 0.00041455602950908913,
      "loss": 2.0636,
      "step": 86467
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.875410258769989,
      "learning_rate": 0.0004145522489072341,
      "loss": 2.1237,
      "step": 86468
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.7562212347984314,
      "learning_rate": 0.00041454846828408164,
      "loss": 2.3181,
      "step": 86469
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.8208537101745605,
      "learning_rate": 0.0004145446876396324,
      "loss": 2.1364,
      "step": 86470
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.7909685373306274,
      "learning_rate": 0.0004145409069738872,
      "loss": 2.1694,
      "step": 86471
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.8577916026115417,
      "learning_rate": 0.0004145371262868466,
      "loss": 2.1306,
      "step": 86472
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.8755542039871216,
      "learning_rate": 0.0004145333455785114,
      "loss": 2.0985,
      "step": 86473
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.8483148217201233,
      "learning_rate": 0.0004145295648488824,
      "loss": 1.9774,
      "step": 86474
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.9483879208564758,
      "learning_rate": 0.00041452578409796007,
      "loss": 2.1125,
      "step": 86475
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.8158230781555176,
      "learning_rate": 0.0004145220033257453,
      "loss": 2.2444,
      "step": 86476
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.8236895799636841,
      "learning_rate": 0.0004145182225322388,
      "loss": 2.0068,
      "step": 86477
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.8290060758590698,
      "learning_rate": 0.00041451444171744105,
      "loss": 2.1227,
      "step": 86478
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.7727197408676147,
      "learning_rate": 0.00041451066088135294,
      "loss": 2.1688,
      "step": 86479
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.885732114315033,
      "learning_rate": 0.0004145068800239752,
      "loss": 2.0811,
      "step": 86480
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.8405069708824158,
      "learning_rate": 0.0004145030991453084,
      "loss": 2.2383,
      "step": 86481
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.7377654314041138,
      "learning_rate": 0.00041449931824535333,
      "loss": 2.079,
      "step": 86482
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.7793629169464111,
      "learning_rate": 0.00041449553732411074,
      "loss": 2.2438,
      "step": 86483
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.7234419584274292,
      "learning_rate": 0.0004144917563815812,
      "loss": 2.0547,
      "step": 86484
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.8115662336349487,
      "learning_rate": 0.00041448797541776544,
      "loss": 2.4032,
      "step": 86485
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.8616853356361389,
      "learning_rate": 0.00041448419443266426,
      "loss": 2.1604,
      "step": 86486
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.7647867798805237,
      "learning_rate": 0.00041448041342627835,
      "loss": 2.4214,
      "step": 86487
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.8395643830299377,
      "learning_rate": 0.00041447663239860825,
      "loss": 2.0832,
      "step": 86488
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.7629625797271729,
      "learning_rate": 0.00041447285134965493,
      "loss": 2.2349,
      "step": 86489
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.7427088618278503,
      "learning_rate": 0.0004144690702794189,
      "loss": 2.2814,
      "step": 86490
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.8801872730255127,
      "learning_rate": 0.0004144652891879009,
      "loss": 2.3007,
      "step": 86491
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.7713785767555237,
      "learning_rate": 0.0004144615080751016,
      "loss": 2.0608,
      "step": 86492
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.8723729252815247,
      "learning_rate": 0.0004144577269410218,
      "loss": 2.4054,
      "step": 86493
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.8262331485748291,
      "learning_rate": 0.00041445394578566204,
      "loss": 2.5263,
      "step": 86494
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.9122350215911865,
      "learning_rate": 0.0004144501646090232,
      "loss": 2.3132,
      "step": 86495
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.8027318120002747,
      "learning_rate": 0.000414446383411106,
      "loss": 2.1967,
      "step": 86496
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.8166528940200806,
      "learning_rate": 0.000414442602191911,
      "loss": 2.1117,
      "step": 86497
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.8128201365470886,
      "learning_rate": 0.0004144388209514388,
      "loss": 2.0684,
      "step": 86498
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.839748203754425,
      "learning_rate": 0.00041443503968969044,
      "loss": 1.9435,
      "step": 86499
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.8064789772033691,
      "learning_rate": 0.0004144312584066665,
      "loss": 1.9819,
      "step": 86500
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.8337401747703552,
      "learning_rate": 0.00041442747710236747,
      "loss": 2.1853,
      "step": 86501
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.8181577920913696,
      "learning_rate": 0.0004144236957767943,
      "loss": 2.1676,
      "step": 86502
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.8193576335906982,
      "learning_rate": 0.00041441991442994763,
      "loss": 2.218,
      "step": 86503
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.8506083488464355,
      "learning_rate": 0.0004144161330618281,
      "loss": 2.2002,
      "step": 86504
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.8195227384567261,
      "learning_rate": 0.0004144123516724364,
      "loss": 1.9462,
      "step": 86505
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.8332070112228394,
      "learning_rate": 0.0004144085702617733,
      "loss": 2.2896,
      "step": 86506
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.8422763347625732,
      "learning_rate": 0.00041440478882983955,
      "loss": 2.3021,
      "step": 86507
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.8301697969436646,
      "learning_rate": 0.0004144010073766357,
      "loss": 2.2231,
      "step": 86508
    },
    {
      "epoch": 1.13,
      "grad_norm": 1.1189473867416382,
      "learning_rate": 0.00041439722590216264,
      "loss": 2.0626,
      "step": 86509
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.8380815982818604,
      "learning_rate": 0.000414393444406421,
      "loss": 2.0302,
      "step": 86510
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.9807650446891785,
      "learning_rate": 0.0004143896628894113,
      "loss": 2.2874,
      "step": 86511
    },
    {
      "epoch": 1.13,
      "grad_norm": 1.0990265607833862,
      "learning_rate": 0.0004143858813511345,
      "loss": 2.2256,
      "step": 86512
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.8539689183235168,
      "learning_rate": 0.0004143820997915912,
      "loss": 2.0372,
      "step": 86513
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.8573094010353088,
      "learning_rate": 0.00041437831821078213,
      "loss": 2.4133,
      "step": 86514
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.9769361615180969,
      "learning_rate": 0.000414374536608708,
      "loss": 2.2249,
      "step": 86515
    },
    {
      "epoch": 1.13,
      "grad_norm": 1.1657615900039673,
      "learning_rate": 0.00041437075498536937,
      "loss": 2.3692,
      "step": 86516
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.8777459859848022,
      "learning_rate": 0.0004143669733407672,
      "loss": 2.3485,
      "step": 86517
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.882594108581543,
      "learning_rate": 0.0004143631916749019,
      "loss": 2.1542,
      "step": 86518
    },
    {
      "epoch": 1.13,
      "grad_norm": 1.035485029220581,
      "learning_rate": 0.0004143594099877744,
      "loss": 2.3013,
      "step": 86519
    },
    {
      "epoch": 1.13,
      "grad_norm": 1.0100421905517578,
      "learning_rate": 0.00041435562827938537,
      "loss": 1.9979,
      "step": 86520
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.8483631014823914,
      "learning_rate": 0.0004143518465497354,
      "loss": 2.0334,
      "step": 86521
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.9375194311141968,
      "learning_rate": 0.0004143480647988253,
      "loss": 2.0693,
      "step": 86522
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.9735219478607178,
      "learning_rate": 0.00041434428302665574,
      "loss": 2.2317,
      "step": 86523
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.8670793175697327,
      "learning_rate": 0.0004143405012332274,
      "loss": 2.1434,
      "step": 86524
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.7105475068092346,
      "learning_rate": 0.00041433671941854103,
      "loss": 2.2926,
      "step": 86525
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.9631904363632202,
      "learning_rate": 0.00041433293758259734,
      "loss": 2.0672,
      "step": 86526
    },
    {
      "epoch": 1.13,
      "grad_norm": 1.0328203439712524,
      "learning_rate": 0.00041432915572539694,
      "loss": 2.2427,
      "step": 86527
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.7741975784301758,
      "learning_rate": 0.00041432537384694055,
      "loss": 2.1295,
      "step": 86528
    },
    {
      "epoch": 1.13,
      "grad_norm": 1.0195790529251099,
      "learning_rate": 0.000414321591947229,
      "loss": 2.0341,
      "step": 86529
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.9942259788513184,
      "learning_rate": 0.0004143178100262629,
      "loss": 2.286,
      "step": 86530
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.8371213674545288,
      "learning_rate": 0.00041431402808404284,
      "loss": 2.3637,
      "step": 86531
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.8504022359848022,
      "learning_rate": 0.00041431024612056976,
      "loss": 2.3193,
      "step": 86532
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.900692880153656,
      "learning_rate": 0.0004143064641358442,
      "loss": 2.2743,
      "step": 86533
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.8227190971374512,
      "learning_rate": 0.000414302682129867,
      "loss": 2.1508,
      "step": 86534
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.8414080739021301,
      "learning_rate": 0.0004142989001026387,
      "loss": 2.169,
      "step": 86535
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.8092540502548218,
      "learning_rate": 0.00041429511805416005,
      "loss": 2.0714,
      "step": 86536
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.7471073269844055,
      "learning_rate": 0.0004142913359844318,
      "loss": 2.107,
      "step": 86537
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.7514888644218445,
      "learning_rate": 0.0004142875538934547,
      "loss": 2.0171,
      "step": 86538
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.7935139536857605,
      "learning_rate": 0.0004142837717812293,
      "loss": 2.3036,
      "step": 86539
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.840779721736908,
      "learning_rate": 0.0004142799896477564,
      "loss": 1.9573,
      "step": 86540
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.81038898229599,
      "learning_rate": 0.00041427620749303674,
      "loss": 2.2409,
      "step": 86541
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.7644456028938293,
      "learning_rate": 0.00041427242531707093,
      "loss": 2.1022,
      "step": 86542
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.7923827767372131,
      "learning_rate": 0.00041426864311985975,
      "loss": 2.0922,
      "step": 86543
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.7878793478012085,
      "learning_rate": 0.0004142648609014039,
      "loss": 2.2294,
      "step": 86544
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.8028210401535034,
      "learning_rate": 0.000414261078661704,
      "loss": 2.0575,
      "step": 86545
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.8933623433113098,
      "learning_rate": 0.00041425729640076083,
      "loss": 2.2011,
      "step": 86546
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.9838581681251526,
      "learning_rate": 0.0004142535141185751,
      "loss": 2.2585,
      "step": 86547
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.8469598889350891,
      "learning_rate": 0.00041424973181514743,
      "loss": 1.9748,
      "step": 86548
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.8060218095779419,
      "learning_rate": 0.0004142459494904786,
      "loss": 2.0567,
      "step": 86549
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.8604914546012878,
      "learning_rate": 0.0004142421671445693,
      "loss": 2.1777,
      "step": 86550
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.7934236526489258,
      "learning_rate": 0.0004142383847774202,
      "loss": 2.2378,
      "step": 86551
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.7201409935951233,
      "learning_rate": 0.0004142346023890321,
      "loss": 2.0517,
      "step": 86552
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.8279311656951904,
      "learning_rate": 0.0004142308199794055,
      "loss": 2.3698,
      "step": 86553
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.7942599654197693,
      "learning_rate": 0.00041422703754854135,
      "loss": 2.4303,
      "step": 86554
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.8208656907081604,
      "learning_rate": 0.00041422325509644014,
      "loss": 2.204,
      "step": 86555
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.78520268201828,
      "learning_rate": 0.00041421947262310283,
      "loss": 2.1372,
      "step": 86556
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.7461094856262207,
      "learning_rate": 0.0004142156901285298,
      "loss": 2.3322,
      "step": 86557
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.7897351384162903,
      "learning_rate": 0.000414211907612722,
      "loss": 2.0611,
      "step": 86558
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.7402001023292542,
      "learning_rate": 0.00041420812507568007,
      "loss": 1.8257,
      "step": 86559
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.9413495063781738,
      "learning_rate": 0.0004142043425174046,
      "loss": 2.1892,
      "step": 86560
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.7823943495750427,
      "learning_rate": 0.00041420055993789644,
      "loss": 2.0982,
      "step": 86561
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.8947314023971558,
      "learning_rate": 0.00041419677733715626,
      "loss": 2.0744,
      "step": 86562
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.9753961563110352,
      "learning_rate": 0.00041419299471518475,
      "loss": 2.0843,
      "step": 86563
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.9440042972564697,
      "learning_rate": 0.00041418921207198256,
      "loss": 2.242,
      "step": 86564
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.9339252710342407,
      "learning_rate": 0.00041418542940755055,
      "loss": 2.2028,
      "step": 86565
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.7523670196533203,
      "learning_rate": 0.00041418164672188916,
      "loss": 2.1208,
      "step": 86566
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.993996798992157,
      "learning_rate": 0.0004141778640149993,
      "loss": 2.1331,
      "step": 86567
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.8525866866111755,
      "learning_rate": 0.00041417408128688164,
      "loss": 2.0389,
      "step": 86568
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.7772798538208008,
      "learning_rate": 0.00041417029853753687,
      "loss": 2.0988,
      "step": 86569
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.8155958652496338,
      "learning_rate": 0.00041416651576696564,
      "loss": 2.134,
      "step": 86570
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.8863745927810669,
      "learning_rate": 0.00041416273297516877,
      "loss": 2.2427,
      "step": 86571
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.8263073563575745,
      "learning_rate": 0.00041415895016214685,
      "loss": 2.227,
      "step": 86572
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.8704020380973816,
      "learning_rate": 0.0004141551673279006,
      "loss": 2.3504,
      "step": 86573
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.8177746534347534,
      "learning_rate": 0.0004141513844724308,
      "loss": 2.3007,
      "step": 86574
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.7985755801200867,
      "learning_rate": 0.00041414760159573803,
      "loss": 2.2003,
      "step": 86575
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.8167378902435303,
      "learning_rate": 0.00041414381869782314,
      "loss": 2.0804,
      "step": 86576
    },
    {
      "epoch": 1.13,
      "grad_norm": 1.0636957883834839,
      "learning_rate": 0.0004141400357786867,
      "loss": 2.0209,
      "step": 86577
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.8795203566551208,
      "learning_rate": 0.00041413625283832954,
      "loss": 2.2071,
      "step": 86578
    },
    {
      "epoch": 1.13,
      "grad_norm": 1.0445518493652344,
      "learning_rate": 0.0004141324698767523,
      "loss": 2.3339,
      "step": 86579
    },
    {
      "epoch": 1.13,
      "grad_norm": 1.0700782537460327,
      "learning_rate": 0.0004141286868939556,
      "loss": 1.9369,
      "step": 86580
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.9193593859672546,
      "learning_rate": 0.0004141249038899403,
      "loss": 2.292,
      "step": 86581
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.8334766030311584,
      "learning_rate": 0.00041412112086470696,
      "loss": 2.0469,
      "step": 86582
    },
    {
      "epoch": 1.13,
      "grad_norm": 1.2376186847686768,
      "learning_rate": 0.0004141173378182564,
      "loss": 1.9898,
      "step": 86583
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.9448060393333435,
      "learning_rate": 0.00041411355475058935,
      "loss": 2.1608,
      "step": 86584
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.8048583269119263,
      "learning_rate": 0.00041410977166170623,
      "loss": 2.007,
      "step": 86585
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.882687509059906,
      "learning_rate": 0.00041410598855160815,
      "loss": 2.3295,
      "step": 86586
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.8890331387519836,
      "learning_rate": 0.0004141022054202954,
      "loss": 2.3894,
      "step": 86587
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.9308112859725952,
      "learning_rate": 0.00041409842226776915,
      "loss": 2.2447,
      "step": 86588
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.9834306240081787,
      "learning_rate": 0.00041409463909402965,
      "loss": 2.3569,
      "step": 86589
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.8082241415977478,
      "learning_rate": 0.0004140908558990779,
      "loss": 2.092,
      "step": 86590
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.8227644562721252,
      "learning_rate": 0.0004140870726829145,
      "loss": 2.0168,
      "step": 86591
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.8633387684822083,
      "learning_rate": 0.00041408328944554017,
      "loss": 2.0181,
      "step": 86592
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.7315240502357483,
      "learning_rate": 0.00041407950618695545,
      "loss": 2.1212,
      "step": 86593
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.7196586728096008,
      "learning_rate": 0.00041407572290716135,
      "loss": 1.946,
      "step": 86594
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.8285544514656067,
      "learning_rate": 0.0004140719396061584,
      "loss": 2.2063,
      "step": 86595
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.7957941889762878,
      "learning_rate": 0.00041406815628394734,
      "loss": 2.2916,
      "step": 86596
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.7837100625038147,
      "learning_rate": 0.0004140643729405288,
      "loss": 2.2295,
      "step": 86597
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.8760349154472351,
      "learning_rate": 0.00041406058957590356,
      "loss": 2.1307,
      "step": 86598
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.8259554505348206,
      "learning_rate": 0.0004140568061900723,
      "loss": 2.1952,
      "step": 86599
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.8371416330337524,
      "learning_rate": 0.0004140530227830357,
      "loss": 2.1711,
      "step": 86600
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.8388106822967529,
      "learning_rate": 0.00041404923935479455,
      "loss": 2.371,
      "step": 86601
    },
    {
      "epoch": 1.13,
      "grad_norm": 1.0402498245239258,
      "learning_rate": 0.0004140454559053496,
      "loss": 2.1788,
      "step": 86602
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.8730145692825317,
      "learning_rate": 0.00041404167243470127,
      "loss": 2.295,
      "step": 86603
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.9401675462722778,
      "learning_rate": 0.00041403788894285045,
      "loss": 2.2068,
      "step": 86604
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.9127192497253418,
      "learning_rate": 0.0004140341054297979,
      "loss": 2.2055,
      "step": 86605
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.8359643220901489,
      "learning_rate": 0.0004140303218955442,
      "loss": 2.1965,
      "step": 86606
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.7903509140014648,
      "learning_rate": 0.0004140265383400901,
      "loss": 2.2787,
      "step": 86607
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.9142720103263855,
      "learning_rate": 0.00041402275476343643,
      "loss": 2.4205,
      "step": 86608
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.8814189434051514,
      "learning_rate": 0.00041401897116558365,
      "loss": 2.0695,
      "step": 86609
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.7851291298866272,
      "learning_rate": 0.00041401518754653267,
      "loss": 1.9674,
      "step": 86610
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.84024578332901,
      "learning_rate": 0.00041401140390628417,
      "loss": 2.241,
      "step": 86611
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.8139327764511108,
      "learning_rate": 0.00041400762024483865,
      "loss": 2.2006,
      "step": 86612
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.866845428943634,
      "learning_rate": 0.000414003836562197,
      "loss": 2.2511,
      "step": 86613
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.9440573453903198,
      "learning_rate": 0.00041400005285836,
      "loss": 2.0582,
      "step": 86614
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.8010278344154358,
      "learning_rate": 0.00041399626913332806,
      "loss": 2.2654,
      "step": 86615
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.777567446231842,
      "learning_rate": 0.00041399248538710216,
      "loss": 2.3589,
      "step": 86616
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.7945156097412109,
      "learning_rate": 0.00041398870161968293,
      "loss": 2.1818,
      "step": 86617
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.8430072665214539,
      "learning_rate": 0.000413984917831071,
      "loss": 2.1206,
      "step": 86618
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.8376155495643616,
      "learning_rate": 0.0004139811340212672,
      "loss": 2.0593,
      "step": 86619
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.7745687961578369,
      "learning_rate": 0.00041397735019027207,
      "loss": 2.2714,
      "step": 86620
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.8175476789474487,
      "learning_rate": 0.00041397356633808644,
      "loss": 1.9786,
      "step": 86621
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.8568418622016907,
      "learning_rate": 0.0004139697824647109,
      "loss": 1.9966,
      "step": 86622
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.7764543294906616,
      "learning_rate": 0.00041396599857014636,
      "loss": 2.1238,
      "step": 86623
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.8097122311592102,
      "learning_rate": 0.00041396221465439325,
      "loss": 2.3653,
      "step": 86624
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.8440244197845459,
      "learning_rate": 0.00041395843071745245,
      "loss": 2.1173,
      "step": 86625
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.8605628609657288,
      "learning_rate": 0.0004139546467593247,
      "loss": 2.2454,
      "step": 86626
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.808379590511322,
      "learning_rate": 0.0004139508627800106,
      "loss": 2.103,
      "step": 86627
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.829745352268219,
      "learning_rate": 0.0004139470787795108,
      "loss": 2.1886,
      "step": 86628
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.924393355846405,
      "learning_rate": 0.0004139432947578262,
      "loss": 2.4278,
      "step": 86629
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.7763485908508301,
      "learning_rate": 0.0004139395107149573,
      "loss": 2.1969,
      "step": 86630
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.8741288185119629,
      "learning_rate": 0.00041393572665090496,
      "loss": 2.1913,
      "step": 86631
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.8145119547843933,
      "learning_rate": 0.0004139319425656698,
      "loss": 2.098,
      "step": 86632
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.7887730598449707,
      "learning_rate": 0.0004139281584592525,
      "loss": 2.353,
      "step": 86633
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.7736915946006775,
      "learning_rate": 0.0004139243743316539,
      "loss": 2.0779,
      "step": 86634
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.7970824241638184,
      "learning_rate": 0.0004139205901828746,
      "loss": 2.0536,
      "step": 86635
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.8232607841491699,
      "learning_rate": 0.00041391680601291517,
      "loss": 2.0763,
      "step": 86636
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.8531394600868225,
      "learning_rate": 0.00041391302182177656,
      "loss": 2.1807,
      "step": 86637
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.9750574827194214,
      "learning_rate": 0.00041390923760945946,
      "loss": 2.3487,
      "step": 86638
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.8197968006134033,
      "learning_rate": 0.00041390545337596427,
      "loss": 2.2639,
      "step": 86639
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.7924865484237671,
      "learning_rate": 0.00041390166912129206,
      "loss": 2.1006,
      "step": 86640
    },
    {
      "epoch": 1.13,
      "grad_norm": 1.3300306797027588,
      "learning_rate": 0.0004138978848454434,
      "loss": 2.265,
      "step": 86641
    },
    {
      "epoch": 1.13,
      "grad_norm": 1.0390574932098389,
      "learning_rate": 0.0004138941005484188,
      "loss": 2.0428,
      "step": 86642
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.8690963387489319,
      "learning_rate": 0.00041389031623021924,
      "loss": 2.3486,
      "step": 86643
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.8491939902305603,
      "learning_rate": 0.00041388653189084536,
      "loss": 2.0296,
      "step": 86644
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.9769256711006165,
      "learning_rate": 0.00041388274753029777,
      "loss": 2.1913,
      "step": 86645
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.8988178968429565,
      "learning_rate": 0.00041387896314857723,
      "loss": 2.1666,
      "step": 86646
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.7746058702468872,
      "learning_rate": 0.00041387517874568445,
      "loss": 2.1062,
      "step": 86647
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.9484955668449402,
      "learning_rate": 0.0004138713943216202,
      "loss": 2.1765,
      "step": 86648
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.7831621766090393,
      "learning_rate": 0.00041386760987638496,
      "loss": 1.8777,
      "step": 86649
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.9034677147865295,
      "learning_rate": 0.00041386382540997967,
      "loss": 2.0598,
      "step": 86650
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.8993801474571228,
      "learning_rate": 0.000413860040922405,
      "loss": 2.1591,
      "step": 86651
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.7809641361236572,
      "learning_rate": 0.0004138562564136615,
      "loss": 2.1359,
      "step": 86652
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.8068220615386963,
      "learning_rate": 0.00041385247188375,
      "loss": 2.1024,
      "step": 86653
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.914043664932251,
      "learning_rate": 0.0004138486873326713,
      "loss": 2.066,
      "step": 86654
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.8756574988365173,
      "learning_rate": 0.0004138449027604258,
      "loss": 2.16,
      "step": 86655
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.8369112610816956,
      "learning_rate": 0.0004138411181670144,
      "loss": 2.1064,
      "step": 86656
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.8340666890144348,
      "learning_rate": 0.00041383733355243786,
      "loss": 2.2859,
      "step": 86657
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.9085386395454407,
      "learning_rate": 0.0004138335489166968,
      "loss": 2.1501,
      "step": 86658
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.7734178900718689,
      "learning_rate": 0.00041382976425979184,
      "loss": 2.1793,
      "step": 86659
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.9725180268287659,
      "learning_rate": 0.000413825979581724,
      "loss": 2.1538,
      "step": 86660
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.8340247273445129,
      "learning_rate": 0.00041382219488249355,
      "loss": 2.215,
      "step": 86661
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.8129907846450806,
      "learning_rate": 0.00041381841016210143,
      "loss": 2.3357,
      "step": 86662
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.9067367315292358,
      "learning_rate": 0.00041381462542054845,
      "loss": 2.1363,
      "step": 86663
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.7863737344741821,
      "learning_rate": 0.0004138108406578351,
      "loss": 2.2353,
      "step": 86664
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.7854729890823364,
      "learning_rate": 0.0004138070558739621,
      "loss": 1.9081,
      "step": 86665
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.7695932388305664,
      "learning_rate": 0.00041380327106893035,
      "loss": 2.2407,
      "step": 86666
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.9017605185508728,
      "learning_rate": 0.00041379948624274037,
      "loss": 2.1779,
      "step": 86667
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.8443371653556824,
      "learning_rate": 0.0004137957013953929,
      "loss": 2.1811,
      "step": 86668
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.8092812299728394,
      "learning_rate": 0.00041379191652688867,
      "loss": 2.0855,
      "step": 86669
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.8034712672233582,
      "learning_rate": 0.0004137881316372283,
      "loss": 2.0389,
      "step": 86670
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.8412032127380371,
      "learning_rate": 0.0004137843467264126,
      "loss": 2.1864,
      "step": 86671
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.9060668349266052,
      "learning_rate": 0.0004137805617944424,
      "loss": 2.0394,
      "step": 86672
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.7771490216255188,
      "learning_rate": 0.00041377677684131814,
      "loss": 2.0375,
      "step": 86673
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.8399307727813721,
      "learning_rate": 0.00041377299186704056,
      "loss": 2.0016,
      "step": 86674
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.8195866346359253,
      "learning_rate": 0.0004137692068716105,
      "loss": 2.3188,
      "step": 86675
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.8024554252624512,
      "learning_rate": 0.00041376542185502863,
      "loss": 2.3169,
      "step": 86676
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.736257791519165,
      "learning_rate": 0.0004137616368172955,
      "loss": 2.0523,
      "step": 86677
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.7681720852851868,
      "learning_rate": 0.000413757851758412,
      "loss": 2.3097,
      "step": 86678
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.7894605398178101,
      "learning_rate": 0.0004137540666783788,
      "loss": 2.1145,
      "step": 86679
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.9130123853683472,
      "learning_rate": 0.00041375028157719656,
      "loss": 2.1793,
      "step": 86680
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.8696109056472778,
      "learning_rate": 0.00041374649645486603,
      "loss": 2.2147,
      "step": 86681
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.8670323491096497,
      "learning_rate": 0.00041374271131138776,
      "loss": 2.0086,
      "step": 86682
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.8172861337661743,
      "learning_rate": 0.0004137389261467627,
      "loss": 1.9873,
      "step": 86683
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.8183755278587341,
      "learning_rate": 0.00041373514096099134,
      "loss": 2.0425,
      "step": 86684
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.7893103361129761,
      "learning_rate": 0.0004137313557540745,
      "loss": 2.1702,
      "step": 86685
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.7541918158531189,
      "learning_rate": 0.00041372757052601284,
      "loss": 1.9808,
      "step": 86686
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.8042681813240051,
      "learning_rate": 0.0004137237852768072,
      "loss": 2.0539,
      "step": 86687
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.8242625594139099,
      "learning_rate": 0.000413720000006458,
      "loss": 2.1899,
      "step": 86688
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.8716530203819275,
      "learning_rate": 0.00041371621471496614,
      "loss": 2.4749,
      "step": 86689
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.9581349492073059,
      "learning_rate": 0.0004137124294023323,
      "loss": 2.2432,
      "step": 86690
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.8820342421531677,
      "learning_rate": 0.0004137086440685572,
      "loss": 2.2276,
      "step": 86691
    },
    {
      "epoch": 1.13,
      "grad_norm": 1.0121240615844727,
      "learning_rate": 0.0004137048587136415,
      "loss": 2.2283,
      "step": 86692
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.8241207599639893,
      "learning_rate": 0.000413701073337586,
      "loss": 2.5027,
      "step": 86693
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.8876075744628906,
      "learning_rate": 0.0004136972879403912,
      "loss": 2.3541,
      "step": 86694
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.9852043986320496,
      "learning_rate": 0.00041369350252205795,
      "loss": 2.0176,
      "step": 86695
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.8216562867164612,
      "learning_rate": 0.000413689717082587,
      "loss": 2.0656,
      "step": 86696
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.9754934310913086,
      "learning_rate": 0.00041368593162197895,
      "loss": 2.2001,
      "step": 86697
    },
    {
      "epoch": 1.13,
      "grad_norm": 1.0691486597061157,
      "learning_rate": 0.0004136821461402345,
      "loss": 2.2611,
      "step": 86698
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.8354629278182983,
      "learning_rate": 0.00041367836063735446,
      "loss": 2.3146,
      "step": 86699
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.785764217376709,
      "learning_rate": 0.00041367457511333944,
      "loss": 2.235,
      "step": 86700
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.9226773381233215,
      "learning_rate": 0.0004136707895681901,
      "loss": 2.2806,
      "step": 86701
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.930027186870575,
      "learning_rate": 0.00041366700400190736,
      "loss": 2.2627,
      "step": 86702
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.908534586429596,
      "learning_rate": 0.00041366321841449167,
      "loss": 2.3056,
      "step": 86703
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.8138155937194824,
      "learning_rate": 0.00041365943280594384,
      "loss": 2.3168,
      "step": 86704
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.8941683769226074,
      "learning_rate": 0.00041365564717626464,
      "loss": 2.189,
      "step": 86705
    },
    {
      "epoch": 1.13,
      "grad_norm": 1.0750789642333984,
      "learning_rate": 0.00041365186152545467,
      "loss": 1.9545,
      "step": 86706
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.7624532580375671,
      "learning_rate": 0.0004136480758535147,
      "loss": 2.2992,
      "step": 86707
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.9161053895950317,
      "learning_rate": 0.0004136442901604454,
      "loss": 2.2704,
      "step": 86708
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.9986797571182251,
      "learning_rate": 0.0004136405044462474,
      "loss": 2.1987,
      "step": 86709
    },
    {
      "epoch": 1.13,
      "grad_norm": 1.055005431175232,
      "learning_rate": 0.0004136367187109216,
      "loss": 2.3293,
      "step": 86710
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.9190179109573364,
      "learning_rate": 0.00041363293295446855,
      "loss": 2.1249,
      "step": 86711
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.9469074606895447,
      "learning_rate": 0.000413629147176889,
      "loss": 2.2877,
      "step": 86712
    },
    {
      "epoch": 1.13,
      "grad_norm": 1.192742109298706,
      "learning_rate": 0.0004136253613781836,
      "loss": 2.2161,
      "step": 86713
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.9239273071289062,
      "learning_rate": 0.00041362157555835307,
      "loss": 2.2646,
      "step": 86714
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.8903929591178894,
      "learning_rate": 0.0004136177897173983,
      "loss": 2.1962,
      "step": 86715
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.8731290102005005,
      "learning_rate": 0.00041361400385531974,
      "loss": 2.3685,
      "step": 86716
    },
    {
      "epoch": 1.13,
      "grad_norm": 1.6664015054702759,
      "learning_rate": 0.00041361021797211813,
      "loss": 2.1272,
      "step": 86717
    },
    {
      "epoch": 1.13,
      "grad_norm": 1.1816134452819824,
      "learning_rate": 0.00041360643206779437,
      "loss": 2.3674,
      "step": 86718
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.8891193270683289,
      "learning_rate": 0.00041360264614234894,
      "loss": 2.17,
      "step": 86719
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.8653639554977417,
      "learning_rate": 0.00041359886019578265,
      "loss": 2.1457,
      "step": 86720
    },
    {
      "epoch": 1.13,
      "grad_norm": 1.0755329132080078,
      "learning_rate": 0.0004135950742280962,
      "loss": 2.2081,
      "step": 86721
    },
    {
      "epoch": 1.13,
      "grad_norm": 1.067421317100525,
      "learning_rate": 0.0004135912882392903,
      "loss": 2.2415,
      "step": 86722
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.7959436178207397,
      "learning_rate": 0.0004135875022293656,
      "loss": 2.1542,
      "step": 86723
    },
    {
      "epoch": 1.13,
      "grad_norm": 1.0655769109725952,
      "learning_rate": 0.00041358371619832284,
      "loss": 2.3527,
      "step": 86724
    },
    {
      "epoch": 1.13,
      "grad_norm": 1.0391522645950317,
      "learning_rate": 0.00041357993014616275,
      "loss": 2.2452,
      "step": 86725
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.8166463971138,
      "learning_rate": 0.00041357614407288597,
      "loss": 2.0506,
      "step": 86726
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.8384370803833008,
      "learning_rate": 0.00041357235797849333,
      "loss": 2.0782,
      "step": 86727
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.9762664437294006,
      "learning_rate": 0.00041356857186298525,
      "loss": 2.314,
      "step": 86728
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.9869099855422974,
      "learning_rate": 0.00041356478572636277,
      "loss": 2.0331,
      "step": 86729
    },
    {
      "epoch": 1.13,
      "grad_norm": 1.0474809408187866,
      "learning_rate": 0.00041356099956862654,
      "loss": 2.1608,
      "step": 86730
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.802830696105957,
      "learning_rate": 0.000413557213389777,
      "loss": 2.0385,
      "step": 86731
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.9164223670959473,
      "learning_rate": 0.00041355342718981503,
      "loss": 2.0891,
      "step": 86732
    },
    {
      "epoch": 1.13,
      "grad_norm": 1.1309996843338013,
      "learning_rate": 0.00041354964096874155,
      "loss": 2.1457,
      "step": 86733
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.8231431841850281,
      "learning_rate": 0.00041354585472655686,
      "loss": 2.2789,
      "step": 86734
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.813145101070404,
      "learning_rate": 0.00041354206846326184,
      "loss": 1.8961,
      "step": 86735
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.976969838142395,
      "learning_rate": 0.0004135382821788573,
      "loss": 2.0909,
      "step": 86736
    },
    {
      "epoch": 1.13,
      "grad_norm": 1.1836774349212646,
      "learning_rate": 0.0004135344958733438,
      "loss": 2.119,
      "step": 86737
    },
    {
      "epoch": 1.13,
      "grad_norm": 1.0458884239196777,
      "learning_rate": 0.0004135307095467221,
      "loss": 2.3364,
      "step": 86738
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.7902629375457764,
      "learning_rate": 0.00041352692319899294,
      "loss": 2.2612,
      "step": 86739
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.9220899343490601,
      "learning_rate": 0.00041352313683015696,
      "loss": 2.2212,
      "step": 86740
    },
    {
      "epoch": 1.13,
      "grad_norm": 1.147680401802063,
      "learning_rate": 0.0004135193504402148,
      "loss": 2.1464,
      "step": 86741
    },
    {
      "epoch": 1.13,
      "grad_norm": 1.0322705507278442,
      "learning_rate": 0.0004135155640291674,
      "loss": 2.3445,
      "step": 86742
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.8558586835861206,
      "learning_rate": 0.00041351177759701524,
      "loss": 2.1663,
      "step": 86743
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.9541861414909363,
      "learning_rate": 0.0004135079911437591,
      "loss": 2.0012,
      "step": 86744
    },
    {
      "epoch": 1.13,
      "grad_norm": 1.2485464811325073,
      "learning_rate": 0.00041350420466939973,
      "loss": 2.0913,
      "step": 86745
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.8032669425010681,
      "learning_rate": 0.00041350041817393767,
      "loss": 2.0584,
      "step": 86746
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.8811048865318298,
      "learning_rate": 0.00041349663165737377,
      "loss": 2.2825,
      "step": 86747
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.9741935133934021,
      "learning_rate": 0.0004134928451197088,
      "loss": 2.298,
      "step": 86748
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.8967857360839844,
      "learning_rate": 0.00041348905856094327,
      "loss": 2.0727,
      "step": 86749
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.8932086825370789,
      "learning_rate": 0.000413485271981078,
      "loss": 2.2191,
      "step": 86750
    },
    {
      "epoch": 1.13,
      "grad_norm": 1.0709178447723389,
      "learning_rate": 0.00041348148538011375,
      "loss": 2.26,
      "step": 86751
    },
    {
      "epoch": 1.13,
      "grad_norm": 1.13752019405365,
      "learning_rate": 0.0004134776987580511,
      "loss": 2.0693,
      "step": 86752
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.9011566042900085,
      "learning_rate": 0.00041347391211489075,
      "loss": 2.2677,
      "step": 86753
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.8606006503105164,
      "learning_rate": 0.00041347012545063353,
      "loss": 2.2951,
      "step": 86754
    },
    {
      "epoch": 1.13,
      "grad_norm": 1.0223121643066406,
      "learning_rate": 0.00041346633876528007,
      "loss": 2.2556,
      "step": 86755
    },
    {
      "epoch": 1.13,
      "grad_norm": 1.2322343587875366,
      "learning_rate": 0.00041346255205883096,
      "loss": 2.2705,
      "step": 86756
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.8849189281463623,
      "learning_rate": 0.0004134587653312872,
      "loss": 2.0555,
      "step": 86757
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.8277934193611145,
      "learning_rate": 0.0004134549785826492,
      "loss": 1.9788,
      "step": 86758
    },
    {
      "epoch": 1.13,
      "grad_norm": 1.1777806282043457,
      "learning_rate": 0.00041345119181291777,
      "loss": 2.1383,
      "step": 86759
    },
    {
      "epoch": 1.13,
      "grad_norm": 1.1445190906524658,
      "learning_rate": 0.0004134474050220936,
      "loss": 2.1253,
      "step": 86760
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.810766875743866,
      "learning_rate": 0.0004134436182101776,
      "loss": 2.3863,
      "step": 86761
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.8029268383979797,
      "learning_rate": 0.0004134398313771701,
      "loss": 2.2836,
      "step": 86762
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.8893110752105713,
      "learning_rate": 0.000413436044523072,
      "loss": 2.5182,
      "step": 86763
    },
    {
      "epoch": 1.13,
      "grad_norm": 1.0773707628250122,
      "learning_rate": 0.0004134322576478842,
      "loss": 2.4153,
      "step": 86764
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.8104235529899597,
      "learning_rate": 0.00041342847075160694,
      "loss": 2.1634,
      "step": 86765
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.8506526947021484,
      "learning_rate": 0.0004134246838342413,
      "loss": 2.1147,
      "step": 86766
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.8245054483413696,
      "learning_rate": 0.0004134208968957879,
      "loss": 2.1306,
      "step": 86767
    },
    {
      "epoch": 1.13,
      "grad_norm": 1.3066277503967285,
      "learning_rate": 0.0004134171099362474,
      "loss": 1.979,
      "step": 86768
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.9277042746543884,
      "learning_rate": 0.0004134133229556205,
      "loss": 2.1272,
      "step": 86769
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.7819560766220093,
      "learning_rate": 0.0004134095359539078,
      "loss": 2.181,
      "step": 86770
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.9709548950195312,
      "learning_rate": 0.0004134057489311103,
      "loss": 2.1783,
      "step": 86771
    },
    {
      "epoch": 1.13,
      "grad_norm": 1.0260387659072876,
      "learning_rate": 0.00041340196188722854,
      "loss": 2.2763,
      "step": 86772
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.8831406831741333,
      "learning_rate": 0.0004133981748222631,
      "loss": 2.1623,
      "step": 86773
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.8392928838729858,
      "learning_rate": 0.0004133943877362149,
      "loss": 2.2134,
      "step": 86774
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.9893087148666382,
      "learning_rate": 0.0004133906006290845,
      "loss": 2.0534,
      "step": 86775
    },
    {
      "epoch": 1.13,
      "grad_norm": 1.0608582496643066,
      "learning_rate": 0.00041338681350087264,
      "loss": 2.1629,
      "step": 86776
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.843281090259552,
      "learning_rate": 0.00041338302635158,
      "loss": 2.0855,
      "step": 86777
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.768854022026062,
      "learning_rate": 0.00041337923918120735,
      "loss": 2.6146,
      "step": 86778
    },
    {
      "epoch": 1.13,
      "grad_norm": 1.082100510597229,
      "learning_rate": 0.0004133754519897554,
      "loss": 2.0809,
      "step": 86779
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.8171884417533875,
      "learning_rate": 0.00041337166477722475,
      "loss": 2.1146,
      "step": 86780
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.7843098044395447,
      "learning_rate": 0.0004133678775436162,
      "loss": 2.4091,
      "step": 86781
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.8531711101531982,
      "learning_rate": 0.0004133640902889304,
      "loss": 2.0274,
      "step": 86782
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.764708399772644,
      "learning_rate": 0.0004133603030131681,
      "loss": 2.0971,
      "step": 86783
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.7962109446525574,
      "learning_rate": 0.00041335651571632997,
      "loss": 1.8914,
      "step": 86784
    },
    {
      "epoch": 1.13,
      "grad_norm": 1.075097680091858,
      "learning_rate": 0.00041335272839841675,
      "loss": 2.1796,
      "step": 86785
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.7639371156692505,
      "learning_rate": 0.0004133489410594291,
      "loss": 2.2359,
      "step": 86786
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.8410763144493103,
      "learning_rate": 0.00041334515369936765,
      "loss": 2.3775,
      "step": 86787
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.8884861469268799,
      "learning_rate": 0.00041334136631823335,
      "loss": 2.1173,
      "step": 86788
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.8618728518486023,
      "learning_rate": 0.0004133375789160267,
      "loss": 2.1396,
      "step": 86789
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.8562257885932922,
      "learning_rate": 0.00041333379149274834,
      "loss": 2.1056,
      "step": 86790
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.7865633368492126,
      "learning_rate": 0.0004133300040483993,
      "loss": 2.1839,
      "step": 86791
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.8207772374153137,
      "learning_rate": 0.00041332621658297995,
      "loss": 1.9396,
      "step": 86792
    },
    {
      "epoch": 1.13,
      "grad_norm": 1.217294692993164,
      "learning_rate": 0.00041332242909649107,
      "loss": 2.1399,
      "step": 86793
    },
    {
      "epoch": 1.13,
      "grad_norm": 1.2778818607330322,
      "learning_rate": 0.0004133186415889335,
      "loss": 1.9179,
      "step": 86794
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.8439632058143616,
      "learning_rate": 0.0004133148540603078,
      "loss": 1.9563,
      "step": 86795
    },
    {
      "epoch": 1.13,
      "grad_norm": 1.2315888404846191,
      "learning_rate": 0.00041331106651061475,
      "loss": 2.0897,
      "step": 86796
    },
    {
      "epoch": 1.13,
      "grad_norm": 1.0775524377822876,
      "learning_rate": 0.00041330727893985507,
      "loss": 2.0976,
      "step": 86797
    },
    {
      "epoch": 1.13,
      "grad_norm": 1.0612010955810547,
      "learning_rate": 0.0004133034913480294,
      "loss": 2.37,
      "step": 86798
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.7746737599372864,
      "learning_rate": 0.0004132997037351384,
      "loss": 2.4116,
      "step": 86799
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.9348337054252625,
      "learning_rate": 0.00041329591610118297,
      "loss": 2.3174,
      "step": 86800
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.9137298464775085,
      "learning_rate": 0.00041329212844616364,
      "loss": 2.2317,
      "step": 86801
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.8619669079780579,
      "learning_rate": 0.00041328834077008114,
      "loss": 2.3627,
      "step": 86802
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.8052416443824768,
      "learning_rate": 0.0004132845530729362,
      "loss": 2.2055,
      "step": 86803
    },
    {
      "epoch": 1.13,
      "grad_norm": 1.1835070848464966,
      "learning_rate": 0.00041328076535472954,
      "loss": 2.2262,
      "step": 86804
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.9080857038497925,
      "learning_rate": 0.00041327697761546186,
      "loss": 2.1871,
      "step": 86805
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.7562054991722107,
      "learning_rate": 0.00041327318985513386,
      "loss": 2.2311,
      "step": 86806
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.8302956819534302,
      "learning_rate": 0.0004132694020737462,
      "loss": 2.2505,
      "step": 86807
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.9666738510131836,
      "learning_rate": 0.0004132656142712996,
      "loss": 2.279,
      "step": 86808
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.7648799419403076,
      "learning_rate": 0.0004132618264477949,
      "loss": 2.3004,
      "step": 86809
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.9596179127693176,
      "learning_rate": 0.00041325803860323254,
      "loss": 2.2086,
      "step": 86810
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.7818176746368408,
      "learning_rate": 0.00041325425073761345,
      "loss": 2.0696,
      "step": 86811
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.750156044960022,
      "learning_rate": 0.0004132504628509383,
      "loss": 2.277,
      "step": 86812
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.8667047619819641,
      "learning_rate": 0.00041324667494320764,
      "loss": 2.051,
      "step": 86813
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.8245537877082825,
      "learning_rate": 0.00041324288701442233,
      "loss": 2.3779,
      "step": 86814
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.8660119771957397,
      "learning_rate": 0.00041323909906458307,
      "loss": 2.0895,
      "step": 86815
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.8766900897026062,
      "learning_rate": 0.00041323531109369055,
      "loss": 2.1846,
      "step": 86816
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.8527098298072815,
      "learning_rate": 0.00041323152310174537,
      "loss": 2.2798,
      "step": 86817
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.8896264433860779,
      "learning_rate": 0.00041322773508874834,
      "loss": 2.1535,
      "step": 86818
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.8349048495292664,
      "learning_rate": 0.0004132239470547001,
      "loss": 2.2689,
      "step": 86819
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.7759386897087097,
      "learning_rate": 0.00041322015899960136,
      "loss": 2.0564,
      "step": 86820
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.8972437381744385,
      "learning_rate": 0.000413216370923453,
      "loss": 2.3905,
      "step": 86821
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.9477682709693909,
      "learning_rate": 0.0004132125828262555,
      "loss": 2.2583,
      "step": 86822
    },
    {
      "epoch": 1.13,
      "grad_norm": 1.1242547035217285,
      "learning_rate": 0.0004132087947080096,
      "loss": 2.2664,
      "step": 86823
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.897523820400238,
      "learning_rate": 0.00041320500656871615,
      "loss": 2.1608,
      "step": 86824
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.8972238898277283,
      "learning_rate": 0.00041320121840837567,
      "loss": 2.2341,
      "step": 86825
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.9950429797172546,
      "learning_rate": 0.000413197430226989,
      "loss": 2.2257,
      "step": 86826
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.8780627846717834,
      "learning_rate": 0.00041319364202455674,
      "loss": 2.1679,
      "step": 86827
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.7768124938011169,
      "learning_rate": 0.00041318985380107965,
      "loss": 2.2242,
      "step": 86828
    },
    {
      "epoch": 1.13,
      "grad_norm": 1.040778398513794,
      "learning_rate": 0.00041318606555655846,
      "loss": 2.1197,
      "step": 86829
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.7809637188911438,
      "learning_rate": 0.0004131822772909939,
      "loss": 2.1614,
      "step": 86830
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.8395169973373413,
      "learning_rate": 0.00041317848900438646,
      "loss": 2.1479,
      "step": 86831
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.8567433953285217,
      "learning_rate": 0.0004131747006967371,
      "loss": 2.1902,
      "step": 86832
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.837966799736023,
      "learning_rate": 0.0004131709123680465,
      "loss": 2.3059,
      "step": 86833
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.9268248081207275,
      "learning_rate": 0.00041316712401831513,
      "loss": 2.1304,
      "step": 86834
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.7889559864997864,
      "learning_rate": 0.0004131633356475439,
      "loss": 2.3299,
      "step": 86835
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.9546900987625122,
      "learning_rate": 0.00041315954725573353,
      "loss": 2.1719,
      "step": 86836
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.833773136138916,
      "learning_rate": 0.00041315575884288465,
      "loss": 2.2258,
      "step": 86837
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.9420397281646729,
      "learning_rate": 0.0004131519704089979,
      "loss": 2.0808,
      "step": 86838
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.8506676554679871,
      "learning_rate": 0.0004131481819540742,
      "loss": 2.196,
      "step": 86839
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.8032541275024414,
      "learning_rate": 0.0004131443934781141,
      "loss": 2.1048,
      "step": 86840
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.8293132781982422,
      "learning_rate": 0.00041314060498111825,
      "loss": 2.1717,
      "step": 86841
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.8200406432151794,
      "learning_rate": 0.0004131368164630874,
      "loss": 1.9236,
      "step": 86842
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.7522515654563904,
      "learning_rate": 0.00041313302792402235,
      "loss": 1.9237,
      "step": 86843
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.8585798740386963,
      "learning_rate": 0.0004131292393639237,
      "loss": 2.0926,
      "step": 86844
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.7275775074958801,
      "learning_rate": 0.00041312545078279214,
      "loss": 2.133,
      "step": 86845
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.8022192716598511,
      "learning_rate": 0.00041312166218062854,
      "loss": 2.136,
      "step": 86846
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.9054273962974548,
      "learning_rate": 0.0004131178735574334,
      "loss": 2.1241,
      "step": 86847
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.9307882785797119,
      "learning_rate": 0.0004131140849132075,
      "loss": 2.5112,
      "step": 86848
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.7925095558166504,
      "learning_rate": 0.0004131102962479517,
      "loss": 2.3038,
      "step": 86849
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.9416965246200562,
      "learning_rate": 0.0004131065075616664,
      "loss": 2.2457,
      "step": 86850
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.7885212898254395,
      "learning_rate": 0.0004131027188543525,
      "loss": 2.1825,
      "step": 86851
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.8673849701881409,
      "learning_rate": 0.0004130989301260107,
      "loss": 2.1502,
      "step": 86852
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.8129704594612122,
      "learning_rate": 0.0004130951413766417,
      "loss": 2.0682,
      "step": 86853
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.775037944316864,
      "learning_rate": 0.0004130913526062461,
      "loss": 2.2967,
      "step": 86854
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.8995419144630432,
      "learning_rate": 0.0004130875638148247,
      "loss": 2.3722,
      "step": 86855
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.7591526508331299,
      "learning_rate": 0.0004130837750023783,
      "loss": 2.0354,
      "step": 86856
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.7515432834625244,
      "learning_rate": 0.00041307998616890734,
      "loss": 2.1892,
      "step": 86857
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.8915225267410278,
      "learning_rate": 0.00041307619731441276,
      "loss": 2.2492,
      "step": 86858
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.7676655650138855,
      "learning_rate": 0.0004130724084388952,
      "loss": 2.1174,
      "step": 86859
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.799258291721344,
      "learning_rate": 0.0004130686195423553,
      "loss": 2.0365,
      "step": 86860
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.9637588262557983,
      "learning_rate": 0.0004130648306247938,
      "loss": 2.1697,
      "step": 86861
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.8324964046478271,
      "learning_rate": 0.0004130610416862114,
      "loss": 2.1599,
      "step": 86862
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.8373430371284485,
      "learning_rate": 0.00041305725272660883,
      "loss": 1.9989,
      "step": 86863
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.804791271686554,
      "learning_rate": 0.0004130534637459868,
      "loss": 2.1011,
      "step": 86864
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.900597333908081,
      "learning_rate": 0.00041304967474434603,
      "loss": 2.3909,
      "step": 86865
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.8879544138908386,
      "learning_rate": 0.00041304588572168714,
      "loss": 2.2459,
      "step": 86866
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.7790929675102234,
      "learning_rate": 0.00041304209667801094,
      "loss": 2.0679,
      "step": 86867
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.8128739595413208,
      "learning_rate": 0.000413038307613318,
      "loss": 2.138,
      "step": 86868
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.8968607783317566,
      "learning_rate": 0.0004130345185276091,
      "loss": 2.3379,
      "step": 86869
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.8300669193267822,
      "learning_rate": 0.0004130307294208851,
      "loss": 2.2626,
      "step": 86870
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.7788952589035034,
      "learning_rate": 0.00041302694029314645,
      "loss": 2.3262,
      "step": 86871
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.7780113816261292,
      "learning_rate": 0.0004130231511443939,
      "loss": 2.289,
      "step": 86872
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.8821353316307068,
      "learning_rate": 0.0004130193619746283,
      "loss": 2.2225,
      "step": 86873
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.8750197887420654,
      "learning_rate": 0.0004130155727838502,
      "loss": 2.3833,
      "step": 86874
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.7916033864021301,
      "learning_rate": 0.0004130117835720604,
      "loss": 2.3735,
      "step": 86875
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.8838302493095398,
      "learning_rate": 0.0004130079943392597,
      "loss": 2.2531,
      "step": 86876
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.8530775904655457,
      "learning_rate": 0.00041300420508544853,
      "loss": 2.1496,
      "step": 86877
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.8863291144371033,
      "learning_rate": 0.00041300041581062775,
      "loss": 2.0603,
      "step": 86878
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.7984259128570557,
      "learning_rate": 0.0004129966265147982,
      "loss": 2.0966,
      "step": 86879
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.8006868362426758,
      "learning_rate": 0.00041299283719796026,
      "loss": 2.2132,
      "step": 86880
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.8028510808944702,
      "learning_rate": 0.0004129890478601149,
      "loss": 2.2084,
      "step": 86881
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.7872610688209534,
      "learning_rate": 0.0004129852585012628,
      "loss": 2.08,
      "step": 86882
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.824105441570282,
      "learning_rate": 0.0004129814691214045,
      "loss": 2.1053,
      "step": 86883
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.946574866771698,
      "learning_rate": 0.0004129776797205409,
      "loss": 2.243,
      "step": 86884
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.8845953941345215,
      "learning_rate": 0.0004129738902986726,
      "loss": 2.0757,
      "step": 86885
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.8179122805595398,
      "learning_rate": 0.00041297010085580025,
      "loss": 2.0265,
      "step": 86886
    },
    {
      "epoch": 1.13,
      "grad_norm": 1.0797351598739624,
      "learning_rate": 0.0004129663113919247,
      "loss": 2.3286,
      "step": 86887
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.941948413848877,
      "learning_rate": 0.00041296252190704656,
      "loss": 2.2096,
      "step": 86888
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.8340083360671997,
      "learning_rate": 0.00041295873240116657,
      "loss": 2.1494,
      "step": 86889
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.8615207672119141,
      "learning_rate": 0.0004129549428742853,
      "loss": 2.1693,
      "step": 86890
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.915521502494812,
      "learning_rate": 0.00041295115332640376,
      "loss": 2.1892,
      "step": 86891
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.7879707217216492,
      "learning_rate": 0.00041294736375752236,
      "loss": 2.2593,
      "step": 86892
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.8945559859275818,
      "learning_rate": 0.0004129435741676419,
      "loss": 2.1471,
      "step": 86893
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.8514530062675476,
      "learning_rate": 0.0004129397845567632,
      "loss": 2.152,
      "step": 86894
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.7516213655471802,
      "learning_rate": 0.00041293599492488674,
      "loss": 2.3362,
      "step": 86895
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.7581504583358765,
      "learning_rate": 0.0004129322052720134,
      "loss": 2.25,
      "step": 86896
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.7870240807533264,
      "learning_rate": 0.00041292841559814386,
      "loss": 2.262,
      "step": 86897
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.8080188632011414,
      "learning_rate": 0.00041292462590327876,
      "loss": 2.0324,
      "step": 86898
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.7483142018318176,
      "learning_rate": 0.00041292083618741884,
      "loss": 2.2171,
      "step": 86899
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.827025294303894,
      "learning_rate": 0.0004129170464505649,
      "loss": 2.2027,
      "step": 86900
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.8043172955513,
      "learning_rate": 0.0004129132566927174,
      "loss": 2.0414,
      "step": 86901
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.7590027451515198,
      "learning_rate": 0.0004129094669138773,
      "loss": 2.1114,
      "step": 86902
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.8146236538887024,
      "learning_rate": 0.0004129056771140452,
      "loss": 2.128,
      "step": 86903
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.7512033581733704,
      "learning_rate": 0.00041290188729322167,
      "loss": 2.2407,
      "step": 86904
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.8080409169197083,
      "learning_rate": 0.00041289809745140765,
      "loss": 1.9217,
      "step": 86905
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.8021172881126404,
      "learning_rate": 0.00041289430758860375,
      "loss": 2.1536,
      "step": 86906
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.8500104546546936,
      "learning_rate": 0.0004128905177048106,
      "loss": 2.1934,
      "step": 86907
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.8263583183288574,
      "learning_rate": 0.000412886727800029,
      "loss": 2.2195,
      "step": 86908
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.8081140518188477,
      "learning_rate": 0.00041288293787425975,
      "loss": 2.2397,
      "step": 86909
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.8386034369468689,
      "learning_rate": 0.0004128791479275033,
      "loss": 2.137,
      "step": 86910
    },
    {
      "epoch": 1.13,
      "grad_norm": 1.1117937564849854,
      "learning_rate": 0.00041287535795976046,
      "loss": 2.0529,
      "step": 86911
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.828198254108429,
      "learning_rate": 0.000412871567971032,
      "loss": 1.9784,
      "step": 86912
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.7784438133239746,
      "learning_rate": 0.00041286777796131866,
      "loss": 2.0866,
      "step": 86913
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.9188300967216492,
      "learning_rate": 0.000412863987930621,
      "loss": 2.0558,
      "step": 86914
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.7880356311798096,
      "learning_rate": 0.00041286019787893977,
      "loss": 2.1442,
      "step": 86915
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.8084493279457092,
      "learning_rate": 0.0004128564078062758,
      "loss": 2.0514,
      "step": 86916
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.8181793689727783,
      "learning_rate": 0.0004128526177126296,
      "loss": 2.226,
      "step": 86917
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.8154630064964294,
      "learning_rate": 0.00041284882759800196,
      "loss": 2.3641,
      "step": 86918
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.9176775217056274,
      "learning_rate": 0.00041284503746239366,
      "loss": 2.179,
      "step": 86919
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.7932053804397583,
      "learning_rate": 0.00041284124730580523,
      "loss": 2.2452,
      "step": 86920
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.8038524389266968,
      "learning_rate": 0.00041283745712823754,
      "loss": 1.9945,
      "step": 86921
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.8197343349456787,
      "learning_rate": 0.00041283366692969136,
      "loss": 2.1909,
      "step": 86922
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.8484361171722412,
      "learning_rate": 0.0004128298767101671,
      "loss": 2.1713,
      "step": 86923
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.8045632839202881,
      "learning_rate": 0.00041282608646966566,
      "loss": 2.2013,
      "step": 86924
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.8482223153114319,
      "learning_rate": 0.00041282229620818783,
      "loss": 2.2556,
      "step": 86925
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.837514340877533,
      "learning_rate": 0.00041281850592573416,
      "loss": 2.2572,
      "step": 86926
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.8789146542549133,
      "learning_rate": 0.0004128147156223053,
      "loss": 2.1107,
      "step": 86927
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.9274330735206604,
      "learning_rate": 0.00041281092529790216,
      "loss": 2.1012,
      "step": 86928
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.8386232852935791,
      "learning_rate": 0.00041280713495252534,
      "loss": 2.197,
      "step": 86929
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.8266681432723999,
      "learning_rate": 0.00041280334458617546,
      "loss": 2.2696,
      "step": 86930
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.7864508628845215,
      "learning_rate": 0.0004127995541988534,
      "loss": 2.1167,
      "step": 86931
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.8681609630584717,
      "learning_rate": 0.00041279576379055975,
      "loss": 2.0379,
      "step": 86932
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.8210896849632263,
      "learning_rate": 0.0004127919733612952,
      "loss": 2.1752,
      "step": 86933
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.8909057974815369,
      "learning_rate": 0.00041278818291106057,
      "loss": 2.0626,
      "step": 86934
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.8968286514282227,
      "learning_rate": 0.00041278439243985635,
      "loss": 2.3423,
      "step": 86935
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.8340248465538025,
      "learning_rate": 0.0004127806019476835,
      "loss": 2.1113,
      "step": 86936
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.8513445854187012,
      "learning_rate": 0.0004127768114345426,
      "loss": 2.0062,
      "step": 86937
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.8293487429618835,
      "learning_rate": 0.00041277302090043434,
      "loss": 2.0535,
      "step": 86938
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.9458770751953125,
      "learning_rate": 0.0004127692303453594,
      "loss": 2.0787,
      "step": 86939
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.825082004070282,
      "learning_rate": 0.0004127654397693185,
      "loss": 2.2008,
      "step": 86940
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.873499870300293,
      "learning_rate": 0.0004127616491723125,
      "loss": 1.9848,
      "step": 86941
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.8882287740707397,
      "learning_rate": 0.0004127578585543419,
      "loss": 2.2972,
      "step": 86942
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.7944682836532593,
      "learning_rate": 0.0004127540679154076,
      "loss": 2.1241,
      "step": 86943
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.7479270100593567,
      "learning_rate": 0.00041275027725551003,
      "loss": 2.1238,
      "step": 86944
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.8543065786361694,
      "learning_rate": 0.0004127464865746502,
      "loss": 2.1704,
      "step": 86945
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.8984062671661377,
      "learning_rate": 0.00041274269587282856,
      "loss": 2.2464,
      "step": 86946
    },
    {
      "epoch": 1.13,
      "grad_norm": 1.0403999090194702,
      "learning_rate": 0.00041273890515004596,
      "loss": 2.2606,
      "step": 86947
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.7917060256004333,
      "learning_rate": 0.00041273511440630313,
      "loss": 2.2662,
      "step": 86948
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.9404358267784119,
      "learning_rate": 0.0004127313236416006,
      "loss": 2.1968,
      "step": 86949
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.8212377429008484,
      "learning_rate": 0.0004127275328559392,
      "loss": 2.3214,
      "step": 86950
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.7859930396080017,
      "learning_rate": 0.0004127237420493197,
      "loss": 2.1026,
      "step": 86951
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.8355738520622253,
      "learning_rate": 0.0004127199512217428,
      "loss": 1.9932,
      "step": 86952
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.9896748661994934,
      "learning_rate": 0.00041271616037320895,
      "loss": 2.0622,
      "step": 86953
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.8464678525924683,
      "learning_rate": 0.0004127123695037191,
      "loss": 2.0034,
      "step": 86954
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.8669583201408386,
      "learning_rate": 0.00041270857861327396,
      "loss": 2.0173,
      "step": 86955
    },
    {
      "epoch": 1.13,
      "grad_norm": 1.0926746129989624,
      "learning_rate": 0.0004127047877018741,
      "loss": 2.1338,
      "step": 86956
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.9184262752532959,
      "learning_rate": 0.00041270099676952033,
      "loss": 2.1816,
      "step": 86957
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.9854763150215149,
      "learning_rate": 0.0004126972058162133,
      "loss": 2.1277,
      "step": 86958
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.955980122089386,
      "learning_rate": 0.00041269341484195376,
      "loss": 2.2483,
      "step": 86959
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.9641957879066467,
      "learning_rate": 0.0004126896238467423,
      "loss": 2.2668,
      "step": 86960
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.8007369637489319,
      "learning_rate": 0.0004126858328305799,
      "loss": 2.1471,
      "step": 86961
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.8563247919082642,
      "learning_rate": 0.0004126820417934669,
      "loss": 2.1823,
      "step": 86962
    },
    {
      "epoch": 1.13,
      "grad_norm": 1.1670851707458496,
      "learning_rate": 0.00041267825073540426,
      "loss": 2.2811,
      "step": 86963
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.8209275603294373,
      "learning_rate": 0.00041267445965639257,
      "loss": 2.2287,
      "step": 86964
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.8665385246276855,
      "learning_rate": 0.00041267066855643256,
      "loss": 2.1499,
      "step": 86965
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.8773414492607117,
      "learning_rate": 0.00041266687743552496,
      "loss": 2.1583,
      "step": 86966
    },
    {
      "epoch": 1.13,
      "grad_norm": 1.0349817276000977,
      "learning_rate": 0.0004126630862936705,
      "loss": 2.4234,
      "step": 86967
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.8071295619010925,
      "learning_rate": 0.00041265929513086984,
      "loss": 2.108,
      "step": 86968
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.9259698987007141,
      "learning_rate": 0.0004126555039471236,
      "loss": 2.252,
      "step": 86969
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.8522130846977234,
      "learning_rate": 0.0004126517127424327,
      "loss": 2.2824,
      "step": 86970
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.8760788440704346,
      "learning_rate": 0.0004126479215167976,
      "loss": 2.2401,
      "step": 86971
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.7896451354026794,
      "learning_rate": 0.0004126441302702192,
      "loss": 2.0841,
      "step": 86972
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.9146748185157776,
      "learning_rate": 0.0004126403390026981,
      "loss": 2.4303,
      "step": 86973
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.7842462658882141,
      "learning_rate": 0.0004126365477142351,
      "loss": 2.0077,
      "step": 86974
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.9189326167106628,
      "learning_rate": 0.0004126327564048307,
      "loss": 2.0165,
      "step": 86975
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.8542414903640747,
      "learning_rate": 0.00041262896507448594,
      "loss": 2.4071,
      "step": 86976
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.808866560459137,
      "learning_rate": 0.00041262517372320115,
      "loss": 2.084,
      "step": 86977
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.7973388433456421,
      "learning_rate": 0.00041262138235097733,
      "loss": 2.2254,
      "step": 86978
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.7998607158660889,
      "learning_rate": 0.00041261759095781496,
      "loss": 2.1644,
      "step": 86979
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.798501193523407,
      "learning_rate": 0.0004126137995437149,
      "loss": 2.0198,
      "step": 86980
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.721464991569519,
      "learning_rate": 0.0004126100081086779,
      "loss": 2.0416,
      "step": 86981
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.7689359784126282,
      "learning_rate": 0.00041260621665270444,
      "loss": 1.9724,
      "step": 86982
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.8294751644134521,
      "learning_rate": 0.0004126024251757955,
      "loss": 2.4021,
      "step": 86983
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.8170986175537109,
      "learning_rate": 0.0004125986336779515,
      "loss": 2.1783,
      "step": 86984
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.8460155129432678,
      "learning_rate": 0.00041259484215917336,
      "loss": 2.1005,
      "step": 86985
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.8214467763900757,
      "learning_rate": 0.0004125910506194617,
      "loss": 2.1622,
      "step": 86986
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.8131238222122192,
      "learning_rate": 0.00041258725905881727,
      "loss": 2.3278,
      "step": 86987
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.8999044895172119,
      "learning_rate": 0.0004125834674772407,
      "loss": 2.1855,
      "step": 86988
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.8334996700286865,
      "learning_rate": 0.0004125796758747328,
      "loss": 2.0623,
      "step": 86989
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.7590895295143127,
      "learning_rate": 0.00041257588425129413,
      "loss": 2.2613,
      "step": 86990
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.7777339816093445,
      "learning_rate": 0.0004125720926069255,
      "loss": 2.2037,
      "step": 86991
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.8223711848258972,
      "learning_rate": 0.0004125683009416276,
      "loss": 2.1736,
      "step": 86992
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.7761660218238831,
      "learning_rate": 0.00041256450925540117,
      "loss": 2.2771,
      "step": 86993
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.8464371562004089,
      "learning_rate": 0.00041256071754824686,
      "loss": 2.0104,
      "step": 86994
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.7944809794425964,
      "learning_rate": 0.0004125569258201654,
      "loss": 2.2864,
      "step": 86995
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.7872968912124634,
      "learning_rate": 0.0004125531340711574,
      "loss": 2.1589,
      "step": 86996
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.8550615906715393,
      "learning_rate": 0.00041254934230122367,
      "loss": 2.312,
      "step": 86997
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.9341951608657837,
      "learning_rate": 0.00041254555051036494,
      "loss": 2.2307,
      "step": 86998
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.8060091137886047,
      "learning_rate": 0.00041254175869858187,
      "loss": 2.3278,
      "step": 86999
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.8556985855102539,
      "learning_rate": 0.0004125379668658751,
      "loss": 2.0612,
      "step": 87000
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.7819380164146423,
      "learning_rate": 0.00041253417501224546,
      "loss": 2.1089,
      "step": 87001
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.7783997058868408,
      "learning_rate": 0.0004125303831376936,
      "loss": 2.1062,
      "step": 87002
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.8031674027442932,
      "learning_rate": 0.0004125265912422202,
      "loss": 2.1444,
      "step": 87003
    },
    {
      "epoch": 1.13,
      "grad_norm": 1.1238058805465698,
      "learning_rate": 0.000412522799325826,
      "loss": 1.971,
      "step": 87004
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.854595959186554,
      "learning_rate": 0.00041251900738851165,
      "loss": 2.1623,
      "step": 87005
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.815805196762085,
      "learning_rate": 0.0004125152154302778,
      "loss": 2.1584,
      "step": 87006
    },
    {
      "epoch": 1.13,
      "grad_norm": 1.1023820638656616,
      "learning_rate": 0.00041251142345112546,
      "loss": 1.9927,
      "step": 87007
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.9968609809875488,
      "learning_rate": 0.00041250763145105496,
      "loss": 2.2627,
      "step": 87008
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.84231036901474,
      "learning_rate": 0.0004125038394300672,
      "loss": 2.0798,
      "step": 87009
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.9570745229721069,
      "learning_rate": 0.00041250004738816294,
      "loss": 2.3754,
      "step": 87010
    },
    {
      "epoch": 1.13,
      "grad_norm": 1.0417202711105347,
      "learning_rate": 0.00041249625532534267,
      "loss": 2.0742,
      "step": 87011
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.9073994159698486,
      "learning_rate": 0.0004124924632416073,
      "loss": 1.8757,
      "step": 87012
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.8514060378074646,
      "learning_rate": 0.0004124886711369575,
      "loss": 2.1532,
      "step": 87013
    },
    {
      "epoch": 1.13,
      "grad_norm": 1.1209405660629272,
      "learning_rate": 0.0004124848790113938,
      "loss": 1.9575,
      "step": 87014
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.9489627480506897,
      "learning_rate": 0.0004124810868649171,
      "loss": 2.1645,
      "step": 87015
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.8492171168327332,
      "learning_rate": 0.0004124772946975281,
      "loss": 2.438,
      "step": 87016
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.9449072480201721,
      "learning_rate": 0.00041247350250922743,
      "loss": 2.1024,
      "step": 87017
    },
    {
      "epoch": 1.13,
      "grad_norm": 1.0508986711502075,
      "learning_rate": 0.00041246971030001574,
      "loss": 2.3393,
      "step": 87018
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.8520936965942383,
      "learning_rate": 0.0004124659180698939,
      "loss": 2.2502,
      "step": 87019
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.847916305065155,
      "learning_rate": 0.00041246212581886243,
      "loss": 2.1801,
      "step": 87020
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.8120469450950623,
      "learning_rate": 0.00041245833354692217,
      "loss": 2.2519,
      "step": 87021
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.8771001696586609,
      "learning_rate": 0.0004124545412540738,
      "loss": 2.2135,
      "step": 87022
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.8507506251335144,
      "learning_rate": 0.000412450748940318,
      "loss": 2.178,
      "step": 87023
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.9068415760993958,
      "learning_rate": 0.0004124469566056555,
      "loss": 2.2004,
      "step": 87024
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.8024471402168274,
      "learning_rate": 0.0004124431642500869,
      "loss": 2.0837,
      "step": 87025
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.9413312077522278,
      "learning_rate": 0.00041243937187361313,
      "loss": 2.2526,
      "step": 87026
    },
    {
      "epoch": 1.13,
      "grad_norm": 1.1804733276367188,
      "learning_rate": 0.0004124355794762347,
      "loss": 2.2815,
      "step": 87027
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.8261745572090149,
      "learning_rate": 0.00041243178705795234,
      "loss": 2.3625,
      "step": 87028
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.8843612670898438,
      "learning_rate": 0.00041242799461876684,
      "loss": 2.343,
      "step": 87029
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.8766646981239319,
      "learning_rate": 0.00041242420215867873,
      "loss": 2.1656,
      "step": 87030
    },
    {
      "epoch": 1.13,
      "grad_norm": 1.0047450065612793,
      "learning_rate": 0.00041242040967768895,
      "loss": 2.0469,
      "step": 87031
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.799676239490509,
      "learning_rate": 0.000412416617175798,
      "loss": 2.0305,
      "step": 87032
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.9158011674880981,
      "learning_rate": 0.00041241282465300687,
      "loss": 1.9811,
      "step": 87033
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.9721419811248779,
      "learning_rate": 0.0004124090321093159,
      "loss": 2.1467,
      "step": 87034
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.9193576574325562,
      "learning_rate": 0.000412405239544726,
      "loss": 2.2928,
      "step": 87035
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.8122303485870361,
      "learning_rate": 0.0004124014469592379,
      "loss": 2.2294,
      "step": 87036
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.8605232834815979,
      "learning_rate": 0.00041239765435285223,
      "loss": 2.173,
      "step": 87037
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.987532377243042,
      "learning_rate": 0.0004123938617255697,
      "loss": 2.2096,
      "step": 87038
    },
    {
      "epoch": 1.13,
      "grad_norm": 1.0012142658233643,
      "learning_rate": 0.000412390069077391,
      "loss": 2.2813,
      "step": 87039
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.8562092185020447,
      "learning_rate": 0.0004123862764083169,
      "loss": 1.9612,
      "step": 87040
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.8536286354064941,
      "learning_rate": 0.0004123824837183481,
      "loss": 2.0303,
      "step": 87041
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.8790737986564636,
      "learning_rate": 0.0004123786910074852,
      "loss": 2.3888,
      "step": 87042
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.8152526021003723,
      "learning_rate": 0.00041237489827572905,
      "loss": 2.1083,
      "step": 87043
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.8579559922218323,
      "learning_rate": 0.00041237110552308026,
      "loss": 2.3346,
      "step": 87044
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.7496479153633118,
      "learning_rate": 0.00041236731274953955,
      "loss": 2.1363,
      "step": 87045
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.7915471196174622,
      "learning_rate": 0.0004123635199551076,
      "loss": 2.1535,
      "step": 87046
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.9775605201721191,
      "learning_rate": 0.0004123597271397853,
      "loss": 2.3769,
      "step": 87047
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.8718344569206238,
      "learning_rate": 0.00041235593430357303,
      "loss": 2.2335,
      "step": 87048
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.9014380574226379,
      "learning_rate": 0.0004123521414464717,
      "loss": 2.1253,
      "step": 87049
    },
    {
      "epoch": 1.13,
      "grad_norm": 1.0756186246871948,
      "learning_rate": 0.0004123483485684821,
      "loss": 2.2316,
      "step": 87050
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.8648675084114075,
      "learning_rate": 0.0004123445556696047,
      "loss": 2.0952,
      "step": 87051
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.8223294019699097,
      "learning_rate": 0.0004123407627498404,
      "loss": 2.3077,
      "step": 87052
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.8990137577056885,
      "learning_rate": 0.00041233696980918984,
      "loss": 2.2227,
      "step": 87053
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.8271760940551758,
      "learning_rate": 0.0004123331768476537,
      "loss": 2.2977,
      "step": 87054
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.9587163925170898,
      "learning_rate": 0.00041232938386523264,
      "loss": 1.9436,
      "step": 87055
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.8210166692733765,
      "learning_rate": 0.00041232559086192753,
      "loss": 2.1473,
      "step": 87056
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.817421555519104,
      "learning_rate": 0.00041232179783773885,
      "loss": 2.0288,
      "step": 87057
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.8522366285324097,
      "learning_rate": 0.0004123180047926675,
      "loss": 2.4079,
      "step": 87058
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.845655083656311,
      "learning_rate": 0.0004123142117267141,
      "loss": 2.221,
      "step": 87059
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.7675703763961792,
      "learning_rate": 0.0004123104186398794,
      "loss": 2.0742,
      "step": 87060
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.7996214628219604,
      "learning_rate": 0.000412306625532164,
      "loss": 2.0389,
      "step": 87061
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.8215267658233643,
      "learning_rate": 0.0004123028324035688,
      "loss": 2.0163,
      "step": 87062
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.8449508547782898,
      "learning_rate": 0.0004122990392540943,
      "loss": 1.9914,
      "step": 87063
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.8865154385566711,
      "learning_rate": 0.0004122952460837413,
      "loss": 2.1892,
      "step": 87064
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.8472089171409607,
      "learning_rate": 0.0004122914528925105,
      "loss": 2.216,
      "step": 87065
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.8300269842147827,
      "learning_rate": 0.00041228765968040254,
      "loss": 2.23,
      "step": 87066
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.7872598171234131,
      "learning_rate": 0.0004122838664474182,
      "loss": 2.4584,
      "step": 87067
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.8045649528503418,
      "learning_rate": 0.00041228007319355826,
      "loss": 1.9876,
      "step": 87068
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.789890706539154,
      "learning_rate": 0.0004122762799188233,
      "loss": 2.3193,
      "step": 87069
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.8184495568275452,
      "learning_rate": 0.000412272486623214,
      "loss": 2.0085,
      "step": 87070
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.8735399842262268,
      "learning_rate": 0.0004122686933067312,
      "loss": 2.1359,
      "step": 87071
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.8413404226303101,
      "learning_rate": 0.0004122648999693754,
      "loss": 2.2579,
      "step": 87072
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.8224233984947205,
      "learning_rate": 0.00041226110661114757,
      "loss": 2.2357,
      "step": 87073
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.8157787322998047,
      "learning_rate": 0.0004122573132320483,
      "loss": 2.0699,
      "step": 87074
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.77450031042099,
      "learning_rate": 0.0004122535198320782,
      "loss": 1.9761,
      "step": 87075
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.8655065894126892,
      "learning_rate": 0.0004122497264112381,
      "loss": 2.1138,
      "step": 87076
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.7827311754226685,
      "learning_rate": 0.0004122459329695286,
      "loss": 2.1451,
      "step": 87077
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.8895755410194397,
      "learning_rate": 0.00041224213950695053,
      "loss": 2.0843,
      "step": 87078
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.777336061000824,
      "learning_rate": 0.00041223834602350444,
      "loss": 2.208,
      "step": 87079
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.9582858681678772,
      "learning_rate": 0.00041223455251919124,
      "loss": 2.296,
      "step": 87080
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.8989657163619995,
      "learning_rate": 0.00041223075899401144,
      "loss": 2.2118,
      "step": 87081
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.7478519678115845,
      "learning_rate": 0.00041222696544796573,
      "loss": 2.0811,
      "step": 87082
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.8059048652648926,
      "learning_rate": 0.00041222317188105515,
      "loss": 2.1821,
      "step": 87083
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.7810732126235962,
      "learning_rate": 0.00041221937829327997,
      "loss": 2.1884,
      "step": 87084
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.8243290185928345,
      "learning_rate": 0.00041221558468464104,
      "loss": 2.232,
      "step": 87085
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.7572200894355774,
      "learning_rate": 0.0004122117910551393,
      "loss": 2.2099,
      "step": 87086
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.7890941500663757,
      "learning_rate": 0.0004122079974047752,
      "loss": 2.1536,
      "step": 87087
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.924301028251648,
      "learning_rate": 0.0004122042037335495,
      "loss": 2.2967,
      "step": 87088
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.8358349800109863,
      "learning_rate": 0.0004122004100414629,
      "loss": 2.3703,
      "step": 87089
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.7846478819847107,
      "learning_rate": 0.0004121966163285162,
      "loss": 2.0087,
      "step": 87090
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.7740612626075745,
      "learning_rate": 0.00041219282259471,
      "loss": 2.2873,
      "step": 87091
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.8416124582290649,
      "learning_rate": 0.000412189028840045,
      "loss": 2.2872,
      "step": 87092
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.7814735174179077,
      "learning_rate": 0.00041218523506452196,
      "loss": 2.2906,
      "step": 87093
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.9020609259605408,
      "learning_rate": 0.0004121814412681415,
      "loss": 1.9939,
      "step": 87094
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.865191638469696,
      "learning_rate": 0.00041217764745090453,
      "loss": 2.3018,
      "step": 87095
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.8384935855865479,
      "learning_rate": 0.0004121738536128115,
      "loss": 2.3532,
      "step": 87096
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.8316450119018555,
      "learning_rate": 0.00041217005975386327,
      "loss": 2.3214,
      "step": 87097
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.8448407649993896,
      "learning_rate": 0.00041216626587406056,
      "loss": 2.2472,
      "step": 87098
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.8066202998161316,
      "learning_rate": 0.000412162471973404,
      "loss": 2.0715,
      "step": 87099
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.7859165668487549,
      "learning_rate": 0.00041215867805189423,
      "loss": 2.1917,
      "step": 87100
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.8386105298995972,
      "learning_rate": 0.0004121548841095322,
      "loss": 2.1005,
      "step": 87101
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.7751619219779968,
      "learning_rate": 0.00041215109014631834,
      "loss": 2.0717,
      "step": 87102
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.7503206133842468,
      "learning_rate": 0.0004121472961622535,
      "loss": 2.2725,
      "step": 87103
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.7810369729995728,
      "learning_rate": 0.0004121435021573384,
      "loss": 2.0332,
      "step": 87104
    },
    {
      "epoch": 1.13,
      "grad_norm": 1.007077693939209,
      "learning_rate": 0.0004121397081315736,
      "loss": 2.1466,
      "step": 87105
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.7863987684249878,
      "learning_rate": 0.00041213591408496,
      "loss": 2.0722,
      "step": 87106
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.756098747253418,
      "learning_rate": 0.0004121321200174983,
      "loss": 2.2541,
      "step": 87107
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.8431797027587891,
      "learning_rate": 0.000412128325929189,
      "loss": 2.2302,
      "step": 87108
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.9636995196342468,
      "learning_rate": 0.000412124531820033,
      "loss": 1.9723,
      "step": 87109
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.8802265524864197,
      "learning_rate": 0.0004121207376900309,
      "loss": 2.1942,
      "step": 87110
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.850909411907196,
      "learning_rate": 0.0004121169435391834,
      "loss": 2.1898,
      "step": 87111
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.8481159210205078,
      "learning_rate": 0.0004121131493674913,
      "loss": 2.2077,
      "step": 87112
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.8012329339981079,
      "learning_rate": 0.0004121093551749551,
      "loss": 1.9407,
      "step": 87113
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.9261874556541443,
      "learning_rate": 0.0004121055609615758,
      "loss": 2.3689,
      "step": 87114
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.7969962954521179,
      "learning_rate": 0.000412101766727354,
      "loss": 2.1908,
      "step": 87115
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.82655930519104,
      "learning_rate": 0.0004120979724722903,
      "loss": 2.3226,
      "step": 87116
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.7205185890197754,
      "learning_rate": 0.0004120941781963855,
      "loss": 2.0108,
      "step": 87117
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.9031713008880615,
      "learning_rate": 0.00041209038389964015,
      "loss": 2.0696,
      "step": 87118
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.8715868592262268,
      "learning_rate": 0.00041208658958205525,
      "loss": 2.1645,
      "step": 87119
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.8033266663551331,
      "learning_rate": 0.00041208279524363123,
      "loss": 2.2224,
      "step": 87120
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.8556246757507324,
      "learning_rate": 0.000412079000884369,
      "loss": 2.1013,
      "step": 87121
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.8613374829292297,
      "learning_rate": 0.00041207520650426906,
      "loss": 2.1177,
      "step": 87122
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.8220230937004089,
      "learning_rate": 0.0004120714121033322,
      "loss": 2.1395,
      "step": 87123
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.8331857919692993,
      "learning_rate": 0.00041206761768155923,
      "loss": 2.1273,
      "step": 87124
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.8779845237731934,
      "learning_rate": 0.00041206382323895075,
      "loss": 2.4562,
      "step": 87125
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.8991895318031311,
      "learning_rate": 0.00041206002877550754,
      "loss": 2.2084,
      "step": 87126
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.9128933548927307,
      "learning_rate": 0.0004120562342912302,
      "loss": 2.1657,
      "step": 87127
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.839444100856781,
      "learning_rate": 0.00041205243978611957,
      "loss": 1.9695,
      "step": 87128
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.9046825766563416,
      "learning_rate": 0.00041204864526017617,
      "loss": 1.9935,
      "step": 87129
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.8162201046943665,
      "learning_rate": 0.00041204485071340085,
      "loss": 1.9781,
      "step": 87130
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.8520997166633606,
      "learning_rate": 0.0004120410561457943,
      "loss": 1.9919,
      "step": 87131
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.7862842679023743,
      "learning_rate": 0.00041203726155735723,
      "loss": 2.1798,
      "step": 87132
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.8770169615745544,
      "learning_rate": 0.00041203346694809024,
      "loss": 2.403,
      "step": 87133
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.8604494333267212,
      "learning_rate": 0.00041202967231799415,
      "loss": 2.3798,
      "step": 87134
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.7995538115501404,
      "learning_rate": 0.00041202587766706966,
      "loss": 2.0777,
      "step": 87135
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.8544203639030457,
      "learning_rate": 0.0004120220829953174,
      "loss": 2.3326,
      "step": 87136
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.8410939574241638,
      "learning_rate": 0.00041201828830273807,
      "loss": 2.1555,
      "step": 87137
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.7383425831794739,
      "learning_rate": 0.0004120144935893326,
      "loss": 2.0424,
      "step": 87138
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.9737675189971924,
      "learning_rate": 0.0004120106988551013,
      "loss": 2.0995,
      "step": 87139
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.8235531449317932,
      "learning_rate": 0.00041200690410004523,
      "loss": 2.3489,
      "step": 87140
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.8164818286895752,
      "learning_rate": 0.00041200310932416493,
      "loss": 2.3714,
      "step": 87141
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.7939198613166809,
      "learning_rate": 0.0004119993145274612,
      "loss": 2.3532,
      "step": 87142
    },
    {
      "epoch": 1.13,
      "grad_norm": 1.2698949575424194,
      "learning_rate": 0.0004119955197099346,
      "loss": 2.0487,
      "step": 87143
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.8634188771247864,
      "learning_rate": 0.000411991724871586,
      "loss": 2.1481,
      "step": 87144
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.759200930595398,
      "learning_rate": 0.0004119879300124159,
      "loss": 2.3536,
      "step": 87145
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.7611199021339417,
      "learning_rate": 0.0004119841351324252,
      "loss": 2.1731,
      "step": 87146
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.7748376727104187,
      "learning_rate": 0.00041198034023161463,
      "loss": 2.0275,
      "step": 87147
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.9273929595947266,
      "learning_rate": 0.0004119765453099847,
      "loss": 2.3012,
      "step": 87148
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.8648070096969604,
      "learning_rate": 0.0004119727503675362,
      "loss": 2.3644,
      "step": 87149
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.8169238567352295,
      "learning_rate": 0.0004119689554042699,
      "loss": 2.3189,
      "step": 87150
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.8118595480918884,
      "learning_rate": 0.00041196516042018644,
      "loss": 2.2263,
      "step": 87151
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.8776057958602905,
      "learning_rate": 0.00041196136541528654,
      "loss": 2.0537,
      "step": 87152
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.9436894655227661,
      "learning_rate": 0.00041195757038957093,
      "loss": 2.4744,
      "step": 87153
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.9076494574546814,
      "learning_rate": 0.00041195377534304026,
      "loss": 2.1948,
      "step": 87154
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.8213127851486206,
      "learning_rate": 0.00041194998027569525,
      "loss": 2.3419,
      "step": 87155
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.8879772424697876,
      "learning_rate": 0.0004119461851875367,
      "loss": 2.112,
      "step": 87156
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.7925137877464294,
      "learning_rate": 0.0004119423900785652,
      "loss": 2.0788,
      "step": 87157
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.8376350402832031,
      "learning_rate": 0.00041193859494878147,
      "loss": 2.0577,
      "step": 87158
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.772290050983429,
      "learning_rate": 0.0004119347997981863,
      "loss": 2.357,
      "step": 87159
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.8529401421546936,
      "learning_rate": 0.0004119310046267803,
      "loss": 2.2486,
      "step": 87160
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.7785086035728455,
      "learning_rate": 0.0004119272094345642,
      "loss": 2.223,
      "step": 87161
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.747085452079773,
      "learning_rate": 0.00041192341422153876,
      "loss": 2.0717,
      "step": 87162
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.8327189087867737,
      "learning_rate": 0.0004119196189877046,
      "loss": 2.3511,
      "step": 87163
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.733913004398346,
      "learning_rate": 0.0004119158237330625,
      "loss": 2.2761,
      "step": 87164
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.7265926599502563,
      "learning_rate": 0.0004119120284576132,
      "loss": 1.9829,
      "step": 87165
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.8307526111602783,
      "learning_rate": 0.00041190823316135725,
      "loss": 2.176,
      "step": 87166
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.8301194310188293,
      "learning_rate": 0.0004119044378442954,
      "loss": 2.2946,
      "step": 87167
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.8902557492256165,
      "learning_rate": 0.00041190064250642847,
      "loss": 2.2583,
      "step": 87168
    },
    {
      "epoch": 1.13,
      "grad_norm": 0.8725361227989197,
      "learning_rate": 0.00041189684714775714,
      "loss": 2.2981,
      "step": 87169
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.9213995337486267,
      "learning_rate": 0.000411893051768282,
      "loss": 2.2688,
      "step": 87170
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.7807721495628357,
      "learning_rate": 0.0004118892563680039,
      "loss": 2.2942,
      "step": 87171
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.8702874779701233,
      "learning_rate": 0.00041188546094692344,
      "loss": 2.1326,
      "step": 87172
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.872154712677002,
      "learning_rate": 0.00041188166550504133,
      "loss": 2.3219,
      "step": 87173
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.9070103168487549,
      "learning_rate": 0.00041187787004235836,
      "loss": 2.2161,
      "step": 87174
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.7617709636688232,
      "learning_rate": 0.0004118740745588752,
      "loss": 2.0841,
      "step": 87175
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.818535327911377,
      "learning_rate": 0.0004118702790545925,
      "loss": 2.2915,
      "step": 87176
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.8244105577468872,
      "learning_rate": 0.0004118664835295109,
      "loss": 2.2955,
      "step": 87177
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.9348508715629578,
      "learning_rate": 0.00041186268798363136,
      "loss": 2.1378,
      "step": 87178
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.8426584005355835,
      "learning_rate": 0.0004118588924169544,
      "loss": 2.2118,
      "step": 87179
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.887866199016571,
      "learning_rate": 0.0004118550968294807,
      "loss": 2.1737,
      "step": 87180
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.9809615015983582,
      "learning_rate": 0.0004118513012212111,
      "loss": 2.0636,
      "step": 87181
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.7929410338401794,
      "learning_rate": 0.0004118475055921462,
      "loss": 2.031,
      "step": 87182
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.8548387885093689,
      "learning_rate": 0.00041184370994228666,
      "loss": 2.2323,
      "step": 87183
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.9064220190048218,
      "learning_rate": 0.00041183991427163344,
      "loss": 2.1157,
      "step": 87184
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.9403924942016602,
      "learning_rate": 0.00041183611858018696,
      "loss": 2.1252,
      "step": 87185
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.8751833438873291,
      "learning_rate": 0.000411832322867948,
      "loss": 2.6266,
      "step": 87186
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.8909383416175842,
      "learning_rate": 0.0004118285271349174,
      "loss": 2.2674,
      "step": 87187
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.7865390777587891,
      "learning_rate": 0.0004118247313810957,
      "loss": 1.9211,
      "step": 87188
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.880352258682251,
      "learning_rate": 0.00041182093560648364,
      "loss": 1.9985,
      "step": 87189
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.7857416868209839,
      "learning_rate": 0.00041181713981108207,
      "loss": 2.3026,
      "step": 87190
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.8846520185470581,
      "learning_rate": 0.0004118133439948914,
      "loss": 2.1938,
      "step": 87191
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.8596081137657166,
      "learning_rate": 0.00041180954815791263,
      "loss": 2.3344,
      "step": 87192
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.7923193573951721,
      "learning_rate": 0.00041180575230014645,
      "loss": 2.3517,
      "step": 87193
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.7695015668869019,
      "learning_rate": 0.0004118019564215933,
      "loss": 2.0803,
      "step": 87194
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.8736974596977234,
      "learning_rate": 0.0004117981605222542,
      "loss": 2.036,
      "step": 87195
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.8601281046867371,
      "learning_rate": 0.0004117943646021296,
      "loss": 2.2083,
      "step": 87196
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.8480299115180969,
      "learning_rate": 0.00041179056866122037,
      "loss": 2.4248,
      "step": 87197
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.7684424519538879,
      "learning_rate": 0.0004117867726995271,
      "loss": 2.4149,
      "step": 87198
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.786960244178772,
      "learning_rate": 0.0004117829767170507,
      "loss": 2.1268,
      "step": 87199
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.8358076214790344,
      "learning_rate": 0.0004117791807137916,
      "loss": 2.2223,
      "step": 87200
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.8498647809028625,
      "learning_rate": 0.00041177538468975067,
      "loss": 2.3415,
      "step": 87201
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.7949463725090027,
      "learning_rate": 0.00041177158864492866,
      "loss": 2.1921,
      "step": 87202
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.7434045076370239,
      "learning_rate": 0.0004117677925793261,
      "loss": 1.8517,
      "step": 87203
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.8339365720748901,
      "learning_rate": 0.0004117639964929438,
      "loss": 1.9996,
      "step": 87204
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.8184765577316284,
      "learning_rate": 0.00041176020038578257,
      "loss": 2.2442,
      "step": 87205
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.794263482093811,
      "learning_rate": 0.00041175640425784296,
      "loss": 2.1331,
      "step": 87206
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.8215741515159607,
      "learning_rate": 0.0004117526081091257,
      "loss": 2.0909,
      "step": 87207
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.8432920575141907,
      "learning_rate": 0.0004117488119396315,
      "loss": 2.41,
      "step": 87208
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.762771725654602,
      "learning_rate": 0.00041174501574936116,
      "loss": 2.0548,
      "step": 87209
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.9255907535552979,
      "learning_rate": 0.00041174121953831523,
      "loss": 2.2548,
      "step": 87210
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.8308632373809814,
      "learning_rate": 0.00041173742330649453,
      "loss": 2.2108,
      "step": 87211
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.8178333640098572,
      "learning_rate": 0.00041173362705389975,
      "loss": 2.3663,
      "step": 87212
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.8334718346595764,
      "learning_rate": 0.00041172983078053156,
      "loss": 2.2262,
      "step": 87213
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.7691740393638611,
      "learning_rate": 0.0004117260344863907,
      "loss": 2.0326,
      "step": 87214
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.8371747732162476,
      "learning_rate": 0.0004117222381714778,
      "loss": 2.1574,
      "step": 87215
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.7126084566116333,
      "learning_rate": 0.00041171844183579375,
      "loss": 2.2299,
      "step": 87216
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.743880569934845,
      "learning_rate": 0.00041171464547933907,
      "loss": 2.0678,
      "step": 87217
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.8358493447303772,
      "learning_rate": 0.0004117108491021145,
      "loss": 2.3561,
      "step": 87218
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.8588376641273499,
      "learning_rate": 0.0004117070527041208,
      "loss": 2.3828,
      "step": 87219
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.8675276041030884,
      "learning_rate": 0.0004117032562853587,
      "loss": 2.1743,
      "step": 87220
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.9116283655166626,
      "learning_rate": 0.0004116994598458288,
      "loss": 2.2059,
      "step": 87221
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.749286413192749,
      "learning_rate": 0.0004116956633855318,
      "loss": 2.2096,
      "step": 87222
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.8903298377990723,
      "learning_rate": 0.0004116918669044686,
      "loss": 2.2201,
      "step": 87223
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.9994492530822754,
      "learning_rate": 0.00041168807040263967,
      "loss": 2.1662,
      "step": 87224
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.8295034170150757,
      "learning_rate": 0.00041168427388004586,
      "loss": 2.2115,
      "step": 87225
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.8262650370597839,
      "learning_rate": 0.00041168047733668785,
      "loss": 2.205,
      "step": 87226
    },
    {
      "epoch": 1.14,
      "grad_norm": 1.1334915161132812,
      "learning_rate": 0.0004116766807725663,
      "loss": 2.4375,
      "step": 87227
    },
    {
      "epoch": 1.14,
      "grad_norm": 1.018241047859192,
      "learning_rate": 0.0004116728841876819,
      "loss": 2.266,
      "step": 87228
    },
    {
      "epoch": 1.14,
      "grad_norm": 1.079798936843872,
      "learning_rate": 0.0004116690875820355,
      "loss": 2.2283,
      "step": 87229
    },
    {
      "epoch": 1.14,
      "grad_norm": 1.0551891326904297,
      "learning_rate": 0.00041166529095562767,
      "loss": 2.3065,
      "step": 87230
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.909145176410675,
      "learning_rate": 0.00041166149430845914,
      "loss": 2.117,
      "step": 87231
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.8346996307373047,
      "learning_rate": 0.0004116576976405307,
      "loss": 2.0289,
      "step": 87232
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.8844238519668579,
      "learning_rate": 0.00041165390095184297,
      "loss": 2.2689,
      "step": 87233
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.898345410823822,
      "learning_rate": 0.00041165010424239657,
      "loss": 2.3498,
      "step": 87234
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.7834979891777039,
      "learning_rate": 0.00041164630751219246,
      "loss": 2.1193,
      "step": 87235
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.8801631331443787,
      "learning_rate": 0.00041164251076123104,
      "loss": 2.3107,
      "step": 87236
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.8088243007659912,
      "learning_rate": 0.0004116387139895132,
      "loss": 1.8963,
      "step": 87237
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.8780319690704346,
      "learning_rate": 0.00041163491719703975,
      "loss": 2.4912,
      "step": 87238
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.8257285952568054,
      "learning_rate": 0.00041163112038381113,
      "loss": 2.3503,
      "step": 87239
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.8247209787368774,
      "learning_rate": 0.0004116273235498282,
      "loss": 2.2552,
      "step": 87240
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.8287224173545837,
      "learning_rate": 0.0004116235266950917,
      "loss": 2.128,
      "step": 87241
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.9544863104820251,
      "learning_rate": 0.0004116197298196022,
      "loss": 2.323,
      "step": 87242
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.7863388061523438,
      "learning_rate": 0.0004116159329233605,
      "loss": 2.3849,
      "step": 87243
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.8333972692489624,
      "learning_rate": 0.0004116121360063673,
      "loss": 1.9875,
      "step": 87244
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.9157539010047913,
      "learning_rate": 0.0004116083390686234,
      "loss": 2.0226,
      "step": 87245
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.8820618391036987,
      "learning_rate": 0.0004116045421101293,
      "loss": 2.1768,
      "step": 87246
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.8127277493476868,
      "learning_rate": 0.0004116007451308858,
      "loss": 1.9541,
      "step": 87247
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.8654027581214905,
      "learning_rate": 0.00041159694813089364,
      "loss": 2.2384,
      "step": 87248
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.9979334473609924,
      "learning_rate": 0.0004115931511101535,
      "loss": 2.1756,
      "step": 87249
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.8694819808006287,
      "learning_rate": 0.00041158935406866606,
      "loss": 2.317,
      "step": 87250
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.9010300040245056,
      "learning_rate": 0.00041158555700643215,
      "loss": 1.9702,
      "step": 87251
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.8908904790878296,
      "learning_rate": 0.00041158175992345224,
      "loss": 2.2026,
      "step": 87252
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.8439866304397583,
      "learning_rate": 0.0004115779628197272,
      "loss": 2.09,
      "step": 87253
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.7895991206169128,
      "learning_rate": 0.0004115741656952578,
      "loss": 1.9405,
      "step": 87254
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.8757488131523132,
      "learning_rate": 0.00041157036855004464,
      "loss": 2.0126,
      "step": 87255
    },
    {
      "epoch": 1.14,
      "grad_norm": 1.0334365367889404,
      "learning_rate": 0.00041156657138408836,
      "loss": 2.2029,
      "step": 87256
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.7960792779922485,
      "learning_rate": 0.00041156277419738987,
      "loss": 2.0569,
      "step": 87257
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.9066171646118164,
      "learning_rate": 0.0004115589769899497,
      "loss": 2.3217,
      "step": 87258
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.7926754355430603,
      "learning_rate": 0.0004115551797617685,
      "loss": 2.1134,
      "step": 87259
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.9847112894058228,
      "learning_rate": 0.0004115513825128472,
      "loss": 2.3573,
      "step": 87260
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.8925875425338745,
      "learning_rate": 0.0004115475852431864,
      "loss": 2.1626,
      "step": 87261
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.8225168585777283,
      "learning_rate": 0.00041154378795278673,
      "loss": 2.3106,
      "step": 87262
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.8529929518699646,
      "learning_rate": 0.00041153999064164905,
      "loss": 1.9834,
      "step": 87263
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.7465571761131287,
      "learning_rate": 0.0004115361933097739,
      "loss": 2.1185,
      "step": 87264
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.855442225933075,
      "learning_rate": 0.00041153239595716207,
      "loss": 1.933,
      "step": 87265
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.8449817299842834,
      "learning_rate": 0.0004115285985838144,
      "loss": 2.0261,
      "step": 87266
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.8399947881698608,
      "learning_rate": 0.00041152480118973134,
      "loss": 2.2802,
      "step": 87267
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.7894905209541321,
      "learning_rate": 0.0004115210037749137,
      "loss": 2.0349,
      "step": 87268
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.8107722401618958,
      "learning_rate": 0.0004115172063393622,
      "loss": 2.005,
      "step": 87269
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.811379075050354,
      "learning_rate": 0.00041151340888307753,
      "loss": 2.0878,
      "step": 87270
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.842688798904419,
      "learning_rate": 0.00041150961140606044,
      "loss": 1.9542,
      "step": 87271
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.8552413582801819,
      "learning_rate": 0.00041150581390831173,
      "loss": 2.2633,
      "step": 87272
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.80857914686203,
      "learning_rate": 0.0004115020163898318,
      "loss": 2.3644,
      "step": 87273
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.9019544124603271,
      "learning_rate": 0.0004114982188506216,
      "loss": 2.3035,
      "step": 87274
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.768060564994812,
      "learning_rate": 0.00041149442129068185,
      "loss": 2.1709,
      "step": 87275
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.7913256883621216,
      "learning_rate": 0.0004114906237100131,
      "loss": 2.2178,
      "step": 87276
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.8277349472045898,
      "learning_rate": 0.00041148682610861616,
      "loss": 2.2282,
      "step": 87277
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.7850583791732788,
      "learning_rate": 0.0004114830284864918,
      "loss": 2.1586,
      "step": 87278
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.7654302716255188,
      "learning_rate": 0.00041147923084364056,
      "loss": 2.1654,
      "step": 87279
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.8745684027671814,
      "learning_rate": 0.0004114754331800631,
      "loss": 2.2906,
      "step": 87280
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.8552700877189636,
      "learning_rate": 0.0004114716354957605,
      "loss": 2.2787,
      "step": 87281
    },
    {
      "epoch": 1.14,
      "grad_norm": 1.0439167022705078,
      "learning_rate": 0.00041146783779073305,
      "loss": 1.9988,
      "step": 87282
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.8053336143493652,
      "learning_rate": 0.00041146404006498166,
      "loss": 2.2955,
      "step": 87283
    },
    {
      "epoch": 1.14,
      "grad_norm": 1.0193910598754883,
      "learning_rate": 0.00041146024231850705,
      "loss": 2.1968,
      "step": 87284
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.8900471329689026,
      "learning_rate": 0.0004114564445513098,
      "loss": 2.3246,
      "step": 87285
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.8218262791633606,
      "learning_rate": 0.00041145264676339076,
      "loss": 2.0731,
      "step": 87286
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.8487718105316162,
      "learning_rate": 0.00041144884895475054,
      "loss": 2.2851,
      "step": 87287
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.9948673248291016,
      "learning_rate": 0.0004114450511253899,
      "loss": 2.3475,
      "step": 87288
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.875840961933136,
      "learning_rate": 0.0004114412532753095,
      "loss": 2.1281,
      "step": 87289
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.8904297947883606,
      "learning_rate": 0.00041143745540451,
      "loss": 2.164,
      "step": 87290
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.9162677526473999,
      "learning_rate": 0.0004114336575129923,
      "loss": 2.0989,
      "step": 87291
    },
    {
      "epoch": 1.14,
      "grad_norm": 1.0400885343551636,
      "learning_rate": 0.00041142985960075683,
      "loss": 2.265,
      "step": 87292
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.9403281211853027,
      "learning_rate": 0.0004114260616678046,
      "loss": 2.1784,
      "step": 87293
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.8924828171730042,
      "learning_rate": 0.0004114222637141361,
      "loss": 2.3315,
      "step": 87294
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.8310673832893372,
      "learning_rate": 0.0004114184657397521,
      "loss": 2.2029,
      "step": 87295
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.8652476072311401,
      "learning_rate": 0.0004114146677446533,
      "loss": 1.9119,
      "step": 87296
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.7950506806373596,
      "learning_rate": 0.00041141086972884035,
      "loss": 2.2615,
      "step": 87297
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.8534578680992126,
      "learning_rate": 0.00041140707169231417,
      "loss": 1.9904,
      "step": 87298
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.8332827687263489,
      "learning_rate": 0.0004114032736350752,
      "loss": 2.3628,
      "step": 87299
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.9086757302284241,
      "learning_rate": 0.00041139947555712423,
      "loss": 2.239,
      "step": 87300
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.8052884340286255,
      "learning_rate": 0.00041139567745846205,
      "loss": 2.3306,
      "step": 87301
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.8583587408065796,
      "learning_rate": 0.00041139187933908933,
      "loss": 2.3012,
      "step": 87302
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.7578859329223633,
      "learning_rate": 0.00041138808119900666,
      "loss": 2.4332,
      "step": 87303
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.8189558386802673,
      "learning_rate": 0.0004113842830382149,
      "loss": 2.2833,
      "step": 87304
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.7588042616844177,
      "learning_rate": 0.0004113804848567147,
      "loss": 2.3552,
      "step": 87305
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.8634153008460999,
      "learning_rate": 0.0004113766866545068,
      "loss": 2.078,
      "step": 87306
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.7831700444221497,
      "learning_rate": 0.0004113728884315919,
      "loss": 2.1048,
      "step": 87307
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.9461381435394287,
      "learning_rate": 0.0004113690901879706,
      "loss": 2.1044,
      "step": 87308
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.9323679804801941,
      "learning_rate": 0.0004113652919236437,
      "loss": 1.8232,
      "step": 87309
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.7889778017997742,
      "learning_rate": 0.0004113614936386119,
      "loss": 2.2288,
      "step": 87310
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.790583610534668,
      "learning_rate": 0.00041135769533287586,
      "loss": 2.1735,
      "step": 87311
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.8305160999298096,
      "learning_rate": 0.00041135389700643635,
      "loss": 2.0861,
      "step": 87312
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.8293344378471375,
      "learning_rate": 0.0004113500986592941,
      "loss": 2.2675,
      "step": 87313
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.905822217464447,
      "learning_rate": 0.0004113463002914497,
      "loss": 2.1649,
      "step": 87314
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.8601540327072144,
      "learning_rate": 0.000411342501902904,
      "loss": 2.0349,
      "step": 87315
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.987397313117981,
      "learning_rate": 0.00041133870349365756,
      "loss": 2.2121,
      "step": 87316
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.9503358006477356,
      "learning_rate": 0.0004113349050637111,
      "loss": 2.1739,
      "step": 87317
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.7949160933494568,
      "learning_rate": 0.00041133110661306546,
      "loss": 2.2552,
      "step": 87318
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.954383909702301,
      "learning_rate": 0.0004113273081417213,
      "loss": 2.2618,
      "step": 87319
    },
    {
      "epoch": 1.14,
      "grad_norm": 1.0125622749328613,
      "learning_rate": 0.0004113235096496792,
      "loss": 2.1072,
      "step": 87320
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.927730143070221,
      "learning_rate": 0.00041131971113694,
      "loss": 2.2134,
      "step": 87321
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.7929105162620544,
      "learning_rate": 0.00041131591260350436,
      "loss": 2.0825,
      "step": 87322
    },
    {
      "epoch": 1.14,
      "grad_norm": 1.0165891647338867,
      "learning_rate": 0.00041131211404937297,
      "loss": 2.2643,
      "step": 87323
    },
    {
      "epoch": 1.14,
      "grad_norm": 1.0002467632293701,
      "learning_rate": 0.00041130831547454665,
      "loss": 2.1941,
      "step": 87324
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.8230862021446228,
      "learning_rate": 0.0004113045168790259,
      "loss": 2.2921,
      "step": 87325
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.8167006373405457,
      "learning_rate": 0.0004113007182628116,
      "loss": 2.1551,
      "step": 87326
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.8551967740058899,
      "learning_rate": 0.0004112969196259044,
      "loss": 2.0953,
      "step": 87327
    },
    {
      "epoch": 1.14,
      "grad_norm": 1.0886908769607544,
      "learning_rate": 0.00041129312096830495,
      "loss": 2.0677,
      "step": 87328
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.8591325879096985,
      "learning_rate": 0.000411289322290014,
      "loss": 2.2607,
      "step": 87329
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.8882420659065247,
      "learning_rate": 0.0004112855235910324,
      "loss": 2.1754,
      "step": 87330
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.8551324605941772,
      "learning_rate": 0.0004112817248713606,
      "loss": 2.005,
      "step": 87331
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.9476715922355652,
      "learning_rate": 0.00041127792613099946,
      "loss": 2.3389,
      "step": 87332
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.8032691478729248,
      "learning_rate": 0.00041127412736994965,
      "loss": 2.3489,
      "step": 87333
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.867967426776886,
      "learning_rate": 0.00041127032858821185,
      "loss": 2.1362,
      "step": 87334
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.8976536393165588,
      "learning_rate": 0.0004112665297857868,
      "loss": 2.4109,
      "step": 87335
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.7909066081047058,
      "learning_rate": 0.00041126273096267523,
      "loss": 2.373,
      "step": 87336
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.852503776550293,
      "learning_rate": 0.00041125893211887776,
      "loss": 2.2452,
      "step": 87337
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.8171125054359436,
      "learning_rate": 0.0004112551332543953,
      "loss": 2.2942,
      "step": 87338
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.893780529499054,
      "learning_rate": 0.0004112513343692283,
      "loss": 2.1174,
      "step": 87339
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.931391716003418,
      "learning_rate": 0.0004112475354633775,
      "loss": 2.2681,
      "step": 87340
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.7652585506439209,
      "learning_rate": 0.00041124373653684377,
      "loss": 2.2435,
      "step": 87341
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.8288301825523376,
      "learning_rate": 0.00041123993758962783,
      "loss": 2.0824,
      "step": 87342
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.8246221542358398,
      "learning_rate": 0.0004112361386217302,
      "loss": 2.2482,
      "step": 87343
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.868013322353363,
      "learning_rate": 0.0004112323396331516,
      "loss": 1.9336,
      "step": 87344
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.9450108408927917,
      "learning_rate": 0.0004112285406238929,
      "loss": 1.8559,
      "step": 87345
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.8879055976867676,
      "learning_rate": 0.00041122474159395467,
      "loss": 2.1153,
      "step": 87346
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.8218618631362915,
      "learning_rate": 0.0004112209425433376,
      "loss": 2.0748,
      "step": 87347
    },
    {
      "epoch": 1.14,
      "grad_norm": 1.0184693336486816,
      "learning_rate": 0.0004112171434720426,
      "loss": 2.1831,
      "step": 87348
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.945509672164917,
      "learning_rate": 0.0004112133443800702,
      "loss": 2.3987,
      "step": 87349
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.8544702529907227,
      "learning_rate": 0.00041120954526742106,
      "loss": 2.095,
      "step": 87350
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.994911789894104,
      "learning_rate": 0.00041120574613409607,
      "loss": 2.2777,
      "step": 87351
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.7782749533653259,
      "learning_rate": 0.0004112019469800957,
      "loss": 2.1611,
      "step": 87352
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.8826868534088135,
      "learning_rate": 0.0004111981478054209,
      "loss": 2.3715,
      "step": 87353
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.9068272709846497,
      "learning_rate": 0.0004111943486100722,
      "loss": 2.3365,
      "step": 87354
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.8705130815505981,
      "learning_rate": 0.0004111905493940504,
      "loss": 2.1291,
      "step": 87355
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.8089882731437683,
      "learning_rate": 0.00041118675015735617,
      "loss": 2.0473,
      "step": 87356
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.8391091227531433,
      "learning_rate": 0.00041118295089999027,
      "loss": 2.2535,
      "step": 87357
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.7926360964775085,
      "learning_rate": 0.0004111791516219533,
      "loss": 2.1319,
      "step": 87358
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.8596510291099548,
      "learning_rate": 0.00041117535232324605,
      "loss": 2.185,
      "step": 87359
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.8097468614578247,
      "learning_rate": 0.00041117155300386926,
      "loss": 2.0383,
      "step": 87360
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.8090812563896179,
      "learning_rate": 0.0004111677536638235,
      "loss": 2.093,
      "step": 87361
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.8614538908004761,
      "learning_rate": 0.00041116395430310955,
      "loss": 2.2843,
      "step": 87362
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.8470592498779297,
      "learning_rate": 0.0004111601549217282,
      "loss": 1.9748,
      "step": 87363
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.7418425679206848,
      "learning_rate": 0.00041115635551967997,
      "loss": 2.2722,
      "step": 87364
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.8457484841346741,
      "learning_rate": 0.00041115255609696575,
      "loss": 2.0498,
      "step": 87365
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.868354320526123,
      "learning_rate": 0.0004111487566535862,
      "loss": 2.1894,
      "step": 87366
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.7872511148452759,
      "learning_rate": 0.0004111449571895419,
      "loss": 2.2239,
      "step": 87367
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.7802908420562744,
      "learning_rate": 0.0004111411577048337,
      "loss": 2.2378,
      "step": 87368
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.8987900018692017,
      "learning_rate": 0.00041113735819946234,
      "loss": 2.159,
      "step": 87369
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.894747257232666,
      "learning_rate": 0.00041113355867342837,
      "loss": 2.2052,
      "step": 87370
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.8406686782836914,
      "learning_rate": 0.0004111297591267326,
      "loss": 2.1847,
      "step": 87371
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.8274365663528442,
      "learning_rate": 0.00041112595955937573,
      "loss": 2.1537,
      "step": 87372
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.8194350004196167,
      "learning_rate": 0.00041112215997135836,
      "loss": 2.1954,
      "step": 87373
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.8326080441474915,
      "learning_rate": 0.00041111836036268124,
      "loss": 2.2917,
      "step": 87374
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.848549485206604,
      "learning_rate": 0.0004111145607333453,
      "loss": 2.2666,
      "step": 87375
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.7935220003128052,
      "learning_rate": 0.0004111107610833509,
      "loss": 1.9916,
      "step": 87376
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.826644778251648,
      "learning_rate": 0.000411106961412699,
      "loss": 2.242,
      "step": 87377
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.9493375420570374,
      "learning_rate": 0.0004111031617213902,
      "loss": 2.3728,
      "step": 87378
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.8376120924949646,
      "learning_rate": 0.00041109936200942526,
      "loss": 2.3667,
      "step": 87379
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.8967647552490234,
      "learning_rate": 0.0004110955622768048,
      "loss": 2.5377,
      "step": 87380
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.9352288842201233,
      "learning_rate": 0.0004110917625235296,
      "loss": 2.3272,
      "step": 87381
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.8790714144706726,
      "learning_rate": 0.0004110879627496004,
      "loss": 2.3087,
      "step": 87382
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.7763688564300537,
      "learning_rate": 0.00041108416295501773,
      "loss": 2.3068,
      "step": 87383
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.807979941368103,
      "learning_rate": 0.0004110803631397825,
      "loss": 2.0309,
      "step": 87384
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.8599072098731995,
      "learning_rate": 0.0004110765633038953,
      "loss": 2.3919,
      "step": 87385
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.9347222447395325,
      "learning_rate": 0.00041107276344735694,
      "loss": 2.2993,
      "step": 87386
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.9934875965118408,
      "learning_rate": 0.00041106896357016794,
      "loss": 2.1937,
      "step": 87387
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.9776736497879028,
      "learning_rate": 0.00041106516367232917,
      "loss": 2.1557,
      "step": 87388
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.9468556642532349,
      "learning_rate": 0.0004110613637538413,
      "loss": 2.1637,
      "step": 87389
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.9152645468711853,
      "learning_rate": 0.0004110575638147051,
      "loss": 1.9761,
      "step": 87390
    },
    {
      "epoch": 1.14,
      "grad_norm": 1.0090546607971191,
      "learning_rate": 0.00041105376385492114,
      "loss": 2.1741,
      "step": 87391
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.916053831577301,
      "learning_rate": 0.0004110499638744901,
      "loss": 2.3378,
      "step": 87392
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.7951684594154358,
      "learning_rate": 0.0004110461638734129,
      "loss": 2.2975,
      "step": 87393
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.9135687351226807,
      "learning_rate": 0.0004110423638516901,
      "loss": 2.1894,
      "step": 87394
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.9675078392028809,
      "learning_rate": 0.00041103856380932243,
      "loss": 2.281,
      "step": 87395
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.7558969259262085,
      "learning_rate": 0.00041103476374631056,
      "loss": 2.0916,
      "step": 87396
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.742088258266449,
      "learning_rate": 0.00041103096366265526,
      "loss": 2.2778,
      "step": 87397
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.8318442702293396,
      "learning_rate": 0.0004110271635583572,
      "loss": 2.2824,
      "step": 87398
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.8294807076454163,
      "learning_rate": 0.00041102336343341704,
      "loss": 2.1541,
      "step": 87399
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.8732804656028748,
      "learning_rate": 0.00041101956328783564,
      "loss": 2.1455,
      "step": 87400
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.8290748596191406,
      "learning_rate": 0.0004110157631216135,
      "loss": 2.2311,
      "step": 87401
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.8129642009735107,
      "learning_rate": 0.0004110119629347515,
      "loss": 2.3378,
      "step": 87402
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.888153076171875,
      "learning_rate": 0.00041100816272725033,
      "loss": 2.0278,
      "step": 87403
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.7278068661689758,
      "learning_rate": 0.00041100436249911056,
      "loss": 1.9159,
      "step": 87404
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.8279461860656738,
      "learning_rate": 0.00041100056225033295,
      "loss": 2.0646,
      "step": 87405
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.8727999925613403,
      "learning_rate": 0.0004109967619809183,
      "loss": 2.311,
      "step": 87406
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.8765288591384888,
      "learning_rate": 0.0004109929616908673,
      "loss": 2.3875,
      "step": 87407
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.8664584755897522,
      "learning_rate": 0.00041098916138018055,
      "loss": 2.3386,
      "step": 87408
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.8131489157676697,
      "learning_rate": 0.0004109853610488589,
      "loss": 2.2514,
      "step": 87409
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.9055478572845459,
      "learning_rate": 0.0004109815606969029,
      "loss": 2.3783,
      "step": 87410
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.8649030327796936,
      "learning_rate": 0.00041097776032431336,
      "loss": 2.1804,
      "step": 87411
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.8070931434631348,
      "learning_rate": 0.00041097395993109097,
      "loss": 2.284,
      "step": 87412
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.8580889105796814,
      "learning_rate": 0.0004109701595172364,
      "loss": 2.419,
      "step": 87413
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.8073985576629639,
      "learning_rate": 0.00041096635908275037,
      "loss": 2.2474,
      "step": 87414
    },
    {
      "epoch": 1.14,
      "grad_norm": 1.096289873123169,
      "learning_rate": 0.00041096255862763364,
      "loss": 2.1653,
      "step": 87415
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.8832220435142517,
      "learning_rate": 0.00041095875815188687,
      "loss": 2.2238,
      "step": 87416
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.7948758602142334,
      "learning_rate": 0.0004109549576555107,
      "loss": 1.9851,
      "step": 87417
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.9529632329940796,
      "learning_rate": 0.00041095115713850604,
      "loss": 2.1303,
      "step": 87418
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.7698020935058594,
      "learning_rate": 0.00041094735660087334,
      "loss": 2.125,
      "step": 87419
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.8176988363265991,
      "learning_rate": 0.00041094355604261347,
      "loss": 2.2728,
      "step": 87420
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.8180516362190247,
      "learning_rate": 0.0004109397554637272,
      "loss": 2.327,
      "step": 87421
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.8402564525604248,
      "learning_rate": 0.000410935954864215,
      "loss": 2.2523,
      "step": 87422
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.8074913620948792,
      "learning_rate": 0.0004109321542440777,
      "loss": 2.3021,
      "step": 87423
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.8171876072883606,
      "learning_rate": 0.00041092835360331614,
      "loss": 2.1238,
      "step": 87424
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.8570005893707275,
      "learning_rate": 0.0004109245529419308,
      "loss": 2.129,
      "step": 87425
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.8189641237258911,
      "learning_rate": 0.00041092075225992254,
      "loss": 2.3235,
      "step": 87426
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.9109671115875244,
      "learning_rate": 0.00041091695155729205,
      "loss": 2.0954,
      "step": 87427
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.7943556904792786,
      "learning_rate": 0.00041091315083404,
      "loss": 2.2925,
      "step": 87428
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.8502237200737,
      "learning_rate": 0.00041090935009016697,
      "loss": 2.2622,
      "step": 87429
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.8646165132522583,
      "learning_rate": 0.00041090554932567396,
      "loss": 2.2895,
      "step": 87430
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.8731975555419922,
      "learning_rate": 0.00041090174854056143,
      "loss": 2.0792,
      "step": 87431
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.8435047268867493,
      "learning_rate": 0.0004108979477348302,
      "loss": 2.0737,
      "step": 87432
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.840927004814148,
      "learning_rate": 0.00041089414690848096,
      "loss": 2.1861,
      "step": 87433
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.8059142231941223,
      "learning_rate": 0.00041089034606151437,
      "loss": 2.2061,
      "step": 87434
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.8656850457191467,
      "learning_rate": 0.0004108865451939312,
      "loss": 2.0604,
      "step": 87435
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.8048956990242004,
      "learning_rate": 0.0004108827443057321,
      "loss": 2.2215,
      "step": 87436
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.9592406749725342,
      "learning_rate": 0.00041087894339691787,
      "loss": 2.2504,
      "step": 87437
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.9740154147148132,
      "learning_rate": 0.00041087514246748903,
      "loss": 2.2486,
      "step": 87438
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.7647045850753784,
      "learning_rate": 0.00041087134151744653,
      "loss": 2.2205,
      "step": 87439
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.9186918139457703,
      "learning_rate": 0.0004108675405467909,
      "loss": 2.2625,
      "step": 87440
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.9921591877937317,
      "learning_rate": 0.0004108637395555229,
      "loss": 2.1819,
      "step": 87441
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.8086058497428894,
      "learning_rate": 0.00041085993854364327,
      "loss": 2.2715,
      "step": 87442
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.9469583630561829,
      "learning_rate": 0.00041085613751115266,
      "loss": 1.934,
      "step": 87443
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.8067176938056946,
      "learning_rate": 0.00041085233645805176,
      "loss": 2.069,
      "step": 87444
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.8122676610946655,
      "learning_rate": 0.0004108485353843414,
      "loss": 2.2738,
      "step": 87445
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.8050715923309326,
      "learning_rate": 0.0004108447342900222,
      "loss": 2.4421,
      "step": 87446
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.7543467283248901,
      "learning_rate": 0.0004108409331750948,
      "loss": 2.3593,
      "step": 87447
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.8114911913871765,
      "learning_rate": 0.00041083713203956,
      "loss": 2.0998,
      "step": 87448
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.849407434463501,
      "learning_rate": 0.0004108333308834185,
      "loss": 2.1641,
      "step": 87449
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.8046072125434875,
      "learning_rate": 0.000410829529706671,
      "loss": 2.2393,
      "step": 87450
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.8465038537979126,
      "learning_rate": 0.00041082572850931816,
      "loss": 2.1698,
      "step": 87451
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.8445777893066406,
      "learning_rate": 0.0004108219272913608,
      "loss": 2.0708,
      "step": 87452
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.7864205837249756,
      "learning_rate": 0.0004108181260527995,
      "loss": 2.0093,
      "step": 87453
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.9205520749092102,
      "learning_rate": 0.00041081432479363504,
      "loss": 1.9899,
      "step": 87454
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.8766093850135803,
      "learning_rate": 0.00041081052351386806,
      "loss": 1.9405,
      "step": 87455
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.9507501721382141,
      "learning_rate": 0.00041080672221349934,
      "loss": 2.1399,
      "step": 87456
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.7885175943374634,
      "learning_rate": 0.0004108029208925296,
      "loss": 2.3265,
      "step": 87457
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.878953754901886,
      "learning_rate": 0.00041079911955095946,
      "loss": 2.0759,
      "step": 87458
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.8376503586769104,
      "learning_rate": 0.0004107953181887897,
      "loss": 2.2178,
      "step": 87459
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.7503078579902649,
      "learning_rate": 0.0004107915168060209,
      "loss": 2.0183,
      "step": 87460
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.940594494342804,
      "learning_rate": 0.00041078771540265397,
      "loss": 2.3229,
      "step": 87461
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.7950137257575989,
      "learning_rate": 0.0004107839139786895,
      "loss": 2.4169,
      "step": 87462
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.802736222743988,
      "learning_rate": 0.0004107801125341282,
      "loss": 2.1166,
      "step": 87463
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.8364951014518738,
      "learning_rate": 0.00041077631106897083,
      "loss": 2.1736,
      "step": 87464
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.7325671911239624,
      "learning_rate": 0.00041077250958321797,
      "loss": 2.4396,
      "step": 87465
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.7973462343215942,
      "learning_rate": 0.0004107687080768704,
      "loss": 2.2061,
      "step": 87466
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.8122342228889465,
      "learning_rate": 0.0004107649065499289,
      "loss": 2.0734,
      "step": 87467
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.8133584260940552,
      "learning_rate": 0.00041076110500239414,
      "loss": 2.0713,
      "step": 87468
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.7887259125709534,
      "learning_rate": 0.00041075730343426674,
      "loss": 2.0891,
      "step": 87469
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.7369795441627502,
      "learning_rate": 0.0004107535018455474,
      "loss": 2.0953,
      "step": 87470
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.8794636726379395,
      "learning_rate": 0.00041074970023623707,
      "loss": 2.2478,
      "step": 87471
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.7535368800163269,
      "learning_rate": 0.0004107458986063361,
      "loss": 2.1696,
      "step": 87472
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.7921860814094543,
      "learning_rate": 0.00041074209695584544,
      "loss": 2.1589,
      "step": 87473
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.8214133977890015,
      "learning_rate": 0.00041073829528476583,
      "loss": 2.1599,
      "step": 87474
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.7704656720161438,
      "learning_rate": 0.00041073449359309777,
      "loss": 2.1025,
      "step": 87475
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.8273688554763794,
      "learning_rate": 0.0004107306918808421,
      "loss": 2.2594,
      "step": 87476
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.8719887137413025,
      "learning_rate": 0.0004107268901479995,
      "loss": 2.1725,
      "step": 87477
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.7447077631950378,
      "learning_rate": 0.00041072308839457073,
      "loss": 1.98,
      "step": 87478
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.7789727449417114,
      "learning_rate": 0.00041071928662055637,
      "loss": 2.164,
      "step": 87479
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.9482824802398682,
      "learning_rate": 0.00041071548482595724,
      "loss": 2.2386,
      "step": 87480
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.8857833743095398,
      "learning_rate": 0.00041071168301077404,
      "loss": 2.1153,
      "step": 87481
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.8206575512886047,
      "learning_rate": 0.00041070788117500743,
      "loss": 2.1997,
      "step": 87482
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.8001135587692261,
      "learning_rate": 0.0004107040793186581,
      "loss": 2.1838,
      "step": 87483
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.9032823443412781,
      "learning_rate": 0.00041070027744172683,
      "loss": 2.1653,
      "step": 87484
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.8356987237930298,
      "learning_rate": 0.00041069647554421436,
      "loss": 1.9801,
      "step": 87485
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.8562047481536865,
      "learning_rate": 0.0004106926736261212,
      "loss": 2.3325,
      "step": 87486
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.8613792061805725,
      "learning_rate": 0.0004106888716874482,
      "loss": 2.0477,
      "step": 87487
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.748478353023529,
      "learning_rate": 0.00041068506972819613,
      "loss": 2.2117,
      "step": 87488
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.8927954435348511,
      "learning_rate": 0.00041068126774836556,
      "loss": 2.3115,
      "step": 87489
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.8479140400886536,
      "learning_rate": 0.00041067746574795726,
      "loss": 2.0554,
      "step": 87490
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.836093544960022,
      "learning_rate": 0.00041067366372697195,
      "loss": 2.0886,
      "step": 87491
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.8234051465988159,
      "learning_rate": 0.0004106698616854103,
      "loss": 2.0634,
      "step": 87492
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.738339900970459,
      "learning_rate": 0.00041066605962327297,
      "loss": 2.0541,
      "step": 87493
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.8312050104141235,
      "learning_rate": 0.0004106622575405609,
      "loss": 2.1069,
      "step": 87494
    },
    {
      "epoch": 1.14,
      "grad_norm": 1.0013235807418823,
      "learning_rate": 0.0004106584554372744,
      "loss": 2.2959,
      "step": 87495
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.8150639533996582,
      "learning_rate": 0.0004106546533134146,
      "loss": 2.2274,
      "step": 87496
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.84461909532547,
      "learning_rate": 0.00041065085116898196,
      "loss": 2.0828,
      "step": 87497
    },
    {
      "epoch": 1.14,
      "grad_norm": 1.0426260232925415,
      "learning_rate": 0.00041064704900397713,
      "loss": 2.0342,
      "step": 87498
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.8011721968650818,
      "learning_rate": 0.00041064324681840104,
      "loss": 2.2746,
      "step": 87499
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.7933465838432312,
      "learning_rate": 0.00041063944461225426,
      "loss": 2.2263,
      "step": 87500
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.7824614644050598,
      "learning_rate": 0.0004106356423855375,
      "loss": 2.1663,
      "step": 87501
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.7797243595123291,
      "learning_rate": 0.00041063184013825145,
      "loss": 1.9052,
      "step": 87502
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.8019309639930725,
      "learning_rate": 0.000410628037870397,
      "loss": 2.3167,
      "step": 87503
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.7870339751243591,
      "learning_rate": 0.0004106242355819745,
      "loss": 2.1305,
      "step": 87504
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.7528378367424011,
      "learning_rate": 0.000410620433272985,
      "loss": 1.9251,
      "step": 87505
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.8177765607833862,
      "learning_rate": 0.0004106166309434291,
      "loss": 2.2607,
      "step": 87506
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.8307044506072998,
      "learning_rate": 0.00041061282859330744,
      "loss": 2.4909,
      "step": 87507
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.7779132723808289,
      "learning_rate": 0.0004106090262226207,
      "loss": 2.2711,
      "step": 87508
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.9652759432792664,
      "learning_rate": 0.00041060522383136976,
      "loss": 2.3055,
      "step": 87509
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.9101171493530273,
      "learning_rate": 0.00041060142141955514,
      "loss": 2.6043,
      "step": 87510
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.9302247762680054,
      "learning_rate": 0.00041059761898717766,
      "loss": 2.0695,
      "step": 87511
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.8271371126174927,
      "learning_rate": 0.00041059381653423796,
      "loss": 2.3943,
      "step": 87512
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.7812416553497314,
      "learning_rate": 0.0004105900140607368,
      "loss": 2.3126,
      "step": 87513
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.8261110782623291,
      "learning_rate": 0.00041058621156667493,
      "loss": 2.3453,
      "step": 87514
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.9078066945075989,
      "learning_rate": 0.0004105824090520529,
      "loss": 2.3196,
      "step": 87515
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.9357293844223022,
      "learning_rate": 0.0004105786065168716,
      "loss": 2.063,
      "step": 87516
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.8291210532188416,
      "learning_rate": 0.0004105748039611316,
      "loss": 2.1286,
      "step": 87517
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.8723433613777161,
      "learning_rate": 0.00041057100138483364,
      "loss": 2.1165,
      "step": 87518
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.8644816279411316,
      "learning_rate": 0.0004105671987879785,
      "loss": 2.0405,
      "step": 87519
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.7881436944007874,
      "learning_rate": 0.0004105633961705668,
      "loss": 2.2092,
      "step": 87520
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.8532318472862244,
      "learning_rate": 0.0004105595935325992,
      "loss": 1.942,
      "step": 87521
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.7403656244277954,
      "learning_rate": 0.0004105557908740766,
      "loss": 2.2886,
      "step": 87522
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.8709083199501038,
      "learning_rate": 0.0004105519881949996,
      "loss": 2.2889,
      "step": 87523
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.8546808362007141,
      "learning_rate": 0.00041054818549536885,
      "loss": 2.2115,
      "step": 87524
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.7692757844924927,
      "learning_rate": 0.00041054438277518513,
      "loss": 2.0756,
      "step": 87525
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.789215624332428,
      "learning_rate": 0.00041054058003444906,
      "loss": 2.216,
      "step": 87526
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.8691260814666748,
      "learning_rate": 0.00041053677727316144,
      "loss": 2.2674,
      "step": 87527
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.8261895179748535,
      "learning_rate": 0.00041053297449132305,
      "loss": 2.3231,
      "step": 87528
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.8244048953056335,
      "learning_rate": 0.00041052917168893436,
      "loss": 2.1282,
      "step": 87529
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.8929557204246521,
      "learning_rate": 0.00041052536886599625,
      "loss": 2.195,
      "step": 87530
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.8049483299255371,
      "learning_rate": 0.00041052156602250936,
      "loss": 2.2257,
      "step": 87531
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.8771987557411194,
      "learning_rate": 0.00041051776315847446,
      "loss": 2.3137,
      "step": 87532
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.829208493232727,
      "learning_rate": 0.0004105139602738922,
      "loss": 2.1166,
      "step": 87533
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.8016890287399292,
      "learning_rate": 0.0004105101573687634,
      "loss": 2.2043,
      "step": 87534
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.8696343302726746,
      "learning_rate": 0.0004105063544430886,
      "loss": 1.919,
      "step": 87535
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.8346652984619141,
      "learning_rate": 0.0004105025514968685,
      "loss": 2.1983,
      "step": 87536
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.7945689558982849,
      "learning_rate": 0.0004104987485301041,
      "loss": 2.2794,
      "step": 87537
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.7874419689178467,
      "learning_rate": 0.00041049494554279566,
      "loss": 2.0778,
      "step": 87538
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.7814546823501587,
      "learning_rate": 0.0004104911425349442,
      "loss": 2.1503,
      "step": 87539
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.8047814965248108,
      "learning_rate": 0.00041048733950655044,
      "loss": 1.8931,
      "step": 87540
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.8023037910461426,
      "learning_rate": 0.00041048353645761494,
      "loss": 2.2387,
      "step": 87541
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.8794761896133423,
      "learning_rate": 0.00041047973338813847,
      "loss": 2.1371,
      "step": 87542
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.8577194809913635,
      "learning_rate": 0.0004104759302981218,
      "loss": 2.4985,
      "step": 87543
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.8300806879997253,
      "learning_rate": 0.00041047212718756543,
      "loss": 2.2451,
      "step": 87544
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.8064510226249695,
      "learning_rate": 0.0004104683240564703,
      "loss": 2.1225,
      "step": 87545
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.8108242154121399,
      "learning_rate": 0.00041046452090483704,
      "loss": 2.2619,
      "step": 87546
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.7712473273277283,
      "learning_rate": 0.0004104607177326663,
      "loss": 2.1748,
      "step": 87547
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.8782209157943726,
      "learning_rate": 0.00041045691453995875,
      "loss": 2.3086,
      "step": 87548
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.7859175801277161,
      "learning_rate": 0.00041045311132671534,
      "loss": 2.4518,
      "step": 87549
    },
    {
      "epoch": 1.14,
      "grad_norm": 1.085606336593628,
      "learning_rate": 0.0004104493080929365,
      "loss": 2.3911,
      "step": 87550
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.8336872458457947,
      "learning_rate": 0.000410445504838623,
      "loss": 2.2857,
      "step": 87551
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.8070226907730103,
      "learning_rate": 0.0004104417015637757,
      "loss": 2.15,
      "step": 87552
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.7733041048049927,
      "learning_rate": 0.00041043789826839515,
      "loss": 2.1511,
      "step": 87553
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.8569861650466919,
      "learning_rate": 0.0004104340949524822,
      "loss": 2.3569,
      "step": 87554
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.945651113986969,
      "learning_rate": 0.0004104302916160374,
      "loss": 2.2231,
      "step": 87555
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.8706373572349548,
      "learning_rate": 0.00041042648825906147,
      "loss": 2.2328,
      "step": 87556
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.828176736831665,
      "learning_rate": 0.0004104226848815552,
      "loss": 2.2007,
      "step": 87557
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.7950783371925354,
      "learning_rate": 0.0004104188814835193,
      "loss": 2.1867,
      "step": 87558
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.9601364135742188,
      "learning_rate": 0.0004104150780649545,
      "loss": 2.2743,
      "step": 87559
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.8373584151268005,
      "learning_rate": 0.00041041127462586134,
      "loss": 2.5289,
      "step": 87560
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.8193674087524414,
      "learning_rate": 0.0004104074711662407,
      "loss": 2.1173,
      "step": 87561
    },
    {
      "epoch": 1.14,
      "grad_norm": 1.123533010482788,
      "learning_rate": 0.0004104036676860932,
      "loss": 2.2911,
      "step": 87562
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.7637609839439392,
      "learning_rate": 0.00041039986418541955,
      "loss": 2.0664,
      "step": 87563
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.8393346071243286,
      "learning_rate": 0.00041039606066422055,
      "loss": 2.0924,
      "step": 87564
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.834662914276123,
      "learning_rate": 0.0004103922571224968,
      "loss": 2.3332,
      "step": 87565
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.905076801776886,
      "learning_rate": 0.00041038845356024907,
      "loss": 2.04,
      "step": 87566
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.9695160984992981,
      "learning_rate": 0.000410384649977478,
      "loss": 2.1482,
      "step": 87567
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.8019319176673889,
      "learning_rate": 0.0004103808463741844,
      "loss": 2.2591,
      "step": 87568
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.8808360695838928,
      "learning_rate": 0.0004103770427503688,
      "loss": 2.2185,
      "step": 87569
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.8310320973396301,
      "learning_rate": 0.0004103732391060322,
      "loss": 2.1082,
      "step": 87570
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.8953994512557983,
      "learning_rate": 0.000410369435441175,
      "loss": 2.2436,
      "step": 87571
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.9392039775848389,
      "learning_rate": 0.00041036563175579806,
      "loss": 2.3649,
      "step": 87572
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.932086706161499,
      "learning_rate": 0.0004103618280499021,
      "loss": 2.3209,
      "step": 87573
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.8182754516601562,
      "learning_rate": 0.0004103580243234878,
      "loss": 2.2546,
      "step": 87574
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.8378714323043823,
      "learning_rate": 0.00041035422057655586,
      "loss": 2.0212,
      "step": 87575
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.8026041984558105,
      "learning_rate": 0.00041035041680910695,
      "loss": 2.1292,
      "step": 87576
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.7553238868713379,
      "learning_rate": 0.00041034661302114184,
      "loss": 1.9782,
      "step": 87577
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.7600242495536804,
      "learning_rate": 0.00041034280921266114,
      "loss": 2.3858,
      "step": 87578
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.7944957613945007,
      "learning_rate": 0.0004103390053836657,
      "loss": 2.024,
      "step": 87579
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.8474504351615906,
      "learning_rate": 0.0004103352015341562,
      "loss": 2.2208,
      "step": 87580
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.8226222395896912,
      "learning_rate": 0.00041033139766413323,
      "loss": 2.344,
      "step": 87581
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.8155467510223389,
      "learning_rate": 0.00041032759377359755,
      "loss": 2.1669,
      "step": 87582
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.8716332316398621,
      "learning_rate": 0.00041032378986254997,
      "loss": 2.1263,
      "step": 87583
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.8452746272087097,
      "learning_rate": 0.00041031998593099106,
      "loss": 2.0324,
      "step": 87584
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.8213579058647156,
      "learning_rate": 0.0004103161819789216,
      "loss": 2.3846,
      "step": 87585
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.8156880140304565,
      "learning_rate": 0.0004103123780063423,
      "loss": 2.259,
      "step": 87586
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.8592559695243835,
      "learning_rate": 0.0004103085740132538,
      "loss": 2.0767,
      "step": 87587
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.8952709436416626,
      "learning_rate": 0.00041030476999965687,
      "loss": 1.9903,
      "step": 87588
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.8372002840042114,
      "learning_rate": 0.00041030096596555223,
      "loss": 2.2304,
      "step": 87589
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.7829644680023193,
      "learning_rate": 0.0004102971619109405,
      "loss": 2.114,
      "step": 87590
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.8297665119171143,
      "learning_rate": 0.0004102933578358225,
      "loss": 2.1131,
      "step": 87591
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.8110776543617249,
      "learning_rate": 0.0004102895537401989,
      "loss": 2.2856,
      "step": 87592
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.8482217788696289,
      "learning_rate": 0.0004102857496240703,
      "loss": 2.1717,
      "step": 87593
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.9018623232841492,
      "learning_rate": 0.00041028194548743755,
      "loss": 2.3923,
      "step": 87594
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.8130152821540833,
      "learning_rate": 0.00041027814133030133,
      "loss": 2.5567,
      "step": 87595
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.8112825751304626,
      "learning_rate": 0.0004102743371526623,
      "loss": 1.8923,
      "step": 87596
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.9074743986129761,
      "learning_rate": 0.0004102705329545212,
      "loss": 2.0499,
      "step": 87597
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.7801645994186401,
      "learning_rate": 0.00041026672873587867,
      "loss": 1.994,
      "step": 87598
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.861689567565918,
      "learning_rate": 0.0004102629244967355,
      "loss": 2.1515,
      "step": 87599
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.8116603493690491,
      "learning_rate": 0.0004102591202370924,
      "loss": 1.9734,
      "step": 87600
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.9294790029525757,
      "learning_rate": 0.0004102553159569501,
      "loss": 2.0291,
      "step": 87601
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.8222384452819824,
      "learning_rate": 0.00041025151165630916,
      "loss": 2.2292,
      "step": 87602
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.8657398819923401,
      "learning_rate": 0.00041024770733517045,
      "loss": 2.2262,
      "step": 87603
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.9258096814155579,
      "learning_rate": 0.00041024390299353457,
      "loss": 2.1845,
      "step": 87604
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.8166555166244507,
      "learning_rate": 0.0004102400986314023,
      "loss": 2.2987,
      "step": 87605
    },
    {
      "epoch": 1.14,
      "grad_norm": 1.0478019714355469,
      "learning_rate": 0.0004102362942487742,
      "loss": 2.2001,
      "step": 87606
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.8918800354003906,
      "learning_rate": 0.0004102324898456512,
      "loss": 2.1275,
      "step": 87607
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.8220179080963135,
      "learning_rate": 0.00041022868542203386,
      "loss": 2.1537,
      "step": 87608
    },
    {
      "epoch": 1.14,
      "grad_norm": 1.0983002185821533,
      "learning_rate": 0.000410224880977923,
      "loss": 2.1093,
      "step": 87609
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.8234692811965942,
      "learning_rate": 0.0004102210765133192,
      "loss": 1.9523,
      "step": 87610
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.8087553381919861,
      "learning_rate": 0.00041021727202822324,
      "loss": 2.1776,
      "step": 87611
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.7937591671943665,
      "learning_rate": 0.0004102134675226357,
      "loss": 1.9152,
      "step": 87612
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.8236989974975586,
      "learning_rate": 0.00041020966299655753,
      "loss": 1.9197,
      "step": 87613
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.8585507869720459,
      "learning_rate": 0.00041020585844998923,
      "loss": 2.1873,
      "step": 87614
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.8277738690376282,
      "learning_rate": 0.0004102020538829316,
      "loss": 2.2739,
      "step": 87615
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.9021012187004089,
      "learning_rate": 0.0004101982492953854,
      "loss": 2.2555,
      "step": 87616
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.7781687378883362,
      "learning_rate": 0.0004101944446873512,
      "loss": 2.1557,
      "step": 87617
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.8657015562057495,
      "learning_rate": 0.00041019064005882967,
      "loss": 2.0633,
      "step": 87618
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.9107025861740112,
      "learning_rate": 0.0004101868354098218,
      "loss": 2.307,
      "step": 87619
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.7972930669784546,
      "learning_rate": 0.00041018303074032803,
      "loss": 2.0227,
      "step": 87620
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.895783007144928,
      "learning_rate": 0.00041017922605034917,
      "loss": 2.1087,
      "step": 87621
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.8335660099983215,
      "learning_rate": 0.000410175421339886,
      "loss": 2.2653,
      "step": 87622
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.8142699003219604,
      "learning_rate": 0.000410171616608939,
      "loss": 2.3247,
      "step": 87623
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.8136391639709473,
      "learning_rate": 0.000410167811857509,
      "loss": 1.9651,
      "step": 87624
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.8116374015808105,
      "learning_rate": 0.00041016400708559684,
      "loss": 1.9909,
      "step": 87625
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.9443821310997009,
      "learning_rate": 0.0004101602022932031,
      "loss": 2.249,
      "step": 87626
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.819982647895813,
      "learning_rate": 0.0004101563974803284,
      "loss": 2.2637,
      "step": 87627
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.8036364316940308,
      "learning_rate": 0.00041015259264697365,
      "loss": 2.114,
      "step": 87628
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.7771296501159668,
      "learning_rate": 0.0004101487877931394,
      "loss": 2.1014,
      "step": 87629
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.8133364319801331,
      "learning_rate": 0.0004101449829188264,
      "loss": 2.1341,
      "step": 87630
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.8550141453742981,
      "learning_rate": 0.0004101411780240354,
      "loss": 2.2207,
      "step": 87631
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.9151557087898254,
      "learning_rate": 0.000410137373108767,
      "loss": 2.4662,
      "step": 87632
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.9445383548736572,
      "learning_rate": 0.00041013356817302206,
      "loss": 2.0191,
      "step": 87633
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.7478774785995483,
      "learning_rate": 0.00041012976321680124,
      "loss": 2.0068,
      "step": 87634
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.848761796951294,
      "learning_rate": 0.00041012595824010517,
      "loss": 2.5306,
      "step": 87635
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.7570796608924866,
      "learning_rate": 0.0004101221532429346,
      "loss": 2.0884,
      "step": 87636
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.7718878984451294,
      "learning_rate": 0.0004101183482252903,
      "loss": 2.0589,
      "step": 87637
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.8264490365982056,
      "learning_rate": 0.0004101145431871728,
      "loss": 2.1617,
      "step": 87638
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.8431583642959595,
      "learning_rate": 0.00041011073812858304,
      "loss": 2.2482,
      "step": 87639
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.803438663482666,
      "learning_rate": 0.00041010693304952163,
      "loss": 2.2259,
      "step": 87640
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.8662858605384827,
      "learning_rate": 0.0004101031279499892,
      "loss": 2.1263,
      "step": 87641
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.8438752293586731,
      "learning_rate": 0.0004100993228299865,
      "loss": 2.2099,
      "step": 87642
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.8187806010246277,
      "learning_rate": 0.00041009551768951427,
      "loss": 2.1205,
      "step": 87643
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.9796940684318542,
      "learning_rate": 0.00041009171252857327,
      "loss": 2.2091,
      "step": 87644
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.838481068611145,
      "learning_rate": 0.0004100879073471641,
      "loss": 2.2054,
      "step": 87645
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.935797929763794,
      "learning_rate": 0.0004100841021452874,
      "loss": 2.0396,
      "step": 87646
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.8247941732406616,
      "learning_rate": 0.0004100802969229442,
      "loss": 2.1176,
      "step": 87647
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.8001497387886047,
      "learning_rate": 0.00041007649168013486,
      "loss": 2.0839,
      "step": 87648
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.8235942125320435,
      "learning_rate": 0.0004100726864168602,
      "loss": 2.1586,
      "step": 87649
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.7902474403381348,
      "learning_rate": 0.00041006888113312103,
      "loss": 2.1095,
      "step": 87650
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.8293656706809998,
      "learning_rate": 0.00041006507582891794,
      "loss": 2.2241,
      "step": 87651
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.8107290267944336,
      "learning_rate": 0.0004100612705042517,
      "loss": 2.029,
      "step": 87652
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.8441546559333801,
      "learning_rate": 0.000410057465159123,
      "loss": 2.3273,
      "step": 87653
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.79625004529953,
      "learning_rate": 0.00041005365979353254,
      "loss": 2.1997,
      "step": 87654
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.8132594227790833,
      "learning_rate": 0.0004100498544074809,
      "loss": 2.254,
      "step": 87655
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.912532389163971,
      "learning_rate": 0.00041004604900096904,
      "loss": 2.3077,
      "step": 87656
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.7518802285194397,
      "learning_rate": 0.0004100422435739975,
      "loss": 2.1079,
      "step": 87657
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.7901032567024231,
      "learning_rate": 0.00041003843812656714,
      "loss": 2.169,
      "step": 87658
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.8288295269012451,
      "learning_rate": 0.00041003463265867847,
      "loss": 2.1123,
      "step": 87659
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.9016769528388977,
      "learning_rate": 0.0004100308271703322,
      "loss": 2.0885,
      "step": 87660
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.7899488806724548,
      "learning_rate": 0.00041002702166152916,
      "loss": 2.471,
      "step": 87661
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.7678868174552917,
      "learning_rate": 0.00041002321613227015,
      "loss": 2.2501,
      "step": 87662
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.9241816401481628,
      "learning_rate": 0.00041001941058255566,
      "loss": 2.1617,
      "step": 87663
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.9897826313972473,
      "learning_rate": 0.0004100156050123864,
      "loss": 2.3432,
      "step": 87664
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.8628855347633362,
      "learning_rate": 0.00041001179942176327,
      "loss": 2.0885,
      "step": 87665
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.8797434568405151,
      "learning_rate": 0.0004100079938106869,
      "loss": 2.2851,
      "step": 87666
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.9065996408462524,
      "learning_rate": 0.00041000418817915786,
      "loss": 2.1027,
      "step": 87667
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.8140385150909424,
      "learning_rate": 0.0004100003825271771,
      "loss": 2.3582,
      "step": 87668
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.8203794956207275,
      "learning_rate": 0.00040999657685474505,
      "loss": 2.3173,
      "step": 87669
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.9235868453979492,
      "learning_rate": 0.00040999277116186264,
      "loss": 2.1867,
      "step": 87670
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.9105530381202698,
      "learning_rate": 0.0004099889654485305,
      "loss": 2.1004,
      "step": 87671
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.8210387229919434,
      "learning_rate": 0.0004099851597147493,
      "loss": 2.0607,
      "step": 87672
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.7516788840293884,
      "learning_rate": 0.0004099813539605198,
      "loss": 2.0798,
      "step": 87673
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.8698597550392151,
      "learning_rate": 0.00040997754818584265,
      "loss": 2.1971,
      "step": 87674
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.8206911087036133,
      "learning_rate": 0.00040997374239071867,
      "loss": 2.2776,
      "step": 87675
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.8251291513442993,
      "learning_rate": 0.00040996993657514843,
      "loss": 2.2561,
      "step": 87676
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.8313484191894531,
      "learning_rate": 0.0004099661307391328,
      "loss": 2.1747,
      "step": 87677
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.8954570293426514,
      "learning_rate": 0.0004099623248826723,
      "loss": 2.1103,
      "step": 87678
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.9561581015586853,
      "learning_rate": 0.0004099585190057677,
      "loss": 2.177,
      "step": 87679
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.8046634197235107,
      "learning_rate": 0.00040995471310841983,
      "loss": 2.0111,
      "step": 87680
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.9935199618339539,
      "learning_rate": 0.0004099509071906293,
      "loss": 2.5467,
      "step": 87681
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.9034082889556885,
      "learning_rate": 0.00040994710125239674,
      "loss": 2.3465,
      "step": 87682
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.7505411505699158,
      "learning_rate": 0.00040994329529372303,
      "loss": 2.2549,
      "step": 87683
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.9470014572143555,
      "learning_rate": 0.0004099394893146087,
      "loss": 2.2527,
      "step": 87684
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.9610489010810852,
      "learning_rate": 0.00040993568331505455,
      "loss": 2.207,
      "step": 87685
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.877689003944397,
      "learning_rate": 0.0004099318772950614,
      "loss": 2.0029,
      "step": 87686
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.8498887419700623,
      "learning_rate": 0.0004099280712546297,
      "loss": 2.0697,
      "step": 87687
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.9684351086616516,
      "learning_rate": 0.0004099242651937603,
      "loss": 2.4531,
      "step": 87688
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.935340404510498,
      "learning_rate": 0.00040992045911245403,
      "loss": 2.3388,
      "step": 87689
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.8422289490699768,
      "learning_rate": 0.00040991665301071134,
      "loss": 2.1111,
      "step": 87690
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.88614422082901,
      "learning_rate": 0.0004099128468885331,
      "loss": 2.2755,
      "step": 87691
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.8875117301940918,
      "learning_rate": 0.00040990904074592006,
      "loss": 1.9803,
      "step": 87692
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.8611968159675598,
      "learning_rate": 0.00040990523458287283,
      "loss": 2.2721,
      "step": 87693
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.7745649814605713,
      "learning_rate": 0.0004099014283993921,
      "loss": 2.0453,
      "step": 87694
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.7983061671257019,
      "learning_rate": 0.0004098976221954786,
      "loss": 2.0789,
      "step": 87695
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.8309337496757507,
      "learning_rate": 0.00040989381597113315,
      "loss": 2.4074,
      "step": 87696
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.9535475969314575,
      "learning_rate": 0.0004098900097263562,
      "loss": 2.4281,
      "step": 87697
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.8058619499206543,
      "learning_rate": 0.00040988620346114885,
      "loss": 2.0947,
      "step": 87698
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.9405885934829712,
      "learning_rate": 0.0004098823971755114,
      "loss": 2.1371,
      "step": 87699
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.9588320255279541,
      "learning_rate": 0.0004098785908694448,
      "loss": 2.2317,
      "step": 87700
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.8469789624214172,
      "learning_rate": 0.00040987478454294977,
      "loss": 2.1141,
      "step": 87701
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.858009934425354,
      "learning_rate": 0.00040987097819602684,
      "loss": 2.2876,
      "step": 87702
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.9111087918281555,
      "learning_rate": 0.0004098671718286768,
      "loss": 2.1601,
      "step": 87703
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.9542606472969055,
      "learning_rate": 0.00040986336544090047,
      "loss": 2.3832,
      "step": 87704
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.8003179430961609,
      "learning_rate": 0.00040985955903269844,
      "loss": 2.4585,
      "step": 87705
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.9098437428474426,
      "learning_rate": 0.00040985575260407144,
      "loss": 2.0523,
      "step": 87706
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.8305588364601135,
      "learning_rate": 0.0004098519461550202,
      "loss": 2.199,
      "step": 87707
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.8656644821166992,
      "learning_rate": 0.0004098481396855453,
      "loss": 2.1748,
      "step": 87708
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.8292815089225769,
      "learning_rate": 0.00040984433319564764,
      "loss": 2.0935,
      "step": 87709
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.8171236515045166,
      "learning_rate": 0.00040984052668532786,
      "loss": 2.0404,
      "step": 87710
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.7798290252685547,
      "learning_rate": 0.00040983672015458664,
      "loss": 2.2119,
      "step": 87711
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.8689716458320618,
      "learning_rate": 0.00040983291360342474,
      "loss": 2.1614,
      "step": 87712
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.8783893585205078,
      "learning_rate": 0.0004098291070318427,
      "loss": 2.1827,
      "step": 87713
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.832273542881012,
      "learning_rate": 0.0004098253004398415,
      "loss": 2.2151,
      "step": 87714
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.9147338271141052,
      "learning_rate": 0.0004098214938274216,
      "loss": 2.1806,
      "step": 87715
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.8089055418968201,
      "learning_rate": 0.0004098176871945839,
      "loss": 2.18,
      "step": 87716
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.9206518530845642,
      "learning_rate": 0.000409813880541329,
      "loss": 2.1611,
      "step": 87717
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.905168354511261,
      "learning_rate": 0.0004098100738676576,
      "loss": 2.1999,
      "step": 87718
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.8358138799667358,
      "learning_rate": 0.00040980626717357035,
      "loss": 2.2503,
      "step": 87719
    },
    {
      "epoch": 1.14,
      "grad_norm": 1.003233551979065,
      "learning_rate": 0.0004098024604590682,
      "loss": 1.9985,
      "step": 87720
    },
    {
      "epoch": 1.14,
      "grad_norm": 1.0354595184326172,
      "learning_rate": 0.0004097986537241516,
      "loss": 2.1908,
      "step": 87721
    },
    {
      "epoch": 1.14,
      "grad_norm": 1.1467665433883667,
      "learning_rate": 0.00040979484696882136,
      "loss": 2.261,
      "step": 87722
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.8778085112571716,
      "learning_rate": 0.00040979104019307826,
      "loss": 2.18,
      "step": 87723
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.8841837048530579,
      "learning_rate": 0.0004097872333969229,
      "loss": 2.0344,
      "step": 87724
    },
    {
      "epoch": 1.14,
      "grad_norm": 1.188733458518982,
      "learning_rate": 0.000409783426580356,
      "loss": 2.2361,
      "step": 87725
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.8598631620407104,
      "learning_rate": 0.00040977961974337834,
      "loss": 2.4488,
      "step": 87726
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.7746044397354126,
      "learning_rate": 0.0004097758128859905,
      "loss": 2.1767,
      "step": 87727
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.9953206181526184,
      "learning_rate": 0.0004097720060081932,
      "loss": 2.2436,
      "step": 87728
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.8905491828918457,
      "learning_rate": 0.0004097681991099874,
      "loss": 2.1895,
      "step": 87729
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.8575575351715088,
      "learning_rate": 0.0004097643921913735,
      "loss": 2.172,
      "step": 87730
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.8621081709861755,
      "learning_rate": 0.00040976058525235234,
      "loss": 2.2537,
      "step": 87731
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.7985430955886841,
      "learning_rate": 0.0004097567782929247,
      "loss": 2.212,
      "step": 87732
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.8151954412460327,
      "learning_rate": 0.0004097529713130911,
      "loss": 2.2817,
      "step": 87733
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.7878478765487671,
      "learning_rate": 0.00040974916431285233,
      "loss": 2.202,
      "step": 87734
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.7758140563964844,
      "learning_rate": 0.00040974535729220926,
      "loss": 2.1582,
      "step": 87735
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.8494619131088257,
      "learning_rate": 0.0004097415502511623,
      "loss": 2.3034,
      "step": 87736
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.7626681923866272,
      "learning_rate": 0.00040973774318971243,
      "loss": 1.9949,
      "step": 87737
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.8799182772636414,
      "learning_rate": 0.00040973393610786013,
      "loss": 2.2695,
      "step": 87738
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.8790793418884277,
      "learning_rate": 0.00040973012900560635,
      "loss": 2.1515,
      "step": 87739
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.7772368788719177,
      "learning_rate": 0.0004097263218829515,
      "loss": 2.138,
      "step": 87740
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.9429930448532104,
      "learning_rate": 0.0004097225147398966,
      "loss": 2.3377,
      "step": 87741
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.861405611038208,
      "learning_rate": 0.00040971870757644215,
      "loss": 2.2638,
      "step": 87742
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.8413007259368896,
      "learning_rate": 0.000409714900392589,
      "loss": 2.0144,
      "step": 87743
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.7370583415031433,
      "learning_rate": 0.0004097110931883377,
      "loss": 2.267,
      "step": 87744
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.9447761178016663,
      "learning_rate": 0.000409707285963689,
      "loss": 2.3144,
      "step": 87745
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.8790377378463745,
      "learning_rate": 0.0004097034787186438,
      "loss": 2.2047,
      "step": 87746
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.8926201462745667,
      "learning_rate": 0.0004096996714532025,
      "loss": 2.1729,
      "step": 87747
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.81472247838974,
      "learning_rate": 0.00040969586416736604,
      "loss": 2.0264,
      "step": 87748
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.8257747888565063,
      "learning_rate": 0.00040969205686113496,
      "loss": 2.1578,
      "step": 87749
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.852549135684967,
      "learning_rate": 0.0004096882495345102,
      "loss": 2.1323,
      "step": 87750
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.7649672627449036,
      "learning_rate": 0.0004096844421874922,
      "loss": 2.1432,
      "step": 87751
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.8211110830307007,
      "learning_rate": 0.00040968063482008186,
      "loss": 2.1212,
      "step": 87752
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.8693577647209167,
      "learning_rate": 0.0004096768274322798,
      "loss": 2.0873,
      "step": 87753
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.8441978096961975,
      "learning_rate": 0.00040967302002408676,
      "loss": 2.2519,
      "step": 87754
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.8246339559555054,
      "learning_rate": 0.0004096692125955034,
      "loss": 2.1479,
      "step": 87755
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.8303712606430054,
      "learning_rate": 0.0004096654051465305,
      "loss": 2.2387,
      "step": 87756
    },
    {
      "epoch": 1.14,
      "grad_norm": 1.2016217708587646,
      "learning_rate": 0.0004096615976771687,
      "loss": 2.4533,
      "step": 87757
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.8319644927978516,
      "learning_rate": 0.0004096577901874187,
      "loss": 2.0866,
      "step": 87758
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.8063223958015442,
      "learning_rate": 0.0004096539826772814,
      "loss": 2.3527,
      "step": 87759
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.8163546919822693,
      "learning_rate": 0.0004096501751467572,
      "loss": 2.2118,
      "step": 87760
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.8957545757293701,
      "learning_rate": 0.000409646367595847,
      "loss": 1.9509,
      "step": 87761
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.918172299861908,
      "learning_rate": 0.0004096425600245515,
      "loss": 2.2357,
      "step": 87762
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.8022223711013794,
      "learning_rate": 0.00040963875243287136,
      "loss": 2.0934,
      "step": 87763
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.9162041544914246,
      "learning_rate": 0.0004096349448208073,
      "loss": 2.1877,
      "step": 87764
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.8409852981567383,
      "learning_rate": 0.0004096311371883601,
      "loss": 1.9002,
      "step": 87765
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.8365981578826904,
      "learning_rate": 0.00040962732953553035,
      "loss": 2.2572,
      "step": 87766
    },
    {
      "epoch": 1.14,
      "grad_norm": 1.0357285737991333,
      "learning_rate": 0.00040962352186231873,
      "loss": 2.19,
      "step": 87767
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.7799445986747742,
      "learning_rate": 0.0004096197141687262,
      "loss": 2.1396,
      "step": 87768
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.8662598133087158,
      "learning_rate": 0.0004096159064547531,
      "loss": 2.0119,
      "step": 87769
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.8502131104469299,
      "learning_rate": 0.00040961209872040046,
      "loss": 2.2506,
      "step": 87770
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.8044908046722412,
      "learning_rate": 0.0004096082909656689,
      "loss": 2.1059,
      "step": 87771
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.8600719571113586,
      "learning_rate": 0.000409604483190559,
      "loss": 2.1472,
      "step": 87772
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.905430793762207,
      "learning_rate": 0.00040960067539507156,
      "loss": 2.3787,
      "step": 87773
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.7579087018966675,
      "learning_rate": 0.00040959686757920734,
      "loss": 2.0813,
      "step": 87774
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.8685979247093201,
      "learning_rate": 0.00040959305974296696,
      "loss": 1.9001,
      "step": 87775
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.84766685962677,
      "learning_rate": 0.0004095892518863511,
      "loss": 2.0943,
      "step": 87776
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.844062864780426,
      "learning_rate": 0.00040958544400936065,
      "loss": 2.1266,
      "step": 87777
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.8543375134468079,
      "learning_rate": 0.00040958163611199615,
      "loss": 2.1663,
      "step": 87778
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.8321424722671509,
      "learning_rate": 0.0004095778281942583,
      "loss": 2.1088,
      "step": 87779
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.8116825222969055,
      "learning_rate": 0.00040957402025614795,
      "loss": 2.1725,
      "step": 87780
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.8604845404624939,
      "learning_rate": 0.00040957021229766566,
      "loss": 2.3009,
      "step": 87781
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.8637414574623108,
      "learning_rate": 0.00040956640431881225,
      "loss": 2.4984,
      "step": 87782
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.812105655670166,
      "learning_rate": 0.00040956259631958825,
      "loss": 2.1833,
      "step": 87783
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.8535630702972412,
      "learning_rate": 0.0004095587882999947,
      "loss": 2.2221,
      "step": 87784
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.8955866694450378,
      "learning_rate": 0.0004095549802600319,
      "loss": 2.1333,
      "step": 87785
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.9604430198669434,
      "learning_rate": 0.00040955117219970087,
      "loss": 2.0385,
      "step": 87786
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.8375470042228699,
      "learning_rate": 0.00040954736411900227,
      "loss": 2.1392,
      "step": 87787
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.8747425675392151,
      "learning_rate": 0.00040954355601793667,
      "loss": 2.1743,
      "step": 87788
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.8563700914382935,
      "learning_rate": 0.0004095397478965048,
      "loss": 2.1103,
      "step": 87789
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.8064656853675842,
      "learning_rate": 0.0004095359397547075,
      "loss": 2.217,
      "step": 87790
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.8138558864593506,
      "learning_rate": 0.00040953213159254545,
      "loss": 2.1934,
      "step": 87791
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.9226033687591553,
      "learning_rate": 0.0004095283234100192,
      "loss": 2.3339,
      "step": 87792
    },
    {
      "epoch": 1.14,
      "grad_norm": 1.0130215883255005,
      "learning_rate": 0.0004095245152071296,
      "loss": 2.0282,
      "step": 87793
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.823077380657196,
      "learning_rate": 0.00040952070698387733,
      "loss": 2.3955,
      "step": 87794
    },
    {
      "epoch": 1.14,
      "grad_norm": 1.020682692527771,
      "learning_rate": 0.00040951689874026305,
      "loss": 2.1677,
      "step": 87795
    },
    {
      "epoch": 1.14,
      "grad_norm": 1.0912010669708252,
      "learning_rate": 0.0004095130904762876,
      "loss": 2.512,
      "step": 87796
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.7939164042472839,
      "learning_rate": 0.0004095092821919516,
      "loss": 2.3154,
      "step": 87797
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.8256858587265015,
      "learning_rate": 0.0004095054738872557,
      "loss": 2.1109,
      "step": 87798
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.9627788662910461,
      "learning_rate": 0.0004095016655622007,
      "loss": 2.4572,
      "step": 87799
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.8147106766700745,
      "learning_rate": 0.00040949785721678723,
      "loss": 2.1185,
      "step": 87800
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.8350608944892883,
      "learning_rate": 0.000409494048851016,
      "loss": 2.3481,
      "step": 87801
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.8466948866844177,
      "learning_rate": 0.00040949024046488795,
      "loss": 2.1679,
      "step": 87802
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.810749351978302,
      "learning_rate": 0.0004094864320584034,
      "loss": 2.1368,
      "step": 87803
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.9357613921165466,
      "learning_rate": 0.00040948262363156335,
      "loss": 2.2658,
      "step": 87804
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.803071141242981,
      "learning_rate": 0.0004094788151843684,
      "loss": 1.8864,
      "step": 87805
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.8705233931541443,
      "learning_rate": 0.00040947500671681923,
      "loss": 2.1205,
      "step": 87806
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.8058461546897888,
      "learning_rate": 0.00040947119822891664,
      "loss": 1.9753,
      "step": 87807
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.8477047085762024,
      "learning_rate": 0.0004094673897206613,
      "loss": 2.2929,
      "step": 87808
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.7982784509658813,
      "learning_rate": 0.0004094635811920539,
      "loss": 2.0196,
      "step": 87809
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.8576846122741699,
      "learning_rate": 0.00040945977264309505,
      "loss": 2.3505,
      "step": 87810
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.8348298072814941,
      "learning_rate": 0.00040945596407378573,
      "loss": 2.1556,
      "step": 87811
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.8298026323318481,
      "learning_rate": 0.00040945215548412637,
      "loss": 2.3448,
      "step": 87812
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.7903141379356384,
      "learning_rate": 0.0004094483468741178,
      "loss": 2.0647,
      "step": 87813
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.7762866020202637,
      "learning_rate": 0.0004094445382437608,
      "loss": 2.0826,
      "step": 87814
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.8171238303184509,
      "learning_rate": 0.00040944072959305585,
      "loss": 2.2623,
      "step": 87815
    },
    {
      "epoch": 1.14,
      "grad_norm": 1.1070423126220703,
      "learning_rate": 0.00040943692092200386,
      "loss": 2.1796,
      "step": 87816
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.8696098327636719,
      "learning_rate": 0.0004094331122306055,
      "loss": 2.2304,
      "step": 87817
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.873725414276123,
      "learning_rate": 0.00040942930351886147,
      "loss": 2.0586,
      "step": 87818
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.8298978805541992,
      "learning_rate": 0.0004094254947867724,
      "loss": 2.1741,
      "step": 87819
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.8283092379570007,
      "learning_rate": 0.0004094216860343392,
      "loss": 2.2973,
      "step": 87820
    },
    {
      "epoch": 1.14,
      "grad_norm": 1.006030797958374,
      "learning_rate": 0.00040941787726156236,
      "loss": 2.2705,
      "step": 87821
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.8425012826919556,
      "learning_rate": 0.00040941406846844265,
      "loss": 2.2466,
      "step": 87822
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.8934367895126343,
      "learning_rate": 0.00040941025965498076,
      "loss": 2.1447,
      "step": 87823
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.8628419637680054,
      "learning_rate": 0.00040940645082117755,
      "loss": 2.2435,
      "step": 87824
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.8614656925201416,
      "learning_rate": 0.0004094026419670336,
      "loss": 2.2902,
      "step": 87825
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.8074628710746765,
      "learning_rate": 0.0004093988330925495,
      "loss": 2.0907,
      "step": 87826
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.8199080228805542,
      "learning_rate": 0.00040939502419772624,
      "loss": 2.0999,
      "step": 87827
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.8294341564178467,
      "learning_rate": 0.0004093912152825643,
      "loss": 2.1406,
      "step": 87828
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.9156913757324219,
      "learning_rate": 0.0004093874063470645,
      "loss": 2.0223,
      "step": 87829
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.966311514377594,
      "learning_rate": 0.0004093835973912275,
      "loss": 2.262,
      "step": 87830
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.7780961990356445,
      "learning_rate": 0.0004093797884150541,
      "loss": 1.9547,
      "step": 87831
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.8879673480987549,
      "learning_rate": 0.0004093759794185448,
      "loss": 2.0486,
      "step": 87832
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.9360967874526978,
      "learning_rate": 0.00040937217040170053,
      "loss": 2.272,
      "step": 87833
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.8741067051887512,
      "learning_rate": 0.00040936836136452187,
      "loss": 2.1542,
      "step": 87834
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.8362383842468262,
      "learning_rate": 0.0004093645523070096,
      "loss": 2.2313,
      "step": 87835
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.8924311399459839,
      "learning_rate": 0.0004093607432291644,
      "loss": 2.1659,
      "step": 87836
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.898324191570282,
      "learning_rate": 0.0004093569341309869,
      "loss": 2.2206,
      "step": 87837
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.9294669032096863,
      "learning_rate": 0.00040935312501247796,
      "loss": 2.1655,
      "step": 87838
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.8386925458908081,
      "learning_rate": 0.0004093493158736382,
      "loss": 2.0089,
      "step": 87839
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.8282775282859802,
      "learning_rate": 0.0004093455067144683,
      "loss": 2.1709,
      "step": 87840
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.871246874332428,
      "learning_rate": 0.0004093416975349691,
      "loss": 2.149,
      "step": 87841
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.9107624292373657,
      "learning_rate": 0.0004093378883351411,
      "loss": 2.2767,
      "step": 87842
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.8133758306503296,
      "learning_rate": 0.00040933407911498515,
      "loss": 1.9388,
      "step": 87843
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.8790730834007263,
      "learning_rate": 0.000409330269874502,
      "loss": 2.2128,
      "step": 87844
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.8135104179382324,
      "learning_rate": 0.0004093264606136922,
      "loss": 2.1282,
      "step": 87845
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.833933413028717,
      "learning_rate": 0.0004093226513325566,
      "loss": 2.1531,
      "step": 87846
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.788615882396698,
      "learning_rate": 0.0004093188420310958,
      "loss": 2.0803,
      "step": 87847
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.7894423007965088,
      "learning_rate": 0.0004093150327093106,
      "loss": 2.1353,
      "step": 87848
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.8542777299880981,
      "learning_rate": 0.0004093112233672017,
      "loss": 2.2099,
      "step": 87849
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.900330662727356,
      "learning_rate": 0.0004093074140047698,
      "loss": 2.2029,
      "step": 87850
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.8433307409286499,
      "learning_rate": 0.0004093036046220155,
      "loss": 2.2123,
      "step": 87851
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.7920348644256592,
      "learning_rate": 0.0004092997952189397,
      "loss": 2.1311,
      "step": 87852
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.9135204553604126,
      "learning_rate": 0.00040929598579554286,
      "loss": 2.3719,
      "step": 87853
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.9196801781654358,
      "learning_rate": 0.000409292176351826,
      "loss": 2.069,
      "step": 87854
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.8005788326263428,
      "learning_rate": 0.0004092883668877895,
      "loss": 2.199,
      "step": 87855
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.7563678026199341,
      "learning_rate": 0.00040928455740343433,
      "loss": 2.1334,
      "step": 87856
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.8425012230873108,
      "learning_rate": 0.0004092807478987611,
      "loss": 2.1861,
      "step": 87857
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.8650756478309631,
      "learning_rate": 0.00040927693837377054,
      "loss": 2.0728,
      "step": 87858
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.7755094766616821,
      "learning_rate": 0.00040927312882846323,
      "loss": 2.2019,
      "step": 87859
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.7671764492988586,
      "learning_rate": 0.0004092693192628401,
      "loss": 1.9206,
      "step": 87860
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.8741239905357361,
      "learning_rate": 0.0004092655096769016,
      "loss": 2.027,
      "step": 87861
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.7556447982788086,
      "learning_rate": 0.0004092617000706487,
      "loss": 2.1999,
      "step": 87862
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.8776613473892212,
      "learning_rate": 0.00040925789044408197,
      "loss": 2.2375,
      "step": 87863
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.7416713833808899,
      "learning_rate": 0.00040925408079720216,
      "loss": 2.0758,
      "step": 87864
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.8437063694000244,
      "learning_rate": 0.00040925027113000983,
      "loss": 2.4288,
      "step": 87865
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.9761872291564941,
      "learning_rate": 0.00040924646144250594,
      "loss": 2.2714,
      "step": 87866
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.8364773988723755,
      "learning_rate": 0.000409242651734691,
      "loss": 2.2942,
      "step": 87867
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.8718950152397156,
      "learning_rate": 0.0004092388420065658,
      "loss": 2.0743,
      "step": 87868
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.7831866145133972,
      "learning_rate": 0.0004092350322581311,
      "loss": 1.7688,
      "step": 87869
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.8609657883644104,
      "learning_rate": 0.00040923122248938745,
      "loss": 2.0806,
      "step": 87870
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.9582258462905884,
      "learning_rate": 0.0004092274127003357,
      "loss": 2.3135,
      "step": 87871
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.8225250244140625,
      "learning_rate": 0.00040922360289097657,
      "loss": 2.2421,
      "step": 87872
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.8016998171806335,
      "learning_rate": 0.00040921979306131063,
      "loss": 1.9384,
      "step": 87873
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.9202911257743835,
      "learning_rate": 0.0004092159832113386,
      "loss": 2.0221,
      "step": 87874
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.8501137495040894,
      "learning_rate": 0.00040921217334106145,
      "loss": 2.3259,
      "step": 87875
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.7600457072257996,
      "learning_rate": 0.00040920836345047955,
      "loss": 2.2574,
      "step": 87876
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.900839626789093,
      "learning_rate": 0.0004092045535395938,
      "loss": 2.1309,
      "step": 87877
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.9506144523620605,
      "learning_rate": 0.0004092007436084049,
      "loss": 2.0784,
      "step": 87878
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.8960310220718384,
      "learning_rate": 0.0004091969336569134,
      "loss": 2.2719,
      "step": 87879
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.9578847885131836,
      "learning_rate": 0.0004091931236851202,
      "loss": 2.2347,
      "step": 87880
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.9363940954208374,
      "learning_rate": 0.000409189313693026,
      "loss": 2.1337,
      "step": 87881
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.8336776494979858,
      "learning_rate": 0.0004091855036806314,
      "loss": 2.4101,
      "step": 87882
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.9973371624946594,
      "learning_rate": 0.0004091816936479371,
      "loss": 2.1289,
      "step": 87883
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.7684697508811951,
      "learning_rate": 0.00040917788359494395,
      "loss": 2.1859,
      "step": 87884
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.8403503894805908,
      "learning_rate": 0.0004091740735216525,
      "loss": 2.1422,
      "step": 87885
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.7656055688858032,
      "learning_rate": 0.0004091702634280636,
      "loss": 2.0024,
      "step": 87886
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.8839820623397827,
      "learning_rate": 0.00040916645331417784,
      "loss": 2.2655,
      "step": 87887
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.9419180750846863,
      "learning_rate": 0.00040916264317999594,
      "loss": 2.0531,
      "step": 87888
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.8334599137306213,
      "learning_rate": 0.0004091588330255187,
      "loss": 2.3792,
      "step": 87889
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.7461581230163574,
      "learning_rate": 0.0004091550228507468,
      "loss": 2.2543,
      "step": 87890
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.7589796781539917,
      "learning_rate": 0.0004091512126556808,
      "loss": 1.9293,
      "step": 87891
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.74904865026474,
      "learning_rate": 0.0004091474024403216,
      "loss": 2.2434,
      "step": 87892
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.8556697964668274,
      "learning_rate": 0.0004091435922046699,
      "loss": 2.1135,
      "step": 87893
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.809733510017395,
      "learning_rate": 0.00040913978194872624,
      "loss": 2.3956,
      "step": 87894
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.7882127165794373,
      "learning_rate": 0.0004091359716724915,
      "loss": 1.9553,
      "step": 87895
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.8629598617553711,
      "learning_rate": 0.00040913216137596634,
      "loss": 1.9825,
      "step": 87896
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.834615170955658,
      "learning_rate": 0.0004091283510591514,
      "loss": 2.4462,
      "step": 87897
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.8295962810516357,
      "learning_rate": 0.00040912454072204745,
      "loss": 2.3311,
      "step": 87898
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.8608757853507996,
      "learning_rate": 0.0004091207303646552,
      "loss": 2.1257,
      "step": 87899
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.7763912677764893,
      "learning_rate": 0.00040911691998697533,
      "loss": 2.3398,
      "step": 87900
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.8037413358688354,
      "learning_rate": 0.00040911310958900856,
      "loss": 2.3367,
      "step": 87901
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.8013197183609009,
      "learning_rate": 0.0004091092991707556,
      "loss": 2.1775,
      "step": 87902
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.9034236669540405,
      "learning_rate": 0.0004091054887322172,
      "loss": 2.3953,
      "step": 87903
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.8028118014335632,
      "learning_rate": 0.000409101678273394,
      "loss": 2.1177,
      "step": 87904
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.851359486579895,
      "learning_rate": 0.0004090978677942868,
      "loss": 2.3821,
      "step": 87905
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.7708142995834351,
      "learning_rate": 0.0004090940572948962,
      "loss": 2.0652,
      "step": 87906
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.8116098642349243,
      "learning_rate": 0.00040909024677522295,
      "loss": 2.4854,
      "step": 87907
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.8820410966873169,
      "learning_rate": 0.00040908643623526776,
      "loss": 2.2997,
      "step": 87908
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.7899349927902222,
      "learning_rate": 0.00040908262567503133,
      "loss": 2.0481,
      "step": 87909
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.8505098819732666,
      "learning_rate": 0.00040907881509451435,
      "loss": 2.1625,
      "step": 87910
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.7846401929855347,
      "learning_rate": 0.00040907500449371764,
      "loss": 2.0552,
      "step": 87911
    },
    {
      "epoch": 1.14,
      "grad_norm": 1.046716332435608,
      "learning_rate": 0.00040907119387264185,
      "loss": 2.1332,
      "step": 87912
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.8336461782455444,
      "learning_rate": 0.00040906738323128757,
      "loss": 2.3021,
      "step": 87913
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.9256038069725037,
      "learning_rate": 0.0004090635725696556,
      "loss": 2.1691,
      "step": 87914
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.774280309677124,
      "learning_rate": 0.0004090597618877467,
      "loss": 2.2505,
      "step": 87915
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.8200413584709167,
      "learning_rate": 0.00040905595118556154,
      "loss": 2.1549,
      "step": 87916
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.8970032930374146,
      "learning_rate": 0.00040905214046310087,
      "loss": 2.3246,
      "step": 87917
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.9025350213050842,
      "learning_rate": 0.00040904832972036523,
      "loss": 2.1347,
      "step": 87918
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.8030627369880676,
      "learning_rate": 0.0004090445189573555,
      "loss": 2.393,
      "step": 87919
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.8797624707221985,
      "learning_rate": 0.0004090407081740724,
      "loss": 2.2858,
      "step": 87920
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.8035526871681213,
      "learning_rate": 0.0004090368973705165,
      "loss": 2.1141,
      "step": 87921
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.8007094860076904,
      "learning_rate": 0.0004090330865466885,
      "loss": 2.0855,
      "step": 87922
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.8855927586555481,
      "learning_rate": 0.0004090292757025894,
      "loss": 2.42,
      "step": 87923
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.9412326216697693,
      "learning_rate": 0.0004090254648382195,
      "loss": 2.327,
      "step": 87924
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.826353907585144,
      "learning_rate": 0.0004090216539535798,
      "loss": 2.137,
      "step": 87925
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.9932587742805481,
      "learning_rate": 0.00040901784304867093,
      "loss": 2.2674,
      "step": 87926
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.8707500100135803,
      "learning_rate": 0.00040901403212349356,
      "loss": 2.3137,
      "step": 87927
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.8538667559623718,
      "learning_rate": 0.00040901022117804844,
      "loss": 2.3103,
      "step": 87928
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.9168582558631897,
      "learning_rate": 0.0004090064102123362,
      "loss": 2.2184,
      "step": 87929
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.8276174664497375,
      "learning_rate": 0.00040900259922635777,
      "loss": 2.3392,
      "step": 87930
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.865835428237915,
      "learning_rate": 0.0004089987882201135,
      "loss": 2.0393,
      "step": 87931
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.8536744117736816,
      "learning_rate": 0.0004089949771936044,
      "loss": 2.3106,
      "step": 87932
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.8993533849716187,
      "learning_rate": 0.00040899116614683114,
      "loss": 2.2622,
      "step": 87933
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.8850628733634949,
      "learning_rate": 0.0004089873550797942,
      "loss": 2.1904,
      "step": 87934
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.858881950378418,
      "learning_rate": 0.00040898354399249463,
      "loss": 2.2999,
      "step": 87935
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.8170873522758484,
      "learning_rate": 0.0004089797328849329,
      "loss": 2.1251,
      "step": 87936
    },
    {
      "epoch": 1.14,
      "grad_norm": 0.9759476184844971,
      "learning_rate": 0.0004089759217571098,
      "loss": 2.3135,
      "step": 87937
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.9651319980621338,
      "learning_rate": 0.0004089721106090259,
      "loss": 1.9541,
      "step": 87938
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.8028312921524048,
      "learning_rate": 0.0004089682994406821,
      "loss": 2.0548,
      "step": 87939
    },
    {
      "epoch": 1.15,
      "grad_norm": 1.672166347503662,
      "learning_rate": 0.000408964488252079,
      "loss": 2.0683,
      "step": 87940
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.8406639695167542,
      "learning_rate": 0.00040896067704321743,
      "loss": 2.1816,
      "step": 87941
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.8708063960075378,
      "learning_rate": 0.000408956865814098,
      "loss": 2.229,
      "step": 87942
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.9361748695373535,
      "learning_rate": 0.0004089530545647214,
      "loss": 1.9748,
      "step": 87943
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.7858689427375793,
      "learning_rate": 0.00040894924329508837,
      "loss": 2.0433,
      "step": 87944
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.894004762172699,
      "learning_rate": 0.0004089454320051996,
      "loss": 2.3509,
      "step": 87945
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.8148865103721619,
      "learning_rate": 0.00040894162069505586,
      "loss": 2.2278,
      "step": 87946
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.9498750567436218,
      "learning_rate": 0.00040893780936465776,
      "loss": 2.2081,
      "step": 87947
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.8460758328437805,
      "learning_rate": 0.0004089339980140061,
      "loss": 2.3863,
      "step": 87948
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.8030781745910645,
      "learning_rate": 0.0004089301866431016,
      "loss": 2.0166,
      "step": 87949
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.982679009437561,
      "learning_rate": 0.00040892637525194485,
      "loss": 2.2018,
      "step": 87950
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.7528368830680847,
      "learning_rate": 0.0004089225638405367,
      "loss": 2.1439,
      "step": 87951
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.9212538003921509,
      "learning_rate": 0.0004089187524088777,
      "loss": 2.2085,
      "step": 87952
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.8820710182189941,
      "learning_rate": 0.00040891494095696867,
      "loss": 2.1567,
      "step": 87953
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.9304940104484558,
      "learning_rate": 0.0004089111294848104,
      "loss": 2.0978,
      "step": 87954
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.8405486345291138,
      "learning_rate": 0.00040890731799240337,
      "loss": 2.3667,
      "step": 87955
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.7886113524436951,
      "learning_rate": 0.00040890350647974845,
      "loss": 2.0223,
      "step": 87956
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.8479344844818115,
      "learning_rate": 0.0004088996949468464,
      "loss": 2.2126,
      "step": 87957
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.8013792037963867,
      "learning_rate": 0.0004088958833936977,
      "loss": 2.0463,
      "step": 87958
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.7889069318771362,
      "learning_rate": 0.0004088920718203033,
      "loss": 2.1453,
      "step": 87959
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.9516282677650452,
      "learning_rate": 0.0004088882602266638,
      "loss": 2.1306,
      "step": 87960
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.8581645488739014,
      "learning_rate": 0.00040888444861277983,
      "loss": 2.1942,
      "step": 87961
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.904278039932251,
      "learning_rate": 0.00040888063697865225,
      "loss": 2.4071,
      "step": 87962
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.7647601366043091,
      "learning_rate": 0.0004088768253242817,
      "loss": 2.1284,
      "step": 87963
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.7885140776634216,
      "learning_rate": 0.00040887301364966893,
      "loss": 2.2745,
      "step": 87964
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.7836154103279114,
      "learning_rate": 0.0004088692019548145,
      "loss": 2.2229,
      "step": 87965
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.9287564754486084,
      "learning_rate": 0.00040886539023971935,
      "loss": 2.1167,
      "step": 87966
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.8644470572471619,
      "learning_rate": 0.00040886157850438405,
      "loss": 2.1401,
      "step": 87967
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.8562723994255066,
      "learning_rate": 0.00040885776674880924,
      "loss": 2.1541,
      "step": 87968
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.8188139200210571,
      "learning_rate": 0.0004088539549729958,
      "loss": 2.1321,
      "step": 87969
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.8998070359230042,
      "learning_rate": 0.00040885014317694436,
      "loss": 2.3345,
      "step": 87970
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.8540557026863098,
      "learning_rate": 0.0004088463313606556,
      "loss": 2.23,
      "step": 87971
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.8449459671974182,
      "learning_rate": 0.00040884251952413023,
      "loss": 2.2459,
      "step": 87972
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.7950334548950195,
      "learning_rate": 0.000408838707667369,
      "loss": 2.114,
      "step": 87973
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.7437964081764221,
      "learning_rate": 0.0004088348957903726,
      "loss": 1.9768,
      "step": 87974
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.8229016065597534,
      "learning_rate": 0.00040883108389314174,
      "loss": 2.0764,
      "step": 87975
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.8146225810050964,
      "learning_rate": 0.0004088272719756771,
      "loss": 2.0481,
      "step": 87976
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.813284695148468,
      "learning_rate": 0.00040882346003797944,
      "loss": 2.2585,
      "step": 87977
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.8848689794540405,
      "learning_rate": 0.00040881964808004947,
      "loss": 2.2138,
      "step": 87978
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.9184584021568298,
      "learning_rate": 0.0004088158361018878,
      "loss": 2.1067,
      "step": 87979
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.7771121263504028,
      "learning_rate": 0.00040881202410349523,
      "loss": 2.3531,
      "step": 87980
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.9816232323646545,
      "learning_rate": 0.0004088082120848725,
      "loss": 2.1483,
      "step": 87981
    },
    {
      "epoch": 1.15,
      "grad_norm": 1.0699881315231323,
      "learning_rate": 0.0004088044000460203,
      "loss": 2.2962,
      "step": 87982
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.8562645316123962,
      "learning_rate": 0.00040880058798693926,
      "loss": 2.1187,
      "step": 87983
    },
    {
      "epoch": 1.15,
      "grad_norm": 1.0989755392074585,
      "learning_rate": 0.0004087967759076301,
      "loss": 2.0872,
      "step": 87984
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.9842356443405151,
      "learning_rate": 0.00040879296380809366,
      "loss": 2.0906,
      "step": 87985
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.7893025279045105,
      "learning_rate": 0.00040878915168833047,
      "loss": 2.0839,
      "step": 87986
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.9608734846115112,
      "learning_rate": 0.0004087853395483413,
      "loss": 2.0916,
      "step": 87987
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.8932884335517883,
      "learning_rate": 0.000408781527388127,
      "loss": 1.9938,
      "step": 87988
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.8845028281211853,
      "learning_rate": 0.00040877771520768803,
      "loss": 2.238,
      "step": 87989
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.8324877023696899,
      "learning_rate": 0.0004087739030070253,
      "loss": 2.1237,
      "step": 87990
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.8761880397796631,
      "learning_rate": 0.00040877009078613943,
      "loss": 2.3674,
      "step": 87991
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.9110502004623413,
      "learning_rate": 0.0004087662785450312,
      "loss": 2.0942,
      "step": 87992
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.9374558925628662,
      "learning_rate": 0.00040876246628370117,
      "loss": 2.3052,
      "step": 87993
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.8672906160354614,
      "learning_rate": 0.0004087586540021503,
      "loss": 2.2678,
      "step": 87994
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.8914951682090759,
      "learning_rate": 0.00040875484170037893,
      "loss": 2.4453,
      "step": 87995
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.9222558736801147,
      "learning_rate": 0.0004087510293783881,
      "loss": 2.2322,
      "step": 87996
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.8590549230575562,
      "learning_rate": 0.00040874721703617847,
      "loss": 2.2382,
      "step": 87997
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.7591068744659424,
      "learning_rate": 0.0004087434046737505,
      "loss": 2.1416,
      "step": 87998
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.8360746502876282,
      "learning_rate": 0.0004087395922911052,
      "loss": 2.1371,
      "step": 87999
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.8902090191841125,
      "learning_rate": 0.0004087357798882432,
      "loss": 2.1605,
      "step": 88000
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.8611855506896973,
      "learning_rate": 0.0004087319674651651,
      "loss": 2.0817,
      "step": 88001
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.8457424640655518,
      "learning_rate": 0.00040872815502187166,
      "loss": 2.1059,
      "step": 88002
    },
    {
      "epoch": 1.15,
      "grad_norm": 1.019550085067749,
      "learning_rate": 0.0004087243425583635,
      "loss": 2.187,
      "step": 88003
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.7729971408843994,
      "learning_rate": 0.00040872053007464163,
      "loss": 2.2159,
      "step": 88004
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.880403995513916,
      "learning_rate": 0.00040871671757070646,
      "loss": 2.2655,
      "step": 88005
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.9679732322692871,
      "learning_rate": 0.00040871290504655884,
      "loss": 2.2523,
      "step": 88006
    },
    {
      "epoch": 1.15,
      "grad_norm": 1.000722885131836,
      "learning_rate": 0.00040870909250219947,
      "loss": 2.2051,
      "step": 88007
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.8312826156616211,
      "learning_rate": 0.00040870527993762894,
      "loss": 2.3892,
      "step": 88008
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.8227306604385376,
      "learning_rate": 0.0004087014673528481,
      "loss": 2.1213,
      "step": 88009
    },
    {
      "epoch": 1.15,
      "grad_norm": 1.2138850688934326,
      "learning_rate": 0.00040869765474785757,
      "loss": 2.2045,
      "step": 88010
    },
    {
      "epoch": 1.15,
      "grad_norm": 1.0373519659042358,
      "learning_rate": 0.00040869384212265813,
      "loss": 2.3522,
      "step": 88011
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.8078628778457642,
      "learning_rate": 0.0004086900294772504,
      "loss": 2.1329,
      "step": 88012
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.9064918160438538,
      "learning_rate": 0.0004086862168116352,
      "loss": 2.3726,
      "step": 88013
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.9911050200462341,
      "learning_rate": 0.00040868240412581316,
      "loss": 2.2134,
      "step": 88014
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.8359789252281189,
      "learning_rate": 0.00040867859141978505,
      "loss": 2.2829,
      "step": 88015
    },
    {
      "epoch": 1.15,
      "grad_norm": 1.0507349967956543,
      "learning_rate": 0.00040867477869355146,
      "loss": 2.3164,
      "step": 88016
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.8759379982948303,
      "learning_rate": 0.00040867096594711323,
      "loss": 2.1234,
      "step": 88017
    },
    {
      "epoch": 1.15,
      "grad_norm": 1.1889581680297852,
      "learning_rate": 0.000408667153180471,
      "loss": 2.4606,
      "step": 88018
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.8287368416786194,
      "learning_rate": 0.0004086633403936255,
      "loss": 2.0388,
      "step": 88019
    },
    {
      "epoch": 1.15,
      "grad_norm": 1.4123879671096802,
      "learning_rate": 0.0004086595275865774,
      "loss": 2.1287,
      "step": 88020
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.8814258575439453,
      "learning_rate": 0.0004086557147593275,
      "loss": 2.1615,
      "step": 88021
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.7636622786521912,
      "learning_rate": 0.00040865190191187646,
      "loss": 2.2555,
      "step": 88022
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.972393810749054,
      "learning_rate": 0.0004086480890442249,
      "loss": 2.3508,
      "step": 88023
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.8385923504829407,
      "learning_rate": 0.0004086442761563737,
      "loss": 2.0684,
      "step": 88024
    },
    {
      "epoch": 1.15,
      "grad_norm": 1.0796902179718018,
      "learning_rate": 0.0004086404632483234,
      "loss": 2.1258,
      "step": 88025
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.7896276116371155,
      "learning_rate": 0.00040863665032007485,
      "loss": 2.5687,
      "step": 88026
    },
    {
      "epoch": 1.15,
      "grad_norm": 1.1934674978256226,
      "learning_rate": 0.00040863283737162866,
      "loss": 2.0726,
      "step": 88027
    },
    {
      "epoch": 1.15,
      "grad_norm": 1.3799058198928833,
      "learning_rate": 0.0004086290244029855,
      "loss": 2.0451,
      "step": 88028
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.7839708924293518,
      "learning_rate": 0.0004086252114141463,
      "loss": 2.0923,
      "step": 88029
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.8101897239685059,
      "learning_rate": 0.00040862139840511154,
      "loss": 2.1876,
      "step": 88030
    },
    {
      "epoch": 1.15,
      "grad_norm": 1.0744911432266235,
      "learning_rate": 0.00040861758537588205,
      "loss": 2.1121,
      "step": 88031
    },
    {
      "epoch": 1.15,
      "grad_norm": 1.068804383277893,
      "learning_rate": 0.0004086137723264585,
      "loss": 2.0937,
      "step": 88032
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.8623402714729309,
      "learning_rate": 0.0004086099592568416,
      "loss": 2.2013,
      "step": 88033
    },
    {
      "epoch": 1.15,
      "grad_norm": 1.0053036212921143,
      "learning_rate": 0.00040860614616703203,
      "loss": 1.9419,
      "step": 88034
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.962903618812561,
      "learning_rate": 0.0004086023330570305,
      "loss": 2.1038,
      "step": 88035
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.9744258522987366,
      "learning_rate": 0.00040859851992683784,
      "loss": 2.211,
      "step": 88036
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.8746882677078247,
      "learning_rate": 0.0004085947067764546,
      "loss": 2.2256,
      "step": 88037
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.9233893752098083,
      "learning_rate": 0.0004085908936058815,
      "loss": 2.2497,
      "step": 88038
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.7705419063568115,
      "learning_rate": 0.0004085870804151194,
      "loss": 2.185,
      "step": 88039
    },
    {
      "epoch": 1.15,
      "grad_norm": 1.0444754362106323,
      "learning_rate": 0.00040858326720416885,
      "loss": 2.1683,
      "step": 88040
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.7766718864440918,
      "learning_rate": 0.00040857945397303067,
      "loss": 2.0648,
      "step": 88041
    },
    {
      "epoch": 1.15,
      "grad_norm": 1.0437990427017212,
      "learning_rate": 0.0004085756407217055,
      "loss": 2.2588,
      "step": 88042
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.7783839702606201,
      "learning_rate": 0.0004085718274501941,
      "loss": 2.1385,
      "step": 88043
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.8733788728713989,
      "learning_rate": 0.00040856801415849707,
      "loss": 2.1887,
      "step": 88044
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.8760418891906738,
      "learning_rate": 0.0004085642008466152,
      "loss": 2.3404,
      "step": 88045
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.9355078935623169,
      "learning_rate": 0.0004085603875145493,
      "loss": 2.1235,
      "step": 88046
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.95632404088974,
      "learning_rate": 0.0004085565741622999,
      "loss": 2.0503,
      "step": 88047
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.8412765264511108,
      "learning_rate": 0.0004085527607898678,
      "loss": 2.1526,
      "step": 88048
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.8635289669036865,
      "learning_rate": 0.0004085489473972537,
      "loss": 2.1357,
      "step": 88049
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.9213899374008179,
      "learning_rate": 0.0004085451339844583,
      "loss": 2.321,
      "step": 88050
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.8566174507141113,
      "learning_rate": 0.0004085413205514823,
      "loss": 2.3099,
      "step": 88051
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.882652223110199,
      "learning_rate": 0.00040853750709832643,
      "loss": 2.0367,
      "step": 88052
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.7608791589736938,
      "learning_rate": 0.0004085336936249913,
      "loss": 2.1323,
      "step": 88053
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.7860694527626038,
      "learning_rate": 0.0004085298801314778,
      "loss": 2.1062,
      "step": 88054
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.771682620048523,
      "learning_rate": 0.0004085260666177866,
      "loss": 2.2834,
      "step": 88055
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.8571553230285645,
      "learning_rate": 0.0004085222530839182,
      "loss": 2.1808,
      "step": 88056
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.9424498677253723,
      "learning_rate": 0.0004085184395298736,
      "loss": 2.4333,
      "step": 88057
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.8772445321083069,
      "learning_rate": 0.0004085146259556534,
      "loss": 2.2491,
      "step": 88058
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.943088948726654,
      "learning_rate": 0.0004085108123612582,
      "loss": 2.1388,
      "step": 88059
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.8464796543121338,
      "learning_rate": 0.00040850699874668875,
      "loss": 2.059,
      "step": 88060
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.8606958389282227,
      "learning_rate": 0.00040850318511194596,
      "loss": 1.9983,
      "step": 88061
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.8709654808044434,
      "learning_rate": 0.0004084993714570302,
      "loss": 1.9891,
      "step": 88062
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.8457356095314026,
      "learning_rate": 0.0004084955577819424,
      "loss": 2.3348,
      "step": 88063
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.7664340734481812,
      "learning_rate": 0.0004084917440866833,
      "loss": 2.07,
      "step": 88064
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.7962859869003296,
      "learning_rate": 0.0004084879303712535,
      "loss": 2.0674,
      "step": 88065
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.8239638209342957,
      "learning_rate": 0.0004084841166356537,
      "loss": 2.2664,
      "step": 88066
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.9208034873008728,
      "learning_rate": 0.0004084803028798847,
      "loss": 2.2485,
      "step": 88067
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.886427640914917,
      "learning_rate": 0.0004084764891039472,
      "loss": 2.0024,
      "step": 88068
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.8658671379089355,
      "learning_rate": 0.0004084726753078418,
      "loss": 2.3507,
      "step": 88069
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.9604989290237427,
      "learning_rate": 0.0004084688614915693,
      "loss": 2.2055,
      "step": 88070
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.8599607348442078,
      "learning_rate": 0.00040846504765513044,
      "loss": 2.0834,
      "step": 88071
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.8322585225105286,
      "learning_rate": 0.00040846123379852583,
      "loss": 1.9841,
      "step": 88072
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.7871001958847046,
      "learning_rate": 0.0004084574199217563,
      "loss": 2.2165,
      "step": 88073
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.8245612382888794,
      "learning_rate": 0.0004084536060248224,
      "loss": 2.0322,
      "step": 88074
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.8575451374053955,
      "learning_rate": 0.00040844979210772497,
      "loss": 2.0668,
      "step": 88075
    },
    {
      "epoch": 1.15,
      "grad_norm": 1.045305848121643,
      "learning_rate": 0.00040844597817046467,
      "loss": 1.973,
      "step": 88076
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.7905447483062744,
      "learning_rate": 0.0004084421642130422,
      "loss": 2.1605,
      "step": 88077
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.9606421589851379,
      "learning_rate": 0.0004084383502354582,
      "loss": 2.216,
      "step": 88078
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.7872782945632935,
      "learning_rate": 0.0004084345362377137,
      "loss": 2.4791,
      "step": 88079
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.8102344870567322,
      "learning_rate": 0.00040843072221980896,
      "loss": 2.3066,
      "step": 88080
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.8964211940765381,
      "learning_rate": 0.000408426908181745,
      "loss": 2.182,
      "step": 88081
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.9413344860076904,
      "learning_rate": 0.0004084230941235225,
      "loss": 1.9666,
      "step": 88082
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.8430569171905518,
      "learning_rate": 0.00040841928004514195,
      "loss": 2.1682,
      "step": 88083
    },
    {
      "epoch": 1.15,
      "grad_norm": 1.0164998769760132,
      "learning_rate": 0.0004084154659466042,
      "loss": 2.2138,
      "step": 88084
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.842187225818634,
      "learning_rate": 0.00040841165182791013,
      "loss": 2.1059,
      "step": 88085
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.9483607411384583,
      "learning_rate": 0.00040840783768906016,
      "loss": 2.1355,
      "step": 88086
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.8280832767486572,
      "learning_rate": 0.00040840402353005514,
      "loss": 2.4843,
      "step": 88087
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.7920989990234375,
      "learning_rate": 0.0004084002093508958,
      "loss": 2.287,
      "step": 88088
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.8322175741195679,
      "learning_rate": 0.0004083963951515828,
      "loss": 2.2716,
      "step": 88089
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.8914021849632263,
      "learning_rate": 0.00040839258093211685,
      "loss": 2.2146,
      "step": 88090
    },
    {
      "epoch": 1.15,
      "grad_norm": 1.0089178085327148,
      "learning_rate": 0.00040838876669249864,
      "loss": 2.5251,
      "step": 88091
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.750777006149292,
      "learning_rate": 0.000408384952432729,
      "loss": 2.1338,
      "step": 88092
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.8694270849227905,
      "learning_rate": 0.00040838113815280856,
      "loss": 2.4475,
      "step": 88093
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.8479688763618469,
      "learning_rate": 0.0004083773238527379,
      "loss": 2.253,
      "step": 88094
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.8693867921829224,
      "learning_rate": 0.00040837350953251786,
      "loss": 2.1716,
      "step": 88095
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.8000121116638184,
      "learning_rate": 0.00040836969519214924,
      "loss": 2.1856,
      "step": 88096
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.7674700021743774,
      "learning_rate": 0.0004083658808316326,
      "loss": 2.173,
      "step": 88097
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.9157819747924805,
      "learning_rate": 0.00040836206645096876,
      "loss": 2.1329,
      "step": 88098
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.826814591884613,
      "learning_rate": 0.00040835825205015836,
      "loss": 1.963,
      "step": 88099
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.7274463176727295,
      "learning_rate": 0.000408354437629202,
      "loss": 2.1909,
      "step": 88100
    },
    {
      "epoch": 1.15,
      "grad_norm": 1.0180258750915527,
      "learning_rate": 0.00040835062318810055,
      "loss": 2.1656,
      "step": 88101
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.8261725902557373,
      "learning_rate": 0.00040834680872685466,
      "loss": 2.2823,
      "step": 88102
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.8385568261146545,
      "learning_rate": 0.0004083429942454652,
      "loss": 2.1259,
      "step": 88103
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.8685630559921265,
      "learning_rate": 0.00040833917974393254,
      "loss": 2.1215,
      "step": 88104
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.898212730884552,
      "learning_rate": 0.0004083353652222576,
      "loss": 2.3016,
      "step": 88105
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.8160332441329956,
      "learning_rate": 0.0004083315506804412,
      "loss": 2.183,
      "step": 88106
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.8078040480613708,
      "learning_rate": 0.0004083277361184838,
      "loss": 2.0522,
      "step": 88107
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.7933624982833862,
      "learning_rate": 0.0004083239215363863,
      "loss": 2.3962,
      "step": 88108
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.9057419300079346,
      "learning_rate": 0.0004083201069341494,
      "loss": 2.1231,
      "step": 88109
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.9003252983093262,
      "learning_rate": 0.0004083162923117737,
      "loss": 2.2348,
      "step": 88110
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.9052772521972656,
      "learning_rate": 0.00040831247766925986,
      "loss": 2.247,
      "step": 88111
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.9384434223175049,
      "learning_rate": 0.0004083086630066088,
      "loss": 2.2651,
      "step": 88112
    },
    {
      "epoch": 1.15,
      "grad_norm": 1.079433560371399,
      "learning_rate": 0.00040830484832382103,
      "loss": 2.3512,
      "step": 88113
    },
    {
      "epoch": 1.15,
      "grad_norm": 1.5644469261169434,
      "learning_rate": 0.00040830103362089737,
      "loss": 2.2703,
      "step": 88114
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.871474027633667,
      "learning_rate": 0.00040829721889783845,
      "loss": 2.162,
      "step": 88115
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.9436244964599609,
      "learning_rate": 0.0004082934041546452,
      "loss": 2.1966,
      "step": 88116
    },
    {
      "epoch": 1.15,
      "grad_norm": 1.110927939414978,
      "learning_rate": 0.00040828958939131797,
      "loss": 2.1713,
      "step": 88117
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.8645585179328918,
      "learning_rate": 0.00040828577460785777,
      "loss": 2.1263,
      "step": 88118
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.8276674747467041,
      "learning_rate": 0.00040828195980426523,
      "loss": 2.0532,
      "step": 88119
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.9845330119132996,
      "learning_rate": 0.0004082781449805409,
      "loss": 2.3219,
      "step": 88120
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.8991442322731018,
      "learning_rate": 0.0004082743301366857,
      "loss": 2.2348,
      "step": 88121
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.8447883129119873,
      "learning_rate": 0.00040827051527270034,
      "loss": 2.1744,
      "step": 88122
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.922319233417511,
      "learning_rate": 0.00040826670038858536,
      "loss": 2.0396,
      "step": 88123
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.929991602897644,
      "learning_rate": 0.0004082628854843415,
      "loss": 2.3268,
      "step": 88124
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.8320079445838928,
      "learning_rate": 0.0004082590705599696,
      "loss": 2.3196,
      "step": 88125
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.8111768960952759,
      "learning_rate": 0.0004082552556154703,
      "loss": 2.1527,
      "step": 88126
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.8753109574317932,
      "learning_rate": 0.00040825144065084426,
      "loss": 1.9896,
      "step": 88127
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.8258299827575684,
      "learning_rate": 0.0004082476256660923,
      "loss": 1.9668,
      "step": 88128
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.8107515573501587,
      "learning_rate": 0.000408243810661215,
      "loss": 2.2512,
      "step": 88129
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.9477384090423584,
      "learning_rate": 0.00040823999563621314,
      "loss": 2.1327,
      "step": 88130
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.7811373472213745,
      "learning_rate": 0.00040823618059108747,
      "loss": 2.2943,
      "step": 88131
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.8021682500839233,
      "learning_rate": 0.0004082323655258386,
      "loss": 2.2224,
      "step": 88132
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.856160044670105,
      "learning_rate": 0.0004082285504404673,
      "loss": 2.1789,
      "step": 88133
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.8100085854530334,
      "learning_rate": 0.0004082247353349743,
      "loss": 2.0578,
      "step": 88134
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.8045557141304016,
      "learning_rate": 0.00040822092020936025,
      "loss": 2.1038,
      "step": 88135
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.9293263554573059,
      "learning_rate": 0.0004082171050636259,
      "loss": 2.2846,
      "step": 88136
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.9022533893585205,
      "learning_rate": 0.00040821328989777196,
      "loss": 2.291,
      "step": 88137
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.8480063080787659,
      "learning_rate": 0.0004082094747117991,
      "loss": 2.2004,
      "step": 88138
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.7701237797737122,
      "learning_rate": 0.00040820565950570805,
      "loss": 2.252,
      "step": 88139
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.7849346399307251,
      "learning_rate": 0.0004082018442794996,
      "loss": 2.262,
      "step": 88140
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.7036592960357666,
      "learning_rate": 0.00040819802903317427,
      "loss": 2.2125,
      "step": 88141
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.764610230922699,
      "learning_rate": 0.00040819421376673295,
      "loss": 2.3975,
      "step": 88142
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.9259567856788635,
      "learning_rate": 0.00040819039848017627,
      "loss": 2.1586,
      "step": 88143
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.800173282623291,
      "learning_rate": 0.000408186583173505,
      "loss": 2.0929,
      "step": 88144
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.8240222930908203,
      "learning_rate": 0.00040818276784671974,
      "loss": 2.1401,
      "step": 88145
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.8322685360908508,
      "learning_rate": 0.0004081789524998213,
      "loss": 2.1003,
      "step": 88146
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.7334988117218018,
      "learning_rate": 0.0004081751371328103,
      "loss": 1.9506,
      "step": 88147
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.862069308757782,
      "learning_rate": 0.0004081713217456875,
      "loss": 2.1602,
      "step": 88148
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.8448018431663513,
      "learning_rate": 0.00040816750633845367,
      "loss": 2.1529,
      "step": 88149
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.8360031843185425,
      "learning_rate": 0.0004081636909111094,
      "loss": 2.3077,
      "step": 88150
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.7991552948951721,
      "learning_rate": 0.00040815987546365546,
      "loss": 2.1027,
      "step": 88151
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.7611677050590515,
      "learning_rate": 0.00040815605999609263,
      "loss": 2.4615,
      "step": 88152
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.82376629114151,
      "learning_rate": 0.0004081522445084215,
      "loss": 2.003,
      "step": 88153
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.9160885810852051,
      "learning_rate": 0.00040814842900064284,
      "loss": 2.3859,
      "step": 88154
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.8780210018157959,
      "learning_rate": 0.00040814461347275734,
      "loss": 2.143,
      "step": 88155
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.885053277015686,
      "learning_rate": 0.0004081407979247657,
      "loss": 1.9999,
      "step": 88156
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.9211807250976562,
      "learning_rate": 0.00040813698235666864,
      "loss": 2.2813,
      "step": 88157
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.7540118098258972,
      "learning_rate": 0.0004081331667684669,
      "loss": 2.1317,
      "step": 88158
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.8372454047203064,
      "learning_rate": 0.00040812935116016114,
      "loss": 2.0669,
      "step": 88159
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.7851012349128723,
      "learning_rate": 0.0004081255355317521,
      "loss": 2.2496,
      "step": 88160
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.9183127880096436,
      "learning_rate": 0.00040812171988324053,
      "loss": 2.0686,
      "step": 88161
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.8483960628509521,
      "learning_rate": 0.000408117904214627,
      "loss": 2.158,
      "step": 88162
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.9059157371520996,
      "learning_rate": 0.00040811408852591235,
      "loss": 2.2632,
      "step": 88163
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.9660734534263611,
      "learning_rate": 0.0004081102728170973,
      "loss": 2.0705,
      "step": 88164
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.8594668507575989,
      "learning_rate": 0.00040810645708818245,
      "loss": 2.3834,
      "step": 88165
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.816787600517273,
      "learning_rate": 0.0004081026413391685,
      "loss": 2.0997,
      "step": 88166
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.8105681538581848,
      "learning_rate": 0.00040809882557005635,
      "loss": 2.3989,
      "step": 88167
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.8142972588539124,
      "learning_rate": 0.00040809500978084657,
      "loss": 2.2046,
      "step": 88168
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.8670711517333984,
      "learning_rate": 0.0004080911939715399,
      "loss": 2.2221,
      "step": 88169
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.8878937363624573,
      "learning_rate": 0.000408087378142137,
      "loss": 2.0878,
      "step": 88170
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.7492647171020508,
      "learning_rate": 0.0004080835622926386,
      "loss": 2.1937,
      "step": 88171
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.8440175652503967,
      "learning_rate": 0.00040807974642304544,
      "loss": 2.2071,
      "step": 88172
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.8056822419166565,
      "learning_rate": 0.00040807593053335825,
      "loss": 2.1191,
      "step": 88173
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.881807267665863,
      "learning_rate": 0.0004080721146235776,
      "loss": 2.2473,
      "step": 88174
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.8099920153617859,
      "learning_rate": 0.0004080682986937044,
      "loss": 2.053,
      "step": 88175
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.821508526802063,
      "learning_rate": 0.0004080644827437392,
      "loss": 2.3327,
      "step": 88176
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.8182429075241089,
      "learning_rate": 0.00040806066677368284,
      "loss": 2.144,
      "step": 88177
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.850021481513977,
      "learning_rate": 0.0004080568507835359,
      "loss": 2.1405,
      "step": 88178
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.7300475835800171,
      "learning_rate": 0.00040805303477329917,
      "loss": 2.0797,
      "step": 88179
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.707942545413971,
      "learning_rate": 0.00040804921874297345,
      "loss": 1.9881,
      "step": 88180
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.9496030807495117,
      "learning_rate": 0.00040804540269255926,
      "loss": 2.1657,
      "step": 88181
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.8497350215911865,
      "learning_rate": 0.00040804158662205735,
      "loss": 2.0989,
      "step": 88182
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.8404912948608398,
      "learning_rate": 0.0004080377705314685,
      "loss": 2.1358,
      "step": 88183
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.8928824067115784,
      "learning_rate": 0.0004080339544207934,
      "loss": 2.1184,
      "step": 88184
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.8941680788993835,
      "learning_rate": 0.0004080301382900327,
      "loss": 2.2724,
      "step": 88185
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.7456635236740112,
      "learning_rate": 0.00040802632213918715,
      "loss": 2.1649,
      "step": 88186
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.8184972405433655,
      "learning_rate": 0.0004080225059682575,
      "loss": 2.1088,
      "step": 88187
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.9258174300193787,
      "learning_rate": 0.0004080186897772445,
      "loss": 2.0319,
      "step": 88188
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.9339517951011658,
      "learning_rate": 0.0004080148735661487,
      "loss": 2.2669,
      "step": 88189
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.8338701128959656,
      "learning_rate": 0.0004080110573349709,
      "loss": 2.0604,
      "step": 88190
    },
    {
      "epoch": 1.15,
      "grad_norm": 1.0013457536697388,
      "learning_rate": 0.0004080072410837119,
      "loss": 2.267,
      "step": 88191
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.7885928750038147,
      "learning_rate": 0.0004080034248123722,
      "loss": 2.1744,
      "step": 88192
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.9207184314727783,
      "learning_rate": 0.0004079996085209527,
      "loss": 2.0026,
      "step": 88193
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.9320035576820374,
      "learning_rate": 0.00040799579220945393,
      "loss": 2.3027,
      "step": 88194
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.8920660018920898,
      "learning_rate": 0.00040799197587787684,
      "loss": 2.0992,
      "step": 88195
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.8023458123207092,
      "learning_rate": 0.00040798815952622196,
      "loss": 2.1987,
      "step": 88196
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.947694718837738,
      "learning_rate": 0.00040798434315449,
      "loss": 2.0697,
      "step": 88197
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.8279853463172913,
      "learning_rate": 0.0004079805267626817,
      "loss": 2.0886,
      "step": 88198
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.7966462969779968,
      "learning_rate": 0.0004079767103507978,
      "loss": 1.9941,
      "step": 88199
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.8763499855995178,
      "learning_rate": 0.000407972893918839,
      "loss": 2.132,
      "step": 88200
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.8301218748092651,
      "learning_rate": 0.000407969077466806,
      "loss": 2.0568,
      "step": 88201
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.7881414294242859,
      "learning_rate": 0.00040796526099469953,
      "loss": 2.2188,
      "step": 88202
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.7590851187705994,
      "learning_rate": 0.0004079614445025203,
      "loss": 2.1142,
      "step": 88203
    },
    {
      "epoch": 1.15,
      "grad_norm": 1.027970314025879,
      "learning_rate": 0.00040795762799026897,
      "loss": 2.1075,
      "step": 88204
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.8515289425849915,
      "learning_rate": 0.00040795381145794626,
      "loss": 2.187,
      "step": 88205
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.7975127100944519,
      "learning_rate": 0.00040794999490555286,
      "loss": 2.3196,
      "step": 88206
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.8301810622215271,
      "learning_rate": 0.0004079461783330897,
      "loss": 2.0245,
      "step": 88207
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.7836794853210449,
      "learning_rate": 0.0004079423617405571,
      "loss": 2.1979,
      "step": 88208
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.8341448307037354,
      "learning_rate": 0.000407938545127956,
      "loss": 2.0596,
      "step": 88209
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.8414402604103088,
      "learning_rate": 0.00040793472849528725,
      "loss": 2.2809,
      "step": 88210
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.7251289486885071,
      "learning_rate": 0.00040793091184255126,
      "loss": 1.9805,
      "step": 88211
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.787815272808075,
      "learning_rate": 0.00040792709516974886,
      "loss": 2.0885,
      "step": 88212
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.8489202260971069,
      "learning_rate": 0.0004079232784768809,
      "loss": 2.3648,
      "step": 88213
    },
    {
      "epoch": 1.15,
      "grad_norm": 1.0111794471740723,
      "learning_rate": 0.00040791946176394785,
      "loss": 2.0536,
      "step": 88214
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.808255136013031,
      "learning_rate": 0.00040791564503095054,
      "loss": 2.0835,
      "step": 88215
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.7775101065635681,
      "learning_rate": 0.0004079118282778898,
      "loss": 2.165,
      "step": 88216
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.7933999300003052,
      "learning_rate": 0.00040790801150476606,
      "loss": 2.2048,
      "step": 88217
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.881736159324646,
      "learning_rate": 0.0004079041947115803,
      "loss": 1.8696,
      "step": 88218
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.834370493888855,
      "learning_rate": 0.0004079003778983331,
      "loss": 2.2223,
      "step": 88219
    },
    {
      "epoch": 1.15,
      "grad_norm": 1.1982346773147583,
      "learning_rate": 0.0004078965610650251,
      "loss": 2.057,
      "step": 88220
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.8822436928749084,
      "learning_rate": 0.0004078927442116571,
      "loss": 2.0307,
      "step": 88221
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.81782466173172,
      "learning_rate": 0.0004078889273382299,
      "loss": 2.1288,
      "step": 88222
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.8404439687728882,
      "learning_rate": 0.000407885110444744,
      "loss": 2.2651,
      "step": 88223
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.9600961804389954,
      "learning_rate": 0.0004078812935312003,
      "loss": 2.2724,
      "step": 88224
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.8781812787055969,
      "learning_rate": 0.00040787747659759945,
      "loss": 1.9695,
      "step": 88225
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.7721512913703918,
      "learning_rate": 0.000407873659643942,
      "loss": 2.1128,
      "step": 88226
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.8412940502166748,
      "learning_rate": 0.00040786984267022895,
      "loss": 2.2827,
      "step": 88227
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.8273325562477112,
      "learning_rate": 0.0004078660256764608,
      "loss": 2.139,
      "step": 88228
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.7984740138053894,
      "learning_rate": 0.0004078622086626383,
      "loss": 2.0045,
      "step": 88229
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.8243209719657898,
      "learning_rate": 0.0004078583916287622,
      "loss": 2.2362,
      "step": 88230
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.8622843027114868,
      "learning_rate": 0.0004078545745748332,
      "loss": 2.376,
      "step": 88231
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.7398465871810913,
      "learning_rate": 0.000407850757500852,
      "loss": 2.0802,
      "step": 88232
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.8875320553779602,
      "learning_rate": 0.00040784694040681927,
      "loss": 2.1417,
      "step": 88233
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.9071840047836304,
      "learning_rate": 0.00040784312329273585,
      "loss": 2.3435,
      "step": 88234
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.8637126088142395,
      "learning_rate": 0.0004078393061586023,
      "loss": 2.4115,
      "step": 88235
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.841788649559021,
      "learning_rate": 0.0004078354890044193,
      "loss": 2.2421,
      "step": 88236
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.8039408922195435,
      "learning_rate": 0.0004078316718301878,
      "loss": 2.0856,
      "step": 88237
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.8494836688041687,
      "learning_rate": 0.00040782785463590825,
      "loss": 2.4704,
      "step": 88238
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.8519460558891296,
      "learning_rate": 0.00040782403742158145,
      "loss": 2.0864,
      "step": 88239
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.8498396873474121,
      "learning_rate": 0.0004078202201872083,
      "loss": 1.8955,
      "step": 88240
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.9503887891769409,
      "learning_rate": 0.0004078164029327892,
      "loss": 2.1505,
      "step": 88241
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.8475704789161682,
      "learning_rate": 0.0004078125856583249,
      "loss": 2.3362,
      "step": 88242
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.8633576035499573,
      "learning_rate": 0.00040780876836381643,
      "loss": 2.0799,
      "step": 88243
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.7597506642341614,
      "learning_rate": 0.00040780495104926414,
      "loss": 2.1367,
      "step": 88244
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.9388669729232788,
      "learning_rate": 0.00040780113371466885,
      "loss": 2.1078,
      "step": 88245
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.8365997076034546,
      "learning_rate": 0.0004077973163600313,
      "loss": 1.9997,
      "step": 88246
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.8463255763053894,
      "learning_rate": 0.00040779349898535227,
      "loss": 2.1339,
      "step": 88247
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.83575040102005,
      "learning_rate": 0.00040778968159063233,
      "loss": 2.3004,
      "step": 88248
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.8218216300010681,
      "learning_rate": 0.00040778586417587233,
      "loss": 2.1878,
      "step": 88249
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.8338868618011475,
      "learning_rate": 0.00040778204674107285,
      "loss": 2.2052,
      "step": 88250
    },
    {
      "epoch": 1.15,
      "grad_norm": 1.0073788166046143,
      "learning_rate": 0.0004077782292862346,
      "loss": 2.1475,
      "step": 88251
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.8595920205116272,
      "learning_rate": 0.0004077744118113584,
      "loss": 2.2555,
      "step": 88252
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.8545529842376709,
      "learning_rate": 0.00040777059431644497,
      "loss": 2.042,
      "step": 88253
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.8406326174736023,
      "learning_rate": 0.00040776677680149487,
      "loss": 2.3027,
      "step": 88254
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.9467542767524719,
      "learning_rate": 0.0004077629592665089,
      "loss": 2.1745,
      "step": 88255
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.8916800618171692,
      "learning_rate": 0.00040775914171148776,
      "loss": 2.3691,
      "step": 88256
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.8766341805458069,
      "learning_rate": 0.00040775532413643215,
      "loss": 2.1919,
      "step": 88257
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.9159137606620789,
      "learning_rate": 0.00040775150654134276,
      "loss": 2.4197,
      "step": 88258
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.9129489064216614,
      "learning_rate": 0.00040774768892622046,
      "loss": 2.2465,
      "step": 88259
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.8303890824317932,
      "learning_rate": 0.00040774387129106573,
      "loss": 2.4391,
      "step": 88260
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.7954140901565552,
      "learning_rate": 0.00040774005363587945,
      "loss": 2.2824,
      "step": 88261
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.8732545375823975,
      "learning_rate": 0.0004077362359606622,
      "loss": 2.2037,
      "step": 88262
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.8009039759635925,
      "learning_rate": 0.0004077324182654147,
      "loss": 1.933,
      "step": 88263
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.8071821331977844,
      "learning_rate": 0.0004077286005501378,
      "loss": 1.9501,
      "step": 88264
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.9030634164810181,
      "learning_rate": 0.00040772478281483215,
      "loss": 1.9566,
      "step": 88265
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.8296165466308594,
      "learning_rate": 0.00040772096505949833,
      "loss": 2.08,
      "step": 88266
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.8330511450767517,
      "learning_rate": 0.0004077171472841372,
      "loss": 2.0393,
      "step": 88267
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.7520802617073059,
      "learning_rate": 0.0004077133294887493,
      "loss": 2.1375,
      "step": 88268
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.8357809782028198,
      "learning_rate": 0.00040770951167333567,
      "loss": 2.3027,
      "step": 88269
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.7986066937446594,
      "learning_rate": 0.00040770569383789666,
      "loss": 2.0163,
      "step": 88270
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.9014894366264343,
      "learning_rate": 0.00040770187598243313,
      "loss": 2.3125,
      "step": 88271
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.8888260722160339,
      "learning_rate": 0.0004076980581069459,
      "loss": 2.1092,
      "step": 88272
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.8795824646949768,
      "learning_rate": 0.0004076942402114355,
      "loss": 2.1925,
      "step": 88273
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.8451125025749207,
      "learning_rate": 0.00040769042229590267,
      "loss": 2.2499,
      "step": 88274
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.8667458295822144,
      "learning_rate": 0.00040768660436034816,
      "loss": 2.0428,
      "step": 88275
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.817053496837616,
      "learning_rate": 0.0004076827864047728,
      "loss": 2.1237,
      "step": 88276
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.7993965148925781,
      "learning_rate": 0.00040767896842917704,
      "loss": 2.2642,
      "step": 88277
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.8400589823722839,
      "learning_rate": 0.0004076751504335618,
      "loss": 2.0855,
      "step": 88278
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.7425418496131897,
      "learning_rate": 0.0004076713324179276,
      "loss": 2.1688,
      "step": 88279
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.8520342111587524,
      "learning_rate": 0.00040766751438227545,
      "loss": 2.0138,
      "step": 88280
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.8694045543670654,
      "learning_rate": 0.00040766369632660575,
      "loss": 2.2037,
      "step": 88281
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.7819096446037292,
      "learning_rate": 0.0004076598782509194,
      "loss": 2.2102,
      "step": 88282
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.7888705134391785,
      "learning_rate": 0.000407656060155217,
      "loss": 1.8868,
      "step": 88283
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.8098922371864319,
      "learning_rate": 0.00040765224203949933,
      "loss": 2.2792,
      "step": 88284
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.9068189859390259,
      "learning_rate": 0.0004076484239037671,
      "loss": 2.4676,
      "step": 88285
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.8075213432312012,
      "learning_rate": 0.000407644605748021,
      "loss": 2.0325,
      "step": 88286
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.7984886765480042,
      "learning_rate": 0.0004076407875722617,
      "loss": 2.105,
      "step": 88287
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.917730987071991,
      "learning_rate": 0.00040763696937648995,
      "loss": 2.2889,
      "step": 88288
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.79146409034729,
      "learning_rate": 0.0004076331511607065,
      "loss": 2.37,
      "step": 88289
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.7953107953071594,
      "learning_rate": 0.00040762933292491193,
      "loss": 2.3202,
      "step": 88290
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.8545888662338257,
      "learning_rate": 0.0004076255146691071,
      "loss": 2.2497,
      "step": 88291
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.8988995552062988,
      "learning_rate": 0.00040762169639329263,
      "loss": 2.2167,
      "step": 88292
    },
    {
      "epoch": 1.15,
      "grad_norm": 1.0155726671218872,
      "learning_rate": 0.00040761787809746923,
      "loss": 2.206,
      "step": 88293
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.8518314957618713,
      "learning_rate": 0.00040761405978163764,
      "loss": 2.1201,
      "step": 88294
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.9126819372177124,
      "learning_rate": 0.00040761024144579867,
      "loss": 2.139,
      "step": 88295
    },
    {
      "epoch": 1.15,
      "grad_norm": 1.0263006687164307,
      "learning_rate": 0.0004076064230899528,
      "loss": 2.1224,
      "step": 88296
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.9610795378684998,
      "learning_rate": 0.0004076026047141009,
      "loss": 2.1304,
      "step": 88297
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.8421462774276733,
      "learning_rate": 0.00040759878631824365,
      "loss": 2.2654,
      "step": 88298
    },
    {
      "epoch": 1.15,
      "grad_norm": 1.4282160997390747,
      "learning_rate": 0.0004075949679023818,
      "loss": 2.1899,
      "step": 88299
    },
    {
      "epoch": 1.15,
      "grad_norm": 1.1837760210037231,
      "learning_rate": 0.0004075911494665159,
      "loss": 2.1449,
      "step": 88300
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.7737247943878174,
      "learning_rate": 0.0004075873310106469,
      "loss": 2.1817,
      "step": 88301
    },
    {
      "epoch": 1.15,
      "grad_norm": 1.025260329246521,
      "learning_rate": 0.0004075835125347753,
      "loss": 2.1065,
      "step": 88302
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.745359480381012,
      "learning_rate": 0.0004075796940389019,
      "loss": 2.0693,
      "step": 88303
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.9728919267654419,
      "learning_rate": 0.0004075758755230275,
      "loss": 2.0613,
      "step": 88304
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.8382325172424316,
      "learning_rate": 0.00040757205698715266,
      "loss": 2.0361,
      "step": 88305
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.9037917852401733,
      "learning_rate": 0.00040756823843127806,
      "loss": 2.067,
      "step": 88306
    },
    {
      "epoch": 1.15,
      "grad_norm": 1.1308000087738037,
      "learning_rate": 0.0004075644198554046,
      "loss": 2.1262,
      "step": 88307
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.8199923634529114,
      "learning_rate": 0.0004075606012595328,
      "loss": 2.2237,
      "step": 88308
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.8880725502967834,
      "learning_rate": 0.00040755678264366347,
      "loss": 2.3027,
      "step": 88309
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.9519597291946411,
      "learning_rate": 0.00040755296400779735,
      "loss": 1.981,
      "step": 88310
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.9383143186569214,
      "learning_rate": 0.00040754914535193513,
      "loss": 2.1298,
      "step": 88311
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.8717637658119202,
      "learning_rate": 0.0004075453266760774,
      "loss": 2.0696,
      "step": 88312
    },
    {
      "epoch": 1.15,
      "grad_norm": 1.1395372152328491,
      "learning_rate": 0.0004075415079802251,
      "loss": 2.2554,
      "step": 88313
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.8676615357398987,
      "learning_rate": 0.00040753768926437867,
      "loss": 2.3321,
      "step": 88314
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.7907096743583679,
      "learning_rate": 0.000407533870528539,
      "loss": 2.2135,
      "step": 88315
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.7997079491615295,
      "learning_rate": 0.0004075300517727067,
      "loss": 2.2805,
      "step": 88316
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.9132505059242249,
      "learning_rate": 0.00040752623299688265,
      "loss": 2.0237,
      "step": 88317
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.8049216270446777,
      "learning_rate": 0.0004075224142010674,
      "loss": 2.0782,
      "step": 88318
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.8089439868927002,
      "learning_rate": 0.00040751859538526165,
      "loss": 2.3106,
      "step": 88319
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.7974227070808411,
      "learning_rate": 0.00040751477654946627,
      "loss": 2.1571,
      "step": 88320
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.8909398317337036,
      "learning_rate": 0.0004075109576936817,
      "loss": 2.3495,
      "step": 88321
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.8399389386177063,
      "learning_rate": 0.0004075071388179089,
      "loss": 2.0788,
      "step": 88322
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.877203643321991,
      "learning_rate": 0.0004075033199221486,
      "loss": 2.1861,
      "step": 88323
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.8868818879127502,
      "learning_rate": 0.00040749950100640127,
      "loss": 2.2145,
      "step": 88324
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.8750771880149841,
      "learning_rate": 0.00040749568207066776,
      "loss": 1.9865,
      "step": 88325
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.8821606040000916,
      "learning_rate": 0.0004074918631149488,
      "loss": 2.0091,
      "step": 88326
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.9588924646377563,
      "learning_rate": 0.0004074880441392451,
      "loss": 2.2073,
      "step": 88327
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.8562143445014954,
      "learning_rate": 0.0004074842251435573,
      "loss": 2.0828,
      "step": 88328
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.9977943897247314,
      "learning_rate": 0.0004074804061278862,
      "loss": 2.2107,
      "step": 88329
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.8490920066833496,
      "learning_rate": 0.0004074765870922324,
      "loss": 2.1394,
      "step": 88330
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.8491015434265137,
      "learning_rate": 0.0004074727680365967,
      "loss": 2.0488,
      "step": 88331
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.8229488134384155,
      "learning_rate": 0.00040746894896097986,
      "loss": 2.1311,
      "step": 88332
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.977853536605835,
      "learning_rate": 0.0004074651298653824,
      "loss": 2.4072,
      "step": 88333
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.9550285935401917,
      "learning_rate": 0.00040746131074980514,
      "loss": 1.9904,
      "step": 88334
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.8784709572792053,
      "learning_rate": 0.0004074574916142489,
      "loss": 2.4053,
      "step": 88335
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.8998745083808899,
      "learning_rate": 0.00040745367245871426,
      "loss": 1.9792,
      "step": 88336
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.9601909518241882,
      "learning_rate": 0.0004074498532832019,
      "loss": 2.1187,
      "step": 88337
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.7799609899520874,
      "learning_rate": 0.00040744603408771263,
      "loss": 1.9994,
      "step": 88338
    },
    {
      "epoch": 1.15,
      "grad_norm": 1.082335114479065,
      "learning_rate": 0.00040744221487224705,
      "loss": 2.0956,
      "step": 88339
    },
    {
      "epoch": 1.15,
      "grad_norm": 1.1305001974105835,
      "learning_rate": 0.00040743839563680596,
      "loss": 2.1966,
      "step": 88340
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.8977283239364624,
      "learning_rate": 0.0004074345763813901,
      "loss": 2.2394,
      "step": 88341
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.8615903258323669,
      "learning_rate": 0.0004074307571060001,
      "loss": 2.2835,
      "step": 88342
    },
    {
      "epoch": 1.15,
      "grad_norm": 1.1993014812469482,
      "learning_rate": 0.00040742693781063667,
      "loss": 2.0566,
      "step": 88343
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.7960038185119629,
      "learning_rate": 0.0004074231184953005,
      "loss": 1.9458,
      "step": 88344
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.8785862922668457,
      "learning_rate": 0.0004074192991599924,
      "loss": 2.1042,
      "step": 88345
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.9168069958686829,
      "learning_rate": 0.000407415479804713,
      "loss": 2.0634,
      "step": 88346
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.9214432835578918,
      "learning_rate": 0.0004074116604294631,
      "loss": 2.4309,
      "step": 88347
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.8598825335502625,
      "learning_rate": 0.00040740784103424326,
      "loss": 2.2101,
      "step": 88348
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.8043554425239563,
      "learning_rate": 0.0004074040216190543,
      "loss": 2.3146,
      "step": 88349
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.9337857961654663,
      "learning_rate": 0.00040740020218389694,
      "loss": 2.2053,
      "step": 88350
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.88239985704422,
      "learning_rate": 0.0004073963827287718,
      "loss": 2.4005,
      "step": 88351
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.8203070759773254,
      "learning_rate": 0.0004073925632536797,
      "loss": 2.0759,
      "step": 88352
    },
    {
      "epoch": 1.15,
      "grad_norm": 1.093141794204712,
      "learning_rate": 0.0004073887437586213,
      "loss": 2.3036,
      "step": 88353
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.9168210625648499,
      "learning_rate": 0.0004073849242435973,
      "loss": 2.1185,
      "step": 88354
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.9041634798049927,
      "learning_rate": 0.0004073811047086084,
      "loss": 2.2933,
      "step": 88355
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.9141364693641663,
      "learning_rate": 0.00040737728515365525,
      "loss": 2.2323,
      "step": 88356
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.984965443611145,
      "learning_rate": 0.0004073734655787388,
      "loss": 2.015,
      "step": 88357
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.8293417692184448,
      "learning_rate": 0.0004073696459838595,
      "loss": 2.1393,
      "step": 88358
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.9856275320053101,
      "learning_rate": 0.0004073658263690181,
      "loss": 2.0687,
      "step": 88359
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.8913910984992981,
      "learning_rate": 0.0004073620067342154,
      "loss": 2.1629,
      "step": 88360
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.8443153500556946,
      "learning_rate": 0.00040735818707945216,
      "loss": 2.19,
      "step": 88361
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.832298994064331,
      "learning_rate": 0.00040735436740472894,
      "loss": 2.0815,
      "step": 88362
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.9610463976860046,
      "learning_rate": 0.00040735054771004643,
      "loss": 2.1481,
      "step": 88363
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.8936040997505188,
      "learning_rate": 0.0004073467279954056,
      "loss": 2.1803,
      "step": 88364
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.7813976407051086,
      "learning_rate": 0.0004073429082608069,
      "loss": 2.164,
      "step": 88365
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.9276866912841797,
      "learning_rate": 0.0004073390885062511,
      "loss": 2.332,
      "step": 88366
    },
    {
      "epoch": 1.15,
      "grad_norm": 1.1951045989990234,
      "learning_rate": 0.00040733526873173894,
      "loss": 1.9279,
      "step": 88367
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.7549406886100769,
      "learning_rate": 0.00040733144893727124,
      "loss": 2.1912,
      "step": 88368
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.901757001876831,
      "learning_rate": 0.0004073276291228485,
      "loss": 2.1605,
      "step": 88369
    },
    {
      "epoch": 1.15,
      "grad_norm": 1.072979211807251,
      "learning_rate": 0.00040732380928847154,
      "loss": 2.2647,
      "step": 88370
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.9371360540390015,
      "learning_rate": 0.00040731998943414107,
      "loss": 2.1394,
      "step": 88371
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.8109437227249146,
      "learning_rate": 0.0004073161695598578,
      "loss": 2.1386,
      "step": 88372
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.862178385257721,
      "learning_rate": 0.00040731234966562236,
      "loss": 2.3374,
      "step": 88373
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.8794690370559692,
      "learning_rate": 0.00040730852975143564,
      "loss": 2.1691,
      "step": 88374
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.9417001605033875,
      "learning_rate": 0.0004073047098172982,
      "loss": 2.0417,
      "step": 88375
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.9169811606407166,
      "learning_rate": 0.0004073008898632107,
      "loss": 2.0878,
      "step": 88376
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.8765475749969482,
      "learning_rate": 0.000407297069889174,
      "loss": 1.962,
      "step": 88377
    },
    {
      "epoch": 1.15,
      "grad_norm": 1.1927944421768188,
      "learning_rate": 0.0004072932498951888,
      "loss": 2.2225,
      "step": 88378
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.8376490473747253,
      "learning_rate": 0.00040728942988125567,
      "loss": 1.9911,
      "step": 88379
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.755351185798645,
      "learning_rate": 0.0004072856098473754,
      "loss": 2.0496,
      "step": 88380
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.9087139964103699,
      "learning_rate": 0.0004072817897935488,
      "loss": 2.175,
      "step": 88381
    },
    {
      "epoch": 1.15,
      "grad_norm": 1.0132943391799927,
      "learning_rate": 0.0004072779697197765,
      "loss": 1.9908,
      "step": 88382
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.8137099742889404,
      "learning_rate": 0.0004072741496260591,
      "loss": 2.11,
      "step": 88383
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.8303378820419312,
      "learning_rate": 0.0004072703295123975,
      "loss": 2.1142,
      "step": 88384
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.8298972845077515,
      "learning_rate": 0.0004072665093787923,
      "loss": 2.0507,
      "step": 88385
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.9410343766212463,
      "learning_rate": 0.0004072626892252442,
      "loss": 2.2719,
      "step": 88386
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.8072484135627747,
      "learning_rate": 0.000407258869051754,
      "loss": 2.1882,
      "step": 88387
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.7492977380752563,
      "learning_rate": 0.0004072550488583223,
      "loss": 2.0121,
      "step": 88388
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.816032886505127,
      "learning_rate": 0.00040725122864494984,
      "loss": 2.1784,
      "step": 88389
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.9271597862243652,
      "learning_rate": 0.00040724740841163737,
      "loss": 2.143,
      "step": 88390
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.8949503302574158,
      "learning_rate": 0.0004072435881583856,
      "loss": 1.9923,
      "step": 88391
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.8462399244308472,
      "learning_rate": 0.00040723976788519526,
      "loss": 2.1613,
      "step": 88392
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.9631560444831848,
      "learning_rate": 0.000407235947592067,
      "loss": 2.1914,
      "step": 88393
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.9773625135421753,
      "learning_rate": 0.0004072321272790015,
      "loss": 2.3338,
      "step": 88394
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.7945072054862976,
      "learning_rate": 0.0004072283069459995,
      "loss": 2.3152,
      "step": 88395
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.8612287640571594,
      "learning_rate": 0.0004072244865930619,
      "loss": 2.4556,
      "step": 88396
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.8922075629234314,
      "learning_rate": 0.0004072206662201891,
      "loss": 2.2471,
      "step": 88397
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.8014281988143921,
      "learning_rate": 0.000407216845827382,
      "loss": 2.2086,
      "step": 88398
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.8544937372207642,
      "learning_rate": 0.0004072130254146412,
      "loss": 2.3243,
      "step": 88399
    },
    {
      "epoch": 1.15,
      "grad_norm": 1.0630323886871338,
      "learning_rate": 0.00040720920498196755,
      "loss": 2.4349,
      "step": 88400
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.8540525436401367,
      "learning_rate": 0.00040720538452936164,
      "loss": 2.165,
      "step": 88401
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.8041746616363525,
      "learning_rate": 0.00040720156405682435,
      "loss": 2.3677,
      "step": 88402
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.7951475977897644,
      "learning_rate": 0.00040719774356435607,
      "loss": 2.0821,
      "step": 88403
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.8009966611862183,
      "learning_rate": 0.00040719392305195776,
      "loss": 2.3704,
      "step": 88404
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.8772698640823364,
      "learning_rate": 0.0004071901025196302,
      "loss": 2.2205,
      "step": 88405
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.9279611706733704,
      "learning_rate": 0.0004071862819673739,
      "loss": 2.0305,
      "step": 88406
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.8677634596824646,
      "learning_rate": 0.0004071824613951896,
      "loss": 2.2278,
      "step": 88407
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.9870067834854126,
      "learning_rate": 0.00040717864080307815,
      "loss": 2.3829,
      "step": 88408
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.8317593336105347,
      "learning_rate": 0.00040717482019104014,
      "loss": 2.1856,
      "step": 88409
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.7968392968177795,
      "learning_rate": 0.0004071709995590762,
      "loss": 2.1107,
      "step": 88410
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.7746737599372864,
      "learning_rate": 0.00040716717890718724,
      "loss": 2.2761,
      "step": 88411
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.9137381911277771,
      "learning_rate": 0.0004071633582353739,
      "loss": 2.249,
      "step": 88412
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.8849868178367615,
      "learning_rate": 0.0004071595375436368,
      "loss": 2.1388,
      "step": 88413
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.8979930877685547,
      "learning_rate": 0.00040715571683197683,
      "loss": 2.1643,
      "step": 88414
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.9394835829734802,
      "learning_rate": 0.00040715189610039454,
      "loss": 2.2256,
      "step": 88415
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.8223353624343872,
      "learning_rate": 0.0004071480753488906,
      "loss": 2.1979,
      "step": 88416
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.8748842477798462,
      "learning_rate": 0.0004071442545774659,
      "loss": 2.1878,
      "step": 88417
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.7903693914413452,
      "learning_rate": 0.000407140433786121,
      "loss": 1.9644,
      "step": 88418
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.843312680721283,
      "learning_rate": 0.00040713661297485674,
      "loss": 2.1709,
      "step": 88419
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.8673500418663025,
      "learning_rate": 0.0004071327921436738,
      "loss": 1.9552,
      "step": 88420
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.8640746474266052,
      "learning_rate": 0.0004071289712925727,
      "loss": 2.2812,
      "step": 88421
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.878299355506897,
      "learning_rate": 0.0004071251504215544,
      "loss": 2.3141,
      "step": 88422
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.8946450352668762,
      "learning_rate": 0.0004071213295306195,
      "loss": 2.1887,
      "step": 88423
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.8253207802772522,
      "learning_rate": 0.0004071175086197687,
      "loss": 2.0351,
      "step": 88424
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.784701943397522,
      "learning_rate": 0.00040711368768900277,
      "loss": 2.1965,
      "step": 88425
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.8562806248664856,
      "learning_rate": 0.0004071098667383224,
      "loss": 2.187,
      "step": 88426
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.9242278933525085,
      "learning_rate": 0.0004071060457677282,
      "loss": 2.1546,
      "step": 88427
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.8171406984329224,
      "learning_rate": 0.000407102224777221,
      "loss": 2.1026,
      "step": 88428
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.7781291604042053,
      "learning_rate": 0.00040709840376680157,
      "loss": 2.0041,
      "step": 88429
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.891510009765625,
      "learning_rate": 0.0004070945827364704,
      "loss": 2.204,
      "step": 88430
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.8524336218833923,
      "learning_rate": 0.00040709076168622834,
      "loss": 2.2092,
      "step": 88431
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.839524507522583,
      "learning_rate": 0.00040708694061607616,
      "loss": 2.2372,
      "step": 88432
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.7980960607528687,
      "learning_rate": 0.00040708311952601444,
      "loss": 2.1646,
      "step": 88433
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.8885543346405029,
      "learning_rate": 0.0004070792984160439,
      "loss": 2.0991,
      "step": 88434
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.9091883301734924,
      "learning_rate": 0.00040707547728616547,
      "loss": 2.4303,
      "step": 88435
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.8962149024009705,
      "learning_rate": 0.00040707165613637956,
      "loss": 2.3697,
      "step": 88436
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.8651623129844666,
      "learning_rate": 0.00040706783496668693,
      "loss": 2.1304,
      "step": 88437
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.8726295232772827,
      "learning_rate": 0.0004070640137770885,
      "loss": 2.2084,
      "step": 88438
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.9861281514167786,
      "learning_rate": 0.00040706019256758487,
      "loss": 2.0901,
      "step": 88439
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.7747549414634705,
      "learning_rate": 0.0004070563713381766,
      "loss": 2.0535,
      "step": 88440
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.7755675315856934,
      "learning_rate": 0.0004070525500888646,
      "loss": 2.1424,
      "step": 88441
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.8313921689987183,
      "learning_rate": 0.0004070487288196495,
      "loss": 2.2175,
      "step": 88442
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.8283339738845825,
      "learning_rate": 0.0004070449075305321,
      "loss": 2.1272,
      "step": 88443
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.7897301316261292,
      "learning_rate": 0.0004070410862215129,
      "loss": 1.9673,
      "step": 88444
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.8298853039741516,
      "learning_rate": 0.0004070372648925928,
      "loss": 2.2213,
      "step": 88445
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.8880888819694519,
      "learning_rate": 0.0004070334435437724,
      "loss": 2.3417,
      "step": 88446
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.8477353453636169,
      "learning_rate": 0.0004070296221750526,
      "loss": 2.3464,
      "step": 88447
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.7970949411392212,
      "learning_rate": 0.00040702580078643384,
      "loss": 2.071,
      "step": 88448
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.7185832858085632,
      "learning_rate": 0.000407021979377917,
      "loss": 2.1564,
      "step": 88449
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.9274612069129944,
      "learning_rate": 0.00040701815794950275,
      "loss": 2.2244,
      "step": 88450
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.8463160991668701,
      "learning_rate": 0.00040701433650119185,
      "loss": 1.9068,
      "step": 88451
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.8860607743263245,
      "learning_rate": 0.000407010515032985,
      "loss": 2.0248,
      "step": 88452
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.9102072715759277,
      "learning_rate": 0.0004070066935448827,
      "loss": 2.0312,
      "step": 88453
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.9831545352935791,
      "learning_rate": 0.00040700287203688596,
      "loss": 2.2161,
      "step": 88454
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.8360318541526794,
      "learning_rate": 0.0004069990505089953,
      "loss": 2.1622,
      "step": 88455
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.8108372688293457,
      "learning_rate": 0.0004069952289612116,
      "loss": 2.1093,
      "step": 88456
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.9603220820426941,
      "learning_rate": 0.0004069914073935354,
      "loss": 2.38,
      "step": 88457
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.8288470506668091,
      "learning_rate": 0.00040698758580596747,
      "loss": 1.968,
      "step": 88458
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.7832897901535034,
      "learning_rate": 0.00040698376419850854,
      "loss": 2.1265,
      "step": 88459
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.8638613224029541,
      "learning_rate": 0.00040697994257115944,
      "loss": 2.2232,
      "step": 88460
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.9366202354431152,
      "learning_rate": 0.00040697612092392054,
      "loss": 2.0291,
      "step": 88461
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.8858979344367981,
      "learning_rate": 0.00040697229925679287,
      "loss": 1.9536,
      "step": 88462
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.9578703045845032,
      "learning_rate": 0.00040696847756977703,
      "loss": 1.9953,
      "step": 88463
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.8634811639785767,
      "learning_rate": 0.00040696465586287367,
      "loss": 2.1254,
      "step": 88464
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.8802748918533325,
      "learning_rate": 0.0004069608341360836,
      "loss": 1.9944,
      "step": 88465
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.9644575715065002,
      "learning_rate": 0.00040695701238940753,
      "loss": 2.2337,
      "step": 88466
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.7971338033676147,
      "learning_rate": 0.0004069531906228461,
      "loss": 2.1283,
      "step": 88467
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.9301930069923401,
      "learning_rate": 0.0004069493688364,
      "loss": 2.0365,
      "step": 88468
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.9152019619941711,
      "learning_rate": 0.00040694554703007014,
      "loss": 2.1175,
      "step": 88469
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.8102722764015198,
      "learning_rate": 0.00040694172520385695,
      "loss": 2.0885,
      "step": 88470
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.8898169994354248,
      "learning_rate": 0.00040693790335776127,
      "loss": 2.2765,
      "step": 88471
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.7795596122741699,
      "learning_rate": 0.0004069340814917839,
      "loss": 2.0694,
      "step": 88472
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.8263857364654541,
      "learning_rate": 0.0004069302596059254,
      "loss": 2.3967,
      "step": 88473
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.8668344020843506,
      "learning_rate": 0.0004069264377001866,
      "loss": 2.1519,
      "step": 88474
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.800183892250061,
      "learning_rate": 0.00040692261577456815,
      "loss": 1.9404,
      "step": 88475
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.9481855034828186,
      "learning_rate": 0.0004069187938290707,
      "loss": 2.1008,
      "step": 88476
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.8769708871841431,
      "learning_rate": 0.00040691497186369507,
      "loss": 1.9926,
      "step": 88477
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.8085472583770752,
      "learning_rate": 0.0004069111498784419,
      "loss": 2.2818,
      "step": 88478
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.7948772311210632,
      "learning_rate": 0.00040690732787331197,
      "loss": 1.9883,
      "step": 88479
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.7959802150726318,
      "learning_rate": 0.0004069035058483059,
      "loss": 2.115,
      "step": 88480
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.8945175409317017,
      "learning_rate": 0.0004068996838034246,
      "loss": 2.4014,
      "step": 88481
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.8383381962776184,
      "learning_rate": 0.00040689586173866843,
      "loss": 2.2643,
      "step": 88482
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.9136112928390503,
      "learning_rate": 0.00040689203965403836,
      "loss": 2.1584,
      "step": 88483
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.8046450018882751,
      "learning_rate": 0.00040688821754953513,
      "loss": 2.1186,
      "step": 88484
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.9861910939216614,
      "learning_rate": 0.0004068843954251593,
      "loss": 2.6528,
      "step": 88485
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.806018590927124,
      "learning_rate": 0.00040688057328091165,
      "loss": 2.1186,
      "step": 88486
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.889938473701477,
      "learning_rate": 0.0004068767511167929,
      "loss": 2.1363,
      "step": 88487
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.8937021493911743,
      "learning_rate": 0.00040687292893280374,
      "loss": 2.1455,
      "step": 88488
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.8453846573829651,
      "learning_rate": 0.0004068691067289449,
      "loss": 1.973,
      "step": 88489
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.9108036756515503,
      "learning_rate": 0.000406865284505217,
      "loss": 2.151,
      "step": 88490
    },
    {
      "epoch": 1.15,
      "grad_norm": 1.1181930303573608,
      "learning_rate": 0.0004068614622616209,
      "loss": 2.3084,
      "step": 88491
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.7951861619949341,
      "learning_rate": 0.00040685763999815717,
      "loss": 2.1672,
      "step": 88492
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.8581668734550476,
      "learning_rate": 0.00040685381771482667,
      "loss": 2.1839,
      "step": 88493
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.926863431930542,
      "learning_rate": 0.00040684999541162995,
      "loss": 2.2119,
      "step": 88494
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.897060215473175,
      "learning_rate": 0.0004068461730885678,
      "loss": 2.3367,
      "step": 88495
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.7922269105911255,
      "learning_rate": 0.00040684235074564106,
      "loss": 2.3295,
      "step": 88496
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.8445848822593689,
      "learning_rate": 0.0004068385283828502,
      "loss": 2.3491,
      "step": 88497
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.8099338412284851,
      "learning_rate": 0.00040683470600019604,
      "loss": 2.2415,
      "step": 88498
    },
    {
      "epoch": 1.15,
      "grad_norm": 1.27718186378479,
      "learning_rate": 0.00040683088359767935,
      "loss": 2.3557,
      "step": 88499
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.7736222147941589,
      "learning_rate": 0.00040682706117530073,
      "loss": 2.1088,
      "step": 88500
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.7769547700881958,
      "learning_rate": 0.00040682323873306094,
      "loss": 1.8724,
      "step": 88501
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.9129093885421753,
      "learning_rate": 0.0004068194162709607,
      "loss": 2.1021,
      "step": 88502
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.9981169104576111,
      "learning_rate": 0.0004068155937890007,
      "loss": 2.215,
      "step": 88503
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.8451490998268127,
      "learning_rate": 0.0004068117712871817,
      "loss": 2.0153,
      "step": 88504
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.8778603076934814,
      "learning_rate": 0.0004068079487655044,
      "loss": 2.2243,
      "step": 88505
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.8503972291946411,
      "learning_rate": 0.0004068041262239694,
      "loss": 2.3217,
      "step": 88506
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.7953847646713257,
      "learning_rate": 0.0004068003036625775,
      "loss": 2.1966,
      "step": 88507
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.8076450228691101,
      "learning_rate": 0.0004067964810813295,
      "loss": 2.3038,
      "step": 88508
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.8678512573242188,
      "learning_rate": 0.00040679265848022597,
      "loss": 2.2982,
      "step": 88509
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.912429928779602,
      "learning_rate": 0.00040678883585926765,
      "loss": 2.085,
      "step": 88510
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.9350414872169495,
      "learning_rate": 0.00040678501321845536,
      "loss": 2.141,
      "step": 88511
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.8008240461349487,
      "learning_rate": 0.0004067811905577896,
      "loss": 2.1365,
      "step": 88512
    },
    {
      "epoch": 1.15,
      "grad_norm": 1.0735194683074951,
      "learning_rate": 0.0004067773678772712,
      "loss": 2.0052,
      "step": 88513
    },
    {
      "epoch": 1.15,
      "grad_norm": 1.0677348375320435,
      "learning_rate": 0.000406773545176901,
      "loss": 1.9744,
      "step": 88514
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.7678468227386475,
      "learning_rate": 0.0004067697224566795,
      "loss": 2.0564,
      "step": 88515
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.8530286550521851,
      "learning_rate": 0.00040676589971660744,
      "loss": 2.0906,
      "step": 88516
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.9201841354370117,
      "learning_rate": 0.0004067620769566856,
      "loss": 2.0497,
      "step": 88517
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.9347488880157471,
      "learning_rate": 0.0004067582541769148,
      "loss": 2.2974,
      "step": 88518
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.9840487241744995,
      "learning_rate": 0.0004067544313772955,
      "loss": 2.091,
      "step": 88519
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.828439474105835,
      "learning_rate": 0.0004067506085578286,
      "loss": 2.1954,
      "step": 88520
    },
    {
      "epoch": 1.15,
      "grad_norm": 1.0046652555465698,
      "learning_rate": 0.00040674678571851476,
      "loss": 2.176,
      "step": 88521
    },
    {
      "epoch": 1.15,
      "grad_norm": 1.1632344722747803,
      "learning_rate": 0.0004067429628593546,
      "loss": 2.0968,
      "step": 88522
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.9088526964187622,
      "learning_rate": 0.00040673913998034896,
      "loss": 2.3086,
      "step": 88523
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.9614778757095337,
      "learning_rate": 0.00040673531708149856,
      "loss": 2.0464,
      "step": 88524
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.9491071701049805,
      "learning_rate": 0.0004067314941628039,
      "loss": 2.1111,
      "step": 88525
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.7726824879646301,
      "learning_rate": 0.0004067276712242659,
      "loss": 1.9973,
      "step": 88526
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.8402042984962463,
      "learning_rate": 0.00040672384826588527,
      "loss": 2.3864,
      "step": 88527
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.8373243808746338,
      "learning_rate": 0.00040672002528766265,
      "loss": 2.3137,
      "step": 88528
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.8550696969032288,
      "learning_rate": 0.0004067162022895987,
      "loss": 2.2004,
      "step": 88529
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.9157741069793701,
      "learning_rate": 0.0004067123792716943,
      "loss": 2.0326,
      "step": 88530
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.853772759437561,
      "learning_rate": 0.00040670855623394994,
      "loss": 2.0856,
      "step": 88531
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.8838766813278198,
      "learning_rate": 0.00040670473317636644,
      "loss": 2.1646,
      "step": 88532
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.8862361311912537,
      "learning_rate": 0.0004067009100989445,
      "loss": 2.1207,
      "step": 88533
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.9771173000335693,
      "learning_rate": 0.00040669708700168496,
      "loss": 2.2492,
      "step": 88534
    },
    {
      "epoch": 1.15,
      "grad_norm": 1.0312594175338745,
      "learning_rate": 0.0004066932638845884,
      "loss": 2.252,
      "step": 88535
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.8588435053825378,
      "learning_rate": 0.0004066894407476555,
      "loss": 2.0499,
      "step": 88536
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.9131376147270203,
      "learning_rate": 0.00040668561759088707,
      "loss": 2.3556,
      "step": 88537
    },
    {
      "epoch": 1.15,
      "grad_norm": 1.0530544519424438,
      "learning_rate": 0.0004066817944142837,
      "loss": 2.2626,
      "step": 88538
    },
    {
      "epoch": 1.15,
      "grad_norm": 1.0350528955459595,
      "learning_rate": 0.0004066779712178462,
      "loss": 2.1585,
      "step": 88539
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.9305988550186157,
      "learning_rate": 0.00040667414800157523,
      "loss": 2.0729,
      "step": 88540
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.8940351605415344,
      "learning_rate": 0.0004066703247654716,
      "loss": 2.2492,
      "step": 88541
    },
    {
      "epoch": 1.15,
      "grad_norm": 1.4820750951766968,
      "learning_rate": 0.00040666650150953586,
      "loss": 2.193,
      "step": 88542
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.7831306457519531,
      "learning_rate": 0.00040666267823376886,
      "loss": 2.1542,
      "step": 88543
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.8467082381248474,
      "learning_rate": 0.00040665885493817126,
      "loss": 2.1822,
      "step": 88544
    },
    {
      "epoch": 1.15,
      "grad_norm": 1.031123161315918,
      "learning_rate": 0.0004066550316227437,
      "loss": 2.2505,
      "step": 88545
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.9453560709953308,
      "learning_rate": 0.000406651208287487,
      "loss": 2.1277,
      "step": 88546
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.827061653137207,
      "learning_rate": 0.0004066473849324018,
      "loss": 2.0659,
      "step": 88547
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.8633909821510315,
      "learning_rate": 0.00040664356155748887,
      "loss": 2.3685,
      "step": 88548
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.8206118941307068,
      "learning_rate": 0.00040663973816274884,
      "loss": 2.1797,
      "step": 88549
    },
    {
      "epoch": 1.15,
      "grad_norm": 1.0998414754867554,
      "learning_rate": 0.0004066359147481825,
      "loss": 2.2255,
      "step": 88550
    },
    {
      "epoch": 1.15,
      "grad_norm": 1.0275276899337769,
      "learning_rate": 0.0004066320913137906,
      "loss": 2.087,
      "step": 88551
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.7978743314743042,
      "learning_rate": 0.00040662826785957374,
      "loss": 2.1222,
      "step": 88552
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.8089885115623474,
      "learning_rate": 0.0004066244443855326,
      "loss": 2.037,
      "step": 88553
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.9343675374984741,
      "learning_rate": 0.000406620620891668,
      "loss": 2.1637,
      "step": 88554
    },
    {
      "epoch": 1.15,
      "grad_norm": 1.0190719366073608,
      "learning_rate": 0.00040661679737798064,
      "loss": 2.217,
      "step": 88555
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.7813451886177063,
      "learning_rate": 0.0004066129738444712,
      "loss": 2.2419,
      "step": 88556
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.9638329744338989,
      "learning_rate": 0.0004066091502911404,
      "loss": 2.4535,
      "step": 88557
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.7897276282310486,
      "learning_rate": 0.0004066053267179889,
      "loss": 2.1122,
      "step": 88558
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.9586350917816162,
      "learning_rate": 0.00040660150312501754,
      "loss": 2.2663,
      "step": 88559
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.8406252861022949,
      "learning_rate": 0.00040659767951222695,
      "loss": 2.3412,
      "step": 88560
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.7601698637008667,
      "learning_rate": 0.00040659385587961774,
      "loss": 2.009,
      "step": 88561
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.8566228151321411,
      "learning_rate": 0.0004065900322271907,
      "loss": 2.1128,
      "step": 88562
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.9176209568977356,
      "learning_rate": 0.0004065862085549467,
      "loss": 2.35,
      "step": 88563
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.8935914039611816,
      "learning_rate": 0.00040658238486288625,
      "loss": 2.2408,
      "step": 88564
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.8487920761108398,
      "learning_rate": 0.00040657856115101015,
      "loss": 2.1416,
      "step": 88565
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.8827618360519409,
      "learning_rate": 0.00040657473741931906,
      "loss": 2.3014,
      "step": 88566
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.9457790851593018,
      "learning_rate": 0.0004065709136678137,
      "loss": 2.2489,
      "step": 88567
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.7903938293457031,
      "learning_rate": 0.0004065670898964948,
      "loss": 2.256,
      "step": 88568
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.8875494599342346,
      "learning_rate": 0.00040656326610536304,
      "loss": 2.1904,
      "step": 88569
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.8550007343292236,
      "learning_rate": 0.00040655944229441916,
      "loss": 2.2214,
      "step": 88570
    },
    {
      "epoch": 1.15,
      "grad_norm": 1.1374772787094116,
      "learning_rate": 0.00040655561846366387,
      "loss": 2.1558,
      "step": 88571
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.8479214310646057,
      "learning_rate": 0.00040655179461309793,
      "loss": 2.1711,
      "step": 88572
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.7818789482116699,
      "learning_rate": 0.00040654797074272193,
      "loss": 2.2347,
      "step": 88573
    },
    {
      "epoch": 1.15,
      "grad_norm": 1.2009224891662598,
      "learning_rate": 0.0004065441468525367,
      "loss": 2.206,
      "step": 88574
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.8815385103225708,
      "learning_rate": 0.0004065403229425429,
      "loss": 2.0322,
      "step": 88575
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.8812114596366882,
      "learning_rate": 0.00040653649901274127,
      "loss": 2.1546,
      "step": 88576
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.9003816843032837,
      "learning_rate": 0.00040653267506313243,
      "loss": 2.0483,
      "step": 88577
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.9157686233520508,
      "learning_rate": 0.0004065288510937172,
      "loss": 2.1373,
      "step": 88578
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.8839501738548279,
      "learning_rate": 0.00040652502710449625,
      "loss": 2.2212,
      "step": 88579
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.8205675482749939,
      "learning_rate": 0.00040652120309547026,
      "loss": 2.0585,
      "step": 88580
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.8911875486373901,
      "learning_rate": 0.00040651737906663996,
      "loss": 1.9972,
      "step": 88581
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.8392283916473389,
      "learning_rate": 0.0004065135550180061,
      "loss": 2.195,
      "step": 88582
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.8310931324958801,
      "learning_rate": 0.00040650973094956937,
      "loss": 2.42,
      "step": 88583
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.8715099692344666,
      "learning_rate": 0.0004065059068613304,
      "loss": 2.3591,
      "step": 88584
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.8466904163360596,
      "learning_rate": 0.00040650208275329005,
      "loss": 2.3418,
      "step": 88585
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.7729007005691528,
      "learning_rate": 0.0004064982586254489,
      "loss": 2.1055,
      "step": 88586
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.9395918846130371,
      "learning_rate": 0.0004064944344778077,
      "loss": 2.0678,
      "step": 88587
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.825326681137085,
      "learning_rate": 0.00040649061031036726,
      "loss": 2.3073,
      "step": 88588
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.7784081697463989,
      "learning_rate": 0.00040648678612312814,
      "loss": 2.2567,
      "step": 88589
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.8795536756515503,
      "learning_rate": 0.0004064829619160911,
      "loss": 2.4105,
      "step": 88590
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.8222148418426514,
      "learning_rate": 0.0004064791376892569,
      "loss": 2.2297,
      "step": 88591
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.896300733089447,
      "learning_rate": 0.0004064753134426262,
      "loss": 2.1837,
      "step": 88592
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.8513239026069641,
      "learning_rate": 0.0004064714891761997,
      "loss": 2.0731,
      "step": 88593
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.8139917254447937,
      "learning_rate": 0.0004064676648899783,
      "loss": 2.1622,
      "step": 88594
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.7826764583587646,
      "learning_rate": 0.00040646384058396244,
      "loss": 2.1581,
      "step": 88595
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.8688595294952393,
      "learning_rate": 0.00040646001625815285,
      "loss": 2.0518,
      "step": 88596
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.8769255876541138,
      "learning_rate": 0.00040645619191255044,
      "loss": 2.1813,
      "step": 88597
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.9083730578422546,
      "learning_rate": 0.0004064523675471558,
      "loss": 2.1912,
      "step": 88598
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.8204594254493713,
      "learning_rate": 0.0004064485431619696,
      "loss": 2.1361,
      "step": 88599
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.83939528465271,
      "learning_rate": 0.0004064447187569927,
      "loss": 2.4739,
      "step": 88600
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.8798115253448486,
      "learning_rate": 0.0004064408943322257,
      "loss": 2.3168,
      "step": 88601
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.7948063015937805,
      "learning_rate": 0.00040643706988766927,
      "loss": 2.1641,
      "step": 88602
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.8325167894363403,
      "learning_rate": 0.00040643324542332423,
      "loss": 1.9829,
      "step": 88603
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.8288729786872864,
      "learning_rate": 0.00040642942093919125,
      "loss": 2.2208,
      "step": 88604
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.9146122336387634,
      "learning_rate": 0.000406425596435271,
      "loss": 1.8436,
      "step": 88605
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.8105718493461609,
      "learning_rate": 0.00040642177191156425,
      "loss": 2.043,
      "step": 88606
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.9279612302780151,
      "learning_rate": 0.0004064179473680716,
      "loss": 2.2844,
      "step": 88607
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.7237548828125,
      "learning_rate": 0.00040641412280479386,
      "loss": 2.0685,
      "step": 88608
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.8604738116264343,
      "learning_rate": 0.0004064102982217319,
      "loss": 2.3134,
      "step": 88609
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.7640409469604492,
      "learning_rate": 0.0004064064736188861,
      "loss": 1.9529,
      "step": 88610
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.8760943412780762,
      "learning_rate": 0.00040640264899625733,
      "loss": 2.4881,
      "step": 88611
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.8758600354194641,
      "learning_rate": 0.00040639882435384634,
      "loss": 2.2531,
      "step": 88612
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.8087360262870789,
      "learning_rate": 0.0004063949996916538,
      "loss": 2.1765,
      "step": 88613
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.8277984261512756,
      "learning_rate": 0.00040639117500968034,
      "loss": 2.1614,
      "step": 88614
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.8349357843399048,
      "learning_rate": 0.0004063873503079269,
      "loss": 2.1258,
      "step": 88615
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.7574730515480042,
      "learning_rate": 0.0004063835255863939,
      "loss": 2.0847,
      "step": 88616
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.9252683520317078,
      "learning_rate": 0.0004063797008450823,
      "loss": 2.273,
      "step": 88617
    },
    {
      "epoch": 1.15,
      "grad_norm": 1.0165659189224243,
      "learning_rate": 0.0004063758760839926,
      "loss": 2.0939,
      "step": 88618
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.7926923632621765,
      "learning_rate": 0.00040637205130312576,
      "loss": 2.4384,
      "step": 88619
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.7486305832862854,
      "learning_rate": 0.0004063682265024822,
      "loss": 1.9136,
      "step": 88620
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.7591419219970703,
      "learning_rate": 0.0004063644016820628,
      "loss": 2.2447,
      "step": 88621
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.8116942048072815,
      "learning_rate": 0.00040636057684186836,
      "loss": 2.3516,
      "step": 88622
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.7971315979957581,
      "learning_rate": 0.00040635675198189935,
      "loss": 1.929,
      "step": 88623
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.7922803163528442,
      "learning_rate": 0.0004063529271021567,
      "loss": 2.3113,
      "step": 88624
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.749620258808136,
      "learning_rate": 0.00040634910220264107,
      "loss": 2.23,
      "step": 88625
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.9114651083946228,
      "learning_rate": 0.00040634527728335303,
      "loss": 2.1123,
      "step": 88626
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.8736875653266907,
      "learning_rate": 0.0004063414523442935,
      "loss": 2.2396,
      "step": 88627
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.8867101669311523,
      "learning_rate": 0.0004063376273854629,
      "loss": 2.1196,
      "step": 88628
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.8763796091079712,
      "learning_rate": 0.00040633380240686235,
      "loss": 2.1967,
      "step": 88629
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.8692596554756165,
      "learning_rate": 0.0004063299774084922,
      "loss": 2.3939,
      "step": 88630
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.7848944664001465,
      "learning_rate": 0.00040632615239035336,
      "loss": 2.1817,
      "step": 88631
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.8194066286087036,
      "learning_rate": 0.0004063223273524464,
      "loss": 2.2904,
      "step": 88632
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.9379748702049255,
      "learning_rate": 0.00040631850229477226,
      "loss": 2.0939,
      "step": 88633
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.8492854237556458,
      "learning_rate": 0.0004063146772173313,
      "loss": 2.2771,
      "step": 88634
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.9120239615440369,
      "learning_rate": 0.00040631085212012463,
      "loss": 2.165,
      "step": 88635
    },
    {
      "epoch": 1.15,
      "grad_norm": 1.0811277627944946,
      "learning_rate": 0.00040630702700315276,
      "loss": 2.2389,
      "step": 88636
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.8643463850021362,
      "learning_rate": 0.00040630320186641627,
      "loss": 2.0615,
      "step": 88637
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.7744169235229492,
      "learning_rate": 0.0004062993767099161,
      "loss": 2.2936,
      "step": 88638
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.8289061188697815,
      "learning_rate": 0.0004062955515336528,
      "loss": 2.1174,
      "step": 88639
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.7995103597640991,
      "learning_rate": 0.00040629172633762723,
      "loss": 2.0567,
      "step": 88640
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.9378554821014404,
      "learning_rate": 0.00040628790112184,
      "loss": 2.0482,
      "step": 88641
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.8219877481460571,
      "learning_rate": 0.00040628407588629185,
      "loss": 2.283,
      "step": 88642
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.8693851232528687,
      "learning_rate": 0.00040628025063098344,
      "loss": 2.1915,
      "step": 88643
    },
    {
      "epoch": 1.15,
      "grad_norm": 1.0755537748336792,
      "learning_rate": 0.00040627642535591554,
      "loss": 2.3024,
      "step": 88644
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.8514158129692078,
      "learning_rate": 0.0004062726000610889,
      "loss": 2.3909,
      "step": 88645
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.8472583293914795,
      "learning_rate": 0.00040626877474650414,
      "loss": 2.3773,
      "step": 88646
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.9944241046905518,
      "learning_rate": 0.000406264949412162,
      "loss": 2.2085,
      "step": 88647
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.8740015029907227,
      "learning_rate": 0.00040626112405806313,
      "loss": 2.1858,
      "step": 88648
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.8992060422897339,
      "learning_rate": 0.0004062572986842085,
      "loss": 2.1939,
      "step": 88649
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.9017229080200195,
      "learning_rate": 0.0004062534732905985,
      "loss": 2.263,
      "step": 88650
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.8577932715415955,
      "learning_rate": 0.0004062496478772339,
      "loss": 2.1936,
      "step": 88651
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.8613621592521667,
      "learning_rate": 0.00040624582244411564,
      "loss": 2.3069,
      "step": 88652
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.8740943074226379,
      "learning_rate": 0.00040624199699124423,
      "loss": 2.3551,
      "step": 88653
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.7932326793670654,
      "learning_rate": 0.00040623817151862033,
      "loss": 2.1898,
      "step": 88654
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.7572082877159119,
      "learning_rate": 0.0004062343460262449,
      "loss": 2.2072,
      "step": 88655
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.8996686935424805,
      "learning_rate": 0.0004062305205141184,
      "loss": 2.2256,
      "step": 88656
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.9604223966598511,
      "learning_rate": 0.00040622669498224166,
      "loss": 2.431,
      "step": 88657
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.7991150617599487,
      "learning_rate": 0.00040622286943061537,
      "loss": 2.2662,
      "step": 88658
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.7832449078559875,
      "learning_rate": 0.0004062190438592403,
      "loss": 2.2264,
      "step": 88659
    },
    {
      "epoch": 1.15,
      "grad_norm": 1.241889238357544,
      "learning_rate": 0.000406215218268117,
      "loss": 2.2372,
      "step": 88660
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.841758131980896,
      "learning_rate": 0.00040621139265724636,
      "loss": 2.2309,
      "step": 88661
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.8976859450340271,
      "learning_rate": 0.000406207567026629,
      "loss": 2.1904,
      "step": 88662
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.9245572686195374,
      "learning_rate": 0.00040620374137626566,
      "loss": 2.3721,
      "step": 88663
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.813865065574646,
      "learning_rate": 0.000406199915706157,
      "loss": 2.1803,
      "step": 88664
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.7957454919815063,
      "learning_rate": 0.0004061960900163038,
      "loss": 2.108,
      "step": 88665
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.8844573497772217,
      "learning_rate": 0.0004061922643067067,
      "loss": 2.0336,
      "step": 88666
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.8206036686897278,
      "learning_rate": 0.0004061884385773666,
      "loss": 2.247,
      "step": 88667
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.8952849507331848,
      "learning_rate": 0.0004061846128282839,
      "loss": 2.1968,
      "step": 88668
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.7903090715408325,
      "learning_rate": 0.0004061807870594596,
      "loss": 2.3988,
      "step": 88669
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.8757895827293396,
      "learning_rate": 0.0004061769612708943,
      "loss": 2.2222,
      "step": 88670
    },
    {
      "epoch": 1.15,
      "grad_norm": 1.4048658609390259,
      "learning_rate": 0.00040617313546258854,
      "loss": 2.2845,
      "step": 88671
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.8345737457275391,
      "learning_rate": 0.00040616930963454327,
      "loss": 2.1618,
      "step": 88672
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.961035430431366,
      "learning_rate": 0.00040616548378675913,
      "loss": 2.0793,
      "step": 88673
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.8947079181671143,
      "learning_rate": 0.0004061616579192368,
      "loss": 2.3284,
      "step": 88674
    },
    {
      "epoch": 1.15,
      "grad_norm": 1.0149801969528198,
      "learning_rate": 0.00040615783203197704,
      "loss": 2.1549,
      "step": 88675
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.8457106947898865,
      "learning_rate": 0.00040615400612498056,
      "loss": 2.0685,
      "step": 88676
    },
    {
      "epoch": 1.15,
      "grad_norm": 1.0472018718719482,
      "learning_rate": 0.000406150180198248,
      "loss": 2.0322,
      "step": 88677
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.9768750667572021,
      "learning_rate": 0.00040614635425178006,
      "loss": 2.2201,
      "step": 88678
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.823567271232605,
      "learning_rate": 0.0004061425282855777,
      "loss": 2.2765,
      "step": 88679
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.8430930972099304,
      "learning_rate": 0.00040613870229964125,
      "loss": 2.1301,
      "step": 88680
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.8513092398643494,
      "learning_rate": 0.0004061348762939717,
      "loss": 2.2326,
      "step": 88681
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.8054596781730652,
      "learning_rate": 0.00040613105026856964,
      "loss": 2.1936,
      "step": 88682
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.853780210018158,
      "learning_rate": 0.0004061272242234359,
      "loss": 2.1955,
      "step": 88683
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.8525252342224121,
      "learning_rate": 0.00040612339815857093,
      "loss": 2.1816,
      "step": 88684
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.9074857831001282,
      "learning_rate": 0.0004061195720739758,
      "loss": 2.1758,
      "step": 88685
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.7398910522460938,
      "learning_rate": 0.00040611574596965093,
      "loss": 2.1736,
      "step": 88686
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.9090545773506165,
      "learning_rate": 0.00040611191984559717,
      "loss": 2.2535,
      "step": 88687
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.850447416305542,
      "learning_rate": 0.0004061080937018152,
      "loss": 2.4471,
      "step": 88688
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.9990489482879639,
      "learning_rate": 0.00040610426753830576,
      "loss": 2.1729,
      "step": 88689
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.8307639360427856,
      "learning_rate": 0.0004061004413550694,
      "loss": 2.1792,
      "step": 88690
    },
    {
      "epoch": 1.15,
      "grad_norm": 1.0488989353179932,
      "learning_rate": 0.00040609661515210715,
      "loss": 2.1537,
      "step": 88691
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.7500322461128235,
      "learning_rate": 0.0004060927889294195,
      "loss": 2.443,
      "step": 88692
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.7903777956962585,
      "learning_rate": 0.0004060889626870071,
      "loss": 2.3234,
      "step": 88693
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.8255867958068848,
      "learning_rate": 0.0004060851364248709,
      "loss": 2.1858,
      "step": 88694
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.8607100248336792,
      "learning_rate": 0.00040608131014301135,
      "loss": 2.326,
      "step": 88695
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.9128870964050293,
      "learning_rate": 0.0004060774838414293,
      "loss": 2.263,
      "step": 88696
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.8461684584617615,
      "learning_rate": 0.0004060736575201255,
      "loss": 2.231,
      "step": 88697
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.7749360203742981,
      "learning_rate": 0.00040606983117910054,
      "loss": 2.0936,
      "step": 88698
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.9520663619041443,
      "learning_rate": 0.00040606600481835525,
      "loss": 2.3422,
      "step": 88699
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.9173080921173096,
      "learning_rate": 0.0004060621784378902,
      "loss": 2.08,
      "step": 88700
    },
    {
      "epoch": 1.15,
      "grad_norm": 1.1466189622879028,
      "learning_rate": 0.0004060583520377063,
      "loss": 2.1791,
      "step": 88701
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.8021843433380127,
      "learning_rate": 0.00040605452561780404,
      "loss": 2.158,
      "step": 88702
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.9869697690010071,
      "learning_rate": 0.00040605069917818435,
      "loss": 2.0675,
      "step": 88703
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.8974420428276062,
      "learning_rate": 0.0004060468727188478,
      "loss": 2.1329,
      "step": 88704
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.8479375839233398,
      "learning_rate": 0.00040604304623979506,
      "loss": 2.1148,
      "step": 88705
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.8130048513412476,
      "learning_rate": 0.00040603921974102704,
      "loss": 2.0357,
      "step": 88706
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.9525576829910278,
      "learning_rate": 0.0004060353932225442,
      "loss": 2.2305,
      "step": 88707
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.9462994337081909,
      "learning_rate": 0.00040603156668434745,
      "loss": 2.0775,
      "step": 88708
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.9272271394729614,
      "learning_rate": 0.0004060277401264374,
      "loss": 2.0746,
      "step": 88709
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.8906847834587097,
      "learning_rate": 0.00040602391354881487,
      "loss": 2.2339,
      "step": 88710
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.7523322105407715,
      "learning_rate": 0.0004060200869514804,
      "loss": 1.8637,
      "step": 88711
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.8441782593727112,
      "learning_rate": 0.0004060162603344348,
      "loss": 2.0955,
      "step": 88712
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.8062295317649841,
      "learning_rate": 0.0004060124336976788,
      "loss": 2.206,
      "step": 88713
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.9407396912574768,
      "learning_rate": 0.00040600860704121316,
      "loss": 2.1533,
      "step": 88714
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.7621864080429077,
      "learning_rate": 0.00040600478036503844,
      "loss": 2.2175,
      "step": 88715
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.8260363936424255,
      "learning_rate": 0.00040600095366915537,
      "loss": 1.9495,
      "step": 88716
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.8598691821098328,
      "learning_rate": 0.00040599712695356486,
      "loss": 2.3166,
      "step": 88717
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.8254950642585754,
      "learning_rate": 0.0004059933002182674,
      "loss": 2.2278,
      "step": 88718
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.9293105602264404,
      "learning_rate": 0.0004059894734632638,
      "loss": 2.142,
      "step": 88719
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.9224562644958496,
      "learning_rate": 0.00040598564668855474,
      "loss": 2.0842,
      "step": 88720
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.8436856865882874,
      "learning_rate": 0.000405981819894141,
      "loss": 2.2622,
      "step": 88721
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.8977502584457397,
      "learning_rate": 0.00040597799308002324,
      "loss": 2.2594,
      "step": 88722
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.822446346282959,
      "learning_rate": 0.0004059741662462021,
      "loss": 2.0425,
      "step": 88723
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.8694379329681396,
      "learning_rate": 0.00040597033939267837,
      "loss": 2.2172,
      "step": 88724
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.9413506388664246,
      "learning_rate": 0.0004059665125194528,
      "loss": 2.1357,
      "step": 88725
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.8096489906311035,
      "learning_rate": 0.00040596268562652604,
      "loss": 2.3168,
      "step": 88726
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.9034485220909119,
      "learning_rate": 0.0004059588587138988,
      "loss": 2.1121,
      "step": 88727
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.8881727457046509,
      "learning_rate": 0.0004059550317815719,
      "loss": 2.297,
      "step": 88728
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.8132075071334839,
      "learning_rate": 0.00040595120482954584,
      "loss": 2.1933,
      "step": 88729
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.8187270760536194,
      "learning_rate": 0.00040594737785782146,
      "loss": 2.1468,
      "step": 88730
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.9136732816696167,
      "learning_rate": 0.00040594355086639956,
      "loss": 2.0344,
      "step": 88731
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.8182518482208252,
      "learning_rate": 0.00040593972385528064,
      "loss": 2.2082,
      "step": 88732
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.7774529457092285,
      "learning_rate": 0.00040593589682446556,
      "loss": 2.3633,
      "step": 88733
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.9446491599082947,
      "learning_rate": 0.0004059320697739551,
      "loss": 2.1075,
      "step": 88734
    },
    {
      "epoch": 1.16,
      "grad_norm": 1.0141944885253906,
      "learning_rate": 0.0004059282427037497,
      "loss": 2.1485,
      "step": 88735
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.8055180907249451,
      "learning_rate": 0.0004059244156138503,
      "loss": 2.3587,
      "step": 88736
    },
    {
      "epoch": 1.16,
      "grad_norm": 1.2337779998779297,
      "learning_rate": 0.00040592058850425763,
      "loss": 2.2226,
      "step": 88737
    },
    {
      "epoch": 1.16,
      "grad_norm": 1.1316864490509033,
      "learning_rate": 0.0004059167613749723,
      "loss": 2.1993,
      "step": 88738
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.8387598395347595,
      "learning_rate": 0.00040591293422599493,
      "loss": 1.9541,
      "step": 88739
    },
    {
      "epoch": 1.16,
      "grad_norm": 1.1682313680648804,
      "learning_rate": 0.00040590910705732647,
      "loss": 1.997,
      "step": 88740
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.9029697775840759,
      "learning_rate": 0.00040590527986896744,
      "loss": 2.1119,
      "step": 88741
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.9129382967948914,
      "learning_rate": 0.00040590145266091864,
      "loss": 2.2042,
      "step": 88742
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.8125735521316528,
      "learning_rate": 0.0004058976254331808,
      "loss": 2.262,
      "step": 88743
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.916926920413971,
      "learning_rate": 0.0004058937981857545,
      "loss": 2.2239,
      "step": 88744
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.8429601788520813,
      "learning_rate": 0.0004058899709186406,
      "loss": 2.1662,
      "step": 88745
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.916385293006897,
      "learning_rate": 0.0004058861436318398,
      "loss": 2.3303,
      "step": 88746
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.8910529613494873,
      "learning_rate": 0.0004058823163253527,
      "loss": 2.2021,
      "step": 88747
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.9121508002281189,
      "learning_rate": 0.0004058784889991801,
      "loss": 2.1842,
      "step": 88748
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.7642521262168884,
      "learning_rate": 0.0004058746616533228,
      "loss": 2.1742,
      "step": 88749
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.9384551048278809,
      "learning_rate": 0.00040587083428778124,
      "loss": 2.0014,
      "step": 88750
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.8318910002708435,
      "learning_rate": 0.00040586700690255635,
      "loss": 2.2786,
      "step": 88751
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.8101125359535217,
      "learning_rate": 0.0004058631794976488,
      "loss": 2.4913,
      "step": 88752
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.8459411263465881,
      "learning_rate": 0.00040585935207305924,
      "loss": 2.1458,
      "step": 88753
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.8492796421051025,
      "learning_rate": 0.00040585552462878843,
      "loss": 2.2756,
      "step": 88754
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.8421133756637573,
      "learning_rate": 0.0004058516971648372,
      "loss": 2.2249,
      "step": 88755
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.7936141490936279,
      "learning_rate": 0.000405847869681206,
      "loss": 2.199,
      "step": 88756
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.8492374420166016,
      "learning_rate": 0.0004058440421778957,
      "loss": 2.1451,
      "step": 88757
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.7886973023414612,
      "learning_rate": 0.0004058402146549071,
      "loss": 2.2487,
      "step": 88758
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.8932790160179138,
      "learning_rate": 0.00040583638711224067,
      "loss": 2.3713,
      "step": 88759
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.9233687520027161,
      "learning_rate": 0.0004058325595498973,
      "loss": 2.0965,
      "step": 88760
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.9017630815505981,
      "learning_rate": 0.00040582873196787775,
      "loss": 2.2924,
      "step": 88761
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.7738617658615112,
      "learning_rate": 0.0004058249043661825,
      "loss": 2.5408,
      "step": 88762
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.794989287853241,
      "learning_rate": 0.0004058210767448125,
      "loss": 2.1689,
      "step": 88763
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.8147201538085938,
      "learning_rate": 0.00040581724910376837,
      "loss": 2.1431,
      "step": 88764
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.8344363570213318,
      "learning_rate": 0.0004058134214430508,
      "loss": 2.3617,
      "step": 88765
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.8118566870689392,
      "learning_rate": 0.0004058095937626605,
      "loss": 2.1027,
      "step": 88766
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.8622261881828308,
      "learning_rate": 0.0004058057660625982,
      "loss": 2.0547,
      "step": 88767
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.9410313367843628,
      "learning_rate": 0.0004058019383428646,
      "loss": 2.175,
      "step": 88768
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.826493501663208,
      "learning_rate": 0.00040579811060346044,
      "loss": 2.025,
      "step": 88769
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.8583333492279053,
      "learning_rate": 0.00040579428284438643,
      "loss": 2.1184,
      "step": 88770
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.9756600856781006,
      "learning_rate": 0.0004057904550656432,
      "loss": 2.0858,
      "step": 88771
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.849176287651062,
      "learning_rate": 0.0004057866272672316,
      "loss": 2.4257,
      "step": 88772
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.8165552616119385,
      "learning_rate": 0.00040578279944915226,
      "loss": 2.306,
      "step": 88773
    },
    {
      "epoch": 1.16,
      "grad_norm": 1.148023247718811,
      "learning_rate": 0.0004057789716114059,
      "loss": 2.285,
      "step": 88774
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.850878894329071,
      "learning_rate": 0.0004057751437539931,
      "loss": 2.2925,
      "step": 88775
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.7947456240653992,
      "learning_rate": 0.0004057713158769149,
      "loss": 2.2305,
      "step": 88776
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.9279507398605347,
      "learning_rate": 0.0004057674879801717,
      "loss": 2.0135,
      "step": 88777
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.9364233613014221,
      "learning_rate": 0.0004057636600637644,
      "loss": 2.1934,
      "step": 88778
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.8574361205101013,
      "learning_rate": 0.0004057598321276936,
      "loss": 2.1464,
      "step": 88779
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.7944440245628357,
      "learning_rate": 0.00040575600417196004,
      "loss": 2.0132,
      "step": 88780
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.8468354344367981,
      "learning_rate": 0.00040575217619656445,
      "loss": 2.0386,
      "step": 88781
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.8215506672859192,
      "learning_rate": 0.0004057483482015075,
      "loss": 2.0535,
      "step": 88782
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.8947082757949829,
      "learning_rate": 0.00040574452018679006,
      "loss": 1.9547,
      "step": 88783
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.9110977053642273,
      "learning_rate": 0.00040574069215241256,
      "loss": 2.1449,
      "step": 88784
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.7961770296096802,
      "learning_rate": 0.00040573686409837593,
      "loss": 2.2903,
      "step": 88785
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.8016935586929321,
      "learning_rate": 0.0004057330360246809,
      "loss": 2.2012,
      "step": 88786
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.8574723601341248,
      "learning_rate": 0.000405729207931328,
      "loss": 1.9457,
      "step": 88787
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.8512651920318604,
      "learning_rate": 0.0004057253798183181,
      "loss": 2.441,
      "step": 88788
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.8444517254829407,
      "learning_rate": 0.0004057215516856518,
      "loss": 2.1865,
      "step": 88789
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.8514799475669861,
      "learning_rate": 0.0004057177235333299,
      "loss": 2.1945,
      "step": 88790
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.8623972535133362,
      "learning_rate": 0.0004057138953613531,
      "loss": 2.2688,
      "step": 88791
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.893100917339325,
      "learning_rate": 0.0004057100671697221,
      "loss": 2.1422,
      "step": 88792
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.977504551410675,
      "learning_rate": 0.00040570623895843755,
      "loss": 2.1081,
      "step": 88793
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.9965136647224426,
      "learning_rate": 0.0004057024107275002,
      "loss": 2.1245,
      "step": 88794
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.832187831401825,
      "learning_rate": 0.00040569858247691085,
      "loss": 2.1562,
      "step": 88795
    },
    {
      "epoch": 1.16,
      "grad_norm": 1.1750965118408203,
      "learning_rate": 0.0004056947542066701,
      "loss": 2.2821,
      "step": 88796
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.8351731300354004,
      "learning_rate": 0.0004056909259167787,
      "loss": 1.9994,
      "step": 88797
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.8514536619186401,
      "learning_rate": 0.00040568709760723746,
      "loss": 2.2275,
      "step": 88798
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.9480715394020081,
      "learning_rate": 0.00040568326927804685,
      "loss": 2.1068,
      "step": 88799
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.8205493688583374,
      "learning_rate": 0.00040567944092920774,
      "loss": 2.3248,
      "step": 88800
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.867638111114502,
      "learning_rate": 0.0004056756125607208,
      "loss": 2.2586,
      "step": 88801
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.8781270384788513,
      "learning_rate": 0.0004056717841725869,
      "loss": 2.315,
      "step": 88802
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.8834848403930664,
      "learning_rate": 0.0004056679557648066,
      "loss": 2.1487,
      "step": 88803
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.8321994543075562,
      "learning_rate": 0.0004056641273373806,
      "loss": 2.1605,
      "step": 88804
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.787979781627655,
      "learning_rate": 0.00040566029889030955,
      "loss": 2.1412,
      "step": 88805
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.859332799911499,
      "learning_rate": 0.00040565647042359435,
      "loss": 1.9657,
      "step": 88806
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.7769086360931396,
      "learning_rate": 0.0004056526419372356,
      "loss": 2.1663,
      "step": 88807
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.817216157913208,
      "learning_rate": 0.0004056488134312341,
      "loss": 2.2054,
      "step": 88808
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.8723353743553162,
      "learning_rate": 0.00040564498490559044,
      "loss": 2.2755,
      "step": 88809
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.8671316504478455,
      "learning_rate": 0.0004056411563603053,
      "loss": 1.9634,
      "step": 88810
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.8837080597877502,
      "learning_rate": 0.0004056373277953795,
      "loss": 2.3813,
      "step": 88811
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.8648873567581177,
      "learning_rate": 0.00040563349921081377,
      "loss": 2.3692,
      "step": 88812
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.8025013208389282,
      "learning_rate": 0.00040562967060660883,
      "loss": 2.0799,
      "step": 88813
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.7864918112754822,
      "learning_rate": 0.00040562584198276525,
      "loss": 2.1806,
      "step": 88814
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.8602262735366821,
      "learning_rate": 0.0004056220133392839,
      "loss": 2.139,
      "step": 88815
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.8742038011550903,
      "learning_rate": 0.0004056181846761654,
      "loss": 2.1925,
      "step": 88816
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.8257430195808411,
      "learning_rate": 0.0004056143559934105,
      "loss": 2.2414,
      "step": 88817
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.7759371995925903,
      "learning_rate": 0.00040561052729101986,
      "loss": 2.286,
      "step": 88818
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.9506016969680786,
      "learning_rate": 0.0004056066985689943,
      "loss": 2.1762,
      "step": 88819
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.7832993865013123,
      "learning_rate": 0.00040560286982733437,
      "loss": 1.9069,
      "step": 88820
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.825508177280426,
      "learning_rate": 0.0004055990410660409,
      "loss": 2.0635,
      "step": 88821
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.8449559807777405,
      "learning_rate": 0.0004055952122851146,
      "loss": 2.0276,
      "step": 88822
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.9117333889007568,
      "learning_rate": 0.00040559138348455614,
      "loss": 2.2756,
      "step": 88823
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.8254528641700745,
      "learning_rate": 0.0004055875546643662,
      "loss": 2.2836,
      "step": 88824
    },
    {
      "epoch": 1.16,
      "grad_norm": 1.1550347805023193,
      "learning_rate": 0.0004055837258245457,
      "loss": 2.2013,
      "step": 88825
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.7782496809959412,
      "learning_rate": 0.00040557989696509507,
      "loss": 2.2825,
      "step": 88826
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.8902334570884705,
      "learning_rate": 0.0004055760680860151,
      "loss": 2.2843,
      "step": 88827
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.7707943916320801,
      "learning_rate": 0.0004055722391873066,
      "loss": 2.0124,
      "step": 88828
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.7869619131088257,
      "learning_rate": 0.00040556841026897025,
      "loss": 1.9626,
      "step": 88829
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.8614915013313293,
      "learning_rate": 0.00040556458133100674,
      "loss": 2.0551,
      "step": 88830
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.7926218509674072,
      "learning_rate": 0.0004055607523734168,
      "loss": 2.2131,
      "step": 88831
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.8913610577583313,
      "learning_rate": 0.0004055569233962011,
      "loss": 2.3514,
      "step": 88832
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.7896665930747986,
      "learning_rate": 0.0004055530943993603,
      "loss": 1.8951,
      "step": 88833
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.8101001977920532,
      "learning_rate": 0.0004055492653828953,
      "loss": 2.0913,
      "step": 88834
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.7864171862602234,
      "learning_rate": 0.00040554543634680666,
      "loss": 2.1043,
      "step": 88835
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.823208212852478,
      "learning_rate": 0.0004055416072910951,
      "loss": 1.9131,
      "step": 88836
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.8497849106788635,
      "learning_rate": 0.0004055377782157614,
      "loss": 2.3233,
      "step": 88837
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.8173945546150208,
      "learning_rate": 0.0004055339491208062,
      "loss": 2.1024,
      "step": 88838
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.7714566588401794,
      "learning_rate": 0.00040553012000623024,
      "loss": 2.1743,
      "step": 88839
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.8455507755279541,
      "learning_rate": 0.0004055262908720343,
      "loss": 2.1873,
      "step": 88840
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.86316978931427,
      "learning_rate": 0.000405522461718219,
      "loss": 2.1242,
      "step": 88841
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.8169863224029541,
      "learning_rate": 0.000405518632544785,
      "loss": 2.1573,
      "step": 88842
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.9155915379524231,
      "learning_rate": 0.0004055148033517333,
      "loss": 2.3364,
      "step": 88843
    },
    {
      "epoch": 1.16,
      "grad_norm": 1.1582475900650024,
      "learning_rate": 0.0004055109741390642,
      "loss": 2.3523,
      "step": 88844
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.8530603051185608,
      "learning_rate": 0.0004055071449067787,
      "loss": 2.2957,
      "step": 88845
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.8329893946647644,
      "learning_rate": 0.0004055033156548774,
      "loss": 2.3917,
      "step": 88846
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.9153298139572144,
      "learning_rate": 0.00040549948638336113,
      "loss": 2.0888,
      "step": 88847
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.7732503414154053,
      "learning_rate": 0.0004054956570922304,
      "loss": 2.4211,
      "step": 88848
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.8038892149925232,
      "learning_rate": 0.0004054918277814861,
      "loss": 2.1056,
      "step": 88849
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.9065954685211182,
      "learning_rate": 0.0004054879984511289,
      "loss": 2.1627,
      "step": 88850
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.8661022782325745,
      "learning_rate": 0.00040548416910115946,
      "loss": 2.4053,
      "step": 88851
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.7997717261314392,
      "learning_rate": 0.0004054803397315785,
      "loss": 2.2475,
      "step": 88852
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.9911451935768127,
      "learning_rate": 0.0004054765103423868,
      "loss": 2.0915,
      "step": 88853
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.7566376328468323,
      "learning_rate": 0.000405472680933585,
      "loss": 2.2356,
      "step": 88854
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.8557209968566895,
      "learning_rate": 0.00040546885150517375,
      "loss": 2.3706,
      "step": 88855
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.8349624276161194,
      "learning_rate": 0.000405465022057154,
      "loss": 2.3085,
      "step": 88856
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.7907875180244446,
      "learning_rate": 0.0004054611925895263,
      "loss": 2.2191,
      "step": 88857
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.8063700795173645,
      "learning_rate": 0.0004054573631022912,
      "loss": 2.2053,
      "step": 88858
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.7727944254875183,
      "learning_rate": 0.0004054535335954498,
      "loss": 2.2196,
      "step": 88859
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.8399001359939575,
      "learning_rate": 0.00040544970406900243,
      "loss": 2.3824,
      "step": 88860
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.7888382077217102,
      "learning_rate": 0.00040544587452295003,
      "loss": 2.0987,
      "step": 88861
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.8342198133468628,
      "learning_rate": 0.0004054420449572933,
      "loss": 2.3486,
      "step": 88862
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.8160564303398132,
      "learning_rate": 0.00040543821537203286,
      "loss": 2.3841,
      "step": 88863
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.7986912131309509,
      "learning_rate": 0.00040543438576716944,
      "loss": 2.2811,
      "step": 88864
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.7984312176704407,
      "learning_rate": 0.0004054305561427039,
      "loss": 2.1049,
      "step": 88865
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.8358063101768494,
      "learning_rate": 0.0004054267264986367,
      "loss": 2.1254,
      "step": 88866
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.9028821587562561,
      "learning_rate": 0.0004054228968349687,
      "loss": 2.2259,
      "step": 88867
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.7301999926567078,
      "learning_rate": 0.00040541906715170067,
      "loss": 2.2078,
      "step": 88868
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.8880369067192078,
      "learning_rate": 0.00040541523744883317,
      "loss": 2.3394,
      "step": 88869
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.8349841833114624,
      "learning_rate": 0.00040541140772636694,
      "loss": 2.0584,
      "step": 88870
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.8355255722999573,
      "learning_rate": 0.00040540757798430293,
      "loss": 2.0148,
      "step": 88871
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.8091684579849243,
      "learning_rate": 0.0004054037482226415,
      "loss": 2.0953,
      "step": 88872
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.8310284614562988,
      "learning_rate": 0.0004053999184413836,
      "loss": 2.0986,
      "step": 88873
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.8997249603271484,
      "learning_rate": 0.0004053960886405298,
      "loss": 2.1234,
      "step": 88874
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.7929096817970276,
      "learning_rate": 0.00040539225882008084,
      "loss": 1.9621,
      "step": 88875
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.7975203394889832,
      "learning_rate": 0.00040538842898003754,
      "loss": 2.1642,
      "step": 88876
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.7732520699501038,
      "learning_rate": 0.0004053845991204006,
      "loss": 2.0257,
      "step": 88877
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.8032568097114563,
      "learning_rate": 0.0004053807692411706,
      "loss": 2.0677,
      "step": 88878
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.8274953961372375,
      "learning_rate": 0.0004053769393423483,
      "loss": 2.1502,
      "step": 88879
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.905326783657074,
      "learning_rate": 0.00040537310942393454,
      "loss": 2.1549,
      "step": 88880
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.8116399645805359,
      "learning_rate": 0.0004053692794859299,
      "loss": 1.9826,
      "step": 88881
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.8441259264945984,
      "learning_rate": 0.000405365449528335,
      "loss": 2.1325,
      "step": 88882
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.8176878690719604,
      "learning_rate": 0.0004053616195511508,
      "loss": 1.9386,
      "step": 88883
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.9167526364326477,
      "learning_rate": 0.0004053577895543778,
      "loss": 2.3862,
      "step": 88884
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.8284594416618347,
      "learning_rate": 0.00040535395953801687,
      "loss": 1.995,
      "step": 88885
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.8928124904632568,
      "learning_rate": 0.00040535012950206864,
      "loss": 2.1412,
      "step": 88886
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.8379560112953186,
      "learning_rate": 0.0004053462994465338,
      "loss": 2.1983,
      "step": 88887
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.7943046689033508,
      "learning_rate": 0.0004053424693714131,
      "loss": 1.9498,
      "step": 88888
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.8962357640266418,
      "learning_rate": 0.00040533863927670724,
      "loss": 2.1975,
      "step": 88889
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.8216618895530701,
      "learning_rate": 0.000405334809162417,
      "loss": 2.1637,
      "step": 88890
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.8647178411483765,
      "learning_rate": 0.000405330979028543,
      "loss": 2.3589,
      "step": 88891
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.8141840696334839,
      "learning_rate": 0.0004053271488750859,
      "loss": 2.0494,
      "step": 88892
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.8423374891281128,
      "learning_rate": 0.0004053233187020466,
      "loss": 2.1893,
      "step": 88893
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.8237054944038391,
      "learning_rate": 0.0004053194885094257,
      "loss": 1.97,
      "step": 88894
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.8756911158561707,
      "learning_rate": 0.00040531565829722387,
      "loss": 2.2529,
      "step": 88895
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.9172561168670654,
      "learning_rate": 0.00040531182806544184,
      "loss": 2.1794,
      "step": 88896
    },
    {
      "epoch": 1.16,
      "grad_norm": 1.0920240879058838,
      "learning_rate": 0.0004053079978140804,
      "loss": 2.1395,
      "step": 88897
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.8883600234985352,
      "learning_rate": 0.00040530416754314027,
      "loss": 2.0539,
      "step": 88898
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.8639088869094849,
      "learning_rate": 0.00040530033725262203,
      "loss": 2.2988,
      "step": 88899
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.8113262057304382,
      "learning_rate": 0.0004052965069425265,
      "loss": 2.3452,
      "step": 88900
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.8971609473228455,
      "learning_rate": 0.00040529267661285436,
      "loss": 2.1985,
      "step": 88901
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.8202341794967651,
      "learning_rate": 0.00040528884626360635,
      "loss": 2.3496,
      "step": 88902
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.783220112323761,
      "learning_rate": 0.00040528501589478307,
      "loss": 2.1603,
      "step": 88903
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.7659010291099548,
      "learning_rate": 0.00040528118550638544,
      "loss": 2.2137,
      "step": 88904
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.9172677397727966,
      "learning_rate": 0.00040527735509841395,
      "loss": 2.3188,
      "step": 88905
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.8821231722831726,
      "learning_rate": 0.00040527352467086946,
      "loss": 2.252,
      "step": 88906
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.800170361995697,
      "learning_rate": 0.00040526969422375263,
      "loss": 2.1173,
      "step": 88907
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.8030064702033997,
      "learning_rate": 0.0004052658637570641,
      "loss": 2.4686,
      "step": 88908
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.8714753985404968,
      "learning_rate": 0.00040526203327080477,
      "loss": 2.33,
      "step": 88909
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.8860753774642944,
      "learning_rate": 0.0004052582027649753,
      "loss": 2.3066,
      "step": 88910
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.8806573152542114,
      "learning_rate": 0.00040525437223957615,
      "loss": 2.0224,
      "step": 88911
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.770991861820221,
      "learning_rate": 0.00040525054169460834,
      "loss": 2.2248,
      "step": 88912
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.7950706481933594,
      "learning_rate": 0.0004052467111300724,
      "loss": 2.1103,
      "step": 88913
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.8036829829216003,
      "learning_rate": 0.0004052428805459692,
      "loss": 2.222,
      "step": 88914
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.8201680779457092,
      "learning_rate": 0.0004052390499422993,
      "loss": 2.1424,
      "step": 88915
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.8079702854156494,
      "learning_rate": 0.0004052352193190635,
      "loss": 1.895,
      "step": 88916
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.7858794331550598,
      "learning_rate": 0.00040523138867626254,
      "loss": 2.152,
      "step": 88917
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.9223650693893433,
      "learning_rate": 0.00040522755801389695,
      "loss": 2.2852,
      "step": 88918
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.9018848538398743,
      "learning_rate": 0.0004052237273319676,
      "loss": 2.1629,
      "step": 88919
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.8456519842147827,
      "learning_rate": 0.0004052198966304753,
      "loss": 1.9612,
      "step": 88920
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.8019607663154602,
      "learning_rate": 0.0004052160659094205,
      "loss": 2.1316,
      "step": 88921
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.860178530216217,
      "learning_rate": 0.0004052122351688041,
      "loss": 2.1606,
      "step": 88922
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.8371264934539795,
      "learning_rate": 0.0004052084044086268,
      "loss": 2.1131,
      "step": 88923
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.7294368147850037,
      "learning_rate": 0.00040520457362888917,
      "loss": 1.9567,
      "step": 88924
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.8525815606117249,
      "learning_rate": 0.0004052007428295921,
      "loss": 2.534,
      "step": 88925
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.8469685316085815,
      "learning_rate": 0.0004051969120107362,
      "loss": 2.2463,
      "step": 88926
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.7994360327720642,
      "learning_rate": 0.00040519308117232226,
      "loss": 2.1778,
      "step": 88927
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.8817224502563477,
      "learning_rate": 0.00040518925031435087,
      "loss": 1.919,
      "step": 88928
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.8157957196235657,
      "learning_rate": 0.00040518541943682286,
      "loss": 2.3559,
      "step": 88929
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.8799976706504822,
      "learning_rate": 0.00040518158853973887,
      "loss": 2.2408,
      "step": 88930
    },
    {
      "epoch": 1.16,
      "grad_norm": 1.0225931406021118,
      "learning_rate": 0.0004051777576230996,
      "loss": 2.053,
      "step": 88931
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.8525317311286926,
      "learning_rate": 0.00040517392668690586,
      "loss": 2.3588,
      "step": 88932
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.8569023013114929,
      "learning_rate": 0.00040517009573115824,
      "loss": 2.1904,
      "step": 88933
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.9739468097686768,
      "learning_rate": 0.0004051662647558576,
      "loss": 2.099,
      "step": 88934
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.8803138136863708,
      "learning_rate": 0.0004051624337610045,
      "loss": 2.3142,
      "step": 88935
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.8085880875587463,
      "learning_rate": 0.00040515860274659973,
      "loss": 2.2202,
      "step": 88936
    },
    {
      "epoch": 1.16,
      "grad_norm": 1.0159286260604858,
      "learning_rate": 0.000405154771712644,
      "loss": 1.9891,
      "step": 88937
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.8654026985168457,
      "learning_rate": 0.00040515094065913805,
      "loss": 2.1962,
      "step": 88938
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.8152778744697571,
      "learning_rate": 0.0004051471095860825,
      "loss": 2.3037,
      "step": 88939
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.8854197859764099,
      "learning_rate": 0.0004051432784934781,
      "loss": 2.0127,
      "step": 88940
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.8853246569633484,
      "learning_rate": 0.00040513944738132567,
      "loss": 2.4059,
      "step": 88941
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.8508790135383606,
      "learning_rate": 0.00040513561624962575,
      "loss": 2.0388,
      "step": 88942
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.9013841152191162,
      "learning_rate": 0.00040513178509837914,
      "loss": 2.1937,
      "step": 88943
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.9869330525398254,
      "learning_rate": 0.00040512795392758663,
      "loss": 2.4394,
      "step": 88944
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.797365128993988,
      "learning_rate": 0.0004051241227372488,
      "loss": 2.1622,
      "step": 88945
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.8187456130981445,
      "learning_rate": 0.00040512029152736635,
      "loss": 2.1185,
      "step": 88946
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.8739846348762512,
      "learning_rate": 0.00040511646029794013,
      "loss": 2.0719,
      "step": 88947
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.846673309803009,
      "learning_rate": 0.0004051126290489707,
      "loss": 2.0921,
      "step": 88948
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.8067785501480103,
      "learning_rate": 0.00040510879778045893,
      "loss": 2.1942,
      "step": 88949
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.8194959759712219,
      "learning_rate": 0.00040510496649240547,
      "loss": 2.1696,
      "step": 88950
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.9136762619018555,
      "learning_rate": 0.00040510113518481093,
      "loss": 2.1422,
      "step": 88951
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.8514983057975769,
      "learning_rate": 0.00040509730385767604,
      "loss": 1.9971,
      "step": 88952
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.8642069101333618,
      "learning_rate": 0.0004050934725110017,
      "loss": 2.0918,
      "step": 88953
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.8663430213928223,
      "learning_rate": 0.0004050896411447885,
      "loss": 2.2721,
      "step": 88954
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.7968068718910217,
      "learning_rate": 0.00040508580975903713,
      "loss": 2.117,
      "step": 88955
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.8430368900299072,
      "learning_rate": 0.00040508197835374834,
      "loss": 2.1698,
      "step": 88956
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.8248867988586426,
      "learning_rate": 0.0004050781469289228,
      "loss": 2.0542,
      "step": 88957
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.8153325915336609,
      "learning_rate": 0.0004050743154845612,
      "loss": 2.2678,
      "step": 88958
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.8493349552154541,
      "learning_rate": 0.0004050704840206644,
      "loss": 2.1051,
      "step": 88959
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.7977831363677979,
      "learning_rate": 0.00040506665253723295,
      "loss": 2.371,
      "step": 88960
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.8908777236938477,
      "learning_rate": 0.0004050628210342676,
      "loss": 2.2196,
      "step": 88961
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.8354683518409729,
      "learning_rate": 0.00040505898951176915,
      "loss": 2.2034,
      "step": 88962
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.8805404305458069,
      "learning_rate": 0.00040505515796973823,
      "loss": 2.0215,
      "step": 88963
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.8710890412330627,
      "learning_rate": 0.00040505132640817563,
      "loss": 2.0056,
      "step": 88964
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.7778328061103821,
      "learning_rate": 0.00040504749482708194,
      "loss": 2.2626,
      "step": 88965
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.8430074453353882,
      "learning_rate": 0.00040504366322645797,
      "loss": 2.0088,
      "step": 88966
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.8874021172523499,
      "learning_rate": 0.0004050398316063043,
      "loss": 2.2335,
      "step": 88967
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.8283694982528687,
      "learning_rate": 0.00040503599996662186,
      "loss": 1.9352,
      "step": 88968
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.7927678823471069,
      "learning_rate": 0.0004050321683074112,
      "loss": 2.0727,
      "step": 88969
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.8645671606063843,
      "learning_rate": 0.0004050283366286731,
      "loss": 2.2973,
      "step": 88970
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.8212618231773376,
      "learning_rate": 0.0004050245049304083,
      "loss": 2.0431,
      "step": 88971
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.911994457244873,
      "learning_rate": 0.0004050206732126174,
      "loss": 2.4145,
      "step": 88972
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.7840511202812195,
      "learning_rate": 0.0004050168414753011,
      "loss": 2.2055,
      "step": 88973
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.8937790393829346,
      "learning_rate": 0.00040501300971846027,
      "loss": 2.121,
      "step": 88974
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.8055081963539124,
      "learning_rate": 0.00040500917794209553,
      "loss": 2.1391,
      "step": 88975
    },
    {
      "epoch": 1.16,
      "grad_norm": 1.0234532356262207,
      "learning_rate": 0.00040500534614620757,
      "loss": 2.1497,
      "step": 88976
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.826213002204895,
      "learning_rate": 0.00040500151433079713,
      "loss": 2.1742,
      "step": 88977
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.8397778272628784,
      "learning_rate": 0.00040499768249586493,
      "loss": 2.4237,
      "step": 88978
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.8555180430412292,
      "learning_rate": 0.0004049938506414118,
      "loss": 2.2881,
      "step": 88979
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.9153323173522949,
      "learning_rate": 0.0004049900187674382,
      "loss": 2.1784,
      "step": 88980
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.9895663261413574,
      "learning_rate": 0.000404986186873945,
      "loss": 2.1729,
      "step": 88981
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.8628551959991455,
      "learning_rate": 0.0004049823549609329,
      "loss": 2.279,
      "step": 88982
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.7749020457267761,
      "learning_rate": 0.00040497852302840255,
      "loss": 2.0363,
      "step": 88983
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.874308705329895,
      "learning_rate": 0.0004049746910763548,
      "loss": 2.1492,
      "step": 88984
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.8587044477462769,
      "learning_rate": 0.0004049708591047902,
      "loss": 2.2795,
      "step": 88985
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.9331351518630981,
      "learning_rate": 0.0004049670271137096,
      "loss": 2.1198,
      "step": 88986
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.7709001302719116,
      "learning_rate": 0.0004049631951031136,
      "loss": 2.047,
      "step": 88987
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.8317400217056274,
      "learning_rate": 0.00040495936307300296,
      "loss": 2.2911,
      "step": 88988
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.8393154740333557,
      "learning_rate": 0.00040495553102337846,
      "loss": 2.1723,
      "step": 88989
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.8445183038711548,
      "learning_rate": 0.0004049516989542407,
      "loss": 2.155,
      "step": 88990
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.8439305424690247,
      "learning_rate": 0.00040494786686559033,
      "loss": 2.4728,
      "step": 88991
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.8086532950401306,
      "learning_rate": 0.0004049440347574283,
      "loss": 2.2406,
      "step": 88992
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.9319931268692017,
      "learning_rate": 0.0004049402026297552,
      "loss": 2.237,
      "step": 88993
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.8330976366996765,
      "learning_rate": 0.00040493637048257165,
      "loss": 2.2232,
      "step": 88994
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.8407315015792847,
      "learning_rate": 0.00040493253831587846,
      "loss": 2.1275,
      "step": 88995
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.9177454710006714,
      "learning_rate": 0.00040492870612967647,
      "loss": 2.3191,
      "step": 88996
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.9379241466522217,
      "learning_rate": 0.00040492487392396615,
      "loss": 1.9607,
      "step": 88997
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.7720638513565063,
      "learning_rate": 0.0004049210416987483,
      "loss": 2.2334,
      "step": 88998
    },
    {
      "epoch": 1.16,
      "grad_norm": 1.0537387132644653,
      "learning_rate": 0.0004049172094540237,
      "loss": 2.0814,
      "step": 88999
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.8484542965888977,
      "learning_rate": 0.000404913377189793,
      "loss": 2.1351,
      "step": 89000
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.7847674489021301,
      "learning_rate": 0.0004049095449060568,
      "loss": 2.3134,
      "step": 89001
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.7649487257003784,
      "learning_rate": 0.0004049057126028161,
      "loss": 2.1584,
      "step": 89002
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.8040497899055481,
      "learning_rate": 0.0004049018802800714,
      "loss": 2.1731,
      "step": 89003
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.8893244862556458,
      "learning_rate": 0.00040489804793782336,
      "loss": 2.1805,
      "step": 89004
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.8731226325035095,
      "learning_rate": 0.00040489421557607293,
      "loss": 2.3714,
      "step": 89005
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.8471665978431702,
      "learning_rate": 0.00040489038319482064,
      "loss": 2.1286,
      "step": 89006
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.8744380474090576,
      "learning_rate": 0.0004048865507940672,
      "loss": 2.293,
      "step": 89007
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.8261361122131348,
      "learning_rate": 0.00040488271837381347,
      "loss": 2.1982,
      "step": 89008
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.8794934749603271,
      "learning_rate": 0.00040487888593405995,
      "loss": 2.2901,
      "step": 89009
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.7788923382759094,
      "learning_rate": 0.0004048750534748075,
      "loss": 2.2092,
      "step": 89010
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.843116044998169,
      "learning_rate": 0.00040487122099605686,
      "loss": 2.2117,
      "step": 89011
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.8336324095726013,
      "learning_rate": 0.00040486738849780863,
      "loss": 2.1186,
      "step": 89012
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.8170480132102966,
      "learning_rate": 0.00040486355598006357,
      "loss": 2.1635,
      "step": 89013
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.8645462393760681,
      "learning_rate": 0.00040485972344282245,
      "loss": 2.0267,
      "step": 89014
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.7446713447570801,
      "learning_rate": 0.0004048558908860858,
      "loss": 2.1464,
      "step": 89015
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.8615607619285583,
      "learning_rate": 0.0004048520583098545,
      "loss": 2.0643,
      "step": 89016
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.8998410701751709,
      "learning_rate": 0.0004048482257141293,
      "loss": 2.0495,
      "step": 89017
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.7674494385719299,
      "learning_rate": 0.0004048443930989108,
      "loss": 2.2205,
      "step": 89018
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.8498814105987549,
      "learning_rate": 0.00040484056046419974,
      "loss": 2.0408,
      "step": 89019
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.7824779152870178,
      "learning_rate": 0.0004048367278099969,
      "loss": 2.2461,
      "step": 89020
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.789894700050354,
      "learning_rate": 0.0004048328951363029,
      "loss": 1.9725,
      "step": 89021
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.8460904955863953,
      "learning_rate": 0.00040482906244311843,
      "loss": 2.3605,
      "step": 89022
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.9408765435218811,
      "learning_rate": 0.0004048252297304443,
      "loss": 2.3528,
      "step": 89023
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.7807167172431946,
      "learning_rate": 0.0004048213969982812,
      "loss": 2.0649,
      "step": 89024
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.9001962542533875,
      "learning_rate": 0.0004048175642466298,
      "loss": 2.2364,
      "step": 89025
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.8620352149009705,
      "learning_rate": 0.0004048137314754908,
      "loss": 2.1051,
      "step": 89026
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.8113021850585938,
      "learning_rate": 0.000404809898684865,
      "loss": 2.1172,
      "step": 89027
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.8807389736175537,
      "learning_rate": 0.00040480606587475304,
      "loss": 2.1428,
      "step": 89028
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.8385382294654846,
      "learning_rate": 0.0004048022330451557,
      "loss": 2.1023,
      "step": 89029
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.7734715342521667,
      "learning_rate": 0.0004047984001960736,
      "loss": 1.8288,
      "step": 89030
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.8585346341133118,
      "learning_rate": 0.0004047945673275075,
      "loss": 2.2507,
      "step": 89031
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.891805112361908,
      "learning_rate": 0.0004047907344394581,
      "loss": 2.1697,
      "step": 89032
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.9427591562271118,
      "learning_rate": 0.0004047869015319262,
      "loss": 2.1461,
      "step": 89033
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.8324044942855835,
      "learning_rate": 0.0004047830686049123,
      "loss": 2.0185,
      "step": 89034
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.8419538140296936,
      "learning_rate": 0.0004047792356584174,
      "loss": 2.1197,
      "step": 89035
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.8403522372245789,
      "learning_rate": 0.00040477540269244196,
      "loss": 2.1,
      "step": 89036
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.8405492901802063,
      "learning_rate": 0.00040477156970698683,
      "loss": 2.1993,
      "step": 89037
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.8201548457145691,
      "learning_rate": 0.00040476773670205276,
      "loss": 1.9601,
      "step": 89038
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.8458893895149231,
      "learning_rate": 0.00040476390367764025,
      "loss": 2.1103,
      "step": 89039
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.9351216554641724,
      "learning_rate": 0.0004047600706337502,
      "loss": 2.2066,
      "step": 89040
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.801119863986969,
      "learning_rate": 0.0004047562375703834,
      "loss": 2.202,
      "step": 89041
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.8585934042930603,
      "learning_rate": 0.0004047524044875402,
      "loss": 2.2493,
      "step": 89042
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.869086503982544,
      "learning_rate": 0.00040474857138522177,
      "loss": 2.217,
      "step": 89043
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.8459897637367249,
      "learning_rate": 0.0004047447382634285,
      "loss": 2.2744,
      "step": 89044
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.7987155914306641,
      "learning_rate": 0.00040474090512216125,
      "loss": 1.996,
      "step": 89045
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.8430237770080566,
      "learning_rate": 0.0004047370719614206,
      "loss": 2.2684,
      "step": 89046
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.8289120197296143,
      "learning_rate": 0.0004047332387812074,
      "loss": 2.3125,
      "step": 89047
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.873746395111084,
      "learning_rate": 0.0004047294055815224,
      "loss": 2.1505,
      "step": 89048
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.9187464118003845,
      "learning_rate": 0.0004047255723623661,
      "loss": 2.1372,
      "step": 89049
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.8790880441665649,
      "learning_rate": 0.00040472173912373935,
      "loss": 2.1488,
      "step": 89050
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.8248251080513,
      "learning_rate": 0.000404717905865643,
      "loss": 2.2543,
      "step": 89051
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.8470265865325928,
      "learning_rate": 0.0004047140725880774,
      "loss": 1.9366,
      "step": 89052
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.9835533499717712,
      "learning_rate": 0.0004047102392910436,
      "loss": 2.1344,
      "step": 89053
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.8482131958007812,
      "learning_rate": 0.0004047064059745422,
      "loss": 2.434,
      "step": 89054
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.875765323638916,
      "learning_rate": 0.0004047025726385739,
      "loss": 2.1348,
      "step": 89055
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.9618349075317383,
      "learning_rate": 0.00040469873928313937,
      "loss": 2.1926,
      "step": 89056
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.8191442489624023,
      "learning_rate": 0.0004046949059082394,
      "loss": 2.2667,
      "step": 89057
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.7998408675193787,
      "learning_rate": 0.00040469107251387467,
      "loss": 2.2569,
      "step": 89058
    },
    {
      "epoch": 1.16,
      "grad_norm": 1.0613566637039185,
      "learning_rate": 0.00040468723910004584,
      "loss": 2.1125,
      "step": 89059
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.9034850001335144,
      "learning_rate": 0.00040468340566675374,
      "loss": 2.2794,
      "step": 89060
    },
    {
      "epoch": 1.16,
      "grad_norm": 1.436758041381836,
      "learning_rate": 0.000404679572213999,
      "loss": 2.2965,
      "step": 89061
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.9459869861602783,
      "learning_rate": 0.00040467573874178233,
      "loss": 1.8857,
      "step": 89062
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.8982343673706055,
      "learning_rate": 0.0004046719052501046,
      "loss": 2.1016,
      "step": 89063
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.7914316058158875,
      "learning_rate": 0.0004046680717389662,
      "loss": 2.256,
      "step": 89064
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.840322732925415,
      "learning_rate": 0.00040466423820836807,
      "loss": 2.3164,
      "step": 89065
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.9064221978187561,
      "learning_rate": 0.00040466040465831087,
      "loss": 2.1618,
      "step": 89066
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.9130581617355347,
      "learning_rate": 0.00040465657108879543,
      "loss": 2.1127,
      "step": 89067
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.8366593718528748,
      "learning_rate": 0.00040465273749982225,
      "loss": 2.318,
      "step": 89068
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.8419715166091919,
      "learning_rate": 0.0004046489038913922,
      "loss": 2.1525,
      "step": 89069
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.8708877563476562,
      "learning_rate": 0.00040464507026350603,
      "loss": 2.2748,
      "step": 89070
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.8776112198829651,
      "learning_rate": 0.00040464123661616423,
      "loss": 2.2281,
      "step": 89071
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.818843424320221,
      "learning_rate": 0.0004046374029493677,
      "loss": 2.1435,
      "step": 89072
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.7859883904457092,
      "learning_rate": 0.000404633569263117,
      "loss": 2.3667,
      "step": 89073
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.7687848806381226,
      "learning_rate": 0.00040462973555741315,
      "loss": 2.1274,
      "step": 89074
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.8342575430870056,
      "learning_rate": 0.00040462590183225653,
      "loss": 2.1028,
      "step": 89075
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.8421902656555176,
      "learning_rate": 0.000404622068087648,
      "loss": 2.0526,
      "step": 89076
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.7712152600288391,
      "learning_rate": 0.00040461823432358824,
      "loss": 2.2735,
      "step": 89077
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.9441550374031067,
      "learning_rate": 0.000404614400540078,
      "loss": 2.0483,
      "step": 89078
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.7721741795539856,
      "learning_rate": 0.0004046105667371179,
      "loss": 2.1508,
      "step": 89079
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.8227483034133911,
      "learning_rate": 0.0004046067329147088,
      "loss": 2.1495,
      "step": 89080
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.7888328433036804,
      "learning_rate": 0.0004046028990728513,
      "loss": 2.183,
      "step": 89081
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.8380240201950073,
      "learning_rate": 0.0004045990652115462,
      "loss": 2.1513,
      "step": 89082
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.9497379660606384,
      "learning_rate": 0.00040459523133079404,
      "loss": 2.1495,
      "step": 89083
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.8570138216018677,
      "learning_rate": 0.0004045913974305958,
      "loss": 2.2435,
      "step": 89084
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.8501450419425964,
      "learning_rate": 0.00040458756351095197,
      "loss": 2.2489,
      "step": 89085
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.7931740880012512,
      "learning_rate": 0.0004045837295718633,
      "loss": 2.1754,
      "step": 89086
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.8567239046096802,
      "learning_rate": 0.0004045798956133306,
      "loss": 2.1444,
      "step": 89087
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.8213877081871033,
      "learning_rate": 0.0004045760616353545,
      "loss": 2.04,
      "step": 89088
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.8597981333732605,
      "learning_rate": 0.00040457222763793575,
      "loss": 2.19,
      "step": 89089
    },
    {
      "epoch": 1.16,
      "grad_norm": 1.168550729751587,
      "learning_rate": 0.0004045683936210751,
      "loss": 2.3892,
      "step": 89090
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.8581012487411499,
      "learning_rate": 0.0004045645595847731,
      "loss": 2.2763,
      "step": 89091
    },
    {
      "epoch": 1.16,
      "grad_norm": 1.0336549282073975,
      "learning_rate": 0.0004045607255290306,
      "loss": 2.1768,
      "step": 89092
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.7931857109069824,
      "learning_rate": 0.00040455689145384833,
      "loss": 2.077,
      "step": 89093
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.8458784222602844,
      "learning_rate": 0.00040455305735922695,
      "loss": 2.1352,
      "step": 89094
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.8261904716491699,
      "learning_rate": 0.0004045492232451672,
      "loss": 2.0966,
      "step": 89095
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.9719256162643433,
      "learning_rate": 0.0004045453891116698,
      "loss": 2.1364,
      "step": 89096
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.8298323750495911,
      "learning_rate": 0.00040454155495873534,
      "loss": 2.1091,
      "step": 89097
    },
    {
      "epoch": 1.16,
      "grad_norm": 1.0383433103561401,
      "learning_rate": 0.0004045377207863647,
      "loss": 2.2556,
      "step": 89098
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.8371740579605103,
      "learning_rate": 0.0004045338865945585,
      "loss": 2.0757,
      "step": 89099
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.7863624095916748,
      "learning_rate": 0.00040453005238331747,
      "loss": 2.3039,
      "step": 89100
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.7622833251953125,
      "learning_rate": 0.0004045262181526423,
      "loss": 2.1686,
      "step": 89101
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.8054872751235962,
      "learning_rate": 0.00040452238390253383,
      "loss": 1.8557,
      "step": 89102
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.81740403175354,
      "learning_rate": 0.00040451854963299263,
      "loss": 2.0704,
      "step": 89103
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.7832443118095398,
      "learning_rate": 0.00040451471534401945,
      "loss": 2.1634,
      "step": 89104
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.7766019105911255,
      "learning_rate": 0.00040451088103561503,
      "loss": 1.974,
      "step": 89105
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.8296065330505371,
      "learning_rate": 0.00040450704670778,
      "loss": 2.2686,
      "step": 89106
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.7815301418304443,
      "learning_rate": 0.00040450321236051525,
      "loss": 2.0647,
      "step": 89107
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.8403380513191223,
      "learning_rate": 0.0004044993779938213,
      "loss": 2.2295,
      "step": 89108
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.8980324268341064,
      "learning_rate": 0.00040449554360769894,
      "loss": 2.328,
      "step": 89109
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.8159474730491638,
      "learning_rate": 0.00040449170920214887,
      "loss": 2.3035,
      "step": 89110
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.8848509192466736,
      "learning_rate": 0.0004044878747771719,
      "loss": 2.1498,
      "step": 89111
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.8112116456031799,
      "learning_rate": 0.0004044840403327686,
      "loss": 2.2145,
      "step": 89112
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.8147615194320679,
      "learning_rate": 0.00040448020586893976,
      "loss": 2.1866,
      "step": 89113
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.7690155506134033,
      "learning_rate": 0.000404476371385686,
      "loss": 2.1018,
      "step": 89114
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.8256399035453796,
      "learning_rate": 0.00040447253688300833,
      "loss": 2.0963,
      "step": 89115
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.8634936213493347,
      "learning_rate": 0.000404468702360907,
      "loss": 2.3225,
      "step": 89116
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.8144631385803223,
      "learning_rate": 0.0004044648678193831,
      "loss": 2.1616,
      "step": 89117
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.7709876894950867,
      "learning_rate": 0.0004044610332584372,
      "loss": 1.9823,
      "step": 89118
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.8504643440246582,
      "learning_rate": 0.00040445719867807,
      "loss": 1.9717,
      "step": 89119
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.7521997094154358,
      "learning_rate": 0.0004044533640782822,
      "loss": 2.2755,
      "step": 89120
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.8506831526756287,
      "learning_rate": 0.00040444952945907465,
      "loss": 2.1651,
      "step": 89121
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.8128541111946106,
      "learning_rate": 0.0004044456948204479,
      "loss": 2.3171,
      "step": 89122
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.8244055509567261,
      "learning_rate": 0.00040444186016240263,
      "loss": 2.4031,
      "step": 89123
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.8265342712402344,
      "learning_rate": 0.0004044380254849398,
      "loss": 2.3821,
      "step": 89124
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.8769586682319641,
      "learning_rate": 0.0004044341907880599,
      "loss": 2.2461,
      "step": 89125
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.8968560695648193,
      "learning_rate": 0.00040443035607176366,
      "loss": 2.3146,
      "step": 89126
    },
    {
      "epoch": 1.16,
      "grad_norm": 1.0750409364700317,
      "learning_rate": 0.00040442652133605196,
      "loss": 2.0389,
      "step": 89127
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.8070706129074097,
      "learning_rate": 0.0004044226865809253,
      "loss": 2.288,
      "step": 89128
    },
    {
      "epoch": 1.16,
      "grad_norm": 1.2385896444320679,
      "learning_rate": 0.0004044188518063845,
      "loss": 1.8486,
      "step": 89129
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.9889370203018188,
      "learning_rate": 0.00040441501701243035,
      "loss": 2.3652,
      "step": 89130
    },
    {
      "epoch": 1.16,
      "grad_norm": 1.0088618993759155,
      "learning_rate": 0.00040441118219906343,
      "loss": 2.4068,
      "step": 89131
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.8821666836738586,
      "learning_rate": 0.00040440734736628446,
      "loss": 2.0802,
      "step": 89132
    },
    {
      "epoch": 1.16,
      "grad_norm": 1.1245990991592407,
      "learning_rate": 0.0004044035125140942,
      "loss": 2.2052,
      "step": 89133
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.9340535998344421,
      "learning_rate": 0.00040439967764249344,
      "loss": 2.1689,
      "step": 89134
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.932476282119751,
      "learning_rate": 0.0004043958427514827,
      "loss": 2.066,
      "step": 89135
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.968166708946228,
      "learning_rate": 0.00040439200784106287,
      "loss": 2.1625,
      "step": 89136
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.9993862509727478,
      "learning_rate": 0.0004043881729112346,
      "loss": 2.2083,
      "step": 89137
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.792326807975769,
      "learning_rate": 0.00040438433796199854,
      "loss": 2.3297,
      "step": 89138
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.9553311467170715,
      "learning_rate": 0.00040438050299335555,
      "loss": 2.2692,
      "step": 89139
    },
    {
      "epoch": 1.16,
      "grad_norm": 1.1137763261795044,
      "learning_rate": 0.00040437666800530615,
      "loss": 2.2541,
      "step": 89140
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.888465166091919,
      "learning_rate": 0.00040437283299785126,
      "loss": 2.1556,
      "step": 89141
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.8969276547431946,
      "learning_rate": 0.00040436899797099144,
      "loss": 2.1657,
      "step": 89142
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.8306415677070618,
      "learning_rate": 0.00040436516292472743,
      "loss": 2.0255,
      "step": 89143
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.833272397518158,
      "learning_rate": 0.00040436132785906,
      "loss": 2.1727,
      "step": 89144
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.8559430837631226,
      "learning_rate": 0.00040435749277398984,
      "loss": 2.2186,
      "step": 89145
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.8597794771194458,
      "learning_rate": 0.00040435365766951757,
      "loss": 2.1617,
      "step": 89146
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.8500164747238159,
      "learning_rate": 0.00040434982254564404,
      "loss": 2.1576,
      "step": 89147
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.8023806214332581,
      "learning_rate": 0.00040434598740236996,
      "loss": 2.1164,
      "step": 89148
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.8374809622764587,
      "learning_rate": 0.0004043421522396959,
      "loss": 2.2829,
      "step": 89149
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.8189197182655334,
      "learning_rate": 0.00040433831705762274,
      "loss": 2.1847,
      "step": 89150
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.863066554069519,
      "learning_rate": 0.0004043344818561511,
      "loss": 1.9796,
      "step": 89151
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.9434211850166321,
      "learning_rate": 0.0004043306466352817,
      "loss": 2.4125,
      "step": 89152
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.8576630353927612,
      "learning_rate": 0.00040432681139501524,
      "loss": 2.1333,
      "step": 89153
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.8508293628692627,
      "learning_rate": 0.0004043229761353525,
      "loss": 2.0146,
      "step": 89154
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.8153666853904724,
      "learning_rate": 0.0004043191408562942,
      "loss": 2.1091,
      "step": 89155
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.8356086015701294,
      "learning_rate": 0.0004043153055578408,
      "loss": 2.2813,
      "step": 89156
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.8121087551116943,
      "learning_rate": 0.00040431147023999335,
      "loss": 2.2318,
      "step": 89157
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.8636875748634338,
      "learning_rate": 0.0004043076349027524,
      "loss": 2.3271,
      "step": 89158
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.9118139147758484,
      "learning_rate": 0.0004043037995461187,
      "loss": 2.3111,
      "step": 89159
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.8138535022735596,
      "learning_rate": 0.00040429996417009305,
      "loss": 2.3829,
      "step": 89160
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.8298824429512024,
      "learning_rate": 0.0004042961287746759,
      "loss": 2.2824,
      "step": 89161
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.8578305244445801,
      "learning_rate": 0.0004042922933598683,
      "loss": 2.5628,
      "step": 89162
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.7572461366653442,
      "learning_rate": 0.0004042884579256707,
      "loss": 2.1702,
      "step": 89163
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.8300344944000244,
      "learning_rate": 0.0004042846224720839,
      "loss": 2.207,
      "step": 89164
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.802568793296814,
      "learning_rate": 0.0004042807869991086,
      "loss": 2.0884,
      "step": 89165
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.9334121942520142,
      "learning_rate": 0.00040427695150674564,
      "loss": 2.3542,
      "step": 89166
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.8911474347114563,
      "learning_rate": 0.0004042731159949955,
      "loss": 2.3389,
      "step": 89167
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.8855552077293396,
      "learning_rate": 0.0004042692804638591,
      "loss": 2.1624,
      "step": 89168
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.9347126483917236,
      "learning_rate": 0.0004042654449133371,
      "loss": 2.1639,
      "step": 89169
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.8991937637329102,
      "learning_rate": 0.00040426160934343004,
      "loss": 2.083,
      "step": 89170
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.8000481724739075,
      "learning_rate": 0.00040425777375413893,
      "loss": 2.3882,
      "step": 89171
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.9498875737190247,
      "learning_rate": 0.0004042539381454643,
      "loss": 2.0742,
      "step": 89172
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.8510631322860718,
      "learning_rate": 0.0004042501025174068,
      "loss": 2.3054,
      "step": 89173
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.8256093859672546,
      "learning_rate": 0.0004042462668699673,
      "loss": 2.2184,
      "step": 89174
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.818993866443634,
      "learning_rate": 0.00040424243120314656,
      "loss": 2.2021,
      "step": 89175
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.8036336898803711,
      "learning_rate": 0.000404238595516945,
      "loss": 2.1044,
      "step": 89176
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.7938863039016724,
      "learning_rate": 0.00040423475981136357,
      "loss": 2.0823,
      "step": 89177
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.8538375496864319,
      "learning_rate": 0.0004042309240864031,
      "loss": 2.2068,
      "step": 89178
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.8186146020889282,
      "learning_rate": 0.0004042270883420639,
      "loss": 2.0627,
      "step": 89179
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.8943175673484802,
      "learning_rate": 0.00040422325257834707,
      "loss": 2.2754,
      "step": 89180
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.8965831398963928,
      "learning_rate": 0.00040421941679525303,
      "loss": 2.2462,
      "step": 89181
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.8371372818946838,
      "learning_rate": 0.0004042155809927828,
      "loss": 2.4299,
      "step": 89182
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.8112363815307617,
      "learning_rate": 0.00040421174517093686,
      "loss": 2.1934,
      "step": 89183
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.8408143520355225,
      "learning_rate": 0.0004042079093297159,
      "loss": 2.0148,
      "step": 89184
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.8298265337944031,
      "learning_rate": 0.0004042040734691209,
      "loss": 1.9103,
      "step": 89185
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.8098543882369995,
      "learning_rate": 0.00040420023758915225,
      "loss": 2.1941,
      "step": 89186
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.8218279480934143,
      "learning_rate": 0.0004041964016898108,
      "loss": 2.2481,
      "step": 89187
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.7796717882156372,
      "learning_rate": 0.00040419256577109734,
      "loss": 2.1217,
      "step": 89188
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.7775071859359741,
      "learning_rate": 0.0004041887298330125,
      "loss": 2.2435,
      "step": 89189
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.797797679901123,
      "learning_rate": 0.000404184893875557,
      "loss": 2.187,
      "step": 89190
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.7582634091377258,
      "learning_rate": 0.00040418105789873164,
      "loss": 2.071,
      "step": 89191
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.9093446731567383,
      "learning_rate": 0.00040417722190253696,
      "loss": 2.1691,
      "step": 89192
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.9107793569564819,
      "learning_rate": 0.00040417338588697376,
      "loss": 2.1357,
      "step": 89193
    },
    {
      "epoch": 1.16,
      "grad_norm": 1.2010668516159058,
      "learning_rate": 0.0004041695498520428,
      "loss": 2.0398,
      "step": 89194
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.7268084287643433,
      "learning_rate": 0.00040416571379774477,
      "loss": 2.0303,
      "step": 89195
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.7588393092155457,
      "learning_rate": 0.0004041618777240803,
      "loss": 2.2349,
      "step": 89196
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.778473436832428,
      "learning_rate": 0.0004041580416310503,
      "loss": 1.9739,
      "step": 89197
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.9713805317878723,
      "learning_rate": 0.0004041542055186552,
      "loss": 2.2536,
      "step": 89198
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.8655984997749329,
      "learning_rate": 0.0004041503693868959,
      "loss": 2.3191,
      "step": 89199
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.8896603584289551,
      "learning_rate": 0.0004041465332357731,
      "loss": 2.0389,
      "step": 89200
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.8032028675079346,
      "learning_rate": 0.00040414269706528757,
      "loss": 2.2878,
      "step": 89201
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.8159340620040894,
      "learning_rate": 0.00040413886087543985,
      "loss": 2.2181,
      "step": 89202
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.9457937479019165,
      "learning_rate": 0.00040413502466623083,
      "loss": 2.2798,
      "step": 89203
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.8211527466773987,
      "learning_rate": 0.0004041311884376611,
      "loss": 2.0912,
      "step": 89204
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.8950309753417969,
      "learning_rate": 0.0004041273521897314,
      "loss": 2.055,
      "step": 89205
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.7315599322319031,
      "learning_rate": 0.00040412351592244246,
      "loss": 2.1557,
      "step": 89206
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.8189666867256165,
      "learning_rate": 0.00040411967963579505,
      "loss": 2.305,
      "step": 89207
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.8737313151359558,
      "learning_rate": 0.0004041158433297897,
      "loss": 2.2306,
      "step": 89208
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.8299084901809692,
      "learning_rate": 0.0004041120070044274,
      "loss": 2.2735,
      "step": 89209
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.814367413520813,
      "learning_rate": 0.0004041081706597086,
      "loss": 2.4313,
      "step": 89210
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.7975728511810303,
      "learning_rate": 0.0004041043342956342,
      "loss": 2.028,
      "step": 89211
    },
    {
      "epoch": 1.16,
      "grad_norm": 1.0492382049560547,
      "learning_rate": 0.0004041004979122048,
      "loss": 1.9535,
      "step": 89212
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.8433868288993835,
      "learning_rate": 0.0004040966615094212,
      "loss": 2.3367,
      "step": 89213
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.8630035519599915,
      "learning_rate": 0.000404092825087284,
      "loss": 2.0675,
      "step": 89214
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.8351590633392334,
      "learning_rate": 0.00040408898864579407,
      "loss": 2.1429,
      "step": 89215
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.7746500372886658,
      "learning_rate": 0.00040408515218495197,
      "loss": 2.0157,
      "step": 89216
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.8583148121833801,
      "learning_rate": 0.00040408131570475843,
      "loss": 2.1646,
      "step": 89217
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.818080723285675,
      "learning_rate": 0.0004040774792052143,
      "loss": 2.228,
      "step": 89218
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.8531943559646606,
      "learning_rate": 0.00040407364268632014,
      "loss": 2.3596,
      "step": 89219
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.8031447529792786,
      "learning_rate": 0.0004040698061480767,
      "loss": 2.0902,
      "step": 89220
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.9460002779960632,
      "learning_rate": 0.00040406596959048483,
      "loss": 2.1794,
      "step": 89221
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.912761390209198,
      "learning_rate": 0.000404062133013545,
      "loss": 2.1739,
      "step": 89222
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.8411341905593872,
      "learning_rate": 0.00040405829641725814,
      "loss": 2.1492,
      "step": 89223
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.7987263202667236,
      "learning_rate": 0.0004040544598016249,
      "loss": 2.0162,
      "step": 89224
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.8144907355308533,
      "learning_rate": 0.0004040506231666459,
      "loss": 2.1256,
      "step": 89225
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.9628547430038452,
      "learning_rate": 0.00040404678651232193,
      "loss": 2.1136,
      "step": 89226
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.8706714510917664,
      "learning_rate": 0.00040404294983865377,
      "loss": 2.2024,
      "step": 89227
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.8072035908699036,
      "learning_rate": 0.000404039113145642,
      "loss": 2.2884,
      "step": 89228
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.8861308097839355,
      "learning_rate": 0.0004040352764332874,
      "loss": 2.338,
      "step": 89229
    },
    {
      "epoch": 1.16,
      "grad_norm": 1.1748857498168945,
      "learning_rate": 0.0004040314397015907,
      "loss": 2.2115,
      "step": 89230
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.8159492611885071,
      "learning_rate": 0.00040402760295055255,
      "loss": 2.2989,
      "step": 89231
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.836093008518219,
      "learning_rate": 0.0004040237661801737,
      "loss": 2.1908,
      "step": 89232
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.7376898527145386,
      "learning_rate": 0.00040401992939045493,
      "loss": 2.2242,
      "step": 89233
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.8809519410133362,
      "learning_rate": 0.0004040160925813969,
      "loss": 2.4466,
      "step": 89234
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.8344237804412842,
      "learning_rate": 0.00040401225575300025,
      "loss": 2.0654,
      "step": 89235
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.9918026328086853,
      "learning_rate": 0.0004040084189052658,
      "loss": 2.3705,
      "step": 89236
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.7731446623802185,
      "learning_rate": 0.0004040045820381942,
      "loss": 2.2679,
      "step": 89237
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.8039185404777527,
      "learning_rate": 0.0004040007451517862,
      "loss": 2.1487,
      "step": 89238
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.8092641830444336,
      "learning_rate": 0.00040399690824604245,
      "loss": 2.3029,
      "step": 89239
    },
    {
      "epoch": 1.16,
      "grad_norm": 1.0139379501342773,
      "learning_rate": 0.0004039930713209638,
      "loss": 2.0761,
      "step": 89240
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.8160774111747742,
      "learning_rate": 0.00040398923437655077,
      "loss": 1.9502,
      "step": 89241
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.8150400519371033,
      "learning_rate": 0.00040398539741280426,
      "loss": 2.3626,
      "step": 89242
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.9176394939422607,
      "learning_rate": 0.00040398156042972485,
      "loss": 2.0764,
      "step": 89243
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.8812207579612732,
      "learning_rate": 0.00040397772342731336,
      "loss": 2.1976,
      "step": 89244
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.7792434692382812,
      "learning_rate": 0.0004039738864055704,
      "loss": 2.3848,
      "step": 89245
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.8509466648101807,
      "learning_rate": 0.0004039700493644967,
      "loss": 2.1957,
      "step": 89246
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.7988594770431519,
      "learning_rate": 0.0004039662123040931,
      "loss": 2.0552,
      "step": 89247
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.8198500275611877,
      "learning_rate": 0.00040396237522436016,
      "loss": 2.3113,
      "step": 89248
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.7962000966072083,
      "learning_rate": 0.00040395853812529866,
      "loss": 2.0019,
      "step": 89249
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.8466730117797852,
      "learning_rate": 0.0004039547010069093,
      "loss": 2.1327,
      "step": 89250
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.7640115022659302,
      "learning_rate": 0.0004039508638691929,
      "loss": 2.0372,
      "step": 89251
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.8122164011001587,
      "learning_rate": 0.0004039470267121499,
      "loss": 2.3409,
      "step": 89252
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.8120146989822388,
      "learning_rate": 0.00040394318953578125,
      "loss": 2.0979,
      "step": 89253
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.8863831758499146,
      "learning_rate": 0.0004039393523400877,
      "loss": 2.061,
      "step": 89254
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.7518359422683716,
      "learning_rate": 0.00040393551512506973,
      "loss": 2.4073,
      "step": 89255
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.797466516494751,
      "learning_rate": 0.00040393167789072826,
      "loss": 2.1427,
      "step": 89256
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.9451103210449219,
      "learning_rate": 0.00040392784063706386,
      "loss": 2.4021,
      "step": 89257
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.8985311985015869,
      "learning_rate": 0.0004039240033640774,
      "loss": 2.2233,
      "step": 89258
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.8130371570587158,
      "learning_rate": 0.00040392016607176945,
      "loss": 1.9259,
      "step": 89259
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.9058540463447571,
      "learning_rate": 0.0004039163287601408,
      "loss": 2.2143,
      "step": 89260
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.7849827408790588,
      "learning_rate": 0.0004039124914291922,
      "loss": 2.0263,
      "step": 89261
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.9262995719909668,
      "learning_rate": 0.0004039086540789242,
      "loss": 2.1499,
      "step": 89262
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.8272832036018372,
      "learning_rate": 0.00040390481670933765,
      "loss": 2.395,
      "step": 89263
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.8517851829528809,
      "learning_rate": 0.00040390097932043324,
      "loss": 2.1085,
      "step": 89264
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.8861585259437561,
      "learning_rate": 0.00040389714191221165,
      "loss": 2.1932,
      "step": 89265
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.8717306852340698,
      "learning_rate": 0.0004038933044846737,
      "loss": 2.2703,
      "step": 89266
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.9219345450401306,
      "learning_rate": 0.00040388946703782,
      "loss": 2.3395,
      "step": 89267
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.8417148590087891,
      "learning_rate": 0.0004038856295716512,
      "loss": 2.161,
      "step": 89268
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.9383631944656372,
      "learning_rate": 0.00040388179208616817,
      "loss": 2.0698,
      "step": 89269
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.7979839444160461,
      "learning_rate": 0.00040387795458137165,
      "loss": 2.0399,
      "step": 89270
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.7759439945220947,
      "learning_rate": 0.0004038741170572621,
      "loss": 2.3638,
      "step": 89271
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.8509613871574402,
      "learning_rate": 0.00040387027951384037,
      "loss": 2.2332,
      "step": 89272
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.8427332043647766,
      "learning_rate": 0.00040386644195110734,
      "loss": 2.1283,
      "step": 89273
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.7637781500816345,
      "learning_rate": 0.00040386260436906353,
      "loss": 2.1505,
      "step": 89274
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.8468614816665649,
      "learning_rate": 0.00040385876676770965,
      "loss": 2.2322,
      "step": 89275
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.8972734808921814,
      "learning_rate": 0.0004038549291470465,
      "loss": 2.1329,
      "step": 89276
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.8144475817680359,
      "learning_rate": 0.00040385109150707475,
      "loss": 2.0545,
      "step": 89277
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.7555680871009827,
      "learning_rate": 0.0004038472538477951,
      "loss": 2.1224,
      "step": 89278
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.8739826679229736,
      "learning_rate": 0.0004038434161692084,
      "loss": 2.1665,
      "step": 89279
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.9222556352615356,
      "learning_rate": 0.0004038395784713151,
      "loss": 2.3113,
      "step": 89280
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.7715970277786255,
      "learning_rate": 0.00040383574075411615,
      "loss": 2.175,
      "step": 89281
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.9256935119628906,
      "learning_rate": 0.00040383190301761213,
      "loss": 2.0015,
      "step": 89282
    },
    {
      "epoch": 1.16,
      "grad_norm": 1.0614296197891235,
      "learning_rate": 0.0004038280652618038,
      "loss": 2.2903,
      "step": 89283
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.9167773127555847,
      "learning_rate": 0.0004038242274866919,
      "loss": 2.2204,
      "step": 89284
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.929499089717865,
      "learning_rate": 0.0004038203896922771,
      "loss": 2.0781,
      "step": 89285
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.8890292048454285,
      "learning_rate": 0.00040381655187856014,
      "loss": 2.0888,
      "step": 89286
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.8514743447303772,
      "learning_rate": 0.0004038127140455417,
      "loss": 2.2797,
      "step": 89287
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.9654334187507629,
      "learning_rate": 0.0004038088761932226,
      "loss": 2.2222,
      "step": 89288
    },
    {
      "epoch": 1.16,
      "grad_norm": 1.053158164024353,
      "learning_rate": 0.0004038050383216034,
      "loss": 2.3507,
      "step": 89289
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.8958539366722107,
      "learning_rate": 0.00040380120043068487,
      "loss": 2.2045,
      "step": 89290
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.7787988185882568,
      "learning_rate": 0.00040379736252046776,
      "loss": 1.9671,
      "step": 89291
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.8681269884109497,
      "learning_rate": 0.00040379352459095276,
      "loss": 2.0317,
      "step": 89292
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.9011009931564331,
      "learning_rate": 0.00040378968664214055,
      "loss": 2.163,
      "step": 89293
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.8180899024009705,
      "learning_rate": 0.00040378584867403197,
      "loss": 2.214,
      "step": 89294
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.8121791481971741,
      "learning_rate": 0.00040378201068662757,
      "loss": 2.3145,
      "step": 89295
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.7755306959152222,
      "learning_rate": 0.0004037781726799281,
      "loss": 2.0184,
      "step": 89296
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.8833619356155396,
      "learning_rate": 0.00040377433465393437,
      "loss": 1.9859,
      "step": 89297
    },
    {
      "epoch": 1.16,
      "grad_norm": 1.0426509380340576,
      "learning_rate": 0.00040377049660864707,
      "loss": 2.1271,
      "step": 89298
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.8136350512504578,
      "learning_rate": 0.0004037666585440668,
      "loss": 2.2461,
      "step": 89299
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.943885862827301,
      "learning_rate": 0.0004037628204601944,
      "loss": 2.3118,
      "step": 89300
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.8421752452850342,
      "learning_rate": 0.0004037589823570305,
      "loss": 2.0834,
      "step": 89301
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.8318750858306885,
      "learning_rate": 0.00040375514423457583,
      "loss": 2.1487,
      "step": 89302
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.8096741437911987,
      "learning_rate": 0.00040375130609283114,
      "loss": 2.1462,
      "step": 89303
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.9059080481529236,
      "learning_rate": 0.00040374746793179713,
      "loss": 1.8874,
      "step": 89304
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.7837887406349182,
      "learning_rate": 0.0004037436297514745,
      "loss": 2.0704,
      "step": 89305
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.8566265106201172,
      "learning_rate": 0.00040373979155186406,
      "loss": 2.2172,
      "step": 89306
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.9058547616004944,
      "learning_rate": 0.0004037359533329663,
      "loss": 2.4814,
      "step": 89307
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.8307982683181763,
      "learning_rate": 0.00040373211509478203,
      "loss": 2.2441,
      "step": 89308
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.8875045776367188,
      "learning_rate": 0.0004037282768373122,
      "loss": 2.2104,
      "step": 89309
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.8314564228057861,
      "learning_rate": 0.00040372443856055717,
      "loss": 2.1077,
      "step": 89310
    },
    {
      "epoch": 1.16,
      "grad_norm": 1.0400222539901733,
      "learning_rate": 0.0004037206002645178,
      "loss": 2.111,
      "step": 89311
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.781575620174408,
      "learning_rate": 0.00040371676194919496,
      "loss": 2.2785,
      "step": 89312
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.9113864898681641,
      "learning_rate": 0.0004037129236145891,
      "loss": 2.2094,
      "step": 89313
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.8840299844741821,
      "learning_rate": 0.0004037090852607011,
      "loss": 2.0163,
      "step": 89314
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.8495413661003113,
      "learning_rate": 0.0004037052468875315,
      "loss": 2.1927,
      "step": 89315
    },
    {
      "epoch": 1.16,
      "grad_norm": 1.095145344734192,
      "learning_rate": 0.0004037014084950813,
      "loss": 2.0708,
      "step": 89316
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.9706998467445374,
      "learning_rate": 0.000403697570083351,
      "loss": 2.333,
      "step": 89317
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.7814632654190063,
      "learning_rate": 0.00040369373165234136,
      "loss": 2.2654,
      "step": 89318
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.7935808897018433,
      "learning_rate": 0.00040368989320205306,
      "loss": 2.1935,
      "step": 89319
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.9312146902084351,
      "learning_rate": 0.0004036860547324869,
      "loss": 2.205,
      "step": 89320
    },
    {
      "epoch": 1.16,
      "grad_norm": 1.2209209203720093,
      "learning_rate": 0.0004036822162436435,
      "loss": 2.3107,
      "step": 89321
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.7975718379020691,
      "learning_rate": 0.00040367837773552367,
      "loss": 2.2123,
      "step": 89322
    },
    {
      "epoch": 1.16,
      "grad_norm": 1.0371980667114258,
      "learning_rate": 0.000403674539208128,
      "loss": 2.0403,
      "step": 89323
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.8848502039909363,
      "learning_rate": 0.0004036707006614573,
      "loss": 2.2127,
      "step": 89324
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.9063469171524048,
      "learning_rate": 0.00040366686209551234,
      "loss": 2.3688,
      "step": 89325
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.8243358731269836,
      "learning_rate": 0.00040366302351029366,
      "loss": 2.1366,
      "step": 89326
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.9007447361946106,
      "learning_rate": 0.00040365918490580205,
      "loss": 2.2471,
      "step": 89327
    },
    {
      "epoch": 1.16,
      "grad_norm": 1.0539203882217407,
      "learning_rate": 0.00040365534628203836,
      "loss": 2.2443,
      "step": 89328
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.8158476948738098,
      "learning_rate": 0.0004036515076390031,
      "loss": 2.2562,
      "step": 89329
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.836231529712677,
      "learning_rate": 0.0004036476689766971,
      "loss": 2.2347,
      "step": 89330
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.8628119230270386,
      "learning_rate": 0.00040364383029512093,
      "loss": 2.2941,
      "step": 89331
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.8881141543388367,
      "learning_rate": 0.0004036399915942756,
      "loss": 2.0361,
      "step": 89332
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.7938628196716309,
      "learning_rate": 0.0004036361528741615,
      "loss": 2.2225,
      "step": 89333
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.9524679183959961,
      "learning_rate": 0.00040363231413477956,
      "loss": 2.0555,
      "step": 89334
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.899243175983429,
      "learning_rate": 0.0004036284753761304,
      "loss": 2.2058,
      "step": 89335
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.7974830865859985,
      "learning_rate": 0.00040362463659821473,
      "loss": 2.1428,
      "step": 89336
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.9089423418045044,
      "learning_rate": 0.00040362079780103323,
      "loss": 2.3076,
      "step": 89337
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.8628255724906921,
      "learning_rate": 0.0004036169589845868,
      "loss": 2.2665,
      "step": 89338
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.9770514369010925,
      "learning_rate": 0.00040361312014887597,
      "loss": 2.0217,
      "step": 89339
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.864530622959137,
      "learning_rate": 0.00040360928129390145,
      "loss": 2.1526,
      "step": 89340
    },
    {
      "epoch": 1.16,
      "grad_norm": 1.0343180894851685,
      "learning_rate": 0.000403605442419664,
      "loss": 2.1254,
      "step": 89341
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.9138121008872986,
      "learning_rate": 0.00040360160352616444,
      "loss": 2.4596,
      "step": 89342
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.8128562569618225,
      "learning_rate": 0.00040359776461340336,
      "loss": 2.1908,
      "step": 89343
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.8928746581077576,
      "learning_rate": 0.0004035939256813814,
      "loss": 2.4239,
      "step": 89344
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.8422815203666687,
      "learning_rate": 0.00040359008673009945,
      "loss": 2.349,
      "step": 89345
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.8985847234725952,
      "learning_rate": 0.0004035862477595582,
      "loss": 2.0911,
      "step": 89346
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.7774916291236877,
      "learning_rate": 0.00040358240876975826,
      "loss": 2.386,
      "step": 89347
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.9484636187553406,
      "learning_rate": 0.00040357856976070035,
      "loss": 2.2345,
      "step": 89348
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.7890692353248596,
      "learning_rate": 0.0004035747307323853,
      "loss": 1.9942,
      "step": 89349
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.8357211947441101,
      "learning_rate": 0.00040357089168481373,
      "loss": 2.0258,
      "step": 89350
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.8471099734306335,
      "learning_rate": 0.0004035670526179864,
      "loss": 2.1793,
      "step": 89351
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.9659099578857422,
      "learning_rate": 0.000403563213531904,
      "loss": 2.031,
      "step": 89352
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.9104084372520447,
      "learning_rate": 0.0004035593744265672,
      "loss": 2.201,
      "step": 89353
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.9675506353378296,
      "learning_rate": 0.00040355553530197673,
      "loss": 2.283,
      "step": 89354
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.7784125208854675,
      "learning_rate": 0.00040355169615813335,
      "loss": 2.1339,
      "step": 89355
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.9681174159049988,
      "learning_rate": 0.0004035478569950378,
      "loss": 2.0864,
      "step": 89356
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.9357589483261108,
      "learning_rate": 0.0004035440178126907,
      "loss": 2.4134,
      "step": 89357
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.9549697041511536,
      "learning_rate": 0.00040354017861109285,
      "loss": 2.2608,
      "step": 89358
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.8194757699966431,
      "learning_rate": 0.000403536339390245,
      "loss": 2.0953,
      "step": 89359
    },
    {
      "epoch": 1.16,
      "grad_norm": 1.1412492990493774,
      "learning_rate": 0.00040353250015014765,
      "loss": 2.1831,
      "step": 89360
    },
    {
      "epoch": 1.16,
      "grad_norm": 1.0153858661651611,
      "learning_rate": 0.0004035286608908018,
      "loss": 2.2152,
      "step": 89361
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.9723572134971619,
      "learning_rate": 0.00040352482161220784,
      "loss": 1.9178,
      "step": 89362
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.9378166198730469,
      "learning_rate": 0.0004035209823143667,
      "loss": 2.0594,
      "step": 89363
    },
    {
      "epoch": 1.16,
      "grad_norm": 1.0289872884750366,
      "learning_rate": 0.0004035171429972792,
      "loss": 2.235,
      "step": 89364
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.7735764980316162,
      "learning_rate": 0.00040351330366094576,
      "loss": 2.1445,
      "step": 89365
    },
    {
      "epoch": 1.16,
      "grad_norm": 1.1075421571731567,
      "learning_rate": 0.0004035094643053673,
      "loss": 2.2503,
      "step": 89366
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.9440389275550842,
      "learning_rate": 0.0004035056249305445,
      "loss": 2.182,
      "step": 89367
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.7467553019523621,
      "learning_rate": 0.00040350178553647805,
      "loss": 2.1444,
      "step": 89368
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.8068192601203918,
      "learning_rate": 0.00040349794612316864,
      "loss": 2.135,
      "step": 89369
    },
    {
      "epoch": 1.16,
      "grad_norm": 1.1920393705368042,
      "learning_rate": 0.000403494106690617,
      "loss": 2.0691,
      "step": 89370
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.8481847047805786,
      "learning_rate": 0.00040349026723882384,
      "loss": 2.1599,
      "step": 89371
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.930580735206604,
      "learning_rate": 0.0004034864277677899,
      "loss": 2.2733,
      "step": 89372
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.86075758934021,
      "learning_rate": 0.0004034825882775159,
      "loss": 1.9011,
      "step": 89373
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.9033012986183167,
      "learning_rate": 0.0004034787487680025,
      "loss": 2.2262,
      "step": 89374
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.8477657437324524,
      "learning_rate": 0.0004034749092392504,
      "loss": 2.1406,
      "step": 89375
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.951758086681366,
      "learning_rate": 0.00040347106969126055,
      "loss": 2.3909,
      "step": 89376
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.8538199663162231,
      "learning_rate": 0.0004034672301240333,
      "loss": 2.2298,
      "step": 89377
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.7859323024749756,
      "learning_rate": 0.0004034633905375695,
      "loss": 2.0465,
      "step": 89378
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.8431488871574402,
      "learning_rate": 0.00040345955093187,
      "loss": 2.0425,
      "step": 89379
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.8119168281555176,
      "learning_rate": 0.0004034557113069355,
      "loss": 2.2459,
      "step": 89380
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.7493367195129395,
      "learning_rate": 0.00040345187166276647,
      "loss": 2.1073,
      "step": 89381
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.7966365218162537,
      "learning_rate": 0.0004034480319993638,
      "loss": 2.2457,
      "step": 89382
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.8254086971282959,
      "learning_rate": 0.00040344419231672833,
      "loss": 2.1239,
      "step": 89383
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.8071736693382263,
      "learning_rate": 0.0004034403526148605,
      "loss": 2.1051,
      "step": 89384
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.8828813433647156,
      "learning_rate": 0.0004034365128937612,
      "loss": 2.129,
      "step": 89385
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.8231304883956909,
      "learning_rate": 0.00040343267315343117,
      "loss": 2.1338,
      "step": 89386
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.8740437030792236,
      "learning_rate": 0.00040342883339387095,
      "loss": 1.9343,
      "step": 89387
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.8004587292671204,
      "learning_rate": 0.0004034249936150813,
      "loss": 2.0289,
      "step": 89388
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.8353673219680786,
      "learning_rate": 0.0004034211538170632,
      "loss": 2.153,
      "step": 89389
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.9144119620323181,
      "learning_rate": 0.000403417313999817,
      "loss": 2.0108,
      "step": 89390
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.8431107997894287,
      "learning_rate": 0.00040341347416334355,
      "loss": 2.0813,
      "step": 89391
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.7570055723190308,
      "learning_rate": 0.0004034096343076437,
      "loss": 2.1223,
      "step": 89392
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.8305166363716125,
      "learning_rate": 0.00040340579443271803,
      "loss": 2.2948,
      "step": 89393
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.8271338939666748,
      "learning_rate": 0.0004034019545385672,
      "loss": 2.2533,
      "step": 89394
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.8263412714004517,
      "learning_rate": 0.000403398114625192,
      "loss": 2.1673,
      "step": 89395
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.7912595868110657,
      "learning_rate": 0.0004033942746925932,
      "loss": 2.1714,
      "step": 89396
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.8964203596115112,
      "learning_rate": 0.0004033904347407714,
      "loss": 2.2795,
      "step": 89397
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.8238699436187744,
      "learning_rate": 0.00040338659476972744,
      "loss": 2.347,
      "step": 89398
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.7881922721862793,
      "learning_rate": 0.00040338275477946184,
      "loss": 2.1487,
      "step": 89399
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.7715065479278564,
      "learning_rate": 0.00040337891476997545,
      "loss": 2.0855,
      "step": 89400
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.787623941898346,
      "learning_rate": 0.00040337507474126903,
      "loss": 2.184,
      "step": 89401
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.9029487371444702,
      "learning_rate": 0.0004033712346933433,
      "loss": 2.3115,
      "step": 89402
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.8304718136787415,
      "learning_rate": 0.00040336739462619876,
      "loss": 2.3286,
      "step": 89403
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.7892179489135742,
      "learning_rate": 0.00040336355453983637,
      "loss": 2.2482,
      "step": 89404
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.7775236368179321,
      "learning_rate": 0.00040335971443425665,
      "loss": 2.2227,
      "step": 89405
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.829177975654602,
      "learning_rate": 0.00040335587430946043,
      "loss": 2.3451,
      "step": 89406
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.8218576312065125,
      "learning_rate": 0.0004033520341654485,
      "loss": 2.1122,
      "step": 89407
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.7494438290596008,
      "learning_rate": 0.0004033481940022214,
      "loss": 2.1055,
      "step": 89408
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.7450383305549622,
      "learning_rate": 0.00040334435381977993,
      "loss": 2.058,
      "step": 89409
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.8535900115966797,
      "learning_rate": 0.0004033405136181248,
      "loss": 2.2339,
      "step": 89410
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.766088604927063,
      "learning_rate": 0.0004033366733972567,
      "loss": 1.9479,
      "step": 89411
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.9335155487060547,
      "learning_rate": 0.0004033328331571764,
      "loss": 2.1789,
      "step": 89412
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.7797447443008423,
      "learning_rate": 0.0004033289928978845,
      "loss": 2.1588,
      "step": 89413
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.7353919148445129,
      "learning_rate": 0.00040332515261938187,
      "loss": 2.0689,
      "step": 89414
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.8880510330200195,
      "learning_rate": 0.0004033213123216691,
      "loss": 2.2258,
      "step": 89415
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.8990995287895203,
      "learning_rate": 0.00040331747200474696,
      "loss": 2.3067,
      "step": 89416
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.8107548356056213,
      "learning_rate": 0.0004033136316686162,
      "loss": 2.1631,
      "step": 89417
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.7784591317176819,
      "learning_rate": 0.0004033097913132774,
      "loss": 2.0153,
      "step": 89418
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.8267906904220581,
      "learning_rate": 0.00040330595093873137,
      "loss": 2.3243,
      "step": 89419
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.7949557900428772,
      "learning_rate": 0.0004033021105449789,
      "loss": 2.1743,
      "step": 89420
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.8500255942344666,
      "learning_rate": 0.0004032982701320205,
      "loss": 2.0435,
      "step": 89421
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.9794522523880005,
      "learning_rate": 0.00040329442969985703,
      "loss": 2.1245,
      "step": 89422
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.7878525257110596,
      "learning_rate": 0.0004032905892484892,
      "loss": 2.1921,
      "step": 89423
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.8551552891731262,
      "learning_rate": 0.00040328674877791775,
      "loss": 2.1519,
      "step": 89424
    },
    {
      "epoch": 1.16,
      "grad_norm": 1.0506505966186523,
      "learning_rate": 0.0004032829082881433,
      "loss": 2.5611,
      "step": 89425
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.8850923776626587,
      "learning_rate": 0.0004032790677791666,
      "loss": 2.4224,
      "step": 89426
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.8247124552726746,
      "learning_rate": 0.00040327522725098846,
      "loss": 2.1771,
      "step": 89427
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.7325146198272705,
      "learning_rate": 0.00040327138670360937,
      "loss": 2.1418,
      "step": 89428
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.82378751039505,
      "learning_rate": 0.00040326754613703025,
      "loss": 2.4366,
      "step": 89429
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.8900942802429199,
      "learning_rate": 0.00040326370555125165,
      "loss": 2.177,
      "step": 89430
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.9599694609642029,
      "learning_rate": 0.00040325986494627454,
      "loss": 2.0833,
      "step": 89431
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.9488366842269897,
      "learning_rate": 0.00040325602432209935,
      "loss": 2.3659,
      "step": 89432
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.85661780834198,
      "learning_rate": 0.0004032521836787269,
      "loss": 1.9685,
      "step": 89433
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.788845956325531,
      "learning_rate": 0.00040324834301615805,
      "loss": 2.0526,
      "step": 89434
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.9041321873664856,
      "learning_rate": 0.0004032445023343933,
      "loss": 2.0949,
      "step": 89435
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.8133382201194763,
      "learning_rate": 0.00040324066163343345,
      "loss": 2.1602,
      "step": 89436
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.861304521560669,
      "learning_rate": 0.00040323682091327926,
      "loss": 2.2179,
      "step": 89437
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.9805386066436768,
      "learning_rate": 0.00040323298017393134,
      "loss": 2.2298,
      "step": 89438
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.8489104509353638,
      "learning_rate": 0.00040322913941539043,
      "loss": 2.1107,
      "step": 89439
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.7905676364898682,
      "learning_rate": 0.00040322529863765736,
      "loss": 2.2419,
      "step": 89440
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.8189701437950134,
      "learning_rate": 0.0004032214578407327,
      "loss": 2.1683,
      "step": 89441
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.7877041101455688,
      "learning_rate": 0.00040321761702461726,
      "loss": 2.2047,
      "step": 89442
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.9680843353271484,
      "learning_rate": 0.0004032137761893117,
      "loss": 2.247,
      "step": 89443
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.7988750338554382,
      "learning_rate": 0.00040320993533481673,
      "loss": 2.1126,
      "step": 89444
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.8890613317489624,
      "learning_rate": 0.00040320609446113306,
      "loss": 2.0029,
      "step": 89445
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.8606209754943848,
      "learning_rate": 0.00040320225356826146,
      "loss": 2.2817,
      "step": 89446
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.8196209073066711,
      "learning_rate": 0.00040319841265620273,
      "loss": 2.2913,
      "step": 89447
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.798990786075592,
      "learning_rate": 0.00040319457172495726,
      "loss": 2.0346,
      "step": 89448
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.7888736128807068,
      "learning_rate": 0.00040319073077452606,
      "loss": 2.0392,
      "step": 89449
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.8421063423156738,
      "learning_rate": 0.00040318688980490986,
      "loss": 2.0997,
      "step": 89450
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.8351227045059204,
      "learning_rate": 0.0004031830488161091,
      "loss": 2.1527,
      "step": 89451
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.845051109790802,
      "learning_rate": 0.00040317920780812475,
      "loss": 2.1626,
      "step": 89452
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.9472783803939819,
      "learning_rate": 0.0004031753667809575,
      "loss": 2.1646,
      "step": 89453
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.7610401511192322,
      "learning_rate": 0.00040317152573460794,
      "loss": 2.3837,
      "step": 89454
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.8372142314910889,
      "learning_rate": 0.00040316768466907676,
      "loss": 2.2201,
      "step": 89455
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.8582380414009094,
      "learning_rate": 0.0004031638435843649,
      "loss": 2.1975,
      "step": 89456
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.8537862300872803,
      "learning_rate": 0.0004031600024804728,
      "loss": 2.1542,
      "step": 89457
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.8333134055137634,
      "learning_rate": 0.0004031561613574014,
      "loss": 2.2968,
      "step": 89458
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.8518905639648438,
      "learning_rate": 0.0004031523202151513,
      "loss": 2.221,
      "step": 89459
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.9259258508682251,
      "learning_rate": 0.00040314847905372327,
      "loss": 2.3643,
      "step": 89460
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.8592706322669983,
      "learning_rate": 0.00040314463787311796,
      "loss": 2.3122,
      "step": 89461
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.8800433278083801,
      "learning_rate": 0.00040314079667333615,
      "loss": 2.0759,
      "step": 89462
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.8025385141372681,
      "learning_rate": 0.00040313695545437844,
      "loss": 2.1563,
      "step": 89463
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.8676366209983826,
      "learning_rate": 0.0004031331142162457,
      "loss": 2.2252,
      "step": 89464
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.7986676096916199,
      "learning_rate": 0.0004031292729589385,
      "loss": 2.4097,
      "step": 89465
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.8940194845199585,
      "learning_rate": 0.0004031254316824577,
      "loss": 2.3929,
      "step": 89466
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.8064142465591431,
      "learning_rate": 0.00040312159038680384,
      "loss": 2.1684,
      "step": 89467
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.8068727850914001,
      "learning_rate": 0.0004031177490719778,
      "loss": 1.9891,
      "step": 89468
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.829069197177887,
      "learning_rate": 0.0004031139077379802,
      "loss": 2.1748,
      "step": 89469
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.9489830136299133,
      "learning_rate": 0.0004031100663848118,
      "loss": 2.2118,
      "step": 89470
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.8525899052619934,
      "learning_rate": 0.0004031062250124733,
      "loss": 2.3823,
      "step": 89471
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.8341885209083557,
      "learning_rate": 0.00040310238362096536,
      "loss": 2.1635,
      "step": 89472
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.8161817789077759,
      "learning_rate": 0.0004030985422102887,
      "loss": 2.1788,
      "step": 89473
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.8035814166069031,
      "learning_rate": 0.00040309470078044425,
      "loss": 2.1098,
      "step": 89474
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.8426828980445862,
      "learning_rate": 0.00040309085933143234,
      "loss": 2.1727,
      "step": 89475
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.8394148349761963,
      "learning_rate": 0.000403087017863254,
      "loss": 2.3067,
      "step": 89476
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.8438199758529663,
      "learning_rate": 0.00040308317637590985,
      "loss": 2.2062,
      "step": 89477
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.8410165905952454,
      "learning_rate": 0.00040307933486940053,
      "loss": 2.0755,
      "step": 89478
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.7893823385238647,
      "learning_rate": 0.00040307549334372685,
      "loss": 2.1646,
      "step": 89479
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.9308720827102661,
      "learning_rate": 0.00040307165179888956,
      "loss": 2.2594,
      "step": 89480
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.8727901577949524,
      "learning_rate": 0.00040306781023488917,
      "loss": 2.2509,
      "step": 89481
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.8069811463356018,
      "learning_rate": 0.0004030639686517266,
      "loss": 2.178,
      "step": 89482
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.7803609371185303,
      "learning_rate": 0.0004030601270494025,
      "loss": 2.2501,
      "step": 89483
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.801310122013092,
      "learning_rate": 0.00040305628542791763,
      "loss": 2.2872,
      "step": 89484
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.8064674139022827,
      "learning_rate": 0.00040305244378727254,
      "loss": 2.1587,
      "step": 89485
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.7998642325401306,
      "learning_rate": 0.00040304860212746815,
      "loss": 2.0774,
      "step": 89486
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.8682433366775513,
      "learning_rate": 0.000403044760448505,
      "loss": 2.1566,
      "step": 89487
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.7596819400787354,
      "learning_rate": 0.0004030409187503839,
      "loss": 2.0386,
      "step": 89488
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.9048723578453064,
      "learning_rate": 0.0004030370770331056,
      "loss": 2.1559,
      "step": 89489
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.7767053842544556,
      "learning_rate": 0.0004030332352966707,
      "loss": 2.1944,
      "step": 89490
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.7898958325386047,
      "learning_rate": 0.00040302939354108,
      "loss": 2.2576,
      "step": 89491
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.8350234031677246,
      "learning_rate": 0.00040302555176633417,
      "loss": 2.1367,
      "step": 89492
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.9085804224014282,
      "learning_rate": 0.00040302170997243397,
      "loss": 2.431,
      "step": 89493
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.9568604826927185,
      "learning_rate": 0.0004030178681593801,
      "loss": 2.0979,
      "step": 89494
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.7991340160369873,
      "learning_rate": 0.0004030140263271733,
      "loss": 2.3918,
      "step": 89495
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.8982091546058655,
      "learning_rate": 0.00040301018447581416,
      "loss": 2.0885,
      "step": 89496
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.8010857105255127,
      "learning_rate": 0.0004030063426053035,
      "loss": 2.2107,
      "step": 89497
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.8017007112503052,
      "learning_rate": 0.0004030025007156421,
      "loss": 2.1428,
      "step": 89498
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.8057723641395569,
      "learning_rate": 0.0004029986588068305,
      "loss": 2.1609,
      "step": 89499
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.7554423213005066,
      "learning_rate": 0.0004029948168788695,
      "loss": 2.0186,
      "step": 89500
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.8260401487350464,
      "learning_rate": 0.00040299097493175993,
      "loss": 1.9453,
      "step": 89501
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.8072686195373535,
      "learning_rate": 0.0004029871329655023,
      "loss": 2.2956,
      "step": 89502
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.89593905210495,
      "learning_rate": 0.0004029832909800974,
      "loss": 1.9562,
      "step": 89503
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.7571979761123657,
      "learning_rate": 0.000402979448975546,
      "loss": 2.0754,
      "step": 89504
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.8719142079353333,
      "learning_rate": 0.00040297560695184883,
      "loss": 2.3401,
      "step": 89505
    },
    {
      "epoch": 1.17,
      "grad_norm": 1.26594877243042,
      "learning_rate": 0.00040297176490900643,
      "loss": 2.2418,
      "step": 89506
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.8168710470199585,
      "learning_rate": 0.00040296792284701973,
      "loss": 2.2476,
      "step": 89507
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.7924215793609619,
      "learning_rate": 0.0004029640807658894,
      "loss": 2.0061,
      "step": 89508
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.8324189186096191,
      "learning_rate": 0.000402960238665616,
      "loss": 2.1158,
      "step": 89509
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.822813868522644,
      "learning_rate": 0.00040295639654620033,
      "loss": 2.0866,
      "step": 89510
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.759470522403717,
      "learning_rate": 0.0004029525544076431,
      "loss": 2.3487,
      "step": 89511
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.8992813229560852,
      "learning_rate": 0.00040294871224994514,
      "loss": 1.946,
      "step": 89512
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.7595551013946533,
      "learning_rate": 0.00040294487007310706,
      "loss": 2.0411,
      "step": 89513
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.9705174565315247,
      "learning_rate": 0.00040294102787712956,
      "loss": 2.3458,
      "step": 89514
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.7717810273170471,
      "learning_rate": 0.00040293718566201345,
      "loss": 2.188,
      "step": 89515
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.8968310356140137,
      "learning_rate": 0.0004029333434277593,
      "loss": 2.1947,
      "step": 89516
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.8182002902030945,
      "learning_rate": 0.0004029295011743678,
      "loss": 2.3127,
      "step": 89517
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.9562832117080688,
      "learning_rate": 0.0004029256589018399,
      "loss": 2.279,
      "step": 89518
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.839352548122406,
      "learning_rate": 0.00040292181661017617,
      "loss": 2.5001,
      "step": 89519
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.8170350790023804,
      "learning_rate": 0.0004029179742993773,
      "loss": 2.2516,
      "step": 89520
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.8733401298522949,
      "learning_rate": 0.00040291413196944404,
      "loss": 2.0336,
      "step": 89521
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.8432015776634216,
      "learning_rate": 0.0004029102896203771,
      "loss": 2.327,
      "step": 89522
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.8469166159629822,
      "learning_rate": 0.0004029064472521772,
      "loss": 1.9314,
      "step": 89523
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.7916465401649475,
      "learning_rate": 0.00040290260486484507,
      "loss": 2.0242,
      "step": 89524
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.9574386477470398,
      "learning_rate": 0.0004028987624583814,
      "loss": 2.2956,
      "step": 89525
    },
    {
      "epoch": 1.17,
      "grad_norm": 1.1430522203445435,
      "learning_rate": 0.0004028949200327869,
      "loss": 2.2134,
      "step": 89526
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.9369150996208191,
      "learning_rate": 0.0004028910775880623,
      "loss": 2.272,
      "step": 89527
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.8974939584732056,
      "learning_rate": 0.00040288723512420824,
      "loss": 2.1521,
      "step": 89528
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.8820043802261353,
      "learning_rate": 0.00040288339264122556,
      "loss": 2.0379,
      "step": 89529
    },
    {
      "epoch": 1.17,
      "grad_norm": 1.060692548751831,
      "learning_rate": 0.0004028795501391149,
      "loss": 2.0542,
      "step": 89530
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.8587432503700256,
      "learning_rate": 0.000402875707617877,
      "loss": 2.2738,
      "step": 89531
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.939024031162262,
      "learning_rate": 0.00040287186507751257,
      "loss": 2.1017,
      "step": 89532
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.9628185033798218,
      "learning_rate": 0.00040286802251802235,
      "loss": 2.1747,
      "step": 89533
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.7901230454444885,
      "learning_rate": 0.00040286417993940697,
      "loss": 2.125,
      "step": 89534
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.9878812432289124,
      "learning_rate": 0.00040286033734166724,
      "loss": 2.2899,
      "step": 89535
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.9088442325592041,
      "learning_rate": 0.0004028564947248037,
      "loss": 2.1071,
      "step": 89536
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.8669801950454712,
      "learning_rate": 0.00040285265208881734,
      "loss": 2.2814,
      "step": 89537
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.8701116442680359,
      "learning_rate": 0.00040284880943370873,
      "loss": 1.9904,
      "step": 89538
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.8878218531608582,
      "learning_rate": 0.00040284496675947854,
      "loss": 2.3029,
      "step": 89539
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.8179814219474792,
      "learning_rate": 0.00040284112406612746,
      "loss": 2.1729,
      "step": 89540
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.771040141582489,
      "learning_rate": 0.00040283728135365643,
      "loss": 2.2835,
      "step": 89541
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.7830716371536255,
      "learning_rate": 0.0004028334386220659,
      "loss": 2.1523,
      "step": 89542
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.8627082109451294,
      "learning_rate": 0.0004028295958713567,
      "loss": 2.3012,
      "step": 89543
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.8633415699005127,
      "learning_rate": 0.00040282575310152954,
      "loss": 2.14,
      "step": 89544
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.7619640827178955,
      "learning_rate": 0.00040282191031258517,
      "loss": 2.0051,
      "step": 89545
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.763324499130249,
      "learning_rate": 0.0004028180675045242,
      "loss": 2.3855,
      "step": 89546
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.8470709323883057,
      "learning_rate": 0.0004028142246773475,
      "loss": 2.3133,
      "step": 89547
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.8437092900276184,
      "learning_rate": 0.0004028103818310556,
      "loss": 2.0052,
      "step": 89548
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.8965861797332764,
      "learning_rate": 0.00040280653896564936,
      "loss": 2.1421,
      "step": 89549
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.9251240491867065,
      "learning_rate": 0.00040280269608112954,
      "loss": 2.2312,
      "step": 89550
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.8242234587669373,
      "learning_rate": 0.0004027988531774966,
      "loss": 2.1827,
      "step": 89551
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.8218323588371277,
      "learning_rate": 0.00040279501025475143,
      "loss": 1.9969,
      "step": 89552
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.8228917121887207,
      "learning_rate": 0.00040279116731289477,
      "loss": 2.3802,
      "step": 89553
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.788797914981842,
      "learning_rate": 0.0004027873243519273,
      "loss": 2.419,
      "step": 89554
    },
    {
      "epoch": 1.17,
      "grad_norm": 1.006354570388794,
      "learning_rate": 0.0004027834813718497,
      "loss": 2.1883,
      "step": 89555
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.8544912934303284,
      "learning_rate": 0.00040277963837266276,
      "loss": 2.005,
      "step": 89556
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.8360743522644043,
      "learning_rate": 0.00040277579535436713,
      "loss": 2.2934,
      "step": 89557
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.9966275095939636,
      "learning_rate": 0.00040277195231696347,
      "loss": 2.3096,
      "step": 89558
    },
    {
      "epoch": 1.17,
      "grad_norm": 1.0421723127365112,
      "learning_rate": 0.0004027681092604527,
      "loss": 2.178,
      "step": 89559
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.7222549319267273,
      "learning_rate": 0.00040276426618483525,
      "loss": 2.1601,
      "step": 89560
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.8277648091316223,
      "learning_rate": 0.00040276042309011204,
      "loss": 2.2524,
      "step": 89561
    },
    {
      "epoch": 1.17,
      "grad_norm": 1.3814153671264648,
      "learning_rate": 0.00040275657997628373,
      "loss": 1.9695,
      "step": 89562
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.764048159122467,
      "learning_rate": 0.0004027527368433511,
      "loss": 2.2671,
      "step": 89563
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.7788968086242676,
      "learning_rate": 0.0004027488936913147,
      "loss": 2.0553,
      "step": 89564
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.9766948223114014,
      "learning_rate": 0.00040274505052017537,
      "loss": 1.9849,
      "step": 89565
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.9261506199836731,
      "learning_rate": 0.0004027412073299338,
      "loss": 2.2319,
      "step": 89566
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.7795146107673645,
      "learning_rate": 0.0004027373641205907,
      "loss": 2.0678,
      "step": 89567
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.8463044762611389,
      "learning_rate": 0.0004027335208921468,
      "loss": 2.2463,
      "step": 89568
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.9051457047462463,
      "learning_rate": 0.00040272967764460276,
      "loss": 2.3523,
      "step": 89569
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.7903969883918762,
      "learning_rate": 0.00040272583437795935,
      "loss": 2.1137,
      "step": 89570
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.8181799054145813,
      "learning_rate": 0.00040272199109221734,
      "loss": 2.2389,
      "step": 89571
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.8384283781051636,
      "learning_rate": 0.0004027181477873772,
      "loss": 2.0319,
      "step": 89572
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.8941256403923035,
      "learning_rate": 0.00040271430446344,
      "loss": 2.3778,
      "step": 89573
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.8789822459220886,
      "learning_rate": 0.00040271046112040615,
      "loss": 2.1578,
      "step": 89574
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.7943218946456909,
      "learning_rate": 0.00040270661775827657,
      "loss": 1.837,
      "step": 89575
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.933489978313446,
      "learning_rate": 0.0004027027743770518,
      "loss": 2.1397,
      "step": 89576
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.8434092402458191,
      "learning_rate": 0.0004026989309767328,
      "loss": 2.3218,
      "step": 89577
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.7857353687286377,
      "learning_rate": 0.00040269508755732,
      "loss": 2.2424,
      "step": 89578
    },
    {
      "epoch": 1.17,
      "grad_norm": 1.1720428466796875,
      "learning_rate": 0.0004026912441188143,
      "loss": 2.185,
      "step": 89579
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.9571992754936218,
      "learning_rate": 0.0004026874006612162,
      "loss": 2.2071,
      "step": 89580
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.8452754020690918,
      "learning_rate": 0.0004026835571845268,
      "loss": 1.9375,
      "step": 89581
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.9917460083961487,
      "learning_rate": 0.00040267971368874647,
      "loss": 2.2601,
      "step": 89582
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.9051693081855774,
      "learning_rate": 0.0004026758701738761,
      "loss": 2.1868,
      "step": 89583
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.7989223003387451,
      "learning_rate": 0.0004026720266399163,
      "loss": 2.2114,
      "step": 89584
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.8270600438117981,
      "learning_rate": 0.00040266818308686784,
      "loss": 1.8673,
      "step": 89585
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.9248800873756409,
      "learning_rate": 0.0004026643395147314,
      "loss": 2.0474,
      "step": 89586
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.9566389322280884,
      "learning_rate": 0.0004026604959235078,
      "loss": 2.112,
      "step": 89587
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.8196879625320435,
      "learning_rate": 0.0004026566523131977,
      "loss": 2.1984,
      "step": 89588
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.926949143409729,
      "learning_rate": 0.00040265280868380165,
      "loss": 2.2184,
      "step": 89589
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.926025390625,
      "learning_rate": 0.0004026489650353206,
      "loss": 2.0962,
      "step": 89590
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.8928958177566528,
      "learning_rate": 0.0004026451213677552,
      "loss": 2.2845,
      "step": 89591
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.8541338443756104,
      "learning_rate": 0.00040264127768110605,
      "loss": 2.2128,
      "step": 89592
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.7835108041763306,
      "learning_rate": 0.000402637433975374,
      "loss": 2.2609,
      "step": 89593
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.8577665090560913,
      "learning_rate": 0.0004026335902505597,
      "loss": 2.2661,
      "step": 89594
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.9181984663009644,
      "learning_rate": 0.0004026297465066639,
      "loss": 2.4833,
      "step": 89595
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.9541023373603821,
      "learning_rate": 0.00040262590274368725,
      "loss": 2.1229,
      "step": 89596
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.964911699295044,
      "learning_rate": 0.0004026220589616305,
      "loss": 1.8895,
      "step": 89597
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.7796120643615723,
      "learning_rate": 0.0004026182151604944,
      "loss": 2.021,
      "step": 89598
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.8447595834732056,
      "learning_rate": 0.00040261437134027963,
      "loss": 2.0581,
      "step": 89599
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.9083365201950073,
      "learning_rate": 0.0004026105275009869,
      "loss": 2.2554,
      "step": 89600
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.9369205832481384,
      "learning_rate": 0.0004026066836426169,
      "loss": 2.4247,
      "step": 89601
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.8314827084541321,
      "learning_rate": 0.0004026028397651705,
      "loss": 2.2504,
      "step": 89602
    },
    {
      "epoch": 1.17,
      "grad_norm": 1.0046595335006714,
      "learning_rate": 0.0004025989958686482,
      "loss": 2.4312,
      "step": 89603
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.7941771745681763,
      "learning_rate": 0.00040259515195305084,
      "loss": 2.0023,
      "step": 89604
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.7871909141540527,
      "learning_rate": 0.0004025913080183791,
      "loss": 2.1702,
      "step": 89605
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.8252044320106506,
      "learning_rate": 0.0004025874640646337,
      "loss": 2.142,
      "step": 89606
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.8695790767669678,
      "learning_rate": 0.0004025836200918154,
      "loss": 2.2579,
      "step": 89607
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.8195542097091675,
      "learning_rate": 0.0004025797760999248,
      "loss": 2.1885,
      "step": 89608
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.8566973209381104,
      "learning_rate": 0.0004025759320889626,
      "loss": 1.8976,
      "step": 89609
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.7781826257705688,
      "learning_rate": 0.00040257208805892974,
      "loss": 2.1607,
      "step": 89610
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.8271004557609558,
      "learning_rate": 0.0004025682440098268,
      "loss": 2.207,
      "step": 89611
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.8821575045585632,
      "learning_rate": 0.0004025643999416544,
      "loss": 2.2331,
      "step": 89612
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.7950425148010254,
      "learning_rate": 0.00040256055585441344,
      "loss": 2.2951,
      "step": 89613
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.8176964521408081,
      "learning_rate": 0.0004025567117481045,
      "loss": 2.2092,
      "step": 89614
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.8621742725372314,
      "learning_rate": 0.00040255286762272825,
      "loss": 2.2032,
      "step": 89615
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.88313227891922,
      "learning_rate": 0.0004025490234782856,
      "loss": 2.1183,
      "step": 89616
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.8424043655395508,
      "learning_rate": 0.00040254517931477703,
      "loss": 2.1195,
      "step": 89617
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.8914281129837036,
      "learning_rate": 0.0004025413351322035,
      "loss": 2.4601,
      "step": 89618
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.9742228984832764,
      "learning_rate": 0.0004025374909305655,
      "loss": 2.1964,
      "step": 89619
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.941315233707428,
      "learning_rate": 0.00040253364670986394,
      "loss": 2.4032,
      "step": 89620
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.8007906079292297,
      "learning_rate": 0.00040252980247009937,
      "loss": 2.223,
      "step": 89621
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.7708243131637573,
      "learning_rate": 0.00040252595821127263,
      "loss": 2.0437,
      "step": 89622
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.8275488615036011,
      "learning_rate": 0.0004025221139333843,
      "loss": 2.3561,
      "step": 89623
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.8278903365135193,
      "learning_rate": 0.00040251826963643523,
      "loss": 2.1339,
      "step": 89624
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.8087332844734192,
      "learning_rate": 0.00040251442532042604,
      "loss": 2.244,
      "step": 89625
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.9119881987571716,
      "learning_rate": 0.00040251058098535755,
      "loss": 2.4524,
      "step": 89626
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.8235213160514832,
      "learning_rate": 0.00040250673663123036,
      "loss": 2.0924,
      "step": 89627
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.8815165162086487,
      "learning_rate": 0.0004025028922580452,
      "loss": 2.1533,
      "step": 89628
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.8117852807044983,
      "learning_rate": 0.0004024990478658029,
      "loss": 2.3457,
      "step": 89629
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.8336823582649231,
      "learning_rate": 0.00040249520345450406,
      "loss": 2.1655,
      "step": 89630
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.8974642157554626,
      "learning_rate": 0.00040249135902414943,
      "loss": 2.1008,
      "step": 89631
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.817616879940033,
      "learning_rate": 0.00040248751457473974,
      "loss": 2.2565,
      "step": 89632
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.8006260991096497,
      "learning_rate": 0.0004024836701062756,
      "loss": 2.1079,
      "step": 89633
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.7951080799102783,
      "learning_rate": 0.0004024798256187579,
      "loss": 2.0407,
      "step": 89634
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.8984225988388062,
      "learning_rate": 0.00040247598111218727,
      "loss": 2.1007,
      "step": 89635
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.8362411856651306,
      "learning_rate": 0.00040247213658656434,
      "loss": 2.2082,
      "step": 89636
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.8499258756637573,
      "learning_rate": 0.0004024682920418899,
      "loss": 2.0403,
      "step": 89637
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.913036048412323,
      "learning_rate": 0.0004024644474781648,
      "loss": 2.1478,
      "step": 89638
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.7892507314682007,
      "learning_rate": 0.0004024606028953895,
      "loss": 2.0037,
      "step": 89639
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.7590685486793518,
      "learning_rate": 0.00040245675829356486,
      "loss": 2.5286,
      "step": 89640
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.8712815642356873,
      "learning_rate": 0.0004024529136726917,
      "loss": 2.418,
      "step": 89641
    },
    {
      "epoch": 1.17,
      "grad_norm": 1.0483297109603882,
      "learning_rate": 0.0004024490690327704,
      "loss": 2.1882,
      "step": 89642
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.8339653015136719,
      "learning_rate": 0.00040244522437380206,
      "loss": 2.0903,
      "step": 89643
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.881732702255249,
      "learning_rate": 0.00040244137969578715,
      "loss": 2.2494,
      "step": 89644
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.8221858739852905,
      "learning_rate": 0.00040243753499872643,
      "loss": 2.1417,
      "step": 89645
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.7942131161689758,
      "learning_rate": 0.0004024336902826206,
      "loss": 2.1795,
      "step": 89646
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.814318060874939,
      "learning_rate": 0.0004024298455474705,
      "loss": 2.3165,
      "step": 89647
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.8202700614929199,
      "learning_rate": 0.00040242600079327685,
      "loss": 2.1534,
      "step": 89648
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.8705912232398987,
      "learning_rate": 0.0004024221560200401,
      "loss": 2.3773,
      "step": 89649
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.8515130877494812,
      "learning_rate": 0.00040241831122776115,
      "loss": 2.1717,
      "step": 89650
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.8483632802963257,
      "learning_rate": 0.00040241446641644076,
      "loss": 2.2477,
      "step": 89651
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.7885958552360535,
      "learning_rate": 0.0004024106215860796,
      "loss": 2.1927,
      "step": 89652
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.8802783489227295,
      "learning_rate": 0.00040240677673667835,
      "loss": 2.0368,
      "step": 89653
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.8312612771987915,
      "learning_rate": 0.0004024029318682378,
      "loss": 2.0709,
      "step": 89654
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.8955448865890503,
      "learning_rate": 0.0004023990869807585,
      "loss": 1.9363,
      "step": 89655
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.9079132080078125,
      "learning_rate": 0.0004023952420742413,
      "loss": 2.2084,
      "step": 89656
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.7988458871841431,
      "learning_rate": 0.000402391397148687,
      "loss": 2.1073,
      "step": 89657
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.9621458053588867,
      "learning_rate": 0.0004023875522040961,
      "loss": 2.3904,
      "step": 89658
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.9297493100166321,
      "learning_rate": 0.00040238370724046945,
      "loss": 2.2768,
      "step": 89659
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.8631444573402405,
      "learning_rate": 0.00040237986225780776,
      "loss": 2.2381,
      "step": 89660
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.7998227477073669,
      "learning_rate": 0.00040237601725611166,
      "loss": 2.1511,
      "step": 89661
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.7805904746055603,
      "learning_rate": 0.000402372172235382,
      "loss": 2.1086,
      "step": 89662
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.7597665190696716,
      "learning_rate": 0.0004023683271956194,
      "loss": 2.1619,
      "step": 89663
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.7938222289085388,
      "learning_rate": 0.00040236448213682455,
      "loss": 2.1778,
      "step": 89664
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.8378859162330627,
      "learning_rate": 0.00040236063705899826,
      "loss": 2.1629,
      "step": 89665
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.790996253490448,
      "learning_rate": 0.0004023567919621412,
      "loss": 2.0298,
      "step": 89666
    },
    {
      "epoch": 1.17,
      "grad_norm": 1.1294609308242798,
      "learning_rate": 0.00040235294684625404,
      "loss": 2.2657,
      "step": 89667
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.7398396134376526,
      "learning_rate": 0.0004023491017113376,
      "loss": 2.0672,
      "step": 89668
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.8442475199699402,
      "learning_rate": 0.00040234525655739243,
      "loss": 2.2365,
      "step": 89669
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.8041819334030151,
      "learning_rate": 0.00040234141138441944,
      "loss": 2.107,
      "step": 89670
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.9366137385368347,
      "learning_rate": 0.00040233756619241913,
      "loss": 2.1873,
      "step": 89671
    },
    {
      "epoch": 1.17,
      "grad_norm": 1.0182076692581177,
      "learning_rate": 0.0004023337209813925,
      "loss": 2.0376,
      "step": 89672
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.8153911828994751,
      "learning_rate": 0.00040232987575134,
      "loss": 2.018,
      "step": 89673
    },
    {
      "epoch": 1.17,
      "grad_norm": 1.1059834957122803,
      "learning_rate": 0.00040232603050226244,
      "loss": 2.3481,
      "step": 89674
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.9739009737968445,
      "learning_rate": 0.00040232218523416057,
      "loss": 2.2711,
      "step": 89675
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.943240225315094,
      "learning_rate": 0.0004023183399470351,
      "loss": 2.0511,
      "step": 89676
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.8026928901672363,
      "learning_rate": 0.00040231449464088665,
      "loss": 2.0738,
      "step": 89677
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.8862236142158508,
      "learning_rate": 0.00040231064931571607,
      "loss": 2.3122,
      "step": 89678
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.8294282555580139,
      "learning_rate": 0.00040230680397152396,
      "loss": 2.2044,
      "step": 89679
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.8326637744903564,
      "learning_rate": 0.00040230295860831106,
      "loss": 1.9082,
      "step": 89680
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.8336541056632996,
      "learning_rate": 0.0004022991132260782,
      "loss": 2.2225,
      "step": 89681
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.9362660050392151,
      "learning_rate": 0.0004022952678248259,
      "loss": 1.8867,
      "step": 89682
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.9479526281356812,
      "learning_rate": 0.000402291422404555,
      "loss": 2.2549,
      "step": 89683
    },
    {
      "epoch": 1.17,
      "grad_norm": 1.150664210319519,
      "learning_rate": 0.00040228757696526623,
      "loss": 2.2793,
      "step": 89684
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.8066385388374329,
      "learning_rate": 0.00040228373150696034,
      "loss": 2.234,
      "step": 89685
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.8472884893417358,
      "learning_rate": 0.0004022798860296378,
      "loss": 1.9306,
      "step": 89686
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.7674321532249451,
      "learning_rate": 0.00040227604053329956,
      "loss": 1.887,
      "step": 89687
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.81967693567276,
      "learning_rate": 0.0004022721950179464,
      "loss": 2.0987,
      "step": 89688
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.7494006752967834,
      "learning_rate": 0.0004022683494835788,
      "loss": 2.1725,
      "step": 89689
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.944063127040863,
      "learning_rate": 0.00040226450393019753,
      "loss": 2.0752,
      "step": 89690
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.8297724723815918,
      "learning_rate": 0.00040226065835780343,
      "loss": 2.1318,
      "step": 89691
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.822110652923584,
      "learning_rate": 0.0004022568127663972,
      "loss": 2.2544,
      "step": 89692
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.924208402633667,
      "learning_rate": 0.0004022529671559794,
      "loss": 2.1479,
      "step": 89693
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.8063805103302002,
      "learning_rate": 0.0004022491215265509,
      "loss": 2.1765,
      "step": 89694
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.7915290594100952,
      "learning_rate": 0.00040224527587811226,
      "loss": 2.311,
      "step": 89695
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.8351384401321411,
      "learning_rate": 0.00040224143021066447,
      "loss": 2.2337,
      "step": 89696
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.8668474555015564,
      "learning_rate": 0.00040223758452420796,
      "loss": 2.0122,
      "step": 89697
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.8350956439971924,
      "learning_rate": 0.0004022337388187435,
      "loss": 2.3802,
      "step": 89698
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.8094944953918457,
      "learning_rate": 0.00040222989309427197,
      "loss": 2.2948,
      "step": 89699
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.7923719882965088,
      "learning_rate": 0.0004022260473507939,
      "loss": 2.3025,
      "step": 89700
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.7567318677902222,
      "learning_rate": 0.0004022222015883101,
      "loss": 2.203,
      "step": 89701
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.7794018983840942,
      "learning_rate": 0.0004022183558068214,
      "loss": 2.0492,
      "step": 89702
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.8294879198074341,
      "learning_rate": 0.00040221451000632815,
      "loss": 2.1788,
      "step": 89703
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.8661878705024719,
      "learning_rate": 0.00040221066418683136,
      "loss": 2.0584,
      "step": 89704
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.7973580956459045,
      "learning_rate": 0.00040220681834833176,
      "loss": 2.1523,
      "step": 89705
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.781960129737854,
      "learning_rate": 0.0004022029724908299,
      "loss": 2.3294,
      "step": 89706
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.7861448526382446,
      "learning_rate": 0.00040219912661432656,
      "loss": 2.0063,
      "step": 89707
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.8461706042289734,
      "learning_rate": 0.0004021952807188226,
      "loss": 2.0205,
      "step": 89708
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.7938281893730164,
      "learning_rate": 0.0004021914348043185,
      "loss": 2.1892,
      "step": 89709
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.8065548539161682,
      "learning_rate": 0.00040218758887081515,
      "loss": 2.2226,
      "step": 89710
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.8281455039978027,
      "learning_rate": 0.0004021837429183132,
      "loss": 2.2851,
      "step": 89711
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.7856205701828003,
      "learning_rate": 0.0004021798969468133,
      "loss": 2.1193,
      "step": 89712
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.8260852694511414,
      "learning_rate": 0.00040217605095631617,
      "loss": 2.0877,
      "step": 89713
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.7820552587509155,
      "learning_rate": 0.0004021722049468227,
      "loss": 2.1514,
      "step": 89714
    },
    {
      "epoch": 1.17,
      "grad_norm": 1.0176454782485962,
      "learning_rate": 0.0004021683589183335,
      "loss": 2.2587,
      "step": 89715
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.8237738609313965,
      "learning_rate": 0.0004021645128708492,
      "loss": 2.1058,
      "step": 89716
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.8257578611373901,
      "learning_rate": 0.00040216066680437055,
      "loss": 2.0282,
      "step": 89717
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.9067708253860474,
      "learning_rate": 0.00040215682071889847,
      "loss": 2.4708,
      "step": 89718
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.840582549571991,
      "learning_rate": 0.00040215297461443334,
      "loss": 1.9783,
      "step": 89719
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.8566523790359497,
      "learning_rate": 0.0004021491284909761,
      "loss": 2.2365,
      "step": 89720
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.8386002779006958,
      "learning_rate": 0.00040214528234852744,
      "loss": 1.9612,
      "step": 89721
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.7892716526985168,
      "learning_rate": 0.00040214143618708803,
      "loss": 2.1055,
      "step": 89722
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.8258928656578064,
      "learning_rate": 0.0004021375900066586,
      "loss": 2.2152,
      "step": 89723
    },
    {
      "epoch": 1.17,
      "grad_norm": 1.339032530784607,
      "learning_rate": 0.00040213374380723986,
      "loss": 2.2423,
      "step": 89724
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.8053070902824402,
      "learning_rate": 0.00040212989758883256,
      "loss": 2.0477,
      "step": 89725
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.8407571315765381,
      "learning_rate": 0.0004021260513514373,
      "loss": 2.3624,
      "step": 89726
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.7424805760383606,
      "learning_rate": 0.000402122205095055,
      "loss": 2.1852,
      "step": 89727
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.7763503193855286,
      "learning_rate": 0.0004021183588196861,
      "loss": 2.2713,
      "step": 89728
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.8985331654548645,
      "learning_rate": 0.0004021145125253316,
      "loss": 2.3338,
      "step": 89729
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.8809018135070801,
      "learning_rate": 0.0004021106662119921,
      "loss": 2.0185,
      "step": 89730
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.8353050351142883,
      "learning_rate": 0.0004021068198796682,
      "loss": 2.3697,
      "step": 89731
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.8367862105369568,
      "learning_rate": 0.00040210297352836073,
      "loss": 2.1385,
      "step": 89732
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.8769671320915222,
      "learning_rate": 0.0004020991271580704,
      "loss": 2.1879,
      "step": 89733
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.8488951325416565,
      "learning_rate": 0.00040209528076879796,
      "loss": 2.178,
      "step": 89734
    },
    {
      "epoch": 1.17,
      "grad_norm": 1.09734046459198,
      "learning_rate": 0.00040209143436054404,
      "loss": 2.0818,
      "step": 89735
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.9076626300811768,
      "learning_rate": 0.0004020875879333094,
      "loss": 2.1086,
      "step": 89736
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.812107503414154,
      "learning_rate": 0.0004020837414870947,
      "loss": 1.9552,
      "step": 89737
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.8514150977134705,
      "learning_rate": 0.0004020798950219008,
      "loss": 2.1081,
      "step": 89738
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.8570624589920044,
      "learning_rate": 0.0004020760485377283,
      "loss": 2.0934,
      "step": 89739
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.8021628260612488,
      "learning_rate": 0.00040207220203457793,
      "loss": 2.3989,
      "step": 89740
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.8064224720001221,
      "learning_rate": 0.00040206835551245037,
      "loss": 2.2018,
      "step": 89741
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.9040669202804565,
      "learning_rate": 0.00040206450897134645,
      "loss": 2.1299,
      "step": 89742
    },
    {
      "epoch": 1.17,
      "grad_norm": 1.0165717601776123,
      "learning_rate": 0.00040206066241126677,
      "loss": 2.0084,
      "step": 89743
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.8155327439308167,
      "learning_rate": 0.00040205681583221204,
      "loss": 2.2245,
      "step": 89744
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.8112696409225464,
      "learning_rate": 0.0004020529692341831,
      "loss": 2.5597,
      "step": 89745
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.9558205604553223,
      "learning_rate": 0.00040204912261718054,
      "loss": 2.205,
      "step": 89746
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.9239606261253357,
      "learning_rate": 0.0004020452759812051,
      "loss": 2.1334,
      "step": 89747
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.8512933850288391,
      "learning_rate": 0.00040204142932625755,
      "loss": 2.2852,
      "step": 89748
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.8925155997276306,
      "learning_rate": 0.00040203758265233854,
      "loss": 2.0521,
      "step": 89749
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.9308164119720459,
      "learning_rate": 0.00040203373595944886,
      "loss": 2.0188,
      "step": 89750
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.790473222732544,
      "learning_rate": 0.00040202988924758923,
      "loss": 2.302,
      "step": 89751
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.8381987810134888,
      "learning_rate": 0.00040202604251676023,
      "loss": 2.1102,
      "step": 89752
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.8522266149520874,
      "learning_rate": 0.0004020221957669626,
      "loss": 2.3065,
      "step": 89753
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.8115983605384827,
      "learning_rate": 0.00040201834899819734,
      "loss": 2.2767,
      "step": 89754
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.8904712200164795,
      "learning_rate": 0.00040201450221046475,
      "loss": 2.2108,
      "step": 89755
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.8444280624389648,
      "learning_rate": 0.00040201065540376577,
      "loss": 2.058,
      "step": 89756
    },
    {
      "epoch": 1.17,
      "grad_norm": 1.010693073272705,
      "learning_rate": 0.0004020068085781012,
      "loss": 2.1126,
      "step": 89757
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.7583610415458679,
      "learning_rate": 0.00040200296173347155,
      "loss": 2.2928,
      "step": 89758
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.8020990490913391,
      "learning_rate": 0.00040199911486987755,
      "loss": 2.0104,
      "step": 89759
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.8214738368988037,
      "learning_rate": 0.0004019952679873201,
      "loss": 2.054,
      "step": 89760
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.836967408657074,
      "learning_rate": 0.0004019914210857998,
      "loss": 2.0885,
      "step": 89761
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.817922830581665,
      "learning_rate": 0.00040198757416531726,
      "loss": 2.119,
      "step": 89762
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.9511311650276184,
      "learning_rate": 0.00040198372722587345,
      "loss": 2.153,
      "step": 89763
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.9093800783157349,
      "learning_rate": 0.00040197988026746884,
      "loss": 2.1147,
      "step": 89764
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.905924916267395,
      "learning_rate": 0.0004019760332901043,
      "loss": 2.2127,
      "step": 89765
    },
    {
      "epoch": 1.17,
      "grad_norm": 1.0810543298721313,
      "learning_rate": 0.0004019721862937805,
      "loss": 2.2423,
      "step": 89766
    },
    {
      "epoch": 1.17,
      "grad_norm": 1.0415544509887695,
      "learning_rate": 0.0004019683392784981,
      "loss": 2.1201,
      "step": 89767
    },
    {
      "epoch": 1.17,
      "grad_norm": 1.1409099102020264,
      "learning_rate": 0.0004019644922442578,
      "loss": 2.1444,
      "step": 89768
    },
    {
      "epoch": 1.17,
      "grad_norm": 1.1064966917037964,
      "learning_rate": 0.0004019606451910605,
      "loss": 2.0461,
      "step": 89769
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.8776496648788452,
      "learning_rate": 0.0004019567981189067,
      "loss": 2.0322,
      "step": 89770
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.8241593241691589,
      "learning_rate": 0.00040195295102779733,
      "loss": 2.2523,
      "step": 89771
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.8728845715522766,
      "learning_rate": 0.0004019491039177328,
      "loss": 2.0653,
      "step": 89772
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.889625072479248,
      "learning_rate": 0.00040194525678871417,
      "loss": 2.2659,
      "step": 89773
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.9484096765518188,
      "learning_rate": 0.0004019414096407419,
      "loss": 2.2535,
      "step": 89774
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.8001035451889038,
      "learning_rate": 0.0004019375624738168,
      "loss": 2.1094,
      "step": 89775
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.7677392959594727,
      "learning_rate": 0.00040193371528793965,
      "loss": 2.0502,
      "step": 89776
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.8918348550796509,
      "learning_rate": 0.0004019298680831111,
      "loss": 2.2873,
      "step": 89777
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.7854282259941101,
      "learning_rate": 0.0004019260208593318,
      "loss": 2.3936,
      "step": 89778
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.8312245607376099,
      "learning_rate": 0.0004019221736166025,
      "loss": 2.2416,
      "step": 89779
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.8098138570785522,
      "learning_rate": 0.00040191832635492407,
      "loss": 2.2085,
      "step": 89780
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.8067868947982788,
      "learning_rate": 0.000401914479074297,
      "loss": 2.2195,
      "step": 89781
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.8730314373970032,
      "learning_rate": 0.0004019106317747221,
      "loss": 2.1788,
      "step": 89782
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.8817920088768005,
      "learning_rate": 0.0004019067844562001,
      "loss": 2.315,
      "step": 89783
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.9019905924797058,
      "learning_rate": 0.0004019029371187318,
      "loss": 2.3445,
      "step": 89784
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.7931225299835205,
      "learning_rate": 0.0004018990897623177,
      "loss": 2.325,
      "step": 89785
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.8314834833145142,
      "learning_rate": 0.0004018952423869587,
      "loss": 2.2937,
      "step": 89786
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.9389349818229675,
      "learning_rate": 0.0004018913949926555,
      "loss": 2.125,
      "step": 89787
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.855936586856842,
      "learning_rate": 0.0004018875475794086,
      "loss": 2.1234,
      "step": 89788
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.9203620553016663,
      "learning_rate": 0.00040188370014721897,
      "loss": 2.1249,
      "step": 89789
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.7917173504829407,
      "learning_rate": 0.00040187985269608725,
      "loss": 2.202,
      "step": 89790
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.8770319223403931,
      "learning_rate": 0.0004018760052260142,
      "loss": 2.2835,
      "step": 89791
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.8808028697967529,
      "learning_rate": 0.00040187215773700033,
      "loss": 2.1186,
      "step": 89792
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.7981262803077698,
      "learning_rate": 0.00040186831022904655,
      "loss": 2.1835,
      "step": 89793
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.8681303262710571,
      "learning_rate": 0.0004018644627021537,
      "loss": 2.1405,
      "step": 89794
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.8866055607795715,
      "learning_rate": 0.0004018606151563221,
      "loss": 2.2089,
      "step": 89795
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.8097560405731201,
      "learning_rate": 0.0004018567675915528,
      "loss": 2.1803,
      "step": 89796
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.9133076071739197,
      "learning_rate": 0.00040185292000784645,
      "loss": 2.0053,
      "step": 89797
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.9570335745811462,
      "learning_rate": 0.00040184907240520365,
      "loss": 2.2714,
      "step": 89798
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.8656386733055115,
      "learning_rate": 0.00040184522478362516,
      "loss": 2.0522,
      "step": 89799
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.8791350722312927,
      "learning_rate": 0.0004018413771431118,
      "loss": 2.3988,
      "step": 89800
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.8888475298881531,
      "learning_rate": 0.0004018375294836641,
      "loss": 2.2643,
      "step": 89801
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.9023309350013733,
      "learning_rate": 0.0004018336818052829,
      "loss": 2.2169,
      "step": 89802
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.7397441864013672,
      "learning_rate": 0.000401829834107969,
      "loss": 2.1249,
      "step": 89803
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.7692099213600159,
      "learning_rate": 0.00040182598639172294,
      "loss": 1.9678,
      "step": 89804
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.8213104009628296,
      "learning_rate": 0.0004018221386565455,
      "loss": 2.2997,
      "step": 89805
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.938219428062439,
      "learning_rate": 0.0004018182909024374,
      "loss": 2.2596,
      "step": 89806
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.8931662440299988,
      "learning_rate": 0.00040181444312939934,
      "loss": 2.3411,
      "step": 89807
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.7795092463493347,
      "learning_rate": 0.0004018105953374321,
      "loss": 2.1738,
      "step": 89808
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.8542953729629517,
      "learning_rate": 0.0004018067475265364,
      "loss": 2.037,
      "step": 89809
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.8512768149375916,
      "learning_rate": 0.00040180289969671284,
      "loss": 2.2159,
      "step": 89810
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.8456195592880249,
      "learning_rate": 0.0004017990518479621,
      "loss": 2.3124,
      "step": 89811
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.8470293283462524,
      "learning_rate": 0.0004017952039802852,
      "loss": 2.255,
      "step": 89812
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.9544427394866943,
      "learning_rate": 0.0004017913560936825,
      "loss": 2.1058,
      "step": 89813
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.8261593580245972,
      "learning_rate": 0.0004017875081881549,
      "loss": 2.1037,
      "step": 89814
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.8643370270729065,
      "learning_rate": 0.0004017836602637031,
      "loss": 2.3159,
      "step": 89815
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.9679250717163086,
      "learning_rate": 0.00040177981232032776,
      "loss": 2.1285,
      "step": 89816
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.8705556392669678,
      "learning_rate": 0.0004017759643580296,
      "loss": 2.256,
      "step": 89817
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.9345922470092773,
      "learning_rate": 0.0004017721163768095,
      "loss": 2.3088,
      "step": 89818
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.8132287263870239,
      "learning_rate": 0.00040176826837666794,
      "loss": 2.0849,
      "step": 89819
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.8250192403793335,
      "learning_rate": 0.00040176442035760575,
      "loss": 2.3019,
      "step": 89820
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.8529081344604492,
      "learning_rate": 0.00040176057231962363,
      "loss": 2.2393,
      "step": 89821
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.8379052877426147,
      "learning_rate": 0.0004017567242627223,
      "loss": 2.2165,
      "step": 89822
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.8188390731811523,
      "learning_rate": 0.0004017528761869025,
      "loss": 2.1544,
      "step": 89823
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.8256063461303711,
      "learning_rate": 0.00040174902809216495,
      "loss": 2.3356,
      "step": 89824
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.939643144607544,
      "learning_rate": 0.0004017451799785103,
      "loss": 2.1334,
      "step": 89825
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.7845926284790039,
      "learning_rate": 0.00040174133184593924,
      "loss": 2.1363,
      "step": 89826
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.8604543805122375,
      "learning_rate": 0.0004017374836944527,
      "loss": 2.3348,
      "step": 89827
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.8050827980041504,
      "learning_rate": 0.00040173363552405107,
      "loss": 2.4114,
      "step": 89828
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.8121865391731262,
      "learning_rate": 0.0004017297873347353,
      "loss": 2.0545,
      "step": 89829
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.9292967915534973,
      "learning_rate": 0.0004017259391265061,
      "loss": 2.3373,
      "step": 89830
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.8200895190238953,
      "learning_rate": 0.00040172209089936407,
      "loss": 1.9493,
      "step": 89831
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.77923184633255,
      "learning_rate": 0.0004017182426533099,
      "loss": 2.0824,
      "step": 89832
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.9286708831787109,
      "learning_rate": 0.0004017143943883446,
      "loss": 2.1783,
      "step": 89833
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.7882888317108154,
      "learning_rate": 0.00040171054610446845,
      "loss": 1.9636,
      "step": 89834
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.8814469575881958,
      "learning_rate": 0.0004017066978016825,
      "loss": 2.1263,
      "step": 89835
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.8428670167922974,
      "learning_rate": 0.00040170284947998737,
      "loss": 2.1571,
      "step": 89836
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.8708901405334473,
      "learning_rate": 0.0004016990011393837,
      "loss": 2.2268,
      "step": 89837
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.7990095019340515,
      "learning_rate": 0.00040169515277987234,
      "loss": 2.2285,
      "step": 89838
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.7982138991355896,
      "learning_rate": 0.00040169130440145394,
      "loss": 2.1317,
      "step": 89839
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.8418697714805603,
      "learning_rate": 0.0004016874560041291,
      "loss": 2.3434,
      "step": 89840
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.8532447218894958,
      "learning_rate": 0.0004016836075878988,
      "loss": 2.3665,
      "step": 89841
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.8517933487892151,
      "learning_rate": 0.0004016797591527635,
      "loss": 2.265,
      "step": 89842
    },
    {
      "epoch": 1.17,
      "grad_norm": 1.0412895679473877,
      "learning_rate": 0.00040167591069872394,
      "loss": 2.473,
      "step": 89843
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.7608902454376221,
      "learning_rate": 0.00040167206222578095,
      "loss": 2.0539,
      "step": 89844
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.8533629775047302,
      "learning_rate": 0.0004016682137339354,
      "loss": 2.0199,
      "step": 89845
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.7643757462501526,
      "learning_rate": 0.00040166436522318756,
      "loss": 2.1668,
      "step": 89846
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.7598844766616821,
      "learning_rate": 0.00040166051669353846,
      "loss": 2.14,
      "step": 89847
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.8827305436134338,
      "learning_rate": 0.00040165666814498874,
      "loss": 2.0483,
      "step": 89848
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.8531562685966492,
      "learning_rate": 0.0004016528195775392,
      "loss": 2.0676,
      "step": 89849
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.9093161225318909,
      "learning_rate": 0.00040164897099119045,
      "loss": 2.2037,
      "step": 89850
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.7803094387054443,
      "learning_rate": 0.00040164512238594324,
      "loss": 1.984,
      "step": 89851
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.8261362910270691,
      "learning_rate": 0.0004016412737617983,
      "loss": 2.2729,
      "step": 89852
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.8909466862678528,
      "learning_rate": 0.00040163742511875627,
      "loss": 2.368,
      "step": 89853
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.8629983067512512,
      "learning_rate": 0.0004016335764568179,
      "loss": 2.0957,
      "step": 89854
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.8882296085357666,
      "learning_rate": 0.00040162972777598404,
      "loss": 2.1102,
      "step": 89855
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.9605945944786072,
      "learning_rate": 0.0004016258790762553,
      "loss": 2.2491,
      "step": 89856
    },
    {
      "epoch": 1.17,
      "grad_norm": 1.0423901081085205,
      "learning_rate": 0.00040162203035763223,
      "loss": 2.2029,
      "step": 89857
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.8136914968490601,
      "learning_rate": 0.0004016181816201159,
      "loss": 2.1297,
      "step": 89858
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.8576812744140625,
      "learning_rate": 0.0004016143328637067,
      "loss": 2.2463,
      "step": 89859
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.8366291522979736,
      "learning_rate": 0.00040161048408840554,
      "loss": 2.3528,
      "step": 89860
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.9166067838668823,
      "learning_rate": 0.00040160663529421306,
      "loss": 2.2263,
      "step": 89861
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.8307201266288757,
      "learning_rate": 0.00040160278648112993,
      "loss": 2.1488,
      "step": 89862
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.8535966277122498,
      "learning_rate": 0.000401598937649157,
      "loss": 2.1668,
      "step": 89863
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.8915811777114868,
      "learning_rate": 0.00040159508879829487,
      "loss": 2.2805,
      "step": 89864
    },
    {
      "epoch": 1.17,
      "grad_norm": 1.0033042430877686,
      "learning_rate": 0.00040159123992854434,
      "loss": 2.1701,
      "step": 89865
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.8039798140525818,
      "learning_rate": 0.000401587391039906,
      "loss": 1.9957,
      "step": 89866
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.7806795835494995,
      "learning_rate": 0.0004015835421323807,
      "loss": 2.0701,
      "step": 89867
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.8952734470367432,
      "learning_rate": 0.00040157969320596913,
      "loss": 2.1761,
      "step": 89868
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.8851902484893799,
      "learning_rate": 0.000401575844260672,
      "loss": 2.1673,
      "step": 89869
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.94231116771698,
      "learning_rate": 0.00040157199529648985,
      "loss": 2.0807,
      "step": 89870
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.9362662434577942,
      "learning_rate": 0.00040156814631342366,
      "loss": 2.1254,
      "step": 89871
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.9335155487060547,
      "learning_rate": 0.00040156429731147403,
      "loss": 1.993,
      "step": 89872
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.9515342116355896,
      "learning_rate": 0.00040156044829064167,
      "loss": 2.1181,
      "step": 89873
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.8250282406806946,
      "learning_rate": 0.00040155659925092727,
      "loss": 2.2231,
      "step": 89874
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.8871235251426697,
      "learning_rate": 0.00040155275019233166,
      "loss": 2.1714,
      "step": 89875
    },
    {
      "epoch": 1.17,
      "grad_norm": 1.645865559577942,
      "learning_rate": 0.0004015489011148554,
      "loss": 2.3136,
      "step": 89876
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.8691148161888123,
      "learning_rate": 0.0004015450520184993,
      "loss": 2.0953,
      "step": 89877
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.9133558869361877,
      "learning_rate": 0.0004015412029032641,
      "loss": 2.1871,
      "step": 89878
    },
    {
      "epoch": 1.17,
      "grad_norm": 1.033936858177185,
      "learning_rate": 0.00040153735376915046,
      "loss": 2.1494,
      "step": 89879
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.7905471920967102,
      "learning_rate": 0.0004015335046161591,
      "loss": 2.0505,
      "step": 89880
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.8223832845687866,
      "learning_rate": 0.0004015296554442907,
      "loss": 2.2083,
      "step": 89881
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.9503321647644043,
      "learning_rate": 0.00040152580625354615,
      "loss": 1.8777,
      "step": 89882
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.8608313202857971,
      "learning_rate": 0.00040152195704392594,
      "loss": 2.0395,
      "step": 89883
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.8753790259361267,
      "learning_rate": 0.0004015181078154308,
      "loss": 2.2025,
      "step": 89884
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.8266850113868713,
      "learning_rate": 0.0004015142585680617,
      "loss": 2.3605,
      "step": 89885
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.8434520363807678,
      "learning_rate": 0.0004015104093018191,
      "loss": 2.3764,
      "step": 89886
    },
    {
      "epoch": 1.17,
      "grad_norm": 1.0151488780975342,
      "learning_rate": 0.00040150656001670373,
      "loss": 2.2011,
      "step": 89887
    },
    {
      "epoch": 1.17,
      "grad_norm": 1.0669153928756714,
      "learning_rate": 0.0004015027107127165,
      "loss": 2.2698,
      "step": 89888
    },
    {
      "epoch": 1.17,
      "grad_norm": 1.1868641376495361,
      "learning_rate": 0.00040149886138985796,
      "loss": 2.3134,
      "step": 89889
    },
    {
      "epoch": 1.17,
      "grad_norm": 1.1127290725708008,
      "learning_rate": 0.00040149501204812886,
      "loss": 2.2137,
      "step": 89890
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.9163381457328796,
      "learning_rate": 0.0004014911626875299,
      "loss": 2.1996,
      "step": 89891
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.836497962474823,
      "learning_rate": 0.00040148731330806185,
      "loss": 2.3046,
      "step": 89892
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.8827043771743774,
      "learning_rate": 0.00040148346390972536,
      "loss": 2.0536,
      "step": 89893
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.8960505723953247,
      "learning_rate": 0.00040147961449252126,
      "loss": 2.1625,
      "step": 89894
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.7681379318237305,
      "learning_rate": 0.0004014757650564501,
      "loss": 2.2872,
      "step": 89895
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.8395863771438599,
      "learning_rate": 0.00040147191560151266,
      "loss": 2.2654,
      "step": 89896
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.7859379649162292,
      "learning_rate": 0.00040146806612770974,
      "loss": 2.0755,
      "step": 89897
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.8756364583969116,
      "learning_rate": 0.0004014642166350419,
      "loss": 2.2351,
      "step": 89898
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.7782276272773743,
      "learning_rate": 0.00040146036712351005,
      "loss": 2.1291,
      "step": 89899
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.789563000202179,
      "learning_rate": 0.0004014565175931148,
      "loss": 2.0986,
      "step": 89900
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.8733699917793274,
      "learning_rate": 0.0004014526680438568,
      "loss": 2.1652,
      "step": 89901
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.839073657989502,
      "learning_rate": 0.00040144881847573683,
      "loss": 2.422,
      "step": 89902
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.8153976202011108,
      "learning_rate": 0.0004014449688887557,
      "loss": 2.3187,
      "step": 89903
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.8683033585548401,
      "learning_rate": 0.00040144111928291394,
      "loss": 2.1019,
      "step": 89904
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.8226991891860962,
      "learning_rate": 0.0004014372696582124,
      "loss": 2.0244,
      "step": 89905
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.8596817851066589,
      "learning_rate": 0.0004014334200146518,
      "loss": 2.2316,
      "step": 89906
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.8415603041648865,
      "learning_rate": 0.0004014295703522327,
      "loss": 2.1976,
      "step": 89907
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.8118256330490112,
      "learning_rate": 0.000401425720670956,
      "loss": 2.2403,
      "step": 89908
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.9851109981536865,
      "learning_rate": 0.0004014218709708224,
      "loss": 2.4482,
      "step": 89909
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.8306847214698792,
      "learning_rate": 0.00040141802125183244,
      "loss": 2.125,
      "step": 89910
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.8277236819267273,
      "learning_rate": 0.00040141417151398704,
      "loss": 2.0284,
      "step": 89911
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.8161390423774719,
      "learning_rate": 0.0004014103217572869,
      "loss": 2.2429,
      "step": 89912
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.8381387591362,
      "learning_rate": 0.00040140647198173255,
      "loss": 2.3964,
      "step": 89913
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.7976883053779602,
      "learning_rate": 0.00040140262218732475,
      "loss": 2.0009,
      "step": 89914
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.8566284775733948,
      "learning_rate": 0.0004013987723740644,
      "loss": 2.0132,
      "step": 89915
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.8988139629364014,
      "learning_rate": 0.00040139492254195213,
      "loss": 2.32,
      "step": 89916
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.8467575907707214,
      "learning_rate": 0.00040139107269098856,
      "loss": 2.1611,
      "step": 89917
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.8398157954216003,
      "learning_rate": 0.0004013872228211745,
      "loss": 2.1098,
      "step": 89918
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.8883293271064758,
      "learning_rate": 0.00040138337293251067,
      "loss": 2.3227,
      "step": 89919
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.8485003113746643,
      "learning_rate": 0.0004013795230249977,
      "loss": 2.2528,
      "step": 89920
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.9270059466362,
      "learning_rate": 0.0004013756730986364,
      "loss": 2.4062,
      "step": 89921
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.8712169528007507,
      "learning_rate": 0.0004013718231534275,
      "loss": 2.1043,
      "step": 89922
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.941482424736023,
      "learning_rate": 0.0004013679731893716,
      "loss": 1.943,
      "step": 89923
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.8618262410163879,
      "learning_rate": 0.00040136412320646947,
      "loss": 2.2397,
      "step": 89924
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.927204430103302,
      "learning_rate": 0.0004013602732047219,
      "loss": 2.0001,
      "step": 89925
    },
    {
      "epoch": 1.17,
      "grad_norm": 1.265752911567688,
      "learning_rate": 0.00040135642318412946,
      "loss": 1.9518,
      "step": 89926
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.854219913482666,
      "learning_rate": 0.000401352573144693,
      "loss": 1.9479,
      "step": 89927
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.8239572644233704,
      "learning_rate": 0.00040134872308641317,
      "loss": 2.2887,
      "step": 89928
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.8307461738586426,
      "learning_rate": 0.00040134487300929066,
      "loss": 2.3446,
      "step": 89929
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.9067736864089966,
      "learning_rate": 0.00040134102291332623,
      "loss": 2.081,
      "step": 89930
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.8131862878799438,
      "learning_rate": 0.00040133717279852065,
      "loss": 2.1847,
      "step": 89931
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.7993960380554199,
      "learning_rate": 0.00040133332266487456,
      "loss": 2.3475,
      "step": 89932
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.9073979258537292,
      "learning_rate": 0.00040132947251238866,
      "loss": 2.0882,
      "step": 89933
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.8408566117286682,
      "learning_rate": 0.0004013256223410638,
      "loss": 2.0019,
      "step": 89934
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.8490335941314697,
      "learning_rate": 0.0004013217721509005,
      "loss": 2.1673,
      "step": 89935
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.8276644945144653,
      "learning_rate": 0.00040131792194189955,
      "loss": 2.2209,
      "step": 89936
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.9132882952690125,
      "learning_rate": 0.00040131407171406177,
      "loss": 2.1043,
      "step": 89937
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.8078615665435791,
      "learning_rate": 0.0004013102214673877,
      "loss": 2.2604,
      "step": 89938
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.7872048020362854,
      "learning_rate": 0.0004013063712018782,
      "loss": 2.2015,
      "step": 89939
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.9404663443565369,
      "learning_rate": 0.0004013025209175339,
      "loss": 1.9648,
      "step": 89940
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.961464524269104,
      "learning_rate": 0.0004012986706143556,
      "loss": 2.2197,
      "step": 89941
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.8150050640106201,
      "learning_rate": 0.00040129482029234394,
      "loss": 2.3271,
      "step": 89942
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.7472123503684998,
      "learning_rate": 0.0004012909699514997,
      "loss": 2.2911,
      "step": 89943
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.8196545243263245,
      "learning_rate": 0.0004012871195918235,
      "loss": 2.0993,
      "step": 89944
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.759236752986908,
      "learning_rate": 0.0004012832692133161,
      "loss": 2.2425,
      "step": 89945
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.9887006282806396,
      "learning_rate": 0.0004012794188159783,
      "loss": 2.2825,
      "step": 89946
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.7746649384498596,
      "learning_rate": 0.00040127556839981065,
      "loss": 2.0123,
      "step": 89947
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.8219746947288513,
      "learning_rate": 0.000401271717964814,
      "loss": 2.1905,
      "step": 89948
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.9657906293869019,
      "learning_rate": 0.000401267867510989,
      "loss": 2.1542,
      "step": 89949
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.8391041159629822,
      "learning_rate": 0.0004012640170383365,
      "loss": 2.2005,
      "step": 89950
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.7650861144065857,
      "learning_rate": 0.00040126016654685705,
      "loss": 2.1669,
      "step": 89951
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.8569956421852112,
      "learning_rate": 0.0004012563160365514,
      "loss": 2.4212,
      "step": 89952
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.7886362671852112,
      "learning_rate": 0.0004012524655074203,
      "loss": 2.2694,
      "step": 89953
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.7839337587356567,
      "learning_rate": 0.00040124861495946447,
      "loss": 2.2604,
      "step": 89954
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.8575893640518188,
      "learning_rate": 0.0004012447643926846,
      "loss": 2.3539,
      "step": 89955
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.8919520378112793,
      "learning_rate": 0.00040124091380708136,
      "loss": 2.2168,
      "step": 89956
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.7842797636985779,
      "learning_rate": 0.00040123706320265565,
      "loss": 1.921,
      "step": 89957
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.8673413395881653,
      "learning_rate": 0.0004012332125794079,
      "loss": 2.252,
      "step": 89958
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.8979164361953735,
      "learning_rate": 0.0004012293619373391,
      "loss": 2.0312,
      "step": 89959
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.9310637712478638,
      "learning_rate": 0.0004012255112764499,
      "loss": 2.3367,
      "step": 89960
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.8393598198890686,
      "learning_rate": 0.0004012216605967408,
      "loss": 2.2908,
      "step": 89961
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.8102156519889832,
      "learning_rate": 0.0004012178098982128,
      "loss": 2.2775,
      "step": 89962
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.9380130171775818,
      "learning_rate": 0.00040121395918086644,
      "loss": 2.2999,
      "step": 89963
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.8694196343421936,
      "learning_rate": 0.00040121010844470253,
      "loss": 2.2171,
      "step": 89964
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.8137558102607727,
      "learning_rate": 0.00040120625768972175,
      "loss": 2.1311,
      "step": 89965
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.7894306778907776,
      "learning_rate": 0.00040120240691592476,
      "loss": 2.2228,
      "step": 89966
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.8575598001480103,
      "learning_rate": 0.00040119855612331246,
      "loss": 2.2333,
      "step": 89967
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.767975926399231,
      "learning_rate": 0.00040119470531188535,
      "loss": 2.1623,
      "step": 89968
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.7599074244499207,
      "learning_rate": 0.0004011908544816442,
      "loss": 2.1531,
      "step": 89969
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.9188701510429382,
      "learning_rate": 0.00040118700363258984,
      "loss": 2.1198,
      "step": 89970
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.894981861114502,
      "learning_rate": 0.00040118315276472285,
      "loss": 2.1036,
      "step": 89971
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.7983301281929016,
      "learning_rate": 0.00040117930187804404,
      "loss": 2.1077,
      "step": 89972
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.7440007328987122,
      "learning_rate": 0.0004011754509725541,
      "loss": 2.1106,
      "step": 89973
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.8674570322036743,
      "learning_rate": 0.00040117160004825373,
      "loss": 2.3435,
      "step": 89974
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.8126793503761292,
      "learning_rate": 0.0004011677491051436,
      "loss": 2.0937,
      "step": 89975
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.9149736762046814,
      "learning_rate": 0.0004011638981432245,
      "loss": 2.1766,
      "step": 89976
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.8875184059143066,
      "learning_rate": 0.0004011600471624971,
      "loss": 2.007,
      "step": 89977
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.8486283421516418,
      "learning_rate": 0.00040115619616296213,
      "loss": 2.2913,
      "step": 89978
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.9885828495025635,
      "learning_rate": 0.0004011523451446204,
      "loss": 2.2901,
      "step": 89979
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.8069902062416077,
      "learning_rate": 0.0004011484941074725,
      "loss": 2.1871,
      "step": 89980
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.7906880974769592,
      "learning_rate": 0.0004011446430515191,
      "loss": 2.0444,
      "step": 89981
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.8697780966758728,
      "learning_rate": 0.00040114079197676105,
      "loss": 2.2448,
      "step": 89982
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.8324502110481262,
      "learning_rate": 0.00040113694088319906,
      "loss": 2.3487,
      "step": 89983
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.7788287997245789,
      "learning_rate": 0.0004011330897708338,
      "loss": 2.1737,
      "step": 89984
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.7503963112831116,
      "learning_rate": 0.00040112923863966593,
      "loss": 1.9745,
      "step": 89985
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.79440838098526,
      "learning_rate": 0.0004011253874896963,
      "loss": 2.3454,
      "step": 89986
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.8781764507293701,
      "learning_rate": 0.0004011215363209255,
      "loss": 2.1262,
      "step": 89987
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.914373517036438,
      "learning_rate": 0.00040111768513335434,
      "loss": 2.1942,
      "step": 89988
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.9055025577545166,
      "learning_rate": 0.00040111383392698344,
      "loss": 2.059,
      "step": 89989
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.8612444996833801,
      "learning_rate": 0.0004011099827018136,
      "loss": 1.9291,
      "step": 89990
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.8117057681083679,
      "learning_rate": 0.00040110613145784545,
      "loss": 2.1053,
      "step": 89991
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.8484948873519897,
      "learning_rate": 0.0004011022801950799,
      "loss": 2.1512,
      "step": 89992
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.8371795415878296,
      "learning_rate": 0.00040109842891351746,
      "loss": 2.2125,
      "step": 89993
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.8326444625854492,
      "learning_rate": 0.0004010945776131588,
      "loss": 2.0878,
      "step": 89994
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.8412660360336304,
      "learning_rate": 0.0004010907262940049,
      "loss": 2.1686,
      "step": 89995
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.8568181991577148,
      "learning_rate": 0.00040108687495605627,
      "loss": 2.2138,
      "step": 89996
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.8574353456497192,
      "learning_rate": 0.0004010830235993137,
      "loss": 2.0615,
      "step": 89997
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.7825426459312439,
      "learning_rate": 0.0004010791722237779,
      "loss": 2.2196,
      "step": 89998
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.8482470512390137,
      "learning_rate": 0.0004010753208294495,
      "loss": 2.2276,
      "step": 89999
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.784147322177887,
      "learning_rate": 0.0004010714694163294,
      "loss": 2.089,
      "step": 90000
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.9321642518043518,
      "learning_rate": 0.0004010676179844181,
      "loss": 2.421,
      "step": 90001
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.8090845942497253,
      "learning_rate": 0.0004010637665337165,
      "loss": 2.0918,
      "step": 90002
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.7987651228904724,
      "learning_rate": 0.00040105991506422513,
      "loss": 2.2387,
      "step": 90003
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.8847371935844421,
      "learning_rate": 0.0004010560635759449,
      "loss": 1.9673,
      "step": 90004
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.7858341932296753,
      "learning_rate": 0.00040105221206887645,
      "loss": 2.0887,
      "step": 90005
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.9300084710121155,
      "learning_rate": 0.00040104836054302045,
      "loss": 2.2098,
      "step": 90006
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.7900713682174683,
      "learning_rate": 0.0004010445089983777,
      "loss": 2.2591,
      "step": 90007
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.9184185862541199,
      "learning_rate": 0.0004010406574349488,
      "loss": 2.1272,
      "step": 90008
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.7658536434173584,
      "learning_rate": 0.00040103680585273455,
      "loss": 2.1221,
      "step": 90009
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.8412161469459534,
      "learning_rate": 0.00040103295425173574,
      "loss": 2.2127,
      "step": 90010
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.8311716318130493,
      "learning_rate": 0.0004010291026319529,
      "loss": 2.2123,
      "step": 90011
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.9249448180198669,
      "learning_rate": 0.00040102525099338683,
      "loss": 2.3011,
      "step": 90012
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.8777102828025818,
      "learning_rate": 0.00040102139933603835,
      "loss": 2.2222,
      "step": 90013
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.9011000990867615,
      "learning_rate": 0.000401017547659908,
      "loss": 2.3207,
      "step": 90014
    },
    {
      "epoch": 1.17,
      "grad_norm": 1.1857227087020874,
      "learning_rate": 0.0004010136959649966,
      "loss": 2.1491,
      "step": 90015
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.9675537943840027,
      "learning_rate": 0.0004010098442513049,
      "loss": 2.0193,
      "step": 90016
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.9250405430793762,
      "learning_rate": 0.0004010059925188335,
      "loss": 2.2651,
      "step": 90017
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.8288120627403259,
      "learning_rate": 0.00040100214076758316,
      "loss": 2.3098,
      "step": 90018
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.8540539145469666,
      "learning_rate": 0.0004009982889975547,
      "loss": 2.1262,
      "step": 90019
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.8601857423782349,
      "learning_rate": 0.00040099443720874875,
      "loss": 2.0302,
      "step": 90020
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.8273597955703735,
      "learning_rate": 0.0004009905854011659,
      "loss": 2.2908,
      "step": 90021
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.8578585386276245,
      "learning_rate": 0.00040098673357480717,
      "loss": 2.3139,
      "step": 90022
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.9263215065002441,
      "learning_rate": 0.000400982881729673,
      "loss": 2.1149,
      "step": 90023
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.8107861280441284,
      "learning_rate": 0.00040097902986576414,
      "loss": 2.1889,
      "step": 90024
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.8646183013916016,
      "learning_rate": 0.00040097517798308154,
      "loss": 2.1559,
      "step": 90025
    },
    {
      "epoch": 1.17,
      "grad_norm": 1.1799135208129883,
      "learning_rate": 0.0004009713260816256,
      "loss": 2.3959,
      "step": 90026
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.8190811276435852,
      "learning_rate": 0.0004009674741613973,
      "loss": 2.2493,
      "step": 90027
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.7903114557266235,
      "learning_rate": 0.0004009636222223972,
      "loss": 2.3482,
      "step": 90028
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.7948648929595947,
      "learning_rate": 0.000400959770264626,
      "loss": 2.1718,
      "step": 90029
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.8130778074264526,
      "learning_rate": 0.0004009559182880845,
      "loss": 2.2701,
      "step": 90030
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.8539745807647705,
      "learning_rate": 0.00040095206629277345,
      "loss": 2.1798,
      "step": 90031
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.8852490186691284,
      "learning_rate": 0.0004009482142786934,
      "loss": 2.2401,
      "step": 90032
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.758630096912384,
      "learning_rate": 0.0004009443622458453,
      "loss": 2.0524,
      "step": 90033
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.8408694267272949,
      "learning_rate": 0.00040094051019422966,
      "loss": 2.3259,
      "step": 90034
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.8243812918663025,
      "learning_rate": 0.0004009366581238472,
      "loss": 2.2155,
      "step": 90035
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.8101397752761841,
      "learning_rate": 0.00040093280603469884,
      "loss": 2.1829,
      "step": 90036
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.8803238868713379,
      "learning_rate": 0.00040092895392678505,
      "loss": 2.1958,
      "step": 90037
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.771092414855957,
      "learning_rate": 0.0004009251018001068,
      "loss": 2.2923,
      "step": 90038
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.8829299211502075,
      "learning_rate": 0.0004009212496546646,
      "loss": 2.254,
      "step": 90039
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.8379895091056824,
      "learning_rate": 0.00040091739749045915,
      "loss": 2.1129,
      "step": 90040
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.7759385704994202,
      "learning_rate": 0.0004009135453074914,
      "loss": 2.1358,
      "step": 90041
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.771054744720459,
      "learning_rate": 0.0004009096931057618,
      "loss": 2.2122,
      "step": 90042
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.8308451771736145,
      "learning_rate": 0.0004009058408852712,
      "loss": 2.3779,
      "step": 90043
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.7609881162643433,
      "learning_rate": 0.00040090198864602036,
      "loss": 2.3976,
      "step": 90044
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.7694195508956909,
      "learning_rate": 0.00040089813638801,
      "loss": 1.919,
      "step": 90045
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.768398106098175,
      "learning_rate": 0.0004008942841112406,
      "loss": 2.1819,
      "step": 90046
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.8071702122688293,
      "learning_rate": 0.0004008904318157131,
      "loss": 1.874,
      "step": 90047
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.8537020087242126,
      "learning_rate": 0.0004008865795014282,
      "loss": 2.0033,
      "step": 90048
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.8421211838722229,
      "learning_rate": 0.0004008827271683866,
      "loss": 2.1927,
      "step": 90049
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.8222622275352478,
      "learning_rate": 0.0004008788748165889,
      "loss": 2.2484,
      "step": 90050
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.8171781897544861,
      "learning_rate": 0.000400875022446036,
      "loss": 2.0765,
      "step": 90051
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.8837154507637024,
      "learning_rate": 0.00040087117005672854,
      "loss": 2.0517,
      "step": 90052
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.7726213932037354,
      "learning_rate": 0.0004008673176486672,
      "loss": 2.253,
      "step": 90053
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.9289036989212036,
      "learning_rate": 0.0004008634652218527,
      "loss": 2.2731,
      "step": 90054
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.7803133726119995,
      "learning_rate": 0.00040085961277628576,
      "loss": 2.1368,
      "step": 90055
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.8376546502113342,
      "learning_rate": 0.0004008557603119672,
      "loss": 2.1546,
      "step": 90056
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.8289148211479187,
      "learning_rate": 0.0004008519078288976,
      "loss": 1.992,
      "step": 90057
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.8094843626022339,
      "learning_rate": 0.0004008480553270777,
      "loss": 2.2944,
      "step": 90058
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.8671641945838928,
      "learning_rate": 0.0004008442028065083,
      "loss": 2.197,
      "step": 90059
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.8840608596801758,
      "learning_rate": 0.00040084035026719005,
      "loss": 2.1229,
      "step": 90060
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.8649691343307495,
      "learning_rate": 0.0004008364977091236,
      "loss": 2.3342,
      "step": 90061
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.7988436222076416,
      "learning_rate": 0.0004008326451323098,
      "loss": 2.1834,
      "step": 90062
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.8243634104728699,
      "learning_rate": 0.0004008287925367493,
      "loss": 2.0936,
      "step": 90063
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.8150726556777954,
      "learning_rate": 0.0004008249399224428,
      "loss": 2.1271,
      "step": 90064
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.8256242275238037,
      "learning_rate": 0.0004008210872893911,
      "loss": 2.0016,
      "step": 90065
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.7823824882507324,
      "learning_rate": 0.00040081723463759487,
      "loss": 2.0796,
      "step": 90066
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.810090959072113,
      "learning_rate": 0.00040081338196705475,
      "loss": 2.3507,
      "step": 90067
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.8004592061042786,
      "learning_rate": 0.0004008095292777716,
      "loss": 2.0711,
      "step": 90068
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.838801383972168,
      "learning_rate": 0.00040080567656974593,
      "loss": 2.4306,
      "step": 90069
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.8005613088607788,
      "learning_rate": 0.00040080182384297865,
      "loss": 2.24,
      "step": 90070
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.8253526091575623,
      "learning_rate": 0.0004007979710974704,
      "loss": 1.8905,
      "step": 90071
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.8370599746704102,
      "learning_rate": 0.0004007941183332219,
      "loss": 2.2494,
      "step": 90072
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.7968084812164307,
      "learning_rate": 0.00040079026555023386,
      "loss": 2.1069,
      "step": 90073
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.8143354058265686,
      "learning_rate": 0.00040078641274850707,
      "loss": 2.1709,
      "step": 90074
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.8738493919372559,
      "learning_rate": 0.0004007825599280421,
      "loss": 2.2456,
      "step": 90075
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.8544164299964905,
      "learning_rate": 0.0004007787070888398,
      "loss": 2.2856,
      "step": 90076
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.8200586438179016,
      "learning_rate": 0.0004007748542309008,
      "loss": 2.2639,
      "step": 90077
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.7766446471214294,
      "learning_rate": 0.00040077100135422594,
      "loss": 1.9849,
      "step": 90078
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.8690178394317627,
      "learning_rate": 0.0004007671484588157,
      "loss": 2.4406,
      "step": 90079
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.865771472454071,
      "learning_rate": 0.0004007632955446711,
      "loss": 2.3512,
      "step": 90080
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.8640820980072021,
      "learning_rate": 0.0004007594426117926,
      "loss": 1.9954,
      "step": 90081
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.8240094184875488,
      "learning_rate": 0.00040075558966018107,
      "loss": 2.2152,
      "step": 90082
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.8461273312568665,
      "learning_rate": 0.0004007517366898372,
      "loss": 2.2339,
      "step": 90083
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.8858910202980042,
      "learning_rate": 0.0004007478837007616,
      "loss": 2.2121,
      "step": 90084
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.8406918048858643,
      "learning_rate": 0.00040074403069295504,
      "loss": 2.3838,
      "step": 90085
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.8905646204948425,
      "learning_rate": 0.00040074017766641843,
      "loss": 2.2925,
      "step": 90086
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.8416619300842285,
      "learning_rate": 0.0004007363246211521,
      "loss": 2.1767,
      "step": 90087
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.8603326082229614,
      "learning_rate": 0.00040073247155715714,
      "loss": 2.145,
      "step": 90088
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.8538756966590881,
      "learning_rate": 0.0004007286184744341,
      "loss": 2.1185,
      "step": 90089
    },
    {
      "epoch": 1.17,
      "grad_norm": 1.105523705482483,
      "learning_rate": 0.00040072476537298367,
      "loss": 2.0515,
      "step": 90090
    },
    {
      "epoch": 1.17,
      "grad_norm": 1.0396411418914795,
      "learning_rate": 0.0004007209122528066,
      "loss": 2.1909,
      "step": 90091
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.801006555557251,
      "learning_rate": 0.0004007170591139037,
      "loss": 2.0863,
      "step": 90092
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.8669003248214722,
      "learning_rate": 0.00040071320595627544,
      "loss": 2.2015,
      "step": 90093
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.868196427822113,
      "learning_rate": 0.0004007093527799228,
      "loss": 2.1148,
      "step": 90094
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.8088834285736084,
      "learning_rate": 0.00040070549958484637,
      "loss": 2.4078,
      "step": 90095
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.8019934296607971,
      "learning_rate": 0.0004007016463710469,
      "loss": 2.1469,
      "step": 90096
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.8801595568656921,
      "learning_rate": 0.00040069779313852506,
      "loss": 2.1071,
      "step": 90097
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.917633593082428,
      "learning_rate": 0.00040069393988728164,
      "loss": 2.3812,
      "step": 90098
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.76081782579422,
      "learning_rate": 0.00040069008661731726,
      "loss": 2.3871,
      "step": 90099
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.7921320199966431,
      "learning_rate": 0.0004006862333286327,
      "loss": 2.2612,
      "step": 90100
    },
    {
      "epoch": 1.17,
      "grad_norm": 1.0531678199768066,
      "learning_rate": 0.00040068238002122876,
      "loss": 2.0087,
      "step": 90101
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.832338809967041,
      "learning_rate": 0.000400678526695106,
      "loss": 2.4336,
      "step": 90102
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.7758522629737854,
      "learning_rate": 0.0004006746733502652,
      "loss": 2.3701,
      "step": 90103
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.845405101776123,
      "learning_rate": 0.00040067081998670706,
      "loss": 2.0782,
      "step": 90104
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.9792962074279785,
      "learning_rate": 0.00040066696660443233,
      "loss": 2.0749,
      "step": 90105
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.7950984835624695,
      "learning_rate": 0.0004006631132034416,
      "loss": 2.2943,
      "step": 90106
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.8135631084442139,
      "learning_rate": 0.0004006592597837359,
      "loss": 2.1522,
      "step": 90107
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.8573670983314514,
      "learning_rate": 0.0004006554063453157,
      "loss": 2.1945,
      "step": 90108
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.8272234201431274,
      "learning_rate": 0.00040065155288818166,
      "loss": 2.0435,
      "step": 90109
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.7528040409088135,
      "learning_rate": 0.0004006476994123347,
      "loss": 2.1119,
      "step": 90110
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.8673415184020996,
      "learning_rate": 0.0004006438459177754,
      "loss": 2.2456,
      "step": 90111
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.7575607895851135,
      "learning_rate": 0.00040063999240450444,
      "loss": 2.2163,
      "step": 90112
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.8101400136947632,
      "learning_rate": 0.0004006361388725226,
      "loss": 2.2059,
      "step": 90113
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.8013922572135925,
      "learning_rate": 0.0004006322853218307,
      "loss": 2.0242,
      "step": 90114
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.903969943523407,
      "learning_rate": 0.0004006284317524293,
      "loss": 2.2334,
      "step": 90115
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.8003788590431213,
      "learning_rate": 0.0004006245781643192,
      "loss": 2.1905,
      "step": 90116
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.8715865612030029,
      "learning_rate": 0.00040062072455750115,
      "loss": 2.2819,
      "step": 90117
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.8117473721504211,
      "learning_rate": 0.00040061687093197564,
      "loss": 2.2111,
      "step": 90118
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.875050961971283,
      "learning_rate": 0.0004006130172877437,
      "loss": 2.2928,
      "step": 90119
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.8378573656082153,
      "learning_rate": 0.00040060916362480583,
      "loss": 2.1525,
      "step": 90120
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.7357155084609985,
      "learning_rate": 0.00040060530994316283,
      "loss": 2.0094,
      "step": 90121
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.7977151274681091,
      "learning_rate": 0.0004006014562428155,
      "loss": 2.0474,
      "step": 90122
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.7933375835418701,
      "learning_rate": 0.0004005976025237643,
      "loss": 1.9935,
      "step": 90123
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.8007722496986389,
      "learning_rate": 0.0004005937487860102,
      "loss": 2.3691,
      "step": 90124
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.7871724963188171,
      "learning_rate": 0.0004005898950295538,
      "loss": 2.087,
      "step": 90125
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.848569929599762,
      "learning_rate": 0.00040058604125439587,
      "loss": 2.2722,
      "step": 90126
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.8268167972564697,
      "learning_rate": 0.0004005821874605371,
      "loss": 2.3828,
      "step": 90127
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.8682774901390076,
      "learning_rate": 0.00040057833364797813,
      "loss": 2.0674,
      "step": 90128
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.80399090051651,
      "learning_rate": 0.0004005744798167198,
      "loss": 2.0385,
      "step": 90129
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.8914331197738647,
      "learning_rate": 0.00040057062596676285,
      "loss": 2.111,
      "step": 90130
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.8956338167190552,
      "learning_rate": 0.00040056677209810777,
      "loss": 2.2397,
      "step": 90131
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.8778209686279297,
      "learning_rate": 0.0004005629182107555,
      "loss": 2.1394,
      "step": 90132
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.834514319896698,
      "learning_rate": 0.0004005590643047067,
      "loss": 1.9886,
      "step": 90133
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.9162383675575256,
      "learning_rate": 0.000400555210379962,
      "loss": 2.3127,
      "step": 90134
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.8403884768486023,
      "learning_rate": 0.0004005513564365223,
      "loss": 2.163,
      "step": 90135
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.8264713883399963,
      "learning_rate": 0.0004005475024743881,
      "loss": 2.107,
      "step": 90136
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.7203083634376526,
      "learning_rate": 0.00040054364849356034,
      "loss": 1.982,
      "step": 90137
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.8547336459159851,
      "learning_rate": 0.00040053979449403956,
      "loss": 2.0246,
      "step": 90138
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.9472405910491943,
      "learning_rate": 0.0004005359404758265,
      "loss": 2.3561,
      "step": 90139
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.7608227133750916,
      "learning_rate": 0.00040053208643892193,
      "loss": 2.2868,
      "step": 90140
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.8578752875328064,
      "learning_rate": 0.0004005282323833266,
      "loss": 2.2357,
      "step": 90141
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.9587484002113342,
      "learning_rate": 0.0004005243783090411,
      "loss": 2.327,
      "step": 90142
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.9354387521743774,
      "learning_rate": 0.0004005205242160662,
      "loss": 2.209,
      "step": 90143
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.8959237337112427,
      "learning_rate": 0.0004005166701044027,
      "loss": 2.0771,
      "step": 90144
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.8740095496177673,
      "learning_rate": 0.00040051281597405124,
      "loss": 2.001,
      "step": 90145
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.8321318030357361,
      "learning_rate": 0.0004005089618250126,
      "loss": 2.1062,
      "step": 90146
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.8308900594711304,
      "learning_rate": 0.00040050510765728733,
      "loss": 2.1398,
      "step": 90147
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.8755014538764954,
      "learning_rate": 0.0004005012534708764,
      "loss": 2.4577,
      "step": 90148
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.8701217770576477,
      "learning_rate": 0.00040049739926578025,
      "loss": 1.9962,
      "step": 90149
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.7783799767494202,
      "learning_rate": 0.00040049354504199984,
      "loss": 2.0529,
      "step": 90150
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.830754816532135,
      "learning_rate": 0.0004004896907995357,
      "loss": 1.9664,
      "step": 90151
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.8545107245445251,
      "learning_rate": 0.0004004858365383887,
      "loss": 2.2154,
      "step": 90152
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.8328207731246948,
      "learning_rate": 0.00040048198225855954,
      "loss": 2.137,
      "step": 90153
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.7823933959007263,
      "learning_rate": 0.00040047812796004876,
      "loss": 2.2864,
      "step": 90154
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.8933680057525635,
      "learning_rate": 0.00040047427364285724,
      "loss": 2.1195,
      "step": 90155
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.9220665693283081,
      "learning_rate": 0.00040047041930698566,
      "loss": 2.1222,
      "step": 90156
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.842703640460968,
      "learning_rate": 0.00040046656495243475,
      "loss": 2.3435,
      "step": 90157
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.8235775828361511,
      "learning_rate": 0.0004004627105792052,
      "loss": 2.2094,
      "step": 90158
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.790769100189209,
      "learning_rate": 0.00040045885618729773,
      "loss": 1.9487,
      "step": 90159
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.7919836640357971,
      "learning_rate": 0.00040045500177671303,
      "loss": 2.0489,
      "step": 90160
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.886364758014679,
      "learning_rate": 0.00040045114734745186,
      "loss": 2.3577,
      "step": 90161
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.9197615385055542,
      "learning_rate": 0.00040044729289951504,
      "loss": 2.2362,
      "step": 90162
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.9300962090492249,
      "learning_rate": 0.00040044343843290305,
      "loss": 2.2299,
      "step": 90163
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.7982882261276245,
      "learning_rate": 0.00040043958394761676,
      "loss": 1.997,
      "step": 90164
    },
    {
      "epoch": 1.17,
      "grad_norm": 1.051806926727295,
      "learning_rate": 0.0004004357294436569,
      "loss": 2.086,
      "step": 90165
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.8229818940162659,
      "learning_rate": 0.00040043187492102406,
      "loss": 2.2088,
      "step": 90166
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.8068850040435791,
      "learning_rate": 0.000400428020379719,
      "loss": 2.1373,
      "step": 90167
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.8015331625938416,
      "learning_rate": 0.00040042416581974257,
      "loss": 2.315,
      "step": 90168
    },
    {
      "epoch": 1.17,
      "grad_norm": 1.0366320610046387,
      "learning_rate": 0.0004004203112410954,
      "loss": 2.1277,
      "step": 90169
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.787227213382721,
      "learning_rate": 0.0004004164566437782,
      "loss": 1.9563,
      "step": 90170
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.7382644414901733,
      "learning_rate": 0.00040041260202779166,
      "loss": 2.1026,
      "step": 90171
    },
    {
      "epoch": 1.17,
      "grad_norm": 1.0880624055862427,
      "learning_rate": 0.00040040874739313647,
      "loss": 2.0627,
      "step": 90172
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.8838061094284058,
      "learning_rate": 0.0004004048927398135,
      "loss": 2.2208,
      "step": 90173
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.8732354044914246,
      "learning_rate": 0.00040040103806782324,
      "loss": 2.1964,
      "step": 90174
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.9819835424423218,
      "learning_rate": 0.0004003971833771667,
      "loss": 2.1809,
      "step": 90175
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.8757851123809814,
      "learning_rate": 0.00040039332866784427,
      "loss": 1.9282,
      "step": 90176
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.8461794257164001,
      "learning_rate": 0.00040038947393985694,
      "loss": 2.058,
      "step": 90177
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.9419361352920532,
      "learning_rate": 0.00040038561919320524,
      "loss": 2.203,
      "step": 90178
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.8661487698554993,
      "learning_rate": 0.0004003817644278899,
      "loss": 2.2313,
      "step": 90179
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.9250831604003906,
      "learning_rate": 0.00040037790964391173,
      "loss": 2.1503,
      "step": 90180
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.8642839193344116,
      "learning_rate": 0.0004003740548412716,
      "loss": 2.2983,
      "step": 90181
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.9021057486534119,
      "learning_rate": 0.0004003702000199698,
      "loss": 2.3648,
      "step": 90182
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.8335120677947998,
      "learning_rate": 0.0004003663451800074,
      "loss": 1.9031,
      "step": 90183
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.9379293322563171,
      "learning_rate": 0.00040036249032138503,
      "loss": 2.0568,
      "step": 90184
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.8633661866188049,
      "learning_rate": 0.0004003586354441033,
      "loss": 2.1888,
      "step": 90185
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.8467293381690979,
      "learning_rate": 0.00040035478054816295,
      "loss": 2.1875,
      "step": 90186
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.8683178424835205,
      "learning_rate": 0.0004003509256335649,
      "loss": 2.349,
      "step": 90187
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.7863580584526062,
      "learning_rate": 0.0004003470707003096,
      "loss": 2.0013,
      "step": 90188
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.822399377822876,
      "learning_rate": 0.00040034321574839793,
      "loss": 2.2327,
      "step": 90189
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.8261585831642151,
      "learning_rate": 0.00040033936077783067,
      "loss": 1.977,
      "step": 90190
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.921076238155365,
      "learning_rate": 0.00040033550578860824,
      "loss": 2.3056,
      "step": 90191
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.8603313565254211,
      "learning_rate": 0.0004003316507807317,
      "loss": 2.1884,
      "step": 90192
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.8850975036621094,
      "learning_rate": 0.00040032779575420154,
      "loss": 2.0136,
      "step": 90193
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.8710170984268188,
      "learning_rate": 0.00040032394070901856,
      "loss": 2.1215,
      "step": 90194
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.9690828919410706,
      "learning_rate": 0.00040032008564518345,
      "loss": 2.0184,
      "step": 90195
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.9884160161018372,
      "learning_rate": 0.00040031623056269703,
      "loss": 2.397,
      "step": 90196
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.9614092707633972,
      "learning_rate": 0.0004003123754615598,
      "loss": 2.2647,
      "step": 90197
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.8028926849365234,
      "learning_rate": 0.00040030852034177264,
      "loss": 2.2036,
      "step": 90198
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.7871819138526917,
      "learning_rate": 0.0004003046652033363,
      "loss": 2.1753,
      "step": 90199
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.946163535118103,
      "learning_rate": 0.0004003008100462513,
      "loss": 2.0706,
      "step": 90200
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.9056556820869446,
      "learning_rate": 0.0004002969548705186,
      "loss": 2.0355,
      "step": 90201
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.8895344138145447,
      "learning_rate": 0.0004002930996761388,
      "loss": 2.3392,
      "step": 90202
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.9404471516609192,
      "learning_rate": 0.0004002892444631126,
      "loss": 2.4184,
      "step": 90203
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.9096652269363403,
      "learning_rate": 0.0004002853892314407,
      "loss": 2.032,
      "step": 90204
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.866227388381958,
      "learning_rate": 0.0004002815339811239,
      "loss": 2.1125,
      "step": 90205
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.7984099984169006,
      "learning_rate": 0.00040027767871216285,
      "loss": 2.3458,
      "step": 90206
    },
    {
      "epoch": 1.17,
      "grad_norm": 1.002942681312561,
      "learning_rate": 0.00040027382342455823,
      "loss": 2.1118,
      "step": 90207
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.8696469068527222,
      "learning_rate": 0.0004002699681183109,
      "loss": 2.2315,
      "step": 90208
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.8027034997940063,
      "learning_rate": 0.00040026611279342145,
      "loss": 2.1028,
      "step": 90209
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.9121668338775635,
      "learning_rate": 0.0004002622574498906,
      "loss": 2.1219,
      "step": 90210
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.8251166343688965,
      "learning_rate": 0.0004002584020877192,
      "loss": 2.0614,
      "step": 90211
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.8528842926025391,
      "learning_rate": 0.00040025454670690783,
      "loss": 2.1238,
      "step": 90212
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.8941330909729004,
      "learning_rate": 0.0004002506913074572,
      "loss": 2.2577,
      "step": 90213
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.9522066712379456,
      "learning_rate": 0.00040024683588936813,
      "loss": 2.0785,
      "step": 90214
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.8281272053718567,
      "learning_rate": 0.00040024298045264126,
      "loss": 2.2715,
      "step": 90215
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.8516919016838074,
      "learning_rate": 0.00040023912499727733,
      "loss": 2.4224,
      "step": 90216
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.8575873970985413,
      "learning_rate": 0.00040023526952327705,
      "loss": 2.2026,
      "step": 90217
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.8535692095756531,
      "learning_rate": 0.0004002314140306412,
      "loss": 2.0182,
      "step": 90218
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.7765775322914124,
      "learning_rate": 0.00040022755851937036,
      "loss": 1.9568,
      "step": 90219
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.8212512731552124,
      "learning_rate": 0.0004002237029894653,
      "loss": 1.8608,
      "step": 90220
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.883584201335907,
      "learning_rate": 0.00040021984744092684,
      "loss": 2.3477,
      "step": 90221
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.8134043216705322,
      "learning_rate": 0.0004002159918737556,
      "loss": 2.2327,
      "step": 90222
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.7816364765167236,
      "learning_rate": 0.00040021213628795233,
      "loss": 2.2422,
      "step": 90223
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.7929102182388306,
      "learning_rate": 0.00040020828068351765,
      "loss": 2.1046,
      "step": 90224
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.8060516119003296,
      "learning_rate": 0.00040020442506045247,
      "loss": 2.1624,
      "step": 90225
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.9101825952529907,
      "learning_rate": 0.00040020056941875734,
      "loss": 2.0897,
      "step": 90226
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.9634788036346436,
      "learning_rate": 0.00040019671375843296,
      "loss": 2.3329,
      "step": 90227
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.7608156204223633,
      "learning_rate": 0.0004001928580794802,
      "loss": 2.1793,
      "step": 90228
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.8616987466812134,
      "learning_rate": 0.0004001890023818997,
      "loss": 2.3049,
      "step": 90229
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.8366566896438599,
      "learning_rate": 0.00040018514666569213,
      "loss": 2.0845,
      "step": 90230
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.784582793712616,
      "learning_rate": 0.00040018129093085834,
      "loss": 2.0602,
      "step": 90231
    },
    {
      "epoch": 1.17,
      "grad_norm": 1.0201220512390137,
      "learning_rate": 0.0004001774351773989,
      "loss": 2.2026,
      "step": 90232
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.8465919494628906,
      "learning_rate": 0.0004001735794053146,
      "loss": 2.2707,
      "step": 90233
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.7839525938034058,
      "learning_rate": 0.0004001697236146061,
      "loss": 1.9814,
      "step": 90234
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.8567702174186707,
      "learning_rate": 0.0004001658678052742,
      "loss": 2.3212,
      "step": 90235
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.9789409041404724,
      "learning_rate": 0.00040016201197731955,
      "loss": 2.0461,
      "step": 90236
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.8598804473876953,
      "learning_rate": 0.00040015815613074284,
      "loss": 2.3595,
      "step": 90237
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.743726909160614,
      "learning_rate": 0.00040015430026554496,
      "loss": 2.2119,
      "step": 90238
    },
    {
      "epoch": 1.17,
      "grad_norm": 1.291060209274292,
      "learning_rate": 0.0004001504443817264,
      "loss": 2.1973,
      "step": 90239
    },
    {
      "epoch": 1.17,
      "grad_norm": 0.8433161973953247,
      "learning_rate": 0.000400146588479288,
      "loss": 2.0794,
      "step": 90240
    },
    {
      "epoch": 1.17,
      "grad_norm": 1.0029990673065186,
      "learning_rate": 0.0004001427325582305,
      "loss": 2.2946,
      "step": 90241
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.93538498878479,
      "learning_rate": 0.00040013887661855457,
      "loss": 2.0632,
      "step": 90242
    },
    {
      "epoch": 1.18,
      "grad_norm": 1.1050050258636475,
      "learning_rate": 0.0004001350206602609,
      "loss": 2.0664,
      "step": 90243
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.9990932941436768,
      "learning_rate": 0.00040013116468335037,
      "loss": 2.474,
      "step": 90244
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.7842112183570862,
      "learning_rate": 0.0004001273086878234,
      "loss": 2.2438,
      "step": 90245
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.9079868197441101,
      "learning_rate": 0.0004001234526736809,
      "loss": 2.0014,
      "step": 90246
    },
    {
      "epoch": 1.18,
      "grad_norm": 1.0793894529342651,
      "learning_rate": 0.00040011959664092357,
      "loss": 2.1803,
      "step": 90247
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.8072783350944519,
      "learning_rate": 0.0004001157405895522,
      "loss": 2.091,
      "step": 90248
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.8319937586784363,
      "learning_rate": 0.0004001118845195673,
      "loss": 2.4111,
      "step": 90249
    },
    {
      "epoch": 1.18,
      "grad_norm": 1.2320700883865356,
      "learning_rate": 0.00040010802843096975,
      "loss": 2.2327,
      "step": 90250
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.8752008676528931,
      "learning_rate": 0.0004001041723237603,
      "loss": 2.3306,
      "step": 90251
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.8429689407348633,
      "learning_rate": 0.00040010031619793954,
      "loss": 2.2578,
      "step": 90252
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.8772560954093933,
      "learning_rate": 0.00040009646005350827,
      "loss": 2.401,
      "step": 90253
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.8824847340583801,
      "learning_rate": 0.0004000926038904672,
      "loss": 2.172,
      "step": 90254
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.715419590473175,
      "learning_rate": 0.0004000887477088169,
      "loss": 2.1596,
      "step": 90255
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.7591357231140137,
      "learning_rate": 0.0004000848915085583,
      "loss": 2.1682,
      "step": 90256
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.7846446633338928,
      "learning_rate": 0.00040008103528969206,
      "loss": 2.1155,
      "step": 90257
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.8616548776626587,
      "learning_rate": 0.00040007717905221884,
      "loss": 2.1697,
      "step": 90258
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.849088191986084,
      "learning_rate": 0.0004000733227961393,
      "loss": 2.0656,
      "step": 90259
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.8289379477500916,
      "learning_rate": 0.0004000694665214544,
      "loss": 2.3166,
      "step": 90260
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.9711260199546814,
      "learning_rate": 0.00040006561022816456,
      "loss": 2.2673,
      "step": 90261
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.9237918257713318,
      "learning_rate": 0.00040006175391627067,
      "loss": 2.0166,
      "step": 90262
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.857545018196106,
      "learning_rate": 0.0004000578975857735,
      "loss": 2.3018,
      "step": 90263
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.8880138993263245,
      "learning_rate": 0.0004000540412366736,
      "loss": 2.0759,
      "step": 90264
    },
    {
      "epoch": 1.18,
      "grad_norm": 1.0141266584396362,
      "learning_rate": 0.0004000501848689717,
      "loss": 2.1911,
      "step": 90265
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.7759765982627869,
      "learning_rate": 0.00040004632848266873,
      "loss": 2.2273,
      "step": 90266
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.9899738430976868,
      "learning_rate": 0.00040004247207776515,
      "loss": 2.2256,
      "step": 90267
    },
    {
      "epoch": 1.18,
      "grad_norm": 1.009729266166687,
      "learning_rate": 0.00040003861565426175,
      "loss": 2.1082,
      "step": 90268
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.8548871278762817,
      "learning_rate": 0.00040003475921215935,
      "loss": 2.2047,
      "step": 90269
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.8709923624992371,
      "learning_rate": 0.00040003090275145866,
      "loss": 2.3945,
      "step": 90270
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.7866628170013428,
      "learning_rate": 0.0004000270462721602,
      "loss": 2.2854,
      "step": 90271
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.9266470074653625,
      "learning_rate": 0.000400023189774265,
      "loss": 2.1413,
      "step": 90272
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.8971371650695801,
      "learning_rate": 0.0004000193332577734,
      "loss": 2.2258,
      "step": 90273
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.8316534757614136,
      "learning_rate": 0.00040001547672268643,
      "loss": 2.3338,
      "step": 90274
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.7574178576469421,
      "learning_rate": 0.0004000116201690048,
      "loss": 2.1531,
      "step": 90275
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.8451496958732605,
      "learning_rate": 0.00040000776359672895,
      "loss": 2.3148,
      "step": 90276
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.8267717957496643,
      "learning_rate": 0.0004000039070058598,
      "loss": 2.3134,
      "step": 90277
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.9034491777420044,
      "learning_rate": 0.0004000000503963981,
      "loss": 2.3738,
      "step": 90278
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.772584855556488,
      "learning_rate": 0.0003999961937683444,
      "loss": 2.0547,
      "step": 90279
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.7865453362464905,
      "learning_rate": 0.0003999923371216996,
      "loss": 2.2512,
      "step": 90280
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.7956176996231079,
      "learning_rate": 0.0003999884804564643,
      "loss": 2.0646,
      "step": 90281
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.7700718641281128,
      "learning_rate": 0.0003999846237726393,
      "loss": 2.1997,
      "step": 90282
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.7916952967643738,
      "learning_rate": 0.00039998076707022524,
      "loss": 2.2524,
      "step": 90283
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.7846227884292603,
      "learning_rate": 0.00039997691034922295,
      "loss": 2.1228,
      "step": 90284
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.7927133440971375,
      "learning_rate": 0.000399973053609633,
      "loss": 2.1698,
      "step": 90285
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.8314520716667175,
      "learning_rate": 0.00039996919685145605,
      "loss": 2.1808,
      "step": 90286
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.8469256162643433,
      "learning_rate": 0.00039996534007469313,
      "loss": 2.0716,
      "step": 90287
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.939021110534668,
      "learning_rate": 0.0003999614832793446,
      "loss": 2.2395,
      "step": 90288
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.7628690004348755,
      "learning_rate": 0.0003999576264654115,
      "loss": 1.9235,
      "step": 90289
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.7927829027175903,
      "learning_rate": 0.0003999537696328943,
      "loss": 2.2085,
      "step": 90290
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.9533740282058716,
      "learning_rate": 0.0003999499127817938,
      "loss": 2.0693,
      "step": 90291
    },
    {
      "epoch": 1.18,
      "grad_norm": 1.0415180921554565,
      "learning_rate": 0.00039994605591211075,
      "loss": 2.0279,
      "step": 90292
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.9025166034698486,
      "learning_rate": 0.0003999421990238459,
      "loss": 2.2825,
      "step": 90293
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.717927873134613,
      "learning_rate": 0.0003999383421169998,
      "loss": 2.1197,
      "step": 90294
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.9232674837112427,
      "learning_rate": 0.0003999344851915733,
      "loss": 2.2117,
      "step": 90295
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.9331706166267395,
      "learning_rate": 0.0003999306282475672,
      "loss": 2.0082,
      "step": 90296
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.9051911234855652,
      "learning_rate": 0.000399926771284982,
      "loss": 2.3043,
      "step": 90297
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.8881973624229431,
      "learning_rate": 0.00039992291430381857,
      "loss": 2.1953,
      "step": 90298
    },
    {
      "epoch": 1.18,
      "grad_norm": 1.0802849531173706,
      "learning_rate": 0.00039991905730407765,
      "loss": 2.0558,
      "step": 90299
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.8525209426879883,
      "learning_rate": 0.0003999152002857597,
      "loss": 1.9473,
      "step": 90300
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.9346863627433777,
      "learning_rate": 0.0003999113432488658,
      "loss": 2.2392,
      "step": 90301
    },
    {
      "epoch": 1.18,
      "grad_norm": 1.0484622716903687,
      "learning_rate": 0.00039990748619339643,
      "loss": 2.2048,
      "step": 90302
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.8097537159919739,
      "learning_rate": 0.0003999036291193524,
      "loss": 2.1823,
      "step": 90303
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.8894623517990112,
      "learning_rate": 0.00039989977202673426,
      "loss": 1.9504,
      "step": 90304
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.8820450305938721,
      "learning_rate": 0.00039989591491554306,
      "loss": 2.2091,
      "step": 90305
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.8137431740760803,
      "learning_rate": 0.00039989205778577925,
      "loss": 2.021,
      "step": 90306
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.7680073380470276,
      "learning_rate": 0.0003998882006374436,
      "loss": 2.3379,
      "step": 90307
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.8313524723052979,
      "learning_rate": 0.0003998843434705369,
      "loss": 2.1269,
      "step": 90308
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.791286051273346,
      "learning_rate": 0.0003998804862850597,
      "loss": 2.0968,
      "step": 90309
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.8344314098358154,
      "learning_rate": 0.00039987662908101294,
      "loss": 2.2433,
      "step": 90310
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.7614889144897461,
      "learning_rate": 0.00039987277185839725,
      "loss": 2.165,
      "step": 90311
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.7810764312744141,
      "learning_rate": 0.00039986891461721324,
      "loss": 2.1267,
      "step": 90312
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.9677190780639648,
      "learning_rate": 0.0003998650573574617,
      "loss": 2.1506,
      "step": 90313
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.7911888957023621,
      "learning_rate": 0.00039986120007914346,
      "loss": 2.0301,
      "step": 90314
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.9492008686065674,
      "learning_rate": 0.0003998573427822591,
      "loss": 2.3363,
      "step": 90315
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.782600998878479,
      "learning_rate": 0.0003998534854668093,
      "loss": 2.0608,
      "step": 90316
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.8512458801269531,
      "learning_rate": 0.000399849628132795,
      "loss": 2.3902,
      "step": 90317
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.9504939317703247,
      "learning_rate": 0.00039984577078021663,
      "loss": 2.2291,
      "step": 90318
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.8976103067398071,
      "learning_rate": 0.00039984191340907516,
      "loss": 2.0602,
      "step": 90319
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.8532973527908325,
      "learning_rate": 0.00039983805601937107,
      "loss": 2.3754,
      "step": 90320
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.9107038974761963,
      "learning_rate": 0.00039983419861110527,
      "loss": 2.1622,
      "step": 90321
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.7964745163917542,
      "learning_rate": 0.0003998303411842785,
      "loss": 2.3089,
      "step": 90322
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.9733678102493286,
      "learning_rate": 0.0003998264837388912,
      "loss": 2.3158,
      "step": 90323
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.8217832446098328,
      "learning_rate": 0.00039982262627494443,
      "loss": 2.0708,
      "step": 90324
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.9707240462303162,
      "learning_rate": 0.00039981876879243865,
      "loss": 1.9961,
      "step": 90325
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.84965580701828,
      "learning_rate": 0.0003998149112913747,
      "loss": 2.1624,
      "step": 90326
    },
    {
      "epoch": 1.18,
      "grad_norm": 1.0782405138015747,
      "learning_rate": 0.0003998110537717533,
      "loss": 2.2938,
      "step": 90327
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.8699584603309631,
      "learning_rate": 0.00039980719623357514,
      "loss": 2.0725,
      "step": 90328
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.828657865524292,
      "learning_rate": 0.00039980333867684087,
      "loss": 2.2186,
      "step": 90329
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.8389546275138855,
      "learning_rate": 0.00039979948110155145,
      "loss": 2.2227,
      "step": 90330
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.9746813178062439,
      "learning_rate": 0.0003997956235077072,
      "loss": 2.1556,
      "step": 90331
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.8585687875747681,
      "learning_rate": 0.0003997917658953091,
      "loss": 2.2232,
      "step": 90332
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.8294153213500977,
      "learning_rate": 0.000399787908264358,
      "loss": 2.1553,
      "step": 90333
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.9031662344932556,
      "learning_rate": 0.0003997840506148543,
      "loss": 2.199,
      "step": 90334
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.8119741082191467,
      "learning_rate": 0.0003997801929467989,
      "loss": 2.2787,
      "step": 90335
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.8804603815078735,
      "learning_rate": 0.0003997763352601925,
      "loss": 2.2582,
      "step": 90336
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.8148902058601379,
      "learning_rate": 0.0003997724775550358,
      "loss": 2.2218,
      "step": 90337
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.8362912535667419,
      "learning_rate": 0.00039976861983132944,
      "loss": 2.1405,
      "step": 90338
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.816461980342865,
      "learning_rate": 0.0003997647620890743,
      "loss": 1.9231,
      "step": 90339
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.8369095921516418,
      "learning_rate": 0.0003997609043282709,
      "loss": 2.0239,
      "step": 90340
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.8940713405609131,
      "learning_rate": 0.00039975704654892014,
      "loss": 2.2874,
      "step": 90341
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.9120449423789978,
      "learning_rate": 0.00039975318875102266,
      "loss": 2.2646,
      "step": 90342
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.8556795716285706,
      "learning_rate": 0.0003997493309345792,
      "loss": 2.0863,
      "step": 90343
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.8336551189422607,
      "learning_rate": 0.0003997454730995904,
      "loss": 2.2489,
      "step": 90344
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.8719415664672852,
      "learning_rate": 0.00039974161524605707,
      "loss": 2.074,
      "step": 90345
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.8227331042289734,
      "learning_rate": 0.00039973775737397983,
      "loss": 2.1395,
      "step": 90346
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.7934443354606628,
      "learning_rate": 0.00039973389948335953,
      "loss": 2.0863,
      "step": 90347
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.860421895980835,
      "learning_rate": 0.00039973004157419677,
      "loss": 2.2817,
      "step": 90348
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.8538632392883301,
      "learning_rate": 0.00039972618364649237,
      "loss": 2.0437,
      "step": 90349
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.815700113773346,
      "learning_rate": 0.0003997223257002469,
      "loss": 2.259,
      "step": 90350
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.9065256714820862,
      "learning_rate": 0.0003997184677354613,
      "loss": 2.2309,
      "step": 90351
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.7676505446434021,
      "learning_rate": 0.00039971460975213606,
      "loss": 2.1572,
      "step": 90352
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.9145210981369019,
      "learning_rate": 0.00039971075175027196,
      "loss": 2.3113,
      "step": 90353
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.8249416351318359,
      "learning_rate": 0.0003997068937298698,
      "loss": 2.2954,
      "step": 90354
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.8863856196403503,
      "learning_rate": 0.00039970303569093025,
      "loss": 2.2572,
      "step": 90355
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.9318037629127502,
      "learning_rate": 0.000399699177633454,
      "loss": 1.9143,
      "step": 90356
    },
    {
      "epoch": 1.18,
      "grad_norm": 1.0566678047180176,
      "learning_rate": 0.00039969531955744186,
      "loss": 2.0408,
      "step": 90357
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.9120609760284424,
      "learning_rate": 0.0003996914614628944,
      "loss": 2.2559,
      "step": 90358
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.8699196577072144,
      "learning_rate": 0.0003996876033498124,
      "loss": 2.179,
      "step": 90359
    },
    {
      "epoch": 1.18,
      "grad_norm": 1.0238889455795288,
      "learning_rate": 0.0003996837452181967,
      "loss": 2.2457,
      "step": 90360
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.9179060459136963,
      "learning_rate": 0.0003996798870680479,
      "loss": 2.2589,
      "step": 90361
    },
    {
      "epoch": 1.18,
      "grad_norm": 1.0233873128890991,
      "learning_rate": 0.0003996760288993666,
      "loss": 2.1276,
      "step": 90362
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.9856701493263245,
      "learning_rate": 0.00039967217071215373,
      "loss": 2.131,
      "step": 90363
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.8850593566894531,
      "learning_rate": 0.0003996683125064099,
      "loss": 2.1721,
      "step": 90364
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.8493430018424988,
      "learning_rate": 0.0003996644542821358,
      "loss": 2.2717,
      "step": 90365
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.9706111550331116,
      "learning_rate": 0.00039966059603933234,
      "loss": 1.9811,
      "step": 90366
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.9147660732269287,
      "learning_rate": 0.00039965673777799996,
      "loss": 2.353,
      "step": 90367
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.8724898099899292,
      "learning_rate": 0.00039965287949813953,
      "loss": 2.0959,
      "step": 90368
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.8837859034538269,
      "learning_rate": 0.0003996490211997518,
      "loss": 2.362,
      "step": 90369
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.9770581126213074,
      "learning_rate": 0.00039964516288283744,
      "loss": 2.2268,
      "step": 90370
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.9238786697387695,
      "learning_rate": 0.00039964130454739707,
      "loss": 2.1624,
      "step": 90371
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.8447418212890625,
      "learning_rate": 0.00039963744619343164,
      "loss": 2.1689,
      "step": 90372
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.907924473285675,
      "learning_rate": 0.00039963358782094165,
      "loss": 2.0993,
      "step": 90373
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.8490608930587769,
      "learning_rate": 0.0003996297294299279,
      "loss": 2.1023,
      "step": 90374
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.7810667753219604,
      "learning_rate": 0.0003996258710203911,
      "loss": 2.1993,
      "step": 90375
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.8517846465110779,
      "learning_rate": 0.000399622012592332,
      "loss": 2.1141,
      "step": 90376
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.918628990650177,
      "learning_rate": 0.0003996181541457512,
      "loss": 2.2292,
      "step": 90377
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.8831946849822998,
      "learning_rate": 0.00039961429568064963,
      "loss": 2.1948,
      "step": 90378
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.9011418223381042,
      "learning_rate": 0.00039961043719702777,
      "loss": 2.0368,
      "step": 90379
    },
    {
      "epoch": 1.18,
      "grad_norm": 1.0624679327011108,
      "learning_rate": 0.0003996065786948865,
      "loss": 2.2468,
      "step": 90380
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.9096336960792542,
      "learning_rate": 0.00039960272017422645,
      "loss": 2.3917,
      "step": 90381
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.8710290789604187,
      "learning_rate": 0.0003995988616350484,
      "loss": 2.1567,
      "step": 90382
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.964611828327179,
      "learning_rate": 0.0003995950030773531,
      "loss": 2.0237,
      "step": 90383
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.8570820689201355,
      "learning_rate": 0.0003995911445011411,
      "loss": 2.0938,
      "step": 90384
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.8096322417259216,
      "learning_rate": 0.0003995872859064133,
      "loss": 2.1183,
      "step": 90385
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.8974338173866272,
      "learning_rate": 0.00039958342729317037,
      "loss": 2.1721,
      "step": 90386
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.9014871716499329,
      "learning_rate": 0.00039957956866141294,
      "loss": 2.3299,
      "step": 90387
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.9458160400390625,
      "learning_rate": 0.00039957571001114184,
      "loss": 2.1266,
      "step": 90388
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.904566764831543,
      "learning_rate": 0.00039957185134235765,
      "loss": 1.9442,
      "step": 90389
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.8872693181037903,
      "learning_rate": 0.00039956799265506126,
      "loss": 2.2103,
      "step": 90390
    },
    {
      "epoch": 1.18,
      "grad_norm": 1.0179710388183594,
      "learning_rate": 0.0003995641339492533,
      "loss": 1.9431,
      "step": 90391
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.8544498682022095,
      "learning_rate": 0.00039956027522493445,
      "loss": 2.1492,
      "step": 90392
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.8958554267883301,
      "learning_rate": 0.00039955641648210544,
      "loss": 2.3096,
      "step": 90393
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.7835381031036377,
      "learning_rate": 0.00039955255772076704,
      "loss": 2.095,
      "step": 90394
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.8393462896347046,
      "learning_rate": 0.0003995486989409199,
      "loss": 2.3111,
      "step": 90395
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.9380742907524109,
      "learning_rate": 0.0003995448401425649,
      "loss": 2.1971,
      "step": 90396
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.8550935983657837,
      "learning_rate": 0.0003995409813257025,
      "loss": 2.1954,
      "step": 90397
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.9250481128692627,
      "learning_rate": 0.0003995371224903336,
      "loss": 2.0291,
      "step": 90398
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.7964857220649719,
      "learning_rate": 0.00039953326363645895,
      "loss": 2.1013,
      "step": 90399
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.8164291977882385,
      "learning_rate": 0.00039952940476407905,
      "loss": 2.2009,
      "step": 90400
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.8060237765312195,
      "learning_rate": 0.0003995255458731948,
      "loss": 2.3643,
      "step": 90401
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.8519904613494873,
      "learning_rate": 0.000399521686963807,
      "loss": 2.1254,
      "step": 90402
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.875759482383728,
      "learning_rate": 0.0003995178280359161,
      "loss": 2.2049,
      "step": 90403
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.7716509103775024,
      "learning_rate": 0.0003995139690895229,
      "loss": 2.0929,
      "step": 90404
    },
    {
      "epoch": 1.18,
      "grad_norm": 1.3575761318206787,
      "learning_rate": 0.00039951011012462833,
      "loss": 2.2532,
      "step": 90405
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.8822563886642456,
      "learning_rate": 0.00039950625114123286,
      "loss": 2.1112,
      "step": 90406
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.8645429611206055,
      "learning_rate": 0.00039950239213933737,
      "loss": 2.0715,
      "step": 90407
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.806955873966217,
      "learning_rate": 0.00039949853311894236,
      "loss": 2.4058,
      "step": 90408
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.8740574717521667,
      "learning_rate": 0.00039949467408004885,
      "loss": 2.084,
      "step": 90409
    },
    {
      "epoch": 1.18,
      "grad_norm": 1.1366664171218872,
      "learning_rate": 0.00039949081502265734,
      "loss": 2.192,
      "step": 90410
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.9123893976211548,
      "learning_rate": 0.00039948695594676863,
      "loss": 2.0563,
      "step": 90411
    },
    {
      "epoch": 1.18,
      "grad_norm": 1.0374023914337158,
      "learning_rate": 0.0003994830968523835,
      "loss": 2.1446,
      "step": 90412
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.9817309379577637,
      "learning_rate": 0.0003994792377395024,
      "loss": 2.2091,
      "step": 90413
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.8940895795822144,
      "learning_rate": 0.00039947537860812634,
      "loss": 2.0098,
      "step": 90414
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.9188106656074524,
      "learning_rate": 0.00039947151945825596,
      "loss": 2.3863,
      "step": 90415
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.8542495965957642,
      "learning_rate": 0.00039946766028989186,
      "loss": 2.0616,
      "step": 90416
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.860793948173523,
      "learning_rate": 0.00039946380110303484,
      "loss": 2.2751,
      "step": 90417
    },
    {
      "epoch": 1.18,
      "grad_norm": 1.0049525499343872,
      "learning_rate": 0.0003994599418976858,
      "loss": 2.3927,
      "step": 90418
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.8099036812782288,
      "learning_rate": 0.00039945608267384506,
      "loss": 2.1655,
      "step": 90419
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.8329578042030334,
      "learning_rate": 0.00039945222343151364,
      "loss": 1.99,
      "step": 90420
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.8200661540031433,
      "learning_rate": 0.0003994483641706922,
      "loss": 2.2846,
      "step": 90421
    },
    {
      "epoch": 1.18,
      "grad_norm": 1.0305733680725098,
      "learning_rate": 0.0003994445048913814,
      "loss": 2.1265,
      "step": 90422
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.9739022850990295,
      "learning_rate": 0.000399440645593582,
      "loss": 2.3262,
      "step": 90423
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.9167571663856506,
      "learning_rate": 0.0003994367862772947,
      "loss": 2.0538,
      "step": 90424
    },
    {
      "epoch": 1.18,
      "grad_norm": 1.1889872550964355,
      "learning_rate": 0.0003994329269425202,
      "loss": 2.3181,
      "step": 90425
    },
    {
      "epoch": 1.18,
      "grad_norm": 1.0416818857192993,
      "learning_rate": 0.00039942906758925925,
      "loss": 2.0707,
      "step": 90426
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.8070554137229919,
      "learning_rate": 0.0003994252082175127,
      "loss": 2.2598,
      "step": 90427
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.9998916983604431,
      "learning_rate": 0.0003994213488272809,
      "loss": 2.201,
      "step": 90428
    },
    {
      "epoch": 1.18,
      "grad_norm": 1.0519499778747559,
      "learning_rate": 0.0003994174894185649,
      "loss": 2.0745,
      "step": 90429
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.8229367136955261,
      "learning_rate": 0.0003994136299913654,
      "loss": 2.1627,
      "step": 90430
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.822838544845581,
      "learning_rate": 0.0003994097705456829,
      "loss": 2.3614,
      "step": 90431
    },
    {
      "epoch": 1.18,
      "grad_norm": 1.0791921615600586,
      "learning_rate": 0.00039940591108151826,
      "loss": 2.12,
      "step": 90432
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.8370486497879028,
      "learning_rate": 0.00039940205159887224,
      "loss": 2.1982,
      "step": 90433
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.860789954662323,
      "learning_rate": 0.0003993981920977455,
      "loss": 2.0688,
      "step": 90434
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.8865454196929932,
      "learning_rate": 0.0003993943325781387,
      "loss": 2.055,
      "step": 90435
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.8151876330375671,
      "learning_rate": 0.0003993904730400527,
      "loss": 2.3225,
      "step": 90436
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.9297641515731812,
      "learning_rate": 0.00039938661348348813,
      "loss": 2.1793,
      "step": 90437
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.9842233061790466,
      "learning_rate": 0.00039938275390844565,
      "loss": 2.162,
      "step": 90438
    },
    {
      "epoch": 1.18,
      "grad_norm": 1.0124187469482422,
      "learning_rate": 0.000399378894314926,
      "loss": 2.1197,
      "step": 90439
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.9304620623588562,
      "learning_rate": 0.00039937503470293003,
      "loss": 2.2145,
      "step": 90440
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.9585233330726624,
      "learning_rate": 0.00039937117507245836,
      "loss": 2.1124,
      "step": 90441
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.8612102270126343,
      "learning_rate": 0.0003993673154235117,
      "loss": 2.371,
      "step": 90442
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.862555742263794,
      "learning_rate": 0.00039936345575609076,
      "loss": 2.2254,
      "step": 90443
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.939824640750885,
      "learning_rate": 0.00039935959607019625,
      "loss": 2.3652,
      "step": 90444
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.8740641474723816,
      "learning_rate": 0.00039935573636582904,
      "loss": 1.9792,
      "step": 90445
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.8623647093772888,
      "learning_rate": 0.00039935187664298953,
      "loss": 2.1031,
      "step": 90446
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.8973150253295898,
      "learning_rate": 0.0003993480169016788,
      "loss": 1.8372,
      "step": 90447
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.8296917676925659,
      "learning_rate": 0.00039934415714189725,
      "loss": 2.2621,
      "step": 90448
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.8248698711395264,
      "learning_rate": 0.0003993402973636459,
      "loss": 2.1715,
      "step": 90449
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.9450652599334717,
      "learning_rate": 0.0003993364375669252,
      "loss": 2.3049,
      "step": 90450
    },
    {
      "epoch": 1.18,
      "grad_norm": 1.0143680572509766,
      "learning_rate": 0.000399332577751736,
      "loss": 2.2113,
      "step": 90451
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.8405209183692932,
      "learning_rate": 0.0003993287179180791,
      "loss": 2.1356,
      "step": 90452
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.8160404562950134,
      "learning_rate": 0.000399324858065955,
      "loss": 2.0404,
      "step": 90453
    },
    {
      "epoch": 1.18,
      "grad_norm": 1.1136226654052734,
      "learning_rate": 0.0003993209981953646,
      "loss": 2.1499,
      "step": 90454
    },
    {
      "epoch": 1.18,
      "grad_norm": 1.0977551937103271,
      "learning_rate": 0.0003993171383063085,
      "loss": 2.3053,
      "step": 90455
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.8332354426383972,
      "learning_rate": 0.0003993132783987875,
      "loss": 2.172,
      "step": 90456
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.7794308662414551,
      "learning_rate": 0.0003993094184728023,
      "loss": 2.0708,
      "step": 90457
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.9946740865707397,
      "learning_rate": 0.00039930555852835364,
      "loss": 2.3998,
      "step": 90458
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.9627904295921326,
      "learning_rate": 0.0003993016985654421,
      "loss": 2.0496,
      "step": 90459
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.8513745665550232,
      "learning_rate": 0.0003992978385840685,
      "loss": 2.065,
      "step": 90460
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.869181215763092,
      "learning_rate": 0.00039929397858423367,
      "loss": 2.137,
      "step": 90461
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.8869834542274475,
      "learning_rate": 0.00039929011856593816,
      "loss": 2.1957,
      "step": 90462
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.8630252480506897,
      "learning_rate": 0.0003992862585291827,
      "loss": 2.2358,
      "step": 90463
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.866809606552124,
      "learning_rate": 0.0003992823984739681,
      "loss": 2.1833,
      "step": 90464
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.873958170413971,
      "learning_rate": 0.00039927853840029503,
      "loss": 2.268,
      "step": 90465
    },
    {
      "epoch": 1.18,
      "grad_norm": 1.0334018468856812,
      "learning_rate": 0.00039927467830816417,
      "loss": 2.2767,
      "step": 90466
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.8078314065933228,
      "learning_rate": 0.00039927081819757633,
      "loss": 2.0402,
      "step": 90467
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.8290252089500427,
      "learning_rate": 0.0003992669580685321,
      "loss": 2.0935,
      "step": 90468
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.825760543346405,
      "learning_rate": 0.00039926309792103225,
      "loss": 1.9027,
      "step": 90469
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.9248058795928955,
      "learning_rate": 0.0003992592377550776,
      "loss": 2.2959,
      "step": 90470
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.7769004106521606,
      "learning_rate": 0.0003992553775706688,
      "loss": 2.4504,
      "step": 90471
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.8555524349212646,
      "learning_rate": 0.0003992515173678064,
      "loss": 2.2786,
      "step": 90472
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.9132354259490967,
      "learning_rate": 0.0003992476571464915,
      "loss": 2.2679,
      "step": 90473
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.8198394179344177,
      "learning_rate": 0.0003992437969067244,
      "loss": 2.3846,
      "step": 90474
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.8034825325012207,
      "learning_rate": 0.00039923993664850606,
      "loss": 1.9924,
      "step": 90475
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.7592182159423828,
      "learning_rate": 0.0003992360763718372,
      "loss": 2.1937,
      "step": 90476
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.9359560012817383,
      "learning_rate": 0.00039923221607671844,
      "loss": 2.1736,
      "step": 90477
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.9245507121086121,
      "learning_rate": 0.00039922835576315045,
      "loss": 2.2185,
      "step": 90478
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.8484228849411011,
      "learning_rate": 0.00039922449543113416,
      "loss": 2.2549,
      "step": 90479
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.8173328638076782,
      "learning_rate": 0.00039922063508067006,
      "loss": 2.1233,
      "step": 90480
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.7802678942680359,
      "learning_rate": 0.000399216774711759,
      "loss": 2.3931,
      "step": 90481
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.8554244637489319,
      "learning_rate": 0.0003992129143244017,
      "loss": 2.2408,
      "step": 90482
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.8562061786651611,
      "learning_rate": 0.0003992090539185989,
      "loss": 2.0004,
      "step": 90483
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.9471389651298523,
      "learning_rate": 0.0003992051934943512,
      "loss": 2.1924,
      "step": 90484
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.9055294990539551,
      "learning_rate": 0.00039920133305165935,
      "loss": 2.174,
      "step": 90485
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.8266515135765076,
      "learning_rate": 0.0003991974725905241,
      "loss": 2.0977,
      "step": 90486
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.8675872087478638,
      "learning_rate": 0.00039919361211094624,
      "loss": 2.2562,
      "step": 90487
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.8478172421455383,
      "learning_rate": 0.0003991897516129264,
      "loss": 2.083,
      "step": 90488
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.876782238483429,
      "learning_rate": 0.0003991858910964652,
      "loss": 2.0445,
      "step": 90489
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.7952460050582886,
      "learning_rate": 0.0003991820305615637,
      "loss": 2.3225,
      "step": 90490
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.9497061371803284,
      "learning_rate": 0.0003991781700082222,
      "loss": 2.0287,
      "step": 90491
    },
    {
      "epoch": 1.18,
      "grad_norm": 1.1744755506515503,
      "learning_rate": 0.0003991743094364416,
      "loss": 2.3603,
      "step": 90492
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.9014901518821716,
      "learning_rate": 0.0003991704488462227,
      "loss": 2.2731,
      "step": 90493
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.8716611266136169,
      "learning_rate": 0.00039916658823756614,
      "loss": 2.1331,
      "step": 90494
    },
    {
      "epoch": 1.18,
      "grad_norm": 1.0433402061462402,
      "learning_rate": 0.0003991627276104726,
      "loss": 2.0125,
      "step": 90495
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.9281392097473145,
      "learning_rate": 0.00039915886696494286,
      "loss": 2.3856,
      "step": 90496
    },
    {
      "epoch": 1.18,
      "grad_norm": 1.28067147731781,
      "learning_rate": 0.0003991550063009776,
      "loss": 2.306,
      "step": 90497
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.9707080721855164,
      "learning_rate": 0.00039915114561857757,
      "loss": 2.0933,
      "step": 90498
    },
    {
      "epoch": 1.18,
      "grad_norm": 1.3614856004714966,
      "learning_rate": 0.0003991472849177434,
      "loss": 1.9841,
      "step": 90499
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.7956293225288391,
      "learning_rate": 0.00039914342419847607,
      "loss": 2.3929,
      "step": 90500
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.8560549020767212,
      "learning_rate": 0.00039913956346077594,
      "loss": 2.11,
      "step": 90501
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.9965440034866333,
      "learning_rate": 0.0003991357027046439,
      "loss": 2.2646,
      "step": 90502
    },
    {
      "epoch": 1.18,
      "grad_norm": 1.0965516567230225,
      "learning_rate": 0.00039913184193008077,
      "loss": 2.2085,
      "step": 90503
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.8673638105392456,
      "learning_rate": 0.00039912798113708704,
      "loss": 2.1681,
      "step": 90504
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.8980491161346436,
      "learning_rate": 0.00039912412032566355,
      "loss": 2.0261,
      "step": 90505
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.979439377784729,
      "learning_rate": 0.00039912025949581106,
      "loss": 2.1349,
      "step": 90506
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.8749734163284302,
      "learning_rate": 0.0003991163986475303,
      "loss": 2.1803,
      "step": 90507
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.7866502404212952,
      "learning_rate": 0.0003991125377808218,
      "loss": 2.282,
      "step": 90508
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.8946229815483093,
      "learning_rate": 0.00039910867689568654,
      "loss": 2.1687,
      "step": 90509
    },
    {
      "epoch": 1.18,
      "grad_norm": 1.0146946907043457,
      "learning_rate": 0.000399104815992125,
      "loss": 2.2715,
      "step": 90510
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.8958380222320557,
      "learning_rate": 0.00039910095507013804,
      "loss": 2.3246,
      "step": 90511
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.7488220930099487,
      "learning_rate": 0.00039909709412972637,
      "loss": 1.9574,
      "step": 90512
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.8699423670768738,
      "learning_rate": 0.00039909323317089063,
      "loss": 2.1276,
      "step": 90513
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.8891984224319458,
      "learning_rate": 0.00039908937219363164,
      "loss": 2.0668,
      "step": 90514
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.9022918939590454,
      "learning_rate": 0.00039908551119795,
      "loss": 2.2652,
      "step": 90515
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.8655213713645935,
      "learning_rate": 0.0003990816501838466,
      "loss": 2.0858,
      "step": 90516
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.8122138381004333,
      "learning_rate": 0.000399077789151322,
      "loss": 2.1873,
      "step": 90517
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.7601317167282104,
      "learning_rate": 0.0003990739281003769,
      "loss": 2.2889,
      "step": 90518
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.8810179233551025,
      "learning_rate": 0.0003990700670310122,
      "loss": 2.1146,
      "step": 90519
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.7792108654975891,
      "learning_rate": 0.0003990662059432284,
      "loss": 2.1645,
      "step": 90520
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.8127318620681763,
      "learning_rate": 0.00039906234483702633,
      "loss": 2.3363,
      "step": 90521
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.848781406879425,
      "learning_rate": 0.00039905848371240676,
      "loss": 2.1149,
      "step": 90522
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.9080432653427124,
      "learning_rate": 0.00039905462256937033,
      "loss": 2.1377,
      "step": 90523
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.777908980846405,
      "learning_rate": 0.00039905076140791775,
      "loss": 2.087,
      "step": 90524
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.8082265257835388,
      "learning_rate": 0.00039904690022804983,
      "loss": 2.1346,
      "step": 90525
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.8995251059532166,
      "learning_rate": 0.0003990430390297672,
      "loss": 2.1337,
      "step": 90526
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.8595025539398193,
      "learning_rate": 0.0003990391778130705,
      "loss": 1.962,
      "step": 90527
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.8563105463981628,
      "learning_rate": 0.00039903531657796074,
      "loss": 2.0259,
      "step": 90528
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.9238274097442627,
      "learning_rate": 0.00039903145532443825,
      "loss": 2.0668,
      "step": 90529
    },
    {
      "epoch": 1.18,
      "grad_norm": 1.1418617963790894,
      "learning_rate": 0.000399027594052504,
      "loss": 2.0451,
      "step": 90530
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.9355007410049438,
      "learning_rate": 0.00039902373276215876,
      "loss": 2.1447,
      "step": 90531
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.8694589138031006,
      "learning_rate": 0.000399019871453403,
      "loss": 2.1486,
      "step": 90532
    },
    {
      "epoch": 1.18,
      "grad_norm": 1.0212860107421875,
      "learning_rate": 0.0003990160101262376,
      "loss": 2.3642,
      "step": 90533
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.8611135482788086,
      "learning_rate": 0.0003990121487806633,
      "loss": 2.3388,
      "step": 90534
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.9788224101066589,
      "learning_rate": 0.00039900828741668075,
      "loss": 2.0053,
      "step": 90535
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.8251885175704956,
      "learning_rate": 0.00039900442603429075,
      "loss": 2.0195,
      "step": 90536
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.9650534391403198,
      "learning_rate": 0.0003990005646334939,
      "loss": 1.9127,
      "step": 90537
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.8070882558822632,
      "learning_rate": 0.000398996703214291,
      "loss": 2.0285,
      "step": 90538
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.8635656833648682,
      "learning_rate": 0.0003989928417766827,
      "loss": 2.1055,
      "step": 90539
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.995276153087616,
      "learning_rate": 0.0003989889803206698,
      "loss": 2.0501,
      "step": 90540
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.8503230214118958,
      "learning_rate": 0.0003989851188462529,
      "loss": 1.939,
      "step": 90541
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.9653828740119934,
      "learning_rate": 0.0003989812573534329,
      "loss": 2.0041,
      "step": 90542
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.8691577315330505,
      "learning_rate": 0.0003989773958422104,
      "loss": 2.272,
      "step": 90543
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.9455661177635193,
      "learning_rate": 0.0003989735343125861,
      "loss": 2.1707,
      "step": 90544
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.8762766718864441,
      "learning_rate": 0.0003989696727645608,
      "loss": 1.8214,
      "step": 90545
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.851076602935791,
      "learning_rate": 0.0003989658111981352,
      "loss": 2.108,
      "step": 90546
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.76651930809021,
      "learning_rate": 0.00039896194961330984,
      "loss": 2.2491,
      "step": 90547
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.8599895238876343,
      "learning_rate": 0.00039895808801008564,
      "loss": 2.1237,
      "step": 90548
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.8391627669334412,
      "learning_rate": 0.00039895422638846334,
      "loss": 1.9605,
      "step": 90549
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.8432830572128296,
      "learning_rate": 0.0003989503647484435,
      "loss": 2.357,
      "step": 90550
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.8244757056236267,
      "learning_rate": 0.00039894650309002695,
      "loss": 2.2975,
      "step": 90551
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.8141725659370422,
      "learning_rate": 0.00039894264141321443,
      "loss": 2.1546,
      "step": 90552
    },
    {
      "epoch": 1.18,
      "grad_norm": 1.07855224609375,
      "learning_rate": 0.0003989387797180065,
      "loss": 2.1813,
      "step": 90553
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.7782959342002869,
      "learning_rate": 0.00039893491800440403,
      "loss": 2.1577,
      "step": 90554
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.8678984045982361,
      "learning_rate": 0.00039893105627240773,
      "loss": 2.2882,
      "step": 90555
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.8306193947792053,
      "learning_rate": 0.00039892719452201827,
      "loss": 2.2061,
      "step": 90556
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.7926983833312988,
      "learning_rate": 0.0003989233327532363,
      "loss": 2.226,
      "step": 90557
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.8554521203041077,
      "learning_rate": 0.0003989194709660627,
      "loss": 2.1778,
      "step": 90558
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.854108452796936,
      "learning_rate": 0.00039891560916049815,
      "loss": 2.3753,
      "step": 90559
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.9420194625854492,
      "learning_rate": 0.0003989117473365432,
      "loss": 2.0082,
      "step": 90560
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.8150070905685425,
      "learning_rate": 0.00039890788549419874,
      "loss": 2.2466,
      "step": 90561
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.7774204611778259,
      "learning_rate": 0.00039890402363346547,
      "loss": 2.1992,
      "step": 90562
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.7649496793746948,
      "learning_rate": 0.00039890016175434396,
      "loss": 2.1644,
      "step": 90563
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.8539411425590515,
      "learning_rate": 0.0003988962998568352,
      "loss": 2.1704,
      "step": 90564
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.8419088125228882,
      "learning_rate": 0.00039889243794093967,
      "loss": 1.9942,
      "step": 90565
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.9711941480636597,
      "learning_rate": 0.00039888857600665814,
      "loss": 2.2863,
      "step": 90566
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.9294037818908691,
      "learning_rate": 0.0003988847140539914,
      "loss": 2.4735,
      "step": 90567
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.8241016268730164,
      "learning_rate": 0.00039888085208294007,
      "loss": 2.3648,
      "step": 90568
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.8013477325439453,
      "learning_rate": 0.00039887699009350495,
      "loss": 1.967,
      "step": 90569
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.7866094708442688,
      "learning_rate": 0.00039887312808568675,
      "loss": 2.2828,
      "step": 90570
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.869877815246582,
      "learning_rate": 0.00039886926605948623,
      "loss": 1.9911,
      "step": 90571
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.815815806388855,
      "learning_rate": 0.000398865404014904,
      "loss": 2.1044,
      "step": 90572
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.8024396300315857,
      "learning_rate": 0.0003988615419519407,
      "loss": 1.9584,
      "step": 90573
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.8524574041366577,
      "learning_rate": 0.0003988576798705973,
      "loss": 1.9311,
      "step": 90574
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.8236387372016907,
      "learning_rate": 0.00039885381777087447,
      "loss": 2.2033,
      "step": 90575
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.8230358362197876,
      "learning_rate": 0.00039884995565277274,
      "loss": 2.1083,
      "step": 90576
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.8523496985435486,
      "learning_rate": 0.00039884609351629286,
      "loss": 2.1617,
      "step": 90577
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.7910041213035583,
      "learning_rate": 0.00039884223136143576,
      "loss": 2.039,
      "step": 90578
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.8568594455718994,
      "learning_rate": 0.0003988383691882019,
      "loss": 2.2124,
      "step": 90579
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.8401247262954712,
      "learning_rate": 0.00039883450699659226,
      "loss": 2.1689,
      "step": 90580
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.8789377808570862,
      "learning_rate": 0.00039883064478660733,
      "loss": 2.0092,
      "step": 90581
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.8094179034233093,
      "learning_rate": 0.000398826782558248,
      "loss": 1.9898,
      "step": 90582
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.8690558075904846,
      "learning_rate": 0.00039882292031151483,
      "loss": 2.2564,
      "step": 90583
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.8389318585395813,
      "learning_rate": 0.0003988190580464086,
      "loss": 2.1932,
      "step": 90584
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.8680243492126465,
      "learning_rate": 0.00039881519576293,
      "loss": 2.3175,
      "step": 90585
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.8908098340034485,
      "learning_rate": 0.00039881133346108,
      "loss": 2.1059,
      "step": 90586
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.8590083718299866,
      "learning_rate": 0.0003988074711408589,
      "loss": 2.2567,
      "step": 90587
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.818998396396637,
      "learning_rate": 0.0003988036088022677,
      "loss": 2.3631,
      "step": 90588
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.828924834728241,
      "learning_rate": 0.0003987997464453071,
      "loss": 2.3885,
      "step": 90589
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.8208015561103821,
      "learning_rate": 0.00039879588406997767,
      "loss": 2.2885,
      "step": 90590
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.8427003622055054,
      "learning_rate": 0.00039879202167628026,
      "loss": 2.1998,
      "step": 90591
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.7374893426895142,
      "learning_rate": 0.0003987881592642156,
      "loss": 2.1088,
      "step": 90592
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.8428910970687866,
      "learning_rate": 0.00039878429683378423,
      "loss": 2.1751,
      "step": 90593
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.9723501801490784,
      "learning_rate": 0.0003987804343849871,
      "loss": 1.9911,
      "step": 90594
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.8196606636047363,
      "learning_rate": 0.00039877657191782483,
      "loss": 2.0163,
      "step": 90595
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.853535532951355,
      "learning_rate": 0.00039877270943229805,
      "loss": 2.321,
      "step": 90596
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.8486524224281311,
      "learning_rate": 0.0003987688469284076,
      "loss": 2.3148,
      "step": 90597
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.881869912147522,
      "learning_rate": 0.00039876498440615416,
      "loss": 2.2302,
      "step": 90598
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.7937051057815552,
      "learning_rate": 0.0003987611218655384,
      "loss": 2.0774,
      "step": 90599
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.8528622388839722,
      "learning_rate": 0.00039875725930656116,
      "loss": 2.2023,
      "step": 90600
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.7603968381881714,
      "learning_rate": 0.00039875339672922305,
      "loss": 2.0636,
      "step": 90601
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.8612654209136963,
      "learning_rate": 0.0003987495341335248,
      "loss": 2.1764,
      "step": 90602
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.8206590414047241,
      "learning_rate": 0.00039874567151946714,
      "loss": 2.0695,
      "step": 90603
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.8548928499221802,
      "learning_rate": 0.0003987418088870509,
      "loss": 1.9941,
      "step": 90604
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.8140450716018677,
      "learning_rate": 0.00039873794623627654,
      "loss": 2.1001,
      "step": 90605
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.8601183891296387,
      "learning_rate": 0.00039873408356714505,
      "loss": 2.2251,
      "step": 90606
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.9252188801765442,
      "learning_rate": 0.000398730220879657,
      "loss": 2.1891,
      "step": 90607
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.7960143089294434,
      "learning_rate": 0.00039872635817381305,
      "loss": 2.142,
      "step": 90608
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.737636148929596,
      "learning_rate": 0.0003987224954496141,
      "loss": 1.9596,
      "step": 90609
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.7598845958709717,
      "learning_rate": 0.0003987186327070608,
      "loss": 2.0362,
      "step": 90610
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.8654311895370483,
      "learning_rate": 0.00039871476994615375,
      "loss": 2.0796,
      "step": 90611
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.8772121071815491,
      "learning_rate": 0.00039871090716689383,
      "loss": 2.0612,
      "step": 90612
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.8455052375793457,
      "learning_rate": 0.00039870704436928173,
      "loss": 2.2019,
      "step": 90613
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.7667494416236877,
      "learning_rate": 0.00039870318155331804,
      "loss": 2.2949,
      "step": 90614
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.9776245951652527,
      "learning_rate": 0.0003986993187190035,
      "loss": 2.3011,
      "step": 90615
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.807108461856842,
      "learning_rate": 0.00039869545586633905,
      "loss": 1.9053,
      "step": 90616
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.9195411801338196,
      "learning_rate": 0.0003986915929953252,
      "loss": 2.1478,
      "step": 90617
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.9142037034034729,
      "learning_rate": 0.00039868773010596267,
      "loss": 2.236,
      "step": 90618
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.8481085300445557,
      "learning_rate": 0.0003986838671982523,
      "loss": 2.352,
      "step": 90619
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.8038387298583984,
      "learning_rate": 0.0003986800042721947,
      "loss": 2.0204,
      "step": 90620
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.8776872158050537,
      "learning_rate": 0.00039867614132779055,
      "loss": 2.1022,
      "step": 90621
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.7654587030410767,
      "learning_rate": 0.0003986722783650408,
      "loss": 2.0809,
      "step": 90622
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.8807197213172913,
      "learning_rate": 0.00039866841538394585,
      "loss": 1.9896,
      "step": 90623
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.7907038927078247,
      "learning_rate": 0.00039866455238450664,
      "loss": 2.0852,
      "step": 90624
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.7968574166297913,
      "learning_rate": 0.0003986606893667239,
      "loss": 2.0334,
      "step": 90625
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.8130964636802673,
      "learning_rate": 0.0003986568263305982,
      "loss": 1.9351,
      "step": 90626
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.8481361269950867,
      "learning_rate": 0.00039865296327613035,
      "loss": 2.2997,
      "step": 90627
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.8195872902870178,
      "learning_rate": 0.00039864910020332106,
      "loss": 2.1782,
      "step": 90628
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.913848340511322,
      "learning_rate": 0.00039864523711217103,
      "loss": 2.1581,
      "step": 90629
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.9101824164390564,
      "learning_rate": 0.00039864137400268096,
      "loss": 2.2966,
      "step": 90630
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.9710148572921753,
      "learning_rate": 0.00039863751087485167,
      "loss": 2.316,
      "step": 90631
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.868980348110199,
      "learning_rate": 0.0003986336477286838,
      "loss": 2.0473,
      "step": 90632
    },
    {
      "epoch": 1.18,
      "grad_norm": 1.0741095542907715,
      "learning_rate": 0.00039862978456417796,
      "loss": 2.4771,
      "step": 90633
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.7803080677986145,
      "learning_rate": 0.00039862592138133516,
      "loss": 2.0393,
      "step": 90634
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.859747588634491,
      "learning_rate": 0.0003986220581801558,
      "loss": 2.0746,
      "step": 90635
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.7870720624923706,
      "learning_rate": 0.00039861819496064073,
      "loss": 2.1683,
      "step": 90636
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.8463173508644104,
      "learning_rate": 0.00039861433172279077,
      "loss": 2.187,
      "step": 90637
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.8769242763519287,
      "learning_rate": 0.00039861046846660647,
      "loss": 2.1485,
      "step": 90638
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.8126452565193176,
      "learning_rate": 0.00039860660519208863,
      "loss": 2.1675,
      "step": 90639
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.8625630140304565,
      "learning_rate": 0.000398602741899238,
      "loss": 2.2378,
      "step": 90640
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.8068616390228271,
      "learning_rate": 0.00039859887858805524,
      "loss": 2.2086,
      "step": 90641
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.8272172808647156,
      "learning_rate": 0.0003985950152585411,
      "loss": 2.1494,
      "step": 90642
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.9629635810852051,
      "learning_rate": 0.00039859115191069627,
      "loss": 2.2392,
      "step": 90643
    },
    {
      "epoch": 1.18,
      "grad_norm": 2.017051935195923,
      "learning_rate": 0.0003985872885445215,
      "loss": 2.3419,
      "step": 90644
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.9812703728675842,
      "learning_rate": 0.0003985834251600174,
      "loss": 2.2479,
      "step": 90645
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.868901789188385,
      "learning_rate": 0.0003985795617571849,
      "loss": 2.4281,
      "step": 90646
    },
    {
      "epoch": 1.18,
      "grad_norm": 1.0997146368026733,
      "learning_rate": 0.0003985756983360246,
      "loss": 2.0325,
      "step": 90647
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.8252359628677368,
      "learning_rate": 0.0003985718348965371,
      "loss": 2.2994,
      "step": 90648
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.8539138436317444,
      "learning_rate": 0.00039856797143872336,
      "loss": 2.3084,
      "step": 90649
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.8269137144088745,
      "learning_rate": 0.00039856410796258393,
      "loss": 2.2748,
      "step": 90650
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.9379040002822876,
      "learning_rate": 0.00039856024446811953,
      "loss": 2.1306,
      "step": 90651
    },
    {
      "epoch": 1.18,
      "grad_norm": 1.1071079969406128,
      "learning_rate": 0.00039855638095533097,
      "loss": 2.1459,
      "step": 90652
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.7840721011161804,
      "learning_rate": 0.0003985525174242189,
      "loss": 2.0148,
      "step": 90653
    },
    {
      "epoch": 1.18,
      "grad_norm": 1.0143166780471802,
      "learning_rate": 0.00039854865387478405,
      "loss": 2.2343,
      "step": 90654
    },
    {
      "epoch": 1.18,
      "grad_norm": 1.0711666345596313,
      "learning_rate": 0.00039854479030702715,
      "loss": 2.2236,
      "step": 90655
    },
    {
      "epoch": 1.18,
      "grad_norm": 1.0292518138885498,
      "learning_rate": 0.00039854092672094896,
      "loss": 2.4019,
      "step": 90656
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.8488122224807739,
      "learning_rate": 0.0003985370631165501,
      "loss": 2.2361,
      "step": 90657
    },
    {
      "epoch": 1.18,
      "grad_norm": 1.1526089906692505,
      "learning_rate": 0.0003985331994938313,
      "loss": 1.9976,
      "step": 90658
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.9577704071998596,
      "learning_rate": 0.0003985293358527934,
      "loss": 2.298,
      "step": 90659
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.8803623914718628,
      "learning_rate": 0.00039852547219343696,
      "loss": 2.1934,
      "step": 90660
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.8833906054496765,
      "learning_rate": 0.00039852160851576283,
      "loss": 2.3145,
      "step": 90661
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.8150774240493774,
      "learning_rate": 0.0003985177448197716,
      "loss": 2.2183,
      "step": 90662
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.858715832233429,
      "learning_rate": 0.00039851388110546417,
      "loss": 2.6657,
      "step": 90663
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.7920517325401306,
      "learning_rate": 0.00039851001737284115,
      "loss": 2.555,
      "step": 90664
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.8223790526390076,
      "learning_rate": 0.00039850615362190314,
      "loss": 2.1927,
      "step": 90665
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.8455782532691956,
      "learning_rate": 0.00039850228985265106,
      "loss": 2.1682,
      "step": 90666
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.8414415121078491,
      "learning_rate": 0.00039849842606508556,
      "loss": 2.1128,
      "step": 90667
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.7871294021606445,
      "learning_rate": 0.00039849456225920734,
      "loss": 2.2879,
      "step": 90668
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.8909830451011658,
      "learning_rate": 0.000398490698435017,
      "loss": 2.2355,
      "step": 90669
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.8916106820106506,
      "learning_rate": 0.0003984868345925156,
      "loss": 2.2205,
      "step": 90670
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.7765231132507324,
      "learning_rate": 0.0003984829707317035,
      "loss": 2.1533,
      "step": 90671
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.7679768204689026,
      "learning_rate": 0.0003984791068525815,
      "loss": 2.1939,
      "step": 90672
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.8300264477729797,
      "learning_rate": 0.00039847524295515043,
      "loss": 2.1396,
      "step": 90673
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.7603100538253784,
      "learning_rate": 0.00039847137903941104,
      "loss": 2.0745,
      "step": 90674
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.8482080698013306,
      "learning_rate": 0.00039846751510536386,
      "loss": 2.1335,
      "step": 90675
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.784024715423584,
      "learning_rate": 0.0003984636511530097,
      "loss": 2.093,
      "step": 90676
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.8335357904434204,
      "learning_rate": 0.00039845978718234935,
      "loss": 2.0102,
      "step": 90677
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.7487614154815674,
      "learning_rate": 0.0003984559231933835,
      "loss": 2.1589,
      "step": 90678
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.7462965250015259,
      "learning_rate": 0.00039845205918611276,
      "loss": 2.2526,
      "step": 90679
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.8377357721328735,
      "learning_rate": 0.000398448195160538,
      "loss": 2.1944,
      "step": 90680
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.8006195425987244,
      "learning_rate": 0.0003984443311166598,
      "loss": 1.917,
      "step": 90681
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.8901649117469788,
      "learning_rate": 0.0003984404670544789,
      "loss": 2.3038,
      "step": 90682
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.7924896478652954,
      "learning_rate": 0.0003984366029739961,
      "loss": 2.1914,
      "step": 90683
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.9199590086936951,
      "learning_rate": 0.0003984327388752121,
      "loss": 2.1487,
      "step": 90684
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.7953084707260132,
      "learning_rate": 0.0003984288747581276,
      "loss": 2.1763,
      "step": 90685
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.9021649360656738,
      "learning_rate": 0.0003984250106227433,
      "loss": 2.0258,
      "step": 90686
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.8976247906684875,
      "learning_rate": 0.00039842114646905985,
      "loss": 2.3228,
      "step": 90687
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.9217321276664734,
      "learning_rate": 0.00039841728229707816,
      "loss": 2.27,
      "step": 90688
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.8365147709846497,
      "learning_rate": 0.0003984134181067989,
      "loss": 2.4248,
      "step": 90689
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.7698301672935486,
      "learning_rate": 0.0003984095538982226,
      "loss": 2.2262,
      "step": 90690
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.9564131498336792,
      "learning_rate": 0.0003984056896713501,
      "loss": 1.9734,
      "step": 90691
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.8592941761016846,
      "learning_rate": 0.0003984018254261822,
      "loss": 2.2454,
      "step": 90692
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.8280971646308899,
      "learning_rate": 0.0003983979611627195,
      "loss": 2.0441,
      "step": 90693
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.9838843941688538,
      "learning_rate": 0.0003983940968809627,
      "loss": 2.2842,
      "step": 90694
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.849004328250885,
      "learning_rate": 0.0003983902325809128,
      "loss": 1.9782,
      "step": 90695
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.9280071258544922,
      "learning_rate": 0.00039838636826257005,
      "loss": 2.3199,
      "step": 90696
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.8113858103752136,
      "learning_rate": 0.0003983825039259355,
      "loss": 2.0192,
      "step": 90697
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.8346484899520874,
      "learning_rate": 0.0003983786395710099,
      "loss": 2.2831,
      "step": 90698
    },
    {
      "epoch": 1.18,
      "grad_norm": 1.0741099119186401,
      "learning_rate": 0.0003983747751977937,
      "loss": 2.2461,
      "step": 90699
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.7412177324295044,
      "learning_rate": 0.00039837091080628786,
      "loss": 2.0739,
      "step": 90700
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.9195617437362671,
      "learning_rate": 0.00039836704639649304,
      "loss": 2.184,
      "step": 90701
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.816356897354126,
      "learning_rate": 0.00039836318196840984,
      "loss": 2.1602,
      "step": 90702
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.9670145511627197,
      "learning_rate": 0.00039835931752203903,
      "loss": 2.0294,
      "step": 90703
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.8477356433868408,
      "learning_rate": 0.0003983554530573815,
      "loss": 2.0571,
      "step": 90704
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.7840749621391296,
      "learning_rate": 0.00039835158857443774,
      "loss": 2.2317,
      "step": 90705
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.9400672316551208,
      "learning_rate": 0.0003983477240732086,
      "loss": 2.1702,
      "step": 90706
    },
    {
      "epoch": 1.18,
      "grad_norm": 1.1194559335708618,
      "learning_rate": 0.00039834385955369477,
      "loss": 2.3394,
      "step": 90707
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.9383228421211243,
      "learning_rate": 0.00039833999501589693,
      "loss": 2.1094,
      "step": 90708
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.8656646013259888,
      "learning_rate": 0.0003983361304598158,
      "loss": 2.1106,
      "step": 90709
    },
    {
      "epoch": 1.18,
      "grad_norm": 1.1070350408554077,
      "learning_rate": 0.0003983322658854522,
      "loss": 2.0306,
      "step": 90710
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.9983418583869934,
      "learning_rate": 0.00039832840129280675,
      "loss": 2.1304,
      "step": 90711
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.8438981771469116,
      "learning_rate": 0.0003983245366818802,
      "loss": 2.2323,
      "step": 90712
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.993865966796875,
      "learning_rate": 0.0003983206720526733,
      "loss": 2.1449,
      "step": 90713
    },
    {
      "epoch": 1.18,
      "grad_norm": 1.206750512123108,
      "learning_rate": 0.0003983168074051866,
      "loss": 2.2816,
      "step": 90714
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.8958913683891296,
      "learning_rate": 0.0003983129427394211,
      "loss": 2.1218,
      "step": 90715
    },
    {
      "epoch": 1.18,
      "grad_norm": 1.0026718378067017,
      "learning_rate": 0.0003983090780553772,
      "loss": 2.1609,
      "step": 90716
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.9558027386665344,
      "learning_rate": 0.000398305213353056,
      "loss": 2.0289,
      "step": 90717
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.861243486404419,
      "learning_rate": 0.00039830134863245786,
      "loss": 2.0848,
      "step": 90718
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.8143578767776489,
      "learning_rate": 0.00039829748389358366,
      "loss": 2.2047,
      "step": 90719
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.8143569827079773,
      "learning_rate": 0.00039829361913643415,
      "loss": 2.3589,
      "step": 90720
    },
    {
      "epoch": 1.18,
      "grad_norm": 1.0764820575714111,
      "learning_rate": 0.00039828975436100993,
      "loss": 2.1052,
      "step": 90721
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.8931829929351807,
      "learning_rate": 0.00039828588956731175,
      "loss": 2.297,
      "step": 90722
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.891735315322876,
      "learning_rate": 0.00039828202475534054,
      "loss": 2.2347,
      "step": 90723
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.8703308701515198,
      "learning_rate": 0.0003982781599250968,
      "loss": 2.0686,
      "step": 90724
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.936222493648529,
      "learning_rate": 0.00039827429507658117,
      "loss": 2.3134,
      "step": 90725
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.8556328415870667,
      "learning_rate": 0.0003982704302097946,
      "loss": 2.2437,
      "step": 90726
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.9294564127922058,
      "learning_rate": 0.00039826656532473764,
      "loss": 1.9723,
      "step": 90727
    },
    {
      "epoch": 1.18,
      "grad_norm": 1.0087835788726807,
      "learning_rate": 0.00039826270042141106,
      "loss": 2.14,
      "step": 90728
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.9276740550994873,
      "learning_rate": 0.00039825883549981573,
      "loss": 2.1633,
      "step": 90729
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.8808481693267822,
      "learning_rate": 0.0003982549705599521,
      "loss": 2.2522,
      "step": 90730
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.8502188324928284,
      "learning_rate": 0.00039825110560182104,
      "loss": 2.3393,
      "step": 90731
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.7782308459281921,
      "learning_rate": 0.0003982472406254233,
      "loss": 2.2744,
      "step": 90732
    },
    {
      "epoch": 1.18,
      "grad_norm": 1.0153110027313232,
      "learning_rate": 0.00039824337563075946,
      "loss": 2.1734,
      "step": 90733
    },
    {
      "epoch": 1.18,
      "grad_norm": 1.009419560432434,
      "learning_rate": 0.00039823951061783035,
      "loss": 2.1526,
      "step": 90734
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.7772400379180908,
      "learning_rate": 0.0003982356455866367,
      "loss": 2.0777,
      "step": 90735
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.9813438057899475,
      "learning_rate": 0.0003982317805371792,
      "loss": 2.2373,
      "step": 90736
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.8093855381011963,
      "learning_rate": 0.0003982279154694584,
      "loss": 2.2016,
      "step": 90737
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.8369582295417786,
      "learning_rate": 0.00039822405038347536,
      "loss": 2.0766,
      "step": 90738
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.8095256090164185,
      "learning_rate": 0.00039822018527923056,
      "loss": 2.1248,
      "step": 90739
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.8231372833251953,
      "learning_rate": 0.0003982163201567247,
      "loss": 2.1972,
      "step": 90740
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.9661749601364136,
      "learning_rate": 0.00039821245501595867,
      "loss": 2.0679,
      "step": 90741
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.7871527075767517,
      "learning_rate": 0.00039820858985693304,
      "loss": 2.2332,
      "step": 90742
    },
    {
      "epoch": 1.18,
      "grad_norm": 1.0642069578170776,
      "learning_rate": 0.0003982047246796486,
      "loss": 2.1164,
      "step": 90743
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.9827120900154114,
      "learning_rate": 0.0003982008594841061,
      "loss": 2.1317,
      "step": 90744
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.789651095867157,
      "learning_rate": 0.0003981969942703061,
      "loss": 2.3213,
      "step": 90745
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.9219282865524292,
      "learning_rate": 0.00039819312903824946,
      "loss": 2.3177,
      "step": 90746
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.879098653793335,
      "learning_rate": 0.0003981892637879368,
      "loss": 2.3091,
      "step": 90747
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.8460771441459656,
      "learning_rate": 0.0003981853985193691,
      "loss": 2.0751,
      "step": 90748
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.8447444438934326,
      "learning_rate": 0.0003981815332325467,
      "loss": 2.143,
      "step": 90749
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.9068114161491394,
      "learning_rate": 0.00039817766792747054,
      "loss": 2.222,
      "step": 90750
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.9366257786750793,
      "learning_rate": 0.00039817380260414133,
      "loss": 2.2094,
      "step": 90751
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.7893492579460144,
      "learning_rate": 0.00039816993726255975,
      "loss": 1.9464,
      "step": 90752
    },
    {
      "epoch": 1.18,
      "grad_norm": 1.2806857824325562,
      "learning_rate": 0.0003981660719027265,
      "loss": 2.0708,
      "step": 90753
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.9227041006088257,
      "learning_rate": 0.0003981622065246423,
      "loss": 2.2421,
      "step": 90754
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.7830472588539124,
      "learning_rate": 0.000398158341128308,
      "loss": 2.0766,
      "step": 90755
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.8192620873451233,
      "learning_rate": 0.0003981544757137241,
      "loss": 2.1968,
      "step": 90756
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.9019118547439575,
      "learning_rate": 0.0003981506102808915,
      "loss": 2.3245,
      "step": 90757
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.8730493187904358,
      "learning_rate": 0.0003981467448298107,
      "loss": 2.3886,
      "step": 90758
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.8906219601631165,
      "learning_rate": 0.0003981428793604828,
      "loss": 2.3909,
      "step": 90759
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.8532494306564331,
      "learning_rate": 0.0003981390138729082,
      "loss": 2.1436,
      "step": 90760
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.9046633243560791,
      "learning_rate": 0.00039813514836708764,
      "loss": 2.3952,
      "step": 90761
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.8602069616317749,
      "learning_rate": 0.00039813128284302203,
      "loss": 2.1905,
      "step": 90762
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.8426994681358337,
      "learning_rate": 0.00039812741730071185,
      "loss": 2.0667,
      "step": 90763
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.8254430294036865,
      "learning_rate": 0.0003981235517401579,
      "loss": 2.4639,
      "step": 90764
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.9172637462615967,
      "learning_rate": 0.0003981196861613611,
      "loss": 2.5198,
      "step": 90765
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.8430493474006653,
      "learning_rate": 0.00039811582056432183,
      "loss": 2.0929,
      "step": 90766
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.8440778851509094,
      "learning_rate": 0.000398111954949041,
      "loss": 2.3546,
      "step": 90767
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.7609801888465881,
      "learning_rate": 0.0003981080893155194,
      "loss": 2.1729,
      "step": 90768
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.8365024328231812,
      "learning_rate": 0.0003981042236637576,
      "loss": 1.9632,
      "step": 90769
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.8201530575752258,
      "learning_rate": 0.0003981003579937563,
      "loss": 2.1274,
      "step": 90770
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.9294760823249817,
      "learning_rate": 0.0003980964923055164,
      "loss": 2.1538,
      "step": 90771
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.9015260338783264,
      "learning_rate": 0.0003980926265990385,
      "loss": 2.1293,
      "step": 90772
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.7957854270935059,
      "learning_rate": 0.0003980887608743233,
      "loss": 2.1958,
      "step": 90773
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.8583466410636902,
      "learning_rate": 0.0003980848951313715,
      "loss": 2.2489,
      "step": 90774
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.7543392181396484,
      "learning_rate": 0.00039808102937018396,
      "loss": 2.0436,
      "step": 90775
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.809227705001831,
      "learning_rate": 0.0003980771635907612,
      "loss": 2.1709,
      "step": 90776
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.7883458137512207,
      "learning_rate": 0.0003980732977931042,
      "loss": 2.1612,
      "step": 90777
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.8476559519767761,
      "learning_rate": 0.00039806943197721336,
      "loss": 2.1658,
      "step": 90778
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.8310476541519165,
      "learning_rate": 0.0003980655661430896,
      "loss": 2.015,
      "step": 90779
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.7929078340530396,
      "learning_rate": 0.0003980617002907336,
      "loss": 2.2853,
      "step": 90780
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.8837698698043823,
      "learning_rate": 0.0003980578344201462,
      "loss": 2.3611,
      "step": 90781
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.8307301998138428,
      "learning_rate": 0.00039805396853132783,
      "loss": 1.9869,
      "step": 90782
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.8211348652839661,
      "learning_rate": 0.0003980501026242795,
      "loss": 2.181,
      "step": 90783
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.913214385509491,
      "learning_rate": 0.00039804623669900175,
      "loss": 2.1086,
      "step": 90784
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.9194164276123047,
      "learning_rate": 0.00039804237075549535,
      "loss": 2.1804,
      "step": 90785
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.8521375060081482,
      "learning_rate": 0.000398038504793761,
      "loss": 2.2796,
      "step": 90786
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.829911470413208,
      "learning_rate": 0.0003980346388137995,
      "loss": 2.1767,
      "step": 90787
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.8142623901367188,
      "learning_rate": 0.0003980307728156115,
      "loss": 2.2018,
      "step": 90788
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.8005778193473816,
      "learning_rate": 0.0003980269067991977,
      "loss": 2.1012,
      "step": 90789
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.9009433388710022,
      "learning_rate": 0.0003980230407645589,
      "loss": 2.0705,
      "step": 90790
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.7785487174987793,
      "learning_rate": 0.0003980191747116957,
      "loss": 2.0262,
      "step": 90791
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.7477966547012329,
      "learning_rate": 0.0003980153086406089,
      "loss": 2.2498,
      "step": 90792
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.8093949556350708,
      "learning_rate": 0.00039801144255129924,
      "loss": 2.0772,
      "step": 90793
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.808044970035553,
      "learning_rate": 0.0003980075764437674,
      "loss": 2.3076,
      "step": 90794
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.7526878714561462,
      "learning_rate": 0.000398003710318014,
      "loss": 2.022,
      "step": 90795
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.9372571110725403,
      "learning_rate": 0.00039799984417404,
      "loss": 2.2084,
      "step": 90796
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.7698568105697632,
      "learning_rate": 0.0003979959780118459,
      "loss": 2.0996,
      "step": 90797
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.901578426361084,
      "learning_rate": 0.00039799211183143247,
      "loss": 2.3802,
      "step": 90798
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.871104896068573,
      "learning_rate": 0.0003979882456328006,
      "loss": 2.3417,
      "step": 90799
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.9311985373497009,
      "learning_rate": 0.00039798437941595074,
      "loss": 2.3298,
      "step": 90800
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.8253499865531921,
      "learning_rate": 0.00039798051318088373,
      "loss": 2.2507,
      "step": 90801
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.8068569302558899,
      "learning_rate": 0.0003979766469276004,
      "loss": 2.4502,
      "step": 90802
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.8376089930534363,
      "learning_rate": 0.0003979727806561012,
      "loss": 2.4004,
      "step": 90803
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.7705984711647034,
      "learning_rate": 0.0003979689143663871,
      "loss": 2.2612,
      "step": 90804
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.8295395374298096,
      "learning_rate": 0.0003979650480584588,
      "loss": 2.0494,
      "step": 90805
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.9362911581993103,
      "learning_rate": 0.0003979611817323168,
      "loss": 2.1479,
      "step": 90806
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.779471755027771,
      "learning_rate": 0.0003979573153879621,
      "loss": 2.141,
      "step": 90807
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.7796382904052734,
      "learning_rate": 0.00039795344902539525,
      "loss": 2.1203,
      "step": 90808
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.8137715458869934,
      "learning_rate": 0.000397949582644617,
      "loss": 2.2857,
      "step": 90809
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.8615071177482605,
      "learning_rate": 0.000397945716245628,
      "loss": 2.2853,
      "step": 90810
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.8676972985267639,
      "learning_rate": 0.00039794184982842916,
      "loss": 2.1286,
      "step": 90811
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.8467165231704712,
      "learning_rate": 0.00039793798339302094,
      "loss": 2.4406,
      "step": 90812
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.8665116429328918,
      "learning_rate": 0.0003979341169394043,
      "loss": 2.1831,
      "step": 90813
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.8719595074653625,
      "learning_rate": 0.0003979302504675799,
      "loss": 2.2674,
      "step": 90814
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.8443176746368408,
      "learning_rate": 0.0003979263839775483,
      "loss": 2.1472,
      "step": 90815
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.8452824950218201,
      "learning_rate": 0.0003979225174693104,
      "loss": 2.2091,
      "step": 90816
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.8145936727523804,
      "learning_rate": 0.0003979186509428669,
      "loss": 2.04,
      "step": 90817
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.801394522190094,
      "learning_rate": 0.0003979147843982184,
      "loss": 2.1891,
      "step": 90818
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.9254831075668335,
      "learning_rate": 0.00039791091783536567,
      "loss": 2.3562,
      "step": 90819
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.8543451428413391,
      "learning_rate": 0.0003979070512543096,
      "loss": 1.9775,
      "step": 90820
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.8988851308822632,
      "learning_rate": 0.0003979031846550506,
      "loss": 2.5393,
      "step": 90821
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.8786851167678833,
      "learning_rate": 0.00039789931803758954,
      "loss": 2.3112,
      "step": 90822
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.9217195510864258,
      "learning_rate": 0.00039789545140192734,
      "loss": 2.2966,
      "step": 90823
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.8287109136581421,
      "learning_rate": 0.00039789158474806436,
      "loss": 2.0604,
      "step": 90824
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.8421269655227661,
      "learning_rate": 0.0003978877180760015,
      "loss": 2.1902,
      "step": 90825
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.8995243906974792,
      "learning_rate": 0.00039788385138573955,
      "loss": 2.1675,
      "step": 90826
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.8696637153625488,
      "learning_rate": 0.0003978799846772791,
      "loss": 2.3563,
      "step": 90827
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.8980440497398376,
      "learning_rate": 0.00039787611795062087,
      "loss": 2.0741,
      "step": 90828
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.8744122385978699,
      "learning_rate": 0.00039787225120576573,
      "loss": 2.2041,
      "step": 90829
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.8902894854545593,
      "learning_rate": 0.00039786838444271416,
      "loss": 2.1019,
      "step": 90830
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.9511258602142334,
      "learning_rate": 0.0003978645176614671,
      "loss": 2.2097,
      "step": 90831
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.8335737586021423,
      "learning_rate": 0.0003978606508620252,
      "loss": 2.0882,
      "step": 90832
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.8784007430076599,
      "learning_rate": 0.00039785678404438907,
      "loss": 1.9953,
      "step": 90833
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.9948261976242065,
      "learning_rate": 0.00039785291720855956,
      "loss": 2.1706,
      "step": 90834
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.8303261399269104,
      "learning_rate": 0.00039784905035453725,
      "loss": 2.1018,
      "step": 90835
    },
    {
      "epoch": 1.18,
      "grad_norm": 1.2044756412506104,
      "learning_rate": 0.0003978451834823231,
      "loss": 2.2492,
      "step": 90836
    },
    {
      "epoch": 1.18,
      "grad_norm": 1.0562571287155151,
      "learning_rate": 0.00039784131659191763,
      "loss": 2.2313,
      "step": 90837
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.7731232643127441,
      "learning_rate": 0.0003978374496833216,
      "loss": 1.9915,
      "step": 90838
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.8145661354064941,
      "learning_rate": 0.0003978335827565357,
      "loss": 2.1664,
      "step": 90839
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.8923631310462952,
      "learning_rate": 0.0003978297158115608,
      "loss": 2.0392,
      "step": 90840
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.8845577836036682,
      "learning_rate": 0.0003978258488483975,
      "loss": 2.1494,
      "step": 90841
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.7507566213607788,
      "learning_rate": 0.0003978219818670464,
      "loss": 2.1325,
      "step": 90842
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.8236594200134277,
      "learning_rate": 0.0003978181148675085,
      "loss": 2.2437,
      "step": 90843
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.8885495066642761,
      "learning_rate": 0.00039781424784978423,
      "loss": 2.1422,
      "step": 90844
    },
    {
      "epoch": 1.18,
      "grad_norm": 1.0823508501052856,
      "learning_rate": 0.0003978103808138745,
      "loss": 2.3008,
      "step": 90845
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.807127833366394,
      "learning_rate": 0.00039780651375977995,
      "loss": 2.224,
      "step": 90846
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.8610473275184631,
      "learning_rate": 0.0003978026466875014,
      "loss": 2.2088,
      "step": 90847
    },
    {
      "epoch": 1.18,
      "grad_norm": 1.0780059099197388,
      "learning_rate": 0.0003977987795970394,
      "loss": 2.1432,
      "step": 90848
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.8696928024291992,
      "learning_rate": 0.0003977949124883948,
      "loss": 2.1103,
      "step": 90849
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.9942916035652161,
      "learning_rate": 0.0003977910453615684,
      "loss": 2.3407,
      "step": 90850
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.9108550548553467,
      "learning_rate": 0.00039778717821656063,
      "loss": 2.0798,
      "step": 90851
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.7834149599075317,
      "learning_rate": 0.0003977833110533724,
      "loss": 2.3786,
      "step": 90852
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.8020594120025635,
      "learning_rate": 0.0003977794438720044,
      "loss": 2.2445,
      "step": 90853
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.8471569418907166,
      "learning_rate": 0.00039777557667245746,
      "loss": 2.1537,
      "step": 90854
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.9526691436767578,
      "learning_rate": 0.0003977717094547321,
      "loss": 2.2505,
      "step": 90855
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.7990580201148987,
      "learning_rate": 0.0003977678422188291,
      "loss": 2.2106,
      "step": 90856
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.8159064054489136,
      "learning_rate": 0.0003977639749647493,
      "loss": 2.1711,
      "step": 90857
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.9005549550056458,
      "learning_rate": 0.0003977601076924933,
      "loss": 2.25,
      "step": 90858
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.8790510296821594,
      "learning_rate": 0.0003977562404020619,
      "loss": 2.0869,
      "step": 90859
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.837820291519165,
      "learning_rate": 0.00039775237309345573,
      "loss": 2.1801,
      "step": 90860
    },
    {
      "epoch": 1.18,
      "grad_norm": 1.0102177858352661,
      "learning_rate": 0.0003977485057666755,
      "loss": 2.1717,
      "step": 90861
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.7525960803031921,
      "learning_rate": 0.000397744638421722,
      "loss": 2.2106,
      "step": 90862
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.8180612325668335,
      "learning_rate": 0.00039774077105859604,
      "loss": 2.2127,
      "step": 90863
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.9183622002601624,
      "learning_rate": 0.0003977369036772982,
      "loss": 2.1617,
      "step": 90864
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.8368081450462341,
      "learning_rate": 0.0003977330362778291,
      "loss": 2.4931,
      "step": 90865
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.9173544645309448,
      "learning_rate": 0.0003977291688601896,
      "loss": 2.0786,
      "step": 90866
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.7899281978607178,
      "learning_rate": 0.0003977253014243805,
      "loss": 2.0796,
      "step": 90867
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.8330773711204529,
      "learning_rate": 0.0003977214339704023,
      "loss": 2.0785,
      "step": 90868
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.8666138052940369,
      "learning_rate": 0.00039771756649825603,
      "loss": 1.9863,
      "step": 90869
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.9609046578407288,
      "learning_rate": 0.00039771369900794203,
      "loss": 2.0838,
      "step": 90870
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.8713641166687012,
      "learning_rate": 0.0003977098314994613,
      "loss": 2.2186,
      "step": 90871
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.9935256242752075,
      "learning_rate": 0.0003977059639728144,
      "loss": 2.2836,
      "step": 90872
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.9008185863494873,
      "learning_rate": 0.0003977020964280022,
      "loss": 2.0089,
      "step": 90873
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.8934549689292908,
      "learning_rate": 0.0003976982288650252,
      "loss": 2.0165,
      "step": 90874
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.9328208565711975,
      "learning_rate": 0.0003976943612838845,
      "loss": 2.1751,
      "step": 90875
    },
    {
      "epoch": 1.18,
      "grad_norm": 1.0293571949005127,
      "learning_rate": 0.0003976904936845803,
      "loss": 2.1193,
      "step": 90876
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.7739840149879456,
      "learning_rate": 0.0003976866260671138,
      "loss": 2.1627,
      "step": 90877
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.8012650012969971,
      "learning_rate": 0.0003976827584314854,
      "loss": 2.1509,
      "step": 90878
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.9232054352760315,
      "learning_rate": 0.000397678890777696,
      "loss": 2.14,
      "step": 90879
    },
    {
      "epoch": 1.18,
      "grad_norm": 1.0930023193359375,
      "learning_rate": 0.0003976750231057462,
      "loss": 2.1725,
      "step": 90880
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.8571339845657349,
      "learning_rate": 0.00039767115541563684,
      "loss": 2.336,
      "step": 90881
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.8523075580596924,
      "learning_rate": 0.0003976672877073685,
      "loss": 2.1802,
      "step": 90882
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.855171799659729,
      "learning_rate": 0.00039766341998094193,
      "loss": 2.1351,
      "step": 90883
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.8669852614402771,
      "learning_rate": 0.00039765955223635796,
      "loss": 2.3519,
      "step": 90884
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.8873988389968872,
      "learning_rate": 0.00039765568447361723,
      "loss": 2.1164,
      "step": 90885
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.8700819611549377,
      "learning_rate": 0.0003976518166927204,
      "loss": 2.2716,
      "step": 90886
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.9184053540229797,
      "learning_rate": 0.00039764794889366833,
      "loss": 2.1669,
      "step": 90887
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.7967288494110107,
      "learning_rate": 0.00039764408107646163,
      "loss": 2.0939,
      "step": 90888
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.8123734593391418,
      "learning_rate": 0.00039764021324110104,
      "loss": 2.0286,
      "step": 90889
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.8624240756034851,
      "learning_rate": 0.0003976363453875874,
      "loss": 2.2339,
      "step": 90890
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.8049443960189819,
      "learning_rate": 0.00039763247751592115,
      "loss": 1.9888,
      "step": 90891
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.7554110288619995,
      "learning_rate": 0.00039762860962610326,
      "loss": 2.2368,
      "step": 90892
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.8016275763511658,
      "learning_rate": 0.0003976247417181344,
      "loss": 2.0184,
      "step": 90893
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.9462225437164307,
      "learning_rate": 0.0003976208737920152,
      "loss": 2.1187,
      "step": 90894
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.843390941619873,
      "learning_rate": 0.00039761700584774645,
      "loss": 2.0184,
      "step": 90895
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.8493566513061523,
      "learning_rate": 0.0003976131378853289,
      "loss": 1.8613,
      "step": 90896
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.9264039397239685,
      "learning_rate": 0.0003976092699047632,
      "loss": 2.2017,
      "step": 90897
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.8950567841529846,
      "learning_rate": 0.0003976054019060501,
      "loss": 2.2202,
      "step": 90898
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.8495104908943176,
      "learning_rate": 0.0003976015338891903,
      "loss": 1.9852,
      "step": 90899
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.8683957457542419,
      "learning_rate": 0.00039759766585418453,
      "loss": 2.2464,
      "step": 90900
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.9989460110664368,
      "learning_rate": 0.0003975937978010335,
      "loss": 2.1253,
      "step": 90901
    },
    {
      "epoch": 1.18,
      "grad_norm": 1.0175213813781738,
      "learning_rate": 0.000397589929729738,
      "loss": 2.305,
      "step": 90902
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.8471230864524841,
      "learning_rate": 0.00039758606164029856,
      "loss": 2.2589,
      "step": 90903
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.8912479281425476,
      "learning_rate": 0.00039758219353271617,
      "loss": 2.2988,
      "step": 90904
    },
    {
      "epoch": 1.18,
      "grad_norm": 1.1503714323043823,
      "learning_rate": 0.00039757832540699133,
      "loss": 2.2329,
      "step": 90905
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.851711630821228,
      "learning_rate": 0.00039757445726312486,
      "loss": 2.0931,
      "step": 90906
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.8244279623031616,
      "learning_rate": 0.0003975705891011174,
      "loss": 2.277,
      "step": 90907
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.8035457134246826,
      "learning_rate": 0.0003975667209209699,
      "loss": 2.3054,
      "step": 90908
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.89769047498703,
      "learning_rate": 0.00039756285272268273,
      "loss": 2.0583,
      "step": 90909
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.8668264150619507,
      "learning_rate": 0.00039755898450625685,
      "loss": 2.2991,
      "step": 90910
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.9156084060668945,
      "learning_rate": 0.0003975551162716929,
      "loss": 2.1909,
      "step": 90911
    },
    {
      "epoch": 1.18,
      "grad_norm": 1.036436676979065,
      "learning_rate": 0.00039755124801899163,
      "loss": 2.3117,
      "step": 90912
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.8753367066383362,
      "learning_rate": 0.0003975473797481537,
      "loss": 2.1978,
      "step": 90913
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.9524524807929993,
      "learning_rate": 0.00039754351145917984,
      "loss": 2.248,
      "step": 90914
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.8590167760848999,
      "learning_rate": 0.00039753964315207096,
      "loss": 2.179,
      "step": 90915
    },
    {
      "epoch": 1.18,
      "grad_norm": 1.1529673337936401,
      "learning_rate": 0.0003975357748268275,
      "loss": 2.0347,
      "step": 90916
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.9974008202552795,
      "learning_rate": 0.00039753190648345027,
      "loss": 2.227,
      "step": 90917
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.8891786336898804,
      "learning_rate": 0.0003975280381219401,
      "loss": 2.195,
      "step": 90918
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.9384071230888367,
      "learning_rate": 0.0003975241697422976,
      "loss": 1.949,
      "step": 90919
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.869942307472229,
      "learning_rate": 0.0003975203013445235,
      "loss": 2.2578,
      "step": 90920
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.8729026317596436,
      "learning_rate": 0.00039751643292861857,
      "loss": 2.2682,
      "step": 90921
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.8143355846405029,
      "learning_rate": 0.00039751256449458344,
      "loss": 2.2468,
      "step": 90922
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.8555055856704712,
      "learning_rate": 0.00039750869604241883,
      "loss": 2.3694,
      "step": 90923
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.9171997308731079,
      "learning_rate": 0.00039750482757212565,
      "loss": 2.1804,
      "step": 90924
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.8703101873397827,
      "learning_rate": 0.00039750095908370445,
      "loss": 2.2797,
      "step": 90925
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.8640586137771606,
      "learning_rate": 0.00039749709057715593,
      "loss": 2.1025,
      "step": 90926
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.9141402840614319,
      "learning_rate": 0.00039749322205248085,
      "loss": 2.0857,
      "step": 90927
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.8292084336280823,
      "learning_rate": 0.00039748935350968,
      "loss": 1.8654,
      "step": 90928
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.8321743011474609,
      "learning_rate": 0.000397485484948754,
      "loss": 2.2024,
      "step": 90929
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.7934942841529846,
      "learning_rate": 0.00039748161636970363,
      "loss": 2.1224,
      "step": 90930
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.8772434592247009,
      "learning_rate": 0.00039747774777252964,
      "loss": 2.1628,
      "step": 90931
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.7808775305747986,
      "learning_rate": 0.0003974738791572326,
      "loss": 2.0311,
      "step": 90932
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.8739285469055176,
      "learning_rate": 0.0003974700105238134,
      "loss": 2.1088,
      "step": 90933
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.8255615234375,
      "learning_rate": 0.0003974661418722725,
      "loss": 2.2016,
      "step": 90934
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.7717438340187073,
      "learning_rate": 0.00039746227320261104,
      "loss": 2.1056,
      "step": 90935
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.8426916003227234,
      "learning_rate": 0.00039745840451482935,
      "loss": 2.3349,
      "step": 90936
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.9050040245056152,
      "learning_rate": 0.0003974545358089284,
      "loss": 2.2783,
      "step": 90937
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.7767096161842346,
      "learning_rate": 0.0003974506670849087,
      "loss": 2.0428,
      "step": 90938
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.8595939874649048,
      "learning_rate": 0.0003974467983427713,
      "loss": 2.0448,
      "step": 90939
    },
    {
      "epoch": 1.18,
      "grad_norm": 1.288484811782837,
      "learning_rate": 0.00039744292958251656,
      "loss": 2.4332,
      "step": 90940
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.8681483268737793,
      "learning_rate": 0.0003974390608041453,
      "loss": 2.0401,
      "step": 90941
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.8688140511512756,
      "learning_rate": 0.0003974351920076584,
      "loss": 2.0167,
      "step": 90942
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.9020826816558838,
      "learning_rate": 0.00039743132319305626,
      "loss": 2.2253,
      "step": 90943
    },
    {
      "epoch": 1.18,
      "grad_norm": 1.0125296115875244,
      "learning_rate": 0.00039742745436033994,
      "loss": 2.2815,
      "step": 90944
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.8921396136283875,
      "learning_rate": 0.00039742358550951006,
      "loss": 2.1614,
      "step": 90945
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.863678514957428,
      "learning_rate": 0.0003974197166405672,
      "loss": 2.1038,
      "step": 90946
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.9834278225898743,
      "learning_rate": 0.00039741584775351217,
      "loss": 2.1929,
      "step": 90947
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.9320855736732483,
      "learning_rate": 0.0003974119788483458,
      "loss": 2.2422,
      "step": 90948
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.7934761047363281,
      "learning_rate": 0.0003974081099250686,
      "loss": 2.0352,
      "step": 90949
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.9424647092819214,
      "learning_rate": 0.0003974042409836815,
      "loss": 2.1805,
      "step": 90950
    },
    {
      "epoch": 1.18,
      "grad_norm": 1.0275098085403442,
      "learning_rate": 0.000397400372024185,
      "loss": 2.2728,
      "step": 90951
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.8423042893409729,
      "learning_rate": 0.00039739650304658,
      "loss": 2.1984,
      "step": 90952
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.9038714170455933,
      "learning_rate": 0.00039739263405086714,
      "loss": 2.5096,
      "step": 90953
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.8799564838409424,
      "learning_rate": 0.00039738876503704715,
      "loss": 2.2524,
      "step": 90954
    },
    {
      "epoch": 1.18,
      "grad_norm": 1.1466891765594482,
      "learning_rate": 0.0003973848960051207,
      "loss": 2.2059,
      "step": 90955
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.9020884037017822,
      "learning_rate": 0.0003973810269550886,
      "loss": 2.0686,
      "step": 90956
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.8632010221481323,
      "learning_rate": 0.00039737715788695153,
      "loss": 2.0696,
      "step": 90957
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.851299524307251,
      "learning_rate": 0.0003973732888007102,
      "loss": 2.2766,
      "step": 90958
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.955202043056488,
      "learning_rate": 0.00039736941969636534,
      "loss": 2.1814,
      "step": 90959
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.7528490424156189,
      "learning_rate": 0.00039736555057391773,
      "loss": 2.2122,
      "step": 90960
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.826490581035614,
      "learning_rate": 0.0003973616814333679,
      "loss": 2.2043,
      "step": 90961
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.8155112862586975,
      "learning_rate": 0.00039735781227471676,
      "loss": 2.2125,
      "step": 90962
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.7881513833999634,
      "learning_rate": 0.00039735394309796504,
      "loss": 2.1723,
      "step": 90963
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.8865702748298645,
      "learning_rate": 0.00039735007390311327,
      "loss": 2.3617,
      "step": 90964
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.8839424252510071,
      "learning_rate": 0.0003973462046901623,
      "loss": 2.0495,
      "step": 90965
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.8041111826896667,
      "learning_rate": 0.00039734233545911287,
      "loss": 2.0932,
      "step": 90966
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.8387958407402039,
      "learning_rate": 0.00039733846620996566,
      "loss": 2.3456,
      "step": 90967
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.9024686813354492,
      "learning_rate": 0.0003973345969427214,
      "loss": 2.2749,
      "step": 90968
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.8885087966918945,
      "learning_rate": 0.0003973307276573808,
      "loss": 2.034,
      "step": 90969
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.7745661735534668,
      "learning_rate": 0.00039732685835394455,
      "loss": 2.2936,
      "step": 90970
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.8880694508552551,
      "learning_rate": 0.00039732298903241334,
      "loss": 2.0941,
      "step": 90971
    },
    {
      "epoch": 1.18,
      "grad_norm": 1.0250227451324463,
      "learning_rate": 0.0003973191196927881,
      "loss": 2.2079,
      "step": 90972
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.820453941822052,
      "learning_rate": 0.00039731525033506927,
      "loss": 2.1789,
      "step": 90973
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.8206115961074829,
      "learning_rate": 0.0003973113809592577,
      "loss": 2.234,
      "step": 90974
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.9308484196662903,
      "learning_rate": 0.0003973075115653542,
      "loss": 2.2374,
      "step": 90975
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.9858174324035645,
      "learning_rate": 0.00039730364215335935,
      "loss": 2.2336,
      "step": 90976
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.8174775838851929,
      "learning_rate": 0.0003972997727232739,
      "loss": 2.0831,
      "step": 90977
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.9001597762107849,
      "learning_rate": 0.00039729590327509874,
      "loss": 2.0866,
      "step": 90978
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.9202966690063477,
      "learning_rate": 0.00039729203380883425,
      "loss": 2.0873,
      "step": 90979
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.8530782461166382,
      "learning_rate": 0.0003972881643244813,
      "loss": 2.1375,
      "step": 90980
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.8045032024383545,
      "learning_rate": 0.0003972842948220407,
      "loss": 2.2099,
      "step": 90981
    },
    {
      "epoch": 1.18,
      "grad_norm": 1.096685528755188,
      "learning_rate": 0.00039728042530151326,
      "loss": 1.9876,
      "step": 90982
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.9039388298988342,
      "learning_rate": 0.00039727655576289936,
      "loss": 2.2043,
      "step": 90983
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.7953824996948242,
      "learning_rate": 0.0003972726862062,
      "loss": 2.0402,
      "step": 90984
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.7950025796890259,
      "learning_rate": 0.00039726881663141586,
      "loss": 2.0266,
      "step": 90985
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.9494908452033997,
      "learning_rate": 0.0003972649470385475,
      "loss": 2.114,
      "step": 90986
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.9115687608718872,
      "learning_rate": 0.0003972610774275958,
      "loss": 2.0982,
      "step": 90987
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.8055788278579712,
      "learning_rate": 0.00039725720779856143,
      "loss": 2.17,
      "step": 90988
    },
    {
      "epoch": 1.18,
      "grad_norm": 1.0950772762298584,
      "learning_rate": 0.00039725333815144513,
      "loss": 2.0916,
      "step": 90989
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.8398791551589966,
      "learning_rate": 0.0003972494684862475,
      "loss": 2.0292,
      "step": 90990
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.7791357636451721,
      "learning_rate": 0.00039724559880296957,
      "loss": 2.176,
      "step": 90991
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.9811667799949646,
      "learning_rate": 0.00039724172910161166,
      "loss": 2.2433,
      "step": 90992
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.7819461226463318,
      "learning_rate": 0.0003972378593821746,
      "loss": 2.2336,
      "step": 90993
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.8302499651908875,
      "learning_rate": 0.00039723398964465944,
      "loss": 2.0672,
      "step": 90994
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.7929007411003113,
      "learning_rate": 0.00039723011988906655,
      "loss": 2.1988,
      "step": 90995
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.8257344961166382,
      "learning_rate": 0.00039722625011539664,
      "loss": 2.3466,
      "step": 90996
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.8915053009986877,
      "learning_rate": 0.00039722238032365064,
      "loss": 2.3554,
      "step": 90997
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.7785541415214539,
      "learning_rate": 0.0003972185105138291,
      "loss": 2.2641,
      "step": 90998
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.8691478371620178,
      "learning_rate": 0.00039721464068593286,
      "loss": 2.165,
      "step": 90999
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.792726457118988,
      "learning_rate": 0.00039721077083996266,
      "loss": 2.1203,
      "step": 91000
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.7833646535873413,
      "learning_rate": 0.00039720690097591897,
      "loss": 2.1089,
      "step": 91001
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.851627767086029,
      "learning_rate": 0.0003972030310938027,
      "loss": 2.2931,
      "step": 91002
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.8011767268180847,
      "learning_rate": 0.0003971991611936147,
      "loss": 2.1862,
      "step": 91003
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.8289008736610413,
      "learning_rate": 0.00039719529127535543,
      "loss": 2.365,
      "step": 91004
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.9452056884765625,
      "learning_rate": 0.0003971914213390257,
      "loss": 2.2031,
      "step": 91005
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.8530043959617615,
      "learning_rate": 0.00039718755138462634,
      "loss": 2.2289,
      "step": 91006
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.9293374419212341,
      "learning_rate": 0.0003971836814121579,
      "loss": 2.433,
      "step": 91007
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.7772154808044434,
      "learning_rate": 0.0003971798114216212,
      "loss": 2.1514,
      "step": 91008
    },
    {
      "epoch": 1.18,
      "grad_norm": 0.8371226191520691,
      "learning_rate": 0.00039717594141301703,
      "loss": 2.043,
      "step": 91009
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.8163880705833435,
      "learning_rate": 0.0003971720713863459,
      "loss": 2.19,
      "step": 91010
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.9641458988189697,
      "learning_rate": 0.0003971682013416087,
      "loss": 2.4146,
      "step": 91011
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.8301536440849304,
      "learning_rate": 0.00039716433127880607,
      "loss": 2.2091,
      "step": 91012
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.8746869564056396,
      "learning_rate": 0.00039716046119793875,
      "loss": 2.358,
      "step": 91013
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.9654548168182373,
      "learning_rate": 0.0003971565910990075,
      "loss": 2.2439,
      "step": 91014
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.864176869392395,
      "learning_rate": 0.000397152720982013,
      "loss": 2.1746,
      "step": 91015
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.941865086555481,
      "learning_rate": 0.00039714885084695593,
      "loss": 2.226,
      "step": 91016
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.8226739764213562,
      "learning_rate": 0.0003971449806938371,
      "loss": 2.0359,
      "step": 91017
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.8977530002593994,
      "learning_rate": 0.00039714111052265716,
      "loss": 2.1111,
      "step": 91018
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.8687080144882202,
      "learning_rate": 0.0003971372403334168,
      "loss": 2.1524,
      "step": 91019
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.7778242826461792,
      "learning_rate": 0.00039713337012611696,
      "loss": 2.1668,
      "step": 91020
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.8677191734313965,
      "learning_rate": 0.000397129499900758,
      "loss": 2.1275,
      "step": 91021
    },
    {
      "epoch": 1.19,
      "grad_norm": 1.153836965560913,
      "learning_rate": 0.000397125629657341,
      "loss": 2.2692,
      "step": 91022
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.8944457173347473,
      "learning_rate": 0.00039712175939586646,
      "loss": 2.2277,
      "step": 91023
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.8182674646377563,
      "learning_rate": 0.00039711788911633514,
      "loss": 2.063,
      "step": 91024
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.9232123494148254,
      "learning_rate": 0.00039711401881874773,
      "loss": 2.2464,
      "step": 91025
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.9202238321304321,
      "learning_rate": 0.000397110148503105,
      "loss": 2.0949,
      "step": 91026
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.7531489729881287,
      "learning_rate": 0.00039710627816940774,
      "loss": 2.1034,
      "step": 91027
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.8278921246528625,
      "learning_rate": 0.0003971024078176565,
      "loss": 2.3704,
      "step": 91028
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.8318527936935425,
      "learning_rate": 0.0003970985374478521,
      "loss": 2.1956,
      "step": 91029
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.7900852560997009,
      "learning_rate": 0.00039709466705999526,
      "loss": 2.012,
      "step": 91030
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.8749650120735168,
      "learning_rate": 0.0003970907966540867,
      "loss": 2.0727,
      "step": 91031
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.804959774017334,
      "learning_rate": 0.0003970869262301271,
      "loss": 2.183,
      "step": 91032
    },
    {
      "epoch": 1.19,
      "grad_norm": 1.2714815139770508,
      "learning_rate": 0.0003970830557881173,
      "loss": 2.2877,
      "step": 91033
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.9106417298316956,
      "learning_rate": 0.0003970791853280578,
      "loss": 2.4634,
      "step": 91034
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.8991137146949768,
      "learning_rate": 0.00039707531484994954,
      "loss": 2.0,
      "step": 91035
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.9483766555786133,
      "learning_rate": 0.00039707144435379316,
      "loss": 2.3632,
      "step": 91036
    },
    {
      "epoch": 1.19,
      "grad_norm": 1.1851143836975098,
      "learning_rate": 0.0003970675738395893,
      "loss": 2.0089,
      "step": 91037
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.8474829792976379,
      "learning_rate": 0.0003970637033073387,
      "loss": 2.3701,
      "step": 91038
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.7689990401268005,
      "learning_rate": 0.0003970598327570423,
      "loss": 2.1534,
      "step": 91039
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.9593091607093811,
      "learning_rate": 0.00039705596218870046,
      "loss": 2.1259,
      "step": 91040
    },
    {
      "epoch": 1.19,
      "grad_norm": 1.0542296171188354,
      "learning_rate": 0.00039705209160231423,
      "loss": 2.3008,
      "step": 91041
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.8864812850952148,
      "learning_rate": 0.0003970482209978841,
      "loss": 2.1212,
      "step": 91042
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.9402897357940674,
      "learning_rate": 0.0003970443503754109,
      "loss": 2.1122,
      "step": 91043
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.892875075340271,
      "learning_rate": 0.0003970404797348953,
      "loss": 1.8772,
      "step": 91044
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.8987325429916382,
      "learning_rate": 0.00039703660907633806,
      "loss": 2.2188,
      "step": 91045
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.8345032930374146,
      "learning_rate": 0.0003970327383997399,
      "loss": 2.1636,
      "step": 91046
    },
    {
      "epoch": 1.19,
      "grad_norm": 1.173494577407837,
      "learning_rate": 0.0003970288677051015,
      "loss": 2.4355,
      "step": 91047
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.8308160901069641,
      "learning_rate": 0.00039702499699242357,
      "loss": 2.0913,
      "step": 91048
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.8347557187080383,
      "learning_rate": 0.0003970211262617069,
      "loss": 1.9789,
      "step": 91049
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.8425018787384033,
      "learning_rate": 0.00039701725551295216,
      "loss": 1.9576,
      "step": 91050
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.9132819771766663,
      "learning_rate": 0.0003970133847461601,
      "loss": 2.3518,
      "step": 91051
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.9215039610862732,
      "learning_rate": 0.00039700951396133143,
      "loss": 2.1269,
      "step": 91052
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.837424099445343,
      "learning_rate": 0.00039700564315846683,
      "loss": 2.3795,
      "step": 91053
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.9576257467269897,
      "learning_rate": 0.00039700177233756707,
      "loss": 2.1657,
      "step": 91054
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.8679834008216858,
      "learning_rate": 0.00039699790149863286,
      "loss": 2.0733,
      "step": 91055
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.7717437148094177,
      "learning_rate": 0.00039699403064166484,
      "loss": 2.0344,
      "step": 91056
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.8519660830497742,
      "learning_rate": 0.0003969901597666638,
      "loss": 1.9733,
      "step": 91057
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.9983721375465393,
      "learning_rate": 0.0003969862888736306,
      "loss": 1.9715,
      "step": 91058
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.796320378780365,
      "learning_rate": 0.0003969824179625657,
      "loss": 2.2999,
      "step": 91059
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.8511447906494141,
      "learning_rate": 0.00039697854703346993,
      "loss": 2.2954,
      "step": 91060
    },
    {
      "epoch": 1.19,
      "grad_norm": 1.2999036312103271,
      "learning_rate": 0.00039697467608634405,
      "loss": 2.0256,
      "step": 91061
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.8281795978546143,
      "learning_rate": 0.0003969708051211888,
      "loss": 2.3078,
      "step": 91062
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.7642538547515869,
      "learning_rate": 0.00039696693413800475,
      "loss": 2.1888,
      "step": 91063
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.9446947574615479,
      "learning_rate": 0.00039696306313679285,
      "loss": 2.3963,
      "step": 91064
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.8614438772201538,
      "learning_rate": 0.00039695919211755353,
      "loss": 2.0116,
      "step": 91065
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.7752594947814941,
      "learning_rate": 0.00039695532108028773,
      "loss": 2.0723,
      "step": 91066
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.8057296276092529,
      "learning_rate": 0.0003969514500249961,
      "loss": 2.1733,
      "step": 91067
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.8617953658103943,
      "learning_rate": 0.0003969475789516794,
      "loss": 2.4699,
      "step": 91068
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.8493148684501648,
      "learning_rate": 0.00039694370786033824,
      "loss": 1.9588,
      "step": 91069
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.7995656728744507,
      "learning_rate": 0.00039693983675097354,
      "loss": 2.1455,
      "step": 91070
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.856768012046814,
      "learning_rate": 0.0003969359656235858,
      "loss": 2.1718,
      "step": 91071
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.8087868690490723,
      "learning_rate": 0.0003969320944781758,
      "loss": 2.4168,
      "step": 91072
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.822601854801178,
      "learning_rate": 0.00039692822331474436,
      "loss": 2.1507,
      "step": 91073
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.755486786365509,
      "learning_rate": 0.0003969243521332921,
      "loss": 2.0799,
      "step": 91074
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.9785404801368713,
      "learning_rate": 0.0003969204809338198,
      "loss": 2.1396,
      "step": 91075
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.8059738278388977,
      "learning_rate": 0.00039691660971632813,
      "loss": 2.1877,
      "step": 91076
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.8479416370391846,
      "learning_rate": 0.00039691273848081786,
      "loss": 2.1912,
      "step": 91077
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.8069285154342651,
      "learning_rate": 0.00039690886722728967,
      "loss": 2.2322,
      "step": 91078
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.8035557270050049,
      "learning_rate": 0.0003969049959557443,
      "loss": 2.0947,
      "step": 91079
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.7446802258491516,
      "learning_rate": 0.00039690112466618245,
      "loss": 2.1525,
      "step": 91080
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.8736081123352051,
      "learning_rate": 0.0003968972533586048,
      "loss": 2.2719,
      "step": 91081
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.7881355285644531,
      "learning_rate": 0.00039689338203301226,
      "loss": 2.0466,
      "step": 91082
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.7678932547569275,
      "learning_rate": 0.0003968895106894053,
      "loss": 2.1317,
      "step": 91083
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.778682291507721,
      "learning_rate": 0.0003968856393277848,
      "loss": 1.8104,
      "step": 91084
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.8586106300354004,
      "learning_rate": 0.00039688176794815144,
      "loss": 2.4221,
      "step": 91085
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.7711000442504883,
      "learning_rate": 0.0003968778965505059,
      "loss": 2.1104,
      "step": 91086
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.7616285681724548,
      "learning_rate": 0.0003968740251348489,
      "loss": 2.2739,
      "step": 91087
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.9958475828170776,
      "learning_rate": 0.00039687015370118124,
      "loss": 2.4845,
      "step": 91088
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.8075900077819824,
      "learning_rate": 0.00039686628224950356,
      "loss": 2.1778,
      "step": 91089
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.8669285178184509,
      "learning_rate": 0.00039686241077981657,
      "loss": 2.4056,
      "step": 91090
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.8723678588867188,
      "learning_rate": 0.00039685853929212115,
      "loss": 2.1238,
      "step": 91091
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.7853654623031616,
      "learning_rate": 0.00039685466778641783,
      "loss": 2.0746,
      "step": 91092
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.8924444913864136,
      "learning_rate": 0.0003968507962627074,
      "loss": 2.3631,
      "step": 91093
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.8278688788414001,
      "learning_rate": 0.00039684692472099055,
      "loss": 2.0697,
      "step": 91094
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.8203152418136597,
      "learning_rate": 0.0003968430531612681,
      "loss": 2.3388,
      "step": 91095
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.8063341975212097,
      "learning_rate": 0.00039683918158354064,
      "loss": 2.1521,
      "step": 91096
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.8492822647094727,
      "learning_rate": 0.00039683530998780897,
      "loss": 2.3245,
      "step": 91097
    },
    {
      "epoch": 1.19,
      "grad_norm": 1.003824234008789,
      "learning_rate": 0.00039683143837407374,
      "loss": 2.2161,
      "step": 91098
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.8624541759490967,
      "learning_rate": 0.0003968275667423357,
      "loss": 2.1412,
      "step": 91099
    },
    {
      "epoch": 1.19,
      "grad_norm": 1.1340270042419434,
      "learning_rate": 0.0003968236950925957,
      "loss": 2.3829,
      "step": 91100
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.9918067455291748,
      "learning_rate": 0.0003968198234248543,
      "loss": 2.3341,
      "step": 91101
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.9133645296096802,
      "learning_rate": 0.00039681595173911223,
      "loss": 2.216,
      "step": 91102
    },
    {
      "epoch": 1.19,
      "grad_norm": 1.1289294958114624,
      "learning_rate": 0.0003968120800353703,
      "loss": 2.181,
      "step": 91103
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.9756403565406799,
      "learning_rate": 0.0003968082083136292,
      "loss": 2.1486,
      "step": 91104
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.8086671829223633,
      "learning_rate": 0.00039680433657388953,
      "loss": 2.4495,
      "step": 91105
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.9230365753173828,
      "learning_rate": 0.0003968004648161522,
      "loss": 2.0508,
      "step": 91106
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.8781006336212158,
      "learning_rate": 0.00039679659304041776,
      "loss": 2.3151,
      "step": 91107
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.8487778306007385,
      "learning_rate": 0.0003967927212466871,
      "loss": 2.0643,
      "step": 91108
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.9278160333633423,
      "learning_rate": 0.0003967888494349607,
      "loss": 2.1532,
      "step": 91109
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.8113784193992615,
      "learning_rate": 0.00039678497760523956,
      "loss": 2.1271,
      "step": 91110
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.8967317342758179,
      "learning_rate": 0.0003967811057575242,
      "loss": 2.1695,
      "step": 91111
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.7472211718559265,
      "learning_rate": 0.00039677723389181543,
      "loss": 1.9921,
      "step": 91112
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.9215486645698547,
      "learning_rate": 0.0003967733620081139,
      "loss": 2.4469,
      "step": 91113
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.8015837073326111,
      "learning_rate": 0.0003967694901064204,
      "loss": 2.2614,
      "step": 91114
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.9311924576759338,
      "learning_rate": 0.0003967656181867357,
      "loss": 2.1595,
      "step": 91115
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.9142133593559265,
      "learning_rate": 0.0003967617462490604,
      "loss": 2.008,
      "step": 91116
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.8614286184310913,
      "learning_rate": 0.00039675787429339525,
      "loss": 2.2085,
      "step": 91117
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.9169965982437134,
      "learning_rate": 0.00039675400231974095,
      "loss": 2.1891,
      "step": 91118
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.813863217830658,
      "learning_rate": 0.00039675013032809827,
      "loss": 2.2082,
      "step": 91119
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.9614153504371643,
      "learning_rate": 0.000396746258318468,
      "loss": 1.9462,
      "step": 91120
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.8197988867759705,
      "learning_rate": 0.0003967423862908506,
      "loss": 2.3575,
      "step": 91121
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.8312098979949951,
      "learning_rate": 0.00039673851424524713,
      "loss": 2.3825,
      "step": 91122
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.9412444233894348,
      "learning_rate": 0.00039673464218165804,
      "loss": 2.2299,
      "step": 91123
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.8787570595741272,
      "learning_rate": 0.0003967307701000842,
      "loss": 2.2162,
      "step": 91124
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.8068603277206421,
      "learning_rate": 0.00039672689800052634,
      "loss": 2.3943,
      "step": 91125
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.8553248047828674,
      "learning_rate": 0.0003967230258829851,
      "loss": 2.0908,
      "step": 91126
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.8298747539520264,
      "learning_rate": 0.0003967191537474611,
      "loss": 2.0331,
      "step": 91127
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.8124974370002747,
      "learning_rate": 0.0003967152815939553,
      "loss": 2.3356,
      "step": 91128
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.811662495136261,
      "learning_rate": 0.0003967114094224683,
      "loss": 2.1531,
      "step": 91129
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.7677944302558899,
      "learning_rate": 0.00039670753723300076,
      "loss": 2.2537,
      "step": 91130
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.91305011510849,
      "learning_rate": 0.00039670366502555355,
      "loss": 2.15,
      "step": 91131
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.8015039563179016,
      "learning_rate": 0.00039669979280012723,
      "loss": 2.237,
      "step": 91132
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.8410902619361877,
      "learning_rate": 0.00039669592055672267,
      "loss": 2.196,
      "step": 91133
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.8556243777275085,
      "learning_rate": 0.00039669204829534047,
      "loss": 2.1244,
      "step": 91134
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.8343644142150879,
      "learning_rate": 0.0003966881760159813,
      "loss": 2.1928,
      "step": 91135
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.8206784725189209,
      "learning_rate": 0.0003966843037186461,
      "loss": 2.4317,
      "step": 91136
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.7852259278297424,
      "learning_rate": 0.0003966804314033354,
      "loss": 1.9671,
      "step": 91137
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.8074914813041687,
      "learning_rate": 0.00039667655907005007,
      "loss": 2.054,
      "step": 91138
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.8102647662162781,
      "learning_rate": 0.0003966726867187906,
      "loss": 2.244,
      "step": 91139
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.7674917578697205,
      "learning_rate": 0.00039666881434955803,
      "loss": 2.0835,
      "step": 91140
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.8743079900741577,
      "learning_rate": 0.0003966649419623528,
      "loss": 2.1608,
      "step": 91141
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.8114981055259705,
      "learning_rate": 0.0003966610695571757,
      "loss": 2.0342,
      "step": 91142
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.7955635190010071,
      "learning_rate": 0.00039665719713402755,
      "loss": 2.1706,
      "step": 91143
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.8475279808044434,
      "learning_rate": 0.00039665332469290896,
      "loss": 2.2156,
      "step": 91144
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.7358623743057251,
      "learning_rate": 0.0003966494522338207,
      "loss": 2.2303,
      "step": 91145
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.7766152620315552,
      "learning_rate": 0.0003966455797567635,
      "loss": 1.9989,
      "step": 91146
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.8160277605056763,
      "learning_rate": 0.00039664170726173804,
      "loss": 2.2522,
      "step": 91147
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.8175623416900635,
      "learning_rate": 0.00039663783474874504,
      "loss": 2.0499,
      "step": 91148
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.8931735754013062,
      "learning_rate": 0.00039663396221778534,
      "loss": 2.2708,
      "step": 91149
    },
    {
      "epoch": 1.19,
      "grad_norm": 1.0792934894561768,
      "learning_rate": 0.00039663008966885944,
      "loss": 2.2459,
      "step": 91150
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.8157449960708618,
      "learning_rate": 0.0003966262171019682,
      "loss": 2.2501,
      "step": 91151
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.8645550012588501,
      "learning_rate": 0.00039662234451711244,
      "loss": 2.2967,
      "step": 91152
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.846072256565094,
      "learning_rate": 0.0003966184719142927,
      "loss": 2.2956,
      "step": 91153
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.9246800541877747,
      "learning_rate": 0.00039661459929350967,
      "loss": 2.0187,
      "step": 91154
    },
    {
      "epoch": 1.19,
      "grad_norm": 1.0022649765014648,
      "learning_rate": 0.0003966107266547643,
      "loss": 1.9167,
      "step": 91155
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.8272903561592102,
      "learning_rate": 0.0003966068539980571,
      "loss": 2.2913,
      "step": 91156
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.8461847901344299,
      "learning_rate": 0.00039660298132338884,
      "loss": 1.9597,
      "step": 91157
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.8446110486984253,
      "learning_rate": 0.00039659910863076034,
      "loss": 2.2783,
      "step": 91158
    },
    {
      "epoch": 1.19,
      "grad_norm": 1.0138483047485352,
      "learning_rate": 0.00039659523592017214,
      "loss": 2.1132,
      "step": 91159
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.8632746934890747,
      "learning_rate": 0.0003965913631916251,
      "loss": 2.1801,
      "step": 91160
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.8523613214492798,
      "learning_rate": 0.00039658749044512,
      "loss": 2.3058,
      "step": 91161
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.9553377628326416,
      "learning_rate": 0.0003965836176806574,
      "loss": 2.3914,
      "step": 91162
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.9666211605072021,
      "learning_rate": 0.000396579744898238,
      "loss": 2.2557,
      "step": 91163
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.9993171691894531,
      "learning_rate": 0.0003965758720978627,
      "loss": 2.1575,
      "step": 91164
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.7982152104377747,
      "learning_rate": 0.00039657199927953214,
      "loss": 2.0421,
      "step": 91165
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.8385298252105713,
      "learning_rate": 0.0003965681264432469,
      "loss": 2.2245,
      "step": 91166
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.8483337759971619,
      "learning_rate": 0.00039656425358900794,
      "loss": 2.34,
      "step": 91167
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.8492454886436462,
      "learning_rate": 0.0003965603807168158,
      "loss": 2.3741,
      "step": 91168
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.8432877063751221,
      "learning_rate": 0.00039655650782667123,
      "loss": 2.136,
      "step": 91169
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.7609742283821106,
      "learning_rate": 0.0003965526349185751,
      "loss": 2.13,
      "step": 91170
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.8090368509292603,
      "learning_rate": 0.0003965487619925279,
      "loss": 2.16,
      "step": 91171
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.8013126254081726,
      "learning_rate": 0.00039654488904853057,
      "loss": 2.1424,
      "step": 91172
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.908414900302887,
      "learning_rate": 0.00039654101608658365,
      "loss": 2.1264,
      "step": 91173
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.8327662944793701,
      "learning_rate": 0.00039653714310668793,
      "loss": 2.2396,
      "step": 91174
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.8054671287536621,
      "learning_rate": 0.0003965332701088441,
      "loss": 2.1018,
      "step": 91175
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.851848840713501,
      "learning_rate": 0.000396529397093053,
      "loss": 2.2519,
      "step": 91176
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.7849598526954651,
      "learning_rate": 0.00039652552405931526,
      "loss": 1.9569,
      "step": 91177
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.8244808316230774,
      "learning_rate": 0.00039652165100763153,
      "loss": 2.022,
      "step": 91178
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.7931734323501587,
      "learning_rate": 0.00039651777793800263,
      "loss": 2.0084,
      "step": 91179
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.9894037842750549,
      "learning_rate": 0.00039651390485042926,
      "loss": 2.0869,
      "step": 91180
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.8286902904510498,
      "learning_rate": 0.0003965100317449121,
      "loss": 2.3556,
      "step": 91181
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.8345472812652588,
      "learning_rate": 0.00039650615862145194,
      "loss": 2.1301,
      "step": 91182
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.8499318361282349,
      "learning_rate": 0.00039650228548004945,
      "loss": 2.2128,
      "step": 91183
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.8504154086112976,
      "learning_rate": 0.0003964984123207054,
      "loss": 2.0735,
      "step": 91184
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.793743371963501,
      "learning_rate": 0.0003964945391434204,
      "loss": 2.4006,
      "step": 91185
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.9110387563705444,
      "learning_rate": 0.0003964906659481954,
      "loss": 2.0471,
      "step": 91186
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.8845949172973633,
      "learning_rate": 0.0003964867927350308,
      "loss": 2.218,
      "step": 91187
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.9364674091339111,
      "learning_rate": 0.00039648291950392756,
      "loss": 2.3444,
      "step": 91188
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.832653820514679,
      "learning_rate": 0.0003964790462548863,
      "loss": 2.2044,
      "step": 91189
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.8426850438117981,
      "learning_rate": 0.00039647517298790777,
      "loss": 2.2094,
      "step": 91190
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.8726382255554199,
      "learning_rate": 0.00039647129970299264,
      "loss": 2.2559,
      "step": 91191
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.9416902661323547,
      "learning_rate": 0.0003964674264001417,
      "loss": 2.1492,
      "step": 91192
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.8882421255111694,
      "learning_rate": 0.00039646355307935575,
      "loss": 2.1926,
      "step": 91193
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.8622132539749146,
      "learning_rate": 0.00039645967974063525,
      "loss": 2.0661,
      "step": 91194
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.7897441387176514,
      "learning_rate": 0.00039645580638398117,
      "loss": 2.0898,
      "step": 91195
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.860066294670105,
      "learning_rate": 0.00039645193300939406,
      "loss": 2.0784,
      "step": 91196
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.83541339635849,
      "learning_rate": 0.00039644805961687484,
      "loss": 2.1254,
      "step": 91197
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.8167396783828735,
      "learning_rate": 0.0003964441862064239,
      "loss": 2.3726,
      "step": 91198
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.7919246554374695,
      "learning_rate": 0.0003964403127780423,
      "loss": 2.3862,
      "step": 91199
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.9792816638946533,
      "learning_rate": 0.00039643643933173073,
      "loss": 2.2266,
      "step": 91200
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.9108197689056396,
      "learning_rate": 0.00039643256586748964,
      "loss": 2.2496,
      "step": 91201
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.794465959072113,
      "learning_rate": 0.0003964286923853199,
      "loss": 2.4273,
      "step": 91202
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.8862876296043396,
      "learning_rate": 0.0003964248188852224,
      "loss": 2.2042,
      "step": 91203
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.8614755868911743,
      "learning_rate": 0.0003964209453671976,
      "loss": 2.1052,
      "step": 91204
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.8853614330291748,
      "learning_rate": 0.0003964170718312463,
      "loss": 2.3899,
      "step": 91205
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.7163801789283752,
      "learning_rate": 0.00039641319827736927,
      "loss": 2.0952,
      "step": 91206
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.8275453448295593,
      "learning_rate": 0.0003964093247055673,
      "loss": 2.1466,
      "step": 91207
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.8654056787490845,
      "learning_rate": 0.00039640545111584093,
      "loss": 2.3788,
      "step": 91208
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.8101810812950134,
      "learning_rate": 0.00039640157750819096,
      "loss": 2.2026,
      "step": 91209
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.9364078044891357,
      "learning_rate": 0.0003963977038826182,
      "loss": 2.236,
      "step": 91210
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.8517974615097046,
      "learning_rate": 0.00039639383023912315,
      "loss": 2.4059,
      "step": 91211
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.8521072268486023,
      "learning_rate": 0.0003963899565777067,
      "loss": 2.1911,
      "step": 91212
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.806075930595398,
      "learning_rate": 0.0003963860828983697,
      "loss": 2.0186,
      "step": 91213
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.8445223569869995,
      "learning_rate": 0.0003963822092011125,
      "loss": 2.3083,
      "step": 91214
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.8124446868896484,
      "learning_rate": 0.0003963783354859361,
      "loss": 2.2654,
      "step": 91215
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.887959361076355,
      "learning_rate": 0.00039637446175284127,
      "loss": 2.0467,
      "step": 91216
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.8318598866462708,
      "learning_rate": 0.00039637058800182844,
      "loss": 2.2894,
      "step": 91217
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.8876240849494934,
      "learning_rate": 0.00039636671423289857,
      "loss": 2.2135,
      "step": 91218
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.9251318573951721,
      "learning_rate": 0.0003963628404460523,
      "loss": 2.0283,
      "step": 91219
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.8233144283294678,
      "learning_rate": 0.0003963589666412903,
      "loss": 2.096,
      "step": 91220
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.8350042104721069,
      "learning_rate": 0.00039635509281861345,
      "loss": 2.083,
      "step": 91221
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.9358330965042114,
      "learning_rate": 0.0003963512189780223,
      "loss": 2.2747,
      "step": 91222
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.8076629638671875,
      "learning_rate": 0.00039634734511951767,
      "loss": 2.0826,
      "step": 91223
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.7906516790390015,
      "learning_rate": 0.0003963434712431003,
      "loss": 2.2072,
      "step": 91224
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.8294724225997925,
      "learning_rate": 0.00039633959734877087,
      "loss": 1.979,
      "step": 91225
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.9336380958557129,
      "learning_rate": 0.00039633572343653,
      "loss": 2.344,
      "step": 91226
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.8237743377685547,
      "learning_rate": 0.0003963318495063785,
      "loss": 2.1422,
      "step": 91227
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.778750479221344,
      "learning_rate": 0.0003963279755583171,
      "loss": 2.1592,
      "step": 91228
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.8608217239379883,
      "learning_rate": 0.00039632410159234655,
      "loss": 2.0759,
      "step": 91229
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.8613149523735046,
      "learning_rate": 0.0003963202276084675,
      "loss": 2.1515,
      "step": 91230
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.8267565965652466,
      "learning_rate": 0.00039631635360668074,
      "loss": 2.1051,
      "step": 91231
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.9164482951164246,
      "learning_rate": 0.0003963124795869869,
      "loss": 2.2307,
      "step": 91232
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.9973974227905273,
      "learning_rate": 0.0003963086055493868,
      "loss": 2.1865,
      "step": 91233
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.8952487707138062,
      "learning_rate": 0.0003963047314938811,
      "loss": 2.306,
      "step": 91234
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.9378046989440918,
      "learning_rate": 0.0003963008574204705,
      "loss": 2.2637,
      "step": 91235
    },
    {
      "epoch": 1.19,
      "grad_norm": 1.2570278644561768,
      "learning_rate": 0.00039629698332915576,
      "loss": 2.3401,
      "step": 91236
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.8686386346817017,
      "learning_rate": 0.0003962931092199376,
      "loss": 1.9817,
      "step": 91237
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.8422550559043884,
      "learning_rate": 0.00039628923509281677,
      "loss": 2.3007,
      "step": 91238
    },
    {
      "epoch": 1.19,
      "grad_norm": 1.3224729299545288,
      "learning_rate": 0.0003962853609477939,
      "loss": 2.3065,
      "step": 91239
    },
    {
      "epoch": 1.19,
      "grad_norm": 1.0981899499893188,
      "learning_rate": 0.00039628148678486986,
      "loss": 2.1945,
      "step": 91240
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.8683134913444519,
      "learning_rate": 0.0003962776126040452,
      "loss": 2.1863,
      "step": 91241
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.8618155717849731,
      "learning_rate": 0.00039627373840532066,
      "loss": 2.4988,
      "step": 91242
    },
    {
      "epoch": 1.19,
      "grad_norm": 1.5207767486572266,
      "learning_rate": 0.0003962698641886971,
      "loss": 2.3882,
      "step": 91243
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.9266989231109619,
      "learning_rate": 0.0003962659899541751,
      "loss": 1.9626,
      "step": 91244
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.8095648884773254,
      "learning_rate": 0.0003962621157017554,
      "loss": 2.1121,
      "step": 91245
    },
    {
      "epoch": 1.19,
      "grad_norm": 1.0100047588348389,
      "learning_rate": 0.00039625824143143894,
      "loss": 2.2086,
      "step": 91246
    },
    {
      "epoch": 1.19,
      "grad_norm": 1.1341512203216553,
      "learning_rate": 0.00039625436714322605,
      "loss": 2.0959,
      "step": 91247
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.8304019570350647,
      "learning_rate": 0.00039625049283711775,
      "loss": 1.9469,
      "step": 91248
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.8440859317779541,
      "learning_rate": 0.00039624661851311464,
      "loss": 1.9728,
      "step": 91249
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.8965013027191162,
      "learning_rate": 0.0003962427441712175,
      "loss": 1.872,
      "step": 91250
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.8464382290840149,
      "learning_rate": 0.000396238869811427,
      "loss": 2.0104,
      "step": 91251
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.923828661441803,
      "learning_rate": 0.0003962349954337438,
      "loss": 2.3266,
      "step": 91252
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.893012285232544,
      "learning_rate": 0.00039623112103816884,
      "loss": 2.0706,
      "step": 91253
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.9591602683067322,
      "learning_rate": 0.0003962272466247027,
      "loss": 2.3208,
      "step": 91254
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.8355313539505005,
      "learning_rate": 0.00039622337219334593,
      "loss": 2.3706,
      "step": 91255
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.9443820714950562,
      "learning_rate": 0.0003962194977440996,
      "loss": 2.238,
      "step": 91256
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.8934730887413025,
      "learning_rate": 0.00039621562327696417,
      "loss": 2.179,
      "step": 91257
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.8178239464759827,
      "learning_rate": 0.00039621174879194044,
      "loss": 2.2755,
      "step": 91258
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.8270232081413269,
      "learning_rate": 0.0003962078742890291,
      "loss": 2.1715,
      "step": 91259
    },
    {
      "epoch": 1.19,
      "grad_norm": 1.1038305759429932,
      "learning_rate": 0.0003962039997682309,
      "loss": 2.242,
      "step": 91260
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.8634650111198425,
      "learning_rate": 0.0003962001252295466,
      "loss": 2.1718,
      "step": 91261
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.8738014101982117,
      "learning_rate": 0.0003961962506729769,
      "loss": 2.3861,
      "step": 91262
    },
    {
      "epoch": 1.19,
      "grad_norm": 1.0885599851608276,
      "learning_rate": 0.00039619237609852246,
      "loss": 2.0585,
      "step": 91263
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.8570824265480042,
      "learning_rate": 0.000396188501506184,
      "loss": 2.329,
      "step": 91264
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.7873321771621704,
      "learning_rate": 0.0003961846268959624,
      "loss": 2.0476,
      "step": 91265
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.9592750668525696,
      "learning_rate": 0.0003961807522678582,
      "loss": 2.3044,
      "step": 91266
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.8483859896659851,
      "learning_rate": 0.00039617687762187213,
      "loss": 2.1354,
      "step": 91267
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.7736667394638062,
      "learning_rate": 0.0003961730029580051,
      "loss": 2.1003,
      "step": 91268
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.8045138716697693,
      "learning_rate": 0.0003961691282762576,
      "loss": 2.1227,
      "step": 91269
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.9583948254585266,
      "learning_rate": 0.0003961652535766304,
      "loss": 2.2574,
      "step": 91270
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.8105787634849548,
      "learning_rate": 0.0003961613788591244,
      "loss": 2.1749,
      "step": 91271
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.8747883439064026,
      "learning_rate": 0.00039615750412374005,
      "loss": 2.3786,
      "step": 91272
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.8777017593383789,
      "learning_rate": 0.0003961536293704782,
      "loss": 2.1566,
      "step": 91273
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.8686002492904663,
      "learning_rate": 0.00039614975459933966,
      "loss": 2.1857,
      "step": 91274
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.8168418407440186,
      "learning_rate": 0.00039614587981032505,
      "loss": 2.1534,
      "step": 91275
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.8207770586013794,
      "learning_rate": 0.00039614200500343505,
      "loss": 2.3043,
      "step": 91276
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.7966262102127075,
      "learning_rate": 0.00039613813017867056,
      "loss": 1.9663,
      "step": 91277
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.9210402965545654,
      "learning_rate": 0.000396134255336032,
      "loss": 2.1776,
      "step": 91278
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.7915151119232178,
      "learning_rate": 0.0003961303804755204,
      "loss": 2.1008,
      "step": 91279
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.8073322772979736,
      "learning_rate": 0.0003961265055971362,
      "loss": 2.1503,
      "step": 91280
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.8597060441970825,
      "learning_rate": 0.0003961226307008805,
      "loss": 1.9731,
      "step": 91281
    },
    {
      "epoch": 1.19,
      "grad_norm": 1.0849629640579224,
      "learning_rate": 0.0003961187557867536,
      "loss": 2.0326,
      "step": 91282
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.9053182601928711,
      "learning_rate": 0.00039611488085475647,
      "loss": 1.9252,
      "step": 91283
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.9182024002075195,
      "learning_rate": 0.00039611100590488974,
      "loss": 2.0814,
      "step": 91284
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.8836013674736023,
      "learning_rate": 0.0003961071309371543,
      "loss": 2.1183,
      "step": 91285
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.8727880120277405,
      "learning_rate": 0.00039610325595155056,
      "loss": 2.3929,
      "step": 91286
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.9209296703338623,
      "learning_rate": 0.00039609938094807947,
      "loss": 2.1923,
      "step": 91287
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.7980157732963562,
      "learning_rate": 0.0003960955059267417,
      "loss": 1.929,
      "step": 91288
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.8440797328948975,
      "learning_rate": 0.0003960916308875379,
      "loss": 2.2819,
      "step": 91289
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.8262974619865417,
      "learning_rate": 0.00039608775583046895,
      "loss": 2.155,
      "step": 91290
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.8571485877037048,
      "learning_rate": 0.0003960838807555354,
      "loss": 2.0578,
      "step": 91291
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.85910564661026,
      "learning_rate": 0.0003960800056627381,
      "loss": 2.4047,
      "step": 91292
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.90156090259552,
      "learning_rate": 0.0003960761305520777,
      "loss": 2.2567,
      "step": 91293
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.8078094720840454,
      "learning_rate": 0.0003960722554235549,
      "loss": 2.2264,
      "step": 91294
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.8705524206161499,
      "learning_rate": 0.0003960683802771705,
      "loss": 2.1876,
      "step": 91295
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.8390953540802002,
      "learning_rate": 0.0003960645051129252,
      "loss": 2.176,
      "step": 91296
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.821570873260498,
      "learning_rate": 0.00039606062993081953,
      "loss": 2.2207,
      "step": 91297
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.8959524631500244,
      "learning_rate": 0.00039605675473085457,
      "loss": 2.2018,
      "step": 91298
    },
    {
      "epoch": 1.19,
      "grad_norm": 1.0321922302246094,
      "learning_rate": 0.00039605287951303075,
      "loss": 2.311,
      "step": 91299
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.7873221635818481,
      "learning_rate": 0.00039604900427734886,
      "loss": 2.3188,
      "step": 91300
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.827561616897583,
      "learning_rate": 0.00039604512902380974,
      "loss": 2.07,
      "step": 91301
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.8546696305274963,
      "learning_rate": 0.0003960412537524139,
      "loss": 1.9597,
      "step": 91302
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.832696795463562,
      "learning_rate": 0.0003960373784631623,
      "loss": 2.3137,
      "step": 91303
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.8160394430160522,
      "learning_rate": 0.0003960335031560555,
      "loss": 2.0065,
      "step": 91304
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.8469973802566528,
      "learning_rate": 0.00039602962783109424,
      "loss": 2.1861,
      "step": 91305
    },
    {
      "epoch": 1.19,
      "grad_norm": 1.0651576519012451,
      "learning_rate": 0.00039602575248827925,
      "loss": 2.2612,
      "step": 91306
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.7889958620071411,
      "learning_rate": 0.00039602187712761125,
      "loss": 2.2431,
      "step": 91307
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.8462273478507996,
      "learning_rate": 0.00039601800174909104,
      "loss": 2.0586,
      "step": 91308
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.9019386768341064,
      "learning_rate": 0.0003960141263527192,
      "loss": 1.8916,
      "step": 91309
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.9334685802459717,
      "learning_rate": 0.00039601025093849657,
      "loss": 2.162,
      "step": 91310
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.7826989889144897,
      "learning_rate": 0.00039600637550642377,
      "loss": 2.1971,
      "step": 91311
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.8434040546417236,
      "learning_rate": 0.0003960025000565016,
      "loss": 2.1142,
      "step": 91312
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.9746065735816956,
      "learning_rate": 0.0003959986245887308,
      "loss": 2.367,
      "step": 91313
    },
    {
      "epoch": 1.19,
      "grad_norm": 1.0299441814422607,
      "learning_rate": 0.00039599474910311194,
      "loss": 2.09,
      "step": 91314
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.7588906288146973,
      "learning_rate": 0.0003959908735996459,
      "loss": 2.2662,
      "step": 91315
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.8326277732849121,
      "learning_rate": 0.0003959869980783333,
      "loss": 2.2431,
      "step": 91316
    },
    {
      "epoch": 1.19,
      "grad_norm": 1.171956181526184,
      "learning_rate": 0.000395983122539175,
      "loss": 2.2081,
      "step": 91317
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.9991202354431152,
      "learning_rate": 0.00039597924698217156,
      "loss": 2.3805,
      "step": 91318
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.8766957521438599,
      "learning_rate": 0.0003959753714073238,
      "loss": 2.3262,
      "step": 91319
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.9420490264892578,
      "learning_rate": 0.0003959714958146324,
      "loss": 1.8541,
      "step": 91320
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.8629319071769714,
      "learning_rate": 0.00039596762020409804,
      "loss": 1.9318,
      "step": 91321
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.8747026324272156,
      "learning_rate": 0.00039596374457572154,
      "loss": 2.374,
      "step": 91322
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.8816156387329102,
      "learning_rate": 0.0003959598689295036,
      "loss": 2.3373,
      "step": 91323
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.9494421482086182,
      "learning_rate": 0.00039595599326544485,
      "loss": 2.0857,
      "step": 91324
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.8146984577178955,
      "learning_rate": 0.000395952117583546,
      "loss": 2.115,
      "step": 91325
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.8455374240875244,
      "learning_rate": 0.00039594824188380804,
      "loss": 2.1469,
      "step": 91326
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.8859502077102661,
      "learning_rate": 0.0003959443661662313,
      "loss": 2.1562,
      "step": 91327
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.9286973476409912,
      "learning_rate": 0.00039594049043081673,
      "loss": 2.2391,
      "step": 91328
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.8266936540603638,
      "learning_rate": 0.0003959366146775651,
      "loss": 2.1298,
      "step": 91329
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.8850982785224915,
      "learning_rate": 0.00039593273890647696,
      "loss": 2.1608,
      "step": 91330
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.8750625252723694,
      "learning_rate": 0.00039592886311755316,
      "loss": 2.1549,
      "step": 91331
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.8592889904975891,
      "learning_rate": 0.00039592498731079434,
      "loss": 2.0699,
      "step": 91332
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.9498464465141296,
      "learning_rate": 0.00039592111148620127,
      "loss": 2.1813,
      "step": 91333
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.9414022564888,
      "learning_rate": 0.00039591723564377465,
      "loss": 2.0586,
      "step": 91334
    },
    {
      "epoch": 1.19,
      "grad_norm": 1.0680928230285645,
      "learning_rate": 0.0003959133597835153,
      "loss": 1.9647,
      "step": 91335
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.9593802690505981,
      "learning_rate": 0.00039590948390542366,
      "loss": 2.0699,
      "step": 91336
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.8147744536399841,
      "learning_rate": 0.0003959056080095007,
      "loss": 2.2174,
      "step": 91337
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.8989497423171997,
      "learning_rate": 0.00039590173209574713,
      "loss": 2.0276,
      "step": 91338
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.8383185863494873,
      "learning_rate": 0.0003958978561641636,
      "loss": 2.3363,
      "step": 91339
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.8679448962211609,
      "learning_rate": 0.00039589398021475087,
      "loss": 2.1184,
      "step": 91340
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.8388385772705078,
      "learning_rate": 0.00039589010424750964,
      "loss": 2.0534,
      "step": 91341
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.8662592172622681,
      "learning_rate": 0.0003958862282624406,
      "loss": 2.3135,
      "step": 91342
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.8644372224807739,
      "learning_rate": 0.0003958823522595444,
      "loss": 2.0402,
      "step": 91343
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.8005437254905701,
      "learning_rate": 0.00039587847623882204,
      "loss": 2.1854,
      "step": 91344
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.8612267374992371,
      "learning_rate": 0.000395874600200274,
      "loss": 2.2673,
      "step": 91345
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.809492826461792,
      "learning_rate": 0.000395870724143901,
      "loss": 2.2216,
      "step": 91346
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.8592437505722046,
      "learning_rate": 0.0003958668480697039,
      "loss": 2.2668,
      "step": 91347
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.7637412548065186,
      "learning_rate": 0.00039586297197768326,
      "loss": 2.1463,
      "step": 91348
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.8561443090438843,
      "learning_rate": 0.00039585909586784,
      "loss": 2.0715,
      "step": 91349
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.9194899797439575,
      "learning_rate": 0.0003958552197401747,
      "loss": 2.2227,
      "step": 91350
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.880175769329071,
      "learning_rate": 0.00039585134359468804,
      "loss": 2.212,
      "step": 91351
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.8562887907028198,
      "learning_rate": 0.00039584746743138077,
      "loss": 2.2507,
      "step": 91352
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.8630054593086243,
      "learning_rate": 0.0003958435912502538,
      "loss": 2.2389,
      "step": 91353
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.8875142335891724,
      "learning_rate": 0.00039583971505130754,
      "loss": 2.1349,
      "step": 91354
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.8955663442611694,
      "learning_rate": 0.00039583583883454293,
      "loss": 2.2237,
      "step": 91355
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.8720908761024475,
      "learning_rate": 0.0003958319625999607,
      "loss": 2.2144,
      "step": 91356
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.8321473598480225,
      "learning_rate": 0.00039582808634756135,
      "loss": 2.2181,
      "step": 91357
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.9503347873687744,
      "learning_rate": 0.0003958242100773459,
      "loss": 2.3602,
      "step": 91358
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.861747682094574,
      "learning_rate": 0.00039582033378931483,
      "loss": 2.2791,
      "step": 91359
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.9173141121864319,
      "learning_rate": 0.000395816457483469,
      "loss": 2.1727,
      "step": 91360
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.962990403175354,
      "learning_rate": 0.00039581258115980897,
      "loss": 1.9804,
      "step": 91361
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.8505811095237732,
      "learning_rate": 0.0003958087048183357,
      "loss": 2.0856,
      "step": 91362
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.8592736124992371,
      "learning_rate": 0.0003958048284590497,
      "loss": 2.1704,
      "step": 91363
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.9200858473777771,
      "learning_rate": 0.0003958009520819518,
      "loss": 2.2501,
      "step": 91364
    },
    {
      "epoch": 1.19,
      "grad_norm": 1.0626572370529175,
      "learning_rate": 0.00039579707568704277,
      "loss": 2.0322,
      "step": 91365
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.9151238203048706,
      "learning_rate": 0.0003957931992743232,
      "loss": 2.2195,
      "step": 91366
    },
    {
      "epoch": 1.19,
      "grad_norm": 1.187228798866272,
      "learning_rate": 0.0003957893228437938,
      "loss": 2.2674,
      "step": 91367
    },
    {
      "epoch": 1.19,
      "grad_norm": 1.059915542602539,
      "learning_rate": 0.00039578544639545544,
      "loss": 2.174,
      "step": 91368
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.8478952050209045,
      "learning_rate": 0.00039578156992930877,
      "loss": 2.1753,
      "step": 91369
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.8259056210517883,
      "learning_rate": 0.00039577769344535446,
      "loss": 2.3658,
      "step": 91370
    },
    {
      "epoch": 1.19,
      "grad_norm": 1.0131168365478516,
      "learning_rate": 0.0003957738169435932,
      "loss": 2.0479,
      "step": 91371
    },
    {
      "epoch": 1.19,
      "grad_norm": 1.0349034070968628,
      "learning_rate": 0.00039576994042402584,
      "loss": 2.0466,
      "step": 91372
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.9241281747817993,
      "learning_rate": 0.00039576606388665316,
      "loss": 2.1732,
      "step": 91373
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.9513338208198547,
      "learning_rate": 0.0003957621873314756,
      "loss": 1.9682,
      "step": 91374
    },
    {
      "epoch": 1.19,
      "grad_norm": 1.0451487302780151,
      "learning_rate": 0.0003957583107584941,
      "loss": 2.1922,
      "step": 91375
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.9154961109161377,
      "learning_rate": 0.0003957544341677093,
      "loss": 2.3444,
      "step": 91376
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.8007086515426636,
      "learning_rate": 0.00039575055755912197,
      "loss": 1.9711,
      "step": 91377
    },
    {
      "epoch": 1.19,
      "grad_norm": 1.3161500692367554,
      "learning_rate": 0.00039574668093273277,
      "loss": 2.3452,
      "step": 91378
    },
    {
      "epoch": 1.19,
      "grad_norm": 1.2580801248550415,
      "learning_rate": 0.0003957428042885425,
      "loss": 2.0865,
      "step": 91379
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.800168514251709,
      "learning_rate": 0.0003957389276265518,
      "loss": 2.275,
      "step": 91380
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.9650306701660156,
      "learning_rate": 0.00039573505094676144,
      "loss": 2.1495,
      "step": 91381
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.9586450457572937,
      "learning_rate": 0.00039573117424917213,
      "loss": 2.147,
      "step": 91382
    },
    {
      "epoch": 1.19,
      "grad_norm": 1.2743812799453735,
      "learning_rate": 0.0003957272975337845,
      "loss": 2.4542,
      "step": 91383
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.7750921845436096,
      "learning_rate": 0.00039572342080059956,
      "loss": 2.2092,
      "step": 91384
    },
    {
      "epoch": 1.19,
      "grad_norm": 1.1168317794799805,
      "learning_rate": 0.0003957195440496177,
      "loss": 2.0788,
      "step": 91385
    },
    {
      "epoch": 1.19,
      "grad_norm": 1.1152485609054565,
      "learning_rate": 0.0003957156672808397,
      "loss": 1.9393,
      "step": 91386
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.9059602618217468,
      "learning_rate": 0.0003957117904942665,
      "loss": 2.1505,
      "step": 91387
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.8381722569465637,
      "learning_rate": 0.00039570791368989856,
      "loss": 1.9255,
      "step": 91388
    },
    {
      "epoch": 1.19,
      "grad_norm": 1.0090445280075073,
      "learning_rate": 0.00039570403686773676,
      "loss": 2.2029,
      "step": 91389
    },
    {
      "epoch": 1.19,
      "grad_norm": 1.180972933769226,
      "learning_rate": 0.0003957001600277818,
      "loss": 2.2442,
      "step": 91390
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.8182428479194641,
      "learning_rate": 0.00039569628317003436,
      "loss": 2.1642,
      "step": 91391
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.9959484338760376,
      "learning_rate": 0.0003956924062944951,
      "loss": 1.9433,
      "step": 91392
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.8147697448730469,
      "learning_rate": 0.0003956885294011649,
      "loss": 1.9994,
      "step": 91393
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.9840831160545349,
      "learning_rate": 0.00039568465249004435,
      "loss": 2.0928,
      "step": 91394
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.7865277528762817,
      "learning_rate": 0.0003956807755611342,
      "loss": 1.8889,
      "step": 91395
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.8331872820854187,
      "learning_rate": 0.00039567689861443533,
      "loss": 2.1002,
      "step": 91396
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.7899542450904846,
      "learning_rate": 0.00039567302164994814,
      "loss": 2.256,
      "step": 91397
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.8036230206489563,
      "learning_rate": 0.0003956691446676736,
      "loss": 2.1719,
      "step": 91398
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.7924294471740723,
      "learning_rate": 0.00039566526766761237,
      "loss": 2.3122,
      "step": 91399
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.7818886041641235,
      "learning_rate": 0.00039566139064976515,
      "loss": 1.9575,
      "step": 91400
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.8732156157493591,
      "learning_rate": 0.00039565751361413267,
      "loss": 2.1187,
      "step": 91401
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.8614296317100525,
      "learning_rate": 0.0003956536365607157,
      "loss": 2.0654,
      "step": 91402
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.9767794013023376,
      "learning_rate": 0.0003956497594895149,
      "loss": 2.2072,
      "step": 91403
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.8023467063903809,
      "learning_rate": 0.0003956458824005309,
      "loss": 2.1735,
      "step": 91404
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.7943524122238159,
      "learning_rate": 0.00039564200529376464,
      "loss": 2.1807,
      "step": 91405
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.8319917917251587,
      "learning_rate": 0.00039563812816921664,
      "loss": 2.1508,
      "step": 91406
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.9791997671127319,
      "learning_rate": 0.0003956342510268878,
      "loss": 2.303,
      "step": 91407
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.8319026231765747,
      "learning_rate": 0.0003956303738667787,
      "loss": 2.2231,
      "step": 91408
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.892388105392456,
      "learning_rate": 0.0003956264966888901,
      "loss": 2.1658,
      "step": 91409
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.8676403164863586,
      "learning_rate": 0.0003956226194932227,
      "loss": 2.0775,
      "step": 91410
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.8255048990249634,
      "learning_rate": 0.00039561874227977734,
      "loss": 1.8946,
      "step": 91411
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.8211084604263306,
      "learning_rate": 0.00039561486504855457,
      "loss": 2.0364,
      "step": 91412
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.8061556220054626,
      "learning_rate": 0.0003956109877995552,
      "loss": 2.1402,
      "step": 91413
    },
    {
      "epoch": 1.19,
      "grad_norm": 1.1291660070419312,
      "learning_rate": 0.00039560711053278,
      "loss": 2.2274,
      "step": 91414
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.8633515238761902,
      "learning_rate": 0.00039560323324822956,
      "loss": 2.2258,
      "step": 91415
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.7635767459869385,
      "learning_rate": 0.0003955993559459047,
      "loss": 2.4808,
      "step": 91416
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.8482922315597534,
      "learning_rate": 0.0003955954786258062,
      "loss": 2.1614,
      "step": 91417
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.8540142178535461,
      "learning_rate": 0.00039559160128793463,
      "loss": 2.0325,
      "step": 91418
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.9035782814025879,
      "learning_rate": 0.0003955877239322907,
      "loss": 2.1223,
      "step": 91419
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.7749481797218323,
      "learning_rate": 0.0003955838465588753,
      "loss": 2.2135,
      "step": 91420
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.9678675532341003,
      "learning_rate": 0.00039557996916768904,
      "loss": 1.981,
      "step": 91421
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.8391733169555664,
      "learning_rate": 0.00039557609175873263,
      "loss": 1.9541,
      "step": 91422
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.8896726965904236,
      "learning_rate": 0.0003955722143320069,
      "loss": 2.166,
      "step": 91423
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.868962824344635,
      "learning_rate": 0.0003955683368875124,
      "loss": 2.1898,
      "step": 91424
    },
    {
      "epoch": 1.19,
      "grad_norm": 1.0021473169326782,
      "learning_rate": 0.00039556445942524996,
      "loss": 2.254,
      "step": 91425
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.9218630790710449,
      "learning_rate": 0.00039556058194522035,
      "loss": 2.1155,
      "step": 91426
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.850396990776062,
      "learning_rate": 0.0003955567044474242,
      "loss": 2.0693,
      "step": 91427
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.9312206506729126,
      "learning_rate": 0.00039555282693186215,
      "loss": 2.2141,
      "step": 91428
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.936282753944397,
      "learning_rate": 0.0003955489493985351,
      "loss": 2.2693,
      "step": 91429
    },
    {
      "epoch": 1.19,
      "grad_norm": 1.0861259698867798,
      "learning_rate": 0.0003955450718474437,
      "loss": 2.0855,
      "step": 91430
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.8351918458938599,
      "learning_rate": 0.00039554119427858867,
      "loss": 2.1672,
      "step": 91431
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.8417282700538635,
      "learning_rate": 0.0003955373166919708,
      "loss": 2.1285,
      "step": 91432
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.9371906518936157,
      "learning_rate": 0.0003955334390875906,
      "loss": 2.0723,
      "step": 91433
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.9019535183906555,
      "learning_rate": 0.00039552956146544904,
      "loss": 2.1415,
      "step": 91434
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.9287500977516174,
      "learning_rate": 0.00039552568382554673,
      "loss": 2.3155,
      "step": 91435
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.9251075387001038,
      "learning_rate": 0.0003955218061678843,
      "loss": 2.2177,
      "step": 91436
    },
    {
      "epoch": 1.19,
      "grad_norm": 1.1308213472366333,
      "learning_rate": 0.0003955179284924626,
      "loss": 2.2161,
      "step": 91437
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.7543553113937378,
      "learning_rate": 0.0003955140507992824,
      "loss": 2.4529,
      "step": 91438
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.7577435970306396,
      "learning_rate": 0.00039551017308834423,
      "loss": 2.1725,
      "step": 91439
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.9576189517974854,
      "learning_rate": 0.00039550629535964896,
      "loss": 2.1986,
      "step": 91440
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.834442138671875,
      "learning_rate": 0.0003955024176131973,
      "loss": 2.2009,
      "step": 91441
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.7786297798156738,
      "learning_rate": 0.0003954985398489899,
      "loss": 2.3724,
      "step": 91442
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.7638933658599854,
      "learning_rate": 0.00039549466206702743,
      "loss": 2.2469,
      "step": 91443
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.9075636267662048,
      "learning_rate": 0.0003954907842673109,
      "loss": 2.3999,
      "step": 91444
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.7524944543838501,
      "learning_rate": 0.0003954869064498406,
      "loss": 1.9459,
      "step": 91445
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.8224500417709351,
      "learning_rate": 0.0003954830286146176,
      "loss": 2.0119,
      "step": 91446
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.8210703134536743,
      "learning_rate": 0.00039547915076164255,
      "loss": 2.1786,
      "step": 91447
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.9237495064735413,
      "learning_rate": 0.0003954752728909161,
      "loss": 2.1739,
      "step": 91448
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.9826359748840332,
      "learning_rate": 0.00039547139500243885,
      "loss": 2.2493,
      "step": 91449
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.849332869052887,
      "learning_rate": 0.0003954675170962118,
      "loss": 2.2553,
      "step": 91450
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.8357903361320496,
      "learning_rate": 0.0003954636391722355,
      "loss": 2.0921,
      "step": 91451
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.8319448828697205,
      "learning_rate": 0.0003954597612305107,
      "loss": 2.1139,
      "step": 91452
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.8967153429985046,
      "learning_rate": 0.0003954558832710382,
      "loss": 2.0366,
      "step": 91453
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.8432835340499878,
      "learning_rate": 0.00039545200529381863,
      "loss": 2.3217,
      "step": 91454
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.8971945643424988,
      "learning_rate": 0.0003954481272988527,
      "loss": 2.425,
      "step": 91455
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.8542948365211487,
      "learning_rate": 0.00039544424928614116,
      "loss": 2.2198,
      "step": 91456
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.8533622622489929,
      "learning_rate": 0.00039544037125568477,
      "loss": 2.1961,
      "step": 91457
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.8493611216545105,
      "learning_rate": 0.0003954364932074841,
      "loss": 2.0446,
      "step": 91458
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.8405565619468689,
      "learning_rate": 0.0003954326151415401,
      "loss": 2.3079,
      "step": 91459
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.9147021770477295,
      "learning_rate": 0.0003954287370578533,
      "loss": 2.2078,
      "step": 91460
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.8602644205093384,
      "learning_rate": 0.0003954248589564246,
      "loss": 2.0837,
      "step": 91461
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.8751770853996277,
      "learning_rate": 0.0003954209808372545,
      "loss": 2.1451,
      "step": 91462
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.9452441930770874,
      "learning_rate": 0.0003954171027003439,
      "loss": 2.2114,
      "step": 91463
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.9217247366905212,
      "learning_rate": 0.00039541322454569346,
      "loss": 2.328,
      "step": 91464
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.9288945198059082,
      "learning_rate": 0.0003954093463733039,
      "loss": 2.2801,
      "step": 91465
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.9371890425682068,
      "learning_rate": 0.0003954054681831759,
      "loss": 2.2077,
      "step": 91466
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.9374260306358337,
      "learning_rate": 0.0003954015899753103,
      "loss": 2.1352,
      "step": 91467
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.8280081152915955,
      "learning_rate": 0.00039539771174970764,
      "loss": 2.1436,
      "step": 91468
    },
    {
      "epoch": 1.19,
      "grad_norm": 1.005371332168579,
      "learning_rate": 0.00039539383350636886,
      "loss": 2.2921,
      "step": 91469
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.956892728805542,
      "learning_rate": 0.0003953899552452945,
      "loss": 2.2797,
      "step": 91470
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.8310815691947937,
      "learning_rate": 0.00039538607696648533,
      "loss": 2.1427,
      "step": 91471
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.9106509685516357,
      "learning_rate": 0.00039538219866994215,
      "loss": 2.195,
      "step": 91472
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.9082131385803223,
      "learning_rate": 0.00039537832035566555,
      "loss": 2.1362,
      "step": 91473
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.8450947403907776,
      "learning_rate": 0.0003953744420236564,
      "loss": 2.192,
      "step": 91474
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.8407156467437744,
      "learning_rate": 0.00039537056367391525,
      "loss": 2.3982,
      "step": 91475
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.7686874270439148,
      "learning_rate": 0.00039536668530644296,
      "loss": 2.118,
      "step": 91476
    },
    {
      "epoch": 1.19,
      "grad_norm": 1.027696132659912,
      "learning_rate": 0.0003953628069212402,
      "loss": 2.3206,
      "step": 91477
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.8619613647460938,
      "learning_rate": 0.00039535892851830766,
      "loss": 2.1356,
      "step": 91478
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.8352236747741699,
      "learning_rate": 0.0003953550500976462,
      "loss": 1.8258,
      "step": 91479
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.976717472076416,
      "learning_rate": 0.0003953511716592563,
      "loss": 2.0688,
      "step": 91480
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.790854275226593,
      "learning_rate": 0.00039534729320313895,
      "loss": 2.0603,
      "step": 91481
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.8906794786453247,
      "learning_rate": 0.0003953434147292947,
      "loss": 2.2529,
      "step": 91482
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.903708815574646,
      "learning_rate": 0.00039533953623772423,
      "loss": 2.1492,
      "step": 91483
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.8693709969520569,
      "learning_rate": 0.00039533565772842844,
      "loss": 2.1263,
      "step": 91484
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.8567737340927124,
      "learning_rate": 0.00039533177920140784,
      "loss": 2.3209,
      "step": 91485
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.8663100600242615,
      "learning_rate": 0.0003953279006566633,
      "loss": 2.1369,
      "step": 91486
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.8018127679824829,
      "learning_rate": 0.00039532402209419557,
      "loss": 2.3401,
      "step": 91487
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.8519169688224792,
      "learning_rate": 0.00039532014351400526,
      "loss": 2.0371,
      "step": 91488
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.7736290693283081,
      "learning_rate": 0.00039531626491609307,
      "loss": 2.038,
      "step": 91489
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.797435998916626,
      "learning_rate": 0.00039531238630046,
      "loss": 2.1199,
      "step": 91490
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.8792836666107178,
      "learning_rate": 0.00039530850766710635,
      "loss": 2.1695,
      "step": 91491
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.801571249961853,
      "learning_rate": 0.0003953046290160331,
      "loss": 2.1434,
      "step": 91492
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.9473614692687988,
      "learning_rate": 0.00039530075034724095,
      "loss": 2.0866,
      "step": 91493
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.9316802620887756,
      "learning_rate": 0.0003952968716607306,
      "loss": 2.1887,
      "step": 91494
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.8853083252906799,
      "learning_rate": 0.0003952929929565027,
      "loss": 2.3071,
      "step": 91495
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.8632752299308777,
      "learning_rate": 0.0003952891142345581,
      "loss": 1.9976,
      "step": 91496
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.9182299375534058,
      "learning_rate": 0.0003952852354948974,
      "loss": 2.1,
      "step": 91497
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.9092400670051575,
      "learning_rate": 0.0003952813567375214,
      "loss": 2.3083,
      "step": 91498
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.7967603206634521,
      "learning_rate": 0.0003952774779624308,
      "loss": 2.2194,
      "step": 91499
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.9181373119354248,
      "learning_rate": 0.0003952735991696263,
      "loss": 2.2365,
      "step": 91500
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.8135700821876526,
      "learning_rate": 0.0003952697203591087,
      "loss": 2.3234,
      "step": 91501
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.8055616021156311,
      "learning_rate": 0.0003952658415308786,
      "loss": 2.1247,
      "step": 91502
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.784437894821167,
      "learning_rate": 0.00039526196268493686,
      "loss": 2.1829,
      "step": 91503
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.7677285075187683,
      "learning_rate": 0.000395258083821284,
      "loss": 2.1734,
      "step": 91504
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.8703528046607971,
      "learning_rate": 0.000395254204939921,
      "loss": 2.362,
      "step": 91505
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.8704087734222412,
      "learning_rate": 0.00039525032604084833,
      "loss": 2.2389,
      "step": 91506
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.9074569940567017,
      "learning_rate": 0.00039524644712406686,
      "loss": 2.1591,
      "step": 91507
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.9001244306564331,
      "learning_rate": 0.00039524256818957733,
      "loss": 2.1434,
      "step": 91508
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.8885638117790222,
      "learning_rate": 0.00039523868923738034,
      "loss": 2.2102,
      "step": 91509
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.776445746421814,
      "learning_rate": 0.00039523481026747665,
      "loss": 2.1457,
      "step": 91510
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.8103981614112854,
      "learning_rate": 0.0003952309312798672,
      "loss": 2.3026,
      "step": 91511
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.8347382545471191,
      "learning_rate": 0.0003952270522745523,
      "loss": 2.2649,
      "step": 91512
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.86297208070755,
      "learning_rate": 0.000395223173251533,
      "loss": 2.237,
      "step": 91513
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.8979238867759705,
      "learning_rate": 0.00039521929421080983,
      "loss": 2.3537,
      "step": 91514
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.8984514474868774,
      "learning_rate": 0.0003952154151523837,
      "loss": 2.3424,
      "step": 91515
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.8391984701156616,
      "learning_rate": 0.0003952115360762552,
      "loss": 2.0285,
      "step": 91516
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.8489910364151001,
      "learning_rate": 0.00039520765698242507,
      "loss": 2.2467,
      "step": 91517
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.809748113155365,
      "learning_rate": 0.00039520377787089406,
      "loss": 2.1881,
      "step": 91518
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.795543909072876,
      "learning_rate": 0.00039519989874166285,
      "loss": 2.0264,
      "step": 91519
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.8716000914573669,
      "learning_rate": 0.00039519601959473214,
      "loss": 2.1435,
      "step": 91520
    },
    {
      "epoch": 1.19,
      "grad_norm": 1.0018556118011475,
      "learning_rate": 0.00039519214043010276,
      "loss": 2.1785,
      "step": 91521
    },
    {
      "epoch": 1.19,
      "grad_norm": 1.0016930103302002,
      "learning_rate": 0.00039518826124777535,
      "loss": 2.1913,
      "step": 91522
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.9895568490028381,
      "learning_rate": 0.0003951843820477506,
      "loss": 2.1099,
      "step": 91523
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.9294525980949402,
      "learning_rate": 0.00039518050283002937,
      "loss": 2.2124,
      "step": 91524
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.8511019945144653,
      "learning_rate": 0.0003951766235946122,
      "loss": 2.2684,
      "step": 91525
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.8595287799835205,
      "learning_rate": 0.00039517274434149984,
      "loss": 1.9812,
      "step": 91526
    },
    {
      "epoch": 1.19,
      "grad_norm": 1.094165563583374,
      "learning_rate": 0.0003951688650706933,
      "loss": 2.1422,
      "step": 91527
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.8725787401199341,
      "learning_rate": 0.00039516498578219284,
      "loss": 2.0723,
      "step": 91528
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.8375799059867859,
      "learning_rate": 0.0003951611064759995,
      "loss": 2.1173,
      "step": 91529
    },
    {
      "epoch": 1.19,
      "grad_norm": 1.0457552671432495,
      "learning_rate": 0.000395157227152114,
      "loss": 2.09,
      "step": 91530
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.9321812391281128,
      "learning_rate": 0.00039515334781053686,
      "loss": 2.1032,
      "step": 91531
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.8878771662712097,
      "learning_rate": 0.0003951494684512689,
      "loss": 2.21,
      "step": 91532
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.9574614763259888,
      "learning_rate": 0.000395145589074311,
      "loss": 2.0226,
      "step": 91533
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.8406484723091125,
      "learning_rate": 0.00039514170967966356,
      "loss": 2.2008,
      "step": 91534
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.8386602401733398,
      "learning_rate": 0.00039513783026732754,
      "loss": 2.1374,
      "step": 91535
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.8218105435371399,
      "learning_rate": 0.0003951339508373037,
      "loss": 2.036,
      "step": 91536
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.8599389791488647,
      "learning_rate": 0.0003951300713895926,
      "loss": 2.3216,
      "step": 91537
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.8331974148750305,
      "learning_rate": 0.00039512619192419495,
      "loss": 2.0497,
      "step": 91538
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.8281758427619934,
      "learning_rate": 0.0003951223124411117,
      "loss": 1.9593,
      "step": 91539
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.7746235132217407,
      "learning_rate": 0.00039511843294034326,
      "loss": 1.9862,
      "step": 91540
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.9052268266677856,
      "learning_rate": 0.0003951145534218906,
      "loss": 2.1619,
      "step": 91541
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.9397371411323547,
      "learning_rate": 0.0003951106738857543,
      "loss": 1.8687,
      "step": 91542
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.8655201196670532,
      "learning_rate": 0.00039510679433193513,
      "loss": 2.0969,
      "step": 91543
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.8590461015701294,
      "learning_rate": 0.00039510291476043384,
      "loss": 2.2437,
      "step": 91544
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.7776639461517334,
      "learning_rate": 0.00039509903517125105,
      "loss": 2.2703,
      "step": 91545
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.9148497581481934,
      "learning_rate": 0.0003950951555643877,
      "loss": 2.0911,
      "step": 91546
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.8059138655662537,
      "learning_rate": 0.00039509127593984425,
      "loss": 2.0744,
      "step": 91547
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.7556548118591309,
      "learning_rate": 0.0003950873962976216,
      "loss": 2.1998,
      "step": 91548
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.8272243738174438,
      "learning_rate": 0.00039508351663772036,
      "loss": 2.2952,
      "step": 91549
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.827103853225708,
      "learning_rate": 0.0003950796369601413,
      "loss": 2.2053,
      "step": 91550
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.8494204878807068,
      "learning_rate": 0.0003950757572648852,
      "loss": 2.2378,
      "step": 91551
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.8522118926048279,
      "learning_rate": 0.0003950718775519526,
      "loss": 2.4592,
      "step": 91552
    },
    {
      "epoch": 1.19,
      "grad_norm": 1.0336148738861084,
      "learning_rate": 0.00039506799782134446,
      "loss": 2.3939,
      "step": 91553
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.8572285175323486,
      "learning_rate": 0.00039506411807306135,
      "loss": 2.0779,
      "step": 91554
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.8815435171127319,
      "learning_rate": 0.000395060238307104,
      "loss": 2.3107,
      "step": 91555
    },
    {
      "epoch": 1.19,
      "grad_norm": 1.05448579788208,
      "learning_rate": 0.0003950563585234732,
      "loss": 2.0685,
      "step": 91556
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.9322043657302856,
      "learning_rate": 0.0003950524787221696,
      "loss": 2.3796,
      "step": 91557
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.8559421896934509,
      "learning_rate": 0.00039504859890319397,
      "loss": 1.9705,
      "step": 91558
    },
    {
      "epoch": 1.19,
      "grad_norm": 1.248970866203308,
      "learning_rate": 0.000395044719066547,
      "loss": 1.9078,
      "step": 91559
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.9109789729118347,
      "learning_rate": 0.0003950408392122295,
      "loss": 2.121,
      "step": 91560
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.8436809778213501,
      "learning_rate": 0.000395036959340242,
      "loss": 2.3198,
      "step": 91561
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.9096749424934387,
      "learning_rate": 0.0003950330794505853,
      "loss": 2.117,
      "step": 91562
    },
    {
      "epoch": 1.19,
      "grad_norm": 1.134514570236206,
      "learning_rate": 0.00039502919954326025,
      "loss": 2.3536,
      "step": 91563
    },
    {
      "epoch": 1.19,
      "grad_norm": 1.0165681838989258,
      "learning_rate": 0.00039502531961826753,
      "loss": 2.2762,
      "step": 91564
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.811212420463562,
      "learning_rate": 0.0003950214396756077,
      "loss": 2.2876,
      "step": 91565
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.8670061230659485,
      "learning_rate": 0.00039501755971528167,
      "loss": 2.2105,
      "step": 91566
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.9706565737724304,
      "learning_rate": 0.00039501367973729005,
      "loss": 2.1218,
      "step": 91567
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.9826804995536804,
      "learning_rate": 0.0003950097997416335,
      "loss": 2.3408,
      "step": 91568
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.7410681247711182,
      "learning_rate": 0.00039500591972831296,
      "loss": 2.2125,
      "step": 91569
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.8703898787498474,
      "learning_rate": 0.000395002039697329,
      "loss": 1.981,
      "step": 91570
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.9657683372497559,
      "learning_rate": 0.00039499815964868227,
      "loss": 2.0784,
      "step": 91571
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.8690030574798584,
      "learning_rate": 0.00039499427958237377,
      "loss": 2.0783,
      "step": 91572
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.7900545597076416,
      "learning_rate": 0.00039499039949840393,
      "loss": 2.1391,
      "step": 91573
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.8154890537261963,
      "learning_rate": 0.0003949865193967736,
      "loss": 2.3573,
      "step": 91574
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.7981997728347778,
      "learning_rate": 0.00039498263927748354,
      "loss": 2.0741,
      "step": 91575
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.93620365858078,
      "learning_rate": 0.00039497875914053433,
      "loss": 2.327,
      "step": 91576
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.8595119714736938,
      "learning_rate": 0.0003949748789859267,
      "loss": 2.0535,
      "step": 91577
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.8485203385353088,
      "learning_rate": 0.00039497099881366163,
      "loss": 2.3356,
      "step": 91578
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.8436183929443359,
      "learning_rate": 0.00039496711862373955,
      "loss": 2.1355,
      "step": 91579
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.8184921741485596,
      "learning_rate": 0.0003949632384161613,
      "loss": 2.1122,
      "step": 91580
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.8218779563903809,
      "learning_rate": 0.0003949593581909276,
      "loss": 2.163,
      "step": 91581
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.8307465314865112,
      "learning_rate": 0.0003949554779480392,
      "loss": 2.1676,
      "step": 91582
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.8564963936805725,
      "learning_rate": 0.0003949515976874967,
      "loss": 2.277,
      "step": 91583
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.8561463356018066,
      "learning_rate": 0.000394947717409301,
      "loss": 2.1072,
      "step": 91584
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.9384720921516418,
      "learning_rate": 0.0003949438371134527,
      "loss": 2.2569,
      "step": 91585
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.7916903495788574,
      "learning_rate": 0.0003949399567999525,
      "loss": 2.3503,
      "step": 91586
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.9096798300743103,
      "learning_rate": 0.00039493607646880115,
      "loss": 2.3666,
      "step": 91587
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.8445542454719543,
      "learning_rate": 0.00039493219611999955,
      "loss": 2.1424,
      "step": 91588
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.7954873442649841,
      "learning_rate": 0.00039492831575354816,
      "loss": 1.9375,
      "step": 91589
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.8213770389556885,
      "learning_rate": 0.00039492443536944774,
      "loss": 1.9816,
      "step": 91590
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.9113101363182068,
      "learning_rate": 0.0003949205549676992,
      "loss": 1.9177,
      "step": 91591
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.9175488948822021,
      "learning_rate": 0.00039491667454830305,
      "loss": 1.8875,
      "step": 91592
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.8996549844741821,
      "learning_rate": 0.00039491279411126007,
      "loss": 2.2528,
      "step": 91593
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.81107497215271,
      "learning_rate": 0.0003949089136565711,
      "loss": 1.9749,
      "step": 91594
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.8717154264450073,
      "learning_rate": 0.00039490503318423675,
      "loss": 2.1635,
      "step": 91595
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.8953549861907959,
      "learning_rate": 0.0003949011526942577,
      "loss": 2.1874,
      "step": 91596
    },
    {
      "epoch": 1.19,
      "grad_norm": 1.0586353540420532,
      "learning_rate": 0.00039489727218663484,
      "loss": 2.1382,
      "step": 91597
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.841274082660675,
      "learning_rate": 0.00039489339166136865,
      "loss": 2.2065,
      "step": 91598
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.8688493967056274,
      "learning_rate": 0.00039488951111846005,
      "loss": 2.3907,
      "step": 91599
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.9068520665168762,
      "learning_rate": 0.00039488563055790985,
      "loss": 2.0351,
      "step": 91600
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.8449715375900269,
      "learning_rate": 0.0003948817499797184,
      "loss": 2.2446,
      "step": 91601
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.8777676224708557,
      "learning_rate": 0.0003948778693838867,
      "loss": 2.1299,
      "step": 91602
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.8058589696884155,
      "learning_rate": 0.00039487398877041555,
      "loss": 2.0891,
      "step": 91603
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.8123148679733276,
      "learning_rate": 0.00039487010813930536,
      "loss": 2.2241,
      "step": 91604
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.8368995189666748,
      "learning_rate": 0.00039486622749055706,
      "loss": 2.3013,
      "step": 91605
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.7933443188667297,
      "learning_rate": 0.00039486234682417143,
      "loss": 1.992,
      "step": 91606
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.8303022980690002,
      "learning_rate": 0.000394858466140149,
      "loss": 2.1005,
      "step": 91607
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.8310586214065552,
      "learning_rate": 0.00039485458543849064,
      "loss": 2.2444,
      "step": 91608
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.9164220690727234,
      "learning_rate": 0.000394850704719197,
      "loss": 2.0814,
      "step": 91609
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.8025525808334351,
      "learning_rate": 0.00039484682398226883,
      "loss": 2.2161,
      "step": 91610
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.8435605764389038,
      "learning_rate": 0.0003948429432277068,
      "loss": 2.1405,
      "step": 91611
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.7893444299697876,
      "learning_rate": 0.0003948390624555118,
      "loss": 2.2159,
      "step": 91612
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.8544732332229614,
      "learning_rate": 0.0003948351816656843,
      "loss": 2.2094,
      "step": 91613
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.8549384474754333,
      "learning_rate": 0.00039483130085822516,
      "loss": 2.2308,
      "step": 91614
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.8281860947608948,
      "learning_rate": 0.0003948274200331352,
      "loss": 2.2511,
      "step": 91615
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.843063473701477,
      "learning_rate": 0.00039482353919041496,
      "loss": 2.3338,
      "step": 91616
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.7498888373374939,
      "learning_rate": 0.00039481965833006524,
      "loss": 2.1044,
      "step": 91617
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.8938072919845581,
      "learning_rate": 0.0003948157774520868,
      "loss": 2.0365,
      "step": 91618
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.8694784641265869,
      "learning_rate": 0.0003948118965564802,
      "loss": 2.2255,
      "step": 91619
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.8000938892364502,
      "learning_rate": 0.0003948080156432463,
      "loss": 2.0458,
      "step": 91620
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.8245246410369873,
      "learning_rate": 0.000394804134712386,
      "loss": 2.0594,
      "step": 91621
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.9097487330436707,
      "learning_rate": 0.0003948002537638996,
      "loss": 2.1646,
      "step": 91622
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.8043942451477051,
      "learning_rate": 0.00039479637279778813,
      "loss": 2.1525,
      "step": 91623
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.8466598987579346,
      "learning_rate": 0.0003947924918140522,
      "loss": 2.3059,
      "step": 91624
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.7833845019340515,
      "learning_rate": 0.00039478861081269256,
      "loss": 1.9486,
      "step": 91625
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.8239210247993469,
      "learning_rate": 0.00039478472979370997,
      "loss": 2.1509,
      "step": 91626
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.8378686904907227,
      "learning_rate": 0.00039478084875710503,
      "loss": 2.2177,
      "step": 91627
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.8174523711204529,
      "learning_rate": 0.0003947769677028786,
      "loss": 2.0956,
      "step": 91628
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.7839862704277039,
      "learning_rate": 0.0003947730866310313,
      "loss": 2.2283,
      "step": 91629
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.8136284947395325,
      "learning_rate": 0.000394769205541564,
      "loss": 1.9927,
      "step": 91630
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.7927032113075256,
      "learning_rate": 0.0003947653244344772,
      "loss": 2.1182,
      "step": 91631
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.7790741324424744,
      "learning_rate": 0.00039476144330977174,
      "loss": 2.0835,
      "step": 91632
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.7487818598747253,
      "learning_rate": 0.00039475756216744836,
      "loss": 2.0818,
      "step": 91633
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.879520833492279,
      "learning_rate": 0.00039475368100750784,
      "loss": 2.2322,
      "step": 91634
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.8479780554771423,
      "learning_rate": 0.0003947497998299507,
      "loss": 2.3674,
      "step": 91635
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.8652098774909973,
      "learning_rate": 0.0003947459186347779,
      "loss": 2.196,
      "step": 91636
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.8106311559677124,
      "learning_rate": 0.0003947420374219899,
      "loss": 2.0347,
      "step": 91637
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.8086651563644409,
      "learning_rate": 0.0003947381561915877,
      "loss": 2.1409,
      "step": 91638
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.7652802467346191,
      "learning_rate": 0.0003947342749435718,
      "loss": 2.1856,
      "step": 91639
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.9993438124656677,
      "learning_rate": 0.00039473039367794304,
      "loss": 2.2578,
      "step": 91640
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.7461583018302917,
      "learning_rate": 0.00039472651239470216,
      "loss": 2.233,
      "step": 91641
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.8521125316619873,
      "learning_rate": 0.0003947226310938498,
      "loss": 2.1417,
      "step": 91642
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.8222424983978271,
      "learning_rate": 0.0003947187497753867,
      "loss": 2.1032,
      "step": 91643
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.8225268125534058,
      "learning_rate": 0.0003947148684393136,
      "loss": 2.1575,
      "step": 91644
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.7935507893562317,
      "learning_rate": 0.0003947109870856312,
      "loss": 2.1273,
      "step": 91645
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.7905954122543335,
      "learning_rate": 0.00039470710571434033,
      "loss": 2.3299,
      "step": 91646
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.9913087487220764,
      "learning_rate": 0.0003947032243254415,
      "loss": 2.2663,
      "step": 91647
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.8388283848762512,
      "learning_rate": 0.00039469934291893563,
      "loss": 2.2924,
      "step": 91648
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.8135167956352234,
      "learning_rate": 0.00039469546149482334,
      "loss": 2.3091,
      "step": 91649
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.8796373605728149,
      "learning_rate": 0.00039469158005310534,
      "loss": 2.2653,
      "step": 91650
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.8583745956420898,
      "learning_rate": 0.0003946876985937824,
      "loss": 2.1939,
      "step": 91651
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.8494903445243835,
      "learning_rate": 0.00039468381711685535,
      "loss": 2.3399,
      "step": 91652
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.809287965297699,
      "learning_rate": 0.00039467993562232467,
      "loss": 2.3704,
      "step": 91653
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.8129223585128784,
      "learning_rate": 0.0003946760541101912,
      "loss": 2.0387,
      "step": 91654
    },
    {
      "epoch": 1.19,
      "grad_norm": 1.021252989768982,
      "learning_rate": 0.0003946721725804557,
      "loss": 2.2269,
      "step": 91655
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.8477537631988525,
      "learning_rate": 0.00039466829103311887,
      "loss": 2.3466,
      "step": 91656
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.9840741157531738,
      "learning_rate": 0.0003946644094681814,
      "loss": 2.1542,
      "step": 91657
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.8736782073974609,
      "learning_rate": 0.00039466052788564407,
      "loss": 2.0928,
      "step": 91658
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.8009153604507446,
      "learning_rate": 0.0003946566462855075,
      "loss": 2.1351,
      "step": 91659
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.8741542100906372,
      "learning_rate": 0.00039465276466777246,
      "loss": 2.0841,
      "step": 91660
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.8258858919143677,
      "learning_rate": 0.00039464888303243975,
      "loss": 2.0833,
      "step": 91661
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.9937379360198975,
      "learning_rate": 0.00039464500137951,
      "loss": 2.0651,
      "step": 91662
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.8599765300750732,
      "learning_rate": 0.00039464111970898393,
      "loss": 2.1321,
      "step": 91663
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.7709571123123169,
      "learning_rate": 0.00039463723802086233,
      "loss": 2.1409,
      "step": 91664
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.8016749620437622,
      "learning_rate": 0.0003946333563151458,
      "loss": 2.3104,
      "step": 91665
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.8769288659095764,
      "learning_rate": 0.00039462947459183525,
      "loss": 2.1858,
      "step": 91666
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.9401578307151794,
      "learning_rate": 0.0003946255928509312,
      "loss": 2.4926,
      "step": 91667
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.8959819078445435,
      "learning_rate": 0.0003946217110924346,
      "loss": 2.1243,
      "step": 91668
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.7907621264457703,
      "learning_rate": 0.0003946178293163459,
      "loss": 2.2948,
      "step": 91669
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.9036370515823364,
      "learning_rate": 0.0003946139475226661,
      "loss": 2.2016,
      "step": 91670
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.9933182597160339,
      "learning_rate": 0.0003946100657113957,
      "loss": 2.1613,
      "step": 91671
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.7821400761604309,
      "learning_rate": 0.00039460618388253543,
      "loss": 2.34,
      "step": 91672
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.8672797083854675,
      "learning_rate": 0.00039460230203608617,
      "loss": 2.2694,
      "step": 91673
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.8901666402816772,
      "learning_rate": 0.00039459842017204854,
      "loss": 2.141,
      "step": 91674
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.791086733341217,
      "learning_rate": 0.0003945945382904232,
      "loss": 1.9661,
      "step": 91675
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.8073869347572327,
      "learning_rate": 0.00039459065639121113,
      "loss": 2.1317,
      "step": 91676
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.8962896466255188,
      "learning_rate": 0.0003945867744744127,
      "loss": 2.0912,
      "step": 91677
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.8357825875282288,
      "learning_rate": 0.00039458289254002885,
      "loss": 2.0798,
      "step": 91678
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.7545943260192871,
      "learning_rate": 0.00039457901058806037,
      "loss": 2.1071,
      "step": 91679
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.9872564077377319,
      "learning_rate": 0.0003945751286185077,
      "loss": 2.3091,
      "step": 91680
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.8611939549446106,
      "learning_rate": 0.0003945712466313718,
      "loss": 2.1569,
      "step": 91681
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.9744551777839661,
      "learning_rate": 0.0003945673646266533,
      "loss": 2.1782,
      "step": 91682
    },
    {
      "epoch": 1.19,
      "grad_norm": 1.0225123167037964,
      "learning_rate": 0.000394563482604353,
      "loss": 2.0632,
      "step": 91683
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.9219233989715576,
      "learning_rate": 0.0003945596005644715,
      "loss": 1.9799,
      "step": 91684
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.8265978693962097,
      "learning_rate": 0.0003945557185070097,
      "loss": 2.1526,
      "step": 91685
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.8487629890441895,
      "learning_rate": 0.00039455183643196805,
      "loss": 2.246,
      "step": 91686
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.8192874193191528,
      "learning_rate": 0.0003945479543393475,
      "loss": 2.0512,
      "step": 91687
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.9380719661712646,
      "learning_rate": 0.0003945440722291487,
      "loss": 2.3025,
      "step": 91688
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.8728495240211487,
      "learning_rate": 0.00039454019010137233,
      "loss": 1.9121,
      "step": 91689
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.8823915719985962,
      "learning_rate": 0.0003945363079560192,
      "loss": 2.1361,
      "step": 91690
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.9902060031890869,
      "learning_rate": 0.00039453242579309003,
      "loss": 2.1397,
      "step": 91691
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.7924383878707886,
      "learning_rate": 0.0003945285436125854,
      "loss": 2.0461,
      "step": 91692
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.8520394563674927,
      "learning_rate": 0.0003945246614145062,
      "loss": 2.2525,
      "step": 91693
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.914190411567688,
      "learning_rate": 0.00039452077919885305,
      "loss": 2.1807,
      "step": 91694
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.8773478865623474,
      "learning_rate": 0.0003945168969656267,
      "loss": 2.1395,
      "step": 91695
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.8175010681152344,
      "learning_rate": 0.00039451301471482786,
      "loss": 2.1808,
      "step": 91696
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.7954050898551941,
      "learning_rate": 0.00039450913244645737,
      "loss": 2.3298,
      "step": 91697
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.8721302151679993,
      "learning_rate": 0.00039450525016051575,
      "loss": 2.2306,
      "step": 91698
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.8449352979660034,
      "learning_rate": 0.0003945013678570038,
      "loss": 2.1539,
      "step": 91699
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.8433314561843872,
      "learning_rate": 0.00039449748553592234,
      "loss": 2.3174,
      "step": 91700
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.8901588916778564,
      "learning_rate": 0.00039449360319727196,
      "loss": 2.4095,
      "step": 91701
    },
    {
      "epoch": 1.19,
      "grad_norm": 1.0708609819412231,
      "learning_rate": 0.0003944897208410534,
      "loss": 2.3583,
      "step": 91702
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.8523502945899963,
      "learning_rate": 0.00039448583846726754,
      "loss": 2.3391,
      "step": 91703
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.8237109184265137,
      "learning_rate": 0.00039448195607591486,
      "loss": 2.063,
      "step": 91704
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.9203377962112427,
      "learning_rate": 0.00039447807366699625,
      "loss": 2.3049,
      "step": 91705
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.8681074976921082,
      "learning_rate": 0.00039447419124051243,
      "loss": 2.2755,
      "step": 91706
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.8213306069374084,
      "learning_rate": 0.000394470308796464,
      "loss": 2.3287,
      "step": 91707
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.7637680172920227,
      "learning_rate": 0.00039446642633485175,
      "loss": 2.1587,
      "step": 91708
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.9168334007263184,
      "learning_rate": 0.00039446254385567647,
      "loss": 2.0128,
      "step": 91709
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.8590618371963501,
      "learning_rate": 0.00039445866135893873,
      "loss": 2.3017,
      "step": 91710
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.7864144444465637,
      "learning_rate": 0.0003944547788446394,
      "loss": 2.3511,
      "step": 91711
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.8488500118255615,
      "learning_rate": 0.0003944508963127792,
      "loss": 2.0538,
      "step": 91712
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.9046639800071716,
      "learning_rate": 0.0003944470137633587,
      "loss": 2.1056,
      "step": 91713
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.884233295917511,
      "learning_rate": 0.0003944431311963788,
      "loss": 2.2156,
      "step": 91714
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.9098717570304871,
      "learning_rate": 0.00039443924861184005,
      "loss": 2.3107,
      "step": 91715
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.802433967590332,
      "learning_rate": 0.00039443536600974335,
      "loss": 2.125,
      "step": 91716
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.7865567803382874,
      "learning_rate": 0.00039443148339008923,
      "loss": 1.9113,
      "step": 91717
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.8192927837371826,
      "learning_rate": 0.0003944276007528786,
      "loss": 2.0198,
      "step": 91718
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.8123789429664612,
      "learning_rate": 0.0003944237180981121,
      "loss": 1.9852,
      "step": 91719
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.7968603372573853,
      "learning_rate": 0.0003944198354257904,
      "loss": 2.2998,
      "step": 91720
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.8542242050170898,
      "learning_rate": 0.00039441595273591426,
      "loss": 2.257,
      "step": 91721
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.8825486302375793,
      "learning_rate": 0.0003944120700284845,
      "loss": 2.1356,
      "step": 91722
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.8173959851264954,
      "learning_rate": 0.00039440818730350166,
      "loss": 2.1251,
      "step": 91723
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.9825518131256104,
      "learning_rate": 0.00039440430456096656,
      "loss": 2.2585,
      "step": 91724
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.9793527126312256,
      "learning_rate": 0.00039440042180087997,
      "loss": 2.08,
      "step": 91725
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.8405850529670715,
      "learning_rate": 0.0003943965390232426,
      "loss": 2.2378,
      "step": 91726
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.8961057662963867,
      "learning_rate": 0.000394392656228055,
      "loss": 2.2077,
      "step": 91727
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.7486016750335693,
      "learning_rate": 0.0003943887734153181,
      "loss": 2.1419,
      "step": 91728
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.8368508219718933,
      "learning_rate": 0.0003943848905850325,
      "loss": 2.1125,
      "step": 91729
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.842474639415741,
      "learning_rate": 0.00039438100773719903,
      "loss": 2.2494,
      "step": 91730
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.863252580165863,
      "learning_rate": 0.00039437712487181835,
      "loss": 2.1271,
      "step": 91731
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.848237931728363,
      "learning_rate": 0.0003943732419888911,
      "loss": 1.949,
      "step": 91732
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.8359989523887634,
      "learning_rate": 0.0003943693590884182,
      "loss": 2.3278,
      "step": 91733
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.8746153712272644,
      "learning_rate": 0.0003943654761704002,
      "loss": 2.2977,
      "step": 91734
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.920570433139801,
      "learning_rate": 0.0003943615932348378,
      "loss": 2.1849,
      "step": 91735
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.8040045499801636,
      "learning_rate": 0.0003943577102817319,
      "loss": 2.2286,
      "step": 91736
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.8486584424972534,
      "learning_rate": 0.00039435382731108315,
      "loss": 2.0747,
      "step": 91737
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.895352303981781,
      "learning_rate": 0.00039434994432289213,
      "loss": 2.13,
      "step": 91738
    },
    {
      "epoch": 1.19,
      "grad_norm": 1.6371002197265625,
      "learning_rate": 0.0003943460613171597,
      "loss": 2.0677,
      "step": 91739
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.7902197241783142,
      "learning_rate": 0.00039434217829388665,
      "loss": 2.0604,
      "step": 91740
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.9585888981819153,
      "learning_rate": 0.00039433829525307355,
      "loss": 2.2929,
      "step": 91741
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.8583475947380066,
      "learning_rate": 0.0003943344121947211,
      "loss": 2.201,
      "step": 91742
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.817454993724823,
      "learning_rate": 0.00039433052911883027,
      "loss": 2.0825,
      "step": 91743
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.9496026039123535,
      "learning_rate": 0.00039432664602540143,
      "loss": 1.983,
      "step": 91744
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.8542998433113098,
      "learning_rate": 0.00039432276291443563,
      "loss": 2.4975,
      "step": 91745
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.8361411094665527,
      "learning_rate": 0.0003943188797859334,
      "loss": 2.1497,
      "step": 91746
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.7702373266220093,
      "learning_rate": 0.0003943149966398955,
      "loss": 2.2093,
      "step": 91747
    },
    {
      "epoch": 1.19,
      "grad_norm": 1.098766803741455,
      "learning_rate": 0.0003943111134763226,
      "loss": 2.102,
      "step": 91748
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.85240238904953,
      "learning_rate": 0.00039430723029521556,
      "loss": 2.094,
      "step": 91749
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.846552848815918,
      "learning_rate": 0.000394303347096575,
      "loss": 2.3511,
      "step": 91750
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.7908089756965637,
      "learning_rate": 0.0003942994638804017,
      "loss": 2.2637,
      "step": 91751
    },
    {
      "epoch": 1.19,
      "grad_norm": 1.0189687013626099,
      "learning_rate": 0.0003942955806466964,
      "loss": 2.2175,
      "step": 91752
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.8695154786109924,
      "learning_rate": 0.00039429169739545964,
      "loss": 2.2018,
      "step": 91753
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.9309644103050232,
      "learning_rate": 0.00039428781412669233,
      "loss": 2.1032,
      "step": 91754
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.8497940301895142,
      "learning_rate": 0.00039428393084039515,
      "loss": 2.136,
      "step": 91755
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.8256933093070984,
      "learning_rate": 0.0003942800475365688,
      "loss": 2.2368,
      "step": 91756
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.9887211322784424,
      "learning_rate": 0.00039427616421521395,
      "loss": 2.2564,
      "step": 91757
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.8091623783111572,
      "learning_rate": 0.0003942722808763315,
      "loss": 2.2549,
      "step": 91758
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.8421741724014282,
      "learning_rate": 0.000394268397519922,
      "loss": 2.2377,
      "step": 91759
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.9822934865951538,
      "learning_rate": 0.00039426451414598617,
      "loss": 2.0669,
      "step": 91760
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.8747441172599792,
      "learning_rate": 0.0003942606307545249,
      "loss": 2.1453,
      "step": 91761
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.8083796501159668,
      "learning_rate": 0.00039425674734553875,
      "loss": 2.284,
      "step": 91762
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.869252622127533,
      "learning_rate": 0.00039425286391902843,
      "loss": 2.3316,
      "step": 91763
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.8815444111824036,
      "learning_rate": 0.00039424898047499485,
      "loss": 2.2426,
      "step": 91764
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.852456271648407,
      "learning_rate": 0.0003942450970134385,
      "loss": 2.0193,
      "step": 91765
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.8256045579910278,
      "learning_rate": 0.00039424121353436015,
      "loss": 2.1655,
      "step": 91766
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.827214241027832,
      "learning_rate": 0.00039423733003776077,
      "loss": 2.1188,
      "step": 91767
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.7970988154411316,
      "learning_rate": 0.0003942334465236407,
      "loss": 2.2262,
      "step": 91768
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.8713565468788147,
      "learning_rate": 0.0003942295629920009,
      "loss": 2.3604,
      "step": 91769
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.8827913403511047,
      "learning_rate": 0.00039422567944284216,
      "loss": 2.4436,
      "step": 91770
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.9236540794372559,
      "learning_rate": 0.000394221795876165,
      "loss": 2.0371,
      "step": 91771
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.8662801384925842,
      "learning_rate": 0.0003942179122919702,
      "loss": 2.3662,
      "step": 91772
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.8161175847053528,
      "learning_rate": 0.00039421402869025866,
      "loss": 2.0174,
      "step": 91773
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.8229676485061646,
      "learning_rate": 0.00039421014507103074,
      "loss": 2.225,
      "step": 91774
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.8949617147445679,
      "learning_rate": 0.0003942062614342875,
      "loss": 2.0865,
      "step": 91775
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.9641573429107666,
      "learning_rate": 0.0003942023777800295,
      "loss": 2.2226,
      "step": 91776
    },
    {
      "epoch": 1.19,
      "grad_norm": 0.9104441404342651,
      "learning_rate": 0.0003941984941082575,
      "loss": 2.2878,
      "step": 91777
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.8446326851844788,
      "learning_rate": 0.00039419461041897226,
      "loss": 2.3198,
      "step": 91778
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.8596380949020386,
      "learning_rate": 0.0003941907267121745,
      "loss": 2.0484,
      "step": 91779
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.9387996792793274,
      "learning_rate": 0.0003941868429878648,
      "loss": 2.1818,
      "step": 91780
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.9164173007011414,
      "learning_rate": 0.00039418295924604406,
      "loss": 2.128,
      "step": 91781
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.7866369485855103,
      "learning_rate": 0.00039417907548671284,
      "loss": 2.1679,
      "step": 91782
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.8200735449790955,
      "learning_rate": 0.00039417519170987207,
      "loss": 2.1867,
      "step": 91783
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.9482112526893616,
      "learning_rate": 0.00039417130791552236,
      "loss": 2.1582,
      "step": 91784
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.7886145710945129,
      "learning_rate": 0.0003941674241036643,
      "loss": 2.0857,
      "step": 91785
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.7799171805381775,
      "learning_rate": 0.0003941635402742988,
      "loss": 2.176,
      "step": 91786
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.9120210409164429,
      "learning_rate": 0.0003941596564274265,
      "loss": 2.0413,
      "step": 91787
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.8707389235496521,
      "learning_rate": 0.00039415577256304824,
      "loss": 2.2474,
      "step": 91788
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.8997530937194824,
      "learning_rate": 0.0003941518886811646,
      "loss": 1.9056,
      "step": 91789
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.8177822232246399,
      "learning_rate": 0.0003941480047817763,
      "loss": 2.4525,
      "step": 91790
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.8023307919502258,
      "learning_rate": 0.0003941441208648841,
      "loss": 2.2755,
      "step": 91791
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.7431899905204773,
      "learning_rate": 0.0003941402369304888,
      "loss": 1.9077,
      "step": 91792
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.8962255120277405,
      "learning_rate": 0.00039413635297859097,
      "loss": 2.2277,
      "step": 91793
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.9067578911781311,
      "learning_rate": 0.0003941324690091915,
      "loss": 2.1193,
      "step": 91794
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.8742855191230774,
      "learning_rate": 0.00039412858502229096,
      "loss": 2.1991,
      "step": 91795
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.891450822353363,
      "learning_rate": 0.00039412470101789015,
      "loss": 2.3966,
      "step": 91796
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.9192182421684265,
      "learning_rate": 0.00039412081699598977,
      "loss": 2.2409,
      "step": 91797
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.8860965967178345,
      "learning_rate": 0.0003941169329565907,
      "loss": 2.0658,
      "step": 91798
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.808427631855011,
      "learning_rate": 0.00039411304889969335,
      "loss": 2.0251,
      "step": 91799
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.8777068257331848,
      "learning_rate": 0.00039410916482529866,
      "loss": 2.3647,
      "step": 91800
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.777128279209137,
      "learning_rate": 0.00039410528073340734,
      "loss": 2.1155,
      "step": 91801
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.8121058344841003,
      "learning_rate": 0.00039410139662402,
      "loss": 2.294,
      "step": 91802
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.9041941165924072,
      "learning_rate": 0.00039409751249713744,
      "loss": 2.5315,
      "step": 91803
    },
    {
      "epoch": 1.2,
      "grad_norm": 1.004225492477417,
      "learning_rate": 0.00039409362835276043,
      "loss": 2.1969,
      "step": 91804
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.8792317509651184,
      "learning_rate": 0.00039408974419088955,
      "loss": 2.1694,
      "step": 91805
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.889568567276001,
      "learning_rate": 0.00039408586001152566,
      "loss": 2.3531,
      "step": 91806
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.8838614821434021,
      "learning_rate": 0.0003940819758146695,
      "loss": 2.0128,
      "step": 91807
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.8404428362846375,
      "learning_rate": 0.00039407809160032164,
      "loss": 2.1263,
      "step": 91808
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.8143512010574341,
      "learning_rate": 0.00039407420736848285,
      "loss": 2.2687,
      "step": 91809
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.8271523118019104,
      "learning_rate": 0.00039407032311915394,
      "loss": 2.0533,
      "step": 91810
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.7642010450363159,
      "learning_rate": 0.00039406643885233565,
      "loss": 2.2962,
      "step": 91811
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.7803175449371338,
      "learning_rate": 0.0003940625545680286,
      "loss": 1.9538,
      "step": 91812
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.9082730412483215,
      "learning_rate": 0.00039405867026623344,
      "loss": 2.1366,
      "step": 91813
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.932032585144043,
      "learning_rate": 0.00039405478594695115,
      "loss": 2.0198,
      "step": 91814
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.8265206813812256,
      "learning_rate": 0.0003940509016101822,
      "loss": 2.178,
      "step": 91815
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.8551540374755859,
      "learning_rate": 0.0003940470172559274,
      "loss": 2.0688,
      "step": 91816
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.8472128510475159,
      "learning_rate": 0.00039404313288418755,
      "loss": 1.969,
      "step": 91817
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.7769277691841125,
      "learning_rate": 0.00039403924849496327,
      "loss": 2.1141,
      "step": 91818
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.8802958130836487,
      "learning_rate": 0.0003940353640882553,
      "loss": 2.3235,
      "step": 91819
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.9165925979614258,
      "learning_rate": 0.0003940314796640644,
      "loss": 2.265,
      "step": 91820
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.7898982763290405,
      "learning_rate": 0.00039402759522239137,
      "loss": 1.9285,
      "step": 91821
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.8032076954841614,
      "learning_rate": 0.0003940237107632367,
      "loss": 1.9361,
      "step": 91822
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.850124716758728,
      "learning_rate": 0.00039401982628660127,
      "loss": 2.0818,
      "step": 91823
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.9007251858711243,
      "learning_rate": 0.00039401594179248574,
      "loss": 2.3661,
      "step": 91824
    },
    {
      "epoch": 1.2,
      "grad_norm": 1.1597943305969238,
      "learning_rate": 0.00039401205728089103,
      "loss": 1.9713,
      "step": 91825
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.765291154384613,
      "learning_rate": 0.00039400817275181757,
      "loss": 2.1513,
      "step": 91826
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.7678432464599609,
      "learning_rate": 0.00039400428820526623,
      "loss": 2.1737,
      "step": 91827
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.9379814863204956,
      "learning_rate": 0.00039400040364123777,
      "loss": 2.1523,
      "step": 91828
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.8105406165122986,
      "learning_rate": 0.00039399651905973283,
      "loss": 2.262,
      "step": 91829
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.9040303230285645,
      "learning_rate": 0.00039399263446075213,
      "loss": 2.1313,
      "step": 91830
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.9101884365081787,
      "learning_rate": 0.00039398874984429653,
      "loss": 2.2741,
      "step": 91831
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.8400135040283203,
      "learning_rate": 0.00039398486521036657,
      "loss": 2.2655,
      "step": 91832
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.9535596370697021,
      "learning_rate": 0.000393980980558963,
      "loss": 2.1171,
      "step": 91833
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.7908989191055298,
      "learning_rate": 0.0003939770958900867,
      "loss": 2.1965,
      "step": 91834
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.8768185377120972,
      "learning_rate": 0.00039397321120373825,
      "loss": 2.1615,
      "step": 91835
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.8492464423179626,
      "learning_rate": 0.0003939693264999184,
      "loss": 2.212,
      "step": 91836
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.8760568499565125,
      "learning_rate": 0.00039396544177862787,
      "loss": 2.2493,
      "step": 91837
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.8226876258850098,
      "learning_rate": 0.0003939615570398674,
      "loss": 2.1922,
      "step": 91838
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.8569003939628601,
      "learning_rate": 0.0003939576722836377,
      "loss": 2.4383,
      "step": 91839
    },
    {
      "epoch": 1.2,
      "grad_norm": 1.0107892751693726,
      "learning_rate": 0.0003939537875099395,
      "loss": 2.3733,
      "step": 91840
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.8871530890464783,
      "learning_rate": 0.0003939499027187735,
      "loss": 2.2919,
      "step": 91841
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.8142459988594055,
      "learning_rate": 0.00039394601791014033,
      "loss": 2.078,
      "step": 91842
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.8382718563079834,
      "learning_rate": 0.000393942133084041,
      "loss": 2.0595,
      "step": 91843
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.8695588707923889,
      "learning_rate": 0.00039393824824047606,
      "loss": 2.0165,
      "step": 91844
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.7857484817504883,
      "learning_rate": 0.0003939343633794461,
      "loss": 2.2091,
      "step": 91845
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.9958765506744385,
      "learning_rate": 0.000393930478500952,
      "loss": 2.1036,
      "step": 91846
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.9180632829666138,
      "learning_rate": 0.0003939265936049945,
      "loss": 2.2281,
      "step": 91847
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.8109950423240662,
      "learning_rate": 0.00039392270869157423,
      "loss": 2.0967,
      "step": 91848
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.8498244285583496,
      "learning_rate": 0.0003939188237606919,
      "loss": 2.3508,
      "step": 91849
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.8217765688896179,
      "learning_rate": 0.0003939149388123485,
      "loss": 2.2835,
      "step": 91850
    },
    {
      "epoch": 1.2,
      "grad_norm": 1.0651130676269531,
      "learning_rate": 0.0003939110538465444,
      "loss": 2.1772,
      "step": 91851
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.8470343351364136,
      "learning_rate": 0.00039390716886328033,
      "loss": 2.124,
      "step": 91852
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.9244207143783569,
      "learning_rate": 0.00039390328386255735,
      "loss": 2.4074,
      "step": 91853
    },
    {
      "epoch": 1.2,
      "grad_norm": 1.016983151435852,
      "learning_rate": 0.00039389939884437595,
      "loss": 2.2008,
      "step": 91854
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.9030957818031311,
      "learning_rate": 0.00039389551380873683,
      "loss": 2.1095,
      "step": 91855
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.8562302589416504,
      "learning_rate": 0.00039389162875564076,
      "loss": 2.2415,
      "step": 91856
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.9006220102310181,
      "learning_rate": 0.00039388774368508844,
      "loss": 2.3148,
      "step": 91857
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.9975855350494385,
      "learning_rate": 0.0003938838585970807,
      "loss": 2.4394,
      "step": 91858
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.8110556602478027,
      "learning_rate": 0.00039387997349161813,
      "loss": 2.2892,
      "step": 91859
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.9238741397857666,
      "learning_rate": 0.0003938760883687015,
      "loss": 2.2415,
      "step": 91860
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.8721435070037842,
      "learning_rate": 0.0003938722032283315,
      "loss": 2.1693,
      "step": 91861
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.8103213310241699,
      "learning_rate": 0.0003938683180705089,
      "loss": 2.1469,
      "step": 91862
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.8049274682998657,
      "learning_rate": 0.0003938644328952344,
      "loss": 2.4109,
      "step": 91863
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.9636561870574951,
      "learning_rate": 0.0003938605477025088,
      "loss": 2.4016,
      "step": 91864
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.8927709460258484,
      "learning_rate": 0.00039385666249233274,
      "loss": 2.1734,
      "step": 91865
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.8341097831726074,
      "learning_rate": 0.0003938527772647069,
      "loss": 2.2554,
      "step": 91866
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.8504601716995239,
      "learning_rate": 0.000393848892019632,
      "loss": 2.0894,
      "step": 91867
    },
    {
      "epoch": 1.2,
      "grad_norm": 1.244897723197937,
      "learning_rate": 0.000393845006757109,
      "loss": 2.1506,
      "step": 91868
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.7577969431877136,
      "learning_rate": 0.0003938411214771383,
      "loss": 2.0698,
      "step": 91869
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.8565559387207031,
      "learning_rate": 0.00039383723617972076,
      "loss": 2.1286,
      "step": 91870
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.8831819295883179,
      "learning_rate": 0.0003938333508648572,
      "loss": 2.4924,
      "step": 91871
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.9042195677757263,
      "learning_rate": 0.0003938294655325482,
      "loss": 2.2584,
      "step": 91872
    },
    {
      "epoch": 1.2,
      "grad_norm": 1.0648280382156372,
      "learning_rate": 0.00039382558018279454,
      "loss": 2.4731,
      "step": 91873
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.9511756300926208,
      "learning_rate": 0.00039382169481559697,
      "loss": 2.0776,
      "step": 91874
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.9107614159584045,
      "learning_rate": 0.00039381780943095605,
      "loss": 2.2153,
      "step": 91875
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.7602345943450928,
      "learning_rate": 0.00039381392402887275,
      "loss": 2.0882,
      "step": 91876
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.8613735437393188,
      "learning_rate": 0.0003938100386093476,
      "loss": 2.3258,
      "step": 91877
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.8964778780937195,
      "learning_rate": 0.0003938061531723814,
      "loss": 2.3514,
      "step": 91878
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.8278912901878357,
      "learning_rate": 0.0003938022677179749,
      "loss": 2.0331,
      "step": 91879
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.9256500005722046,
      "learning_rate": 0.00039379838224612885,
      "loss": 2.2046,
      "step": 91880
    },
    {
      "epoch": 1.2,
      "grad_norm": 1.0387468338012695,
      "learning_rate": 0.00039379449675684377,
      "loss": 2.3448,
      "step": 91881
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.8395124077796936,
      "learning_rate": 0.0003937906112501206,
      "loss": 2.1023,
      "step": 91882
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.8158820867538452,
      "learning_rate": 0.00039378672572596,
      "loss": 2.1797,
      "step": 91883
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.8730787038803101,
      "learning_rate": 0.00039378284018436264,
      "loss": 2.1165,
      "step": 91884
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.8173992037773132,
      "learning_rate": 0.00039377895462532926,
      "loss": 2.1265,
      "step": 91885
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.8026074171066284,
      "learning_rate": 0.00039377506904886067,
      "loss": 1.9625,
      "step": 91886
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.7712098360061646,
      "learning_rate": 0.00039377118345495745,
      "loss": 2.0931,
      "step": 91887
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.7910305261611938,
      "learning_rate": 0.0003937672978436204,
      "loss": 2.1543,
      "step": 91888
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.8064157962799072,
      "learning_rate": 0.0003937634122148503,
      "loss": 2.0021,
      "step": 91889
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.7952155470848083,
      "learning_rate": 0.00039375952656864777,
      "loss": 2.0634,
      "step": 91890
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.8178537487983704,
      "learning_rate": 0.0003937556409050136,
      "loss": 2.0555,
      "step": 91891
    },
    {
      "epoch": 1.2,
      "grad_norm": 1.2010868787765503,
      "learning_rate": 0.00039375175522394856,
      "loss": 2.1561,
      "step": 91892
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.8647993803024292,
      "learning_rate": 0.00039374786952545317,
      "loss": 2.4204,
      "step": 91893
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.8638995885848999,
      "learning_rate": 0.0003937439838095283,
      "loss": 2.1494,
      "step": 91894
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.8547366857528687,
      "learning_rate": 0.0003937400980761747,
      "loss": 1.9972,
      "step": 91895
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.8245375156402588,
      "learning_rate": 0.000393736212325393,
      "loss": 2.1733,
      "step": 91896
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.8436304926872253,
      "learning_rate": 0.000393732326557184,
      "loss": 2.4007,
      "step": 91897
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.8864557147026062,
      "learning_rate": 0.00039372844077154836,
      "loss": 1.991,
      "step": 91898
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.8869200348854065,
      "learning_rate": 0.0003937245549684869,
      "loss": 2.039,
      "step": 91899
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.8027036786079407,
      "learning_rate": 0.00039372066914800027,
      "loss": 2.0278,
      "step": 91900
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.9773789048194885,
      "learning_rate": 0.00039371678331008906,
      "loss": 2.2918,
      "step": 91901
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.8686439990997314,
      "learning_rate": 0.0003937128974547543,
      "loss": 2.0393,
      "step": 91902
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.7335454225540161,
      "learning_rate": 0.00039370901158199644,
      "loss": 2.0722,
      "step": 91903
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.8809876441955566,
      "learning_rate": 0.0003937051256918164,
      "loss": 2.2809,
      "step": 91904
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.9665905833244324,
      "learning_rate": 0.00039370123978421466,
      "loss": 2.0962,
      "step": 91905
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.8226388096809387,
      "learning_rate": 0.00039369735385919225,
      "loss": 2.3147,
      "step": 91906
    },
    {
      "epoch": 1.2,
      "grad_norm": 1.002302646636963,
      "learning_rate": 0.0003936934679167496,
      "loss": 2.1601,
      "step": 91907
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.8658433556556702,
      "learning_rate": 0.00039368958195688765,
      "loss": 2.3944,
      "step": 91908
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.7780405282974243,
      "learning_rate": 0.000393685695979607,
      "loss": 2.3219,
      "step": 91909
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.8187298774719238,
      "learning_rate": 0.00039368180998490846,
      "loss": 2.2835,
      "step": 91910
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.897791862487793,
      "learning_rate": 0.0003936779239727927,
      "loss": 2.1508,
      "step": 91911
    },
    {
      "epoch": 1.2,
      "grad_norm": 1.1032143831253052,
      "learning_rate": 0.0003936740379432604,
      "loss": 2.1083,
      "step": 91912
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.8794866800308228,
      "learning_rate": 0.00039367015189631236,
      "loss": 2.1169,
      "step": 91913
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.9398777484893799,
      "learning_rate": 0.0003936662658319492,
      "loss": 2.1612,
      "step": 91914
    },
    {
      "epoch": 1.2,
      "grad_norm": 1.0015997886657715,
      "learning_rate": 0.00039366237975017176,
      "loss": 2.1359,
      "step": 91915
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.8184986710548401,
      "learning_rate": 0.00039365849365098073,
      "loss": 2.3977,
      "step": 91916
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.8122619986534119,
      "learning_rate": 0.0003936546075343769,
      "loss": 2.1579,
      "step": 91917
    },
    {
      "epoch": 1.2,
      "grad_norm": 1.120669960975647,
      "learning_rate": 0.0003936507214003608,
      "loss": 2.3084,
      "step": 91918
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.901985764503479,
      "learning_rate": 0.0003936468352489332,
      "loss": 2.2953,
      "step": 91919
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.8906448483467102,
      "learning_rate": 0.00039364294908009504,
      "loss": 2.2562,
      "step": 91920
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.8549010157585144,
      "learning_rate": 0.00039363906289384686,
      "loss": 2.121,
      "step": 91921
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.9039953351020813,
      "learning_rate": 0.0003936351766901893,
      "loss": 2.4747,
      "step": 91922
    },
    {
      "epoch": 1.2,
      "grad_norm": 1.0379146337509155,
      "learning_rate": 0.0003936312904691233,
      "loss": 2.0899,
      "step": 91923
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.8670070171356201,
      "learning_rate": 0.00039362740423064946,
      "loss": 2.3679,
      "step": 91924
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.9724870324134827,
      "learning_rate": 0.0003936235179747685,
      "loss": 2.2228,
      "step": 91925
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.8446558713912964,
      "learning_rate": 0.00039361963170148116,
      "loss": 2.229,
      "step": 91926
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.9340135455131531,
      "learning_rate": 0.00039361574541078815,
      "loss": 2.1463,
      "step": 91927
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.8482652902603149,
      "learning_rate": 0.0003936118591026902,
      "loss": 2.1774,
      "step": 91928
    },
    {
      "epoch": 1.2,
      "grad_norm": 1.006574273109436,
      "learning_rate": 0.0003936079727771881,
      "loss": 2.206,
      "step": 91929
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.9611555933952332,
      "learning_rate": 0.00039360408643428246,
      "loss": 2.3179,
      "step": 91930
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.7794974446296692,
      "learning_rate": 0.000393600200073974,
      "loss": 2.3017,
      "step": 91931
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.836940586566925,
      "learning_rate": 0.00039359631369626364,
      "loss": 2.3091,
      "step": 91932
    },
    {
      "epoch": 1.2,
      "grad_norm": 1.1277779340744019,
      "learning_rate": 0.0003935924273011519,
      "loss": 2.1138,
      "step": 91933
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.9058656692504883,
      "learning_rate": 0.00039358854088863943,
      "loss": 2.1847,
      "step": 91934
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.8542189598083496,
      "learning_rate": 0.0003935846544587273,
      "loss": 2.3007,
      "step": 91935
    },
    {
      "epoch": 1.2,
      "grad_norm": 1.041116714477539,
      "learning_rate": 0.00039358076801141584,
      "loss": 2.0555,
      "step": 91936
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.985372006893158,
      "learning_rate": 0.000393576881546706,
      "loss": 2.1424,
      "step": 91937
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.9180707931518555,
      "learning_rate": 0.0003935729950645985,
      "loss": 2.3048,
      "step": 91938
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.9783005118370056,
      "learning_rate": 0.00039356910856509395,
      "loss": 2.1545,
      "step": 91939
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.9201836585998535,
      "learning_rate": 0.00039356522204819316,
      "loss": 2.2783,
      "step": 91940
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.8869601488113403,
      "learning_rate": 0.00039356133551389684,
      "loss": 2.1695,
      "step": 91941
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.8347461223602295,
      "learning_rate": 0.00039355744896220574,
      "loss": 2.2534,
      "step": 91942
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.8719432950019836,
      "learning_rate": 0.00039355356239312045,
      "loss": 2.113,
      "step": 91943
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.9708431959152222,
      "learning_rate": 0.0003935496758066419,
      "loss": 2.3494,
      "step": 91944
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.8057951331138611,
      "learning_rate": 0.0003935457892027706,
      "loss": 2.4328,
      "step": 91945
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.9303943514823914,
      "learning_rate": 0.0003935419025815074,
      "loss": 2.1199,
      "step": 91946
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.9040839672088623,
      "learning_rate": 0.0003935380159428531,
      "loss": 2.3641,
      "step": 91947
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.8640149235725403,
      "learning_rate": 0.0003935341292868082,
      "loss": 2.1376,
      "step": 91948
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.8753683567047119,
      "learning_rate": 0.0003935302426133735,
      "loss": 2.4362,
      "step": 91949
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.8741356730461121,
      "learning_rate": 0.00039352635592254984,
      "loss": 2.105,
      "step": 91950
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.9132145643234253,
      "learning_rate": 0.00039352246921433784,
      "loss": 2.3465,
      "step": 91951
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.8230871558189392,
      "learning_rate": 0.00039351858248873827,
      "loss": 2.1538,
      "step": 91952
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.8167902827262878,
      "learning_rate": 0.00039351469574575184,
      "loss": 1.8963,
      "step": 91953
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.797372579574585,
      "learning_rate": 0.0003935108089853792,
      "loss": 1.9752,
      "step": 91954
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.9938725233078003,
      "learning_rate": 0.00039350692220762115,
      "loss": 2.3179,
      "step": 91955
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.8345511555671692,
      "learning_rate": 0.0003935030354124785,
      "loss": 2.2263,
      "step": 91956
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.8020298480987549,
      "learning_rate": 0.0003934991485999518,
      "loss": 2.1692,
      "step": 91957
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.935052752494812,
      "learning_rate": 0.0003934952617700418,
      "loss": 1.9919,
      "step": 91958
    },
    {
      "epoch": 1.2,
      "grad_norm": 1.1118453741073608,
      "learning_rate": 0.00039349137492274936,
      "loss": 2.2576,
      "step": 91959
    },
    {
      "epoch": 1.2,
      "grad_norm": 1.1458747386932373,
      "learning_rate": 0.000393487488058075,
      "loss": 2.2455,
      "step": 91960
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.9187663197517395,
      "learning_rate": 0.0003934836011760196,
      "loss": 1.9546,
      "step": 91961
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.8667886257171631,
      "learning_rate": 0.00039347971427658386,
      "loss": 2.1376,
      "step": 91962
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.8998621106147766,
      "learning_rate": 0.00039347582735976843,
      "loss": 2.0889,
      "step": 91963
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.8410664796829224,
      "learning_rate": 0.00039347194042557406,
      "loss": 2.1797,
      "step": 91964
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.8274677991867065,
      "learning_rate": 0.0003934680534740016,
      "loss": 2.1391,
      "step": 91965
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.8871549963951111,
      "learning_rate": 0.00039346416650505166,
      "loss": 2.4234,
      "step": 91966
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.808489203453064,
      "learning_rate": 0.00039346027951872483,
      "loss": 2.0821,
      "step": 91967
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.7828064560890198,
      "learning_rate": 0.00039345639251502215,
      "loss": 2.2328,
      "step": 91968
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.8793902397155762,
      "learning_rate": 0.000393452505493944,
      "loss": 2.1761,
      "step": 91969
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.7731233835220337,
      "learning_rate": 0.00039344861845549126,
      "loss": 2.1931,
      "step": 91970
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.8140872120857239,
      "learning_rate": 0.0003934447313996648,
      "loss": 2.0839,
      "step": 91971
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.8287685513496399,
      "learning_rate": 0.0003934408443264651,
      "loss": 2.1598,
      "step": 91972
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.8006244897842407,
      "learning_rate": 0.000393436957235893,
      "loss": 2.332,
      "step": 91973
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.8508228063583374,
      "learning_rate": 0.0003934330701279493,
      "loss": 2.2002,
      "step": 91974
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.8254305124282837,
      "learning_rate": 0.00039342918300263447,
      "loss": 1.9859,
      "step": 91975
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.8312649726867676,
      "learning_rate": 0.0003934252958599494,
      "loss": 1.9923,
      "step": 91976
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.823327362537384,
      "learning_rate": 0.0003934214086998949,
      "loss": 1.9359,
      "step": 91977
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.8321346640586853,
      "learning_rate": 0.0003934175215224716,
      "loss": 2.0135,
      "step": 91978
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.9256381392478943,
      "learning_rate": 0.0003934136343276801,
      "loss": 2.1946,
      "step": 91979
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.878921627998352,
      "learning_rate": 0.00039340974711552147,
      "loss": 2.0582,
      "step": 91980
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.8269830346107483,
      "learning_rate": 0.000393405859885996,
      "loss": 2.2095,
      "step": 91981
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.872961699962616,
      "learning_rate": 0.00039340197263910466,
      "loss": 2.0589,
      "step": 91982
    },
    {
      "epoch": 1.2,
      "grad_norm": 1.012506127357483,
      "learning_rate": 0.00039339808537484813,
      "loss": 2.3561,
      "step": 91983
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.8556932210922241,
      "learning_rate": 0.0003933941980932272,
      "loss": 2.3048,
      "step": 91984
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.8319684267044067,
      "learning_rate": 0.0003933903107942424,
      "loss": 2.2467,
      "step": 91985
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.8562904000282288,
      "learning_rate": 0.00039338642347789464,
      "loss": 2.048,
      "step": 91986
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.7982422709465027,
      "learning_rate": 0.00039338253614418467,
      "loss": 2.1797,
      "step": 91987
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.7829063534736633,
      "learning_rate": 0.00039337864879311304,
      "loss": 2.2051,
      "step": 91988
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.791021466255188,
      "learning_rate": 0.0003933747614246805,
      "loss": 2.1541,
      "step": 91989
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.8954469561576843,
      "learning_rate": 0.00039337087403888785,
      "loss": 2.2966,
      "step": 91990
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.7998474836349487,
      "learning_rate": 0.00039336698663573587,
      "loss": 1.9814,
      "step": 91991
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.8688755631446838,
      "learning_rate": 0.0003933630992152252,
      "loss": 2.1292,
      "step": 91992
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.8469921350479126,
      "learning_rate": 0.00039335921177735653,
      "loss": 2.3009,
      "step": 91993
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.7829398512840271,
      "learning_rate": 0.00039335532432213064,
      "loss": 2.0757,
      "step": 91994
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.8485740423202515,
      "learning_rate": 0.00039335143684954824,
      "loss": 2.2736,
      "step": 91995
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.8116173148155212,
      "learning_rate": 0.00039334754935961,
      "loss": 2.0894,
      "step": 91996
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.9214678406715393,
      "learning_rate": 0.00039334366185231665,
      "loss": 2.1735,
      "step": 91997
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.8576850891113281,
      "learning_rate": 0.00039333977432766913,
      "loss": 2.14,
      "step": 91998
    },
    {
      "epoch": 1.2,
      "grad_norm": 1.246201515197754,
      "learning_rate": 0.0003933358867856678,
      "loss": 2.1472,
      "step": 91999
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.8528873324394226,
      "learning_rate": 0.00039333199922631364,
      "loss": 2.1236,
      "step": 92000
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.8789364695549011,
      "learning_rate": 0.0003933281116496073,
      "loss": 2.2381,
      "step": 92001
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.8802611231803894,
      "learning_rate": 0.0003933242240555496,
      "loss": 2.1381,
      "step": 92002
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.8600547909736633,
      "learning_rate": 0.00039332033644414095,
      "loss": 2.0541,
      "step": 92003
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.7816699147224426,
      "learning_rate": 0.0003933164488153824,
      "loss": 1.9534,
      "step": 92004
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.9655657410621643,
      "learning_rate": 0.0003933125611692746,
      "loss": 2.0944,
      "step": 92005
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.960199773311615,
      "learning_rate": 0.0003933086735058182,
      "loss": 2.1829,
      "step": 92006
    },
    {
      "epoch": 1.2,
      "grad_norm": 1.02432382106781,
      "learning_rate": 0.0003933047858250139,
      "loss": 2.2617,
      "step": 92007
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.8761254549026489,
      "learning_rate": 0.00039330089812686256,
      "loss": 2.0779,
      "step": 92008
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.8923717737197876,
      "learning_rate": 0.0003932970104113648,
      "loss": 2.2163,
      "step": 92009
    },
    {
      "epoch": 1.2,
      "grad_norm": 1.2473535537719727,
      "learning_rate": 0.00039329312267852133,
      "loss": 2.1173,
      "step": 92010
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.9065300226211548,
      "learning_rate": 0.000393289234928333,
      "loss": 2.1881,
      "step": 92011
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.7868112325668335,
      "learning_rate": 0.0003932853471608004,
      "loss": 2.2496,
      "step": 92012
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.7967430353164673,
      "learning_rate": 0.00039328145937592417,
      "loss": 2.241,
      "step": 92013
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.9383425712585449,
      "learning_rate": 0.0003932775715737053,
      "loss": 1.7485,
      "step": 92014
    },
    {
      "epoch": 1.2,
      "grad_norm": 1.0397111177444458,
      "learning_rate": 0.00039327368375414437,
      "loss": 2.2519,
      "step": 92015
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.8422236442565918,
      "learning_rate": 0.000393269795917242,
      "loss": 2.2688,
      "step": 92016
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.843311607837677,
      "learning_rate": 0.0003932659080629991,
      "loss": 2.0079,
      "step": 92017
    },
    {
      "epoch": 1.2,
      "grad_norm": 1.0006498098373413,
      "learning_rate": 0.00039326202019141626,
      "loss": 2.1357,
      "step": 92018
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.9499483108520508,
      "learning_rate": 0.0003932581323024942,
      "loss": 2.3927,
      "step": 92019
    },
    {
      "epoch": 1.2,
      "grad_norm": 1.0709866285324097,
      "learning_rate": 0.00039325424439623383,
      "loss": 2.0561,
      "step": 92020
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.8843681812286377,
      "learning_rate": 0.0003932503564726357,
      "loss": 2.1957,
      "step": 92021
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.8723013401031494,
      "learning_rate": 0.0003932464685317005,
      "loss": 2.1414,
      "step": 92022
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.8360607624053955,
      "learning_rate": 0.00039324258057342904,
      "loss": 1.8849,
      "step": 92023
    },
    {
      "epoch": 1.2,
      "grad_norm": 1.0696567296981812,
      "learning_rate": 0.00039323869259782205,
      "loss": 2.2471,
      "step": 92024
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.8864482045173645,
      "learning_rate": 0.00039323480460488024,
      "loss": 1.8831,
      "step": 92025
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.8967605829238892,
      "learning_rate": 0.0003932309165946043,
      "loss": 2.3143,
      "step": 92026
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.7928686738014221,
      "learning_rate": 0.0003932270285669949,
      "loss": 2.3712,
      "step": 92027
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.8516708016395569,
      "learning_rate": 0.0003932231405220529,
      "loss": 2.0485,
      "step": 92028
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.8391264081001282,
      "learning_rate": 0.00039321925245977904,
      "loss": 2.1533,
      "step": 92029
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.8658486604690552,
      "learning_rate": 0.0003932153643801738,
      "loss": 2.2716,
      "step": 92030
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.8137093186378479,
      "learning_rate": 0.0003932114762832382,
      "loss": 2.1628,
      "step": 92031
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.9449018836021423,
      "learning_rate": 0.0003932075881689728,
      "loss": 2.265,
      "step": 92032
    },
    {
      "epoch": 1.2,
      "grad_norm": 1.0140116214752197,
      "learning_rate": 0.00039320370003737833,
      "loss": 2.1877,
      "step": 92033
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.87189120054245,
      "learning_rate": 0.0003931998118884555,
      "loss": 2.3814,
      "step": 92034
    },
    {
      "epoch": 1.2,
      "grad_norm": 1.090353012084961,
      "learning_rate": 0.0003931959237222051,
      "loss": 2.2788,
      "step": 92035
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.9646457433700562,
      "learning_rate": 0.0003931920355386278,
      "loss": 2.3332,
      "step": 92036
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.8556869029998779,
      "learning_rate": 0.00039318814733772437,
      "loss": 1.8404,
      "step": 92037
    },
    {
      "epoch": 1.2,
      "grad_norm": 1.0575605630874634,
      "learning_rate": 0.00039318425911949554,
      "loss": 2.1645,
      "step": 92038
    },
    {
      "epoch": 1.2,
      "grad_norm": 1.0405617952346802,
      "learning_rate": 0.0003931803708839419,
      "loss": 2.2536,
      "step": 92039
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.8833212852478027,
      "learning_rate": 0.0003931764826310643,
      "loss": 2.2305,
      "step": 92040
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.8074382543563843,
      "learning_rate": 0.0003931725943608635,
      "loss": 2.0516,
      "step": 92041
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.8757803440093994,
      "learning_rate": 0.0003931687060733401,
      "loss": 2.2978,
      "step": 92042
    },
    {
      "epoch": 1.2,
      "grad_norm": 1.163044810295105,
      "learning_rate": 0.00039316481776849484,
      "loss": 2.2968,
      "step": 92043
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.9357290267944336,
      "learning_rate": 0.00039316092944632853,
      "loss": 2.0708,
      "step": 92044
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.7957725524902344,
      "learning_rate": 0.0003931570411068419,
      "loss": 2.335,
      "step": 92045
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.8076647520065308,
      "learning_rate": 0.0003931531527500355,
      "loss": 2.1916,
      "step": 92046
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.8278781771659851,
      "learning_rate": 0.00039314926437591024,
      "loss": 1.9919,
      "step": 92047
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.8276902437210083,
      "learning_rate": 0.0003931453759844668,
      "loss": 2.2846,
      "step": 92048
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.803226888179779,
      "learning_rate": 0.0003931414875757058,
      "loss": 2.1077,
      "step": 92049
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.8668732047080994,
      "learning_rate": 0.0003931375991496281,
      "loss": 2.1425,
      "step": 92050
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.8415587544441223,
      "learning_rate": 0.0003931337107062343,
      "loss": 2.1796,
      "step": 92051
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.8182016611099243,
      "learning_rate": 0.00039312982224552523,
      "loss": 2.3189,
      "step": 92052
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.8355304598808289,
      "learning_rate": 0.0003931259337675015,
      "loss": 2.3175,
      "step": 92053
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.8326001167297363,
      "learning_rate": 0.0003931220452721641,
      "loss": 2.0555,
      "step": 92054
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.7923699617385864,
      "learning_rate": 0.00039311815675951333,
      "loss": 2.408,
      "step": 92055
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.8747590780258179,
      "learning_rate": 0.00039311426822955024,
      "loss": 2.1599,
      "step": 92056
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.790330708026886,
      "learning_rate": 0.00039311037968227543,
      "loss": 2.2784,
      "step": 92057
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.7918851375579834,
      "learning_rate": 0.00039310649111768965,
      "loss": 1.9548,
      "step": 92058
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.8363959193229675,
      "learning_rate": 0.00039310260253579366,
      "loss": 2.1943,
      "step": 92059
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.8015446662902832,
      "learning_rate": 0.00039309871393658805,
      "loss": 1.9011,
      "step": 92060
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.8221489191055298,
      "learning_rate": 0.0003930948253200737,
      "loss": 2.2791,
      "step": 92061
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.8119249939918518,
      "learning_rate": 0.0003930909366862512,
      "loss": 2.0006,
      "step": 92062
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.8409883379936218,
      "learning_rate": 0.0003930870480351214,
      "loss": 2.2591,
      "step": 92063
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.8534052968025208,
      "learning_rate": 0.0003930831593666849,
      "loss": 2.3711,
      "step": 92064
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.8050901889801025,
      "learning_rate": 0.00039307927068094257,
      "loss": 2.0392,
      "step": 92065
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.9001665711402893,
      "learning_rate": 0.000393075381977895,
      "loss": 1.9511,
      "step": 92066
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.9008851051330566,
      "learning_rate": 0.00039307149325754293,
      "loss": 2.1585,
      "step": 92067
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.7769064903259277,
      "learning_rate": 0.0003930676045198871,
      "loss": 2.0041,
      "step": 92068
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.8464775681495667,
      "learning_rate": 0.00039306371576492837,
      "loss": 2.1489,
      "step": 92069
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.8756234645843506,
      "learning_rate": 0.00039305982699266724,
      "loss": 2.269,
      "step": 92070
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.9597014784812927,
      "learning_rate": 0.00039305593820310446,
      "loss": 2.2669,
      "step": 92071
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.8142089247703552,
      "learning_rate": 0.000393052049396241,
      "loss": 2.178,
      "step": 92072
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.8659184575080872,
      "learning_rate": 0.0003930481605720773,
      "loss": 1.9921,
      "step": 92073
    },
    {
      "epoch": 1.2,
      "grad_norm": 1.0107911825180054,
      "learning_rate": 0.00039304427173061414,
      "loss": 2.2271,
      "step": 92074
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.7824833393096924,
      "learning_rate": 0.0003930403828718525,
      "loss": 2.0407,
      "step": 92075
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.8018669486045837,
      "learning_rate": 0.0003930364939957927,
      "loss": 2.1103,
      "step": 92076
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.9244962930679321,
      "learning_rate": 0.0003930326051024357,
      "loss": 1.972,
      "step": 92077
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.8888424038887024,
      "learning_rate": 0.00039302871619178217,
      "loss": 2.0662,
      "step": 92078
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.906731128692627,
      "learning_rate": 0.0003930248272638329,
      "loss": 2.2551,
      "step": 92079
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.8320916891098022,
      "learning_rate": 0.0003930209383185885,
      "loss": 2.0659,
      "step": 92080
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.9498911499977112,
      "learning_rate": 0.00039301704935604975,
      "loss": 2.2208,
      "step": 92081
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.9984198808670044,
      "learning_rate": 0.00039301316037621746,
      "loss": 2.2345,
      "step": 92082
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.9008825421333313,
      "learning_rate": 0.00039300927137909226,
      "loss": 2.2723,
      "step": 92083
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.8225730061531067,
      "learning_rate": 0.00039300538236467474,
      "loss": 2.1812,
      "step": 92084
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.966249942779541,
      "learning_rate": 0.00039300149333296587,
      "loss": 2.1506,
      "step": 92085
    },
    {
      "epoch": 1.2,
      "grad_norm": 1.0154991149902344,
      "learning_rate": 0.0003929976042839663,
      "loss": 2.1675,
      "step": 92086
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.785690188407898,
      "learning_rate": 0.0003929937152176767,
      "loss": 2.029,
      "step": 92087
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.8523579835891724,
      "learning_rate": 0.0003929898261340977,
      "loss": 1.9316,
      "step": 92088
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.7911614775657654,
      "learning_rate": 0.00039298593703323024,
      "loss": 2.0956,
      "step": 92089
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.8309239745140076,
      "learning_rate": 0.00039298204791507496,
      "loss": 2.2522,
      "step": 92090
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.7974122166633606,
      "learning_rate": 0.00039297815877963255,
      "loss": 2.172,
      "step": 92091
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.8808034658432007,
      "learning_rate": 0.0003929742696269037,
      "loss": 2.2323,
      "step": 92092
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.8627590537071228,
      "learning_rate": 0.0003929703804568892,
      "loss": 2.4958,
      "step": 92093
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.9130857586860657,
      "learning_rate": 0.00039296649126958975,
      "loss": 2.3068,
      "step": 92094
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.8110401630401611,
      "learning_rate": 0.00039296260206500604,
      "loss": 1.993,
      "step": 92095
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.8085811734199524,
      "learning_rate": 0.00039295871284313896,
      "loss": 2.1669,
      "step": 92096
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.9846055507659912,
      "learning_rate": 0.00039295482360398893,
      "loss": 2.2637,
      "step": 92097
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.9915322065353394,
      "learning_rate": 0.00039295093434755694,
      "loss": 2.2354,
      "step": 92098
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.8716519474983215,
      "learning_rate": 0.00039294704507384363,
      "loss": 2.0717,
      "step": 92099
    },
    {
      "epoch": 1.2,
      "grad_norm": 1.0780248641967773,
      "learning_rate": 0.0003929431557828497,
      "loss": 2.1943,
      "step": 92100
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.7702184319496155,
      "learning_rate": 0.0003929392664745758,
      "loss": 2.1036,
      "step": 92101
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.7924127578735352,
      "learning_rate": 0.00039293537714902286,
      "loss": 2.2148,
      "step": 92102
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.845281183719635,
      "learning_rate": 0.0003929314878061914,
      "loss": 2.2304,
      "step": 92103
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.7781499624252319,
      "learning_rate": 0.0003929275984460822,
      "loss": 2.1533,
      "step": 92104
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.990824282169342,
      "learning_rate": 0.00039292370906869606,
      "loss": 2.1202,
      "step": 92105
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.9077883362770081,
      "learning_rate": 0.00039291981967403365,
      "loss": 2.2037,
      "step": 92106
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.811435341835022,
      "learning_rate": 0.00039291593026209563,
      "loss": 2.1604,
      "step": 92107
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.9709530472755432,
      "learning_rate": 0.0003929120408328829,
      "loss": 1.9111,
      "step": 92108
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.8696619868278503,
      "learning_rate": 0.0003929081513863959,
      "loss": 2.2507,
      "step": 92109
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.7941406965255737,
      "learning_rate": 0.0003929042619226356,
      "loss": 2.0967,
      "step": 92110
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.865159809589386,
      "learning_rate": 0.00039290037244160274,
      "loss": 2.1509,
      "step": 92111
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.937786877155304,
      "learning_rate": 0.0003928964829432979,
      "loss": 2.1685,
      "step": 92112
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.9389442205429077,
      "learning_rate": 0.0003928925934277217,
      "loss": 2.1214,
      "step": 92113
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.9518669247627258,
      "learning_rate": 0.00039288870389487526,
      "loss": 2.3079,
      "step": 92114
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.8566224575042725,
      "learning_rate": 0.00039288481434475886,
      "loss": 1.9248,
      "step": 92115
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.7948272824287415,
      "learning_rate": 0.00039288092477737346,
      "loss": 2.3897,
      "step": 92116
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.8061774373054504,
      "learning_rate": 0.0003928770351927198,
      "loss": 2.2931,
      "step": 92117
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.8114239573478699,
      "learning_rate": 0.0003928731455907985,
      "loss": 2.2219,
      "step": 92118
    },
    {
      "epoch": 1.2,
      "grad_norm": 1.1525262594223022,
      "learning_rate": 0.00039286925597161034,
      "loss": 2.1063,
      "step": 92119
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.8335409760475159,
      "learning_rate": 0.00039286536633515595,
      "loss": 2.3792,
      "step": 92120
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.948920488357544,
      "learning_rate": 0.00039286147668143634,
      "loss": 2.1954,
      "step": 92121
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.7939452528953552,
      "learning_rate": 0.0003928575870104519,
      "loss": 1.8985,
      "step": 92122
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.9072569608688354,
      "learning_rate": 0.0003928536973222035,
      "loss": 2.2965,
      "step": 92123
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.8737580180168152,
      "learning_rate": 0.00039284980761669186,
      "loss": 2.0299,
      "step": 92124
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.9065803289413452,
      "learning_rate": 0.00039284591789391757,
      "loss": 2.0387,
      "step": 92125
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.7938209176063538,
      "learning_rate": 0.0003928420281538816,
      "loss": 2.2085,
      "step": 92126
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.8848769068717957,
      "learning_rate": 0.00039283813839658457,
      "loss": 2.0342,
      "step": 92127
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.911021888256073,
      "learning_rate": 0.00039283424862202705,
      "loss": 2.1758,
      "step": 92128
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.869864821434021,
      "learning_rate": 0.00039283035883020995,
      "loss": 2.233,
      "step": 92129
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.8004449009895325,
      "learning_rate": 0.000392826469021134,
      "loss": 2.2782,
      "step": 92130
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.8574962615966797,
      "learning_rate": 0.00039282257919479973,
      "loss": 2.12,
      "step": 92131
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.915793240070343,
      "learning_rate": 0.0003928186893512081,
      "loss": 2.1951,
      "step": 92132
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.8111717700958252,
      "learning_rate": 0.0003928147994903597,
      "loss": 1.9338,
      "step": 92133
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.8573805689811707,
      "learning_rate": 0.00039281090961225526,
      "loss": 2.1161,
      "step": 92134
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.8083704710006714,
      "learning_rate": 0.0003928070197168955,
      "loss": 2.1602,
      "step": 92135
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.8049250245094299,
      "learning_rate": 0.0003928031298042812,
      "loss": 2.1028,
      "step": 92136
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.8534253835678101,
      "learning_rate": 0.000392799239874413,
      "loss": 2.1411,
      "step": 92137
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.8218088150024414,
      "learning_rate": 0.0003927953499272917,
      "loss": 2.1389,
      "step": 92138
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.8200560212135315,
      "learning_rate": 0.00039279145996291804,
      "loss": 2.1236,
      "step": 92139
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.9255539774894714,
      "learning_rate": 0.0003927875699812926,
      "loss": 2.1457,
      "step": 92140
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.881571888923645,
      "learning_rate": 0.00039278367998241625,
      "loss": 2.2287,
      "step": 92141
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.8420958518981934,
      "learning_rate": 0.0003927797899662897,
      "loss": 2.3583,
      "step": 92142
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.834862232208252,
      "learning_rate": 0.0003927758999329136,
      "loss": 1.8317,
      "step": 92143
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.8678163290023804,
      "learning_rate": 0.00039277200988228867,
      "loss": 2.1258,
      "step": 92144
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.7771912813186646,
      "learning_rate": 0.00039276811981441576,
      "loss": 2.1752,
      "step": 92145
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.8911210894584656,
      "learning_rate": 0.0003927642297292954,
      "loss": 2.2823,
      "step": 92146
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.7974682450294495,
      "learning_rate": 0.00039276033962692844,
      "loss": 2.2197,
      "step": 92147
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.8881458640098572,
      "learning_rate": 0.00039275644950731567,
      "loss": 2.024,
      "step": 92148
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.817514955997467,
      "learning_rate": 0.0003927525593704576,
      "loss": 1.9596,
      "step": 92149
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.8466382026672363,
      "learning_rate": 0.0003927486692163552,
      "loss": 2.1374,
      "step": 92150
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.8620581030845642,
      "learning_rate": 0.00039274477904500905,
      "loss": 2.0843,
      "step": 92151
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.8478686213493347,
      "learning_rate": 0.0003927408888564199,
      "loss": 2.19,
      "step": 92152
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.8648035526275635,
      "learning_rate": 0.00039273699865058836,
      "loss": 2.2071,
      "step": 92153
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.9288891553878784,
      "learning_rate": 0.0003927331084275154,
      "loss": 2.0888,
      "step": 92154
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.9212361574172974,
      "learning_rate": 0.0003927292181872015,
      "loss": 2.2393,
      "step": 92155
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.8506578207015991,
      "learning_rate": 0.0003927253279296475,
      "loss": 2.3219,
      "step": 92156
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.8989076018333435,
      "learning_rate": 0.00039272143765485415,
      "loss": 2.3377,
      "step": 92157
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.8794626593589783,
      "learning_rate": 0.0003927175473628221,
      "loss": 2.2981,
      "step": 92158
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.8593240976333618,
      "learning_rate": 0.00039271365705355213,
      "loss": 2.2535,
      "step": 92159
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.9744725823402405,
      "learning_rate": 0.000392709766727045,
      "loss": 1.9292,
      "step": 92160
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.7664142847061157,
      "learning_rate": 0.0003927058763833013,
      "loss": 2.2867,
      "step": 92161
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.8206644058227539,
      "learning_rate": 0.0003927019860223218,
      "loss": 2.3907,
      "step": 92162
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.7433715462684631,
      "learning_rate": 0.0003926980956441073,
      "loss": 2.1935,
      "step": 92163
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.9476458430290222,
      "learning_rate": 0.00039269420524865853,
      "loss": 1.947,
      "step": 92164
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.8821390271186829,
      "learning_rate": 0.00039269031483597604,
      "loss": 2.2918,
      "step": 92165
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.899334728717804,
      "learning_rate": 0.0003926864244060607,
      "loss": 2.2946,
      "step": 92166
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.8427270650863647,
      "learning_rate": 0.00039268253395891326,
      "loss": 2.4164,
      "step": 92167
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.8383402824401855,
      "learning_rate": 0.00039267864349453433,
      "loss": 2.4154,
      "step": 92168
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.9006280899047852,
      "learning_rate": 0.0003926747530129247,
      "loss": 2.2962,
      "step": 92169
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.8763596415519714,
      "learning_rate": 0.00039267086251408513,
      "loss": 2.1833,
      "step": 92170
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.9206781387329102,
      "learning_rate": 0.00039266697199801627,
      "loss": 2.0963,
      "step": 92171
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.8673955798149109,
      "learning_rate": 0.0003926630814647189,
      "loss": 2.0821,
      "step": 92172
    },
    {
      "epoch": 1.2,
      "grad_norm": 1.1065034866333008,
      "learning_rate": 0.00039265919091419364,
      "loss": 2.1737,
      "step": 92173
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.9325170516967773,
      "learning_rate": 0.0003926553003464413,
      "loss": 2.2126,
      "step": 92174
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.8829278945922852,
      "learning_rate": 0.00039265140976146265,
      "loss": 2.3423,
      "step": 92175
    },
    {
      "epoch": 1.2,
      "grad_norm": 1.092092514038086,
      "learning_rate": 0.0003926475191592583,
      "loss": 2.2862,
      "step": 92176
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.772648274898529,
      "learning_rate": 0.00039264362853982904,
      "loss": 2.3564,
      "step": 92177
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.8889312744140625,
      "learning_rate": 0.00039263973790317565,
      "loss": 1.9592,
      "step": 92178
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.865458071231842,
      "learning_rate": 0.0003926358472492987,
      "loss": 1.8664,
      "step": 92179
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.9276660084724426,
      "learning_rate": 0.00039263195657819903,
      "loss": 2.3446,
      "step": 92180
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.9022243022918701,
      "learning_rate": 0.0003926280658898774,
      "loss": 2.1626,
      "step": 92181
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.8678380846977234,
      "learning_rate": 0.00039262417518433427,
      "loss": 2.293,
      "step": 92182
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.9381794929504395,
      "learning_rate": 0.0003926202844615707,
      "loss": 2.1065,
      "step": 92183
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.8515107035636902,
      "learning_rate": 0.0003926163937215872,
      "loss": 2.3256,
      "step": 92184
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.9051094055175781,
      "learning_rate": 0.00039261250296438463,
      "loss": 2.4321,
      "step": 92185
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.8599492907524109,
      "learning_rate": 0.00039260861218996367,
      "loss": 2.171,
      "step": 92186
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.7830777168273926,
      "learning_rate": 0.000392604721398325,
      "loss": 1.9498,
      "step": 92187
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.9315332174301147,
      "learning_rate": 0.0003926008305894694,
      "loss": 2.2512,
      "step": 92188
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.8513717651367188,
      "learning_rate": 0.0003925969397633974,
      "loss": 2.2097,
      "step": 92189
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.8961988091468811,
      "learning_rate": 0.00039259304892011,
      "loss": 2.3306,
      "step": 92190
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.9351962804794312,
      "learning_rate": 0.0003925891580596078,
      "loss": 2.5387,
      "step": 92191
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.9997246861457825,
      "learning_rate": 0.00039258526718189156,
      "loss": 2.0814,
      "step": 92192
    },
    {
      "epoch": 1.2,
      "grad_norm": 1.0029348134994507,
      "learning_rate": 0.0003925813762869619,
      "loss": 2.0225,
      "step": 92193
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.8235092163085938,
      "learning_rate": 0.0003925774853748197,
      "loss": 2.0116,
      "step": 92194
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.9525555372238159,
      "learning_rate": 0.0003925735944454655,
      "loss": 2.1725,
      "step": 92195
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.9478490352630615,
      "learning_rate": 0.0003925697034989001,
      "loss": 2.2061,
      "step": 92196
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.893424391746521,
      "learning_rate": 0.0003925658125351243,
      "loss": 2.3982,
      "step": 92197
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.8116775751113892,
      "learning_rate": 0.00039256192155413876,
      "loss": 2.1444,
      "step": 92198
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.9335798025131226,
      "learning_rate": 0.00039255803055594427,
      "loss": 2.2468,
      "step": 92199
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.7947080731391907,
      "learning_rate": 0.00039255413954054154,
      "loss": 2.1727,
      "step": 92200
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.899858295917511,
      "learning_rate": 0.00039255024850793106,
      "loss": 2.2207,
      "step": 92201
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.9519472122192383,
      "learning_rate": 0.00039254635745811385,
      "loss": 2.378,
      "step": 92202
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.8863983750343323,
      "learning_rate": 0.0003925424663910905,
      "loss": 2.008,
      "step": 92203
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.8324970602989197,
      "learning_rate": 0.0003925385753068618,
      "loss": 2.3196,
      "step": 92204
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.866607666015625,
      "learning_rate": 0.00039253468420542836,
      "loss": 2.1954,
      "step": 92205
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.9291989803314209,
      "learning_rate": 0.00039253079308679105,
      "loss": 2.2779,
      "step": 92206
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.8980293869972229,
      "learning_rate": 0.00039252690195095053,
      "loss": 2.1625,
      "step": 92207
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.8956687450408936,
      "learning_rate": 0.0003925230107979074,
      "loss": 2.1121,
      "step": 92208
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.8380481600761414,
      "learning_rate": 0.0003925191196276626,
      "loss": 2.1828,
      "step": 92209
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.8645245432853699,
      "learning_rate": 0.00039251522844021674,
      "loss": 1.8849,
      "step": 92210
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.8371585011482239,
      "learning_rate": 0.00039251133723557053,
      "loss": 2.1303,
      "step": 92211
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.8244235515594482,
      "learning_rate": 0.0003925074460137247,
      "loss": 2.1619,
      "step": 92212
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.8094143867492676,
      "learning_rate": 0.00039250355477467996,
      "loss": 2.1173,
      "step": 92213
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.8242261409759521,
      "learning_rate": 0.00039249966351843717,
      "loss": 2.2759,
      "step": 92214
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.8066189289093018,
      "learning_rate": 0.0003924957722449969,
      "loss": 2.1513,
      "step": 92215
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.8206921815872192,
      "learning_rate": 0.0003924918809543599,
      "loss": 2.259,
      "step": 92216
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.8606070876121521,
      "learning_rate": 0.0003924879896465269,
      "loss": 2.1413,
      "step": 92217
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.7887663245201111,
      "learning_rate": 0.0003924840983214987,
      "loss": 2.2685,
      "step": 92218
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.9960910677909851,
      "learning_rate": 0.0003924802069792759,
      "loss": 2.084,
      "step": 92219
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.7892648577690125,
      "learning_rate": 0.00039247631561985924,
      "loss": 2.0252,
      "step": 92220
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.8481630086898804,
      "learning_rate": 0.00039247242424324957,
      "loss": 2.0086,
      "step": 92221
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.832780659198761,
      "learning_rate": 0.00039246853284944755,
      "loss": 2.307,
      "step": 92222
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.8846940398216248,
      "learning_rate": 0.0003924646414384538,
      "loss": 2.2877,
      "step": 92223
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.7599503397941589,
      "learning_rate": 0.0003924607500102692,
      "loss": 1.8675,
      "step": 92224
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.8200922608375549,
      "learning_rate": 0.0003924568585648944,
      "loss": 2.0871,
      "step": 92225
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.8652411699295044,
      "learning_rate": 0.0003924529671023301,
      "loss": 2.245,
      "step": 92226
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.939389705657959,
      "learning_rate": 0.00039244907562257706,
      "loss": 2.0812,
      "step": 92227
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.9091742634773254,
      "learning_rate": 0.00039244518412563596,
      "loss": 1.9644,
      "step": 92228
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.8406777381896973,
      "learning_rate": 0.00039244129261150755,
      "loss": 2.1192,
      "step": 92229
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.7671658992767334,
      "learning_rate": 0.00039243740108019264,
      "loss": 2.4033,
      "step": 92230
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.9711133241653442,
      "learning_rate": 0.00039243350953169177,
      "loss": 2.37,
      "step": 92231
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.8908398151397705,
      "learning_rate": 0.00039242961796600575,
      "loss": 2.2008,
      "step": 92232
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.7848413586616516,
      "learning_rate": 0.00039242572638313546,
      "loss": 2.2569,
      "step": 92233
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.8544573783874512,
      "learning_rate": 0.00039242183478308137,
      "loss": 1.9449,
      "step": 92234
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.8885983824729919,
      "learning_rate": 0.0003924179431658443,
      "loss": 2.3328,
      "step": 92235
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.806177020072937,
      "learning_rate": 0.00039241405153142507,
      "loss": 2.1961,
      "step": 92236
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.836359441280365,
      "learning_rate": 0.00039241015987982427,
      "loss": 2.3852,
      "step": 92237
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.8036307096481323,
      "learning_rate": 0.00039240626821104266,
      "loss": 2.1602,
      "step": 92238
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.8358331918716431,
      "learning_rate": 0.0003924023765250811,
      "loss": 2.2176,
      "step": 92239
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.8235830664634705,
      "learning_rate": 0.0003923984848219401,
      "loss": 2.3036,
      "step": 92240
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.7390557527542114,
      "learning_rate": 0.0003923945931016205,
      "loss": 1.9726,
      "step": 92241
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.8335355520248413,
      "learning_rate": 0.000392390701364123,
      "loss": 2.2226,
      "step": 92242
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.92336106300354,
      "learning_rate": 0.0003923868096094482,
      "loss": 2.2283,
      "step": 92243
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.7982982397079468,
      "learning_rate": 0.00039238291783759714,
      "loss": 2.0312,
      "step": 92244
    },
    {
      "epoch": 1.2,
      "grad_norm": 1.064683437347412,
      "learning_rate": 0.0003923790260485703,
      "loss": 2.4356,
      "step": 92245
    },
    {
      "epoch": 1.2,
      "grad_norm": 1.003724455833435,
      "learning_rate": 0.00039237513424236845,
      "loss": 2.1499,
      "step": 92246
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.9679692983627319,
      "learning_rate": 0.0003923712424189922,
      "loss": 2.2378,
      "step": 92247
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.8480722308158875,
      "learning_rate": 0.0003923673505784425,
      "loss": 2.111,
      "step": 92248
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.8396137952804565,
      "learning_rate": 0.00039236345872072,
      "loss": 2.081,
      "step": 92249
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.9631035923957825,
      "learning_rate": 0.0003923595668458253,
      "loss": 2.2376,
      "step": 92250
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.8575108051300049,
      "learning_rate": 0.0003923556749537593,
      "loss": 2.2221,
      "step": 92251
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.8435726761817932,
      "learning_rate": 0.0003923517830445226,
      "loss": 2.4053,
      "step": 92252
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.9165648818016052,
      "learning_rate": 0.0003923478911181159,
      "loss": 2.1967,
      "step": 92253
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.8833425045013428,
      "learning_rate": 0.00039234399917454006,
      "loss": 2.21,
      "step": 92254
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.7992122173309326,
      "learning_rate": 0.0003923401072137957,
      "loss": 2.2591,
      "step": 92255
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.8088119626045227,
      "learning_rate": 0.0003923362152358836,
      "loss": 2.2091,
      "step": 92256
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.9070948362350464,
      "learning_rate": 0.0003923323232408044,
      "loss": 2.5197,
      "step": 92257
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.7953592538833618,
      "learning_rate": 0.0003923284312285589,
      "loss": 1.9631,
      "step": 92258
    },
    {
      "epoch": 1.2,
      "grad_norm": 1.1597449779510498,
      "learning_rate": 0.0003923245391991479,
      "loss": 1.8269,
      "step": 92259
    },
    {
      "epoch": 1.2,
      "grad_norm": 1.100222110748291,
      "learning_rate": 0.00039232064715257186,
      "loss": 1.87,
      "step": 92260
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.9373072385787964,
      "learning_rate": 0.00039231675508883177,
      "loss": 2.0041,
      "step": 92261
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.8329758048057556,
      "learning_rate": 0.0003923128630079282,
      "loss": 2.2171,
      "step": 92262
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.8094652891159058,
      "learning_rate": 0.00039230897090986196,
      "loss": 2.0504,
      "step": 92263
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.8175979852676392,
      "learning_rate": 0.0003923050787946337,
      "loss": 2.1847,
      "step": 92264
    },
    {
      "epoch": 1.2,
      "grad_norm": 1.08579421043396,
      "learning_rate": 0.00039230118666224427,
      "loss": 2.0951,
      "step": 92265
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.9117980599403381,
      "learning_rate": 0.0003922972945126943,
      "loss": 2.1582,
      "step": 92266
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.8090262413024902,
      "learning_rate": 0.00039229340234598443,
      "loss": 2.0265,
      "step": 92267
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.9378347396850586,
      "learning_rate": 0.0003922895101621155,
      "loss": 2.176,
      "step": 92268
    },
    {
      "epoch": 1.2,
      "grad_norm": 1.059808611869812,
      "learning_rate": 0.0003922856179610882,
      "loss": 2.3081,
      "step": 92269
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.9633922576904297,
      "learning_rate": 0.0003922817257429033,
      "loss": 2.2639,
      "step": 92270
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.8193164467811584,
      "learning_rate": 0.0003922778335075615,
      "loss": 2.3756,
      "step": 92271
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.9564948678016663,
      "learning_rate": 0.00039227394125506353,
      "loss": 2.3515,
      "step": 92272
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.93320631980896,
      "learning_rate": 0.00039227004898541,
      "loss": 2.3558,
      "step": 92273
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.9856606125831604,
      "learning_rate": 0.0003922661566986018,
      "loss": 2.1497,
      "step": 92274
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.8733406066894531,
      "learning_rate": 0.00039226226439463953,
      "loss": 2.3493,
      "step": 92275
    },
    {
      "epoch": 1.2,
      "grad_norm": 1.0904021263122559,
      "learning_rate": 0.00039225837207352405,
      "loss": 2.2717,
      "step": 92276
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.9670783877372742,
      "learning_rate": 0.0003922544797352559,
      "loss": 2.3593,
      "step": 92277
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.8620144128799438,
      "learning_rate": 0.0003922505873798359,
      "loss": 2.3328,
      "step": 92278
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.8659243583679199,
      "learning_rate": 0.00039224669500726486,
      "loss": 2.206,
      "step": 92279
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.8605863451957703,
      "learning_rate": 0.0003922428026175433,
      "loss": 2.1329,
      "step": 92280
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.8318843245506287,
      "learning_rate": 0.00039223891021067213,
      "loss": 2.1891,
      "step": 92281
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.8063672780990601,
      "learning_rate": 0.00039223501778665204,
      "loss": 2.3326,
      "step": 92282
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.8573846817016602,
      "learning_rate": 0.00039223112534548366,
      "loss": 2.103,
      "step": 92283
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.9089944362640381,
      "learning_rate": 0.0003922272328871678,
      "loss": 2.5667,
      "step": 92284
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.840204119682312,
      "learning_rate": 0.0003922233404117052,
      "loss": 2.0875,
      "step": 92285
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.8055837154388428,
      "learning_rate": 0.0003922194479190965,
      "loss": 1.8636,
      "step": 92286
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.800041913986206,
      "learning_rate": 0.00039221555540934244,
      "loss": 2.1339,
      "step": 92287
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.8906679153442383,
      "learning_rate": 0.0003922116628824438,
      "loss": 2.1356,
      "step": 92288
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.7891391515731812,
      "learning_rate": 0.00039220777033840124,
      "loss": 2.1638,
      "step": 92289
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.7832901477813721,
      "learning_rate": 0.00039220387777721554,
      "loss": 2.2013,
      "step": 92290
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.7768498659133911,
      "learning_rate": 0.0003921999851988874,
      "loss": 2.2894,
      "step": 92291
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.9600847363471985,
      "learning_rate": 0.0003921960926034175,
      "loss": 2.2257,
      "step": 92292
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.8458352088928223,
      "learning_rate": 0.0003921921999908066,
      "loss": 2.3653,
      "step": 92293
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.755481481552124,
      "learning_rate": 0.0003921883073610555,
      "loss": 2.2846,
      "step": 92294
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.9190346598625183,
      "learning_rate": 0.0003921844147141648,
      "loss": 2.3891,
      "step": 92295
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.8883048892021179,
      "learning_rate": 0.00039218052205013533,
      "loss": 2.2726,
      "step": 92296
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.8523612022399902,
      "learning_rate": 0.0003921766293689678,
      "loss": 2.2806,
      "step": 92297
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.848397433757782,
      "learning_rate": 0.00039217273667066277,
      "loss": 2.3511,
      "step": 92298
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.9143309593200684,
      "learning_rate": 0.0003921688439552211,
      "loss": 1.9647,
      "step": 92299
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.8064361214637756,
      "learning_rate": 0.00039216495122264364,
      "loss": 2.1021,
      "step": 92300
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.8744651675224304,
      "learning_rate": 0.00039216105847293084,
      "loss": 2.0916,
      "step": 92301
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.8135058283805847,
      "learning_rate": 0.0003921571657060836,
      "loss": 2.1865,
      "step": 92302
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.8422753810882568,
      "learning_rate": 0.0003921532729221026,
      "loss": 2.1797,
      "step": 92303
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.8359680771827698,
      "learning_rate": 0.00039214938012098856,
      "loss": 2.2266,
      "step": 92304
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.8145028948783875,
      "learning_rate": 0.0003921454873027422,
      "loss": 2.041,
      "step": 92305
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.9455961585044861,
      "learning_rate": 0.00039214159446736437,
      "loss": 1.9648,
      "step": 92306
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.8644506335258484,
      "learning_rate": 0.0003921377016148555,
      "loss": 2.3491,
      "step": 92307
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.841946005821228,
      "learning_rate": 0.00039213380874521654,
      "loss": 2.134,
      "step": 92308
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.7845275402069092,
      "learning_rate": 0.0003921299158584483,
      "loss": 2.1807,
      "step": 92309
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.9380821585655212,
      "learning_rate": 0.0003921260229545512,
      "loss": 1.9075,
      "step": 92310
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.8637820482254028,
      "learning_rate": 0.00039212213003352617,
      "loss": 2.2557,
      "step": 92311
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.8610314726829529,
      "learning_rate": 0.000392118237095374,
      "loss": 2.0082,
      "step": 92312
    },
    {
      "epoch": 1.2,
      "grad_norm": 1.056240200996399,
      "learning_rate": 0.00039211434414009527,
      "loss": 2.5188,
      "step": 92313
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.8573477864265442,
      "learning_rate": 0.0003921104511676907,
      "loss": 2.0655,
      "step": 92314
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.8411900401115417,
      "learning_rate": 0.000392106558178161,
      "loss": 2.0347,
      "step": 92315
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.9928432703018188,
      "learning_rate": 0.0003921026651715071,
      "loss": 2.2265,
      "step": 92316
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.8134759068489075,
      "learning_rate": 0.00039209877214772945,
      "loss": 2.2991,
      "step": 92317
    },
    {
      "epoch": 1.2,
      "grad_norm": 1.0350264310836792,
      "learning_rate": 0.00039209487910682895,
      "loss": 2.0466,
      "step": 92318
    },
    {
      "epoch": 1.2,
      "grad_norm": 1.023879885673523,
      "learning_rate": 0.00039209098604880635,
      "loss": 2.1691,
      "step": 92319
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.8089674115180969,
      "learning_rate": 0.0003920870929736622,
      "loss": 2.1846,
      "step": 92320
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.7998495697975159,
      "learning_rate": 0.0003920831998813973,
      "loss": 2.2678,
      "step": 92321
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.9180430173873901,
      "learning_rate": 0.00039207930677201245,
      "loss": 2.2373,
      "step": 92322
    },
    {
      "epoch": 1.2,
      "grad_norm": 1.188560128211975,
      "learning_rate": 0.0003920754136455083,
      "loss": 2.195,
      "step": 92323
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.7859539985656738,
      "learning_rate": 0.0003920715205018856,
      "loss": 2.3232,
      "step": 92324
    },
    {
      "epoch": 1.2,
      "grad_norm": 1.01596200466156,
      "learning_rate": 0.0003920676273411451,
      "loss": 2.1249,
      "step": 92325
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.9935163855552673,
      "learning_rate": 0.00039206373416328746,
      "loss": 2.2617,
      "step": 92326
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.8477074503898621,
      "learning_rate": 0.0003920598409683134,
      "loss": 2.124,
      "step": 92327
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.9186862707138062,
      "learning_rate": 0.00039205594775622375,
      "loss": 2.2281,
      "step": 92328
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.9288061857223511,
      "learning_rate": 0.00039205205452701907,
      "loss": 2.1288,
      "step": 92329
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.8232191801071167,
      "learning_rate": 0.00039204816128070026,
      "loss": 2.2334,
      "step": 92330
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.8451463580131531,
      "learning_rate": 0.00039204426801726797,
      "loss": 2.1911,
      "step": 92331
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.9077122807502747,
      "learning_rate": 0.0003920403747367228,
      "loss": 2.2859,
      "step": 92332
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.7844404578208923,
      "learning_rate": 0.00039203648143906565,
      "loss": 2.176,
      "step": 92333
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.809130847454071,
      "learning_rate": 0.0003920325881242972,
      "loss": 2.2528,
      "step": 92334
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.8787561058998108,
      "learning_rate": 0.00039202869479241815,
      "loss": 2.2401,
      "step": 92335
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.7828165888786316,
      "learning_rate": 0.0003920248014434292,
      "loss": 1.952,
      "step": 92336
    },
    {
      "epoch": 1.2,
      "grad_norm": 1.0779720544815063,
      "learning_rate": 0.00039202090807733113,
      "loss": 2.3291,
      "step": 92337
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.9227176308631897,
      "learning_rate": 0.00039201701469412464,
      "loss": 2.2999,
      "step": 92338
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.8480694890022278,
      "learning_rate": 0.0003920131212938104,
      "loss": 2.1619,
      "step": 92339
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.8075722455978394,
      "learning_rate": 0.00039200922787638933,
      "loss": 2.0871,
      "step": 92340
    },
    {
      "epoch": 1.2,
      "grad_norm": 1.1107171773910522,
      "learning_rate": 0.00039200533444186186,
      "loss": 2.0578,
      "step": 92341
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.874671220779419,
      "learning_rate": 0.0003920014409902288,
      "loss": 2.1926,
      "step": 92342
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.7848601937294006,
      "learning_rate": 0.0003919975475214911,
      "loss": 2.0816,
      "step": 92343
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.9843243360519409,
      "learning_rate": 0.0003919936540356493,
      "loss": 2.281,
      "step": 92344
    },
    {
      "epoch": 1.2,
      "grad_norm": 1.0768921375274658,
      "learning_rate": 0.000391989760532704,
      "loss": 2.22,
      "step": 92345
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.9434704184532166,
      "learning_rate": 0.0003919858670126562,
      "loss": 2.2952,
      "step": 92346
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.9952141046524048,
      "learning_rate": 0.0003919819734755065,
      "loss": 2.3825,
      "step": 92347
    },
    {
      "epoch": 1.2,
      "grad_norm": 1.4395368099212646,
      "learning_rate": 0.00039197807992125556,
      "loss": 2.2678,
      "step": 92348
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.8504041433334351,
      "learning_rate": 0.00039197418634990416,
      "loss": 2.22,
      "step": 92349
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.8798782229423523,
      "learning_rate": 0.000391970292761453,
      "loss": 1.9974,
      "step": 92350
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.8731552958488464,
      "learning_rate": 0.00039196639915590293,
      "loss": 2.3149,
      "step": 92351
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.8348525762557983,
      "learning_rate": 0.00039196250553325446,
      "loss": 2.0065,
      "step": 92352
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.9043435454368591,
      "learning_rate": 0.0003919586118935085,
      "loss": 2.2903,
      "step": 92353
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.8959417939186096,
      "learning_rate": 0.0003919547182366656,
      "loss": 2.121,
      "step": 92354
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.8933684825897217,
      "learning_rate": 0.00039195082456272665,
      "loss": 2.3174,
      "step": 92355
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.8397478461265564,
      "learning_rate": 0.00039194693087169234,
      "loss": 2.1812,
      "step": 92356
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.8584198355674744,
      "learning_rate": 0.00039194303716356325,
      "loss": 2.3275,
      "step": 92357
    },
    {
      "epoch": 1.2,
      "grad_norm": 1.028992772102356,
      "learning_rate": 0.00039193914343834036,
      "loss": 2.1252,
      "step": 92358
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.9738476276397705,
      "learning_rate": 0.0003919352496960241,
      "loss": 2.0679,
      "step": 92359
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.9089959859848022,
      "learning_rate": 0.0003919313559366154,
      "loss": 2.0635,
      "step": 92360
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.8299207091331482,
      "learning_rate": 0.000391927462160115,
      "loss": 2.3019,
      "step": 92361
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.9099791646003723,
      "learning_rate": 0.0003919235683665235,
      "loss": 2.0517,
      "step": 92362
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.7680360078811646,
      "learning_rate": 0.0003919196745558416,
      "loss": 2.0146,
      "step": 92363
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.8794628381729126,
      "learning_rate": 0.00039191578072807025,
      "loss": 2.0917,
      "step": 92364
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.87812739610672,
      "learning_rate": 0.00039191188688320993,
      "loss": 2.0339,
      "step": 92365
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.8057205677032471,
      "learning_rate": 0.00039190799302126134,
      "loss": 2.2324,
      "step": 92366
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.9065952897071838,
      "learning_rate": 0.00039190409914222556,
      "loss": 2.3336,
      "step": 92367
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.7729970812797546,
      "learning_rate": 0.00039190020524610287,
      "loss": 2.0253,
      "step": 92368
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.8148874044418335,
      "learning_rate": 0.00039189631133289434,
      "loss": 2.3644,
      "step": 92369
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.846331000328064,
      "learning_rate": 0.0003918924174026005,
      "loss": 2.0638,
      "step": 92370
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.8212701082229614,
      "learning_rate": 0.0003918885234552221,
      "loss": 2.0495,
      "step": 92371
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.8221237063407898,
      "learning_rate": 0.00039188462949075987,
      "loss": 2.2328,
      "step": 92372
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.8406381011009216,
      "learning_rate": 0.0003918807355092146,
      "loss": 2.1248,
      "step": 92373
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.8744922280311584,
      "learning_rate": 0.00039187684151058695,
      "loss": 2.2149,
      "step": 92374
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.825104832649231,
      "learning_rate": 0.0003918729474948777,
      "loss": 2.2987,
      "step": 92375
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.9221318364143372,
      "learning_rate": 0.00039186905346208744,
      "loss": 2.0709,
      "step": 92376
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.8625631928443909,
      "learning_rate": 0.000391865159412217,
      "loss": 2.3323,
      "step": 92377
    },
    {
      "epoch": 1.2,
      "grad_norm": 1.248416543006897,
      "learning_rate": 0.0003918612653452672,
      "loss": 2.3405,
      "step": 92378
    },
    {
      "epoch": 1.2,
      "grad_norm": 1.048017978668213,
      "learning_rate": 0.0003918573712612386,
      "loss": 2.1712,
      "step": 92379
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.8498184680938721,
      "learning_rate": 0.000391853477160132,
      "loss": 2.2399,
      "step": 92380
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.7371783256530762,
      "learning_rate": 0.00039184958304194805,
      "loss": 2.0675,
      "step": 92381
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.8680197596549988,
      "learning_rate": 0.0003918456889066876,
      "loss": 2.29,
      "step": 92382
    },
    {
      "epoch": 1.2,
      "grad_norm": 1.0811842679977417,
      "learning_rate": 0.00039184179475435123,
      "loss": 2.2402,
      "step": 92383
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.9212614297866821,
      "learning_rate": 0.00039183790058493976,
      "loss": 2.0853,
      "step": 92384
    },
    {
      "epoch": 1.2,
      "grad_norm": 1.0383014678955078,
      "learning_rate": 0.00039183400639845385,
      "loss": 2.0814,
      "step": 92385
    },
    {
      "epoch": 1.2,
      "grad_norm": 1.08202064037323,
      "learning_rate": 0.00039183011219489435,
      "loss": 2.1863,
      "step": 92386
    },
    {
      "epoch": 1.2,
      "grad_norm": 1.0125231742858887,
      "learning_rate": 0.00039182621797426186,
      "loss": 2.2411,
      "step": 92387
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.8837522864341736,
      "learning_rate": 0.00039182232373655715,
      "loss": 2.0747,
      "step": 92388
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.8637949824333191,
      "learning_rate": 0.00039181842948178097,
      "loss": 2.1321,
      "step": 92389
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.8722559809684753,
      "learning_rate": 0.0003918145352099339,
      "loss": 2.0592,
      "step": 92390
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.9165039658546448,
      "learning_rate": 0.0003918106409210168,
      "loss": 2.0729,
      "step": 92391
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.880342423915863,
      "learning_rate": 0.00039180674661503057,
      "loss": 2.1734,
      "step": 92392
    },
    {
      "epoch": 1.2,
      "grad_norm": 1.005704641342163,
      "learning_rate": 0.0003918028522919755,
      "loss": 2.3834,
      "step": 92393
    },
    {
      "epoch": 1.2,
      "grad_norm": 1.1019173860549927,
      "learning_rate": 0.0003917989579518526,
      "loss": 2.0178,
      "step": 92394
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.7988553643226624,
      "learning_rate": 0.0003917950635946627,
      "loss": 2.0789,
      "step": 92395
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.8582890629768372,
      "learning_rate": 0.00039179116922040614,
      "loss": 2.217,
      "step": 92396
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.8732059001922607,
      "learning_rate": 0.0003917872748290839,
      "loss": 2.1417,
      "step": 92397
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.9039947390556335,
      "learning_rate": 0.00039178338042069683,
      "loss": 2.2161,
      "step": 92398
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.8944591879844666,
      "learning_rate": 0.0003917794859952454,
      "loss": 2.1754,
      "step": 92399
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.8976030945777893,
      "learning_rate": 0.00039177559155273043,
      "loss": 2.2459,
      "step": 92400
    },
    {
      "epoch": 1.2,
      "grad_norm": 1.0999383926391602,
      "learning_rate": 0.00039177169709315264,
      "loss": 2.2846,
      "step": 92401
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.7635039687156677,
      "learning_rate": 0.00039176780261651275,
      "loss": 2.0127,
      "step": 92402
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.8853189945220947,
      "learning_rate": 0.0003917639081228115,
      "loss": 2.0293,
      "step": 92403
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.8532865047454834,
      "learning_rate": 0.0003917600136120497,
      "loss": 2.0149,
      "step": 92404
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.8020985126495361,
      "learning_rate": 0.0003917561190842278,
      "loss": 1.7847,
      "step": 92405
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.9036762118339539,
      "learning_rate": 0.00039175222453934685,
      "loss": 2.0859,
      "step": 92406
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.7820833325386047,
      "learning_rate": 0.00039174832997740746,
      "loss": 2.223,
      "step": 92407
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.9066985845565796,
      "learning_rate": 0.00039174443539841016,
      "loss": 2.2488,
      "step": 92408
    },
    {
      "epoch": 1.2,
      "grad_norm": 1.008811116218567,
      "learning_rate": 0.00039174054080235593,
      "loss": 2.1527,
      "step": 92409
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.7913796901702881,
      "learning_rate": 0.0003917366461892454,
      "loss": 2.1667,
      "step": 92410
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.961638331413269,
      "learning_rate": 0.0003917327515590793,
      "loss": 2.1505,
      "step": 92411
    },
    {
      "epoch": 1.2,
      "grad_norm": 1.0001412630081177,
      "learning_rate": 0.0003917288569118583,
      "loss": 2.0714,
      "step": 92412
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.8124580979347229,
      "learning_rate": 0.0003917249622475833,
      "loss": 2.0605,
      "step": 92413
    },
    {
      "epoch": 1.2,
      "grad_norm": 1.3268003463745117,
      "learning_rate": 0.00039172106756625477,
      "loss": 1.8647,
      "step": 92414
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.8885436654090881,
      "learning_rate": 0.0003917171728678736,
      "loss": 1.9133,
      "step": 92415
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.7856780886650085,
      "learning_rate": 0.0003917132781524405,
      "loss": 1.8216,
      "step": 92416
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.9192419052124023,
      "learning_rate": 0.00039170938341995617,
      "loss": 2.1468,
      "step": 92417
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.9957629442214966,
      "learning_rate": 0.0003917054886704213,
      "loss": 2.3128,
      "step": 92418
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.8713034391403198,
      "learning_rate": 0.0003917015939038367,
      "loss": 2.0689,
      "step": 92419
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.8713534474372864,
      "learning_rate": 0.00039169769912020296,
      "loss": 2.2632,
      "step": 92420
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.8272109031677246,
      "learning_rate": 0.00039169380431952096,
      "loss": 2.016,
      "step": 92421
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.9150679111480713,
      "learning_rate": 0.00039168990950179134,
      "loss": 2.0763,
      "step": 92422
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.8007465600967407,
      "learning_rate": 0.0003916860146670148,
      "loss": 2.4171,
      "step": 92423
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.8176863789558411,
      "learning_rate": 0.00039168211981519213,
      "loss": 2.3581,
      "step": 92424
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.8587584495544434,
      "learning_rate": 0.000391678224946324,
      "loss": 2.201,
      "step": 92425
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.8496100306510925,
      "learning_rate": 0.00039167433006041114,
      "loss": 2.2842,
      "step": 92426
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.7923869490623474,
      "learning_rate": 0.00039167043515745434,
      "loss": 2.1911,
      "step": 92427
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.8463762402534485,
      "learning_rate": 0.00039166654023745426,
      "loss": 2.2523,
      "step": 92428
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.8381211161613464,
      "learning_rate": 0.00039166264530041166,
      "loss": 2.2553,
      "step": 92429
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.8575815558433533,
      "learning_rate": 0.0003916587503463272,
      "loss": 2.2415,
      "step": 92430
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.82025545835495,
      "learning_rate": 0.0003916548553752017,
      "loss": 2.471,
      "step": 92431
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.8441411256790161,
      "learning_rate": 0.00039165096038703585,
      "loss": 2.0117,
      "step": 92432
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.8202729225158691,
      "learning_rate": 0.0003916470653818303,
      "loss": 2.2234,
      "step": 92433
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.7950331568717957,
      "learning_rate": 0.00039164317035958576,
      "loss": 1.9959,
      "step": 92434
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.8461004495620728,
      "learning_rate": 0.00039163927532030306,
      "loss": 2.0376,
      "step": 92435
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.8077170252799988,
      "learning_rate": 0.000391635380263983,
      "loss": 2.0805,
      "step": 92436
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.8786081671714783,
      "learning_rate": 0.0003916314851906261,
      "loss": 2.0686,
      "step": 92437
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.9007052779197693,
      "learning_rate": 0.0003916275901002332,
      "loss": 2.3475,
      "step": 92438
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.831511378288269,
      "learning_rate": 0.0003916236949928051,
      "loss": 2.1087,
      "step": 92439
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.7788582444190979,
      "learning_rate": 0.00039161979986834225,
      "loss": 2.0475,
      "step": 92440
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.836633563041687,
      "learning_rate": 0.00039161590472684563,
      "loss": 2.3763,
      "step": 92441
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.7874855399131775,
      "learning_rate": 0.0003916120095683158,
      "loss": 2.022,
      "step": 92442
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.8075104355812073,
      "learning_rate": 0.00039160811439275373,
      "loss": 2.1396,
      "step": 92443
    },
    {
      "epoch": 1.2,
      "grad_norm": 1.0539605617523193,
      "learning_rate": 0.0003916042192001599,
      "loss": 2.1088,
      "step": 92444
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.8719131946563721,
      "learning_rate": 0.00039160032399053506,
      "loss": 2.3831,
      "step": 92445
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.8844865560531616,
      "learning_rate": 0.0003915964287638801,
      "loss": 2.1582,
      "step": 92446
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.882390022277832,
      "learning_rate": 0.0003915925335201956,
      "loss": 2.2385,
      "step": 92447
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.9054157137870789,
      "learning_rate": 0.00039158863825948227,
      "loss": 2.2576,
      "step": 92448
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.7896001935005188,
      "learning_rate": 0.0003915847429817409,
      "loss": 2.389,
      "step": 92449
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.8061541318893433,
      "learning_rate": 0.0003915808476869723,
      "loss": 2.2185,
      "step": 92450
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.9741584658622742,
      "learning_rate": 0.00039157695237517696,
      "loss": 2.1322,
      "step": 92451
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.7968098521232605,
      "learning_rate": 0.00039157305704635574,
      "loss": 2.1349,
      "step": 92452
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.8509932160377502,
      "learning_rate": 0.00039156916170050946,
      "loss": 2.3289,
      "step": 92453
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.8139662146568298,
      "learning_rate": 0.00039156526633763865,
      "loss": 2.1312,
      "step": 92454
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.8052682280540466,
      "learning_rate": 0.00039156137095774414,
      "loss": 2.2037,
      "step": 92455
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.881993293762207,
      "learning_rate": 0.0003915574755608268,
      "loss": 2.2801,
      "step": 92456
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.8259057402610779,
      "learning_rate": 0.000391553580146887,
      "loss": 2.2681,
      "step": 92457
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.8336668610572815,
      "learning_rate": 0.0003915496847159257,
      "loss": 2.1727,
      "step": 92458
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.8937198519706726,
      "learning_rate": 0.0003915457892679436,
      "loss": 2.1557,
      "step": 92459
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.7798980474472046,
      "learning_rate": 0.00039154189380294146,
      "loss": 2.2603,
      "step": 92460
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.8441879153251648,
      "learning_rate": 0.0003915379983209199,
      "loss": 2.1069,
      "step": 92461
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.8931244015693665,
      "learning_rate": 0.0003915341028218798,
      "loss": 2.1576,
      "step": 92462
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.8538731336593628,
      "learning_rate": 0.00039153020730582167,
      "loss": 2.3924,
      "step": 92463
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.9167554974555969,
      "learning_rate": 0.00039152631177274635,
      "loss": 2.267,
      "step": 92464
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.842513918876648,
      "learning_rate": 0.00039152241622265465,
      "loss": 2.0227,
      "step": 92465
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.8447369933128357,
      "learning_rate": 0.0003915185206555471,
      "loss": 2.0611,
      "step": 92466
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.8108542561531067,
      "learning_rate": 0.0003915146250714245,
      "loss": 1.9685,
      "step": 92467
    },
    {
      "epoch": 1.2,
      "grad_norm": 1.0256736278533936,
      "learning_rate": 0.00039151072947028775,
      "loss": 2.1247,
      "step": 92468
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.7882694005966187,
      "learning_rate": 0.0003915068338521373,
      "loss": 2.1416,
      "step": 92469
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.8553889393806458,
      "learning_rate": 0.00039150293821697406,
      "loss": 2.2005,
      "step": 92470
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.8545663952827454,
      "learning_rate": 0.00039149904256479876,
      "loss": 2.2099,
      "step": 92471
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.7969509363174438,
      "learning_rate": 0.00039149514689561194,
      "loss": 2.1554,
      "step": 92472
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.8533521890640259,
      "learning_rate": 0.0003914912512094145,
      "loss": 2.2679,
      "step": 92473
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.8110066652297974,
      "learning_rate": 0.00039148735550620714,
      "loss": 2.2094,
      "step": 92474
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.8248550295829773,
      "learning_rate": 0.0003914834597859905,
      "loss": 1.9843,
      "step": 92475
    },
    {
      "epoch": 1.2,
      "grad_norm": 1.0032844543457031,
      "learning_rate": 0.0003914795640487653,
      "loss": 2.2422,
      "step": 92476
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.7989527583122253,
      "learning_rate": 0.0003914756682945325,
      "loss": 2.1636,
      "step": 92477
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.827563464641571,
      "learning_rate": 0.00039147177252329247,
      "loss": 2.2416,
      "step": 92478
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.7942273616790771,
      "learning_rate": 0.0003914678767350462,
      "loss": 1.8704,
      "step": 92479
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.8163598775863647,
      "learning_rate": 0.00039146398092979436,
      "loss": 2.1996,
      "step": 92480
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.9378194808959961,
      "learning_rate": 0.00039146008510753755,
      "loss": 2.3806,
      "step": 92481
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.8849923014640808,
      "learning_rate": 0.00039145618926827666,
      "loss": 2.306,
      "step": 92482
    },
    {
      "epoch": 1.2,
      "grad_norm": 1.059096097946167,
      "learning_rate": 0.00039145229341201233,
      "loss": 2.3126,
      "step": 92483
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.8860353231430054,
      "learning_rate": 0.0003914483975387452,
      "loss": 2.1812,
      "step": 92484
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.8080748319625854,
      "learning_rate": 0.00039144450164847614,
      "loss": 2.1611,
      "step": 92485
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.8729425072669983,
      "learning_rate": 0.0003914406057412059,
      "loss": 2.357,
      "step": 92486
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.9181926846504211,
      "learning_rate": 0.00039143670981693504,
      "loss": 2.4134,
      "step": 92487
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.8623121380805969,
      "learning_rate": 0.0003914328138756644,
      "loss": 2.268,
      "step": 92488
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.8037983179092407,
      "learning_rate": 0.0003914289179173947,
      "loss": 2.0922,
      "step": 92489
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.8363027572631836,
      "learning_rate": 0.0003914250219421266,
      "loss": 2.3147,
      "step": 92490
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.8372597098350525,
      "learning_rate": 0.0003914211259498608,
      "loss": 2.2745,
      "step": 92491
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.7803137898445129,
      "learning_rate": 0.00039141722994059816,
      "loss": 1.9629,
      "step": 92492
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.8535628318786621,
      "learning_rate": 0.00039141333391433934,
      "loss": 1.9375,
      "step": 92493
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.8354459404945374,
      "learning_rate": 0.000391409437871085,
      "loss": 2.2127,
      "step": 92494
    },
    {
      "epoch": 1.2,
      "grad_norm": 1.0063023567199707,
      "learning_rate": 0.000391405541810836,
      "loss": 2.0274,
      "step": 92495
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.7977749705314636,
      "learning_rate": 0.00039140164573359294,
      "loss": 2.3034,
      "step": 92496
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.8603148460388184,
      "learning_rate": 0.0003913977496393566,
      "loss": 2.2541,
      "step": 92497
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.8108282089233398,
      "learning_rate": 0.0003913938535281277,
      "loss": 2.1252,
      "step": 92498
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.8014984726905823,
      "learning_rate": 0.00039138995739990693,
      "loss": 2.2736,
      "step": 92499
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.8947073221206665,
      "learning_rate": 0.00039138606125469497,
      "loss": 2.0326,
      "step": 92500
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.8034877777099609,
      "learning_rate": 0.0003913821650924928,
      "loss": 2.0176,
      "step": 92501
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.8690451979637146,
      "learning_rate": 0.00039137826891330087,
      "loss": 2.2091,
      "step": 92502
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.9487408995628357,
      "learning_rate": 0.00039137437271711997,
      "loss": 2.0565,
      "step": 92503
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.7382123470306396,
      "learning_rate": 0.00039137047650395087,
      "loss": 2.1262,
      "step": 92504
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.8579257130622864,
      "learning_rate": 0.0003913665802737942,
      "loss": 2.1517,
      "step": 92505
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.9313939809799194,
      "learning_rate": 0.00039136268402665083,
      "loss": 2.083,
      "step": 92506
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.8602336645126343,
      "learning_rate": 0.0003913587877625215,
      "loss": 2.1155,
      "step": 92507
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.8937511444091797,
      "learning_rate": 0.00039135489148140666,
      "loss": 2.0286,
      "step": 92508
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.9149526953697205,
      "learning_rate": 0.00039135099518330734,
      "loss": 2.2254,
      "step": 92509
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.9249700307846069,
      "learning_rate": 0.00039134709886822416,
      "loss": 2.0425,
      "step": 92510
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.8234723806381226,
      "learning_rate": 0.00039134320253615776,
      "loss": 2.2674,
      "step": 92511
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.9637719392776489,
      "learning_rate": 0.00039133930618710897,
      "loss": 2.1586,
      "step": 92512
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.8269864916801453,
      "learning_rate": 0.00039133540982107853,
      "loss": 2.1139,
      "step": 92513
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.8396340012550354,
      "learning_rate": 0.00039133151343806704,
      "loss": 2.2267,
      "step": 92514
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.9462974667549133,
      "learning_rate": 0.0003913276170380753,
      "loss": 2.2491,
      "step": 92515
    },
    {
      "epoch": 1.2,
      "grad_norm": 1.0470253229141235,
      "learning_rate": 0.000391323720621104,
      "loss": 2.5134,
      "step": 92516
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.8689756989479065,
      "learning_rate": 0.000391319824187154,
      "loss": 2.1219,
      "step": 92517
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.8434659242630005,
      "learning_rate": 0.00039131592773622583,
      "loss": 2.2326,
      "step": 92518
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.9339041709899902,
      "learning_rate": 0.00039131203126832034,
      "loss": 1.963,
      "step": 92519
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.903387188911438,
      "learning_rate": 0.00039130813478343825,
      "loss": 2.3123,
      "step": 92520
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.8560200929641724,
      "learning_rate": 0.0003913042382815802,
      "loss": 2.1727,
      "step": 92521
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.8360046744346619,
      "learning_rate": 0.0003913003417627469,
      "loss": 2.1534,
      "step": 92522
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.9048632383346558,
      "learning_rate": 0.0003912964452269393,
      "loss": 2.1134,
      "step": 92523
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.8766961693763733,
      "learning_rate": 0.000391292548674158,
      "loss": 2.2183,
      "step": 92524
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.8533429503440857,
      "learning_rate": 0.0003912886521044035,
      "loss": 1.9957,
      "step": 92525
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.9381287693977356,
      "learning_rate": 0.00039128475551767684,
      "loss": 2.0778,
      "step": 92526
    },
    {
      "epoch": 1.2,
      "grad_norm": 1.1736220121383667,
      "learning_rate": 0.0003912808589139785,
      "loss": 2.2477,
      "step": 92527
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.8867688775062561,
      "learning_rate": 0.0003912769622933095,
      "loss": 1.8647,
      "step": 92528
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.8700571656227112,
      "learning_rate": 0.0003912730656556703,
      "loss": 1.9514,
      "step": 92529
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.9489212036132812,
      "learning_rate": 0.00039126916900106165,
      "loss": 2.2125,
      "step": 92530
    },
    {
      "epoch": 1.2,
      "grad_norm": 1.063576102256775,
      "learning_rate": 0.0003912652723294845,
      "loss": 1.9998,
      "step": 92531
    },
    {
      "epoch": 1.2,
      "grad_norm": 1.1182371377944946,
      "learning_rate": 0.00039126137564093926,
      "loss": 2.1052,
      "step": 92532
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.8458132147789001,
      "learning_rate": 0.00039125747893542685,
      "loss": 2.2315,
      "step": 92533
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.9931900501251221,
      "learning_rate": 0.000391253582212948,
      "loss": 2.2259,
      "step": 92534
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.8005113005638123,
      "learning_rate": 0.0003912496854735034,
      "loss": 2.4307,
      "step": 92535
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.8110736012458801,
      "learning_rate": 0.0003912457887170937,
      "loss": 2.298,
      "step": 92536
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.785875678062439,
      "learning_rate": 0.0003912418919437197,
      "loss": 2.2252,
      "step": 92537
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.8084511756896973,
      "learning_rate": 0.0003912379951533821,
      "loss": 2.2621,
      "step": 92538
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.8305302262306213,
      "learning_rate": 0.0003912340983460817,
      "loss": 2.354,
      "step": 92539
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.7871800661087036,
      "learning_rate": 0.0003912302015218191,
      "loss": 2.1735,
      "step": 92540
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.8529016971588135,
      "learning_rate": 0.0003912263046805951,
      "loss": 2.0489,
      "step": 92541
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.7732135057449341,
      "learning_rate": 0.0003912224078224104,
      "loss": 2.1165,
      "step": 92542
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.8308782577514648,
      "learning_rate": 0.00039121851094726565,
      "loss": 2.3521,
      "step": 92543
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.7560194134712219,
      "learning_rate": 0.0003912146140551618,
      "loss": 2.2517,
      "step": 92544
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.8779910206794739,
      "learning_rate": 0.00039121071714609936,
      "loss": 2.4012,
      "step": 92545
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.8090597987174988,
      "learning_rate": 0.00039120682022007915,
      "loss": 2.3526,
      "step": 92546
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.9617592096328735,
      "learning_rate": 0.0003912029232771019,
      "loss": 2.1383,
      "step": 92547
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.8466718196868896,
      "learning_rate": 0.00039119902631716823,
      "loss": 2.1963,
      "step": 92548
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.839469313621521,
      "learning_rate": 0.0003911951293402789,
      "loss": 2.3198,
      "step": 92549
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.7702010869979858,
      "learning_rate": 0.00039119123234643483,
      "loss": 2.2331,
      "step": 92550
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.9820094704627991,
      "learning_rate": 0.00039118733533563646,
      "loss": 2.2289,
      "step": 92551
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.758798360824585,
      "learning_rate": 0.0003911834383078847,
      "loss": 2.203,
      "step": 92552
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.8368285298347473,
      "learning_rate": 0.00039117954126318023,
      "loss": 2.2082,
      "step": 92553
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.8611103296279907,
      "learning_rate": 0.00039117564420152373,
      "loss": 2.1448,
      "step": 92554
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.9306322932243347,
      "learning_rate": 0.00039117174712291594,
      "loss": 2.3082,
      "step": 92555
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.879203736782074,
      "learning_rate": 0.0003911678500273577,
      "loss": 2.2662,
      "step": 92556
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.8204546570777893,
      "learning_rate": 0.0003911639529148495,
      "loss": 2.318,
      "step": 92557
    },
    {
      "epoch": 1.21,
      "grad_norm": 1.0524557828903198,
      "learning_rate": 0.00039116005578539224,
      "loss": 2.2115,
      "step": 92558
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.9102075695991516,
      "learning_rate": 0.0003911561586389867,
      "loss": 2.3297,
      "step": 92559
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.8106778264045715,
      "learning_rate": 0.00039115226147563343,
      "loss": 2.2899,
      "step": 92560
    },
    {
      "epoch": 1.21,
      "grad_norm": 1.064164161682129,
      "learning_rate": 0.00039114836429533323,
      "loss": 2.3321,
      "step": 92561
    },
    {
      "epoch": 1.21,
      "grad_norm": 1.0969905853271484,
      "learning_rate": 0.00039114446709808683,
      "loss": 2.4232,
      "step": 92562
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.8342824578285217,
      "learning_rate": 0.0003911405698838949,
      "loss": 2.1715,
      "step": 92563
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.9887476563453674,
      "learning_rate": 0.0003911366726527583,
      "loss": 2.2606,
      "step": 92564
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.8535800576210022,
      "learning_rate": 0.00039113277540467773,
      "loss": 2.2522,
      "step": 92565
    },
    {
      "epoch": 1.21,
      "grad_norm": 1.0512404441833496,
      "learning_rate": 0.0003911288781396538,
      "loss": 2.1435,
      "step": 92566
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.7996166348457336,
      "learning_rate": 0.00039112498085768714,
      "loss": 2.0177,
      "step": 92567
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.9985887408256531,
      "learning_rate": 0.00039112108355877884,
      "loss": 2.0712,
      "step": 92568
    },
    {
      "epoch": 1.21,
      "grad_norm": 1.0100152492523193,
      "learning_rate": 0.0003911171862429293,
      "loss": 2.027,
      "step": 92569
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.872161328792572,
      "learning_rate": 0.00039111328891013936,
      "loss": 2.0583,
      "step": 92570
    },
    {
      "epoch": 1.21,
      "grad_norm": 1.0218113660812378,
      "learning_rate": 0.0003911093915604098,
      "loss": 2.285,
      "step": 92571
    },
    {
      "epoch": 1.21,
      "grad_norm": 1.1166363954544067,
      "learning_rate": 0.0003911054941937412,
      "loss": 2.1824,
      "step": 92572
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.8775424957275391,
      "learning_rate": 0.00039110159681013444,
      "loss": 2.1914,
      "step": 92573
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.8755151033401489,
      "learning_rate": 0.00039109769940959017,
      "loss": 1.9035,
      "step": 92574
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.8902116417884827,
      "learning_rate": 0.00039109380199210913,
      "loss": 2.173,
      "step": 92575
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.9385483264923096,
      "learning_rate": 0.00039108990455769194,
      "loss": 2.2749,
      "step": 92576
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.8871963024139404,
      "learning_rate": 0.00039108600710633946,
      "loss": 2.1764,
      "step": 92577
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.8447055816650391,
      "learning_rate": 0.00039108210963805244,
      "loss": 2.0797,
      "step": 92578
    },
    {
      "epoch": 1.21,
      "grad_norm": 1.2098913192749023,
      "learning_rate": 0.0003910782121528315,
      "loss": 2.0817,
      "step": 92579
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.9049286246299744,
      "learning_rate": 0.0003910743146506774,
      "loss": 2.0725,
      "step": 92580
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.9294094443321228,
      "learning_rate": 0.0003910704171315908,
      "loss": 2.1038,
      "step": 92581
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.9031423330307007,
      "learning_rate": 0.00039106651959557254,
      "loss": 2.0423,
      "step": 92582
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.8715634942054749,
      "learning_rate": 0.00039106262204262327,
      "loss": 2.1369,
      "step": 92583
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.9467308521270752,
      "learning_rate": 0.00039105872447274385,
      "loss": 2.0745,
      "step": 92584
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.8304599523544312,
      "learning_rate": 0.0003910548268859347,
      "loss": 2.3029,
      "step": 92585
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.9053027629852295,
      "learning_rate": 0.00039105092928219687,
      "loss": 2.3841,
      "step": 92586
    },
    {
      "epoch": 1.21,
      "grad_norm": 1.1303569078445435,
      "learning_rate": 0.000391047031661531,
      "loss": 2.1272,
      "step": 92587
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.8204120397567749,
      "learning_rate": 0.0003910431340239376,
      "loss": 2.3579,
      "step": 92588
    },
    {
      "epoch": 1.21,
      "grad_norm": 1.0334848165512085,
      "learning_rate": 0.00039103923636941767,
      "loss": 2.1596,
      "step": 92589
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.8182181119918823,
      "learning_rate": 0.0003910353386979719,
      "loss": 2.0825,
      "step": 92590
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.793544590473175,
      "learning_rate": 0.0003910314410096008,
      "loss": 2.1496,
      "step": 92591
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.8222700357437134,
      "learning_rate": 0.0003910275433043053,
      "loss": 2.0898,
      "step": 92592
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.8220700025558472,
      "learning_rate": 0.00039102364558208616,
      "loss": 1.9956,
      "step": 92593
    },
    {
      "epoch": 1.21,
      "grad_norm": 1.0973280668258667,
      "learning_rate": 0.0003910197478429438,
      "loss": 1.9524,
      "step": 92594
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.8692630529403687,
      "learning_rate": 0.00039101585008687934,
      "loss": 2.0909,
      "step": 92595
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.8612714409828186,
      "learning_rate": 0.00039101195231389324,
      "loss": 2.0607,
      "step": 92596
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.942498505115509,
      "learning_rate": 0.00039100805452398627,
      "loss": 2.0133,
      "step": 92597
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.8345301747322083,
      "learning_rate": 0.0003910041567171592,
      "loss": 2.1227,
      "step": 92598
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.8916497230529785,
      "learning_rate": 0.0003910002588934128,
      "loss": 2.0453,
      "step": 92599
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.877175509929657,
      "learning_rate": 0.0003909963610527476,
      "loss": 2.2533,
      "step": 92600
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.8603724241256714,
      "learning_rate": 0.00039099246319516457,
      "loss": 2.0674,
      "step": 92601
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.9908543825149536,
      "learning_rate": 0.00039098856532066435,
      "loss": 2.0328,
      "step": 92602
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.894413411617279,
      "learning_rate": 0.0003909846674292476,
      "loss": 1.9583,
      "step": 92603
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.8306755423545837,
      "learning_rate": 0.000390980769520915,
      "loss": 2.105,
      "step": 92604
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.8860388398170471,
      "learning_rate": 0.0003909768715956674,
      "loss": 2.3861,
      "step": 92605
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.9063660502433777,
      "learning_rate": 0.0003909729736535055,
      "loss": 2.1049,
      "step": 92606
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.8622214794158936,
      "learning_rate": 0.00039096907569443007,
      "loss": 2.2175,
      "step": 92607
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.8664332032203674,
      "learning_rate": 0.00039096517771844164,
      "loss": 2.2382,
      "step": 92608
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.8576102256774902,
      "learning_rate": 0.00039096127972554123,
      "loss": 2.0217,
      "step": 92609
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.9327130317687988,
      "learning_rate": 0.0003909573817157292,
      "loss": 2.23,
      "step": 92610
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.8825425505638123,
      "learning_rate": 0.0003909534836890066,
      "loss": 2.2775,
      "step": 92611
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.8482481837272644,
      "learning_rate": 0.0003909495856453741,
      "loss": 2.1394,
      "step": 92612
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.8882706165313721,
      "learning_rate": 0.00039094568758483224,
      "loss": 2.2389,
      "step": 92613
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.8150797486305237,
      "learning_rate": 0.0003909417895073819,
      "loss": 2.1705,
      "step": 92614
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.845323383808136,
      "learning_rate": 0.0003909378914130237,
      "loss": 2.2117,
      "step": 92615
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.8045508861541748,
      "learning_rate": 0.00039093399330175855,
      "loss": 2.121,
      "step": 92616
    },
    {
      "epoch": 1.21,
      "grad_norm": 1.129699945449829,
      "learning_rate": 0.00039093009517358695,
      "loss": 2.2665,
      "step": 92617
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.721563458442688,
      "learning_rate": 0.00039092619702850976,
      "loss": 2.3712,
      "step": 92618
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.8137356638908386,
      "learning_rate": 0.0003909222988665276,
      "loss": 2.141,
      "step": 92619
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.9609707593917847,
      "learning_rate": 0.00039091840068764146,
      "loss": 2.13,
      "step": 92620
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.8795451521873474,
      "learning_rate": 0.00039091450249185175,
      "loss": 2.1807,
      "step": 92621
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.8509706258773804,
      "learning_rate": 0.00039091060427915927,
      "loss": 2.167,
      "step": 92622
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.8665722012519836,
      "learning_rate": 0.00039090670604956495,
      "loss": 2.135,
      "step": 92623
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.8339154124259949,
      "learning_rate": 0.0003909028078030692,
      "loss": 2.2213,
      "step": 92624
    },
    {
      "epoch": 1.21,
      "grad_norm": 1.028987169265747,
      "learning_rate": 0.000390898909539673,
      "loss": 2.2646,
      "step": 92625
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.8468186855316162,
      "learning_rate": 0.00039089501125937694,
      "loss": 2.0993,
      "step": 92626
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.8141187429428101,
      "learning_rate": 0.00039089111296218175,
      "loss": 1.8859,
      "step": 92627
    },
    {
      "epoch": 1.21,
      "grad_norm": 1.012826681137085,
      "learning_rate": 0.00039088721464808823,
      "loss": 2.1467,
      "step": 92628
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.813342809677124,
      "learning_rate": 0.0003908833163170971,
      "loss": 2.1021,
      "step": 92629
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.8550342321395874,
      "learning_rate": 0.0003908794179692089,
      "loss": 2.1229,
      "step": 92630
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.8952913880348206,
      "learning_rate": 0.0003908755196044246,
      "loss": 2.3502,
      "step": 92631
    },
    {
      "epoch": 1.21,
      "grad_norm": 1.015038251876831,
      "learning_rate": 0.0003908716212227449,
      "loss": 2.1454,
      "step": 92632
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.8167315125465393,
      "learning_rate": 0.0003908677228241704,
      "loss": 2.1005,
      "step": 92633
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.7540097236633301,
      "learning_rate": 0.0003908638244087018,
      "loss": 1.9013,
      "step": 92634
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.8048154711723328,
      "learning_rate": 0.00039085992597633995,
      "loss": 2.2701,
      "step": 92635
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.8094382286071777,
      "learning_rate": 0.0003908560275270855,
      "loss": 2.205,
      "step": 92636
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.7819294333457947,
      "learning_rate": 0.00039085212906093925,
      "loss": 1.8926,
      "step": 92637
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.8189448714256287,
      "learning_rate": 0.00039084823057790183,
      "loss": 2.1018,
      "step": 92638
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.8238987922668457,
      "learning_rate": 0.000390844332077974,
      "loss": 2.3763,
      "step": 92639
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.846072256565094,
      "learning_rate": 0.00039084043356115653,
      "loss": 2.2361,
      "step": 92640
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.863012433052063,
      "learning_rate": 0.0003908365350274501,
      "loss": 2.1994,
      "step": 92641
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.8451650738716125,
      "learning_rate": 0.00039083263647685546,
      "loss": 2.2592,
      "step": 92642
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.861547589302063,
      "learning_rate": 0.0003908287379093733,
      "loss": 2.2815,
      "step": 92643
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.9429839253425598,
      "learning_rate": 0.0003908248393250044,
      "loss": 2.0767,
      "step": 92644
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.8407978415489197,
      "learning_rate": 0.0003908209407237494,
      "loss": 2.1127,
      "step": 92645
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.9205266833305359,
      "learning_rate": 0.00039081704210560905,
      "loss": 2.3442,
      "step": 92646
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.8642603158950806,
      "learning_rate": 0.0003908131434705842,
      "loss": 2.2322,
      "step": 92647
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.8342409729957581,
      "learning_rate": 0.00039080924481867536,
      "loss": 2.2916,
      "step": 92648
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.7853479385375977,
      "learning_rate": 0.0003908053461498834,
      "loss": 2.1608,
      "step": 92649
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.8761562705039978,
      "learning_rate": 0.000390801447464209,
      "loss": 2.1783,
      "step": 92650
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.817798912525177,
      "learning_rate": 0.000390797548761653,
      "loss": 2.5632,
      "step": 92651
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.8213850259780884,
      "learning_rate": 0.0003907936500422158,
      "loss": 2.1166,
      "step": 92652
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.8184368014335632,
      "learning_rate": 0.00039078975130589857,
      "loss": 2.2394,
      "step": 92653
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.7491927742958069,
      "learning_rate": 0.00039078585255270177,
      "loss": 2.2654,
      "step": 92654
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.8722456693649292,
      "learning_rate": 0.00039078195378262605,
      "loss": 2.0933,
      "step": 92655
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.8674866557121277,
      "learning_rate": 0.00039077805499567234,
      "loss": 2.1518,
      "step": 92656
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.8639129400253296,
      "learning_rate": 0.0003907741561918413,
      "loss": 2.1349,
      "step": 92657
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.8232486844062805,
      "learning_rate": 0.00039077025737113355,
      "loss": 2.3303,
      "step": 92658
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.7386752367019653,
      "learning_rate": 0.00039076635853354993,
      "loss": 1.9395,
      "step": 92659
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.8016143441200256,
      "learning_rate": 0.0003907624596790912,
      "loss": 2.2595,
      "step": 92660
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.8955720663070679,
      "learning_rate": 0.0003907585608077579,
      "loss": 2.3069,
      "step": 92661
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.7977784872055054,
      "learning_rate": 0.00039075466191955096,
      "loss": 2.1433,
      "step": 92662
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.8390867114067078,
      "learning_rate": 0.000390750763014471,
      "loss": 2.3306,
      "step": 92663
    },
    {
      "epoch": 1.21,
      "grad_norm": 1.0486125946044922,
      "learning_rate": 0.00039074686409251876,
      "loss": 1.9775,
      "step": 92664
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.8607181310653687,
      "learning_rate": 0.0003907429651536949,
      "loss": 1.9748,
      "step": 92665
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.7664312124252319,
      "learning_rate": 0.0003907390661980003,
      "loss": 2.2327,
      "step": 92666
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.8204205632209778,
      "learning_rate": 0.0003907351672254355,
      "loss": 2.258,
      "step": 92667
    },
    {
      "epoch": 1.21,
      "grad_norm": 1.226444125175476,
      "learning_rate": 0.0003907312682360014,
      "loss": 2.1324,
      "step": 92668
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.8796067237854004,
      "learning_rate": 0.00039072736922969863,
      "loss": 2.4143,
      "step": 92669
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.8428590893745422,
      "learning_rate": 0.0003907234702065279,
      "loss": 2.2499,
      "step": 92670
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.8740059733390808,
      "learning_rate": 0.00039071957116648996,
      "loss": 2.2191,
      "step": 92671
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.9679437875747681,
      "learning_rate": 0.00039071567210958557,
      "loss": 2.255,
      "step": 92672
    },
    {
      "epoch": 1.21,
      "grad_norm": 1.0048034191131592,
      "learning_rate": 0.0003907117730358154,
      "loss": 2.3233,
      "step": 92673
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.8118441700935364,
      "learning_rate": 0.00039070787394518023,
      "loss": 2.2727,
      "step": 92674
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.9773760437965393,
      "learning_rate": 0.00039070397483768074,
      "loss": 1.8693,
      "step": 92675
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.838162899017334,
      "learning_rate": 0.0003907000757133176,
      "loss": 2.2066,
      "step": 92676
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.9271218180656433,
      "learning_rate": 0.0003906961765720917,
      "loss": 2.3301,
      "step": 92677
    },
    {
      "epoch": 1.21,
      "grad_norm": 1.0280029773712158,
      "learning_rate": 0.0003906922774140037,
      "loss": 2.2397,
      "step": 92678
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.7876444458961487,
      "learning_rate": 0.0003906883782390542,
      "loss": 2.1988,
      "step": 92679
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.8398628234863281,
      "learning_rate": 0.000390684479047244,
      "loss": 2.0644,
      "step": 92680
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.7985846996307373,
      "learning_rate": 0.00039068057983857393,
      "loss": 2.1278,
      "step": 92681
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.8096298575401306,
      "learning_rate": 0.0003906766806130446,
      "loss": 2.2037,
      "step": 92682
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.8313522934913635,
      "learning_rate": 0.00039067278137065677,
      "loss": 2.2274,
      "step": 92683
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.8665987253189087,
      "learning_rate": 0.00039066888211141114,
      "loss": 2.0928,
      "step": 92684
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.8290319442749023,
      "learning_rate": 0.00039066498283530846,
      "loss": 2.1095,
      "step": 92685
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.9709165096282959,
      "learning_rate": 0.0003906610835423494,
      "loss": 2.083,
      "step": 92686
    },
    {
      "epoch": 1.21,
      "grad_norm": 1.0780309438705444,
      "learning_rate": 0.00039065718423253486,
      "loss": 2.099,
      "step": 92687
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.8562632203102112,
      "learning_rate": 0.0003906532849058653,
      "loss": 2.2774,
      "step": 92688
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.9759197235107422,
      "learning_rate": 0.0003906493855623416,
      "loss": 2.3595,
      "step": 92689
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.9146836400032043,
      "learning_rate": 0.00039064548620196455,
      "loss": 2.0563,
      "step": 92690
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.9147529006004333,
      "learning_rate": 0.0003906415868247348,
      "loss": 2.0548,
      "step": 92691
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.8225149512290955,
      "learning_rate": 0.000390637687430653,
      "loss": 2.0184,
      "step": 92692
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.9948586821556091,
      "learning_rate": 0.00039063378801971994,
      "loss": 2.5008,
      "step": 92693
    },
    {
      "epoch": 1.21,
      "grad_norm": 1.2771399021148682,
      "learning_rate": 0.00039062988859193634,
      "loss": 2.118,
      "step": 92694
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.9570960402488708,
      "learning_rate": 0.00039062598914730293,
      "loss": 2.2922,
      "step": 92695
    },
    {
      "epoch": 1.21,
      "grad_norm": 1.071459174156189,
      "learning_rate": 0.0003906220896858205,
      "loss": 2.0885,
      "step": 92696
    },
    {
      "epoch": 1.21,
      "grad_norm": 1.0535639524459839,
      "learning_rate": 0.0003906181902074897,
      "loss": 2.2009,
      "step": 92697
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.9197811484336853,
      "learning_rate": 0.0003906142907123112,
      "loss": 2.0352,
      "step": 92698
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.8974238038063049,
      "learning_rate": 0.00039061039120028587,
      "loss": 2.469,
      "step": 92699
    },
    {
      "epoch": 1.21,
      "grad_norm": 1.0277174711227417,
      "learning_rate": 0.00039060649167141424,
      "loss": 2.2978,
      "step": 92700
    },
    {
      "epoch": 1.21,
      "grad_norm": 1.1338495016098022,
      "learning_rate": 0.0003906025921256973,
      "loss": 1.9246,
      "step": 92701
    },
    {
      "epoch": 1.21,
      "grad_norm": 1.0891456604003906,
      "learning_rate": 0.0003905986925631356,
      "loss": 1.9613,
      "step": 92702
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.989454984664917,
      "learning_rate": 0.0003905947929837298,
      "loss": 2.2214,
      "step": 92703
    },
    {
      "epoch": 1.21,
      "grad_norm": 1.2325592041015625,
      "learning_rate": 0.0003905908933874808,
      "loss": 2.1586,
      "step": 92704
    },
    {
      "epoch": 1.21,
      "grad_norm": 1.280094861984253,
      "learning_rate": 0.0003905869937743893,
      "loss": 2.1496,
      "step": 92705
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.8767051100730896,
      "learning_rate": 0.00039058309414445586,
      "loss": 2.1665,
      "step": 92706
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.7893862128257751,
      "learning_rate": 0.0003905791944976813,
      "loss": 2.099,
      "step": 92707
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.9452749490737915,
      "learning_rate": 0.00039057529483406644,
      "loss": 2.1199,
      "step": 92708
    },
    {
      "epoch": 1.21,
      "grad_norm": 1.1101531982421875,
      "learning_rate": 0.0003905713951536119,
      "loss": 1.9556,
      "step": 92709
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.8742304444313049,
      "learning_rate": 0.0003905674954563184,
      "loss": 2.4156,
      "step": 92710
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.9063519835472107,
      "learning_rate": 0.0003905635957421867,
      "loss": 2.0814,
      "step": 92711
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.9653174877166748,
      "learning_rate": 0.0003905596960112176,
      "loss": 2.1707,
      "step": 92712
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.8957406282424927,
      "learning_rate": 0.00039055579626341155,
      "loss": 2.1448,
      "step": 92713
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.8103469014167786,
      "learning_rate": 0.0003905518964987697,
      "loss": 2.0226,
      "step": 92714
    },
    {
      "epoch": 1.21,
      "grad_norm": 1.0344818830490112,
      "learning_rate": 0.0003905479967172924,
      "loss": 2.3408,
      "step": 92715
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.8669765591621399,
      "learning_rate": 0.00039054409691898054,
      "loss": 2.2545,
      "step": 92716
    },
    {
      "epoch": 1.21,
      "grad_norm": 1.0229284763336182,
      "learning_rate": 0.00039054019710383476,
      "loss": 2.2746,
      "step": 92717
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.8506069779396057,
      "learning_rate": 0.00039053629727185603,
      "loss": 2.3247,
      "step": 92718
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.8778465986251831,
      "learning_rate": 0.00039053239742304474,
      "loss": 2.1916,
      "step": 92719
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.8426674008369446,
      "learning_rate": 0.0003905284975574018,
      "loss": 2.3521,
      "step": 92720
    },
    {
      "epoch": 1.21,
      "grad_norm": 1.2084119319915771,
      "learning_rate": 0.00039052459767492794,
      "loss": 2.1971,
      "step": 92721
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.8405429720878601,
      "learning_rate": 0.0003905206977756238,
      "loss": 1.992,
      "step": 92722
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.8431429862976074,
      "learning_rate": 0.00039051679785949016,
      "loss": 2.2432,
      "step": 92723
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.8516724109649658,
      "learning_rate": 0.0003905128979265278,
      "loss": 2.0422,
      "step": 92724
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.9778727293014526,
      "learning_rate": 0.0003905089979767373,
      "loss": 2.0891,
      "step": 92725
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.8634687066078186,
      "learning_rate": 0.0003905050980101195,
      "loss": 2.0698,
      "step": 92726
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.7675467729568481,
      "learning_rate": 0.0003905011980266752,
      "loss": 1.9819,
      "step": 92727
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.9011937975883484,
      "learning_rate": 0.00039049729802640484,
      "loss": 2.2898,
      "step": 92728
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.9202120304107666,
      "learning_rate": 0.0003904933980093094,
      "loss": 2.145,
      "step": 92729
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.8546010851860046,
      "learning_rate": 0.0003904894979753896,
      "loss": 2.0522,
      "step": 92730
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.9470445513725281,
      "learning_rate": 0.000390485597924646,
      "loss": 2.1655,
      "step": 92731
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.9010990858078003,
      "learning_rate": 0.00039048169785707945,
      "loss": 2.04,
      "step": 92732
    },
    {
      "epoch": 1.21,
      "grad_norm": 1.0422803163528442,
      "learning_rate": 0.0003904777977726907,
      "loss": 2.1654,
      "step": 92733
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.869977593421936,
      "learning_rate": 0.0003904738976714803,
      "loss": 2.0914,
      "step": 92734
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.909940779209137,
      "learning_rate": 0.0003904699975534491,
      "loss": 2.1404,
      "step": 92735
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.8061879277229309,
      "learning_rate": 0.00039046609741859794,
      "loss": 2.165,
      "step": 92736
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.8683686852455139,
      "learning_rate": 0.00039046219726692735,
      "loss": 2.1025,
      "step": 92737
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.8643568158149719,
      "learning_rate": 0.0003904582970984381,
      "loss": 2.0497,
      "step": 92738
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.8239340782165527,
      "learning_rate": 0.00039045439691313104,
      "loss": 2.289,
      "step": 92739
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.8238736391067505,
      "learning_rate": 0.0003904504967110067,
      "loss": 2.222,
      "step": 92740
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.8161696791648865,
      "learning_rate": 0.000390446596492066,
      "loss": 2.0233,
      "step": 92741
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.9709975719451904,
      "learning_rate": 0.00039044269625630956,
      "loss": 2.3699,
      "step": 92742
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.8960158228874207,
      "learning_rate": 0.000390438796003738,
      "loss": 2.2632,
      "step": 92743
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.8340932726860046,
      "learning_rate": 0.00039043489573435225,
      "loss": 2.227,
      "step": 92744
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.7936429977416992,
      "learning_rate": 0.000390430995448153,
      "loss": 1.8723,
      "step": 92745
    },
    {
      "epoch": 1.21,
      "grad_norm": 1.0862032175064087,
      "learning_rate": 0.00039042709514514084,
      "loss": 2.3627,
      "step": 92746
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.9864262342453003,
      "learning_rate": 0.00039042319482531656,
      "loss": 2.2303,
      "step": 92747
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.9193960428237915,
      "learning_rate": 0.000390419294488681,
      "loss": 2.1634,
      "step": 92748
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.8175013065338135,
      "learning_rate": 0.00039041539413523467,
      "loss": 2.2706,
      "step": 92749
    },
    {
      "epoch": 1.21,
      "grad_norm": 1.114940881729126,
      "learning_rate": 0.00039041149376497846,
      "loss": 2.225,
      "step": 92750
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.8990631103515625,
      "learning_rate": 0.00039040759337791305,
      "loss": 1.9239,
      "step": 92751
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.8714750409126282,
      "learning_rate": 0.0003904036929740391,
      "loss": 2.2787,
      "step": 92752
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.915285050868988,
      "learning_rate": 0.00039039979255335746,
      "loss": 2.2583,
      "step": 92753
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.8768242001533508,
      "learning_rate": 0.0003903958921158689,
      "loss": 2.2672,
      "step": 92754
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.857291042804718,
      "learning_rate": 0.00039039199166157384,
      "loss": 2.177,
      "step": 92755
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.7442957758903503,
      "learning_rate": 0.0003903880911904733,
      "loss": 2.282,
      "step": 92756
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.8397358059883118,
      "learning_rate": 0.0003903841907025679,
      "loss": 2.3637,
      "step": 92757
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.9257894158363342,
      "learning_rate": 0.0003903802901978583,
      "loss": 2.3074,
      "step": 92758
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.8192451000213623,
      "learning_rate": 0.00039037638967634545,
      "loss": 2.2474,
      "step": 92759
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.8663390278816223,
      "learning_rate": 0.0003903724891380298,
      "loss": 2.1823,
      "step": 92760
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.8454979062080383,
      "learning_rate": 0.00039036858858291227,
      "loss": 2.2941,
      "step": 92761
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.8439953923225403,
      "learning_rate": 0.0003903646880109935,
      "loss": 2.0701,
      "step": 92762
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.8330093026161194,
      "learning_rate": 0.00039036078742227416,
      "loss": 2.1356,
      "step": 92763
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.836015522480011,
      "learning_rate": 0.00039035688681675514,
      "loss": 2.2952,
      "step": 92764
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.9227760434150696,
      "learning_rate": 0.000390352986194437,
      "loss": 2.312,
      "step": 92765
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.8434181809425354,
      "learning_rate": 0.00039034908555532054,
      "loss": 2.1837,
      "step": 92766
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.942127525806427,
      "learning_rate": 0.0003903451848994065,
      "loss": 2.0997,
      "step": 92767
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.8795592784881592,
      "learning_rate": 0.0003903412842266956,
      "loss": 2.396,
      "step": 92768
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.8656880259513855,
      "learning_rate": 0.00039033738353718854,
      "loss": 2.2649,
      "step": 92769
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.8436987996101379,
      "learning_rate": 0.00039033348283088604,
      "loss": 2.0139,
      "step": 92770
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.8512672185897827,
      "learning_rate": 0.00039032958210778886,
      "loss": 2.1414,
      "step": 92771
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.7650206089019775,
      "learning_rate": 0.00039032568136789766,
      "loss": 2.1107,
      "step": 92772
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.9305499792098999,
      "learning_rate": 0.00039032178061121326,
      "loss": 2.2108,
      "step": 92773
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.8625169992446899,
      "learning_rate": 0.0003903178798377363,
      "loss": 2.1309,
      "step": 92774
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.9215855598449707,
      "learning_rate": 0.0003903139790474676,
      "loss": 2.1377,
      "step": 92775
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.7977268099784851,
      "learning_rate": 0.00039031007824040775,
      "loss": 2.2191,
      "step": 92776
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.854415237903595,
      "learning_rate": 0.0003903061774165576,
      "loss": 2.146,
      "step": 92777
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.7661834955215454,
      "learning_rate": 0.00039030227657591785,
      "loss": 2.0649,
      "step": 92778
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.8493055701255798,
      "learning_rate": 0.0003902983757184891,
      "loss": 2.2125,
      "step": 92779
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.8179815411567688,
      "learning_rate": 0.00039029447484427226,
      "loss": 2.1834,
      "step": 92780
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.8115429282188416,
      "learning_rate": 0.00039029057395326795,
      "loss": 2.1081,
      "step": 92781
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.8653932809829712,
      "learning_rate": 0.000390286673045477,
      "loss": 2.2163,
      "step": 92782
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.7533186674118042,
      "learning_rate": 0.0003902827721208999,
      "loss": 2.2545,
      "step": 92783
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.9245119690895081,
      "learning_rate": 0.00039027887117953767,
      "loss": 2.3162,
      "step": 92784
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.8950535655021667,
      "learning_rate": 0.00039027497022139083,
      "loss": 2.1976,
      "step": 92785
    },
    {
      "epoch": 1.21,
      "grad_norm": 1.1376092433929443,
      "learning_rate": 0.0003902710692464601,
      "loss": 2.2782,
      "step": 92786
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.8086826801300049,
      "learning_rate": 0.0003902671682547463,
      "loss": 2.3043,
      "step": 92787
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.8647747039794922,
      "learning_rate": 0.0003902632672462502,
      "loss": 2.2866,
      "step": 92788
    },
    {
      "epoch": 1.21,
      "grad_norm": 1.0719823837280273,
      "learning_rate": 0.0003902593662209725,
      "loss": 2.2694,
      "step": 92789
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.8393144607543945,
      "learning_rate": 0.0003902554651789138,
      "loss": 2.1738,
      "step": 92790
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.8470333218574524,
      "learning_rate": 0.00039025156412007486,
      "loss": 2.0332,
      "step": 92791
    },
    {
      "epoch": 1.21,
      "grad_norm": 1.137558937072754,
      "learning_rate": 0.0003902476630444566,
      "loss": 2.2818,
      "step": 92792
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.9802145957946777,
      "learning_rate": 0.00039024376195205947,
      "loss": 2.3314,
      "step": 92793
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.7347210645675659,
      "learning_rate": 0.0003902398608428843,
      "loss": 2.1317,
      "step": 92794
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.8514446020126343,
      "learning_rate": 0.0003902359597169319,
      "loss": 2.011,
      "step": 92795
    },
    {
      "epoch": 1.21,
      "grad_norm": 1.0966867208480835,
      "learning_rate": 0.0003902320585742029,
      "loss": 2.1914,
      "step": 92796
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.9155769944190979,
      "learning_rate": 0.0003902281574146981,
      "loss": 2.1948,
      "step": 92797
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.8421998620033264,
      "learning_rate": 0.0003902242562384181,
      "loss": 2.152,
      "step": 92798
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.9795229434967041,
      "learning_rate": 0.0003902203550453638,
      "loss": 2.093,
      "step": 92799
    },
    {
      "epoch": 1.21,
      "grad_norm": 1.1092859506607056,
      "learning_rate": 0.0003902164538355358,
      "loss": 2.1927,
      "step": 92800
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.7994393110275269,
      "learning_rate": 0.00039021255260893487,
      "loss": 2.2958,
      "step": 92801
    },
    {
      "epoch": 1.21,
      "grad_norm": 1.0087312459945679,
      "learning_rate": 0.0003902086513655617,
      "loss": 2.1611,
      "step": 92802
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.9640604257583618,
      "learning_rate": 0.0003902047501054171,
      "loss": 2.17,
      "step": 92803
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.9266268014907837,
      "learning_rate": 0.0003902008488285016,
      "loss": 1.9098,
      "step": 92804
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.8204320669174194,
      "learning_rate": 0.0003901969475348162,
      "loss": 2.291,
      "step": 92805
    },
    {
      "epoch": 1.21,
      "grad_norm": 1.0846681594848633,
      "learning_rate": 0.00039019304622436147,
      "loss": 2.2758,
      "step": 92806
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.9143531322479248,
      "learning_rate": 0.00039018914489713806,
      "loss": 2.227,
      "step": 92807
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.8336869478225708,
      "learning_rate": 0.00039018524355314676,
      "loss": 2.3744,
      "step": 92808
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.8905468583106995,
      "learning_rate": 0.00039018134219238844,
      "loss": 2.0888,
      "step": 92809
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.904362142086029,
      "learning_rate": 0.0003901774408148637,
      "loss": 2.1041,
      "step": 92810
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.9396591186523438,
      "learning_rate": 0.0003901735394205732,
      "loss": 2.1559,
      "step": 92811
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.8035548329353333,
      "learning_rate": 0.00039016963800951783,
      "loss": 2.2948,
      "step": 92812
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.8806495070457458,
      "learning_rate": 0.0003901657365816981,
      "loss": 2.2492,
      "step": 92813
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.9632166028022766,
      "learning_rate": 0.000390161835137115,
      "loss": 2.4767,
      "step": 92814
    },
    {
      "epoch": 1.21,
      "grad_norm": 1.2804145812988281,
      "learning_rate": 0.000390157933675769,
      "loss": 2.1164,
      "step": 92815
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.86064213514328,
      "learning_rate": 0.0003901540321976609,
      "loss": 2.0945,
      "step": 92816
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.8629500269889832,
      "learning_rate": 0.00039015013070279155,
      "loss": 2.1282,
      "step": 92817
    },
    {
      "epoch": 1.21,
      "grad_norm": 1.197676420211792,
      "learning_rate": 0.00039014622919116167,
      "loss": 2.1637,
      "step": 92818
    },
    {
      "epoch": 1.21,
      "grad_norm": 1.0383402109146118,
      "learning_rate": 0.00039014232766277176,
      "loss": 2.1117,
      "step": 92819
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.9550004601478577,
      "learning_rate": 0.00039013842611762274,
      "loss": 2.2015,
      "step": 92820
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.9022150635719299,
      "learning_rate": 0.0003901345245557153,
      "loss": 2.1062,
      "step": 92821
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.8468334674835205,
      "learning_rate": 0.00039013062297705013,
      "loss": 2.1248,
      "step": 92822
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.9538134932518005,
      "learning_rate": 0.00039012672138162796,
      "loss": 2.1176,
      "step": 92823
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.8144134283065796,
      "learning_rate": 0.00039012281976944955,
      "loss": 2.2212,
      "step": 92824
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.834865927696228,
      "learning_rate": 0.0003901189181405156,
      "loss": 2.2524,
      "step": 92825
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.8723632097244263,
      "learning_rate": 0.00039011501649482684,
      "loss": 2.2108,
      "step": 92826
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.9118698239326477,
      "learning_rate": 0.00039011111483238406,
      "loss": 2.4482,
      "step": 92827
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.8617120981216431,
      "learning_rate": 0.00039010721315318786,
      "loss": 2.1724,
      "step": 92828
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.8864555954933167,
      "learning_rate": 0.000390103311457239,
      "loss": 2.1374,
      "step": 92829
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.9480841159820557,
      "learning_rate": 0.0003900994097445383,
      "loss": 2.2939,
      "step": 92830
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.8556404709815979,
      "learning_rate": 0.0003900955080150864,
      "loss": 2.5385,
      "step": 92831
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.8216344714164734,
      "learning_rate": 0.000390091606268884,
      "loss": 2.2901,
      "step": 92832
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.930618405342102,
      "learning_rate": 0.00039008770450593195,
      "loss": 2.1878,
      "step": 92833
    },
    {
      "epoch": 1.21,
      "grad_norm": 1.0242671966552734,
      "learning_rate": 0.0003900838027262308,
      "loss": 2.1962,
      "step": 92834
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.7913795113563538,
      "learning_rate": 0.0003900799009297814,
      "loss": 2.2335,
      "step": 92835
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.978766679763794,
      "learning_rate": 0.00039007599911658453,
      "loss": 2.1838,
      "step": 92836
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.9255920052528381,
      "learning_rate": 0.00039007209728664075,
      "loss": 2.2399,
      "step": 92837
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.7998701333999634,
      "learning_rate": 0.0003900681954399508,
      "loss": 2.284,
      "step": 92838
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.819189727306366,
      "learning_rate": 0.0003900642935765157,
      "loss": 2.2203,
      "step": 92839
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.934135913848877,
      "learning_rate": 0.00039006039169633576,
      "loss": 2.1683,
      "step": 92840
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.9271395206451416,
      "learning_rate": 0.0003900564897994119,
      "loss": 1.8804,
      "step": 92841
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.9038575291633606,
      "learning_rate": 0.0003900525878857449,
      "loss": 2.2896,
      "step": 92842
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.8931303024291992,
      "learning_rate": 0.00039004868595533544,
      "loss": 2.3313,
      "step": 92843
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.8592356443405151,
      "learning_rate": 0.0003900447840081842,
      "loss": 2.146,
      "step": 92844
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.7975561022758484,
      "learning_rate": 0.0003900408820442919,
      "loss": 2.2002,
      "step": 92845
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.8604877591133118,
      "learning_rate": 0.0003900369800636593,
      "loss": 2.3757,
      "step": 92846
    },
    {
      "epoch": 1.21,
      "grad_norm": 1.0011262893676758,
      "learning_rate": 0.0003900330780662872,
      "loss": 2.0224,
      "step": 92847
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.7888371348381042,
      "learning_rate": 0.0003900291760521762,
      "loss": 2.2492,
      "step": 92848
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.7970616221427917,
      "learning_rate": 0.000390025274021327,
      "loss": 2.1872,
      "step": 92849
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.8348972797393799,
      "learning_rate": 0.00039002137197374054,
      "loss": 2.3757,
      "step": 92850
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.8919828534126282,
      "learning_rate": 0.00039001746990941725,
      "loss": 2.2195,
      "step": 92851
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.7844417095184326,
      "learning_rate": 0.0003900135678283582,
      "loss": 2.2806,
      "step": 92852
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.8950238823890686,
      "learning_rate": 0.00039000966573056386,
      "loss": 2.0932,
      "step": 92853
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.8128701448440552,
      "learning_rate": 0.00039000576361603495,
      "loss": 2.0645,
      "step": 92854
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.9054847359657288,
      "learning_rate": 0.00039000186148477236,
      "loss": 2.3513,
      "step": 92855
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.8851852416992188,
      "learning_rate": 0.0003899979593367767,
      "loss": 2.1111,
      "step": 92856
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.8336651921272278,
      "learning_rate": 0.0003899940571720486,
      "loss": 2.3942,
      "step": 92857
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.8682360053062439,
      "learning_rate": 0.00038999015499058915,
      "loss": 2.0742,
      "step": 92858
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.8243355751037598,
      "learning_rate": 0.00038998625279239866,
      "loss": 2.2024,
      "step": 92859
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.856063723564148,
      "learning_rate": 0.00038998235057747805,
      "loss": 2.2304,
      "step": 92860
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.8499491214752197,
      "learning_rate": 0.0003899784483458281,
      "loss": 2.3341,
      "step": 92861
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.8633828163146973,
      "learning_rate": 0.00038997454609744934,
      "loss": 2.2042,
      "step": 92862
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.8435317277908325,
      "learning_rate": 0.00038997064383234265,
      "loss": 2.0737,
      "step": 92863
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.9218726754188538,
      "learning_rate": 0.00038996674155050877,
      "loss": 2.265,
      "step": 92864
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.8412052392959595,
      "learning_rate": 0.0003899628392519483,
      "loss": 2.25,
      "step": 92865
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.7551280856132507,
      "learning_rate": 0.0003899589369366621,
      "loss": 2.2983,
      "step": 92866
    },
    {
      "epoch": 1.21,
      "grad_norm": 1.0422735214233398,
      "learning_rate": 0.00038995503460465084,
      "loss": 2.1368,
      "step": 92867
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.9325482845306396,
      "learning_rate": 0.0003899511322559152,
      "loss": 2.0612,
      "step": 92868
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.8602277040481567,
      "learning_rate": 0.000389947229890456,
      "loss": 2.1209,
      "step": 92869
    },
    {
      "epoch": 1.21,
      "grad_norm": 1.0856322050094604,
      "learning_rate": 0.0003899433275082739,
      "loss": 2.2915,
      "step": 92870
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.9003051519393921,
      "learning_rate": 0.00038993942510936954,
      "loss": 2.0842,
      "step": 92871
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.9058757424354553,
      "learning_rate": 0.00038993552269374386,
      "loss": 2.4715,
      "step": 92872
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.8839882016181946,
      "learning_rate": 0.0003899316202613974,
      "loss": 2.1511,
      "step": 92873
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.9266034960746765,
      "learning_rate": 0.000389927717812331,
      "loss": 2.1904,
      "step": 92874
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.8764623999595642,
      "learning_rate": 0.0003899238153465453,
      "loss": 2.1515,
      "step": 92875
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.95281982421875,
      "learning_rate": 0.00038991991286404105,
      "loss": 2.2795,
      "step": 92876
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.9344595670700073,
      "learning_rate": 0.00038991601036481905,
      "loss": 2.0554,
      "step": 92877
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.9142400026321411,
      "learning_rate": 0.00038991210784887996,
      "loss": 2.1742,
      "step": 92878
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.9219827055931091,
      "learning_rate": 0.0003899082053162245,
      "loss": 1.9544,
      "step": 92879
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.9392098188400269,
      "learning_rate": 0.00038990430276685336,
      "loss": 2.1674,
      "step": 92880
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.8498556613922119,
      "learning_rate": 0.00038990040020076737,
      "loss": 2.0816,
      "step": 92881
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.7651716470718384,
      "learning_rate": 0.0003898964976179672,
      "loss": 2.2457,
      "step": 92882
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.8482109308242798,
      "learning_rate": 0.00038989259501845354,
      "loss": 2.1534,
      "step": 92883
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.8621588945388794,
      "learning_rate": 0.0003898886924022272,
      "loss": 2.3197,
      "step": 92884
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.7626882791519165,
      "learning_rate": 0.00038988478976928877,
      "loss": 1.882,
      "step": 92885
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.8581913709640503,
      "learning_rate": 0.00038988088711963914,
      "loss": 2.2369,
      "step": 92886
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.8777769207954407,
      "learning_rate": 0.00038987698445327887,
      "loss": 2.2825,
      "step": 92887
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.8186362385749817,
      "learning_rate": 0.0003898730817702089,
      "loss": 2.1675,
      "step": 92888
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.8799759745597839,
      "learning_rate": 0.00038986917907042975,
      "loss": 2.1007,
      "step": 92889
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.7911610007286072,
      "learning_rate": 0.0003898652763539422,
      "loss": 2.1972,
      "step": 92890
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.8142202496528625,
      "learning_rate": 0.00038986137362074705,
      "loss": 2.2308,
      "step": 92891
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.8199106454849243,
      "learning_rate": 0.00038985747087084494,
      "loss": 2.0983,
      "step": 92892
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.8474363088607788,
      "learning_rate": 0.0003898535681042366,
      "loss": 2.302,
      "step": 92893
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.9403558373451233,
      "learning_rate": 0.00038984966532092287,
      "loss": 2.0794,
      "step": 92894
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.8813556432723999,
      "learning_rate": 0.0003898457625209044,
      "loss": 2.1299,
      "step": 92895
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.9484084844589233,
      "learning_rate": 0.0003898418597041818,
      "loss": 2.1729,
      "step": 92896
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.9245476126670837,
      "learning_rate": 0.000389837956870756,
      "loss": 2.1774,
      "step": 92897
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.8222016096115112,
      "learning_rate": 0.0003898340540206276,
      "loss": 2.0686,
      "step": 92898
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.8304422497749329,
      "learning_rate": 0.00038983015115379735,
      "loss": 2.2342,
      "step": 92899
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.8696776628494263,
      "learning_rate": 0.00038982624827026596,
      "loss": 2.3464,
      "step": 92900
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.9135789275169373,
      "learning_rate": 0.0003898223453700342,
      "loss": 2.0957,
      "step": 92901
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.7219896912574768,
      "learning_rate": 0.00038981844245310275,
      "loss": 2.1036,
      "step": 92902
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.9140698909759521,
      "learning_rate": 0.0003898145395194724,
      "loss": 2.2257,
      "step": 92903
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.9919061660766602,
      "learning_rate": 0.00038981063656914373,
      "loss": 2.157,
      "step": 92904
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.8239730596542358,
      "learning_rate": 0.0003898067336021176,
      "loss": 2.2072,
      "step": 92905
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.7697424292564392,
      "learning_rate": 0.00038980283061839483,
      "loss": 2.133,
      "step": 92906
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.9301751852035522,
      "learning_rate": 0.0003897989276179759,
      "loss": 2.129,
      "step": 92907
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.9128156900405884,
      "learning_rate": 0.0003897950246008617,
      "loss": 2.1096,
      "step": 92908
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.8602675795555115,
      "learning_rate": 0.0003897911215670529,
      "loss": 2.2287,
      "step": 92909
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.8302600979804993,
      "learning_rate": 0.0003897872185165502,
      "loss": 2.1549,
      "step": 92910
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.8446561694145203,
      "learning_rate": 0.00038978331544935435,
      "loss": 2.2178,
      "step": 92911
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.8310613036155701,
      "learning_rate": 0.0003897794123654662,
      "loss": 2.2452,
      "step": 92912
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.7643123865127563,
      "learning_rate": 0.00038977550926488626,
      "loss": 2.2328,
      "step": 92913
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.7994091510772705,
      "learning_rate": 0.0003897716061476153,
      "loss": 2.1886,
      "step": 92914
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.8846218585968018,
      "learning_rate": 0.0003897677030136543,
      "loss": 2.1892,
      "step": 92915
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.7801792621612549,
      "learning_rate": 0.00038976379986300363,
      "loss": 2.1298,
      "step": 92916
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.8864582180976868,
      "learning_rate": 0.00038975989669566417,
      "loss": 2.0035,
      "step": 92917
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.96917724609375,
      "learning_rate": 0.0003897559935116367,
      "loss": 2.1146,
      "step": 92918
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.7873293161392212,
      "learning_rate": 0.000389752090310922,
      "loss": 2.2887,
      "step": 92919
    },
    {
      "epoch": 1.21,
      "grad_norm": 1.0939607620239258,
      "learning_rate": 0.00038974818709352056,
      "loss": 1.9757,
      "step": 92920
    },
    {
      "epoch": 1.21,
      "grad_norm": 1.0421067476272583,
      "learning_rate": 0.0003897442838594332,
      "loss": 2.2831,
      "step": 92921
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.8467457294464111,
      "learning_rate": 0.0003897403806086608,
      "loss": 2.1351,
      "step": 92922
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.8419665694236755,
      "learning_rate": 0.00038973647734120395,
      "loss": 2.0932,
      "step": 92923
    },
    {
      "epoch": 1.21,
      "grad_norm": 1.0010173320770264,
      "learning_rate": 0.00038973257405706333,
      "loss": 1.9092,
      "step": 92924
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.9365845322608948,
      "learning_rate": 0.00038972867075623977,
      "loss": 2.2557,
      "step": 92925
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.9540339708328247,
      "learning_rate": 0.0003897247674387339,
      "loss": 2.2547,
      "step": 92926
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.9128720164299011,
      "learning_rate": 0.00038972086410454657,
      "loss": 2.432,
      "step": 92927
    },
    {
      "epoch": 1.21,
      "grad_norm": 1.2293529510498047,
      "learning_rate": 0.00038971696075367846,
      "loss": 2.0748,
      "step": 92928
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.8131119012832642,
      "learning_rate": 0.00038971305738613027,
      "loss": 2.2732,
      "step": 92929
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.8572548627853394,
      "learning_rate": 0.0003897091540019026,
      "loss": 2.147,
      "step": 92930
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.9105451107025146,
      "learning_rate": 0.00038970525060099645,
      "loss": 1.9848,
      "step": 92931
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.8104694485664368,
      "learning_rate": 0.0003897013471834124,
      "loss": 1.9761,
      "step": 92932
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.9376713633537292,
      "learning_rate": 0.0003896974437491511,
      "loss": 2.2365,
      "step": 92933
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.8512026071548462,
      "learning_rate": 0.0003896935402982134,
      "loss": 2.2012,
      "step": 92934
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.9330690503120422,
      "learning_rate": 0.00038968963683059995,
      "loss": 2.268,
      "step": 92935
    },
    {
      "epoch": 1.21,
      "grad_norm": 1.0512800216674805,
      "learning_rate": 0.00038968573334631154,
      "loss": 2.5148,
      "step": 92936
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.7913661599159241,
      "learning_rate": 0.0003896818298453488,
      "loss": 2.1574,
      "step": 92937
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.8843454122543335,
      "learning_rate": 0.0003896779263277125,
      "loss": 2.3558,
      "step": 92938
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.9182115197181702,
      "learning_rate": 0.00038967402279340346,
      "loss": 2.2368,
      "step": 92939
    },
    {
      "epoch": 1.21,
      "grad_norm": 1.0409317016601562,
      "learning_rate": 0.00038967011924242236,
      "loss": 2.1441,
      "step": 92940
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.8582848310470581,
      "learning_rate": 0.00038966621567476976,
      "loss": 2.1528,
      "step": 92941
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.8976194262504578,
      "learning_rate": 0.00038966231209044657,
      "loss": 2.3992,
      "step": 92942
    },
    {
      "epoch": 1.21,
      "grad_norm": 1.209837794303894,
      "learning_rate": 0.00038965840848945355,
      "loss": 2.1581,
      "step": 92943
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.9283168315887451,
      "learning_rate": 0.0003896545048717912,
      "loss": 2.0639,
      "step": 92944
    },
    {
      "epoch": 1.21,
      "grad_norm": 1.0076425075531006,
      "learning_rate": 0.00038965060123746044,
      "loss": 2.1075,
      "step": 92945
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.8240746259689331,
      "learning_rate": 0.000389646697586462,
      "loss": 2.0662,
      "step": 92946
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.8492333889007568,
      "learning_rate": 0.0003896427939187965,
      "loss": 2.2828,
      "step": 92947
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.8142862915992737,
      "learning_rate": 0.0003896388902344647,
      "loss": 1.9752,
      "step": 92948
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.8056126236915588,
      "learning_rate": 0.00038963498653346736,
      "loss": 2.044,
      "step": 92949
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.93641597032547,
      "learning_rate": 0.0003896310828158052,
      "loss": 2.1105,
      "step": 92950
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.8125507831573486,
      "learning_rate": 0.00038962717908147884,
      "loss": 2.0039,
      "step": 92951
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.9060901403427124,
      "learning_rate": 0.00038962327533048917,
      "loss": 2.0196,
      "step": 92952
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.8521976470947266,
      "learning_rate": 0.0003896193715628368,
      "loss": 2.2718,
      "step": 92953
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.8451990485191345,
      "learning_rate": 0.00038961546777852257,
      "loss": 2.1802,
      "step": 92954
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.8532938361167908,
      "learning_rate": 0.0003896115639775471,
      "loss": 2.1183,
      "step": 92955
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.8280973434448242,
      "learning_rate": 0.00038960766015991115,
      "loss": 2.1926,
      "step": 92956
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.8043786883354187,
      "learning_rate": 0.00038960375632561545,
      "loss": 2.1676,
      "step": 92957
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.8634020090103149,
      "learning_rate": 0.00038959985247466065,
      "loss": 2.224,
      "step": 92958
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.787842333316803,
      "learning_rate": 0.00038959594860704763,
      "loss": 2.0898,
      "step": 92959
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.8610693216323853,
      "learning_rate": 0.000389592044722777,
      "loss": 2.1669,
      "step": 92960
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.8753900527954102,
      "learning_rate": 0.0003895881408218495,
      "loss": 1.9898,
      "step": 92961
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.849358856678009,
      "learning_rate": 0.0003895842369042659,
      "loss": 2.1743,
      "step": 92962
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.8121724128723145,
      "learning_rate": 0.0003895803329700269,
      "loss": 2.2109,
      "step": 92963
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.778396725654602,
      "learning_rate": 0.0003895764290191332,
      "loss": 2.1055,
      "step": 92964
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.9049407243728638,
      "learning_rate": 0.00038957252505158557,
      "loss": 2.059,
      "step": 92965
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.8016354441642761,
      "learning_rate": 0.0003895686210673847,
      "loss": 2.0617,
      "step": 92966
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.8839825987815857,
      "learning_rate": 0.00038956471706653133,
      "loss": 2.0476,
      "step": 92967
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.9689002633094788,
      "learning_rate": 0.0003895608130490262,
      "loss": 2.0306,
      "step": 92968
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.9426239728927612,
      "learning_rate": 0.0003895569090148701,
      "loss": 2.2791,
      "step": 92969
    },
    {
      "epoch": 1.21,
      "grad_norm": 1.0165510177612305,
      "learning_rate": 0.0003895530049640636,
      "loss": 2.1728,
      "step": 92970
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.8734689950942993,
      "learning_rate": 0.00038954910089660747,
      "loss": 2.1375,
      "step": 92971
    },
    {
      "epoch": 1.21,
      "grad_norm": 1.2619893550872803,
      "learning_rate": 0.0003895451968125025,
      "loss": 2.2668,
      "step": 92972
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.8547557592391968,
      "learning_rate": 0.00038954129271174945,
      "loss": 2.2241,
      "step": 92973
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.7888010144233704,
      "learning_rate": 0.0003895373885943489,
      "loss": 2.0224,
      "step": 92974
    },
    {
      "epoch": 1.21,
      "grad_norm": 1.0356361865997314,
      "learning_rate": 0.00038953348446030175,
      "loss": 2.5103,
      "step": 92975
    },
    {
      "epoch": 1.21,
      "grad_norm": 1.0246964693069458,
      "learning_rate": 0.0003895295803096086,
      "loss": 2.3386,
      "step": 92976
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.8682824969291687,
      "learning_rate": 0.00038952567614227015,
      "loss": 2.3684,
      "step": 92977
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.8517475128173828,
      "learning_rate": 0.0003895217719582872,
      "loss": 2.2697,
      "step": 92978
    },
    {
      "epoch": 1.21,
      "grad_norm": 1.0481880903244019,
      "learning_rate": 0.0003895178677576605,
      "loss": 2.1262,
      "step": 92979
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.7932940721511841,
      "learning_rate": 0.0003895139635403907,
      "loss": 1.846,
      "step": 92980
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.8075224161148071,
      "learning_rate": 0.00038951005930647865,
      "loss": 2.1469,
      "step": 92981
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.9675930142402649,
      "learning_rate": 0.0003895061550559249,
      "loss": 2.3773,
      "step": 92982
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.880054771900177,
      "learning_rate": 0.00038950225078873034,
      "loss": 2.1853,
      "step": 92983
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.9166305661201477,
      "learning_rate": 0.0003894983465048956,
      "loss": 2.108,
      "step": 92984
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.8438926935195923,
      "learning_rate": 0.0003894944422044214,
      "loss": 2.2004,
      "step": 92985
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.878273606300354,
      "learning_rate": 0.0003894905378873085,
      "loss": 2.0913,
      "step": 92986
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.8370222449302673,
      "learning_rate": 0.00038948663355355763,
      "loss": 2.104,
      "step": 92987
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.8609746694564819,
      "learning_rate": 0.0003894827292031695,
      "loss": 2.3524,
      "step": 92988
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.8273454308509827,
      "learning_rate": 0.0003894788248361449,
      "loss": 2.223,
      "step": 92989
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.8345036506652832,
      "learning_rate": 0.0003894749204524844,
      "loss": 2.0142,
      "step": 92990
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.7695195078849792,
      "learning_rate": 0.00038947101605218883,
      "loss": 2.1332,
      "step": 92991
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.8987578749656677,
      "learning_rate": 0.000389467111635259,
      "loss": 2.2207,
      "step": 92992
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.7666851282119751,
      "learning_rate": 0.00038946320720169547,
      "loss": 2.2317,
      "step": 92993
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.9042965769767761,
      "learning_rate": 0.00038945930275149904,
      "loss": 1.9216,
      "step": 92994
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.7467396855354309,
      "learning_rate": 0.0003894553982846705,
      "loss": 2.0481,
      "step": 92995
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.9011765718460083,
      "learning_rate": 0.0003894514938012105,
      "loss": 2.121,
      "step": 92996
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.8866257071495056,
      "learning_rate": 0.00038944758930111963,
      "loss": 2.2374,
      "step": 92997
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.8173410892486572,
      "learning_rate": 0.000389443684784399,
      "loss": 1.9814,
      "step": 92998
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.8079481720924377,
      "learning_rate": 0.00038943978025104893,
      "loss": 2.0709,
      "step": 92999
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.9776856303215027,
      "learning_rate": 0.0003894358757010704,
      "loss": 2.1811,
      "step": 93000
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.8665763735771179,
      "learning_rate": 0.000389431971134464,
      "loss": 2.3306,
      "step": 93001
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.9651163816452026,
      "learning_rate": 0.00038942806655123055,
      "loss": 2.3279,
      "step": 93002
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.8092299699783325,
      "learning_rate": 0.0003894241619513707,
      "loss": 1.9738,
      "step": 93003
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.8854440450668335,
      "learning_rate": 0.00038942025733488525,
      "loss": 2.4568,
      "step": 93004
    },
    {
      "epoch": 1.21,
      "grad_norm": 1.0098382234573364,
      "learning_rate": 0.00038941635270177486,
      "loss": 2.1234,
      "step": 93005
    },
    {
      "epoch": 1.21,
      "grad_norm": 1.0368131399154663,
      "learning_rate": 0.00038941244805204024,
      "loss": 1.9886,
      "step": 93006
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.8174501061439514,
      "learning_rate": 0.00038940854338568226,
      "loss": 2.3184,
      "step": 93007
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.865307629108429,
      "learning_rate": 0.00038940463870270145,
      "loss": 2.2089,
      "step": 93008
    },
    {
      "epoch": 1.21,
      "grad_norm": 1.0941194295883179,
      "learning_rate": 0.0003894007340030987,
      "loss": 1.827,
      "step": 93009
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.9255127906799316,
      "learning_rate": 0.00038939682928687465,
      "loss": 2.2313,
      "step": 93010
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.8265114426612854,
      "learning_rate": 0.00038939292455403,
      "loss": 2.0046,
      "step": 93011
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.9374824166297913,
      "learning_rate": 0.00038938901980456556,
      "loss": 2.1262,
      "step": 93012
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.9951289296150208,
      "learning_rate": 0.000389385115038482,
      "loss": 2.2828,
      "step": 93013
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.9011035561561584,
      "learning_rate": 0.00038938121025578,
      "loss": 2.2686,
      "step": 93014
    },
    {
      "epoch": 1.21,
      "grad_norm": 1.061347246170044,
      "learning_rate": 0.0003893773054564604,
      "loss": 2.2086,
      "step": 93015
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.8968451023101807,
      "learning_rate": 0.00038937340064052394,
      "loss": 2.1629,
      "step": 93016
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.8387722373008728,
      "learning_rate": 0.00038936949580797126,
      "loss": 1.9907,
      "step": 93017
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.8432427048683167,
      "learning_rate": 0.00038936559095880297,
      "loss": 2.1218,
      "step": 93018
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.9336920380592346,
      "learning_rate": 0.00038936168609302014,
      "loss": 2.0448,
      "step": 93019
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.8573054671287537,
      "learning_rate": 0.0003893577812106231,
      "loss": 2.2891,
      "step": 93020
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.8325388431549072,
      "learning_rate": 0.0003893538763116128,
      "loss": 2.1292,
      "step": 93021
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.8403118848800659,
      "learning_rate": 0.00038934997139598995,
      "loss": 2.3194,
      "step": 93022
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.825423538684845,
      "learning_rate": 0.00038934606646375523,
      "loss": 2.1677,
      "step": 93023
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.841934859752655,
      "learning_rate": 0.00038934216151490945,
      "loss": 2.1428,
      "step": 93024
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.8404172658920288,
      "learning_rate": 0.00038933825654945326,
      "loss": 2.0214,
      "step": 93025
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.7896185517311096,
      "learning_rate": 0.00038933435156738737,
      "loss": 2.0768,
      "step": 93026
    },
    {
      "epoch": 1.21,
      "grad_norm": 1.082136869430542,
      "learning_rate": 0.0003893304465687125,
      "loss": 2.261,
      "step": 93027
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.8024260997772217,
      "learning_rate": 0.00038932654155342955,
      "loss": 2.178,
      "step": 93028
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.9015231132507324,
      "learning_rate": 0.00038932263652153903,
      "loss": 2.2054,
      "step": 93029
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.9729301333427429,
      "learning_rate": 0.0003893187314730417,
      "loss": 2.2663,
      "step": 93030
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.9301193356513977,
      "learning_rate": 0.0003893148264079384,
      "loss": 2.1933,
      "step": 93031
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.8472633957862854,
      "learning_rate": 0.00038931092132622974,
      "loss": 1.8406,
      "step": 93032
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.8274227976799011,
      "learning_rate": 0.0003893070162279165,
      "loss": 2.1437,
      "step": 93033
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.9783489108085632,
      "learning_rate": 0.00038930311111299947,
      "loss": 2.1811,
      "step": 93034
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.8011177182197571,
      "learning_rate": 0.0003892992059814792,
      "loss": 2.287,
      "step": 93035
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.873485803604126,
      "learning_rate": 0.0003892953008333566,
      "loss": 2.2369,
      "step": 93036
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.8733351826667786,
      "learning_rate": 0.00038929139566863224,
      "loss": 2.0525,
      "step": 93037
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.9022006392478943,
      "learning_rate": 0.000389287490487307,
      "loss": 2.0456,
      "step": 93038
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.8695530295372009,
      "learning_rate": 0.0003892835852893815,
      "loss": 2.3089,
      "step": 93039
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.9528171420097351,
      "learning_rate": 0.00038927968007485653,
      "loss": 2.3176,
      "step": 93040
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.9063896536827087,
      "learning_rate": 0.00038927577484373276,
      "loss": 2.2523,
      "step": 93041
    },
    {
      "epoch": 1.21,
      "grad_norm": 1.0036417245864868,
      "learning_rate": 0.0003892718695960109,
      "loss": 2.2885,
      "step": 93042
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.9344175457954407,
      "learning_rate": 0.00038926796433169175,
      "loss": 2.0255,
      "step": 93043
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.9231246709823608,
      "learning_rate": 0.00038926405905077594,
      "loss": 1.9967,
      "step": 93044
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.861947238445282,
      "learning_rate": 0.0003892601537532643,
      "loss": 2.1842,
      "step": 93045
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.8731759786605835,
      "learning_rate": 0.00038925624843915763,
      "loss": 2.394,
      "step": 93046
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.9622982144355774,
      "learning_rate": 0.0003892523431084564,
      "loss": 2.3331,
      "step": 93047
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.8412367701530457,
      "learning_rate": 0.00038924843776116144,
      "loss": 2.2641,
      "step": 93048
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.8433009386062622,
      "learning_rate": 0.0003892445323972736,
      "loss": 2.1183,
      "step": 93049
    },
    {
      "epoch": 1.21,
      "grad_norm": 1.3167023658752441,
      "learning_rate": 0.00038924062701679353,
      "loss": 2.302,
      "step": 93050
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.975882887840271,
      "learning_rate": 0.0003892367216197219,
      "loss": 2.175,
      "step": 93051
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.8250065445899963,
      "learning_rate": 0.00038923281620605943,
      "loss": 2.2625,
      "step": 93052
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.9711488485336304,
      "learning_rate": 0.00038922891077580697,
      "loss": 2.2491,
      "step": 93053
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.9678226113319397,
      "learning_rate": 0.0003892250053289652,
      "loss": 2.0395,
      "step": 93054
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.9417842626571655,
      "learning_rate": 0.0003892210998655347,
      "loss": 2.4225,
      "step": 93055
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.8170856833457947,
      "learning_rate": 0.00038921719438551633,
      "loss": 2.0966,
      "step": 93056
    },
    {
      "epoch": 1.21,
      "grad_norm": 1.0980676412582397,
      "learning_rate": 0.00038921328888891096,
      "loss": 2.1062,
      "step": 93057
    },
    {
      "epoch": 1.21,
      "grad_norm": 1.149579644203186,
      "learning_rate": 0.000389209383375719,
      "loss": 2.14,
      "step": 93058
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.7709311246871948,
      "learning_rate": 0.00038920547784594136,
      "loss": 2.01,
      "step": 93059
    },
    {
      "epoch": 1.21,
      "grad_norm": 1.0762971639633179,
      "learning_rate": 0.00038920157229957876,
      "loss": 2.2725,
      "step": 93060
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.9120942950248718,
      "learning_rate": 0.00038919766673663187,
      "loss": 2.1785,
      "step": 93061
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.8205544352531433,
      "learning_rate": 0.0003891937611571015,
      "loss": 2.0465,
      "step": 93062
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.8510544896125793,
      "learning_rate": 0.00038918985556098834,
      "loss": 2.2909,
      "step": 93063
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.9571109414100647,
      "learning_rate": 0.000389185949948293,
      "loss": 2.3066,
      "step": 93064
    },
    {
      "epoch": 1.21,
      "grad_norm": 1.025400996208191,
      "learning_rate": 0.0003891820443190164,
      "loss": 2.0931,
      "step": 93065
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.8007238507270813,
      "learning_rate": 0.00038917813867315917,
      "loss": 2.199,
      "step": 93066
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.8914167284965515,
      "learning_rate": 0.000389174233010722,
      "loss": 2.0937,
      "step": 93067
    },
    {
      "epoch": 1.21,
      "grad_norm": 1.22218918800354,
      "learning_rate": 0.0003891703273317057,
      "loss": 2.3416,
      "step": 93068
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.9174089431762695,
      "learning_rate": 0.00038916642163611093,
      "loss": 2.2605,
      "step": 93069
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.885502278804779,
      "learning_rate": 0.00038916251592393836,
      "loss": 2.0839,
      "step": 93070
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.9100987315177917,
      "learning_rate": 0.000389158610195189,
      "loss": 2.3278,
      "step": 93071
    },
    {
      "epoch": 1.21,
      "grad_norm": 1.053828239440918,
      "learning_rate": 0.0003891547044498632,
      "loss": 2.2148,
      "step": 93072
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.8183104395866394,
      "learning_rate": 0.0003891507986879619,
      "loss": 2.0335,
      "step": 93073
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.8443206548690796,
      "learning_rate": 0.00038914689290948587,
      "loss": 2.0407,
      "step": 93074
    },
    {
      "epoch": 1.21,
      "grad_norm": 1.025187611579895,
      "learning_rate": 0.00038914298711443564,
      "loss": 2.1177,
      "step": 93075
    },
    {
      "epoch": 1.21,
      "grad_norm": 1.0448215007781982,
      "learning_rate": 0.000389139081302812,
      "loss": 2.1686,
      "step": 93076
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.8774629831314087,
      "learning_rate": 0.0003891351754746159,
      "loss": 2.1538,
      "step": 93077
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.9279550909996033,
      "learning_rate": 0.0003891312696298477,
      "loss": 2.0722,
      "step": 93078
    },
    {
      "epoch": 1.21,
      "grad_norm": 1.1589972972869873,
      "learning_rate": 0.00038912736376850844,
      "loss": 2.4418,
      "step": 93079
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.8980478644371033,
      "learning_rate": 0.00038912345789059866,
      "loss": 2.0565,
      "step": 93080
    },
    {
      "epoch": 1.21,
      "grad_norm": 1.0897212028503418,
      "learning_rate": 0.00038911955199611917,
      "loss": 1.7984,
      "step": 93081
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.8807510137557983,
      "learning_rate": 0.00038911564608507063,
      "loss": 2.1211,
      "step": 93082
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.808996319770813,
      "learning_rate": 0.0003891117401574539,
      "loss": 2.1127,
      "step": 93083
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.9227397441864014,
      "learning_rate": 0.00038910783421326956,
      "loss": 2.2761,
      "step": 93084
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.8228713274002075,
      "learning_rate": 0.0003891039282525183,
      "loss": 1.9615,
      "step": 93085
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.8575473427772522,
      "learning_rate": 0.0003891000222752012,
      "loss": 2.1986,
      "step": 93086
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.9398495554924011,
      "learning_rate": 0.00038909611628131847,
      "loss": 2.3201,
      "step": 93087
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.7969574332237244,
      "learning_rate": 0.0003890922102708712,
      "loss": 2.1952,
      "step": 93088
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.8131879568099976,
      "learning_rate": 0.00038908830424386003,
      "loss": 2.1364,
      "step": 93089
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.803418755531311,
      "learning_rate": 0.00038908439820028556,
      "loss": 2.1658,
      "step": 93090
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.9634210467338562,
      "learning_rate": 0.00038908049214014866,
      "loss": 2.1942,
      "step": 93091
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.7633801698684692,
      "learning_rate": 0.0003890765860634501,
      "loss": 2.1523,
      "step": 93092
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.8888912200927734,
      "learning_rate": 0.00038907267997019036,
      "loss": 2.354,
      "step": 93093
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.8328922390937805,
      "learning_rate": 0.00038906877386037043,
      "loss": 1.9611,
      "step": 93094
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.8562024831771851,
      "learning_rate": 0.00038906486773399094,
      "loss": 2.2427,
      "step": 93095
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.7872514128684998,
      "learning_rate": 0.00038906096159105265,
      "loss": 2.0674,
      "step": 93096
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.8023411631584167,
      "learning_rate": 0.00038905705543155615,
      "loss": 1.9539,
      "step": 93097
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.8115645051002502,
      "learning_rate": 0.0003890531492555023,
      "loss": 2.009,
      "step": 93098
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.8327994346618652,
      "learning_rate": 0.00038904924306289184,
      "loss": 2.0583,
      "step": 93099
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.8744221329689026,
      "learning_rate": 0.0003890453368537254,
      "loss": 2.1083,
      "step": 93100
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.9349684119224548,
      "learning_rate": 0.00038904143062800376,
      "loss": 1.9022,
      "step": 93101
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.8209947347640991,
      "learning_rate": 0.00038903752438572756,
      "loss": 2.2002,
      "step": 93102
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.8560273051261902,
      "learning_rate": 0.00038903361812689765,
      "loss": 2.2367,
      "step": 93103
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.8591229915618896,
      "learning_rate": 0.0003890297118515148,
      "loss": 2.0823,
      "step": 93104
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.8277791738510132,
      "learning_rate": 0.0003890258055595796,
      "loss": 2.1195,
      "step": 93105
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.828644335269928,
      "learning_rate": 0.00038902189925109267,
      "loss": 2.1663,
      "step": 93106
    },
    {
      "epoch": 1.21,
      "grad_norm": 1.184604287147522,
      "learning_rate": 0.0003890179929260551,
      "loss": 2.2105,
      "step": 93107
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.876467764377594,
      "learning_rate": 0.0003890140865844673,
      "loss": 2.3883,
      "step": 93108
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.9758866429328918,
      "learning_rate": 0.0003890101802263301,
      "loss": 2.0969,
      "step": 93109
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.8115234375,
      "learning_rate": 0.0003890062738516443,
      "loss": 2.1776,
      "step": 93110
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.8497124314308167,
      "learning_rate": 0.00038900236746041046,
      "loss": 2.1942,
      "step": 93111
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.8330417275428772,
      "learning_rate": 0.00038899846105262944,
      "loss": 2.2702,
      "step": 93112
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.9773218035697937,
      "learning_rate": 0.0003889945546283019,
      "loss": 2.4009,
      "step": 93113
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.8416180610656738,
      "learning_rate": 0.0003889906481874286,
      "loss": 2.094,
      "step": 93114
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.8666987419128418,
      "learning_rate": 0.0003889867417300103,
      "loss": 2.299,
      "step": 93115
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.8928297162055969,
      "learning_rate": 0.0003889828352560477,
      "loss": 2.2283,
      "step": 93116
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.8956851363182068,
      "learning_rate": 0.0003889789287655415,
      "loss": 2.1154,
      "step": 93117
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.8588771224021912,
      "learning_rate": 0.00038897502225849237,
      "loss": 2.3507,
      "step": 93118
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.8178935050964355,
      "learning_rate": 0.0003889711157349011,
      "loss": 2.0913,
      "step": 93119
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.8508459329605103,
      "learning_rate": 0.00038896720919476854,
      "loss": 2.1137,
      "step": 93120
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.8324509263038635,
      "learning_rate": 0.0003889633026380952,
      "loss": 1.9826,
      "step": 93121
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.985071063041687,
      "learning_rate": 0.00038895939606488193,
      "loss": 2.0745,
      "step": 93122
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.8145065307617188,
      "learning_rate": 0.00038895548947512945,
      "loss": 2.2194,
      "step": 93123
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.835051953792572,
      "learning_rate": 0.0003889515828688384,
      "loss": 1.9523,
      "step": 93124
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.8026615381240845,
      "learning_rate": 0.0003889476762460096,
      "loss": 2.0265,
      "step": 93125
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.858721137046814,
      "learning_rate": 0.0003889437696066438,
      "loss": 2.2294,
      "step": 93126
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.9673652052879333,
      "learning_rate": 0.0003889398629507416,
      "loss": 2.3644,
      "step": 93127
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.7852728366851807,
      "learning_rate": 0.0003889359562783038,
      "loss": 1.8191,
      "step": 93128
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.8797996044158936,
      "learning_rate": 0.0003889320495893312,
      "loss": 2.2634,
      "step": 93129
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.8957793712615967,
      "learning_rate": 0.0003889281428838244,
      "loss": 2.2029,
      "step": 93130
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.8034120798110962,
      "learning_rate": 0.0003889242361617842,
      "loss": 2.177,
      "step": 93131
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.7965644001960754,
      "learning_rate": 0.0003889203294232113,
      "loss": 2.0442,
      "step": 93132
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.9919873476028442,
      "learning_rate": 0.00038891642266810646,
      "loss": 2.3151,
      "step": 93133
    },
    {
      "epoch": 1.21,
      "grad_norm": 1.0392074584960938,
      "learning_rate": 0.00038891251589647034,
      "loss": 2.1543,
      "step": 93134
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.9035704135894775,
      "learning_rate": 0.00038890860910830376,
      "loss": 2.213,
      "step": 93135
    },
    {
      "epoch": 1.21,
      "grad_norm": 1.1641852855682373,
      "learning_rate": 0.0003889047023036073,
      "loss": 2.0443,
      "step": 93136
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.8658809661865234,
      "learning_rate": 0.0003889007954823818,
      "loss": 1.9598,
      "step": 93137
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.8075590133666992,
      "learning_rate": 0.000388896888644628,
      "loss": 2.0109,
      "step": 93138
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.8957878947257996,
      "learning_rate": 0.0003888929817903466,
      "loss": 2.0539,
      "step": 93139
    },
    {
      "epoch": 1.21,
      "grad_norm": 1.1111555099487305,
      "learning_rate": 0.00038888907491953825,
      "loss": 1.8495,
      "step": 93140
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.7791529297828674,
      "learning_rate": 0.00038888516803220377,
      "loss": 2.0438,
      "step": 93141
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.8860107064247131,
      "learning_rate": 0.00038888126112834393,
      "loss": 2.2849,
      "step": 93142
    },
    {
      "epoch": 1.21,
      "grad_norm": 1.054064393043518,
      "learning_rate": 0.00038887735420795933,
      "loss": 2.2552,
      "step": 93143
    },
    {
      "epoch": 1.21,
      "grad_norm": 1.0350042581558228,
      "learning_rate": 0.00038887344727105074,
      "loss": 2.3908,
      "step": 93144
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.9361807703971863,
      "learning_rate": 0.0003888695403176189,
      "loss": 1.9506,
      "step": 93145
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.8903665542602539,
      "learning_rate": 0.0003888656333476645,
      "loss": 2.1434,
      "step": 93146
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.9218287467956543,
      "learning_rate": 0.0003888617263611884,
      "loss": 2.1704,
      "step": 93147
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.8803573250770569,
      "learning_rate": 0.0003888578193581911,
      "loss": 2.2721,
      "step": 93148
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.7948148846626282,
      "learning_rate": 0.0003888539123386736,
      "loss": 2.0084,
      "step": 93149
    },
    {
      "epoch": 1.21,
      "grad_norm": 1.1111973524093628,
      "learning_rate": 0.0003888500053026364,
      "loss": 2.225,
      "step": 93150
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.7837857604026794,
      "learning_rate": 0.0003888460982500803,
      "loss": 2.2483,
      "step": 93151
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.7959555983543396,
      "learning_rate": 0.000388842191181006,
      "loss": 2.2338,
      "step": 93152
    },
    {
      "epoch": 1.21,
      "grad_norm": 1.0852203369140625,
      "learning_rate": 0.0003888382840954144,
      "loss": 2.1546,
      "step": 93153
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.8313446640968323,
      "learning_rate": 0.000388834376993306,
      "loss": 2.2323,
      "step": 93154
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.7969337701797485,
      "learning_rate": 0.00038883046987468154,
      "loss": 2.2208,
      "step": 93155
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.8209498524665833,
      "learning_rate": 0.00038882656273954197,
      "loss": 2.2399,
      "step": 93156
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.846606433391571,
      "learning_rate": 0.0003888226555878878,
      "loss": 2.347,
      "step": 93157
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.8968396186828613,
      "learning_rate": 0.0003888187484197197,
      "loss": 2.2065,
      "step": 93158
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.8467826247215271,
      "learning_rate": 0.0003888148412350387,
      "loss": 2.1111,
      "step": 93159
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.8723135590553284,
      "learning_rate": 0.00038881093403384524,
      "loss": 2.1274,
      "step": 93160
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.8394969701766968,
      "learning_rate": 0.0003888070268161402,
      "loss": 2.1113,
      "step": 93161
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.7985907793045044,
      "learning_rate": 0.00038880311958192423,
      "loss": 2.1457,
      "step": 93162
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.8001049160957336,
      "learning_rate": 0.00038879921233119804,
      "loss": 2.2531,
      "step": 93163
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.7950908541679382,
      "learning_rate": 0.00038879530506396253,
      "loss": 2.2394,
      "step": 93164
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.7737933993339539,
      "learning_rate": 0.0003887913977802182,
      "loss": 1.9826,
      "step": 93165
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.7932347655296326,
      "learning_rate": 0.00038878749047996586,
      "loss": 2.1476,
      "step": 93166
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.8585358262062073,
      "learning_rate": 0.0003887835831632063,
      "loss": 2.1079,
      "step": 93167
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.8013097047805786,
      "learning_rate": 0.0003887796758299402,
      "loss": 2.1231,
      "step": 93168
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.8444375395774841,
      "learning_rate": 0.00038877576848016823,
      "loss": 2.3407,
      "step": 93169
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.8495272397994995,
      "learning_rate": 0.0003887718611138912,
      "loss": 2.3434,
      "step": 93170
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.9222646355628967,
      "learning_rate": 0.0003887679537311098,
      "loss": 2.2973,
      "step": 93171
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.8618137836456299,
      "learning_rate": 0.00038876404633182476,
      "loss": 2.2426,
      "step": 93172
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.9077682495117188,
      "learning_rate": 0.0003887601389160368,
      "loss": 2.2275,
      "step": 93173
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.8916158080101013,
      "learning_rate": 0.00038875623148374675,
      "loss": 2.1863,
      "step": 93174
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.8049255013465881,
      "learning_rate": 0.00038875232403495514,
      "loss": 2.1443,
      "step": 93175
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.8580437302589417,
      "learning_rate": 0.00038874841656966273,
      "loss": 2.1215,
      "step": 93176
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.8707733750343323,
      "learning_rate": 0.0003887445090878705,
      "loss": 2.2076,
      "step": 93177
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.8731000423431396,
      "learning_rate": 0.0003887406015895789,
      "loss": 2.2449,
      "step": 93178
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.8578161001205444,
      "learning_rate": 0.0003887366940747887,
      "loss": 2.0674,
      "step": 93179
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.7559298872947693,
      "learning_rate": 0.0003887327865435007,
      "loss": 2.1796,
      "step": 93180
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.8438846468925476,
      "learning_rate": 0.0003887288789957156,
      "loss": 2.3001,
      "step": 93181
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.8463073968887329,
      "learning_rate": 0.0003887249714314341,
      "loss": 2.4734,
      "step": 93182
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.8742703795433044,
      "learning_rate": 0.00038872106385065693,
      "loss": 2.0468,
      "step": 93183
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.797956645488739,
      "learning_rate": 0.00038871715625338497,
      "loss": 2.2627,
      "step": 93184
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.9218577146530151,
      "learning_rate": 0.00038871324863961877,
      "loss": 2.0897,
      "step": 93185
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.7721000909805298,
      "learning_rate": 0.0003887093410093591,
      "loss": 2.0662,
      "step": 93186
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.8249897360801697,
      "learning_rate": 0.0003887054333626067,
      "loss": 2.0936,
      "step": 93187
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.9568586945533752,
      "learning_rate": 0.00038870152569936217,
      "loss": 2.2791,
      "step": 93188
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.8680073022842407,
      "learning_rate": 0.00038869761801962644,
      "loss": 2.1417,
      "step": 93189
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.8629295229911804,
      "learning_rate": 0.00038869371032340016,
      "loss": 2.305,
      "step": 93190
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.8670880794525146,
      "learning_rate": 0.00038868980261068403,
      "loss": 2.1172,
      "step": 93191
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.7698138356208801,
      "learning_rate": 0.00038868589488147876,
      "loss": 2.1756,
      "step": 93192
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.8283542394638062,
      "learning_rate": 0.00038868198713578516,
      "loss": 2.2328,
      "step": 93193
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.7885751128196716,
      "learning_rate": 0.0003886780793736039,
      "loss": 2.2496,
      "step": 93194
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.8440084457397461,
      "learning_rate": 0.0003886741715949357,
      "loss": 2.2447,
      "step": 93195
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.7478704452514648,
      "learning_rate": 0.00038867026379978126,
      "loss": 2.3103,
      "step": 93196
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.998228132724762,
      "learning_rate": 0.0003886663559881414,
      "loss": 2.0068,
      "step": 93197
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.8670225739479065,
      "learning_rate": 0.00038866244816001676,
      "loss": 2.2287,
      "step": 93198
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.8992283940315247,
      "learning_rate": 0.00038865854031540814,
      "loss": 2.1564,
      "step": 93199
    },
    {
      "epoch": 1.21,
      "grad_norm": 1.005973219871521,
      "learning_rate": 0.00038865463245431613,
      "loss": 2.2785,
      "step": 93200
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.9661834239959717,
      "learning_rate": 0.0003886507245767416,
      "loss": 1.9652,
      "step": 93201
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.8595868945121765,
      "learning_rate": 0.0003886468166826853,
      "loss": 2.2695,
      "step": 93202
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.8695574998855591,
      "learning_rate": 0.00038864290877214784,
      "loss": 2.0534,
      "step": 93203
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.874333381652832,
      "learning_rate": 0.0003886390008451299,
      "loss": 2.2949,
      "step": 93204
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.8374610543251038,
      "learning_rate": 0.0003886350929016324,
      "loss": 2.0378,
      "step": 93205
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.8601245284080505,
      "learning_rate": 0.0003886311849416559,
      "loss": 2.3724,
      "step": 93206
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.8622042536735535,
      "learning_rate": 0.0003886272769652012,
      "loss": 1.9048,
      "step": 93207
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.8830162882804871,
      "learning_rate": 0.0003886233689722691,
      "loss": 2.2247,
      "step": 93208
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.7968443036079407,
      "learning_rate": 0.00038861946096286015,
      "loss": 2.105,
      "step": 93209
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.7982190251350403,
      "learning_rate": 0.00038861555293697516,
      "loss": 2.0796,
      "step": 93210
    },
    {
      "epoch": 1.21,
      "grad_norm": 1.0817515850067139,
      "learning_rate": 0.0003886116448946149,
      "loss": 2.0901,
      "step": 93211
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.8120672702789307,
      "learning_rate": 0.0003886077368357801,
      "loss": 1.9043,
      "step": 93212
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.8567584156990051,
      "learning_rate": 0.0003886038287604714,
      "loss": 2.1848,
      "step": 93213
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.8067266345024109,
      "learning_rate": 0.0003885999206686896,
      "loss": 1.9135,
      "step": 93214
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.877103865146637,
      "learning_rate": 0.0003885960125604354,
      "loss": 2.3504,
      "step": 93215
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.9458510875701904,
      "learning_rate": 0.0003885921044357095,
      "loss": 2.3554,
      "step": 93216
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.8678216338157654,
      "learning_rate": 0.0003885881962945127,
      "loss": 2.209,
      "step": 93217
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.8446016907691956,
      "learning_rate": 0.0003885842881368457,
      "loss": 1.983,
      "step": 93218
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.8194669485092163,
      "learning_rate": 0.0003885803799627091,
      "loss": 2.1134,
      "step": 93219
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.7637121081352234,
      "learning_rate": 0.0003885764717721038,
      "loss": 2.1161,
      "step": 93220
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.8591048121452332,
      "learning_rate": 0.00038857256356503053,
      "loss": 2.1438,
      "step": 93221
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.896589457988739,
      "learning_rate": 0.0003885686553414898,
      "loss": 2.2788,
      "step": 93222
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.8055634498596191,
      "learning_rate": 0.00038856474710148266,
      "loss": 2.302,
      "step": 93223
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.8560386300086975,
      "learning_rate": 0.00038856083884500955,
      "loss": 2.2351,
      "step": 93224
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.8384794592857361,
      "learning_rate": 0.0003885569305720713,
      "loss": 2.173,
      "step": 93225
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.8210586309432983,
      "learning_rate": 0.0003885530222826687,
      "loss": 2.4355,
      "step": 93226
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.9147976040840149,
      "learning_rate": 0.0003885491139768024,
      "loss": 2.2481,
      "step": 93227
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.9323984980583191,
      "learning_rate": 0.00038854520565447307,
      "loss": 2.0859,
      "step": 93228
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.9802724719047546,
      "learning_rate": 0.00038854129731568163,
      "loss": 2.1291,
      "step": 93229
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.8808222413063049,
      "learning_rate": 0.00038853738896042864,
      "loss": 2.215,
      "step": 93230
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.8853110074996948,
      "learning_rate": 0.00038853348058871486,
      "loss": 2.2309,
      "step": 93231
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.9276121854782104,
      "learning_rate": 0.0003885295722005411,
      "loss": 2.1708,
      "step": 93232
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.9767367839813232,
      "learning_rate": 0.00038852566379590797,
      "loss": 2.1014,
      "step": 93233
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.9642325043678284,
      "learning_rate": 0.00038852175537481626,
      "loss": 2.2278,
      "step": 93234
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.9185264706611633,
      "learning_rate": 0.00038851784693726664,
      "loss": 2.044,
      "step": 93235
    },
    {
      "epoch": 1.21,
      "grad_norm": 1.0926694869995117,
      "learning_rate": 0.0003885139384832599,
      "loss": 2.2131,
      "step": 93236
    },
    {
      "epoch": 1.21,
      "grad_norm": 1.0740292072296143,
      "learning_rate": 0.00038851003001279683,
      "loss": 2.1083,
      "step": 93237
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.8897374272346497,
      "learning_rate": 0.000388506121525878,
      "loss": 2.1669,
      "step": 93238
    },
    {
      "epoch": 1.21,
      "grad_norm": 1.0546538829803467,
      "learning_rate": 0.00038850221302250425,
      "loss": 2.2992,
      "step": 93239
    },
    {
      "epoch": 1.21,
      "grad_norm": 1.0118930339813232,
      "learning_rate": 0.0003884983045026762,
      "loss": 2.0702,
      "step": 93240
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.8894206285476685,
      "learning_rate": 0.0003884943959663948,
      "loss": 2.1309,
      "step": 93241
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.9320549368858337,
      "learning_rate": 0.00038849048741366045,
      "loss": 2.0574,
      "step": 93242
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.8194547295570374,
      "learning_rate": 0.0003884865788444741,
      "loss": 2.0991,
      "step": 93243
    },
    {
      "epoch": 1.21,
      "grad_norm": 1.0090962648391724,
      "learning_rate": 0.00038848267025883644,
      "loss": 1.966,
      "step": 93244
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.8112492561340332,
      "learning_rate": 0.00038847876165674813,
      "loss": 1.9642,
      "step": 93245
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.8633599281311035,
      "learning_rate": 0.00038847485303821004,
      "loss": 2.1459,
      "step": 93246
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.9155470728874207,
      "learning_rate": 0.00038847094440322275,
      "loss": 2.2239,
      "step": 93247
    },
    {
      "epoch": 1.21,
      "grad_norm": 1.0422754287719727,
      "learning_rate": 0.00038846703575178703,
      "loss": 2.2662,
      "step": 93248
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.9621487855911255,
      "learning_rate": 0.00038846312708390365,
      "loss": 2.3362,
      "step": 93249
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.8800550699234009,
      "learning_rate": 0.0003884592183995733,
      "loss": 2.19,
      "step": 93250
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.8893261551856995,
      "learning_rate": 0.0003884553096987967,
      "loss": 2.1933,
      "step": 93251
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.8648464679718018,
      "learning_rate": 0.00038845140098157466,
      "loss": 2.1591,
      "step": 93252
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.8486631512641907,
      "learning_rate": 0.0003884474922479077,
      "loss": 2.2357,
      "step": 93253
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.8339464664459229,
      "learning_rate": 0.00038844358349779677,
      "loss": 2.1349,
      "step": 93254
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.9021934866905212,
      "learning_rate": 0.0003884396747312425,
      "loss": 2.2111,
      "step": 93255
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.8391464948654175,
      "learning_rate": 0.0003884357659482456,
      "loss": 2.0724,
      "step": 93256
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.7671005725860596,
      "learning_rate": 0.0003884318571488069,
      "loss": 2.0455,
      "step": 93257
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.9445789456367493,
      "learning_rate": 0.000388427948332927,
      "loss": 2.2371,
      "step": 93258
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.841101884841919,
      "learning_rate": 0.00038842403950060657,
      "loss": 2.1505,
      "step": 93259
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.8461519479751587,
      "learning_rate": 0.00038842013065184665,
      "loss": 2.1223,
      "step": 93260
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.9198477864265442,
      "learning_rate": 0.0003884162217866476,
      "loss": 2.4308,
      "step": 93261
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.7926951050758362,
      "learning_rate": 0.00038841231290501026,
      "loss": 2.2001,
      "step": 93262
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.8520049452781677,
      "learning_rate": 0.00038840840400693557,
      "loss": 2.0747,
      "step": 93263
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.8887084126472473,
      "learning_rate": 0.000388404495092424,
      "loss": 2.1542,
      "step": 93264
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.8435636758804321,
      "learning_rate": 0.0003884005861614763,
      "loss": 2.164,
      "step": 93265
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.9441483616828918,
      "learning_rate": 0.0003883966772140934,
      "loss": 2.1115,
      "step": 93266
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.784656286239624,
      "learning_rate": 0.00038839276825027583,
      "loss": 2.2104,
      "step": 93267
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.817513644695282,
      "learning_rate": 0.00038838885927002436,
      "loss": 2.0977,
      "step": 93268
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.7996295094490051,
      "learning_rate": 0.00038838495027333975,
      "loss": 2.192,
      "step": 93269
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.8318368196487427,
      "learning_rate": 0.00038838104126022265,
      "loss": 1.9003,
      "step": 93270
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.8441727757453918,
      "learning_rate": 0.00038837713223067397,
      "loss": 2.0989,
      "step": 93271
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.9363895058631897,
      "learning_rate": 0.0003883732231846942,
      "loss": 2.2542,
      "step": 93272
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.7918187379837036,
      "learning_rate": 0.0003883693141222843,
      "loss": 2.2493,
      "step": 93273
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.7784909009933472,
      "learning_rate": 0.0003883654050434447,
      "loss": 2.101,
      "step": 93274
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.8357552886009216,
      "learning_rate": 0.0003883614959481765,
      "loss": 2.1399,
      "step": 93275
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.8949214816093445,
      "learning_rate": 0.0003883575868364801,
      "loss": 2.1993,
      "step": 93276
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.8662218451499939,
      "learning_rate": 0.00038835367770835637,
      "loss": 2.003,
      "step": 93277
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.8030586838722229,
      "learning_rate": 0.0003883497685638061,
      "loss": 2.1534,
      "step": 93278
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.982933521270752,
      "learning_rate": 0.00038834585940282983,
      "loss": 2.4053,
      "step": 93279
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.8738968372344971,
      "learning_rate": 0.00038834195022542845,
      "loss": 2.1639,
      "step": 93280
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.8326554298400879,
      "learning_rate": 0.00038833804103160266,
      "loss": 2.0839,
      "step": 93281
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.8475250005722046,
      "learning_rate": 0.0003883341318213531,
      "loss": 2.0205,
      "step": 93282
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.7748280763626099,
      "learning_rate": 0.0003883302225946806,
      "loss": 2.1426,
      "step": 93283
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.8288679122924805,
      "learning_rate": 0.00038832631335158587,
      "loss": 2.2868,
      "step": 93284
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.839858889579773,
      "learning_rate": 0.0003883224040920695,
      "loss": 2.2359,
      "step": 93285
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.8206257820129395,
      "learning_rate": 0.0003883184948161324,
      "loss": 2.2726,
      "step": 93286
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.8562619090080261,
      "learning_rate": 0.00038831458552377525,
      "loss": 2.1466,
      "step": 93287
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.8628108501434326,
      "learning_rate": 0.00038831067621499876,
      "loss": 2.2983,
      "step": 93288
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.9206447005271912,
      "learning_rate": 0.0003883067668898036,
      "loss": 2.2132,
      "step": 93289
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.8073323965072632,
      "learning_rate": 0.0003883028575481906,
      "loss": 2.1299,
      "step": 93290
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.8671293258666992,
      "learning_rate": 0.00038829894819016037,
      "loss": 2.0994,
      "step": 93291
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.8103281855583191,
      "learning_rate": 0.0003882950388157137,
      "loss": 2.2635,
      "step": 93292
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.919941782951355,
      "learning_rate": 0.0003882911294248514,
      "loss": 2.1944,
      "step": 93293
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.787355899810791,
      "learning_rate": 0.00038828722001757406,
      "loss": 2.2907,
      "step": 93294
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.8716183304786682,
      "learning_rate": 0.0003882833105938824,
      "loss": 1.995,
      "step": 93295
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.8517023324966431,
      "learning_rate": 0.0003882794011537773,
      "loss": 2.3913,
      "step": 93296
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.9420843124389648,
      "learning_rate": 0.0003882754916972593,
      "loss": 2.1488,
      "step": 93297
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.8433448076248169,
      "learning_rate": 0.0003882715822243293,
      "loss": 2.1621,
      "step": 93298
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.8775234818458557,
      "learning_rate": 0.0003882676727349879,
      "loss": 2.0291,
      "step": 93299
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.8987282514572144,
      "learning_rate": 0.00038826376322923596,
      "loss": 1.9678,
      "step": 93300
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.9494194388389587,
      "learning_rate": 0.00038825985370707396,
      "loss": 2.2605,
      "step": 93301
    },
    {
      "epoch": 1.21,
      "grad_norm": 1.1425799131393433,
      "learning_rate": 0.00038825594416850293,
      "loss": 2.2636,
      "step": 93302
    },
    {
      "epoch": 1.21,
      "grad_norm": 1.1104004383087158,
      "learning_rate": 0.0003882520346135234,
      "loss": 1.8955,
      "step": 93303
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.9555855393409729,
      "learning_rate": 0.00038824812504213615,
      "loss": 2.1637,
      "step": 93304
    },
    {
      "epoch": 1.21,
      "grad_norm": 1.2254244089126587,
      "learning_rate": 0.0003882442154543419,
      "loss": 2.2244,
      "step": 93305
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.9949620962142944,
      "learning_rate": 0.00038824030585014137,
      "loss": 2.1703,
      "step": 93306
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.9465696811676025,
      "learning_rate": 0.0003882363962295353,
      "loss": 2.2789,
      "step": 93307
    },
    {
      "epoch": 1.21,
      "grad_norm": 1.0310591459274292,
      "learning_rate": 0.00038823248659252454,
      "loss": 2.136,
      "step": 93308
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.7628488540649414,
      "learning_rate": 0.00038822857693910956,
      "loss": 2.1789,
      "step": 93309
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.8267950415611267,
      "learning_rate": 0.0003882246672692912,
      "loss": 2.2386,
      "step": 93310
    },
    {
      "epoch": 1.21,
      "grad_norm": 1.0002286434173584,
      "learning_rate": 0.0003882207575830704,
      "loss": 2.1788,
      "step": 93311
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.9573462605476379,
      "learning_rate": 0.00038821684788044753,
      "loss": 2.4347,
      "step": 93312
    },
    {
      "epoch": 1.21,
      "grad_norm": 0.8137099146842957,
      "learning_rate": 0.00038821293816142347,
      "loss": 2.388,
      "step": 93313
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.9456624984741211,
      "learning_rate": 0.000388209028425999,
      "loss": 1.9923,
      "step": 93314
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.8636379241943359,
      "learning_rate": 0.00038820511867417484,
      "loss": 2.0743,
      "step": 93315
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.9237992763519287,
      "learning_rate": 0.00038820120890595164,
      "loss": 2.2741,
      "step": 93316
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.8670749664306641,
      "learning_rate": 0.0003881972991213302,
      "loss": 2.0779,
      "step": 93317
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.7754112482070923,
      "learning_rate": 0.00038819338932031123,
      "loss": 2.1594,
      "step": 93318
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.9370229840278625,
      "learning_rate": 0.00038818947950289544,
      "loss": 2.2107,
      "step": 93319
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.9393096566200256,
      "learning_rate": 0.00038818556966908354,
      "loss": 1.953,
      "step": 93320
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.8942291736602783,
      "learning_rate": 0.0003881816598188762,
      "loss": 2.0474,
      "step": 93321
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.9146619439125061,
      "learning_rate": 0.0003881777499522744,
      "loss": 2.3053,
      "step": 93322
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.8312810063362122,
      "learning_rate": 0.00038817384006927853,
      "loss": 2.4411,
      "step": 93323
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.779132604598999,
      "learning_rate": 0.0003881699301698896,
      "loss": 2.1841,
      "step": 93324
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.9852145314216614,
      "learning_rate": 0.0003881660202541081,
      "loss": 1.8381,
      "step": 93325
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.8073728084564209,
      "learning_rate": 0.0003881621103219349,
      "loss": 2.2095,
      "step": 93326
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.8487042188644409,
      "learning_rate": 0.0003881582003733707,
      "loss": 2.3215,
      "step": 93327
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.8577512502670288,
      "learning_rate": 0.0003881542904084163,
      "loss": 2.1432,
      "step": 93328
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.8581956028938293,
      "learning_rate": 0.00038815038042707236,
      "loss": 2.3272,
      "step": 93329
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.8572989702224731,
      "learning_rate": 0.0003881464704293395,
      "loss": 2.2787,
      "step": 93330
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.803863525390625,
      "learning_rate": 0.00038814256041521864,
      "loss": 2.2127,
      "step": 93331
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.9010215997695923,
      "learning_rate": 0.00038813865038471034,
      "loss": 2.1429,
      "step": 93332
    },
    {
      "epoch": 1.22,
      "grad_norm": 1.0827051401138306,
      "learning_rate": 0.00038813474033781546,
      "loss": 2.1737,
      "step": 93333
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.7705273628234863,
      "learning_rate": 0.00038813083027453466,
      "loss": 2.1232,
      "step": 93334
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.8372567296028137,
      "learning_rate": 0.0003881269201948686,
      "loss": 2.1933,
      "step": 93335
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.7921773195266724,
      "learning_rate": 0.0003881230100988182,
      "loss": 2.503,
      "step": 93336
    },
    {
      "epoch": 1.22,
      "grad_norm": 1.041408658027649,
      "learning_rate": 0.00038811909998638395,
      "loss": 2.0458,
      "step": 93337
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.7976492047309875,
      "learning_rate": 0.00038811518985756675,
      "loss": 2.2062,
      "step": 93338
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.8043756484985352,
      "learning_rate": 0.0003881112797123673,
      "loss": 2.1743,
      "step": 93339
    },
    {
      "epoch": 1.22,
      "grad_norm": 1.3707534074783325,
      "learning_rate": 0.0003881073695507863,
      "loss": 2.3211,
      "step": 93340
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.9078829884529114,
      "learning_rate": 0.00038810345937282444,
      "loss": 2.2343,
      "step": 93341
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.9558263421058655,
      "learning_rate": 0.0003880995491784825,
      "loss": 2.3023,
      "step": 93342
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.7880396246910095,
      "learning_rate": 0.00038809563896776117,
      "loss": 2.2614,
      "step": 93343
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.8172673583030701,
      "learning_rate": 0.0003880917287406612,
      "loss": 2.1194,
      "step": 93344
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.8736545443534851,
      "learning_rate": 0.0003880878184971834,
      "loss": 2.302,
      "step": 93345
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.7758046388626099,
      "learning_rate": 0.0003880839082373283,
      "loss": 2.079,
      "step": 93346
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.7747616171836853,
      "learning_rate": 0.0003880799979610968,
      "loss": 2.0426,
      "step": 93347
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.8086344003677368,
      "learning_rate": 0.00038807608766848955,
      "loss": 1.7838,
      "step": 93348
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.7633072733879089,
      "learning_rate": 0.0003880721773595073,
      "loss": 2.0832,
      "step": 93349
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.8396452069282532,
      "learning_rate": 0.00038806826703415074,
      "loss": 2.2588,
      "step": 93350
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.9014073014259338,
      "learning_rate": 0.0003880643566924207,
      "loss": 1.9936,
      "step": 93351
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.8603913187980652,
      "learning_rate": 0.0003880604463343178,
      "loss": 2.4674,
      "step": 93352
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.8638797998428345,
      "learning_rate": 0.00038805653595984276,
      "loss": 2.1335,
      "step": 93353
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.984091579914093,
      "learning_rate": 0.0003880526255689964,
      "loss": 2.1721,
      "step": 93354
    },
    {
      "epoch": 1.22,
      "grad_norm": 1.1230945587158203,
      "learning_rate": 0.0003880487151617793,
      "loss": 2.3005,
      "step": 93355
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.9029632806777954,
      "learning_rate": 0.00038804480473819236,
      "loss": 2.1284,
      "step": 93356
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.9419621229171753,
      "learning_rate": 0.00038804089429823625,
      "loss": 2.4804,
      "step": 93357
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.8419412970542908,
      "learning_rate": 0.00038803698384191166,
      "loss": 2.2304,
      "step": 93358
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.8699415922164917,
      "learning_rate": 0.00038803307336921926,
      "loss": 2.1502,
      "step": 93359
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.9685324430465698,
      "learning_rate": 0.00038802916288015996,
      "loss": 2.3423,
      "step": 93360
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.7779132723808289,
      "learning_rate": 0.0003880252523747343,
      "loss": 1.9627,
      "step": 93361
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.7933803200721741,
      "learning_rate": 0.0003880213418529431,
      "loss": 1.8624,
      "step": 93362
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.771048367023468,
      "learning_rate": 0.00038801743131478705,
      "loss": 2.1898,
      "step": 93363
    },
    {
      "epoch": 1.22,
      "grad_norm": 1.0125786066055298,
      "learning_rate": 0.0003880135207602669,
      "loss": 2.0133,
      "step": 93364
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.8658391833305359,
      "learning_rate": 0.00038800961018938343,
      "loss": 2.072,
      "step": 93365
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.8446136116981506,
      "learning_rate": 0.0003880056996021373,
      "loss": 2.08,
      "step": 93366
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.950287938117981,
      "learning_rate": 0.00038800178899852923,
      "loss": 2.2834,
      "step": 93367
    },
    {
      "epoch": 1.22,
      "grad_norm": 1.0990427732467651,
      "learning_rate": 0.00038799787837855994,
      "loss": 2.281,
      "step": 93368
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.9691352248191833,
      "learning_rate": 0.00038799396774223026,
      "loss": 2.3936,
      "step": 93369
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.8787713050842285,
      "learning_rate": 0.0003879900570895408,
      "loss": 2.1165,
      "step": 93370
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.8588567972183228,
      "learning_rate": 0.0003879861464204923,
      "loss": 2.0344,
      "step": 93371
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.8270351886749268,
      "learning_rate": 0.0003879822357350855,
      "loss": 2.1222,
      "step": 93372
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.9147716164588928,
      "learning_rate": 0.0003879783250333212,
      "loss": 2.1337,
      "step": 93373
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.9316762685775757,
      "learning_rate": 0.00038797441431520006,
      "loss": 2.4477,
      "step": 93374
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.8328734636306763,
      "learning_rate": 0.0003879705035807228,
      "loss": 2.3791,
      "step": 93375
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.8350819945335388,
      "learning_rate": 0.00038796659282989016,
      "loss": 2.0851,
      "step": 93376
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.8005717396736145,
      "learning_rate": 0.00038796268206270286,
      "loss": 2.3498,
      "step": 93377
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.8058960437774658,
      "learning_rate": 0.0003879587712791617,
      "loss": 2.1361,
      "step": 93378
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.8603962659835815,
      "learning_rate": 0.00038795486047926725,
      "loss": 2.2264,
      "step": 93379
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.8572115898132324,
      "learning_rate": 0.00038795094966302034,
      "loss": 2.1726,
      "step": 93380
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.7865982055664062,
      "learning_rate": 0.0003879470388304217,
      "loss": 2.4834,
      "step": 93381
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.8566601276397705,
      "learning_rate": 0.0003879431279814721,
      "loss": 2.1046,
      "step": 93382
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.8415782451629639,
      "learning_rate": 0.00038793921711617217,
      "loss": 2.1637,
      "step": 93383
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.8456988334655762,
      "learning_rate": 0.00038793530623452266,
      "loss": 2.0807,
      "step": 93384
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.8586384654045105,
      "learning_rate": 0.0003879313953365244,
      "loss": 2.2553,
      "step": 93385
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.7874134182929993,
      "learning_rate": 0.000387927484422178,
      "loss": 2.0715,
      "step": 93386
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.9085007309913635,
      "learning_rate": 0.00038792357349148415,
      "loss": 2.0702,
      "step": 93387
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.9390143156051636,
      "learning_rate": 0.0003879196625444437,
      "loss": 2.2919,
      "step": 93388
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.865476667881012,
      "learning_rate": 0.0003879157515810574,
      "loss": 2.1743,
      "step": 93389
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.8889089226722717,
      "learning_rate": 0.00038791184060132575,
      "loss": 2.1894,
      "step": 93390
    },
    {
      "epoch": 1.22,
      "grad_norm": 1.1899099349975586,
      "learning_rate": 0.00038790792960524976,
      "loss": 2.4027,
      "step": 93391
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.7928814888000488,
      "learning_rate": 0.00038790401859282997,
      "loss": 2.4752,
      "step": 93392
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.9163756966590881,
      "learning_rate": 0.00038790010756406713,
      "loss": 2.1742,
      "step": 93393
    },
    {
      "epoch": 1.22,
      "grad_norm": 1.4290690422058105,
      "learning_rate": 0.0003878961965189621,
      "loss": 2.0854,
      "step": 93394
    },
    {
      "epoch": 1.22,
      "grad_norm": 1.016506314277649,
      "learning_rate": 0.0003878922854575154,
      "loss": 2.4442,
      "step": 93395
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.8712076544761658,
      "learning_rate": 0.00038788837437972787,
      "loss": 2.2104,
      "step": 93396
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.8936472535133362,
      "learning_rate": 0.00038788446328560033,
      "loss": 2.2272,
      "step": 93397
    },
    {
      "epoch": 1.22,
      "grad_norm": 1.2127015590667725,
      "learning_rate": 0.00038788055217513337,
      "loss": 2.0024,
      "step": 93398
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.7750919461250305,
      "learning_rate": 0.0003878766410483277,
      "loss": 2.2261,
      "step": 93399
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.8432972431182861,
      "learning_rate": 0.00038787272990518413,
      "loss": 2.2192,
      "step": 93400
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.9533600211143494,
      "learning_rate": 0.0003878688187457034,
      "loss": 2.2405,
      "step": 93401
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.8823792934417725,
      "learning_rate": 0.0003878649075698861,
      "loss": 2.2804,
      "step": 93402
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.8271822929382324,
      "learning_rate": 0.0003878609963777332,
      "loss": 2.1329,
      "step": 93403
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.8148502111434937,
      "learning_rate": 0.00038785708516924516,
      "loss": 1.9759,
      "step": 93404
    },
    {
      "epoch": 1.22,
      "grad_norm": 1.0670585632324219,
      "learning_rate": 0.0003878531739444229,
      "loss": 2.2761,
      "step": 93405
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.9276244640350342,
      "learning_rate": 0.000387849262703267,
      "loss": 2.2036,
      "step": 93406
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.7809697985649109,
      "learning_rate": 0.0003878453514457784,
      "loss": 2.2762,
      "step": 93407
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.9870374202728271,
      "learning_rate": 0.0003878414401719576,
      "loss": 2.159,
      "step": 93408
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.8421658277511597,
      "learning_rate": 0.00038783752888180536,
      "loss": 2.3257,
      "step": 93409
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.8759230375289917,
      "learning_rate": 0.0003878336175753226,
      "loss": 2.2675,
      "step": 93410
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.9260761141777039,
      "learning_rate": 0.00038782970625250983,
      "loss": 2.0145,
      "step": 93411
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.8036335706710815,
      "learning_rate": 0.00038782579491336783,
      "loss": 2.0046,
      "step": 93412
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.8480566740036011,
      "learning_rate": 0.0003878218835578974,
      "loss": 1.9858,
      "step": 93413
    },
    {
      "epoch": 1.22,
      "grad_norm": 1.1637243032455444,
      "learning_rate": 0.00038781797218609925,
      "loss": 2.183,
      "step": 93414
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.8838183879852295,
      "learning_rate": 0.000387814060797974,
      "loss": 2.1608,
      "step": 93415
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.7689601182937622,
      "learning_rate": 0.0003878101493935225,
      "loss": 2.204,
      "step": 93416
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.8614078164100647,
      "learning_rate": 0.00038780623797274544,
      "loss": 2.412,
      "step": 93417
    },
    {
      "epoch": 1.22,
      "grad_norm": 1.17519211769104,
      "learning_rate": 0.0003878023265356436,
      "loss": 2.4357,
      "step": 93418
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.7976915836334229,
      "learning_rate": 0.0003877984150822176,
      "loss": 2.2731,
      "step": 93419
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.9070968627929688,
      "learning_rate": 0.0003877945036124682,
      "loss": 2.0681,
      "step": 93420
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.9037885665893555,
      "learning_rate": 0.0003877905921263962,
      "loss": 2.1347,
      "step": 93421
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.8538275957107544,
      "learning_rate": 0.00038778668062400224,
      "loss": 2.0904,
      "step": 93422
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.9080206751823425,
      "learning_rate": 0.00038778276910528703,
      "loss": 2.2214,
      "step": 93423
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.8215089440345764,
      "learning_rate": 0.00038777885757025144,
      "loss": 2.2027,
      "step": 93424
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.9042001366615295,
      "learning_rate": 0.000387774946018896,
      "loss": 2.3198,
      "step": 93425
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.8206911087036133,
      "learning_rate": 0.0003877710344512216,
      "loss": 2.186,
      "step": 93426
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.8339467644691467,
      "learning_rate": 0.00038776712286722894,
      "loss": 2.3994,
      "step": 93427
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.9214707612991333,
      "learning_rate": 0.0003877632112669187,
      "loss": 2.1301,
      "step": 93428
    },
    {
      "epoch": 1.22,
      "grad_norm": 1.1952487230300903,
      "learning_rate": 0.00038775929965029156,
      "loss": 2.0845,
      "step": 93429
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.7479108572006226,
      "learning_rate": 0.0003877553880173484,
      "loss": 2.0354,
      "step": 93430
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.777481734752655,
      "learning_rate": 0.00038775147636808976,
      "loss": 2.1371,
      "step": 93431
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.9454470872879028,
      "learning_rate": 0.0003877475647025165,
      "loss": 2.2185,
      "step": 93432
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.8929804563522339,
      "learning_rate": 0.00038774365302062934,
      "loss": 2.3141,
      "step": 93433
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.8587974905967712,
      "learning_rate": 0.000387739741322429,
      "loss": 2.3406,
      "step": 93434
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.8252085447311401,
      "learning_rate": 0.0003877358296079161,
      "loss": 2.2335,
      "step": 93435
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.8065347075462341,
      "learning_rate": 0.00038773191787709154,
      "loss": 2.1924,
      "step": 93436
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.8959062695503235,
      "learning_rate": 0.0003877280061299559,
      "loss": 2.0555,
      "step": 93437
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.7791236042976379,
      "learning_rate": 0.00038772409436651,
      "loss": 2.1181,
      "step": 93438
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.7316829562187195,
      "learning_rate": 0.00038772018258675456,
      "loss": 2.2178,
      "step": 93439
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.9518300294876099,
      "learning_rate": 0.0003877162707906902,
      "loss": 2.1042,
      "step": 93440
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.8504706621170044,
      "learning_rate": 0.0003877123589783178,
      "loss": 2.0078,
      "step": 93441
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.9355578422546387,
      "learning_rate": 0.000387708447149638,
      "loss": 2.0548,
      "step": 93442
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.8627389669418335,
      "learning_rate": 0.0003877045353046515,
      "loss": 2.1986,
      "step": 93443
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.8535771369934082,
      "learning_rate": 0.0003877006234433591,
      "loss": 2.1163,
      "step": 93444
    },
    {
      "epoch": 1.22,
      "grad_norm": 1.1065130233764648,
      "learning_rate": 0.00038769671156576154,
      "loss": 2.2147,
      "step": 93445
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.8364512920379639,
      "learning_rate": 0.00038769279967185947,
      "loss": 1.9513,
      "step": 93446
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.9484800696372986,
      "learning_rate": 0.0003876888877616536,
      "loss": 2.1418,
      "step": 93447
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.9424477219581604,
      "learning_rate": 0.0003876849758351448,
      "loss": 2.0393,
      "step": 93448
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.9705683588981628,
      "learning_rate": 0.0003876810638923337,
      "loss": 1.958,
      "step": 93449
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.8848170042037964,
      "learning_rate": 0.00038767715193322095,
      "loss": 2.069,
      "step": 93450
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.9189810752868652,
      "learning_rate": 0.0003876732399578074,
      "loss": 2.3425,
      "step": 93451
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.8595840334892273,
      "learning_rate": 0.00038766932796609384,
      "loss": 2.3772,
      "step": 93452
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.8470784425735474,
      "learning_rate": 0.0003876654159580808,
      "loss": 2.1282,
      "step": 93453
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.8769731521606445,
      "learning_rate": 0.0003876615039337691,
      "loss": 2.2957,
      "step": 93454
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.9179043769836426,
      "learning_rate": 0.00038765759189315956,
      "loss": 2.1873,
      "step": 93455
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.8236531615257263,
      "learning_rate": 0.0003876536798362527,
      "loss": 2.2413,
      "step": 93456
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.7926729321479797,
      "learning_rate": 0.00038764976776304937,
      "loss": 2.2275,
      "step": 93457
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.8051574230194092,
      "learning_rate": 0.0003876458556735504,
      "loss": 2.0163,
      "step": 93458
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.9336159825325012,
      "learning_rate": 0.00038764194356775636,
      "loss": 1.9583,
      "step": 93459
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.8126592040061951,
      "learning_rate": 0.00038763803144566795,
      "loss": 2.142,
      "step": 93460
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.8403641581535339,
      "learning_rate": 0.0003876341193072861,
      "loss": 1.9914,
      "step": 93461
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.8040621876716614,
      "learning_rate": 0.0003876302071526114,
      "loss": 2.1085,
      "step": 93462
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.899695098400116,
      "learning_rate": 0.00038762629498164454,
      "loss": 2.2752,
      "step": 93463
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.8574495911598206,
      "learning_rate": 0.0003876223827943863,
      "loss": 2.0482,
      "step": 93464
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.8810205459594727,
      "learning_rate": 0.0003876184705908374,
      "loss": 2.1826,
      "step": 93465
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.8418119549751282,
      "learning_rate": 0.0003876145583709986,
      "loss": 2.1303,
      "step": 93466
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.8565068244934082,
      "learning_rate": 0.00038761064613487063,
      "loss": 2.0233,
      "step": 93467
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.8678230047225952,
      "learning_rate": 0.0003876067338824541,
      "loss": 2.2118,
      "step": 93468
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.8760842680931091,
      "learning_rate": 0.0003876028216137499,
      "loss": 2.1468,
      "step": 93469
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.7892515659332275,
      "learning_rate": 0.0003875989093287587,
      "loss": 2.0696,
      "step": 93470
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.9174894094467163,
      "learning_rate": 0.00038759499702748114,
      "loss": 2.3003,
      "step": 93471
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.7549600601196289,
      "learning_rate": 0.000387591084709918,
      "loss": 2.09,
      "step": 93472
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.8896599411964417,
      "learning_rate": 0.00038758717237607007,
      "loss": 1.9683,
      "step": 93473
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.8821456432342529,
      "learning_rate": 0.000387583260025938,
      "loss": 2.1517,
      "step": 93474
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.8781800866127014,
      "learning_rate": 0.0003875793476595225,
      "loss": 1.9739,
      "step": 93475
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.849314272403717,
      "learning_rate": 0.0003875754352768245,
      "loss": 1.9034,
      "step": 93476
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.9688982963562012,
      "learning_rate": 0.00038757152287784445,
      "loss": 2.2351,
      "step": 93477
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.8614053726196289,
      "learning_rate": 0.0003875676104625832,
      "loss": 2.1101,
      "step": 93478
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.871188759803772,
      "learning_rate": 0.00038756369803104155,
      "loss": 2.156,
      "step": 93479
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.765134871006012,
      "learning_rate": 0.0003875597855832201,
      "loss": 2.1777,
      "step": 93480
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.8515992760658264,
      "learning_rate": 0.0003875558731191197,
      "loss": 2.1682,
      "step": 93481
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.8439270853996277,
      "learning_rate": 0.000387551960638741,
      "loss": 2.1729,
      "step": 93482
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.9760951399803162,
      "learning_rate": 0.0003875480481420847,
      "loss": 2.1644,
      "step": 93483
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.8519662022590637,
      "learning_rate": 0.0003875441356291515,
      "loss": 2.0674,
      "step": 93484
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.9136447906494141,
      "learning_rate": 0.0003875402230999423,
      "loss": 2.2898,
      "step": 93485
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.9003794193267822,
      "learning_rate": 0.0003875363105544577,
      "loss": 2.0829,
      "step": 93486
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.9530027508735657,
      "learning_rate": 0.00038753239799269837,
      "loss": 1.8645,
      "step": 93487
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.7934730052947998,
      "learning_rate": 0.0003875284854146652,
      "loss": 1.9601,
      "step": 93488
    },
    {
      "epoch": 1.22,
      "grad_norm": 1.0465632677078247,
      "learning_rate": 0.00038752457282035884,
      "loss": 2.195,
      "step": 93489
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.8882856369018555,
      "learning_rate": 0.00038752066020977994,
      "loss": 2.1263,
      "step": 93490
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.8108710050582886,
      "learning_rate": 0.0003875167475829294,
      "loss": 2.4174,
      "step": 93491
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.8282361030578613,
      "learning_rate": 0.00038751283493980773,
      "loss": 2.0661,
      "step": 93492
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.925102174282074,
      "learning_rate": 0.00038750892228041583,
      "loss": 2.1442,
      "step": 93493
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.8843822479248047,
      "learning_rate": 0.00038750500960475434,
      "loss": 2.3973,
      "step": 93494
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.8437144160270691,
      "learning_rate": 0.00038750109691282407,
      "loss": 2.0149,
      "step": 93495
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.7854602336883545,
      "learning_rate": 0.00038749718420462567,
      "loss": 2.0042,
      "step": 93496
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.9274214506149292,
      "learning_rate": 0.00038749327148015984,
      "loss": 2.1817,
      "step": 93497
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.8460880517959595,
      "learning_rate": 0.0003874893587394274,
      "loss": 2.0301,
      "step": 93498
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.9609349370002747,
      "learning_rate": 0.00038748544598242907,
      "loss": 2.2273,
      "step": 93499
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.8842619061470032,
      "learning_rate": 0.00038748153320916547,
      "loss": 2.1935,
      "step": 93500
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.9486784934997559,
      "learning_rate": 0.00038747762041963744,
      "loss": 2.2085,
      "step": 93501
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.9202361702919006,
      "learning_rate": 0.0003874737076138457,
      "loss": 2.1864,
      "step": 93502
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.9296567440032959,
      "learning_rate": 0.0003874697947917909,
      "loss": 2.2552,
      "step": 93503
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.8576041460037231,
      "learning_rate": 0.0003874658819534738,
      "loss": 2.2344,
      "step": 93504
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.8884187936782837,
      "learning_rate": 0.0003874619690988952,
      "loss": 2.1899,
      "step": 93505
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.7891598343849182,
      "learning_rate": 0.00038745805622805577,
      "loss": 2.3149,
      "step": 93506
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.9872046113014221,
      "learning_rate": 0.0003874541433409562,
      "loss": 2.1151,
      "step": 93507
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.8836475610733032,
      "learning_rate": 0.00038745023043759724,
      "loss": 2.1109,
      "step": 93508
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.9151671528816223,
      "learning_rate": 0.00038744631751797974,
      "loss": 2.0784,
      "step": 93509
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.828176736831665,
      "learning_rate": 0.00038744240458210414,
      "loss": 2.104,
      "step": 93510
    },
    {
      "epoch": 1.22,
      "grad_norm": 1.0054404735565186,
      "learning_rate": 0.0003874384916299715,
      "loss": 2.1815,
      "step": 93511
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.8244953751564026,
      "learning_rate": 0.00038743457866158233,
      "loss": 2.2556,
      "step": 93512
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.8683175444602966,
      "learning_rate": 0.0003874306656769374,
      "loss": 2.1132,
      "step": 93513
    },
    {
      "epoch": 1.22,
      "grad_norm": 1.0217804908752441,
      "learning_rate": 0.00038742675267603745,
      "loss": 2.1809,
      "step": 93514
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.8092304468154907,
      "learning_rate": 0.0003874228396588834,
      "loss": 2.472,
      "step": 93515
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.8400992751121521,
      "learning_rate": 0.0003874189266254756,
      "loss": 2.1899,
      "step": 93516
    },
    {
      "epoch": 1.22,
      "grad_norm": 1.0871520042419434,
      "learning_rate": 0.00038741501357581495,
      "loss": 2.2096,
      "step": 93517
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.8059853911399841,
      "learning_rate": 0.00038741110050990226,
      "loss": 2.0275,
      "step": 93518
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.8793325424194336,
      "learning_rate": 0.00038740718742773825,
      "loss": 2.2134,
      "step": 93519
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.8206905126571655,
      "learning_rate": 0.00038740327432932356,
      "loss": 2.1061,
      "step": 93520
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.9557631611824036,
      "learning_rate": 0.0003873993612146589,
      "loss": 2.1881,
      "step": 93521
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.7843638062477112,
      "learning_rate": 0.00038739544808374514,
      "loss": 2.2066,
      "step": 93522
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.8332721590995789,
      "learning_rate": 0.00038739153493658283,
      "loss": 2.1103,
      "step": 93523
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.9166368246078491,
      "learning_rate": 0.0003873876217731728,
      "loss": 2.0088,
      "step": 93524
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.8421202898025513,
      "learning_rate": 0.00038738370859351583,
      "loss": 2.1628,
      "step": 93525
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.8813470005989075,
      "learning_rate": 0.0003873797953976125,
      "loss": 2.1993,
      "step": 93526
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.8204176425933838,
      "learning_rate": 0.00038737588218546365,
      "loss": 2.2402,
      "step": 93527
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.8749650716781616,
      "learning_rate": 0.00038737196895707,
      "loss": 2.2306,
      "step": 93528
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.7814149260520935,
      "learning_rate": 0.0003873680557124322,
      "loss": 1.986,
      "step": 93529
    },
    {
      "epoch": 1.22,
      "grad_norm": 1.0175179243087769,
      "learning_rate": 0.00038736414245155106,
      "loss": 2.4004,
      "step": 93530
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.7725915908813477,
      "learning_rate": 0.0003873602291744273,
      "loss": 2.4241,
      "step": 93531
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.8419443964958191,
      "learning_rate": 0.00038735631588106154,
      "loss": 2.2729,
      "step": 93532
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.8074020743370056,
      "learning_rate": 0.0003873524025714547,
      "loss": 2.0533,
      "step": 93533
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.8243134617805481,
      "learning_rate": 0.00038734848924560733,
      "loss": 2.2463,
      "step": 93534
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.8035179376602173,
      "learning_rate": 0.00038734457590352026,
      "loss": 2.456,
      "step": 93535
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.8889216184616089,
      "learning_rate": 0.0003873406625451941,
      "loss": 2.0502,
      "step": 93536
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.8895719051361084,
      "learning_rate": 0.0003873367491706298,
      "loss": 2.2186,
      "step": 93537
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.8554181456565857,
      "learning_rate": 0.00038733283577982783,
      "loss": 2.2043,
      "step": 93538
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.8198782205581665,
      "learning_rate": 0.0003873289223727891,
      "loss": 2.132,
      "step": 93539
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.8355281352996826,
      "learning_rate": 0.0003873250089495143,
      "loss": 2.1846,
      "step": 93540
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.8135033249855042,
      "learning_rate": 0.000387321095510004,
      "loss": 2.0402,
      "step": 93541
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.8628909587860107,
      "learning_rate": 0.0003873171820542592,
      "loss": 2.1756,
      "step": 93542
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.7877898812294006,
      "learning_rate": 0.00038731326858228046,
      "loss": 2.2175,
      "step": 93543
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.8720546364784241,
      "learning_rate": 0.0003873093550940685,
      "loss": 2.1768,
      "step": 93544
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.805413007736206,
      "learning_rate": 0.00038730544158962403,
      "loss": 2.2696,
      "step": 93545
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.8192307353019714,
      "learning_rate": 0.000387301528068948,
      "loss": 2.1926,
      "step": 93546
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.9069907069206238,
      "learning_rate": 0.0003872976145320408,
      "loss": 1.9277,
      "step": 93547
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.7785258889198303,
      "learning_rate": 0.0003872937009789034,
      "loss": 2.4039,
      "step": 93548
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.9672679901123047,
      "learning_rate": 0.0003872897874095365,
      "loss": 2.2998,
      "step": 93549
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.8859328031539917,
      "learning_rate": 0.00038728587382394063,
      "loss": 2.0501,
      "step": 93550
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.8852153420448303,
      "learning_rate": 0.0003872819602221168,
      "loss": 2.2964,
      "step": 93551
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.8233169317245483,
      "learning_rate": 0.0003872780466040656,
      "loss": 2.2144,
      "step": 93552
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.7727267146110535,
      "learning_rate": 0.00038727413296978775,
      "loss": 2.1254,
      "step": 93553
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.8876051902770996,
      "learning_rate": 0.0003872702193192839,
      "loss": 2.1762,
      "step": 93554
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.8392176628112793,
      "learning_rate": 0.000387266305652555,
      "loss": 2.2406,
      "step": 93555
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.8767449259757996,
      "learning_rate": 0.0003872623919696016,
      "loss": 2.1565,
      "step": 93556
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.8328553438186646,
      "learning_rate": 0.0003872584782704244,
      "loss": 2.0903,
      "step": 93557
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.8562953472137451,
      "learning_rate": 0.0003872545645550244,
      "loss": 2.2044,
      "step": 93558
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.8267685770988464,
      "learning_rate": 0.0003872506508234019,
      "loss": 2.1908,
      "step": 93559
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.8612523078918457,
      "learning_rate": 0.0003872467370755579,
      "loss": 2.2959,
      "step": 93560
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.8408039808273315,
      "learning_rate": 0.0003872428233114932,
      "loss": 1.9774,
      "step": 93561
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.7611677646636963,
      "learning_rate": 0.00038723890953120836,
      "loss": 2.2373,
      "step": 93562
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.9404417276382446,
      "learning_rate": 0.00038723499573470414,
      "loss": 2.2185,
      "step": 93563
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.915199339389801,
      "learning_rate": 0.00038723108192198133,
      "loss": 2.0973,
      "step": 93564
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.7972303032875061,
      "learning_rate": 0.00038722716809304054,
      "loss": 2.0873,
      "step": 93565
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.7449790835380554,
      "learning_rate": 0.0003872232542478826,
      "loss": 2.0304,
      "step": 93566
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.7539566159248352,
      "learning_rate": 0.00038721934038650827,
      "loss": 2.3959,
      "step": 93567
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.8225350379943848,
      "learning_rate": 0.0003872154265089182,
      "loss": 2.3055,
      "step": 93568
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.8346830010414124,
      "learning_rate": 0.00038721151261511305,
      "loss": 2.2573,
      "step": 93569
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.831741452217102,
      "learning_rate": 0.0003872075987050938,
      "loss": 2.3011,
      "step": 93570
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.7916224002838135,
      "learning_rate": 0.0003872036847788608,
      "loss": 2.2243,
      "step": 93571
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.7826968431472778,
      "learning_rate": 0.0003871997708364151,
      "loss": 2.3829,
      "step": 93572
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.8539107441902161,
      "learning_rate": 0.00038719585687775734,
      "loss": 2.1986,
      "step": 93573
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.9138230085372925,
      "learning_rate": 0.0003871919429028882,
      "loss": 2.2003,
      "step": 93574
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.8081119060516357,
      "learning_rate": 0.00038718802891180845,
      "loss": 2.1682,
      "step": 93575
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.7929344773292542,
      "learning_rate": 0.00038718411490451873,
      "loss": 2.1745,
      "step": 93576
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.7925371527671814,
      "learning_rate": 0.0003871802008810199,
      "loss": 2.116,
      "step": 93577
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.8074197769165039,
      "learning_rate": 0.00038717628684131256,
      "loss": 2.1392,
      "step": 93578
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.8707441091537476,
      "learning_rate": 0.0003871723727853975,
      "loss": 2.1978,
      "step": 93579
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.8547412753105164,
      "learning_rate": 0.00038716845871327555,
      "loss": 1.9648,
      "step": 93580
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.8458847403526306,
      "learning_rate": 0.0003871645446249473,
      "loss": 2.1045,
      "step": 93581
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.9200289845466614,
      "learning_rate": 0.0003871606305204135,
      "loss": 2.1768,
      "step": 93582
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.9365640878677368,
      "learning_rate": 0.0003871567163996749,
      "loss": 2.2364,
      "step": 93583
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.9896007180213928,
      "learning_rate": 0.0003871528022627322,
      "loss": 1.9349,
      "step": 93584
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.8558160662651062,
      "learning_rate": 0.00038714888810958617,
      "loss": 2.1739,
      "step": 93585
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.8825773000717163,
      "learning_rate": 0.0003871449739402375,
      "loss": 2.2543,
      "step": 93586
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.8568035364151001,
      "learning_rate": 0.000387141059754687,
      "loss": 2.2516,
      "step": 93587
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.7879536747932434,
      "learning_rate": 0.00038713714555293523,
      "loss": 2.3218,
      "step": 93588
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.7936029434204102,
      "learning_rate": 0.0003871332313349831,
      "loss": 2.4472,
      "step": 93589
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.8465499877929688,
      "learning_rate": 0.0003871293171008312,
      "loss": 2.2687,
      "step": 93590
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.8407159447669983,
      "learning_rate": 0.0003871254028504804,
      "loss": 2.1173,
      "step": 93591
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.8741011619567871,
      "learning_rate": 0.00038712148858393127,
      "loss": 2.2404,
      "step": 93592
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.7855728268623352,
      "learning_rate": 0.00038711757430118454,
      "loss": 2.0997,
      "step": 93593
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.8427106142044067,
      "learning_rate": 0.0003871136600022412,
      "loss": 2.1045,
      "step": 93594
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.879055917263031,
      "learning_rate": 0.00038710974568710165,
      "loss": 2.3113,
      "step": 93595
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.8091045618057251,
      "learning_rate": 0.0003871058313557668,
      "loss": 2.0499,
      "step": 93596
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.8091285228729248,
      "learning_rate": 0.00038710191700823724,
      "loss": 2.4407,
      "step": 93597
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.8707683682441711,
      "learning_rate": 0.0003870980026445139,
      "loss": 2.1484,
      "step": 93598
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.7694981694221497,
      "learning_rate": 0.00038709408826459734,
      "loss": 2.1223,
      "step": 93599
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.7990462779998779,
      "learning_rate": 0.0003870901738684884,
      "loss": 2.132,
      "step": 93600
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.8354822397232056,
      "learning_rate": 0.00038708625945618774,
      "loss": 2.2261,
      "step": 93601
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.8806118369102478,
      "learning_rate": 0.00038708234502769605,
      "loss": 2.3363,
      "step": 93602
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.7973757386207581,
      "learning_rate": 0.00038707843058301413,
      "loss": 2.1008,
      "step": 93603
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.9470233917236328,
      "learning_rate": 0.00038707451612214275,
      "loss": 2.2943,
      "step": 93604
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.846295177936554,
      "learning_rate": 0.00038707060164508254,
      "loss": 2.2133,
      "step": 93605
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.9845947027206421,
      "learning_rate": 0.00038706668715183417,
      "loss": 2.3041,
      "step": 93606
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.8394702076911926,
      "learning_rate": 0.0003870627726423986,
      "loss": 2.112,
      "step": 93607
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.8299950957298279,
      "learning_rate": 0.00038705885811677633,
      "loss": 2.3119,
      "step": 93608
    },
    {
      "epoch": 1.22,
      "grad_norm": 1.0218689441680908,
      "learning_rate": 0.00038705494357496817,
      "loss": 2.2558,
      "step": 93609
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.9204788208007812,
      "learning_rate": 0.00038705102901697493,
      "loss": 2.118,
      "step": 93610
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.7396873831748962,
      "learning_rate": 0.0003870471144427972,
      "loss": 1.9703,
      "step": 93611
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.857202410697937,
      "learning_rate": 0.0003870431998524357,
      "loss": 2.0369,
      "step": 93612
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.8948597311973572,
      "learning_rate": 0.0003870392852458914,
      "loss": 2.1878,
      "step": 93613
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.8763881921768188,
      "learning_rate": 0.0003870353706231647,
      "loss": 2.1029,
      "step": 93614
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.8927613496780396,
      "learning_rate": 0.0003870314559842565,
      "loss": 2.0458,
      "step": 93615
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.958952784538269,
      "learning_rate": 0.00038702754132916766,
      "loss": 2.0648,
      "step": 93616
    },
    {
      "epoch": 1.22,
      "grad_norm": 1.0526964664459229,
      "learning_rate": 0.00038702362665789854,
      "loss": 2.2594,
      "step": 93617
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.8825473785400391,
      "learning_rate": 0.0003870197119704502,
      "loss": 2.0659,
      "step": 93618
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.9090809226036072,
      "learning_rate": 0.0003870157972668233,
      "loss": 2.2846,
      "step": 93619
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.8979657292366028,
      "learning_rate": 0.00038701188254701847,
      "loss": 1.9789,
      "step": 93620
    },
    {
      "epoch": 1.22,
      "grad_norm": 1.1527247428894043,
      "learning_rate": 0.0003870079678110364,
      "loss": 2.2134,
      "step": 93621
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.9259821176528931,
      "learning_rate": 0.0003870040530588781,
      "loss": 2.2551,
      "step": 93622
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.894221305847168,
      "learning_rate": 0.0003870001382905439,
      "loss": 2.0623,
      "step": 93623
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.906882643699646,
      "learning_rate": 0.0003869962235060348,
      "loss": 2.2956,
      "step": 93624
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.9545174837112427,
      "learning_rate": 0.0003869923087053516,
      "loss": 2.277,
      "step": 93625
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.8982749581336975,
      "learning_rate": 0.00038698839388849474,
      "loss": 2.1246,
      "step": 93626
    },
    {
      "epoch": 1.22,
      "grad_norm": 1.043535828590393,
      "learning_rate": 0.0003869844790554651,
      "loss": 2.0579,
      "step": 93627
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.9961869120597839,
      "learning_rate": 0.0003869805642062634,
      "loss": 2.1215,
      "step": 93628
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.8005955219268799,
      "learning_rate": 0.0003869766493408905,
      "loss": 2.2203,
      "step": 93629
    },
    {
      "epoch": 1.22,
      "grad_norm": 1.077775239944458,
      "learning_rate": 0.0003869727344593468,
      "loss": 2.3043,
      "step": 93630
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.8612028360366821,
      "learning_rate": 0.0003869688195616334,
      "loss": 2.1063,
      "step": 93631
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.862877368927002,
      "learning_rate": 0.0003869649046477507,
      "loss": 2.1234,
      "step": 93632
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.7913288474082947,
      "learning_rate": 0.0003869609897176997,
      "loss": 2.0962,
      "step": 93633
    },
    {
      "epoch": 1.22,
      "grad_norm": 1.0369590520858765,
      "learning_rate": 0.00038695707477148106,
      "loss": 2.0149,
      "step": 93634
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.7458047866821289,
      "learning_rate": 0.0003869531598090953,
      "loss": 2.1907,
      "step": 93635
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.8758389949798584,
      "learning_rate": 0.00038694924483054336,
      "loss": 2.3882,
      "step": 93636
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.9268695712089539,
      "learning_rate": 0.000386945329835826,
      "loss": 2.1428,
      "step": 93637
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.8940938115119934,
      "learning_rate": 0.0003869414148249438,
      "loss": 2.08,
      "step": 93638
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.8215296268463135,
      "learning_rate": 0.0003869374997978975,
      "loss": 2.1705,
      "step": 93639
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.8447727560997009,
      "learning_rate": 0.0003869335847546879,
      "loss": 2.2007,
      "step": 93640
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.8692430257797241,
      "learning_rate": 0.0003869296696953158,
      "loss": 2.1512,
      "step": 93641
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.7937301397323608,
      "learning_rate": 0.00038692575461978174,
      "loss": 2.0501,
      "step": 93642
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.8505222201347351,
      "learning_rate": 0.00038692183952808657,
      "loss": 2.1954,
      "step": 93643
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.8277139067649841,
      "learning_rate": 0.000386917924420231,
      "loss": 2.2333,
      "step": 93644
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.8040342330932617,
      "learning_rate": 0.0003869140092962157,
      "loss": 2.2568,
      "step": 93645
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.8659228682518005,
      "learning_rate": 0.0003869100941560415,
      "loss": 2.2435,
      "step": 93646
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.8831489682197571,
      "learning_rate": 0.000386906178999709,
      "loss": 2.3845,
      "step": 93647
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.8537209033966064,
      "learning_rate": 0.0003869022638272191,
      "loss": 2.0832,
      "step": 93648
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.8438025116920471,
      "learning_rate": 0.0003868983486385724,
      "loss": 2.1507,
      "step": 93649
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.8782505393028259,
      "learning_rate": 0.0003868944334337696,
      "loss": 2.3802,
      "step": 93650
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.8283505439758301,
      "learning_rate": 0.0003868905182128115,
      "loss": 2.1187,
      "step": 93651
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.9011846780776978,
      "learning_rate": 0.0003868866029756988,
      "loss": 2.2737,
      "step": 93652
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.8875617384910583,
      "learning_rate": 0.00038688268772243235,
      "loss": 2.1098,
      "step": 93653
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.8312456011772156,
      "learning_rate": 0.00038687877245301264,
      "loss": 2.3128,
      "step": 93654
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.8255608081817627,
      "learning_rate": 0.0003868748571674406,
      "loss": 2.2168,
      "step": 93655
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.9196488857269287,
      "learning_rate": 0.00038687094186571686,
      "loss": 2.2214,
      "step": 93656
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.7914440631866455,
      "learning_rate": 0.00038686702654784215,
      "loss": 2.2728,
      "step": 93657
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.8288446664810181,
      "learning_rate": 0.0003868631112138172,
      "loss": 2.3181,
      "step": 93658
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.8746106624603271,
      "learning_rate": 0.0003868591958636429,
      "loss": 2.1971,
      "step": 93659
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.9298259019851685,
      "learning_rate": 0.00038685528049731973,
      "loss": 2.1397,
      "step": 93660
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.8250073194503784,
      "learning_rate": 0.00038685136511484847,
      "loss": 2.1622,
      "step": 93661
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.8302279710769653,
      "learning_rate": 0.00038684744971623,
      "loss": 2.3587,
      "step": 93662
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.8061493635177612,
      "learning_rate": 0.0003868435343014649,
      "loss": 2.069,
      "step": 93663
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.8573988080024719,
      "learning_rate": 0.0003868396188705539,
      "loss": 1.9342,
      "step": 93664
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.7796431183815002,
      "learning_rate": 0.0003868357034234979,
      "loss": 2.1212,
      "step": 93665
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.8134298324584961,
      "learning_rate": 0.00038683178796029746,
      "loss": 2.2742,
      "step": 93666
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.9040966629981995,
      "learning_rate": 0.0003868278724809533,
      "loss": 2.3016,
      "step": 93667
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.8372673988342285,
      "learning_rate": 0.0003868239569854663,
      "loss": 2.1198,
      "step": 93668
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.8283501863479614,
      "learning_rate": 0.00038682004147383696,
      "loss": 2.0757,
      "step": 93669
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.7693381905555725,
      "learning_rate": 0.0003868161259460662,
      "loss": 2.4249,
      "step": 93670
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.9853623509407043,
      "learning_rate": 0.0003868122104021546,
      "loss": 2.3047,
      "step": 93671
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.809535026550293,
      "learning_rate": 0.00038680829484210314,
      "loss": 2.35,
      "step": 93672
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.8922232389450073,
      "learning_rate": 0.0003868043792659122,
      "loss": 2.1388,
      "step": 93673
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.8276000618934631,
      "learning_rate": 0.0003868004636735828,
      "loss": 2.1823,
      "step": 93674
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.9336802959442139,
      "learning_rate": 0.0003867965480651156,
      "loss": 2.3318,
      "step": 93675
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.8540804386138916,
      "learning_rate": 0.00038679263244051115,
      "loss": 2.2558,
      "step": 93676
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.9445028305053711,
      "learning_rate": 0.0003867887167997704,
      "loss": 2.3563,
      "step": 93677
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.8937026858329773,
      "learning_rate": 0.00038678480114289393,
      "loss": 2.1617,
      "step": 93678
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.8838561773300171,
      "learning_rate": 0.0003867808854698826,
      "loss": 2.1051,
      "step": 93679
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.7565919756889343,
      "learning_rate": 0.000386776969780737,
      "loss": 2.0791,
      "step": 93680
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.8780032992362976,
      "learning_rate": 0.00038677305407545796,
      "loss": 1.993,
      "step": 93681
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.9850114583969116,
      "learning_rate": 0.0003867691383540461,
      "loss": 2.1286,
      "step": 93682
    },
    {
      "epoch": 1.22,
      "grad_norm": 1.0108190774917603,
      "learning_rate": 0.00038676522261650234,
      "loss": 2.3765,
      "step": 93683
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.8393893837928772,
      "learning_rate": 0.0003867613068628272,
      "loss": 2.0358,
      "step": 93684
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.9428362250328064,
      "learning_rate": 0.00038675739109302146,
      "loss": 2.1238,
      "step": 93685
    },
    {
      "epoch": 1.22,
      "grad_norm": 1.1158030033111572,
      "learning_rate": 0.00038675347530708605,
      "loss": 2.2176,
      "step": 93686
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.9447388052940369,
      "learning_rate": 0.0003867495595050214,
      "loss": 1.9964,
      "step": 93687
    },
    {
      "epoch": 1.22,
      "grad_norm": 1.0752918720245361,
      "learning_rate": 0.0003867456436868284,
      "loss": 2.1034,
      "step": 93688
    },
    {
      "epoch": 1.22,
      "grad_norm": 1.9014416933059692,
      "learning_rate": 0.0003867417278525077,
      "loss": 2.1952,
      "step": 93689
    },
    {
      "epoch": 1.22,
      "grad_norm": 1.3863970041275024,
      "learning_rate": 0.00038673781200206016,
      "loss": 2.1477,
      "step": 93690
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.9235509634017944,
      "learning_rate": 0.0003867338961354864,
      "loss": 1.9882,
      "step": 93691
    },
    {
      "epoch": 1.22,
      "grad_norm": 1.075774908065796,
      "learning_rate": 0.00038672998025278713,
      "loss": 2.259,
      "step": 93692
    },
    {
      "epoch": 1.22,
      "grad_norm": 1.3971384763717651,
      "learning_rate": 0.00038672606435396313,
      "loss": 2.3231,
      "step": 93693
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.9280847907066345,
      "learning_rate": 0.0003867221484390151,
      "loss": 2.1099,
      "step": 93694
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.9589519500732422,
      "learning_rate": 0.00038671823250794387,
      "loss": 2.2002,
      "step": 93695
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.9950979948043823,
      "learning_rate": 0.00038671431656075,
      "loss": 2.1327,
      "step": 93696
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.8851667046546936,
      "learning_rate": 0.00038671040059743436,
      "loss": 2.1808,
      "step": 93697
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.774723470211029,
      "learning_rate": 0.00038670648461799765,
      "loss": 2.2983,
      "step": 93698
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.8146947026252747,
      "learning_rate": 0.00038670256862244047,
      "loss": 2.1413,
      "step": 93699
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.7925690412521362,
      "learning_rate": 0.0003866986526107637,
      "loss": 2.1241,
      "step": 93700
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.8072047829627991,
      "learning_rate": 0.000386694736582968,
      "loss": 2.1132,
      "step": 93701
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.9282036423683167,
      "learning_rate": 0.0003866908205390541,
      "loss": 1.8654,
      "step": 93702
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.8451675176620483,
      "learning_rate": 0.00038668690447902273,
      "loss": 2.0721,
      "step": 93703
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.8680562376976013,
      "learning_rate": 0.0003866829884028747,
      "loss": 2.1726,
      "step": 93704
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.8761387467384338,
      "learning_rate": 0.0003866790723106106,
      "loss": 2.3363,
      "step": 93705
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.9094669818878174,
      "learning_rate": 0.00038667515620223117,
      "loss": 2.248,
      "step": 93706
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.8525984287261963,
      "learning_rate": 0.00038667124007773736,
      "loss": 2.1495,
      "step": 93707
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.8769549131393433,
      "learning_rate": 0.0003866673239371296,
      "loss": 2.3016,
      "step": 93708
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.8942551612854004,
      "learning_rate": 0.00038666340778040876,
      "loss": 2.1505,
      "step": 93709
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.8976256251335144,
      "learning_rate": 0.0003866594916075756,
      "loss": 2.2845,
      "step": 93710
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.868399977684021,
      "learning_rate": 0.0003866555754186308,
      "loss": 2.3299,
      "step": 93711
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.9347005486488342,
      "learning_rate": 0.000386651659213575,
      "loss": 2.253,
      "step": 93712
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.8377557396888733,
      "learning_rate": 0.0003866477429924092,
      "loss": 2.2695,
      "step": 93713
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.8290286064147949,
      "learning_rate": 0.0003866438267551338,
      "loss": 1.9927,
      "step": 93714
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.8094969987869263,
      "learning_rate": 0.0003866399105017497,
      "loss": 2.1236,
      "step": 93715
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.84604811668396,
      "learning_rate": 0.00038663599423225764,
      "loss": 2.0824,
      "step": 93716
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.8714594841003418,
      "learning_rate": 0.0003866320779466583,
      "loss": 2.3281,
      "step": 93717
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.8605619668960571,
      "learning_rate": 0.00038662816164495244,
      "loss": 2.0937,
      "step": 93718
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.8390880227088928,
      "learning_rate": 0.0003866242453271407,
      "loss": 2.1863,
      "step": 93719
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.78374844789505,
      "learning_rate": 0.000386620328993224,
      "loss": 2.153,
      "step": 93720
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.8965654969215393,
      "learning_rate": 0.00038661641264320286,
      "loss": 2.1163,
      "step": 93721
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.8595349192619324,
      "learning_rate": 0.00038661249627707807,
      "loss": 2.4817,
      "step": 93722
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.859919548034668,
      "learning_rate": 0.0003866085798948505,
      "loss": 2.2232,
      "step": 93723
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.92052161693573,
      "learning_rate": 0.00038660466349652067,
      "loss": 2.4378,
      "step": 93724
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.9678916931152344,
      "learning_rate": 0.00038660074708208935,
      "loss": 2.2667,
      "step": 93725
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.8743575215339661,
      "learning_rate": 0.0003865968306515574,
      "loss": 2.4009,
      "step": 93726
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.9474013447761536,
      "learning_rate": 0.00038659291420492547,
      "loss": 2.0484,
      "step": 93727
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.8591078519821167,
      "learning_rate": 0.0003865889977421942,
      "loss": 2.051,
      "step": 93728
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.8618503212928772,
      "learning_rate": 0.00038658508126336447,
      "loss": 2.3102,
      "step": 93729
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.8271157741546631,
      "learning_rate": 0.00038658116476843695,
      "loss": 1.9808,
      "step": 93730
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.863028347492218,
      "learning_rate": 0.0003865772482574123,
      "loss": 2.2986,
      "step": 93731
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.916823148727417,
      "learning_rate": 0.0003865733317302913,
      "loss": 2.177,
      "step": 93732
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.7533767819404602,
      "learning_rate": 0.0003865694151870748,
      "loss": 1.9654,
      "step": 93733
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.7668265700340271,
      "learning_rate": 0.0003865654986277633,
      "loss": 2.1385,
      "step": 93734
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.964604914188385,
      "learning_rate": 0.0003865615820523577,
      "loss": 2.2399,
      "step": 93735
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.8744791150093079,
      "learning_rate": 0.0003865576654608586,
      "loss": 2.286,
      "step": 93736
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.9298447966575623,
      "learning_rate": 0.00038655374885326687,
      "loss": 2.0737,
      "step": 93737
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.899581253528595,
      "learning_rate": 0.00038654983222958313,
      "loss": 2.1979,
      "step": 93738
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.857664942741394,
      "learning_rate": 0.0003865459155898081,
      "loss": 2.2609,
      "step": 93739
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.8099071383476257,
      "learning_rate": 0.0003865419989339426,
      "loss": 1.9971,
      "step": 93740
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.8667165637016296,
      "learning_rate": 0.0003865380822619874,
      "loss": 2.0757,
      "step": 93741
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.9080754518508911,
      "learning_rate": 0.00038653416557394297,
      "loss": 2.1531,
      "step": 93742
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.8451882004737854,
      "learning_rate": 0.00038653024886981023,
      "loss": 2.1977,
      "step": 93743
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.8700289726257324,
      "learning_rate": 0.00038652633214958994,
      "loss": 2.4232,
      "step": 93744
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.8400221467018127,
      "learning_rate": 0.00038652241541328275,
      "loss": 2.1392,
      "step": 93745
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.8833936452865601,
      "learning_rate": 0.0003865184986608894,
      "loss": 2.2214,
      "step": 93746
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.7991616129875183,
      "learning_rate": 0.0003865145818924107,
      "loss": 1.9876,
      "step": 93747
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.9064564108848572,
      "learning_rate": 0.0003865106651078472,
      "loss": 2.0651,
      "step": 93748
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.8088630437850952,
      "learning_rate": 0.0003865067483071998,
      "loss": 2.2165,
      "step": 93749
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.8705615401268005,
      "learning_rate": 0.0003865028314904692,
      "loss": 2.2237,
      "step": 93750
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.8308112025260925,
      "learning_rate": 0.000386498914657656,
      "loss": 1.9709,
      "step": 93751
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.8545401692390442,
      "learning_rate": 0.0003864949978087611,
      "loss": 1.9731,
      "step": 93752
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.8341441750526428,
      "learning_rate": 0.00038649108094378513,
      "loss": 2.2429,
      "step": 93753
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.7746773958206177,
      "learning_rate": 0.00038648716406272876,
      "loss": 2.0415,
      "step": 93754
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.8171632289886475,
      "learning_rate": 0.00038648324716559286,
      "loss": 1.9726,
      "step": 93755
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.992419421672821,
      "learning_rate": 0.0003864793302523781,
      "loss": 2.2111,
      "step": 93756
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.794248104095459,
      "learning_rate": 0.0003864754133230852,
      "loss": 2.296,
      "step": 93757
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.891361653804779,
      "learning_rate": 0.00038647149637771475,
      "loss": 2.1275,
      "step": 93758
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.955296516418457,
      "learning_rate": 0.0003864675794162677,
      "loss": 2.0735,
      "step": 93759
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.8747338652610779,
      "learning_rate": 0.00038646366243874484,
      "loss": 2.0777,
      "step": 93760
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.874394416809082,
      "learning_rate": 0.00038645974544514655,
      "loss": 2.2156,
      "step": 93761
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.8227454423904419,
      "learning_rate": 0.0003864558284354739,
      "loss": 1.9777,
      "step": 93762
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.8526372313499451,
      "learning_rate": 0.0003864519114097275,
      "loss": 2.0627,
      "step": 93763
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.851608395576477,
      "learning_rate": 0.0003864479943679079,
      "loss": 2.2366,
      "step": 93764
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.8178184628486633,
      "learning_rate": 0.00038644407731001603,
      "loss": 2.4566,
      "step": 93765
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.9153255820274353,
      "learning_rate": 0.0003864401602360526,
      "loss": 2.0086,
      "step": 93766
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.8486509323120117,
      "learning_rate": 0.0003864362431460184,
      "loss": 2.3079,
      "step": 93767
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.8458197712898254,
      "learning_rate": 0.000386432326039914,
      "loss": 2.0081,
      "step": 93768
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.819727897644043,
      "learning_rate": 0.0003864284089177402,
      "loss": 2.2968,
      "step": 93769
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.8622077107429504,
      "learning_rate": 0.00038642449177949767,
      "loss": 1.9996,
      "step": 93770
    },
    {
      "epoch": 1.22,
      "grad_norm": 1.0212923288345337,
      "learning_rate": 0.00038642057462518733,
      "loss": 2.0266,
      "step": 93771
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.8430885076522827,
      "learning_rate": 0.00038641665745480966,
      "loss": 2.218,
      "step": 93772
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.9841453433036804,
      "learning_rate": 0.0003864127402683655,
      "loss": 2.0443,
      "step": 93773
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.8751956224441528,
      "learning_rate": 0.0003864088230658557,
      "loss": 2.3591,
      "step": 93774
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.9231245517730713,
      "learning_rate": 0.0003864049058472807,
      "loss": 2.246,
      "step": 93775
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.8439788222312927,
      "learning_rate": 0.0003864009886126415,
      "loss": 2.1643,
      "step": 93776
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.8798419237136841,
      "learning_rate": 0.0003863970713619388,
      "loss": 1.9148,
      "step": 93777
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.9787272810935974,
      "learning_rate": 0.0003863931540951731,
      "loss": 2.1089,
      "step": 93778
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.8303635120391846,
      "learning_rate": 0.0003863892368123453,
      "loss": 2.1277,
      "step": 93779
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.806243896484375,
      "learning_rate": 0.0003863853195134562,
      "loss": 1.9715,
      "step": 93780
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.9410203099250793,
      "learning_rate": 0.0003863814021985064,
      "loss": 2.0209,
      "step": 93781
    },
    {
      "epoch": 1.22,
      "grad_norm": 1.012834906578064,
      "learning_rate": 0.00038637748486749666,
      "loss": 2.1648,
      "step": 93782
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.8097728490829468,
      "learning_rate": 0.0003863735675204277,
      "loss": 2.0842,
      "step": 93783
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.8357135653495789,
      "learning_rate": 0.0003863696501573003,
      "loss": 2.1791,
      "step": 93784
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.8324261903762817,
      "learning_rate": 0.00038636573277811516,
      "loss": 2.1242,
      "step": 93785
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.9066418409347534,
      "learning_rate": 0.00038636181538287287,
      "loss": 1.9785,
      "step": 93786
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.8484653830528259,
      "learning_rate": 0.00038635789797157447,
      "loss": 2.0679,
      "step": 93787
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.8378857970237732,
      "learning_rate": 0.0003863539805442204,
      "loss": 2.2976,
      "step": 93788
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.8597121238708496,
      "learning_rate": 0.0003863500631008115,
      "loss": 2.0747,
      "step": 93789
    },
    {
      "epoch": 1.22,
      "grad_norm": 1.0249955654144287,
      "learning_rate": 0.0003863461456413485,
      "loss": 2.2822,
      "step": 93790
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.9407898187637329,
      "learning_rate": 0.00038634222816583217,
      "loss": 2.541,
      "step": 93791
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.7822743654251099,
      "learning_rate": 0.00038633831067426313,
      "loss": 2.3477,
      "step": 93792
    },
    {
      "epoch": 1.22,
      "grad_norm": 1.2879959344863892,
      "learning_rate": 0.00038633439316664226,
      "loss": 1.9519,
      "step": 93793
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.8765513896942139,
      "learning_rate": 0.0003863304756429701,
      "loss": 2.0665,
      "step": 93794
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.805971622467041,
      "learning_rate": 0.0003863265581032475,
      "loss": 2.3339,
      "step": 93795
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.8661971688270569,
      "learning_rate": 0.0003863226405474752,
      "loss": 2.1692,
      "step": 93796
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.9477578997612,
      "learning_rate": 0.00038631872297565387,
      "loss": 2.2367,
      "step": 93797
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.8172412514686584,
      "learning_rate": 0.0003863148053877842,
      "loss": 2.4485,
      "step": 93798
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.8552762866020203,
      "learning_rate": 0.0003863108877838671,
      "loss": 2.1369,
      "step": 93799
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.8669562935829163,
      "learning_rate": 0.00038630697016390304,
      "loss": 2.1558,
      "step": 93800
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.877077579498291,
      "learning_rate": 0.00038630305252789296,
      "loss": 2.2106,
      "step": 93801
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.8575328588485718,
      "learning_rate": 0.00038629913487583755,
      "loss": 2.1428,
      "step": 93802
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.9351742267608643,
      "learning_rate": 0.0003862952172077374,
      "loss": 2.0847,
      "step": 93803
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.8158247470855713,
      "learning_rate": 0.0003862912995235934,
      "loss": 2.2235,
      "step": 93804
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.880655825138092,
      "learning_rate": 0.00038628738182340625,
      "loss": 2.1051,
      "step": 93805
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.822747528553009,
      "learning_rate": 0.00038628346410717664,
      "loss": 2.0512,
      "step": 93806
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.8988993167877197,
      "learning_rate": 0.00038627954637490523,
      "loss": 2.1102,
      "step": 93807
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.827425479888916,
      "learning_rate": 0.00038627562862659294,
      "loss": 2.085,
      "step": 93808
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.8908020853996277,
      "learning_rate": 0.00038627171086224026,
      "loss": 2.2704,
      "step": 93809
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.9568995833396912,
      "learning_rate": 0.00038626779308184806,
      "loss": 2.1729,
      "step": 93810
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.9155206680297852,
      "learning_rate": 0.00038626387528541714,
      "loss": 2.4869,
      "step": 93811
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.8255177736282349,
      "learning_rate": 0.00038625995747294806,
      "loss": 2.0664,
      "step": 93812
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.8403735160827637,
      "learning_rate": 0.0003862560396444416,
      "loss": 2.4058,
      "step": 93813
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.8551434874534607,
      "learning_rate": 0.00038625212179989863,
      "loss": 2.2365,
      "step": 93814
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.9457610845565796,
      "learning_rate": 0.0003862482039393197,
      "loss": 2.1542,
      "step": 93815
    },
    {
      "epoch": 1.22,
      "grad_norm": 1.037076473236084,
      "learning_rate": 0.00038624428606270553,
      "loss": 2.3325,
      "step": 93816
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.8914788365364075,
      "learning_rate": 0.00038624036817005704,
      "loss": 2.178,
      "step": 93817
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.8456154465675354,
      "learning_rate": 0.00038623645026137477,
      "loss": 2.2451,
      "step": 93818
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.8234896659851074,
      "learning_rate": 0.0003862325323366595,
      "loss": 1.9615,
      "step": 93819
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.8071687817573547,
      "learning_rate": 0.000386228614395912,
      "loss": 2.2589,
      "step": 93820
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.8945556282997131,
      "learning_rate": 0.000386224696439133,
      "loss": 2.27,
      "step": 93821
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.7550144791603088,
      "learning_rate": 0.00038622077846632314,
      "loss": 2.2384,
      "step": 93822
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.7852558493614197,
      "learning_rate": 0.0003862168604774833,
      "loss": 2.349,
      "step": 93823
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.8742351531982422,
      "learning_rate": 0.00038621294247261404,
      "loss": 2.1,
      "step": 93824
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.930128812789917,
      "learning_rate": 0.0003862090244517162,
      "loss": 2.1732,
      "step": 93825
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.9265585541725159,
      "learning_rate": 0.0003862051064147905,
      "loss": 2.4227,
      "step": 93826
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.8550072908401489,
      "learning_rate": 0.00038620118836183755,
      "loss": 2.252,
      "step": 93827
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.9431473612785339,
      "learning_rate": 0.0003861972702928582,
      "loss": 2.1333,
      "step": 93828
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.8472043871879578,
      "learning_rate": 0.0003861933522078532,
      "loss": 2.1886,
      "step": 93829
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.9106246829032898,
      "learning_rate": 0.00038618943410682324,
      "loss": 2.0805,
      "step": 93830
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.8160102963447571,
      "learning_rate": 0.00038618551598976897,
      "loss": 2.0669,
      "step": 93831
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.8672460317611694,
      "learning_rate": 0.00038618159785669127,
      "loss": 2.3372,
      "step": 93832
    },
    {
      "epoch": 1.22,
      "grad_norm": 1.2374868392944336,
      "learning_rate": 0.00038617767970759075,
      "loss": 2.2698,
      "step": 93833
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.9893401861190796,
      "learning_rate": 0.0003861737615424681,
      "loss": 2.2322,
      "step": 93834
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.7804430723190308,
      "learning_rate": 0.00038616984336132425,
      "loss": 1.936,
      "step": 93835
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.7633060216903687,
      "learning_rate": 0.00038616592516415964,
      "loss": 2.1424,
      "step": 93836
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.9698523879051208,
      "learning_rate": 0.0003861620069509753,
      "loss": 1.9924,
      "step": 93837
    },
    {
      "epoch": 1.22,
      "grad_norm": 1.0060081481933594,
      "learning_rate": 0.0003861580887217718,
      "loss": 2.0807,
      "step": 93838
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.9271277785301208,
      "learning_rate": 0.00038615417047654975,
      "loss": 2.1222,
      "step": 93839
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.9720779657363892,
      "learning_rate": 0.0003861502522153101,
      "loss": 2.0725,
      "step": 93840
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.9424976706504822,
      "learning_rate": 0.0003861463339380536,
      "loss": 2.1478,
      "step": 93841
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.8366636633872986,
      "learning_rate": 0.00038614241564478076,
      "loss": 2.1982,
      "step": 93842
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.9950581789016724,
      "learning_rate": 0.0003861384973354924,
      "loss": 2.3095,
      "step": 93843
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.9749472141265869,
      "learning_rate": 0.0003861345790101893,
      "loss": 2.0395,
      "step": 93844
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.8484806418418884,
      "learning_rate": 0.00038613066066887214,
      "loss": 2.1095,
      "step": 93845
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.8583514094352722,
      "learning_rate": 0.0003861267423115417,
      "loss": 2.2073,
      "step": 93846
    },
    {
      "epoch": 1.22,
      "grad_norm": 1.1167579889297485,
      "learning_rate": 0.0003861228239381986,
      "loss": 2.0854,
      "step": 93847
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.7890644669532776,
      "learning_rate": 0.0003861189055488438,
      "loss": 2.2155,
      "step": 93848
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.9403137564659119,
      "learning_rate": 0.00038611498714347765,
      "loss": 2.2499,
      "step": 93849
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.8883985877037048,
      "learning_rate": 0.0003861110687221012,
      "loss": 2.1016,
      "step": 93850
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.8635745048522949,
      "learning_rate": 0.00038610715028471513,
      "loss": 1.9862,
      "step": 93851
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.7507983446121216,
      "learning_rate": 0.0003861032318313201,
      "loss": 1.9267,
      "step": 93852
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.8114463686943054,
      "learning_rate": 0.0003860993133619168,
      "loss": 2.3449,
      "step": 93853
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.8311106562614441,
      "learning_rate": 0.000386095394876506,
      "loss": 2.2144,
      "step": 93854
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.85752934217453,
      "learning_rate": 0.00038609147637508855,
      "loss": 2.2358,
      "step": 93855
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.7973061203956604,
      "learning_rate": 0.000386087557857665,
      "loss": 2.1634,
      "step": 93856
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.9782909154891968,
      "learning_rate": 0.00038608363932423614,
      "loss": 1.9394,
      "step": 93857
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.9075478911399841,
      "learning_rate": 0.00038607972077480264,
      "loss": 2.3703,
      "step": 93858
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.8484888076782227,
      "learning_rate": 0.00038607580220936544,
      "loss": 2.3668,
      "step": 93859
    },
    {
      "epoch": 1.22,
      "grad_norm": 1.0304533243179321,
      "learning_rate": 0.00038607188362792505,
      "loss": 2.1585,
      "step": 93860
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.8435097932815552,
      "learning_rate": 0.00038606796503048226,
      "loss": 2.1401,
      "step": 93861
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.8200833797454834,
      "learning_rate": 0.0003860640464170378,
      "loss": 2.0834,
      "step": 93862
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.8806338906288147,
      "learning_rate": 0.00038606012778759244,
      "loss": 2.1633,
      "step": 93863
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.8952205777168274,
      "learning_rate": 0.00038605620914214687,
      "loss": 2.1293,
      "step": 93864
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.8203211426734924,
      "learning_rate": 0.0003860522904807018,
      "loss": 2.196,
      "step": 93865
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.7930205464363098,
      "learning_rate": 0.00038604837180325805,
      "loss": 2.1599,
      "step": 93866
    },
    {
      "epoch": 1.22,
      "grad_norm": 1.0185999870300293,
      "learning_rate": 0.00038604445310981626,
      "loss": 2.1432,
      "step": 93867
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.770842432975769,
      "learning_rate": 0.0003860405344003772,
      "loss": 2.136,
      "step": 93868
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.9923830628395081,
      "learning_rate": 0.00038603661567494156,
      "loss": 2.0781,
      "step": 93869
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.8097230792045593,
      "learning_rate": 0.00038603269693351,
      "loss": 2.1877,
      "step": 93870
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.9767060279846191,
      "learning_rate": 0.0003860287781760834,
      "loss": 2.2805,
      "step": 93871
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.9350929856300354,
      "learning_rate": 0.00038602485940266253,
      "loss": 2.4011,
      "step": 93872
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.8720487356185913,
      "learning_rate": 0.00038602094061324793,
      "loss": 2.3438,
      "step": 93873
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.8091267347335815,
      "learning_rate": 0.0003860170218078404,
      "loss": 2.1721,
      "step": 93874
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.9463298916816711,
      "learning_rate": 0.00038601310298644066,
      "loss": 2.0576,
      "step": 93875
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.8559858202934265,
      "learning_rate": 0.0003860091841490495,
      "loss": 2.2625,
      "step": 93876
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.9685370922088623,
      "learning_rate": 0.0003860052652956676,
      "loss": 2.0541,
      "step": 93877
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.8433654308319092,
      "learning_rate": 0.00038600134642629576,
      "loss": 2.1473,
      "step": 93878
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.8508958220481873,
      "learning_rate": 0.0003859974275409346,
      "loss": 2.1974,
      "step": 93879
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.778477668762207,
      "learning_rate": 0.0003859935086395848,
      "loss": 1.8979,
      "step": 93880
    },
    {
      "epoch": 1.22,
      "grad_norm": 1.050546646118164,
      "learning_rate": 0.0003859895897222473,
      "loss": 1.9249,
      "step": 93881
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.8719611167907715,
      "learning_rate": 0.00038598567078892266,
      "loss": 1.8979,
      "step": 93882
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.880329430103302,
      "learning_rate": 0.00038598175183961165,
      "loss": 2.1639,
      "step": 93883
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.9538164138793945,
      "learning_rate": 0.0003859778328743151,
      "loss": 2.2506,
      "step": 93884
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.8879431486129761,
      "learning_rate": 0.0003859739138930336,
      "loss": 2.3175,
      "step": 93885
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.8281237483024597,
      "learning_rate": 0.0003859699948957679,
      "loss": 2.0927,
      "step": 93886
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.9824280738830566,
      "learning_rate": 0.0003859660758825188,
      "loss": 2.1832,
      "step": 93887
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.9089869260787964,
      "learning_rate": 0.0003859621568532869,
      "loss": 2.5039,
      "step": 93888
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.8292155265808105,
      "learning_rate": 0.00038595823780807303,
      "loss": 2.1083,
      "step": 93889
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.8120296001434326,
      "learning_rate": 0.000385954318746878,
      "loss": 2.0161,
      "step": 93890
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.9416046142578125,
      "learning_rate": 0.00038595039966970237,
      "loss": 2.4428,
      "step": 93891
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.81779545545578,
      "learning_rate": 0.0003859464805765469,
      "loss": 2.1505,
      "step": 93892
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.8013460636138916,
      "learning_rate": 0.00038594256146741245,
      "loss": 2.1172,
      "step": 93893
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.8334774971008301,
      "learning_rate": 0.0003859386423422996,
      "loss": 2.1369,
      "step": 93894
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.9207749962806702,
      "learning_rate": 0.0003859347232012091,
      "loss": 2.1648,
      "step": 93895
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.7945849299430847,
      "learning_rate": 0.00038593080404414185,
      "loss": 2.3385,
      "step": 93896
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.8202307820320129,
      "learning_rate": 0.00038592688487109825,
      "loss": 2.2004,
      "step": 93897
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.8474522233009338,
      "learning_rate": 0.00038592296568207935,
      "loss": 2.2092,
      "step": 93898
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.7627399563789368,
      "learning_rate": 0.0003859190464770857,
      "loss": 1.9364,
      "step": 93899
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.8369829058647156,
      "learning_rate": 0.0003859151272561181,
      "loss": 2.2634,
      "step": 93900
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.874594509601593,
      "learning_rate": 0.00038591120801917725,
      "loss": 2.0069,
      "step": 93901
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.8535833358764648,
      "learning_rate": 0.0003859072887662639,
      "loss": 1.9011,
      "step": 93902
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.964898943901062,
      "learning_rate": 0.00038590336949737865,
      "loss": 2.2165,
      "step": 93903
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.8615849018096924,
      "learning_rate": 0.00038589945021252247,
      "loss": 2.1496,
      "step": 93904
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.8884503245353699,
      "learning_rate": 0.000385895530911696,
      "loss": 2.1688,
      "step": 93905
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.7467501759529114,
      "learning_rate": 0.00038589161159489986,
      "loss": 2.2478,
      "step": 93906
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.7630156874656677,
      "learning_rate": 0.0003858876922621348,
      "loss": 2.2742,
      "step": 93907
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.9031552076339722,
      "learning_rate": 0.0003858837729134017,
      "loss": 2.075,
      "step": 93908
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.8869284987449646,
      "learning_rate": 0.00038587985354870106,
      "loss": 2.2339,
      "step": 93909
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.7972515821456909,
      "learning_rate": 0.00038587593416803377,
      "loss": 1.9127,
      "step": 93910
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.9351168274879456,
      "learning_rate": 0.0003858720147714006,
      "loss": 2.157,
      "step": 93911
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.7609407305717468,
      "learning_rate": 0.00038586809535880217,
      "loss": 2.1861,
      "step": 93912
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.8453615307807922,
      "learning_rate": 0.00038586417593023916,
      "loss": 2.0308,
      "step": 93913
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.9594821929931641,
      "learning_rate": 0.0003858602564857125,
      "loss": 2.0702,
      "step": 93914
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.8397193551063538,
      "learning_rate": 0.00038585633702522274,
      "loss": 2.329,
      "step": 93915
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.8650248646736145,
      "learning_rate": 0.0003858524175487707,
      "loss": 2.0644,
      "step": 93916
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.8002477288246155,
      "learning_rate": 0.000385848498056357,
      "loss": 2.2911,
      "step": 93917
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.955439031124115,
      "learning_rate": 0.0003858445785479825,
      "loss": 2.1,
      "step": 93918
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.990408718585968,
      "learning_rate": 0.0003858406590236478,
      "loss": 2.3296,
      "step": 93919
    },
    {
      "epoch": 1.22,
      "grad_norm": 1.0759140253067017,
      "learning_rate": 0.00038583673948335375,
      "loss": 2.2611,
      "step": 93920
    },
    {
      "epoch": 1.22,
      "grad_norm": 1.0975185632705688,
      "learning_rate": 0.00038583281992710106,
      "loss": 2.1511,
      "step": 93921
    },
    {
      "epoch": 1.22,
      "grad_norm": 1.1610230207443237,
      "learning_rate": 0.0003858289003548904,
      "loss": 2.1401,
      "step": 93922
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.8712261319160461,
      "learning_rate": 0.00038582498076672255,
      "loss": 2.3373,
      "step": 93923
    },
    {
      "epoch": 1.22,
      "grad_norm": 1.0006579160690308,
      "learning_rate": 0.0003858210611625982,
      "loss": 2.0616,
      "step": 93924
    },
    {
      "epoch": 1.22,
      "grad_norm": 1.5233761072158813,
      "learning_rate": 0.00038581714154251815,
      "loss": 2.0964,
      "step": 93925
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.7940706610679626,
      "learning_rate": 0.000385813221906483,
      "loss": 2.1566,
      "step": 93926
    },
    {
      "epoch": 1.22,
      "grad_norm": 1.177594780921936,
      "learning_rate": 0.0003858093022544936,
      "loss": 2.1068,
      "step": 93927
    },
    {
      "epoch": 1.22,
      "grad_norm": 1.3319898843765259,
      "learning_rate": 0.00038580538258655057,
      "loss": 2.3412,
      "step": 93928
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.9793803691864014,
      "learning_rate": 0.0003858014629026547,
      "loss": 2.3,
      "step": 93929
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.9100165963172913,
      "learning_rate": 0.00038579754320280685,
      "loss": 2.1862,
      "step": 93930
    },
    {
      "epoch": 1.22,
      "grad_norm": 1.1761255264282227,
      "learning_rate": 0.0003857936234870074,
      "loss": 2.2915,
      "step": 93931
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.8508106470108032,
      "learning_rate": 0.00038578970375525745,
      "loss": 2.1841,
      "step": 93932
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.9366937875747681,
      "learning_rate": 0.00038578578400755753,
      "loss": 2.346,
      "step": 93933
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.7462396621704102,
      "learning_rate": 0.0003857818642439084,
      "loss": 1.9511,
      "step": 93934
    },
    {
      "epoch": 1.22,
      "grad_norm": 1.023159384727478,
      "learning_rate": 0.00038577794446431087,
      "loss": 2.0343,
      "step": 93935
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.9900116324424744,
      "learning_rate": 0.0003857740246687656,
      "loss": 2.0803,
      "step": 93936
    },
    {
      "epoch": 1.22,
      "grad_norm": 1.043084979057312,
      "learning_rate": 0.00038577010485727327,
      "loss": 2.1048,
      "step": 93937
    },
    {
      "epoch": 1.22,
      "grad_norm": 1.0698416233062744,
      "learning_rate": 0.0003857661850298346,
      "loss": 2.1399,
      "step": 93938
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.862703263759613,
      "learning_rate": 0.00038576226518645043,
      "loss": 1.8374,
      "step": 93939
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.8862258791923523,
      "learning_rate": 0.00038575834532712154,
      "loss": 2.2709,
      "step": 93940
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.8271195292472839,
      "learning_rate": 0.0003857544254518484,
      "loss": 2.2204,
      "step": 93941
    },
    {
      "epoch": 1.22,
      "grad_norm": 1.0195281505584717,
      "learning_rate": 0.00038575050556063196,
      "loss": 2.0825,
      "step": 93942
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.9683443903923035,
      "learning_rate": 0.0003857465856534728,
      "loss": 2.3089,
      "step": 93943
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.7941735982894897,
      "learning_rate": 0.0003857426657303719,
      "loss": 2.1925,
      "step": 93944
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.8987180590629578,
      "learning_rate": 0.00038573874579132973,
      "loss": 2.1205,
      "step": 93945
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.7829390168190002,
      "learning_rate": 0.00038573482583634705,
      "loss": 2.3427,
      "step": 93946
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.8095346689224243,
      "learning_rate": 0.0003857309058654247,
      "loss": 2.1362,
      "step": 93947
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.8205597400665283,
      "learning_rate": 0.0003857269858785634,
      "loss": 2.1645,
      "step": 93948
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.8137980699539185,
      "learning_rate": 0.0003857230658757638,
      "loss": 1.9597,
      "step": 93949
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.8439041972160339,
      "learning_rate": 0.00038571914585702657,
      "loss": 2.3499,
      "step": 93950
    },
    {
      "epoch": 1.22,
      "grad_norm": 1.2099964618682861,
      "learning_rate": 0.0003857152258223527,
      "loss": 2.1761,
      "step": 93951
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.7859372496604919,
      "learning_rate": 0.0003857113057717427,
      "loss": 2.3749,
      "step": 93952
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.9210597276687622,
      "learning_rate": 0.00038570738570519727,
      "loss": 2.2137,
      "step": 93953
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.8365838527679443,
      "learning_rate": 0.00038570346562271735,
      "loss": 2.3606,
      "step": 93954
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.8689596056938171,
      "learning_rate": 0.0003856995455243034,
      "loss": 2.2947,
      "step": 93955
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.7832974791526794,
      "learning_rate": 0.0003856956254099563,
      "loss": 2.1153,
      "step": 93956
    },
    {
      "epoch": 1.22,
      "grad_norm": 1.0261015892028809,
      "learning_rate": 0.00038569170527967685,
      "loss": 2.1111,
      "step": 93957
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.8759766817092896,
      "learning_rate": 0.0003856877851334657,
      "loss": 2.2712,
      "step": 93958
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.7856269478797913,
      "learning_rate": 0.0003856838649713235,
      "loss": 2.1035,
      "step": 93959
    },
    {
      "epoch": 1.22,
      "grad_norm": 1.0006476640701294,
      "learning_rate": 0.00038567994479325106,
      "loss": 2.3402,
      "step": 93960
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.8870353102684021,
      "learning_rate": 0.0003856760245992491,
      "loss": 2.1698,
      "step": 93961
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.8084384799003601,
      "learning_rate": 0.0003856721043893184,
      "loss": 1.9238,
      "step": 93962
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.9103220701217651,
      "learning_rate": 0.00038566818416345964,
      "loss": 2.1957,
      "step": 93963
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.8588762283325195,
      "learning_rate": 0.00038566426392167355,
      "loss": 1.9715,
      "step": 93964
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.8225334286689758,
      "learning_rate": 0.00038566034366396075,
      "loss": 2.2072,
      "step": 93965
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.8738853931427002,
      "learning_rate": 0.0003856564233903222,
      "loss": 2.2436,
      "step": 93966
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.8051856160163879,
      "learning_rate": 0.0003856525031007585,
      "loss": 2.1966,
      "step": 93967
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.857302725315094,
      "learning_rate": 0.0003856485827952703,
      "loss": 2.3459,
      "step": 93968
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.8885042667388916,
      "learning_rate": 0.00038564466247385847,
      "loss": 2.2494,
      "step": 93969
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.8262380361557007,
      "learning_rate": 0.00038564074213652365,
      "loss": 2.1275,
      "step": 93970
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.9311004281044006,
      "learning_rate": 0.0003856368217832666,
      "loss": 1.9867,
      "step": 93971
    },
    {
      "epoch": 1.22,
      "grad_norm": 1.041371464729309,
      "learning_rate": 0.00038563290141408813,
      "loss": 2.3876,
      "step": 93972
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.8026537895202637,
      "learning_rate": 0.0003856289810289888,
      "loss": 2.2105,
      "step": 93973
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.7947921752929688,
      "learning_rate": 0.00038562506062796934,
      "loss": 2.2405,
      "step": 93974
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.8278080224990845,
      "learning_rate": 0.0003856211402110308,
      "loss": 2.0692,
      "step": 93975
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.8069414496421814,
      "learning_rate": 0.0003856172197781735,
      "loss": 2.3453,
      "step": 93976
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.8196004033088684,
      "learning_rate": 0.0003856132993293983,
      "loss": 2.5007,
      "step": 93977
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.871423065662384,
      "learning_rate": 0.00038560937886470613,
      "loss": 2.2912,
      "step": 93978
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.8186788558959961,
      "learning_rate": 0.00038560545838409746,
      "loss": 2.0718,
      "step": 93979
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.8484766483306885,
      "learning_rate": 0.00038560153788757314,
      "loss": 2.0012,
      "step": 93980
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.779172420501709,
      "learning_rate": 0.00038559761737513393,
      "loss": 2.3316,
      "step": 93981
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.8329629302024841,
      "learning_rate": 0.0003855936968467804,
      "loss": 2.3122,
      "step": 93982
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.9726415872573853,
      "learning_rate": 0.00038558977630251347,
      "loss": 2.2693,
      "step": 93983
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.7410762310028076,
      "learning_rate": 0.0003855858557423337,
      "loss": 2.1296,
      "step": 93984
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.9299248456954956,
      "learning_rate": 0.00038558193516624205,
      "loss": 2.4389,
      "step": 93985
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.9019216895103455,
      "learning_rate": 0.000385578014574239,
      "loss": 2.0841,
      "step": 93986
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.7818056344985962,
      "learning_rate": 0.00038557409396632535,
      "loss": 2.0856,
      "step": 93987
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.8525591492652893,
      "learning_rate": 0.00038557017334250193,
      "loss": 2.2592,
      "step": 93988
    },
    {
      "epoch": 1.22,
      "grad_norm": 1.0169689655303955,
      "learning_rate": 0.0003855662527027694,
      "loss": 2.2259,
      "step": 93989
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.8232703804969788,
      "learning_rate": 0.0003855623320471285,
      "loss": 2.271,
      "step": 93990
    },
    {
      "epoch": 1.22,
      "grad_norm": 1.0098298788070679,
      "learning_rate": 0.0003855584113755799,
      "loss": 2.2434,
      "step": 93991
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.8449119329452515,
      "learning_rate": 0.0003855544906881244,
      "loss": 2.341,
      "step": 93992
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.9970188140869141,
      "learning_rate": 0.00038555056998476264,
      "loss": 2.2859,
      "step": 93993
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.8643110394477844,
      "learning_rate": 0.00038554664926549555,
      "loss": 2.2258,
      "step": 93994
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.8262730836868286,
      "learning_rate": 0.00038554272853032364,
      "loss": 2.0691,
      "step": 93995
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.8653867244720459,
      "learning_rate": 0.0003855388077792477,
      "loss": 2.2762,
      "step": 93996
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.8326102495193481,
      "learning_rate": 0.0003855348870122685,
      "loss": 2.1835,
      "step": 93997
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.8334202766418457,
      "learning_rate": 0.0003855309662293868,
      "loss": 2.1836,
      "step": 93998
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.9238207340240479,
      "learning_rate": 0.0003855270454306032,
      "loss": 2.1051,
      "step": 93999
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.9758351445198059,
      "learning_rate": 0.0003855231246159185,
      "loss": 2.2554,
      "step": 94000
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.8279000520706177,
      "learning_rate": 0.0003855192037853335,
      "loss": 2.2044,
      "step": 94001
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.8682078123092651,
      "learning_rate": 0.00038551528293884885,
      "loss": 2.0636,
      "step": 94002
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.8653808236122131,
      "learning_rate": 0.0003855113620764653,
      "loss": 2.341,
      "step": 94003
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.7848419547080994,
      "learning_rate": 0.00038550744119818356,
      "loss": 2.318,
      "step": 94004
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.8210131525993347,
      "learning_rate": 0.0003855035203040044,
      "loss": 2.2444,
      "step": 94005
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.785982072353363,
      "learning_rate": 0.0003854995993939285,
      "loss": 2.0915,
      "step": 94006
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.8129271864891052,
      "learning_rate": 0.0003854956784679566,
      "loss": 2.4908,
      "step": 94007
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.8143530488014221,
      "learning_rate": 0.0003854917575260894,
      "loss": 2.0295,
      "step": 94008
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.7663519978523254,
      "learning_rate": 0.0003854878365683278,
      "loss": 2.4083,
      "step": 94009
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.863158106803894,
      "learning_rate": 0.0003854839155946723,
      "loss": 1.9105,
      "step": 94010
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.7723687291145325,
      "learning_rate": 0.0003854799946051237,
      "loss": 2.1754,
      "step": 94011
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.7883620262145996,
      "learning_rate": 0.0003854760735996829,
      "loss": 2.0363,
      "step": 94012
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.9498263597488403,
      "learning_rate": 0.00038547215257835033,
      "loss": 2.3535,
      "step": 94013
    },
    {
      "epoch": 1.22,
      "grad_norm": 1.1354796886444092,
      "learning_rate": 0.00038546823154112685,
      "loss": 2.2213,
      "step": 94014
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.9221222996711731,
      "learning_rate": 0.00038546431048801336,
      "loss": 2.2955,
      "step": 94015
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.9780751466751099,
      "learning_rate": 0.0003854603894190104,
      "loss": 2.0347,
      "step": 94016
    },
    {
      "epoch": 1.22,
      "grad_norm": 1.0070829391479492,
      "learning_rate": 0.00038545646833411867,
      "loss": 2.2988,
      "step": 94017
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.8145137429237366,
      "learning_rate": 0.000385452547233339,
      "loss": 2.1529,
      "step": 94018
    },
    {
      "epoch": 1.22,
      "grad_norm": 1.1764615774154663,
      "learning_rate": 0.0003854486261166721,
      "loss": 1.9355,
      "step": 94019
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.9226595759391785,
      "learning_rate": 0.00038544470498411864,
      "loss": 2.1426,
      "step": 94020
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.8675881624221802,
      "learning_rate": 0.0003854407838356795,
      "loss": 2.2597,
      "step": 94021
    },
    {
      "epoch": 1.22,
      "grad_norm": 1.0013645887374878,
      "learning_rate": 0.00038543686267135526,
      "loss": 2.2815,
      "step": 94022
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.9130496978759766,
      "learning_rate": 0.0003854329414911466,
      "loss": 2.1605,
      "step": 94023
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.8818677663803101,
      "learning_rate": 0.00038542902029505444,
      "loss": 2.0046,
      "step": 94024
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.8845403790473938,
      "learning_rate": 0.00038542509908307947,
      "loss": 2.0938,
      "step": 94025
    },
    {
      "epoch": 1.22,
      "grad_norm": 1.006413459777832,
      "learning_rate": 0.00038542117785522224,
      "loss": 2.2635,
      "step": 94026
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.9391209483146667,
      "learning_rate": 0.0003854172566114837,
      "loss": 2.2798,
      "step": 94027
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.8190628886222839,
      "learning_rate": 0.00038541333535186437,
      "loss": 2.1296,
      "step": 94028
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.9429672360420227,
      "learning_rate": 0.0003854094140763651,
      "loss": 2.1857,
      "step": 94029
    },
    {
      "epoch": 1.22,
      "grad_norm": 1.1037877798080444,
      "learning_rate": 0.00038540549278498666,
      "loss": 2.1581,
      "step": 94030
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.9696435928344727,
      "learning_rate": 0.00038540157147772967,
      "loss": 2.2089,
      "step": 94031
    },
    {
      "epoch": 1.22,
      "grad_norm": 1.0138903856277466,
      "learning_rate": 0.0003853976501545951,
      "loss": 2.2509,
      "step": 94032
    },
    {
      "epoch": 1.22,
      "grad_norm": 1.1329349279403687,
      "learning_rate": 0.00038539372881558334,
      "loss": 2.4394,
      "step": 94033
    },
    {
      "epoch": 1.22,
      "grad_norm": 1.2691919803619385,
      "learning_rate": 0.00038538980746069524,
      "loss": 2.2155,
      "step": 94034
    },
    {
      "epoch": 1.22,
      "grad_norm": 1.0850261449813843,
      "learning_rate": 0.0003853858860899316,
      "loss": 1.9238,
      "step": 94035
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.8329902291297913,
      "learning_rate": 0.0003853819647032931,
      "loss": 2.4659,
      "step": 94036
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.9335665106773376,
      "learning_rate": 0.00038537804330078057,
      "loss": 1.9992,
      "step": 94037
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.9274067878723145,
      "learning_rate": 0.00038537412188239454,
      "loss": 2.2264,
      "step": 94038
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.9856181144714355,
      "learning_rate": 0.00038537020044813596,
      "loss": 2.2852,
      "step": 94039
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.8840373754501343,
      "learning_rate": 0.0003853662789980054,
      "loss": 2.2003,
      "step": 94040
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.8158826231956482,
      "learning_rate": 0.00038536235753200367,
      "loss": 2.1178,
      "step": 94041
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.9241328835487366,
      "learning_rate": 0.0003853584360501314,
      "loss": 2.3942,
      "step": 94042
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.8393216133117676,
      "learning_rate": 0.00038535451455238933,
      "loss": 2.1965,
      "step": 94043
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.8826835751533508,
      "learning_rate": 0.00038535059303877837,
      "loss": 2.1281,
      "step": 94044
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.7481151819229126,
      "learning_rate": 0.0003853466715092991,
      "loss": 2.0798,
      "step": 94045
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.9768670797348022,
      "learning_rate": 0.00038534274996395227,
      "loss": 2.4369,
      "step": 94046
    },
    {
      "epoch": 1.22,
      "grad_norm": 1.1268905401229858,
      "learning_rate": 0.00038533882840273854,
      "loss": 2.1868,
      "step": 94047
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.7846325635910034,
      "learning_rate": 0.0003853349068256588,
      "loss": 2.1842,
      "step": 94048
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.8725501894950867,
      "learning_rate": 0.00038533098523271364,
      "loss": 2.3821,
      "step": 94049
    },
    {
      "epoch": 1.22,
      "grad_norm": 1.1010115146636963,
      "learning_rate": 0.0003853270636239038,
      "loss": 2.4978,
      "step": 94050
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.8632315993309021,
      "learning_rate": 0.0003853231419992301,
      "loss": 2.1794,
      "step": 94051
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.9853323698043823,
      "learning_rate": 0.0003853192203586932,
      "loss": 2.2116,
      "step": 94052
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.8062126636505127,
      "learning_rate": 0.0003853152987022939,
      "loss": 2.1516,
      "step": 94053
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.9053832292556763,
      "learning_rate": 0.00038531137703003284,
      "loss": 2.1084,
      "step": 94054
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.860177218914032,
      "learning_rate": 0.0003853074553419108,
      "loss": 2.2316,
      "step": 94055
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.8592917919158936,
      "learning_rate": 0.0003853035336379285,
      "loss": 2.2473,
      "step": 94056
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.8579545617103577,
      "learning_rate": 0.00038529961191808665,
      "loss": 2.3025,
      "step": 94057
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.8396880030632019,
      "learning_rate": 0.000385295690182386,
      "loss": 2.1102,
      "step": 94058
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.785016655921936,
      "learning_rate": 0.0003852917684308272,
      "loss": 2.0429,
      "step": 94059
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.7929885387420654,
      "learning_rate": 0.00038528784666341104,
      "loss": 2.1542,
      "step": 94060
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.9835507273674011,
      "learning_rate": 0.0003852839248801384,
      "loss": 2.3,
      "step": 94061
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.7720860242843628,
      "learning_rate": 0.00038528000308100973,
      "loss": 2.1659,
      "step": 94062
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.8186091780662537,
      "learning_rate": 0.00038527608126602596,
      "loss": 2.3099,
      "step": 94063
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.7905296683311462,
      "learning_rate": 0.00038527215943518785,
      "loss": 2.1389,
      "step": 94064
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.8453647494316101,
      "learning_rate": 0.0003852682375884959,
      "loss": 2.0665,
      "step": 94065
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.8618143200874329,
      "learning_rate": 0.000385264315725951,
      "loss": 2.2705,
      "step": 94066
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.8556624054908752,
      "learning_rate": 0.0003852603938475539,
      "loss": 2.2539,
      "step": 94067
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.7352030277252197,
      "learning_rate": 0.00038525647195330527,
      "loss": 2.1891,
      "step": 94068
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.7984911203384399,
      "learning_rate": 0.00038525255004320574,
      "loss": 2.3192,
      "step": 94069
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.8173933029174805,
      "learning_rate": 0.0003852486281172564,
      "loss": 2.1253,
      "step": 94070
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.754587709903717,
      "learning_rate": 0.0003852447061754575,
      "loss": 1.9386,
      "step": 94071
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.810731828212738,
      "learning_rate": 0.00038524078421781007,
      "loss": 2.204,
      "step": 94072
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.8465047478675842,
      "learning_rate": 0.0003852368622443148,
      "loss": 2.3254,
      "step": 94073
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.7482575178146362,
      "learning_rate": 0.0003852329402549724,
      "loss": 2.0875,
      "step": 94074
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.7893854379653931,
      "learning_rate": 0.00038522901824978353,
      "loss": 2.0441,
      "step": 94075
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.9146608114242554,
      "learning_rate": 0.00038522509622874903,
      "loss": 1.9906,
      "step": 94076
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.9530184864997864,
      "learning_rate": 0.0003852211741918695,
      "loss": 2.0851,
      "step": 94077
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.8173010945320129,
      "learning_rate": 0.0003852172521391458,
      "loss": 2.0548,
      "step": 94078
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.8446654081344604,
      "learning_rate": 0.0003852133300705786,
      "loss": 2.0736,
      "step": 94079
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.928857684135437,
      "learning_rate": 0.00038520940798616863,
      "loss": 2.0331,
      "step": 94080
    },
    {
      "epoch": 1.22,
      "grad_norm": 0.9229807257652283,
      "learning_rate": 0.00038520548588591663,
      "loss": 2.2396,
      "step": 94081
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.7662559747695923,
      "learning_rate": 0.00038520156376982336,
      "loss": 2.272,
      "step": 94082
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.8169102072715759,
      "learning_rate": 0.00038519764163788946,
      "loss": 2.2279,
      "step": 94083
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.929025411605835,
      "learning_rate": 0.00038519371949011564,
      "loss": 2.2563,
      "step": 94084
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.9257906675338745,
      "learning_rate": 0.0003851897973265028,
      "loss": 2.2123,
      "step": 94085
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.870399534702301,
      "learning_rate": 0.0003851858751470516,
      "loss": 2.3374,
      "step": 94086
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.9071251153945923,
      "learning_rate": 0.0003851819529517626,
      "loss": 2.1145,
      "step": 94087
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.8726695775985718,
      "learning_rate": 0.0003851780307406368,
      "loss": 2.2784,
      "step": 94088
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.9610780477523804,
      "learning_rate": 0.0003851741085136747,
      "loss": 2.1709,
      "step": 94089
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.8880822658538818,
      "learning_rate": 0.00038517018627087714,
      "loss": 2.1724,
      "step": 94090
    },
    {
      "epoch": 1.23,
      "grad_norm": 1.0454128980636597,
      "learning_rate": 0.00038516626401224495,
      "loss": 2.2598,
      "step": 94091
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.9173315167427063,
      "learning_rate": 0.0003851623417377787,
      "loss": 2.1185,
      "step": 94092
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.890670895576477,
      "learning_rate": 0.00038515841944747906,
      "loss": 2.3366,
      "step": 94093
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.8256529569625854,
      "learning_rate": 0.000385154497141347,
      "loss": 2.1774,
      "step": 94094
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.7824119925498962,
      "learning_rate": 0.00038515057481938297,
      "loss": 2.2694,
      "step": 94095
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.8328878283500671,
      "learning_rate": 0.0003851466524815879,
      "loss": 2.1961,
      "step": 94096
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.7881881594657898,
      "learning_rate": 0.0003851427301279625,
      "loss": 2.2998,
      "step": 94097
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.8127003908157349,
      "learning_rate": 0.00038513880775850736,
      "loss": 2.0585,
      "step": 94098
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.8732307553291321,
      "learning_rate": 0.0003851348853732234,
      "loss": 2.2235,
      "step": 94099
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.856106162071228,
      "learning_rate": 0.00038513096297211125,
      "loss": 2.2591,
      "step": 94100
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.8390441536903381,
      "learning_rate": 0.0003851270405551716,
      "loss": 2.1203,
      "step": 94101
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.8358861207962036,
      "learning_rate": 0.00038512311812240527,
      "loss": 2.1125,
      "step": 94102
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.9422858953475952,
      "learning_rate": 0.00038511919567381294,
      "loss": 2.1602,
      "step": 94103
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.8331666588783264,
      "learning_rate": 0.0003851152732093953,
      "loss": 2.3437,
      "step": 94104
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.9866182208061218,
      "learning_rate": 0.0003851113507291532,
      "loss": 2.3604,
      "step": 94105
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.9526065587997437,
      "learning_rate": 0.00038510742823308727,
      "loss": 2.1788,
      "step": 94106
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.8741582632064819,
      "learning_rate": 0.0003851035057211982,
      "loss": 2.168,
      "step": 94107
    },
    {
      "epoch": 1.23,
      "grad_norm": 1.0859291553497314,
      "learning_rate": 0.0003850995831934868,
      "loss": 2.1576,
      "step": 94108
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.9123268127441406,
      "learning_rate": 0.0003850956606499539,
      "loss": 2.0358,
      "step": 94109
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.8540319204330444,
      "learning_rate": 0.0003850917380906,
      "loss": 2.2385,
      "step": 94110
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.9129671454429626,
      "learning_rate": 0.00038508781551542596,
      "loss": 2.224,
      "step": 94111
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.9991889595985413,
      "learning_rate": 0.0003850838929244324,
      "loss": 2.1539,
      "step": 94112
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.8033132553100586,
      "learning_rate": 0.0003850799703176203,
      "loss": 2.2037,
      "step": 94113
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.8573119640350342,
      "learning_rate": 0.00038507604769499014,
      "loss": 2.3197,
      "step": 94114
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.9053441286087036,
      "learning_rate": 0.00038507212505654273,
      "loss": 2.1933,
      "step": 94115
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.8846114277839661,
      "learning_rate": 0.00038506820240227877,
      "loss": 2.3039,
      "step": 94116
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.8126371502876282,
      "learning_rate": 0.00038506427973219917,
      "loss": 1.956,
      "step": 94117
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.9013044238090515,
      "learning_rate": 0.00038506035704630436,
      "loss": 2.1119,
      "step": 94118
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.7925348877906799,
      "learning_rate": 0.0003850564343445953,
      "loss": 2.0939,
      "step": 94119
    },
    {
      "epoch": 1.23,
      "grad_norm": 1.0702850818634033,
      "learning_rate": 0.00038505251162707267,
      "loss": 2.1614,
      "step": 94120
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.8948254585266113,
      "learning_rate": 0.0003850485888937371,
      "loss": 2.3929,
      "step": 94121
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.8413371443748474,
      "learning_rate": 0.00038504466614458944,
      "loss": 1.9879,
      "step": 94122
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.8726986646652222,
      "learning_rate": 0.0003850407433796303,
      "loss": 2.2838,
      "step": 94123
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.9332106709480286,
      "learning_rate": 0.0003850368205988606,
      "loss": 2.1411,
      "step": 94124
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.830572247505188,
      "learning_rate": 0.00038503289780228087,
      "loss": 2.0839,
      "step": 94125
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.8030413389205933,
      "learning_rate": 0.0003850289749898919,
      "loss": 2.189,
      "step": 94126
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.8767715096473694,
      "learning_rate": 0.0003850250521616945,
      "loss": 1.9414,
      "step": 94127
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.7859722375869751,
      "learning_rate": 0.00038502112931768935,
      "loss": 2.1688,
      "step": 94128
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.8438982367515564,
      "learning_rate": 0.0003850172064578771,
      "loss": 2.2561,
      "step": 94129
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.8488173484802246,
      "learning_rate": 0.00038501328358225855,
      "loss": 2.008,
      "step": 94130
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.8881440162658691,
      "learning_rate": 0.0003850093606908345,
      "loss": 2.0756,
      "step": 94131
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.7688820362091064,
      "learning_rate": 0.00038500543778360554,
      "loss": 2.0337,
      "step": 94132
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.821816086769104,
      "learning_rate": 0.00038500151486057247,
      "loss": 2.2273,
      "step": 94133
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.8553999662399292,
      "learning_rate": 0.00038499759192173606,
      "loss": 2.0658,
      "step": 94134
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.8767535090446472,
      "learning_rate": 0.0003849936689670969,
      "loss": 2.0573,
      "step": 94135
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.8169050216674805,
      "learning_rate": 0.0003849897459966559,
      "loss": 2.2961,
      "step": 94136
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.9013545513153076,
      "learning_rate": 0.00038498582301041374,
      "loss": 2.3959,
      "step": 94137
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.8072664737701416,
      "learning_rate": 0.00038498190000837095,
      "loss": 2.1063,
      "step": 94138
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.8412729501724243,
      "learning_rate": 0.0003849779769905285,
      "loss": 2.0465,
      "step": 94139
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.8496372699737549,
      "learning_rate": 0.00038497405395688714,
      "loss": 2.5012,
      "step": 94140
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.7953763008117676,
      "learning_rate": 0.0003849701309074474,
      "loss": 2.1944,
      "step": 94141
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.9118503928184509,
      "learning_rate": 0.00038496620784221006,
      "loss": 2.4772,
      "step": 94142
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.789648175239563,
      "learning_rate": 0.00038496228476117604,
      "loss": 2.2466,
      "step": 94143
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.8891419172286987,
      "learning_rate": 0.00038495836166434585,
      "loss": 2.2441,
      "step": 94144
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.9739599823951721,
      "learning_rate": 0.0003849544385517203,
      "loss": 2.0195,
      "step": 94145
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.850967526435852,
      "learning_rate": 0.0003849505154233002,
      "loss": 2.3339,
      "step": 94146
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.8518719673156738,
      "learning_rate": 0.000384946592279086,
      "loss": 2.0764,
      "step": 94147
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.8848732709884644,
      "learning_rate": 0.00038494266911907876,
      "loss": 2.2402,
      "step": 94148
    },
    {
      "epoch": 1.23,
      "grad_norm": 1.0665936470031738,
      "learning_rate": 0.0003849387459432791,
      "loss": 2.1549,
      "step": 94149
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.7941968441009521,
      "learning_rate": 0.0003849348227516876,
      "loss": 2.0067,
      "step": 94150
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.8812691569328308,
      "learning_rate": 0.00038493089954430516,
      "loss": 2.2563,
      "step": 94151
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.8924147486686707,
      "learning_rate": 0.00038492697632113256,
      "loss": 2.1721,
      "step": 94152
    },
    {
      "epoch": 1.23,
      "grad_norm": 1.011495590209961,
      "learning_rate": 0.00038492305308217035,
      "loss": 2.2522,
      "step": 94153
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.9222906827926636,
      "learning_rate": 0.00038491912982741933,
      "loss": 2.3673,
      "step": 94154
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.9021216034889221,
      "learning_rate": 0.0003849152065568803,
      "loss": 2.4413,
      "step": 94155
    },
    {
      "epoch": 1.23,
      "grad_norm": 1.0246665477752686,
      "learning_rate": 0.0003849112832705539,
      "loss": 2.1517,
      "step": 94156
    },
    {
      "epoch": 1.23,
      "grad_norm": 1.4095216989517212,
      "learning_rate": 0.00038490735996844083,
      "loss": 2.023,
      "step": 94157
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.8066607713699341,
      "learning_rate": 0.000384903436650542,
      "loss": 2.0946,
      "step": 94158
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.8346040844917297,
      "learning_rate": 0.0003848995133168579,
      "loss": 2.0117,
      "step": 94159
    },
    {
      "epoch": 1.23,
      "grad_norm": 1.2226883172988892,
      "learning_rate": 0.0003848955899673894,
      "loss": 2.0012,
      "step": 94160
    },
    {
      "epoch": 1.23,
      "grad_norm": 1.1304136514663696,
      "learning_rate": 0.0003848916666021373,
      "loss": 2.2068,
      "step": 94161
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.9088578820228577,
      "learning_rate": 0.0003848877432211021,
      "loss": 2.2178,
      "step": 94162
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.9828610420227051,
      "learning_rate": 0.0003848838198242848,
      "loss": 2.3311,
      "step": 94163
    },
    {
      "epoch": 1.23,
      "grad_norm": 1.3807094097137451,
      "learning_rate": 0.00038487989641168595,
      "loss": 2.1508,
      "step": 94164
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.8460904359817505,
      "learning_rate": 0.0003848759729833063,
      "loss": 2.0867,
      "step": 94165
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.8939948081970215,
      "learning_rate": 0.0003848720495391465,
      "loss": 2.2631,
      "step": 94166
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.9412001967430115,
      "learning_rate": 0.0003848681260792076,
      "loss": 1.852,
      "step": 94167
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.9649364352226257,
      "learning_rate": 0.00038486420260349004,
      "loss": 1.9422,
      "step": 94168
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.8517218828201294,
      "learning_rate": 0.00038486027911199453,
      "loss": 2.2325,
      "step": 94169
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.8235869407653809,
      "learning_rate": 0.000384856355604722,
      "loss": 2.318,
      "step": 94170
    },
    {
      "epoch": 1.23,
      "grad_norm": 1.1191301345825195,
      "learning_rate": 0.000384852432081673,
      "loss": 2.0875,
      "step": 94171
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.8630365133285522,
      "learning_rate": 0.0003848485085428484,
      "loss": 2.2779,
      "step": 94172
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.8160399794578552,
      "learning_rate": 0.00038484458498824883,
      "loss": 2.0823,
      "step": 94173
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.8046332597732544,
      "learning_rate": 0.000384840661417875,
      "loss": 2.4826,
      "step": 94174
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.8646311163902283,
      "learning_rate": 0.0003848367378317277,
      "loss": 2.1084,
      "step": 94175
    },
    {
      "epoch": 1.23,
      "grad_norm": 1.5664228200912476,
      "learning_rate": 0.00038483281422980775,
      "loss": 2.321,
      "step": 94176
    },
    {
      "epoch": 1.23,
      "grad_norm": 1.1400948762893677,
      "learning_rate": 0.00038482889061211567,
      "loss": 2.2339,
      "step": 94177
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.8857890963554382,
      "learning_rate": 0.00038482496697865234,
      "loss": 1.8523,
      "step": 94178
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.9384742975234985,
      "learning_rate": 0.0003848210433294185,
      "loss": 2.3771,
      "step": 94179
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.9565339684486389,
      "learning_rate": 0.0003848171196644147,
      "loss": 2.3594,
      "step": 94180
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.8151049613952637,
      "learning_rate": 0.0003848131959836419,
      "loss": 2.1292,
      "step": 94181
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.8525887727737427,
      "learning_rate": 0.0003848092722871007,
      "loss": 2.0055,
      "step": 94182
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.8802593946456909,
      "learning_rate": 0.0003848053485747918,
      "loss": 2.1669,
      "step": 94183
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.8507543206214905,
      "learning_rate": 0.000384801424846716,
      "loss": 2.2331,
      "step": 94184
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.8750978112220764,
      "learning_rate": 0.0003847975011028741,
      "loss": 2.1771,
      "step": 94185
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.8506460189819336,
      "learning_rate": 0.0003847935773432667,
      "loss": 2.0773,
      "step": 94186
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.992131769657135,
      "learning_rate": 0.00038478965356789454,
      "loss": 1.9548,
      "step": 94187
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.8430218696594238,
      "learning_rate": 0.00038478572977675835,
      "loss": 2.2447,
      "step": 94188
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.7710174918174744,
      "learning_rate": 0.000384781805969859,
      "loss": 2.1658,
      "step": 94189
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.8918041586875916,
      "learning_rate": 0.0003847778821471971,
      "loss": 2.1478,
      "step": 94190
    },
    {
      "epoch": 1.23,
      "grad_norm": 1.2177455425262451,
      "learning_rate": 0.0003847739583087733,
      "loss": 2.1265,
      "step": 94191
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.8550278544425964,
      "learning_rate": 0.00038477003445458853,
      "loss": 2.2005,
      "step": 94192
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.9620959162712097,
      "learning_rate": 0.0003847661105846433,
      "loss": 2.3255,
      "step": 94193
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.9964462518692017,
      "learning_rate": 0.00038476218669893845,
      "loss": 2.112,
      "step": 94194
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.772843062877655,
      "learning_rate": 0.0003847582627974749,
      "loss": 2.0149,
      "step": 94195
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.9023983478546143,
      "learning_rate": 0.00038475433888025297,
      "loss": 2.4792,
      "step": 94196
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.8321839570999146,
      "learning_rate": 0.00038475041494727365,
      "loss": 2.2366,
      "step": 94197
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.9018643498420715,
      "learning_rate": 0.0003847464909985377,
      "loss": 1.9834,
      "step": 94198
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.9014692306518555,
      "learning_rate": 0.00038474256703404573,
      "loss": 2.1375,
      "step": 94199
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.8296590447425842,
      "learning_rate": 0.00038473864305379855,
      "loss": 2.2947,
      "step": 94200
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.8371555805206299,
      "learning_rate": 0.0003847347190577969,
      "loss": 2.1322,
      "step": 94201
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.9529421329498291,
      "learning_rate": 0.0003847307950460414,
      "loss": 2.3393,
      "step": 94202
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.8839572668075562,
      "learning_rate": 0.00038472687101853275,
      "loss": 2.2059,
      "step": 94203
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.8209508061408997,
      "learning_rate": 0.0003847229469752719,
      "loss": 2.0662,
      "step": 94204
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.9498443007469177,
      "learning_rate": 0.0003847190229162595,
      "loss": 2.1247,
      "step": 94205
    },
    {
      "epoch": 1.23,
      "grad_norm": 1.0889983177185059,
      "learning_rate": 0.0003847150988414961,
      "loss": 2.3026,
      "step": 94206
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.8907216787338257,
      "learning_rate": 0.00038471117475098266,
      "loss": 2.1331,
      "step": 94207
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.8810898661613464,
      "learning_rate": 0.0003847072506447198,
      "loss": 2.366,
      "step": 94208
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.9548719525337219,
      "learning_rate": 0.0003847033265227082,
      "loss": 2.3312,
      "step": 94209
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.8529475927352905,
      "learning_rate": 0.0003846994023849487,
      "loss": 2.107,
      "step": 94210
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.7694559693336487,
      "learning_rate": 0.00038469547823144197,
      "loss": 2.2457,
      "step": 94211
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.8762230277061462,
      "learning_rate": 0.0003846915540621888,
      "loss": 1.969,
      "step": 94212
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.8883614540100098,
      "learning_rate": 0.00038468762987718983,
      "loss": 2.3731,
      "step": 94213
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.9259026646614075,
      "learning_rate": 0.0003846837056764458,
      "loss": 2.0866,
      "step": 94214
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.8880531191825867,
      "learning_rate": 0.00038467978145995753,
      "loss": 2.2225,
      "step": 94215
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.8481050133705139,
      "learning_rate": 0.00038467585722772567,
      "loss": 1.8368,
      "step": 94216
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.7925577759742737,
      "learning_rate": 0.00038467193297975097,
      "loss": 2.179,
      "step": 94217
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.8570266366004944,
      "learning_rate": 0.0003846680087160341,
      "loss": 2.2966,
      "step": 94218
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.8931347727775574,
      "learning_rate": 0.000384664084436576,
      "loss": 2.2367,
      "step": 94219
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.8264577984809875,
      "learning_rate": 0.00038466016014137704,
      "loss": 1.9891,
      "step": 94220
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.7966234087944031,
      "learning_rate": 0.00038465623583043833,
      "loss": 2.146,
      "step": 94221
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.8816819787025452,
      "learning_rate": 0.00038465231150376037,
      "loss": 1.8038,
      "step": 94222
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.7915287613868713,
      "learning_rate": 0.00038464838716134397,
      "loss": 2.2646,
      "step": 94223
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.843313455581665,
      "learning_rate": 0.00038464446280318977,
      "loss": 2.2168,
      "step": 94224
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.9616535305976868,
      "learning_rate": 0.00038464053842929865,
      "loss": 2.2641,
      "step": 94225
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.8426799178123474,
      "learning_rate": 0.0003846366140396712,
      "loss": 2.4413,
      "step": 94226
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.8143842816352844,
      "learning_rate": 0.0003846326896343082,
      "loss": 2.1898,
      "step": 94227
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.7594031691551208,
      "learning_rate": 0.00038462876521321043,
      "loss": 1.8056,
      "step": 94228
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.8223464488983154,
      "learning_rate": 0.0003846248407763785,
      "loss": 1.9857,
      "step": 94229
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.9691091775894165,
      "learning_rate": 0.0003846209163238133,
      "loss": 2.0784,
      "step": 94230
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.846820056438446,
      "learning_rate": 0.0003846169918555155,
      "loss": 2.255,
      "step": 94231
    },
    {
      "epoch": 1.23,
      "grad_norm": 1.0763719081878662,
      "learning_rate": 0.00038461306737148574,
      "loss": 2.3514,
      "step": 94232
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.958671510219574,
      "learning_rate": 0.00038460914287172476,
      "loss": 2.2784,
      "step": 94233
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.7624472975730896,
      "learning_rate": 0.0003846052183562334,
      "loss": 2.1724,
      "step": 94234
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.9076208472251892,
      "learning_rate": 0.0003846012938250124,
      "loss": 2.2422,
      "step": 94235
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.9919973015785217,
      "learning_rate": 0.00038459736927806226,
      "loss": 2.3551,
      "step": 94236
    },
    {
      "epoch": 1.23,
      "grad_norm": 1.069403886795044,
      "learning_rate": 0.00038459344471538406,
      "loss": 2.2294,
      "step": 94237
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.7840549945831299,
      "learning_rate": 0.0003845895201369782,
      "loss": 1.8537,
      "step": 94238
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.8741151094436646,
      "learning_rate": 0.00038458559554284557,
      "loss": 2.1014,
      "step": 94239
    },
    {
      "epoch": 1.23,
      "grad_norm": 1.0244017839431763,
      "learning_rate": 0.000384581670932987,
      "loss": 2.302,
      "step": 94240
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.892224907875061,
      "learning_rate": 0.00038457774630740296,
      "loss": 2.1699,
      "step": 94241
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.7394649982452393,
      "learning_rate": 0.00038457382166609435,
      "loss": 2.0419,
      "step": 94242
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.9151251912117004,
      "learning_rate": 0.000384569897009062,
      "loss": 2.4252,
      "step": 94243
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.8888706564903259,
      "learning_rate": 0.0003845659723363064,
      "loss": 2.2632,
      "step": 94244
    },
    {
      "epoch": 1.23,
      "grad_norm": 1.1096786260604858,
      "learning_rate": 0.00038456204764782836,
      "loss": 2.3747,
      "step": 94245
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.8325554728507996,
      "learning_rate": 0.0003845581229436287,
      "loss": 2.1761,
      "step": 94246
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.8341354727745056,
      "learning_rate": 0.00038455419822370807,
      "loss": 2.108,
      "step": 94247
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.801294207572937,
      "learning_rate": 0.00038455027348806715,
      "loss": 2.1272,
      "step": 94248
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.9812988638877869,
      "learning_rate": 0.00038454634873670687,
      "loss": 2.1317,
      "step": 94249
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.9501768350601196,
      "learning_rate": 0.00038454242396962767,
      "loss": 2.0113,
      "step": 94250
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.8468992114067078,
      "learning_rate": 0.0003845384991868305,
      "loss": 2.1322,
      "step": 94251
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.7983776330947876,
      "learning_rate": 0.00038453457438831606,
      "loss": 2.1641,
      "step": 94252
    },
    {
      "epoch": 1.23,
      "grad_norm": 1.2111635208129883,
      "learning_rate": 0.0003845306495740851,
      "loss": 2.0076,
      "step": 94253
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.8655706644058228,
      "learning_rate": 0.0003845267247441382,
      "loss": 2.2289,
      "step": 94254
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.8088374733924866,
      "learning_rate": 0.00038452279989847617,
      "loss": 2.2347,
      "step": 94255
    },
    {
      "epoch": 1.23,
      "grad_norm": 1.3695242404937744,
      "learning_rate": 0.0003845188750370999,
      "loss": 2.2816,
      "step": 94256
    },
    {
      "epoch": 1.23,
      "grad_norm": 1.0285345315933228,
      "learning_rate": 0.00038451495016000983,
      "loss": 2.3131,
      "step": 94257
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.830406129360199,
      "learning_rate": 0.0003845110252672069,
      "loss": 2.2453,
      "step": 94258
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.8405269384384155,
      "learning_rate": 0.0003845071003586918,
      "loss": 2.1716,
      "step": 94259
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.874116063117981,
      "learning_rate": 0.00038450317543446515,
      "loss": 2.094,
      "step": 94260
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.9070589542388916,
      "learning_rate": 0.00038449925049452785,
      "loss": 2.3366,
      "step": 94261
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.8586772084236145,
      "learning_rate": 0.0003844953255388805,
      "loss": 2.1614,
      "step": 94262
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.7842024564743042,
      "learning_rate": 0.00038449140056752385,
      "loss": 2.1524,
      "step": 94263
    },
    {
      "epoch": 1.23,
      "grad_norm": 1.029023289680481,
      "learning_rate": 0.00038448747558045873,
      "loss": 2.1403,
      "step": 94264
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.8563808798789978,
      "learning_rate": 0.0003844835505776858,
      "loss": 2.2078,
      "step": 94265
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.8780157566070557,
      "learning_rate": 0.0003844796255592057,
      "loss": 2.1342,
      "step": 94266
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.8233429789543152,
      "learning_rate": 0.00038447570052501924,
      "loss": 2.0005,
      "step": 94267
    },
    {
      "epoch": 1.23,
      "grad_norm": 1.0241762399673462,
      "learning_rate": 0.00038447177547512725,
      "loss": 1.7532,
      "step": 94268
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.8122938871383667,
      "learning_rate": 0.0003844678504095302,
      "loss": 2.1969,
      "step": 94269
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.865347683429718,
      "learning_rate": 0.0003844639253282291,
      "loss": 2.3888,
      "step": 94270
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.8383030891418457,
      "learning_rate": 0.00038446000023122464,
      "loss": 2.3261,
      "step": 94271
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.8743953108787537,
      "learning_rate": 0.0003844560751185173,
      "loss": 2.1373,
      "step": 94272
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.8396098613739014,
      "learning_rate": 0.0003844521499901081,
      "loss": 2.1616,
      "step": 94273
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.8509519100189209,
      "learning_rate": 0.0003844482248459976,
      "loss": 2.2972,
      "step": 94274
    },
    {
      "epoch": 1.23,
      "grad_norm": 1.0006903409957886,
      "learning_rate": 0.0003844442996861866,
      "loss": 2.3626,
      "step": 94275
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.8498915433883667,
      "learning_rate": 0.0003844403745106757,
      "loss": 2.2593,
      "step": 94276
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.8066983819007874,
      "learning_rate": 0.00038443644931946596,
      "loss": 2.1976,
      "step": 94277
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.8630594611167908,
      "learning_rate": 0.0003844325241125577,
      "loss": 2.4033,
      "step": 94278
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.8981731534004211,
      "learning_rate": 0.0003844285988899519,
      "loss": 2.2478,
      "step": 94279
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.9311367273330688,
      "learning_rate": 0.0003844246736516493,
      "loss": 1.9748,
      "step": 94280
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.8696997761726379,
      "learning_rate": 0.0003844207483976505,
      "loss": 2.2523,
      "step": 94281
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.9650859236717224,
      "learning_rate": 0.0003844168231279562,
      "loss": 2.3644,
      "step": 94282
    },
    {
      "epoch": 1.23,
      "grad_norm": 1.014651894569397,
      "learning_rate": 0.00038441289784256743,
      "loss": 2.2026,
      "step": 94283
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.7876683473587036,
      "learning_rate": 0.0003844089725414845,
      "loss": 2.3393,
      "step": 94284
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.8426215648651123,
      "learning_rate": 0.00038440504722470845,
      "loss": 1.9867,
      "step": 94285
    },
    {
      "epoch": 1.23,
      "grad_norm": 1.0043277740478516,
      "learning_rate": 0.0003844011218922399,
      "loss": 2.3385,
      "step": 94286
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.8437826633453369,
      "learning_rate": 0.0003843971965440796,
      "loss": 2.2586,
      "step": 94287
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.7861312031745911,
      "learning_rate": 0.00038439327118022825,
      "loss": 2.1527,
      "step": 94288
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.8754347562789917,
      "learning_rate": 0.0003843893458006866,
      "loss": 2.3079,
      "step": 94289
    },
    {
      "epoch": 1.23,
      "grad_norm": 1.4814027547836304,
      "learning_rate": 0.0003843854204054554,
      "loss": 2.0779,
      "step": 94290
    },
    {
      "epoch": 1.23,
      "grad_norm": 1.0002055168151855,
      "learning_rate": 0.0003843814949945353,
      "loss": 2.1744,
      "step": 94291
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.8280722498893738,
      "learning_rate": 0.0003843775695679271,
      "loss": 2.1732,
      "step": 94292
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.9437412619590759,
      "learning_rate": 0.0003843736441256316,
      "loss": 2.07,
      "step": 94293
    },
    {
      "epoch": 1.23,
      "grad_norm": 1.0325206518173218,
      "learning_rate": 0.00038436971866764934,
      "loss": 2.365,
      "step": 94294
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.8843398690223694,
      "learning_rate": 0.0003843657931939812,
      "loss": 2.1808,
      "step": 94295
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.827781617641449,
      "learning_rate": 0.0003843618677046278,
      "loss": 2.1273,
      "step": 94296
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.8422756791114807,
      "learning_rate": 0.00038435794219959004,
      "loss": 2.1112,
      "step": 94297
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.9664314985275269,
      "learning_rate": 0.0003843540166788685,
      "loss": 2.0453,
      "step": 94298
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.9554715752601624,
      "learning_rate": 0.00038435009114246396,
      "loss": 2.0596,
      "step": 94299
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.7855443954467773,
      "learning_rate": 0.00038434616559037717,
      "loss": 1.9981,
      "step": 94300
    },
    {
      "epoch": 1.23,
      "grad_norm": 1.5383498668670654,
      "learning_rate": 0.0003843422400226088,
      "loss": 2.3424,
      "step": 94301
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.8461619019508362,
      "learning_rate": 0.0003843383144391596,
      "loss": 2.1756,
      "step": 94302
    },
    {
      "epoch": 1.23,
      "grad_norm": 1.0316894054412842,
      "learning_rate": 0.0003843343888400304,
      "loss": 2.4316,
      "step": 94303
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.82901930809021,
      "learning_rate": 0.0003843304632252218,
      "loss": 2.14,
      "step": 94304
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.8009141087532043,
      "learning_rate": 0.00038432653759473456,
      "loss": 2.0425,
      "step": 94305
    },
    {
      "epoch": 1.23,
      "grad_norm": 1.0642180442810059,
      "learning_rate": 0.00038432261194856944,
      "loss": 2.3602,
      "step": 94306
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.8060882091522217,
      "learning_rate": 0.0003843186862867272,
      "loss": 2.0379,
      "step": 94307
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.8037207126617432,
      "learning_rate": 0.0003843147606092084,
      "loss": 1.9462,
      "step": 94308
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.8963572382926941,
      "learning_rate": 0.000384310834916014,
      "loss": 2.1769,
      "step": 94309
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.875188410282135,
      "learning_rate": 0.0003843069092071446,
      "loss": 2.2779,
      "step": 94310
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.8844254612922668,
      "learning_rate": 0.0003843029834826009,
      "loss": 2.151,
      "step": 94311
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.9180737733840942,
      "learning_rate": 0.00038429905774238374,
      "loss": 2.014,
      "step": 94312
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.9028496146202087,
      "learning_rate": 0.00038429513198649383,
      "loss": 2.2727,
      "step": 94313
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.8066049218177795,
      "learning_rate": 0.0003842912062149317,
      "loss": 2.2139,
      "step": 94314
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.8291598558425903,
      "learning_rate": 0.00038428728042769845,
      "loss": 2.1422,
      "step": 94315
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.8537386059761047,
      "learning_rate": 0.0003842833546247945,
      "loss": 2.0638,
      "step": 94316
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.9377192258834839,
      "learning_rate": 0.0003842794288062207,
      "loss": 2.0327,
      "step": 94317
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.7987467646598816,
      "learning_rate": 0.00038427550297197776,
      "loss": 2.053,
      "step": 94318
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.8442325592041016,
      "learning_rate": 0.00038427157712206633,
      "loss": 2.3494,
      "step": 94319
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.998259961605072,
      "learning_rate": 0.0003842676512564874,
      "loss": 2.2821,
      "step": 94320
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.8509659171104431,
      "learning_rate": 0.00038426372537524146,
      "loss": 2.3501,
      "step": 94321
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.8598535656929016,
      "learning_rate": 0.0003842597994783292,
      "loss": 2.0907,
      "step": 94322
    },
    {
      "epoch": 1.23,
      "grad_norm": 1.0190322399139404,
      "learning_rate": 0.0003842558735657516,
      "loss": 2.1796,
      "step": 94323
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.798336386680603,
      "learning_rate": 0.00038425194763750916,
      "loss": 2.2999,
      "step": 94324
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.81687992811203,
      "learning_rate": 0.00038424802169360265,
      "loss": 2.135,
      "step": 94325
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.8701822757720947,
      "learning_rate": 0.00038424409573403295,
      "loss": 2.1554,
      "step": 94326
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.9618560671806335,
      "learning_rate": 0.00038424016975880065,
      "loss": 2.2272,
      "step": 94327
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.8670221567153931,
      "learning_rate": 0.0003842362437679064,
      "loss": 2.1096,
      "step": 94328
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.971864640712738,
      "learning_rate": 0.00038423231776135117,
      "loss": 2.1523,
      "step": 94329
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.8716244697570801,
      "learning_rate": 0.0003842283917391355,
      "loss": 2.0916,
      "step": 94330
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.8716387748718262,
      "learning_rate": 0.0003842244657012602,
      "loss": 2.1818,
      "step": 94331
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.7732208371162415,
      "learning_rate": 0.00038422053964772603,
      "loss": 2.1384,
      "step": 94332
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.8197274208068848,
      "learning_rate": 0.00038421661357853357,
      "loss": 2.2093,
      "step": 94333
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.953172504901886,
      "learning_rate": 0.00038421268749368377,
      "loss": 2.1589,
      "step": 94334
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.8533800840377808,
      "learning_rate": 0.00038420876139317716,
      "loss": 2.1476,
      "step": 94335
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.8164883852005005,
      "learning_rate": 0.0003842048352770146,
      "loss": 2.2238,
      "step": 94336
    },
    {
      "epoch": 1.23,
      "grad_norm": 1.0106383562088013,
      "learning_rate": 0.00038420090914519675,
      "loss": 2.1003,
      "step": 94337
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.9445918798446655,
      "learning_rate": 0.00038419698299772435,
      "loss": 2.1608,
      "step": 94338
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.7990074157714844,
      "learning_rate": 0.00038419305683459813,
      "loss": 2.104,
      "step": 94339
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.8308649063110352,
      "learning_rate": 0.00038418913065581885,
      "loss": 2.3189,
      "step": 94340
    },
    {
      "epoch": 1.23,
      "grad_norm": 1.071488380432129,
      "learning_rate": 0.0003841852044613872,
      "loss": 2.4074,
      "step": 94341
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.9442607164382935,
      "learning_rate": 0.000384181278251304,
      "loss": 2.0311,
      "step": 94342
    },
    {
      "epoch": 1.23,
      "grad_norm": 1.0242547988891602,
      "learning_rate": 0.0003841773520255698,
      "loss": 1.9932,
      "step": 94343
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.928912878036499,
      "learning_rate": 0.00038417342578418555,
      "loss": 2.2092,
      "step": 94344
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.952978789806366,
      "learning_rate": 0.00038416949952715186,
      "loss": 2.1903,
      "step": 94345
    },
    {
      "epoch": 1.23,
      "grad_norm": 1.0645681619644165,
      "learning_rate": 0.0003841655732544694,
      "loss": 2.0331,
      "step": 94346
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.8443456292152405,
      "learning_rate": 0.000384161646966139,
      "loss": 2.1173,
      "step": 94347
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.8175550103187561,
      "learning_rate": 0.0003841577206621614,
      "loss": 2.3777,
      "step": 94348
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.8403319716453552,
      "learning_rate": 0.00038415379434253726,
      "loss": 2.0685,
      "step": 94349
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.8150030374526978,
      "learning_rate": 0.0003841498680072673,
      "loss": 2.0946,
      "step": 94350
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.9940323829650879,
      "learning_rate": 0.00038414594165635236,
      "loss": 2.2643,
      "step": 94351
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.7720608115196228,
      "learning_rate": 0.00038414201528979305,
      "loss": 2.1898,
      "step": 94352
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.8485679030418396,
      "learning_rate": 0.00038413808890759023,
      "loss": 2.4253,
      "step": 94353
    },
    {
      "epoch": 1.23,
      "grad_norm": 1.0399407148361206,
      "learning_rate": 0.00038413416250974444,
      "loss": 2.2185,
      "step": 94354
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.9368253350257874,
      "learning_rate": 0.00038413023609625664,
      "loss": 2.0296,
      "step": 94355
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.8446964621543884,
      "learning_rate": 0.00038412630966712743,
      "loss": 2.0517,
      "step": 94356
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.8356271386146545,
      "learning_rate": 0.0003841223832223574,
      "loss": 2.2721,
      "step": 94357
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.9153135418891907,
      "learning_rate": 0.00038411845676194757,
      "loss": 2.0538,
      "step": 94358
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.8000203967094421,
      "learning_rate": 0.0003841145302858985,
      "loss": 2.3684,
      "step": 94359
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.8046140670776367,
      "learning_rate": 0.0003841106037942109,
      "loss": 2.1872,
      "step": 94360
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.9055169820785522,
      "learning_rate": 0.0003841066772868856,
      "loss": 2.0647,
      "step": 94361
    },
    {
      "epoch": 1.23,
      "grad_norm": 1.0667121410369873,
      "learning_rate": 0.00038410275076392335,
      "loss": 2.2778,
      "step": 94362
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.8134752511978149,
      "learning_rate": 0.0003840988242253247,
      "loss": 2.2534,
      "step": 94363
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.8847882151603699,
      "learning_rate": 0.00038409489767109054,
      "loss": 2.3034,
      "step": 94364
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.9143786430358887,
      "learning_rate": 0.00038409097110122153,
      "loss": 2.2154,
      "step": 94365
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.9890561103820801,
      "learning_rate": 0.0003840870445157184,
      "loss": 2.5377,
      "step": 94366
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.9993104934692383,
      "learning_rate": 0.000384083117914582,
      "loss": 2.2604,
      "step": 94367
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.8129554390907288,
      "learning_rate": 0.0003840791912978129,
      "loss": 2.0889,
      "step": 94368
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.8886605501174927,
      "learning_rate": 0.00038407526466541187,
      "loss": 2.0366,
      "step": 94369
    },
    {
      "epoch": 1.23,
      "grad_norm": 1.3025429248809814,
      "learning_rate": 0.0003840713380173796,
      "loss": 2.0761,
      "step": 94370
    },
    {
      "epoch": 1.23,
      "grad_norm": 1.0352767705917358,
      "learning_rate": 0.000384067411353717,
      "loss": 2.1415,
      "step": 94371
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.8954464197158813,
      "learning_rate": 0.00038406348467442464,
      "loss": 2.4827,
      "step": 94372
    },
    {
      "epoch": 1.23,
      "grad_norm": 1.1360163688659668,
      "learning_rate": 0.0003840595579795033,
      "loss": 2.2585,
      "step": 94373
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.9388149380683899,
      "learning_rate": 0.00038405563126895373,
      "loss": 2.3084,
      "step": 94374
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.8154794573783875,
      "learning_rate": 0.0003840517045427765,
      "loss": 2.1701,
      "step": 94375
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.9485568404197693,
      "learning_rate": 0.0003840477778009726,
      "loss": 2.0501,
      "step": 94376
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.9998763799667358,
      "learning_rate": 0.0003840438510435426,
      "loss": 2.0497,
      "step": 94377
    },
    {
      "epoch": 1.23,
      "grad_norm": 1.0299925804138184,
      "learning_rate": 0.00038403992427048727,
      "loss": 2.09,
      "step": 94378
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.941826343536377,
      "learning_rate": 0.0003840359974818073,
      "loss": 1.957,
      "step": 94379
    },
    {
      "epoch": 1.23,
      "grad_norm": 1.0030274391174316,
      "learning_rate": 0.0003840320706775034,
      "loss": 1.9975,
      "step": 94380
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.8926615118980408,
      "learning_rate": 0.0003840281438575765,
      "loss": 2.3254,
      "step": 94381
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.8036274313926697,
      "learning_rate": 0.0003840242170220271,
      "loss": 2.1315,
      "step": 94382
    },
    {
      "epoch": 1.23,
      "grad_norm": 1.0682018995285034,
      "learning_rate": 0.000384020290170856,
      "loss": 2.3118,
      "step": 94383
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.8325913548469543,
      "learning_rate": 0.0003840163633040639,
      "loss": 2.161,
      "step": 94384
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.7868139743804932,
      "learning_rate": 0.0003840124364216517,
      "loss": 2.2182,
      "step": 94385
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.8871325254440308,
      "learning_rate": 0.00038400850952361994,
      "loss": 2.0425,
      "step": 94386
    },
    {
      "epoch": 1.23,
      "grad_norm": 1.0426812171936035,
      "learning_rate": 0.00038400458260996935,
      "loss": 2.0294,
      "step": 94387
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.8733598589897156,
      "learning_rate": 0.0003840006556807007,
      "loss": 2.1594,
      "step": 94388
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.8960857391357422,
      "learning_rate": 0.0003839967287358149,
      "loss": 2.2561,
      "step": 94389
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.9075037837028503,
      "learning_rate": 0.0003839928017753124,
      "loss": 2.4533,
      "step": 94390
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.8339905142784119,
      "learning_rate": 0.0003839888747991941,
      "loss": 2.1649,
      "step": 94391
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.8569658398628235,
      "learning_rate": 0.00038398494780746075,
      "loss": 2.2077,
      "step": 94392
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.982597291469574,
      "learning_rate": 0.00038398102080011286,
      "loss": 2.1804,
      "step": 94393
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.8583249449729919,
      "learning_rate": 0.0003839770937771514,
      "loss": 2.2965,
      "step": 94394
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.8538334369659424,
      "learning_rate": 0.00038397316673857696,
      "loss": 1.9708,
      "step": 94395
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.8148295283317566,
      "learning_rate": 0.00038396923968439045,
      "loss": 2.3331,
      "step": 94396
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.9190484285354614,
      "learning_rate": 0.00038396531261459236,
      "loss": 2.2316,
      "step": 94397
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.8120017647743225,
      "learning_rate": 0.00038396138552918355,
      "loss": 2.1328,
      "step": 94398
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.827355682849884,
      "learning_rate": 0.00038395745842816477,
      "loss": 1.9897,
      "step": 94399
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.8198946118354797,
      "learning_rate": 0.0003839535313115367,
      "loss": 2.188,
      "step": 94400
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.7899056077003479,
      "learning_rate": 0.00038394960417930003,
      "loss": 2.2107,
      "step": 94401
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.9105680584907532,
      "learning_rate": 0.00038394567703145565,
      "loss": 2.1848,
      "step": 94402
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.912408173084259,
      "learning_rate": 0.00038394174986800406,
      "loss": 2.1285,
      "step": 94403
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.9305923581123352,
      "learning_rate": 0.0003839378226889462,
      "loss": 2.2802,
      "step": 94404
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.7692078948020935,
      "learning_rate": 0.0003839338954942827,
      "loss": 2.0861,
      "step": 94405
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.8078293204307556,
      "learning_rate": 0.00038392996828401425,
      "loss": 2.0016,
      "step": 94406
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.8688041567802429,
      "learning_rate": 0.00038392604105814173,
      "loss": 2.06,
      "step": 94407
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.7385097146034241,
      "learning_rate": 0.00038392211381666576,
      "loss": 1.9031,
      "step": 94408
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.8422111868858337,
      "learning_rate": 0.000383918186559587,
      "loss": 2.1574,
      "step": 94409
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.9143832325935364,
      "learning_rate": 0.00038391425928690636,
      "loss": 2.3452,
      "step": 94410
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.8218292593955994,
      "learning_rate": 0.00038391033199862445,
      "loss": 2.126,
      "step": 94411
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.7868998050689697,
      "learning_rate": 0.00038390640469474196,
      "loss": 1.87,
      "step": 94412
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.8917967677116394,
      "learning_rate": 0.0003839024773752597,
      "loss": 2.1383,
      "step": 94413
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.8391685485839844,
      "learning_rate": 0.00038389855004017853,
      "loss": 2.0819,
      "step": 94414
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.8236411213874817,
      "learning_rate": 0.0003838946226894989,
      "loss": 2.3059,
      "step": 94415
    },
    {
      "epoch": 1.23,
      "grad_norm": 1.03081214427948,
      "learning_rate": 0.00038389069532322165,
      "loss": 2.065,
      "step": 94416
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.8402748107910156,
      "learning_rate": 0.00038388676794134767,
      "loss": 2.2909,
      "step": 94417
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.8522099256515503,
      "learning_rate": 0.00038388284054387744,
      "loss": 2.2648,
      "step": 94418
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.991383969783783,
      "learning_rate": 0.0003838789131308118,
      "loss": 2.159,
      "step": 94419
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.7578330636024475,
      "learning_rate": 0.0003838749857021516,
      "loss": 2.1005,
      "step": 94420
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.8508972525596619,
      "learning_rate": 0.0003838710582578973,
      "loss": 2.4013,
      "step": 94421
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.791589617729187,
      "learning_rate": 0.0003838671307980499,
      "loss": 2.0978,
      "step": 94422
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.8540692925453186,
      "learning_rate": 0.00038386320332261,
      "loss": 2.2078,
      "step": 94423
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.9118332862854004,
      "learning_rate": 0.0003838592758315783,
      "loss": 2.2917,
      "step": 94424
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.8238278031349182,
      "learning_rate": 0.00038385534832495563,
      "loss": 2.303,
      "step": 94425
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.8885286450386047,
      "learning_rate": 0.00038385142080274263,
      "loss": 2.1944,
      "step": 94426
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.8054096698760986,
      "learning_rate": 0.0003838474932649401,
      "loss": 2.1672,
      "step": 94427
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.814685583114624,
      "learning_rate": 0.00038384356571154876,
      "loss": 2.1782,
      "step": 94428
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.851951003074646,
      "learning_rate": 0.0003838396381425693,
      "loss": 2.3412,
      "step": 94429
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.8580515384674072,
      "learning_rate": 0.0003838357105580025,
      "loss": 2.1119,
      "step": 94430
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.8602046370506287,
      "learning_rate": 0.00038383178295784895,
      "loss": 2.2348,
      "step": 94431
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.8825775384902954,
      "learning_rate": 0.0003838278553421096,
      "loss": 2.1818,
      "step": 94432
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.8643022775650024,
      "learning_rate": 0.0003838239277107849,
      "loss": 2.233,
      "step": 94433
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.888699471950531,
      "learning_rate": 0.0003838200000638759,
      "loss": 2.1698,
      "step": 94434
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.8863571882247925,
      "learning_rate": 0.0003838160724013832,
      "loss": 2.3344,
      "step": 94435
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.7715273499488831,
      "learning_rate": 0.0003838121447233075,
      "loss": 2.0167,
      "step": 94436
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.840632975101471,
      "learning_rate": 0.00038380821702964943,
      "loss": 2.0651,
      "step": 94437
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.9136049747467041,
      "learning_rate": 0.00038380428932040993,
      "loss": 2.2138,
      "step": 94438
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.9012436866760254,
      "learning_rate": 0.0003838003615955896,
      "loss": 2.421,
      "step": 94439
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.852156400680542,
      "learning_rate": 0.00038379643385518913,
      "loss": 2.088,
      "step": 94440
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.891939640045166,
      "learning_rate": 0.0003837925060992095,
      "loss": 2.2729,
      "step": 94441
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.83180832862854,
      "learning_rate": 0.0003837885783276511,
      "loss": 2.0431,
      "step": 94442
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.8310071229934692,
      "learning_rate": 0.0003837846505405149,
      "loss": 1.9915,
      "step": 94443
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.9513556957244873,
      "learning_rate": 0.00038378072273780153,
      "loss": 2.1183,
      "step": 94444
    },
    {
      "epoch": 1.23,
      "grad_norm": 1.0178701877593994,
      "learning_rate": 0.00038377679491951173,
      "loss": 2.3759,
      "step": 94445
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.8268485069274902,
      "learning_rate": 0.00038377286708564623,
      "loss": 2.1854,
      "step": 94446
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.9345678091049194,
      "learning_rate": 0.0003837689392362059,
      "loss": 2.2755,
      "step": 94447
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.8217344284057617,
      "learning_rate": 0.00038376501137119115,
      "loss": 2.2917,
      "step": 94448
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.8855531215667725,
      "learning_rate": 0.000383761083490603,
      "loss": 2.207,
      "step": 94449
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.8708673119544983,
      "learning_rate": 0.00038375715559444204,
      "loss": 2.2671,
      "step": 94450
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.8512431979179382,
      "learning_rate": 0.0003837532276827091,
      "loss": 2.1201,
      "step": 94451
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.844961941242218,
      "learning_rate": 0.0003837492997554048,
      "loss": 2.1282,
      "step": 94452
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.8363212943077087,
      "learning_rate": 0.00038374537181253,
      "loss": 2.3054,
      "step": 94453
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.9242340922355652,
      "learning_rate": 0.0003837414438540853,
      "loss": 2.1111,
      "step": 94454
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.8830618858337402,
      "learning_rate": 0.0003837375158800715,
      "loss": 2.1135,
      "step": 94455
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.8951067924499512,
      "learning_rate": 0.0003837335878904893,
      "loss": 2.2134,
      "step": 94456
    },
    {
      "epoch": 1.23,
      "grad_norm": 1.0228372812271118,
      "learning_rate": 0.0003837296598853395,
      "loss": 2.0378,
      "step": 94457
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.8045778274536133,
      "learning_rate": 0.00038372573186462274,
      "loss": 2.2676,
      "step": 94458
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.9405062198638916,
      "learning_rate": 0.00038372180382833973,
      "loss": 2.2035,
      "step": 94459
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.941742479801178,
      "learning_rate": 0.00038371787577649137,
      "loss": 2.3719,
      "step": 94460
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.9134712219238281,
      "learning_rate": 0.0003837139477090782,
      "loss": 2.3186,
      "step": 94461
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.8855618238449097,
      "learning_rate": 0.00038371001962610103,
      "loss": 2.3031,
      "step": 94462
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.9263092875480652,
      "learning_rate": 0.0003837060915275606,
      "loss": 1.9053,
      "step": 94463
    },
    {
      "epoch": 1.23,
      "grad_norm": 1.1356160640716553,
      "learning_rate": 0.0003837021634134576,
      "loss": 2.1606,
      "step": 94464
    },
    {
      "epoch": 1.23,
      "grad_norm": 1.0401408672332764,
      "learning_rate": 0.0003836982352837928,
      "loss": 2.2484,
      "step": 94465
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.9337605237960815,
      "learning_rate": 0.00038369430713856693,
      "loss": 2.276,
      "step": 94466
    },
    {
      "epoch": 1.23,
      "grad_norm": 1.4455819129943848,
      "learning_rate": 0.00038369037897778075,
      "loss": 2.2168,
      "step": 94467
    },
    {
      "epoch": 1.23,
      "grad_norm": 1.0832579135894775,
      "learning_rate": 0.0003836864508014348,
      "loss": 1.991,
      "step": 94468
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.9496357440948486,
      "learning_rate": 0.00038368252260953013,
      "loss": 2.0664,
      "step": 94469
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.9083839654922485,
      "learning_rate": 0.0003836785944020672,
      "loss": 2.002,
      "step": 94470
    },
    {
      "epoch": 1.23,
      "grad_norm": 1.019818663597107,
      "learning_rate": 0.0003836746661790469,
      "loss": 2.0618,
      "step": 94471
    },
    {
      "epoch": 1.23,
      "grad_norm": 1.0526280403137207,
      "learning_rate": 0.00038367073794046987,
      "loss": 2.4439,
      "step": 94472
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.869707465171814,
      "learning_rate": 0.0003836668096863369,
      "loss": 2.2172,
      "step": 94473
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.8885738849639893,
      "learning_rate": 0.0003836628814166487,
      "loss": 2.1053,
      "step": 94474
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.846462070941925,
      "learning_rate": 0.0003836589531314059,
      "loss": 2.0653,
      "step": 94475
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.8670960068702698,
      "learning_rate": 0.00038365502483060936,
      "loss": 2.2427,
      "step": 94476
    },
    {
      "epoch": 1.23,
      "grad_norm": 1.0960252285003662,
      "learning_rate": 0.0003836510965142598,
      "loss": 2.5138,
      "step": 94477
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.850743293762207,
      "learning_rate": 0.0003836471681823579,
      "loss": 1.9536,
      "step": 94478
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.8700737357139587,
      "learning_rate": 0.00038364323983490446,
      "loss": 1.8447,
      "step": 94479
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.8317525386810303,
      "learning_rate": 0.0003836393114719001,
      "loss": 2.0868,
      "step": 94480
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.8295864462852478,
      "learning_rate": 0.00038363538309334563,
      "loss": 2.0076,
      "step": 94481
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.7552893757820129,
      "learning_rate": 0.0003836314546992418,
      "loss": 2.0951,
      "step": 94482
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.837637186050415,
      "learning_rate": 0.0003836275262895892,
      "loss": 2.1613,
      "step": 94483
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.8599657416343689,
      "learning_rate": 0.00038362359786438884,
      "loss": 2.1439,
      "step": 94484
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.8759039044380188,
      "learning_rate": 0.0003836196694236411,
      "loss": 2.0261,
      "step": 94485
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.8825744986534119,
      "learning_rate": 0.000383615740967347,
      "loss": 2.437,
      "step": 94486
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.8820479512214661,
      "learning_rate": 0.0003836118124955072,
      "loss": 2.312,
      "step": 94487
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.8469564318656921,
      "learning_rate": 0.00038360788400812224,
      "loss": 2.2947,
      "step": 94488
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.8808885216712952,
      "learning_rate": 0.0003836039555051931,
      "loss": 2.0774,
      "step": 94489
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.8006967306137085,
      "learning_rate": 0.00038360002698672037,
      "loss": 2.1567,
      "step": 94490
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.9222157001495361,
      "learning_rate": 0.0003835960984527048,
      "loss": 2.3796,
      "step": 94491
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.9328845143318176,
      "learning_rate": 0.00038359216990314714,
      "loss": 2.2308,
      "step": 94492
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.8333610892295837,
      "learning_rate": 0.00038358824133804815,
      "loss": 2.0161,
      "step": 94493
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.8618033528327942,
      "learning_rate": 0.0003835843127574085,
      "loss": 2.3402,
      "step": 94494
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.8620893955230713,
      "learning_rate": 0.00038358038416122893,
      "loss": 2.1254,
      "step": 94495
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.7723391652107239,
      "learning_rate": 0.0003835764555495102,
      "loss": 2.2024,
      "step": 94496
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.7606170177459717,
      "learning_rate": 0.00038357252692225306,
      "loss": 2.338,
      "step": 94497
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.8813797235488892,
      "learning_rate": 0.0003835685982794582,
      "loss": 1.9127,
      "step": 94498
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.948153018951416,
      "learning_rate": 0.00038356466962112637,
      "loss": 2.2901,
      "step": 94499
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.9662436842918396,
      "learning_rate": 0.0003835607409472582,
      "loss": 2.3669,
      "step": 94500
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.8466799855232239,
      "learning_rate": 0.00038355681225785463,
      "loss": 2.0924,
      "step": 94501
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.8753079771995544,
      "learning_rate": 0.00038355288355291627,
      "loss": 2.2516,
      "step": 94502
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.7577493190765381,
      "learning_rate": 0.0003835489548324438,
      "loss": 2.1085,
      "step": 94503
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.8746489882469177,
      "learning_rate": 0.00038354502609643795,
      "loss": 2.0343,
      "step": 94504
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.9264629483222961,
      "learning_rate": 0.00038354109734489966,
      "loss": 2.038,
      "step": 94505
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.8420142531394958,
      "learning_rate": 0.00038353716857782936,
      "loss": 2.0535,
      "step": 94506
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.8152690529823303,
      "learning_rate": 0.0003835332397952279,
      "loss": 2.3214,
      "step": 94507
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.8439496159553528,
      "learning_rate": 0.00038352931099709615,
      "loss": 2.3414,
      "step": 94508
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.80705726146698,
      "learning_rate": 0.0003835253821834347,
      "loss": 2.396,
      "step": 94509
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.9942978620529175,
      "learning_rate": 0.00038352145335424423,
      "loss": 2.0831,
      "step": 94510
    },
    {
      "epoch": 1.23,
      "grad_norm": 1.0998772382736206,
      "learning_rate": 0.0003835175245095256,
      "loss": 2.2337,
      "step": 94511
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.9779279232025146,
      "learning_rate": 0.00038351359564927956,
      "loss": 2.1696,
      "step": 94512
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.9047145247459412,
      "learning_rate": 0.0003835096667735066,
      "loss": 2.2823,
      "step": 94513
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.8798591494560242,
      "learning_rate": 0.0003835057378822078,
      "loss": 2.4241,
      "step": 94514
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.9618213772773743,
      "learning_rate": 0.0003835018089753836,
      "loss": 2.3808,
      "step": 94515
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.8507815003395081,
      "learning_rate": 0.0003834978800530348,
      "loss": 2.1212,
      "step": 94516
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.9779883623123169,
      "learning_rate": 0.0003834939511151622,
      "loss": 2.0424,
      "step": 94517
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.8956652879714966,
      "learning_rate": 0.00038349002216176657,
      "loss": 2.3078,
      "step": 94518
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.9021045565605164,
      "learning_rate": 0.0003834860931928485,
      "loss": 2.132,
      "step": 94519
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.8209378719329834,
      "learning_rate": 0.0003834821642084088,
      "loss": 2.0657,
      "step": 94520
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.7905271053314209,
      "learning_rate": 0.0003834782352084482,
      "loss": 2.1427,
      "step": 94521
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.8379026651382446,
      "learning_rate": 0.0003834743061929674,
      "loss": 2.2003,
      "step": 94522
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.8956561088562012,
      "learning_rate": 0.00038347037716196714,
      "loss": 2.1541,
      "step": 94523
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.861482560634613,
      "learning_rate": 0.00038346644811544824,
      "loss": 2.2951,
      "step": 94524
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.8654389381408691,
      "learning_rate": 0.0003834625190534113,
      "loss": 2.2057,
      "step": 94525
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.893267810344696,
      "learning_rate": 0.00038345858997585705,
      "loss": 2.3413,
      "step": 94526
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.8304938673973083,
      "learning_rate": 0.00038345466088278635,
      "loss": 2.1634,
      "step": 94527
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.9204034209251404,
      "learning_rate": 0.0003834507317741998,
      "loss": 2.0912,
      "step": 94528
    },
    {
      "epoch": 1.23,
      "grad_norm": 1.0385463237762451,
      "learning_rate": 0.0003834468026500982,
      "loss": 2.0885,
      "step": 94529
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.815169632434845,
      "learning_rate": 0.0003834428735104823,
      "loss": 2.3508,
      "step": 94530
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.8814437389373779,
      "learning_rate": 0.0003834389443553528,
      "loss": 2.0735,
      "step": 94531
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.8419860601425171,
      "learning_rate": 0.00038343501518471037,
      "loss": 1.985,
      "step": 94532
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.8608744144439697,
      "learning_rate": 0.0003834310859985559,
      "loss": 2.1048,
      "step": 94533
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.9020824432373047,
      "learning_rate": 0.0003834271567968899,
      "loss": 2.2233,
      "step": 94534
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.8333731889724731,
      "learning_rate": 0.0003834232275797132,
      "loss": 2.2252,
      "step": 94535
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.8559769988059998,
      "learning_rate": 0.00038341929834702663,
      "loss": 2.2385,
      "step": 94536
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.7906403541564941,
      "learning_rate": 0.0003834153690988308,
      "loss": 2.1113,
      "step": 94537
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.7652894258499146,
      "learning_rate": 0.0003834114398351265,
      "loss": 2.3744,
      "step": 94538
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.9185097217559814,
      "learning_rate": 0.00038340751055591444,
      "loss": 2.1249,
      "step": 94539
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.8326297998428345,
      "learning_rate": 0.0003834035812611953,
      "loss": 2.1514,
      "step": 94540
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.828222393989563,
      "learning_rate": 0.00038339965195096986,
      "loss": 2.1283,
      "step": 94541
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.7705504894256592,
      "learning_rate": 0.0003833957226252389,
      "loss": 2.3285,
      "step": 94542
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.8403809070587158,
      "learning_rate": 0.0003833917932840031,
      "loss": 2.1251,
      "step": 94543
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.8558493256568909,
      "learning_rate": 0.00038338786392726323,
      "loss": 2.2801,
      "step": 94544
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.9672616124153137,
      "learning_rate": 0.0003833839345550199,
      "loss": 1.8998,
      "step": 94545
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.8909211754798889,
      "learning_rate": 0.0003833800051672739,
      "loss": 2.1192,
      "step": 94546
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.8995699286460876,
      "learning_rate": 0.00038337607576402607,
      "loss": 2.2111,
      "step": 94547
    },
    {
      "epoch": 1.23,
      "grad_norm": 1.0571072101593018,
      "learning_rate": 0.00038337214634527703,
      "loss": 2.0752,
      "step": 94548
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.8994976282119751,
      "learning_rate": 0.0003833682169110276,
      "loss": 2.0759,
      "step": 94549
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.8633432984352112,
      "learning_rate": 0.0003833642874612783,
      "loss": 2.268,
      "step": 94550
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.9575048089027405,
      "learning_rate": 0.00038336035799603013,
      "loss": 2.1097,
      "step": 94551
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.8931920528411865,
      "learning_rate": 0.0003833564285152836,
      "loss": 2.4712,
      "step": 94552
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.7980123162269592,
      "learning_rate": 0.0003833524990190396,
      "loss": 2.1982,
      "step": 94553
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.824833869934082,
      "learning_rate": 0.00038334856950729873,
      "loss": 2.128,
      "step": 94554
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.8416064381599426,
      "learning_rate": 0.0003833446399800618,
      "loss": 2.6114,
      "step": 94555
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.8793853521347046,
      "learning_rate": 0.0003833407104373296,
      "loss": 2.3207,
      "step": 94556
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.8082777857780457,
      "learning_rate": 0.00038333678087910266,
      "loss": 2.1392,
      "step": 94557
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.8028059005737305,
      "learning_rate": 0.000383332851305382,
      "loss": 2.3488,
      "step": 94558
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.8451026082038879,
      "learning_rate": 0.00038332892171616807,
      "loss": 2.1047,
      "step": 94559
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.9343202114105225,
      "learning_rate": 0.00038332499211146177,
      "loss": 2.1646,
      "step": 94560
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.7937892079353333,
      "learning_rate": 0.00038332106249126374,
      "loss": 2.1533,
      "step": 94561
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.8457414507865906,
      "learning_rate": 0.0003833171328555749,
      "loss": 2.2628,
      "step": 94562
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.9674023985862732,
      "learning_rate": 0.0003833132032043957,
      "loss": 2.1252,
      "step": 94563
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.8633180260658264,
      "learning_rate": 0.000383309273537727,
      "loss": 2.0325,
      "step": 94564
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.8001572489738464,
      "learning_rate": 0.0003833053438555696,
      "loss": 2.1765,
      "step": 94565
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.7842010259628296,
      "learning_rate": 0.00038330141415792406,
      "loss": 2.078,
      "step": 94566
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.9752591848373413,
      "learning_rate": 0.00038329748444479125,
      "loss": 2.1235,
      "step": 94567
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.8042636513710022,
      "learning_rate": 0.00038329355471617186,
      "loss": 2.1829,
      "step": 94568
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.8170108199119568,
      "learning_rate": 0.00038328962497206676,
      "loss": 2.201,
      "step": 94569
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.8766227960586548,
      "learning_rate": 0.00038328569521247643,
      "loss": 2.3133,
      "step": 94570
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.8745291829109192,
      "learning_rate": 0.00038328176543740163,
      "loss": 1.9971,
      "step": 94571
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.873507559299469,
      "learning_rate": 0.00038327783564684334,
      "loss": 2.1234,
      "step": 94572
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.8342449069023132,
      "learning_rate": 0.00038327390584080205,
      "loss": 2.1807,
      "step": 94573
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.8336698412895203,
      "learning_rate": 0.0003832699760192786,
      "loss": 2.3615,
      "step": 94574
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.8071514368057251,
      "learning_rate": 0.0003832660461822737,
      "loss": 2.2444,
      "step": 94575
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.7998070120811462,
      "learning_rate": 0.0003832621163297881,
      "loss": 2.0872,
      "step": 94576
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.7345758676528931,
      "learning_rate": 0.00038325818646182236,
      "loss": 1.9221,
      "step": 94577
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.833091676235199,
      "learning_rate": 0.0003832542565783775,
      "loss": 2.0035,
      "step": 94578
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.8892419338226318,
      "learning_rate": 0.000383250326679454,
      "loss": 2.2204,
      "step": 94579
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.9070884585380554,
      "learning_rate": 0.00038324639676505273,
      "loss": 2.0725,
      "step": 94580
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.9358533024787903,
      "learning_rate": 0.0003832424668351744,
      "loss": 2.207,
      "step": 94581
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.7939937710762024,
      "learning_rate": 0.0003832385368898197,
      "loss": 2.4602,
      "step": 94582
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.893200695514679,
      "learning_rate": 0.0003832346069289894,
      "loss": 2.1505,
      "step": 94583
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.8037870526313782,
      "learning_rate": 0.00038323067695268415,
      "loss": 2.1971,
      "step": 94584
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.8488667011260986,
      "learning_rate": 0.0003832267469609049,
      "loss": 1.8344,
      "step": 94585
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.8971806764602661,
      "learning_rate": 0.0003832228169536521,
      "loss": 2.5007,
      "step": 94586
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.9083906412124634,
      "learning_rate": 0.0003832188869309266,
      "loss": 2.2756,
      "step": 94587
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.8772572875022888,
      "learning_rate": 0.00038321495689272927,
      "loss": 2.1065,
      "step": 94588
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.8248671293258667,
      "learning_rate": 0.0003832110268390606,
      "loss": 2.108,
      "step": 94589
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.8481725454330444,
      "learning_rate": 0.0003832070967699214,
      "loss": 2.1456,
      "step": 94590
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.8456966876983643,
      "learning_rate": 0.0003832031666853125,
      "loss": 2.1818,
      "step": 94591
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.9185490608215332,
      "learning_rate": 0.0003831992365852345,
      "loss": 2.0182,
      "step": 94592
    },
    {
      "epoch": 1.23,
      "grad_norm": 1.041687250137329,
      "learning_rate": 0.00038319530646968824,
      "loss": 2.1889,
      "step": 94593
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.794589102268219,
      "learning_rate": 0.00038319137633867443,
      "loss": 2.0847,
      "step": 94594
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.8945133090019226,
      "learning_rate": 0.00038318744619219376,
      "loss": 2.1431,
      "step": 94595
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.877013623714447,
      "learning_rate": 0.0003831835160302469,
      "loss": 2.2571,
      "step": 94596
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.9356976747512817,
      "learning_rate": 0.00038317958585283476,
      "loss": 2.2104,
      "step": 94597
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.7937724590301514,
      "learning_rate": 0.0003831756556599579,
      "loss": 2.2807,
      "step": 94598
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.8500490188598633,
      "learning_rate": 0.0003831717254516171,
      "loss": 2.0931,
      "step": 94599
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.958977222442627,
      "learning_rate": 0.00038316779522781314,
      "loss": 2.0475,
      "step": 94600
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.8417274951934814,
      "learning_rate": 0.00038316386498854667,
      "loss": 1.9279,
      "step": 94601
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.8107316493988037,
      "learning_rate": 0.0003831599347338185,
      "loss": 2.1148,
      "step": 94602
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.8403077125549316,
      "learning_rate": 0.0003831560044636294,
      "loss": 2.2641,
      "step": 94603
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.9446978569030762,
      "learning_rate": 0.0003831520741779799,
      "loss": 2.4437,
      "step": 94604
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.8582890033721924,
      "learning_rate": 0.00038314814387687096,
      "loss": 2.0895,
      "step": 94605
    },
    {
      "epoch": 1.23,
      "grad_norm": 1.0231214761734009,
      "learning_rate": 0.0003831442135603032,
      "loss": 2.2419,
      "step": 94606
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.9497390985488892,
      "learning_rate": 0.0003831402832282773,
      "loss": 2.1391,
      "step": 94607
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.8261036276817322,
      "learning_rate": 0.00038313635288079404,
      "loss": 2.1392,
      "step": 94608
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.8977041244506836,
      "learning_rate": 0.00038313242251785426,
      "loss": 2.2325,
      "step": 94609
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.9976968169212341,
      "learning_rate": 0.0003831284921394586,
      "loss": 2.0905,
      "step": 94610
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.8403719067573547,
      "learning_rate": 0.00038312456174560764,
      "loss": 2.1104,
      "step": 94611
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.8947774171829224,
      "learning_rate": 0.0003831206313363024,
      "loss": 2.156,
      "step": 94612
    },
    {
      "epoch": 1.23,
      "grad_norm": 1.0482442378997803,
      "learning_rate": 0.0003831167009115434,
      "loss": 2.2434,
      "step": 94613
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.8556633591651917,
      "learning_rate": 0.00038311277047133133,
      "loss": 2.1713,
      "step": 94614
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.8274900317192078,
      "learning_rate": 0.0003831088400156672,
      "loss": 1.9954,
      "step": 94615
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.941943347454071,
      "learning_rate": 0.0003831049095445515,
      "loss": 2.1255,
      "step": 94616
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.8025490045547485,
      "learning_rate": 0.000383100979057985,
      "loss": 2.419,
      "step": 94617
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.8332234025001526,
      "learning_rate": 0.00038309704855596854,
      "loss": 2.3735,
      "step": 94618
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.7616430521011353,
      "learning_rate": 0.00038309311803850273,
      "loss": 1.9737,
      "step": 94619
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.8227289915084839,
      "learning_rate": 0.00038308918750558834,
      "loss": 2.2475,
      "step": 94620
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.8441923260688782,
      "learning_rate": 0.00038308525695722614,
      "loss": 1.9895,
      "step": 94621
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.8025772571563721,
      "learning_rate": 0.00038308132639341676,
      "loss": 2.2585,
      "step": 94622
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.8470727205276489,
      "learning_rate": 0.00038307739581416097,
      "loss": 2.2842,
      "step": 94623
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.8797887563705444,
      "learning_rate": 0.00038307346521945963,
      "loss": 2.2513,
      "step": 94624
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.7813882231712341,
      "learning_rate": 0.0003830695346093133,
      "loss": 2.3136,
      "step": 94625
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.796953558921814,
      "learning_rate": 0.00038306560398372277,
      "loss": 2.1077,
      "step": 94626
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.7454445958137512,
      "learning_rate": 0.0003830616733426888,
      "loss": 1.9613,
      "step": 94627
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.97527015209198,
      "learning_rate": 0.0003830577426862121,
      "loss": 2.1964,
      "step": 94628
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.9256069660186768,
      "learning_rate": 0.0003830538120142933,
      "loss": 2.1152,
      "step": 94629
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.9358211159706116,
      "learning_rate": 0.00038304988132693336,
      "loss": 2.1645,
      "step": 94630
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.9028724431991577,
      "learning_rate": 0.0003830459506241328,
      "loss": 2.2673,
      "step": 94631
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.8152300119400024,
      "learning_rate": 0.0003830420199058924,
      "loss": 2.3103,
      "step": 94632
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.8765671849250793,
      "learning_rate": 0.00038303808917221306,
      "loss": 1.9679,
      "step": 94633
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.8537095189094543,
      "learning_rate": 0.00038303415842309526,
      "loss": 2.2314,
      "step": 94634
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.9016675353050232,
      "learning_rate": 0.00038303022765853983,
      "loss": 2.3917,
      "step": 94635
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.8305500149726868,
      "learning_rate": 0.0003830262968785477,
      "loss": 2.1766,
      "step": 94636
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.8715608716011047,
      "learning_rate": 0.0003830223660831192,
      "loss": 2.0149,
      "step": 94637
    },
    {
      "epoch": 1.23,
      "grad_norm": 1.0280681848526,
      "learning_rate": 0.0003830184352722553,
      "loss": 2.3176,
      "step": 94638
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.8822221159934998,
      "learning_rate": 0.0003830145044459568,
      "loss": 2.1499,
      "step": 94639
    },
    {
      "epoch": 1.23,
      "grad_norm": 1.0324897766113281,
      "learning_rate": 0.00038301057360422427,
      "loss": 2.1971,
      "step": 94640
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.8695284128189087,
      "learning_rate": 0.0003830066427470585,
      "loss": 2.2246,
      "step": 94641
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.9224611520767212,
      "learning_rate": 0.0003830027118744602,
      "loss": 2.1452,
      "step": 94642
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.9443570971488953,
      "learning_rate": 0.00038299878098643027,
      "loss": 2.2906,
      "step": 94643
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.9471124410629272,
      "learning_rate": 0.0003829948500829692,
      "loss": 2.2358,
      "step": 94644
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.8101771473884583,
      "learning_rate": 0.0003829909191640778,
      "loss": 1.9987,
      "step": 94645
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.8924679756164551,
      "learning_rate": 0.0003829869882297569,
      "loss": 2.2763,
      "step": 94646
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.7847338914871216,
      "learning_rate": 0.00038298305728000706,
      "loss": 2.0288,
      "step": 94647
    },
    {
      "epoch": 1.23,
      "grad_norm": 1.0462920665740967,
      "learning_rate": 0.00038297912631482913,
      "loss": 2.2034,
      "step": 94648
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.8473559021949768,
      "learning_rate": 0.00038297519533422386,
      "loss": 2.1927,
      "step": 94649
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.9006155133247375,
      "learning_rate": 0.0003829712643381919,
      "loss": 1.9531,
      "step": 94650
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.959163248538971,
      "learning_rate": 0.000382967333326734,
      "loss": 2.0637,
      "step": 94651
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.8399814963340759,
      "learning_rate": 0.00038296340229985093,
      "loss": 2.1853,
      "step": 94652
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.8975058197975159,
      "learning_rate": 0.0003829594712575434,
      "loss": 2.1864,
      "step": 94653
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.8432493805885315,
      "learning_rate": 0.00038295554019981216,
      "loss": 2.1323,
      "step": 94654
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.8670333623886108,
      "learning_rate": 0.0003829516091266579,
      "loss": 2.1725,
      "step": 94655
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.8141884207725525,
      "learning_rate": 0.00038294767803808127,
      "loss": 2.1068,
      "step": 94656
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.9278773069381714,
      "learning_rate": 0.0003829437469340833,
      "loss": 2.3377,
      "step": 94657
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.7874312400817871,
      "learning_rate": 0.00038293981581466436,
      "loss": 2.4135,
      "step": 94658
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.8500888347625732,
      "learning_rate": 0.0003829358846798254,
      "loss": 2.09,
      "step": 94659
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.8416839241981506,
      "learning_rate": 0.0003829319535295671,
      "loss": 2.2065,
      "step": 94660
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.8113781809806824,
      "learning_rate": 0.00038292802236389014,
      "loss": 2.0393,
      "step": 94661
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.8422855734825134,
      "learning_rate": 0.00038292409118279535,
      "loss": 2.4247,
      "step": 94662
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.8344301581382751,
      "learning_rate": 0.00038292015998628336,
      "loss": 2.0866,
      "step": 94663
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.834301769733429,
      "learning_rate": 0.00038291622877435506,
      "loss": 2.1996,
      "step": 94664
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.8865115642547607,
      "learning_rate": 0.0003829122975470109,
      "loss": 2.3097,
      "step": 94665
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.7713276147842407,
      "learning_rate": 0.0003829083663042519,
      "loss": 2.0248,
      "step": 94666
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.8442140221595764,
      "learning_rate": 0.0003829044350460787,
      "loss": 2.1662,
      "step": 94667
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.8007364869117737,
      "learning_rate": 0.00038290050377249186,
      "loss": 2.1236,
      "step": 94668
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.8688359260559082,
      "learning_rate": 0.00038289657248349233,
      "loss": 2.2193,
      "step": 94669
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.8202524781227112,
      "learning_rate": 0.0003828926411790808,
      "loss": 2.0584,
      "step": 94670
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.8597320318222046,
      "learning_rate": 0.00038288870985925787,
      "loss": 2.1617,
      "step": 94671
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.8589019775390625,
      "learning_rate": 0.00038288477852402444,
      "loss": 2.2906,
      "step": 94672
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.7881651520729065,
      "learning_rate": 0.00038288084717338114,
      "loss": 1.9943,
      "step": 94673
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.8377363085746765,
      "learning_rate": 0.0003828769158073287,
      "loss": 2.2675,
      "step": 94674
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.8547800183296204,
      "learning_rate": 0.00038287298442586795,
      "loss": 2.2345,
      "step": 94675
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.9439942240715027,
      "learning_rate": 0.0003828690530289996,
      "loss": 2.2919,
      "step": 94676
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.8593795895576477,
      "learning_rate": 0.0003828651216167241,
      "loss": 2.0725,
      "step": 94677
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.8044166564941406,
      "learning_rate": 0.00038286119018904254,
      "loss": 2.202,
      "step": 94678
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.8769918084144592,
      "learning_rate": 0.00038285725874595555,
      "loss": 2.21,
      "step": 94679
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.8529384732246399,
      "learning_rate": 0.0003828533272874638,
      "loss": 2.2042,
      "step": 94680
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.8906022906303406,
      "learning_rate": 0.000382849395813568,
      "loss": 2.0527,
      "step": 94681
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.8965041041374207,
      "learning_rate": 0.000382845464324269,
      "loss": 2.0686,
      "step": 94682
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.8830259442329407,
      "learning_rate": 0.0003828415328195675,
      "loss": 2.1071,
      "step": 94683
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.8987444043159485,
      "learning_rate": 0.0003828376012994641,
      "loss": 2.3104,
      "step": 94684
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.9643908739089966,
      "learning_rate": 0.00038283366976395975,
      "loss": 2.234,
      "step": 94685
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.9870932698249817,
      "learning_rate": 0.0003828297382130549,
      "loss": 2.0061,
      "step": 94686
    },
    {
      "epoch": 1.23,
      "grad_norm": 1.1948809623718262,
      "learning_rate": 0.00038282580664675053,
      "loss": 2.2401,
      "step": 94687
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.8904845714569092,
      "learning_rate": 0.0003828218750650473,
      "loss": 1.9679,
      "step": 94688
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.885963499546051,
      "learning_rate": 0.00038281794346794594,
      "loss": 1.8867,
      "step": 94689
    },
    {
      "epoch": 1.23,
      "grad_norm": 1.0139156579971313,
      "learning_rate": 0.000382814011855447,
      "loss": 2.0676,
      "step": 94690
    },
    {
      "epoch": 1.23,
      "grad_norm": 1.027266263961792,
      "learning_rate": 0.0003828100802275516,
      "loss": 2.0801,
      "step": 94691
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.9788540601730347,
      "learning_rate": 0.00038280614858426004,
      "loss": 2.0564,
      "step": 94692
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.888033390045166,
      "learning_rate": 0.0003828022169255734,
      "loss": 2.1831,
      "step": 94693
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.8614052534103394,
      "learning_rate": 0.0003827982852514922,
      "loss": 1.9802,
      "step": 94694
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.8298830389976501,
      "learning_rate": 0.0003827943535620172,
      "loss": 1.9586,
      "step": 94695
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.8580881357192993,
      "learning_rate": 0.0003827904218571492,
      "loss": 2.2255,
      "step": 94696
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.8871672749519348,
      "learning_rate": 0.00038278649013688895,
      "loss": 2.1678,
      "step": 94697
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.8722251057624817,
      "learning_rate": 0.00038278255840123706,
      "loss": 2.2156,
      "step": 94698
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.8671209216117859,
      "learning_rate": 0.0003827786266501943,
      "loss": 2.2481,
      "step": 94699
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.8556146025657654,
      "learning_rate": 0.00038277469488376154,
      "loss": 2.1619,
      "step": 94700
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.9326722621917725,
      "learning_rate": 0.00038277076310193933,
      "loss": 2.1288,
      "step": 94701
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.9213071465492249,
      "learning_rate": 0.0003827668313047285,
      "loss": 2.2066,
      "step": 94702
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.8453187346458435,
      "learning_rate": 0.0003827628994921298,
      "loss": 2.023,
      "step": 94703
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.9536416530609131,
      "learning_rate": 0.00038275896766414377,
      "loss": 2.2331,
      "step": 94704
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.8093844652175903,
      "learning_rate": 0.00038275503582077135,
      "loss": 2.1607,
      "step": 94705
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.8282230496406555,
      "learning_rate": 0.00038275110396201323,
      "loss": 2.0995,
      "step": 94706
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.8393397331237793,
      "learning_rate": 0.00038274717208787017,
      "loss": 2.194,
      "step": 94707
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.7896828651428223,
      "learning_rate": 0.0003827432401983428,
      "loss": 2.1891,
      "step": 94708
    },
    {
      "epoch": 1.23,
      "grad_norm": 1.2696901559829712,
      "learning_rate": 0.0003827393082934319,
      "loss": 2.1127,
      "step": 94709
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.9082801938056946,
      "learning_rate": 0.00038273537637313815,
      "loss": 2.402,
      "step": 94710
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.8280712962150574,
      "learning_rate": 0.00038273144443746234,
      "loss": 2.0809,
      "step": 94711
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.9938317537307739,
      "learning_rate": 0.0003827275124864053,
      "loss": 2.2596,
      "step": 94712
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.8839277029037476,
      "learning_rate": 0.00038272358051996756,
      "loss": 2.135,
      "step": 94713
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.9482874274253845,
      "learning_rate": 0.00038271964853814995,
      "loss": 2.2207,
      "step": 94714
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.9830823540687561,
      "learning_rate": 0.00038271571654095325,
      "loss": 2.041,
      "step": 94715
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.9045015573501587,
      "learning_rate": 0.0003827117845283781,
      "loss": 2.3386,
      "step": 94716
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.8040948510169983,
      "learning_rate": 0.0003827078525004252,
      "loss": 2.0924,
      "step": 94717
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.8712854385375977,
      "learning_rate": 0.0003827039204570954,
      "loss": 2.1891,
      "step": 94718
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.9052286148071289,
      "learning_rate": 0.00038269998839838946,
      "loss": 2.3893,
      "step": 94719
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.8779988884925842,
      "learning_rate": 0.000382696056324308,
      "loss": 2.2471,
      "step": 94720
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.8837127089500427,
      "learning_rate": 0.0003826921242348517,
      "loss": 2.3491,
      "step": 94721
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.9524793028831482,
      "learning_rate": 0.0003826881921300215,
      "loss": 2.2135,
      "step": 94722
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.8971977829933167,
      "learning_rate": 0.00038268426000981787,
      "loss": 2.1783,
      "step": 94723
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.841620683670044,
      "learning_rate": 0.0003826803278742417,
      "loss": 2.0955,
      "step": 94724
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.8106679320335388,
      "learning_rate": 0.00038267639572329374,
      "loss": 2.3356,
      "step": 94725
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.9783040881156921,
      "learning_rate": 0.0003826724635569747,
      "loss": 2.2872,
      "step": 94726
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.8061697483062744,
      "learning_rate": 0.0003826685313752852,
      "loss": 2.1046,
      "step": 94727
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.908734142780304,
      "learning_rate": 0.0003826645991782262,
      "loss": 2.2228,
      "step": 94728
    },
    {
      "epoch": 1.23,
      "grad_norm": 1.0606083869934082,
      "learning_rate": 0.0003826606669657982,
      "loss": 2.3087,
      "step": 94729
    },
    {
      "epoch": 1.23,
      "grad_norm": 1.0188031196594238,
      "learning_rate": 0.000382656734738002,
      "loss": 2.1978,
      "step": 94730
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.8231010437011719,
      "learning_rate": 0.00038265280249483836,
      "loss": 2.0704,
      "step": 94731
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.8858817219734192,
      "learning_rate": 0.00038264887023630806,
      "loss": 2.085,
      "step": 94732
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.8198027610778809,
      "learning_rate": 0.0003826449379624117,
      "loss": 2.1856,
      "step": 94733
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.84531569480896,
      "learning_rate": 0.0003826410056731501,
      "loss": 2.2704,
      "step": 94734
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.8183465600013733,
      "learning_rate": 0.0003826370733685241,
      "loss": 2.2975,
      "step": 94735
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.8466594219207764,
      "learning_rate": 0.00038263314104853413,
      "loss": 2.217,
      "step": 94736
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.9032260179519653,
      "learning_rate": 0.0003826292087131812,
      "loss": 2.1783,
      "step": 94737
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.8478610515594482,
      "learning_rate": 0.00038262527636246594,
      "loss": 2.2563,
      "step": 94738
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.7630190849304199,
      "learning_rate": 0.000382621343996389,
      "loss": 2.1172,
      "step": 94739
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.8447912931442261,
      "learning_rate": 0.00038261741161495124,
      "loss": 2.1829,
      "step": 94740
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.8760302662849426,
      "learning_rate": 0.00038261347921815335,
      "loss": 2.1676,
      "step": 94741
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.8845071196556091,
      "learning_rate": 0.00038260954680599616,
      "loss": 2.1895,
      "step": 94742
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.8027557134628296,
      "learning_rate": 0.0003826056143784801,
      "loss": 2.0551,
      "step": 94743
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.8820207715034485,
      "learning_rate": 0.0003826016819356062,
      "loss": 2.1646,
      "step": 94744
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.9332505464553833,
      "learning_rate": 0.00038259774947737516,
      "loss": 1.9881,
      "step": 94745
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.8295027017593384,
      "learning_rate": 0.0003825938170037875,
      "loss": 2.2848,
      "step": 94746
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.9013848304748535,
      "learning_rate": 0.00038258988451484415,
      "loss": 2.0973,
      "step": 94747
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.8100948929786682,
      "learning_rate": 0.00038258595201054576,
      "loss": 2.1548,
      "step": 94748
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.8850510120391846,
      "learning_rate": 0.0003825820194908932,
      "loss": 2.2617,
      "step": 94749
    },
    {
      "epoch": 1.23,
      "grad_norm": 1.176418423652649,
      "learning_rate": 0.00038257808695588694,
      "loss": 2.1084,
      "step": 94750
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.8178606629371643,
      "learning_rate": 0.00038257415440552786,
      "loss": 2.2689,
      "step": 94751
    },
    {
      "epoch": 1.23,
      "grad_norm": 1.1193263530731201,
      "learning_rate": 0.0003825702218398168,
      "loss": 2.2237,
      "step": 94752
    },
    {
      "epoch": 1.23,
      "grad_norm": 1.0419285297393799,
      "learning_rate": 0.00038256628925875425,
      "loss": 2.1617,
      "step": 94753
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.8797603845596313,
      "learning_rate": 0.00038256235666234113,
      "loss": 2.2404,
      "step": 94754
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.8234107494354248,
      "learning_rate": 0.00038255842405057815,
      "loss": 1.959,
      "step": 94755
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.9075702428817749,
      "learning_rate": 0.00038255449142346595,
      "loss": 2.0827,
      "step": 94756
    },
    {
      "epoch": 1.23,
      "grad_norm": 1.2642706632614136,
      "learning_rate": 0.00038255055878100525,
      "loss": 2.043,
      "step": 94757
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.8467271327972412,
      "learning_rate": 0.000382546626123197,
      "loss": 2.2583,
      "step": 94758
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.8433104753494263,
      "learning_rate": 0.0003825426934500417,
      "loss": 2.4127,
      "step": 94759
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.8836593627929688,
      "learning_rate": 0.0003825387607615401,
      "loss": 2.0095,
      "step": 94760
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.9367268085479736,
      "learning_rate": 0.000382534828057693,
      "loss": 2.0506,
      "step": 94761
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.8473638296127319,
      "learning_rate": 0.0003825308953385012,
      "loss": 2.0694,
      "step": 94762
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.884391188621521,
      "learning_rate": 0.00038252696260396524,
      "loss": 2.1264,
      "step": 94763
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.8311881422996521,
      "learning_rate": 0.0003825230298540861,
      "loss": 2.3362,
      "step": 94764
    },
    {
      "epoch": 1.23,
      "grad_norm": 1.0553171634674072,
      "learning_rate": 0.00038251909708886424,
      "loss": 2.3006,
      "step": 94765
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.8273673057556152,
      "learning_rate": 0.0003825151643083005,
      "loss": 2.2598,
      "step": 94766
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.9211512207984924,
      "learning_rate": 0.0003825112315123958,
      "loss": 2.3661,
      "step": 94767
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.9532338976860046,
      "learning_rate": 0.00038250729870115054,
      "loss": 2.401,
      "step": 94768
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.8838189244270325,
      "learning_rate": 0.0003825033658745657,
      "loss": 2.3512,
      "step": 94769
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.8275169730186462,
      "learning_rate": 0.00038249943303264193,
      "loss": 1.9643,
      "step": 94770
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.9007828831672668,
      "learning_rate": 0.0003824955001753799,
      "loss": 2.2554,
      "step": 94771
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.9985030293464661,
      "learning_rate": 0.00038249156730278043,
      "loss": 2.3773,
      "step": 94772
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.8219286799430847,
      "learning_rate": 0.00038248763441484427,
      "loss": 2.3578,
      "step": 94773
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.8932993412017822,
      "learning_rate": 0.000382483701511572,
      "loss": 2.2004,
      "step": 94774
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.8176220059394836,
      "learning_rate": 0.0003824797685929645,
      "loss": 2.1237,
      "step": 94775
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.828144371509552,
      "learning_rate": 0.00038247583565902257,
      "loss": 2.0557,
      "step": 94776
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.8327077627182007,
      "learning_rate": 0.00038247190270974664,
      "loss": 2.3758,
      "step": 94777
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.978134274482727,
      "learning_rate": 0.0003824679697451377,
      "loss": 2.0979,
      "step": 94778
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.868134081363678,
      "learning_rate": 0.00038246403676519647,
      "loss": 2.3169,
      "step": 94779
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.8523785471916199,
      "learning_rate": 0.00038246010376992353,
      "loss": 2.2225,
      "step": 94780
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.9153534173965454,
      "learning_rate": 0.00038245617075931965,
      "loss": 2.1187,
      "step": 94781
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.8214977383613586,
      "learning_rate": 0.0003824522377333858,
      "loss": 1.9805,
      "step": 94782
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.8944437503814697,
      "learning_rate": 0.0003824483046921224,
      "loss": 2.177,
      "step": 94783
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.8341051340103149,
      "learning_rate": 0.00038244437163553025,
      "loss": 2.2607,
      "step": 94784
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.8196382522583008,
      "learning_rate": 0.0003824404385636102,
      "loss": 2.226,
      "step": 94785
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.7949358224868774,
      "learning_rate": 0.00038243650547636296,
      "loss": 2.186,
      "step": 94786
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.8744991421699524,
      "learning_rate": 0.0003824325723737892,
      "loss": 2.0857,
      "step": 94787
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.8283096551895142,
      "learning_rate": 0.00038242863925588963,
      "loss": 2.4432,
      "step": 94788
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.8233734369277954,
      "learning_rate": 0.00038242470612266503,
      "loss": 2.0041,
      "step": 94789
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.7534247636795044,
      "learning_rate": 0.00038242077297411615,
      "loss": 2.1636,
      "step": 94790
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.8064558506011963,
      "learning_rate": 0.00038241683981024365,
      "loss": 2.3689,
      "step": 94791
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.892570972442627,
      "learning_rate": 0.00038241290663104834,
      "loss": 2.262,
      "step": 94792
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.9848843216896057,
      "learning_rate": 0.00038240897343653086,
      "loss": 2.0989,
      "step": 94793
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.8027659058570862,
      "learning_rate": 0.00038240504022669204,
      "loss": 2.0783,
      "step": 94794
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.9152215123176575,
      "learning_rate": 0.00038240110700153257,
      "loss": 2.1456,
      "step": 94795
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.8636303544044495,
      "learning_rate": 0.0003823971737610532,
      "loss": 1.9369,
      "step": 94796
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.8239777684211731,
      "learning_rate": 0.00038239324050525457,
      "loss": 2.0612,
      "step": 94797
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.8566985726356506,
      "learning_rate": 0.0003823893072341376,
      "loss": 2.2985,
      "step": 94798
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.8959490060806274,
      "learning_rate": 0.0003823853739477028,
      "loss": 2.5194,
      "step": 94799
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.8817920684814453,
      "learning_rate": 0.00038238144064595104,
      "loss": 2.206,
      "step": 94800
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.8289949893951416,
      "learning_rate": 0.000382377507328883,
      "loss": 2.0013,
      "step": 94801
    },
    {
      "epoch": 1.23,
      "grad_norm": 1.0630978345870972,
      "learning_rate": 0.00038237357399649945,
      "loss": 2.0507,
      "step": 94802
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.8872594833374023,
      "learning_rate": 0.000382369640648801,
      "loss": 2.242,
      "step": 94803
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.8205710053443909,
      "learning_rate": 0.0003823657072857886,
      "loss": 2.2821,
      "step": 94804
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.859531581401825,
      "learning_rate": 0.0003823617739074628,
      "loss": 2.0341,
      "step": 94805
    },
    {
      "epoch": 1.23,
      "grad_norm": 1.1271321773529053,
      "learning_rate": 0.0003823578405138244,
      "loss": 2.3523,
      "step": 94806
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.7969144582748413,
      "learning_rate": 0.00038235390710487417,
      "loss": 2.1305,
      "step": 94807
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.8420615792274475,
      "learning_rate": 0.00038234997368061273,
      "loss": 2.0173,
      "step": 94808
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.9544979929924011,
      "learning_rate": 0.0003823460402410409,
      "loss": 2.1079,
      "step": 94809
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.9444714188575745,
      "learning_rate": 0.00038234210678615943,
      "loss": 2.2576,
      "step": 94810
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.8032059669494629,
      "learning_rate": 0.00038233817331596897,
      "loss": 2.2858,
      "step": 94811
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.8916457891464233,
      "learning_rate": 0.0003823342398304703,
      "loss": 2.3237,
      "step": 94812
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.8725741505622864,
      "learning_rate": 0.0003823303063296641,
      "loss": 2.3461,
      "step": 94813
    },
    {
      "epoch": 1.23,
      "grad_norm": 1.0239615440368652,
      "learning_rate": 0.0003823263728135512,
      "loss": 2.3273,
      "step": 94814
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.8729003667831421,
      "learning_rate": 0.0003823224392821322,
      "loss": 2.3035,
      "step": 94815
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.8243657350540161,
      "learning_rate": 0.00038231850573540796,
      "loss": 2.3537,
      "step": 94816
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.7591322064399719,
      "learning_rate": 0.0003823145721733791,
      "loss": 1.8322,
      "step": 94817
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.8362765908241272,
      "learning_rate": 0.00038231063859604645,
      "loss": 2.2719,
      "step": 94818
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.9030651450157166,
      "learning_rate": 0.0003823067050034108,
      "loss": 2.1053,
      "step": 94819
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.8648353219032288,
      "learning_rate": 0.0003823027713954726,
      "loss": 2.3334,
      "step": 94820
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.9419980049133301,
      "learning_rate": 0.00038229883777223287,
      "loss": 2.0934,
      "step": 94821
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.7764524817466736,
      "learning_rate": 0.00038229490413369216,
      "loss": 2.1185,
      "step": 94822
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.8780238032341003,
      "learning_rate": 0.0003822909704798513,
      "loss": 2.1697,
      "step": 94823
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.9082143306732178,
      "learning_rate": 0.00038228703681071104,
      "loss": 2.2643,
      "step": 94824
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.8850861191749573,
      "learning_rate": 0.00038228310312627203,
      "loss": 1.9341,
      "step": 94825
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.9490033388137817,
      "learning_rate": 0.0003822791694265351,
      "loss": 2.1367,
      "step": 94826
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.7996860146522522,
      "learning_rate": 0.0003822752357115008,
      "loss": 2.0698,
      "step": 94827
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.8716555833816528,
      "learning_rate": 0.00038227130198117005,
      "loss": 2.1284,
      "step": 94828
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.8535351157188416,
      "learning_rate": 0.00038226736823554345,
      "loss": 2.1763,
      "step": 94829
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.7770561575889587,
      "learning_rate": 0.00038226343447462195,
      "loss": 2.3247,
      "step": 94830
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.8700219392776489,
      "learning_rate": 0.00038225950069840594,
      "loss": 1.982,
      "step": 94831
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.8453314304351807,
      "learning_rate": 0.00038225556690689645,
      "loss": 1.8146,
      "step": 94832
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.8741366267204285,
      "learning_rate": 0.00038225163310009406,
      "loss": 2.2809,
      "step": 94833
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.940894365310669,
      "learning_rate": 0.0003822476992779996,
      "loss": 2.0241,
      "step": 94834
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.9502739310264587,
      "learning_rate": 0.0003822437654406136,
      "loss": 2.24,
      "step": 94835
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.8382943272590637,
      "learning_rate": 0.000382239831587937,
      "loss": 2.1723,
      "step": 94836
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.9346253275871277,
      "learning_rate": 0.00038223589771997055,
      "loss": 2.1262,
      "step": 94837
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.9657649397850037,
      "learning_rate": 0.00038223196383671483,
      "loss": 1.9507,
      "step": 94838
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.851173996925354,
      "learning_rate": 0.0003822280299381706,
      "loss": 2.0092,
      "step": 94839
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.9006849527359009,
      "learning_rate": 0.0003822240960243387,
      "loss": 2.2154,
      "step": 94840
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.9265116453170776,
      "learning_rate": 0.00038222016209521974,
      "loss": 2.0036,
      "step": 94841
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.8296328783035278,
      "learning_rate": 0.00038221622815081443,
      "loss": 2.0235,
      "step": 94842
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.7948670983314514,
      "learning_rate": 0.00038221229419112377,
      "loss": 2.1584,
      "step": 94843
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.8408280611038208,
      "learning_rate": 0.0003822083602161481,
      "loss": 2.0255,
      "step": 94844
    },
    {
      "epoch": 1.23,
      "grad_norm": 1.030806303024292,
      "learning_rate": 0.00038220442622588843,
      "loss": 2.1878,
      "step": 94845
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.861474335193634,
      "learning_rate": 0.0003822004922203454,
      "loss": 2.2086,
      "step": 94846
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.7693079710006714,
      "learning_rate": 0.00038219655819951977,
      "loss": 2.2557,
      "step": 94847
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.8521096110343933,
      "learning_rate": 0.00038219262416341215,
      "loss": 2.3315,
      "step": 94848
    },
    {
      "epoch": 1.23,
      "grad_norm": 0.8966348767280579,
      "learning_rate": 0.0003821886901120235,
      "loss": 2.0675,
      "step": 94849
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.8605670928955078,
      "learning_rate": 0.00038218475604535435,
      "loss": 2.2178,
      "step": 94850
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.934138834476471,
      "learning_rate": 0.00038218082196340546,
      "loss": 2.3213,
      "step": 94851
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.8433021903038025,
      "learning_rate": 0.0003821768878661777,
      "loss": 2.2824,
      "step": 94852
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.7789985537528992,
      "learning_rate": 0.00038217295375367165,
      "loss": 2.0929,
      "step": 94853
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.8089863061904907,
      "learning_rate": 0.00038216901962588807,
      "loss": 2.2003,
      "step": 94854
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.8395354747772217,
      "learning_rate": 0.00038216508548282775,
      "loss": 2.3061,
      "step": 94855
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.7977077960968018,
      "learning_rate": 0.00038216115132449146,
      "loss": 1.9198,
      "step": 94856
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.9510762691497803,
      "learning_rate": 0.0003821572171508798,
      "loss": 2.1213,
      "step": 94857
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.8867294192314148,
      "learning_rate": 0.0003821532829619935,
      "loss": 1.8148,
      "step": 94858
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.8703052997589111,
      "learning_rate": 0.00038214934875783346,
      "loss": 2.3485,
      "step": 94859
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.8801649212837219,
      "learning_rate": 0.0003821454145384002,
      "loss": 2.0797,
      "step": 94860
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.8341217637062073,
      "learning_rate": 0.0003821414803036946,
      "loss": 2.2347,
      "step": 94861
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.824883759021759,
      "learning_rate": 0.0003821375460537174,
      "loss": 2.273,
      "step": 94862
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.9176898002624512,
      "learning_rate": 0.0003821336117884692,
      "loss": 2.2898,
      "step": 94863
    },
    {
      "epoch": 1.24,
      "grad_norm": 1.1050069332122803,
      "learning_rate": 0.0003821296775079508,
      "loss": 2.1076,
      "step": 94864
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.8139151930809021,
      "learning_rate": 0.00038212574321216307,
      "loss": 2.2636,
      "step": 94865
    },
    {
      "epoch": 1.24,
      "grad_norm": 1.1100324392318726,
      "learning_rate": 0.00038212180890110655,
      "loss": 2.1789,
      "step": 94866
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.7592031359672546,
      "learning_rate": 0.000382117874574782,
      "loss": 2.1441,
      "step": 94867
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.8896073698997498,
      "learning_rate": 0.00038211394023319025,
      "loss": 2.1341,
      "step": 94868
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.8468211889266968,
      "learning_rate": 0.0003821100058763319,
      "loss": 2.2878,
      "step": 94869
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.7947489619255066,
      "learning_rate": 0.0003821060715042078,
      "loss": 1.9967,
      "step": 94870
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.8213880062103271,
      "learning_rate": 0.00038210213711681865,
      "loss": 2.0104,
      "step": 94871
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.7994445562362671,
      "learning_rate": 0.0003820982027141651,
      "loss": 2.007,
      "step": 94872
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.9341797232627869,
      "learning_rate": 0.000382094268296248,
      "loss": 2.2146,
      "step": 94873
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.8532963395118713,
      "learning_rate": 0.00038209033386306797,
      "loss": 2.4214,
      "step": 94874
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.9179748892784119,
      "learning_rate": 0.0003820863994146258,
      "loss": 2.2352,
      "step": 94875
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.8259716629981995,
      "learning_rate": 0.0003820824649509222,
      "loss": 2.0478,
      "step": 94876
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.9382782578468323,
      "learning_rate": 0.000382078530471958,
      "loss": 2.2501,
      "step": 94877
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.9364871978759766,
      "learning_rate": 0.0003820745959777338,
      "loss": 2.1286,
      "step": 94878
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.8883347511291504,
      "learning_rate": 0.0003820706614682504,
      "loss": 2.1758,
      "step": 94879
    },
    {
      "epoch": 1.24,
      "grad_norm": 1.0172889232635498,
      "learning_rate": 0.00038206672694350856,
      "loss": 2.4607,
      "step": 94880
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.85884690284729,
      "learning_rate": 0.0003820627924035089,
      "loss": 1.9369,
      "step": 94881
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.8930890560150146,
      "learning_rate": 0.0003820588578482522,
      "loss": 2.0729,
      "step": 94882
    },
    {
      "epoch": 1.24,
      "grad_norm": 1.0123381614685059,
      "learning_rate": 0.00038205492327773925,
      "loss": 2.3571,
      "step": 94883
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.8918641209602356,
      "learning_rate": 0.00038205098869197073,
      "loss": 2.1912,
      "step": 94884
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.8030427694320679,
      "learning_rate": 0.0003820470540909474,
      "loss": 2.1341,
      "step": 94885
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.8401533961296082,
      "learning_rate": 0.00038204311947466995,
      "loss": 2.1689,
      "step": 94886
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.7876310348510742,
      "learning_rate": 0.0003820391848431392,
      "loss": 1.9819,
      "step": 94887
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.8194173574447632,
      "learning_rate": 0.00038203525019635574,
      "loss": 2.0663,
      "step": 94888
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.8751510977745056,
      "learning_rate": 0.0003820313155343204,
      "loss": 2.1579,
      "step": 94889
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.8236806392669678,
      "learning_rate": 0.0003820273808570339,
      "loss": 2.1882,
      "step": 94890
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.8395264744758606,
      "learning_rate": 0.0003820234461644969,
      "loss": 2.1368,
      "step": 94891
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.9035307168960571,
      "learning_rate": 0.0003820195114567103,
      "loss": 2.2252,
      "step": 94892
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.8548246026039124,
      "learning_rate": 0.00038201557673367463,
      "loss": 2.1006,
      "step": 94893
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.9346504211425781,
      "learning_rate": 0.0003820116419953908,
      "loss": 2.1149,
      "step": 94894
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.7976332306861877,
      "learning_rate": 0.0003820077072418594,
      "loss": 1.9646,
      "step": 94895
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.9660432934761047,
      "learning_rate": 0.0003820037724730812,
      "loss": 2.0924,
      "step": 94896
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.8765924572944641,
      "learning_rate": 0.00038199983768905695,
      "loss": 2.2255,
      "step": 94897
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.8748909831047058,
      "learning_rate": 0.0003819959028897875,
      "loss": 2.1846,
      "step": 94898
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.8186120390892029,
      "learning_rate": 0.00038199196807527336,
      "loss": 2.2577,
      "step": 94899
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.8238480091094971,
      "learning_rate": 0.0003819880332455154,
      "loss": 2.2649,
      "step": 94900
    },
    {
      "epoch": 1.24,
      "grad_norm": 1.1340546607971191,
      "learning_rate": 0.0003819840984005143,
      "loss": 2.1038,
      "step": 94901
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.8917712569236755,
      "learning_rate": 0.0003819801635402708,
      "loss": 2.2386,
      "step": 94902
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.9759207963943481,
      "learning_rate": 0.00038197622866478564,
      "loss": 2.103,
      "step": 94903
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.8864732384681702,
      "learning_rate": 0.0003819722937740596,
      "loss": 2.2089,
      "step": 94904
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.9491866230964661,
      "learning_rate": 0.0003819683588680933,
      "loss": 2.093,
      "step": 94905
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.9146832823753357,
      "learning_rate": 0.00038196442394688757,
      "loss": 2.2588,
      "step": 94906
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.9984584450721741,
      "learning_rate": 0.0003819604890104431,
      "loss": 2.2827,
      "step": 94907
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.8907075524330139,
      "learning_rate": 0.00038195655405876065,
      "loss": 2.1042,
      "step": 94908
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.8616296648979187,
      "learning_rate": 0.00038195261909184085,
      "loss": 2.2724,
      "step": 94909
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.9401915073394775,
      "learning_rate": 0.00038194868410968455,
      "loss": 2.2408,
      "step": 94910
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.9307218194007874,
      "learning_rate": 0.0003819447491122925,
      "loss": 1.9842,
      "step": 94911
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.8779902458190918,
      "learning_rate": 0.00038194081409966533,
      "loss": 2.1476,
      "step": 94912
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.8253142833709717,
      "learning_rate": 0.0003819368790718038,
      "loss": 2.2443,
      "step": 94913
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.8346466422080994,
      "learning_rate": 0.0003819329440287087,
      "loss": 2.2323,
      "step": 94914
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.8556522130966187,
      "learning_rate": 0.0003819290089703807,
      "loss": 2.0537,
      "step": 94915
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.8865342736244202,
      "learning_rate": 0.00038192507389682055,
      "loss": 2.1448,
      "step": 94916
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.8008215427398682,
      "learning_rate": 0.00038192113880802895,
      "loss": 2.015,
      "step": 94917
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.8154293894767761,
      "learning_rate": 0.0003819172037040067,
      "loss": 2.3204,
      "step": 94918
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.8383607864379883,
      "learning_rate": 0.00038191326858475454,
      "loss": 2.2139,
      "step": 94919
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.8229764103889465,
      "learning_rate": 0.0003819093334502731,
      "loss": 2.0065,
      "step": 94920
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.7774920463562012,
      "learning_rate": 0.0003819053983005631,
      "loss": 2.2013,
      "step": 94921
    },
    {
      "epoch": 1.24,
      "grad_norm": 1.049414873123169,
      "learning_rate": 0.0003819014631356255,
      "loss": 2.1326,
      "step": 94922
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.8107484579086304,
      "learning_rate": 0.00038189752795546073,
      "loss": 2.1925,
      "step": 94923
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.9433892369270325,
      "learning_rate": 0.00038189359276006966,
      "loss": 2.13,
      "step": 94924
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.8370677828788757,
      "learning_rate": 0.0003818896575494531,
      "loss": 2.045,
      "step": 94925
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.9440562129020691,
      "learning_rate": 0.0003818857223236118,
      "loss": 1.964,
      "step": 94926
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.8441503047943115,
      "learning_rate": 0.0003818817870825462,
      "loss": 2.205,
      "step": 94927
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.7850871682167053,
      "learning_rate": 0.0003818778518262573,
      "loss": 2.1567,
      "step": 94928
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.8628842830657959,
      "learning_rate": 0.0003818739165547458,
      "loss": 2.3635,
      "step": 94929
    },
    {
      "epoch": 1.24,
      "grad_norm": 1.106141209602356,
      "learning_rate": 0.0003818699812680124,
      "loss": 2.6617,
      "step": 94930
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.7936644554138184,
      "learning_rate": 0.0003818660459660578,
      "loss": 2.0069,
      "step": 94931
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.8264908194541931,
      "learning_rate": 0.00038186211064888275,
      "loss": 2.3434,
      "step": 94932
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.8195525407791138,
      "learning_rate": 0.000381858175316488,
      "loss": 2.1247,
      "step": 94933
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.8105414509773254,
      "learning_rate": 0.0003818542399688742,
      "loss": 2.1968,
      "step": 94934
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.7909901738166809,
      "learning_rate": 0.00038185030460604236,
      "loss": 2.3033,
      "step": 94935
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.8516099452972412,
      "learning_rate": 0.0003818463692279928,
      "loss": 2.1697,
      "step": 94936
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.7985185384750366,
      "learning_rate": 0.0003818424338347265,
      "loss": 2.1711,
      "step": 94937
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.8041640520095825,
      "learning_rate": 0.0003818384984262442,
      "loss": 2.1681,
      "step": 94938
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.9325999021530151,
      "learning_rate": 0.00038183456300254655,
      "loss": 2.1643,
      "step": 94939
    },
    {
      "epoch": 1.24,
      "grad_norm": 1.1731019020080566,
      "learning_rate": 0.0003818306275636342,
      "loss": 2.1605,
      "step": 94940
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.9130677580833435,
      "learning_rate": 0.00038182669210950815,
      "loss": 2.3738,
      "step": 94941
    },
    {
      "epoch": 1.24,
      "grad_norm": 1.0288528203964233,
      "learning_rate": 0.00038182275664016895,
      "loss": 2.4061,
      "step": 94942
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.9509080052375793,
      "learning_rate": 0.0003818188211556173,
      "loss": 2.1474,
      "step": 94943
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.8882237672805786,
      "learning_rate": 0.000381814885655854,
      "loss": 2.3486,
      "step": 94944
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.8749069571495056,
      "learning_rate": 0.00038181095014087977,
      "loss": 2.2053,
      "step": 94945
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.9245705008506775,
      "learning_rate": 0.00038180701461069537,
      "loss": 2.1101,
      "step": 94946
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.9042525887489319,
      "learning_rate": 0.00038180307906530146,
      "loss": 2.4446,
      "step": 94947
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.8859807252883911,
      "learning_rate": 0.00038179914350469884,
      "loss": 2.2036,
      "step": 94948
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.9936931729316711,
      "learning_rate": 0.00038179520792888823,
      "loss": 2.246,
      "step": 94949
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.841474175453186,
      "learning_rate": 0.00038179127233787033,
      "loss": 2.2195,
      "step": 94950
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.9419023394584656,
      "learning_rate": 0.00038178733673164584,
      "loss": 1.9915,
      "step": 94951
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.870299220085144,
      "learning_rate": 0.00038178340111021564,
      "loss": 2.3526,
      "step": 94952
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.9120058417320251,
      "learning_rate": 0.0003817794654735803,
      "loss": 2.3006,
      "step": 94953
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.8306275010108948,
      "learning_rate": 0.0003817755298217406,
      "loss": 2.063,
      "step": 94954
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.868025004863739,
      "learning_rate": 0.00038177159415469727,
      "loss": 2.0628,
      "step": 94955
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.7872204780578613,
      "learning_rate": 0.0003817676584724512,
      "loss": 1.964,
      "step": 94956
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.8285651803016663,
      "learning_rate": 0.0003817637227750028,
      "loss": 1.9444,
      "step": 94957
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.9221881031990051,
      "learning_rate": 0.000381759787062353,
      "loss": 2.2185,
      "step": 94958
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.8130984902381897,
      "learning_rate": 0.00038175585133450263,
      "loss": 2.0768,
      "step": 94959
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.8767569065093994,
      "learning_rate": 0.0003817519155914522,
      "loss": 2.3425,
      "step": 94960
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.8837764263153076,
      "learning_rate": 0.0003817479798332026,
      "loss": 2.3011,
      "step": 94961
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.8271650671958923,
      "learning_rate": 0.0003817440440597545,
      "loss": 2.1564,
      "step": 94962
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.7994689345359802,
      "learning_rate": 0.00038174010827110865,
      "loss": 2.0737,
      "step": 94963
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.8830505609512329,
      "learning_rate": 0.00038173617246726565,
      "loss": 2.0927,
      "step": 94964
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.8996939659118652,
      "learning_rate": 0.0003817322366482266,
      "loss": 2.3167,
      "step": 94965
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.8732757568359375,
      "learning_rate": 0.0003817283008139918,
      "loss": 2.1718,
      "step": 94966
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.8926630616188049,
      "learning_rate": 0.00038172436496456216,
      "loss": 2.1711,
      "step": 94967
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.8523141741752625,
      "learning_rate": 0.00038172042909993847,
      "loss": 2.3012,
      "step": 94968
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.792498767375946,
      "learning_rate": 0.0003817164932201214,
      "loss": 2.0773,
      "step": 94969
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.9626453518867493,
      "learning_rate": 0.00038171255732511166,
      "loss": 2.1941,
      "step": 94970
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.8379306793212891,
      "learning_rate": 0.00038170862141491017,
      "loss": 2.2922,
      "step": 94971
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.8043102025985718,
      "learning_rate": 0.00038170468548951727,
      "loss": 2.2039,
      "step": 94972
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.9244291186332703,
      "learning_rate": 0.000381700749548934,
      "loss": 2.1815,
      "step": 94973
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.857714831829071,
      "learning_rate": 0.0003816968135931611,
      "loss": 2.1819,
      "step": 94974
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.8612378835678101,
      "learning_rate": 0.00038169287762219913,
      "loss": 2.2822,
      "step": 94975
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.8274553418159485,
      "learning_rate": 0.000381688941636049,
      "loss": 2.1011,
      "step": 94976
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.9293942451477051,
      "learning_rate": 0.0003816850056347114,
      "loss": 2.2057,
      "step": 94977
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.8542874455451965,
      "learning_rate": 0.00038168106961818685,
      "loss": 2.2047,
      "step": 94978
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.9228222370147705,
      "learning_rate": 0.00038167713358647627,
      "loss": 2.2388,
      "step": 94979
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.8375787734985352,
      "learning_rate": 0.0003816731975395805,
      "loss": 2.3626,
      "step": 94980
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.8008981943130493,
      "learning_rate": 0.0003816692614775001,
      "loss": 2.0606,
      "step": 94981
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.8027002215385437,
      "learning_rate": 0.0003816653254002358,
      "loss": 1.9389,
      "step": 94982
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.8473520874977112,
      "learning_rate": 0.00038166138930778836,
      "loss": 2.2743,
      "step": 94983
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.8705434799194336,
      "learning_rate": 0.00038165745320015853,
      "loss": 2.1483,
      "step": 94984
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.9079391956329346,
      "learning_rate": 0.00038165351707734706,
      "loss": 1.9953,
      "step": 94985
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.8685329556465149,
      "learning_rate": 0.0003816495809393546,
      "loss": 1.9119,
      "step": 94986
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.8964998126029968,
      "learning_rate": 0.00038164564478618203,
      "loss": 2.2063,
      "step": 94987
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.8105648756027222,
      "learning_rate": 0.00038164170861782994,
      "loss": 2.1513,
      "step": 94988
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.8560488820075989,
      "learning_rate": 0.00038163777243429913,
      "loss": 2.0657,
      "step": 94989
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.7723397612571716,
      "learning_rate": 0.00038163383623559036,
      "loss": 2.234,
      "step": 94990
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.8408728241920471,
      "learning_rate": 0.0003816299000217043,
      "loss": 2.0462,
      "step": 94991
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.933621346950531,
      "learning_rate": 0.00038162596379264163,
      "loss": 2.1911,
      "step": 94992
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.8725109696388245,
      "learning_rate": 0.00038162202754840326,
      "loss": 2.1665,
      "step": 94993
    },
    {
      "epoch": 1.24,
      "grad_norm": 1.0768886804580688,
      "learning_rate": 0.00038161809128898973,
      "loss": 2.3818,
      "step": 94994
    },
    {
      "epoch": 1.24,
      "grad_norm": 1.02993643283844,
      "learning_rate": 0.0003816141550144018,
      "loss": 2.0335,
      "step": 94995
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.8557446599006653,
      "learning_rate": 0.00038161021872464045,
      "loss": 2.03,
      "step": 94996
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.930936336517334,
      "learning_rate": 0.0003816062824197061,
      "loss": 2.0266,
      "step": 94997
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.939292848110199,
      "learning_rate": 0.0003816023460995996,
      "loss": 2.1757,
      "step": 94998
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.9140956997871399,
      "learning_rate": 0.0003815984097643217,
      "loss": 2.0156,
      "step": 94999
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.8535177111625671,
      "learning_rate": 0.0003815944734138731,
      "loss": 2.3289,
      "step": 95000
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.8804454803466797,
      "learning_rate": 0.00038159053704825457,
      "loss": 2.0329,
      "step": 95001
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.9608508944511414,
      "learning_rate": 0.0003815866006674668,
      "loss": 2.2838,
      "step": 95002
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.9442126154899597,
      "learning_rate": 0.00038158266427151057,
      "loss": 2.0771,
      "step": 95003
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.9657636880874634,
      "learning_rate": 0.00038157872786038653,
      "loss": 2.0768,
      "step": 95004
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.86882483959198,
      "learning_rate": 0.0003815747914340955,
      "loss": 2.0997,
      "step": 95005
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.9281606078147888,
      "learning_rate": 0.0003815708549926381,
      "loss": 2.3704,
      "step": 95006
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.891554057598114,
      "learning_rate": 0.00038156691853601533,
      "loss": 2.0509,
      "step": 95007
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.9165340065956116,
      "learning_rate": 0.0003815629820642276,
      "loss": 2.1548,
      "step": 95008
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.8571286201477051,
      "learning_rate": 0.00038155904557727574,
      "loss": 2.218,
      "step": 95009
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.8873360753059387,
      "learning_rate": 0.0003815551090751606,
      "loss": 2.1991,
      "step": 95010
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.8737542033195496,
      "learning_rate": 0.00038155117255788274,
      "loss": 2.2105,
      "step": 95011
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.829145610332489,
      "learning_rate": 0.000381547236025443,
      "loss": 2.2,
      "step": 95012
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.906337559223175,
      "learning_rate": 0.00038154329947784213,
      "loss": 2.2126,
      "step": 95013
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.889853835105896,
      "learning_rate": 0.0003815393629150808,
      "loss": 2.1509,
      "step": 95014
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.8296612501144409,
      "learning_rate": 0.0003815354263371598,
      "loss": 2.3884,
      "step": 95015
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.8603962063789368,
      "learning_rate": 0.00038153148974407977,
      "loss": 2.2248,
      "step": 95016
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.9220597147941589,
      "learning_rate": 0.00038152755313584154,
      "loss": 2.2093,
      "step": 95017
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.845282793045044,
      "learning_rate": 0.00038152361651244574,
      "loss": 2.1457,
      "step": 95018
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.8538560271263123,
      "learning_rate": 0.0003815196798738932,
      "loss": 1.9914,
      "step": 95019
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.8695106506347656,
      "learning_rate": 0.00038151574322018464,
      "loss": 2.3219,
      "step": 95020
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.8522026538848877,
      "learning_rate": 0.0003815118065513207,
      "loss": 2.1663,
      "step": 95021
    },
    {
      "epoch": 1.24,
      "grad_norm": 1.0571843385696411,
      "learning_rate": 0.0003815078698673022,
      "loss": 2.1619,
      "step": 95022
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.9508640170097351,
      "learning_rate": 0.0003815039331681299,
      "loss": 2.2294,
      "step": 95023
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.7551062107086182,
      "learning_rate": 0.00038149999645380444,
      "loss": 2.1877,
      "step": 95024
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.8044446110725403,
      "learning_rate": 0.00038149605972432655,
      "loss": 2.1829,
      "step": 95025
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.8293607234954834,
      "learning_rate": 0.0003814921229796971,
      "loss": 2.2074,
      "step": 95026
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.9699999690055847,
      "learning_rate": 0.00038148818621991663,
      "loss": 2.1975,
      "step": 95027
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.8131282925605774,
      "learning_rate": 0.000381484249444986,
      "loss": 2.5114,
      "step": 95028
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.8297426700592041,
      "learning_rate": 0.000381480312654906,
      "loss": 2.2702,
      "step": 95029
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.8686494827270508,
      "learning_rate": 0.0003814763758496771,
      "loss": 2.2448,
      "step": 95030
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.9030948281288147,
      "learning_rate": 0.0003814724390293004,
      "loss": 2.152,
      "step": 95031
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.8401514291763306,
      "learning_rate": 0.00038146850219377635,
      "loss": 2.1769,
      "step": 95032
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.8126506209373474,
      "learning_rate": 0.0003814645653431057,
      "loss": 2.4165,
      "step": 95033
    },
    {
      "epoch": 1.24,
      "grad_norm": 1.1007558107376099,
      "learning_rate": 0.0003814606284772893,
      "loss": 1.9831,
      "step": 95034
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.8565950393676758,
      "learning_rate": 0.00038145669159632786,
      "loss": 2.1945,
      "step": 95035
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.8058601021766663,
      "learning_rate": 0.00038145275470022203,
      "loss": 2.1519,
      "step": 95036
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.8402548432350159,
      "learning_rate": 0.0003814488177889726,
      "loss": 2.1316,
      "step": 95037
    },
    {
      "epoch": 1.24,
      "grad_norm": 1.2430553436279297,
      "learning_rate": 0.00038144488086258035,
      "loss": 2.1354,
      "step": 95038
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.8314982652664185,
      "learning_rate": 0.0003814409439210459,
      "loss": 2.1607,
      "step": 95039
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.8694687485694885,
      "learning_rate": 0.00038143700696437004,
      "loss": 2.2486,
      "step": 95040
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.8345483541488647,
      "learning_rate": 0.0003814330699925536,
      "loss": 2.4005,
      "step": 95041
    },
    {
      "epoch": 1.24,
      "grad_norm": 1.1071034669876099,
      "learning_rate": 0.0003814291330055971,
      "loss": 2.1766,
      "step": 95042
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.9679797291755676,
      "learning_rate": 0.00038142519600350137,
      "loss": 2.2125,
      "step": 95043
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.8440513014793396,
      "learning_rate": 0.0003814212589862673,
      "loss": 2.1972,
      "step": 95044
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.9908653497695923,
      "learning_rate": 0.0003814173219538954,
      "loss": 2.1427,
      "step": 95045
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.8632960319519043,
      "learning_rate": 0.0003814133849063865,
      "loss": 2.2737,
      "step": 95046
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.7999590039253235,
      "learning_rate": 0.0003814094478437413,
      "loss": 2.2025,
      "step": 95047
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.8654642105102539,
      "learning_rate": 0.00038140551076596053,
      "loss": 2.0221,
      "step": 95048
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.9487912654876709,
      "learning_rate": 0.00038140157367304496,
      "loss": 2.1939,
      "step": 95049
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.9554847478866577,
      "learning_rate": 0.0003813976365649953,
      "loss": 2.2183,
      "step": 95050
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.8868615031242371,
      "learning_rate": 0.0003813936994418123,
      "loss": 2.2332,
      "step": 95051
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.832580029964447,
      "learning_rate": 0.00038138976230349663,
      "loss": 2.3398,
      "step": 95052
    },
    {
      "epoch": 1.24,
      "grad_norm": 1.2190333604812622,
      "learning_rate": 0.0003813858251500491,
      "loss": 2.0655,
      "step": 95053
    },
    {
      "epoch": 1.24,
      "grad_norm": 1.0442439317703247,
      "learning_rate": 0.00038138188798147044,
      "loss": 1.8352,
      "step": 95054
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.9121828675270081,
      "learning_rate": 0.00038137795079776123,
      "loss": 2.3904,
      "step": 95055
    },
    {
      "epoch": 1.24,
      "grad_norm": 1.1867444515228271,
      "learning_rate": 0.0003813740135989224,
      "loss": 2.1626,
      "step": 95056
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.8131657838821411,
      "learning_rate": 0.0003813700763849547,
      "loss": 2.3654,
      "step": 95057
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.9764209985733032,
      "learning_rate": 0.0003813661391558586,
      "loss": 2.1475,
      "step": 95058
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.7870388627052307,
      "learning_rate": 0.0003813622019116351,
      "loss": 2.1579,
      "step": 95059
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.8487241864204407,
      "learning_rate": 0.00038135826465228483,
      "loss": 2.0221,
      "step": 95060
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.9552677869796753,
      "learning_rate": 0.0003813543273778085,
      "loss": 2.2297,
      "step": 95061
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.8794766664505005,
      "learning_rate": 0.0003813503900882068,
      "loss": 2.256,
      "step": 95062
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.8123955130577087,
      "learning_rate": 0.0003813464527834807,
      "loss": 2.2384,
      "step": 95063
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.8584151864051819,
      "learning_rate": 0.0003813425154636306,
      "loss": 2.1795,
      "step": 95064
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.9911616444587708,
      "learning_rate": 0.0003813385781286574,
      "loss": 2.1022,
      "step": 95065
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.9679984450340271,
      "learning_rate": 0.00038133464077856193,
      "loss": 2.2588,
      "step": 95066
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.8582208752632141,
      "learning_rate": 0.0003813307034133447,
      "loss": 2.043,
      "step": 95067
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.9565668106079102,
      "learning_rate": 0.0003813267660330066,
      "loss": 2.3174,
      "step": 95068
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.9430316090583801,
      "learning_rate": 0.0003813228286375484,
      "loss": 2.0557,
      "step": 95069
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.9217216372489929,
      "learning_rate": 0.0003813188912269706,
      "loss": 2.2523,
      "step": 95070
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.9364989995956421,
      "learning_rate": 0.00038131495380127413,
      "loss": 1.9792,
      "step": 95071
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.9851758480072021,
      "learning_rate": 0.0003813110163604598,
      "loss": 2.161,
      "step": 95072
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.9455491900444031,
      "learning_rate": 0.0003813070789045281,
      "loss": 2.1314,
      "step": 95073
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.8374528884887695,
      "learning_rate": 0.00038130314143347983,
      "loss": 2.4141,
      "step": 95074
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.904684841632843,
      "learning_rate": 0.0003812992039473159,
      "loss": 2.0786,
      "step": 95075
    },
    {
      "epoch": 1.24,
      "grad_norm": 1.3407939672470093,
      "learning_rate": 0.00038129526644603687,
      "loss": 2.1844,
      "step": 95076
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.8324173092842102,
      "learning_rate": 0.00038129132892964347,
      "loss": 2.2397,
      "step": 95077
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.9911448955535889,
      "learning_rate": 0.00038128739139813657,
      "loss": 2.0919,
      "step": 95078
    },
    {
      "epoch": 1.24,
      "grad_norm": 1.188660979270935,
      "learning_rate": 0.00038128345385151665,
      "loss": 2.2916,
      "step": 95079
    },
    {
      "epoch": 1.24,
      "grad_norm": 1.1627919673919678,
      "learning_rate": 0.00038127951628978473,
      "loss": 2.4194,
      "step": 95080
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.8823956847190857,
      "learning_rate": 0.00038127557871294143,
      "loss": 2.0922,
      "step": 95081
    },
    {
      "epoch": 1.24,
      "grad_norm": 1.0132313966751099,
      "learning_rate": 0.0003812716411209874,
      "loss": 2.1541,
      "step": 95082
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.9360503554344177,
      "learning_rate": 0.00038126770351392335,
      "loss": 2.2034,
      "step": 95083
    },
    {
      "epoch": 1.24,
      "grad_norm": 1.043481469154358,
      "learning_rate": 0.0003812637658917503,
      "loss": 2.3926,
      "step": 95084
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.8647186160087585,
      "learning_rate": 0.0003812598282544687,
      "loss": 1.9714,
      "step": 95085
    },
    {
      "epoch": 1.24,
      "grad_norm": 1.0950974225997925,
      "learning_rate": 0.00038125589060207933,
      "loss": 2.2563,
      "step": 95086
    },
    {
      "epoch": 1.24,
      "grad_norm": 1.0194352865219116,
      "learning_rate": 0.00038125195293458295,
      "loss": 2.3327,
      "step": 95087
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.7810841202735901,
      "learning_rate": 0.00038124801525198034,
      "loss": 2.1527,
      "step": 95088
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.9991023540496826,
      "learning_rate": 0.0003812440775542722,
      "loss": 2.0509,
      "step": 95089
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.8056820631027222,
      "learning_rate": 0.00038124013984145913,
      "loss": 2.0723,
      "step": 95090
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.7995372414588928,
      "learning_rate": 0.0003812362021135422,
      "loss": 2.0352,
      "step": 95091
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.9156971573829651,
      "learning_rate": 0.00038123226437052174,
      "loss": 2.2766,
      "step": 95092
    },
    {
      "epoch": 1.24,
      "grad_norm": 1.0434930324554443,
      "learning_rate": 0.00038122832661239875,
      "loss": 2.1157,
      "step": 95093
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.9043428301811218,
      "learning_rate": 0.00038122438883917385,
      "loss": 1.8363,
      "step": 95094
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.9617646932601929,
      "learning_rate": 0.0003812204510508478,
      "loss": 2.1356,
      "step": 95095
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.959365725517273,
      "learning_rate": 0.00038121651324742143,
      "loss": 2.0384,
      "step": 95096
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.9948912858963013,
      "learning_rate": 0.0003812125754288952,
      "loss": 2.1674,
      "step": 95097
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.8207255601882935,
      "learning_rate": 0.00038120863759527016,
      "loss": 2.0496,
      "step": 95098
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.9231554865837097,
      "learning_rate": 0.0003812046997465469,
      "loss": 2.0543,
      "step": 95099
    },
    {
      "epoch": 1.24,
      "grad_norm": 1.0229227542877197,
      "learning_rate": 0.00038120076188272607,
      "loss": 1.8979,
      "step": 95100
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.9250190854072571,
      "learning_rate": 0.0003811968240038086,
      "loss": 2.041,
      "step": 95101
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.8920192122459412,
      "learning_rate": 0.00038119288610979506,
      "loss": 2.1866,
      "step": 95102
    },
    {
      "epoch": 1.24,
      "grad_norm": 1.003299355506897,
      "learning_rate": 0.0003811889482006862,
      "loss": 1.9515,
      "step": 95103
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.9278847575187683,
      "learning_rate": 0.0003811850102764827,
      "loss": 2.2583,
      "step": 95104
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.8630682826042175,
      "learning_rate": 0.0003811810723371855,
      "loss": 2.3428,
      "step": 95105
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.8826584815979004,
      "learning_rate": 0.0003811771343827952,
      "loss": 2.0193,
      "step": 95106
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.8171201944351196,
      "learning_rate": 0.00038117319641331243,
      "loss": 2.266,
      "step": 95107
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.9362824559211731,
      "learning_rate": 0.0003811692584287382,
      "loss": 2.1422,
      "step": 95108
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.9223228693008423,
      "learning_rate": 0.000381165320429073,
      "loss": 2.1983,
      "step": 95109
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.9018136262893677,
      "learning_rate": 0.0003811613824143176,
      "loss": 2.1055,
      "step": 95110
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.9288283586502075,
      "learning_rate": 0.0003811574443844728,
      "loss": 2.199,
      "step": 95111
    },
    {
      "epoch": 1.24,
      "grad_norm": 1.0221527814865112,
      "learning_rate": 0.00038115350633953927,
      "loss": 2.1313,
      "step": 95112
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.9059779047966003,
      "learning_rate": 0.00038114956827951774,
      "loss": 2.4813,
      "step": 95113
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.9099768400192261,
      "learning_rate": 0.00038114563020440905,
      "loss": 2.4645,
      "step": 95114
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.8337173461914062,
      "learning_rate": 0.00038114169211421386,
      "loss": 2.1262,
      "step": 95115
    },
    {
      "epoch": 1.24,
      "grad_norm": 1.0559685230255127,
      "learning_rate": 0.00038113775400893285,
      "loss": 2.1871,
      "step": 95116
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.9228476285934448,
      "learning_rate": 0.0003811338158885668,
      "loss": 2.305,
      "step": 95117
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.8256500363349915,
      "learning_rate": 0.00038112987775311646,
      "loss": 2.1068,
      "step": 95118
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.8243878483772278,
      "learning_rate": 0.00038112593960258254,
      "loss": 2.4123,
      "step": 95119
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.9021176099777222,
      "learning_rate": 0.00038112200143696574,
      "loss": 2.1621,
      "step": 95120
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.7876430153846741,
      "learning_rate": 0.00038111806325626693,
      "loss": 2.2457,
      "step": 95121
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.8591034412384033,
      "learning_rate": 0.0003811141250604866,
      "loss": 2.2485,
      "step": 95122
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.8875012993812561,
      "learning_rate": 0.0003811101868496257,
      "loss": 2.0501,
      "step": 95123
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.9124780893325806,
      "learning_rate": 0.000381106248623685,
      "loss": 2.2586,
      "step": 95124
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.8811543583869934,
      "learning_rate": 0.00038110231038266496,
      "loss": 2.1635,
      "step": 95125
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.8167721629142761,
      "learning_rate": 0.00038109837212656645,
      "loss": 2.4896,
      "step": 95126
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.7730928063392639,
      "learning_rate": 0.0003810944338553903,
      "loss": 2.1595,
      "step": 95127
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.8243553638458252,
      "learning_rate": 0.00038109049556913716,
      "loss": 2.0704,
      "step": 95128
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.9545049667358398,
      "learning_rate": 0.0003810865572678078,
      "loss": 2.1462,
      "step": 95129
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.7361269593238831,
      "learning_rate": 0.00038108261895140286,
      "loss": 2.2505,
      "step": 95130
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.8263640403747559,
      "learning_rate": 0.0003810786806199231,
      "loss": 2.243,
      "step": 95131
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.8569650053977966,
      "learning_rate": 0.0003810747422733693,
      "loss": 1.8651,
      "step": 95132
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.8323276042938232,
      "learning_rate": 0.00038107080391174225,
      "loss": 2.0452,
      "step": 95133
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.8249744176864624,
      "learning_rate": 0.00038106686553504256,
      "loss": 1.9724,
      "step": 95134
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.9235860109329224,
      "learning_rate": 0.00038106292714327095,
      "loss": 2.2419,
      "step": 95135
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.8569403290748596,
      "learning_rate": 0.0003810589887364283,
      "loss": 2.1144,
      "step": 95136
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.7649361491203308,
      "learning_rate": 0.00038105505031451524,
      "loss": 2.0481,
      "step": 95137
    },
    {
      "epoch": 1.24,
      "grad_norm": 1.000411868095398,
      "learning_rate": 0.0003810511118775325,
      "loss": 2.044,
      "step": 95138
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.8297264575958252,
      "learning_rate": 0.00038104717342548087,
      "loss": 2.1326,
      "step": 95139
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.8717120885848999,
      "learning_rate": 0.00038104323495836094,
      "loss": 2.2479,
      "step": 95140
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.8681648373603821,
      "learning_rate": 0.0003810392964761736,
      "loss": 2.2559,
      "step": 95141
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.8050861358642578,
      "learning_rate": 0.00038103535797891957,
      "loss": 2.3255,
      "step": 95142
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.8850453495979309,
      "learning_rate": 0.0003810314194665995,
      "loss": 2.2193,
      "step": 95143
    },
    {
      "epoch": 1.24,
      "grad_norm": 1.071894884109497,
      "learning_rate": 0.00038102748093921405,
      "loss": 2.3447,
      "step": 95144
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.8697682023048401,
      "learning_rate": 0.0003810235423967643,
      "loss": 2.2462,
      "step": 95145
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.9283099174499512,
      "learning_rate": 0.00038101960383925054,
      "loss": 2.272,
      "step": 95146
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.975073516368866,
      "learning_rate": 0.00038101566526667374,
      "loss": 2.1688,
      "step": 95147
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.9065352082252502,
      "learning_rate": 0.0003810117266790347,
      "loss": 2.516,
      "step": 95148
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.7548681497573853,
      "learning_rate": 0.00038100778807633395,
      "loss": 1.939,
      "step": 95149
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.8883154392242432,
      "learning_rate": 0.0003810038494585723,
      "loss": 2.0268,
      "step": 95150
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.8399969339370728,
      "learning_rate": 0.0003809999108257507,
      "loss": 2.2086,
      "step": 95151
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.9420726299285889,
      "learning_rate": 0.0003809959721778695,
      "loss": 2.2602,
      "step": 95152
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.8298723697662354,
      "learning_rate": 0.00038099203351492963,
      "loss": 2.1791,
      "step": 95153
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.8797097206115723,
      "learning_rate": 0.0003809880948369319,
      "loss": 2.0318,
      "step": 95154
    },
    {
      "epoch": 1.24,
      "grad_norm": 1.1704185009002686,
      "learning_rate": 0.0003809841561438769,
      "loss": 2.2517,
      "step": 95155
    },
    {
      "epoch": 1.24,
      "grad_norm": 1.017568588256836,
      "learning_rate": 0.00038098021743576536,
      "loss": 2.2572,
      "step": 95156
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.9495511651039124,
      "learning_rate": 0.00038097627871259813,
      "loss": 2.4523,
      "step": 95157
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.9117209315299988,
      "learning_rate": 0.0003809723399743759,
      "loss": 2.1938,
      "step": 95158
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.8235606551170349,
      "learning_rate": 0.0003809684012210994,
      "loss": 2.2534,
      "step": 95159
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.8331851959228516,
      "learning_rate": 0.0003809644624527693,
      "loss": 2.2041,
      "step": 95160
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.9517487287521362,
      "learning_rate": 0.00038096052366938635,
      "loss": 2.0565,
      "step": 95161
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.8516626358032227,
      "learning_rate": 0.00038095658487095137,
      "loss": 2.3962,
      "step": 95162
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.9917020797729492,
      "learning_rate": 0.00038095264605746506,
      "loss": 2.3583,
      "step": 95163
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.8035535216331482,
      "learning_rate": 0.0003809487072289281,
      "loss": 2.2618,
      "step": 95164
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.8779018521308899,
      "learning_rate": 0.00038094476838534117,
      "loss": 2.1335,
      "step": 95165
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.7736124396324158,
      "learning_rate": 0.00038094082952670516,
      "loss": 2.0298,
      "step": 95166
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.8691033720970154,
      "learning_rate": 0.0003809368906530207,
      "loss": 2.1645,
      "step": 95167
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.808000385761261,
      "learning_rate": 0.00038093295176428853,
      "loss": 2.197,
      "step": 95168
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.9076477289199829,
      "learning_rate": 0.00038092901286050944,
      "loss": 2.0058,
      "step": 95169
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.8271815776824951,
      "learning_rate": 0.00038092507394168405,
      "loss": 2.2819,
      "step": 95170
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.9211326837539673,
      "learning_rate": 0.0003809211350078132,
      "loss": 2.0769,
      "step": 95171
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.8070668578147888,
      "learning_rate": 0.0003809171960588976,
      "loss": 2.293,
      "step": 95172
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.8525922894477844,
      "learning_rate": 0.0003809132570949379,
      "loss": 2.2538,
      "step": 95173
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.7974445819854736,
      "learning_rate": 0.000380909318115935,
      "loss": 2.3068,
      "step": 95174
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.9179664254188538,
      "learning_rate": 0.00038090537912188944,
      "loss": 2.202,
      "step": 95175
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.8107883334159851,
      "learning_rate": 0.00038090144011280206,
      "loss": 1.8986,
      "step": 95176
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.7967137694358826,
      "learning_rate": 0.0003808975010886736,
      "loss": 2.2748,
      "step": 95177
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.8587924242019653,
      "learning_rate": 0.00038089356204950474,
      "loss": 2.1655,
      "step": 95178
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.8134405016899109,
      "learning_rate": 0.00038088962299529633,
      "loss": 2.2837,
      "step": 95179
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.8924757242202759,
      "learning_rate": 0.0003808856839260489,
      "loss": 2.3194,
      "step": 95180
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.8688940405845642,
      "learning_rate": 0.0003808817448417633,
      "loss": 2.2686,
      "step": 95181
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.9676674604415894,
      "learning_rate": 0.00038087780574244033,
      "loss": 2.1666,
      "step": 95182
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.8304467797279358,
      "learning_rate": 0.00038087386662808066,
      "loss": 2.0686,
      "step": 95183
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.920944333076477,
      "learning_rate": 0.0003808699274986849,
      "loss": 2.2332,
      "step": 95184
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.8733590841293335,
      "learning_rate": 0.00038086598835425394,
      "loss": 2.2547,
      "step": 95185
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.7895981073379517,
      "learning_rate": 0.00038086204919478853,
      "loss": 2.1598,
      "step": 95186
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.8302128314971924,
      "learning_rate": 0.00038085811002028924,
      "loss": 2.1991,
      "step": 95187
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.8020514249801636,
      "learning_rate": 0.00038085417083075695,
      "loss": 2.3276,
      "step": 95188
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.8277769684791565,
      "learning_rate": 0.00038085023162619236,
      "loss": 1.8338,
      "step": 95189
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.8918243646621704,
      "learning_rate": 0.00038084629240659623,
      "loss": 2.1371,
      "step": 95190
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.9161606431007385,
      "learning_rate": 0.00038084235317196915,
      "loss": 2.1659,
      "step": 95191
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.9214919209480286,
      "learning_rate": 0.000380838413922312,
      "loss": 2.1417,
      "step": 95192
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.8258557319641113,
      "learning_rate": 0.0003808344746576254,
      "loss": 2.2432,
      "step": 95193
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.8800202012062073,
      "learning_rate": 0.00038083053537791027,
      "loss": 2.2238,
      "step": 95194
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.8258949518203735,
      "learning_rate": 0.0003808265960831671,
      "loss": 2.2382,
      "step": 95195
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.857224702835083,
      "learning_rate": 0.0003808226567733968,
      "loss": 2.0632,
      "step": 95196
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.8938482999801636,
      "learning_rate": 0.00038081871744860004,
      "loss": 2.1048,
      "step": 95197
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.8317682147026062,
      "learning_rate": 0.00038081477810877754,
      "loss": 2.0881,
      "step": 95198
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.8741863965988159,
      "learning_rate": 0.00038081083875393006,
      "loss": 2.1645,
      "step": 95199
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.8802415728569031,
      "learning_rate": 0.0003808068993840583,
      "loss": 2.2997,
      "step": 95200
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.8383619785308838,
      "learning_rate": 0.000380802959999163,
      "loss": 2.221,
      "step": 95201
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.9294973611831665,
      "learning_rate": 0.000380799020599245,
      "loss": 2.2028,
      "step": 95202
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.8915176391601562,
      "learning_rate": 0.00038079508118430485,
      "loss": 2.1628,
      "step": 95203
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.8971629738807678,
      "learning_rate": 0.0003807911417543434,
      "loss": 2.3079,
      "step": 95204
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.8565099835395813,
      "learning_rate": 0.0003807872023093613,
      "loss": 2.0466,
      "step": 95205
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.9370788931846619,
      "learning_rate": 0.0003807832628493594,
      "loss": 2.0857,
      "step": 95206
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.8643019795417786,
      "learning_rate": 0.00038077932337433836,
      "loss": 2.25,
      "step": 95207
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.8235759735107422,
      "learning_rate": 0.00038077538388429883,
      "loss": 2.1177,
      "step": 95208
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.8736279010772705,
      "learning_rate": 0.00038077144437924175,
      "loss": 2.3362,
      "step": 95209
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.8395076990127563,
      "learning_rate": 0.00038076750485916766,
      "loss": 2.0526,
      "step": 95210
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.8893010020256042,
      "learning_rate": 0.00038076356532407737,
      "loss": 2.143,
      "step": 95211
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.8371720910072327,
      "learning_rate": 0.0003807596257739717,
      "loss": 2.1442,
      "step": 95212
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.880555272102356,
      "learning_rate": 0.00038075568620885117,
      "loss": 2.0145,
      "step": 95213
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.8054815530776978,
      "learning_rate": 0.00038075174662871667,
      "loss": 2.1507,
      "step": 95214
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.7759124636650085,
      "learning_rate": 0.00038074780703356895,
      "loss": 2.0161,
      "step": 95215
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.8955773711204529,
      "learning_rate": 0.0003807438674234086,
      "loss": 2.4194,
      "step": 95216
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.8315297961235046,
      "learning_rate": 0.0003807399277982365,
      "loss": 2.2002,
      "step": 95217
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.8499147891998291,
      "learning_rate": 0.00038073598815805337,
      "loss": 2.1807,
      "step": 95218
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.8703789114952087,
      "learning_rate": 0.0003807320485028598,
      "loss": 2.2929,
      "step": 95219
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.9093947410583496,
      "learning_rate": 0.0003807281088326566,
      "loss": 2.2098,
      "step": 95220
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.7792257070541382,
      "learning_rate": 0.0003807241691474446,
      "loss": 2.0645,
      "step": 95221
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.9454873204231262,
      "learning_rate": 0.00038072022944722443,
      "loss": 2.21,
      "step": 95222
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.9147692322731018,
      "learning_rate": 0.0003807162897319968,
      "loss": 1.9299,
      "step": 95223
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.9378246665000916,
      "learning_rate": 0.0003807123500017626,
      "loss": 2.1036,
      "step": 95224
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.9712052941322327,
      "learning_rate": 0.0003807084102565224,
      "loss": 1.9785,
      "step": 95225
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.7939456701278687,
      "learning_rate": 0.0003807044704962769,
      "loss": 2.3495,
      "step": 95226
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.971518337726593,
      "learning_rate": 0.00038070053072102706,
      "loss": 2.1914,
      "step": 95227
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.9404546618461609,
      "learning_rate": 0.00038069659093077336,
      "loss": 2.3669,
      "step": 95228
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.8182876110076904,
      "learning_rate": 0.00038069265112551664,
      "loss": 2.2077,
      "step": 95229
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.8999013900756836,
      "learning_rate": 0.0003806887113052577,
      "loss": 2.2072,
      "step": 95230
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.8138515949249268,
      "learning_rate": 0.0003806847714699972,
      "loss": 2.2677,
      "step": 95231
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.8716062903404236,
      "learning_rate": 0.0003806808316197358,
      "loss": 2.0968,
      "step": 95232
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.863492488861084,
      "learning_rate": 0.0003806768917544744,
      "loss": 2.099,
      "step": 95233
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.8849173784255981,
      "learning_rate": 0.0003806729518742136,
      "loss": 1.8393,
      "step": 95234
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.9054695963859558,
      "learning_rate": 0.00038066901197895416,
      "loss": 2.1462,
      "step": 95235
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.9392178058624268,
      "learning_rate": 0.00038066507206869685,
      "loss": 2.1798,
      "step": 95236
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.865418016910553,
      "learning_rate": 0.0003806611321434423,
      "loss": 2.2317,
      "step": 95237
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.8394814133644104,
      "learning_rate": 0.0003806571922031914,
      "loss": 2.072,
      "step": 95238
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.8947535753250122,
      "learning_rate": 0.0003806532522479449,
      "loss": 2.0384,
      "step": 95239
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.8730847835540771,
      "learning_rate": 0.0003806493122777033,
      "loss": 2.3263,
      "step": 95240
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.8494974970817566,
      "learning_rate": 0.0003806453722924675,
      "loss": 2.2241,
      "step": 95241
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.8495809435844421,
      "learning_rate": 0.0003806414322922383,
      "loss": 2.1465,
      "step": 95242
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.9699842929840088,
      "learning_rate": 0.0003806374922770162,
      "loss": 2.124,
      "step": 95243
    },
    {
      "epoch": 1.24,
      "grad_norm": 1.0856562852859497,
      "learning_rate": 0.0003806335522468021,
      "loss": 1.9736,
      "step": 95244
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.8466454148292542,
      "learning_rate": 0.00038062961220159677,
      "loss": 2.1386,
      "step": 95245
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.8092003464698792,
      "learning_rate": 0.0003806256721414008,
      "loss": 1.9912,
      "step": 95246
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.8837389945983887,
      "learning_rate": 0.000380621732066215,
      "loss": 2.1327,
      "step": 95247
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.8237816691398621,
      "learning_rate": 0.0003806177919760401,
      "loss": 2.1289,
      "step": 95248
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.838189959526062,
      "learning_rate": 0.00038061385187087687,
      "loss": 2.4194,
      "step": 95249
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.859684407711029,
      "learning_rate": 0.00038060991175072593,
      "loss": 2.324,
      "step": 95250
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.8885034322738647,
      "learning_rate": 0.0003806059716155881,
      "loss": 2.1555,
      "step": 95251
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.9093291759490967,
      "learning_rate": 0.0003806020314654642,
      "loss": 2.1417,
      "step": 95252
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.7740298509597778,
      "learning_rate": 0.00038059809130035476,
      "loss": 2.2308,
      "step": 95253
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.8431734442710876,
      "learning_rate": 0.0003805941511202606,
      "loss": 2.3106,
      "step": 95254
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.9803064465522766,
      "learning_rate": 0.0003805902109251826,
      "loss": 2.1575,
      "step": 95255
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.851142942905426,
      "learning_rate": 0.00038058627071512125,
      "loss": 2.3036,
      "step": 95256
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.852442741394043,
      "learning_rate": 0.0003805823304900774,
      "loss": 2.2632,
      "step": 95257
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.9130292534828186,
      "learning_rate": 0.0003805783902500518,
      "loss": 2.2479,
      "step": 95258
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.8552655577659607,
      "learning_rate": 0.00038057444999504514,
      "loss": 2.0949,
      "step": 95259
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.8048385381698608,
      "learning_rate": 0.0003805705097250581,
      "loss": 2.0709,
      "step": 95260
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.8309203386306763,
      "learning_rate": 0.0003805665694400917,
      "loss": 1.9797,
      "step": 95261
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.7916891574859619,
      "learning_rate": 0.0003805626291401462,
      "loss": 2.0088,
      "step": 95262
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.8501478433609009,
      "learning_rate": 0.00038055868882522274,
      "loss": 2.2659,
      "step": 95263
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.8937523365020752,
      "learning_rate": 0.00038055474849532184,
      "loss": 2.2277,
      "step": 95264
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.9187261462211609,
      "learning_rate": 0.0003805508081504443,
      "loss": 1.9647,
      "step": 95265
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.7908741235733032,
      "learning_rate": 0.00038054686779059085,
      "loss": 2.1408,
      "step": 95266
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.8242483735084534,
      "learning_rate": 0.0003805429274157622,
      "loss": 2.0164,
      "step": 95267
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.8487094640731812,
      "learning_rate": 0.0003805389870259591,
      "loss": 1.9661,
      "step": 95268
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.8685838580131531,
      "learning_rate": 0.0003805350466211823,
      "loss": 2.2525,
      "step": 95269
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.7706146836280823,
      "learning_rate": 0.0003805311062014325,
      "loss": 2.0912,
      "step": 95270
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.8034771680831909,
      "learning_rate": 0.0003805271657667105,
      "loss": 2.1622,
      "step": 95271
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.8786409497261047,
      "learning_rate": 0.0003805232253170169,
      "loss": 2.2533,
      "step": 95272
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.8269001245498657,
      "learning_rate": 0.00038051928485235254,
      "loss": 2.0518,
      "step": 95273
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.9333718419075012,
      "learning_rate": 0.0003805153443727181,
      "loss": 2.3747,
      "step": 95274
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.8415974378585815,
      "learning_rate": 0.0003805114038781144,
      "loss": 2.0532,
      "step": 95275
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.9085764288902283,
      "learning_rate": 0.0003805074633685421,
      "loss": 2.0761,
      "step": 95276
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.7711530923843384,
      "learning_rate": 0.0003805035228440019,
      "loss": 1.9542,
      "step": 95277
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.7891209125518799,
      "learning_rate": 0.00038049958230449457,
      "loss": 2.3219,
      "step": 95278
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.9097629189491272,
      "learning_rate": 0.0003804956417500209,
      "loss": 2.0938,
      "step": 95279
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.9230667948722839,
      "learning_rate": 0.0003804917011805816,
      "loss": 2.2184,
      "step": 95280
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.8738462924957275,
      "learning_rate": 0.00038048776059617723,
      "loss": 2.0732,
      "step": 95281
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.8841379880905151,
      "learning_rate": 0.0003804838199968088,
      "loss": 2.2026,
      "step": 95282
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.9039700031280518,
      "learning_rate": 0.00038047987938247686,
      "loss": 2.2099,
      "step": 95283
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.9262575507164001,
      "learning_rate": 0.0003804759387531822,
      "loss": 2.1233,
      "step": 95284
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.806225597858429,
      "learning_rate": 0.0003804719981089256,
      "loss": 1.95,
      "step": 95285
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.8009236454963684,
      "learning_rate": 0.00038046805744970766,
      "loss": 2.2258,
      "step": 95286
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.9533716440200806,
      "learning_rate": 0.0003804641167755292,
      "loss": 2.0065,
      "step": 95287
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.8374699354171753,
      "learning_rate": 0.00038046017608639096,
      "loss": 2.136,
      "step": 95288
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.8521671891212463,
      "learning_rate": 0.00038045623538229367,
      "loss": 2.3848,
      "step": 95289
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.8629457950592041,
      "learning_rate": 0.00038045229466323796,
      "loss": 2.3034,
      "step": 95290
    },
    {
      "epoch": 1.24,
      "grad_norm": 1.013537049293518,
      "learning_rate": 0.00038044835392922477,
      "loss": 2.0193,
      "step": 95291
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.8203476071357727,
      "learning_rate": 0.0003804444131802546,
      "loss": 1.9611,
      "step": 95292
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.8709245920181274,
      "learning_rate": 0.00038044047241632834,
      "loss": 2.397,
      "step": 95293
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.9064986109733582,
      "learning_rate": 0.00038043653163744674,
      "loss": 2.2059,
      "step": 95294
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.9526792764663696,
      "learning_rate": 0.0003804325908436104,
      "loss": 2.1388,
      "step": 95295
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.8450632095336914,
      "learning_rate": 0.0003804286500348201,
      "loss": 2.2771,
      "step": 95296
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.8056923747062683,
      "learning_rate": 0.0003804247092110767,
      "loss": 2.1353,
      "step": 95297
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.9838370680809021,
      "learning_rate": 0.00038042076837238077,
      "loss": 2.1551,
      "step": 95298
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.8876822590827942,
      "learning_rate": 0.0003804168275187331,
      "loss": 2.5481,
      "step": 95299
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.8711075782775879,
      "learning_rate": 0.00038041288665013445,
      "loss": 2.3748,
      "step": 95300
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.8610633611679077,
      "learning_rate": 0.0003804089457665855,
      "loss": 1.8362,
      "step": 95301
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.8245605826377869,
      "learning_rate": 0.00038040500486808695,
      "loss": 2.457,
      "step": 95302
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.8787966966629028,
      "learning_rate": 0.0003804010639546397,
      "loss": 2.2894,
      "step": 95303
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.8465878367424011,
      "learning_rate": 0.0003803971230262443,
      "loss": 2.2644,
      "step": 95304
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.8256511092185974,
      "learning_rate": 0.0003803931820829016,
      "loss": 2.1317,
      "step": 95305
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.8816028237342834,
      "learning_rate": 0.00038038924112461227,
      "loss": 2.1424,
      "step": 95306
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.8380045890808105,
      "learning_rate": 0.000380385300151377,
      "loss": 2.165,
      "step": 95307
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.8754558563232422,
      "learning_rate": 0.00038038135916319664,
      "loss": 2.2065,
      "step": 95308
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.8823079466819763,
      "learning_rate": 0.00038037741816007194,
      "loss": 2.2126,
      "step": 95309
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.7741186618804932,
      "learning_rate": 0.0003803734771420035,
      "loss": 2.051,
      "step": 95310
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.9034506678581238,
      "learning_rate": 0.00038036953610899203,
      "loss": 2.2232,
      "step": 95311
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.8191304802894592,
      "learning_rate": 0.0003803655950610385,
      "loss": 2.1672,
      "step": 95312
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.7828216552734375,
      "learning_rate": 0.0003803616539981434,
      "loss": 2.1603,
      "step": 95313
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.8171110153198242,
      "learning_rate": 0.00038035771292030754,
      "loss": 2.1852,
      "step": 95314
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.9413872957229614,
      "learning_rate": 0.00038035377182753173,
      "loss": 2.0618,
      "step": 95315
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.8427433371543884,
      "learning_rate": 0.0003803498307198166,
      "loss": 2.2445,
      "step": 95316
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.8434390425682068,
      "learning_rate": 0.0003803458895971629,
      "loss": 2.0814,
      "step": 95317
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.8418967127799988,
      "learning_rate": 0.0003803419484595713,
      "loss": 2.3021,
      "step": 95318
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.8694270253181458,
      "learning_rate": 0.0003803380073070428,
      "loss": 2.0958,
      "step": 95319
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.8850294947624207,
      "learning_rate": 0.00038033406613957783,
      "loss": 2.2946,
      "step": 95320
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.9343432784080505,
      "learning_rate": 0.00038033012495717727,
      "loss": 2.0711,
      "step": 95321
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.7621220350265503,
      "learning_rate": 0.0003803261837598418,
      "loss": 1.8253,
      "step": 95322
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.8806433081626892,
      "learning_rate": 0.00038032224254757225,
      "loss": 2.1691,
      "step": 95323
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.8345158696174622,
      "learning_rate": 0.00038031830132036925,
      "loss": 1.9269,
      "step": 95324
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.9965152740478516,
      "learning_rate": 0.0003803143600782335,
      "loss": 2.2306,
      "step": 95325
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.7983110547065735,
      "learning_rate": 0.00038031041882116587,
      "loss": 2.2114,
      "step": 95326
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.8467268943786621,
      "learning_rate": 0.00038030647754916695,
      "loss": 2.3984,
      "step": 95327
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.8462619781494141,
      "learning_rate": 0.00038030253626223764,
      "loss": 2.2469,
      "step": 95328
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.8186536431312561,
      "learning_rate": 0.0003802985949603785,
      "loss": 2.0534,
      "step": 95329
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.8823931217193604,
      "learning_rate": 0.0003802946536435903,
      "loss": 2.1797,
      "step": 95330
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.9487574696540833,
      "learning_rate": 0.00038029071231187385,
      "loss": 2.1036,
      "step": 95331
    },
    {
      "epoch": 1.24,
      "grad_norm": 1.020677924156189,
      "learning_rate": 0.0003802867709652299,
      "loss": 2.1238,
      "step": 95332
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.9205299019813538,
      "learning_rate": 0.00038028282960365906,
      "loss": 2.2491,
      "step": 95333
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.906456470489502,
      "learning_rate": 0.00038027888822716213,
      "loss": 2.1105,
      "step": 95334
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.8184148073196411,
      "learning_rate": 0.0003802749468357398,
      "loss": 2.1302,
      "step": 95335
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.859870970249176,
      "learning_rate": 0.00038027100542939283,
      "loss": 1.9858,
      "step": 95336
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.8741141557693481,
      "learning_rate": 0.0003802670640081221,
      "loss": 2.2399,
      "step": 95337
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.9494068622589111,
      "learning_rate": 0.0003802631225719281,
      "loss": 2.3531,
      "step": 95338
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.7990452647209167,
      "learning_rate": 0.00038025918112081174,
      "loss": 2.262,
      "step": 95339
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.8924829959869385,
      "learning_rate": 0.00038025523965477364,
      "loss": 2.1775,
      "step": 95340
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.8770936727523804,
      "learning_rate": 0.0003802512981738146,
      "loss": 2.2965,
      "step": 95341
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.8150348663330078,
      "learning_rate": 0.00038024735667793524,
      "loss": 2.0461,
      "step": 95342
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.832647979259491,
      "learning_rate": 0.00038024341516713645,
      "loss": 2.0831,
      "step": 95343
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.8940922617912292,
      "learning_rate": 0.00038023947364141894,
      "loss": 2.1,
      "step": 95344
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.8194050192832947,
      "learning_rate": 0.00038023553210078334,
      "loss": 2.1237,
      "step": 95345
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.9051132798194885,
      "learning_rate": 0.00038023159054523047,
      "loss": 2.2066,
      "step": 95346
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.8441011309623718,
      "learning_rate": 0.000380227648974761,
      "loss": 2.385,
      "step": 95347
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.9072028994560242,
      "learning_rate": 0.0003802237073893757,
      "loss": 2.15,
      "step": 95348
    },
    {
      "epoch": 1.24,
      "grad_norm": 1.048645257949829,
      "learning_rate": 0.00038021976578907527,
      "loss": 2.1719,
      "step": 95349
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.8852716684341431,
      "learning_rate": 0.0003802158241738605,
      "loss": 2.2265,
      "step": 95350
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.828762412071228,
      "learning_rate": 0.0003802118825437321,
      "loss": 2.2511,
      "step": 95351
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.9111725091934204,
      "learning_rate": 0.0003802079408986908,
      "loss": 2.1243,
      "step": 95352
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.8792862892150879,
      "learning_rate": 0.0003802039992387373,
      "loss": 2.2515,
      "step": 95353
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.773186981678009,
      "learning_rate": 0.00038020005756387237,
      "loss": 2.3043,
      "step": 95354
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.8880637884140015,
      "learning_rate": 0.00038019611587409674,
      "loss": 2.1604,
      "step": 95355
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.9114627242088318,
      "learning_rate": 0.0003801921741694112,
      "loss": 2.2079,
      "step": 95356
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.8442078232765198,
      "learning_rate": 0.0003801882324498164,
      "loss": 2.1079,
      "step": 95357
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.8810914754867554,
      "learning_rate": 0.00038018429071531303,
      "loss": 2.2764,
      "step": 95358
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.9763261675834656,
      "learning_rate": 0.0003801803489659019,
      "loss": 2.2885,
      "step": 95359
    },
    {
      "epoch": 1.24,
      "grad_norm": 1.040527105331421,
      "learning_rate": 0.00038017640720158384,
      "loss": 2.0775,
      "step": 95360
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.8254236578941345,
      "learning_rate": 0.0003801724654223593,
      "loss": 2.0419,
      "step": 95361
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.8797101974487305,
      "learning_rate": 0.0003801685236282293,
      "loss": 2.1582,
      "step": 95362
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.9445735216140747,
      "learning_rate": 0.00038016458181919444,
      "loss": 2.3023,
      "step": 95363
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.9534039497375488,
      "learning_rate": 0.00038016063999525546,
      "loss": 2.2001,
      "step": 95364
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.7815572023391724,
      "learning_rate": 0.00038015669815641314,
      "loss": 2.187,
      "step": 95365
    },
    {
      "epoch": 1.24,
      "grad_norm": 1.021986484527588,
      "learning_rate": 0.00038015275630266804,
      "loss": 2.212,
      "step": 95366
    },
    {
      "epoch": 1.24,
      "grad_norm": 1.0683469772338867,
      "learning_rate": 0.0003801488144340212,
      "loss": 2.1943,
      "step": 95367
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.8184527158737183,
      "learning_rate": 0.0003801448725504731,
      "loss": 2.293,
      "step": 95368
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.9790398478507996,
      "learning_rate": 0.0003801409306520246,
      "loss": 2.3299,
      "step": 95369
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.9988834261894226,
      "learning_rate": 0.0003801369887386763,
      "loss": 2.008,
      "step": 95370
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.8544870615005493,
      "learning_rate": 0.00038013304681042907,
      "loss": 2.1449,
      "step": 95371
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.7999396920204163,
      "learning_rate": 0.00038012910486728363,
      "loss": 2.0606,
      "step": 95372
    },
    {
      "epoch": 1.24,
      "grad_norm": 1.1931331157684326,
      "learning_rate": 0.00038012516290924066,
      "loss": 2.1028,
      "step": 95373
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.9330223202705383,
      "learning_rate": 0.0003801212209363009,
      "loss": 2.2596,
      "step": 95374
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.7780163884162903,
      "learning_rate": 0.00038011727894846504,
      "loss": 2.3561,
      "step": 95375
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.8876418471336365,
      "learning_rate": 0.000380113336945734,
      "loss": 2.1336,
      "step": 95376
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.8104932904243469,
      "learning_rate": 0.0003801093949281083,
      "loss": 1.9478,
      "step": 95377
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.8237712383270264,
      "learning_rate": 0.00038010545289558865,
      "loss": 2.3044,
      "step": 95378
    },
    {
      "epoch": 1.24,
      "grad_norm": 1.5446323156356812,
      "learning_rate": 0.00038010151084817603,
      "loss": 2.3439,
      "step": 95379
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.9003846049308777,
      "learning_rate": 0.00038009756878587097,
      "loss": 2.1899,
      "step": 95380
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.8153195381164551,
      "learning_rate": 0.00038009362670867424,
      "loss": 2.3548,
      "step": 95381
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.8020586371421814,
      "learning_rate": 0.0003800896846165867,
      "loss": 2.1165,
      "step": 95382
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.86029452085495,
      "learning_rate": 0.0003800857425096088,
      "loss": 2.2664,
      "step": 95383
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.7999352216720581,
      "learning_rate": 0.00038008180038774165,
      "loss": 2.4428,
      "step": 95384
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.8622595071792603,
      "learning_rate": 0.0003800778582509856,
      "loss": 2.3667,
      "step": 95385
    },
    {
      "epoch": 1.24,
      "grad_norm": 1.0033180713653564,
      "learning_rate": 0.00038007391609934165,
      "loss": 2.0547,
      "step": 95386
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.8054071664810181,
      "learning_rate": 0.00038006997393281045,
      "loss": 2.2858,
      "step": 95387
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.8426637649536133,
      "learning_rate": 0.0003800660317513927,
      "loss": 2.2938,
      "step": 95388
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.8416342735290527,
      "learning_rate": 0.0003800620895550892,
      "loss": 2.1933,
      "step": 95389
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.781119704246521,
      "learning_rate": 0.0003800581473439006,
      "loss": 2.2097,
      "step": 95390
    },
    {
      "epoch": 1.24,
      "grad_norm": 1.0272165536880493,
      "learning_rate": 0.0003800542051178277,
      "loss": 2.4442,
      "step": 95391
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.8412665724754333,
      "learning_rate": 0.0003800502628768713,
      "loss": 2.2466,
      "step": 95392
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.9137885570526123,
      "learning_rate": 0.000380046320621032,
      "loss": 2.2092,
      "step": 95393
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.8932220339775085,
      "learning_rate": 0.00038004237835031053,
      "loss": 2.2013,
      "step": 95394
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.8960920572280884,
      "learning_rate": 0.0003800384360647077,
      "loss": 2.1837,
      "step": 95395
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.8448416590690613,
      "learning_rate": 0.00038003449376422413,
      "loss": 1.9984,
      "step": 95396
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.9377579689025879,
      "learning_rate": 0.0003800305514488607,
      "loss": 2.2724,
      "step": 95397
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.844706654548645,
      "learning_rate": 0.0003800266091186181,
      "loss": 2.3362,
      "step": 95398
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.8679140210151672,
      "learning_rate": 0.000380022666773497,
      "loss": 2.0056,
      "step": 95399
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.8761197924613953,
      "learning_rate": 0.0003800187244134982,
      "loss": 2.0848,
      "step": 95400
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.9098755717277527,
      "learning_rate": 0.0003800147820386224,
      "loss": 2.0289,
      "step": 95401
    },
    {
      "epoch": 1.24,
      "grad_norm": 1.194638729095459,
      "learning_rate": 0.00038001083964887047,
      "loss": 2.422,
      "step": 95402
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.8179331421852112,
      "learning_rate": 0.00038000689724424283,
      "loss": 2.0532,
      "step": 95403
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.8696945309638977,
      "learning_rate": 0.0003800029548247405,
      "loss": 2.0988,
      "step": 95404
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.9194008111953735,
      "learning_rate": 0.000379999012390364,
      "loss": 2.075,
      "step": 95405
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.9438509345054626,
      "learning_rate": 0.00037999506994111426,
      "loss": 2.1598,
      "step": 95406
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.8503131866455078,
      "learning_rate": 0.000379991127476992,
      "loss": 2.1052,
      "step": 95407
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.8159940242767334,
      "learning_rate": 0.0003799871849979977,
      "loss": 2.3186,
      "step": 95408
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.889595091342926,
      "learning_rate": 0.0003799832425041324,
      "loss": 2.1481,
      "step": 95409
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.8879415392875671,
      "learning_rate": 0.00037997929999539676,
      "loss": 2.2981,
      "step": 95410
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.8927336931228638,
      "learning_rate": 0.00037997535747179127,
      "loss": 2.2291,
      "step": 95411
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.7942034006118774,
      "learning_rate": 0.000379971414933317,
      "loss": 2.0468,
      "step": 95412
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.846598207950592,
      "learning_rate": 0.00037996747237997454,
      "loss": 2.1635,
      "step": 95413
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.8583148121833801,
      "learning_rate": 0.0003799635298117645,
      "loss": 2.2247,
      "step": 95414
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.8897974491119385,
      "learning_rate": 0.00037995958722868774,
      "loss": 2.2408,
      "step": 95415
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.9597006440162659,
      "learning_rate": 0.0003799556446307451,
      "loss": 2.1645,
      "step": 95416
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.8579723238945007,
      "learning_rate": 0.0003799517020179371,
      "loss": 2.0688,
      "step": 95417
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.9142129421234131,
      "learning_rate": 0.00037994775939026465,
      "loss": 2.2248,
      "step": 95418
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.8761176466941833,
      "learning_rate": 0.00037994381674772836,
      "loss": 2.1729,
      "step": 95419
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.7639676332473755,
      "learning_rate": 0.000379939874090329,
      "loss": 2.2531,
      "step": 95420
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.8128931522369385,
      "learning_rate": 0.00037993593141806727,
      "loss": 1.9584,
      "step": 95421
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.8579932451248169,
      "learning_rate": 0.0003799319887309441,
      "loss": 2.0531,
      "step": 95422
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.8345962762832642,
      "learning_rate": 0.00037992804602895986,
      "loss": 2.21,
      "step": 95423
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.9483782649040222,
      "learning_rate": 0.0003799241033121156,
      "loss": 1.9966,
      "step": 95424
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.8377285003662109,
      "learning_rate": 0.00037992016058041193,
      "loss": 1.9666,
      "step": 95425
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.9009523391723633,
      "learning_rate": 0.00037991621783384954,
      "loss": 2.0262,
      "step": 95426
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.9589229822158813,
      "learning_rate": 0.0003799122750724293,
      "loss": 2.0965,
      "step": 95427
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.841697096824646,
      "learning_rate": 0.0003799083322961518,
      "loss": 2.2081,
      "step": 95428
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.9996150732040405,
      "learning_rate": 0.00037990438950501784,
      "loss": 2.1495,
      "step": 95429
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.8573275208473206,
      "learning_rate": 0.0003799004466990281,
      "loss": 2.3604,
      "step": 95430
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.8622638583183289,
      "learning_rate": 0.0003798965038781835,
      "loss": 2.3265,
      "step": 95431
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.840960681438446,
      "learning_rate": 0.0003798925610424845,
      "loss": 2.214,
      "step": 95432
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.8245953321456909,
      "learning_rate": 0.000379888618191932,
      "loss": 2.1254,
      "step": 95433
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.800652027130127,
      "learning_rate": 0.0003798846753265268,
      "loss": 2.227,
      "step": 95434
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.8496225476264954,
      "learning_rate": 0.0003798807324462694,
      "loss": 2.4466,
      "step": 95435
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.789543867111206,
      "learning_rate": 0.00037987678955116064,
      "loss": 2.2314,
      "step": 95436
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.7922129034996033,
      "learning_rate": 0.0003798728466412014,
      "loss": 2.0747,
      "step": 95437
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.8734498023986816,
      "learning_rate": 0.0003798689037163922,
      "loss": 2.1616,
      "step": 95438
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.880646824836731,
      "learning_rate": 0.0003798649607767339,
      "loss": 2.2735,
      "step": 95439
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.8779601454734802,
      "learning_rate": 0.00037986101782222715,
      "loss": 2.1746,
      "step": 95440
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.8393647074699402,
      "learning_rate": 0.00037985707485287275,
      "loss": 2.0136,
      "step": 95441
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.827281653881073,
      "learning_rate": 0.0003798531318686715,
      "loss": 2.0516,
      "step": 95442
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.9039394855499268,
      "learning_rate": 0.0003798491888696239,
      "loss": 2.4585,
      "step": 95443
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.8067350387573242,
      "learning_rate": 0.000379845245855731,
      "loss": 2.2528,
      "step": 95444
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.9293991327285767,
      "learning_rate": 0.0003798413028269932,
      "loss": 2.1524,
      "step": 95445
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.8082493543624878,
      "learning_rate": 0.0003798373597834114,
      "loss": 2.2153,
      "step": 95446
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.8981379270553589,
      "learning_rate": 0.00037983341672498643,
      "loss": 2.1422,
      "step": 95447
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.7568396329879761,
      "learning_rate": 0.00037982947365171896,
      "loss": 2.1741,
      "step": 95448
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.829302966594696,
      "learning_rate": 0.0003798255305636095,
      "loss": 2.0911,
      "step": 95449
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.885996401309967,
      "learning_rate": 0.0003798215874606591,
      "loss": 2.2962,
      "step": 95450
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.9140482544898987,
      "learning_rate": 0.0003798176443428683,
      "loss": 2.2061,
      "step": 95451
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.8505818843841553,
      "learning_rate": 0.00037981370121023794,
      "loss": 1.9885,
      "step": 95452
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.8378947973251343,
      "learning_rate": 0.0003798097580627687,
      "loss": 2.0728,
      "step": 95453
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.9271906614303589,
      "learning_rate": 0.0003798058149004612,
      "loss": 2.1444,
      "step": 95454
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.8104698657989502,
      "learning_rate": 0.0003798018717233165,
      "loss": 2.1967,
      "step": 95455
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.8336992263793945,
      "learning_rate": 0.000379797928531335,
      "loss": 2.3875,
      "step": 95456
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.8797906637191772,
      "learning_rate": 0.0003797939853245176,
      "loss": 2.2915,
      "step": 95457
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.8829441666603088,
      "learning_rate": 0.00037979004210286494,
      "loss": 2.4043,
      "step": 95458
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.8854854702949524,
      "learning_rate": 0.0003797860988663779,
      "loss": 2.3018,
      "step": 95459
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.9983609914779663,
      "learning_rate": 0.00037978215561505703,
      "loss": 2.196,
      "step": 95460
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.8608375787734985,
      "learning_rate": 0.00037977821234890316,
      "loss": 2.2105,
      "step": 95461
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.8313824534416199,
      "learning_rate": 0.0003797742690679171,
      "loss": 2.1186,
      "step": 95462
    },
    {
      "epoch": 1.24,
      "grad_norm": 1.010077714920044,
      "learning_rate": 0.0003797703257720994,
      "loss": 2.1679,
      "step": 95463
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.8587547540664673,
      "learning_rate": 0.00037976638246145093,
      "loss": 2.0964,
      "step": 95464
    },
    {
      "epoch": 1.24,
      "grad_norm": 1.0429612398147583,
      "learning_rate": 0.0003797624391359724,
      "loss": 2.3055,
      "step": 95465
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.8786342144012451,
      "learning_rate": 0.0003797584957956645,
      "loss": 2.2241,
      "step": 95466
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.8805845975875854,
      "learning_rate": 0.00037975455244052794,
      "loss": 2.074,
      "step": 95467
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.8053204417228699,
      "learning_rate": 0.0003797506090705636,
      "loss": 2.1443,
      "step": 95468
    },
    {
      "epoch": 1.24,
      "grad_norm": 1.0039082765579224,
      "learning_rate": 0.00037974666568577206,
      "loss": 2.0885,
      "step": 95469
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.8263996839523315,
      "learning_rate": 0.0003797427222861541,
      "loss": 2.183,
      "step": 95470
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.8047197461128235,
      "learning_rate": 0.0003797387788717106,
      "loss": 2.1813,
      "step": 95471
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.8763678073883057,
      "learning_rate": 0.00037973483544244196,
      "loss": 2.3159,
      "step": 95472
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.8879523873329163,
      "learning_rate": 0.00037973089199834915,
      "loss": 2.1955,
      "step": 95473
    },
    {
      "epoch": 1.24,
      "grad_norm": 1.143113374710083,
      "learning_rate": 0.00037972694853943297,
      "loss": 2.3338,
      "step": 95474
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.8900927305221558,
      "learning_rate": 0.000379723005065694,
      "loss": 2.3234,
      "step": 95475
    },
    {
      "epoch": 1.24,
      "grad_norm": 1.1729546785354614,
      "learning_rate": 0.000379719061577133,
      "loss": 2.0998,
      "step": 95476
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.9339869022369385,
      "learning_rate": 0.00037971511807375073,
      "loss": 2.227,
      "step": 95477
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.7765811085700989,
      "learning_rate": 0.0003797111745555479,
      "loss": 1.9617,
      "step": 95478
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.8969103693962097,
      "learning_rate": 0.0003797072310225252,
      "loss": 2.0988,
      "step": 95479
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.866978645324707,
      "learning_rate": 0.00037970328747468355,
      "loss": 2.2584,
      "step": 95480
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.8518845438957214,
      "learning_rate": 0.0003796993439120234,
      "loss": 2.0344,
      "step": 95481
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.84596848487854,
      "learning_rate": 0.0003796954003345457,
      "loss": 2.2327,
      "step": 95482
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.8135333061218262,
      "learning_rate": 0.00037969145674225126,
      "loss": 1.9667,
      "step": 95483
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.9863857626914978,
      "learning_rate": 0.0003796875131351405,
      "loss": 2.1867,
      "step": 95484
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.8402283191680908,
      "learning_rate": 0.0003796835695132144,
      "loss": 2.2895,
      "step": 95485
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.7789064049720764,
      "learning_rate": 0.00037967962587647365,
      "loss": 1.9275,
      "step": 95486
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.8776150941848755,
      "learning_rate": 0.0003796756822249189,
      "loss": 2.1846,
      "step": 95487
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.8431465029716492,
      "learning_rate": 0.0003796717385585509,
      "loss": 2.2251,
      "step": 95488
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.8908127546310425,
      "learning_rate": 0.0003796677948773705,
      "loss": 2.5248,
      "step": 95489
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.9238951206207275,
      "learning_rate": 0.00037966385118137834,
      "loss": 2.3152,
      "step": 95490
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.870979905128479,
      "learning_rate": 0.00037965990747057505,
      "loss": 2.0443,
      "step": 95491
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.7979370355606079,
      "learning_rate": 0.00037965596374496167,
      "loss": 2.0456,
      "step": 95492
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.8747810125350952,
      "learning_rate": 0.0003796520200045386,
      "loss": 2.0356,
      "step": 95493
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.9083532691001892,
      "learning_rate": 0.0003796480762493068,
      "loss": 2.0594,
      "step": 95494
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.9576937556266785,
      "learning_rate": 0.0003796441324792669,
      "loss": 2.349,
      "step": 95495
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.8547353744506836,
      "learning_rate": 0.0003796401886944196,
      "loss": 2.168,
      "step": 95496
    },
    {
      "epoch": 1.24,
      "grad_norm": 1.0244550704956055,
      "learning_rate": 0.0003796362448947657,
      "loss": 2.1227,
      "step": 95497
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.8444005250930786,
      "learning_rate": 0.000379632301080306,
      "loss": 2.1646,
      "step": 95498
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.9337426424026489,
      "learning_rate": 0.00037962835725104105,
      "loss": 2.1898,
      "step": 95499
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.7988766431808472,
      "learning_rate": 0.00037962441340697175,
      "loss": 1.956,
      "step": 95500
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.9288736581802368,
      "learning_rate": 0.0003796204695480988,
      "loss": 2.0798,
      "step": 95501
    },
    {
      "epoch": 1.24,
      "grad_norm": 1.0046803951263428,
      "learning_rate": 0.0003796165256744228,
      "loss": 2.1083,
      "step": 95502
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.901682436466217,
      "learning_rate": 0.0003796125817859446,
      "loss": 2.2242,
      "step": 95503
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.9363892674446106,
      "learning_rate": 0.000379608637882665,
      "loss": 2.0927,
      "step": 95504
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.8506883978843689,
      "learning_rate": 0.0003796046939645846,
      "loss": 2.1765,
      "step": 95505
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.8048860430717468,
      "learning_rate": 0.00037960075003170413,
      "loss": 2.0031,
      "step": 95506
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.9582500457763672,
      "learning_rate": 0.0003795968060840245,
      "loss": 2.2202,
      "step": 95507
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.8855029344558716,
      "learning_rate": 0.00037959286212154624,
      "loss": 2.3259,
      "step": 95508
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.8389607071876526,
      "learning_rate": 0.0003795889181442701,
      "loss": 2.267,
      "step": 95509
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.8095302581787109,
      "learning_rate": 0.0003795849741521971,
      "loss": 2.2845,
      "step": 95510
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.8275256156921387,
      "learning_rate": 0.00037958103014532757,
      "loss": 2.1085,
      "step": 95511
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.9549157023429871,
      "learning_rate": 0.00037957708612366243,
      "loss": 1.9199,
      "step": 95512
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.8730834126472473,
      "learning_rate": 0.0003795731420872025,
      "loss": 2.3146,
      "step": 95513
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.849854588508606,
      "learning_rate": 0.0003795691980359484,
      "loss": 1.956,
      "step": 95514
    },
    {
      "epoch": 1.24,
      "grad_norm": 1.0044260025024414,
      "learning_rate": 0.0003795652539699007,
      "loss": 2.4017,
      "step": 95515
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.9144585132598877,
      "learning_rate": 0.00037956130988906053,
      "loss": 2.0327,
      "step": 95516
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.9047535061836243,
      "learning_rate": 0.0003795573657934283,
      "loss": 2.2231,
      "step": 95517
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.8127443790435791,
      "learning_rate": 0.00037955342168300495,
      "loss": 2.2854,
      "step": 95518
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.8520740270614624,
      "learning_rate": 0.000379549477557791,
      "loss": 2.294,
      "step": 95519
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.995732843875885,
      "learning_rate": 0.00037954553341778744,
      "loss": 2.2853,
      "step": 95520
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.8166241645812988,
      "learning_rate": 0.0003795415892629948,
      "loss": 1.956,
      "step": 95521
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.8098912835121155,
      "learning_rate": 0.00037953764509341385,
      "loss": 2.0026,
      "step": 95522
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.8988529443740845,
      "learning_rate": 0.0003795337009090454,
      "loss": 2.4467,
      "step": 95523
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.9521389603614807,
      "learning_rate": 0.0003795297567098901,
      "loss": 2.28,
      "step": 95524
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.905949592590332,
      "learning_rate": 0.0003795258124959487,
      "loss": 2.1991,
      "step": 95525
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.8431792855262756,
      "learning_rate": 0.00037952186826722197,
      "loss": 2.1902,
      "step": 95526
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.8726080656051636,
      "learning_rate": 0.0003795179240237106,
      "loss": 2.1593,
      "step": 95527
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.8652877807617188,
      "learning_rate": 0.00037951397976541533,
      "loss": 2.098,
      "step": 95528
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.8417388796806335,
      "learning_rate": 0.000379510035492337,
      "loss": 2.3426,
      "step": 95529
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.8270983695983887,
      "learning_rate": 0.0003795060912044762,
      "loss": 2.2784,
      "step": 95530
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.9530045986175537,
      "learning_rate": 0.00037950214690183373,
      "loss": 2.0138,
      "step": 95531
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.8611847758293152,
      "learning_rate": 0.0003794982025844102,
      "loss": 2.0903,
      "step": 95532
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.9748663902282715,
      "learning_rate": 0.00037949425825220664,
      "loss": 2.1413,
      "step": 95533
    },
    {
      "epoch": 1.24,
      "grad_norm": 1.0019711256027222,
      "learning_rate": 0.0003794903139052235,
      "loss": 2.1613,
      "step": 95534
    },
    {
      "epoch": 1.24,
      "grad_norm": 1.0137879848480225,
      "learning_rate": 0.0003794863695434616,
      "loss": 2.1866,
      "step": 95535
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.7901710271835327,
      "learning_rate": 0.00037948242516692174,
      "loss": 1.9909,
      "step": 95536
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.9547908306121826,
      "learning_rate": 0.00037947848077560454,
      "loss": 2.1806,
      "step": 95537
    },
    {
      "epoch": 1.24,
      "grad_norm": 1.0190552473068237,
      "learning_rate": 0.0003794745363695108,
      "loss": 2.2065,
      "step": 95538
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.8811897039413452,
      "learning_rate": 0.0003794705919486412,
      "loss": 2.2669,
      "step": 95539
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.8776928186416626,
      "learning_rate": 0.0003794666475129967,
      "loss": 2.1825,
      "step": 95540
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.8934705257415771,
      "learning_rate": 0.0003794627030625776,
      "loss": 2.353,
      "step": 95541
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.9224043488502502,
      "learning_rate": 0.00037945875859738505,
      "loss": 2.2471,
      "step": 95542
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.8253328800201416,
      "learning_rate": 0.00037945481411741957,
      "loss": 2.1183,
      "step": 95543
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.9898739457130432,
      "learning_rate": 0.00037945086962268195,
      "loss": 2.1389,
      "step": 95544
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.9799591302871704,
      "learning_rate": 0.00037944692511317294,
      "loss": 2.2749,
      "step": 95545
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.8812167644500732,
      "learning_rate": 0.0003794429805888932,
      "loss": 2.1813,
      "step": 95546
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.7586795687675476,
      "learning_rate": 0.00037943903604984353,
      "loss": 1.9452,
      "step": 95547
    },
    {
      "epoch": 1.24,
      "grad_norm": 1.0542868375778198,
      "learning_rate": 0.0003794350914960246,
      "loss": 2.1256,
      "step": 95548
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.9577633738517761,
      "learning_rate": 0.00037943114692743724,
      "loss": 2.1053,
      "step": 95549
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.8473849296569824,
      "learning_rate": 0.0003794272023440821,
      "loss": 2.1395,
      "step": 95550
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.8016011118888855,
      "learning_rate": 0.0003794232577459599,
      "loss": 2.0402,
      "step": 95551
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.8393533825874329,
      "learning_rate": 0.0003794193131330715,
      "loss": 2.3279,
      "step": 95552
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.8196122050285339,
      "learning_rate": 0.00037941536850541756,
      "loss": 2.2287,
      "step": 95553
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.8688640594482422,
      "learning_rate": 0.0003794114238629988,
      "loss": 1.979,
      "step": 95554
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.773802638053894,
      "learning_rate": 0.0003794074792058159,
      "loss": 1.9726,
      "step": 95555
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.8552516102790833,
      "learning_rate": 0.00037940353453386975,
      "loss": 2.4527,
      "step": 95556
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.9857103824615479,
      "learning_rate": 0.00037939958984716085,
      "loss": 2.1688,
      "step": 95557
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.820406436920166,
      "learning_rate": 0.0003793956451456901,
      "loss": 2.149,
      "step": 95558
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.8916566967964172,
      "learning_rate": 0.0003793917004294583,
      "loss": 2.1845,
      "step": 95559
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.9357226490974426,
      "learning_rate": 0.000379387755698466,
      "loss": 2.1132,
      "step": 95560
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.8677136301994324,
      "learning_rate": 0.00037938381095271395,
      "loss": 2.0636,
      "step": 95561
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.8790416717529297,
      "learning_rate": 0.0003793798661922031,
      "loss": 2.1855,
      "step": 95562
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.8900894522666931,
      "learning_rate": 0.00037937592141693396,
      "loss": 2.2103,
      "step": 95563
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.9096850156784058,
      "learning_rate": 0.0003793719766269073,
      "loss": 2.2054,
      "step": 95564
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.8756580352783203,
      "learning_rate": 0.00037936803182212403,
      "loss": 2.0622,
      "step": 95565
    },
    {
      "epoch": 1.24,
      "grad_norm": 2.126434564590454,
      "learning_rate": 0.0003793640870025846,
      "loss": 2.3555,
      "step": 95566
    },
    {
      "epoch": 1.24,
      "grad_norm": 1.3761241436004639,
      "learning_rate": 0.00037936014216828997,
      "loss": 2.1415,
      "step": 95567
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.8652989268302917,
      "learning_rate": 0.0003793561973192407,
      "loss": 2.0268,
      "step": 95568
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.8753793239593506,
      "learning_rate": 0.0003793522524554377,
      "loss": 2.2062,
      "step": 95569
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.8993375301361084,
      "learning_rate": 0.00037934830757688155,
      "loss": 1.9859,
      "step": 95570
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.9905130863189697,
      "learning_rate": 0.00037934436268357316,
      "loss": 2.3015,
      "step": 95571
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.7542805075645447,
      "learning_rate": 0.00037934041777551306,
      "loss": 2.36,
      "step": 95572
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.82927006483078,
      "learning_rate": 0.0003793364728527021,
      "loss": 2.2367,
      "step": 95573
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.9420542120933533,
      "learning_rate": 0.00037933252791514104,
      "loss": 1.9365,
      "step": 95574
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.8621981143951416,
      "learning_rate": 0.0003793285829628305,
      "loss": 2.0738,
      "step": 95575
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.9552692770957947,
      "learning_rate": 0.00037932463799577134,
      "loss": 2.0563,
      "step": 95576
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.8496197462081909,
      "learning_rate": 0.00037932069301396424,
      "loss": 2.0408,
      "step": 95577
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.8291279673576355,
      "learning_rate": 0.0003793167480174099,
      "loss": 2.2788,
      "step": 95578
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.7875543832778931,
      "learning_rate": 0.00037931280300610896,
      "loss": 2.2016,
      "step": 95579
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.8985128998756409,
      "learning_rate": 0.0003793088579800624,
      "loss": 2.2891,
      "step": 95580
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.8436250686645508,
      "learning_rate": 0.00037930491293927085,
      "loss": 2.1192,
      "step": 95581
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.8045081496238708,
      "learning_rate": 0.0003793009678837349,
      "loss": 2.1753,
      "step": 95582
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.8602752685546875,
      "learning_rate": 0.0003792970228134555,
      "loss": 1.9337,
      "step": 95583
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.9142042994499207,
      "learning_rate": 0.0003792930777284333,
      "loss": 2.1786,
      "step": 95584
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.9177417159080505,
      "learning_rate": 0.00037928913262866895,
      "loss": 2.373,
      "step": 95585
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.8731919527053833,
      "learning_rate": 0.00037928518751416325,
      "loss": 2.2818,
      "step": 95586
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.934855043888092,
      "learning_rate": 0.000379281242384917,
      "loss": 2.1878,
      "step": 95587
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.7618799805641174,
      "learning_rate": 0.0003792772972409309,
      "loss": 2.3216,
      "step": 95588
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.8410208225250244,
      "learning_rate": 0.0003792733520822055,
      "loss": 2.0888,
      "step": 95589
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.8284435868263245,
      "learning_rate": 0.00037926940690874184,
      "loss": 1.8826,
      "step": 95590
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.9211506247520447,
      "learning_rate": 0.0003792654617205405,
      "loss": 2.2677,
      "step": 95591
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.8245164752006531,
      "learning_rate": 0.00037926151651760206,
      "loss": 2.3667,
      "step": 95592
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.9432772994041443,
      "learning_rate": 0.0003792575712999276,
      "loss": 2.1254,
      "step": 95593
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.8544232249259949,
      "learning_rate": 0.00037925362606751757,
      "loss": 2.0723,
      "step": 95594
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.8109775185585022,
      "learning_rate": 0.0003792496808203727,
      "loss": 2.1349,
      "step": 95595
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.8028135895729065,
      "learning_rate": 0.000379245735558494,
      "loss": 2.2513,
      "step": 95596
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.8831213116645813,
      "learning_rate": 0.0003792417902818819,
      "loss": 2.1286,
      "step": 95597
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.8338068723678589,
      "learning_rate": 0.00037923784499053727,
      "loss": 2.2088,
      "step": 95598
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.7967861294746399,
      "learning_rate": 0.0003792338996844609,
      "loss": 2.0644,
      "step": 95599
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.9113783836364746,
      "learning_rate": 0.0003792299543636534,
      "loss": 2.3279,
      "step": 95600
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.8628491163253784,
      "learning_rate": 0.0003792260090281155,
      "loss": 2.1245,
      "step": 95601
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.8239733576774597,
      "learning_rate": 0.0003792220636778481,
      "loss": 2.1784,
      "step": 95602
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.8295032978057861,
      "learning_rate": 0.0003792181183128518,
      "loss": 2.3224,
      "step": 95603
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.8617444634437561,
      "learning_rate": 0.0003792141729331273,
      "loss": 2.1418,
      "step": 95604
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.8487313389778137,
      "learning_rate": 0.00037921022753867545,
      "loss": 2.0109,
      "step": 95605
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.9972054958343506,
      "learning_rate": 0.00037920628212949684,
      "loss": 2.0684,
      "step": 95606
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.799615204334259,
      "learning_rate": 0.0003792023367055923,
      "loss": 2.3246,
      "step": 95607
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.8908523917198181,
      "learning_rate": 0.00037919839126696265,
      "loss": 2.2302,
      "step": 95608
    },
    {
      "epoch": 1.24,
      "grad_norm": 1.0657716989517212,
      "learning_rate": 0.00037919444581360843,
      "loss": 1.9823,
      "step": 95609
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.9004680514335632,
      "learning_rate": 0.00037919050034553053,
      "loss": 2.1039,
      "step": 95610
    },
    {
      "epoch": 1.24,
      "grad_norm": 1.035249948501587,
      "learning_rate": 0.0003791865548627296,
      "loss": 2.3158,
      "step": 95611
    },
    {
      "epoch": 1.24,
      "grad_norm": 1.14060640335083,
      "learning_rate": 0.00037918260936520634,
      "loss": 2.2911,
      "step": 95612
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.9513477087020874,
      "learning_rate": 0.00037917866385296157,
      "loss": 2.0791,
      "step": 95613
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.811996579170227,
      "learning_rate": 0.0003791747183259961,
      "loss": 2.3451,
      "step": 95614
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.7835733294487,
      "learning_rate": 0.0003791707727843104,
      "loss": 2.0785,
      "step": 95615
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.9611560702323914,
      "learning_rate": 0.0003791668272279054,
      "loss": 2.0531,
      "step": 95616
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.8565278649330139,
      "learning_rate": 0.0003791628816567819,
      "loss": 1.9314,
      "step": 95617
    },
    {
      "epoch": 1.25,
      "grad_norm": 1.0319273471832275,
      "learning_rate": 0.00037915893607094037,
      "loss": 2.3731,
      "step": 95618
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.7313459515571594,
      "learning_rate": 0.0003791549904703818,
      "loss": 2.0796,
      "step": 95619
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.9043715596199036,
      "learning_rate": 0.00037915104485510676,
      "loss": 2.2043,
      "step": 95620
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.9347652196884155,
      "learning_rate": 0.00037914709922511606,
      "loss": 2.4788,
      "step": 95621
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.7591302394866943,
      "learning_rate": 0.00037914315358041046,
      "loss": 2.304,
      "step": 95622
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.8037439584732056,
      "learning_rate": 0.0003791392079209906,
      "loss": 2.1369,
      "step": 95623
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.8315821886062622,
      "learning_rate": 0.0003791352622468573,
      "loss": 2.1237,
      "step": 95624
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.9738702178001404,
      "learning_rate": 0.00037913131655801125,
      "loss": 2.1366,
      "step": 95625
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.8729681968688965,
      "learning_rate": 0.00037912737085445326,
      "loss": 2.188,
      "step": 95626
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.7213784456253052,
      "learning_rate": 0.00037912342513618383,
      "loss": 2.1584,
      "step": 95627
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.877232551574707,
      "learning_rate": 0.00037911947940320406,
      "loss": 2.2307,
      "step": 95628
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.9570826888084412,
      "learning_rate": 0.0003791155336555144,
      "loss": 2.3383,
      "step": 95629
    },
    {
      "epoch": 1.25,
      "grad_norm": 1.0260186195373535,
      "learning_rate": 0.00037911158789311565,
      "loss": 2.0159,
      "step": 95630
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.7445664405822754,
      "learning_rate": 0.00037910764211600853,
      "loss": 2.2534,
      "step": 95631
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.8659618496894836,
      "learning_rate": 0.0003791036963241939,
      "loss": 2.4196,
      "step": 95632
    },
    {
      "epoch": 1.25,
      "grad_norm": 1.2212755680084229,
      "learning_rate": 0.0003790997505176723,
      "loss": 2.2179,
      "step": 95633
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.8860995769500732,
      "learning_rate": 0.00037909580469644465,
      "loss": 2.0684,
      "step": 95634
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.9018166661262512,
      "learning_rate": 0.00037909185886051165,
      "loss": 2.1676,
      "step": 95635
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.8637462854385376,
      "learning_rate": 0.0003790879130098738,
      "loss": 2.093,
      "step": 95636
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.9089314341545105,
      "learning_rate": 0.0003790839671445321,
      "loss": 2.1687,
      "step": 95637
    },
    {
      "epoch": 1.25,
      "grad_norm": 1.0056488513946533,
      "learning_rate": 0.0003790800212644873,
      "loss": 2.2875,
      "step": 95638
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.7618979811668396,
      "learning_rate": 0.00037907607536973986,
      "loss": 1.9168,
      "step": 95639
    },
    {
      "epoch": 1.25,
      "grad_norm": 1.069666862487793,
      "learning_rate": 0.00037907212946029073,
      "loss": 2.2669,
      "step": 95640
    },
    {
      "epoch": 1.25,
      "grad_norm": 1.234158992767334,
      "learning_rate": 0.00037906818353614075,
      "loss": 2.3046,
      "step": 95641
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.8198824524879456,
      "learning_rate": 0.00037906423759729034,
      "loss": 2.0551,
      "step": 95642
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.9574794769287109,
      "learning_rate": 0.00037906029164374043,
      "loss": 2.2384,
      "step": 95643
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.8474948406219482,
      "learning_rate": 0.0003790563456754917,
      "loss": 1.9588,
      "step": 95644
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.8980891108512878,
      "learning_rate": 0.000379052399692545,
      "loss": 2.19,
      "step": 95645
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.8634989857673645,
      "learning_rate": 0.0003790484536949008,
      "loss": 2.2279,
      "step": 95646
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.926566481590271,
      "learning_rate": 0.0003790445076825602,
      "loss": 2.1289,
      "step": 95647
    },
    {
      "epoch": 1.25,
      "grad_norm": 1.400372862815857,
      "learning_rate": 0.0003790405616555236,
      "loss": 2.4765,
      "step": 95648
    },
    {
      "epoch": 1.25,
      "grad_norm": 1.0557550191879272,
      "learning_rate": 0.0003790366156137919,
      "loss": 2.0288,
      "step": 95649
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.9196557998657227,
      "learning_rate": 0.00037903266955736584,
      "loss": 2.0624,
      "step": 95650
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.8210316896438599,
      "learning_rate": 0.00037902872348624607,
      "loss": 2.2258,
      "step": 95651
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.8738781213760376,
      "learning_rate": 0.0003790247774004333,
      "loss": 2.0266,
      "step": 95652
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.9451029300689697,
      "learning_rate": 0.0003790208312999284,
      "loss": 2.4596,
      "step": 95653
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.9238409996032715,
      "learning_rate": 0.00037901688518473213,
      "loss": 2.0846,
      "step": 95654
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.985809862613678,
      "learning_rate": 0.00037901293905484503,
      "loss": 2.1972,
      "step": 95655
    },
    {
      "epoch": 1.25,
      "grad_norm": 1.6105536222457886,
      "learning_rate": 0.0003790089929102679,
      "loss": 2.4693,
      "step": 95656
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.8785350918769836,
      "learning_rate": 0.00037900504675100164,
      "loss": 2.1541,
      "step": 95657
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.8372781276702881,
      "learning_rate": 0.0003790011005770467,
      "loss": 2.0119,
      "step": 95658
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.9221480488777161,
      "learning_rate": 0.00037899715438840403,
      "loss": 2.2791,
      "step": 95659
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.9251615405082703,
      "learning_rate": 0.0003789932081850743,
      "loss": 2.2537,
      "step": 95660
    },
    {
      "epoch": 1.25,
      "grad_norm": 1.0019065141677856,
      "learning_rate": 0.0003789892619670582,
      "loss": 2.13,
      "step": 95661
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.9463825225830078,
      "learning_rate": 0.00037898531573435655,
      "loss": 2.2621,
      "step": 95662
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.8804594278335571,
      "learning_rate": 0.00037898136948697007,
      "loss": 2.0885,
      "step": 95663
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.9146782755851746,
      "learning_rate": 0.0003789774232248994,
      "loss": 2.1021,
      "step": 95664
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.8270623683929443,
      "learning_rate": 0.00037897347694814537,
      "loss": 2.0405,
      "step": 95665
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.8212814927101135,
      "learning_rate": 0.0003789695306567087,
      "loss": 2.0726,
      "step": 95666
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.8025502562522888,
      "learning_rate": 0.0003789655843505901,
      "loss": 2.2782,
      "step": 95667
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.7968557476997375,
      "learning_rate": 0.0003789616380297903,
      "loss": 1.9845,
      "step": 95668
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.9616790413856506,
      "learning_rate": 0.00037895769169431,
      "loss": 2.1043,
      "step": 95669
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.9183815717697144,
      "learning_rate": 0.00037895374534414997,
      "loss": 2.1898,
      "step": 95670
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.90438312292099,
      "learning_rate": 0.000378949798979311,
      "loss": 2.2015,
      "step": 95671
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.8738244771957397,
      "learning_rate": 0.0003789458525997937,
      "loss": 2.1131,
      "step": 95672
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.8461360931396484,
      "learning_rate": 0.00037894190620559894,
      "loss": 2.29,
      "step": 95673
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.858600914478302,
      "learning_rate": 0.0003789379597967273,
      "loss": 2.1666,
      "step": 95674
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.8367795944213867,
      "learning_rate": 0.0003789340133731798,
      "loss": 2.3507,
      "step": 95675
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.9476701617240906,
      "learning_rate": 0.0003789300669349568,
      "loss": 1.9793,
      "step": 95676
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.8268342018127441,
      "learning_rate": 0.00037892612048205925,
      "loss": 2.1497,
      "step": 95677
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.8232461214065552,
      "learning_rate": 0.0003789221740144879,
      "loss": 2.1759,
      "step": 95678
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.9070119261741638,
      "learning_rate": 0.0003789182275322434,
      "loss": 1.8554,
      "step": 95679
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.8218206167221069,
      "learning_rate": 0.00037891428103532646,
      "loss": 2.0822,
      "step": 95680
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.8730512261390686,
      "learning_rate": 0.0003789103345237379,
      "loss": 2.2233,
      "step": 95681
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.9830804467201233,
      "learning_rate": 0.0003789063879974785,
      "loss": 2.1779,
      "step": 95682
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.8258435726165771,
      "learning_rate": 0.00037890244145654875,
      "loss": 1.9548,
      "step": 95683
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.8515554666519165,
      "learning_rate": 0.00037889849490094967,
      "loss": 2.047,
      "step": 95684
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.9025765061378479,
      "learning_rate": 0.0003788945483306818,
      "loss": 2.0816,
      "step": 95685
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.9002768397331238,
      "learning_rate": 0.00037889060174574595,
      "loss": 2.0805,
      "step": 95686
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.793231725692749,
      "learning_rate": 0.0003788866551461429,
      "loss": 2.1732,
      "step": 95687
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.8497949242591858,
      "learning_rate": 0.00037888270853187335,
      "loss": 2.1668,
      "step": 95688
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.9551472663879395,
      "learning_rate": 0.00037887876190293796,
      "loss": 2.2529,
      "step": 95689
    },
    {
      "epoch": 1.25,
      "grad_norm": 1.039750337600708,
      "learning_rate": 0.00037887481525933754,
      "loss": 2.2756,
      "step": 95690
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.9878454208374023,
      "learning_rate": 0.0003788708686010728,
      "loss": 2.1092,
      "step": 95691
    },
    {
      "epoch": 1.25,
      "grad_norm": 1.115108847618103,
      "learning_rate": 0.00037886692192814445,
      "loss": 2.2022,
      "step": 95692
    },
    {
      "epoch": 1.25,
      "grad_norm": 1.114752173423767,
      "learning_rate": 0.00037886297524055334,
      "loss": 2.1588,
      "step": 95693
    },
    {
      "epoch": 1.25,
      "grad_norm": 1.023127555847168,
      "learning_rate": 0.0003788590285383,
      "loss": 2.2251,
      "step": 95694
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.9276705384254456,
      "learning_rate": 0.0003788550818213853,
      "loss": 2.1988,
      "step": 95695
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.9029868245124817,
      "learning_rate": 0.00037885113508981003,
      "loss": 2.1003,
      "step": 95696
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.8452470898628235,
      "learning_rate": 0.0003788471883435747,
      "loss": 2.316,
      "step": 95697
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.9065276980400085,
      "learning_rate": 0.00037884324158268037,
      "loss": 2.2512,
      "step": 95698
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.8677434921264648,
      "learning_rate": 0.0003788392948071275,
      "loss": 2.317,
      "step": 95699
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.924561083316803,
      "learning_rate": 0.00037883534801691693,
      "loss": 2.1964,
      "step": 95700
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.9018638730049133,
      "learning_rate": 0.00037883140121204936,
      "loss": 2.2839,
      "step": 95701
    },
    {
      "epoch": 1.25,
      "grad_norm": 1.0150961875915527,
      "learning_rate": 0.00037882745439252554,
      "loss": 2.1319,
      "step": 95702
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.9014958143234253,
      "learning_rate": 0.00037882350755834624,
      "loss": 2.0933,
      "step": 95703
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.8631067276000977,
      "learning_rate": 0.0003788195607095121,
      "loss": 2.3053,
      "step": 95704
    },
    {
      "epoch": 1.25,
      "grad_norm": 1.091796875,
      "learning_rate": 0.000378815613846024,
      "loss": 2.042,
      "step": 95705
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.9513736367225647,
      "learning_rate": 0.00037881166696788256,
      "loss": 2.0184,
      "step": 95706
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.8090340495109558,
      "learning_rate": 0.00037880772007508847,
      "loss": 2.0889,
      "step": 95707
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.9550212025642395,
      "learning_rate": 0.00037880377316764263,
      "loss": 2.5725,
      "step": 95708
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.9138612151145935,
      "learning_rate": 0.00037879982624554565,
      "loss": 2.3691,
      "step": 95709
    },
    {
      "epoch": 1.25,
      "grad_norm": 1.0112241506576538,
      "learning_rate": 0.00037879587930879827,
      "loss": 2.3117,
      "step": 95710
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.8270821571350098,
      "learning_rate": 0.0003787919323574013,
      "loss": 2.2103,
      "step": 95711
    },
    {
      "epoch": 1.25,
      "grad_norm": 1.046679139137268,
      "learning_rate": 0.0003787879853913554,
      "loss": 2.3507,
      "step": 95712
    },
    {
      "epoch": 1.25,
      "grad_norm": 1.008791446685791,
      "learning_rate": 0.00037878403841066127,
      "loss": 2.349,
      "step": 95713
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.977023184299469,
      "learning_rate": 0.0003787800914153198,
      "loss": 2.1695,
      "step": 95714
    },
    {
      "epoch": 1.25,
      "grad_norm": 1.0468822717666626,
      "learning_rate": 0.00037877614440533154,
      "loss": 2.1883,
      "step": 95715
    },
    {
      "epoch": 1.25,
      "grad_norm": 1.1943997144699097,
      "learning_rate": 0.00037877219738069744,
      "loss": 2.1181,
      "step": 95716
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.9820142388343811,
      "learning_rate": 0.00037876825034141793,
      "loss": 2.0737,
      "step": 95717
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.8166996240615845,
      "learning_rate": 0.00037876430328749404,
      "loss": 1.9838,
      "step": 95718
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.8440111875534058,
      "learning_rate": 0.00037876035621892633,
      "loss": 2.1864,
      "step": 95719
    },
    {
      "epoch": 1.25,
      "grad_norm": 1.036590337753296,
      "learning_rate": 0.00037875640913571554,
      "loss": 2.1646,
      "step": 95720
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.8665857315063477,
      "learning_rate": 0.00037875246203786255,
      "loss": 2.2514,
      "step": 95721
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.7690652012825012,
      "learning_rate": 0.0003787485149253679,
      "loss": 1.998,
      "step": 95722
    },
    {
      "epoch": 1.25,
      "grad_norm": 1.019579529762268,
      "learning_rate": 0.0003787445677982324,
      "loss": 2.2912,
      "step": 95723
    },
    {
      "epoch": 1.25,
      "grad_norm": 1.0008078813552856,
      "learning_rate": 0.00037874062065645686,
      "loss": 2.1192,
      "step": 95724
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.8250686526298523,
      "learning_rate": 0.00037873667350004194,
      "loss": 1.9943,
      "step": 95725
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.882187008857727,
      "learning_rate": 0.0003787327263289884,
      "loss": 2.3298,
      "step": 95726
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.9672505855560303,
      "learning_rate": 0.00037872877914329697,
      "loss": 2.0518,
      "step": 95727
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.8250595331192017,
      "learning_rate": 0.0003787248319429683,
      "loss": 2.245,
      "step": 95728
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.8842807412147522,
      "learning_rate": 0.00037872088472800324,
      "loss": 2.1723,
      "step": 95729
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.888391375541687,
      "learning_rate": 0.00037871693749840256,
      "loss": 1.9313,
      "step": 95730
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.8499546051025391,
      "learning_rate": 0.0003787129902541668,
      "loss": 2.0154,
      "step": 95731
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.7285254001617432,
      "learning_rate": 0.0003787090429952968,
      "loss": 1.9776,
      "step": 95732
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.9170553088188171,
      "learning_rate": 0.0003787050957217934,
      "loss": 2.0883,
      "step": 95733
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.915176510810852,
      "learning_rate": 0.0003787011484336572,
      "loss": 2.2997,
      "step": 95734
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.9771654605865479,
      "learning_rate": 0.00037869720113088896,
      "loss": 2.2968,
      "step": 95735
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.9053380489349365,
      "learning_rate": 0.0003786932538134895,
      "loss": 2.0374,
      "step": 95736
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.9060936570167542,
      "learning_rate": 0.0003786893064814594,
      "loss": 2.0787,
      "step": 95737
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.8834011554718018,
      "learning_rate": 0.00037868535913479945,
      "loss": 2.1954,
      "step": 95738
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.8377317786216736,
      "learning_rate": 0.00037868141177351043,
      "loss": 2.2444,
      "step": 95739
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.8927419185638428,
      "learning_rate": 0.00037867746439759303,
      "loss": 2.2614,
      "step": 95740
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.8855484127998352,
      "learning_rate": 0.000378673517007048,
      "loss": 2.3421,
      "step": 95741
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.8338202834129333,
      "learning_rate": 0.0003786695696018762,
      "loss": 1.8877,
      "step": 95742
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.8927385210990906,
      "learning_rate": 0.00037866562218207814,
      "loss": 2.101,
      "step": 95743
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.8355101943016052,
      "learning_rate": 0.00037866167474765464,
      "loss": 2.208,
      "step": 95744
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.8700199723243713,
      "learning_rate": 0.0003786577272986065,
      "loss": 2.1899,
      "step": 95745
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.8739989399909973,
      "learning_rate": 0.0003786537798349343,
      "loss": 2.0502,
      "step": 95746
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.8456985354423523,
      "learning_rate": 0.00037864983235663896,
      "loss": 2.3725,
      "step": 95747
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.820850133895874,
      "learning_rate": 0.00037864588486372124,
      "loss": 2.0975,
      "step": 95748
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.8072528839111328,
      "learning_rate": 0.0003786419373561816,
      "loss": 2.0618,
      "step": 95749
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.7814874053001404,
      "learning_rate": 0.000378637989834021,
      "loss": 1.9916,
      "step": 95750
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.8734728097915649,
      "learning_rate": 0.0003786340422972402,
      "loss": 2.2185,
      "step": 95751
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.8402875065803528,
      "learning_rate": 0.00037863009474583967,
      "loss": 2.0719,
      "step": 95752
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.8788182735443115,
      "learning_rate": 0.00037862614717982043,
      "loss": 2.1397,
      "step": 95753
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.8737267255783081,
      "learning_rate": 0.0003786221995991831,
      "loss": 2.1455,
      "step": 95754
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.9106146693229675,
      "learning_rate": 0.00037861825200392836,
      "loss": 2.1452,
      "step": 95755
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.7987862825393677,
      "learning_rate": 0.00037861430439405704,
      "loss": 2.1301,
      "step": 95756
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.8569633364677429,
      "learning_rate": 0.00037861035676956987,
      "loss": 2.1088,
      "step": 95757
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.848656177520752,
      "learning_rate": 0.0003786064091304675,
      "loss": 2.235,
      "step": 95758
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.8230098485946655,
      "learning_rate": 0.0003786024614767507,
      "loss": 2.253,
      "step": 95759
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.7824263572692871,
      "learning_rate": 0.0003785985138084203,
      "loss": 2.4402,
      "step": 95760
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.9614007472991943,
      "learning_rate": 0.0003785945661254769,
      "loss": 2.0731,
      "step": 95761
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.8823630213737488,
      "learning_rate": 0.0003785906184279212,
      "loss": 2.1507,
      "step": 95762
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.8792799115180969,
      "learning_rate": 0.00037858667071575424,
      "loss": 2.3932,
      "step": 95763
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.8081644773483276,
      "learning_rate": 0.0003785827229889764,
      "loss": 2.1593,
      "step": 95764
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.9961488842964172,
      "learning_rate": 0.0003785787752475885,
      "loss": 2.2047,
      "step": 95765
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.8191153407096863,
      "learning_rate": 0.00037857482749159143,
      "loss": 2.1016,
      "step": 95766
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.8370306491851807,
      "learning_rate": 0.0003785708797209858,
      "loss": 2.1407,
      "step": 95767
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.8520517349243164,
      "learning_rate": 0.00037856693193577225,
      "loss": 1.9647,
      "step": 95768
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.9452130198478699,
      "learning_rate": 0.0003785629841359518,
      "loss": 2.0188,
      "step": 95769
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.7984347343444824,
      "learning_rate": 0.0003785590363215248,
      "loss": 2.1512,
      "step": 95770
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.8959227800369263,
      "learning_rate": 0.00037855508849249234,
      "loss": 2.1597,
      "step": 95771
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.8253505825996399,
      "learning_rate": 0.000378551140648855,
      "loss": 2.2697,
      "step": 95772
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.8049551844596863,
      "learning_rate": 0.0003785471927906134,
      "loss": 2.1918,
      "step": 95773
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.845594048500061,
      "learning_rate": 0.00037854324491776853,
      "loss": 2.024,
      "step": 95774
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.8291651606559753,
      "learning_rate": 0.00037853929703032097,
      "loss": 2.1495,
      "step": 95775
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.8172093033790588,
      "learning_rate": 0.0003785353491282714,
      "loss": 2.2224,
      "step": 95776
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.8333799242973328,
      "learning_rate": 0.00037853140121162066,
      "loss": 2.1501,
      "step": 95777
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.8407124876976013,
      "learning_rate": 0.00037852745328036954,
      "loss": 2.1684,
      "step": 95778
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.8836663365364075,
      "learning_rate": 0.00037852350533451855,
      "loss": 2.2194,
      "step": 95779
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.8511409163475037,
      "learning_rate": 0.0003785195573740686,
      "loss": 2.3511,
      "step": 95780
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.8328919410705566,
      "learning_rate": 0.0003785156093990204,
      "loss": 2.2079,
      "step": 95781
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.8048282861709595,
      "learning_rate": 0.00037851166140937464,
      "loss": 2.0787,
      "step": 95782
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.8342725038528442,
      "learning_rate": 0.00037850771340513205,
      "loss": 2.2779,
      "step": 95783
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.7880653142929077,
      "learning_rate": 0.00037850376538629355,
      "loss": 2.1804,
      "step": 95784
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.8469204306602478,
      "learning_rate": 0.00037849981735285957,
      "loss": 2.0865,
      "step": 95785
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.8005702495574951,
      "learning_rate": 0.000378495869304831,
      "loss": 1.875,
      "step": 95786
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.8586471676826477,
      "learning_rate": 0.00037849192124220855,
      "loss": 2.4409,
      "step": 95787
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.8312831521034241,
      "learning_rate": 0.00037848797316499314,
      "loss": 2.2942,
      "step": 95788
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.8123698830604553,
      "learning_rate": 0.0003784840250731851,
      "loss": 2.3135,
      "step": 95789
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.8263167142868042,
      "learning_rate": 0.0003784800769667856,
      "loss": 2.1,
      "step": 95790
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.8750483393669128,
      "learning_rate": 0.0003784761288457951,
      "loss": 2.5363,
      "step": 95791
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.9301266074180603,
      "learning_rate": 0.00037847218071021435,
      "loss": 2.2566,
      "step": 95792
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.8572970032691956,
      "learning_rate": 0.0003784682325600442,
      "loss": 2.09,
      "step": 95793
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.7685519456863403,
      "learning_rate": 0.00037846428439528533,
      "loss": 1.9952,
      "step": 95794
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.8699326515197754,
      "learning_rate": 0.0003784603362159384,
      "loss": 2.0424,
      "step": 95795
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.7988945245742798,
      "learning_rate": 0.0003784563880220042,
      "loss": 2.1641,
      "step": 95796
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.7560880184173584,
      "learning_rate": 0.00037845243981348363,
      "loss": 2.2179,
      "step": 95797
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.8537948727607727,
      "learning_rate": 0.0003784484915903771,
      "loss": 2.2767,
      "step": 95798
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.8070707321166992,
      "learning_rate": 0.00037844454335268557,
      "loss": 2.1547,
      "step": 95799
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.7911636233329773,
      "learning_rate": 0.0003784405951004097,
      "loss": 2.1903,
      "step": 95800
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.8218724131584167,
      "learning_rate": 0.0003784366468335503,
      "loss": 2.3035,
      "step": 95801
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.8190218806266785,
      "learning_rate": 0.0003784326985521081,
      "loss": 2.3199,
      "step": 95802
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.9012787938117981,
      "learning_rate": 0.0003784287502560837,
      "loss": 2.3281,
      "step": 95803
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.8795484304428101,
      "learning_rate": 0.0003784248019454779,
      "loss": 2.0506,
      "step": 95804
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.7942749857902527,
      "learning_rate": 0.0003784208536202915,
      "loss": 2.1243,
      "step": 95805
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.8577693700790405,
      "learning_rate": 0.0003784169052805251,
      "loss": 2.2249,
      "step": 95806
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.953483521938324,
      "learning_rate": 0.0003784129569261796,
      "loss": 2.2412,
      "step": 95807
    },
    {
      "epoch": 1.25,
      "grad_norm": 1.0642242431640625,
      "learning_rate": 0.0003784090085572556,
      "loss": 2.1071,
      "step": 95808
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.8555088639259338,
      "learning_rate": 0.00037840506017375395,
      "loss": 2.0753,
      "step": 95809
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.8967868685722351,
      "learning_rate": 0.00037840111177567525,
      "loss": 2.0469,
      "step": 95810
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.9484193921089172,
      "learning_rate": 0.00037839716336302026,
      "loss": 2.1728,
      "step": 95811
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.8363212943077087,
      "learning_rate": 0.0003783932149357899,
      "loss": 2.0447,
      "step": 95812
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.9044272303581238,
      "learning_rate": 0.0003783892664939847,
      "loss": 2.1872,
      "step": 95813
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.8736817240715027,
      "learning_rate": 0.00037838531803760533,
      "loss": 2.1122,
      "step": 95814
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.8553819060325623,
      "learning_rate": 0.0003783813695666529,
      "loss": 2.1098,
      "step": 95815
    },
    {
      "epoch": 1.25,
      "grad_norm": 1.0196951627731323,
      "learning_rate": 0.00037837742108112765,
      "loss": 2.3959,
      "step": 95816
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.8670305013656616,
      "learning_rate": 0.00037837347258103067,
      "loss": 2.0685,
      "step": 95817
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.9195502400398254,
      "learning_rate": 0.0003783695240663626,
      "loss": 2.1095,
      "step": 95818
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.8670923709869385,
      "learning_rate": 0.0003783655755371241,
      "loss": 2.2955,
      "step": 95819
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.7605099081993103,
      "learning_rate": 0.000378361626993316,
      "loss": 2.1944,
      "step": 95820
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.9085057973861694,
      "learning_rate": 0.0003783576784349391,
      "loss": 2.2339,
      "step": 95821
    },
    {
      "epoch": 1.25,
      "grad_norm": 1.0130083560943604,
      "learning_rate": 0.00037835372986199387,
      "loss": 2.0918,
      "step": 95822
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.7881784439086914,
      "learning_rate": 0.00037834978127448125,
      "loss": 2.2885,
      "step": 95823
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.828873872756958,
      "learning_rate": 0.00037834583267240193,
      "loss": 2.1512,
      "step": 95824
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.81898033618927,
      "learning_rate": 0.0003783418840557566,
      "loss": 2.046,
      "step": 95825
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.9628421068191528,
      "learning_rate": 0.00037833793542454606,
      "loss": 2.267,
      "step": 95826
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.9328550696372986,
      "learning_rate": 0.00037833398677877107,
      "loss": 1.8656,
      "step": 95827
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.9255630373954773,
      "learning_rate": 0.00037833003811843226,
      "loss": 2.3024,
      "step": 95828
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.9837791919708252,
      "learning_rate": 0.00037832608944353045,
      "loss": 2.1512,
      "step": 95829
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.8673064112663269,
      "learning_rate": 0.00037832214075406633,
      "loss": 2.3979,
      "step": 95830
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.9717276692390442,
      "learning_rate": 0.0003783181920500407,
      "loss": 2.3604,
      "step": 95831
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.997681736946106,
      "learning_rate": 0.0003783142433314541,
      "loss": 2.2315,
      "step": 95832
    },
    {
      "epoch": 1.25,
      "grad_norm": 1.369599461555481,
      "learning_rate": 0.0003783102945983075,
      "loss": 2.2606,
      "step": 95833
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.9920674562454224,
      "learning_rate": 0.00037830634585060154,
      "loss": 2.2907,
      "step": 95834
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.9317678809165955,
      "learning_rate": 0.0003783023970883369,
      "loss": 2.0283,
      "step": 95835
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.9502075910568237,
      "learning_rate": 0.00037829844831151443,
      "loss": 2.3245,
      "step": 95836
    },
    {
      "epoch": 1.25,
      "grad_norm": 1.011812448501587,
      "learning_rate": 0.0003782944995201348,
      "loss": 2.1577,
      "step": 95837
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.8704630136489868,
      "learning_rate": 0.0003782905507141986,
      "loss": 2.1807,
      "step": 95838
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.7848531603813171,
      "learning_rate": 0.00037828660189370694,
      "loss": 2.0617,
      "step": 95839
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.8793898820877075,
      "learning_rate": 0.0003782826530586602,
      "loss": 2.1004,
      "step": 95840
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.9406379461288452,
      "learning_rate": 0.0003782787042090592,
      "loss": 2.3064,
      "step": 95841
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.7880294322967529,
      "learning_rate": 0.00037827475534490474,
      "loss": 2.0217,
      "step": 95842
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.9120059013366699,
      "learning_rate": 0.0003782708064661976,
      "loss": 2.1828,
      "step": 95843
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.9077435731887817,
      "learning_rate": 0.0003782668575729383,
      "loss": 1.8839,
      "step": 95844
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.7783873081207275,
      "learning_rate": 0.0003782629086651278,
      "loss": 2.1151,
      "step": 95845
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.8482793569564819,
      "learning_rate": 0.00037825895974276673,
      "loss": 2.2167,
      "step": 95846
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.9902245402336121,
      "learning_rate": 0.00037825501080585586,
      "loss": 2.1646,
      "step": 95847
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.9847378730773926,
      "learning_rate": 0.000378251061854396,
      "loss": 2.3458,
      "step": 95848
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.8448750376701355,
      "learning_rate": 0.0003782471128883876,
      "loss": 2.2649,
      "step": 95849
    },
    {
      "epoch": 1.25,
      "grad_norm": 1.072986125946045,
      "learning_rate": 0.00037824316390783165,
      "loss": 2.3627,
      "step": 95850
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.8673216700553894,
      "learning_rate": 0.0003782392149127289,
      "loss": 2.0959,
      "step": 95851
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.848573625087738,
      "learning_rate": 0.0003782352659030799,
      "loss": 2.0534,
      "step": 95852
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.8425368070602417,
      "learning_rate": 0.0003782313168788855,
      "loss": 2.1392,
      "step": 95853
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.854105532169342,
      "learning_rate": 0.00037822736784014644,
      "loss": 2.276,
      "step": 95854
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.8811786770820618,
      "learning_rate": 0.0003782234187868635,
      "loss": 2.2426,
      "step": 95855
    },
    {
      "epoch": 1.25,
      "grad_norm": 1.3266898393630981,
      "learning_rate": 0.00037821946971903727,
      "loss": 2.0301,
      "step": 95856
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.9459395408630371,
      "learning_rate": 0.00037821552063666865,
      "loss": 2.1056,
      "step": 95857
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.8661355376243591,
      "learning_rate": 0.00037821157153975827,
      "loss": 2.0182,
      "step": 95858
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.8959873914718628,
      "learning_rate": 0.00037820762242830685,
      "loss": 2.1626,
      "step": 95859
    },
    {
      "epoch": 1.25,
      "grad_norm": 1.2560193538665771,
      "learning_rate": 0.0003782036733023151,
      "loss": 2.1185,
      "step": 95860
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.8938807249069214,
      "learning_rate": 0.0003781997241617839,
      "loss": 2.1959,
      "step": 95861
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.8081650733947754,
      "learning_rate": 0.00037819577500671387,
      "loss": 2.0156,
      "step": 95862
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.846893310546875,
      "learning_rate": 0.0003781918258371057,
      "loss": 2.2198,
      "step": 95863
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.9747481346130371,
      "learning_rate": 0.0003781878766529603,
      "loss": 2.0075,
      "step": 95864
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.8440404534339905,
      "learning_rate": 0.00037818392745427823,
      "loss": 2.1495,
      "step": 95865
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.8929485082626343,
      "learning_rate": 0.0003781799782410603,
      "loss": 2.29,
      "step": 95866
    },
    {
      "epoch": 1.25,
      "grad_norm": 1.1420553922653198,
      "learning_rate": 0.0003781760290133072,
      "loss": 2.077,
      "step": 95867
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.8903850317001343,
      "learning_rate": 0.00037817207977101973,
      "loss": 2.2735,
      "step": 95868
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.7700453996658325,
      "learning_rate": 0.00037816813051419864,
      "loss": 2.0721,
      "step": 95869
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.9074376225471497,
      "learning_rate": 0.00037816418124284463,
      "loss": 2.0586,
      "step": 95870
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.9548936486244202,
      "learning_rate": 0.0003781602319569584,
      "loss": 2.4624,
      "step": 95871
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.8463836908340454,
      "learning_rate": 0.00037815628265654063,
      "loss": 2.1275,
      "step": 95872
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.8301526308059692,
      "learning_rate": 0.0003781523333415922,
      "loss": 2.2865,
      "step": 95873
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.8270131945610046,
      "learning_rate": 0.0003781483840121138,
      "loss": 2.2438,
      "step": 95874
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.8409159183502197,
      "learning_rate": 0.0003781444346681061,
      "loss": 2.2109,
      "step": 95875
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.9361531138420105,
      "learning_rate": 0.0003781404853095699,
      "loss": 2.1372,
      "step": 95876
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.8351232409477234,
      "learning_rate": 0.0003781365359365059,
      "loss": 2.1643,
      "step": 95877
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.9371201992034912,
      "learning_rate": 0.0003781325865489148,
      "loss": 2.0484,
      "step": 95878
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.9746679067611694,
      "learning_rate": 0.0003781286371467975,
      "loss": 2.23,
      "step": 95879
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.8127063512802124,
      "learning_rate": 0.0003781246877301545,
      "loss": 2.2691,
      "step": 95880
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.8629660606384277,
      "learning_rate": 0.00037812073829898664,
      "loss": 1.9245,
      "step": 95881
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.9326102137565613,
      "learning_rate": 0.0003781167888532947,
      "loss": 2.113,
      "step": 95882
    },
    {
      "epoch": 1.25,
      "grad_norm": 1.3475526571273804,
      "learning_rate": 0.00037811283939307935,
      "loss": 1.9623,
      "step": 95883
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.951571524143219,
      "learning_rate": 0.0003781088899183414,
      "loss": 2.2175,
      "step": 95884
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.9814239740371704,
      "learning_rate": 0.00037810494042908143,
      "loss": 2.1828,
      "step": 95885
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.8863241672515869,
      "learning_rate": 0.0003781009909253004,
      "loss": 2.202,
      "step": 95886
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.8062447309494019,
      "learning_rate": 0.00037809704140699885,
      "loss": 1.9508,
      "step": 95887
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.8232729434967041,
      "learning_rate": 0.00037809309187417756,
      "loss": 2.1756,
      "step": 95888
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.8523977398872375,
      "learning_rate": 0.00037808914232683735,
      "loss": 2.0274,
      "step": 95889
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.7697912454605103,
      "learning_rate": 0.0003780851927649789,
      "loss": 2.5031,
      "step": 95890
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.8751901388168335,
      "learning_rate": 0.0003780812431886029,
      "loss": 2.3692,
      "step": 95891
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.8770937323570251,
      "learning_rate": 0.0003780772935977101,
      "loss": 2.3082,
      "step": 95892
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.8687336444854736,
      "learning_rate": 0.0003780733439923013,
      "loss": 1.9723,
      "step": 95893
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.8390392065048218,
      "learning_rate": 0.0003780693943723772,
      "loss": 2.2017,
      "step": 95894
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.8368526697158813,
      "learning_rate": 0.00037806544473793845,
      "loss": 2.1868,
      "step": 95895
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.9308425188064575,
      "learning_rate": 0.000378061495088986,
      "loss": 2.0022,
      "step": 95896
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.8658046126365662,
      "learning_rate": 0.00037805754542552034,
      "loss": 2.2731,
      "step": 95897
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.9252356290817261,
      "learning_rate": 0.00037805359574754233,
      "loss": 2.211,
      "step": 95898
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.8527442216873169,
      "learning_rate": 0.0003780496460550527,
      "loss": 2.3281,
      "step": 95899
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.77574622631073,
      "learning_rate": 0.00037804569634805217,
      "loss": 2.2376,
      "step": 95900
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.8712121844291687,
      "learning_rate": 0.00037804174662654143,
      "loss": 2.1966,
      "step": 95901
    },
    {
      "epoch": 1.25,
      "grad_norm": 1.1070246696472168,
      "learning_rate": 0.00037803779689052125,
      "loss": 2.1158,
      "step": 95902
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.8536878228187561,
      "learning_rate": 0.0003780338471399925,
      "loss": 2.2635,
      "step": 95903
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.8726274371147156,
      "learning_rate": 0.00037802989737495564,
      "loss": 2.1851,
      "step": 95904
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.8377850651741028,
      "learning_rate": 0.0003780259475954115,
      "loss": 2.2387,
      "step": 95905
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.8464467525482178,
      "learning_rate": 0.0003780219978013611,
      "loss": 2.3288,
      "step": 95906
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.7937713265419006,
      "learning_rate": 0.00037801804799280476,
      "loss": 2.3714,
      "step": 95907
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.8688459992408752,
      "learning_rate": 0.0003780140981697434,
      "loss": 2.1181,
      "step": 95908
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.786702036857605,
      "learning_rate": 0.00037801014833217783,
      "loss": 2.3204,
      "step": 95909
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.8425282835960388,
      "learning_rate": 0.0003780061984801086,
      "loss": 2.3299,
      "step": 95910
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.9010969996452332,
      "learning_rate": 0.0003780022486135366,
      "loss": 2.2893,
      "step": 95911
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.8896418213844299,
      "learning_rate": 0.00037799829873246254,
      "loss": 2.3364,
      "step": 95912
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.8457832336425781,
      "learning_rate": 0.0003779943488368871,
      "loss": 2.2779,
      "step": 95913
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.7882376313209534,
      "learning_rate": 0.000377990398926811,
      "loss": 2.0742,
      "step": 95914
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.8521284461021423,
      "learning_rate": 0.00037798644900223505,
      "loss": 2.0817,
      "step": 95915
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.8816295266151428,
      "learning_rate": 0.0003779824990631599,
      "loss": 2.1507,
      "step": 95916
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.8194339871406555,
      "learning_rate": 0.00037797854910958637,
      "loss": 2.1836,
      "step": 95917
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.8204417824745178,
      "learning_rate": 0.0003779745991415152,
      "loss": 2.1443,
      "step": 95918
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.8583785891532898,
      "learning_rate": 0.000377970649158947,
      "loss": 2.0845,
      "step": 95919
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.84371018409729,
      "learning_rate": 0.0003779666991618826,
      "loss": 2.1318,
      "step": 95920
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.8262590169906616,
      "learning_rate": 0.00037796274915032275,
      "loss": 2.2285,
      "step": 95921
    },
    {
      "epoch": 1.25,
      "grad_norm": 1.0420501232147217,
      "learning_rate": 0.0003779587991242682,
      "loss": 2.0245,
      "step": 95922
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.8366538882255554,
      "learning_rate": 0.0003779548490837195,
      "loss": 2.2108,
      "step": 95923
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.8665717244148254,
      "learning_rate": 0.0003779508990286776,
      "loss": 2.2132,
      "step": 95924
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.9374237656593323,
      "learning_rate": 0.0003779469489591432,
      "loss": 1.9645,
      "step": 95925
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.8708800673484802,
      "learning_rate": 0.00037794299887511693,
      "loss": 2.2032,
      "step": 95926
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.8796845078468323,
      "learning_rate": 0.0003779390487765996,
      "loss": 2.2759,
      "step": 95927
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.9447883367538452,
      "learning_rate": 0.000377935098663592,
      "loss": 2.2898,
      "step": 95928
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.8915639519691467,
      "learning_rate": 0.0003779311485360947,
      "loss": 2.0995,
      "step": 95929
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.8435386419296265,
      "learning_rate": 0.0003779271983941085,
      "loss": 2.0703,
      "step": 95930
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.9627689719200134,
      "learning_rate": 0.00037792324823763423,
      "loss": 2.4103,
      "step": 95931
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.9091755747795105,
      "learning_rate": 0.0003779192980666726,
      "loss": 2.223,
      "step": 95932
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.7829894423484802,
      "learning_rate": 0.00037791534788122425,
      "loss": 2.1219,
      "step": 95933
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.8262215852737427,
      "learning_rate": 0.00037791139768128995,
      "loss": 1.9719,
      "step": 95934
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.9718349575996399,
      "learning_rate": 0.00037790744746687044,
      "loss": 2.2147,
      "step": 95935
    },
    {
      "epoch": 1.25,
      "grad_norm": 1.0404560565948486,
      "learning_rate": 0.0003779034972379665,
      "loss": 2.3619,
      "step": 95936
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.9153035879135132,
      "learning_rate": 0.00037789954699457887,
      "loss": 2.1399,
      "step": 95937
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.9904718995094299,
      "learning_rate": 0.0003778955967367081,
      "loss": 2.1733,
      "step": 95938
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.8728963732719421,
      "learning_rate": 0.0003778916464643552,
      "loss": 2.3325,
      "step": 95939
    },
    {
      "epoch": 1.25,
      "grad_norm": 1.145871639251709,
      "learning_rate": 0.0003778876961775208,
      "loss": 2.1476,
      "step": 95940
    },
    {
      "epoch": 1.25,
      "grad_norm": 1.0448096990585327,
      "learning_rate": 0.00037788374587620544,
      "loss": 2.1259,
      "step": 95941
    },
    {
      "epoch": 1.25,
      "grad_norm": 1.0602343082427979,
      "learning_rate": 0.0003778797955604101,
      "loss": 2.2068,
      "step": 95942
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.9026216864585876,
      "learning_rate": 0.0003778758452301355,
      "loss": 2.2649,
      "step": 95943
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.8330212831497192,
      "learning_rate": 0.0003778718948853823,
      "loss": 1.9513,
      "step": 95944
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.8732285499572754,
      "learning_rate": 0.0003778679445261512,
      "loss": 1.9704,
      "step": 95945
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.993170440196991,
      "learning_rate": 0.00037786399415244297,
      "loss": 2.1286,
      "step": 95946
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.8276412487030029,
      "learning_rate": 0.0003778600437642584,
      "loss": 2.4587,
      "step": 95947
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.882845401763916,
      "learning_rate": 0.00037785609336159817,
      "loss": 2.2509,
      "step": 95948
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.9886480569839478,
      "learning_rate": 0.000377852142944463,
      "loss": 2.3632,
      "step": 95949
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.915883481502533,
      "learning_rate": 0.0003778481925128536,
      "loss": 2.1721,
      "step": 95950
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.7507032752037048,
      "learning_rate": 0.0003778442420667708,
      "loss": 2.0798,
      "step": 95951
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.9340132474899292,
      "learning_rate": 0.00037784029160621536,
      "loss": 2.0756,
      "step": 95952
    },
    {
      "epoch": 1.25,
      "grad_norm": 1.1777925491333008,
      "learning_rate": 0.00037783634113118785,
      "loss": 2.081,
      "step": 95953
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.8727258443832397,
      "learning_rate": 0.00037783239064168906,
      "loss": 1.957,
      "step": 95954
    },
    {
      "epoch": 1.25,
      "grad_norm": 1.0143978595733643,
      "learning_rate": 0.00037782844013771986,
      "loss": 2.0788,
      "step": 95955
    },
    {
      "epoch": 1.25,
      "grad_norm": 1.1096444129943848,
      "learning_rate": 0.00037782448961928085,
      "loss": 2.3153,
      "step": 95956
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.8186267018318176,
      "learning_rate": 0.0003778205390863728,
      "loss": 2.287,
      "step": 95957
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.861821711063385,
      "learning_rate": 0.00037781658853899646,
      "loss": 2.3425,
      "step": 95958
    },
    {
      "epoch": 1.25,
      "grad_norm": 1.1480873823165894,
      "learning_rate": 0.00037781263797715244,
      "loss": 2.123,
      "step": 95959
    },
    {
      "epoch": 1.25,
      "grad_norm": 1.222086787223816,
      "learning_rate": 0.0003778086874008417,
      "loss": 2.0216,
      "step": 95960
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.8434967994689941,
      "learning_rate": 0.00037780473681006487,
      "loss": 2.1831,
      "step": 95961
    },
    {
      "epoch": 1.25,
      "grad_norm": 1.1480543613433838,
      "learning_rate": 0.00037780078620482257,
      "loss": 2.2196,
      "step": 95962
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.9946768879890442,
      "learning_rate": 0.00037779683558511565,
      "loss": 2.3963,
      "step": 95963
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.8311654925346375,
      "learning_rate": 0.00037779288495094494,
      "loss": 2.2448,
      "step": 95964
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.7969911694526672,
      "learning_rate": 0.0003777889343023109,
      "loss": 2.153,
      "step": 95965
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.902651309967041,
      "learning_rate": 0.0003777849836392145,
      "loss": 2.333,
      "step": 95966
    },
    {
      "epoch": 1.25,
      "grad_norm": 1.038720726966858,
      "learning_rate": 0.00037778103296165646,
      "loss": 2.0769,
      "step": 95967
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.9246187210083008,
      "learning_rate": 0.00037777708226963743,
      "loss": 2.1816,
      "step": 95968
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.9281442761421204,
      "learning_rate": 0.00037777313156315813,
      "loss": 2.2322,
      "step": 95969
    },
    {
      "epoch": 1.25,
      "grad_norm": 1.0170187950134277,
      "learning_rate": 0.00037776918084221937,
      "loss": 2.1473,
      "step": 95970
    },
    {
      "epoch": 1.25,
      "grad_norm": 1.1224018335342407,
      "learning_rate": 0.00037776523010682185,
      "loss": 2.3299,
      "step": 95971
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.9046185612678528,
      "learning_rate": 0.0003777612793569663,
      "loss": 2.28,
      "step": 95972
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.9718775749206543,
      "learning_rate": 0.0003777573285926534,
      "loss": 2.0046,
      "step": 95973
    },
    {
      "epoch": 1.25,
      "grad_norm": 1.1247460842132568,
      "learning_rate": 0.00037775337781388406,
      "loss": 2.0947,
      "step": 95974
    },
    {
      "epoch": 1.25,
      "grad_norm": 1.1535927057266235,
      "learning_rate": 0.00037774942702065883,
      "loss": 2.2039,
      "step": 95975
    },
    {
      "epoch": 1.25,
      "grad_norm": 1.1104310750961304,
      "learning_rate": 0.00037774547621297847,
      "loss": 2.1222,
      "step": 95976
    },
    {
      "epoch": 1.25,
      "grad_norm": 1.1749142408370972,
      "learning_rate": 0.00037774152539084385,
      "loss": 2.0522,
      "step": 95977
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.7470481395721436,
      "learning_rate": 0.00037773757455425563,
      "loss": 1.9879,
      "step": 95978
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.9173930883407593,
      "learning_rate": 0.00037773362370321445,
      "loss": 2.113,
      "step": 95979
    },
    {
      "epoch": 1.25,
      "grad_norm": 1.0116493701934814,
      "learning_rate": 0.00037772967283772113,
      "loss": 2.2041,
      "step": 95980
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.9329285621643066,
      "learning_rate": 0.00037772572195777647,
      "loss": 2.0685,
      "step": 95981
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.8949904441833496,
      "learning_rate": 0.00037772177106338103,
      "loss": 2.0621,
      "step": 95982
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.835322916507721,
      "learning_rate": 0.00037771782015453566,
      "loss": 2.3096,
      "step": 95983
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.9980617165565491,
      "learning_rate": 0.00037771386923124114,
      "loss": 1.9987,
      "step": 95984
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.7809913754463196,
      "learning_rate": 0.00037770991829349815,
      "loss": 2.2227,
      "step": 95985
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.8514241576194763,
      "learning_rate": 0.00037770596734130736,
      "loss": 2.1591,
      "step": 95986
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.9720931053161621,
      "learning_rate": 0.0003777020163746696,
      "loss": 2.1342,
      "step": 95987
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.8583678007125854,
      "learning_rate": 0.00037769806539358555,
      "loss": 2.1913,
      "step": 95988
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.810601532459259,
      "learning_rate": 0.000377694114398056,
      "loss": 1.9909,
      "step": 95989
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.8321210741996765,
      "learning_rate": 0.0003776901633880816,
      "loss": 2.348,
      "step": 95990
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.9132347702980042,
      "learning_rate": 0.0003776862123636631,
      "loss": 2.0876,
      "step": 95991
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.9407815337181091,
      "learning_rate": 0.0003776822613248013,
      "loss": 2.3679,
      "step": 95992
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.8426848649978638,
      "learning_rate": 0.00037767831027149697,
      "loss": 2.1027,
      "step": 95993
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.8410668969154358,
      "learning_rate": 0.0003776743592037507,
      "loss": 2.0223,
      "step": 95994
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.8461706042289734,
      "learning_rate": 0.0003776704081215633,
      "loss": 2.4151,
      "step": 95995
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.8532496690750122,
      "learning_rate": 0.0003776664570249356,
      "loss": 2.2003,
      "step": 95996
    },
    {
      "epoch": 1.25,
      "grad_norm": 1.3957833051681519,
      "learning_rate": 0.0003776625059138681,
      "loss": 2.0762,
      "step": 95997
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.8614977598190308,
      "learning_rate": 0.00037765855478836185,
      "loss": 2.2508,
      "step": 95998
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.8098522424697876,
      "learning_rate": 0.0003776546036484173,
      "loss": 2.216,
      "step": 95999
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.90935218334198,
      "learning_rate": 0.0003776506524940352,
      "loss": 2.1365,
      "step": 96000
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.8317690491676331,
      "learning_rate": 0.0003776467013252165,
      "loss": 2.234,
      "step": 96001
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.8306635022163391,
      "learning_rate": 0.0003776427501419618,
      "loss": 2.1489,
      "step": 96002
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.9466466307640076,
      "learning_rate": 0.0003776387989442718,
      "loss": 2.1812,
      "step": 96003
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.8393067717552185,
      "learning_rate": 0.0003776348477321474,
      "loss": 2.2338,
      "step": 96004
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.9740830659866333,
      "learning_rate": 0.0003776308965055891,
      "loss": 2.174,
      "step": 96005
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.9710103869438171,
      "learning_rate": 0.00037762694526459776,
      "loss": 2.0714,
      "step": 96006
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.863963782787323,
      "learning_rate": 0.0003776229940091742,
      "loss": 2.2733,
      "step": 96007
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.8806717991828918,
      "learning_rate": 0.00037761904273931896,
      "loss": 2.1859,
      "step": 96008
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.842616081237793,
      "learning_rate": 0.0003776150914550329,
      "loss": 2.0554,
      "step": 96009
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.8392276167869568,
      "learning_rate": 0.0003776111401563167,
      "loss": 2.2245,
      "step": 96010
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.9745429754257202,
      "learning_rate": 0.00037760718884317115,
      "loss": 2.1734,
      "step": 96011
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.840449333190918,
      "learning_rate": 0.00037760323751559694,
      "loss": 1.918,
      "step": 96012
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.9549130201339722,
      "learning_rate": 0.0003775992861735949,
      "loss": 2.206,
      "step": 96013
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.9762747883796692,
      "learning_rate": 0.00037759533481716563,
      "loss": 2.4285,
      "step": 96014
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.8991851210594177,
      "learning_rate": 0.00037759138344630995,
      "loss": 2.2003,
      "step": 96015
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.8648878931999207,
      "learning_rate": 0.00037758743206102853,
      "loss": 2.0395,
      "step": 96016
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.8181470632553101,
      "learning_rate": 0.0003775834806613222,
      "loss": 2.256,
      "step": 96017
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.899300754070282,
      "learning_rate": 0.00037757952924719155,
      "loss": 2.1905,
      "step": 96018
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.8352858424186707,
      "learning_rate": 0.0003775755778186375,
      "loss": 2.198,
      "step": 96019
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.9000073671340942,
      "learning_rate": 0.00037757162637566064,
      "loss": 2.1413,
      "step": 96020
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.9595441222190857,
      "learning_rate": 0.00037756767491826167,
      "loss": 2.0858,
      "step": 96021
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.9185848832130432,
      "learning_rate": 0.0003775637234464415,
      "loss": 2.2268,
      "step": 96022
    },
    {
      "epoch": 1.25,
      "grad_norm": 1.0100921392440796,
      "learning_rate": 0.00037755977196020075,
      "loss": 2.0922,
      "step": 96023
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.873290479183197,
      "learning_rate": 0.00037755582045954016,
      "loss": 2.1019,
      "step": 96024
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.9619118571281433,
      "learning_rate": 0.00037755186894446055,
      "loss": 2.202,
      "step": 96025
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.8975304961204529,
      "learning_rate": 0.0003775479174149625,
      "loss": 2.1831,
      "step": 96026
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.7769743204116821,
      "learning_rate": 0.00037754396587104687,
      "loss": 2.119,
      "step": 96027
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.9741042852401733,
      "learning_rate": 0.0003775400143127143,
      "loss": 2.3912,
      "step": 96028
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.9204932451248169,
      "learning_rate": 0.0003775360627399656,
      "loss": 2.3627,
      "step": 96029
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.8456143736839294,
      "learning_rate": 0.0003775321111528015,
      "loss": 2.1549,
      "step": 96030
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.9090761542320251,
      "learning_rate": 0.0003775281595512228,
      "loss": 2.3981,
      "step": 96031
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.9238156080245972,
      "learning_rate": 0.00037752420793523003,
      "loss": 1.8667,
      "step": 96032
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.8264799118041992,
      "learning_rate": 0.00037752025630482405,
      "loss": 2.2863,
      "step": 96033
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.8254212141036987,
      "learning_rate": 0.00037751630466000566,
      "loss": 2.2043,
      "step": 96034
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.8514968752861023,
      "learning_rate": 0.00037751235300077546,
      "loss": 2.0101,
      "step": 96035
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.7970923781394958,
      "learning_rate": 0.00037750840132713426,
      "loss": 2.2041,
      "step": 96036
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.8843916058540344,
      "learning_rate": 0.0003775044496390829,
      "loss": 2.0671,
      "step": 96037
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.7974969148635864,
      "learning_rate": 0.0003775004979366218,
      "loss": 2.133,
      "step": 96038
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.8421044945716858,
      "learning_rate": 0.00037749654621975205,
      "loss": 2.1613,
      "step": 96039
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.8679066300392151,
      "learning_rate": 0.0003774925944884742,
      "loss": 2.2018,
      "step": 96040
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.9094911217689514,
      "learning_rate": 0.00037748864274278894,
      "loss": 2.4286,
      "step": 96041
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.8690980076789856,
      "learning_rate": 0.0003774846909826971,
      "loss": 2.1094,
      "step": 96042
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.8354507684707642,
      "learning_rate": 0.0003774807392081995,
      "loss": 2.3483,
      "step": 96043
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.882847011089325,
      "learning_rate": 0.0003774767874192966,
      "loss": 2.0046,
      "step": 96044
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.8344719409942627,
      "learning_rate": 0.00037747283561598944,
      "loss": 1.9833,
      "step": 96045
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.9418998956680298,
      "learning_rate": 0.0003774688837982786,
      "loss": 2.132,
      "step": 96046
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.8930766582489014,
      "learning_rate": 0.00037746493196616477,
      "loss": 1.9941,
      "step": 96047
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.9347193837165833,
      "learning_rate": 0.0003774609801196487,
      "loss": 2.1366,
      "step": 96048
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.8835973143577576,
      "learning_rate": 0.0003774570282587314,
      "loss": 2.0883,
      "step": 96049
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.8855295181274414,
      "learning_rate": 0.0003774530763834132,
      "loss": 2.0849,
      "step": 96050
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.8639792203903198,
      "learning_rate": 0.000377449124493695,
      "loss": 2.1223,
      "step": 96051
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.8605936765670776,
      "learning_rate": 0.0003774451725895776,
      "loss": 2.1594,
      "step": 96052
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.8891992568969727,
      "learning_rate": 0.00037744122067106167,
      "loss": 2.3592,
      "step": 96053
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.8619934320449829,
      "learning_rate": 0.00037743726873814794,
      "loss": 2.11,
      "step": 96054
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.9097234010696411,
      "learning_rate": 0.00037743331679083715,
      "loss": 1.9477,
      "step": 96055
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.8064280152320862,
      "learning_rate": 0.00037742936482913005,
      "loss": 2.2334,
      "step": 96056
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.877071738243103,
      "learning_rate": 0.00037742541285302734,
      "loss": 2.1557,
      "step": 96057
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.8702065348625183,
      "learning_rate": 0.00037742146086252985,
      "loss": 2.1184,
      "step": 96058
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.8578085899353027,
      "learning_rate": 0.0003774175088576383,
      "loss": 2.0861,
      "step": 96059
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.9498971700668335,
      "learning_rate": 0.00037741355683835326,
      "loss": 2.1428,
      "step": 96060
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.8858411312103271,
      "learning_rate": 0.00037740960480467557,
      "loss": 2.159,
      "step": 96061
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.7696965336799622,
      "learning_rate": 0.00037740565275660613,
      "loss": 2.2274,
      "step": 96062
    },
    {
      "epoch": 1.25,
      "grad_norm": 1.1768580675125122,
      "learning_rate": 0.00037740170069414537,
      "loss": 1.9165,
      "step": 96063
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.891139805316925,
      "learning_rate": 0.00037739774861729415,
      "loss": 2.3734,
      "step": 96064
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.9073325991630554,
      "learning_rate": 0.0003773937965260533,
      "loss": 2.2862,
      "step": 96065
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.9999533295631409,
      "learning_rate": 0.00037738984442042356,
      "loss": 2.1743,
      "step": 96066
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.8718078136444092,
      "learning_rate": 0.00037738589230040543,
      "loss": 2.2003,
      "step": 96067
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.8454574942588806,
      "learning_rate": 0.0003773819401659999,
      "loss": 2.0304,
      "step": 96068
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.8338883519172668,
      "learning_rate": 0.0003773779880172076,
      "loss": 2.3694,
      "step": 96069
    },
    {
      "epoch": 1.25,
      "grad_norm": 1.0611867904663086,
      "learning_rate": 0.0003773740358540292,
      "loss": 2.1219,
      "step": 96070
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.9498296976089478,
      "learning_rate": 0.0003773700836764655,
      "loss": 2.2055,
      "step": 96071
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.7781501412391663,
      "learning_rate": 0.00037736613148451736,
      "loss": 2.2445,
      "step": 96072
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.822502076625824,
      "learning_rate": 0.00037736217927818535,
      "loss": 2.0365,
      "step": 96073
    },
    {
      "epoch": 1.25,
      "grad_norm": 1.3794631958007812,
      "learning_rate": 0.00037735822705747027,
      "loss": 2.1671,
      "step": 96074
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.7977731227874756,
      "learning_rate": 0.00037735427482237275,
      "loss": 2.0998,
      "step": 96075
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.8621395826339722,
      "learning_rate": 0.0003773503225728936,
      "loss": 2.2634,
      "step": 96076
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.9325057864189148,
      "learning_rate": 0.00037734637030903374,
      "loss": 2.0766,
      "step": 96077
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.8980264663696289,
      "learning_rate": 0.0003773424180307936,
      "loss": 2.1132,
      "step": 96078
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.8275741338729858,
      "learning_rate": 0.000377338465738174,
      "loss": 2.0547,
      "step": 96079
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.9468640685081482,
      "learning_rate": 0.0003773345134311758,
      "loss": 2.2542,
      "step": 96080
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.9172917604446411,
      "learning_rate": 0.0003773305611097996,
      "loss": 2.4378,
      "step": 96081
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.8888850808143616,
      "learning_rate": 0.0003773266087740462,
      "loss": 2.1473,
      "step": 96082
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.874363124370575,
      "learning_rate": 0.0003773226564239164,
      "loss": 2.1396,
      "step": 96083
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.8726010918617249,
      "learning_rate": 0.0003773187040594108,
      "loss": 2.1914,
      "step": 96084
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.843669056892395,
      "learning_rate": 0.00037731475168053017,
      "loss": 2.3717,
      "step": 96085
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.8070566058158875,
      "learning_rate": 0.00037731079928727537,
      "loss": 2.0233,
      "step": 96086
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.8228449821472168,
      "learning_rate": 0.0003773068468796469,
      "loss": 2.1177,
      "step": 96087
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.8036385774612427,
      "learning_rate": 0.0003773028944576457,
      "loss": 1.9625,
      "step": 96088
    },
    {
      "epoch": 1.25,
      "grad_norm": 1.0411344766616821,
      "learning_rate": 0.00037729894202127243,
      "loss": 2.2148,
      "step": 96089
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.8713628053665161,
      "learning_rate": 0.0003772949895705278,
      "loss": 2.171,
      "step": 96090
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.8952273726463318,
      "learning_rate": 0.0003772910371054125,
      "loss": 2.1221,
      "step": 96091
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.8749905824661255,
      "learning_rate": 0.0003772870846259275,
      "loss": 2.4403,
      "step": 96092
    },
    {
      "epoch": 1.25,
      "grad_norm": 1.5209825038909912,
      "learning_rate": 0.0003772831321320733,
      "loss": 2.0201,
      "step": 96093
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.9273594617843628,
      "learning_rate": 0.0003772791796238507,
      "loss": 2.0514,
      "step": 96094
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.9363444447517395,
      "learning_rate": 0.0003772752271012605,
      "loss": 2.0407,
      "step": 96095
    },
    {
      "epoch": 1.25,
      "grad_norm": 1.0601861476898193,
      "learning_rate": 0.0003772712745643033,
      "loss": 2.1303,
      "step": 96096
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.8879122138023376,
      "learning_rate": 0.0003772673220129799,
      "loss": 1.8698,
      "step": 96097
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.833517849445343,
      "learning_rate": 0.00037726336944729114,
      "loss": 2.1221,
      "step": 96098
    },
    {
      "epoch": 1.25,
      "grad_norm": 1.0336277484893799,
      "learning_rate": 0.0003772594168672376,
      "loss": 2.0964,
      "step": 96099
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.9458556175231934,
      "learning_rate": 0.00037725546427282,
      "loss": 2.1548,
      "step": 96100
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.8480797410011292,
      "learning_rate": 0.0003772515116640393,
      "loss": 2.1538,
      "step": 96101
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.9150289297103882,
      "learning_rate": 0.00037724755904089603,
      "loss": 2.2232,
      "step": 96102
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.9413920044898987,
      "learning_rate": 0.0003772436064033909,
      "loss": 2.1726,
      "step": 96103
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.8656726479530334,
      "learning_rate": 0.00037723965375152484,
      "loss": 2.0884,
      "step": 96104
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.9160420894622803,
      "learning_rate": 0.00037723570108529845,
      "loss": 2.3423,
      "step": 96105
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.9000684022903442,
      "learning_rate": 0.0003772317484047124,
      "loss": 2.176,
      "step": 96106
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.8565242886543274,
      "learning_rate": 0.00037722779570976755,
      "loss": 2.2559,
      "step": 96107
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.9447035789489746,
      "learning_rate": 0.0003772238430004646,
      "loss": 2.1873,
      "step": 96108
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.9199645519256592,
      "learning_rate": 0.00037721989027680435,
      "loss": 2.2231,
      "step": 96109
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.879360556602478,
      "learning_rate": 0.0003772159375387874,
      "loss": 2.0768,
      "step": 96110
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.8789564967155457,
      "learning_rate": 0.00037721198478641454,
      "loss": 2.1996,
      "step": 96111
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.8783397674560547,
      "learning_rate": 0.00037720803201968657,
      "loss": 2.115,
      "step": 96112
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.8394860029220581,
      "learning_rate": 0.0003772040792386041,
      "loss": 2.0574,
      "step": 96113
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.8322760462760925,
      "learning_rate": 0.000377200126443168,
      "loss": 2.0185,
      "step": 96114
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.9375719428062439,
      "learning_rate": 0.00037719617363337883,
      "loss": 2.1821,
      "step": 96115
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.931410014629364,
      "learning_rate": 0.0003771922208092376,
      "loss": 2.151,
      "step": 96116
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.8531737923622131,
      "learning_rate": 0.0003771882679707447,
      "loss": 2.239,
      "step": 96117
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.8164745569229126,
      "learning_rate": 0.00037718431511790115,
      "loss": 2.094,
      "step": 96118
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.8444005846977234,
      "learning_rate": 0.0003771803622507076,
      "loss": 2.5392,
      "step": 96119
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.7986249923706055,
      "learning_rate": 0.00037717640936916466,
      "loss": 1.9745,
      "step": 96120
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.9473913908004761,
      "learning_rate": 0.00037717245647327324,
      "loss": 2.0959,
      "step": 96121
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.8911657333374023,
      "learning_rate": 0.000377168503563034,
      "loss": 2.2184,
      "step": 96122
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.8169296383857727,
      "learning_rate": 0.00037716455063844774,
      "loss": 2.2082,
      "step": 96123
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.8603453040122986,
      "learning_rate": 0.000377160597699515,
      "loss": 2.0597,
      "step": 96124
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.8468630909919739,
      "learning_rate": 0.0003771566447462367,
      "loss": 2.2417,
      "step": 96125
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.8446101546287537,
      "learning_rate": 0.0003771526917786136,
      "loss": 1.9773,
      "step": 96126
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.7829409241676331,
      "learning_rate": 0.0003771487387966463,
      "loss": 1.933,
      "step": 96127
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.8870376944541931,
      "learning_rate": 0.00037714478580033555,
      "loss": 2.0749,
      "step": 96128
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.8921277523040771,
      "learning_rate": 0.0003771408327896822,
      "loss": 2.2166,
      "step": 96129
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.8348807096481323,
      "learning_rate": 0.00037713687976468686,
      "loss": 2.0684,
      "step": 96130
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.868074357509613,
      "learning_rate": 0.00037713292672535034,
      "loss": 2.3692,
      "step": 96131
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.8481525182723999,
      "learning_rate": 0.0003771289736716734,
      "loss": 1.9552,
      "step": 96132
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.8622639179229736,
      "learning_rate": 0.00037712502060365666,
      "loss": 2.3934,
      "step": 96133
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.8492266535758972,
      "learning_rate": 0.0003771210675213009,
      "loss": 2.1379,
      "step": 96134
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.9781566858291626,
      "learning_rate": 0.00037711711442460694,
      "loss": 2.2358,
      "step": 96135
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.8375810384750366,
      "learning_rate": 0.00037711316131357545,
      "loss": 2.1429,
      "step": 96136
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.7576289772987366,
      "learning_rate": 0.0003771092081882071,
      "loss": 2.3774,
      "step": 96137
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.8820127248764038,
      "learning_rate": 0.0003771052550485028,
      "loss": 2.1949,
      "step": 96138
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.9011525511741638,
      "learning_rate": 0.0003771013018944631,
      "loss": 2.2204,
      "step": 96139
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.9935120940208435,
      "learning_rate": 0.00037709734872608887,
      "loss": 2.0407,
      "step": 96140
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.8892686367034912,
      "learning_rate": 0.0003770933955433807,
      "loss": 2.2763,
      "step": 96141
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.8613776564598083,
      "learning_rate": 0.0003770894423463395,
      "loss": 2.0737,
      "step": 96142
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.8567877411842346,
      "learning_rate": 0.0003770854891349659,
      "loss": 2.3704,
      "step": 96143
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.8727461099624634,
      "learning_rate": 0.00037708153590926063,
      "loss": 2.2466,
      "step": 96144
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.8704726099967957,
      "learning_rate": 0.00037707758266922447,
      "loss": 2.1196,
      "step": 96145
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.9859043955802917,
      "learning_rate": 0.0003770736294148581,
      "loss": 2.4143,
      "step": 96146
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.828273355960846,
      "learning_rate": 0.00037706967614616234,
      "loss": 2.2779,
      "step": 96147
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.9764162302017212,
      "learning_rate": 0.0003770657228631378,
      "loss": 2.3007,
      "step": 96148
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.8311954140663147,
      "learning_rate": 0.0003770617695657853,
      "loss": 2.2028,
      "step": 96149
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.8327317833900452,
      "learning_rate": 0.0003770578162541056,
      "loss": 2.1582,
      "step": 96150
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.9622195959091187,
      "learning_rate": 0.00037705386292809933,
      "loss": 2.1999,
      "step": 96151
    },
    {
      "epoch": 1.25,
      "grad_norm": 1.0377368927001953,
      "learning_rate": 0.00037704990958776737,
      "loss": 2.1935,
      "step": 96152
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.8856587409973145,
      "learning_rate": 0.00037704595623311033,
      "loss": 2.1929,
      "step": 96153
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.8829204440116882,
      "learning_rate": 0.0003770420028641291,
      "loss": 2.1424,
      "step": 96154
    },
    {
      "epoch": 1.25,
      "grad_norm": 1.0738104581832886,
      "learning_rate": 0.0003770380494808242,
      "loss": 2.1275,
      "step": 96155
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.9246405363082886,
      "learning_rate": 0.00037703409608319646,
      "loss": 2.0517,
      "step": 96156
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.8363962173461914,
      "learning_rate": 0.0003770301426712466,
      "loss": 2.375,
      "step": 96157
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.8502927422523499,
      "learning_rate": 0.00037702618924497546,
      "loss": 2.3486,
      "step": 96158
    },
    {
      "epoch": 1.25,
      "grad_norm": 1.0001765489578247,
      "learning_rate": 0.0003770222358043837,
      "loss": 2.1965,
      "step": 96159
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.8909844756126404,
      "learning_rate": 0.000377018282349472,
      "loss": 2.1986,
      "step": 96160
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.9197293519973755,
      "learning_rate": 0.0003770143288802412,
      "loss": 1.9641,
      "step": 96161
    },
    {
      "epoch": 1.25,
      "grad_norm": 1.0377612113952637,
      "learning_rate": 0.0003770103753966919,
      "loss": 2.2348,
      "step": 96162
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.8511267900466919,
      "learning_rate": 0.000377006421898825,
      "loss": 2.3505,
      "step": 96163
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.7866608500480652,
      "learning_rate": 0.0003770024683866412,
      "loss": 1.9441,
      "step": 96164
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.8642232418060303,
      "learning_rate": 0.00037699851486014115,
      "loss": 2.1373,
      "step": 96165
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.8464785218238831,
      "learning_rate": 0.0003769945613193255,
      "loss": 2.1251,
      "step": 96166
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.8470568060874939,
      "learning_rate": 0.0003769906077641952,
      "loss": 2.1513,
      "step": 96167
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.9738559126853943,
      "learning_rate": 0.00037698665419475094,
      "loss": 2.435,
      "step": 96168
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.8173040747642517,
      "learning_rate": 0.00037698270061099336,
      "loss": 2.1041,
      "step": 96169
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.8900270462036133,
      "learning_rate": 0.00037697874701292323,
      "loss": 2.315,
      "step": 96170
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.8429808616638184,
      "learning_rate": 0.00037697479340054136,
      "loss": 2.199,
      "step": 96171
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.8820272088050842,
      "learning_rate": 0.00037697083977384835,
      "loss": 1.9353,
      "step": 96172
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.9439142346382141,
      "learning_rate": 0.000376966886132845,
      "loss": 2.2004,
      "step": 96173
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.8407213687896729,
      "learning_rate": 0.00037696293247753215,
      "loss": 2.1655,
      "step": 96174
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.9237492084503174,
      "learning_rate": 0.0003769589788079103,
      "loss": 2.1357,
      "step": 96175
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.9477881193161011,
      "learning_rate": 0.0003769550251239804,
      "loss": 2.3061,
      "step": 96176
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.7995514869689941,
      "learning_rate": 0.00037695107142574313,
      "loss": 2.1917,
      "step": 96177
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.8500715494155884,
      "learning_rate": 0.0003769471177131992,
      "loss": 2.182,
      "step": 96178
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.8135961294174194,
      "learning_rate": 0.00037694316398634927,
      "loss": 2.1078,
      "step": 96179
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.873241662979126,
      "learning_rate": 0.00037693921024519426,
      "loss": 2.1371,
      "step": 96180
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.8840436339378357,
      "learning_rate": 0.0003769352564897348,
      "loss": 1.9006,
      "step": 96181
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.8343308568000793,
      "learning_rate": 0.00037693130271997154,
      "loss": 2.0378,
      "step": 96182
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.8576573729515076,
      "learning_rate": 0.0003769273489359053,
      "loss": 1.9922,
      "step": 96183
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.874535858631134,
      "learning_rate": 0.0003769233951375369,
      "loss": 2.0919,
      "step": 96184
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.8527173399925232,
      "learning_rate": 0.00037691944132486687,
      "loss": 2.0843,
      "step": 96185
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.8687745332717896,
      "learning_rate": 0.00037691548749789613,
      "loss": 2.0063,
      "step": 96186
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.9845132827758789,
      "learning_rate": 0.0003769115336566254,
      "loss": 2.0067,
      "step": 96187
    },
    {
      "epoch": 1.25,
      "grad_norm": 1.0003788471221924,
      "learning_rate": 0.00037690757980105526,
      "loss": 2.248,
      "step": 96188
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.9658569097518921,
      "learning_rate": 0.0003769036259311866,
      "loss": 2.2562,
      "step": 96189
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.8592303991317749,
      "learning_rate": 0.0003768996720470201,
      "loss": 2.2254,
      "step": 96190
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.7986487150192261,
      "learning_rate": 0.0003768957181485565,
      "loss": 2.1575,
      "step": 96191
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.8356277346611023,
      "learning_rate": 0.00037689176423579647,
      "loss": 2.0323,
      "step": 96192
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.812940239906311,
      "learning_rate": 0.00037688781030874096,
      "loss": 2.3513,
      "step": 96193
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.8332267999649048,
      "learning_rate": 0.00037688385636739047,
      "loss": 2.2019,
      "step": 96194
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.8427032232284546,
      "learning_rate": 0.0003768799024117458,
      "loss": 2.3091,
      "step": 96195
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.8811721205711365,
      "learning_rate": 0.00037687594844180777,
      "loss": 2.1272,
      "step": 96196
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.9077767729759216,
      "learning_rate": 0.00037687199445757697,
      "loss": 2.246,
      "step": 96197
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.8003485798835754,
      "learning_rate": 0.0003768680404590543,
      "loss": 2.0942,
      "step": 96198
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.83024001121521,
      "learning_rate": 0.00037686408644624033,
      "loss": 2.1212,
      "step": 96199
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.8193426132202148,
      "learning_rate": 0.0003768601324191359,
      "loss": 2.082,
      "step": 96200
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.8279872536659241,
      "learning_rate": 0.00037685617837774173,
      "loss": 2.421,
      "step": 96201
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.8085801005363464,
      "learning_rate": 0.00037685222432205854,
      "loss": 2.1656,
      "step": 96202
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.8314194679260254,
      "learning_rate": 0.00037684827025208703,
      "loss": 2.069,
      "step": 96203
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.9743257761001587,
      "learning_rate": 0.00037684431616782807,
      "loss": 2.1109,
      "step": 96204
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.9449266791343689,
      "learning_rate": 0.0003768403620692823,
      "loss": 2.3467,
      "step": 96205
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.8836844563484192,
      "learning_rate": 0.0003768364079564504,
      "loss": 2.1678,
      "step": 96206
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.896251916885376,
      "learning_rate": 0.00037683245382933307,
      "loss": 2.3982,
      "step": 96207
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.8543868660926819,
      "learning_rate": 0.00037682849968793134,
      "loss": 2.1879,
      "step": 96208
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.8924111723899841,
      "learning_rate": 0.00037682454553224557,
      "loss": 2.3164,
      "step": 96209
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.7913429737091064,
      "learning_rate": 0.0003768205913622768,
      "loss": 2.1578,
      "step": 96210
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.934834897518158,
      "learning_rate": 0.00037681663717802554,
      "loss": 2.21,
      "step": 96211
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.9141948223114014,
      "learning_rate": 0.0003768126829794927,
      "loss": 2.1004,
      "step": 96212
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.9405149817466736,
      "learning_rate": 0.0003768087287666788,
      "loss": 2.2866,
      "step": 96213
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.8472540974617004,
      "learning_rate": 0.0003768047745395848,
      "loss": 2.1243,
      "step": 96214
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.9316514730453491,
      "learning_rate": 0.0003768008202982113,
      "loss": 2.2569,
      "step": 96215
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.9242135286331177,
      "learning_rate": 0.00037679686604255916,
      "loss": 2.1264,
      "step": 96216
    },
    {
      "epoch": 1.25,
      "grad_norm": 1.1900150775909424,
      "learning_rate": 0.000376792911772629,
      "loss": 1.8186,
      "step": 96217
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.9669782519340515,
      "learning_rate": 0.00037678895748842153,
      "loss": 2.0388,
      "step": 96218
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.9965440630912781,
      "learning_rate": 0.0003767850031899375,
      "loss": 2.0707,
      "step": 96219
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.7946636080741882,
      "learning_rate": 0.00037678104887717783,
      "loss": 2.1558,
      "step": 96220
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.9599279165267944,
      "learning_rate": 0.00037677709455014307,
      "loss": 2.1012,
      "step": 96221
    },
    {
      "epoch": 1.25,
      "grad_norm": 1.0174975395202637,
      "learning_rate": 0.0003767731402088339,
      "loss": 2.0368,
      "step": 96222
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.7708531618118286,
      "learning_rate": 0.0003767691858532513,
      "loss": 2.3432,
      "step": 96223
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.8453661799430847,
      "learning_rate": 0.0003767652314833958,
      "loss": 2.138,
      "step": 96224
    },
    {
      "epoch": 1.25,
      "grad_norm": 1.0707206726074219,
      "learning_rate": 0.0003767612770992681,
      "loss": 2.2221,
      "step": 96225
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.7961606383323669,
      "learning_rate": 0.00037675732270086925,
      "loss": 1.9893,
      "step": 96226
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.8157179355621338,
      "learning_rate": 0.00037675336828819954,
      "loss": 2.1639,
      "step": 96227
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.8623899221420288,
      "learning_rate": 0.00037674941386126005,
      "loss": 2.0755,
      "step": 96228
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.9750533699989319,
      "learning_rate": 0.0003767454594200514,
      "loss": 2.3579,
      "step": 96229
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.887447714805603,
      "learning_rate": 0.00037674150496457427,
      "loss": 2.293,
      "step": 96230
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.8675890564918518,
      "learning_rate": 0.0003767375504948294,
      "loss": 2.2213,
      "step": 96231
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.9050207138061523,
      "learning_rate": 0.0003767335960108177,
      "loss": 2.1533,
      "step": 96232
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.8528733253479004,
      "learning_rate": 0.0003767296415125397,
      "loss": 2.2507,
      "step": 96233
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.8504395484924316,
      "learning_rate": 0.0003767256869999962,
      "loss": 2.2456,
      "step": 96234
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.9337859749794006,
      "learning_rate": 0.000376721732473188,
      "loss": 2.3321,
      "step": 96235
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.8296734094619751,
      "learning_rate": 0.00037671777793211574,
      "loss": 2.093,
      "step": 96236
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.876965343952179,
      "learning_rate": 0.00037671382337678015,
      "loss": 2.1325,
      "step": 96237
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.8889798521995544,
      "learning_rate": 0.00037670986880718204,
      "loss": 1.8417,
      "step": 96238
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.8624171614646912,
      "learning_rate": 0.0003767059142233222,
      "loss": 2.1951,
      "step": 96239
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.8525269031524658,
      "learning_rate": 0.0003767019596252012,
      "loss": 2.3565,
      "step": 96240
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.9079080820083618,
      "learning_rate": 0.00037669800501281993,
      "loss": 2.192,
      "step": 96241
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.9135522842407227,
      "learning_rate": 0.000376694050386179,
      "loss": 2.1195,
      "step": 96242
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.7778637409210205,
      "learning_rate": 0.0003766900957452792,
      "loss": 2.0192,
      "step": 96243
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.8308922648429871,
      "learning_rate": 0.0003766861410901213,
      "loss": 2.0825,
      "step": 96244
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.8641499876976013,
      "learning_rate": 0.00037668218642070584,
      "loss": 2.0345,
      "step": 96245
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.9271581172943115,
      "learning_rate": 0.00037667823173703394,
      "loss": 1.8836,
      "step": 96246
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.7750930786132812,
      "learning_rate": 0.000376674277039106,
      "loss": 2.2607,
      "step": 96247
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.906671404838562,
      "learning_rate": 0.00037667032232692287,
      "loss": 2.0411,
      "step": 96248
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.8098429441452026,
      "learning_rate": 0.0003766663676004852,
      "loss": 2.2847,
      "step": 96249
    },
    {
      "epoch": 1.25,
      "grad_norm": 1.100376844406128,
      "learning_rate": 0.00037666241285979403,
      "loss": 2.2037,
      "step": 96250
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.8219330310821533,
      "learning_rate": 0.00037665845810484967,
      "loss": 2.1698,
      "step": 96251
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.878038763999939,
      "learning_rate": 0.0003766545033356531,
      "loss": 2.3401,
      "step": 96252
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.8422019481658936,
      "learning_rate": 0.00037665054855220504,
      "loss": 2.0486,
      "step": 96253
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.811190664768219,
      "learning_rate": 0.0003766465937545062,
      "loss": 2.1696,
      "step": 96254
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.896919846534729,
      "learning_rate": 0.0003766426389425573,
      "loss": 2.2088,
      "step": 96255
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.9079989194869995,
      "learning_rate": 0.000376638684116359,
      "loss": 2.0549,
      "step": 96256
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.8609172701835632,
      "learning_rate": 0.00037663472927591227,
      "loss": 2.3669,
      "step": 96257
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.7919818758964539,
      "learning_rate": 0.0003766307744212176,
      "loss": 2.0639,
      "step": 96258
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.9240964651107788,
      "learning_rate": 0.0003766268195522759,
      "loss": 2.1768,
      "step": 96259
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.8683239817619324,
      "learning_rate": 0.0003766228646690878,
      "loss": 2.1692,
      "step": 96260
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.7581874132156372,
      "learning_rate": 0.000376618909771654,
      "loss": 2.4263,
      "step": 96261
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.8102935552597046,
      "learning_rate": 0.0003766149548599753,
      "loss": 2.2307,
      "step": 96262
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.8343861103057861,
      "learning_rate": 0.00037661099993405254,
      "loss": 2.297,
      "step": 96263
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.848587155342102,
      "learning_rate": 0.00037660704499388627,
      "loss": 2.1691,
      "step": 96264
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.8225795030593872,
      "learning_rate": 0.00037660309003947727,
      "loss": 2.1509,
      "step": 96265
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.8145701289176941,
      "learning_rate": 0.00037659913507082646,
      "loss": 2.1251,
      "step": 96266
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.8039411306381226,
      "learning_rate": 0.0003765951800879343,
      "loss": 2.1185,
      "step": 96267
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.8362880349159241,
      "learning_rate": 0.0003765912250908016,
      "loss": 2.1936,
      "step": 96268
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.8593350648880005,
      "learning_rate": 0.0003765872700794293,
      "loss": 2.1523,
      "step": 96269
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.9231131672859192,
      "learning_rate": 0.00037658331505381794,
      "loss": 2.2008,
      "step": 96270
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.8693766593933105,
      "learning_rate": 0.00037657936001396817,
      "loss": 2.152,
      "step": 96271
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.8498697280883789,
      "learning_rate": 0.000376575404959881,
      "loss": 2.2112,
      "step": 96272
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.8947733640670776,
      "learning_rate": 0.0003765714498915569,
      "loss": 2.1891,
      "step": 96273
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.8818089962005615,
      "learning_rate": 0.0003765674948089968,
      "loss": 2.2298,
      "step": 96274
    },
    {
      "epoch": 1.25,
      "grad_norm": 1.0252983570098877,
      "learning_rate": 0.00037656353971220137,
      "loss": 1.9956,
      "step": 96275
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.8757503032684326,
      "learning_rate": 0.00037655958460117123,
      "loss": 2.4351,
      "step": 96276
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.8401963114738464,
      "learning_rate": 0.0003765556294759073,
      "loss": 2.0881,
      "step": 96277
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.978790819644928,
      "learning_rate": 0.00037655167433641026,
      "loss": 2.1184,
      "step": 96278
    },
    {
      "epoch": 1.25,
      "grad_norm": 1.0375654697418213,
      "learning_rate": 0.0003765477191826808,
      "loss": 2.2008,
      "step": 96279
    },
    {
      "epoch": 1.25,
      "grad_norm": 1.0949393510818481,
      "learning_rate": 0.00037654376401471956,
      "loss": 2.0422,
      "step": 96280
    },
    {
      "epoch": 1.25,
      "grad_norm": 1.1394007205963135,
      "learning_rate": 0.00037653980883252755,
      "loss": 2.2597,
      "step": 96281
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.9064439535140991,
      "learning_rate": 0.0003765358536361053,
      "loss": 2.333,
      "step": 96282
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.9930917620658875,
      "learning_rate": 0.0003765318984254535,
      "loss": 2.1658,
      "step": 96283
    },
    {
      "epoch": 1.25,
      "grad_norm": 1.042407751083374,
      "learning_rate": 0.0003765279432005731,
      "loss": 2.1448,
      "step": 96284
    },
    {
      "epoch": 1.25,
      "grad_norm": 1.1270484924316406,
      "learning_rate": 0.00037652398796146467,
      "loss": 2.1589,
      "step": 96285
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.848129391670227,
      "learning_rate": 0.00037652003270812894,
      "loss": 2.21,
      "step": 96286
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.8650285601615906,
      "learning_rate": 0.00037651607744056673,
      "loss": 2.1806,
      "step": 96287
    },
    {
      "epoch": 1.25,
      "grad_norm": 1.006384015083313,
      "learning_rate": 0.0003765121221587787,
      "loss": 2.1807,
      "step": 96288
    },
    {
      "epoch": 1.25,
      "grad_norm": 1.1971839666366577,
      "learning_rate": 0.00037650816686276564,
      "loss": 1.9314,
      "step": 96289
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.8940063714981079,
      "learning_rate": 0.0003765042115525283,
      "loss": 2.3387,
      "step": 96290
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.9100732207298279,
      "learning_rate": 0.00037650025622806735,
      "loss": 2.1907,
      "step": 96291
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.9138044118881226,
      "learning_rate": 0.0003764963008893836,
      "loss": 1.9626,
      "step": 96292
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.9228641986846924,
      "learning_rate": 0.0003764923455364777,
      "loss": 2.3247,
      "step": 96293
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.8788984417915344,
      "learning_rate": 0.0003764883901693504,
      "loss": 2.1471,
      "step": 96294
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.9479967355728149,
      "learning_rate": 0.0003764844347880025,
      "loss": 2.0905,
      "step": 96295
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.7966241836547852,
      "learning_rate": 0.00037648047939243477,
      "loss": 2.2308,
      "step": 96296
    },
    {
      "epoch": 1.25,
      "grad_norm": 1.0453011989593506,
      "learning_rate": 0.0003764765239826478,
      "loss": 2.3722,
      "step": 96297
    },
    {
      "epoch": 1.25,
      "grad_norm": 1.2219899892807007,
      "learning_rate": 0.00037647256855864237,
      "loss": 2.3527,
      "step": 96298
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.8696079850196838,
      "learning_rate": 0.00037646861312041937,
      "loss": 2.4361,
      "step": 96299
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.9013124704360962,
      "learning_rate": 0.00037646465766797923,
      "loss": 2.0786,
      "step": 96300
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.95676589012146,
      "learning_rate": 0.00037646070220132303,
      "loss": 2.4554,
      "step": 96301
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.8939682245254517,
      "learning_rate": 0.00037645674672045126,
      "loss": 2.218,
      "step": 96302
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.9393090009689331,
      "learning_rate": 0.0003764527912253648,
      "loss": 1.9921,
      "step": 96303
    },
    {
      "epoch": 1.25,
      "grad_norm": 1.007493257522583,
      "learning_rate": 0.0003764488357160642,
      "loss": 2.1866,
      "step": 96304
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.8424314260482788,
      "learning_rate": 0.0003764448801925505,
      "loss": 2.1486,
      "step": 96305
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.8134247064590454,
      "learning_rate": 0.00037644092465482414,
      "loss": 2.1074,
      "step": 96306
    },
    {
      "epoch": 1.25,
      "grad_norm": 1.1022613048553467,
      "learning_rate": 0.0003764369691028859,
      "loss": 2.1882,
      "step": 96307
    },
    {
      "epoch": 1.25,
      "grad_norm": 1.0282701253890991,
      "learning_rate": 0.0003764330135367367,
      "loss": 1.9812,
      "step": 96308
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.9022017121315002,
      "learning_rate": 0.0003764290579563771,
      "loss": 2.2571,
      "step": 96309
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.8800550699234009,
      "learning_rate": 0.00037642510236180794,
      "loss": 1.9157,
      "step": 96310
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.840388834476471,
      "learning_rate": 0.00037642114675302997,
      "loss": 2.1124,
      "step": 96311
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.8708289861679077,
      "learning_rate": 0.00037641719113004374,
      "loss": 2.0793,
      "step": 96312
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.9688116312026978,
      "learning_rate": 0.00037641323549285013,
      "loss": 2.1371,
      "step": 96313
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.8496149182319641,
      "learning_rate": 0.0003764092798414499,
      "loss": 2.2411,
      "step": 96314
    },
    {
      "epoch": 1.25,
      "grad_norm": 1.1025644540786743,
      "learning_rate": 0.0003764053241758437,
      "loss": 2.3129,
      "step": 96315
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.9191195368766785,
      "learning_rate": 0.0003764013684960323,
      "loss": 2.3016,
      "step": 96316
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.8321658372879028,
      "learning_rate": 0.0003763974128020166,
      "loss": 2.0831,
      "step": 96317
    },
    {
      "epoch": 1.25,
      "grad_norm": 1.1177862882614136,
      "learning_rate": 0.000376393457093797,
      "loss": 2.2337,
      "step": 96318
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.8536747694015503,
      "learning_rate": 0.0003763895013713744,
      "loss": 1.9237,
      "step": 96319
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.8061145544052124,
      "learning_rate": 0.00037638554563474964,
      "loss": 2.183,
      "step": 96320
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.904472291469574,
      "learning_rate": 0.0003763815898839234,
      "loss": 2.3277,
      "step": 96321
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.8360368013381958,
      "learning_rate": 0.00037637763411889623,
      "loss": 2.1896,
      "step": 96322
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.8908101916313171,
      "learning_rate": 0.00037637367833966915,
      "loss": 2.2895,
      "step": 96323
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.8295714259147644,
      "learning_rate": 0.00037636972254624277,
      "loss": 2.0956,
      "step": 96324
    },
    {
      "epoch": 1.25,
      "grad_norm": 1.0344220399856567,
      "learning_rate": 0.0003763657667386178,
      "loss": 2.3526,
      "step": 96325
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.9486791491508484,
      "learning_rate": 0.0003763618109167949,
      "loss": 2.2218,
      "step": 96326
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.7963657975196838,
      "learning_rate": 0.000376357855080775,
      "loss": 2.1392,
      "step": 96327
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.8836432695388794,
      "learning_rate": 0.0003763538992305587,
      "loss": 2.2065,
      "step": 96328
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.8645151257514954,
      "learning_rate": 0.00037634994336614675,
      "loss": 2.2364,
      "step": 96329
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.8227644562721252,
      "learning_rate": 0.00037634598748753986,
      "loss": 1.9625,
      "step": 96330
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.7796361446380615,
      "learning_rate": 0.0003763420315947389,
      "loss": 2.07,
      "step": 96331
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.8526892066001892,
      "learning_rate": 0.00037633807568774447,
      "loss": 2.1119,
      "step": 96332
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.9149073958396912,
      "learning_rate": 0.0003763341197665574,
      "loss": 2.2152,
      "step": 96333
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.9145228266716003,
      "learning_rate": 0.00037633016383117836,
      "loss": 2.2125,
      "step": 96334
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.8895462155342102,
      "learning_rate": 0.00037632620788160805,
      "loss": 2.0582,
      "step": 96335
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.8548429012298584,
      "learning_rate": 0.00037632225191784727,
      "loss": 2.3169,
      "step": 96336
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.9948709607124329,
      "learning_rate": 0.0003763182959398967,
      "loss": 2.2037,
      "step": 96337
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.807659387588501,
      "learning_rate": 0.00037631433994775727,
      "loss": 2.2566,
      "step": 96338
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.797079861164093,
      "learning_rate": 0.00037631038394142946,
      "loss": 2.2107,
      "step": 96339
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.8175395131111145,
      "learning_rate": 0.0003763064279209141,
      "loss": 2.3665,
      "step": 96340
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.8699837923049927,
      "learning_rate": 0.0003763024718862119,
      "loss": 2.221,
      "step": 96341
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.8737620115280151,
      "learning_rate": 0.0003762985158373237,
      "loss": 2.1803,
      "step": 96342
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.8182474374771118,
      "learning_rate": 0.0003762945597742502,
      "loss": 2.1438,
      "step": 96343
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.874687671661377,
      "learning_rate": 0.000376290603696992,
      "loss": 2.359,
      "step": 96344
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.8381779789924622,
      "learning_rate": 0.00037628664760555006,
      "loss": 2.2289,
      "step": 96345
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.9028302431106567,
      "learning_rate": 0.0003762826914999249,
      "loss": 2.137,
      "step": 96346
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.8800027966499329,
      "learning_rate": 0.00037627873538011733,
      "loss": 2.1466,
      "step": 96347
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.8430805206298828,
      "learning_rate": 0.0003762747792461282,
      "loss": 2.232,
      "step": 96348
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.8485854268074036,
      "learning_rate": 0.00037627082309795806,
      "loss": 2.2548,
      "step": 96349
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.8746171593666077,
      "learning_rate": 0.00037626686693560777,
      "loss": 2.3794,
      "step": 96350
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.847331166267395,
      "learning_rate": 0.00037626291075907806,
      "loss": 2.2963,
      "step": 96351
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.883009672164917,
      "learning_rate": 0.00037625895456836964,
      "loss": 2.0862,
      "step": 96352
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.8041309714317322,
      "learning_rate": 0.00037625499836348316,
      "loss": 2.2284,
      "step": 96353
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.8551996350288391,
      "learning_rate": 0.00037625104214441955,
      "loss": 2.3693,
      "step": 96354
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.8305766582489014,
      "learning_rate": 0.00037624708591117934,
      "loss": 2.1477,
      "step": 96355
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.773190438747406,
      "learning_rate": 0.0003762431296637633,
      "loss": 2.0145,
      "step": 96356
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.81584632396698,
      "learning_rate": 0.0003762391734021724,
      "loss": 2.0252,
      "step": 96357
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.8930182456970215,
      "learning_rate": 0.00037623521712640707,
      "loss": 2.2183,
      "step": 96358
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.7877891659736633,
      "learning_rate": 0.0003762312608364682,
      "loss": 2.3115,
      "step": 96359
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.8145621418952942,
      "learning_rate": 0.00037622730453235655,
      "loss": 2.0452,
      "step": 96360
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.7686071991920471,
      "learning_rate": 0.0003762233482140727,
      "loss": 2.1307,
      "step": 96361
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.8467879891395569,
      "learning_rate": 0.0003762193918816175,
      "loss": 2.2446,
      "step": 96362
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.8489752411842346,
      "learning_rate": 0.0003762154355349918,
      "loss": 2.2442,
      "step": 96363
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.8017669320106506,
      "learning_rate": 0.0003762114791741961,
      "loss": 2.1836,
      "step": 96364
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.9038551449775696,
      "learning_rate": 0.00037620752279923125,
      "loss": 2.2508,
      "step": 96365
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.834686815738678,
      "learning_rate": 0.00037620356641009805,
      "loss": 1.9645,
      "step": 96366
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.8349443674087524,
      "learning_rate": 0.00037619961000679716,
      "loss": 2.267,
      "step": 96367
    },
    {
      "epoch": 1.25,
      "grad_norm": 1.0202062129974365,
      "learning_rate": 0.00037619565358932926,
      "loss": 2.1559,
      "step": 96368
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.7659645080566406,
      "learning_rate": 0.0003761916971576953,
      "loss": 2.226,
      "step": 96369
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.9418690800666809,
      "learning_rate": 0.0003761877407118957,
      "loss": 2.2363,
      "step": 96370
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.9481667280197144,
      "learning_rate": 0.0003761837842519314,
      "loss": 2.2039,
      "step": 96371
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.8314533829689026,
      "learning_rate": 0.00037617982777780315,
      "loss": 2.2129,
      "step": 96372
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.8963268399238586,
      "learning_rate": 0.00037617587128951154,
      "loss": 2.2523,
      "step": 96373
    },
    {
      "epoch": 1.25,
      "grad_norm": 1.011939525604248,
      "learning_rate": 0.0003761719147870575,
      "loss": 2.0421,
      "step": 96374
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.8460261821746826,
      "learning_rate": 0.00037616795827044163,
      "loss": 1.9328,
      "step": 96375
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.8673396706581116,
      "learning_rate": 0.0003761640017396647,
      "loss": 2.1386,
      "step": 96376
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.8137896656990051,
      "learning_rate": 0.0003761600451947274,
      "loss": 2.1871,
      "step": 96377
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.8457430601119995,
      "learning_rate": 0.00037615608863563057,
      "loss": 1.9895,
      "step": 96378
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.8426579236984253,
      "learning_rate": 0.0003761521320623748,
      "loss": 2.058,
      "step": 96379
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.9978563189506531,
      "learning_rate": 0.000376148175474961,
      "loss": 2.4504,
      "step": 96380
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.8251218199729919,
      "learning_rate": 0.0003761442188733898,
      "loss": 2.1956,
      "step": 96381
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.8798932433128357,
      "learning_rate": 0.0003761402622576619,
      "loss": 2.1836,
      "step": 96382
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.7775979042053223,
      "learning_rate": 0.0003761363056277781,
      "loss": 2.4388,
      "step": 96383
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.8710717558860779,
      "learning_rate": 0.00037613234898373926,
      "loss": 2.1309,
      "step": 96384
    },
    {
      "epoch": 1.25,
      "grad_norm": 0.8499701023101807,
      "learning_rate": 0.0003761283923255458,
      "loss": 2.3892,
      "step": 96385
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.8805164694786072,
      "learning_rate": 0.0003761244356531987,
      "loss": 2.2338,
      "step": 96386
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.9059950709342957,
      "learning_rate": 0.00037612047896669863,
      "loss": 2.3912,
      "step": 96387
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.8215526938438416,
      "learning_rate": 0.00037611652226604634,
      "loss": 2.0982,
      "step": 96388
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.9294642806053162,
      "learning_rate": 0.0003761125655512425,
      "loss": 2.0472,
      "step": 96389
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.9263695478439331,
      "learning_rate": 0.00037610860882228794,
      "loss": 2.227,
      "step": 96390
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.8756455779075623,
      "learning_rate": 0.0003761046520791834,
      "loss": 2.3926,
      "step": 96391
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.7695773839950562,
      "learning_rate": 0.00037610069532192946,
      "loss": 1.9427,
      "step": 96392
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.9179719686508179,
      "learning_rate": 0.000376096738550527,
      "loss": 2.0729,
      "step": 96393
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.8183976411819458,
      "learning_rate": 0.0003760927817649768,
      "loss": 2.0275,
      "step": 96394
    },
    {
      "epoch": 1.26,
      "grad_norm": 1.0122363567352295,
      "learning_rate": 0.0003760888249652795,
      "loss": 2.1138,
      "step": 96395
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.8425448536872864,
      "learning_rate": 0.0003760848681514357,
      "loss": 1.9087,
      "step": 96396
    },
    {
      "epoch": 1.26,
      "grad_norm": 1.3487628698349,
      "learning_rate": 0.0003760809113234465,
      "loss": 2.1549,
      "step": 96397
    },
    {
      "epoch": 1.26,
      "grad_norm": 1.044069528579712,
      "learning_rate": 0.0003760769544813123,
      "loss": 1.8525,
      "step": 96398
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.7928313612937927,
      "learning_rate": 0.00037607299762503393,
      "loss": 2.0221,
      "step": 96399
    },
    {
      "epoch": 1.26,
      "grad_norm": 1.5617281198501587,
      "learning_rate": 0.0003760690407546123,
      "loss": 2.1465,
      "step": 96400
    },
    {
      "epoch": 1.26,
      "grad_norm": 1.3261955976486206,
      "learning_rate": 0.00037606508387004787,
      "loss": 2.1541,
      "step": 96401
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.859714925289154,
      "learning_rate": 0.0003760611269713415,
      "loss": 2.0424,
      "step": 96402
    },
    {
      "epoch": 1.26,
      "grad_norm": 1.2041279077529907,
      "learning_rate": 0.000376057170058494,
      "loss": 2.4581,
      "step": 96403
    },
    {
      "epoch": 1.26,
      "grad_norm": 1.0633989572525024,
      "learning_rate": 0.000376053213131506,
      "loss": 2.281,
      "step": 96404
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.9028322696685791,
      "learning_rate": 0.0003760492561903782,
      "loss": 2.0986,
      "step": 96405
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.9649943709373474,
      "learning_rate": 0.0003760452992351115,
      "loss": 2.0056,
      "step": 96406
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.9817235469818115,
      "learning_rate": 0.0003760413422657066,
      "loss": 1.9755,
      "step": 96407
    },
    {
      "epoch": 1.26,
      "grad_norm": 1.0772875547409058,
      "learning_rate": 0.000376037385282164,
      "loss": 2.3197,
      "step": 96408
    },
    {
      "epoch": 1.26,
      "grad_norm": 1.1565979719161987,
      "learning_rate": 0.0003760334282844848,
      "loss": 2.0738,
      "step": 96409
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.9258221387863159,
      "learning_rate": 0.0003760294712726695,
      "loss": 2.2533,
      "step": 96410
    },
    {
      "epoch": 1.26,
      "grad_norm": 1.0028338432312012,
      "learning_rate": 0.00037602551424671877,
      "loss": 2.0511,
      "step": 96411
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.9384626150131226,
      "learning_rate": 0.0003760215572066336,
      "loss": 2.1316,
      "step": 96412
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.8994836807250977,
      "learning_rate": 0.00037601760015241456,
      "loss": 2.2295,
      "step": 96413
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.7872065901756287,
      "learning_rate": 0.00037601364308406237,
      "loss": 2.2427,
      "step": 96414
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.9078750610351562,
      "learning_rate": 0.0003760096860015779,
      "loss": 2.1689,
      "step": 96415
    },
    {
      "epoch": 1.26,
      "grad_norm": 1.2391023635864258,
      "learning_rate": 0.0003760057289049616,
      "loss": 2.1747,
      "step": 96416
    },
    {
      "epoch": 1.26,
      "grad_norm": 1.1167231798171997,
      "learning_rate": 0.0003760017717942145,
      "loss": 2.1146,
      "step": 96417
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.8259081840515137,
      "learning_rate": 0.0003759978146693373,
      "loss": 2.1161,
      "step": 96418
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.8174542784690857,
      "learning_rate": 0.00037599385753033063,
      "loss": 2.1804,
      "step": 96419
    },
    {
      "epoch": 1.26,
      "grad_norm": 1.188466191291809,
      "learning_rate": 0.00037598990037719524,
      "loss": 2.0714,
      "step": 96420
    },
    {
      "epoch": 1.26,
      "grad_norm": 1.1315003633499146,
      "learning_rate": 0.00037598594320993194,
      "loss": 2.0139,
      "step": 96421
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.8258755803108215,
      "learning_rate": 0.00037598198602854137,
      "loss": 2.2377,
      "step": 96422
    },
    {
      "epoch": 1.26,
      "grad_norm": 1.0690561532974243,
      "learning_rate": 0.00037597802883302436,
      "loss": 2.1157,
      "step": 96423
    },
    {
      "epoch": 1.26,
      "grad_norm": 1.0719540119171143,
      "learning_rate": 0.0003759740716233816,
      "loss": 2.0917,
      "step": 96424
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.9770822525024414,
      "learning_rate": 0.0003759701143996138,
      "loss": 1.917,
      "step": 96425
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.9982665777206421,
      "learning_rate": 0.0003759661571617217,
      "loss": 2.2412,
      "step": 96426
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.9204577207565308,
      "learning_rate": 0.0003759621999097061,
      "loss": 2.2615,
      "step": 96427
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.9567046761512756,
      "learning_rate": 0.0003759582426435677,
      "loss": 2.0723,
      "step": 96428
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.8691510558128357,
      "learning_rate": 0.00037595428536330716,
      "loss": 1.9478,
      "step": 96429
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.9543552398681641,
      "learning_rate": 0.00037595032806892544,
      "loss": 2.1442,
      "step": 96430
    },
    {
      "epoch": 1.26,
      "grad_norm": 1.0593838691711426,
      "learning_rate": 0.00037594637076042295,
      "loss": 1.8038,
      "step": 96431
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.8388186097145081,
      "learning_rate": 0.0003759424134378006,
      "loss": 2.253,
      "step": 96432
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.9925191402435303,
      "learning_rate": 0.00037593845610105924,
      "loss": 2.0761,
      "step": 96433
    },
    {
      "epoch": 1.26,
      "grad_norm": 1.0598974227905273,
      "learning_rate": 0.0003759344987501994,
      "loss": 2.2068,
      "step": 96434
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.9528648257255554,
      "learning_rate": 0.00037593054138522193,
      "loss": 1.9702,
      "step": 96435
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.8764578104019165,
      "learning_rate": 0.00037592658400612764,
      "loss": 2.403,
      "step": 96436
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.8643307685852051,
      "learning_rate": 0.00037592262661291704,
      "loss": 2.2179,
      "step": 96437
    },
    {
      "epoch": 1.26,
      "grad_norm": 1.09445059299469,
      "learning_rate": 0.00037591866920559094,
      "loss": 2.1849,
      "step": 96438
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.8235682845115662,
      "learning_rate": 0.00037591471178415026,
      "loss": 2.1043,
      "step": 96439
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.9231974482536316,
      "learning_rate": 0.00037591075434859555,
      "loss": 2.2243,
      "step": 96440
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.9596215486526489,
      "learning_rate": 0.00037590679689892756,
      "loss": 2.1121,
      "step": 96441
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.9063663482666016,
      "learning_rate": 0.00037590283943514716,
      "loss": 2.218,
      "step": 96442
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.9174761176109314,
      "learning_rate": 0.0003758988819572549,
      "loss": 2.1776,
      "step": 96443
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.8462967872619629,
      "learning_rate": 0.00037589492446525163,
      "loss": 2.2474,
      "step": 96444
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.8881447315216064,
      "learning_rate": 0.0003758909669591381,
      "loss": 2.0634,
      "step": 96445
    },
    {
      "epoch": 1.26,
      "grad_norm": 1.2177326679229736,
      "learning_rate": 0.0003758870094389149,
      "loss": 2.1171,
      "step": 96446
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.9122373461723328,
      "learning_rate": 0.00037588305190458293,
      "loss": 2.0136,
      "step": 96447
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.8597570061683655,
      "learning_rate": 0.000375879094356143,
      "loss": 2.4034,
      "step": 96448
    },
    {
      "epoch": 1.26,
      "grad_norm": 1.1105012893676758,
      "learning_rate": 0.00037587513679359556,
      "loss": 2.1946,
      "step": 96449
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.9219031929969788,
      "learning_rate": 0.0003758711792169415,
      "loss": 1.9625,
      "step": 96450
    },
    {
      "epoch": 1.26,
      "grad_norm": 1.0060495138168335,
      "learning_rate": 0.0003758672216261816,
      "loss": 2.1669,
      "step": 96451
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.8543128371238708,
      "learning_rate": 0.00037586326402131657,
      "loss": 2.2852,
      "step": 96452
    },
    {
      "epoch": 1.26,
      "grad_norm": 1.171367883682251,
      "learning_rate": 0.0003758593064023471,
      "loss": 2.2108,
      "step": 96453
    },
    {
      "epoch": 1.26,
      "grad_norm": 1.106946587562561,
      "learning_rate": 0.0003758553487692739,
      "loss": 2.2668,
      "step": 96454
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.9376106262207031,
      "learning_rate": 0.00037585139112209783,
      "loss": 2.4514,
      "step": 96455
    },
    {
      "epoch": 1.26,
      "grad_norm": 1.0964967012405396,
      "learning_rate": 0.00037584743346081955,
      "loss": 2.1909,
      "step": 96456
    },
    {
      "epoch": 1.26,
      "grad_norm": 1.1579537391662598,
      "learning_rate": 0.00037584347578543975,
      "loss": 2.3671,
      "step": 96457
    },
    {
      "epoch": 1.26,
      "grad_norm": 1.064667820930481,
      "learning_rate": 0.00037583951809595925,
      "loss": 2.2947,
      "step": 96458
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.8964782953262329,
      "learning_rate": 0.00037583556039237875,
      "loss": 2.1735,
      "step": 96459
    },
    {
      "epoch": 1.26,
      "grad_norm": 1.112353801727295,
      "learning_rate": 0.000375831602674699,
      "loss": 1.9521,
      "step": 96460
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.9718388915061951,
      "learning_rate": 0.0003758276449429207,
      "loss": 2.2809,
      "step": 96461
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.91997230052948,
      "learning_rate": 0.0003758236871970446,
      "loss": 2.1777,
      "step": 96462
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.9473426342010498,
      "learning_rate": 0.00037581972943707156,
      "loss": 2.28,
      "step": 96463
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.811270534992218,
      "learning_rate": 0.00037581577166300215,
      "loss": 2.0721,
      "step": 96464
    },
    {
      "epoch": 1.26,
      "grad_norm": 1.052682638168335,
      "learning_rate": 0.0003758118138748371,
      "loss": 2.3949,
      "step": 96465
    },
    {
      "epoch": 1.26,
      "grad_norm": 1.0272234678268433,
      "learning_rate": 0.00037580785607257717,
      "loss": 2.1556,
      "step": 96466
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.831100583076477,
      "learning_rate": 0.00037580389825622317,
      "loss": 2.3449,
      "step": 96467
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.8247830867767334,
      "learning_rate": 0.0003757999404257758,
      "loss": 2.1848,
      "step": 96468
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.8413463830947876,
      "learning_rate": 0.00037579598258123583,
      "loss": 2.3233,
      "step": 96469
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.9571269154548645,
      "learning_rate": 0.0003757920247226039,
      "loss": 2.041,
      "step": 96470
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.8993868231773376,
      "learning_rate": 0.00037578806684988083,
      "loss": 2.2125,
      "step": 96471
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.764923632144928,
      "learning_rate": 0.0003757841089630673,
      "loss": 1.8836,
      "step": 96472
    },
    {
      "epoch": 1.26,
      "grad_norm": 1.1398155689239502,
      "learning_rate": 0.0003757801510621641,
      "loss": 2.181,
      "step": 96473
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.81600421667099,
      "learning_rate": 0.000375776193147172,
      "loss": 2.1374,
      "step": 96474
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.8950331211090088,
      "learning_rate": 0.00037577223521809155,
      "loss": 2.2621,
      "step": 96475
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.960448682308197,
      "learning_rate": 0.0003757682772749237,
      "loss": 2.1208,
      "step": 96476
    },
    {
      "epoch": 1.26,
      "grad_norm": 1.0195006132125854,
      "learning_rate": 0.00037576431931766904,
      "loss": 2.0543,
      "step": 96477
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.8794830441474915,
      "learning_rate": 0.00037576036134632834,
      "loss": 1.9812,
      "step": 96478
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.9217759966850281,
      "learning_rate": 0.00037575640336090255,
      "loss": 2.0025,
      "step": 96479
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.918430507183075,
      "learning_rate": 0.000375752445361392,
      "loss": 2.3173,
      "step": 96480
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.9305789470672607,
      "learning_rate": 0.00037574848734779766,
      "loss": 2.0318,
      "step": 96481
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.8169556260108948,
      "learning_rate": 0.00037574452932012035,
      "loss": 2.416,
      "step": 96482
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.886657178401947,
      "learning_rate": 0.0003757405712783606,
      "loss": 2.0904,
      "step": 96483
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.9978962540626526,
      "learning_rate": 0.00037573661322251936,
      "loss": 2.1109,
      "step": 96484
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.8973181247711182,
      "learning_rate": 0.0003757326551525972,
      "loss": 2.236,
      "step": 96485
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.8803990483283997,
      "learning_rate": 0.00037572869706859494,
      "loss": 2.4555,
      "step": 96486
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.9093903303146362,
      "learning_rate": 0.0003757247389705132,
      "loss": 2.2211,
      "step": 96487
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.9807648062705994,
      "learning_rate": 0.0003757207808583529,
      "loss": 2.2025,
      "step": 96488
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.8806191682815552,
      "learning_rate": 0.00037571682273211454,
      "loss": 2.2914,
      "step": 96489
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.8174343109130859,
      "learning_rate": 0.00037571286459179914,
      "loss": 2.4309,
      "step": 96490
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.8147807121276855,
      "learning_rate": 0.00037570890643740727,
      "loss": 2.1748,
      "step": 96491
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.9024645090103149,
      "learning_rate": 0.00037570494826893964,
      "loss": 2.033,
      "step": 96492
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.9595741629600525,
      "learning_rate": 0.000375700990086397,
      "loss": 2.1741,
      "step": 96493
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.9062941074371338,
      "learning_rate": 0.00037569703188978013,
      "loss": 2.1632,
      "step": 96494
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.7953348159790039,
      "learning_rate": 0.0003756930736790898,
      "loss": 2.2298,
      "step": 96495
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.9001317620277405,
      "learning_rate": 0.00037568911545432667,
      "loss": 2.2532,
      "step": 96496
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.8826402425765991,
      "learning_rate": 0.00037568515721549155,
      "loss": 2.1724,
      "step": 96497
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.8313009142875671,
      "learning_rate": 0.0003756811989625851,
      "loss": 2.0319,
      "step": 96498
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.8097653388977051,
      "learning_rate": 0.00037567724069560807,
      "loss": 2.1429,
      "step": 96499
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.8970871567726135,
      "learning_rate": 0.0003756732824145612,
      "loss": 2.3575,
      "step": 96500
    },
    {
      "epoch": 1.26,
      "grad_norm": 1.0421186685562134,
      "learning_rate": 0.00037566932411944523,
      "loss": 2.273,
      "step": 96501
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.9261997938156128,
      "learning_rate": 0.00037566536581026095,
      "loss": 2.3756,
      "step": 96502
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.9582251906394958,
      "learning_rate": 0.00037566140748700904,
      "loss": 2.1311,
      "step": 96503
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.9258038997650146,
      "learning_rate": 0.00037565744914969026,
      "loss": 2.1972,
      "step": 96504
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.9430882930755615,
      "learning_rate": 0.0003756534907983053,
      "loss": 2.1817,
      "step": 96505
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.8193619251251221,
      "learning_rate": 0.0003756495324328549,
      "loss": 2.1679,
      "step": 96506
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.9384127259254456,
      "learning_rate": 0.00037564557405333994,
      "loss": 2.0662,
      "step": 96507
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.9351166486740112,
      "learning_rate": 0.00037564161565976093,
      "loss": 2.206,
      "step": 96508
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.9259543418884277,
      "learning_rate": 0.00037563765725211873,
      "loss": 2.1902,
      "step": 96509
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.7994047403335571,
      "learning_rate": 0.00037563369883041406,
      "loss": 2.0914,
      "step": 96510
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.8703056573867798,
      "learning_rate": 0.0003756297403946477,
      "loss": 2.3172,
      "step": 96511
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.9149304628372192,
      "learning_rate": 0.0003756257819448203,
      "loss": 2.1694,
      "step": 96512
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.825097382068634,
      "learning_rate": 0.00037562182348093267,
      "loss": 2.1752,
      "step": 96513
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.8310514092445374,
      "learning_rate": 0.00037561786500298555,
      "loss": 2.3269,
      "step": 96514
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.9745118021965027,
      "learning_rate": 0.0003756139065109796,
      "loss": 2.2735,
      "step": 96515
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.9004051685333252,
      "learning_rate": 0.0003756099480049155,
      "loss": 2.1018,
      "step": 96516
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.8617162704467773,
      "learning_rate": 0.00037560598948479417,
      "loss": 2.104,
      "step": 96517
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.8693459033966064,
      "learning_rate": 0.00037560203095061636,
      "loss": 2.1373,
      "step": 96518
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.7833705544471741,
      "learning_rate": 0.0003755980724023826,
      "loss": 2.0424,
      "step": 96519
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.9656617641448975,
      "learning_rate": 0.0003755941138400937,
      "loss": 2.1404,
      "step": 96520
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.8167803287506104,
      "learning_rate": 0.0003755901552637505,
      "loss": 2.0934,
      "step": 96521
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.8570382595062256,
      "learning_rate": 0.0003755861966733536,
      "loss": 2.2193,
      "step": 96522
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.8443797826766968,
      "learning_rate": 0.00037558223806890385,
      "loss": 2.2746,
      "step": 96523
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.8035374283790588,
      "learning_rate": 0.0003755782794504019,
      "loss": 2.1933,
      "step": 96524
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.892241895198822,
      "learning_rate": 0.00037557432081784857,
      "loss": 2.2586,
      "step": 96525
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.8355537056922913,
      "learning_rate": 0.0003755703621712445,
      "loss": 2.1097,
      "step": 96526
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.9687674045562744,
      "learning_rate": 0.0003755664035105905,
      "loss": 2.2414,
      "step": 96527
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.7878159880638123,
      "learning_rate": 0.00037556244483588733,
      "loss": 2.258,
      "step": 96528
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.7949642539024353,
      "learning_rate": 0.0003755584861471356,
      "loss": 2.0025,
      "step": 96529
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.9130784869194031,
      "learning_rate": 0.0003755545274443361,
      "loss": 2.1717,
      "step": 96530
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.9658219218254089,
      "learning_rate": 0.0003755505687274897,
      "loss": 2.223,
      "step": 96531
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.924266517162323,
      "learning_rate": 0.0003755466099965969,
      "loss": 2.2631,
      "step": 96532
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.8598567247390747,
      "learning_rate": 0.0003755426512516586,
      "loss": 2.2215,
      "step": 96533
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.797606885433197,
      "learning_rate": 0.0003755386924926755,
      "loss": 2.2503,
      "step": 96534
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.8920338749885559,
      "learning_rate": 0.00037553473371964837,
      "loss": 2.2723,
      "step": 96535
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.8632913827896118,
      "learning_rate": 0.0003755307749325778,
      "loss": 2.0642,
      "step": 96536
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.8106177449226379,
      "learning_rate": 0.00037552681613146476,
      "loss": 2.0984,
      "step": 96537
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.8031882643699646,
      "learning_rate": 0.0003755228573163098,
      "loss": 2.0825,
      "step": 96538
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.8535186052322388,
      "learning_rate": 0.0003755188984871137,
      "loss": 1.9908,
      "step": 96539
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.8581464290618896,
      "learning_rate": 0.00037551493964387727,
      "loss": 1.9943,
      "step": 96540
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.9307357668876648,
      "learning_rate": 0.0003755109807866011,
      "loss": 2.0848,
      "step": 96541
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.8826558589935303,
      "learning_rate": 0.00037550702191528607,
      "loss": 2.2167,
      "step": 96542
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.8893588781356812,
      "learning_rate": 0.00037550306302993285,
      "loss": 2.3695,
      "step": 96543
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.8187817931175232,
      "learning_rate": 0.00037549910413054225,
      "loss": 2.0596,
      "step": 96544
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.8353139162063599,
      "learning_rate": 0.0003754951452171148,
      "loss": 2.4402,
      "step": 96545
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.8800382614135742,
      "learning_rate": 0.0003754911862896515,
      "loss": 2.2255,
      "step": 96546
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.8734380006790161,
      "learning_rate": 0.0003754872273481529,
      "loss": 2.2737,
      "step": 96547
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.9127148389816284,
      "learning_rate": 0.0003754832683926198,
      "loss": 2.3147,
      "step": 96548
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.843626081943512,
      "learning_rate": 0.00037547930942305295,
      "loss": 2.1498,
      "step": 96549
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.8698298335075378,
      "learning_rate": 0.00037547535043945307,
      "loss": 2.3004,
      "step": 96550
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.8661239147186279,
      "learning_rate": 0.0003754713914418209,
      "loss": 2.1491,
      "step": 96551
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.8043383359909058,
      "learning_rate": 0.00037546743243015726,
      "loss": 2.0862,
      "step": 96552
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.939267635345459,
      "learning_rate": 0.00037546347340446264,
      "loss": 2.1541,
      "step": 96553
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.8307921886444092,
      "learning_rate": 0.00037545951436473806,
      "loss": 2.1498,
      "step": 96554
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.864504337310791,
      "learning_rate": 0.00037545555531098406,
      "loss": 2.225,
      "step": 96555
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.9507396817207336,
      "learning_rate": 0.00037545159624320147,
      "loss": 1.9633,
      "step": 96556
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.8101539015769958,
      "learning_rate": 0.0003754476371613909,
      "loss": 2.2509,
      "step": 96557
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.8389894366264343,
      "learning_rate": 0.0003754436780655534,
      "loss": 2.1992,
      "step": 96558
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.7777434587478638,
      "learning_rate": 0.00037543971895568934,
      "loss": 2.2271,
      "step": 96559
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.8713054060935974,
      "learning_rate": 0.0003754357598317997,
      "loss": 2.1338,
      "step": 96560
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.7400813102722168,
      "learning_rate": 0.00037543180069388513,
      "loss": 2.136,
      "step": 96561
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.8915568590164185,
      "learning_rate": 0.00037542784154194627,
      "loss": 2.3499,
      "step": 96562
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.8291895985603333,
      "learning_rate": 0.000375423882375984,
      "loss": 2.0654,
      "step": 96563
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.9813942909240723,
      "learning_rate": 0.0003754199231959991,
      "loss": 2.3217,
      "step": 96564
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.8560211062431335,
      "learning_rate": 0.00037541596400199204,
      "loss": 2.0619,
      "step": 96565
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.8619744181632996,
      "learning_rate": 0.0003754120047939638,
      "loss": 2.0506,
      "step": 96566
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.8000332117080688,
      "learning_rate": 0.00037540804557191515,
      "loss": 2.0393,
      "step": 96567
    },
    {
      "epoch": 1.26,
      "grad_norm": 1.1593692302703857,
      "learning_rate": 0.00037540408633584654,
      "loss": 2.0962,
      "step": 96568
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.9559175968170166,
      "learning_rate": 0.00037540012708575903,
      "loss": 2.1249,
      "step": 96569
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.8469840884208679,
      "learning_rate": 0.00037539616782165323,
      "loss": 2.206,
      "step": 96570
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.8888496160507202,
      "learning_rate": 0.0003753922085435297,
      "loss": 2.1014,
      "step": 96571
    },
    {
      "epoch": 1.26,
      "grad_norm": 1.2425915002822876,
      "learning_rate": 0.00037538824925138943,
      "loss": 2.3222,
      "step": 96572
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.8258742094039917,
      "learning_rate": 0.0003753842899452331,
      "loss": 2.2029,
      "step": 96573
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.8813108801841736,
      "learning_rate": 0.0003753803306250613,
      "loss": 2.4114,
      "step": 96574
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.9944722056388855,
      "learning_rate": 0.00037537637129087496,
      "loss": 2.3416,
      "step": 96575
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.9878983497619629,
      "learning_rate": 0.00037537241194267475,
      "loss": 2.1029,
      "step": 96576
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.862246572971344,
      "learning_rate": 0.00037536845258046134,
      "loss": 2.1412,
      "step": 96577
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.8366127610206604,
      "learning_rate": 0.0003753644932042355,
      "loss": 2.2054,
      "step": 96578
    },
    {
      "epoch": 1.26,
      "grad_norm": 1.0076758861541748,
      "learning_rate": 0.000375360533813998,
      "loss": 2.3944,
      "step": 96579
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.9189107418060303,
      "learning_rate": 0.00037535657440974956,
      "loss": 2.2137,
      "step": 96580
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.823030948638916,
      "learning_rate": 0.00037535261499149086,
      "loss": 2.0217,
      "step": 96581
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.9644820690155029,
      "learning_rate": 0.00037534865555922275,
      "loss": 2.1311,
      "step": 96582
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.8012136220932007,
      "learning_rate": 0.0003753446961129458,
      "loss": 2.1601,
      "step": 96583
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.9297817945480347,
      "learning_rate": 0.00037534073665266096,
      "loss": 2.1705,
      "step": 96584
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.8641180992126465,
      "learning_rate": 0.0003753367771783689,
      "loss": 2.1876,
      "step": 96585
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.933673083782196,
      "learning_rate": 0.0003753328176900702,
      "loss": 2.1607,
      "step": 96586
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.7974353432655334,
      "learning_rate": 0.0003753288581877657,
      "loss": 2.1184,
      "step": 96587
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.7896596789360046,
      "learning_rate": 0.00037532489867145617,
      "loss": 2.1313,
      "step": 96588
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.8295149207115173,
      "learning_rate": 0.0003753209391411424,
      "loss": 2.2853,
      "step": 96589
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.8798078894615173,
      "learning_rate": 0.0003753169795968249,
      "loss": 2.1899,
      "step": 96590
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.8385721445083618,
      "learning_rate": 0.0003753130200385046,
      "loss": 2.1882,
      "step": 96591
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.9662569761276245,
      "learning_rate": 0.0003753090604661823,
      "loss": 2.2562,
      "step": 96592
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.930033802986145,
      "learning_rate": 0.0003753051008798585,
      "loss": 2.2666,
      "step": 96593
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.9022141695022583,
      "learning_rate": 0.00037530114127953406,
      "loss": 2.2684,
      "step": 96594
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.9921855330467224,
      "learning_rate": 0.00037529718166520986,
      "loss": 2.2356,
      "step": 96595
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.8010918498039246,
      "learning_rate": 0.00037529322203688634,
      "loss": 2.1736,
      "step": 96596
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.9467045664787292,
      "learning_rate": 0.0003752892623945644,
      "loss": 2.3674,
      "step": 96597
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.9288710951805115,
      "learning_rate": 0.0003752853027382448,
      "loss": 2.3876,
      "step": 96598
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.8911312222480774,
      "learning_rate": 0.0003752813430679283,
      "loss": 2.274,
      "step": 96599
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.8221340775489807,
      "learning_rate": 0.0003752773833836155,
      "loss": 1.8783,
      "step": 96600
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.8390357494354248,
      "learning_rate": 0.0003752734236853072,
      "loss": 2.2157,
      "step": 96601
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.8948750495910645,
      "learning_rate": 0.00037526946397300413,
      "loss": 2.5092,
      "step": 96602
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.8693435192108154,
      "learning_rate": 0.00037526550424670717,
      "loss": 2.1659,
      "step": 96603
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.8750720024108887,
      "learning_rate": 0.00037526154450641684,
      "loss": 2.3265,
      "step": 96604
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.9804334044456482,
      "learning_rate": 0.00037525758475213396,
      "loss": 2.3597,
      "step": 96605
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.838379979133606,
      "learning_rate": 0.00037525362498385935,
      "loss": 2.0991,
      "step": 96606
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.8473598957061768,
      "learning_rate": 0.0003752496652015936,
      "loss": 2.1309,
      "step": 96607
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.8032165765762329,
      "learning_rate": 0.00037524570540533757,
      "loss": 2.2217,
      "step": 96608
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.8716276288032532,
      "learning_rate": 0.00037524174559509186,
      "loss": 2.3523,
      "step": 96609
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.798348069190979,
      "learning_rate": 0.00037523778577085734,
      "loss": 2.0714,
      "step": 96610
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.9780677556991577,
      "learning_rate": 0.0003752338259326347,
      "loss": 1.9787,
      "step": 96611
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.8346880078315735,
      "learning_rate": 0.0003752298660804247,
      "loss": 2.2051,
      "step": 96612
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.8658642768859863,
      "learning_rate": 0.00037522590621422803,
      "loss": 2.1033,
      "step": 96613
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.8198816776275635,
      "learning_rate": 0.0003752219463340455,
      "loss": 2.1907,
      "step": 96614
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.8632017374038696,
      "learning_rate": 0.00037521798643987765,
      "loss": 2.2863,
      "step": 96615
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.9150511026382446,
      "learning_rate": 0.00037521402653172545,
      "loss": 2.2087,
      "step": 96616
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.9100390076637268,
      "learning_rate": 0.00037521006660958954,
      "loss": 2.0591,
      "step": 96617
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.9047939777374268,
      "learning_rate": 0.00037520610667347057,
      "loss": 2.3809,
      "step": 96618
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.868261456489563,
      "learning_rate": 0.00037520214672336956,
      "loss": 2.2697,
      "step": 96619
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.8636639714241028,
      "learning_rate": 0.0003751981867592869,
      "loss": 1.9636,
      "step": 96620
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.8747545480728149,
      "learning_rate": 0.00037519422678122344,
      "loss": 1.8324,
      "step": 96621
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.9391133785247803,
      "learning_rate": 0.00037519026678918007,
      "loss": 2.2504,
      "step": 96622
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.8093376159667969,
      "learning_rate": 0.0003751863067831573,
      "loss": 2.0319,
      "step": 96623
    },
    {
      "epoch": 1.26,
      "grad_norm": 1.0603786706924438,
      "learning_rate": 0.000375182346763156,
      "loss": 2.3105,
      "step": 96624
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.8797305822372437,
      "learning_rate": 0.000375178386729177,
      "loss": 1.9581,
      "step": 96625
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.8261018991470337,
      "learning_rate": 0.0003751744266812208,
      "loss": 2.0602,
      "step": 96626
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.7388485074043274,
      "learning_rate": 0.00037517046661928827,
      "loss": 1.9316,
      "step": 96627
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.8522148132324219,
      "learning_rate": 0.00037516650654338023,
      "loss": 2.026,
      "step": 96628
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.8834466338157654,
      "learning_rate": 0.0003751625464534972,
      "loss": 2.2191,
      "step": 96629
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.9397525787353516,
      "learning_rate": 0.0003751585863496401,
      "loss": 2.2147,
      "step": 96630
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.8731273412704468,
      "learning_rate": 0.00037515462623180957,
      "loss": 2.2148,
      "step": 96631
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.8105534911155701,
      "learning_rate": 0.00037515066610000635,
      "loss": 2.0175,
      "step": 96632
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.9166232347488403,
      "learning_rate": 0.00037514670595423125,
      "loss": 2.3041,
      "step": 96633
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.8017518520355225,
      "learning_rate": 0.000375142745794485,
      "loss": 2.062,
      "step": 96634
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.9305739998817444,
      "learning_rate": 0.00037513878562076827,
      "loss": 2.1586,
      "step": 96635
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.8242538571357727,
      "learning_rate": 0.00037513482543308175,
      "loss": 2.021,
      "step": 96636
    },
    {
      "epoch": 1.26,
      "grad_norm": 1.0135084390640259,
      "learning_rate": 0.00037513086523142627,
      "loss": 1.9478,
      "step": 96637
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.8940662145614624,
      "learning_rate": 0.0003751269050158026,
      "loss": 1.9789,
      "step": 96638
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.9220470190048218,
      "learning_rate": 0.0003751229447862114,
      "loss": 2.0231,
      "step": 96639
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.8645339012145996,
      "learning_rate": 0.0003751189845426533,
      "loss": 2.1811,
      "step": 96640
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.8793497681617737,
      "learning_rate": 0.0003751150242851293,
      "loss": 2.1639,
      "step": 96641
    },
    {
      "epoch": 1.26,
      "grad_norm": 1.0891382694244385,
      "learning_rate": 0.00037511106401364,
      "loss": 2.2582,
      "step": 96642
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.8865762948989868,
      "learning_rate": 0.0003751071037281861,
      "loss": 2.3946,
      "step": 96643
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.8381751775741577,
      "learning_rate": 0.0003751031434287684,
      "loss": 2.1988,
      "step": 96644
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.9523918032646179,
      "learning_rate": 0.00037509918311538754,
      "loss": 2.0471,
      "step": 96645
    },
    {
      "epoch": 1.26,
      "grad_norm": 1.0560760498046875,
      "learning_rate": 0.0003750952227880444,
      "loss": 2.1232,
      "step": 96646
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.9088619947433472,
      "learning_rate": 0.00037509126244673956,
      "loss": 2.0491,
      "step": 96647
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.8488808274269104,
      "learning_rate": 0.00037508730209147393,
      "loss": 2.015,
      "step": 96648
    },
    {
      "epoch": 1.26,
      "grad_norm": 1.0014557838439941,
      "learning_rate": 0.00037508334172224817,
      "loss": 2.2698,
      "step": 96649
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.8276917338371277,
      "learning_rate": 0.00037507938133906287,
      "loss": 2.2377,
      "step": 96650
    },
    {
      "epoch": 1.26,
      "grad_norm": 1.0312166213989258,
      "learning_rate": 0.00037507542094191896,
      "loss": 2.0951,
      "step": 96651
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.9194272756576538,
      "learning_rate": 0.00037507146053081714,
      "loss": 2.285,
      "step": 96652
    },
    {
      "epoch": 1.26,
      "grad_norm": 1.1437511444091797,
      "learning_rate": 0.00037506750010575805,
      "loss": 2.4614,
      "step": 96653
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.8883817195892334,
      "learning_rate": 0.00037506353966674263,
      "loss": 2.2506,
      "step": 96654
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.7780577540397644,
      "learning_rate": 0.0003750595792137713,
      "loss": 1.9295,
      "step": 96655
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.7969009876251221,
      "learning_rate": 0.00037505561874684516,
      "loss": 2.1537,
      "step": 96656
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.9169773459434509,
      "learning_rate": 0.00037505165826596463,
      "loss": 2.0327,
      "step": 96657
    },
    {
      "epoch": 1.26,
      "grad_norm": 1.0878221988677979,
      "learning_rate": 0.00037504769777113063,
      "loss": 2.1915,
      "step": 96658
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.89473956823349,
      "learning_rate": 0.0003750437372623438,
      "loss": 2.1962,
      "step": 96659
    },
    {
      "epoch": 1.26,
      "grad_norm": 1.163198471069336,
      "learning_rate": 0.000375039776739605,
      "loss": 2.1536,
      "step": 96660
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.9778228402137756,
      "learning_rate": 0.0003750358162029148,
      "loss": 2.0347,
      "step": 96661
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.9888395667076111,
      "learning_rate": 0.00037503185565227414,
      "loss": 2.1917,
      "step": 96662
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.817810595035553,
      "learning_rate": 0.0003750278950876836,
      "loss": 2.1945,
      "step": 96663
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.9025707840919495,
      "learning_rate": 0.00037502393450914383,
      "loss": 2.1112,
      "step": 96664
    },
    {
      "epoch": 1.26,
      "grad_norm": 1.0681328773498535,
      "learning_rate": 0.00037501997391665585,
      "loss": 2.0074,
      "step": 96665
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.9319225549697876,
      "learning_rate": 0.0003750160133102202,
      "loss": 2.1295,
      "step": 96666
    },
    {
      "epoch": 1.26,
      "grad_norm": 1.030488133430481,
      "learning_rate": 0.0003750120526898376,
      "loss": 2.323,
      "step": 96667
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.8630404472351074,
      "learning_rate": 0.00037500809205550897,
      "loss": 2.2307,
      "step": 96668
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.8784899711608887,
      "learning_rate": 0.00037500413140723483,
      "loss": 2.2576,
      "step": 96669
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.8567376136779785,
      "learning_rate": 0.00037500017074501595,
      "loss": 2.2558,
      "step": 96670
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.8901110887527466,
      "learning_rate": 0.00037499621006885326,
      "loss": 2.2945,
      "step": 96671
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.8886313438415527,
      "learning_rate": 0.0003749922493787473,
      "loss": 2.3159,
      "step": 96672
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.8079471588134766,
      "learning_rate": 0.0003749882886746988,
      "loss": 2.0955,
      "step": 96673
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.87541264295578,
      "learning_rate": 0.00037498432795670864,
      "loss": 1.9685,
      "step": 96674
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.8633517026901245,
      "learning_rate": 0.0003749803672247775,
      "loss": 2.2013,
      "step": 96675
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.9568312764167786,
      "learning_rate": 0.00037497640647890594,
      "loss": 2.1794,
      "step": 96676
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.8794578909873962,
      "learning_rate": 0.00037497244571909504,
      "loss": 2.0885,
      "step": 96677
    },
    {
      "epoch": 1.26,
      "grad_norm": 1.039689302444458,
      "learning_rate": 0.0003749684849453452,
      "loss": 1.9507,
      "step": 96678
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.9262988567352295,
      "learning_rate": 0.0003749645241576574,
      "loss": 2.223,
      "step": 96679
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.9162020683288574,
      "learning_rate": 0.00037496056335603226,
      "loss": 2.3707,
      "step": 96680
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.9578669667243958,
      "learning_rate": 0.00037495660254047055,
      "loss": 2.2555,
      "step": 96681
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.9476731419563293,
      "learning_rate": 0.00037495264171097294,
      "loss": 2.0248,
      "step": 96682
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.9428699612617493,
      "learning_rate": 0.00037494868086754025,
      "loss": 2.1915,
      "step": 96683
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.8361914753913879,
      "learning_rate": 0.0003749447200101732,
      "loss": 2.0638,
      "step": 96684
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.9958751201629639,
      "learning_rate": 0.00037494075913887246,
      "loss": 2.0705,
      "step": 96685
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.8527840971946716,
      "learning_rate": 0.0003749367982536389,
      "loss": 2.0347,
      "step": 96686
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.8160186409950256,
      "learning_rate": 0.00037493283735447316,
      "loss": 2.1047,
      "step": 96687
    },
    {
      "epoch": 1.26,
      "grad_norm": 1.0751757621765137,
      "learning_rate": 0.0003749288764413759,
      "loss": 2.157,
      "step": 96688
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.7667233943939209,
      "learning_rate": 0.000374924915514348,
      "loss": 2.1151,
      "step": 96689
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.8968043923377991,
      "learning_rate": 0.0003749209545733901,
      "loss": 2.1203,
      "step": 96690
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.9493723511695862,
      "learning_rate": 0.0003749169936185031,
      "loss": 2.2982,
      "step": 96691
    },
    {
      "epoch": 1.26,
      "grad_norm": 1.008902668952942,
      "learning_rate": 0.00037491303264968747,
      "loss": 2.3792,
      "step": 96692
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.8798700571060181,
      "learning_rate": 0.0003749090716669442,
      "loss": 2.279,
      "step": 96693
    },
    {
      "epoch": 1.26,
      "grad_norm": 1.0464272499084473,
      "learning_rate": 0.00037490511067027386,
      "loss": 2.3731,
      "step": 96694
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.8221026062965393,
      "learning_rate": 0.0003749011496596773,
      "loss": 1.9103,
      "step": 96695
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.8851242065429688,
      "learning_rate": 0.0003748971886351552,
      "loss": 2.4121,
      "step": 96696
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.8655893802642822,
      "learning_rate": 0.0003748932275967082,
      "loss": 2.1712,
      "step": 96697
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.9398083090782166,
      "learning_rate": 0.0003748892665443373,
      "loss": 2.0643,
      "step": 96698
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.9111671447753906,
      "learning_rate": 0.00037488530547804297,
      "loss": 2.1624,
      "step": 96699
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.8602215051651001,
      "learning_rate": 0.00037488134439782604,
      "loss": 1.9352,
      "step": 96700
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.7965035438537598,
      "learning_rate": 0.00037487738330368726,
      "loss": 2.2731,
      "step": 96701
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.9701313376426697,
      "learning_rate": 0.0003748734221956274,
      "loss": 2.0592,
      "step": 96702
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.9329246282577515,
      "learning_rate": 0.0003748694610736471,
      "loss": 2.1411,
      "step": 96703
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.8857477307319641,
      "learning_rate": 0.00037486549993774723,
      "loss": 2.012,
      "step": 96704
    },
    {
      "epoch": 1.26,
      "grad_norm": 1.1059473752975464,
      "learning_rate": 0.0003748615387879284,
      "loss": 2.3829,
      "step": 96705
    },
    {
      "epoch": 1.26,
      "grad_norm": 1.0457152128219604,
      "learning_rate": 0.00037485757762419135,
      "loss": 2.2407,
      "step": 96706
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.7930549383163452,
      "learning_rate": 0.000374853616446537,
      "loss": 2.2445,
      "step": 96707
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.9337067604064941,
      "learning_rate": 0.00037484965525496586,
      "loss": 2.331,
      "step": 96708
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.9529046416282654,
      "learning_rate": 0.0003748456940494787,
      "loss": 2.1022,
      "step": 96709
    },
    {
      "epoch": 1.26,
      "grad_norm": 1.115724802017212,
      "learning_rate": 0.00037484173283007646,
      "loss": 2.3544,
      "step": 96710
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.7759381532669067,
      "learning_rate": 0.0003748377715967597,
      "loss": 2.3741,
      "step": 96711
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.9433904886245728,
      "learning_rate": 0.000374833810349529,
      "loss": 2.2836,
      "step": 96712
    },
    {
      "epoch": 1.26,
      "grad_norm": 1.0011018514633179,
      "learning_rate": 0.0003748298490883855,
      "loss": 2.046,
      "step": 96713
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.8426714539527893,
      "learning_rate": 0.0003748258878133296,
      "loss": 1.9072,
      "step": 96714
    },
    {
      "epoch": 1.26,
      "grad_norm": 1.01358163356781,
      "learning_rate": 0.0003748219265243622,
      "loss": 2.0386,
      "step": 96715
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.9803590774536133,
      "learning_rate": 0.00037481796522148396,
      "loss": 2.237,
      "step": 96716
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.8565914630889893,
      "learning_rate": 0.0003748140039046957,
      "loss": 1.9688,
      "step": 96717
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.937714159488678,
      "learning_rate": 0.00037481004257399804,
      "loss": 2.0259,
      "step": 96718
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.8932325839996338,
      "learning_rate": 0.0003748060812293918,
      "loss": 2.3283,
      "step": 96719
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.9575846791267395,
      "learning_rate": 0.00037480211987087774,
      "loss": 2.0492,
      "step": 96720
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.7714217901229858,
      "learning_rate": 0.00037479815849845646,
      "loss": 2.0653,
      "step": 96721
    },
    {
      "epoch": 1.26,
      "grad_norm": 1.131799340248108,
      "learning_rate": 0.0003747941971121289,
      "loss": 2.247,
      "step": 96722
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.8076916933059692,
      "learning_rate": 0.00037479023571189565,
      "loss": 1.8658,
      "step": 96723
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.7961933016777039,
      "learning_rate": 0.00037478627429775747,
      "loss": 1.8572,
      "step": 96724
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.9301255941390991,
      "learning_rate": 0.0003747823128697151,
      "loss": 2.3709,
      "step": 96725
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.8924649953842163,
      "learning_rate": 0.00037477835142776935,
      "loss": 2.374,
      "step": 96726
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.9418450593948364,
      "learning_rate": 0.00037477438997192083,
      "loss": 2.2209,
      "step": 96727
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.8399847745895386,
      "learning_rate": 0.00037477042850217034,
      "loss": 2.2167,
      "step": 96728
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.8501148819923401,
      "learning_rate": 0.00037476646701851865,
      "loss": 2.0997,
      "step": 96729
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.9587569832801819,
      "learning_rate": 0.00037476250552096645,
      "loss": 1.9072,
      "step": 96730
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.980350136756897,
      "learning_rate": 0.0003747585440095144,
      "loss": 2.1352,
      "step": 96731
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.8143690228462219,
      "learning_rate": 0.0003747545824841635,
      "loss": 2.2367,
      "step": 96732
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.931214451789856,
      "learning_rate": 0.0003747506209449142,
      "loss": 2.2259,
      "step": 96733
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.844618022441864,
      "learning_rate": 0.00037474665939176734,
      "loss": 1.9614,
      "step": 96734
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.8877764940261841,
      "learning_rate": 0.00037474269782472374,
      "loss": 2.3687,
      "step": 96735
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.8042131662368774,
      "learning_rate": 0.000374738736243784,
      "loss": 2.1222,
      "step": 96736
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.793554961681366,
      "learning_rate": 0.0003747347746489489,
      "loss": 1.9112,
      "step": 96737
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.8018994331359863,
      "learning_rate": 0.0003747308130402193,
      "loss": 2.0205,
      "step": 96738
    },
    {
      "epoch": 1.26,
      "grad_norm": 1.2969255447387695,
      "learning_rate": 0.0003747268514175957,
      "loss": 2.1114,
      "step": 96739
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.8173564672470093,
      "learning_rate": 0.000374722889781079,
      "loss": 2.076,
      "step": 96740
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.7993014454841614,
      "learning_rate": 0.00037471892813067,
      "loss": 2.0368,
      "step": 96741
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.9261019229888916,
      "learning_rate": 0.00037471496646636923,
      "loss": 2.1134,
      "step": 96742
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.8179057240486145,
      "learning_rate": 0.0003747110047881775,
      "loss": 2.3351,
      "step": 96743
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.9829554557800293,
      "learning_rate": 0.00037470704309609576,
      "loss": 1.8935,
      "step": 96744
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.8159583806991577,
      "learning_rate": 0.00037470308139012445,
      "loss": 2.1281,
      "step": 96745
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.8622555136680603,
      "learning_rate": 0.0003746991196702644,
      "loss": 2.3176,
      "step": 96746
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.9199249148368835,
      "learning_rate": 0.0003746951579365165,
      "loss": 2.2913,
      "step": 96747
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.8184623718261719,
      "learning_rate": 0.0003746911961888813,
      "loss": 2.0837,
      "step": 96748
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.8410694003105164,
      "learning_rate": 0.0003746872344273595,
      "loss": 2.1748,
      "step": 96749
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.9406951069831848,
      "learning_rate": 0.0003746832726519521,
      "loss": 2.0818,
      "step": 96750
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.990290105342865,
      "learning_rate": 0.00037467931086265956,
      "loss": 2.2753,
      "step": 96751
    },
    {
      "epoch": 1.26,
      "grad_norm": 1.0139766931533813,
      "learning_rate": 0.0003746753490594827,
      "loss": 2.2487,
      "step": 96752
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.8217230439186096,
      "learning_rate": 0.0003746713872424224,
      "loss": 1.9725,
      "step": 96753
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.989264965057373,
      "learning_rate": 0.0003746674254114792,
      "loss": 2.2938,
      "step": 96754
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.9310277700424194,
      "learning_rate": 0.0003746634635666539,
      "loss": 2.2053,
      "step": 96755
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.8784465789794922,
      "learning_rate": 0.00037465950170794733,
      "loss": 2.2271,
      "step": 96756
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.8323063850402832,
      "learning_rate": 0.00037465553983536006,
      "loss": 2.1441,
      "step": 96757
    },
    {
      "epoch": 1.26,
      "grad_norm": 1.1329631805419922,
      "learning_rate": 0.00037465157794889294,
      "loss": 2.2225,
      "step": 96758
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.9634782075881958,
      "learning_rate": 0.00037464761604854674,
      "loss": 2.0207,
      "step": 96759
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.8974910974502563,
      "learning_rate": 0.0003746436541343221,
      "loss": 2.2663,
      "step": 96760
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.9975048899650574,
      "learning_rate": 0.00037463969220621977,
      "loss": 2.2896,
      "step": 96761
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.9573435187339783,
      "learning_rate": 0.0003746357302642406,
      "loss": 2.0399,
      "step": 96762
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.9380531311035156,
      "learning_rate": 0.0003746317683083851,
      "loss": 2.1574,
      "step": 96763
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.9184753894805908,
      "learning_rate": 0.00037462780633865426,
      "loss": 2.4615,
      "step": 96764
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.8734329342842102,
      "learning_rate": 0.00037462384435504864,
      "loss": 2.0188,
      "step": 96765
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.8682299256324768,
      "learning_rate": 0.0003746198823575691,
      "loss": 1.9721,
      "step": 96766
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.8935691118240356,
      "learning_rate": 0.0003746159203462162,
      "loss": 2.04,
      "step": 96767
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.8830497860908508,
      "learning_rate": 0.00037461195832099093,
      "loss": 2.3013,
      "step": 96768
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.9176236391067505,
      "learning_rate": 0.0003746079962818938,
      "loss": 2.1807,
      "step": 96769
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.7791648507118225,
      "learning_rate": 0.0003746040342289257,
      "loss": 2.1355,
      "step": 96770
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.9323375821113586,
      "learning_rate": 0.0003746000721620872,
      "loss": 2.3484,
      "step": 96771
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.9112266898155212,
      "learning_rate": 0.0003745961100813792,
      "loss": 2.0617,
      "step": 96772
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.9005705118179321,
      "learning_rate": 0.00037459214798680237,
      "loss": 2.3589,
      "step": 96773
    },
    {
      "epoch": 1.26,
      "grad_norm": 1.004724383354187,
      "learning_rate": 0.00037458818587835744,
      "loss": 2.1548,
      "step": 96774
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.8498942255973816,
      "learning_rate": 0.0003745842237560452,
      "loss": 2.3909,
      "step": 96775
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.8056104779243469,
      "learning_rate": 0.0003745802616198663,
      "loss": 1.9116,
      "step": 96776
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.8005945682525635,
      "learning_rate": 0.0003745762994698216,
      "loss": 2.1696,
      "step": 96777
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.9043033123016357,
      "learning_rate": 0.00037457233730591165,
      "loss": 2.1946,
      "step": 96778
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.8727604150772095,
      "learning_rate": 0.0003745683751281374,
      "loss": 2.0496,
      "step": 96779
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.7768720388412476,
      "learning_rate": 0.00037456441293649937,
      "loss": 1.9612,
      "step": 96780
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.8516333103179932,
      "learning_rate": 0.0003745604507309985,
      "loss": 2.1417,
      "step": 96781
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.8381823897361755,
      "learning_rate": 0.00037455648851163534,
      "loss": 2.0842,
      "step": 96782
    },
    {
      "epoch": 1.26,
      "grad_norm": 1.1878433227539062,
      "learning_rate": 0.0003745525262784108,
      "loss": 2.0232,
      "step": 96783
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.8124640583992004,
      "learning_rate": 0.0003745485640313255,
      "loss": 2.0753,
      "step": 96784
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.8715353012084961,
      "learning_rate": 0.00037454460177038025,
      "loss": 2.2975,
      "step": 96785
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.9643591046333313,
      "learning_rate": 0.0003745406394955757,
      "loss": 2.3675,
      "step": 96786
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.8199042081832886,
      "learning_rate": 0.00037453667720691275,
      "loss": 2.1309,
      "step": 96787
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.8544071912765503,
      "learning_rate": 0.00037453271490439194,
      "loss": 2.1623,
      "step": 96788
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.8550640344619751,
      "learning_rate": 0.00037452875258801405,
      "loss": 1.9543,
      "step": 96789
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.8417055606842041,
      "learning_rate": 0.00037452479025777994,
      "loss": 2.1301,
      "step": 96790
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.9813838601112366,
      "learning_rate": 0.0003745208279136902,
      "loss": 2.1858,
      "step": 96791
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.8058136105537415,
      "learning_rate": 0.00037451686555574565,
      "loss": 2.2074,
      "step": 96792
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.840981662273407,
      "learning_rate": 0.0003745129031839471,
      "loss": 2.1593,
      "step": 96793
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.8533795475959778,
      "learning_rate": 0.0003745089407982951,
      "loss": 2.0645,
      "step": 96794
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.9715626239776611,
      "learning_rate": 0.00037450497839879043,
      "loss": 2.1854,
      "step": 96795
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.9086946249008179,
      "learning_rate": 0.000374501015985434,
      "loss": 2.2473,
      "step": 96796
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.8343465328216553,
      "learning_rate": 0.0003744970535582264,
      "loss": 2.2154,
      "step": 96797
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.8722289800643921,
      "learning_rate": 0.00037449309111716833,
      "loss": 2.2978,
      "step": 96798
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.9127566814422607,
      "learning_rate": 0.0003744891286622607,
      "loss": 2.2851,
      "step": 96799
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.9263059496879578,
      "learning_rate": 0.00037448516619350403,
      "loss": 2.1984,
      "step": 96800
    },
    {
      "epoch": 1.26,
      "grad_norm": 1.0282410383224487,
      "learning_rate": 0.00037448120371089914,
      "loss": 2.0465,
      "step": 96801
    },
    {
      "epoch": 1.26,
      "grad_norm": 1.004642367362976,
      "learning_rate": 0.00037447724121444687,
      "loss": 2.3547,
      "step": 96802
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.8151868581771851,
      "learning_rate": 0.0003744732787041478,
      "loss": 2.2999,
      "step": 96803
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.8971341848373413,
      "learning_rate": 0.0003744693161800028,
      "loss": 2.1981,
      "step": 96804
    },
    {
      "epoch": 1.26,
      "grad_norm": 1.023257851600647,
      "learning_rate": 0.0003744653536420125,
      "loss": 2.1502,
      "step": 96805
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.9432553052902222,
      "learning_rate": 0.00037446139109017774,
      "loss": 2.2187,
      "step": 96806
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.8554568886756897,
      "learning_rate": 0.00037445742852449917,
      "loss": 2.272,
      "step": 96807
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.7948863506317139,
      "learning_rate": 0.00037445346594497766,
      "loss": 2.0202,
      "step": 96808
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.9573250412940979,
      "learning_rate": 0.00037444950335161365,
      "loss": 1.9887,
      "step": 96809
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.9176079034805298,
      "learning_rate": 0.0003744455407444082,
      "loss": 2.3206,
      "step": 96810
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.9840620756149292,
      "learning_rate": 0.00037444157812336194,
      "loss": 1.9852,
      "step": 96811
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.8349171876907349,
      "learning_rate": 0.00037443761548847554,
      "loss": 2.4652,
      "step": 96812
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.8373454809188843,
      "learning_rate": 0.00037443365283974973,
      "loss": 2.0509,
      "step": 96813
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.9216889142990112,
      "learning_rate": 0.0003744296901771854,
      "loss": 2.2116,
      "step": 96814
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.7820981740951538,
      "learning_rate": 0.00037442572750078315,
      "loss": 2.405,
      "step": 96815
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.9474029541015625,
      "learning_rate": 0.00037442176481054367,
      "loss": 2.2243,
      "step": 96816
    },
    {
      "epoch": 1.26,
      "grad_norm": 1.0108311176300049,
      "learning_rate": 0.00037441780210646793,
      "loss": 2.2587,
      "step": 96817
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.8694159984588623,
      "learning_rate": 0.00037441383938855636,
      "loss": 2.4331,
      "step": 96818
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.7904818654060364,
      "learning_rate": 0.0003744098766568099,
      "loss": 2.2057,
      "step": 96819
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.8064547181129456,
      "learning_rate": 0.00037440591391122937,
      "loss": 2.0295,
      "step": 96820
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.8523193001747131,
      "learning_rate": 0.0003744019511518152,
      "loss": 2.1204,
      "step": 96821
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.8561540246009827,
      "learning_rate": 0.00037439798837856833,
      "loss": 2.0633,
      "step": 96822
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.8918929100036621,
      "learning_rate": 0.0003743940255914896,
      "loss": 2.3805,
      "step": 96823
    },
    {
      "epoch": 1.26,
      "grad_norm": 1.004315972328186,
      "learning_rate": 0.0003743900627905795,
      "loss": 2.2505,
      "step": 96824
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.9523802995681763,
      "learning_rate": 0.0003743860999758389,
      "loss": 2.2906,
      "step": 96825
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.8051999807357788,
      "learning_rate": 0.00037438213714726857,
      "loss": 2.1549,
      "step": 96826
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.8979448080062866,
      "learning_rate": 0.0003743781743048692,
      "loss": 2.2655,
      "step": 96827
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.9675576090812683,
      "learning_rate": 0.0003743742114486414,
      "loss": 2.345,
      "step": 96828
    },
    {
      "epoch": 1.26,
      "grad_norm": 1.1776237487792969,
      "learning_rate": 0.00037437024857858616,
      "loss": 2.1748,
      "step": 96829
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.8775249123573303,
      "learning_rate": 0.00037436628569470406,
      "loss": 1.9087,
      "step": 96830
    },
    {
      "epoch": 1.26,
      "grad_norm": 1.1758755445480347,
      "learning_rate": 0.0003743623227969958,
      "loss": 2.1776,
      "step": 96831
    },
    {
      "epoch": 1.26,
      "grad_norm": 1.104142665863037,
      "learning_rate": 0.00037435835988546223,
      "loss": 2.2272,
      "step": 96832
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.8822007179260254,
      "learning_rate": 0.00037435439696010403,
      "loss": 2.3618,
      "step": 96833
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.8551732301712036,
      "learning_rate": 0.0003743504340209219,
      "loss": 2.0843,
      "step": 96834
    },
    {
      "epoch": 1.26,
      "grad_norm": 1.0698983669281006,
      "learning_rate": 0.0003743464710679167,
      "loss": 2.1184,
      "step": 96835
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.9669567346572876,
      "learning_rate": 0.00037434250810108904,
      "loss": 2.2928,
      "step": 96836
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.9497036337852478,
      "learning_rate": 0.0003743385451204398,
      "loss": 2.0037,
      "step": 96837
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.8270329236984253,
      "learning_rate": 0.0003743345821259695,
      "loss": 2.0328,
      "step": 96838
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.93475341796875,
      "learning_rate": 0.00037433061911767903,
      "loss": 2.2728,
      "step": 96839
    },
    {
      "epoch": 1.26,
      "grad_norm": 1.0218957662582397,
      "learning_rate": 0.00037432665609556906,
      "loss": 2.2436,
      "step": 96840
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.9948592782020569,
      "learning_rate": 0.00037432269305964045,
      "loss": 2.4286,
      "step": 96841
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.8430532813072205,
      "learning_rate": 0.00037431873000989374,
      "loss": 2.0851,
      "step": 96842
    },
    {
      "epoch": 1.26,
      "grad_norm": 1.211431860923767,
      "learning_rate": 0.00037431476694632984,
      "loss": 2.3547,
      "step": 96843
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.8115602135658264,
      "learning_rate": 0.00037431080386894953,
      "loss": 2.2442,
      "step": 96844
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.7894163131713867,
      "learning_rate": 0.0003743068407777532,
      "loss": 2.1167,
      "step": 96845
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.9122933149337769,
      "learning_rate": 0.000374302877672742,
      "loss": 2.2603,
      "step": 96846
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.7946262359619141,
      "learning_rate": 0.0003742989145539165,
      "loss": 1.8,
      "step": 96847
    },
    {
      "epoch": 1.26,
      "grad_norm": 1.0561610460281372,
      "learning_rate": 0.0003742949514212774,
      "loss": 2.2448,
      "step": 96848
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.9393606185913086,
      "learning_rate": 0.00037429098827482536,
      "loss": 2.0659,
      "step": 96849
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.895319402217865,
      "learning_rate": 0.0003742870251145613,
      "loss": 1.9815,
      "step": 96850
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.9231306910514832,
      "learning_rate": 0.00037428306194048594,
      "loss": 2.1374,
      "step": 96851
    },
    {
      "epoch": 1.26,
      "grad_norm": 1.1481759548187256,
      "learning_rate": 0.0003742790987525999,
      "loss": 2.2309,
      "step": 96852
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.944490909576416,
      "learning_rate": 0.0003742751355509039,
      "loss": 2.4557,
      "step": 96853
    },
    {
      "epoch": 1.26,
      "grad_norm": 1.0980936288833618,
      "learning_rate": 0.00037427117233539885,
      "loss": 2.1335,
      "step": 96854
    },
    {
      "epoch": 1.26,
      "grad_norm": 1.0192630290985107,
      "learning_rate": 0.0003742672091060853,
      "loss": 2.332,
      "step": 96855
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.8929259181022644,
      "learning_rate": 0.0003742632458629642,
      "loss": 2.2194,
      "step": 96856
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.801766574382782,
      "learning_rate": 0.00037425928260603617,
      "loss": 2.053,
      "step": 96857
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.8965469002723694,
      "learning_rate": 0.00037425531933530184,
      "loss": 2.1478,
      "step": 96858
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.9159029126167297,
      "learning_rate": 0.000374251356050762,
      "loss": 2.1131,
      "step": 96859
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.8436530828475952,
      "learning_rate": 0.0003742473927524175,
      "loss": 2.2498,
      "step": 96860
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.8201740980148315,
      "learning_rate": 0.000374243429440269,
      "loss": 2.1274,
      "step": 96861
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.9722356796264648,
      "learning_rate": 0.00037423946611431723,
      "loss": 2.4395,
      "step": 96862
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.8490339517593384,
      "learning_rate": 0.000374235502774563,
      "loss": 2.2294,
      "step": 96863
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.7822145223617554,
      "learning_rate": 0.000374231539421007,
      "loss": 2.0239,
      "step": 96864
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.8549874424934387,
      "learning_rate": 0.00037422757605364986,
      "loss": 2.1444,
      "step": 96865
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.8467262387275696,
      "learning_rate": 0.00037422361267249246,
      "loss": 1.9911,
      "step": 96866
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.9532305002212524,
      "learning_rate": 0.00037421964927753545,
      "loss": 2.1836,
      "step": 96867
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.8508836627006531,
      "learning_rate": 0.00037421568586877975,
      "loss": 2.0418,
      "step": 96868
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.76829594373703,
      "learning_rate": 0.0003742117224462258,
      "loss": 2.2981,
      "step": 96869
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.831401526927948,
      "learning_rate": 0.0003742077590098745,
      "loss": 2.3931,
      "step": 96870
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.8808642029762268,
      "learning_rate": 0.00037420379555972664,
      "loss": 2.0923,
      "step": 96871
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.8295574188232422,
      "learning_rate": 0.0003741998320957829,
      "loss": 2.0102,
      "step": 96872
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.8543877601623535,
      "learning_rate": 0.00037419586861804394,
      "loss": 2.0156,
      "step": 96873
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.8477613925933838,
      "learning_rate": 0.0003741919051265106,
      "loss": 2.0277,
      "step": 96874
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.9664698839187622,
      "learning_rate": 0.0003741879416211836,
      "loss": 2.1498,
      "step": 96875
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.9068416953086853,
      "learning_rate": 0.0003741839781020637,
      "loss": 2.0526,
      "step": 96876
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.96994948387146,
      "learning_rate": 0.0003741800145691515,
      "loss": 2.3069,
      "step": 96877
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.9185328483581543,
      "learning_rate": 0.0003741760510224479,
      "loss": 1.9723,
      "step": 96878
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.8842527270317078,
      "learning_rate": 0.00037417208746195357,
      "loss": 1.9917,
      "step": 96879
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.8293368816375732,
      "learning_rate": 0.0003741681238876692,
      "loss": 2.2473,
      "step": 96880
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.9894386529922485,
      "learning_rate": 0.00037416416029959566,
      "loss": 2.2669,
      "step": 96881
    },
    {
      "epoch": 1.26,
      "grad_norm": 1.019911289215088,
      "learning_rate": 0.0003741601966977336,
      "loss": 2.3039,
      "step": 96882
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.8608133792877197,
      "learning_rate": 0.0003741562330820837,
      "loss": 2.2386,
      "step": 96883
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.8063059449195862,
      "learning_rate": 0.0003741522694526467,
      "loss": 2.063,
      "step": 96884
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.948621928691864,
      "learning_rate": 0.0003741483058094235,
      "loss": 2.0947,
      "step": 96885
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.9397775530815125,
      "learning_rate": 0.00037414434215241466,
      "loss": 2.3639,
      "step": 96886
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.7982927560806274,
      "learning_rate": 0.0003741403784816211,
      "loss": 1.9528,
      "step": 96887
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.8820589184761047,
      "learning_rate": 0.00037413641479704335,
      "loss": 2.2108,
      "step": 96888
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.8979405760765076,
      "learning_rate": 0.00037413245109868223,
      "loss": 1.9375,
      "step": 96889
    },
    {
      "epoch": 1.26,
      "grad_norm": 1.004472255706787,
      "learning_rate": 0.00037412848738653856,
      "loss": 2.0324,
      "step": 96890
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.8304687142372131,
      "learning_rate": 0.0003741245236606129,
      "loss": 1.9394,
      "step": 96891
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.8164806365966797,
      "learning_rate": 0.00037412055992090617,
      "loss": 2.1001,
      "step": 96892
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.9608996510505676,
      "learning_rate": 0.00037411659616741905,
      "loss": 2.4727,
      "step": 96893
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.9424388408660889,
      "learning_rate": 0.00037411263240015217,
      "loss": 2.053,
      "step": 96894
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.8096939325332642,
      "learning_rate": 0.0003741086686191064,
      "loss": 2.0285,
      "step": 96895
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.9121333360671997,
      "learning_rate": 0.0003741047048242825,
      "loss": 2.1334,
      "step": 96896
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.9033697843551636,
      "learning_rate": 0.00037410074101568104,
      "loss": 2.3409,
      "step": 96897
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.9314121603965759,
      "learning_rate": 0.0003740967771933028,
      "loss": 2.2566,
      "step": 96898
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.9460298418998718,
      "learning_rate": 0.0003740928133571487,
      "loss": 2.4057,
      "step": 96899
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.9084135890007019,
      "learning_rate": 0.0003740888495072193,
      "loss": 2.3375,
      "step": 96900
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.8053960800170898,
      "learning_rate": 0.00037408488564351534,
      "loss": 2.1518,
      "step": 96901
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.8103638887405396,
      "learning_rate": 0.0003740809217660377,
      "loss": 2.1147,
      "step": 96902
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.924256443977356,
      "learning_rate": 0.000374076957874787,
      "loss": 2.3652,
      "step": 96903
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.9478479027748108,
      "learning_rate": 0.0003740729939697639,
      "loss": 2.2356,
      "step": 96904
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.9020678997039795,
      "learning_rate": 0.0003740690300509693,
      "loss": 2.0141,
      "step": 96905
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.8305255770683289,
      "learning_rate": 0.00037406506611840386,
      "loss": 2.1555,
      "step": 96906
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.9609909057617188,
      "learning_rate": 0.00037406110217206835,
      "loss": 2.1298,
      "step": 96907
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.8984440565109253,
      "learning_rate": 0.0003740571382119635,
      "loss": 2.2607,
      "step": 96908
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.9149348735809326,
      "learning_rate": 0.0003740531742380899,
      "loss": 2.282,
      "step": 96909
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.902329683303833,
      "learning_rate": 0.0003740492102504485,
      "loss": 2.1745,
      "step": 96910
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.9363004565238953,
      "learning_rate": 0.00037404524624903997,
      "loss": 2.3112,
      "step": 96911
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.8518936634063721,
      "learning_rate": 0.000374041282233865,
      "loss": 2.118,
      "step": 96912
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.8199718594551086,
      "learning_rate": 0.0003740373182049244,
      "loss": 2.1615,
      "step": 96913
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.898963987827301,
      "learning_rate": 0.00037403335416221894,
      "loss": 2.1861,
      "step": 96914
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.8818196654319763,
      "learning_rate": 0.0003740293901057491,
      "loss": 2.1408,
      "step": 96915
    },
    {
      "epoch": 1.26,
      "grad_norm": 1.5484672784805298,
      "learning_rate": 0.0003740254260355159,
      "loss": 2.3838,
      "step": 96916
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.8824197053909302,
      "learning_rate": 0.00037402146195152,
      "loss": 2.15,
      "step": 96917
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.9185665249824524,
      "learning_rate": 0.000374017497853762,
      "loss": 2.0437,
      "step": 96918
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.9730631709098816,
      "learning_rate": 0.0003740135337422428,
      "loss": 2.4499,
      "step": 96919
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.9096906781196594,
      "learning_rate": 0.00037400956961696315,
      "loss": 2.2028,
      "step": 96920
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.8344506025314331,
      "learning_rate": 0.0003740056054779237,
      "loss": 2.2581,
      "step": 96921
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.9554895162582397,
      "learning_rate": 0.00037400164132512517,
      "loss": 2.4678,
      "step": 96922
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.8420812487602234,
      "learning_rate": 0.00037399767715856836,
      "loss": 2.1648,
      "step": 96923
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.8818303346633911,
      "learning_rate": 0.000373993712978254,
      "loss": 2.1449,
      "step": 96924
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.8713687062263489,
      "learning_rate": 0.0003739897487841828,
      "loss": 1.9647,
      "step": 96925
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.8375057578086853,
      "learning_rate": 0.00037398578457635545,
      "loss": 2.1712,
      "step": 96926
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.8650307059288025,
      "learning_rate": 0.0003739818203547729,
      "loss": 2.0918,
      "step": 96927
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.8365552425384521,
      "learning_rate": 0.0003739778561194356,
      "loss": 2.0911,
      "step": 96928
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.7906819581985474,
      "learning_rate": 0.0003739738918703444,
      "loss": 2.001,
      "step": 96929
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.9425572156906128,
      "learning_rate": 0.00037396992760750024,
      "loss": 2.3113,
      "step": 96930
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.9281967878341675,
      "learning_rate": 0.0003739659633309035,
      "loss": 2.1662,
      "step": 96931
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.8056100010871887,
      "learning_rate": 0.0003739619990405551,
      "loss": 2.2248,
      "step": 96932
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.9537798762321472,
      "learning_rate": 0.00037395803473645587,
      "loss": 2.2119,
      "step": 96933
    },
    {
      "epoch": 1.26,
      "grad_norm": 1.1713505983352661,
      "learning_rate": 0.00037395407041860635,
      "loss": 2.2784,
      "step": 96934
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.8558042645454407,
      "learning_rate": 0.0003739501060870074,
      "loss": 2.1921,
      "step": 96935
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.9022029638290405,
      "learning_rate": 0.00037394614174165975,
      "loss": 2.1486,
      "step": 96936
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.9249921441078186,
      "learning_rate": 0.0003739421773825641,
      "loss": 2.0655,
      "step": 96937
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.7846373915672302,
      "learning_rate": 0.00037393821300972117,
      "loss": 1.9444,
      "step": 96938
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.8796257972717285,
      "learning_rate": 0.0003739342486231319,
      "loss": 1.8859,
      "step": 96939
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.8793131709098816,
      "learning_rate": 0.00037393028422279663,
      "loss": 2.382,
      "step": 96940
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.8307341933250427,
      "learning_rate": 0.00037392631980871636,
      "loss": 2.1762,
      "step": 96941
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.9071762561798096,
      "learning_rate": 0.0003739223553808919,
      "loss": 1.8802,
      "step": 96942
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.8469108939170837,
      "learning_rate": 0.0003739183909393239,
      "loss": 1.9898,
      "step": 96943
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.8863497972488403,
      "learning_rate": 0.00037391442648401297,
      "loss": 1.9767,
      "step": 96944
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.8795323371887207,
      "learning_rate": 0.00037391046201495996,
      "loss": 1.9335,
      "step": 96945
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.998369038105011,
      "learning_rate": 0.0003739064975321656,
      "loss": 2.0072,
      "step": 96946
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.8408412337303162,
      "learning_rate": 0.0003739025330356307,
      "loss": 2.2874,
      "step": 96947
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.8404040932655334,
      "learning_rate": 0.0003738985685253558,
      "loss": 1.9748,
      "step": 96948
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.8911270499229431,
      "learning_rate": 0.00037389460400134194,
      "loss": 2.1435,
      "step": 96949
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.8000326752662659,
      "learning_rate": 0.0003738906394635895,
      "loss": 2.0302,
      "step": 96950
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.7782068252563477,
      "learning_rate": 0.0003738866749120995,
      "loss": 1.9743,
      "step": 96951
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.7864102125167847,
      "learning_rate": 0.00037388271034687255,
      "loss": 2.0795,
      "step": 96952
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.845102846622467,
      "learning_rate": 0.00037387874576790936,
      "loss": 2.2127,
      "step": 96953
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.8181338310241699,
      "learning_rate": 0.0003738747811752108,
      "loss": 1.9648,
      "step": 96954
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.7882232069969177,
      "learning_rate": 0.00037387081656877744,
      "loss": 2.2259,
      "step": 96955
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.8645337224006653,
      "learning_rate": 0.00037386685194861017,
      "loss": 2.0606,
      "step": 96956
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.8277808427810669,
      "learning_rate": 0.0003738628873147096,
      "loss": 2.1828,
      "step": 96957
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.8098856806755066,
      "learning_rate": 0.00037385892266707656,
      "loss": 2.4063,
      "step": 96958
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.8580788969993591,
      "learning_rate": 0.00037385495800571183,
      "loss": 2.282,
      "step": 96959
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.8883601427078247,
      "learning_rate": 0.0003738509933306159,
      "loss": 2.2954,
      "step": 96960
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.8387253284454346,
      "learning_rate": 0.0003738470286417898,
      "loss": 1.9733,
      "step": 96961
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.9448668956756592,
      "learning_rate": 0.00037384306393923407,
      "loss": 2.2514,
      "step": 96962
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.8167269229888916,
      "learning_rate": 0.0003738390992229496,
      "loss": 2.0507,
      "step": 96963
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.8813208937644958,
      "learning_rate": 0.00037383513449293696,
      "loss": 2.1134,
      "step": 96964
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.8582174777984619,
      "learning_rate": 0.00037383116974919705,
      "loss": 2.2363,
      "step": 96965
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.772042453289032,
      "learning_rate": 0.00037382720499173054,
      "loss": 2.3823,
      "step": 96966
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.778343915939331,
      "learning_rate": 0.00037382324022053814,
      "loss": 1.9699,
      "step": 96967
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.8340643644332886,
      "learning_rate": 0.00037381927543562055,
      "loss": 2.3277,
      "step": 96968
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.8602989315986633,
      "learning_rate": 0.0003738153106369786,
      "loss": 2.2169,
      "step": 96969
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.9489806294441223,
      "learning_rate": 0.00037381134582461297,
      "loss": 2.3422,
      "step": 96970
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.8682985901832581,
      "learning_rate": 0.0003738073809985245,
      "loss": 2.0695,
      "step": 96971
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.8242203593254089,
      "learning_rate": 0.0003738034161587138,
      "loss": 2.1291,
      "step": 96972
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.8411280512809753,
      "learning_rate": 0.0003737994513051816,
      "loss": 2.0991,
      "step": 96973
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.8359754085540771,
      "learning_rate": 0.0003737954864379287,
      "loss": 2.1726,
      "step": 96974
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.8381456732749939,
      "learning_rate": 0.00037379152155695593,
      "loss": 1.9567,
      "step": 96975
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.8746582865715027,
      "learning_rate": 0.00037378755666226375,
      "loss": 2.3686,
      "step": 96976
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.7588681578636169,
      "learning_rate": 0.00037378359175385315,
      "loss": 2.1351,
      "step": 96977
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.8264269232749939,
      "learning_rate": 0.0003737796268317249,
      "loss": 2.2932,
      "step": 96978
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.8740427494049072,
      "learning_rate": 0.0003737756618958795,
      "loss": 2.1152,
      "step": 96979
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.7998185753822327,
      "learning_rate": 0.00037377169694631783,
      "loss": 2.0707,
      "step": 96980
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.7803019881248474,
      "learning_rate": 0.00037376773198304064,
      "loss": 2.291,
      "step": 96981
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.7663216590881348,
      "learning_rate": 0.0003737637670060486,
      "loss": 2.1358,
      "step": 96982
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.9344611763954163,
      "learning_rate": 0.0003737598020153425,
      "loss": 2.1138,
      "step": 96983
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.8380919694900513,
      "learning_rate": 0.0003737558370109231,
      "loss": 2.2087,
      "step": 96984
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.9790704846382141,
      "learning_rate": 0.00037375187199279104,
      "loss": 2.1721,
      "step": 96985
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.7967758774757385,
      "learning_rate": 0.0003737479069609471,
      "loss": 2.0077,
      "step": 96986
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.8747299313545227,
      "learning_rate": 0.000373743941915392,
      "loss": 1.9908,
      "step": 96987
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.8352243900299072,
      "learning_rate": 0.00037373997685612664,
      "loss": 1.996,
      "step": 96988
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.7731814384460449,
      "learning_rate": 0.00037373601178315155,
      "loss": 1.887,
      "step": 96989
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.8395601511001587,
      "learning_rate": 0.00037373204669646754,
      "loss": 2.0335,
      "step": 96990
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.9568008184432983,
      "learning_rate": 0.00037372808159607537,
      "loss": 2.062,
      "step": 96991
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.9708102345466614,
      "learning_rate": 0.00037372411648197573,
      "loss": 2.2829,
      "step": 96992
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.8907249569892883,
      "learning_rate": 0.00037372015135416945,
      "loss": 1.8968,
      "step": 96993
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.9540151953697205,
      "learning_rate": 0.0003737161862126572,
      "loss": 2.2727,
      "step": 96994
    },
    {
      "epoch": 1.26,
      "grad_norm": 1.013155460357666,
      "learning_rate": 0.0003737122210574396,
      "loss": 2.1464,
      "step": 96995
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.9290206432342529,
      "learning_rate": 0.0003737082558885176,
      "loss": 2.1828,
      "step": 96996
    },
    {
      "epoch": 1.26,
      "grad_norm": 1.095495343208313,
      "learning_rate": 0.00037370429070589185,
      "loss": 2.149,
      "step": 96997
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.9358848929405212,
      "learning_rate": 0.000373700325509563,
      "loss": 2.2846,
      "step": 96998
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.9710443615913391,
      "learning_rate": 0.000373696360299532,
      "loss": 2.0585,
      "step": 96999
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.983045756816864,
      "learning_rate": 0.0003736923950757994,
      "loss": 2.0731,
      "step": 97000
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.8565540313720703,
      "learning_rate": 0.0003736884298383659,
      "loss": 2.1187,
      "step": 97001
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.9848629236221313,
      "learning_rate": 0.00037368446458723245,
      "loss": 1.9658,
      "step": 97002
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.9361057877540588,
      "learning_rate": 0.00037368049932239966,
      "loss": 2.144,
      "step": 97003
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.9054426550865173,
      "learning_rate": 0.0003736765340438682,
      "loss": 2.2466,
      "step": 97004
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.82377028465271,
      "learning_rate": 0.0003736725687516389,
      "loss": 2.0669,
      "step": 97005
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.8647047877311707,
      "learning_rate": 0.0003736686034457125,
      "loss": 2.2385,
      "step": 97006
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.9773432016372681,
      "learning_rate": 0.00037366463812608974,
      "loss": 2.2352,
      "step": 97007
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.846483588218689,
      "learning_rate": 0.00037366067279277125,
      "loss": 2.1265,
      "step": 97008
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.898417592048645,
      "learning_rate": 0.000373656707445758,
      "loss": 1.9538,
      "step": 97009
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.87650066614151,
      "learning_rate": 0.0003736527420850504,
      "loss": 2.1354,
      "step": 97010
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.9493138194084167,
      "learning_rate": 0.00037364877671064944,
      "loss": 2.1757,
      "step": 97011
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.9169378280639648,
      "learning_rate": 0.00037364481132255587,
      "loss": 1.9562,
      "step": 97012
    },
    {
      "epoch": 1.26,
      "grad_norm": 1.0204097032546997,
      "learning_rate": 0.00037364084592077023,
      "loss": 1.9844,
      "step": 97013
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.8808056712150574,
      "learning_rate": 0.00037363688050529335,
      "loss": 2.0765,
      "step": 97014
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.913646399974823,
      "learning_rate": 0.00037363291507612614,
      "loss": 1.9338,
      "step": 97015
    },
    {
      "epoch": 1.26,
      "grad_norm": 1.1330668926239014,
      "learning_rate": 0.000373628949633269,
      "loss": 2.1145,
      "step": 97016
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.8898374438285828,
      "learning_rate": 0.00037362498417672295,
      "loss": 2.0675,
      "step": 97017
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.8192319869995117,
      "learning_rate": 0.0003736210187064886,
      "loss": 1.9754,
      "step": 97018
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.9535366892814636,
      "learning_rate": 0.00037361705322256673,
      "loss": 2.308,
      "step": 97019
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.8988124132156372,
      "learning_rate": 0.00037361308772495796,
      "loss": 2.3886,
      "step": 97020
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.8607481122016907,
      "learning_rate": 0.0003736091222136633,
      "loss": 2.1452,
      "step": 97021
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.8760993480682373,
      "learning_rate": 0.0003736051566886832,
      "loss": 2.2665,
      "step": 97022
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.8692441582679749,
      "learning_rate": 0.0003736011911500185,
      "loss": 2.2482,
      "step": 97023
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.906568706035614,
      "learning_rate": 0.00037359722559767003,
      "loss": 2.2677,
      "step": 97024
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.8171733021736145,
      "learning_rate": 0.00037359326003163834,
      "loss": 2.1389,
      "step": 97025
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.8148661255836487,
      "learning_rate": 0.00037358929445192434,
      "loss": 2.1559,
      "step": 97026
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.7801304459571838,
      "learning_rate": 0.00037358532885852874,
      "loss": 2.117,
      "step": 97027
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.8330914974212646,
      "learning_rate": 0.0003735813632514522,
      "loss": 2.0661,
      "step": 97028
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.8820485472679138,
      "learning_rate": 0.00037357739763069545,
      "loss": 2.1757,
      "step": 97029
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.8703331351280212,
      "learning_rate": 0.0003735734319962594,
      "loss": 2.0704,
      "step": 97030
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.8218626976013184,
      "learning_rate": 0.0003735694663481445,
      "loss": 2.1931,
      "step": 97031
    },
    {
      "epoch": 1.26,
      "grad_norm": 1.0061635971069336,
      "learning_rate": 0.0003735655006863517,
      "loss": 2.2887,
      "step": 97032
    },
    {
      "epoch": 1.26,
      "grad_norm": 1.1323179006576538,
      "learning_rate": 0.00037356153501088174,
      "loss": 2.1592,
      "step": 97033
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.8234570622444153,
      "learning_rate": 0.00037355756932173526,
      "loss": 1.9676,
      "step": 97034
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.833615243434906,
      "learning_rate": 0.0003735536036189131,
      "loss": 2.2764,
      "step": 97035
    },
    {
      "epoch": 1.26,
      "grad_norm": 1.0404934883117676,
      "learning_rate": 0.0003735496379024158,
      "loss": 2.0892,
      "step": 97036
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.929168701171875,
      "learning_rate": 0.0003735456721722444,
      "loss": 2.2083,
      "step": 97037
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.8453235030174255,
      "learning_rate": 0.0003735417064283993,
      "loss": 1.8776,
      "step": 97038
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.9961510300636292,
      "learning_rate": 0.00037353774067088146,
      "loss": 1.9205,
      "step": 97039
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.898486852645874,
      "learning_rate": 0.00037353377489969167,
      "loss": 2.2045,
      "step": 97040
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.8574023842811584,
      "learning_rate": 0.0003735298091148305,
      "loss": 2.1166,
      "step": 97041
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.8110498785972595,
      "learning_rate": 0.00037352584331629867,
      "loss": 2.264,
      "step": 97042
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.8429887294769287,
      "learning_rate": 0.00037352187750409714,
      "loss": 2.1647,
      "step": 97043
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.8863925933837891,
      "learning_rate": 0.0003735179116782264,
      "loss": 2.1853,
      "step": 97044
    },
    {
      "epoch": 1.26,
      "grad_norm": 1.0161793231964111,
      "learning_rate": 0.00037351394583868733,
      "loss": 2.1559,
      "step": 97045
    },
    {
      "epoch": 1.26,
      "grad_norm": 1.0531442165374756,
      "learning_rate": 0.00037350997998548063,
      "loss": 2.1638,
      "step": 97046
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.8428269624710083,
      "learning_rate": 0.000373506014118607,
      "loss": 1.9036,
      "step": 97047
    },
    {
      "epoch": 1.26,
      "grad_norm": 1.0345240831375122,
      "learning_rate": 0.0003735020482380673,
      "loss": 2.073,
      "step": 97048
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.8765251636505127,
      "learning_rate": 0.00037349808234386214,
      "loss": 2.0927,
      "step": 97049
    },
    {
      "epoch": 1.26,
      "grad_norm": 1.027505874633789,
      "learning_rate": 0.0003734941164359923,
      "loss": 2.0353,
      "step": 97050
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.8576120734214783,
      "learning_rate": 0.0003734901505144585,
      "loss": 2.1345,
      "step": 97051
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.9523823261260986,
      "learning_rate": 0.0003734861845792615,
      "loss": 2.3542,
      "step": 97052
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.9132861495018005,
      "learning_rate": 0.000373482218630402,
      "loss": 2.2998,
      "step": 97053
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.7998063564300537,
      "learning_rate": 0.0003734782526678808,
      "loss": 2.2864,
      "step": 97054
    },
    {
      "epoch": 1.26,
      "grad_norm": 1.017309308052063,
      "learning_rate": 0.0003734742866916987,
      "loss": 2.2133,
      "step": 97055
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.7997348308563232,
      "learning_rate": 0.0003734703207018562,
      "loss": 1.9988,
      "step": 97056
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.9131895899772644,
      "learning_rate": 0.0003734663546983542,
      "loss": 2.183,
      "step": 97057
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.8997374773025513,
      "learning_rate": 0.00037346238868119354,
      "loss": 2.2419,
      "step": 97058
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.8549675941467285,
      "learning_rate": 0.0003734584226503747,
      "loss": 2.2639,
      "step": 97059
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.8101258277893066,
      "learning_rate": 0.0003734544566058985,
      "loss": 2.3382,
      "step": 97060
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.8002078533172607,
      "learning_rate": 0.0003734504905477659,
      "loss": 2.1018,
      "step": 97061
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.8886443376541138,
      "learning_rate": 0.00037344652447597745,
      "loss": 2.0773,
      "step": 97062
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.8700275421142578,
      "learning_rate": 0.00037344255839053377,
      "loss": 1.8723,
      "step": 97063
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.8800903558731079,
      "learning_rate": 0.00037343859229143587,
      "loss": 2.2034,
      "step": 97064
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.8532734513282776,
      "learning_rate": 0.0003734346261786843,
      "loss": 2.052,
      "step": 97065
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.8163793683052063,
      "learning_rate": 0.00037343066005227977,
      "loss": 2.2877,
      "step": 97066
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.7920229434967041,
      "learning_rate": 0.0003734266939122232,
      "loss": 2.264,
      "step": 97067
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.8242791295051575,
      "learning_rate": 0.0003734227277585152,
      "loss": 2.1938,
      "step": 97068
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.8751845359802246,
      "learning_rate": 0.0003734187615911565,
      "loss": 2.2571,
      "step": 97069
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.8876715302467346,
      "learning_rate": 0.00037341479541014793,
      "loss": 2.2871,
      "step": 97070
    },
    {
      "epoch": 1.26,
      "grad_norm": 1.087314486503601,
      "learning_rate": 0.0003734108292154901,
      "loss": 2.2772,
      "step": 97071
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.8936702609062195,
      "learning_rate": 0.0003734068630071838,
      "loss": 1.9668,
      "step": 97072
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.8921303749084473,
      "learning_rate": 0.00037340289678522984,
      "loss": 2.1743,
      "step": 97073
    },
    {
      "epoch": 1.26,
      "grad_norm": 1.2527670860290527,
      "learning_rate": 0.0003733989305496288,
      "loss": 2.1612,
      "step": 97074
    },
    {
      "epoch": 1.26,
      "grad_norm": 1.0973325967788696,
      "learning_rate": 0.0003733949643003816,
      "loss": 2.3032,
      "step": 97075
    },
    {
      "epoch": 1.26,
      "grad_norm": 1.0273560285568237,
      "learning_rate": 0.0003733909980374889,
      "loss": 2.2487,
      "step": 97076
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.8915390372276306,
      "learning_rate": 0.0003733870317609513,
      "loss": 2.2921,
      "step": 97077
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.8192155361175537,
      "learning_rate": 0.0003733830654707698,
      "loss": 2.1192,
      "step": 97078
    },
    {
      "epoch": 1.26,
      "grad_norm": 1.467083215713501,
      "learning_rate": 0.000373379099166945,
      "loss": 2.0363,
      "step": 97079
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.8647664189338684,
      "learning_rate": 0.0003733751328494775,
      "loss": 2.3038,
      "step": 97080
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.9158477783203125,
      "learning_rate": 0.0003733711665183683,
      "loss": 2.1548,
      "step": 97081
    },
    {
      "epoch": 1.26,
      "grad_norm": 1.093706727027893,
      "learning_rate": 0.00037336720017361807,
      "loss": 2.1923,
      "step": 97082
    },
    {
      "epoch": 1.26,
      "grad_norm": 1.0514674186706543,
      "learning_rate": 0.0003733632338152274,
      "loss": 2.3887,
      "step": 97083
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.8747155666351318,
      "learning_rate": 0.0003733592674431971,
      "loss": 2.506,
      "step": 97084
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.93181312084198,
      "learning_rate": 0.00037335530105752805,
      "loss": 2.1179,
      "step": 97085
    },
    {
      "epoch": 1.26,
      "grad_norm": 1.049691915512085,
      "learning_rate": 0.0003733513346582207,
      "loss": 2.114,
      "step": 97086
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.8862226009368896,
      "learning_rate": 0.000373347368245276,
      "loss": 2.1492,
      "step": 97087
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.9390487670898438,
      "learning_rate": 0.00037334340181869474,
      "loss": 2.2128,
      "step": 97088
    },
    {
      "epoch": 1.26,
      "grad_norm": 1.1507655382156372,
      "learning_rate": 0.0003733394353784775,
      "loss": 2.1563,
      "step": 97089
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.8299365639686584,
      "learning_rate": 0.00037333546892462496,
      "loss": 2.1276,
      "step": 97090
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.8230882287025452,
      "learning_rate": 0.0003733315024571382,
      "loss": 2.3423,
      "step": 97091
    },
    {
      "epoch": 1.26,
      "grad_norm": 1.1473861932754517,
      "learning_rate": 0.00037332753597601754,
      "loss": 2.1168,
      "step": 97092
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.8974113464355469,
      "learning_rate": 0.0003733235694812639,
      "loss": 2.1237,
      "step": 97093
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.9117029309272766,
      "learning_rate": 0.00037331960297287817,
      "loss": 2.2449,
      "step": 97094
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.9025036096572876,
      "learning_rate": 0.00037331563645086086,
      "loss": 2.3122,
      "step": 97095
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.9840084314346313,
      "learning_rate": 0.0003733116699152128,
      "loss": 2.2602,
      "step": 97096
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.7735291123390198,
      "learning_rate": 0.00037330770336593475,
      "loss": 2.0289,
      "step": 97097
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.9314519166946411,
      "learning_rate": 0.00037330373680302735,
      "loss": 2.1114,
      "step": 97098
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.8744893074035645,
      "learning_rate": 0.0003732997702264914,
      "loss": 2.2733,
      "step": 97099
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.9678951501846313,
      "learning_rate": 0.00037329580363632773,
      "loss": 2.1952,
      "step": 97100
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.9920635223388672,
      "learning_rate": 0.0003732918370325369,
      "loss": 2.2723,
      "step": 97101
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.906451940536499,
      "learning_rate": 0.00037328787041511973,
      "loss": 2.0415,
      "step": 97102
    },
    {
      "epoch": 1.26,
      "grad_norm": 1.0302002429962158,
      "learning_rate": 0.000373283903784077,
      "loss": 2.0679,
      "step": 97103
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.8726657032966614,
      "learning_rate": 0.00037327993713940944,
      "loss": 2.122,
      "step": 97104
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.8018044829368591,
      "learning_rate": 0.0003732759704811177,
      "loss": 2.1739,
      "step": 97105
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.8751664161682129,
      "learning_rate": 0.0003732720038092026,
      "loss": 2.0757,
      "step": 97106
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.9518973231315613,
      "learning_rate": 0.0003732680371236648,
      "loss": 2.4923,
      "step": 97107
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.7919948697090149,
      "learning_rate": 0.0003732640704245051,
      "loss": 1.8806,
      "step": 97108
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.8411424160003662,
      "learning_rate": 0.0003732601037117243,
      "loss": 2.2384,
      "step": 97109
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.8438461422920227,
      "learning_rate": 0.000373256136985323,
      "loss": 1.9647,
      "step": 97110
    },
    {
      "epoch": 1.26,
      "grad_norm": 1.0199780464172363,
      "learning_rate": 0.00037325217024530204,
      "loss": 2.2323,
      "step": 97111
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.888591468334198,
      "learning_rate": 0.0003732482034916621,
      "loss": 2.2802,
      "step": 97112
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.9273078441619873,
      "learning_rate": 0.0003732442367244039,
      "loss": 2.1767,
      "step": 97113
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.9467478394508362,
      "learning_rate": 0.00037324026994352825,
      "loss": 2.1273,
      "step": 97114
    },
    {
      "epoch": 1.26,
      "grad_norm": 1.0542515516281128,
      "learning_rate": 0.00037323630314903586,
      "loss": 2.1351,
      "step": 97115
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.9654065370559692,
      "learning_rate": 0.0003732323363409275,
      "loss": 2.2719,
      "step": 97116
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.9101082682609558,
      "learning_rate": 0.0003732283695192038,
      "loss": 2.1771,
      "step": 97117
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.8113194704055786,
      "learning_rate": 0.0003732244026838656,
      "loss": 2.1385,
      "step": 97118
    },
    {
      "epoch": 1.26,
      "grad_norm": 1.2280560731887817,
      "learning_rate": 0.0003732204358349135,
      "loss": 2.1635,
      "step": 97119
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.9240541458129883,
      "learning_rate": 0.0003732164689723484,
      "loss": 2.3018,
      "step": 97120
    },
    {
      "epoch": 1.26,
      "grad_norm": 1.0066640377044678,
      "learning_rate": 0.000373212502096171,
      "loss": 2.1854,
      "step": 97121
    },
    {
      "epoch": 1.26,
      "grad_norm": 1.025399923324585,
      "learning_rate": 0.0003732085352063821,
      "loss": 2.3568,
      "step": 97122
    },
    {
      "epoch": 1.26,
      "grad_norm": 1.2218283414840698,
      "learning_rate": 0.0003732045683029822,
      "loss": 2.1629,
      "step": 97123
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.9950084090232849,
      "learning_rate": 0.00037320060138597223,
      "loss": 2.2501,
      "step": 97124
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.8594568967819214,
      "learning_rate": 0.0003731966344553529,
      "loss": 2.1612,
      "step": 97125
    },
    {
      "epoch": 1.26,
      "grad_norm": 1.1132702827453613,
      "learning_rate": 0.00037319266751112493,
      "loss": 2.1364,
      "step": 97126
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.9446285963058472,
      "learning_rate": 0.0003731887005532891,
      "loss": 2.288,
      "step": 97127
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.9701792001724243,
      "learning_rate": 0.0003731847335818461,
      "loss": 2.0652,
      "step": 97128
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.8333526849746704,
      "learning_rate": 0.00037318076659679665,
      "loss": 2.2267,
      "step": 97129
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.8553217053413391,
      "learning_rate": 0.0003731767995981415,
      "loss": 1.9762,
      "step": 97130
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.9571149945259094,
      "learning_rate": 0.00037317283258588143,
      "loss": 2.0905,
      "step": 97131
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.9866562485694885,
      "learning_rate": 0.00037316886556001716,
      "loss": 2.2393,
      "step": 97132
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.7990008592605591,
      "learning_rate": 0.00037316489852054935,
      "loss": 2.096,
      "step": 97133
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.9550011157989502,
      "learning_rate": 0.00037316093146747887,
      "loss": 2.3527,
      "step": 97134
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.8899355530738831,
      "learning_rate": 0.00037315696440080637,
      "loss": 1.9966,
      "step": 97135
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.9113997220993042,
      "learning_rate": 0.00037315299732053266,
      "loss": 2.3335,
      "step": 97136
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.7663559913635254,
      "learning_rate": 0.00037314903022665834,
      "loss": 2.05,
      "step": 97137
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.8862844705581665,
      "learning_rate": 0.00037314506311918427,
      "loss": 2.3275,
      "step": 97138
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.910441517829895,
      "learning_rate": 0.0003731410959981111,
      "loss": 2.0784,
      "step": 97139
    },
    {
      "epoch": 1.26,
      "grad_norm": 1.0259990692138672,
      "learning_rate": 0.0003731371288634397,
      "loss": 2.1876,
      "step": 97140
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.9141934514045715,
      "learning_rate": 0.0003731331617151707,
      "loss": 2.1926,
      "step": 97141
    },
    {
      "epoch": 1.26,
      "grad_norm": 1.1047468185424805,
      "learning_rate": 0.0003731291945533049,
      "loss": 2.097,
      "step": 97142
    },
    {
      "epoch": 1.26,
      "grad_norm": 1.2169960737228394,
      "learning_rate": 0.00037312522737784304,
      "loss": 2.4079,
      "step": 97143
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.8141289353370667,
      "learning_rate": 0.0003731212601887857,
      "loss": 2.0606,
      "step": 97144
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.9581187963485718,
      "learning_rate": 0.0003731172929861338,
      "loss": 2.1471,
      "step": 97145
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.9041684865951538,
      "learning_rate": 0.000373113325769888,
      "loss": 2.1363,
      "step": 97146
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.9149066805839539,
      "learning_rate": 0.00037310935854004904,
      "loss": 2.1504,
      "step": 97147
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.9175355434417725,
      "learning_rate": 0.00037310539129661765,
      "loss": 2.115,
      "step": 97148
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.9700149297714233,
      "learning_rate": 0.0003731014240395947,
      "loss": 2.1754,
      "step": 97149
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.8755766153335571,
      "learning_rate": 0.0003730974567689807,
      "loss": 2.13,
      "step": 97150
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.9968675374984741,
      "learning_rate": 0.0003730934894847765,
      "loss": 2.4901,
      "step": 97151
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.9096992611885071,
      "learning_rate": 0.00037308952218698295,
      "loss": 2.0973,
      "step": 97152
    },
    {
      "epoch": 1.26,
      "grad_norm": 0.811398983001709,
      "learning_rate": 0.0003730855548756006,
      "loss": 2.0782,
      "step": 97153
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.8912457823753357,
      "learning_rate": 0.00037308158755063026,
      "loss": 2.1309,
      "step": 97154
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.931160032749176,
      "learning_rate": 0.0003730776202120727,
      "loss": 2.078,
      "step": 97155
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.8390951156616211,
      "learning_rate": 0.00037307365285992863,
      "loss": 2.1501,
      "step": 97156
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.8795733451843262,
      "learning_rate": 0.0003730696854941988,
      "loss": 2.3725,
      "step": 97157
    },
    {
      "epoch": 1.27,
      "grad_norm": 1.0127031803131104,
      "learning_rate": 0.0003730657181148839,
      "loss": 2.1929,
      "step": 97158
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.9444608092308044,
      "learning_rate": 0.0003730617507219847,
      "loss": 2.3053,
      "step": 97159
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.8869538307189941,
      "learning_rate": 0.00037305778331550197,
      "loss": 2.1775,
      "step": 97160
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.7900993824005127,
      "learning_rate": 0.0003730538158954364,
      "loss": 2.2444,
      "step": 97161
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.9104862809181213,
      "learning_rate": 0.0003730498484617888,
      "loss": 2.1796,
      "step": 97162
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.7946403622627258,
      "learning_rate": 0.00037304588101455977,
      "loss": 2.2854,
      "step": 97163
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.8789615631103516,
      "learning_rate": 0.0003730419135537502,
      "loss": 2.139,
      "step": 97164
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.8402348756790161,
      "learning_rate": 0.00037303794607936074,
      "loss": 2.0157,
      "step": 97165
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.9031738042831421,
      "learning_rate": 0.0003730339785913921,
      "loss": 2.114,
      "step": 97166
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.7977598905563354,
      "learning_rate": 0.0003730300110898452,
      "loss": 2.2066,
      "step": 97167
    },
    {
      "epoch": 1.27,
      "grad_norm": 1.0021867752075195,
      "learning_rate": 0.0003730260435747205,
      "loss": 2.1778,
      "step": 97168
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.8996961712837219,
      "learning_rate": 0.00037302207604601885,
      "loss": 1.9686,
      "step": 97169
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.8186383843421936,
      "learning_rate": 0.0003730181085037412,
      "loss": 1.9712,
      "step": 97170
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.8670438528060913,
      "learning_rate": 0.000373014140947888,
      "loss": 2.1277,
      "step": 97171
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.8371589779853821,
      "learning_rate": 0.00037301017337846003,
      "loss": 2.2045,
      "step": 97172
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.8459741473197937,
      "learning_rate": 0.00037300620579545815,
      "loss": 1.872,
      "step": 97173
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.8435482382774353,
      "learning_rate": 0.0003730022381988831,
      "loss": 2.1721,
      "step": 97174
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.9093371629714966,
      "learning_rate": 0.0003729982705887355,
      "loss": 2.154,
      "step": 97175
    },
    {
      "epoch": 1.27,
      "grad_norm": 1.0128837823867798,
      "learning_rate": 0.00037299430296501617,
      "loss": 2.2937,
      "step": 97176
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.8365377187728882,
      "learning_rate": 0.00037299033532772573,
      "loss": 2.2387,
      "step": 97177
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.8803156018257141,
      "learning_rate": 0.00037298636767686504,
      "loss": 2.1945,
      "step": 97178
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.8382694125175476,
      "learning_rate": 0.0003729824000124349,
      "loss": 2.1971,
      "step": 97179
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.8502631187438965,
      "learning_rate": 0.0003729784323344359,
      "loss": 2.3834,
      "step": 97180
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.8340789675712585,
      "learning_rate": 0.0003729744646428687,
      "loss": 2.2007,
      "step": 97181
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.848143994808197,
      "learning_rate": 0.0003729704969377344,
      "loss": 2.2582,
      "step": 97182
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.9603206515312195,
      "learning_rate": 0.00037296652921903336,
      "loss": 2.0148,
      "step": 97183
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.8311928510665894,
      "learning_rate": 0.0003729625614867665,
      "loss": 2.3106,
      "step": 97184
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.8535539507865906,
      "learning_rate": 0.0003729585937409346,
      "loss": 2.2687,
      "step": 97185
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.8581302165985107,
      "learning_rate": 0.00037295462598153825,
      "loss": 1.945,
      "step": 97186
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.8395888805389404,
      "learning_rate": 0.0003729506582085782,
      "loss": 2.1692,
      "step": 97187
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.875378429889679,
      "learning_rate": 0.0003729466904220553,
      "loss": 2.0857,
      "step": 97188
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.9129027128219604,
      "learning_rate": 0.00037294272262197026,
      "loss": 1.9343,
      "step": 97189
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.8784631490707397,
      "learning_rate": 0.00037293875480832377,
      "loss": 2.2223,
      "step": 97190
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.8842375874519348,
      "learning_rate": 0.00037293478698111656,
      "loss": 2.055,
      "step": 97191
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.8552897572517395,
      "learning_rate": 0.0003729308191403494,
      "loss": 2.0796,
      "step": 97192
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.8883494138717651,
      "learning_rate": 0.00037292685128602305,
      "loss": 2.1007,
      "step": 97193
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.8798695802688599,
      "learning_rate": 0.0003729228834181382,
      "loss": 2.2175,
      "step": 97194
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.8929063677787781,
      "learning_rate": 0.0003729189155366957,
      "loss": 2.1742,
      "step": 97195
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.8547946810722351,
      "learning_rate": 0.0003729149476416961,
      "loss": 2.4068,
      "step": 97196
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.8912805318832397,
      "learning_rate": 0.0003729109797331402,
      "loss": 2.173,
      "step": 97197
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.8720017671585083,
      "learning_rate": 0.0003729070118110289,
      "loss": 2.1119,
      "step": 97198
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.8571404814720154,
      "learning_rate": 0.0003729030438753627,
      "loss": 2.0564,
      "step": 97199
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.8739292025566101,
      "learning_rate": 0.0003728990759261425,
      "loss": 2.2114,
      "step": 97200
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.8499115705490112,
      "learning_rate": 0.000372895107963369,
      "loss": 2.3918,
      "step": 97201
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.9151257872581482,
      "learning_rate": 0.0003728911399870429,
      "loss": 2.3864,
      "step": 97202
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.829796314239502,
      "learning_rate": 0.0003728871719971649,
      "loss": 2.1166,
      "step": 97203
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.8999640345573425,
      "learning_rate": 0.0003728832039937359,
      "loss": 2.1843,
      "step": 97204
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.8950218558311462,
      "learning_rate": 0.0003728792359767565,
      "loss": 2.2547,
      "step": 97205
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.8710623383522034,
      "learning_rate": 0.00037287526794622743,
      "loss": 2.0966,
      "step": 97206
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.7952364087104797,
      "learning_rate": 0.0003728712999021496,
      "loss": 1.9989,
      "step": 97207
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.8765355944633484,
      "learning_rate": 0.00037286733184452346,
      "loss": 2.0683,
      "step": 97208
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.8576532602310181,
      "learning_rate": 0.00037286336377335,
      "loss": 2.1368,
      "step": 97209
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.8049238324165344,
      "learning_rate": 0.0003728593956886299,
      "loss": 2.2497,
      "step": 97210
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.8284928798675537,
      "learning_rate": 0.00037285542759036373,
      "loss": 1.8423,
      "step": 97211
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.8273950219154358,
      "learning_rate": 0.0003728514594785524,
      "loss": 2.1701,
      "step": 97212
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.8502134084701538,
      "learning_rate": 0.0003728474913531967,
      "loss": 2.3998,
      "step": 97213
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.8067120313644409,
      "learning_rate": 0.0003728435232142972,
      "loss": 2.0946,
      "step": 97214
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.9381849765777588,
      "learning_rate": 0.00037283955506185474,
      "loss": 2.3097,
      "step": 97215
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.8879900574684143,
      "learning_rate": 0.00037283558689587003,
      "loss": 2.099,
      "step": 97216
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.8711227178573608,
      "learning_rate": 0.00037283161871634385,
      "loss": 2.138,
      "step": 97217
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.9789298176765442,
      "learning_rate": 0.00037282765052327685,
      "loss": 1.9752,
      "step": 97218
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.8110612630844116,
      "learning_rate": 0.0003728236823166698,
      "loss": 2.1176,
      "step": 97219
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.8760451674461365,
      "learning_rate": 0.00037281971409652343,
      "loss": 2.1377,
      "step": 97220
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.8287932872772217,
      "learning_rate": 0.0003728157458628386,
      "loss": 2.1212,
      "step": 97221
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.9276020526885986,
      "learning_rate": 0.00037281177761561585,
      "loss": 2.1581,
      "step": 97222
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.8112326860427856,
      "learning_rate": 0.00037280780935485604,
      "loss": 1.8135,
      "step": 97223
    },
    {
      "epoch": 1.27,
      "grad_norm": 1.1432932615280151,
      "learning_rate": 0.00037280384108055996,
      "loss": 2.1428,
      "step": 97224
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.7371507287025452,
      "learning_rate": 0.00037279987279272817,
      "loss": 1.8802,
      "step": 97225
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.7672635912895203,
      "learning_rate": 0.0003727959044913615,
      "loss": 2.2223,
      "step": 97226
    },
    {
      "epoch": 1.27,
      "grad_norm": 1.0377923250198364,
      "learning_rate": 0.00037279193617646076,
      "loss": 2.4153,
      "step": 97227
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.7925582528114319,
      "learning_rate": 0.0003727879678480266,
      "loss": 2.0587,
      "step": 97228
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.9139032959938049,
      "learning_rate": 0.0003727839995060598,
      "loss": 2.1475,
      "step": 97229
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.8592064380645752,
      "learning_rate": 0.0003727800311505611,
      "loss": 2.1791,
      "step": 97230
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.9087476134300232,
      "learning_rate": 0.00037277606278153123,
      "loss": 2.3264,
      "step": 97231
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.8310891389846802,
      "learning_rate": 0.00037277209439897085,
      "loss": 2.1183,
      "step": 97232
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.7886747121810913,
      "learning_rate": 0.0003727681260028808,
      "loss": 2.0142,
      "step": 97233
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.781999409198761,
      "learning_rate": 0.00037276415759326183,
      "loss": 2.2154,
      "step": 97234
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.8101428747177124,
      "learning_rate": 0.0003727601891701145,
      "loss": 2.015,
      "step": 97235
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.8325276374816895,
      "learning_rate": 0.0003727562207334398,
      "loss": 1.9752,
      "step": 97236
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.9020853638648987,
      "learning_rate": 0.0003727522522832383,
      "loss": 2.3584,
      "step": 97237
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.8284746408462524,
      "learning_rate": 0.00037274828381951076,
      "loss": 2.1872,
      "step": 97238
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.8946000337600708,
      "learning_rate": 0.00037274431534225794,
      "loss": 2.0598,
      "step": 97239
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.907071590423584,
      "learning_rate": 0.0003727403468514807,
      "loss": 2.3111,
      "step": 97240
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.8852397203445435,
      "learning_rate": 0.0003727363783471795,
      "loss": 2.223,
      "step": 97241
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.9348841309547424,
      "learning_rate": 0.00037273240982935526,
      "loss": 2.0822,
      "step": 97242
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.9086207151412964,
      "learning_rate": 0.0003727284412980088,
      "loss": 2.1818,
      "step": 97243
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.9680471420288086,
      "learning_rate": 0.0003727244727531407,
      "loss": 2.2736,
      "step": 97244
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.8218459486961365,
      "learning_rate": 0.00037272050419475165,
      "loss": 2.1284,
      "step": 97245
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.8671156167984009,
      "learning_rate": 0.0003727165356228426,
      "loss": 2.1406,
      "step": 97246
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.9033741354942322,
      "learning_rate": 0.0003727125670374141,
      "loss": 1.9753,
      "step": 97247
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.9865826964378357,
      "learning_rate": 0.000372708598438467,
      "loss": 2.526,
      "step": 97248
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.8108954429626465,
      "learning_rate": 0.000372704629826002,
      "loss": 1.7775,
      "step": 97249
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.858688473701477,
      "learning_rate": 0.0003727006612000198,
      "loss": 2.1116,
      "step": 97250
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.892068088054657,
      "learning_rate": 0.00037269669256052124,
      "loss": 2.531,
      "step": 97251
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.8578595519065857,
      "learning_rate": 0.00037269272390750695,
      "loss": 2.1566,
      "step": 97252
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.8597835302352905,
      "learning_rate": 0.00037268875524097776,
      "loss": 2.2028,
      "step": 97253
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.9032580852508545,
      "learning_rate": 0.00037268478656093425,
      "loss": 2.195,
      "step": 97254
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.9138942360877991,
      "learning_rate": 0.0003726808178673773,
      "loss": 2.2592,
      "step": 97255
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.9331119656562805,
      "learning_rate": 0.0003726768491603077,
      "loss": 2.1222,
      "step": 97256
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.8131568431854248,
      "learning_rate": 0.000372672880439726,
      "loss": 2.4815,
      "step": 97257
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.9580865502357483,
      "learning_rate": 0.0003726689117056331,
      "loss": 2.2921,
      "step": 97258
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.8483872413635254,
      "learning_rate": 0.00037266494295802973,
      "loss": 2.1756,
      "step": 97259
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.8540382981300354,
      "learning_rate": 0.00037266097419691647,
      "loss": 2.1295,
      "step": 97260
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.8727170825004578,
      "learning_rate": 0.0003726570054222942,
      "loss": 2.048,
      "step": 97261
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.8436409831047058,
      "learning_rate": 0.00037265303663416366,
      "loss": 2.2989,
      "step": 97262
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.8167213201522827,
      "learning_rate": 0.0003726490678325255,
      "loss": 2.0067,
      "step": 97263
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.8654354810714722,
      "learning_rate": 0.0003726450990173805,
      "loss": 2.2201,
      "step": 97264
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.8068645000457764,
      "learning_rate": 0.0003726411301887295,
      "loss": 2.0342,
      "step": 97265
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.9913325309753418,
      "learning_rate": 0.00037263716134657303,
      "loss": 2.0162,
      "step": 97266
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.8253157734870911,
      "learning_rate": 0.0003726331924909119,
      "loss": 2.3678,
      "step": 97267
    },
    {
      "epoch": 1.27,
      "grad_norm": 1.193830132484436,
      "learning_rate": 0.00037262922362174707,
      "loss": 2.1454,
      "step": 97268
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.9114087820053101,
      "learning_rate": 0.00037262525473907897,
      "loss": 1.93,
      "step": 97269
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.8799310326576233,
      "learning_rate": 0.00037262128584290843,
      "loss": 2.0782,
      "step": 97270
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.8202173709869385,
      "learning_rate": 0.00037261731693323633,
      "loss": 2.0158,
      "step": 97271
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.7850794196128845,
      "learning_rate": 0.0003726133480100632,
      "loss": 2.1255,
      "step": 97272
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.7579099535942078,
      "learning_rate": 0.00037260937907338996,
      "loss": 2.0438,
      "step": 97273
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.8776741623878479,
      "learning_rate": 0.00037260541012321726,
      "loss": 2.0876,
      "step": 97274
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.8107260465621948,
      "learning_rate": 0.0003726014411595458,
      "loss": 2.2014,
      "step": 97275
    },
    {
      "epoch": 1.27,
      "grad_norm": 1.0084789991378784,
      "learning_rate": 0.0003725974721823763,
      "loss": 2.3315,
      "step": 97276
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.9467636942863464,
      "learning_rate": 0.00037259350319170976,
      "loss": 2.0917,
      "step": 97277
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.9149200916290283,
      "learning_rate": 0.00037258953418754657,
      "loss": 2.0107,
      "step": 97278
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.9430275559425354,
      "learning_rate": 0.0003725855651698876,
      "loss": 2.3202,
      "step": 97279
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.9272522926330566,
      "learning_rate": 0.0003725815961387337,
      "loss": 2.1671,
      "step": 97280
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.7849472165107727,
      "learning_rate": 0.00037257762709408545,
      "loss": 2.0416,
      "step": 97281
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.9375802278518677,
      "learning_rate": 0.00037257365803594367,
      "loss": 2.2502,
      "step": 97282
    },
    {
      "epoch": 1.27,
      "grad_norm": 1.0370771884918213,
      "learning_rate": 0.0003725696889643091,
      "loss": 2.2215,
      "step": 97283
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.8846195340156555,
      "learning_rate": 0.0003725657198791824,
      "loss": 1.9443,
      "step": 97284
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.8716344237327576,
      "learning_rate": 0.0003725617507805644,
      "loss": 2.4073,
      "step": 97285
    },
    {
      "epoch": 1.27,
      "grad_norm": 1.419765591621399,
      "learning_rate": 0.0003725577816684558,
      "loss": 1.9314,
      "step": 97286
    },
    {
      "epoch": 1.27,
      "grad_norm": 1.0662058591842651,
      "learning_rate": 0.00037255381254285724,
      "loss": 2.096,
      "step": 97287
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.8337138891220093,
      "learning_rate": 0.00037254984340376963,
      "loss": 1.9665,
      "step": 97288
    },
    {
      "epoch": 1.27,
      "grad_norm": 1.1067248582839966,
      "learning_rate": 0.0003725458742511937,
      "loss": 2.1002,
      "step": 97289
    },
    {
      "epoch": 1.27,
      "grad_norm": 1.5747740268707275,
      "learning_rate": 0.00037254190508513,
      "loss": 1.8818,
      "step": 97290
    },
    {
      "epoch": 1.27,
      "grad_norm": 1.0703836679458618,
      "learning_rate": 0.0003725379359055795,
      "loss": 2.2064,
      "step": 97291
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.9034708142280579,
      "learning_rate": 0.0003725339667125428,
      "loss": 2.0461,
      "step": 97292
    },
    {
      "epoch": 1.27,
      "grad_norm": 1.2092124223709106,
      "learning_rate": 0.0003725299975060206,
      "loss": 2.0652,
      "step": 97293
    },
    {
      "epoch": 1.27,
      "grad_norm": 1.1236672401428223,
      "learning_rate": 0.00037252602828601375,
      "loss": 2.1567,
      "step": 97294
    },
    {
      "epoch": 1.27,
      "grad_norm": 1.1398438215255737,
      "learning_rate": 0.000372522059052523,
      "loss": 2.2627,
      "step": 97295
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.9309415221214294,
      "learning_rate": 0.00037251808980554897,
      "loss": 2.3597,
      "step": 97296
    },
    {
      "epoch": 1.27,
      "grad_norm": 1.1282669305801392,
      "learning_rate": 0.00037251412054509247,
      "loss": 2.1524,
      "step": 97297
    },
    {
      "epoch": 1.27,
      "grad_norm": 1.0987275838851929,
      "learning_rate": 0.0003725101512711542,
      "loss": 2.144,
      "step": 97298
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.8500550389289856,
      "learning_rate": 0.000372506181983735,
      "loss": 2.1218,
      "step": 97299
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.895517885684967,
      "learning_rate": 0.0003725022126828354,
      "loss": 2.2021,
      "step": 97300
    },
    {
      "epoch": 1.27,
      "grad_norm": 1.0864473581314087,
      "learning_rate": 0.00037249824336845627,
      "loss": 2.0521,
      "step": 97301
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.9081149697303772,
      "learning_rate": 0.00037249427404059855,
      "loss": 2.0084,
      "step": 97302
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.8532822728157043,
      "learning_rate": 0.00037249030469926257,
      "loss": 2.3053,
      "step": 97303
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.9675427079200745,
      "learning_rate": 0.00037248633534444936,
      "loss": 2.1579,
      "step": 97304
    },
    {
      "epoch": 1.27,
      "grad_norm": 1.1155556440353394,
      "learning_rate": 0.00037248236597615963,
      "loss": 2.4079,
      "step": 97305
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.8442215323448181,
      "learning_rate": 0.00037247839659439397,
      "loss": 2.2392,
      "step": 97306
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.8099876046180725,
      "learning_rate": 0.0003724744271991532,
      "loss": 2.064,
      "step": 97307
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.9422726035118103,
      "learning_rate": 0.00037247045779043806,
      "loss": 1.8774,
      "step": 97308
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.9948776364326477,
      "learning_rate": 0.0003724664883682494,
      "loss": 1.9842,
      "step": 97309
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.7713310122489929,
      "learning_rate": 0.00037246251893258776,
      "loss": 2.2609,
      "step": 97310
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.9568735361099243,
      "learning_rate": 0.00037245854948345395,
      "loss": 2.3113,
      "step": 97311
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.9139445424079895,
      "learning_rate": 0.00037245458002084877,
      "loss": 2.0491,
      "step": 97312
    },
    {
      "epoch": 1.27,
      "grad_norm": 1.0358984470367432,
      "learning_rate": 0.00037245061054477297,
      "loss": 2.2007,
      "step": 97313
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.7673754096031189,
      "learning_rate": 0.0003724466410552272,
      "loss": 2.0211,
      "step": 97314
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.9471326470375061,
      "learning_rate": 0.0003724426715522122,
      "loss": 2.2482,
      "step": 97315
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.9078440070152283,
      "learning_rate": 0.0003724387020357288,
      "loss": 2.2334,
      "step": 97316
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.9138370752334595,
      "learning_rate": 0.0003724347325057776,
      "loss": 2.3166,
      "step": 97317
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.783518373966217,
      "learning_rate": 0.0003724307629623594,
      "loss": 2.1252,
      "step": 97318
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.921012282371521,
      "learning_rate": 0.0003724267934054751,
      "loss": 2.1965,
      "step": 97319
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.8528754115104675,
      "learning_rate": 0.00037242282383512514,
      "loss": 1.9344,
      "step": 97320
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.8627349138259888,
      "learning_rate": 0.0003724188542513105,
      "loss": 2.1884,
      "step": 97321
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.8867339491844177,
      "learning_rate": 0.0003724148846540317,
      "loss": 2.23,
      "step": 97322
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.8781380653381348,
      "learning_rate": 0.0003724109150432897,
      "loss": 2.2205,
      "step": 97323
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.8131891489028931,
      "learning_rate": 0.0003724069454190852,
      "loss": 2.2838,
      "step": 97324
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.8533205389976501,
      "learning_rate": 0.0003724029757814188,
      "loss": 1.8433,
      "step": 97325
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.9109026789665222,
      "learning_rate": 0.0003723990061302914,
      "loss": 1.9157,
      "step": 97326
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.8810418248176575,
      "learning_rate": 0.00037239503646570353,
      "loss": 2.1184,
      "step": 97327
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.9512127041816711,
      "learning_rate": 0.00037239106678765615,
      "loss": 2.1097,
      "step": 97328
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.8628310561180115,
      "learning_rate": 0.00037238709709614987,
      "loss": 2.0247,
      "step": 97329
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.8141366243362427,
      "learning_rate": 0.0003723831273911855,
      "loss": 2.3587,
      "step": 97330
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.8069188594818115,
      "learning_rate": 0.0003723791576727637,
      "loss": 2.1091,
      "step": 97331
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.9767966866493225,
      "learning_rate": 0.0003723751879408853,
      "loss": 2.2522,
      "step": 97332
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.9236155152320862,
      "learning_rate": 0.00037237121819555096,
      "loss": 2.1757,
      "step": 97333
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.9880012273788452,
      "learning_rate": 0.0003723672484367614,
      "loss": 1.9284,
      "step": 97334
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.7916489243507385,
      "learning_rate": 0.00037236327866451747,
      "loss": 2.2319,
      "step": 97335
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.9918246269226074,
      "learning_rate": 0.00037235930887881977,
      "loss": 2.0222,
      "step": 97336
    },
    {
      "epoch": 1.27,
      "grad_norm": 1.006821870803833,
      "learning_rate": 0.0003723553390796692,
      "loss": 2.0887,
      "step": 97337
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.880234956741333,
      "learning_rate": 0.00037235136926706635,
      "loss": 2.2315,
      "step": 97338
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.8988878726959229,
      "learning_rate": 0.000372347399441012,
      "loss": 2.2018,
      "step": 97339
    },
    {
      "epoch": 1.27,
      "grad_norm": 1.056298851966858,
      "learning_rate": 0.0003723434296015069,
      "loss": 2.0841,
      "step": 97340
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.8238789439201355,
      "learning_rate": 0.0003723394597485519,
      "loss": 2.2985,
      "step": 97341
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.9616500735282898,
      "learning_rate": 0.0003723354898821475,
      "loss": 2.2469,
      "step": 97342
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.8858741521835327,
      "learning_rate": 0.0003723315200022946,
      "loss": 2.0116,
      "step": 97343
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.9377516508102417,
      "learning_rate": 0.00037232755010899394,
      "loss": 2.1936,
      "step": 97344
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.8778926134109497,
      "learning_rate": 0.00037232358020224614,
      "loss": 2.1045,
      "step": 97345
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.8668817281723022,
      "learning_rate": 0.0003723196102820521,
      "loss": 2.131,
      "step": 97346
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.847939133644104,
      "learning_rate": 0.00037231564034841244,
      "loss": 2.1365,
      "step": 97347
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.8835625052452087,
      "learning_rate": 0.000372311670401328,
      "loss": 2.2695,
      "step": 97348
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.8768340945243835,
      "learning_rate": 0.0003723077004407993,
      "loss": 2.2062,
      "step": 97349
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.8769230842590332,
      "learning_rate": 0.0003723037304668274,
      "loss": 2.1966,
      "step": 97350
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.8966086506843567,
      "learning_rate": 0.0003722997604794128,
      "loss": 2.3352,
      "step": 97351
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.9097976684570312,
      "learning_rate": 0.00037229579047855634,
      "loss": 2.1638,
      "step": 97352
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.8934652209281921,
      "learning_rate": 0.0003722918204642587,
      "loss": 2.378,
      "step": 97353
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.8172905445098877,
      "learning_rate": 0.0003722878504365207,
      "loss": 2.3064,
      "step": 97354
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.7889859080314636,
      "learning_rate": 0.000372283880395343,
      "loss": 1.9112,
      "step": 97355
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.9258763790130615,
      "learning_rate": 0.00037227991034072635,
      "loss": 2.1531,
      "step": 97356
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.92453533411026,
      "learning_rate": 0.0003722759402726715,
      "loss": 2.3372,
      "step": 97357
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.9102230668067932,
      "learning_rate": 0.00037227197019117916,
      "loss": 2.148,
      "step": 97358
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.924156129360199,
      "learning_rate": 0.00037226800009625014,
      "loss": 2.4399,
      "step": 97359
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.8498280644416809,
      "learning_rate": 0.0003722640299878851,
      "loss": 2.2664,
      "step": 97360
    },
    {
      "epoch": 1.27,
      "grad_norm": 1.024275541305542,
      "learning_rate": 0.0003722600598660848,
      "loss": 2.1779,
      "step": 97361
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.767848789691925,
      "learning_rate": 0.00037225608973085004,
      "loss": 1.9661,
      "step": 97362
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.9950034618377686,
      "learning_rate": 0.0003722521195821815,
      "loss": 2.1456,
      "step": 97363
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.9130331873893738,
      "learning_rate": 0.0003722481494200799,
      "loss": 2.1229,
      "step": 97364
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.9608629941940308,
      "learning_rate": 0.00037224417924454605,
      "loss": 2.265,
      "step": 97365
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.9367518424987793,
      "learning_rate": 0.0003722402090555806,
      "loss": 2.2063,
      "step": 97366
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.94756680727005,
      "learning_rate": 0.00037223623885318427,
      "loss": 2.1284,
      "step": 97367
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.8513237237930298,
      "learning_rate": 0.00037223226863735803,
      "loss": 2.1127,
      "step": 97368
    },
    {
      "epoch": 1.27,
      "grad_norm": 1.0129139423370361,
      "learning_rate": 0.00037222829840810227,
      "loss": 2.1252,
      "step": 97369
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.9395594000816345,
      "learning_rate": 0.000372224328165418,
      "loss": 2.4907,
      "step": 97370
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.8332235217094421,
      "learning_rate": 0.0003722203579093059,
      "loss": 2.1167,
      "step": 97371
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.9004045128822327,
      "learning_rate": 0.0003722163876397666,
      "loss": 2.1061,
      "step": 97372
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.9183700084686279,
      "learning_rate": 0.00037221241735680094,
      "loss": 2.2056,
      "step": 97373
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.8732370138168335,
      "learning_rate": 0.0003722084470604097,
      "loss": 2.0523,
      "step": 97374
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.9148663878440857,
      "learning_rate": 0.00037220447675059343,
      "loss": 2.0562,
      "step": 97375
    },
    {
      "epoch": 1.27,
      "grad_norm": 1.0194140672683716,
      "learning_rate": 0.000372200506427353,
      "loss": 1.975,
      "step": 97376
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.801531970500946,
      "learning_rate": 0.00037219653609068913,
      "loss": 2.2203,
      "step": 97377
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.9755275249481201,
      "learning_rate": 0.00037219256574060256,
      "loss": 2.1862,
      "step": 97378
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.9999444484710693,
      "learning_rate": 0.0003721885953770941,
      "loss": 2.1271,
      "step": 97379
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.8576372265815735,
      "learning_rate": 0.0003721846250001644,
      "loss": 2.1799,
      "step": 97380
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.8426454067230225,
      "learning_rate": 0.00037218065460981417,
      "loss": 1.9731,
      "step": 97381
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.9277846813201904,
      "learning_rate": 0.0003721766842060442,
      "loss": 2.2961,
      "step": 97382
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.9237218499183655,
      "learning_rate": 0.0003721727137888553,
      "loss": 2.0602,
      "step": 97383
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.89432692527771,
      "learning_rate": 0.00037216874335824804,
      "loss": 2.3023,
      "step": 97384
    },
    {
      "epoch": 1.27,
      "grad_norm": 1.1356443166732788,
      "learning_rate": 0.00037216477291422326,
      "loss": 2.3305,
      "step": 97385
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.8515768647193909,
      "learning_rate": 0.0003721608024567817,
      "loss": 2.2067,
      "step": 97386
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.805368185043335,
      "learning_rate": 0.0003721568319859241,
      "loss": 1.9839,
      "step": 97387
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.873134195804596,
      "learning_rate": 0.00037215286150165116,
      "loss": 2.2783,
      "step": 97388
    },
    {
      "epoch": 1.27,
      "grad_norm": 1.1507396697998047,
      "learning_rate": 0.00037214889100396363,
      "loss": 1.9177,
      "step": 97389
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.8569852113723755,
      "learning_rate": 0.0003721449204928623,
      "loss": 2.0963,
      "step": 97390
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.9725026488304138,
      "learning_rate": 0.0003721409499683479,
      "loss": 2.08,
      "step": 97391
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.9277521967887878,
      "learning_rate": 0.000372136979430421,
      "loss": 2.1377,
      "step": 97392
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.8473777770996094,
      "learning_rate": 0.0003721330088790825,
      "loss": 2.2243,
      "step": 97393
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.7785001993179321,
      "learning_rate": 0.0003721290383143333,
      "loss": 2.0465,
      "step": 97394
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.9390265345573425,
      "learning_rate": 0.00037212506773617377,
      "loss": 2.1219,
      "step": 97395
    },
    {
      "epoch": 1.27,
      "grad_norm": 1.0692671537399292,
      "learning_rate": 0.00037212109714460486,
      "loss": 2.1999,
      "step": 97396
    },
    {
      "epoch": 1.27,
      "grad_norm": 1.8794124126434326,
      "learning_rate": 0.0003721171265396273,
      "loss": 2.1662,
      "step": 97397
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.837822675704956,
      "learning_rate": 0.00037211315592124177,
      "loss": 2.2232,
      "step": 97398
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.9428783655166626,
      "learning_rate": 0.00037210918528944906,
      "loss": 2.0652,
      "step": 97399
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.8863013386726379,
      "learning_rate": 0.00037210521464424987,
      "loss": 1.8985,
      "step": 97400
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.8803994655609131,
      "learning_rate": 0.00037210124398564506,
      "loss": 2.1901,
      "step": 97401
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.8421145677566528,
      "learning_rate": 0.0003720972733136352,
      "loss": 2.3011,
      "step": 97402
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.9743654727935791,
      "learning_rate": 0.0003720933026282211,
      "loss": 2.262,
      "step": 97403
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.8161494135856628,
      "learning_rate": 0.0003720893319294035,
      "loss": 2.2104,
      "step": 97404
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.8552822470664978,
      "learning_rate": 0.0003720853612171831,
      "loss": 2.1474,
      "step": 97405
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.8701430559158325,
      "learning_rate": 0.00037208139049156065,
      "loss": 2.3037,
      "step": 97406
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.8640089631080627,
      "learning_rate": 0.00037207741975253694,
      "loss": 2.2658,
      "step": 97407
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.8350055813789368,
      "learning_rate": 0.00037207344900011275,
      "loss": 2.1145,
      "step": 97408
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.8278467059135437,
      "learning_rate": 0.00037206947823428864,
      "loss": 2.1689,
      "step": 97409
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.8618268370628357,
      "learning_rate": 0.00037206550745506547,
      "loss": 2.3299,
      "step": 97410
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.8063551783561707,
      "learning_rate": 0.00037206153666244406,
      "loss": 2.0475,
      "step": 97411
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.7851836085319519,
      "learning_rate": 0.00037205756585642494,
      "loss": 2.3087,
      "step": 97412
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.8063056468963623,
      "learning_rate": 0.000372053595037009,
      "loss": 1.9875,
      "step": 97413
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.8868807554244995,
      "learning_rate": 0.00037204962420419696,
      "loss": 2.2417,
      "step": 97414
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.8682365417480469,
      "learning_rate": 0.0003720456533579895,
      "loss": 1.9188,
      "step": 97415
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.9307907223701477,
      "learning_rate": 0.0003720416824983874,
      "loss": 2.1622,
      "step": 97416
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.8357630968093872,
      "learning_rate": 0.0003720377116253914,
      "loss": 2.2961,
      "step": 97417
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.7482035756111145,
      "learning_rate": 0.0003720337407390023,
      "loss": 2.0356,
      "step": 97418
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.8486887216567993,
      "learning_rate": 0.0003720297698392206,
      "loss": 2.2415,
      "step": 97419
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.872463047504425,
      "learning_rate": 0.00037202579892604734,
      "loss": 2.2817,
      "step": 97420
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.9562608003616333,
      "learning_rate": 0.0003720218279994831,
      "loss": 2.3565,
      "step": 97421
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.9795603156089783,
      "learning_rate": 0.00037201785705952853,
      "loss": 2.1742,
      "step": 97422
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.8974328637123108,
      "learning_rate": 0.00037201388610618466,
      "loss": 2.2326,
      "step": 97423
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.7414222955703735,
      "learning_rate": 0.0003720099151394519,
      "loss": 2.0664,
      "step": 97424
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.9287363886833191,
      "learning_rate": 0.00037200594415933124,
      "loss": 2.2626,
      "step": 97425
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.808506190776825,
      "learning_rate": 0.0003720019731658233,
      "loss": 1.9885,
      "step": 97426
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.8124724626541138,
      "learning_rate": 0.0003719980021589288,
      "loss": 2.1647,
      "step": 97427
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.8704129457473755,
      "learning_rate": 0.0003719940311386485,
      "loss": 1.9122,
      "step": 97428
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.7999243140220642,
      "learning_rate": 0.0003719900601049832,
      "loss": 2.1286,
      "step": 97429
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.9532246589660645,
      "learning_rate": 0.0003719860890579336,
      "loss": 2.1407,
      "step": 97430
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.9233155846595764,
      "learning_rate": 0.00037198211799750037,
      "loss": 2.0992,
      "step": 97431
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.8202729225158691,
      "learning_rate": 0.0003719781469236843,
      "loss": 2.2364,
      "step": 97432
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.9439225196838379,
      "learning_rate": 0.00037197417583648617,
      "loss": 2.1638,
      "step": 97433
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.9794498682022095,
      "learning_rate": 0.00037197020473590667,
      "loss": 2.3787,
      "step": 97434
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.7871471643447876,
      "learning_rate": 0.0003719662336219466,
      "loss": 2.1091,
      "step": 97435
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.8675836324691772,
      "learning_rate": 0.00037196226249460655,
      "loss": 2.1587,
      "step": 97436
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.9426521062850952,
      "learning_rate": 0.00037195829135388745,
      "loss": 2.3007,
      "step": 97437
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.8856551051139832,
      "learning_rate": 0.0003719543201997899,
      "loss": 1.9743,
      "step": 97438
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.8106070160865784,
      "learning_rate": 0.0003719503490323147,
      "loss": 2.1268,
      "step": 97439
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.9240131378173828,
      "learning_rate": 0.00037194637785146253,
      "loss": 2.1496,
      "step": 97440
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.7928985953330994,
      "learning_rate": 0.0003719424066572342,
      "loss": 2.0087,
      "step": 97441
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.8434135317802429,
      "learning_rate": 0.00037193843544963044,
      "loss": 1.992,
      "step": 97442
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.8509232401847839,
      "learning_rate": 0.0003719344642286519,
      "loss": 2.2002,
      "step": 97443
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.905791699886322,
      "learning_rate": 0.0003719304929942995,
      "loss": 1.9189,
      "step": 97444
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.8055478930473328,
      "learning_rate": 0.00037192652174657375,
      "loss": 2.048,
      "step": 97445
    },
    {
      "epoch": 1.27,
      "grad_norm": 1.195087194442749,
      "learning_rate": 0.0003719225504854755,
      "loss": 2.2232,
      "step": 97446
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.9034953117370605,
      "learning_rate": 0.0003719185792110056,
      "loss": 2.5506,
      "step": 97447
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.8261107802391052,
      "learning_rate": 0.0003719146079231646,
      "loss": 1.9771,
      "step": 97448
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.8709319829940796,
      "learning_rate": 0.0003719106366219533,
      "loss": 2.3038,
      "step": 97449
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.8846036195755005,
      "learning_rate": 0.00037190666530737253,
      "loss": 2.0511,
      "step": 97450
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.8920002579689026,
      "learning_rate": 0.00037190269397942294,
      "loss": 2.1817,
      "step": 97451
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.8449590802192688,
      "learning_rate": 0.0003718987226381052,
      "loss": 2.156,
      "step": 97452
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.9724043011665344,
      "learning_rate": 0.00037189475128342013,
      "loss": 2.2211,
      "step": 97453
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.812658429145813,
      "learning_rate": 0.0003718907799153686,
      "loss": 1.9824,
      "step": 97454
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.8769766092300415,
      "learning_rate": 0.0003718868085339511,
      "loss": 2.0674,
      "step": 97455
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.8532787561416626,
      "learning_rate": 0.0003718828371391685,
      "loss": 2.0977,
      "step": 97456
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.8705229759216309,
      "learning_rate": 0.00037187886573102156,
      "loss": 2.4065,
      "step": 97457
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.848288357257843,
      "learning_rate": 0.000371874894309511,
      "loss": 2.1013,
      "step": 97458
    },
    {
      "epoch": 1.27,
      "grad_norm": 1.1765409708023071,
      "learning_rate": 0.0003718709228746375,
      "loss": 2.2366,
      "step": 97459
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.8715884685516357,
      "learning_rate": 0.0003718669514264019,
      "loss": 2.1357,
      "step": 97460
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.8640521168708801,
      "learning_rate": 0.00037186297996480476,
      "loss": 2.1887,
      "step": 97461
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.9177508354187012,
      "learning_rate": 0.000371859008489847,
      "loss": 2.0505,
      "step": 97462
    },
    {
      "epoch": 1.27,
      "grad_norm": 1.1531648635864258,
      "learning_rate": 0.0003718550370015293,
      "loss": 2.2286,
      "step": 97463
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.8601267337799072,
      "learning_rate": 0.00037185106549985237,
      "loss": 2.1738,
      "step": 97464
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.8866494297981262,
      "learning_rate": 0.000371847093984817,
      "loss": 2.2413,
      "step": 97465
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.8334510326385498,
      "learning_rate": 0.0003718431224564239,
      "loss": 2.1824,
      "step": 97466
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.8706845641136169,
      "learning_rate": 0.0003718391509146738,
      "loss": 2.4077,
      "step": 97467
    },
    {
      "epoch": 1.27,
      "grad_norm": 1.0197994709014893,
      "learning_rate": 0.0003718351793595674,
      "loss": 1.9298,
      "step": 97468
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.8126695156097412,
      "learning_rate": 0.0003718312077911056,
      "loss": 2.263,
      "step": 97469
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.9090316891670227,
      "learning_rate": 0.00037182723620928887,
      "loss": 2.2443,
      "step": 97470
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.9019102454185486,
      "learning_rate": 0.00037182326461411824,
      "loss": 2.4619,
      "step": 97471
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.9692159295082092,
      "learning_rate": 0.00037181929300559425,
      "loss": 2.2659,
      "step": 97472
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.87774658203125,
      "learning_rate": 0.00037181532138371766,
      "loss": 2.0648,
      "step": 97473
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.8936373591423035,
      "learning_rate": 0.0003718113497484893,
      "loss": 2.1771,
      "step": 97474
    },
    {
      "epoch": 1.27,
      "grad_norm": 1.0716904401779175,
      "learning_rate": 0.0003718073780999099,
      "loss": 2.2239,
      "step": 97475
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.8403726816177368,
      "learning_rate": 0.00037180340643798004,
      "loss": 2.2339,
      "step": 97476
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.990265429019928,
      "learning_rate": 0.00037179943476270064,
      "loss": 2.141,
      "step": 97477
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.9297761917114258,
      "learning_rate": 0.0003717954630740724,
      "loss": 2.136,
      "step": 97478
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.8281004428863525,
      "learning_rate": 0.00037179149137209596,
      "loss": 2.4097,
      "step": 97479
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.860230565071106,
      "learning_rate": 0.00037178751965677204,
      "loss": 2.2947,
      "step": 97480
    },
    {
      "epoch": 1.27,
      "grad_norm": 1.1194614171981812,
      "learning_rate": 0.0003717835479281016,
      "loss": 2.2796,
      "step": 97481
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.845379114151001,
      "learning_rate": 0.0003717795761860852,
      "loss": 2.1571,
      "step": 97482
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.833618700504303,
      "learning_rate": 0.00037177560443072365,
      "loss": 2.1662,
      "step": 97483
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.8730733394622803,
      "learning_rate": 0.0003717716326620176,
      "loss": 2.0801,
      "step": 97484
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.9641719460487366,
      "learning_rate": 0.000371767660879968,
      "loss": 2.0796,
      "step": 97485
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.8636353611946106,
      "learning_rate": 0.00037176368908457526,
      "loss": 1.9549,
      "step": 97486
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.8152925968170166,
      "learning_rate": 0.0003717597172758404,
      "loss": 1.8802,
      "step": 97487
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.8417961597442627,
      "learning_rate": 0.00037175574545376396,
      "loss": 2.1701,
      "step": 97488
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.9523545503616333,
      "learning_rate": 0.00037175177361834687,
      "loss": 2.1437,
      "step": 97489
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.9135241508483887,
      "learning_rate": 0.0003717478017695897,
      "loss": 2.4522,
      "step": 97490
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.747848391532898,
      "learning_rate": 0.00037174382990749327,
      "loss": 1.9968,
      "step": 97491
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.88233882188797,
      "learning_rate": 0.0003717398580320583,
      "loss": 2.3873,
      "step": 97492
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.8423033952713013,
      "learning_rate": 0.00037173588614328564,
      "loss": 2.0478,
      "step": 97493
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.8820106983184814,
      "learning_rate": 0.00037173191424117586,
      "loss": 2.131,
      "step": 97494
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.7771198749542236,
      "learning_rate": 0.0003717279423257297,
      "loss": 2.0688,
      "step": 97495
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.9270646572113037,
      "learning_rate": 0.0003717239703969481,
      "loss": 2.2309,
      "step": 97496
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.884982168674469,
      "learning_rate": 0.0003717199984548315,
      "loss": 2.3016,
      "step": 97497
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.8501297831535339,
      "learning_rate": 0.0003717160264993809,
      "loss": 2.3337,
      "step": 97498
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.9185452461242676,
      "learning_rate": 0.0003717120545305969,
      "loss": 2.2605,
      "step": 97499
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.9619141817092896,
      "learning_rate": 0.0003717080825484803,
      "loss": 2.059,
      "step": 97500
    },
    {
      "epoch": 1.27,
      "grad_norm": 1.042678952217102,
      "learning_rate": 0.0003717041105530317,
      "loss": 2.3834,
      "step": 97501
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.9136056900024414,
      "learning_rate": 0.0003717001385442522,
      "loss": 2.0183,
      "step": 97502
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.9434909224510193,
      "learning_rate": 0.00037169616652214213,
      "loss": 2.1245,
      "step": 97503
    },
    {
      "epoch": 1.27,
      "grad_norm": 1.0154168605804443,
      "learning_rate": 0.00037169219448670237,
      "loss": 2.1539,
      "step": 97504
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.8997606039047241,
      "learning_rate": 0.00037168822243793375,
      "loss": 2.1476,
      "step": 97505
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.7570874691009521,
      "learning_rate": 0.00037168425037583687,
      "loss": 1.9752,
      "step": 97506
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.814613938331604,
      "learning_rate": 0.0003716802783004125,
      "loss": 2.1634,
      "step": 97507
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.8078292608261108,
      "learning_rate": 0.00037167630621166153,
      "loss": 2.0741,
      "step": 97508
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.8358138799667358,
      "learning_rate": 0.00037167233410958454,
      "loss": 2.0974,
      "step": 97509
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.9059548377990723,
      "learning_rate": 0.00037166836199418226,
      "loss": 2.0004,
      "step": 97510
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.8215038180351257,
      "learning_rate": 0.0003716643898654555,
      "loss": 2.2162,
      "step": 97511
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.8813949823379517,
      "learning_rate": 0.000371660417723405,
      "loss": 2.0594,
      "step": 97512
    },
    {
      "epoch": 1.27,
      "grad_norm": 1.0134557485580444,
      "learning_rate": 0.0003716564455680315,
      "loss": 1.9371,
      "step": 97513
    },
    {
      "epoch": 1.27,
      "grad_norm": 1.280478835105896,
      "learning_rate": 0.0003716524733993357,
      "loss": 2.1117,
      "step": 97514
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.8672871589660645,
      "learning_rate": 0.00037164850121731833,
      "loss": 2.2327,
      "step": 97515
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.8904746770858765,
      "learning_rate": 0.0003716445290219801,
      "loss": 2.1155,
      "step": 97516
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.8644921779632568,
      "learning_rate": 0.0003716405568133219,
      "loss": 2.131,
      "step": 97517
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.9587911367416382,
      "learning_rate": 0.0003716365845913443,
      "loss": 2.0984,
      "step": 97518
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.807776927947998,
      "learning_rate": 0.00037163261235604813,
      "loss": 2.3262,
      "step": 97519
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.9137119650840759,
      "learning_rate": 0.0003716286401074341,
      "loss": 2.2927,
      "step": 97520
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.9233542680740356,
      "learning_rate": 0.000371624667845503,
      "loss": 2.1223,
      "step": 97521
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.9310930967330933,
      "learning_rate": 0.00037162069557025544,
      "loss": 2.3036,
      "step": 97522
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.828269362449646,
      "learning_rate": 0.0003716167232816923,
      "loss": 2.0333,
      "step": 97523
    },
    {
      "epoch": 1.27,
      "grad_norm": 1.0749568939208984,
      "learning_rate": 0.0003716127509798143,
      "loss": 2.0373,
      "step": 97524
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.838279664516449,
      "learning_rate": 0.00037160877866462204,
      "loss": 2.2474,
      "step": 97525
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.9193928241729736,
      "learning_rate": 0.00037160480633611635,
      "loss": 2.1337,
      "step": 97526
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.8358353972434998,
      "learning_rate": 0.0003716008339942981,
      "loss": 2.3687,
      "step": 97527
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.8716017603874207,
      "learning_rate": 0.0003715968616391678,
      "loss": 2.185,
      "step": 97528
    },
    {
      "epoch": 1.27,
      "grad_norm": 1.0688451528549194,
      "learning_rate": 0.0003715928892707263,
      "loss": 2.2017,
      "step": 97529
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.93137526512146,
      "learning_rate": 0.00037158891688897437,
      "loss": 2.096,
      "step": 97530
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.8869883418083191,
      "learning_rate": 0.00037158494449391266,
      "loss": 2.1573,
      "step": 97531
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.9092071056365967,
      "learning_rate": 0.000371580972085542,
      "loss": 2.1075,
      "step": 97532
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.9903943538665771,
      "learning_rate": 0.00037157699966386306,
      "loss": 2.1905,
      "step": 97533
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.8777880072593689,
      "learning_rate": 0.00037157302722887664,
      "loss": 2.1789,
      "step": 97534
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.7926355004310608,
      "learning_rate": 0.0003715690547805834,
      "loss": 2.2074,
      "step": 97535
    },
    {
      "epoch": 1.27,
      "grad_norm": 1.0309350490570068,
      "learning_rate": 0.0003715650823189841,
      "loss": 2.1867,
      "step": 97536
    },
    {
      "epoch": 1.27,
      "grad_norm": 1.0742292404174805,
      "learning_rate": 0.0003715611098440796,
      "loss": 2.2782,
      "step": 97537
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.8165940046310425,
      "learning_rate": 0.00037155713735587045,
      "loss": 2.0147,
      "step": 97538
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.9739518165588379,
      "learning_rate": 0.00037155316485435754,
      "loss": 1.9672,
      "step": 97539
    },
    {
      "epoch": 1.27,
      "grad_norm": 1.200485348701477,
      "learning_rate": 0.00037154919233954153,
      "loss": 1.9065,
      "step": 97540
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.8643597960472107,
      "learning_rate": 0.0003715452198114231,
      "loss": 2.071,
      "step": 97541
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.9048675894737244,
      "learning_rate": 0.00037154124727000316,
      "loss": 2.234,
      "step": 97542
    },
    {
      "epoch": 1.27,
      "grad_norm": 1.0873178243637085,
      "learning_rate": 0.00037153727471528233,
      "loss": 2.3101,
      "step": 97543
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.974920928478241,
      "learning_rate": 0.0003715333021472613,
      "loss": 2.0458,
      "step": 97544
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.8232133388519287,
      "learning_rate": 0.00037152932956594094,
      "loss": 2.3268,
      "step": 97545
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.9343146085739136,
      "learning_rate": 0.00037152535697132196,
      "loss": 2.0515,
      "step": 97546
    },
    {
      "epoch": 1.27,
      "grad_norm": 1.0324488878250122,
      "learning_rate": 0.0003715213843634049,
      "loss": 2.107,
      "step": 97547
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.9607656002044678,
      "learning_rate": 0.0003715174117421909,
      "loss": 2.0988,
      "step": 97548
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.8073098063468933,
      "learning_rate": 0.0003715134391076803,
      "loss": 2.119,
      "step": 97549
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.9016832709312439,
      "learning_rate": 0.00037150946645987405,
      "loss": 2.2333,
      "step": 97550
    },
    {
      "epoch": 1.27,
      "grad_norm": 1.11008882522583,
      "learning_rate": 0.0003715054937987729,
      "loss": 2.191,
      "step": 97551
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.9015742540359497,
      "learning_rate": 0.0003715015211243775,
      "loss": 1.9334,
      "step": 97552
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.7940180897712708,
      "learning_rate": 0.0003714975484366885,
      "loss": 2.0453,
      "step": 97553
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.9661076664924622,
      "learning_rate": 0.00037149357573570686,
      "loss": 2.2559,
      "step": 97554
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.831932008266449,
      "learning_rate": 0.0003714896030214332,
      "loss": 2.3269,
      "step": 97555
    },
    {
      "epoch": 1.27,
      "grad_norm": 1.0509198904037476,
      "learning_rate": 0.00037148563029386827,
      "loss": 2.0073,
      "step": 97556
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.9315856099128723,
      "learning_rate": 0.00037148165755301286,
      "loss": 2.0668,
      "step": 97557
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.8657909631729126,
      "learning_rate": 0.0003714776847988676,
      "loss": 2.1674,
      "step": 97558
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.9860310554504395,
      "learning_rate": 0.00037147371203143326,
      "loss": 2.4203,
      "step": 97559
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.8346751928329468,
      "learning_rate": 0.00037146973925071075,
      "loss": 2.2873,
      "step": 97560
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.8375080823898315,
      "learning_rate": 0.0003714657664567005,
      "loss": 2.1898,
      "step": 97561
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.882067084312439,
      "learning_rate": 0.0003714617936494035,
      "loss": 1.7469,
      "step": 97562
    },
    {
      "epoch": 1.27,
      "grad_norm": 1.052473783493042,
      "learning_rate": 0.00037145782082882045,
      "loss": 2.061,
      "step": 97563
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.823142945766449,
      "learning_rate": 0.0003714538479949519,
      "loss": 2.2677,
      "step": 97564
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.8735289573669434,
      "learning_rate": 0.00037144987514779883,
      "loss": 2.3242,
      "step": 97565
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.956272304058075,
      "learning_rate": 0.00037144590228736183,
      "loss": 2.1051,
      "step": 97566
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.8259121179580688,
      "learning_rate": 0.0003714419294136417,
      "loss": 2.2443,
      "step": 97567
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.8535693883895874,
      "learning_rate": 0.0003714379565266392,
      "loss": 2.0814,
      "step": 97568
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.8341212868690491,
      "learning_rate": 0.000371433983626355,
      "loss": 2.2733,
      "step": 97569
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.8315748572349548,
      "learning_rate": 0.00037143001071278993,
      "loss": 2.2544,
      "step": 97570
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.8676363229751587,
      "learning_rate": 0.0003714260377859446,
      "loss": 2.183,
      "step": 97571
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.7816982269287109,
      "learning_rate": 0.0003714220648458199,
      "loss": 2.1469,
      "step": 97572
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.9058188796043396,
      "learning_rate": 0.0003714180918924164,
      "loss": 2.1068,
      "step": 97573
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.9457792043685913,
      "learning_rate": 0.00037141411892573507,
      "loss": 2.259,
      "step": 97574
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.914009153842926,
      "learning_rate": 0.0003714101459457764,
      "loss": 2.444,
      "step": 97575
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.756264328956604,
      "learning_rate": 0.00037140617295254125,
      "loss": 2.0926,
      "step": 97576
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.925897479057312,
      "learning_rate": 0.00037140219994603033,
      "loss": 2.1931,
      "step": 97577
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.9658790230751038,
      "learning_rate": 0.0003713982269262444,
      "loss": 2.1628,
      "step": 97578
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.8028470277786255,
      "learning_rate": 0.0003713942538931842,
      "loss": 1.9425,
      "step": 97579
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.8631658554077148,
      "learning_rate": 0.0003713902808468504,
      "loss": 2.3611,
      "step": 97580
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.8824260830879211,
      "learning_rate": 0.0003713863077872439,
      "loss": 2.2506,
      "step": 97581
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.7400556206703186,
      "learning_rate": 0.0003713823347143653,
      "loss": 2.0366,
      "step": 97582
    },
    {
      "epoch": 1.27,
      "grad_norm": 1.0115289688110352,
      "learning_rate": 0.00037137836162821534,
      "loss": 1.9448,
      "step": 97583
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.9142807126045227,
      "learning_rate": 0.00037137438852879486,
      "loss": 2.3704,
      "step": 97584
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.8231096267700195,
      "learning_rate": 0.0003713704154161045,
      "loss": 2.2149,
      "step": 97585
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.8046329021453857,
      "learning_rate": 0.00037136644229014503,
      "loss": 2.1592,
      "step": 97586
    },
    {
      "epoch": 1.27,
      "grad_norm": 1.1176291704177856,
      "learning_rate": 0.00037136246915091714,
      "loss": 1.9916,
      "step": 97587
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.833228588104248,
      "learning_rate": 0.00037135849599842177,
      "loss": 2.134,
      "step": 97588
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.7725687026977539,
      "learning_rate": 0.00037135452283265934,
      "loss": 1.9478,
      "step": 97589
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.9572176933288574,
      "learning_rate": 0.0003713505496536308,
      "loss": 2.329,
      "step": 97590
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.9681053757667542,
      "learning_rate": 0.00037134657646133694,
      "loss": 2.2979,
      "step": 97591
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.8309944272041321,
      "learning_rate": 0.0003713426032557784,
      "loss": 2.1829,
      "step": 97592
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.7991262078285217,
      "learning_rate": 0.00037133863003695574,
      "loss": 2.3107,
      "step": 97593
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.8871723413467407,
      "learning_rate": 0.0003713346568048701,
      "loss": 2.0837,
      "step": 97594
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.9302531480789185,
      "learning_rate": 0.0003713306835595219,
      "loss": 2.1255,
      "step": 97595
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.840040922164917,
      "learning_rate": 0.00037132671030091195,
      "loss": 2.0098,
      "step": 97596
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.9025099277496338,
      "learning_rate": 0.00037132273702904114,
      "loss": 2.3402,
      "step": 97597
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.9708803296089172,
      "learning_rate": 0.00037131876374390995,
      "loss": 2.297,
      "step": 97598
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.9776355624198914,
      "learning_rate": 0.0003713147904455193,
      "loss": 2.2735,
      "step": 97599
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.8260123133659363,
      "learning_rate": 0.0003713108171338699,
      "loss": 2.1513,
      "step": 97600
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.7572734355926514,
      "learning_rate": 0.0003713068438089625,
      "loss": 2.0043,
      "step": 97601
    },
    {
      "epoch": 1.27,
      "grad_norm": 1.0656524896621704,
      "learning_rate": 0.00037130287047079777,
      "loss": 2.3685,
      "step": 97602
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.8731191158294678,
      "learning_rate": 0.00037129889711937656,
      "loss": 2.1097,
      "step": 97603
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.7509345412254333,
      "learning_rate": 0.0003712949237546994,
      "loss": 1.9804,
      "step": 97604
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.8778806924819946,
      "learning_rate": 0.0003712909503767673,
      "loss": 2.0773,
      "step": 97605
    },
    {
      "epoch": 1.27,
      "grad_norm": 1.0916163921356201,
      "learning_rate": 0.0003712869769855808,
      "loss": 2.0149,
      "step": 97606
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.9397467970848083,
      "learning_rate": 0.00037128300358114073,
      "loss": 2.2687,
      "step": 97607
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.9403010010719299,
      "learning_rate": 0.0003712790301634477,
      "loss": 2.334,
      "step": 97608
    },
    {
      "epoch": 1.27,
      "grad_norm": 1.0225965976715088,
      "learning_rate": 0.0003712750567325027,
      "loss": 2.2466,
      "step": 97609
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.9988383650779724,
      "learning_rate": 0.0003712710832883063,
      "loss": 2.1269,
      "step": 97610
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.9021579027175903,
      "learning_rate": 0.0003712671098308592,
      "loss": 2.1545,
      "step": 97611
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.8461304903030396,
      "learning_rate": 0.00037126313636016224,
      "loss": 2.2794,
      "step": 97612
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.8675190210342407,
      "learning_rate": 0.00037125916287621607,
      "loss": 2.1839,
      "step": 97613
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.9088851809501648,
      "learning_rate": 0.00037125518937902155,
      "loss": 2.0578,
      "step": 97614
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.8295682072639465,
      "learning_rate": 0.0003712512158685793,
      "loss": 1.9748,
      "step": 97615
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.8342861533164978,
      "learning_rate": 0.0003712472423448901,
      "loss": 2.1619,
      "step": 97616
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.9392724633216858,
      "learning_rate": 0.0003712432688079547,
      "loss": 2.152,
      "step": 97617
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.8674825429916382,
      "learning_rate": 0.0003712392952577739,
      "loss": 2.1375,
      "step": 97618
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.7468091249465942,
      "learning_rate": 0.0003712353216943482,
      "loss": 2.0384,
      "step": 97619
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.9723129868507385,
      "learning_rate": 0.0003712313481176787,
      "loss": 2.0225,
      "step": 97620
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.8908321857452393,
      "learning_rate": 0.0003712273745277658,
      "loss": 2.2744,
      "step": 97621
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.8701359629631042,
      "learning_rate": 0.00037122340092461047,
      "loss": 2.1443,
      "step": 97622
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.8796759843826294,
      "learning_rate": 0.00037121942730821333,
      "loss": 2.1466,
      "step": 97623
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.8962298631668091,
      "learning_rate": 0.00037121545367857517,
      "loss": 2.1715,
      "step": 97624
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.8333399891853333,
      "learning_rate": 0.0003712114800356967,
      "loss": 2.1626,
      "step": 97625
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.8550923466682434,
      "learning_rate": 0.0003712075063795787,
      "loss": 2.2208,
      "step": 97626
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.8370183110237122,
      "learning_rate": 0.00037120353271022187,
      "loss": 2.3107,
      "step": 97627
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.7817015647888184,
      "learning_rate": 0.000371199559027627,
      "loss": 2.0295,
      "step": 97628
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.9871018528938293,
      "learning_rate": 0.00037119558533179464,
      "loss": 2.1066,
      "step": 97629
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.8156163096427917,
      "learning_rate": 0.00037119161162272587,
      "loss": 2.0804,
      "step": 97630
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.8838117122650146,
      "learning_rate": 0.0003711876379004211,
      "loss": 2.0533,
      "step": 97631
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.7819247245788574,
      "learning_rate": 0.0003711836641648812,
      "loss": 2.1,
      "step": 97632
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.8222612142562866,
      "learning_rate": 0.00037117969041610706,
      "loss": 2.0429,
      "step": 97633
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.8971751928329468,
      "learning_rate": 0.00037117571665409914,
      "loss": 1.939,
      "step": 97634
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.886384129524231,
      "learning_rate": 0.0003711717428788583,
      "loss": 2.1122,
      "step": 97635
    },
    {
      "epoch": 1.27,
      "grad_norm": 1.0773576498031616,
      "learning_rate": 0.0003711677690903854,
      "loss": 1.9249,
      "step": 97636
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.9288716912269592,
      "learning_rate": 0.000371163795288681,
      "loss": 2.1294,
      "step": 97637
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.858308732509613,
      "learning_rate": 0.0003711598214737458,
      "loss": 2.0744,
      "step": 97638
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.9401355385780334,
      "learning_rate": 0.0003711558476455808,
      "loss": 2.0317,
      "step": 97639
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.8029796481132507,
      "learning_rate": 0.0003711518738041865,
      "loss": 2.2104,
      "step": 97640
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.8618212342262268,
      "learning_rate": 0.00037114789994956377,
      "loss": 2.184,
      "step": 97641
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.8161454796791077,
      "learning_rate": 0.00037114392608171337,
      "loss": 2.2077,
      "step": 97642
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.9501190185546875,
      "learning_rate": 0.00037113995220063587,
      "loss": 2.2669,
      "step": 97643
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.8835022449493408,
      "learning_rate": 0.0003711359783063321,
      "loss": 2.2456,
      "step": 97644
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.8385156393051147,
      "learning_rate": 0.0003711320043988029,
      "loss": 2.0381,
      "step": 97645
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.9055206775665283,
      "learning_rate": 0.0003711280304780489,
      "loss": 2.015,
      "step": 97646
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.9536764025688171,
      "learning_rate": 0.0003711240565440707,
      "loss": 2.1314,
      "step": 97647
    },
    {
      "epoch": 1.27,
      "grad_norm": 1.0026377439498901,
      "learning_rate": 0.0003711200825968694,
      "loss": 2.4058,
      "step": 97648
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.8878577351570129,
      "learning_rate": 0.0003711161086364454,
      "loss": 2.1457,
      "step": 97649
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.8557074666023254,
      "learning_rate": 0.0003711121346627997,
      "loss": 2.0983,
      "step": 97650
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.9333294034004211,
      "learning_rate": 0.00037110816067593283,
      "loss": 2.2425,
      "step": 97651
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.8734149932861328,
      "learning_rate": 0.00037110418667584555,
      "loss": 2.3247,
      "step": 97652
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.8844798803329468,
      "learning_rate": 0.0003711002126625387,
      "loss": 2.2763,
      "step": 97653
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.9097490906715393,
      "learning_rate": 0.000371096238636013,
      "loss": 2.3653,
      "step": 97654
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.806279182434082,
      "learning_rate": 0.00037109226459626916,
      "loss": 2.3164,
      "step": 97655
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.9147189855575562,
      "learning_rate": 0.00037108829054330794,
      "loss": 2.0561,
      "step": 97656
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.8460276126861572,
      "learning_rate": 0.0003710843164771301,
      "loss": 2.2474,
      "step": 97657
    },
    {
      "epoch": 1.27,
      "grad_norm": 1.0936287641525269,
      "learning_rate": 0.0003710803423977363,
      "loss": 2.1986,
      "step": 97658
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.8555496335029602,
      "learning_rate": 0.0003710763683051273,
      "loss": 2.4008,
      "step": 97659
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.937592625617981,
      "learning_rate": 0.00037107239419930386,
      "loss": 2.1942,
      "step": 97660
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.9006474614143372,
      "learning_rate": 0.00037106842008026675,
      "loss": 2.3088,
      "step": 97661
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.8642085790634155,
      "learning_rate": 0.0003710644459480167,
      "loss": 2.1918,
      "step": 97662
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.9029778242111206,
      "learning_rate": 0.00037106047180255437,
      "loss": 2.3252,
      "step": 97663
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.8588172197341919,
      "learning_rate": 0.0003710564976438806,
      "loss": 2.2184,
      "step": 97664
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.8731042146682739,
      "learning_rate": 0.000371052523471996,
      "loss": 1.9021,
      "step": 97665
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.841629683971405,
      "learning_rate": 0.00037104854928690157,
      "loss": 2.2467,
      "step": 97666
    },
    {
      "epoch": 1.27,
      "grad_norm": 1.0945981740951538,
      "learning_rate": 0.0003710445750885977,
      "loss": 2.3837,
      "step": 97667
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.8818631172180176,
      "learning_rate": 0.00037104060087708545,
      "loss": 2.2036,
      "step": 97668
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.8456904888153076,
      "learning_rate": 0.0003710366266523653,
      "loss": 2.0891,
      "step": 97669
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.9016616940498352,
      "learning_rate": 0.00037103265241443816,
      "loss": 1.9445,
      "step": 97670
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.7866500616073608,
      "learning_rate": 0.0003710286781633046,
      "loss": 1.9902,
      "step": 97671
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.8335856795310974,
      "learning_rate": 0.0003710247038989656,
      "loss": 2.1126,
      "step": 97672
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.8593577742576599,
      "learning_rate": 0.0003710207296214218,
      "loss": 2.2265,
      "step": 97673
    },
    {
      "epoch": 1.27,
      "grad_norm": 1.0099066495895386,
      "learning_rate": 0.00037101675533067377,
      "loss": 2.2253,
      "step": 97674
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.9473097324371338,
      "learning_rate": 0.00037101278102672245,
      "loss": 2.0805,
      "step": 97675
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.823335587978363,
      "learning_rate": 0.00037100880670956854,
      "loss": 2.3188,
      "step": 97676
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.8373199105262756,
      "learning_rate": 0.0003710048323792127,
      "loss": 2.2265,
      "step": 97677
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.8977344036102295,
      "learning_rate": 0.0003710008580356557,
      "loss": 2.2159,
      "step": 97678
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.8874219059944153,
      "learning_rate": 0.00037099688367889835,
      "loss": 2.032,
      "step": 97679
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.9059575200080872,
      "learning_rate": 0.0003709929093089413,
      "loss": 2.1003,
      "step": 97680
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.9310628175735474,
      "learning_rate": 0.0003709889349257854,
      "loss": 2.2143,
      "step": 97681
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.9506047368049622,
      "learning_rate": 0.0003709849605294313,
      "loss": 2.2051,
      "step": 97682
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.9984771013259888,
      "learning_rate": 0.0003709809861198797,
      "loss": 2.3891,
      "step": 97683
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.8697807788848877,
      "learning_rate": 0.00037097701169713136,
      "loss": 2.343,
      "step": 97684
    },
    {
      "epoch": 1.27,
      "grad_norm": 1.0032449960708618,
      "learning_rate": 0.0003709730372611872,
      "loss": 2.2193,
      "step": 97685
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.9201045632362366,
      "learning_rate": 0.0003709690628120477,
      "loss": 1.9095,
      "step": 97686
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.816259503364563,
      "learning_rate": 0.00037096508834971373,
      "loss": 2.1496,
      "step": 97687
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.762634813785553,
      "learning_rate": 0.0003709611138741861,
      "loss": 2.1271,
      "step": 97688
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.9828559160232544,
      "learning_rate": 0.00037095713938546535,
      "loss": 2.1768,
      "step": 97689
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.9630188941955566,
      "learning_rate": 0.0003709531648835523,
      "loss": 2.433,
      "step": 97690
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.8983672261238098,
      "learning_rate": 0.00037094919036844787,
      "loss": 2.2385,
      "step": 97691
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.9109818339347839,
      "learning_rate": 0.0003709452158401525,
      "loss": 2.187,
      "step": 97692
    },
    {
      "epoch": 1.27,
      "grad_norm": 1.0482611656188965,
      "learning_rate": 0.00037094124129866715,
      "loss": 2.3426,
      "step": 97693
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.9057289361953735,
      "learning_rate": 0.0003709372667439925,
      "loss": 2.4713,
      "step": 97694
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.9691604375839233,
      "learning_rate": 0.0003709332921761292,
      "loss": 2.3807,
      "step": 97695
    },
    {
      "epoch": 1.27,
      "grad_norm": 1.5580308437347412,
      "learning_rate": 0.00037092931759507804,
      "loss": 2.1578,
      "step": 97696
    },
    {
      "epoch": 1.27,
      "grad_norm": 1.0805014371871948,
      "learning_rate": 0.0003709253430008399,
      "loss": 2.0448,
      "step": 97697
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.8786227107048035,
      "learning_rate": 0.00037092136839341535,
      "loss": 2.2705,
      "step": 97698
    },
    {
      "epoch": 1.27,
      "grad_norm": 1.2674871683120728,
      "learning_rate": 0.0003709173937728052,
      "loss": 2.1574,
      "step": 97699
    },
    {
      "epoch": 1.27,
      "grad_norm": 1.0661135911941528,
      "learning_rate": 0.0003709134191390101,
      "loss": 2.122,
      "step": 97700
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.9316908717155457,
      "learning_rate": 0.0003709094444920309,
      "loss": 2.1001,
      "step": 97701
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.8926284909248352,
      "learning_rate": 0.0003709054698318683,
      "loss": 2.1877,
      "step": 97702
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.8678241968154907,
      "learning_rate": 0.000370901495158523,
      "loss": 2.004,
      "step": 97703
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.9356529116630554,
      "learning_rate": 0.0003708975204719958,
      "loss": 1.9731,
      "step": 97704
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.9370190501213074,
      "learning_rate": 0.0003708935457722874,
      "loss": 2.2367,
      "step": 97705
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.9179943203926086,
      "learning_rate": 0.00037088957105939856,
      "loss": 2.2523,
      "step": 97706
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.966221809387207,
      "learning_rate": 0.00037088559633333004,
      "loss": 2.0713,
      "step": 97707
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.8502190709114075,
      "learning_rate": 0.0003708816215940825,
      "loss": 2.2141,
      "step": 97708
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.9307364821434021,
      "learning_rate": 0.00037087764684165677,
      "loss": 2.1325,
      "step": 97709
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.787540853023529,
      "learning_rate": 0.0003708736720760535,
      "loss": 2.0877,
      "step": 97710
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.8519092202186584,
      "learning_rate": 0.00037086969729727354,
      "loss": 2.0498,
      "step": 97711
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.8711866736412048,
      "learning_rate": 0.00037086572250531757,
      "loss": 2.4183,
      "step": 97712
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.8610890507698059,
      "learning_rate": 0.00037086174770018627,
      "loss": 2.2038,
      "step": 97713
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.8244082927703857,
      "learning_rate": 0.00037085777288188045,
      "loss": 2.0795,
      "step": 97714
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.8290753364562988,
      "learning_rate": 0.0003708537980504008,
      "loss": 2.033,
      "step": 97715
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.9582028985023499,
      "learning_rate": 0.0003708498232057481,
      "loss": 2.2793,
      "step": 97716
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.8075973987579346,
      "learning_rate": 0.0003708458483479231,
      "loss": 2.0687,
      "step": 97717
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.8582260608673096,
      "learning_rate": 0.00037084187347692654,
      "loss": 2.067,
      "step": 97718
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.9066739678382874,
      "learning_rate": 0.0003708378985927591,
      "loss": 2.2338,
      "step": 97719
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.8481118679046631,
      "learning_rate": 0.0003708339236954216,
      "loss": 2.1693,
      "step": 97720
    },
    {
      "epoch": 1.27,
      "grad_norm": 1.0026288032531738,
      "learning_rate": 0.0003708299487849147,
      "loss": 2.2507,
      "step": 97721
    },
    {
      "epoch": 1.27,
      "grad_norm": 1.0209920406341553,
      "learning_rate": 0.00037082597386123924,
      "loss": 2.123,
      "step": 97722
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.9089497327804565,
      "learning_rate": 0.00037082199892439586,
      "loss": 2.2202,
      "step": 97723
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.874033510684967,
      "learning_rate": 0.0003708180239743852,
      "loss": 2.4492,
      "step": 97724
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.8849186301231384,
      "learning_rate": 0.00037081404901120836,
      "loss": 2.271,
      "step": 97725
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.8671238422393799,
      "learning_rate": 0.00037081007403486576,
      "loss": 2.2575,
      "step": 97726
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.8714619874954224,
      "learning_rate": 0.0003708060990453581,
      "loss": 1.9631,
      "step": 97727
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.9527518153190613,
      "learning_rate": 0.00037080212404268643,
      "loss": 2.3749,
      "step": 97728
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.8518770933151245,
      "learning_rate": 0.0003707981490268512,
      "loss": 2.0451,
      "step": 97729
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.883281409740448,
      "learning_rate": 0.0003707941739978533,
      "loss": 2.0305,
      "step": 97730
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.8796981573104858,
      "learning_rate": 0.00037079019895569347,
      "loss": 2.1477,
      "step": 97731
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.9289690852165222,
      "learning_rate": 0.0003707862239003724,
      "loss": 2.1718,
      "step": 97732
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.8416315913200378,
      "learning_rate": 0.00037078224883189075,
      "loss": 2.0331,
      "step": 97733
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.9088961482048035,
      "learning_rate": 0.0003707782737502494,
      "loss": 2.3427,
      "step": 97734
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.8464286923408508,
      "learning_rate": 0.00037077429865544904,
      "loss": 2.2925,
      "step": 97735
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.8589717149734497,
      "learning_rate": 0.00037077032354749037,
      "loss": 1.8664,
      "step": 97736
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.9183205366134644,
      "learning_rate": 0.00037076634842637425,
      "loss": 2.24,
      "step": 97737
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.9131268858909607,
      "learning_rate": 0.0003707623732921011,
      "loss": 2.0445,
      "step": 97738
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.9552828073501587,
      "learning_rate": 0.00037075839814467213,
      "loss": 2.3748,
      "step": 97739
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.862698495388031,
      "learning_rate": 0.00037075442298408775,
      "loss": 2.3047,
      "step": 97740
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.852630078792572,
      "learning_rate": 0.0003707504478103488,
      "loss": 2.1298,
      "step": 97741
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.9325474500656128,
      "learning_rate": 0.000370746472623456,
      "loss": 2.2456,
      "step": 97742
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.9351797699928284,
      "learning_rate": 0.0003707424974234101,
      "loss": 2.0471,
      "step": 97743
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.8553064465522766,
      "learning_rate": 0.00037073852221021183,
      "loss": 2.2746,
      "step": 97744
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.8661884069442749,
      "learning_rate": 0.00037073454698386186,
      "loss": 2.1339,
      "step": 97745
    },
    {
      "epoch": 1.27,
      "grad_norm": 1.1275347471237183,
      "learning_rate": 0.0003707305717443611,
      "loss": 2.3351,
      "step": 97746
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.8189599514007568,
      "learning_rate": 0.0003707265964917102,
      "loss": 2.1007,
      "step": 97747
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.8053777813911438,
      "learning_rate": 0.00037072262122590983,
      "loss": 1.9789,
      "step": 97748
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.9030117392539978,
      "learning_rate": 0.00037071864594696077,
      "loss": 2.2531,
      "step": 97749
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.8131181597709656,
      "learning_rate": 0.0003707146706548639,
      "loss": 2.3215,
      "step": 97750
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.8464570045471191,
      "learning_rate": 0.00037071069534961974,
      "loss": 2.169,
      "step": 97751
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.8482025265693665,
      "learning_rate": 0.00037070672003122914,
      "loss": 2.0599,
      "step": 97752
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.8433298468589783,
      "learning_rate": 0.00037070274469969274,
      "loss": 2.1202,
      "step": 97753
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.8611611127853394,
      "learning_rate": 0.00037069876935501154,
      "loss": 2.1619,
      "step": 97754
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.8929179310798645,
      "learning_rate": 0.000370694793997186,
      "loss": 2.2039,
      "step": 97755
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.91375732421875,
      "learning_rate": 0.000370690818626217,
      "loss": 1.9858,
      "step": 97756
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.8574888110160828,
      "learning_rate": 0.00037068684324210516,
      "loss": 1.9013,
      "step": 97757
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.9413275718688965,
      "learning_rate": 0.00037068286784485143,
      "loss": 2.2668,
      "step": 97758
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.9470993280410767,
      "learning_rate": 0.00037067889243445635,
      "loss": 2.2777,
      "step": 97759
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.8047124743461609,
      "learning_rate": 0.00037067491701092074,
      "loss": 2.0746,
      "step": 97760
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.7368241548538208,
      "learning_rate": 0.00037067094157424536,
      "loss": 2.1802,
      "step": 97761
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.9299473762512207,
      "learning_rate": 0.00037066696612443086,
      "loss": 2.3943,
      "step": 97762
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.8793920874595642,
      "learning_rate": 0.00037066299066147804,
      "loss": 2.3115,
      "step": 97763
    },
    {
      "epoch": 1.27,
      "grad_norm": 1.130165696144104,
      "learning_rate": 0.00037065901518538774,
      "loss": 2.1229,
      "step": 97764
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.9672716856002808,
      "learning_rate": 0.00037065503969616043,
      "loss": 2.2107,
      "step": 97765
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.8933173418045044,
      "learning_rate": 0.00037065106419379714,
      "loss": 2.083,
      "step": 97766
    },
    {
      "epoch": 1.27,
      "grad_norm": 1.2379300594329834,
      "learning_rate": 0.0003706470886782985,
      "loss": 2.0907,
      "step": 97767
    },
    {
      "epoch": 1.27,
      "grad_norm": 1.1306047439575195,
      "learning_rate": 0.00037064311314966513,
      "loss": 2.175,
      "step": 97768
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.8786335587501526,
      "learning_rate": 0.00037063913760789787,
      "loss": 2.0154,
      "step": 97769
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.8675678372383118,
      "learning_rate": 0.00037063516205299756,
      "loss": 2.217,
      "step": 97770
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.9585796594619751,
      "learning_rate": 0.0003706311864849648,
      "loss": 2.1774,
      "step": 97771
    },
    {
      "epoch": 1.27,
      "grad_norm": 1.0128401517868042,
      "learning_rate": 0.00037062721090380034,
      "loss": 1.9003,
      "step": 97772
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.8841124773025513,
      "learning_rate": 0.00037062323530950495,
      "loss": 2.0603,
      "step": 97773
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.8504026532173157,
      "learning_rate": 0.0003706192597020794,
      "loss": 2.2689,
      "step": 97774
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.961746871471405,
      "learning_rate": 0.00037061528408152434,
      "loss": 2.2237,
      "step": 97775
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.9348375201225281,
      "learning_rate": 0.00037061130844784064,
      "loss": 2.1334,
      "step": 97776
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.7997708320617676,
      "learning_rate": 0.000370607332801029,
      "loss": 2.1755,
      "step": 97777
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.8641436100006104,
      "learning_rate": 0.00037060335714109,
      "loss": 2.0482,
      "step": 97778
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.9670592546463013,
      "learning_rate": 0.00037059938146802453,
      "loss": 2.0777,
      "step": 97779
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.948907196521759,
      "learning_rate": 0.00037059540578183336,
      "loss": 2.0061,
      "step": 97780
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.824674665927887,
      "learning_rate": 0.0003705914300825171,
      "loss": 2.3525,
      "step": 97781
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.9747745990753174,
      "learning_rate": 0.0003705874543700767,
      "loss": 2.262,
      "step": 97782
    },
    {
      "epoch": 1.27,
      "grad_norm": 1.1190893650054932,
      "learning_rate": 0.00037058347864451266,
      "loss": 2.1891,
      "step": 97783
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.9086553454399109,
      "learning_rate": 0.00037057950290582577,
      "loss": 2.2713,
      "step": 97784
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.9028155207633972,
      "learning_rate": 0.0003705755271540169,
      "loss": 2.2021,
      "step": 97785
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.9425101280212402,
      "learning_rate": 0.00037057155138908667,
      "loss": 2.0901,
      "step": 97786
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.9515548944473267,
      "learning_rate": 0.0003705675756110358,
      "loss": 2.1829,
      "step": 97787
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.7856143712997437,
      "learning_rate": 0.00037056359981986513,
      "loss": 1.9297,
      "step": 97788
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.8836188316345215,
      "learning_rate": 0.0003705596240155754,
      "loss": 2.0343,
      "step": 97789
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.8738917708396912,
      "learning_rate": 0.0003705556481981673,
      "loss": 2.2113,
      "step": 97790
    },
    {
      "epoch": 1.27,
      "grad_norm": 1.022935390472412,
      "learning_rate": 0.0003705516723676415,
      "loss": 2.1795,
      "step": 97791
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.8066747188568115,
      "learning_rate": 0.0003705476965239989,
      "loss": 1.9683,
      "step": 97792
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.902192234992981,
      "learning_rate": 0.0003705437206672401,
      "loss": 2.2353,
      "step": 97793
    },
    {
      "epoch": 1.27,
      "grad_norm": 1.1196990013122559,
      "learning_rate": 0.0003705397447973659,
      "loss": 2.4253,
      "step": 97794
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.9348576068878174,
      "learning_rate": 0.00037053576891437706,
      "loss": 1.9925,
      "step": 97795
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.8624193668365479,
      "learning_rate": 0.00037053179301827426,
      "loss": 2.3056,
      "step": 97796
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.8561015129089355,
      "learning_rate": 0.00037052781710905826,
      "loss": 2.2639,
      "step": 97797
    },
    {
      "epoch": 1.27,
      "grad_norm": 1.0171250104904175,
      "learning_rate": 0.0003705238411867299,
      "loss": 2.3686,
      "step": 97798
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.831666111946106,
      "learning_rate": 0.00037051986525128965,
      "loss": 2.1189,
      "step": 97799
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.8779531121253967,
      "learning_rate": 0.0003705158893027385,
      "loss": 2.1693,
      "step": 97800
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.8825636506080627,
      "learning_rate": 0.0003705119133410772,
      "loss": 2.2223,
      "step": 97801
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.9798402190208435,
      "learning_rate": 0.00037050793736630633,
      "loss": 2.1879,
      "step": 97802
    },
    {
      "epoch": 1.27,
      "grad_norm": 1.0336040258407593,
      "learning_rate": 0.0003705039613784267,
      "loss": 2.2808,
      "step": 97803
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.9404039978981018,
      "learning_rate": 0.0003704999853774391,
      "loss": 2.1165,
      "step": 97804
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.8897691369056702,
      "learning_rate": 0.0003704960093633442,
      "loss": 2.0769,
      "step": 97805
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.9100127816200256,
      "learning_rate": 0.0003704920333361427,
      "loss": 2.0804,
      "step": 97806
    },
    {
      "epoch": 1.27,
      "grad_norm": 1.0613352060317993,
      "learning_rate": 0.00037048805729583554,
      "loss": 2.1131,
      "step": 97807
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.8722458481788635,
      "learning_rate": 0.0003704840812424232,
      "loss": 2.1029,
      "step": 97808
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.8373533487319946,
      "learning_rate": 0.0003704801051759065,
      "loss": 1.7638,
      "step": 97809
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.8024895191192627,
      "learning_rate": 0.00037047612909628635,
      "loss": 1.9542,
      "step": 97810
    },
    {
      "epoch": 1.27,
      "grad_norm": 1.0267131328582764,
      "learning_rate": 0.0003704721530035633,
      "loss": 2.4125,
      "step": 97811
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.8467469215393066,
      "learning_rate": 0.0003704681768977381,
      "loss": 2.2911,
      "step": 97812
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.9090699553489685,
      "learning_rate": 0.0003704642007788117,
      "loss": 2.1,
      "step": 97813
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.8958908915519714,
      "learning_rate": 0.00037046022464678444,
      "loss": 2.0415,
      "step": 97814
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.8448190689086914,
      "learning_rate": 0.0003704562485016574,
      "loss": 1.9568,
      "step": 97815
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.8174898624420166,
      "learning_rate": 0.0003704522723434313,
      "loss": 2.2583,
      "step": 97816
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.8998745083808899,
      "learning_rate": 0.0003704482961721067,
      "loss": 2.3375,
      "step": 97817
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.7949833869934082,
      "learning_rate": 0.0003704443199876844,
      "loss": 2.2211,
      "step": 97818
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.8516706228256226,
      "learning_rate": 0.00037044034379016526,
      "loss": 2.1565,
      "step": 97819
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.7742664813995361,
      "learning_rate": 0.00037043636757954984,
      "loss": 2.0766,
      "step": 97820
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.8839136958122253,
      "learning_rate": 0.00037043239135583903,
      "loss": 1.992,
      "step": 97821
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.8969150185585022,
      "learning_rate": 0.00037042841511903356,
      "loss": 2.11,
      "step": 97822
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.8489599227905273,
      "learning_rate": 0.000370424438869134,
      "loss": 2.2149,
      "step": 97823
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.895084023475647,
      "learning_rate": 0.00037042046260614127,
      "loss": 2.0792,
      "step": 97824
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.8256464600563049,
      "learning_rate": 0.00037041648633005607,
      "loss": 2.2491,
      "step": 97825
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.8837612271308899,
      "learning_rate": 0.00037041251004087903,
      "loss": 2.2603,
      "step": 97826
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.7914807200431824,
      "learning_rate": 0.00037040853373861106,
      "loss": 2.0638,
      "step": 97827
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.8854008913040161,
      "learning_rate": 0.0003704045574232528,
      "loss": 2.0443,
      "step": 97828
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.8701367378234863,
      "learning_rate": 0.00037040058109480493,
      "loss": 2.0287,
      "step": 97829
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.8828514814376831,
      "learning_rate": 0.0003703966047532683,
      "loss": 2.4,
      "step": 97830
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.8603107333183289,
      "learning_rate": 0.00037039262839864365,
      "loss": 2.0676,
      "step": 97831
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.8514584302902222,
      "learning_rate": 0.0003703886520309317,
      "loss": 2.3101,
      "step": 97832
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.7705677151679993,
      "learning_rate": 0.00037038467565013314,
      "loss": 1.8897,
      "step": 97833
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.8481903672218323,
      "learning_rate": 0.00037038069925624873,
      "loss": 2.2133,
      "step": 97834
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.8322850465774536,
      "learning_rate": 0.0003703767228492792,
      "loss": 2.2728,
      "step": 97835
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.8270867466926575,
      "learning_rate": 0.00037037274642922534,
      "loss": 2.1504,
      "step": 97836
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.8377052545547485,
      "learning_rate": 0.0003703687699960878,
      "loss": 2.2479,
      "step": 97837
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.8328843712806702,
      "learning_rate": 0.00037036479354986745,
      "loss": 2.2653,
      "step": 97838
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.7946515679359436,
      "learning_rate": 0.00037036081709056494,
      "loss": 2.2108,
      "step": 97839
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.8970595598220825,
      "learning_rate": 0.00037035684061818104,
      "loss": 2.1581,
      "step": 97840
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.8315832018852234,
      "learning_rate": 0.0003703528641327164,
      "loss": 2.072,
      "step": 97841
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.9137168526649475,
      "learning_rate": 0.00037034888763417195,
      "loss": 2.0791,
      "step": 97842
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.8560943007469177,
      "learning_rate": 0.0003703449111225483,
      "loss": 2.1995,
      "step": 97843
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.8087642788887024,
      "learning_rate": 0.00037034093459784604,
      "loss": 2.0991,
      "step": 97844
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.8768714070320129,
      "learning_rate": 0.0003703369580600662,
      "loss": 2.2523,
      "step": 97845
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.9377992153167725,
      "learning_rate": 0.0003703329815092095,
      "loss": 2.1811,
      "step": 97846
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.8749052286148071,
      "learning_rate": 0.0003703290049452764,
      "loss": 2.2868,
      "step": 97847
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.8753104209899902,
      "learning_rate": 0.0003703250283682679,
      "loss": 2.3157,
      "step": 97848
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.851708173751831,
      "learning_rate": 0.00037032105177818456,
      "loss": 2.2563,
      "step": 97849
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.9880666136741638,
      "learning_rate": 0.0003703170751750273,
      "loss": 2.1711,
      "step": 97850
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.8082640171051025,
      "learning_rate": 0.0003703130985587967,
      "loss": 2.0837,
      "step": 97851
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.8283251523971558,
      "learning_rate": 0.00037030912192949356,
      "loss": 2.0869,
      "step": 97852
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.8705013990402222,
      "learning_rate": 0.00037030514528711873,
      "loss": 2.2719,
      "step": 97853
    },
    {
      "epoch": 1.27,
      "grad_norm": 1.2192045450210571,
      "learning_rate": 0.0003703011686316727,
      "loss": 2.2217,
      "step": 97854
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.8479539752006531,
      "learning_rate": 0.0003702971919631564,
      "loss": 2.1777,
      "step": 97855
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.823608934879303,
      "learning_rate": 0.0003702932152815707,
      "loss": 2.2343,
      "step": 97856
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.886371374130249,
      "learning_rate": 0.00037028923858691595,
      "loss": 2.0779,
      "step": 97857
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.8908605575561523,
      "learning_rate": 0.00037028526187919317,
      "loss": 2.2846,
      "step": 97858
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.9530543088912964,
      "learning_rate": 0.0003702812851584031,
      "loss": 2.2111,
      "step": 97859
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.8467422127723694,
      "learning_rate": 0.0003702773084245463,
      "loss": 2.211,
      "step": 97860
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.8931568264961243,
      "learning_rate": 0.0003702733316776237,
      "loss": 2.0431,
      "step": 97861
    },
    {
      "epoch": 1.27,
      "grad_norm": 1.0639978647232056,
      "learning_rate": 0.00037026935491763597,
      "loss": 2.1554,
      "step": 97862
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.8998844027519226,
      "learning_rate": 0.0003702653781445838,
      "loss": 2.0909,
      "step": 97863
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.8984927535057068,
      "learning_rate": 0.000370261401358468,
      "loss": 2.0475,
      "step": 97864
    },
    {
      "epoch": 1.27,
      "grad_norm": 1.339133620262146,
      "learning_rate": 0.0003702574245592893,
      "loss": 2.0844,
      "step": 97865
    },
    {
      "epoch": 1.27,
      "grad_norm": 1.0994800329208374,
      "learning_rate": 0.0003702534477470483,
      "loss": 1.9674,
      "step": 97866
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.8381998538970947,
      "learning_rate": 0.00037024947092174595,
      "loss": 2.0825,
      "step": 97867
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.813854455947876,
      "learning_rate": 0.0003702454940833829,
      "loss": 2.0912,
      "step": 97868
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.8867412805557251,
      "learning_rate": 0.00037024151723195984,
      "loss": 2.2385,
      "step": 97869
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.9090791344642639,
      "learning_rate": 0.0003702375403674776,
      "loss": 2.1247,
      "step": 97870
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.9489503502845764,
      "learning_rate": 0.00037023356348993687,
      "loss": 2.0071,
      "step": 97871
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.829222559928894,
      "learning_rate": 0.0003702295865993384,
      "loss": 2.1565,
      "step": 97872
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.9630869030952454,
      "learning_rate": 0.00037022560969568284,
      "loss": 2.0774,
      "step": 97873
    },
    {
      "epoch": 1.27,
      "grad_norm": 1.1720292568206787,
      "learning_rate": 0.0003702216327789712,
      "loss": 2.2144,
      "step": 97874
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.8657166361808777,
      "learning_rate": 0.0003702176558492038,
      "loss": 2.0538,
      "step": 97875
    },
    {
      "epoch": 1.27,
      "grad_norm": 1.0160887241363525,
      "learning_rate": 0.0003702136789063818,
      "loss": 2.2315,
      "step": 97876
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.8669856786727905,
      "learning_rate": 0.0003702097019505057,
      "loss": 2.2416,
      "step": 97877
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.8969664573669434,
      "learning_rate": 0.00037020572498157626,
      "loss": 2.1181,
      "step": 97878
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.8909700512886047,
      "learning_rate": 0.00037020174799959425,
      "loss": 2.0336,
      "step": 97879
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.8322502970695496,
      "learning_rate": 0.0003701977710045604,
      "loss": 2.2217,
      "step": 97880
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.7792573571205139,
      "learning_rate": 0.0003701937939964755,
      "loss": 2.3762,
      "step": 97881
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.8967667818069458,
      "learning_rate": 0.0003701898169753402,
      "loss": 2.2627,
      "step": 97882
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.8021142482757568,
      "learning_rate": 0.00037018583994115535,
      "loss": 2.2987,
      "step": 97883
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.8783514499664307,
      "learning_rate": 0.0003701818628939216,
      "loss": 2.0014,
      "step": 97884
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.8307021260261536,
      "learning_rate": 0.00037017788583363964,
      "loss": 1.8981,
      "step": 97885
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.8834204077720642,
      "learning_rate": 0.00037017390876031045,
      "loss": 2.2699,
      "step": 97886
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.83629310131073,
      "learning_rate": 0.0003701699316739345,
      "loss": 2.1678,
      "step": 97887
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.88653963804245,
      "learning_rate": 0.00037016595457451263,
      "loss": 2.2181,
      "step": 97888
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.8876470923423767,
      "learning_rate": 0.0003701619774620456,
      "loss": 2.0317,
      "step": 97889
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.9234925508499146,
      "learning_rate": 0.0003701580003365342,
      "loss": 1.9631,
      "step": 97890
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.7715022563934326,
      "learning_rate": 0.000370154023197979,
      "loss": 2.0952,
      "step": 97891
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.9104474186897278,
      "learning_rate": 0.00037015004604638087,
      "loss": 2.1617,
      "step": 97892
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.8659257292747498,
      "learning_rate": 0.0003701460688817405,
      "loss": 2.0514,
      "step": 97893
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.9069700837135315,
      "learning_rate": 0.00037014209170405874,
      "loss": 2.1014,
      "step": 97894
    },
    {
      "epoch": 1.27,
      "grad_norm": 1.007043480873108,
      "learning_rate": 0.00037013811451333626,
      "loss": 1.9449,
      "step": 97895
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.836327850818634,
      "learning_rate": 0.00037013413730957364,
      "loss": 2.1869,
      "step": 97896
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.968454122543335,
      "learning_rate": 0.0003701301600927718,
      "loss": 2.0559,
      "step": 97897
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.9660010933876038,
      "learning_rate": 0.00037012618286293153,
      "loss": 2.2344,
      "step": 97898
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.9899651408195496,
      "learning_rate": 0.0003701222056200534,
      "loss": 2.1338,
      "step": 97899
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.985539436340332,
      "learning_rate": 0.0003701182283641382,
      "loss": 2.1985,
      "step": 97900
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.8593795895576477,
      "learning_rate": 0.0003701142510951868,
      "loss": 2.0355,
      "step": 97901
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.9114359021186829,
      "learning_rate": 0.00037011027381319975,
      "loss": 2.1325,
      "step": 97902
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.9982810020446777,
      "learning_rate": 0.0003701062965181779,
      "loss": 2.1714,
      "step": 97903
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.8617250919342041,
      "learning_rate": 0.00037010231921012207,
      "loss": 2.098,
      "step": 97904
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.9419636726379395,
      "learning_rate": 0.00037009834188903273,
      "loss": 2.2092,
      "step": 97905
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.8120498657226562,
      "learning_rate": 0.00037009436455491085,
      "loss": 2.2871,
      "step": 97906
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.7872208952903748,
      "learning_rate": 0.0003700903872077572,
      "loss": 2.1295,
      "step": 97907
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.8870699405670166,
      "learning_rate": 0.0003700864098475722,
      "loss": 2.0507,
      "step": 97908
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.7662898898124695,
      "learning_rate": 0.000370082432474357,
      "loss": 2.003,
      "step": 97909
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.8521799445152283,
      "learning_rate": 0.0003700784550881121,
      "loss": 2.0227,
      "step": 97910
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.8746744990348816,
      "learning_rate": 0.00037007447768883827,
      "loss": 2.205,
      "step": 97911
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.8222479224205017,
      "learning_rate": 0.0003700705002765363,
      "loss": 2.0832,
      "step": 97912
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.9648885726928711,
      "learning_rate": 0.00037006652285120694,
      "loss": 2.1362,
      "step": 97913
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.8474863171577454,
      "learning_rate": 0.0003700625454128508,
      "loss": 2.1093,
      "step": 97914
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.8535182476043701,
      "learning_rate": 0.00037005856796146875,
      "loss": 2.2003,
      "step": 97915
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.845017671585083,
      "learning_rate": 0.00037005459049706154,
      "loss": 2.0844,
      "step": 97916
    },
    {
      "epoch": 1.27,
      "grad_norm": 1.046100378036499,
      "learning_rate": 0.00037005061301962976,
      "loss": 2.1073,
      "step": 97917
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.8354191184043884,
      "learning_rate": 0.00037004663552917433,
      "loss": 2.1466,
      "step": 97918
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.8803008794784546,
      "learning_rate": 0.00037004265802569584,
      "loss": 2.1742,
      "step": 97919
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.9021702408790588,
      "learning_rate": 0.0003700386805091952,
      "loss": 2.1761,
      "step": 97920
    },
    {
      "epoch": 1.27,
      "grad_norm": 0.7613050937652588,
      "learning_rate": 0.00037003470297967294,
      "loss": 2.1128,
      "step": 97921
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.7864131331443787,
      "learning_rate": 0.00037003072543712994,
      "loss": 2.0794,
      "step": 97922
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.8631594777107239,
      "learning_rate": 0.000370026747881567,
      "loss": 2.2678,
      "step": 97923
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.827754020690918,
      "learning_rate": 0.00037002277031298464,
      "loss": 2.1579,
      "step": 97924
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.9501103758811951,
      "learning_rate": 0.0003700187927313838,
      "loss": 1.9215,
      "step": 97925
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.8351958990097046,
      "learning_rate": 0.00037001481513676505,
      "loss": 1.9417,
      "step": 97926
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.8097123503684998,
      "learning_rate": 0.00037001083752912937,
      "loss": 2.0734,
      "step": 97927
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.917846143245697,
      "learning_rate": 0.00037000685990847725,
      "loss": 2.0393,
      "step": 97928
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.8384203314781189,
      "learning_rate": 0.0003700028822748095,
      "loss": 2.2051,
      "step": 97929
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.8749582171440125,
      "learning_rate": 0.000369998904628127,
      "loss": 2.1227,
      "step": 97930
    },
    {
      "epoch": 1.28,
      "grad_norm": 1.1208981275558472,
      "learning_rate": 0.0003699949269684303,
      "loss": 2.2334,
      "step": 97931
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.9667282104492188,
      "learning_rate": 0.0003699909492957202,
      "loss": 2.2411,
      "step": 97932
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.8285638689994812,
      "learning_rate": 0.0003699869716099975,
      "loss": 2.2933,
      "step": 97933
    },
    {
      "epoch": 1.28,
      "grad_norm": 1.0696635246276855,
      "learning_rate": 0.000369982993911263,
      "loss": 2.264,
      "step": 97934
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.8443225622177124,
      "learning_rate": 0.00036997901619951725,
      "loss": 2.3542,
      "step": 97935
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.8545307517051697,
      "learning_rate": 0.00036997503847476106,
      "loss": 2.0602,
      "step": 97936
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.8978962898254395,
      "learning_rate": 0.0003699710607369952,
      "loss": 2.3055,
      "step": 97937
    },
    {
      "epoch": 1.28,
      "grad_norm": 1.0091878175735474,
      "learning_rate": 0.00036996708298622045,
      "loss": 2.2432,
      "step": 97938
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.8495324850082397,
      "learning_rate": 0.00036996310522243743,
      "loss": 2.4995,
      "step": 97939
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.8603386282920837,
      "learning_rate": 0.00036995912744564697,
      "loss": 2.1093,
      "step": 97940
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.9297297596931458,
      "learning_rate": 0.0003699551496558499,
      "loss": 2.0019,
      "step": 97941
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.807891845703125,
      "learning_rate": 0.00036995117185304664,
      "loss": 2.0417,
      "step": 97942
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.9975658059120178,
      "learning_rate": 0.00036994719403723824,
      "loss": 2.1846,
      "step": 97943
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.8842437267303467,
      "learning_rate": 0.00036994321620842537,
      "loss": 2.1589,
      "step": 97944
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.8328912258148193,
      "learning_rate": 0.0003699392383666087,
      "loss": 2.1295,
      "step": 97945
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.9374631643295288,
      "learning_rate": 0.000369935260511789,
      "loss": 1.9099,
      "step": 97946
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.8764816522598267,
      "learning_rate": 0.0003699312826439671,
      "loss": 2.1512,
      "step": 97947
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.8908044099807739,
      "learning_rate": 0.0003699273047631436,
      "loss": 2.1918,
      "step": 97948
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.9487879872322083,
      "learning_rate": 0.0003699233268693192,
      "loss": 2.1572,
      "step": 97949
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.8461498618125916,
      "learning_rate": 0.0003699193489624948,
      "loss": 2.3348,
      "step": 97950
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.8961308002471924,
      "learning_rate": 0.0003699153710426711,
      "loss": 2.2117,
      "step": 97951
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.9109858274459839,
      "learning_rate": 0.00036991139310984877,
      "loss": 2.2998,
      "step": 97952
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.8257408142089844,
      "learning_rate": 0.00036990741516402864,
      "loss": 2.0971,
      "step": 97953
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.8233646750450134,
      "learning_rate": 0.00036990343720521136,
      "loss": 1.8683,
      "step": 97954
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.9071213603019714,
      "learning_rate": 0.0003698994592333977,
      "loss": 2.2119,
      "step": 97955
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.8898829221725464,
      "learning_rate": 0.00036989548124858853,
      "loss": 2.3448,
      "step": 97956
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.8016321659088135,
      "learning_rate": 0.0003698915032507843,
      "loss": 2.3267,
      "step": 97957
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.8638981580734253,
      "learning_rate": 0.000369887525239986,
      "loss": 2.1214,
      "step": 97958
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.9627346992492676,
      "learning_rate": 0.0003698835472161944,
      "loss": 2.2624,
      "step": 97959
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.9066141247749329,
      "learning_rate": 0.00036987956917940997,
      "loss": 2.3162,
      "step": 97960
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.7824603915214539,
      "learning_rate": 0.0003698755911296336,
      "loss": 2.1173,
      "step": 97961
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.9820733070373535,
      "learning_rate": 0.0003698716130668661,
      "loss": 2.3178,
      "step": 97962
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.9182317852973938,
      "learning_rate": 0.0003698676349911082,
      "loss": 2.0573,
      "step": 97963
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.8675232529640198,
      "learning_rate": 0.00036986365690236053,
      "loss": 2.2006,
      "step": 97964
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.8104952573776245,
      "learning_rate": 0.0003698596788006239,
      "loss": 2.1955,
      "step": 97965
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.9416711330413818,
      "learning_rate": 0.000369855700685899,
      "loss": 2.2564,
      "step": 97966
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.8974320888519287,
      "learning_rate": 0.0003698517225581866,
      "loss": 2.058,
      "step": 97967
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.8424621224403381,
      "learning_rate": 0.0003698477444174876,
      "loss": 2.1591,
      "step": 97968
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.8849013447761536,
      "learning_rate": 0.0003698437662638024,
      "loss": 2.2028,
      "step": 97969
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.9913801550865173,
      "learning_rate": 0.00036983978809713194,
      "loss": 2.1136,
      "step": 97970
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.9087719917297363,
      "learning_rate": 0.00036983580991747705,
      "loss": 2.1697,
      "step": 97971
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.8778363466262817,
      "learning_rate": 0.00036983183172483835,
      "loss": 2.296,
      "step": 97972
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.872819721698761,
      "learning_rate": 0.0003698278535192165,
      "loss": 2.1855,
      "step": 97973
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.9249194264411926,
      "learning_rate": 0.0003698238753006124,
      "loss": 1.927,
      "step": 97974
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.9398555755615234,
      "learning_rate": 0.0003698198970690267,
      "loss": 2.1199,
      "step": 97975
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.819281280040741,
      "learning_rate": 0.0003698159188244602,
      "loss": 2.3865,
      "step": 97976
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.886223316192627,
      "learning_rate": 0.00036981194056691363,
      "loss": 2.1393,
      "step": 97977
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.9591232538223267,
      "learning_rate": 0.00036980796229638764,
      "loss": 2.2612,
      "step": 97978
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.9099708199501038,
      "learning_rate": 0.000369803984012883,
      "loss": 2.1206,
      "step": 97979
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.858498752117157,
      "learning_rate": 0.00036980000571640054,
      "loss": 2.1743,
      "step": 97980
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.8253882527351379,
      "learning_rate": 0.00036979602740694096,
      "loss": 2.2097,
      "step": 97981
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.867577850818634,
      "learning_rate": 0.00036979204908450493,
      "loss": 2.346,
      "step": 97982
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.8794089555740356,
      "learning_rate": 0.0003697880707490933,
      "loss": 2.2816,
      "step": 97983
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.9531741738319397,
      "learning_rate": 0.00036978409240070667,
      "loss": 2.2626,
      "step": 97984
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.9413313865661621,
      "learning_rate": 0.0003697801140393459,
      "loss": 2.1611,
      "step": 97985
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.8456212878227234,
      "learning_rate": 0.00036977613566501177,
      "loss": 1.9092,
      "step": 97986
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.8630843162536621,
      "learning_rate": 0.00036977215727770484,
      "loss": 2.2153,
      "step": 97987
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.9626142978668213,
      "learning_rate": 0.00036976817887742587,
      "loss": 2.1423,
      "step": 97988
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.8796635866165161,
      "learning_rate": 0.00036976420046417577,
      "loss": 1.9346,
      "step": 97989
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.8182818293571472,
      "learning_rate": 0.00036976022203795525,
      "loss": 2.1129,
      "step": 97990
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.8811280727386475,
      "learning_rate": 0.00036975624359876486,
      "loss": 2.2106,
      "step": 97991
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.9127559065818787,
      "learning_rate": 0.00036975226514660557,
      "loss": 2.2646,
      "step": 97992
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.8709209561347961,
      "learning_rate": 0.000369748286681478,
      "loss": 2.1492,
      "step": 97993
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.8662002682685852,
      "learning_rate": 0.00036974430820338293,
      "loss": 2.1231,
      "step": 97994
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.9556289315223694,
      "learning_rate": 0.000369740329712321,
      "loss": 2.1641,
      "step": 97995
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.78654545545578,
      "learning_rate": 0.0003697363512082931,
      "loss": 2.2335,
      "step": 97996
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.7976709008216858,
      "learning_rate": 0.00036973237269129986,
      "loss": 2.3601,
      "step": 97997
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.8851398825645447,
      "learning_rate": 0.000369728394161342,
      "loss": 2.072,
      "step": 97998
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.8420338034629822,
      "learning_rate": 0.00036972441561842047,
      "loss": 2.1966,
      "step": 97999
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.9492038488388062,
      "learning_rate": 0.0003697204370625357,
      "loss": 2.1851,
      "step": 98000
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.8692975044250488,
      "learning_rate": 0.0003697164584936886,
      "loss": 2.2204,
      "step": 98001
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.7920264005661011,
      "learning_rate": 0.00036971247991188,
      "loss": 1.8952,
      "step": 98002
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.8177841305732727,
      "learning_rate": 0.0003697085013171105,
      "loss": 2.0025,
      "step": 98003
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.8701744675636292,
      "learning_rate": 0.0003697045227093808,
      "loss": 2.4905,
      "step": 98004
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.9132468104362488,
      "learning_rate": 0.0003697005440886918,
      "loss": 2.2072,
      "step": 98005
    },
    {
      "epoch": 1.28,
      "grad_norm": 1.1116209030151367,
      "learning_rate": 0.0003696965654550441,
      "loss": 2.3051,
      "step": 98006
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.9664498567581177,
      "learning_rate": 0.00036969258680843845,
      "loss": 2.0906,
      "step": 98007
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.8825789093971252,
      "learning_rate": 0.00036968860814887573,
      "loss": 2.1389,
      "step": 98008
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.8548176288604736,
      "learning_rate": 0.00036968462947635657,
      "loss": 2.0845,
      "step": 98009
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.8001404404640198,
      "learning_rate": 0.0003696806507908817,
      "loss": 2.0848,
      "step": 98010
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.917245626449585,
      "learning_rate": 0.00036967667209245183,
      "loss": 2.1688,
      "step": 98011
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.8862665891647339,
      "learning_rate": 0.0003696726933810678,
      "loss": 2.2954,
      "step": 98012
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.873155415058136,
      "learning_rate": 0.00036966871465673025,
      "loss": 2.0175,
      "step": 98013
    },
    {
      "epoch": 1.28,
      "grad_norm": 1.0031638145446777,
      "learning_rate": 0.00036966473591944004,
      "loss": 2.2537,
      "step": 98014
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.8074946403503418,
      "learning_rate": 0.00036966075716919785,
      "loss": 2.1698,
      "step": 98015
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.9383465647697449,
      "learning_rate": 0.00036965677840600437,
      "loss": 2.4132,
      "step": 98016
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.8588472008705139,
      "learning_rate": 0.00036965279962986036,
      "loss": 2.465,
      "step": 98017
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.8448910117149353,
      "learning_rate": 0.00036964882084076654,
      "loss": 2.2376,
      "step": 98018
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.8286255598068237,
      "learning_rate": 0.0003696448420387239,
      "loss": 2.2129,
      "step": 98019
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.8732878565788269,
      "learning_rate": 0.0003696408632237328,
      "loss": 2.1304,
      "step": 98020
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.8967863321304321,
      "learning_rate": 0.00036963688439579413,
      "loss": 2.3107,
      "step": 98021
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.8461942076683044,
      "learning_rate": 0.0003696329055549087,
      "loss": 2.1789,
      "step": 98022
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.7898326516151428,
      "learning_rate": 0.00036962892670107714,
      "loss": 2.1674,
      "step": 98023
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.8974694609642029,
      "learning_rate": 0.0003696249478343003,
      "loss": 1.9401,
      "step": 98024
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.986656129360199,
      "learning_rate": 0.0003696209689545789,
      "loss": 2.1324,
      "step": 98025
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.8275165557861328,
      "learning_rate": 0.00036961699006191366,
      "loss": 2.1474,
      "step": 98026
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.8351776599884033,
      "learning_rate": 0.0003696130111563052,
      "loss": 2.197,
      "step": 98027
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.9050534963607788,
      "learning_rate": 0.0003696090322377544,
      "loss": 2.0053,
      "step": 98028
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.8399314284324646,
      "learning_rate": 0.0003696050533062621,
      "loss": 2.1296,
      "step": 98029
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.8164360523223877,
      "learning_rate": 0.0003696010743618287,
      "loss": 1.9719,
      "step": 98030
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.8125045895576477,
      "learning_rate": 0.00036959709540445527,
      "loss": 2.173,
      "step": 98031
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.7840482592582703,
      "learning_rate": 0.0003695931164341424,
      "loss": 2.114,
      "step": 98032
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.8201563358306885,
      "learning_rate": 0.00036958913745089087,
      "loss": 2.1624,
      "step": 98033
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.847317636013031,
      "learning_rate": 0.0003695851584547014,
      "loss": 1.8959,
      "step": 98034
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.9236482381820679,
      "learning_rate": 0.0003695811794455748,
      "loss": 2.1893,
      "step": 98035
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.9172638654708862,
      "learning_rate": 0.00036957720042351167,
      "loss": 2.2402,
      "step": 98036
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.990562915802002,
      "learning_rate": 0.00036957322138851277,
      "loss": 2.0866,
      "step": 98037
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.8728283643722534,
      "learning_rate": 0.000369569242340579,
      "loss": 2.2439,
      "step": 98038
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.9270635843276978,
      "learning_rate": 0.00036956526327971094,
      "loss": 2.4042,
      "step": 98039
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.8442248702049255,
      "learning_rate": 0.00036956128420590937,
      "loss": 2.1074,
      "step": 98040
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.8272258043289185,
      "learning_rate": 0.00036955730511917516,
      "loss": 2.2537,
      "step": 98041
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.9700557589530945,
      "learning_rate": 0.00036955332601950887,
      "loss": 2.2322,
      "step": 98042
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.8703809380531311,
      "learning_rate": 0.00036954934690691123,
      "loss": 2.1857,
      "step": 98043
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.8416350483894348,
      "learning_rate": 0.00036954536778138313,
      "loss": 1.9647,
      "step": 98044
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.9114974141120911,
      "learning_rate": 0.00036954138864292517,
      "loss": 2.2019,
      "step": 98045
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.8912805914878845,
      "learning_rate": 0.00036953740949153814,
      "loss": 2.1135,
      "step": 98046
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.8609374165534973,
      "learning_rate": 0.0003695334303272229,
      "loss": 2.028,
      "step": 98047
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.8754103779792786,
      "learning_rate": 0.00036952945114998,
      "loss": 2.3441,
      "step": 98048
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.9136238694190979,
      "learning_rate": 0.0003695254719598103,
      "loss": 2.0487,
      "step": 98049
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.8988152146339417,
      "learning_rate": 0.00036952149275671454,
      "loss": 2.1457,
      "step": 98050
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.8221673965454102,
      "learning_rate": 0.00036951751354069335,
      "loss": 2.1345,
      "step": 98051
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.8715649843215942,
      "learning_rate": 0.0003695135343117475,
      "loss": 2.4344,
      "step": 98052
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.8873856663703918,
      "learning_rate": 0.0003695095550698779,
      "loss": 2.1937,
      "step": 98053
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.8520938754081726,
      "learning_rate": 0.0003695055758150851,
      "loss": 2.0648,
      "step": 98054
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.8347030282020569,
      "learning_rate": 0.0003695015965473699,
      "loss": 1.845,
      "step": 98055
    },
    {
      "epoch": 1.28,
      "grad_norm": 1.0058231353759766,
      "learning_rate": 0.000369497617266733,
      "loss": 2.1808,
      "step": 98056
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.8882647752761841,
      "learning_rate": 0.0003694936379731753,
      "loss": 2.5072,
      "step": 98057
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.8622666001319885,
      "learning_rate": 0.0003694896586666973,
      "loss": 2.2989,
      "step": 98058
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.966990053653717,
      "learning_rate": 0.0003694856793472999,
      "loss": 2.0052,
      "step": 98059
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.8625115752220154,
      "learning_rate": 0.0003694817000149838,
      "loss": 2.4967,
      "step": 98060
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.8745571970939636,
      "learning_rate": 0.0003694777206697498,
      "loss": 2.2493,
      "step": 98061
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.8028920888900757,
      "learning_rate": 0.0003694737413115984,
      "loss": 2.4537,
      "step": 98062
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.8779381513595581,
      "learning_rate": 0.0003694697619405308,
      "loss": 2.1031,
      "step": 98063
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.8628004193305969,
      "learning_rate": 0.00036946578255654726,
      "loss": 2.0367,
      "step": 98064
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.7961117625236511,
      "learning_rate": 0.0003694618031596487,
      "loss": 2.0171,
      "step": 98065
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.8761612176895142,
      "learning_rate": 0.000369457823749836,
      "loss": 2.0825,
      "step": 98066
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.8233599066734314,
      "learning_rate": 0.0003694538443271097,
      "loss": 2.128,
      "step": 98067
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.822247326374054,
      "learning_rate": 0.0003694498648914706,
      "loss": 1.816,
      "step": 98068
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.8691991567611694,
      "learning_rate": 0.0003694458854429196,
      "loss": 1.8702,
      "step": 98069
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.9924868941307068,
      "learning_rate": 0.0003694419059814572,
      "loss": 2.1945,
      "step": 98070
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.9113412499427795,
      "learning_rate": 0.00036943792650708413,
      "loss": 2.3542,
      "step": 98071
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.957106351852417,
      "learning_rate": 0.00036943394701980146,
      "loss": 2.0788,
      "step": 98072
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.8739374279975891,
      "learning_rate": 0.0003694299675196095,
      "loss": 2.1277,
      "step": 98073
    },
    {
      "epoch": 1.28,
      "grad_norm": 1.0041329860687256,
      "learning_rate": 0.00036942598800650934,
      "loss": 2.255,
      "step": 98074
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.8007447719573975,
      "learning_rate": 0.00036942200848050157,
      "loss": 2.2196,
      "step": 98075
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.8908689618110657,
      "learning_rate": 0.0003694180289415869,
      "loss": 2.162,
      "step": 98076
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.9806926846504211,
      "learning_rate": 0.0003694140493897661,
      "loss": 2.2461,
      "step": 98077
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.9048973917961121,
      "learning_rate": 0.00036941006982503993,
      "loss": 2.3119,
      "step": 98078
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.9204557538032532,
      "learning_rate": 0.0003694060902474091,
      "loss": 2.0707,
      "step": 98079
    },
    {
      "epoch": 1.28,
      "grad_norm": 1.1321470737457275,
      "learning_rate": 0.0003694021106568744,
      "loss": 2.3385,
      "step": 98080
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.8141705989837646,
      "learning_rate": 0.00036939813105343655,
      "loss": 2.4704,
      "step": 98081
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.8299152255058289,
      "learning_rate": 0.0003693941514370962,
      "loss": 2.1149,
      "step": 98082
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.9226096868515015,
      "learning_rate": 0.00036939017180785417,
      "loss": 1.9438,
      "step": 98083
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.9031182527542114,
      "learning_rate": 0.0003693861921657113,
      "loss": 2.2105,
      "step": 98084
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.8566308617591858,
      "learning_rate": 0.0003693822125106682,
      "loss": 1.9722,
      "step": 98085
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.9823435544967651,
      "learning_rate": 0.00036937823284272553,
      "loss": 2.0935,
      "step": 98086
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.8546844124794006,
      "learning_rate": 0.00036937425316188426,
      "loss": 2.2256,
      "step": 98087
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.8892474174499512,
      "learning_rate": 0.000369370273468145,
      "loss": 2.2517,
      "step": 98088
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.8617905974388123,
      "learning_rate": 0.00036936629376150843,
      "loss": 2.3106,
      "step": 98089
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.8924515247344971,
      "learning_rate": 0.00036936231404197545,
      "loss": 2.1152,
      "step": 98090
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.8815918564796448,
      "learning_rate": 0.0003693583343095466,
      "loss": 2.1742,
      "step": 98091
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.884477436542511,
      "learning_rate": 0.0003693543545642227,
      "loss": 2.0394,
      "step": 98092
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.9110785722732544,
      "learning_rate": 0.0003693503748060047,
      "loss": 2.1254,
      "step": 98093
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.9250278472900391,
      "learning_rate": 0.000369346395034893,
      "loss": 2.0561,
      "step": 98094
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.8599076867103577,
      "learning_rate": 0.00036934241525088846,
      "loss": 2.2169,
      "step": 98095
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.8473698496818542,
      "learning_rate": 0.000369338435453992,
      "loss": 2.27,
      "step": 98096
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.8862435221672058,
      "learning_rate": 0.00036933445564420414,
      "loss": 2.0887,
      "step": 98097
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.8816662430763245,
      "learning_rate": 0.0003693304758215257,
      "loss": 2.0761,
      "step": 98098
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.8658509254455566,
      "learning_rate": 0.0003693264959859574,
      "loss": 2.1741,
      "step": 98099
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.871924877166748,
      "learning_rate": 0.00036932251613750003,
      "loss": 2.1203,
      "step": 98100
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.8966900706291199,
      "learning_rate": 0.0003693185362761543,
      "loss": 2.0697,
      "step": 98101
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.8180153965950012,
      "learning_rate": 0.0003693145564019209,
      "loss": 2.0086,
      "step": 98102
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.8261520862579346,
      "learning_rate": 0.0003693105765148007,
      "loss": 2.1387,
      "step": 98103
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.8915006518363953,
      "learning_rate": 0.00036930659661479426,
      "loss": 2.2113,
      "step": 98104
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.9522327780723572,
      "learning_rate": 0.00036930261670190244,
      "loss": 2.2214,
      "step": 98105
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.8475140333175659,
      "learning_rate": 0.00036929863677612597,
      "loss": 2.0805,
      "step": 98106
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.9931635856628418,
      "learning_rate": 0.0003692946568374656,
      "loss": 2.4113,
      "step": 98107
    },
    {
      "epoch": 1.28,
      "grad_norm": 1.0059239864349365,
      "learning_rate": 0.00036929067688592197,
      "loss": 2.2139,
      "step": 98108
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.840686023235321,
      "learning_rate": 0.00036928669692149596,
      "loss": 2.3704,
      "step": 98109
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.8537137508392334,
      "learning_rate": 0.0003692827169441882,
      "loss": 2.1216,
      "step": 98110
    },
    {
      "epoch": 1.28,
      "grad_norm": 1.0080851316452026,
      "learning_rate": 0.0003692787369539995,
      "loss": 2.0496,
      "step": 98111
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.8779574632644653,
      "learning_rate": 0.00036927475695093065,
      "loss": 2.17,
      "step": 98112
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.843890905380249,
      "learning_rate": 0.00036927077693498216,
      "loss": 2.4294,
      "step": 98113
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.9665573835372925,
      "learning_rate": 0.00036926679690615504,
      "loss": 2.1501,
      "step": 98114
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.9845128655433655,
      "learning_rate": 0.0003692628168644499,
      "loss": 2.1176,
      "step": 98115
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.9232884049415588,
      "learning_rate": 0.00036925883680986746,
      "loss": 2.1128,
      "step": 98116
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.8388685584068298,
      "learning_rate": 0.00036925485674240853,
      "loss": 2.3369,
      "step": 98117
    },
    {
      "epoch": 1.28,
      "grad_norm": 1.101932168006897,
      "learning_rate": 0.0003692508766620738,
      "loss": 2.1457,
      "step": 98118
    },
    {
      "epoch": 1.28,
      "grad_norm": 1.017274260520935,
      "learning_rate": 0.000369246896568864,
      "loss": 2.1775,
      "step": 98119
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.8407726883888245,
      "learning_rate": 0.0003692429164627799,
      "loss": 2.1129,
      "step": 98120
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.8937073945999146,
      "learning_rate": 0.0003692389363438223,
      "loss": 2.4502,
      "step": 98121
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.9148039817810059,
      "learning_rate": 0.0003692349562119918,
      "loss": 1.8955,
      "step": 98122
    },
    {
      "epoch": 1.28,
      "grad_norm": 1.0493474006652832,
      "learning_rate": 0.0003692309760672893,
      "loss": 2.306,
      "step": 98123
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.7911824584007263,
      "learning_rate": 0.0003692269959097153,
      "loss": 2.1808,
      "step": 98124
    },
    {
      "epoch": 1.28,
      "grad_norm": 1.0187036991119385,
      "learning_rate": 0.00036922301573927085,
      "loss": 2.0566,
      "step": 98125
    },
    {
      "epoch": 1.28,
      "grad_norm": 1.3040552139282227,
      "learning_rate": 0.00036921903555595646,
      "loss": 2.0896,
      "step": 98126
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.7989833354949951,
      "learning_rate": 0.000369215055359773,
      "loss": 2.0815,
      "step": 98127
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.841428279876709,
      "learning_rate": 0.00036921107515072114,
      "loss": 2.1358,
      "step": 98128
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.9424060583114624,
      "learning_rate": 0.00036920709492880156,
      "loss": 2.1005,
      "step": 98129
    },
    {
      "epoch": 1.28,
      "grad_norm": 1.2494571208953857,
      "learning_rate": 0.00036920311469401516,
      "loss": 2.1361,
      "step": 98130
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.9188781380653381,
      "learning_rate": 0.00036919913444636253,
      "loss": 2.0243,
      "step": 98131
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.8361650109291077,
      "learning_rate": 0.0003691951541858445,
      "loss": 1.9296,
      "step": 98132
    },
    {
      "epoch": 1.28,
      "grad_norm": 1.1023588180541992,
      "learning_rate": 0.00036919117391246185,
      "loss": 2.1493,
      "step": 98133
    },
    {
      "epoch": 1.28,
      "grad_norm": 1.0910344123840332,
      "learning_rate": 0.0003691871936262152,
      "loss": 2.3046,
      "step": 98134
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.8302300572395325,
      "learning_rate": 0.00036918321332710525,
      "loss": 2.2457,
      "step": 98135
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.8154672384262085,
      "learning_rate": 0.000369179233015133,
      "loss": 2.1916,
      "step": 98136
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.8499454855918884,
      "learning_rate": 0.00036917525269029893,
      "loss": 2.1721,
      "step": 98137
    },
    {
      "epoch": 1.28,
      "grad_norm": 1.0436663627624512,
      "learning_rate": 0.0003691712723526039,
      "loss": 2.1738,
      "step": 98138
    },
    {
      "epoch": 1.28,
      "grad_norm": 1.182785987854004,
      "learning_rate": 0.0003691672920020486,
      "loss": 2.3417,
      "step": 98139
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.7601277232170105,
      "learning_rate": 0.0003691633116386338,
      "loss": 2.267,
      "step": 98140
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.9355464577674866,
      "learning_rate": 0.0003691593312623603,
      "loss": 2.0482,
      "step": 98141
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.8859689235687256,
      "learning_rate": 0.0003691553508732287,
      "loss": 2.0913,
      "step": 98142
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.8794100284576416,
      "learning_rate": 0.00036915137047123986,
      "loss": 2.2153,
      "step": 98143
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.7620164155960083,
      "learning_rate": 0.00036914739005639436,
      "loss": 2.3456,
      "step": 98144
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.7591941952705383,
      "learning_rate": 0.0003691434096286932,
      "loss": 2.1932,
      "step": 98145
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.8581095337867737,
      "learning_rate": 0.00036913942918813695,
      "loss": 2.0322,
      "step": 98146
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.854902446269989,
      "learning_rate": 0.0003691354487347263,
      "loss": 2.1929,
      "step": 98147
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.8703321218490601,
      "learning_rate": 0.0003691314682684622,
      "loss": 2.4683,
      "step": 98148
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.8234686255455017,
      "learning_rate": 0.0003691274877893451,
      "loss": 2.3987,
      "step": 98149
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.7989399433135986,
      "learning_rate": 0.00036912350729737594,
      "loss": 2.1328,
      "step": 98150
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.7970271706581116,
      "learning_rate": 0.00036911952679255554,
      "loss": 2.3488,
      "step": 98151
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.8998937606811523,
      "learning_rate": 0.00036911554627488437,
      "loss": 2.0314,
      "step": 98152
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.7599282264709473,
      "learning_rate": 0.0003691115657443633,
      "loss": 2.1944,
      "step": 98153
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.8972724676132202,
      "learning_rate": 0.0003691075852009932,
      "loss": 2.1265,
      "step": 98154
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.9709770679473877,
      "learning_rate": 0.00036910360464477464,
      "loss": 2.0722,
      "step": 98155
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.8598114848136902,
      "learning_rate": 0.0003690996240757084,
      "loss": 2.2419,
      "step": 98156
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.9505510926246643,
      "learning_rate": 0.0003690956434937953,
      "loss": 2.0384,
      "step": 98157
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.8142566084861755,
      "learning_rate": 0.00036909166289903595,
      "loss": 2.0117,
      "step": 98158
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.8772894144058228,
      "learning_rate": 0.00036908768229143113,
      "loss": 2.0267,
      "step": 98159
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.9343976974487305,
      "learning_rate": 0.0003690837016709817,
      "loss": 2.3484,
      "step": 98160
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.8351538181304932,
      "learning_rate": 0.00036907972103768824,
      "loss": 2.1736,
      "step": 98161
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.927865743637085,
      "learning_rate": 0.0003690757403915515,
      "loss": 2.3908,
      "step": 98162
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.923323929309845,
      "learning_rate": 0.00036907175973257243,
      "loss": 2.2004,
      "step": 98163
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.7855128645896912,
      "learning_rate": 0.0003690677790607515,
      "loss": 2.0587,
      "step": 98164
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.9525250792503357,
      "learning_rate": 0.0003690637983760896,
      "loss": 2.0925,
      "step": 98165
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.805466890335083,
      "learning_rate": 0.00036905981767858747,
      "loss": 2.2715,
      "step": 98166
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.8903037309646606,
      "learning_rate": 0.00036905583696824576,
      "loss": 2.4004,
      "step": 98167
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.9513407349586487,
      "learning_rate": 0.00036905185624506526,
      "loss": 2.1944,
      "step": 98168
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.8704217076301575,
      "learning_rate": 0.0003690478755090468,
      "loss": 2.3256,
      "step": 98169
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.8713609576225281,
      "learning_rate": 0.0003690438947601909,
      "loss": 2.0731,
      "step": 98170
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.9306651949882507,
      "learning_rate": 0.0003690399139984985,
      "loss": 2.2214,
      "step": 98171
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.8129138350486755,
      "learning_rate": 0.00036903593322397036,
      "loss": 2.0818,
      "step": 98172
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.8269823789596558,
      "learning_rate": 0.0003690319524366071,
      "loss": 2.3806,
      "step": 98173
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.855302095413208,
      "learning_rate": 0.0003690279716364094,
      "loss": 2.1755,
      "step": 98174
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.8229990005493164,
      "learning_rate": 0.0003690239908233782,
      "loss": 2.1354,
      "step": 98175
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.8918314576148987,
      "learning_rate": 0.0003690200099975141,
      "loss": 2.1553,
      "step": 98176
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.9210292100906372,
      "learning_rate": 0.00036901602915881784,
      "loss": 2.1804,
      "step": 98177
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.9067524671554565,
      "learning_rate": 0.0003690120483072903,
      "loss": 2.0057,
      "step": 98178
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.9005430340766907,
      "learning_rate": 0.00036900806744293203,
      "loss": 2.3296,
      "step": 98179
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.867240309715271,
      "learning_rate": 0.0003690040865657438,
      "loss": 2.3104,
      "step": 98180
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.8222641944885254,
      "learning_rate": 0.0003690001056757266,
      "loss": 2.162,
      "step": 98181
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.7991429567337036,
      "learning_rate": 0.00036899612477288085,
      "loss": 2.2083,
      "step": 98182
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.8395220637321472,
      "learning_rate": 0.0003689921438572074,
      "loss": 2.1941,
      "step": 98183
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.8224610090255737,
      "learning_rate": 0.0003689881629287071,
      "loss": 2.1563,
      "step": 98184
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.85649174451828,
      "learning_rate": 0.0003689841819873805,
      "loss": 2.4051,
      "step": 98185
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.7928721308708191,
      "learning_rate": 0.0003689802010332284,
      "loss": 2.043,
      "step": 98186
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.8752744197845459,
      "learning_rate": 0.0003689762200662517,
      "loss": 2.1409,
      "step": 98187
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.9467682838439941,
      "learning_rate": 0.000368972239086451,
      "loss": 2.3475,
      "step": 98188
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.9960838556289673,
      "learning_rate": 0.00036896825809382697,
      "loss": 2.3134,
      "step": 98189
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.8229803442955017,
      "learning_rate": 0.00036896427708838056,
      "loss": 2.0194,
      "step": 98190
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.9145663976669312,
      "learning_rate": 0.00036896029607011227,
      "loss": 2.0614,
      "step": 98191
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.9652326107025146,
      "learning_rate": 0.00036895631503902306,
      "loss": 2.2763,
      "step": 98192
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.8581153154373169,
      "learning_rate": 0.00036895233399511355,
      "loss": 2.1919,
      "step": 98193
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.8625971078872681,
      "learning_rate": 0.0003689483529383845,
      "loss": 2.3417,
      "step": 98194
    },
    {
      "epoch": 1.28,
      "grad_norm": 1.0043474435806274,
      "learning_rate": 0.0003689443718688367,
      "loss": 2.1405,
      "step": 98195
    },
    {
      "epoch": 1.28,
      "grad_norm": 1.1033499240875244,
      "learning_rate": 0.0003689403907864707,
      "loss": 2.1652,
      "step": 98196
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.8583364486694336,
      "learning_rate": 0.0003689364096912874,
      "loss": 2.02,
      "step": 98197
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.9377740621566772,
      "learning_rate": 0.00036893242858328755,
      "loss": 2.1059,
      "step": 98198
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.9932096004486084,
      "learning_rate": 0.0003689284474624719,
      "loss": 1.8344,
      "step": 98199
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.8397831320762634,
      "learning_rate": 0.0003689244663288411,
      "loss": 2.0286,
      "step": 98200
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.9140469431877136,
      "learning_rate": 0.00036892048518239595,
      "loss": 2.1812,
      "step": 98201
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.9212721586227417,
      "learning_rate": 0.00036891650402313725,
      "loss": 2.0093,
      "step": 98202
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.9298825263977051,
      "learning_rate": 0.0003689125228510656,
      "loss": 2.1622,
      "step": 98203
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.8885992169380188,
      "learning_rate": 0.0003689085416661819,
      "loss": 2.1694,
      "step": 98204
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.9906848073005676,
      "learning_rate": 0.0003689045604684867,
      "loss": 2.2034,
      "step": 98205
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.8480827212333679,
      "learning_rate": 0.0003689005792579809,
      "loss": 2.253,
      "step": 98206
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.9053277969360352,
      "learning_rate": 0.00036889659803466515,
      "loss": 2.3421,
      "step": 98207
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.9049986600875854,
      "learning_rate": 0.00036889261679854025,
      "loss": 2.1027,
      "step": 98208
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.8677626848220825,
      "learning_rate": 0.0003688886355496069,
      "loss": 2.2743,
      "step": 98209
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.8223641514778137,
      "learning_rate": 0.00036888465428786584,
      "loss": 2.4554,
      "step": 98210
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.9221435785293579,
      "learning_rate": 0.0003688806730133178,
      "loss": 2.2587,
      "step": 98211
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.8733350038528442,
      "learning_rate": 0.0003688766917259636,
      "loss": 2.2403,
      "step": 98212
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.8494841456413269,
      "learning_rate": 0.00036887271042580387,
      "loss": 2.4507,
      "step": 98213
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.8335246443748474,
      "learning_rate": 0.0003688687291128394,
      "loss": 1.9516,
      "step": 98214
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.8352561593055725,
      "learning_rate": 0.000368864747787071,
      "loss": 2.3285,
      "step": 98215
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.8589246869087219,
      "learning_rate": 0.00036886076644849933,
      "loss": 1.9415,
      "step": 98216
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.8183338046073914,
      "learning_rate": 0.00036885678509712504,
      "loss": 2.3302,
      "step": 98217
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.9726741909980774,
      "learning_rate": 0.00036885280373294914,
      "loss": 2.2101,
      "step": 98218
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.9011144638061523,
      "learning_rate": 0.00036884882235597207,
      "loss": 2.0713,
      "step": 98219
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.835781991481781,
      "learning_rate": 0.00036884484096619475,
      "loss": 2.31,
      "step": 98220
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.817866861820221,
      "learning_rate": 0.00036884085956361793,
      "loss": 2.0773,
      "step": 98221
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.8595983982086182,
      "learning_rate": 0.0003688368781482422,
      "loss": 2.0935,
      "step": 98222
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.8596215844154358,
      "learning_rate": 0.0003688328967200684,
      "loss": 2.1497,
      "step": 98223
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.9342564940452576,
      "learning_rate": 0.00036882891527909733,
      "loss": 2.1648,
      "step": 98224
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.803412139415741,
      "learning_rate": 0.0003688249338253296,
      "loss": 2.168,
      "step": 98225
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.9584909677505493,
      "learning_rate": 0.0003688209523587661,
      "loss": 2.0501,
      "step": 98226
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.8229634165763855,
      "learning_rate": 0.0003688169708794074,
      "loss": 2.1205,
      "step": 98227
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.880410373210907,
      "learning_rate": 0.0003688129893872544,
      "loss": 2.1518,
      "step": 98228
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.9247163534164429,
      "learning_rate": 0.0003688090078823077,
      "loss": 2.2877,
      "step": 98229
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.9211753606796265,
      "learning_rate": 0.0003688050263645681,
      "loss": 2.0412,
      "step": 98230
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.8731140494346619,
      "learning_rate": 0.0003688010448340364,
      "loss": 2.1232,
      "step": 98231
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.7986003160476685,
      "learning_rate": 0.0003687970632907133,
      "loss": 2.1241,
      "step": 98232
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.8487497568130493,
      "learning_rate": 0.0003687930817345996,
      "loss": 2.0791,
      "step": 98233
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.8477683663368225,
      "learning_rate": 0.0003687891001656958,
      "loss": 2.1448,
      "step": 98234
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.8686730265617371,
      "learning_rate": 0.00036878511858400286,
      "loss": 2.0877,
      "step": 98235
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.8749473690986633,
      "learning_rate": 0.00036878113698952144,
      "loss": 2.2849,
      "step": 98236
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.9127641916275024,
      "learning_rate": 0.00036877715538225235,
      "loss": 2.1112,
      "step": 98237
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.8871804475784302,
      "learning_rate": 0.0003687731737621963,
      "loss": 2.1397,
      "step": 98238
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.8394006490707397,
      "learning_rate": 0.000368769192129354,
      "loss": 2.2111,
      "step": 98239
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.9624348878860474,
      "learning_rate": 0.0003687652104837262,
      "loss": 2.0143,
      "step": 98240
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.8372267484664917,
      "learning_rate": 0.0003687612288253137,
      "loss": 2.1381,
      "step": 98241
    },
    {
      "epoch": 1.28,
      "grad_norm": 1.0197393894195557,
      "learning_rate": 0.00036875724715411715,
      "loss": 2.0918,
      "step": 98242
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.8250766396522522,
      "learning_rate": 0.00036875326547013734,
      "loss": 1.8903,
      "step": 98243
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.8708399534225464,
      "learning_rate": 0.000368749283773375,
      "loss": 2.2842,
      "step": 98244
    },
    {
      "epoch": 1.28,
      "grad_norm": 1.0343694686889648,
      "learning_rate": 0.0003687453020638309,
      "loss": 2.0984,
      "step": 98245
    },
    {
      "epoch": 1.28,
      "grad_norm": 1.013197660446167,
      "learning_rate": 0.00036874132034150567,
      "loss": 2.0573,
      "step": 98246
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.8273101449012756,
      "learning_rate": 0.00036873733860640014,
      "loss": 2.0056,
      "step": 98247
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.8716663122177124,
      "learning_rate": 0.00036873335685851517,
      "loss": 1.9479,
      "step": 98248
    },
    {
      "epoch": 1.28,
      "grad_norm": 1.0521281957626343,
      "learning_rate": 0.00036872937509785126,
      "loss": 2.3489,
      "step": 98249
    },
    {
      "epoch": 1.28,
      "grad_norm": 1.0507080554962158,
      "learning_rate": 0.0003687253933244092,
      "loss": 2.2024,
      "step": 98250
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.9711031317710876,
      "learning_rate": 0.00036872141153818994,
      "loss": 2.1853,
      "step": 98251
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.9895671606063843,
      "learning_rate": 0.000368717429739194,
      "loss": 2.4901,
      "step": 98252
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.9120597243309021,
      "learning_rate": 0.00036871344792742216,
      "loss": 2.2185,
      "step": 98253
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.8775768280029297,
      "learning_rate": 0.0003687094661028752,
      "loss": 2.3975,
      "step": 98254
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.8528416156768799,
      "learning_rate": 0.00036870548426555397,
      "loss": 2.2313,
      "step": 98255
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.8098129630088806,
      "learning_rate": 0.000368701502415459,
      "loss": 2.1301,
      "step": 98256
    },
    {
      "epoch": 1.28,
      "grad_norm": 1.0962061882019043,
      "learning_rate": 0.0003686975205525911,
      "loss": 2.0809,
      "step": 98257
    },
    {
      "epoch": 1.28,
      "grad_norm": 1.1220858097076416,
      "learning_rate": 0.00036869353867695104,
      "loss": 2.2267,
      "step": 98258
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.9305862188339233,
      "learning_rate": 0.0003686895567885396,
      "loss": 2.3863,
      "step": 98259
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.9017544984817505,
      "learning_rate": 0.0003686855748873574,
      "loss": 2.31,
      "step": 98260
    },
    {
      "epoch": 1.28,
      "grad_norm": 1.1122584342956543,
      "learning_rate": 0.0003686815929734054,
      "loss": 2.3353,
      "step": 98261
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.9376477003097534,
      "learning_rate": 0.00036867761104668404,
      "loss": 2.2459,
      "step": 98262
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.8335005044937134,
      "learning_rate": 0.0003686736291071943,
      "loss": 2.4206,
      "step": 98263
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.9693224430084229,
      "learning_rate": 0.0003686696471549368,
      "loss": 2.1818,
      "step": 98264
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.8661639094352722,
      "learning_rate": 0.0003686656651899124,
      "loss": 2.0561,
      "step": 98265
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.8523560166358948,
      "learning_rate": 0.00036866168321212164,
      "loss": 2.428,
      "step": 98266
    },
    {
      "epoch": 1.28,
      "grad_norm": 1.1156634092330933,
      "learning_rate": 0.00036865770122156545,
      "loss": 2.1462,
      "step": 98267
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.8945983052253723,
      "learning_rate": 0.00036865371921824446,
      "loss": 1.9674,
      "step": 98268
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.9089639186859131,
      "learning_rate": 0.0003686497372021594,
      "loss": 2.1062,
      "step": 98269
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.8769204020500183,
      "learning_rate": 0.00036864575517331116,
      "loss": 2.1501,
      "step": 98270
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.8605599403381348,
      "learning_rate": 0.0003686417731317003,
      "loss": 2.1539,
      "step": 98271
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.8832492828369141,
      "learning_rate": 0.00036863779107732767,
      "loss": 2.1631,
      "step": 98272
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.9718114137649536,
      "learning_rate": 0.00036863380901019406,
      "loss": 2.2076,
      "step": 98273
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.9354174733161926,
      "learning_rate": 0.00036862982693030004,
      "loss": 2.2067,
      "step": 98274
    },
    {
      "epoch": 1.28,
      "grad_norm": 1.04843008518219,
      "learning_rate": 0.00036862584483764643,
      "loss": 2.1937,
      "step": 98275
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.8177289366722107,
      "learning_rate": 0.0003686218627322341,
      "loss": 2.2847,
      "step": 98276
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.8522377610206604,
      "learning_rate": 0.0003686178806140635,
      "loss": 2.1102,
      "step": 98277
    },
    {
      "epoch": 1.28,
      "grad_norm": 1.0202898979187012,
      "learning_rate": 0.0003686138984831356,
      "loss": 2.2505,
      "step": 98278
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.8072625398635864,
      "learning_rate": 0.00036860991633945104,
      "loss": 2.0912,
      "step": 98279
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.8796627521514893,
      "learning_rate": 0.0003686059341830107,
      "loss": 2.0231,
      "step": 98280
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.8038963079452515,
      "learning_rate": 0.0003686019520138152,
      "loss": 2.1714,
      "step": 98281
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.9400935769081116,
      "learning_rate": 0.0003685979698318653,
      "loss": 2.041,
      "step": 98282
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.9153552651405334,
      "learning_rate": 0.0003685939876371617,
      "loss": 2.28,
      "step": 98283
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.894563615322113,
      "learning_rate": 0.0003685900054297052,
      "loss": 2.2416,
      "step": 98284
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.9441385865211487,
      "learning_rate": 0.0003685860232094965,
      "loss": 2.1936,
      "step": 98285
    },
    {
      "epoch": 1.28,
      "grad_norm": 1.1249147653579712,
      "learning_rate": 0.0003685820409765364,
      "loss": 2.1855,
      "step": 98286
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.8206512331962585,
      "learning_rate": 0.00036857805873082564,
      "loss": 2.1346,
      "step": 98287
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.8312304019927979,
      "learning_rate": 0.0003685740764723649,
      "loss": 2.1512,
      "step": 98288
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.8908900618553162,
      "learning_rate": 0.0003685700942011549,
      "loss": 2.1096,
      "step": 98289
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.8715852499008179,
      "learning_rate": 0.00036856611191719645,
      "loss": 2.0483,
      "step": 98290
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.8203057646751404,
      "learning_rate": 0.0003685621296204903,
      "loss": 2.1624,
      "step": 98291
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.8865100741386414,
      "learning_rate": 0.0003685581473110371,
      "loss": 2.211,
      "step": 98292
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.8203650712966919,
      "learning_rate": 0.00036855416498883764,
      "loss": 2.0285,
      "step": 98293
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.8275206089019775,
      "learning_rate": 0.00036855018265389276,
      "loss": 2.3272,
      "step": 98294
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.9027823805809021,
      "learning_rate": 0.000368546200306203,
      "loss": 1.9801,
      "step": 98295
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.857675313949585,
      "learning_rate": 0.0003685422179457692,
      "loss": 2.1265,
      "step": 98296
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.8466322422027588,
      "learning_rate": 0.0003685382355725922,
      "loss": 2.161,
      "step": 98297
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.871100902557373,
      "learning_rate": 0.0003685342531866726,
      "loss": 2.0184,
      "step": 98298
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.963572084903717,
      "learning_rate": 0.00036853027078801115,
      "loss": 2.1056,
      "step": 98299
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.7835228443145752,
      "learning_rate": 0.00036852628837660877,
      "loss": 2.2619,
      "step": 98300
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.8854021430015564,
      "learning_rate": 0.0003685223059524659,
      "loss": 2.1626,
      "step": 98301
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.865844190120697,
      "learning_rate": 0.00036851832351558354,
      "loss": 2.3261,
      "step": 98302
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.9126316905021667,
      "learning_rate": 0.0003685143410659623,
      "loss": 2.308,
      "step": 98303
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.983397901058197,
      "learning_rate": 0.000368510358603603,
      "loss": 2.1123,
      "step": 98304
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.9309066534042358,
      "learning_rate": 0.0003685063761285062,
      "loss": 2.1828,
      "step": 98305
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.9499678015708923,
      "learning_rate": 0.0003685023936406729,
      "loss": 2.1858,
      "step": 98306
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.8312178254127502,
      "learning_rate": 0.00036849841114010366,
      "loss": 2.1238,
      "step": 98307
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.9028398394584656,
      "learning_rate": 0.00036849442862679923,
      "loss": 2.2723,
      "step": 98308
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.8616141080856323,
      "learning_rate": 0.00036849044610076046,
      "loss": 2.1532,
      "step": 98309
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.9845004081726074,
      "learning_rate": 0.000368486463561988,
      "loss": 2.0131,
      "step": 98310
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.8879608511924744,
      "learning_rate": 0.00036848248101048253,
      "loss": 2.1139,
      "step": 98311
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.822425127029419,
      "learning_rate": 0.00036847849844624504,
      "loss": 2.0108,
      "step": 98312
    },
    {
      "epoch": 1.28,
      "grad_norm": 1.0513931512832642,
      "learning_rate": 0.000368474515869276,
      "loss": 2.0932,
      "step": 98313
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.8167040348052979,
      "learning_rate": 0.00036847053327957623,
      "loss": 2.2511,
      "step": 98314
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.8927530646324158,
      "learning_rate": 0.0003684665506771466,
      "loss": 2.0224,
      "step": 98315
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.9478036761283875,
      "learning_rate": 0.0003684625680619877,
      "loss": 2.2166,
      "step": 98316
    },
    {
      "epoch": 1.28,
      "grad_norm": 1.1190505027770996,
      "learning_rate": 0.0003684585854341003,
      "loss": 2.2241,
      "step": 98317
    },
    {
      "epoch": 1.28,
      "grad_norm": 1.0074539184570312,
      "learning_rate": 0.00036845460279348515,
      "loss": 2.1835,
      "step": 98318
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.8867776393890381,
      "learning_rate": 0.000368450620140143,
      "loss": 2.2489,
      "step": 98319
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.9018731713294983,
      "learning_rate": 0.0003684466374740746,
      "loss": 1.8669,
      "step": 98320
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.8200444579124451,
      "learning_rate": 0.0003684426547952806,
      "loss": 2.4792,
      "step": 98321
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.920643150806427,
      "learning_rate": 0.000368438672103762,
      "loss": 2.175,
      "step": 98322
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.8854502439498901,
      "learning_rate": 0.0003684346893995192,
      "loss": 2.4259,
      "step": 98323
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.893276035785675,
      "learning_rate": 0.00036843070668255313,
      "loss": 2.1572,
      "step": 98324
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.8411206007003784,
      "learning_rate": 0.00036842672395286455,
      "loss": 2.0238,
      "step": 98325
    },
    {
      "epoch": 1.28,
      "grad_norm": 1.01485013961792,
      "learning_rate": 0.0003684227412104541,
      "loss": 2.0136,
      "step": 98326
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.862566351890564,
      "learning_rate": 0.00036841875845532255,
      "loss": 2.1617,
      "step": 98327
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.8958032131195068,
      "learning_rate": 0.00036841477568747076,
      "loss": 2.1769,
      "step": 98328
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.8229080438613892,
      "learning_rate": 0.0003684107929068993,
      "loss": 2.041,
      "step": 98329
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.9775904417037964,
      "learning_rate": 0.000368406810113609,
      "loss": 2.1899,
      "step": 98330
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.8843317627906799,
      "learning_rate": 0.0003684028273076006,
      "loss": 2.2342,
      "step": 98331
    },
    {
      "epoch": 1.28,
      "grad_norm": 1.112632155418396,
      "learning_rate": 0.0003683988444888748,
      "loss": 2.2777,
      "step": 98332
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.9658653736114502,
      "learning_rate": 0.00036839486165743233,
      "loss": 2.1603,
      "step": 98333
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.8425899147987366,
      "learning_rate": 0.00036839087881327407,
      "loss": 2.2278,
      "step": 98334
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.9271824359893799,
      "learning_rate": 0.00036838689595640055,
      "loss": 2.0617,
      "step": 98335
    },
    {
      "epoch": 1.28,
      "grad_norm": 1.0534950494766235,
      "learning_rate": 0.0003683829130868126,
      "loss": 2.2767,
      "step": 98336
    },
    {
      "epoch": 1.28,
      "grad_norm": 1.101586103439331,
      "learning_rate": 0.0003683789302045111,
      "loss": 2.128,
      "step": 98337
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.9450793266296387,
      "learning_rate": 0.00036837494730949656,
      "loss": 2.1073,
      "step": 98338
    },
    {
      "epoch": 1.28,
      "grad_norm": 1.0700658559799194,
      "learning_rate": 0.0003683709644017698,
      "loss": 2.1428,
      "step": 98339
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.8368713855743408,
      "learning_rate": 0.00036836698148133177,
      "loss": 2.0504,
      "step": 98340
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.8901944756507874,
      "learning_rate": 0.00036836299854818284,
      "loss": 2.0971,
      "step": 98341
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.9930815100669861,
      "learning_rate": 0.00036835901560232397,
      "loss": 2.2499,
      "step": 98342
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.978911280632019,
      "learning_rate": 0.000368355032643756,
      "loss": 2.3718,
      "step": 98343
    },
    {
      "epoch": 1.28,
      "grad_norm": 1.179302453994751,
      "learning_rate": 0.00036835104967247934,
      "loss": 2.0989,
      "step": 98344
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.968559741973877,
      "learning_rate": 0.000368347066688495,
      "loss": 2.1969,
      "step": 98345
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.9115623831748962,
      "learning_rate": 0.00036834308369180374,
      "loss": 2.2694,
      "step": 98346
    },
    {
      "epoch": 1.28,
      "grad_norm": 1.1008597612380981,
      "learning_rate": 0.0003683391006824062,
      "loss": 2.4459,
      "step": 98347
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.9549007415771484,
      "learning_rate": 0.00036833511766030296,
      "loss": 2.0472,
      "step": 98348
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.8864195346832275,
      "learning_rate": 0.0003683311346254951,
      "loss": 2.1511,
      "step": 98349
    },
    {
      "epoch": 1.28,
      "grad_norm": 1.1201550960540771,
      "learning_rate": 0.0003683271515779832,
      "loss": 2.0531,
      "step": 98350
    },
    {
      "epoch": 1.28,
      "grad_norm": 1.1617132425308228,
      "learning_rate": 0.00036832316851776787,
      "loss": 2.2078,
      "step": 98351
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.8392584323883057,
      "learning_rate": 0.0003683191854448501,
      "loss": 2.0562,
      "step": 98352
    },
    {
      "epoch": 1.28,
      "grad_norm": 1.014159917831421,
      "learning_rate": 0.0003683152023592304,
      "loss": 2.0689,
      "step": 98353
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.8579323291778564,
      "learning_rate": 0.0003683112192609097,
      "loss": 2.0381,
      "step": 98354
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.7956748008728027,
      "learning_rate": 0.0003683072361498886,
      "loss": 1.9436,
      "step": 98355
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.8318254947662354,
      "learning_rate": 0.0003683032530261679,
      "loss": 1.9291,
      "step": 98356
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.9921247363090515,
      "learning_rate": 0.0003682992698897483,
      "loss": 2.1077,
      "step": 98357
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.8679496049880981,
      "learning_rate": 0.00036829528674063073,
      "loss": 2.1517,
      "step": 98358
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.8368533253669739,
      "learning_rate": 0.0003682913035788156,
      "loss": 1.9922,
      "step": 98359
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.8885061144828796,
      "learning_rate": 0.0003682873204043039,
      "loss": 2.0699,
      "step": 98360
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.8965079188346863,
      "learning_rate": 0.0003682833372170963,
      "loss": 2.2035,
      "step": 98361
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.905079185962677,
      "learning_rate": 0.00036827935401719355,
      "loss": 2.0776,
      "step": 98362
    },
    {
      "epoch": 1.28,
      "grad_norm": 1.0553958415985107,
      "learning_rate": 0.0003682753708045963,
      "loss": 2.0746,
      "step": 98363
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.9424277544021606,
      "learning_rate": 0.00036827138757930544,
      "loss": 2.1059,
      "step": 98364
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.8785401582717896,
      "learning_rate": 0.00036826740434132163,
      "loss": 2.3117,
      "step": 98365
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.852221667766571,
      "learning_rate": 0.0003682634210906456,
      "loss": 2.0429,
      "step": 98366
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.7877495884895325,
      "learning_rate": 0.00036825943782727813,
      "loss": 1.9742,
      "step": 98367
    },
    {
      "epoch": 1.28,
      "grad_norm": 1.0693615674972534,
      "learning_rate": 0.00036825545455122,
      "loss": 2.3387,
      "step": 98368
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.8756763935089111,
      "learning_rate": 0.00036825147126247186,
      "loss": 2.2316,
      "step": 98369
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.879348874092102,
      "learning_rate": 0.0003682474879610344,
      "loss": 2.1955,
      "step": 98370
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.8972699046134949,
      "learning_rate": 0.00036824350464690844,
      "loss": 2.2271,
      "step": 98371
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.8658642172813416,
      "learning_rate": 0.00036823952132009486,
      "loss": 2.0075,
      "step": 98372
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.8491480350494385,
      "learning_rate": 0.0003682355379805942,
      "loss": 2.1321,
      "step": 98373
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.9606938362121582,
      "learning_rate": 0.0003682315546284073,
      "loss": 2.2271,
      "step": 98374
    },
    {
      "epoch": 1.28,
      "grad_norm": 1.0556392669677734,
      "learning_rate": 0.00036822757126353485,
      "loss": 2.1113,
      "step": 98375
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.7923076152801514,
      "learning_rate": 0.0003682235878859775,
      "loss": 2.2058,
      "step": 98376
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.8203656077384949,
      "learning_rate": 0.00036821960449573616,
      "loss": 2.0085,
      "step": 98377
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.8369422554969788,
      "learning_rate": 0.00036821562109281154,
      "loss": 2.2885,
      "step": 98378
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.876844584941864,
      "learning_rate": 0.00036821163767720435,
      "loss": 2.1276,
      "step": 98379
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.901948869228363,
      "learning_rate": 0.00036820765424891536,
      "loss": 2.096,
      "step": 98380
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.880024790763855,
      "learning_rate": 0.0003682036708079452,
      "loss": 2.3544,
      "step": 98381
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.8421719074249268,
      "learning_rate": 0.00036819968735429473,
      "loss": 2.2692,
      "step": 98382
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.8244146108627319,
      "learning_rate": 0.0003681957038879646,
      "loss": 2.1816,
      "step": 98383
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.8778316378593445,
      "learning_rate": 0.00036819172040895566,
      "loss": 2.1232,
      "step": 98384
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.8401651978492737,
      "learning_rate": 0.00036818773691726865,
      "loss": 1.9962,
      "step": 98385
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.8358206748962402,
      "learning_rate": 0.0003681837534129041,
      "loss": 2.0033,
      "step": 98386
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.9711260199546814,
      "learning_rate": 0.000368179769895863,
      "loss": 2.3067,
      "step": 98387
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.8935549259185791,
      "learning_rate": 0.00036817578636614596,
      "loss": 2.0198,
      "step": 98388
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.8277960419654846,
      "learning_rate": 0.0003681718028237538,
      "loss": 2.1466,
      "step": 98389
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.8158117532730103,
      "learning_rate": 0.0003681678192686872,
      "loss": 2.298,
      "step": 98390
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.860393226146698,
      "learning_rate": 0.00036816383570094683,
      "loss": 2.0025,
      "step": 98391
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.8709431886672974,
      "learning_rate": 0.00036815985212053365,
      "loss": 2.5665,
      "step": 98392
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.8230881094932556,
      "learning_rate": 0.0003681558685274482,
      "loss": 1.9424,
      "step": 98393
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.8312448859214783,
      "learning_rate": 0.00036815188492169126,
      "loss": 2.1258,
      "step": 98394
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.8381666541099548,
      "learning_rate": 0.0003681479013032637,
      "loss": 2.1334,
      "step": 98395
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.9597269296646118,
      "learning_rate": 0.00036814391767216604,
      "loss": 2.1864,
      "step": 98396
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.8988912105560303,
      "learning_rate": 0.0003681399340283992,
      "loss": 2.0753,
      "step": 98397
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.8965219259262085,
      "learning_rate": 0.00036813595037196386,
      "loss": 2.2942,
      "step": 98398
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.8681580424308777,
      "learning_rate": 0.0003681319667028608,
      "loss": 2.1342,
      "step": 98399
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.9446913599967957,
      "learning_rate": 0.00036812798302109063,
      "loss": 2.3976,
      "step": 98400
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.9068864583969116,
      "learning_rate": 0.00036812399932665424,
      "loss": 2.1659,
      "step": 98401
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.9117903113365173,
      "learning_rate": 0.00036812001561955225,
      "loss": 2.326,
      "step": 98402
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.8455125093460083,
      "learning_rate": 0.00036811603189978553,
      "loss": 2.161,
      "step": 98403
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.8795740008354187,
      "learning_rate": 0.0003681120481673548,
      "loss": 2.2971,
      "step": 98404
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.9524585008621216,
      "learning_rate": 0.0003681080644222606,
      "loss": 2.3413,
      "step": 98405
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.9561650156974792,
      "learning_rate": 0.00036810408066450395,
      "loss": 2.0784,
      "step": 98406
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.7871922254562378,
      "learning_rate": 0.0003681000968940855,
      "loss": 2.0304,
      "step": 98407
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.7522339820861816,
      "learning_rate": 0.0003680961131110058,
      "loss": 2.0191,
      "step": 98408
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.8014207482337952,
      "learning_rate": 0.00036809212931526584,
      "loss": 2.1649,
      "step": 98409
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.8873946666717529,
      "learning_rate": 0.0003680881455068663,
      "loss": 2.0527,
      "step": 98410
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.9355093836784363,
      "learning_rate": 0.00036808416168580783,
      "loss": 2.3874,
      "step": 98411
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.9268833994865417,
      "learning_rate": 0.00036808017785209125,
      "loss": 2.3122,
      "step": 98412
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.9466809630393982,
      "learning_rate": 0.00036807619400571726,
      "loss": 2.1452,
      "step": 98413
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.910147488117218,
      "learning_rate": 0.00036807221014668667,
      "loss": 2.1444,
      "step": 98414
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.9066002368927002,
      "learning_rate": 0.00036806822627500013,
      "loss": 2.016,
      "step": 98415
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.9410775899887085,
      "learning_rate": 0.0003680642423906584,
      "loss": 2.2367,
      "step": 98416
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.7956919074058533,
      "learning_rate": 0.00036806025849366226,
      "loss": 2.0389,
      "step": 98417
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.8899592161178589,
      "learning_rate": 0.0003680562745840125,
      "loss": 2.2276,
      "step": 98418
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.8418521881103516,
      "learning_rate": 0.00036805229066170977,
      "loss": 2.1937,
      "step": 98419
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.852114200592041,
      "learning_rate": 0.0003680483067267548,
      "loss": 2.0225,
      "step": 98420
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.850943922996521,
      "learning_rate": 0.0003680443227791483,
      "loss": 2.255,
      "step": 98421
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.8382213115692139,
      "learning_rate": 0.0003680403388188912,
      "loss": 2.0742,
      "step": 98422
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.8540952801704407,
      "learning_rate": 0.00036803635484598406,
      "loss": 2.2094,
      "step": 98423
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.9502615928649902,
      "learning_rate": 0.00036803237086042767,
      "loss": 2.3568,
      "step": 98424
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.8017470836639404,
      "learning_rate": 0.0003680283868622228,
      "loss": 2.1226,
      "step": 98425
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.8224010467529297,
      "learning_rate": 0.00036802440285137015,
      "loss": 2.1804,
      "step": 98426
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.8660781383514404,
      "learning_rate": 0.00036802041882787043,
      "loss": 1.981,
      "step": 98427
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.9006985425949097,
      "learning_rate": 0.0003680164347917246,
      "loss": 2.0654,
      "step": 98428
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.9497690200805664,
      "learning_rate": 0.0003680124507429331,
      "loss": 2.5067,
      "step": 98429
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.9293608665466309,
      "learning_rate": 0.0003680084666814968,
      "loss": 2.1624,
      "step": 98430
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.9140436053276062,
      "learning_rate": 0.00036800448260741644,
      "loss": 2.1686,
      "step": 98431
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.9006564021110535,
      "learning_rate": 0.0003680004985206928,
      "loss": 2.2675,
      "step": 98432
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.9673943519592285,
      "learning_rate": 0.0003679965144213265,
      "loss": 2.1115,
      "step": 98433
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.8428177833557129,
      "learning_rate": 0.0003679925303093185,
      "loss": 2.2472,
      "step": 98434
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.75620436668396,
      "learning_rate": 0.00036798854618466935,
      "loss": 2.14,
      "step": 98435
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.8643755316734314,
      "learning_rate": 0.0003679845620473798,
      "loss": 2.3698,
      "step": 98436
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.9477918148040771,
      "learning_rate": 0.0003679805778974507,
      "loss": 1.9962,
      "step": 98437
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.9441950917243958,
      "learning_rate": 0.0003679765937348827,
      "loss": 2.2445,
      "step": 98438
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.8396373391151428,
      "learning_rate": 0.00036797260955967654,
      "loss": 2.0869,
      "step": 98439
    },
    {
      "epoch": 1.28,
      "grad_norm": 1.0383356809616089,
      "learning_rate": 0.000367968625371833,
      "loss": 2.0397,
      "step": 98440
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.9678059220314026,
      "learning_rate": 0.00036796464117135286,
      "loss": 2.1599,
      "step": 98441
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.9880562424659729,
      "learning_rate": 0.00036796065695823675,
      "loss": 2.1755,
      "step": 98442
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.9514987468719482,
      "learning_rate": 0.00036795667273248556,
      "loss": 2.175,
      "step": 98443
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.8303797245025635,
      "learning_rate": 0.0003679526884940999,
      "loss": 2.1899,
      "step": 98444
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.8510274291038513,
      "learning_rate": 0.00036794870424308046,
      "loss": 2.1364,
      "step": 98445
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.9198312163352966,
      "learning_rate": 0.00036794471997942817,
      "loss": 2.1902,
      "step": 98446
    },
    {
      "epoch": 1.28,
      "grad_norm": 1.1119269132614136,
      "learning_rate": 0.0003679407357031437,
      "loss": 2.0266,
      "step": 98447
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.8368023037910461,
      "learning_rate": 0.00036793675141422763,
      "loss": 2.0649,
      "step": 98448
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.8666858673095703,
      "learning_rate": 0.000367932767112681,
      "loss": 2.299,
      "step": 98449
    },
    {
      "epoch": 1.28,
      "grad_norm": 1.080731749534607,
      "learning_rate": 0.0003679287827985043,
      "loss": 2.3774,
      "step": 98450
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.8406919836997986,
      "learning_rate": 0.0003679247984716983,
      "loss": 2.0021,
      "step": 98451
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.8422974348068237,
      "learning_rate": 0.0003679208141322638,
      "loss": 2.2427,
      "step": 98452
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.9786880612373352,
      "learning_rate": 0.00036791682978020167,
      "loss": 2.1893,
      "step": 98453
    },
    {
      "epoch": 1.28,
      "grad_norm": 1.0418365001678467,
      "learning_rate": 0.0003679128454155124,
      "loss": 2.2162,
      "step": 98454
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.906606912612915,
      "learning_rate": 0.00036790886103819687,
      "loss": 2.2178,
      "step": 98455
    },
    {
      "epoch": 1.28,
      "grad_norm": 1.1328771114349365,
      "learning_rate": 0.00036790487664825584,
      "loss": 2.2807,
      "step": 98456
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.865496814250946,
      "learning_rate": 0.0003679008922456901,
      "loss": 2.0787,
      "step": 98457
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.8848302960395813,
      "learning_rate": 0.00036789690783050016,
      "loss": 2.239,
      "step": 98458
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.932148277759552,
      "learning_rate": 0.00036789292340268687,
      "loss": 2.3044,
      "step": 98459
    },
    {
      "epoch": 1.28,
      "grad_norm": 1.0382506847381592,
      "learning_rate": 0.00036788893896225113,
      "loss": 2.2522,
      "step": 98460
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.9307076334953308,
      "learning_rate": 0.00036788495450919354,
      "loss": 2.3057,
      "step": 98461
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.8512136340141296,
      "learning_rate": 0.0003678809700435147,
      "loss": 2.2692,
      "step": 98462
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.8502410650253296,
      "learning_rate": 0.00036787698556521565,
      "loss": 2.1105,
      "step": 98463
    },
    {
      "epoch": 1.28,
      "grad_norm": 1.1742788553237915,
      "learning_rate": 0.000367873001074297,
      "loss": 2.5318,
      "step": 98464
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.8985942006111145,
      "learning_rate": 0.00036786901657075946,
      "loss": 1.9695,
      "step": 98465
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.7766064405441284,
      "learning_rate": 0.00036786503205460375,
      "loss": 2.0586,
      "step": 98466
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.8727867007255554,
      "learning_rate": 0.0003678610475258307,
      "loss": 2.2385,
      "step": 98467
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.8434113264083862,
      "learning_rate": 0.00036785706298444093,
      "loss": 2.3473,
      "step": 98468
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.8518121242523193,
      "learning_rate": 0.00036785307843043523,
      "loss": 2.0529,
      "step": 98469
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.9481995105743408,
      "learning_rate": 0.00036784909386381446,
      "loss": 2.2403,
      "step": 98470
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.9787684679031372,
      "learning_rate": 0.00036784510928457927,
      "loss": 2.3132,
      "step": 98471
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.9314976334571838,
      "learning_rate": 0.0003678411246927303,
      "loss": 2.0538,
      "step": 98472
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.807830274105072,
      "learning_rate": 0.00036783714008826844,
      "loss": 1.9074,
      "step": 98473
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.8508031368255615,
      "learning_rate": 0.00036783315547119443,
      "loss": 2.2814,
      "step": 98474
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.8197047710418701,
      "learning_rate": 0.0003678291708415088,
      "loss": 1.9773,
      "step": 98475
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.9073926210403442,
      "learning_rate": 0.0003678251861992126,
      "loss": 1.9561,
      "step": 98476
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.8417775630950928,
      "learning_rate": 0.00036782120154430636,
      "loss": 2.2835,
      "step": 98477
    },
    {
      "epoch": 1.28,
      "grad_norm": 1.0127508640289307,
      "learning_rate": 0.0003678172168767909,
      "loss": 2.2709,
      "step": 98478
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.8583118319511414,
      "learning_rate": 0.0003678132321966669,
      "loss": 2.1996,
      "step": 98479
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.8836914300918579,
      "learning_rate": 0.0003678092475039352,
      "loss": 2.2593,
      "step": 98480
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.9353424906730652,
      "learning_rate": 0.0003678052627985963,
      "loss": 2.3175,
      "step": 98481
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.8999482989311218,
      "learning_rate": 0.00036780127808065134,
      "loss": 2.0321,
      "step": 98482
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.8810166716575623,
      "learning_rate": 0.0003677972933501008,
      "loss": 2.1626,
      "step": 98483
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.8423642516136169,
      "learning_rate": 0.0003677933086069454,
      "loss": 2.2955,
      "step": 98484
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.9973910450935364,
      "learning_rate": 0.000367789323851186,
      "loss": 2.0649,
      "step": 98485
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.7912921905517578,
      "learning_rate": 0.0003677853390828232,
      "loss": 2.0144,
      "step": 98486
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.8453190326690674,
      "learning_rate": 0.0003677813543018579,
      "loss": 2.1833,
      "step": 98487
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.9066462516784668,
      "learning_rate": 0.00036777736950829073,
      "loss": 2.2357,
      "step": 98488
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.846859872341156,
      "learning_rate": 0.0003677733847021225,
      "loss": 2.2696,
      "step": 98489
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.8485261797904968,
      "learning_rate": 0.0003677693998833539,
      "loss": 2.1088,
      "step": 98490
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.7863194346427917,
      "learning_rate": 0.00036776541505198566,
      "loss": 2.2168,
      "step": 98491
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.8365481495857239,
      "learning_rate": 0.00036776143020801866,
      "loss": 1.8759,
      "step": 98492
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.8701496124267578,
      "learning_rate": 0.0003677574453514534,
      "loss": 2.182,
      "step": 98493
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.9338286519050598,
      "learning_rate": 0.00036775346048229077,
      "loss": 2.2991,
      "step": 98494
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.8101832270622253,
      "learning_rate": 0.00036774947560053157,
      "loss": 1.9694,
      "step": 98495
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.8906552791595459,
      "learning_rate": 0.00036774549070617643,
      "loss": 2.1264,
      "step": 98496
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.8631389737129211,
      "learning_rate": 0.00036774150579922606,
      "loss": 2.3838,
      "step": 98497
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.7961487770080566,
      "learning_rate": 0.0003677375208796814,
      "loss": 2.0453,
      "step": 98498
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.8560906648635864,
      "learning_rate": 0.00036773353594754294,
      "loss": 1.9047,
      "step": 98499
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.9111064672470093,
      "learning_rate": 0.00036772955100281154,
      "loss": 2.0915,
      "step": 98500
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.8375548124313354,
      "learning_rate": 0.000367725566045488,
      "loss": 2.2566,
      "step": 98501
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.9180167317390442,
      "learning_rate": 0.00036772158107557297,
      "loss": 2.2459,
      "step": 98502
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.8201436400413513,
      "learning_rate": 0.0003677175960930672,
      "loss": 2.16,
      "step": 98503
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.8825517296791077,
      "learning_rate": 0.0003677136110979715,
      "loss": 2.1501,
      "step": 98504
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.8280651569366455,
      "learning_rate": 0.0003677096260902866,
      "loss": 2.0767,
      "step": 98505
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.8421335816383362,
      "learning_rate": 0.00036770564107001306,
      "loss": 2.1935,
      "step": 98506
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.9750060439109802,
      "learning_rate": 0.0003677016560371519,
      "loss": 2.5436,
      "step": 98507
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.8380107879638672,
      "learning_rate": 0.0003676976709917036,
      "loss": 2.347,
      "step": 98508
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.9481422305107117,
      "learning_rate": 0.0003676936859336691,
      "loss": 2.2107,
      "step": 98509
    },
    {
      "epoch": 1.28,
      "grad_norm": 1.0611639022827148,
      "learning_rate": 0.0003676897008630491,
      "loss": 2.116,
      "step": 98510
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.8623849749565125,
      "learning_rate": 0.0003676857157798442,
      "loss": 2.3671,
      "step": 98511
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.9345235228538513,
      "learning_rate": 0.0003676817306840553,
      "loss": 2.126,
      "step": 98512
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.8706006407737732,
      "learning_rate": 0.0003676777455756831,
      "loss": 2.0738,
      "step": 98513
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.8167320489883423,
      "learning_rate": 0.0003676737604547284,
      "loss": 2.1621,
      "step": 98514
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.8212386965751648,
      "learning_rate": 0.00036766977532119174,
      "loss": 2.0058,
      "step": 98515
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.9567433595657349,
      "learning_rate": 0.0003676657901750741,
      "loss": 2.2827,
      "step": 98516
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.9134544134140015,
      "learning_rate": 0.000367661805016376,
      "loss": 2.2926,
      "step": 98517
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.9431430101394653,
      "learning_rate": 0.00036765781984509835,
      "loss": 2.0852,
      "step": 98518
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.9570369124412537,
      "learning_rate": 0.00036765383466124194,
      "loss": 2.1183,
      "step": 98519
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.8535969853401184,
      "learning_rate": 0.00036764984946480726,
      "loss": 2.1216,
      "step": 98520
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.9246536493301392,
      "learning_rate": 0.00036764586425579523,
      "loss": 2.2966,
      "step": 98521
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.9133745431900024,
      "learning_rate": 0.0003676418790342065,
      "loss": 2.0515,
      "step": 98522
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.9820749163627625,
      "learning_rate": 0.000367637893800042,
      "loss": 2.0457,
      "step": 98523
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.853702962398529,
      "learning_rate": 0.0003676339085533023,
      "loss": 2.042,
      "step": 98524
    },
    {
      "epoch": 1.28,
      "grad_norm": 1.080669641494751,
      "learning_rate": 0.0003676299232939881,
      "loss": 1.8745,
      "step": 98525
    },
    {
      "epoch": 1.28,
      "grad_norm": 1.4593819379806519,
      "learning_rate": 0.00036762593802210027,
      "loss": 2.3421,
      "step": 98526
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.8985602259635925,
      "learning_rate": 0.0003676219527376395,
      "loss": 2.1301,
      "step": 98527
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.8763489723205566,
      "learning_rate": 0.0003676179674406065,
      "loss": 2.0251,
      "step": 98528
    },
    {
      "epoch": 1.28,
      "grad_norm": 1.0554169416427612,
      "learning_rate": 0.0003676139821310021,
      "loss": 2.0166,
      "step": 98529
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.8446998000144958,
      "learning_rate": 0.00036760999680882697,
      "loss": 2.1558,
      "step": 98530
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.904334545135498,
      "learning_rate": 0.00036760601147408186,
      "loss": 2.1583,
      "step": 98531
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.9355929493904114,
      "learning_rate": 0.00036760202612676755,
      "loss": 2.2888,
      "step": 98532
    },
    {
      "epoch": 1.28,
      "grad_norm": 1.0691314935684204,
      "learning_rate": 0.00036759804076688473,
      "loss": 2.1021,
      "step": 98533
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.833687961101532,
      "learning_rate": 0.0003675940553944341,
      "loss": 2.177,
      "step": 98534
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.8061984777450562,
      "learning_rate": 0.0003675900700094166,
      "loss": 2.0891,
      "step": 98535
    },
    {
      "epoch": 1.28,
      "grad_norm": 1.1248700618743896,
      "learning_rate": 0.0003675860846118326,
      "loss": 2.4583,
      "step": 98536
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.8997222185134888,
      "learning_rate": 0.00036758209920168325,
      "loss": 2.3144,
      "step": 98537
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.8361509442329407,
      "learning_rate": 0.00036757811377896907,
      "loss": 2.2177,
      "step": 98538
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.8309251666069031,
      "learning_rate": 0.00036757412834369083,
      "loss": 2.2105,
      "step": 98539
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.8492460250854492,
      "learning_rate": 0.0003675701428958493,
      "loss": 2.4742,
      "step": 98540
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.8359025120735168,
      "learning_rate": 0.00036756615743544527,
      "loss": 1.9285,
      "step": 98541
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.9581711888313293,
      "learning_rate": 0.00036756217196247924,
      "loss": 1.9514,
      "step": 98542
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.9083722233772278,
      "learning_rate": 0.00036755818647695224,
      "loss": 2.231,
      "step": 98543
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.8551942706108093,
      "learning_rate": 0.0003675542009788649,
      "loss": 2.168,
      "step": 98544
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.8951666951179504,
      "learning_rate": 0.000367550215468218,
      "loss": 2.0618,
      "step": 98545
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.9129778742790222,
      "learning_rate": 0.00036754622994501215,
      "loss": 2.2804,
      "step": 98546
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.8691818118095398,
      "learning_rate": 0.00036754224440924816,
      "loss": 2.201,
      "step": 98547
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.9030781984329224,
      "learning_rate": 0.00036753825886092693,
      "loss": 2.157,
      "step": 98548
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.8377328515052795,
      "learning_rate": 0.00036753427330004897,
      "loss": 2.0702,
      "step": 98549
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.9633382558822632,
      "learning_rate": 0.0003675302877266151,
      "loss": 2.2879,
      "step": 98550
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.9192052483558655,
      "learning_rate": 0.00036752630214062607,
      "loss": 1.9264,
      "step": 98551
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.829276442527771,
      "learning_rate": 0.0003675223165420827,
      "loss": 2.0693,
      "step": 98552
    },
    {
      "epoch": 1.28,
      "grad_norm": 1.0134742259979248,
      "learning_rate": 0.0003675183309309856,
      "loss": 2.0748,
      "step": 98553
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.8234869241714478,
      "learning_rate": 0.00036751434530733556,
      "loss": 1.9199,
      "step": 98554
    },
    {
      "epoch": 1.28,
      "grad_norm": 1.0758742094039917,
      "learning_rate": 0.00036751035967113335,
      "loss": 2.3389,
      "step": 98555
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.9898413419723511,
      "learning_rate": 0.0003675063740223797,
      "loss": 2.1976,
      "step": 98556
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.8982981443405151,
      "learning_rate": 0.0003675023883610753,
      "loss": 2.1602,
      "step": 98557
    },
    {
      "epoch": 1.28,
      "grad_norm": 1.1961866617202759,
      "learning_rate": 0.00036749840268722093,
      "loss": 2.2156,
      "step": 98558
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.9169967770576477,
      "learning_rate": 0.00036749441700081745,
      "loss": 2.1711,
      "step": 98559
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.8382245898246765,
      "learning_rate": 0.00036749043130186536,
      "loss": 2.0744,
      "step": 98560
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.7936698198318481,
      "learning_rate": 0.00036748644559036553,
      "loss": 2.2324,
      "step": 98561
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.856046736240387,
      "learning_rate": 0.0003674824598663188,
      "loss": 2.3853,
      "step": 98562
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.841927707195282,
      "learning_rate": 0.0003674784741297257,
      "loss": 2.264,
      "step": 98563
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.8573065996170044,
      "learning_rate": 0.00036747448838058707,
      "loss": 1.9761,
      "step": 98564
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.9022623300552368,
      "learning_rate": 0.00036747050261890375,
      "loss": 2.1852,
      "step": 98565
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.8692317008972168,
      "learning_rate": 0.0003674665168446763,
      "loss": 2.169,
      "step": 98566
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.8435993790626526,
      "learning_rate": 0.0003674625310579056,
      "loss": 2.1641,
      "step": 98567
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.9262094497680664,
      "learning_rate": 0.0003674585452585924,
      "loss": 2.1182,
      "step": 98568
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.8069328665733337,
      "learning_rate": 0.0003674545594467373,
      "loss": 2.3133,
      "step": 98569
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.8793882131576538,
      "learning_rate": 0.0003674505736223411,
      "loss": 2.3097,
      "step": 98570
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.8328864574432373,
      "learning_rate": 0.00036744658778540466,
      "loss": 2.0009,
      "step": 98571
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.8361687660217285,
      "learning_rate": 0.00036744260193592856,
      "loss": 2.0934,
      "step": 98572
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.9463460445404053,
      "learning_rate": 0.0003674386160739136,
      "loss": 2.006,
      "step": 98573
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.9848343729972839,
      "learning_rate": 0.0003674346301993606,
      "loss": 2.3767,
      "step": 98574
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.8416457176208496,
      "learning_rate": 0.0003674306443122701,
      "loss": 2.3114,
      "step": 98575
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.981611430644989,
      "learning_rate": 0.00036742665841264305,
      "loss": 2.0797,
      "step": 98576
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.9267690181732178,
      "learning_rate": 0.00036742267250048017,
      "loss": 2.2276,
      "step": 98577
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.9222166538238525,
      "learning_rate": 0.00036741868657578207,
      "loss": 2.4014,
      "step": 98578
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.9084031581878662,
      "learning_rate": 0.00036741470063854955,
      "loss": 2.0744,
      "step": 98579
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.9638945460319519,
      "learning_rate": 0.00036741071468878345,
      "loss": 2.2244,
      "step": 98580
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.7763540148735046,
      "learning_rate": 0.00036740672872648435,
      "loss": 2.2054,
      "step": 98581
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.926706850528717,
      "learning_rate": 0.00036740274275165306,
      "loss": 2.1252,
      "step": 98582
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.8672738075256348,
      "learning_rate": 0.00036739875676429035,
      "loss": 2.1069,
      "step": 98583
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.7744833827018738,
      "learning_rate": 0.0003673947707643969,
      "loss": 2.2368,
      "step": 98584
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.8476147651672363,
      "learning_rate": 0.0003673907847519735,
      "loss": 2.0852,
      "step": 98585
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.9754524230957031,
      "learning_rate": 0.00036738679872702103,
      "loss": 2.4702,
      "step": 98586
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.8758974671363831,
      "learning_rate": 0.00036738281268953993,
      "loss": 2.1926,
      "step": 98587
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.8418878316879272,
      "learning_rate": 0.0003673788266395311,
      "loss": 2.1382,
      "step": 98588
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.9740959405899048,
      "learning_rate": 0.00036737484057699533,
      "loss": 2.1849,
      "step": 98589
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.8131762742996216,
      "learning_rate": 0.0003673708545019333,
      "loss": 2.4729,
      "step": 98590
    },
    {
      "epoch": 1.28,
      "grad_norm": 1.1521828174591064,
      "learning_rate": 0.0003673668684143457,
      "loss": 2.1082,
      "step": 98591
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.8628407120704651,
      "learning_rate": 0.0003673628823142333,
      "loss": 2.0712,
      "step": 98592
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.8233036398887634,
      "learning_rate": 0.000367358896201597,
      "loss": 2.1129,
      "step": 98593
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.8735062479972839,
      "learning_rate": 0.0003673549100764373,
      "loss": 2.139,
      "step": 98594
    },
    {
      "epoch": 1.28,
      "grad_norm": 1.0528017282485962,
      "learning_rate": 0.0003673509239387551,
      "loss": 2.3331,
      "step": 98595
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.8071422576904297,
      "learning_rate": 0.00036734693778855114,
      "loss": 2.3286,
      "step": 98596
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.8993945121765137,
      "learning_rate": 0.00036734295162582604,
      "loss": 2.1322,
      "step": 98597
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.9775543212890625,
      "learning_rate": 0.0003673389654505805,
      "loss": 2.1979,
      "step": 98598
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.8619586825370789,
      "learning_rate": 0.0003673349792628156,
      "loss": 2.3846,
      "step": 98599
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.8328344821929932,
      "learning_rate": 0.00036733099306253175,
      "loss": 2.0639,
      "step": 98600
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.812250018119812,
      "learning_rate": 0.0003673270068497298,
      "loss": 2.2901,
      "step": 98601
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.8712878227233887,
      "learning_rate": 0.00036732302062441057,
      "loss": 2.0781,
      "step": 98602
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.8892613649368286,
      "learning_rate": 0.0003673190343865746,
      "loss": 2.0906,
      "step": 98603
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.7811167240142822,
      "learning_rate": 0.00036731504813622277,
      "loss": 2.2061,
      "step": 98604
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.8581506013870239,
      "learning_rate": 0.0003673110618733559,
      "loss": 2.1731,
      "step": 98605
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.9867326021194458,
      "learning_rate": 0.0003673070755979745,
      "loss": 2.0219,
      "step": 98606
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.833658754825592,
      "learning_rate": 0.0003673030893100795,
      "loss": 2.0708,
      "step": 98607
    },
    {
      "epoch": 1.28,
      "grad_norm": 1.0360133647918701,
      "learning_rate": 0.0003672991030096716,
      "loss": 2.3395,
      "step": 98608
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.9432004690170288,
      "learning_rate": 0.0003672951166967515,
      "loss": 2.1819,
      "step": 98609
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.8165406584739685,
      "learning_rate": 0.00036729113037131997,
      "loss": 2.2154,
      "step": 98610
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.9150300621986389,
      "learning_rate": 0.00036728714403337783,
      "loss": 2.1062,
      "step": 98611
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.9823417067527771,
      "learning_rate": 0.0003672831576829256,
      "loss": 2.0681,
      "step": 98612
    },
    {
      "epoch": 1.28,
      "grad_norm": 1.17691969871521,
      "learning_rate": 0.0003672791713199642,
      "loss": 2.0201,
      "step": 98613
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.8295847177505493,
      "learning_rate": 0.00036727518494449453,
      "loss": 2.0209,
      "step": 98614
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.8913447260856628,
      "learning_rate": 0.00036727119855651695,
      "loss": 2.2979,
      "step": 98615
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.9355307817459106,
      "learning_rate": 0.00036726721215603233,
      "loss": 2.102,
      "step": 98616
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.798517107963562,
      "learning_rate": 0.00036726322574304155,
      "loss": 2.2115,
      "step": 98617
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.7985079884529114,
      "learning_rate": 0.00036725923931754526,
      "loss": 2.2681,
      "step": 98618
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.8827319741249084,
      "learning_rate": 0.00036725525287954416,
      "loss": 2.2945,
      "step": 98619
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.9084373712539673,
      "learning_rate": 0.0003672512664290391,
      "loss": 2.1712,
      "step": 98620
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.9462850093841553,
      "learning_rate": 0.00036724727996603073,
      "loss": 2.0372,
      "step": 98621
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.7847900390625,
      "learning_rate": 0.00036724329349051987,
      "loss": 2.2725,
      "step": 98622
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.8676093220710754,
      "learning_rate": 0.0003672393070025072,
      "loss": 2.1068,
      "step": 98623
    },
    {
      "epoch": 1.28,
      "grad_norm": 1.2000892162322998,
      "learning_rate": 0.0003672353205019934,
      "loss": 2.2124,
      "step": 98624
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.8578318357467651,
      "learning_rate": 0.00036723133398897936,
      "loss": 2.1352,
      "step": 98625
    },
    {
      "epoch": 1.28,
      "grad_norm": 1.3426240682601929,
      "learning_rate": 0.00036722734746346574,
      "loss": 2.348,
      "step": 98626
    },
    {
      "epoch": 1.28,
      "grad_norm": 1.1655057668685913,
      "learning_rate": 0.0003672233609254532,
      "loss": 2.1206,
      "step": 98627
    },
    {
      "epoch": 1.28,
      "grad_norm": 1.0003645420074463,
      "learning_rate": 0.0003672193743749427,
      "loss": 2.2528,
      "step": 98628
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.836042582988739,
      "learning_rate": 0.0003672153878119348,
      "loss": 2.3111,
      "step": 98629
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.8984490633010864,
      "learning_rate": 0.00036721140123643025,
      "loss": 2.2401,
      "step": 98630
    },
    {
      "epoch": 1.28,
      "grad_norm": 1.0485326051712036,
      "learning_rate": 0.0003672074146484299,
      "loss": 2.0449,
      "step": 98631
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.9053646326065063,
      "learning_rate": 0.00036720342804793426,
      "loss": 2.2434,
      "step": 98632
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.8017616868019104,
      "learning_rate": 0.00036719944143494444,
      "loss": 2.3203,
      "step": 98633
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.9131001234054565,
      "learning_rate": 0.0003671954548094609,
      "loss": 1.9492,
      "step": 98634
    },
    {
      "epoch": 1.28,
      "grad_norm": 1.336241602897644,
      "learning_rate": 0.00036719146817148444,
      "loss": 2.1378,
      "step": 98635
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.7927589416503906,
      "learning_rate": 0.00036718748152101576,
      "loss": 2.3262,
      "step": 98636
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.8286481499671936,
      "learning_rate": 0.0003671834948580558,
      "loss": 1.944,
      "step": 98637
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.8701414465904236,
      "learning_rate": 0.000367179508182605,
      "loss": 2.3155,
      "step": 98638
    },
    {
      "epoch": 1.28,
      "grad_norm": 1.215603232383728,
      "learning_rate": 0.00036717552149466437,
      "loss": 2.1919,
      "step": 98639
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.9473965167999268,
      "learning_rate": 0.0003671715347942345,
      "loss": 2.1907,
      "step": 98640
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.9973123073577881,
      "learning_rate": 0.00036716754808131623,
      "loss": 2.1131,
      "step": 98641
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.9591516256332397,
      "learning_rate": 0.00036716356135591016,
      "loss": 2.1142,
      "step": 98642
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.8686679005622864,
      "learning_rate": 0.0003671595746180171,
      "loss": 2.1145,
      "step": 98643
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.9659823775291443,
      "learning_rate": 0.00036715558786763796,
      "loss": 2.0247,
      "step": 98644
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.9910885691642761,
      "learning_rate": 0.0003671516011047732,
      "loss": 2.1693,
      "step": 98645
    },
    {
      "epoch": 1.28,
      "grad_norm": 1.791308879852295,
      "learning_rate": 0.00036714761432942377,
      "loss": 2.1175,
      "step": 98646
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.8816343545913696,
      "learning_rate": 0.00036714362754159026,
      "loss": 2.2063,
      "step": 98647
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.9279497265815735,
      "learning_rate": 0.0003671396407412735,
      "loss": 2.0682,
      "step": 98648
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.9992627501487732,
      "learning_rate": 0.00036713565392847424,
      "loss": 2.0469,
      "step": 98649
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.9891794323921204,
      "learning_rate": 0.0003671316671031932,
      "loss": 2.2535,
      "step": 98650
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.8212514519691467,
      "learning_rate": 0.0003671276802654311,
      "loss": 2.2106,
      "step": 98651
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.9390556812286377,
      "learning_rate": 0.0003671236934151887,
      "loss": 2.1041,
      "step": 98652
    },
    {
      "epoch": 1.28,
      "grad_norm": 1.1006985902786255,
      "learning_rate": 0.0003671197065524667,
      "loss": 2.0595,
      "step": 98653
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.8705479502677917,
      "learning_rate": 0.0003671157196772659,
      "loss": 2.1178,
      "step": 98654
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.9168892502784729,
      "learning_rate": 0.000367111732789587,
      "loss": 2.0841,
      "step": 98655
    },
    {
      "epoch": 1.28,
      "grad_norm": 1.1136709451675415,
      "learning_rate": 0.00036710774588943077,
      "loss": 2.0579,
      "step": 98656
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.9895751476287842,
      "learning_rate": 0.000367103758976798,
      "loss": 2.2566,
      "step": 98657
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.8036046028137207,
      "learning_rate": 0.00036709977205168936,
      "loss": 2.3012,
      "step": 98658
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.8840785622596741,
      "learning_rate": 0.0003670957851141055,
      "loss": 2.2577,
      "step": 98659
    },
    {
      "epoch": 1.28,
      "grad_norm": 1.0033258199691772,
      "learning_rate": 0.00036709179816404744,
      "loss": 2.2519,
      "step": 98660
    },
    {
      "epoch": 1.28,
      "grad_norm": 1.025840401649475,
      "learning_rate": 0.0003670878112015157,
      "loss": 2.1699,
      "step": 98661
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.9554508924484253,
      "learning_rate": 0.00036708382422651097,
      "loss": 1.9875,
      "step": 98662
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.9650566577911377,
      "learning_rate": 0.0003670798372390342,
      "loss": 2.1568,
      "step": 98663
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.896984875202179,
      "learning_rate": 0.00036707585023908597,
      "loss": 2.0192,
      "step": 98664
    },
    {
      "epoch": 1.28,
      "grad_norm": 1.2626712322235107,
      "learning_rate": 0.0003670718632266671,
      "loss": 2.1854,
      "step": 98665
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.9746719598770142,
      "learning_rate": 0.0003670678762017783,
      "loss": 2.0492,
      "step": 98666
    },
    {
      "epoch": 1.28,
      "grad_norm": 1.019859790802002,
      "learning_rate": 0.0003670638891644202,
      "loss": 2.1098,
      "step": 98667
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.8286156058311462,
      "learning_rate": 0.00036705990211459373,
      "loss": 2.0517,
      "step": 98668
    },
    {
      "epoch": 1.28,
      "grad_norm": 1.0415973663330078,
      "learning_rate": 0.0003670559150522997,
      "loss": 2.0024,
      "step": 98669
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.8655554056167603,
      "learning_rate": 0.00036705192797753853,
      "loss": 2.2811,
      "step": 98670
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.8538329601287842,
      "learning_rate": 0.0003670479408903112,
      "loss": 2.125,
      "step": 98671
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.9186115264892578,
      "learning_rate": 0.00036704395379061843,
      "loss": 2.1466,
      "step": 98672
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.9135866165161133,
      "learning_rate": 0.00036703996667846085,
      "loss": 2.1415,
      "step": 98673
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.9161661863327026,
      "learning_rate": 0.00036703597955383934,
      "loss": 2.1897,
      "step": 98674
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.9713737964630127,
      "learning_rate": 0.00036703199241675455,
      "loss": 2.0225,
      "step": 98675
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.9244167804718018,
      "learning_rate": 0.0003670280052672072,
      "loss": 2.095,
      "step": 98676
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.9647074937820435,
      "learning_rate": 0.00036702401810519814,
      "loss": 2.1567,
      "step": 98677
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.8417748808860779,
      "learning_rate": 0.0003670200309307281,
      "loss": 2.0599,
      "step": 98678
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.8336970806121826,
      "learning_rate": 0.00036701604374379766,
      "loss": 2.0363,
      "step": 98679
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.9250370860099792,
      "learning_rate": 0.00036701205654440767,
      "loss": 2.1722,
      "step": 98680
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.939144492149353,
      "learning_rate": 0.000367008069332559,
      "loss": 2.2098,
      "step": 98681
    },
    {
      "epoch": 1.28,
      "grad_norm": 1.0025348663330078,
      "learning_rate": 0.00036700408210825215,
      "loss": 2.2115,
      "step": 98682
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.8053387999534607,
      "learning_rate": 0.00036700009487148803,
      "loss": 1.9099,
      "step": 98683
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.8604979515075684,
      "learning_rate": 0.00036699610762226735,
      "loss": 2.3819,
      "step": 98684
    },
    {
      "epoch": 1.28,
      "grad_norm": 1.0420219898223877,
      "learning_rate": 0.00036699212036059076,
      "loss": 2.2229,
      "step": 98685
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.847384512424469,
      "learning_rate": 0.00036698813308645907,
      "loss": 2.0996,
      "step": 98686
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.8507968187332153,
      "learning_rate": 0.00036698414579987315,
      "loss": 2.3677,
      "step": 98687
    },
    {
      "epoch": 1.28,
      "grad_norm": 1.0024149417877197,
      "learning_rate": 0.00036698015850083343,
      "loss": 2.3112,
      "step": 98688
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.9375738501548767,
      "learning_rate": 0.00036697617118934095,
      "loss": 2.1762,
      "step": 98689
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.9030793309211731,
      "learning_rate": 0.0003669721838653963,
      "loss": 2.1789,
      "step": 98690
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.7693650126457214,
      "learning_rate": 0.00036696819652900023,
      "loss": 2.3304,
      "step": 98691
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.8438075184822083,
      "learning_rate": 0.00036696420918015356,
      "loss": 2.3824,
      "step": 98692
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.886371374130249,
      "learning_rate": 0.000366960221818857,
      "loss": 2.1577,
      "step": 98693
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.8716925382614136,
      "learning_rate": 0.0003669562344451112,
      "loss": 2.1769,
      "step": 98694
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.925414502620697,
      "learning_rate": 0.000366952247058917,
      "loss": 2.4077,
      "step": 98695
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.8331462740898132,
      "learning_rate": 0.00036694825966027513,
      "loss": 2.2258,
      "step": 98696
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.8188999891281128,
      "learning_rate": 0.00036694427224918626,
      "loss": 2.095,
      "step": 98697
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.8850101232528687,
      "learning_rate": 0.00036694028482565124,
      "loss": 2.154,
      "step": 98698
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.8278873562812805,
      "learning_rate": 0.0003669362973896708,
      "loss": 2.1461,
      "step": 98699
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.7809970378875732,
      "learning_rate": 0.00036693230994124556,
      "loss": 2.1592,
      "step": 98700
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.8684226870536804,
      "learning_rate": 0.0003669283224803763,
      "loss": 2.1253,
      "step": 98701
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.9837160110473633,
      "learning_rate": 0.0003669243350070639,
      "loss": 2.1676,
      "step": 98702
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.9269702434539795,
      "learning_rate": 0.000366920347521309,
      "loss": 1.966,
      "step": 98703
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.8417218923568726,
      "learning_rate": 0.0003669163600231123,
      "loss": 2.1896,
      "step": 98704
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.7989289164543152,
      "learning_rate": 0.0003669123725124746,
      "loss": 2.0654,
      "step": 98705
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.861354410648346,
      "learning_rate": 0.0003669083849893966,
      "loss": 2.309,
      "step": 98706
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.8887221217155457,
      "learning_rate": 0.0003669043974538791,
      "loss": 2.1896,
      "step": 98707
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.7963987588882446,
      "learning_rate": 0.00036690040990592287,
      "loss": 2.1925,
      "step": 98708
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.8247557878494263,
      "learning_rate": 0.00036689642234552847,
      "loss": 1.9487,
      "step": 98709
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.8511395454406738,
      "learning_rate": 0.0003668924347726968,
      "loss": 2.4301,
      "step": 98710
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.955610454082489,
      "learning_rate": 0.0003668884471874286,
      "loss": 2.2228,
      "step": 98711
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.802263081073761,
      "learning_rate": 0.00036688445958972456,
      "loss": 2.1384,
      "step": 98712
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.9481539726257324,
      "learning_rate": 0.0003668804719795854,
      "loss": 2.1958,
      "step": 98713
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.8249249458312988,
      "learning_rate": 0.000366876484357012,
      "loss": 2.1284,
      "step": 98714
    },
    {
      "epoch": 1.29,
      "grad_norm": 1.232054352760315,
      "learning_rate": 0.00036687249672200486,
      "loss": 2.0291,
      "step": 98715
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.895232081413269,
      "learning_rate": 0.0003668685090745649,
      "loss": 2.1807,
      "step": 98716
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.9573867321014404,
      "learning_rate": 0.00036686452141469284,
      "loss": 2.4177,
      "step": 98717
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.8610703945159912,
      "learning_rate": 0.00036686053374238947,
      "loss": 2.1835,
      "step": 98718
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.9360353350639343,
      "learning_rate": 0.0003668565460576554,
      "loss": 2.0403,
      "step": 98719
    },
    {
      "epoch": 1.29,
      "grad_norm": 1.146877646446228,
      "learning_rate": 0.0003668525583604914,
      "loss": 2.2273,
      "step": 98720
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.846635103225708,
      "learning_rate": 0.0003668485706508983,
      "loss": 2.1217,
      "step": 98721
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.8608344197273254,
      "learning_rate": 0.0003668445829288768,
      "loss": 2.1097,
      "step": 98722
    },
    {
      "epoch": 1.29,
      "grad_norm": 1.0847197771072388,
      "learning_rate": 0.0003668405951944276,
      "loss": 2.1472,
      "step": 98723
    },
    {
      "epoch": 1.29,
      "grad_norm": 1.0301190614700317,
      "learning_rate": 0.00036683660744755146,
      "loss": 2.3153,
      "step": 98724
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.8311353921890259,
      "learning_rate": 0.0003668326196882492,
      "loss": 2.0429,
      "step": 98725
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.9755411148071289,
      "learning_rate": 0.0003668286319165214,
      "loss": 2.3148,
      "step": 98726
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.8937636017799377,
      "learning_rate": 0.00036682464413236894,
      "loss": 2.2578,
      "step": 98727
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.8964657187461853,
      "learning_rate": 0.00036682065633579253,
      "loss": 2.1546,
      "step": 98728
    },
    {
      "epoch": 1.29,
      "grad_norm": 1.017250657081604,
      "learning_rate": 0.0003668166685267929,
      "loss": 2.2637,
      "step": 98729
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.828025221824646,
      "learning_rate": 0.00036681268070537076,
      "loss": 1.9681,
      "step": 98730
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.8741338849067688,
      "learning_rate": 0.00036680869287152685,
      "loss": 2.0924,
      "step": 98731
    },
    {
      "epoch": 1.29,
      "grad_norm": 1.1668425798416138,
      "learning_rate": 0.00036680470502526206,
      "loss": 2.1514,
      "step": 98732
    },
    {
      "epoch": 1.29,
      "grad_norm": 1.091853141784668,
      "learning_rate": 0.0003668007171665769,
      "loss": 2.2382,
      "step": 98733
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.7937376499176025,
      "learning_rate": 0.00036679672929547234,
      "loss": 2.4333,
      "step": 98734
    },
    {
      "epoch": 1.29,
      "grad_norm": 1.1266106367111206,
      "learning_rate": 0.00036679274141194893,
      "loss": 2.0526,
      "step": 98735
    },
    {
      "epoch": 1.29,
      "grad_norm": 1.135807752609253,
      "learning_rate": 0.0003667887535160075,
      "loss": 2.2058,
      "step": 98736
    },
    {
      "epoch": 1.29,
      "grad_norm": 1.0672615766525269,
      "learning_rate": 0.0003667847656076488,
      "loss": 1.9992,
      "step": 98737
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.8934386968612671,
      "learning_rate": 0.0003667807776868735,
      "loss": 2.227,
      "step": 98738
    },
    {
      "epoch": 1.29,
      "grad_norm": 1.2109869718551636,
      "learning_rate": 0.00036677678975368245,
      "loss": 2.1549,
      "step": 98739
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.8605210781097412,
      "learning_rate": 0.0003667728018080763,
      "loss": 2.2082,
      "step": 98740
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.8928049206733704,
      "learning_rate": 0.0003667688138500558,
      "loss": 2.0999,
      "step": 98741
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.9804349541664124,
      "learning_rate": 0.0003667648258796218,
      "loss": 2.2399,
      "step": 98742
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.8062388896942139,
      "learning_rate": 0.0003667608378967749,
      "loss": 2.2262,
      "step": 98743
    },
    {
      "epoch": 1.29,
      "grad_norm": 1.0777692794799805,
      "learning_rate": 0.0003667568499015159,
      "loss": 2.2873,
      "step": 98744
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.9570872187614441,
      "learning_rate": 0.0003667528618938456,
      "loss": 2.2728,
      "step": 98745
    },
    {
      "epoch": 1.29,
      "grad_norm": 1.2662665843963623,
      "learning_rate": 0.0003667488738737646,
      "loss": 2.209,
      "step": 98746
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.9064745306968689,
      "learning_rate": 0.0003667448858412737,
      "loss": 2.3437,
      "step": 98747
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.8537775278091431,
      "learning_rate": 0.0003667408977963738,
      "loss": 2.4474,
      "step": 98748
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.9407549500465393,
      "learning_rate": 0.0003667369097390654,
      "loss": 2.3072,
      "step": 98749
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.8558276295661926,
      "learning_rate": 0.0003667329216693494,
      "loss": 2.0805,
      "step": 98750
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.8365955948829651,
      "learning_rate": 0.00036672893358722644,
      "loss": 2.0107,
      "step": 98751
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.8309130072593689,
      "learning_rate": 0.0003667249454926974,
      "loss": 2.0558,
      "step": 98752
    },
    {
      "epoch": 1.29,
      "grad_norm": 1.0408810377120972,
      "learning_rate": 0.0003667209573857628,
      "loss": 2.0454,
      "step": 98753
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.934047281742096,
      "learning_rate": 0.00036671696926642364,
      "loss": 2.3214,
      "step": 98754
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.8132582306861877,
      "learning_rate": 0.00036671298113468044,
      "loss": 2.2224,
      "step": 98755
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.8924217224121094,
      "learning_rate": 0.00036670899299053413,
      "loss": 2.2744,
      "step": 98756
    },
    {
      "epoch": 1.29,
      "grad_norm": 1.1185591220855713,
      "learning_rate": 0.0003667050048339853,
      "loss": 2.0935,
      "step": 98757
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.9820084571838379,
      "learning_rate": 0.0003667010166650348,
      "loss": 2.1513,
      "step": 98758
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.8372491002082825,
      "learning_rate": 0.00036669702848368327,
      "loss": 2.2491,
      "step": 98759
    },
    {
      "epoch": 1.29,
      "grad_norm": 1.01585054397583,
      "learning_rate": 0.00036669304028993157,
      "loss": 2.3079,
      "step": 98760
    },
    {
      "epoch": 1.29,
      "grad_norm": 1.243223786354065,
      "learning_rate": 0.0003666890520837803,
      "loss": 2.1761,
      "step": 98761
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.8101227283477783,
      "learning_rate": 0.0003666850638652303,
      "loss": 2.0612,
      "step": 98762
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.9444047808647156,
      "learning_rate": 0.00036668107563428225,
      "loss": 2.2748,
      "step": 98763
    },
    {
      "epoch": 1.29,
      "grad_norm": 1.2513630390167236,
      "learning_rate": 0.000366677087390937,
      "loss": 2.1359,
      "step": 98764
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.9124813675880432,
      "learning_rate": 0.00036667309913519515,
      "loss": 2.2913,
      "step": 98765
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.8667031526565552,
      "learning_rate": 0.00036666911086705764,
      "loss": 2.237,
      "step": 98766
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.8025360703468323,
      "learning_rate": 0.00036666512258652493,
      "loss": 2.0308,
      "step": 98767
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.8453229665756226,
      "learning_rate": 0.00036666113429359795,
      "loss": 2.2433,
      "step": 98768
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.9224358201026917,
      "learning_rate": 0.00036665714598827745,
      "loss": 2.1229,
      "step": 98769
    },
    {
      "epoch": 1.29,
      "grad_norm": 1.0394505262374878,
      "learning_rate": 0.0003666531576705641,
      "loss": 2.1429,
      "step": 98770
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.7680955529212952,
      "learning_rate": 0.00036664916934045874,
      "loss": 2.0613,
      "step": 98771
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.9365329742431641,
      "learning_rate": 0.00036664518099796203,
      "loss": 2.4468,
      "step": 98772
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.9249572157859802,
      "learning_rate": 0.00036664119264307463,
      "loss": 2.2512,
      "step": 98773
    },
    {
      "epoch": 1.29,
      "grad_norm": 1.0405526161193848,
      "learning_rate": 0.0003666372042757974,
      "loss": 1.9587,
      "step": 98774
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.8765777349472046,
      "learning_rate": 0.00036663321589613116,
      "loss": 2.2474,
      "step": 98775
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.9729783535003662,
      "learning_rate": 0.0003666292275040764,
      "loss": 2.2941,
      "step": 98776
    },
    {
      "epoch": 1.29,
      "grad_norm": 1.1107510328292847,
      "learning_rate": 0.00036662523909963406,
      "loss": 2.1587,
      "step": 98777
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.9592156410217285,
      "learning_rate": 0.0003666212506828049,
      "loss": 2.0992,
      "step": 98778
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.9573073983192444,
      "learning_rate": 0.0003666172622535895,
      "loss": 2.2991,
      "step": 98779
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.8124768137931824,
      "learning_rate": 0.0003666132738119887,
      "loss": 2.104,
      "step": 98780
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.846849262714386,
      "learning_rate": 0.00036660928535800334,
      "loss": 2.1853,
      "step": 98781
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.7917734384536743,
      "learning_rate": 0.00036660529689163394,
      "loss": 2.2417,
      "step": 98782
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.8833884000778198,
      "learning_rate": 0.0003666013084128814,
      "loss": 2.3041,
      "step": 98783
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.843544065952301,
      "learning_rate": 0.00036659731992174646,
      "loss": 1.8976,
      "step": 98784
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.8598634600639343,
      "learning_rate": 0.0003665933314182298,
      "loss": 2.2895,
      "step": 98785
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.978212296962738,
      "learning_rate": 0.0003665893429023321,
      "loss": 2.5979,
      "step": 98786
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.9422559142112732,
      "learning_rate": 0.0003665853543740543,
      "loss": 2.3916,
      "step": 98787
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.7778717875480652,
      "learning_rate": 0.00036658136583339693,
      "loss": 1.9561,
      "step": 98788
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.7905868887901306,
      "learning_rate": 0.0003665773772803609,
      "loss": 2.0932,
      "step": 98789
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.9458755850791931,
      "learning_rate": 0.0003665733887149468,
      "loss": 2.0727,
      "step": 98790
    },
    {
      "epoch": 1.29,
      "grad_norm": 1.023797631263733,
      "learning_rate": 0.0003665694001371556,
      "loss": 2.2644,
      "step": 98791
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.9676910638809204,
      "learning_rate": 0.0003665654115469877,
      "loss": 2.0473,
      "step": 98792
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.873245120048523,
      "learning_rate": 0.0003665614229444441,
      "loss": 2.1289,
      "step": 98793
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.835578978061676,
      "learning_rate": 0.00036655743432952557,
      "loss": 2.176,
      "step": 98794
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.9234142899513245,
      "learning_rate": 0.00036655344570223265,
      "loss": 2.1984,
      "step": 98795
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.9083157181739807,
      "learning_rate": 0.0003665494570625662,
      "loss": 2.2199,
      "step": 98796
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.8715697526931763,
      "learning_rate": 0.000366545468410527,
      "loss": 2.2161,
      "step": 98797
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.9144651889801025,
      "learning_rate": 0.00036654147974611573,
      "loss": 2.4304,
      "step": 98798
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.84330153465271,
      "learning_rate": 0.00036653749106933315,
      "loss": 2.3155,
      "step": 98799
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.8540379405021667,
      "learning_rate": 0.00036653350238018,
      "loss": 1.8864,
      "step": 98800
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.8861416578292847,
      "learning_rate": 0.000366529513678657,
      "loss": 2.1479,
      "step": 98801
    },
    {
      "epoch": 1.29,
      "grad_norm": 1.045373558998108,
      "learning_rate": 0.0003665255249647649,
      "loss": 2.1079,
      "step": 98802
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.8730297684669495,
      "learning_rate": 0.0003665215362385044,
      "loss": 2.1829,
      "step": 98803
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.8066524267196655,
      "learning_rate": 0.00036651754749987634,
      "loss": 2.1333,
      "step": 98804
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.9978553652763367,
      "learning_rate": 0.00036651355874888137,
      "loss": 2.1513,
      "step": 98805
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.8060687184333801,
      "learning_rate": 0.00036650956998552037,
      "loss": 2.1785,
      "step": 98806
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.8279905915260315,
      "learning_rate": 0.0003665055812097939,
      "loss": 2.1268,
      "step": 98807
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.8223564624786377,
      "learning_rate": 0.0003665015924217028,
      "loss": 2.2661,
      "step": 98808
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.9029942750930786,
      "learning_rate": 0.0003664976036212478,
      "loss": 2.1372,
      "step": 98809
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.8491416573524475,
      "learning_rate": 0.0003664936148084297,
      "loss": 2.1261,
      "step": 98810
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.88639897108078,
      "learning_rate": 0.00036648962598324913,
      "loss": 2.1062,
      "step": 98811
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.9723880887031555,
      "learning_rate": 0.00036648563714570685,
      "loss": 2.5639,
      "step": 98812
    },
    {
      "epoch": 1.29,
      "grad_norm": 1.0144588947296143,
      "learning_rate": 0.0003664816482958037,
      "loss": 2.1112,
      "step": 98813
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.8541169166564941,
      "learning_rate": 0.0003664776594335403,
      "loss": 2.1964,
      "step": 98814
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.834200382232666,
      "learning_rate": 0.0003664736705589175,
      "loss": 2.0526,
      "step": 98815
    },
    {
      "epoch": 1.29,
      "grad_norm": 1.094395637512207,
      "learning_rate": 0.00036646968167193593,
      "loss": 2.1646,
      "step": 98816
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.8490678071975708,
      "learning_rate": 0.0003664656927725965,
      "loss": 2.3023,
      "step": 98817
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.8718973398208618,
      "learning_rate": 0.00036646170386089974,
      "loss": 2.2741,
      "step": 98818
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.953283965587616,
      "learning_rate": 0.0003664577149368465,
      "loss": 2.2676,
      "step": 98819
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.8373638391494751,
      "learning_rate": 0.00036645372600043754,
      "loss": 2.2046,
      "step": 98820
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.8199650049209595,
      "learning_rate": 0.0003664497370516736,
      "loss": 2.1658,
      "step": 98821
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.895466148853302,
      "learning_rate": 0.0003664457480905554,
      "loss": 2.2513,
      "step": 98822
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.845218300819397,
      "learning_rate": 0.00036644175911708366,
      "loss": 2.0888,
      "step": 98823
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.8898456692695618,
      "learning_rate": 0.00036643777013125917,
      "loss": 1.9873,
      "step": 98824
    },
    {
      "epoch": 1.29,
      "grad_norm": 1.0078861713409424,
      "learning_rate": 0.00036643378113308263,
      "loss": 2.179,
      "step": 98825
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.9110655188560486,
      "learning_rate": 0.0003664297921225547,
      "loss": 2.3757,
      "step": 98826
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.8166179656982422,
      "learning_rate": 0.00036642580309967636,
      "loss": 2.0491,
      "step": 98827
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.8610459566116333,
      "learning_rate": 0.0003664218140644482,
      "loss": 2.2575,
      "step": 98828
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.8179525136947632,
      "learning_rate": 0.00036641782501687086,
      "loss": 2.1203,
      "step": 98829
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.8784757852554321,
      "learning_rate": 0.0003664138359569453,
      "loss": 2.0895,
      "step": 98830
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.8388949036598206,
      "learning_rate": 0.0003664098468846721,
      "loss": 2.2003,
      "step": 98831
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.9501452445983887,
      "learning_rate": 0.00036640585780005206,
      "loss": 2.1003,
      "step": 98832
    },
    {
      "epoch": 1.29,
      "grad_norm": 1.0580120086669922,
      "learning_rate": 0.0003664018687030859,
      "loss": 2.1445,
      "step": 98833
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.8617745041847229,
      "learning_rate": 0.00036639787959377447,
      "loss": 1.9252,
      "step": 98834
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.89577317237854,
      "learning_rate": 0.0003663938904721183,
      "loss": 2.1982,
      "step": 98835
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.8354947566986084,
      "learning_rate": 0.0003663899013381183,
      "loss": 2.0854,
      "step": 98836
    },
    {
      "epoch": 1.29,
      "grad_norm": 1.0602601766586304,
      "learning_rate": 0.0003663859121917752,
      "loss": 2.3001,
      "step": 98837
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.826904296875,
      "learning_rate": 0.0003663819230330897,
      "loss": 2.0071,
      "step": 98838
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.9305035471916199,
      "learning_rate": 0.00036637793386206253,
      "loss": 2.3188,
      "step": 98839
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.9634577035903931,
      "learning_rate": 0.0003663739446786944,
      "loss": 2.1836,
      "step": 98840
    },
    {
      "epoch": 1.29,
      "grad_norm": 1.0006694793701172,
      "learning_rate": 0.0003663699554829862,
      "loss": 2.2436,
      "step": 98841
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.9758203029632568,
      "learning_rate": 0.0003663659662749385,
      "loss": 2.2386,
      "step": 98842
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.9059287309646606,
      "learning_rate": 0.0003663619770545521,
      "loss": 2.3088,
      "step": 98843
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.938893735408783,
      "learning_rate": 0.00036635798782182774,
      "loss": 2.2483,
      "step": 98844
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.8807771801948547,
      "learning_rate": 0.0003663539985767663,
      "loss": 2.305,
      "step": 98845
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.8577018976211548,
      "learning_rate": 0.00036635000931936834,
      "loss": 2.2979,
      "step": 98846
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.918767511844635,
      "learning_rate": 0.00036634602004963457,
      "loss": 2.0788,
      "step": 98847
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.8379306793212891,
      "learning_rate": 0.000366342030767566,
      "loss": 2.2096,
      "step": 98848
    },
    {
      "epoch": 1.29,
      "grad_norm": 1.0106902122497559,
      "learning_rate": 0.0003663380414731631,
      "loss": 2.2947,
      "step": 98849
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.9264925718307495,
      "learning_rate": 0.0003663340521664266,
      "loss": 2.2528,
      "step": 98850
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.7955231070518494,
      "learning_rate": 0.0003663300628473575,
      "loss": 2.3406,
      "step": 98851
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.7976225018501282,
      "learning_rate": 0.00036632607351595633,
      "loss": 2.0966,
      "step": 98852
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.8463118672370911,
      "learning_rate": 0.00036632208417222386,
      "loss": 2.1881,
      "step": 98853
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.811741054058075,
      "learning_rate": 0.0003663180948161609,
      "loss": 2.286,
      "step": 98854
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.8630146384239197,
      "learning_rate": 0.0003663141054477682,
      "loss": 2.0316,
      "step": 98855
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.8898752927780151,
      "learning_rate": 0.0003663101160670464,
      "loss": 2.0714,
      "step": 98856
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.9426694512367249,
      "learning_rate": 0.00036630612667399627,
      "loss": 2.1554,
      "step": 98857
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.8576640486717224,
      "learning_rate": 0.00036630213726861866,
      "loss": 2.1676,
      "step": 98858
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.8756137490272522,
      "learning_rate": 0.00036629814785091416,
      "loss": 2.2857,
      "step": 98859
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.9495912194252014,
      "learning_rate": 0.0003662941584208836,
      "loss": 2.095,
      "step": 98860
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.9303774237632751,
      "learning_rate": 0.00036629016897852775,
      "loss": 2.1753,
      "step": 98861
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.8244590163230896,
      "learning_rate": 0.00036628617952384724,
      "loss": 2.3136,
      "step": 98862
    },
    {
      "epoch": 1.29,
      "grad_norm": 1.0009868144989014,
      "learning_rate": 0.0003662821900568429,
      "loss": 2.0123,
      "step": 98863
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.9068808555603027,
      "learning_rate": 0.0003662782005775155,
      "loss": 2.1115,
      "step": 98864
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.8150415420532227,
      "learning_rate": 0.0003662742110858657,
      "loss": 2.2669,
      "step": 98865
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.866929292678833,
      "learning_rate": 0.00036627022158189425,
      "loss": 2.2283,
      "step": 98866
    },
    {
      "epoch": 1.29,
      "grad_norm": 1.0013453960418701,
      "learning_rate": 0.0003662662320656019,
      "loss": 2.1215,
      "step": 98867
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.8820121884346008,
      "learning_rate": 0.0003662622425369894,
      "loss": 2.2272,
      "step": 98868
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.9222077131271362,
      "learning_rate": 0.00036625825299605754,
      "loss": 2.2912,
      "step": 98869
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.9159741997718811,
      "learning_rate": 0.00036625426344280706,
      "loss": 2.0136,
      "step": 98870
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.923326849937439,
      "learning_rate": 0.00036625027387723864,
      "loss": 1.9966,
      "step": 98871
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.880898118019104,
      "learning_rate": 0.00036624628429935297,
      "loss": 2.1573,
      "step": 98872
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.858527421951294,
      "learning_rate": 0.000366242294709151,
      "loss": 2.2338,
      "step": 98873
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.8889861702919006,
      "learning_rate": 0.00036623830510663317,
      "loss": 2.1595,
      "step": 98874
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.9885287880897522,
      "learning_rate": 0.00036623431549180044,
      "loss": 2.2566,
      "step": 98875
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.8653396964073181,
      "learning_rate": 0.0003662303258646536,
      "loss": 2.022,
      "step": 98876
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.8335219621658325,
      "learning_rate": 0.0003662263362251932,
      "loss": 2.2955,
      "step": 98877
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.8353819847106934,
      "learning_rate": 0.00036622234657342006,
      "loss": 2.1641,
      "step": 98878
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.8123951554298401,
      "learning_rate": 0.000366218356909335,
      "loss": 2.3489,
      "step": 98879
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.86528480052948,
      "learning_rate": 0.0003662143672329387,
      "loss": 1.9157,
      "step": 98880
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.8652944564819336,
      "learning_rate": 0.00036621037754423183,
      "loss": 2.2454,
      "step": 98881
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.8463975191116333,
      "learning_rate": 0.0003662063878432153,
      "loss": 2.1127,
      "step": 98882
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.9142770171165466,
      "learning_rate": 0.0003662023981298896,
      "loss": 2.1786,
      "step": 98883
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.8516234159469604,
      "learning_rate": 0.00036619840840425576,
      "loss": 2.2321,
      "step": 98884
    },
    {
      "epoch": 1.29,
      "grad_norm": 1.0732252597808838,
      "learning_rate": 0.00036619441866631437,
      "loss": 2.4384,
      "step": 98885
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.8388611078262329,
      "learning_rate": 0.00036619042891606615,
      "loss": 2.196,
      "step": 98886
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.7868493795394897,
      "learning_rate": 0.00036618643915351185,
      "loss": 1.8855,
      "step": 98887
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.8777517080307007,
      "learning_rate": 0.0003661824493786523,
      "loss": 2.0549,
      "step": 98888
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.9101560711860657,
      "learning_rate": 0.0003661784595914882,
      "loss": 2.1686,
      "step": 98889
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.8750910758972168,
      "learning_rate": 0.0003661744697920202,
      "loss": 2.237,
      "step": 98890
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.9501637816429138,
      "learning_rate": 0.0003661704799802492,
      "loss": 2.1715,
      "step": 98891
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.891828715801239,
      "learning_rate": 0.0003661664901561758,
      "loss": 2.1606,
      "step": 98892
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.8524889349937439,
      "learning_rate": 0.0003661625003198008,
      "loss": 2.2045,
      "step": 98893
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.8474975824356079,
      "learning_rate": 0.00036615851047112495,
      "loss": 2.2019,
      "step": 98894
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.8423582911491394,
      "learning_rate": 0.000366154520610149,
      "loss": 2.0397,
      "step": 98895
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.7818876504898071,
      "learning_rate": 0.00036615053073687365,
      "loss": 2.0608,
      "step": 98896
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.8261070847511292,
      "learning_rate": 0.00036614654085129965,
      "loss": 2.355,
      "step": 98897
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.9264740347862244,
      "learning_rate": 0.00036614255095342786,
      "loss": 2.1374,
      "step": 98898
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.8692164421081543,
      "learning_rate": 0.00036613856104325884,
      "loss": 2.2971,
      "step": 98899
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.8586277365684509,
      "learning_rate": 0.00036613457112079345,
      "loss": 2.2725,
      "step": 98900
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.8113690614700317,
      "learning_rate": 0.0003661305811860324,
      "loss": 2.1395,
      "step": 98901
    },
    {
      "epoch": 1.29,
      "grad_norm": 1.10438871383667,
      "learning_rate": 0.0003661265912389764,
      "loss": 1.9442,
      "step": 98902
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.8083108067512512,
      "learning_rate": 0.00036612260127962616,
      "loss": 1.9628,
      "step": 98903
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.9845625162124634,
      "learning_rate": 0.0003661186113079825,
      "loss": 2.0785,
      "step": 98904
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.9302516579627991,
      "learning_rate": 0.00036611462132404626,
      "loss": 2.1845,
      "step": 98905
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.9332789182662964,
      "learning_rate": 0.000366110631327818,
      "loss": 2.1527,
      "step": 98906
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.9014244079589844,
      "learning_rate": 0.00036610664131929844,
      "loss": 2.2159,
      "step": 98907
    },
    {
      "epoch": 1.29,
      "grad_norm": 1.0482367277145386,
      "learning_rate": 0.00036610265129848843,
      "loss": 2.0984,
      "step": 98908
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.8909845948219299,
      "learning_rate": 0.00036609866126538883,
      "loss": 2.3136,
      "step": 98909
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.8915461301803589,
      "learning_rate": 0.00036609467122000007,
      "loss": 2.2482,
      "step": 98910
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.8420231342315674,
      "learning_rate": 0.0003660906811623232,
      "loss": 2.2666,
      "step": 98911
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.876800000667572,
      "learning_rate": 0.0003660866910923588,
      "loss": 1.9514,
      "step": 98912
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.8552722930908203,
      "learning_rate": 0.0003660827010101076,
      "loss": 2.2756,
      "step": 98913
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.8022280335426331,
      "learning_rate": 0.0003660787109155704,
      "loss": 2.1979,
      "step": 98914
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.829670250415802,
      "learning_rate": 0.00036607472080874785,
      "loss": 2.1373,
      "step": 98915
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.8620654940605164,
      "learning_rate": 0.0003660707306896408,
      "loss": 2.3084,
      "step": 98916
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.8675370216369629,
      "learning_rate": 0.00036606674055825,
      "loss": 2.0019,
      "step": 98917
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.8564225435256958,
      "learning_rate": 0.00036606275041457613,
      "loss": 2.1658,
      "step": 98918
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.888680100440979,
      "learning_rate": 0.00036605876025861994,
      "loss": 2.1838,
      "step": 98919
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.9557342529296875,
      "learning_rate": 0.00036605477009038213,
      "loss": 2.1835,
      "step": 98920
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.9017528891563416,
      "learning_rate": 0.0003660507799098635,
      "loss": 2.1201,
      "step": 98921
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.8145807385444641,
      "learning_rate": 0.0003660467897170649,
      "loss": 2.4053,
      "step": 98922
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.9025323390960693,
      "learning_rate": 0.0003660427995119868,
      "loss": 2.383,
      "step": 98923
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.8773486018180847,
      "learning_rate": 0.0003660388092946302,
      "loss": 2.0716,
      "step": 98924
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.9512557983398438,
      "learning_rate": 0.00036603481906499573,
      "loss": 2.1883,
      "step": 98925
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.954652726650238,
      "learning_rate": 0.0003660308288230841,
      "loss": 2.3178,
      "step": 98926
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.9399676322937012,
      "learning_rate": 0.00036602683856889605,
      "loss": 2.1548,
      "step": 98927
    },
    {
      "epoch": 1.29,
      "grad_norm": 1.2690414190292358,
      "learning_rate": 0.00036602284830243245,
      "loss": 2.1899,
      "step": 98928
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.8435590267181396,
      "learning_rate": 0.000366018858023694,
      "loss": 2.0719,
      "step": 98929
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.8876422047615051,
      "learning_rate": 0.0003660148677326813,
      "loss": 2.1396,
      "step": 98930
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.8207477331161499,
      "learning_rate": 0.00036601087742939525,
      "loss": 2.269,
      "step": 98931
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.9203250408172607,
      "learning_rate": 0.00036600688711383643,
      "loss": 2.1023,
      "step": 98932
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.8786200284957886,
      "learning_rate": 0.00036600289678600574,
      "loss": 1.8636,
      "step": 98933
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.8279545307159424,
      "learning_rate": 0.00036599890644590393,
      "loss": 2.1143,
      "step": 98934
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.8191681504249573,
      "learning_rate": 0.0003659949160935316,
      "loss": 2.02,
      "step": 98935
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.822296142578125,
      "learning_rate": 0.00036599092572888966,
      "loss": 2.1363,
      "step": 98936
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.8735918402671814,
      "learning_rate": 0.00036598693535197865,
      "loss": 2.1637,
      "step": 98937
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.8996680974960327,
      "learning_rate": 0.0003659829449627995,
      "loss": 2.4664,
      "step": 98938
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.9217032194137573,
      "learning_rate": 0.0003659789545613528,
      "loss": 2.0942,
      "step": 98939
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.9613948464393616,
      "learning_rate": 0.00036597496414763947,
      "loss": 2.1801,
      "step": 98940
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.7333122491836548,
      "learning_rate": 0.0003659709737216601,
      "loss": 1.9727,
      "step": 98941
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.9422796964645386,
      "learning_rate": 0.0003659669832834154,
      "loss": 2.4197,
      "step": 98942
    },
    {
      "epoch": 1.29,
      "grad_norm": 1.0586880445480347,
      "learning_rate": 0.00036596299283290633,
      "loss": 2.0537,
      "step": 98943
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.9133875966072083,
      "learning_rate": 0.00036595900237013345,
      "loss": 2.2737,
      "step": 98944
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.9160730242729187,
      "learning_rate": 0.00036595501189509743,
      "loss": 2.2814,
      "step": 98945
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.8394070267677307,
      "learning_rate": 0.0003659510214077993,
      "loss": 2.1036,
      "step": 98946
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.9063891768455505,
      "learning_rate": 0.0003659470309082395,
      "loss": 2.1125,
      "step": 98947
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.8609286546707153,
      "learning_rate": 0.000365943040396419,
      "loss": 1.9935,
      "step": 98948
    },
    {
      "epoch": 1.29,
      "grad_norm": 1.0250765085220337,
      "learning_rate": 0.00036593904987233843,
      "loss": 2.0022,
      "step": 98949
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.8449626564979553,
      "learning_rate": 0.00036593505933599853,
      "loss": 2.2802,
      "step": 98950
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.8776041865348816,
      "learning_rate": 0.00036593106878740004,
      "loss": 2.1933,
      "step": 98951
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.9420304894447327,
      "learning_rate": 0.00036592707822654373,
      "loss": 2.2316,
      "step": 98952
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.8103165030479431,
      "learning_rate": 0.0003659230876534303,
      "loss": 1.9123,
      "step": 98953
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.8726080060005188,
      "learning_rate": 0.00036591909706806054,
      "loss": 2.0268,
      "step": 98954
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.863204836845398,
      "learning_rate": 0.0003659151064704352,
      "loss": 2.0433,
      "step": 98955
    },
    {
      "epoch": 1.29,
      "grad_norm": 1.018454670906067,
      "learning_rate": 0.000365911115860555,
      "loss": 2.1969,
      "step": 98956
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.8947611451148987,
      "learning_rate": 0.00036590712523842066,
      "loss": 2.0897,
      "step": 98957
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.8831087350845337,
      "learning_rate": 0.0003659031346040329,
      "loss": 2.0332,
      "step": 98958
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.7781087160110474,
      "learning_rate": 0.0003658991439573926,
      "loss": 2.233,
      "step": 98959
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.8894413113594055,
      "learning_rate": 0.0003658951532985003,
      "loss": 2.2155,
      "step": 98960
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.805437445640564,
      "learning_rate": 0.00036589116262735697,
      "loss": 2.4367,
      "step": 98961
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.8121804594993591,
      "learning_rate": 0.0003658871719439631,
      "loss": 2.1212,
      "step": 98962
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.8811702728271484,
      "learning_rate": 0.00036588318124831964,
      "loss": 2.2511,
      "step": 98963
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.842354416847229,
      "learning_rate": 0.0003658791905404273,
      "loss": 2.0465,
      "step": 98964
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.8302247524261475,
      "learning_rate": 0.00036587519982028666,
      "loss": 2.1416,
      "step": 98965
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.8210916519165039,
      "learning_rate": 0.00036587120908789856,
      "loss": 2.3823,
      "step": 98966
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.8619142174720764,
      "learning_rate": 0.0003658672183432639,
      "loss": 2.343,
      "step": 98967
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.8674594163894653,
      "learning_rate": 0.00036586322758638313,
      "loss": 2.0957,
      "step": 98968
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.8482036590576172,
      "learning_rate": 0.0003658592368172572,
      "loss": 2.0016,
      "step": 98969
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.8643491268157959,
      "learning_rate": 0.00036585524603588687,
      "loss": 2.2918,
      "step": 98970
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.9208253026008606,
      "learning_rate": 0.0003658512552422728,
      "loss": 2.3642,
      "step": 98971
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.8664607405662537,
      "learning_rate": 0.00036584726443641556,
      "loss": 2.1231,
      "step": 98972
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.9400724172592163,
      "learning_rate": 0.0003658432736183163,
      "loss": 2.184,
      "step": 98973
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.9405821561813354,
      "learning_rate": 0.0003658392827879754,
      "loss": 2.2614,
      "step": 98974
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.9226588010787964,
      "learning_rate": 0.00036583529194539376,
      "loss": 2.0508,
      "step": 98975
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.8536309599876404,
      "learning_rate": 0.0003658313010905721,
      "loss": 2.1316,
      "step": 98976
    },
    {
      "epoch": 1.29,
      "grad_norm": 1.0083929300308228,
      "learning_rate": 0.00036582731022351115,
      "loss": 2.1147,
      "step": 98977
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.9295017123222351,
      "learning_rate": 0.00036582331934421165,
      "loss": 1.939,
      "step": 98978
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.8666023015975952,
      "learning_rate": 0.00036581932845267443,
      "loss": 2.1462,
      "step": 98979
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.9470264315605164,
      "learning_rate": 0.00036581533754890004,
      "loss": 2.0747,
      "step": 98980
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.7735211253166199,
      "learning_rate": 0.00036581134663288945,
      "loss": 2.0043,
      "step": 98981
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.903718888759613,
      "learning_rate": 0.00036580735570464315,
      "loss": 2.2702,
      "step": 98982
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.8711787462234497,
      "learning_rate": 0.0003658033647641621,
      "loss": 2.115,
      "step": 98983
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.8457559943199158,
      "learning_rate": 0.00036579937381144706,
      "loss": 2.1792,
      "step": 98984
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.895862877368927,
      "learning_rate": 0.0003657953828464985,
      "loss": 2.2927,
      "step": 98985
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.9259741902351379,
      "learning_rate": 0.00036579139186931745,
      "loss": 2.017,
      "step": 98986
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.9595707058906555,
      "learning_rate": 0.0003657874008799045,
      "loss": 1.9996,
      "step": 98987
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.9321489334106445,
      "learning_rate": 0.0003657834098782604,
      "loss": 2.2626,
      "step": 98988
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.8242424130439758,
      "learning_rate": 0.00036577941886438593,
      "loss": 1.9092,
      "step": 98989
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.8209723234176636,
      "learning_rate": 0.0003657754278382819,
      "loss": 2.3734,
      "step": 98990
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.8517667651176453,
      "learning_rate": 0.00036577143679994894,
      "loss": 2.1298,
      "step": 98991
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.8865947723388672,
      "learning_rate": 0.0003657674457493878,
      "loss": 2.0498,
      "step": 98992
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.7748454809188843,
      "learning_rate": 0.00036576345468659926,
      "loss": 1.8154,
      "step": 98993
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.9140937924385071,
      "learning_rate": 0.000365759463611584,
      "loss": 2.2346,
      "step": 98994
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.869598388671875,
      "learning_rate": 0.0003657554725243429,
      "loss": 2.1849,
      "step": 98995
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.8341230750083923,
      "learning_rate": 0.00036575148142487657,
      "loss": 2.2013,
      "step": 98996
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.8979765176773071,
      "learning_rate": 0.00036574749031318587,
      "loss": 2.0053,
      "step": 98997
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.792850136756897,
      "learning_rate": 0.0003657434991892714,
      "loss": 1.8451,
      "step": 98998
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.7778702974319458,
      "learning_rate": 0.0003657395080531339,
      "loss": 2.465,
      "step": 98999
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.8716051578521729,
      "learning_rate": 0.00036573551690477436,
      "loss": 2.073,
      "step": 99000
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.7610414028167725,
      "learning_rate": 0.00036573152574419324,
      "loss": 1.9309,
      "step": 99001
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.7910438776016235,
      "learning_rate": 0.0003657275345713914,
      "loss": 1.9802,
      "step": 99002
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.881375253200531,
      "learning_rate": 0.0003657235433863696,
      "loss": 2.2564,
      "step": 99003
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.7579172849655151,
      "learning_rate": 0.00036571955218912854,
      "loss": 1.9868,
      "step": 99004
    },
    {
      "epoch": 1.29,
      "grad_norm": 1.229479193687439,
      "learning_rate": 0.00036571556097966893,
      "loss": 2.1803,
      "step": 99005
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.9217856526374817,
      "learning_rate": 0.0003657115697579916,
      "loss": 2.19,
      "step": 99006
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.8185182213783264,
      "learning_rate": 0.0003657075785240973,
      "loss": 2.2441,
      "step": 99007
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.8850781321525574,
      "learning_rate": 0.00036570358727798665,
      "loss": 2.309,
      "step": 99008
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.9273954629898071,
      "learning_rate": 0.0003656995960196604,
      "loss": 2.0935,
      "step": 99009
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.8880651593208313,
      "learning_rate": 0.00036569560474911946,
      "loss": 2.1703,
      "step": 99010
    },
    {
      "epoch": 1.29,
      "grad_norm": 1.0674408674240112,
      "learning_rate": 0.0003656916134663644,
      "loss": 2.0168,
      "step": 99011
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.8269085884094238,
      "learning_rate": 0.0003656876221713961,
      "loss": 1.9694,
      "step": 99012
    },
    {
      "epoch": 1.29,
      "grad_norm": 1.2250436544418335,
      "learning_rate": 0.0003656836308642153,
      "loss": 2.1575,
      "step": 99013
    },
    {
      "epoch": 1.29,
      "grad_norm": 1.248329520225525,
      "learning_rate": 0.0003656796395448225,
      "loss": 2.2153,
      "step": 99014
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.8704069256782532,
      "learning_rate": 0.00036567564821321865,
      "loss": 2.1216,
      "step": 99015
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.9557719826698303,
      "learning_rate": 0.00036567165686940455,
      "loss": 2.0941,
      "step": 99016
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.9594807028770447,
      "learning_rate": 0.0003656676655133808,
      "loss": 2.1332,
      "step": 99017
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.9285221099853516,
      "learning_rate": 0.00036566367414514814,
      "loss": 2.1928,
      "step": 99018
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.8397213816642761,
      "learning_rate": 0.0003656596827647075,
      "loss": 2.1632,
      "step": 99019
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.8684588670730591,
      "learning_rate": 0.00036565569137205937,
      "loss": 2.3814,
      "step": 99020
    },
    {
      "epoch": 1.29,
      "grad_norm": 1.0045348405838013,
      "learning_rate": 0.0003656516999672046,
      "loss": 2.2078,
      "step": 99021
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.889397144317627,
      "learning_rate": 0.000365647708550144,
      "loss": 1.8442,
      "step": 99022
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.8081014156341553,
      "learning_rate": 0.00036564371712087816,
      "loss": 2.0716,
      "step": 99023
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.8759844303131104,
      "learning_rate": 0.00036563972567940796,
      "loss": 2.2684,
      "step": 99024
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.9611153602600098,
      "learning_rate": 0.0003656357342257342,
      "loss": 2.2219,
      "step": 99025
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.8737790584564209,
      "learning_rate": 0.00036563174275985736,
      "loss": 2.0816,
      "step": 99026
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.889065146446228,
      "learning_rate": 0.0003656277512817784,
      "loss": 2.3411,
      "step": 99027
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.7832828164100647,
      "learning_rate": 0.0003656237597914981,
      "loss": 2.0837,
      "step": 99028
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.9208701848983765,
      "learning_rate": 0.000365619768289017,
      "loss": 2.1687,
      "step": 99029
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.7902826070785522,
      "learning_rate": 0.00036561577677433595,
      "loss": 1.9788,
      "step": 99030
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.9516963362693787,
      "learning_rate": 0.0003656117852474557,
      "loss": 2.2364,
      "step": 99031
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.8805915713310242,
      "learning_rate": 0.000365607793708377,
      "loss": 1.977,
      "step": 99032
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.9409262537956238,
      "learning_rate": 0.0003656038021571005,
      "loss": 1.9659,
      "step": 99033
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.8931277990341187,
      "learning_rate": 0.00036559981059362714,
      "loss": 2.2413,
      "step": 99034
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.7999675273895264,
      "learning_rate": 0.00036559581901795743,
      "loss": 2.1516,
      "step": 99035
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.8972323536872864,
      "learning_rate": 0.00036559182743009223,
      "loss": 2.1129,
      "step": 99036
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.9094506502151489,
      "learning_rate": 0.00036558783583003236,
      "loss": 2.0666,
      "step": 99037
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.8179619908332825,
      "learning_rate": 0.00036558384421777837,
      "loss": 2.3006,
      "step": 99038
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.8043579459190369,
      "learning_rate": 0.00036557985259333116,
      "loss": 2.0309,
      "step": 99039
    },
    {
      "epoch": 1.29,
      "grad_norm": 1.0183755159378052,
      "learning_rate": 0.00036557586095669145,
      "loss": 1.9856,
      "step": 99040
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.8750817179679871,
      "learning_rate": 0.00036557186930785983,
      "loss": 2.3335,
      "step": 99041
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.9446452260017395,
      "learning_rate": 0.0003655678776468373,
      "loss": 2.192,
      "step": 99042
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.9122574329376221,
      "learning_rate": 0.00036556388597362444,
      "loss": 2.1515,
      "step": 99043
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.9646031856536865,
      "learning_rate": 0.0003655598942882219,
      "loss": 2.0853,
      "step": 99044
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.9144853949546814,
      "learning_rate": 0.0003655559025906306,
      "loss": 2.3448,
      "step": 99045
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.9219053983688354,
      "learning_rate": 0.0003655519108808513,
      "loss": 2.0409,
      "step": 99046
    },
    {
      "epoch": 1.29,
      "grad_norm": 1.157747745513916,
      "learning_rate": 0.0003655479191588845,
      "loss": 2.4906,
      "step": 99047
    },
    {
      "epoch": 1.29,
      "grad_norm": 1.2510383129119873,
      "learning_rate": 0.0003655439274247312,
      "loss": 2.2492,
      "step": 99048
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.8831518292427063,
      "learning_rate": 0.0003655399356783921,
      "loss": 2.3297,
      "step": 99049
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.9471028447151184,
      "learning_rate": 0.00036553594391986784,
      "loss": 2.1425,
      "step": 99050
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.8248763084411621,
      "learning_rate": 0.0003655319521491592,
      "loss": 2.1251,
      "step": 99051
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.9025756120681763,
      "learning_rate": 0.00036552796036626696,
      "loss": 2.2856,
      "step": 99052
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.7959377765655518,
      "learning_rate": 0.0003655239685711918,
      "loss": 2.0009,
      "step": 99053
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.9479884505271912,
      "learning_rate": 0.00036551997676393446,
      "loss": 2.1179,
      "step": 99054
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.8726339936256409,
      "learning_rate": 0.0003655159849444958,
      "loss": 2.3905,
      "step": 99055
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.8576540350914001,
      "learning_rate": 0.0003655119931128765,
      "loss": 2.1182,
      "step": 99056
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.8899893164634705,
      "learning_rate": 0.0003655080012690772,
      "loss": 1.9566,
      "step": 99057
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.810530960559845,
      "learning_rate": 0.0003655040094130987,
      "loss": 2.296,
      "step": 99058
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.9515299797058105,
      "learning_rate": 0.0003655000175449419,
      "loss": 2.2707,
      "step": 99059
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.8763865232467651,
      "learning_rate": 0.00036549602566460736,
      "loss": 2.0554,
      "step": 99060
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.9154310822486877,
      "learning_rate": 0.00036549203377209583,
      "loss": 2.3002,
      "step": 99061
    },
    {
      "epoch": 1.29,
      "grad_norm": 1.0259599685668945,
      "learning_rate": 0.00036548804186740816,
      "loss": 2.3764,
      "step": 99062
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.8483070731163025,
      "learning_rate": 0.00036548404995054496,
      "loss": 2.0378,
      "step": 99063
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.9142945408821106,
      "learning_rate": 0.00036548005802150704,
      "loss": 2.2001,
      "step": 99064
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.9322261214256287,
      "learning_rate": 0.0003654760660802952,
      "loss": 2.195,
      "step": 99065
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.9695561528205872,
      "learning_rate": 0.0003654720741269101,
      "loss": 2.1681,
      "step": 99066
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.8792819976806641,
      "learning_rate": 0.00036546808216135246,
      "loss": 1.9033,
      "step": 99067
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.8263148665428162,
      "learning_rate": 0.0003654640901836231,
      "loss": 2.2317,
      "step": 99068
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.8551109433174133,
      "learning_rate": 0.00036546009819372275,
      "loss": 2.2236,
      "step": 99069
    },
    {
      "epoch": 1.29,
      "grad_norm": 1.0608360767364502,
      "learning_rate": 0.00036545610619165205,
      "loss": 2.3026,
      "step": 99070
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.8666654229164124,
      "learning_rate": 0.00036545211417741193,
      "loss": 2.2444,
      "step": 99071
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.9294285774230957,
      "learning_rate": 0.000365448122151003,
      "loss": 2.261,
      "step": 99072
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.7963178753852844,
      "learning_rate": 0.000365444130112426,
      "loss": 2.0845,
      "step": 99073
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.8312440514564514,
      "learning_rate": 0.0003654401380616818,
      "loss": 2.0718,
      "step": 99074
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.8899110555648804,
      "learning_rate": 0.00036543614599877087,
      "loss": 2.3544,
      "step": 99075
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.9652425050735474,
      "learning_rate": 0.0003654321539236942,
      "loss": 2.3508,
      "step": 99076
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.8339117765426636,
      "learning_rate": 0.00036542816183645244,
      "loss": 2.2463,
      "step": 99077
    },
    {
      "epoch": 1.29,
      "grad_norm": 1.0370433330535889,
      "learning_rate": 0.00036542416973704646,
      "loss": 2.4466,
      "step": 99078
    },
    {
      "epoch": 1.29,
      "grad_norm": 1.1009910106658936,
      "learning_rate": 0.0003654201776254768,
      "loss": 2.0848,
      "step": 99079
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.8979104161262512,
      "learning_rate": 0.0003654161855017443,
      "loss": 2.29,
      "step": 99080
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.8388379812240601,
      "learning_rate": 0.00036541219336584966,
      "loss": 2.1158,
      "step": 99081
    },
    {
      "epoch": 1.29,
      "grad_norm": 1.2199070453643799,
      "learning_rate": 0.0003654082012177938,
      "loss": 2.123,
      "step": 99082
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.7952158451080322,
      "learning_rate": 0.0003654042090575772,
      "loss": 2.0383,
      "step": 99083
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.8505934476852417,
      "learning_rate": 0.0003654002168852007,
      "loss": 2.1981,
      "step": 99084
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.8425386548042297,
      "learning_rate": 0.0003653962247006651,
      "loss": 2.0862,
      "step": 99085
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.9215219020843506,
      "learning_rate": 0.00036539223250397113,
      "loss": 2.3293,
      "step": 99086
    },
    {
      "epoch": 1.29,
      "grad_norm": 1.0683255195617676,
      "learning_rate": 0.0003653882402951195,
      "loss": 2.2787,
      "step": 99087
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.8551444411277771,
      "learning_rate": 0.0003653842480741109,
      "loss": 1.9945,
      "step": 99088
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.8971149325370789,
      "learning_rate": 0.00036538025584094625,
      "loss": 2.2859,
      "step": 99089
    },
    {
      "epoch": 1.29,
      "grad_norm": 1.2087804079055786,
      "learning_rate": 0.00036537626359562606,
      "loss": 2.3068,
      "step": 99090
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.9852071404457092,
      "learning_rate": 0.00036537227133815126,
      "loss": 2.0667,
      "step": 99091
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.9322394728660583,
      "learning_rate": 0.00036536827906852255,
      "loss": 2.3467,
      "step": 99092
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.9255576729774475,
      "learning_rate": 0.0003653642867867406,
      "loss": 2.2927,
      "step": 99093
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.9166461229324341,
      "learning_rate": 0.0003653602944928062,
      "loss": 2.1834,
      "step": 99094
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.8758575320243835,
      "learning_rate": 0.0003653563021867201,
      "loss": 2.2215,
      "step": 99095
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.8973143696784973,
      "learning_rate": 0.00036535230986848297,
      "loss": 1.9039,
      "step": 99096
    },
    {
      "epoch": 1.29,
      "grad_norm": 1.1940572261810303,
      "learning_rate": 0.00036534831753809563,
      "loss": 2.0863,
      "step": 99097
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.9255607724189758,
      "learning_rate": 0.0003653443251955589,
      "loss": 2.2428,
      "step": 99098
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.8951156735420227,
      "learning_rate": 0.0003653403328408733,
      "loss": 2.172,
      "step": 99099
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.901124894618988,
      "learning_rate": 0.00036533634047403976,
      "loss": 2.103,
      "step": 99100
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.9069522619247437,
      "learning_rate": 0.00036533234809505897,
      "loss": 2.0902,
      "step": 99101
    },
    {
      "epoch": 1.29,
      "grad_norm": 1.0453693866729736,
      "learning_rate": 0.00036532835570393164,
      "loss": 2.238,
      "step": 99102
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.8454744815826416,
      "learning_rate": 0.00036532436330065853,
      "loss": 2.2126,
      "step": 99103
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.7685173749923706,
      "learning_rate": 0.00036532037088524047,
      "loss": 2.0208,
      "step": 99104
    },
    {
      "epoch": 1.29,
      "grad_norm": 1.0129907131195068,
      "learning_rate": 0.000365316378457678,
      "loss": 2.3054,
      "step": 99105
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.8748045563697815,
      "learning_rate": 0.00036531238601797205,
      "loss": 2.1008,
      "step": 99106
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.8199349641799927,
      "learning_rate": 0.00036530839356612333,
      "loss": 2.2597,
      "step": 99107
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.932339072227478,
      "learning_rate": 0.00036530440110213246,
      "loss": 2.0285,
      "step": 99108
    },
    {
      "epoch": 1.29,
      "grad_norm": 1.0030407905578613,
      "learning_rate": 0.00036530040862600027,
      "loss": 2.2309,
      "step": 99109
    },
    {
      "epoch": 1.29,
      "grad_norm": 1.055939793586731,
      "learning_rate": 0.00036529641613772756,
      "loss": 2.199,
      "step": 99110
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.8943917751312256,
      "learning_rate": 0.00036529242363731503,
      "loss": 2.204,
      "step": 99111
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.9316608309745789,
      "learning_rate": 0.0003652884311247633,
      "loss": 2.2135,
      "step": 99112
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.8735337853431702,
      "learning_rate": 0.0003652844386000733,
      "loss": 2.4808,
      "step": 99113
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.8843289613723755,
      "learning_rate": 0.0003652804460632457,
      "loss": 2.0306,
      "step": 99114
    },
    {
      "epoch": 1.29,
      "grad_norm": 1.0636889934539795,
      "learning_rate": 0.0003652764535142812,
      "loss": 2.105,
      "step": 99115
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.8530943989753723,
      "learning_rate": 0.0003652724609531806,
      "loss": 2.0961,
      "step": 99116
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.8646048903465271,
      "learning_rate": 0.0003652684683799446,
      "loss": 2.1221,
      "step": 99117
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.8345473408699036,
      "learning_rate": 0.00036526447579457395,
      "loss": 2.1194,
      "step": 99118
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.899139940738678,
      "learning_rate": 0.00036526048319706944,
      "loss": 1.9525,
      "step": 99119
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.9086064696311951,
      "learning_rate": 0.0003652564905874317,
      "loss": 2.1607,
      "step": 99120
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.9716189503669739,
      "learning_rate": 0.00036525249796566154,
      "loss": 2.2007,
      "step": 99121
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.856280505657196,
      "learning_rate": 0.0003652485053317598,
      "loss": 2.3146,
      "step": 99122
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.8675520420074463,
      "learning_rate": 0.0003652445126857271,
      "loss": 2.0962,
      "step": 99123
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.9529899954795837,
      "learning_rate": 0.00036524052002756416,
      "loss": 2.1277,
      "step": 99124
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.9682608842849731,
      "learning_rate": 0.00036523652735727185,
      "loss": 2.2481,
      "step": 99125
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.8935679793357849,
      "learning_rate": 0.00036523253467485084,
      "loss": 2.1579,
      "step": 99126
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.8890315294265747,
      "learning_rate": 0.0003652285419803019,
      "loss": 2.1479,
      "step": 99127
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.8885127902030945,
      "learning_rate": 0.0003652245492736256,
      "loss": 2.1129,
      "step": 99128
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.8528231382369995,
      "learning_rate": 0.00036522055655482293,
      "loss": 2.3342,
      "step": 99129
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.8623372316360474,
      "learning_rate": 0.00036521656382389455,
      "loss": 2.1635,
      "step": 99130
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.8497438430786133,
      "learning_rate": 0.0003652125710808411,
      "loss": 2.247,
      "step": 99131
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.9178338050842285,
      "learning_rate": 0.00036520857832566346,
      "loss": 2.2013,
      "step": 99132
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.8476186990737915,
      "learning_rate": 0.00036520458555836226,
      "loss": 1.8224,
      "step": 99133
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.9269756078720093,
      "learning_rate": 0.00036520059277893833,
      "loss": 2.1642,
      "step": 99134
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.8964713215827942,
      "learning_rate": 0.0003651965999873924,
      "loss": 2.155,
      "step": 99135
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.86310875415802,
      "learning_rate": 0.00036519260718372514,
      "loss": 2.2292,
      "step": 99136
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.7969814538955688,
      "learning_rate": 0.00036518861436793734,
      "loss": 2.0305,
      "step": 99137
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.7898100018501282,
      "learning_rate": 0.00036518462154002984,
      "loss": 2.0447,
      "step": 99138
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.8417295813560486,
      "learning_rate": 0.00036518062870000324,
      "loss": 1.957,
      "step": 99139
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.812633752822876,
      "learning_rate": 0.00036517663584785835,
      "loss": 2.1405,
      "step": 99140
    },
    {
      "epoch": 1.29,
      "grad_norm": 1.0255054235458374,
      "learning_rate": 0.0003651726429835958,
      "loss": 2.106,
      "step": 99141
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.9327936172485352,
      "learning_rate": 0.0003651686501072165,
      "loss": 2.4528,
      "step": 99142
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.9312446713447571,
      "learning_rate": 0.00036516465721872114,
      "loss": 2.1176,
      "step": 99143
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.8209248781204224,
      "learning_rate": 0.00036516066431811046,
      "loss": 1.9825,
      "step": 99144
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.806479811668396,
      "learning_rate": 0.00036515667140538506,
      "loss": 2.1376,
      "step": 99145
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.8533922433853149,
      "learning_rate": 0.0003651526784805459,
      "loss": 2.1258,
      "step": 99146
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.888898491859436,
      "learning_rate": 0.0003651486855435936,
      "loss": 2.0245,
      "step": 99147
    },
    {
      "epoch": 1.29,
      "grad_norm": 1.0578755140304565,
      "learning_rate": 0.0003651446925945289,
      "loss": 2.2009,
      "step": 99148
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.809263288974762,
      "learning_rate": 0.00036514069963335257,
      "loss": 2.1571,
      "step": 99149
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.8545551300048828,
      "learning_rate": 0.00036513670666006543,
      "loss": 2.2105,
      "step": 99150
    },
    {
      "epoch": 1.29,
      "grad_norm": 1.0859988927841187,
      "learning_rate": 0.0003651327136746681,
      "loss": 2.1838,
      "step": 99151
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.8592785000801086,
      "learning_rate": 0.00036512872067716133,
      "loss": 2.1794,
      "step": 99152
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.8740987777709961,
      "learning_rate": 0.000365124727667546,
      "loss": 2.1432,
      "step": 99153
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.9890936017036438,
      "learning_rate": 0.0003651207346458227,
      "loss": 1.978,
      "step": 99154
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.9338659048080444,
      "learning_rate": 0.00036511674161199216,
      "loss": 2.2806,
      "step": 99155
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.8388867974281311,
      "learning_rate": 0.00036511274856605534,
      "loss": 2.1741,
      "step": 99156
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.8546507954597473,
      "learning_rate": 0.0003651087555080127,
      "loss": 2.0981,
      "step": 99157
    },
    {
      "epoch": 1.29,
      "grad_norm": 1.0127201080322266,
      "learning_rate": 0.0003651047624378651,
      "loss": 2.0457,
      "step": 99158
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.9564342498779297,
      "learning_rate": 0.00036510076935561343,
      "loss": 2.1117,
      "step": 99159
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.8550097942352295,
      "learning_rate": 0.00036509677626125816,
      "loss": 2.1839,
      "step": 99160
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.8925451040267944,
      "learning_rate": 0.0003650927831548003,
      "loss": 2.1994,
      "step": 99161
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.9103229641914368,
      "learning_rate": 0.0003650887900362403,
      "loss": 2.282,
      "step": 99162
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.8726749420166016,
      "learning_rate": 0.00036508479690557923,
      "loss": 2.2088,
      "step": 99163
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.8672164082527161,
      "learning_rate": 0.0003650808037628176,
      "loss": 1.94,
      "step": 99164
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.9327970743179321,
      "learning_rate": 0.0003650768106079562,
      "loss": 2.2782,
      "step": 99165
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.8437781929969788,
      "learning_rate": 0.00036507281744099584,
      "loss": 2.1899,
      "step": 99166
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.8076496720314026,
      "learning_rate": 0.00036506882426193716,
      "loss": 2.2007,
      "step": 99167
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.799623966217041,
      "learning_rate": 0.00036506483107078095,
      "loss": 2.24,
      "step": 99168
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.8529192209243774,
      "learning_rate": 0.00036506083786752797,
      "loss": 1.965,
      "step": 99169
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.8409850597381592,
      "learning_rate": 0.000365056844652179,
      "loss": 2.2942,
      "step": 99170
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.8053392767906189,
      "learning_rate": 0.0003650528514247347,
      "loss": 2.1367,
      "step": 99171
    },
    {
      "epoch": 1.29,
      "grad_norm": 1.0341397523880005,
      "learning_rate": 0.00036504885818519585,
      "loss": 2.0672,
      "step": 99172
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.8890342712402344,
      "learning_rate": 0.0003650448649335631,
      "loss": 2.272,
      "step": 99173
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.9002478718757629,
      "learning_rate": 0.00036504087166983745,
      "loss": 2.144,
      "step": 99174
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.902951717376709,
      "learning_rate": 0.00036503687839401937,
      "loss": 2.1875,
      "step": 99175
    },
    {
      "epoch": 1.29,
      "grad_norm": 1.0190916061401367,
      "learning_rate": 0.00036503288510610977,
      "loss": 2.269,
      "step": 99176
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.8017322421073914,
      "learning_rate": 0.00036502889180610925,
      "loss": 2.0383,
      "step": 99177
    },
    {
      "epoch": 1.29,
      "grad_norm": 1.0114563703536987,
      "learning_rate": 0.0003650248984940187,
      "loss": 2.1388,
      "step": 99178
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.9668562412261963,
      "learning_rate": 0.00036502090516983874,
      "loss": 2.3816,
      "step": 99179
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.8185790777206421,
      "learning_rate": 0.0003650169118335703,
      "loss": 2.2319,
      "step": 99180
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.8399853706359863,
      "learning_rate": 0.00036501291848521384,
      "loss": 2.0537,
      "step": 99181
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.7939586639404297,
      "learning_rate": 0.00036500892512477025,
      "loss": 1.9715,
      "step": 99182
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.8319969177246094,
      "learning_rate": 0.0003650049317522404,
      "loss": 1.994,
      "step": 99183
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.8213514089584351,
      "learning_rate": 0.0003650009383676248,
      "loss": 1.9408,
      "step": 99184
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.8382408022880554,
      "learning_rate": 0.0003649969449709243,
      "loss": 2.1992,
      "step": 99185
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.8440523147583008,
      "learning_rate": 0.0003649929515621397,
      "loss": 2.2206,
      "step": 99186
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.8397672176361084,
      "learning_rate": 0.00036498895814127163,
      "loss": 2.1617,
      "step": 99187
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.7982668876647949,
      "learning_rate": 0.0003649849647083209,
      "loss": 2.1507,
      "step": 99188
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.8314505219459534,
      "learning_rate": 0.0003649809712632882,
      "loss": 2.3551,
      "step": 99189
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.950178325176239,
      "learning_rate": 0.0003649769778061744,
      "loss": 2.1222,
      "step": 99190
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.8038967847824097,
      "learning_rate": 0.00036497298433698006,
      "loss": 2.2143,
      "step": 99191
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.8825164437294006,
      "learning_rate": 0.0003649689908557061,
      "loss": 2.3131,
      "step": 99192
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.7997260689735413,
      "learning_rate": 0.00036496499736235314,
      "loss": 2.344,
      "step": 99193
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.7940142750740051,
      "learning_rate": 0.000364961003856922,
      "loss": 2.2778,
      "step": 99194
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.9003022909164429,
      "learning_rate": 0.0003649570103394133,
      "loss": 2.2335,
      "step": 99195
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.8559842109680176,
      "learning_rate": 0.00036495301680982794,
      "loss": 2.0261,
      "step": 99196
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.8016697764396667,
      "learning_rate": 0.0003649490232681665,
      "loss": 2.1136,
      "step": 99197
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.9623448848724365,
      "learning_rate": 0.0003649450297144299,
      "loss": 2.1052,
      "step": 99198
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.9310051798820496,
      "learning_rate": 0.0003649410361486188,
      "loss": 2.0483,
      "step": 99199
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.9316121339797974,
      "learning_rate": 0.00036493704257073385,
      "loss": 2.2385,
      "step": 99200
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.8734175562858582,
      "learning_rate": 0.00036493304898077593,
      "loss": 2.0782,
      "step": 99201
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.9236263036727905,
      "learning_rate": 0.0003649290553787458,
      "loss": 1.9483,
      "step": 99202
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.8526853322982788,
      "learning_rate": 0.00036492506176464407,
      "loss": 2.0889,
      "step": 99203
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.9411941170692444,
      "learning_rate": 0.00036492106813847147,
      "loss": 2.1823,
      "step": 99204
    },
    {
      "epoch": 1.29,
      "grad_norm": 1.0262609720230103,
      "learning_rate": 0.00036491707450022893,
      "loss": 2.285,
      "step": 99205
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.8856165409088135,
      "learning_rate": 0.000364913080849917,
      "loss": 2.1252,
      "step": 99206
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.8812627792358398,
      "learning_rate": 0.00036490908718753655,
      "loss": 2.214,
      "step": 99207
    },
    {
      "epoch": 1.29,
      "grad_norm": 1.012107253074646,
      "learning_rate": 0.0003649050935130882,
      "loss": 2.115,
      "step": 99208
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.7959034442901611,
      "learning_rate": 0.0003649010998265729,
      "loss": 2.2135,
      "step": 99209
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.8655562996864319,
      "learning_rate": 0.0003648971061279912,
      "loss": 2.0996,
      "step": 99210
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.8904680013656616,
      "learning_rate": 0.00036489311241734386,
      "loss": 2.1605,
      "step": 99211
    },
    {
      "epoch": 1.29,
      "grad_norm": 1.0034630298614502,
      "learning_rate": 0.00036488911869463173,
      "loss": 2.227,
      "step": 99212
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.939708411693573,
      "learning_rate": 0.00036488512495985547,
      "loss": 2.1212,
      "step": 99213
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.9119586944580078,
      "learning_rate": 0.0003648811312130158,
      "loss": 2.1437,
      "step": 99214
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.9048295617103577,
      "learning_rate": 0.00036487713745411357,
      "loss": 2.0197,
      "step": 99215
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.8160120248794556,
      "learning_rate": 0.00036487314368314935,
      "loss": 2.3719,
      "step": 99216
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.9361015558242798,
      "learning_rate": 0.00036486914990012414,
      "loss": 2.2623,
      "step": 99217
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.9254422783851624,
      "learning_rate": 0.0003648651561050384,
      "loss": 1.9969,
      "step": 99218
    },
    {
      "epoch": 1.29,
      "grad_norm": 1.0060197114944458,
      "learning_rate": 0.000364861162297893,
      "loss": 2.2184,
      "step": 99219
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.8846530914306641,
      "learning_rate": 0.0003648571684786888,
      "loss": 2.1497,
      "step": 99220
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.8088858127593994,
      "learning_rate": 0.00036485317464742635,
      "loss": 2.2871,
      "step": 99221
    },
    {
      "epoch": 1.29,
      "grad_norm": 1.1795697212219238,
      "learning_rate": 0.00036484918080410643,
      "loss": 2.2886,
      "step": 99222
    },
    {
      "epoch": 1.29,
      "grad_norm": 1.0501928329467773,
      "learning_rate": 0.00036484518694872997,
      "loss": 2.2467,
      "step": 99223
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.8768154978752136,
      "learning_rate": 0.0003648411930812974,
      "loss": 1.8958,
      "step": 99224
    },
    {
      "epoch": 1.29,
      "grad_norm": 1.0878149271011353,
      "learning_rate": 0.00036483719920180974,
      "loss": 2.0504,
      "step": 99225
    },
    {
      "epoch": 1.29,
      "grad_norm": 1.2843921184539795,
      "learning_rate": 0.00036483320531026755,
      "loss": 2.2807,
      "step": 99226
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.988578200340271,
      "learning_rate": 0.00036482921140667174,
      "loss": 2.1936,
      "step": 99227
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.8642789125442505,
      "learning_rate": 0.0003648252174910228,
      "loss": 2.2705,
      "step": 99228
    },
    {
      "epoch": 1.29,
      "grad_norm": 1.0237425565719604,
      "learning_rate": 0.00036482122356332175,
      "loss": 2.2184,
      "step": 99229
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.9731934666633606,
      "learning_rate": 0.00036481722962356916,
      "loss": 1.9931,
      "step": 99230
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.903598427772522,
      "learning_rate": 0.0003648132356717658,
      "loss": 2.2541,
      "step": 99231
    },
    {
      "epoch": 1.29,
      "grad_norm": 1.0990206003189087,
      "learning_rate": 0.0003648092417079126,
      "loss": 2.2004,
      "step": 99232
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.9221460819244385,
      "learning_rate": 0.00036480524773201,
      "loss": 2.0541,
      "step": 99233
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.8390182256698608,
      "learning_rate": 0.00036480125374405885,
      "loss": 2.0866,
      "step": 99234
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.8657435774803162,
      "learning_rate": 0.00036479725974406,
      "loss": 2.0542,
      "step": 99235
    },
    {
      "epoch": 1.29,
      "grad_norm": 1.0350615978240967,
      "learning_rate": 0.00036479326573201406,
      "loss": 2.1381,
      "step": 99236
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.7725866436958313,
      "learning_rate": 0.00036478927170792186,
      "loss": 1.8582,
      "step": 99237
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.9882652163505554,
      "learning_rate": 0.0003647852776717842,
      "loss": 2.2405,
      "step": 99238
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.9582182765007019,
      "learning_rate": 0.00036478128362360155,
      "loss": 2.1038,
      "step": 99239
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.9763112664222717,
      "learning_rate": 0.0003647772895633749,
      "loss": 2.198,
      "step": 99240
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.8770830035209656,
      "learning_rate": 0.00036477329549110504,
      "loss": 2.345,
      "step": 99241
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.9936624765396118,
      "learning_rate": 0.00036476930140679246,
      "loss": 1.9986,
      "step": 99242
    },
    {
      "epoch": 1.29,
      "grad_norm": 1.1102807521820068,
      "learning_rate": 0.0003647653073104381,
      "loss": 2.1164,
      "step": 99243
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.8712990880012512,
      "learning_rate": 0.00036476131320204266,
      "loss": 2.0566,
      "step": 99244
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.8164505362510681,
      "learning_rate": 0.0003647573190816069,
      "loss": 2.0089,
      "step": 99245
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.9803053736686707,
      "learning_rate": 0.00036475332494913145,
      "loss": 2.1732,
      "step": 99246
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.9052250385284424,
      "learning_rate": 0.0003647493308046172,
      "loss": 2.2715,
      "step": 99247
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.8484998345375061,
      "learning_rate": 0.0003647453366480648,
      "loss": 2.2035,
      "step": 99248
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.8424698114395142,
      "learning_rate": 0.00036474134247947503,
      "loss": 2.2238,
      "step": 99249
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.8429019451141357,
      "learning_rate": 0.0003647373482988486,
      "loss": 2.3142,
      "step": 99250
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.845093846321106,
      "learning_rate": 0.0003647333541061863,
      "loss": 2.0128,
      "step": 99251
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.9381037354469299,
      "learning_rate": 0.0003647293599014888,
      "loss": 2.3941,
      "step": 99252
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.8985596299171448,
      "learning_rate": 0.00036472536568475693,
      "loss": 2.0897,
      "step": 99253
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.8737955093383789,
      "learning_rate": 0.0003647213714559913,
      "loss": 2.1869,
      "step": 99254
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.8903859257698059,
      "learning_rate": 0.00036471737721519297,
      "loss": 2.1449,
      "step": 99255
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.9244731068611145,
      "learning_rate": 0.00036471338296236225,
      "loss": 2.293,
      "step": 99256
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.9230650663375854,
      "learning_rate": 0.00036470938869750014,
      "loss": 2.1692,
      "step": 99257
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.932752788066864,
      "learning_rate": 0.00036470539442060745,
      "loss": 2.1583,
      "step": 99258
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.8682935237884521,
      "learning_rate": 0.0003647014001316847,
      "loss": 2.0869,
      "step": 99259
    },
    {
      "epoch": 1.29,
      "grad_norm": 1.1429027318954468,
      "learning_rate": 0.0003646974058307327,
      "loss": 1.9272,
      "step": 99260
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.8026271462440491,
      "learning_rate": 0.00036469341151775225,
      "loss": 2.0797,
      "step": 99261
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.9947972297668457,
      "learning_rate": 0.0003646894171927442,
      "loss": 2.2307,
      "step": 99262
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.9665388464927673,
      "learning_rate": 0.00036468542285570906,
      "loss": 2.1842,
      "step": 99263
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.9266566038131714,
      "learning_rate": 0.0003646814285066476,
      "loss": 2.1203,
      "step": 99264
    },
    {
      "epoch": 1.29,
      "grad_norm": 1.0149041414260864,
      "learning_rate": 0.0003646774341455608,
      "loss": 2.0207,
      "step": 99265
    },
    {
      "epoch": 1.29,
      "grad_norm": 1.0094943046569824,
      "learning_rate": 0.0003646734397724492,
      "loss": 2.3283,
      "step": 99266
    },
    {
      "epoch": 1.29,
      "grad_norm": 1.1261311769485474,
      "learning_rate": 0.00036466944538731354,
      "loss": 2.0119,
      "step": 99267
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.7969895601272583,
      "learning_rate": 0.0003646654509901546,
      "loss": 2.2701,
      "step": 99268
    },
    {
      "epoch": 1.29,
      "grad_norm": 1.0612539052963257,
      "learning_rate": 0.00036466145658097324,
      "loss": 2.1118,
      "step": 99269
    },
    {
      "epoch": 1.29,
      "grad_norm": 1.1954056024551392,
      "learning_rate": 0.00036465746215976993,
      "loss": 2.1118,
      "step": 99270
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.8362734913825989,
      "learning_rate": 0.0003646534677265457,
      "loss": 1.9175,
      "step": 99271
    },
    {
      "epoch": 1.29,
      "grad_norm": 1.0623619556427002,
      "learning_rate": 0.0003646494732813012,
      "loss": 2.1851,
      "step": 99272
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.9844880700111389,
      "learning_rate": 0.00036464547882403704,
      "loss": 1.9356,
      "step": 99273
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.8660603165626526,
      "learning_rate": 0.0003646414843547541,
      "loss": 1.9805,
      "step": 99274
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.8234456181526184,
      "learning_rate": 0.00036463748987345307,
      "loss": 1.829,
      "step": 99275
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.960309624671936,
      "learning_rate": 0.0003646334953801347,
      "loss": 2.2071,
      "step": 99276
    },
    {
      "epoch": 1.29,
      "grad_norm": 1.056477427482605,
      "learning_rate": 0.0003646295008747998,
      "loss": 2.2538,
      "step": 99277
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.8655970096588135,
      "learning_rate": 0.00036462550635744907,
      "loss": 2.2455,
      "step": 99278
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.8059938549995422,
      "learning_rate": 0.00036462151182808315,
      "loss": 1.9968,
      "step": 99279
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.9079359173774719,
      "learning_rate": 0.0003646175172867029,
      "loss": 2.3234,
      "step": 99280
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.9104678630828857,
      "learning_rate": 0.0003646135227333091,
      "loss": 2.1195,
      "step": 99281
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.9209959506988525,
      "learning_rate": 0.0003646095281679024,
      "loss": 2.1709,
      "step": 99282
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.8698281049728394,
      "learning_rate": 0.00036460553359048353,
      "loss": 2.2528,
      "step": 99283
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.9063788652420044,
      "learning_rate": 0.0003646015390010533,
      "loss": 2.0114,
      "step": 99284
    },
    {
      "epoch": 1.29,
      "grad_norm": 1.1660223007202148,
      "learning_rate": 0.0003645975443996124,
      "loss": 2.069,
      "step": 99285
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.8545970916748047,
      "learning_rate": 0.0003645935497861616,
      "loss": 2.2274,
      "step": 99286
    },
    {
      "epoch": 1.29,
      "grad_norm": 1.112492322921753,
      "learning_rate": 0.0003645895551607017,
      "loss": 2.1568,
      "step": 99287
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.9651398658752441,
      "learning_rate": 0.0003645855605232333,
      "loss": 2.1236,
      "step": 99288
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.9106976389884949,
      "learning_rate": 0.00036458156587375725,
      "loss": 2.0713,
      "step": 99289
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.9186147451400757,
      "learning_rate": 0.0003645775712122743,
      "loss": 2.2926,
      "step": 99290
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.9537140727043152,
      "learning_rate": 0.00036457357653878517,
      "loss": 2.2047,
      "step": 99291
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.8981457352638245,
      "learning_rate": 0.0003645695818532905,
      "loss": 2.022,
      "step": 99292
    },
    {
      "epoch": 1.29,
      "grad_norm": 1.056148648262024,
      "learning_rate": 0.0003645655871557913,
      "loss": 2.1323,
      "step": 99293
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.8437434434890747,
      "learning_rate": 0.00036456159244628795,
      "loss": 2.0278,
      "step": 99294
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.9010863304138184,
      "learning_rate": 0.00036455759772478145,
      "loss": 2.154,
      "step": 99295
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.8346824049949646,
      "learning_rate": 0.0003645536029912725,
      "loss": 2.1296,
      "step": 99296
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.9276286363601685,
      "learning_rate": 0.0003645496082457618,
      "loss": 2.1017,
      "step": 99297
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.8359804153442383,
      "learning_rate": 0.0003645456134882501,
      "loss": 2.1926,
      "step": 99298
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.9121800661087036,
      "learning_rate": 0.00036454161871873826,
      "loss": 2.1604,
      "step": 99299
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.8181336522102356,
      "learning_rate": 0.0003645376239372267,
      "loss": 1.9561,
      "step": 99300
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.8089604377746582,
      "learning_rate": 0.0003645336291437165,
      "loss": 2.1512,
      "step": 99301
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.8341910243034363,
      "learning_rate": 0.00036452963433820836,
      "loss": 2.1665,
      "step": 99302
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.8318922519683838,
      "learning_rate": 0.0003645256395207028,
      "loss": 2.0964,
      "step": 99303
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.8093885779380798,
      "learning_rate": 0.00036452164469120074,
      "loss": 1.9263,
      "step": 99304
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.8838644027709961,
      "learning_rate": 0.00036451764984970293,
      "loss": 2.1057,
      "step": 99305
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.8919748663902283,
      "learning_rate": 0.00036451365499621006,
      "loss": 2.1848,
      "step": 99306
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.786636233329773,
      "learning_rate": 0.0003645096601307229,
      "loss": 2.3337,
      "step": 99307
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.8650479912757874,
      "learning_rate": 0.0003645056652532422,
      "loss": 2.2682,
      "step": 99308
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.9681350588798523,
      "learning_rate": 0.00036450167036376855,
      "loss": 2.0143,
      "step": 99309
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.8599870800971985,
      "learning_rate": 0.00036449767546230293,
      "loss": 2.1361,
      "step": 99310
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.851497232913971,
      "learning_rate": 0.000364493680548846,
      "loss": 2.0726,
      "step": 99311
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.8438341021537781,
      "learning_rate": 0.0003644896856233984,
      "loss": 2.1537,
      "step": 99312
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.8693965673446655,
      "learning_rate": 0.0003644856906859609,
      "loss": 1.9921,
      "step": 99313
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.8337506055831909,
      "learning_rate": 0.00036448169573653446,
      "loss": 2.0189,
      "step": 99314
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.9229089021682739,
      "learning_rate": 0.0003644777007751195,
      "loss": 2.2055,
      "step": 99315
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.8810128569602966,
      "learning_rate": 0.000364473705801717,
      "loss": 2.085,
      "step": 99316
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.9417696595191956,
      "learning_rate": 0.00036446971081632764,
      "loss": 1.9794,
      "step": 99317
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.8390821814537048,
      "learning_rate": 0.0003644657158189521,
      "loss": 1.9296,
      "step": 99318
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.8699348568916321,
      "learning_rate": 0.00036446172080959117,
      "loss": 2.2172,
      "step": 99319
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.8758577704429626,
      "learning_rate": 0.00036445772578824564,
      "loss": 2.0952,
      "step": 99320
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.824810802936554,
      "learning_rate": 0.00036445373075491604,
      "loss": 2.192,
      "step": 99321
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.8257240653038025,
      "learning_rate": 0.0003644497357096034,
      "loss": 2.1272,
      "step": 99322
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.8216051459312439,
      "learning_rate": 0.00036444574065230834,
      "loss": 1.9402,
      "step": 99323
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.7915748953819275,
      "learning_rate": 0.00036444174558303164,
      "loss": 2.1349,
      "step": 99324
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.8773565292358398,
      "learning_rate": 0.0003644377505017739,
      "loss": 2.1029,
      "step": 99325
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.8783555626869202,
      "learning_rate": 0.000364433755408536,
      "loss": 2.116,
      "step": 99326
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.8575559854507446,
      "learning_rate": 0.0003644297603033187,
      "loss": 2.0075,
      "step": 99327
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.8853194117546082,
      "learning_rate": 0.0003644257651861226,
      "loss": 2.0857,
      "step": 99328
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.8048029541969299,
      "learning_rate": 0.0003644217700569486,
      "loss": 1.9294,
      "step": 99329
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.8535996079444885,
      "learning_rate": 0.00036441777491579746,
      "loss": 2.16,
      "step": 99330
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.8635879755020142,
      "learning_rate": 0.0003644137797626697,
      "loss": 2.2537,
      "step": 99331
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.9459025263786316,
      "learning_rate": 0.0003644097845975662,
      "loss": 2.1865,
      "step": 99332
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.8306983709335327,
      "learning_rate": 0.0003644057894204878,
      "loss": 1.9895,
      "step": 99333
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.8051620125770569,
      "learning_rate": 0.0003644017942314351,
      "loss": 2.1375,
      "step": 99334
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.8635170459747314,
      "learning_rate": 0.00036439779903040887,
      "loss": 2.1028,
      "step": 99335
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.893064558506012,
      "learning_rate": 0.00036439380381740995,
      "loss": 2.1543,
      "step": 99336
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.8437791466712952,
      "learning_rate": 0.00036438980859243895,
      "loss": 2.094,
      "step": 99337
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.8395819664001465,
      "learning_rate": 0.00036438581335549667,
      "loss": 2.0578,
      "step": 99338
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.9580256342887878,
      "learning_rate": 0.0003643818181065839,
      "loss": 2.2779,
      "step": 99339
    },
    {
      "epoch": 1.29,
      "grad_norm": 1.1320834159851074,
      "learning_rate": 0.00036437782284570123,
      "loss": 2.0706,
      "step": 99340
    },
    {
      "epoch": 1.29,
      "grad_norm": 1.0233104228973389,
      "learning_rate": 0.0003643738275728495,
      "loss": 1.9126,
      "step": 99341
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.8894739747047424,
      "learning_rate": 0.0003643698322880296,
      "loss": 2.2541,
      "step": 99342
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.9358533024787903,
      "learning_rate": 0.000364365836991242,
      "loss": 2.2376,
      "step": 99343
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.7532934546470642,
      "learning_rate": 0.00036436184168248765,
      "loss": 2.1411,
      "step": 99344
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.8033161759376526,
      "learning_rate": 0.0003643578463617672,
      "loss": 2.1908,
      "step": 99345
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.8504988551139832,
      "learning_rate": 0.0003643538510290815,
      "loss": 2.1271,
      "step": 99346
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.8731110692024231,
      "learning_rate": 0.00036434985568443106,
      "loss": 2.327,
      "step": 99347
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.9698110818862915,
      "learning_rate": 0.0003643458603278168,
      "loss": 2.1273,
      "step": 99348
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.9033982157707214,
      "learning_rate": 0.00036434186495923945,
      "loss": 2.1847,
      "step": 99349
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.8921182751655579,
      "learning_rate": 0.0003643378695786998,
      "loss": 2.0932,
      "step": 99350
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.9053018689155579,
      "learning_rate": 0.0003643338741861984,
      "loss": 2.1214,
      "step": 99351
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.9886526465415955,
      "learning_rate": 0.00036432987878173617,
      "loss": 2.1203,
      "step": 99352
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.9336824417114258,
      "learning_rate": 0.0003643258833653138,
      "loss": 2.4766,
      "step": 99353
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.9478357434272766,
      "learning_rate": 0.0003643218879369321,
      "loss": 2.0659,
      "step": 99354
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.9305179119110107,
      "learning_rate": 0.0003643178924965917,
      "loss": 2.1992,
      "step": 99355
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.9292802214622498,
      "learning_rate": 0.00036431389704429335,
      "loss": 2.1095,
      "step": 99356
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.8813247680664062,
      "learning_rate": 0.0003643099015800379,
      "loss": 2.1604,
      "step": 99357
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.966651201248169,
      "learning_rate": 0.00036430590610382597,
      "loss": 2.385,
      "step": 99358
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.9141510725021362,
      "learning_rate": 0.0003643019106156584,
      "loss": 2.1225,
      "step": 99359
    },
    {
      "epoch": 1.29,
      "grad_norm": 1.0531151294708252,
      "learning_rate": 0.0003642979151155359,
      "loss": 2.1979,
      "step": 99360
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.9075552225112915,
      "learning_rate": 0.00036429391960345915,
      "loss": 2.1231,
      "step": 99361
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.8303531408309937,
      "learning_rate": 0.000364289924079429,
      "loss": 2.2885,
      "step": 99362
    },
    {
      "epoch": 1.29,
      "grad_norm": 1.0832781791687012,
      "learning_rate": 0.0003642859285434461,
      "loss": 1.9627,
      "step": 99363
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.8959673643112183,
      "learning_rate": 0.0003642819329955112,
      "loss": 2.2266,
      "step": 99364
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.8787301182746887,
      "learning_rate": 0.00036427793743562515,
      "loss": 2.23,
      "step": 99365
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.9941489696502686,
      "learning_rate": 0.0003642739418637886,
      "loss": 2.2063,
      "step": 99366
    },
    {
      "epoch": 1.29,
      "grad_norm": 1.2924801111221313,
      "learning_rate": 0.0003642699462800023,
      "loss": 2.0921,
      "step": 99367
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.8707271814346313,
      "learning_rate": 0.00036426595068426694,
      "loss": 2.1244,
      "step": 99368
    },
    {
      "epoch": 1.29,
      "grad_norm": 1.204058051109314,
      "learning_rate": 0.00036426195507658347,
      "loss": 2.199,
      "step": 99369
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.9517562985420227,
      "learning_rate": 0.0003642579594569524,
      "loss": 2.2445,
      "step": 99370
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.991134762763977,
      "learning_rate": 0.00036425396382537453,
      "loss": 2.3649,
      "step": 99371
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.9533840417861938,
      "learning_rate": 0.0003642499681818507,
      "loss": 2.1723,
      "step": 99372
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.9659368395805359,
      "learning_rate": 0.00036424597252638155,
      "loss": 2.034,
      "step": 99373
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.8569825887680054,
      "learning_rate": 0.00036424197685896786,
      "loss": 2.313,
      "step": 99374
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.79924476146698,
      "learning_rate": 0.00036423798117961044,
      "loss": 2.1858,
      "step": 99375
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.821994960308075,
      "learning_rate": 0.00036423398548830993,
      "loss": 2.0848,
      "step": 99376
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.9269108176231384,
      "learning_rate": 0.00036422998978506704,
      "loss": 2.1392,
      "step": 99377
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.8635095357894897,
      "learning_rate": 0.00036422599406988275,
      "loss": 2.3001,
      "step": 99378
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.8829192519187927,
      "learning_rate": 0.0003642219983427575,
      "loss": 2.2784,
      "step": 99379
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.8853980898857117,
      "learning_rate": 0.00036421800260369214,
      "loss": 2.169,
      "step": 99380
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.8268671035766602,
      "learning_rate": 0.00036421400685268755,
      "loss": 2.532,
      "step": 99381
    },
    {
      "epoch": 1.29,
      "grad_norm": 1.0799728631973267,
      "learning_rate": 0.00036421001108974426,
      "loss": 2.3232,
      "step": 99382
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.8823160529136658,
      "learning_rate": 0.00036420601531486315,
      "loss": 2.0097,
      "step": 99383
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.9965046644210815,
      "learning_rate": 0.00036420201952804503,
      "loss": 2.3762,
      "step": 99384
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.9592118859291077,
      "learning_rate": 0.00036419802372929044,
      "loss": 2.0558,
      "step": 99385
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.9390116930007935,
      "learning_rate": 0.00036419402791860013,
      "loss": 2.0325,
      "step": 99386
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.8376674652099609,
      "learning_rate": 0.0003641900320959751,
      "loss": 2.0887,
      "step": 99387
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.81593257188797,
      "learning_rate": 0.00036418603626141587,
      "loss": 2.1398,
      "step": 99388
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.8993404507637024,
      "learning_rate": 0.00036418204041492326,
      "loss": 2.2996,
      "step": 99389
    },
    {
      "epoch": 1.29,
      "grad_norm": 1.0774688720703125,
      "learning_rate": 0.00036417804455649797,
      "loss": 2.2596,
      "step": 99390
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.922393262386322,
      "learning_rate": 0.0003641740486861408,
      "loss": 2.3844,
      "step": 99391
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.9718217849731445,
      "learning_rate": 0.0003641700528038525,
      "loss": 2.3965,
      "step": 99392
    },
    {
      "epoch": 1.29,
      "grad_norm": 1.1507647037506104,
      "learning_rate": 0.00036416605690963366,
      "loss": 2.2049,
      "step": 99393
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.9205292463302612,
      "learning_rate": 0.0003641620610034853,
      "loss": 2.2456,
      "step": 99394
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.9266291856765747,
      "learning_rate": 0.0003641580650854079,
      "loss": 2.2149,
      "step": 99395
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.9235864281654358,
      "learning_rate": 0.00036415406915540227,
      "loss": 2.2768,
      "step": 99396
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.9915255308151245,
      "learning_rate": 0.0003641500732134693,
      "loss": 2.1642,
      "step": 99397
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.758574903011322,
      "learning_rate": 0.0003641460772596095,
      "loss": 1.9502,
      "step": 99398
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.927400529384613,
      "learning_rate": 0.0003641420812938238,
      "loss": 2.0757,
      "step": 99399
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.93233323097229,
      "learning_rate": 0.0003641380853161129,
      "loss": 2.2452,
      "step": 99400
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.9285228252410889,
      "learning_rate": 0.0003641340893264775,
      "loss": 2.3449,
      "step": 99401
    },
    {
      "epoch": 1.29,
      "grad_norm": 1.0698384046554565,
      "learning_rate": 0.00036413009332491827,
      "loss": 2.1472,
      "step": 99402
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.9942648410797119,
      "learning_rate": 0.0003641260973114362,
      "loss": 2.0821,
      "step": 99403
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.92859947681427,
      "learning_rate": 0.00036412210128603173,
      "loss": 2.3615,
      "step": 99404
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.9398454427719116,
      "learning_rate": 0.00036411810524870583,
      "loss": 2.1225,
      "step": 99405
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.8531479835510254,
      "learning_rate": 0.0003641141091994592,
      "loss": 2.2176,
      "step": 99406
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.9261795282363892,
      "learning_rate": 0.0003641101131382925,
      "loss": 2.016,
      "step": 99407
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.8959211111068726,
      "learning_rate": 0.0003641061170652065,
      "loss": 2.0796,
      "step": 99408
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.9162717461585999,
      "learning_rate": 0.00036410212098020203,
      "loss": 2.0518,
      "step": 99409
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.8562474846839905,
      "learning_rate": 0.0003640981248832797,
      "loss": 2.2386,
      "step": 99410
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.9100279211997986,
      "learning_rate": 0.0003640941287744403,
      "loss": 2.0051,
      "step": 99411
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.9124030470848083,
      "learning_rate": 0.00036409013265368466,
      "loss": 2.0931,
      "step": 99412
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.9451156258583069,
      "learning_rate": 0.0003640861365210134,
      "loss": 2.2058,
      "step": 99413
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.9144366383552551,
      "learning_rate": 0.00036408214037642733,
      "loss": 2.0047,
      "step": 99414
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.9784601926803589,
      "learning_rate": 0.0003640781442199272,
      "loss": 2.1929,
      "step": 99415
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.9779165983200073,
      "learning_rate": 0.0003640741480515138,
      "loss": 2.2266,
      "step": 99416
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.9035618305206299,
      "learning_rate": 0.00036407015187118764,
      "loss": 2.0481,
      "step": 99417
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.951271116733551,
      "learning_rate": 0.0003640661556789498,
      "loss": 2.1708,
      "step": 99418
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.8884739875793457,
      "learning_rate": 0.00036406215947480073,
      "loss": 2.2628,
      "step": 99419
    },
    {
      "epoch": 1.29,
      "grad_norm": 1.0430340766906738,
      "learning_rate": 0.00036405816325874133,
      "loss": 2.2269,
      "step": 99420
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.9663410782814026,
      "learning_rate": 0.00036405416703077235,
      "loss": 2.0316,
      "step": 99421
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.8961566090583801,
      "learning_rate": 0.0003640501707908945,
      "loss": 2.2708,
      "step": 99422
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.8375818133354187,
      "learning_rate": 0.0003640461745391084,
      "loss": 2.0354,
      "step": 99423
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.9575011134147644,
      "learning_rate": 0.000364042178275415,
      "loss": 2.2965,
      "step": 99424
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.8275957107543945,
      "learning_rate": 0.0003640381819998149,
      "loss": 2.0376,
      "step": 99425
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.9300507307052612,
      "learning_rate": 0.0003640341857123089,
      "loss": 2.2748,
      "step": 99426
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.9018104672431946,
      "learning_rate": 0.0003640301894128978,
      "loss": 2.0451,
      "step": 99427
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.841200053691864,
      "learning_rate": 0.0003640261931015823,
      "loss": 1.8979,
      "step": 99428
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.8583739995956421,
      "learning_rate": 0.000364022196778363,
      "loss": 2.0495,
      "step": 99429
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.8979661464691162,
      "learning_rate": 0.00036401820044324086,
      "loss": 2.1798,
      "step": 99430
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.9496749639511108,
      "learning_rate": 0.00036401420409621645,
      "loss": 2.2557,
      "step": 99431
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.9119287729263306,
      "learning_rate": 0.0003640102077372907,
      "loss": 2.1857,
      "step": 99432
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.925717294216156,
      "learning_rate": 0.0003640062113664641,
      "loss": 1.7559,
      "step": 99433
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.9637891054153442,
      "learning_rate": 0.0003640022149837376,
      "loss": 2.2443,
      "step": 99434
    },
    {
      "epoch": 1.29,
      "grad_norm": 1.014931321144104,
      "learning_rate": 0.00036399821858911196,
      "loss": 2.1597,
      "step": 99435
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.9312857389450073,
      "learning_rate": 0.00036399422218258773,
      "loss": 2.1937,
      "step": 99436
    },
    {
      "epoch": 1.29,
      "grad_norm": 1.0029397010803223,
      "learning_rate": 0.0003639902257641658,
      "loss": 2.1296,
      "step": 99437
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.9413896203041077,
      "learning_rate": 0.00036398622933384695,
      "loss": 2.138,
      "step": 99438
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.9487640857696533,
      "learning_rate": 0.00036398223289163176,
      "loss": 2.2196,
      "step": 99439
    },
    {
      "epoch": 1.29,
      "grad_norm": 1.2142761945724487,
      "learning_rate": 0.00036397823643752114,
      "loss": 2.5162,
      "step": 99440
    },
    {
      "epoch": 1.29,
      "grad_norm": 1.0264809131622314,
      "learning_rate": 0.0003639742399715157,
      "loss": 2.265,
      "step": 99441
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.9394388794898987,
      "learning_rate": 0.00036397024349361625,
      "loss": 2.3584,
      "step": 99442
    },
    {
      "epoch": 1.29,
      "grad_norm": 1.1006027460098267,
      "learning_rate": 0.00036396624700382355,
      "loss": 2.0312,
      "step": 99443
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.8614352345466614,
      "learning_rate": 0.00036396225050213825,
      "loss": 2.1667,
      "step": 99444
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.9752868413925171,
      "learning_rate": 0.0003639582539885613,
      "loss": 2.0258,
      "step": 99445
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.9151197671890259,
      "learning_rate": 0.0003639542574630932,
      "loss": 1.9057,
      "step": 99446
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.9627470374107361,
      "learning_rate": 0.0003639502609257347,
      "loss": 2.387,
      "step": 99447
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.8250824213027954,
      "learning_rate": 0.0003639462643764869,
      "loss": 2.4021,
      "step": 99448
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.812214732170105,
      "learning_rate": 0.0003639422678153501,
      "loss": 2.1387,
      "step": 99449
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.8612669706344604,
      "learning_rate": 0.0003639382712423252,
      "loss": 2.1249,
      "step": 99450
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.8944598436355591,
      "learning_rate": 0.0003639342746574131,
      "loss": 2.1799,
      "step": 99451
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.9063867330551147,
      "learning_rate": 0.0003639302780606143,
      "loss": 2.5906,
      "step": 99452
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.8414433598518372,
      "learning_rate": 0.00036392628145192967,
      "loss": 2.037,
      "step": 99453
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.8370549082756042,
      "learning_rate": 0.00036392228483136,
      "loss": 2.1279,
      "step": 99454
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.9265297651290894,
      "learning_rate": 0.0003639182881989059,
      "loss": 2.3203,
      "step": 99455
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.820370614528656,
      "learning_rate": 0.00036391429155456826,
      "loss": 2.0451,
      "step": 99456
    },
    {
      "epoch": 1.29,
      "grad_norm": 0.7964966297149658,
      "learning_rate": 0.00036391029489834766,
      "loss": 2.1823,
      "step": 99457
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.9339701533317566,
      "learning_rate": 0.000363906298230245,
      "loss": 2.3514,
      "step": 99458
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.8661603331565857,
      "learning_rate": 0.0003639023015502609,
      "loss": 2.1576,
      "step": 99459
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.9640686511993408,
      "learning_rate": 0.00036389830485839615,
      "loss": 1.9921,
      "step": 99460
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.836120069026947,
      "learning_rate": 0.0003638943081546516,
      "loss": 2.2139,
      "step": 99461
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.8612567186355591,
      "learning_rate": 0.00036389031143902785,
      "loss": 1.9719,
      "step": 99462
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.9031834006309509,
      "learning_rate": 0.0003638863147115256,
      "loss": 2.2674,
      "step": 99463
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.9542892575263977,
      "learning_rate": 0.00036388231797214573,
      "loss": 2.1387,
      "step": 99464
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.8123321533203125,
      "learning_rate": 0.00036387832122088896,
      "loss": 2.3154,
      "step": 99465
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.8168947100639343,
      "learning_rate": 0.00036387432445775594,
      "loss": 2.2585,
      "step": 99466
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.8247846961021423,
      "learning_rate": 0.00036387032768274755,
      "loss": 2.199,
      "step": 99467
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.7695209980010986,
      "learning_rate": 0.00036386633089586444,
      "loss": 2.1451,
      "step": 99468
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.9131715297698975,
      "learning_rate": 0.0003638623340971073,
      "loss": 2.2797,
      "step": 99469
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.869013786315918,
      "learning_rate": 0.00036385833728647707,
      "loss": 1.9826,
      "step": 99470
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.8798261284828186,
      "learning_rate": 0.0003638543404639743,
      "loss": 2.4273,
      "step": 99471
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.8465765118598938,
      "learning_rate": 0.0003638503436295998,
      "loss": 1.9551,
      "step": 99472
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.9783928394317627,
      "learning_rate": 0.00036384634678335434,
      "loss": 2.1824,
      "step": 99473
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.8206304311752319,
      "learning_rate": 0.00036384234992523865,
      "loss": 2.3113,
      "step": 99474
    },
    {
      "epoch": 1.3,
      "grad_norm": 1.021364688873291,
      "learning_rate": 0.00036383835305525344,
      "loss": 2.5739,
      "step": 99475
    },
    {
      "epoch": 1.3,
      "grad_norm": 1.050968885421753,
      "learning_rate": 0.0003638343561733995,
      "loss": 2.2819,
      "step": 99476
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.9260392785072327,
      "learning_rate": 0.00036383035927967745,
      "loss": 2.49,
      "step": 99477
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.890641450881958,
      "learning_rate": 0.00036382636237408824,
      "loss": 2.4129,
      "step": 99478
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.9601067304611206,
      "learning_rate": 0.0003638223654566325,
      "loss": 2.1515,
      "step": 99479
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.8410441279411316,
      "learning_rate": 0.0003638183685273109,
      "loss": 2.2427,
      "step": 99480
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.8305577039718628,
      "learning_rate": 0.0003638143715861243,
      "loss": 2.1147,
      "step": 99481
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.9354700446128845,
      "learning_rate": 0.0003638103746330734,
      "loss": 2.0477,
      "step": 99482
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.9369547963142395,
      "learning_rate": 0.000363806377668159,
      "loss": 2.027,
      "step": 99483
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.8429822325706482,
      "learning_rate": 0.0003638023806913817,
      "loss": 2.2132,
      "step": 99484
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.9223834872245789,
      "learning_rate": 0.00036379838370274246,
      "loss": 2.1648,
      "step": 99485
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.873676598072052,
      "learning_rate": 0.0003637943867022418,
      "loss": 2.2131,
      "step": 99486
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.8971732258796692,
      "learning_rate": 0.00036379038968988053,
      "loss": 2.3119,
      "step": 99487
    },
    {
      "epoch": 1.3,
      "grad_norm": 1.0646895170211792,
      "learning_rate": 0.0003637863926656595,
      "loss": 1.9736,
      "step": 99488
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.9265927672386169,
      "learning_rate": 0.00036378239562957933,
      "loss": 2.0852,
      "step": 99489
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.8184841275215149,
      "learning_rate": 0.00036377839858164077,
      "loss": 2.1379,
      "step": 99490
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.9706614017486572,
      "learning_rate": 0.0003637744015218447,
      "loss": 2.1156,
      "step": 99491
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.8185774683952332,
      "learning_rate": 0.0003637704044501917,
      "loss": 2.1109,
      "step": 99492
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.8924909830093384,
      "learning_rate": 0.00036376640736668255,
      "loss": 2.4427,
      "step": 99493
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.8679850697517395,
      "learning_rate": 0.0003637624102713181,
      "loss": 2.079,
      "step": 99494
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.8400579690933228,
      "learning_rate": 0.000363758413164099,
      "loss": 2.3447,
      "step": 99495
    },
    {
      "epoch": 1.3,
      "grad_norm": 1.0190508365631104,
      "learning_rate": 0.000363754416045026,
      "loss": 2.1431,
      "step": 99496
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.9145325422286987,
      "learning_rate": 0.0003637504189140999,
      "loss": 2.1879,
      "step": 99497
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.8955850601196289,
      "learning_rate": 0.00036374642177132123,
      "loss": 2.0118,
      "step": 99498
    },
    {
      "epoch": 1.3,
      "grad_norm": 1.3580150604248047,
      "learning_rate": 0.00036374242461669105,
      "loss": 2.2339,
      "step": 99499
    },
    {
      "epoch": 1.3,
      "grad_norm": 1.0256476402282715,
      "learning_rate": 0.0003637384274502099,
      "loss": 2.3728,
      "step": 99500
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.8717281818389893,
      "learning_rate": 0.00036373443027187856,
      "loss": 2.1213,
      "step": 99501
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.9652872681617737,
      "learning_rate": 0.00036373043308169777,
      "loss": 2.2293,
      "step": 99502
    },
    {
      "epoch": 1.3,
      "grad_norm": 1.1691391468048096,
      "learning_rate": 0.0003637264358796684,
      "loss": 2.0895,
      "step": 99503
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.8524531126022339,
      "learning_rate": 0.000363722438665791,
      "loss": 2.1704,
      "step": 99504
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.9254133105278015,
      "learning_rate": 0.0003637184414400664,
      "loss": 2.2555,
      "step": 99505
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.8597067594528198,
      "learning_rate": 0.00036371444420249527,
      "loss": 2.1094,
      "step": 99506
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.795387327671051,
      "learning_rate": 0.0003637104469530785,
      "loss": 2.1133,
      "step": 99507
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.8293598294258118,
      "learning_rate": 0.00036370644969181676,
      "loss": 2.1779,
      "step": 99508
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.9342770576477051,
      "learning_rate": 0.00036370245241871077,
      "loss": 2.4131,
      "step": 99509
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.78858482837677,
      "learning_rate": 0.00036369845513376127,
      "loss": 2.0143,
      "step": 99510
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.8367199897766113,
      "learning_rate": 0.00036369445783696904,
      "loss": 1.9435,
      "step": 99511
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.8083975315093994,
      "learning_rate": 0.0003636904605283349,
      "loss": 2.1111,
      "step": 99512
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.8797096014022827,
      "learning_rate": 0.0003636864632078594,
      "loss": 2.2078,
      "step": 99513
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.838480532169342,
      "learning_rate": 0.0003636824658755434,
      "loss": 2.2323,
      "step": 99514
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.8930140733718872,
      "learning_rate": 0.00036367846853138755,
      "loss": 2.167,
      "step": 99515
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.8677928447723389,
      "learning_rate": 0.00036367447117539284,
      "loss": 2.0057,
      "step": 99516
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.8897655606269836,
      "learning_rate": 0.0003636704738075597,
      "loss": 2.3752,
      "step": 99517
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.9682179689407349,
      "learning_rate": 0.00036366647642788907,
      "loss": 2.1038,
      "step": 99518
    },
    {
      "epoch": 1.3,
      "grad_norm": 1.0497374534606934,
      "learning_rate": 0.00036366247903638164,
      "loss": 2.081,
      "step": 99519
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.8222398161888123,
      "learning_rate": 0.0003636584816330382,
      "loss": 2.0011,
      "step": 99520
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.8289079666137695,
      "learning_rate": 0.00036365448421785937,
      "loss": 2.1541,
      "step": 99521
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.9637348055839539,
      "learning_rate": 0.000363650486790846,
      "loss": 2.2286,
      "step": 99522
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.8683087229728699,
      "learning_rate": 0.00036364648935199883,
      "loss": 1.9736,
      "step": 99523
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.8138126730918884,
      "learning_rate": 0.0003636424919013185,
      "loss": 2.2475,
      "step": 99524
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.8204519152641296,
      "learning_rate": 0.00036363849443880593,
      "loss": 2.2905,
      "step": 99525
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.9147999882698059,
      "learning_rate": 0.00036363449696446166,
      "loss": 2.3409,
      "step": 99526
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.7845576405525208,
      "learning_rate": 0.0003636304994782866,
      "loss": 2.1986,
      "step": 99527
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.9023059606552124,
      "learning_rate": 0.00036362650198028143,
      "loss": 2.1563,
      "step": 99528
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.8907008171081543,
      "learning_rate": 0.00036362250447044683,
      "loss": 2.2328,
      "step": 99529
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.9212561845779419,
      "learning_rate": 0.0003636185069487837,
      "loss": 2.4009,
      "step": 99530
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.8373464941978455,
      "learning_rate": 0.0003636145094152926,
      "loss": 2.1439,
      "step": 99531
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.8921578526496887,
      "learning_rate": 0.0003636105118699744,
      "loss": 2.1359,
      "step": 99532
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.8924800753593445,
      "learning_rate": 0.0003636065143128298,
      "loss": 2.1683,
      "step": 99533
    },
    {
      "epoch": 1.3,
      "grad_norm": 1.0155340433120728,
      "learning_rate": 0.0003636025167438596,
      "loss": 2.0719,
      "step": 99534
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.8242939114570618,
      "learning_rate": 0.0003635985191630644,
      "loss": 2.0609,
      "step": 99535
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.9206989407539368,
      "learning_rate": 0.0003635945215704451,
      "loss": 2.1483,
      "step": 99536
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.9210777878761292,
      "learning_rate": 0.00036359052396600243,
      "loss": 2.246,
      "step": 99537
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.8173942565917969,
      "learning_rate": 0.000363586526349737,
      "loss": 1.9539,
      "step": 99538
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.8815869092941284,
      "learning_rate": 0.00036358252872164957,
      "loss": 2.3974,
      "step": 99539
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.9117376208305359,
      "learning_rate": 0.0003635785310817411,
      "loss": 2.4036,
      "step": 99540
    },
    {
      "epoch": 1.3,
      "grad_norm": 1.138871431350708,
      "learning_rate": 0.0003635745334300121,
      "loss": 2.1616,
      "step": 99541
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.9909799098968506,
      "learning_rate": 0.00036357053576646333,
      "loss": 2.2177,
      "step": 99542
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.9463845491409302,
      "learning_rate": 0.0003635665380910957,
      "loss": 2.0129,
      "step": 99543
    },
    {
      "epoch": 1.3,
      "grad_norm": 1.0320287942886353,
      "learning_rate": 0.0003635625404039099,
      "loss": 1.974,
      "step": 99544
    },
    {
      "epoch": 1.3,
      "grad_norm": 1.194047212600708,
      "learning_rate": 0.0003635585427049065,
      "loss": 2.122,
      "step": 99545
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.8222474455833435,
      "learning_rate": 0.00036355454499408646,
      "loss": 2.0126,
      "step": 99546
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.9666790962219238,
      "learning_rate": 0.0003635505472714503,
      "loss": 2.1525,
      "step": 99547
    },
    {
      "epoch": 1.3,
      "grad_norm": 1.164129614830017,
      "learning_rate": 0.000363546549536999,
      "loss": 2.2039,
      "step": 99548
    },
    {
      "epoch": 1.3,
      "grad_norm": 1.0424309968948364,
      "learning_rate": 0.00036354255179073323,
      "loss": 2.1174,
      "step": 99549
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.9125619530677795,
      "learning_rate": 0.0003635385540326536,
      "loss": 2.1783,
      "step": 99550
    },
    {
      "epoch": 1.3,
      "grad_norm": 1.312262773513794,
      "learning_rate": 0.000363534556262761,
      "loss": 2.1353,
      "step": 99551
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.8417596220970154,
      "learning_rate": 0.00036353055848105614,
      "loss": 2.2769,
      "step": 99552
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.8476528525352478,
      "learning_rate": 0.0003635265606875397,
      "loss": 2.2361,
      "step": 99553
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.8272147178649902,
      "learning_rate": 0.0003635225628822125,
      "loss": 2.2426,
      "step": 99554
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.9174523949623108,
      "learning_rate": 0.0003635185650650753,
      "loss": 2.1848,
      "step": 99555
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.813177764415741,
      "learning_rate": 0.00036351456723612873,
      "loss": 2.1753,
      "step": 99556
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.9122198820114136,
      "learning_rate": 0.0003635105693953736,
      "loss": 2.0185,
      "step": 99557
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.8651317954063416,
      "learning_rate": 0.00036350657154281073,
      "loss": 1.9964,
      "step": 99558
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.9859456419944763,
      "learning_rate": 0.0003635025736784408,
      "loss": 2.1702,
      "step": 99559
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.8898937702178955,
      "learning_rate": 0.0003634985758022644,
      "loss": 2.3096,
      "step": 99560
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.9188416600227356,
      "learning_rate": 0.0003634945779142826,
      "loss": 2.1021,
      "step": 99561
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.9197000861167908,
      "learning_rate": 0.00036349058001449586,
      "loss": 2.041,
      "step": 99562
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.8719446063041687,
      "learning_rate": 0.00036348658210290504,
      "loss": 2.0242,
      "step": 99563
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.9466880559921265,
      "learning_rate": 0.0003634825841795108,
      "loss": 2.2136,
      "step": 99564
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.8342450261116028,
      "learning_rate": 0.00036347858624431403,
      "loss": 2.0275,
      "step": 99565
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.8483179807662964,
      "learning_rate": 0.0003634745882973154,
      "loss": 2.2145,
      "step": 99566
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.8608319759368896,
      "learning_rate": 0.00036347059033851566,
      "loss": 2.3265,
      "step": 99567
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.8331023454666138,
      "learning_rate": 0.00036346659236791547,
      "loss": 2.0209,
      "step": 99568
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.7938666939735413,
      "learning_rate": 0.0003634625943855156,
      "loss": 2.1195,
      "step": 99569
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.9241169095039368,
      "learning_rate": 0.00036345859639131696,
      "loss": 2.2154,
      "step": 99570
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.9639610052108765,
      "learning_rate": 0.0003634545983853201,
      "loss": 2.2804,
      "step": 99571
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.872495174407959,
      "learning_rate": 0.0003634506003675258,
      "loss": 1.9827,
      "step": 99572
    },
    {
      "epoch": 1.3,
      "grad_norm": 1.0405056476593018,
      "learning_rate": 0.00036344660233793497,
      "loss": 2.1159,
      "step": 99573
    },
    {
      "epoch": 1.3,
      "grad_norm": 1.1503312587738037,
      "learning_rate": 0.00036344260429654803,
      "loss": 2.2257,
      "step": 99574
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.8521419763565063,
      "learning_rate": 0.000363438606243366,
      "loss": 2.3536,
      "step": 99575
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.8148472905158997,
      "learning_rate": 0.00036343460817838964,
      "loss": 2.11,
      "step": 99576
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.9602285027503967,
      "learning_rate": 0.0003634306101016195,
      "loss": 2.2639,
      "step": 99577
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.8813337683677673,
      "learning_rate": 0.0003634266120130563,
      "loss": 2.1285,
      "step": 99578
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.8413169384002686,
      "learning_rate": 0.0003634226139127011,
      "loss": 2.0647,
      "step": 99579
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.8582797050476074,
      "learning_rate": 0.00036341861580055433,
      "loss": 2.1011,
      "step": 99580
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.8514699339866638,
      "learning_rate": 0.0003634146176766168,
      "loss": 1.991,
      "step": 99581
    },
    {
      "epoch": 1.3,
      "grad_norm": 1.0837494134902954,
      "learning_rate": 0.00036341061954088945,
      "loss": 2.0431,
      "step": 99582
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.8973871469497681,
      "learning_rate": 0.0003634066213933727,
      "loss": 2.1914,
      "step": 99583
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.8649733066558838,
      "learning_rate": 0.00036340262323406757,
      "loss": 2.043,
      "step": 99584
    },
    {
      "epoch": 1.3,
      "grad_norm": 1.052473783493042,
      "learning_rate": 0.0003633986250629747,
      "loss": 2.225,
      "step": 99585
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.9513629674911499,
      "learning_rate": 0.0003633946268800947,
      "loss": 2.2837,
      "step": 99586
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.8183738589286804,
      "learning_rate": 0.0003633906286854285,
      "loss": 2.1231,
      "step": 99587
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.8659033179283142,
      "learning_rate": 0.00036338663047897685,
      "loss": 2.0764,
      "step": 99588
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.8449370265007019,
      "learning_rate": 0.00036338263226074044,
      "loss": 2.1472,
      "step": 99589
    },
    {
      "epoch": 1.3,
      "grad_norm": 1.1048403978347778,
      "learning_rate": 0.0003633786340307199,
      "loss": 2.0244,
      "step": 99590
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.9730960726737976,
      "learning_rate": 0.0003633746357889161,
      "loss": 2.2664,
      "step": 99591
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.9415176510810852,
      "learning_rate": 0.0003633706375353298,
      "loss": 2.1089,
      "step": 99592
    },
    {
      "epoch": 1.3,
      "grad_norm": 1.0306928157806396,
      "learning_rate": 0.00036336663926996173,
      "loss": 2.2549,
      "step": 99593
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.9244044423103333,
      "learning_rate": 0.0003633626409928125,
      "loss": 2.0814,
      "step": 99594
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.8452944755554199,
      "learning_rate": 0.0003633586427038831,
      "loss": 2.1288,
      "step": 99595
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.925993800163269,
      "learning_rate": 0.00036335464440317403,
      "loss": 2.1472,
      "step": 99596
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.9887963533401489,
      "learning_rate": 0.00036335064609068613,
      "loss": 2.1472,
      "step": 99597
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.8485077619552612,
      "learning_rate": 0.00036334664776642027,
      "loss": 2.1397,
      "step": 99598
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.9121379852294922,
      "learning_rate": 0.00036334264943037694,
      "loss": 2.0732,
      "step": 99599
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.8894482851028442,
      "learning_rate": 0.00036333865108255706,
      "loss": 2.1558,
      "step": 99600
    },
    {
      "epoch": 1.3,
      "grad_norm": 1.0068737268447876,
      "learning_rate": 0.00036333465272296133,
      "loss": 1.9768,
      "step": 99601
    },
    {
      "epoch": 1.3,
      "grad_norm": 1.051445722579956,
      "learning_rate": 0.0003633306543515905,
      "loss": 1.9822,
      "step": 99602
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.9721022248268127,
      "learning_rate": 0.0003633266559684453,
      "loss": 2.0913,
      "step": 99603
    },
    {
      "epoch": 1.3,
      "grad_norm": 1.1098843812942505,
      "learning_rate": 0.0003633226575735265,
      "loss": 2.1961,
      "step": 99604
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.9107272028923035,
      "learning_rate": 0.0003633186591668348,
      "loss": 2.1438,
      "step": 99605
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.8326626420021057,
      "learning_rate": 0.0003633146607483709,
      "loss": 2.3207,
      "step": 99606
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.9116513729095459,
      "learning_rate": 0.00036331066231813565,
      "loss": 2.2294,
      "step": 99607
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.8956721425056458,
      "learning_rate": 0.00036330666387612983,
      "loss": 2.1194,
      "step": 99608
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.8624877333641052,
      "learning_rate": 0.00036330266542235407,
      "loss": 1.9976,
      "step": 99609
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.8688316941261292,
      "learning_rate": 0.0003632986669568091,
      "loss": 2.2998,
      "step": 99610
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.8949124813079834,
      "learning_rate": 0.0003632946684794958,
      "loss": 2.1267,
      "step": 99611
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.9386194348335266,
      "learning_rate": 0.00036329066999041476,
      "loss": 2.247,
      "step": 99612
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.886904239654541,
      "learning_rate": 0.0003632866714895668,
      "loss": 2.1759,
      "step": 99613
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.8566303849220276,
      "learning_rate": 0.00036328267297695263,
      "loss": 2.0828,
      "step": 99614
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.8986591100692749,
      "learning_rate": 0.00036327867445257303,
      "loss": 2.4056,
      "step": 99615
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.8607423901557922,
      "learning_rate": 0.0003632746759164288,
      "loss": 2.0525,
      "step": 99616
    },
    {
      "epoch": 1.3,
      "grad_norm": 1.0040209293365479,
      "learning_rate": 0.00036327067736852044,
      "loss": 2.4345,
      "step": 99617
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.9548507332801819,
      "learning_rate": 0.00036326667880884903,
      "loss": 2.2137,
      "step": 99618
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.8396925926208496,
      "learning_rate": 0.00036326268023741516,
      "loss": 2.1248,
      "step": 99619
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.8902414441108704,
      "learning_rate": 0.00036325868165421944,
      "loss": 2.1166,
      "step": 99620
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.892194926738739,
      "learning_rate": 0.00036325468305926277,
      "loss": 2.237,
      "step": 99621
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.8924185037612915,
      "learning_rate": 0.000363250684452546,
      "loss": 1.9341,
      "step": 99622
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.8441236615180969,
      "learning_rate": 0.00036324668583406956,
      "loss": 2.2531,
      "step": 99623
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.8309017419815063,
      "learning_rate": 0.0003632426872038344,
      "loss": 1.9825,
      "step": 99624
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.8551861047744751,
      "learning_rate": 0.00036323868856184133,
      "loss": 2.0679,
      "step": 99625
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.8909287452697754,
      "learning_rate": 0.00036323468990809093,
      "loss": 2.2081,
      "step": 99626
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.8143578171730042,
      "learning_rate": 0.000363230691242584,
      "loss": 2.1916,
      "step": 99627
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.8274610638618469,
      "learning_rate": 0.00036322669256532136,
      "loss": 2.2894,
      "step": 99628
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.91102534532547,
      "learning_rate": 0.00036322269387630354,
      "loss": 2.2185,
      "step": 99629
    },
    {
      "epoch": 1.3,
      "grad_norm": 1.0629935264587402,
      "learning_rate": 0.0003632186951755315,
      "loss": 2.203,
      "step": 99630
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.9090973138809204,
      "learning_rate": 0.00036321469646300597,
      "loss": 2.191,
      "step": 99631
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.9669338464736938,
      "learning_rate": 0.0003632106977387276,
      "loss": 2.4614,
      "step": 99632
    },
    {
      "epoch": 1.3,
      "grad_norm": 1.0906398296356201,
      "learning_rate": 0.0003632066990026971,
      "loss": 2.2842,
      "step": 99633
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.9245464205741882,
      "learning_rate": 0.0003632027002549154,
      "loss": 2.2775,
      "step": 99634
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.8719089031219482,
      "learning_rate": 0.00036319870149538306,
      "loss": 2.2358,
      "step": 99635
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.9592685103416443,
      "learning_rate": 0.0003631947027241008,
      "loss": 2.2787,
      "step": 99636
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.9304536581039429,
      "learning_rate": 0.00036319070394106966,
      "loss": 2.4156,
      "step": 99637
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.8237966895103455,
      "learning_rate": 0.00036318670514629,
      "loss": 1.93,
      "step": 99638
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.8837960362434387,
      "learning_rate": 0.00036318270633976275,
      "loss": 2.1941,
      "step": 99639
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.9152011871337891,
      "learning_rate": 0.0003631787075214888,
      "loss": 2.1795,
      "step": 99640
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.8164907693862915,
      "learning_rate": 0.00036317470869146854,
      "loss": 2.0854,
      "step": 99641
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.8628028035163879,
      "learning_rate": 0.000363170709849703,
      "loss": 2.1856,
      "step": 99642
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.8129327297210693,
      "learning_rate": 0.00036316671099619285,
      "loss": 2.1482,
      "step": 99643
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.8427343368530273,
      "learning_rate": 0.0003631627121309388,
      "loss": 2.0726,
      "step": 99644
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.9400268197059631,
      "learning_rate": 0.00036315871325394157,
      "loss": 2.2269,
      "step": 99645
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.9305302500724792,
      "learning_rate": 0.000363154714365202,
      "loss": 1.8884,
      "step": 99646
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.9050828218460083,
      "learning_rate": 0.00036315071546472074,
      "loss": 2.1488,
      "step": 99647
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.9300446510314941,
      "learning_rate": 0.0003631467165524986,
      "loss": 2.3014,
      "step": 99648
    },
    {
      "epoch": 1.3,
      "grad_norm": 1.0834792852401733,
      "learning_rate": 0.0003631427176285363,
      "loss": 2.4509,
      "step": 99649
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.9116827845573425,
      "learning_rate": 0.0003631387186928345,
      "loss": 2.3107,
      "step": 99650
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.8218060731887817,
      "learning_rate": 0.00036313471974539407,
      "loss": 2.1321,
      "step": 99651
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.8242014050483704,
      "learning_rate": 0.00036313072078621574,
      "loss": 2.1753,
      "step": 99652
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.999347984790802,
      "learning_rate": 0.0003631267218153002,
      "loss": 2.24,
      "step": 99653
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.9409846663475037,
      "learning_rate": 0.0003631227228326482,
      "loss": 1.9837,
      "step": 99654
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.8544363975524902,
      "learning_rate": 0.0003631187238382604,
      "loss": 2.0258,
      "step": 99655
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.905603289604187,
      "learning_rate": 0.00036311472483213777,
      "loss": 2.2844,
      "step": 99656
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.9685705900192261,
      "learning_rate": 0.00036311072581428084,
      "loss": 2.0409,
      "step": 99657
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.7639092803001404,
      "learning_rate": 0.00036310672678469046,
      "loss": 2.0509,
      "step": 99658
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.935880720615387,
      "learning_rate": 0.0003631027277433674,
      "loss": 1.9949,
      "step": 99659
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.9729979634284973,
      "learning_rate": 0.0003630987286903123,
      "loss": 1.9289,
      "step": 99660
    },
    {
      "epoch": 1.3,
      "grad_norm": 1.0122647285461426,
      "learning_rate": 0.00036309472962552596,
      "loss": 2.2813,
      "step": 99661
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.9401289224624634,
      "learning_rate": 0.0003630907305490092,
      "loss": 2.1775,
      "step": 99662
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.8613508343696594,
      "learning_rate": 0.00036308673146076255,
      "loss": 2.1087,
      "step": 99663
    },
    {
      "epoch": 1.3,
      "grad_norm": 1.1786259412765503,
      "learning_rate": 0.0003630827323607869,
      "loss": 2.183,
      "step": 99664
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.9365384578704834,
      "learning_rate": 0.00036307873324908306,
      "loss": 2.1759,
      "step": 99665
    },
    {
      "epoch": 1.3,
      "grad_norm": 1.015556812286377,
      "learning_rate": 0.00036307473412565167,
      "loss": 1.9166,
      "step": 99666
    },
    {
      "epoch": 1.3,
      "grad_norm": 1.703962802886963,
      "learning_rate": 0.00036307073499049344,
      "loss": 2.0057,
      "step": 99667
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.9907877445220947,
      "learning_rate": 0.00036306673584360924,
      "loss": 1.9202,
      "step": 99668
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.8766751289367676,
      "learning_rate": 0.0003630627366849997,
      "loss": 2.158,
      "step": 99669
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.9596279859542847,
      "learning_rate": 0.0003630587375146656,
      "loss": 2.1381,
      "step": 99670
    },
    {
      "epoch": 1.3,
      "grad_norm": 1.1550555229187012,
      "learning_rate": 0.00036305473833260775,
      "loss": 1.9902,
      "step": 99671
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.9055739045143127,
      "learning_rate": 0.0003630507391388268,
      "loss": 2.2918,
      "step": 99672
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.8737882375717163,
      "learning_rate": 0.0003630467399333235,
      "loss": 2.0904,
      "step": 99673
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.9492724537849426,
      "learning_rate": 0.0003630427407160987,
      "loss": 2.2045,
      "step": 99674
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.8491173982620239,
      "learning_rate": 0.00036303874148715297,
      "loss": 2.1059,
      "step": 99675
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.8103801012039185,
      "learning_rate": 0.0003630347422464872,
      "loss": 1.8445,
      "step": 99676
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.8870543241500854,
      "learning_rate": 0.0003630307429941021,
      "loss": 2.3538,
      "step": 99677
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.8480987548828125,
      "learning_rate": 0.00036302674372999834,
      "loss": 2.1477,
      "step": 99678
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.8647763729095459,
      "learning_rate": 0.0003630227444541767,
      "loss": 2.2387,
      "step": 99679
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.9129132032394409,
      "learning_rate": 0.0003630187451666381,
      "loss": 2.171,
      "step": 99680
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.9831234216690063,
      "learning_rate": 0.00036301474586738293,
      "loss": 2.3735,
      "step": 99681
    },
    {
      "epoch": 1.3,
      "grad_norm": 1.0864341259002686,
      "learning_rate": 0.00036301074655641216,
      "loss": 2.3323,
      "step": 99682
    },
    {
      "epoch": 1.3,
      "grad_norm": 1.089604139328003,
      "learning_rate": 0.00036300674723372664,
      "loss": 2.2262,
      "step": 99683
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.9731342792510986,
      "learning_rate": 0.00036300274789932685,
      "loss": 2.1664,
      "step": 99684
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.8043718338012695,
      "learning_rate": 0.0003629987485532137,
      "loss": 1.9675,
      "step": 99685
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.842739462852478,
      "learning_rate": 0.0003629947491953879,
      "loss": 2.1441,
      "step": 99686
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.8492487668991089,
      "learning_rate": 0.00036299074982585017,
      "loss": 1.9344,
      "step": 99687
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.8393047451972961,
      "learning_rate": 0.00036298675044460125,
      "loss": 2.1185,
      "step": 99688
    },
    {
      "epoch": 1.3,
      "grad_norm": 1.1135854721069336,
      "learning_rate": 0.00036298275105164197,
      "loss": 2.3074,
      "step": 99689
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.8090960383415222,
      "learning_rate": 0.00036297875164697293,
      "loss": 2.2473,
      "step": 99690
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.7912443280220032,
      "learning_rate": 0.000362974752230595,
      "loss": 2.2538,
      "step": 99691
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.8258365392684937,
      "learning_rate": 0.0003629707528025089,
      "loss": 2.29,
      "step": 99692
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.7730636596679688,
      "learning_rate": 0.0003629667533627153,
      "loss": 1.9763,
      "step": 99693
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.8335542678833008,
      "learning_rate": 0.00036296275391121496,
      "loss": 2.2981,
      "step": 99694
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.8356214761734009,
      "learning_rate": 0.0003629587544480087,
      "loss": 2.1202,
      "step": 99695
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.8855748772621155,
      "learning_rate": 0.0003629547549730973,
      "loss": 2.4147,
      "step": 99696
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.8145650029182434,
      "learning_rate": 0.0003629507554864813,
      "loss": 2.1949,
      "step": 99697
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.8284308314323425,
      "learning_rate": 0.0003629467559881616,
      "loss": 2.0162,
      "step": 99698
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.9335461258888245,
      "learning_rate": 0.0003629427564781389,
      "loss": 2.0381,
      "step": 99699
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.8848322033882141,
      "learning_rate": 0.00036293875695641404,
      "loss": 2.009,
      "step": 99700
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.928477942943573,
      "learning_rate": 0.0003629347574229876,
      "loss": 2.0415,
      "step": 99701
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.8425346612930298,
      "learning_rate": 0.0003629307578778604,
      "loss": 2.1987,
      "step": 99702
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.7606443166732788,
      "learning_rate": 0.00036292675832103315,
      "loss": 2.2708,
      "step": 99703
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.8609470129013062,
      "learning_rate": 0.0003629227587525067,
      "loss": 2.1452,
      "step": 99704
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.8024550676345825,
      "learning_rate": 0.0003629187591722817,
      "loss": 2.1568,
      "step": 99705
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.8924041986465454,
      "learning_rate": 0.0003629147595803589,
      "loss": 2.1415,
      "step": 99706
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.8595337867736816,
      "learning_rate": 0.000362910759976739,
      "loss": 2.4449,
      "step": 99707
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.8906998038291931,
      "learning_rate": 0.0003629067603614229,
      "loss": 2.0569,
      "step": 99708
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.9097541570663452,
      "learning_rate": 0.00036290276073441124,
      "loss": 1.9634,
      "step": 99709
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.9343203902244568,
      "learning_rate": 0.00036289876109570475,
      "loss": 2.2329,
      "step": 99710
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.8073104023933411,
      "learning_rate": 0.0003628947614453042,
      "loss": 2.0305,
      "step": 99711
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.881706178188324,
      "learning_rate": 0.0003628907617832103,
      "loss": 2.2101,
      "step": 99712
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.8855684995651245,
      "learning_rate": 0.00036288676210942384,
      "loss": 2.222,
      "step": 99713
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.9567460417747498,
      "learning_rate": 0.00036288276242394556,
      "loss": 2.3749,
      "step": 99714
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.8881877064704895,
      "learning_rate": 0.00036287876272677607,
      "loss": 2.3522,
      "step": 99715
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.8538572788238525,
      "learning_rate": 0.00036287476301791644,
      "loss": 1.9101,
      "step": 99716
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.8946995735168457,
      "learning_rate": 0.0003628707632973671,
      "loss": 2.1578,
      "step": 99717
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.864159345626831,
      "learning_rate": 0.00036286676356512887,
      "loss": 2.2158,
      "step": 99718
    },
    {
      "epoch": 1.3,
      "grad_norm": 1.080314040184021,
      "learning_rate": 0.00036286276382120256,
      "loss": 2.1639,
      "step": 99719
    },
    {
      "epoch": 1.3,
      "grad_norm": 1.0399733781814575,
      "learning_rate": 0.0003628587640655888,
      "loss": 2.2691,
      "step": 99720
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.837212324142456,
      "learning_rate": 0.00036285476429828847,
      "loss": 2.2799,
      "step": 99721
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.9138435125350952,
      "learning_rate": 0.00036285076451930226,
      "loss": 2.4944,
      "step": 99722
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.9582090973854065,
      "learning_rate": 0.0003628467647286309,
      "loss": 2.3758,
      "step": 99723
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.8230448961257935,
      "learning_rate": 0.0003628427649262751,
      "loss": 2.3664,
      "step": 99724
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.9444952011108398,
      "learning_rate": 0.00036283876511223565,
      "loss": 2.1714,
      "step": 99725
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.8253520131111145,
      "learning_rate": 0.00036283476528651344,
      "loss": 2.0859,
      "step": 99726
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.8284644484519958,
      "learning_rate": 0.00036283076544910885,
      "loss": 2.219,
      "step": 99727
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.9646987318992615,
      "learning_rate": 0.00036282676560002295,
      "loss": 2.1886,
      "step": 99728
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.8406069278717041,
      "learning_rate": 0.0003628227657392564,
      "loss": 2.1104,
      "step": 99729
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.8947879672050476,
      "learning_rate": 0.0003628187658668098,
      "loss": 2.1695,
      "step": 99730
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.8363794088363647,
      "learning_rate": 0.000362814765982684,
      "loss": 2.1761,
      "step": 99731
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.8322806358337402,
      "learning_rate": 0.0003628107660868799,
      "loss": 2.2782,
      "step": 99732
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.7962417602539062,
      "learning_rate": 0.000362806766179398,
      "loss": 2.1432,
      "step": 99733
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.7831336855888367,
      "learning_rate": 0.00036280276626023914,
      "loss": 2.3195,
      "step": 99734
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.8897673487663269,
      "learning_rate": 0.000362798766329404,
      "loss": 2.2104,
      "step": 99735
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.8760198950767517,
      "learning_rate": 0.0003627947663868935,
      "loss": 2.3395,
      "step": 99736
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.9646205902099609,
      "learning_rate": 0.00036279076643270814,
      "loss": 2.1084,
      "step": 99737
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.9230587482452393,
      "learning_rate": 0.00036278676646684895,
      "loss": 2.1072,
      "step": 99738
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.7898602485656738,
      "learning_rate": 0.00036278276648931643,
      "loss": 2.0169,
      "step": 99739
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.8352834582328796,
      "learning_rate": 0.00036277876650011134,
      "loss": 1.9985,
      "step": 99740
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.9454070925712585,
      "learning_rate": 0.00036277476649923464,
      "loss": 2.4428,
      "step": 99741
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.949893593788147,
      "learning_rate": 0.00036277076648668684,
      "loss": 2.2626,
      "step": 99742
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.885010838508606,
      "learning_rate": 0.00036276676646246873,
      "loss": 2.3293,
      "step": 99743
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.8115013837814331,
      "learning_rate": 0.0003627627664265812,
      "loss": 2.192,
      "step": 99744
    },
    {
      "epoch": 1.3,
      "grad_norm": 1.0611158609390259,
      "learning_rate": 0.0003627587663790247,
      "loss": 2.0288,
      "step": 99745
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.86652672290802,
      "learning_rate": 0.0003627547663198004,
      "loss": 2.1657,
      "step": 99746
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.7957596778869629,
      "learning_rate": 0.0003627507662489086,
      "loss": 2.0562,
      "step": 99747
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.8228670954704285,
      "learning_rate": 0.0003627467661663504,
      "loss": 2.0633,
      "step": 99748
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.783351719379425,
      "learning_rate": 0.00036274276607212624,
      "loss": 2.1709,
      "step": 99749
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.8731739521026611,
      "learning_rate": 0.0003627387659662372,
      "loss": 2.0359,
      "step": 99750
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.8710045218467712,
      "learning_rate": 0.00036273476584868376,
      "loss": 2.1946,
      "step": 99751
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.9913055300712585,
      "learning_rate": 0.0003627307657194667,
      "loss": 2.2728,
      "step": 99752
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.8720443844795227,
      "learning_rate": 0.0003627267655785868,
      "loss": 2.0813,
      "step": 99753
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.8303614258766174,
      "learning_rate": 0.00036272276542604487,
      "loss": 2.0703,
      "step": 99754
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.786518394947052,
      "learning_rate": 0.0003627187652618415,
      "loss": 2.1013,
      "step": 99755
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.8914003968238831,
      "learning_rate": 0.00036271476508597763,
      "loss": 2.2143,
      "step": 99756
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.8575184941291809,
      "learning_rate": 0.0003627107648984539,
      "loss": 2.2392,
      "step": 99757
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.8722200393676758,
      "learning_rate": 0.000362706764699271,
      "loss": 1.8933,
      "step": 99758
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.851581335067749,
      "learning_rate": 0.00036270276448842977,
      "loss": 2.0763,
      "step": 99759
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.8321493864059448,
      "learning_rate": 0.0003626987642659309,
      "loss": 2.1726,
      "step": 99760
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.9172354936599731,
      "learning_rate": 0.0003626947640317751,
      "loss": 2.0586,
      "step": 99761
    },
    {
      "epoch": 1.3,
      "grad_norm": 1.0031118392944336,
      "learning_rate": 0.0003626907637859633,
      "loss": 2.2415,
      "step": 99762
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.9372648000717163,
      "learning_rate": 0.00036268676352849594,
      "loss": 2.3495,
      "step": 99763
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.9088677167892456,
      "learning_rate": 0.00036268276325937397,
      "loss": 2.5504,
      "step": 99764
    },
    {
      "epoch": 1.3,
      "grad_norm": 1.0305947065353394,
      "learning_rate": 0.0003626787629785982,
      "loss": 2.0898,
      "step": 99765
    },
    {
      "epoch": 1.3,
      "grad_norm": 1.0867999792099,
      "learning_rate": 0.0003626747626861691,
      "loss": 2.2796,
      "step": 99766
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.861733615398407,
      "learning_rate": 0.00036267076238208763,
      "loss": 2.0792,
      "step": 99767
    },
    {
      "epoch": 1.3,
      "grad_norm": 1.0842245817184448,
      "learning_rate": 0.0003626667620663545,
      "loss": 2.2685,
      "step": 99768
    },
    {
      "epoch": 1.3,
      "grad_norm": 1.3298029899597168,
      "learning_rate": 0.00036266276173897044,
      "loss": 2.0363,
      "step": 99769
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.9441957473754883,
      "learning_rate": 0.00036265876139993623,
      "loss": 2.2642,
      "step": 99770
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.9430158734321594,
      "learning_rate": 0.00036265476104925255,
      "loss": 1.9897,
      "step": 99771
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.950764000415802,
      "learning_rate": 0.00036265076068692014,
      "loss": 2.1252,
      "step": 99772
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.921310305595398,
      "learning_rate": 0.0003626467603129398,
      "loss": 2.2167,
      "step": 99773
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.7967374920845032,
      "learning_rate": 0.0003626427599273122,
      "loss": 2.039,
      "step": 99774
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.912858247756958,
      "learning_rate": 0.0003626387595300382,
      "loss": 2.1742,
      "step": 99775
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.8695324063301086,
      "learning_rate": 0.00036263475912111837,
      "loss": 2.0967,
      "step": 99776
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.9160517454147339,
      "learning_rate": 0.0003626307587005536,
      "loss": 2.0859,
      "step": 99777
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.9713367819786072,
      "learning_rate": 0.0003626267582683446,
      "loss": 2.0803,
      "step": 99778
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.862088680267334,
      "learning_rate": 0.0003626227578244921,
      "loss": 2.294,
      "step": 99779
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.9486137628555298,
      "learning_rate": 0.0003626187573689968,
      "loss": 2.3038,
      "step": 99780
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.8456131219863892,
      "learning_rate": 0.0003626147569018597,
      "loss": 2.0216,
      "step": 99781
    },
    {
      "epoch": 1.3,
      "grad_norm": 1.4023126363754272,
      "learning_rate": 0.0003626107564230811,
      "loss": 1.9687,
      "step": 99782
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.8971751928329468,
      "learning_rate": 0.000362606755932662,
      "loss": 2.3622,
      "step": 99783
    },
    {
      "epoch": 1.3,
      "grad_norm": 1.0687761306762695,
      "learning_rate": 0.0003626027554306033,
      "loss": 2.2793,
      "step": 99784
    },
    {
      "epoch": 1.3,
      "grad_norm": 1.0283684730529785,
      "learning_rate": 0.00036259875491690536,
      "loss": 2.3641,
      "step": 99785
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.7960861921310425,
      "learning_rate": 0.0003625947543915692,
      "loss": 2.3419,
      "step": 99786
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.8810299038887024,
      "learning_rate": 0.00036259075385459554,
      "loss": 2.0477,
      "step": 99787
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.7703080773353577,
      "learning_rate": 0.0003625867533059851,
      "loss": 2.1562,
      "step": 99788
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.9934650659561157,
      "learning_rate": 0.0003625827527457385,
      "loss": 2.2518,
      "step": 99789
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.9889007210731506,
      "learning_rate": 0.00036257875217385663,
      "loss": 2.1852,
      "step": 99790
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.9270790219306946,
      "learning_rate": 0.0003625747515903402,
      "loss": 2.192,
      "step": 99791
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.9303449392318726,
      "learning_rate": 0.00036257075099518996,
      "loss": 2.3524,
      "step": 99792
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.8565173149108887,
      "learning_rate": 0.0003625667503884066,
      "loss": 2.0513,
      "step": 99793
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.8493187427520752,
      "learning_rate": 0.00036256274976999086,
      "loss": 2.2462,
      "step": 99794
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.9221097826957703,
      "learning_rate": 0.0003625587491399436,
      "loss": 2.0535,
      "step": 99795
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.9196679592132568,
      "learning_rate": 0.00036255474849826544,
      "loss": 2.2017,
      "step": 99796
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.8600611686706543,
      "learning_rate": 0.00036255074784495724,
      "loss": 2.2878,
      "step": 99797
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.8693209886550903,
      "learning_rate": 0.00036254674718001956,
      "loss": 2.2715,
      "step": 99798
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.8466044068336487,
      "learning_rate": 0.0003625427465034534,
      "loss": 2.1758,
      "step": 99799
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.7695504426956177,
      "learning_rate": 0.0003625387458152593,
      "loss": 1.8564,
      "step": 99800
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.869290292263031,
      "learning_rate": 0.00036253474511543796,
      "loss": 2.2217,
      "step": 99801
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.8373793363571167,
      "learning_rate": 0.0003625307444039904,
      "loss": 2.085,
      "step": 99802
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.9403889775276184,
      "learning_rate": 0.0003625267436809171,
      "loss": 2.3683,
      "step": 99803
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.9406334161758423,
      "learning_rate": 0.0003625227429462189,
      "loss": 2.3084,
      "step": 99804
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.8914557099342346,
      "learning_rate": 0.0003625187421998966,
      "loss": 2.258,
      "step": 99805
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.9127079248428345,
      "learning_rate": 0.0003625147414419509,
      "loss": 2.2795,
      "step": 99806
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.8602083325386047,
      "learning_rate": 0.00036251074067238245,
      "loss": 2.1457,
      "step": 99807
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.9281019568443298,
      "learning_rate": 0.0003625067398911921,
      "loss": 2.257,
      "step": 99808
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.8368148803710938,
      "learning_rate": 0.00036250273909838054,
      "loss": 2.0872,
      "step": 99809
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.8831707239151001,
      "learning_rate": 0.0003624987382939486,
      "loss": 2.219,
      "step": 99810
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.8280202746391296,
      "learning_rate": 0.00036249473747789694,
      "loss": 2.2684,
      "step": 99811
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.8660690188407898,
      "learning_rate": 0.0003624907366502263,
      "loss": 2.333,
      "step": 99812
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.8785293102264404,
      "learning_rate": 0.00036248673581093743,
      "loss": 2.2761,
      "step": 99813
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.8809915781021118,
      "learning_rate": 0.0003624827349600312,
      "loss": 2.2545,
      "step": 99814
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.8546820878982544,
      "learning_rate": 0.00036247873409750816,
      "loss": 2.2676,
      "step": 99815
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.8832768797874451,
      "learning_rate": 0.00036247473322336917,
      "loss": 2.4063,
      "step": 99816
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.8673285245895386,
      "learning_rate": 0.000362470732337615,
      "loss": 2.0709,
      "step": 99817
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.8157020211219788,
      "learning_rate": 0.0003624667314402463,
      "loss": 2.1541,
      "step": 99818
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.8500437140464783,
      "learning_rate": 0.00036246273053126376,
      "loss": 2.188,
      "step": 99819
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.8996621370315552,
      "learning_rate": 0.0003624587296106684,
      "loss": 2.2839,
      "step": 99820
    },
    {
      "epoch": 1.3,
      "grad_norm": 1.00831139087677,
      "learning_rate": 0.0003624547286784607,
      "loss": 2.218,
      "step": 99821
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.7622085809707642,
      "learning_rate": 0.00036245072773464143,
      "loss": 2.1301,
      "step": 99822
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.983282744884491,
      "learning_rate": 0.00036244672677921146,
      "loss": 2.4204,
      "step": 99823
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.8023209571838379,
      "learning_rate": 0.00036244272581217147,
      "loss": 2.1639,
      "step": 99824
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.8915663361549377,
      "learning_rate": 0.0003624387248335221,
      "loss": 1.9616,
      "step": 99825
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.8233210444450378,
      "learning_rate": 0.0003624347238432644,
      "loss": 2.0618,
      "step": 99826
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.8846690654754639,
      "learning_rate": 0.0003624307228413987,
      "loss": 1.9523,
      "step": 99827
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.8229922652244568,
      "learning_rate": 0.00036242672182792605,
      "loss": 2.3169,
      "step": 99828
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.8545451760292053,
      "learning_rate": 0.00036242272080284704,
      "loss": 1.9696,
      "step": 99829
    },
    {
      "epoch": 1.3,
      "grad_norm": 1.0055631399154663,
      "learning_rate": 0.00036241871976616245,
      "loss": 2.2566,
      "step": 99830
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.818706750869751,
      "learning_rate": 0.0003624147187178731,
      "loss": 1.9577,
      "step": 99831
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.893943727016449,
      "learning_rate": 0.0003624107176579797,
      "loss": 2.1496,
      "step": 99832
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.8402358889579773,
      "learning_rate": 0.00036240671658648293,
      "loss": 2.2276,
      "step": 99833
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.878228485584259,
      "learning_rate": 0.0003624027155033836,
      "loss": 2.1886,
      "step": 99834
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.7821966409683228,
      "learning_rate": 0.00036239871440868243,
      "loss": 1.8862,
      "step": 99835
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.8818812370300293,
      "learning_rate": 0.0003623947133023801,
      "loss": 2.2988,
      "step": 99836
    },
    {
      "epoch": 1.3,
      "grad_norm": 1.004171371459961,
      "learning_rate": 0.00036239071218447744,
      "loss": 2.1808,
      "step": 99837
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.8277145028114319,
      "learning_rate": 0.00036238671105497523,
      "loss": 2.2531,
      "step": 99838
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.7830390930175781,
      "learning_rate": 0.00036238270991387407,
      "loss": 1.9594,
      "step": 99839
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.8589043617248535,
      "learning_rate": 0.00036237870876117476,
      "loss": 2.1718,
      "step": 99840
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.9049071073532104,
      "learning_rate": 0.0003623747075968782,
      "loss": 2.3626,
      "step": 99841
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.8745073080062866,
      "learning_rate": 0.00036237070642098485,
      "loss": 2.1506,
      "step": 99842
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.9044693112373352,
      "learning_rate": 0.0003623667052334957,
      "loss": 2.0706,
      "step": 99843
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.8449416160583496,
      "learning_rate": 0.00036236270403441147,
      "loss": 2.0996,
      "step": 99844
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.872071385383606,
      "learning_rate": 0.0003623587028237327,
      "loss": 2.1739,
      "step": 99845
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.8108618259429932,
      "learning_rate": 0.0003623547016014603,
      "loss": 2.2885,
      "step": 99846
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.9229552149772644,
      "learning_rate": 0.0003623507003675951,
      "loss": 2.2111,
      "step": 99847
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.8102187514305115,
      "learning_rate": 0.00036234669912213756,
      "loss": 2.1157,
      "step": 99848
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.8940376043319702,
      "learning_rate": 0.00036234269786508867,
      "loss": 2.2175,
      "step": 99849
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.8246634006500244,
      "learning_rate": 0.0003623386965964492,
      "loss": 2.1111,
      "step": 99850
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.7928494811058044,
      "learning_rate": 0.00036233469531621966,
      "loss": 2.3031,
      "step": 99851
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.8208925127983093,
      "learning_rate": 0.0003623306940244009,
      "loss": 1.9817,
      "step": 99852
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.8757632970809937,
      "learning_rate": 0.0003623266927209938,
      "loss": 2.1967,
      "step": 99853
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.9191933870315552,
      "learning_rate": 0.0003623226914059989,
      "loss": 1.8741,
      "step": 99854
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.9457378387451172,
      "learning_rate": 0.00036231869007941704,
      "loss": 1.9551,
      "step": 99855
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.9027923345565796,
      "learning_rate": 0.00036231468874124896,
      "loss": 2.0742,
      "step": 99856
    },
    {
      "epoch": 1.3,
      "grad_norm": 1.0321050882339478,
      "learning_rate": 0.00036231068739149547,
      "loss": 2.1078,
      "step": 99857
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.8938823342323303,
      "learning_rate": 0.0003623066860301572,
      "loss": 2.0217,
      "step": 99858
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.9514987468719482,
      "learning_rate": 0.00036230268465723493,
      "loss": 2.3345,
      "step": 99859
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.9255778789520264,
      "learning_rate": 0.00036229868327272946,
      "loss": 2.2556,
      "step": 99860
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.9294226765632629,
      "learning_rate": 0.0003622946818766414,
      "loss": 2.0251,
      "step": 99861
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.8543075919151306,
      "learning_rate": 0.00036229068046897167,
      "loss": 2.2004,
      "step": 99862
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.8863992094993591,
      "learning_rate": 0.00036228667904972093,
      "loss": 2.1379,
      "step": 99863
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.9712591767311096,
      "learning_rate": 0.00036228267761888983,
      "loss": 2.2685,
      "step": 99864
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.9272552728652954,
      "learning_rate": 0.0003622786761764793,
      "loss": 2.6219,
      "step": 99865
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.9409364461898804,
      "learning_rate": 0.00036227467472249,
      "loss": 2.0629,
      "step": 99866
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.8589962124824524,
      "learning_rate": 0.0003622706732569226,
      "loss": 2.3951,
      "step": 99867
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.8391345739364624,
      "learning_rate": 0.0003622666717797779,
      "loss": 2.1615,
      "step": 99868
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.8796704411506653,
      "learning_rate": 0.00036226267029105673,
      "loss": 2.4026,
      "step": 99869
    },
    {
      "epoch": 1.3,
      "grad_norm": 1.0238815546035767,
      "learning_rate": 0.00036225866879075966,
      "loss": 2.4593,
      "step": 99870
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.7789070010185242,
      "learning_rate": 0.00036225466727888753,
      "loss": 2.0735,
      "step": 99871
    },
    {
      "epoch": 1.3,
      "grad_norm": 1.1545276641845703,
      "learning_rate": 0.00036225066575544106,
      "loss": 2.2414,
      "step": 99872
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.8428651690483093,
      "learning_rate": 0.00036224666422042117,
      "loss": 2.1257,
      "step": 99873
    },
    {
      "epoch": 1.3,
      "grad_norm": 4.98989725112915,
      "learning_rate": 0.00036224266267382833,
      "loss": 2.1775,
      "step": 99874
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.9468457102775574,
      "learning_rate": 0.0003622386611156634,
      "loss": 2.174,
      "step": 99875
    },
    {
      "epoch": 1.3,
      "grad_norm": 1.3829833269119263,
      "learning_rate": 0.0003622346595459272,
      "loss": 2.1642,
      "step": 99876
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.8622170090675354,
      "learning_rate": 0.0003622306579646203,
      "loss": 2.0561,
      "step": 99877
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.8928980827331543,
      "learning_rate": 0.0003622266563717436,
      "loss": 2.2349,
      "step": 99878
    },
    {
      "epoch": 1.3,
      "grad_norm": 1.0167330503463745,
      "learning_rate": 0.00036222265476729777,
      "loss": 2.2471,
      "step": 99879
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.9167066216468811,
      "learning_rate": 0.00036221865315128364,
      "loss": 2.0992,
      "step": 99880
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.9536371827125549,
      "learning_rate": 0.00036221465152370177,
      "loss": 2.128,
      "step": 99881
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.9788069128990173,
      "learning_rate": 0.00036221064988455314,
      "loss": 2.2454,
      "step": 99882
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.9764955639839172,
      "learning_rate": 0.00036220664823383836,
      "loss": 2.2639,
      "step": 99883
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.8042130470275879,
      "learning_rate": 0.0003622026465715582,
      "loss": 2.0629,
      "step": 99884
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.9473218321800232,
      "learning_rate": 0.00036219864489771324,
      "loss": 2.1653,
      "step": 99885
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.989608645439148,
      "learning_rate": 0.00036219464321230447,
      "loss": 2.0052,
      "step": 99886
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.8610125184059143,
      "learning_rate": 0.00036219064151533264,
      "loss": 2.0352,
      "step": 99887
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.8866193294525146,
      "learning_rate": 0.00036218663980679827,
      "loss": 2.1052,
      "step": 99888
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.8533074259757996,
      "learning_rate": 0.00036218263808670235,
      "loss": 2.2284,
      "step": 99889
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.9378204941749573,
      "learning_rate": 0.00036217863635504543,
      "loss": 2.0938,
      "step": 99890
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.9211593866348267,
      "learning_rate": 0.0003621746346118283,
      "loss": 2.2275,
      "step": 99891
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.8649412989616394,
      "learning_rate": 0.00036217063285705174,
      "loss": 2.3174,
      "step": 99892
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.9367876052856445,
      "learning_rate": 0.00036216663109071655,
      "loss": 2.2324,
      "step": 99893
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.9137036800384521,
      "learning_rate": 0.0003621626293128234,
      "loss": 2.2581,
      "step": 99894
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.9167113304138184,
      "learning_rate": 0.00036215862752337296,
      "loss": 2.2646,
      "step": 99895
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.9155150651931763,
      "learning_rate": 0.0003621546257223661,
      "loss": 2.1095,
      "step": 99896
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.8489274978637695,
      "learning_rate": 0.00036215062390980353,
      "loss": 2.0024,
      "step": 99897
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.8980873823165894,
      "learning_rate": 0.000362146622085686,
      "loss": 2.286,
      "step": 99898
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.8250086307525635,
      "learning_rate": 0.0003621426202500143,
      "loss": 1.9879,
      "step": 99899
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.7694607973098755,
      "learning_rate": 0.000362138618402789,
      "loss": 2.0843,
      "step": 99900
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.8355188369750977,
      "learning_rate": 0.000362134616544011,
      "loss": 2.0375,
      "step": 99901
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.9520686268806458,
      "learning_rate": 0.000362130614673681,
      "loss": 2.082,
      "step": 99902
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.8908504843711853,
      "learning_rate": 0.0003621266127917997,
      "loss": 2.2594,
      "step": 99903
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.8946961760520935,
      "learning_rate": 0.00036212261089836796,
      "loss": 2.0662,
      "step": 99904
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.8832141757011414,
      "learning_rate": 0.0003621186089933865,
      "loss": 2.2935,
      "step": 99905
    },
    {
      "epoch": 1.3,
      "grad_norm": 1.214857816696167,
      "learning_rate": 0.00036211460707685593,
      "loss": 2.1752,
      "step": 99906
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.9393501877784729,
      "learning_rate": 0.0003621106051487771,
      "loss": 2.2778,
      "step": 99907
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.8635163307189941,
      "learning_rate": 0.0003621066032091507,
      "loss": 1.9858,
      "step": 99908
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.9745829105377197,
      "learning_rate": 0.0003621026012579776,
      "loss": 2.0978,
      "step": 99909
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.9774081110954285,
      "learning_rate": 0.00036209859929525837,
      "loss": 2.1144,
      "step": 99910
    },
    {
      "epoch": 1.3,
      "grad_norm": 1.094225287437439,
      "learning_rate": 0.00036209459732099386,
      "loss": 2.1239,
      "step": 99911
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.9288245439529419,
      "learning_rate": 0.0003620905953351848,
      "loss": 2.2903,
      "step": 99912
    },
    {
      "epoch": 1.3,
      "grad_norm": 1.042142629623413,
      "learning_rate": 0.00036208659333783195,
      "loss": 1.9069,
      "step": 99913
    },
    {
      "epoch": 1.3,
      "grad_norm": 1.0727678537368774,
      "learning_rate": 0.000362082591328936,
      "loss": 2.3023,
      "step": 99914
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.993982195854187,
      "learning_rate": 0.0003620785893084978,
      "loss": 2.3036,
      "step": 99915
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.9428754448890686,
      "learning_rate": 0.0003620745872765179,
      "loss": 2.0403,
      "step": 99916
    },
    {
      "epoch": 1.3,
      "grad_norm": 1.0119683742523193,
      "learning_rate": 0.00036207058523299724,
      "loss": 2.207,
      "step": 99917
    },
    {
      "epoch": 1.3,
      "grad_norm": 1.2038826942443848,
      "learning_rate": 0.00036206658317793647,
      "loss": 2.3349,
      "step": 99918
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.8417060375213623,
      "learning_rate": 0.0003620625811113363,
      "loss": 1.8634,
      "step": 99919
    },
    {
      "epoch": 1.3,
      "grad_norm": 1.035214900970459,
      "learning_rate": 0.00036205857903319765,
      "loss": 2.0454,
      "step": 99920
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.8281021118164062,
      "learning_rate": 0.00036205457694352095,
      "loss": 2.2955,
      "step": 99921
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.9117532968521118,
      "learning_rate": 0.00036205057484230726,
      "loss": 2.2708,
      "step": 99922
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.8722087144851685,
      "learning_rate": 0.0003620465727295572,
      "loss": 2.2336,
      "step": 99923
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.8143497109413147,
      "learning_rate": 0.0003620425706052715,
      "loss": 2.0866,
      "step": 99924
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.8845544457435608,
      "learning_rate": 0.0003620385684694509,
      "loss": 2.2413,
      "step": 99925
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.8479722142219543,
      "learning_rate": 0.0003620345663220961,
      "loss": 2.3639,
      "step": 99926
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.8588060736656189,
      "learning_rate": 0.000362030564163208,
      "loss": 2.0121,
      "step": 99927
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.8666947484016418,
      "learning_rate": 0.0003620265619927872,
      "loss": 2.2469,
      "step": 99928
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.8250753879547119,
      "learning_rate": 0.0003620225598108345,
      "loss": 1.9811,
      "step": 99929
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.9134896397590637,
      "learning_rate": 0.0003620185576173507,
      "loss": 2.2271,
      "step": 99930
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.8639379739761353,
      "learning_rate": 0.00036201455541233637,
      "loss": 2.1564,
      "step": 99931
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.8574692010879517,
      "learning_rate": 0.00036201055319579235,
      "loss": 2.2035,
      "step": 99932
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.8709779381752014,
      "learning_rate": 0.00036200655096771953,
      "loss": 2.0045,
      "step": 99933
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.904033362865448,
      "learning_rate": 0.00036200254872811844,
      "loss": 2.0298,
      "step": 99934
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.8933192491531372,
      "learning_rate": 0.0003619985464769899,
      "loss": 2.2439,
      "step": 99935
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.8435684442520142,
      "learning_rate": 0.0003619945442143347,
      "loss": 2.1256,
      "step": 99936
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.8344502449035645,
      "learning_rate": 0.0003619905419401535,
      "loss": 2.1607,
      "step": 99937
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.827621579170227,
      "learning_rate": 0.0003619865396544471,
      "loss": 1.938,
      "step": 99938
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.9039148092269897,
      "learning_rate": 0.0003619825373572163,
      "loss": 2.2241,
      "step": 99939
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.8975744843482971,
      "learning_rate": 0.00036197853504846164,
      "loss": 2.2664,
      "step": 99940
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.8347770571708679,
      "learning_rate": 0.00036197453272818414,
      "loss": 2.2806,
      "step": 99941
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.9662255644798279,
      "learning_rate": 0.00036197053039638434,
      "loss": 1.9728,
      "step": 99942
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.8487005233764648,
      "learning_rate": 0.00036196652805306305,
      "loss": 2.1066,
      "step": 99943
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.9401911497116089,
      "learning_rate": 0.00036196252569822097,
      "loss": 1.9977,
      "step": 99944
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.8273640275001526,
      "learning_rate": 0.000361958523331859,
      "loss": 1.9861,
      "step": 99945
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.8153151869773865,
      "learning_rate": 0.00036195452095397765,
      "loss": 1.974,
      "step": 99946
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.8360571265220642,
      "learning_rate": 0.0003619505185645778,
      "loss": 2.0171,
      "step": 99947
    },
    {
      "epoch": 1.3,
      "grad_norm": 1.0350593328475952,
      "learning_rate": 0.00036194651616366024,
      "loss": 2.4107,
      "step": 99948
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.8845027685165405,
      "learning_rate": 0.00036194251375122567,
      "loss": 1.9594,
      "step": 99949
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.8416405320167542,
      "learning_rate": 0.0003619385113272747,
      "loss": 2.3096,
      "step": 99950
    },
    {
      "epoch": 1.3,
      "grad_norm": 1.0308068990707397,
      "learning_rate": 0.0003619345088918083,
      "loss": 2.0249,
      "step": 99951
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.8795455694198608,
      "learning_rate": 0.0003619305064448271,
      "loss": 2.1722,
      "step": 99952
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.8951412439346313,
      "learning_rate": 0.00036192650398633177,
      "loss": 2.1811,
      "step": 99953
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.8217231631278992,
      "learning_rate": 0.00036192250151632324,
      "loss": 2.1851,
      "step": 99954
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.9283117651939392,
      "learning_rate": 0.00036191849903480206,
      "loss": 2.2616,
      "step": 99955
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.9545412063598633,
      "learning_rate": 0.0003619144965417691,
      "loss": 2.3086,
      "step": 99956
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.8509579300880432,
      "learning_rate": 0.0003619104940372251,
      "loss": 2.1815,
      "step": 99957
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.8927929401397705,
      "learning_rate": 0.00036190649152117066,
      "loss": 2.2354,
      "step": 99958
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.9028434753417969,
      "learning_rate": 0.0003619024889936068,
      "loss": 2.1706,
      "step": 99959
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.9856845140457153,
      "learning_rate": 0.0003618984864545339,
      "loss": 2.1909,
      "step": 99960
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.8244068026542664,
      "learning_rate": 0.00036189448390395307,
      "loss": 2.2312,
      "step": 99961
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.9122551679611206,
      "learning_rate": 0.00036189048134186477,
      "loss": 2.4625,
      "step": 99962
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.97478187084198,
      "learning_rate": 0.00036188647876826994,
      "loss": 2.0646,
      "step": 99963
    },
    {
      "epoch": 1.3,
      "grad_norm": 1.0545161962509155,
      "learning_rate": 0.00036188247618316923,
      "loss": 2.0872,
      "step": 99964
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.8739042282104492,
      "learning_rate": 0.00036187847358656346,
      "loss": 2.1737,
      "step": 99965
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.8779751658439636,
      "learning_rate": 0.0003618744709784532,
      "loss": 2.3096,
      "step": 99966
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.9446209073066711,
      "learning_rate": 0.0003618704683588393,
      "loss": 2.1305,
      "step": 99967
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.9328943490982056,
      "learning_rate": 0.00036186646572772263,
      "loss": 2.3254,
      "step": 99968
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.8983559012413025,
      "learning_rate": 0.0003618624630851037,
      "loss": 2.024,
      "step": 99969
    },
    {
      "epoch": 1.3,
      "grad_norm": 1.1345707178115845,
      "learning_rate": 0.00036185846043098344,
      "loss": 2.2897,
      "step": 99970
    },
    {
      "epoch": 1.3,
      "grad_norm": 1.0342944860458374,
      "learning_rate": 0.00036185445776536245,
      "loss": 1.9415,
      "step": 99971
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.9056568145751953,
      "learning_rate": 0.00036185045508824165,
      "loss": 2.0578,
      "step": 99972
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.9460691213607788,
      "learning_rate": 0.00036184645239962165,
      "loss": 2.1506,
      "step": 99973
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.9589000344276428,
      "learning_rate": 0.0003618424496995032,
      "loss": 2.2166,
      "step": 99974
    },
    {
      "epoch": 1.3,
      "grad_norm": 1.0088768005371094,
      "learning_rate": 0.0003618384469878871,
      "loss": 2.1261,
      "step": 99975
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.9311256408691406,
      "learning_rate": 0.00036183444426477406,
      "loss": 2.484,
      "step": 99976
    },
    {
      "epoch": 1.3,
      "grad_norm": 1.1204140186309814,
      "learning_rate": 0.0003618304415301648,
      "loss": 2.2947,
      "step": 99977
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.886494517326355,
      "learning_rate": 0.0003618264387840601,
      "loss": 2.0753,
      "step": 99978
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.8886162638664246,
      "learning_rate": 0.0003618224360264608,
      "loss": 2.2042,
      "step": 99979
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.9493678212165833,
      "learning_rate": 0.0003618184332573674,
      "loss": 2.1278,
      "step": 99980
    },
    {
      "epoch": 1.3,
      "grad_norm": 1.021430492401123,
      "learning_rate": 0.00036181443047678087,
      "loss": 2.4063,
      "step": 99981
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.8400924205780029,
      "learning_rate": 0.0003618104276847018,
      "loss": 1.9526,
      "step": 99982
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.9372426271438599,
      "learning_rate": 0.0003618064248811311,
      "loss": 2.0756,
      "step": 99983
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.8930134773254395,
      "learning_rate": 0.0003618024220660693,
      "loss": 2.1345,
      "step": 99984
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.8431650996208191,
      "learning_rate": 0.0003617984192395174,
      "loss": 2.1001,
      "step": 99985
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.8210866451263428,
      "learning_rate": 0.0003617944164014759,
      "loss": 2.3051,
      "step": 99986
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.8235087990760803,
      "learning_rate": 0.0003617904135519457,
      "loss": 2.038,
      "step": 99987
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.8603664636611938,
      "learning_rate": 0.00036178641069092746,
      "loss": 2.2784,
      "step": 99988
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.8527628779411316,
      "learning_rate": 0.000361782407818422,
      "loss": 1.9735,
      "step": 99989
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.9167450070381165,
      "learning_rate": 0.00036177840493443,
      "loss": 2.2336,
      "step": 99990
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.8429102897644043,
      "learning_rate": 0.00036177440203895224,
      "loss": 2.233,
      "step": 99991
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.8375817537307739,
      "learning_rate": 0.0003617703991319894,
      "loss": 2.2912,
      "step": 99992
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.9353029131889343,
      "learning_rate": 0.00036176639621354235,
      "loss": 1.973,
      "step": 99993
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.8555074334144592,
      "learning_rate": 0.0003617623932836118,
      "loss": 2.1479,
      "step": 99994
    },
    {
      "epoch": 1.3,
      "grad_norm": 1.1162596940994263,
      "learning_rate": 0.0003617583903421984,
      "loss": 2.3155,
      "step": 99995
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.9349202513694763,
      "learning_rate": 0.00036175438738930287,
      "loss": 2.2594,
      "step": 99996
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.8596281409263611,
      "learning_rate": 0.0003617503844249262,
      "loss": 2.0493,
      "step": 99997
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.8317598104476929,
      "learning_rate": 0.0003617463814490688,
      "loss": 2.2837,
      "step": 99998
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.8080947995185852,
      "learning_rate": 0.00036174237846173165,
      "loss": 2.0327,
      "step": 99999
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.8956518769264221,
      "learning_rate": 0.00036173837546291544,
      "loss": 2.0296,
      "step": 100000
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.8816113471984863,
      "learning_rate": 0.00036173437245262094,
      "loss": 2.5344,
      "step": 100001
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.8557977676391602,
      "learning_rate": 0.00036173036943084875,
      "loss": 2.0341,
      "step": 100002
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.8239787817001343,
      "learning_rate": 0.0003617263663975998,
      "loss": 2.2437,
      "step": 100003
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.9081307053565979,
      "learning_rate": 0.0003617223633528747,
      "loss": 2.0297,
      "step": 100004
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.9494733810424805,
      "learning_rate": 0.0003617183602966743,
      "loss": 2.3704,
      "step": 100005
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.8187143802642822,
      "learning_rate": 0.0003617143572289993,
      "loss": 1.8787,
      "step": 100006
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.8358161449432373,
      "learning_rate": 0.0003617103541498504,
      "loss": 2.0403,
      "step": 100007
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.9082042574882507,
      "learning_rate": 0.00036170635105922834,
      "loss": 2.1878,
      "step": 100008
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.8949214816093445,
      "learning_rate": 0.000361702347957134,
      "loss": 2.0565,
      "step": 100009
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.944190263748169,
      "learning_rate": 0.0003616983448435679,
      "loss": 2.2938,
      "step": 100010
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.8727756142616272,
      "learning_rate": 0.000361694341718531,
      "loss": 1.9798,
      "step": 100011
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.7742614150047302,
      "learning_rate": 0.00036169033858202403,
      "loss": 2.1068,
      "step": 100012
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.8348312973976135,
      "learning_rate": 0.00036168633543404757,
      "loss": 2.2388,
      "step": 100013
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.8630173802375793,
      "learning_rate": 0.0003616823322746024,
      "loss": 2.1411,
      "step": 100014
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.8856641054153442,
      "learning_rate": 0.00036167832910368944,
      "loss": 2.0013,
      "step": 100015
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.8337949514389038,
      "learning_rate": 0.0003616743259213093,
      "loss": 1.8455,
      "step": 100016
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.9227664470672607,
      "learning_rate": 0.00036167032272746267,
      "loss": 2.0997,
      "step": 100017
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.830075740814209,
      "learning_rate": 0.0003616663195221504,
      "loss": 2.2483,
      "step": 100018
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.8417523503303528,
      "learning_rate": 0.00036166231630537323,
      "loss": 2.1847,
      "step": 100019
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.8374970555305481,
      "learning_rate": 0.00036165831307713176,
      "loss": 2.2365,
      "step": 100020
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.8278445601463318,
      "learning_rate": 0.00036165430983742697,
      "loss": 2.0401,
      "step": 100021
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.825923502445221,
      "learning_rate": 0.0003616503065862594,
      "loss": 2.2505,
      "step": 100022
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.8316976428031921,
      "learning_rate": 0.0003616463033236299,
      "loss": 2.2038,
      "step": 100023
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.8303305506706238,
      "learning_rate": 0.0003616423000495392,
      "loss": 2.3578,
      "step": 100024
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.8861360549926758,
      "learning_rate": 0.000361638296763988,
      "loss": 2.3182,
      "step": 100025
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.7613140344619751,
      "learning_rate": 0.00036163429346697707,
      "loss": 2.2312,
      "step": 100026
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.9223392605781555,
      "learning_rate": 0.00036163029015850724,
      "loss": 2.3828,
      "step": 100027
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.9494432210922241,
      "learning_rate": 0.00036162628683857913,
      "loss": 2.2602,
      "step": 100028
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.9747969508171082,
      "learning_rate": 0.0003616222835071935,
      "loss": 2.4039,
      "step": 100029
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.8813138008117676,
      "learning_rate": 0.0003616182801643512,
      "loss": 2.1182,
      "step": 100030
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.9691392183303833,
      "learning_rate": 0.0003616142768100528,
      "loss": 1.9986,
      "step": 100031
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.8406317234039307,
      "learning_rate": 0.00036161027344429916,
      "loss": 2.054,
      "step": 100032
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.8806326985359192,
      "learning_rate": 0.00036160627006709113,
      "loss": 2.2941,
      "step": 100033
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.7944072484970093,
      "learning_rate": 0.0003616022666784292,
      "loss": 2.2855,
      "step": 100034
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.8129589557647705,
      "learning_rate": 0.0003615982632783143,
      "loss": 2.2791,
      "step": 100035
    },
    {
      "epoch": 1.3,
      "grad_norm": 1.211402416229248,
      "learning_rate": 0.0003615942598667471,
      "loss": 2.1347,
      "step": 100036
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.841722846031189,
      "learning_rate": 0.00036159025644372835,
      "loss": 2.0938,
      "step": 100037
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.9212601184844971,
      "learning_rate": 0.00036158625300925884,
      "loss": 2.2359,
      "step": 100038
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.8355740904808044,
      "learning_rate": 0.00036158224956333934,
      "loss": 2.3462,
      "step": 100039
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.9282846450805664,
      "learning_rate": 0.00036157824610597044,
      "loss": 2.0933,
      "step": 100040
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.8452036380767822,
      "learning_rate": 0.00036157424263715297,
      "loss": 2.4122,
      "step": 100041
    },
    {
      "epoch": 1.3,
      "grad_norm": 3.5344414710998535,
      "learning_rate": 0.0003615702391568878,
      "loss": 2.2695,
      "step": 100042
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.8405085802078247,
      "learning_rate": 0.0003615662356651755,
      "loss": 2.106,
      "step": 100043
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.8942487239837646,
      "learning_rate": 0.00036156223216201683,
      "loss": 2.1925,
      "step": 100044
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.9658867716789246,
      "learning_rate": 0.00036155822864741263,
      "loss": 2.0571,
      "step": 100045
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.8944794535636902,
      "learning_rate": 0.00036155422512136354,
      "loss": 2.038,
      "step": 100046
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.8408738374710083,
      "learning_rate": 0.00036155022158387037,
      "loss": 2.2433,
      "step": 100047
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.8971253037452698,
      "learning_rate": 0.0003615462180349339,
      "loss": 2.1707,
      "step": 100048
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.9010732173919678,
      "learning_rate": 0.0003615422144745548,
      "loss": 2.0464,
      "step": 100049
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.886873185634613,
      "learning_rate": 0.00036153821090273387,
      "loss": 2.2889,
      "step": 100050
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.86909419298172,
      "learning_rate": 0.00036153420731947176,
      "loss": 2.2969,
      "step": 100051
    },
    {
      "epoch": 1.3,
      "grad_norm": 1.0223015546798706,
      "learning_rate": 0.00036153020372476925,
      "loss": 2.274,
      "step": 100052
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.8645559549331665,
      "learning_rate": 0.0003615262001186272,
      "loss": 2.1434,
      "step": 100053
    },
    {
      "epoch": 1.3,
      "grad_norm": 1.0459915399551392,
      "learning_rate": 0.00036152219650104625,
      "loss": 2.2645,
      "step": 100054
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.9644142389297485,
      "learning_rate": 0.0003615181928720271,
      "loss": 2.1578,
      "step": 100055
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.8541263937950134,
      "learning_rate": 0.0003615141892315707,
      "loss": 2.2894,
      "step": 100056
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.9058852791786194,
      "learning_rate": 0.00036151018557967755,
      "loss": 2.3226,
      "step": 100057
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.8887525200843811,
      "learning_rate": 0.00036150618191634846,
      "loss": 2.2139,
      "step": 100058
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.8533379435539246,
      "learning_rate": 0.0003615021782415843,
      "loss": 2.1599,
      "step": 100059
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.8048311471939087,
      "learning_rate": 0.00036149817455538574,
      "loss": 2.3121,
      "step": 100060
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.846230685710907,
      "learning_rate": 0.0003614941708577534,
      "loss": 2.258,
      "step": 100061
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.8774369955062866,
      "learning_rate": 0.00036149016714868813,
      "loss": 2.0557,
      "step": 100062
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.9481293559074402,
      "learning_rate": 0.00036148616342819074,
      "loss": 2.3042,
      "step": 100063
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.8546720743179321,
      "learning_rate": 0.00036148215969626196,
      "loss": 2.0485,
      "step": 100064
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.8546949625015259,
      "learning_rate": 0.00036147815595290243,
      "loss": 2.1755,
      "step": 100065
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.9661757349967957,
      "learning_rate": 0.0003614741521981129,
      "loss": 2.1656,
      "step": 100066
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.8192513585090637,
      "learning_rate": 0.0003614701484318943,
      "loss": 2.1694,
      "step": 100067
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.8821629881858826,
      "learning_rate": 0.00036146614465424714,
      "loss": 1.9249,
      "step": 100068
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.8057874441146851,
      "learning_rate": 0.0003614621408651723,
      "loss": 2.1101,
      "step": 100069
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.8579088449478149,
      "learning_rate": 0.0003614581370646705,
      "loss": 2.3818,
      "step": 100070
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.8642517328262329,
      "learning_rate": 0.0003614541332527424,
      "loss": 2.1787,
      "step": 100071
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.9367608428001404,
      "learning_rate": 0.0003614501294293888,
      "loss": 2.1547,
      "step": 100072
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.977006733417511,
      "learning_rate": 0.00036144612559461063,
      "loss": 2.2806,
      "step": 100073
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.8901026844978333,
      "learning_rate": 0.0003614421217484083,
      "loss": 2.0379,
      "step": 100074
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.898546040058136,
      "learning_rate": 0.0003614381178907827,
      "loss": 2.3948,
      "step": 100075
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.9687163829803467,
      "learning_rate": 0.00036143411402173476,
      "loss": 2.2539,
      "step": 100076
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.9041305184364319,
      "learning_rate": 0.00036143011014126496,
      "loss": 1.9312,
      "step": 100077
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.8760011196136475,
      "learning_rate": 0.00036142610624937413,
      "loss": 2.399,
      "step": 100078
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.8998295068740845,
      "learning_rate": 0.0003614221023460631,
      "loss": 2.2017,
      "step": 100079
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.8345619440078735,
      "learning_rate": 0.0003614180984313324,
      "loss": 2.3093,
      "step": 100080
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.9109202027320862,
      "learning_rate": 0.00036141409450518304,
      "loss": 2.1308,
      "step": 100081
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.9056165218353271,
      "learning_rate": 0.00036141009056761563,
      "loss": 2.1976,
      "step": 100082
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.906167209148407,
      "learning_rate": 0.00036140608661863093,
      "loss": 2.4237,
      "step": 100083
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.8428558707237244,
      "learning_rate": 0.00036140208265822963,
      "loss": 2.0397,
      "step": 100084
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.8500540256500244,
      "learning_rate": 0.00036139807868641255,
      "loss": 2.1406,
      "step": 100085
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.8693564534187317,
      "learning_rate": 0.0003613940747031804,
      "loss": 2.2358,
      "step": 100086
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.9177345633506775,
      "learning_rate": 0.00036139007070853387,
      "loss": 2.2296,
      "step": 100087
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.8410706520080566,
      "learning_rate": 0.00036138606670247384,
      "loss": 2.0374,
      "step": 100088
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.8146323561668396,
      "learning_rate": 0.000361382062685001,
      "loss": 2.0347,
      "step": 100089
    },
    {
      "epoch": 1.3,
      "grad_norm": 1.0487756729125977,
      "learning_rate": 0.000361378058656116,
      "loss": 2.1236,
      "step": 100090
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.9765694737434387,
      "learning_rate": 0.00036137405461581976,
      "loss": 2.1539,
      "step": 100091
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.9516168832778931,
      "learning_rate": 0.0003613700505641128,
      "loss": 2.1684,
      "step": 100092
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.8436043858528137,
      "learning_rate": 0.0003613660465009961,
      "loss": 2.2293,
      "step": 100093
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.7995219230651855,
      "learning_rate": 0.00036136204242647027,
      "loss": 1.9615,
      "step": 100094
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.8173941373825073,
      "learning_rate": 0.0003613580383405361,
      "loss": 1.9641,
      "step": 100095
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.9341987371444702,
      "learning_rate": 0.00036135403424319425,
      "loss": 2.1938,
      "step": 100096
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.8375560641288757,
      "learning_rate": 0.0003613500301344456,
      "loss": 2.2637,
      "step": 100097
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.9214732646942139,
      "learning_rate": 0.0003613460260142908,
      "loss": 2.208,
      "step": 100098
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.9248552322387695,
      "learning_rate": 0.0003613420218827305,
      "loss": 2.2984,
      "step": 100099
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.7924216389656067,
      "learning_rate": 0.0003613380177397657,
      "loss": 2.3247,
      "step": 100100
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.8489139080047607,
      "learning_rate": 0.0003613340135853969,
      "loss": 2.0923,
      "step": 100101
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.8607426881790161,
      "learning_rate": 0.000361330009419625,
      "loss": 2.3699,
      "step": 100102
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.9627777338027954,
      "learning_rate": 0.00036132600524245076,
      "loss": 2.0346,
      "step": 100103
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.7982950210571289,
      "learning_rate": 0.00036132200105387484,
      "loss": 2.1904,
      "step": 100104
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.8186106085777283,
      "learning_rate": 0.00036131799685389787,
      "loss": 2.3195,
      "step": 100105
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.8322160840034485,
      "learning_rate": 0.0003613139926425209,
      "loss": 2.0342,
      "step": 100106
    },
    {
      "epoch": 1.3,
      "grad_norm": 1.0573927164077759,
      "learning_rate": 0.0003613099884197444,
      "loss": 2.1838,
      "step": 100107
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.9283764362335205,
      "learning_rate": 0.00036130598418556917,
      "loss": 2.2261,
      "step": 100108
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.8721290826797485,
      "learning_rate": 0.0003613019799399961,
      "loss": 2.0405,
      "step": 100109
    },
    {
      "epoch": 1.3,
      "grad_norm": 1.0613104104995728,
      "learning_rate": 0.0003612979756830258,
      "loss": 2.0387,
      "step": 100110
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.9376421570777893,
      "learning_rate": 0.000361293971414659,
      "loss": 2.1572,
      "step": 100111
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.9166694283485413,
      "learning_rate": 0.0003612899671348966,
      "loss": 2.1291,
      "step": 100112
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.9704486727714539,
      "learning_rate": 0.0003612859628437392,
      "loss": 1.9894,
      "step": 100113
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.8476982116699219,
      "learning_rate": 0.00036128195854118753,
      "loss": 2.1547,
      "step": 100114
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.8609374165534973,
      "learning_rate": 0.00036127795422724243,
      "loss": 1.9155,
      "step": 100115
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.7854978442192078,
      "learning_rate": 0.0003612739499019046,
      "loss": 2.1706,
      "step": 100116
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.9170974493026733,
      "learning_rate": 0.00036126994556517477,
      "loss": 2.1288,
      "step": 100117
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.9772602319717407,
      "learning_rate": 0.00036126594121705367,
      "loss": 2.3059,
      "step": 100118
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.8510516881942749,
      "learning_rate": 0.0003612619368575421,
      "loss": 2.2652,
      "step": 100119
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.9105267524719238,
      "learning_rate": 0.00036125793248664083,
      "loss": 2.1721,
      "step": 100120
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.859300971031189,
      "learning_rate": 0.00036125392810435055,
      "loss": 1.9913,
      "step": 100121
    },
    {
      "epoch": 1.3,
      "grad_norm": 1.0391803979873657,
      "learning_rate": 0.000361249923710672,
      "loss": 2.1114,
      "step": 100122
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.9275705814361572,
      "learning_rate": 0.0003612459193056058,
      "loss": 2.1212,
      "step": 100123
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.8654230237007141,
      "learning_rate": 0.00036124191488915293,
      "loss": 2.115,
      "step": 100124
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.8185024857521057,
      "learning_rate": 0.00036123791046131407,
      "loss": 1.9697,
      "step": 100125
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.9999626278877258,
      "learning_rate": 0.00036123390602208985,
      "loss": 2.1377,
      "step": 100126
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.8876969218254089,
      "learning_rate": 0.0003612299015714812,
      "loss": 2.084,
      "step": 100127
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.8498640656471252,
      "learning_rate": 0.0003612258971094887,
      "loss": 2.1258,
      "step": 100128
    },
    {
      "epoch": 1.3,
      "grad_norm": 1.207594633102417,
      "learning_rate": 0.00036122189263611314,
      "loss": 2.1609,
      "step": 100129
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.8834888935089111,
      "learning_rate": 0.0003612178881513552,
      "loss": 2.2261,
      "step": 100130
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.8905568718910217,
      "learning_rate": 0.0003612138836552158,
      "loss": 2.1333,
      "step": 100131
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.9919979572296143,
      "learning_rate": 0.0003612098791476956,
      "loss": 2.1691,
      "step": 100132
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.9428267478942871,
      "learning_rate": 0.00036120587462879525,
      "loss": 2.2707,
      "step": 100133
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.9258954524993896,
      "learning_rate": 0.00036120187009851566,
      "loss": 2.0024,
      "step": 100134
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.917009174823761,
      "learning_rate": 0.00036119786555685737,
      "loss": 2.4945,
      "step": 100135
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.9540759921073914,
      "learning_rate": 0.0003611938610038213,
      "loss": 2.2515,
      "step": 100136
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.9274119138717651,
      "learning_rate": 0.0003611898564394082,
      "loss": 2.0397,
      "step": 100137
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.9642177820205688,
      "learning_rate": 0.00036118585186361863,
      "loss": 2.1264,
      "step": 100138
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.9271225333213806,
      "learning_rate": 0.00036118184727645356,
      "loss": 2.0845,
      "step": 100139
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.8416492938995361,
      "learning_rate": 0.0003611778426779135,
      "loss": 2.0663,
      "step": 100140
    },
    {
      "epoch": 1.3,
      "grad_norm": 1.092215895652771,
      "learning_rate": 0.0003611738380679995,
      "loss": 2.1137,
      "step": 100141
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.8578045964241028,
      "learning_rate": 0.000361169833446712,
      "loss": 2.0857,
      "step": 100142
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.8555128574371338,
      "learning_rate": 0.00036116582881405194,
      "loss": 2.4231,
      "step": 100143
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.8878229260444641,
      "learning_rate": 0.0003611618241700199,
      "loss": 2.2359,
      "step": 100144
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.9304516315460205,
      "learning_rate": 0.00036115781951461684,
      "loss": 2.0668,
      "step": 100145
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.811011552810669,
      "learning_rate": 0.0003611538148478433,
      "loss": 1.958,
      "step": 100146
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.8281837701797485,
      "learning_rate": 0.00036114981016970016,
      "loss": 2.0705,
      "step": 100147
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.9097738862037659,
      "learning_rate": 0.00036114580548018814,
      "loss": 2.2212,
      "step": 100148
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.8692024946212769,
      "learning_rate": 0.0003611418007793079,
      "loss": 2.2868,
      "step": 100149
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.8393908739089966,
      "learning_rate": 0.0003611377960670603,
      "loss": 2.1421,
      "step": 100150
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.9071306586265564,
      "learning_rate": 0.00036113379134344593,
      "loss": 2.4907,
      "step": 100151
    },
    {
      "epoch": 1.3,
      "grad_norm": 1.0246450901031494,
      "learning_rate": 0.0003611297866084658,
      "loss": 2.4851,
      "step": 100152
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.9448847770690918,
      "learning_rate": 0.00036112578186212036,
      "loss": 2.1736,
      "step": 100153
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.9108692407608032,
      "learning_rate": 0.00036112177710441044,
      "loss": 2.0601,
      "step": 100154
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.8974342942237854,
      "learning_rate": 0.000361117772335337,
      "loss": 2.2129,
      "step": 100155
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.844965398311615,
      "learning_rate": 0.0003611137675549005,
      "loss": 2.2446,
      "step": 100156
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.7865193486213684,
      "learning_rate": 0.00036110976276310184,
      "loss": 2.3198,
      "step": 100157
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.8409500122070312,
      "learning_rate": 0.0003611057579599417,
      "loss": 2.1249,
      "step": 100158
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.8454638719558716,
      "learning_rate": 0.0003611017531454208,
      "loss": 2.3474,
      "step": 100159
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.9159564971923828,
      "learning_rate": 0.00036109774831953996,
      "loss": 2.0577,
      "step": 100160
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.8778267502784729,
      "learning_rate": 0.00036109374348229997,
      "loss": 2.1183,
      "step": 100161
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.9335465431213379,
      "learning_rate": 0.00036108973863370144,
      "loss": 2.1174,
      "step": 100162
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.8366283774375916,
      "learning_rate": 0.0003610857337737452,
      "loss": 2.1132,
      "step": 100163
    },
    {
      "epoch": 1.3,
      "grad_norm": 1.049196720123291,
      "learning_rate": 0.0003610817289024319,
      "loss": 2.1859,
      "step": 100164
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.9194693565368652,
      "learning_rate": 0.00036107772401976245,
      "loss": 2.1078,
      "step": 100165
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.906396210193634,
      "learning_rate": 0.00036107371912573737,
      "loss": 2.2158,
      "step": 100166
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.8770772814750671,
      "learning_rate": 0.0003610697142203577,
      "loss": 2.005,
      "step": 100167
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.9384452104568481,
      "learning_rate": 0.0003610657093036239,
      "loss": 2.2409,
      "step": 100168
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.9294329881668091,
      "learning_rate": 0.0003610617043755368,
      "loss": 2.176,
      "step": 100169
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.9457398653030396,
      "learning_rate": 0.0003610576994360973,
      "loss": 1.9603,
      "step": 100170
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.8375750780105591,
      "learning_rate": 0.0003610536944853059,
      "loss": 1.8958,
      "step": 100171
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.8206568956375122,
      "learning_rate": 0.0003610496895231635,
      "loss": 2.091,
      "step": 100172
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.8797917366027832,
      "learning_rate": 0.0003610456845496709,
      "loss": 2.1448,
      "step": 100173
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.8788460493087769,
      "learning_rate": 0.00036104167956482866,
      "loss": 2.3232,
      "step": 100174
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.8506510257720947,
      "learning_rate": 0.00036103767456863764,
      "loss": 2.1427,
      "step": 100175
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.849999189376831,
      "learning_rate": 0.00036103366956109865,
      "loss": 2.1237,
      "step": 100176
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.8517388105392456,
      "learning_rate": 0.00036102966454221224,
      "loss": 2.1491,
      "step": 100177
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.9508448243141174,
      "learning_rate": 0.0003610256595119792,
      "loss": 2.0661,
      "step": 100178
    },
    {
      "epoch": 1.3,
      "grad_norm": 1.2409220933914185,
      "learning_rate": 0.0003610216544704005,
      "loss": 2.1524,
      "step": 100179
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.8719319105148315,
      "learning_rate": 0.0003610176494174766,
      "loss": 2.1204,
      "step": 100180
    },
    {
      "epoch": 1.3,
      "grad_norm": 1.3053869009017944,
      "learning_rate": 0.00036101364435320847,
      "loss": 2.0352,
      "step": 100181
    },
    {
      "epoch": 1.3,
      "grad_norm": 1.1940667629241943,
      "learning_rate": 0.0003610096392775967,
      "loss": 2.0768,
      "step": 100182
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.9784340858459473,
      "learning_rate": 0.0003610056341906421,
      "loss": 1.9761,
      "step": 100183
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.8718701601028442,
      "learning_rate": 0.00036100162909234534,
      "loss": 1.9905,
      "step": 100184
    },
    {
      "epoch": 1.3,
      "grad_norm": 1.0515385866165161,
      "learning_rate": 0.00036099762398270735,
      "loss": 2.0345,
      "step": 100185
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.8640998601913452,
      "learning_rate": 0.00036099361886172867,
      "loss": 2.3783,
      "step": 100186
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.9557546973228455,
      "learning_rate": 0.00036098961372941005,
      "loss": 2.3037,
      "step": 100187
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.820119321346283,
      "learning_rate": 0.00036098560858575246,
      "loss": 2.2234,
      "step": 100188
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.9816020727157593,
      "learning_rate": 0.0003609816034307564,
      "loss": 2.0708,
      "step": 100189
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.8408923149108887,
      "learning_rate": 0.0003609775982644227,
      "loss": 2.1979,
      "step": 100190
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.8218601942062378,
      "learning_rate": 0.00036097359308675213,
      "loss": 2.1203,
      "step": 100191
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.8919483423233032,
      "learning_rate": 0.00036096958789774545,
      "loss": 2.0352,
      "step": 100192
    },
    {
      "epoch": 1.3,
      "grad_norm": 1.0288059711456299,
      "learning_rate": 0.00036096558269740333,
      "loss": 2.1535,
      "step": 100193
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.9154008626937866,
      "learning_rate": 0.00036096157748572664,
      "loss": 2.0869,
      "step": 100194
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.8327580094337463,
      "learning_rate": 0.00036095757226271597,
      "loss": 2.2705,
      "step": 100195
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.9039774537086487,
      "learning_rate": 0.00036095356702837213,
      "loss": 2.1407,
      "step": 100196
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.9475288987159729,
      "learning_rate": 0.0003609495617826959,
      "loss": 2.2878,
      "step": 100197
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.7994703650474548,
      "learning_rate": 0.000360945556525688,
      "loss": 1.905,
      "step": 100198
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.8468812704086304,
      "learning_rate": 0.00036094155125734916,
      "loss": 2.1956,
      "step": 100199
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.9507272243499756,
      "learning_rate": 0.0003609375459776801,
      "loss": 2.2021,
      "step": 100200
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.9821213483810425,
      "learning_rate": 0.0003609335406866817,
      "loss": 1.9633,
      "step": 100201
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.7932247519493103,
      "learning_rate": 0.00036092953538435454,
      "loss": 2.2021,
      "step": 100202
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.9165907502174377,
      "learning_rate": 0.0003609255300706994,
      "loss": 2.1755,
      "step": 100203
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.9567089676856995,
      "learning_rate": 0.00036092152474571714,
      "loss": 1.8707,
      "step": 100204
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.7936774492263794,
      "learning_rate": 0.0003609175194094083,
      "loss": 2.1249,
      "step": 100205
    },
    {
      "epoch": 1.3,
      "grad_norm": 1.071326494216919,
      "learning_rate": 0.0003609135140617738,
      "loss": 2.1777,
      "step": 100206
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.9368844628334045,
      "learning_rate": 0.0003609095087028144,
      "loss": 2.0174,
      "step": 100207
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.9624311923980713,
      "learning_rate": 0.00036090550333253067,
      "loss": 2.1225,
      "step": 100208
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.9250324964523315,
      "learning_rate": 0.0003609014979509235,
      "loss": 2.3308,
      "step": 100209
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.8141917586326599,
      "learning_rate": 0.00036089749255799363,
      "loss": 2.2231,
      "step": 100210
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.9393648505210876,
      "learning_rate": 0.0003608934871537417,
      "loss": 2.129,
      "step": 100211
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.9033696055412292,
      "learning_rate": 0.0003608894817381685,
      "loss": 2.1973,
      "step": 100212
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.8644282817840576,
      "learning_rate": 0.0003608854763112749,
      "loss": 2.1148,
      "step": 100213
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.9086257815361023,
      "learning_rate": 0.0003608814708730614,
      "loss": 2.2165,
      "step": 100214
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.8474107980728149,
      "learning_rate": 0.00036087746542352893,
      "loss": 2.3565,
      "step": 100215
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.9221301674842834,
      "learning_rate": 0.0003608734599626783,
      "loss": 2.212,
      "step": 100216
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.9540925621986389,
      "learning_rate": 0.0003608694544905101,
      "loss": 2.2037,
      "step": 100217
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.8744568228721619,
      "learning_rate": 0.00036086544900702506,
      "loss": 2.4174,
      "step": 100218
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.9401612281799316,
      "learning_rate": 0.000360861443512224,
      "loss": 2.085,
      "step": 100219
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.9185319542884827,
      "learning_rate": 0.0003608574380061077,
      "loss": 2.451,
      "step": 100220
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.8863418102264404,
      "learning_rate": 0.00036085343248867677,
      "loss": 2.1165,
      "step": 100221
    },
    {
      "epoch": 1.3,
      "grad_norm": 1.0044636726379395,
      "learning_rate": 0.00036084942695993215,
      "loss": 2.1868,
      "step": 100222
    },
    {
      "epoch": 1.3,
      "grad_norm": 1.139681339263916,
      "learning_rate": 0.00036084542141987436,
      "loss": 2.068,
      "step": 100223
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.8380978107452393,
      "learning_rate": 0.0003608414158685043,
      "loss": 2.316,
      "step": 100224
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.9212021231651306,
      "learning_rate": 0.0003608374103058226,
      "loss": 2.2423,
      "step": 100225
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.8133242726325989,
      "learning_rate": 0.0003608334047318302,
      "loss": 1.9458,
      "step": 100226
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.9062836766242981,
      "learning_rate": 0.0003608293991465277,
      "loss": 2.3868,
      "step": 100227
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.856354296207428,
      "learning_rate": 0.0003608253935499158,
      "loss": 2.3132,
      "step": 100228
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.8608959913253784,
      "learning_rate": 0.0003608213879419954,
      "loss": 2.1952,
      "step": 100229
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.9645413756370544,
      "learning_rate": 0.00036081738232276707,
      "loss": 2.1046,
      "step": 100230
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.9701926708221436,
      "learning_rate": 0.0003608133766922317,
      "loss": 2.0479,
      "step": 100231
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.8298092484474182,
      "learning_rate": 0.0003608093710503899,
      "loss": 2.1951,
      "step": 100232
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.8303772807121277,
      "learning_rate": 0.00036080536539724264,
      "loss": 2.2717,
      "step": 100233
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.9405670762062073,
      "learning_rate": 0.0003608013597327904,
      "loss": 2.1795,
      "step": 100234
    },
    {
      "epoch": 1.31,
      "grad_norm": 1.009634017944336,
      "learning_rate": 0.000360797354057034,
      "loss": 2.2926,
      "step": 100235
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.9810323715209961,
      "learning_rate": 0.00036079334836997435,
      "loss": 2.1486,
      "step": 100236
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.8426530361175537,
      "learning_rate": 0.00036078934267161193,
      "loss": 1.9677,
      "step": 100237
    },
    {
      "epoch": 1.31,
      "grad_norm": 1.0312172174453735,
      "learning_rate": 0.00036078533696194773,
      "loss": 2.318,
      "step": 100238
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.9042968153953552,
      "learning_rate": 0.0003607813312409823,
      "loss": 2.4063,
      "step": 100239
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.9379608631134033,
      "learning_rate": 0.00036077732550871656,
      "loss": 2.1391,
      "step": 100240
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.9322319030761719,
      "learning_rate": 0.0003607733197651511,
      "loss": 2.0554,
      "step": 100241
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.8840616345405579,
      "learning_rate": 0.0003607693140102868,
      "loss": 1.8407,
      "step": 100242
    },
    {
      "epoch": 1.31,
      "grad_norm": 1.2324705123901367,
      "learning_rate": 0.0003607653082441243,
      "loss": 2.1354,
      "step": 100243
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.8046956658363342,
      "learning_rate": 0.00036076130246666437,
      "loss": 2.1201,
      "step": 100244
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.9631211757659912,
      "learning_rate": 0.00036075729667790773,
      "loss": 1.8992,
      "step": 100245
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.9378550052642822,
      "learning_rate": 0.0003607532908778553,
      "loss": 2.1354,
      "step": 100246
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.922856330871582,
      "learning_rate": 0.0003607492850665076,
      "loss": 2.1492,
      "step": 100247
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.8395770192146301,
      "learning_rate": 0.0003607452792438655,
      "loss": 2.0098,
      "step": 100248
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.8898767232894897,
      "learning_rate": 0.00036074127340992966,
      "loss": 2.2216,
      "step": 100249
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.9024558067321777,
      "learning_rate": 0.00036073726756470084,
      "loss": 2.0997,
      "step": 100250
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.8589144945144653,
      "learning_rate": 0.0003607332617081799,
      "loss": 2.1703,
      "step": 100251
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.9678365588188171,
      "learning_rate": 0.0003607292558403674,
      "loss": 1.9419,
      "step": 100252
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.8361292481422424,
      "learning_rate": 0.00036072524996126425,
      "loss": 2.0685,
      "step": 100253
    },
    {
      "epoch": 1.31,
      "grad_norm": 1.004953384399414,
      "learning_rate": 0.00036072124407087114,
      "loss": 2.1717,
      "step": 100254
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.9032599329948425,
      "learning_rate": 0.0003607172381691888,
      "loss": 2.0707,
      "step": 100255
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.9097862243652344,
      "learning_rate": 0.000360713232256218,
      "loss": 2.3029,
      "step": 100256
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.8461470603942871,
      "learning_rate": 0.0003607092263319594,
      "loss": 2.0756,
      "step": 100257
    },
    {
      "epoch": 1.31,
      "grad_norm": 1.0394498109817505,
      "learning_rate": 0.00036070522039641375,
      "loss": 2.194,
      "step": 100258
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.9207237362861633,
      "learning_rate": 0.00036070121444958206,
      "loss": 2.1773,
      "step": 100259
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.8426440358161926,
      "learning_rate": 0.0003606972084914647,
      "loss": 1.8331,
      "step": 100260
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.880085825920105,
      "learning_rate": 0.00036069320252206263,
      "loss": 2.1015,
      "step": 100261
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.8842537999153137,
      "learning_rate": 0.0003606891965413766,
      "loss": 2.0648,
      "step": 100262
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.8052964806556702,
      "learning_rate": 0.0003606851905494072,
      "loss": 2.107,
      "step": 100263
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.8828736543655396,
      "learning_rate": 0.00036068118454615533,
      "loss": 2.2668,
      "step": 100264
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.860774576663971,
      "learning_rate": 0.00036067717853162167,
      "loss": 2.2186,
      "step": 100265
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.9045006036758423,
      "learning_rate": 0.000360673172505807,
      "loss": 2.3007,
      "step": 100266
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.8374832272529602,
      "learning_rate": 0.000360669166468712,
      "loss": 2.1258,
      "step": 100267
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.8516067266464233,
      "learning_rate": 0.0003606651604203375,
      "loss": 2.0294,
      "step": 100268
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.8126991391181946,
      "learning_rate": 0.0003606611543606842,
      "loss": 2.0376,
      "step": 100269
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.785145103931427,
      "learning_rate": 0.00036065714828975283,
      "loss": 2.0748,
      "step": 100270
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.7623597383499146,
      "learning_rate": 0.00036065314220754425,
      "loss": 2.0361,
      "step": 100271
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.859211802482605,
      "learning_rate": 0.000360649136114059,
      "loss": 2.2387,
      "step": 100272
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.8157690167427063,
      "learning_rate": 0.0003606451300092979,
      "loss": 2.009,
      "step": 100273
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.8532027006149292,
      "learning_rate": 0.0003606411238932618,
      "loss": 2.0889,
      "step": 100274
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.8532536625862122,
      "learning_rate": 0.0003606371177659513,
      "loss": 2.3476,
      "step": 100275
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.8973679542541504,
      "learning_rate": 0.00036063311162736725,
      "loss": 2.2742,
      "step": 100276
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.9360791444778442,
      "learning_rate": 0.0003606291054775104,
      "loss": 2.0707,
      "step": 100277
    },
    {
      "epoch": 1.31,
      "grad_norm": 1.0006612539291382,
      "learning_rate": 0.00036062509931638145,
      "loss": 2.4029,
      "step": 100278
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.872090756893158,
      "learning_rate": 0.00036062109314398107,
      "loss": 2.0971,
      "step": 100279
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.8942567706108093,
      "learning_rate": 0.00036061708696031013,
      "loss": 2.3034,
      "step": 100280
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.8655203580856323,
      "learning_rate": 0.00036061308076536935,
      "loss": 2.2856,
      "step": 100281
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.8413435816764832,
      "learning_rate": 0.00036060907455915937,
      "loss": 2.1838,
      "step": 100282
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.8402669429779053,
      "learning_rate": 0.00036060506834168117,
      "loss": 2.2322,
      "step": 100283
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.999221920967102,
      "learning_rate": 0.0003606010621129353,
      "loss": 2.1861,
      "step": 100284
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.9286792278289795,
      "learning_rate": 0.00036059705587292244,
      "loss": 2.1438,
      "step": 100285
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.8757005333900452,
      "learning_rate": 0.0003605930496216436,
      "loss": 2.1359,
      "step": 100286
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.8843501210212708,
      "learning_rate": 0.00036058904335909927,
      "loss": 2.415,
      "step": 100287
    },
    {
      "epoch": 1.31,
      "grad_norm": 1.229718565940857,
      "learning_rate": 0.0003605850370852902,
      "loss": 2.2523,
      "step": 100288
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.8585480451583862,
      "learning_rate": 0.0003605810308002174,
      "loss": 1.9903,
      "step": 100289
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.836743175983429,
      "learning_rate": 0.0003605770245038814,
      "loss": 2.1782,
      "step": 100290
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.8198928236961365,
      "learning_rate": 0.0003605730181962829,
      "loss": 2.1256,
      "step": 100291
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.8537623882293701,
      "learning_rate": 0.00036056901187742286,
      "loss": 2.0017,
      "step": 100292
    },
    {
      "epoch": 1.31,
      "grad_norm": 1.2047909498214722,
      "learning_rate": 0.0003605650055473018,
      "loss": 2.1962,
      "step": 100293
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.8172696828842163,
      "learning_rate": 0.0003605609992059206,
      "loss": 2.1459,
      "step": 100294
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.8930568695068359,
      "learning_rate": 0.00036055699285328,
      "loss": 2.3266,
      "step": 100295
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.9009937644004822,
      "learning_rate": 0.0003605529864893806,
      "loss": 2.2324,
      "step": 100296
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.8737270832061768,
      "learning_rate": 0.0003605489801142233,
      "loss": 2.1895,
      "step": 100297
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.8635531663894653,
      "learning_rate": 0.0003605449737278089,
      "loss": 2.1187,
      "step": 100298
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.853980302810669,
      "learning_rate": 0.00036054096733013793,
      "loss": 2.1202,
      "step": 100299
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.772560715675354,
      "learning_rate": 0.0003605369609212113,
      "loss": 2.1045,
      "step": 100300
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.9043160676956177,
      "learning_rate": 0.0003605329545010298,
      "loss": 2.1722,
      "step": 100301
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.8185281753540039,
      "learning_rate": 0.00036052894806959396,
      "loss": 2.0754,
      "step": 100302
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.9136041402816772,
      "learning_rate": 0.00036052494162690463,
      "loss": 2.1091,
      "step": 100303
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.9429486393928528,
      "learning_rate": 0.0003605209351729627,
      "loss": 2.3771,
      "step": 100304
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.829031229019165,
      "learning_rate": 0.00036051692870776864,
      "loss": 2.3297,
      "step": 100305
    },
    {
      "epoch": 1.31,
      "grad_norm": 1.0332669019699097,
      "learning_rate": 0.00036051292223132343,
      "loss": 2.2398,
      "step": 100306
    },
    {
      "epoch": 1.31,
      "grad_norm": 1.2338372468948364,
      "learning_rate": 0.0003605089157436277,
      "loss": 2.2706,
      "step": 100307
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.9608136415481567,
      "learning_rate": 0.0003605049092446822,
      "loss": 2.318,
      "step": 100308
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.9373246431350708,
      "learning_rate": 0.0003605009027344877,
      "loss": 2.0802,
      "step": 100309
    },
    {
      "epoch": 1.31,
      "grad_norm": 1.001939058303833,
      "learning_rate": 0.000360496896213045,
      "loss": 2.3394,
      "step": 100310
    },
    {
      "epoch": 1.31,
      "grad_norm": 1.311840534210205,
      "learning_rate": 0.0003604928896803547,
      "loss": 2.289,
      "step": 100311
    },
    {
      "epoch": 1.31,
      "grad_norm": 1.0485944747924805,
      "learning_rate": 0.0003604888831364177,
      "loss": 2.2392,
      "step": 100312
    },
    {
      "epoch": 1.31,
      "grad_norm": 1.010048747062683,
      "learning_rate": 0.0003604848765812346,
      "loss": 2.0318,
      "step": 100313
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.9280490875244141,
      "learning_rate": 0.0003604808700148063,
      "loss": 2.4122,
      "step": 100314
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.9016765356063843,
      "learning_rate": 0.0003604768634371334,
      "loss": 1.8567,
      "step": 100315
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.9048379063606262,
      "learning_rate": 0.0003604728568482167,
      "loss": 2.0957,
      "step": 100316
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.9602977633476257,
      "learning_rate": 0.000360468850248057,
      "loss": 2.2803,
      "step": 100317
    },
    {
      "epoch": 1.31,
      "grad_norm": 1.144111156463623,
      "learning_rate": 0.000360464843636655,
      "loss": 2.0969,
      "step": 100318
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.8010972738265991,
      "learning_rate": 0.0003604608370140114,
      "loss": 2.0915,
      "step": 100319
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.9844216704368591,
      "learning_rate": 0.000360456830380127,
      "loss": 1.9933,
      "step": 100320
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.9454295039176941,
      "learning_rate": 0.00036045282373500266,
      "loss": 2.0468,
      "step": 100321
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.9840987324714661,
      "learning_rate": 0.00036044881707863883,
      "loss": 2.1242,
      "step": 100322
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.8307557702064514,
      "learning_rate": 0.0003604448104110365,
      "loss": 2.1926,
      "step": 100323
    },
    {
      "epoch": 1.31,
      "grad_norm": 1.0489184856414795,
      "learning_rate": 0.0003604408037321963,
      "loss": 2.1354,
      "step": 100324
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.8545236587524414,
      "learning_rate": 0.00036043679704211905,
      "loss": 1.9886,
      "step": 100325
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.9366869330406189,
      "learning_rate": 0.00036043279034080544,
      "loss": 2.0762,
      "step": 100326
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.9390407204627991,
      "learning_rate": 0.00036042878362825624,
      "loss": 2.131,
      "step": 100327
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.9357013702392578,
      "learning_rate": 0.0003604247769044722,
      "loss": 2.2949,
      "step": 100328
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.7580236792564392,
      "learning_rate": 0.00036042077016945406,
      "loss": 2.3039,
      "step": 100329
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.8455811142921448,
      "learning_rate": 0.00036041676342320254,
      "loss": 2.0748,
      "step": 100330
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.8123035430908203,
      "learning_rate": 0.0003604127566657184,
      "loss": 2.3322,
      "step": 100331
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.9840955138206482,
      "learning_rate": 0.0003604087498970024,
      "loss": 2.125,
      "step": 100332
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.8278086185455322,
      "learning_rate": 0.00036040474311705524,
      "loss": 2.2174,
      "step": 100333
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.9022943377494812,
      "learning_rate": 0.0003604007363258777,
      "loss": 1.9364,
      "step": 100334
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.858097493648529,
      "learning_rate": 0.00036039672952347063,
      "loss": 2.3771,
      "step": 100335
    },
    {
      "epoch": 1.31,
      "grad_norm": 1.0013000965118408,
      "learning_rate": 0.0003603927227098346,
      "loss": 2.0285,
      "step": 100336
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.8459968566894531,
      "learning_rate": 0.00036038871588497034,
      "loss": 2.1422,
      "step": 100337
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.8484508395195007,
      "learning_rate": 0.0003603847090488788,
      "loss": 2.1517,
      "step": 100338
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.9766495227813721,
      "learning_rate": 0.00036038070220156056,
      "loss": 2.2527,
      "step": 100339
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.9557934999465942,
      "learning_rate": 0.0003603766953430163,
      "loss": 2.1677,
      "step": 100340
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.8714126348495483,
      "learning_rate": 0.0003603726884732471,
      "loss": 1.9505,
      "step": 100341
    },
    {
      "epoch": 1.31,
      "grad_norm": 1.1799629926681519,
      "learning_rate": 0.00036036868159225325,
      "loss": 1.9793,
      "step": 100342
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.9826768040657043,
      "learning_rate": 0.00036036467470003587,
      "loss": 2.0728,
      "step": 100343
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.8681771755218506,
      "learning_rate": 0.0003603606677965955,
      "loss": 2.1981,
      "step": 100344
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.9405130743980408,
      "learning_rate": 0.00036035666088193294,
      "loss": 2.2195,
      "step": 100345
    },
    {
      "epoch": 1.31,
      "grad_norm": 1.0336742401123047,
      "learning_rate": 0.00036035265395604896,
      "loss": 2.2075,
      "step": 100346
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.9660769701004028,
      "learning_rate": 0.0003603486470189443,
      "loss": 1.922,
      "step": 100347
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.9070929288864136,
      "learning_rate": 0.0003603446400706196,
      "loss": 2.1298,
      "step": 100348
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.8113507032394409,
      "learning_rate": 0.00036034063311107573,
      "loss": 1.9442,
      "step": 100349
    },
    {
      "epoch": 1.31,
      "grad_norm": 1.0281031131744385,
      "learning_rate": 0.0003603366261403135,
      "loss": 1.8285,
      "step": 100350
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.8459390997886658,
      "learning_rate": 0.0003603326191583335,
      "loss": 2.0911,
      "step": 100351
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.9427932500839233,
      "learning_rate": 0.00036032861216513644,
      "loss": 2.1231,
      "step": 100352
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.8757040500640869,
      "learning_rate": 0.0003603246051607232,
      "loss": 2.2037,
      "step": 100353
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.9046732187271118,
      "learning_rate": 0.0003603205981450945,
      "loss": 2.3691,
      "step": 100354
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.9635646939277649,
      "learning_rate": 0.00036031659111825104,
      "loss": 1.9772,
      "step": 100355
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.885519802570343,
      "learning_rate": 0.0003603125840801936,
      "loss": 2.3571,
      "step": 100356
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.8942747712135315,
      "learning_rate": 0.00036030857703092293,
      "loss": 2.0543,
      "step": 100357
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.9226411581039429,
      "learning_rate": 0.00036030456997043975,
      "loss": 1.8617,
      "step": 100358
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.8338443636894226,
      "learning_rate": 0.0003603005628987448,
      "loss": 2.0622,
      "step": 100359
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.8433949947357178,
      "learning_rate": 0.0003602965558158388,
      "loss": 1.9712,
      "step": 100360
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.8570016622543335,
      "learning_rate": 0.00036029254872172253,
      "loss": 2.4352,
      "step": 100361
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.8183960914611816,
      "learning_rate": 0.00036028854161639686,
      "loss": 2.0354,
      "step": 100362
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.8248745203018188,
      "learning_rate": 0.00036028453449986233,
      "loss": 2.1967,
      "step": 100363
    },
    {
      "epoch": 1.31,
      "grad_norm": 1.11574387550354,
      "learning_rate": 0.0003602805273721197,
      "loss": 2.1094,
      "step": 100364
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.8695654273033142,
      "learning_rate": 0.0003602765202331699,
      "loss": 2.2173,
      "step": 100365
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.8476163148880005,
      "learning_rate": 0.0003602725130830135,
      "loss": 2.2209,
      "step": 100366
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.9296231865882874,
      "learning_rate": 0.00036026850592165127,
      "loss": 1.9874,
      "step": 100367
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.9397103190422058,
      "learning_rate": 0.00036026449874908403,
      "loss": 2.0708,
      "step": 100368
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.8071795105934143,
      "learning_rate": 0.00036026049156531243,
      "loss": 2.1112,
      "step": 100369
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.8713149428367615,
      "learning_rate": 0.00036025648437033734,
      "loss": 2.3293,
      "step": 100370
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.8377674221992493,
      "learning_rate": 0.00036025247716415945,
      "loss": 2.1078,
      "step": 100371
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.8184419870376587,
      "learning_rate": 0.00036024846994677943,
      "loss": 2.2695,
      "step": 100372
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.831847071647644,
      "learning_rate": 0.00036024446271819803,
      "loss": 2.1866,
      "step": 100373
    },
    {
      "epoch": 1.31,
      "grad_norm": 1.0202605724334717,
      "learning_rate": 0.00036024045547841617,
      "loss": 2.1644,
      "step": 100374
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.8686848878860474,
      "learning_rate": 0.0003602364482274344,
      "loss": 2.3161,
      "step": 100375
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.9008985161781311,
      "learning_rate": 0.0003602324409652535,
      "loss": 2.3259,
      "step": 100376
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.85044926404953,
      "learning_rate": 0.0003602284336918744,
      "loss": 2.1933,
      "step": 100377
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.8876394033432007,
      "learning_rate": 0.00036022442640729757,
      "loss": 2.3668,
      "step": 100378
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.8775465488433838,
      "learning_rate": 0.0003602204191115239,
      "loss": 2.1188,
      "step": 100379
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.9155294895172119,
      "learning_rate": 0.00036021641180455417,
      "loss": 2.2206,
      "step": 100380
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.8158979415893555,
      "learning_rate": 0.000360212404486389,
      "loss": 1.943,
      "step": 100381
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.9360153079032898,
      "learning_rate": 0.0003602083971570292,
      "loss": 2.3189,
      "step": 100382
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.8042770624160767,
      "learning_rate": 0.0003602043898164756,
      "loss": 2.0986,
      "step": 100383
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.8521285653114319,
      "learning_rate": 0.0003602003824647288,
      "loss": 1.9309,
      "step": 100384
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.9034854769706726,
      "learning_rate": 0.0003601963751017896,
      "loss": 2.1589,
      "step": 100385
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.8967206478118896,
      "learning_rate": 0.0003601923677276589,
      "loss": 2.0493,
      "step": 100386
    },
    {
      "epoch": 1.31,
      "grad_norm": 1.039819598197937,
      "learning_rate": 0.00036018836034233713,
      "loss": 2.1863,
      "step": 100387
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.9840915203094482,
      "learning_rate": 0.0003601843529458253,
      "loss": 2.3063,
      "step": 100388
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.8263115882873535,
      "learning_rate": 0.000360180345538124,
      "loss": 2.3077,
      "step": 100389
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.9103010296821594,
      "learning_rate": 0.00036017633811923414,
      "loss": 2.2624,
      "step": 100390
    },
    {
      "epoch": 1.31,
      "grad_norm": 1.150343894958496,
      "learning_rate": 0.00036017233068915625,
      "loss": 2.0012,
      "step": 100391
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.9295356869697571,
      "learning_rate": 0.00036016832324789125,
      "loss": 2.2813,
      "step": 100392
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.9098798632621765,
      "learning_rate": 0.0003601643157954398,
      "loss": 2.1555,
      "step": 100393
    },
    {
      "epoch": 1.31,
      "grad_norm": 1.1384978294372559,
      "learning_rate": 0.0003601603083318027,
      "loss": 2.0791,
      "step": 100394
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.8275436758995056,
      "learning_rate": 0.00036015630085698064,
      "loss": 2.0415,
      "step": 100395
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.883348286151886,
      "learning_rate": 0.0003601522933709744,
      "loss": 2.1397,
      "step": 100396
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.9030851721763611,
      "learning_rate": 0.0003601482858737847,
      "loss": 1.9571,
      "step": 100397
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.9506941437721252,
      "learning_rate": 0.0003601442783654123,
      "loss": 2.0779,
      "step": 100398
    },
    {
      "epoch": 1.31,
      "grad_norm": 1.006206750869751,
      "learning_rate": 0.000360140270845858,
      "loss": 2.2993,
      "step": 100399
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.9221525192260742,
      "learning_rate": 0.0003601362633151224,
      "loss": 2.2106,
      "step": 100400
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.934961199760437,
      "learning_rate": 0.00036013225577320637,
      "loss": 2.1859,
      "step": 100401
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.9531802535057068,
      "learning_rate": 0.0003601282482201107,
      "loss": 2.1251,
      "step": 100402
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.9127957224845886,
      "learning_rate": 0.0003601242406558359,
      "loss": 2.4741,
      "step": 100403
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.8942206501960754,
      "learning_rate": 0.00036012023308038286,
      "loss": 2.2067,
      "step": 100404
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.8551236391067505,
      "learning_rate": 0.0003601162254937525,
      "loss": 2.1073,
      "step": 100405
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.9089004397392273,
      "learning_rate": 0.0003601122178959453,
      "loss": 1.8672,
      "step": 100406
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.8170942664146423,
      "learning_rate": 0.0003601082102869621,
      "loss": 2.2528,
      "step": 100407
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.8304987549781799,
      "learning_rate": 0.0003601042026668036,
      "loss": 2.0614,
      "step": 100408
    },
    {
      "epoch": 1.31,
      "grad_norm": 1.0809943675994873,
      "learning_rate": 0.0003601001950354707,
      "loss": 1.9833,
      "step": 100409
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.9479173421859741,
      "learning_rate": 0.000360096187392964,
      "loss": 2.273,
      "step": 100410
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.8467960953712463,
      "learning_rate": 0.0003600921797392843,
      "loss": 2.3835,
      "step": 100411
    },
    {
      "epoch": 1.31,
      "grad_norm": 1.0713839530944824,
      "learning_rate": 0.00036008817207443223,
      "loss": 1.9484,
      "step": 100412
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.8238242268562317,
      "learning_rate": 0.0003600841643984088,
      "loss": 2.1798,
      "step": 100413
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.8484728932380676,
      "learning_rate": 0.00036008015671121446,
      "loss": 2.1241,
      "step": 100414
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.9837633967399597,
      "learning_rate": 0.0003600761490128501,
      "loss": 2.1941,
      "step": 100415
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.942333996295929,
      "learning_rate": 0.00036007214130331645,
      "loss": 2.2205,
      "step": 100416
    },
    {
      "epoch": 1.31,
      "grad_norm": 1.0962865352630615,
      "learning_rate": 0.0003600681335826144,
      "loss": 2.2588,
      "step": 100417
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.8525424599647522,
      "learning_rate": 0.0003600641258507444,
      "loss": 2.3929,
      "step": 100418
    },
    {
      "epoch": 1.31,
      "grad_norm": 1.154398798942566,
      "learning_rate": 0.00036006011810770736,
      "loss": 2.0685,
      "step": 100419
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.8475926518440247,
      "learning_rate": 0.00036005611035350404,
      "loss": 2.3577,
      "step": 100420
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.8785361647605896,
      "learning_rate": 0.00036005210258813517,
      "loss": 1.9667,
      "step": 100421
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.8223380446434021,
      "learning_rate": 0.00036004809481160146,
      "loss": 2.1197,
      "step": 100422
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.971272885799408,
      "learning_rate": 0.00036004408702390366,
      "loss": 2.0968,
      "step": 100423
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.9416089057922363,
      "learning_rate": 0.00036004007922504254,
      "loss": 2.3192,
      "step": 100424
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.8478834629058838,
      "learning_rate": 0.0003600360714150188,
      "loss": 2.1481,
      "step": 100425
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.885662317276001,
      "learning_rate": 0.0003600320635938333,
      "loss": 2.2887,
      "step": 100426
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.8171464204788208,
      "learning_rate": 0.0003600280557614867,
      "loss": 2.0811,
      "step": 100427
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.8783013820648193,
      "learning_rate": 0.0003600240479179796,
      "loss": 2.2125,
      "step": 100428
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.8583087921142578,
      "learning_rate": 0.00036002004006331314,
      "loss": 2.3842,
      "step": 100429
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.9035416841506958,
      "learning_rate": 0.0003600160321974877,
      "loss": 2.3823,
      "step": 100430
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.8433849215507507,
      "learning_rate": 0.0003600120243205041,
      "loss": 2.1766,
      "step": 100431
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.8241079449653625,
      "learning_rate": 0.0003600080164323632,
      "loss": 2.0121,
      "step": 100432
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.8287251591682434,
      "learning_rate": 0.0003600040085330656,
      "loss": 2.1075,
      "step": 100433
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.9745700359344482,
      "learning_rate": 0.0003600000006226122,
      "loss": 2.4478,
      "step": 100434
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.8102699518203735,
      "learning_rate": 0.00035999599270100365,
      "loss": 2.1145,
      "step": 100435
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.9569061994552612,
      "learning_rate": 0.00035999198476824074,
      "loss": 2.1256,
      "step": 100436
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.8721792101860046,
      "learning_rate": 0.0003599879768243241,
      "loss": 2.2494,
      "step": 100437
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.9238486289978027,
      "learning_rate": 0.00035998396886925465,
      "loss": 2.0661,
      "step": 100438
    },
    {
      "epoch": 1.31,
      "grad_norm": 1.0951248407363892,
      "learning_rate": 0.00035997996090303305,
      "loss": 2.0962,
      "step": 100439
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.9392113089561462,
      "learning_rate": 0.00035997595292566,
      "loss": 2.1054,
      "step": 100440
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.9771912097930908,
      "learning_rate": 0.00035997194493713624,
      "loss": 2.1006,
      "step": 100441
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.9187836050987244,
      "learning_rate": 0.0003599679369374626,
      "loss": 2.2815,
      "step": 100442
    },
    {
      "epoch": 1.31,
      "grad_norm": 1.0161097049713135,
      "learning_rate": 0.0003599639289266398,
      "loss": 2.1199,
      "step": 100443
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.8641250133514404,
      "learning_rate": 0.0003599599209046687,
      "loss": 2.1215,
      "step": 100444
    },
    {
      "epoch": 1.31,
      "grad_norm": 1.0337480306625366,
      "learning_rate": 0.0003599559128715497,
      "loss": 2.0473,
      "step": 100445
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.8067247271537781,
      "learning_rate": 0.00035995190482728386,
      "loss": 2.1165,
      "step": 100446
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.9066470265388489,
      "learning_rate": 0.0003599478967718719,
      "loss": 2.2063,
      "step": 100447
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.8629900217056274,
      "learning_rate": 0.0003599438887053144,
      "loss": 2.2663,
      "step": 100448
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.8183560371398926,
      "learning_rate": 0.0003599398806276122,
      "loss": 2.0487,
      "step": 100449
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.8332877159118652,
      "learning_rate": 0.0003599358725387661,
      "loss": 2.1122,
      "step": 100450
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.8366001844406128,
      "learning_rate": 0.0003599318644387767,
      "loss": 2.115,
      "step": 100451
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.8861768841743469,
      "learning_rate": 0.00035992785632764493,
      "loss": 2.2623,
      "step": 100452
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.849067211151123,
      "learning_rate": 0.00035992384820537145,
      "loss": 1.9,
      "step": 100453
    },
    {
      "epoch": 1.31,
      "grad_norm": 1.0346949100494385,
      "learning_rate": 0.00035991984007195687,
      "loss": 2.2489,
      "step": 100454
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.8319962620735168,
      "learning_rate": 0.00035991583192740217,
      "loss": 2.0313,
      "step": 100455
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.8534992337226868,
      "learning_rate": 0.00035991182377170794,
      "loss": 2.018,
      "step": 100456
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.845754861831665,
      "learning_rate": 0.00035990781560487505,
      "loss": 2.0987,
      "step": 100457
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.8430221080780029,
      "learning_rate": 0.00035990380742690405,
      "loss": 2.2064,
      "step": 100458
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.8080028891563416,
      "learning_rate": 0.00035989979923779585,
      "loss": 2.285,
      "step": 100459
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.8389022350311279,
      "learning_rate": 0.0003598957910375512,
      "loss": 2.1359,
      "step": 100460
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.9612796902656555,
      "learning_rate": 0.00035989178282617074,
      "loss": 2.2071,
      "step": 100461
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.8873617649078369,
      "learning_rate": 0.00035988777460365523,
      "loss": 2.1891,
      "step": 100462
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.8936271071434021,
      "learning_rate": 0.0003598837663700055,
      "loss": 2.0976,
      "step": 100463
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.8462548851966858,
      "learning_rate": 0.00035987975812522223,
      "loss": 2.1107,
      "step": 100464
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.8468086123466492,
      "learning_rate": 0.0003598757498693062,
      "loss": 2.1609,
      "step": 100465
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.8376944661140442,
      "learning_rate": 0.00035987174160225817,
      "loss": 2.243,
      "step": 100466
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.8281148076057434,
      "learning_rate": 0.0003598677333240788,
      "loss": 2.116,
      "step": 100467
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.8601563572883606,
      "learning_rate": 0.00035986372503476886,
      "loss": 2.1461,
      "step": 100468
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.8877249956130981,
      "learning_rate": 0.00035985971673432925,
      "loss": 1.9466,
      "step": 100469
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.8496847152709961,
      "learning_rate": 0.00035985570842276047,
      "loss": 2.0172,
      "step": 100470
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.8494630455970764,
      "learning_rate": 0.0003598517001000634,
      "loss": 2.1613,
      "step": 100471
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.8035219311714172,
      "learning_rate": 0.00035984769176623885,
      "loss": 2.0415,
      "step": 100472
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.9369176626205444,
      "learning_rate": 0.0003598436834212874,
      "loss": 2.2235,
      "step": 100473
    },
    {
      "epoch": 1.31,
      "grad_norm": 1.0233590602874756,
      "learning_rate": 0.0003598396750652098,
      "loss": 2.2051,
      "step": 100474
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.8130218386650085,
      "learning_rate": 0.00035983566669800705,
      "loss": 2.0828,
      "step": 100475
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.9205514788627625,
      "learning_rate": 0.00035983165831967966,
      "loss": 2.2509,
      "step": 100476
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.8950427770614624,
      "learning_rate": 0.0003598276499302284,
      "loss": 2.2175,
      "step": 100477
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.8801559209823608,
      "learning_rate": 0.0003598236415296541,
      "loss": 2.1476,
      "step": 100478
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.956578254699707,
      "learning_rate": 0.0003598196331179574,
      "loss": 2.0795,
      "step": 100479
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.9429908394813538,
      "learning_rate": 0.0003598156246951391,
      "loss": 2.2392,
      "step": 100480
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.950831413269043,
      "learning_rate": 0.00035981161626120003,
      "loss": 1.9228,
      "step": 100481
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.9384665489196777,
      "learning_rate": 0.00035980760781614075,
      "loss": 2.236,
      "step": 100482
    },
    {
      "epoch": 1.31,
      "grad_norm": 1.241620659828186,
      "learning_rate": 0.00035980359935996217,
      "loss": 2.1858,
      "step": 100483
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.9554986953735352,
      "learning_rate": 0.00035979959089266504,
      "loss": 2.0535,
      "step": 100484
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.7850503325462341,
      "learning_rate": 0.0003597955824142499,
      "loss": 2.1229,
      "step": 100485
    },
    {
      "epoch": 1.31,
      "grad_norm": 1.0946069955825806,
      "learning_rate": 0.0003597915739247177,
      "loss": 2.3987,
      "step": 100486
    },
    {
      "epoch": 1.31,
      "grad_norm": 1.3594118356704712,
      "learning_rate": 0.00035978756542406914,
      "loss": 2.1222,
      "step": 100487
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.9063031673431396,
      "learning_rate": 0.00035978355691230487,
      "loss": 2.0472,
      "step": 100488
    },
    {
      "epoch": 1.31,
      "grad_norm": 1.1177833080291748,
      "learning_rate": 0.00035977954838942573,
      "loss": 2.4668,
      "step": 100489
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.9883657097816467,
      "learning_rate": 0.0003597755398554325,
      "loss": 2.3282,
      "step": 100490
    },
    {
      "epoch": 1.31,
      "grad_norm": 1.1736876964569092,
      "learning_rate": 0.0003597715313103258,
      "loss": 2.0452,
      "step": 100491
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.9489961266517639,
      "learning_rate": 0.00035976752275410643,
      "loss": 2.0559,
      "step": 100492
    },
    {
      "epoch": 1.31,
      "grad_norm": 1.8361306190490723,
      "learning_rate": 0.0003597635141867752,
      "loss": 2.1167,
      "step": 100493
    },
    {
      "epoch": 1.31,
      "grad_norm": 1.0858458280563354,
      "learning_rate": 0.00035975950560833284,
      "loss": 2.1337,
      "step": 100494
    },
    {
      "epoch": 1.31,
      "grad_norm": 1.026637077331543,
      "learning_rate": 0.00035975549701878,
      "loss": 2.107,
      "step": 100495
    },
    {
      "epoch": 1.31,
      "grad_norm": 1.0524762868881226,
      "learning_rate": 0.0003597514884181175,
      "loss": 2.1791,
      "step": 100496
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.8911661505699158,
      "learning_rate": 0.000359747479806346,
      "loss": 2.3777,
      "step": 100497
    },
    {
      "epoch": 1.31,
      "grad_norm": 1.1827760934829712,
      "learning_rate": 0.00035974347118346647,
      "loss": 2.1936,
      "step": 100498
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.9817790985107422,
      "learning_rate": 0.0003597394625494794,
      "loss": 2.1103,
      "step": 100499
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.9151616096496582,
      "learning_rate": 0.0003597354539043856,
      "loss": 2.2154,
      "step": 100500
    },
    {
      "epoch": 1.31,
      "grad_norm": 1.0422123670578003,
      "learning_rate": 0.000359731445248186,
      "loss": 2.0911,
      "step": 100501
    },
    {
      "epoch": 1.31,
      "grad_norm": 1.0725661516189575,
      "learning_rate": 0.000359727436580881,
      "loss": 2.3664,
      "step": 100502
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.9932274222373962,
      "learning_rate": 0.00035972342790247165,
      "loss": 2.1426,
      "step": 100503
    },
    {
      "epoch": 1.31,
      "grad_norm": 1.0026724338531494,
      "learning_rate": 0.0003597194192129586,
      "loss": 1.9915,
      "step": 100504
    },
    {
      "epoch": 1.31,
      "grad_norm": 1.037556767463684,
      "learning_rate": 0.00035971541051234263,
      "loss": 2.201,
      "step": 100505
    },
    {
      "epoch": 1.31,
      "grad_norm": 1.0381035804748535,
      "learning_rate": 0.0003597114018006243,
      "loss": 2.0557,
      "step": 100506
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.9718140363693237,
      "learning_rate": 0.00035970739307780454,
      "loss": 2.1441,
      "step": 100507
    },
    {
      "epoch": 1.31,
      "grad_norm": 1.2271854877471924,
      "learning_rate": 0.00035970338434388415,
      "loss": 2.0356,
      "step": 100508
    },
    {
      "epoch": 1.31,
      "grad_norm": 1.1887836456298828,
      "learning_rate": 0.00035969937559886364,
      "loss": 2.0487,
      "step": 100509
    },
    {
      "epoch": 1.31,
      "grad_norm": 1.0017675161361694,
      "learning_rate": 0.00035969536684274394,
      "loss": 2.1779,
      "step": 100510
    },
    {
      "epoch": 1.31,
      "grad_norm": 1.0700287818908691,
      "learning_rate": 0.00035969135807552576,
      "loss": 2.1229,
      "step": 100511
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.8081332445144653,
      "learning_rate": 0.00035968734929720984,
      "loss": 1.9739,
      "step": 100512
    },
    {
      "epoch": 1.31,
      "grad_norm": 1.0509350299835205,
      "learning_rate": 0.00035968334050779684,
      "loss": 2.0646,
      "step": 100513
    },
    {
      "epoch": 1.31,
      "grad_norm": 1.4032838344573975,
      "learning_rate": 0.0003596793317072877,
      "loss": 2.1826,
      "step": 100514
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.8828133344650269,
      "learning_rate": 0.00035967532289568296,
      "loss": 1.99,
      "step": 100515
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.8387428522109985,
      "learning_rate": 0.0003596713140729835,
      "loss": 2.1292,
      "step": 100516
    },
    {
      "epoch": 1.31,
      "grad_norm": 1.25431227684021,
      "learning_rate": 0.00035966730523919003,
      "loss": 1.9843,
      "step": 100517
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.9895257949829102,
      "learning_rate": 0.0003596632963943032,
      "loss": 2.0673,
      "step": 100518
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.9782959818840027,
      "learning_rate": 0.00035965928753832385,
      "loss": 2.1675,
      "step": 100519
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.9461504817008972,
      "learning_rate": 0.00035965527867125275,
      "loss": 2.1697,
      "step": 100520
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.888627290725708,
      "learning_rate": 0.0003596512697930906,
      "loss": 2.034,
      "step": 100521
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.8986217379570007,
      "learning_rate": 0.00035964726090383813,
      "loss": 2.2856,
      "step": 100522
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.8036502599716187,
      "learning_rate": 0.0003596432520034961,
      "loss": 1.9134,
      "step": 100523
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.8557614684104919,
      "learning_rate": 0.00035963924309206533,
      "loss": 2.2547,
      "step": 100524
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.8342230916023254,
      "learning_rate": 0.00035963523416954646,
      "loss": 2.1212,
      "step": 100525
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.8369894623756409,
      "learning_rate": 0.0003596312252359403,
      "loss": 1.9614,
      "step": 100526
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.923762857913971,
      "learning_rate": 0.00035962721629124756,
      "loss": 2.2191,
      "step": 100527
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.8389261960983276,
      "learning_rate": 0.00035962320733546894,
      "loss": 2.104,
      "step": 100528
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.8298258781433105,
      "learning_rate": 0.0003596191983686053,
      "loss": 2.2642,
      "step": 100529
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.9008733034133911,
      "learning_rate": 0.0003596151893906573,
      "loss": 2.2208,
      "step": 100530
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.9464177489280701,
      "learning_rate": 0.00035961118040162574,
      "loss": 2.2137,
      "step": 100531
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.8237135410308838,
      "learning_rate": 0.00035960717140151127,
      "loss": 2.3687,
      "step": 100532
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.9084128737449646,
      "learning_rate": 0.0003596031623903148,
      "loss": 2.1878,
      "step": 100533
    },
    {
      "epoch": 1.31,
      "grad_norm": 1.0529597997665405,
      "learning_rate": 0.0003595991533680369,
      "loss": 2.2682,
      "step": 100534
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.9602306485176086,
      "learning_rate": 0.00035959514433467837,
      "loss": 2.3787,
      "step": 100535
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.8770765662193298,
      "learning_rate": 0.0003595911352902401,
      "loss": 2.2289,
      "step": 100536
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.909932553768158,
      "learning_rate": 0.00035958712623472257,
      "loss": 2.1521,
      "step": 100537
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.8748867511749268,
      "learning_rate": 0.0003595831171681268,
      "loss": 2.1282,
      "step": 100538
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.9344933032989502,
      "learning_rate": 0.0003595791080904533,
      "loss": 2.087,
      "step": 100539
    },
    {
      "epoch": 1.31,
      "grad_norm": 1.0208286046981812,
      "learning_rate": 0.000359575099001703,
      "loss": 2.2738,
      "step": 100540
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.9817028641700745,
      "learning_rate": 0.00035957108990187646,
      "loss": 2.146,
      "step": 100541
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.8317157030105591,
      "learning_rate": 0.00035956708079097464,
      "loss": 2.1297,
      "step": 100542
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.9763641357421875,
      "learning_rate": 0.0003595630716689981,
      "loss": 2.2907,
      "step": 100543
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.7974100708961487,
      "learning_rate": 0.0003595590625359477,
      "loss": 2.31,
      "step": 100544
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.9378670454025269,
      "learning_rate": 0.0003595550533918242,
      "loss": 1.9207,
      "step": 100545
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.8884111046791077,
      "learning_rate": 0.0003595510442366282,
      "loss": 2.2475,
      "step": 100546
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.8985435962677002,
      "learning_rate": 0.0003595470350703605,
      "loss": 2.4802,
      "step": 100547
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.8681119680404663,
      "learning_rate": 0.00035954302589302203,
      "loss": 2.1996,
      "step": 100548
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.8923453092575073,
      "learning_rate": 0.0003595390167046132,
      "loss": 1.9648,
      "step": 100549
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.854793906211853,
      "learning_rate": 0.0003595350075051351,
      "loss": 1.9044,
      "step": 100550
    },
    {
      "epoch": 1.31,
      "grad_norm": 1.0318195819854736,
      "learning_rate": 0.00035953099829458827,
      "loss": 2.0954,
      "step": 100551
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.8604530692100525,
      "learning_rate": 0.0003595269890729735,
      "loss": 2.1024,
      "step": 100552
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.912020742893219,
      "learning_rate": 0.0003595229798402915,
      "loss": 2.2309,
      "step": 100553
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.9201868772506714,
      "learning_rate": 0.00035951897059654316,
      "loss": 2.2378,
      "step": 100554
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.8854290246963501,
      "learning_rate": 0.000359514961341729,
      "loss": 2.2292,
      "step": 100555
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.8407402038574219,
      "learning_rate": 0.0003595109520758499,
      "loss": 1.9661,
      "step": 100556
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.8019218444824219,
      "learning_rate": 0.00035950694279890665,
      "loss": 1.9414,
      "step": 100557
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.897998034954071,
      "learning_rate": 0.0003595029335108999,
      "loss": 2.4276,
      "step": 100558
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.8582013249397278,
      "learning_rate": 0.00035949892421183047,
      "loss": 2.0959,
      "step": 100559
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.9800292253494263,
      "learning_rate": 0.000359494914901699,
      "loss": 2.1073,
      "step": 100560
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.8928595781326294,
      "learning_rate": 0.0003594909055805064,
      "loss": 2.0368,
      "step": 100561
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.9171552062034607,
      "learning_rate": 0.0003594868962482532,
      "loss": 2.2342,
      "step": 100562
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.9076564908027649,
      "learning_rate": 0.0003594828869049404,
      "loss": 2.1742,
      "step": 100563
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.8947778344154358,
      "learning_rate": 0.0003594788775505685,
      "loss": 2.2888,
      "step": 100564
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.8411091566085815,
      "learning_rate": 0.00035947486818513834,
      "loss": 2.1837,
      "step": 100565
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.9488379955291748,
      "learning_rate": 0.0003594708588086508,
      "loss": 2.2206,
      "step": 100566
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.9360446333885193,
      "learning_rate": 0.0003594668494211064,
      "loss": 2.3028,
      "step": 100567
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.9458901286125183,
      "learning_rate": 0.000359462840022506,
      "loss": 2.0339,
      "step": 100568
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.904941976070404,
      "learning_rate": 0.0003594588306128504,
      "loss": 2.0114,
      "step": 100569
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.9468773007392883,
      "learning_rate": 0.0003594548211921402,
      "loss": 2.2192,
      "step": 100570
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.8853285312652588,
      "learning_rate": 0.0003594508117603763,
      "loss": 2.1926,
      "step": 100571
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.8483681678771973,
      "learning_rate": 0.0003594468023175593,
      "loss": 2.1674,
      "step": 100572
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.8976724743843079,
      "learning_rate": 0.0003594427928636901,
      "loss": 2.1257,
      "step": 100573
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.9103361964225769,
      "learning_rate": 0.00035943878339876926,
      "loss": 2.2445,
      "step": 100574
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.9257677793502808,
      "learning_rate": 0.00035943477392279776,
      "loss": 2.4339,
      "step": 100575
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.9257340431213379,
      "learning_rate": 0.0003594307644357761,
      "loss": 2.1347,
      "step": 100576
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.8113570809364319,
      "learning_rate": 0.00035942675493770516,
      "loss": 2.1789,
      "step": 100577
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.7851617932319641,
      "learning_rate": 0.00035942274542858564,
      "loss": 2.2122,
      "step": 100578
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.9135043025016785,
      "learning_rate": 0.0003594187359084184,
      "loss": 2.0698,
      "step": 100579
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.9650675654411316,
      "learning_rate": 0.00035941472637720406,
      "loss": 2.0448,
      "step": 100580
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.8708619475364685,
      "learning_rate": 0.00035941071683494335,
      "loss": 2.0686,
      "step": 100581
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.8104445338249207,
      "learning_rate": 0.0003594067072816371,
      "loss": 2.2453,
      "step": 100582
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.7582848072052002,
      "learning_rate": 0.00035940269771728606,
      "loss": 2.099,
      "step": 100583
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.9003318548202515,
      "learning_rate": 0.0003593986881418909,
      "loss": 2.0276,
      "step": 100584
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.9243074059486389,
      "learning_rate": 0.00035939467855545235,
      "loss": 1.9848,
      "step": 100585
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.9385536909103394,
      "learning_rate": 0.00035939066895797133,
      "loss": 2.1153,
      "step": 100586
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.8751667141914368,
      "learning_rate": 0.00035938665934944836,
      "loss": 2.061,
      "step": 100587
    },
    {
      "epoch": 1.31,
      "grad_norm": 1.0512561798095703,
      "learning_rate": 0.00035938264972988426,
      "loss": 2.095,
      "step": 100588
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.8654295802116394,
      "learning_rate": 0.0003593786400992799,
      "loss": 2.2218,
      "step": 100589
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.9092991352081299,
      "learning_rate": 0.0003593746304576359,
      "loss": 2.3006,
      "step": 100590
    },
    {
      "epoch": 1.31,
      "grad_norm": 1.0742179155349731,
      "learning_rate": 0.000359370620804953,
      "loss": 2.3805,
      "step": 100591
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.9032977223396301,
      "learning_rate": 0.00035936661114123204,
      "loss": 2.1943,
      "step": 100592
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.8409359455108643,
      "learning_rate": 0.00035936260146647375,
      "loss": 2.0932,
      "step": 100593
    },
    {
      "epoch": 1.31,
      "grad_norm": 1.0009855031967163,
      "learning_rate": 0.00035935859178067866,
      "loss": 2.0209,
      "step": 100594
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.9074554443359375,
      "learning_rate": 0.00035935458208384785,
      "loss": 2.1412,
      "step": 100595
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.9123432636260986,
      "learning_rate": 0.00035935057237598177,
      "loss": 2.2623,
      "step": 100596
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.8883728384971619,
      "learning_rate": 0.00035934656265708143,
      "loss": 2.3115,
      "step": 100597
    },
    {
      "epoch": 1.31,
      "grad_norm": 1.1032944917678833,
      "learning_rate": 0.0003593425529271474,
      "loss": 2.2878,
      "step": 100598
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.8948070406913757,
      "learning_rate": 0.0003593385431861804,
      "loss": 2.2403,
      "step": 100599
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.8904642462730408,
      "learning_rate": 0.0003593345334341814,
      "loss": 2.1361,
      "step": 100600
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.8511441946029663,
      "learning_rate": 0.0003593305236711508,
      "loss": 2.1402,
      "step": 100601
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.8436427712440491,
      "learning_rate": 0.00035932651389708963,
      "loss": 1.97,
      "step": 100602
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.8444304466247559,
      "learning_rate": 0.0003593225041119986,
      "loss": 2.0338,
      "step": 100603
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.8640093207359314,
      "learning_rate": 0.0003593184943158783,
      "loss": 2.054,
      "step": 100604
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.8148703575134277,
      "learning_rate": 0.0003593144845087296,
      "loss": 1.9445,
      "step": 100605
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.8753774762153625,
      "learning_rate": 0.00035931047469055325,
      "loss": 2.0693,
      "step": 100606
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.7516974210739136,
      "learning_rate": 0.00035930646486134987,
      "loss": 2.0933,
      "step": 100607
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.8787501454353333,
      "learning_rate": 0.0003593024550211204,
      "loss": 2.1407,
      "step": 100608
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.8566400408744812,
      "learning_rate": 0.00035929844516986546,
      "loss": 2.1326,
      "step": 100609
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.8911845684051514,
      "learning_rate": 0.0003592944353075858,
      "loss": 2.0959,
      "step": 100610
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.8556499481201172,
      "learning_rate": 0.00035929042543428216,
      "loss": 1.9736,
      "step": 100611
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.9506291151046753,
      "learning_rate": 0.0003592864155499554,
      "loss": 2.1179,
      "step": 100612
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.8604143857955933,
      "learning_rate": 0.0003592824056546061,
      "loss": 1.9534,
      "step": 100613
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.7935823202133179,
      "learning_rate": 0.00035927839574823506,
      "loss": 2.2835,
      "step": 100614
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.9257225394248962,
      "learning_rate": 0.00035927438583084316,
      "loss": 2.2537,
      "step": 100615
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.8166871666908264,
      "learning_rate": 0.00035927037590243095,
      "loss": 2.1415,
      "step": 100616
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.7977597117424011,
      "learning_rate": 0.00035926636596299923,
      "loss": 2.1476,
      "step": 100617
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.8553510904312134,
      "learning_rate": 0.00035926235601254887,
      "loss": 2.2499,
      "step": 100618
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.8925040364265442,
      "learning_rate": 0.0003592583460510804,
      "loss": 2.152,
      "step": 100619
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.8590319156646729,
      "learning_rate": 0.00035925433607859474,
      "loss": 2.0971,
      "step": 100620
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.8639402389526367,
      "learning_rate": 0.0003592503260950927,
      "loss": 2.2275,
      "step": 100621
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.8220581412315369,
      "learning_rate": 0.0003592463161005747,
      "loss": 2.3645,
      "step": 100622
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.8897759914398193,
      "learning_rate": 0.00035924230609504185,
      "loss": 2.0331,
      "step": 100623
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.8825838565826416,
      "learning_rate": 0.0003592382960784947,
      "loss": 2.3186,
      "step": 100624
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.9187501668930054,
      "learning_rate": 0.0003592342860509339,
      "loss": 2.1411,
      "step": 100625
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.8739774227142334,
      "learning_rate": 0.00035923027601236047,
      "loss": 1.9447,
      "step": 100626
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.9005387425422668,
      "learning_rate": 0.000359226265962775,
      "loss": 2.3773,
      "step": 100627
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.8863621950149536,
      "learning_rate": 0.0003592222559021782,
      "loss": 2.3296,
      "step": 100628
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.84633868932724,
      "learning_rate": 0.0003592182458305709,
      "loss": 2.177,
      "step": 100629
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.9609747529029846,
      "learning_rate": 0.0003592142357479538,
      "loss": 2.1455,
      "step": 100630
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.8262999057769775,
      "learning_rate": 0.0003592102256543277,
      "loss": 2.3661,
      "step": 100631
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.8734146356582642,
      "learning_rate": 0.00035920621554969326,
      "loss": 2.1308,
      "step": 100632
    },
    {
      "epoch": 1.31,
      "grad_norm": 1.0345168113708496,
      "learning_rate": 0.0003592022054340513,
      "loss": 2.0334,
      "step": 100633
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.8804629445075989,
      "learning_rate": 0.0003591981953074025,
      "loss": 2.1511,
      "step": 100634
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.9366840720176697,
      "learning_rate": 0.0003591941851697476,
      "loss": 2.2321,
      "step": 100635
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.814234733581543,
      "learning_rate": 0.0003591901750210875,
      "loss": 2.4362,
      "step": 100636
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.872525691986084,
      "learning_rate": 0.0003591861648614227,
      "loss": 1.9697,
      "step": 100637
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.866786539554596,
      "learning_rate": 0.00035918215469075407,
      "loss": 2.1515,
      "step": 100638
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.8958797454833984,
      "learning_rate": 0.00035917814450908247,
      "loss": 2.0934,
      "step": 100639
    },
    {
      "epoch": 1.31,
      "grad_norm": 1.0352824926376343,
      "learning_rate": 0.00035917413431640845,
      "loss": 2.0562,
      "step": 100640
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.8752229809761047,
      "learning_rate": 0.00035917012411273286,
      "loss": 2.0359,
      "step": 100641
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.9122368693351746,
      "learning_rate": 0.00035916611389805647,
      "loss": 1.9432,
      "step": 100642
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.8815972208976746,
      "learning_rate": 0.0003591621036723799,
      "loss": 2.0411,
      "step": 100643
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.880274772644043,
      "learning_rate": 0.0003591580934357041,
      "loss": 2.2062,
      "step": 100644
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.8492830395698547,
      "learning_rate": 0.0003591540831880296,
      "loss": 2.2665,
      "step": 100645
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.9083777666091919,
      "learning_rate": 0.00035915007292935724,
      "loss": 2.2601,
      "step": 100646
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.8629615306854248,
      "learning_rate": 0.0003591460626596878,
      "loss": 2.1694,
      "step": 100647
    },
    {
      "epoch": 1.31,
      "grad_norm": 1.4346369504928589,
      "learning_rate": 0.00035914205237902203,
      "loss": 2.1848,
      "step": 100648
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.8624560236930847,
      "learning_rate": 0.00035913804208736056,
      "loss": 2.127,
      "step": 100649
    },
    {
      "epoch": 1.31,
      "grad_norm": 1.1451740264892578,
      "learning_rate": 0.0003591340317847042,
      "loss": 2.0483,
      "step": 100650
    },
    {
      "epoch": 1.31,
      "grad_norm": 1.1524224281311035,
      "learning_rate": 0.00035913002147105376,
      "loss": 2.2736,
      "step": 100651
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.9827368259429932,
      "learning_rate": 0.00035912601114640994,
      "loss": 2.2297,
      "step": 100652
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.7705265283584595,
      "learning_rate": 0.00035912200081077345,
      "loss": 2.2223,
      "step": 100653
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.96208256483078,
      "learning_rate": 0.00035911799046414505,
      "loss": 1.9038,
      "step": 100654
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.9284048676490784,
      "learning_rate": 0.00035911398010652555,
      "loss": 2.2577,
      "step": 100655
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.857314944267273,
      "learning_rate": 0.00035910996973791554,
      "loss": 2.0736,
      "step": 100656
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.7776820063591003,
      "learning_rate": 0.00035910595935831596,
      "loss": 2.1748,
      "step": 100657
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.8184090852737427,
      "learning_rate": 0.00035910194896772744,
      "loss": 2.2883,
      "step": 100658
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.8582740426063538,
      "learning_rate": 0.00035909793856615075,
      "loss": 2.2721,
      "step": 100659
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.9065975546836853,
      "learning_rate": 0.00035909392815358665,
      "loss": 1.9669,
      "step": 100660
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.8894851803779602,
      "learning_rate": 0.00035908991773003594,
      "loss": 2.3907,
      "step": 100661
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.9100746512413025,
      "learning_rate": 0.00035908590729549924,
      "loss": 2.3669,
      "step": 100662
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.9308614730834961,
      "learning_rate": 0.0003590818968499773,
      "loss": 2.0966,
      "step": 100663
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.870431661605835,
      "learning_rate": 0.000359077886393471,
      "loss": 2.0592,
      "step": 100664
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.8510481119155884,
      "learning_rate": 0.00035907387592598095,
      "loss": 2.1182,
      "step": 100665
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.9248523712158203,
      "learning_rate": 0.000359069865447508,
      "loss": 2.1758,
      "step": 100666
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.9052905440330505,
      "learning_rate": 0.00035906585495805286,
      "loss": 1.9634,
      "step": 100667
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.890014111995697,
      "learning_rate": 0.0003590618444576162,
      "loss": 2.353,
      "step": 100668
    },
    {
      "epoch": 1.31,
      "grad_norm": 1.0592598915100098,
      "learning_rate": 0.0003590578339461988,
      "loss": 2.26,
      "step": 100669
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.847642719745636,
      "learning_rate": 0.0003590538234238015,
      "loss": 2.0469,
      "step": 100670
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.8266707062721252,
      "learning_rate": 0.000359049812890425,
      "loss": 2.0117,
      "step": 100671
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.869073748588562,
      "learning_rate": 0.00035904580234606996,
      "loss": 2.4376,
      "step": 100672
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.8926184177398682,
      "learning_rate": 0.0003590417917907372,
      "loss": 2.2028,
      "step": 100673
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.8998291492462158,
      "learning_rate": 0.0003590377812244275,
      "loss": 2.2298,
      "step": 100674
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.8157260417938232,
      "learning_rate": 0.0003590337706471415,
      "loss": 2.2384,
      "step": 100675
    },
    {
      "epoch": 1.31,
      "grad_norm": 1.0799213647842407,
      "learning_rate": 0.0003590297600588801,
      "loss": 2.1351,
      "step": 100676
    },
    {
      "epoch": 1.31,
      "grad_norm": 1.1025285720825195,
      "learning_rate": 0.0003590257494596438,
      "loss": 2.2104,
      "step": 100677
    },
    {
      "epoch": 1.31,
      "grad_norm": 1.0593165159225464,
      "learning_rate": 0.0003590217388494337,
      "loss": 2.1425,
      "step": 100678
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.8625229001045227,
      "learning_rate": 0.0003590177282282502,
      "loss": 2.1449,
      "step": 100679
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.875449001789093,
      "learning_rate": 0.00035901371759609424,
      "loss": 2.1771,
      "step": 100680
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.8495705127716064,
      "learning_rate": 0.0003590097069529666,
      "loss": 2.3488,
      "step": 100681
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.9328632354736328,
      "learning_rate": 0.0003590056962988678,
      "loss": 2.0493,
      "step": 100682
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.8274416327476501,
      "learning_rate": 0.0003590016856337987,
      "loss": 1.9161,
      "step": 100683
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.8900114297866821,
      "learning_rate": 0.0003589976749577602,
      "loss": 2.0771,
      "step": 100684
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.943271815776825,
      "learning_rate": 0.0003589936642707529,
      "loss": 2.176,
      "step": 100685
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.8885066509246826,
      "learning_rate": 0.0003589896535727775,
      "loss": 2.1981,
      "step": 100686
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.8500044941902161,
      "learning_rate": 0.00035898564286383484,
      "loss": 2.0361,
      "step": 100687
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.9685720205307007,
      "learning_rate": 0.00035898163214392566,
      "loss": 2.0911,
      "step": 100688
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.8164588809013367,
      "learning_rate": 0.00035897762141305064,
      "loss": 2.0612,
      "step": 100689
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.9968740940093994,
      "learning_rate": 0.0003589736106712106,
      "loss": 2.1976,
      "step": 100690
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.9047020077705383,
      "learning_rate": 0.0003589695999184063,
      "loss": 2.1808,
      "step": 100691
    },
    {
      "epoch": 1.31,
      "grad_norm": 1.0015292167663574,
      "learning_rate": 0.0003589655891546383,
      "loss": 2.0291,
      "step": 100692
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.9306011199951172,
      "learning_rate": 0.0003589615783799076,
      "loss": 2.0311,
      "step": 100693
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.8104346394538879,
      "learning_rate": 0.0003589575675942148,
      "loss": 2.1052,
      "step": 100694
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.9392861127853394,
      "learning_rate": 0.00035895355679756066,
      "loss": 1.9868,
      "step": 100695
    },
    {
      "epoch": 1.31,
      "grad_norm": 1.274091362953186,
      "learning_rate": 0.0003589495459899459,
      "loss": 2.2781,
      "step": 100696
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.9711132049560547,
      "learning_rate": 0.0003589455351713714,
      "loss": 2.2661,
      "step": 100697
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.9024415612220764,
      "learning_rate": 0.00035894152434183773,
      "loss": 2.2947,
      "step": 100698
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.9275081157684326,
      "learning_rate": 0.00035893751350134576,
      "loss": 2.4213,
      "step": 100699
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.9377233982086182,
      "learning_rate": 0.0003589335026498963,
      "loss": 2.236,
      "step": 100700
    },
    {
      "epoch": 1.31,
      "grad_norm": 1.1037330627441406,
      "learning_rate": 0.0003589294917874898,
      "loss": 1.9467,
      "step": 100701
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.8907148241996765,
      "learning_rate": 0.00035892548091412727,
      "loss": 2.135,
      "step": 100702
    },
    {
      "epoch": 1.31,
      "grad_norm": 1.0062187910079956,
      "learning_rate": 0.0003589214700298094,
      "loss": 1.9924,
      "step": 100703
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.9822407960891724,
      "learning_rate": 0.0003589174591345369,
      "loss": 1.8475,
      "step": 100704
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.8919492959976196,
      "learning_rate": 0.0003589134482283105,
      "loss": 2.2174,
      "step": 100705
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.9233836531639099,
      "learning_rate": 0.00035890943731113113,
      "loss": 2.1049,
      "step": 100706
    },
    {
      "epoch": 1.31,
      "grad_norm": 1.010138750076294,
      "learning_rate": 0.00035890542638299924,
      "loss": 1.966,
      "step": 100707
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.9503770470619202,
      "learning_rate": 0.00035890141544391566,
      "loss": 2.4975,
      "step": 100708
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.8837775588035583,
      "learning_rate": 0.0003588974044938813,
      "loss": 2.2697,
      "step": 100709
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.902746856212616,
      "learning_rate": 0.00035889339353289675,
      "loss": 2.0415,
      "step": 100710
    },
    {
      "epoch": 1.31,
      "grad_norm": 1.0494155883789062,
      "learning_rate": 0.0003588893825609628,
      "loss": 2.0261,
      "step": 100711
    },
    {
      "epoch": 1.31,
      "grad_norm": 1.0557228326797485,
      "learning_rate": 0.00035888537157808035,
      "loss": 2.1141,
      "step": 100712
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.878908634185791,
      "learning_rate": 0.00035888136058424987,
      "loss": 2.0782,
      "step": 100713
    },
    {
      "epoch": 1.31,
      "grad_norm": 1.0256707668304443,
      "learning_rate": 0.0003588773495794722,
      "loss": 2.0903,
      "step": 100714
    },
    {
      "epoch": 1.31,
      "grad_norm": 1.208873987197876,
      "learning_rate": 0.0003588733385637483,
      "loss": 2.3351,
      "step": 100715
    },
    {
      "epoch": 1.31,
      "grad_norm": 1.0449955463409424,
      "learning_rate": 0.0003588693275370785,
      "loss": 2.1843,
      "step": 100716
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.9489710330963135,
      "learning_rate": 0.0003588653164994639,
      "loss": 2.2793,
      "step": 100717
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.8762987852096558,
      "learning_rate": 0.0003588613054509051,
      "loss": 2.4004,
      "step": 100718
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.9165139198303223,
      "learning_rate": 0.00035885729439140293,
      "loss": 2.3762,
      "step": 100719
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.8286368250846863,
      "learning_rate": 0.000358853283320958,
      "loss": 2.0034,
      "step": 100720
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.8818387985229492,
      "learning_rate": 0.0003588492722395712,
      "loss": 2.1701,
      "step": 100721
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.9062589406967163,
      "learning_rate": 0.00035884526114724315,
      "loss": 2.2038,
      "step": 100722
    },
    {
      "epoch": 1.31,
      "grad_norm": 1.1898221969604492,
      "learning_rate": 0.00035884125004397466,
      "loss": 2.1289,
      "step": 100723
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.8158443570137024,
      "learning_rate": 0.0003588372389297665,
      "loss": 2.583,
      "step": 100724
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.7814287543296814,
      "learning_rate": 0.00035883322780461943,
      "loss": 1.9504,
      "step": 100725
    },
    {
      "epoch": 1.31,
      "grad_norm": 1.0596327781677246,
      "learning_rate": 0.00035882921666853404,
      "loss": 2.1914,
      "step": 100726
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.8801177144050598,
      "learning_rate": 0.0003588252055215113,
      "loss": 2.3144,
      "step": 100727
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.8708023428916931,
      "learning_rate": 0.00035882119436355184,
      "loss": 2.2388,
      "step": 100728
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.9530036449432373,
      "learning_rate": 0.0003588171831946563,
      "loss": 2.2596,
      "step": 100729
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.8959347009658813,
      "learning_rate": 0.00035881317201482554,
      "loss": 2.2214,
      "step": 100730
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.8388424515724182,
      "learning_rate": 0.0003588091608240604,
      "loss": 2.0825,
      "step": 100731
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.8527013063430786,
      "learning_rate": 0.0003588051496223615,
      "loss": 2.0876,
      "step": 100732
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.9495924711227417,
      "learning_rate": 0.0003588011384097296,
      "loss": 2.224,
      "step": 100733
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.8827431797981262,
      "learning_rate": 0.00035879712718616546,
      "loss": 2.1468,
      "step": 100734
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.919508159160614,
      "learning_rate": 0.00035879311595166986,
      "loss": 2.3064,
      "step": 100735
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.8546004891395569,
      "learning_rate": 0.00035878910470624344,
      "loss": 2.1611,
      "step": 100736
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.9299113750457764,
      "learning_rate": 0.00035878509344988707,
      "loss": 2.415,
      "step": 100737
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.9325588941574097,
      "learning_rate": 0.00035878108218260135,
      "loss": 2.1782,
      "step": 100738
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.8292627334594727,
      "learning_rate": 0.0003587770709043872,
      "loss": 2.1955,
      "step": 100739
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.8979249596595764,
      "learning_rate": 0.0003587730596152453,
      "loss": 2.1553,
      "step": 100740
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.9447973370552063,
      "learning_rate": 0.00035876904831517634,
      "loss": 2.0701,
      "step": 100741
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.8034889698028564,
      "learning_rate": 0.0003587650370041811,
      "loss": 2.1978,
      "step": 100742
    },
    {
      "epoch": 1.31,
      "grad_norm": 1.0214372873306274,
      "learning_rate": 0.00035876102568226033,
      "loss": 2.1329,
      "step": 100743
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.970911979675293,
      "learning_rate": 0.0003587570143494148,
      "loss": 2.3272,
      "step": 100744
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.8525816202163696,
      "learning_rate": 0.00035875300300564517,
      "loss": 2.1149,
      "step": 100745
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.9267726540565491,
      "learning_rate": 0.00035874899165095237,
      "loss": 2.2521,
      "step": 100746
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.8844862580299377,
      "learning_rate": 0.00035874498028533693,
      "loss": 2.0642,
      "step": 100747
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.9480177164077759,
      "learning_rate": 0.0003587409689087997,
      "loss": 1.9598,
      "step": 100748
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.9027274250984192,
      "learning_rate": 0.0003587369575213415,
      "loss": 2.1834,
      "step": 100749
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.8313961029052734,
      "learning_rate": 0.00035873294612296286,
      "loss": 2.0263,
      "step": 100750
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.8843539953231812,
      "learning_rate": 0.0003587289347136647,
      "loss": 2.1648,
      "step": 100751
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.7596728801727295,
      "learning_rate": 0.00035872492329344776,
      "loss": 2.1186,
      "step": 100752
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.9654299020767212,
      "learning_rate": 0.00035872091186231274,
      "loss": 2.2008,
      "step": 100753
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.8482765555381775,
      "learning_rate": 0.00035871690042026035,
      "loss": 2.1515,
      "step": 100754
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.7812345027923584,
      "learning_rate": 0.0003587128889672914,
      "loss": 1.807,
      "step": 100755
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.8957112431526184,
      "learning_rate": 0.00035870887750340664,
      "loss": 1.8625,
      "step": 100756
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.7612897157669067,
      "learning_rate": 0.00035870486602860674,
      "loss": 2.0342,
      "step": 100757
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.8487971425056458,
      "learning_rate": 0.00035870085454289255,
      "loss": 2.0487,
      "step": 100758
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.8452998399734497,
      "learning_rate": 0.0003586968430462648,
      "loss": 2.2622,
      "step": 100759
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.8844459652900696,
      "learning_rate": 0.00035869283153872414,
      "loss": 2.2901,
      "step": 100760
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.8241969347000122,
      "learning_rate": 0.0003586888200202713,
      "loss": 2.1492,
      "step": 100761
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.9157957434654236,
      "learning_rate": 0.00035868480849090715,
      "loss": 2.3154,
      "step": 100762
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.964650571346283,
      "learning_rate": 0.00035868079695063244,
      "loss": 2.171,
      "step": 100763
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.8043715953826904,
      "learning_rate": 0.0003586767853994479,
      "loss": 2.1714,
      "step": 100764
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.9436050653457642,
      "learning_rate": 0.0003586727738373541,
      "loss": 2.1536,
      "step": 100765
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.8303318619728088,
      "learning_rate": 0.00035866876226435204,
      "loss": 2.2265,
      "step": 100766
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.8721738457679749,
      "learning_rate": 0.00035866475068044236,
      "loss": 2.2217,
      "step": 100767
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.913532018661499,
      "learning_rate": 0.00035866073908562567,
      "loss": 1.9655,
      "step": 100768
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.8449124693870544,
      "learning_rate": 0.0003586567274799029,
      "loss": 2.0879,
      "step": 100769
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.8229154348373413,
      "learning_rate": 0.0003586527158632748,
      "loss": 2.138,
      "step": 100770
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.8340559005737305,
      "learning_rate": 0.00035864870423574197,
      "loss": 2.0964,
      "step": 100771
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.9196112155914307,
      "learning_rate": 0.0003586446925973052,
      "loss": 2.2067,
      "step": 100772
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.8963063955307007,
      "learning_rate": 0.00035864068094796546,
      "loss": 2.1377,
      "step": 100773
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.9141696095466614,
      "learning_rate": 0.0003586366692877232,
      "loss": 2.2463,
      "step": 100774
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.8792460560798645,
      "learning_rate": 0.00035863265761657925,
      "loss": 2.1414,
      "step": 100775
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.871823251247406,
      "learning_rate": 0.0003586286459345344,
      "loss": 2.1617,
      "step": 100776
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.886106550693512,
      "learning_rate": 0.00035862463424158936,
      "loss": 2.1257,
      "step": 100777
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.8534912467002869,
      "learning_rate": 0.0003586206225377449,
      "loss": 2.1211,
      "step": 100778
    },
    {
      "epoch": 1.31,
      "grad_norm": 1.3836464881896973,
      "learning_rate": 0.00035861661082300185,
      "loss": 2.0933,
      "step": 100779
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.8224385380744934,
      "learning_rate": 0.00035861259909736077,
      "loss": 2.226,
      "step": 100780
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.8579925298690796,
      "learning_rate": 0.00035860858736082253,
      "loss": 2.5475,
      "step": 100781
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.8038328289985657,
      "learning_rate": 0.0003586045756133879,
      "loss": 2.2911,
      "step": 100782
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.9073050022125244,
      "learning_rate": 0.00035860056385505747,
      "loss": 1.9289,
      "step": 100783
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.8611823916435242,
      "learning_rate": 0.0003585965520858321,
      "loss": 2.3638,
      "step": 100784
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.861682653427124,
      "learning_rate": 0.00035859254030571263,
      "loss": 1.9996,
      "step": 100785
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.8449245095252991,
      "learning_rate": 0.00035858852851469963,
      "loss": 2.0262,
      "step": 100786
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.8524137139320374,
      "learning_rate": 0.00035858451671279387,
      "loss": 2.2282,
      "step": 100787
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.8151864409446716,
      "learning_rate": 0.00035858050489999627,
      "loss": 2.1247,
      "step": 100788
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.844571053981781,
      "learning_rate": 0.0003585764930763073,
      "loss": 2.3178,
      "step": 100789
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.8344385623931885,
      "learning_rate": 0.0003585724812417279,
      "loss": 2.0997,
      "step": 100790
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.8687645792961121,
      "learning_rate": 0.0003585684693962588,
      "loss": 2.063,
      "step": 100791
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.8186935186386108,
      "learning_rate": 0.0003585644575399008,
      "loss": 2.121,
      "step": 100792
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.8429558873176575,
      "learning_rate": 0.00035856044567265444,
      "loss": 2.2165,
      "step": 100793
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.8505473732948303,
      "learning_rate": 0.0003585564337945206,
      "loss": 2.0224,
      "step": 100794
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.8610619306564331,
      "learning_rate": 0.00035855242190550006,
      "loss": 2.2563,
      "step": 100795
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.8431385159492493,
      "learning_rate": 0.0003585484100055934,
      "loss": 2.3164,
      "step": 100796
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.7760546803474426,
      "learning_rate": 0.0003585443980948016,
      "loss": 2.2252,
      "step": 100797
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.9023579955101013,
      "learning_rate": 0.00035854038617312534,
      "loss": 2.0134,
      "step": 100798
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.9180982708930969,
      "learning_rate": 0.00035853637424056525,
      "loss": 2.1758,
      "step": 100799
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.7685848474502563,
      "learning_rate": 0.0003585323622971221,
      "loss": 2.3905,
      "step": 100800
    },
    {
      "epoch": 1.31,
      "grad_norm": 1.030150055885315,
      "learning_rate": 0.00035852835034279676,
      "loss": 2.3338,
      "step": 100801
    },
    {
      "epoch": 1.31,
      "grad_norm": 1.0549436807632446,
      "learning_rate": 0.0003585243383775898,
      "loss": 2.342,
      "step": 100802
    },
    {
      "epoch": 1.31,
      "grad_norm": 1.0184952020645142,
      "learning_rate": 0.00035852032640150216,
      "loss": 2.1778,
      "step": 100803
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.9309825897216797,
      "learning_rate": 0.0003585163144145344,
      "loss": 2.1787,
      "step": 100804
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.90733402967453,
      "learning_rate": 0.0003585123024166874,
      "loss": 1.9033,
      "step": 100805
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.8983455300331116,
      "learning_rate": 0.00035850829040796184,
      "loss": 2.1511,
      "step": 100806
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.8369847536087036,
      "learning_rate": 0.0003585042783883586,
      "loss": 1.9962,
      "step": 100807
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.8976080417633057,
      "learning_rate": 0.0003585002663578781,
      "loss": 1.9307,
      "step": 100808
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.96236652135849,
      "learning_rate": 0.0003584962543165214,
      "loss": 2.0637,
      "step": 100809
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.8957398533821106,
      "learning_rate": 0.0003584922422642892,
      "loss": 1.9263,
      "step": 100810
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.7971165776252747,
      "learning_rate": 0.0003584882302011821,
      "loss": 2.0406,
      "step": 100811
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.8526230454444885,
      "learning_rate": 0.000358484218127201,
      "loss": 2.0816,
      "step": 100812
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.7930300831794739,
      "learning_rate": 0.0003584802060423466,
      "loss": 2.1761,
      "step": 100813
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.8864510655403137,
      "learning_rate": 0.00035847619394661955,
      "loss": 2.3251,
      "step": 100814
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.9393038153648376,
      "learning_rate": 0.00035847218184002067,
      "loss": 2.0379,
      "step": 100815
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.8575194478034973,
      "learning_rate": 0.0003584681697225507,
      "loss": 1.9475,
      "step": 100816
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.8146263360977173,
      "learning_rate": 0.00035846415759421043,
      "loss": 2.092,
      "step": 100817
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.8812976479530334,
      "learning_rate": 0.0003584601454550006,
      "loss": 2.3347,
      "step": 100818
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.8403597474098206,
      "learning_rate": 0.00035845613330492184,
      "loss": 2.0069,
      "step": 100819
    },
    {
      "epoch": 1.31,
      "grad_norm": 1.0536835193634033,
      "learning_rate": 0.00035845212114397505,
      "loss": 2.1436,
      "step": 100820
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.8505685329437256,
      "learning_rate": 0.00035844810897216087,
      "loss": 2.4016,
      "step": 100821
    },
    {
      "epoch": 1.31,
      "grad_norm": 1.0791057348251343,
      "learning_rate": 0.0003584440967894802,
      "loss": 2.0933,
      "step": 100822
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.9158863425254822,
      "learning_rate": 0.0003584400845959335,
      "loss": 2.144,
      "step": 100823
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.9798341393470764,
      "learning_rate": 0.00035843607239152177,
      "loss": 2.0274,
      "step": 100824
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.8128821849822998,
      "learning_rate": 0.00035843206017624566,
      "loss": 2.4327,
      "step": 100825
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.7960453033447266,
      "learning_rate": 0.00035842804795010595,
      "loss": 1.9842,
      "step": 100826
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.8811999559402466,
      "learning_rate": 0.0003584240357131033,
      "loss": 2.2417,
      "step": 100827
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.9617837071418762,
      "learning_rate": 0.0003584200234652386,
      "loss": 2.2411,
      "step": 100828
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.9063600301742554,
      "learning_rate": 0.0003584160112065124,
      "loss": 2.3408,
      "step": 100829
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.8877288103103638,
      "learning_rate": 0.00035841199893692563,
      "loss": 2.0813,
      "step": 100830
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.972821831703186,
      "learning_rate": 0.00035840798665647905,
      "loss": 2.24,
      "step": 100831
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.9689785838127136,
      "learning_rate": 0.0003584039743651732,
      "loss": 2.2433,
      "step": 100832
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.8970662355422974,
      "learning_rate": 0.000358399962063009,
      "loss": 2.4047,
      "step": 100833
    },
    {
      "epoch": 1.31,
      "grad_norm": 1.0328567028045654,
      "learning_rate": 0.00035839594974998717,
      "loss": 2.0752,
      "step": 100834
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.9318500757217407,
      "learning_rate": 0.00035839193742610837,
      "loss": 2.2236,
      "step": 100835
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.761349081993103,
      "learning_rate": 0.0003583879250913734,
      "loss": 2.3463,
      "step": 100836
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.8514224290847778,
      "learning_rate": 0.00035838391274578313,
      "loss": 2.1344,
      "step": 100837
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.932569682598114,
      "learning_rate": 0.0003583799003893381,
      "loss": 2.095,
      "step": 100838
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.8589640855789185,
      "learning_rate": 0.00035837588802203906,
      "loss": 1.9704,
      "step": 100839
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.8409230709075928,
      "learning_rate": 0.000358371875643887,
      "loss": 2.031,
      "step": 100840
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.9053312540054321,
      "learning_rate": 0.0003583678632548824,
      "loss": 2.2808,
      "step": 100841
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.8465416431427002,
      "learning_rate": 0.00035836385085502614,
      "loss": 2.0591,
      "step": 100842
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.8493056297302246,
      "learning_rate": 0.0003583598384443189,
      "loss": 2.0398,
      "step": 100843
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.8253966569900513,
      "learning_rate": 0.00035835582602276156,
      "loss": 2.1165,
      "step": 100844
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.919971764087677,
      "learning_rate": 0.00035835181359035467,
      "loss": 2.3033,
      "step": 100845
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.8693320155143738,
      "learning_rate": 0.0003583478011470992,
      "loss": 2.1467,
      "step": 100846
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.8646649122238159,
      "learning_rate": 0.0003583437886929957,
      "loss": 2.3564,
      "step": 100847
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.9890104532241821,
      "learning_rate": 0.00035833977622804495,
      "loss": 2.2511,
      "step": 100848
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.8150073289871216,
      "learning_rate": 0.00035833576375224774,
      "loss": 2.2282,
      "step": 100849
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.91684889793396,
      "learning_rate": 0.0003583317512656049,
      "loss": 2.1929,
      "step": 100850
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.8534292578697205,
      "learning_rate": 0.000358327738768117,
      "loss": 2.079,
      "step": 100851
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.943347156047821,
      "learning_rate": 0.00035832372625978486,
      "loss": 2.0481,
      "step": 100852
    },
    {
      "epoch": 1.31,
      "grad_norm": 1.0141950845718384,
      "learning_rate": 0.00035831971374060927,
      "loss": 2.0193,
      "step": 100853
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.8987794518470764,
      "learning_rate": 0.00035831570121059106,
      "loss": 2.1322,
      "step": 100854
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.8827064633369446,
      "learning_rate": 0.0003583116886697307,
      "loss": 2.2018,
      "step": 100855
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.9666050672531128,
      "learning_rate": 0.00035830767611802917,
      "loss": 2.0421,
      "step": 100856
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.9160124063491821,
      "learning_rate": 0.00035830366355548706,
      "loss": 1.9733,
      "step": 100857
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.8085713982582092,
      "learning_rate": 0.00035829965098210526,
      "loss": 2.2697,
      "step": 100858
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.9695277810096741,
      "learning_rate": 0.0003582956383978845,
      "loss": 2.3007,
      "step": 100859
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.8811779618263245,
      "learning_rate": 0.00035829162580282536,
      "loss": 2.09,
      "step": 100860
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.9127143025398254,
      "learning_rate": 0.00035828761319692883,
      "loss": 2.0457,
      "step": 100861
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.8743169903755188,
      "learning_rate": 0.0003582836005801955,
      "loss": 2.115,
      "step": 100862
    },
    {
      "epoch": 1.31,
      "grad_norm": 1.0378901958465576,
      "learning_rate": 0.0003582795879526261,
      "loss": 2.1128,
      "step": 100863
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.8933442234992981,
      "learning_rate": 0.0003582755753142214,
      "loss": 2.1247,
      "step": 100864
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.838050365447998,
      "learning_rate": 0.0003582715626649824,
      "loss": 2.1518,
      "step": 100865
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.8437451720237732,
      "learning_rate": 0.00035826755000490935,
      "loss": 2.0949,
      "step": 100866
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.815628170967102,
      "learning_rate": 0.00035826353733400335,
      "loss": 2.0426,
      "step": 100867
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.9019090533256531,
      "learning_rate": 0.0003582595246522651,
      "loss": 2.0228,
      "step": 100868
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.8604775071144104,
      "learning_rate": 0.0003582555119596952,
      "loss": 2.1919,
      "step": 100869
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.9815124869346619,
      "learning_rate": 0.0003582514992562946,
      "loss": 2.0404,
      "step": 100870
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.9105361104011536,
      "learning_rate": 0.000358247486542064,
      "loss": 2.078,
      "step": 100871
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.9020628333091736,
      "learning_rate": 0.0003582434738170039,
      "loss": 2.239,
      "step": 100872
    },
    {
      "epoch": 1.31,
      "grad_norm": 1.0022108554840088,
      "learning_rate": 0.00035823946108111534,
      "loss": 2.2329,
      "step": 100873
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.8800070285797119,
      "learning_rate": 0.00035823544833439905,
      "loss": 1.9012,
      "step": 100874
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.9446054100990295,
      "learning_rate": 0.0003582314355768556,
      "loss": 2.0107,
      "step": 100875
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.815139651298523,
      "learning_rate": 0.00035822742280848584,
      "loss": 2.0365,
      "step": 100876
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.9175779819488525,
      "learning_rate": 0.00035822341002929054,
      "loss": 2.1321,
      "step": 100877
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.9687721133232117,
      "learning_rate": 0.00035821939723927035,
      "loss": 2.125,
      "step": 100878
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.8698660731315613,
      "learning_rate": 0.0003582153844384261,
      "loss": 2.2032,
      "step": 100879
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.8662556409835815,
      "learning_rate": 0.00035821137162675854,
      "loss": 2.0722,
      "step": 100880
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.837125837802887,
      "learning_rate": 0.0003582073588042683,
      "loss": 2.0974,
      "step": 100881
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.8682249188423157,
      "learning_rate": 0.0003582033459709563,
      "loss": 2.1737,
      "step": 100882
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.8504132628440857,
      "learning_rate": 0.0003581993331268232,
      "loss": 2.0517,
      "step": 100883
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.9610158205032349,
      "learning_rate": 0.0003581953202718697,
      "loss": 2.1646,
      "step": 100884
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.871900737285614,
      "learning_rate": 0.0003581913074060966,
      "loss": 2.058,
      "step": 100885
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.9790490865707397,
      "learning_rate": 0.00035818729452950467,
      "loss": 2.2218,
      "step": 100886
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.8959269523620605,
      "learning_rate": 0.00035818328164209456,
      "loss": 2.2474,
      "step": 100887
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.7715156674385071,
      "learning_rate": 0.0003581792687438671,
      "loss": 1.9584,
      "step": 100888
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.889362633228302,
      "learning_rate": 0.0003581752558348231,
      "loss": 2.2639,
      "step": 100889
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.9342796206474304,
      "learning_rate": 0.0003581712429149631,
      "loss": 2.2056,
      "step": 100890
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.8137036561965942,
      "learning_rate": 0.000358167229984288,
      "loss": 2.1679,
      "step": 100891
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.9028295278549194,
      "learning_rate": 0.00035816321704279866,
      "loss": 2.3212,
      "step": 100892
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.8336091637611389,
      "learning_rate": 0.00035815920409049546,
      "loss": 2.0491,
      "step": 100893
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.9300352334976196,
      "learning_rate": 0.00035815519112737947,
      "loss": 1.9871,
      "step": 100894
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.8076826930046082,
      "learning_rate": 0.0003581511781534514,
      "loss": 2.046,
      "step": 100895
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.7958354353904724,
      "learning_rate": 0.0003581471651687118,
      "loss": 2.2085,
      "step": 100896
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.8769867420196533,
      "learning_rate": 0.0003581431521731616,
      "loss": 1.9914,
      "step": 100897
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.8505240082740784,
      "learning_rate": 0.0003581391391668015,
      "loss": 2.3948,
      "step": 100898
    },
    {
      "epoch": 1.31,
      "grad_norm": 1.0361493825912476,
      "learning_rate": 0.0003581351261496322,
      "loss": 2.2053,
      "step": 100899
    },
    {
      "epoch": 1.31,
      "grad_norm": 1.15885329246521,
      "learning_rate": 0.0003581311131216545,
      "loss": 1.8866,
      "step": 100900
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.8117268085479736,
      "learning_rate": 0.0003581271000828692,
      "loss": 2.0196,
      "step": 100901
    },
    {
      "epoch": 1.31,
      "grad_norm": 1.0858664512634277,
      "learning_rate": 0.0003581230870332769,
      "loss": 2.106,
      "step": 100902
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.9719740152359009,
      "learning_rate": 0.0003581190739728784,
      "loss": 2.316,
      "step": 100903
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.8300475478172302,
      "learning_rate": 0.00035811506090167454,
      "loss": 2.0056,
      "step": 100904
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.9781128168106079,
      "learning_rate": 0.0003581110478196659,
      "loss": 2.3326,
      "step": 100905
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.9243867993354797,
      "learning_rate": 0.0003581070347268534,
      "loss": 2.1877,
      "step": 100906
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.7963089942932129,
      "learning_rate": 0.00035810302162323773,
      "loss": 2.0974,
      "step": 100907
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.8348010778427124,
      "learning_rate": 0.0003580990085088195,
      "loss": 2.0685,
      "step": 100908
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.9754681587219238,
      "learning_rate": 0.0003580949953835996,
      "loss": 2.1802,
      "step": 100909
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.9629614353179932,
      "learning_rate": 0.00035809098224757883,
      "loss": 2.6368,
      "step": 100910
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.8516770005226135,
      "learning_rate": 0.0003580869691007578,
      "loss": 2.0377,
      "step": 100911
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.9874422550201416,
      "learning_rate": 0.0003580829559431372,
      "loss": 2.2143,
      "step": 100912
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.9719792008399963,
      "learning_rate": 0.00035807894277471806,
      "loss": 2.05,
      "step": 100913
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.9278317093849182,
      "learning_rate": 0.00035807492959550084,
      "loss": 2.4497,
      "step": 100914
    },
    {
      "epoch": 1.31,
      "grad_norm": 1.148329734802246,
      "learning_rate": 0.00035807091640548637,
      "loss": 2.1467,
      "step": 100915
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.9054102897644043,
      "learning_rate": 0.0003580669032046756,
      "loss": 2.3864,
      "step": 100916
    },
    {
      "epoch": 1.31,
      "grad_norm": 1.0638355016708374,
      "learning_rate": 0.0003580628899930689,
      "loss": 2.4153,
      "step": 100917
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.9543741345405579,
      "learning_rate": 0.00035805887677066726,
      "loss": 2.2436,
      "step": 100918
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.883722186088562,
      "learning_rate": 0.00035805486353747147,
      "loss": 2.1773,
      "step": 100919
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.7937111854553223,
      "learning_rate": 0.00035805085029348206,
      "loss": 2.2873,
      "step": 100920
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.9783906936645508,
      "learning_rate": 0.00035804683703869996,
      "loss": 2.2814,
      "step": 100921
    },
    {
      "epoch": 1.31,
      "grad_norm": 1.254150629043579,
      "learning_rate": 0.0003580428237731259,
      "loss": 2.2098,
      "step": 100922
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.9832748770713806,
      "learning_rate": 0.00035803881049676045,
      "loss": 2.3127,
      "step": 100923
    },
    {
      "epoch": 1.31,
      "grad_norm": 1.0349336862564087,
      "learning_rate": 0.0003580347972096046,
      "loss": 2.1547,
      "step": 100924
    },
    {
      "epoch": 1.31,
      "grad_norm": 1.0842466354370117,
      "learning_rate": 0.00035803078391165896,
      "loss": 2.0924,
      "step": 100925
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.9164372682571411,
      "learning_rate": 0.00035802677060292436,
      "loss": 2.2581,
      "step": 100926
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.8413845300674438,
      "learning_rate": 0.00035802275728340135,
      "loss": 2.0278,
      "step": 100927
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.9304758310317993,
      "learning_rate": 0.0003580187439530909,
      "loss": 2.0919,
      "step": 100928
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.9130685329437256,
      "learning_rate": 0.0003580147306119937,
      "loss": 1.8792,
      "step": 100929
    },
    {
      "epoch": 1.31,
      "grad_norm": 1.1050511598587036,
      "learning_rate": 0.00035801071726011043,
      "loss": 2.2313,
      "step": 100930
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.8430792093276978,
      "learning_rate": 0.00035800670389744187,
      "loss": 2.0458,
      "step": 100931
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.9528040885925293,
      "learning_rate": 0.0003580026905239888,
      "loss": 2.1915,
      "step": 100932
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.9036223292350769,
      "learning_rate": 0.0003579986771397519,
      "loss": 2.3577,
      "step": 100933
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.8294379711151123,
      "learning_rate": 0.0003579946637447319,
      "loss": 2.1348,
      "step": 100934
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.8762314915657043,
      "learning_rate": 0.00035799065033892977,
      "loss": 2.3327,
      "step": 100935
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.8525457978248596,
      "learning_rate": 0.00035798663692234594,
      "loss": 2.1497,
      "step": 100936
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.9866005182266235,
      "learning_rate": 0.0003579826234949813,
      "loss": 2.1293,
      "step": 100937
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.93732750415802,
      "learning_rate": 0.0003579786100568366,
      "loss": 2.2924,
      "step": 100938
    },
    {
      "epoch": 1.31,
      "grad_norm": 1.082929253578186,
      "learning_rate": 0.0003579745966079126,
      "loss": 2.1917,
      "step": 100939
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.842196524143219,
      "learning_rate": 0.00035797058314821005,
      "loss": 2.3178,
      "step": 100940
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.854173481464386,
      "learning_rate": 0.00035796656967772967,
      "loss": 2.0857,
      "step": 100941
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.8602780699729919,
      "learning_rate": 0.00035796255619647215,
      "loss": 2.3193,
      "step": 100942
    },
    {
      "epoch": 1.31,
      "grad_norm": 1.0781632661819458,
      "learning_rate": 0.0003579585427044384,
      "loss": 2.141,
      "step": 100943
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.9034246206283569,
      "learning_rate": 0.000357954529201629,
      "loss": 2.2037,
      "step": 100944
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.8926360607147217,
      "learning_rate": 0.00035795051568804473,
      "loss": 2.1416,
      "step": 100945
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.9159747958183289,
      "learning_rate": 0.00035794650216368644,
      "loss": 1.724,
      "step": 100946
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.8901574015617371,
      "learning_rate": 0.00035794248862855476,
      "loss": 2.4602,
      "step": 100947
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.9124808311462402,
      "learning_rate": 0.00035793847508265044,
      "loss": 2.2648,
      "step": 100948
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.8936755657196045,
      "learning_rate": 0.0003579344615259743,
      "loss": 2.1186,
      "step": 100949
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.9015068411827087,
      "learning_rate": 0.0003579304479585271,
      "loss": 2.2987,
      "step": 100950
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.9351286888122559,
      "learning_rate": 0.0003579264343803095,
      "loss": 2.1386,
      "step": 100951
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.8296180963516235,
      "learning_rate": 0.00035792242079132224,
      "loss": 2.3758,
      "step": 100952
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.8685789108276367,
      "learning_rate": 0.00035791840719156615,
      "loss": 2.1224,
      "step": 100953
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.8846048712730408,
      "learning_rate": 0.000357914393581042,
      "loss": 2.2339,
      "step": 100954
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.9887121915817261,
      "learning_rate": 0.0003579103799597503,
      "loss": 2.0082,
      "step": 100955
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.8326301574707031,
      "learning_rate": 0.00035790636632769215,
      "loss": 2.2297,
      "step": 100956
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.8521835803985596,
      "learning_rate": 0.00035790235268486797,
      "loss": 2.1966,
      "step": 100957
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.7856200933456421,
      "learning_rate": 0.0003578983390312787,
      "loss": 2.3938,
      "step": 100958
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.8576560020446777,
      "learning_rate": 0.00035789432536692507,
      "loss": 2.2264,
      "step": 100959
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.9630952477455139,
      "learning_rate": 0.0003578903116918078,
      "loss": 2.2656,
      "step": 100960
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.9221829175949097,
      "learning_rate": 0.0003578862980059276,
      "loss": 2.2416,
      "step": 100961
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.811657190322876,
      "learning_rate": 0.0003578822843092853,
      "loss": 2.3304,
      "step": 100962
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.9415715336799622,
      "learning_rate": 0.00035787827060188146,
      "loss": 2.2186,
      "step": 100963
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.848223090171814,
      "learning_rate": 0.00035787425688371704,
      "loss": 2.0874,
      "step": 100964
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.9163410663604736,
      "learning_rate": 0.0003578702431547927,
      "loss": 2.0948,
      "step": 100965
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.8914470076560974,
      "learning_rate": 0.00035786622941510923,
      "loss": 1.9755,
      "step": 100966
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.941281795501709,
      "learning_rate": 0.0003578622156646672,
      "loss": 2.3494,
      "step": 100967
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.9621013402938843,
      "learning_rate": 0.00035785820190346765,
      "loss": 2.1774,
      "step": 100968
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.9120457768440247,
      "learning_rate": 0.0003578541881315111,
      "loss": 2.0799,
      "step": 100969
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.8160279989242554,
      "learning_rate": 0.0003578501743487983,
      "loss": 2.1801,
      "step": 100970
    },
    {
      "epoch": 1.31,
      "grad_norm": 1.0898433923721313,
      "learning_rate": 0.00035784616055533014,
      "loss": 2.2775,
      "step": 100971
    },
    {
      "epoch": 1.31,
      "grad_norm": 1.0087374448776245,
      "learning_rate": 0.0003578421467511073,
      "loss": 2.1087,
      "step": 100972
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.8717654943466187,
      "learning_rate": 0.0003578381329361304,
      "loss": 2.1569,
      "step": 100973
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.8726628422737122,
      "learning_rate": 0.00035783411911040043,
      "loss": 2.2817,
      "step": 100974
    },
    {
      "epoch": 1.31,
      "grad_norm": 1.0148053169250488,
      "learning_rate": 0.00035783010527391795,
      "loss": 2.1048,
      "step": 100975
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.8589392900466919,
      "learning_rate": 0.00035782609142668377,
      "loss": 2.2269,
      "step": 100976
    },
    {
      "epoch": 1.31,
      "grad_norm": 1.0249159336090088,
      "learning_rate": 0.0003578220775686987,
      "loss": 2.2346,
      "step": 100977
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.9645391702651978,
      "learning_rate": 0.0003578180636999633,
      "loss": 2.1902,
      "step": 100978
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.8525276184082031,
      "learning_rate": 0.00035781404982047844,
      "loss": 1.9591,
      "step": 100979
    },
    {
      "epoch": 1.31,
      "grad_norm": 1.0466076135635376,
      "learning_rate": 0.0003578100359302449,
      "loss": 2.1255,
      "step": 100980
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.8586350679397583,
      "learning_rate": 0.0003578060220292634,
      "loss": 2.0431,
      "step": 100981
    },
    {
      "epoch": 1.31,
      "grad_norm": 1.0193523168563843,
      "learning_rate": 0.0003578020081175346,
      "loss": 2.0934,
      "step": 100982
    },
    {
      "epoch": 1.31,
      "grad_norm": 1.0251226425170898,
      "learning_rate": 0.00035779799419505937,
      "loss": 2.2519,
      "step": 100983
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.9613674879074097,
      "learning_rate": 0.0003577939802618384,
      "loss": 2.1558,
      "step": 100984
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.91938316822052,
      "learning_rate": 0.00035778996631787247,
      "loss": 2.1929,
      "step": 100985
    },
    {
      "epoch": 1.31,
      "grad_norm": 1.02407705783844,
      "learning_rate": 0.00035778595236316224,
      "loss": 2.0587,
      "step": 100986
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.8928531408309937,
      "learning_rate": 0.0003577819383977085,
      "loss": 1.8909,
      "step": 100987
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.9291930794715881,
      "learning_rate": 0.000357777924421512,
      "loss": 2.3571,
      "step": 100988
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.8946777582168579,
      "learning_rate": 0.00035777391043457347,
      "loss": 2.2541,
      "step": 100989
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.9476655125617981,
      "learning_rate": 0.00035776989643689387,
      "loss": 2.3407,
      "step": 100990
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.8728339076042175,
      "learning_rate": 0.0003577658824284735,
      "loss": 2.0386,
      "step": 100991
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.8497956991195679,
      "learning_rate": 0.00035776186840931354,
      "loss": 2.257,
      "step": 100992
    },
    {
      "epoch": 1.31,
      "grad_norm": 0.9251359105110168,
      "learning_rate": 0.00035775785437941454,
      "loss": 2.09,
      "step": 100993
    },
    {
      "epoch": 1.32,
      "grad_norm": 1.0385092496871948,
      "learning_rate": 0.0003577538403387772,
      "loss": 2.0692,
      "step": 100994
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.8720044493675232,
      "learning_rate": 0.0003577498262874023,
      "loss": 2.1651,
      "step": 100995
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.8324500918388367,
      "learning_rate": 0.00035774581222529074,
      "loss": 1.9964,
      "step": 100996
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.9330112934112549,
      "learning_rate": 0.000357741798152443,
      "loss": 2.0348,
      "step": 100997
    },
    {
      "epoch": 1.32,
      "grad_norm": 1.0259451866149902,
      "learning_rate": 0.00035773778406886003,
      "loss": 2.1641,
      "step": 100998
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.9178081750869751,
      "learning_rate": 0.0003577337699745426,
      "loss": 2.224,
      "step": 100999
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.841027557849884,
      "learning_rate": 0.0003577297558694913,
      "loss": 2.1476,
      "step": 101000
    },
    {
      "epoch": 1.32,
      "grad_norm": 1.23219895362854,
      "learning_rate": 0.0003577257417537069,
      "loss": 2.1311,
      "step": 101001
    },
    {
      "epoch": 1.32,
      "grad_norm": 1.0028644800186157,
      "learning_rate": 0.0003577217276271903,
      "loss": 2.1234,
      "step": 101002
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.967859148979187,
      "learning_rate": 0.00035771771348994206,
      "loss": 2.4051,
      "step": 101003
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.9065113067626953,
      "learning_rate": 0.00035771369934196304,
      "loss": 2.2363,
      "step": 101004
    },
    {
      "epoch": 1.32,
      "grad_norm": 1.0756882429122925,
      "learning_rate": 0.00035770968518325394,
      "loss": 2.16,
      "step": 101005
    },
    {
      "epoch": 1.32,
      "grad_norm": 1.1004996299743652,
      "learning_rate": 0.0003577056710138156,
      "loss": 2.2016,
      "step": 101006
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.83228600025177,
      "learning_rate": 0.00035770165683364854,
      "loss": 2.0858,
      "step": 101007
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.9612585306167603,
      "learning_rate": 0.0003576976426427537,
      "loss": 2.2134,
      "step": 101008
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.897402822971344,
      "learning_rate": 0.00035769362844113184,
      "loss": 1.9851,
      "step": 101009
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.9990962147712708,
      "learning_rate": 0.00035768961422878357,
      "loss": 2.4143,
      "step": 101010
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.9281439185142517,
      "learning_rate": 0.00035768560000570983,
      "loss": 2.2835,
      "step": 101011
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.9153311848640442,
      "learning_rate": 0.0003576815857719111,
      "loss": 2.2095,
      "step": 101012
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.9305990934371948,
      "learning_rate": 0.00035767757152738833,
      "loss": 2.1709,
      "step": 101013
    },
    {
      "epoch": 1.32,
      "grad_norm": 1.1515820026397705,
      "learning_rate": 0.0003576735572721423,
      "loss": 2.2866,
      "step": 101014
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.8001006245613098,
      "learning_rate": 0.0003576695430061736,
      "loss": 1.976,
      "step": 101015
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.9264094829559326,
      "learning_rate": 0.00035766552872948295,
      "loss": 2.434,
      "step": 101016
    },
    {
      "epoch": 1.32,
      "grad_norm": 1.0842231512069702,
      "learning_rate": 0.00035766151444207136,
      "loss": 2.2251,
      "step": 101017
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.8619613647460938,
      "learning_rate": 0.00035765750014393933,
      "loss": 2.3358,
      "step": 101018
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.9246510863304138,
      "learning_rate": 0.00035765348583508764,
      "loss": 2.322,
      "step": 101019
    },
    {
      "epoch": 1.32,
      "grad_norm": 1.0511093139648438,
      "learning_rate": 0.00035764947151551715,
      "loss": 2.1854,
      "step": 101020
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.8047193288803101,
      "learning_rate": 0.00035764545718522845,
      "loss": 2.2439,
      "step": 101021
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.9453924298286438,
      "learning_rate": 0.00035764144284422246,
      "loss": 2.2916,
      "step": 101022
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.8840305805206299,
      "learning_rate": 0.00035763742849249973,
      "loss": 2.0587,
      "step": 101023
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.8515818119049072,
      "learning_rate": 0.0003576334141300613,
      "loss": 2.1476,
      "step": 101024
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.9344007968902588,
      "learning_rate": 0.0003576293997569075,
      "loss": 2.0179,
      "step": 101025
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.8771353363990784,
      "learning_rate": 0.00035762538537303945,
      "loss": 2.2385,
      "step": 101026
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.8482117056846619,
      "learning_rate": 0.0003576213709784578,
      "loss": 2.2144,
      "step": 101027
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.8074012398719788,
      "learning_rate": 0.0003576173565731631,
      "loss": 2.464,
      "step": 101028
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.8687171339988708,
      "learning_rate": 0.0003576133421571563,
      "loss": 2.0751,
      "step": 101029
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.993786096572876,
      "learning_rate": 0.00035760932773043805,
      "loss": 2.3352,
      "step": 101030
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.9264588952064514,
      "learning_rate": 0.0003576053132930093,
      "loss": 1.987,
      "step": 101031
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.8619999289512634,
      "learning_rate": 0.0003576012988448705,
      "loss": 2.3903,
      "step": 101032
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.802172064781189,
      "learning_rate": 0.0003575972843860225,
      "loss": 2.0649,
      "step": 101033
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.810063898563385,
      "learning_rate": 0.00035759326991646615,
      "loss": 2.0579,
      "step": 101034
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.8074391484260559,
      "learning_rate": 0.00035758925543620215,
      "loss": 2.0453,
      "step": 101035
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.8914066553115845,
      "learning_rate": 0.00035758524094523116,
      "loss": 2.198,
      "step": 101036
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.8796019554138184,
      "learning_rate": 0.000357581226443554,
      "loss": 2.1805,
      "step": 101037
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.774832546710968,
      "learning_rate": 0.0003575772119311715,
      "loss": 2.0142,
      "step": 101038
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.9373033046722412,
      "learning_rate": 0.0003575731974080842,
      "loss": 2.2583,
      "step": 101039
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.8497488498687744,
      "learning_rate": 0.00035756918287429293,
      "loss": 2.0943,
      "step": 101040
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.8237014412879944,
      "learning_rate": 0.00035756516832979857,
      "loss": 2.1148,
      "step": 101041
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.768658459186554,
      "learning_rate": 0.00035756115377460167,
      "loss": 2.2917,
      "step": 101042
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.8614761829376221,
      "learning_rate": 0.0003575571392087031,
      "loss": 2.0088,
      "step": 101043
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.8027951717376709,
      "learning_rate": 0.0003575531246321036,
      "loss": 2.2402,
      "step": 101044
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.8306759595870972,
      "learning_rate": 0.0003575491100448038,
      "loss": 2.2674,
      "step": 101045
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.8365916013717651,
      "learning_rate": 0.00035754509544680456,
      "loss": 2.1005,
      "step": 101046
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.791868269443512,
      "learning_rate": 0.00035754108083810673,
      "loss": 2.151,
      "step": 101047
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.7642980217933655,
      "learning_rate": 0.0003575370662187108,
      "loss": 2.0839,
      "step": 101048
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.943583071231842,
      "learning_rate": 0.00035753305158861765,
      "loss": 2.2756,
      "step": 101049
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.9877232313156128,
      "learning_rate": 0.0003575290369478281,
      "loss": 2.0866,
      "step": 101050
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.8869127035140991,
      "learning_rate": 0.0003575250222963427,
      "loss": 2.2613,
      "step": 101051
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.9606234431266785,
      "learning_rate": 0.0003575210076341624,
      "loss": 2.1869,
      "step": 101052
    },
    {
      "epoch": 1.32,
      "grad_norm": 1.0553321838378906,
      "learning_rate": 0.00035751699296128784,
      "loss": 2.2214,
      "step": 101053
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.872230052947998,
      "learning_rate": 0.0003575129782777197,
      "loss": 1.8046,
      "step": 101054
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.8434809446334839,
      "learning_rate": 0.00035750896358345893,
      "loss": 2.0956,
      "step": 101055
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.8475940227508545,
      "learning_rate": 0.00035750494887850607,
      "loss": 2.096,
      "step": 101056
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.8221124410629272,
      "learning_rate": 0.00035750093416286204,
      "loss": 2.1003,
      "step": 101057
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.8485114574432373,
      "learning_rate": 0.00035749691943652746,
      "loss": 1.9809,
      "step": 101058
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.9152734279632568,
      "learning_rate": 0.0003574929046995031,
      "loss": 2.0253,
      "step": 101059
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.8859662413597107,
      "learning_rate": 0.0003574888899517898,
      "loss": 2.3483,
      "step": 101060
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.9368849396705627,
      "learning_rate": 0.0003574848751933882,
      "loss": 2.1684,
      "step": 101061
    },
    {
      "epoch": 1.32,
      "grad_norm": 1.0040817260742188,
      "learning_rate": 0.000357480860424299,
      "loss": 2.1314,
      "step": 101062
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.8958969116210938,
      "learning_rate": 0.00035747684564452307,
      "loss": 2.1573,
      "step": 101063
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.8464104533195496,
      "learning_rate": 0.00035747283085406107,
      "loss": 2.0908,
      "step": 101064
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.9036747813224792,
      "learning_rate": 0.0003574688160529138,
      "loss": 2.3963,
      "step": 101065
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.9525331258773804,
      "learning_rate": 0.0003574648012410821,
      "loss": 2.4216,
      "step": 101066
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.9216309785842896,
      "learning_rate": 0.0003574607864185665,
      "loss": 2.2486,
      "step": 101067
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.890731155872345,
      "learning_rate": 0.0003574567715853679,
      "loss": 2.1146,
      "step": 101068
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.8514645099639893,
      "learning_rate": 0.000357452756741487,
      "loss": 2.0762,
      "step": 101069
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.9740893244743347,
      "learning_rate": 0.0003574487418869245,
      "loss": 2.0975,
      "step": 101070
    },
    {
      "epoch": 1.32,
      "grad_norm": 1.05951988697052,
      "learning_rate": 0.0003574447270216812,
      "loss": 1.8869,
      "step": 101071
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.8781504034996033,
      "learning_rate": 0.0003574407121457579,
      "loss": 2.3826,
      "step": 101072
    },
    {
      "epoch": 1.32,
      "grad_norm": 1.025349497795105,
      "learning_rate": 0.00035743669725915524,
      "loss": 2.2668,
      "step": 101073
    },
    {
      "epoch": 1.32,
      "grad_norm": 1.0978338718414307,
      "learning_rate": 0.000357432682361874,
      "loss": 2.1597,
      "step": 101074
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.8420050740242004,
      "learning_rate": 0.000357428667453915,
      "loss": 1.9441,
      "step": 101075
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.9096100926399231,
      "learning_rate": 0.00035742465253527886,
      "loss": 2.1643,
      "step": 101076
    },
    {
      "epoch": 1.32,
      "grad_norm": 1.117990255355835,
      "learning_rate": 0.00035742063760596645,
      "loss": 2.1336,
      "step": 101077
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.8818736672401428,
      "learning_rate": 0.00035741662266597845,
      "loss": 2.0315,
      "step": 101078
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.9110532999038696,
      "learning_rate": 0.0003574126077153156,
      "loss": 2.0799,
      "step": 101079
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.7913700342178345,
      "learning_rate": 0.00035740859275397864,
      "loss": 1.9814,
      "step": 101080
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.9043756127357483,
      "learning_rate": 0.0003574045777819684,
      "loss": 2.1377,
      "step": 101081
    },
    {
      "epoch": 1.32,
      "grad_norm": 1.0748450756072998,
      "learning_rate": 0.0003574005627992855,
      "loss": 2.1431,
      "step": 101082
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.9220943450927734,
      "learning_rate": 0.0003573965478059308,
      "loss": 2.2098,
      "step": 101083
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.9388107061386108,
      "learning_rate": 0.00035739253280190497,
      "loss": 2.1229,
      "step": 101084
    },
    {
      "epoch": 1.32,
      "grad_norm": 1.082696557044983,
      "learning_rate": 0.0003573885177872088,
      "loss": 2.3497,
      "step": 101085
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.9204136729240417,
      "learning_rate": 0.00035738450276184304,
      "loss": 2.2821,
      "step": 101086
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.8442865014076233,
      "learning_rate": 0.0003573804877258084,
      "loss": 2.0661,
      "step": 101087
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.9893065690994263,
      "learning_rate": 0.0003573764726791056,
      "loss": 2.3662,
      "step": 101088
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.9024512767791748,
      "learning_rate": 0.00035737245762173543,
      "loss": 2.2934,
      "step": 101089
    },
    {
      "epoch": 1.32,
      "grad_norm": 1.029780626296997,
      "learning_rate": 0.0003573684425536987,
      "loss": 2.174,
      "step": 101090
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.8787658214569092,
      "learning_rate": 0.0003573644274749961,
      "loss": 1.9083,
      "step": 101091
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.9329316020011902,
      "learning_rate": 0.0003573604123856283,
      "loss": 2.2011,
      "step": 101092
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.9975858330726624,
      "learning_rate": 0.00035735639728559627,
      "loss": 2.1002,
      "step": 101093
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.9364154934883118,
      "learning_rate": 0.0003573523821749004,
      "loss": 2.2117,
      "step": 101094
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.9608187675476074,
      "learning_rate": 0.0003573483670535417,
      "loss": 2.1199,
      "step": 101095
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.9056286215782166,
      "learning_rate": 0.00035734435192152095,
      "loss": 2.1276,
      "step": 101096
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.8965222239494324,
      "learning_rate": 0.00035734033677883875,
      "loss": 2.3676,
      "step": 101097
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.8396919965744019,
      "learning_rate": 0.00035733632162549586,
      "loss": 2.1563,
      "step": 101098
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.9218782782554626,
      "learning_rate": 0.0003573323064614931,
      "loss": 2.3357,
      "step": 101099
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.9600847959518433,
      "learning_rate": 0.00035732829128683115,
      "loss": 2.0649,
      "step": 101100
    },
    {
      "epoch": 1.32,
      "grad_norm": 1.0293455123901367,
      "learning_rate": 0.0003573242761015108,
      "loss": 2.0483,
      "step": 101101
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.86672443151474,
      "learning_rate": 0.0003573202609055328,
      "loss": 2.0442,
      "step": 101102
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.9608147144317627,
      "learning_rate": 0.0003573162456988979,
      "loss": 2.3002,
      "step": 101103
    },
    {
      "epoch": 1.32,
      "grad_norm": 1.0393779277801514,
      "learning_rate": 0.0003573122304816068,
      "loss": 2.2134,
      "step": 101104
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.880110502243042,
      "learning_rate": 0.00035730821525366036,
      "loss": 1.9108,
      "step": 101105
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.8302690982818604,
      "learning_rate": 0.0003573042000150591,
      "loss": 2.0971,
      "step": 101106
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.8045526146888733,
      "learning_rate": 0.0003573001847658039,
      "loss": 2.1881,
      "step": 101107
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.9416801333427429,
      "learning_rate": 0.0003572961695058957,
      "loss": 2.1394,
      "step": 101108
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.8982117176055908,
      "learning_rate": 0.0003572921542353349,
      "loss": 2.1799,
      "step": 101109
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.8352992534637451,
      "learning_rate": 0.00035728813895412244,
      "loss": 2.0468,
      "step": 101110
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.8934805989265442,
      "learning_rate": 0.00035728412366225907,
      "loss": 1.9291,
      "step": 101111
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.9431558847427368,
      "learning_rate": 0.0003572801083597455,
      "loss": 2.1275,
      "step": 101112
    },
    {
      "epoch": 1.32,
      "grad_norm": 1.0661967992782593,
      "learning_rate": 0.0003572760930465824,
      "loss": 2.2636,
      "step": 101113
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.8099893927574158,
      "learning_rate": 0.0003572720777227706,
      "loss": 2.0826,
      "step": 101114
    },
    {
      "epoch": 1.32,
      "grad_norm": 1.099078893661499,
      "learning_rate": 0.0003572680623883109,
      "loss": 2.1537,
      "step": 101115
    },
    {
      "epoch": 1.32,
      "grad_norm": 1.3049427270889282,
      "learning_rate": 0.000357264047043204,
      "loss": 2.4419,
      "step": 101116
    },
    {
      "epoch": 1.32,
      "grad_norm": 1.437278389930725,
      "learning_rate": 0.0003572600316874506,
      "loss": 2.3828,
      "step": 101117
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.9588773250579834,
      "learning_rate": 0.00035725601632105147,
      "loss": 2.1197,
      "step": 101118
    },
    {
      "epoch": 1.32,
      "grad_norm": 1.195039987564087,
      "learning_rate": 0.0003572520009440074,
      "loss": 2.1857,
      "step": 101119
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.8803972601890564,
      "learning_rate": 0.00035724798555631905,
      "loss": 2.1728,
      "step": 101120
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.917711615562439,
      "learning_rate": 0.00035724397015798724,
      "loss": 2.4008,
      "step": 101121
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.9064181447029114,
      "learning_rate": 0.0003572399547490127,
      "loss": 2.141,
      "step": 101122
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.8451969027519226,
      "learning_rate": 0.0003572359393293962,
      "loss": 2.2096,
      "step": 101123
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.7657859921455383,
      "learning_rate": 0.0003572319238991384,
      "loss": 2.1075,
      "step": 101124
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.8159144520759583,
      "learning_rate": 0.00035722790845824017,
      "loss": 2.3669,
      "step": 101125
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.8090810179710388,
      "learning_rate": 0.00035722389300670216,
      "loss": 2.1993,
      "step": 101126
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.8585577607154846,
      "learning_rate": 0.0003572198775445251,
      "loss": 2.1847,
      "step": 101127
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.8745132088661194,
      "learning_rate": 0.00035721586207170984,
      "loss": 2.185,
      "step": 101128
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.8961560726165771,
      "learning_rate": 0.00035721184658825704,
      "loss": 2.0404,
      "step": 101129
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.855512261390686,
      "learning_rate": 0.0003572078310941676,
      "loss": 2.136,
      "step": 101130
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.8278306126594543,
      "learning_rate": 0.000357203815589442,
      "loss": 2.1509,
      "step": 101131
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.9155570864677429,
      "learning_rate": 0.0003571998000740812,
      "loss": 2.3727,
      "step": 101132
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.8281530141830444,
      "learning_rate": 0.0003571957845480859,
      "loss": 2.1923,
      "step": 101133
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.8742157220840454,
      "learning_rate": 0.0003571917690114567,
      "loss": 1.9288,
      "step": 101134
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.916092574596405,
      "learning_rate": 0.00035718775346419454,
      "loss": 1.7774,
      "step": 101135
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.8697824478149414,
      "learning_rate": 0.00035718373790630015,
      "loss": 2.1247,
      "step": 101136
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.8996335864067078,
      "learning_rate": 0.0003571797223377742,
      "loss": 2.2036,
      "step": 101137
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.928000807762146,
      "learning_rate": 0.0003571757067586174,
      "loss": 2.2484,
      "step": 101138
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.9213644862174988,
      "learning_rate": 0.00035717169116883065,
      "loss": 1.9519,
      "step": 101139
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.844012975692749,
      "learning_rate": 0.0003571676755684145,
      "loss": 2.0046,
      "step": 101140
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.8449925184249878,
      "learning_rate": 0.00035716365995736986,
      "loss": 1.9966,
      "step": 101141
    },
    {
      "epoch": 1.32,
      "grad_norm": 1.0942028760910034,
      "learning_rate": 0.0003571596443356974,
      "loss": 2.2499,
      "step": 101142
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.8090705275535583,
      "learning_rate": 0.0003571556287033979,
      "loss": 2.0635,
      "step": 101143
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.8896687626838684,
      "learning_rate": 0.00035715161306047207,
      "loss": 1.9859,
      "step": 101144
    },
    {
      "epoch": 1.32,
      "grad_norm": 1.017769455909729,
      "learning_rate": 0.00035714759740692074,
      "loss": 2.1093,
      "step": 101145
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.8943774104118347,
      "learning_rate": 0.00035714358174274445,
      "loss": 2.209,
      "step": 101146
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.8707471489906311,
      "learning_rate": 0.00035713956606794417,
      "loss": 2.2071,
      "step": 101147
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.8862414360046387,
      "learning_rate": 0.00035713555038252066,
      "loss": 2.3822,
      "step": 101148
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.969224750995636,
      "learning_rate": 0.00035713153468647446,
      "loss": 2.1511,
      "step": 101149
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.9127227067947388,
      "learning_rate": 0.00035712751897980645,
      "loss": 2.1198,
      "step": 101150
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.8952112197875977,
      "learning_rate": 0.0003571235032625174,
      "loss": 2.2294,
      "step": 101151
    },
    {
      "epoch": 1.32,
      "grad_norm": 1.0907379388809204,
      "learning_rate": 0.00035711948753460793,
      "loss": 2.4022,
      "step": 101152
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.9511131644248962,
      "learning_rate": 0.0003571154717960789,
      "loss": 2.1558,
      "step": 101153
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.8140376210212708,
      "learning_rate": 0.0003571114560469311,
      "loss": 2.1833,
      "step": 101154
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.8263858556747437,
      "learning_rate": 0.00035710744028716513,
      "loss": 2.2788,
      "step": 101155
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.9197941422462463,
      "learning_rate": 0.0003571034245167818,
      "loss": 2.3039,
      "step": 101156
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.920781672000885,
      "learning_rate": 0.000357099408735782,
      "loss": 2.119,
      "step": 101157
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.8329476714134216,
      "learning_rate": 0.0003570953929441661,
      "loss": 2.1296,
      "step": 101158
    },
    {
      "epoch": 1.32,
      "grad_norm": 1.0797721147537231,
      "learning_rate": 0.0003570913771419352,
      "loss": 2.0045,
      "step": 101159
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.8868549466133118,
      "learning_rate": 0.00035708736132909004,
      "loss": 2.2315,
      "step": 101160
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.9763880968093872,
      "learning_rate": 0.0003570833455056311,
      "loss": 2.3579,
      "step": 101161
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.8824886083602905,
      "learning_rate": 0.00035707932967155937,
      "loss": 2.0396,
      "step": 101162
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.8604017496109009,
      "learning_rate": 0.0003570753138268756,
      "loss": 2.0887,
      "step": 101163
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.8306867480278015,
      "learning_rate": 0.0003570712979715803,
      "loss": 2.0458,
      "step": 101164
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.9212838411331177,
      "learning_rate": 0.0003570672821056744,
      "loss": 2.2622,
      "step": 101165
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.8987596035003662,
      "learning_rate": 0.0003570632662291587,
      "loss": 2.2849,
      "step": 101166
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.9653782248497009,
      "learning_rate": 0.0003570592503420338,
      "loss": 2.272,
      "step": 101167
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.8556475043296814,
      "learning_rate": 0.0003570552344443005,
      "loss": 2.0024,
      "step": 101168
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.8688185811042786,
      "learning_rate": 0.0003570512185359596,
      "loss": 1.9777,
      "step": 101169
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.9607017636299133,
      "learning_rate": 0.00035704720261701175,
      "loss": 2.0797,
      "step": 101170
    },
    {
      "epoch": 1.32,
      "grad_norm": 1.0496193170547485,
      "learning_rate": 0.00035704318668745773,
      "loss": 2.1851,
      "step": 101171
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.8306235074996948,
      "learning_rate": 0.00035703917074729843,
      "loss": 2.1423,
      "step": 101172
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.9034716486930847,
      "learning_rate": 0.0003570351547965344,
      "loss": 2.0826,
      "step": 101173
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.8608361482620239,
      "learning_rate": 0.0003570311388351664,
      "loss": 2.169,
      "step": 101174
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.9065226912498474,
      "learning_rate": 0.0003570271228631953,
      "loss": 2.1837,
      "step": 101175
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.8802343010902405,
      "learning_rate": 0.00035702310688062174,
      "loss": 2.1268,
      "step": 101176
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.8813744783401489,
      "learning_rate": 0.0003570190908874465,
      "loss": 2.263,
      "step": 101177
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.8991307616233826,
      "learning_rate": 0.00035701507488367036,
      "loss": 2.1525,
      "step": 101178
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.856646716594696,
      "learning_rate": 0.00035701105886929406,
      "loss": 2.1755,
      "step": 101179
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.8662245869636536,
      "learning_rate": 0.00035700704284431827,
      "loss": 2.0239,
      "step": 101180
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.8302227258682251,
      "learning_rate": 0.0003570030268087439,
      "loss": 2.1485,
      "step": 101181
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.8745244741439819,
      "learning_rate": 0.00035699901076257154,
      "loss": 2.2465,
      "step": 101182
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.876677930355072,
      "learning_rate": 0.0003569949947058019,
      "loss": 2.0622,
      "step": 101183
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.8813450336456299,
      "learning_rate": 0.000356990978638436,
      "loss": 2.3401,
      "step": 101184
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.9049585461616516,
      "learning_rate": 0.00035698696256047425,
      "loss": 2.1252,
      "step": 101185
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.8306467533111572,
      "learning_rate": 0.00035698294647191754,
      "loss": 2.183,
      "step": 101186
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.8939933776855469,
      "learning_rate": 0.00035697893037276677,
      "loss": 2.1883,
      "step": 101187
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.8602795600891113,
      "learning_rate": 0.0003569749142630224,
      "loss": 1.9869,
      "step": 101188
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.8368529677391052,
      "learning_rate": 0.0003569708981426853,
      "loss": 2.434,
      "step": 101189
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.9161385893821716,
      "learning_rate": 0.0003569668820117563,
      "loss": 1.9289,
      "step": 101190
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.8674174547195435,
      "learning_rate": 0.00035696286587023614,
      "loss": 1.9047,
      "step": 101191
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.9103767275810242,
      "learning_rate": 0.0003569588497181254,
      "loss": 2.0512,
      "step": 101192
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.8203454613685608,
      "learning_rate": 0.000356954833555425,
      "loss": 2.065,
      "step": 101193
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.9244928359985352,
      "learning_rate": 0.0003569508173821356,
      "loss": 2.0359,
      "step": 101194
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.8556910753250122,
      "learning_rate": 0.000356946801198258,
      "loss": 2.0966,
      "step": 101195
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.8601921200752258,
      "learning_rate": 0.0003569427850037929,
      "loss": 2.2293,
      "step": 101196
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.9266791343688965,
      "learning_rate": 0.00035693876879874104,
      "loss": 2.1115,
      "step": 101197
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.9321150779724121,
      "learning_rate": 0.00035693475258310314,
      "loss": 2.2036,
      "step": 101198
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.8182681798934937,
      "learning_rate": 0.00035693073635688,
      "loss": 1.9347,
      "step": 101199
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.8888769149780273,
      "learning_rate": 0.0003569267201200725,
      "loss": 2.1639,
      "step": 101200
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.8687552213668823,
      "learning_rate": 0.0003569227038726812,
      "loss": 2.1233,
      "step": 101201
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.9416182637214661,
      "learning_rate": 0.0003569186876147068,
      "loss": 2.329,
      "step": 101202
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.9026204347610474,
      "learning_rate": 0.00035691467134615017,
      "loss": 2.2609,
      "step": 101203
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.8458927273750305,
      "learning_rate": 0.00035691065506701213,
      "loss": 2.2516,
      "step": 101204
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.8629118800163269,
      "learning_rate": 0.0003569066387772932,
      "loss": 2.1315,
      "step": 101205
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.8197803497314453,
      "learning_rate": 0.00035690262247699424,
      "loss": 1.9963,
      "step": 101206
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.9324390292167664,
      "learning_rate": 0.0003568986061661161,
      "loss": 2.1973,
      "step": 101207
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.8490986227989197,
      "learning_rate": 0.00035689458984465946,
      "loss": 2.2125,
      "step": 101208
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.9387019276618958,
      "learning_rate": 0.00035689057351262493,
      "loss": 2.1399,
      "step": 101209
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.9223814010620117,
      "learning_rate": 0.0003568865571700134,
      "loss": 2.0652,
      "step": 101210
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.855243444442749,
      "learning_rate": 0.0003568825408168257,
      "loss": 2.2334,
      "step": 101211
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.9724690318107605,
      "learning_rate": 0.00035687852445306234,
      "loss": 2.1119,
      "step": 101212
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.9647262096405029,
      "learning_rate": 0.00035687450807872414,
      "loss": 2.2881,
      "step": 101213
    },
    {
      "epoch": 1.32,
      "grad_norm": 1.0190339088439941,
      "learning_rate": 0.000356870491693812,
      "loss": 2.1539,
      "step": 101214
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.9147252440452576,
      "learning_rate": 0.0003568664752983266,
      "loss": 2.2342,
      "step": 101215
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.8800873756408691,
      "learning_rate": 0.0003568624588922686,
      "loss": 2.2923,
      "step": 101216
    },
    {
      "epoch": 1.32,
      "grad_norm": 1.0198063850402832,
      "learning_rate": 0.0003568584424756387,
      "loss": 2.1488,
      "step": 101217
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.9199331402778625,
      "learning_rate": 0.0003568544260484379,
      "loss": 2.2149,
      "step": 101218
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.9956762790679932,
      "learning_rate": 0.0003568504096106667,
      "loss": 2.3367,
      "step": 101219
    },
    {
      "epoch": 1.32,
      "grad_norm": 1.0895918607711792,
      "learning_rate": 0.0003568463931623259,
      "loss": 2.2867,
      "step": 101220
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.9246203303337097,
      "learning_rate": 0.00035684237670341635,
      "loss": 2.1465,
      "step": 101221
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.8436192274093628,
      "learning_rate": 0.0003568383602339387,
      "loss": 2.3439,
      "step": 101222
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.9328855276107788,
      "learning_rate": 0.0003568343437538937,
      "loss": 2.1509,
      "step": 101223
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.8526571393013,
      "learning_rate": 0.00035683032726328227,
      "loss": 2.21,
      "step": 101224
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.7859551310539246,
      "learning_rate": 0.0003568263107621049,
      "loss": 2.0391,
      "step": 101225
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.8578904867172241,
      "learning_rate": 0.0003568222942503623,
      "loss": 2.2123,
      "step": 101226
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.8414753675460815,
      "learning_rate": 0.00035681827772805557,
      "loss": 2.0745,
      "step": 101227
    },
    {
      "epoch": 1.32,
      "grad_norm": 1.0259979963302612,
      "learning_rate": 0.0003568142611951852,
      "loss": 2.2894,
      "step": 101228
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.9305891990661621,
      "learning_rate": 0.000356810244651752,
      "loss": 2.0459,
      "step": 101229
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.819107174873352,
      "learning_rate": 0.0003568062280977567,
      "loss": 2.0716,
      "step": 101230
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.9516634941101074,
      "learning_rate": 0.0003568022115332,
      "loss": 2.0584,
      "step": 101231
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.8928630948066711,
      "learning_rate": 0.0003567981949580827,
      "loss": 2.1951,
      "step": 101232
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.8478876948356628,
      "learning_rate": 0.00035679417837240556,
      "loss": 2.2563,
      "step": 101233
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.9665379524230957,
      "learning_rate": 0.0003567901617761694,
      "loss": 2.1026,
      "step": 101234
    },
    {
      "epoch": 1.32,
      "grad_norm": 1.046988844871521,
      "learning_rate": 0.0003567861451693747,
      "loss": 2.091,
      "step": 101235
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.925510585308075,
      "learning_rate": 0.00035678212855202253,
      "loss": 2.2008,
      "step": 101236
    },
    {
      "epoch": 1.32,
      "grad_norm": 1.0220396518707275,
      "learning_rate": 0.0003567781119241134,
      "loss": 2.3517,
      "step": 101237
    },
    {
      "epoch": 1.32,
      "grad_norm": 1.1437269449234009,
      "learning_rate": 0.00035677409528564816,
      "loss": 2.0686,
      "step": 101238
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.8862602114677429,
      "learning_rate": 0.00035677007863662766,
      "loss": 2.2534,
      "step": 101239
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.863477885723114,
      "learning_rate": 0.0003567660619770524,
      "loss": 2.2915,
      "step": 101240
    },
    {
      "epoch": 1.32,
      "grad_norm": 1.023437738418579,
      "learning_rate": 0.00035676204530692325,
      "loss": 1.8665,
      "step": 101241
    },
    {
      "epoch": 1.32,
      "grad_norm": 1.1775703430175781,
      "learning_rate": 0.00035675802862624107,
      "loss": 2.1809,
      "step": 101242
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.7757124900817871,
      "learning_rate": 0.00035675401193500646,
      "loss": 2.2433,
      "step": 101243
    },
    {
      "epoch": 1.32,
      "grad_norm": 1.1240636110305786,
      "learning_rate": 0.0003567499952332201,
      "loss": 2.2109,
      "step": 101244
    },
    {
      "epoch": 1.32,
      "grad_norm": 1.030890703201294,
      "learning_rate": 0.00035674597852088303,
      "loss": 2.2591,
      "step": 101245
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.7799170613288879,
      "learning_rate": 0.0003567419617979957,
      "loss": 2.0658,
      "step": 101246
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.8587276339530945,
      "learning_rate": 0.000356737945064559,
      "loss": 2.1996,
      "step": 101247
    },
    {
      "epoch": 1.32,
      "grad_norm": 1.0939407348632812,
      "learning_rate": 0.0003567339283205736,
      "loss": 2.2749,
      "step": 101248
    },
    {
      "epoch": 1.32,
      "grad_norm": 1.0371328592300415,
      "learning_rate": 0.00035672991156604034,
      "loss": 2.2936,
      "step": 101249
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.8332449197769165,
      "learning_rate": 0.0003567258948009599,
      "loss": 2.0539,
      "step": 101250
    },
    {
      "epoch": 1.32,
      "grad_norm": 1.0935593843460083,
      "learning_rate": 0.0003567218780253331,
      "loss": 2.4905,
      "step": 101251
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.9765326380729675,
      "learning_rate": 0.00035671786123916055,
      "loss": 2.2411,
      "step": 101252
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.8227740526199341,
      "learning_rate": 0.00035671384444244304,
      "loss": 2.2406,
      "step": 101253
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.8916548490524292,
      "learning_rate": 0.0003567098276351815,
      "loss": 1.8138,
      "step": 101254
    },
    {
      "epoch": 1.32,
      "grad_norm": 1.1553571224212646,
      "learning_rate": 0.0003567058108173765,
      "loss": 2.1335,
      "step": 101255
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.8689852952957153,
      "learning_rate": 0.00035670179398902867,
      "loss": 2.1002,
      "step": 101256
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.8713539838790894,
      "learning_rate": 0.000356697777150139,
      "loss": 2.1119,
      "step": 101257
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.882391631603241,
      "learning_rate": 0.0003566937603007082,
      "loss": 2.1698,
      "step": 101258
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.8892930150032043,
      "learning_rate": 0.0003566897434407369,
      "loss": 2.2727,
      "step": 101259
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.9109513163566589,
      "learning_rate": 0.0003566857265702259,
      "loss": 2.127,
      "step": 101260
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.8757396936416626,
      "learning_rate": 0.00035668170968917605,
      "loss": 2.137,
      "step": 101261
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.9377902150154114,
      "learning_rate": 0.00035667769279758787,
      "loss": 2.0355,
      "step": 101262
    },
    {
      "epoch": 1.32,
      "grad_norm": 1.027021884918213,
      "learning_rate": 0.00035667367589546227,
      "loss": 2.1513,
      "step": 101263
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.8796443343162537,
      "learning_rate": 0.0003566696589828,
      "loss": 2.3976,
      "step": 101264
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.8229315280914307,
      "learning_rate": 0.00035666564205960176,
      "loss": 2.1788,
      "step": 101265
    },
    {
      "epoch": 1.32,
      "grad_norm": 1.2323131561279297,
      "learning_rate": 0.00035666162512586826,
      "loss": 2.2299,
      "step": 101266
    },
    {
      "epoch": 1.32,
      "grad_norm": 1.0599286556243896,
      "learning_rate": 0.0003566576081816004,
      "loss": 2.4183,
      "step": 101267
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.8779906630516052,
      "learning_rate": 0.0003566535912267987,
      "loss": 2.2409,
      "step": 101268
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.8857791423797607,
      "learning_rate": 0.0003566495742614641,
      "loss": 2.3364,
      "step": 101269
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.8585559129714966,
      "learning_rate": 0.0003566455572855973,
      "loss": 2.3361,
      "step": 101270
    },
    {
      "epoch": 1.32,
      "grad_norm": 1.2808823585510254,
      "learning_rate": 0.000356641540299199,
      "loss": 2.2177,
      "step": 101271
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.8147010803222656,
      "learning_rate": 0.0003566375233022699,
      "loss": 2.2141,
      "step": 101272
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.8822939991950989,
      "learning_rate": 0.0003566335062948109,
      "loss": 1.9063,
      "step": 101273
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.9985077977180481,
      "learning_rate": 0.00035662948927682263,
      "loss": 1.8663,
      "step": 101274
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.7850659489631653,
      "learning_rate": 0.00035662547224830584,
      "loss": 2.0532,
      "step": 101275
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.8975929617881775,
      "learning_rate": 0.00035662145520926134,
      "loss": 2.2406,
      "step": 101276
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.8957112431526184,
      "learning_rate": 0.00035661743815968984,
      "loss": 2.173,
      "step": 101277
    },
    {
      "epoch": 1.32,
      "grad_norm": 1.0962969064712524,
      "learning_rate": 0.0003566134210995921,
      "loss": 2.1525,
      "step": 101278
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.9264845252037048,
      "learning_rate": 0.0003566094040289689,
      "loss": 2.1861,
      "step": 101279
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.8155291080474854,
      "learning_rate": 0.0003566053869478208,
      "loss": 2.0376,
      "step": 101280
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.8640187978744507,
      "learning_rate": 0.00035660136985614876,
      "loss": 2.1194,
      "step": 101281
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.9022015333175659,
      "learning_rate": 0.00035659735275395354,
      "loss": 2.1004,
      "step": 101282
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.8920800685882568,
      "learning_rate": 0.00035659333564123566,
      "loss": 2.0715,
      "step": 101283
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.9106515049934387,
      "learning_rate": 0.0003565893185179961,
      "loss": 2.2681,
      "step": 101284
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.8747574090957642,
      "learning_rate": 0.0003565853013842356,
      "loss": 2.0684,
      "step": 101285
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.860577404499054,
      "learning_rate": 0.0003565812842399546,
      "loss": 2.0915,
      "step": 101286
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.9094148874282837,
      "learning_rate": 0.00035657726708515424,
      "loss": 2.1838,
      "step": 101287
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.9593949913978577,
      "learning_rate": 0.00035657324991983505,
      "loss": 2.1195,
      "step": 101288
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.9472630620002747,
      "learning_rate": 0.00035656923274399784,
      "loss": 2.1497,
      "step": 101289
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.88950115442276,
      "learning_rate": 0.0003565652155576434,
      "loss": 2.2931,
      "step": 101290
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.8296186327934265,
      "learning_rate": 0.0003565611983607723,
      "loss": 2.0668,
      "step": 101291
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.8941987156867981,
      "learning_rate": 0.0003565571811533855,
      "loss": 1.8712,
      "step": 101292
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.9885492324829102,
      "learning_rate": 0.00035655316393548364,
      "loss": 2.1068,
      "step": 101293
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.8888205289840698,
      "learning_rate": 0.00035654914670706734,
      "loss": 2.2602,
      "step": 101294
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.8014568090438843,
      "learning_rate": 0.0003565451294681376,
      "loss": 2.0339,
      "step": 101295
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.8296434283256531,
      "learning_rate": 0.0003565411122186951,
      "loss": 2.1292,
      "step": 101296
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.9757315516471863,
      "learning_rate": 0.00035653709495874045,
      "loss": 2.2771,
      "step": 101297
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.8323931097984314,
      "learning_rate": 0.0003565330776882745,
      "loss": 1.9912,
      "step": 101298
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.8844203948974609,
      "learning_rate": 0.0003565290604072981,
      "loss": 2.0746,
      "step": 101299
    },
    {
      "epoch": 1.32,
      "grad_norm": 1.0400159358978271,
      "learning_rate": 0.00035652504311581176,
      "loss": 2.289,
      "step": 101300
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.9729970097541809,
      "learning_rate": 0.00035652102581381636,
      "loss": 2.1585,
      "step": 101301
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.9233561158180237,
      "learning_rate": 0.00035651700850131265,
      "loss": 2.1708,
      "step": 101302
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.9672884941101074,
      "learning_rate": 0.0003565129911783014,
      "loss": 2.3113,
      "step": 101303
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.9227334856987,
      "learning_rate": 0.00035650897384478325,
      "loss": 2.2104,
      "step": 101304
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.8534502983093262,
      "learning_rate": 0.000356504956500759,
      "loss": 2.1103,
      "step": 101305
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.9239787459373474,
      "learning_rate": 0.0003565009391462295,
      "loss": 1.9734,
      "step": 101306
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.9027302265167236,
      "learning_rate": 0.00035649692178119536,
      "loss": 2.1888,
      "step": 101307
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.9695752263069153,
      "learning_rate": 0.0003564929044056574,
      "loss": 2.1934,
      "step": 101308
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.854764461517334,
      "learning_rate": 0.00035648888701961636,
      "loss": 2.1234,
      "step": 101309
    },
    {
      "epoch": 1.32,
      "grad_norm": 1.1060694456100464,
      "learning_rate": 0.00035648486962307294,
      "loss": 2.0004,
      "step": 101310
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.9581952095031738,
      "learning_rate": 0.00035648085221602784,
      "loss": 2.1994,
      "step": 101311
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.8706266283988953,
      "learning_rate": 0.000356476834798482,
      "loss": 2.2315,
      "step": 101312
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.894520103931427,
      "learning_rate": 0.000356472817370436,
      "loss": 2.2623,
      "step": 101313
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.9249105453491211,
      "learning_rate": 0.0003564687999318906,
      "loss": 2.4149,
      "step": 101314
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.9339934587478638,
      "learning_rate": 0.0003564647824828467,
      "loss": 1.9335,
      "step": 101315
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.851327657699585,
      "learning_rate": 0.00035646076502330486,
      "loss": 2.1443,
      "step": 101316
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.815934419631958,
      "learning_rate": 0.0003564567475532659,
      "loss": 2.1248,
      "step": 101317
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.9995046257972717,
      "learning_rate": 0.00035645273007273056,
      "loss": 2.1838,
      "step": 101318
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.7742558121681213,
      "learning_rate": 0.0003564487125816996,
      "loss": 2.1293,
      "step": 101319
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.8242416977882385,
      "learning_rate": 0.00035644469508017373,
      "loss": 2.1073,
      "step": 101320
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.9970258474349976,
      "learning_rate": 0.00035644067756815373,
      "loss": 2.2303,
      "step": 101321
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.8533373475074768,
      "learning_rate": 0.0003564366600456404,
      "loss": 2.1851,
      "step": 101322
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.8486340641975403,
      "learning_rate": 0.0003564326425126343,
      "loss": 2.0963,
      "step": 101323
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.7935063242912292,
      "learning_rate": 0.00035642862496913644,
      "loss": 1.9664,
      "step": 101324
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.9893922209739685,
      "learning_rate": 0.0003564246074151474,
      "loss": 2.1068,
      "step": 101325
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.982480525970459,
      "learning_rate": 0.0003564205898506679,
      "loss": 2.3413,
      "step": 101326
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.9839118719100952,
      "learning_rate": 0.0003564165722756988,
      "loss": 2.1637,
      "step": 101327
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.9086236357688904,
      "learning_rate": 0.00035641255469024086,
      "loss": 2.1353,
      "step": 101328
    },
    {
      "epoch": 1.32,
      "grad_norm": 1.0013455152511597,
      "learning_rate": 0.00035640853709429466,
      "loss": 2.2848,
      "step": 101329
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.8016378283500671,
      "learning_rate": 0.00035640451948786104,
      "loss": 2.2497,
      "step": 101330
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.8375269770622253,
      "learning_rate": 0.0003564005018709409,
      "loss": 2.0771,
      "step": 101331
    },
    {
      "epoch": 1.32,
      "grad_norm": 1.0006598234176636,
      "learning_rate": 0.0003563964842435347,
      "loss": 2.1434,
      "step": 101332
    },
    {
      "epoch": 1.32,
      "grad_norm": 1.0657366514205933,
      "learning_rate": 0.00035639246660564336,
      "loss": 2.2851,
      "step": 101333
    },
    {
      "epoch": 1.32,
      "grad_norm": 1.0015816688537598,
      "learning_rate": 0.00035638844895726765,
      "loss": 2.4757,
      "step": 101334
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.9369299411773682,
      "learning_rate": 0.0003563844312984082,
      "loss": 2.1575,
      "step": 101335
    },
    {
      "epoch": 1.32,
      "grad_norm": 1.0051445960998535,
      "learning_rate": 0.00035638041362906584,
      "loss": 2.1213,
      "step": 101336
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.9482840895652771,
      "learning_rate": 0.00035637639594924127,
      "loss": 2.178,
      "step": 101337
    },
    {
      "epoch": 1.32,
      "grad_norm": 1.4178082942962646,
      "learning_rate": 0.0003563723782589353,
      "loss": 2.3387,
      "step": 101338
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.89412921667099,
      "learning_rate": 0.00035636836055814863,
      "loss": 2.3812,
      "step": 101339
    },
    {
      "epoch": 1.32,
      "grad_norm": 1.0261754989624023,
      "learning_rate": 0.000356364342846882,
      "loss": 2.1514,
      "step": 101340
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.9413368701934814,
      "learning_rate": 0.0003563603251251362,
      "loss": 1.9924,
      "step": 101341
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.9059465527534485,
      "learning_rate": 0.0003563563073929119,
      "loss": 2.1232,
      "step": 101342
    },
    {
      "epoch": 1.32,
      "grad_norm": 1.0727430582046509,
      "learning_rate": 0.00035635228965021,
      "loss": 1.8913,
      "step": 101343
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.8605508804321289,
      "learning_rate": 0.000356348271897031,
      "loss": 2.2007,
      "step": 101344
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.8827469944953918,
      "learning_rate": 0.00035634425413337594,
      "loss": 2.2252,
      "step": 101345
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.921645998954773,
      "learning_rate": 0.0003563402363592454,
      "loss": 2.0496,
      "step": 101346
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.9193776249885559,
      "learning_rate": 0.00035633621857464006,
      "loss": 2.245,
      "step": 101347
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.8820477724075317,
      "learning_rate": 0.00035633220077956084,
      "loss": 2.1763,
      "step": 101348
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.8936354517936707,
      "learning_rate": 0.00035632818297400833,
      "loss": 2.1167,
      "step": 101349
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.820443332195282,
      "learning_rate": 0.00035632416515798336,
      "loss": 2.0518,
      "step": 101350
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.8258249759674072,
      "learning_rate": 0.0003563201473314866,
      "loss": 2.1513,
      "step": 101351
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.9088808298110962,
      "learning_rate": 0.000356316129494519,
      "loss": 2.3478,
      "step": 101352
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.9595761299133301,
      "learning_rate": 0.00035631211164708113,
      "loss": 2.3024,
      "step": 101353
    },
    {
      "epoch": 1.32,
      "grad_norm": 1.1072053909301758,
      "learning_rate": 0.00035630809378917375,
      "loss": 2.1975,
      "step": 101354
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.9727959036827087,
      "learning_rate": 0.0003563040759207977,
      "loss": 2.138,
      "step": 101355
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.9301041960716248,
      "learning_rate": 0.00035630005804195355,
      "loss": 2.1183,
      "step": 101356
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.8782199621200562,
      "learning_rate": 0.0003562960401526422,
      "loss": 2.392,
      "step": 101357
    },
    {
      "epoch": 1.32,
      "grad_norm": 1.0534242391586304,
      "learning_rate": 0.0003562920222528644,
      "loss": 2.3622,
      "step": 101358
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.8434368371963501,
      "learning_rate": 0.0003562880043426208,
      "loss": 2.2364,
      "step": 101359
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.848388135433197,
      "learning_rate": 0.0003562839864219122,
      "loss": 1.9101,
      "step": 101360
    },
    {
      "epoch": 1.32,
      "grad_norm": 1.0740077495574951,
      "learning_rate": 0.00035627996849073936,
      "loss": 2.2255,
      "step": 101361
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.841076135635376,
      "learning_rate": 0.00035627595054910305,
      "loss": 2.2141,
      "step": 101362
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.9005298018455505,
      "learning_rate": 0.00035627193259700387,
      "loss": 2.1145,
      "step": 101363
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.8676465749740601,
      "learning_rate": 0.0003562679146344428,
      "loss": 2.1597,
      "step": 101364
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.8333991765975952,
      "learning_rate": 0.00035626389666142035,
      "loss": 2.1196,
      "step": 101365
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.9787273406982422,
      "learning_rate": 0.00035625987867793737,
      "loss": 2.1134,
      "step": 101366
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.9041128158569336,
      "learning_rate": 0.00035625586068399476,
      "loss": 2.1771,
      "step": 101367
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.8940152525901794,
      "learning_rate": 0.000356251842679593,
      "loss": 1.8305,
      "step": 101368
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.8892686367034912,
      "learning_rate": 0.000356247824664733,
      "loss": 2.1132,
      "step": 101369
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.9268772006034851,
      "learning_rate": 0.00035624380663941554,
      "loss": 2.1256,
      "step": 101370
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.9202102422714233,
      "learning_rate": 0.00035623978860364116,
      "loss": 2.2309,
      "step": 101371
    },
    {
      "epoch": 1.32,
      "grad_norm": 1.0130689144134521,
      "learning_rate": 0.00035623577055741084,
      "loss": 2.114,
      "step": 101372
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.8950028419494629,
      "learning_rate": 0.0003562317525007252,
      "loss": 2.0982,
      "step": 101373
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.8698019981384277,
      "learning_rate": 0.00035622773443358494,
      "loss": 2.209,
      "step": 101374
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.9518160820007324,
      "learning_rate": 0.000356223716355991,
      "loss": 2.183,
      "step": 101375
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.978035032749176,
      "learning_rate": 0.0003562196982679439,
      "loss": 2.1124,
      "step": 101376
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.8090733289718628,
      "learning_rate": 0.00035621568016944464,
      "loss": 2.1759,
      "step": 101377
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.9316860437393188,
      "learning_rate": 0.00035621166206049366,
      "loss": 2.185,
      "step": 101378
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.8718730211257935,
      "learning_rate": 0.00035620764394109195,
      "loss": 2.3009,
      "step": 101379
    },
    {
      "epoch": 1.32,
      "grad_norm": 1.0381263494491577,
      "learning_rate": 0.0003562036258112402,
      "loss": 2.1027,
      "step": 101380
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.849295973777771,
      "learning_rate": 0.00035619960767093914,
      "loss": 2.1815,
      "step": 101381
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.8251743316650391,
      "learning_rate": 0.00035619558952018946,
      "loss": 2.2971,
      "step": 101382
    },
    {
      "epoch": 1.32,
      "grad_norm": 1.0205646753311157,
      "learning_rate": 0.00035619157135899197,
      "loss": 2.0867,
      "step": 101383
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.9846821427345276,
      "learning_rate": 0.0003561875531873475,
      "loss": 2.0678,
      "step": 101384
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.8314400911331177,
      "learning_rate": 0.00035618353500525656,
      "loss": 2.3644,
      "step": 101385
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.8622444868087769,
      "learning_rate": 0.00035617951681272004,
      "loss": 2.1098,
      "step": 101386
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.8482855558395386,
      "learning_rate": 0.00035617549860973875,
      "loss": 1.8701,
      "step": 101387
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.8830922245979309,
      "learning_rate": 0.00035617148039631346,
      "loss": 2.1981,
      "step": 101388
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.9580780267715454,
      "learning_rate": 0.00035616746217244474,
      "loss": 2.1403,
      "step": 101389
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.9396240711212158,
      "learning_rate": 0.0003561634439381334,
      "loss": 2.0427,
      "step": 101390
    },
    {
      "epoch": 1.32,
      "grad_norm": 1.4933044910430908,
      "learning_rate": 0.00035615942569338033,
      "loss": 2.3352,
      "step": 101391
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.9369208812713623,
      "learning_rate": 0.000356155407438186,
      "loss": 1.9179,
      "step": 101392
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.907541036605835,
      "learning_rate": 0.0003561513891725514,
      "loss": 2.1558,
      "step": 101393
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.9887869954109192,
      "learning_rate": 0.00035614737089647723,
      "loss": 2.3824,
      "step": 101394
    },
    {
      "epoch": 1.32,
      "grad_norm": 1.213384747505188,
      "learning_rate": 0.0003561433526099642,
      "loss": 2.2415,
      "step": 101395
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.8058918118476868,
      "learning_rate": 0.000356139334313013,
      "loss": 2.4887,
      "step": 101396
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.9093547463417053,
      "learning_rate": 0.0003561353160056245,
      "loss": 1.9693,
      "step": 101397
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.8874359726905823,
      "learning_rate": 0.0003561312976877993,
      "loss": 2.0992,
      "step": 101398
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.8196064233779907,
      "learning_rate": 0.00035612727935953835,
      "loss": 2.1955,
      "step": 101399
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.9421346187591553,
      "learning_rate": 0.0003561232610208422,
      "loss": 2.2126,
      "step": 101400
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.9075758457183838,
      "learning_rate": 0.00035611924267171173,
      "loss": 2.319,
      "step": 101401
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.8951401710510254,
      "learning_rate": 0.0003561152243121476,
      "loss": 1.9646,
      "step": 101402
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.8442907929420471,
      "learning_rate": 0.00035611120594215053,
      "loss": 2.0754,
      "step": 101403
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.8519947528839111,
      "learning_rate": 0.00035610718756172145,
      "loss": 2.0973,
      "step": 101404
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.8487573266029358,
      "learning_rate": 0.00035610316917086094,
      "loss": 2.1945,
      "step": 101405
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.8605829477310181,
      "learning_rate": 0.00035609915076956976,
      "loss": 2.2324,
      "step": 101406
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.813395082950592,
      "learning_rate": 0.0003560951323578487,
      "loss": 1.9967,
      "step": 101407
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.8535061478614807,
      "learning_rate": 0.00035609111393569844,
      "loss": 2.249,
      "step": 101408
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.7938389778137207,
      "learning_rate": 0.0003560870955031199,
      "loss": 1.9728,
      "step": 101409
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.9209691882133484,
      "learning_rate": 0.0003560830770601137,
      "loss": 1.9329,
      "step": 101410
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.8738346695899963,
      "learning_rate": 0.00035607905860668047,
      "loss": 2.1082,
      "step": 101411
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.959965705871582,
      "learning_rate": 0.0003560750401428212,
      "loss": 2.2055,
      "step": 101412
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.8508623242378235,
      "learning_rate": 0.00035607102166853655,
      "loss": 2.0983,
      "step": 101413
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.9522421956062317,
      "learning_rate": 0.0003560670031838271,
      "loss": 2.4492,
      "step": 101414
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.9157698154449463,
      "learning_rate": 0.00035606298468869384,
      "loss": 2.3007,
      "step": 101415
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.9319608211517334,
      "learning_rate": 0.00035605896618313743,
      "loss": 2.2334,
      "step": 101416
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.9342395067214966,
      "learning_rate": 0.0003560549476671585,
      "loss": 2.2935,
      "step": 101417
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.8863401412963867,
      "learning_rate": 0.00035605092914075794,
      "loss": 2.3057,
      "step": 101418
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.8359551429748535,
      "learning_rate": 0.0003560469106039365,
      "loss": 2.3521,
      "step": 101419
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.889809250831604,
      "learning_rate": 0.00035604289205669483,
      "loss": 2.106,
      "step": 101420
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.9523336291313171,
      "learning_rate": 0.0003560388734990338,
      "loss": 1.9628,
      "step": 101421
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.8732526302337646,
      "learning_rate": 0.0003560348549309541,
      "loss": 2.4567,
      "step": 101422
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.8878746628761292,
      "learning_rate": 0.0003560308363524563,
      "loss": 2.238,
      "step": 101423
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.9085782766342163,
      "learning_rate": 0.0003560268177635414,
      "loss": 2.3414,
      "step": 101424
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.8417546153068542,
      "learning_rate": 0.00035602279916421007,
      "loss": 2.0969,
      "step": 101425
    },
    {
      "epoch": 1.32,
      "grad_norm": 1.09819495677948,
      "learning_rate": 0.00035601878055446307,
      "loss": 2.2714,
      "step": 101426
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.9022160172462463,
      "learning_rate": 0.000356014761934301,
      "loss": 2.08,
      "step": 101427
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.9235037565231323,
      "learning_rate": 0.0003560107433037249,
      "loss": 2.3369,
      "step": 101428
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.9461697936058044,
      "learning_rate": 0.0003560067246627353,
      "loss": 2.1278,
      "step": 101429
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.848468542098999,
      "learning_rate": 0.00035600270601133285,
      "loss": 1.9211,
      "step": 101430
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.9308121800422668,
      "learning_rate": 0.0003559986873495186,
      "loss": 2.1564,
      "step": 101431
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.8609482049942017,
      "learning_rate": 0.000355994668677293,
      "loss": 2.2362,
      "step": 101432
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.8981351852416992,
      "learning_rate": 0.00035599064999465704,
      "loss": 2.1604,
      "step": 101433
    },
    {
      "epoch": 1.32,
      "grad_norm": 1.1719751358032227,
      "learning_rate": 0.00035598663130161133,
      "loss": 2.0804,
      "step": 101434
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.8950806260108948,
      "learning_rate": 0.0003559826125981567,
      "loss": 2.0447,
      "step": 101435
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.9180543422698975,
      "learning_rate": 0.0003559785938842938,
      "loss": 2.0797,
      "step": 101436
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.9283645749092102,
      "learning_rate": 0.00035597457516002345,
      "loss": 1.9229,
      "step": 101437
    },
    {
      "epoch": 1.32,
      "grad_norm": 1.039770483970642,
      "learning_rate": 0.00035597055642534633,
      "loss": 2.0514,
      "step": 101438
    },
    {
      "epoch": 1.32,
      "grad_norm": 1.0181961059570312,
      "learning_rate": 0.0003559665376802632,
      "loss": 2.1557,
      "step": 101439
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.9982489347457886,
      "learning_rate": 0.0003559625189247749,
      "loss": 2.213,
      "step": 101440
    },
    {
      "epoch": 1.32,
      "grad_norm": 1.0145403146743774,
      "learning_rate": 0.0003559585001588821,
      "loss": 2.3563,
      "step": 101441
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.8712449669837952,
      "learning_rate": 0.0003559544813825855,
      "loss": 2.2198,
      "step": 101442
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.9089406728744507,
      "learning_rate": 0.000355950462595886,
      "loss": 2.1736,
      "step": 101443
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.9430093169212341,
      "learning_rate": 0.0003559464437987842,
      "loss": 2.0616,
      "step": 101444
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.9143939018249512,
      "learning_rate": 0.00035594242499128087,
      "loss": 2.4945,
      "step": 101445
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.8414613008499146,
      "learning_rate": 0.00035593840617337687,
      "loss": 2.257,
      "step": 101446
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.8562028408050537,
      "learning_rate": 0.00035593438734507276,
      "loss": 2.0904,
      "step": 101447
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.8602509498596191,
      "learning_rate": 0.0003559303685063694,
      "loss": 2.1742,
      "step": 101448
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.8456363081932068,
      "learning_rate": 0.00035592634965726767,
      "loss": 2.0307,
      "step": 101449
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.9108904600143433,
      "learning_rate": 0.0003559223307977681,
      "loss": 2.1979,
      "step": 101450
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.8613208532333374,
      "learning_rate": 0.00035591831192787143,
      "loss": 2.0861,
      "step": 101451
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.903332531452179,
      "learning_rate": 0.00035591429304757864,
      "loss": 2.2407,
      "step": 101452
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.8371202945709229,
      "learning_rate": 0.0003559102741568902,
      "loss": 2.1028,
      "step": 101453
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.8012620210647583,
      "learning_rate": 0.000355906255255807,
      "loss": 2.0331,
      "step": 101454
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.8720331192016602,
      "learning_rate": 0.0003559022363443298,
      "loss": 2.1238,
      "step": 101455
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.9267851114273071,
      "learning_rate": 0.0003558982174224593,
      "loss": 2.1518,
      "step": 101456
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.9829385280609131,
      "learning_rate": 0.00035589419849019633,
      "loss": 2.0439,
      "step": 101457
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.90260249376297,
      "learning_rate": 0.0003558901795475414,
      "loss": 2.0784,
      "step": 101458
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.869235098361969,
      "learning_rate": 0.0003558861605944957,
      "loss": 2.0127,
      "step": 101459
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.8681407570838928,
      "learning_rate": 0.0003558821416310595,
      "loss": 2.191,
      "step": 101460
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.8204209208488464,
      "learning_rate": 0.00035587812265723377,
      "loss": 2.3565,
      "step": 101461
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.8744118809700012,
      "learning_rate": 0.00035587410367301934,
      "loss": 2.3309,
      "step": 101462
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.8411940336227417,
      "learning_rate": 0.00035587008467841683,
      "loss": 2.2273,
      "step": 101463
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.9282894134521484,
      "learning_rate": 0.00035586606567342697,
      "loss": 2.0292,
      "step": 101464
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.8560674786567688,
      "learning_rate": 0.00035586204665805066,
      "loss": 2.3825,
      "step": 101465
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.8721662163734436,
      "learning_rate": 0.0003558580276322884,
      "loss": 2.2327,
      "step": 101466
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.8971641659736633,
      "learning_rate": 0.0003558540085961411,
      "loss": 2.1828,
      "step": 101467
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.8225038051605225,
      "learning_rate": 0.0003558499895496095,
      "loss": 2.0331,
      "step": 101468
    },
    {
      "epoch": 1.32,
      "grad_norm": 1.0413098335266113,
      "learning_rate": 0.00035584597049269443,
      "loss": 2.0224,
      "step": 101469
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.8773921132087708,
      "learning_rate": 0.00035584195142539644,
      "loss": 2.4315,
      "step": 101470
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.8310727477073669,
      "learning_rate": 0.0003558379323477164,
      "loss": 2.2483,
      "step": 101471
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.9015318751335144,
      "learning_rate": 0.0003558339132596551,
      "loss": 2.0785,
      "step": 101472
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.84550940990448,
      "learning_rate": 0.00035582989416121317,
      "loss": 1.9977,
      "step": 101473
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.8809425234794617,
      "learning_rate": 0.0003558258750523913,
      "loss": 2.2412,
      "step": 101474
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.907370924949646,
      "learning_rate": 0.00035582185593319046,
      "loss": 2.2928,
      "step": 101475
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.9418450593948364,
      "learning_rate": 0.00035581783680361135,
      "loss": 2.0481,
      "step": 101476
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.833082914352417,
      "learning_rate": 0.00035581381766365447,
      "loss": 2.2792,
      "step": 101477
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.8832457661628723,
      "learning_rate": 0.0003558097985133209,
      "loss": 2.1984,
      "step": 101478
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.8978390693664551,
      "learning_rate": 0.0003558057793526111,
      "loss": 2.2822,
      "step": 101479
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.8616344928741455,
      "learning_rate": 0.00035580176018152606,
      "loss": 2.165,
      "step": 101480
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.8838837742805481,
      "learning_rate": 0.00035579774100006646,
      "loss": 1.96,
      "step": 101481
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.8092490434646606,
      "learning_rate": 0.0003557937218082329,
      "loss": 2.1157,
      "step": 101482
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.9620712399482727,
      "learning_rate": 0.00035578970260602633,
      "loss": 2.1169,
      "step": 101483
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.8489503264427185,
      "learning_rate": 0.0003557856833934473,
      "loss": 2.1328,
      "step": 101484
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.9899225234985352,
      "learning_rate": 0.0003557816641704967,
      "loss": 1.9821,
      "step": 101485
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.832757830619812,
      "learning_rate": 0.00035577764493717527,
      "loss": 2.1748,
      "step": 101486
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.8873366713523865,
      "learning_rate": 0.00035577362569348366,
      "loss": 2.1461,
      "step": 101487
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.9357647895812988,
      "learning_rate": 0.0003557696064394227,
      "loss": 2.2516,
      "step": 101488
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.9773862361907959,
      "learning_rate": 0.0003557655871749932,
      "loss": 2.0351,
      "step": 101489
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.856889545917511,
      "learning_rate": 0.0003557615679001957,
      "loss": 2.0728,
      "step": 101490
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.8572024703025818,
      "learning_rate": 0.0003557575486150311,
      "loss": 2.1477,
      "step": 101491
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.869596004486084,
      "learning_rate": 0.0003557535293195002,
      "loss": 2.0318,
      "step": 101492
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.9933111667633057,
      "learning_rate": 0.00035574951001360365,
      "loss": 2.0179,
      "step": 101493
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.8259966969490051,
      "learning_rate": 0.00035574549069734216,
      "loss": 2.0219,
      "step": 101494
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.8214914202690125,
      "learning_rate": 0.00035574147137071657,
      "loss": 1.9587,
      "step": 101495
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.9604012966156006,
      "learning_rate": 0.0003557374520337275,
      "loss": 2.2669,
      "step": 101496
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.871486485004425,
      "learning_rate": 0.00035573343268637583,
      "loss": 2.352,
      "step": 101497
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.8031863570213318,
      "learning_rate": 0.00035572941332866236,
      "loss": 2.1419,
      "step": 101498
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.8547133207321167,
      "learning_rate": 0.0003557253939605876,
      "loss": 2.022,
      "step": 101499
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.8535143136978149,
      "learning_rate": 0.00035572137458215255,
      "loss": 1.9822,
      "step": 101500
    },
    {
      "epoch": 1.32,
      "grad_norm": 1.0678311586380005,
      "learning_rate": 0.0003557173551933578,
      "loss": 2.2673,
      "step": 101501
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.8757668733596802,
      "learning_rate": 0.00035571333579420414,
      "loss": 2.1899,
      "step": 101502
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.8794326782226562,
      "learning_rate": 0.00035570931638469227,
      "loss": 2.0357,
      "step": 101503
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.7978987693786621,
      "learning_rate": 0.0003557052969648231,
      "loss": 2.2593,
      "step": 101504
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.8164926767349243,
      "learning_rate": 0.0003557012775345972,
      "loss": 2.2763,
      "step": 101505
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.8083850145339966,
      "learning_rate": 0.0003556972580940154,
      "loss": 2.1374,
      "step": 101506
    },
    {
      "epoch": 1.32,
      "grad_norm": 1.0445631742477417,
      "learning_rate": 0.0003556932386430784,
      "loss": 2.1846,
      "step": 101507
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.857197642326355,
      "learning_rate": 0.00035568921918178694,
      "loss": 2.253,
      "step": 101508
    },
    {
      "epoch": 1.32,
      "grad_norm": 1.013171911239624,
      "learning_rate": 0.0003556851997101418,
      "loss": 2.2323,
      "step": 101509
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.9287766814231873,
      "learning_rate": 0.0003556811802281438,
      "loss": 2.144,
      "step": 101510
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.8888167142868042,
      "learning_rate": 0.00035567716073579364,
      "loss": 2.1427,
      "step": 101511
    },
    {
      "epoch": 1.32,
      "grad_norm": 1.1994233131408691,
      "learning_rate": 0.00035567314123309193,
      "loss": 1.9869,
      "step": 101512
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.9688516855239868,
      "learning_rate": 0.00035566912172003967,
      "loss": 2.4921,
      "step": 101513
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.9365752935409546,
      "learning_rate": 0.0003556651021966374,
      "loss": 2.1056,
      "step": 101514
    },
    {
      "epoch": 1.32,
      "grad_norm": 1.000339150428772,
      "learning_rate": 0.00035566108266288593,
      "loss": 2.3285,
      "step": 101515
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.9580523371696472,
      "learning_rate": 0.0003556570631187861,
      "loss": 1.989,
      "step": 101516
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.9801767468452454,
      "learning_rate": 0.00035565304356433843,
      "loss": 2.1891,
      "step": 101517
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.9077235460281372,
      "learning_rate": 0.00035564902399954387,
      "loss": 2.2035,
      "step": 101518
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.8627371788024902,
      "learning_rate": 0.0003556450044244031,
      "loss": 2.2918,
      "step": 101519
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.8780589699745178,
      "learning_rate": 0.0003556409848389169,
      "loss": 2.0234,
      "step": 101520
    },
    {
      "epoch": 1.32,
      "grad_norm": 1.034842848777771,
      "learning_rate": 0.0003556369652430859,
      "loss": 2.2338,
      "step": 101521
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.864946186542511,
      "learning_rate": 0.0003556329456369111,
      "loss": 1.9164,
      "step": 101522
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.7742577791213989,
      "learning_rate": 0.000355628926020393,
      "loss": 2.3821,
      "step": 101523
    },
    {
      "epoch": 1.32,
      "grad_norm": 1.1053452491760254,
      "learning_rate": 0.00035562490639353235,
      "loss": 2.2779,
      "step": 101524
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.8857076168060303,
      "learning_rate": 0.00035562088675633005,
      "loss": 2.2065,
      "step": 101525
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.9069889783859253,
      "learning_rate": 0.00035561686710878685,
      "loss": 2.2773,
      "step": 101526
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.8371722102165222,
      "learning_rate": 0.0003556128474509033,
      "loss": 2.3383,
      "step": 101527
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.8098952174186707,
      "learning_rate": 0.0003556088277826804,
      "loss": 2.0603,
      "step": 101528
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.9038671255111694,
      "learning_rate": 0.0003556048081041187,
      "loss": 2.226,
      "step": 101529
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.7767361402511597,
      "learning_rate": 0.00035560078841521893,
      "loss": 2.4386,
      "step": 101530
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.894096851348877,
      "learning_rate": 0.000355596768715982,
      "loss": 2.1312,
      "step": 101531
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.9246311187744141,
      "learning_rate": 0.00035559274900640866,
      "loss": 2.3081,
      "step": 101532
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.9247376322746277,
      "learning_rate": 0.00035558872928649946,
      "loss": 2.3316,
      "step": 101533
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.8642542362213135,
      "learning_rate": 0.0003555847095562554,
      "loss": 2.3194,
      "step": 101534
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.8601668477058411,
      "learning_rate": 0.00035558068981567694,
      "loss": 2.2816,
      "step": 101535
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.8622601628303528,
      "learning_rate": 0.0003555766700647651,
      "loss": 2.184,
      "step": 101536
    },
    {
      "epoch": 1.32,
      "grad_norm": 1.7540112733840942,
      "learning_rate": 0.0003555726503035204,
      "loss": 2.1534,
      "step": 101537
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.8794637322425842,
      "learning_rate": 0.00035556863053194383,
      "loss": 2.2051,
      "step": 101538
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.8501335382461548,
      "learning_rate": 0.0003555646107500359,
      "loss": 2.3435,
      "step": 101539
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.8397205471992493,
      "learning_rate": 0.0003555605909577975,
      "loss": 2.1281,
      "step": 101540
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.8187829256057739,
      "learning_rate": 0.0003555565711552293,
      "loss": 2.2329,
      "step": 101541
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.8233550786972046,
      "learning_rate": 0.0003555525513423321,
      "loss": 2.0252,
      "step": 101542
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.7901600003242493,
      "learning_rate": 0.00035554853151910663,
      "loss": 2.0586,
      "step": 101543
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.8931440114974976,
      "learning_rate": 0.0003555445116855537,
      "loss": 2.2631,
      "step": 101544
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.8962798118591309,
      "learning_rate": 0.0003555404918416739,
      "loss": 2.2265,
      "step": 101545
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.8094667792320251,
      "learning_rate": 0.00035553647198746814,
      "loss": 2.1128,
      "step": 101546
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.8922085762023926,
      "learning_rate": 0.0003555324521229371,
      "loss": 1.9481,
      "step": 101547
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.827688992023468,
      "learning_rate": 0.0003555284322480815,
      "loss": 2.1779,
      "step": 101548
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.9049224853515625,
      "learning_rate": 0.0003555244123629021,
      "loss": 2.2936,
      "step": 101549
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.916155993938446,
      "learning_rate": 0.00035552039246739974,
      "loss": 1.9842,
      "step": 101550
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.9339388608932495,
      "learning_rate": 0.000355516372561575,
      "loss": 2.4069,
      "step": 101551
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.8575513958930969,
      "learning_rate": 0.0003555123526454288,
      "loss": 2.1542,
      "step": 101552
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.869654655456543,
      "learning_rate": 0.00035550833271896174,
      "loss": 2.218,
      "step": 101553
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.9613012075424194,
      "learning_rate": 0.00035550431278217465,
      "loss": 2.08,
      "step": 101554
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.976654589176178,
      "learning_rate": 0.00035550029283506823,
      "loss": 2.2871,
      "step": 101555
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.9184238910675049,
      "learning_rate": 0.0003554962728776433,
      "loss": 2.4289,
      "step": 101556
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.7846184372901917,
      "learning_rate": 0.0003554922529099006,
      "loss": 2.2294,
      "step": 101557
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.8262834548950195,
      "learning_rate": 0.0003554882329318408,
      "loss": 1.8607,
      "step": 101558
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.9028535485267639,
      "learning_rate": 0.00035548421294346466,
      "loss": 2.0101,
      "step": 101559
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.8582965731620789,
      "learning_rate": 0.00035548019294477304,
      "loss": 2.2227,
      "step": 101560
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.856143593788147,
      "learning_rate": 0.00035547617293576653,
      "loss": 2.1974,
      "step": 101561
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.880693793296814,
      "learning_rate": 0.00035547215291644596,
      "loss": 2.1025,
      "step": 101562
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.9764044284820557,
      "learning_rate": 0.00035546813288681213,
      "loss": 2.1572,
      "step": 101563
    },
    {
      "epoch": 1.32,
      "grad_norm": 1.0187139511108398,
      "learning_rate": 0.0003554641128468657,
      "loss": 2.554,
      "step": 101564
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.8198924660682678,
      "learning_rate": 0.0003554600927966074,
      "loss": 2.075,
      "step": 101565
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.9112504720687866,
      "learning_rate": 0.00035545607273603804,
      "loss": 2.3,
      "step": 101566
    },
    {
      "epoch": 1.32,
      "grad_norm": 1.0905890464782715,
      "learning_rate": 0.0003554520526651583,
      "loss": 2.0733,
      "step": 101567
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.9781780242919922,
      "learning_rate": 0.0003554480325839691,
      "loss": 2.1091,
      "step": 101568
    },
    {
      "epoch": 1.32,
      "grad_norm": 1.1077613830566406,
      "learning_rate": 0.000355444012492471,
      "loss": 1.998,
      "step": 101569
    },
    {
      "epoch": 1.32,
      "grad_norm": 1.0097085237503052,
      "learning_rate": 0.00035543999239066485,
      "loss": 2.2167,
      "step": 101570
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.8443113565444946,
      "learning_rate": 0.0003554359722785513,
      "loss": 2.0177,
      "step": 101571
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.9596110582351685,
      "learning_rate": 0.0003554319521561312,
      "loss": 2.269,
      "step": 101572
    },
    {
      "epoch": 1.32,
      "grad_norm": 1.0388357639312744,
      "learning_rate": 0.00035542793202340513,
      "loss": 2.2887,
      "step": 101573
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.9575850367546082,
      "learning_rate": 0.00035542391188037416,
      "loss": 2.0752,
      "step": 101574
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.885334849357605,
      "learning_rate": 0.0003554198917270387,
      "loss": 2.1635,
      "step": 101575
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.8404681086540222,
      "learning_rate": 0.00035541587156339967,
      "loss": 1.991,
      "step": 101576
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.875511884689331,
      "learning_rate": 0.0003554118513894579,
      "loss": 2.4665,
      "step": 101577
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.9975028038024902,
      "learning_rate": 0.00035540783120521384,
      "loss": 2.2432,
      "step": 101578
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.8538311123847961,
      "learning_rate": 0.0003554038110106685,
      "loss": 2.3552,
      "step": 101579
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.9617778658866882,
      "learning_rate": 0.0003553997908058225,
      "loss": 2.1152,
      "step": 101580
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.8811752796173096,
      "learning_rate": 0.0003553957705906767,
      "loss": 2.3682,
      "step": 101581
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.8243811726570129,
      "learning_rate": 0.00035539175036523174,
      "loss": 2.0184,
      "step": 101582
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.8600219488143921,
      "learning_rate": 0.0003553877301294884,
      "loss": 2.1114,
      "step": 101583
    },
    {
      "epoch": 1.32,
      "grad_norm": 1.045366883277893,
      "learning_rate": 0.0003553837098834475,
      "loss": 2.1317,
      "step": 101584
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.9586094617843628,
      "learning_rate": 0.00035537968962710966,
      "loss": 2.2408,
      "step": 101585
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.9219191670417786,
      "learning_rate": 0.00035537566936047574,
      "loss": 2.313,
      "step": 101586
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.9834632277488708,
      "learning_rate": 0.0003553716490835464,
      "loss": 2.0876,
      "step": 101587
    },
    {
      "epoch": 1.32,
      "grad_norm": 1.1578599214553833,
      "learning_rate": 0.0003553676287963224,
      "loss": 2.2473,
      "step": 101588
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.9291166663169861,
      "learning_rate": 0.00035536360849880455,
      "loss": 2.0231,
      "step": 101589
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.8325362801551819,
      "learning_rate": 0.00035535958819099355,
      "loss": 2.1633,
      "step": 101590
    },
    {
      "epoch": 1.32,
      "grad_norm": 1.248151183128357,
      "learning_rate": 0.00035535556787289017,
      "loss": 2.3267,
      "step": 101591
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.9356124401092529,
      "learning_rate": 0.0003553515475444951,
      "loss": 2.0406,
      "step": 101592
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.9358693957328796,
      "learning_rate": 0.00035534752720580917,
      "loss": 2.0838,
      "step": 101593
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.851223886013031,
      "learning_rate": 0.00035534350685683313,
      "loss": 2.1626,
      "step": 101594
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.8738873600959778,
      "learning_rate": 0.00035533948649756763,
      "loss": 2.1624,
      "step": 101595
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.8888406157493591,
      "learning_rate": 0.0003553354661280135,
      "loss": 2.1291,
      "step": 101596
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.8861188292503357,
      "learning_rate": 0.00035533144574817146,
      "loss": 2.0578,
      "step": 101597
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.8024939894676208,
      "learning_rate": 0.00035532742535804225,
      "loss": 2.0451,
      "step": 101598
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.8600123524665833,
      "learning_rate": 0.0003553234049576267,
      "loss": 2.3995,
      "step": 101599
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.8081066608428955,
      "learning_rate": 0.0003553193845469254,
      "loss": 1.8852,
      "step": 101600
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.8891449570655823,
      "learning_rate": 0.0003553153641259391,
      "loss": 2.205,
      "step": 101601
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.8920888304710388,
      "learning_rate": 0.0003553113436946688,
      "loss": 2.1841,
      "step": 101602
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.8485461473464966,
      "learning_rate": 0.000355307323253115,
      "loss": 2.0972,
      "step": 101603
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.8581995368003845,
      "learning_rate": 0.0003553033028012785,
      "loss": 2.0358,
      "step": 101604
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.8564586639404297,
      "learning_rate": 0.00035529928233916014,
      "loss": 2.3109,
      "step": 101605
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.8397106528282166,
      "learning_rate": 0.00035529526186676053,
      "loss": 2.1322,
      "step": 101606
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.9393043518066406,
      "learning_rate": 0.00035529124138408046,
      "loss": 2.2627,
      "step": 101607
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.8932763338088989,
      "learning_rate": 0.00035528722089112085,
      "loss": 2.4062,
      "step": 101608
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.8062936067581177,
      "learning_rate": 0.0003552832003878822,
      "loss": 2.3291,
      "step": 101609
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.8881320953369141,
      "learning_rate": 0.00035527917987436536,
      "loss": 1.9638,
      "step": 101610
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.9379337430000305,
      "learning_rate": 0.0003552751593505711,
      "loss": 1.9174,
      "step": 101611
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.9749612212181091,
      "learning_rate": 0.0003552711388165001,
      "loss": 2.3079,
      "step": 101612
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.8402755856513977,
      "learning_rate": 0.0003552671182721532,
      "loss": 2.1118,
      "step": 101613
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.8339772820472717,
      "learning_rate": 0.0003552630977175311,
      "loss": 2.1289,
      "step": 101614
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.8856161236763,
      "learning_rate": 0.00035525907715263456,
      "loss": 2.0662,
      "step": 101615
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.9479069709777832,
      "learning_rate": 0.00035525505657746424,
      "loss": 2.2024,
      "step": 101616
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.863921046257019,
      "learning_rate": 0.00035525103599202105,
      "loss": 2.1096,
      "step": 101617
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.8692783117294312,
      "learning_rate": 0.00035524701539630565,
      "loss": 2.1481,
      "step": 101618
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.8092696666717529,
      "learning_rate": 0.00035524299479031867,
      "loss": 2.092,
      "step": 101619
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.8358729481697083,
      "learning_rate": 0.0003552389741740611,
      "loss": 2.2185,
      "step": 101620
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.9295811653137207,
      "learning_rate": 0.00035523495354753353,
      "loss": 2.1327,
      "step": 101621
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.9357185959815979,
      "learning_rate": 0.00035523093291073667,
      "loss": 2.0337,
      "step": 101622
    },
    {
      "epoch": 1.32,
      "grad_norm": 1.007254958152771,
      "learning_rate": 0.0003552269122636714,
      "loss": 2.2211,
      "step": 101623
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.9329538345336914,
      "learning_rate": 0.0003552228916063384,
      "loss": 2.0434,
      "step": 101624
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.9174211025238037,
      "learning_rate": 0.0003552188709387384,
      "loss": 2.0185,
      "step": 101625
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.8925053477287292,
      "learning_rate": 0.0003552148502608722,
      "loss": 2.3979,
      "step": 101626
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.8018405437469482,
      "learning_rate": 0.00035521082957274053,
      "loss": 2.1435,
      "step": 101627
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.9944870471954346,
      "learning_rate": 0.000355206808874344,
      "loss": 2.1219,
      "step": 101628
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.9332563281059265,
      "learning_rate": 0.0003552027881656837,
      "loss": 2.1561,
      "step": 101629
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.8817505836486816,
      "learning_rate": 0.00035519876744676,
      "loss": 2.0117,
      "step": 101630
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.8465269804000854,
      "learning_rate": 0.0003551947467175738,
      "loss": 1.8549,
      "step": 101631
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.9630212187767029,
      "learning_rate": 0.00035519072597812594,
      "loss": 2.2167,
      "step": 101632
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.898367702960968,
      "learning_rate": 0.00035518670522841707,
      "loss": 2.3351,
      "step": 101633
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.7717511057853699,
      "learning_rate": 0.0003551826844684479,
      "loss": 2.0319,
      "step": 101634
    },
    {
      "epoch": 1.32,
      "grad_norm": 1.0638697147369385,
      "learning_rate": 0.00035517866369821926,
      "loss": 2.3096,
      "step": 101635
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.9872023463249207,
      "learning_rate": 0.00035517464291773184,
      "loss": 2.2148,
      "step": 101636
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.8291665315628052,
      "learning_rate": 0.0003551706221269864,
      "loss": 2.1245,
      "step": 101637
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.9487178325653076,
      "learning_rate": 0.0003551666013259838,
      "loss": 2.2422,
      "step": 101638
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.93265700340271,
      "learning_rate": 0.0003551625805147246,
      "loss": 2.1156,
      "step": 101639
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.8979513049125671,
      "learning_rate": 0.0003551585596932096,
      "loss": 2.1046,
      "step": 101640
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.8488380312919617,
      "learning_rate": 0.00035515453886143957,
      "loss": 2.2004,
      "step": 101641
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.9331960678100586,
      "learning_rate": 0.0003551505180194154,
      "loss": 2.3104,
      "step": 101642
    },
    {
      "epoch": 1.32,
      "grad_norm": 1.267330527305603,
      "learning_rate": 0.0003551464971671377,
      "loss": 2.2031,
      "step": 101643
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.981343150138855,
      "learning_rate": 0.00035514247630460705,
      "loss": 2.2058,
      "step": 101644
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.7934556603431702,
      "learning_rate": 0.00035513845543182454,
      "loss": 2.1968,
      "step": 101645
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.8895010352134705,
      "learning_rate": 0.0003551344345487907,
      "loss": 2.1281,
      "step": 101646
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.973152756690979,
      "learning_rate": 0.0003551304136555063,
      "loss": 2.441,
      "step": 101647
    },
    {
      "epoch": 1.32,
      "grad_norm": 1.2761473655700684,
      "learning_rate": 0.00035512639275197215,
      "loss": 2.0887,
      "step": 101648
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.890920877456665,
      "learning_rate": 0.00035512237183818896,
      "loss": 2.3558,
      "step": 101649
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.8660756945610046,
      "learning_rate": 0.00035511835091415746,
      "loss": 2.2835,
      "step": 101650
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.9477977156639099,
      "learning_rate": 0.0003551143299798784,
      "loss": 2.4108,
      "step": 101651
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.8232834935188293,
      "learning_rate": 0.0003551103090353526,
      "loss": 2.0962,
      "step": 101652
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.798514187335968,
      "learning_rate": 0.0003551062880805807,
      "loss": 2.1097,
      "step": 101653
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.9265995621681213,
      "learning_rate": 0.0003551022671155636,
      "loss": 1.9336,
      "step": 101654
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.8456501364707947,
      "learning_rate": 0.0003550982461403018,
      "loss": 2.387,
      "step": 101655
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.9297744035720825,
      "learning_rate": 0.0003550942251547963,
      "loss": 2.0112,
      "step": 101656
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.8180761933326721,
      "learning_rate": 0.00035509020415904776,
      "loss": 2.1341,
      "step": 101657
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.8696499466896057,
      "learning_rate": 0.00035508618315305686,
      "loss": 2.1371,
      "step": 101658
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.8076090216636658,
      "learning_rate": 0.00035508216213682437,
      "loss": 2.1962,
      "step": 101659
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.8967541456222534,
      "learning_rate": 0.00035507814111035115,
      "loss": 1.8945,
      "step": 101660
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.8962213397026062,
      "learning_rate": 0.0003550741200736378,
      "loss": 2.3237,
      "step": 101661
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.8953835368156433,
      "learning_rate": 0.00035507009902668515,
      "loss": 2.2113,
      "step": 101662
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.9862362742424011,
      "learning_rate": 0.00035506607796949394,
      "loss": 2.1991,
      "step": 101663
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.9117991328239441,
      "learning_rate": 0.0003550620569020649,
      "loss": 2.3059,
      "step": 101664
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.9705334901809692,
      "learning_rate": 0.00035505803582439873,
      "loss": 2.1301,
      "step": 101665
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.8397902250289917,
      "learning_rate": 0.00035505401473649636,
      "loss": 2.3471,
      "step": 101666
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.8942074179649353,
      "learning_rate": 0.0003550499936383583,
      "loss": 2.1977,
      "step": 101667
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.8780379295349121,
      "learning_rate": 0.00035504597252998543,
      "loss": 2.0907,
      "step": 101668
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.9029498100280762,
      "learning_rate": 0.0003550419514113785,
      "loss": 2.1723,
      "step": 101669
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.9218766093254089,
      "learning_rate": 0.00035503793028253823,
      "loss": 2.1911,
      "step": 101670
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.804402768611908,
      "learning_rate": 0.0003550339091434653,
      "loss": 2.0858,
      "step": 101671
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.8441992998123169,
      "learning_rate": 0.00035502988799416066,
      "loss": 2.3488,
      "step": 101672
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.9178524017333984,
      "learning_rate": 0.00035502586683462483,
      "loss": 2.1017,
      "step": 101673
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.8483853936195374,
      "learning_rate": 0.00035502184566485863,
      "loss": 2.1629,
      "step": 101674
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.8564188480377197,
      "learning_rate": 0.00035501782448486293,
      "loss": 2.2837,
      "step": 101675
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.8617733716964722,
      "learning_rate": 0.00035501380329463827,
      "loss": 2.0297,
      "step": 101676
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.994591236114502,
      "learning_rate": 0.00035500978209418557,
      "loss": 2.2563,
      "step": 101677
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.8180150389671326,
      "learning_rate": 0.00035500576088350554,
      "loss": 2.0769,
      "step": 101678
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.891265869140625,
      "learning_rate": 0.00035500173966259883,
      "loss": 2.3546,
      "step": 101679
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.9247442483901978,
      "learning_rate": 0.0003549977184314663,
      "loss": 1.9966,
      "step": 101680
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.8183590173721313,
      "learning_rate": 0.00035499369719010866,
      "loss": 2.2649,
      "step": 101681
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.8721021413803101,
      "learning_rate": 0.00035498967593852656,
      "loss": 2.1629,
      "step": 101682
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.8826063275337219,
      "learning_rate": 0.000354985654676721,
      "loss": 2.1448,
      "step": 101683
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.8874698877334595,
      "learning_rate": 0.0003549816334046925,
      "loss": 2.1136,
      "step": 101684
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.8412950038909912,
      "learning_rate": 0.00035497761212244185,
      "loss": 2.2349,
      "step": 101685
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.867979884147644,
      "learning_rate": 0.0003549735908299698,
      "loss": 2.3746,
      "step": 101686
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.8440576195716858,
      "learning_rate": 0.0003549695695272772,
      "loss": 2.2172,
      "step": 101687
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.8122638463973999,
      "learning_rate": 0.00035496554821436464,
      "loss": 2.1681,
      "step": 101688
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.7952166199684143,
      "learning_rate": 0.00035496152689123295,
      "loss": 2.1769,
      "step": 101689
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.940634548664093,
      "learning_rate": 0.000354957505557883,
      "loss": 2.1599,
      "step": 101690
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.8535249829292297,
      "learning_rate": 0.0003549534842143153,
      "loss": 2.1884,
      "step": 101691
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.989784836769104,
      "learning_rate": 0.0003549494628605307,
      "loss": 2.391,
      "step": 101692
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.852569043636322,
      "learning_rate": 0.0003549454414965301,
      "loss": 2.1247,
      "step": 101693
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.8093953728675842,
      "learning_rate": 0.00035494142012231395,
      "loss": 2.179,
      "step": 101694
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.8553715348243713,
      "learning_rate": 0.00035493739873788315,
      "loss": 2.1516,
      "step": 101695
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.8573736548423767,
      "learning_rate": 0.0003549333773432385,
      "loss": 2.3045,
      "step": 101696
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.9301313161849976,
      "learning_rate": 0.00035492935593838074,
      "loss": 2.1191,
      "step": 101697
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.7929182648658752,
      "learning_rate": 0.00035492533452331044,
      "loss": 2.1023,
      "step": 101698
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.8227084875106812,
      "learning_rate": 0.0003549213130980287,
      "loss": 2.245,
      "step": 101699
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.842766284942627,
      "learning_rate": 0.00035491729166253587,
      "loss": 2.0803,
      "step": 101700
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.9555577635765076,
      "learning_rate": 0.0003549132702168329,
      "loss": 2.3428,
      "step": 101701
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.8089897632598877,
      "learning_rate": 0.00035490924876092065,
      "loss": 2.0991,
      "step": 101702
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.9067333936691284,
      "learning_rate": 0.0003549052272947996,
      "loss": 1.8829,
      "step": 101703
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.7943199276924133,
      "learning_rate": 0.00035490120581847063,
      "loss": 2.1827,
      "step": 101704
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.9460431933403015,
      "learning_rate": 0.0003548971843319346,
      "loss": 2.2271,
      "step": 101705
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.859114944934845,
      "learning_rate": 0.00035489316283519206,
      "loss": 2.1919,
      "step": 101706
    },
    {
      "epoch": 1.32,
      "grad_norm": 1.0471975803375244,
      "learning_rate": 0.0003548891413282438,
      "loss": 2.0833,
      "step": 101707
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.8618994355201721,
      "learning_rate": 0.0003548851198110907,
      "loss": 2.1635,
      "step": 101708
    },
    {
      "epoch": 1.32,
      "grad_norm": 1.2966893911361694,
      "learning_rate": 0.0003548810982837334,
      "loss": 2.1437,
      "step": 101709
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.8638579845428467,
      "learning_rate": 0.00035487707674617265,
      "loss": 2.2335,
      "step": 101710
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.8368918895721436,
      "learning_rate": 0.00035487305519840934,
      "loss": 2.2017,
      "step": 101711
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.8732354640960693,
      "learning_rate": 0.0003548690336404439,
      "loss": 2.2708,
      "step": 101712
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.9375280737876892,
      "learning_rate": 0.0003548650120722773,
      "loss": 2.0907,
      "step": 101713
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.9419721364974976,
      "learning_rate": 0.0003548609904939104,
      "loss": 2.239,
      "step": 101714
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.8131513595581055,
      "learning_rate": 0.0003548569689053437,
      "loss": 2.1483,
      "step": 101715
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.9854085445404053,
      "learning_rate": 0.000354852947306578,
      "loss": 2.1508,
      "step": 101716
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.9391918182373047,
      "learning_rate": 0.00035484892569761427,
      "loss": 2.0335,
      "step": 101717
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.8712798357009888,
      "learning_rate": 0.000354844904078453,
      "loss": 1.9645,
      "step": 101718
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.9077289700508118,
      "learning_rate": 0.0003548408824490949,
      "loss": 2.1993,
      "step": 101719
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.8541074395179749,
      "learning_rate": 0.0003548368608095411,
      "loss": 2.3267,
      "step": 101720
    },
    {
      "epoch": 1.32,
      "grad_norm": 1.0174301862716675,
      "learning_rate": 0.0003548328391597919,
      "loss": 2.1714,
      "step": 101721
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.7984763979911804,
      "learning_rate": 0.00035482881749984826,
      "loss": 2.1517,
      "step": 101722
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.9002288579940796,
      "learning_rate": 0.00035482479582971095,
      "loss": 2.0311,
      "step": 101723
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.8532171249389648,
      "learning_rate": 0.00035482077414938066,
      "loss": 2.1069,
      "step": 101724
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.8946475982666016,
      "learning_rate": 0.0003548167524588582,
      "loss": 2.189,
      "step": 101725
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.8006603717803955,
      "learning_rate": 0.0003548127307581442,
      "loss": 2.1363,
      "step": 101726
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.9249882698059082,
      "learning_rate": 0.00035480870904723956,
      "loss": 2.2421,
      "step": 101727
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.983470618724823,
      "learning_rate": 0.0003548046873261449,
      "loss": 2.0197,
      "step": 101728
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.8245841860771179,
      "learning_rate": 0.00035480066559486097,
      "loss": 2.0586,
      "step": 101729
    },
    {
      "epoch": 1.32,
      "grad_norm": 1.0121794939041138,
      "learning_rate": 0.0003547966438533886,
      "loss": 2.4096,
      "step": 101730
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.8812842965126038,
      "learning_rate": 0.0003547926221017285,
      "loss": 2.2534,
      "step": 101731
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.8518321514129639,
      "learning_rate": 0.0003547886003398814,
      "loss": 2.0946,
      "step": 101732
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.8723189830780029,
      "learning_rate": 0.0003547845785678481,
      "loss": 2.2483,
      "step": 101733
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.8368874192237854,
      "learning_rate": 0.00035478055678562934,
      "loss": 2.1803,
      "step": 101734
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.8277992606163025,
      "learning_rate": 0.0003547765349932258,
      "loss": 2.1665,
      "step": 101735
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.9106060862541199,
      "learning_rate": 0.0003547725131906382,
      "loss": 2.222,
      "step": 101736
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.8148492574691772,
      "learning_rate": 0.0003547684913778674,
      "loss": 2.1095,
      "step": 101737
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.8584705591201782,
      "learning_rate": 0.0003547644695549142,
      "loss": 2.2314,
      "step": 101738
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.8819918036460876,
      "learning_rate": 0.0003547604477217791,
      "loss": 2.214,
      "step": 101739
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.8759834170341492,
      "learning_rate": 0.00035475642587846305,
      "loss": 2.0983,
      "step": 101740
    },
    {
      "epoch": 1.32,
      "grad_norm": 1.0006062984466553,
      "learning_rate": 0.0003547524040249669,
      "loss": 2.2173,
      "step": 101741
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.9305576086044312,
      "learning_rate": 0.00035474838216129106,
      "loss": 2.1847,
      "step": 101742
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.9146662354469299,
      "learning_rate": 0.00035474436028743654,
      "loss": 2.0673,
      "step": 101743
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.9199634194374084,
      "learning_rate": 0.000354740338403404,
      "loss": 2.1608,
      "step": 101744
    },
    {
      "epoch": 1.32,
      "grad_norm": 1.0891945362091064,
      "learning_rate": 0.00035473631650919415,
      "loss": 2.3052,
      "step": 101745
    },
    {
      "epoch": 1.32,
      "grad_norm": 1.0971550941467285,
      "learning_rate": 0.00035473229460480785,
      "loss": 2.0725,
      "step": 101746
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.8100802302360535,
      "learning_rate": 0.0003547282726902457,
      "loss": 2.0844,
      "step": 101747
    },
    {
      "epoch": 1.32,
      "grad_norm": 1.036965250968933,
      "learning_rate": 0.0003547242507655087,
      "loss": 2.2582,
      "step": 101748
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.9127706289291382,
      "learning_rate": 0.00035472022883059726,
      "loss": 2.3719,
      "step": 101749
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.8450414538383484,
      "learning_rate": 0.0003547162068855123,
      "loss": 2.0348,
      "step": 101750
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.8664343357086182,
      "learning_rate": 0.0003547121849302547,
      "loss": 2.1088,
      "step": 101751
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.8943890929222107,
      "learning_rate": 0.000354708162964825,
      "loss": 2.3105,
      "step": 101752
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.8357481956481934,
      "learning_rate": 0.00035470414098922394,
      "loss": 1.9841,
      "step": 101753
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.9094321727752686,
      "learning_rate": 0.0003547001190034525,
      "loss": 2.2303,
      "step": 101754
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.8454054594039917,
      "learning_rate": 0.0003546960970075111,
      "loss": 2.1333,
      "step": 101755
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.808693528175354,
      "learning_rate": 0.0003546920750014008,
      "loss": 2.2819,
      "step": 101756
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.8263242244720459,
      "learning_rate": 0.0003546880529851221,
      "loss": 2.1525,
      "step": 101757
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.8557476997375488,
      "learning_rate": 0.00035468403095867597,
      "loss": 2.4399,
      "step": 101758
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.875546395778656,
      "learning_rate": 0.000354680008922063,
      "loss": 1.906,
      "step": 101759
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.9022508859634399,
      "learning_rate": 0.00035467598687528406,
      "loss": 2.1648,
      "step": 101760
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.891488254070282,
      "learning_rate": 0.00035467196481833973,
      "loss": 2.3479,
      "step": 101761
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.9473453760147095,
      "learning_rate": 0.0003546679427512308,
      "loss": 2.111,
      "step": 101762
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.8719332814216614,
      "learning_rate": 0.00035466392067395826,
      "loss": 2.1075,
      "step": 101763
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.88474041223526,
      "learning_rate": 0.00035465989858652247,
      "loss": 2.2793,
      "step": 101764
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.8349394202232361,
      "learning_rate": 0.0003546558764889244,
      "loss": 2.2047,
      "step": 101765
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.9192646145820618,
      "learning_rate": 0.00035465185438116493,
      "loss": 2.0311,
      "step": 101766
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.8336718678474426,
      "learning_rate": 0.00035464783226324447,
      "loss": 2.0867,
      "step": 101767
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.9282455444335938,
      "learning_rate": 0.000354643810135164,
      "loss": 1.9858,
      "step": 101768
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.8787310719490051,
      "learning_rate": 0.00035463978799692424,
      "loss": 2.3438,
      "step": 101769
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.8925874829292297,
      "learning_rate": 0.00035463576584852583,
      "loss": 2.0119,
      "step": 101770
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.8813178539276123,
      "learning_rate": 0.0003546317436899697,
      "loss": 2.3979,
      "step": 101771
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.8858360648155212,
      "learning_rate": 0.00035462772152125647,
      "loss": 2.271,
      "step": 101772
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.8813438415527344,
      "learning_rate": 0.0003546236993423869,
      "loss": 2.3719,
      "step": 101773
    },
    {
      "epoch": 1.33,
      "grad_norm": 1.0117906332015991,
      "learning_rate": 0.00035461967715336173,
      "loss": 2.2933,
      "step": 101774
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.9240934252738953,
      "learning_rate": 0.0003546156549541819,
      "loss": 2.0124,
      "step": 101775
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.911740779876709,
      "learning_rate": 0.00035461163274484774,
      "loss": 2.2233,
      "step": 101776
    },
    {
      "epoch": 1.33,
      "grad_norm": 1.0196458101272583,
      "learning_rate": 0.0003546076105253604,
      "loss": 2.0656,
      "step": 101777
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.8482131958007812,
      "learning_rate": 0.00035460358829572045,
      "loss": 2.0502,
      "step": 101778
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.8075911402702332,
      "learning_rate": 0.0003545995660559286,
      "loss": 2.0825,
      "step": 101779
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.7812655568122864,
      "learning_rate": 0.0003545955438059857,
      "loss": 2.1007,
      "step": 101780
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.8149378895759583,
      "learning_rate": 0.0003545915215458925,
      "loss": 2.2115,
      "step": 101781
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.8629953861236572,
      "learning_rate": 0.00035458749927564964,
      "loss": 2.2738,
      "step": 101782
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.9123886227607727,
      "learning_rate": 0.0003545834769952579,
      "loss": 2.1045,
      "step": 101783
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.8941959738731384,
      "learning_rate": 0.0003545794547047182,
      "loss": 2.1965,
      "step": 101784
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.9229388236999512,
      "learning_rate": 0.0003545754324040311,
      "loss": 1.9934,
      "step": 101785
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.8740630149841309,
      "learning_rate": 0.0003545714100931973,
      "loss": 2.192,
      "step": 101786
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.9038137793540955,
      "learning_rate": 0.0003545673877722178,
      "loss": 2.2402,
      "step": 101787
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.9204628467559814,
      "learning_rate": 0.0003545633654410931,
      "loss": 1.934,
      "step": 101788
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.9042059779167175,
      "learning_rate": 0.00035455934309982395,
      "loss": 2.143,
      "step": 101789
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.8238481879234314,
      "learning_rate": 0.0003545553207484113,
      "loss": 2.1353,
      "step": 101790
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.9390270709991455,
      "learning_rate": 0.00035455129838685585,
      "loss": 1.9586,
      "step": 101791
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.9758656620979309,
      "learning_rate": 0.0003545472760151582,
      "loss": 1.9592,
      "step": 101792
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.9808533787727356,
      "learning_rate": 0.0003545432536333191,
      "loss": 2.3182,
      "step": 101793
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.9975523352622986,
      "learning_rate": 0.00035453923124133957,
      "loss": 2.0591,
      "step": 101794
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.9112301468849182,
      "learning_rate": 0.00035453520883922007,
      "loss": 2.0568,
      "step": 101795
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.8506905436515808,
      "learning_rate": 0.0003545311864269614,
      "loss": 2.2173,
      "step": 101796
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.9986172318458557,
      "learning_rate": 0.0003545271640045644,
      "loss": 2.1138,
      "step": 101797
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.8548327088356018,
      "learning_rate": 0.00035452314157202974,
      "loss": 2.0189,
      "step": 101798
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.9570323824882507,
      "learning_rate": 0.0003545191191293582,
      "loss": 2.1452,
      "step": 101799
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.8000806570053101,
      "learning_rate": 0.0003545150966765506,
      "loss": 1.9028,
      "step": 101800
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.8140029907226562,
      "learning_rate": 0.00035451107421360754,
      "loss": 2.1143,
      "step": 101801
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.9584673643112183,
      "learning_rate": 0.0003545070517405299,
      "loss": 1.9861,
      "step": 101802
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.8803461194038391,
      "learning_rate": 0.00035450302925731835,
      "loss": 2.1621,
      "step": 101803
    },
    {
      "epoch": 1.33,
      "grad_norm": 1.0274332761764526,
      "learning_rate": 0.0003544990067639736,
      "loss": 2.5631,
      "step": 101804
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.9232206344604492,
      "learning_rate": 0.00035449498426049647,
      "loss": 2.0921,
      "step": 101805
    },
    {
      "epoch": 1.33,
      "grad_norm": 1.0025544166564941,
      "learning_rate": 0.00035449096174688775,
      "loss": 2.0289,
      "step": 101806
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.8195977807044983,
      "learning_rate": 0.00035448693922314814,
      "loss": 2.2974,
      "step": 101807
    },
    {
      "epoch": 1.33,
      "grad_norm": 1.0067253112792969,
      "learning_rate": 0.00035448291668927834,
      "loss": 2.5493,
      "step": 101808
    },
    {
      "epoch": 1.33,
      "grad_norm": 1.2052122354507446,
      "learning_rate": 0.00035447889414527917,
      "loss": 2.2153,
      "step": 101809
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.8344449996948242,
      "learning_rate": 0.00035447487159115127,
      "loss": 2.1896,
      "step": 101810
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.9146899580955505,
      "learning_rate": 0.0003544708490268955,
      "loss": 2.4244,
      "step": 101811
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.9140070080757141,
      "learning_rate": 0.00035446682645251266,
      "loss": 1.9462,
      "step": 101812
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.8070754408836365,
      "learning_rate": 0.0003544628038680033,
      "loss": 2.1479,
      "step": 101813
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.8689841032028198,
      "learning_rate": 0.0003544587812733682,
      "loss": 2.2275,
      "step": 101814
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.8528104424476624,
      "learning_rate": 0.0003544547586686084,
      "loss": 2.0522,
      "step": 101815
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.9485543966293335,
      "learning_rate": 0.0003544507360537243,
      "loss": 2.2664,
      "step": 101816
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.8980487585067749,
      "learning_rate": 0.0003544467134287168,
      "loss": 2.1173,
      "step": 101817
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.8347746729850769,
      "learning_rate": 0.00035444269079358667,
      "loss": 2.1365,
      "step": 101818
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.9919302463531494,
      "learning_rate": 0.00035443866814833456,
      "loss": 2.1586,
      "step": 101819
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.9509589076042175,
      "learning_rate": 0.00035443464549296126,
      "loss": 2.2745,
      "step": 101820
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.782460629940033,
      "learning_rate": 0.0003544306228274675,
      "loss": 2.2073,
      "step": 101821
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.8409086465835571,
      "learning_rate": 0.00035442660015185424,
      "loss": 2.2905,
      "step": 101822
    },
    {
      "epoch": 1.33,
      "grad_norm": 1.026354193687439,
      "learning_rate": 0.0003544225774661218,
      "loss": 2.0234,
      "step": 101823
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.8839911818504333,
      "learning_rate": 0.00035441855477027134,
      "loss": 1.9224,
      "step": 101824
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.8301615118980408,
      "learning_rate": 0.0003544145320643035,
      "loss": 2.305,
      "step": 101825
    },
    {
      "epoch": 1.33,
      "grad_norm": 1.2025151252746582,
      "learning_rate": 0.00035441050934821887,
      "loss": 2.2445,
      "step": 101826
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.9263905882835388,
      "learning_rate": 0.0003544064866220182,
      "loss": 2.1103,
      "step": 101827
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.9380027055740356,
      "learning_rate": 0.0003544024638857025,
      "loss": 2.129,
      "step": 101828
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.8547399640083313,
      "learning_rate": 0.0003543984411392723,
      "loss": 2.2725,
      "step": 101829
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.9547371864318848,
      "learning_rate": 0.0003543944183827284,
      "loss": 2.0781,
      "step": 101830
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.8361653685569763,
      "learning_rate": 0.00035439039561607153,
      "loss": 2.0217,
      "step": 101831
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.9238890409469604,
      "learning_rate": 0.0003543863728393025,
      "loss": 2.1549,
      "step": 101832
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.9264211654663086,
      "learning_rate": 0.00035438235005242204,
      "loss": 2.1959,
      "step": 101833
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.8602219820022583,
      "learning_rate": 0.0003543783272554308,
      "loss": 1.9633,
      "step": 101834
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.8185477256774902,
      "learning_rate": 0.00035437430444832963,
      "loss": 2.1816,
      "step": 101835
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.928444504737854,
      "learning_rate": 0.00035437028163111926,
      "loss": 1.888,
      "step": 101836
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.8137827515602112,
      "learning_rate": 0.0003543662588038005,
      "loss": 2.0619,
      "step": 101837
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.8080826997756958,
      "learning_rate": 0.0003543622359663739,
      "loss": 2.1605,
      "step": 101838
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.9414470791816711,
      "learning_rate": 0.00035435821311884043,
      "loss": 2.3917,
      "step": 101839
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.9376333355903625,
      "learning_rate": 0.00035435419026120065,
      "loss": 1.9604,
      "step": 101840
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.8743999600410461,
      "learning_rate": 0.0003543501673934554,
      "loss": 2.0737,
      "step": 101841
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.9856323599815369,
      "learning_rate": 0.00035434614451560555,
      "loss": 2.0807,
      "step": 101842
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.9156754016876221,
      "learning_rate": 0.0003543421216276517,
      "loss": 2.1544,
      "step": 101843
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.9082235097885132,
      "learning_rate": 0.00035433809872959454,
      "loss": 2.0533,
      "step": 101844
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.8626349568367004,
      "learning_rate": 0.00035433407582143495,
      "loss": 2.219,
      "step": 101845
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.8703412413597107,
      "learning_rate": 0.0003543300529031736,
      "loss": 2.4398,
      "step": 101846
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.9011076092720032,
      "learning_rate": 0.00035432602997481123,
      "loss": 2.152,
      "step": 101847
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.932185173034668,
      "learning_rate": 0.0003543220070363488,
      "loss": 2.1286,
      "step": 101848
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.8189056515693665,
      "learning_rate": 0.00035431798408778667,
      "loss": 2.0607,
      "step": 101849
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.8555753231048584,
      "learning_rate": 0.00035431396112912587,
      "loss": 2.1395,
      "step": 101850
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.8727672100067139,
      "learning_rate": 0.0003543099381603672,
      "loss": 2.3019,
      "step": 101851
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.9025281667709351,
      "learning_rate": 0.0003543059151815112,
      "loss": 2.155,
      "step": 101852
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.9789599180221558,
      "learning_rate": 0.0003543018921925586,
      "loss": 2.1946,
      "step": 101853
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.8751584887504578,
      "learning_rate": 0.00035429786919351046,
      "loss": 2.0271,
      "step": 101854
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.7995894551277161,
      "learning_rate": 0.0003542938461843672,
      "loss": 2.1308,
      "step": 101855
    },
    {
      "epoch": 1.33,
      "grad_norm": 1.0974781513214111,
      "learning_rate": 0.00035428982316512963,
      "loss": 2.267,
      "step": 101856
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.8796629905700684,
      "learning_rate": 0.00035428580013579863,
      "loss": 2.0576,
      "step": 101857
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.8399832844734192,
      "learning_rate": 0.0003542817770963749,
      "loss": 1.9695,
      "step": 101858
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.8869006037712097,
      "learning_rate": 0.0003542777540468592,
      "loss": 2.2444,
      "step": 101859
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.9858626127243042,
      "learning_rate": 0.0003542737309872521,
      "loss": 1.9215,
      "step": 101860
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.9153911471366882,
      "learning_rate": 0.00035426970791755465,
      "loss": 2.1883,
      "step": 101861
    },
    {
      "epoch": 1.33,
      "grad_norm": 1.0218008756637573,
      "learning_rate": 0.0003542656848377673,
      "loss": 2.1388,
      "step": 101862
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.9260887503623962,
      "learning_rate": 0.000354261661747891,
      "loss": 2.2827,
      "step": 101863
    },
    {
      "epoch": 1.33,
      "grad_norm": 1.0137536525726318,
      "learning_rate": 0.00035425763864792645,
      "loss": 2.1021,
      "step": 101864
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.8407866954803467,
      "learning_rate": 0.00035425361553787437,
      "loss": 2.2388,
      "step": 101865
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.9044962525367737,
      "learning_rate": 0.0003542495924177355,
      "loss": 2.0419,
      "step": 101866
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.8700737953186035,
      "learning_rate": 0.0003542455692875106,
      "loss": 2.1365,
      "step": 101867
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.8700549602508545,
      "learning_rate": 0.0003542415461472004,
      "loss": 2.1591,
      "step": 101868
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.8520527482032776,
      "learning_rate": 0.0003542375229968057,
      "loss": 2.2425,
      "step": 101869
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.8025491237640381,
      "learning_rate": 0.0003542334998363273,
      "loss": 2.1158,
      "step": 101870
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.8719367384910583,
      "learning_rate": 0.00035422947666576574,
      "loss": 2.1888,
      "step": 101871
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.8182231187820435,
      "learning_rate": 0.00035422545348512196,
      "loss": 2.0903,
      "step": 101872
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.8590489029884338,
      "learning_rate": 0.0003542214302943966,
      "loss": 2.2485,
      "step": 101873
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.864356279373169,
      "learning_rate": 0.00035421740709359054,
      "loss": 2.0943,
      "step": 101874
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.8998120427131653,
      "learning_rate": 0.00035421338388270436,
      "loss": 2.2755,
      "step": 101875
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.8621618747711182,
      "learning_rate": 0.00035420936066173894,
      "loss": 2.3244,
      "step": 101876
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.8620297908782959,
      "learning_rate": 0.00035420533743069495,
      "loss": 2.0451,
      "step": 101877
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.9227392673492432,
      "learning_rate": 0.0003542013141895732,
      "loss": 2.2412,
      "step": 101878
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.8566162586212158,
      "learning_rate": 0.00035419729093837435,
      "loss": 1.9919,
      "step": 101879
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.8531134128570557,
      "learning_rate": 0.0003541932676770992,
      "loss": 2.139,
      "step": 101880
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.9667989015579224,
      "learning_rate": 0.0003541892444057485,
      "loss": 2.2952,
      "step": 101881
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.9019176363945007,
      "learning_rate": 0.0003541852211243231,
      "loss": 2.2269,
      "step": 101882
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.9567548036575317,
      "learning_rate": 0.00035418119783282354,
      "loss": 2.1714,
      "step": 101883
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.8092859387397766,
      "learning_rate": 0.0003541771745312506,
      "loss": 2.1559,
      "step": 101884
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.8386353254318237,
      "learning_rate": 0.0003541731512196053,
      "loss": 2.2151,
      "step": 101885
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.856023371219635,
      "learning_rate": 0.00035416912789788807,
      "loss": 2.2516,
      "step": 101886
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.8119059205055237,
      "learning_rate": 0.00035416510456609974,
      "loss": 2.4132,
      "step": 101887
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.8016961812973022,
      "learning_rate": 0.0003541610812242412,
      "loss": 2.019,
      "step": 101888
    },
    {
      "epoch": 1.33,
      "grad_norm": 1.0332229137420654,
      "learning_rate": 0.0003541570578723131,
      "loss": 2.3879,
      "step": 101889
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.9100506901741028,
      "learning_rate": 0.000354153034510316,
      "loss": 2.2232,
      "step": 101890
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.8636853694915771,
      "learning_rate": 0.0003541490111382511,
      "loss": 2.0776,
      "step": 101891
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.8173783421516418,
      "learning_rate": 0.0003541449877561187,
      "loss": 2.123,
      "step": 101892
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.843832790851593,
      "learning_rate": 0.00035414096436391963,
      "loss": 2.1382,
      "step": 101893
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.8265442252159119,
      "learning_rate": 0.0003541369409616549,
      "loss": 2.164,
      "step": 101894
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.8585302829742432,
      "learning_rate": 0.00035413291754932504,
      "loss": 1.9456,
      "step": 101895
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.8349010348320007,
      "learning_rate": 0.0003541288941269308,
      "loss": 1.9873,
      "step": 101896
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.9508680701255798,
      "learning_rate": 0.0003541248706944731,
      "loss": 2.2179,
      "step": 101897
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.8988555073738098,
      "learning_rate": 0.0003541208472519525,
      "loss": 2.1043,
      "step": 101898
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.8254488706588745,
      "learning_rate": 0.0003541168237993698,
      "loss": 2.2277,
      "step": 101899
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.8743171095848083,
      "learning_rate": 0.0003541128003367258,
      "loss": 2.3744,
      "step": 101900
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.9030312895774841,
      "learning_rate": 0.0003541087768640211,
      "loss": 2.332,
      "step": 101901
    },
    {
      "epoch": 1.33,
      "grad_norm": 1.1048611402511597,
      "learning_rate": 0.00035410475338125663,
      "loss": 2.3002,
      "step": 101902
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.7860015034675598,
      "learning_rate": 0.0003541007298884331,
      "loss": 2.3367,
      "step": 101903
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.7963560223579407,
      "learning_rate": 0.00035409670638555116,
      "loss": 2.0777,
      "step": 101904
    },
    {
      "epoch": 1.33,
      "grad_norm": 1.1764296293258667,
      "learning_rate": 0.00035409268287261167,
      "loss": 2.3372,
      "step": 101905
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.9003210663795471,
      "learning_rate": 0.00035408865934961534,
      "loss": 2.0494,
      "step": 101906
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.8870478868484497,
      "learning_rate": 0.0003540846358165629,
      "loss": 2.2386,
      "step": 101907
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.8747329711914062,
      "learning_rate": 0.000354080612273455,
      "loss": 2.3305,
      "step": 101908
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.9900134801864624,
      "learning_rate": 0.00035407658872029256,
      "loss": 2.0141,
      "step": 101909
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.859341561794281,
      "learning_rate": 0.00035407256515707634,
      "loss": 2.0486,
      "step": 101910
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.8401476740837097,
      "learning_rate": 0.0003540685415838069,
      "loss": 2.072,
      "step": 101911
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.8227859139442444,
      "learning_rate": 0.00035406451800048513,
      "loss": 1.9191,
      "step": 101912
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.9627869129180908,
      "learning_rate": 0.0003540604944071118,
      "loss": 2.1732,
      "step": 101913
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.8155714273452759,
      "learning_rate": 0.00035405647080368754,
      "loss": 1.9619,
      "step": 101914
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.8774077296257019,
      "learning_rate": 0.00035405244719021324,
      "loss": 2.0882,
      "step": 101915
    },
    {
      "epoch": 1.33,
      "grad_norm": 1.044448971748352,
      "learning_rate": 0.00035404842356668946,
      "loss": 2.3303,
      "step": 101916
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.9650627374649048,
      "learning_rate": 0.0003540443999331172,
      "loss": 2.2641,
      "step": 101917
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.8882577419281006,
      "learning_rate": 0.0003540403762894969,
      "loss": 2.2203,
      "step": 101918
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.986769437789917,
      "learning_rate": 0.0003540363526358295,
      "loss": 2.0159,
      "step": 101919
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.8623687624931335,
      "learning_rate": 0.00035403232897211577,
      "loss": 1.9517,
      "step": 101920
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.8787113428115845,
      "learning_rate": 0.0003540283052983565,
      "loss": 2.1873,
      "step": 101921
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.8534218668937683,
      "learning_rate": 0.00035402428161455225,
      "loss": 2.3533,
      "step": 101922
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.8336189985275269,
      "learning_rate": 0.00035402025792070387,
      "loss": 2.4137,
      "step": 101923
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.8294584155082703,
      "learning_rate": 0.0003540162342168121,
      "loss": 1.9576,
      "step": 101924
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.8915808796882629,
      "learning_rate": 0.0003540122105028778,
      "loss": 2.1867,
      "step": 101925
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.7603245973587036,
      "learning_rate": 0.00035400818677890143,
      "loss": 1.9424,
      "step": 101926
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.9758329391479492,
      "learning_rate": 0.00035400416304488396,
      "loss": 2.0959,
      "step": 101927
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.9410552978515625,
      "learning_rate": 0.00035400013930082623,
      "loss": 2.2358,
      "step": 101928
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.8683072924613953,
      "learning_rate": 0.0003539961155467287,
      "loss": 2.1174,
      "step": 101929
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.876167356967926,
      "learning_rate": 0.00035399209178259234,
      "loss": 2.4413,
      "step": 101930
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.858691394329071,
      "learning_rate": 0.0003539880680084179,
      "loss": 2.1761,
      "step": 101931
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.8683724999427795,
      "learning_rate": 0.00035398404422420595,
      "loss": 2.0384,
      "step": 101932
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.8739365339279175,
      "learning_rate": 0.00035398002042995736,
      "loss": 2.0193,
      "step": 101933
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.8983083963394165,
      "learning_rate": 0.0003539759966256729,
      "loss": 2.4654,
      "step": 101934
    },
    {
      "epoch": 1.33,
      "grad_norm": 1.0657941102981567,
      "learning_rate": 0.0003539719728113533,
      "loss": 2.2531,
      "step": 101935
    },
    {
      "epoch": 1.33,
      "grad_norm": 1.0067354440689087,
      "learning_rate": 0.00035396794898699923,
      "loss": 2.0201,
      "step": 101936
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.938065767288208,
      "learning_rate": 0.00035396392515261164,
      "loss": 1.9583,
      "step": 101937
    },
    {
      "epoch": 1.33,
      "grad_norm": 1.133695125579834,
      "learning_rate": 0.00035395990130819094,
      "loss": 2.1684,
      "step": 101938
    },
    {
      "epoch": 1.33,
      "grad_norm": 1.1268092393875122,
      "learning_rate": 0.0003539558774537382,
      "loss": 2.0135,
      "step": 101939
    },
    {
      "epoch": 1.33,
      "grad_norm": 1.079284906387329,
      "learning_rate": 0.00035395185358925405,
      "loss": 2.2318,
      "step": 101940
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.8680458068847656,
      "learning_rate": 0.0003539478297147392,
      "loss": 2.1363,
      "step": 101941
    },
    {
      "epoch": 1.33,
      "grad_norm": 1.0521736145019531,
      "learning_rate": 0.00035394380583019436,
      "loss": 2.3955,
      "step": 101942
    },
    {
      "epoch": 1.33,
      "grad_norm": 1.1234363317489624,
      "learning_rate": 0.0003539397819356205,
      "loss": 2.2048,
      "step": 101943
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.8963669538497925,
      "learning_rate": 0.0003539357580310181,
      "loss": 2.1743,
      "step": 101944
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.8238986134529114,
      "learning_rate": 0.00035393173411638805,
      "loss": 2.2929,
      "step": 101945
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.9455544948577881,
      "learning_rate": 0.0003539277101917311,
      "loss": 2.1755,
      "step": 101946
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.9714503288269043,
      "learning_rate": 0.0003539236862570479,
      "loss": 2.3209,
      "step": 101947
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.9293712973594666,
      "learning_rate": 0.0003539196623123393,
      "loss": 2.167,
      "step": 101948
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.840868353843689,
      "learning_rate": 0.0003539156383576061,
      "loss": 2.2784,
      "step": 101949
    },
    {
      "epoch": 1.33,
      "grad_norm": 1.1427717208862305,
      "learning_rate": 0.0003539116143928489,
      "loss": 2.176,
      "step": 101950
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.991793692111969,
      "learning_rate": 0.0003539075904180684,
      "loss": 2.02,
      "step": 101951
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.8775174021720886,
      "learning_rate": 0.00035390356643326566,
      "loss": 2.1078,
      "step": 101952
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.92933189868927,
      "learning_rate": 0.00035389954243844114,
      "loss": 2.2206,
      "step": 101953
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.9833164811134338,
      "learning_rate": 0.00035389551843359565,
      "loss": 2.2717,
      "step": 101954
    },
    {
      "epoch": 1.33,
      "grad_norm": 1.1936132907867432,
      "learning_rate": 0.0003538914944187301,
      "loss": 2.0012,
      "step": 101955
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.8812965154647827,
      "learning_rate": 0.00035388747039384496,
      "loss": 2.153,
      "step": 101956
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.9717960357666016,
      "learning_rate": 0.0003538834463589411,
      "loss": 2.0534,
      "step": 101957
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.9222596287727356,
      "learning_rate": 0.00035387942231401945,
      "loss": 2.0342,
      "step": 101958
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.9820281863212585,
      "learning_rate": 0.00035387539825908047,
      "loss": 2.1297,
      "step": 101959
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.8994133472442627,
      "learning_rate": 0.00035387137419412497,
      "loss": 2.2687,
      "step": 101960
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.8947696685791016,
      "learning_rate": 0.00035386735011915394,
      "loss": 2.016,
      "step": 101961
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.9096561670303345,
      "learning_rate": 0.00035386332603416787,
      "loss": 2.2761,
      "step": 101962
    },
    {
      "epoch": 1.33,
      "grad_norm": 1.3757470846176147,
      "learning_rate": 0.00035385930193916756,
      "loss": 2.2791,
      "step": 101963
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.9297409653663635,
      "learning_rate": 0.0003538552778341539,
      "loss": 2.2205,
      "step": 101964
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.8897780179977417,
      "learning_rate": 0.00035385125371912744,
      "loss": 2.1618,
      "step": 101965
    },
    {
      "epoch": 1.33,
      "grad_norm": 1.1587289571762085,
      "learning_rate": 0.00035384722959408904,
      "loss": 2.1502,
      "step": 101966
    },
    {
      "epoch": 1.33,
      "grad_norm": 1.0394632816314697,
      "learning_rate": 0.00035384320545903944,
      "loss": 2.024,
      "step": 101967
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.9976487755775452,
      "learning_rate": 0.00035383918131397935,
      "loss": 2.4407,
      "step": 101968
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.8395757675170898,
      "learning_rate": 0.00035383515715890946,
      "loss": 2.2571,
      "step": 101969
    },
    {
      "epoch": 1.33,
      "grad_norm": 1.0275373458862305,
      "learning_rate": 0.0003538311329938307,
      "loss": 2.5319,
      "step": 101970
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.9756741523742676,
      "learning_rate": 0.0003538271088187437,
      "loss": 2.2851,
      "step": 101971
    },
    {
      "epoch": 1.33,
      "grad_norm": 1.005405306816101,
      "learning_rate": 0.0003538230846336491,
      "loss": 1.8576,
      "step": 101972
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.7958038449287415,
      "learning_rate": 0.000353819060438548,
      "loss": 2.1114,
      "step": 101973
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.9953424334526062,
      "learning_rate": 0.0003538150362334408,
      "loss": 2.1668,
      "step": 101974
    },
    {
      "epoch": 1.33,
      "grad_norm": 1.0628494024276733,
      "learning_rate": 0.0003538110120183283,
      "loss": 2.0724,
      "step": 101975
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.9484390020370483,
      "learning_rate": 0.00035380698779321145,
      "loss": 2.252,
      "step": 101976
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.8777814507484436,
      "learning_rate": 0.00035380296355809086,
      "loss": 2.2131,
      "step": 101977
    },
    {
      "epoch": 1.33,
      "grad_norm": 1.0488027334213257,
      "learning_rate": 0.00035379893931296715,
      "loss": 2.0226,
      "step": 101978
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.9369560480117798,
      "learning_rate": 0.00035379491505784134,
      "loss": 2.1697,
      "step": 101979
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.9635660648345947,
      "learning_rate": 0.000353790890792714,
      "loss": 2.2431,
      "step": 101980
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.8870922923088074,
      "learning_rate": 0.00035378686651758585,
      "loss": 2.3154,
      "step": 101981
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.9479213953018188,
      "learning_rate": 0.0003537828422324578,
      "loss": 2.0648,
      "step": 101982
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.9163810610771179,
      "learning_rate": 0.0003537788179373304,
      "loss": 2.3211,
      "step": 101983
    },
    {
      "epoch": 1.33,
      "grad_norm": 1.0588372945785522,
      "learning_rate": 0.00035377479363220456,
      "loss": 2.2878,
      "step": 101984
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.829190194606781,
      "learning_rate": 0.000353770769317081,
      "loss": 2.2048,
      "step": 101985
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.9517602920532227,
      "learning_rate": 0.0003537667449919604,
      "loss": 2.4432,
      "step": 101986
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.9044358134269714,
      "learning_rate": 0.0003537627206568436,
      "loss": 1.9464,
      "step": 101987
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.9962720274925232,
      "learning_rate": 0.0003537586963117312,
      "loss": 2.1304,
      "step": 101988
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.7722988724708557,
      "learning_rate": 0.00035375467195662415,
      "loss": 2.0023,
      "step": 101989
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.8939984440803528,
      "learning_rate": 0.000353750647591523,
      "loss": 1.9669,
      "step": 101990
    },
    {
      "epoch": 1.33,
      "grad_norm": 1.1487764120101929,
      "learning_rate": 0.0003537466232164286,
      "loss": 2.2892,
      "step": 101991
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.9779523015022278,
      "learning_rate": 0.00035374259883134176,
      "loss": 2.3582,
      "step": 101992
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.9055575728416443,
      "learning_rate": 0.00035373857443626313,
      "loss": 2.2603,
      "step": 101993
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.883979320526123,
      "learning_rate": 0.00035373455003119347,
      "loss": 2.0369,
      "step": 101994
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.8594631552696228,
      "learning_rate": 0.00035373052561613353,
      "loss": 2.0795,
      "step": 101995
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.8540389537811279,
      "learning_rate": 0.00035372650119108403,
      "loss": 2.0313,
      "step": 101996
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.8725116848945618,
      "learning_rate": 0.00035372247675604583,
      "loss": 2.0978,
      "step": 101997
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.7776719331741333,
      "learning_rate": 0.0003537184523110197,
      "loss": 2.2319,
      "step": 101998
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.8740249276161194,
      "learning_rate": 0.0003537144278560061,
      "loss": 2.1291,
      "step": 101999
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.8310279846191406,
      "learning_rate": 0.00035371040339100606,
      "loss": 1.9488,
      "step": 102000
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.8408117294311523,
      "learning_rate": 0.00035370637891602027,
      "loss": 2.1506,
      "step": 102001
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.8712387681007385,
      "learning_rate": 0.00035370235443104946,
      "loss": 2.0103,
      "step": 102002
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.8503139019012451,
      "learning_rate": 0.0003536983299360943,
      "loss": 2.1452,
      "step": 102003
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.8662859201431274,
      "learning_rate": 0.00035369430543115557,
      "loss": 2.2685,
      "step": 102004
    },
    {
      "epoch": 1.33,
      "grad_norm": 1.0578298568725586,
      "learning_rate": 0.0003536902809162341,
      "loss": 2.0423,
      "step": 102005
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.8054379820823669,
      "learning_rate": 0.00035368625639133067,
      "loss": 2.0809,
      "step": 102006
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.9388865232467651,
      "learning_rate": 0.0003536822318564459,
      "loss": 2.1261,
      "step": 102007
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.8393632769584656,
      "learning_rate": 0.00035367820731158057,
      "loss": 2.1332,
      "step": 102008
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.8711570501327515,
      "learning_rate": 0.0003536741827567355,
      "loss": 2.0429,
      "step": 102009
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.8442320823669434,
      "learning_rate": 0.0003536701581919113,
      "loss": 2.1746,
      "step": 102010
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.82670658826828,
      "learning_rate": 0.00035366613361710884,
      "loss": 1.9381,
      "step": 102011
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.8538627624511719,
      "learning_rate": 0.0003536621090323288,
      "loss": 2.019,
      "step": 102012
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.8551917672157288,
      "learning_rate": 0.00035365808443757214,
      "loss": 2.4244,
      "step": 102013
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.8554579615592957,
      "learning_rate": 0.0003536540598328392,
      "loss": 2.0764,
      "step": 102014
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.8942277431488037,
      "learning_rate": 0.00035365003521813104,
      "loss": 2.0456,
      "step": 102015
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.8722657561302185,
      "learning_rate": 0.0003536460105934484,
      "loss": 2.3113,
      "step": 102016
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.9233272671699524,
      "learning_rate": 0.00035364198595879185,
      "loss": 2.1139,
      "step": 102017
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.8914003968238831,
      "learning_rate": 0.0003536379613141623,
      "loss": 2.182,
      "step": 102018
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.9406610131263733,
      "learning_rate": 0.0003536339366595605,
      "loss": 2.3248,
      "step": 102019
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.7960619926452637,
      "learning_rate": 0.00035362991199498707,
      "loss": 2.2363,
      "step": 102020
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.9277374148368835,
      "learning_rate": 0.0003536258873204428,
      "loss": 2.0722,
      "step": 102021
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.8368510603904724,
      "learning_rate": 0.0003536218626359285,
      "loss": 2.051,
      "step": 102022
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.8236247301101685,
      "learning_rate": 0.0003536178379414449,
      "loss": 2.031,
      "step": 102023
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.8756991028785706,
      "learning_rate": 0.00035361381323699263,
      "loss": 2.242,
      "step": 102024
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.9085758924484253,
      "learning_rate": 0.0003536097885225727,
      "loss": 2.2713,
      "step": 102025
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.895265519618988,
      "learning_rate": 0.0003536057637981856,
      "loss": 2.1894,
      "step": 102026
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.961036205291748,
      "learning_rate": 0.00035360173906383215,
      "loss": 2.135,
      "step": 102027
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.8787271976470947,
      "learning_rate": 0.0003535977143195132,
      "loss": 2.2285,
      "step": 102028
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.8524850606918335,
      "learning_rate": 0.0003535936895652294,
      "loss": 2.3927,
      "step": 102029
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.8243494629859924,
      "learning_rate": 0.00035358966480098143,
      "loss": 2.1672,
      "step": 102030
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.8143397569656372,
      "learning_rate": 0.00035358564002677025,
      "loss": 2.1239,
      "step": 102031
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.811603844165802,
      "learning_rate": 0.00035358161524259645,
      "loss": 2.1119,
      "step": 102032
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.838933527469635,
      "learning_rate": 0.00035357759044846084,
      "loss": 2.1672,
      "step": 102033
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.8551326990127563,
      "learning_rate": 0.0003535735656443642,
      "loss": 2.297,
      "step": 102034
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.9987497925758362,
      "learning_rate": 0.00035356954083030705,
      "loss": 2.2682,
      "step": 102035
    },
    {
      "epoch": 1.33,
      "grad_norm": 1.0120205879211426,
      "learning_rate": 0.0003535655160062904,
      "loss": 2.1204,
      "step": 102036
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.9331040978431702,
      "learning_rate": 0.00035356149117231496,
      "loss": 2.1662,
      "step": 102037
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.8521360754966736,
      "learning_rate": 0.00035355746632838135,
      "loss": 2.0429,
      "step": 102038
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.9047674536705017,
      "learning_rate": 0.00035355344147449037,
      "loss": 2.2285,
      "step": 102039
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.8171906471252441,
      "learning_rate": 0.0003535494166106429,
      "loss": 2.0068,
      "step": 102040
    },
    {
      "epoch": 1.33,
      "grad_norm": 1.0093564987182617,
      "learning_rate": 0.00035354539173683956,
      "loss": 2.2183,
      "step": 102041
    },
    {
      "epoch": 1.33,
      "grad_norm": 2.1120333671569824,
      "learning_rate": 0.00035354136685308103,
      "loss": 2.197,
      "step": 102042
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.8174012303352356,
      "learning_rate": 0.0003535373419593683,
      "loss": 2.2433,
      "step": 102043
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.9088853001594543,
      "learning_rate": 0.00035353331705570185,
      "loss": 2.1561,
      "step": 102044
    },
    {
      "epoch": 1.33,
      "grad_norm": 1.1482473611831665,
      "learning_rate": 0.0003535292921420825,
      "loss": 2.1019,
      "step": 102045
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.8983883857727051,
      "learning_rate": 0.0003535252672185112,
      "loss": 2.3112,
      "step": 102046
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.9373814463615417,
      "learning_rate": 0.0003535212422849885,
      "loss": 2.2387,
      "step": 102047
    },
    {
      "epoch": 1.33,
      "grad_norm": 1.1139076948165894,
      "learning_rate": 0.0003535172173415151,
      "loss": 2.1382,
      "step": 102048
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.9647906422615051,
      "learning_rate": 0.00035351319238809196,
      "loss": 2.1755,
      "step": 102049
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.8387983441352844,
      "learning_rate": 0.00035350916742471955,
      "loss": 2.1205,
      "step": 102050
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.9441040754318237,
      "learning_rate": 0.00035350514245139893,
      "loss": 2.1974,
      "step": 102051
    },
    {
      "epoch": 1.33,
      "grad_norm": 1.1699166297912598,
      "learning_rate": 0.00035350111746813064,
      "loss": 2.2939,
      "step": 102052
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.9460716247558594,
      "learning_rate": 0.00035349709247491545,
      "loss": 1.9818,
      "step": 102053
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.8756694793701172,
      "learning_rate": 0.0003534930674717541,
      "loss": 2.1545,
      "step": 102054
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.8953455090522766,
      "learning_rate": 0.0003534890424586475,
      "loss": 2.1676,
      "step": 102055
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.9231140613555908,
      "learning_rate": 0.0003534850174355962,
      "loss": 2.065,
      "step": 102056
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.9339456558227539,
      "learning_rate": 0.00035348099240260105,
      "loss": 2.1398,
      "step": 102057
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.8508598804473877,
      "learning_rate": 0.0003534769673596628,
      "loss": 2.1821,
      "step": 102058
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.9259350299835205,
      "learning_rate": 0.0003534729423067821,
      "loss": 2.1496,
      "step": 102059
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.8450961112976074,
      "learning_rate": 0.00035346891724395983,
      "loss": 1.9711,
      "step": 102060
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.8588524460792542,
      "learning_rate": 0.0003534648921711966,
      "loss": 2.0092,
      "step": 102061
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.9723039865493774,
      "learning_rate": 0.00035346086708849334,
      "loss": 2.2582,
      "step": 102062
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.8324386477470398,
      "learning_rate": 0.0003534568419958507,
      "loss": 2.1805,
      "step": 102063
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.8624488115310669,
      "learning_rate": 0.0003534528168932693,
      "loss": 2.0988,
      "step": 102064
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.899213433265686,
      "learning_rate": 0.00035344879178075014,
      "loss": 2.2989,
      "step": 102065
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.9044967293739319,
      "learning_rate": 0.0003534447666582938,
      "loss": 2.1336,
      "step": 102066
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.8422393202781677,
      "learning_rate": 0.000353440741525901,
      "loss": 2.1974,
      "step": 102067
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.8563713431358337,
      "learning_rate": 0.00035343671638357274,
      "loss": 2.0765,
      "step": 102068
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.8946992754936218,
      "learning_rate": 0.00035343269123130943,
      "loss": 2.3282,
      "step": 102069
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.8380218148231506,
      "learning_rate": 0.00035342866606911195,
      "loss": 2.3006,
      "step": 102070
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.9858136177062988,
      "learning_rate": 0.0003534246408969812,
      "loss": 2.4248,
      "step": 102071
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.867842435836792,
      "learning_rate": 0.00035342061571491765,
      "loss": 2.2113,
      "step": 102072
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.7884306907653809,
      "learning_rate": 0.0003534165905229223,
      "loss": 2.0286,
      "step": 102073
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.9024397134780884,
      "learning_rate": 0.00035341256532099585,
      "loss": 2.2423,
      "step": 102074
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.8661966919898987,
      "learning_rate": 0.0003534085401091389,
      "loss": 2.084,
      "step": 102075
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.8004311919212341,
      "learning_rate": 0.00035340451488735224,
      "loss": 2.0426,
      "step": 102076
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.8958428502082825,
      "learning_rate": 0.0003534004896556368,
      "loss": 2.1866,
      "step": 102077
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.8731486797332764,
      "learning_rate": 0.00035339646441399317,
      "loss": 2.2918,
      "step": 102078
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.8854759335517883,
      "learning_rate": 0.00035339243916242204,
      "loss": 2.2265,
      "step": 102079
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.9393767714500427,
      "learning_rate": 0.0003533884139009244,
      "loss": 1.9524,
      "step": 102080
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.8334276676177979,
      "learning_rate": 0.00035338438862950075,
      "loss": 2.2612,
      "step": 102081
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.8847731947898865,
      "learning_rate": 0.0003533803633481519,
      "loss": 2.1894,
      "step": 102082
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.8819879293441772,
      "learning_rate": 0.00035337633805687874,
      "loss": 2.1475,
      "step": 102083
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.9508745670318604,
      "learning_rate": 0.00035337231275568183,
      "loss": 2.3482,
      "step": 102084
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.9138373732566833,
      "learning_rate": 0.00035336828744456203,
      "loss": 2.0351,
      "step": 102085
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.9101406931877136,
      "learning_rate": 0.00035336426212352,
      "loss": 2.2028,
      "step": 102086
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.8562107682228088,
      "learning_rate": 0.00035336023679255664,
      "loss": 2.3025,
      "step": 102087
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.7419806122779846,
      "learning_rate": 0.00035335621145167256,
      "loss": 2.038,
      "step": 102088
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.9381018877029419,
      "learning_rate": 0.00035335218610086847,
      "loss": 2.0689,
      "step": 102089
    },
    {
      "epoch": 1.33,
      "grad_norm": 1.0258082151412964,
      "learning_rate": 0.0003533481607401454,
      "loss": 2.3838,
      "step": 102090
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.826923668384552,
      "learning_rate": 0.0003533441353695037,
      "loss": 2.2791,
      "step": 102091
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.7741665840148926,
      "learning_rate": 0.00035334010998894444,
      "loss": 2.1689,
      "step": 102092
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.8609970211982727,
      "learning_rate": 0.00035333608459846814,
      "loss": 1.8479,
      "step": 102093
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.8464415073394775,
      "learning_rate": 0.0003533320591980758,
      "loss": 2.1684,
      "step": 102094
    },
    {
      "epoch": 1.33,
      "grad_norm": 2.37917423248291,
      "learning_rate": 0.0003533280337877679,
      "loss": 1.972,
      "step": 102095
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.8557173013687134,
      "learning_rate": 0.0003533240083675454,
      "loss": 2.3754,
      "step": 102096
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.8675032258033752,
      "learning_rate": 0.00035331998293740887,
      "loss": 2.053,
      "step": 102097
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.8938998579978943,
      "learning_rate": 0.0003533159574973593,
      "loss": 2.1343,
      "step": 102098
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.9366175532341003,
      "learning_rate": 0.0003533119320473971,
      "loss": 2.2005,
      "step": 102099
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.8933957815170288,
      "learning_rate": 0.00035330790658752334,
      "loss": 2.2561,
      "step": 102100
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.8500643372535706,
      "learning_rate": 0.0003533038811177386,
      "loss": 2.2743,
      "step": 102101
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.9023555517196655,
      "learning_rate": 0.00035329985563804355,
      "loss": 2.078,
      "step": 102102
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.8455548286437988,
      "learning_rate": 0.0003532958301484392,
      "loss": 2.2469,
      "step": 102103
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.9299938082695007,
      "learning_rate": 0.00035329180464892615,
      "loss": 1.9498,
      "step": 102104
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.9236696362495422,
      "learning_rate": 0.00035328777913950506,
      "loss": 2.249,
      "step": 102105
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.8385637998580933,
      "learning_rate": 0.00035328375362017676,
      "loss": 1.8802,
      "step": 102106
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.9385573267936707,
      "learning_rate": 0.0003532797280909421,
      "loss": 2.2823,
      "step": 102107
    },
    {
      "epoch": 1.33,
      "grad_norm": 1.419396996498108,
      "learning_rate": 0.0003532757025518017,
      "loss": 2.3426,
      "step": 102108
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.8753951191902161,
      "learning_rate": 0.0003532716770027563,
      "loss": 2.2444,
      "step": 102109
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.9005609750747681,
      "learning_rate": 0.0003532676514438067,
      "loss": 2.2124,
      "step": 102110
    },
    {
      "epoch": 1.33,
      "grad_norm": 1.3440696001052856,
      "learning_rate": 0.0003532636258749536,
      "loss": 2.2504,
      "step": 102111
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.9664481282234192,
      "learning_rate": 0.00035325960029619787,
      "loss": 2.1305,
      "step": 102112
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.7863603234291077,
      "learning_rate": 0.00035325557470754015,
      "loss": 2.2757,
      "step": 102113
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.9230297803878784,
      "learning_rate": 0.00035325154910898117,
      "loss": 2.2185,
      "step": 102114
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.8711452484130859,
      "learning_rate": 0.00035324752350052174,
      "loss": 2.269,
      "step": 102115
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.8760937452316284,
      "learning_rate": 0.0003532434978821627,
      "loss": 2.237,
      "step": 102116
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.8996148705482483,
      "learning_rate": 0.00035323947225390453,
      "loss": 2.1547,
      "step": 102117
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.9596124291419983,
      "learning_rate": 0.0003532354466157482,
      "loss": 2.0609,
      "step": 102118
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.8553422093391418,
      "learning_rate": 0.0003532314209676945,
      "loss": 2.2656,
      "step": 102119
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.8596550226211548,
      "learning_rate": 0.0003532273953097439,
      "loss": 2.3039,
      "step": 102120
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.8731824159622192,
      "learning_rate": 0.00035322336964189733,
      "loss": 2.2999,
      "step": 102121
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.8992897272109985,
      "learning_rate": 0.00035321934396415565,
      "loss": 2.3836,
      "step": 102122
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.8421475887298584,
      "learning_rate": 0.0003532153182765195,
      "loss": 2.0398,
      "step": 102123
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.8317567706108093,
      "learning_rate": 0.0003532112925789895,
      "loss": 2.2099,
      "step": 102124
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.9101839661598206,
      "learning_rate": 0.00035320726687156653,
      "loss": 2.0059,
      "step": 102125
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.8730121850967407,
      "learning_rate": 0.0003532032411542514,
      "loss": 2.1291,
      "step": 102126
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.9409615397453308,
      "learning_rate": 0.0003531992154270447,
      "loss": 1.9039,
      "step": 102127
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.8823844194412231,
      "learning_rate": 0.0003531951896899473,
      "loss": 2.0885,
      "step": 102128
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.8059414625167847,
      "learning_rate": 0.00035319116394296,
      "loss": 2.0836,
      "step": 102129
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.9271318316459656,
      "learning_rate": 0.0003531871381860833,
      "loss": 2.1257,
      "step": 102130
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.986452579498291,
      "learning_rate": 0.0003531831124193182,
      "loss": 2.1455,
      "step": 102131
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.8434915542602539,
      "learning_rate": 0.0003531790866426654,
      "loss": 1.9334,
      "step": 102132
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.9157059192657471,
      "learning_rate": 0.0003531750608561255,
      "loss": 2.1512,
      "step": 102133
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.8603766560554504,
      "learning_rate": 0.0003531710350596994,
      "loss": 2.246,
      "step": 102134
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.9122754335403442,
      "learning_rate": 0.0003531670092533878,
      "loss": 2.4129,
      "step": 102135
    },
    {
      "epoch": 1.33,
      "grad_norm": 1.0071948766708374,
      "learning_rate": 0.0003531629834371914,
      "loss": 2.1337,
      "step": 102136
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.8848347663879395,
      "learning_rate": 0.000353158957611111,
      "loss": 2.3004,
      "step": 102137
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.9043081402778625,
      "learning_rate": 0.0003531549317751475,
      "loss": 2.3218,
      "step": 102138
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.9599729180335999,
      "learning_rate": 0.00035315090592930135,
      "loss": 2.1147,
      "step": 102139
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.8799035549163818,
      "learning_rate": 0.00035314688007357346,
      "loss": 2.3331,
      "step": 102140
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.9460906386375427,
      "learning_rate": 0.00035314285420796464,
      "loss": 2.1636,
      "step": 102141
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.8279620409011841,
      "learning_rate": 0.00035313882833247544,
      "loss": 2.0763,
      "step": 102142
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.8877261877059937,
      "learning_rate": 0.00035313480244710677,
      "loss": 2.0954,
      "step": 102143
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.8162766098976135,
      "learning_rate": 0.0003531307765518594,
      "loss": 2.206,
      "step": 102144
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.8377014398574829,
      "learning_rate": 0.000353126750646734,
      "loss": 2.1116,
      "step": 102145
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.937149167060852,
      "learning_rate": 0.00035312272473173126,
      "loss": 2.0628,
      "step": 102146
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.8633756041526794,
      "learning_rate": 0.0003531186988068521,
      "loss": 2.2416,
      "step": 102147
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.8687186241149902,
      "learning_rate": 0.00035311467287209704,
      "loss": 2.1305,
      "step": 102148
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.8909791111946106,
      "learning_rate": 0.00035311064692746707,
      "loss": 2.1652,
      "step": 102149
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.8370943665504456,
      "learning_rate": 0.00035310662097296283,
      "loss": 2.4324,
      "step": 102150
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.8754417300224304,
      "learning_rate": 0.000353102595008585,
      "loss": 2.21,
      "step": 102151
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.9023578763008118,
      "learning_rate": 0.0003530985690343344,
      "loss": 2.1053,
      "step": 102152
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.8399730324745178,
      "learning_rate": 0.00035309454305021187,
      "loss": 2.1935,
      "step": 102153
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.845971405506134,
      "learning_rate": 0.00035309051705621796,
      "loss": 2.1746,
      "step": 102154
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.9400502443313599,
      "learning_rate": 0.0003530864910523535,
      "loss": 2.1124,
      "step": 102155
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.8918445706367493,
      "learning_rate": 0.00035308246503861943,
      "loss": 2.1584,
      "step": 102156
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.8400307297706604,
      "learning_rate": 0.0003530784390150162,
      "loss": 2.0257,
      "step": 102157
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.8609297871589661,
      "learning_rate": 0.0003530744129815446,
      "loss": 2.0151,
      "step": 102158
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.8752703070640564,
      "learning_rate": 0.0003530703869382057,
      "loss": 1.8814,
      "step": 102159
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.8957207202911377,
      "learning_rate": 0.00035306636088499986,
      "loss": 2.5147,
      "step": 102160
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.9390122890472412,
      "learning_rate": 0.0003530623348219279,
      "loss": 2.0895,
      "step": 102161
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.9220560789108276,
      "learning_rate": 0.0003530583087489908,
      "loss": 2.1911,
      "step": 102162
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.8875412940979004,
      "learning_rate": 0.0003530542826661891,
      "loss": 2.1125,
      "step": 102163
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.8832400441169739,
      "learning_rate": 0.0003530502565735236,
      "loss": 2.3396,
      "step": 102164
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.8682483434677124,
      "learning_rate": 0.00035304623047099513,
      "loss": 2.0956,
      "step": 102165
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.8132207989692688,
      "learning_rate": 0.0003530422043586043,
      "loss": 2.1677,
      "step": 102166
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.8597379922866821,
      "learning_rate": 0.0003530381782363519,
      "loss": 2.1773,
      "step": 102167
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.8384603261947632,
      "learning_rate": 0.00035303415210423877,
      "loss": 2.1264,
      "step": 102168
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.8720672130584717,
      "learning_rate": 0.0003530301259622655,
      "loss": 1.981,
      "step": 102169
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.8665257692337036,
      "learning_rate": 0.000353026099810433,
      "loss": 2.083,
      "step": 102170
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.9128865599632263,
      "learning_rate": 0.000353022073648742,
      "loss": 2.1907,
      "step": 102171
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.9932038187980652,
      "learning_rate": 0.00035301804747719305,
      "loss": 2.2188,
      "step": 102172
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.9493569135665894,
      "learning_rate": 0.0003530140212957871,
      "loss": 2.1618,
      "step": 102173
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.8873466849327087,
      "learning_rate": 0.0003530099951045248,
      "loss": 1.9545,
      "step": 102174
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.8992818593978882,
      "learning_rate": 0.0003530059689034071,
      "loss": 2.1013,
      "step": 102175
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.8216956257820129,
      "learning_rate": 0.00035300194269243445,
      "loss": 2.3217,
      "step": 102176
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.9105704426765442,
      "learning_rate": 0.0003529979164716078,
      "loss": 2.0411,
      "step": 102177
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.9834833145141602,
      "learning_rate": 0.0003529938902409277,
      "loss": 2.2844,
      "step": 102178
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.9977811574935913,
      "learning_rate": 0.0003529898640003952,
      "loss": 2.1928,
      "step": 102179
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.9517040848731995,
      "learning_rate": 0.00035298583775001085,
      "loss": 2.1307,
      "step": 102180
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.9652444124221802,
      "learning_rate": 0.0003529818114897754,
      "loss": 2.2977,
      "step": 102181
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.87360018491745,
      "learning_rate": 0.0003529777852196897,
      "loss": 2.0935,
      "step": 102182
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.8935019373893738,
      "learning_rate": 0.00035297375893975433,
      "loss": 2.0949,
      "step": 102183
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.795080840587616,
      "learning_rate": 0.0003529697326499702,
      "loss": 2.2064,
      "step": 102184
    },
    {
      "epoch": 1.33,
      "grad_norm": 1.0037044286727905,
      "learning_rate": 0.0003529657063503379,
      "loss": 2.1364,
      "step": 102185
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.985828161239624,
      "learning_rate": 0.0003529616800408584,
      "loss": 2.2963,
      "step": 102186
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.8791229724884033,
      "learning_rate": 0.0003529576537215323,
      "loss": 2.058,
      "step": 102187
    },
    {
      "epoch": 1.33,
      "grad_norm": 1.0182291269302368,
      "learning_rate": 0.0003529536273923604,
      "loss": 1.9781,
      "step": 102188
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.8502145409584045,
      "learning_rate": 0.00035294960105334335,
      "loss": 1.9774,
      "step": 102189
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.843140721321106,
      "learning_rate": 0.000352945574704482,
      "loss": 2.1033,
      "step": 102190
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.8726504445075989,
      "learning_rate": 0.00035294154834577705,
      "loss": 2.2806,
      "step": 102191
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.880642294883728,
      "learning_rate": 0.00035293752197722923,
      "loss": 2.2168,
      "step": 102192
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.961887776851654,
      "learning_rate": 0.00035293349559883946,
      "loss": 2.2938,
      "step": 102193
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.8607177734375,
      "learning_rate": 0.00035292946921060823,
      "loss": 2.3019,
      "step": 102194
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.8642792701721191,
      "learning_rate": 0.0003529254428125364,
      "loss": 1.984,
      "step": 102195
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.8481797575950623,
      "learning_rate": 0.0003529214164046249,
      "loss": 1.9924,
      "step": 102196
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.8738968372344971,
      "learning_rate": 0.00035291738998687413,
      "loss": 2.1885,
      "step": 102197
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.9489679932594299,
      "learning_rate": 0.00035291336355928505,
      "loss": 2.1823,
      "step": 102198
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.8534315824508667,
      "learning_rate": 0.0003529093371218584,
      "loss": 2.435,
      "step": 102199
    },
    {
      "epoch": 1.33,
      "grad_norm": 1.0415849685668945,
      "learning_rate": 0.000352905310674595,
      "loss": 2.1124,
      "step": 102200
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.9002100825309753,
      "learning_rate": 0.0003529012842174953,
      "loss": 2.2591,
      "step": 102201
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.9728562235832214,
      "learning_rate": 0.0003528972577505605,
      "loss": 2.1771,
      "step": 102202
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.9960105419158936,
      "learning_rate": 0.0003528932312737909,
      "loss": 2.1485,
      "step": 102203
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.8962040543556213,
      "learning_rate": 0.00035288920478718754,
      "loss": 2.2159,
      "step": 102204
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.9852985739707947,
      "learning_rate": 0.0003528851782907511,
      "loss": 2.0976,
      "step": 102205
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.8418355584144592,
      "learning_rate": 0.0003528811517844822,
      "loss": 2.2397,
      "step": 102206
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.9043181538581848,
      "learning_rate": 0.0003528771252683818,
      "loss": 2.1587,
      "step": 102207
    },
    {
      "epoch": 1.33,
      "grad_norm": 1.0761783123016357,
      "learning_rate": 0.0003528730987424505,
      "loss": 2.3333,
      "step": 102208
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.8898783922195435,
      "learning_rate": 0.0003528690722066891,
      "loss": 2.3807,
      "step": 102209
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.8325095176696777,
      "learning_rate": 0.00035286504566109833,
      "loss": 2.1952,
      "step": 102210
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.9758875370025635,
      "learning_rate": 0.000352861019105679,
      "loss": 2.1873,
      "step": 102211
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.9783862233161926,
      "learning_rate": 0.00035285699254043176,
      "loss": 2.2203,
      "step": 102212
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.9006587862968445,
      "learning_rate": 0.0003528529659653574,
      "loss": 1.9482,
      "step": 102213
    },
    {
      "epoch": 1.33,
      "grad_norm": 1.0016309022903442,
      "learning_rate": 0.0003528489393804567,
      "loss": 2.2195,
      "step": 102214
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.9985727071762085,
      "learning_rate": 0.00035284491278573035,
      "loss": 2.0767,
      "step": 102215
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.9306846261024475,
      "learning_rate": 0.0003528408861811792,
      "loss": 2.0404,
      "step": 102216
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.8977829217910767,
      "learning_rate": 0.00035283685956680395,
      "loss": 2.3511,
      "step": 102217
    },
    {
      "epoch": 1.33,
      "grad_norm": 1.678885817527771,
      "learning_rate": 0.00035283283294260516,
      "loss": 1.881,
      "step": 102218
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.9716612100601196,
      "learning_rate": 0.0003528288063085839,
      "loss": 2.4959,
      "step": 102219
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.8256523013114929,
      "learning_rate": 0.00035282477966474076,
      "loss": 2.1049,
      "step": 102220
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.8920108675956726,
      "learning_rate": 0.00035282075301107645,
      "loss": 2.3211,
      "step": 102221
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.8972562551498413,
      "learning_rate": 0.0003528167263475918,
      "loss": 2.2722,
      "step": 102222
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.9735680818557739,
      "learning_rate": 0.0003528126996742875,
      "loss": 2.4017,
      "step": 102223
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.8317853808403015,
      "learning_rate": 0.00035280867299116434,
      "loss": 2.1729,
      "step": 102224
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.98136967420578,
      "learning_rate": 0.00035280464629822297,
      "loss": 2.187,
      "step": 102225
    },
    {
      "epoch": 1.33,
      "grad_norm": 1.0119765996932983,
      "learning_rate": 0.00035280061959546436,
      "loss": 1.9739,
      "step": 102226
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.8232935667037964,
      "learning_rate": 0.00035279659288288905,
      "loss": 2.2085,
      "step": 102227
    },
    {
      "epoch": 1.33,
      "grad_norm": 1.0217821598052979,
      "learning_rate": 0.00035279256616049776,
      "loss": 2.1393,
      "step": 102228
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.8442388772964478,
      "learning_rate": 0.0003527885394282915,
      "loss": 2.0651,
      "step": 102229
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.8382906913757324,
      "learning_rate": 0.00035278451268627075,
      "loss": 2.1202,
      "step": 102230
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.8727065920829773,
      "learning_rate": 0.0003527804859344364,
      "loss": 2.114,
      "step": 102231
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.950496256351471,
      "learning_rate": 0.00035277645917278916,
      "loss": 2.1935,
      "step": 102232
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.8893601894378662,
      "learning_rate": 0.00035277243240132975,
      "loss": 2.2021,
      "step": 102233
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.8662327527999878,
      "learning_rate": 0.000352768405620059,
      "loss": 2.2575,
      "step": 102234
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.9081529378890991,
      "learning_rate": 0.0003527643788289776,
      "loss": 2.0229,
      "step": 102235
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.8806940317153931,
      "learning_rate": 0.0003527603520280863,
      "loss": 2.1487,
      "step": 102236
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.8511571288108826,
      "learning_rate": 0.0003527563252173858,
      "loss": 2.1544,
      "step": 102237
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.9507798552513123,
      "learning_rate": 0.00035275229839687705,
      "loss": 2.3256,
      "step": 102238
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.8567734360694885,
      "learning_rate": 0.00035274827156656053,
      "loss": 2.3507,
      "step": 102239
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.9067920446395874,
      "learning_rate": 0.00035274424472643703,
      "loss": 2.1948,
      "step": 102240
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.8478638529777527,
      "learning_rate": 0.0003527402178765075,
      "loss": 2.1108,
      "step": 102241
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.8997869491577148,
      "learning_rate": 0.0003527361910167726,
      "loss": 2.1477,
      "step": 102242
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.9527527689933777,
      "learning_rate": 0.000352732164147233,
      "loss": 1.8527,
      "step": 102243
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.8469969630241394,
      "learning_rate": 0.00035272813726788945,
      "loss": 1.9616,
      "step": 102244
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.8543707132339478,
      "learning_rate": 0.0003527241103787428,
      "loss": 2.2907,
      "step": 102245
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.9079710841178894,
      "learning_rate": 0.0003527200834797937,
      "loss": 2.4355,
      "step": 102246
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.8708628416061401,
      "learning_rate": 0.000352716056571043,
      "loss": 2.2671,
      "step": 102247
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.916312575340271,
      "learning_rate": 0.0003527120296524914,
      "loss": 2.0567,
      "step": 102248
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.9032613039016724,
      "learning_rate": 0.0003527080027241395,
      "loss": 2.1076,
      "step": 102249
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.9244236946105957,
      "learning_rate": 0.0003527039757859883,
      "loss": 2.3593,
      "step": 102250
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.8574557900428772,
      "learning_rate": 0.0003526999488380384,
      "loss": 2.1892,
      "step": 102251
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.9284526705741882,
      "learning_rate": 0.0003526959218802906,
      "loss": 2.3027,
      "step": 102252
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.9169368147850037,
      "learning_rate": 0.00035269189491274566,
      "loss": 2.0602,
      "step": 102253
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.8916515111923218,
      "learning_rate": 0.0003526878679354043,
      "loss": 2.1897,
      "step": 102254
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.8538107872009277,
      "learning_rate": 0.0003526838409482672,
      "loss": 1.9407,
      "step": 102255
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.8167568445205688,
      "learning_rate": 0.00035267981395133526,
      "loss": 1.9578,
      "step": 102256
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.8337283730506897,
      "learning_rate": 0.00035267578694460913,
      "loss": 1.9431,
      "step": 102257
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.8800259232521057,
      "learning_rate": 0.00035267175992808957,
      "loss": 1.9974,
      "step": 102258
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.9783957600593567,
      "learning_rate": 0.0003526677329017773,
      "loss": 2.0715,
      "step": 102259
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.7764402031898499,
      "learning_rate": 0.00035266370586567317,
      "loss": 2.0559,
      "step": 102260
    },
    {
      "epoch": 1.33,
      "grad_norm": 1.0067551136016846,
      "learning_rate": 0.0003526596788197778,
      "loss": 2.2182,
      "step": 102261
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.9689599275588989,
      "learning_rate": 0.00035265565176409203,
      "loss": 2.128,
      "step": 102262
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.9437766671180725,
      "learning_rate": 0.0003526516246986166,
      "loss": 2.2136,
      "step": 102263
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.8454376459121704,
      "learning_rate": 0.0003526475976233522,
      "loss": 1.8915,
      "step": 102264
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.9617363810539246,
      "learning_rate": 0.0003526435705382996,
      "loss": 2.2744,
      "step": 102265
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.8070040345191956,
      "learning_rate": 0.0003526395434434596,
      "loss": 2.0862,
      "step": 102266
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.9304336905479431,
      "learning_rate": 0.000352635516338833,
      "loss": 1.9578,
      "step": 102267
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.9396935701370239,
      "learning_rate": 0.00035263148922442026,
      "loss": 2.2408,
      "step": 102268
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.901206910610199,
      "learning_rate": 0.0003526274621002225,
      "loss": 2.1506,
      "step": 102269
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.8386086225509644,
      "learning_rate": 0.00035262343496624026,
      "loss": 2.0799,
      "step": 102270
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.8363543152809143,
      "learning_rate": 0.00035261940782247434,
      "loss": 2.0165,
      "step": 102271
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.8690324425697327,
      "learning_rate": 0.00035261538066892544,
      "loss": 2.3659,
      "step": 102272
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.8094519972801208,
      "learning_rate": 0.0003526113535055944,
      "loss": 2.1484,
      "step": 102273
    },
    {
      "epoch": 1.33,
      "grad_norm": 1.0669595003128052,
      "learning_rate": 0.0003526073263324819,
      "loss": 2.1832,
      "step": 102274
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.8295063972473145,
      "learning_rate": 0.0003526032991495886,
      "loss": 2.241,
      "step": 102275
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.9497986435890198,
      "learning_rate": 0.0003525992719569155,
      "loss": 2.2347,
      "step": 102276
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.9430768489837646,
      "learning_rate": 0.0003525952447544631,
      "loss": 2.0509,
      "step": 102277
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.9532936811447144,
      "learning_rate": 0.00035259121754223236,
      "loss": 2.1738,
      "step": 102278
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.8629323840141296,
      "learning_rate": 0.0003525871903202238,
      "loss": 2.2308,
      "step": 102279
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.9111002683639526,
      "learning_rate": 0.0003525831630884383,
      "loss": 2.3082,
      "step": 102280
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.8254532814025879,
      "learning_rate": 0.0003525791358468767,
      "loss": 1.987,
      "step": 102281
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.877292275428772,
      "learning_rate": 0.0003525751085955396,
      "loss": 2.2184,
      "step": 102282
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.9323709607124329,
      "learning_rate": 0.0003525710813344278,
      "loss": 2.0866,
      "step": 102283
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.9487209320068359,
      "learning_rate": 0.0003525670540635421,
      "loss": 2.5158,
      "step": 102284
    },
    {
      "epoch": 1.33,
      "grad_norm": 1.014231562614441,
      "learning_rate": 0.00035256302678288306,
      "loss": 2.1452,
      "step": 102285
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.9170213341712952,
      "learning_rate": 0.00035255899949245163,
      "loss": 1.9872,
      "step": 102286
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.8917849063873291,
      "learning_rate": 0.0003525549721922485,
      "loss": 2.2486,
      "step": 102287
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.7712129354476929,
      "learning_rate": 0.0003525509448822744,
      "loss": 2.2729,
      "step": 102288
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.8552981019020081,
      "learning_rate": 0.00035254691756253005,
      "loss": 2.0912,
      "step": 102289
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.9934305548667908,
      "learning_rate": 0.0003525428902330163,
      "loss": 2.0973,
      "step": 102290
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.8667909502983093,
      "learning_rate": 0.00035253886289373374,
      "loss": 2.0103,
      "step": 102291
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.9280675649642944,
      "learning_rate": 0.0003525348355446833,
      "loss": 2.1653,
      "step": 102292
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.9218590259552002,
      "learning_rate": 0.00035253080818586566,
      "loss": 2.0368,
      "step": 102293
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.8334301114082336,
      "learning_rate": 0.00035252678081728147,
      "loss": 2.155,
      "step": 102294
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.8521060347557068,
      "learning_rate": 0.00035252275343893155,
      "loss": 2.1679,
      "step": 102295
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.8186811208724976,
      "learning_rate": 0.00035251872605081676,
      "loss": 2.0289,
      "step": 102296
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.8452374339103699,
      "learning_rate": 0.0003525146986529377,
      "loss": 2.1032,
      "step": 102297
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.9065632820129395,
      "learning_rate": 0.00035251067124529513,
      "loss": 2.1518,
      "step": 102298
    },
    {
      "epoch": 1.33,
      "grad_norm": 1.0166807174682617,
      "learning_rate": 0.0003525066438278899,
      "loss": 2.3356,
      "step": 102299
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.844245433807373,
      "learning_rate": 0.00035250261640072265,
      "loss": 2.2318,
      "step": 102300
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.9027232527732849,
      "learning_rate": 0.00035249858896379416,
      "loss": 2.0921,
      "step": 102301
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.972265362739563,
      "learning_rate": 0.00035249456151710524,
      "loss": 2.1586,
      "step": 102302
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.8937918543815613,
      "learning_rate": 0.00035249053406065655,
      "loss": 2.02,
      "step": 102303
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.8733685612678528,
      "learning_rate": 0.0003524865065944489,
      "loss": 2.0951,
      "step": 102304
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.8592061996459961,
      "learning_rate": 0.000352482479118483,
      "loss": 2.1451,
      "step": 102305
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.9303984045982361,
      "learning_rate": 0.00035247845163275966,
      "loss": 2.5248,
      "step": 102306
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.90523761510849,
      "learning_rate": 0.0003524744241372795,
      "loss": 2.1807,
      "step": 102307
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.8083969354629517,
      "learning_rate": 0.00035247039663204347,
      "loss": 2.0998,
      "step": 102308
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.7754637002944946,
      "learning_rate": 0.0003524663691170521,
      "loss": 2.2386,
      "step": 102309
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.9356725215911865,
      "learning_rate": 0.00035246234159230625,
      "loss": 2.4529,
      "step": 102310
    },
    {
      "epoch": 1.33,
      "grad_norm": 1.126592755317688,
      "learning_rate": 0.0003524583140578067,
      "loss": 2.2028,
      "step": 102311
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.895523726940155,
      "learning_rate": 0.0003524542865135542,
      "loss": 2.2158,
      "step": 102312
    },
    {
      "epoch": 1.33,
      "grad_norm": 1.0374306440353394,
      "learning_rate": 0.0003524502589595494,
      "loss": 2.1204,
      "step": 102313
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.9571374654769897,
      "learning_rate": 0.00035244623139579317,
      "loss": 2.1764,
      "step": 102314
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.95223468542099,
      "learning_rate": 0.00035244220382228605,
      "loss": 2.2018,
      "step": 102315
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.9322738647460938,
      "learning_rate": 0.000352438176239029,
      "loss": 2.1521,
      "step": 102316
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.9987814426422119,
      "learning_rate": 0.0003524341486460228,
      "loss": 1.9191,
      "step": 102317
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.9596610069274902,
      "learning_rate": 0.000352430121043268,
      "loss": 2.2297,
      "step": 102318
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.8881986737251282,
      "learning_rate": 0.0003524260934307655,
      "loss": 2.3584,
      "step": 102319
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.8865122199058533,
      "learning_rate": 0.000352422065808516,
      "loss": 2.0426,
      "step": 102320
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.9441144466400146,
      "learning_rate": 0.00035241803817652017,
      "loss": 2.1712,
      "step": 102321
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.9473505020141602,
      "learning_rate": 0.00035241401053477895,
      "loss": 2.3618,
      "step": 102322
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.9288435578346252,
      "learning_rate": 0.00035240998288329296,
      "loss": 2.2358,
      "step": 102323
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.8731145858764648,
      "learning_rate": 0.0003524059552220629,
      "loss": 2.2034,
      "step": 102324
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.8367503881454468,
      "learning_rate": 0.00035240192755108955,
      "loss": 2.2679,
      "step": 102325
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.9413453936576843,
      "learning_rate": 0.0003523978998703737,
      "loss": 1.9982,
      "step": 102326
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.9078735113143921,
      "learning_rate": 0.00035239387217991623,
      "loss": 2.1195,
      "step": 102327
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.8910889029502869,
      "learning_rate": 0.0003523898444797176,
      "loss": 1.9987,
      "step": 102328
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.8528926372528076,
      "learning_rate": 0.0003523858167697788,
      "loss": 1.9668,
      "step": 102329
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.9256451725959778,
      "learning_rate": 0.0003523817890501005,
      "loss": 2.2262,
      "step": 102330
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.8895583152770996,
      "learning_rate": 0.0003523777613206833,
      "loss": 2.3338,
      "step": 102331
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.9263929724693298,
      "learning_rate": 0.0003523737335815282,
      "loss": 1.9702,
      "step": 102332
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.9291607737541199,
      "learning_rate": 0.00035236970583263594,
      "loss": 2.0352,
      "step": 102333
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.8453540205955505,
      "learning_rate": 0.000352365678074007,
      "loss": 2.1862,
      "step": 102334
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.8391184210777283,
      "learning_rate": 0.00035236165030564227,
      "loss": 2.1516,
      "step": 102335
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.9321960210800171,
      "learning_rate": 0.0003523576225275426,
      "loss": 2.0824,
      "step": 102336
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.8853241205215454,
      "learning_rate": 0.0003523535947397087,
      "loss": 2.3542,
      "step": 102337
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.8724544644355774,
      "learning_rate": 0.00035234956694214114,
      "loss": 1.9425,
      "step": 102338
    },
    {
      "epoch": 1.33,
      "grad_norm": 1.0361827611923218,
      "learning_rate": 0.00035234553913484094,
      "loss": 2.3332,
      "step": 102339
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.9301707148551941,
      "learning_rate": 0.0003523415113178086,
      "loss": 1.9675,
      "step": 102340
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.8145341873168945,
      "learning_rate": 0.0003523374834910451,
      "loss": 2.2172,
      "step": 102341
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.855766773223877,
      "learning_rate": 0.00035233345565455107,
      "loss": 2.1193,
      "step": 102342
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.9458364844322205,
      "learning_rate": 0.00035232942780832714,
      "loss": 2.0657,
      "step": 102343
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.8592237234115601,
      "learning_rate": 0.00035232539995237424,
      "loss": 2.0455,
      "step": 102344
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.9018085598945618,
      "learning_rate": 0.0003523213720866931,
      "loss": 2.075,
      "step": 102345
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.8702539205551147,
      "learning_rate": 0.00035231734421128447,
      "loss": 1.9742,
      "step": 102346
    },
    {
      "epoch": 1.33,
      "grad_norm": 1.0525684356689453,
      "learning_rate": 0.00035231331632614896,
      "loss": 2.2691,
      "step": 102347
    },
    {
      "epoch": 1.33,
      "grad_norm": 1.0282946825027466,
      "learning_rate": 0.0003523092884312875,
      "loss": 2.0731,
      "step": 102348
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.7891978621482849,
      "learning_rate": 0.0003523052605267007,
      "loss": 2.0242,
      "step": 102349
    },
    {
      "epoch": 1.33,
      "grad_norm": 1.0276871919631958,
      "learning_rate": 0.0003523012326123894,
      "loss": 2.0319,
      "step": 102350
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.8790152072906494,
      "learning_rate": 0.00035229720468835427,
      "loss": 2.1978,
      "step": 102351
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.9765366911888123,
      "learning_rate": 0.00035229317675459614,
      "loss": 2.255,
      "step": 102352
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.9452334642410278,
      "learning_rate": 0.00035228914881111565,
      "loss": 2.158,
      "step": 102353
    },
    {
      "epoch": 1.33,
      "grad_norm": 1.1416049003601074,
      "learning_rate": 0.00035228512085791366,
      "loss": 2.2305,
      "step": 102354
    },
    {
      "epoch": 1.33,
      "grad_norm": 1.021318793296814,
      "learning_rate": 0.00035228109289499093,
      "loss": 2.2362,
      "step": 102355
    },
    {
      "epoch": 1.33,
      "grad_norm": 1.0554158687591553,
      "learning_rate": 0.00035227706492234806,
      "loss": 2.2299,
      "step": 102356
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.922002375125885,
      "learning_rate": 0.00035227303693998596,
      "loss": 2.4396,
      "step": 102357
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.9735899567604065,
      "learning_rate": 0.0003522690089479053,
      "loss": 2.2119,
      "step": 102358
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.8449839949607849,
      "learning_rate": 0.00035226498094610687,
      "loss": 2.0326,
      "step": 102359
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.9815467000007629,
      "learning_rate": 0.0003522609529345914,
      "loss": 2.1327,
      "step": 102360
    },
    {
      "epoch": 1.33,
      "grad_norm": 1.055991768836975,
      "learning_rate": 0.00035225692491335956,
      "loss": 2.1023,
      "step": 102361
    },
    {
      "epoch": 1.33,
      "grad_norm": 1.0190857648849487,
      "learning_rate": 0.00035225289688241225,
      "loss": 1.979,
      "step": 102362
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.7974834442138672,
      "learning_rate": 0.00035224886884175006,
      "loss": 2.0464,
      "step": 102363
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.8132507801055908,
      "learning_rate": 0.0003522448407913739,
      "loss": 2.1079,
      "step": 102364
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.9718533158302307,
      "learning_rate": 0.0003522408127312844,
      "loss": 2.0698,
      "step": 102365
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.9409072995185852,
      "learning_rate": 0.00035223678466148235,
      "loss": 2.1239,
      "step": 102366
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.997336208820343,
      "learning_rate": 0.0003522327565819685,
      "loss": 2.1875,
      "step": 102367
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.8552548289299011,
      "learning_rate": 0.0003522287284927436,
      "loss": 2.1816,
      "step": 102368
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.9021852612495422,
      "learning_rate": 0.0003522247003938084,
      "loss": 2.0765,
      "step": 102369
    },
    {
      "epoch": 1.33,
      "grad_norm": 1.013870120048523,
      "learning_rate": 0.0003522206722851636,
      "loss": 2.1712,
      "step": 102370
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.867118775844574,
      "learning_rate": 0.00035221664416681004,
      "loss": 2.3522,
      "step": 102371
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.9047839045524597,
      "learning_rate": 0.00035221261603874846,
      "loss": 2.0337,
      "step": 102372
    },
    {
      "epoch": 1.33,
      "grad_norm": 1.0687702894210815,
      "learning_rate": 0.0003522085879009795,
      "loss": 2.2419,
      "step": 102373
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.8867177963256836,
      "learning_rate": 0.0003522045597535039,
      "loss": 2.0433,
      "step": 102374
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.9818161725997925,
      "learning_rate": 0.0003522005315963226,
      "loss": 2.2265,
      "step": 102375
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.8630509376525879,
      "learning_rate": 0.00035219650342943625,
      "loss": 2.3738,
      "step": 102376
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.958958625793457,
      "learning_rate": 0.0003521924752528455,
      "loss": 2.1379,
      "step": 102377
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.878144383430481,
      "learning_rate": 0.00035218844706655134,
      "loss": 2.2277,
      "step": 102378
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.9027686715126038,
      "learning_rate": 0.0003521844188705542,
      "loss": 2.2496,
      "step": 102379
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.9222776889801025,
      "learning_rate": 0.000352180390664855,
      "loss": 2.1125,
      "step": 102380
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.900815486907959,
      "learning_rate": 0.0003521763624494546,
      "loss": 2.2519,
      "step": 102381
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.8274499773979187,
      "learning_rate": 0.0003521723342243536,
      "loss": 2.2214,
      "step": 102382
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.8397937417030334,
      "learning_rate": 0.00035216830598955263,
      "loss": 2.1559,
      "step": 102383
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.86881422996521,
      "learning_rate": 0.00035216427774505273,
      "loss": 2.2295,
      "step": 102384
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.8842880129814148,
      "learning_rate": 0.0003521602494908545,
      "loss": 2.0113,
      "step": 102385
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.8231884241104126,
      "learning_rate": 0.00035215622122695863,
      "loss": 2.0696,
      "step": 102386
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.9586843252182007,
      "learning_rate": 0.000352152192953366,
      "loss": 2.2652,
      "step": 102387
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.790976345539093,
      "learning_rate": 0.00035214816467007726,
      "loss": 2.0861,
      "step": 102388
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.9449216723442078,
      "learning_rate": 0.00035214413637709314,
      "loss": 2.4667,
      "step": 102389
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.9307122230529785,
      "learning_rate": 0.00035214010807441457,
      "loss": 2.2769,
      "step": 102390
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.8268819451332092,
      "learning_rate": 0.00035213607976204204,
      "loss": 2.1517,
      "step": 102391
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.7929515242576599,
      "learning_rate": 0.0003521320514399765,
      "loss": 1.964,
      "step": 102392
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.860380232334137,
      "learning_rate": 0.00035212802310821857,
      "loss": 2.2314,
      "step": 102393
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.7975110411643982,
      "learning_rate": 0.00035212399476676915,
      "loss": 1.948,
      "step": 102394
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.847930908203125,
      "learning_rate": 0.0003521199664156288,
      "loss": 2.3618,
      "step": 102395
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.9343814849853516,
      "learning_rate": 0.00035211593805479835,
      "loss": 2.0152,
      "step": 102396
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.9862581491470337,
      "learning_rate": 0.00035211190968427865,
      "loss": 2.2913,
      "step": 102397
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.9042206406593323,
      "learning_rate": 0.00035210788130407035,
      "loss": 2.2634,
      "step": 102398
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.8847629427909851,
      "learning_rate": 0.0003521038529141741,
      "loss": 2.2907,
      "step": 102399
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.9402316212654114,
      "learning_rate": 0.00035209982451459097,
      "loss": 1.9659,
      "step": 102400
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.8957716226577759,
      "learning_rate": 0.00035209579610532134,
      "loss": 2.2468,
      "step": 102401
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.8838050961494446,
      "learning_rate": 0.0003520917676863661,
      "loss": 2.1749,
      "step": 102402
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.913327693939209,
      "learning_rate": 0.00035208773925772617,
      "loss": 2.0212,
      "step": 102403
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.8318420052528381,
      "learning_rate": 0.000352083710819402,
      "loss": 2.1702,
      "step": 102404
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.9129395484924316,
      "learning_rate": 0.00035207968237139457,
      "loss": 2.0589,
      "step": 102405
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.8406118154525757,
      "learning_rate": 0.0003520756539137045,
      "loss": 2.0529,
      "step": 102406
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.851208508014679,
      "learning_rate": 0.00035207162544633255,
      "loss": 2.1156,
      "step": 102407
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.851170003414154,
      "learning_rate": 0.0003520675969692795,
      "loss": 2.1802,
      "step": 102408
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.8611300587654114,
      "learning_rate": 0.0003520635684825462,
      "loss": 2.102,
      "step": 102409
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.7637587189674377,
      "learning_rate": 0.0003520595399861333,
      "loss": 2.0694,
      "step": 102410
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.9973673820495605,
      "learning_rate": 0.00035205551148004147,
      "loss": 2.1871,
      "step": 102411
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.923327624797821,
      "learning_rate": 0.00035205148296427164,
      "loss": 2.0248,
      "step": 102412
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.8937641978263855,
      "learning_rate": 0.0003520474544388243,
      "loss": 2.1283,
      "step": 102413
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.8822699189186096,
      "learning_rate": 0.0003520434259037004,
      "loss": 2.2185,
      "step": 102414
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.872188150882721,
      "learning_rate": 0.00035203939735890073,
      "loss": 2.2032,
      "step": 102415
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.8239359259605408,
      "learning_rate": 0.00035203536880442585,
      "loss": 1.988,
      "step": 102416
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.8465452194213867,
      "learning_rate": 0.00035203134024027664,
      "loss": 2.0894,
      "step": 102417
    },
    {
      "epoch": 1.33,
      "grad_norm": 1.0617433786392212,
      "learning_rate": 0.0003520273116664539,
      "loss": 2.0708,
      "step": 102418
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.8754098415374756,
      "learning_rate": 0.00035202328308295825,
      "loss": 2.1147,
      "step": 102419
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.9178510308265686,
      "learning_rate": 0.0003520192544897905,
      "loss": 2.2432,
      "step": 102420
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.8618130087852478,
      "learning_rate": 0.0003520152258869514,
      "loss": 2.078,
      "step": 102421
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.917392373085022,
      "learning_rate": 0.00035201119727444163,
      "loss": 2.0795,
      "step": 102422
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.8701136112213135,
      "learning_rate": 0.000352007168652262,
      "loss": 2.2686,
      "step": 102423
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.8988713026046753,
      "learning_rate": 0.0003520031400204134,
      "loss": 2.099,
      "step": 102424
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.956080436706543,
      "learning_rate": 0.00035199911137889626,
      "loss": 2.1415,
      "step": 102425
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.8174378275871277,
      "learning_rate": 0.00035199508272771153,
      "loss": 2.2075,
      "step": 102426
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.842929482460022,
      "learning_rate": 0.00035199105406686005,
      "loss": 2.1931,
      "step": 102427
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.8960585594177246,
      "learning_rate": 0.0003519870253963423,
      "loss": 2.3127,
      "step": 102428
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.9538388252258301,
      "learning_rate": 0.00035198299671615924,
      "loss": 2.1845,
      "step": 102429
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.9579855799674988,
      "learning_rate": 0.0003519789680263116,
      "loss": 2.4099,
      "step": 102430
    },
    {
      "epoch": 1.33,
      "grad_norm": 1.0012965202331543,
      "learning_rate": 0.00035197493932680006,
      "loss": 2.1086,
      "step": 102431
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.8264316320419312,
      "learning_rate": 0.00035197091061762537,
      "loss": 2.034,
      "step": 102432
    },
    {
      "epoch": 1.33,
      "grad_norm": 1.1604524850845337,
      "learning_rate": 0.0003519668818987884,
      "loss": 1.9898,
      "step": 102433
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.9404374361038208,
      "learning_rate": 0.0003519628531702897,
      "loss": 2.1763,
      "step": 102434
    },
    {
      "epoch": 1.33,
      "grad_norm": 1.1274495124816895,
      "learning_rate": 0.00035195882443213013,
      "loss": 2.0441,
      "step": 102435
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.8976953625679016,
      "learning_rate": 0.0003519547956843105,
      "loss": 2.2287,
      "step": 102436
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.896075963973999,
      "learning_rate": 0.00035195076692683147,
      "loss": 2.2024,
      "step": 102437
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.9026116132736206,
      "learning_rate": 0.0003519467381596938,
      "loss": 2.269,
      "step": 102438
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.9561416506767273,
      "learning_rate": 0.00035194270938289823,
      "loss": 2.479,
      "step": 102439
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.8617717027664185,
      "learning_rate": 0.00035193868059644557,
      "loss": 2.2814,
      "step": 102440
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.803067147731781,
      "learning_rate": 0.0003519346518003365,
      "loss": 2.2612,
      "step": 102441
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.9318975210189819,
      "learning_rate": 0.00035193062299457176,
      "loss": 2.1918,
      "step": 102442
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.920540988445282,
      "learning_rate": 0.00035192659417915224,
      "loss": 2.2332,
      "step": 102443
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.7999686002731323,
      "learning_rate": 0.00035192256535407856,
      "loss": 1.9855,
      "step": 102444
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.8995078802108765,
      "learning_rate": 0.00035191853651935135,
      "loss": 2.1752,
      "step": 102445
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.8665199279785156,
      "learning_rate": 0.00035191450767497166,
      "loss": 2.0756,
      "step": 102446
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.8887113928794861,
      "learning_rate": 0.0003519104788209401,
      "loss": 2.3123,
      "step": 102447
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.9734994173049927,
      "learning_rate": 0.00035190644995725735,
      "loss": 2.0954,
      "step": 102448
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.9311331510543823,
      "learning_rate": 0.00035190242108392415,
      "loss": 2.2665,
      "step": 102449
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.9230351448059082,
      "learning_rate": 0.0003518983922009414,
      "loss": 2.2355,
      "step": 102450
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.8532891869544983,
      "learning_rate": 0.0003518943633083097,
      "loss": 2.1595,
      "step": 102451
    },
    {
      "epoch": 1.33,
      "grad_norm": 1.047396183013916,
      "learning_rate": 0.0003518903344060298,
      "loss": 2.0395,
      "step": 102452
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.9138665795326233,
      "learning_rate": 0.00035188630549410266,
      "loss": 2.1397,
      "step": 102453
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.9445183873176575,
      "learning_rate": 0.0003518822765725288,
      "loss": 2.0765,
      "step": 102454
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.843071460723877,
      "learning_rate": 0.0003518782476413091,
      "loss": 2.2061,
      "step": 102455
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.9130065441131592,
      "learning_rate": 0.00035187421870044416,
      "loss": 2.3466,
      "step": 102456
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.9809563159942627,
      "learning_rate": 0.0003518701897499349,
      "loss": 2.0912,
      "step": 102457
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.9355993866920471,
      "learning_rate": 0.000351866160789782,
      "loss": 1.8778,
      "step": 102458
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.8441067337989807,
      "learning_rate": 0.0003518621318199861,
      "loss": 2.0421,
      "step": 102459
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.8548503518104553,
      "learning_rate": 0.0003518581028405481,
      "loss": 2.1478,
      "step": 102460
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.8323373198509216,
      "learning_rate": 0.00035185407385146875,
      "loss": 1.8888,
      "step": 102461
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.8280138373374939,
      "learning_rate": 0.0003518500448527487,
      "loss": 2.1947,
      "step": 102462
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.8985836505889893,
      "learning_rate": 0.00035184601584438874,
      "loss": 2.0908,
      "step": 102463
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.8818292021751404,
      "learning_rate": 0.0003518419868263897,
      "loss": 2.1544,
      "step": 102464
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.904423177242279,
      "learning_rate": 0.00035183795779875223,
      "loss": 1.99,
      "step": 102465
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.9126985669136047,
      "learning_rate": 0.000351833928761477,
      "loss": 2.2273,
      "step": 102466
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.7642514705657959,
      "learning_rate": 0.000351829899714565,
      "loss": 1.9245,
      "step": 102467
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.9884825348854065,
      "learning_rate": 0.00035182587065801677,
      "loss": 2.2606,
      "step": 102468
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.9279314279556274,
      "learning_rate": 0.00035182184159183316,
      "loss": 2.2272,
      "step": 102469
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.8389534950256348,
      "learning_rate": 0.0003518178125160149,
      "loss": 2.289,
      "step": 102470
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.8632403612136841,
      "learning_rate": 0.0003518137834305627,
      "loss": 2.1811,
      "step": 102471
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.8151851892471313,
      "learning_rate": 0.00035180975433547735,
      "loss": 2.2291,
      "step": 102472
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.7942214012145996,
      "learning_rate": 0.0003518057252307596,
      "loss": 2.2543,
      "step": 102473
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.9144030809402466,
      "learning_rate": 0.0003518016961164102,
      "loss": 2.0021,
      "step": 102474
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.8845804929733276,
      "learning_rate": 0.0003517976669924298,
      "loss": 2.1474,
      "step": 102475
    },
    {
      "epoch": 1.33,
      "grad_norm": 1.00191068649292,
      "learning_rate": 0.0003517936378588194,
      "loss": 2.0401,
      "step": 102476
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.8715801239013672,
      "learning_rate": 0.0003517896087155794,
      "loss": 2.2006,
      "step": 102477
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.9468638896942139,
      "learning_rate": 0.00035178557956271083,
      "loss": 2.153,
      "step": 102478
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.9215702414512634,
      "learning_rate": 0.00035178155040021435,
      "loss": 2.1366,
      "step": 102479
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.9025761485099792,
      "learning_rate": 0.00035177752122809064,
      "loss": 2.4352,
      "step": 102480
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.9125861525535583,
      "learning_rate": 0.00035177349204634057,
      "loss": 2.3427,
      "step": 102481
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.901779055595398,
      "learning_rate": 0.0003517694628549649,
      "loss": 2.0329,
      "step": 102482
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.9088339805603027,
      "learning_rate": 0.00035176543365396417,
      "loss": 2.1944,
      "step": 102483
    },
    {
      "epoch": 1.33,
      "grad_norm": 1.0175009965896606,
      "learning_rate": 0.00035176140444333936,
      "loss": 2.1482,
      "step": 102484
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.9437641501426697,
      "learning_rate": 0.0003517573752230911,
      "loss": 2.2608,
      "step": 102485
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.9139347672462463,
      "learning_rate": 0.00035175334599322014,
      "loss": 2.0802,
      "step": 102486
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.9215494394302368,
      "learning_rate": 0.00035174931675372725,
      "loss": 2.1114,
      "step": 102487
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.841202974319458,
      "learning_rate": 0.00035174528750461326,
      "loss": 2.3277,
      "step": 102488
    },
    {
      "epoch": 1.33,
      "grad_norm": 1.0985321998596191,
      "learning_rate": 0.0003517412582458788,
      "loss": 2.2046,
      "step": 102489
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.8591364622116089,
      "learning_rate": 0.00035173722897752463,
      "loss": 2.1766,
      "step": 102490
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.8615685105323792,
      "learning_rate": 0.00035173319969955156,
      "loss": 1.8258,
      "step": 102491
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.9070526361465454,
      "learning_rate": 0.0003517291704119603,
      "loss": 2.0012,
      "step": 102492
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.9622400403022766,
      "learning_rate": 0.00035172514111475166,
      "loss": 2.0895,
      "step": 102493
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.8301343321800232,
      "learning_rate": 0.00035172111180792635,
      "loss": 2.2355,
      "step": 102494
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.8248885869979858,
      "learning_rate": 0.00035171708249148503,
      "loss": 2.0855,
      "step": 102495
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.915469229221344,
      "learning_rate": 0.0003517130531654286,
      "loss": 1.9862,
      "step": 102496
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.8956125974655151,
      "learning_rate": 0.0003517090238297577,
      "loss": 2.0664,
      "step": 102497
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.816152811050415,
      "learning_rate": 0.00035170499448447306,
      "loss": 2.2608,
      "step": 102498
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.9180059432983398,
      "learning_rate": 0.0003517009651295755,
      "loss": 2.0803,
      "step": 102499
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.8368458151817322,
      "learning_rate": 0.0003516969357650659,
      "loss": 2.0648,
      "step": 102500
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.9334172606468201,
      "learning_rate": 0.00035169290639094474,
      "loss": 2.1348,
      "step": 102501
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.8653347492218018,
      "learning_rate": 0.00035168887700721285,
      "loss": 2.1947,
      "step": 102502
    },
    {
      "epoch": 1.33,
      "grad_norm": 1.0031324625015259,
      "learning_rate": 0.00035168484761387113,
      "loss": 2.1381,
      "step": 102503
    },
    {
      "epoch": 1.33,
      "grad_norm": 1.0643832683563232,
      "learning_rate": 0.0003516808182109202,
      "loss": 2.1296,
      "step": 102504
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.9863523244857788,
      "learning_rate": 0.00035167678879836076,
      "loss": 2.0427,
      "step": 102505
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.8389155864715576,
      "learning_rate": 0.00035167275937619374,
      "loss": 1.9428,
      "step": 102506
    },
    {
      "epoch": 1.33,
      "grad_norm": 1.3116830587387085,
      "learning_rate": 0.0003516687299444197,
      "loss": 2.1297,
      "step": 102507
    },
    {
      "epoch": 1.33,
      "grad_norm": 1.7260103225708008,
      "learning_rate": 0.00035166470050303943,
      "loss": 2.3375,
      "step": 102508
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.8243181109428406,
      "learning_rate": 0.00035166067105205377,
      "loss": 2.3174,
      "step": 102509
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.8986958265304565,
      "learning_rate": 0.0003516566415914635,
      "loss": 2.0717,
      "step": 102510
    },
    {
      "epoch": 1.33,
      "grad_norm": 1.2667688131332397,
      "learning_rate": 0.0003516526121212691,
      "loss": 2.3089,
      "step": 102511
    },
    {
      "epoch": 1.33,
      "grad_norm": 1.0408982038497925,
      "learning_rate": 0.00035164858264147167,
      "loss": 2.1823,
      "step": 102512
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.9175557494163513,
      "learning_rate": 0.0003516445531520717,
      "loss": 2.1197,
      "step": 102513
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.9785417914390564,
      "learning_rate": 0.00035164052365307003,
      "loss": 2.2114,
      "step": 102514
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.8499466180801392,
      "learning_rate": 0.00035163649414446754,
      "loss": 2.1634,
      "step": 102515
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.8787890076637268,
      "learning_rate": 0.0003516324646262647,
      "loss": 2.1855,
      "step": 102516
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.9143869280815125,
      "learning_rate": 0.00035162843509846246,
      "loss": 2.2717,
      "step": 102517
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.8708230257034302,
      "learning_rate": 0.00035162440556106155,
      "loss": 1.9845,
      "step": 102518
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.8848506212234497,
      "learning_rate": 0.00035162037601406263,
      "loss": 2.1701,
      "step": 102519
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.8667040467262268,
      "learning_rate": 0.0003516163464574666,
      "loss": 1.9938,
      "step": 102520
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.8524702191352844,
      "learning_rate": 0.000351612316891274,
      "loss": 2.0162,
      "step": 102521
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.863030195236206,
      "learning_rate": 0.00035160828731548575,
      "loss": 2.274,
      "step": 102522
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.8400306701660156,
      "learning_rate": 0.00035160425773010255,
      "loss": 2.1384,
      "step": 102523
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.8804792165756226,
      "learning_rate": 0.00035160022813512514,
      "loss": 1.934,
      "step": 102524
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.8607087135314941,
      "learning_rate": 0.00035159619853055433,
      "loss": 2.0167,
      "step": 102525
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.912874162197113,
      "learning_rate": 0.0003515921689163907,
      "loss": 2.289,
      "step": 102526
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.9351032972335815,
      "learning_rate": 0.00035158813929263516,
      "loss": 2.1186,
      "step": 102527
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.9929875731468201,
      "learning_rate": 0.0003515841096592885,
      "loss": 2.2968,
      "step": 102528
    },
    {
      "epoch": 1.33,
      "grad_norm": 0.8932254910469055,
      "learning_rate": 0.00035158008001635123,
      "loss": 2.4784,
      "step": 102529
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.9100604057312012,
      "learning_rate": 0.0003515760503638243,
      "loss": 2.2425,
      "step": 102530
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.8572356104850769,
      "learning_rate": 0.00035157202070170844,
      "loss": 1.9741,
      "step": 102531
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.9548785090446472,
      "learning_rate": 0.0003515679910300044,
      "loss": 2.2081,
      "step": 102532
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.9055275321006775,
      "learning_rate": 0.0003515639613487128,
      "loss": 2.0581,
      "step": 102533
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.7752280831336975,
      "learning_rate": 0.0003515599316578346,
      "loss": 2.09,
      "step": 102534
    },
    {
      "epoch": 1.34,
      "grad_norm": 1.0583713054656982,
      "learning_rate": 0.00035155590195737037,
      "loss": 2.1915,
      "step": 102535
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.9727471470832825,
      "learning_rate": 0.00035155187224732094,
      "loss": 2.1797,
      "step": 102536
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.7938289642333984,
      "learning_rate": 0.000351547842527687,
      "loss": 2.056,
      "step": 102537
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.9042090773582458,
      "learning_rate": 0.0003515438127984694,
      "loss": 2.0706,
      "step": 102538
    },
    {
      "epoch": 1.34,
      "grad_norm": 1.0137284994125366,
      "learning_rate": 0.0003515397830596688,
      "loss": 2.3261,
      "step": 102539
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.8357288241386414,
      "learning_rate": 0.00035153575331128595,
      "loss": 2.0972,
      "step": 102540
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.8949124217033386,
      "learning_rate": 0.0003515317235533217,
      "loss": 2.1982,
      "step": 102541
    },
    {
      "epoch": 1.34,
      "grad_norm": 1.227734088897705,
      "learning_rate": 0.00035152769378577663,
      "loss": 2.2993,
      "step": 102542
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.9632149338722229,
      "learning_rate": 0.0003515236640086517,
      "loss": 2.4339,
      "step": 102543
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.8989814519882202,
      "learning_rate": 0.00035151963422194754,
      "loss": 2.0556,
      "step": 102544
    },
    {
      "epoch": 1.34,
      "grad_norm": 1.0452368259429932,
      "learning_rate": 0.00035151560442566484,
      "loss": 2.1055,
      "step": 102545
    },
    {
      "epoch": 1.34,
      "grad_norm": 1.1607853174209595,
      "learning_rate": 0.00035151157461980453,
      "loss": 1.9356,
      "step": 102546
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.8255844712257385,
      "learning_rate": 0.0003515075448043671,
      "loss": 2.0576,
      "step": 102547
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.9868668913841248,
      "learning_rate": 0.00035150351497935346,
      "loss": 2.2151,
      "step": 102548
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.9879882335662842,
      "learning_rate": 0.0003514994851447645,
      "loss": 2.1003,
      "step": 102549
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.8741970658302307,
      "learning_rate": 0.00035149545530060067,
      "loss": 2.143,
      "step": 102550
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.8993427157402039,
      "learning_rate": 0.0003514914254468629,
      "loss": 2.3256,
      "step": 102551
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.8199240565299988,
      "learning_rate": 0.0003514873955835519,
      "loss": 2.3222,
      "step": 102552
    },
    {
      "epoch": 1.34,
      "grad_norm": 1.1769760847091675,
      "learning_rate": 0.00035148336571066844,
      "loss": 2.2926,
      "step": 102553
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.8814632892608643,
      "learning_rate": 0.00035147933582821325,
      "loss": 2.0816,
      "step": 102554
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.8382623791694641,
      "learning_rate": 0.0003514753059361871,
      "loss": 2.1267,
      "step": 102555
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.8955453038215637,
      "learning_rate": 0.0003514712760345907,
      "loss": 2.1884,
      "step": 102556
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.8619335889816284,
      "learning_rate": 0.0003514672461234248,
      "loss": 2.1664,
      "step": 102557
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.9092409014701843,
      "learning_rate": 0.0003514632162026902,
      "loss": 2.0269,
      "step": 102558
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.9676869511604309,
      "learning_rate": 0.0003514591862723875,
      "loss": 2.355,
      "step": 102559
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.9081055521965027,
      "learning_rate": 0.00035145515633251765,
      "loss": 2.1897,
      "step": 102560
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.819213330745697,
      "learning_rate": 0.0003514511263830814,
      "loss": 2.1215,
      "step": 102561
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.8548740744590759,
      "learning_rate": 0.00035144709642407934,
      "loss": 2.2322,
      "step": 102562
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.8665767312049866,
      "learning_rate": 0.0003514430664555122,
      "loss": 2.2047,
      "step": 102563
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.8615961670875549,
      "learning_rate": 0.000351439036477381,
      "loss": 2.1334,
      "step": 102564
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.930878758430481,
      "learning_rate": 0.0003514350064896862,
      "loss": 2.1563,
      "step": 102565
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.7929721474647522,
      "learning_rate": 0.0003514309764924287,
      "loss": 2.013,
      "step": 102566
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.9166209101676941,
      "learning_rate": 0.00035142694648560924,
      "loss": 2.0971,
      "step": 102567
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.8649281859397888,
      "learning_rate": 0.0003514229164692284,
      "loss": 2.0406,
      "step": 102568
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.8719085454940796,
      "learning_rate": 0.0003514188864432872,
      "loss": 2.116,
      "step": 102569
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.8841924667358398,
      "learning_rate": 0.0003514148564077863,
      "loss": 1.9334,
      "step": 102570
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.8469985723495483,
      "learning_rate": 0.00035141082636272627,
      "loss": 2.2791,
      "step": 102571
    },
    {
      "epoch": 1.34,
      "grad_norm": 1.0571043491363525,
      "learning_rate": 0.00035140679630810807,
      "loss": 1.9884,
      "step": 102572
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.9120269417762756,
      "learning_rate": 0.00035140276624393235,
      "loss": 2.0568,
      "step": 102573
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.9151214361190796,
      "learning_rate": 0.0003513987361701999,
      "loss": 2.3643,
      "step": 102574
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.820652186870575,
      "learning_rate": 0.0003513947060869115,
      "loss": 2.1292,
      "step": 102575
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.8749601244926453,
      "learning_rate": 0.0003513906759940678,
      "loss": 1.956,
      "step": 102576
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.854698657989502,
      "learning_rate": 0.00035138664589166955,
      "loss": 2.2147,
      "step": 102577
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.7830439805984497,
      "learning_rate": 0.0003513826157797176,
      "loss": 2.0671,
      "step": 102578
    },
    {
      "epoch": 1.34,
      "grad_norm": 1.0255799293518066,
      "learning_rate": 0.00035137858565821274,
      "loss": 1.9623,
      "step": 102579
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.8780707716941833,
      "learning_rate": 0.00035137455552715553,
      "loss": 1.9349,
      "step": 102580
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.8070458769798279,
      "learning_rate": 0.00035137052538654687,
      "loss": 2.0375,
      "step": 102581
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.8954545855522156,
      "learning_rate": 0.00035136649523638747,
      "loss": 2.09,
      "step": 102582
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.7475388646125793,
      "learning_rate": 0.00035136246507667804,
      "loss": 1.8622,
      "step": 102583
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.9163066744804382,
      "learning_rate": 0.0003513584349074193,
      "loss": 2.2994,
      "step": 102584
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.8813797831535339,
      "learning_rate": 0.0003513544047286122,
      "loss": 2.1175,
      "step": 102585
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.9274595975875854,
      "learning_rate": 0.0003513503745402572,
      "loss": 2.3393,
      "step": 102586
    },
    {
      "epoch": 1.34,
      "grad_norm": 1.0852892398834229,
      "learning_rate": 0.00035134634434235524,
      "loss": 1.8746,
      "step": 102587
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.9535820484161377,
      "learning_rate": 0.0003513423141349071,
      "loss": 2.2058,
      "step": 102588
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.8754776120185852,
      "learning_rate": 0.00035133828391791335,
      "loss": 2.0289,
      "step": 102589
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.892747163772583,
      "learning_rate": 0.00035133425369137484,
      "loss": 2.1219,
      "step": 102590
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.9388410449028015,
      "learning_rate": 0.00035133022345529237,
      "loss": 2.2822,
      "step": 102591
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.9570922255516052,
      "learning_rate": 0.0003513261932096667,
      "loss": 2.2319,
      "step": 102592
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.8644598722457886,
      "learning_rate": 0.0003513221629544984,
      "loss": 2.1675,
      "step": 102593
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.9411078095436096,
      "learning_rate": 0.0003513181326897884,
      "loss": 2.2423,
      "step": 102594
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.8123376965522766,
      "learning_rate": 0.00035131410241553745,
      "loss": 2.2614,
      "step": 102595
    },
    {
      "epoch": 1.34,
      "grad_norm": 1.2821743488311768,
      "learning_rate": 0.0003513100721317462,
      "loss": 2.0727,
      "step": 102596
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.8499405980110168,
      "learning_rate": 0.00035130604183841534,
      "loss": 2.1806,
      "step": 102597
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.914160430431366,
      "learning_rate": 0.0003513020115355459,
      "loss": 2.2095,
      "step": 102598
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.9392300844192505,
      "learning_rate": 0.0003512979812231383,
      "loss": 2.0605,
      "step": 102599
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.8994289636611938,
      "learning_rate": 0.00035129395090119353,
      "loss": 2.3768,
      "step": 102600
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.8400670886039734,
      "learning_rate": 0.0003512899205697122,
      "loss": 2.0417,
      "step": 102601
    },
    {
      "epoch": 1.34,
      "grad_norm": 1.0793395042419434,
      "learning_rate": 0.00035128589022869513,
      "loss": 2.3288,
      "step": 102602
    },
    {
      "epoch": 1.34,
      "grad_norm": 1.0156183242797852,
      "learning_rate": 0.00035128185987814303,
      "loss": 2.1549,
      "step": 102603
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.8405780792236328,
      "learning_rate": 0.00035127782951805667,
      "loss": 2.2056,
      "step": 102604
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.9436637759208679,
      "learning_rate": 0.00035127379914843675,
      "loss": 2.109,
      "step": 102605
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.9037337303161621,
      "learning_rate": 0.0003512697687692841,
      "loss": 2.0682,
      "step": 102606
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.8584232926368713,
      "learning_rate": 0.0003512657383805994,
      "loss": 2.167,
      "step": 102607
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.8780213594436646,
      "learning_rate": 0.0003512617079823835,
      "loss": 2.2365,
      "step": 102608
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.912504255771637,
      "learning_rate": 0.000351257677574637,
      "loss": 1.9736,
      "step": 102609
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.9453991055488586,
      "learning_rate": 0.00035125364715736084,
      "loss": 2.4822,
      "step": 102610
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.8878340125083923,
      "learning_rate": 0.00035124961673055555,
      "loss": 1.9031,
      "step": 102611
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.8347781300544739,
      "learning_rate": 0.00035124558629422197,
      "loss": 2.0312,
      "step": 102612
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.8189873099327087,
      "learning_rate": 0.000351241555848361,
      "loss": 2.2474,
      "step": 102613
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.8911311030387878,
      "learning_rate": 0.00035123752539297317,
      "loss": 2.054,
      "step": 102614
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.8965120911598206,
      "learning_rate": 0.00035123349492805937,
      "loss": 2.1039,
      "step": 102615
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.8977495431900024,
      "learning_rate": 0.0003512294644536203,
      "loss": 2.076,
      "step": 102616
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.8416714668273926,
      "learning_rate": 0.00035122543396965664,
      "loss": 2.0644,
      "step": 102617
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.8074325323104858,
      "learning_rate": 0.0003512214034761692,
      "loss": 1.8471,
      "step": 102618
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.9524263739585876,
      "learning_rate": 0.00035121737297315873,
      "loss": 2.0692,
      "step": 102619
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.8219344019889832,
      "learning_rate": 0.00035121334246062606,
      "loss": 2.1812,
      "step": 102620
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.8547934889793396,
      "learning_rate": 0.00035120931193857177,
      "loss": 1.9973,
      "step": 102621
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.9770851731300354,
      "learning_rate": 0.0003512052814069967,
      "loss": 2.261,
      "step": 102622
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.8147939443588257,
      "learning_rate": 0.0003512012508659017,
      "loss": 2.1263,
      "step": 102623
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.9828964471817017,
      "learning_rate": 0.00035119722031528737,
      "loss": 2.2035,
      "step": 102624
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.9648349285125732,
      "learning_rate": 0.00035119318975515454,
      "loss": 2.058,
      "step": 102625
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.8171917796134949,
      "learning_rate": 0.0003511891591855039,
      "loss": 2.1776,
      "step": 102626
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.8674570322036743,
      "learning_rate": 0.00035118512860633625,
      "loss": 2.2626,
      "step": 102627
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.8993281722068787,
      "learning_rate": 0.00035118109801765226,
      "loss": 2.2475,
      "step": 102628
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.9162035584449768,
      "learning_rate": 0.0003511770674194528,
      "loss": 2.4545,
      "step": 102629
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.8233543038368225,
      "learning_rate": 0.0003511730368117385,
      "loss": 2.1223,
      "step": 102630
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.802299976348877,
      "learning_rate": 0.0003511690061945102,
      "loss": 1.8645,
      "step": 102631
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.9084767699241638,
      "learning_rate": 0.00035116497556776866,
      "loss": 2.2807,
      "step": 102632
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.934334933757782,
      "learning_rate": 0.0003511609449315145,
      "loss": 2.1951,
      "step": 102633
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.8827372789382935,
      "learning_rate": 0.00035115691428574866,
      "loss": 2.119,
      "step": 102634
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.9643961787223816,
      "learning_rate": 0.0003511528836304717,
      "loss": 2.1904,
      "step": 102635
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.8818914294242859,
      "learning_rate": 0.0003511488529656845,
      "loss": 2.4207,
      "step": 102636
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.9615791440010071,
      "learning_rate": 0.0003511448222913877,
      "loss": 2.3263,
      "step": 102637
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.8725897669792175,
      "learning_rate": 0.00035114079160758217,
      "loss": 2.0785,
      "step": 102638
    },
    {
      "epoch": 1.34,
      "grad_norm": 1.0165317058563232,
      "learning_rate": 0.00035113676091426855,
      "loss": 2.4901,
      "step": 102639
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.8103771209716797,
      "learning_rate": 0.0003511327302114477,
      "loss": 2.2579,
      "step": 102640
    },
    {
      "epoch": 1.34,
      "grad_norm": 1.0145525932312012,
      "learning_rate": 0.0003511286994991203,
      "loss": 2.1401,
      "step": 102641
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.9537423849105835,
      "learning_rate": 0.00035112466877728704,
      "loss": 2.1322,
      "step": 102642
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.8851475119590759,
      "learning_rate": 0.0003511206380459489,
      "loss": 2.0859,
      "step": 102643
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.8917242884635925,
      "learning_rate": 0.00035111660730510633,
      "loss": 2.138,
      "step": 102644
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.8985814452171326,
      "learning_rate": 0.0003511125765547602,
      "loss": 2.3694,
      "step": 102645
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.8585076332092285,
      "learning_rate": 0.00035110854579491137,
      "loss": 2.2776,
      "step": 102646
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.828420102596283,
      "learning_rate": 0.0003511045150255604,
      "loss": 2.1957,
      "step": 102647
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.9336673021316528,
      "learning_rate": 0.0003511004842467082,
      "loss": 2.1707,
      "step": 102648
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.9530413746833801,
      "learning_rate": 0.0003510964534583555,
      "loss": 2.2113,
      "step": 102649
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.9900693893432617,
      "learning_rate": 0.00035109242266050296,
      "loss": 2.2898,
      "step": 102650
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.8444066643714905,
      "learning_rate": 0.0003510883918531513,
      "loss": 2.2592,
      "step": 102651
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.850638747215271,
      "learning_rate": 0.0003510843610363015,
      "loss": 2.2178,
      "step": 102652
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.890497088432312,
      "learning_rate": 0.000351080330209954,
      "loss": 2.1731,
      "step": 102653
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.8170623779296875,
      "learning_rate": 0.00035107629937410973,
      "loss": 2.1424,
      "step": 102654
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.8251374363899231,
      "learning_rate": 0.0003510722685287695,
      "loss": 2.1482,
      "step": 102655
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.7930969595909119,
      "learning_rate": 0.00035106823767393387,
      "loss": 2.434,
      "step": 102656
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.8511115908622742,
      "learning_rate": 0.00035106420680960373,
      "loss": 2.1251,
      "step": 102657
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.8402445912361145,
      "learning_rate": 0.00035106017593577977,
      "loss": 2.2418,
      "step": 102658
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.8613340854644775,
      "learning_rate": 0.0003510561450524628,
      "loss": 2.213,
      "step": 102659
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.8560953140258789,
      "learning_rate": 0.00035105211415965354,
      "loss": 2.1664,
      "step": 102660
    },
    {
      "epoch": 1.34,
      "grad_norm": 1.037858247756958,
      "learning_rate": 0.0003510480832573527,
      "loss": 2.1484,
      "step": 102661
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.774114727973938,
      "learning_rate": 0.00035104405234556105,
      "loss": 2.2265,
      "step": 102662
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.9510698914527893,
      "learning_rate": 0.00035104002142427934,
      "loss": 2.1756,
      "step": 102663
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.8725636005401611,
      "learning_rate": 0.0003510359904935083,
      "loss": 2.2016,
      "step": 102664
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.9057077765464783,
      "learning_rate": 0.0003510319595532489,
      "loss": 1.9708,
      "step": 102665
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.8989322781562805,
      "learning_rate": 0.0003510279286035015,
      "loss": 2.2917,
      "step": 102666
    },
    {
      "epoch": 1.34,
      "grad_norm": 1.0048116445541382,
      "learning_rate": 0.000351023897644267,
      "loss": 2.1672,
      "step": 102667
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.88909512758255,
      "learning_rate": 0.00035101986667554643,
      "loss": 2.3471,
      "step": 102668
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.8910518288612366,
      "learning_rate": 0.0003510158356973401,
      "loss": 2.2237,
      "step": 102669
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.9142335653305054,
      "learning_rate": 0.000351011804709649,
      "loss": 2.1866,
      "step": 102670
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.8748500943183899,
      "learning_rate": 0.0003510077737124739,
      "loss": 2.3981,
      "step": 102671
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.8241271376609802,
      "learning_rate": 0.00035100374270581543,
      "loss": 2.1623,
      "step": 102672
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.8860055208206177,
      "learning_rate": 0.0003509997116896744,
      "loss": 1.8946,
      "step": 102673
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.8747870326042175,
      "learning_rate": 0.00035099568066405165,
      "loss": 2.4239,
      "step": 102674
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.8307413458824158,
      "learning_rate": 0.00035099164962894774,
      "loss": 2.1813,
      "step": 102675
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.9658716320991516,
      "learning_rate": 0.0003509876185843635,
      "loss": 1.9906,
      "step": 102676
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.8489815592765808,
      "learning_rate": 0.0003509835875302998,
      "loss": 2.1417,
      "step": 102677
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.8862722516059875,
      "learning_rate": 0.0003509795564667573,
      "loss": 2.3408,
      "step": 102678
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.9088699817657471,
      "learning_rate": 0.0003509755253937366,
      "loss": 2.1814,
      "step": 102679
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.8433170914649963,
      "learning_rate": 0.00035097149431123875,
      "loss": 2.2576,
      "step": 102680
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.9008014798164368,
      "learning_rate": 0.00035096746321926426,
      "loss": 2.1663,
      "step": 102681
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.87203049659729,
      "learning_rate": 0.0003509634321178139,
      "loss": 2.3732,
      "step": 102682
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.8853473663330078,
      "learning_rate": 0.0003509594010068886,
      "loss": 2.2709,
      "step": 102683
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.8909183740615845,
      "learning_rate": 0.0003509553698864889,
      "loss": 2.148,
      "step": 102684
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.8040838837623596,
      "learning_rate": 0.0003509513387566156,
      "loss": 2.1455,
      "step": 102685
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.9492589831352234,
      "learning_rate": 0.0003509473076172696,
      "loss": 2.0657,
      "step": 102686
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.934904932975769,
      "learning_rate": 0.0003509432764684514,
      "loss": 2.3106,
      "step": 102687
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.871262788772583,
      "learning_rate": 0.000350939245310162,
      "loss": 2.1263,
      "step": 102688
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.8225842118263245,
      "learning_rate": 0.000350935214142402,
      "loss": 2.1685,
      "step": 102689
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.9903296828269958,
      "learning_rate": 0.00035093118296517217,
      "loss": 2.0629,
      "step": 102690
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.8395224213600159,
      "learning_rate": 0.00035092715177847326,
      "loss": 2.0468,
      "step": 102691
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.8930495977401733,
      "learning_rate": 0.0003509231205823061,
      "loss": 1.9513,
      "step": 102692
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.8773922324180603,
      "learning_rate": 0.00035091908937667126,
      "loss": 2.2527,
      "step": 102693
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.9029332399368286,
      "learning_rate": 0.0003509150581615696,
      "loss": 2.0863,
      "step": 102694
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.9036465287208557,
      "learning_rate": 0.000350911026937002,
      "loss": 2.16,
      "step": 102695
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.9040282964706421,
      "learning_rate": 0.00035090699570296903,
      "loss": 2.4827,
      "step": 102696
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.8896111845970154,
      "learning_rate": 0.0003509029644594714,
      "loss": 2.1614,
      "step": 102697
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.9127069711685181,
      "learning_rate": 0.00035089893320651,
      "loss": 2.1723,
      "step": 102698
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.8746190071105957,
      "learning_rate": 0.00035089490194408553,
      "loss": 2.2664,
      "step": 102699
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.9844139814376831,
      "learning_rate": 0.00035089087067219875,
      "loss": 2.1034,
      "step": 102700
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.9226425886154175,
      "learning_rate": 0.0003508868393908504,
      "loss": 2.1663,
      "step": 102701
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.8874468803405762,
      "learning_rate": 0.0003508828081000412,
      "loss": 2.1853,
      "step": 102702
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.9202528595924377,
      "learning_rate": 0.0003508787767997719,
      "loss": 2.1793,
      "step": 102703
    },
    {
      "epoch": 1.34,
      "grad_norm": 1.0113439559936523,
      "learning_rate": 0.00035087474549004333,
      "loss": 2.1812,
      "step": 102704
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.8547044396400452,
      "learning_rate": 0.0003508707141708562,
      "loss": 2.245,
      "step": 102705
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.8702062368392944,
      "learning_rate": 0.0003508666828422112,
      "loss": 2.0692,
      "step": 102706
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.9102945923805237,
      "learning_rate": 0.0003508626515041091,
      "loss": 2.2145,
      "step": 102707
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.872529923915863,
      "learning_rate": 0.0003508586201565508,
      "loss": 1.9294,
      "step": 102708
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.9075803160667419,
      "learning_rate": 0.0003508545887995368,
      "loss": 2.2159,
      "step": 102709
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.9145265221595764,
      "learning_rate": 0.000350850557433068,
      "loss": 2.317,
      "step": 102710
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.9666286110877991,
      "learning_rate": 0.00035084652605714504,
      "loss": 2.2217,
      "step": 102711
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.9571205377578735,
      "learning_rate": 0.0003508424946717689,
      "loss": 2.2823,
      "step": 102712
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.8469634652137756,
      "learning_rate": 0.0003508384632769401,
      "loss": 2.1779,
      "step": 102713
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.8871021270751953,
      "learning_rate": 0.0003508344318726595,
      "loss": 2.0544,
      "step": 102714
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.9110179543495178,
      "learning_rate": 0.00035083040045892785,
      "loss": 2.0591,
      "step": 102715
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.8420738577842712,
      "learning_rate": 0.00035082636903574575,
      "loss": 2.1213,
      "step": 102716
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.9300453066825867,
      "learning_rate": 0.0003508223376031142,
      "loss": 1.9689,
      "step": 102717
    },
    {
      "epoch": 1.34,
      "grad_norm": 1.0092536211013794,
      "learning_rate": 0.0003508183061610337,
      "loss": 2.1906,
      "step": 102718
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.8099769949913025,
      "learning_rate": 0.0003508142747095052,
      "loss": 2.0179,
      "step": 102719
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.832852840423584,
      "learning_rate": 0.00035081024324852935,
      "loss": 2.0506,
      "step": 102720
    },
    {
      "epoch": 1.34,
      "grad_norm": 1.0001728534698486,
      "learning_rate": 0.00035080621177810693,
      "loss": 2.1701,
      "step": 102721
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.838729977607727,
      "learning_rate": 0.0003508021802982387,
      "loss": 2.1415,
      "step": 102722
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.8014836311340332,
      "learning_rate": 0.0003507981488089253,
      "loss": 2.1238,
      "step": 102723
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.8797474503517151,
      "learning_rate": 0.0003507941173101676,
      "loss": 2.089,
      "step": 102724
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.8544462323188782,
      "learning_rate": 0.0003507900858019663,
      "loss": 2.2786,
      "step": 102725
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.887200117111206,
      "learning_rate": 0.0003507860542843223,
      "loss": 2.2039,
      "step": 102726
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.7919784784317017,
      "learning_rate": 0.00035078202275723604,
      "loss": 2.2277,
      "step": 102727
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.7893924713134766,
      "learning_rate": 0.00035077799122070846,
      "loss": 2.172,
      "step": 102728
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.9109826683998108,
      "learning_rate": 0.0003507739596747404,
      "loss": 2.1251,
      "step": 102729
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.9565306901931763,
      "learning_rate": 0.00035076992811933243,
      "loss": 1.923,
      "step": 102730
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.9312365651130676,
      "learning_rate": 0.00035076589655448534,
      "loss": 2.1109,
      "step": 102731
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.9212836027145386,
      "learning_rate": 0.0003507618649802,
      "loss": 2.275,
      "step": 102732
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.864230215549469,
      "learning_rate": 0.000350757833396477,
      "loss": 1.985,
      "step": 102733
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.8589146733283997,
      "learning_rate": 0.00035075380180331714,
      "loss": 1.993,
      "step": 102734
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.8999952077865601,
      "learning_rate": 0.0003507497702007213,
      "loss": 2.28,
      "step": 102735
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.8346385955810547,
      "learning_rate": 0.00035074573858869006,
      "loss": 2.1859,
      "step": 102736
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.9145962595939636,
      "learning_rate": 0.0003507417069672242,
      "loss": 2.1263,
      "step": 102737
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.9176850914955139,
      "learning_rate": 0.00035073767533632457,
      "loss": 2.0872,
      "step": 102738
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.8707577586174011,
      "learning_rate": 0.0003507336436959918,
      "loss": 2.1748,
      "step": 102739
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.8485844731330872,
      "learning_rate": 0.00035072961204622667,
      "loss": 2.0817,
      "step": 102740
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.7927107214927673,
      "learning_rate": 0.00035072558038702997,
      "loss": 2.0761,
      "step": 102741
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.8913496136665344,
      "learning_rate": 0.00035072154871840244,
      "loss": 2.1957,
      "step": 102742
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.9175770282745361,
      "learning_rate": 0.00035071751704034474,
      "loss": 2.0492,
      "step": 102743
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.9288004040718079,
      "learning_rate": 0.00035071348535285784,
      "loss": 1.988,
      "step": 102744
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.8658407330513,
      "learning_rate": 0.00035070945365594223,
      "loss": 2.0552,
      "step": 102745
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.8764784336090088,
      "learning_rate": 0.00035070542194959883,
      "loss": 2.1249,
      "step": 102746
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.8484211564064026,
      "learning_rate": 0.0003507013902338283,
      "loss": 2.0368,
      "step": 102747
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.9524476528167725,
      "learning_rate": 0.0003506973585086315,
      "loss": 2.4897,
      "step": 102748
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.885452151298523,
      "learning_rate": 0.000350693326774009,
      "loss": 2.1895,
      "step": 102749
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.928244948387146,
      "learning_rate": 0.0003506892950299617,
      "loss": 2.2271,
      "step": 102750
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.8497669696807861,
      "learning_rate": 0.0003506852632764903,
      "loss": 2.0231,
      "step": 102751
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.8547126054763794,
      "learning_rate": 0.00035068123151359555,
      "loss": 2.1671,
      "step": 102752
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.8600753545761108,
      "learning_rate": 0.00035067719974127824,
      "loss": 1.9556,
      "step": 102753
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.8795002102851868,
      "learning_rate": 0.00035067316795953904,
      "loss": 2.2833,
      "step": 102754
    },
    {
      "epoch": 1.34,
      "grad_norm": 1.0086342096328735,
      "learning_rate": 0.0003506691361683787,
      "loss": 2.0363,
      "step": 102755
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.8007491827011108,
      "learning_rate": 0.00035066510436779815,
      "loss": 2.1685,
      "step": 102756
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.845925509929657,
      "learning_rate": 0.0003506610725577979,
      "loss": 2.1399,
      "step": 102757
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.8929721713066101,
      "learning_rate": 0.00035065704073837875,
      "loss": 2.019,
      "step": 102758
    },
    {
      "epoch": 1.34,
      "grad_norm": 1.0340255498886108,
      "learning_rate": 0.0003506530089095416,
      "loss": 2.1757,
      "step": 102759
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.8989657163619995,
      "learning_rate": 0.00035064897707128706,
      "loss": 2.1582,
      "step": 102760
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.8143423795700073,
      "learning_rate": 0.0003506449452236159,
      "loss": 2.266,
      "step": 102761
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.9010813236236572,
      "learning_rate": 0.00035064091336652894,
      "loss": 2.0327,
      "step": 102762
    },
    {
      "epoch": 1.34,
      "grad_norm": 1.0425924062728882,
      "learning_rate": 0.0003506368815000268,
      "loss": 2.3105,
      "step": 102763
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.881224513053894,
      "learning_rate": 0.00035063284962411035,
      "loss": 2.0505,
      "step": 102764
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.8029662370681763,
      "learning_rate": 0.00035062881773878035,
      "loss": 2.2688,
      "step": 102765
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.8815725445747375,
      "learning_rate": 0.0003506247858440375,
      "loss": 2.0873,
      "step": 102766
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.9478844404220581,
      "learning_rate": 0.0003506207539398824,
      "loss": 2.4492,
      "step": 102767
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.9173095226287842,
      "learning_rate": 0.0003506167220263161,
      "loss": 2.1983,
      "step": 102768
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.8564584851264954,
      "learning_rate": 0.00035061269010333915,
      "loss": 2.2999,
      "step": 102769
    },
    {
      "epoch": 1.34,
      "grad_norm": 1.1053961515426636,
      "learning_rate": 0.00035060865817095234,
      "loss": 2.1226,
      "step": 102770
    },
    {
      "epoch": 1.34,
      "grad_norm": 1.0701966285705566,
      "learning_rate": 0.00035060462622915644,
      "loss": 2.3367,
      "step": 102771
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.8355247974395752,
      "learning_rate": 0.0003506005942779521,
      "loss": 2.3167,
      "step": 102772
    },
    {
      "epoch": 1.34,
      "grad_norm": 1.1418288946151733,
      "learning_rate": 0.0003505965623173402,
      "loss": 1.9499,
      "step": 102773
    },
    {
      "epoch": 1.34,
      "grad_norm": 1.1803983449935913,
      "learning_rate": 0.00035059253034732155,
      "loss": 2.276,
      "step": 102774
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.9452685117721558,
      "learning_rate": 0.0003505884983678967,
      "loss": 2.1037,
      "step": 102775
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.906662106513977,
      "learning_rate": 0.00035058446637906647,
      "loss": 2.0033,
      "step": 102776
    },
    {
      "epoch": 1.34,
      "grad_norm": 1.0086095333099365,
      "learning_rate": 0.00035058043438083167,
      "loss": 2.3233,
      "step": 102777
    },
    {
      "epoch": 1.34,
      "grad_norm": 1.065066933631897,
      "learning_rate": 0.000350576402373193,
      "loss": 2.1856,
      "step": 102778
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.891407310962677,
      "learning_rate": 0.00035057237035615124,
      "loss": 2.3665,
      "step": 102779
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.9025174379348755,
      "learning_rate": 0.0003505683383297071,
      "loss": 2.1293,
      "step": 102780
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.9721927046775818,
      "learning_rate": 0.00035056430629386135,
      "loss": 2.1871,
      "step": 102781
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.9231087565422058,
      "learning_rate": 0.00035056027424861473,
      "loss": 2.0958,
      "step": 102782
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.8696544170379639,
      "learning_rate": 0.00035055624219396803,
      "loss": 2.09,
      "step": 102783
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.8099400997161865,
      "learning_rate": 0.000350552210129922,
      "loss": 2.2791,
      "step": 102784
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.8194465637207031,
      "learning_rate": 0.0003505481780564773,
      "loss": 2.0248,
      "step": 102785
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.8513517379760742,
      "learning_rate": 0.0003505441459736348,
      "loss": 2.1593,
      "step": 102786
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.9077621102333069,
      "learning_rate": 0.00035054011388139513,
      "loss": 2.0848,
      "step": 102787
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.832603394985199,
      "learning_rate": 0.00035053608177975905,
      "loss": 2.0872,
      "step": 102788
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.8303961753845215,
      "learning_rate": 0.0003505320496687275,
      "loss": 2.1167,
      "step": 102789
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.9096629023551941,
      "learning_rate": 0.000350528017548301,
      "loss": 2.0396,
      "step": 102790
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.8626229763031006,
      "learning_rate": 0.00035052398541848037,
      "loss": 2.2898,
      "step": 102791
    },
    {
      "epoch": 1.34,
      "grad_norm": 1.0367422103881836,
      "learning_rate": 0.0003505199532792664,
      "loss": 2.185,
      "step": 102792
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.8383360505104065,
      "learning_rate": 0.00035051592113065984,
      "loss": 2.1072,
      "step": 102793
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.8256728649139404,
      "learning_rate": 0.0003505118889726614,
      "loss": 2.1018,
      "step": 102794
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.8187040686607361,
      "learning_rate": 0.00035050785680527187,
      "loss": 1.9889,
      "step": 102795
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.8986522555351257,
      "learning_rate": 0.00035050382462849196,
      "loss": 2.1582,
      "step": 102796
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.8342492580413818,
      "learning_rate": 0.0003504997924423224,
      "loss": 1.942,
      "step": 102797
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.9798849821090698,
      "learning_rate": 0.00035049576024676407,
      "loss": 2.3845,
      "step": 102798
    },
    {
      "epoch": 1.34,
      "grad_norm": 1.0506603717803955,
      "learning_rate": 0.0003504917280418175,
      "loss": 1.8735,
      "step": 102799
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.8309759497642517,
      "learning_rate": 0.0003504876958274837,
      "loss": 2.1067,
      "step": 102800
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.9106535911560059,
      "learning_rate": 0.00035048366360376315,
      "loss": 2.2294,
      "step": 102801
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.9103690981864929,
      "learning_rate": 0.0003504796313706568,
      "loss": 2.1141,
      "step": 102802
    },
    {
      "epoch": 1.34,
      "grad_norm": 1.1286879777908325,
      "learning_rate": 0.0003504755991281653,
      "loss": 2.1689,
      "step": 102803
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.9223870635032654,
      "learning_rate": 0.00035047156687628955,
      "loss": 2.0653,
      "step": 102804
    },
    {
      "epoch": 1.34,
      "grad_norm": 1.004183053970337,
      "learning_rate": 0.00035046753461503,
      "loss": 2.3027,
      "step": 102805
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.994831919670105,
      "learning_rate": 0.00035046350234438774,
      "loss": 2.1723,
      "step": 102806
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.8330473899841309,
      "learning_rate": 0.0003504594700643633,
      "loss": 2.1198,
      "step": 102807
    },
    {
      "epoch": 1.34,
      "grad_norm": 1.067051887512207,
      "learning_rate": 0.00035045543777495746,
      "loss": 2.3298,
      "step": 102808
    },
    {
      "epoch": 1.34,
      "grad_norm": 1.1164859533309937,
      "learning_rate": 0.000350451405476171,
      "loss": 2.1302,
      "step": 102809
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.8125698566436768,
      "learning_rate": 0.0003504473731680047,
      "loss": 2.0659,
      "step": 102810
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.9347037672996521,
      "learning_rate": 0.0003504433408504593,
      "loss": 2.3019,
      "step": 102811
    },
    {
      "epoch": 1.34,
      "grad_norm": 1.2580209970474243,
      "learning_rate": 0.0003504393085235355,
      "loss": 2.2096,
      "step": 102812
    },
    {
      "epoch": 1.34,
      "grad_norm": 1.060247540473938,
      "learning_rate": 0.0003504352761872341,
      "loss": 2.1181,
      "step": 102813
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.9129323363304138,
      "learning_rate": 0.0003504312438415559,
      "loss": 2.1081,
      "step": 102814
    },
    {
      "epoch": 1.34,
      "grad_norm": 1.2353123426437378,
      "learning_rate": 0.00035042721148650143,
      "loss": 2.2051,
      "step": 102815
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.9103253483772278,
      "learning_rate": 0.00035042317912207166,
      "loss": 2.3585,
      "step": 102816
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.8303169012069702,
      "learning_rate": 0.0003504191467482673,
      "loss": 1.8266,
      "step": 102817
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.8702498078346252,
      "learning_rate": 0.0003504151143650891,
      "loss": 2.1384,
      "step": 102818
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.8705256581306458,
      "learning_rate": 0.00035041108197253766,
      "loss": 2.0444,
      "step": 102819
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.870400607585907,
      "learning_rate": 0.000350407049570614,
      "loss": 2.1126,
      "step": 102820
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.833747148513794,
      "learning_rate": 0.0003504030171593186,
      "loss": 2.0405,
      "step": 102821
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.9566189050674438,
      "learning_rate": 0.00035039898473865236,
      "loss": 2.2544,
      "step": 102822
    },
    {
      "epoch": 1.34,
      "grad_norm": 1.117110252380371,
      "learning_rate": 0.000350394952308616,
      "loss": 2.0562,
      "step": 102823
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.847720742225647,
      "learning_rate": 0.0003503909198692102,
      "loss": 2.258,
      "step": 102824
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.8873925805091858,
      "learning_rate": 0.00035038688742043586,
      "loss": 2.1637,
      "step": 102825
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.8514949679374695,
      "learning_rate": 0.00035038285496229366,
      "loss": 2.061,
      "step": 102826
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.9959533214569092,
      "learning_rate": 0.0003503788224947843,
      "loss": 2.3068,
      "step": 102827
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.8492602109909058,
      "learning_rate": 0.00035037479001790854,
      "loss": 2.0943,
      "step": 102828
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.78826904296875,
      "learning_rate": 0.0003503707575316672,
      "loss": 2.1549,
      "step": 102829
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.900878369808197,
      "learning_rate": 0.000350366725036061,
      "loss": 2.3802,
      "step": 102830
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.912400484085083,
      "learning_rate": 0.00035036269253109064,
      "loss": 2.1973,
      "step": 102831
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.8594517707824707,
      "learning_rate": 0.0003503586600167569,
      "loss": 2.1329,
      "step": 102832
    },
    {
      "epoch": 1.34,
      "grad_norm": 1.0399774312973022,
      "learning_rate": 0.00035035462749306054,
      "loss": 2.0008,
      "step": 102833
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.8813562393188477,
      "learning_rate": 0.0003503505949600023,
      "loss": 2.3893,
      "step": 102834
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.8463728427886963,
      "learning_rate": 0.00035034656241758296,
      "loss": 2.1739,
      "step": 102835
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.9069309830665588,
      "learning_rate": 0.0003503425298658032,
      "loss": 2.312,
      "step": 102836
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.8842479586601257,
      "learning_rate": 0.00035033849730466387,
      "loss": 2.0205,
      "step": 102837
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.89913010597229,
      "learning_rate": 0.00035033446473416564,
      "loss": 2.3225,
      "step": 102838
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.8763708472251892,
      "learning_rate": 0.0003503304321543093,
      "loss": 2.1292,
      "step": 102839
    },
    {
      "epoch": 1.34,
      "grad_norm": 1.1629425287246704,
      "learning_rate": 0.00035032639956509545,
      "loss": 2.0615,
      "step": 102840
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.8338073492050171,
      "learning_rate": 0.0003503223669665252,
      "loss": 2.1261,
      "step": 102841
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.8458184003829956,
      "learning_rate": 0.0003503183343585989,
      "loss": 2.0877,
      "step": 102842
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.8804086446762085,
      "learning_rate": 0.00035031430174131754,
      "loss": 2.3109,
      "step": 102843
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.9705203175544739,
      "learning_rate": 0.0003503102691146818,
      "loss": 2.0948,
      "step": 102844
    },
    {
      "epoch": 1.34,
      "grad_norm": 1.0393184423446655,
      "learning_rate": 0.0003503062364786924,
      "loss": 1.9891,
      "step": 102845
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.8769916296005249,
      "learning_rate": 0.0003503022038333501,
      "loss": 2.2243,
      "step": 102846
    },
    {
      "epoch": 1.34,
      "grad_norm": 1.31564462184906,
      "learning_rate": 0.0003502981711786558,
      "loss": 2.2356,
      "step": 102847
    },
    {
      "epoch": 1.34,
      "grad_norm": 1.6286399364471436,
      "learning_rate": 0.00035029413851460996,
      "loss": 2.0832,
      "step": 102848
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.8818348050117493,
      "learning_rate": 0.00035029010584121357,
      "loss": 2.4084,
      "step": 102849
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.9635701775550842,
      "learning_rate": 0.00035028607315846735,
      "loss": 1.9531,
      "step": 102850
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.9903550744056702,
      "learning_rate": 0.0003502820404663719,
      "loss": 2.1692,
      "step": 102851
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.8595141768455505,
      "learning_rate": 0.00035027800776492806,
      "loss": 2.199,
      "step": 102852
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.8672268390655518,
      "learning_rate": 0.0003502739750541367,
      "loss": 2.2028,
      "step": 102853
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.8572511076927185,
      "learning_rate": 0.0003502699423339984,
      "loss": 2.3781,
      "step": 102854
    },
    {
      "epoch": 1.34,
      "grad_norm": 1.1434922218322754,
      "learning_rate": 0.0003502659096045139,
      "loss": 2.1648,
      "step": 102855
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.9342847466468811,
      "learning_rate": 0.00035026187686568416,
      "loss": 2.2488,
      "step": 102856
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.8946489095687866,
      "learning_rate": 0.0003502578441175097,
      "loss": 2.1262,
      "step": 102857
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.8798868656158447,
      "learning_rate": 0.0003502538113599913,
      "loss": 2.2538,
      "step": 102858
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.8615244030952454,
      "learning_rate": 0.00035024977859312987,
      "loss": 2.4546,
      "step": 102859
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.9252325892448425,
      "learning_rate": 0.000350245745816926,
      "loss": 2.2862,
      "step": 102860
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.8743487000465393,
      "learning_rate": 0.00035024171303138057,
      "loss": 2.0412,
      "step": 102861
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.9685444831848145,
      "learning_rate": 0.00035023768023649413,
      "loss": 2.1333,
      "step": 102862
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.9520020484924316,
      "learning_rate": 0.0003502336474322677,
      "loss": 2.1363,
      "step": 102863
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.882503867149353,
      "learning_rate": 0.0003502296146187018,
      "loss": 2.0431,
      "step": 102864
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.8790062069892883,
      "learning_rate": 0.00035022558179579726,
      "loss": 2.1351,
      "step": 102865
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.8533623814582825,
      "learning_rate": 0.0003502215489635549,
      "loss": 2.1577,
      "step": 102866
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.974808394908905,
      "learning_rate": 0.0003502175161219754,
      "loss": 2.1037,
      "step": 102867
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.8932872414588928,
      "learning_rate": 0.0003502134832710595,
      "loss": 2.1585,
      "step": 102868
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.8374416828155518,
      "learning_rate": 0.00035020945041080797,
      "loss": 2.1123,
      "step": 102869
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.9448373913764954,
      "learning_rate": 0.0003502054175412215,
      "loss": 2.1691,
      "step": 102870
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.9216015338897705,
      "learning_rate": 0.00035020138466230094,
      "loss": 2.0173,
      "step": 102871
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.9090297818183899,
      "learning_rate": 0.000350197351774047,
      "loss": 2.585,
      "step": 102872
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.8115150928497314,
      "learning_rate": 0.00035019331887646045,
      "loss": 2.0722,
      "step": 102873
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.9016033411026001,
      "learning_rate": 0.00035018928596954193,
      "loss": 1.9991,
      "step": 102874
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.8526520133018494,
      "learning_rate": 0.0003501852530532924,
      "loss": 2.1215,
      "step": 102875
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.8485196232795715,
      "learning_rate": 0.00035018122012771236,
      "loss": 2.0905,
      "step": 102876
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.9185701012611389,
      "learning_rate": 0.00035017718719280267,
      "loss": 2.0465,
      "step": 102877
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.8938535451889038,
      "learning_rate": 0.00035017315424856424,
      "loss": 2.197,
      "step": 102878
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.9306392669677734,
      "learning_rate": 0.00035016912129499754,
      "loss": 2.2324,
      "step": 102879
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.9371951818466187,
      "learning_rate": 0.0003501650883321035,
      "loss": 2.2622,
      "step": 102880
    },
    {
      "epoch": 1.34,
      "grad_norm": 1.1925005912780762,
      "learning_rate": 0.00035016105535988284,
      "loss": 2.3315,
      "step": 102881
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.8983577489852905,
      "learning_rate": 0.00035015702237833625,
      "loss": 2.1286,
      "step": 102882
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.9555920362472534,
      "learning_rate": 0.0003501529893874646,
      "loss": 2.2338,
      "step": 102883
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.9370794296264648,
      "learning_rate": 0.0003501489563872685,
      "loss": 2.0663,
      "step": 102884
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.8426997661590576,
      "learning_rate": 0.0003501449233777488,
      "loss": 2.2447,
      "step": 102885
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.9021822214126587,
      "learning_rate": 0.00035014089035890616,
      "loss": 2.1803,
      "step": 102886
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.933078408241272,
      "learning_rate": 0.00035013685733074136,
      "loss": 2.2029,
      "step": 102887
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.8602592945098877,
      "learning_rate": 0.0003501328242932553,
      "loss": 2.1804,
      "step": 102888
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.8441979885101318,
      "learning_rate": 0.00035012879124644847,
      "loss": 2.2248,
      "step": 102889
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.9424152970314026,
      "learning_rate": 0.0003501247581903218,
      "loss": 1.9939,
      "step": 102890
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.8879417777061462,
      "learning_rate": 0.000350120725124876,
      "loss": 2.2854,
      "step": 102891
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.9384087920188904,
      "learning_rate": 0.00035011669205011186,
      "loss": 2.0228,
      "step": 102892
    },
    {
      "epoch": 1.34,
      "grad_norm": 1.0849237442016602,
      "learning_rate": 0.00035011265896603,
      "loss": 2.3258,
      "step": 102893
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.866855263710022,
      "learning_rate": 0.00035010862587263124,
      "loss": 2.5088,
      "step": 102894
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.8175002336502075,
      "learning_rate": 0.0003501045927699164,
      "loss": 2.161,
      "step": 102895
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.8378339409828186,
      "learning_rate": 0.00035010055965788617,
      "loss": 2.1563,
      "step": 102896
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.7955731153488159,
      "learning_rate": 0.00035009652653654125,
      "loss": 2.1679,
      "step": 102897
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.9649171829223633,
      "learning_rate": 0.00035009249340588243,
      "loss": 2.2777,
      "step": 102898
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.8908215165138245,
      "learning_rate": 0.00035008846026591054,
      "loss": 2.067,
      "step": 102899
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.8972659111022949,
      "learning_rate": 0.00035008442711662626,
      "loss": 2.1778,
      "step": 102900
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.9206421375274658,
      "learning_rate": 0.00035008039395803027,
      "loss": 2.2686,
      "step": 102901
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.8038588762283325,
      "learning_rate": 0.00035007636079012347,
      "loss": 2.2073,
      "step": 102902
    },
    {
      "epoch": 1.34,
      "grad_norm": 1.0285999774932861,
      "learning_rate": 0.00035007232761290647,
      "loss": 2.0227,
      "step": 102903
    },
    {
      "epoch": 1.34,
      "grad_norm": 1.1417312622070312,
      "learning_rate": 0.00035006829442638007,
      "loss": 2.1168,
      "step": 102904
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.9594606161117554,
      "learning_rate": 0.00035006426123054514,
      "loss": 2.3027,
      "step": 102905
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.9719054102897644,
      "learning_rate": 0.00035006022802540224,
      "loss": 2.1316,
      "step": 102906
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.8448721766471863,
      "learning_rate": 0.0003500561948109521,
      "loss": 2.0724,
      "step": 102907
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.8825246095657349,
      "learning_rate": 0.0003500521615871958,
      "loss": 2.1427,
      "step": 102908
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.8848810791969299,
      "learning_rate": 0.00035004812835413363,
      "loss": 2.1098,
      "step": 102909
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.8035045862197876,
      "learning_rate": 0.0003500440951117666,
      "loss": 2.1235,
      "step": 102910
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.9475666880607605,
      "learning_rate": 0.0003500400618600956,
      "loss": 2.1943,
      "step": 102911
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.8218812346458435,
      "learning_rate": 0.00035003602859912103,
      "loss": 2.2384,
      "step": 102912
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.8965258002281189,
      "learning_rate": 0.00035003199532884386,
      "loss": 1.9361,
      "step": 102913
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.9055341482162476,
      "learning_rate": 0.0003500279620492649,
      "loss": 2.1805,
      "step": 102914
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.7913602590560913,
      "learning_rate": 0.00035002392876038466,
      "loss": 2.1944,
      "step": 102915
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.8546275496482849,
      "learning_rate": 0.0003500198954622041,
      "loss": 2.2719,
      "step": 102916
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.9771749973297119,
      "learning_rate": 0.0003500158621547239,
      "loss": 2.1755,
      "step": 102917
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.9291344881057739,
      "learning_rate": 0.0003500118288379448,
      "loss": 1.94,
      "step": 102918
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.822029709815979,
      "learning_rate": 0.0003500077955118675,
      "loss": 1.9918,
      "step": 102919
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.8869739770889282,
      "learning_rate": 0.0003500037621764929,
      "loss": 2.1777,
      "step": 102920
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.8417843580245972,
      "learning_rate": 0.0003499997288318216,
      "loss": 2.0071,
      "step": 102921
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.8568942546844482,
      "learning_rate": 0.0003499956954778544,
      "loss": 2.3327,
      "step": 102922
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.9544559121131897,
      "learning_rate": 0.00034999166211459207,
      "loss": 2.0747,
      "step": 102923
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.8651971817016602,
      "learning_rate": 0.00034998762874203533,
      "loss": 2.0296,
      "step": 102924
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.9925895929336548,
      "learning_rate": 0.000349983595360185,
      "loss": 2.0627,
      "step": 102925
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.8569381237030029,
      "learning_rate": 0.0003499795619690417,
      "loss": 2.4545,
      "step": 102926
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.9455121755599976,
      "learning_rate": 0.0003499755285686064,
      "loss": 2.1863,
      "step": 102927
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.9008747935295105,
      "learning_rate": 0.0003499714951588795,
      "loss": 2.2323,
      "step": 102928
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.8669717907905579,
      "learning_rate": 0.0003499674617398621,
      "loss": 2.3916,
      "step": 102929
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.9218987226486206,
      "learning_rate": 0.0003499634283115548,
      "loss": 2.221,
      "step": 102930
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.8053724765777588,
      "learning_rate": 0.00034995939487395824,
      "loss": 2.1468,
      "step": 102931
    },
    {
      "epoch": 1.34,
      "grad_norm": 1.0200151205062866,
      "learning_rate": 0.0003499553614270734,
      "loss": 2.1514,
      "step": 102932
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.9056196808815002,
      "learning_rate": 0.0003499513279709009,
      "loss": 2.5052,
      "step": 102933
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.939551830291748,
      "learning_rate": 0.0003499472945054415,
      "loss": 2.0,
      "step": 102934
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.881873369216919,
      "learning_rate": 0.0003499432610306959,
      "loss": 2.2709,
      "step": 102935
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.9178209900856018,
      "learning_rate": 0.00034993922754666495,
      "loss": 2.3637,
      "step": 102936
    },
    {
      "epoch": 1.34,
      "grad_norm": 1.003395915031433,
      "learning_rate": 0.00034993519405334935,
      "loss": 2.0806,
      "step": 102937
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.9786250591278076,
      "learning_rate": 0.0003499311605507498,
      "loss": 2.0218,
      "step": 102938
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.975783109664917,
      "learning_rate": 0.00034992712703886726,
      "loss": 2.0294,
      "step": 102939
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.8882451057434082,
      "learning_rate": 0.0003499230935177022,
      "loss": 1.9954,
      "step": 102940
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.8406006097793579,
      "learning_rate": 0.0003499190599872555,
      "loss": 2.1216,
      "step": 102941
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.815445601940155,
      "learning_rate": 0.000349915026447528,
      "loss": 2.0348,
      "step": 102942
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.8588355183601379,
      "learning_rate": 0.0003499109928985202,
      "loss": 1.9603,
      "step": 102943
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.9792916774749756,
      "learning_rate": 0.0003499069593402331,
      "loss": 2.1421,
      "step": 102944
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.8315536379814148,
      "learning_rate": 0.00034990292577266743,
      "loss": 2.0495,
      "step": 102945
    },
    {
      "epoch": 1.34,
      "grad_norm": 1.0528584718704224,
      "learning_rate": 0.0003498988921958237,
      "loss": 2.3411,
      "step": 102946
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.9129324555397034,
      "learning_rate": 0.0003498948586097029,
      "loss": 2.105,
      "step": 102947
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.9760680794715881,
      "learning_rate": 0.00034989082501430573,
      "loss": 2.2601,
      "step": 102948
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.8359334468841553,
      "learning_rate": 0.00034988679140963286,
      "loss": 2.3038,
      "step": 102949
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.873370885848999,
      "learning_rate": 0.00034988275779568507,
      "loss": 2.0413,
      "step": 102950
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.8558604717254639,
      "learning_rate": 0.00034987872417246323,
      "loss": 2.1611,
      "step": 102951
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.8255003094673157,
      "learning_rate": 0.000349874690539968,
      "loss": 2.4251,
      "step": 102952
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.8813507556915283,
      "learning_rate": 0.00034987065689819996,
      "loss": 2.3357,
      "step": 102953
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.8959315419197083,
      "learning_rate": 0.0003498666232471602,
      "loss": 2.1655,
      "step": 102954
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.8927869200706482,
      "learning_rate": 0.0003498625895868492,
      "loss": 2.4089,
      "step": 102955
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.8909792900085449,
      "learning_rate": 0.0003498585559172678,
      "loss": 2.2896,
      "step": 102956
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.809432327747345,
      "learning_rate": 0.0003498545222384169,
      "loss": 2.3304,
      "step": 102957
    },
    {
      "epoch": 1.34,
      "grad_norm": 1.0472493171691895,
      "learning_rate": 0.0003498504885502969,
      "loss": 2.462,
      "step": 102958
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.8769208192825317,
      "learning_rate": 0.0003498464548529089,
      "loss": 2.2157,
      "step": 102959
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.8505194783210754,
      "learning_rate": 0.00034984242114625345,
      "loss": 2.285,
      "step": 102960
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.899133563041687,
      "learning_rate": 0.0003498383874303313,
      "loss": 2.3373,
      "step": 102961
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.9199471473693848,
      "learning_rate": 0.0003498343537051433,
      "loss": 2.0127,
      "step": 102962
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.8942369818687439,
      "learning_rate": 0.00034983031997069016,
      "loss": 2.2293,
      "step": 102963
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.8488209247589111,
      "learning_rate": 0.0003498262862269726,
      "loss": 2.2887,
      "step": 102964
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.8794844150543213,
      "learning_rate": 0.0003498222524739914,
      "loss": 1.9517,
      "step": 102965
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.88596510887146,
      "learning_rate": 0.0003498182187117473,
      "loss": 2.1281,
      "step": 102966
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.8209123611450195,
      "learning_rate": 0.00034981418494024107,
      "loss": 2.2852,
      "step": 102967
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.8131546974182129,
      "learning_rate": 0.00034981015115947343,
      "loss": 1.8971,
      "step": 102968
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.9188861846923828,
      "learning_rate": 0.0003498061173694452,
      "loss": 2.246,
      "step": 102969
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.9356393814086914,
      "learning_rate": 0.000349802083570157,
      "loss": 2.198,
      "step": 102970
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.9548165798187256,
      "learning_rate": 0.0003497980497616096,
      "loss": 2.0322,
      "step": 102971
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.9342555999755859,
      "learning_rate": 0.0003497940159438039,
      "loss": 1.9814,
      "step": 102972
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.8860859274864197,
      "learning_rate": 0.00034978998211674056,
      "loss": 2.2561,
      "step": 102973
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.8916812539100647,
      "learning_rate": 0.0003497859482804203,
      "loss": 2.0213,
      "step": 102974
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.8405705094337463,
      "learning_rate": 0.0003497819144348438,
      "loss": 2.3936,
      "step": 102975
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.8730934262275696,
      "learning_rate": 0.00034977788058001197,
      "loss": 2.0605,
      "step": 102976
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.8958613276481628,
      "learning_rate": 0.0003497738467159256,
      "loss": 2.0066,
      "step": 102977
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.8430079221725464,
      "learning_rate": 0.00034976981284258517,
      "loss": 2.0476,
      "step": 102978
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.893815815448761,
      "learning_rate": 0.0003497657789599917,
      "loss": 2.4588,
      "step": 102979
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.8548080325126648,
      "learning_rate": 0.0003497617450681458,
      "loss": 1.9524,
      "step": 102980
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.9476867318153381,
      "learning_rate": 0.00034975771116704823,
      "loss": 2.2098,
      "step": 102981
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.9028056263923645,
      "learning_rate": 0.00034975367725669973,
      "loss": 2.1314,
      "step": 102982
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.9830178022384644,
      "learning_rate": 0.00034974964333710115,
      "loss": 2.0535,
      "step": 102983
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.9528473615646362,
      "learning_rate": 0.0003497456094082532,
      "loss": 2.2348,
      "step": 102984
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.8270090222358704,
      "learning_rate": 0.00034974157547015654,
      "loss": 1.9056,
      "step": 102985
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.9096464514732361,
      "learning_rate": 0.000349737541522812,
      "loss": 1.9745,
      "step": 102986
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.8960926532745361,
      "learning_rate": 0.0003497335075662203,
      "loss": 2.1822,
      "step": 102987
    },
    {
      "epoch": 1.34,
      "grad_norm": 1.805726408958435,
      "learning_rate": 0.0003497294736003822,
      "loss": 2.302,
      "step": 102988
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.8481078743934631,
      "learning_rate": 0.0003497254396252984,
      "loss": 2.2238,
      "step": 102989
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.9183220267295837,
      "learning_rate": 0.00034972140564096986,
      "loss": 2.0287,
      "step": 102990
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.8297486305236816,
      "learning_rate": 0.00034971737164739703,
      "loss": 2.314,
      "step": 102991
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.9047824144363403,
      "learning_rate": 0.0003497133376445809,
      "loss": 2.1637,
      "step": 102992
    },
    {
      "epoch": 1.34,
      "grad_norm": 1.0431278944015503,
      "learning_rate": 0.00034970930363252203,
      "loss": 2.2915,
      "step": 102993
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.9270706176757812,
      "learning_rate": 0.00034970526961122133,
      "loss": 2.1272,
      "step": 102994
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.8055630922317505,
      "learning_rate": 0.0003497012355806795,
      "loss": 1.9896,
      "step": 102995
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.8620568513870239,
      "learning_rate": 0.0003496972015408972,
      "loss": 2.0584,
      "step": 102996
    },
    {
      "epoch": 1.34,
      "grad_norm": 1.0564297437667847,
      "learning_rate": 0.0003496931674918753,
      "loss": 2.1813,
      "step": 102997
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.8794786334037781,
      "learning_rate": 0.00034968913343361445,
      "loss": 2.1574,
      "step": 102998
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.865289568901062,
      "learning_rate": 0.00034968509936611546,
      "loss": 1.9312,
      "step": 102999
    },
    {
      "epoch": 1.34,
      "grad_norm": 1.0004812479019165,
      "learning_rate": 0.0003496810652893791,
      "loss": 2.1569,
      "step": 103000
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.9256055951118469,
      "learning_rate": 0.0003496770312034061,
      "loss": 2.2508,
      "step": 103001
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.783194363117218,
      "learning_rate": 0.0003496729971081972,
      "loss": 2.1638,
      "step": 103002
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.8917178511619568,
      "learning_rate": 0.00034966896300375316,
      "loss": 2.3046,
      "step": 103003
    },
    {
      "epoch": 1.34,
      "grad_norm": 1.0624427795410156,
      "learning_rate": 0.00034966492889007474,
      "loss": 2.1923,
      "step": 103004
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.9002083539962769,
      "learning_rate": 0.0003496608947671626,
      "loss": 2.1046,
      "step": 103005
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.9660166501998901,
      "learning_rate": 0.00034965686063501757,
      "loss": 2.2165,
      "step": 103006
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.9794668555259705,
      "learning_rate": 0.00034965282649364045,
      "loss": 2.1708,
      "step": 103007
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.8205257058143616,
      "learning_rate": 0.0003496487923430319,
      "loss": 2.2377,
      "step": 103008
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.8892875909805298,
      "learning_rate": 0.00034964475818319274,
      "loss": 2.1263,
      "step": 103009
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.9352273941040039,
      "learning_rate": 0.00034964072401412363,
      "loss": 2.1644,
      "step": 103010
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.8825114369392395,
      "learning_rate": 0.00034963668983582536,
      "loss": 1.8797,
      "step": 103011
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.8515563607215881,
      "learning_rate": 0.0003496326556482988,
      "loss": 2.0339,
      "step": 103012
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.920672595500946,
      "learning_rate": 0.00034962862145154446,
      "loss": 2.078,
      "step": 103013
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.8700534701347351,
      "learning_rate": 0.00034962458724556325,
      "loss": 2.1616,
      "step": 103014
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.85968017578125,
      "learning_rate": 0.000349620553030356,
      "loss": 1.9553,
      "step": 103015
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.8286287188529968,
      "learning_rate": 0.0003496165188059232,
      "loss": 2.1741,
      "step": 103016
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.9047691822052002,
      "learning_rate": 0.00034961248457226584,
      "loss": 2.2003,
      "step": 103017
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.8391762971878052,
      "learning_rate": 0.0003496084503293846,
      "loss": 2.0485,
      "step": 103018
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.9252243638038635,
      "learning_rate": 0.00034960441607728016,
      "loss": 2.2481,
      "step": 103019
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.9085294604301453,
      "learning_rate": 0.00034960038181595335,
      "loss": 2.2105,
      "step": 103020
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.8913161158561707,
      "learning_rate": 0.0003495963475454049,
      "loss": 1.9718,
      "step": 103021
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.9320980310440063,
      "learning_rate": 0.00034959231326563557,
      "loss": 2.0295,
      "step": 103022
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.820499062538147,
      "learning_rate": 0.000349588278976646,
      "loss": 1.9457,
      "step": 103023
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.8531608581542969,
      "learning_rate": 0.0003495842446784372,
      "loss": 2.0905,
      "step": 103024
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.7715494632720947,
      "learning_rate": 0.00034958021037100956,
      "loss": 1.9994,
      "step": 103025
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.953229546546936,
      "learning_rate": 0.0003495761760543641,
      "loss": 2.1695,
      "step": 103026
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.8643356561660767,
      "learning_rate": 0.0003495721417285016,
      "loss": 2.2414,
      "step": 103027
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.894968569278717,
      "learning_rate": 0.0003495681073934226,
      "loss": 2.1638,
      "step": 103028
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.9392814636230469,
      "learning_rate": 0.0003495640730491279,
      "loss": 2.5004,
      "step": 103029
    },
    {
      "epoch": 1.34,
      "grad_norm": 1.0269839763641357,
      "learning_rate": 0.00034956003869561844,
      "loss": 2.252,
      "step": 103030
    },
    {
      "epoch": 1.34,
      "grad_norm": 1.1015647649765015,
      "learning_rate": 0.0003495560043328947,
      "loss": 2.2282,
      "step": 103031
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.7854104042053223,
      "learning_rate": 0.0003495519699609576,
      "loss": 2.2736,
      "step": 103032
    },
    {
      "epoch": 1.34,
      "grad_norm": 1.1214429140090942,
      "learning_rate": 0.0003495479355798079,
      "loss": 2.3577,
      "step": 103033
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.9053740501403809,
      "learning_rate": 0.00034954390118944627,
      "loss": 2.3485,
      "step": 103034
    },
    {
      "epoch": 1.34,
      "grad_norm": 1.0042117834091187,
      "learning_rate": 0.0003495398667898735,
      "loss": 2.0777,
      "step": 103035
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.9224932193756104,
      "learning_rate": 0.00034953583238109037,
      "loss": 1.9946,
      "step": 103036
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.9268734455108643,
      "learning_rate": 0.0003495317979630975,
      "loss": 2.2391,
      "step": 103037
    },
    {
      "epoch": 1.34,
      "grad_norm": 1.3890057802200317,
      "learning_rate": 0.0003495277635358958,
      "loss": 2.0857,
      "step": 103038
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.9424539804458618,
      "learning_rate": 0.000349523729099486,
      "loss": 2.2079,
      "step": 103039
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.8316209316253662,
      "learning_rate": 0.00034951969465386873,
      "loss": 2.1898,
      "step": 103040
    },
    {
      "epoch": 1.34,
      "grad_norm": 1.034820318222046,
      "learning_rate": 0.00034951566019904485,
      "loss": 2.3331,
      "step": 103041
    },
    {
      "epoch": 1.34,
      "grad_norm": 1.3510373830795288,
      "learning_rate": 0.00034951162573501507,
      "loss": 2.0855,
      "step": 103042
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.9078734517097473,
      "learning_rate": 0.0003495075912617801,
      "loss": 2.0215,
      "step": 103043
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.8894177675247192,
      "learning_rate": 0.00034950355677934076,
      "loss": 2.0151,
      "step": 103044
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.9356296062469482,
      "learning_rate": 0.00034949952228769787,
      "loss": 2.1354,
      "step": 103045
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.8461335897445679,
      "learning_rate": 0.00034949548778685194,
      "loss": 1.9484,
      "step": 103046
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.8471173644065857,
      "learning_rate": 0.0003494914532768039,
      "loss": 2.1108,
      "step": 103047
    },
    {
      "epoch": 1.34,
      "grad_norm": 1.0789204835891724,
      "learning_rate": 0.0003494874187575545,
      "loss": 2.2671,
      "step": 103048
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.8878984451293945,
      "learning_rate": 0.00034948338422910447,
      "loss": 2.2316,
      "step": 103049
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.9390659332275391,
      "learning_rate": 0.00034947934969145443,
      "loss": 2.2085,
      "step": 103050
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.851604163646698,
      "learning_rate": 0.0003494753151446054,
      "loss": 2.0656,
      "step": 103051
    },
    {
      "epoch": 1.34,
      "grad_norm": 1.3043842315673828,
      "learning_rate": 0.00034947128058855794,
      "loss": 2.0732,
      "step": 103052
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.8644875884056091,
      "learning_rate": 0.00034946724602331273,
      "loss": 1.8926,
      "step": 103053
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.9077752232551575,
      "learning_rate": 0.0003494632114488707,
      "loss": 2.2045,
      "step": 103054
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.8945568799972534,
      "learning_rate": 0.00034945917686523257,
      "loss": 1.8977,
      "step": 103055
    },
    {
      "epoch": 1.34,
      "grad_norm": 1.1621884107589722,
      "learning_rate": 0.0003494551422723989,
      "loss": 2.3625,
      "step": 103056
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.9181365370750427,
      "learning_rate": 0.0003494511076703707,
      "loss": 2.1231,
      "step": 103057
    },
    {
      "epoch": 1.34,
      "grad_norm": 1.0026793479919434,
      "learning_rate": 0.00034944707305914865,
      "loss": 2.0964,
      "step": 103058
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.9912410378456116,
      "learning_rate": 0.0003494430384387333,
      "loss": 2.1543,
      "step": 103059
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.8067406415939331,
      "learning_rate": 0.00034943900380912565,
      "loss": 1.9751,
      "step": 103060
    },
    {
      "epoch": 1.34,
      "grad_norm": 1.01056969165802,
      "learning_rate": 0.0003494349691703265,
      "loss": 2.257,
      "step": 103061
    },
    {
      "epoch": 1.34,
      "grad_norm": 1.0397977828979492,
      "learning_rate": 0.0003494309345223362,
      "loss": 2.0546,
      "step": 103062
    },
    {
      "epoch": 1.34,
      "grad_norm": 1.0763798952102661,
      "learning_rate": 0.00034942689986515587,
      "loss": 2.1942,
      "step": 103063
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.9614147543907166,
      "learning_rate": 0.00034942286519878615,
      "loss": 2.4233,
      "step": 103064
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.9108810424804688,
      "learning_rate": 0.00034941883052322777,
      "loss": 2.2409,
      "step": 103065
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.9357238411903381,
      "learning_rate": 0.0003494147958384815,
      "loss": 2.386,
      "step": 103066
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.8845564126968384,
      "learning_rate": 0.00034941076114454814,
      "loss": 2.0596,
      "step": 103067
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.966621994972229,
      "learning_rate": 0.00034940672644142825,
      "loss": 2.1224,
      "step": 103068
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.8417696356773376,
      "learning_rate": 0.00034940269172912283,
      "loss": 2.2307,
      "step": 103069
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.905060350894928,
      "learning_rate": 0.00034939865700763255,
      "loss": 2.0652,
      "step": 103070
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.9919875264167786,
      "learning_rate": 0.000349394622276958,
      "loss": 2.1596,
      "step": 103071
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.8582249879837036,
      "learning_rate": 0.0003493905875371002,
      "loss": 2.2018,
      "step": 103072
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.959936797618866,
      "learning_rate": 0.0003493865527880596,
      "loss": 2.27,
      "step": 103073
    },
    {
      "epoch": 1.34,
      "grad_norm": 1.1077195405960083,
      "learning_rate": 0.00034938251802983726,
      "loss": 2.1606,
      "step": 103074
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.9968662261962891,
      "learning_rate": 0.00034937848326243367,
      "loss": 2.3308,
      "step": 103075
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.8956586718559265,
      "learning_rate": 0.00034937444848584976,
      "loss": 2.2392,
      "step": 103076
    },
    {
      "epoch": 1.34,
      "grad_norm": 1.0383325815200806,
      "learning_rate": 0.0003493704137000861,
      "loss": 2.0232,
      "step": 103077
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.9283466339111328,
      "learning_rate": 0.00034936637890514363,
      "loss": 2.34,
      "step": 103078
    },
    {
      "epoch": 1.34,
      "grad_norm": 1.0641239881515503,
      "learning_rate": 0.0003493623441010231,
      "loss": 2.2828,
      "step": 103079
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.9495524764060974,
      "learning_rate": 0.000349358309287725,
      "loss": 2.0586,
      "step": 103080
    },
    {
      "epoch": 1.34,
      "grad_norm": 1.104996681213379,
      "learning_rate": 0.00034935427446525035,
      "loss": 2.0699,
      "step": 103081
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.947677493095398,
      "learning_rate": 0.0003493502396335999,
      "loss": 2.2074,
      "step": 103082
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.916838526725769,
      "learning_rate": 0.00034934620479277415,
      "loss": 2.2678,
      "step": 103083
    },
    {
      "epoch": 1.34,
      "grad_norm": 1.0107417106628418,
      "learning_rate": 0.0003493421699427741,
      "loss": 2.1198,
      "step": 103084
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.8916004300117493,
      "learning_rate": 0.00034933813508360043,
      "loss": 2.271,
      "step": 103085
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.8789266347885132,
      "learning_rate": 0.00034933410021525377,
      "loss": 2.1254,
      "step": 103086
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.8848817348480225,
      "learning_rate": 0.000349330065337735,
      "loss": 2.1699,
      "step": 103087
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.8830703496932983,
      "learning_rate": 0.0003493260304510449,
      "loss": 2.1492,
      "step": 103088
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.8048170208930969,
      "learning_rate": 0.0003493219955551842,
      "loss": 2.1607,
      "step": 103089
    },
    {
      "epoch": 1.34,
      "grad_norm": 1.0310585498809814,
      "learning_rate": 0.0003493179606501535,
      "loss": 2.2541,
      "step": 103090
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.9640181064605713,
      "learning_rate": 0.00034931392573595374,
      "loss": 2.1533,
      "step": 103091
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.9748937487602234,
      "learning_rate": 0.0003493098908125855,
      "loss": 2.2907,
      "step": 103092
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.871679425239563,
      "learning_rate": 0.0003493058558800496,
      "loss": 2.1407,
      "step": 103093
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.8567557334899902,
      "learning_rate": 0.0003493018209383469,
      "loss": 2.1005,
      "step": 103094
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.7922907471656799,
      "learning_rate": 0.000349297785987478,
      "loss": 2.2218,
      "step": 103095
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.9471791386604309,
      "learning_rate": 0.00034929375102744377,
      "loss": 2.1144,
      "step": 103096
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.9851934313774109,
      "learning_rate": 0.0003492897160582449,
      "loss": 2.0503,
      "step": 103097
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.9485320448875427,
      "learning_rate": 0.0003492856810798821,
      "loss": 2.2013,
      "step": 103098
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.8707631826400757,
      "learning_rate": 0.00034928164609235616,
      "loss": 1.986,
      "step": 103099
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.9147183895111084,
      "learning_rate": 0.00034927761109566793,
      "loss": 2.1349,
      "step": 103100
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.9019205570220947,
      "learning_rate": 0.00034927357608981793,
      "loss": 2.1247,
      "step": 103101
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.8187074661254883,
      "learning_rate": 0.0003492695410748071,
      "loss": 1.9664,
      "step": 103102
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.943662166595459,
      "learning_rate": 0.00034926550605063615,
      "loss": 2.301,
      "step": 103103
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.9342747926712036,
      "learning_rate": 0.00034926147101730585,
      "loss": 2.2976,
      "step": 103104
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.875152587890625,
      "learning_rate": 0.00034925743597481675,
      "loss": 2.047,
      "step": 103105
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.869873583316803,
      "learning_rate": 0.00034925340092316993,
      "loss": 2.1354,
      "step": 103106
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.9362471103668213,
      "learning_rate": 0.00034924936586236595,
      "loss": 2.212,
      "step": 103107
    },
    {
      "epoch": 1.34,
      "grad_norm": 1.4110485315322876,
      "learning_rate": 0.0003492453307924055,
      "loss": 2.0338,
      "step": 103108
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.8943987488746643,
      "learning_rate": 0.0003492412957132895,
      "loss": 2.0737,
      "step": 103109
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.9384899139404297,
      "learning_rate": 0.00034923726062501855,
      "loss": 2.198,
      "step": 103110
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.9196416139602661,
      "learning_rate": 0.0003492332255275934,
      "loss": 2.2479,
      "step": 103111
    },
    {
      "epoch": 1.34,
      "grad_norm": 1.0500210523605347,
      "learning_rate": 0.000349229190421015,
      "loss": 2.2645,
      "step": 103112
    },
    {
      "epoch": 1.34,
      "grad_norm": 1.0657638311386108,
      "learning_rate": 0.0003492251553052839,
      "loss": 2.1847,
      "step": 103113
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.8634571433067322,
      "learning_rate": 0.000349221120180401,
      "loss": 2.367,
      "step": 103114
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.9525700807571411,
      "learning_rate": 0.0003492170850463669,
      "loss": 1.8889,
      "step": 103115
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.8838956952095032,
      "learning_rate": 0.00034921304990318235,
      "loss": 2.3585,
      "step": 103116
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.8890100717544556,
      "learning_rate": 0.00034920901475084823,
      "loss": 2.3292,
      "step": 103117
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.9019689559936523,
      "learning_rate": 0.0003492049795893653,
      "loss": 2.1675,
      "step": 103118
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.9136479496955872,
      "learning_rate": 0.0003492009444187341,
      "loss": 2.2067,
      "step": 103119
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.9654284119606018,
      "learning_rate": 0.0003491969092389555,
      "loss": 1.9659,
      "step": 103120
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.8688579797744751,
      "learning_rate": 0.00034919287405003045,
      "loss": 1.9209,
      "step": 103121
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.8281576633453369,
      "learning_rate": 0.00034918883885195936,
      "loss": 2.1895,
      "step": 103122
    },
    {
      "epoch": 1.34,
      "grad_norm": 1.2343796491622925,
      "learning_rate": 0.00034918480364474314,
      "loss": 2.4574,
      "step": 103123
    },
    {
      "epoch": 1.34,
      "grad_norm": 1.0854837894439697,
      "learning_rate": 0.0003491807684283826,
      "loss": 2.0417,
      "step": 103124
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.9789657592773438,
      "learning_rate": 0.0003491767332028784,
      "loss": 2.0357,
      "step": 103125
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.8855815529823303,
      "learning_rate": 0.0003491726979682313,
      "loss": 2.2081,
      "step": 103126
    },
    {
      "epoch": 1.34,
      "grad_norm": 1.4202113151550293,
      "learning_rate": 0.00034916866272444207,
      "loss": 2.0042,
      "step": 103127
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.8778662085533142,
      "learning_rate": 0.00034916462747151155,
      "loss": 2.1667,
      "step": 103128
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.9487211108207703,
      "learning_rate": 0.00034916059220944023,
      "loss": 1.9663,
      "step": 103129
    },
    {
      "epoch": 1.34,
      "grad_norm": 1.0608021020889282,
      "learning_rate": 0.00034915655693822914,
      "loss": 2.1826,
      "step": 103130
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.9863710403442383,
      "learning_rate": 0.00034915252165787893,
      "loss": 2.1574,
      "step": 103131
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.8526144027709961,
      "learning_rate": 0.0003491484863683903,
      "loss": 2.0861,
      "step": 103132
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.9550017714500427,
      "learning_rate": 0.00034914445106976397,
      "loss": 2.1407,
      "step": 103133
    },
    {
      "epoch": 1.34,
      "grad_norm": 1.1938648223876953,
      "learning_rate": 0.0003491404157620009,
      "loss": 2.1135,
      "step": 103134
    },
    {
      "epoch": 1.34,
      "grad_norm": 1.0705511569976807,
      "learning_rate": 0.00034913638044510164,
      "loss": 2.1621,
      "step": 103135
    },
    {
      "epoch": 1.34,
      "grad_norm": 1.0752363204956055,
      "learning_rate": 0.00034913234511906694,
      "loss": 2.3003,
      "step": 103136
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.9876445531845093,
      "learning_rate": 0.0003491283097838977,
      "loss": 2.3296,
      "step": 103137
    },
    {
      "epoch": 1.34,
      "grad_norm": 1.2539947032928467,
      "learning_rate": 0.00034912427443959456,
      "loss": 2.1331,
      "step": 103138
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.9801250696182251,
      "learning_rate": 0.0003491202390861582,
      "loss": 2.0272,
      "step": 103139
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.9229834079742432,
      "learning_rate": 0.0003491162037235896,
      "loss": 2.2349,
      "step": 103140
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.9712006449699402,
      "learning_rate": 0.0003491121683518893,
      "loss": 2.2562,
      "step": 103141
    },
    {
      "epoch": 1.34,
      "grad_norm": 1.093379259109497,
      "learning_rate": 0.0003491081329710581,
      "loss": 2.1158,
      "step": 103142
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.9621391892433167,
      "learning_rate": 0.0003491040975810968,
      "loss": 2.0636,
      "step": 103143
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.9075879454612732,
      "learning_rate": 0.0003491000621820061,
      "loss": 2.2914,
      "step": 103144
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.9858332872390747,
      "learning_rate": 0.00034909602677378677,
      "loss": 2.3012,
      "step": 103145
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.9246906638145447,
      "learning_rate": 0.00034909199135643964,
      "loss": 2.1388,
      "step": 103146
    },
    {
      "epoch": 1.34,
      "grad_norm": 1.0158579349517822,
      "learning_rate": 0.00034908795592996527,
      "loss": 2.2027,
      "step": 103147
    },
    {
      "epoch": 1.34,
      "grad_norm": 1.0579380989074707,
      "learning_rate": 0.0003490839204943646,
      "loss": 2.0478,
      "step": 103148
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.8221401572227478,
      "learning_rate": 0.0003490798850496382,
      "loss": 1.8928,
      "step": 103149
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.9091869592666626,
      "learning_rate": 0.0003490758495957871,
      "loss": 2.1461,
      "step": 103150
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.9290849566459656,
      "learning_rate": 0.00034907181413281174,
      "loss": 2.1528,
      "step": 103151
    },
    {
      "epoch": 1.34,
      "grad_norm": 1.0547711849212646,
      "learning_rate": 0.000349067778660713,
      "loss": 2.462,
      "step": 103152
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.8518182039260864,
      "learning_rate": 0.00034906374317949173,
      "loss": 2.0768,
      "step": 103153
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.8198959827423096,
      "learning_rate": 0.0003490597076891484,
      "loss": 2.2728,
      "step": 103154
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.9072535037994385,
      "learning_rate": 0.0003490556721896841,
      "loss": 2.214,
      "step": 103155
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.923984706401825,
      "learning_rate": 0.0003490516366810994,
      "loss": 2.1849,
      "step": 103156
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.8129146695137024,
      "learning_rate": 0.00034904760116339506,
      "loss": 2.0189,
      "step": 103157
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.852383017539978,
      "learning_rate": 0.0003490435656365719,
      "loss": 2.1603,
      "step": 103158
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.9082701206207275,
      "learning_rate": 0.0003490395301006305,
      "loss": 2.0673,
      "step": 103159
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.7964521646499634,
      "learning_rate": 0.0003490354945555718,
      "loss": 2.1796,
      "step": 103160
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.8247721195220947,
      "learning_rate": 0.0003490314590013965,
      "loss": 2.2209,
      "step": 103161
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.8232585787773132,
      "learning_rate": 0.00034902742343810524,
      "loss": 2.3631,
      "step": 103162
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.8649235367774963,
      "learning_rate": 0.00034902338786569893,
      "loss": 2.2425,
      "step": 103163
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.8633543252944946,
      "learning_rate": 0.00034901935228417823,
      "loss": 2.106,
      "step": 103164
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.8668312430381775,
      "learning_rate": 0.00034901531669354394,
      "loss": 2.1896,
      "step": 103165
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.8912299871444702,
      "learning_rate": 0.00034901128109379666,
      "loss": 2.0956,
      "step": 103166
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.8768914341926575,
      "learning_rate": 0.0003490072454849374,
      "loss": 2.2051,
      "step": 103167
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.9463256597518921,
      "learning_rate": 0.0003490032098669667,
      "loss": 2.1172,
      "step": 103168
    },
    {
      "epoch": 1.34,
      "grad_norm": 1.0323225259780884,
      "learning_rate": 0.00034899917423988533,
      "loss": 2.1472,
      "step": 103169
    },
    {
      "epoch": 1.34,
      "grad_norm": 1.0214128494262695,
      "learning_rate": 0.0003489951386036943,
      "loss": 2.3449,
      "step": 103170
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.8767509460449219,
      "learning_rate": 0.0003489911029583939,
      "loss": 2.2228,
      "step": 103171
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.918976902961731,
      "learning_rate": 0.0003489870673039852,
      "loss": 2.0795,
      "step": 103172
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.9615650177001953,
      "learning_rate": 0.000348983031640469,
      "loss": 2.1172,
      "step": 103173
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.9636361002922058,
      "learning_rate": 0.00034897899596784574,
      "loss": 2.2406,
      "step": 103174
    },
    {
      "epoch": 1.34,
      "grad_norm": 1.1312110424041748,
      "learning_rate": 0.0003489749602861164,
      "loss": 2.1613,
      "step": 103175
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.8977580666542053,
      "learning_rate": 0.00034897092459528184,
      "loss": 2.0944,
      "step": 103176
    },
    {
      "epoch": 1.34,
      "grad_norm": 1.0495904684066772,
      "learning_rate": 0.00034896688889534257,
      "loss": 2.1226,
      "step": 103177
    },
    {
      "epoch": 1.34,
      "grad_norm": 1.0632834434509277,
      "learning_rate": 0.0003489628531862994,
      "loss": 2.232,
      "step": 103178
    },
    {
      "epoch": 1.34,
      "grad_norm": 1.1169013977050781,
      "learning_rate": 0.00034895881746815313,
      "loss": 2.1182,
      "step": 103179
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.9914416074752808,
      "learning_rate": 0.0003489547817409045,
      "loss": 2.2286,
      "step": 103180
    },
    {
      "epoch": 1.34,
      "grad_norm": 1.574413776397705,
      "learning_rate": 0.00034895074600455424,
      "loss": 2.0774,
      "step": 103181
    },
    {
      "epoch": 1.34,
      "grad_norm": 1.26980459690094,
      "learning_rate": 0.00034894671025910315,
      "loss": 1.9452,
      "step": 103182
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.9043335914611816,
      "learning_rate": 0.00034894267450455186,
      "loss": 2.0263,
      "step": 103183
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.9648139476776123,
      "learning_rate": 0.0003489386387409013,
      "loss": 2.3633,
      "step": 103184
    },
    {
      "epoch": 1.34,
      "grad_norm": 1.318762183189392,
      "learning_rate": 0.0003489346029681521,
      "loss": 1.8831,
      "step": 103185
    },
    {
      "epoch": 1.34,
      "grad_norm": 1.4176249504089355,
      "learning_rate": 0.000348930567186305,
      "loss": 2.2883,
      "step": 103186
    },
    {
      "epoch": 1.34,
      "grad_norm": 1.2047675848007202,
      "learning_rate": 0.0003489265313953607,
      "loss": 1.9588,
      "step": 103187
    },
    {
      "epoch": 1.34,
      "grad_norm": 1.1112784147262573,
      "learning_rate": 0.00034892249559532025,
      "loss": 2.2349,
      "step": 103188
    },
    {
      "epoch": 1.34,
      "grad_norm": 1.2452598810195923,
      "learning_rate": 0.00034891845978618404,
      "loss": 2.3423,
      "step": 103189
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.890909731388092,
      "learning_rate": 0.00034891442396795295,
      "loss": 2.1337,
      "step": 103190
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.8906123042106628,
      "learning_rate": 0.00034891038814062773,
      "loss": 2.1153,
      "step": 103191
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.8846955299377441,
      "learning_rate": 0.0003489063523042093,
      "loss": 1.9277,
      "step": 103192
    },
    {
      "epoch": 1.34,
      "grad_norm": 1.4177738428115845,
      "learning_rate": 0.0003489023164586981,
      "loss": 2.064,
      "step": 103193
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.9085233211517334,
      "learning_rate": 0.00034889828060409513,
      "loss": 2.0333,
      "step": 103194
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.8924965262413025,
      "learning_rate": 0.000348894244740401,
      "loss": 2.1799,
      "step": 103195
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.9681503176689148,
      "learning_rate": 0.00034889020886761654,
      "loss": 2.3235,
      "step": 103196
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.8710439801216125,
      "learning_rate": 0.0003488861729857424,
      "loss": 2.273,
      "step": 103197
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.8835867643356323,
      "learning_rate": 0.00034888213709477955,
      "loss": 2.2965,
      "step": 103198
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.976947009563446,
      "learning_rate": 0.00034887810119472844,
      "loss": 1.9555,
      "step": 103199
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.9215591549873352,
      "learning_rate": 0.0003488740652855899,
      "loss": 2.2606,
      "step": 103200
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.8860144019126892,
      "learning_rate": 0.00034887002936736497,
      "loss": 2.2144,
      "step": 103201
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.8459317684173584,
      "learning_rate": 0.00034886599344005403,
      "loss": 2.3029,
      "step": 103202
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.9820109009742737,
      "learning_rate": 0.000348861957503658,
      "loss": 2.1561,
      "step": 103203
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.9515042901039124,
      "learning_rate": 0.00034885792155817764,
      "loss": 2.1051,
      "step": 103204
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.9429589509963989,
      "learning_rate": 0.00034885388560361365,
      "loss": 2.2057,
      "step": 103205
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.9130643010139465,
      "learning_rate": 0.0003488498496399668,
      "loss": 2.3034,
      "step": 103206
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.9192810654640198,
      "learning_rate": 0.0003488458136672379,
      "loss": 2.0473,
      "step": 103207
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.8886831998825073,
      "learning_rate": 0.0003488417776854276,
      "loss": 2.2067,
      "step": 103208
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.8443422913551331,
      "learning_rate": 0.00034883774169453663,
      "loss": 2.3861,
      "step": 103209
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.976099967956543,
      "learning_rate": 0.0003488337056945659,
      "loss": 2.1611,
      "step": 103210
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.8972024917602539,
      "learning_rate": 0.00034882966968551596,
      "loss": 2.3592,
      "step": 103211
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.9616914391517639,
      "learning_rate": 0.0003488256336673877,
      "loss": 2.1829,
      "step": 103212
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.8509105443954468,
      "learning_rate": 0.00034882159764018194,
      "loss": 1.998,
      "step": 103213
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.902850866317749,
      "learning_rate": 0.00034881756160389915,
      "loss": 2.1762,
      "step": 103214
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.9868054986000061,
      "learning_rate": 0.0003488135255585404,
      "loss": 2.3882,
      "step": 103215
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.8344311118125916,
      "learning_rate": 0.00034880948950410625,
      "loss": 2.3774,
      "step": 103216
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.8437107801437378,
      "learning_rate": 0.00034880545344059745,
      "loss": 2.26,
      "step": 103217
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.8484984636306763,
      "learning_rate": 0.0003488014173680148,
      "loss": 2.2764,
      "step": 103218
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.9110132455825806,
      "learning_rate": 0.0003487973812863591,
      "loss": 2.1337,
      "step": 103219
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.9266926050186157,
      "learning_rate": 0.00034879334519563105,
      "loss": 2.0442,
      "step": 103220
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.8975201845169067,
      "learning_rate": 0.0003487893090958313,
      "loss": 2.1343,
      "step": 103221
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.9157931804656982,
      "learning_rate": 0.0003487852729869608,
      "loss": 2.1428,
      "step": 103222
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.9489758610725403,
      "learning_rate": 0.0003487812368690201,
      "loss": 2.2476,
      "step": 103223
    },
    {
      "epoch": 1.34,
      "grad_norm": 1.0043810606002808,
      "learning_rate": 0.00034877720074201013,
      "loss": 1.9939,
      "step": 103224
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.8844403028488159,
      "learning_rate": 0.00034877316460593156,
      "loss": 2.3088,
      "step": 103225
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.8439474701881409,
      "learning_rate": 0.00034876912846078505,
      "loss": 2.0853,
      "step": 103226
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.8871162533760071,
      "learning_rate": 0.00034876509230657147,
      "loss": 2.1712,
      "step": 103227
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.8882266283035278,
      "learning_rate": 0.0003487610561432916,
      "loss": 2.2271,
      "step": 103228
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.7850232124328613,
      "learning_rate": 0.00034875701997094606,
      "loss": 2.1921,
      "step": 103229
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.8718945384025574,
      "learning_rate": 0.0003487529837895357,
      "loss": 2.1197,
      "step": 103230
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.8756312131881714,
      "learning_rate": 0.00034874894759906125,
      "loss": 2.1485,
      "step": 103231
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.8896723389625549,
      "learning_rate": 0.00034874491139952346,
      "loss": 2.1705,
      "step": 103232
    },
    {
      "epoch": 1.34,
      "grad_norm": 1.0402358770370483,
      "learning_rate": 0.000348740875190923,
      "loss": 2.3727,
      "step": 103233
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.9162307977676392,
      "learning_rate": 0.0003487368389732608,
      "loss": 2.0356,
      "step": 103234
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.9708837866783142,
      "learning_rate": 0.00034873280274653746,
      "loss": 1.9242,
      "step": 103235
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.8012188076972961,
      "learning_rate": 0.0003487287665107537,
      "loss": 1.9791,
      "step": 103236
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.8219633102416992,
      "learning_rate": 0.00034872473026591037,
      "loss": 2.0294,
      "step": 103237
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.9605114459991455,
      "learning_rate": 0.00034872069401200824,
      "loss": 1.9823,
      "step": 103238
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.9122166633605957,
      "learning_rate": 0.000348716657749048,
      "loss": 2.0742,
      "step": 103239
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.8469257354736328,
      "learning_rate": 0.00034871262147703037,
      "loss": 2.1422,
      "step": 103240
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.9468416571617126,
      "learning_rate": 0.0003487085851959563,
      "loss": 2.1396,
      "step": 103241
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.8796488642692566,
      "learning_rate": 0.0003487045489058262,
      "loss": 1.983,
      "step": 103242
    },
    {
      "epoch": 1.34,
      "grad_norm": 1.003887414932251,
      "learning_rate": 0.00034870051260664107,
      "loss": 2.0948,
      "step": 103243
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.8970571160316467,
      "learning_rate": 0.00034869647629840155,
      "loss": 2.1083,
      "step": 103244
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.9140486121177673,
      "learning_rate": 0.00034869243998110857,
      "loss": 2.2231,
      "step": 103245
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.7841572165489197,
      "learning_rate": 0.0003486884036547626,
      "loss": 2.2133,
      "step": 103246
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.9600726366043091,
      "learning_rate": 0.00034868436731936463,
      "loss": 2.0274,
      "step": 103247
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.8142293691635132,
      "learning_rate": 0.0003486803309749152,
      "loss": 1.8847,
      "step": 103248
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.9677190184593201,
      "learning_rate": 0.0003486762946214153,
      "loss": 2.0909,
      "step": 103249
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.8962714672088623,
      "learning_rate": 0.00034867225825886553,
      "loss": 2.2733,
      "step": 103250
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.8312480449676514,
      "learning_rate": 0.0003486682218872667,
      "loss": 2.1316,
      "step": 103251
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.9541851878166199,
      "learning_rate": 0.0003486641855066195,
      "loss": 2.12,
      "step": 103252
    },
    {
      "epoch": 1.34,
      "grad_norm": 1.0169427394866943,
      "learning_rate": 0.0003486601491169247,
      "loss": 2.1246,
      "step": 103253
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.906627357006073,
      "learning_rate": 0.00034865611271818296,
      "loss": 2.1362,
      "step": 103254
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.8955191373825073,
      "learning_rate": 0.0003486520763103953,
      "loss": 2.2468,
      "step": 103255
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.8072725534439087,
      "learning_rate": 0.00034864803989356215,
      "loss": 2.0404,
      "step": 103256
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.9173694849014282,
      "learning_rate": 0.00034864400346768453,
      "loss": 2.1906,
      "step": 103257
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.8412432074546814,
      "learning_rate": 0.000348639967032763,
      "loss": 2.3559,
      "step": 103258
    },
    {
      "epoch": 1.34,
      "grad_norm": 1.0584150552749634,
      "learning_rate": 0.0003486359305887985,
      "loss": 2.001,
      "step": 103259
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.8354697227478027,
      "learning_rate": 0.00034863189413579147,
      "loss": 2.0146,
      "step": 103260
    },
    {
      "epoch": 1.34,
      "grad_norm": 1.0016080141067505,
      "learning_rate": 0.00034862785767374296,
      "loss": 2.34,
      "step": 103261
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.8849844932556152,
      "learning_rate": 0.00034862382120265373,
      "loss": 2.1256,
      "step": 103262
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.9258372783660889,
      "learning_rate": 0.00034861978472252426,
      "loss": 2.182,
      "step": 103263
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.810822606086731,
      "learning_rate": 0.00034861574823335546,
      "loss": 2.259,
      "step": 103264
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.9286839962005615,
      "learning_rate": 0.0003486117117351481,
      "loss": 2.2692,
      "step": 103265
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.9746461510658264,
      "learning_rate": 0.0003486076752279029,
      "loss": 2.2247,
      "step": 103266
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.84715336561203,
      "learning_rate": 0.0003486036387116206,
      "loss": 2.3015,
      "step": 103267
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.9103923439979553,
      "learning_rate": 0.000348599602186302,
      "loss": 2.0502,
      "step": 103268
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.9152241349220276,
      "learning_rate": 0.00034859556565194785,
      "loss": 2.0936,
      "step": 103269
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.9685128927230835,
      "learning_rate": 0.00034859152910855876,
      "loss": 2.2339,
      "step": 103270
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.8689389824867249,
      "learning_rate": 0.0003485874925561357,
      "loss": 2.0227,
      "step": 103271
    },
    {
      "epoch": 1.34,
      "grad_norm": 1.0403889417648315,
      "learning_rate": 0.00034858345599467926,
      "loss": 2.2072,
      "step": 103272
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.9499880075454712,
      "learning_rate": 0.00034857941942419016,
      "loss": 2.0764,
      "step": 103273
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.8456888198852539,
      "learning_rate": 0.00034857538284466943,
      "loss": 2.1999,
      "step": 103274
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.8945415616035461,
      "learning_rate": 0.00034857134625611747,
      "loss": 2.2874,
      "step": 103275
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.93083256483078,
      "learning_rate": 0.0003485673096585351,
      "loss": 2.0131,
      "step": 103276
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.94835364818573,
      "learning_rate": 0.0003485632730519234,
      "loss": 2.229,
      "step": 103277
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.923868715763092,
      "learning_rate": 0.00034855923643628263,
      "loss": 2.0104,
      "step": 103278
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.9802767634391785,
      "learning_rate": 0.0003485551998116139,
      "loss": 2.2623,
      "step": 103279
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.8720517158508301,
      "learning_rate": 0.00034855116317791785,
      "loss": 1.9707,
      "step": 103280
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.8328245878219604,
      "learning_rate": 0.0003485471265351952,
      "loss": 1.9672,
      "step": 103281
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.987478494644165,
      "learning_rate": 0.00034854308988344665,
      "loss": 2.2273,
      "step": 103282
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.8663038015365601,
      "learning_rate": 0.00034853905322267315,
      "loss": 2.007,
      "step": 103283
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.8286756873130798,
      "learning_rate": 0.00034853501655287525,
      "loss": 2.0508,
      "step": 103284
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.7647543549537659,
      "learning_rate": 0.0003485309798740538,
      "loss": 2.0941,
      "step": 103285
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.974001944065094,
      "learning_rate": 0.0003485269431862095,
      "loss": 2.2292,
      "step": 103286
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.8559262752532959,
      "learning_rate": 0.00034852290648934315,
      "loss": 2.009,
      "step": 103287
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.8360581398010254,
      "learning_rate": 0.0003485188697834554,
      "loss": 2.1888,
      "step": 103288
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.8155883550643921,
      "learning_rate": 0.0003485148330685472,
      "loss": 2.0854,
      "step": 103289
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.9592740535736084,
      "learning_rate": 0.00034851079634461905,
      "loss": 2.1877,
      "step": 103290
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.8554089069366455,
      "learning_rate": 0.00034850675961167196,
      "loss": 2.0399,
      "step": 103291
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.8404315710067749,
      "learning_rate": 0.00034850272286970647,
      "loss": 2.2249,
      "step": 103292
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.8000892996788025,
      "learning_rate": 0.0003484986861187234,
      "loss": 1.9971,
      "step": 103293
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.8874762058258057,
      "learning_rate": 0.0003484946493587235,
      "loss": 2.1839,
      "step": 103294
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.8790642023086548,
      "learning_rate": 0.00034849061258970766,
      "loss": 2.2127,
      "step": 103295
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.8932701945304871,
      "learning_rate": 0.00034848657581167635,
      "loss": 2.2712,
      "step": 103296
    },
    {
      "epoch": 1.34,
      "grad_norm": 0.8994302749633789,
      "learning_rate": 0.0003484825390246305,
      "loss": 2.1998,
      "step": 103297
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.8775629997253418,
      "learning_rate": 0.0003484785022285709,
      "loss": 2.0255,
      "step": 103298
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.9447400569915771,
      "learning_rate": 0.0003484744654234981,
      "loss": 2.2897,
      "step": 103299
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.9007098078727722,
      "learning_rate": 0.0003484704286094131,
      "loss": 2.0153,
      "step": 103300
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.8256339430809021,
      "learning_rate": 0.0003484663917863165,
      "loss": 2.0286,
      "step": 103301
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.9063467979431152,
      "learning_rate": 0.00034846235495420905,
      "loss": 1.9286,
      "step": 103302
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.9538320899009705,
      "learning_rate": 0.00034845831811309154,
      "loss": 2.0532,
      "step": 103303
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.8793887495994568,
      "learning_rate": 0.0003484542812629648,
      "loss": 2.1884,
      "step": 103304
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.8005654215812683,
      "learning_rate": 0.0003484502444038294,
      "loss": 2.0662,
      "step": 103305
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.9136793613433838,
      "learning_rate": 0.00034844620753568615,
      "loss": 2.1415,
      "step": 103306
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.8265224695205688,
      "learning_rate": 0.00034844217065853594,
      "loss": 2.1855,
      "step": 103307
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.8657368421554565,
      "learning_rate": 0.0003484381337723794,
      "loss": 2.2525,
      "step": 103308
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.9834380149841309,
      "learning_rate": 0.00034843409687721725,
      "loss": 2.3762,
      "step": 103309
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.848639190196991,
      "learning_rate": 0.00034843005997305034,
      "loss": 1.8613,
      "step": 103310
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.935355544090271,
      "learning_rate": 0.00034842602305987935,
      "loss": 2.1217,
      "step": 103311
    },
    {
      "epoch": 1.35,
      "grad_norm": 1.0125473737716675,
      "learning_rate": 0.000348421986137705,
      "loss": 2.1866,
      "step": 103312
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.9093877077102661,
      "learning_rate": 0.0003484179492065281,
      "loss": 2.3364,
      "step": 103313
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.8845562934875488,
      "learning_rate": 0.00034841391226634943,
      "loss": 2.2125,
      "step": 103314
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.8639451265335083,
      "learning_rate": 0.0003484098753171697,
      "loss": 2.1306,
      "step": 103315
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.9795364737510681,
      "learning_rate": 0.0003484058383589897,
      "loss": 1.922,
      "step": 103316
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.9007606506347656,
      "learning_rate": 0.00034840180139181004,
      "loss": 2.1527,
      "step": 103317
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.9208512306213379,
      "learning_rate": 0.00034839776441563156,
      "loss": 2.2456,
      "step": 103318
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.8850675225257874,
      "learning_rate": 0.00034839372743045513,
      "loss": 1.9765,
      "step": 103319
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.9098014831542969,
      "learning_rate": 0.00034838969043628134,
      "loss": 2.2946,
      "step": 103320
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.8000900149345398,
      "learning_rate": 0.00034838565343311094,
      "loss": 2.0004,
      "step": 103321
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.9358575344085693,
      "learning_rate": 0.00034838161642094487,
      "loss": 2.1361,
      "step": 103322
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.9615339636802673,
      "learning_rate": 0.00034837757939978366,
      "loss": 2.2361,
      "step": 103323
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.9223779439926147,
      "learning_rate": 0.00034837354236962807,
      "loss": 2.1718,
      "step": 103324
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.8507043123245239,
      "learning_rate": 0.00034836950533047897,
      "loss": 2.0128,
      "step": 103325
    },
    {
      "epoch": 1.35,
      "grad_norm": 1.0037366151809692,
      "learning_rate": 0.00034836546828233717,
      "loss": 2.3607,
      "step": 103326
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.9513846635818481,
      "learning_rate": 0.0003483614312252032,
      "loss": 1.9416,
      "step": 103327
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.9616134166717529,
      "learning_rate": 0.000348357394159078,
      "loss": 2.0864,
      "step": 103328
    },
    {
      "epoch": 1.35,
      "grad_norm": 1.3947391510009766,
      "learning_rate": 0.0003483533570839622,
      "loss": 1.9295,
      "step": 103329
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.9875310063362122,
      "learning_rate": 0.00034834931999985666,
      "loss": 2.3267,
      "step": 103330
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.8739559650421143,
      "learning_rate": 0.00034834528290676194,
      "loss": 1.8527,
      "step": 103331
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.832054615020752,
      "learning_rate": 0.00034834124580467897,
      "loss": 2.2292,
      "step": 103332
    },
    {
      "epoch": 1.35,
      "grad_norm": 1.2298952341079712,
      "learning_rate": 0.00034833720869360856,
      "loss": 2.097,
      "step": 103333
    },
    {
      "epoch": 1.35,
      "grad_norm": 1.1653505563735962,
      "learning_rate": 0.00034833317157355125,
      "loss": 2.3064,
      "step": 103334
    },
    {
      "epoch": 1.35,
      "grad_norm": 1.0133863687515259,
      "learning_rate": 0.0003483291344445079,
      "loss": 2.4496,
      "step": 103335
    },
    {
      "epoch": 1.35,
      "grad_norm": 1.1945785284042358,
      "learning_rate": 0.0003483250973064792,
      "loss": 2.2506,
      "step": 103336
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.955282986164093,
      "learning_rate": 0.00034832106015946613,
      "loss": 2.134,
      "step": 103337
    },
    {
      "epoch": 1.35,
      "grad_norm": 1.2091600894927979,
      "learning_rate": 0.0003483170230034691,
      "loss": 2.2468,
      "step": 103338
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.9552040696144104,
      "learning_rate": 0.00034831298583848906,
      "loss": 2.2015,
      "step": 103339
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.8625777959823608,
      "learning_rate": 0.0003483089486645267,
      "loss": 2.1376,
      "step": 103340
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.9977741837501526,
      "learning_rate": 0.0003483049114815829,
      "loss": 2.1585,
      "step": 103341
    },
    {
      "epoch": 1.35,
      "grad_norm": 1.356732726097107,
      "learning_rate": 0.00034830087428965826,
      "loss": 1.9422,
      "step": 103342
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.8174633979797363,
      "learning_rate": 0.00034829683708875346,
      "loss": 2.1158,
      "step": 103343
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.8772875070571899,
      "learning_rate": 0.0003482927998788695,
      "loss": 2.2973,
      "step": 103344
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.825252890586853,
      "learning_rate": 0.00034828876266000696,
      "loss": 1.8997,
      "step": 103345
    },
    {
      "epoch": 1.35,
      "grad_norm": 1.1005973815917969,
      "learning_rate": 0.0003482847254321666,
      "loss": 2.2393,
      "step": 103346
    },
    {
      "epoch": 1.35,
      "grad_norm": 1.0015482902526855,
      "learning_rate": 0.0003482806881953493,
      "loss": 1.9311,
      "step": 103347
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.9376789927482605,
      "learning_rate": 0.0003482766509495556,
      "loss": 2.1183,
      "step": 103348
    },
    {
      "epoch": 1.35,
      "grad_norm": 1.0110942125320435,
      "learning_rate": 0.0003482726136947864,
      "loss": 2.1187,
      "step": 103349
    },
    {
      "epoch": 1.35,
      "grad_norm": 1.443152904510498,
      "learning_rate": 0.0003482685764310424,
      "loss": 2.0995,
      "step": 103350
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.881215512752533,
      "learning_rate": 0.0003482645391583243,
      "loss": 2.3409,
      "step": 103351
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.8596259951591492,
      "learning_rate": 0.00034826050187663295,
      "loss": 2.0876,
      "step": 103352
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.980323076248169,
      "learning_rate": 0.00034825646458596916,
      "loss": 2.2239,
      "step": 103353
    },
    {
      "epoch": 1.35,
      "grad_norm": 1.0152312517166138,
      "learning_rate": 0.0003482524272863335,
      "loss": 2.1214,
      "step": 103354
    },
    {
      "epoch": 1.35,
      "grad_norm": 1.041562795639038,
      "learning_rate": 0.0003482483899777268,
      "loss": 2.3326,
      "step": 103355
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.9055113196372986,
      "learning_rate": 0.0003482443526601499,
      "loss": 2.4031,
      "step": 103356
    },
    {
      "epoch": 1.35,
      "grad_norm": 1.0255916118621826,
      "learning_rate": 0.00034824031533360334,
      "loss": 2.2542,
      "step": 103357
    },
    {
      "epoch": 1.35,
      "grad_norm": 1.005145788192749,
      "learning_rate": 0.00034823627799808796,
      "loss": 2.0252,
      "step": 103358
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.8556785583496094,
      "learning_rate": 0.0003482322406536047,
      "loss": 2.3032,
      "step": 103359
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.8905066847801208,
      "learning_rate": 0.00034822820330015407,
      "loss": 2.2729,
      "step": 103360
    },
    {
      "epoch": 1.35,
      "grad_norm": 1.2398934364318848,
      "learning_rate": 0.0003482241659377369,
      "loss": 2.096,
      "step": 103361
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.9875058531761169,
      "learning_rate": 0.000348220128566354,
      "loss": 2.1956,
      "step": 103362
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.8676436543464661,
      "learning_rate": 0.00034821609118600605,
      "loss": 2.0258,
      "step": 103363
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.9579089879989624,
      "learning_rate": 0.00034821205379669375,
      "loss": 2.2328,
      "step": 103364
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.9542371034622192,
      "learning_rate": 0.000348208016398418,
      "loss": 2.151,
      "step": 103365
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.8551306128501892,
      "learning_rate": 0.00034820397899117946,
      "loss": 2.001,
      "step": 103366
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.9138990044593811,
      "learning_rate": 0.00034819994157497884,
      "loss": 2.251,
      "step": 103367
    },
    {
      "epoch": 1.35,
      "grad_norm": 1.0738087892532349,
      "learning_rate": 0.00034819590414981706,
      "loss": 2.1476,
      "step": 103368
    },
    {
      "epoch": 1.35,
      "grad_norm": 1.0498558282852173,
      "learning_rate": 0.00034819186671569463,
      "loss": 2.2413,
      "step": 103369
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.9005071520805359,
      "learning_rate": 0.0003481878292726124,
      "loss": 2.2849,
      "step": 103370
    },
    {
      "epoch": 1.35,
      "grad_norm": 1.0025354623794556,
      "learning_rate": 0.00034818379182057125,
      "loss": 2.1871,
      "step": 103371
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.8310536742210388,
      "learning_rate": 0.00034817975435957177,
      "loss": 1.8605,
      "step": 103372
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.8697198629379272,
      "learning_rate": 0.0003481757168896148,
      "loss": 2.1087,
      "step": 103373
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.8435718417167664,
      "learning_rate": 0.000348171679410701,
      "loss": 1.9913,
      "step": 103374
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.8066931962966919,
      "learning_rate": 0.0003481676419228312,
      "loss": 2.3646,
      "step": 103375
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.973810613155365,
      "learning_rate": 0.00034816360442600607,
      "loss": 2.2309,
      "step": 103376
    },
    {
      "epoch": 1.35,
      "grad_norm": 1.0245803594589233,
      "learning_rate": 0.00034815956692022654,
      "loss": 2.0251,
      "step": 103377
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.7933157086372375,
      "learning_rate": 0.00034815552940549313,
      "loss": 2.2197,
      "step": 103378
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.895913302898407,
      "learning_rate": 0.00034815149188180676,
      "loss": 2.2393,
      "step": 103379
    },
    {
      "epoch": 1.35,
      "grad_norm": 1.0016440153121948,
      "learning_rate": 0.00034814745434916814,
      "loss": 1.994,
      "step": 103380
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.9072571396827698,
      "learning_rate": 0.00034814341680757796,
      "loss": 2.2535,
      "step": 103381
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.9899271726608276,
      "learning_rate": 0.000348139379257037,
      "loss": 2.3042,
      "step": 103382
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.879023015499115,
      "learning_rate": 0.00034813534169754603,
      "loss": 1.9996,
      "step": 103383
    },
    {
      "epoch": 1.35,
      "grad_norm": 1.0071567296981812,
      "learning_rate": 0.00034813130412910575,
      "loss": 2.1985,
      "step": 103384
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.9084800481796265,
      "learning_rate": 0.00034812726655171695,
      "loss": 2.2905,
      "step": 103385
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.852176308631897,
      "learning_rate": 0.0003481232289653805,
      "loss": 1.988,
      "step": 103386
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.9386837482452393,
      "learning_rate": 0.0003481191913700969,
      "loss": 2.1425,
      "step": 103387
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.9527920484542847,
      "learning_rate": 0.0003481151537658671,
      "loss": 2.0746,
      "step": 103388
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.9209485650062561,
      "learning_rate": 0.0003481111161526918,
      "loss": 2.2799,
      "step": 103389
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.8645557165145874,
      "learning_rate": 0.0003481070785305717,
      "loss": 1.7875,
      "step": 103390
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.8634777665138245,
      "learning_rate": 0.0003481030408995075,
      "loss": 2.0944,
      "step": 103391
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.9507015347480774,
      "learning_rate": 0.0003480990032595002,
      "loss": 2.3742,
      "step": 103392
    },
    {
      "epoch": 1.35,
      "grad_norm": 1.0694787502288818,
      "learning_rate": 0.00034809496561055036,
      "loss": 2.1855,
      "step": 103393
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.8823683857917786,
      "learning_rate": 0.0003480909279526586,
      "loss": 2.1964,
      "step": 103394
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.9459847807884216,
      "learning_rate": 0.000348086890285826,
      "loss": 2.1492,
      "step": 103395
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.9401680827140808,
      "learning_rate": 0.00034808285261005314,
      "loss": 2.1064,
      "step": 103396
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.9510693550109863,
      "learning_rate": 0.0003480788149253407,
      "loss": 2.1733,
      "step": 103397
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.8319426774978638,
      "learning_rate": 0.0003480747772316895,
      "loss": 1.9452,
      "step": 103398
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.997104287147522,
      "learning_rate": 0.0003480707395291003,
      "loss": 2.1255,
      "step": 103399
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.9388498663902283,
      "learning_rate": 0.00034806670181757385,
      "loss": 2.3353,
      "step": 103400
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.8604276776313782,
      "learning_rate": 0.00034806266409711086,
      "loss": 2.009,
      "step": 103401
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.8662512302398682,
      "learning_rate": 0.00034805862636771225,
      "loss": 1.9435,
      "step": 103402
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.9019715189933777,
      "learning_rate": 0.00034805458862937846,
      "loss": 2.1717,
      "step": 103403
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.905756413936615,
      "learning_rate": 0.00034805055088211047,
      "loss": 1.988,
      "step": 103404
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.8759602308273315,
      "learning_rate": 0.000348046513125909,
      "loss": 2.0582,
      "step": 103405
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.8685587644577026,
      "learning_rate": 0.0003480424753607748,
      "loss": 2.3195,
      "step": 103406
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.872869610786438,
      "learning_rate": 0.0003480384375867085,
      "loss": 2.3035,
      "step": 103407
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.8332251906394958,
      "learning_rate": 0.00034803439980371113,
      "loss": 2.1808,
      "step": 103408
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.8959951400756836,
      "learning_rate": 0.0003480303620117831,
      "loss": 2.1931,
      "step": 103409
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.8932310342788696,
      "learning_rate": 0.0003480263242109253,
      "loss": 2.2392,
      "step": 103410
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.849998414516449,
      "learning_rate": 0.0003480222864011386,
      "loss": 2.2167,
      "step": 103411
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.8405364751815796,
      "learning_rate": 0.0003480182485824236,
      "loss": 2.2479,
      "step": 103412
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.9606748223304749,
      "learning_rate": 0.0003480142107547811,
      "loss": 2.0015,
      "step": 103413
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.9266317486763,
      "learning_rate": 0.0003480101729182119,
      "loss": 2.1539,
      "step": 103414
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.9356921315193176,
      "learning_rate": 0.00034800613507271667,
      "loss": 2.1614,
      "step": 103415
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.9448626637458801,
      "learning_rate": 0.00034800209721829616,
      "loss": 1.9874,
      "step": 103416
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.8621374368667603,
      "learning_rate": 0.00034799805935495115,
      "loss": 2.0436,
      "step": 103417
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.8718190789222717,
      "learning_rate": 0.00034799402148268246,
      "loss": 1.8468,
      "step": 103418
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.8829830288887024,
      "learning_rate": 0.0003479899836014907,
      "loss": 2.1332,
      "step": 103419
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.7953304648399353,
      "learning_rate": 0.0003479859457113767,
      "loss": 2.1456,
      "step": 103420
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.8902776837348938,
      "learning_rate": 0.0003479819078123413,
      "loss": 2.14,
      "step": 103421
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.8350642919540405,
      "learning_rate": 0.00034797786990438514,
      "loss": 2.4054,
      "step": 103422
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.87034672498703,
      "learning_rate": 0.0003479738319875089,
      "loss": 2.176,
      "step": 103423
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.8605735301971436,
      "learning_rate": 0.00034796979406171347,
      "loss": 2.0686,
      "step": 103424
    },
    {
      "epoch": 1.35,
      "grad_norm": 1.055072546005249,
      "learning_rate": 0.0003479657561269996,
      "loss": 2.3256,
      "step": 103425
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.9196012616157532,
      "learning_rate": 0.00034796171818336796,
      "loss": 2.1009,
      "step": 103426
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.8890871405601501,
      "learning_rate": 0.0003479576802308193,
      "loss": 2.1919,
      "step": 103427
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.8449400663375854,
      "learning_rate": 0.0003479536422693543,
      "loss": 2.3848,
      "step": 103428
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.7871021628379822,
      "learning_rate": 0.000347949604298974,
      "loss": 2.096,
      "step": 103429
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.9620193839073181,
      "learning_rate": 0.00034794556631967887,
      "loss": 2.2305,
      "step": 103430
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.8541322946548462,
      "learning_rate": 0.0003479415283314697,
      "loss": 2.305,
      "step": 103431
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.9564256072044373,
      "learning_rate": 0.00034793749033434744,
      "loss": 2.097,
      "step": 103432
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.9150901436805725,
      "learning_rate": 0.00034793345232831264,
      "loss": 2.0121,
      "step": 103433
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.8449050784111023,
      "learning_rate": 0.0003479294143133661,
      "loss": 2.2187,
      "step": 103434
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.9616733193397522,
      "learning_rate": 0.0003479253762895086,
      "loss": 2.1126,
      "step": 103435
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.9247040748596191,
      "learning_rate": 0.00034792133825674076,
      "loss": 2.3172,
      "step": 103436
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.8069385290145874,
      "learning_rate": 0.0003479173002150635,
      "loss": 2.4304,
      "step": 103437
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.933665931224823,
      "learning_rate": 0.00034791326216447756,
      "loss": 2.4306,
      "step": 103438
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.8498672246932983,
      "learning_rate": 0.00034790922410498357,
      "loss": 1.8341,
      "step": 103439
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.9068117737770081,
      "learning_rate": 0.0003479051860365824,
      "loss": 2.2422,
      "step": 103440
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.8429556488990784,
      "learning_rate": 0.0003479011479592748,
      "loss": 1.9751,
      "step": 103441
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.8255571126937866,
      "learning_rate": 0.0003478971098730614,
      "loss": 2.2523,
      "step": 103442
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.9123071432113647,
      "learning_rate": 0.000347893071777943,
      "loss": 2.1958,
      "step": 103443
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.920024573802948,
      "learning_rate": 0.00034788903367392047,
      "loss": 2.3504,
      "step": 103444
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.9356433153152466,
      "learning_rate": 0.00034788499556099434,
      "loss": 2.2171,
      "step": 103445
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.8229392170906067,
      "learning_rate": 0.00034788095743916556,
      "loss": 2.2549,
      "step": 103446
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.934619665145874,
      "learning_rate": 0.0003478769193084348,
      "loss": 2.0217,
      "step": 103447
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.8789925575256348,
      "learning_rate": 0.00034787288116880285,
      "loss": 2.0229,
      "step": 103448
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.8705059885978699,
      "learning_rate": 0.00034786884302027033,
      "loss": 2.2724,
      "step": 103449
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.9584152102470398,
      "learning_rate": 0.0003478648048628382,
      "loss": 2.1681,
      "step": 103450
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.869496762752533,
      "learning_rate": 0.000347860766696507,
      "loss": 2.1286,
      "step": 103451
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.9340983629226685,
      "learning_rate": 0.00034785672852127766,
      "loss": 2.2362,
      "step": 103452
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.9534547924995422,
      "learning_rate": 0.00034785269033715084,
      "loss": 1.937,
      "step": 103453
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.9191527366638184,
      "learning_rate": 0.00034784865214412727,
      "loss": 2.0583,
      "step": 103454
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.9554520845413208,
      "learning_rate": 0.00034784461394220766,
      "loss": 2.0808,
      "step": 103455
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.854337751865387,
      "learning_rate": 0.000347840575731393,
      "loss": 2.0539,
      "step": 103456
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.8776764869689941,
      "learning_rate": 0.0003478365375116838,
      "loss": 2.2875,
      "step": 103457
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.8996947407722473,
      "learning_rate": 0.00034783249928308077,
      "loss": 2.2028,
      "step": 103458
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.9463124871253967,
      "learning_rate": 0.00034782846104558485,
      "loss": 2.3554,
      "step": 103459
    },
    {
      "epoch": 1.35,
      "grad_norm": 1.1930369138717651,
      "learning_rate": 0.00034782442279919685,
      "loss": 2.1817,
      "step": 103460
    },
    {
      "epoch": 1.35,
      "grad_norm": 1.1227874755859375,
      "learning_rate": 0.0003478203845439172,
      "loss": 2.3065,
      "step": 103461
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.8864834904670715,
      "learning_rate": 0.00034781634627974693,
      "loss": 2.0283,
      "step": 103462
    },
    {
      "epoch": 1.35,
      "grad_norm": 1.2449216842651367,
      "learning_rate": 0.00034781230800668675,
      "loss": 2.3845,
      "step": 103463
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.9727108478546143,
      "learning_rate": 0.00034780826972473724,
      "loss": 2.2207,
      "step": 103464
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.9246826171875,
      "learning_rate": 0.0003478042314338993,
      "loss": 2.0987,
      "step": 103465
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.8125324845314026,
      "learning_rate": 0.00034780019313417374,
      "loss": 2.1873,
      "step": 103466
    },
    {
      "epoch": 1.35,
      "grad_norm": 1.2196099758148193,
      "learning_rate": 0.00034779615482556116,
      "loss": 2.3988,
      "step": 103467
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.9220503568649292,
      "learning_rate": 0.0003477921165080623,
      "loss": 2.1745,
      "step": 103468
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.8598064184188843,
      "learning_rate": 0.00034778807818167817,
      "loss": 2.3726,
      "step": 103469
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.9132322669029236,
      "learning_rate": 0.0003477840398464092,
      "loss": 2.2298,
      "step": 103470
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.8849971294403076,
      "learning_rate": 0.0003477800015022562,
      "loss": 2.2419,
      "step": 103471
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.882515549659729,
      "learning_rate": 0.0003477759631492202,
      "loss": 2.4217,
      "step": 103472
    },
    {
      "epoch": 1.35,
      "grad_norm": 1.0243929624557495,
      "learning_rate": 0.00034777192478730156,
      "loss": 2.1393,
      "step": 103473
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.8876596689224243,
      "learning_rate": 0.0003477678864165013,
      "loss": 2.0519,
      "step": 103474
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.8414279818534851,
      "learning_rate": 0.00034776384803682014,
      "loss": 2.0488,
      "step": 103475
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.8198544979095459,
      "learning_rate": 0.00034775980964825865,
      "loss": 2.0627,
      "step": 103476
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.9753032326698303,
      "learning_rate": 0.0003477557712508178,
      "loss": 2.2938,
      "step": 103477
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.835110068321228,
      "learning_rate": 0.0003477517328444983,
      "loss": 2.0511,
      "step": 103478
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.8632134795188904,
      "learning_rate": 0.00034774769442930077,
      "loss": 2.0567,
      "step": 103479
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.8160306811332703,
      "learning_rate": 0.000347743656005226,
      "loss": 2.1476,
      "step": 103480
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.8859197497367859,
      "learning_rate": 0.0003477396175722749,
      "loss": 2.0875,
      "step": 103481
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.8682297468185425,
      "learning_rate": 0.00034773557913044803,
      "loss": 2.1828,
      "step": 103482
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.8768282532691956,
      "learning_rate": 0.00034773154067974614,
      "loss": 2.2537,
      "step": 103483
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.8083635568618774,
      "learning_rate": 0.0003477275022201702,
      "loss": 2.1737,
      "step": 103484
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.9232274293899536,
      "learning_rate": 0.00034772346375172076,
      "loss": 2.16,
      "step": 103485
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.8586903810501099,
      "learning_rate": 0.0003477194252743986,
      "loss": 2.2857,
      "step": 103486
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.8287203907966614,
      "learning_rate": 0.0003477153867882046,
      "loss": 2.5044,
      "step": 103487
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.9013432860374451,
      "learning_rate": 0.0003477113482931393,
      "loss": 2.313,
      "step": 103488
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.9568458199501038,
      "learning_rate": 0.00034770730978920356,
      "loss": 2.2075,
      "step": 103489
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.8565291166305542,
      "learning_rate": 0.0003477032712763982,
      "loss": 2.2984,
      "step": 103490
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.8842243552207947,
      "learning_rate": 0.0003476992327547239,
      "loss": 2.1328,
      "step": 103491
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.8373056054115295,
      "learning_rate": 0.00034769519422418136,
      "loss": 2.2011,
      "step": 103492
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.8245093822479248,
      "learning_rate": 0.0003476911556847714,
      "loss": 1.9027,
      "step": 103493
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.8071956634521484,
      "learning_rate": 0.00034768711713649474,
      "loss": 1.9247,
      "step": 103494
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.8201264142990112,
      "learning_rate": 0.00034768307857935217,
      "loss": 2.0257,
      "step": 103495
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.9148543477058411,
      "learning_rate": 0.0003476790400133444,
      "loss": 2.1932,
      "step": 103496
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.8423961997032166,
      "learning_rate": 0.0003476750014384722,
      "loss": 1.9317,
      "step": 103497
    },
    {
      "epoch": 1.35,
      "grad_norm": 1.028202772140503,
      "learning_rate": 0.00034767096285473633,
      "loss": 2.0087,
      "step": 103498
    },
    {
      "epoch": 1.35,
      "grad_norm": 1.0116304159164429,
      "learning_rate": 0.00034766692426213753,
      "loss": 2.0574,
      "step": 103499
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.8980385065078735,
      "learning_rate": 0.0003476628856606765,
      "loss": 2.0463,
      "step": 103500
    },
    {
      "epoch": 1.35,
      "grad_norm": 1.2857836484909058,
      "learning_rate": 0.0003476588470503541,
      "loss": 2.0681,
      "step": 103501
    },
    {
      "epoch": 1.35,
      "grad_norm": 1.1278371810913086,
      "learning_rate": 0.00034765480843117094,
      "loss": 2.0493,
      "step": 103502
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.9553531408309937,
      "learning_rate": 0.00034765076980312795,
      "loss": 2.2903,
      "step": 103503
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.9229039549827576,
      "learning_rate": 0.0003476467311662257,
      "loss": 2.1615,
      "step": 103504
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.9433224201202393,
      "learning_rate": 0.00034764269252046513,
      "loss": 2.0586,
      "step": 103505
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.9264320731163025,
      "learning_rate": 0.0003476386538658468,
      "loss": 2.13,
      "step": 103506
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.832877516746521,
      "learning_rate": 0.0003476346152023716,
      "loss": 2.2259,
      "step": 103507
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.8703519701957703,
      "learning_rate": 0.0003476305765300401,
      "loss": 2.0377,
      "step": 103508
    },
    {
      "epoch": 1.35,
      "grad_norm": 1.0388838052749634,
      "learning_rate": 0.00034762653784885326,
      "loss": 2.0958,
      "step": 103509
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.9530155062675476,
      "learning_rate": 0.0003476224991588118,
      "loss": 2.1154,
      "step": 103510
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.8396946787834167,
      "learning_rate": 0.00034761846045991635,
      "loss": 2.1398,
      "step": 103511
    },
    {
      "epoch": 1.35,
      "grad_norm": 1.0472862720489502,
      "learning_rate": 0.00034761442175216775,
      "loss": 1.9708,
      "step": 103512
    },
    {
      "epoch": 1.35,
      "grad_norm": 1.1089956760406494,
      "learning_rate": 0.00034761038303556674,
      "loss": 2.1844,
      "step": 103513
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.8348288536071777,
      "learning_rate": 0.00034760634431011404,
      "loss": 2.1351,
      "step": 103514
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.8339268565177917,
      "learning_rate": 0.00034760230557581045,
      "loss": 2.0937,
      "step": 103515
    },
    {
      "epoch": 1.35,
      "grad_norm": 1.5700123310089111,
      "learning_rate": 0.0003475982668326566,
      "loss": 1.9681,
      "step": 103516
    },
    {
      "epoch": 1.35,
      "grad_norm": 1.097590684890747,
      "learning_rate": 0.0003475942280806535,
      "loss": 2.0081,
      "step": 103517
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.892026960849762,
      "learning_rate": 0.00034759018931980164,
      "loss": 2.5575,
      "step": 103518
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.8777492046356201,
      "learning_rate": 0.00034758615055010183,
      "loss": 2.1593,
      "step": 103519
    },
    {
      "epoch": 1.35,
      "grad_norm": 1.168967366218567,
      "learning_rate": 0.00034758211177155494,
      "loss": 1.922,
      "step": 103520
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.9953716993331909,
      "learning_rate": 0.00034757807298416155,
      "loss": 2.2165,
      "step": 103521
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.8623983263969421,
      "learning_rate": 0.00034757403418792253,
      "loss": 2.2315,
      "step": 103522
    },
    {
      "epoch": 1.35,
      "grad_norm": 1.0238394737243652,
      "learning_rate": 0.0003475699953828387,
      "loss": 2.2385,
      "step": 103523
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.8481125831604004,
      "learning_rate": 0.0003475659565689106,
      "loss": 2.1775,
      "step": 103524
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.9410504698753357,
      "learning_rate": 0.0003475619177461391,
      "loss": 2.2457,
      "step": 103525
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.9431092143058777,
      "learning_rate": 0.0003475578789145249,
      "loss": 2.2695,
      "step": 103526
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.8463686108589172,
      "learning_rate": 0.00034755384007406886,
      "loss": 2.0038,
      "step": 103527
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.9766293168067932,
      "learning_rate": 0.00034754980122477164,
      "loss": 2.1188,
      "step": 103528
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.9475608468055725,
      "learning_rate": 0.00034754576236663406,
      "loss": 1.9161,
      "step": 103529
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.8761089444160461,
      "learning_rate": 0.00034754172349965676,
      "loss": 2.1494,
      "step": 103530
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.8508656024932861,
      "learning_rate": 0.0003475376846238406,
      "loss": 2.3739,
      "step": 103531
    },
    {
      "epoch": 1.35,
      "grad_norm": 1.2007602453231812,
      "learning_rate": 0.00034753364573918627,
      "loss": 1.9786,
      "step": 103532
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.9311692714691162,
      "learning_rate": 0.00034752960684569453,
      "loss": 2.0251,
      "step": 103533
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.8553751111030579,
      "learning_rate": 0.00034752556794336617,
      "loss": 2.1211,
      "step": 103534
    },
    {
      "epoch": 1.35,
      "grad_norm": 1.2664680480957031,
      "learning_rate": 0.00034752152903220183,
      "loss": 2.1035,
      "step": 103535
    },
    {
      "epoch": 1.35,
      "grad_norm": 1.2069426774978638,
      "learning_rate": 0.0003475174901122025,
      "loss": 2.1403,
      "step": 103536
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.965136706829071,
      "learning_rate": 0.00034751345118336864,
      "loss": 2.1061,
      "step": 103537
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.9594425559043884,
      "learning_rate": 0.0003475094122457012,
      "loss": 1.9616,
      "step": 103538
    },
    {
      "epoch": 1.35,
      "grad_norm": 1.365220308303833,
      "learning_rate": 0.0003475053732992008,
      "loss": 1.9792,
      "step": 103539
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.9801514148712158,
      "learning_rate": 0.00034750133434386824,
      "loss": 2.0168,
      "step": 103540
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.8856872320175171,
      "learning_rate": 0.00034749729537970436,
      "loss": 2.3504,
      "step": 103541
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.9911265969276428,
      "learning_rate": 0.0003474932564067098,
      "loss": 2.1712,
      "step": 103542
    },
    {
      "epoch": 1.35,
      "grad_norm": 1.4483119249343872,
      "learning_rate": 0.00034748921742488537,
      "loss": 2.0339,
      "step": 103543
    },
    {
      "epoch": 1.35,
      "grad_norm": 1.1726058721542358,
      "learning_rate": 0.00034748517843423177,
      "loss": 1.9502,
      "step": 103544
    },
    {
      "epoch": 1.35,
      "grad_norm": 1.0207127332687378,
      "learning_rate": 0.00034748113943474984,
      "loss": 1.958,
      "step": 103545
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.9703654646873474,
      "learning_rate": 0.0003474771004264401,
      "loss": 2.0055,
      "step": 103546
    },
    {
      "epoch": 1.35,
      "grad_norm": 1.0437040328979492,
      "learning_rate": 0.0003474730614093036,
      "loss": 2.2865,
      "step": 103547
    },
    {
      "epoch": 1.35,
      "grad_norm": 1.1090830564498901,
      "learning_rate": 0.00034746902238334105,
      "loss": 2.0352,
      "step": 103548
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.845348060131073,
      "learning_rate": 0.00034746498334855294,
      "loss": 1.9699,
      "step": 103549
    },
    {
      "epoch": 1.35,
      "grad_norm": 1.0881896018981934,
      "learning_rate": 0.00034746094430494024,
      "loss": 2.0672,
      "step": 103550
    },
    {
      "epoch": 1.35,
      "grad_norm": 1.0452001094818115,
      "learning_rate": 0.0003474569052525037,
      "loss": 2.1953,
      "step": 103551
    },
    {
      "epoch": 1.35,
      "grad_norm": 1.0265095233917236,
      "learning_rate": 0.000347452866191244,
      "loss": 2.1174,
      "step": 103552
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.8724562525749207,
      "learning_rate": 0.00034744882712116196,
      "loss": 2.2166,
      "step": 103553
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.9067151546478271,
      "learning_rate": 0.00034744478804225827,
      "loss": 2.4455,
      "step": 103554
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.8937398195266724,
      "learning_rate": 0.00034744074895453367,
      "loss": 2.2307,
      "step": 103555
    },
    {
      "epoch": 1.35,
      "grad_norm": 1.0780525207519531,
      "learning_rate": 0.0003474367098579889,
      "loss": 2.2311,
      "step": 103556
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.8822110295295715,
      "learning_rate": 0.0003474326707526249,
      "loss": 2.1481,
      "step": 103557
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.9710730314254761,
      "learning_rate": 0.0003474286316384421,
      "loss": 2.226,
      "step": 103558
    },
    {
      "epoch": 1.35,
      "grad_norm": 1.0060955286026,
      "learning_rate": 0.0003474245925154415,
      "loss": 2.2113,
      "step": 103559
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.9916970729827881,
      "learning_rate": 0.0003474205533836238,
      "loss": 2.0546,
      "step": 103560
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.8371286392211914,
      "learning_rate": 0.00034741651424298963,
      "loss": 2.1675,
      "step": 103561
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.992867112159729,
      "learning_rate": 0.00034741247509353994,
      "loss": 2.1515,
      "step": 103562
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.9169800877571106,
      "learning_rate": 0.00034740843593527535,
      "loss": 2.084,
      "step": 103563
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.9322173595428467,
      "learning_rate": 0.0003474043967681966,
      "loss": 2.1394,
      "step": 103564
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.9649112224578857,
      "learning_rate": 0.00034740035759230444,
      "loss": 2.036,
      "step": 103565
    },
    {
      "epoch": 1.35,
      "grad_norm": 1.2031934261322021,
      "learning_rate": 0.0003473963184075998,
      "loss": 2.4261,
      "step": 103566
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.8378291130065918,
      "learning_rate": 0.00034739227921408314,
      "loss": 2.1323,
      "step": 103567
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.9570276737213135,
      "learning_rate": 0.0003473882400117554,
      "loss": 2.1255,
      "step": 103568
    },
    {
      "epoch": 1.35,
      "grad_norm": 1.2352330684661865,
      "learning_rate": 0.00034738420080061735,
      "loss": 1.9915,
      "step": 103569
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.9445509910583496,
      "learning_rate": 0.00034738016158066963,
      "loss": 2.1186,
      "step": 103570
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.9961945414543152,
      "learning_rate": 0.00034737612235191304,
      "loss": 2.0261,
      "step": 103571
    },
    {
      "epoch": 1.35,
      "grad_norm": 1.0076767206192017,
      "learning_rate": 0.00034737208311434845,
      "loss": 2.0798,
      "step": 103572
    },
    {
      "epoch": 1.35,
      "grad_norm": 1.0643768310546875,
      "learning_rate": 0.0003473680438679764,
      "loss": 2.1477,
      "step": 103573
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.9043801426887512,
      "learning_rate": 0.00034736400461279774,
      "loss": 2.2607,
      "step": 103574
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.8167315721511841,
      "learning_rate": 0.00034735996534881325,
      "loss": 2.1713,
      "step": 103575
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.9440408945083618,
      "learning_rate": 0.0003473559260760235,
      "loss": 2.171,
      "step": 103576
    },
    {
      "epoch": 1.35,
      "grad_norm": 1.3370157480239868,
      "learning_rate": 0.00034735188679442953,
      "loss": 1.9973,
      "step": 103577
    },
    {
      "epoch": 1.35,
      "grad_norm": 1.0408318042755127,
      "learning_rate": 0.0003473478475040319,
      "loss": 2.1591,
      "step": 103578
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.9189195036888123,
      "learning_rate": 0.0003473438082048315,
      "loss": 2.1817,
      "step": 103579
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.9619342684745789,
      "learning_rate": 0.00034733976889682886,
      "loss": 1.9573,
      "step": 103580
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.8871842622756958,
      "learning_rate": 0.00034733572958002497,
      "loss": 2.1254,
      "step": 103581
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.9180060029029846,
      "learning_rate": 0.00034733169025442045,
      "loss": 2.1355,
      "step": 103582
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.9506019353866577,
      "learning_rate": 0.000347327650920016,
      "loss": 2.1,
      "step": 103583
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.9608646631240845,
      "learning_rate": 0.00034732361157681254,
      "loss": 2.4027,
      "step": 103584
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.9487802386283875,
      "learning_rate": 0.00034731957222481073,
      "loss": 2.0715,
      "step": 103585
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.8886263370513916,
      "learning_rate": 0.0003473155328640112,
      "loss": 2.28,
      "step": 103586
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.9198957085609436,
      "learning_rate": 0.00034731149349441503,
      "loss": 1.949,
      "step": 103587
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.9347777962684631,
      "learning_rate": 0.0003473074541160226,
      "loss": 2.3025,
      "step": 103588
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.8294476270675659,
      "learning_rate": 0.00034730341472883486,
      "loss": 2.0989,
      "step": 103589
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.8584650158882141,
      "learning_rate": 0.0003472993753328525,
      "loss": 2.0706,
      "step": 103590
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.9498084187507629,
      "learning_rate": 0.0003472953359280763,
      "loss": 2.3684,
      "step": 103591
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.8984650373458862,
      "learning_rate": 0.00034729129651450696,
      "loss": 2.2548,
      "step": 103592
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.9457340836524963,
      "learning_rate": 0.0003472872570921453,
      "loss": 1.9448,
      "step": 103593
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.8458304405212402,
      "learning_rate": 0.0003472832176609921,
      "loss": 2.2178,
      "step": 103594
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.8910511136054993,
      "learning_rate": 0.0003472791782210481,
      "loss": 2.3055,
      "step": 103595
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.8258482217788696,
      "learning_rate": 0.0003472751387723139,
      "loss": 2.2076,
      "step": 103596
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.8860763907432556,
      "learning_rate": 0.00034727109931479035,
      "loss": 2.296,
      "step": 103597
    },
    {
      "epoch": 1.35,
      "grad_norm": 1.0969709157943726,
      "learning_rate": 0.0003472670598484783,
      "loss": 1.9244,
      "step": 103598
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.8684486150741577,
      "learning_rate": 0.0003472630203733784,
      "loss": 2.2143,
      "step": 103599
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.8896920680999756,
      "learning_rate": 0.0003472589808894913,
      "loss": 2.1401,
      "step": 103600
    },
    {
      "epoch": 1.35,
      "grad_norm": 1.2998589277267456,
      "learning_rate": 0.00034725494139681794,
      "loss": 1.9965,
      "step": 103601
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.7908372282981873,
      "learning_rate": 0.0003472509018953591,
      "loss": 2.0457,
      "step": 103602
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.9349909424781799,
      "learning_rate": 0.00034724686238511525,
      "loss": 2.187,
      "step": 103603
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.7833223342895508,
      "learning_rate": 0.0003472428228660874,
      "loss": 2.2747,
      "step": 103604
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.8726475238800049,
      "learning_rate": 0.0003472387833382763,
      "loss": 2.1689,
      "step": 103605
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.9083079099655151,
      "learning_rate": 0.00034723474380168244,
      "loss": 1.9773,
      "step": 103606
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.9727077484130859,
      "learning_rate": 0.00034723070425630683,
      "loss": 2.0972,
      "step": 103607
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.7717581391334534,
      "learning_rate": 0.0003472266647021501,
      "loss": 2.0876,
      "step": 103608
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.870606541633606,
      "learning_rate": 0.0003472226251392132,
      "loss": 2.2087,
      "step": 103609
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.8981802463531494,
      "learning_rate": 0.00034721858556749656,
      "loss": 2.1901,
      "step": 103610
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.8191710114479065,
      "learning_rate": 0.00034721454598700115,
      "loss": 2.079,
      "step": 103611
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.8577390909194946,
      "learning_rate": 0.0003472105063977277,
      "loss": 2.2277,
      "step": 103612
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.9118903279304504,
      "learning_rate": 0.00034720646679967684,
      "loss": 2.0559,
      "step": 103613
    },
    {
      "epoch": 1.35,
      "grad_norm": 1.1439837217330933,
      "learning_rate": 0.00034720242719284947,
      "loss": 2.0637,
      "step": 103614
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.9078656435012817,
      "learning_rate": 0.00034719838757724624,
      "loss": 2.225,
      "step": 103615
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.9864633679389954,
      "learning_rate": 0.00034719434795286796,
      "loss": 2.1924,
      "step": 103616
    },
    {
      "epoch": 1.35,
      "grad_norm": 1.079763412475586,
      "learning_rate": 0.0003471903083197154,
      "loss": 2.1633,
      "step": 103617
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.9195396900177002,
      "learning_rate": 0.00034718626867778925,
      "loss": 2.1499,
      "step": 103618
    },
    {
      "epoch": 1.35,
      "grad_norm": 1.0529118776321411,
      "learning_rate": 0.00034718222902709023,
      "loss": 2.2867,
      "step": 103619
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.9992960095405579,
      "learning_rate": 0.00034717818936761915,
      "loss": 2.2304,
      "step": 103620
    },
    {
      "epoch": 1.35,
      "grad_norm": 1.1066617965698242,
      "learning_rate": 0.0003471741496993768,
      "loss": 2.0825,
      "step": 103621
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.9166644811630249,
      "learning_rate": 0.00034717011002236383,
      "loss": 2.1892,
      "step": 103622
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.905103325843811,
      "learning_rate": 0.0003471660703365811,
      "loss": 2.2419,
      "step": 103623
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.9454163312911987,
      "learning_rate": 0.00034716203064202937,
      "loss": 2.5139,
      "step": 103624
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.9242677092552185,
      "learning_rate": 0.0003471579909387092,
      "loss": 2.0464,
      "step": 103625
    },
    {
      "epoch": 1.35,
      "grad_norm": 1.2391608953475952,
      "learning_rate": 0.0003471539512266215,
      "loss": 2.0625,
      "step": 103626
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.8821431398391724,
      "learning_rate": 0.000347149911505767,
      "loss": 2.2021,
      "step": 103627
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.9267821907997131,
      "learning_rate": 0.0003471458717761465,
      "loss": 2.1391,
      "step": 103628
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.9314389228820801,
      "learning_rate": 0.0003471418320377606,
      "loss": 2.2328,
      "step": 103629
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.9078039526939392,
      "learning_rate": 0.0003471377922906102,
      "loss": 2.1608,
      "step": 103630
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.929104745388031,
      "learning_rate": 0.000347133752534696,
      "loss": 2.2751,
      "step": 103631
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.8243297934532166,
      "learning_rate": 0.0003471297127700187,
      "loss": 2.1254,
      "step": 103632
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.9370238780975342,
      "learning_rate": 0.00034712567299657914,
      "loss": 2.183,
      "step": 103633
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.9591995477676392,
      "learning_rate": 0.000347121633214378,
      "loss": 2.1426,
      "step": 103634
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.8406413793563843,
      "learning_rate": 0.000347117593423416,
      "loss": 1.8726,
      "step": 103635
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.8500834107398987,
      "learning_rate": 0.0003471135536236941,
      "loss": 2.1457,
      "step": 103636
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.8336347937583923,
      "learning_rate": 0.00034710951381521283,
      "loss": 2.0399,
      "step": 103637
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.8265148401260376,
      "learning_rate": 0.00034710547399797293,
      "loss": 2.2534,
      "step": 103638
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.8441410660743713,
      "learning_rate": 0.00034710143417197533,
      "loss": 2.2254,
      "step": 103639
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.8577682971954346,
      "learning_rate": 0.0003470973943372207,
      "loss": 2.1792,
      "step": 103640
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.8117326498031616,
      "learning_rate": 0.0003470933544937097,
      "loss": 2.1967,
      "step": 103641
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.9927439093589783,
      "learning_rate": 0.0003470893146414432,
      "loss": 2.3755,
      "step": 103642
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.9070309996604919,
      "learning_rate": 0.0003470852747804219,
      "loss": 2.1308,
      "step": 103643
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.7971017360687256,
      "learning_rate": 0.0003470812349106465,
      "loss": 2.1859,
      "step": 103644
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.885345995426178,
      "learning_rate": 0.00034707719503211795,
      "loss": 2.196,
      "step": 103645
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.85785311460495,
      "learning_rate": 0.0003470731551448368,
      "loss": 2.0586,
      "step": 103646
    },
    {
      "epoch": 1.35,
      "grad_norm": 1.0203561782836914,
      "learning_rate": 0.00034706911524880375,
      "loss": 2.1532,
      "step": 103647
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.8591804504394531,
      "learning_rate": 0.00034706507534401984,
      "loss": 2.3151,
      "step": 103648
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.8812288641929626,
      "learning_rate": 0.00034706103543048555,
      "loss": 2.1757,
      "step": 103649
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.8613908886909485,
      "learning_rate": 0.0003470569955082017,
      "loss": 2.1536,
      "step": 103650
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.8224246501922607,
      "learning_rate": 0.00034705295557716913,
      "loss": 2.0875,
      "step": 103651
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.8837522864341736,
      "learning_rate": 0.0003470489156373885,
      "loss": 2.1737,
      "step": 103652
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.8336840867996216,
      "learning_rate": 0.00034704487568886053,
      "loss": 2.0962,
      "step": 103653
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.8399158120155334,
      "learning_rate": 0.0003470408357315861,
      "loss": 1.979,
      "step": 103654
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.8993809819221497,
      "learning_rate": 0.00034703679576556586,
      "loss": 2.1386,
      "step": 103655
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.9054907560348511,
      "learning_rate": 0.0003470327557908006,
      "loss": 2.2901,
      "step": 103656
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.9210052490234375,
      "learning_rate": 0.0003470287158072911,
      "loss": 2.1908,
      "step": 103657
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.9106101989746094,
      "learning_rate": 0.000347024675815038,
      "loss": 2.3125,
      "step": 103658
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.8164160847663879,
      "learning_rate": 0.0003470206358140422,
      "loss": 2.0768,
      "step": 103659
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.8901501893997192,
      "learning_rate": 0.00034701659580430434,
      "loss": 2.1498,
      "step": 103660
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.9002466797828674,
      "learning_rate": 0.00034701255578582524,
      "loss": 2.1833,
      "step": 103661
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.9136377573013306,
      "learning_rate": 0.0003470085157586055,
      "loss": 2.057,
      "step": 103662
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.8558035492897034,
      "learning_rate": 0.0003470044757226461,
      "loss": 2.2077,
      "step": 103663
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.8949844241142273,
      "learning_rate": 0.0003470004356779477,
      "loss": 2.1195,
      "step": 103664
    },
    {
      "epoch": 1.35,
      "grad_norm": 1.014557123184204,
      "learning_rate": 0.000346996395624511,
      "loss": 2.1168,
      "step": 103665
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.8254344463348389,
      "learning_rate": 0.0003469923555623367,
      "loss": 2.0344,
      "step": 103666
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.8686848282814026,
      "learning_rate": 0.00034698831549142574,
      "loss": 2.1834,
      "step": 103667
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.8569790124893188,
      "learning_rate": 0.0003469842754117787,
      "loss": 2.1649,
      "step": 103668
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.8820415735244751,
      "learning_rate": 0.0003469802353233965,
      "loss": 2.1481,
      "step": 103669
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.9555544853210449,
      "learning_rate": 0.00034697619522627973,
      "loss": 2.2557,
      "step": 103670
    },
    {
      "epoch": 1.35,
      "grad_norm": 1.011406421661377,
      "learning_rate": 0.00034697215512042917,
      "loss": 2.1503,
      "step": 103671
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.8914182782173157,
      "learning_rate": 0.0003469681150058456,
      "loss": 2.0285,
      "step": 103672
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.8520926833152771,
      "learning_rate": 0.0003469640748825298,
      "loss": 1.8982,
      "step": 103673
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.9468550086021423,
      "learning_rate": 0.0003469600347504825,
      "loss": 2.0242,
      "step": 103674
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.9013235569000244,
      "learning_rate": 0.00034695599460970436,
      "loss": 2.0118,
      "step": 103675
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.9102456569671631,
      "learning_rate": 0.00034695195446019634,
      "loss": 2.2801,
      "step": 103676
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.8515563011169434,
      "learning_rate": 0.00034694791430195897,
      "loss": 2.2053,
      "step": 103677
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.8208289742469788,
      "learning_rate": 0.0003469438741349931,
      "loss": 2.1044,
      "step": 103678
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.9251750707626343,
      "learning_rate": 0.00034693983395929956,
      "loss": 2.1079,
      "step": 103679
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.9054683446884155,
      "learning_rate": 0.00034693579377487893,
      "loss": 1.9421,
      "step": 103680
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.8948158621788025,
      "learning_rate": 0.0003469317535817321,
      "loss": 2.054,
      "step": 103681
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.9874926805496216,
      "learning_rate": 0.0003469277133798597,
      "loss": 2.2584,
      "step": 103682
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.9752634763717651,
      "learning_rate": 0.0003469236731692627,
      "loss": 1.9786,
      "step": 103683
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.9310144782066345,
      "learning_rate": 0.0003469196329499416,
      "loss": 2.1952,
      "step": 103684
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.9611350297927856,
      "learning_rate": 0.0003469155927218972,
      "loss": 2.2504,
      "step": 103685
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.8230834603309631,
      "learning_rate": 0.0003469115524851304,
      "loss": 1.9577,
      "step": 103686
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.8684766292572021,
      "learning_rate": 0.0003469075122396418,
      "loss": 2.1023,
      "step": 103687
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.9761071801185608,
      "learning_rate": 0.0003469034719854322,
      "loss": 2.0227,
      "step": 103688
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.898317277431488,
      "learning_rate": 0.00034689943172250243,
      "loss": 2.3717,
      "step": 103689
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.9955306053161621,
      "learning_rate": 0.0003468953914508532,
      "loss": 2.3504,
      "step": 103690
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.9650223851203918,
      "learning_rate": 0.0003468913511704851,
      "loss": 1.9618,
      "step": 103691
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.9180464148521423,
      "learning_rate": 0.00034688731088139906,
      "loss": 2.2932,
      "step": 103692
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.9420782923698425,
      "learning_rate": 0.0003468832705835958,
      "loss": 2.0179,
      "step": 103693
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.8697232604026794,
      "learning_rate": 0.0003468792302770761,
      "loss": 2.3723,
      "step": 103694
    },
    {
      "epoch": 1.35,
      "grad_norm": 1.030596375465393,
      "learning_rate": 0.00034687518996184065,
      "loss": 2.1473,
      "step": 103695
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.8954954147338867,
      "learning_rate": 0.00034687114963789014,
      "loss": 1.9985,
      "step": 103696
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.8914076685905457,
      "learning_rate": 0.00034686710930522546,
      "loss": 2.1962,
      "step": 103697
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.9564724564552307,
      "learning_rate": 0.00034686306896384726,
      "loss": 2.1234,
      "step": 103698
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.9094050526618958,
      "learning_rate": 0.00034685902861375636,
      "loss": 2.1164,
      "step": 103699
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.881754994392395,
      "learning_rate": 0.0003468549882549536,
      "loss": 2.3171,
      "step": 103700
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.9595435857772827,
      "learning_rate": 0.0003468509478874394,
      "loss": 1.919,
      "step": 103701
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.8682671785354614,
      "learning_rate": 0.0003468469075112148,
      "loss": 2.0315,
      "step": 103702
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.90969318151474,
      "learning_rate": 0.00034684286712628054,
      "loss": 2.0787,
      "step": 103703
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.8762585520744324,
      "learning_rate": 0.0003468388267326373,
      "loss": 2.1682,
      "step": 103704
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.8828988075256348,
      "learning_rate": 0.00034683478633028574,
      "loss": 2.2073,
      "step": 103705
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.8963786959648132,
      "learning_rate": 0.0003468307459192268,
      "loss": 2.1212,
      "step": 103706
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.8412906527519226,
      "learning_rate": 0.0003468267054994611,
      "loss": 1.9187,
      "step": 103707
    },
    {
      "epoch": 1.35,
      "grad_norm": 1.2317417860031128,
      "learning_rate": 0.00034682266507098943,
      "loss": 2.2513,
      "step": 103708
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.8288553953170776,
      "learning_rate": 0.0003468186246338126,
      "loss": 2.1062,
      "step": 103709
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.9785693883895874,
      "learning_rate": 0.0003468145841879312,
      "loss": 2.1492,
      "step": 103710
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.8703629374504089,
      "learning_rate": 0.0003468105437333462,
      "loss": 2.0553,
      "step": 103711
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.8767224550247192,
      "learning_rate": 0.00034680650327005813,
      "loss": 2.0786,
      "step": 103712
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.8142492771148682,
      "learning_rate": 0.0003468024627980679,
      "loss": 2.0409,
      "step": 103713
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.8686646819114685,
      "learning_rate": 0.0003467984223173762,
      "loss": 2.2665,
      "step": 103714
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.9205982685089111,
      "learning_rate": 0.0003467943818279839,
      "loss": 2.1074,
      "step": 103715
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.8822802901268005,
      "learning_rate": 0.0003467903413298915,
      "loss": 2.0544,
      "step": 103716
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.8902536630630493,
      "learning_rate": 0.0003467863008230999,
      "loss": 2.2892,
      "step": 103717
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.9888067841529846,
      "learning_rate": 0.0003467822603076099,
      "loss": 2.1728,
      "step": 103718
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.7709395885467529,
      "learning_rate": 0.00034677821978342215,
      "loss": 2.1057,
      "step": 103719
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.865324079990387,
      "learning_rate": 0.0003467741792505375,
      "loss": 2.1302,
      "step": 103720
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.8751693367958069,
      "learning_rate": 0.0003467701387089566,
      "loss": 2.0662,
      "step": 103721
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.9175614714622498,
      "learning_rate": 0.0003467660981586802,
      "loss": 2.1837,
      "step": 103722
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.9688865542411804,
      "learning_rate": 0.00034676205759970915,
      "loss": 2.3176,
      "step": 103723
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.9033135175704956,
      "learning_rate": 0.00034675801703204425,
      "loss": 2.3312,
      "step": 103724
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.9563446044921875,
      "learning_rate": 0.000346753976455686,
      "loss": 2.2111,
      "step": 103725
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.810840368270874,
      "learning_rate": 0.0003467499358706353,
      "loss": 2.3033,
      "step": 103726
    },
    {
      "epoch": 1.35,
      "grad_norm": 1.2055364847183228,
      "learning_rate": 0.000346745895276893,
      "loss": 2.2565,
      "step": 103727
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.8753641247749329,
      "learning_rate": 0.00034674185467445975,
      "loss": 2.1217,
      "step": 103728
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.9309158325195312,
      "learning_rate": 0.0003467378140633363,
      "loss": 2.2635,
      "step": 103729
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.9203385710716248,
      "learning_rate": 0.0003467337734435233,
      "loss": 2.0063,
      "step": 103730
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.8883395195007324,
      "learning_rate": 0.0003467297328150218,
      "loss": 2.0863,
      "step": 103731
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.8452456593513489,
      "learning_rate": 0.00034672569217783223,
      "loss": 2.0272,
      "step": 103732
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.8645204901695251,
      "learning_rate": 0.0003467216515319555,
      "loss": 2.0872,
      "step": 103733
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.9029160141944885,
      "learning_rate": 0.0003467176108773923,
      "loss": 2.1724,
      "step": 103734
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.9329334497451782,
      "learning_rate": 0.0003467135702141435,
      "loss": 2.1642,
      "step": 103735
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.9240878820419312,
      "learning_rate": 0.0003467095295422097,
      "loss": 2.5268,
      "step": 103736
    },
    {
      "epoch": 1.35,
      "grad_norm": 1.08729088306427,
      "learning_rate": 0.0003467054888615918,
      "loss": 2.1711,
      "step": 103737
    },
    {
      "epoch": 1.35,
      "grad_norm": 1.4010823965072632,
      "learning_rate": 0.0003467014481722904,
      "loss": 2.0566,
      "step": 103738
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.9261764883995056,
      "learning_rate": 0.0003466974074743062,
      "loss": 1.9213,
      "step": 103739
    },
    {
      "epoch": 1.35,
      "grad_norm": 1.029152750968933,
      "learning_rate": 0.00034669336676764026,
      "loss": 2.3446,
      "step": 103740
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.9764014482498169,
      "learning_rate": 0.00034668932605229303,
      "loss": 2.3091,
      "step": 103741
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.8885525465011597,
      "learning_rate": 0.0003466852853282654,
      "loss": 2.061,
      "step": 103742
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.8813461065292358,
      "learning_rate": 0.00034668124459555816,
      "loss": 2.2323,
      "step": 103743
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.8739721775054932,
      "learning_rate": 0.000346677203854172,
      "loss": 2.02,
      "step": 103744
    },
    {
      "epoch": 1.35,
      "grad_norm": 1.2211812734603882,
      "learning_rate": 0.0003466731631041075,
      "loss": 2.1061,
      "step": 103745
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.9912188053131104,
      "learning_rate": 0.0003466691223453658,
      "loss": 2.2732,
      "step": 103746
    },
    {
      "epoch": 1.35,
      "grad_norm": 1.0153167247772217,
      "learning_rate": 0.00034666508157794727,
      "loss": 2.3264,
      "step": 103747
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.8628356456756592,
      "learning_rate": 0.00034666104080185284,
      "loss": 2.2,
      "step": 103748
    },
    {
      "epoch": 1.35,
      "grad_norm": 1.0382035970687866,
      "learning_rate": 0.0003466570000170833,
      "loss": 2.1414,
      "step": 103749
    },
    {
      "epoch": 1.35,
      "grad_norm": 1.0041890144348145,
      "learning_rate": 0.0003466529592236393,
      "loss": 2.2062,
      "step": 103750
    },
    {
      "epoch": 1.35,
      "grad_norm": 1.1033247709274292,
      "learning_rate": 0.00034664891842152163,
      "loss": 2.1678,
      "step": 103751
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.9032941460609436,
      "learning_rate": 0.00034664487761073106,
      "loss": 2.0508,
      "step": 103752
    },
    {
      "epoch": 1.35,
      "grad_norm": 1.071158766746521,
      "learning_rate": 0.0003466408367912684,
      "loss": 2.0179,
      "step": 103753
    },
    {
      "epoch": 1.35,
      "grad_norm": 1.2579054832458496,
      "learning_rate": 0.0003466367959631342,
      "loss": 2.2014,
      "step": 103754
    },
    {
      "epoch": 1.35,
      "grad_norm": 1.1472455263137817,
      "learning_rate": 0.00034663275512632945,
      "loss": 2.0904,
      "step": 103755
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.8873713612556458,
      "learning_rate": 0.00034662871428085465,
      "loss": 2.1417,
      "step": 103756
    },
    {
      "epoch": 1.35,
      "grad_norm": 1.0939375162124634,
      "learning_rate": 0.0003466246734267108,
      "loss": 2.2012,
      "step": 103757
    },
    {
      "epoch": 1.35,
      "grad_norm": 1.2376598119735718,
      "learning_rate": 0.00034662063256389857,
      "loss": 1.947,
      "step": 103758
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.921897828578949,
      "learning_rate": 0.0003466165916924186,
      "loss": 2.2071,
      "step": 103759
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.9496933817863464,
      "learning_rate": 0.00034661255081227176,
      "loss": 2.3029,
      "step": 103760
    },
    {
      "epoch": 1.35,
      "grad_norm": 1.1688270568847656,
      "learning_rate": 0.0003466085099234588,
      "loss": 2.2185,
      "step": 103761
    },
    {
      "epoch": 1.35,
      "grad_norm": 1.2638349533081055,
      "learning_rate": 0.00034660446902598036,
      "loss": 2.3187,
      "step": 103762
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.9059704542160034,
      "learning_rate": 0.00034660042811983735,
      "loss": 2.2776,
      "step": 103763
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.84456866979599,
      "learning_rate": 0.0003465963872050304,
      "loss": 1.903,
      "step": 103764
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.9553253650665283,
      "learning_rate": 0.00034659234628156034,
      "loss": 2.4304,
      "step": 103765
    },
    {
      "epoch": 1.35,
      "grad_norm": 1.1485470533370972,
      "learning_rate": 0.0003465883053494278,
      "loss": 2.018,
      "step": 103766
    },
    {
      "epoch": 1.35,
      "grad_norm": 1.1048481464385986,
      "learning_rate": 0.0003465842644086337,
      "loss": 2.1214,
      "step": 103767
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.8790717124938965,
      "learning_rate": 0.00034658022345917866,
      "loss": 2.0844,
      "step": 103768
    },
    {
      "epoch": 1.35,
      "grad_norm": 1.3122875690460205,
      "learning_rate": 0.0003465761825010635,
      "loss": 2.3991,
      "step": 103769
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.9171444177627563,
      "learning_rate": 0.00034657214153428886,
      "loss": 2.3129,
      "step": 103770
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.9639313817024231,
      "learning_rate": 0.00034656810055885574,
      "loss": 2.0976,
      "step": 103771
    },
    {
      "epoch": 1.35,
      "grad_norm": 1.09967041015625,
      "learning_rate": 0.00034656405957476466,
      "loss": 2.45,
      "step": 103772
    },
    {
      "epoch": 1.35,
      "grad_norm": 1.2470189332962036,
      "learning_rate": 0.00034656001858201634,
      "loss": 2.3528,
      "step": 103773
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.971676766872406,
      "learning_rate": 0.00034655597758061175,
      "loss": 2.14,
      "step": 103774
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.9030789732933044,
      "learning_rate": 0.00034655193657055154,
      "loss": 1.9794,
      "step": 103775
    },
    {
      "epoch": 1.35,
      "grad_norm": 1.036858081817627,
      "learning_rate": 0.00034654789555183636,
      "loss": 2.0596,
      "step": 103776
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.9552948474884033,
      "learning_rate": 0.00034654385452446713,
      "loss": 2.0213,
      "step": 103777
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.8790835738182068,
      "learning_rate": 0.0003465398134884445,
      "loss": 2.2383,
      "step": 103778
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.9036867618560791,
      "learning_rate": 0.00034653577244376924,
      "loss": 2.3427,
      "step": 103779
    },
    {
      "epoch": 1.35,
      "grad_norm": 1.106528878211975,
      "learning_rate": 0.00034653173139044205,
      "loss": 2.2793,
      "step": 103780
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.8634822368621826,
      "learning_rate": 0.00034652769032846373,
      "loss": 2.1717,
      "step": 103781
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.9064980745315552,
      "learning_rate": 0.0003465236492578351,
      "loss": 2.1633,
      "step": 103782
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.8672783374786377,
      "learning_rate": 0.0003465196081785568,
      "loss": 2.1903,
      "step": 103783
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.8475381135940552,
      "learning_rate": 0.0003465155670906296,
      "loss": 1.9884,
      "step": 103784
    },
    {
      "epoch": 1.35,
      "grad_norm": 1.0133867263793945,
      "learning_rate": 0.00034651152599405444,
      "loss": 2.1074,
      "step": 103785
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.8254137635231018,
      "learning_rate": 0.0003465074848888317,
      "loss": 2.073,
      "step": 103786
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.9445365071296692,
      "learning_rate": 0.00034650344377496243,
      "loss": 1.9461,
      "step": 103787
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.9068765044212341,
      "learning_rate": 0.00034649940265244736,
      "loss": 2.2562,
      "step": 103788
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.8116558194160461,
      "learning_rate": 0.0003464953615212871,
      "loss": 2.0774,
      "step": 103789
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.9094984531402588,
      "learning_rate": 0.0003464913203814824,
      "loss": 2.1664,
      "step": 103790
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.9464695453643799,
      "learning_rate": 0.00034648727923303424,
      "loss": 2.1549,
      "step": 103791
    },
    {
      "epoch": 1.35,
      "grad_norm": 1.0148844718933105,
      "learning_rate": 0.00034648323807594315,
      "loss": 1.9792,
      "step": 103792
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.9765273332595825,
      "learning_rate": 0.0003464791969102099,
      "loss": 2.2705,
      "step": 103793
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.88344806432724,
      "learning_rate": 0.0003464751557358353,
      "loss": 2.2212,
      "step": 103794
    },
    {
      "epoch": 1.35,
      "grad_norm": 1.0090028047561646,
      "learning_rate": 0.00034647111455282013,
      "loss": 2.3249,
      "step": 103795
    },
    {
      "epoch": 1.35,
      "grad_norm": 1.2536097764968872,
      "learning_rate": 0.0003464670733611651,
      "loss": 2.0963,
      "step": 103796
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.9303262233734131,
      "learning_rate": 0.00034646303216087095,
      "loss": 2.1792,
      "step": 103797
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.9044976830482483,
      "learning_rate": 0.0003464589909519385,
      "loss": 2.3066,
      "step": 103798
    },
    {
      "epoch": 1.35,
      "grad_norm": 1.2098209857940674,
      "learning_rate": 0.0003464549497343684,
      "loss": 1.9021,
      "step": 103799
    },
    {
      "epoch": 1.35,
      "grad_norm": 1.029292345046997,
      "learning_rate": 0.00034645090850816143,
      "loss": 2.1365,
      "step": 103800
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.8527030944824219,
      "learning_rate": 0.00034644686727331847,
      "loss": 1.9946,
      "step": 103801
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.8394966721534729,
      "learning_rate": 0.00034644282602984,
      "loss": 2.0574,
      "step": 103802
    },
    {
      "epoch": 1.35,
      "grad_norm": 1.1446601152420044,
      "learning_rate": 0.000346438784777727,
      "loss": 1.9861,
      "step": 103803
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.9832909107208252,
      "learning_rate": 0.0003464347435169802,
      "loss": 2.1014,
      "step": 103804
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.8355933427810669,
      "learning_rate": 0.00034643070224760024,
      "loss": 2.1737,
      "step": 103805
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.9779499769210815,
      "learning_rate": 0.000346426660969588,
      "loss": 2.1179,
      "step": 103806
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.9720650315284729,
      "learning_rate": 0.0003464226196829441,
      "loss": 1.9554,
      "step": 103807
    },
    {
      "epoch": 1.35,
      "grad_norm": 1.0545634031295776,
      "learning_rate": 0.00034641857838766933,
      "loss": 2.2802,
      "step": 103808
    },
    {
      "epoch": 1.35,
      "grad_norm": 1.0290179252624512,
      "learning_rate": 0.00034641453708376453,
      "loss": 2.4239,
      "step": 103809
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.9133167266845703,
      "learning_rate": 0.00034641049577123045,
      "loss": 2.067,
      "step": 103810
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.8502717614173889,
      "learning_rate": 0.0003464064544500677,
      "loss": 2.1685,
      "step": 103811
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.840514600276947,
      "learning_rate": 0.0003464024131202771,
      "loss": 2.2419,
      "step": 103812
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.9549155235290527,
      "learning_rate": 0.00034639837178185946,
      "loss": 2.0317,
      "step": 103813
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.8736713528633118,
      "learning_rate": 0.00034639433043481545,
      "loss": 1.9902,
      "step": 103814
    },
    {
      "epoch": 1.35,
      "grad_norm": 1.0734096765518188,
      "learning_rate": 0.0003463902890791459,
      "loss": 2.3452,
      "step": 103815
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.8896276354789734,
      "learning_rate": 0.0003463862477148515,
      "loss": 2.1969,
      "step": 103816
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.906437337398529,
      "learning_rate": 0.000346382206341933,
      "loss": 2.2136,
      "step": 103817
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.8813779354095459,
      "learning_rate": 0.0003463781649603912,
      "loss": 2.0692,
      "step": 103818
    },
    {
      "epoch": 1.35,
      "grad_norm": 1.004063367843628,
      "learning_rate": 0.00034637412357022685,
      "loss": 2.3306,
      "step": 103819
    },
    {
      "epoch": 1.35,
      "grad_norm": 1.0002845525741577,
      "learning_rate": 0.00034637008217144064,
      "loss": 2.3656,
      "step": 103820
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.8920012712478638,
      "learning_rate": 0.00034636604076403334,
      "loss": 2.0645,
      "step": 103821
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.8517255187034607,
      "learning_rate": 0.00034636199934800577,
      "loss": 2.2861,
      "step": 103822
    },
    {
      "epoch": 1.35,
      "grad_norm": 1.187228798866272,
      "learning_rate": 0.00034635795792335863,
      "loss": 2.029,
      "step": 103823
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.8267697691917419,
      "learning_rate": 0.00034635391649009257,
      "loss": 2.208,
      "step": 103824
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.8719910979270935,
      "learning_rate": 0.00034634987504820856,
      "loss": 1.9302,
      "step": 103825
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.8274446129798889,
      "learning_rate": 0.00034634583359770716,
      "loss": 2.2035,
      "step": 103826
    },
    {
      "epoch": 1.35,
      "grad_norm": 1.0137633085250854,
      "learning_rate": 0.0003463417921385892,
      "loss": 2.0735,
      "step": 103827
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.893302321434021,
      "learning_rate": 0.00034633775067085545,
      "loss": 2.2996,
      "step": 103828
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.8149821758270264,
      "learning_rate": 0.0003463337091945066,
      "loss": 2.0738,
      "step": 103829
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.8997015357017517,
      "learning_rate": 0.0003463296677095435,
      "loss": 2.1139,
      "step": 103830
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.8749298453330994,
      "learning_rate": 0.00034632562621596686,
      "loss": 2.1977,
      "step": 103831
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.9555675983428955,
      "learning_rate": 0.0003463215847137773,
      "loss": 2.2001,
      "step": 103832
    },
    {
      "epoch": 1.35,
      "grad_norm": 1.0133765935897827,
      "learning_rate": 0.0003463175432029757,
      "loss": 2.0482,
      "step": 103833
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.8982499837875366,
      "learning_rate": 0.0003463135016835629,
      "loss": 2.2974,
      "step": 103834
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.8578851222991943,
      "learning_rate": 0.00034630946015553953,
      "loss": 2.1679,
      "step": 103835
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.9581714272499084,
      "learning_rate": 0.00034630541861890623,
      "loss": 2.2198,
      "step": 103836
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.8013396263122559,
      "learning_rate": 0.00034630137707366404,
      "loss": 2.2605,
      "step": 103837
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.9504246711730957,
      "learning_rate": 0.00034629733551981345,
      "loss": 2.1821,
      "step": 103838
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.8367882370948792,
      "learning_rate": 0.00034629329395735527,
      "loss": 2.2345,
      "step": 103839
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.9071590304374695,
      "learning_rate": 0.0003462892523862904,
      "loss": 2.0796,
      "step": 103840
    },
    {
      "epoch": 1.35,
      "grad_norm": 1.018566370010376,
      "learning_rate": 0.00034628521080661944,
      "loss": 2.4322,
      "step": 103841
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.9628108739852905,
      "learning_rate": 0.00034628116921834315,
      "loss": 2.0278,
      "step": 103842
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.9778763055801392,
      "learning_rate": 0.0003462771276214624,
      "loss": 1.972,
      "step": 103843
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.8898997902870178,
      "learning_rate": 0.00034627308601597784,
      "loss": 2.0362,
      "step": 103844
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.9820608496665955,
      "learning_rate": 0.00034626904440189017,
      "loss": 2.2781,
      "step": 103845
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.8915877342224121,
      "learning_rate": 0.0003462650027792003,
      "loss": 2.1423,
      "step": 103846
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.9482290148735046,
      "learning_rate": 0.0003462609611479088,
      "loss": 2.1213,
      "step": 103847
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.9787999987602234,
      "learning_rate": 0.0003462569195080166,
      "loss": 2.1991,
      "step": 103848
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.8582066297531128,
      "learning_rate": 0.0003462528778595243,
      "loss": 2.2839,
      "step": 103849
    },
    {
      "epoch": 1.35,
      "grad_norm": 1.088529348373413,
      "learning_rate": 0.00034624883620243277,
      "loss": 2.1259,
      "step": 103850
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.9277384877204895,
      "learning_rate": 0.0003462447945367426,
      "loss": 2.4707,
      "step": 103851
    },
    {
      "epoch": 1.35,
      "grad_norm": 1.0149344205856323,
      "learning_rate": 0.00034624075286245477,
      "loss": 2.0998,
      "step": 103852
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.9955025911331177,
      "learning_rate": 0.00034623671117956986,
      "loss": 2.2,
      "step": 103853
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.8730282187461853,
      "learning_rate": 0.00034623266948808864,
      "loss": 2.0615,
      "step": 103854
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.9282300472259521,
      "learning_rate": 0.00034622862778801197,
      "loss": 2.137,
      "step": 103855
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.9802640676498413,
      "learning_rate": 0.0003462245860793405,
      "loss": 2.3556,
      "step": 103856
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.9766401052474976,
      "learning_rate": 0.00034622054436207495,
      "loss": 2.0203,
      "step": 103857
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.9117588996887207,
      "learning_rate": 0.00034621650263621617,
      "loss": 2.3529,
      "step": 103858
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.9654451608657837,
      "learning_rate": 0.0003462124609017649,
      "loss": 2.2522,
      "step": 103859
    },
    {
      "epoch": 1.35,
      "grad_norm": 1.0339205265045166,
      "learning_rate": 0.00034620841915872187,
      "loss": 2.0707,
      "step": 103860
    },
    {
      "epoch": 1.35,
      "grad_norm": 1.1018515825271606,
      "learning_rate": 0.00034620437740708775,
      "loss": 2.2597,
      "step": 103861
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.867793083190918,
      "learning_rate": 0.0003462003356468634,
      "loss": 2.1633,
      "step": 103862
    },
    {
      "epoch": 1.35,
      "grad_norm": 1.2175546884536743,
      "learning_rate": 0.00034619629387804956,
      "loss": 2.0613,
      "step": 103863
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.9887521266937256,
      "learning_rate": 0.00034619225210064694,
      "loss": 2.1539,
      "step": 103864
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.9300262928009033,
      "learning_rate": 0.0003461882103146563,
      "loss": 2.2052,
      "step": 103865
    },
    {
      "epoch": 1.35,
      "grad_norm": 1.0376710891723633,
      "learning_rate": 0.0003461841685200784,
      "loss": 2.0512,
      "step": 103866
    },
    {
      "epoch": 1.35,
      "grad_norm": 1.174233317375183,
      "learning_rate": 0.00034618012671691395,
      "loss": 2.3359,
      "step": 103867
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.9024737477302551,
      "learning_rate": 0.0003461760849051638,
      "loss": 2.0394,
      "step": 103868
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.8288981318473816,
      "learning_rate": 0.0003461720430848287,
      "loss": 2.0731,
      "step": 103869
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.8784121870994568,
      "learning_rate": 0.0003461680012559092,
      "loss": 2.2385,
      "step": 103870
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.8453227281570435,
      "learning_rate": 0.0003461639594184063,
      "loss": 2.1697,
      "step": 103871
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.8819019794464111,
      "learning_rate": 0.0003461599175723206,
      "loss": 2.1537,
      "step": 103872
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.886483371257782,
      "learning_rate": 0.0003461558757176529,
      "loss": 2.4533,
      "step": 103873
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.9076414704322815,
      "learning_rate": 0.00034615183385440404,
      "loss": 2.0197,
      "step": 103874
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.9390166401863098,
      "learning_rate": 0.00034614779198257456,
      "loss": 2.2905,
      "step": 103875
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.9880982637405396,
      "learning_rate": 0.0003461437501021654,
      "loss": 1.9684,
      "step": 103876
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.8590795993804932,
      "learning_rate": 0.0003461397082131773,
      "loss": 2.1703,
      "step": 103877
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.941441535949707,
      "learning_rate": 0.0003461356663156109,
      "loss": 2.4329,
      "step": 103878
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.8572946786880493,
      "learning_rate": 0.00034613162440946696,
      "loss": 1.9742,
      "step": 103879
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.9377133846282959,
      "learning_rate": 0.0003461275824947464,
      "loss": 2.1314,
      "step": 103880
    },
    {
      "epoch": 1.35,
      "grad_norm": 1.140189528465271,
      "learning_rate": 0.0003461235405714497,
      "loss": 2.189,
      "step": 103881
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.9714452028274536,
      "learning_rate": 0.00034611949863957785,
      "loss": 2.2484,
      "step": 103882
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.9653593301773071,
      "learning_rate": 0.00034611545669913156,
      "loss": 2.3494,
      "step": 103883
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.8948551416397095,
      "learning_rate": 0.0003461114147501114,
      "loss": 2.0443,
      "step": 103884
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.831086277961731,
      "learning_rate": 0.00034610737279251833,
      "loss": 2.2335,
      "step": 103885
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.8803477883338928,
      "learning_rate": 0.0003461033308263531,
      "loss": 2.2848,
      "step": 103886
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.8469727039337158,
      "learning_rate": 0.0003460992888516163,
      "loss": 2.2645,
      "step": 103887
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.871239423751831,
      "learning_rate": 0.00034609524686830877,
      "loss": 2.0375,
      "step": 103888
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.9189020395278931,
      "learning_rate": 0.0003460912048764314,
      "loss": 2.194,
      "step": 103889
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.9106581211090088,
      "learning_rate": 0.00034608716287598464,
      "loss": 2.1263,
      "step": 103890
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.8959406614303589,
      "learning_rate": 0.0003460831208669695,
      "loss": 2.1491,
      "step": 103891
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.9971683621406555,
      "learning_rate": 0.00034607907884938655,
      "loss": 2.1654,
      "step": 103892
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.8789724111557007,
      "learning_rate": 0.00034607503682323673,
      "loss": 2.1999,
      "step": 103893
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.9119694828987122,
      "learning_rate": 0.00034607099478852057,
      "loss": 2.1317,
      "step": 103894
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.8848897218704224,
      "learning_rate": 0.0003460669527452391,
      "loss": 2.0496,
      "step": 103895
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.9145779013633728,
      "learning_rate": 0.00034606291069339283,
      "loss": 2.4166,
      "step": 103896
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.8515293002128601,
      "learning_rate": 0.00034605886863298254,
      "loss": 1.9354,
      "step": 103897
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.9250609278678894,
      "learning_rate": 0.0003460548265640092,
      "loss": 2.0456,
      "step": 103898
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.9257309436798096,
      "learning_rate": 0.0003460507844864732,
      "loss": 2.2299,
      "step": 103899
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.9791780710220337,
      "learning_rate": 0.0003460467424003756,
      "loss": 2.3696,
      "step": 103900
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.8143526911735535,
      "learning_rate": 0.0003460427003057171,
      "loss": 2.0966,
      "step": 103901
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.8638534545898438,
      "learning_rate": 0.00034603865820249827,
      "loss": 2.2193,
      "step": 103902
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.8490225076675415,
      "learning_rate": 0.00034603461609071995,
      "loss": 2.1693,
      "step": 103903
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.7798429131507874,
      "learning_rate": 0.00034603057397038304,
      "loss": 2.0539,
      "step": 103904
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.8609791994094849,
      "learning_rate": 0.0003460265318414881,
      "loss": 2.1502,
      "step": 103905
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.789271891117096,
      "learning_rate": 0.00034602248970403603,
      "loss": 2.455,
      "step": 103906
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.905064582824707,
      "learning_rate": 0.0003460184475580275,
      "loss": 2.2714,
      "step": 103907
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.8627705574035645,
      "learning_rate": 0.00034601440540346325,
      "loss": 2.0223,
      "step": 103908
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.8137589693069458,
      "learning_rate": 0.00034601036324034406,
      "loss": 2.343,
      "step": 103909
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.9026473760604858,
      "learning_rate": 0.0003460063210686707,
      "loss": 2.2726,
      "step": 103910
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.8879621028900146,
      "learning_rate": 0.00034600227888844384,
      "loss": 2.0769,
      "step": 103911
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.9582769274711609,
      "learning_rate": 0.0003459982366996643,
      "loss": 1.9592,
      "step": 103912
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.9005249738693237,
      "learning_rate": 0.0003459941945023329,
      "loss": 1.9886,
      "step": 103913
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.8709399700164795,
      "learning_rate": 0.0003459901522964502,
      "loss": 2.0271,
      "step": 103914
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.9187030792236328,
      "learning_rate": 0.00034598611008201713,
      "loss": 2.2431,
      "step": 103915
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.8735105991363525,
      "learning_rate": 0.00034598206785903436,
      "loss": 2.0591,
      "step": 103916
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.9511801600456238,
      "learning_rate": 0.0003459780256275026,
      "loss": 2.1611,
      "step": 103917
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.8688077926635742,
      "learning_rate": 0.0003459739833874227,
      "loss": 2.1759,
      "step": 103918
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.8278377652168274,
      "learning_rate": 0.0003459699411387955,
      "loss": 2.1066,
      "step": 103919
    },
    {
      "epoch": 1.35,
      "grad_norm": 1.066282033920288,
      "learning_rate": 0.0003459658988816214,
      "loss": 2.1491,
      "step": 103920
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.7822925448417664,
      "learning_rate": 0.00034596185661590147,
      "loss": 2.1835,
      "step": 103921
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.9717354774475098,
      "learning_rate": 0.0003459578143416364,
      "loss": 2.1858,
      "step": 103922
    },
    {
      "epoch": 1.35,
      "grad_norm": 1.0811517238616943,
      "learning_rate": 0.00034595377205882684,
      "loss": 2.0683,
      "step": 103923
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.8067125082015991,
      "learning_rate": 0.00034594972976747363,
      "loss": 1.8487,
      "step": 103924
    },
    {
      "epoch": 1.35,
      "grad_norm": 1.0443329811096191,
      "learning_rate": 0.00034594568746757745,
      "loss": 2.076,
      "step": 103925
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.9940233826637268,
      "learning_rate": 0.0003459416451591392,
      "loss": 2.074,
      "step": 103926
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.8399497270584106,
      "learning_rate": 0.0003459376028421595,
      "loss": 2.3353,
      "step": 103927
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.9324161410331726,
      "learning_rate": 0.00034593356051663905,
      "loss": 2.3462,
      "step": 103928
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.9439419507980347,
      "learning_rate": 0.0003459295181825788,
      "loss": 2.0681,
      "step": 103929
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.9109356999397278,
      "learning_rate": 0.0003459254758399793,
      "loss": 2.2773,
      "step": 103930
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.8341131806373596,
      "learning_rate": 0.0003459214334888414,
      "loss": 2.1992,
      "step": 103931
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.9018682837486267,
      "learning_rate": 0.0003459173911291659,
      "loss": 1.9026,
      "step": 103932
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.9859175682067871,
      "learning_rate": 0.00034591334876095343,
      "loss": 2.1716,
      "step": 103933
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.8929532170295715,
      "learning_rate": 0.0003459093063842048,
      "loss": 1.9613,
      "step": 103934
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.8690482974052429,
      "learning_rate": 0.0003459052639989208,
      "loss": 2.1534,
      "step": 103935
    },
    {
      "epoch": 1.35,
      "grad_norm": 1.0509724617004395,
      "learning_rate": 0.00034590122160510213,
      "loss": 2.2845,
      "step": 103936
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.973454475402832,
      "learning_rate": 0.00034589717920274955,
      "loss": 1.8887,
      "step": 103937
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.8817905187606812,
      "learning_rate": 0.0003458931367918638,
      "loss": 2.3127,
      "step": 103938
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.917980968952179,
      "learning_rate": 0.00034588909437244567,
      "loss": 2.0935,
      "step": 103939
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.9415597915649414,
      "learning_rate": 0.0003458850519444959,
      "loss": 2.3456,
      "step": 103940
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.9122049808502197,
      "learning_rate": 0.0003458810095080153,
      "loss": 2.1909,
      "step": 103941
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.8912001848220825,
      "learning_rate": 0.00034587696706300444,
      "loss": 2.0126,
      "step": 103942
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.899431586265564,
      "learning_rate": 0.0003458729246094642,
      "loss": 2.1925,
      "step": 103943
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.9147305488586426,
      "learning_rate": 0.0003458688821473954,
      "loss": 2.1717,
      "step": 103944
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.9047912359237671,
      "learning_rate": 0.0003458648396767986,
      "loss": 2.3745,
      "step": 103945
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.8161434531211853,
      "learning_rate": 0.0003458607971976747,
      "loss": 1.9685,
      "step": 103946
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.8954946398735046,
      "learning_rate": 0.0003458567547100244,
      "loss": 2.2161,
      "step": 103947
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.8604336977005005,
      "learning_rate": 0.00034585271221384856,
      "loss": 2.1156,
      "step": 103948
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.9214105606079102,
      "learning_rate": 0.00034584866970914773,
      "loss": 2.0937,
      "step": 103949
    },
    {
      "epoch": 1.35,
      "grad_norm": 1.0248775482177734,
      "learning_rate": 0.0003458446271959228,
      "loss": 2.1845,
      "step": 103950
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.8587905168533325,
      "learning_rate": 0.0003458405846741745,
      "loss": 2.2474,
      "step": 103951
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.8792462348937988,
      "learning_rate": 0.00034583654214390353,
      "loss": 2.0353,
      "step": 103952
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.9123836755752563,
      "learning_rate": 0.00034583249960511066,
      "loss": 2.3509,
      "step": 103953
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.9762401580810547,
      "learning_rate": 0.00034582845705779673,
      "loss": 2.0596,
      "step": 103954
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.9169553518295288,
      "learning_rate": 0.00034582441450196243,
      "loss": 2.0296,
      "step": 103955
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.9238544702529907,
      "learning_rate": 0.00034582037193760847,
      "loss": 2.3954,
      "step": 103956
    },
    {
      "epoch": 1.35,
      "grad_norm": 1.0919599533081055,
      "learning_rate": 0.00034581632936473566,
      "loss": 2.1883,
      "step": 103957
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.9076681137084961,
      "learning_rate": 0.00034581228678334466,
      "loss": 2.2968,
      "step": 103958
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.9220505952835083,
      "learning_rate": 0.00034580824419343643,
      "loss": 2.0263,
      "step": 103959
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.9355211853981018,
      "learning_rate": 0.0003458042015950115,
      "loss": 2.2706,
      "step": 103960
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.9067978858947754,
      "learning_rate": 0.0003458001589880707,
      "loss": 2.2593,
      "step": 103961
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.8714017271995544,
      "learning_rate": 0.0003457961163726148,
      "loss": 1.9689,
      "step": 103962
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.8938935399055481,
      "learning_rate": 0.00034579207374864455,
      "loss": 2.1319,
      "step": 103963
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.8474253416061401,
      "learning_rate": 0.0003457880311161606,
      "loss": 2.1574,
      "step": 103964
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.944060742855072,
      "learning_rate": 0.0003457839884751639,
      "loss": 2.1064,
      "step": 103965
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.880930483341217,
      "learning_rate": 0.00034577994582565504,
      "loss": 2.1565,
      "step": 103966
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.8153226971626282,
      "learning_rate": 0.00034577590316763485,
      "loss": 1.9642,
      "step": 103967
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.8692672252655029,
      "learning_rate": 0.0003457718605011041,
      "loss": 2.4189,
      "step": 103968
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.9182121157646179,
      "learning_rate": 0.00034576781782606346,
      "loss": 2.2576,
      "step": 103969
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.892477810382843,
      "learning_rate": 0.00034576377514251367,
      "loss": 2.3767,
      "step": 103970
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.8741456270217896,
      "learning_rate": 0.0003457597324504556,
      "loss": 1.9888,
      "step": 103971
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.9012622833251953,
      "learning_rate": 0.00034575568974988985,
      "loss": 2.1085,
      "step": 103972
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.9452100992202759,
      "learning_rate": 0.00034575164704081724,
      "loss": 2.3566,
      "step": 103973
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.8680111169815063,
      "learning_rate": 0.0003457476043232387,
      "loss": 2.2908,
      "step": 103974
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.823880672454834,
      "learning_rate": 0.00034574356159715466,
      "loss": 1.9629,
      "step": 103975
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.8664239645004272,
      "learning_rate": 0.00034573951886256605,
      "loss": 1.9427,
      "step": 103976
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.9699902534484863,
      "learning_rate": 0.00034573547611947366,
      "loss": 2.1634,
      "step": 103977
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.9745699167251587,
      "learning_rate": 0.0003457314333678781,
      "loss": 2.1138,
      "step": 103978
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.9856576323509216,
      "learning_rate": 0.00034572739060778027,
      "loss": 2.3179,
      "step": 103979
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.9046851992607117,
      "learning_rate": 0.0003457233478391809,
      "loss": 2.1978,
      "step": 103980
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.8826253414154053,
      "learning_rate": 0.0003457193050620806,
      "loss": 2.31,
      "step": 103981
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.9042630791664124,
      "learning_rate": 0.00034571526227648023,
      "loss": 1.8765,
      "step": 103982
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.7977219223976135,
      "learning_rate": 0.00034571121948238063,
      "loss": 2.216,
      "step": 103983
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.7984111309051514,
      "learning_rate": 0.0003457071766797823,
      "loss": 2.2975,
      "step": 103984
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.7936133146286011,
      "learning_rate": 0.0003457031338686862,
      "loss": 2.2481,
      "step": 103985
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.873495876789093,
      "learning_rate": 0.0003456990910490931,
      "loss": 2.2085,
      "step": 103986
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.8574898838996887,
      "learning_rate": 0.00034569504822100355,
      "loss": 1.9815,
      "step": 103987
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.9010515809059143,
      "learning_rate": 0.0003456910053844185,
      "loss": 2.2339,
      "step": 103988
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.87906813621521,
      "learning_rate": 0.00034568696253933865,
      "loss": 2.1359,
      "step": 103989
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.9040146470069885,
      "learning_rate": 0.00034568291968576476,
      "loss": 2.1747,
      "step": 103990
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.8370913863182068,
      "learning_rate": 0.00034567887682369745,
      "loss": 2.0762,
      "step": 103991
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.9536092877388,
      "learning_rate": 0.0003456748339531376,
      "loss": 2.1172,
      "step": 103992
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.8525908589363098,
      "learning_rate": 0.000345670791074086,
      "loss": 2.0133,
      "step": 103993
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.8606365919113159,
      "learning_rate": 0.00034566674818654327,
      "loss": 2.112,
      "step": 103994
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.9212384819984436,
      "learning_rate": 0.00034566270529051024,
      "loss": 2.1235,
      "step": 103995
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.8180646896362305,
      "learning_rate": 0.00034565866238598774,
      "loss": 2.1269,
      "step": 103996
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.8827380537986755,
      "learning_rate": 0.00034565461947297637,
      "loss": 1.9069,
      "step": 103997
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.9618393778800964,
      "learning_rate": 0.00034565057655147683,
      "loss": 1.9236,
      "step": 103998
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.8825638890266418,
      "learning_rate": 0.00034564653362149017,
      "loss": 2.1371,
      "step": 103999
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.959371030330658,
      "learning_rate": 0.00034564249068301685,
      "loss": 2.1043,
      "step": 104000
    },
    {
      "epoch": 1.35,
      "grad_norm": 1.1584827899932861,
      "learning_rate": 0.00034563844773605776,
      "loss": 2.3552,
      "step": 104001
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.9562665820121765,
      "learning_rate": 0.00034563440478061364,
      "loss": 2.0826,
      "step": 104002
    },
    {
      "epoch": 1.35,
      "grad_norm": 1.0047885179519653,
      "learning_rate": 0.0003456303618166852,
      "loss": 2.015,
      "step": 104003
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.9512091875076294,
      "learning_rate": 0.00034562631884427317,
      "loss": 1.8648,
      "step": 104004
    },
    {
      "epoch": 1.35,
      "grad_norm": 1.1888515949249268,
      "learning_rate": 0.00034562227586337844,
      "loss": 2.2499,
      "step": 104005
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.9660938382148743,
      "learning_rate": 0.0003456182328740016,
      "loss": 2.0688,
      "step": 104006
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.9191613793373108,
      "learning_rate": 0.00034561418987614346,
      "loss": 2.0416,
      "step": 104007
    },
    {
      "epoch": 1.35,
      "grad_norm": 1.1404352188110352,
      "learning_rate": 0.0003456101468698049,
      "loss": 2.5081,
      "step": 104008
    },
    {
      "epoch": 1.35,
      "grad_norm": 1.0013359785079956,
      "learning_rate": 0.0003456061038549864,
      "loss": 2.1568,
      "step": 104009
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.9001590609550476,
      "learning_rate": 0.0003456020608316889,
      "loss": 2.2547,
      "step": 104010
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.8677302598953247,
      "learning_rate": 0.0003455980177999132,
      "loss": 2.1607,
      "step": 104011
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.8997803926467896,
      "learning_rate": 0.00034559397475965987,
      "loss": 1.9478,
      "step": 104012
    },
    {
      "epoch": 1.35,
      "grad_norm": 1.0181304216384888,
      "learning_rate": 0.0003455899317109298,
      "loss": 2.3494,
      "step": 104013
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.8486839532852173,
      "learning_rate": 0.0003455858886537238,
      "loss": 2.0927,
      "step": 104014
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.8302570581436157,
      "learning_rate": 0.0003455818455880423,
      "loss": 1.9638,
      "step": 104015
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.9070016145706177,
      "learning_rate": 0.00034557780251388647,
      "loss": 2.0852,
      "step": 104016
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.9789110422134399,
      "learning_rate": 0.00034557375943125676,
      "loss": 2.066,
      "step": 104017
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.9997525215148926,
      "learning_rate": 0.0003455697163401541,
      "loss": 2.2378,
      "step": 104018
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.9733458757400513,
      "learning_rate": 0.00034556567324057916,
      "loss": 2.3028,
      "step": 104019
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.9838710427284241,
      "learning_rate": 0.0003455616301325327,
      "loss": 2.2392,
      "step": 104020
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.8757193088531494,
      "learning_rate": 0.0003455575870160154,
      "loss": 2.1838,
      "step": 104021
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.8499149680137634,
      "learning_rate": 0.0003455535438910281,
      "loss": 2.2804,
      "step": 104022
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.8221368193626404,
      "learning_rate": 0.0003455495007575716,
      "loss": 1.9962,
      "step": 104023
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.7983033061027527,
      "learning_rate": 0.00034554545761564655,
      "loss": 2.0505,
      "step": 104024
    },
    {
      "epoch": 1.35,
      "grad_norm": 1.0168731212615967,
      "learning_rate": 0.0003455414144652537,
      "loss": 2.072,
      "step": 104025
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.9199690818786621,
      "learning_rate": 0.000345537371306394,
      "loss": 1.9041,
      "step": 104026
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.8878887891769409,
      "learning_rate": 0.00034553332813906787,
      "loss": 2.3695,
      "step": 104027
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.8277866244316101,
      "learning_rate": 0.00034552928496327625,
      "loss": 2.0418,
      "step": 104028
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.9447411298751831,
      "learning_rate": 0.00034552524177902,
      "loss": 1.9961,
      "step": 104029
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.9458581209182739,
      "learning_rate": 0.00034552119858629964,
      "loss": 2.1852,
      "step": 104030
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.9683769345283508,
      "learning_rate": 0.00034551715538511606,
      "loss": 2.2309,
      "step": 104031
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.858433723449707,
      "learning_rate": 0.00034551311217547,
      "loss": 2.1057,
      "step": 104032
    },
    {
      "epoch": 1.35,
      "grad_norm": 1.0630357265472412,
      "learning_rate": 0.00034550906895736215,
      "loss": 2.3431,
      "step": 104033
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.8506661653518677,
      "learning_rate": 0.00034550502573079333,
      "loss": 2.3808,
      "step": 104034
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.9067783951759338,
      "learning_rate": 0.0003455009824957642,
      "loss": 2.0677,
      "step": 104035
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.9095520377159119,
      "learning_rate": 0.00034549693925227577,
      "loss": 2.163,
      "step": 104036
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.886700451374054,
      "learning_rate": 0.00034549289600032843,
      "loss": 2.2258,
      "step": 104037
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.9390051364898682,
      "learning_rate": 0.0003454888527399231,
      "loss": 2.4563,
      "step": 104038
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.9332392811775208,
      "learning_rate": 0.00034548480947106056,
      "loss": 2.1263,
      "step": 104039
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.9083262085914612,
      "learning_rate": 0.0003454807661937416,
      "loss": 2.2932,
      "step": 104040
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.951829195022583,
      "learning_rate": 0.00034547672290796683,
      "loss": 2.0997,
      "step": 104041
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.8212149143218994,
      "learning_rate": 0.00034547267961373713,
      "loss": 2.2469,
      "step": 104042
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.9121295213699341,
      "learning_rate": 0.0003454686363110532,
      "loss": 2.1579,
      "step": 104043
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.9158843159675598,
      "learning_rate": 0.00034546459299991574,
      "loss": 2.0916,
      "step": 104044
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.8765835165977478,
      "learning_rate": 0.00034546054968032563,
      "loss": 2.2293,
      "step": 104045
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.8458824753761292,
      "learning_rate": 0.00034545650635228346,
      "loss": 2.1764,
      "step": 104046
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.8621934056282043,
      "learning_rate": 0.00034545246301579015,
      "loss": 2.3129,
      "step": 104047
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.8883171081542969,
      "learning_rate": 0.0003454484196708463,
      "loss": 1.9392,
      "step": 104048
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.9360311031341553,
      "learning_rate": 0.0003454443763174527,
      "loss": 2.3206,
      "step": 104049
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.8693563938140869,
      "learning_rate": 0.0003454403329556103,
      "loss": 2.2777,
      "step": 104050
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.9138466715812683,
      "learning_rate": 0.0003454362895853195,
      "loss": 2.1281,
      "step": 104051
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.9011656045913696,
      "learning_rate": 0.0003454322462065814,
      "loss": 2.1557,
      "step": 104052
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.8228915333747864,
      "learning_rate": 0.0003454282028193965,
      "loss": 2.2769,
      "step": 104053
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.899196982383728,
      "learning_rate": 0.0003454241594237656,
      "loss": 2.1479,
      "step": 104054
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.9102370142936707,
      "learning_rate": 0.0003454201160196895,
      "loss": 2.1952,
      "step": 104055
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.8210738301277161,
      "learning_rate": 0.000345416072607169,
      "loss": 2.1305,
      "step": 104056
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.9808359146118164,
      "learning_rate": 0.00034541202918620474,
      "loss": 2.1424,
      "step": 104057
    },
    {
      "epoch": 1.35,
      "grad_norm": 1.2851535081863403,
      "learning_rate": 0.0003454079857567976,
      "loss": 2.0135,
      "step": 104058
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.8942426443099976,
      "learning_rate": 0.00034540394231894817,
      "loss": 2.367,
      "step": 104059
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.8373443484306335,
      "learning_rate": 0.0003453998988726574,
      "loss": 2.1453,
      "step": 104060
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.9267455339431763,
      "learning_rate": 0.0003453958554179258,
      "loss": 2.1905,
      "step": 104061
    },
    {
      "epoch": 1.35,
      "grad_norm": 1.0628061294555664,
      "learning_rate": 0.0003453918119547543,
      "loss": 1.9398,
      "step": 104062
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.8193541765213013,
      "learning_rate": 0.0003453877684831437,
      "loss": 2.2828,
      "step": 104063
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.8923174142837524,
      "learning_rate": 0.00034538372500309456,
      "loss": 2.1418,
      "step": 104064
    },
    {
      "epoch": 1.35,
      "grad_norm": 0.9194362163543701,
      "learning_rate": 0.00034537968151460776,
      "loss": 2.2578,
      "step": 104065
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.9551258683204651,
      "learning_rate": 0.000345375638017684,
      "loss": 2.0046,
      "step": 104066
    },
    {
      "epoch": 1.36,
      "grad_norm": 1.0478558540344238,
      "learning_rate": 0.0003453715945123241,
      "loss": 2.2023,
      "step": 104067
    },
    {
      "epoch": 1.36,
      "grad_norm": 1.0543278455734253,
      "learning_rate": 0.00034536755099852867,
      "loss": 2.2079,
      "step": 104068
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.7854470014572144,
      "learning_rate": 0.00034536350747629873,
      "loss": 2.1951,
      "step": 104069
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.8800806403160095,
      "learning_rate": 0.0003453594639456347,
      "loss": 2.5758,
      "step": 104070
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.8431838154792786,
      "learning_rate": 0.00034535542040653753,
      "loss": 2.0135,
      "step": 104071
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.9691519141197205,
      "learning_rate": 0.00034535137685900796,
      "loss": 2.2139,
      "step": 104072
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.8479833006858826,
      "learning_rate": 0.00034534733330304665,
      "loss": 2.0745,
      "step": 104073
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.9463688731193542,
      "learning_rate": 0.00034534328973865447,
      "loss": 2.2338,
      "step": 104074
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.956461489200592,
      "learning_rate": 0.0003453392461658321,
      "loss": 2.0803,
      "step": 104075
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.8386402726173401,
      "learning_rate": 0.0003453352025845803,
      "loss": 2.1384,
      "step": 104076
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.8686890006065369,
      "learning_rate": 0.0003453311589948998,
      "loss": 2.2771,
      "step": 104077
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.8802275657653809,
      "learning_rate": 0.0003453271153967915,
      "loss": 2.2514,
      "step": 104078
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.866849958896637,
      "learning_rate": 0.00034532307179025595,
      "loss": 2.0933,
      "step": 104079
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.8648186922073364,
      "learning_rate": 0.00034531902817529393,
      "loss": 2.1883,
      "step": 104080
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.9903842210769653,
      "learning_rate": 0.00034531498455190636,
      "loss": 2.3897,
      "step": 104081
    },
    {
      "epoch": 1.36,
      "grad_norm": 1.0293145179748535,
      "learning_rate": 0.00034531094092009383,
      "loss": 1.9432,
      "step": 104082
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.8451953530311584,
      "learning_rate": 0.00034530689727985705,
      "loss": 2.2164,
      "step": 104083
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.8497431874275208,
      "learning_rate": 0.000345302853631197,
      "loss": 2.0134,
      "step": 104084
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.8398836851119995,
      "learning_rate": 0.0003452988099741142,
      "loss": 2.1551,
      "step": 104085
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.8771308064460754,
      "learning_rate": 0.00034529476630860956,
      "loss": 1.9961,
      "step": 104086
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.8378835916519165,
      "learning_rate": 0.0003452907226346838,
      "loss": 2.0878,
      "step": 104087
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.804394543170929,
      "learning_rate": 0.0003452866789523376,
      "loss": 2.1226,
      "step": 104088
    },
    {
      "epoch": 1.36,
      "grad_norm": 1.08132803440094,
      "learning_rate": 0.0003452826352615717,
      "loss": 2.0925,
      "step": 104089
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.9658142924308777,
      "learning_rate": 0.000345278591562387,
      "loss": 2.0421,
      "step": 104090
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.8516309857368469,
      "learning_rate": 0.000345274547854784,
      "loss": 1.8298,
      "step": 104091
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.8393357396125793,
      "learning_rate": 0.00034527050413876373,
      "loss": 2.1026,
      "step": 104092
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.9487091898918152,
      "learning_rate": 0.00034526646041432683,
      "loss": 2.0564,
      "step": 104093
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.8325502872467041,
      "learning_rate": 0.00034526241668147393,
      "loss": 2.343,
      "step": 104094
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.8299185633659363,
      "learning_rate": 0.00034525837294020593,
      "loss": 2.1813,
      "step": 104095
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.8814324140548706,
      "learning_rate": 0.00034525432919052366,
      "loss": 2.1811,
      "step": 104096
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.927966833114624,
      "learning_rate": 0.00034525028543242766,
      "loss": 2.5105,
      "step": 104097
    },
    {
      "epoch": 1.36,
      "grad_norm": 1.0791577100753784,
      "learning_rate": 0.0003452462416659188,
      "loss": 2.023,
      "step": 104098
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.8365002870559692,
      "learning_rate": 0.0003452421978909978,
      "loss": 2.1154,
      "step": 104099
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.88837730884552,
      "learning_rate": 0.0003452381541076654,
      "loss": 2.0085,
      "step": 104100
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.9283479452133179,
      "learning_rate": 0.0003452341103159224,
      "loss": 2.1119,
      "step": 104101
    },
    {
      "epoch": 1.36,
      "grad_norm": 1.4008357524871826,
      "learning_rate": 0.00034523006651576954,
      "loss": 2.1774,
      "step": 104102
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.8450425267219543,
      "learning_rate": 0.00034522602270720755,
      "loss": 2.0265,
      "step": 104103
    },
    {
      "epoch": 1.36,
      "grad_norm": 1.0044658184051514,
      "learning_rate": 0.00034522197889023715,
      "loss": 2.358,
      "step": 104104
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.9483307600021362,
      "learning_rate": 0.0003452179350648592,
      "loss": 2.0265,
      "step": 104105
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.8508949279785156,
      "learning_rate": 0.00034521389123107427,
      "loss": 2.1965,
      "step": 104106
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.9189192652702332,
      "learning_rate": 0.0003452098473888833,
      "loss": 2.198,
      "step": 104107
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.8977217674255371,
      "learning_rate": 0.000345205803538287,
      "loss": 2.0598,
      "step": 104108
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.8180071711540222,
      "learning_rate": 0.00034520175967928606,
      "loss": 1.8829,
      "step": 104109
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.8703745603561401,
      "learning_rate": 0.0003451977158118812,
      "loss": 2.2197,
      "step": 104110
    },
    {
      "epoch": 1.36,
      "grad_norm": 1.2711122035980225,
      "learning_rate": 0.00034519367193607334,
      "loss": 2.3194,
      "step": 104111
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.8645989298820496,
      "learning_rate": 0.00034518962805186304,
      "loss": 2.1988,
      "step": 104112
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.8915887475013733,
      "learning_rate": 0.0003451855841592511,
      "loss": 1.993,
      "step": 104113
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.9207841157913208,
      "learning_rate": 0.0003451815402582384,
      "loss": 2.0409,
      "step": 104114
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.8986603617668152,
      "learning_rate": 0.00034517749634882554,
      "loss": 2.1986,
      "step": 104115
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.877260684967041,
      "learning_rate": 0.0003451734524310134,
      "loss": 2.3726,
      "step": 104116
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.9814417958259583,
      "learning_rate": 0.0003451694085048026,
      "loss": 2.2385,
      "step": 104117
    },
    {
      "epoch": 1.36,
      "grad_norm": 1.1920217275619507,
      "learning_rate": 0.000345165364570194,
      "loss": 2.3901,
      "step": 104118
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.9190370440483093,
      "learning_rate": 0.0003451613206271882,
      "loss": 2.1789,
      "step": 104119
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.9966797232627869,
      "learning_rate": 0.00034515727667578614,
      "loss": 2.4608,
      "step": 104120
    },
    {
      "epoch": 1.36,
      "grad_norm": 1.185727596282959,
      "learning_rate": 0.0003451532327159886,
      "loss": 2.4395,
      "step": 104121
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.919662356376648,
      "learning_rate": 0.00034514918874779604,
      "loss": 2.3874,
      "step": 104122
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.9460103511810303,
      "learning_rate": 0.00034514514477120944,
      "loss": 2.3952,
      "step": 104123
    },
    {
      "epoch": 1.36,
      "grad_norm": 1.2715970277786255,
      "learning_rate": 0.00034514110078622955,
      "loss": 2.2298,
      "step": 104124
    },
    {
      "epoch": 1.36,
      "grad_norm": 1.225372552871704,
      "learning_rate": 0.00034513705679285706,
      "loss": 2.1464,
      "step": 104125
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.8476801514625549,
      "learning_rate": 0.0003451330127910927,
      "loss": 2.3045,
      "step": 104126
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.8875923752784729,
      "learning_rate": 0.00034512896878093725,
      "loss": 2.1772,
      "step": 104127
    },
    {
      "epoch": 1.36,
      "grad_norm": 1.011947751045227,
      "learning_rate": 0.00034512492476239154,
      "loss": 2.0407,
      "step": 104128
    },
    {
      "epoch": 1.36,
      "grad_norm": 1.0665279626846313,
      "learning_rate": 0.0003451208807354562,
      "loss": 2.101,
      "step": 104129
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.9116562008857727,
      "learning_rate": 0.0003451168367001321,
      "loss": 2.2088,
      "step": 104130
    },
    {
      "epoch": 1.36,
      "grad_norm": 1.0105398893356323,
      "learning_rate": 0.0003451127926564199,
      "loss": 2.0596,
      "step": 104131
    },
    {
      "epoch": 1.36,
      "grad_norm": 1.2015644311904907,
      "learning_rate": 0.0003451087486043204,
      "loss": 2.2663,
      "step": 104132
    },
    {
      "epoch": 1.36,
      "grad_norm": 1.017809510231018,
      "learning_rate": 0.00034510470454383424,
      "loss": 2.1507,
      "step": 104133
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.8965940475463867,
      "learning_rate": 0.0003451006604749623,
      "loss": 2.1588,
      "step": 104134
    },
    {
      "epoch": 1.36,
      "grad_norm": 1.1985151767730713,
      "learning_rate": 0.00034509661639770537,
      "loss": 2.132,
      "step": 104135
    },
    {
      "epoch": 1.36,
      "grad_norm": 1.0384882688522339,
      "learning_rate": 0.0003450925723120641,
      "loss": 2.3186,
      "step": 104136
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.876328706741333,
      "learning_rate": 0.00034508852821803925,
      "loss": 2.0295,
      "step": 104137
    },
    {
      "epoch": 1.36,
      "grad_norm": 1.0640134811401367,
      "learning_rate": 0.00034508448411563154,
      "loss": 2.0718,
      "step": 104138
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.8933802247047424,
      "learning_rate": 0.0003450804400048419,
      "loss": 2.1063,
      "step": 104139
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.9273322820663452,
      "learning_rate": 0.0003450763958856708,
      "loss": 2.1656,
      "step": 104140
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.8857737183570862,
      "learning_rate": 0.00034507235175811925,
      "loss": 2.2079,
      "step": 104141
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.8806352615356445,
      "learning_rate": 0.0003450683076221879,
      "loss": 2.0688,
      "step": 104142
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.9817662835121155,
      "learning_rate": 0.0003450642634778775,
      "loss": 2.1262,
      "step": 104143
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.9543930292129517,
      "learning_rate": 0.00034506021932518877,
      "loss": 2.0969,
      "step": 104144
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.9010326266288757,
      "learning_rate": 0.0003450561751641225,
      "loss": 2.1314,
      "step": 104145
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.922843873500824,
      "learning_rate": 0.00034505213099467945,
      "loss": 1.86,
      "step": 104146
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.8771277070045471,
      "learning_rate": 0.0003450480868168604,
      "loss": 2.0738,
      "step": 104147
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.915711522102356,
      "learning_rate": 0.000345044042630666,
      "loss": 1.9357,
      "step": 104148
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.8744582533836365,
      "learning_rate": 0.00034503999843609704,
      "loss": 2.0362,
      "step": 104149
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.8549262881278992,
      "learning_rate": 0.00034503595423315435,
      "loss": 2.178,
      "step": 104150
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.8726505637168884,
      "learning_rate": 0.00034503191002183863,
      "loss": 2.0963,
      "step": 104151
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.8259748220443726,
      "learning_rate": 0.00034502786580215054,
      "loss": 2.173,
      "step": 104152
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.8647961020469666,
      "learning_rate": 0.000345023821574091,
      "loss": 1.8846,
      "step": 104153
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.9366759061813354,
      "learning_rate": 0.0003450197773376607,
      "loss": 2.359,
      "step": 104154
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.9192949533462524,
      "learning_rate": 0.00034501573309286035,
      "loss": 2.2705,
      "step": 104155
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.9454835653305054,
      "learning_rate": 0.0003450116888396907,
      "loss": 2.1916,
      "step": 104156
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.7927192449569702,
      "learning_rate": 0.00034500764457815256,
      "loss": 2.2869,
      "step": 104157
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.915805995464325,
      "learning_rate": 0.0003450036003082466,
      "loss": 2.0948,
      "step": 104158
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.8775951266288757,
      "learning_rate": 0.00034499955602997366,
      "loss": 2.0584,
      "step": 104159
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.8377430438995361,
      "learning_rate": 0.00034499551174333446,
      "loss": 1.9243,
      "step": 104160
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.8834714293479919,
      "learning_rate": 0.0003449914674483297,
      "loss": 2.1554,
      "step": 104161
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.9192392230033875,
      "learning_rate": 0.0003449874231449602,
      "loss": 1.9933,
      "step": 104162
    },
    {
      "epoch": 1.36,
      "grad_norm": 1.0770416259765625,
      "learning_rate": 0.00034498337883322677,
      "loss": 2.3211,
      "step": 104163
    },
    {
      "epoch": 1.36,
      "grad_norm": 1.0189906358718872,
      "learning_rate": 0.00034497933451312996,
      "loss": 2.101,
      "step": 104164
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.8106305003166199,
      "learning_rate": 0.0003449752901846707,
      "loss": 2.2321,
      "step": 104165
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.9420619010925293,
      "learning_rate": 0.0003449712458478497,
      "loss": 2.0182,
      "step": 104166
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.888312816619873,
      "learning_rate": 0.0003449672015026677,
      "loss": 1.9323,
      "step": 104167
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.9542002081871033,
      "learning_rate": 0.00034496315714912536,
      "loss": 2.194,
      "step": 104168
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.9194405674934387,
      "learning_rate": 0.0003449591127872236,
      "loss": 2.1799,
      "step": 104169
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.8862060308456421,
      "learning_rate": 0.0003449550684169631,
      "loss": 2.2841,
      "step": 104170
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.9296929240226746,
      "learning_rate": 0.00034495102403834453,
      "loss": 2.1547,
      "step": 104171
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.8706483244895935,
      "learning_rate": 0.0003449469796513688,
      "loss": 2.2623,
      "step": 104172
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.8570572137832642,
      "learning_rate": 0.0003449429352560365,
      "loss": 2.3148,
      "step": 104173
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.8696526288986206,
      "learning_rate": 0.00034493889085234847,
      "loss": 2.2318,
      "step": 104174
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.7890912890434265,
      "learning_rate": 0.00034493484644030557,
      "loss": 2.2161,
      "step": 104175
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.824645459651947,
      "learning_rate": 0.0003449308020199083,
      "loss": 2.0113,
      "step": 104176
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.9915635585784912,
      "learning_rate": 0.0003449267575911576,
      "loss": 2.3053,
      "step": 104177
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.8620980978012085,
      "learning_rate": 0.00034492271315405415,
      "loss": 2.1829,
      "step": 104178
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.8165149092674255,
      "learning_rate": 0.00034491866870859874,
      "loss": 2.1263,
      "step": 104179
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.845904529094696,
      "learning_rate": 0.00034491462425479205,
      "loss": 2.4285,
      "step": 104180
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.9019092917442322,
      "learning_rate": 0.000344910579792635,
      "loss": 2.166,
      "step": 104181
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.909247100353241,
      "learning_rate": 0.00034490653532212806,
      "loss": 2.2755,
      "step": 104182
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.8541257977485657,
      "learning_rate": 0.00034490249084327224,
      "loss": 1.9869,
      "step": 104183
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.9801391363143921,
      "learning_rate": 0.00034489844635606827,
      "loss": 2.2133,
      "step": 104184
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.8656567931175232,
      "learning_rate": 0.00034489440186051675,
      "loss": 2.1777,
      "step": 104185
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.8767040967941284,
      "learning_rate": 0.0003448903573566185,
      "loss": 2.2568,
      "step": 104186
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.819273054599762,
      "learning_rate": 0.00034488631284437435,
      "loss": 2.2996,
      "step": 104187
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.8912479281425476,
      "learning_rate": 0.00034488226832378487,
      "loss": 2.0062,
      "step": 104188
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.8878504037857056,
      "learning_rate": 0.000344878223794851,
      "loss": 2.2346,
      "step": 104189
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.9271354079246521,
      "learning_rate": 0.0003448741792575735,
      "loss": 2.1525,
      "step": 104190
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.9555431604385376,
      "learning_rate": 0.00034487013471195296,
      "loss": 2.1944,
      "step": 104191
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.8659369945526123,
      "learning_rate": 0.0003448660901579902,
      "loss": 2.1463,
      "step": 104192
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.8692161440849304,
      "learning_rate": 0.000344862045595686,
      "loss": 2.2299,
      "step": 104193
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.9129090905189514,
      "learning_rate": 0.00034485800102504113,
      "loss": 2.1434,
      "step": 104194
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.7446790933609009,
      "learning_rate": 0.00034485395644605625,
      "loss": 2.0417,
      "step": 104195
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.8211334347724915,
      "learning_rate": 0.00034484991185873223,
      "loss": 1.8218,
      "step": 104196
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.8721709847450256,
      "learning_rate": 0.00034484586726306977,
      "loss": 2.1538,
      "step": 104197
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.8804314732551575,
      "learning_rate": 0.0003448418226590696,
      "loss": 2.1174,
      "step": 104198
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.8103379011154175,
      "learning_rate": 0.0003448377780467325,
      "loss": 2.133,
      "step": 104199
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.8369250893592834,
      "learning_rate": 0.00034483373342605915,
      "loss": 2.0636,
      "step": 104200
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.9036833047866821,
      "learning_rate": 0.00034482968879705043,
      "loss": 2.1147,
      "step": 104201
    },
    {
      "epoch": 1.36,
      "grad_norm": 1.1139683723449707,
      "learning_rate": 0.0003448256441597069,
      "loss": 2.0709,
      "step": 104202
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.9095147252082825,
      "learning_rate": 0.0003448215995140296,
      "loss": 2.0304,
      "step": 104203
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.8420792818069458,
      "learning_rate": 0.00034481755486001907,
      "loss": 1.8634,
      "step": 104204
    },
    {
      "epoch": 1.36,
      "grad_norm": 1.0568532943725586,
      "learning_rate": 0.000344813510197676,
      "loss": 1.8968,
      "step": 104205
    },
    {
      "epoch": 1.36,
      "grad_norm": 1.17385995388031,
      "learning_rate": 0.00034480946552700145,
      "loss": 2.2517,
      "step": 104206
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.9461387395858765,
      "learning_rate": 0.00034480542084799583,
      "loss": 2.3365,
      "step": 104207
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.9972856640815735,
      "learning_rate": 0.00034480137616066007,
      "loss": 2.0973,
      "step": 104208
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.906603991985321,
      "learning_rate": 0.0003447973314649949,
      "loss": 2.0499,
      "step": 104209
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.8492358326911926,
      "learning_rate": 0.00034479328676100104,
      "loss": 2.0021,
      "step": 104210
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.8450164198875427,
      "learning_rate": 0.00034478924204867935,
      "loss": 1.9022,
      "step": 104211
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.8425950407981873,
      "learning_rate": 0.00034478519732803045,
      "loss": 2.4085,
      "step": 104212
    },
    {
      "epoch": 1.36,
      "grad_norm": 1.1243174076080322,
      "learning_rate": 0.0003447811525990551,
      "loss": 2.3998,
      "step": 104213
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.9032387137413025,
      "learning_rate": 0.00034477710786175403,
      "loss": 1.9497,
      "step": 104214
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.9342382550239563,
      "learning_rate": 0.0003447730631161281,
      "loss": 2.3538,
      "step": 104215
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.9646520018577576,
      "learning_rate": 0.0003447690183621781,
      "loss": 2.0667,
      "step": 104216
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.9214195609092712,
      "learning_rate": 0.00034476497359990454,
      "loss": 2.1958,
      "step": 104217
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.9848027229309082,
      "learning_rate": 0.00034476092882930844,
      "loss": 2.1256,
      "step": 104218
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.9499572515487671,
      "learning_rate": 0.0003447568840503904,
      "loss": 2.347,
      "step": 104219
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.8357728719711304,
      "learning_rate": 0.0003447528392631513,
      "loss": 2.2734,
      "step": 104220
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.8322255611419678,
      "learning_rate": 0.0003447487944675917,
      "loss": 2.2977,
      "step": 104221
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.9062113165855408,
      "learning_rate": 0.0003447447496637124,
      "loss": 2.056,
      "step": 104222
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.9797325134277344,
      "learning_rate": 0.00034474070485151435,
      "loss": 2.1213,
      "step": 104223
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.8993774652481079,
      "learning_rate": 0.00034473666003099806,
      "loss": 1.915,
      "step": 104224
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.8852319717407227,
      "learning_rate": 0.0003447326152021644,
      "loss": 2.0122,
      "step": 104225
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.9209756255149841,
      "learning_rate": 0.0003447285703650141,
      "loss": 1.9582,
      "step": 104226
    },
    {
      "epoch": 1.36,
      "grad_norm": 1.2416337728500366,
      "learning_rate": 0.000344724525519548,
      "loss": 2.1245,
      "step": 104227
    },
    {
      "epoch": 1.36,
      "grad_norm": 1.0333703756332397,
      "learning_rate": 0.00034472048066576665,
      "loss": 2.0558,
      "step": 104228
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.8893260359764099,
      "learning_rate": 0.00034471643580367094,
      "loss": 2.2455,
      "step": 104229
    },
    {
      "epoch": 1.36,
      "grad_norm": 1.2798171043395996,
      "learning_rate": 0.0003447123909332617,
      "loss": 2.1153,
      "step": 104230
    },
    {
      "epoch": 1.36,
      "grad_norm": 1.0555387735366821,
      "learning_rate": 0.00034470834605453946,
      "loss": 2.0749,
      "step": 104231
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.8939752578735352,
      "learning_rate": 0.00034470430116750514,
      "loss": 2.369,
      "step": 104232
    },
    {
      "epoch": 1.36,
      "grad_norm": 1.0242472887039185,
      "learning_rate": 0.00034470025627215945,
      "loss": 1.95,
      "step": 104233
    },
    {
      "epoch": 1.36,
      "grad_norm": 1.2502750158309937,
      "learning_rate": 0.0003446962113685032,
      "loss": 2.3738,
      "step": 104234
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.8496201038360596,
      "learning_rate": 0.00034469216645653693,
      "loss": 2.1954,
      "step": 104235
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.8520134687423706,
      "learning_rate": 0.0003446881215362617,
      "loss": 2.1888,
      "step": 104236
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.9990006685256958,
      "learning_rate": 0.000344684076607678,
      "loss": 2.0729,
      "step": 104237
    },
    {
      "epoch": 1.36,
      "grad_norm": 1.2075271606445312,
      "learning_rate": 0.0003446800316707867,
      "loss": 2.1932,
      "step": 104238
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.8436748385429382,
      "learning_rate": 0.0003446759867255886,
      "loss": 1.8686,
      "step": 104239
    },
    {
      "epoch": 1.36,
      "grad_norm": 1.1106436252593994,
      "learning_rate": 0.00034467194177208437,
      "loss": 2.1917,
      "step": 104240
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.9989900588989258,
      "learning_rate": 0.00034466789681027475,
      "loss": 2.1581,
      "step": 104241
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.8539714813232422,
      "learning_rate": 0.0003446638518401605,
      "loss": 2.0732,
      "step": 104242
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.9338600039482117,
      "learning_rate": 0.00034465980686174244,
      "loss": 2.2218,
      "step": 104243
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.9209067821502686,
      "learning_rate": 0.00034465576187502127,
      "loss": 2.2817,
      "step": 104244
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.8798046112060547,
      "learning_rate": 0.0003446517168799978,
      "loss": 2.2154,
      "step": 104245
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.8223780989646912,
      "learning_rate": 0.00034464767187667263,
      "loss": 2.1188,
      "step": 104246
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.9434332251548767,
      "learning_rate": 0.0003446436268650467,
      "loss": 2.2438,
      "step": 104247
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.8685503602027893,
      "learning_rate": 0.00034463958184512074,
      "loss": 2.306,
      "step": 104248
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.8551766872406006,
      "learning_rate": 0.0003446355368168953,
      "loss": 2.3728,
      "step": 104249
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.9511174559593201,
      "learning_rate": 0.0003446314917803713,
      "loss": 2.2248,
      "step": 104250
    },
    {
      "epoch": 1.36,
      "grad_norm": 1.0123517513275146,
      "learning_rate": 0.0003446274467355495,
      "loss": 2.2331,
      "step": 104251
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.9551084041595459,
      "learning_rate": 0.0003446234016824306,
      "loss": 2.3661,
      "step": 104252
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.9673098921775818,
      "learning_rate": 0.0003446193566210154,
      "loss": 2.3917,
      "step": 104253
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.8971622586250305,
      "learning_rate": 0.00034461531155130455,
      "loss": 2.0769,
      "step": 104254
    },
    {
      "epoch": 1.36,
      "grad_norm": 1.004494071006775,
      "learning_rate": 0.000344611266473299,
      "loss": 2.1886,
      "step": 104255
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.8487215042114258,
      "learning_rate": 0.00034460722138699915,
      "loss": 2.1675,
      "step": 104256
    },
    {
      "epoch": 1.36,
      "grad_norm": 1.1137601137161255,
      "learning_rate": 0.00034460317629240615,
      "loss": 2.0102,
      "step": 104257
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.7809095978736877,
      "learning_rate": 0.0003445991311895206,
      "loss": 1.9029,
      "step": 104258
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.8580820560455322,
      "learning_rate": 0.0003445950860783431,
      "loss": 2.324,
      "step": 104259
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.8719901442527771,
      "learning_rate": 0.0003445910409588746,
      "loss": 2.3358,
      "step": 104260
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.8341807126998901,
      "learning_rate": 0.0003445869958311158,
      "loss": 2.1099,
      "step": 104261
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.8850396275520325,
      "learning_rate": 0.0003445829506950674,
      "loss": 2.1528,
      "step": 104262
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.9159557819366455,
      "learning_rate": 0.00034457890555073017,
      "loss": 2.1944,
      "step": 104263
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.9851928949356079,
      "learning_rate": 0.000344574860398105,
      "loss": 2.1276,
      "step": 104264
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.94305819272995,
      "learning_rate": 0.00034457081523719236,
      "loss": 2.2966,
      "step": 104265
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.8484123945236206,
      "learning_rate": 0.0003445667700679932,
      "loss": 2.333,
      "step": 104266
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.901485800743103,
      "learning_rate": 0.0003445627248905084,
      "loss": 2.2856,
      "step": 104267
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.910376250743866,
      "learning_rate": 0.00034455867970473837,
      "loss": 2.3533,
      "step": 104268
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.8795316815376282,
      "learning_rate": 0.0003445546345106841,
      "loss": 2.0733,
      "step": 104269
    },
    {
      "epoch": 1.36,
      "grad_norm": 1.0843274593353271,
      "learning_rate": 0.0003445505893083463,
      "loss": 2.3125,
      "step": 104270
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.9270513653755188,
      "learning_rate": 0.00034454654409772563,
      "loss": 2.3079,
      "step": 104271
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.823383092880249,
      "learning_rate": 0.00034454249887882295,
      "loss": 2.0698,
      "step": 104272
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.9731114506721497,
      "learning_rate": 0.000344538453651639,
      "loss": 2.0119,
      "step": 104273
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.9929746985435486,
      "learning_rate": 0.00034453440841617454,
      "loss": 2.2413,
      "step": 104274
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.8388345241546631,
      "learning_rate": 0.00034453036317243024,
      "loss": 2.4283,
      "step": 104275
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.846537172794342,
      "learning_rate": 0.00034452631792040696,
      "loss": 2.0252,
      "step": 104276
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.9257421493530273,
      "learning_rate": 0.00034452227266010537,
      "loss": 1.923,
      "step": 104277
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.8368085026741028,
      "learning_rate": 0.00034451822739152617,
      "loss": 2.007,
      "step": 104278
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.8363930583000183,
      "learning_rate": 0.0003445141821146704,
      "loss": 2.1558,
      "step": 104279
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.9454553723335266,
      "learning_rate": 0.0003445101368295384,
      "loss": 2.2251,
      "step": 104280
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.8828439712524414,
      "learning_rate": 0.0003445060915361311,
      "loss": 2.2965,
      "step": 104281
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.8779656887054443,
      "learning_rate": 0.0003445020462344495,
      "loss": 2.0513,
      "step": 104282
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.929266631603241,
      "learning_rate": 0.00034449800092449393,
      "loss": 2.1359,
      "step": 104283
    },
    {
      "epoch": 1.36,
      "grad_norm": 1.1065466403961182,
      "learning_rate": 0.0003444939556062654,
      "loss": 2.3063,
      "step": 104284
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.8528085350990295,
      "learning_rate": 0.0003444899102797646,
      "loss": 2.0087,
      "step": 104285
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.8876297473907471,
      "learning_rate": 0.0003444858649449923,
      "loss": 2.2245,
      "step": 104286
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.8943425416946411,
      "learning_rate": 0.00034448181960194924,
      "loss": 2.3211,
      "step": 104287
    },
    {
      "epoch": 1.36,
      "grad_norm": 1.041496753692627,
      "learning_rate": 0.00034447777425063625,
      "loss": 2.1388,
      "step": 104288
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.8339239954948425,
      "learning_rate": 0.0003444737288910539,
      "loss": 2.2645,
      "step": 104289
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.9008848071098328,
      "learning_rate": 0.000344469683523203,
      "loss": 2.1532,
      "step": 104290
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.9352630972862244,
      "learning_rate": 0.00034446563814708445,
      "loss": 2.1315,
      "step": 104291
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.9140294194221497,
      "learning_rate": 0.00034446159276269885,
      "loss": 1.9687,
      "step": 104292
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.8425377607345581,
      "learning_rate": 0.000344457547370047,
      "loss": 1.8623,
      "step": 104293
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.9575153589248657,
      "learning_rate": 0.00034445350196912966,
      "loss": 2.389,
      "step": 104294
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.9205752611160278,
      "learning_rate": 0.0003444494565599476,
      "loss": 2.3912,
      "step": 104295
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.9057984352111816,
      "learning_rate": 0.00034444541114250147,
      "loss": 2.0531,
      "step": 104296
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.9417630434036255,
      "learning_rate": 0.0003444413657167922,
      "loss": 2.2322,
      "step": 104297
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.8645332455635071,
      "learning_rate": 0.0003444373202828204,
      "loss": 2.0994,
      "step": 104298
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.9110602140426636,
      "learning_rate": 0.00034443327484058687,
      "loss": 2.2504,
      "step": 104299
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.838374137878418,
      "learning_rate": 0.0003444292293900923,
      "loss": 1.9533,
      "step": 104300
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.8783674240112305,
      "learning_rate": 0.0003444251839313376,
      "loss": 2.3201,
      "step": 104301
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.8824843764305115,
      "learning_rate": 0.00034442113846432326,
      "loss": 2.1795,
      "step": 104302
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.8217692375183105,
      "learning_rate": 0.0003444170929890503,
      "loss": 2.0347,
      "step": 104303
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.8359466791152954,
      "learning_rate": 0.0003444130475055194,
      "loss": 2.3195,
      "step": 104304
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.8734397888183594,
      "learning_rate": 0.00034440900201373125,
      "loss": 2.2315,
      "step": 104305
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.8887486457824707,
      "learning_rate": 0.00034440495651368654,
      "loss": 2.288,
      "step": 104306
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.9671005606651306,
      "learning_rate": 0.0003444009110053862,
      "loss": 2.0059,
      "step": 104307
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.9221046566963196,
      "learning_rate": 0.0003443968654888309,
      "loss": 2.5196,
      "step": 104308
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.8749248385429382,
      "learning_rate": 0.0003443928199640214,
      "loss": 2.189,
      "step": 104309
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.958925187587738,
      "learning_rate": 0.0003443887744309584,
      "loss": 2.2577,
      "step": 104310
    },
    {
      "epoch": 1.36,
      "grad_norm": 1.0693451166152954,
      "learning_rate": 0.0003443847288896426,
      "loss": 2.2563,
      "step": 104311
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.9406530261039734,
      "learning_rate": 0.00034438068334007493,
      "loss": 2.1856,
      "step": 104312
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.9122692942619324,
      "learning_rate": 0.00034437663778225607,
      "loss": 1.9546,
      "step": 104313
    },
    {
      "epoch": 1.36,
      "grad_norm": 1.0357086658477783,
      "learning_rate": 0.0003443725922161867,
      "loss": 2.0373,
      "step": 104314
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.8831934332847595,
      "learning_rate": 0.0003443685466418677,
      "loss": 2.0328,
      "step": 104315
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.8296087980270386,
      "learning_rate": 0.0003443645010592997,
      "loss": 2.202,
      "step": 104316
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.861506462097168,
      "learning_rate": 0.0003443604554684835,
      "loss": 2.0792,
      "step": 104317
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.8654764294624329,
      "learning_rate": 0.0003443564098694199,
      "loss": 2.2754,
      "step": 104318
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.929080069065094,
      "learning_rate": 0.0003443523642621096,
      "loss": 2.2253,
      "step": 104319
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.9590145349502563,
      "learning_rate": 0.0003443483186465533,
      "loss": 2.2864,
      "step": 104320
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.9330073595046997,
      "learning_rate": 0.0003443442730227518,
      "loss": 2.1132,
      "step": 104321
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.8797597885131836,
      "learning_rate": 0.00034434022739070593,
      "loss": 2.1913,
      "step": 104322
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.9747802019119263,
      "learning_rate": 0.0003443361817504163,
      "loss": 2.141,
      "step": 104323
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.8324536085128784,
      "learning_rate": 0.00034433213610188387,
      "loss": 2.4192,
      "step": 104324
    },
    {
      "epoch": 1.36,
      "grad_norm": 1.0722824335098267,
      "learning_rate": 0.00034432809044510917,
      "loss": 2.0058,
      "step": 104325
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.8572233319282532,
      "learning_rate": 0.000344324044780093,
      "loss": 2.3724,
      "step": 104326
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.9583659768104553,
      "learning_rate": 0.00034431999910683616,
      "loss": 2.3187,
      "step": 104327
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.9501374363899231,
      "learning_rate": 0.0003443159534253395,
      "loss": 2.4362,
      "step": 104328
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.8550300598144531,
      "learning_rate": 0.00034431190773560357,
      "loss": 2.0574,
      "step": 104329
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.8242319822311401,
      "learning_rate": 0.00034430786203762924,
      "loss": 2.05,
      "step": 104330
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.9038629531860352,
      "learning_rate": 0.0003443038163314173,
      "loss": 2.1735,
      "step": 104331
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.8800516724586487,
      "learning_rate": 0.0003442997706169684,
      "loss": 2.207,
      "step": 104332
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.8374073505401611,
      "learning_rate": 0.00034429572489428327,
      "loss": 2.0342,
      "step": 104333
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.9536840319633484,
      "learning_rate": 0.00034429167916336285,
      "loss": 2.063,
      "step": 104334
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.890025794506073,
      "learning_rate": 0.0003442876334242077,
      "loss": 2.1748,
      "step": 104335
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.8407217264175415,
      "learning_rate": 0.00034428358767681864,
      "loss": 2.2735,
      "step": 104336
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.8653063178062439,
      "learning_rate": 0.00034427954192119646,
      "loss": 2.248,
      "step": 104337
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.9170616269111633,
      "learning_rate": 0.0003442754961573419,
      "loss": 2.102,
      "step": 104338
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.9129036068916321,
      "learning_rate": 0.00034427145038525556,
      "loss": 2.1006,
      "step": 104339
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.9330204725265503,
      "learning_rate": 0.00034426740460493846,
      "loss": 2.1779,
      "step": 104340
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.8929764628410339,
      "learning_rate": 0.0003442633588163912,
      "loss": 2.4817,
      "step": 104341
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.8791602849960327,
      "learning_rate": 0.00034425931301961443,
      "loss": 2.1716,
      "step": 104342
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.9262537360191345,
      "learning_rate": 0.0003442552672146092,
      "loss": 1.9847,
      "step": 104343
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.8198686838150024,
      "learning_rate": 0.0003442512214013759,
      "loss": 1.8754,
      "step": 104344
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.8653417229652405,
      "learning_rate": 0.00034424717557991557,
      "loss": 2.231,
      "step": 104345
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.9039036631584167,
      "learning_rate": 0.0003442431297502289,
      "loss": 2.33,
      "step": 104346
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.9501795172691345,
      "learning_rate": 0.0003442390839123165,
      "loss": 2.2543,
      "step": 104347
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.9064369797706604,
      "learning_rate": 0.0003442350380661792,
      "loss": 2.3318,
      "step": 104348
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.8585633635520935,
      "learning_rate": 0.00034423099221181786,
      "loss": 2.2517,
      "step": 104349
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.8343244194984436,
      "learning_rate": 0.000344226946349233,
      "loss": 1.9694,
      "step": 104350
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.8289769887924194,
      "learning_rate": 0.00034422290047842564,
      "loss": 2.3061,
      "step": 104351
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.9408442378044128,
      "learning_rate": 0.0003442188545993965,
      "loss": 2.1575,
      "step": 104352
    },
    {
      "epoch": 1.36,
      "grad_norm": 1.0185426473617554,
      "learning_rate": 0.00034421480871214605,
      "loss": 2.0566,
      "step": 104353
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.8619874715805054,
      "learning_rate": 0.00034421076281667524,
      "loss": 2.2013,
      "step": 104354
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.8617737889289856,
      "learning_rate": 0.000344206716912985,
      "loss": 2.1437,
      "step": 104355
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.8483268022537231,
      "learning_rate": 0.00034420267100107566,
      "loss": 2.1271,
      "step": 104356
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.9355399012565613,
      "learning_rate": 0.00034419862508094836,
      "loss": 2.2662,
      "step": 104357
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.9518517851829529,
      "learning_rate": 0.00034419457915260373,
      "loss": 1.9885,
      "step": 104358
    },
    {
      "epoch": 1.36,
      "grad_norm": 1.0158674716949463,
      "learning_rate": 0.00034419053321604236,
      "loss": 1.9956,
      "step": 104359
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.9610220193862915,
      "learning_rate": 0.00034418648727126524,
      "loss": 2.1314,
      "step": 104360
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.9006214141845703,
      "learning_rate": 0.00034418244131827295,
      "loss": 2.1803,
      "step": 104361
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.949590802192688,
      "learning_rate": 0.00034417839535706636,
      "loss": 2.3587,
      "step": 104362
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.886488139629364,
      "learning_rate": 0.00034417434938764614,
      "loss": 2.0373,
      "step": 104363
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.9545381665229797,
      "learning_rate": 0.00034417030341001313,
      "loss": 2.2734,
      "step": 104364
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.8319272398948669,
      "learning_rate": 0.000344166257424168,
      "loss": 2.0493,
      "step": 104365
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.8792047500610352,
      "learning_rate": 0.00034416221143011144,
      "loss": 1.9732,
      "step": 104366
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.8886505365371704,
      "learning_rate": 0.00034415816542784444,
      "loss": 2.0559,
      "step": 104367
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.862988293170929,
      "learning_rate": 0.00034415411941736754,
      "loss": 2.1614,
      "step": 104368
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.9237654805183411,
      "learning_rate": 0.0003441500733986815,
      "loss": 2.1675,
      "step": 104369
    },
    {
      "epoch": 1.36,
      "grad_norm": 1.001451849937439,
      "learning_rate": 0.00034414602737178727,
      "loss": 2.1878,
      "step": 104370
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.9947901964187622,
      "learning_rate": 0.0003441419813366853,
      "loss": 2.0857,
      "step": 104371
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.8670941591262817,
      "learning_rate": 0.00034413793529337655,
      "loss": 1.9793,
      "step": 104372
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.9520888924598694,
      "learning_rate": 0.0003441338892418618,
      "loss": 2.2257,
      "step": 104373
    },
    {
      "epoch": 1.36,
      "grad_norm": 1.1857231855392456,
      "learning_rate": 0.0003441298431821416,
      "loss": 1.8879,
      "step": 104374
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.8648536205291748,
      "learning_rate": 0.0003441257971142169,
      "loss": 2.3081,
      "step": 104375
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.9234303832054138,
      "learning_rate": 0.00034412175103808846,
      "loss": 2.264,
      "step": 104376
    },
    {
      "epoch": 1.36,
      "grad_norm": 1.1029311418533325,
      "learning_rate": 0.00034411770495375684,
      "loss": 1.9553,
      "step": 104377
    },
    {
      "epoch": 1.36,
      "grad_norm": 1.0162781476974487,
      "learning_rate": 0.0003441136588612229,
      "loss": 2.1082,
      "step": 104378
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.8535696268081665,
      "learning_rate": 0.00034410961276048753,
      "loss": 2.1802,
      "step": 104379
    },
    {
      "epoch": 1.36,
      "grad_norm": 1.0550568103790283,
      "learning_rate": 0.00034410556665155124,
      "loss": 2.2201,
      "step": 104380
    },
    {
      "epoch": 1.36,
      "grad_norm": 1.284462332725525,
      "learning_rate": 0.0003441015205344148,
      "loss": 2.2256,
      "step": 104381
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.9348188042640686,
      "learning_rate": 0.00034409747440907926,
      "loss": 2.1795,
      "step": 104382
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.9401499032974243,
      "learning_rate": 0.00034409342827554503,
      "loss": 2.007,
      "step": 104383
    },
    {
      "epoch": 1.36,
      "grad_norm": 1.1743426322937012,
      "learning_rate": 0.00034408938213381304,
      "loss": 2.0489,
      "step": 104384
    },
    {
      "epoch": 1.36,
      "grad_norm": 1.0173990726470947,
      "learning_rate": 0.00034408533598388405,
      "loss": 2.1614,
      "step": 104385
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.8761963844299316,
      "learning_rate": 0.0003440812898257587,
      "loss": 2.2354,
      "step": 104386
    },
    {
      "epoch": 1.36,
      "grad_norm": 1.012856125831604,
      "learning_rate": 0.00034407724365943777,
      "loss": 2.1934,
      "step": 104387
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.9489965438842773,
      "learning_rate": 0.0003440731974849221,
      "loss": 2.0491,
      "step": 104388
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.82377690076828,
      "learning_rate": 0.0003440691513022125,
      "loss": 2.0733,
      "step": 104389
    },
    {
      "epoch": 1.36,
      "grad_norm": 1.0045053958892822,
      "learning_rate": 0.0003440651051113094,
      "loss": 2.2894,
      "step": 104390
    },
    {
      "epoch": 1.36,
      "grad_norm": 1.34438955783844,
      "learning_rate": 0.0003440610589122139,
      "loss": 2.204,
      "step": 104391
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.9382531046867371,
      "learning_rate": 0.00034405701270492656,
      "loss": 2.2836,
      "step": 104392
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.8357137441635132,
      "learning_rate": 0.0003440529664894483,
      "loss": 2.1199,
      "step": 104393
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.9259453415870667,
      "learning_rate": 0.0003440489202657796,
      "loss": 1.9799,
      "step": 104394
    },
    {
      "epoch": 1.36,
      "grad_norm": 1.0380287170410156,
      "learning_rate": 0.00034404487403392147,
      "loss": 2.1496,
      "step": 104395
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.9387900829315186,
      "learning_rate": 0.00034404082779387456,
      "loss": 1.9944,
      "step": 104396
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.8308089375495911,
      "learning_rate": 0.0003440367815456396,
      "loss": 2.2225,
      "step": 104397
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.9692738056182861,
      "learning_rate": 0.0003440327352892174,
      "loss": 2.0894,
      "step": 104398
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.9770649075508118,
      "learning_rate": 0.0003440286890246086,
      "loss": 2.5182,
      "step": 104399
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.8287914991378784,
      "learning_rate": 0.0003440246427518142,
      "loss": 1.9746,
      "step": 104400
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.8368139863014221,
      "learning_rate": 0.00034402059647083464,
      "loss": 2.1207,
      "step": 104401
    },
    {
      "epoch": 1.36,
      "grad_norm": 1.0850390195846558,
      "learning_rate": 0.00034401655018167083,
      "loss": 2.031,
      "step": 104402
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.9768994450569153,
      "learning_rate": 0.0003440125038843236,
      "loss": 2.0481,
      "step": 104403
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.8695816397666931,
      "learning_rate": 0.0003440084575787936,
      "loss": 2.103,
      "step": 104404
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.8846014142036438,
      "learning_rate": 0.00034400441126508153,
      "loss": 2.1084,
      "step": 104405
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.9357031583786011,
      "learning_rate": 0.00034400036494318826,
      "loss": 2.1634,
      "step": 104406
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.8969452381134033,
      "learning_rate": 0.0003439963186131145,
      "loss": 1.9968,
      "step": 104407
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.8381344676017761,
      "learning_rate": 0.00034399227227486096,
      "loss": 2.0588,
      "step": 104408
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.8599967360496521,
      "learning_rate": 0.0003439882259284284,
      "loss": 2.2292,
      "step": 104409
    },
    {
      "epoch": 1.36,
      "grad_norm": 1.0886383056640625,
      "learning_rate": 0.0003439841795738177,
      "loss": 2.1889,
      "step": 104410
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.8569456338882446,
      "learning_rate": 0.00034398013321102947,
      "loss": 2.2033,
      "step": 104411
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.9694580435752869,
      "learning_rate": 0.00034397608684006445,
      "loss": 2.0433,
      "step": 104412
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.9625193476676941,
      "learning_rate": 0.00034397204046092357,
      "loss": 2.1546,
      "step": 104413
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.9928145408630371,
      "learning_rate": 0.00034396799407360727,
      "loss": 2.1537,
      "step": 104414
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.8284410238265991,
      "learning_rate": 0.0003439639476781166,
      "loss": 2.171,
      "step": 104415
    },
    {
      "epoch": 1.36,
      "grad_norm": 1.1479544639587402,
      "learning_rate": 0.0003439599012744523,
      "loss": 2.191,
      "step": 104416
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.9658551216125488,
      "learning_rate": 0.00034395585486261484,
      "loss": 2.1555,
      "step": 104417
    },
    {
      "epoch": 1.36,
      "grad_norm": 1.0214378833770752,
      "learning_rate": 0.0003439518084426052,
      "loss": 1.9829,
      "step": 104418
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.8986537456512451,
      "learning_rate": 0.00034394776201442415,
      "loss": 2.0459,
      "step": 104419
    },
    {
      "epoch": 1.36,
      "grad_norm": 1.0443934202194214,
      "learning_rate": 0.00034394371557807237,
      "loss": 2.2365,
      "step": 104420
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.8515704274177551,
      "learning_rate": 0.00034393966913355057,
      "loss": 2.2436,
      "step": 104421
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.8741779327392578,
      "learning_rate": 0.00034393562268085964,
      "loss": 2.0666,
      "step": 104422
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.8670694828033447,
      "learning_rate": 0.00034393157622000017,
      "loss": 2.3352,
      "step": 104423
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.9049208760261536,
      "learning_rate": 0.000343927529750973,
      "loss": 2.1365,
      "step": 104424
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.9429786205291748,
      "learning_rate": 0.00034392348327377895,
      "loss": 2.1021,
      "step": 104425
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.9824033379554749,
      "learning_rate": 0.00034391943678841857,
      "loss": 2.2267,
      "step": 104426
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.900664210319519,
      "learning_rate": 0.00034391539029489284,
      "loss": 2.1281,
      "step": 104427
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.9578198790550232,
      "learning_rate": 0.00034391134379320237,
      "loss": 2.2238,
      "step": 104428
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.8276395201683044,
      "learning_rate": 0.00034390729728334796,
      "loss": 2.2862,
      "step": 104429
    },
    {
      "epoch": 1.36,
      "grad_norm": 1.0472124814987183,
      "learning_rate": 0.0003439032507653303,
      "loss": 2.239,
      "step": 104430
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.9853582978248596,
      "learning_rate": 0.0003438992042391503,
      "loss": 2.0569,
      "step": 104431
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.8826338052749634,
      "learning_rate": 0.00034389515770480856,
      "loss": 1.9723,
      "step": 104432
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.8677815198898315,
      "learning_rate": 0.0003438911111623059,
      "loss": 2.124,
      "step": 104433
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.9936290383338928,
      "learning_rate": 0.0003438870646116431,
      "loss": 2.0433,
      "step": 104434
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.8965116143226624,
      "learning_rate": 0.0003438830180528207,
      "loss": 2.2333,
      "step": 104435
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.8675057888031006,
      "learning_rate": 0.0003438789714858397,
      "loss": 2.2249,
      "step": 104436
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.9134480357170105,
      "learning_rate": 0.00034387492491070085,
      "loss": 2.2624,
      "step": 104437
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.8557217121124268,
      "learning_rate": 0.00034387087832740476,
      "loss": 1.9683,
      "step": 104438
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.9962297677993774,
      "learning_rate": 0.0003438668317359522,
      "loss": 2.4054,
      "step": 104439
    },
    {
      "epoch": 1.36,
      "grad_norm": 1.0542609691619873,
      "learning_rate": 0.0003438627851363441,
      "loss": 2.0685,
      "step": 104440
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.8835738897323608,
      "learning_rate": 0.00034385873852858096,
      "loss": 2.0041,
      "step": 104441
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.8379163146018982,
      "learning_rate": 0.00034385469191266366,
      "loss": 2.3345,
      "step": 104442
    },
    {
      "epoch": 1.36,
      "grad_norm": 1.0620932579040527,
      "learning_rate": 0.00034385064528859305,
      "loss": 2.0036,
      "step": 104443
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.9766561985015869,
      "learning_rate": 0.0003438465986563697,
      "loss": 2.1234,
      "step": 104444
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.8438494801521301,
      "learning_rate": 0.0003438425520159944,
      "loss": 2.2285,
      "step": 104445
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.9023354053497314,
      "learning_rate": 0.000343838505367468,
      "loss": 2.2204,
      "step": 104446
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.8586165904998779,
      "learning_rate": 0.0003438344587107912,
      "loss": 2.0595,
      "step": 104447
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.89313143491745,
      "learning_rate": 0.00034383041204596467,
      "loss": 2.1577,
      "step": 104448
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.8558594584465027,
      "learning_rate": 0.0003438263653729894,
      "loss": 2.0811,
      "step": 104449
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.8387202620506287,
      "learning_rate": 0.0003438223186918658,
      "loss": 2.196,
      "step": 104450
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.8635851144790649,
      "learning_rate": 0.00034381827200259484,
      "loss": 2.4653,
      "step": 104451
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.8521791696548462,
      "learning_rate": 0.00034381422530517736,
      "loss": 1.9769,
      "step": 104452
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.8333319425582886,
      "learning_rate": 0.00034381017859961383,
      "loss": 2.1193,
      "step": 104453
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.9191516041755676,
      "learning_rate": 0.00034380613188590524,
      "loss": 2.2337,
      "step": 104454
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.9433336853981018,
      "learning_rate": 0.0003438020851640523,
      "loss": 2.0534,
      "step": 104455
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.8954829573631287,
      "learning_rate": 0.00034379803843405567,
      "loss": 2.0625,
      "step": 104456
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.9035190939903259,
      "learning_rate": 0.00034379399169591615,
      "loss": 2.0168,
      "step": 104457
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.9515174627304077,
      "learning_rate": 0.00034378994494963454,
      "loss": 2.0392,
      "step": 104458
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.9130548238754272,
      "learning_rate": 0.0003437858981952115,
      "loss": 2.232,
      "step": 104459
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.8943283557891846,
      "learning_rate": 0.0003437818514326479,
      "loss": 2.0839,
      "step": 104460
    },
    {
      "epoch": 1.36,
      "grad_norm": 1.1740972995758057,
      "learning_rate": 0.0003437778046619444,
      "loss": 2.0691,
      "step": 104461
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.8869372606277466,
      "learning_rate": 0.0003437737578831018,
      "loss": 2.0069,
      "step": 104462
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.8804717063903809,
      "learning_rate": 0.0003437697110961208,
      "loss": 2.5158,
      "step": 104463
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.8156939744949341,
      "learning_rate": 0.00034376566430100217,
      "loss": 2.1403,
      "step": 104464
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.8431115746498108,
      "learning_rate": 0.0003437616174977468,
      "loss": 2.0674,
      "step": 104465
    },
    {
      "epoch": 1.36,
      "grad_norm": 1.0385628938674927,
      "learning_rate": 0.00034375757068635514,
      "loss": 1.9602,
      "step": 104466
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.8840585947036743,
      "learning_rate": 0.00034375352386682815,
      "loss": 1.9364,
      "step": 104467
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.8931901454925537,
      "learning_rate": 0.00034374947703916665,
      "loss": 2.1651,
      "step": 104468
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.9028491973876953,
      "learning_rate": 0.00034374543020337125,
      "loss": 2.3517,
      "step": 104469
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.871570348739624,
      "learning_rate": 0.00034374138335944277,
      "loss": 2.4706,
      "step": 104470
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.877537190914154,
      "learning_rate": 0.00034373733650738196,
      "loss": 2.2644,
      "step": 104471
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.856948733329773,
      "learning_rate": 0.00034373328964718946,
      "loss": 2.0725,
      "step": 104472
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.8800350427627563,
      "learning_rate": 0.00034372924277886616,
      "loss": 2.1059,
      "step": 104473
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.9841619729995728,
      "learning_rate": 0.0003437251959024128,
      "loss": 1.9704,
      "step": 104474
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.9582526683807373,
      "learning_rate": 0.0003437211490178301,
      "loss": 2.0352,
      "step": 104475
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.8911653161048889,
      "learning_rate": 0.00034371710212511875,
      "loss": 2.2283,
      "step": 104476
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.8699649572372437,
      "learning_rate": 0.0003437130552242797,
      "loss": 2.2364,
      "step": 104477
    },
    {
      "epoch": 1.36,
      "grad_norm": 1.1320433616638184,
      "learning_rate": 0.0003437090083153134,
      "loss": 1.9588,
      "step": 104478
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.8429677486419678,
      "learning_rate": 0.0003437049613982208,
      "loss": 2.1972,
      "step": 104479
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.9660887718200684,
      "learning_rate": 0.0003437009144730027,
      "loss": 2.1209,
      "step": 104480
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.9236554503440857,
      "learning_rate": 0.0003436968675396598,
      "loss": 2.1715,
      "step": 104481
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.920865535736084,
      "learning_rate": 0.0003436928205981927,
      "loss": 2.11,
      "step": 104482
    },
    {
      "epoch": 1.36,
      "grad_norm": 1.2026556730270386,
      "learning_rate": 0.0003436887736486024,
      "loss": 2.3351,
      "step": 104483
    },
    {
      "epoch": 1.36,
      "grad_norm": 1.1186233758926392,
      "learning_rate": 0.00034368472669088945,
      "loss": 2.1969,
      "step": 104484
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.8753136396408081,
      "learning_rate": 0.00034368067972505474,
      "loss": 2.081,
      "step": 104485
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.9445422887802124,
      "learning_rate": 0.0003436766327510989,
      "loss": 2.1749,
      "step": 104486
    },
    {
      "epoch": 1.36,
      "grad_norm": 1.0206178426742554,
      "learning_rate": 0.0003436725857690228,
      "loss": 2.101,
      "step": 104487
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.9153528213500977,
      "learning_rate": 0.00034366853877882715,
      "loss": 2.2149,
      "step": 104488
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.8740186095237732,
      "learning_rate": 0.00034366449178051267,
      "loss": 2.0323,
      "step": 104489
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.8552651405334473,
      "learning_rate": 0.0003436604447740801,
      "loss": 2.1481,
      "step": 104490
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.9375013113021851,
      "learning_rate": 0.0003436563977595303,
      "loss": 2.2503,
      "step": 104491
    },
    {
      "epoch": 1.36,
      "grad_norm": 1.0608476400375366,
      "learning_rate": 0.000343652350736864,
      "loss": 2.2114,
      "step": 104492
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.9429129362106323,
      "learning_rate": 0.00034364830370608177,
      "loss": 2.1485,
      "step": 104493
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.9134773015975952,
      "learning_rate": 0.00034364425666718457,
      "loss": 2.1338,
      "step": 104494
    },
    {
      "epoch": 1.36,
      "grad_norm": 1.279671311378479,
      "learning_rate": 0.0003436402096201731,
      "loss": 1.9578,
      "step": 104495
    },
    {
      "epoch": 1.36,
      "grad_norm": 1.0985335111618042,
      "learning_rate": 0.00034363616256504804,
      "loss": 2.2311,
      "step": 104496
    },
    {
      "epoch": 1.36,
      "grad_norm": 1.0713626146316528,
      "learning_rate": 0.0003436321155018102,
      "loss": 2.1144,
      "step": 104497
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.8711981177330017,
      "learning_rate": 0.00034362806843046036,
      "loss": 2.0297,
      "step": 104498
    },
    {
      "epoch": 1.36,
      "grad_norm": 1.4501031637191772,
      "learning_rate": 0.0003436240213509992,
      "loss": 2.1199,
      "step": 104499
    },
    {
      "epoch": 1.36,
      "grad_norm": 1.0696327686309814,
      "learning_rate": 0.0003436199742634275,
      "loss": 2.191,
      "step": 104500
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.9727180600166321,
      "learning_rate": 0.0003436159271677461,
      "loss": 2.0214,
      "step": 104501
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.9288794994354248,
      "learning_rate": 0.0003436118800639556,
      "loss": 2.0326,
      "step": 104502
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.9649456143379211,
      "learning_rate": 0.0003436078329520569,
      "loss": 2.1582,
      "step": 104503
    },
    {
      "epoch": 1.36,
      "grad_norm": 1.254074215888977,
      "learning_rate": 0.00034360378583205063,
      "loss": 2.2484,
      "step": 104504
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.9521270394325256,
      "learning_rate": 0.0003435997387039376,
      "loss": 2.0457,
      "step": 104505
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.8802230358123779,
      "learning_rate": 0.0003435956915677185,
      "loss": 2.2269,
      "step": 104506
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.8883374929428101,
      "learning_rate": 0.0003435916444233943,
      "loss": 2.176,
      "step": 104507
    },
    {
      "epoch": 1.36,
      "grad_norm": 1.1534526348114014,
      "learning_rate": 0.00034358759727096545,
      "loss": 2.0732,
      "step": 104508
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.9815986156463623,
      "learning_rate": 0.0003435835501104328,
      "loss": 2.0409,
      "step": 104509
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.8733547329902649,
      "learning_rate": 0.0003435795029417973,
      "loss": 2.2774,
      "step": 104510
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.908324658870697,
      "learning_rate": 0.0003435754557650594,
      "loss": 2.1174,
      "step": 104511
    },
    {
      "epoch": 1.36,
      "grad_norm": 1.1015253067016602,
      "learning_rate": 0.0003435714085802201,
      "loss": 2.4947,
      "step": 104512
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.9021843075752258,
      "learning_rate": 0.00034356736138728004,
      "loss": 2.2561,
      "step": 104513
    },
    {
      "epoch": 1.36,
      "grad_norm": 1.0722663402557373,
      "learning_rate": 0.0003435633141862399,
      "loss": 1.9892,
      "step": 104514
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.8647483587265015,
      "learning_rate": 0.00034355926697710065,
      "loss": 2.1043,
      "step": 104515
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.9518237113952637,
      "learning_rate": 0.0003435552197598629,
      "loss": 2.2608,
      "step": 104516
    },
    {
      "epoch": 1.36,
      "grad_norm": 1.4758332967758179,
      "learning_rate": 0.0003435511725345273,
      "loss": 2.2116,
      "step": 104517
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.9147453308105469,
      "learning_rate": 0.0003435471253010948,
      "loss": 2.0823,
      "step": 104518
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.8582720160484314,
      "learning_rate": 0.00034354307805956605,
      "loss": 2.2631,
      "step": 104519
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.9251633286476135,
      "learning_rate": 0.00034353903080994175,
      "loss": 2.0179,
      "step": 104520
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.9712053537368774,
      "learning_rate": 0.0003435349835522228,
      "loss": 1.9875,
      "step": 104521
    },
    {
      "epoch": 1.36,
      "grad_norm": 1.03416907787323,
      "learning_rate": 0.00034353093628640993,
      "loss": 2.16,
      "step": 104522
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.8820942640304565,
      "learning_rate": 0.00034352688901250367,
      "loss": 2.1826,
      "step": 104523
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.981112003326416,
      "learning_rate": 0.0003435228417305051,
      "loss": 2.0186,
      "step": 104524
    },
    {
      "epoch": 1.36,
      "grad_norm": 1.06689453125,
      "learning_rate": 0.0003435187944404147,
      "loss": 2.1755,
      "step": 104525
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.817859411239624,
      "learning_rate": 0.0003435147471422334,
      "loss": 2.2839,
      "step": 104526
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.8327040672302246,
      "learning_rate": 0.0003435106998359619,
      "loss": 2.311,
      "step": 104527
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.9221220016479492,
      "learning_rate": 0.0003435066525216008,
      "loss": 2.419,
      "step": 104528
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.9463211297988892,
      "learning_rate": 0.0003435026051991512,
      "loss": 2.38,
      "step": 104529
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.9148015975952148,
      "learning_rate": 0.0003434985578686135,
      "loss": 2.127,
      "step": 104530
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.9094772934913635,
      "learning_rate": 0.00034349451052998863,
      "loss": 2.2007,
      "step": 104531
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.8871210217475891,
      "learning_rate": 0.0003434904631832774,
      "loss": 2.2584,
      "step": 104532
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.9373549222946167,
      "learning_rate": 0.0003434864158284803,
      "loss": 2.3128,
      "step": 104533
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.9488090872764587,
      "learning_rate": 0.00034348236846559835,
      "loss": 2.3718,
      "step": 104534
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.8816648721694946,
      "learning_rate": 0.00034347832109463225,
      "loss": 2.2202,
      "step": 104535
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.9686968922615051,
      "learning_rate": 0.00034347427371558267,
      "loss": 2.0632,
      "step": 104536
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.9036991000175476,
      "learning_rate": 0.00034347022632845036,
      "loss": 2.2371,
      "step": 104537
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.9389192461967468,
      "learning_rate": 0.0003434661789332362,
      "loss": 2.2174,
      "step": 104538
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.8588056564331055,
      "learning_rate": 0.0003434621315299408,
      "loss": 2.0243,
      "step": 104539
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.9676125645637512,
      "learning_rate": 0.00034345808411856497,
      "loss": 2.338,
      "step": 104540
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.9696446061134338,
      "learning_rate": 0.0003434540366991095,
      "loss": 2.0641,
      "step": 104541
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.8265761137008667,
      "learning_rate": 0.00034344998927157504,
      "loss": 1.9617,
      "step": 104542
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.9216710925102234,
      "learning_rate": 0.0003434459418359624,
      "loss": 2.1826,
      "step": 104543
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.9643678665161133,
      "learning_rate": 0.0003434418943922724,
      "loss": 2.2437,
      "step": 104544
    },
    {
      "epoch": 1.36,
      "grad_norm": 1.10728120803833,
      "learning_rate": 0.0003434378469405058,
      "loss": 2.1393,
      "step": 104545
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.8541377782821655,
      "learning_rate": 0.00034343379948066307,
      "loss": 2.1916,
      "step": 104546
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.8659640550613403,
      "learning_rate": 0.0003434297520127454,
      "loss": 2.2981,
      "step": 104547
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.9634110331535339,
      "learning_rate": 0.00034342570453675317,
      "loss": 2.2256,
      "step": 104548
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.9565485715866089,
      "learning_rate": 0.00034342165705268733,
      "loss": 2.0009,
      "step": 104549
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.8293758630752563,
      "learning_rate": 0.0003434176095605486,
      "loss": 2.0477,
      "step": 104550
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.9030287861824036,
      "learning_rate": 0.00034341356206033766,
      "loss": 2.149,
      "step": 104551
    },
    {
      "epoch": 1.36,
      "grad_norm": 1.0408967733383179,
      "learning_rate": 0.0003434095145520554,
      "loss": 2.1246,
      "step": 104552
    },
    {
      "epoch": 1.36,
      "grad_norm": 1.1227989196777344,
      "learning_rate": 0.00034340546703570245,
      "loss": 2.1661,
      "step": 104553
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.9720864295959473,
      "learning_rate": 0.0003434014195112796,
      "loss": 2.2727,
      "step": 104554
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.9002785682678223,
      "learning_rate": 0.0003433973719787876,
      "loss": 2.2667,
      "step": 104555
    },
    {
      "epoch": 1.36,
      "grad_norm": 1.0016655921936035,
      "learning_rate": 0.0003433933244382273,
      "loss": 2.1122,
      "step": 104556
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.9666067361831665,
      "learning_rate": 0.0003433892768895993,
      "loss": 2.1849,
      "step": 104557
    },
    {
      "epoch": 1.36,
      "grad_norm": 1.0470412969589233,
      "learning_rate": 0.00034338522933290437,
      "loss": 2.1353,
      "step": 104558
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.9637776613235474,
      "learning_rate": 0.0003433811817681433,
      "loss": 2.068,
      "step": 104559
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.8873411417007446,
      "learning_rate": 0.0003433771341953169,
      "loss": 2.2062,
      "step": 104560
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.8326921463012695,
      "learning_rate": 0.0003433730866144258,
      "loss": 2.1555,
      "step": 104561
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.94199138879776,
      "learning_rate": 0.00034336903902547095,
      "loss": 2.1946,
      "step": 104562
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.8870912194252014,
      "learning_rate": 0.0003433649914284529,
      "loss": 2.314,
      "step": 104563
    },
    {
      "epoch": 1.36,
      "grad_norm": 1.0931148529052734,
      "learning_rate": 0.0003433609438233725,
      "loss": 2.4495,
      "step": 104564
    },
    {
      "epoch": 1.36,
      "grad_norm": 1.0294095277786255,
      "learning_rate": 0.00034335689621023044,
      "loss": 2.0179,
      "step": 104565
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.9389997720718384,
      "learning_rate": 0.0003433528485890276,
      "loss": 1.9796,
      "step": 104566
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.8816097378730774,
      "learning_rate": 0.00034334880095976454,
      "loss": 2.2029,
      "step": 104567
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.8996949791908264,
      "learning_rate": 0.00034334475332244217,
      "loss": 1.8197,
      "step": 104568
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.9869848489761353,
      "learning_rate": 0.00034334070567706125,
      "loss": 2.0316,
      "step": 104569
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.9023219347000122,
      "learning_rate": 0.00034333665802362243,
      "loss": 2.1814,
      "step": 104570
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.9488848447799683,
      "learning_rate": 0.00034333261036212647,
      "loss": 2.2132,
      "step": 104571
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.8340548276901245,
      "learning_rate": 0.0003433285626925742,
      "loss": 1.8257,
      "step": 104572
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.817330539226532,
      "learning_rate": 0.00034332451501496627,
      "loss": 2.2923,
      "step": 104573
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.9549697637557983,
      "learning_rate": 0.0003433204673293036,
      "loss": 1.945,
      "step": 104574
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.7964352965354919,
      "learning_rate": 0.00034331641963558673,
      "loss": 2.0705,
      "step": 104575
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.9822198748588562,
      "learning_rate": 0.0003433123719338166,
      "loss": 2.0733,
      "step": 104576
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.9136847853660583,
      "learning_rate": 0.00034330832422399386,
      "loss": 2.2773,
      "step": 104577
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.9389266967773438,
      "learning_rate": 0.0003433042765061192,
      "loss": 2.1903,
      "step": 104578
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.9418291449546814,
      "learning_rate": 0.00034330022878019357,
      "loss": 2.047,
      "step": 104579
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.8201209306716919,
      "learning_rate": 0.0003432961810462176,
      "loss": 1.8315,
      "step": 104580
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.9163196682929993,
      "learning_rate": 0.00034329213330419204,
      "loss": 2.0933,
      "step": 104581
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.9075296521186829,
      "learning_rate": 0.00034328808555411763,
      "loss": 2.0166,
      "step": 104582
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.8366791605949402,
      "learning_rate": 0.00034328403779599524,
      "loss": 2.164,
      "step": 104583
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.8986117839813232,
      "learning_rate": 0.00034327999002982536,
      "loss": 2.1426,
      "step": 104584
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.8160516619682312,
      "learning_rate": 0.000343275942255609,
      "loss": 2.2756,
      "step": 104585
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.8775726556777954,
      "learning_rate": 0.00034327189447334687,
      "loss": 2.153,
      "step": 104586
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.846971869468689,
      "learning_rate": 0.00034326784668303967,
      "loss": 2.165,
      "step": 104587
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.9609295129776001,
      "learning_rate": 0.00034326379888468807,
      "loss": 2.0725,
      "step": 104588
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.8635637164115906,
      "learning_rate": 0.00034325975107829304,
      "loss": 1.9685,
      "step": 104589
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.8566458821296692,
      "learning_rate": 0.00034325570326385513,
      "loss": 2.1982,
      "step": 104590
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.8710500001907349,
      "learning_rate": 0.0003432516554413751,
      "loss": 2.1979,
      "step": 104591
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.9210518598556519,
      "learning_rate": 0.00034324760761085397,
      "loss": 2.0829,
      "step": 104592
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.9983537793159485,
      "learning_rate": 0.0003432435597722921,
      "loss": 2.2124,
      "step": 104593
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.8250354528427124,
      "learning_rate": 0.0003432395119256905,
      "loss": 2.1591,
      "step": 104594
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.9231072664260864,
      "learning_rate": 0.0003432354640710498,
      "loss": 2.015,
      "step": 104595
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.8749488592147827,
      "learning_rate": 0.000343231416208371,
      "loss": 2.0633,
      "step": 104596
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.9661178588867188,
      "learning_rate": 0.00034322736833765447,
      "loss": 2.4305,
      "step": 104597
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.8319380879402161,
      "learning_rate": 0.00034322332045890125,
      "loss": 1.9888,
      "step": 104598
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.9187709093093872,
      "learning_rate": 0.00034321927257211204,
      "loss": 2.1825,
      "step": 104599
    },
    {
      "epoch": 1.36,
      "grad_norm": 1.0825895071029663,
      "learning_rate": 0.0003432152246772874,
      "loss": 2.2371,
      "step": 104600
    },
    {
      "epoch": 1.36,
      "grad_norm": 1.089187741279602,
      "learning_rate": 0.0003432111767744283,
      "loss": 2.0372,
      "step": 104601
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.905288577079773,
      "learning_rate": 0.0003432071288635355,
      "loss": 2.0874,
      "step": 104602
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.9473776817321777,
      "learning_rate": 0.0003432030809446096,
      "loss": 1.9944,
      "step": 104603
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.9890720248222351,
      "learning_rate": 0.00034319903301765147,
      "loss": 2.0514,
      "step": 104604
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.9534056782722473,
      "learning_rate": 0.0003431949850826618,
      "loss": 2.0959,
      "step": 104605
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.8747648000717163,
      "learning_rate": 0.00034319093713964144,
      "loss": 2.1698,
      "step": 104606
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.9746960401535034,
      "learning_rate": 0.00034318688918859097,
      "loss": 2.4854,
      "step": 104607
    },
    {
      "epoch": 1.36,
      "grad_norm": 1.2509126663208008,
      "learning_rate": 0.00034318284122951127,
      "loss": 1.8813,
      "step": 104608
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.7829001545906067,
      "learning_rate": 0.0003431787932624031,
      "loss": 1.8594,
      "step": 104609
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.8036813735961914,
      "learning_rate": 0.00034317474528726714,
      "loss": 2.1723,
      "step": 104610
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.8904659152030945,
      "learning_rate": 0.0003431706973041042,
      "loss": 2.1379,
      "step": 104611
    },
    {
      "epoch": 1.36,
      "grad_norm": 1.3412082195281982,
      "learning_rate": 0.000343166649312915,
      "loss": 2.1712,
      "step": 104612
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.8393624424934387,
      "learning_rate": 0.0003431626013137003,
      "loss": 2.2805,
      "step": 104613
    },
    {
      "epoch": 1.36,
      "grad_norm": 1.0152801275253296,
      "learning_rate": 0.0003431585533064609,
      "loss": 2.0118,
      "step": 104614
    },
    {
      "epoch": 1.36,
      "grad_norm": 1.0121983289718628,
      "learning_rate": 0.0003431545052911975,
      "loss": 2.0993,
      "step": 104615
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.7850813269615173,
      "learning_rate": 0.00034315045726791084,
      "loss": 2.2398,
      "step": 104616
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.8393896222114563,
      "learning_rate": 0.0003431464092366017,
      "loss": 2.206,
      "step": 104617
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.8871505260467529,
      "learning_rate": 0.00034314236119727077,
      "loss": 2.2799,
      "step": 104618
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.8730263113975525,
      "learning_rate": 0.0003431383131499189,
      "loss": 2.0238,
      "step": 104619
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.8862823843955994,
      "learning_rate": 0.0003431342650945469,
      "loss": 2.5417,
      "step": 104620
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.8558267951011658,
      "learning_rate": 0.00034313021703115534,
      "loss": 2.2199,
      "step": 104621
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.8500820994377136,
      "learning_rate": 0.000343126168959745,
      "loss": 2.0534,
      "step": 104622
    },
    {
      "epoch": 1.36,
      "grad_norm": 1.0612598657608032,
      "learning_rate": 0.0003431221208803168,
      "loss": 1.8957,
      "step": 104623
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.9425238966941833,
      "learning_rate": 0.00034311807279287134,
      "loss": 2.113,
      "step": 104624
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.9072261452674866,
      "learning_rate": 0.00034311402469740937,
      "loss": 2.3379,
      "step": 104625
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.8312849998474121,
      "learning_rate": 0.0003431099765939318,
      "loss": 2.22,
      "step": 104626
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.9460166096687317,
      "learning_rate": 0.00034310592848243917,
      "loss": 2.1418,
      "step": 104627
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.9475997686386108,
      "learning_rate": 0.00034310188036293235,
      "loss": 2.2844,
      "step": 104628
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.8213220834732056,
      "learning_rate": 0.0003430978322354122,
      "loss": 1.9753,
      "step": 104629
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.8975902199745178,
      "learning_rate": 0.0003430937840998792,
      "loss": 2.0198,
      "step": 104630
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.8939328789710999,
      "learning_rate": 0.0003430897359563342,
      "loss": 2.2016,
      "step": 104631
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.9436609148979187,
      "learning_rate": 0.00034308568780477816,
      "loss": 2.1073,
      "step": 104632
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.8660857081413269,
      "learning_rate": 0.00034308163964521165,
      "loss": 2.2385,
      "step": 104633
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.777106523513794,
      "learning_rate": 0.0003430775914776354,
      "loss": 2.1715,
      "step": 104634
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.880504846572876,
      "learning_rate": 0.00034307354330205027,
      "loss": 2.3536,
      "step": 104635
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.7797117233276367,
      "learning_rate": 0.0003430694951184569,
      "loss": 2.1399,
      "step": 104636
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.9008627533912659,
      "learning_rate": 0.00034306544692685613,
      "loss": 2.2498,
      "step": 104637
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.8592478632926941,
      "learning_rate": 0.0003430613987272487,
      "loss": 2.1462,
      "step": 104638
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.8517701029777527,
      "learning_rate": 0.00034305735051963525,
      "loss": 2.1562,
      "step": 104639
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.8875588178634644,
      "learning_rate": 0.0003430533023040167,
      "loss": 2.1375,
      "step": 104640
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.849147617816925,
      "learning_rate": 0.0003430492540803937,
      "loss": 2.2481,
      "step": 104641
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.8834007382392883,
      "learning_rate": 0.000343045205848767,
      "loss": 2.0508,
      "step": 104642
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.8816559910774231,
      "learning_rate": 0.0003430411576091374,
      "loss": 2.1835,
      "step": 104643
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.8261557221412659,
      "learning_rate": 0.0003430371093615058,
      "loss": 2.2124,
      "step": 104644
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.7972180843353271,
      "learning_rate": 0.0003430330611058726,
      "loss": 2.364,
      "step": 104645
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.83038330078125,
      "learning_rate": 0.0003430290128422388,
      "loss": 2.1487,
      "step": 104646
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.9498569965362549,
      "learning_rate": 0.0003430249645706051,
      "loss": 2.1637,
      "step": 104647
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.959904670715332,
      "learning_rate": 0.00034302091629097217,
      "loss": 2.2483,
      "step": 104648
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.9150230288505554,
      "learning_rate": 0.00034301686800334094,
      "loss": 2.1848,
      "step": 104649
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.8450998663902283,
      "learning_rate": 0.0003430128197077121,
      "loss": 2.0687,
      "step": 104650
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.9090068340301514,
      "learning_rate": 0.00034300877140408623,
      "loss": 1.9509,
      "step": 104651
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.8312807083129883,
      "learning_rate": 0.0003430047230924643,
      "loss": 2.0822,
      "step": 104652
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.8950293064117432,
      "learning_rate": 0.0003430006747728469,
      "loss": 2.2015,
      "step": 104653
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.9278160929679871,
      "learning_rate": 0.00034299662644523497,
      "loss": 2.1274,
      "step": 104654
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.8579943180084229,
      "learning_rate": 0.0003429925781096291,
      "loss": 2.0922,
      "step": 104655
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.8181191086769104,
      "learning_rate": 0.00034298852976603005,
      "loss": 2.1531,
      "step": 104656
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.875190794467926,
      "learning_rate": 0.00034298448141443876,
      "loss": 2.284,
      "step": 104657
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.8547647595405579,
      "learning_rate": 0.00034298043305485575,
      "loss": 1.9777,
      "step": 104658
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.9471500515937805,
      "learning_rate": 0.00034297638468728186,
      "loss": 1.9813,
      "step": 104659
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.900583028793335,
      "learning_rate": 0.0003429723363117179,
      "loss": 2.3886,
      "step": 104660
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.9299806952476501,
      "learning_rate": 0.0003429682879281645,
      "loss": 2.0703,
      "step": 104661
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.882041335105896,
      "learning_rate": 0.00034296423953662254,
      "loss": 2.1458,
      "step": 104662
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.9674188494682312,
      "learning_rate": 0.00034296019113709265,
      "loss": 2.1182,
      "step": 104663
    },
    {
      "epoch": 1.36,
      "grad_norm": 1.0545063018798828,
      "learning_rate": 0.0003429561427295757,
      "loss": 2.1834,
      "step": 104664
    },
    {
      "epoch": 1.36,
      "grad_norm": 1.277172565460205,
      "learning_rate": 0.0003429520943140724,
      "loss": 2.1259,
      "step": 104665
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.8419874906539917,
      "learning_rate": 0.00034294804589058347,
      "loss": 2.3236,
      "step": 104666
    },
    {
      "epoch": 1.36,
      "grad_norm": 1.0197820663452148,
      "learning_rate": 0.0003429439974591097,
      "loss": 2.1609,
      "step": 104667
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.977548360824585,
      "learning_rate": 0.0003429399490196519,
      "loss": 2.1029,
      "step": 104668
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.9836965203285217,
      "learning_rate": 0.00034293590057221066,
      "loss": 2.1022,
      "step": 104669
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.8426653146743774,
      "learning_rate": 0.00034293185211678684,
      "loss": 2.2314,
      "step": 104670
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.940375030040741,
      "learning_rate": 0.0003429278036533811,
      "loss": 1.9994,
      "step": 104671
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.8578681349754333,
      "learning_rate": 0.0003429237551819945,
      "loss": 2.1515,
      "step": 104672
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.9281249642372131,
      "learning_rate": 0.00034291970670262737,
      "loss": 2.1885,
      "step": 104673
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.9320061206817627,
      "learning_rate": 0.0003429156582152807,
      "loss": 2.3201,
      "step": 104674
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.9056050181388855,
      "learning_rate": 0.00034291160971995524,
      "loss": 2.0121,
      "step": 104675
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.9825049638748169,
      "learning_rate": 0.00034290756121665164,
      "loss": 2.1839,
      "step": 104676
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.9480145573616028,
      "learning_rate": 0.00034290351270537076,
      "loss": 2.0223,
      "step": 104677
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.9029080867767334,
      "learning_rate": 0.0003428994641861133,
      "loss": 2.2091,
      "step": 104678
    },
    {
      "epoch": 1.36,
      "grad_norm": 1.1302179098129272,
      "learning_rate": 0.00034289541565888,
      "loss": 2.3013,
      "step": 104679
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.8996073007583618,
      "learning_rate": 0.00034289136712367164,
      "loss": 2.1007,
      "step": 104680
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.8901967406272888,
      "learning_rate": 0.00034288731858048903,
      "loss": 2.3143,
      "step": 104681
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.9524012804031372,
      "learning_rate": 0.00034288327002933276,
      "loss": 2.1055,
      "step": 104682
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.87184739112854,
      "learning_rate": 0.0003428792214702037,
      "loss": 1.9497,
      "step": 104683
    },
    {
      "epoch": 1.36,
      "grad_norm": 1.0708080530166626,
      "learning_rate": 0.00034287517290310267,
      "loss": 1.9957,
      "step": 104684
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.9980147480964661,
      "learning_rate": 0.0003428711243280302,
      "loss": 2.2034,
      "step": 104685
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.9561323523521423,
      "learning_rate": 0.0003428670757449872,
      "loss": 2.1704,
      "step": 104686
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.818374514579773,
      "learning_rate": 0.00034286302715397454,
      "loss": 1.9856,
      "step": 104687
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.8830657005310059,
      "learning_rate": 0.0003428589785549927,
      "loss": 2.1576,
      "step": 104688
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.9251055121421814,
      "learning_rate": 0.0003428549299480426,
      "loss": 2.3073,
      "step": 104689
    },
    {
      "epoch": 1.36,
      "grad_norm": 1.0596187114715576,
      "learning_rate": 0.000342850881333125,
      "loss": 2.1708,
      "step": 104690
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.8681593537330627,
      "learning_rate": 0.0003428468327102406,
      "loss": 1.9013,
      "step": 104691
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.8945856690406799,
      "learning_rate": 0.00034284278407939017,
      "loss": 2.3239,
      "step": 104692
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.9149158000946045,
      "learning_rate": 0.0003428387354405744,
      "loss": 1.9063,
      "step": 104693
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.9377583861351013,
      "learning_rate": 0.0003428346867937941,
      "loss": 1.9824,
      "step": 104694
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.8860605359077454,
      "learning_rate": 0.00034283063813905003,
      "loss": 2.2711,
      "step": 104695
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.9210764169692993,
      "learning_rate": 0.000342826589476343,
      "loss": 2.1351,
      "step": 104696
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.8564526438713074,
      "learning_rate": 0.0003428225408056737,
      "loss": 2.2514,
      "step": 104697
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.9117650985717773,
      "learning_rate": 0.00034281849212704284,
      "loss": 2.0773,
      "step": 104698
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.8770020008087158,
      "learning_rate": 0.00034281444344045123,
      "loss": 2.1549,
      "step": 104699
    },
    {
      "epoch": 1.36,
      "grad_norm": 1.030532717704773,
      "learning_rate": 0.0003428103947458996,
      "loss": 2.4726,
      "step": 104700
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.867681086063385,
      "learning_rate": 0.0003428063460433887,
      "loss": 2.2888,
      "step": 104701
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.908904492855072,
      "learning_rate": 0.00034280229733291926,
      "loss": 2.1904,
      "step": 104702
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.8889241814613342,
      "learning_rate": 0.0003427982486144921,
      "loss": 2.2233,
      "step": 104703
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.9850786328315735,
      "learning_rate": 0.00034279419988810796,
      "loss": 2.2591,
      "step": 104704
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.8625666499137878,
      "learning_rate": 0.0003427901511537676,
      "loss": 2.4041,
      "step": 104705
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.976077139377594,
      "learning_rate": 0.00034278610241147164,
      "loss": 2.3883,
      "step": 104706
    },
    {
      "epoch": 1.36,
      "grad_norm": 1.016418218612671,
      "learning_rate": 0.000342782053661221,
      "loss": 2.2314,
      "step": 104707
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.9741951823234558,
      "learning_rate": 0.0003427780049030164,
      "loss": 2.1664,
      "step": 104708
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.8299299478530884,
      "learning_rate": 0.0003427739561368585,
      "loss": 2.0711,
      "step": 104709
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.8922903537750244,
      "learning_rate": 0.0003427699073627481,
      "loss": 2.0678,
      "step": 104710
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.875532329082489,
      "learning_rate": 0.000342765858580686,
      "loss": 2.0615,
      "step": 104711
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.8679497241973877,
      "learning_rate": 0.00034276180979067296,
      "loss": 1.9907,
      "step": 104712
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.8402084112167358,
      "learning_rate": 0.0003427577609927096,
      "loss": 2.1468,
      "step": 104713
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.8922256231307983,
      "learning_rate": 0.0003427537121867969,
      "loss": 2.2779,
      "step": 104714
    },
    {
      "epoch": 1.36,
      "grad_norm": 1.0358545780181885,
      "learning_rate": 0.00034274966337293535,
      "loss": 2.1808,
      "step": 104715
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.900268018245697,
      "learning_rate": 0.0003427456145511259,
      "loss": 2.1821,
      "step": 104716
    },
    {
      "epoch": 1.36,
      "grad_norm": 1.009911060333252,
      "learning_rate": 0.0003427415657213692,
      "loss": 1.995,
      "step": 104717
    },
    {
      "epoch": 1.36,
      "grad_norm": 1.2182726860046387,
      "learning_rate": 0.00034273751688366604,
      "loss": 2.152,
      "step": 104718
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.8943233489990234,
      "learning_rate": 0.00034273346803801716,
      "loss": 1.9477,
      "step": 104719
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.8509635329246521,
      "learning_rate": 0.00034272941918442334,
      "loss": 2.122,
      "step": 104720
    },
    {
      "epoch": 1.36,
      "grad_norm": 1.021536111831665,
      "learning_rate": 0.0003427253703228853,
      "loss": 2.1217,
      "step": 104721
    },
    {
      "epoch": 1.36,
      "grad_norm": 1.0302271842956543,
      "learning_rate": 0.0003427213214534038,
      "loss": 2.3864,
      "step": 104722
    },
    {
      "epoch": 1.36,
      "grad_norm": 1.0356844663619995,
      "learning_rate": 0.00034271727257597967,
      "loss": 2.0448,
      "step": 104723
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.8636422753334045,
      "learning_rate": 0.00034271322369061353,
      "loss": 2.1302,
      "step": 104724
    },
    {
      "epoch": 1.36,
      "grad_norm": 1.0848984718322754,
      "learning_rate": 0.0003427091747973062,
      "loss": 2.1632,
      "step": 104725
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.8439134359359741,
      "learning_rate": 0.0003427051258960584,
      "loss": 2.412,
      "step": 104726
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.791267991065979,
      "learning_rate": 0.0003427010769868709,
      "loss": 1.9663,
      "step": 104727
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.8611876368522644,
      "learning_rate": 0.0003426970280697446,
      "loss": 2.2926,
      "step": 104728
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.841577410697937,
      "learning_rate": 0.00034269297914468,
      "loss": 2.0844,
      "step": 104729
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.9300891160964966,
      "learning_rate": 0.000342688930211678,
      "loss": 2.011,
      "step": 104730
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.8930085897445679,
      "learning_rate": 0.00034268488127073927,
      "loss": 1.9876,
      "step": 104731
    },
    {
      "epoch": 1.36,
      "grad_norm": 1.2604540586471558,
      "learning_rate": 0.0003426808323218647,
      "loss": 2.0067,
      "step": 104732
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.9537439346313477,
      "learning_rate": 0.00034267678336505494,
      "loss": 2.0287,
      "step": 104733
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.9349929094314575,
      "learning_rate": 0.0003426727344003107,
      "loss": 2.2085,
      "step": 104734
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.9233977198600769,
      "learning_rate": 0.00034266868542763286,
      "loss": 2.081,
      "step": 104735
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.894655704498291,
      "learning_rate": 0.0003426646364470221,
      "loss": 2.1288,
      "step": 104736
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.8943525552749634,
      "learning_rate": 0.0003426605874584791,
      "loss": 2.0816,
      "step": 104737
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.9551821351051331,
      "learning_rate": 0.00034265653846200474,
      "loss": 2.2743,
      "step": 104738
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.9132353067398071,
      "learning_rate": 0.0003426524894575998,
      "loss": 2.1633,
      "step": 104739
    },
    {
      "epoch": 1.36,
      "grad_norm": 1.2723441123962402,
      "learning_rate": 0.00034264844044526487,
      "loss": 2.0529,
      "step": 104740
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.922415018081665,
      "learning_rate": 0.0003426443914250008,
      "loss": 2.287,
      "step": 104741
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.9239178895950317,
      "learning_rate": 0.0003426403423968084,
      "loss": 2.2073,
      "step": 104742
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.8738502264022827,
      "learning_rate": 0.00034263629336068823,
      "loss": 1.9742,
      "step": 104743
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.9131990671157837,
      "learning_rate": 0.0003426322443166412,
      "loss": 2.504,
      "step": 104744
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.8340638875961304,
      "learning_rate": 0.0003426281952646681,
      "loss": 2.3453,
      "step": 104745
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.9908307194709778,
      "learning_rate": 0.00034262414620476957,
      "loss": 2.1308,
      "step": 104746
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.8606910109519958,
      "learning_rate": 0.0003426200971369464,
      "loss": 2.2561,
      "step": 104747
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.8547644019126892,
      "learning_rate": 0.0003426160480611994,
      "loss": 2.2211,
      "step": 104748
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.8363927602767944,
      "learning_rate": 0.00034261199897752927,
      "loss": 2.0353,
      "step": 104749
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.8697701692581177,
      "learning_rate": 0.0003426079498859367,
      "loss": 2.2104,
      "step": 104750
    },
    {
      "epoch": 1.36,
      "grad_norm": 1.008380651473999,
      "learning_rate": 0.00034260390078642253,
      "loss": 2.1399,
      "step": 104751
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.8501943945884705,
      "learning_rate": 0.00034259985167898744,
      "loss": 2.0734,
      "step": 104752
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.8545557260513306,
      "learning_rate": 0.00034259580256363237,
      "loss": 2.0796,
      "step": 104753
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.8818799257278442,
      "learning_rate": 0.00034259175344035785,
      "loss": 2.1044,
      "step": 104754
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.940180778503418,
      "learning_rate": 0.0003425877043091647,
      "loss": 2.244,
      "step": 104755
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.9505763053894043,
      "learning_rate": 0.0003425836551700537,
      "loss": 2.1268,
      "step": 104756
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.8405218720436096,
      "learning_rate": 0.0003425796060230257,
      "loss": 2.5227,
      "step": 104757
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.9605136513710022,
      "learning_rate": 0.00034257555686808123,
      "loss": 2.2269,
      "step": 104758
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.877515971660614,
      "learning_rate": 0.00034257150770522114,
      "loss": 2.1985,
      "step": 104759
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.9122867584228516,
      "learning_rate": 0.00034256745853444636,
      "loss": 2.2548,
      "step": 104760
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.9314054250717163,
      "learning_rate": 0.00034256340935575735,
      "loss": 2.1338,
      "step": 104761
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.884723961353302,
      "learning_rate": 0.00034255936016915496,
      "loss": 2.1721,
      "step": 104762
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.8462978005409241,
      "learning_rate": 0.00034255531097464017,
      "loss": 2.066,
      "step": 104763
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.9197004437446594,
      "learning_rate": 0.0003425512617722134,
      "loss": 2.2197,
      "step": 104764
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.8542550206184387,
      "learning_rate": 0.00034254721256187556,
      "loss": 1.9784,
      "step": 104765
    },
    {
      "epoch": 1.36,
      "grad_norm": 1.0309736728668213,
      "learning_rate": 0.0003425431633436274,
      "loss": 2.2354,
      "step": 104766
    },
    {
      "epoch": 1.36,
      "grad_norm": 1.0120303630828857,
      "learning_rate": 0.0003425391141174697,
      "loss": 2.4289,
      "step": 104767
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.920200765132904,
      "learning_rate": 0.0003425350648834031,
      "loss": 2.0577,
      "step": 104768
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.877630889415741,
      "learning_rate": 0.0003425310156414285,
      "loss": 1.9761,
      "step": 104769
    },
    {
      "epoch": 1.36,
      "grad_norm": 1.0708931684494019,
      "learning_rate": 0.0003425269663915466,
      "loss": 1.7335,
      "step": 104770
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.9152357578277588,
      "learning_rate": 0.00034252291713375804,
      "loss": 2.2988,
      "step": 104771
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.8852457404136658,
      "learning_rate": 0.00034251886786806375,
      "loss": 2.3833,
      "step": 104772
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.8910915851593018,
      "learning_rate": 0.00034251481859446433,
      "loss": 2.1685,
      "step": 104773
    },
    {
      "epoch": 1.36,
      "grad_norm": 1.1834499835968018,
      "learning_rate": 0.00034251076931296064,
      "loss": 2.0572,
      "step": 104774
    },
    {
      "epoch": 1.36,
      "grad_norm": 1.0669714212417603,
      "learning_rate": 0.00034250672002355344,
      "loss": 2.1104,
      "step": 104775
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.9329902529716492,
      "learning_rate": 0.0003425026707262434,
      "loss": 2.2317,
      "step": 104776
    },
    {
      "epoch": 1.36,
      "grad_norm": 1.04485023021698,
      "learning_rate": 0.0003424986214210312,
      "loss": 2.0482,
      "step": 104777
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.8608264923095703,
      "learning_rate": 0.0003424945721079179,
      "loss": 2.0929,
      "step": 104778
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.943252444267273,
      "learning_rate": 0.0003424905227869039,
      "loss": 2.0758,
      "step": 104779
    },
    {
      "epoch": 1.36,
      "grad_norm": 1.017999291419983,
      "learning_rate": 0.0003424864734579902,
      "loss": 2.1431,
      "step": 104780
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.8720254302024841,
      "learning_rate": 0.0003424824241211775,
      "loss": 2.2533,
      "step": 104781
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.8994837999343872,
      "learning_rate": 0.0003424783747764664,
      "loss": 2.0719,
      "step": 104782
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.8575490713119507,
      "learning_rate": 0.0003424743254238578,
      "loss": 2.0205,
      "step": 104783
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.994835376739502,
      "learning_rate": 0.00034247027606335245,
      "loss": 2.2125,
      "step": 104784
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.8647803664207458,
      "learning_rate": 0.0003424662266949511,
      "loss": 2.2536,
      "step": 104785
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.9057521224021912,
      "learning_rate": 0.0003424621773186544,
      "loss": 2.1837,
      "step": 104786
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.9451169371604919,
      "learning_rate": 0.0003424581279344633,
      "loss": 1.9345,
      "step": 104787
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.8229593634605408,
      "learning_rate": 0.00034245407854237834,
      "loss": 2.0623,
      "step": 104788
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.8305239677429199,
      "learning_rate": 0.0003424500291424003,
      "loss": 2.0071,
      "step": 104789
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.9400737285614014,
      "learning_rate": 0.00034244597973453014,
      "loss": 2.0757,
      "step": 104790
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.9367561340332031,
      "learning_rate": 0.0003424419303187684,
      "loss": 2.1737,
      "step": 104791
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.9026094079017639,
      "learning_rate": 0.0003424378808951159,
      "loss": 2.1289,
      "step": 104792
    },
    {
      "epoch": 1.36,
      "grad_norm": 1.0889127254486084,
      "learning_rate": 0.00034243383146357335,
      "loss": 1.9513,
      "step": 104793
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.9031553268432617,
      "learning_rate": 0.0003424297820241417,
      "loss": 1.9139,
      "step": 104794
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.8461360931396484,
      "learning_rate": 0.00034242573257682146,
      "loss": 2.0007,
      "step": 104795
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.8752743601799011,
      "learning_rate": 0.00034242168312161346,
      "loss": 2.4173,
      "step": 104796
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.918390691280365,
      "learning_rate": 0.00034241763365851853,
      "loss": 2.1266,
      "step": 104797
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.8661690950393677,
      "learning_rate": 0.0003424135841875373,
      "loss": 2.0089,
      "step": 104798
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.8380450010299683,
      "learning_rate": 0.0003424095347086706,
      "loss": 2.1169,
      "step": 104799
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.8804355263710022,
      "learning_rate": 0.0003424054852219192,
      "loss": 2.0356,
      "step": 104800
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.8891299962997437,
      "learning_rate": 0.0003424014357272838,
      "loss": 2.3008,
      "step": 104801
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.9214855432510376,
      "learning_rate": 0.00034239738622476504,
      "loss": 1.9707,
      "step": 104802
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.8523989319801331,
      "learning_rate": 0.00034239333671436405,
      "loss": 2.2163,
      "step": 104803
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.9598169922828674,
      "learning_rate": 0.0003423892871960811,
      "loss": 2.067,
      "step": 104804
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.85821533203125,
      "learning_rate": 0.00034238523766991734,
      "loss": 2.0501,
      "step": 104805
    },
    {
      "epoch": 1.36,
      "grad_norm": 1.0121406316757202,
      "learning_rate": 0.00034238118813587336,
      "loss": 2.2256,
      "step": 104806
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.8481581211090088,
      "learning_rate": 0.00034237713859394984,
      "loss": 2.2213,
      "step": 104807
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.8868175745010376,
      "learning_rate": 0.00034237308904414764,
      "loss": 2.1749,
      "step": 104808
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.87522953748703,
      "learning_rate": 0.0003423690394864675,
      "loss": 1.9417,
      "step": 104809
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.9111196994781494,
      "learning_rate": 0.00034236498992091017,
      "loss": 2.0565,
      "step": 104810
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.8826591372489929,
      "learning_rate": 0.0003423609403474763,
      "loss": 2.1728,
      "step": 104811
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.9318596124649048,
      "learning_rate": 0.0003423568907661669,
      "loss": 2.1277,
      "step": 104812
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.8746623992919922,
      "learning_rate": 0.0003423528411769824,
      "loss": 2.0399,
      "step": 104813
    },
    {
      "epoch": 1.36,
      "grad_norm": 1.0440900325775146,
      "learning_rate": 0.00034234879157992376,
      "loss": 2.3875,
      "step": 104814
    },
    {
      "epoch": 1.36,
      "grad_norm": 1.190923810005188,
      "learning_rate": 0.0003423447419749917,
      "loss": 2.1157,
      "step": 104815
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.9130246639251709,
      "learning_rate": 0.00034234069236218696,
      "loss": 2.2116,
      "step": 104816
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.8180124163627625,
      "learning_rate": 0.00034233664274151024,
      "loss": 2.0444,
      "step": 104817
    },
    {
      "epoch": 1.36,
      "grad_norm": 1.1769052743911743,
      "learning_rate": 0.0003423325931129624,
      "loss": 2.2421,
      "step": 104818
    },
    {
      "epoch": 1.36,
      "grad_norm": 1.1164995431900024,
      "learning_rate": 0.0003423285434765441,
      "loss": 2.0973,
      "step": 104819
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.8958364725112915,
      "learning_rate": 0.00034232449383225604,
      "loss": 2.2327,
      "step": 104820
    },
    {
      "epoch": 1.36,
      "grad_norm": 1.0224190950393677,
      "learning_rate": 0.00034232044418009913,
      "loss": 2.1969,
      "step": 104821
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.9732798337936401,
      "learning_rate": 0.0003423163945200741,
      "loss": 1.7817,
      "step": 104822
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.8132263422012329,
      "learning_rate": 0.00034231234485218153,
      "loss": 2.148,
      "step": 104823
    },
    {
      "epoch": 1.36,
      "grad_norm": 1.0036219358444214,
      "learning_rate": 0.00034230829517642247,
      "loss": 2.3022,
      "step": 104824
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.9051971435546875,
      "learning_rate": 0.00034230424549279735,
      "loss": 2.1392,
      "step": 104825
    },
    {
      "epoch": 1.36,
      "grad_norm": 1.0426520109176636,
      "learning_rate": 0.00034230019580130716,
      "loss": 1.989,
      "step": 104826
    },
    {
      "epoch": 1.36,
      "grad_norm": 1.3185871839523315,
      "learning_rate": 0.0003422961461019525,
      "loss": 2.0368,
      "step": 104827
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.8284090161323547,
      "learning_rate": 0.00034229209639473427,
      "loss": 2.0451,
      "step": 104828
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.8811830282211304,
      "learning_rate": 0.00034228804667965304,
      "loss": 2.1521,
      "step": 104829
    },
    {
      "epoch": 1.36,
      "grad_norm": 0.9983415603637695,
      "learning_rate": 0.0003422839969567097,
      "loss": 2.2113,
      "step": 104830
    },
    {
      "epoch": 1.36,
      "grad_norm": 1.027655839920044,
      "learning_rate": 0.000342279947225905,
      "loss": 1.9545,
      "step": 104831
    },
    {
      "epoch": 1.36,
      "grad_norm": 1.0108219385147095,
      "learning_rate": 0.0003422758974872396,
      "loss": 2.0558,
      "step": 104832
    },
    {
      "epoch": 1.36,
      "grad_norm": 1.1199995279312134,
      "learning_rate": 0.00034227184774071433,
      "loss": 2.1818,
      "step": 104833
    },
    {
      "epoch": 1.37,
      "grad_norm": 1.0045552253723145,
      "learning_rate": 0.00034226779798632995,
      "loss": 2.2226,
      "step": 104834
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.9868437051773071,
      "learning_rate": 0.0003422637482240872,
      "loss": 2.4008,
      "step": 104835
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.8226099014282227,
      "learning_rate": 0.00034225969845398684,
      "loss": 2.3154,
      "step": 104836
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.8625010848045349,
      "learning_rate": 0.0003422556486760295,
      "loss": 2.1103,
      "step": 104837
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.8542978167533875,
      "learning_rate": 0.00034225159889021613,
      "loss": 2.1591,
      "step": 104838
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.905627965927124,
      "learning_rate": 0.00034224754909654737,
      "loss": 1.9967,
      "step": 104839
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.9624574184417725,
      "learning_rate": 0.000342243499295024,
      "loss": 2.2147,
      "step": 104840
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.874072253704071,
      "learning_rate": 0.0003422394494856468,
      "loss": 2.4216,
      "step": 104841
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.9006741642951965,
      "learning_rate": 0.00034223539966841645,
      "loss": 2.2716,
      "step": 104842
    },
    {
      "epoch": 1.37,
      "grad_norm": 1.0720078945159912,
      "learning_rate": 0.0003422313498433337,
      "loss": 2.2256,
      "step": 104843
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.9192943572998047,
      "learning_rate": 0.00034222730001039935,
      "loss": 2.3378,
      "step": 104844
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.8480350971221924,
      "learning_rate": 0.00034222325016961425,
      "loss": 2.0045,
      "step": 104845
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.8946819305419922,
      "learning_rate": 0.00034221920032097893,
      "loss": 2.2954,
      "step": 104846
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.8397250175476074,
      "learning_rate": 0.00034221515046449426,
      "loss": 2.0458,
      "step": 104847
    },
    {
      "epoch": 1.37,
      "grad_norm": 1.1102339029312134,
      "learning_rate": 0.0003422111006001611,
      "loss": 2.1778,
      "step": 104848
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.9174938797950745,
      "learning_rate": 0.00034220705072798006,
      "loss": 2.1466,
      "step": 104849
    },
    {
      "epoch": 1.37,
      "grad_norm": 1.0338127613067627,
      "learning_rate": 0.00034220300084795193,
      "loss": 2.2793,
      "step": 104850
    },
    {
      "epoch": 1.37,
      "grad_norm": 1.264326572418213,
      "learning_rate": 0.00034219895096007744,
      "loss": 2.1012,
      "step": 104851
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.8889267444610596,
      "learning_rate": 0.0003421949010643574,
      "loss": 2.1752,
      "step": 104852
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.8675603270530701,
      "learning_rate": 0.0003421908511607925,
      "loss": 2.1163,
      "step": 104853
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.9732601642608643,
      "learning_rate": 0.0003421868012493836,
      "loss": 2.0451,
      "step": 104854
    },
    {
      "epoch": 1.37,
      "grad_norm": 1.1701654195785522,
      "learning_rate": 0.0003421827513301313,
      "loss": 1.9748,
      "step": 104855
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.9698779582977295,
      "learning_rate": 0.0003421787014030364,
      "loss": 2.4102,
      "step": 104856
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.9080132246017456,
      "learning_rate": 0.00034217465146809984,
      "loss": 2.1885,
      "step": 104857
    },
    {
      "epoch": 1.37,
      "grad_norm": 1.165466070175171,
      "learning_rate": 0.00034217060152532205,
      "loss": 2.2285,
      "step": 104858
    },
    {
      "epoch": 1.37,
      "grad_norm": 1.1967976093292236,
      "learning_rate": 0.000342166551574704,
      "loss": 2.3234,
      "step": 104859
    },
    {
      "epoch": 1.37,
      "grad_norm": 1.006858229637146,
      "learning_rate": 0.0003421625016162464,
      "loss": 2.0854,
      "step": 104860
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.8551278114318848,
      "learning_rate": 0.00034215845164995004,
      "loss": 2.019,
      "step": 104861
    },
    {
      "epoch": 1.37,
      "grad_norm": 1.0311099290847778,
      "learning_rate": 0.00034215440167581555,
      "loss": 2.1595,
      "step": 104862
    },
    {
      "epoch": 1.37,
      "grad_norm": 1.2762845754623413,
      "learning_rate": 0.00034215035169384375,
      "loss": 2.3074,
      "step": 104863
    },
    {
      "epoch": 1.37,
      "grad_norm": 1.0870513916015625,
      "learning_rate": 0.0003421463017040355,
      "loss": 2.0277,
      "step": 104864
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.8571540117263794,
      "learning_rate": 0.0003421422517063914,
      "loss": 1.904,
      "step": 104865
    },
    {
      "epoch": 1.37,
      "grad_norm": 1.115644931793213,
      "learning_rate": 0.00034213820170091223,
      "loss": 2.1077,
      "step": 104866
    },
    {
      "epoch": 1.37,
      "grad_norm": 1.0210835933685303,
      "learning_rate": 0.0003421341516875989,
      "loss": 2.0745,
      "step": 104867
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.9986028075218201,
      "learning_rate": 0.0003421301016664519,
      "loss": 2.3829,
      "step": 104868
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.8529683947563171,
      "learning_rate": 0.00034212605163747213,
      "loss": 2.1717,
      "step": 104869
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.9248512983322144,
      "learning_rate": 0.0003421220016006604,
      "loss": 2.1533,
      "step": 104870
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.8838097453117371,
      "learning_rate": 0.0003421179515560174,
      "loss": 2.1423,
      "step": 104871
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.9678062796592712,
      "learning_rate": 0.00034211390150354375,
      "loss": 2.1464,
      "step": 104872
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.9591724276542664,
      "learning_rate": 0.0003421098514432405,
      "loss": 2.1845,
      "step": 104873
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.8097015619277954,
      "learning_rate": 0.0003421058013751082,
      "loss": 2.0997,
      "step": 104874
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.9099205732345581,
      "learning_rate": 0.0003421017512991475,
      "loss": 1.9612,
      "step": 104875
    },
    {
      "epoch": 1.37,
      "grad_norm": 1.0553114414215088,
      "learning_rate": 0.0003420977012153594,
      "loss": 2.27,
      "step": 104876
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.9831666946411133,
      "learning_rate": 0.0003420936511237445,
      "loss": 2.0557,
      "step": 104877
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.874974250793457,
      "learning_rate": 0.00034208960102430363,
      "loss": 1.9971,
      "step": 104878
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.9219703674316406,
      "learning_rate": 0.0003420855509170376,
      "loss": 2.0816,
      "step": 104879
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.9995768070220947,
      "learning_rate": 0.00034208150080194694,
      "loss": 2.1842,
      "step": 104880
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.9280738234519958,
      "learning_rate": 0.0003420774506790325,
      "loss": 2.3404,
      "step": 104881
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.8637232780456543,
      "learning_rate": 0.00034207340054829514,
      "loss": 2.0533,
      "step": 104882
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.8055769801139832,
      "learning_rate": 0.00034206935040973554,
      "loss": 2.0218,
      "step": 104883
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.9541252851486206,
      "learning_rate": 0.0003420653002633544,
      "loss": 2.082,
      "step": 104884
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.9087453484535217,
      "learning_rate": 0.0003420612501091527,
      "loss": 2.3042,
      "step": 104885
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.8892679214477539,
      "learning_rate": 0.00034205719994713084,
      "loss": 2.1202,
      "step": 104886
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.8847424387931824,
      "learning_rate": 0.00034205314977728973,
      "loss": 2.1951,
      "step": 104887
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.8404808044433594,
      "learning_rate": 0.0003420490995996303,
      "loss": 2.2775,
      "step": 104888
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.9712596535682678,
      "learning_rate": 0.0003420450494141531,
      "loss": 2.1026,
      "step": 104889
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.9444950222969055,
      "learning_rate": 0.0003420409992208588,
      "loss": 2.0621,
      "step": 104890
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.8957239985466003,
      "learning_rate": 0.0003420369490197484,
      "loss": 2.0831,
      "step": 104891
    },
    {
      "epoch": 1.37,
      "grad_norm": 1.200990915298462,
      "learning_rate": 0.0003420328988108225,
      "loss": 2.0498,
      "step": 104892
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.9016168117523193,
      "learning_rate": 0.0003420288485940819,
      "loss": 2.1202,
      "step": 104893
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.8612833619117737,
      "learning_rate": 0.00034202479836952746,
      "loss": 2.3812,
      "step": 104894
    },
    {
      "epoch": 1.37,
      "grad_norm": 1.0114339590072632,
      "learning_rate": 0.0003420207481371597,
      "loss": 2.2697,
      "step": 104895
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.912651002407074,
      "learning_rate": 0.00034201669789697946,
      "loss": 2.0554,
      "step": 104896
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.9598044157028198,
      "learning_rate": 0.0003420126476489876,
      "loss": 2.2862,
      "step": 104897
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.8073103427886963,
      "learning_rate": 0.00034200859739318476,
      "loss": 2.0492,
      "step": 104898
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.8234939575195312,
      "learning_rate": 0.00034200454712957166,
      "loss": 2.0652,
      "step": 104899
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.8713805675506592,
      "learning_rate": 0.0003420004968581493,
      "loss": 2.3099,
      "step": 104900
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.865777313709259,
      "learning_rate": 0.0003419964465789181,
      "loss": 2.0515,
      "step": 104901
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.8231232166290283,
      "learning_rate": 0.000341992396291879,
      "loss": 2.0985,
      "step": 104902
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.9317395687103271,
      "learning_rate": 0.0003419883459970328,
      "loss": 2.1795,
      "step": 104903
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.8530188798904419,
      "learning_rate": 0.0003419842956943801,
      "loss": 1.9483,
      "step": 104904
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.8977480530738831,
      "learning_rate": 0.00034198024538392165,
      "loss": 2.3949,
      "step": 104905
    },
    {
      "epoch": 1.37,
      "grad_norm": 1.0750932693481445,
      "learning_rate": 0.00034197619506565847,
      "loss": 2.3014,
      "step": 104906
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.8181314468383789,
      "learning_rate": 0.000341972144739591,
      "loss": 2.1335,
      "step": 104907
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.8336837291717529,
      "learning_rate": 0.0003419680944057201,
      "loss": 2.2107,
      "step": 104908
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.8814674019813538,
      "learning_rate": 0.00034196404406404657,
      "loss": 2.1568,
      "step": 104909
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.8705272674560547,
      "learning_rate": 0.00034195999371457116,
      "loss": 1.9754,
      "step": 104910
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.8389512300491333,
      "learning_rate": 0.00034195594335729453,
      "loss": 2.2197,
      "step": 104911
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.8779077529907227,
      "learning_rate": 0.0003419518929922176,
      "loss": 2.1636,
      "step": 104912
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.9533011317253113,
      "learning_rate": 0.00034194784261934094,
      "loss": 2.2702,
      "step": 104913
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.9216870665550232,
      "learning_rate": 0.0003419437922386654,
      "loss": 2.1665,
      "step": 104914
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.9881739020347595,
      "learning_rate": 0.00034193974185019177,
      "loss": 1.9709,
      "step": 104915
    },
    {
      "epoch": 1.37,
      "grad_norm": 1.3490674495697021,
      "learning_rate": 0.0003419356914539206,
      "loss": 2.4637,
      "step": 104916
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.8604651093482971,
      "learning_rate": 0.000341931641049853,
      "loss": 1.9763,
      "step": 104917
    },
    {
      "epoch": 1.37,
      "grad_norm": 1.0206389427185059,
      "learning_rate": 0.0003419275906379893,
      "loss": 2.099,
      "step": 104918
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.9037783741950989,
      "learning_rate": 0.0003419235402183306,
      "loss": 2.1444,
      "step": 104919
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.9994903206825256,
      "learning_rate": 0.00034191948979087755,
      "loss": 2.0724,
      "step": 104920
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.8906804919242859,
      "learning_rate": 0.0003419154393556308,
      "loss": 2.1145,
      "step": 104921
    },
    {
      "epoch": 1.37,
      "grad_norm": 1.0426921844482422,
      "learning_rate": 0.0003419113889125913,
      "loss": 2.1313,
      "step": 104922
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.8392193913459778,
      "learning_rate": 0.00034190733846175956,
      "loss": 2.0462,
      "step": 104923
    },
    {
      "epoch": 1.37,
      "grad_norm": 1.0332285165786743,
      "learning_rate": 0.00034190328800313643,
      "loss": 2.2512,
      "step": 104924
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.8767758011817932,
      "learning_rate": 0.0003418992375367227,
      "loss": 2.407,
      "step": 104925
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.7385926842689514,
      "learning_rate": 0.00034189518706251917,
      "loss": 2.0906,
      "step": 104926
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.8942164182662964,
      "learning_rate": 0.0003418911365805266,
      "loss": 2.1292,
      "step": 104927
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.9145551323890686,
      "learning_rate": 0.0003418870860907455,
      "loss": 2.2433,
      "step": 104928
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.9578608870506287,
      "learning_rate": 0.0003418830355931769,
      "loss": 2.1299,
      "step": 104929
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.8019828796386719,
      "learning_rate": 0.00034187898508782153,
      "loss": 2.0811,
      "step": 104930
    },
    {
      "epoch": 1.37,
      "grad_norm": 1.007460355758667,
      "learning_rate": 0.0003418749345746799,
      "loss": 2.0796,
      "step": 104931
    },
    {
      "epoch": 1.37,
      "grad_norm": 1.0864050388336182,
      "learning_rate": 0.00034187088405375303,
      "loss": 2.1693,
      "step": 104932
    },
    {
      "epoch": 1.37,
      "grad_norm": 1.0274794101715088,
      "learning_rate": 0.0003418668335250416,
      "loss": 2.0996,
      "step": 104933
    },
    {
      "epoch": 1.37,
      "grad_norm": 1.1247882843017578,
      "learning_rate": 0.00034186278298854625,
      "loss": 2.149,
      "step": 104934
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.8278002738952637,
      "learning_rate": 0.00034185873244426784,
      "loss": 2.1193,
      "step": 104935
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.9336608052253723,
      "learning_rate": 0.0003418546818922071,
      "loss": 2.1513,
      "step": 104936
    },
    {
      "epoch": 1.37,
      "grad_norm": 1.0531013011932373,
      "learning_rate": 0.0003418506313323649,
      "loss": 2.0956,
      "step": 104937
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.8551723957061768,
      "learning_rate": 0.00034184658076474167,
      "loss": 2.0814,
      "step": 104938
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.8510938882827759,
      "learning_rate": 0.00034184253018933853,
      "loss": 2.3762,
      "step": 104939
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.8637583255767822,
      "learning_rate": 0.0003418384796061561,
      "loss": 2.2699,
      "step": 104940
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.9174222946166992,
      "learning_rate": 0.00034183442901519496,
      "loss": 2.195,
      "step": 104941
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.8406504988670349,
      "learning_rate": 0.00034183037841645605,
      "loss": 2.0457,
      "step": 104942
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.8743927478790283,
      "learning_rate": 0.0003418263278099402,
      "loss": 2.0148,
      "step": 104943
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.8745906352996826,
      "learning_rate": 0.0003418222771956479,
      "loss": 2.0683,
      "step": 104944
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.870039701461792,
      "learning_rate": 0.00034181822657358013,
      "loss": 2.0523,
      "step": 104945
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.8605341911315918,
      "learning_rate": 0.00034181417594373757,
      "loss": 2.1416,
      "step": 104946
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.8517887592315674,
      "learning_rate": 0.0003418101253061209,
      "loss": 2.2807,
      "step": 104947
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.9211676120758057,
      "learning_rate": 0.000341806074660731,
      "loss": 2.1083,
      "step": 104948
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.8881479501724243,
      "learning_rate": 0.0003418020240075686,
      "loss": 1.9906,
      "step": 104949
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.8861917853355408,
      "learning_rate": 0.0003417979733466343,
      "loss": 2.2489,
      "step": 104950
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.8206145763397217,
      "learning_rate": 0.000341793922677929,
      "loss": 1.9984,
      "step": 104951
    },
    {
      "epoch": 1.37,
      "grad_norm": 1.048068881034851,
      "learning_rate": 0.0003417898720014535,
      "loss": 2.2473,
      "step": 104952
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.9343592524528503,
      "learning_rate": 0.00034178582131720836,
      "loss": 2.3899,
      "step": 104953
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.9299870133399963,
      "learning_rate": 0.0003417817706251945,
      "loss": 2.2624,
      "step": 104954
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.9360056519508362,
      "learning_rate": 0.0003417777199254127,
      "loss": 2.2425,
      "step": 104955
    },
    {
      "epoch": 1.37,
      "grad_norm": 1.0170328617095947,
      "learning_rate": 0.0003417736692178635,
      "loss": 2.2209,
      "step": 104956
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.8861998915672302,
      "learning_rate": 0.00034176961850254786,
      "loss": 2.0204,
      "step": 104957
    },
    {
      "epoch": 1.37,
      "grad_norm": 1.1858106851577759,
      "learning_rate": 0.0003417655677794665,
      "loss": 2.1851,
      "step": 104958
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.9340661764144897,
      "learning_rate": 0.00034176151704862,
      "loss": 2.1511,
      "step": 104959
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.8688004016876221,
      "learning_rate": 0.00034175746631000937,
      "loss": 2.2731,
      "step": 104960
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.8439331650733948,
      "learning_rate": 0.00034175341556363524,
      "loss": 1.9193,
      "step": 104961
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.9205878973007202,
      "learning_rate": 0.0003417493648094982,
      "loss": 2.0004,
      "step": 104962
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.906906247138977,
      "learning_rate": 0.0003417453140475993,
      "loss": 2.1241,
      "step": 104963
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.8588403463363647,
      "learning_rate": 0.0003417412632779392,
      "loss": 2.0952,
      "step": 104964
    },
    {
      "epoch": 1.37,
      "grad_norm": 1.004797339439392,
      "learning_rate": 0.00034173721250051854,
      "loss": 2.0175,
      "step": 104965
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.8782194256782532,
      "learning_rate": 0.0003417331617153381,
      "loss": 2.207,
      "step": 104966
    },
    {
      "epoch": 1.37,
      "grad_norm": 1.0015971660614014,
      "learning_rate": 0.00034172911092239874,
      "loss": 2.3956,
      "step": 104967
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.884097158908844,
      "learning_rate": 0.00034172506012170115,
      "loss": 2.3441,
      "step": 104968
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.9571478366851807,
      "learning_rate": 0.0003417210093132461,
      "loss": 2.194,
      "step": 104969
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.8293672204017639,
      "learning_rate": 0.0003417169584970343,
      "loss": 2.1307,
      "step": 104970
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.9333224892616272,
      "learning_rate": 0.0003417129076730666,
      "loss": 2.2776,
      "step": 104971
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.8552352786064148,
      "learning_rate": 0.0003417088568413435,
      "loss": 2.2253,
      "step": 104972
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.8584941029548645,
      "learning_rate": 0.0003417048060018661,
      "loss": 2.3523,
      "step": 104973
    },
    {
      "epoch": 1.37,
      "grad_norm": 1.15530264377594,
      "learning_rate": 0.0003417007551546349,
      "loss": 2.0575,
      "step": 104974
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.9282251596450806,
      "learning_rate": 0.00034169670429965083,
      "loss": 1.977,
      "step": 104975
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.868893027305603,
      "learning_rate": 0.0003416926534369145,
      "loss": 1.9512,
      "step": 104976
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.991214394569397,
      "learning_rate": 0.0003416886025664267,
      "loss": 2.0641,
      "step": 104977
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.9667812585830688,
      "learning_rate": 0.0003416845516881882,
      "loss": 2.1993,
      "step": 104978
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.9029436111450195,
      "learning_rate": 0.00034168050080219986,
      "loss": 2.2614,
      "step": 104979
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.9522824287414551,
      "learning_rate": 0.0003416764499084622,
      "loss": 2.0851,
      "step": 104980
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.8485557436943054,
      "learning_rate": 0.0003416723990069761,
      "loss": 2.0667,
      "step": 104981
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.8869506120681763,
      "learning_rate": 0.00034166834809774247,
      "loss": 2.2038,
      "step": 104982
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.797527551651001,
      "learning_rate": 0.0003416642971807618,
      "loss": 2.0337,
      "step": 104983
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.9471709132194519,
      "learning_rate": 0.00034166024625603496,
      "loss": 2.3079,
      "step": 104984
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.935076892375946,
      "learning_rate": 0.0003416561953235627,
      "loss": 2.2416,
      "step": 104985
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.8836243152618408,
      "learning_rate": 0.00034165214438334573,
      "loss": 2.2694,
      "step": 104986
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.9163283705711365,
      "learning_rate": 0.00034164809343538487,
      "loss": 2.1601,
      "step": 104987
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.9982303977012634,
      "learning_rate": 0.00034164404247968085,
      "loss": 2.39,
      "step": 104988
    },
    {
      "epoch": 1.37,
      "grad_norm": 1.0076556205749512,
      "learning_rate": 0.0003416399915162344,
      "loss": 2.3098,
      "step": 104989
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.872424304485321,
      "learning_rate": 0.00034163594054504627,
      "loss": 2.2283,
      "step": 104990
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.8555969595909119,
      "learning_rate": 0.0003416318895661173,
      "loss": 2.3009,
      "step": 104991
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.9096537232398987,
      "learning_rate": 0.0003416278385794481,
      "loss": 2.0939,
      "step": 104992
    },
    {
      "epoch": 1.37,
      "grad_norm": 1.0837229490280151,
      "learning_rate": 0.0003416237875850395,
      "loss": 2.2607,
      "step": 104993
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.9051842093467712,
      "learning_rate": 0.00034161973658289227,
      "loss": 2.3807,
      "step": 104994
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.8035933375358582,
      "learning_rate": 0.0003416156855730072,
      "loss": 2.1204,
      "step": 104995
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.9610502123832703,
      "learning_rate": 0.000341611634555385,
      "loss": 2.1572,
      "step": 104996
    },
    {
      "epoch": 1.37,
      "grad_norm": 1.083457112312317,
      "learning_rate": 0.0003416075835300263,
      "loss": 2.2456,
      "step": 104997
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.8315836191177368,
      "learning_rate": 0.000341603532496932,
      "loss": 2.0181,
      "step": 104998
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.9620196223258972,
      "learning_rate": 0.0003415994814561029,
      "loss": 2.2148,
      "step": 104999
    },
    {
      "epoch": 1.37,
      "grad_norm": 1.0167922973632812,
      "learning_rate": 0.00034159543040753956,
      "loss": 2.0612,
      "step": 105000
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.876279890537262,
      "learning_rate": 0.00034159137935124294,
      "loss": 2.2826,
      "step": 105001
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.8784785270690918,
      "learning_rate": 0.0003415873282872136,
      "loss": 2.2559,
      "step": 105002
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.8858507871627808,
      "learning_rate": 0.0003415832772154525,
      "loss": 2.2609,
      "step": 105003
    },
    {
      "epoch": 1.37,
      "grad_norm": 1.0337738990783691,
      "learning_rate": 0.0003415792261359603,
      "loss": 2.1133,
      "step": 105004
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.8544345498085022,
      "learning_rate": 0.0003415751750487376,
      "loss": 2.308,
      "step": 105005
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.9037474989891052,
      "learning_rate": 0.0003415711239537853,
      "loss": 2.0828,
      "step": 105006
    },
    {
      "epoch": 1.37,
      "grad_norm": 1.1012372970581055,
      "learning_rate": 0.00034156707285110425,
      "loss": 1.9751,
      "step": 105007
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.9489445686340332,
      "learning_rate": 0.00034156302174069504,
      "loss": 2.2325,
      "step": 105008
    },
    {
      "epoch": 1.37,
      "grad_norm": 1.0098111629486084,
      "learning_rate": 0.00034155897062255845,
      "loss": 1.967,
      "step": 105009
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.9372521638870239,
      "learning_rate": 0.00034155491949669525,
      "loss": 2.085,
      "step": 105010
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.8897272348403931,
      "learning_rate": 0.00034155086836310634,
      "loss": 2.2601,
      "step": 105011
    },
    {
      "epoch": 1.37,
      "grad_norm": 1.0703192949295044,
      "learning_rate": 0.0003415468172217922,
      "loss": 2.0558,
      "step": 105012
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.9886776208877563,
      "learning_rate": 0.00034154276607275375,
      "loss": 2.1668,
      "step": 105013
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.8969171643257141,
      "learning_rate": 0.00034153871491599187,
      "loss": 2.1122,
      "step": 105014
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.9773629903793335,
      "learning_rate": 0.000341534663751507,
      "loss": 2.3102,
      "step": 105015
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.9353089332580566,
      "learning_rate": 0.0003415306125793,
      "loss": 2.1693,
      "step": 105016
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.8790772557258606,
      "learning_rate": 0.00034152656139937174,
      "loss": 2.2272,
      "step": 105017
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.8753125667572021,
      "learning_rate": 0.000341522510211723,
      "loss": 2.1435,
      "step": 105018
    },
    {
      "epoch": 1.37,
      "grad_norm": 1.0094163417816162,
      "learning_rate": 0.00034151845901635435,
      "loss": 2.2133,
      "step": 105019
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.9560821652412415,
      "learning_rate": 0.0003415144078132666,
      "loss": 2.3161,
      "step": 105020
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.8428516983985901,
      "learning_rate": 0.00034151035660246067,
      "loss": 2.0155,
      "step": 105021
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.830119788646698,
      "learning_rate": 0.0003415063053839371,
      "loss": 2.1736,
      "step": 105022
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.8428546190261841,
      "learning_rate": 0.0003415022541576967,
      "loss": 2.1116,
      "step": 105023
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.8350826501846313,
      "learning_rate": 0.0003414982029237403,
      "loss": 2.3477,
      "step": 105024
    },
    {
      "epoch": 1.37,
      "grad_norm": 1.1603162288665771,
      "learning_rate": 0.0003414941516820685,
      "loss": 2.2341,
      "step": 105025
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.93830806016922,
      "learning_rate": 0.0003414901004326823,
      "loss": 2.1225,
      "step": 105026
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.879701554775238,
      "learning_rate": 0.0003414860491755822,
      "loss": 2.053,
      "step": 105027
    },
    {
      "epoch": 1.37,
      "grad_norm": 1.041271686553955,
      "learning_rate": 0.00034148199791076913,
      "loss": 2.1402,
      "step": 105028
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.9397279024124146,
      "learning_rate": 0.00034147794663824373,
      "loss": 2.0587,
      "step": 105029
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.9186453223228455,
      "learning_rate": 0.00034147389535800677,
      "loss": 2.2127,
      "step": 105030
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.9237343072891235,
      "learning_rate": 0.0003414698440700591,
      "loss": 1.9355,
      "step": 105031
    },
    {
      "epoch": 1.37,
      "grad_norm": 1.2684521675109863,
      "learning_rate": 0.00034146579277440135,
      "loss": 2.2105,
      "step": 105032
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.838160514831543,
      "learning_rate": 0.00034146174147103435,
      "loss": 1.8243,
      "step": 105033
    },
    {
      "epoch": 1.37,
      "grad_norm": 1.4341778755187988,
      "learning_rate": 0.00034145769015995884,
      "loss": 2.2812,
      "step": 105034
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.9599182605743408,
      "learning_rate": 0.00034145363884117557,
      "loss": 2.1659,
      "step": 105035
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.8767738342285156,
      "learning_rate": 0.00034144958751468525,
      "loss": 2.221,
      "step": 105036
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.9375341534614563,
      "learning_rate": 0.0003414455361804887,
      "loss": 2.2721,
      "step": 105037
    },
    {
      "epoch": 1.37,
      "grad_norm": 1.084006428718567,
      "learning_rate": 0.0003414414848385866,
      "loss": 2.1675,
      "step": 105038
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.9432727694511414,
      "learning_rate": 0.0003414374334889798,
      "loss": 2.0718,
      "step": 105039
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.8063501119613647,
      "learning_rate": 0.000341433382131669,
      "loss": 2.2639,
      "step": 105040
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.8768215179443359,
      "learning_rate": 0.00034142933076665487,
      "loss": 2.055,
      "step": 105041
    },
    {
      "epoch": 1.37,
      "grad_norm": 1.0028870105743408,
      "learning_rate": 0.0003414252793939383,
      "loss": 2.2976,
      "step": 105042
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.8368369936943054,
      "learning_rate": 0.00034142122801352,
      "loss": 2.0339,
      "step": 105043
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.7779968976974487,
      "learning_rate": 0.00034141717662540067,
      "loss": 2.1677,
      "step": 105044
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.8888458609580994,
      "learning_rate": 0.00034141312522958115,
      "loss": 2.3926,
      "step": 105045
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.8750908374786377,
      "learning_rate": 0.0003414090738260622,
      "loss": 2.3062,
      "step": 105046
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.8256924748420715,
      "learning_rate": 0.0003414050224148444,
      "loss": 2.1223,
      "step": 105047
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.815314531326294,
      "learning_rate": 0.0003414009709959287,
      "loss": 1.7966,
      "step": 105048
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.8117974996566772,
      "learning_rate": 0.00034139691956931576,
      "loss": 2.3019,
      "step": 105049
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.8251373171806335,
      "learning_rate": 0.0003413928681350063,
      "loss": 2.0324,
      "step": 105050
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.8699812889099121,
      "learning_rate": 0.00034138881669300113,
      "loss": 2.0932,
      "step": 105051
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.9715054631233215,
      "learning_rate": 0.00034138476524330115,
      "loss": 2.0011,
      "step": 105052
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.883371889591217,
      "learning_rate": 0.0003413807137859068,
      "loss": 2.1844,
      "step": 105053
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.9577105641365051,
      "learning_rate": 0.000341376662320819,
      "loss": 2.2548,
      "step": 105054
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.869477391242981,
      "learning_rate": 0.00034137261084803865,
      "loss": 1.9866,
      "step": 105055
    },
    {
      "epoch": 1.37,
      "grad_norm": 1.0059603452682495,
      "learning_rate": 0.0003413685593675662,
      "loss": 2.1287,
      "step": 105056
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.851690411567688,
      "learning_rate": 0.0003413645078794026,
      "loss": 1.9518,
      "step": 105057
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.8436154723167419,
      "learning_rate": 0.00034136045638354854,
      "loss": 1.9071,
      "step": 105058
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.9231732487678528,
      "learning_rate": 0.00034135640488000486,
      "loss": 2.0533,
      "step": 105059
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.9534745216369629,
      "learning_rate": 0.00034135235336877213,
      "loss": 2.1846,
      "step": 105060
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.9741222858428955,
      "learning_rate": 0.00034134830184985127,
      "loss": 2.1523,
      "step": 105061
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.8530286550521851,
      "learning_rate": 0.000341344250323243,
      "loss": 2.0199,
      "step": 105062
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.9239881038665771,
      "learning_rate": 0.00034134019878894807,
      "loss": 1.9278,
      "step": 105063
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.9118849635124207,
      "learning_rate": 0.00034133614724696714,
      "loss": 2.1502,
      "step": 105064
    },
    {
      "epoch": 1.37,
      "grad_norm": 1.0167726278305054,
      "learning_rate": 0.0003413320956973011,
      "loss": 2.3858,
      "step": 105065
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.8522677421569824,
      "learning_rate": 0.00034132804413995066,
      "loss": 2.1422,
      "step": 105066
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.9350742101669312,
      "learning_rate": 0.0003413239925749165,
      "loss": 2.2216,
      "step": 105067
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.9880009293556213,
      "learning_rate": 0.0003413199410021994,
      "loss": 2.0498,
      "step": 105068
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.855961799621582,
      "learning_rate": 0.0003413158894218003,
      "loss": 2.1064,
      "step": 105069
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.928112268447876,
      "learning_rate": 0.0003413118378337196,
      "loss": 2.1165,
      "step": 105070
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.9461079835891724,
      "learning_rate": 0.0003413077862379584,
      "loss": 2.1937,
      "step": 105071
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.981344997882843,
      "learning_rate": 0.0003413037346345173,
      "loss": 2.0548,
      "step": 105072
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.9333033561706543,
      "learning_rate": 0.00034129968302339694,
      "loss": 1.9998,
      "step": 105073
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.9734131693840027,
      "learning_rate": 0.00034129563140459823,
      "loss": 2.0226,
      "step": 105074
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.8894299864768982,
      "learning_rate": 0.0003412915797781219,
      "loss": 2.3499,
      "step": 105075
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.8556932806968689,
      "learning_rate": 0.0003412875281439687,
      "loss": 2.1941,
      "step": 105076
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.8463736176490784,
      "learning_rate": 0.0003412834765021394,
      "loss": 2.325,
      "step": 105077
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.8979331254959106,
      "learning_rate": 0.0003412794248526347,
      "loss": 2.1924,
      "step": 105078
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.954295814037323,
      "learning_rate": 0.0003412753731954553,
      "loss": 2.3566,
      "step": 105079
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.8712135553359985,
      "learning_rate": 0.0003412713215306022,
      "loss": 2.045,
      "step": 105080
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.9251344799995422,
      "learning_rate": 0.00034126726985807574,
      "loss": 2.2967,
      "step": 105081
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.9627904295921326,
      "learning_rate": 0.0003412632181778771,
      "loss": 2.059,
      "step": 105082
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.903285562992096,
      "learning_rate": 0.0003412591664900068,
      "loss": 2.1825,
      "step": 105083
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.9324002265930176,
      "learning_rate": 0.0003412551147944656,
      "loss": 2.0425,
      "step": 105084
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.8928875923156738,
      "learning_rate": 0.0003412510630912543,
      "loss": 2.1763,
      "step": 105085
    },
    {
      "epoch": 1.37,
      "grad_norm": 1.0496495962142944,
      "learning_rate": 0.0003412470113803738,
      "loss": 1.927,
      "step": 105086
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.8800150752067566,
      "learning_rate": 0.0003412429596618245,
      "loss": 2.1209,
      "step": 105087
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.8157214522361755,
      "learning_rate": 0.00034123890793560745,
      "loss": 2.134,
      "step": 105088
    },
    {
      "epoch": 1.37,
      "grad_norm": 1.0173780918121338,
      "learning_rate": 0.0003412348562017233,
      "loss": 2.058,
      "step": 105089
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.8853080868721008,
      "learning_rate": 0.00034123080446017283,
      "loss": 2.2201,
      "step": 105090
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.9941275715827942,
      "learning_rate": 0.0003412267527109567,
      "loss": 2.1676,
      "step": 105091
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.971608579158783,
      "learning_rate": 0.00034122270095407584,
      "loss": 2.0294,
      "step": 105092
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.9167671799659729,
      "learning_rate": 0.00034121864918953087,
      "loss": 2.0557,
      "step": 105093
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.8166186809539795,
      "learning_rate": 0.00034121459741732254,
      "loss": 2.119,
      "step": 105094
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.895902156829834,
      "learning_rate": 0.00034121054563745166,
      "loss": 2.0131,
      "step": 105095
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.9074946045875549,
      "learning_rate": 0.00034120649384991895,
      "loss": 2.4041,
      "step": 105096
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.8573988080024719,
      "learning_rate": 0.00034120244205472516,
      "loss": 1.925,
      "step": 105097
    },
    {
      "epoch": 1.37,
      "grad_norm": 1.0968433618545532,
      "learning_rate": 0.00034119839025187104,
      "loss": 2.3461,
      "step": 105098
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.8288449048995972,
      "learning_rate": 0.00034119433844135747,
      "loss": 2.2686,
      "step": 105099
    },
    {
      "epoch": 1.37,
      "grad_norm": 1.0826364755630493,
      "learning_rate": 0.00034119028662318504,
      "loss": 2.1499,
      "step": 105100
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.9673289060592651,
      "learning_rate": 0.0003411862347973545,
      "loss": 2.3886,
      "step": 105101
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.8889797329902649,
      "learning_rate": 0.00034118218296386675,
      "loss": 2.3612,
      "step": 105102
    },
    {
      "epoch": 1.37,
      "grad_norm": 1.0510271787643433,
      "learning_rate": 0.00034117813112272236,
      "loss": 2.2889,
      "step": 105103
    },
    {
      "epoch": 1.37,
      "grad_norm": 1.174186110496521,
      "learning_rate": 0.0003411740792739222,
      "loss": 2.208,
      "step": 105104
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.889590859413147,
      "learning_rate": 0.0003411700274174671,
      "loss": 2.2249,
      "step": 105105
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.8921202421188354,
      "learning_rate": 0.00034116597555335765,
      "loss": 2.0286,
      "step": 105106
    },
    {
      "epoch": 1.37,
      "grad_norm": 1.230317234992981,
      "learning_rate": 0.0003411619236815946,
      "loss": 2.3155,
      "step": 105107
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.949932873249054,
      "learning_rate": 0.00034115787180217887,
      "loss": 2.1843,
      "step": 105108
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.851661205291748,
      "learning_rate": 0.0003411538199151111,
      "loss": 1.9562,
      "step": 105109
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.9324007630348206,
      "learning_rate": 0.00034114976802039203,
      "loss": 2.2571,
      "step": 105110
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.8260027170181274,
      "learning_rate": 0.0003411457161180224,
      "loss": 2.1397,
      "step": 105111
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.8638578057289124,
      "learning_rate": 0.00034114166420800304,
      "loss": 2.5012,
      "step": 105112
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.9007530212402344,
      "learning_rate": 0.00034113761229033474,
      "loss": 2.0942,
      "step": 105113
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.9663012623786926,
      "learning_rate": 0.0003411335603650181,
      "loss": 2.202,
      "step": 105114
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.9413843154907227,
      "learning_rate": 0.00034112950843205397,
      "loss": 2.2198,
      "step": 105115
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.9189561605453491,
      "learning_rate": 0.00034112545649144313,
      "loss": 2.1346,
      "step": 105116
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.9115141034126282,
      "learning_rate": 0.0003411214045431863,
      "loss": 2.3801,
      "step": 105117
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.8570591807365417,
      "learning_rate": 0.0003411173525872841,
      "loss": 2.1267,
      "step": 105118
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.8448038101196289,
      "learning_rate": 0.0003411133006237376,
      "loss": 2.1964,
      "step": 105119
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.970892071723938,
      "learning_rate": 0.0003411092486525472,
      "loss": 2.2444,
      "step": 105120
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.8735144734382629,
      "learning_rate": 0.0003411051966737138,
      "loss": 2.2871,
      "step": 105121
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.8311735391616821,
      "learning_rate": 0.0003411011446872383,
      "loss": 2.1559,
      "step": 105122
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.8452258706092834,
      "learning_rate": 0.00034109709269312125,
      "loss": 2.0374,
      "step": 105123
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.923462450504303,
      "learning_rate": 0.0003410930406913634,
      "loss": 2.179,
      "step": 105124
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.9288644790649414,
      "learning_rate": 0.00034108898868196575,
      "loss": 2.0886,
      "step": 105125
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.8551164269447327,
      "learning_rate": 0.00034108493666492874,
      "loss": 2.3235,
      "step": 105126
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.9192535877227783,
      "learning_rate": 0.0003410808846402533,
      "loss": 2.0665,
      "step": 105127
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.8982378244400024,
      "learning_rate": 0.00034107683260794017,
      "loss": 2.3216,
      "step": 105128
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.8400693535804749,
      "learning_rate": 0.0003410727805679901,
      "loss": 2.134,
      "step": 105129
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.831539511680603,
      "learning_rate": 0.00034106872852040376,
      "loss": 2.0841,
      "step": 105130
    },
    {
      "epoch": 1.37,
      "grad_norm": 1.0137077569961548,
      "learning_rate": 0.000341064676465182,
      "loss": 2.1724,
      "step": 105131
    },
    {
      "epoch": 1.37,
      "grad_norm": 1.084706425666809,
      "learning_rate": 0.0003410606244023256,
      "loss": 1.9823,
      "step": 105132
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.8384203910827637,
      "learning_rate": 0.0003410565723318352,
      "loss": 1.9337,
      "step": 105133
    },
    {
      "epoch": 1.37,
      "grad_norm": 1.3072679042816162,
      "learning_rate": 0.0003410525202537116,
      "loss": 1.7792,
      "step": 105134
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.869881272315979,
      "learning_rate": 0.0003410484681679556,
      "loss": 1.84,
      "step": 105135
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.9704849720001221,
      "learning_rate": 0.00034104441607456785,
      "loss": 2.1266,
      "step": 105136
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.8998124003410339,
      "learning_rate": 0.0003410403639735492,
      "loss": 2.0727,
      "step": 105137
    },
    {
      "epoch": 1.37,
      "grad_norm": 1.027206301689148,
      "learning_rate": 0.00034103631186490046,
      "loss": 2.1343,
      "step": 105138
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.8314452171325684,
      "learning_rate": 0.00034103225974862214,
      "loss": 2.1061,
      "step": 105139
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.9253509044647217,
      "learning_rate": 0.00034102820762471517,
      "loss": 2.0544,
      "step": 105140
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.873318612575531,
      "learning_rate": 0.0003410241554931804,
      "loss": 2.2069,
      "step": 105141
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.9975460767745972,
      "learning_rate": 0.0003410201033540184,
      "loss": 2.1999,
      "step": 105142
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.8637335300445557,
      "learning_rate": 0.00034101605120723,
      "loss": 2.1829,
      "step": 105143
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.874046802520752,
      "learning_rate": 0.00034101199905281596,
      "loss": 1.8922,
      "step": 105144
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.8725420236587524,
      "learning_rate": 0.00034100794689077694,
      "loss": 2.1312,
      "step": 105145
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.8399572372436523,
      "learning_rate": 0.0003410038947211138,
      "loss": 2.1336,
      "step": 105146
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.9372943043708801,
      "learning_rate": 0.00034099984254382733,
      "loss": 1.9803,
      "step": 105147
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.93853360414505,
      "learning_rate": 0.00034099579035891815,
      "loss": 2.474,
      "step": 105148
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.9336866140365601,
      "learning_rate": 0.0003409917381663871,
      "loss": 2.2447,
      "step": 105149
    },
    {
      "epoch": 1.37,
      "grad_norm": 1.026740312576294,
      "learning_rate": 0.00034098768596623494,
      "loss": 2.1544,
      "step": 105150
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.8974576592445374,
      "learning_rate": 0.0003409836337584623,
      "loss": 2.2059,
      "step": 105151
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.8469417095184326,
      "learning_rate": 0.0003409795815430701,
      "loss": 2.023,
      "step": 105152
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.979687511920929,
      "learning_rate": 0.000340975529320059,
      "loss": 2.3569,
      "step": 105153
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.9914140701293945,
      "learning_rate": 0.0003409714770894298,
      "loss": 2.1841,
      "step": 105154
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.8733741044998169,
      "learning_rate": 0.0003409674248511832,
      "loss": 2.0992,
      "step": 105155
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.9254757165908813,
      "learning_rate": 0.00034096337260532003,
      "loss": 2.4551,
      "step": 105156
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.9901871085166931,
      "learning_rate": 0.0003409593203518409,
      "loss": 2.1488,
      "step": 105157
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.8875457644462585,
      "learning_rate": 0.0003409552680907467,
      "loss": 2.0851,
      "step": 105158
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.9846431612968445,
      "learning_rate": 0.00034095121582203823,
      "loss": 2.2841,
      "step": 105159
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.9390484690666199,
      "learning_rate": 0.00034094716354571606,
      "loss": 1.9626,
      "step": 105160
    },
    {
      "epoch": 1.37,
      "grad_norm": 1.0861517190933228,
      "learning_rate": 0.00034094311126178106,
      "loss": 2.3435,
      "step": 105161
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.8328655362129211,
      "learning_rate": 0.00034093905897023403,
      "loss": 2.2511,
      "step": 105162
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.8655686378479004,
      "learning_rate": 0.00034093500667107553,
      "loss": 1.9228,
      "step": 105163
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.9040662050247192,
      "learning_rate": 0.0003409309543643065,
      "loss": 2.0245,
      "step": 105164
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.9172111749649048,
      "learning_rate": 0.0003409269020499277,
      "loss": 2.3056,
      "step": 105165
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.8709011673927307,
      "learning_rate": 0.00034092284972793977,
      "loss": 2.1193,
      "step": 105166
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.919175922870636,
      "learning_rate": 0.0003409187973983434,
      "loss": 2.0309,
      "step": 105167
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.8371763229370117,
      "learning_rate": 0.0003409147450611396,
      "loss": 1.9695,
      "step": 105168
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.8870915174484253,
      "learning_rate": 0.0003409106927163289,
      "loss": 2.2262,
      "step": 105169
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.8995185494422913,
      "learning_rate": 0.00034090664036391216,
      "loss": 2.0445,
      "step": 105170
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.9784365892410278,
      "learning_rate": 0.0003409025880038901,
      "loss": 2.0216,
      "step": 105171
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.8959603905677795,
      "learning_rate": 0.00034089853563626344,
      "loss": 2.158,
      "step": 105172
    },
    {
      "epoch": 1.37,
      "grad_norm": 1.0282256603240967,
      "learning_rate": 0.00034089448326103303,
      "loss": 2.2901,
      "step": 105173
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.8929814696311951,
      "learning_rate": 0.0003408904308781996,
      "loss": 2.2168,
      "step": 105174
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.9009043574333191,
      "learning_rate": 0.00034088637848776374,
      "loss": 2.3077,
      "step": 105175
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.9911966919898987,
      "learning_rate": 0.00034088232608972644,
      "loss": 2.1714,
      "step": 105176
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.8720995783805847,
      "learning_rate": 0.00034087827368408833,
      "loss": 2.2061,
      "step": 105177
    },
    {
      "epoch": 1.37,
      "grad_norm": 1.0720667839050293,
      "learning_rate": 0.00034087422127085013,
      "loss": 2.2278,
      "step": 105178
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.9315302968025208,
      "learning_rate": 0.00034087016885001265,
      "loss": 2.256,
      "step": 105179
    },
    {
      "epoch": 1.37,
      "grad_norm": 1.0432525873184204,
      "learning_rate": 0.0003408661164215767,
      "loss": 2.1733,
      "step": 105180
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.8483054637908936,
      "learning_rate": 0.0003408620639855429,
      "loss": 2.1498,
      "step": 105181
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.8831222653388977,
      "learning_rate": 0.000340858011541912,
      "loss": 1.9285,
      "step": 105182
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.9373754858970642,
      "learning_rate": 0.000340853959090685,
      "loss": 2.2524,
      "step": 105183
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.9081518054008484,
      "learning_rate": 0.0003408499066318623,
      "loss": 2.1483,
      "step": 105184
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.9657779932022095,
      "learning_rate": 0.00034084585416544497,
      "loss": 2.2352,
      "step": 105185
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.9379510283470154,
      "learning_rate": 0.0003408418016914335,
      "loss": 2.168,
      "step": 105186
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.9175072312355042,
      "learning_rate": 0.0003408377492098289,
      "loss": 2.0648,
      "step": 105187
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.8831772208213806,
      "learning_rate": 0.0003408336967206317,
      "loss": 2.3074,
      "step": 105188
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.9711019396781921,
      "learning_rate": 0.00034082964422384274,
      "loss": 2.2089,
      "step": 105189
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.9966387152671814,
      "learning_rate": 0.0003408255917194628,
      "loss": 2.2173,
      "step": 105190
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.8091490864753723,
      "learning_rate": 0.00034082153920749267,
      "loss": 2.2021,
      "step": 105191
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.873104989528656,
      "learning_rate": 0.00034081748668793294,
      "loss": 2.2132,
      "step": 105192
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.8827027678489685,
      "learning_rate": 0.00034081343416078456,
      "loss": 2.4574,
      "step": 105193
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.985934317111969,
      "learning_rate": 0.00034080938162604817,
      "loss": 1.8424,
      "step": 105194
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.8375295996665955,
      "learning_rate": 0.00034080532908372453,
      "loss": 2.2142,
      "step": 105195
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.8780571818351746,
      "learning_rate": 0.0003408012765338144,
      "loss": 2.24,
      "step": 105196
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.881030261516571,
      "learning_rate": 0.00034079722397631855,
      "loss": 2.328,
      "step": 105197
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.8269165754318237,
      "learning_rate": 0.0003407931714112378,
      "loss": 2.1455,
      "step": 105198
    },
    {
      "epoch": 1.37,
      "grad_norm": 1.1743426322937012,
      "learning_rate": 0.00034078911883857264,
      "loss": 2.1978,
      "step": 105199
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.9437445998191833,
      "learning_rate": 0.00034078506625832416,
      "loss": 2.023,
      "step": 105200
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.941897988319397,
      "learning_rate": 0.00034078101367049287,
      "loss": 2.0467,
      "step": 105201
    },
    {
      "epoch": 1.37,
      "grad_norm": 1.07757568359375,
      "learning_rate": 0.00034077696107507975,
      "loss": 2.2272,
      "step": 105202
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.8800976872444153,
      "learning_rate": 0.00034077290847208536,
      "loss": 2.3002,
      "step": 105203
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.8156529068946838,
      "learning_rate": 0.00034076885586151054,
      "loss": 2.2394,
      "step": 105204
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.8563256859779358,
      "learning_rate": 0.000340764803243356,
      "loss": 2.0655,
      "step": 105205
    },
    {
      "epoch": 1.37,
      "grad_norm": 1.0079374313354492,
      "learning_rate": 0.00034076075061762254,
      "loss": 2.3136,
      "step": 105206
    },
    {
      "epoch": 1.37,
      "grad_norm": 1.1268264055252075,
      "learning_rate": 0.0003407566979843108,
      "loss": 2.1043,
      "step": 105207
    },
    {
      "epoch": 1.37,
      "grad_norm": 1.2150624990463257,
      "learning_rate": 0.0003407526453434217,
      "loss": 2.1805,
      "step": 105208
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.8923569917678833,
      "learning_rate": 0.0003407485926949559,
      "loss": 2.0256,
      "step": 105209
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.8862576484680176,
      "learning_rate": 0.00034074454003891417,
      "loss": 2.2005,
      "step": 105210
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.8910805583000183,
      "learning_rate": 0.00034074048737529726,
      "loss": 2.054,
      "step": 105211
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.7958377599716187,
      "learning_rate": 0.0003407364347041059,
      "loss": 2.3247,
      "step": 105212
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.850502073764801,
      "learning_rate": 0.00034073238202534094,
      "loss": 2.3099,
      "step": 105213
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.9554179906845093,
      "learning_rate": 0.000340728329339003,
      "loss": 2.1515,
      "step": 105214
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.9347426295280457,
      "learning_rate": 0.0003407242766450929,
      "loss": 2.2086,
      "step": 105215
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.8070256114006042,
      "learning_rate": 0.00034072022394361135,
      "loss": 2.0688,
      "step": 105216
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.8487520813941956,
      "learning_rate": 0.00034071617123455927,
      "loss": 2.2288,
      "step": 105217
    },
    {
      "epoch": 1.37,
      "grad_norm": 1.0100576877593994,
      "learning_rate": 0.0003407121185179372,
      "loss": 2.0483,
      "step": 105218
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.9128474593162537,
      "learning_rate": 0.00034070806579374594,
      "loss": 2.025,
      "step": 105219
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.858916163444519,
      "learning_rate": 0.00034070401306198633,
      "loss": 2.1881,
      "step": 105220
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.9088643193244934,
      "learning_rate": 0.0003406999603226591,
      "loss": 2.249,
      "step": 105221
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.9681529402732849,
      "learning_rate": 0.0003406959075757649,
      "loss": 2.3107,
      "step": 105222
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.8070769309997559,
      "learning_rate": 0.0003406918548213046,
      "loss": 2.2401,
      "step": 105223
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.8545351624488831,
      "learning_rate": 0.0003406878020592789,
      "loss": 2.1748,
      "step": 105224
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.8224226236343384,
      "learning_rate": 0.0003406837492896886,
      "loss": 2.2869,
      "step": 105225
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.9461170434951782,
      "learning_rate": 0.0003406796965125345,
      "loss": 2.3511,
      "step": 105226
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.8181290626525879,
      "learning_rate": 0.0003406756437278171,
      "loss": 2.3458,
      "step": 105227
    },
    {
      "epoch": 1.37,
      "grad_norm": 1.1829169988632202,
      "learning_rate": 0.0003406715909355374,
      "loss": 2.2442,
      "step": 105228
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.9522585272789001,
      "learning_rate": 0.0003406675381356962,
      "loss": 2.131,
      "step": 105229
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.8715254068374634,
      "learning_rate": 0.000340663485328294,
      "loss": 2.0963,
      "step": 105230
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.9843224287033081,
      "learning_rate": 0.0003406594325133317,
      "loss": 2.2529,
      "step": 105231
    },
    {
      "epoch": 1.37,
      "grad_norm": 1.029144048690796,
      "learning_rate": 0.00034065537969081013,
      "loss": 2.1976,
      "step": 105232
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.8170986175537109,
      "learning_rate": 0.0003406513268607299,
      "loss": 1.9214,
      "step": 105233
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.8746001720428467,
      "learning_rate": 0.0003406472740230918,
      "loss": 2.0594,
      "step": 105234
    },
    {
      "epoch": 1.37,
      "grad_norm": 1.0210092067718506,
      "learning_rate": 0.00034064322117789665,
      "loss": 2.1211,
      "step": 105235
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.9196314811706543,
      "learning_rate": 0.0003406391683251451,
      "loss": 2.1437,
      "step": 105236
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.8581269383430481,
      "learning_rate": 0.000340635115464838,
      "loss": 2.1553,
      "step": 105237
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.9916620850563049,
      "learning_rate": 0.0003406310625969761,
      "loss": 2.3342,
      "step": 105238
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.9994103312492371,
      "learning_rate": 0.0003406270097215601,
      "loss": 2.2298,
      "step": 105239
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.8592016696929932,
      "learning_rate": 0.00034062295683859073,
      "loss": 2.0629,
      "step": 105240
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.8801556825637817,
      "learning_rate": 0.0003406189039480688,
      "loss": 2.1297,
      "step": 105241
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.9087192416191101,
      "learning_rate": 0.00034061485104999506,
      "loss": 2.1245,
      "step": 105242
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.9046370983123779,
      "learning_rate": 0.00034061079814437024,
      "loss": 2.2638,
      "step": 105243
    },
    {
      "epoch": 1.37,
      "grad_norm": 1.016933560371399,
      "learning_rate": 0.00034060674523119514,
      "loss": 2.2765,
      "step": 105244
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.7951169013977051,
      "learning_rate": 0.00034060269231047047,
      "loss": 2.147,
      "step": 105245
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.9186074733734131,
      "learning_rate": 0.00034059863938219693,
      "loss": 1.9809,
      "step": 105246
    },
    {
      "epoch": 1.37,
      "grad_norm": 1.0540916919708252,
      "learning_rate": 0.00034059458644637546,
      "loss": 1.9699,
      "step": 105247
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.9269645810127258,
      "learning_rate": 0.00034059053350300663,
      "loss": 2.1333,
      "step": 105248
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.956208348274231,
      "learning_rate": 0.0003405864805520912,
      "loss": 2.3968,
      "step": 105249
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.9000177383422852,
      "learning_rate": 0.00034058242759363,
      "loss": 2.0593,
      "step": 105250
    },
    {
      "epoch": 1.37,
      "grad_norm": 1.0234516859054565,
      "learning_rate": 0.0003405783746276238,
      "loss": 2.01,
      "step": 105251
    },
    {
      "epoch": 1.37,
      "grad_norm": 1.0864681005477905,
      "learning_rate": 0.00034057432165407323,
      "loss": 2.3566,
      "step": 105252
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.8696235418319702,
      "learning_rate": 0.0003405702686729793,
      "loss": 2.0265,
      "step": 105253
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.8628786206245422,
      "learning_rate": 0.00034056621568434245,
      "loss": 2.1287,
      "step": 105254
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.9363346695899963,
      "learning_rate": 0.0003405621626881636,
      "loss": 2.1972,
      "step": 105255
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.8773364424705505,
      "learning_rate": 0.00034055810968444347,
      "loss": 1.9906,
      "step": 105256
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.8846352696418762,
      "learning_rate": 0.0003405540566731828,
      "loss": 2.0964,
      "step": 105257
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.7976359128952026,
      "learning_rate": 0.0003405500036543824,
      "loss": 1.9864,
      "step": 105258
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.8918688893318176,
      "learning_rate": 0.00034054595062804307,
      "loss": 2.1057,
      "step": 105259
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.8280799984931946,
      "learning_rate": 0.00034054189759416537,
      "loss": 1.8948,
      "step": 105260
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.7741450667381287,
      "learning_rate": 0.00034053784455275013,
      "loss": 2.0741,
      "step": 105261
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.905401885509491,
      "learning_rate": 0.0003405337915037982,
      "loss": 2.4928,
      "step": 105262
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.9184345006942749,
      "learning_rate": 0.0003405297384473103,
      "loss": 2.2311,
      "step": 105263
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.8644351363182068,
      "learning_rate": 0.0003405256853832872,
      "loss": 2.0236,
      "step": 105264
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.8189010620117188,
      "learning_rate": 0.0003405216323117295,
      "loss": 2.203,
      "step": 105265
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.9451979994773865,
      "learning_rate": 0.0003405175792326381,
      "loss": 2.1863,
      "step": 105266
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.809913158416748,
      "learning_rate": 0.00034051352614601377,
      "loss": 2.2484,
      "step": 105267
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.8265882730484009,
      "learning_rate": 0.00034050947305185714,
      "loss": 2.3168,
      "step": 105268
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.9365066885948181,
      "learning_rate": 0.0003405054199501691,
      "loss": 2.0895,
      "step": 105269
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.9263861179351807,
      "learning_rate": 0.00034050136684095036,
      "loss": 2.2782,
      "step": 105270
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.9679443836212158,
      "learning_rate": 0.0003404973137242015,
      "loss": 2.1988,
      "step": 105271
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.8835831880569458,
      "learning_rate": 0.00034049326059992357,
      "loss": 2.1557,
      "step": 105272
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.8704610466957092,
      "learning_rate": 0.0003404892074681171,
      "loss": 2.1462,
      "step": 105273
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.8588532209396362,
      "learning_rate": 0.00034048515432878295,
      "loss": 2.069,
      "step": 105274
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.9587857127189636,
      "learning_rate": 0.0003404811011819219,
      "loss": 2.2494,
      "step": 105275
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.8641815185546875,
      "learning_rate": 0.0003404770480275346,
      "loss": 2.0566,
      "step": 105276
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.8111510276794434,
      "learning_rate": 0.0003404729948656217,
      "loss": 2.1818,
      "step": 105277
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.9625665545463562,
      "learning_rate": 0.0003404689416961843,
      "loss": 2.1414,
      "step": 105278
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.8820651173591614,
      "learning_rate": 0.00034046488851922294,
      "loss": 2.166,
      "step": 105279
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.8457480669021606,
      "learning_rate": 0.00034046083533473834,
      "loss": 2.2441,
      "step": 105280
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.8608981370925903,
      "learning_rate": 0.0003404567821427313,
      "loss": 1.9823,
      "step": 105281
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.8727515935897827,
      "learning_rate": 0.00034045272894320266,
      "loss": 2.0436,
      "step": 105282
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.9695591926574707,
      "learning_rate": 0.00034044867573615304,
      "loss": 1.7713,
      "step": 105283
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.9445779919624329,
      "learning_rate": 0.0003404446225215833,
      "loss": 2.2709,
      "step": 105284
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.873356819152832,
      "learning_rate": 0.00034044056929949406,
      "loss": 2.0195,
      "step": 105285
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.9902983903884888,
      "learning_rate": 0.00034043651606988616,
      "loss": 1.8841,
      "step": 105286
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.9066458940505981,
      "learning_rate": 0.0003404324628327604,
      "loss": 2.0665,
      "step": 105287
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.857513427734375,
      "learning_rate": 0.0003404284095881175,
      "loss": 2.0955,
      "step": 105288
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.9518187642097473,
      "learning_rate": 0.00034042435633595806,
      "loss": 2.318,
      "step": 105289
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.9739640355110168,
      "learning_rate": 0.0003404203030762831,
      "loss": 2.2594,
      "step": 105290
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.9400728344917297,
      "learning_rate": 0.00034041624980909324,
      "loss": 2.2582,
      "step": 105291
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.8923599123954773,
      "learning_rate": 0.00034041219653438914,
      "loss": 2.2913,
      "step": 105292
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.9217231273651123,
      "learning_rate": 0.0003404081432521718,
      "loss": 2.0847,
      "step": 105293
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.9473361372947693,
      "learning_rate": 0.00034040408996244173,
      "loss": 2.1816,
      "step": 105294
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.9380565881729126,
      "learning_rate": 0.0003404000366651997,
      "loss": 2.2115,
      "step": 105295
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.97674161195755,
      "learning_rate": 0.0003403959833604467,
      "loss": 2.1615,
      "step": 105296
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.8581716418266296,
      "learning_rate": 0.00034039193004818327,
      "loss": 2.0983,
      "step": 105297
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.9951075315475464,
      "learning_rate": 0.0003403878767284101,
      "loss": 2.0339,
      "step": 105298
    },
    {
      "epoch": 1.37,
      "grad_norm": 1.0435768365859985,
      "learning_rate": 0.00034038382340112824,
      "loss": 2.0872,
      "step": 105299
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.9745673537254333,
      "learning_rate": 0.00034037977006633814,
      "loss": 2.0444,
      "step": 105300
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.9374616742134094,
      "learning_rate": 0.00034037571672404076,
      "loss": 2.2158,
      "step": 105301
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.848399817943573,
      "learning_rate": 0.00034037166337423674,
      "loss": 2.1712,
      "step": 105302
    },
    {
      "epoch": 1.37,
      "grad_norm": 1.133634328842163,
      "learning_rate": 0.0003403676100169269,
      "loss": 2.267,
      "step": 105303
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.9270059466362,
      "learning_rate": 0.0003403635566521118,
      "loss": 2.0767,
      "step": 105304
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.8755248785018921,
      "learning_rate": 0.00034035950327979255,
      "loss": 1.9155,
      "step": 105305
    },
    {
      "epoch": 1.37,
      "grad_norm": 1.1676045656204224,
      "learning_rate": 0.00034035544989996963,
      "loss": 2.1531,
      "step": 105306
    },
    {
      "epoch": 1.37,
      "grad_norm": 1.1777195930480957,
      "learning_rate": 0.00034035139651264387,
      "loss": 2.5024,
      "step": 105307
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.8507381677627563,
      "learning_rate": 0.000340347343117816,
      "loss": 2.2207,
      "step": 105308
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.933431088924408,
      "learning_rate": 0.0003403432897154868,
      "loss": 2.2439,
      "step": 105309
    },
    {
      "epoch": 1.37,
      "grad_norm": 1.012935996055603,
      "learning_rate": 0.000340339236305657,
      "loss": 2.3478,
      "step": 105310
    },
    {
      "epoch": 1.37,
      "grad_norm": 1.306390404701233,
      "learning_rate": 0.0003403351828883275,
      "loss": 2.118,
      "step": 105311
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.9524794220924377,
      "learning_rate": 0.00034033112946349887,
      "loss": 1.8662,
      "step": 105312
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.9776872396469116,
      "learning_rate": 0.0003403270760311719,
      "loss": 2.2422,
      "step": 105313
    },
    {
      "epoch": 1.37,
      "grad_norm": 1.239725947380066,
      "learning_rate": 0.0003403230225913473,
      "loss": 1.9572,
      "step": 105314
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.9593533277511597,
      "learning_rate": 0.0003403189691440259,
      "loss": 2.1096,
      "step": 105315
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.8636461496353149,
      "learning_rate": 0.0003403149156892085,
      "loss": 2.1341,
      "step": 105316
    },
    {
      "epoch": 1.37,
      "grad_norm": 1.0028855800628662,
      "learning_rate": 0.00034031086222689586,
      "loss": 2.2179,
      "step": 105317
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.8994293808937073,
      "learning_rate": 0.0003403068087570886,
      "loss": 2.0735,
      "step": 105318
    },
    {
      "epoch": 1.37,
      "grad_norm": 1.0862476825714111,
      "learning_rate": 0.0003403027552797875,
      "loss": 2.1385,
      "step": 105319
    },
    {
      "epoch": 1.37,
      "grad_norm": 1.0215760469436646,
      "learning_rate": 0.00034029870179499343,
      "loss": 2.1037,
      "step": 105320
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.9411314725875854,
      "learning_rate": 0.00034029464830270703,
      "loss": 2.0636,
      "step": 105321
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.9924456477165222,
      "learning_rate": 0.0003402905948029291,
      "loss": 2.0699,
      "step": 105322
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.8707526326179504,
      "learning_rate": 0.0003402865412956604,
      "loss": 2.0505,
      "step": 105323
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.8552878499031067,
      "learning_rate": 0.0003402824877809017,
      "loss": 2.1129,
      "step": 105324
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.9029509425163269,
      "learning_rate": 0.0003402784342586536,
      "loss": 2.2174,
      "step": 105325
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.8579456210136414,
      "learning_rate": 0.00034027438072891714,
      "loss": 2.2212,
      "step": 105326
    },
    {
      "epoch": 1.37,
      "grad_norm": 1.1226396560668945,
      "learning_rate": 0.0003402703271916928,
      "loss": 2.096,
      "step": 105327
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.8325591683387756,
      "learning_rate": 0.0003402662736469815,
      "loss": 2.2472,
      "step": 105328
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.8749786615371704,
      "learning_rate": 0.0003402622200947839,
      "loss": 2.0065,
      "step": 105329
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.9300228953361511,
      "learning_rate": 0.00034025816653510083,
      "loss": 2.2164,
      "step": 105330
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.844169020652771,
      "learning_rate": 0.00034025411296793303,
      "loss": 1.8799,
      "step": 105331
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.8814578056335449,
      "learning_rate": 0.00034025005939328115,
      "loss": 2.4133,
      "step": 105332
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.8695613741874695,
      "learning_rate": 0.00034024600581114605,
      "loss": 2.1107,
      "step": 105333
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.8576890826225281,
      "learning_rate": 0.0003402419522215285,
      "loss": 2.0906,
      "step": 105334
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.9437593817710876,
      "learning_rate": 0.00034023789862442917,
      "loss": 2.2185,
      "step": 105335
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.9129024147987366,
      "learning_rate": 0.0003402338450198489,
      "loss": 2.2108,
      "step": 105336
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.8342543244361877,
      "learning_rate": 0.00034022979140778837,
      "loss": 2.1636,
      "step": 105337
    },
    {
      "epoch": 1.37,
      "grad_norm": 1.0123728513717651,
      "learning_rate": 0.00034022573778824834,
      "loss": 2.0573,
      "step": 105338
    },
    {
      "epoch": 1.37,
      "grad_norm": 1.008028507232666,
      "learning_rate": 0.00034022168416122966,
      "loss": 2.1032,
      "step": 105339
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.8235791325569153,
      "learning_rate": 0.00034021763052673296,
      "loss": 2.0134,
      "step": 105340
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.9374477863311768,
      "learning_rate": 0.000340213576884759,
      "loss": 2.1783,
      "step": 105341
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.910247266292572,
      "learning_rate": 0.00034020952323530866,
      "loss": 2.0298,
      "step": 105342
    },
    {
      "epoch": 1.37,
      "grad_norm": 1.0166622400283813,
      "learning_rate": 0.0003402054695783825,
      "loss": 2.1339,
      "step": 105343
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.8950659036636353,
      "learning_rate": 0.00034020141591398147,
      "loss": 2.1844,
      "step": 105344
    },
    {
      "epoch": 1.37,
      "grad_norm": 1.0226675271987915,
      "learning_rate": 0.0003401973622421063,
      "loss": 2.2482,
      "step": 105345
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.89723801612854,
      "learning_rate": 0.00034019330856275757,
      "loss": 2.0972,
      "step": 105346
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.9115262627601624,
      "learning_rate": 0.00034018925487593613,
      "loss": 2.1345,
      "step": 105347
    },
    {
      "epoch": 1.37,
      "grad_norm": 1.0477871894836426,
      "learning_rate": 0.0003401852011816429,
      "loss": 2.1939,
      "step": 105348
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.8996912240982056,
      "learning_rate": 0.0003401811474798783,
      "loss": 2.1396,
      "step": 105349
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.8806056976318359,
      "learning_rate": 0.00034017709377064337,
      "loss": 2.1791,
      "step": 105350
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.976157546043396,
      "learning_rate": 0.0003401730400539388,
      "loss": 2.1477,
      "step": 105351
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.9271097183227539,
      "learning_rate": 0.0003401689863297652,
      "loss": 2.0558,
      "step": 105352
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.8693876266479492,
      "learning_rate": 0.0003401649325981234,
      "loss": 2.1568,
      "step": 105353
    },
    {
      "epoch": 1.37,
      "grad_norm": 1.0098764896392822,
      "learning_rate": 0.00034016087885901437,
      "loss": 2.069,
      "step": 105354
    },
    {
      "epoch": 1.37,
      "grad_norm": 1.852100133895874,
      "learning_rate": 0.0003401568251124385,
      "loss": 2.2828,
      "step": 105355
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.8458263874053955,
      "learning_rate": 0.0003401527713583968,
      "loss": 2.2248,
      "step": 105356
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.90650475025177,
      "learning_rate": 0.0003401487175968899,
      "loss": 2.1684,
      "step": 105357
    },
    {
      "epoch": 1.37,
      "grad_norm": 1.050903081893921,
      "learning_rate": 0.0003401446638279186,
      "loss": 2.2543,
      "step": 105358
    },
    {
      "epoch": 1.37,
      "grad_norm": 1.3515114784240723,
      "learning_rate": 0.0003401406100514837,
      "loss": 2.3877,
      "step": 105359
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.9094094038009644,
      "learning_rate": 0.00034013655626758586,
      "loss": 2.1571,
      "step": 105360
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.9090128540992737,
      "learning_rate": 0.00034013250247622587,
      "loss": 2.061,
      "step": 105361
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.9328492283821106,
      "learning_rate": 0.00034012844867740445,
      "loss": 2.1597,
      "step": 105362
    },
    {
      "epoch": 1.37,
      "grad_norm": 1.1407227516174316,
      "learning_rate": 0.0003401243948711224,
      "loss": 2.1305,
      "step": 105363
    },
    {
      "epoch": 1.37,
      "grad_norm": 1.2351295948028564,
      "learning_rate": 0.0003401203410573806,
      "loss": 2.2608,
      "step": 105364
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.9305551052093506,
      "learning_rate": 0.00034011628723617956,
      "loss": 2.0899,
      "step": 105365
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.9582388401031494,
      "learning_rate": 0.00034011223340752014,
      "loss": 2.1628,
      "step": 105366
    },
    {
      "epoch": 1.37,
      "grad_norm": 1.0216343402862549,
      "learning_rate": 0.00034010817957140315,
      "loss": 2.1481,
      "step": 105367
    },
    {
      "epoch": 1.37,
      "grad_norm": 1.064523458480835,
      "learning_rate": 0.0003401041257278293,
      "loss": 2.0845,
      "step": 105368
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.8557673096656799,
      "learning_rate": 0.00034010007187679924,
      "loss": 2.3027,
      "step": 105369
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.9938151836395264,
      "learning_rate": 0.00034009601801831386,
      "loss": 2.1181,
      "step": 105370
    },
    {
      "epoch": 1.37,
      "grad_norm": 1.110231876373291,
      "learning_rate": 0.000340091964152374,
      "loss": 2.2233,
      "step": 105371
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.9783088564872742,
      "learning_rate": 0.0003400879102789801,
      "loss": 2.1939,
      "step": 105372
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.9515185952186584,
      "learning_rate": 0.00034008385639813316,
      "loss": 2.2783,
      "step": 105373
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.8755408525466919,
      "learning_rate": 0.0003400798025098339,
      "loss": 2.0049,
      "step": 105374
    },
    {
      "epoch": 1.37,
      "grad_norm": 1.099949598312378,
      "learning_rate": 0.00034007574861408306,
      "loss": 2.1037,
      "step": 105375
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.9215933084487915,
      "learning_rate": 0.0003400716947108814,
      "loss": 2.1025,
      "step": 105376
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.8351098895072937,
      "learning_rate": 0.00034006764080022955,
      "loss": 2.1315,
      "step": 105377
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.8776986598968506,
      "learning_rate": 0.00034006358688212855,
      "loss": 2.2344,
      "step": 105378
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.9549275040626526,
      "learning_rate": 0.0003400595329565788,
      "loss": 2.1751,
      "step": 105379
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.9223237037658691,
      "learning_rate": 0.00034005547902358124,
      "loss": 2.3533,
      "step": 105380
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.8504732847213745,
      "learning_rate": 0.00034005142508313673,
      "loss": 2.0887,
      "step": 105381
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.8826135993003845,
      "learning_rate": 0.0003400473711352458,
      "loss": 2.1201,
      "step": 105382
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.9052810668945312,
      "learning_rate": 0.0003400433171799094,
      "loss": 2.1083,
      "step": 105383
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.9710864424705505,
      "learning_rate": 0.0003400392632171281,
      "loss": 2.0961,
      "step": 105384
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.8949624300003052,
      "learning_rate": 0.00034003520924690277,
      "loss": 2.2934,
      "step": 105385
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.9025180339813232,
      "learning_rate": 0.0003400311552692342,
      "loss": 2.2912,
      "step": 105386
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.902945339679718,
      "learning_rate": 0.00034002710128412303,
      "loss": 2.0464,
      "step": 105387
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.9366313815116882,
      "learning_rate": 0.00034002304729157,
      "loss": 2.3072,
      "step": 105388
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.8411122560501099,
      "learning_rate": 0.000340018993291576,
      "loss": 2.2352,
      "step": 105389
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.8581629395484924,
      "learning_rate": 0.00034001493928414176,
      "loss": 2.0896,
      "step": 105390
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.8515493273735046,
      "learning_rate": 0.0003400108852692679,
      "loss": 2.0874,
      "step": 105391
    },
    {
      "epoch": 1.37,
      "grad_norm": 1.0063397884368896,
      "learning_rate": 0.00034000683124695534,
      "loss": 2.073,
      "step": 105392
    },
    {
      "epoch": 1.37,
      "grad_norm": 1.1543110609054565,
      "learning_rate": 0.0003400027772172047,
      "loss": 2.0936,
      "step": 105393
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.9569587111473083,
      "learning_rate": 0.00033999872318001684,
      "loss": 2.0863,
      "step": 105394
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.9197187423706055,
      "learning_rate": 0.0003399946691353924,
      "loss": 2.156,
      "step": 105395
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.92380291223526,
      "learning_rate": 0.0003399906150833322,
      "loss": 2.0578,
      "step": 105396
    },
    {
      "epoch": 1.37,
      "grad_norm": 1.053903579711914,
      "learning_rate": 0.00033998656102383705,
      "loss": 2.0891,
      "step": 105397
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.9448107481002808,
      "learning_rate": 0.0003399825069569076,
      "loss": 2.3634,
      "step": 105398
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.9638064503669739,
      "learning_rate": 0.0003399784528825446,
      "loss": 2.3203,
      "step": 105399
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.9718546271324158,
      "learning_rate": 0.00033997439880074904,
      "loss": 2.133,
      "step": 105400
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.905302882194519,
      "learning_rate": 0.0003399703447115213,
      "loss": 2.1279,
      "step": 105401
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.8236689567565918,
      "learning_rate": 0.00033996629061486234,
      "loss": 2.1265,
      "step": 105402
    },
    {
      "epoch": 1.37,
      "grad_norm": 1.0353652238845825,
      "learning_rate": 0.00033996223651077296,
      "loss": 1.9936,
      "step": 105403
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.9413166046142578,
      "learning_rate": 0.0003399581823992538,
      "loss": 2.1262,
      "step": 105404
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.8715804815292358,
      "learning_rate": 0.00033995412828030565,
      "loss": 1.9403,
      "step": 105405
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.9012866020202637,
      "learning_rate": 0.0003399500741539293,
      "loss": 1.9816,
      "step": 105406
    },
    {
      "epoch": 1.37,
      "grad_norm": 1.0554062128067017,
      "learning_rate": 0.0003399460200201254,
      "loss": 2.2671,
      "step": 105407
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.825899064540863,
      "learning_rate": 0.00033994196587889484,
      "loss": 2.1923,
      "step": 105408
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.8451429009437561,
      "learning_rate": 0.0003399379117302384,
      "loss": 2.3066,
      "step": 105409
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.8805256485939026,
      "learning_rate": 0.00033993385757415665,
      "loss": 1.9474,
      "step": 105410
    },
    {
      "epoch": 1.37,
      "grad_norm": 1.1255745887756348,
      "learning_rate": 0.00033992980341065044,
      "loss": 2.1126,
      "step": 105411
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.9032387733459473,
      "learning_rate": 0.0003399257492397206,
      "loss": 1.9064,
      "step": 105412
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.8517321944236755,
      "learning_rate": 0.0003399216950613677,
      "loss": 2.2763,
      "step": 105413
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.9727795124053955,
      "learning_rate": 0.0003399176408755927,
      "loss": 1.9898,
      "step": 105414
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.9473298788070679,
      "learning_rate": 0.0003399135866823962,
      "loss": 2.2918,
      "step": 105415
    },
    {
      "epoch": 1.37,
      "grad_norm": 1.027596354484558,
      "learning_rate": 0.00033990953248177905,
      "loss": 2.3926,
      "step": 105416
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.9452863335609436,
      "learning_rate": 0.0003399054782737419,
      "loss": 2.0945,
      "step": 105417
    },
    {
      "epoch": 1.37,
      "grad_norm": 1.0250340700149536,
      "learning_rate": 0.00033990142405828565,
      "loss": 2.1063,
      "step": 105418
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.8973882794380188,
      "learning_rate": 0.0003398973698354109,
      "loss": 2.0074,
      "step": 105419
    },
    {
      "epoch": 1.37,
      "grad_norm": 1.025590181350708,
      "learning_rate": 0.0003398933156051185,
      "loss": 2.2073,
      "step": 105420
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.9582241177558899,
      "learning_rate": 0.0003398892613674092,
      "loss": 2.0882,
      "step": 105421
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.8257347941398621,
      "learning_rate": 0.00033988520712228374,
      "loss": 2.3243,
      "step": 105422
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.8942044973373413,
      "learning_rate": 0.00033988115286974285,
      "loss": 2.2318,
      "step": 105423
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.8624157309532166,
      "learning_rate": 0.00033987709860978737,
      "loss": 2.1308,
      "step": 105424
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.8682435154914856,
      "learning_rate": 0.00033987304434241784,
      "loss": 2.1011,
      "step": 105425
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.928890585899353,
      "learning_rate": 0.00033986899006763525,
      "loss": 2.112,
      "step": 105426
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.9298461675643921,
      "learning_rate": 0.0003398649357854403,
      "loss": 2.1934,
      "step": 105427
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.9615416526794434,
      "learning_rate": 0.00033986088149583356,
      "loss": 2.0247,
      "step": 105428
    },
    {
      "epoch": 1.37,
      "grad_norm": 1.146550178527832,
      "learning_rate": 0.00033985682719881604,
      "loss": 2.2016,
      "step": 105429
    },
    {
      "epoch": 1.37,
      "grad_norm": 1.0600879192352295,
      "learning_rate": 0.00033985277289438843,
      "loss": 1.996,
      "step": 105430
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.8109734058380127,
      "learning_rate": 0.00033984871858255133,
      "loss": 2.0803,
      "step": 105431
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.837988018989563,
      "learning_rate": 0.0003398446642633056,
      "loss": 1.993,
      "step": 105432
    },
    {
      "epoch": 1.37,
      "grad_norm": 1.1917061805725098,
      "learning_rate": 0.0003398406099366521,
      "loss": 2.1739,
      "step": 105433
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.9410598278045654,
      "learning_rate": 0.00033983655560259143,
      "loss": 2.0943,
      "step": 105434
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.8737201690673828,
      "learning_rate": 0.00033983250126112433,
      "loss": 2.1223,
      "step": 105435
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.9577955603599548,
      "learning_rate": 0.0003398284469122517,
      "loss": 2.2033,
      "step": 105436
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.9463978409767151,
      "learning_rate": 0.00033982439255597413,
      "loss": 2.3501,
      "step": 105437
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.8815115690231323,
      "learning_rate": 0.0003398203381922925,
      "loss": 2.2602,
      "step": 105438
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.9906864762306213,
      "learning_rate": 0.0003398162838212076,
      "loss": 2.0942,
      "step": 105439
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.9314732551574707,
      "learning_rate": 0.00033981222944271997,
      "loss": 2.3872,
      "step": 105440
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.8699579238891602,
      "learning_rate": 0.00033980817505683056,
      "loss": 2.1166,
      "step": 105441
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.9128488898277283,
      "learning_rate": 0.00033980412066354006,
      "loss": 2.0904,
      "step": 105442
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.9481762647628784,
      "learning_rate": 0.0003398000662628492,
      "loss": 2.0415,
      "step": 105443
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.8836494088172913,
      "learning_rate": 0.0003397960118547587,
      "loss": 2.1186,
      "step": 105444
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.8150844573974609,
      "learning_rate": 0.00033979195743926945,
      "loss": 2.2037,
      "step": 105445
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.9929764866828918,
      "learning_rate": 0.0003397879030163821,
      "loss": 2.2091,
      "step": 105446
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.9326667785644531,
      "learning_rate": 0.0003397838485860974,
      "loss": 2.2979,
      "step": 105447
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.8607356548309326,
      "learning_rate": 0.0003397797941484162,
      "loss": 2.2171,
      "step": 105448
    },
    {
      "epoch": 1.37,
      "grad_norm": 1.1419248580932617,
      "learning_rate": 0.0003397757397033391,
      "loss": 2.2712,
      "step": 105449
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.9816635251045227,
      "learning_rate": 0.00033977168525086706,
      "loss": 2.2942,
      "step": 105450
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.9151907563209534,
      "learning_rate": 0.0003397676307910006,
      "loss": 2.164,
      "step": 105451
    },
    {
      "epoch": 1.37,
      "grad_norm": 1.002536654472351,
      "learning_rate": 0.00033976357632374066,
      "loss": 1.9228,
      "step": 105452
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.9686702489852905,
      "learning_rate": 0.00033975952184908784,
      "loss": 1.9551,
      "step": 105453
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.9376950860023499,
      "learning_rate": 0.00033975546736704303,
      "loss": 2.2429,
      "step": 105454
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.8446204662322998,
      "learning_rate": 0.0003397514128776069,
      "loss": 2.1583,
      "step": 105455
    },
    {
      "epoch": 1.37,
      "grad_norm": 1.045576572418213,
      "learning_rate": 0.0003397473583807803,
      "loss": 2.2741,
      "step": 105456
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.8674914836883545,
      "learning_rate": 0.0003397433038765639,
      "loss": 2.1763,
      "step": 105457
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.869787335395813,
      "learning_rate": 0.00033973924936495835,
      "loss": 2.114,
      "step": 105458
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.8423636555671692,
      "learning_rate": 0.00033973519484596464,
      "loss": 2.1726,
      "step": 105459
    },
    {
      "epoch": 1.37,
      "grad_norm": 1.0296013355255127,
      "learning_rate": 0.00033973114031958334,
      "loss": 2.2913,
      "step": 105460
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.8374065160751343,
      "learning_rate": 0.00033972708578581536,
      "loss": 2.2252,
      "step": 105461
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.9329709410667419,
      "learning_rate": 0.00033972303124466126,
      "loss": 2.0498,
      "step": 105462
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.8787054419517517,
      "learning_rate": 0.000339718976696122,
      "loss": 1.9593,
      "step": 105463
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.8875816464424133,
      "learning_rate": 0.00033971492214019815,
      "loss": 2.1872,
      "step": 105464
    },
    {
      "epoch": 1.37,
      "grad_norm": 1.0089308023452759,
      "learning_rate": 0.0003397108675768906,
      "loss": 2.2428,
      "step": 105465
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.9095314145088196,
      "learning_rate": 0.00033970681300620007,
      "loss": 2.0406,
      "step": 105466
    },
    {
      "epoch": 1.37,
      "grad_norm": 1.0235122442245483,
      "learning_rate": 0.00033970275842812726,
      "loss": 2.2114,
      "step": 105467
    },
    {
      "epoch": 1.37,
      "grad_norm": 1.4237719774246216,
      "learning_rate": 0.00033969870384267293,
      "loss": 2.2186,
      "step": 105468
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.9615887403488159,
      "learning_rate": 0.00033969464924983784,
      "loss": 2.1051,
      "step": 105469
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.8947567939758301,
      "learning_rate": 0.0003396905946496228,
      "loss": 1.9838,
      "step": 105470
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.882843017578125,
      "learning_rate": 0.0003396865400420285,
      "loss": 2.261,
      "step": 105471
    },
    {
      "epoch": 1.37,
      "grad_norm": 1.0215810537338257,
      "learning_rate": 0.0003396824854270558,
      "loss": 2.284,
      "step": 105472
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.9129270911216736,
      "learning_rate": 0.0003396784308047054,
      "loss": 2.0714,
      "step": 105473
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.9485273361206055,
      "learning_rate": 0.00033967437617497793,
      "loss": 2.1038,
      "step": 105474
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.9114347696304321,
      "learning_rate": 0.00033967032153787427,
      "loss": 2.1998,
      "step": 105475
    },
    {
      "epoch": 1.37,
      "grad_norm": 1.071012020111084,
      "learning_rate": 0.0003396662668933952,
      "loss": 2.0289,
      "step": 105476
    },
    {
      "epoch": 1.37,
      "grad_norm": 1.1017334461212158,
      "learning_rate": 0.00033966221224154136,
      "loss": 2.0157,
      "step": 105477
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.8669363856315613,
      "learning_rate": 0.0003396581575823136,
      "loss": 2.195,
      "step": 105478
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.8816256523132324,
      "learning_rate": 0.0003396541029157126,
      "loss": 2.0885,
      "step": 105479
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.9843491315841675,
      "learning_rate": 0.0003396500482417392,
      "loss": 1.9842,
      "step": 105480
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.9266684055328369,
      "learning_rate": 0.00033964599356039413,
      "loss": 2.2483,
      "step": 105481
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.8622922897338867,
      "learning_rate": 0.00033964193887167806,
      "loss": 2.151,
      "step": 105482
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.8046783804893494,
      "learning_rate": 0.00033963788417559185,
      "loss": 2.1743,
      "step": 105483
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.9759461283683777,
      "learning_rate": 0.00033963382947213615,
      "loss": 2.0351,
      "step": 105484
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.8388956785202026,
      "learning_rate": 0.00033962977476131184,
      "loss": 2.1342,
      "step": 105485
    },
    {
      "epoch": 1.37,
      "grad_norm": 1.0332616567611694,
      "learning_rate": 0.00033962572004311956,
      "loss": 2.2716,
      "step": 105486
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.8825153112411499,
      "learning_rate": 0.0003396216653175601,
      "loss": 2.1219,
      "step": 105487
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.9039750099182129,
      "learning_rate": 0.0003396176105846343,
      "loss": 2.2151,
      "step": 105488
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.9909451603889465,
      "learning_rate": 0.0003396135558443428,
      "loss": 2.1876,
      "step": 105489
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.8722031116485596,
      "learning_rate": 0.00033960950109668633,
      "loss": 2.0928,
      "step": 105490
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.9023967981338501,
      "learning_rate": 0.00033960544634166587,
      "loss": 2.2043,
      "step": 105491
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.9768064022064209,
      "learning_rate": 0.0003396013915792819,
      "loss": 2.0067,
      "step": 105492
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.9821826815605164,
      "learning_rate": 0.0003395973368095352,
      "loss": 2.2882,
      "step": 105493
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.9154091477394104,
      "learning_rate": 0.0003395932820324268,
      "loss": 2.1216,
      "step": 105494
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.947292685508728,
      "learning_rate": 0.00033958922724795706,
      "loss": 2.2975,
      "step": 105495
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.8441590666770935,
      "learning_rate": 0.000339585172456127,
      "loss": 2.5216,
      "step": 105496
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.9193596839904785,
      "learning_rate": 0.0003395811176569374,
      "loss": 2.2213,
      "step": 105497
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.9886738657951355,
      "learning_rate": 0.0003395770628503889,
      "loss": 2.3877,
      "step": 105498
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.9296786785125732,
      "learning_rate": 0.0003395730080364822,
      "loss": 2.2233,
      "step": 105499
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.9340580105781555,
      "learning_rate": 0.00033956895321521813,
      "loss": 2.2991,
      "step": 105500
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.8665855526924133,
      "learning_rate": 0.0003395648983865975,
      "loss": 2.0003,
      "step": 105501
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.9740222692489624,
      "learning_rate": 0.00033956084355062094,
      "loss": 2.2016,
      "step": 105502
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.8329911231994629,
      "learning_rate": 0.0003395567887072894,
      "loss": 2.2266,
      "step": 105503
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.8335551619529724,
      "learning_rate": 0.00033955273385660343,
      "loss": 2.1441,
      "step": 105504
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.9223210215568542,
      "learning_rate": 0.0003395486789985638,
      "loss": 1.9767,
      "step": 105505
    },
    {
      "epoch": 1.37,
      "grad_norm": 1.0512497425079346,
      "learning_rate": 0.00033954462413317147,
      "loss": 1.8395,
      "step": 105506
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.8522105813026428,
      "learning_rate": 0.00033954056926042693,
      "loss": 2.145,
      "step": 105507
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.9652092456817627,
      "learning_rate": 0.00033953651438033106,
      "loss": 2.3288,
      "step": 105508
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.9592653512954712,
      "learning_rate": 0.0003395324594928847,
      "loss": 2.0604,
      "step": 105509
    },
    {
      "epoch": 1.37,
      "grad_norm": 1.0054216384887695,
      "learning_rate": 0.00033952840459808845,
      "loss": 2.0403,
      "step": 105510
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.8818702697753906,
      "learning_rate": 0.00033952434969594306,
      "loss": 2.2184,
      "step": 105511
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.9281243681907654,
      "learning_rate": 0.0003395202947864495,
      "loss": 2.1453,
      "step": 105512
    },
    {
      "epoch": 1.37,
      "grad_norm": 1.0194385051727295,
      "learning_rate": 0.00033951623986960816,
      "loss": 2.2742,
      "step": 105513
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.9004603624343872,
      "learning_rate": 0.00033951218494542013,
      "loss": 2.0155,
      "step": 105514
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.846798300743103,
      "learning_rate": 0.0003395081300138861,
      "loss": 2.2487,
      "step": 105515
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.9931315183639526,
      "learning_rate": 0.00033950407507500667,
      "loss": 1.9449,
      "step": 105516
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.8619812726974487,
      "learning_rate": 0.00033950002012878266,
      "loss": 2.1217,
      "step": 105517
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.8940689563751221,
      "learning_rate": 0.0003394959651752149,
      "loss": 2.1763,
      "step": 105518
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.9623900651931763,
      "learning_rate": 0.0003394919102143041,
      "loss": 2.2266,
      "step": 105519
    },
    {
      "epoch": 1.37,
      "grad_norm": 1.1505708694458008,
      "learning_rate": 0.000339487855246051,
      "loss": 2.1484,
      "step": 105520
    },
    {
      "epoch": 1.37,
      "grad_norm": 1.0207196474075317,
      "learning_rate": 0.0003394838002704564,
      "loss": 2.0825,
      "step": 105521
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.8742876052856445,
      "learning_rate": 0.0003394797452875209,
      "loss": 1.9464,
      "step": 105522
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.8914895057678223,
      "learning_rate": 0.00033947569029724547,
      "loss": 2.0539,
      "step": 105523
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.9080018997192383,
      "learning_rate": 0.0003394716352996308,
      "loss": 2.0753,
      "step": 105524
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.968531608581543,
      "learning_rate": 0.0003394675802946775,
      "loss": 2.2391,
      "step": 105525
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.9181066751480103,
      "learning_rate": 0.0003394635252823865,
      "loss": 2.1647,
      "step": 105526
    },
    {
      "epoch": 1.37,
      "grad_norm": 1.058152437210083,
      "learning_rate": 0.00033945947026275844,
      "loss": 2.2059,
      "step": 105527
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.9656702876091003,
      "learning_rate": 0.00033945541523579416,
      "loss": 1.9583,
      "step": 105528
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.8880947232246399,
      "learning_rate": 0.00033945136020149434,
      "loss": 2.2602,
      "step": 105529
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.9492307305335999,
      "learning_rate": 0.0003394473051598598,
      "loss": 2.1137,
      "step": 105530
    },
    {
      "epoch": 1.37,
      "grad_norm": 1.041279673576355,
      "learning_rate": 0.00033944325011089124,
      "loss": 2.324,
      "step": 105531
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.8063094615936279,
      "learning_rate": 0.0003394391950545894,
      "loss": 1.9809,
      "step": 105532
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.863246738910675,
      "learning_rate": 0.00033943513999095515,
      "loss": 2.0748,
      "step": 105533
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.8967410922050476,
      "learning_rate": 0.00033943108491998913,
      "loss": 2.2784,
      "step": 105534
    },
    {
      "epoch": 1.37,
      "grad_norm": 1.1255296468734741,
      "learning_rate": 0.0003394270298416921,
      "loss": 2.1958,
      "step": 105535
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.9166228175163269,
      "learning_rate": 0.0003394229747560649,
      "loss": 2.2158,
      "step": 105536
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.988323986530304,
      "learning_rate": 0.0003394189196631082,
      "loss": 2.2183,
      "step": 105537
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.9021956920623779,
      "learning_rate": 0.00033941486456282277,
      "loss": 2.1356,
      "step": 105538
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.9331368803977966,
      "learning_rate": 0.00033941080945520935,
      "loss": 2.1049,
      "step": 105539
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.9154418706893921,
      "learning_rate": 0.0003394067543402688,
      "loss": 2.124,
      "step": 105540
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.8669907450675964,
      "learning_rate": 0.0003394026992180017,
      "loss": 2.2152,
      "step": 105541
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.9899133443832397,
      "learning_rate": 0.0003393986440884089,
      "loss": 2.1468,
      "step": 105542
    },
    {
      "epoch": 1.37,
      "grad_norm": 1.0218164920806885,
      "learning_rate": 0.00033939458895149117,
      "loss": 2.3261,
      "step": 105543
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.8672780990600586,
      "learning_rate": 0.00033939053380724923,
      "loss": 1.9628,
      "step": 105544
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.9695227742195129,
      "learning_rate": 0.00033938647865568383,
      "loss": 2.102,
      "step": 105545
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.8530851006507874,
      "learning_rate": 0.0003393824234967958,
      "loss": 2.0942,
      "step": 105546
    },
    {
      "epoch": 1.37,
      "grad_norm": 1.027768850326538,
      "learning_rate": 0.0003393783683305858,
      "loss": 2.2032,
      "step": 105547
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.8651174902915955,
      "learning_rate": 0.00033937431315705463,
      "loss": 2.2176,
      "step": 105548
    },
    {
      "epoch": 1.37,
      "grad_norm": 1.071396827697754,
      "learning_rate": 0.000339370257976203,
      "loss": 2.1095,
      "step": 105549
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.9148478507995605,
      "learning_rate": 0.0003393662027880317,
      "loss": 2.0808,
      "step": 105550
    },
    {
      "epoch": 1.37,
      "grad_norm": 1.0084483623504639,
      "learning_rate": 0.00033936214759254155,
      "loss": 2.3486,
      "step": 105551
    },
    {
      "epoch": 1.37,
      "grad_norm": 1.0449215173721313,
      "learning_rate": 0.0003393580923897331,
      "loss": 2.2441,
      "step": 105552
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.8940521478652954,
      "learning_rate": 0.0003393540371796073,
      "loss": 2.0715,
      "step": 105553
    },
    {
      "epoch": 1.37,
      "grad_norm": 1.1612688302993774,
      "learning_rate": 0.00033934998196216487,
      "loss": 2.071,
      "step": 105554
    },
    {
      "epoch": 1.37,
      "grad_norm": 1.0220779180526733,
      "learning_rate": 0.0003393459267374066,
      "loss": 2.1738,
      "step": 105555
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.8564106225967407,
      "learning_rate": 0.00033934187150533307,
      "loss": 1.9068,
      "step": 105556
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.982977032661438,
      "learning_rate": 0.0003393378162659452,
      "loss": 2.2899,
      "step": 105557
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.930680513381958,
      "learning_rate": 0.0003393337610192437,
      "loss": 2.2496,
      "step": 105558
    },
    {
      "epoch": 1.37,
      "grad_norm": 1.0514694452285767,
      "learning_rate": 0.0003393297057652292,
      "loss": 2.1268,
      "step": 105559
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.8980682492256165,
      "learning_rate": 0.00033932565050390266,
      "loss": 2.1428,
      "step": 105560
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.8724448680877686,
      "learning_rate": 0.0003393215952352648,
      "loss": 2.2416,
      "step": 105561
    },
    {
      "epoch": 1.37,
      "grad_norm": 1.1043182611465454,
      "learning_rate": 0.0003393175399593162,
      "loss": 2.0619,
      "step": 105562
    },
    {
      "epoch": 1.37,
      "grad_norm": 1.0374668836593628,
      "learning_rate": 0.00033931348467605775,
      "loss": 2.3135,
      "step": 105563
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.8458759784698486,
      "learning_rate": 0.0003393094293854902,
      "loss": 1.924,
      "step": 105564
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.8269603252410889,
      "learning_rate": 0.00033930537408761427,
      "loss": 2.2954,
      "step": 105565
    },
    {
      "epoch": 1.37,
      "grad_norm": 1.1238248348236084,
      "learning_rate": 0.0003393013187824307,
      "loss": 2.1132,
      "step": 105566
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.9889182448387146,
      "learning_rate": 0.0003392972634699403,
      "loss": 1.973,
      "step": 105567
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.8534659147262573,
      "learning_rate": 0.00033929320815014383,
      "loss": 2.1978,
      "step": 105568
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.9107702374458313,
      "learning_rate": 0.000339289152823042,
      "loss": 2.2734,
      "step": 105569
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.9250601530075073,
      "learning_rate": 0.00033928509748863564,
      "loss": 2.1625,
      "step": 105570
    },
    {
      "epoch": 1.37,
      "grad_norm": 1.257631540298462,
      "learning_rate": 0.0003392810421469253,
      "loss": 2.2317,
      "step": 105571
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.7896707653999329,
      "learning_rate": 0.00033927698679791195,
      "loss": 2.2204,
      "step": 105572
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.9173612594604492,
      "learning_rate": 0.0003392729314415963,
      "loss": 2.2496,
      "step": 105573
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.8955510854721069,
      "learning_rate": 0.000339268876077979,
      "loss": 2.0271,
      "step": 105574
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.9569264650344849,
      "learning_rate": 0.00033926482070706083,
      "loss": 2.0825,
      "step": 105575
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.9171401858329773,
      "learning_rate": 0.00033926076532884275,
      "loss": 2.2918,
      "step": 105576
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.9012147784233093,
      "learning_rate": 0.00033925670994332523,
      "loss": 2.1251,
      "step": 105577
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.8835138082504272,
      "learning_rate": 0.00033925265455050913,
      "loss": 2.3994,
      "step": 105578
    },
    {
      "epoch": 1.37,
      "grad_norm": 1.0098886489868164,
      "learning_rate": 0.0003392485991503953,
      "loss": 2.2469,
      "step": 105579
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.8396874070167542,
      "learning_rate": 0.0003392445437429844,
      "loss": 2.2373,
      "step": 105580
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.9526689052581787,
      "learning_rate": 0.0003392404883282771,
      "loss": 2.1426,
      "step": 105581
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.8546109795570374,
      "learning_rate": 0.00033923643290627444,
      "loss": 2.3097,
      "step": 105582
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.9756127595901489,
      "learning_rate": 0.0003392323774769769,
      "loss": 2.2675,
      "step": 105583
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.9257087111473083,
      "learning_rate": 0.00033922832204038526,
      "loss": 2.2922,
      "step": 105584
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.8151795268058777,
      "learning_rate": 0.0003392242665965004,
      "loss": 2.0208,
      "step": 105585
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.9263833165168762,
      "learning_rate": 0.00033922021114532295,
      "loss": 2.0023,
      "step": 105586
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.9038638472557068,
      "learning_rate": 0.0003392161556868537,
      "loss": 2.1669,
      "step": 105587
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.8781387209892273,
      "learning_rate": 0.00033921210022109355,
      "loss": 2.149,
      "step": 105588
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.826755166053772,
      "learning_rate": 0.000339208044748043,
      "loss": 2.1562,
      "step": 105589
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.9322923421859741,
      "learning_rate": 0.000339203989267703,
      "loss": 2.2568,
      "step": 105590
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.8980060815811157,
      "learning_rate": 0.0003391999337800743,
      "loss": 2.3524,
      "step": 105591
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.8109309673309326,
      "learning_rate": 0.00033919587828515747,
      "loss": 1.9574,
      "step": 105592
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.9803030490875244,
      "learning_rate": 0.00033919182278295345,
      "loss": 2.1476,
      "step": 105593
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.8930419683456421,
      "learning_rate": 0.00033918776727346294,
      "loss": 1.9707,
      "step": 105594
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.8413490056991577,
      "learning_rate": 0.00033918371175668664,
      "loss": 2.0287,
      "step": 105595
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.9721240401268005,
      "learning_rate": 0.0003391796562326254,
      "loss": 2.1252,
      "step": 105596
    },
    {
      "epoch": 1.37,
      "grad_norm": 1.30643892288208,
      "learning_rate": 0.0003391756007012799,
      "loss": 1.8404,
      "step": 105597
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.865282416343689,
      "learning_rate": 0.00033917154516265095,
      "loss": 2.1331,
      "step": 105598
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.9301139712333679,
      "learning_rate": 0.00033916748961673916,
      "loss": 2.1109,
      "step": 105599
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.9128060340881348,
      "learning_rate": 0.00033916343406354543,
      "loss": 2.1757,
      "step": 105600
    },
    {
      "epoch": 1.37,
      "grad_norm": 0.9080715775489807,
      "learning_rate": 0.0003391593785030706,
      "loss": 2.1832,
      "step": 105601
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.8883801102638245,
      "learning_rate": 0.00033915532293531515,
      "loss": 2.2543,
      "step": 105602
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.8969346284866333,
      "learning_rate": 0.0003391512673602801,
      "loss": 2.0974,
      "step": 105603
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.910243570804596,
      "learning_rate": 0.0003391472117779661,
      "loss": 2.0907,
      "step": 105604
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.9136362671852112,
      "learning_rate": 0.0003391431561883738,
      "loss": 2.1791,
      "step": 105605
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.8963757157325745,
      "learning_rate": 0.0003391391005915041,
      "loss": 2.2311,
      "step": 105606
    },
    {
      "epoch": 1.38,
      "grad_norm": 1.1010500192642212,
      "learning_rate": 0.0003391350449873577,
      "loss": 2.0622,
      "step": 105607
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.8829947710037231,
      "learning_rate": 0.0003391309893759354,
      "loss": 2.0388,
      "step": 105608
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.8777186870574951,
      "learning_rate": 0.00033912693375723774,
      "loss": 2.1767,
      "step": 105609
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.9514949321746826,
      "learning_rate": 0.00033912287813126586,
      "loss": 2.1345,
      "step": 105610
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.870322048664093,
      "learning_rate": 0.0003391188224980201,
      "loss": 2.315,
      "step": 105611
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.8515765070915222,
      "learning_rate": 0.0003391147668575016,
      "loss": 2.1975,
      "step": 105612
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.892348051071167,
      "learning_rate": 0.00033911071120971086,
      "loss": 2.0715,
      "step": 105613
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.9130329489707947,
      "learning_rate": 0.00033910665555464864,
      "loss": 2.32,
      "step": 105614
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.9178891777992249,
      "learning_rate": 0.00033910259989231576,
      "loss": 2.1943,
      "step": 105615
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.9436580538749695,
      "learning_rate": 0.0003390985442227131,
      "loss": 2.234,
      "step": 105616
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.9527488946914673,
      "learning_rate": 0.0003390944885458411,
      "loss": 2.1596,
      "step": 105617
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.9932137131690979,
      "learning_rate": 0.0003390904328617008,
      "loss": 2.11,
      "step": 105618
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.9065428376197815,
      "learning_rate": 0.0003390863771702929,
      "loss": 2.05,
      "step": 105619
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.8800243735313416,
      "learning_rate": 0.000339082321471618,
      "loss": 2.0496,
      "step": 105620
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.8837738037109375,
      "learning_rate": 0.00033907826576567697,
      "loss": 2.2341,
      "step": 105621
    },
    {
      "epoch": 1.38,
      "grad_norm": 1.0247950553894043,
      "learning_rate": 0.0003390742100524706,
      "loss": 2.0669,
      "step": 105622
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.8627461194992065,
      "learning_rate": 0.00033907015433199956,
      "loss": 2.2942,
      "step": 105623
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.9138631224632263,
      "learning_rate": 0.0003390660986042647,
      "loss": 2.0941,
      "step": 105624
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.8187224268913269,
      "learning_rate": 0.0003390620428692667,
      "loss": 1.9795,
      "step": 105625
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.8309436440467834,
      "learning_rate": 0.00033905798712700625,
      "loss": 2.138,
      "step": 105626
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.866358757019043,
      "learning_rate": 0.0003390539313774842,
      "loss": 2.1287,
      "step": 105627
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.8904163241386414,
      "learning_rate": 0.0003390498756207014,
      "loss": 2.3197,
      "step": 105628
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.8844089508056641,
      "learning_rate": 0.00033904581985665837,
      "loss": 2.2179,
      "step": 105629
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.8825900554656982,
      "learning_rate": 0.000339041764085356,
      "loss": 2.3429,
      "step": 105630
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.8412845730781555,
      "learning_rate": 0.00033903770830679504,
      "loss": 2.0039,
      "step": 105631
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.8269718885421753,
      "learning_rate": 0.00033903365252097636,
      "loss": 2.0853,
      "step": 105632
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.8670117855072021,
      "learning_rate": 0.0003390295967279004,
      "loss": 2.1347,
      "step": 105633
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.9279666543006897,
      "learning_rate": 0.00033902554092756816,
      "loss": 2.2277,
      "step": 105634
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.9293105602264404,
      "learning_rate": 0.0003390214851199803,
      "loss": 2.1266,
      "step": 105635
    },
    {
      "epoch": 1.38,
      "grad_norm": 1.0751396417617798,
      "learning_rate": 0.00033901742930513777,
      "loss": 2.3407,
      "step": 105636
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.890389084815979,
      "learning_rate": 0.000339013373483041,
      "loss": 2.3167,
      "step": 105637
    },
    {
      "epoch": 1.38,
      "grad_norm": 1.0426172018051147,
      "learning_rate": 0.00033900931765369093,
      "loss": 2.1957,
      "step": 105638
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.9950293898582458,
      "learning_rate": 0.0003390052618170884,
      "loss": 2.1568,
      "step": 105639
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.9644994139671326,
      "learning_rate": 0.00033900120597323393,
      "loss": 2.3784,
      "step": 105640
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.8614163994789124,
      "learning_rate": 0.00033899715012212844,
      "loss": 2.2563,
      "step": 105641
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.9024088978767395,
      "learning_rate": 0.0003389930942637726,
      "loss": 2.2626,
      "step": 105642
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.8721799254417419,
      "learning_rate": 0.0003389890383981673,
      "loss": 2.2464,
      "step": 105643
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.8430954813957214,
      "learning_rate": 0.0003389849825253132,
      "loss": 2.1621,
      "step": 105644
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.8191952109336853,
      "learning_rate": 0.000338980926645211,
      "loss": 2.1016,
      "step": 105645
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.966284453868866,
      "learning_rate": 0.00033897687075786154,
      "loss": 2.0632,
      "step": 105646
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.881898820400238,
      "learning_rate": 0.00033897281486326556,
      "loss": 2.279,
      "step": 105647
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.9198142886161804,
      "learning_rate": 0.0003389687589614237,
      "loss": 2.124,
      "step": 105648
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.8792635798454285,
      "learning_rate": 0.0003389647030523369,
      "loss": 2.1498,
      "step": 105649
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.9942840337753296,
      "learning_rate": 0.0003389606471360059,
      "loss": 2.2609,
      "step": 105650
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.9722182750701904,
      "learning_rate": 0.0003389565912124312,
      "loss": 2.1822,
      "step": 105651
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.9170197248458862,
      "learning_rate": 0.00033895253528161385,
      "loss": 2.2776,
      "step": 105652
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.8625652194023132,
      "learning_rate": 0.0003389484793435544,
      "loss": 2.3703,
      "step": 105653
    },
    {
      "epoch": 1.38,
      "grad_norm": 1.1876178979873657,
      "learning_rate": 0.0003389444233982538,
      "loss": 2.2427,
      "step": 105654
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.9236636161804199,
      "learning_rate": 0.00033894036744571267,
      "loss": 2.1474,
      "step": 105655
    },
    {
      "epoch": 1.38,
      "grad_norm": 1.122902750968933,
      "learning_rate": 0.00033893631148593174,
      "loss": 2.2451,
      "step": 105656
    },
    {
      "epoch": 1.38,
      "grad_norm": 1.0909724235534668,
      "learning_rate": 0.00033893225551891187,
      "loss": 2.2068,
      "step": 105657
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.9109954833984375,
      "learning_rate": 0.00033892819954465375,
      "loss": 2.3083,
      "step": 105658
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.9938237071037292,
      "learning_rate": 0.00033892414356315816,
      "loss": 2.1446,
      "step": 105659
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.866256058216095,
      "learning_rate": 0.0003389200875744257,
      "loss": 2.1456,
      "step": 105660
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.9405998587608337,
      "learning_rate": 0.0003389160315784574,
      "loss": 2.0897,
      "step": 105661
    },
    {
      "epoch": 1.38,
      "grad_norm": 1.0311826467514038,
      "learning_rate": 0.0003389119755752539,
      "loss": 2.1282,
      "step": 105662
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.9209864735603333,
      "learning_rate": 0.00033890791956481584,
      "loss": 1.8683,
      "step": 105663
    },
    {
      "epoch": 1.38,
      "grad_norm": 1.1532458066940308,
      "learning_rate": 0.00033890386354714405,
      "loss": 2.2592,
      "step": 105664
    },
    {
      "epoch": 1.38,
      "grad_norm": 1.127909779548645,
      "learning_rate": 0.00033889980752223943,
      "loss": 2.0261,
      "step": 105665
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.8952838182449341,
      "learning_rate": 0.0003388957514901025,
      "loss": 2.1042,
      "step": 105666
    },
    {
      "epoch": 1.38,
      "grad_norm": 1.1463831663131714,
      "learning_rate": 0.0003388916954507341,
      "loss": 2.2374,
      "step": 105667
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.8934096693992615,
      "learning_rate": 0.00033888763940413506,
      "loss": 2.043,
      "step": 105668
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.947563886642456,
      "learning_rate": 0.00033888358335030605,
      "loss": 2.2793,
      "step": 105669
    },
    {
      "epoch": 1.38,
      "grad_norm": 1.0691120624542236,
      "learning_rate": 0.00033887952728924774,
      "loss": 2.0907,
      "step": 105670
    },
    {
      "epoch": 1.38,
      "grad_norm": 1.0811107158660889,
      "learning_rate": 0.00033887547122096116,
      "loss": 2.4298,
      "step": 105671
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.9273285269737244,
      "learning_rate": 0.00033887141514544686,
      "loss": 2.3195,
      "step": 105672
    },
    {
      "epoch": 1.38,
      "grad_norm": 1.048344373703003,
      "learning_rate": 0.0003388673590627055,
      "loss": 2.1913,
      "step": 105673
    },
    {
      "epoch": 1.38,
      "grad_norm": 1.333471655845642,
      "learning_rate": 0.00033886330297273816,
      "loss": 2.2968,
      "step": 105674
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.8302818536758423,
      "learning_rate": 0.0003388592468755453,
      "loss": 2.1506,
      "step": 105675
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.9823098182678223,
      "learning_rate": 0.0003388551907711277,
      "loss": 1.9985,
      "step": 105676
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.9089959859848022,
      "learning_rate": 0.00033885113465948635,
      "loss": 2.1421,
      "step": 105677
    },
    {
      "epoch": 1.38,
      "grad_norm": 1.1595078706741333,
      "learning_rate": 0.0003388470785406217,
      "loss": 2.2283,
      "step": 105678
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.9345402121543884,
      "learning_rate": 0.00033884302241453466,
      "loss": 2.3985,
      "step": 105679
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.84180748462677,
      "learning_rate": 0.00033883896628122604,
      "loss": 2.1477,
      "step": 105680
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.8610353469848633,
      "learning_rate": 0.00033883491014069647,
      "loss": 2.3136,
      "step": 105681
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.9030670523643494,
      "learning_rate": 0.00033883085399294676,
      "loss": 2.2148,
      "step": 105682
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.9486516714096069,
      "learning_rate": 0.00033882679783797773,
      "loss": 2.1431,
      "step": 105683
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.8717252612113953,
      "learning_rate": 0.00033882274167579,
      "loss": 1.9371,
      "step": 105684
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.8759649395942688,
      "learning_rate": 0.0003388186855063844,
      "loss": 2.022,
      "step": 105685
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.9127877354621887,
      "learning_rate": 0.0003388146293297617,
      "loss": 2.1871,
      "step": 105686
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.9402355551719666,
      "learning_rate": 0.00033881057314592253,
      "loss": 1.9561,
      "step": 105687
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.9993191361427307,
      "learning_rate": 0.00033880651695486784,
      "loss": 2.2194,
      "step": 105688
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.89915531873703,
      "learning_rate": 0.0003388024607565983,
      "loss": 2.1284,
      "step": 105689
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.8779665231704712,
      "learning_rate": 0.0003387984045511146,
      "loss": 2.3168,
      "step": 105690
    },
    {
      "epoch": 1.38,
      "grad_norm": 1.005477786064148,
      "learning_rate": 0.0003387943483384175,
      "loss": 2.2082,
      "step": 105691
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.8915992379188538,
      "learning_rate": 0.00033879029211850793,
      "loss": 2.0692,
      "step": 105692
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.8952187895774841,
      "learning_rate": 0.00033878623589138644,
      "loss": 2.1948,
      "step": 105693
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.8830829858779907,
      "learning_rate": 0.0003387821796570538,
      "loss": 2.0805,
      "step": 105694
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.8659319877624512,
      "learning_rate": 0.0003387781234155109,
      "loss": 2.0678,
      "step": 105695
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.860784649848938,
      "learning_rate": 0.00033877406716675834,
      "loss": 2.1754,
      "step": 105696
    },
    {
      "epoch": 1.38,
      "grad_norm": 1.0909442901611328,
      "learning_rate": 0.000338770010910797,
      "loss": 2.4689,
      "step": 105697
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.8775157332420349,
      "learning_rate": 0.0003387659546476276,
      "loss": 1.9606,
      "step": 105698
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.9351593255996704,
      "learning_rate": 0.0003387618983772508,
      "loss": 2.1719,
      "step": 105699
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.9988588094711304,
      "learning_rate": 0.0003387578420996675,
      "loss": 2.1916,
      "step": 105700
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.9489551782608032,
      "learning_rate": 0.0003387537858148784,
      "loss": 2.0067,
      "step": 105701
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.8546392321586609,
      "learning_rate": 0.0003387497295228843,
      "loss": 2.0955,
      "step": 105702
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.9662507772445679,
      "learning_rate": 0.0003387456732236857,
      "loss": 2.0352,
      "step": 105703
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.811211884021759,
      "learning_rate": 0.0003387416169172837,
      "loss": 2.0273,
      "step": 105704
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.8613198399543762,
      "learning_rate": 0.0003387375606036789,
      "loss": 1.9784,
      "step": 105705
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.9470968842506409,
      "learning_rate": 0.00033873350428287194,
      "loss": 2.2174,
      "step": 105706
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.8698247671127319,
      "learning_rate": 0.0003387294479548638,
      "loss": 2.214,
      "step": 105707
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.9226539134979248,
      "learning_rate": 0.0003387253916196551,
      "loss": 2.1522,
      "step": 105708
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.9471014738082886,
      "learning_rate": 0.0003387213352772466,
      "loss": 1.9793,
      "step": 105709
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.8560740351676941,
      "learning_rate": 0.00033871727892763906,
      "loss": 2.1148,
      "step": 105710
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.9026809930801392,
      "learning_rate": 0.00033871322257083327,
      "loss": 2.342,
      "step": 105711
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.872148334980011,
      "learning_rate": 0.00033870916620682995,
      "loss": 1.972,
      "step": 105712
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.9522204399108887,
      "learning_rate": 0.0003387051098356299,
      "loss": 2.4084,
      "step": 105713
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.8393787741661072,
      "learning_rate": 0.0003387010534572338,
      "loss": 2.0519,
      "step": 105714
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.9140699505805969,
      "learning_rate": 0.00033869699707164247,
      "loss": 2.103,
      "step": 105715
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.8612747192382812,
      "learning_rate": 0.0003386929406788566,
      "loss": 2.0742,
      "step": 105716
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.8819273114204407,
      "learning_rate": 0.00033868888427887703,
      "loss": 2.146,
      "step": 105717
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.824346125125885,
      "learning_rate": 0.0003386848278717044,
      "loss": 2.0756,
      "step": 105718
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.902981698513031,
      "learning_rate": 0.0003386807714573396,
      "loss": 2.1632,
      "step": 105719
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.9309509992599487,
      "learning_rate": 0.00033867671503578334,
      "loss": 2.2448,
      "step": 105720
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.8701977133750916,
      "learning_rate": 0.00033867265860703624,
      "loss": 2.1621,
      "step": 105721
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.8400290608406067,
      "learning_rate": 0.0003386686021710992,
      "loss": 2.1264,
      "step": 105722
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.890097439289093,
      "learning_rate": 0.0003386645457279729,
      "loss": 2.1125,
      "step": 105723
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.8759548664093018,
      "learning_rate": 0.0003386604892776582,
      "loss": 2.1877,
      "step": 105724
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.8934414386749268,
      "learning_rate": 0.0003386564328201558,
      "loss": 2.1889,
      "step": 105725
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.8813864588737488,
      "learning_rate": 0.0003386523763554664,
      "loss": 2.2526,
      "step": 105726
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.9095334410667419,
      "learning_rate": 0.0003386483198835908,
      "loss": 2.2683,
      "step": 105727
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.8726522922515869,
      "learning_rate": 0.0003386442634045298,
      "loss": 2.2498,
      "step": 105728
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.8142856359481812,
      "learning_rate": 0.000338640206918284,
      "loss": 2.0303,
      "step": 105729
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.8275474309921265,
      "learning_rate": 0.00033863615042485434,
      "loss": 2.3128,
      "step": 105730
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.8639844059944153,
      "learning_rate": 0.00033863209392424144,
      "loss": 2.0997,
      "step": 105731
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.9189818501472473,
      "learning_rate": 0.00033862803741644613,
      "loss": 2.1968,
      "step": 105732
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.9437538385391235,
      "learning_rate": 0.0003386239809014691,
      "loss": 2.1403,
      "step": 105733
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.943178653717041,
      "learning_rate": 0.0003386199243793111,
      "loss": 2.1791,
      "step": 105734
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.8475208878517151,
      "learning_rate": 0.0003386158678499731,
      "loss": 2.0783,
      "step": 105735
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.9886991381645203,
      "learning_rate": 0.0003386118113134556,
      "loss": 2.2168,
      "step": 105736
    },
    {
      "epoch": 1.38,
      "grad_norm": 1.1017887592315674,
      "learning_rate": 0.0003386077547697594,
      "loss": 2.1279,
      "step": 105737
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.8015367388725281,
      "learning_rate": 0.0003386036982188853,
      "loss": 2.1105,
      "step": 105738
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.8993395566940308,
      "learning_rate": 0.00033859964166083403,
      "loss": 2.1213,
      "step": 105739
    },
    {
      "epoch": 1.38,
      "grad_norm": 1.0397586822509766,
      "learning_rate": 0.00033859558509560637,
      "loss": 2.0142,
      "step": 105740
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.9672624468803406,
      "learning_rate": 0.00033859152852320313,
      "loss": 2.0706,
      "step": 105741
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.8541112542152405,
      "learning_rate": 0.00033858747194362497,
      "loss": 1.9339,
      "step": 105742
    },
    {
      "epoch": 1.38,
      "grad_norm": 1.0352367162704468,
      "learning_rate": 0.0003385834153568726,
      "loss": 2.0893,
      "step": 105743
    },
    {
      "epoch": 1.38,
      "grad_norm": 1.0022947788238525,
      "learning_rate": 0.0003385793587629469,
      "loss": 1.9641,
      "step": 105744
    },
    {
      "epoch": 1.38,
      "grad_norm": 1.0995513200759888,
      "learning_rate": 0.0003385753021618486,
      "loss": 2.1352,
      "step": 105745
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.917732298374176,
      "learning_rate": 0.0003385712455535783,
      "loss": 2.1268,
      "step": 105746
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.9459049701690674,
      "learning_rate": 0.000338567188938137,
      "loss": 2.0856,
      "step": 105747
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.9864811301231384,
      "learning_rate": 0.0003385631323155253,
      "loss": 2.143,
      "step": 105748
    },
    {
      "epoch": 1.38,
      "grad_norm": 1.0296870470046997,
      "learning_rate": 0.0003385590756857439,
      "loss": 2.0833,
      "step": 105749
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.9222937822341919,
      "learning_rate": 0.0003385550190487938,
      "loss": 2.2857,
      "step": 105750
    },
    {
      "epoch": 1.38,
      "grad_norm": 1.0787014961242676,
      "learning_rate": 0.0003385509624046755,
      "loss": 2.1691,
      "step": 105751
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.9485759139060974,
      "learning_rate": 0.0003385469057533898,
      "loss": 2.2116,
      "step": 105752
    },
    {
      "epoch": 1.38,
      "grad_norm": 1.1802295446395874,
      "learning_rate": 0.0003385428490949376,
      "loss": 2.0716,
      "step": 105753
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.817876398563385,
      "learning_rate": 0.00033853879242931947,
      "loss": 2.2283,
      "step": 105754
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.8623647093772888,
      "learning_rate": 0.0003385347357565363,
      "loss": 2.1524,
      "step": 105755
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.8644736409187317,
      "learning_rate": 0.0003385306790765888,
      "loss": 2.2842,
      "step": 105756
    },
    {
      "epoch": 1.38,
      "grad_norm": 1.0277687311172485,
      "learning_rate": 0.0003385266223894777,
      "loss": 2.1393,
      "step": 105757
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.8501251339912415,
      "learning_rate": 0.00033852256569520374,
      "loss": 2.1608,
      "step": 105758
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.8686923980712891,
      "learning_rate": 0.00033851850899376787,
      "loss": 2.0879,
      "step": 105759
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.8814866542816162,
      "learning_rate": 0.0003385144522851705,
      "loss": 2.3697,
      "step": 105760
    },
    {
      "epoch": 1.38,
      "grad_norm": 1.036534309387207,
      "learning_rate": 0.0003385103955694126,
      "loss": 2.0998,
      "step": 105761
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.9668492674827576,
      "learning_rate": 0.00033850633884649494,
      "loss": 1.9967,
      "step": 105762
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.935117781162262,
      "learning_rate": 0.00033850228211641825,
      "loss": 2.3082,
      "step": 105763
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.9141243100166321,
      "learning_rate": 0.00033849822537918315,
      "loss": 2.3048,
      "step": 105764
    },
    {
      "epoch": 1.38,
      "grad_norm": 1.1233625411987305,
      "learning_rate": 0.0003384941686347906,
      "loss": 2.1554,
      "step": 105765
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.8915858268737793,
      "learning_rate": 0.00033849011188324127,
      "loss": 2.1543,
      "step": 105766
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.8717681765556335,
      "learning_rate": 0.0003384860551245358,
      "loss": 2.3544,
      "step": 105767
    },
    {
      "epoch": 1.38,
      "grad_norm": 1.1002970933914185,
      "learning_rate": 0.00033848199835867515,
      "loss": 2.3125,
      "step": 105768
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.969258189201355,
      "learning_rate": 0.00033847794158565987,
      "loss": 2.2159,
      "step": 105769
    },
    {
      "epoch": 1.38,
      "grad_norm": 1.0602478981018066,
      "learning_rate": 0.0003384738848054909,
      "loss": 2.0935,
      "step": 105770
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.9113838076591492,
      "learning_rate": 0.0003384698280181689,
      "loss": 2.1856,
      "step": 105771
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.8853234648704529,
      "learning_rate": 0.0003384657712236946,
      "loss": 2.0341,
      "step": 105772
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.9032059907913208,
      "learning_rate": 0.0003384617144220688,
      "loss": 2.1921,
      "step": 105773
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.8171226978302002,
      "learning_rate": 0.0003384576576132923,
      "loss": 2.1915,
      "step": 105774
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.8854013085365295,
      "learning_rate": 0.0003384536007973657,
      "loss": 2.1312,
      "step": 105775
    },
    {
      "epoch": 1.38,
      "grad_norm": 1.000515341758728,
      "learning_rate": 0.0003384495439742898,
      "loss": 1.9876,
      "step": 105776
    },
    {
      "epoch": 1.38,
      "grad_norm": 1.191245198249817,
      "learning_rate": 0.0003384454871440655,
      "loss": 2.0795,
      "step": 105777
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.9911051392555237,
      "learning_rate": 0.00033844143030669347,
      "loss": 2.0847,
      "step": 105778
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.9292831420898438,
      "learning_rate": 0.00033843737346217435,
      "loss": 2.2675,
      "step": 105779
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.9402467608451843,
      "learning_rate": 0.00033843331661050915,
      "loss": 2.041,
      "step": 105780
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.9340152144432068,
      "learning_rate": 0.0003384292597516984,
      "loss": 2.2324,
      "step": 105781
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.9015084505081177,
      "learning_rate": 0.00033842520288574293,
      "loss": 1.883,
      "step": 105782
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.9415680170059204,
      "learning_rate": 0.00033842114601264343,
      "loss": 2.3189,
      "step": 105783
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.9175294637680054,
      "learning_rate": 0.00033841708913240077,
      "loss": 2.2068,
      "step": 105784
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.9163174033164978,
      "learning_rate": 0.00033841303224501566,
      "loss": 2.413,
      "step": 105785
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.8610701560974121,
      "learning_rate": 0.00033840897535048885,
      "loss": 2.2195,
      "step": 105786
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.8300681114196777,
      "learning_rate": 0.000338404918448821,
      "loss": 1.9031,
      "step": 105787
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.9405106902122498,
      "learning_rate": 0.000338400861540013,
      "loss": 2.1196,
      "step": 105788
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.8522809147834778,
      "learning_rate": 0.0003383968046240656,
      "loss": 2.1606,
      "step": 105789
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.8818759322166443,
      "learning_rate": 0.0003383927477009795,
      "loss": 2.0593,
      "step": 105790
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.9552816152572632,
      "learning_rate": 0.0003383886907707554,
      "loss": 2.2158,
      "step": 105791
    },
    {
      "epoch": 1.38,
      "grad_norm": 1.1884180307388306,
      "learning_rate": 0.0003383846338333942,
      "loss": 2.3947,
      "step": 105792
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.9625298380851746,
      "learning_rate": 0.0003383805768888965,
      "loss": 2.3209,
      "step": 105793
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.8147479295730591,
      "learning_rate": 0.00033837651993726315,
      "loss": 1.877,
      "step": 105794
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.8238709568977356,
      "learning_rate": 0.0003383724629784948,
      "loss": 1.9805,
      "step": 105795
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.8703508377075195,
      "learning_rate": 0.0003383684060125925,
      "loss": 2.1482,
      "step": 105796
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.9011709094047546,
      "learning_rate": 0.0003383643490395566,
      "loss": 1.9407,
      "step": 105797
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.926138162612915,
      "learning_rate": 0.0003383602920593881,
      "loss": 2.2083,
      "step": 105798
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.9655241966247559,
      "learning_rate": 0.0003383562350720877,
      "loss": 2.1604,
      "step": 105799
    },
    {
      "epoch": 1.38,
      "grad_norm": 1.0731706619262695,
      "learning_rate": 0.0003383521780776562,
      "loss": 2.036,
      "step": 105800
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.9819887280464172,
      "learning_rate": 0.0003383481210760942,
      "loss": 2.0645,
      "step": 105801
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.9471184611320496,
      "learning_rate": 0.00033834406406740267,
      "loss": 2.2511,
      "step": 105802
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.9069100022315979,
      "learning_rate": 0.0003383400070515822,
      "loss": 2.2186,
      "step": 105803
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.916635274887085,
      "learning_rate": 0.00033833595002863355,
      "loss": 1.9596,
      "step": 105804
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.8772612810134888,
      "learning_rate": 0.00033833189299855757,
      "loss": 2.1481,
      "step": 105805
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.8564366698265076,
      "learning_rate": 0.000338327835961355,
      "loss": 2.2945,
      "step": 105806
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.8501756191253662,
      "learning_rate": 0.0003383237789170265,
      "loss": 2.2193,
      "step": 105807
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.9312950968742371,
      "learning_rate": 0.0003383197218655729,
      "loss": 2.2364,
      "step": 105808
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.9563605785369873,
      "learning_rate": 0.00033831566480699505,
      "loss": 2.0718,
      "step": 105809
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.9148920774459839,
      "learning_rate": 0.00033831160774129343,
      "loss": 2.088,
      "step": 105810
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.8621289134025574,
      "learning_rate": 0.00033830755066846903,
      "loss": 2.1816,
      "step": 105811
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.8270887732505798,
      "learning_rate": 0.00033830349358852257,
      "loss": 2.0674,
      "step": 105812
    },
    {
      "epoch": 1.38,
      "grad_norm": 1.0309224128723145,
      "learning_rate": 0.00033829943650145474,
      "loss": 1.908,
      "step": 105813
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.852658748626709,
      "learning_rate": 0.00033829537940726625,
      "loss": 2.1203,
      "step": 105814
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.8387948870658875,
      "learning_rate": 0.000338291322305958,
      "loss": 2.1681,
      "step": 105815
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.8984763026237488,
      "learning_rate": 0.0003382872651975307,
      "loss": 2.0305,
      "step": 105816
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.8749175667762756,
      "learning_rate": 0.000338283208081985,
      "loss": 2.0943,
      "step": 105817
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.8727245926856995,
      "learning_rate": 0.0003382791509593218,
      "loss": 2.4287,
      "step": 105818
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.839092493057251,
      "learning_rate": 0.0003382750938295417,
      "loss": 2.1344,
      "step": 105819
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.9028778672218323,
      "learning_rate": 0.0003382710366926456,
      "loss": 1.9289,
      "step": 105820
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.8800114989280701,
      "learning_rate": 0.0003382669795486342,
      "loss": 2.3128,
      "step": 105821
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.8781729340553284,
      "learning_rate": 0.0003382629223975082,
      "loss": 2.0538,
      "step": 105822
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.8412539958953857,
      "learning_rate": 0.00033825886523926844,
      "loss": 2.2757,
      "step": 105823
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.8071202039718628,
      "learning_rate": 0.00033825480807391563,
      "loss": 1.9336,
      "step": 105824
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.9401443600654602,
      "learning_rate": 0.0003382507509014505,
      "loss": 2.1978,
      "step": 105825
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.9712238907814026,
      "learning_rate": 0.0003382466937218739,
      "loss": 2.321,
      "step": 105826
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.8942240476608276,
      "learning_rate": 0.0003382426365351865,
      "loss": 2.2186,
      "step": 105827
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.8204828500747681,
      "learning_rate": 0.0003382385793413889,
      "loss": 2.1445,
      "step": 105828
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.9527459144592285,
      "learning_rate": 0.0003382345221404823,
      "loss": 2.3955,
      "step": 105829
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.9778416752815247,
      "learning_rate": 0.000338230464932467,
      "loss": 2.1414,
      "step": 105830
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.8600473999977112,
      "learning_rate": 0.00033822640771734395,
      "loss": 2.0029,
      "step": 105831
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.8510386347770691,
      "learning_rate": 0.00033822235049511395,
      "loss": 2.2571,
      "step": 105832
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.9445803165435791,
      "learning_rate": 0.00033821829326577765,
      "loss": 2.126,
      "step": 105833
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.8888631463050842,
      "learning_rate": 0.00033821423602933587,
      "loss": 2.2601,
      "step": 105834
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.7915300130844116,
      "learning_rate": 0.0003382101787857894,
      "loss": 2.2797,
      "step": 105835
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.8978673815727234,
      "learning_rate": 0.0003382061215351388,
      "loss": 2.2023,
      "step": 105836
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.9557475447654724,
      "learning_rate": 0.0003382020642773851,
      "loss": 2.196,
      "step": 105837
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.8414986729621887,
      "learning_rate": 0.00033819800701252885,
      "loss": 2.0234,
      "step": 105838
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.794571042060852,
      "learning_rate": 0.00033819394974057083,
      "loss": 2.0619,
      "step": 105839
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.8663721680641174,
      "learning_rate": 0.00033818989246151183,
      "loss": 2.094,
      "step": 105840
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.9970552921295166,
      "learning_rate": 0.0003381858351753527,
      "loss": 2.1761,
      "step": 105841
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.9058713316917419,
      "learning_rate": 0.00033818177788209403,
      "loss": 2.3075,
      "step": 105842
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.9543907046318054,
      "learning_rate": 0.0003381777205817367,
      "loss": 1.9733,
      "step": 105843
    },
    {
      "epoch": 1.38,
      "grad_norm": 1.0102033615112305,
      "learning_rate": 0.0003381736632742814,
      "loss": 2.1042,
      "step": 105844
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.987563967704773,
      "learning_rate": 0.0003381696059597289,
      "loss": 2.4204,
      "step": 105845
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.9471909999847412,
      "learning_rate": 0.0003381655486380798,
      "loss": 2.2926,
      "step": 105846
    },
    {
      "epoch": 1.38,
      "grad_norm": 1.2562675476074219,
      "learning_rate": 0.00033816149130933525,
      "loss": 2.228,
      "step": 105847
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.9466981887817383,
      "learning_rate": 0.0003381574339734956,
      "loss": 2.0797,
      "step": 105848
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.8475826978683472,
      "learning_rate": 0.0003381533766305618,
      "loss": 2.132,
      "step": 105849
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.9468978047370911,
      "learning_rate": 0.00033814931928053456,
      "loss": 1.9591,
      "step": 105850
    },
    {
      "epoch": 1.38,
      "grad_norm": 1.1034479141235352,
      "learning_rate": 0.0003381452619234147,
      "loss": 2.0828,
      "step": 105851
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.8816711902618408,
      "learning_rate": 0.00033814120455920283,
      "loss": 2.0893,
      "step": 105852
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.8934699296951294,
      "learning_rate": 0.0003381371471878999,
      "loss": 2.4525,
      "step": 105853
    },
    {
      "epoch": 1.38,
      "grad_norm": 1.1223605871200562,
      "learning_rate": 0.0003381330898095064,
      "loss": 2.0865,
      "step": 105854
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.9906130433082581,
      "learning_rate": 0.0003381290324240233,
      "loss": 2.2325,
      "step": 105855
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.9043527841567993,
      "learning_rate": 0.0003381249750314514,
      "loss": 2.2145,
      "step": 105856
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.9689680933952332,
      "learning_rate": 0.0003381209176317912,
      "loss": 2.0558,
      "step": 105857
    },
    {
      "epoch": 1.38,
      "grad_norm": 1.1493468284606934,
      "learning_rate": 0.0003381168602250436,
      "loss": 1.9363,
      "step": 105858
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.8680739402770996,
      "learning_rate": 0.0003381128028112095,
      "loss": 2.0203,
      "step": 105859
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.8542172908782959,
      "learning_rate": 0.00033810874539028936,
      "loss": 2.0145,
      "step": 105860
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.8803815245628357,
      "learning_rate": 0.0003381046879622842,
      "loss": 2.1218,
      "step": 105861
    },
    {
      "epoch": 1.38,
      "grad_norm": 1.8276907205581665,
      "learning_rate": 0.00033810063052719456,
      "loss": 2.202,
      "step": 105862
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.8501574397087097,
      "learning_rate": 0.00033809657308502136,
      "loss": 2.4867,
      "step": 105863
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.9945268034934998,
      "learning_rate": 0.0003380925156357653,
      "loss": 2.0336,
      "step": 105864
    },
    {
      "epoch": 1.38,
      "grad_norm": 1.1100150346755981,
      "learning_rate": 0.0003380884581794271,
      "loss": 1.9736,
      "step": 105865
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.82737135887146,
      "learning_rate": 0.0003380844007160076,
      "loss": 1.9411,
      "step": 105866
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.8171452283859253,
      "learning_rate": 0.0003380803432455073,
      "loss": 2.0885,
      "step": 105867
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.8318457007408142,
      "learning_rate": 0.00033807628576792726,
      "loss": 2.0505,
      "step": 105868
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.8466634154319763,
      "learning_rate": 0.00033807222828326824,
      "loss": 1.9839,
      "step": 105869
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.8711932301521301,
      "learning_rate": 0.00033806817079153073,
      "loss": 2.1456,
      "step": 105870
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.9061662554740906,
      "learning_rate": 0.0003380641132927156,
      "loss": 2.404,
      "step": 105871
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.9063293933868408,
      "learning_rate": 0.00033806005578682373,
      "loss": 2.1256,
      "step": 105872
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.9223964214324951,
      "learning_rate": 0.0003380559982738557,
      "loss": 2.0715,
      "step": 105873
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.8725878596305847,
      "learning_rate": 0.00033805194075381244,
      "loss": 1.9944,
      "step": 105874
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.8644914627075195,
      "learning_rate": 0.00033804788322669454,
      "loss": 2.2116,
      "step": 105875
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.8719805479049683,
      "learning_rate": 0.0003380438256925028,
      "loss": 2.0902,
      "step": 105876
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.8331231474876404,
      "learning_rate": 0.00033803976815123803,
      "loss": 1.9541,
      "step": 105877
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.8467023372650146,
      "learning_rate": 0.000338035710602901,
      "loss": 2.2474,
      "step": 105878
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.9861302971839905,
      "learning_rate": 0.00033803165304749236,
      "loss": 2.447,
      "step": 105879
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.8545295000076294,
      "learning_rate": 0.0003380275954850129,
      "loss": 2.0849,
      "step": 105880
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.8346349000930786,
      "learning_rate": 0.00033802353791546343,
      "loss": 2.4968,
      "step": 105881
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.9003905653953552,
      "learning_rate": 0.0003380194803388446,
      "loss": 2.3533,
      "step": 105882
    },
    {
      "epoch": 1.38,
      "grad_norm": 1.0065141916275024,
      "learning_rate": 0.0003380154227551572,
      "loss": 2.154,
      "step": 105883
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.9334018230438232,
      "learning_rate": 0.0003380113651644022,
      "loss": 2.212,
      "step": 105884
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.8840409517288208,
      "learning_rate": 0.00033800730756658004,
      "loss": 2.1865,
      "step": 105885
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.8329263925552368,
      "learning_rate": 0.0003380032499616916,
      "loss": 2.0499,
      "step": 105886
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.9912413358688354,
      "learning_rate": 0.0003379991923497377,
      "loss": 2.093,
      "step": 105887
    },
    {
      "epoch": 1.38,
      "grad_norm": 1.1426652669906616,
      "learning_rate": 0.000337995134730719,
      "loss": 2.3559,
      "step": 105888
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.9755942821502686,
      "learning_rate": 0.00033799107710463626,
      "loss": 2.2881,
      "step": 105889
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.8635478019714355,
      "learning_rate": 0.0003379870194714903,
      "loss": 2.1967,
      "step": 105890
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.8792474865913391,
      "learning_rate": 0.0003379829618312818,
      "loss": 2.032,
      "step": 105891
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.8502762317657471,
      "learning_rate": 0.0003379789041840116,
      "loss": 2.0705,
      "step": 105892
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.8851622343063354,
      "learning_rate": 0.00033797484652968034,
      "loss": 2.1753,
      "step": 105893
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.8975895643234253,
      "learning_rate": 0.0003379707888682889,
      "loss": 2.0791,
      "step": 105894
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.8627769351005554,
      "learning_rate": 0.0003379667311998379,
      "loss": 2.1655,
      "step": 105895
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.9179853796958923,
      "learning_rate": 0.0003379626735243282,
      "loss": 2.3031,
      "step": 105896
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.9853943586349487,
      "learning_rate": 0.0003379586158417606,
      "loss": 1.945,
      "step": 105897
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.956494152545929,
      "learning_rate": 0.00033795455815213563,
      "loss": 2.0999,
      "step": 105898
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.8272284865379333,
      "learning_rate": 0.00033795050045545426,
      "loss": 2.0066,
      "step": 105899
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.8907662034034729,
      "learning_rate": 0.00033794644275171717,
      "loss": 2.0402,
      "step": 105900
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.8861128091812134,
      "learning_rate": 0.00033794238504092523,
      "loss": 1.9493,
      "step": 105901
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.8670912384986877,
      "learning_rate": 0.00033793832732307894,
      "loss": 1.9715,
      "step": 105902
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.9423537254333496,
      "learning_rate": 0.00033793426959817925,
      "loss": 2.1766,
      "step": 105903
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.8875874876976013,
      "learning_rate": 0.00033793021186622694,
      "loss": 2.2359,
      "step": 105904
    },
    {
      "epoch": 1.38,
      "grad_norm": 1.1345876455307007,
      "learning_rate": 0.00033792615412722254,
      "loss": 1.9554,
      "step": 105905
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.8008459210395813,
      "learning_rate": 0.00033792209638116704,
      "loss": 1.9907,
      "step": 105906
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.9575395584106445,
      "learning_rate": 0.0003379180386280611,
      "loss": 2.2082,
      "step": 105907
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.9274693131446838,
      "learning_rate": 0.0003379139808679055,
      "loss": 2.0881,
      "step": 105908
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.8716181516647339,
      "learning_rate": 0.0003379099231007009,
      "loss": 2.1453,
      "step": 105909
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.8231106996536255,
      "learning_rate": 0.0003379058653264482,
      "loss": 2.0713,
      "step": 105910
    },
    {
      "epoch": 1.38,
      "grad_norm": 1.0636738538742065,
      "learning_rate": 0.0003379018075451481,
      "loss": 2.1216,
      "step": 105911
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.8659628629684448,
      "learning_rate": 0.0003378977497568013,
      "loss": 2.3407,
      "step": 105912
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.9513229727745056,
      "learning_rate": 0.0003378936919614085,
      "loss": 1.9232,
      "step": 105913
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.9509941339492798,
      "learning_rate": 0.00033788963415897074,
      "loss": 2.2551,
      "step": 105914
    },
    {
      "epoch": 1.38,
      "grad_norm": 1.004467248916626,
      "learning_rate": 0.00033788557634948843,
      "loss": 1.9136,
      "step": 105915
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.8600372076034546,
      "learning_rate": 0.0003378815185329625,
      "loss": 2.0714,
      "step": 105916
    },
    {
      "epoch": 1.38,
      "grad_norm": 1.0314850807189941,
      "learning_rate": 0.00033787746070939375,
      "loss": 2.1694,
      "step": 105917
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.8299143314361572,
      "learning_rate": 0.00033787340287878283,
      "loss": 2.0749,
      "step": 105918
    },
    {
      "epoch": 1.38,
      "grad_norm": 1.0234861373901367,
      "learning_rate": 0.0003378693450411305,
      "loss": 2.0928,
      "step": 105919
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.8910098671913147,
      "learning_rate": 0.0003378652871964376,
      "loss": 2.1812,
      "step": 105920
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.9789823889732361,
      "learning_rate": 0.00033786122934470475,
      "loss": 1.9231,
      "step": 105921
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.9883298277854919,
      "learning_rate": 0.0003378571714859328,
      "loss": 2.3286,
      "step": 105922
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.9865829944610596,
      "learning_rate": 0.0003378531136201226,
      "loss": 2.1902,
      "step": 105923
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.9086606502532959,
      "learning_rate": 0.0003378490557472747,
      "loss": 2.0493,
      "step": 105924
    },
    {
      "epoch": 1.38,
      "grad_norm": 1.0170495510101318,
      "learning_rate": 0.00033784499786738997,
      "loss": 2.1651,
      "step": 105925
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.930635392665863,
      "learning_rate": 0.0003378409399804691,
      "loss": 2.2442,
      "step": 105926
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.9806348085403442,
      "learning_rate": 0.00033783688208651293,
      "loss": 2.0827,
      "step": 105927
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.9763756990432739,
      "learning_rate": 0.0003378328241855221,
      "loss": 2.2117,
      "step": 105928
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.9338486194610596,
      "learning_rate": 0.0003378287662774975,
      "loss": 2.1176,
      "step": 105929
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.8366190791130066,
      "learning_rate": 0.00033782470836243986,
      "loss": 2.0189,
      "step": 105930
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.9011461734771729,
      "learning_rate": 0.00033782065044034976,
      "loss": 2.2465,
      "step": 105931
    },
    {
      "epoch": 1.38,
      "grad_norm": 1.1329078674316406,
      "learning_rate": 0.0003378165925112282,
      "loss": 2.0584,
      "step": 105932
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.9446346759796143,
      "learning_rate": 0.0003378125345750758,
      "loss": 2.2304,
      "step": 105933
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.8621842861175537,
      "learning_rate": 0.0003378084766318933,
      "loss": 2.1815,
      "step": 105934
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.8589674234390259,
      "learning_rate": 0.0003378044186816815,
      "loss": 2.3834,
      "step": 105935
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.8396742343902588,
      "learning_rate": 0.0003378003607244412,
      "loss": 2.1266,
      "step": 105936
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.8187181353569031,
      "learning_rate": 0.00033779630276017307,
      "loss": 1.8514,
      "step": 105937
    },
    {
      "epoch": 1.38,
      "grad_norm": 1.0795912742614746,
      "learning_rate": 0.00033779224478887776,
      "loss": 2.02,
      "step": 105938
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.8920131325721741,
      "learning_rate": 0.00033778818681055633,
      "loss": 2.2105,
      "step": 105939
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.8715544939041138,
      "learning_rate": 0.00033778412882520933,
      "loss": 2.1054,
      "step": 105940
    },
    {
      "epoch": 1.38,
      "grad_norm": 1.0298843383789062,
      "learning_rate": 0.0003377800708328375,
      "loss": 2.4436,
      "step": 105941
    },
    {
      "epoch": 1.38,
      "grad_norm": 1.1385440826416016,
      "learning_rate": 0.0003377760128334417,
      "loss": 2.2018,
      "step": 105942
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.8758759498596191,
      "learning_rate": 0.00033777195482702257,
      "loss": 2.1883,
      "step": 105943
    },
    {
      "epoch": 1.38,
      "grad_norm": 1.0038896799087524,
      "learning_rate": 0.00033776789681358096,
      "loss": 2.1261,
      "step": 105944
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.9948481321334839,
      "learning_rate": 0.0003377638387931176,
      "loss": 2.1117,
      "step": 105945
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.9473931193351746,
      "learning_rate": 0.00033775978076563317,
      "loss": 2.2467,
      "step": 105946
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.843906581401825,
      "learning_rate": 0.00033775572273112855,
      "loss": 2.3389,
      "step": 105947
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.9524312615394592,
      "learning_rate": 0.00033775166468960437,
      "loss": 2.0727,
      "step": 105948
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.8877089023590088,
      "learning_rate": 0.00033774760664106144,
      "loss": 2.2572,
      "step": 105949
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.8731799125671387,
      "learning_rate": 0.00033774354858550053,
      "loss": 2.1415,
      "step": 105950
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.801118016242981,
      "learning_rate": 0.0003377394905229225,
      "loss": 2.0522,
      "step": 105951
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.8578583598136902,
      "learning_rate": 0.0003377354324533278,
      "loss": 2.1552,
      "step": 105952
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.9139521718025208,
      "learning_rate": 0.0003377313743767174,
      "loss": 2.1398,
      "step": 105953
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.8791734576225281,
      "learning_rate": 0.00033772731629309215,
      "loss": 2.2484,
      "step": 105954
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.8627509474754333,
      "learning_rate": 0.0003377232582024526,
      "loss": 1.8888,
      "step": 105955
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.8461677432060242,
      "learning_rate": 0.0003377192001047996,
      "loss": 2.1967,
      "step": 105956
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.8577977418899536,
      "learning_rate": 0.0003377151420001339,
      "loss": 2.3767,
      "step": 105957
    },
    {
      "epoch": 1.38,
      "grad_norm": 1.0318968296051025,
      "learning_rate": 0.0003377110838884562,
      "loss": 2.1493,
      "step": 105958
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.8509214520454407,
      "learning_rate": 0.0003377070257697673,
      "loss": 2.2659,
      "step": 105959
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.9250643253326416,
      "learning_rate": 0.000337702967644068,
      "loss": 2.2934,
      "step": 105960
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.909921407699585,
      "learning_rate": 0.00033769890951135896,
      "loss": 2.1854,
      "step": 105961
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.935527503490448,
      "learning_rate": 0.000337694851371641,
      "loss": 2.3529,
      "step": 105962
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.8458881974220276,
      "learning_rate": 0.0003376907932249149,
      "loss": 2.2931,
      "step": 105963
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.8500065803527832,
      "learning_rate": 0.0003376867350711813,
      "loss": 1.9727,
      "step": 105964
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.8713389039039612,
      "learning_rate": 0.000337682676910441,
      "loss": 2.2991,
      "step": 105965
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.8534385561943054,
      "learning_rate": 0.0003376786187426949,
      "loss": 2.1259,
      "step": 105966
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.9387760162353516,
      "learning_rate": 0.00033767456056794356,
      "loss": 2.1338,
      "step": 105967
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.8916864395141602,
      "learning_rate": 0.0003376705023861878,
      "loss": 2.1704,
      "step": 105968
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.9085649847984314,
      "learning_rate": 0.00033766644419742844,
      "loss": 2.0432,
      "step": 105969
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.8388660550117493,
      "learning_rate": 0.0003376623860016661,
      "loss": 2.1473,
      "step": 105970
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.9015154838562012,
      "learning_rate": 0.00033765832779890164,
      "loss": 2.0732,
      "step": 105971
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.9122612476348877,
      "learning_rate": 0.0003376542695891358,
      "loss": 2.2813,
      "step": 105972
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.8721194267272949,
      "learning_rate": 0.0003376502113723693,
      "loss": 2.4387,
      "step": 105973
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.8620423674583435,
      "learning_rate": 0.00033764615314860283,
      "loss": 2.1005,
      "step": 105974
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.8783677220344543,
      "learning_rate": 0.0003376420949178374,
      "loss": 2.3576,
      "step": 105975
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.8962301015853882,
      "learning_rate": 0.0003376380366800735,
      "loss": 2.2406,
      "step": 105976
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.9150483012199402,
      "learning_rate": 0.000337633978435312,
      "loss": 2.1408,
      "step": 105977
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.8643935322761536,
      "learning_rate": 0.00033762992018355364,
      "loss": 1.9,
      "step": 105978
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.852675199508667,
      "learning_rate": 0.0003376258619247992,
      "loss": 2.2002,
      "step": 105979
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.9005244970321655,
      "learning_rate": 0.0003376218036590493,
      "loss": 2.1773,
      "step": 105980
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.8772096633911133,
      "learning_rate": 0.00033761774538630483,
      "loss": 2.2339,
      "step": 105981
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.9892892241477966,
      "learning_rate": 0.00033761368710656664,
      "loss": 2.0814,
      "step": 105982
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.9214094281196594,
      "learning_rate": 0.0003376096288198352,
      "loss": 2.1171,
      "step": 105983
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.9555317163467407,
      "learning_rate": 0.0003376055705261114,
      "loss": 2.0712,
      "step": 105984
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.8679895401000977,
      "learning_rate": 0.00033760151222539614,
      "loss": 2.1639,
      "step": 105985
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.9022120833396912,
      "learning_rate": 0.00033759745391769,
      "loss": 2.2597,
      "step": 105986
    },
    {
      "epoch": 1.38,
      "grad_norm": 1.005357027053833,
      "learning_rate": 0.00033759339560299374,
      "loss": 2.0841,
      "step": 105987
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.904681921005249,
      "learning_rate": 0.00033758933728130814,
      "loss": 2.1067,
      "step": 105988
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.9230065941810608,
      "learning_rate": 0.0003375852789526341,
      "loss": 2.4679,
      "step": 105989
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.9295792579650879,
      "learning_rate": 0.00033758122061697217,
      "loss": 2.2556,
      "step": 105990
    },
    {
      "epoch": 1.38,
      "grad_norm": 1.115749716758728,
      "learning_rate": 0.0003375771622743232,
      "loss": 2.094,
      "step": 105991
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.8416808247566223,
      "learning_rate": 0.0003375731039246879,
      "loss": 2.1958,
      "step": 105992
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.9141592979431152,
      "learning_rate": 0.0003375690455680671,
      "loss": 2.147,
      "step": 105993
    },
    {
      "epoch": 1.38,
      "grad_norm": 1.0344126224517822,
      "learning_rate": 0.00033756498720446146,
      "loss": 2.1869,
      "step": 105994
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.888727068901062,
      "learning_rate": 0.00033756092883387173,
      "loss": 2.2407,
      "step": 105995
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.9503845572471619,
      "learning_rate": 0.00033755687045629887,
      "loss": 2.0117,
      "step": 105996
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.8245825171470642,
      "learning_rate": 0.0003375528120717433,
      "loss": 1.8685,
      "step": 105997
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.8902027010917664,
      "learning_rate": 0.000337548753680206,
      "loss": 2.1751,
      "step": 105998
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.8836636543273926,
      "learning_rate": 0.00033754469528168775,
      "loss": 2.225,
      "step": 105999
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.9625490307807922,
      "learning_rate": 0.0003375406368761892,
      "loss": 2.1529,
      "step": 106000
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.9707117080688477,
      "learning_rate": 0.00033753657846371116,
      "loss": 2.1337,
      "step": 106001
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.8814305067062378,
      "learning_rate": 0.00033753252004425433,
      "loss": 2.1879,
      "step": 106002
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.9818302989006042,
      "learning_rate": 0.0003375284616178195,
      "loss": 2.2472,
      "step": 106003
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.9021428823471069,
      "learning_rate": 0.00033752440318440737,
      "loss": 2.1728,
      "step": 106004
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.919251024723053,
      "learning_rate": 0.00033752034474401877,
      "loss": 2.0838,
      "step": 106005
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.8884474635124207,
      "learning_rate": 0.0003375162862966545,
      "loss": 1.9963,
      "step": 106006
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.9383111596107483,
      "learning_rate": 0.0003375122278423152,
      "loss": 2.3241,
      "step": 106007
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.9170196056365967,
      "learning_rate": 0.00033750816938100166,
      "loss": 2.3348,
      "step": 106008
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.8845292925834656,
      "learning_rate": 0.0003375041109127147,
      "loss": 2.0308,
      "step": 106009
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.893150269985199,
      "learning_rate": 0.00033750005243745495,
      "loss": 2.3184,
      "step": 106010
    },
    {
      "epoch": 1.38,
      "grad_norm": 1.0031501054763794,
      "learning_rate": 0.00033749599395522324,
      "loss": 2.2483,
      "step": 106011
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.8424952626228333,
      "learning_rate": 0.00033749193546602033,
      "loss": 2.0306,
      "step": 106012
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.9244915843009949,
      "learning_rate": 0.000337487876969847,
      "loss": 2.1852,
      "step": 106013
    },
    {
      "epoch": 1.38,
      "grad_norm": 1.0047335624694824,
      "learning_rate": 0.0003374838184667039,
      "loss": 2.0489,
      "step": 106014
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.9259185194969177,
      "learning_rate": 0.00033747975995659186,
      "loss": 2.1958,
      "step": 106015
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.9474926590919495,
      "learning_rate": 0.0003374757014395117,
      "loss": 2.2976,
      "step": 106016
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.8630908131599426,
      "learning_rate": 0.00033747164291546397,
      "loss": 2.0269,
      "step": 106017
    },
    {
      "epoch": 1.38,
      "grad_norm": 1.0300487279891968,
      "learning_rate": 0.0003374675843844497,
      "loss": 2.1435,
      "step": 106018
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.8336251974105835,
      "learning_rate": 0.0003374635258464694,
      "loss": 2.2495,
      "step": 106019
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.8765172958374023,
      "learning_rate": 0.0003374594673015239,
      "loss": 1.9999,
      "step": 106020
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.9525994062423706,
      "learning_rate": 0.0003374554087496141,
      "loss": 2.0288,
      "step": 106021
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.9645287394523621,
      "learning_rate": 0.0003374513501907405,
      "loss": 2.2563,
      "step": 106022
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.9644055366516113,
      "learning_rate": 0.00033744729162490407,
      "loss": 2.2165,
      "step": 106023
    },
    {
      "epoch": 1.38,
      "grad_norm": 1.0403674840927124,
      "learning_rate": 0.0003374432330521055,
      "loss": 2.1187,
      "step": 106024
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.8389461040496826,
      "learning_rate": 0.00033743917447234546,
      "loss": 2.0637,
      "step": 106025
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.8251623511314392,
      "learning_rate": 0.0003374351158856248,
      "loss": 2.136,
      "step": 106026
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.9205735325813293,
      "learning_rate": 0.00033743105729194426,
      "loss": 2.2917,
      "step": 106027
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.93996661901474,
      "learning_rate": 0.00033742699869130456,
      "loss": 2.331,
      "step": 106028
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.8182680606842041,
      "learning_rate": 0.00033742294008370645,
      "loss": 2.2157,
      "step": 106029
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.8187704086303711,
      "learning_rate": 0.0003374188814691508,
      "loss": 2.1243,
      "step": 106030
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.8843543529510498,
      "learning_rate": 0.0003374148228476382,
      "loss": 1.8893,
      "step": 106031
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.9933049082756042,
      "learning_rate": 0.0003374107642191694,
      "loss": 2.1641,
      "step": 106032
    },
    {
      "epoch": 1.38,
      "grad_norm": 1.0027316808700562,
      "learning_rate": 0.0003374067055837454,
      "loss": 1.9448,
      "step": 106033
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.8579859733581543,
      "learning_rate": 0.0003374026469413667,
      "loss": 2.1589,
      "step": 106034
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.8786427974700928,
      "learning_rate": 0.0003373985882920341,
      "loss": 2.2218,
      "step": 106035
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.9101779460906982,
      "learning_rate": 0.0003373945296357485,
      "loss": 2.2023,
      "step": 106036
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.9399449229240417,
      "learning_rate": 0.0003373904709725104,
      "loss": 1.8329,
      "step": 106037
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.9485193490982056,
      "learning_rate": 0.0003373864123023208,
      "loss": 2.0567,
      "step": 106038
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.9170982241630554,
      "learning_rate": 0.00033738235362518046,
      "loss": 2.0098,
      "step": 106039
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.802733302116394,
      "learning_rate": 0.00033737829494108994,
      "loss": 2.113,
      "step": 106040
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.8954957127571106,
      "learning_rate": 0.00033737423625005,
      "loss": 2.1744,
      "step": 106041
    },
    {
      "epoch": 1.38,
      "grad_norm": 1.0450061559677124,
      "learning_rate": 0.0003373701775520616,
      "loss": 2.1988,
      "step": 106042
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.9088910818099976,
      "learning_rate": 0.00033736611884712536,
      "loss": 2.171,
      "step": 106043
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.968105137348175,
      "learning_rate": 0.00033736206013524207,
      "loss": 2.1186,
      "step": 106044
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.9754831194877625,
      "learning_rate": 0.0003373580014164124,
      "loss": 2.177,
      "step": 106045
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.8304501175880432,
      "learning_rate": 0.00033735394269063723,
      "loss": 2.1246,
      "step": 106046
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.9167531728744507,
      "learning_rate": 0.00033734988395791725,
      "loss": 2.3085,
      "step": 106047
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.8655388355255127,
      "learning_rate": 0.00033734582521825323,
      "loss": 2.2162,
      "step": 106048
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.9304793477058411,
      "learning_rate": 0.00033734176647164587,
      "loss": 2.1155,
      "step": 106049
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.8151758313179016,
      "learning_rate": 0.000337337707718096,
      "loss": 2.1808,
      "step": 106050
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.8674752712249756,
      "learning_rate": 0.0003373336489576044,
      "loss": 2.2782,
      "step": 106051
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.8908616900444031,
      "learning_rate": 0.00033732959019017167,
      "loss": 2.1233,
      "step": 106052
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.8217060565948486,
      "learning_rate": 0.00033732553141579865,
      "loss": 1.9572,
      "step": 106053
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.9152014255523682,
      "learning_rate": 0.0003373214726344862,
      "loss": 2.0165,
      "step": 106054
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.8655322194099426,
      "learning_rate": 0.00033731741384623496,
      "loss": 2.2071,
      "step": 106055
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.9755747318267822,
      "learning_rate": 0.0003373133550510457,
      "loss": 2.253,
      "step": 106056
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.8391243815422058,
      "learning_rate": 0.00033730929624891916,
      "loss": 2.116,
      "step": 106057
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.873867928981781,
      "learning_rate": 0.0003373052374398561,
      "loss": 2.198,
      "step": 106058
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.8850721716880798,
      "learning_rate": 0.0003373011786238574,
      "loss": 2.2922,
      "step": 106059
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.9148004055023193,
      "learning_rate": 0.0003372971198009236,
      "loss": 2.1063,
      "step": 106060
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.8797204494476318,
      "learning_rate": 0.00033729306097105564,
      "loss": 2.2636,
      "step": 106061
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.9120399355888367,
      "learning_rate": 0.0003372890021342541,
      "loss": 2.1519,
      "step": 106062
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.8245095610618591,
      "learning_rate": 0.0003372849432905199,
      "loss": 2.1328,
      "step": 106063
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.9008610248565674,
      "learning_rate": 0.00033728088443985373,
      "loss": 2.1409,
      "step": 106064
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.9217045307159424,
      "learning_rate": 0.00033727682558225633,
      "loss": 2.0685,
      "step": 106065
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.8775584101676941,
      "learning_rate": 0.00033727276671772843,
      "loss": 2.2676,
      "step": 106066
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.7919285893440247,
      "learning_rate": 0.00033726870784627085,
      "loss": 2.1454,
      "step": 106067
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.857395589351654,
      "learning_rate": 0.0003372646489678843,
      "loss": 2.1032,
      "step": 106068
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.9121689796447754,
      "learning_rate": 0.00033726059008256954,
      "loss": 2.2456,
      "step": 106069
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.9709591865539551,
      "learning_rate": 0.0003372565311903274,
      "loss": 1.9276,
      "step": 106070
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.9986193776130676,
      "learning_rate": 0.0003372524722911585,
      "loss": 2.2922,
      "step": 106071
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.8784605860710144,
      "learning_rate": 0.00033724841338506365,
      "loss": 2.005,
      "step": 106072
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.8690974712371826,
      "learning_rate": 0.0003372443544720437,
      "loss": 2.0063,
      "step": 106073
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.904270350933075,
      "learning_rate": 0.0003372402955520993,
      "loss": 2.321,
      "step": 106074
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.837762713432312,
      "learning_rate": 0.0003372362366252311,
      "loss": 2.2799,
      "step": 106075
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.8852826952934265,
      "learning_rate": 0.00033723217769144004,
      "loss": 2.0458,
      "step": 106076
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.9191261529922485,
      "learning_rate": 0.0003372281187507269,
      "loss": 2.0154,
      "step": 106077
    },
    {
      "epoch": 1.38,
      "grad_norm": 1.0045543909072876,
      "learning_rate": 0.0003372240598030923,
      "loss": 2.1338,
      "step": 106078
    },
    {
      "epoch": 1.38,
      "grad_norm": 1.0426183938980103,
      "learning_rate": 0.00033722000084853706,
      "loss": 1.9973,
      "step": 106079
    },
    {
      "epoch": 1.38,
      "grad_norm": 1.0379903316497803,
      "learning_rate": 0.00033721594188706186,
      "loss": 2.0057,
      "step": 106080
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.980814516544342,
      "learning_rate": 0.00033721188291866756,
      "loss": 2.4269,
      "step": 106081
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.8956027030944824,
      "learning_rate": 0.00033720782394335485,
      "loss": 2.1913,
      "step": 106082
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.962173342704773,
      "learning_rate": 0.00033720376496112456,
      "loss": 2.1316,
      "step": 106083
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.9697239995002747,
      "learning_rate": 0.0003371997059719773,
      "loss": 2.2133,
      "step": 106084
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.9796885848045349,
      "learning_rate": 0.000337195646975914,
      "loss": 2.1335,
      "step": 106085
    },
    {
      "epoch": 1.38,
      "grad_norm": 1.3951336145401,
      "learning_rate": 0.00033719158797293524,
      "loss": 2.0002,
      "step": 106086
    },
    {
      "epoch": 1.38,
      "grad_norm": 1.0714178085327148,
      "learning_rate": 0.0003371875289630419,
      "loss": 2.1842,
      "step": 106087
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.8342587947845459,
      "learning_rate": 0.00033718346994623477,
      "loss": 2.0209,
      "step": 106088
    },
    {
      "epoch": 1.38,
      "grad_norm": 1.0143041610717773,
      "learning_rate": 0.0003371794109225144,
      "loss": 2.1565,
      "step": 106089
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.9043737053871155,
      "learning_rate": 0.00033717535189188173,
      "loss": 1.9849,
      "step": 106090
    },
    {
      "epoch": 1.38,
      "grad_norm": 1.0134460926055908,
      "learning_rate": 0.0003371712928543375,
      "loss": 2.4101,
      "step": 106091
    },
    {
      "epoch": 1.38,
      "grad_norm": 1.011081576347351,
      "learning_rate": 0.00033716723380988235,
      "loss": 2.1792,
      "step": 106092
    },
    {
      "epoch": 1.38,
      "grad_norm": 1.3255565166473389,
      "learning_rate": 0.0003371631747585172,
      "loss": 2.0819,
      "step": 106093
    },
    {
      "epoch": 1.38,
      "grad_norm": 1.1527832746505737,
      "learning_rate": 0.0003371591157002427,
      "loss": 2.4976,
      "step": 106094
    },
    {
      "epoch": 1.38,
      "grad_norm": 1.0157086849212646,
      "learning_rate": 0.00033715505663505955,
      "loss": 2.1657,
      "step": 106095
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.9046047925949097,
      "learning_rate": 0.0003371509975629686,
      "loss": 1.9376,
      "step": 106096
    },
    {
      "epoch": 1.38,
      "grad_norm": 1.1759618520736694,
      "learning_rate": 0.00033714693848397063,
      "loss": 2.0831,
      "step": 106097
    },
    {
      "epoch": 1.38,
      "grad_norm": 1.0637497901916504,
      "learning_rate": 0.0003371428793980663,
      "loss": 2.3456,
      "step": 106098
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.9092873334884644,
      "learning_rate": 0.0003371388203052564,
      "loss": 2.3427,
      "step": 106099
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.9199640154838562,
      "learning_rate": 0.00033713476120554174,
      "loss": 2.1065,
      "step": 106100
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.8934067487716675,
      "learning_rate": 0.00033713070209892294,
      "loss": 2.1605,
      "step": 106101
    },
    {
      "epoch": 1.38,
      "grad_norm": 1.0268545150756836,
      "learning_rate": 0.0003371266429854009,
      "loss": 2.2149,
      "step": 106102
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.861835241317749,
      "learning_rate": 0.0003371225838649763,
      "loss": 2.2472,
      "step": 106103
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.9238477349281311,
      "learning_rate": 0.00033711852473764995,
      "loss": 2.2613,
      "step": 106104
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.8889025449752808,
      "learning_rate": 0.00033711446560342253,
      "loss": 2.3655,
      "step": 106105
    },
    {
      "epoch": 1.38,
      "grad_norm": 1.0368634462356567,
      "learning_rate": 0.0003371104064622948,
      "loss": 2.3527,
      "step": 106106
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.8800509572029114,
      "learning_rate": 0.0003371063473142676,
      "loss": 2.1826,
      "step": 106107
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.826251745223999,
      "learning_rate": 0.00033710228815934164,
      "loss": 2.0639,
      "step": 106108
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.9379958510398865,
      "learning_rate": 0.0003370982289975177,
      "loss": 2.2508,
      "step": 106109
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.9388294219970703,
      "learning_rate": 0.00033709416982879634,
      "loss": 2.2358,
      "step": 106110
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.8788569569587708,
      "learning_rate": 0.0003370901106531786,
      "loss": 2.3473,
      "step": 106111
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.8667799830436707,
      "learning_rate": 0.00033708605147066514,
      "loss": 2.0036,
      "step": 106112
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.8663079142570496,
      "learning_rate": 0.00033708199228125655,
      "loss": 1.9387,
      "step": 106113
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.9068949222564697,
      "learning_rate": 0.0003370779330849538,
      "loss": 2.1247,
      "step": 106114
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.797144889831543,
      "learning_rate": 0.0003370738738817575,
      "loss": 2.0074,
      "step": 106115
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.9050621390342712,
      "learning_rate": 0.0003370698146716685,
      "loss": 2.0615,
      "step": 106116
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.8118857741355896,
      "learning_rate": 0.0003370657554546876,
      "loss": 2.0244,
      "step": 106117
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.9585614204406738,
      "learning_rate": 0.00033706169623081545,
      "loss": 2.2822,
      "step": 106118
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.8798186779022217,
      "learning_rate": 0.0003370576370000528,
      "loss": 2.1613,
      "step": 106119
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.9090662598609924,
      "learning_rate": 0.00033705357776240034,
      "loss": 2.1554,
      "step": 106120
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.9598129391670227,
      "learning_rate": 0.00033704951851785904,
      "loss": 1.8892,
      "step": 106121
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.9499902725219727,
      "learning_rate": 0.00033704545926642953,
      "loss": 2.2919,
      "step": 106122
    },
    {
      "epoch": 1.38,
      "grad_norm": 1.0412310361862183,
      "learning_rate": 0.0003370414000081125,
      "loss": 2.4053,
      "step": 106123
    },
    {
      "epoch": 1.38,
      "grad_norm": 1.0258773565292358,
      "learning_rate": 0.00033703734074290887,
      "loss": 2.1664,
      "step": 106124
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.8510246872901917,
      "learning_rate": 0.00033703328147081924,
      "loss": 2.1357,
      "step": 106125
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.9948853254318237,
      "learning_rate": 0.0003370292221918444,
      "loss": 2.2046,
      "step": 106126
    },
    {
      "epoch": 1.38,
      "grad_norm": 1.041340708732605,
      "learning_rate": 0.00033702516290598516,
      "loss": 1.9952,
      "step": 106127
    },
    {
      "epoch": 1.38,
      "grad_norm": 1.146366834640503,
      "learning_rate": 0.0003370211036132423,
      "loss": 2.1926,
      "step": 106128
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.8735383152961731,
      "learning_rate": 0.00033701704431361647,
      "loss": 2.0955,
      "step": 106129
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.8982213735580444,
      "learning_rate": 0.0003370129850071084,
      "loss": 2.2478,
      "step": 106130
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.8907575011253357,
      "learning_rate": 0.000337008925693719,
      "loss": 2.1438,
      "step": 106131
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.9810653328895569,
      "learning_rate": 0.0003370048663734489,
      "loss": 2.313,
      "step": 106132
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.926682710647583,
      "learning_rate": 0.00033700080704629894,
      "loss": 2.2233,
      "step": 106133
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.900675892829895,
      "learning_rate": 0.0003369967477122698,
      "loss": 2.024,
      "step": 106134
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.8580861687660217,
      "learning_rate": 0.0003369926883713623,
      "loss": 2.1491,
      "step": 106135
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.8887682557106018,
      "learning_rate": 0.00033698862902357713,
      "loss": 2.1493,
      "step": 106136
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.8957467675209045,
      "learning_rate": 0.00033698456966891515,
      "loss": 2.2652,
      "step": 106137
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.8907364010810852,
      "learning_rate": 0.00033698051030737694,
      "loss": 2.0838,
      "step": 106138
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.9553897380828857,
      "learning_rate": 0.00033697645093896337,
      "loss": 2.2824,
      "step": 106139
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.9895813465118408,
      "learning_rate": 0.00033697239156367525,
      "loss": 2.1136,
      "step": 106140
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.8590937256813049,
      "learning_rate": 0.0003369683321815132,
      "loss": 2.3817,
      "step": 106141
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.8709893226623535,
      "learning_rate": 0.000336964272792478,
      "loss": 2.0882,
      "step": 106142
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.8616853356361389,
      "learning_rate": 0.00033696021339657057,
      "loss": 2.0596,
      "step": 106143
    },
    {
      "epoch": 1.38,
      "grad_norm": 1.1329448223114014,
      "learning_rate": 0.0003369561539937915,
      "loss": 2.2361,
      "step": 106144
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.9174606204032898,
      "learning_rate": 0.00033695209458414147,
      "loss": 1.9708,
      "step": 106145
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.8992353677749634,
      "learning_rate": 0.0003369480351676215,
      "loss": 2.26,
      "step": 106146
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.8465089797973633,
      "learning_rate": 0.0003369439757442321,
      "loss": 2.2058,
      "step": 106147
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.849444568157196,
      "learning_rate": 0.0003369399163139741,
      "loss": 2.325,
      "step": 106148
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.8782840371131897,
      "learning_rate": 0.00033693585687684835,
      "loss": 2.2836,
      "step": 106149
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.9753190875053406,
      "learning_rate": 0.00033693179743285543,
      "loss": 2.0631,
      "step": 106150
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.8881868720054626,
      "learning_rate": 0.00033692773798199625,
      "loss": 2.0471,
      "step": 106151
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.9738892912864685,
      "learning_rate": 0.00033692367852427153,
      "loss": 2.0664,
      "step": 106152
    },
    {
      "epoch": 1.38,
      "grad_norm": 1.0318807363510132,
      "learning_rate": 0.000336919619059682,
      "loss": 1.9345,
      "step": 106153
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.8789315223693848,
      "learning_rate": 0.00033691555958822837,
      "loss": 2.3282,
      "step": 106154
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.9656167030334473,
      "learning_rate": 0.0003369115001099115,
      "loss": 1.9776,
      "step": 106155
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.8738663792610168,
      "learning_rate": 0.00033690744062473207,
      "loss": 2.2226,
      "step": 106156
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.9698391556739807,
      "learning_rate": 0.0003369033811326908,
      "loss": 2.2244,
      "step": 106157
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.893047034740448,
      "learning_rate": 0.00033689932163378855,
      "loss": 2.1719,
      "step": 106158
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.9618669748306274,
      "learning_rate": 0.00033689526212802597,
      "loss": 2.3561,
      "step": 106159
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.8833048343658447,
      "learning_rate": 0.0003368912026154039,
      "loss": 2.238,
      "step": 106160
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.8995581865310669,
      "learning_rate": 0.000336887143095923,
      "loss": 2.1881,
      "step": 106161
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.8360835313796997,
      "learning_rate": 0.0003368830835695842,
      "loss": 2.2587,
      "step": 106162
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.9318139553070068,
      "learning_rate": 0.00033687902403638806,
      "loss": 2.3971,
      "step": 106163
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.9377683401107788,
      "learning_rate": 0.0003368749644963354,
      "loss": 2.0445,
      "step": 106164
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.9157235622406006,
      "learning_rate": 0.00033687090494942704,
      "loss": 2.1818,
      "step": 106165
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.8656181693077087,
      "learning_rate": 0.00033686684539566364,
      "loss": 2.0061,
      "step": 106166
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.8983341455459595,
      "learning_rate": 0.00033686278583504596,
      "loss": 2.3075,
      "step": 106167
    },
    {
      "epoch": 1.38,
      "grad_norm": 1.0367234945297241,
      "learning_rate": 0.0003368587262675748,
      "loss": 2.0964,
      "step": 106168
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.9936960339546204,
      "learning_rate": 0.000336854666693251,
      "loss": 2.0964,
      "step": 106169
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.9167405366897583,
      "learning_rate": 0.0003368506071120752,
      "loss": 2.3011,
      "step": 106170
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.9235268235206604,
      "learning_rate": 0.0003368465475240481,
      "loss": 2.111,
      "step": 106171
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.9181755781173706,
      "learning_rate": 0.0003368424879291706,
      "loss": 2.3236,
      "step": 106172
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.9247270822525024,
      "learning_rate": 0.0003368384283274434,
      "loss": 2.2246,
      "step": 106173
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.940606415271759,
      "learning_rate": 0.0003368343687188672,
      "loss": 2.2383,
      "step": 106174
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.9736003279685974,
      "learning_rate": 0.00033683030910344275,
      "loss": 2.1283,
      "step": 106175
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.9482579231262207,
      "learning_rate": 0.00033682624948117094,
      "loss": 2.3285,
      "step": 106176
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.937324047088623,
      "learning_rate": 0.0003368221898520524,
      "loss": 2.0712,
      "step": 106177
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.9741924405097961,
      "learning_rate": 0.00033681813021608787,
      "loss": 1.9598,
      "step": 106178
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.9801791906356812,
      "learning_rate": 0.0003368140705732783,
      "loss": 2.1255,
      "step": 106179
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.9279784560203552,
      "learning_rate": 0.00033681001092362414,
      "loss": 2.0938,
      "step": 106180
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.9952864646911621,
      "learning_rate": 0.00033680595126712635,
      "loss": 2.198,
      "step": 106181
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.8505128026008606,
      "learning_rate": 0.0003368018916037857,
      "loss": 2.0627,
      "step": 106182
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.934260368347168,
      "learning_rate": 0.00033679783193360285,
      "loss": 2.0898,
      "step": 106183
    },
    {
      "epoch": 1.38,
      "grad_norm": 1.0423245429992676,
      "learning_rate": 0.00033679377225657856,
      "loss": 2.1604,
      "step": 106184
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.8198818564414978,
      "learning_rate": 0.0003367897125727136,
      "loss": 2.0924,
      "step": 106185
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.9333913326263428,
      "learning_rate": 0.0003367856528820088,
      "loss": 1.9322,
      "step": 106186
    },
    {
      "epoch": 1.38,
      "grad_norm": 1.1056032180786133,
      "learning_rate": 0.0003367815931844648,
      "loss": 2.2675,
      "step": 106187
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.9076785445213318,
      "learning_rate": 0.00033677753348008245,
      "loss": 2.1223,
      "step": 106188
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.9143353700637817,
      "learning_rate": 0.0003367734737688624,
      "loss": 2.2591,
      "step": 106189
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.8650328516960144,
      "learning_rate": 0.0003367694140508056,
      "loss": 1.9474,
      "step": 106190
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.994536280632019,
      "learning_rate": 0.00033676535432591257,
      "loss": 2.3494,
      "step": 106191
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.9573734998703003,
      "learning_rate": 0.00033676129459418415,
      "loss": 2.0969,
      "step": 106192
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.9269607663154602,
      "learning_rate": 0.0003367572348556211,
      "loss": 2.1767,
      "step": 106193
    },
    {
      "epoch": 1.38,
      "grad_norm": 1.1722842454910278,
      "learning_rate": 0.0003367531751102242,
      "loss": 2.1847,
      "step": 106194
    },
    {
      "epoch": 1.38,
      "grad_norm": 1.2652785778045654,
      "learning_rate": 0.00033674911535799427,
      "loss": 2.3983,
      "step": 106195
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.8546084761619568,
      "learning_rate": 0.0003367450555989319,
      "loss": 2.1496,
      "step": 106196
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.8577078580856323,
      "learning_rate": 0.00033674099583303797,
      "loss": 2.184,
      "step": 106197
    },
    {
      "epoch": 1.38,
      "grad_norm": 1.038993000984192,
      "learning_rate": 0.0003367369360603132,
      "loss": 2.1116,
      "step": 106198
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.8665173053741455,
      "learning_rate": 0.0003367328762807583,
      "loss": 2.1418,
      "step": 106199
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.8513891100883484,
      "learning_rate": 0.00033672881649437407,
      "loss": 2.2269,
      "step": 106200
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.90081387758255,
      "learning_rate": 0.00033672475670116136,
      "loss": 2.0939,
      "step": 106201
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.9496854543685913,
      "learning_rate": 0.00033672069690112073,
      "loss": 2.0381,
      "step": 106202
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.887895941734314,
      "learning_rate": 0.000336716637094253,
      "loss": 2.0198,
      "step": 106203
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.8989525437355042,
      "learning_rate": 0.00033671257728055905,
      "loss": 2.3337,
      "step": 106204
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.874005138874054,
      "learning_rate": 0.00033670851746003946,
      "loss": 1.89,
      "step": 106205
    },
    {
      "epoch": 1.38,
      "grad_norm": 1.046248197555542,
      "learning_rate": 0.0003367044576326951,
      "loss": 2.0334,
      "step": 106206
    },
    {
      "epoch": 1.38,
      "grad_norm": 1.085261344909668,
      "learning_rate": 0.0003367003977985267,
      "loss": 2.2531,
      "step": 106207
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.9738870859146118,
      "learning_rate": 0.000336696337957535,
      "loss": 2.0998,
      "step": 106208
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.898463785648346,
      "learning_rate": 0.0003366922781097207,
      "loss": 2.2417,
      "step": 106209
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.9108679294586182,
      "learning_rate": 0.0003366882182550847,
      "loss": 2.0895,
      "step": 106210
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.872689962387085,
      "learning_rate": 0.00033668415839362755,
      "loss": 2.2504,
      "step": 106211
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.9026445150375366,
      "learning_rate": 0.0003366800985253502,
      "loss": 2.1273,
      "step": 106212
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.9389908313751221,
      "learning_rate": 0.00033667603865025335,
      "loss": 2.2563,
      "step": 106213
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.8573412299156189,
      "learning_rate": 0.00033667197876833773,
      "loss": 2.2897,
      "step": 106214
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.8828665614128113,
      "learning_rate": 0.00033666791887960405,
      "loss": 2.0012,
      "step": 106215
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.9359121322631836,
      "learning_rate": 0.00033666385898405317,
      "loss": 2.2137,
      "step": 106216
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.9435034990310669,
      "learning_rate": 0.00033665979908168575,
      "loss": 2.0717,
      "step": 106217
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.8581243753433228,
      "learning_rate": 0.00033665573917250255,
      "loss": 2.5568,
      "step": 106218
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.8894219994544983,
      "learning_rate": 0.00033665167925650444,
      "loss": 2.1409,
      "step": 106219
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.8359277844429016,
      "learning_rate": 0.000336647619333692,
      "loss": 2.121,
      "step": 106220
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.9724753499031067,
      "learning_rate": 0.00033664355940406606,
      "loss": 2.1202,
      "step": 106221
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.8963128328323364,
      "learning_rate": 0.00033663949946762754,
      "loss": 2.1914,
      "step": 106222
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.8402953743934631,
      "learning_rate": 0.0003366354395243769,
      "loss": 2.0925,
      "step": 106223
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.9348763227462769,
      "learning_rate": 0.00033663137957431506,
      "loss": 2.2206,
      "step": 106224
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.863240659236908,
      "learning_rate": 0.00033662731961744285,
      "loss": 2.3121,
      "step": 106225
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.8939200043678284,
      "learning_rate": 0.0003366232596537608,
      "loss": 2.1894,
      "step": 106226
    },
    {
      "epoch": 1.38,
      "grad_norm": 1.0128341913223267,
      "learning_rate": 0.00033661919968326983,
      "loss": 2.0131,
      "step": 106227
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.8701549768447876,
      "learning_rate": 0.00033661513970597077,
      "loss": 2.0559,
      "step": 106228
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.978956937789917,
      "learning_rate": 0.0003366110797218642,
      "loss": 2.1304,
      "step": 106229
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.959561824798584,
      "learning_rate": 0.00033660701973095087,
      "loss": 2.4152,
      "step": 106230
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.8471099138259888,
      "learning_rate": 0.0003366029597332317,
      "loss": 1.9989,
      "step": 106231
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.8344279527664185,
      "learning_rate": 0.0003365988997287073,
      "loss": 2.1645,
      "step": 106232
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.8979983925819397,
      "learning_rate": 0.0003365948397173785,
      "loss": 1.8698,
      "step": 106233
    },
    {
      "epoch": 1.38,
      "grad_norm": 1.0726962089538574,
      "learning_rate": 0.00033659077969924603,
      "loss": 2.4663,
      "step": 106234
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.9214297533035278,
      "learning_rate": 0.0003365867196743106,
      "loss": 2.074,
      "step": 106235
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.8706345558166504,
      "learning_rate": 0.00033658265964257297,
      "loss": 2.2402,
      "step": 106236
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.8751536011695862,
      "learning_rate": 0.0003365785996040341,
      "loss": 2.3895,
      "step": 106237
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.8636868000030518,
      "learning_rate": 0.0003365745395586944,
      "loss": 2.1746,
      "step": 106238
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.810387134552002,
      "learning_rate": 0.00033657047950655486,
      "loss": 1.9963,
      "step": 106239
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.9311338067054749,
      "learning_rate": 0.00033656641944761624,
      "loss": 2.0746,
      "step": 106240
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.8526797890663147,
      "learning_rate": 0.0003365623593818791,
      "loss": 2.0959,
      "step": 106241
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.88741534948349,
      "learning_rate": 0.00033655829930934444,
      "loss": 1.9057,
      "step": 106242
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.8259417414665222,
      "learning_rate": 0.0003365542392300129,
      "loss": 2.2151,
      "step": 106243
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.8242536783218384,
      "learning_rate": 0.0003365501791438852,
      "loss": 2.251,
      "step": 106244
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.8717566132545471,
      "learning_rate": 0.00033654611905096214,
      "loss": 2.1803,
      "step": 106245
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.9427387118339539,
      "learning_rate": 0.0003365420589512445,
      "loss": 2.1499,
      "step": 106246
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.9079916477203369,
      "learning_rate": 0.000336537998844733,
      "loss": 2.0615,
      "step": 106247
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.9004547595977783,
      "learning_rate": 0.0003365339387314283,
      "loss": 2.0765,
      "step": 106248
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.8222251534461975,
      "learning_rate": 0.0003365298786113313,
      "loss": 2.1149,
      "step": 106249
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.8868497014045715,
      "learning_rate": 0.0003365258184844427,
      "loss": 1.927,
      "step": 106250
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.7968342900276184,
      "learning_rate": 0.0003365217583507633,
      "loss": 2.1356,
      "step": 106251
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.9713367819786072,
      "learning_rate": 0.00033651769821029377,
      "loss": 2.3672,
      "step": 106252
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.8914603590965271,
      "learning_rate": 0.00033651363806303495,
      "loss": 2.2429,
      "step": 106253
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.8508316278457642,
      "learning_rate": 0.0003365095779089876,
      "loss": 1.9486,
      "step": 106254
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.8705137968063354,
      "learning_rate": 0.0003365055177481523,
      "loss": 1.965,
      "step": 106255
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.9345521330833435,
      "learning_rate": 0.00033650145758053,
      "loss": 2.1968,
      "step": 106256
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.8331261277198792,
      "learning_rate": 0.00033649739740612147,
      "loss": 2.2177,
      "step": 106257
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.917417049407959,
      "learning_rate": 0.0003364933372249273,
      "loss": 2.0096,
      "step": 106258
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.8580313324928284,
      "learning_rate": 0.0003364892770369483,
      "loss": 2.0691,
      "step": 106259
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.8025978803634644,
      "learning_rate": 0.00033648521684218535,
      "loss": 2.0569,
      "step": 106260
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.8717888593673706,
      "learning_rate": 0.00033648115664063906,
      "loss": 2.2315,
      "step": 106261
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.8854489326477051,
      "learning_rate": 0.00033647709643231023,
      "loss": 2.0587,
      "step": 106262
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.8705074787139893,
      "learning_rate": 0.0003364730362171996,
      "loss": 2.1855,
      "step": 106263
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.8706569075584412,
      "learning_rate": 0.00033646897599530793,
      "loss": 2.0744,
      "step": 106264
    },
    {
      "epoch": 1.38,
      "grad_norm": 1.1147725582122803,
      "learning_rate": 0.0003364649157666361,
      "loss": 2.3411,
      "step": 106265
    },
    {
      "epoch": 1.38,
      "grad_norm": 1.1023224592208862,
      "learning_rate": 0.00033646085553118465,
      "loss": 2.1601,
      "step": 106266
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.9343286752700806,
      "learning_rate": 0.00033645679528895443,
      "loss": 2.2344,
      "step": 106267
    },
    {
      "epoch": 1.38,
      "grad_norm": 1.018157958984375,
      "learning_rate": 0.0003364527350399463,
      "loss": 2.1265,
      "step": 106268
    },
    {
      "epoch": 1.38,
      "grad_norm": 1.3392341136932373,
      "learning_rate": 0.00033644867478416083,
      "loss": 2.3564,
      "step": 106269
    },
    {
      "epoch": 1.38,
      "grad_norm": 1.0865815877914429,
      "learning_rate": 0.00033644461452159884,
      "loss": 2.1835,
      "step": 106270
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.9177111387252808,
      "learning_rate": 0.00033644055425226124,
      "loss": 2.1471,
      "step": 106271
    },
    {
      "epoch": 1.38,
      "grad_norm": 1.0311334133148193,
      "learning_rate": 0.0003364364939761486,
      "loss": 2.206,
      "step": 106272
    },
    {
      "epoch": 1.38,
      "grad_norm": 1.2061070203781128,
      "learning_rate": 0.0003364324336932616,
      "loss": 2.3589,
      "step": 106273
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.9774112105369568,
      "learning_rate": 0.00033642837340360133,
      "loss": 2.1993,
      "step": 106274
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.9383080005645752,
      "learning_rate": 0.00033642431310716817,
      "loss": 2.0069,
      "step": 106275
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.932486891746521,
      "learning_rate": 0.00033642025280396313,
      "loss": 2.0119,
      "step": 106276
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.9035506844520569,
      "learning_rate": 0.00033641619249398687,
      "loss": 2.181,
      "step": 106277
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.8392202258110046,
      "learning_rate": 0.00033641213217724014,
      "loss": 2.217,
      "step": 106278
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.9768018126487732,
      "learning_rate": 0.00033640807185372366,
      "loss": 2.1898,
      "step": 106279
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.9816322326660156,
      "learning_rate": 0.00033640401152343834,
      "loss": 2.144,
      "step": 106280
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.9664838910102844,
      "learning_rate": 0.00033639995118638473,
      "loss": 2.2467,
      "step": 106281
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.8604744672775269,
      "learning_rate": 0.0003363958908425637,
      "loss": 2.4074,
      "step": 106282
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.9703757762908936,
      "learning_rate": 0.00033639183049197604,
      "loss": 2.1904,
      "step": 106283
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.8451900482177734,
      "learning_rate": 0.0003363877701346224,
      "loss": 2.1038,
      "step": 106284
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.8578360676765442,
      "learning_rate": 0.00033638370977050354,
      "loss": 2.2288,
      "step": 106285
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.8837850093841553,
      "learning_rate": 0.0003363796493996204,
      "loss": 2.2251,
      "step": 106286
    },
    {
      "epoch": 1.38,
      "grad_norm": 1.0721752643585205,
      "learning_rate": 0.00033637558902197353,
      "loss": 1.9901,
      "step": 106287
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.9099077582359314,
      "learning_rate": 0.0003363715286375637,
      "loss": 2.2588,
      "step": 106288
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.8307567834854126,
      "learning_rate": 0.0003363674682463918,
      "loss": 2.2602,
      "step": 106289
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.9784882068634033,
      "learning_rate": 0.0003363634078484584,
      "loss": 2.0309,
      "step": 106290
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.9310339689254761,
      "learning_rate": 0.0003363593474437644,
      "loss": 2.3196,
      "step": 106291
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.7909137010574341,
      "learning_rate": 0.00033635528703231056,
      "loss": 2.2954,
      "step": 106292
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.8582414984703064,
      "learning_rate": 0.00033635122661409753,
      "loss": 2.3634,
      "step": 106293
    },
    {
      "epoch": 1.38,
      "grad_norm": 1.0557048320770264,
      "learning_rate": 0.0003363471661891261,
      "loss": 2.2067,
      "step": 106294
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.983106255531311,
      "learning_rate": 0.00033634310575739716,
      "loss": 2.1222,
      "step": 106295
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.8189228177070618,
      "learning_rate": 0.00033633904531891123,
      "loss": 2.0428,
      "step": 106296
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.8438263535499573,
      "learning_rate": 0.0003363349848736692,
      "loss": 2.1532,
      "step": 106297
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.9857704639434814,
      "learning_rate": 0.0003363309244216719,
      "loss": 2.1149,
      "step": 106298
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.8831541538238525,
      "learning_rate": 0.00033632686396291994,
      "loss": 1.8881,
      "step": 106299
    },
    {
      "epoch": 1.38,
      "grad_norm": 1.0839992761611938,
      "learning_rate": 0.0003363228034974141,
      "loss": 2.0387,
      "step": 106300
    },
    {
      "epoch": 1.38,
      "grad_norm": 1.0829120874404907,
      "learning_rate": 0.00033631874302515524,
      "loss": 2.2026,
      "step": 106301
    },
    {
      "epoch": 1.38,
      "grad_norm": 1.3272370100021362,
      "learning_rate": 0.000336314682546144,
      "loss": 1.976,
      "step": 106302
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.9773498773574829,
      "learning_rate": 0.0003363106220603811,
      "loss": 2.0556,
      "step": 106303
    },
    {
      "epoch": 1.38,
      "grad_norm": 1.0673209428787231,
      "learning_rate": 0.00033630656156786754,
      "loss": 2.0893,
      "step": 106304
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.9587961435317993,
      "learning_rate": 0.0003363025010686038,
      "loss": 2.2868,
      "step": 106305
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.9586649537086487,
      "learning_rate": 0.0003362984405625907,
      "loss": 2.1403,
      "step": 106306
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.8552802801132202,
      "learning_rate": 0.0003362943800498291,
      "loss": 2.1661,
      "step": 106307
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.8451172113418579,
      "learning_rate": 0.0003362903195303196,
      "loss": 2.4548,
      "step": 106308
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.909299373626709,
      "learning_rate": 0.00033628625900406314,
      "loss": 2.2537,
      "step": 106309
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.8984209895133972,
      "learning_rate": 0.0003362821984710604,
      "loss": 2.1849,
      "step": 106310
    },
    {
      "epoch": 1.38,
      "grad_norm": 1.0138294696807861,
      "learning_rate": 0.00033627813793131206,
      "loss": 2.1422,
      "step": 106311
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.8336928486824036,
      "learning_rate": 0.0003362740773848189,
      "loss": 2.0824,
      "step": 106312
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.8673495054244995,
      "learning_rate": 0.00033627001683158176,
      "loss": 2.2377,
      "step": 106313
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.9111653566360474,
      "learning_rate": 0.0003362659562716013,
      "loss": 2.1152,
      "step": 106314
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.9028632640838623,
      "learning_rate": 0.0003362618957048783,
      "loss": 2.2856,
      "step": 106315
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.8826249241828918,
      "learning_rate": 0.00033625783513141364,
      "loss": 2.1571,
      "step": 106316
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.8686530590057373,
      "learning_rate": 0.0003362537745512079,
      "loss": 2.206,
      "step": 106317
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.9082404971122742,
      "learning_rate": 0.0003362497139642618,
      "loss": 2.1645,
      "step": 106318
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.9614983201026917,
      "learning_rate": 0.0003362456533705764,
      "loss": 2.4223,
      "step": 106319
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.920312225818634,
      "learning_rate": 0.00033624159277015206,
      "loss": 2.1925,
      "step": 106320
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.8076892495155334,
      "learning_rate": 0.00033623753216298975,
      "loss": 2.0155,
      "step": 106321
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.8770982623100281,
      "learning_rate": 0.0003362334715490903,
      "loss": 2.0642,
      "step": 106322
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.9441884756088257,
      "learning_rate": 0.0003362294109284543,
      "loss": 2.1768,
      "step": 106323
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.9219558238983154,
      "learning_rate": 0.0003362253503010826,
      "loss": 2.3113,
      "step": 106324
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.8865523338317871,
      "learning_rate": 0.00033622128966697594,
      "loss": 2.1871,
      "step": 106325
    },
    {
      "epoch": 1.38,
      "grad_norm": 1.0628701448440552,
      "learning_rate": 0.00033621722902613495,
      "loss": 2.2348,
      "step": 106326
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.8256204724311829,
      "learning_rate": 0.0003362131683785606,
      "loss": 2.1714,
      "step": 106327
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.8337538838386536,
      "learning_rate": 0.0003362091077242534,
      "loss": 1.9769,
      "step": 106328
    },
    {
      "epoch": 1.38,
      "grad_norm": 1.0365604162216187,
      "learning_rate": 0.0003362050470632144,
      "loss": 2.3824,
      "step": 106329
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.8468925952911377,
      "learning_rate": 0.0003362009863954441,
      "loss": 1.976,
      "step": 106330
    },
    {
      "epoch": 1.38,
      "grad_norm": 1.0174869298934937,
      "learning_rate": 0.0003361969257209434,
      "loss": 2.3237,
      "step": 106331
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.966748833656311,
      "learning_rate": 0.00033619286503971306,
      "loss": 2.3081,
      "step": 106332
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.8715634346008301,
      "learning_rate": 0.00033618880435175373,
      "loss": 2.1385,
      "step": 106333
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.9200212359428406,
      "learning_rate": 0.00033618474365706616,
      "loss": 2.1203,
      "step": 106334
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.968127965927124,
      "learning_rate": 0.00033618068295565123,
      "loss": 1.9637,
      "step": 106335
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.8672447800636292,
      "learning_rate": 0.0003361766222475096,
      "loss": 2.065,
      "step": 106336
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.9082915186882019,
      "learning_rate": 0.00033617256153264203,
      "loss": 2.0986,
      "step": 106337
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.909589946269989,
      "learning_rate": 0.0003361685008110494,
      "loss": 1.9962,
      "step": 106338
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.9801072478294373,
      "learning_rate": 0.00033616444008273226,
      "loss": 2.0244,
      "step": 106339
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.934058427810669,
      "learning_rate": 0.0003361603793476915,
      "loss": 2.0656,
      "step": 106340
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.8458530306816101,
      "learning_rate": 0.0003361563186059278,
      "loss": 2.368,
      "step": 106341
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.8956889510154724,
      "learning_rate": 0.00033615225785744204,
      "loss": 2.015,
      "step": 106342
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.8922885060310364,
      "learning_rate": 0.00033614819710223485,
      "loss": 2.2225,
      "step": 106343
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.8106686472892761,
      "learning_rate": 0.00033614413634030707,
      "loss": 1.9778,
      "step": 106344
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.9096041917800903,
      "learning_rate": 0.0003361400755716593,
      "loss": 2.082,
      "step": 106345
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.8503653407096863,
      "learning_rate": 0.00033613601479629256,
      "loss": 2.1333,
      "step": 106346
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.8666764497756958,
      "learning_rate": 0.00033613195401420735,
      "loss": 2.1454,
      "step": 106347
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.9637230038642883,
      "learning_rate": 0.00033612789322540453,
      "loss": 2.0569,
      "step": 106348
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.980815052986145,
      "learning_rate": 0.0003361238324298849,
      "loss": 2.0003,
      "step": 106349
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.9446771144866943,
      "learning_rate": 0.00033611977162764916,
      "loss": 2.1936,
      "step": 106350
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.8869243264198303,
      "learning_rate": 0.00033611571081869807,
      "loss": 2.2324,
      "step": 106351
    },
    {
      "epoch": 1.38,
      "grad_norm": 1.1457282304763794,
      "learning_rate": 0.00033611165000303235,
      "loss": 2.11,
      "step": 106352
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.9452008605003357,
      "learning_rate": 0.0003361075891806528,
      "loss": 2.0189,
      "step": 106353
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.8739910125732422,
      "learning_rate": 0.0003361035283515602,
      "loss": 2.201,
      "step": 106354
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.8605028986930847,
      "learning_rate": 0.00033609946751575524,
      "loss": 2.1613,
      "step": 106355
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.9278501868247986,
      "learning_rate": 0.00033609540667323877,
      "loss": 2.0402,
      "step": 106356
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.8580073118209839,
      "learning_rate": 0.0003360913458240114,
      "loss": 2.1378,
      "step": 106357
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.8485433459281921,
      "learning_rate": 0.00033608728496807404,
      "loss": 2.0936,
      "step": 106358
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.9512392282485962,
      "learning_rate": 0.00033608322410542735,
      "loss": 2.2817,
      "step": 106359
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.9152942299842834,
      "learning_rate": 0.0003360791632360721,
      "loss": 2.1931,
      "step": 106360
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.8185677528381348,
      "learning_rate": 0.00033607510236000906,
      "loss": 2.1548,
      "step": 106361
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.8283686637878418,
      "learning_rate": 0.000336071041477239,
      "loss": 2.1555,
      "step": 106362
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.9215417504310608,
      "learning_rate": 0.00033606698058776256,
      "loss": 1.9918,
      "step": 106363
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.8511353135108948,
      "learning_rate": 0.00033606291969158063,
      "loss": 2.2451,
      "step": 106364
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.8826562762260437,
      "learning_rate": 0.000336058858788694,
      "loss": 2.2009,
      "step": 106365
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.8871774673461914,
      "learning_rate": 0.0003360547978791033,
      "loss": 2.0398,
      "step": 106366
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.8399690389633179,
      "learning_rate": 0.00033605073696280926,
      "loss": 2.0006,
      "step": 106367
    },
    {
      "epoch": 1.38,
      "grad_norm": 0.9348657727241516,
      "learning_rate": 0.00033604667603981284,
      "loss": 2.3784,
      "step": 106368
    },
    {
      "epoch": 1.38,
      "grad_norm": 1.1299564838409424,
      "learning_rate": 0.00033604261511011453,
      "loss": 2.0975,
      "step": 106369
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.8740651607513428,
      "learning_rate": 0.0003360385541737153,
      "loss": 2.4498,
      "step": 106370
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.8257160782814026,
      "learning_rate": 0.00033603449323061583,
      "loss": 2.2174,
      "step": 106371
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.8713957071304321,
      "learning_rate": 0.0003360304322808168,
      "loss": 2.2531,
      "step": 106372
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.9885223507881165,
      "learning_rate": 0.00033602637132431907,
      "loss": 2.3028,
      "step": 106373
    },
    {
      "epoch": 1.39,
      "grad_norm": 1.0593652725219727,
      "learning_rate": 0.0003360223103611234,
      "loss": 2.1643,
      "step": 106374
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.9720442295074463,
      "learning_rate": 0.0003360182493912305,
      "loss": 2.0474,
      "step": 106375
    },
    {
      "epoch": 1.39,
      "grad_norm": 1.1429435014724731,
      "learning_rate": 0.00033601418841464097,
      "loss": 2.0126,
      "step": 106376
    },
    {
      "epoch": 1.39,
      "grad_norm": 1.1549104452133179,
      "learning_rate": 0.0003360101274313559,
      "loss": 2.2629,
      "step": 106377
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.9022423624992371,
      "learning_rate": 0.00033600606644137575,
      "loss": 1.8801,
      "step": 106378
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.9739823937416077,
      "learning_rate": 0.0003360020054447015,
      "loss": 2.2111,
      "step": 106379
    },
    {
      "epoch": 1.39,
      "grad_norm": 1.2409460544586182,
      "learning_rate": 0.0003359979444413338,
      "loss": 2.14,
      "step": 106380
    },
    {
      "epoch": 1.39,
      "grad_norm": 1.0513182878494263,
      "learning_rate": 0.0003359938834312733,
      "loss": 2.0915,
      "step": 106381
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.8629159331321716,
      "learning_rate": 0.0003359898224145209,
      "loss": 2.0706,
      "step": 106382
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.993922233581543,
      "learning_rate": 0.00033598576139107737,
      "loss": 2.1179,
      "step": 106383
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.9318490028381348,
      "learning_rate": 0.0003359817003609433,
      "loss": 2.2237,
      "step": 106384
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.9683981537818909,
      "learning_rate": 0.00033597763932411963,
      "loss": 2.0969,
      "step": 106385
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.9255120158195496,
      "learning_rate": 0.00033597357828060704,
      "loss": 2.2028,
      "step": 106386
    },
    {
      "epoch": 1.39,
      "grad_norm": 1.067636251449585,
      "learning_rate": 0.0003359695172304062,
      "loss": 2.0564,
      "step": 106387
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.8092429637908936,
      "learning_rate": 0.00033596545617351804,
      "loss": 2.2775,
      "step": 106388
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.9211950302124023,
      "learning_rate": 0.00033596139510994326,
      "loss": 2.3453,
      "step": 106389
    },
    {
      "epoch": 1.39,
      "grad_norm": 1.0669420957565308,
      "learning_rate": 0.0003359573340396824,
      "loss": 2.0938,
      "step": 106390
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.8329840302467346,
      "learning_rate": 0.00033595327296273656,
      "loss": 1.8752,
      "step": 106391
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.9221535325050354,
      "learning_rate": 0.0003359492118791063,
      "loss": 2.4583,
      "step": 106392
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.8780578374862671,
      "learning_rate": 0.00033594515078879235,
      "loss": 2.152,
      "step": 106393
    },
    {
      "epoch": 1.39,
      "grad_norm": 1.0031254291534424,
      "learning_rate": 0.0003359410896917955,
      "loss": 2.2669,
      "step": 106394
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.8553540706634521,
      "learning_rate": 0.0003359370285881166,
      "loss": 1.9812,
      "step": 106395
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.9462462067604065,
      "learning_rate": 0.0003359329674777563,
      "loss": 2.2051,
      "step": 106396
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.8543885350227356,
      "learning_rate": 0.00033592890636071536,
      "loss": 1.9301,
      "step": 106397
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.9585274457931519,
      "learning_rate": 0.0003359248452369946,
      "loss": 2.1722,
      "step": 106398
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.8205134272575378,
      "learning_rate": 0.0003359207841065947,
      "loss": 2.1239,
      "step": 106399
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.8841438889503479,
      "learning_rate": 0.00033591672296951646,
      "loss": 2.2899,
      "step": 106400
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.8466120362281799,
      "learning_rate": 0.00033591266182576054,
      "loss": 2.251,
      "step": 106401
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.8339555859565735,
      "learning_rate": 0.00033590860067532794,
      "loss": 2.2124,
      "step": 106402
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.8639004230499268,
      "learning_rate": 0.00033590453951821915,
      "loss": 2.3095,
      "step": 106403
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.915500819683075,
      "learning_rate": 0.00033590047835443503,
      "loss": 2.0376,
      "step": 106404
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.912875235080719,
      "learning_rate": 0.0003358964171839764,
      "loss": 2.1674,
      "step": 106405
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.8637531995773315,
      "learning_rate": 0.0003358923560068439,
      "loss": 2.1625,
      "step": 106406
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.8552027940750122,
      "learning_rate": 0.00033588829482303833,
      "loss": 2.2903,
      "step": 106407
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.9365695118904114,
      "learning_rate": 0.00033588423363256045,
      "loss": 1.9939,
      "step": 106408
    },
    {
      "epoch": 1.39,
      "grad_norm": 1.003408670425415,
      "learning_rate": 0.00033588017243541105,
      "loss": 2.247,
      "step": 106409
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.8555105328559875,
      "learning_rate": 0.0003358761112315907,
      "loss": 2.0002,
      "step": 106410
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.8981253504753113,
      "learning_rate": 0.0003358720500211005,
      "loss": 2.2178,
      "step": 106411
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.8543847799301147,
      "learning_rate": 0.0003358679888039409,
      "loss": 1.9949,
      "step": 106412
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.8543574213981628,
      "learning_rate": 0.0003358639275801128,
      "loss": 2.0509,
      "step": 106413
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.915267825126648,
      "learning_rate": 0.0003358598663496169,
      "loss": 2.0282,
      "step": 106414
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.8825504183769226,
      "learning_rate": 0.000335855805112454,
      "loss": 2.1021,
      "step": 106415
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.8733260035514832,
      "learning_rate": 0.0003358517438686248,
      "loss": 1.9947,
      "step": 106416
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.8724581003189087,
      "learning_rate": 0.00033584768261813016,
      "loss": 2.2426,
      "step": 106417
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.847974956035614,
      "learning_rate": 0.00033584362136097065,
      "loss": 2.1519,
      "step": 106418
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.9283368587493896,
      "learning_rate": 0.0003358395600971472,
      "loss": 2.2072,
      "step": 106419
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.9347910284996033,
      "learning_rate": 0.0003358354988266605,
      "loss": 1.9135,
      "step": 106420
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.8933266997337341,
      "learning_rate": 0.00033583143754951125,
      "loss": 1.9806,
      "step": 106421
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.9016793370246887,
      "learning_rate": 0.00033582737626570026,
      "loss": 2.0885,
      "step": 106422
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.8317147493362427,
      "learning_rate": 0.0003358233149752284,
      "loss": 2.135,
      "step": 106423
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.8671600222587585,
      "learning_rate": 0.0003358192536780962,
      "loss": 2.3871,
      "step": 106424
    },
    {
      "epoch": 1.39,
      "grad_norm": 1.01792573928833,
      "learning_rate": 0.0003358151923743045,
      "loss": 2.0672,
      "step": 106425
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.9010617136955261,
      "learning_rate": 0.00033581113106385415,
      "loss": 1.9751,
      "step": 106426
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.8041061758995056,
      "learning_rate": 0.00033580706974674585,
      "loss": 2.1495,
      "step": 106427
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.7859892249107361,
      "learning_rate": 0.0003358030084229803,
      "loss": 1.9515,
      "step": 106428
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.8592151403427124,
      "learning_rate": 0.00033579894709255824,
      "loss": 2.081,
      "step": 106429
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.815028965473175,
      "learning_rate": 0.0003357948857554806,
      "loss": 2.0111,
      "step": 106430
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.9155353307723999,
      "learning_rate": 0.00033579082441174794,
      "loss": 2.0129,
      "step": 106431
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.8512593507766724,
      "learning_rate": 0.0003357867630613611,
      "loss": 2.299,
      "step": 106432
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.9165082573890686,
      "learning_rate": 0.0003357827017043208,
      "loss": 2.4333,
      "step": 106433
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.8854071497917175,
      "learning_rate": 0.00033577864034062786,
      "loss": 2.2433,
      "step": 106434
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.8582013249397278,
      "learning_rate": 0.000335774578970283,
      "loss": 2.2512,
      "step": 106435
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.9152280688285828,
      "learning_rate": 0.0003357705175932869,
      "loss": 2.1324,
      "step": 106436
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.8968582153320312,
      "learning_rate": 0.0003357664562096405,
      "loss": 2.1846,
      "step": 106437
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.8920044898986816,
      "learning_rate": 0.0003357623948193444,
      "loss": 2.0428,
      "step": 106438
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.9406458735466003,
      "learning_rate": 0.00033575833342239936,
      "loss": 2.1721,
      "step": 106439
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.844900906085968,
      "learning_rate": 0.00033575427201880615,
      "loss": 2.2269,
      "step": 106440
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.9418528079986572,
      "learning_rate": 0.00033575021060856564,
      "loss": 2.217,
      "step": 106441
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.8606879115104675,
      "learning_rate": 0.00033574614919167843,
      "loss": 2.146,
      "step": 106442
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.9245173335075378,
      "learning_rate": 0.00033574208776814533,
      "loss": 1.9976,
      "step": 106443
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.9037930369377136,
      "learning_rate": 0.00033573802633796716,
      "loss": 2.3444,
      "step": 106444
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.8192846775054932,
      "learning_rate": 0.0003357339649011445,
      "loss": 2.1339,
      "step": 106445
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.8319515585899353,
      "learning_rate": 0.0003357299034576783,
      "loss": 2.1808,
      "step": 106446
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.8798612952232361,
      "learning_rate": 0.0003357258420075693,
      "loss": 2.161,
      "step": 106447
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.9512109160423279,
      "learning_rate": 0.0003357217805508181,
      "loss": 2.2681,
      "step": 106448
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.8994218111038208,
      "learning_rate": 0.00033571771908742553,
      "loss": 2.1391,
      "step": 106449
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.8716151714324951,
      "learning_rate": 0.00033571365761739247,
      "loss": 2.2112,
      "step": 106450
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.995189368724823,
      "learning_rate": 0.0003357095961407194,
      "loss": 2.1572,
      "step": 106451
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.8170015811920166,
      "learning_rate": 0.0003357055346574074,
      "loss": 2.1121,
      "step": 106452
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.8509278893470764,
      "learning_rate": 0.000335701473167457,
      "loss": 2.0682,
      "step": 106453
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.9842243194580078,
      "learning_rate": 0.00033569741167086903,
      "loss": 2.3495,
      "step": 106454
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.9077950119972229,
      "learning_rate": 0.0003356933501676442,
      "loss": 2.0974,
      "step": 106455
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.8967396020889282,
      "learning_rate": 0.00033568928865778335,
      "loss": 2.0063,
      "step": 106456
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.8880720734596252,
      "learning_rate": 0.0003356852271412872,
      "loss": 2.3054,
      "step": 106457
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.9061763286590576,
      "learning_rate": 0.00033568116561815645,
      "loss": 1.8164,
      "step": 106458
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.802966296672821,
      "learning_rate": 0.0003356771040883919,
      "loss": 1.9641,
      "step": 106459
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.8091400861740112,
      "learning_rate": 0.0003356730425519943,
      "loss": 2.1199,
      "step": 106460
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.8950836658477783,
      "learning_rate": 0.0003356689810089644,
      "loss": 2.2561,
      "step": 106461
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.8004696369171143,
      "learning_rate": 0.000335664919459303,
      "loss": 2.1326,
      "step": 106462
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.8811209797859192,
      "learning_rate": 0.0003356608579030109,
      "loss": 2.228,
      "step": 106463
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.839859127998352,
      "learning_rate": 0.00033565679634008864,
      "loss": 2.2145,
      "step": 106464
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.9814794659614563,
      "learning_rate": 0.00033565273477053714,
      "loss": 2.1305,
      "step": 106465
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.9057345986366272,
      "learning_rate": 0.00033564867319435716,
      "loss": 2.1557,
      "step": 106466
    },
    {
      "epoch": 1.39,
      "grad_norm": 1.0724146366119385,
      "learning_rate": 0.00033564461161154934,
      "loss": 2.0528,
      "step": 106467
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.913506031036377,
      "learning_rate": 0.00033564055002211456,
      "loss": 2.1679,
      "step": 106468
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.9117051362991333,
      "learning_rate": 0.0003356364884260536,
      "loss": 2.1237,
      "step": 106469
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.7920956611633301,
      "learning_rate": 0.00033563242682336706,
      "loss": 2.03,
      "step": 106470
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.8285825848579407,
      "learning_rate": 0.00033562836521405575,
      "loss": 2.0124,
      "step": 106471
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.8564910292625427,
      "learning_rate": 0.00033562430359812057,
      "loss": 2.0517,
      "step": 106472
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.9417871832847595,
      "learning_rate": 0.00033562024197556205,
      "loss": 2.0673,
      "step": 106473
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.9348173141479492,
      "learning_rate": 0.00033561618034638106,
      "loss": 2.278,
      "step": 106474
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.8988676071166992,
      "learning_rate": 0.0003356121187105784,
      "loss": 2.3178,
      "step": 106475
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.9713464975357056,
      "learning_rate": 0.00033560805706815477,
      "loss": 2.0883,
      "step": 106476
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.9133530259132385,
      "learning_rate": 0.0003356039954191109,
      "loss": 2.1953,
      "step": 106477
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.8708692789077759,
      "learning_rate": 0.00033559993376344765,
      "loss": 2.0239,
      "step": 106478
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.9590029120445251,
      "learning_rate": 0.00033559587210116564,
      "loss": 2.1311,
      "step": 106479
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.8676667213439941,
      "learning_rate": 0.00033559181043226567,
      "loss": 2.0759,
      "step": 106480
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.9529253244400024,
      "learning_rate": 0.0003355877487567486,
      "loss": 2.0234,
      "step": 106481
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.857880711555481,
      "learning_rate": 0.000335583687074615,
      "loss": 2.2878,
      "step": 106482
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.8892860412597656,
      "learning_rate": 0.0003355796253858657,
      "loss": 2.2131,
      "step": 106483
    },
    {
      "epoch": 1.39,
      "grad_norm": 1.1306703090667725,
      "learning_rate": 0.00033557556369050157,
      "loss": 2.0417,
      "step": 106484
    },
    {
      "epoch": 1.39,
      "grad_norm": 1.0982118844985962,
      "learning_rate": 0.0003355715019885232,
      "loss": 2.0643,
      "step": 106485
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.8645979762077332,
      "learning_rate": 0.00033556744027993143,
      "loss": 2.2163,
      "step": 106486
    },
    {
      "epoch": 1.39,
      "grad_norm": 1.1077426671981812,
      "learning_rate": 0.000335563378564727,
      "loss": 2.2608,
      "step": 106487
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.9939706325531006,
      "learning_rate": 0.00033555931684291077,
      "loss": 2.1392,
      "step": 106488
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.8901245594024658,
      "learning_rate": 0.0003355552551144832,
      "loss": 1.9342,
      "step": 106489
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.8706710934638977,
      "learning_rate": 0.00033555119337944545,
      "loss": 2.1613,
      "step": 106490
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.9687420129776001,
      "learning_rate": 0.0003355471316377979,
      "loss": 2.1438,
      "step": 106491
    },
    {
      "epoch": 1.39,
      "grad_norm": 1.0467110872268677,
      "learning_rate": 0.00033554306988954147,
      "loss": 2.1952,
      "step": 106492
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.9725613594055176,
      "learning_rate": 0.00033553900813467706,
      "loss": 2.2786,
      "step": 106493
    },
    {
      "epoch": 1.39,
      "grad_norm": 1.0289369821548462,
      "learning_rate": 0.00033553494637320513,
      "loss": 2.1433,
      "step": 106494
    },
    {
      "epoch": 1.39,
      "grad_norm": 1.0226444005966187,
      "learning_rate": 0.0003355308846051266,
      "loss": 1.976,
      "step": 106495
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.9093486666679382,
      "learning_rate": 0.0003355268228304423,
      "loss": 2.1044,
      "step": 106496
    },
    {
      "epoch": 1.39,
      "grad_norm": 1.004883885383606,
      "learning_rate": 0.00033552276104915283,
      "loss": 2.2124,
      "step": 106497
    },
    {
      "epoch": 1.39,
      "grad_norm": 1.2221792936325073,
      "learning_rate": 0.00033551869926125896,
      "loss": 2.107,
      "step": 106498
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.873051106929779,
      "learning_rate": 0.0003355146374667616,
      "loss": 2.4286,
      "step": 106499
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.992326557636261,
      "learning_rate": 0.0003355105756656613,
      "loss": 2.2208,
      "step": 106500
    },
    {
      "epoch": 1.39,
      "grad_norm": 1.049128770828247,
      "learning_rate": 0.00033550651385795896,
      "loss": 2.1002,
      "step": 106501
    },
    {
      "epoch": 1.39,
      "grad_norm": 1.0474328994750977,
      "learning_rate": 0.00033550245204365533,
      "loss": 2.1531,
      "step": 106502
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.8806153535842896,
      "learning_rate": 0.000335498390222751,
      "loss": 2.3405,
      "step": 106503
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.9781337976455688,
      "learning_rate": 0.00033549432839524696,
      "loss": 2.3082,
      "step": 106504
    },
    {
      "epoch": 1.39,
      "grad_norm": 1.270401120185852,
      "learning_rate": 0.0003354902665611439,
      "loss": 2.0388,
      "step": 106505
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.9895059466362,
      "learning_rate": 0.00033548620472044244,
      "loss": 2.0325,
      "step": 106506
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.9217440485954285,
      "learning_rate": 0.0003354821428731434,
      "loss": 2.245,
      "step": 106507
    },
    {
      "epoch": 1.39,
      "grad_norm": 1.151620626449585,
      "learning_rate": 0.00033547808101924763,
      "loss": 2.0026,
      "step": 106508
    },
    {
      "epoch": 1.39,
      "grad_norm": 1.0495786666870117,
      "learning_rate": 0.00033547401915875574,
      "loss": 1.9451,
      "step": 106509
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.8520181179046631,
      "learning_rate": 0.0003354699572916686,
      "loss": 2.0307,
      "step": 106510
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.9773613214492798,
      "learning_rate": 0.000335465895417987,
      "loss": 2.1239,
      "step": 106511
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.8906951546669006,
      "learning_rate": 0.00033546183353771146,
      "loss": 2.1345,
      "step": 106512
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.9218320250511169,
      "learning_rate": 0.000335457771650843,
      "loss": 2.1285,
      "step": 106513
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.8134576678276062,
      "learning_rate": 0.0003354537097573822,
      "loss": 2.1647,
      "step": 106514
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.9049282670021057,
      "learning_rate": 0.00033544964785733,
      "loss": 2.3206,
      "step": 106515
    },
    {
      "epoch": 1.39,
      "grad_norm": 1.3443243503570557,
      "learning_rate": 0.0003354455859506869,
      "loss": 1.9591,
      "step": 106516
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.8333304524421692,
      "learning_rate": 0.0003354415240374539,
      "loss": 1.9535,
      "step": 106517
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.8467863202095032,
      "learning_rate": 0.00033543746211763166,
      "loss": 2.175,
      "step": 106518
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.9028959274291992,
      "learning_rate": 0.0003354334001912209,
      "loss": 2.2702,
      "step": 106519
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.8132104873657227,
      "learning_rate": 0.00033542933825822237,
      "loss": 1.9817,
      "step": 106520
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.924400269985199,
      "learning_rate": 0.0003354252763186369,
      "loss": 2.0084,
      "step": 106521
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.9679450988769531,
      "learning_rate": 0.00033542121437246517,
      "loss": 2.2753,
      "step": 106522
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.9251593947410583,
      "learning_rate": 0.00033541715241970806,
      "loss": 2.0479,
      "step": 106523
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.8136669993400574,
      "learning_rate": 0.0003354130904603661,
      "loss": 2.1159,
      "step": 106524
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.9017419219017029,
      "learning_rate": 0.0003354090284944402,
      "loss": 2.1956,
      "step": 106525
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.9679194688796997,
      "learning_rate": 0.0003354049665219312,
      "loss": 2.0516,
      "step": 106526
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.9387909770011902,
      "learning_rate": 0.00033540090454283963,
      "loss": 2.1762,
      "step": 106527
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.8849808573722839,
      "learning_rate": 0.0003353968425571664,
      "loss": 2.2382,
      "step": 106528
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.8555554747581482,
      "learning_rate": 0.00033539278056491224,
      "loss": 2.3985,
      "step": 106529
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.869078516960144,
      "learning_rate": 0.0003353887185660779,
      "loss": 2.2351,
      "step": 106530
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.8670134544372559,
      "learning_rate": 0.0003353846565606641,
      "loss": 2.1489,
      "step": 106531
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.8570722341537476,
      "learning_rate": 0.0003353805945486716,
      "loss": 1.8864,
      "step": 106532
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.9107595086097717,
      "learning_rate": 0.0003353765325301013,
      "loss": 2.1704,
      "step": 106533
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.8944588303565979,
      "learning_rate": 0.00033537247050495375,
      "loss": 2.2435,
      "step": 106534
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.8438491821289062,
      "learning_rate": 0.00033536840847322975,
      "loss": 2.1505,
      "step": 106535
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.9135357141494751,
      "learning_rate": 0.0003353643464349302,
      "loss": 2.0423,
      "step": 106536
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.8713518381118774,
      "learning_rate": 0.0003353602843900557,
      "loss": 2.2069,
      "step": 106537
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.873293399810791,
      "learning_rate": 0.00033535622233860703,
      "loss": 2.159,
      "step": 106538
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.9099924564361572,
      "learning_rate": 0.0003353521602805851,
      "loss": 2.0758,
      "step": 106539
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.969840407371521,
      "learning_rate": 0.0003353480982159904,
      "loss": 2.4189,
      "step": 106540
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.8720968961715698,
      "learning_rate": 0.0003353440361448238,
      "loss": 2.0879,
      "step": 106541
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.9241102933883667,
      "learning_rate": 0.0003353399740670862,
      "loss": 2.2609,
      "step": 106542
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.9473528861999512,
      "learning_rate": 0.0003353359119827781,
      "loss": 1.9869,
      "step": 106543
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.8482012152671814,
      "learning_rate": 0.0003353318498919005,
      "loss": 1.9644,
      "step": 106544
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.9225366711616516,
      "learning_rate": 0.000335327787794454,
      "loss": 2.1209,
      "step": 106545
    },
    {
      "epoch": 1.39,
      "grad_norm": 1.0463182926177979,
      "learning_rate": 0.0003353237256904394,
      "loss": 2.1836,
      "step": 106546
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.849282443523407,
      "learning_rate": 0.00033531966357985734,
      "loss": 1.8867,
      "step": 106547
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.9045788645744324,
      "learning_rate": 0.0003353156014627089,
      "loss": 2.0506,
      "step": 106548
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.8832208514213562,
      "learning_rate": 0.0003353115393389944,
      "loss": 2.3775,
      "step": 106549
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.9894400835037231,
      "learning_rate": 0.00033530747720871495,
      "loss": 2.0464,
      "step": 106550
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.8543254137039185,
      "learning_rate": 0.0003353034150718712,
      "loss": 2.014,
      "step": 106551
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.9067248106002808,
      "learning_rate": 0.0003352993529284638,
      "loss": 2.1891,
      "step": 106552
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.8612204194068909,
      "learning_rate": 0.0003352952907784936,
      "loss": 2.2049,
      "step": 106553
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.9842202663421631,
      "learning_rate": 0.0003352912286219614,
      "loss": 2.2772,
      "step": 106554
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.9037559032440186,
      "learning_rate": 0.0003352871664588678,
      "loss": 2.1954,
      "step": 106555
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.8871421813964844,
      "learning_rate": 0.0003352831042892137,
      "loss": 2.0169,
      "step": 106556
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.9094529747962952,
      "learning_rate": 0.00033527904211299984,
      "loss": 2.3274,
      "step": 106557
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.7803496718406677,
      "learning_rate": 0.00033527497993022685,
      "loss": 2.1123,
      "step": 106558
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.9055166840553284,
      "learning_rate": 0.00033527091774089563,
      "loss": 2.1673,
      "step": 106559
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.850213348865509,
      "learning_rate": 0.000335266855545007,
      "loss": 2.1616,
      "step": 106560
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.892891526222229,
      "learning_rate": 0.0003352627933425614,
      "loss": 2.0239,
      "step": 106561
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.9136143922805786,
      "learning_rate": 0.00033525873113355984,
      "loss": 2.192,
      "step": 106562
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.8256438970565796,
      "learning_rate": 0.00033525466891800307,
      "loss": 2.2232,
      "step": 106563
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.8922789096832275,
      "learning_rate": 0.00033525060669589176,
      "loss": 2.0773,
      "step": 106564
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.8712252378463745,
      "learning_rate": 0.00033524654446722666,
      "loss": 2.4257,
      "step": 106565
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.8922476768493652,
      "learning_rate": 0.00033524248223200864,
      "loss": 2.343,
      "step": 106566
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.8450016379356384,
      "learning_rate": 0.00033523841999023835,
      "loss": 2.0883,
      "step": 106567
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.8312814831733704,
      "learning_rate": 0.00033523435774191645,
      "loss": 1.8919,
      "step": 106568
    },
    {
      "epoch": 1.39,
      "grad_norm": 1.0055525302886963,
      "learning_rate": 0.000335230295487044,
      "loss": 2.3477,
      "step": 106569
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.8967122435569763,
      "learning_rate": 0.0003352262332256215,
      "loss": 2.1031,
      "step": 106570
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.8916680812835693,
      "learning_rate": 0.0003352221709576497,
      "loss": 2.0816,
      "step": 106571
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.8864608407020569,
      "learning_rate": 0.0003352181086831296,
      "loss": 2.1925,
      "step": 106572
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.9398401379585266,
      "learning_rate": 0.00033521404640206163,
      "loss": 2.1622,
      "step": 106573
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.8858940601348877,
      "learning_rate": 0.0003352099841144468,
      "loss": 2.0511,
      "step": 106574
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.9370513558387756,
      "learning_rate": 0.0003352059218202858,
      "loss": 2.198,
      "step": 106575
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.9786795973777771,
      "learning_rate": 0.0003352018595195792,
      "loss": 2.1646,
      "step": 106576
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.8501338958740234,
      "learning_rate": 0.000335197797212328,
      "loss": 2.0733,
      "step": 106577
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.843731164932251,
      "learning_rate": 0.0003351937348985329,
      "loss": 2.1626,
      "step": 106578
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.9520155191421509,
      "learning_rate": 0.00033518967257819453,
      "loss": 2.0155,
      "step": 106579
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.8818060159683228,
      "learning_rate": 0.00033518561025131383,
      "loss": 2.0743,
      "step": 106580
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.8467785716056824,
      "learning_rate": 0.00033518154791789144,
      "loss": 2.099,
      "step": 106581
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.854709804058075,
      "learning_rate": 0.00033517748557792816,
      "loss": 1.9993,
      "step": 106582
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.8824769258499146,
      "learning_rate": 0.0003351734232314246,
      "loss": 2.1496,
      "step": 106583
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.9584268927574158,
      "learning_rate": 0.00033516936087838177,
      "loss": 2.2142,
      "step": 106584
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.8507466316223145,
      "learning_rate": 0.0003351652985188002,
      "loss": 2.353,
      "step": 106585
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.8640273809432983,
      "learning_rate": 0.0003351612361526808,
      "loss": 2.0385,
      "step": 106586
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.9593188166618347,
      "learning_rate": 0.00033515717378002427,
      "loss": 2.0833,
      "step": 106587
    },
    {
      "epoch": 1.39,
      "grad_norm": 1.0161161422729492,
      "learning_rate": 0.00033515311140083125,
      "loss": 2.0455,
      "step": 106588
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.8343182802200317,
      "learning_rate": 0.0003351490490151027,
      "loss": 2.1916,
      "step": 106589
    },
    {
      "epoch": 1.39,
      "grad_norm": 1.0225244760513306,
      "learning_rate": 0.0003351449866228393,
      "loss": 2.2507,
      "step": 106590
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.9668446779251099,
      "learning_rate": 0.00033514092422404175,
      "loss": 2.2215,
      "step": 106591
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.8224474191665649,
      "learning_rate": 0.0003351368618187108,
      "loss": 2.1127,
      "step": 106592
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.9626038074493408,
      "learning_rate": 0.0003351327994068473,
      "loss": 2.2947,
      "step": 106593
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.9949658513069153,
      "learning_rate": 0.00033512873698845195,
      "loss": 2.2097,
      "step": 106594
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.8771011233329773,
      "learning_rate": 0.00033512467456352544,
      "loss": 2.0138,
      "step": 106595
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.9972328543663025,
      "learning_rate": 0.0003351206121320686,
      "loss": 2.2778,
      "step": 106596
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.9448290467262268,
      "learning_rate": 0.00033511654969408224,
      "loss": 2.0088,
      "step": 106597
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.8555278182029724,
      "learning_rate": 0.000335112487249567,
      "loss": 1.9604,
      "step": 106598
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.9080071449279785,
      "learning_rate": 0.0003351084247985237,
      "loss": 2.0671,
      "step": 106599
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.903767466545105,
      "learning_rate": 0.0003351043623409531,
      "loss": 2.0829,
      "step": 106600
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.9497079849243164,
      "learning_rate": 0.0003351002998768559,
      "loss": 2.1233,
      "step": 106601
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.8942360877990723,
      "learning_rate": 0.00033509623740623286,
      "loss": 2.174,
      "step": 106602
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.9141138195991516,
      "learning_rate": 0.00033509217492908494,
      "loss": 2.27,
      "step": 106603
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.9384035468101501,
      "learning_rate": 0.00033508811244541256,
      "loss": 2.0636,
      "step": 106604
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.9066590070724487,
      "learning_rate": 0.00033508404995521665,
      "loss": 2.1845,
      "step": 106605
    },
    {
      "epoch": 1.39,
      "grad_norm": 1.0336304903030396,
      "learning_rate": 0.00033507998745849795,
      "loss": 2.1683,
      "step": 106606
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.8889484405517578,
      "learning_rate": 0.00033507592495525724,
      "loss": 2.0972,
      "step": 106607
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.9078130125999451,
      "learning_rate": 0.0003350718624454953,
      "loss": 2.2473,
      "step": 106608
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.842703640460968,
      "learning_rate": 0.00033506779992921275,
      "loss": 2.1015,
      "step": 106609
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.9741955399513245,
      "learning_rate": 0.00033506373740641054,
      "loss": 2.1962,
      "step": 106610
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.8737365007400513,
      "learning_rate": 0.00033505967487708924,
      "loss": 2.0781,
      "step": 106611
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.934766411781311,
      "learning_rate": 0.0003350556123412497,
      "loss": 2.3115,
      "step": 106612
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.8527297377586365,
      "learning_rate": 0.00033505154979889267,
      "loss": 2.2148,
      "step": 106613
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.8427301049232483,
      "learning_rate": 0.00033504748725001886,
      "loss": 2.1847,
      "step": 106614
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.863849401473999,
      "learning_rate": 0.0003350434246946291,
      "loss": 2.1385,
      "step": 106615
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.8713822364807129,
      "learning_rate": 0.0003350393621327241,
      "loss": 2.1941,
      "step": 106616
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.9391623735427856,
      "learning_rate": 0.00033503529956430464,
      "loss": 2.3441,
      "step": 106617
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.8954176902770996,
      "learning_rate": 0.00033503123698937144,
      "loss": 1.9497,
      "step": 106618
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.8188630938529968,
      "learning_rate": 0.0003350271744079253,
      "loss": 2.3856,
      "step": 106619
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.8712759017944336,
      "learning_rate": 0.0003350231118199669,
      "loss": 1.9798,
      "step": 106620
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.9704671502113342,
      "learning_rate": 0.00033501904922549704,
      "loss": 2.1329,
      "step": 106621
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.9096572995185852,
      "learning_rate": 0.0003350149866245165,
      "loss": 2.0495,
      "step": 106622
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.9044073820114136,
      "learning_rate": 0.000335010924017026,
      "loss": 2.0301,
      "step": 106623
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.8927881121635437,
      "learning_rate": 0.0003350068614030263,
      "loss": 2.1479,
      "step": 106624
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.8373247981071472,
      "learning_rate": 0.0003350027987825182,
      "loss": 2.0536,
      "step": 106625
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.8643741607666016,
      "learning_rate": 0.00033499873615550237,
      "loss": 2.1651,
      "step": 106626
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.8621935844421387,
      "learning_rate": 0.00033499467352197975,
      "loss": 2.2001,
      "step": 106627
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.8507304787635803,
      "learning_rate": 0.0003349906108819508,
      "loss": 2.1418,
      "step": 106628
    },
    {
      "epoch": 1.39,
      "grad_norm": 1.0220433473587036,
      "learning_rate": 0.0003349865482354165,
      "loss": 2.3402,
      "step": 106629
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.7587164640426636,
      "learning_rate": 0.00033498248558237753,
      "loss": 1.9252,
      "step": 106630
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.9633629322052002,
      "learning_rate": 0.00033497842292283466,
      "loss": 1.9596,
      "step": 106631
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.9870012402534485,
      "learning_rate": 0.00033497436025678863,
      "loss": 2.1751,
      "step": 106632
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.8555634617805481,
      "learning_rate": 0.0003349702975842402,
      "loss": 2.105,
      "step": 106633
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.8579099774360657,
      "learning_rate": 0.0003349662349051901,
      "loss": 1.9823,
      "step": 106634
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.7645847201347351,
      "learning_rate": 0.0003349621722196392,
      "loss": 2.3216,
      "step": 106635
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.8707054853439331,
      "learning_rate": 0.00033495810952758817,
      "loss": 2.2945,
      "step": 106636
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.8730872273445129,
      "learning_rate": 0.0003349540468290377,
      "loss": 2.0615,
      "step": 106637
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.9218773245811462,
      "learning_rate": 0.0003349499841239886,
      "loss": 2.3182,
      "step": 106638
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.8323825001716614,
      "learning_rate": 0.00033494592141244175,
      "loss": 2.0501,
      "step": 106639
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.8796305060386658,
      "learning_rate": 0.0003349418586943977,
      "loss": 1.7849,
      "step": 106640
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.8843153119087219,
      "learning_rate": 0.00033493779596985735,
      "loss": 2.2608,
      "step": 106641
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.8938025236129761,
      "learning_rate": 0.0003349337332388214,
      "loss": 2.2604,
      "step": 106642
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.8945969939231873,
      "learning_rate": 0.00033492967050129053,
      "loss": 2.3982,
      "step": 106643
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.8648959398269653,
      "learning_rate": 0.0003349256077572656,
      "loss": 2.2798,
      "step": 106644
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.9119892716407776,
      "learning_rate": 0.0003349215450067475,
      "loss": 2.2356,
      "step": 106645
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.8070598840713501,
      "learning_rate": 0.0003349174822497366,
      "loss": 2.2247,
      "step": 106646
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.8240814805030823,
      "learning_rate": 0.00033491341948623394,
      "loss": 1.9961,
      "step": 106647
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.963478684425354,
      "learning_rate": 0.00033490935671624036,
      "loss": 2.1269,
      "step": 106648
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.8849138021469116,
      "learning_rate": 0.00033490529393975636,
      "loss": 2.1261,
      "step": 106649
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.9396818280220032,
      "learning_rate": 0.00033490123115678276,
      "loss": 2.1791,
      "step": 106650
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.8020410537719727,
      "learning_rate": 0.0003348971683673205,
      "loss": 1.7673,
      "step": 106651
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.8780354261398315,
      "learning_rate": 0.00033489310557137007,
      "loss": 2.0417,
      "step": 106652
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.9407936334609985,
      "learning_rate": 0.0003348890427689324,
      "loss": 2.2027,
      "step": 106653
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.9679338932037354,
      "learning_rate": 0.00033488497996000824,
      "loss": 2.1648,
      "step": 106654
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.9652184844017029,
      "learning_rate": 0.0003348809171445982,
      "loss": 2.2242,
      "step": 106655
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.8253530263900757,
      "learning_rate": 0.0003348768543227032,
      "loss": 2.0806,
      "step": 106656
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.932854950428009,
      "learning_rate": 0.00033487279149432397,
      "loss": 2.068,
      "step": 106657
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.9345899820327759,
      "learning_rate": 0.00033486872865946116,
      "loss": 2.0929,
      "step": 106658
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.865402102470398,
      "learning_rate": 0.0003348646658181156,
      "loss": 2.1028,
      "step": 106659
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.8315898776054382,
      "learning_rate": 0.0003348606029702881,
      "loss": 1.8311,
      "step": 106660
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.9474422931671143,
      "learning_rate": 0.0003348565401159793,
      "loss": 2.2729,
      "step": 106661
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.8598122596740723,
      "learning_rate": 0.00033485247725519,
      "loss": 2.0117,
      "step": 106662
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.9633011221885681,
      "learning_rate": 0.000334848414387921,
      "loss": 2.1036,
      "step": 106663
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.9683293700218201,
      "learning_rate": 0.0003348443515141731,
      "loss": 2.1874,
      "step": 106664
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.8532314300537109,
      "learning_rate": 0.00033484028863394686,
      "loss": 2.1145,
      "step": 106665
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.8342334032058716,
      "learning_rate": 0.0003348362257472432,
      "loss": 2.0554,
      "step": 106666
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.9164614677429199,
      "learning_rate": 0.00033483216285406283,
      "loss": 2.3184,
      "step": 106667
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.8895350694656372,
      "learning_rate": 0.0003348280999544065,
      "loss": 1.9344,
      "step": 106668
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.8726590871810913,
      "learning_rate": 0.00033482403704827493,
      "loss": 2.1296,
      "step": 106669
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.8782938122749329,
      "learning_rate": 0.000334819974135669,
      "loss": 2.0392,
      "step": 106670
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.9490272998809814,
      "learning_rate": 0.0003348159112165893,
      "loss": 1.9921,
      "step": 106671
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.8968340754508972,
      "learning_rate": 0.00033481184829103666,
      "loss": 2.2475,
      "step": 106672
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.8699517846107483,
      "learning_rate": 0.00033480778535901194,
      "loss": 1.9146,
      "step": 106673
    },
    {
      "epoch": 1.39,
      "grad_norm": 1.3316454887390137,
      "learning_rate": 0.0003348037224205157,
      "loss": 2.1545,
      "step": 106674
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.8567380309104919,
      "learning_rate": 0.0003347996594755488,
      "loss": 2.0248,
      "step": 106675
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.9438401460647583,
      "learning_rate": 0.0003347955965241121,
      "loss": 2.1242,
      "step": 106676
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.8931201696395874,
      "learning_rate": 0.0003347915335662061,
      "loss": 2.2883,
      "step": 106677
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.8373270630836487,
      "learning_rate": 0.00033478747060183166,
      "loss": 2.0982,
      "step": 106678
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.9169008731842041,
      "learning_rate": 0.0003347834076309898,
      "loss": 2.1057,
      "step": 106679
    },
    {
      "epoch": 1.39,
      "grad_norm": 1.0409507751464844,
      "learning_rate": 0.00033477934465368087,
      "loss": 2.2292,
      "step": 106680
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.8400971293449402,
      "learning_rate": 0.00033477528166990576,
      "loss": 2.1299,
      "step": 106681
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.8509631752967834,
      "learning_rate": 0.00033477121867966545,
      "loss": 2.1898,
      "step": 106682
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.8871915340423584,
      "learning_rate": 0.00033476715568296037,
      "loss": 2.1846,
      "step": 106683
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.8816994428634644,
      "learning_rate": 0.0003347630926797915,
      "loss": 2.2682,
      "step": 106684
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.8632510900497437,
      "learning_rate": 0.00033475902967015946,
      "loss": 2.1734,
      "step": 106685
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.9038241505622864,
      "learning_rate": 0.00033475496665406505,
      "loss": 2.2594,
      "step": 106686
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.8751286268234253,
      "learning_rate": 0.000334750903631509,
      "loss": 2.1726,
      "step": 106687
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.9528641104698181,
      "learning_rate": 0.00033474684060249226,
      "loss": 2.0942,
      "step": 106688
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.9722926020622253,
      "learning_rate": 0.00033474277756701535,
      "loss": 2.0572,
      "step": 106689
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.8967514634132385,
      "learning_rate": 0.00033473871452507903,
      "loss": 2.3318,
      "step": 106690
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.9092307090759277,
      "learning_rate": 0.0003347346514766842,
      "loss": 2.226,
      "step": 106691
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.8794898986816406,
      "learning_rate": 0.0003347305884218315,
      "loss": 2.2235,
      "step": 106692
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.9357765316963196,
      "learning_rate": 0.0003347265253605217,
      "loss": 2.1197,
      "step": 106693
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.9463691711425781,
      "learning_rate": 0.00033472246229275564,
      "loss": 2.042,
      "step": 106694
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.850296676158905,
      "learning_rate": 0.0003347183992185341,
      "loss": 2.1187,
      "step": 106695
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.8542408347129822,
      "learning_rate": 0.0003347143361378576,
      "loss": 2.1072,
      "step": 106696
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.9516465067863464,
      "learning_rate": 0.0003347102730507271,
      "loss": 1.7946,
      "step": 106697
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.9984957575798035,
      "learning_rate": 0.00033470620995714335,
      "loss": 2.266,
      "step": 106698
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.9154533743858337,
      "learning_rate": 0.000334702146857107,
      "loss": 2.0289,
      "step": 106699
    },
    {
      "epoch": 1.39,
      "grad_norm": 1.1241071224212646,
      "learning_rate": 0.0003346980837506189,
      "loss": 2.1622,
      "step": 106700
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.9661675691604614,
      "learning_rate": 0.00033469402063767973,
      "loss": 2.3279,
      "step": 106701
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.8750805258750916,
      "learning_rate": 0.0003346899575182904,
      "loss": 2.1056,
      "step": 106702
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.9250605702400208,
      "learning_rate": 0.0003346858943924514,
      "loss": 2.2279,
      "step": 106703
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.8148314952850342,
      "learning_rate": 0.0003346818312601637,
      "loss": 2.4134,
      "step": 106704
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.8567646741867065,
      "learning_rate": 0.0003346777681214281,
      "loss": 2.0639,
      "step": 106705
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.8326691389083862,
      "learning_rate": 0.00033467370497624514,
      "loss": 2.0082,
      "step": 106706
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.8928850293159485,
      "learning_rate": 0.00033466964182461565,
      "loss": 2.3826,
      "step": 106707
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.9798686504364014,
      "learning_rate": 0.00033466557866654046,
      "loss": 2.0975,
      "step": 106708
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.8829858899116516,
      "learning_rate": 0.00033466151550202034,
      "loss": 2.2479,
      "step": 106709
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.8386214971542358,
      "learning_rate": 0.00033465745233105594,
      "loss": 2.0648,
      "step": 106710
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.8711012601852417,
      "learning_rate": 0.000334653389153648,
      "loss": 2.1606,
      "step": 106711
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.9890093207359314,
      "learning_rate": 0.0003346493259697975,
      "loss": 2.1595,
      "step": 106712
    },
    {
      "epoch": 1.39,
      "grad_norm": 1.017871618270874,
      "learning_rate": 0.00033464526277950487,
      "loss": 2.4436,
      "step": 106713
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.9587106704711914,
      "learning_rate": 0.00033464119958277115,
      "loss": 2.1035,
      "step": 106714
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.9174647927284241,
      "learning_rate": 0.00033463713637959695,
      "loss": 2.3033,
      "step": 106715
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.9056066870689392,
      "learning_rate": 0.00033463307316998307,
      "loss": 2.274,
      "step": 106716
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.9626824855804443,
      "learning_rate": 0.0003346290099539302,
      "loss": 2.2464,
      "step": 106717
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.8208815455436707,
      "learning_rate": 0.0003346249467314392,
      "loss": 2.1708,
      "step": 106718
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.9183309078216553,
      "learning_rate": 0.0003346208835025106,
      "loss": 2.0154,
      "step": 106719
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.8787531852722168,
      "learning_rate": 0.0003346168202671455,
      "loss": 2.0616,
      "step": 106720
    },
    {
      "epoch": 1.39,
      "grad_norm": 1.071471929550171,
      "learning_rate": 0.00033461275702534446,
      "loss": 2.1458,
      "step": 106721
    },
    {
      "epoch": 1.39,
      "grad_norm": 1.0526504516601562,
      "learning_rate": 0.00033460869377710816,
      "loss": 2.2868,
      "step": 106722
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.9133713841438293,
      "learning_rate": 0.0003346046305224375,
      "loss": 1.9282,
      "step": 106723
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.9253136515617371,
      "learning_rate": 0.0003346005672613333,
      "loss": 2.1247,
      "step": 106724
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.9057279825210571,
      "learning_rate": 0.00033459650399379607,
      "loss": 2.1488,
      "step": 106725
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.8961867094039917,
      "learning_rate": 0.0003345924407198267,
      "loss": 2.0253,
      "step": 106726
    },
    {
      "epoch": 1.39,
      "grad_norm": 1.0248692035675049,
      "learning_rate": 0.000334588377439426,
      "loss": 1.9754,
      "step": 106727
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.972626268863678,
      "learning_rate": 0.00033458431415259464,
      "loss": 2.2538,
      "step": 106728
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.8322595953941345,
      "learning_rate": 0.00033458025085933337,
      "loss": 2.0881,
      "step": 106729
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.9466055631637573,
      "learning_rate": 0.000334576187559643,
      "loss": 2.0938,
      "step": 106730
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.9609929919242859,
      "learning_rate": 0.0003345721242535243,
      "loss": 2.2481,
      "step": 106731
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.8520897626876831,
      "learning_rate": 0.00033456806094097794,
      "loss": 2.1232,
      "step": 106732
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.9576174020767212,
      "learning_rate": 0.00033456399762200477,
      "loss": 2.0905,
      "step": 106733
    },
    {
      "epoch": 1.39,
      "grad_norm": 1.380995273590088,
      "learning_rate": 0.00033455993429660553,
      "loss": 2.2814,
      "step": 106734
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.9027932286262512,
      "learning_rate": 0.00033455587096478087,
      "loss": 2.3538,
      "step": 106735
    },
    {
      "epoch": 1.39,
      "grad_norm": 1.0261818170547485,
      "learning_rate": 0.0003345518076265316,
      "loss": 2.2887,
      "step": 106736
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.7897863388061523,
      "learning_rate": 0.0003345477442818586,
      "loss": 2.2009,
      "step": 106737
    },
    {
      "epoch": 1.39,
      "grad_norm": 1.0287960767745972,
      "learning_rate": 0.0003345436809307625,
      "loss": 2.2001,
      "step": 106738
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.8990294933319092,
      "learning_rate": 0.00033453961757324397,
      "loss": 2.1841,
      "step": 106739
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.8656529188156128,
      "learning_rate": 0.000334535554209304,
      "loss": 2.1914,
      "step": 106740
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.9570363163948059,
      "learning_rate": 0.0003345314908389431,
      "loss": 2.0582,
      "step": 106741
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.9003979563713074,
      "learning_rate": 0.00033452742746216223,
      "loss": 2.3103,
      "step": 106742
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.9212514758110046,
      "learning_rate": 0.0003345233640789621,
      "loss": 2.1323,
      "step": 106743
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.8846253752708435,
      "learning_rate": 0.0003345193006893433,
      "loss": 2.0064,
      "step": 106744
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.8412187099456787,
      "learning_rate": 0.0003345152372933068,
      "loss": 2.3078,
      "step": 106745
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.9227560758590698,
      "learning_rate": 0.0003345111738908533,
      "loss": 2.3193,
      "step": 106746
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.9059765934944153,
      "learning_rate": 0.0003345071104819834,
      "loss": 2.0881,
      "step": 106747
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.97469162940979,
      "learning_rate": 0.0003345030470666981,
      "loss": 2.3091,
      "step": 106748
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.908847987651825,
      "learning_rate": 0.00033449898364499794,
      "loss": 2.1336,
      "step": 106749
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.8107373118400574,
      "learning_rate": 0.0003344949202168838,
      "loss": 2.0493,
      "step": 106750
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.9470014572143555,
      "learning_rate": 0.0003344908567823564,
      "loss": 2.1126,
      "step": 106751
    },
    {
      "epoch": 1.39,
      "grad_norm": 1.115805745124817,
      "learning_rate": 0.0003344867933414165,
      "loss": 2.3308,
      "step": 106752
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.9479139447212219,
      "learning_rate": 0.0003344827298940648,
      "loss": 2.1705,
      "step": 106753
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.9686686992645264,
      "learning_rate": 0.0003344786664403022,
      "loss": 2.1335,
      "step": 106754
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.9726681709289551,
      "learning_rate": 0.00033447460298012934,
      "loss": 2.249,
      "step": 106755
    },
    {
      "epoch": 1.39,
      "grad_norm": 2.0728163719177246,
      "learning_rate": 0.000334470539513547,
      "loss": 2.109,
      "step": 106756
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.9363278150558472,
      "learning_rate": 0.00033446647604055594,
      "loss": 2.325,
      "step": 106757
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.9477719068527222,
      "learning_rate": 0.0003344624125611569,
      "loss": 2.1262,
      "step": 106758
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.9173376560211182,
      "learning_rate": 0.00033445834907535064,
      "loss": 2.0899,
      "step": 106759
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.852450966835022,
      "learning_rate": 0.00033445428558313796,
      "loss": 2.0789,
      "step": 106760
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.864142656326294,
      "learning_rate": 0.00033445022208451953,
      "loss": 2.3464,
      "step": 106761
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.9374392032623291,
      "learning_rate": 0.0003344461585794961,
      "loss": 2.1391,
      "step": 106762
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.8275817036628723,
      "learning_rate": 0.0003344420950680686,
      "loss": 2.1234,
      "step": 106763
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.8641891479492188,
      "learning_rate": 0.00033443803155023763,
      "loss": 2.2011,
      "step": 106764
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.837752103805542,
      "learning_rate": 0.00033443396802600395,
      "loss": 2.1618,
      "step": 106765
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.8625434637069702,
      "learning_rate": 0.0003344299044953683,
      "loss": 2.0081,
      "step": 106766
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.949372410774231,
      "learning_rate": 0.00033442584095833163,
      "loss": 2.2912,
      "step": 106767
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.8721356391906738,
      "learning_rate": 0.00033442177741489445,
      "loss": 1.8422,
      "step": 106768
    },
    {
      "epoch": 1.39,
      "grad_norm": 1.0994786024093628,
      "learning_rate": 0.00033441771386505755,
      "loss": 2.1923,
      "step": 106769
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.9780049324035645,
      "learning_rate": 0.00033441365030882186,
      "loss": 2.2627,
      "step": 106770
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.8553688526153564,
      "learning_rate": 0.000334409586746188,
      "loss": 2.2138,
      "step": 106771
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.8408094048500061,
      "learning_rate": 0.0003344055231771567,
      "loss": 1.9964,
      "step": 106772
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.8948525190353394,
      "learning_rate": 0.0003344014596017288,
      "loss": 2.191,
      "step": 106773
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.8320726752281189,
      "learning_rate": 0.000334397396019905,
      "loss": 2.0335,
      "step": 106774
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.8960140943527222,
      "learning_rate": 0.00033439333243168607,
      "loss": 2.1915,
      "step": 106775
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.8687978982925415,
      "learning_rate": 0.00033438926883707287,
      "loss": 2.0229,
      "step": 106776
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.8937349319458008,
      "learning_rate": 0.0003343852052360659,
      "loss": 2.3515,
      "step": 106777
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.9216251373291016,
      "learning_rate": 0.0003343811416286661,
      "loss": 2.1204,
      "step": 106778
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.8786496520042419,
      "learning_rate": 0.00033437707801487425,
      "loss": 2.22,
      "step": 106779
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.8653784990310669,
      "learning_rate": 0.00033437301439469107,
      "loss": 2.4958,
      "step": 106780
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.8478708267211914,
      "learning_rate": 0.00033436895076811727,
      "loss": 2.2695,
      "step": 106781
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.8518769145011902,
      "learning_rate": 0.00033436488713515356,
      "loss": 2.1449,
      "step": 106782
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.841256856918335,
      "learning_rate": 0.0003343608234958009,
      "loss": 2.1991,
      "step": 106783
    },
    {
      "epoch": 1.39,
      "grad_norm": 1.0798697471618652,
      "learning_rate": 0.00033435675985005983,
      "loss": 2.1326,
      "step": 106784
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.9583293199539185,
      "learning_rate": 0.0003343526961979312,
      "loss": 2.0936,
      "step": 106785
    },
    {
      "epoch": 1.39,
      "grad_norm": 1.0930070877075195,
      "learning_rate": 0.00033434863253941575,
      "loss": 2.0789,
      "step": 106786
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.9148935675621033,
      "learning_rate": 0.0003343445688745143,
      "loss": 1.9655,
      "step": 106787
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.869462251663208,
      "learning_rate": 0.0003343405052032275,
      "loss": 1.8705,
      "step": 106788
    },
    {
      "epoch": 1.39,
      "grad_norm": 1.032616376876831,
      "learning_rate": 0.00033433644152555615,
      "loss": 2.2852,
      "step": 106789
    },
    {
      "epoch": 1.39,
      "grad_norm": 1.0574532747268677,
      "learning_rate": 0.00033433237784150094,
      "loss": 2.0188,
      "step": 106790
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.8846917748451233,
      "learning_rate": 0.00033432831415106286,
      "loss": 2.1707,
      "step": 106791
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.8382801413536072,
      "learning_rate": 0.00033432425045424235,
      "loss": 2.1857,
      "step": 106792
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.9799150228500366,
      "learning_rate": 0.00033432018675104034,
      "loss": 1.9917,
      "step": 106793
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.927649974822998,
      "learning_rate": 0.00033431612304145764,
      "loss": 2.3105,
      "step": 106794
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.863483190536499,
      "learning_rate": 0.00033431205932549484,
      "loss": 2.3279,
      "step": 106795
    },
    {
      "epoch": 1.39,
      "grad_norm": 1.115906834602356,
      "learning_rate": 0.0003343079956031528,
      "loss": 1.9979,
      "step": 106796
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.9791671633720398,
      "learning_rate": 0.0003343039318744323,
      "loss": 1.8753,
      "step": 106797
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.9987245202064514,
      "learning_rate": 0.00033429986813933405,
      "loss": 2.1047,
      "step": 106798
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.9561507105827332,
      "learning_rate": 0.00033429580439785873,
      "loss": 2.0262,
      "step": 106799
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.8869634866714478,
      "learning_rate": 0.0003342917406500072,
      "loss": 2.167,
      "step": 106800
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.952605128288269,
      "learning_rate": 0.00033428767689578026,
      "loss": 2.1585,
      "step": 106801
    },
    {
      "epoch": 1.39,
      "grad_norm": 1.1122981309890747,
      "learning_rate": 0.00033428361313517853,
      "loss": 2.0617,
      "step": 106802
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.883686900138855,
      "learning_rate": 0.00033427954936820283,
      "loss": 2.1609,
      "step": 106803
    },
    {
      "epoch": 1.39,
      "grad_norm": 1.0910484790802002,
      "learning_rate": 0.00033427548559485397,
      "loss": 1.985,
      "step": 106804
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.8822446465492249,
      "learning_rate": 0.00033427142181513254,
      "loss": 2.461,
      "step": 106805
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.8739825487136841,
      "learning_rate": 0.00033426735802903954,
      "loss": 2.3017,
      "step": 106806
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.8885687589645386,
      "learning_rate": 0.00033426329423657554,
      "loss": 2.1831,
      "step": 106807
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.8897128105163574,
      "learning_rate": 0.00033425923043774126,
      "loss": 2.3068,
      "step": 106808
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.8766783475875854,
      "learning_rate": 0.0003342551666325377,
      "loss": 2.2453,
      "step": 106809
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.9453144669532776,
      "learning_rate": 0.0003342511028209653,
      "loss": 2.0841,
      "step": 106810
    },
    {
      "epoch": 1.39,
      "grad_norm": 1.0050710439682007,
      "learning_rate": 0.00033424703900302507,
      "loss": 2.0125,
      "step": 106811
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.9084523916244507,
      "learning_rate": 0.0003342429751787177,
      "loss": 2.1423,
      "step": 106812
    },
    {
      "epoch": 1.39,
      "grad_norm": 1.0511796474456787,
      "learning_rate": 0.0003342389113480438,
      "loss": 2.0717,
      "step": 106813
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.8701675534248352,
      "learning_rate": 0.00033423484751100434,
      "loss": 2.0408,
      "step": 106814
    },
    {
      "epoch": 1.39,
      "grad_norm": 1.064974069595337,
      "learning_rate": 0.00033423078366759986,
      "loss": 2.2071,
      "step": 106815
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.8477790951728821,
      "learning_rate": 0.0003342267198178314,
      "loss": 2.0573,
      "step": 106816
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.9108451008796692,
      "learning_rate": 0.0003342226559616994,
      "loss": 2.2456,
      "step": 106817
    },
    {
      "epoch": 1.39,
      "grad_norm": 1.1235414743423462,
      "learning_rate": 0.0003342185920992048,
      "loss": 2.069,
      "step": 106818
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.9194337725639343,
      "learning_rate": 0.00033421452823034835,
      "loss": 2.1883,
      "step": 106819
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.9632341265678406,
      "learning_rate": 0.0003342104643551307,
      "loss": 2.0644,
      "step": 106820
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.8555608987808228,
      "learning_rate": 0.00033420640047355274,
      "loss": 2.1072,
      "step": 106821
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.960997998714447,
      "learning_rate": 0.00033420233658561525,
      "loss": 2.1608,
      "step": 106822
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.8711665868759155,
      "learning_rate": 0.00033419827269131876,
      "loss": 2.2895,
      "step": 106823
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.8609083294868469,
      "learning_rate": 0.0003341942087906642,
      "loss": 2.0248,
      "step": 106824
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.988360583782196,
      "learning_rate": 0.00033419014488365233,
      "loss": 2.1019,
      "step": 106825
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.934945821762085,
      "learning_rate": 0.00033418608097028385,
      "loss": 2.3317,
      "step": 106826
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.9381653666496277,
      "learning_rate": 0.0003341820170505595,
      "loss": 2.0473,
      "step": 106827
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.9840930700302124,
      "learning_rate": 0.0003341779531244801,
      "loss": 2.1502,
      "step": 106828
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.8633204102516174,
      "learning_rate": 0.00033417388919204644,
      "loss": 2.0209,
      "step": 106829
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.9272099137306213,
      "learning_rate": 0.00033416982525325906,
      "loss": 2.3007,
      "step": 106830
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.9268577098846436,
      "learning_rate": 0.00033416576130811896,
      "loss": 1.9465,
      "step": 106831
    },
    {
      "epoch": 1.39,
      "grad_norm": 1.1607805490493774,
      "learning_rate": 0.0003341616973566268,
      "loss": 2.1053,
      "step": 106832
    },
    {
      "epoch": 1.39,
      "grad_norm": 1.0050442218780518,
      "learning_rate": 0.00033415763339878326,
      "loss": 2.1509,
      "step": 106833
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.9193062782287598,
      "learning_rate": 0.0003341535694345893,
      "loss": 2.1001,
      "step": 106834
    },
    {
      "epoch": 1.39,
      "grad_norm": 1.0310426950454712,
      "learning_rate": 0.0003341495054640454,
      "loss": 2.3006,
      "step": 106835
    },
    {
      "epoch": 1.39,
      "grad_norm": 1.0239582061767578,
      "learning_rate": 0.00033414544148715255,
      "loss": 2.2407,
      "step": 106836
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.9401761889457703,
      "learning_rate": 0.00033414137750391147,
      "loss": 2.0986,
      "step": 106837
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.9585452675819397,
      "learning_rate": 0.00033413731351432277,
      "loss": 2.0552,
      "step": 106838
    },
    {
      "epoch": 1.39,
      "grad_norm": 1.0233440399169922,
      "learning_rate": 0.0003341332495183873,
      "loss": 2.1118,
      "step": 106839
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.9886011481285095,
      "learning_rate": 0.00033412918551610585,
      "loss": 2.2351,
      "step": 106840
    },
    {
      "epoch": 1.39,
      "grad_norm": 1.0868165493011475,
      "learning_rate": 0.0003341251215074791,
      "loss": 2.0255,
      "step": 106841
    },
    {
      "epoch": 1.39,
      "grad_norm": 1.026711344718933,
      "learning_rate": 0.00033412105749250793,
      "loss": 2.1071,
      "step": 106842
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.9563255310058594,
      "learning_rate": 0.000334116993471193,
      "loss": 2.2518,
      "step": 106843
    },
    {
      "epoch": 1.39,
      "grad_norm": 1.001348614692688,
      "learning_rate": 0.00033411292944353496,
      "loss": 2.0251,
      "step": 106844
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.9194226264953613,
      "learning_rate": 0.0003341088654095348,
      "loss": 2.0176,
      "step": 106845
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.9027774333953857,
      "learning_rate": 0.0003341048013691931,
      "loss": 2.082,
      "step": 106846
    },
    {
      "epoch": 1.39,
      "grad_norm": 1.018588662147522,
      "learning_rate": 0.0003341007373225107,
      "loss": 2.0854,
      "step": 106847
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.9838930368423462,
      "learning_rate": 0.00033409667326948826,
      "loss": 2.1668,
      "step": 106848
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.8465133309364319,
      "learning_rate": 0.00033409260921012676,
      "loss": 2.0249,
      "step": 106849
    },
    {
      "epoch": 1.39,
      "grad_norm": 1.0551905632019043,
      "learning_rate": 0.0003340885451444267,
      "loss": 2.0881,
      "step": 106850
    },
    {
      "epoch": 1.39,
      "grad_norm": 1.1019490957260132,
      "learning_rate": 0.0003340844810723889,
      "loss": 2.095,
      "step": 106851
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.936647891998291,
      "learning_rate": 0.0003340804169940142,
      "loss": 2.1424,
      "step": 106852
    },
    {
      "epoch": 1.39,
      "grad_norm": 1.0182459354400635,
      "learning_rate": 0.0003340763529093033,
      "loss": 2.1303,
      "step": 106853
    },
    {
      "epoch": 1.39,
      "grad_norm": 1.014766812324524,
      "learning_rate": 0.00033407228881825694,
      "loss": 2.14,
      "step": 106854
    },
    {
      "epoch": 1.39,
      "grad_norm": 1.0657150745391846,
      "learning_rate": 0.000334068224720876,
      "loss": 2.2205,
      "step": 106855
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.9327172636985779,
      "learning_rate": 0.00033406416061716103,
      "loss": 1.9594,
      "step": 106856
    },
    {
      "epoch": 1.39,
      "grad_norm": 1.1661782264709473,
      "learning_rate": 0.00033406009650711286,
      "loss": 2.2167,
      "step": 106857
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.8764646649360657,
      "learning_rate": 0.00033405603239073235,
      "loss": 2.1959,
      "step": 106858
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.9477849006652832,
      "learning_rate": 0.0003340519682680201,
      "loss": 2.2791,
      "step": 106859
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.9147315621376038,
      "learning_rate": 0.00033404790413897703,
      "loss": 2.1958,
      "step": 106860
    },
    {
      "epoch": 1.39,
      "grad_norm": 1.0068825483322144,
      "learning_rate": 0.0003340438400036038,
      "loss": 2.134,
      "step": 106861
    },
    {
      "epoch": 1.39,
      "grad_norm": 1.0100661516189575,
      "learning_rate": 0.0003340397758619012,
      "loss": 2.2227,
      "step": 106862
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.9226674437522888,
      "learning_rate": 0.0003340357117138699,
      "loss": 2.0025,
      "step": 106863
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.8762862682342529,
      "learning_rate": 0.0003340316475595107,
      "loss": 2.2846,
      "step": 106864
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.8803648352622986,
      "learning_rate": 0.00033402758339882445,
      "loss": 2.1436,
      "step": 106865
    },
    {
      "epoch": 1.39,
      "grad_norm": 1.2396283149719238,
      "learning_rate": 0.0003340235192318118,
      "loss": 2.197,
      "step": 106866
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.8673203587532043,
      "learning_rate": 0.0003340194550584735,
      "loss": 2.124,
      "step": 106867
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.936138391494751,
      "learning_rate": 0.00033401539087881045,
      "loss": 2.1633,
      "step": 106868
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.9492008686065674,
      "learning_rate": 0.00033401132669282324,
      "loss": 2.1793,
      "step": 106869
    },
    {
      "epoch": 1.39,
      "grad_norm": 1.0714547634124756,
      "learning_rate": 0.0003340072625005126,
      "loss": 2.1808,
      "step": 106870
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.8823375105857849,
      "learning_rate": 0.0003340031983018794,
      "loss": 2.2211,
      "step": 106871
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.9307523965835571,
      "learning_rate": 0.00033399913409692445,
      "loss": 2.4279,
      "step": 106872
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.9252802133560181,
      "learning_rate": 0.00033399506988564834,
      "loss": 2.3549,
      "step": 106873
    },
    {
      "epoch": 1.39,
      "grad_norm": 1.0308067798614502,
      "learning_rate": 0.0003339910056680519,
      "loss": 2.1479,
      "step": 106874
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.9113757610321045,
      "learning_rate": 0.00033398694144413596,
      "loss": 2.0154,
      "step": 106875
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.9340634346008301,
      "learning_rate": 0.00033398287721390116,
      "loss": 2.0852,
      "step": 106876
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.938917875289917,
      "learning_rate": 0.00033397881297734833,
      "loss": 2.0837,
      "step": 106877
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.9118757247924805,
      "learning_rate": 0.00033397474873447813,
      "loss": 2.0901,
      "step": 106878
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.9208688139915466,
      "learning_rate": 0.0003339706844852915,
      "loss": 2.2875,
      "step": 106879
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.8921613693237305,
      "learning_rate": 0.0003339666202297889,
      "loss": 2.2218,
      "step": 106880
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.975030243396759,
      "learning_rate": 0.00033396255596797146,
      "loss": 2.033,
      "step": 106881
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.8934946060180664,
      "learning_rate": 0.0003339584916998397,
      "loss": 2.2827,
      "step": 106882
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.924124538898468,
      "learning_rate": 0.00033395442742539425,
      "loss": 2.1754,
      "step": 106883
    },
    {
      "epoch": 1.39,
      "grad_norm": 1.271677851676941,
      "learning_rate": 0.0003339503631446362,
      "loss": 2.1381,
      "step": 106884
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.9647272229194641,
      "learning_rate": 0.00033394629885756603,
      "loss": 2.1562,
      "step": 106885
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.8383074402809143,
      "learning_rate": 0.0003339422345641847,
      "loss": 2.1417,
      "step": 106886
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.9944760799407959,
      "learning_rate": 0.0003339381702644928,
      "loss": 2.1171,
      "step": 106887
    },
    {
      "epoch": 1.39,
      "grad_norm": 1.020950436592102,
      "learning_rate": 0.00033393410595849115,
      "loss": 2.0829,
      "step": 106888
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.923858106136322,
      "learning_rate": 0.0003339300416461806,
      "loss": 2.1551,
      "step": 106889
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.9673399329185486,
      "learning_rate": 0.00033392597732756176,
      "loss": 2.3878,
      "step": 106890
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.8993720412254333,
      "learning_rate": 0.0003339219130026354,
      "loss": 2.1425,
      "step": 106891
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.9724448323249817,
      "learning_rate": 0.00033391784867140234,
      "loss": 2.3599,
      "step": 106892
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.843492865562439,
      "learning_rate": 0.0003339137843338633,
      "loss": 2.0879,
      "step": 106893
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.8271090388298035,
      "learning_rate": 0.0003339097199900191,
      "loss": 2.2014,
      "step": 106894
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.876642644405365,
      "learning_rate": 0.0003339056556398705,
      "loss": 2.1384,
      "step": 106895
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.8484744429588318,
      "learning_rate": 0.000333901591283418,
      "loss": 2.2879,
      "step": 106896
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.9254544377326965,
      "learning_rate": 0.00033389752692066264,
      "loss": 2.1041,
      "step": 106897
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.8774563074111938,
      "learning_rate": 0.0003338934625516052,
      "loss": 2.1446,
      "step": 106898
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.8724089860916138,
      "learning_rate": 0.0003338893981762462,
      "loss": 2.3166,
      "step": 106899
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.8630479574203491,
      "learning_rate": 0.00033388533379458655,
      "loss": 2.2126,
      "step": 106900
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.8291933536529541,
      "learning_rate": 0.00033388126940662706,
      "loss": 2.0122,
      "step": 106901
    },
    {
      "epoch": 1.39,
      "grad_norm": 1.0162384510040283,
      "learning_rate": 0.0003338772050123683,
      "loss": 2.1052,
      "step": 106902
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.9460369944572449,
      "learning_rate": 0.00033387314061181114,
      "loss": 1.9866,
      "step": 106903
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.8365392684936523,
      "learning_rate": 0.0003338690762049564,
      "loss": 1.9146,
      "step": 106904
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.9205787777900696,
      "learning_rate": 0.00033386501179180465,
      "loss": 2.279,
      "step": 106905
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.8717803955078125,
      "learning_rate": 0.0003338609473723568,
      "loss": 2.0914,
      "step": 106906
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.9657118916511536,
      "learning_rate": 0.00033385688294661355,
      "loss": 2.1809,
      "step": 106907
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.9541972875595093,
      "learning_rate": 0.0003338528185145757,
      "loss": 2.2307,
      "step": 106908
    },
    {
      "epoch": 1.39,
      "grad_norm": 1.011997103691101,
      "learning_rate": 0.0003338487540762439,
      "loss": 2.1166,
      "step": 106909
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.9932839870452881,
      "learning_rate": 0.00033384468963161914,
      "loss": 2.1935,
      "step": 106910
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.963439404964447,
      "learning_rate": 0.0003338406251807019,
      "loss": 2.1905,
      "step": 106911
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.9045746326446533,
      "learning_rate": 0.000333836560723493,
      "loss": 2.3441,
      "step": 106912
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.8337365388870239,
      "learning_rate": 0.00033383249625999333,
      "loss": 2.3038,
      "step": 106913
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.9333354830741882,
      "learning_rate": 0.00033382843179020346,
      "loss": 2.1371,
      "step": 106914
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.8568692803382874,
      "learning_rate": 0.0003338243673141243,
      "loss": 2.1654,
      "step": 106915
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.9730628132820129,
      "learning_rate": 0.00033382030283175666,
      "loss": 2.165,
      "step": 106916
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.9698078632354736,
      "learning_rate": 0.0003338162383431011,
      "loss": 2.129,
      "step": 106917
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.8965858221054077,
      "learning_rate": 0.00033381217384815837,
      "loss": 2.0953,
      "step": 106918
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.9152681827545166,
      "learning_rate": 0.00033380810934692946,
      "loss": 1.9232,
      "step": 106919
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.9187016487121582,
      "learning_rate": 0.00033380404483941494,
      "loss": 2.0545,
      "step": 106920
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.8954223394393921,
      "learning_rate": 0.0003337999803256156,
      "loss": 2.1551,
      "step": 106921
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.8439406156539917,
      "learning_rate": 0.0003337959158055322,
      "loss": 2.163,
      "step": 106922
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.9506474137306213,
      "learning_rate": 0.00033379185127916554,
      "loss": 2.1245,
      "step": 106923
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.9203947186470032,
      "learning_rate": 0.0003337877867465163,
      "loss": 2.2467,
      "step": 106924
    },
    {
      "epoch": 1.39,
      "grad_norm": 1.0625722408294678,
      "learning_rate": 0.0003337837222075853,
      "loss": 2.2636,
      "step": 106925
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.9062755107879639,
      "learning_rate": 0.0003337796576623732,
      "loss": 2.2875,
      "step": 106926
    },
    {
      "epoch": 1.39,
      "grad_norm": 1.1523425579071045,
      "learning_rate": 0.0003337755931108809,
      "loss": 2.1682,
      "step": 106927
    },
    {
      "epoch": 1.39,
      "grad_norm": 1.027564287185669,
      "learning_rate": 0.000333771528553109,
      "loss": 2.2368,
      "step": 106928
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.9038848876953125,
      "learning_rate": 0.0003337674639890584,
      "loss": 2.3212,
      "step": 106929
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.8945881724357605,
      "learning_rate": 0.00033376339941872975,
      "loss": 2.0967,
      "step": 106930
    },
    {
      "epoch": 1.39,
      "grad_norm": 1.08263099193573,
      "learning_rate": 0.00033375933484212385,
      "loss": 2.2834,
      "step": 106931
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.9360790848731995,
      "learning_rate": 0.00033375527025924156,
      "loss": 2.0543,
      "step": 106932
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.8966464996337891,
      "learning_rate": 0.0003337512056700834,
      "loss": 2.0395,
      "step": 106933
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.9811808466911316,
      "learning_rate": 0.00033374714107465024,
      "loss": 2.1026,
      "step": 106934
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.9878215789794922,
      "learning_rate": 0.000333743076472943,
      "loss": 2.3628,
      "step": 106935
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.9438232779502869,
      "learning_rate": 0.0003337390118649622,
      "loss": 2.0741,
      "step": 106936
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.7815536856651306,
      "learning_rate": 0.0003337349472507086,
      "loss": 2.0583,
      "step": 106937
    },
    {
      "epoch": 1.39,
      "grad_norm": 1.1219120025634766,
      "learning_rate": 0.0003337308826301832,
      "loss": 2.3224,
      "step": 106938
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.916522741317749,
      "learning_rate": 0.00033372681800338644,
      "loss": 1.9169,
      "step": 106939
    },
    {
      "epoch": 1.39,
      "grad_norm": 1.0182379484176636,
      "learning_rate": 0.00033372275337031925,
      "loss": 2.0623,
      "step": 106940
    },
    {
      "epoch": 1.39,
      "grad_norm": 1.0387370586395264,
      "learning_rate": 0.0003337186887309825,
      "loss": 2.1503,
      "step": 106941
    },
    {
      "epoch": 1.39,
      "grad_norm": 1.073827862739563,
      "learning_rate": 0.0003337146240853767,
      "loss": 2.1067,
      "step": 106942
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.9001465439796448,
      "learning_rate": 0.00033371055943350264,
      "loss": 2.1144,
      "step": 106943
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.9546301364898682,
      "learning_rate": 0.0003337064947753613,
      "loss": 2.2679,
      "step": 106944
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.9203866720199585,
      "learning_rate": 0.0003337024301109532,
      "loss": 2.0824,
      "step": 106945
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.9766663908958435,
      "learning_rate": 0.0003336983654402792,
      "loss": 2.3043,
      "step": 106946
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.8508042097091675,
      "learning_rate": 0.0003336943007633401,
      "loss": 1.986,
      "step": 106947
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.8839635848999023,
      "learning_rate": 0.0003336902360801365,
      "loss": 2.2049,
      "step": 106948
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.9056875705718994,
      "learning_rate": 0.00033368617139066927,
      "loss": 2.1362,
      "step": 106949
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.8730140924453735,
      "learning_rate": 0.0003336821066949392,
      "loss": 2.2231,
      "step": 106950
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.9346170425415039,
      "learning_rate": 0.00033367804199294695,
      "loss": 2.2119,
      "step": 106951
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.8791786432266235,
      "learning_rate": 0.00033367397728469326,
      "loss": 2.0749,
      "step": 106952
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.8696102499961853,
      "learning_rate": 0.00033366991257017907,
      "loss": 2.2535,
      "step": 106953
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.8741630911827087,
      "learning_rate": 0.0003336658478494049,
      "loss": 2.1986,
      "step": 106954
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.9174394011497498,
      "learning_rate": 0.0003336617831223716,
      "loss": 2.1067,
      "step": 106955
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.8281317949295044,
      "learning_rate": 0.00033365771838908006,
      "loss": 2.2288,
      "step": 106956
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.8485584259033203,
      "learning_rate": 0.0003336536536495309,
      "loss": 2.1563,
      "step": 106957
    },
    {
      "epoch": 1.39,
      "grad_norm": 1.022782325744629,
      "learning_rate": 0.00033364958890372477,
      "loss": 1.8495,
      "step": 106958
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.8484640717506409,
      "learning_rate": 0.0003336455241516626,
      "loss": 2.1585,
      "step": 106959
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.9478453993797302,
      "learning_rate": 0.0003336414593933451,
      "loss": 2.2323,
      "step": 106960
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.9325463175773621,
      "learning_rate": 0.00033363739462877304,
      "loss": 2.1999,
      "step": 106961
    },
    {
      "epoch": 1.39,
      "grad_norm": 1.1806840896606445,
      "learning_rate": 0.0003336333298579471,
      "loss": 2.1394,
      "step": 106962
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.898651659488678,
      "learning_rate": 0.00033362926508086817,
      "loss": 1.9129,
      "step": 106963
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.9279483556747437,
      "learning_rate": 0.00033362520029753685,
      "loss": 2.0939,
      "step": 106964
    },
    {
      "epoch": 1.39,
      "grad_norm": 1.173136591911316,
      "learning_rate": 0.000333621135507954,
      "loss": 2.2171,
      "step": 106965
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.8833224177360535,
      "learning_rate": 0.0003336170707121203,
      "loss": 2.248,
      "step": 106966
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.8875637054443359,
      "learning_rate": 0.0003336130059100367,
      "loss": 2.1235,
      "step": 106967
    },
    {
      "epoch": 1.39,
      "grad_norm": 1.1604278087615967,
      "learning_rate": 0.0003336089411017036,
      "loss": 2.1898,
      "step": 106968
    },
    {
      "epoch": 1.39,
      "grad_norm": 1.0407658815383911,
      "learning_rate": 0.0003336048762871221,
      "loss": 2.3394,
      "step": 106969
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.9202550649642944,
      "learning_rate": 0.00033360081146629274,
      "loss": 1.9439,
      "step": 106970
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.9282056093215942,
      "learning_rate": 0.0003335967466392164,
      "loss": 2.0836,
      "step": 106971
    },
    {
      "epoch": 1.39,
      "grad_norm": 1.027969241142273,
      "learning_rate": 0.0003335926818058938,
      "loss": 2.3748,
      "step": 106972
    },
    {
      "epoch": 1.39,
      "grad_norm": 1.2538870573043823,
      "learning_rate": 0.0003335886169663257,
      "loss": 2.1743,
      "step": 106973
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.8000818490982056,
      "learning_rate": 0.0003335845521205129,
      "loss": 2.2347,
      "step": 106974
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.9401729702949524,
      "learning_rate": 0.00033358048726845593,
      "loss": 2.1374,
      "step": 106975
    },
    {
      "epoch": 1.39,
      "grad_norm": 1.1441069841384888,
      "learning_rate": 0.0003335764224101558,
      "loss": 1.9662,
      "step": 106976
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.9066903591156006,
      "learning_rate": 0.0003335723575456132,
      "loss": 2.2221,
      "step": 106977
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.8147912621498108,
      "learning_rate": 0.00033356829267482883,
      "loss": 2.0368,
      "step": 106978
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.8920685052871704,
      "learning_rate": 0.0003335642277978035,
      "loss": 2.1156,
      "step": 106979
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.9153696298599243,
      "learning_rate": 0.000333560162914538,
      "loss": 2.3,
      "step": 106980
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.9505626559257507,
      "learning_rate": 0.00033355609802503296,
      "loss": 2.1823,
      "step": 106981
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.8610169887542725,
      "learning_rate": 0.0003335520331292892,
      "loss": 2.0356,
      "step": 106982
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.8936008214950562,
      "learning_rate": 0.00033354796822730753,
      "loss": 1.9796,
      "step": 106983
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.9226457476615906,
      "learning_rate": 0.00033354390331908864,
      "loss": 2.149,
      "step": 106984
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.9593625664710999,
      "learning_rate": 0.00033353983840463325,
      "loss": 2.0099,
      "step": 106985
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.8981665372848511,
      "learning_rate": 0.00033353577348394224,
      "loss": 2.1994,
      "step": 106986
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.7940689325332642,
      "learning_rate": 0.00033353170855701637,
      "loss": 2.0214,
      "step": 106987
    },
    {
      "epoch": 1.39,
      "grad_norm": 1.0401557683944702,
      "learning_rate": 0.0003335276436238561,
      "loss": 2.1477,
      "step": 106988
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.9843124151229858,
      "learning_rate": 0.00033352357868446267,
      "loss": 2.1595,
      "step": 106989
    },
    {
      "epoch": 1.39,
      "grad_norm": 1.0920008420944214,
      "learning_rate": 0.0003335195137388364,
      "loss": 2.1058,
      "step": 106990
    },
    {
      "epoch": 1.39,
      "grad_norm": 1.1969934701919556,
      "learning_rate": 0.0003335154487869782,
      "loss": 2.2802,
      "step": 106991
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.9010578393936157,
      "learning_rate": 0.000333511383828889,
      "loss": 2.4621,
      "step": 106992
    },
    {
      "epoch": 1.39,
      "grad_norm": 1.033416986465454,
      "learning_rate": 0.0003335073188645693,
      "loss": 1.9861,
      "step": 106993
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.9953233003616333,
      "learning_rate": 0.0003335032538940199,
      "loss": 2.2318,
      "step": 106994
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.8472177982330322,
      "learning_rate": 0.00033349918891724167,
      "loss": 2.0119,
      "step": 106995
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.9570607542991638,
      "learning_rate": 0.0003334951239342354,
      "loss": 1.9341,
      "step": 106996
    },
    {
      "epoch": 1.39,
      "grad_norm": 1.0259042978286743,
      "learning_rate": 0.00033349105894500165,
      "loss": 2.1923,
      "step": 106997
    },
    {
      "epoch": 1.39,
      "grad_norm": 1.0539261102676392,
      "learning_rate": 0.00033348699394954124,
      "loss": 2.2641,
      "step": 106998
    },
    {
      "epoch": 1.39,
      "grad_norm": 1.1558431386947632,
      "learning_rate": 0.00033348292894785507,
      "loss": 1.9325,
      "step": 106999
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.983260452747345,
      "learning_rate": 0.0003334788639399438,
      "loss": 2.3409,
      "step": 107000
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.9086118936538696,
      "learning_rate": 0.0003334747989258081,
      "loss": 2.2219,
      "step": 107001
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.9395821690559387,
      "learning_rate": 0.00033347073390544883,
      "loss": 2.2656,
      "step": 107002
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.950874388217926,
      "learning_rate": 0.00033346666887886666,
      "loss": 2.2245,
      "step": 107003
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.91498863697052,
      "learning_rate": 0.00033346260384606253,
      "loss": 2.0567,
      "step": 107004
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.8777956962585449,
      "learning_rate": 0.000333458538807037,
      "loss": 2.1501,
      "step": 107005
    },
    {
      "epoch": 1.39,
      "grad_norm": 1.0233380794525146,
      "learning_rate": 0.00033345447376179093,
      "loss": 2.4826,
      "step": 107006
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.9088029265403748,
      "learning_rate": 0.0003334504087103249,
      "loss": 2.0647,
      "step": 107007
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.9014736413955688,
      "learning_rate": 0.00033344634365264003,
      "loss": 2.2478,
      "step": 107008
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.8552175760269165,
      "learning_rate": 0.0003334422785887367,
      "loss": 2.0728,
      "step": 107009
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.9852939248085022,
      "learning_rate": 0.00033343821351861585,
      "loss": 2.2269,
      "step": 107010
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.861193835735321,
      "learning_rate": 0.00033343414844227824,
      "loss": 2.2474,
      "step": 107011
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.8523274660110474,
      "learning_rate": 0.0003334300833597245,
      "loss": 1.881,
      "step": 107012
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.8354742527008057,
      "learning_rate": 0.0003334260182709556,
      "loss": 2.1169,
      "step": 107013
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.8887704014778137,
      "learning_rate": 0.00033342195317597216,
      "loss": 2.1409,
      "step": 107014
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.8731985688209534,
      "learning_rate": 0.0003334178880747749,
      "loss": 2.2228,
      "step": 107015
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.9525648355484009,
      "learning_rate": 0.00033341382296736457,
      "loss": 2.1317,
      "step": 107016
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.8893502950668335,
      "learning_rate": 0.0003334097578537421,
      "loss": 2.3638,
      "step": 107017
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.9173997044563293,
      "learning_rate": 0.00033340569273390805,
      "loss": 2.0012,
      "step": 107018
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.9228319525718689,
      "learning_rate": 0.00033340162760786324,
      "loss": 2.2681,
      "step": 107019
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.9738901853561401,
      "learning_rate": 0.0003333975624756085,
      "loss": 2.258,
      "step": 107020
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.8662154674530029,
      "learning_rate": 0.0003333934973371444,
      "loss": 2.0849,
      "step": 107021
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.9459743499755859,
      "learning_rate": 0.000333389432192472,
      "loss": 2.1931,
      "step": 107022
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.8930475115776062,
      "learning_rate": 0.0003333853670415918,
      "loss": 2.1232,
      "step": 107023
    },
    {
      "epoch": 1.39,
      "grad_norm": 1.1275910139083862,
      "learning_rate": 0.00033338130188450453,
      "loss": 2.146,
      "step": 107024
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.9247937798500061,
      "learning_rate": 0.0003333772367212111,
      "loss": 2.0044,
      "step": 107025
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.92237389087677,
      "learning_rate": 0.00033337317155171227,
      "loss": 2.1687,
      "step": 107026
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.9212864637374878,
      "learning_rate": 0.00033336910637600867,
      "loss": 2.0849,
      "step": 107027
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.8841125965118408,
      "learning_rate": 0.0003333650411941012,
      "loss": 1.9412,
      "step": 107028
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.9013514518737793,
      "learning_rate": 0.00033336097600599054,
      "loss": 2.1497,
      "step": 107029
    },
    {
      "epoch": 1.39,
      "grad_norm": 1.0520652532577515,
      "learning_rate": 0.00033335691081167737,
      "loss": 1.9771,
      "step": 107030
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.8232957720756531,
      "learning_rate": 0.00033335284561116256,
      "loss": 2.0623,
      "step": 107031
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.9031516313552856,
      "learning_rate": 0.0003333487804044468,
      "loss": 2.1583,
      "step": 107032
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.8940789103507996,
      "learning_rate": 0.00033334471519153085,
      "loss": 2.178,
      "step": 107033
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.9160163998603821,
      "learning_rate": 0.0003333406499724155,
      "loss": 1.9947,
      "step": 107034
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.9128877520561218,
      "learning_rate": 0.0003333365847471016,
      "loss": 1.9948,
      "step": 107035
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.8806897401809692,
      "learning_rate": 0.0003333325195155897,
      "loss": 2.2143,
      "step": 107036
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.9601919651031494,
      "learning_rate": 0.00033332845427788064,
      "loss": 2.4771,
      "step": 107037
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.8851073980331421,
      "learning_rate": 0.0003333243890339753,
      "loss": 2.1196,
      "step": 107038
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.9434154629707336,
      "learning_rate": 0.0003333203237838742,
      "loss": 2.0895,
      "step": 107039
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.8976216316223145,
      "learning_rate": 0.0003333162585275783,
      "loss": 2.2675,
      "step": 107040
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.905850350856781,
      "learning_rate": 0.00033331219326508823,
      "loss": 1.9603,
      "step": 107041
    },
    {
      "epoch": 1.39,
      "grad_norm": 1.082871913909912,
      "learning_rate": 0.0003333081279964048,
      "loss": 2.0626,
      "step": 107042
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.9049601554870605,
      "learning_rate": 0.00033330406272152885,
      "loss": 1.9916,
      "step": 107043
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.9072253108024597,
      "learning_rate": 0.00033329999744046096,
      "loss": 2.3309,
      "step": 107044
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.960202693939209,
      "learning_rate": 0.00033329593215320196,
      "loss": 2.1359,
      "step": 107045
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.9338017106056213,
      "learning_rate": 0.00033329186685975264,
      "loss": 2.1046,
      "step": 107046
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.9017788171768188,
      "learning_rate": 0.0003332878015601137,
      "loss": 2.0554,
      "step": 107047
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.9656022787094116,
      "learning_rate": 0.00033328373625428597,
      "loss": 2.2616,
      "step": 107048
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.9500899910926819,
      "learning_rate": 0.0003332796709422702,
      "loss": 2.3438,
      "step": 107049
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.8888802528381348,
      "learning_rate": 0.000333275605624067,
      "loss": 1.9533,
      "step": 107050
    },
    {
      "epoch": 1.39,
      "grad_norm": 1.030104160308838,
      "learning_rate": 0.0003332715402996774,
      "loss": 1.984,
      "step": 107051
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.8412884473800659,
      "learning_rate": 0.00033326747496910186,
      "loss": 2.1584,
      "step": 107052
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.8825098872184753,
      "learning_rate": 0.0003332634096323413,
      "loss": 1.9945,
      "step": 107053
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.8652815818786621,
      "learning_rate": 0.0003332593442893965,
      "loss": 2.087,
      "step": 107054
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.852135181427002,
      "learning_rate": 0.0003332552789402681,
      "loss": 2.0814,
      "step": 107055
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.9265480041503906,
      "learning_rate": 0.0003332512135849569,
      "loss": 2.0343,
      "step": 107056
    },
    {
      "epoch": 1.39,
      "grad_norm": 1.013606071472168,
      "learning_rate": 0.0003332471482234637,
      "loss": 2.4663,
      "step": 107057
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.939807116985321,
      "learning_rate": 0.0003332430828557892,
      "loss": 2.0865,
      "step": 107058
    },
    {
      "epoch": 1.39,
      "grad_norm": 1.0263699293136597,
      "learning_rate": 0.0003332390174819343,
      "loss": 2.0557,
      "step": 107059
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.9326382279396057,
      "learning_rate": 0.0003332349521018995,
      "loss": 2.3986,
      "step": 107060
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.932680606842041,
      "learning_rate": 0.0003332308867156857,
      "loss": 2.0428,
      "step": 107061
    },
    {
      "epoch": 1.39,
      "grad_norm": 1.107100486755371,
      "learning_rate": 0.0003332268213232937,
      "loss": 2.0396,
      "step": 107062
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.8456616401672363,
      "learning_rate": 0.0003332227559247243,
      "loss": 2.0225,
      "step": 107063
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.9166334867477417,
      "learning_rate": 0.00033321869051997804,
      "loss": 2.2385,
      "step": 107064
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.9008832573890686,
      "learning_rate": 0.0003332146251090558,
      "loss": 2.3252,
      "step": 107065
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.9878013134002686,
      "learning_rate": 0.00033321055969195843,
      "loss": 2.0502,
      "step": 107066
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.8792589902877808,
      "learning_rate": 0.0003332064942686865,
      "loss": 2.1753,
      "step": 107067
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.9361578822135925,
      "learning_rate": 0.00033320242883924086,
      "loss": 2.2536,
      "step": 107068
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.9349333643913269,
      "learning_rate": 0.0003331983634036223,
      "loss": 2.1672,
      "step": 107069
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.9377155900001526,
      "learning_rate": 0.00033319429796183146,
      "loss": 2.1391,
      "step": 107070
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.8656238913536072,
      "learning_rate": 0.0003331902325138692,
      "loss": 2.3008,
      "step": 107071
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.8673404455184937,
      "learning_rate": 0.00033318616705973634,
      "loss": 2.2906,
      "step": 107072
    },
    {
      "epoch": 1.39,
      "grad_norm": 1.149156928062439,
      "learning_rate": 0.00033318210159943347,
      "loss": 1.994,
      "step": 107073
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.9283957481384277,
      "learning_rate": 0.0003331780361329615,
      "loss": 2.1974,
      "step": 107074
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.9570361375808716,
      "learning_rate": 0.000333173970660321,
      "loss": 1.9923,
      "step": 107075
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.9413865804672241,
      "learning_rate": 0.00033316990518151283,
      "loss": 2.3071,
      "step": 107076
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.8158121109008789,
      "learning_rate": 0.0003331658396965378,
      "loss": 2.0249,
      "step": 107077
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.9491742849349976,
      "learning_rate": 0.00033316177420539663,
      "loss": 2.1823,
      "step": 107078
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.8687212467193604,
      "learning_rate": 0.00033315770870809004,
      "loss": 2.2294,
      "step": 107079
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.8874955773353577,
      "learning_rate": 0.0003331536432046188,
      "loss": 2.234,
      "step": 107080
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.8895645141601562,
      "learning_rate": 0.0003331495776949837,
      "loss": 2.0948,
      "step": 107081
    },
    {
      "epoch": 1.39,
      "grad_norm": 1.0892289876937866,
      "learning_rate": 0.0003331455121791854,
      "loss": 2.209,
      "step": 107082
    },
    {
      "epoch": 1.39,
      "grad_norm": 1.0291327238082886,
      "learning_rate": 0.0003331414466572248,
      "loss": 1.8796,
      "step": 107083
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.8762280941009521,
      "learning_rate": 0.00033313738112910257,
      "loss": 2.0578,
      "step": 107084
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.8403573036193848,
      "learning_rate": 0.0003331333155948194,
      "loss": 2.0628,
      "step": 107085
    },
    {
      "epoch": 1.39,
      "grad_norm": 1.179684042930603,
      "learning_rate": 0.0003331292500543762,
      "loss": 1.8703,
      "step": 107086
    },
    {
      "epoch": 1.39,
      "grad_norm": 1.0022536516189575,
      "learning_rate": 0.0003331251845077737,
      "loss": 1.931,
      "step": 107087
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.8786500096321106,
      "learning_rate": 0.0003331211189550124,
      "loss": 2.1872,
      "step": 107088
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.9692187309265137,
      "learning_rate": 0.00033311705339609344,
      "loss": 2.2986,
      "step": 107089
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.9242833852767944,
      "learning_rate": 0.0003331129878310174,
      "loss": 2.2093,
      "step": 107090
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.9315523505210876,
      "learning_rate": 0.0003331089222597849,
      "loss": 2.3252,
      "step": 107091
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.8876832723617554,
      "learning_rate": 0.0003331048566823969,
      "loss": 2.1971,
      "step": 107092
    },
    {
      "epoch": 1.39,
      "grad_norm": 1.2062945365905762,
      "learning_rate": 0.0003331007910988541,
      "loss": 2.0888,
      "step": 107093
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.9184115529060364,
      "learning_rate": 0.00033309672550915716,
      "loss": 2.0682,
      "step": 107094
    },
    {
      "epoch": 1.39,
      "grad_norm": 1.0042046308517456,
      "learning_rate": 0.000333092659913307,
      "loss": 2.2021,
      "step": 107095
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.8666970729827881,
      "learning_rate": 0.00033308859431130426,
      "loss": 1.9716,
      "step": 107096
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.8784924149513245,
      "learning_rate": 0.00033308452870314977,
      "loss": 2.0927,
      "step": 107097
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.8504295349121094,
      "learning_rate": 0.0003330804630888441,
      "loss": 2.2521,
      "step": 107098
    },
    {
      "epoch": 1.39,
      "grad_norm": 1.1494998931884766,
      "learning_rate": 0.0003330763974683883,
      "loss": 2.4045,
      "step": 107099
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.9066864848136902,
      "learning_rate": 0.0003330723318417829,
      "loss": 2.1327,
      "step": 107100
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.9467986822128296,
      "learning_rate": 0.00033306826620902873,
      "loss": 2.2151,
      "step": 107101
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.8534879088401794,
      "learning_rate": 0.0003330642005701266,
      "loss": 2.0702,
      "step": 107102
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.8290718197822571,
      "learning_rate": 0.0003330601349250772,
      "loss": 2.3328,
      "step": 107103
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.9151210784912109,
      "learning_rate": 0.0003330560692738812,
      "loss": 2.0086,
      "step": 107104
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.9191531538963318,
      "learning_rate": 0.0003330520036165396,
      "loss": 2.3392,
      "step": 107105
    },
    {
      "epoch": 1.39,
      "grad_norm": 1.1183290481567383,
      "learning_rate": 0.00033304793795305284,
      "loss": 2.2089,
      "step": 107106
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.8370711803436279,
      "learning_rate": 0.00033304387228342197,
      "loss": 2.1949,
      "step": 107107
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.9755862355232239,
      "learning_rate": 0.00033303980660764757,
      "loss": 2.1257,
      "step": 107108
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.8540290594100952,
      "learning_rate": 0.00033303574092573046,
      "loss": 2.1807,
      "step": 107109
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.8981806039810181,
      "learning_rate": 0.0003330316752376714,
      "loss": 2.1866,
      "step": 107110
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.8194161653518677,
      "learning_rate": 0.00033302760954347115,
      "loss": 2.0155,
      "step": 107111
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.9239370226860046,
      "learning_rate": 0.0003330235438431304,
      "loss": 2.1839,
      "step": 107112
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.907211422920227,
      "learning_rate": 0.00033301947813664995,
      "loss": 2.0336,
      "step": 107113
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.9028758406639099,
      "learning_rate": 0.00033301541242403056,
      "loss": 2.2252,
      "step": 107114
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.8807001709938049,
      "learning_rate": 0.00033301134670527297,
      "loss": 2.2111,
      "step": 107115
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.9513935446739197,
      "learning_rate": 0.0003330072809803779,
      "loss": 2.2443,
      "step": 107116
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.9840776324272156,
      "learning_rate": 0.0003330032152493463,
      "loss": 2.3621,
      "step": 107117
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.9866469502449036,
      "learning_rate": 0.0003329991495121787,
      "loss": 2.087,
      "step": 107118
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.8190455436706543,
      "learning_rate": 0.0003329950837688759,
      "loss": 2.2057,
      "step": 107119
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.8980712294578552,
      "learning_rate": 0.00033299101801943874,
      "loss": 2.2597,
      "step": 107120
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.8471217155456543,
      "learning_rate": 0.00033298695226386787,
      "loss": 2.2086,
      "step": 107121
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.9074408411979675,
      "learning_rate": 0.00033298288650216416,
      "loss": 2.2333,
      "step": 107122
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.9143083691596985,
      "learning_rate": 0.0003329788207343283,
      "loss": 2.1476,
      "step": 107123
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.8500462174415588,
      "learning_rate": 0.00033297475496036103,
      "loss": 2.2517,
      "step": 107124
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.9634230136871338,
      "learning_rate": 0.00033297068918026313,
      "loss": 2.0863,
      "step": 107125
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.9099753499031067,
      "learning_rate": 0.0003329666233940354,
      "loss": 2.2115,
      "step": 107126
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.8367984890937805,
      "learning_rate": 0.0003329625576016785,
      "loss": 2.1395,
      "step": 107127
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.9805203080177307,
      "learning_rate": 0.00033295849180319323,
      "loss": 2.3346,
      "step": 107128
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.9990098476409912,
      "learning_rate": 0.0003329544259985804,
      "loss": 2.2432,
      "step": 107129
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.8474997878074646,
      "learning_rate": 0.00033295036018784075,
      "loss": 2.0197,
      "step": 107130
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.9349613189697266,
      "learning_rate": 0.0003329462943709749,
      "loss": 1.9772,
      "step": 107131
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.9945234656333923,
      "learning_rate": 0.00033294222854798373,
      "loss": 2.2125,
      "step": 107132
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.8474094271659851,
      "learning_rate": 0.0003329381627188681,
      "loss": 2.1853,
      "step": 107133
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.9192079901695251,
      "learning_rate": 0.00033293409688362857,
      "loss": 2.2946,
      "step": 107134
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.9406020641326904,
      "learning_rate": 0.0003329300310422659,
      "loss": 2.0862,
      "step": 107135
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.903681218624115,
      "learning_rate": 0.00033292596519478106,
      "loss": 2.0898,
      "step": 107136
    },
    {
      "epoch": 1.39,
      "grad_norm": 0.8470005989074707,
      "learning_rate": 0.0003329218993411745,
      "loss": 1.9738,
      "step": 107137
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.8743112683296204,
      "learning_rate": 0.00033291783348144723,
      "loss": 2.0435,
      "step": 107138
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.9037350416183472,
      "learning_rate": 0.00033291376761559994,
      "loss": 2.1232,
      "step": 107139
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.828075647354126,
      "learning_rate": 0.00033290970174363336,
      "loss": 2.0263,
      "step": 107140
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.9521462917327881,
      "learning_rate": 0.00033290563586554813,
      "loss": 2.1594,
      "step": 107141
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.9142955541610718,
      "learning_rate": 0.0003329015699813452,
      "loss": 2.1579,
      "step": 107142
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.9712029099464417,
      "learning_rate": 0.00033289750409102524,
      "loss": 2.054,
      "step": 107143
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.8966227769851685,
      "learning_rate": 0.000332893438194589,
      "loss": 2.014,
      "step": 107144
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.8604188561439514,
      "learning_rate": 0.00033288937229203726,
      "loss": 2.0918,
      "step": 107145
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.9116252064704895,
      "learning_rate": 0.00033288530638337076,
      "loss": 2.1861,
      "step": 107146
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.9251354336738586,
      "learning_rate": 0.00033288124046859034,
      "loss": 2.1338,
      "step": 107147
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.8455456495285034,
      "learning_rate": 0.00033287717454769653,
      "loss": 2.3636,
      "step": 107148
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.9189292788505554,
      "learning_rate": 0.0003328731086206903,
      "loss": 2.1935,
      "step": 107149
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.9118226766586304,
      "learning_rate": 0.00033286904268757235,
      "loss": 2.0971,
      "step": 107150
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.8980169892311096,
      "learning_rate": 0.0003328649767483435,
      "loss": 2.129,
      "step": 107151
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.8712851405143738,
      "learning_rate": 0.0003328609108030043,
      "loss": 2.1609,
      "step": 107152
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.9443178772926331,
      "learning_rate": 0.0003328568448515557,
      "loss": 2.1388,
      "step": 107153
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.951503336429596,
      "learning_rate": 0.0003328527788939984,
      "loss": 2.2848,
      "step": 107154
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.9994409084320068,
      "learning_rate": 0.0003328487129303331,
      "loss": 2.1266,
      "step": 107155
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.8701984286308289,
      "learning_rate": 0.00033284464696056064,
      "loss": 2.1201,
      "step": 107156
    },
    {
      "epoch": 1.4,
      "grad_norm": 1.0443278551101685,
      "learning_rate": 0.0003328405809846818,
      "loss": 2.0649,
      "step": 107157
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.8260439038276672,
      "learning_rate": 0.0003328365150026972,
      "loss": 2.1422,
      "step": 107158
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.9472991228103638,
      "learning_rate": 0.00033283244901460764,
      "loss": 2.1841,
      "step": 107159
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.8756827116012573,
      "learning_rate": 0.000332828383020414,
      "loss": 2.2286,
      "step": 107160
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.9287291169166565,
      "learning_rate": 0.0003328243170201169,
      "loss": 2.1098,
      "step": 107161
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.9339002966880798,
      "learning_rate": 0.0003328202510137171,
      "loss": 2.2918,
      "step": 107162
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.9430277347564697,
      "learning_rate": 0.00033281618500121546,
      "loss": 2.121,
      "step": 107163
    },
    {
      "epoch": 1.4,
      "grad_norm": 1.0845143795013428,
      "learning_rate": 0.00033281211898261265,
      "loss": 2.1833,
      "step": 107164
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.8611167073249817,
      "learning_rate": 0.0003328080529579094,
      "loss": 2.181,
      "step": 107165
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.9795612692832947,
      "learning_rate": 0.0003328039869271066,
      "loss": 2.1357,
      "step": 107166
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.8848904371261597,
      "learning_rate": 0.00033279992089020485,
      "loss": 2.0689,
      "step": 107167
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.9732447862625122,
      "learning_rate": 0.000332795854847205,
      "loss": 2.1219,
      "step": 107168
    },
    {
      "epoch": 1.4,
      "grad_norm": 1.0054938793182373,
      "learning_rate": 0.0003327917887981078,
      "loss": 2.1374,
      "step": 107169
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.8520838022232056,
      "learning_rate": 0.000332787722742914,
      "loss": 2.1506,
      "step": 107170
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.9235086441040039,
      "learning_rate": 0.00033278365668162424,
      "loss": 1.9177,
      "step": 107171
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.9011343717575073,
      "learning_rate": 0.00033277959061423954,
      "loss": 2.1635,
      "step": 107172
    },
    {
      "epoch": 1.4,
      "grad_norm": 1.0033178329467773,
      "learning_rate": 0.0003327755245407603,
      "loss": 2.1096,
      "step": 107173
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.8943896889686584,
      "learning_rate": 0.0003327714584611876,
      "loss": 2.0382,
      "step": 107174
    },
    {
      "epoch": 1.4,
      "grad_norm": 1.0183985233306885,
      "learning_rate": 0.0003327673923755221,
      "loss": 2.0674,
      "step": 107175
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.9744682908058167,
      "learning_rate": 0.00033276332628376446,
      "loss": 2.2593,
      "step": 107176
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.9267878532409668,
      "learning_rate": 0.0003327592601859155,
      "loss": 2.2534,
      "step": 107177
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.9052540063858032,
      "learning_rate": 0.00033275519408197603,
      "loss": 1.9264,
      "step": 107178
    },
    {
      "epoch": 1.4,
      "grad_norm": 1.3252310752868652,
      "learning_rate": 0.0003327511279719467,
      "loss": 2.271,
      "step": 107179
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.8981457948684692,
      "learning_rate": 0.0003327470618558283,
      "loss": 2.2324,
      "step": 107180
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.924632728099823,
      "learning_rate": 0.00033274299573362166,
      "loss": 2.2762,
      "step": 107181
    },
    {
      "epoch": 1.4,
      "grad_norm": 1.374284267425537,
      "learning_rate": 0.0003327389296053274,
      "loss": 2.0615,
      "step": 107182
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.8920811414718628,
      "learning_rate": 0.0003327348634709464,
      "loss": 2.0881,
      "step": 107183
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.8427188992500305,
      "learning_rate": 0.0003327307973304794,
      "loss": 2.2788,
      "step": 107184
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.9054439663887024,
      "learning_rate": 0.0003327267311839271,
      "loss": 2.2709,
      "step": 107185
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.9173508286476135,
      "learning_rate": 0.0003327226650312902,
      "loss": 2.2119,
      "step": 107186
    },
    {
      "epoch": 1.4,
      "grad_norm": 1.0774413347244263,
      "learning_rate": 0.00033271859887256967,
      "loss": 2.0206,
      "step": 107187
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.8948072791099548,
      "learning_rate": 0.0003327145327077661,
      "loss": 2.22,
      "step": 107188
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.9149422645568848,
      "learning_rate": 0.0003327104665368802,
      "loss": 2.3177,
      "step": 107189
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.9693366885185242,
      "learning_rate": 0.00033270640035991297,
      "loss": 2.4285,
      "step": 107190
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.8440714478492737,
      "learning_rate": 0.0003327023341768648,
      "loss": 2.0986,
      "step": 107191
    },
    {
      "epoch": 1.4,
      "grad_norm": 1.0696027278900146,
      "learning_rate": 0.00033269826798773674,
      "loss": 2.0964,
      "step": 107192
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.8667312860488892,
      "learning_rate": 0.00033269420179252954,
      "loss": 2.3237,
      "step": 107193
    },
    {
      "epoch": 1.4,
      "grad_norm": 1.007685661315918,
      "learning_rate": 0.0003326901355912438,
      "loss": 2.2291,
      "step": 107194
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.9538922309875488,
      "learning_rate": 0.0003326860693838803,
      "loss": 2.0821,
      "step": 107195
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.8397325873374939,
      "learning_rate": 0.0003326820031704399,
      "loss": 2.4293,
      "step": 107196
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.9472004175186157,
      "learning_rate": 0.00033267793695092334,
      "loss": 2.3786,
      "step": 107197
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.9788883924484253,
      "learning_rate": 0.0003326738707253312,
      "loss": 2.2171,
      "step": 107198
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.8247020244598389,
      "learning_rate": 0.00033266980449366446,
      "loss": 1.8267,
      "step": 107199
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.9904685020446777,
      "learning_rate": 0.00033266573825592384,
      "loss": 2.1329,
      "step": 107200
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.8768450617790222,
      "learning_rate": 0.00033266167201210996,
      "loss": 2.1467,
      "step": 107201
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.8230037689208984,
      "learning_rate": 0.0003326576057622237,
      "loss": 1.9407,
      "step": 107202
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.9063090085983276,
      "learning_rate": 0.0003326535395062657,
      "loss": 2.2007,
      "step": 107203
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.862407386302948,
      "learning_rate": 0.0003326494732442369,
      "loss": 2.0071,
      "step": 107204
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.9970992207527161,
      "learning_rate": 0.00033264540697613786,
      "loss": 2.5336,
      "step": 107205
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.9557850956916809,
      "learning_rate": 0.00033264134070196946,
      "loss": 2.2463,
      "step": 107206
    },
    {
      "epoch": 1.4,
      "grad_norm": 1.066071629524231,
      "learning_rate": 0.0003326372744217324,
      "loss": 2.3365,
      "step": 107207
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.8757259845733643,
      "learning_rate": 0.0003326332081354275,
      "loss": 1.9998,
      "step": 107208
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.9374480843544006,
      "learning_rate": 0.00033262914184305545,
      "loss": 1.9328,
      "step": 107209
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.8589695692062378,
      "learning_rate": 0.00033262507554461697,
      "loss": 2.4282,
      "step": 107210
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.9937874674797058,
      "learning_rate": 0.0003326210092401129,
      "loss": 2.1316,
      "step": 107211
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.9623659253120422,
      "learning_rate": 0.00033261694292954406,
      "loss": 2.2247,
      "step": 107212
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.8943698406219482,
      "learning_rate": 0.000332612876612911,
      "loss": 2.5475,
      "step": 107213
    },
    {
      "epoch": 1.4,
      "grad_norm": 1.0237082242965698,
      "learning_rate": 0.0003326088102902146,
      "loss": 2.1976,
      "step": 107214
    },
    {
      "epoch": 1.4,
      "grad_norm": 1.1885586977005005,
      "learning_rate": 0.0003326047439614557,
      "loss": 2.224,
      "step": 107215
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.9036431908607483,
      "learning_rate": 0.0003326006776266348,
      "loss": 2.1166,
      "step": 107216
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.8905921578407288,
      "learning_rate": 0.00033259661128575297,
      "loss": 2.2371,
      "step": 107217
    },
    {
      "epoch": 1.4,
      "grad_norm": 1.2348350286483765,
      "learning_rate": 0.0003325925449388108,
      "loss": 2.1529,
      "step": 107218
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.8556588292121887,
      "learning_rate": 0.00033258847858580904,
      "loss": 2.1654,
      "step": 107219
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.8307039737701416,
      "learning_rate": 0.0003325844122267484,
      "loss": 2.1353,
      "step": 107220
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.8553304672241211,
      "learning_rate": 0.00033258034586162977,
      "loss": 2.2078,
      "step": 107221
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.8825990557670593,
      "learning_rate": 0.0003325762794904538,
      "loss": 1.8815,
      "step": 107222
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.9205097556114197,
      "learning_rate": 0.00033257221311322135,
      "loss": 2.0498,
      "step": 107223
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.8194427490234375,
      "learning_rate": 0.00033256814672993305,
      "loss": 2.0717,
      "step": 107224
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.9101717472076416,
      "learning_rate": 0.00033256408034058974,
      "loss": 1.9965,
      "step": 107225
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.990200936794281,
      "learning_rate": 0.00033256001394519214,
      "loss": 2.0275,
      "step": 107226
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.8849873542785645,
      "learning_rate": 0.000332555947543741,
      "loss": 2.0974,
      "step": 107227
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.9535980820655823,
      "learning_rate": 0.00033255188113623716,
      "loss": 2.1092,
      "step": 107228
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.9367755651473999,
      "learning_rate": 0.00033254781472268123,
      "loss": 2.2097,
      "step": 107229
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.9264126420021057,
      "learning_rate": 0.0003325437483030741,
      "loss": 2.2531,
      "step": 107230
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.9612894058227539,
      "learning_rate": 0.0003325396818774164,
      "loss": 2.4297,
      "step": 107231
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.873676061630249,
      "learning_rate": 0.00033253561544570906,
      "loss": 2.1933,
      "step": 107232
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.9408944845199585,
      "learning_rate": 0.0003325315490079527,
      "loss": 2.018,
      "step": 107233
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.839249312877655,
      "learning_rate": 0.0003325274825641481,
      "loss": 2.0836,
      "step": 107234
    },
    {
      "epoch": 1.4,
      "grad_norm": 1.0110275745391846,
      "learning_rate": 0.000332523416114296,
      "loss": 2.1681,
      "step": 107235
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.9003468155860901,
      "learning_rate": 0.00033251934965839727,
      "loss": 2.2655,
      "step": 107236
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.8752057552337646,
      "learning_rate": 0.00033251528319645253,
      "loss": 1.9724,
      "step": 107237
    },
    {
      "epoch": 1.4,
      "grad_norm": 1.1129511594772339,
      "learning_rate": 0.0003325112167284625,
      "loss": 2.3333,
      "step": 107238
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.9668287038803101,
      "learning_rate": 0.0003325071502544282,
      "loss": 2.3722,
      "step": 107239
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.9199426174163818,
      "learning_rate": 0.0003325030837743501,
      "loss": 2.4524,
      "step": 107240
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.8758572936058044,
      "learning_rate": 0.00033249901728822905,
      "loss": 2.1721,
      "step": 107241
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.8826455473899841,
      "learning_rate": 0.0003324949507960658,
      "loss": 2.0665,
      "step": 107242
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.8317742943763733,
      "learning_rate": 0.0003324908842978612,
      "loss": 2.0047,
      "step": 107243
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.9242072105407715,
      "learning_rate": 0.00033248681779361587,
      "loss": 2.2229,
      "step": 107244
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.9337494373321533,
      "learning_rate": 0.0003324827512833307,
      "loss": 2.1956,
      "step": 107245
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.8916343450546265,
      "learning_rate": 0.00033247868476700627,
      "loss": 2.2657,
      "step": 107246
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.9814175963401794,
      "learning_rate": 0.0003324746182446435,
      "loss": 2.1142,
      "step": 107247
    },
    {
      "epoch": 1.4,
      "grad_norm": 1.0208739042282104,
      "learning_rate": 0.00033247055171624313,
      "loss": 2.1501,
      "step": 107248
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.8564399480819702,
      "learning_rate": 0.0003324664851818058,
      "loss": 2.1963,
      "step": 107249
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.8919188976287842,
      "learning_rate": 0.0003324624186413323,
      "loss": 2.3441,
      "step": 107250
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.8929471969604492,
      "learning_rate": 0.0003324583520948236,
      "loss": 2.1285,
      "step": 107251
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.8629993200302124,
      "learning_rate": 0.0003324542855422801,
      "loss": 2.1796,
      "step": 107252
    },
    {
      "epoch": 1.4,
      "grad_norm": 1.016608476638794,
      "learning_rate": 0.00033245021898370284,
      "loss": 2.2867,
      "step": 107253
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.9363292455673218,
      "learning_rate": 0.00033244615241909245,
      "loss": 2.2177,
      "step": 107254
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.9391738772392273,
      "learning_rate": 0.0003324420858484497,
      "loss": 2.2974,
      "step": 107255
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.8619625568389893,
      "learning_rate": 0.0003324380192717753,
      "loss": 2.0259,
      "step": 107256
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.8567590117454529,
      "learning_rate": 0.00033243395268907014,
      "loss": 2.0729,
      "step": 107257
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.9701572060585022,
      "learning_rate": 0.00033242988610033484,
      "loss": 2.0985,
      "step": 107258
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.9615484476089478,
      "learning_rate": 0.00033242581950557025,
      "loss": 2.2657,
      "step": 107259
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.9665728807449341,
      "learning_rate": 0.00033242175290477705,
      "loss": 1.9349,
      "step": 107260
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.883000910282135,
      "learning_rate": 0.000332417686297956,
      "loss": 2.1765,
      "step": 107261
    },
    {
      "epoch": 1.4,
      "grad_norm": 1.038733959197998,
      "learning_rate": 0.00033241361968510796,
      "loss": 2.3047,
      "step": 107262
    },
    {
      "epoch": 1.4,
      "grad_norm": 1.0526267290115356,
      "learning_rate": 0.0003324095530662336,
      "loss": 2.0109,
      "step": 107263
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.9466868042945862,
      "learning_rate": 0.0003324054864413337,
      "loss": 2.0295,
      "step": 107264
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.9301683902740479,
      "learning_rate": 0.00033240141981040906,
      "loss": 1.9652,
      "step": 107265
    },
    {
      "epoch": 1.4,
      "grad_norm": 1.0880199670791626,
      "learning_rate": 0.00033239735317346027,
      "loss": 2.2175,
      "step": 107266
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.9569319486618042,
      "learning_rate": 0.0003323932865304883,
      "loss": 2.1377,
      "step": 107267
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.9244626760482788,
      "learning_rate": 0.00033238921988149376,
      "loss": 2.0861,
      "step": 107268
    },
    {
      "epoch": 1.4,
      "grad_norm": 1.0247207880020142,
      "learning_rate": 0.0003323851532264774,
      "loss": 2.2453,
      "step": 107269
    },
    {
      "epoch": 1.4,
      "grad_norm": 1.0801268815994263,
      "learning_rate": 0.0003323810865654402,
      "loss": 2.3051,
      "step": 107270
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.8449658751487732,
      "learning_rate": 0.00033237701989838255,
      "loss": 1.8969,
      "step": 107271
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.8826913237571716,
      "learning_rate": 0.00033237295322530543,
      "loss": 1.8463,
      "step": 107272
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.9782616496086121,
      "learning_rate": 0.0003323688865462097,
      "loss": 2.1684,
      "step": 107273
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.8993188142776489,
      "learning_rate": 0.0003323648198610958,
      "loss": 2.108,
      "step": 107274
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.8576810359954834,
      "learning_rate": 0.0003323607531699648,
      "loss": 1.8318,
      "step": 107275
    },
    {
      "epoch": 1.4,
      "grad_norm": 1.0249451398849487,
      "learning_rate": 0.00033235668647281735,
      "loss": 2.1615,
      "step": 107276
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.9153081178665161,
      "learning_rate": 0.00033235261976965406,
      "loss": 2.1269,
      "step": 107277
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.9629676938056946,
      "learning_rate": 0.0003323485530604759,
      "loss": 2.1075,
      "step": 107278
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.9108864068984985,
      "learning_rate": 0.00033234448634528353,
      "loss": 2.0391,
      "step": 107279
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.9799332022666931,
      "learning_rate": 0.0003323404196240777,
      "loss": 2.1101,
      "step": 107280
    },
    {
      "epoch": 1.4,
      "grad_norm": 1.0451384782791138,
      "learning_rate": 0.0003323363528968591,
      "loss": 2.0059,
      "step": 107281
    },
    {
      "epoch": 1.4,
      "grad_norm": 1.092842698097229,
      "learning_rate": 0.00033233228616362867,
      "loss": 2.1655,
      "step": 107282
    },
    {
      "epoch": 1.4,
      "grad_norm": 1.0115485191345215,
      "learning_rate": 0.000332328219424387,
      "loss": 2.2836,
      "step": 107283
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.8220394849777222,
      "learning_rate": 0.0003323241526791349,
      "loss": 2.1969,
      "step": 107284
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.9130257368087769,
      "learning_rate": 0.0003323200859278732,
      "loss": 2.2608,
      "step": 107285
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.9243661165237427,
      "learning_rate": 0.0003323160191706025,
      "loss": 2.4064,
      "step": 107286
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.8915476202964783,
      "learning_rate": 0.0003323119524073237,
      "loss": 2.1409,
      "step": 107287
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.8944147825241089,
      "learning_rate": 0.0003323078856380375,
      "loss": 2.1261,
      "step": 107288
    },
    {
      "epoch": 1.4,
      "grad_norm": 1.040467619895935,
      "learning_rate": 0.0003323038188627446,
      "loss": 2.1314,
      "step": 107289
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.8590195775032043,
      "learning_rate": 0.00033229975208144577,
      "loss": 2.1066,
      "step": 107290
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.9446895718574524,
      "learning_rate": 0.00033229568529414194,
      "loss": 2.2898,
      "step": 107291
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.9258169531822205,
      "learning_rate": 0.00033229161850083363,
      "loss": 1.9446,
      "step": 107292
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.9075629115104675,
      "learning_rate": 0.0003322875517015217,
      "loss": 2.2106,
      "step": 107293
    },
    {
      "epoch": 1.4,
      "grad_norm": 1.0101231336593628,
      "learning_rate": 0.000332283484896207,
      "loss": 2.1211,
      "step": 107294
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.9013828039169312,
      "learning_rate": 0.0003322794180848901,
      "loss": 2.0986,
      "step": 107295
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.9766267538070679,
      "learning_rate": 0.0003322753512675719,
      "loss": 2.3634,
      "step": 107296
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.900840163230896,
      "learning_rate": 0.0003322712844442531,
      "loss": 2.2248,
      "step": 107297
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.8782782554626465,
      "learning_rate": 0.0003322672176149344,
      "loss": 2.2582,
      "step": 107298
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.8339619040489197,
      "learning_rate": 0.0003322631507796166,
      "loss": 1.9717,
      "step": 107299
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.8509660959243774,
      "learning_rate": 0.0003322590839383006,
      "loss": 2.2129,
      "step": 107300
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.8855876922607422,
      "learning_rate": 0.00033225501709098693,
      "loss": 2.0078,
      "step": 107301
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.877769410610199,
      "learning_rate": 0.00033225095023767644,
      "loss": 2.2393,
      "step": 107302
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.9538889527320862,
      "learning_rate": 0.00033224688337837,
      "loss": 2.0665,
      "step": 107303
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.9066206812858582,
      "learning_rate": 0.0003322428165130681,
      "loss": 2.1313,
      "step": 107304
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.8118975758552551,
      "learning_rate": 0.0003322387496417717,
      "loss": 1.9782,
      "step": 107305
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.858699381351471,
      "learning_rate": 0.0003322346827644816,
      "loss": 2.0875,
      "step": 107306
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.9008456468582153,
      "learning_rate": 0.00033223061588119834,
      "loss": 2.1386,
      "step": 107307
    },
    {
      "epoch": 1.4,
      "grad_norm": 1.192700982093811,
      "learning_rate": 0.00033222654899192286,
      "loss": 2.3051,
      "step": 107308
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.8933433890342712,
      "learning_rate": 0.0003322224820966559,
      "loss": 2.2705,
      "step": 107309
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.9574825167655945,
      "learning_rate": 0.0003322184151953981,
      "loss": 2.3107,
      "step": 107310
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.8859197497367859,
      "learning_rate": 0.0003322143482881502,
      "loss": 1.9906,
      "step": 107311
    },
    {
      "epoch": 1.4,
      "grad_norm": 1.0176899433135986,
      "learning_rate": 0.0003322102813749132,
      "loss": 2.4762,
      "step": 107312
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.9108718037605286,
      "learning_rate": 0.0003322062144556877,
      "loss": 2.1016,
      "step": 107313
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.8682255148887634,
      "learning_rate": 0.0003322021475304744,
      "loss": 2.1215,
      "step": 107314
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.9499485492706299,
      "learning_rate": 0.0003321980805992741,
      "loss": 2.2478,
      "step": 107315
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.907397449016571,
      "learning_rate": 0.0003321940136620876,
      "loss": 2.097,
      "step": 107316
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.9021600484848022,
      "learning_rate": 0.0003321899467189156,
      "loss": 2.1636,
      "step": 107317
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.9607886075973511,
      "learning_rate": 0.0003321858797697589,
      "loss": 2.2122,
      "step": 107318
    },
    {
      "epoch": 1.4,
      "grad_norm": 1.267532467842102,
      "learning_rate": 0.00033218181281461826,
      "loss": 2.0637,
      "step": 107319
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.8505947589874268,
      "learning_rate": 0.0003321777458534944,
      "loss": 2.094,
      "step": 107320
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.8312457203865051,
      "learning_rate": 0.0003321736788863881,
      "loss": 1.8648,
      "step": 107321
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.9218612313270569,
      "learning_rate": 0.00033216961191330003,
      "loss": 2.4382,
      "step": 107322
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.8436973094940186,
      "learning_rate": 0.0003321655449342311,
      "loss": 2.4461,
      "step": 107323
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.9824755787849426,
      "learning_rate": 0.00033216147794918195,
      "loss": 2.2516,
      "step": 107324
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.976159930229187,
      "learning_rate": 0.0003321574109581534,
      "loss": 1.9789,
      "step": 107325
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.9818938970565796,
      "learning_rate": 0.00033215334396114615,
      "loss": 1.9132,
      "step": 107326
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.8840223550796509,
      "learning_rate": 0.000332149276958161,
      "loss": 2.1067,
      "step": 107327
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.8734919428825378,
      "learning_rate": 0.0003321452099491987,
      "loss": 2.1406,
      "step": 107328
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.8607740998268127,
      "learning_rate": 0.00033214114293426,
      "loss": 2.1708,
      "step": 107329
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.8573445677757263,
      "learning_rate": 0.00033213707591334563,
      "loss": 2.0316,
      "step": 107330
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.887083888053894,
      "learning_rate": 0.00033213300888645643,
      "loss": 1.9799,
      "step": 107331
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.8086350560188293,
      "learning_rate": 0.00033212894185359306,
      "loss": 2.0428,
      "step": 107332
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.9066353440284729,
      "learning_rate": 0.0003321248748147563,
      "loss": 2.2763,
      "step": 107333
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.9009888172149658,
      "learning_rate": 0.00033212080776994694,
      "loss": 2.1135,
      "step": 107334
    },
    {
      "epoch": 1.4,
      "grad_norm": 1.075365662574768,
      "learning_rate": 0.00033211674071916567,
      "loss": 2.1037,
      "step": 107335
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.887747049331665,
      "learning_rate": 0.0003321126736624133,
      "loss": 2.006,
      "step": 107336
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.9163615703582764,
      "learning_rate": 0.0003321086065996907,
      "loss": 2.1979,
      "step": 107337
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.9281124472618103,
      "learning_rate": 0.0003321045395309984,
      "loss": 2.2941,
      "step": 107338
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.9239251613616943,
      "learning_rate": 0.00033210047245633723,
      "loss": 1.9422,
      "step": 107339
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.913895845413208,
      "learning_rate": 0.0003320964053757081,
      "loss": 2.1635,
      "step": 107340
    },
    {
      "epoch": 1.4,
      "grad_norm": 1.079716682434082,
      "learning_rate": 0.0003320923382891115,
      "loss": 2.1552,
      "step": 107341
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.9286401271820068,
      "learning_rate": 0.00033208827119654837,
      "loss": 2.2337,
      "step": 107342
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.8645438551902771,
      "learning_rate": 0.0003320842040980195,
      "loss": 1.9051,
      "step": 107343
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.876764714717865,
      "learning_rate": 0.00033208013699352547,
      "loss": 1.9498,
      "step": 107344
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.898411214351654,
      "learning_rate": 0.0003320760698830672,
      "loss": 1.9666,
      "step": 107345
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.8720080256462097,
      "learning_rate": 0.00033207200276664535,
      "loss": 2.1838,
      "step": 107346
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.8694899082183838,
      "learning_rate": 0.00033206793564426076,
      "loss": 2.0629,
      "step": 107347
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.9333947896957397,
      "learning_rate": 0.00033206386851591404,
      "loss": 2.1447,
      "step": 107348
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.8763803243637085,
      "learning_rate": 0.00033205980138160615,
      "loss": 2.2037,
      "step": 107349
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.9900644421577454,
      "learning_rate": 0.00033205573424133764,
      "loss": 2.1852,
      "step": 107350
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.925433337688446,
      "learning_rate": 0.00033205166709510944,
      "loss": 2.0839,
      "step": 107351
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.9154631495475769,
      "learning_rate": 0.00033204759994292225,
      "loss": 2.0622,
      "step": 107352
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.8250280022621155,
      "learning_rate": 0.00033204353278477677,
      "loss": 2.1578,
      "step": 107353
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.8387013077735901,
      "learning_rate": 0.0003320394656206737,
      "loss": 2.33,
      "step": 107354
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.8678967952728271,
      "learning_rate": 0.00033203539845061403,
      "loss": 2.3909,
      "step": 107355
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.8891915082931519,
      "learning_rate": 0.00033203133127459824,
      "loss": 1.9326,
      "step": 107356
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.8538893461227417,
      "learning_rate": 0.0003320272640926273,
      "loss": 2.1019,
      "step": 107357
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.9313010573387146,
      "learning_rate": 0.00033202319690470194,
      "loss": 2.0997,
      "step": 107358
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.8814405798912048,
      "learning_rate": 0.0003320191297108228,
      "loss": 2.3118,
      "step": 107359
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.9071831107139587,
      "learning_rate": 0.00033201506251099064,
      "loss": 2.4663,
      "step": 107360
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.952298104763031,
      "learning_rate": 0.0003320109953052064,
      "loss": 2.2544,
      "step": 107361
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.8148847818374634,
      "learning_rate": 0.0003320069280934707,
      "loss": 2.0162,
      "step": 107362
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.9397063255310059,
      "learning_rate": 0.0003320028608757842,
      "loss": 2.1273,
      "step": 107363
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.8946309685707092,
      "learning_rate": 0.0003319987936521479,
      "loss": 2.2723,
      "step": 107364
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.8595229983329773,
      "learning_rate": 0.0003319947264225623,
      "loss": 2.3734,
      "step": 107365
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.8467739820480347,
      "learning_rate": 0.00033199065918702826,
      "loss": 2.0854,
      "step": 107366
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.9523979425430298,
      "learning_rate": 0.0003319865919455467,
      "loss": 2.2447,
      "step": 107367
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.8733259439468384,
      "learning_rate": 0.00033198252469811816,
      "loss": 2.2117,
      "step": 107368
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.9488372802734375,
      "learning_rate": 0.0003319784574447434,
      "loss": 2.2312,
      "step": 107369
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.9032377004623413,
      "learning_rate": 0.0003319743901854233,
      "loss": 2.0893,
      "step": 107370
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.8478891849517822,
      "learning_rate": 0.0003319703229201585,
      "loss": 2.3697,
      "step": 107371
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.9169663190841675,
      "learning_rate": 0.00033196625564894987,
      "loss": 2.1363,
      "step": 107372
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.8419005870819092,
      "learning_rate": 0.00033196218837179813,
      "loss": 1.9521,
      "step": 107373
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.8538004159927368,
      "learning_rate": 0.00033195812108870404,
      "loss": 2.1647,
      "step": 107374
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.947124719619751,
      "learning_rate": 0.0003319540537996682,
      "loss": 2.1484,
      "step": 107375
    },
    {
      "epoch": 1.4,
      "grad_norm": 1.0364584922790527,
      "learning_rate": 0.00033194998650469163,
      "loss": 2.0877,
      "step": 107376
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.8645282983779907,
      "learning_rate": 0.00033194591920377494,
      "loss": 2.194,
      "step": 107377
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.8527110815048218,
      "learning_rate": 0.0003319418518969188,
      "loss": 2.2812,
      "step": 107378
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.9432849287986755,
      "learning_rate": 0.00033193778458412415,
      "loss": 2.131,
      "step": 107379
    },
    {
      "epoch": 1.4,
      "grad_norm": 1.1345853805541992,
      "learning_rate": 0.00033193371726539163,
      "loss": 2.332,
      "step": 107380
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.8367570638656616,
      "learning_rate": 0.000331929649940722,
      "loss": 2.0327,
      "step": 107381
    },
    {
      "epoch": 1.4,
      "grad_norm": 1.0406938791275024,
      "learning_rate": 0.0003319255826101162,
      "loss": 2.0325,
      "step": 107382
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.9710463285446167,
      "learning_rate": 0.00033192151527357467,
      "loss": 2.072,
      "step": 107383
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.9150031208992004,
      "learning_rate": 0.0003319174479310983,
      "loss": 2.1545,
      "step": 107384
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.9071357846260071,
      "learning_rate": 0.00033191338058268806,
      "loss": 2.2906,
      "step": 107385
    },
    {
      "epoch": 1.4,
      "grad_norm": 1.0952669382095337,
      "learning_rate": 0.0003319093132283443,
      "loss": 1.9904,
      "step": 107386
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.8935889601707458,
      "learning_rate": 0.0003319052458680681,
      "loss": 2.1786,
      "step": 107387
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.844864547252655,
      "learning_rate": 0.00033190117850186016,
      "loss": 2.1825,
      "step": 107388
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.9594845175743103,
      "learning_rate": 0.00033189711112972115,
      "loss": 2.1466,
      "step": 107389
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.946999192237854,
      "learning_rate": 0.00033189304375165187,
      "loss": 2.181,
      "step": 107390
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.8458777666091919,
      "learning_rate": 0.0003318889763676531,
      "loss": 1.929,
      "step": 107391
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.8109493851661682,
      "learning_rate": 0.0003318849089777255,
      "loss": 2.0808,
      "step": 107392
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.779590368270874,
      "learning_rate": 0.0003318808415818699,
      "loss": 2.0887,
      "step": 107393
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.8480624556541443,
      "learning_rate": 0.0003318767741800871,
      "loss": 2.1694,
      "step": 107394
    },
    {
      "epoch": 1.4,
      "grad_norm": 1.0251615047454834,
      "learning_rate": 0.00033187270677237775,
      "loss": 2.1859,
      "step": 107395
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.8600510954856873,
      "learning_rate": 0.0003318686393587427,
      "loss": 1.983,
      "step": 107396
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.8366572260856628,
      "learning_rate": 0.0003318645719391826,
      "loss": 2.2534,
      "step": 107397
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.8817541003227234,
      "learning_rate": 0.0003318605045136984,
      "loss": 2.1804,
      "step": 107398
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.8976021409034729,
      "learning_rate": 0.00033185643708229066,
      "loss": 2.1548,
      "step": 107399
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.8795929551124573,
      "learning_rate": 0.0003318523696449602,
      "loss": 2.5373,
      "step": 107400
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.9536616802215576,
      "learning_rate": 0.0003318483022017078,
      "loss": 2.2615,
      "step": 107401
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.8593329191207886,
      "learning_rate": 0.0003318442347525342,
      "loss": 2.0401,
      "step": 107402
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.9459431767463684,
      "learning_rate": 0.0003318401672974401,
      "loss": 1.9664,
      "step": 107403
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.8752529621124268,
      "learning_rate": 0.00033183609983642636,
      "loss": 2.2431,
      "step": 107404
    },
    {
      "epoch": 1.4,
      "grad_norm": 1.0520490407943726,
      "learning_rate": 0.00033183203236949375,
      "loss": 2.2948,
      "step": 107405
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.9409931302070618,
      "learning_rate": 0.00033182796489664287,
      "loss": 2.1783,
      "step": 107406
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.93227618932724,
      "learning_rate": 0.0003318238974178746,
      "loss": 1.8826,
      "step": 107407
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.8837616443634033,
      "learning_rate": 0.0003318198299331897,
      "loss": 2.2136,
      "step": 107408
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.9442088007926941,
      "learning_rate": 0.00033181576244258883,
      "loss": 2.2032,
      "step": 107409
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.8918901681900024,
      "learning_rate": 0.0003318116949460728,
      "loss": 2.2438,
      "step": 107410
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.9180275201797485,
      "learning_rate": 0.00033180762744364244,
      "loss": 2.1224,
      "step": 107411
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.9542666673660278,
      "learning_rate": 0.00033180355993529847,
      "loss": 2.0452,
      "step": 107412
    },
    {
      "epoch": 1.4,
      "grad_norm": 2.0369083881378174,
      "learning_rate": 0.0003317994924210415,
      "loss": 2.1808,
      "step": 107413
    },
    {
      "epoch": 1.4,
      "grad_norm": 1.0473319292068481,
      "learning_rate": 0.0003317954249008725,
      "loss": 2.242,
      "step": 107414
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.9163020849227905,
      "learning_rate": 0.00033179135737479204,
      "loss": 2.1951,
      "step": 107415
    },
    {
      "epoch": 1.4,
      "grad_norm": 1.0668936967849731,
      "learning_rate": 0.00033178728984280107,
      "loss": 2.2404,
      "step": 107416
    },
    {
      "epoch": 1.4,
      "grad_norm": 1.1523414850234985,
      "learning_rate": 0.0003317832223049002,
      "loss": 1.908,
      "step": 107417
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.9524844288825989,
      "learning_rate": 0.0003317791547610902,
      "loss": 2.2109,
      "step": 107418
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.8475098609924316,
      "learning_rate": 0.00033177508721137194,
      "loss": 2.1748,
      "step": 107419
    },
    {
      "epoch": 1.4,
      "grad_norm": 1.0350449085235596,
      "learning_rate": 0.000331771019655746,
      "loss": 2.3013,
      "step": 107420
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.9770719408988953,
      "learning_rate": 0.00033176695209421325,
      "loss": 2.1961,
      "step": 107421
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.9422669410705566,
      "learning_rate": 0.0003317628845267745,
      "loss": 2.1824,
      "step": 107422
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.9616259336471558,
      "learning_rate": 0.00033175881695343034,
      "loss": 2.3147,
      "step": 107423
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.9066694378852844,
      "learning_rate": 0.00033175474937418165,
      "loss": 1.9996,
      "step": 107424
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.9989470839500427,
      "learning_rate": 0.00033175068178902916,
      "loss": 2.0376,
      "step": 107425
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.9871692061424255,
      "learning_rate": 0.00033174661419797357,
      "loss": 2.056,
      "step": 107426
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.9453673362731934,
      "learning_rate": 0.00033174254660101575,
      "loss": 2.18,
      "step": 107427
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.889167308807373,
      "learning_rate": 0.0003317384789981564,
      "loss": 2.0904,
      "step": 107428
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.8949268460273743,
      "learning_rate": 0.00033173441138939616,
      "loss": 2.0376,
      "step": 107429
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.8963636755943298,
      "learning_rate": 0.000331730343774736,
      "loss": 1.8975,
      "step": 107430
    },
    {
      "epoch": 1.4,
      "grad_norm": 1.182693362236023,
      "learning_rate": 0.0003317262761541766,
      "loss": 2.1042,
      "step": 107431
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.8383427262306213,
      "learning_rate": 0.0003317222085277186,
      "loss": 2.0605,
      "step": 107432
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.9597041606903076,
      "learning_rate": 0.0003317181408953628,
      "loss": 2.2106,
      "step": 107433
    },
    {
      "epoch": 1.4,
      "grad_norm": 1.132257103919983,
      "learning_rate": 0.00033171407325711017,
      "loss": 1.8716,
      "step": 107434
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.912116527557373,
      "learning_rate": 0.0003317100056129611,
      "loss": 2.3327,
      "step": 107435
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.8950542211532593,
      "learning_rate": 0.00033170593796291666,
      "loss": 2.0287,
      "step": 107436
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.9150840044021606,
      "learning_rate": 0.00033170187030697753,
      "loss": 1.9914,
      "step": 107437
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.8848912119865417,
      "learning_rate": 0.0003316978026451443,
      "loss": 2.1443,
      "step": 107438
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.9074572920799255,
      "learning_rate": 0.0003316937349774179,
      "loss": 1.9779,
      "step": 107439
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.8564225435256958,
      "learning_rate": 0.000331689667303799,
      "loss": 2.2078,
      "step": 107440
    },
    {
      "epoch": 1.4,
      "grad_norm": 1.0494670867919922,
      "learning_rate": 0.00033168559962428845,
      "loss": 2.2116,
      "step": 107441
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.9202081561088562,
      "learning_rate": 0.0003316815319388869,
      "loss": 2.251,
      "step": 107442
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.8856181502342224,
      "learning_rate": 0.00033167746424759523,
      "loss": 2.3335,
      "step": 107443
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.8953697085380554,
      "learning_rate": 0.0003316733965504141,
      "loss": 2.314,
      "step": 107444
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.884790301322937,
      "learning_rate": 0.00033166932884734425,
      "loss": 2.1085,
      "step": 107445
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.9759727716445923,
      "learning_rate": 0.0003316652611383865,
      "loss": 2.0619,
      "step": 107446
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.9902263879776001,
      "learning_rate": 0.00033166119342354153,
      "loss": 2.0725,
      "step": 107447
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.8905193209648132,
      "learning_rate": 0.00033165712570281015,
      "loss": 2.0995,
      "step": 107448
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.8701704740524292,
      "learning_rate": 0.00033165305797619324,
      "loss": 2.1752,
      "step": 107449
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.8255254626274109,
      "learning_rate": 0.00033164899024369133,
      "loss": 2.2793,
      "step": 107450
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.8693220019340515,
      "learning_rate": 0.00033164492250530524,
      "loss": 2.3148,
      "step": 107451
    },
    {
      "epoch": 1.4,
      "grad_norm": 1.0277801752090454,
      "learning_rate": 0.00033164085476103584,
      "loss": 2.1908,
      "step": 107452
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.8749437928199768,
      "learning_rate": 0.0003316367870108837,
      "loss": 2.2684,
      "step": 107453
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.9652124643325806,
      "learning_rate": 0.0003316327192548498,
      "loss": 2.19,
      "step": 107454
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.9430580735206604,
      "learning_rate": 0.00033162865149293473,
      "loss": 2.306,
      "step": 107455
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.9101949334144592,
      "learning_rate": 0.00033162458372513933,
      "loss": 1.9907,
      "step": 107456
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.9686787724494934,
      "learning_rate": 0.00033162051595146425,
      "loss": 2.0647,
      "step": 107457
    },
    {
      "epoch": 1.4,
      "grad_norm": 1.0067002773284912,
      "learning_rate": 0.00033161644817191046,
      "loss": 1.9731,
      "step": 107458
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.8718030452728271,
      "learning_rate": 0.0003316123803864784,
      "loss": 2.0435,
      "step": 107459
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.9670677185058594,
      "learning_rate": 0.0003316083125951691,
      "loss": 2.2042,
      "step": 107460
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.9648498296737671,
      "learning_rate": 0.0003316042447979832,
      "loss": 2.1518,
      "step": 107461
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.9565796852111816,
      "learning_rate": 0.00033160017699492147,
      "loss": 2.4083,
      "step": 107462
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.9289007782936096,
      "learning_rate": 0.00033159610918598466,
      "loss": 2.2831,
      "step": 107463
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.9444945454597473,
      "learning_rate": 0.00033159204137117354,
      "loss": 2.076,
      "step": 107464
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.9539023637771606,
      "learning_rate": 0.0003315879735504889,
      "loss": 2.1137,
      "step": 107465
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.8887772560119629,
      "learning_rate": 0.00033158390572393144,
      "loss": 2.3141,
      "step": 107466
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.9540774822235107,
      "learning_rate": 0.00033157983789150187,
      "loss": 2.1582,
      "step": 107467
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.9349396228790283,
      "learning_rate": 0.00033157577005320113,
      "loss": 2.2809,
      "step": 107468
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.8642511367797852,
      "learning_rate": 0.0003315717022090298,
      "loss": 2.2114,
      "step": 107469
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.9604918360710144,
      "learning_rate": 0.0003315676343589887,
      "loss": 2.1217,
      "step": 107470
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.8609734177589417,
      "learning_rate": 0.00033156356650307855,
      "loss": 2.1066,
      "step": 107471
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.8926195502281189,
      "learning_rate": 0.0003315594986413002,
      "loss": 2.0201,
      "step": 107472
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.8502361178398132,
      "learning_rate": 0.00033155543077365425,
      "loss": 2.2735,
      "step": 107473
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.8747310042381287,
      "learning_rate": 0.00033155136290014166,
      "loss": 2.1485,
      "step": 107474
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.8356794714927673,
      "learning_rate": 0.00033154729502076304,
      "loss": 2.2923,
      "step": 107475
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.8477950096130371,
      "learning_rate": 0.0003315432271355191,
      "loss": 2.2715,
      "step": 107476
    },
    {
      "epoch": 1.4,
      "grad_norm": 1.1274033784866333,
      "learning_rate": 0.0003315391592444108,
      "loss": 2.1316,
      "step": 107477
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.896742582321167,
      "learning_rate": 0.00033153509134743866,
      "loss": 2.1623,
      "step": 107478
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.9529361128807068,
      "learning_rate": 0.0003315310234446036,
      "loss": 2.1185,
      "step": 107479
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.9085919857025146,
      "learning_rate": 0.0003315269555359064,
      "loss": 2.1262,
      "step": 107480
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.9666039943695068,
      "learning_rate": 0.0003315228876213477,
      "loss": 2.1717,
      "step": 107481
    },
    {
      "epoch": 1.4,
      "grad_norm": 1.0161134004592896,
      "learning_rate": 0.0003315188197009282,
      "loss": 2.1858,
      "step": 107482
    },
    {
      "epoch": 1.4,
      "grad_norm": 1.0345052480697632,
      "learning_rate": 0.0003315147517746489,
      "loss": 2.0507,
      "step": 107483
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.8836507201194763,
      "learning_rate": 0.0003315106838425103,
      "loss": 2.1551,
      "step": 107484
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.9046567678451538,
      "learning_rate": 0.00033150661590451334,
      "loss": 2.1567,
      "step": 107485
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.9145501852035522,
      "learning_rate": 0.0003315025479606587,
      "loss": 2.1184,
      "step": 107486
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.9143242835998535,
      "learning_rate": 0.0003314984800109471,
      "loss": 1.8755,
      "step": 107487
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.9470338225364685,
      "learning_rate": 0.00033149441205537934,
      "loss": 2.3468,
      "step": 107488
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.9344531297683716,
      "learning_rate": 0.0003314903440939563,
      "loss": 2.1793,
      "step": 107489
    },
    {
      "epoch": 1.4,
      "grad_norm": 1.062570333480835,
      "learning_rate": 0.00033148627612667847,
      "loss": 2.1765,
      "step": 107490
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.9476791620254517,
      "learning_rate": 0.0003314822081535468,
      "loss": 2.2663,
      "step": 107491
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.9512296319007874,
      "learning_rate": 0.0003314781401745619,
      "loss": 2.0403,
      "step": 107492
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.8560313582420349,
      "learning_rate": 0.0003314740721897248,
      "loss": 2.1033,
      "step": 107493
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.8152588605880737,
      "learning_rate": 0.00033147000419903597,
      "loss": 1.9597,
      "step": 107494
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.9135557413101196,
      "learning_rate": 0.0003314659362024962,
      "loss": 2.2382,
      "step": 107495
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.8121302127838135,
      "learning_rate": 0.0003314618682001064,
      "loss": 2.0996,
      "step": 107496
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.8334968090057373,
      "learning_rate": 0.00033145780019186726,
      "loss": 2.3218,
      "step": 107497
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.9404900670051575,
      "learning_rate": 0.00033145373217777945,
      "loss": 2.1148,
      "step": 107498
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.9937604069709778,
      "learning_rate": 0.0003314496641578439,
      "loss": 2.1796,
      "step": 107499
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.9153819680213928,
      "learning_rate": 0.00033144559613206123,
      "loss": 2.2708,
      "step": 107500
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.9037567973136902,
      "learning_rate": 0.00033144152810043217,
      "loss": 2.3694,
      "step": 107501
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.9397082328796387,
      "learning_rate": 0.00033143746006295753,
      "loss": 2.3052,
      "step": 107502
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.884888231754303,
      "learning_rate": 0.00033143339201963817,
      "loss": 2.2315,
      "step": 107503
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.9175670146942139,
      "learning_rate": 0.00033142932397047474,
      "loss": 2.2727,
      "step": 107504
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.8955758810043335,
      "learning_rate": 0.0003314252559154679,
      "loss": 2.0372,
      "step": 107505
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.8791843056678772,
      "learning_rate": 0.00033142118785461856,
      "loss": 2.0427,
      "step": 107506
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.8037012815475464,
      "learning_rate": 0.00033141711978792754,
      "loss": 2.1857,
      "step": 107507
    },
    {
      "epoch": 1.4,
      "grad_norm": 1.0230244398117065,
      "learning_rate": 0.0003314130517153954,
      "loss": 2.1236,
      "step": 107508
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.9248667359352112,
      "learning_rate": 0.00033140898363702285,
      "loss": 2.0794,
      "step": 107509
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.8412101864814758,
      "learning_rate": 0.00033140491555281096,
      "loss": 2.0156,
      "step": 107510
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.8303079009056091,
      "learning_rate": 0.00033140084746276026,
      "loss": 2.1643,
      "step": 107511
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.9105386734008789,
      "learning_rate": 0.0003313967793668715,
      "loss": 2.0622,
      "step": 107512
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.9390608072280884,
      "learning_rate": 0.0003313927112651456,
      "loss": 2.0378,
      "step": 107513
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.9684833884239197,
      "learning_rate": 0.0003313886431575831,
      "loss": 2.2483,
      "step": 107514
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.9463142156600952,
      "learning_rate": 0.0003313845750441848,
      "loss": 2.0982,
      "step": 107515
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.8844488263130188,
      "learning_rate": 0.00033138050692495166,
      "loss": 2.1904,
      "step": 107516
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.8633354902267456,
      "learning_rate": 0.00033137643879988416,
      "loss": 2.0203,
      "step": 107517
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.9381723999977112,
      "learning_rate": 0.0003313723706689833,
      "loss": 2.1315,
      "step": 107518
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.8803773522377014,
      "learning_rate": 0.0003313683025322497,
      "loss": 2.2677,
      "step": 107519
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.848446786403656,
      "learning_rate": 0.00033136423438968405,
      "loss": 1.9291,
      "step": 107520
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.838390588760376,
      "learning_rate": 0.00033136016624128723,
      "loss": 2.1794,
      "step": 107521
    },
    {
      "epoch": 1.4,
      "grad_norm": 1.004857063293457,
      "learning_rate": 0.0003313560980870601,
      "loss": 2.0318,
      "step": 107522
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.8776109218597412,
      "learning_rate": 0.00033135202992700313,
      "loss": 2.1211,
      "step": 107523
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.827269434928894,
      "learning_rate": 0.0003313479617611173,
      "loss": 2.1477,
      "step": 107524
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.9510936737060547,
      "learning_rate": 0.00033134389358940327,
      "loss": 2.1862,
      "step": 107525
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.8768898248672485,
      "learning_rate": 0.00033133982541186185,
      "loss": 2.2907,
      "step": 107526
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.8980127573013306,
      "learning_rate": 0.00033133575722849365,
      "loss": 1.9733,
      "step": 107527
    },
    {
      "epoch": 1.4,
      "grad_norm": 1.3316476345062256,
      "learning_rate": 0.00033133168903929966,
      "loss": 2.046,
      "step": 107528
    },
    {
      "epoch": 1.4,
      "grad_norm": 1.0860666036605835,
      "learning_rate": 0.00033132762084428047,
      "loss": 2.1428,
      "step": 107529
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.9425156712532043,
      "learning_rate": 0.0003313235526434369,
      "loss": 2.3657,
      "step": 107530
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.9164997935295105,
      "learning_rate": 0.0003313194844367696,
      "loss": 2.2063,
      "step": 107531
    },
    {
      "epoch": 1.4,
      "grad_norm": 1.127787709236145,
      "learning_rate": 0.0003313154162242796,
      "loss": 2.2796,
      "step": 107532
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.937290370464325,
      "learning_rate": 0.0003313113480059673,
      "loss": 2.0962,
      "step": 107533
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.8547398447990417,
      "learning_rate": 0.00033130727978183375,
      "loss": 2.329,
      "step": 107534
    },
    {
      "epoch": 1.4,
      "grad_norm": 1.0849242210388184,
      "learning_rate": 0.00033130321155187955,
      "loss": 2.1825,
      "step": 107535
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.8566309213638306,
      "learning_rate": 0.0003312991433161054,
      "loss": 2.1627,
      "step": 107536
    },
    {
      "epoch": 1.4,
      "grad_norm": 1.0671887397766113,
      "learning_rate": 0.0003312950750745122,
      "loss": 2.0374,
      "step": 107537
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.8418931365013123,
      "learning_rate": 0.00033129100682710073,
      "loss": 1.9127,
      "step": 107538
    },
    {
      "epoch": 1.4,
      "grad_norm": 1.1462318897247314,
      "learning_rate": 0.00033128693857387164,
      "loss": 2.0911,
      "step": 107539
    },
    {
      "epoch": 1.4,
      "grad_norm": 1.0534147024154663,
      "learning_rate": 0.00033128287031482565,
      "loss": 2.2173,
      "step": 107540
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.9545023441314697,
      "learning_rate": 0.00033127880204996367,
      "loss": 2.1085,
      "step": 107541
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.9460635781288147,
      "learning_rate": 0.00033127473377928636,
      "loss": 2.2709,
      "step": 107542
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.9478428363800049,
      "learning_rate": 0.00033127066550279447,
      "loss": 2.1711,
      "step": 107543
    },
    {
      "epoch": 1.4,
      "grad_norm": 1.1117300987243652,
      "learning_rate": 0.00033126659722048876,
      "loss": 2.2384,
      "step": 107544
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.9503247737884521,
      "learning_rate": 0.00033126252893237,
      "loss": 2.121,
      "step": 107545
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.8546181321144104,
      "learning_rate": 0.0003312584606384389,
      "loss": 2.1716,
      "step": 107546
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.8491435647010803,
      "learning_rate": 0.0003312543923386964,
      "loss": 2.4124,
      "step": 107547
    },
    {
      "epoch": 1.4,
      "grad_norm": 1.246209740638733,
      "learning_rate": 0.000331250324033143,
      "loss": 2.1916,
      "step": 107548
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.9635222554206848,
      "learning_rate": 0.0003312462557217796,
      "loss": 2.0946,
      "step": 107549
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.8656010627746582,
      "learning_rate": 0.00033124218740460695,
      "loss": 1.9693,
      "step": 107550
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.9872825145721436,
      "learning_rate": 0.0003312381190816257,
      "loss": 2.1398,
      "step": 107551
    },
    {
      "epoch": 1.4,
      "grad_norm": 1.174027681350708,
      "learning_rate": 0.0003312340507528368,
      "loss": 2.2017,
      "step": 107552
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.8911834955215454,
      "learning_rate": 0.00033122998241824094,
      "loss": 2.2909,
      "step": 107553
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.9630509614944458,
      "learning_rate": 0.0003312259140778387,
      "loss": 1.9982,
      "step": 107554
    },
    {
      "epoch": 1.4,
      "grad_norm": 1.1477645635604858,
      "learning_rate": 0.0003312218457316311,
      "loss": 1.9476,
      "step": 107555
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.8857994675636292,
      "learning_rate": 0.0003312177773796187,
      "loss": 2.0751,
      "step": 107556
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.9607238173484802,
      "learning_rate": 0.0003312137090218023,
      "loss": 2.1463,
      "step": 107557
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.8254221081733704,
      "learning_rate": 0.0003312096406581827,
      "loss": 2.0563,
      "step": 107558
    },
    {
      "epoch": 1.4,
      "grad_norm": 1.0228426456451416,
      "learning_rate": 0.0003312055722887607,
      "loss": 2.1776,
      "step": 107559
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.8333329558372498,
      "learning_rate": 0.0003312015039135369,
      "loss": 2.1217,
      "step": 107560
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.8849629163742065,
      "learning_rate": 0.00033119743553251216,
      "loss": 2.0213,
      "step": 107561
    },
    {
      "epoch": 1.4,
      "grad_norm": 1.0413269996643066,
      "learning_rate": 0.00033119336714568736,
      "loss": 2.2102,
      "step": 107562
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.8899047374725342,
      "learning_rate": 0.000331189298753063,
      "loss": 2.2939,
      "step": 107563
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.8741581439971924,
      "learning_rate": 0.00033118523035464,
      "loss": 2.0605,
      "step": 107564
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.9620802402496338,
      "learning_rate": 0.00033118116195041906,
      "loss": 2.2229,
      "step": 107565
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.856388509273529,
      "learning_rate": 0.000331177093540401,
      "loss": 1.9822,
      "step": 107566
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.9251256585121155,
      "learning_rate": 0.0003311730251245864,
      "loss": 2.1218,
      "step": 107567
    },
    {
      "epoch": 1.4,
      "grad_norm": 1.0644665956497192,
      "learning_rate": 0.00033116895670297635,
      "loss": 2.0677,
      "step": 107568
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.9261651039123535,
      "learning_rate": 0.00033116488827557126,
      "loss": 2.0656,
      "step": 107569
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.8678356409072876,
      "learning_rate": 0.00033116081984237204,
      "loss": 2.1956,
      "step": 107570
    },
    {
      "epoch": 1.4,
      "grad_norm": 1.061102271080017,
      "learning_rate": 0.0003311567514033795,
      "loss": 2.0645,
      "step": 107571
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.8288306593894958,
      "learning_rate": 0.0003311526829585943,
      "loss": 2.0681,
      "step": 107572
    },
    {
      "epoch": 1.4,
      "grad_norm": 1.2763594388961792,
      "learning_rate": 0.00033114861450801714,
      "loss": 2.0606,
      "step": 107573
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.9407132267951965,
      "learning_rate": 0.000331144546051649,
      "loss": 2.2991,
      "step": 107574
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.8531593680381775,
      "learning_rate": 0.0003311404775894904,
      "loss": 2.3992,
      "step": 107575
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.8892054557800293,
      "learning_rate": 0.00033113640912154225,
      "loss": 2.0948,
      "step": 107576
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.8562685251235962,
      "learning_rate": 0.00033113234064780525,
      "loss": 2.0511,
      "step": 107577
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.8480265736579895,
      "learning_rate": 0.0003311282721682802,
      "loss": 2.0638,
      "step": 107578
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.8541269302368164,
      "learning_rate": 0.0003311242036829677,
      "loss": 2.1981,
      "step": 107579
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.9856419563293457,
      "learning_rate": 0.00033112013519186873,
      "loss": 2.0744,
      "step": 107580
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.949731707572937,
      "learning_rate": 0.0003311160666949839,
      "loss": 2.0439,
      "step": 107581
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.9492773413658142,
      "learning_rate": 0.000331111998192314,
      "loss": 2.039,
      "step": 107582
    },
    {
      "epoch": 1.4,
      "grad_norm": 1.1077593564987183,
      "learning_rate": 0.00033110792968385976,
      "loss": 2.2164,
      "step": 107583
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.9231317043304443,
      "learning_rate": 0.00033110386116962196,
      "loss": 2.191,
      "step": 107584
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.8926500082015991,
      "learning_rate": 0.0003310997926496015,
      "loss": 2.1677,
      "step": 107585
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.9789733290672302,
      "learning_rate": 0.0003310957241237989,
      "loss": 2.0529,
      "step": 107586
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.9562223553657532,
      "learning_rate": 0.000331091655592215,
      "loss": 2.0485,
      "step": 107587
    },
    {
      "epoch": 1.4,
      "grad_norm": 1.0027421712875366,
      "learning_rate": 0.00033108758705485057,
      "loss": 1.9923,
      "step": 107588
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.8988752365112305,
      "learning_rate": 0.00033108351851170647,
      "loss": 2.1741,
      "step": 107589
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.8868988156318665,
      "learning_rate": 0.0003310794499627833,
      "loss": 2.5566,
      "step": 107590
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.9832762479782104,
      "learning_rate": 0.00033107538140808183,
      "loss": 2.1317,
      "step": 107591
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.8830057978630066,
      "learning_rate": 0.00033107131284760297,
      "loss": 2.1217,
      "step": 107592
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.8984032273292542,
      "learning_rate": 0.00033106724428134726,
      "loss": 2.0913,
      "step": 107593
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.8754139542579651,
      "learning_rate": 0.00033106317570931557,
      "loss": 2.2241,
      "step": 107594
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.8426833152770996,
      "learning_rate": 0.00033105910713150866,
      "loss": 2.0932,
      "step": 107595
    },
    {
      "epoch": 1.4,
      "grad_norm": 1.0742011070251465,
      "learning_rate": 0.00033105503854792735,
      "loss": 2.2554,
      "step": 107596
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.8573343753814697,
      "learning_rate": 0.0003310509699585722,
      "loss": 2.1542,
      "step": 107597
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.8530012369155884,
      "learning_rate": 0.00033104690136344416,
      "loss": 2.228,
      "step": 107598
    },
    {
      "epoch": 1.4,
      "grad_norm": 1.037920594215393,
      "learning_rate": 0.00033104283276254396,
      "loss": 1.9951,
      "step": 107599
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.8761951327323914,
      "learning_rate": 0.00033103876415587223,
      "loss": 2.2402,
      "step": 107600
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.8874178528785706,
      "learning_rate": 0.00033103469554342983,
      "loss": 2.4077,
      "step": 107601
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.8974960446357727,
      "learning_rate": 0.00033103062692521753,
      "loss": 2.3339,
      "step": 107602
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.8210716843605042,
      "learning_rate": 0.000331026558301236,
      "loss": 2.2172,
      "step": 107603
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.8711641430854797,
      "learning_rate": 0.0003310224896714861,
      "loss": 2.3506,
      "step": 107604
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.8904229998588562,
      "learning_rate": 0.0003310184210359685,
      "loss": 2.1301,
      "step": 107605
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.8906047940254211,
      "learning_rate": 0.000331014352394684,
      "loss": 2.4426,
      "step": 107606
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.879662036895752,
      "learning_rate": 0.00033101028374763326,
      "loss": 2.2456,
      "step": 107607
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.8627097010612488,
      "learning_rate": 0.0003310062150948173,
      "loss": 2.2261,
      "step": 107608
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.9391369223594666,
      "learning_rate": 0.00033100214643623656,
      "loss": 1.923,
      "step": 107609
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.8703442215919495,
      "learning_rate": 0.000330998077771892,
      "loss": 1.9591,
      "step": 107610
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.931583821773529,
      "learning_rate": 0.0003309940091017843,
      "loss": 1.953,
      "step": 107611
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.9185729622840881,
      "learning_rate": 0.00033098994042591424,
      "loss": 2.3207,
      "step": 107612
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.940560519695282,
      "learning_rate": 0.0003309858717442825,
      "loss": 2.0424,
      "step": 107613
    },
    {
      "epoch": 1.4,
      "grad_norm": 1.016602873802185,
      "learning_rate": 0.00033098180305689,
      "loss": 2.3402,
      "step": 107614
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.9281715154647827,
      "learning_rate": 0.0003309777343637373,
      "loss": 2.3273,
      "step": 107615
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.9037227034568787,
      "learning_rate": 0.0003309736656648253,
      "loss": 2.3537,
      "step": 107616
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.938803493976593,
      "learning_rate": 0.0003309695969601547,
      "loss": 2.1701,
      "step": 107617
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.9468309283256531,
      "learning_rate": 0.0003309655282497263,
      "loss": 2.0281,
      "step": 107618
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.9878635406494141,
      "learning_rate": 0.0003309614595335408,
      "loss": 2.0869,
      "step": 107619
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.8767794966697693,
      "learning_rate": 0.000330957390811599,
      "loss": 1.9598,
      "step": 107620
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.871577799320221,
      "learning_rate": 0.0003309533220839016,
      "loss": 2.0719,
      "step": 107621
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.954507052898407,
      "learning_rate": 0.0003309492533504494,
      "loss": 2.1186,
      "step": 107622
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.9676472544670105,
      "learning_rate": 0.0003309451846112432,
      "loss": 2.149,
      "step": 107623
    },
    {
      "epoch": 1.4,
      "grad_norm": 1.056180477142334,
      "learning_rate": 0.00033094111586628365,
      "loss": 1.9987,
      "step": 107624
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.8798641562461853,
      "learning_rate": 0.00033093704711557154,
      "loss": 2.0532,
      "step": 107625
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.9913492798805237,
      "learning_rate": 0.0003309329783591078,
      "loss": 2.1732,
      "step": 107626
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.9237061142921448,
      "learning_rate": 0.00033092890959689284,
      "loss": 1.939,
      "step": 107627
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.8721116185188293,
      "learning_rate": 0.00033092484082892766,
      "loss": 1.9757,
      "step": 107628
    },
    {
      "epoch": 1.4,
      "grad_norm": 1.0167641639709473,
      "learning_rate": 0.00033092077205521306,
      "loss": 2.1425,
      "step": 107629
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.885092556476593,
      "learning_rate": 0.00033091670327574965,
      "loss": 2.294,
      "step": 107630
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.9615945219993591,
      "learning_rate": 0.0003309126344905382,
      "loss": 2.2745,
      "step": 107631
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.9746133685112,
      "learning_rate": 0.0003309085656995796,
      "loss": 2.2099,
      "step": 107632
    },
    {
      "epoch": 1.4,
      "grad_norm": 1.0872774124145508,
      "learning_rate": 0.00033090449690287445,
      "loss": 2.2961,
      "step": 107633
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.9229221343994141,
      "learning_rate": 0.00033090042810042357,
      "loss": 2.1368,
      "step": 107634
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.8857308626174927,
      "learning_rate": 0.00033089635929222776,
      "loss": 1.9956,
      "step": 107635
    },
    {
      "epoch": 1.4,
      "grad_norm": 1.0036662817001343,
      "learning_rate": 0.0003308922904782877,
      "loss": 2.1763,
      "step": 107636
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.8637824058532715,
      "learning_rate": 0.00033088822165860417,
      "loss": 2.2902,
      "step": 107637
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.8878174424171448,
      "learning_rate": 0.00033088415283317796,
      "loss": 2.1891,
      "step": 107638
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.9177708029747009,
      "learning_rate": 0.0003308800840020098,
      "loss": 2.1706,
      "step": 107639
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.9346484541893005,
      "learning_rate": 0.00033087601516510037,
      "loss": 2.2635,
      "step": 107640
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.8585402965545654,
      "learning_rate": 0.00033087194632245063,
      "loss": 2.0627,
      "step": 107641
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.9332046508789062,
      "learning_rate": 0.0003308678774740612,
      "loss": 2.1341,
      "step": 107642
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.971505343914032,
      "learning_rate": 0.0003308638086199327,
      "loss": 2.2431,
      "step": 107643
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.9406254291534424,
      "learning_rate": 0.0003308597397600662,
      "loss": 1.9995,
      "step": 107644
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.8992322087287903,
      "learning_rate": 0.00033085567089446223,
      "loss": 2.1271,
      "step": 107645
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.8908255696296692,
      "learning_rate": 0.0003308516020231216,
      "loss": 2.1118,
      "step": 107646
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.8430687785148621,
      "learning_rate": 0.0003308475331460451,
      "loss": 2.1856,
      "step": 107647
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.8435870409011841,
      "learning_rate": 0.0003308434642632334,
      "loss": 2.2886,
      "step": 107648
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.8835589289665222,
      "learning_rate": 0.0003308393953746874,
      "loss": 2.2088,
      "step": 107649
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.9304400682449341,
      "learning_rate": 0.0003308353264804078,
      "loss": 2.3049,
      "step": 107650
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.8847489356994629,
      "learning_rate": 0.00033083125758039524,
      "loss": 2.0126,
      "step": 107651
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.8954277634620667,
      "learning_rate": 0.00033082718867465056,
      "loss": 2.163,
      "step": 107652
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.8912068009376526,
      "learning_rate": 0.00033082311976317464,
      "loss": 2.1656,
      "step": 107653
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.8873984813690186,
      "learning_rate": 0.000330819050845968,
      "loss": 2.0766,
      "step": 107654
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.946408748626709,
      "learning_rate": 0.00033081498192303157,
      "loss": 2.2976,
      "step": 107655
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.8805640339851379,
      "learning_rate": 0.0003308109129943661,
      "loss": 2.0778,
      "step": 107656
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.9782811403274536,
      "learning_rate": 0.0003308068440599722,
      "loss": 2.2792,
      "step": 107657
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.9490643739700317,
      "learning_rate": 0.0003308027751198508,
      "loss": 2.3066,
      "step": 107658
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.9863924384117126,
      "learning_rate": 0.00033079870617400256,
      "loss": 2.2407,
      "step": 107659
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.92078697681427,
      "learning_rate": 0.00033079463722242823,
      "loss": 2.2634,
      "step": 107660
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.9579569697380066,
      "learning_rate": 0.0003307905682651286,
      "loss": 2.2716,
      "step": 107661
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.8659576177597046,
      "learning_rate": 0.0003307864993021044,
      "loss": 2.2381,
      "step": 107662
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.9427602291107178,
      "learning_rate": 0.0003307824303333565,
      "loss": 2.2206,
      "step": 107663
    },
    {
      "epoch": 1.4,
      "grad_norm": 1.0476810932159424,
      "learning_rate": 0.0003307783613588855,
      "loss": 1.9874,
      "step": 107664
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.8842645287513733,
      "learning_rate": 0.0003307742923786922,
      "loss": 1.9296,
      "step": 107665
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.8552092909812927,
      "learning_rate": 0.00033077022339277745,
      "loss": 2.1123,
      "step": 107666
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.8905211091041565,
      "learning_rate": 0.0003307661544011419,
      "loss": 1.8426,
      "step": 107667
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.9704659581184387,
      "learning_rate": 0.0003307620854037863,
      "loss": 2.161,
      "step": 107668
    },
    {
      "epoch": 1.4,
      "grad_norm": 1.0166893005371094,
      "learning_rate": 0.00033075801640071147,
      "loss": 2.2351,
      "step": 107669
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.8759713768959045,
      "learning_rate": 0.00033075394739191826,
      "loss": 2.3327,
      "step": 107670
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.8641865849494934,
      "learning_rate": 0.0003307498783774072,
      "loss": 2.1749,
      "step": 107671
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.8622090816497803,
      "learning_rate": 0.0003307458093571791,
      "loss": 2.303,
      "step": 107672
    },
    {
      "epoch": 1.4,
      "grad_norm": 1.0546602010726929,
      "learning_rate": 0.0003307417403312349,
      "loss": 2.0644,
      "step": 107673
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.9818538427352905,
      "learning_rate": 0.0003307376712995752,
      "loss": 2.1759,
      "step": 107674
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.8642311096191406,
      "learning_rate": 0.0003307336022622007,
      "loss": 2.1479,
      "step": 107675
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.885714590549469,
      "learning_rate": 0.00033072953321911224,
      "loss": 1.9652,
      "step": 107676
    },
    {
      "epoch": 1.4,
      "grad_norm": 1.3715914487838745,
      "learning_rate": 0.0003307254641703107,
      "loss": 2.3465,
      "step": 107677
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.9209262132644653,
      "learning_rate": 0.00033072139511579666,
      "loss": 2.1385,
      "step": 107678
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.949250340461731,
      "learning_rate": 0.0003307173260555709,
      "loss": 2.2219,
      "step": 107679
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.9498229622840881,
      "learning_rate": 0.0003307132569896342,
      "loss": 1.9729,
      "step": 107680
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.8992311358451843,
      "learning_rate": 0.0003307091879179874,
      "loss": 2.4654,
      "step": 107681
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.903492271900177,
      "learning_rate": 0.00033070511884063117,
      "loss": 2.2543,
      "step": 107682
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.840143084526062,
      "learning_rate": 0.00033070104975756626,
      "loss": 2.2984,
      "step": 107683
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.8885307312011719,
      "learning_rate": 0.00033069698066879346,
      "loss": 2.0684,
      "step": 107684
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.9112576842308044,
      "learning_rate": 0.0003306929115743135,
      "loss": 2.1318,
      "step": 107685
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.9393675327301025,
      "learning_rate": 0.0003306888424741271,
      "loss": 2.2946,
      "step": 107686
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.8667438626289368,
      "learning_rate": 0.00033068477336823515,
      "loss": 2.018,
      "step": 107687
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.8962125182151794,
      "learning_rate": 0.0003306807042566383,
      "loss": 1.8864,
      "step": 107688
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.9579212665557861,
      "learning_rate": 0.00033067663513933725,
      "loss": 1.987,
      "step": 107689
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.9231646656990051,
      "learning_rate": 0.00033067256601633293,
      "loss": 1.9974,
      "step": 107690
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.9899562001228333,
      "learning_rate": 0.00033066849688762593,
      "loss": 2.1813,
      "step": 107691
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.9426532983779907,
      "learning_rate": 0.0003306644277532171,
      "loss": 2.229,
      "step": 107692
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.9128355979919434,
      "learning_rate": 0.0003306603586131072,
      "loss": 2.3984,
      "step": 107693
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.8901107311248779,
      "learning_rate": 0.00033065628946729695,
      "loss": 2.0721,
      "step": 107694
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.8691154718399048,
      "learning_rate": 0.00033065222031578706,
      "loss": 2.3927,
      "step": 107695
    },
    {
      "epoch": 1.4,
      "grad_norm": 1.0578272342681885,
      "learning_rate": 0.0003306481511585785,
      "loss": 2.0438,
      "step": 107696
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.8547021150588989,
      "learning_rate": 0.00033064408199567175,
      "loss": 2.2369,
      "step": 107697
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.8284894824028015,
      "learning_rate": 0.0003306400128270677,
      "loss": 2.201,
      "step": 107698
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.917941689491272,
      "learning_rate": 0.0003306359436527671,
      "loss": 2.2658,
      "step": 107699
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.89969402551651,
      "learning_rate": 0.0003306318744727707,
      "loss": 2.1792,
      "step": 107700
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.8573833107948303,
      "learning_rate": 0.0003306278052870792,
      "loss": 2.2927,
      "step": 107701
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.8872530460357666,
      "learning_rate": 0.00033062373609569345,
      "loss": 1.861,
      "step": 107702
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.917902410030365,
      "learning_rate": 0.0003306196668986143,
      "loss": 2.2702,
      "step": 107703
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.8569161891937256,
      "learning_rate": 0.0003306155976958422,
      "loss": 2.3548,
      "step": 107704
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.9014899134635925,
      "learning_rate": 0.0003306115284873781,
      "loss": 2.0133,
      "step": 107705
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.8984004855155945,
      "learning_rate": 0.0003306074592732228,
      "loss": 2.2123,
      "step": 107706
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.9545167088508606,
      "learning_rate": 0.000330603390053377,
      "loss": 2.1779,
      "step": 107707
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.8535210490226746,
      "learning_rate": 0.0003305993208278414,
      "loss": 2.1117,
      "step": 107708
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.9624866247177124,
      "learning_rate": 0.00033059525159661685,
      "loss": 2.2001,
      "step": 107709
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.9237391352653503,
      "learning_rate": 0.00033059118235970403,
      "loss": 2.2016,
      "step": 107710
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.9023614525794983,
      "learning_rate": 0.0003305871131171038,
      "loss": 2.0927,
      "step": 107711
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.8308365345001221,
      "learning_rate": 0.0003305830438688168,
      "loss": 2.1433,
      "step": 107712
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.8423336148262024,
      "learning_rate": 0.0003305789746148438,
      "loss": 2.246,
      "step": 107713
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.8884781002998352,
      "learning_rate": 0.00033057490535518566,
      "loss": 2.2735,
      "step": 107714
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.8570695519447327,
      "learning_rate": 0.000330570836089843,
      "loss": 2.1662,
      "step": 107715
    },
    {
      "epoch": 1.4,
      "grad_norm": 1.0155162811279297,
      "learning_rate": 0.0003305667668188167,
      "loss": 2.1895,
      "step": 107716
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.9958300590515137,
      "learning_rate": 0.0003305626975421075,
      "loss": 2.0944,
      "step": 107717
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.8084706664085388,
      "learning_rate": 0.00033055862825971606,
      "loss": 2.1438,
      "step": 107718
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.9790553450584412,
      "learning_rate": 0.00033055455897164316,
      "loss": 2.2111,
      "step": 107719
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.9267316460609436,
      "learning_rate": 0.00033055048967788964,
      "loss": 1.9429,
      "step": 107720
    },
    {
      "epoch": 1.4,
      "grad_norm": 1.0708049535751343,
      "learning_rate": 0.00033054642037845614,
      "loss": 2.3355,
      "step": 107721
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.9233191609382629,
      "learning_rate": 0.00033054235107334354,
      "loss": 2.1311,
      "step": 107722
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.9984055757522583,
      "learning_rate": 0.0003305382817625526,
      "loss": 2.1133,
      "step": 107723
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.9211373925209045,
      "learning_rate": 0.00033053421244608395,
      "loss": 2.2965,
      "step": 107724
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.9233425259590149,
      "learning_rate": 0.00033053014312393837,
      "loss": 1.9231,
      "step": 107725
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.9158897995948792,
      "learning_rate": 0.00033052607379611677,
      "loss": 2.1621,
      "step": 107726
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.9003127813339233,
      "learning_rate": 0.0003305220044626197,
      "loss": 1.9679,
      "step": 107727
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.9403506517410278,
      "learning_rate": 0.0003305179351234481,
      "loss": 2.3419,
      "step": 107728
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.8791427612304688,
      "learning_rate": 0.0003305138657786025,
      "loss": 2.1475,
      "step": 107729
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.9705556035041809,
      "learning_rate": 0.00033050979642808395,
      "loss": 2.1109,
      "step": 107730
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.8470805287361145,
      "learning_rate": 0.0003305057270718929,
      "loss": 2.2298,
      "step": 107731
    },
    {
      "epoch": 1.4,
      "grad_norm": 1.0102802515029907,
      "learning_rate": 0.0003305016577100304,
      "loss": 2.1843,
      "step": 107732
    },
    {
      "epoch": 1.4,
      "grad_norm": 1.0169321298599243,
      "learning_rate": 0.000330497588342497,
      "loss": 2.3127,
      "step": 107733
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.8113622665405273,
      "learning_rate": 0.00033049351896929355,
      "loss": 2.2673,
      "step": 107734
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.9145401120185852,
      "learning_rate": 0.0003304894495904208,
      "loss": 2.2317,
      "step": 107735
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.8193338513374329,
      "learning_rate": 0.0003304853802058794,
      "loss": 2.0804,
      "step": 107736
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.9129905700683594,
      "learning_rate": 0.0003304813108156702,
      "loss": 2.2573,
      "step": 107737
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.866124153137207,
      "learning_rate": 0.000330477241419794,
      "loss": 2.1494,
      "step": 107738
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.9802135825157166,
      "learning_rate": 0.00033047317201825155,
      "loss": 2.0098,
      "step": 107739
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.767380952835083,
      "learning_rate": 0.0003304691026110435,
      "loss": 2.0346,
      "step": 107740
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.9627434611320496,
      "learning_rate": 0.00033046503319817067,
      "loss": 2.2992,
      "step": 107741
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.8862755298614502,
      "learning_rate": 0.0003304609637796338,
      "loss": 2.1128,
      "step": 107742
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.8608644008636475,
      "learning_rate": 0.0003304568943554337,
      "loss": 2.0876,
      "step": 107743
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.8498750329017639,
      "learning_rate": 0.00033045282492557106,
      "loss": 2.2846,
      "step": 107744
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.9296244382858276,
      "learning_rate": 0.0003304487554900467,
      "loss": 2.2084,
      "step": 107745
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.9083082675933838,
      "learning_rate": 0.00033044468604886127,
      "loss": 2.2512,
      "step": 107746
    },
    {
      "epoch": 1.4,
      "grad_norm": 1.0217372179031372,
      "learning_rate": 0.00033044061660201564,
      "loss": 2.1428,
      "step": 107747
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.8946401476860046,
      "learning_rate": 0.00033043654714951063,
      "loss": 2.0994,
      "step": 107748
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.8499002456665039,
      "learning_rate": 0.0003304324776913468,
      "loss": 2.1747,
      "step": 107749
    },
    {
      "epoch": 1.4,
      "grad_norm": 1.0014443397521973,
      "learning_rate": 0.00033042840822752493,
      "loss": 2.3475,
      "step": 107750
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.9328103065490723,
      "learning_rate": 0.0003304243387580459,
      "loss": 2.1388,
      "step": 107751
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.872222900390625,
      "learning_rate": 0.0003304202692829105,
      "loss": 2.1662,
      "step": 107752
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.9262091517448425,
      "learning_rate": 0.00033041619980211923,
      "loss": 2.0829,
      "step": 107753
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.9300075173377991,
      "learning_rate": 0.00033041213031567316,
      "loss": 2.2192,
      "step": 107754
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.7866128087043762,
      "learning_rate": 0.0003304080608235729,
      "loss": 2.1421,
      "step": 107755
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.9949888586997986,
      "learning_rate": 0.00033040399132581907,
      "loss": 2.4895,
      "step": 107756
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.9210785627365112,
      "learning_rate": 0.0003303999218224126,
      "loss": 2.133,
      "step": 107757
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.9191768765449524,
      "learning_rate": 0.00033039585231335426,
      "loss": 2.2389,
      "step": 107758
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.9033174514770508,
      "learning_rate": 0.00033039178279864473,
      "loss": 2.1084,
      "step": 107759
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.89586341381073,
      "learning_rate": 0.0003303877132782848,
      "loss": 1.9616,
      "step": 107760
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.8214660286903381,
      "learning_rate": 0.00033038364375227524,
      "loss": 2.0402,
      "step": 107761
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.9063019752502441,
      "learning_rate": 0.0003303795742206168,
      "loss": 2.1572,
      "step": 107762
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.9135559797286987,
      "learning_rate": 0.0003303755046833102,
      "loss": 2.2729,
      "step": 107763
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.9192730784416199,
      "learning_rate": 0.0003303714351403562,
      "loss": 2.266,
      "step": 107764
    },
    {
      "epoch": 1.4,
      "grad_norm": 1.0495651960372925,
      "learning_rate": 0.0003303673655917557,
      "loss": 2.0959,
      "step": 107765
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.9887930750846863,
      "learning_rate": 0.00033036329603750914,
      "loss": 2.1115,
      "step": 107766
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.9537707567214966,
      "learning_rate": 0.00033035922647761754,
      "loss": 2.1058,
      "step": 107767
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.9626094102859497,
      "learning_rate": 0.00033035515691208164,
      "loss": 2.0017,
      "step": 107768
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.885115921497345,
      "learning_rate": 0.00033035108734090214,
      "loss": 2.179,
      "step": 107769
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.942348062992096,
      "learning_rate": 0.00033034701776407974,
      "loss": 2.0915,
      "step": 107770
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.860966145992279,
      "learning_rate": 0.0003303429481816153,
      "loss": 1.9948,
      "step": 107771
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.8687440752983093,
      "learning_rate": 0.0003303388785935096,
      "loss": 2.1568,
      "step": 107772
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.9009369015693665,
      "learning_rate": 0.00033033480899976323,
      "loss": 1.8974,
      "step": 107773
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.8605125546455383,
      "learning_rate": 0.000330330739400377,
      "loss": 2.0518,
      "step": 107774
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.9746959209442139,
      "learning_rate": 0.00033032666979535187,
      "loss": 2.148,
      "step": 107775
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.7925148606300354,
      "learning_rate": 0.0003303226001846883,
      "loss": 2.2437,
      "step": 107776
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.9457053542137146,
      "learning_rate": 0.00033031853056838726,
      "loss": 2.029,
      "step": 107777
    },
    {
      "epoch": 1.4,
      "grad_norm": 1.104036808013916,
      "learning_rate": 0.0003303144609464495,
      "loss": 1.976,
      "step": 107778
    },
    {
      "epoch": 1.4,
      "grad_norm": 1.5860698223114014,
      "learning_rate": 0.0003303103913188755,
      "loss": 2.165,
      "step": 107779
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.9131898283958435,
      "learning_rate": 0.0003303063216856664,
      "loss": 2.0502,
      "step": 107780
    },
    {
      "epoch": 1.4,
      "grad_norm": 1.004010796546936,
      "learning_rate": 0.00033030225204682274,
      "loss": 2.1392,
      "step": 107781
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.9483758807182312,
      "learning_rate": 0.0003302981824023453,
      "loss": 1.9928,
      "step": 107782
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.8427128195762634,
      "learning_rate": 0.0003302941127522348,
      "loss": 1.8968,
      "step": 107783
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.8499931693077087,
      "learning_rate": 0.0003302900430964922,
      "loss": 2.1882,
      "step": 107784
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.9965794682502747,
      "learning_rate": 0.000330285973435118,
      "loss": 2.0775,
      "step": 107785
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.9253382682800293,
      "learning_rate": 0.00033028190376811306,
      "loss": 2.1881,
      "step": 107786
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.8463160395622253,
      "learning_rate": 0.00033027783409547826,
      "loss": 1.9974,
      "step": 107787
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.8842825889587402,
      "learning_rate": 0.0003302737644172141,
      "loss": 2.2391,
      "step": 107788
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.9513768553733826,
      "learning_rate": 0.0003302696947333215,
      "loss": 2.2489,
      "step": 107789
    },
    {
      "epoch": 1.4,
      "grad_norm": 1.0108989477157593,
      "learning_rate": 0.0003302656250438013,
      "loss": 2.2148,
      "step": 107790
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.9288374781608582,
      "learning_rate": 0.00033026155534865404,
      "loss": 2.1858,
      "step": 107791
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.8977051973342896,
      "learning_rate": 0.0003302574856478806,
      "loss": 2.1965,
      "step": 107792
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.9270327687263489,
      "learning_rate": 0.0003302534159414818,
      "loss": 2.1635,
      "step": 107793
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.8319073915481567,
      "learning_rate": 0.0003302493462294582,
      "loss": 2.2237,
      "step": 107794
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.8971366286277771,
      "learning_rate": 0.00033024527651181074,
      "loss": 2.3133,
      "step": 107795
    },
    {
      "epoch": 1.4,
      "grad_norm": 1.0126861333847046,
      "learning_rate": 0.00033024120678854007,
      "loss": 2.0986,
      "step": 107796
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.8514066934585571,
      "learning_rate": 0.0003302371370596471,
      "loss": 2.3119,
      "step": 107797
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.8780657052993774,
      "learning_rate": 0.0003302330673251324,
      "loss": 2.0622,
      "step": 107798
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.8718441724777222,
      "learning_rate": 0.00033022899758499677,
      "loss": 2.2294,
      "step": 107799
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.8951879143714905,
      "learning_rate": 0.00033022492783924104,
      "loss": 1.9828,
      "step": 107800
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.8825219869613647,
      "learning_rate": 0.00033022085808786595,
      "loss": 2.0061,
      "step": 107801
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.8878625631332397,
      "learning_rate": 0.00033021678833087214,
      "loss": 2.1061,
      "step": 107802
    },
    {
      "epoch": 1.4,
      "grad_norm": 1.2123416662216187,
      "learning_rate": 0.00033021271856826053,
      "loss": 2.2649,
      "step": 107803
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.9448791146278381,
      "learning_rate": 0.0003302086488000318,
      "loss": 2.14,
      "step": 107804
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.8292258977890015,
      "learning_rate": 0.0003302045790261867,
      "loss": 2.1022,
      "step": 107805
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.9182083010673523,
      "learning_rate": 0.000330200509246726,
      "loss": 1.8946,
      "step": 107806
    },
    {
      "epoch": 1.4,
      "grad_norm": 1.0200650691986084,
      "learning_rate": 0.0003301964394616505,
      "loss": 2.4226,
      "step": 107807
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.8673064112663269,
      "learning_rate": 0.0003301923696709608,
      "loss": 2.2609,
      "step": 107808
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.8716528415679932,
      "learning_rate": 0.00033018829987465796,
      "loss": 2.0763,
      "step": 107809
    },
    {
      "epoch": 1.4,
      "grad_norm": 1.0008984804153442,
      "learning_rate": 0.0003301842300727424,
      "loss": 2.2198,
      "step": 107810
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.8972683548927307,
      "learning_rate": 0.000330180160265215,
      "loss": 2.2176,
      "step": 107811
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.8176690340042114,
      "learning_rate": 0.00033017609045207666,
      "loss": 2.2494,
      "step": 107812
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.9701607823371887,
      "learning_rate": 0.0003301720206333279,
      "loss": 2.2702,
      "step": 107813
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.8563412427902222,
      "learning_rate": 0.00033016795080896956,
      "loss": 2.054,
      "step": 107814
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.8997110724449158,
      "learning_rate": 0.0003301638809790026,
      "loss": 2.4551,
      "step": 107815
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.9236973524093628,
      "learning_rate": 0.00033015981114342744,
      "loss": 2.2776,
      "step": 107816
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.9821558594703674,
      "learning_rate": 0.00033015574130224507,
      "loss": 2.4331,
      "step": 107817
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.8424767255783081,
      "learning_rate": 0.0003301516714554562,
      "loss": 2.1445,
      "step": 107818
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.8192883133888245,
      "learning_rate": 0.0003301476016030615,
      "loss": 2.0802,
      "step": 107819
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.8789733648300171,
      "learning_rate": 0.00033014353174506186,
      "loss": 2.1022,
      "step": 107820
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.9794433116912842,
      "learning_rate": 0.000330139461881458,
      "loss": 2.2024,
      "step": 107821
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.868168830871582,
      "learning_rate": 0.0003301353920122505,
      "loss": 2.2945,
      "step": 107822
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.8714686632156372,
      "learning_rate": 0.00033013132213744037,
      "loss": 2.1208,
      "step": 107823
    },
    {
      "epoch": 1.4,
      "grad_norm": 1.179561972618103,
      "learning_rate": 0.0003301272522570283,
      "loss": 2.1986,
      "step": 107824
    },
    {
      "epoch": 1.4,
      "grad_norm": 1.0713579654693604,
      "learning_rate": 0.000330123182371015,
      "loss": 2.1329,
      "step": 107825
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.9090813994407654,
      "learning_rate": 0.0003301191124794011,
      "loss": 2.0311,
      "step": 107826
    },
    {
      "epoch": 1.4,
      "grad_norm": 1.2318042516708374,
      "learning_rate": 0.0003301150425821876,
      "loss": 2.17,
      "step": 107827
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.8328928351402283,
      "learning_rate": 0.0003301109726793751,
      "loss": 2.1062,
      "step": 107828
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.9325256943702698,
      "learning_rate": 0.00033010690277096436,
      "loss": 2.096,
      "step": 107829
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.8117227554321289,
      "learning_rate": 0.0003301028328569563,
      "loss": 2.3727,
      "step": 107830
    },
    {
      "epoch": 1.4,
      "grad_norm": 1.5235596895217896,
      "learning_rate": 0.00033009876293735145,
      "loss": 2.0906,
      "step": 107831
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.9433001279830933,
      "learning_rate": 0.00033009469301215066,
      "loss": 2.3125,
      "step": 107832
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.9024372100830078,
      "learning_rate": 0.00033009062308135485,
      "loss": 2.1079,
      "step": 107833
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.9301201105117798,
      "learning_rate": 0.00033008655314496445,
      "loss": 2.2526,
      "step": 107834
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.9157330989837646,
      "learning_rate": 0.00033008248320298044,
      "loss": 2.1283,
      "step": 107835
    },
    {
      "epoch": 1.4,
      "grad_norm": 1.0829236507415771,
      "learning_rate": 0.0003300784132554036,
      "loss": 2.1743,
      "step": 107836
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.9281128644943237,
      "learning_rate": 0.00033007434330223454,
      "loss": 2.2103,
      "step": 107837
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.9731569290161133,
      "learning_rate": 0.0003300702733434741,
      "loss": 1.9812,
      "step": 107838
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.915429949760437,
      "learning_rate": 0.0003300662033791231,
      "loss": 2.22,
      "step": 107839
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.8688449263572693,
      "learning_rate": 0.0003300621334091821,
      "loss": 2.1861,
      "step": 107840
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.8745405673980713,
      "learning_rate": 0.00033005806343365204,
      "loss": 2.2434,
      "step": 107841
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.8615485429763794,
      "learning_rate": 0.00033005399345253364,
      "loss": 2.2193,
      "step": 107842
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.8725729584693909,
      "learning_rate": 0.00033004992346582763,
      "loss": 2.0527,
      "step": 107843
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.9281947612762451,
      "learning_rate": 0.0003300458534735348,
      "loss": 2.1853,
      "step": 107844
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.8588955402374268,
      "learning_rate": 0.0003300417834756558,
      "loss": 2.2658,
      "step": 107845
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.8568205833435059,
      "learning_rate": 0.0003300377134721916,
      "loss": 2.4781,
      "step": 107846
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.8683196306228638,
      "learning_rate": 0.00033003364346314266,
      "loss": 2.0665,
      "step": 107847
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.8716052174568176,
      "learning_rate": 0.00033002957344851,
      "loss": 2.3879,
      "step": 107848
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.9279086589813232,
      "learning_rate": 0.00033002550342829424,
      "loss": 2.1623,
      "step": 107849
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.9482243061065674,
      "learning_rate": 0.0003300214334024962,
      "loss": 2.2213,
      "step": 107850
    },
    {
      "epoch": 1.4,
      "grad_norm": 1.0957709550857544,
      "learning_rate": 0.00033001736337111656,
      "loss": 2.1801,
      "step": 107851
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.9243539571762085,
      "learning_rate": 0.0003300132933341561,
      "loss": 2.1246,
      "step": 107852
    },
    {
      "epoch": 1.4,
      "grad_norm": 1.0641428232192993,
      "learning_rate": 0.0003300092232916157,
      "loss": 1.9236,
      "step": 107853
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.8629197478294373,
      "learning_rate": 0.000330005153243496,
      "loss": 2.1613,
      "step": 107854
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.9125906825065613,
      "learning_rate": 0.00033000108318979773,
      "loss": 2.2823,
      "step": 107855
    },
    {
      "epoch": 1.4,
      "grad_norm": 1.0080314874649048,
      "learning_rate": 0.00032999701313052163,
      "loss": 2.1325,
      "step": 107856
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.9128079414367676,
      "learning_rate": 0.0003299929430656687,
      "loss": 2.2817,
      "step": 107857
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.8334223031997681,
      "learning_rate": 0.00032998887299523944,
      "loss": 2.0467,
      "step": 107858
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.8918508887290955,
      "learning_rate": 0.0003299848029192346,
      "loss": 2.2469,
      "step": 107859
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.9313272833824158,
      "learning_rate": 0.0003299807328376551,
      "loss": 2.2334,
      "step": 107860
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.8944722414016724,
      "learning_rate": 0.00032997666275050156,
      "loss": 2.2915,
      "step": 107861
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.963279664516449,
      "learning_rate": 0.00032997259265777484,
      "loss": 2.0313,
      "step": 107862
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.9385107755661011,
      "learning_rate": 0.0003299685225594756,
      "loss": 2.1083,
      "step": 107863
    },
    {
      "epoch": 1.4,
      "grad_norm": 1.0076795816421509,
      "learning_rate": 0.00032996445245560475,
      "loss": 1.9773,
      "step": 107864
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.9249826669692993,
      "learning_rate": 0.0003299603823461629,
      "loss": 2.1543,
      "step": 107865
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.8944032788276672,
      "learning_rate": 0.0003299563122311508,
      "loss": 2.395,
      "step": 107866
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.8474336266517639,
      "learning_rate": 0.0003299522421105693,
      "loss": 1.9077,
      "step": 107867
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.900562047958374,
      "learning_rate": 0.00032994817198441905,
      "loss": 2.3194,
      "step": 107868
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.8727556467056274,
      "learning_rate": 0.0003299441018527009,
      "loss": 1.8446,
      "step": 107869
    },
    {
      "epoch": 1.4,
      "grad_norm": 1.4262739419937134,
      "learning_rate": 0.00032994003171541564,
      "loss": 2.2146,
      "step": 107870
    },
    {
      "epoch": 1.4,
      "grad_norm": 1.041701078414917,
      "learning_rate": 0.0003299359615725639,
      "loss": 1.9651,
      "step": 107871
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.9132078289985657,
      "learning_rate": 0.0003299318914241465,
      "loss": 2.1489,
      "step": 107872
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.8647263646125793,
      "learning_rate": 0.0003299278212701643,
      "loss": 2.1264,
      "step": 107873
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.9815427660942078,
      "learning_rate": 0.0003299237511106178,
      "loss": 2.2185,
      "step": 107874
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.9751908779144287,
      "learning_rate": 0.0003299196809455079,
      "loss": 2.1911,
      "step": 107875
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.9406566023826599,
      "learning_rate": 0.0003299156107748355,
      "loss": 2.04,
      "step": 107876
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.9136301279067993,
      "learning_rate": 0.00032991154059860116,
      "loss": 1.9934,
      "step": 107877
    },
    {
      "epoch": 1.4,
      "grad_norm": 1.0042825937271118,
      "learning_rate": 0.0003299074704168057,
      "loss": 2.1429,
      "step": 107878
    },
    {
      "epoch": 1.4,
      "grad_norm": 1.0165255069732666,
      "learning_rate": 0.0003299034002294499,
      "loss": 2.4032,
      "step": 107879
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.944424033164978,
      "learning_rate": 0.00032989933003653443,
      "loss": 2.0607,
      "step": 107880
    },
    {
      "epoch": 1.4,
      "grad_norm": 1.0278429985046387,
      "learning_rate": 0.00032989525983806016,
      "loss": 2.3177,
      "step": 107881
    },
    {
      "epoch": 1.4,
      "grad_norm": 1.006877064704895,
      "learning_rate": 0.00032989118963402784,
      "loss": 2.1882,
      "step": 107882
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.8741374015808105,
      "learning_rate": 0.0003298871194244381,
      "loss": 2.2365,
      "step": 107883
    },
    {
      "epoch": 1.4,
      "grad_norm": 1.2940274477005005,
      "learning_rate": 0.0003298830492092918,
      "loss": 1.9794,
      "step": 107884
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.9187095165252686,
      "learning_rate": 0.0003298789789885897,
      "loss": 2.3355,
      "step": 107885
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.8581894636154175,
      "learning_rate": 0.00032987490876233255,
      "loss": 2.0306,
      "step": 107886
    },
    {
      "epoch": 1.4,
      "grad_norm": 1.227731466293335,
      "learning_rate": 0.000329870838530521,
      "loss": 2.1981,
      "step": 107887
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.8144177794456482,
      "learning_rate": 0.00032986676829315604,
      "loss": 1.9904,
      "step": 107888
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.9411221742630005,
      "learning_rate": 0.00032986269805023816,
      "loss": 2.337,
      "step": 107889
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.8395439386367798,
      "learning_rate": 0.0003298586278017683,
      "loss": 2.0606,
      "step": 107890
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.8204277157783508,
      "learning_rate": 0.00032985455754774715,
      "loss": 2.0356,
      "step": 107891
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.8543090224266052,
      "learning_rate": 0.0003298504872881755,
      "loss": 2.096,
      "step": 107892
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.9092928767204285,
      "learning_rate": 0.000329846417023054,
      "loss": 2.2311,
      "step": 107893
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.8462936282157898,
      "learning_rate": 0.0003298423467523836,
      "loss": 2.1574,
      "step": 107894
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.8927861452102661,
      "learning_rate": 0.0003298382764761648,
      "loss": 2.2607,
      "step": 107895
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.8264256119728088,
      "learning_rate": 0.0003298342061943986,
      "loss": 2.2796,
      "step": 107896
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.8932408094406128,
      "learning_rate": 0.0003298301359070857,
      "loss": 2.2152,
      "step": 107897
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.9100937843322754,
      "learning_rate": 0.0003298260656142267,
      "loss": 2.1532,
      "step": 107898
    },
    {
      "epoch": 1.4,
      "grad_norm": 1.0265791416168213,
      "learning_rate": 0.00032982199531582254,
      "loss": 2.025,
      "step": 107899
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.8852290511131287,
      "learning_rate": 0.0003298179250118739,
      "loss": 2.2368,
      "step": 107900
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.9386380910873413,
      "learning_rate": 0.0003298138547023815,
      "loss": 2.1141,
      "step": 107901
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.9725342392921448,
      "learning_rate": 0.0003298097843873462,
      "loss": 2.0786,
      "step": 107902
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.9755359292030334,
      "learning_rate": 0.0003298057140667687,
      "loss": 2.1425,
      "step": 107903
    },
    {
      "epoch": 1.4,
      "grad_norm": 1.1458951234817505,
      "learning_rate": 0.00032980164374064977,
      "loss": 2.2207,
      "step": 107904
    },
    {
      "epoch": 1.4,
      "grad_norm": 0.8831833600997925,
      "learning_rate": 0.0003297975734089901,
      "loss": 1.8207,
      "step": 107905
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.9527754187583923,
      "learning_rate": 0.00032979350307179054,
      "loss": 2.2952,
      "step": 107906
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.9272066354751587,
      "learning_rate": 0.00032978943272905174,
      "loss": 2.0446,
      "step": 107907
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.9491319060325623,
      "learning_rate": 0.00032978536238077456,
      "loss": 2.1197,
      "step": 107908
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.8395523428916931,
      "learning_rate": 0.00032978129202695976,
      "loss": 2.2951,
      "step": 107909
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.9346839785575867,
      "learning_rate": 0.000329777221667608,
      "loss": 2.2541,
      "step": 107910
    },
    {
      "epoch": 1.41,
      "grad_norm": 1.0972076654434204,
      "learning_rate": 0.0003297731513027201,
      "loss": 2.0626,
      "step": 107911
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.9238670468330383,
      "learning_rate": 0.00032976908093229683,
      "loss": 2.2502,
      "step": 107912
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.9476768374443054,
      "learning_rate": 0.00032976501055633894,
      "loss": 2.3517,
      "step": 107913
    },
    {
      "epoch": 1.41,
      "grad_norm": 1.2249401807785034,
      "learning_rate": 0.00032976094017484706,
      "loss": 2.0452,
      "step": 107914
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.9032697081565857,
      "learning_rate": 0.00032975686978782223,
      "loss": 2.0473,
      "step": 107915
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.9499280452728271,
      "learning_rate": 0.00032975279939526493,
      "loss": 2.1237,
      "step": 107916
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.8574235439300537,
      "learning_rate": 0.00032974872899717603,
      "loss": 2.0351,
      "step": 107917
    },
    {
      "epoch": 1.41,
      "grad_norm": 1.1094450950622559,
      "learning_rate": 0.00032974465859355635,
      "loss": 2.2295,
      "step": 107918
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.910781979560852,
      "learning_rate": 0.0003297405881844065,
      "loss": 2.1279,
      "step": 107919
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.8923547863960266,
      "learning_rate": 0.0003297365177697273,
      "loss": 2.0724,
      "step": 107920
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.8802916407585144,
      "learning_rate": 0.0003297324473495196,
      "loss": 1.917,
      "step": 107921
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.8457378149032593,
      "learning_rate": 0.000329728376923784,
      "loss": 2.1211,
      "step": 107922
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.8981829285621643,
      "learning_rate": 0.00032972430649252135,
      "loss": 2.3061,
      "step": 107923
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.9713062047958374,
      "learning_rate": 0.00032972023605573245,
      "loss": 2.2192,
      "step": 107924
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.8658709526062012,
      "learning_rate": 0.0003297161656134179,
      "loss": 2.0815,
      "step": 107925
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.9455400109291077,
      "learning_rate": 0.0003297120951655786,
      "loss": 2.0581,
      "step": 107926
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.8640819191932678,
      "learning_rate": 0.00032970802471221536,
      "loss": 2.2994,
      "step": 107927
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.9087071418762207,
      "learning_rate": 0.0003297039542533287,
      "loss": 2.0397,
      "step": 107928
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.9205935597419739,
      "learning_rate": 0.0003296998837889195,
      "loss": 1.9343,
      "step": 107929
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.8538536429405212,
      "learning_rate": 0.0003296958133189886,
      "loss": 1.9928,
      "step": 107930
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.8785665035247803,
      "learning_rate": 0.0003296917428435367,
      "loss": 2.1385,
      "step": 107931
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.891252875328064,
      "learning_rate": 0.0003296876723625645,
      "loss": 2.135,
      "step": 107932
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.9171789884567261,
      "learning_rate": 0.00032968360187607285,
      "loss": 2.0416,
      "step": 107933
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.9136303663253784,
      "learning_rate": 0.00032967953138406244,
      "loss": 2.1996,
      "step": 107934
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.8333099484443665,
      "learning_rate": 0.000329675460886534,
      "loss": 2.0009,
      "step": 107935
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.9051647782325745,
      "learning_rate": 0.00032967139038348836,
      "loss": 1.9878,
      "step": 107936
    },
    {
      "epoch": 1.41,
      "grad_norm": 1.0885595083236694,
      "learning_rate": 0.0003296673198749262,
      "loss": 2.1263,
      "step": 107937
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.8708935379981995,
      "learning_rate": 0.0003296632493608485,
      "loss": 2.1856,
      "step": 107938
    },
    {
      "epoch": 1.41,
      "grad_norm": 1.0628952980041504,
      "learning_rate": 0.0003296591788412557,
      "loss": 2.1962,
      "step": 107939
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.9215812087059021,
      "learning_rate": 0.00032965510831614873,
      "loss": 2.0672,
      "step": 107940
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.8797670602798462,
      "learning_rate": 0.00032965103778552827,
      "loss": 2.1055,
      "step": 107941
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.9189686179161072,
      "learning_rate": 0.00032964696724939517,
      "loss": 2.0776,
      "step": 107942
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.9107223749160767,
      "learning_rate": 0.0003296428967077501,
      "loss": 2.142,
      "step": 107943
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.9087758660316467,
      "learning_rate": 0.0003296388261605939,
      "loss": 2.0185,
      "step": 107944
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.9556981325149536,
      "learning_rate": 0.0003296347556079273,
      "loss": 2.0898,
      "step": 107945
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.8744390606880188,
      "learning_rate": 0.00032963068504975095,
      "loss": 2.0586,
      "step": 107946
    },
    {
      "epoch": 1.41,
      "grad_norm": 1.0517762899398804,
      "learning_rate": 0.0003296266144860658,
      "loss": 2.2883,
      "step": 107947
    },
    {
      "epoch": 1.41,
      "grad_norm": 1.0461217164993286,
      "learning_rate": 0.0003296225439168724,
      "loss": 2.0577,
      "step": 107948
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.9002725481987,
      "learning_rate": 0.0003296184733421718,
      "loss": 2.1717,
      "step": 107949
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.9451943635940552,
      "learning_rate": 0.0003296144027619643,
      "loss": 2.1884,
      "step": 107950
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.990729808807373,
      "learning_rate": 0.00032961033217625106,
      "loss": 2.4929,
      "step": 107951
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.9059406518936157,
      "learning_rate": 0.0003296062615850328,
      "loss": 2.0382,
      "step": 107952
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.8946505784988403,
      "learning_rate": 0.00032960219098831004,
      "loss": 2.2179,
      "step": 107953
    },
    {
      "epoch": 1.41,
      "grad_norm": 1.0686513185501099,
      "learning_rate": 0.00032959812038608365,
      "loss": 2.299,
      "step": 107954
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.9313505291938782,
      "learning_rate": 0.0003295940497783545,
      "loss": 2.0675,
      "step": 107955
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.8975502848625183,
      "learning_rate": 0.0003295899791651232,
      "loss": 2.2126,
      "step": 107956
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.9555890560150146,
      "learning_rate": 0.0003295859085463906,
      "loss": 2.2682,
      "step": 107957
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.9048227667808533,
      "learning_rate": 0.00032958183792215746,
      "loss": 1.9433,
      "step": 107958
    },
    {
      "epoch": 1.41,
      "grad_norm": 1.0416340827941895,
      "learning_rate": 0.00032957776729242443,
      "loss": 2.0845,
      "step": 107959
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.9297664761543274,
      "learning_rate": 0.0003295736966571923,
      "loss": 2.0745,
      "step": 107960
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.9628893733024597,
      "learning_rate": 0.00032956962601646196,
      "loss": 2.1424,
      "step": 107961
    },
    {
      "epoch": 1.41,
      "grad_norm": 1.0896440744400024,
      "learning_rate": 0.000329565555370234,
      "loss": 2.1206,
      "step": 107962
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.8677113056182861,
      "learning_rate": 0.00032956148471850933,
      "loss": 2.2099,
      "step": 107963
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.8450146913528442,
      "learning_rate": 0.00032955741406128853,
      "loss": 2.0973,
      "step": 107964
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.8294072151184082,
      "learning_rate": 0.0003295533433985725,
      "loss": 2.0208,
      "step": 107965
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.9015811085700989,
      "learning_rate": 0.00032954927273036195,
      "loss": 2.191,
      "step": 107966
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.9551223516464233,
      "learning_rate": 0.00032954520205665764,
      "loss": 2.2112,
      "step": 107967
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.8780817985534668,
      "learning_rate": 0.0003295411313774602,
      "loss": 2.2661,
      "step": 107968
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.8350005745887756,
      "learning_rate": 0.00032953706069277066,
      "loss": 2.1115,
      "step": 107969
    },
    {
      "epoch": 1.41,
      "grad_norm": 1.0688729286193848,
      "learning_rate": 0.00032953299000258957,
      "loss": 2.0975,
      "step": 107970
    },
    {
      "epoch": 1.41,
      "grad_norm": 1.0805734395980835,
      "learning_rate": 0.0003295289193069177,
      "loss": 2.1512,
      "step": 107971
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.884839653968811,
      "learning_rate": 0.00032952484860575585,
      "loss": 2.1367,
      "step": 107972
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.902550995349884,
      "learning_rate": 0.00032952077789910486,
      "loss": 2.0831,
      "step": 107973
    },
    {
      "epoch": 1.41,
      "grad_norm": 1.1224654912948608,
      "learning_rate": 0.00032951670718696526,
      "loss": 2.2269,
      "step": 107974
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.9050338268280029,
      "learning_rate": 0.000329512636469338,
      "loss": 2.1909,
      "step": 107975
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.8625040054321289,
      "learning_rate": 0.0003295085657462239,
      "loss": 2.4157,
      "step": 107976
    },
    {
      "epoch": 1.41,
      "grad_norm": 1.0454919338226318,
      "learning_rate": 0.0003295044950176235,
      "loss": 2.0012,
      "step": 107977
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.896805465221405,
      "learning_rate": 0.0003295004242835376,
      "loss": 2.0378,
      "step": 107978
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.9757972359657288,
      "learning_rate": 0.00032949635354396714,
      "loss": 2.3939,
      "step": 107979
    },
    {
      "epoch": 1.41,
      "grad_norm": 1.0049471855163574,
      "learning_rate": 0.0003294922827989127,
      "loss": 2.1092,
      "step": 107980
    },
    {
      "epoch": 1.41,
      "grad_norm": 1.2685335874557495,
      "learning_rate": 0.000329488212048375,
      "loss": 2.1859,
      "step": 107981
    },
    {
      "epoch": 1.41,
      "grad_norm": 1.056326150894165,
      "learning_rate": 0.00032948414129235505,
      "loss": 2.2386,
      "step": 107982
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.9239629507064819,
      "learning_rate": 0.00032948007053085334,
      "loss": 2.2225,
      "step": 107983
    },
    {
      "epoch": 1.41,
      "grad_norm": 1.1035289764404297,
      "learning_rate": 0.00032947599976387073,
      "loss": 2.2514,
      "step": 107984
    },
    {
      "epoch": 1.41,
      "grad_norm": 1.1695719957351685,
      "learning_rate": 0.000329471928991408,
      "loss": 2.0929,
      "step": 107985
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.9657862782478333,
      "learning_rate": 0.00032946785821346585,
      "loss": 2.2347,
      "step": 107986
    },
    {
      "epoch": 1.41,
      "grad_norm": 1.0295714139938354,
      "learning_rate": 0.0003294637874300451,
      "loss": 2.219,
      "step": 107987
    },
    {
      "epoch": 1.41,
      "grad_norm": 1.0615094900131226,
      "learning_rate": 0.0003294597166411465,
      "loss": 1.9826,
      "step": 107988
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.8816750645637512,
      "learning_rate": 0.00032945564584677074,
      "loss": 2.0132,
      "step": 107989
    },
    {
      "epoch": 1.41,
      "grad_norm": 1.0109835863113403,
      "learning_rate": 0.00032945157504691864,
      "loss": 2.1217,
      "step": 107990
    },
    {
      "epoch": 1.41,
      "grad_norm": 1.0711681842803955,
      "learning_rate": 0.00032944750424159094,
      "loss": 2.0853,
      "step": 107991
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.9820513725280762,
      "learning_rate": 0.00032944343343078834,
      "loss": 2.2138,
      "step": 107992
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.9610797762870789,
      "learning_rate": 0.00032943936261451166,
      "loss": 2.3323,
      "step": 107993
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.8986378908157349,
      "learning_rate": 0.0003294352917927617,
      "loss": 2.2233,
      "step": 107994
    },
    {
      "epoch": 1.41,
      "grad_norm": 1.0172255039215088,
      "learning_rate": 0.00032943122096553914,
      "loss": 2.2884,
      "step": 107995
    },
    {
      "epoch": 1.41,
      "grad_norm": 1.0378726720809937,
      "learning_rate": 0.0003294271501328447,
      "loss": 2.197,
      "step": 107996
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.8792535066604614,
      "learning_rate": 0.0003294230792946793,
      "loss": 2.2222,
      "step": 107997
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.925626277923584,
      "learning_rate": 0.0003294190084510436,
      "loss": 2.1426,
      "step": 107998
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.9820306301116943,
      "learning_rate": 0.00032941493760193824,
      "loss": 2.464,
      "step": 107999
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.8808335065841675,
      "learning_rate": 0.0003294108667473641,
      "loss": 2.4775,
      "step": 108000
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.910832405090332,
      "learning_rate": 0.000329406795887322,
      "loss": 1.9474,
      "step": 108001
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.9007843136787415,
      "learning_rate": 0.0003294027250218126,
      "loss": 2.2948,
      "step": 108002
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.9300707578659058,
      "learning_rate": 0.0003293986541508367,
      "loss": 2.1831,
      "step": 108003
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.8677937984466553,
      "learning_rate": 0.000329394583274395,
      "loss": 2.0735,
      "step": 108004
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.8578896522521973,
      "learning_rate": 0.0003293905123924883,
      "loss": 2.161,
      "step": 108005
    },
    {
      "epoch": 1.41,
      "grad_norm": 1.2227078676223755,
      "learning_rate": 0.0003293864415051174,
      "loss": 2.0504,
      "step": 108006
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.8766089081764221,
      "learning_rate": 0.00032938237061228297,
      "loss": 2.1924,
      "step": 108007
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.9325107336044312,
      "learning_rate": 0.00032937829971398574,
      "loss": 2.0937,
      "step": 108008
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.8672115802764893,
      "learning_rate": 0.0003293742288102265,
      "loss": 2.3633,
      "step": 108009
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.9583311676979065,
      "learning_rate": 0.0003293701579010062,
      "loss": 2.3284,
      "step": 108010
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.9038520455360413,
      "learning_rate": 0.0003293660869863253,
      "loss": 2.1198,
      "step": 108011
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.8395993709564209,
      "learning_rate": 0.0003293620160661847,
      "loss": 1.8736,
      "step": 108012
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.983116865158081,
      "learning_rate": 0.00032935794514058533,
      "loss": 2.0716,
      "step": 108013
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.8973697423934937,
      "learning_rate": 0.0003293538742095275,
      "loss": 2.3334,
      "step": 108014
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.885184645652771,
      "learning_rate": 0.0003293498032730124,
      "loss": 2.385,
      "step": 108015
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.8729262351989746,
      "learning_rate": 0.0003293457323310406,
      "loss": 2.2665,
      "step": 108016
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.8756343722343445,
      "learning_rate": 0.00032934166138361284,
      "loss": 2.3346,
      "step": 108017
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.9133453965187073,
      "learning_rate": 0.00032933759043072983,
      "loss": 2.0594,
      "step": 108018
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.8520922660827637,
      "learning_rate": 0.0003293335194723926,
      "loss": 2.1157,
      "step": 108019
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.9301323890686035,
      "learning_rate": 0.0003293294485086015,
      "loss": 2.1427,
      "step": 108020
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.9303227066993713,
      "learning_rate": 0.00032932537753935763,
      "loss": 2.0217,
      "step": 108021
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.9481642246246338,
      "learning_rate": 0.0003293213065646616,
      "loss": 2.2418,
      "step": 108022
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.918793797492981,
      "learning_rate": 0.0003293172355845142,
      "loss": 1.7455,
      "step": 108023
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.9720147252082825,
      "learning_rate": 0.0003293131645989161,
      "loss": 2.0732,
      "step": 108024
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.8900590538978577,
      "learning_rate": 0.0003293090936078682,
      "loss": 2.2299,
      "step": 108025
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.8813393115997314,
      "learning_rate": 0.0003293050226113712,
      "loss": 2.135,
      "step": 108026
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.8439174294471741,
      "learning_rate": 0.0003293009516094258,
      "loss": 2.0853,
      "step": 108027
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.8826686143875122,
      "learning_rate": 0.00032929688060203283,
      "loss": 2.2214,
      "step": 108028
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.8641334176063538,
      "learning_rate": 0.00032929280958919296,
      "loss": 2.0457,
      "step": 108029
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.9171136021614075,
      "learning_rate": 0.0003292887385709071,
      "loss": 2.4041,
      "step": 108030
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.8685181736946106,
      "learning_rate": 0.0003292846675471758,
      "loss": 1.8785,
      "step": 108031
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.9289743304252625,
      "learning_rate": 0.0003292805965179999,
      "loss": 1.9278,
      "step": 108032
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.9988596439361572,
      "learning_rate": 0.0003292765254833803,
      "loss": 2.41,
      "step": 108033
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.9443636536598206,
      "learning_rate": 0.0003292724544433177,
      "loss": 2.146,
      "step": 108034
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.897102952003479,
      "learning_rate": 0.00032926838339781265,
      "loss": 2.2825,
      "step": 108035
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.9074530005455017,
      "learning_rate": 0.0003292643123468661,
      "loss": 1.8968,
      "step": 108036
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.9914476871490479,
      "learning_rate": 0.0003292602412904788,
      "loss": 2.0736,
      "step": 108037
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.9764223098754883,
      "learning_rate": 0.0003292561702286514,
      "loss": 2.0146,
      "step": 108038
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.9524499773979187,
      "learning_rate": 0.0003292520991613848,
      "loss": 2.1036,
      "step": 108039
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.9265197515487671,
      "learning_rate": 0.00032924802808867963,
      "loss": 2.2071,
      "step": 108040
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.9161713123321533,
      "learning_rate": 0.00032924395701053674,
      "loss": 2.0624,
      "step": 108041
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.8837074041366577,
      "learning_rate": 0.0003292398859269568,
      "loss": 2.1537,
      "step": 108042
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.9643598198890686,
      "learning_rate": 0.0003292358148379407,
      "loss": 2.2052,
      "step": 108043
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.8797056078910828,
      "learning_rate": 0.00032923174374348903,
      "loss": 2.116,
      "step": 108044
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.9575600624084473,
      "learning_rate": 0.0003292276726436026,
      "loss": 1.973,
      "step": 108045
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.9221838116645813,
      "learning_rate": 0.0003292236015382823,
      "loss": 2.272,
      "step": 108046
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.8759576678276062,
      "learning_rate": 0.00032921953042752873,
      "loss": 2.1876,
      "step": 108047
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.829003632068634,
      "learning_rate": 0.0003292154593113426,
      "loss": 2.2735,
      "step": 108048
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.8905554413795471,
      "learning_rate": 0.00032921138818972493,
      "loss": 2.0623,
      "step": 108049
    },
    {
      "epoch": 1.41,
      "grad_norm": 1.0786489248275757,
      "learning_rate": 0.00032920731706267617,
      "loss": 2.2198,
      "step": 108050
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.9959378838539124,
      "learning_rate": 0.0003292032459301973,
      "loss": 2.1058,
      "step": 108051
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.9013168215751648,
      "learning_rate": 0.000329199174792289,
      "loss": 2.288,
      "step": 108052
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.8279496431350708,
      "learning_rate": 0.000329195103648952,
      "loss": 1.9926,
      "step": 108053
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.8102792501449585,
      "learning_rate": 0.00032919103250018706,
      "loss": 2.0557,
      "step": 108054
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.9104573130607605,
      "learning_rate": 0.000329186961345995,
      "loss": 2.2391,
      "step": 108055
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.9282839894294739,
      "learning_rate": 0.0003291828901863765,
      "loss": 2.1119,
      "step": 108056
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.8321645259857178,
      "learning_rate": 0.0003291788190213324,
      "loss": 2.1946,
      "step": 108057
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.8538901209831238,
      "learning_rate": 0.0003291747478508633,
      "loss": 2.4092,
      "step": 108058
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.8705660104751587,
      "learning_rate": 0.0003291706766749701,
      "loss": 2.2577,
      "step": 108059
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.9523157477378845,
      "learning_rate": 0.00032916660549365363,
      "loss": 2.2065,
      "step": 108060
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.8037582039833069,
      "learning_rate": 0.0003291625343069145,
      "loss": 2.3624,
      "step": 108061
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.9233992695808411,
      "learning_rate": 0.00032915846311475344,
      "loss": 2.0855,
      "step": 108062
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.8533992767333984,
      "learning_rate": 0.00032915439191717126,
      "loss": 2.1871,
      "step": 108063
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.8009389638900757,
      "learning_rate": 0.0003291503207141687,
      "loss": 2.1843,
      "step": 108064
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.8678573369979858,
      "learning_rate": 0.0003291462495057467,
      "loss": 2.1086,
      "step": 108065
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.8565648198127747,
      "learning_rate": 0.0003291421782919057,
      "loss": 2.2238,
      "step": 108066
    },
    {
      "epoch": 1.41,
      "grad_norm": 1.16985285282135,
      "learning_rate": 0.0003291381070726467,
      "loss": 1.9637,
      "step": 108067
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.8039721846580505,
      "learning_rate": 0.0003291340358479704,
      "loss": 2.0818,
      "step": 108068
    },
    {
      "epoch": 1.41,
      "grad_norm": 1.0940403938293457,
      "learning_rate": 0.00032912996461787744,
      "loss": 2.0917,
      "step": 108069
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.8159663081169128,
      "learning_rate": 0.0003291258933823687,
      "loss": 2.2935,
      "step": 108070
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.9434748888015747,
      "learning_rate": 0.000329121822141445,
      "loss": 2.3202,
      "step": 108071
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.9532023072242737,
      "learning_rate": 0.0003291177508951069,
      "loss": 2.172,
      "step": 108072
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.8503543734550476,
      "learning_rate": 0.0003291136796433553,
      "loss": 2.0335,
      "step": 108073
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.9025867581367493,
      "learning_rate": 0.00032910960838619094,
      "loss": 2.164,
      "step": 108074
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.9714050889015198,
      "learning_rate": 0.0003291055371236145,
      "loss": 1.9225,
      "step": 108075
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.9568392634391785,
      "learning_rate": 0.0003291014658556268,
      "loss": 2.2085,
      "step": 108076
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.9783837795257568,
      "learning_rate": 0.00032909739458222864,
      "loss": 2.246,
      "step": 108077
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.8433436751365662,
      "learning_rate": 0.0003290933233034207,
      "loss": 2.1412,
      "step": 108078
    },
    {
      "epoch": 1.41,
      "grad_norm": 1.0379467010498047,
      "learning_rate": 0.00032908925201920376,
      "loss": 2.2115,
      "step": 108079
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.9101744294166565,
      "learning_rate": 0.00032908518072957856,
      "loss": 2.1312,
      "step": 108080
    },
    {
      "epoch": 1.41,
      "grad_norm": 1.0193369388580322,
      "learning_rate": 0.00032908110943454585,
      "loss": 2.2374,
      "step": 108081
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.9250664710998535,
      "learning_rate": 0.0003290770381341065,
      "loss": 2.074,
      "step": 108082
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.9203833341598511,
      "learning_rate": 0.00032907296682826114,
      "loss": 2.0555,
      "step": 108083
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.9775055050849915,
      "learning_rate": 0.00032906889551701055,
      "loss": 2.3679,
      "step": 108084
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.8904313445091248,
      "learning_rate": 0.0003290648242003555,
      "loss": 2.0008,
      "step": 108085
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.9153993129730225,
      "learning_rate": 0.0003290607528782968,
      "loss": 2.2018,
      "step": 108086
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.9545137286186218,
      "learning_rate": 0.0003290566815508351,
      "loss": 1.9448,
      "step": 108087
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.953975260257721,
      "learning_rate": 0.0003290526102179712,
      "loss": 2.1184,
      "step": 108088
    },
    {
      "epoch": 1.41,
      "grad_norm": 1.1505545377731323,
      "learning_rate": 0.00032904853887970597,
      "loss": 2.2792,
      "step": 108089
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.9347606897354126,
      "learning_rate": 0.00032904446753603996,
      "loss": 2.1723,
      "step": 108090
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.8911188840866089,
      "learning_rate": 0.0003290403961869741,
      "loss": 2.1909,
      "step": 108091
    },
    {
      "epoch": 1.41,
      "grad_norm": 1.045880675315857,
      "learning_rate": 0.00032903632483250903,
      "loss": 2.3227,
      "step": 108092
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.9165623188018799,
      "learning_rate": 0.0003290322534726456,
      "loss": 2.1763,
      "step": 108093
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.8764321208000183,
      "learning_rate": 0.00032902818210738455,
      "loss": 1.9805,
      "step": 108094
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.8149137496948242,
      "learning_rate": 0.00032902411073672656,
      "loss": 2.0663,
      "step": 108095
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.8918251991271973,
      "learning_rate": 0.00032902003936067245,
      "loss": 1.9475,
      "step": 108096
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.9233492612838745,
      "learning_rate": 0.000329015967979223,
      "loss": 2.2887,
      "step": 108097
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.8606706857681274,
      "learning_rate": 0.00032901189659237894,
      "loss": 2.1181,
      "step": 108098
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.8498026132583618,
      "learning_rate": 0.0003290078252001409,
      "loss": 2.0675,
      "step": 108099
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.9546363949775696,
      "learning_rate": 0.00032900375380250985,
      "loss": 1.8464,
      "step": 108100
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.928749144077301,
      "learning_rate": 0.0003289996823994865,
      "loss": 2.1986,
      "step": 108101
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.875494658946991,
      "learning_rate": 0.0003289956109910715,
      "loss": 2.2502,
      "step": 108102
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.8626434206962585,
      "learning_rate": 0.0003289915395772657,
      "loss": 2.1597,
      "step": 108103
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.9097857475280762,
      "learning_rate": 0.0003289874681580698,
      "loss": 2.1493,
      "step": 108104
    },
    {
      "epoch": 1.41,
      "grad_norm": 1.418054461479187,
      "learning_rate": 0.0003289833967334846,
      "loss": 2.4067,
      "step": 108105
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.9803465604782104,
      "learning_rate": 0.0003289793253035108,
      "loss": 2.1442,
      "step": 108106
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.9038952589035034,
      "learning_rate": 0.0003289752538681492,
      "loss": 2.2834,
      "step": 108107
    },
    {
      "epoch": 1.41,
      "grad_norm": 1.0639865398406982,
      "learning_rate": 0.0003289711824274006,
      "loss": 2.0375,
      "step": 108108
    },
    {
      "epoch": 1.41,
      "grad_norm": 1.1318284273147583,
      "learning_rate": 0.00032896711098126573,
      "loss": 2.0775,
      "step": 108109
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.8147632479667664,
      "learning_rate": 0.0003289630395297452,
      "loss": 2.129,
      "step": 108110
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.9094247817993164,
      "learning_rate": 0.00032895896807284006,
      "loss": 2.3093,
      "step": 108111
    },
    {
      "epoch": 1.41,
      "grad_norm": 1.040399193763733,
      "learning_rate": 0.0003289548966105508,
      "loss": 2.2292,
      "step": 108112
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.9385697841644287,
      "learning_rate": 0.0003289508251428783,
      "loss": 1.989,
      "step": 108113
    },
    {
      "epoch": 1.41,
      "grad_norm": 1.093057632446289,
      "learning_rate": 0.00032894675366982326,
      "loss": 2.33,
      "step": 108114
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.9262003898620605,
      "learning_rate": 0.0003289426821913866,
      "loss": 2.1316,
      "step": 108115
    },
    {
      "epoch": 1.41,
      "grad_norm": 1.0655404329299927,
      "learning_rate": 0.0003289386107075688,
      "loss": 1.9617,
      "step": 108116
    },
    {
      "epoch": 1.41,
      "grad_norm": 1.0774695873260498,
      "learning_rate": 0.0003289345392183708,
      "loss": 2.2472,
      "step": 108117
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.9816790223121643,
      "learning_rate": 0.0003289304677237934,
      "loss": 2.0295,
      "step": 108118
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.8854055404663086,
      "learning_rate": 0.0003289263962238372,
      "loss": 2.0405,
      "step": 108119
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.8867905139923096,
      "learning_rate": 0.000328922324718503,
      "loss": 2.0307,
      "step": 108120
    },
    {
      "epoch": 1.41,
      "grad_norm": 1.2231309413909912,
      "learning_rate": 0.0003289182532077917,
      "loss": 2.2597,
      "step": 108121
    },
    {
      "epoch": 1.41,
      "grad_norm": 1.1512967348098755,
      "learning_rate": 0.000328914181691704,
      "loss": 2.0656,
      "step": 108122
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.8967441916465759,
      "learning_rate": 0.00032891011017024046,
      "loss": 2.148,
      "step": 108123
    },
    {
      "epoch": 1.41,
      "grad_norm": 1.360459804534912,
      "learning_rate": 0.00032890603864340197,
      "loss": 2.0766,
      "step": 108124
    },
    {
      "epoch": 1.41,
      "grad_norm": 1.226046085357666,
      "learning_rate": 0.00032890196711118946,
      "loss": 2.2692,
      "step": 108125
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.9621158838272095,
      "learning_rate": 0.0003288978955736034,
      "loss": 1.9758,
      "step": 108126
    },
    {
      "epoch": 1.41,
      "grad_norm": 1.0958786010742188,
      "learning_rate": 0.00032889382403064474,
      "loss": 2.1056,
      "step": 108127
    },
    {
      "epoch": 1.41,
      "grad_norm": 1.0751121044158936,
      "learning_rate": 0.00032888975248231416,
      "loss": 2.1132,
      "step": 108128
    },
    {
      "epoch": 1.41,
      "grad_norm": 1.3034089803695679,
      "learning_rate": 0.0003288856809286124,
      "loss": 2.0122,
      "step": 108129
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.8536435961723328,
      "learning_rate": 0.00032888160936954023,
      "loss": 2.1195,
      "step": 108130
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.9560658931732178,
      "learning_rate": 0.0003288775378050984,
      "loss": 2.1377,
      "step": 108131
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.9110459089279175,
      "learning_rate": 0.0003288734662352879,
      "loss": 2.4648,
      "step": 108132
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.9879226088523865,
      "learning_rate": 0.00032886939466010903,
      "loss": 2.019,
      "step": 108133
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.9242355823516846,
      "learning_rate": 0.00032886532307956285,
      "loss": 2.122,
      "step": 108134
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.9679683446884155,
      "learning_rate": 0.0003288612514936502,
      "loss": 2.0761,
      "step": 108135
    },
    {
      "epoch": 1.41,
      "grad_norm": 1.133741855621338,
      "learning_rate": 0.0003288571799023716,
      "loss": 2.2269,
      "step": 108136
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.9248847365379333,
      "learning_rate": 0.0003288531083057279,
      "loss": 2.3357,
      "step": 108137
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.9192577600479126,
      "learning_rate": 0.00032884903670371985,
      "loss": 2.1224,
      "step": 108138
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.9515834450721741,
      "learning_rate": 0.0003288449650963483,
      "loss": 2.1153,
      "step": 108139
    },
    {
      "epoch": 1.41,
      "grad_norm": 1.4680087566375732,
      "learning_rate": 0.00032884089348361376,
      "loss": 2.0549,
      "step": 108140
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.8939274549484253,
      "learning_rate": 0.00032883682186551737,
      "loss": 2.2282,
      "step": 108141
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.8935341238975525,
      "learning_rate": 0.0003288327502420595,
      "loss": 2.1869,
      "step": 108142
    },
    {
      "epoch": 1.41,
      "grad_norm": 1.4574437141418457,
      "learning_rate": 0.0003288286786132411,
      "loss": 1.9581,
      "step": 108143
    },
    {
      "epoch": 1.41,
      "grad_norm": 1.0196338891983032,
      "learning_rate": 0.000328824606979063,
      "loss": 2.1866,
      "step": 108144
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.8786540627479553,
      "learning_rate": 0.00032882053533952576,
      "loss": 2.1521,
      "step": 108145
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.8835749626159668,
      "learning_rate": 0.0003288164636946303,
      "loss": 2.2999,
      "step": 108146
    },
    {
      "epoch": 1.41,
      "grad_norm": 1.0222105979919434,
      "learning_rate": 0.0003288123920443773,
      "loss": 2.2178,
      "step": 108147
    },
    {
      "epoch": 1.41,
      "grad_norm": 1.1471967697143555,
      "learning_rate": 0.0003288083203887676,
      "loss": 2.263,
      "step": 108148
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.9559027552604675,
      "learning_rate": 0.00032880424872780173,
      "loss": 2.2149,
      "step": 108149
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.927841305732727,
      "learning_rate": 0.00032880017706148073,
      "loss": 2.1266,
      "step": 108150
    },
    {
      "epoch": 1.41,
      "grad_norm": 1.0048412084579468,
      "learning_rate": 0.0003287961053898052,
      "loss": 2.0294,
      "step": 108151
    },
    {
      "epoch": 1.41,
      "grad_norm": 1.0174295902252197,
      "learning_rate": 0.0003287920337127759,
      "loss": 2.1169,
      "step": 108152
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.9229374527931213,
      "learning_rate": 0.00032878796203039375,
      "loss": 2.2572,
      "step": 108153
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.9457613229751587,
      "learning_rate": 0.0003287838903426592,
      "loss": 2.0445,
      "step": 108154
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.9026344418525696,
      "learning_rate": 0.0003287798186495733,
      "loss": 2.1295,
      "step": 108155
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.8876156806945801,
      "learning_rate": 0.00032877574695113664,
      "loss": 2.2398,
      "step": 108156
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.9290544986724854,
      "learning_rate": 0.00032877167524734995,
      "loss": 1.9202,
      "step": 108157
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.8592559695243835,
      "learning_rate": 0.0003287676035382142,
      "loss": 2.0661,
      "step": 108158
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.9380192160606384,
      "learning_rate": 0.00032876353182372995,
      "loss": 2.2368,
      "step": 108159
    },
    {
      "epoch": 1.41,
      "grad_norm": 1.1497083902359009,
      "learning_rate": 0.000328759460103898,
      "loss": 2.161,
      "step": 108160
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.9505919218063354,
      "learning_rate": 0.00032875538837871907,
      "loss": 2.1802,
      "step": 108161
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.9949914216995239,
      "learning_rate": 0.00032875131664819417,
      "loss": 2.1662,
      "step": 108162
    },
    {
      "epoch": 1.41,
      "grad_norm": 1.1112815141677856,
      "learning_rate": 0.00032874724491232366,
      "loss": 2.3397,
      "step": 108163
    },
    {
      "epoch": 1.41,
      "grad_norm": 1.2282501459121704,
      "learning_rate": 0.0003287431731711085,
      "loss": 2.0067,
      "step": 108164
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.9982602000236511,
      "learning_rate": 0.0003287391014245496,
      "loss": 2.451,
      "step": 108165
    },
    {
      "epoch": 1.41,
      "grad_norm": 1.0583879947662354,
      "learning_rate": 0.00032873502967264745,
      "loss": 2.0916,
      "step": 108166
    },
    {
      "epoch": 1.41,
      "grad_norm": 1.0597493648529053,
      "learning_rate": 0.0003287309579154029,
      "loss": 2.1831,
      "step": 108167
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.8723554611206055,
      "learning_rate": 0.00032872688615281676,
      "loss": 2.2488,
      "step": 108168
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.9448915719985962,
      "learning_rate": 0.00032872281438488975,
      "loss": 2.2257,
      "step": 108169
    },
    {
      "epoch": 1.41,
      "grad_norm": 1.0435597896575928,
      "learning_rate": 0.00032871874261162256,
      "loss": 2.2551,
      "step": 108170
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.9931875467300415,
      "learning_rate": 0.00032871467083301617,
      "loss": 2.1408,
      "step": 108171
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.8648949861526489,
      "learning_rate": 0.000328710599049071,
      "loss": 2.1104,
      "step": 108172
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.9975138902664185,
      "learning_rate": 0.0003287065272597881,
      "loss": 2.3755,
      "step": 108173
    },
    {
      "epoch": 1.41,
      "grad_norm": 1.121901035308838,
      "learning_rate": 0.00032870245546516813,
      "loss": 2.2615,
      "step": 108174
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.8253113031387329,
      "learning_rate": 0.00032869838366521173,
      "loss": 2.0225,
      "step": 108175
    },
    {
      "epoch": 1.41,
      "grad_norm": 1.0036160945892334,
      "learning_rate": 0.0003286943118599198,
      "loss": 2.3801,
      "step": 108176
    },
    {
      "epoch": 1.41,
      "grad_norm": 1.0888545513153076,
      "learning_rate": 0.00032869024004929313,
      "loss": 2.1047,
      "step": 108177
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.9875221848487854,
      "learning_rate": 0.00032868616823333234,
      "loss": 2.0729,
      "step": 108178
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.9399693012237549,
      "learning_rate": 0.00032868209641203825,
      "loss": 2.1029,
      "step": 108179
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.9152134656906128,
      "learning_rate": 0.0003286780245854116,
      "loss": 2.125,
      "step": 108180
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.8637371063232422,
      "learning_rate": 0.00032867395275345315,
      "loss": 2.1265,
      "step": 108181
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.9024850130081177,
      "learning_rate": 0.0003286698809161637,
      "loss": 2.0629,
      "step": 108182
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.8919167518615723,
      "learning_rate": 0.00032866580907354407,
      "loss": 2.0802,
      "step": 108183
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.9702630639076233,
      "learning_rate": 0.0003286617372255948,
      "loss": 2.2684,
      "step": 108184
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.9531562924385071,
      "learning_rate": 0.00032865766537231675,
      "loss": 2.2998,
      "step": 108185
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.9975880980491638,
      "learning_rate": 0.0003286535935137108,
      "loss": 2.295,
      "step": 108186
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.9158644080162048,
      "learning_rate": 0.0003286495216497776,
      "loss": 2.1953,
      "step": 108187
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.9589008688926697,
      "learning_rate": 0.0003286454497805178,
      "loss": 2.1298,
      "step": 108188
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.884350061416626,
      "learning_rate": 0.0003286413779059324,
      "loss": 2.1783,
      "step": 108189
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.9536256194114685,
      "learning_rate": 0.00032863730602602197,
      "loss": 2.2825,
      "step": 108190
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.8852646350860596,
      "learning_rate": 0.0003286332341407873,
      "loss": 2.1946,
      "step": 108191
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.9517644047737122,
      "learning_rate": 0.00032862916225022924,
      "loss": 2.2589,
      "step": 108192
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.8380120992660522,
      "learning_rate": 0.00032862509035434844,
      "loss": 1.975,
      "step": 108193
    },
    {
      "epoch": 1.41,
      "grad_norm": 1.0225682258605957,
      "learning_rate": 0.0003286210184531457,
      "loss": 2.156,
      "step": 108194
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.9411590099334717,
      "learning_rate": 0.00032861694654662174,
      "loss": 2.2759,
      "step": 108195
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.8644618391990662,
      "learning_rate": 0.00032861287463477735,
      "loss": 2.1428,
      "step": 108196
    },
    {
      "epoch": 1.41,
      "grad_norm": 1.0908535718917847,
      "learning_rate": 0.00032860880271761334,
      "loss": 2.0195,
      "step": 108197
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.8919999003410339,
      "learning_rate": 0.0003286047307951304,
      "loss": 2.1299,
      "step": 108198
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.855574905872345,
      "learning_rate": 0.0003286006588673293,
      "loss": 2.0693,
      "step": 108199
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.8540953993797302,
      "learning_rate": 0.00032859658693421077,
      "loss": 1.9278,
      "step": 108200
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.9065451622009277,
      "learning_rate": 0.00032859251499577557,
      "loss": 1.9848,
      "step": 108201
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.9563349485397339,
      "learning_rate": 0.0003285884430520245,
      "loss": 2.3299,
      "step": 108202
    },
    {
      "epoch": 1.41,
      "grad_norm": 1.0300464630126953,
      "learning_rate": 0.00032858437110295835,
      "loss": 2.0921,
      "step": 108203
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.9676086902618408,
      "learning_rate": 0.0003285802991485778,
      "loss": 2.082,
      "step": 108204
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.9418956637382507,
      "learning_rate": 0.00032857622718888356,
      "loss": 2.1264,
      "step": 108205
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.9637989401817322,
      "learning_rate": 0.0003285721552238765,
      "loss": 2.2679,
      "step": 108206
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.9462597966194153,
      "learning_rate": 0.0003285680832535575,
      "loss": 2.3342,
      "step": 108207
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.8300558924674988,
      "learning_rate": 0.000328564011277927,
      "loss": 1.9048,
      "step": 108208
    },
    {
      "epoch": 1.41,
      "grad_norm": 1.0103881359100342,
      "learning_rate": 0.00032855993929698586,
      "loss": 2.3276,
      "step": 108209
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.8484769463539124,
      "learning_rate": 0.000328555867310735,
      "loss": 2.3923,
      "step": 108210
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.8748043179512024,
      "learning_rate": 0.000328551795319175,
      "loss": 2.25,
      "step": 108211
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.9847287535667419,
      "learning_rate": 0.00032854772332230667,
      "loss": 2.0873,
      "step": 108212
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.8534132242202759,
      "learning_rate": 0.0003285436513201308,
      "loss": 2.3591,
      "step": 108213
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.8828771710395813,
      "learning_rate": 0.0003285395793126482,
      "loss": 2.346,
      "step": 108214
    },
    {
      "epoch": 1.41,
      "grad_norm": 1.093332052230835,
      "learning_rate": 0.0003285355072998594,
      "loss": 2.1351,
      "step": 108215
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.9514167904853821,
      "learning_rate": 0.00032853143528176545,
      "loss": 2.1332,
      "step": 108216
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.8845559358596802,
      "learning_rate": 0.00032852736325836697,
      "loss": 2.012,
      "step": 108217
    },
    {
      "epoch": 1.41,
      "grad_norm": 1.01598060131073,
      "learning_rate": 0.0003285232912296646,
      "loss": 2.2898,
      "step": 108218
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.9593123197555542,
      "learning_rate": 0.00032851921919565933,
      "loss": 1.9894,
      "step": 108219
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.9315134882926941,
      "learning_rate": 0.00032851514715635176,
      "loss": 2.3832,
      "step": 108220
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.9408571124076843,
      "learning_rate": 0.00032851107511174264,
      "loss": 1.9243,
      "step": 108221
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.9071856737136841,
      "learning_rate": 0.0003285070030618328,
      "loss": 2.1188,
      "step": 108222
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.9632108807563782,
      "learning_rate": 0.000328502931006623,
      "loss": 2.4048,
      "step": 108223
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.8877872824668884,
      "learning_rate": 0.000328498858946114,
      "loss": 2.2557,
      "step": 108224
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.9914640784263611,
      "learning_rate": 0.0003284947868803064,
      "loss": 2.2495,
      "step": 108225
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.9561679363250732,
      "learning_rate": 0.0003284907148092012,
      "loss": 2.1377,
      "step": 108226
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.8502196073532104,
      "learning_rate": 0.00032848664273279896,
      "loss": 2.048,
      "step": 108227
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.9890832901000977,
      "learning_rate": 0.00032848257065110054,
      "loss": 2.0449,
      "step": 108228
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.9974655508995056,
      "learning_rate": 0.00032847849856410666,
      "loss": 1.9967,
      "step": 108229
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.8754194974899292,
      "learning_rate": 0.00032847442647181813,
      "loss": 2.0898,
      "step": 108230
    },
    {
      "epoch": 1.41,
      "grad_norm": 1.0018582344055176,
      "learning_rate": 0.0003284703543742356,
      "loss": 2.1702,
      "step": 108231
    },
    {
      "epoch": 1.41,
      "grad_norm": 1.000595211982727,
      "learning_rate": 0.00032846628227136,
      "loss": 2.1967,
      "step": 108232
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.9510379433631897,
      "learning_rate": 0.0003284622101631919,
      "loss": 2.3381,
      "step": 108233
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.8935540318489075,
      "learning_rate": 0.00032845813804973213,
      "loss": 2.179,
      "step": 108234
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.9385708570480347,
      "learning_rate": 0.00032845406593098154,
      "loss": 1.941,
      "step": 108235
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.921112060546875,
      "learning_rate": 0.0003284499938069407,
      "loss": 2.0496,
      "step": 108236
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.8081091642379761,
      "learning_rate": 0.0003284459216776105,
      "loss": 2.0676,
      "step": 108237
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.9414486885070801,
      "learning_rate": 0.00032844184954299174,
      "loss": 2.1591,
      "step": 108238
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.8980088829994202,
      "learning_rate": 0.000328437777403085,
      "loss": 2.1578,
      "step": 108239
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.8341771960258484,
      "learning_rate": 0.00032843370525789114,
      "loss": 2.0697,
      "step": 108240
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.802905261516571,
      "learning_rate": 0.000328429633107411,
      "loss": 1.8958,
      "step": 108241
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.9753924012184143,
      "learning_rate": 0.0003284255609516452,
      "loss": 2.2093,
      "step": 108242
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.9877192378044128,
      "learning_rate": 0.00032842148879059455,
      "loss": 2.262,
      "step": 108243
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.9295611381530762,
      "learning_rate": 0.00032841741662425984,
      "loss": 2.1891,
      "step": 108244
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.9127046465873718,
      "learning_rate": 0.0003284133444526418,
      "loss": 2.1083,
      "step": 108245
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.8690125346183777,
      "learning_rate": 0.0003284092722757411,
      "loss": 2.0668,
      "step": 108246
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.8139014840126038,
      "learning_rate": 0.00032840520009355873,
      "loss": 2.0209,
      "step": 108247
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.9211660027503967,
      "learning_rate": 0.00032840112790609514,
      "loss": 1.9124,
      "step": 108248
    },
    {
      "epoch": 1.41,
      "grad_norm": 1.2330303192138672,
      "learning_rate": 0.00032839705571335127,
      "loss": 2.0498,
      "step": 108249
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.922585666179657,
      "learning_rate": 0.000328392983515328,
      "loss": 2.2523,
      "step": 108250
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.8978206515312195,
      "learning_rate": 0.00032838891131202583,
      "loss": 2.1816,
      "step": 108251
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.8844589591026306,
      "learning_rate": 0.0003283848391034456,
      "loss": 2.0415,
      "step": 108252
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.9142856001853943,
      "learning_rate": 0.00032838076688958815,
      "loss": 2.1389,
      "step": 108253
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.9773228764533997,
      "learning_rate": 0.0003283766946704541,
      "loss": 2.1129,
      "step": 108254
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.9690389633178711,
      "learning_rate": 0.00032837262244604437,
      "loss": 2.0775,
      "step": 108255
    },
    {
      "epoch": 1.41,
      "grad_norm": 1.002379059791565,
      "learning_rate": 0.0003283685502163596,
      "loss": 2.0453,
      "step": 108256
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.8860685229301453,
      "learning_rate": 0.0003283644779814006,
      "loss": 1.991,
      "step": 108257
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.8718129396438599,
      "learning_rate": 0.000328360405741168,
      "loss": 2.2506,
      "step": 108258
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.8668993711471558,
      "learning_rate": 0.00032835633349566286,
      "loss": 2.2359,
      "step": 108259
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.9716302156448364,
      "learning_rate": 0.0003283522612448856,
      "loss": 2.361,
      "step": 108260
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.9267394542694092,
      "learning_rate": 0.0003283481889888371,
      "loss": 1.9941,
      "step": 108261
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.7985746264457703,
      "learning_rate": 0.0003283441167275183,
      "loss": 2.055,
      "step": 108262
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.9417716264724731,
      "learning_rate": 0.00032834004446092967,
      "loss": 2.079,
      "step": 108263
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.9062893390655518,
      "learning_rate": 0.00032833597218907205,
      "loss": 1.9949,
      "step": 108264
    },
    {
      "epoch": 1.41,
      "grad_norm": 1.1462143659591675,
      "learning_rate": 0.0003283318999119463,
      "loss": 2.0818,
      "step": 108265
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.8523281216621399,
      "learning_rate": 0.0003283278276295531,
      "loss": 2.4063,
      "step": 108266
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.9872894883155823,
      "learning_rate": 0.00032832375534189325,
      "loss": 2.157,
      "step": 108267
    },
    {
      "epoch": 1.41,
      "grad_norm": 1.0393118858337402,
      "learning_rate": 0.0003283196830489674,
      "loss": 2.0479,
      "step": 108268
    },
    {
      "epoch": 1.41,
      "grad_norm": 1.0327427387237549,
      "learning_rate": 0.0003283156107507765,
      "loss": 2.3114,
      "step": 108269
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.9111501574516296,
      "learning_rate": 0.0003283115384473211,
      "loss": 2.2767,
      "step": 108270
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.8461046814918518,
      "learning_rate": 0.0003283074661386021,
      "loss": 2.1437,
      "step": 108271
    },
    {
      "epoch": 1.41,
      "grad_norm": 1.0851258039474487,
      "learning_rate": 0.00032830339382462027,
      "loss": 1.9192,
      "step": 108272
    },
    {
      "epoch": 1.41,
      "grad_norm": 1.1262743473052979,
      "learning_rate": 0.00032829932150537625,
      "loss": 2.0433,
      "step": 108273
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.8505817651748657,
      "learning_rate": 0.00032829524918087075,
      "loss": 2.2781,
      "step": 108274
    },
    {
      "epoch": 1.41,
      "grad_norm": 1.1715471744537354,
      "learning_rate": 0.0003282911768511048,
      "loss": 2.2094,
      "step": 108275
    },
    {
      "epoch": 1.41,
      "grad_norm": 1.2195721864700317,
      "learning_rate": 0.00032828710451607887,
      "loss": 2.4201,
      "step": 108276
    },
    {
      "epoch": 1.41,
      "grad_norm": 1.0218702554702759,
      "learning_rate": 0.00032828303217579384,
      "loss": 2.1781,
      "step": 108277
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.9374240636825562,
      "learning_rate": 0.00032827895983025053,
      "loss": 2.102,
      "step": 108278
    },
    {
      "epoch": 1.41,
      "grad_norm": 1.276147484779358,
      "learning_rate": 0.00032827488747944956,
      "loss": 2.3904,
      "step": 108279
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.9340892434120178,
      "learning_rate": 0.00032827081512339177,
      "loss": 2.1432,
      "step": 108280
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.8505265116691589,
      "learning_rate": 0.00032826674276207794,
      "loss": 2.1424,
      "step": 108281
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.9156564474105835,
      "learning_rate": 0.00032826267039550875,
      "loss": 2.2556,
      "step": 108282
    },
    {
      "epoch": 1.41,
      "grad_norm": 1.0833402872085571,
      "learning_rate": 0.000328258598023685,
      "loss": 2.105,
      "step": 108283
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.8839297294616699,
      "learning_rate": 0.00032825452564660745,
      "loss": 1.8345,
      "step": 108284
    },
    {
      "epoch": 1.41,
      "grad_norm": 1.0645757913589478,
      "learning_rate": 0.00032825045326427684,
      "loss": 2.3114,
      "step": 108285
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.9287512898445129,
      "learning_rate": 0.00032824638087669393,
      "loss": 2.4273,
      "step": 108286
    },
    {
      "epoch": 1.41,
      "grad_norm": 1.0386483669281006,
      "learning_rate": 0.00032824230848385946,
      "loss": 2.1045,
      "step": 108287
    },
    {
      "epoch": 1.41,
      "grad_norm": 1.034376621246338,
      "learning_rate": 0.0003282382360857743,
      "loss": 2.1195,
      "step": 108288
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.8385164141654968,
      "learning_rate": 0.0003282341636824391,
      "loss": 2.2137,
      "step": 108289
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.9928187131881714,
      "learning_rate": 0.0003282300912738545,
      "loss": 2.2126,
      "step": 108290
    },
    {
      "epoch": 1.41,
      "grad_norm": 1.06150484085083,
      "learning_rate": 0.00032822601886002166,
      "loss": 2.1217,
      "step": 108291
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.8468798995018005,
      "learning_rate": 0.0003282219464409408,
      "loss": 2.0848,
      "step": 108292
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.9830037951469421,
      "learning_rate": 0.00032821787401661305,
      "loss": 2.1964,
      "step": 108293
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.8715113401412964,
      "learning_rate": 0.0003282138015870391,
      "loss": 2.0743,
      "step": 108294
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.9312788248062134,
      "learning_rate": 0.00032820972915221966,
      "loss": 2.0701,
      "step": 108295
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.9337587356567383,
      "learning_rate": 0.0003282056567121555,
      "loss": 2.1046,
      "step": 108296
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.9483609795570374,
      "learning_rate": 0.00032820158426684733,
      "loss": 2.109,
      "step": 108297
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.887155294418335,
      "learning_rate": 0.000328197511816296,
      "loss": 2.1473,
      "step": 108298
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.8477723598480225,
      "learning_rate": 0.0003281934393605022,
      "loss": 2.0939,
      "step": 108299
    },
    {
      "epoch": 1.41,
      "grad_norm": 1.0328532457351685,
      "learning_rate": 0.00032818936689946664,
      "loss": 2.234,
      "step": 108300
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.990975558757782,
      "learning_rate": 0.0003281852944331903,
      "loss": 1.9215,
      "step": 108301
    },
    {
      "epoch": 1.41,
      "grad_norm": 1.008199691772461,
      "learning_rate": 0.00032818122196167363,
      "loss": 2.1464,
      "step": 108302
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.8546015620231628,
      "learning_rate": 0.00032817714948491763,
      "loss": 2.0399,
      "step": 108303
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.8401480317115784,
      "learning_rate": 0.00032817307700292294,
      "loss": 2.0901,
      "step": 108304
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.8765089511871338,
      "learning_rate": 0.00032816900451569035,
      "loss": 2.2036,
      "step": 108305
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.8904579877853394,
      "learning_rate": 0.00032816493202322063,
      "loss": 2.0979,
      "step": 108306
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.9084742665290833,
      "learning_rate": 0.00032816085952551443,
      "loss": 2.2499,
      "step": 108307
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.863957941532135,
      "learning_rate": 0.0003281567870225727,
      "loss": 2.284,
      "step": 108308
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.9566771984100342,
      "learning_rate": 0.00032815271451439607,
      "loss": 2.2423,
      "step": 108309
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.9038106203079224,
      "learning_rate": 0.00032814864200098525,
      "loss": 2.2109,
      "step": 108310
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.8650845289230347,
      "learning_rate": 0.00032814456948234116,
      "loss": 2.1009,
      "step": 108311
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.8814075589179993,
      "learning_rate": 0.0003281404969584644,
      "loss": 2.1823,
      "step": 108312
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.9221534132957458,
      "learning_rate": 0.0003281364244293558,
      "loss": 2.0184,
      "step": 108313
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.9291163086891174,
      "learning_rate": 0.00032813235189501613,
      "loss": 2.2489,
      "step": 108314
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.8343915343284607,
      "learning_rate": 0.0003281282793554461,
      "loss": 2.0532,
      "step": 108315
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.8685081601142883,
      "learning_rate": 0.0003281242068106465,
      "loss": 2.0469,
      "step": 108316
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.8507367372512817,
      "learning_rate": 0.0003281201342606181,
      "loss": 2.3232,
      "step": 108317
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.9931545257568359,
      "learning_rate": 0.0003281160617053616,
      "loss": 2.1516,
      "step": 108318
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.8531166911125183,
      "learning_rate": 0.00032811198914487784,
      "loss": 1.9835,
      "step": 108319
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.8886667490005493,
      "learning_rate": 0.0003281079165791675,
      "loss": 2.5288,
      "step": 108320
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.8868642449378967,
      "learning_rate": 0.0003281038440082313,
      "loss": 2.0293,
      "step": 108321
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.9267057180404663,
      "learning_rate": 0.0003280997714320702,
      "loss": 2.136,
      "step": 108322
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.8486685156822205,
      "learning_rate": 0.00032809569885068477,
      "loss": 2.187,
      "step": 108323
    },
    {
      "epoch": 1.41,
      "grad_norm": 1.0134379863739014,
      "learning_rate": 0.00032809162626407573,
      "loss": 1.9942,
      "step": 108324
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.9434437155723572,
      "learning_rate": 0.00032808755367224403,
      "loss": 2.0965,
      "step": 108325
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.9044403433799744,
      "learning_rate": 0.00032808348107519033,
      "loss": 2.171,
      "step": 108326
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.9215784668922424,
      "learning_rate": 0.00032807940847291527,
      "loss": 2.0142,
      "step": 108327
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.9074531197547913,
      "learning_rate": 0.00032807533586541983,
      "loss": 2.0655,
      "step": 108328
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.8678842186927795,
      "learning_rate": 0.0003280712632527046,
      "loss": 2.1687,
      "step": 108329
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.9242986440658569,
      "learning_rate": 0.0003280671906347704,
      "loss": 2.3655,
      "step": 108330
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.9053328633308411,
      "learning_rate": 0.000328063118011618,
      "loss": 2.3123,
      "step": 108331
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.9540641903877258,
      "learning_rate": 0.0003280590453832481,
      "loss": 2.2683,
      "step": 108332
    },
    {
      "epoch": 1.41,
      "grad_norm": 1.0084731578826904,
      "learning_rate": 0.00032805497274966157,
      "loss": 2.2829,
      "step": 108333
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.883574903011322,
      "learning_rate": 0.00032805090011085904,
      "loss": 2.0343,
      "step": 108334
    },
    {
      "epoch": 1.41,
      "grad_norm": 1.031563401222229,
      "learning_rate": 0.0003280468274668413,
      "loss": 2.2279,
      "step": 108335
    },
    {
      "epoch": 1.41,
      "grad_norm": 1.0928317308425903,
      "learning_rate": 0.0003280427548176092,
      "loss": 2.0695,
      "step": 108336
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.8282361626625061,
      "learning_rate": 0.00032803868216316337,
      "loss": 1.9346,
      "step": 108337
    },
    {
      "epoch": 1.41,
      "grad_norm": 1.0185281038284302,
      "learning_rate": 0.0003280346095035046,
      "loss": 2.1255,
      "step": 108338
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.9872558116912842,
      "learning_rate": 0.0003280305368386337,
      "loss": 2.1281,
      "step": 108339
    },
    {
      "epoch": 1.41,
      "grad_norm": 1.0552252531051636,
      "learning_rate": 0.0003280264641685514,
      "loss": 2.3764,
      "step": 108340
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.8147933483123779,
      "learning_rate": 0.00032802239149325836,
      "loss": 2.1185,
      "step": 108341
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.9004658460617065,
      "learning_rate": 0.0003280183188127555,
      "loss": 2.1553,
      "step": 108342
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.9446712732315063,
      "learning_rate": 0.0003280142461270435,
      "loss": 2.2481,
      "step": 108343
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.9000644683837891,
      "learning_rate": 0.00032801017343612315,
      "loss": 2.048,
      "step": 108344
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.9652746915817261,
      "learning_rate": 0.0003280061007399952,
      "loss": 2.1257,
      "step": 108345
    },
    {
      "epoch": 1.41,
      "grad_norm": 1.0725270509719849,
      "learning_rate": 0.0003280020280386603,
      "loss": 2.3577,
      "step": 108346
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.9666911363601685,
      "learning_rate": 0.00032799795533211936,
      "loss": 2.2957,
      "step": 108347
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.9727658629417419,
      "learning_rate": 0.0003279938826203731,
      "loss": 2.1307,
      "step": 108348
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.9398951530456543,
      "learning_rate": 0.00032798980990342215,
      "loss": 2.2515,
      "step": 108349
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.9022248983383179,
      "learning_rate": 0.0003279857371812674,
      "loss": 2.1896,
      "step": 108350
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.8776193857192993,
      "learning_rate": 0.00032798166445390957,
      "loss": 2.2065,
      "step": 108351
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.8435453772544861,
      "learning_rate": 0.00032797759172134946,
      "loss": 2.0196,
      "step": 108352
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.9951428771018982,
      "learning_rate": 0.0003279735189835877,
      "loss": 2.2565,
      "step": 108353
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.8642644286155701,
      "learning_rate": 0.0003279694462406253,
      "loss": 2.1418,
      "step": 108354
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.9757751226425171,
      "learning_rate": 0.00032796537349246276,
      "loss": 1.9738,
      "step": 108355
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.9885565638542175,
      "learning_rate": 0.00032796130073910087,
      "loss": 2.053,
      "step": 108356
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.8647324442863464,
      "learning_rate": 0.0003279572279805405,
      "loss": 2.3342,
      "step": 108357
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.8340619802474976,
      "learning_rate": 0.00032795315521678243,
      "loss": 1.9811,
      "step": 108358
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.8208437561988831,
      "learning_rate": 0.00032794908244782717,
      "loss": 2.1979,
      "step": 108359
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.90585857629776,
      "learning_rate": 0.00032794500967367577,
      "loss": 2.1382,
      "step": 108360
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.8947412371635437,
      "learning_rate": 0.00032794093689432886,
      "loss": 2.2443,
      "step": 108361
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.9127969145774841,
      "learning_rate": 0.0003279368641097871,
      "loss": 2.2499,
      "step": 108362
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.8837850689888,
      "learning_rate": 0.00032793279132005155,
      "loss": 2.0604,
      "step": 108363
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.8219357132911682,
      "learning_rate": 0.0003279287185251226,
      "loss": 1.9457,
      "step": 108364
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.8736233711242676,
      "learning_rate": 0.0003279246457250012,
      "loss": 2.0218,
      "step": 108365
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.8878676891326904,
      "learning_rate": 0.00032792057291968814,
      "loss": 2.0364,
      "step": 108366
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.8537921905517578,
      "learning_rate": 0.0003279165001091841,
      "loss": 2.2436,
      "step": 108367
    },
    {
      "epoch": 1.41,
      "grad_norm": 1.1350986957550049,
      "learning_rate": 0.0003279124272934898,
      "loss": 2.1022,
      "step": 108368
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.8149097561836243,
      "learning_rate": 0.00032790835447260607,
      "loss": 2.0695,
      "step": 108369
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.9270318150520325,
      "learning_rate": 0.0003279042816465337,
      "loss": 1.9918,
      "step": 108370
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.9751095175743103,
      "learning_rate": 0.0003279002088152733,
      "loss": 2.1834,
      "step": 108371
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.8789824843406677,
      "learning_rate": 0.00032789613597882585,
      "loss": 2.1571,
      "step": 108372
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.921125054359436,
      "learning_rate": 0.0003278920631371919,
      "loss": 2.3363,
      "step": 108373
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.8896322846412659,
      "learning_rate": 0.00032788799029037234,
      "loss": 2.1388,
      "step": 108374
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.9695920348167419,
      "learning_rate": 0.0003278839174383678,
      "loss": 2.037,
      "step": 108375
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.8485496640205383,
      "learning_rate": 0.00032787984458117913,
      "loss": 2.1948,
      "step": 108376
    },
    {
      "epoch": 1.41,
      "grad_norm": 1.1742348670959473,
      "learning_rate": 0.0003278757717188071,
      "loss": 2.0983,
      "step": 108377
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.9365637302398682,
      "learning_rate": 0.00032787169885125245,
      "loss": 2.0693,
      "step": 108378
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.9935708045959473,
      "learning_rate": 0.00032786762597851595,
      "loss": 2.1246,
      "step": 108379
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.9643787145614624,
      "learning_rate": 0.00032786355310059827,
      "loss": 2.1923,
      "step": 108380
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.8668461441993713,
      "learning_rate": 0.00032785948021750027,
      "loss": 2.3125,
      "step": 108381
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.9568639993667603,
      "learning_rate": 0.0003278554073292226,
      "loss": 2.088,
      "step": 108382
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.9048768281936646,
      "learning_rate": 0.0003278513344357662,
      "loss": 2.0301,
      "step": 108383
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.9525381326675415,
      "learning_rate": 0.0003278472615371316,
      "loss": 1.9634,
      "step": 108384
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.9449300765991211,
      "learning_rate": 0.0003278431886333197,
      "loss": 2.1053,
      "step": 108385
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.8748555183410645,
      "learning_rate": 0.00032783911572433123,
      "loss": 2.1325,
      "step": 108386
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.9521676898002625,
      "learning_rate": 0.00032783504281016704,
      "loss": 2.2606,
      "step": 108387
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.8976139426231384,
      "learning_rate": 0.00032783096989082763,
      "loss": 2.2783,
      "step": 108388
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.9927066564559937,
      "learning_rate": 0.00032782689696631397,
      "loss": 2.0655,
      "step": 108389
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.8652578592300415,
      "learning_rate": 0.0003278228240366269,
      "loss": 2.2615,
      "step": 108390
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.9529139399528503,
      "learning_rate": 0.0003278187511017668,
      "loss": 1.945,
      "step": 108391
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.9004449248313904,
      "learning_rate": 0.00032781467816173483,
      "loss": 2.3205,
      "step": 108392
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.9064311385154724,
      "learning_rate": 0.0003278106052165316,
      "loss": 2.3364,
      "step": 108393
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.9445629119873047,
      "learning_rate": 0.0003278065322661578,
      "loss": 2.159,
      "step": 108394
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.8425231575965881,
      "learning_rate": 0.0003278024593106142,
      "loss": 2.3104,
      "step": 108395
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.9773007035255432,
      "learning_rate": 0.00032779838634990164,
      "loss": 2.1537,
      "step": 108396
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.9555060267448425,
      "learning_rate": 0.0003277943133840209,
      "loss": 2.3048,
      "step": 108397
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.8554258942604065,
      "learning_rate": 0.0003277902404129726,
      "loss": 1.9126,
      "step": 108398
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.9402533769607544,
      "learning_rate": 0.00032778616743675754,
      "loss": 2.0826,
      "step": 108399
    },
    {
      "epoch": 1.41,
      "grad_norm": 1.0685081481933594,
      "learning_rate": 0.00032778209445537656,
      "loss": 2.3707,
      "step": 108400
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.8737490177154541,
      "learning_rate": 0.0003277780214688303,
      "loss": 2.2855,
      "step": 108401
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.885762631893158,
      "learning_rate": 0.00032777394847711966,
      "loss": 2.2024,
      "step": 108402
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.9633713364601135,
      "learning_rate": 0.0003277698754802453,
      "loss": 2.0149,
      "step": 108403
    },
    {
      "epoch": 1.41,
      "grad_norm": 1.1849303245544434,
      "learning_rate": 0.000327765802478208,
      "loss": 2.294,
      "step": 108404
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.9166290163993835,
      "learning_rate": 0.00032776172947100843,
      "loss": 2.0247,
      "step": 108405
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.9166858196258545,
      "learning_rate": 0.0003277576564586476,
      "loss": 2.0083,
      "step": 108406
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.8740671277046204,
      "learning_rate": 0.000327753583441126,
      "loss": 2.1958,
      "step": 108407
    },
    {
      "epoch": 1.41,
      "grad_norm": 1.0627354383468628,
      "learning_rate": 0.0003277495104184444,
      "loss": 2.2801,
      "step": 108408
    },
    {
      "epoch": 1.41,
      "grad_norm": 1.055200219154358,
      "learning_rate": 0.0003277454373906038,
      "loss": 2.1869,
      "step": 108409
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.893266499042511,
      "learning_rate": 0.00032774136435760473,
      "loss": 2.2315,
      "step": 108410
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.9676496982574463,
      "learning_rate": 0.000327737291319448,
      "loss": 2.2043,
      "step": 108411
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.9624595642089844,
      "learning_rate": 0.00032773321827613445,
      "loss": 2.2428,
      "step": 108412
    },
    {
      "epoch": 1.41,
      "grad_norm": 1.0311192274093628,
      "learning_rate": 0.00032772914522766466,
      "loss": 2.171,
      "step": 108413
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.9305251240730286,
      "learning_rate": 0.00032772507217403954,
      "loss": 2.2992,
      "step": 108414
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.9891179800033569,
      "learning_rate": 0.0003277209991152599,
      "loss": 2.2051,
      "step": 108415
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.9889330863952637,
      "learning_rate": 0.0003277169260513263,
      "loss": 1.9716,
      "step": 108416
    },
    {
      "epoch": 1.41,
      "grad_norm": 1.1368205547332764,
      "learning_rate": 0.00032771285298223955,
      "loss": 2.0123,
      "step": 108417
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.9294390082359314,
      "learning_rate": 0.0003277087799080006,
      "loss": 2.2136,
      "step": 108418
    },
    {
      "epoch": 1.41,
      "grad_norm": 1.0533852577209473,
      "learning_rate": 0.00032770470682860997,
      "loss": 2.028,
      "step": 108419
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.8827372193336487,
      "learning_rate": 0.00032770063374406856,
      "loss": 2.2516,
      "step": 108420
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.9994432926177979,
      "learning_rate": 0.0003276965606543771,
      "loss": 1.9542,
      "step": 108421
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.9668020009994507,
      "learning_rate": 0.0003276924875595363,
      "loss": 2.2073,
      "step": 108422
    },
    {
      "epoch": 1.41,
      "grad_norm": 1.128058671951294,
      "learning_rate": 0.00032768841445954684,
      "loss": 2.0818,
      "step": 108423
    },
    {
      "epoch": 1.41,
      "grad_norm": 1.033393144607544,
      "learning_rate": 0.00032768434135440976,
      "loss": 2.193,
      "step": 108424
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.8778643608093262,
      "learning_rate": 0.0003276802682441255,
      "loss": 2.2169,
      "step": 108425
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.8560969829559326,
      "learning_rate": 0.00032767619512869505,
      "loss": 1.9801,
      "step": 108426
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.8667870759963989,
      "learning_rate": 0.000327672122008119,
      "loss": 2.0134,
      "step": 108427
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.8852720856666565,
      "learning_rate": 0.0003276680488823982,
      "loss": 2.4441,
      "step": 108428
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.8124589323997498,
      "learning_rate": 0.0003276639757515334,
      "loss": 2.0951,
      "step": 108429
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.8975794315338135,
      "learning_rate": 0.00032765990261552535,
      "loss": 2.2994,
      "step": 108430
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.9298554062843323,
      "learning_rate": 0.0003276558294743748,
      "loss": 2.2265,
      "step": 108431
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.8575894236564636,
      "learning_rate": 0.0003276517563280825,
      "loss": 2.0213,
      "step": 108432
    },
    {
      "epoch": 1.41,
      "grad_norm": 1.3009696006774902,
      "learning_rate": 0.00032764768317664923,
      "loss": 2.3279,
      "step": 108433
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.9319273829460144,
      "learning_rate": 0.0003276436100200757,
      "loss": 2.207,
      "step": 108434
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.8767277002334595,
      "learning_rate": 0.00032763953685836274,
      "loss": 2.2616,
      "step": 108435
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.9249305129051208,
      "learning_rate": 0.0003276354636915111,
      "loss": 2.0803,
      "step": 108436
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.9538129568099976,
      "learning_rate": 0.0003276313905195214,
      "loss": 2.0097,
      "step": 108437
    },
    {
      "epoch": 1.41,
      "grad_norm": 1.3842352628707886,
      "learning_rate": 0.00032762731734239454,
      "loss": 2.0226,
      "step": 108438
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.9313584566116333,
      "learning_rate": 0.0003276232441601313,
      "loss": 2.1279,
      "step": 108439
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.8991650938987732,
      "learning_rate": 0.00032761917097273236,
      "loss": 1.9556,
      "step": 108440
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.8757677674293518,
      "learning_rate": 0.00032761509778019845,
      "loss": 1.8846,
      "step": 108441
    },
    {
      "epoch": 1.41,
      "grad_norm": 1.0467489957809448,
      "learning_rate": 0.0003276110245825304,
      "loss": 2.0114,
      "step": 108442
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.9936872124671936,
      "learning_rate": 0.000327606951379729,
      "loss": 2.3045,
      "step": 108443
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.9466583132743835,
      "learning_rate": 0.0003276028781717948,
      "loss": 2.3449,
      "step": 108444
    },
    {
      "epoch": 1.41,
      "grad_norm": 1.171170711517334,
      "learning_rate": 0.00032759880495872887,
      "loss": 2.083,
      "step": 108445
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.9872701168060303,
      "learning_rate": 0.0003275947317405317,
      "loss": 2.1099,
      "step": 108446
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.8653262257575989,
      "learning_rate": 0.0003275906585172041,
      "loss": 2.1392,
      "step": 108447
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.9515483379364014,
      "learning_rate": 0.000327586585288747,
      "loss": 2.0604,
      "step": 108448
    },
    {
      "epoch": 1.41,
      "grad_norm": 1.2857118844985962,
      "learning_rate": 0.00032758251205516095,
      "loss": 1.7833,
      "step": 108449
    },
    {
      "epoch": 1.41,
      "grad_norm": 1.2145057916641235,
      "learning_rate": 0.00032757843881644675,
      "loss": 2.1882,
      "step": 108450
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.9053441882133484,
      "learning_rate": 0.00032757436557260535,
      "loss": 2.1706,
      "step": 108451
    },
    {
      "epoch": 1.41,
      "grad_norm": 1.1056009531021118,
      "learning_rate": 0.0003275702923236372,
      "loss": 1.9838,
      "step": 108452
    },
    {
      "epoch": 1.41,
      "grad_norm": 1.1775695085525513,
      "learning_rate": 0.00032756621906954327,
      "loss": 2.0526,
      "step": 108453
    },
    {
      "epoch": 1.41,
      "grad_norm": 1.054257869720459,
      "learning_rate": 0.00032756214581032436,
      "loss": 2.0688,
      "step": 108454
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.9043323993682861,
      "learning_rate": 0.00032755807254598103,
      "loss": 2.1569,
      "step": 108455
    },
    {
      "epoch": 1.41,
      "grad_norm": 1.2834659814834595,
      "learning_rate": 0.0003275539992765141,
      "loss": 2.3963,
      "step": 108456
    },
    {
      "epoch": 1.41,
      "grad_norm": 1.2270981073379517,
      "learning_rate": 0.0003275499260019244,
      "loss": 1.9512,
      "step": 108457
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.9372332692146301,
      "learning_rate": 0.00032754585272221266,
      "loss": 2.2045,
      "step": 108458
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.8300833106040955,
      "learning_rate": 0.0003275417794373796,
      "loss": 2.0264,
      "step": 108459
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.8864255547523499,
      "learning_rate": 0.000327537706147426,
      "loss": 2.2645,
      "step": 108460
    },
    {
      "epoch": 1.41,
      "grad_norm": 1.059401035308838,
      "learning_rate": 0.0003275336328523527,
      "loss": 2.1956,
      "step": 108461
    },
    {
      "epoch": 1.41,
      "grad_norm": 1.176248550415039,
      "learning_rate": 0.00032752955955216026,
      "loss": 2.1731,
      "step": 108462
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.8832524418830872,
      "learning_rate": 0.00032752548624684965,
      "loss": 2.0889,
      "step": 108463
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.9144102931022644,
      "learning_rate": 0.0003275214129364215,
      "loss": 2.1777,
      "step": 108464
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.9388395547866821,
      "learning_rate": 0.00032751733962087655,
      "loss": 2.1628,
      "step": 108465
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.9767181277275085,
      "learning_rate": 0.00032751326630021564,
      "loss": 2.1282,
      "step": 108466
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.8533377647399902,
      "learning_rate": 0.0003275091929744395,
      "loss": 2.1916,
      "step": 108467
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.86238694190979,
      "learning_rate": 0.0003275051196435489,
      "loss": 2.0286,
      "step": 108468
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.999068021774292,
      "learning_rate": 0.00032750104630754456,
      "loss": 2.3285,
      "step": 108469
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.9155808091163635,
      "learning_rate": 0.0003274969729664273,
      "loss": 2.1317,
      "step": 108470
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.9718897342681885,
      "learning_rate": 0.00032749289962019775,
      "loss": 1.9723,
      "step": 108471
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.8808242678642273,
      "learning_rate": 0.0003274888262688568,
      "loss": 2.2395,
      "step": 108472
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.9857341051101685,
      "learning_rate": 0.0003274847529124051,
      "loss": 1.9145,
      "step": 108473
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.9471290707588196,
      "learning_rate": 0.00032748067955084355,
      "loss": 2.1335,
      "step": 108474
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.9711546301841736,
      "learning_rate": 0.00032747660618417284,
      "loss": 2.0738,
      "step": 108475
    },
    {
      "epoch": 1.41,
      "grad_norm": 1.0575562715530396,
      "learning_rate": 0.00032747253281239364,
      "loss": 2.1978,
      "step": 108476
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.8665145635604858,
      "learning_rate": 0.00032746845943550676,
      "loss": 2.2587,
      "step": 108477
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.9670331478118896,
      "learning_rate": 0.000327464386053513,
      "loss": 2.2071,
      "step": 108478
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.8763589262962341,
      "learning_rate": 0.00032746031266641317,
      "loss": 2.2741,
      "step": 108479
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.8925915360450745,
      "learning_rate": 0.00032745623927420787,
      "loss": 2.3694,
      "step": 108480
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.829804003238678,
      "learning_rate": 0.00032745216587689793,
      "loss": 2.0623,
      "step": 108481
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.8333681225776672,
      "learning_rate": 0.0003274480924744842,
      "loss": 2.1969,
      "step": 108482
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.9015402793884277,
      "learning_rate": 0.0003274440190669672,
      "loss": 2.1724,
      "step": 108483
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.9244281649589539,
      "learning_rate": 0.00032743994565434797,
      "loss": 2.2286,
      "step": 108484
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.8961076736450195,
      "learning_rate": 0.00032743587223662716,
      "loss": 1.8773,
      "step": 108485
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.9493849873542786,
      "learning_rate": 0.0003274317988138054,
      "loss": 2.2689,
      "step": 108486
    },
    {
      "epoch": 1.41,
      "grad_norm": 1.0355708599090576,
      "learning_rate": 0.00032742772538588356,
      "loss": 2.0368,
      "step": 108487
    },
    {
      "epoch": 1.41,
      "grad_norm": 1.0108224153518677,
      "learning_rate": 0.0003274236519528625,
      "loss": 2.0942,
      "step": 108488
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.9431854486465454,
      "learning_rate": 0.0003274195785147427,
      "loss": 1.9637,
      "step": 108489
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.9018478989601135,
      "learning_rate": 0.0003274155050715252,
      "loss": 2.194,
      "step": 108490
    },
    {
      "epoch": 1.41,
      "grad_norm": 1.001848816871643,
      "learning_rate": 0.00032741143162321064,
      "loss": 2.0652,
      "step": 108491
    },
    {
      "epoch": 1.41,
      "grad_norm": 1.026479721069336,
      "learning_rate": 0.00032740735816979975,
      "loss": 1.849,
      "step": 108492
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.8939732909202576,
      "learning_rate": 0.00032740328471129326,
      "loss": 2.1382,
      "step": 108493
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.8090766668319702,
      "learning_rate": 0.0003273992112476921,
      "loss": 2.0114,
      "step": 108494
    },
    {
      "epoch": 1.41,
      "grad_norm": 1.030030608177185,
      "learning_rate": 0.0003273951377789968,
      "loss": 2.1403,
      "step": 108495
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.989873468875885,
      "learning_rate": 0.00032739106430520825,
      "loss": 2.2381,
      "step": 108496
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.9320708513259888,
      "learning_rate": 0.0003273869908263272,
      "loss": 2.2041,
      "step": 108497
    },
    {
      "epoch": 1.41,
      "grad_norm": 1.0680948495864868,
      "learning_rate": 0.00032738291734235445,
      "loss": 2.2259,
      "step": 108498
    },
    {
      "epoch": 1.41,
      "grad_norm": 1.0591427087783813,
      "learning_rate": 0.0003273788438532906,
      "loss": 2.0923,
      "step": 108499
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.9033148288726807,
      "learning_rate": 0.0003273747703591366,
      "loss": 2.1884,
      "step": 108500
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.9250344634056091,
      "learning_rate": 0.00032737069685989304,
      "loss": 1.9984,
      "step": 108501
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.9855780005455017,
      "learning_rate": 0.00032736662335556074,
      "loss": 1.9341,
      "step": 108502
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.9222858548164368,
      "learning_rate": 0.0003273625498461405,
      "loss": 2.1177,
      "step": 108503
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.8658369183540344,
      "learning_rate": 0.0003273584763316331,
      "loss": 2.1078,
      "step": 108504
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.9802647829055786,
      "learning_rate": 0.0003273544028120391,
      "loss": 2.2258,
      "step": 108505
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.9498332738876343,
      "learning_rate": 0.0003273503292873596,
      "loss": 2.2758,
      "step": 108506
    },
    {
      "epoch": 1.41,
      "grad_norm": 1.2262881994247437,
      "learning_rate": 0.00032734625575759496,
      "loss": 1.7971,
      "step": 108507
    },
    {
      "epoch": 1.41,
      "grad_norm": 1.0047717094421387,
      "learning_rate": 0.0003273421822227463,
      "loss": 2.072,
      "step": 108508
    },
    {
      "epoch": 1.41,
      "grad_norm": 1.1264816522598267,
      "learning_rate": 0.0003273381086828141,
      "loss": 2.0296,
      "step": 108509
    },
    {
      "epoch": 1.41,
      "grad_norm": 1.0934689044952393,
      "learning_rate": 0.0003273340351377993,
      "loss": 2.2274,
      "step": 108510
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.9233244061470032,
      "learning_rate": 0.0003273299615877025,
      "loss": 2.2846,
      "step": 108511
    },
    {
      "epoch": 1.41,
      "grad_norm": 1.0386366844177246,
      "learning_rate": 0.00032732588803252463,
      "loss": 1.9001,
      "step": 108512
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.9586784839630127,
      "learning_rate": 0.0003273218144722663,
      "loss": 2.1319,
      "step": 108513
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.9204176664352417,
      "learning_rate": 0.00032731774090692835,
      "loss": 1.9795,
      "step": 108514
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.9092826843261719,
      "learning_rate": 0.00032731366733651155,
      "loss": 2.2082,
      "step": 108515
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.9160207509994507,
      "learning_rate": 0.00032730959376101656,
      "loss": 2.0141,
      "step": 108516
    },
    {
      "epoch": 1.41,
      "grad_norm": 1.221375823020935,
      "learning_rate": 0.00032730552018044425,
      "loss": 2.2267,
      "step": 108517
    },
    {
      "epoch": 1.41,
      "grad_norm": 1.0619925260543823,
      "learning_rate": 0.0003273014465947953,
      "loss": 2.2949,
      "step": 108518
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.9020814299583435,
      "learning_rate": 0.0003272973730040705,
      "loss": 2.3257,
      "step": 108519
    },
    {
      "epoch": 1.41,
      "grad_norm": 1.1144641637802124,
      "learning_rate": 0.0003272932994082706,
      "loss": 2.1026,
      "step": 108520
    },
    {
      "epoch": 1.41,
      "grad_norm": 1.148335576057434,
      "learning_rate": 0.00032728922580739647,
      "loss": 2.0696,
      "step": 108521
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.9267200231552124,
      "learning_rate": 0.0003272851522014486,
      "loss": 1.8688,
      "step": 108522
    },
    {
      "epoch": 1.41,
      "grad_norm": 1.0309600830078125,
      "learning_rate": 0.00032728107859042796,
      "loss": 2.1708,
      "step": 108523
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.9002004861831665,
      "learning_rate": 0.0003272770049743353,
      "loss": 2.1012,
      "step": 108524
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.957915186882019,
      "learning_rate": 0.0003272729313531713,
      "loss": 2.3531,
      "step": 108525
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.8246921896934509,
      "learning_rate": 0.0003272688577269367,
      "loss": 2.0069,
      "step": 108526
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.8713402152061462,
      "learning_rate": 0.00032726478409563235,
      "loss": 2.0872,
      "step": 108527
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.9690300822257996,
      "learning_rate": 0.0003272607104592589,
      "loss": 2.1998,
      "step": 108528
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.8483604788780212,
      "learning_rate": 0.0003272566368178173,
      "loss": 2.0032,
      "step": 108529
    },
    {
      "epoch": 1.41,
      "grad_norm": 1.0169543027877808,
      "learning_rate": 0.000327252563171308,
      "loss": 2.0823,
      "step": 108530
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.9393964409828186,
      "learning_rate": 0.00032724848951973213,
      "loss": 2.1614,
      "step": 108531
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.9643117785453796,
      "learning_rate": 0.0003272444158630901,
      "loss": 2.3596,
      "step": 108532
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.8587554693222046,
      "learning_rate": 0.00032724034220138286,
      "loss": 2.0909,
      "step": 108533
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.9467964172363281,
      "learning_rate": 0.0003272362685346112,
      "loss": 2.422,
      "step": 108534
    },
    {
      "epoch": 1.41,
      "grad_norm": 1.062550663948059,
      "learning_rate": 0.0003272321948627757,
      "loss": 2.0445,
      "step": 108535
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.8981165289878845,
      "learning_rate": 0.0003272281211858773,
      "loss": 2.0346,
      "step": 108536
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.9820835590362549,
      "learning_rate": 0.0003272240475039167,
      "loss": 2.12,
      "step": 108537
    },
    {
      "epoch": 1.41,
      "grad_norm": 1.0749852657318115,
      "learning_rate": 0.0003272199738168945,
      "loss": 2.2132,
      "step": 108538
    },
    {
      "epoch": 1.41,
      "grad_norm": 1.080688238143921,
      "learning_rate": 0.0003272159001248116,
      "loss": 2.0936,
      "step": 108539
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.8706082105636597,
      "learning_rate": 0.00032721182642766897,
      "loss": 2.2158,
      "step": 108540
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.9162786602973938,
      "learning_rate": 0.0003272077527254669,
      "loss": 2.2407,
      "step": 108541
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.8878217935562134,
      "learning_rate": 0.00032720367901820655,
      "loss": 2.122,
      "step": 108542
    },
    {
      "epoch": 1.41,
      "grad_norm": 1.1844635009765625,
      "learning_rate": 0.00032719960530588844,
      "loss": 2.4124,
      "step": 108543
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.8572021126747131,
      "learning_rate": 0.0003271955315885134,
      "loss": 2.0659,
      "step": 108544
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.9852377772331238,
      "learning_rate": 0.00032719145786608224,
      "loss": 2.2886,
      "step": 108545
    },
    {
      "epoch": 1.41,
      "grad_norm": 1.0727730989456177,
      "learning_rate": 0.00032718738413859577,
      "loss": 2.1813,
      "step": 108546
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.9622519016265869,
      "learning_rate": 0.00032718331040605447,
      "loss": 2.0131,
      "step": 108547
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.9703963398933411,
      "learning_rate": 0.00032717923666845933,
      "loss": 2.2158,
      "step": 108548
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.8785829544067383,
      "learning_rate": 0.00032717516292581116,
      "loss": 2.1306,
      "step": 108549
    },
    {
      "epoch": 1.41,
      "grad_norm": 1.0072966814041138,
      "learning_rate": 0.00032717108917811055,
      "loss": 2.3027,
      "step": 108550
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.8545089960098267,
      "learning_rate": 0.00032716701542535826,
      "loss": 2.1651,
      "step": 108551
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.8572826981544495,
      "learning_rate": 0.00032716294166755517,
      "loss": 2.1265,
      "step": 108552
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.9164966344833374,
      "learning_rate": 0.000327158867904702,
      "loss": 2.0742,
      "step": 108553
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.9843006134033203,
      "learning_rate": 0.0003271547941367994,
      "loss": 2.1218,
      "step": 108554
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.9984400868415833,
      "learning_rate": 0.0003271507203638483,
      "loss": 2.0631,
      "step": 108555
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.9023975133895874,
      "learning_rate": 0.00032714664658584937,
      "loss": 2.0887,
      "step": 108556
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.9224827289581299,
      "learning_rate": 0.00032714257280280323,
      "loss": 2.0479,
      "step": 108557
    },
    {
      "epoch": 1.41,
      "grad_norm": 1.036680817604065,
      "learning_rate": 0.00032713849901471094,
      "loss": 2.4697,
      "step": 108558
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.9668455719947815,
      "learning_rate": 0.00032713442522157293,
      "loss": 2.172,
      "step": 108559
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.9013110399246216,
      "learning_rate": 0.0003271303514233903,
      "loss": 1.8682,
      "step": 108560
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.9921749234199524,
      "learning_rate": 0.00032712627762016347,
      "loss": 2.1778,
      "step": 108561
    },
    {
      "epoch": 1.41,
      "grad_norm": 1.0704689025878906,
      "learning_rate": 0.0003271222038118934,
      "loss": 2.2565,
      "step": 108562
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.9446659684181213,
      "learning_rate": 0.00032711812999858085,
      "loss": 1.8912,
      "step": 108563
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.9287890791893005,
      "learning_rate": 0.00032711405618022647,
      "loss": 2.2214,
      "step": 108564
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.9695106744766235,
      "learning_rate": 0.00032710998235683115,
      "loss": 2.1161,
      "step": 108565
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.9352185726165771,
      "learning_rate": 0.0003271059085283955,
      "loss": 2.0311,
      "step": 108566
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.9616519808769226,
      "learning_rate": 0.0003271018346949204,
      "loss": 2.0779,
      "step": 108567
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.975280225276947,
      "learning_rate": 0.0003270977608564065,
      "loss": 2.1459,
      "step": 108568
    },
    {
      "epoch": 1.41,
      "grad_norm": 1.108842134475708,
      "learning_rate": 0.0003270936870128546,
      "loss": 2.1448,
      "step": 108569
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.8897761702537537,
      "learning_rate": 0.0003270896131642656,
      "loss": 2.1529,
      "step": 108570
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.8451244831085205,
      "learning_rate": 0.00032708553931064005,
      "loss": 2.2849,
      "step": 108571
    },
    {
      "epoch": 1.41,
      "grad_norm": 1.0207791328430176,
      "learning_rate": 0.0003270814654519788,
      "loss": 2.1968,
      "step": 108572
    },
    {
      "epoch": 1.41,
      "grad_norm": 1.017945408821106,
      "learning_rate": 0.0003270773915882826,
      "loss": 2.2431,
      "step": 108573
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.9541627764701843,
      "learning_rate": 0.00032707331771955216,
      "loss": 2.2687,
      "step": 108574
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.8351380228996277,
      "learning_rate": 0.0003270692438457882,
      "loss": 2.1829,
      "step": 108575
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.8091501593589783,
      "learning_rate": 0.0003270651699669918,
      "loss": 2.2737,
      "step": 108576
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.8871859312057495,
      "learning_rate": 0.0003270610960831632,
      "loss": 2.0735,
      "step": 108577
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.9905198812484741,
      "learning_rate": 0.00032705702219430364,
      "loss": 2.2366,
      "step": 108578
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.8971520662307739,
      "learning_rate": 0.0003270529483004136,
      "loss": 2.1111,
      "step": 108579
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.8830329775810242,
      "learning_rate": 0.00032704887440149384,
      "loss": 2.1979,
      "step": 108580
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.8868189454078674,
      "learning_rate": 0.0003270448004975453,
      "loss": 2.1378,
      "step": 108581
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.8963599801063538,
      "learning_rate": 0.00032704072658856857,
      "loss": 2.0142,
      "step": 108582
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.9231048822402954,
      "learning_rate": 0.0003270366526745644,
      "loss": 2.054,
      "step": 108583
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.9884297251701355,
      "learning_rate": 0.0003270325787555336,
      "loss": 2.1332,
      "step": 108584
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.9506561160087585,
      "learning_rate": 0.00032702850483147706,
      "loss": 2.307,
      "step": 108585
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.8908669948577881,
      "learning_rate": 0.00032702443090239534,
      "loss": 1.9771,
      "step": 108586
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.8704133629798889,
      "learning_rate": 0.00032702035696828926,
      "loss": 2.0486,
      "step": 108587
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.9874587655067444,
      "learning_rate": 0.0003270162830291596,
      "loss": 1.8588,
      "step": 108588
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.9147014617919922,
      "learning_rate": 0.0003270122090850071,
      "loss": 1.9573,
      "step": 108589
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.878753125667572,
      "learning_rate": 0.0003270081351358325,
      "loss": 2.0783,
      "step": 108590
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.8872435688972473,
      "learning_rate": 0.0003270040611816366,
      "loss": 2.1242,
      "step": 108591
    },
    {
      "epoch": 1.41,
      "grad_norm": 1.0093315839767456,
      "learning_rate": 0.00032699998722242015,
      "loss": 2.2017,
      "step": 108592
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.9656901955604553,
      "learning_rate": 0.0003269959132581838,
      "loss": 2.0632,
      "step": 108593
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.8749889135360718,
      "learning_rate": 0.00032699183928892856,
      "loss": 2.104,
      "step": 108594
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.9396313428878784,
      "learning_rate": 0.00032698776531465496,
      "loss": 2.2033,
      "step": 108595
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.8390352725982666,
      "learning_rate": 0.0003269836913353638,
      "loss": 2.3541,
      "step": 108596
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.9348466396331787,
      "learning_rate": 0.0003269796173510558,
      "loss": 2.1869,
      "step": 108597
    },
    {
      "epoch": 1.41,
      "grad_norm": 1.0322511196136475,
      "learning_rate": 0.00032697554336173193,
      "loss": 2.1696,
      "step": 108598
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.9326604008674622,
      "learning_rate": 0.0003269714693673927,
      "loss": 2.1238,
      "step": 108599
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.9164006114006042,
      "learning_rate": 0.00032696739536803893,
      "loss": 2.2707,
      "step": 108600
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.9506893157958984,
      "learning_rate": 0.0003269633213636715,
      "loss": 2.0531,
      "step": 108601
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.8348382711410522,
      "learning_rate": 0.00032695924735429106,
      "loss": 1.9986,
      "step": 108602
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.8598727583885193,
      "learning_rate": 0.00032695517333989835,
      "loss": 2.1193,
      "step": 108603
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.9751155376434326,
      "learning_rate": 0.00032695109932049423,
      "loss": 2.0444,
      "step": 108604
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.9339765906333923,
      "learning_rate": 0.00032694702529607926,
      "loss": 2.0706,
      "step": 108605
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.9829598665237427,
      "learning_rate": 0.00032694295126665447,
      "loss": 2.1577,
      "step": 108606
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.8696407675743103,
      "learning_rate": 0.00032693887723222044,
      "loss": 1.9684,
      "step": 108607
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.8722341060638428,
      "learning_rate": 0.00032693480319277794,
      "loss": 2.1533,
      "step": 108608
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.8798664212226868,
      "learning_rate": 0.0003269307291483277,
      "loss": 2.0179,
      "step": 108609
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.8575729727745056,
      "learning_rate": 0.00032692665509887066,
      "loss": 2.1965,
      "step": 108610
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.8468266725540161,
      "learning_rate": 0.00032692258104440736,
      "loss": 2.008,
      "step": 108611
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.9729311466217041,
      "learning_rate": 0.0003269185069849386,
      "loss": 2.3012,
      "step": 108612
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.7828167080879211,
      "learning_rate": 0.0003269144329204653,
      "loss": 2.0987,
      "step": 108613
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.8604930639266968,
      "learning_rate": 0.000326910358850988,
      "loss": 2.2087,
      "step": 108614
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.9290704131126404,
      "learning_rate": 0.0003269062847765076,
      "loss": 2.2557,
      "step": 108615
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.9644862413406372,
      "learning_rate": 0.00032690221069702483,
      "loss": 2.0698,
      "step": 108616
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.8825365900993347,
      "learning_rate": 0.0003268981366125404,
      "loss": 2.2069,
      "step": 108617
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.9384909868240356,
      "learning_rate": 0.0003268940625230551,
      "loss": 2.1069,
      "step": 108618
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.8568137884140015,
      "learning_rate": 0.00032688998842856975,
      "loss": 2.1794,
      "step": 108619
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.8396363854408264,
      "learning_rate": 0.00032688591432908495,
      "loss": 2.274,
      "step": 108620
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.903821587562561,
      "learning_rate": 0.0003268818402246016,
      "loss": 2.2009,
      "step": 108621
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.844916045665741,
      "learning_rate": 0.0003268777661151204,
      "loss": 1.997,
      "step": 108622
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.8958213925361633,
      "learning_rate": 0.0003268736920006421,
      "loss": 2.0489,
      "step": 108623
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.9622798562049866,
      "learning_rate": 0.00032686961788116743,
      "loss": 2.0891,
      "step": 108624
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.9638642072677612,
      "learning_rate": 0.0003268655437566973,
      "loss": 1.9698,
      "step": 108625
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.9292877912521362,
      "learning_rate": 0.0003268614696272323,
      "loss": 1.9112,
      "step": 108626
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.8697751760482788,
      "learning_rate": 0.00032685739549277325,
      "loss": 2.1059,
      "step": 108627
    },
    {
      "epoch": 1.41,
      "grad_norm": 1.0038846731185913,
      "learning_rate": 0.0003268533213533209,
      "loss": 2.0248,
      "step": 108628
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.814400851726532,
      "learning_rate": 0.000326849247208876,
      "loss": 2.2093,
      "step": 108629
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.860011100769043,
      "learning_rate": 0.00032684517305943937,
      "loss": 2.0962,
      "step": 108630
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.8528251647949219,
      "learning_rate": 0.00032684109890501175,
      "loss": 2.1339,
      "step": 108631
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.9094553589820862,
      "learning_rate": 0.0003268370247455937,
      "loss": 2.2503,
      "step": 108632
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.9452036023139954,
      "learning_rate": 0.00032683295058118624,
      "loss": 2.3454,
      "step": 108633
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.8870192170143127,
      "learning_rate": 0.0003268288764117901,
      "loss": 2.2149,
      "step": 108634
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.936680257320404,
      "learning_rate": 0.0003268248022374058,
      "loss": 2.1055,
      "step": 108635
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.8766093254089355,
      "learning_rate": 0.00032682072805803434,
      "loss": 2.135,
      "step": 108636
    },
    {
      "epoch": 1.41,
      "grad_norm": 1.0035516023635864,
      "learning_rate": 0.00032681665387367646,
      "loss": 2.24,
      "step": 108637
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.8507251143455505,
      "learning_rate": 0.00032681257968433276,
      "loss": 2.156,
      "step": 108638
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.8861856460571289,
      "learning_rate": 0.0003268085054900041,
      "loss": 2.1494,
      "step": 108639
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.8712167739868164,
      "learning_rate": 0.00032680443129069135,
      "loss": 2.2698,
      "step": 108640
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.8255742192268372,
      "learning_rate": 0.0003268003570863951,
      "loss": 2.0762,
      "step": 108641
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.8915194869041443,
      "learning_rate": 0.00032679628287711607,
      "loss": 2.1183,
      "step": 108642
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.8807289004325867,
      "learning_rate": 0.0003267922086628552,
      "loss": 2.2334,
      "step": 108643
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.9010451436042786,
      "learning_rate": 0.0003267881344436131,
      "loss": 2.1329,
      "step": 108644
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.8870994448661804,
      "learning_rate": 0.0003267840602193906,
      "loss": 2.2129,
      "step": 108645
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.8985149264335632,
      "learning_rate": 0.00032677998599018844,
      "loss": 2.3513,
      "step": 108646
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.9189450740814209,
      "learning_rate": 0.00032677591175600735,
      "loss": 2.1482,
      "step": 108647
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.956326961517334,
      "learning_rate": 0.00032677183751684814,
      "loss": 2.0466,
      "step": 108648
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.895599365234375,
      "learning_rate": 0.0003267677632727116,
      "loss": 2.0098,
      "step": 108649
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.9435893297195435,
      "learning_rate": 0.0003267636890235983,
      "loss": 1.9266,
      "step": 108650
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.8439342379570007,
      "learning_rate": 0.0003267596147695092,
      "loss": 2.1126,
      "step": 108651
    },
    {
      "epoch": 1.41,
      "grad_norm": 1.0275450944900513,
      "learning_rate": 0.00032675554051044497,
      "loss": 1.8971,
      "step": 108652
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.9022439122200012,
      "learning_rate": 0.00032675146624640635,
      "loss": 2.1164,
      "step": 108653
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.9500269889831543,
      "learning_rate": 0.0003267473919773941,
      "loss": 2.2148,
      "step": 108654
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.9236766695976257,
      "learning_rate": 0.00032674331770340915,
      "loss": 2.3155,
      "step": 108655
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.8563188910484314,
      "learning_rate": 0.000326739243424452,
      "loss": 1.9867,
      "step": 108656
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.8982744812965393,
      "learning_rate": 0.0003267351691405235,
      "loss": 1.9675,
      "step": 108657
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.8616458177566528,
      "learning_rate": 0.0003267310948516246,
      "loss": 2.1264,
      "step": 108658
    },
    {
      "epoch": 1.41,
      "grad_norm": 1.0340310335159302,
      "learning_rate": 0.0003267270205577557,
      "loss": 2.2521,
      "step": 108659
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.9715821743011475,
      "learning_rate": 0.0003267229462589178,
      "loss": 2.3248,
      "step": 108660
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.9513821601867676,
      "learning_rate": 0.00032671887195511167,
      "loss": 2.2867,
      "step": 108661
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.8678787350654602,
      "learning_rate": 0.00032671479764633784,
      "loss": 2.1589,
      "step": 108662
    },
    {
      "epoch": 1.41,
      "grad_norm": 1.0747545957565308,
      "learning_rate": 0.0003267107233325973,
      "loss": 1.9886,
      "step": 108663
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.8888067603111267,
      "learning_rate": 0.00032670664901389077,
      "loss": 2.4306,
      "step": 108664
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.9416006803512573,
      "learning_rate": 0.000326702574690219,
      "loss": 2.2345,
      "step": 108665
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.9433720111846924,
      "learning_rate": 0.00032669850036158257,
      "loss": 2.089,
      "step": 108666
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.9612030982971191,
      "learning_rate": 0.00032669442602798246,
      "loss": 2.2746,
      "step": 108667
    },
    {
      "epoch": 1.41,
      "grad_norm": 1.0085371732711792,
      "learning_rate": 0.0003266903516894194,
      "loss": 2.2093,
      "step": 108668
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.9576559066772461,
      "learning_rate": 0.0003266862773458941,
      "loss": 2.1131,
      "step": 108669
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.9737867116928101,
      "learning_rate": 0.00032668220299740725,
      "loss": 2.2832,
      "step": 108670
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.8964850902557373,
      "learning_rate": 0.00032667812864395975,
      "loss": 2.1438,
      "step": 108671
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.8583598732948303,
      "learning_rate": 0.00032667405428555223,
      "loss": 2.176,
      "step": 108672
    },
    {
      "epoch": 1.41,
      "grad_norm": 0.913672149181366,
      "learning_rate": 0.00032666997992218547,
      "loss": 2.1662,
      "step": 108673
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.8718640804290771,
      "learning_rate": 0.0003266659055538604,
      "loss": 2.0534,
      "step": 108674
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.8455730676651001,
      "learning_rate": 0.00032666183118057744,
      "loss": 2.2145,
      "step": 108675
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.9666218757629395,
      "learning_rate": 0.0003266577568023377,
      "loss": 2.1646,
      "step": 108676
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.946058452129364,
      "learning_rate": 0.0003266536824191417,
      "loss": 2.1543,
      "step": 108677
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.9679601192474365,
      "learning_rate": 0.00032664960803099034,
      "loss": 2.2635,
      "step": 108678
    },
    {
      "epoch": 1.42,
      "grad_norm": 1.0688773393630981,
      "learning_rate": 0.00032664553363788423,
      "loss": 2.1575,
      "step": 108679
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.9807886481285095,
      "learning_rate": 0.0003266414592398243,
      "loss": 2.1783,
      "step": 108680
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.8253764510154724,
      "learning_rate": 0.0003266373848368111,
      "loss": 2.0574,
      "step": 108681
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.9377798438072205,
      "learning_rate": 0.0003266333104288456,
      "loss": 2.191,
      "step": 108682
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.9000945091247559,
      "learning_rate": 0.0003266292360159285,
      "loss": 2.1898,
      "step": 108683
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.866481602191925,
      "learning_rate": 0.0003266251615980604,
      "loss": 2.2258,
      "step": 108684
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.8576299548149109,
      "learning_rate": 0.00032662108717524226,
      "loss": 2.0769,
      "step": 108685
    },
    {
      "epoch": 1.42,
      "grad_norm": 1.1101776361465454,
      "learning_rate": 0.00032661701274747475,
      "loss": 2.211,
      "step": 108686
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.9796733260154724,
      "learning_rate": 0.0003266129383147586,
      "loss": 2.2637,
      "step": 108687
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.9495770931243896,
      "learning_rate": 0.00032660886387709463,
      "loss": 2.0088,
      "step": 108688
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.9311002492904663,
      "learning_rate": 0.00032660478943448364,
      "loss": 2.3056,
      "step": 108689
    },
    {
      "epoch": 1.42,
      "grad_norm": 1.1063390970230103,
      "learning_rate": 0.00032660071498692624,
      "loss": 2.2054,
      "step": 108690
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.8311771750450134,
      "learning_rate": 0.00032659664053442326,
      "loss": 2.2383,
      "step": 108691
    },
    {
      "epoch": 1.42,
      "grad_norm": 1.2215540409088135,
      "learning_rate": 0.0003265925660769755,
      "loss": 2.1221,
      "step": 108692
    },
    {
      "epoch": 1.42,
      "grad_norm": 1.0845026969909668,
      "learning_rate": 0.0003265884916145836,
      "loss": 2.345,
      "step": 108693
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.9467114210128784,
      "learning_rate": 0.0003265844171472485,
      "loss": 2.24,
      "step": 108694
    },
    {
      "epoch": 1.42,
      "grad_norm": 1.2209128141403198,
      "learning_rate": 0.0003265803426749708,
      "loss": 2.2729,
      "step": 108695
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.9345064759254456,
      "learning_rate": 0.0003265762681977513,
      "loss": 2.2403,
      "step": 108696
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.9184468388557434,
      "learning_rate": 0.00032657219371559073,
      "loss": 2.2652,
      "step": 108697
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.8881169557571411,
      "learning_rate": 0.00032656811922849004,
      "loss": 2.1398,
      "step": 108698
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.8383423686027527,
      "learning_rate": 0.00032656404473644973,
      "loss": 2.2962,
      "step": 108699
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.9730811715126038,
      "learning_rate": 0.00032655997023947066,
      "loss": 2.2025,
      "step": 108700
    },
    {
      "epoch": 1.42,
      "grad_norm": 1.1071803569793701,
      "learning_rate": 0.00032655589573755365,
      "loss": 2.1689,
      "step": 108701
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.9048678278923035,
      "learning_rate": 0.0003265518212306993,
      "loss": 2.1973,
      "step": 108702
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.8815432190895081,
      "learning_rate": 0.00032654774671890855,
      "loss": 1.8816,
      "step": 108703
    },
    {
      "epoch": 1.42,
      "grad_norm": 1.0042588710784912,
      "learning_rate": 0.00032654367220218204,
      "loss": 2.3138,
      "step": 108704
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.900265634059906,
      "learning_rate": 0.00032653959768052057,
      "loss": 2.2163,
      "step": 108705
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.8484264612197876,
      "learning_rate": 0.00032653552315392483,
      "loss": 2.4296,
      "step": 108706
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.9259070754051208,
      "learning_rate": 0.00032653144862239576,
      "loss": 2.3709,
      "step": 108707
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.8988468647003174,
      "learning_rate": 0.00032652737408593385,
      "loss": 2.058,
      "step": 108708
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.85514235496521,
      "learning_rate": 0.0003265232995445401,
      "loss": 2.3802,
      "step": 108709
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.8669207096099854,
      "learning_rate": 0.00032651922499821515,
      "loss": 2.107,
      "step": 108710
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.9113501310348511,
      "learning_rate": 0.0003265151504469598,
      "loss": 2.0918,
      "step": 108711
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.8916858434677124,
      "learning_rate": 0.00032651107589077465,
      "loss": 2.1287,
      "step": 108712
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.8909223079681396,
      "learning_rate": 0.00032650700132966074,
      "loss": 2.1963,
      "step": 108713
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.9054040312767029,
      "learning_rate": 0.00032650292676361865,
      "loss": 2.2241,
      "step": 108714
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.8387059569358826,
      "learning_rate": 0.00032649885219264907,
      "loss": 2.2327,
      "step": 108715
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.9231637120246887,
      "learning_rate": 0.000326494777616753,
      "loss": 2.397,
      "step": 108716
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.9908857941627502,
      "learning_rate": 0.00032649070303593093,
      "loss": 2.1146,
      "step": 108717
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.9194179773330688,
      "learning_rate": 0.0003264866284501838,
      "loss": 2.222,
      "step": 108718
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.8567391037940979,
      "learning_rate": 0.0003264825538595123,
      "loss": 2.2364,
      "step": 108719
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.9593641757965088,
      "learning_rate": 0.0003264784792639172,
      "loss": 2.0119,
      "step": 108720
    },
    {
      "epoch": 1.42,
      "grad_norm": 1.189931869506836,
      "learning_rate": 0.0003264744046633992,
      "loss": 2.0208,
      "step": 108721
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.8699798583984375,
      "learning_rate": 0.00032647033005795914,
      "loss": 2.0473,
      "step": 108722
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.9236763119697571,
      "learning_rate": 0.0003264662554475978,
      "loss": 2.1198,
      "step": 108723
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.8391635417938232,
      "learning_rate": 0.0003264621808323158,
      "loss": 1.8782,
      "step": 108724
    },
    {
      "epoch": 1.42,
      "grad_norm": 1.22511625289917,
      "learning_rate": 0.00032645810621211404,
      "loss": 2.1559,
      "step": 108725
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.9281709790229797,
      "learning_rate": 0.00032645403158699317,
      "loss": 2.1136,
      "step": 108726
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.8809415698051453,
      "learning_rate": 0.000326449956956954,
      "loss": 2.0782,
      "step": 108727
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.9591582417488098,
      "learning_rate": 0.0003264458823219972,
      "loss": 2.1484,
      "step": 108728
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.9980943202972412,
      "learning_rate": 0.0003264418076821238,
      "loss": 2.124,
      "step": 108729
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.8230236768722534,
      "learning_rate": 0.0003264377330373343,
      "loss": 2.1871,
      "step": 108730
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.952445924282074,
      "learning_rate": 0.0003264336583876294,
      "loss": 2.3089,
      "step": 108731
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.8795400261878967,
      "learning_rate": 0.0003264295837330101,
      "loss": 2.1874,
      "step": 108732
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.9157143235206604,
      "learning_rate": 0.0003264255090734771,
      "loss": 2.1802,
      "step": 108733
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.9115877151489258,
      "learning_rate": 0.000326421434409031,
      "loss": 2.179,
      "step": 108734
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.8023250102996826,
      "learning_rate": 0.0003264173597396727,
      "loss": 2.0426,
      "step": 108735
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.8938873410224915,
      "learning_rate": 0.0003264132850654029,
      "loss": 2.1733,
      "step": 108736
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.9338772892951965,
      "learning_rate": 0.0003264092103862224,
      "loss": 1.949,
      "step": 108737
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.9505406618118286,
      "learning_rate": 0.0003264051357021318,
      "loss": 2.2809,
      "step": 108738
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.9380146265029907,
      "learning_rate": 0.0003264010610131321,
      "loss": 2.269,
      "step": 108739
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.8853442668914795,
      "learning_rate": 0.000326396986319224,
      "loss": 2.1859,
      "step": 108740
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.845233142375946,
      "learning_rate": 0.0003263929116204081,
      "loss": 2.2634,
      "step": 108741
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.9603481292724609,
      "learning_rate": 0.00032638883691668525,
      "loss": 1.9831,
      "step": 108742
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.8782588839530945,
      "learning_rate": 0.00032638476220805637,
      "loss": 2.1011,
      "step": 108743
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.8076871037483215,
      "learning_rate": 0.0003263806874945219,
      "loss": 1.8699,
      "step": 108744
    },
    {
      "epoch": 1.42,
      "grad_norm": 1.228010892868042,
      "learning_rate": 0.00032637661277608277,
      "loss": 2.1288,
      "step": 108745
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.8925217986106873,
      "learning_rate": 0.0003263725380527398,
      "loss": 2.1528,
      "step": 108746
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.9290056824684143,
      "learning_rate": 0.0003263684633244937,
      "loss": 2.4626,
      "step": 108747
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.8504126667976379,
      "learning_rate": 0.00032636438859134513,
      "loss": 2.1298,
      "step": 108748
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.8637979030609131,
      "learning_rate": 0.00032636031385329496,
      "loss": 2.0231,
      "step": 108749
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.8726223707199097,
      "learning_rate": 0.0003263562391103439,
      "loss": 2.1893,
      "step": 108750
    },
    {
      "epoch": 1.42,
      "grad_norm": 1.0149908065795898,
      "learning_rate": 0.0003263521643624927,
      "loss": 2.2257,
      "step": 108751
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.9037815928459167,
      "learning_rate": 0.00032634808960974217,
      "loss": 2.0326,
      "step": 108752
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.8635126948356628,
      "learning_rate": 0.00032634401485209303,
      "loss": 2.3483,
      "step": 108753
    },
    {
      "epoch": 1.42,
      "grad_norm": 1.0216559171676636,
      "learning_rate": 0.000326339940089546,
      "loss": 1.8753,
      "step": 108754
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.9552944302558899,
      "learning_rate": 0.0003263358653221019,
      "loss": 2.1181,
      "step": 108755
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.868010938167572,
      "learning_rate": 0.0003263317905497615,
      "loss": 2.2779,
      "step": 108756
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.972293496131897,
      "learning_rate": 0.0003263277157725255,
      "loss": 2.0835,
      "step": 108757
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.979016900062561,
      "learning_rate": 0.00032632364099039466,
      "loss": 2.0901,
      "step": 108758
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.8312845826148987,
      "learning_rate": 0.00032631956620336977,
      "loss": 2.2142,
      "step": 108759
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.9337733387947083,
      "learning_rate": 0.00032631549141145166,
      "loss": 2.1704,
      "step": 108760
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.9214165806770325,
      "learning_rate": 0.00032631141661464087,
      "loss": 2.089,
      "step": 108761
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.8558635711669922,
      "learning_rate": 0.0003263073418129383,
      "loss": 2.1858,
      "step": 108762
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.8104104399681091,
      "learning_rate": 0.0003263032670063448,
      "loss": 2.1021,
      "step": 108763
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.9462884664535522,
      "learning_rate": 0.00032629919219486094,
      "loss": 2.4467,
      "step": 108764
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.9453328847885132,
      "learning_rate": 0.0003262951173784876,
      "loss": 2.3854,
      "step": 108765
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.916291356086731,
      "learning_rate": 0.00032629104255722547,
      "loss": 2.1787,
      "step": 108766
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.8369653224945068,
      "learning_rate": 0.0003262869677310754,
      "loss": 2.2256,
      "step": 108767
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.8479477167129517,
      "learning_rate": 0.00032628289290003804,
      "loss": 2.0919,
      "step": 108768
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.9182388186454773,
      "learning_rate": 0.0003262788180641142,
      "loss": 2.2215,
      "step": 108769
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.9110948443412781,
      "learning_rate": 0.00032627474322330466,
      "loss": 2.1235,
      "step": 108770
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.9413723945617676,
      "learning_rate": 0.0003262706683776101,
      "loss": 2.1873,
      "step": 108771
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.9923654198646545,
      "learning_rate": 0.00032626659352703137,
      "loss": 2.2946,
      "step": 108772
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.8717057108879089,
      "learning_rate": 0.0003262625186715692,
      "loss": 2.2188,
      "step": 108773
    },
    {
      "epoch": 1.42,
      "grad_norm": 1.0213409662246704,
      "learning_rate": 0.0003262584438112243,
      "loss": 1.8851,
      "step": 108774
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.9252035617828369,
      "learning_rate": 0.00032625436894599745,
      "loss": 2.1572,
      "step": 108775
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.9711927175521851,
      "learning_rate": 0.0003262502940758894,
      "loss": 2.0424,
      "step": 108776
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.9829970598220825,
      "learning_rate": 0.000326246219200901,
      "loss": 2.1826,
      "step": 108777
    },
    {
      "epoch": 1.42,
      "grad_norm": 1.0334464311599731,
      "learning_rate": 0.0003262421443210328,
      "loss": 2.0536,
      "step": 108778
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.9402886033058167,
      "learning_rate": 0.0003262380694362858,
      "loss": 2.0633,
      "step": 108779
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.9930317997932434,
      "learning_rate": 0.0003262339945466607,
      "loss": 2.0817,
      "step": 108780
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.9459593892097473,
      "learning_rate": 0.0003262299196521581,
      "loss": 2.2764,
      "step": 108781
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.861984133720398,
      "learning_rate": 0.0003262258447527788,
      "loss": 2.223,
      "step": 108782
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.9247944355010986,
      "learning_rate": 0.0003262217698485238,
      "loss": 2.412,
      "step": 108783
    },
    {
      "epoch": 1.42,
      "grad_norm": 1.0167367458343506,
      "learning_rate": 0.0003262176949393936,
      "loss": 2.1134,
      "step": 108784
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.9830087423324585,
      "learning_rate": 0.00032621362002538894,
      "loss": 2.1367,
      "step": 108785
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.9087136387825012,
      "learning_rate": 0.0003262095451065108,
      "loss": 2.2465,
      "step": 108786
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.9396078586578369,
      "learning_rate": 0.0003262054701827598,
      "loss": 2.271,
      "step": 108787
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.8763730525970459,
      "learning_rate": 0.00032620139525413663,
      "loss": 2.019,
      "step": 108788
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.8666651844978333,
      "learning_rate": 0.0003261973203206422,
      "loss": 2.1762,
      "step": 108789
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.886623203754425,
      "learning_rate": 0.0003261932453822771,
      "loss": 2.0223,
      "step": 108790
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.9293209314346313,
      "learning_rate": 0.0003261891704390423,
      "loss": 2.0941,
      "step": 108791
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.8718895316123962,
      "learning_rate": 0.0003261850954909384,
      "loss": 2.0541,
      "step": 108792
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.9864819049835205,
      "learning_rate": 0.00032618102053796617,
      "loss": 2.0886,
      "step": 108793
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.891811728477478,
      "learning_rate": 0.00032617694558012637,
      "loss": 2.2194,
      "step": 108794
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.9533876776695251,
      "learning_rate": 0.00032617287061741985,
      "loss": 2.1424,
      "step": 108795
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.9038378596305847,
      "learning_rate": 0.0003261687956498473,
      "loss": 1.9597,
      "step": 108796
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.8418002128601074,
      "learning_rate": 0.0003261647206774094,
      "loss": 1.9762,
      "step": 108797
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.8911892175674438,
      "learning_rate": 0.000326160645700107,
      "loss": 2.1295,
      "step": 108798
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.9291282892227173,
      "learning_rate": 0.0003261565707179409,
      "loss": 2.1682,
      "step": 108799
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.8366933465003967,
      "learning_rate": 0.00032615249573091174,
      "loss": 2.2229,
      "step": 108800
    },
    {
      "epoch": 1.42,
      "grad_norm": 1.0210809707641602,
      "learning_rate": 0.00032614842073902035,
      "loss": 2.3117,
      "step": 108801
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.9821939468383789,
      "learning_rate": 0.00032614434574226755,
      "loss": 2.292,
      "step": 108802
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.943419337272644,
      "learning_rate": 0.0003261402707406539,
      "loss": 2.187,
      "step": 108803
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.9091391563415527,
      "learning_rate": 0.0003261361957341803,
      "loss": 2.1786,
      "step": 108804
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.8876587748527527,
      "learning_rate": 0.0003261321207228476,
      "loss": 2.1447,
      "step": 108805
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.8360325694084167,
      "learning_rate": 0.0003261280457066564,
      "loss": 2.1406,
      "step": 108806
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.9039812088012695,
      "learning_rate": 0.00032612397068560746,
      "loss": 2.3863,
      "step": 108807
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.8322685360908508,
      "learning_rate": 0.0003261198956597016,
      "loss": 2.3323,
      "step": 108808
    },
    {
      "epoch": 1.42,
      "grad_norm": 1.0444175004959106,
      "learning_rate": 0.0003261158206289396,
      "loss": 2.0659,
      "step": 108809
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.8478409051895142,
      "learning_rate": 0.00032611174559332213,
      "loss": 2.0026,
      "step": 108810
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.8856072425842285,
      "learning_rate": 0.00032610767055284997,
      "loss": 2.5017,
      "step": 108811
    },
    {
      "epoch": 1.42,
      "grad_norm": 1.030409812927246,
      "learning_rate": 0.00032610359550752394,
      "loss": 2.1541,
      "step": 108812
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.9038981199264526,
      "learning_rate": 0.0003260995204573447,
      "loss": 2.2914,
      "step": 108813
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.8930478096008301,
      "learning_rate": 0.0003260954454023132,
      "loss": 2.1926,
      "step": 108814
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.891126275062561,
      "learning_rate": 0.00032609137034242993,
      "loss": 2.1009,
      "step": 108815
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.8824462890625,
      "learning_rate": 0.0003260872952776958,
      "loss": 2.108,
      "step": 108816
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.9299062490463257,
      "learning_rate": 0.0003260832202081117,
      "loss": 2.1933,
      "step": 108817
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.8848918676376343,
      "learning_rate": 0.00032607914513367803,
      "loss": 2.2824,
      "step": 108818
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.9545299410820007,
      "learning_rate": 0.0003260750700543959,
      "loss": 2.0557,
      "step": 108819
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.8723298907279968,
      "learning_rate": 0.0003260709949702658,
      "loss": 2.2156,
      "step": 108820
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.8299873471260071,
      "learning_rate": 0.0003260669198812887,
      "loss": 2.2794,
      "step": 108821
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.851016640663147,
      "learning_rate": 0.00032606284478746525,
      "loss": 1.9903,
      "step": 108822
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.99775230884552,
      "learning_rate": 0.0003260587696887962,
      "loss": 2.2458,
      "step": 108823
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.8757480978965759,
      "learning_rate": 0.00032605469458528235,
      "loss": 2.1383,
      "step": 108824
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.9252134561538696,
      "learning_rate": 0.00032605061947692444,
      "loss": 2.1044,
      "step": 108825
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.8487370610237122,
      "learning_rate": 0.00032604654436372324,
      "loss": 2.1287,
      "step": 108826
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.8957195281982422,
      "learning_rate": 0.00032604246924567944,
      "loss": 2.1955,
      "step": 108827
    },
    {
      "epoch": 1.42,
      "grad_norm": 1.0632209777832031,
      "learning_rate": 0.000326038394122794,
      "loss": 2.0367,
      "step": 108828
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.8751576542854309,
      "learning_rate": 0.00032603431899506734,
      "loss": 1.9816,
      "step": 108829
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.8931722044944763,
      "learning_rate": 0.0003260302438625005,
      "loss": 1.9512,
      "step": 108830
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.9594506621360779,
      "learning_rate": 0.0003260261687250941,
      "loss": 2.4072,
      "step": 108831
    },
    {
      "epoch": 1.42,
      "grad_norm": 1.0090093612670898,
      "learning_rate": 0.0003260220935828491,
      "loss": 2.0679,
      "step": 108832
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.861819863319397,
      "learning_rate": 0.00032601801843576596,
      "loss": 2.186,
      "step": 108833
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.88001549243927,
      "learning_rate": 0.00032601394328384556,
      "loss": 2.2003,
      "step": 108834
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.8729088306427002,
      "learning_rate": 0.00032600986812708876,
      "loss": 2.1531,
      "step": 108835
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.8549567461013794,
      "learning_rate": 0.0003260057929654962,
      "loss": 1.9252,
      "step": 108836
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.836661696434021,
      "learning_rate": 0.00032600171779906864,
      "loss": 1.9604,
      "step": 108837
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.9631165266036987,
      "learning_rate": 0.00032599764262780694,
      "loss": 2.1042,
      "step": 108838
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.9092703461647034,
      "learning_rate": 0.00032599356745171177,
      "loss": 2.1259,
      "step": 108839
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.9364928007125854,
      "learning_rate": 0.0003259894922707838,
      "loss": 2.2245,
      "step": 108840
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.9242961406707764,
      "learning_rate": 0.00032598541708502406,
      "loss": 2.1773,
      "step": 108841
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.9172163009643555,
      "learning_rate": 0.000325981341894433,
      "loss": 1.9127,
      "step": 108842
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.8901487588882446,
      "learning_rate": 0.0003259772666990116,
      "loss": 2.1463,
      "step": 108843
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.9836193919181824,
      "learning_rate": 0.0003259731914987605,
      "loss": 2.1654,
      "step": 108844
    },
    {
      "epoch": 1.42,
      "grad_norm": 1.1155977249145508,
      "learning_rate": 0.00032596911629368053,
      "loss": 2.1408,
      "step": 108845
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.901425838470459,
      "learning_rate": 0.00032596504108377234,
      "loss": 2.2483,
      "step": 108846
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.9173000454902649,
      "learning_rate": 0.00032596096586903685,
      "loss": 2.2537,
      "step": 108847
    },
    {
      "epoch": 1.42,
      "grad_norm": 1.0091538429260254,
      "learning_rate": 0.00032595689064947466,
      "loss": 2.2373,
      "step": 108848
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.9201929569244385,
      "learning_rate": 0.0003259528154250866,
      "loss": 2.3097,
      "step": 108849
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.8741560578346252,
      "learning_rate": 0.0003259487401958735,
      "loss": 2.3531,
      "step": 108850
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.8568539023399353,
      "learning_rate": 0.0003259446649618359,
      "loss": 2.3895,
      "step": 108851
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.9311424493789673,
      "learning_rate": 0.0003259405897229748,
      "loss": 2.2136,
      "step": 108852
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.9780515432357788,
      "learning_rate": 0.00032593651447929085,
      "loss": 2.1505,
      "step": 108853
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.9203444123268127,
      "learning_rate": 0.0003259324392307848,
      "loss": 2.2008,
      "step": 108854
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.9653360247612,
      "learning_rate": 0.0003259283639774574,
      "loss": 2.2452,
      "step": 108855
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.8814720511436462,
      "learning_rate": 0.0003259242887193095,
      "loss": 2.1021,
      "step": 108856
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.8503081202507019,
      "learning_rate": 0.00032592021345634163,
      "loss": 2.3323,
      "step": 108857
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.9247824549674988,
      "learning_rate": 0.0003259161381885548,
      "loss": 2.1063,
      "step": 108858
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.8738468885421753,
      "learning_rate": 0.0003259120629159497,
      "loss": 2.1514,
      "step": 108859
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.9136305451393127,
      "learning_rate": 0.00032590798763852707,
      "loss": 2.3551,
      "step": 108860
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.9524834752082825,
      "learning_rate": 0.0003259039123562876,
      "loss": 2.1322,
      "step": 108861
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.9062405824661255,
      "learning_rate": 0.0003258998370692321,
      "loss": 1.9707,
      "step": 108862
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.9800912737846375,
      "learning_rate": 0.00032589576177736133,
      "loss": 2.0759,
      "step": 108863
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.9606562852859497,
      "learning_rate": 0.0003258916864806761,
      "loss": 2.2168,
      "step": 108864
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.8812320232391357,
      "learning_rate": 0.00032588761117917706,
      "loss": 2.117,
      "step": 108865
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.8341593742370605,
      "learning_rate": 0.0003258835358728651,
      "loss": 2.1617,
      "step": 108866
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.9305365681648254,
      "learning_rate": 0.00032587946056174076,
      "loss": 2.277,
      "step": 108867
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.9013237953186035,
      "learning_rate": 0.00032587538524580503,
      "loss": 1.952,
      "step": 108868
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.9973753094673157,
      "learning_rate": 0.00032587130992505864,
      "loss": 2.289,
      "step": 108869
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.9252580404281616,
      "learning_rate": 0.0003258672345995022,
      "loss": 2.3379,
      "step": 108870
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.9315178394317627,
      "learning_rate": 0.00032586315926913653,
      "loss": 2.0475,
      "step": 108871
    },
    {
      "epoch": 1.42,
      "grad_norm": 1.0970453023910522,
      "learning_rate": 0.0003258590839339625,
      "loss": 2.1564,
      "step": 108872
    },
    {
      "epoch": 1.42,
      "grad_norm": 1.0293419361114502,
      "learning_rate": 0.00032585500859398073,
      "loss": 2.1082,
      "step": 108873
    },
    {
      "epoch": 1.42,
      "grad_norm": 1.0807924270629883,
      "learning_rate": 0.000325850933249192,
      "loss": 2.06,
      "step": 108874
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.9416017532348633,
      "learning_rate": 0.0003258468578995972,
      "loss": 2.2632,
      "step": 108875
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.9296957850456238,
      "learning_rate": 0.00032584278254519684,
      "loss": 2.2634,
      "step": 108876
    },
    {
      "epoch": 1.42,
      "grad_norm": 1.088153600692749,
      "learning_rate": 0.0003258387071859919,
      "loss": 2.2033,
      "step": 108877
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.950375497341156,
      "learning_rate": 0.0003258346318219831,
      "loss": 2.3537,
      "step": 108878
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.8894863724708557,
      "learning_rate": 0.0003258305564531711,
      "loss": 2.2857,
      "step": 108879
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.8777981996536255,
      "learning_rate": 0.0003258264810795566,
      "loss": 1.9699,
      "step": 108880
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.927782416343689,
      "learning_rate": 0.0003258224057011406,
      "loss": 2.0047,
      "step": 108881
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.9704490303993225,
      "learning_rate": 0.00032581833031792375,
      "loss": 2.2968,
      "step": 108882
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.9251960515975952,
      "learning_rate": 0.0003258142549299067,
      "loss": 2.1735,
      "step": 108883
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.915416419506073,
      "learning_rate": 0.0003258101795370904,
      "loss": 2.1837,
      "step": 108884
    },
    {
      "epoch": 1.42,
      "grad_norm": 1.0082381963729858,
      "learning_rate": 0.0003258061041394754,
      "loss": 2.0998,
      "step": 108885
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.8448992967605591,
      "learning_rate": 0.0003258020287370625,
      "loss": 2.1185,
      "step": 108886
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.9665535688400269,
      "learning_rate": 0.0003257979533298527,
      "loss": 2.1232,
      "step": 108887
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.945190966129303,
      "learning_rate": 0.00032579387791784644,
      "loss": 2.1015,
      "step": 108888
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.9339834451675415,
      "learning_rate": 0.00032578980250104467,
      "loss": 2.2265,
      "step": 108889
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.862704336643219,
      "learning_rate": 0.00032578572707944807,
      "loss": 2.0318,
      "step": 108890
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.9361692667007446,
      "learning_rate": 0.00032578165165305745,
      "loss": 2.3118,
      "step": 108891
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.8585200309753418,
      "learning_rate": 0.00032577757622187346,
      "loss": 2.2568,
      "step": 108892
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.938134491443634,
      "learning_rate": 0.00032577350078589703,
      "loss": 2.2879,
      "step": 108893
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.9843297600746155,
      "learning_rate": 0.0003257694253451287,
      "loss": 2.1884,
      "step": 108894
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.8705231547355652,
      "learning_rate": 0.00032576534989956944,
      "loss": 2.153,
      "step": 108895
    },
    {
      "epoch": 1.42,
      "grad_norm": 1.0842695236206055,
      "learning_rate": 0.0003257612744492199,
      "loss": 2.1441,
      "step": 108896
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.8774034380912781,
      "learning_rate": 0.0003257571989940808,
      "loss": 2.0916,
      "step": 108897
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.9131739139556885,
      "learning_rate": 0.000325753123534153,
      "loss": 2.2208,
      "step": 108898
    },
    {
      "epoch": 1.42,
      "grad_norm": 1.0085439682006836,
      "learning_rate": 0.0003257490480694372,
      "loss": 2.0937,
      "step": 108899
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.9130509495735168,
      "learning_rate": 0.0003257449725999342,
      "loss": 2.0539,
      "step": 108900
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.96485835313797,
      "learning_rate": 0.00032574089712564464,
      "loss": 2.1275,
      "step": 108901
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.8561186194419861,
      "learning_rate": 0.00032573682164656944,
      "loss": 1.9858,
      "step": 108902
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.8838199377059937,
      "learning_rate": 0.0003257327461627092,
      "loss": 2.0257,
      "step": 108903
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.8710765838623047,
      "learning_rate": 0.0003257286706740648,
      "loss": 2.1225,
      "step": 108904
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.8702523708343506,
      "learning_rate": 0.0003257245951806369,
      "loss": 2.3912,
      "step": 108905
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.8115681409835815,
      "learning_rate": 0.0003257205196824264,
      "loss": 2.0546,
      "step": 108906
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.8683674335479736,
      "learning_rate": 0.00032571644417943394,
      "loss": 2.2407,
      "step": 108907
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.8915483355522156,
      "learning_rate": 0.0003257123686716603,
      "loss": 2.1024,
      "step": 108908
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.9677603840827942,
      "learning_rate": 0.00032570829315910633,
      "loss": 1.9787,
      "step": 108909
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.8838797807693481,
      "learning_rate": 0.00032570421764177256,
      "loss": 2.1682,
      "step": 108910
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.9093384146690369,
      "learning_rate": 0.00032570014211965996,
      "loss": 2.1375,
      "step": 108911
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.913784384727478,
      "learning_rate": 0.0003256960665927692,
      "loss": 2.2678,
      "step": 108912
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.8850146532058716,
      "learning_rate": 0.00032569199106110105,
      "loss": 2.186,
      "step": 108913
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.8680799007415771,
      "learning_rate": 0.0003256879155246563,
      "loss": 2.1243,
      "step": 108914
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.8705061078071594,
      "learning_rate": 0.0003256838399834357,
      "loss": 2.2602,
      "step": 108915
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.8782206177711487,
      "learning_rate": 0.00032567976443743994,
      "loss": 2.3681,
      "step": 108916
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.8839313387870789,
      "learning_rate": 0.00032567568888666986,
      "loss": 2.1134,
      "step": 108917
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.8497122526168823,
      "learning_rate": 0.00032567161333112615,
      "loss": 1.9301,
      "step": 108918
    },
    {
      "epoch": 1.42,
      "grad_norm": 1.016649603843689,
      "learning_rate": 0.0003256675377708096,
      "loss": 1.8906,
      "step": 108919
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.8923486471176147,
      "learning_rate": 0.00032566346220572103,
      "loss": 2.0188,
      "step": 108920
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.9456464052200317,
      "learning_rate": 0.00032565938663586114,
      "loss": 2.1093,
      "step": 108921
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.907659649848938,
      "learning_rate": 0.0003256553110612306,
      "loss": 2.0847,
      "step": 108922
    },
    {
      "epoch": 1.42,
      "grad_norm": 1.0246604681015015,
      "learning_rate": 0.00032565123548183037,
      "loss": 2.0256,
      "step": 108923
    },
    {
      "epoch": 1.42,
      "grad_norm": 1.0103728771209717,
      "learning_rate": 0.000325647159897661,
      "loss": 2.1065,
      "step": 108924
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.902463972568512,
      "learning_rate": 0.0003256430843087234,
      "loss": 2.3298,
      "step": 108925
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.8872471451759338,
      "learning_rate": 0.00032563900871501825,
      "loss": 2.037,
      "step": 108926
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.8981527090072632,
      "learning_rate": 0.00032563493311654626,
      "loss": 1.8648,
      "step": 108927
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.8270477652549744,
      "learning_rate": 0.0003256308575133083,
      "loss": 2.0333,
      "step": 108928
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.9831875562667847,
      "learning_rate": 0.00032562678190530507,
      "loss": 2.1295,
      "step": 108929
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.990121603012085,
      "learning_rate": 0.00032562270629253733,
      "loss": 2.1527,
      "step": 108930
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.8376742005348206,
      "learning_rate": 0.0003256186306750059,
      "loss": 2.2738,
      "step": 108931
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.9233556985855103,
      "learning_rate": 0.00032561455505271137,
      "loss": 2.188,
      "step": 108932
    },
    {
      "epoch": 1.42,
      "grad_norm": 1.0475218296051025,
      "learning_rate": 0.00032561047942565477,
      "loss": 2.1369,
      "step": 108933
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.8744919300079346,
      "learning_rate": 0.00032560640379383657,
      "loss": 2.1368,
      "step": 108934
    },
    {
      "epoch": 1.42,
      "grad_norm": 1.0551012754440308,
      "learning_rate": 0.00032560232815725764,
      "loss": 1.9565,
      "step": 108935
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.8340764045715332,
      "learning_rate": 0.0003255982525159189,
      "loss": 2.085,
      "step": 108936
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.8829272985458374,
      "learning_rate": 0.00032559417686982083,
      "loss": 2.1571,
      "step": 108937
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.8755813241004944,
      "learning_rate": 0.00032559010121896437,
      "loss": 2.3366,
      "step": 108938
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.9151419997215271,
      "learning_rate": 0.0003255860255633503,
      "loss": 2.334,
      "step": 108939
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.8658672571182251,
      "learning_rate": 0.00032558194990297916,
      "loss": 2.1166,
      "step": 108940
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.8968296647071838,
      "learning_rate": 0.0003255778742378519,
      "loss": 2.0523,
      "step": 108941
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.9926050901412964,
      "learning_rate": 0.0003255737985679693,
      "loss": 2.2048,
      "step": 108942
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.8453152775764465,
      "learning_rate": 0.000325569722893332,
      "loss": 2.1412,
      "step": 108943
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.9127559661865234,
      "learning_rate": 0.00032556564721394074,
      "loss": 2.1445,
      "step": 108944
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.9666513800621033,
      "learning_rate": 0.0003255615715297965,
      "loss": 2.2045,
      "step": 108945
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.9454749226570129,
      "learning_rate": 0.0003255574958408997,
      "loss": 2.435,
      "step": 108946
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.8211957812309265,
      "learning_rate": 0.0003255534201472514,
      "loss": 2.0551,
      "step": 108947
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.9076093435287476,
      "learning_rate": 0.00032554934444885226,
      "loss": 2.0627,
      "step": 108948
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.9765791893005371,
      "learning_rate": 0.00032554526874570294,
      "loss": 2.168,
      "step": 108949
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.8220617771148682,
      "learning_rate": 0.0003255411930378042,
      "loss": 2.3071,
      "step": 108950
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.8990205526351929,
      "learning_rate": 0.00032553711732515706,
      "loss": 2.2152,
      "step": 108951
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.9465317726135254,
      "learning_rate": 0.00032553304160776197,
      "loss": 2.2288,
      "step": 108952
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.9014778137207031,
      "learning_rate": 0.0003255289658856198,
      "loss": 2.0497,
      "step": 108953
    },
    {
      "epoch": 1.42,
      "grad_norm": 1.0190086364746094,
      "learning_rate": 0.0003255248901587314,
      "loss": 2.2274,
      "step": 108954
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.9001133441925049,
      "learning_rate": 0.0003255208144270973,
      "loss": 2.1262,
      "step": 108955
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.8962533473968506,
      "learning_rate": 0.00032551673869071854,
      "loss": 2.3157,
      "step": 108956
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.8743425607681274,
      "learning_rate": 0.00032551266294959563,
      "loss": 1.8917,
      "step": 108957
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.9273393750190735,
      "learning_rate": 0.0003255085872037295,
      "loss": 2.0917,
      "step": 108958
    },
    {
      "epoch": 1.42,
      "grad_norm": 1.0376750230789185,
      "learning_rate": 0.00032550451145312085,
      "loss": 1.8987,
      "step": 108959
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.9669119119644165,
      "learning_rate": 0.00032550043569777045,
      "loss": 2.0926,
      "step": 108960
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.8493865728378296,
      "learning_rate": 0.00032549635993767893,
      "loss": 2.354,
      "step": 108961
    },
    {
      "epoch": 1.42,
      "grad_norm": 1.0389728546142578,
      "learning_rate": 0.0003254922841728472,
      "loss": 2.1102,
      "step": 108962
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.8556060194969177,
      "learning_rate": 0.00032548820840327605,
      "loss": 2.051,
      "step": 108963
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.8648684024810791,
      "learning_rate": 0.00032548413262896604,
      "loss": 2.0112,
      "step": 108964
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.8727983236312866,
      "learning_rate": 0.00032548005684991806,
      "loss": 2.1913,
      "step": 108965
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.8451173901557922,
      "learning_rate": 0.000325475981066133,
      "loss": 2.1105,
      "step": 108966
    },
    {
      "epoch": 1.42,
      "grad_norm": 1.1798969507217407,
      "learning_rate": 0.0003254719052776113,
      "loss": 2.388,
      "step": 108967
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.9144905209541321,
      "learning_rate": 0.000325467829484354,
      "loss": 1.9717,
      "step": 108968
    },
    {
      "epoch": 1.42,
      "grad_norm": 1.0426017045974731,
      "learning_rate": 0.00032546375368636176,
      "loss": 2.382,
      "step": 108969
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.872218668460846,
      "learning_rate": 0.00032545967788363524,
      "loss": 2.183,
      "step": 108970
    },
    {
      "epoch": 1.42,
      "grad_norm": 1.0137649774551392,
      "learning_rate": 0.0003254556020761753,
      "loss": 1.9613,
      "step": 108971
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.8953109979629517,
      "learning_rate": 0.0003254515262639828,
      "loss": 2.1177,
      "step": 108972
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.9462977647781372,
      "learning_rate": 0.00032544745044705826,
      "loss": 2.3238,
      "step": 108973
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.9316161274909973,
      "learning_rate": 0.00032544337462540255,
      "loss": 2.2704,
      "step": 108974
    },
    {
      "epoch": 1.42,
      "grad_norm": 1.0011993646621704,
      "learning_rate": 0.0003254392987990166,
      "loss": 2.0821,
      "step": 108975
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.8634130358695984,
      "learning_rate": 0.0003254352229679008,
      "loss": 2.2773,
      "step": 108976
    },
    {
      "epoch": 1.42,
      "grad_norm": 1.0297679901123047,
      "learning_rate": 0.00032543114713205615,
      "loss": 2.1573,
      "step": 108977
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.8294282555580139,
      "learning_rate": 0.00032542707129148355,
      "loss": 1.915,
      "step": 108978
    },
    {
      "epoch": 1.42,
      "grad_norm": 1.0351996421813965,
      "learning_rate": 0.0003254229954461834,
      "loss": 2.31,
      "step": 108979
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.8430230617523193,
      "learning_rate": 0.00032541891959615666,
      "loss": 2.1275,
      "step": 108980
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.9451324939727783,
      "learning_rate": 0.00032541484374140416,
      "loss": 2.1452,
      "step": 108981
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.9770740866661072,
      "learning_rate": 0.00032541076788192644,
      "loss": 2.0538,
      "step": 108982
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.996110200881958,
      "learning_rate": 0.0003254066920177244,
      "loss": 2.0438,
      "step": 108983
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.9184967875480652,
      "learning_rate": 0.00032540261614879884,
      "loss": 2.0661,
      "step": 108984
    },
    {
      "epoch": 1.42,
      "grad_norm": 1.0199658870697021,
      "learning_rate": 0.0003253985402751504,
      "loss": 1.9858,
      "step": 108985
    },
    {
      "epoch": 1.42,
      "grad_norm": 1.3320776224136353,
      "learning_rate": 0.00032539446439677983,
      "loss": 2.2433,
      "step": 108986
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.8357792496681213,
      "learning_rate": 0.0003253903885136881,
      "loss": 2.2625,
      "step": 108987
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.9743344783782959,
      "learning_rate": 0.00032538631262587575,
      "loss": 2.1135,
      "step": 108988
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.8586598634719849,
      "learning_rate": 0.0003253822367333435,
      "loss": 2.1442,
      "step": 108989
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.97232985496521,
      "learning_rate": 0.0003253781608360924,
      "loss": 2.136,
      "step": 108990
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.9079285264015198,
      "learning_rate": 0.00032537408493412296,
      "loss": 1.9798,
      "step": 108991
    },
    {
      "epoch": 1.42,
      "grad_norm": 1.0121564865112305,
      "learning_rate": 0.00032537000902743593,
      "loss": 2.5571,
      "step": 108992
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.908168613910675,
      "learning_rate": 0.00032536593311603213,
      "loss": 2.0889,
      "step": 108993
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.8875241279602051,
      "learning_rate": 0.00032536185719991243,
      "loss": 2.1985,
      "step": 108994
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.8841003775596619,
      "learning_rate": 0.00032535778127907737,
      "loss": 2.011,
      "step": 108995
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.8476383686065674,
      "learning_rate": 0.0003253537053535278,
      "loss": 2.1575,
      "step": 108996
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.9768639206886292,
      "learning_rate": 0.00032534962942326465,
      "loss": 2.1444,
      "step": 108997
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.9914592504501343,
      "learning_rate": 0.00032534555348828846,
      "loss": 2.216,
      "step": 108998
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.8415462970733643,
      "learning_rate": 0.0003253414775486,
      "loss": 1.8757,
      "step": 108999
    },
    {
      "epoch": 1.42,
      "grad_norm": 1.0770164728164673,
      "learning_rate": 0.0003253374016042001,
      "loss": 2.1188,
      "step": 109000
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.8753491640090942,
      "learning_rate": 0.00032533332565508956,
      "loss": 2.2765,
      "step": 109001
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.9998079538345337,
      "learning_rate": 0.000325329249701269,
      "loss": 2.2344,
      "step": 109002
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.9867575764656067,
      "learning_rate": 0.0003253251737427393,
      "loss": 2.2045,
      "step": 109003
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.9094947576522827,
      "learning_rate": 0.00032532109777950113,
      "loss": 2.0903,
      "step": 109004
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.9715675711631775,
      "learning_rate": 0.00032531702181155537,
      "loss": 2.0848,
      "step": 109005
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.9224940538406372,
      "learning_rate": 0.0003253129458389026,
      "loss": 2.0281,
      "step": 109006
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.8669465780258179,
      "learning_rate": 0.0003253088698615437,
      "loss": 2.0369,
      "step": 109007
    },
    {
      "epoch": 1.42,
      "grad_norm": 1.225917935371399,
      "learning_rate": 0.0003253047938794795,
      "loss": 2.0067,
      "step": 109008
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.9649251699447632,
      "learning_rate": 0.0003253007178927106,
      "loss": 2.1914,
      "step": 109009
    },
    {
      "epoch": 1.42,
      "grad_norm": 1.0378063917160034,
      "learning_rate": 0.0003252966419012377,
      "loss": 2.1916,
      "step": 109010
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.8606073260307312,
      "learning_rate": 0.0003252925659050618,
      "loss": 2.1355,
      "step": 109011
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.856867253780365,
      "learning_rate": 0.0003252884899041836,
      "loss": 2.0846,
      "step": 109012
    },
    {
      "epoch": 1.42,
      "grad_norm": 1.111201286315918,
      "learning_rate": 0.00032528441389860366,
      "loss": 2.259,
      "step": 109013
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.9629143476486206,
      "learning_rate": 0.0003252803378883229,
      "loss": 2.1946,
      "step": 109014
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.9068132638931274,
      "learning_rate": 0.0003252762618733421,
      "loss": 1.8416,
      "step": 109015
    },
    {
      "epoch": 1.42,
      "grad_norm": 1.1018263101577759,
      "learning_rate": 0.0003252721858536619,
      "loss": 2.1683,
      "step": 109016
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.9444063901901245,
      "learning_rate": 0.0003252681098292832,
      "loss": 2.1781,
      "step": 109017
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.9147486686706543,
      "learning_rate": 0.00032526403380020663,
      "loss": 2.4268,
      "step": 109018
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.8881849050521851,
      "learning_rate": 0.000325259957766433,
      "loss": 2.1928,
      "step": 109019
    },
    {
      "epoch": 1.42,
      "grad_norm": 1.0470178127288818,
      "learning_rate": 0.00032525588172796304,
      "loss": 1.9727,
      "step": 109020
    },
    {
      "epoch": 1.42,
      "grad_norm": 1.1299316883087158,
      "learning_rate": 0.00032525180568479767,
      "loss": 2.281,
      "step": 109021
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.9098279476165771,
      "learning_rate": 0.0003252477296369374,
      "loss": 2.1794,
      "step": 109022
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.9554760456085205,
      "learning_rate": 0.0003252436535843831,
      "loss": 2.0881,
      "step": 109023
    },
    {
      "epoch": 1.42,
      "grad_norm": 1.0181807279586792,
      "learning_rate": 0.00032523957752713555,
      "loss": 2.0471,
      "step": 109024
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.9903833270072937,
      "learning_rate": 0.0003252355014651955,
      "loss": 2.0021,
      "step": 109025
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.9163921475410461,
      "learning_rate": 0.00032523142539856364,
      "loss": 2.1186,
      "step": 109026
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.9144022464752197,
      "learning_rate": 0.00032522734932724083,
      "loss": 2.1359,
      "step": 109027
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.8965736627578735,
      "learning_rate": 0.0003252232732512277,
      "loss": 1.9545,
      "step": 109028
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.9993308186531067,
      "learning_rate": 0.00032521919717052515,
      "loss": 2.2603,
      "step": 109029
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.9695993661880493,
      "learning_rate": 0.00032521512108513395,
      "loss": 2.3078,
      "step": 109030
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.9804081916809082,
      "learning_rate": 0.00032521104499505467,
      "loss": 2.0345,
      "step": 109031
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.8855313062667847,
      "learning_rate": 0.0003252069689002882,
      "loss": 1.9984,
      "step": 109032
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.8429263830184937,
      "learning_rate": 0.0003252028928008354,
      "loss": 2.3643,
      "step": 109033
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.9202084541320801,
      "learning_rate": 0.0003251988166966968,
      "loss": 2.1112,
      "step": 109034
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.8979921936988831,
      "learning_rate": 0.0003251947405878732,
      "loss": 2.0059,
      "step": 109035
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.8249958753585815,
      "learning_rate": 0.0003251906644743655,
      "loss": 1.8959,
      "step": 109036
    },
    {
      "epoch": 1.42,
      "grad_norm": 1.0445955991744995,
      "learning_rate": 0.0003251865883561744,
      "loss": 2.3015,
      "step": 109037
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.9605159759521484,
      "learning_rate": 0.00032518251223330057,
      "loss": 2.4049,
      "step": 109038
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.9371368288993835,
      "learning_rate": 0.00032517843610574496,
      "loss": 2.1744,
      "step": 109039
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.8783795237541199,
      "learning_rate": 0.0003251743599735081,
      "loss": 2.0215,
      "step": 109040
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.9510579109191895,
      "learning_rate": 0.00032517028383659085,
      "loss": 2.074,
      "step": 109041
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.9529417753219604,
      "learning_rate": 0.00032516620769499396,
      "loss": 1.9066,
      "step": 109042
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.7986515760421753,
      "learning_rate": 0.00032516213154871825,
      "loss": 2.1006,
      "step": 109043
    },
    {
      "epoch": 1.42,
      "grad_norm": 1.1463454961776733,
      "learning_rate": 0.00032515805539776437,
      "loss": 2.2277,
      "step": 109044
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.8305240869522095,
      "learning_rate": 0.0003251539792421332,
      "loss": 2.3117,
      "step": 109045
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.9296672344207764,
      "learning_rate": 0.00032514990308182536,
      "loss": 2.1187,
      "step": 109046
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.8465682864189148,
      "learning_rate": 0.0003251458269168417,
      "loss": 1.9947,
      "step": 109047
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.9942271709442139,
      "learning_rate": 0.00032514175074718304,
      "loss": 2.0625,
      "step": 109048
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.9499667286872864,
      "learning_rate": 0.00032513767457284997,
      "loss": 2.1364,
      "step": 109049
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.8916130065917969,
      "learning_rate": 0.0003251335983938433,
      "loss": 2.2166,
      "step": 109050
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.867281973361969,
      "learning_rate": 0.0003251295222101639,
      "loss": 2.202,
      "step": 109051
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.9971227049827576,
      "learning_rate": 0.00032512544602181237,
      "loss": 2.0243,
      "step": 109052
    },
    {
      "epoch": 1.42,
      "grad_norm": 1.2810152769088745,
      "learning_rate": 0.0003251213698287896,
      "loss": 2.1873,
      "step": 109053
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.8600361943244934,
      "learning_rate": 0.00032511729363109636,
      "loss": 2.2955,
      "step": 109054
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.9018213748931885,
      "learning_rate": 0.00032511321742873325,
      "loss": 2.1315,
      "step": 109055
    },
    {
      "epoch": 1.42,
      "grad_norm": 1.0052711963653564,
      "learning_rate": 0.0003251091412217011,
      "loss": 2.1276,
      "step": 109056
    },
    {
      "epoch": 1.42,
      "grad_norm": 1.0753130912780762,
      "learning_rate": 0.00032510506501000076,
      "loss": 2.0362,
      "step": 109057
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.9550663232803345,
      "learning_rate": 0.00032510098879363284,
      "loss": 2.3449,
      "step": 109058
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.8970576524734497,
      "learning_rate": 0.00032509691257259814,
      "loss": 2.1291,
      "step": 109059
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.9638381004333496,
      "learning_rate": 0.0003250928363468976,
      "loss": 2.2009,
      "step": 109060
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.9983571171760559,
      "learning_rate": 0.0003250887601165317,
      "loss": 2.0915,
      "step": 109061
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.8130504488945007,
      "learning_rate": 0.00032508468388150135,
      "loss": 2.128,
      "step": 109062
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.8728798627853394,
      "learning_rate": 0.00032508060764180726,
      "loss": 2.1156,
      "step": 109063
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.8437493443489075,
      "learning_rate": 0.0003250765313974503,
      "loss": 2.1605,
      "step": 109064
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.8221173882484436,
      "learning_rate": 0.0003250724551484311,
      "loss": 2.2773,
      "step": 109065
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.947520911693573,
      "learning_rate": 0.00032506837889475045,
      "loss": 2.142,
      "step": 109066
    },
    {
      "epoch": 1.42,
      "grad_norm": 1.1898319721221924,
      "learning_rate": 0.0003250643026364091,
      "loss": 2.4522,
      "step": 109067
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.897115170955658,
      "learning_rate": 0.0003250602263734078,
      "loss": 2.1609,
      "step": 109068
    },
    {
      "epoch": 1.42,
      "grad_norm": 1.001466155052185,
      "learning_rate": 0.00032505615010574744,
      "loss": 2.0559,
      "step": 109069
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.8792204856872559,
      "learning_rate": 0.0003250520738334286,
      "loss": 2.3243,
      "step": 109070
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.9196714162826538,
      "learning_rate": 0.0003250479975564521,
      "loss": 2.1571,
      "step": 109071
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.9073208570480347,
      "learning_rate": 0.00032504392127481865,
      "loss": 2.2202,
      "step": 109072
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.8746318817138672,
      "learning_rate": 0.00032503984498852913,
      "loss": 2.2262,
      "step": 109073
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.8306217193603516,
      "learning_rate": 0.00032503576869758426,
      "loss": 1.9229,
      "step": 109074
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.9282665848731995,
      "learning_rate": 0.0003250316924019847,
      "loss": 2.117,
      "step": 109075
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.9331004023551941,
      "learning_rate": 0.00032502761610173135,
      "loss": 1.8284,
      "step": 109076
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.8893274664878845,
      "learning_rate": 0.00032502353979682483,
      "loss": 2.3181,
      "step": 109077
    },
    {
      "epoch": 1.42,
      "grad_norm": 1.0061476230621338,
      "learning_rate": 0.000325019463487266,
      "loss": 2.2784,
      "step": 109078
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.8980743885040283,
      "learning_rate": 0.00032501538717305556,
      "loss": 2.1448,
      "step": 109079
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.8786199688911438,
      "learning_rate": 0.00032501131085419427,
      "loss": 1.9663,
      "step": 109080
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.8495588898658752,
      "learning_rate": 0.0003250072345306829,
      "loss": 1.9831,
      "step": 109081
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.9040653705596924,
      "learning_rate": 0.00032500315820252224,
      "loss": 2.145,
      "step": 109082
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.949867308139801,
      "learning_rate": 0.000324999081869713,
      "loss": 2.2226,
      "step": 109083
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.87135910987854,
      "learning_rate": 0.000324995005532256,
      "loss": 2.1875,
      "step": 109084
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.9985013604164124,
      "learning_rate": 0.0003249909291901519,
      "loss": 2.2146,
      "step": 109085
    },
    {
      "epoch": 1.42,
      "grad_norm": 1.0235443115234375,
      "learning_rate": 0.00032498685284340157,
      "loss": 2.3129,
      "step": 109086
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.919597864151001,
      "learning_rate": 0.00032498277649200565,
      "loss": 2.1394,
      "step": 109087
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.8234529495239258,
      "learning_rate": 0.000324978700135965,
      "loss": 1.9501,
      "step": 109088
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.9210216999053955,
      "learning_rate": 0.0003249746237752804,
      "loss": 2.0503,
      "step": 109089
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.8791350722312927,
      "learning_rate": 0.00032497054740995244,
      "loss": 2.2854,
      "step": 109090
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.8835986256599426,
      "learning_rate": 0.000324966471039982,
      "loss": 2.1648,
      "step": 109091
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.9600919485092163,
      "learning_rate": 0.0003249623946653698,
      "loss": 2.1814,
      "step": 109092
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.8175342082977295,
      "learning_rate": 0.00032495831828611673,
      "loss": 2.4661,
      "step": 109093
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.948827862739563,
      "learning_rate": 0.00032495424190222337,
      "loss": 2.1866,
      "step": 109094
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.9337355494499207,
      "learning_rate": 0.0003249501655136906,
      "loss": 2.1458,
      "step": 109095
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.9640403389930725,
      "learning_rate": 0.000324946089120519,
      "loss": 2.1638,
      "step": 109096
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.8706884980201721,
      "learning_rate": 0.0003249420127227096,
      "loss": 2.2233,
      "step": 109097
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.8109347820281982,
      "learning_rate": 0.0003249379363202629,
      "loss": 2.0686,
      "step": 109098
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.8437332510948181,
      "learning_rate": 0.0003249338599131798,
      "loss": 2.221,
      "step": 109099
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.9458763003349304,
      "learning_rate": 0.00032492978350146104,
      "loss": 2.2067,
      "step": 109100
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.9210492372512817,
      "learning_rate": 0.0003249257070851073,
      "loss": 2.0966,
      "step": 109101
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.9146968722343445,
      "learning_rate": 0.00032492163066411944,
      "loss": 2.0992,
      "step": 109102
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.9975591897964478,
      "learning_rate": 0.00032491755423849823,
      "loss": 2.1937,
      "step": 109103
    },
    {
      "epoch": 1.42,
      "grad_norm": 1.0630017518997192,
      "learning_rate": 0.00032491347780824433,
      "loss": 2.109,
      "step": 109104
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.8576087951660156,
      "learning_rate": 0.0003249094013733585,
      "loss": 2.1083,
      "step": 109105
    },
    {
      "epoch": 1.42,
      "grad_norm": 1.1876894235610962,
      "learning_rate": 0.0003249053249338416,
      "loss": 2.2179,
      "step": 109106
    },
    {
      "epoch": 1.42,
      "grad_norm": 1.1951038837432861,
      "learning_rate": 0.0003249012484896943,
      "loss": 2.3055,
      "step": 109107
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.9268895387649536,
      "learning_rate": 0.00032489717204091747,
      "loss": 1.9114,
      "step": 109108
    },
    {
      "epoch": 1.42,
      "grad_norm": 1.1749382019042969,
      "learning_rate": 0.00032489309558751174,
      "loss": 2.3635,
      "step": 109109
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.8609443306922913,
      "learning_rate": 0.00032488901912947787,
      "loss": 2.0513,
      "step": 109110
    },
    {
      "epoch": 1.42,
      "grad_norm": 1.0654528141021729,
      "learning_rate": 0.0003248849426668167,
      "loss": 2.2331,
      "step": 109111
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.9871187806129456,
      "learning_rate": 0.0003248808661995289,
      "loss": 2.1514,
      "step": 109112
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.9871729016304016,
      "learning_rate": 0.00032487678972761537,
      "loss": 2.1033,
      "step": 109113
    },
    {
      "epoch": 1.42,
      "grad_norm": 1.139571189880371,
      "learning_rate": 0.0003248727132510767,
      "loss": 2.1433,
      "step": 109114
    },
    {
      "epoch": 1.42,
      "grad_norm": 1.0205049514770508,
      "learning_rate": 0.00032486863676991373,
      "loss": 2.2007,
      "step": 109115
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.8515167236328125,
      "learning_rate": 0.0003248645602841272,
      "loss": 2.1897,
      "step": 109116
    },
    {
      "epoch": 1.42,
      "grad_norm": 1.2370939254760742,
      "learning_rate": 0.0003248604837937179,
      "loss": 2.1323,
      "step": 109117
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.9893661737442017,
      "learning_rate": 0.00032485640729868663,
      "loss": 2.2761,
      "step": 109118
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.9092854261398315,
      "learning_rate": 0.000324852330799034,
      "loss": 2.1659,
      "step": 109119
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.9459622502326965,
      "learning_rate": 0.00032484825429476084,
      "loss": 1.9541,
      "step": 109120
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.9369596838951111,
      "learning_rate": 0.000324844177785868,
      "loss": 2.301,
      "step": 109121
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.8560547232627869,
      "learning_rate": 0.00032484010127235613,
      "loss": 2.084,
      "step": 109122
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.8450219631195068,
      "learning_rate": 0.000324836024754226,
      "loss": 2.2532,
      "step": 109123
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.8998631834983826,
      "learning_rate": 0.00032483194823147845,
      "loss": 2.2347,
      "step": 109124
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.9478685855865479,
      "learning_rate": 0.00032482787170411403,
      "loss": 2.3314,
      "step": 109125
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.9258409738540649,
      "learning_rate": 0.0003248237951721337,
      "loss": 2.3762,
      "step": 109126
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.9630237221717834,
      "learning_rate": 0.00032481971863553824,
      "loss": 2.1942,
      "step": 109127
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.9576582908630371,
      "learning_rate": 0.00032481564209432823,
      "loss": 2.1538,
      "step": 109128
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.9025982618331909,
      "learning_rate": 0.00032481156554850456,
      "loss": 2.1583,
      "step": 109129
    },
    {
      "epoch": 1.42,
      "grad_norm": 1.0300782918930054,
      "learning_rate": 0.00032480748899806793,
      "loss": 2.1792,
      "step": 109130
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.9655986428260803,
      "learning_rate": 0.0003248034124430192,
      "loss": 2.0393,
      "step": 109131
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.8766465187072754,
      "learning_rate": 0.00032479933588335896,
      "loss": 1.9681,
      "step": 109132
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.9304198622703552,
      "learning_rate": 0.0003247952593190881,
      "loss": 2.2836,
      "step": 109133
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.8900645971298218,
      "learning_rate": 0.0003247911827502073,
      "loss": 2.3265,
      "step": 109134
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.9856216907501221,
      "learning_rate": 0.00032478710617671737,
      "loss": 2.0117,
      "step": 109135
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.9013920426368713,
      "learning_rate": 0.000324783029598619,
      "loss": 2.2688,
      "step": 109136
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.9218877553939819,
      "learning_rate": 0.0003247789530159131,
      "loss": 2.1282,
      "step": 109137
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.920957088470459,
      "learning_rate": 0.0003247748764286002,
      "loss": 1.995,
      "step": 109138
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.9658687114715576,
      "learning_rate": 0.00032477079983668126,
      "loss": 2.3828,
      "step": 109139
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.9671367406845093,
      "learning_rate": 0.000324766723240157,
      "loss": 2.2827,
      "step": 109140
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.8616886138916016,
      "learning_rate": 0.00032476264663902805,
      "loss": 2.0095,
      "step": 109141
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.8561856150627136,
      "learning_rate": 0.0003247585700332953,
      "loss": 2.4225,
      "step": 109142
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.970443606376648,
      "learning_rate": 0.0003247544934229595,
      "loss": 2.2008,
      "step": 109143
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.935164749622345,
      "learning_rate": 0.00032475041680802126,
      "loss": 1.8689,
      "step": 109144
    },
    {
      "epoch": 1.42,
      "grad_norm": 1.1912267208099365,
      "learning_rate": 0.00032474634018848155,
      "loss": 2.24,
      "step": 109145
    },
    {
      "epoch": 1.42,
      "grad_norm": 1.0302610397338867,
      "learning_rate": 0.000324742263564341,
      "loss": 2.2646,
      "step": 109146
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.9796963334083557,
      "learning_rate": 0.00032473818693560044,
      "loss": 2.2132,
      "step": 109147
    },
    {
      "epoch": 1.42,
      "grad_norm": 1.0563955307006836,
      "learning_rate": 0.00032473411030226046,
      "loss": 2.2657,
      "step": 109148
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.892317533493042,
      "learning_rate": 0.0003247300336643221,
      "loss": 2.0629,
      "step": 109149
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.8230689167976379,
      "learning_rate": 0.0003247259570217858,
      "loss": 2.15,
      "step": 109150
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.9117147922515869,
      "learning_rate": 0.00032472188037465256,
      "loss": 2.2905,
      "step": 109151
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.9389111399650574,
      "learning_rate": 0.0003247178037229231,
      "loss": 2.1507,
      "step": 109152
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.8142596483230591,
      "learning_rate": 0.00032471372706659814,
      "loss": 2.2028,
      "step": 109153
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.8346421122550964,
      "learning_rate": 0.0003247096504056784,
      "loss": 2.1408,
      "step": 109154
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.8913424611091614,
      "learning_rate": 0.0003247055737401646,
      "loss": 2.1317,
      "step": 109155
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.9583101272583008,
      "learning_rate": 0.00032470149707005763,
      "loss": 2.112,
      "step": 109156
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.9581846594810486,
      "learning_rate": 0.00032469742039535816,
      "loss": 2.1223,
      "step": 109157
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.9119543433189392,
      "learning_rate": 0.00032469334371606703,
      "loss": 2.1603,
      "step": 109158
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.8559362292289734,
      "learning_rate": 0.00032468926703218485,
      "loss": 2.1785,
      "step": 109159
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.8818721175193787,
      "learning_rate": 0.0003246851903437125,
      "loss": 2.1223,
      "step": 109160
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.9263718724250793,
      "learning_rate": 0.00032468111365065074,
      "loss": 2.2679,
      "step": 109161
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.8979472517967224,
      "learning_rate": 0.0003246770369530003,
      "loss": 2.0993,
      "step": 109162
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.8905248045921326,
      "learning_rate": 0.0003246729602507619,
      "loss": 2.0765,
      "step": 109163
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.8681391477584839,
      "learning_rate": 0.0003246688835439364,
      "loss": 2.2451,
      "step": 109164
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.9663434624671936,
      "learning_rate": 0.0003246648068325244,
      "loss": 2.2255,
      "step": 109165
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.9367373585700989,
      "learning_rate": 0.00032466073011652675,
      "loss": 2.2078,
      "step": 109166
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.8842475414276123,
      "learning_rate": 0.0003246566533959442,
      "loss": 2.463,
      "step": 109167
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.9608306884765625,
      "learning_rate": 0.00032465257667077757,
      "loss": 2.1157,
      "step": 109168
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.7695921063423157,
      "learning_rate": 0.00032464849994102757,
      "loss": 1.9861,
      "step": 109169
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.8820075392723083,
      "learning_rate": 0.00032464442320669497,
      "loss": 2.2244,
      "step": 109170
    },
    {
      "epoch": 1.42,
      "grad_norm": 1.1106256246566772,
      "learning_rate": 0.00032464034646778036,
      "loss": 2.2177,
      "step": 109171
    },
    {
      "epoch": 1.42,
      "grad_norm": 1.033915400505066,
      "learning_rate": 0.00032463626972428474,
      "loss": 2.2606,
      "step": 109172
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.9184296727180481,
      "learning_rate": 0.00032463219297620873,
      "loss": 2.2706,
      "step": 109173
    },
    {
      "epoch": 1.42,
      "grad_norm": 1.0558747053146362,
      "learning_rate": 0.0003246281162235532,
      "loss": 1.9127,
      "step": 109174
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.8608380556106567,
      "learning_rate": 0.0003246240394663188,
      "loss": 2.3517,
      "step": 109175
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.9795218110084534,
      "learning_rate": 0.00032461996270450634,
      "loss": 2.224,
      "step": 109176
    },
    {
      "epoch": 1.42,
      "grad_norm": 1.0082412958145142,
      "learning_rate": 0.00032461588593811654,
      "loss": 2.2967,
      "step": 109177
    },
    {
      "epoch": 1.42,
      "grad_norm": 1.0260015726089478,
      "learning_rate": 0.00032461180916715026,
      "loss": 2.2766,
      "step": 109178
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.8444933891296387,
      "learning_rate": 0.00032460773239160806,
      "loss": 2.2327,
      "step": 109179
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.9128385782241821,
      "learning_rate": 0.0003246036556114909,
      "loss": 2.258,
      "step": 109180
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.8765206336975098,
      "learning_rate": 0.0003245995788267995,
      "loss": 2.1226,
      "step": 109181
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.8416591286659241,
      "learning_rate": 0.00032459550203753443,
      "loss": 2.2068,
      "step": 109182
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.9331367611885071,
      "learning_rate": 0.00032459142524369664,
      "loss": 2.0528,
      "step": 109183
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.9298574924468994,
      "learning_rate": 0.0003245873484452869,
      "loss": 2.2202,
      "step": 109184
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.9083724617958069,
      "learning_rate": 0.0003245832716423059,
      "loss": 2.2361,
      "step": 109185
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.8296628594398499,
      "learning_rate": 0.00032457919483475437,
      "loss": 2.1309,
      "step": 109186
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.9553390741348267,
      "learning_rate": 0.0003245751180226331,
      "loss": 1.9391,
      "step": 109187
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.8760870695114136,
      "learning_rate": 0.0003245710412059429,
      "loss": 2.1452,
      "step": 109188
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.9005774855613708,
      "learning_rate": 0.0003245669643846845,
      "loss": 2.148,
      "step": 109189
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.9838858842849731,
      "learning_rate": 0.00032456288755885847,
      "loss": 2.0858,
      "step": 109190
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.9178541302680969,
      "learning_rate": 0.00032455881072846593,
      "loss": 2.1209,
      "step": 109191
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.9066615700721741,
      "learning_rate": 0.00032455473389350734,
      "loss": 2.398,
      "step": 109192
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.8963969945907593,
      "learning_rate": 0.00032455065705398354,
      "loss": 2.1669,
      "step": 109193
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.9430689811706543,
      "learning_rate": 0.0003245465802098954,
      "loss": 2.2856,
      "step": 109194
    },
    {
      "epoch": 1.42,
      "grad_norm": 1.1192283630371094,
      "learning_rate": 0.00032454250336124357,
      "loss": 2.1356,
      "step": 109195
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.9667301774024963,
      "learning_rate": 0.00032453842650802874,
      "loss": 2.1713,
      "step": 109196
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.9951456785202026,
      "learning_rate": 0.00032453434965025186,
      "loss": 2.2964,
      "step": 109197
    },
    {
      "epoch": 1.42,
      "grad_norm": 1.1115421056747437,
      "learning_rate": 0.0003245302727879135,
      "loss": 2.0989,
      "step": 109198
    },
    {
      "epoch": 1.42,
      "grad_norm": 1.0269992351531982,
      "learning_rate": 0.0003245261959210146,
      "loss": 2.5107,
      "step": 109199
    },
    {
      "epoch": 1.42,
      "grad_norm": 1.0189845561981201,
      "learning_rate": 0.0003245221190495557,
      "loss": 2.2141,
      "step": 109200
    },
    {
      "epoch": 1.42,
      "grad_norm": 1.0541936159133911,
      "learning_rate": 0.00032451804217353776,
      "loss": 2.1918,
      "step": 109201
    },
    {
      "epoch": 1.42,
      "grad_norm": 1.0508074760437012,
      "learning_rate": 0.0003245139652929614,
      "loss": 1.9686,
      "step": 109202
    },
    {
      "epoch": 1.42,
      "grad_norm": 1.0443174839019775,
      "learning_rate": 0.00032450988840782745,
      "loss": 2.1287,
      "step": 109203
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.8325443863868713,
      "learning_rate": 0.00032450581151813665,
      "loss": 1.977,
      "step": 109204
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.8830262422561646,
      "learning_rate": 0.00032450173462388976,
      "loss": 2.0555,
      "step": 109205
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.9538608193397522,
      "learning_rate": 0.00032449765772508754,
      "loss": 1.8419,
      "step": 109206
    },
    {
      "epoch": 1.42,
      "grad_norm": 1.0724027156829834,
      "learning_rate": 0.00032449358082173074,
      "loss": 2.1237,
      "step": 109207
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.8811385631561279,
      "learning_rate": 0.0003244895039138201,
      "loss": 2.2736,
      "step": 109208
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.9687040448188782,
      "learning_rate": 0.0003244854270013564,
      "loss": 2.1172,
      "step": 109209
    },
    {
      "epoch": 1.42,
      "grad_norm": 1.1283996105194092,
      "learning_rate": 0.0003244813500843405,
      "loss": 2.2423,
      "step": 109210
    },
    {
      "epoch": 1.42,
      "grad_norm": 1.2002631425857544,
      "learning_rate": 0.0003244772731627729,
      "loss": 2.2857,
      "step": 109211
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.8390625715255737,
      "learning_rate": 0.00032447319623665457,
      "loss": 1.9353,
      "step": 109212
    },
    {
      "epoch": 1.42,
      "grad_norm": 1.1829813718795776,
      "learning_rate": 0.00032446911930598623,
      "loss": 2.3544,
      "step": 109213
    },
    {
      "epoch": 1.42,
      "grad_norm": 1.0553969144821167,
      "learning_rate": 0.0003244650423707686,
      "loss": 2.1754,
      "step": 109214
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.9986478090286255,
      "learning_rate": 0.00032446096543100244,
      "loss": 1.9374,
      "step": 109215
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.8842123746871948,
      "learning_rate": 0.0003244568884866886,
      "loss": 2.2474,
      "step": 109216
    },
    {
      "epoch": 1.42,
      "grad_norm": 1.1404991149902344,
      "learning_rate": 0.0003244528115378277,
      "loss": 2.2308,
      "step": 109217
    },
    {
      "epoch": 1.42,
      "grad_norm": 1.0282151699066162,
      "learning_rate": 0.00032444873458442055,
      "loss": 2.2388,
      "step": 109218
    },
    {
      "epoch": 1.42,
      "grad_norm": 1.020017147064209,
      "learning_rate": 0.000324444657626468,
      "loss": 2.2838,
      "step": 109219
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.9371975660324097,
      "learning_rate": 0.00032444058066397063,
      "loss": 2.1283,
      "step": 109220
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.9495496153831482,
      "learning_rate": 0.00032443650369692927,
      "loss": 2.1995,
      "step": 109221
    },
    {
      "epoch": 1.42,
      "grad_norm": 1.4537218809127808,
      "learning_rate": 0.0003244324267253448,
      "loss": 2.2332,
      "step": 109222
    },
    {
      "epoch": 1.42,
      "grad_norm": 1.3149755001068115,
      "learning_rate": 0.00032442834974921785,
      "loss": 2.0984,
      "step": 109223
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.9210739731788635,
      "learning_rate": 0.0003244242727685492,
      "loss": 2.2995,
      "step": 109224
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.9416360259056091,
      "learning_rate": 0.0003244201957833397,
      "loss": 2.1125,
      "step": 109225
    },
    {
      "epoch": 1.42,
      "grad_norm": 1.2664819955825806,
      "learning_rate": 0.0003244161187935899,
      "loss": 2.2161,
      "step": 109226
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.9929904937744141,
      "learning_rate": 0.0003244120417993007,
      "loss": 2.0925,
      "step": 109227
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.9616749882698059,
      "learning_rate": 0.0003244079648004729,
      "loss": 2.0256,
      "step": 109228
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.9354140758514404,
      "learning_rate": 0.0003244038877971072,
      "loss": 1.9608,
      "step": 109229
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.9660035967826843,
      "learning_rate": 0.00032439981078920423,
      "loss": 2.3388,
      "step": 109230
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.8924241662025452,
      "learning_rate": 0.00032439573377676504,
      "loss": 2.1262,
      "step": 109231
    },
    {
      "epoch": 1.42,
      "grad_norm": 1.007439374923706,
      "learning_rate": 0.00032439165675979015,
      "loss": 2.1764,
      "step": 109232
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.9538918733596802,
      "learning_rate": 0.00032438757973828037,
      "loss": 2.1157,
      "step": 109233
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.939700722694397,
      "learning_rate": 0.00032438350271223656,
      "loss": 2.1155,
      "step": 109234
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.8746232390403748,
      "learning_rate": 0.0003243794256816593,
      "loss": 2.1036,
      "step": 109235
    },
    {
      "epoch": 1.42,
      "grad_norm": 1.1222771406173706,
      "learning_rate": 0.00032437534864654947,
      "loss": 2.0487,
      "step": 109236
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.9736124277114868,
      "learning_rate": 0.00032437127160690786,
      "loss": 2.2397,
      "step": 109237
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.9009499549865723,
      "learning_rate": 0.0003243671945627351,
      "loss": 2.2573,
      "step": 109238
    },
    {
      "epoch": 1.42,
      "grad_norm": 1.3713324069976807,
      "learning_rate": 0.000324363117514032,
      "loss": 2.1689,
      "step": 109239
    },
    {
      "epoch": 1.42,
      "grad_norm": 1.0095725059509277,
      "learning_rate": 0.00032435904046079945,
      "loss": 2.2013,
      "step": 109240
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.8262987732887268,
      "learning_rate": 0.000324354963403038,
      "loss": 2.0247,
      "step": 109241
    },
    {
      "epoch": 1.42,
      "grad_norm": 1.098642349243164,
      "learning_rate": 0.00032435088634074857,
      "loss": 2.1607,
      "step": 109242
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.8873627185821533,
      "learning_rate": 0.0003243468092739318,
      "loss": 2.1346,
      "step": 109243
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.9990530610084534,
      "learning_rate": 0.0003243427322025885,
      "loss": 2.059,
      "step": 109244
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.9749256372451782,
      "learning_rate": 0.0003243386551267194,
      "loss": 2.0313,
      "step": 109245
    },
    {
      "epoch": 1.42,
      "grad_norm": 1.0579032897949219,
      "learning_rate": 0.0003243345780463254,
      "loss": 2.1562,
      "step": 109246
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.8645742535591125,
      "learning_rate": 0.00032433050096140697,
      "loss": 2.1276,
      "step": 109247
    },
    {
      "epoch": 1.42,
      "grad_norm": 1.3315801620483398,
      "learning_rate": 0.00032432642387196514,
      "loss": 2.252,
      "step": 109248
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.9474667906761169,
      "learning_rate": 0.0003243223467780006,
      "loss": 2.1877,
      "step": 109249
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.8515605926513672,
      "learning_rate": 0.000324318269679514,
      "loss": 2.1571,
      "step": 109250
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.9532709121704102,
      "learning_rate": 0.0003243141925765062,
      "loss": 2.2863,
      "step": 109251
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.8499199748039246,
      "learning_rate": 0.000324310115468978,
      "loss": 2.1483,
      "step": 109252
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.8752698302268982,
      "learning_rate": 0.00032430603835693005,
      "loss": 2.0658,
      "step": 109253
    },
    {
      "epoch": 1.42,
      "grad_norm": 1.0514343976974487,
      "learning_rate": 0.00032430196124036306,
      "loss": 2.0041,
      "step": 109254
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.9336526393890381,
      "learning_rate": 0.000324297884119278,
      "loss": 2.2388,
      "step": 109255
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.9156041145324707,
      "learning_rate": 0.00032429380699367546,
      "loss": 2.252,
      "step": 109256
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.8523234724998474,
      "learning_rate": 0.0003242897298635562,
      "loss": 1.8523,
      "step": 109257
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.9137225151062012,
      "learning_rate": 0.00032428565272892105,
      "loss": 2.1478,
      "step": 109258
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.90935879945755,
      "learning_rate": 0.0003242815755897708,
      "loss": 2.143,
      "step": 109259
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.9889180660247803,
      "learning_rate": 0.00032427749844610604,
      "loss": 1.9974,
      "step": 109260
    },
    {
      "epoch": 1.42,
      "grad_norm": 1.056238055229187,
      "learning_rate": 0.0003242734212979277,
      "loss": 2.3893,
      "step": 109261
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.9063456058502197,
      "learning_rate": 0.00032426934414523653,
      "loss": 2.0974,
      "step": 109262
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.9719317555427551,
      "learning_rate": 0.00032426526698803315,
      "loss": 2.0578,
      "step": 109263
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.8918770551681519,
      "learning_rate": 0.00032426118982631835,
      "loss": 2.3182,
      "step": 109264
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.959811806678772,
      "learning_rate": 0.000324257112660093,
      "loss": 2.0782,
      "step": 109265
    },
    {
      "epoch": 1.42,
      "grad_norm": 1.0891903638839722,
      "learning_rate": 0.00032425303548935786,
      "loss": 2.1712,
      "step": 109266
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.9207801222801208,
      "learning_rate": 0.00032424895831411346,
      "loss": 1.811,
      "step": 109267
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.9033584594726562,
      "learning_rate": 0.0003242448811343608,
      "loss": 2.2004,
      "step": 109268
    },
    {
      "epoch": 1.42,
      "grad_norm": 1.5194579362869263,
      "learning_rate": 0.00032424080395010066,
      "loss": 2.1002,
      "step": 109269
    },
    {
      "epoch": 1.42,
      "grad_norm": 1.3034218549728394,
      "learning_rate": 0.0003242367267613336,
      "loss": 2.1876,
      "step": 109270
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.9550269246101379,
      "learning_rate": 0.00032423264956806045,
      "loss": 2.1778,
      "step": 109271
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.9955640435218811,
      "learning_rate": 0.000324228572370282,
      "loss": 2.1432,
      "step": 109272
    },
    {
      "epoch": 1.42,
      "grad_norm": 1.0710502862930298,
      "learning_rate": 0.00032422449516799907,
      "loss": 1.8972,
      "step": 109273
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.9417079091072083,
      "learning_rate": 0.00032422041796121225,
      "loss": 2.1521,
      "step": 109274
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.8885870575904846,
      "learning_rate": 0.0003242163407499225,
      "loss": 2.1129,
      "step": 109275
    },
    {
      "epoch": 1.42,
      "grad_norm": 1.0507662296295166,
      "learning_rate": 0.00032421226353413044,
      "loss": 2.1437,
      "step": 109276
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.9618041515350342,
      "learning_rate": 0.00032420818631383685,
      "loss": 2.1685,
      "step": 109277
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.7921088933944702,
      "learning_rate": 0.00032420410908904246,
      "loss": 2.1142,
      "step": 109278
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.9630962610244751,
      "learning_rate": 0.00032420003185974815,
      "loss": 2.159,
      "step": 109279
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.8574070334434509,
      "learning_rate": 0.0003241959546259546,
      "loss": 2.3635,
      "step": 109280
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.8503022193908691,
      "learning_rate": 0.0003241918773876625,
      "loss": 2.1234,
      "step": 109281
    },
    {
      "epoch": 1.42,
      "grad_norm": 1.1149126291275024,
      "learning_rate": 0.00032418780014487265,
      "loss": 2.0618,
      "step": 109282
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.901924192905426,
      "learning_rate": 0.000324183722897586,
      "loss": 2.1386,
      "step": 109283
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.9367682933807373,
      "learning_rate": 0.000324179645645803,
      "loss": 2.2038,
      "step": 109284
    },
    {
      "epoch": 1.42,
      "grad_norm": 1.0578596591949463,
      "learning_rate": 0.0003241755683895245,
      "loss": 2.1396,
      "step": 109285
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.9082845449447632,
      "learning_rate": 0.00032417149112875143,
      "loss": 2.1121,
      "step": 109286
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.8466026186943054,
      "learning_rate": 0.0003241674138634844,
      "loss": 2.0982,
      "step": 109287
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.9013246297836304,
      "learning_rate": 0.00032416333659372406,
      "loss": 2.0852,
      "step": 109288
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.9695388674736023,
      "learning_rate": 0.00032415925931947147,
      "loss": 2.1855,
      "step": 109289
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.8913897275924683,
      "learning_rate": 0.0003241551820407271,
      "loss": 2.2485,
      "step": 109290
    },
    {
      "epoch": 1.42,
      "grad_norm": 1.0124760866165161,
      "learning_rate": 0.00032415110475749186,
      "loss": 2.1725,
      "step": 109291
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.8779125213623047,
      "learning_rate": 0.00032414702746976653,
      "loss": 2.1343,
      "step": 109292
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.8876447081565857,
      "learning_rate": 0.00032414295017755174,
      "loss": 2.1669,
      "step": 109293
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.9738571047782898,
      "learning_rate": 0.00032413887288084835,
      "loss": 2.1886,
      "step": 109294
    },
    {
      "epoch": 1.42,
      "grad_norm": 1.0890218019485474,
      "learning_rate": 0.0003241347955796572,
      "loss": 2.1559,
      "step": 109295
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.9021865129470825,
      "learning_rate": 0.00032413071827397876,
      "loss": 2.1368,
      "step": 109296
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.8744252920150757,
      "learning_rate": 0.000324126640963814,
      "loss": 1.9888,
      "step": 109297
    },
    {
      "epoch": 1.42,
      "grad_norm": 1.2294399738311768,
      "learning_rate": 0.00032412256364916373,
      "loss": 2.3977,
      "step": 109298
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.8873533606529236,
      "learning_rate": 0.0003241184863300285,
      "loss": 2.1276,
      "step": 109299
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.8556157350540161,
      "learning_rate": 0.0003241144090064092,
      "loss": 2.1516,
      "step": 109300
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.9505943059921265,
      "learning_rate": 0.00032411033167830674,
      "loss": 2.1286,
      "step": 109301
    },
    {
      "epoch": 1.42,
      "grad_norm": 1.0182090997695923,
      "learning_rate": 0.00032410625434572157,
      "loss": 2.3942,
      "step": 109302
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.8637223243713379,
      "learning_rate": 0.0003241021770086546,
      "loss": 2.44,
      "step": 109303
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.9441612362861633,
      "learning_rate": 0.0003240980996671066,
      "loss": 2.1904,
      "step": 109304
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.9858081340789795,
      "learning_rate": 0.00032409402232107834,
      "loss": 2.0293,
      "step": 109305
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.8765324950218201,
      "learning_rate": 0.00032408994497057044,
      "loss": 1.9935,
      "step": 109306
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.8284426927566528,
      "learning_rate": 0.0003240858676155839,
      "loss": 2.2317,
      "step": 109307
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.9002572298049927,
      "learning_rate": 0.00032408179025611927,
      "loss": 2.2584,
      "step": 109308
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.8970214128494263,
      "learning_rate": 0.0003240777128921774,
      "loss": 2.278,
      "step": 109309
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.9051191210746765,
      "learning_rate": 0.000324073635523759,
      "loss": 2.2358,
      "step": 109310
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.9823311567306519,
      "learning_rate": 0.0003240695581508649,
      "loss": 2.1839,
      "step": 109311
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.9057099223136902,
      "learning_rate": 0.0003240654807734958,
      "loss": 2.164,
      "step": 109312
    },
    {
      "epoch": 1.42,
      "grad_norm": 1.0635621547698975,
      "learning_rate": 0.0003240614033916525,
      "loss": 2.2046,
      "step": 109313
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.9881113767623901,
      "learning_rate": 0.0003240573260053356,
      "loss": 2.2984,
      "step": 109314
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.9229632616043091,
      "learning_rate": 0.0003240532486145461,
      "loss": 2.2998,
      "step": 109315
    },
    {
      "epoch": 1.42,
      "grad_norm": 1.0034170150756836,
      "learning_rate": 0.0003240491712192847,
      "loss": 2.2489,
      "step": 109316
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.8432086706161499,
      "learning_rate": 0.000324045093819552,
      "loss": 2.1693,
      "step": 109317
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.9476341605186462,
      "learning_rate": 0.00032404101641534885,
      "loss": 2.4147,
      "step": 109318
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.9085972905158997,
      "learning_rate": 0.00032403693900667617,
      "loss": 1.9483,
      "step": 109319
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.849622905254364,
      "learning_rate": 0.00032403286159353445,
      "loss": 1.9956,
      "step": 109320
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.8974447846412659,
      "learning_rate": 0.0003240287841759246,
      "loss": 2.0021,
      "step": 109321
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.9816859364509583,
      "learning_rate": 0.00032402470675384735,
      "loss": 2.1005,
      "step": 109322
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.9125298857688904,
      "learning_rate": 0.0003240206293273034,
      "loss": 2.1827,
      "step": 109323
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.8956084847450256,
      "learning_rate": 0.0003240165518962936,
      "loss": 2.2803,
      "step": 109324
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.8705909252166748,
      "learning_rate": 0.0003240124744608187,
      "loss": 2.0932,
      "step": 109325
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.9593257904052734,
      "learning_rate": 0.00032400839702087937,
      "loss": 2.3923,
      "step": 109326
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.897932231426239,
      "learning_rate": 0.0003240043195764764,
      "loss": 2.1851,
      "step": 109327
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.8904814124107361,
      "learning_rate": 0.0003240002421276107,
      "loss": 2.1289,
      "step": 109328
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.904183030128479,
      "learning_rate": 0.00032399616467428284,
      "loss": 1.9505,
      "step": 109329
    },
    {
      "epoch": 1.42,
      "grad_norm": 1.026868462562561,
      "learning_rate": 0.0003239920872164936,
      "loss": 2.1016,
      "step": 109330
    },
    {
      "epoch": 1.42,
      "grad_norm": 1.0161209106445312,
      "learning_rate": 0.0003239880097542438,
      "loss": 2.1177,
      "step": 109331
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.909729540348053,
      "learning_rate": 0.0003239839322875342,
      "loss": 2.1125,
      "step": 109332
    },
    {
      "epoch": 1.42,
      "grad_norm": 1.0015009641647339,
      "learning_rate": 0.0003239798548163655,
      "loss": 2.2275,
      "step": 109333
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.8816866278648376,
      "learning_rate": 0.0003239757773407385,
      "loss": 2.0416,
      "step": 109334
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.9019011855125427,
      "learning_rate": 0.00032397169986065396,
      "loss": 2.2139,
      "step": 109335
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.9470581412315369,
      "learning_rate": 0.0003239676223761126,
      "loss": 2.1138,
      "step": 109336
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.9830474853515625,
      "learning_rate": 0.0003239635448871152,
      "loss": 2.0987,
      "step": 109337
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.9997397661209106,
      "learning_rate": 0.00032395946739366264,
      "loss": 2.1461,
      "step": 109338
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.9071122407913208,
      "learning_rate": 0.0003239553898957555,
      "loss": 2.2903,
      "step": 109339
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.9604768753051758,
      "learning_rate": 0.0003239513123933946,
      "loss": 2.0493,
      "step": 109340
    },
    {
      "epoch": 1.42,
      "grad_norm": 1.0351406335830688,
      "learning_rate": 0.0003239472348865807,
      "loss": 1.9424,
      "step": 109341
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.9894205927848816,
      "learning_rate": 0.0003239431573753145,
      "loss": 2.3783,
      "step": 109342
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.8801076412200928,
      "learning_rate": 0.0003239390798595968,
      "loss": 2.0383,
      "step": 109343
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.860508143901825,
      "learning_rate": 0.0003239350023394285,
      "loss": 2.3804,
      "step": 109344
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.8562191724777222,
      "learning_rate": 0.0003239309248148101,
      "loss": 2.0651,
      "step": 109345
    },
    {
      "epoch": 1.42,
      "grad_norm": 1.0087324380874634,
      "learning_rate": 0.00032392684728574253,
      "loss": 2.3181,
      "step": 109346
    },
    {
      "epoch": 1.42,
      "grad_norm": 1.0347349643707275,
      "learning_rate": 0.00032392276975222657,
      "loss": 2.2209,
      "step": 109347
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.9222063422203064,
      "learning_rate": 0.0003239186922142629,
      "loss": 2.2062,
      "step": 109348
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.8928059935569763,
      "learning_rate": 0.0003239146146718522,
      "loss": 2.2395,
      "step": 109349
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.8675331473350525,
      "learning_rate": 0.00032391053712499537,
      "loss": 2.1859,
      "step": 109350
    },
    {
      "epoch": 1.42,
      "grad_norm": 1.050514817237854,
      "learning_rate": 0.00032390645957369323,
      "loss": 2.1896,
      "step": 109351
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.8637217879295349,
      "learning_rate": 0.0003239023820179463,
      "loss": 2.2041,
      "step": 109352
    },
    {
      "epoch": 1.42,
      "grad_norm": 1.0889002084732056,
      "learning_rate": 0.0003238983044577555,
      "loss": 2.0961,
      "step": 109353
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.8892748951911926,
      "learning_rate": 0.0003238942268931216,
      "loss": 1.8737,
      "step": 109354
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.9881296753883362,
      "learning_rate": 0.0003238901493240452,
      "loss": 2.1221,
      "step": 109355
    },
    {
      "epoch": 1.42,
      "grad_norm": 1.0177764892578125,
      "learning_rate": 0.00032388607175052724,
      "loss": 1.9151,
      "step": 109356
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.9681305885314941,
      "learning_rate": 0.00032388199417256836,
      "loss": 2.0076,
      "step": 109357
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.9506043195724487,
      "learning_rate": 0.00032387791659016946,
      "loss": 2.1915,
      "step": 109358
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.927671492099762,
      "learning_rate": 0.0003238738390033311,
      "loss": 2.1226,
      "step": 109359
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.8516724705696106,
      "learning_rate": 0.0003238697614120542,
      "loss": 2.1835,
      "step": 109360
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.953124463558197,
      "learning_rate": 0.00032386568381633955,
      "loss": 2.1401,
      "step": 109361
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.8525187969207764,
      "learning_rate": 0.00032386160621618765,
      "loss": 2.1505,
      "step": 109362
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.9070055484771729,
      "learning_rate": 0.00032385752861159947,
      "loss": 2.0013,
      "step": 109363
    },
    {
      "epoch": 1.42,
      "grad_norm": 1.002461552619934,
      "learning_rate": 0.00032385345100257576,
      "loss": 2.1562,
      "step": 109364
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.9184761047363281,
      "learning_rate": 0.0003238493733891173,
      "loss": 2.0402,
      "step": 109365
    },
    {
      "epoch": 1.42,
      "grad_norm": 1.0307494401931763,
      "learning_rate": 0.0003238452957712247,
      "loss": 2.045,
      "step": 109366
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.9648844003677368,
      "learning_rate": 0.00032384121814889885,
      "loss": 2.1124,
      "step": 109367
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.8594639897346497,
      "learning_rate": 0.0003238371405221405,
      "loss": 2.1063,
      "step": 109368
    },
    {
      "epoch": 1.42,
      "grad_norm": 1.0193345546722412,
      "learning_rate": 0.00032383306289095036,
      "loss": 2.0509,
      "step": 109369
    },
    {
      "epoch": 1.42,
      "grad_norm": 1.1284432411193848,
      "learning_rate": 0.0003238289852553291,
      "loss": 2.1689,
      "step": 109370
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.9262242317199707,
      "learning_rate": 0.0003238249076152777,
      "loss": 2.1637,
      "step": 109371
    },
    {
      "epoch": 1.42,
      "grad_norm": 1.1480575799942017,
      "learning_rate": 0.0003238208299707968,
      "loss": 2.4109,
      "step": 109372
    },
    {
      "epoch": 1.42,
      "grad_norm": 1.0749200582504272,
      "learning_rate": 0.000323816752321887,
      "loss": 2.2337,
      "step": 109373
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.8295587301254272,
      "learning_rate": 0.00032381267466854947,
      "loss": 1.9186,
      "step": 109374
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.9466401934623718,
      "learning_rate": 0.00032380859701078453,
      "loss": 2.1556,
      "step": 109375
    },
    {
      "epoch": 1.42,
      "grad_norm": 1.0238720178604126,
      "learning_rate": 0.0003238045193485932,
      "loss": 2.017,
      "step": 109376
    },
    {
      "epoch": 1.42,
      "grad_norm": 1.194138765335083,
      "learning_rate": 0.00032380044168197616,
      "loss": 2.0692,
      "step": 109377
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.8801193237304688,
      "learning_rate": 0.00032379636401093414,
      "loss": 1.9222,
      "step": 109378
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.8777874708175659,
      "learning_rate": 0.0003237922863354679,
      "loss": 2.3149,
      "step": 109379
    },
    {
      "epoch": 1.42,
      "grad_norm": 1.0731202363967896,
      "learning_rate": 0.00032378820865557826,
      "loss": 2.1561,
      "step": 109380
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.9289072155952454,
      "learning_rate": 0.0003237841309712659,
      "loss": 1.981,
      "step": 109381
    },
    {
      "epoch": 1.42,
      "grad_norm": 1.0334094762802124,
      "learning_rate": 0.0003237800532825317,
      "loss": 2.0004,
      "step": 109382
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.9981951713562012,
      "learning_rate": 0.0003237759755893763,
      "loss": 2.3043,
      "step": 109383
    },
    {
      "epoch": 1.42,
      "grad_norm": 1.0942747592926025,
      "learning_rate": 0.00032377189789180045,
      "loss": 2.3382,
      "step": 109384
    },
    {
      "epoch": 1.42,
      "grad_norm": 1.0634177923202515,
      "learning_rate": 0.000323767820189805,
      "loss": 2.1965,
      "step": 109385
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.9005782604217529,
      "learning_rate": 0.00032376374248339077,
      "loss": 2.1006,
      "step": 109386
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.9038125872612,
      "learning_rate": 0.0003237596647725582,
      "loss": 2.1198,
      "step": 109387
    },
    {
      "epoch": 1.42,
      "grad_norm": 1.2544022798538208,
      "learning_rate": 0.00032375558705730837,
      "loss": 1.9703,
      "step": 109388
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.9266854524612427,
      "learning_rate": 0.00032375150933764194,
      "loss": 1.943,
      "step": 109389
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.9051744341850281,
      "learning_rate": 0.0003237474316135596,
      "loss": 2.2045,
      "step": 109390
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.8520072102546692,
      "learning_rate": 0.00032374335388506224,
      "loss": 2.1817,
      "step": 109391
    },
    {
      "epoch": 1.42,
      "grad_norm": 1.0182667970657349,
      "learning_rate": 0.00032373927615215054,
      "loss": 2.1414,
      "step": 109392
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.9594374895095825,
      "learning_rate": 0.0003237351984148252,
      "loss": 1.9434,
      "step": 109393
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.8655202388763428,
      "learning_rate": 0.0003237311206730871,
      "loss": 2.089,
      "step": 109394
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.9266190528869629,
      "learning_rate": 0.00032372704292693685,
      "loss": 1.9551,
      "step": 109395
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.9129145741462708,
      "learning_rate": 0.0003237229651763754,
      "loss": 2.1659,
      "step": 109396
    },
    {
      "epoch": 1.42,
      "grad_norm": 1.0412496328353882,
      "learning_rate": 0.0003237188874214033,
      "loss": 2.1477,
      "step": 109397
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.9182745218276978,
      "learning_rate": 0.00032371480966202147,
      "loss": 2.1101,
      "step": 109398
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.849502444267273,
      "learning_rate": 0.00032371073189823066,
      "loss": 2.2189,
      "step": 109399
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.9601293802261353,
      "learning_rate": 0.0003237066541300315,
      "loss": 2.0344,
      "step": 109400
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.8898752927780151,
      "learning_rate": 0.00032370257635742477,
      "loss": 2.154,
      "step": 109401
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.85330730676651,
      "learning_rate": 0.0003236984985804114,
      "loss": 2.1958,
      "step": 109402
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.9276201725006104,
      "learning_rate": 0.000323694420798992,
      "loss": 1.9378,
      "step": 109403
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.865428626537323,
      "learning_rate": 0.0003236903430131673,
      "loss": 2.0665,
      "step": 109404
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.8956477642059326,
      "learning_rate": 0.00032368626522293823,
      "loss": 2.2497,
      "step": 109405
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.8523276448249817,
      "learning_rate": 0.0003236821874283053,
      "loss": 1.935,
      "step": 109406
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.8092678189277649,
      "learning_rate": 0.00032367810962926946,
      "loss": 2.1175,
      "step": 109407
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.8702409863471985,
      "learning_rate": 0.0003236740318258315,
      "loss": 2.2726,
      "step": 109408
    },
    {
      "epoch": 1.42,
      "grad_norm": 1.020668625831604,
      "learning_rate": 0.00032366995401799205,
      "loss": 2.1548,
      "step": 109409
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.8957826495170593,
      "learning_rate": 0.00032366587620575183,
      "loss": 2.0501,
      "step": 109410
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.986972451210022,
      "learning_rate": 0.0003236617983891118,
      "loss": 2.1963,
      "step": 109411
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.9167982935905457,
      "learning_rate": 0.0003236577205680725,
      "loss": 2.0636,
      "step": 109412
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.9603719115257263,
      "learning_rate": 0.00032365364274263475,
      "loss": 2.127,
      "step": 109413
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.8844633102416992,
      "learning_rate": 0.00032364956491279945,
      "loss": 2.2108,
      "step": 109414
    },
    {
      "epoch": 1.42,
      "grad_norm": 1.0328868627548218,
      "learning_rate": 0.00032364548707856716,
      "loss": 2.0566,
      "step": 109415
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.9915620684623718,
      "learning_rate": 0.0003236414092399387,
      "loss": 1.9938,
      "step": 109416
    },
    {
      "epoch": 1.42,
      "grad_norm": 1.0148367881774902,
      "learning_rate": 0.000323637331396915,
      "loss": 2.1365,
      "step": 109417
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.9995827674865723,
      "learning_rate": 0.0003236332535494966,
      "loss": 2.5205,
      "step": 109418
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.9719393253326416,
      "learning_rate": 0.0003236291756976843,
      "loss": 2.3956,
      "step": 109419
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.848685622215271,
      "learning_rate": 0.0003236250978414789,
      "loss": 2.1685,
      "step": 109420
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.8893858194351196,
      "learning_rate": 0.0003236210199808812,
      "loss": 2.1403,
      "step": 109421
    },
    {
      "epoch": 1.42,
      "grad_norm": 1.0189064741134644,
      "learning_rate": 0.00032361694211589185,
      "loss": 2.397,
      "step": 109422
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.9669288396835327,
      "learning_rate": 0.0003236128642465117,
      "loss": 2.0281,
      "step": 109423
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.8379732370376587,
      "learning_rate": 0.0003236087863727414,
      "loss": 2.0889,
      "step": 109424
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.9976087212562561,
      "learning_rate": 0.00032360470849458186,
      "loss": 1.9911,
      "step": 109425
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.8336192965507507,
      "learning_rate": 0.0003236006306120338,
      "loss": 2.3328,
      "step": 109426
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.93724524974823,
      "learning_rate": 0.00032359655272509784,
      "loss": 2.1764,
      "step": 109427
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.9911121726036072,
      "learning_rate": 0.0003235924748337748,
      "loss": 2.1738,
      "step": 109428
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.8454456925392151,
      "learning_rate": 0.0003235883969380656,
      "loss": 2.2792,
      "step": 109429
    },
    {
      "epoch": 1.42,
      "grad_norm": 1.0364521741867065,
      "learning_rate": 0.0003235843190379708,
      "loss": 2.0064,
      "step": 109430
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.958997368812561,
      "learning_rate": 0.00032358024113349123,
      "loss": 2.2395,
      "step": 109431
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.9163298010826111,
      "learning_rate": 0.00032357616322462767,
      "loss": 2.2206,
      "step": 109432
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.8933451771736145,
      "learning_rate": 0.0003235720853113809,
      "loss": 2.2688,
      "step": 109433
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.8337783217430115,
      "learning_rate": 0.0003235680073937515,
      "loss": 1.9771,
      "step": 109434
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.9452637434005737,
      "learning_rate": 0.0003235639294717405,
      "loss": 2.0077,
      "step": 109435
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.8693098425865173,
      "learning_rate": 0.00032355985154534846,
      "loss": 2.1142,
      "step": 109436
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.924019992351532,
      "learning_rate": 0.0003235557736145762,
      "loss": 2.1644,
      "step": 109437
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.8280619978904724,
      "learning_rate": 0.00032355169567942443,
      "loss": 1.8119,
      "step": 109438
    },
    {
      "epoch": 1.42,
      "grad_norm": 1.015785574913025,
      "learning_rate": 0.00032354761773989405,
      "loss": 2.0778,
      "step": 109439
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.8665688633918762,
      "learning_rate": 0.0003235435397959857,
      "loss": 1.9835,
      "step": 109440
    },
    {
      "epoch": 1.42,
      "grad_norm": 0.9202423095703125,
      "learning_rate": 0.00032353946184770013,
      "loss": 2.2314,
      "step": 109441
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.9016109108924866,
      "learning_rate": 0.00032353538389503817,
      "loss": 2.4147,
      "step": 109442
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.9652578830718994,
      "learning_rate": 0.0003235313059380005,
      "loss": 2.3885,
      "step": 109443
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.9434787034988403,
      "learning_rate": 0.0003235272279765879,
      "loss": 2.1381,
      "step": 109444
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.8924371600151062,
      "learning_rate": 0.00032352315001080113,
      "loss": 2.0256,
      "step": 109445
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.9906436204910278,
      "learning_rate": 0.000323519072040641,
      "loss": 2.4867,
      "step": 109446
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.9366194605827332,
      "learning_rate": 0.00032351499406610824,
      "loss": 2.3843,
      "step": 109447
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.90850430727005,
      "learning_rate": 0.0003235109160872036,
      "loss": 2.0978,
      "step": 109448
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.958401620388031,
      "learning_rate": 0.0003235068381039278,
      "loss": 2.0598,
      "step": 109449
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.8059225678443909,
      "learning_rate": 0.0003235027601162817,
      "loss": 2.1641,
      "step": 109450
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.9273338317871094,
      "learning_rate": 0.00032349868212426595,
      "loss": 2.0825,
      "step": 109451
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.9614661931991577,
      "learning_rate": 0.0003234946041278813,
      "loss": 2.0915,
      "step": 109452
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.8871712684631348,
      "learning_rate": 0.0003234905261271287,
      "loss": 2.141,
      "step": 109453
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.9829561114311218,
      "learning_rate": 0.0003234864481220087,
      "loss": 2.2825,
      "step": 109454
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.8467950820922852,
      "learning_rate": 0.000323482370112522,
      "loss": 2.1761,
      "step": 109455
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.9575187563896179,
      "learning_rate": 0.00032347829209866967,
      "loss": 2.0318,
      "step": 109456
    },
    {
      "epoch": 1.43,
      "grad_norm": 1.0029003620147705,
      "learning_rate": 0.00032347421408045223,
      "loss": 1.981,
      "step": 109457
    },
    {
      "epoch": 1.43,
      "grad_norm": 1.074833631515503,
      "learning_rate": 0.00032347013605787045,
      "loss": 2.2699,
      "step": 109458
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.9152922630310059,
      "learning_rate": 0.0003234660580309252,
      "loss": 2.3083,
      "step": 109459
    },
    {
      "epoch": 1.43,
      "grad_norm": 1.022769570350647,
      "learning_rate": 0.00032346197999961703,
      "loss": 2.1049,
      "step": 109460
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.9208916425704956,
      "learning_rate": 0.000323457901963947,
      "loss": 1.9806,
      "step": 109461
    },
    {
      "epoch": 1.43,
      "grad_norm": 1.0524144172668457,
      "learning_rate": 0.00032345382392391566,
      "loss": 2.0824,
      "step": 109462
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.9267499446868896,
      "learning_rate": 0.00032344974587952374,
      "loss": 2.2756,
      "step": 109463
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.9341731071472168,
      "learning_rate": 0.000323445667830772,
      "loss": 2.3683,
      "step": 109464
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.9876185655593872,
      "learning_rate": 0.0003234415897776614,
      "loss": 2.2472,
      "step": 109465
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.9217232465744019,
      "learning_rate": 0.0003234375117201926,
      "loss": 2.3399,
      "step": 109466
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.9235213994979858,
      "learning_rate": 0.00032343343365836625,
      "loss": 2.0476,
      "step": 109467
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.9180731773376465,
      "learning_rate": 0.00032342935559218316,
      "loss": 2.2187,
      "step": 109468
    },
    {
      "epoch": 1.43,
      "grad_norm": 1.0605567693710327,
      "learning_rate": 0.00032342527752164413,
      "loss": 2.1186,
      "step": 109469
    },
    {
      "epoch": 1.43,
      "grad_norm": 1.0149152278900146,
      "learning_rate": 0.0003234211994467499,
      "loss": 2.2223,
      "step": 109470
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.9246652126312256,
      "learning_rate": 0.0003234171213675012,
      "loss": 2.1483,
      "step": 109471
    },
    {
      "epoch": 1.43,
      "grad_norm": 1.3177752494812012,
      "learning_rate": 0.00032341304328389897,
      "loss": 2.1943,
      "step": 109472
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.872266948223114,
      "learning_rate": 0.0003234089651959437,
      "loss": 2.1963,
      "step": 109473
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.9187871813774109,
      "learning_rate": 0.00032340488710363616,
      "loss": 2.3603,
      "step": 109474
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.945320725440979,
      "learning_rate": 0.0003234008090069774,
      "loss": 2.2306,
      "step": 109475
    },
    {
      "epoch": 1.43,
      "grad_norm": 1.0071440935134888,
      "learning_rate": 0.00032339673090596785,
      "loss": 2.2925,
      "step": 109476
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.9812726974487305,
      "learning_rate": 0.0003233926528006084,
      "loss": 2.1901,
      "step": 109477
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.8656948208808899,
      "learning_rate": 0.0003233885746909,
      "loss": 2.161,
      "step": 109478
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.8683257102966309,
      "learning_rate": 0.000323384496576843,
      "loss": 2.1948,
      "step": 109479
    },
    {
      "epoch": 1.43,
      "grad_norm": 1.0119503736495972,
      "learning_rate": 0.0003233804184584385,
      "loss": 2.0116,
      "step": 109480
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.9037065505981445,
      "learning_rate": 0.00032337634033568715,
      "loss": 2.1544,
      "step": 109481
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.9346593618392944,
      "learning_rate": 0.0003233722622085896,
      "loss": 2.0684,
      "step": 109482
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.8818958401679993,
      "learning_rate": 0.0003233681840771468,
      "loss": 2.1014,
      "step": 109483
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.8738386631011963,
      "learning_rate": 0.0003233641059413594,
      "loss": 2.0522,
      "step": 109484
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.8849300146102905,
      "learning_rate": 0.00032336002780122814,
      "loss": 2.3012,
      "step": 109485
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.9514132142066956,
      "learning_rate": 0.00032335594965675373,
      "loss": 2.2266,
      "step": 109486
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.9839832782745361,
      "learning_rate": 0.00032335187150793715,
      "loss": 2.1106,
      "step": 109487
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.9079087376594543,
      "learning_rate": 0.000323347793354779,
      "loss": 2.2382,
      "step": 109488
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.907739520072937,
      "learning_rate": 0.00032334371519727995,
      "loss": 2.0751,
      "step": 109489
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.9702982902526855,
      "learning_rate": 0.000323339637035441,
      "loss": 2.2393,
      "step": 109490
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.9757633209228516,
      "learning_rate": 0.0003233355588692626,
      "loss": 2.0725,
      "step": 109491
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.9686340093612671,
      "learning_rate": 0.00032333148069874583,
      "loss": 2.1604,
      "step": 109492
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.9766845703125,
      "learning_rate": 0.00032332740252389127,
      "loss": 2.0256,
      "step": 109493
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.8647180199623108,
      "learning_rate": 0.0003233233243446997,
      "loss": 2.1442,
      "step": 109494
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.9593665599822998,
      "learning_rate": 0.00032331924616117177,
      "loss": 2.1354,
      "step": 109495
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.9134303331375122,
      "learning_rate": 0.0003233151679733085,
      "loss": 2.1879,
      "step": 109496
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.9102267026901245,
      "learning_rate": 0.0003233110897811104,
      "loss": 1.9705,
      "step": 109497
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.8842475414276123,
      "learning_rate": 0.0003233070115845784,
      "loss": 2.0608,
      "step": 109498
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.9588133096694946,
      "learning_rate": 0.0003233029333837132,
      "loss": 2.1498,
      "step": 109499
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.8986556529998779,
      "learning_rate": 0.00032329885517851553,
      "loss": 2.0117,
      "step": 109500
    },
    {
      "epoch": 1.43,
      "grad_norm": 1.2330175638198853,
      "learning_rate": 0.0003232947769689861,
      "loss": 2.1421,
      "step": 109501
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.823262631893158,
      "learning_rate": 0.0003232906987551258,
      "loss": 2.0983,
      "step": 109502
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.7968621253967285,
      "learning_rate": 0.0003232866205369353,
      "loss": 1.8887,
      "step": 109503
    },
    {
      "epoch": 1.43,
      "grad_norm": 1.0012731552124023,
      "learning_rate": 0.0003232825423144153,
      "loss": 2.4243,
      "step": 109504
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.9617233872413635,
      "learning_rate": 0.00032327846408756675,
      "loss": 2.0374,
      "step": 109505
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.8387290239334106,
      "learning_rate": 0.00032327438585639026,
      "loss": 2.1319,
      "step": 109506
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.9197981953620911,
      "learning_rate": 0.00032327030762088657,
      "loss": 2.3951,
      "step": 109507
    },
    {
      "epoch": 1.43,
      "grad_norm": 1.0635956525802612,
      "learning_rate": 0.0003232662293810566,
      "loss": 1.9793,
      "step": 109508
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.824421763420105,
      "learning_rate": 0.0003232621511369009,
      "loss": 2.1531,
      "step": 109509
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.9004408121109009,
      "learning_rate": 0.0003232580728884204,
      "loss": 1.9802,
      "step": 109510
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.9116532206535339,
      "learning_rate": 0.0003232539946356157,
      "loss": 2.3674,
      "step": 109511
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.8189514875411987,
      "learning_rate": 0.0003232499163784877,
      "loss": 2.2302,
      "step": 109512
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.8910608291625977,
      "learning_rate": 0.0003232458381170371,
      "loss": 2.0063,
      "step": 109513
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.9645036458969116,
      "learning_rate": 0.00032324175985126464,
      "loss": 2.2671,
      "step": 109514
    },
    {
      "epoch": 1.43,
      "grad_norm": 1.0912652015686035,
      "learning_rate": 0.0003232376815811711,
      "loss": 2.2065,
      "step": 109515
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.9962239861488342,
      "learning_rate": 0.00032323360330675717,
      "loss": 2.1888,
      "step": 109516
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.8540268540382385,
      "learning_rate": 0.00032322952502802384,
      "loss": 2.0993,
      "step": 109517
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.857398509979248,
      "learning_rate": 0.0003232254467449716,
      "loss": 2.2114,
      "step": 109518
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.9043616056442261,
      "learning_rate": 0.00032322136845760124,
      "loss": 2.0092,
      "step": 109519
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.9688428640365601,
      "learning_rate": 0.00032321729016591375,
      "loss": 1.9412,
      "step": 109520
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.8939171433448792,
      "learning_rate": 0.0003232132118699097,
      "loss": 2.0639,
      "step": 109521
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.954634428024292,
      "learning_rate": 0.0003232091335695897,
      "loss": 1.9977,
      "step": 109522
    },
    {
      "epoch": 1.43,
      "grad_norm": 1.0381687879562378,
      "learning_rate": 0.00032320505526495485,
      "loss": 2.2133,
      "step": 109523
    },
    {
      "epoch": 1.43,
      "grad_norm": 1.3429515361785889,
      "learning_rate": 0.0003232009769560057,
      "loss": 2.1993,
      "step": 109524
    },
    {
      "epoch": 1.43,
      "grad_norm": 1.0477967262268066,
      "learning_rate": 0.000323196898642743,
      "loss": 2.0524,
      "step": 109525
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.8906946778297424,
      "learning_rate": 0.00032319282032516764,
      "loss": 2.1411,
      "step": 109526
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.8636936545372009,
      "learning_rate": 0.0003231887420032803,
      "loss": 2.0392,
      "step": 109527
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.9176242351531982,
      "learning_rate": 0.0003231846636770817,
      "loss": 1.949,
      "step": 109528
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.9601520895957947,
      "learning_rate": 0.00032318058534657254,
      "loss": 2.387,
      "step": 109529
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.9898953437805176,
      "learning_rate": 0.00032317650701175376,
      "loss": 2.0356,
      "step": 109530
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.8604317903518677,
      "learning_rate": 0.0003231724286726261,
      "loss": 1.9493,
      "step": 109531
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.963061511516571,
      "learning_rate": 0.00032316835032919017,
      "loss": 2.1498,
      "step": 109532
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.9039312601089478,
      "learning_rate": 0.00032316427198144683,
      "loss": 2.339,
      "step": 109533
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.942615270614624,
      "learning_rate": 0.0003231601936293968,
      "loss": 2.199,
      "step": 109534
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.9446364045143127,
      "learning_rate": 0.00032315611527304085,
      "loss": 2.1128,
      "step": 109535
    },
    {
      "epoch": 1.43,
      "grad_norm": 1.1186250448226929,
      "learning_rate": 0.0003231520369123797,
      "loss": 2.005,
      "step": 109536
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.9209852814674377,
      "learning_rate": 0.00032314795854741416,
      "loss": 2.2683,
      "step": 109537
    },
    {
      "epoch": 1.43,
      "grad_norm": 1.0788613557815552,
      "learning_rate": 0.00032314388017814507,
      "loss": 2.3205,
      "step": 109538
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.998699963092804,
      "learning_rate": 0.0003231398018045731,
      "loss": 2.0514,
      "step": 109539
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.9127340316772461,
      "learning_rate": 0.00032313572342669886,
      "loss": 2.1226,
      "step": 109540
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.9440438151359558,
      "learning_rate": 0.0003231316450445234,
      "loss": 2.0064,
      "step": 109541
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.8719225525856018,
      "learning_rate": 0.00032312756665804724,
      "loss": 2.3137,
      "step": 109542
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.9210988879203796,
      "learning_rate": 0.00032312348826727126,
      "loss": 2.1419,
      "step": 109543
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.9524054527282715,
      "learning_rate": 0.00032311940987219615,
      "loss": 1.8685,
      "step": 109544
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.9392396211624146,
      "learning_rate": 0.0003231153314728228,
      "loss": 1.9709,
      "step": 109545
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.9248737692832947,
      "learning_rate": 0.0003231112530691518,
      "loss": 2.1535,
      "step": 109546
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.9069942235946655,
      "learning_rate": 0.0003231071746611839,
      "loss": 2.028,
      "step": 109547
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.9852665066719055,
      "learning_rate": 0.00032310309624892017,
      "loss": 2.2239,
      "step": 109548
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.9130972623825073,
      "learning_rate": 0.00032309901783236094,
      "loss": 2.0804,
      "step": 109549
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.9752072095870972,
      "learning_rate": 0.0003230949394115072,
      "loss": 2.2396,
      "step": 109550
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.8624514937400818,
      "learning_rate": 0.0003230908609863598,
      "loss": 1.9795,
      "step": 109551
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.9738919734954834,
      "learning_rate": 0.0003230867825569192,
      "loss": 2.177,
      "step": 109552
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.8916496634483337,
      "learning_rate": 0.0003230827041231864,
      "loss": 2.0245,
      "step": 109553
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.9192136526107788,
      "learning_rate": 0.0003230786256851622,
      "loss": 1.9555,
      "step": 109554
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.891403079032898,
      "learning_rate": 0.00032307454724284703,
      "loss": 2.1047,
      "step": 109555
    },
    {
      "epoch": 1.43,
      "grad_norm": 1.0445753335952759,
      "learning_rate": 0.000323070468796242,
      "loss": 2.2188,
      "step": 109556
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.8822174072265625,
      "learning_rate": 0.0003230663903453478,
      "loss": 2.1375,
      "step": 109557
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.9414072036743164,
      "learning_rate": 0.00032306231189016494,
      "loss": 2.0351,
      "step": 109558
    },
    {
      "epoch": 1.43,
      "grad_norm": 1.0892778635025024,
      "learning_rate": 0.00032305823343069446,
      "loss": 2.3679,
      "step": 109559
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.9413769841194153,
      "learning_rate": 0.0003230541549669371,
      "loss": 2.3834,
      "step": 109560
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.915602445602417,
      "learning_rate": 0.0003230500764988934,
      "loss": 2.3824,
      "step": 109561
    },
    {
      "epoch": 1.43,
      "grad_norm": 1.1442080736160278,
      "learning_rate": 0.0003230459980265643,
      "loss": 2.0718,
      "step": 109562
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.8932153582572937,
      "learning_rate": 0.00032304191954995055,
      "loss": 2.1675,
      "step": 109563
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.9284954071044922,
      "learning_rate": 0.0003230378410690528,
      "loss": 2.168,
      "step": 109564
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.9370249509811401,
      "learning_rate": 0.000323033762583872,
      "loss": 2.354,
      "step": 109565
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.881402313709259,
      "learning_rate": 0.0003230296840944087,
      "loss": 2.2767,
      "step": 109566
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.9475009441375732,
      "learning_rate": 0.00032302560560066375,
      "loss": 2.1099,
      "step": 109567
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.8992804884910583,
      "learning_rate": 0.0003230215271026379,
      "loss": 2.1531,
      "step": 109568
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.8699727654457092,
      "learning_rate": 0.00032301744860033194,
      "loss": 2.1099,
      "step": 109569
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.9680474400520325,
      "learning_rate": 0.00032301337009374664,
      "loss": 2.1763,
      "step": 109570
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.9031559228897095,
      "learning_rate": 0.0003230092915828826,
      "loss": 2.127,
      "step": 109571
    },
    {
      "epoch": 1.43,
      "grad_norm": 1.0280734300613403,
      "learning_rate": 0.0003230052130677409,
      "loss": 2.0315,
      "step": 109572
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.9863491058349609,
      "learning_rate": 0.00032300113454832187,
      "loss": 2.0691,
      "step": 109573
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.9037105441093445,
      "learning_rate": 0.0003229970560246266,
      "loss": 2.1997,
      "step": 109574
    },
    {
      "epoch": 1.43,
      "grad_norm": 1.0049360990524292,
      "learning_rate": 0.0003229929774966558,
      "loss": 2.2552,
      "step": 109575
    },
    {
      "epoch": 1.43,
      "grad_norm": 1.0612258911132812,
      "learning_rate": 0.0003229888989644101,
      "loss": 2.2114,
      "step": 109576
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.9480857253074646,
      "learning_rate": 0.0003229848204278903,
      "loss": 2.1424,
      "step": 109577
    },
    {
      "epoch": 1.43,
      "grad_norm": 1.0055879354476929,
      "learning_rate": 0.0003229807418870973,
      "loss": 1.9548,
      "step": 109578
    },
    {
      "epoch": 1.43,
      "grad_norm": 1.0152968168258667,
      "learning_rate": 0.0003229766633420316,
      "loss": 1.9595,
      "step": 109579
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.9607505202293396,
      "learning_rate": 0.0003229725847926942,
      "loss": 2.1945,
      "step": 109580
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.9299020171165466,
      "learning_rate": 0.00032296850623908573,
      "loss": 2.3168,
      "step": 109581
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.9450481534004211,
      "learning_rate": 0.000322964427681207,
      "loss": 2.0936,
      "step": 109582
    },
    {
      "epoch": 1.43,
      "grad_norm": 1.042970895767212,
      "learning_rate": 0.0003229603491190587,
      "loss": 2.0204,
      "step": 109583
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.9201440215110779,
      "learning_rate": 0.0003229562705526417,
      "loss": 2.0944,
      "step": 109584
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.9062198400497437,
      "learning_rate": 0.00032295219198195664,
      "loss": 2.0805,
      "step": 109585
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.9197188019752502,
      "learning_rate": 0.0003229481134070044,
      "loss": 2.1904,
      "step": 109586
    },
    {
      "epoch": 1.43,
      "grad_norm": 1.1086862087249756,
      "learning_rate": 0.00032294403482778565,
      "loss": 2.28,
      "step": 109587
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.9727179408073425,
      "learning_rate": 0.0003229399562443012,
      "loss": 2.2038,
      "step": 109588
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.862434983253479,
      "learning_rate": 0.0003229358776565517,
      "loss": 2.1407,
      "step": 109589
    },
    {
      "epoch": 1.43,
      "grad_norm": 1.1819311380386353,
      "learning_rate": 0.00032293179906453805,
      "loss": 2.1796,
      "step": 109590
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.9470458030700684,
      "learning_rate": 0.00032292772046826093,
      "loss": 2.2535,
      "step": 109591
    },
    {
      "epoch": 1.43,
      "grad_norm": 1.0250869989395142,
      "learning_rate": 0.0003229236418677211,
      "loss": 2.0454,
      "step": 109592
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.8836197853088379,
      "learning_rate": 0.00032291956326291934,
      "loss": 2.2403,
      "step": 109593
    },
    {
      "epoch": 1.43,
      "grad_norm": 1.057202935218811,
      "learning_rate": 0.0003229154846538564,
      "loss": 2.0048,
      "step": 109594
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.9170026779174805,
      "learning_rate": 0.00032291140604053306,
      "loss": 2.052,
      "step": 109595
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.9257073402404785,
      "learning_rate": 0.00032290732742295007,
      "loss": 2.0936,
      "step": 109596
    },
    {
      "epoch": 1.43,
      "grad_norm": 1.0880000591278076,
      "learning_rate": 0.0003229032488011081,
      "loss": 2.0771,
      "step": 109597
    },
    {
      "epoch": 1.43,
      "grad_norm": 1.0267677307128906,
      "learning_rate": 0.000322899170175008,
      "loss": 2.1609,
      "step": 109598
    },
    {
      "epoch": 1.43,
      "grad_norm": 1.3110305070877075,
      "learning_rate": 0.00032289509154465057,
      "loss": 2.2585,
      "step": 109599
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.8803179264068604,
      "learning_rate": 0.0003228910129100365,
      "loss": 2.2336,
      "step": 109600
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.9784607887268066,
      "learning_rate": 0.00032288693427116655,
      "loss": 2.0845,
      "step": 109601
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.8899387717247009,
      "learning_rate": 0.00032288285562804143,
      "loss": 2.1959,
      "step": 109602
    },
    {
      "epoch": 1.43,
      "grad_norm": 1.0613142251968384,
      "learning_rate": 0.00032287877698066196,
      "loss": 2.0981,
      "step": 109603
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.852864146232605,
      "learning_rate": 0.00032287469832902895,
      "loss": 2.1033,
      "step": 109604
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.8828626871109009,
      "learning_rate": 0.00032287061967314304,
      "loss": 2.1477,
      "step": 109605
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.96168452501297,
      "learning_rate": 0.00032286654101300517,
      "loss": 2.1827,
      "step": 109606
    },
    {
      "epoch": 1.43,
      "grad_norm": 1.0495173931121826,
      "learning_rate": 0.00032286246234861587,
      "loss": 2.0816,
      "step": 109607
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.8976017236709595,
      "learning_rate": 0.000322858383679976,
      "loss": 2.128,
      "step": 109608
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.9062225222587585,
      "learning_rate": 0.00032285430500708645,
      "loss": 2.1123,
      "step": 109609
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.8794295191764832,
      "learning_rate": 0.00032285022632994774,
      "loss": 2.098,
      "step": 109610
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.9214357137680054,
      "learning_rate": 0.00032284614764856075,
      "loss": 2.0752,
      "step": 109611
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.9560592174530029,
      "learning_rate": 0.0003228420689629263,
      "loss": 2.1127,
      "step": 109612
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.9027817845344543,
      "learning_rate": 0.000322837990273045,
      "loss": 2.1893,
      "step": 109613
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.9116343855857849,
      "learning_rate": 0.0003228339115789177,
      "loss": 2.0255,
      "step": 109614
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.8630028367042542,
      "learning_rate": 0.0003228298328805452,
      "loss": 1.9667,
      "step": 109615
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.8855759501457214,
      "learning_rate": 0.0003228257541779281,
      "loss": 2.1356,
      "step": 109616
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.9027855396270752,
      "learning_rate": 0.00032282167547106736,
      "loss": 2.0741,
      "step": 109617
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.8598947525024414,
      "learning_rate": 0.00032281759675996353,
      "loss": 2.2111,
      "step": 109618
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.9334117770195007,
      "learning_rate": 0.0003228135180446176,
      "loss": 1.956,
      "step": 109619
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.9974979758262634,
      "learning_rate": 0.00032280943932503015,
      "loss": 2.1475,
      "step": 109620
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.931894063949585,
      "learning_rate": 0.000322805360601202,
      "loss": 2.0197,
      "step": 109621
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.9773574471473694,
      "learning_rate": 0.0003228012818731339,
      "loss": 2.3238,
      "step": 109622
    },
    {
      "epoch": 1.43,
      "grad_norm": 1.0091410875320435,
      "learning_rate": 0.0003227972031408267,
      "loss": 2.2479,
      "step": 109623
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.8945237398147583,
      "learning_rate": 0.0003227931244042809,
      "loss": 2.1935,
      "step": 109624
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.8743345141410828,
      "learning_rate": 0.0003227890456634975,
      "loss": 2.2506,
      "step": 109625
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.8456850647926331,
      "learning_rate": 0.00032278496691847726,
      "loss": 2.0403,
      "step": 109626
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.983665406703949,
      "learning_rate": 0.0003227808881692208,
      "loss": 2.1946,
      "step": 109627
    },
    {
      "epoch": 1.43,
      "grad_norm": 1.351397156715393,
      "learning_rate": 0.0003227768094157289,
      "loss": 2.172,
      "step": 109628
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.9216736555099487,
      "learning_rate": 0.0003227727306580024,
      "loss": 2.0806,
      "step": 109629
    },
    {
      "epoch": 1.43,
      "grad_norm": 1.098919153213501,
      "learning_rate": 0.00032276865189604205,
      "loss": 2.1819,
      "step": 109630
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.9908563494682312,
      "learning_rate": 0.00032276457312984855,
      "loss": 2.291,
      "step": 109631
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.8781213164329529,
      "learning_rate": 0.00032276049435942265,
      "loss": 2.032,
      "step": 109632
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.9884200692176819,
      "learning_rate": 0.0003227564155847652,
      "loss": 2.0561,
      "step": 109633
    },
    {
      "epoch": 1.43,
      "grad_norm": 1.0494626760482788,
      "learning_rate": 0.0003227523368058769,
      "loss": 2.1609,
      "step": 109634
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.9875375628471375,
      "learning_rate": 0.00032274825802275847,
      "loss": 2.0721,
      "step": 109635
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.8529254198074341,
      "learning_rate": 0.0003227441792354108,
      "loss": 2.2422,
      "step": 109636
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.8868149518966675,
      "learning_rate": 0.00032274010044383453,
      "loss": 2.1999,
      "step": 109637
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.8896933794021606,
      "learning_rate": 0.00032273602164803033,
      "loss": 2.1352,
      "step": 109638
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.964096188545227,
      "learning_rate": 0.0003227319428479992,
      "loss": 2.2278,
      "step": 109639
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.9408670663833618,
      "learning_rate": 0.00032272786404374175,
      "loss": 2.4191,
      "step": 109640
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.9526143670082092,
      "learning_rate": 0.0003227237852352587,
      "loss": 2.1026,
      "step": 109641
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.9866341352462769,
      "learning_rate": 0.000322719706422551,
      "loss": 2.1297,
      "step": 109642
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.9190725088119507,
      "learning_rate": 0.00032271562760561917,
      "loss": 2.1131,
      "step": 109643
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.9660619497299194,
      "learning_rate": 0.0003227115487844641,
      "loss": 2.1494,
      "step": 109644
    },
    {
      "epoch": 1.43,
      "grad_norm": 1.0857298374176025,
      "learning_rate": 0.0003227074699590865,
      "loss": 2.1207,
      "step": 109645
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.9386284351348877,
      "learning_rate": 0.0003227033911294872,
      "loss": 2.0546,
      "step": 109646
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.9473342895507812,
      "learning_rate": 0.0003226993122956668,
      "loss": 2.0399,
      "step": 109647
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.8747634291648865,
      "learning_rate": 0.00032269523345762634,
      "loss": 2.0115,
      "step": 109648
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.8822370767593384,
      "learning_rate": 0.00032269115461536625,
      "loss": 2.2005,
      "step": 109649
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.9019580483436584,
      "learning_rate": 0.0003226870757688875,
      "loss": 2.0654,
      "step": 109650
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.9451156258583069,
      "learning_rate": 0.0003226829969181909,
      "loss": 2.0952,
      "step": 109651
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.9228783845901489,
      "learning_rate": 0.000322678918063277,
      "loss": 2.1323,
      "step": 109652
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.8576143383979797,
      "learning_rate": 0.0003226748392041466,
      "loss": 2.1192,
      "step": 109653
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.8985827565193176,
      "learning_rate": 0.0003226707603408007,
      "loss": 2.1028,
      "step": 109654
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.9000029563903809,
      "learning_rate": 0.00032266668147323975,
      "loss": 2.1497,
      "step": 109655
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.9897325038909912,
      "learning_rate": 0.00032266260260146464,
      "loss": 2.183,
      "step": 109656
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.8644812107086182,
      "learning_rate": 0.0003226585237254762,
      "loss": 2.1723,
      "step": 109657
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.9195541143417358,
      "learning_rate": 0.000322654444845275,
      "loss": 2.0692,
      "step": 109658
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.9269863963127136,
      "learning_rate": 0.00032265036596086194,
      "loss": 2.1626,
      "step": 109659
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.8874879479408264,
      "learning_rate": 0.00032264628707223783,
      "loss": 2.1203,
      "step": 109660
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.9964756369590759,
      "learning_rate": 0.0003226422081794033,
      "loss": 1.9097,
      "step": 109661
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.8633010983467102,
      "learning_rate": 0.0003226381292823591,
      "loss": 2.2269,
      "step": 109662
    },
    {
      "epoch": 1.43,
      "grad_norm": 1.0528374910354614,
      "learning_rate": 0.0003226340503811061,
      "loss": 1.9618,
      "step": 109663
    },
    {
      "epoch": 1.43,
      "grad_norm": 1.1057701110839844,
      "learning_rate": 0.000322629971475645,
      "loss": 2.3429,
      "step": 109664
    },
    {
      "epoch": 1.43,
      "grad_norm": 1.0553468465805054,
      "learning_rate": 0.00032262589256597653,
      "loss": 2.2307,
      "step": 109665
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.9646596312522888,
      "learning_rate": 0.0003226218136521015,
      "loss": 1.8924,
      "step": 109666
    },
    {
      "epoch": 1.43,
      "grad_norm": 1.2278938293457031,
      "learning_rate": 0.00032261773473402067,
      "loss": 2.2189,
      "step": 109667
    },
    {
      "epoch": 1.43,
      "grad_norm": 1.0900812149047852,
      "learning_rate": 0.00032261365581173475,
      "loss": 2.0498,
      "step": 109668
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.8229725360870361,
      "learning_rate": 0.0003226095768852445,
      "loss": 2.0074,
      "step": 109669
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.9869776964187622,
      "learning_rate": 0.00032260549795455086,
      "loss": 2.0156,
      "step": 109670
    },
    {
      "epoch": 1.43,
      "grad_norm": 1.1698904037475586,
      "learning_rate": 0.00032260141901965426,
      "loss": 2.115,
      "step": 109671
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.8845454454421997,
      "learning_rate": 0.00032259734008055566,
      "loss": 1.9537,
      "step": 109672
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.9586138725280762,
      "learning_rate": 0.00032259326113725584,
      "loss": 2.1805,
      "step": 109673
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.8858553767204285,
      "learning_rate": 0.00032258918218975544,
      "loss": 2.0884,
      "step": 109674
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.8633840680122375,
      "learning_rate": 0.0003225851032380553,
      "loss": 2.2024,
      "step": 109675
    },
    {
      "epoch": 1.43,
      "grad_norm": 1.087240219116211,
      "learning_rate": 0.00032258102428215626,
      "loss": 2.1605,
      "step": 109676
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.7882654666900635,
      "learning_rate": 0.00032257694532205884,
      "loss": 2.0971,
      "step": 109677
    },
    {
      "epoch": 1.43,
      "grad_norm": 1.2842576503753662,
      "learning_rate": 0.00032257286635776404,
      "loss": 2.185,
      "step": 109678
    },
    {
      "epoch": 1.43,
      "grad_norm": 1.0565003156661987,
      "learning_rate": 0.00032256878738927245,
      "loss": 2.0459,
      "step": 109679
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.9243922829627991,
      "learning_rate": 0.0003225647084165849,
      "loss": 2.016,
      "step": 109680
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.9739524722099304,
      "learning_rate": 0.0003225606294397022,
      "loss": 2.146,
      "step": 109681
    },
    {
      "epoch": 1.43,
      "grad_norm": 1.249833106994629,
      "learning_rate": 0.00032255655045862507,
      "loss": 2.2705,
      "step": 109682
    },
    {
      "epoch": 1.43,
      "grad_norm": 1.043717384338379,
      "learning_rate": 0.0003225524714733542,
      "loss": 2.1347,
      "step": 109683
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.9728503823280334,
      "learning_rate": 0.0003225483924838903,
      "loss": 2.0187,
      "step": 109684
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.9000105261802673,
      "learning_rate": 0.00032254431349023444,
      "loss": 2.047,
      "step": 109685
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.8310726881027222,
      "learning_rate": 0.000322540234492387,
      "loss": 2.1682,
      "step": 109686
    },
    {
      "epoch": 1.43,
      "grad_norm": 1.052108645439148,
      "learning_rate": 0.000322536155490349,
      "loss": 2.2796,
      "step": 109687
    },
    {
      "epoch": 1.43,
      "grad_norm": 1.094536542892456,
      "learning_rate": 0.00032253207648412106,
      "loss": 2.0848,
      "step": 109688
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.8630480766296387,
      "learning_rate": 0.000322527997473704,
      "loss": 2.0491,
      "step": 109689
    },
    {
      "epoch": 1.43,
      "grad_norm": 1.0019711256027222,
      "learning_rate": 0.00032252391845909856,
      "loss": 2.1996,
      "step": 109690
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.9576636552810669,
      "learning_rate": 0.00032251983944030547,
      "loss": 2.0242,
      "step": 109691
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.9671276211738586,
      "learning_rate": 0.00032251576041732557,
      "loss": 1.9939,
      "step": 109692
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.8669222593307495,
      "learning_rate": 0.00032251168139015945,
      "loss": 2.0587,
      "step": 109693
    },
    {
      "epoch": 1.43,
      "grad_norm": 1.0248280763626099,
      "learning_rate": 0.00032250760235880816,
      "loss": 2.0988,
      "step": 109694
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.9238118529319763,
      "learning_rate": 0.00032250352332327217,
      "loss": 2.0482,
      "step": 109695
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.8565703630447388,
      "learning_rate": 0.0003224994442835523,
      "loss": 1.8629,
      "step": 109696
    },
    {
      "epoch": 1.43,
      "grad_norm": 1.0532984733581543,
      "learning_rate": 0.0003224953652396495,
      "loss": 2.2039,
      "step": 109697
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.8600922226905823,
      "learning_rate": 0.00032249128619156426,
      "loss": 2.175,
      "step": 109698
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.9688898921012878,
      "learning_rate": 0.0003224872071392975,
      "loss": 2.4244,
      "step": 109699
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.9685349464416504,
      "learning_rate": 0.00032248312808285,
      "loss": 1.8636,
      "step": 109700
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.925595760345459,
      "learning_rate": 0.00032247904902222245,
      "loss": 2.074,
      "step": 109701
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.9202814698219299,
      "learning_rate": 0.00032247496995741555,
      "loss": 1.8269,
      "step": 109702
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.9068545699119568,
      "learning_rate": 0.0003224708908884302,
      "loss": 2.0679,
      "step": 109703
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.8829591274261475,
      "learning_rate": 0.00032246681181526707,
      "loss": 2.1062,
      "step": 109704
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.9679060578346252,
      "learning_rate": 0.0003224627327379269,
      "loss": 2.0948,
      "step": 109705
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.8963397145271301,
      "learning_rate": 0.0003224586536564105,
      "loss": 2.0402,
      "step": 109706
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.8432238101959229,
      "learning_rate": 0.00032245457457071865,
      "loss": 2.238,
      "step": 109707
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.9346317052841187,
      "learning_rate": 0.00032245049548085204,
      "loss": 2.2505,
      "step": 109708
    },
    {
      "epoch": 1.43,
      "grad_norm": 1.1655690670013428,
      "learning_rate": 0.0003224464163868114,
      "loss": 2.3941,
      "step": 109709
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.9096105098724365,
      "learning_rate": 0.00032244233728859765,
      "loss": 1.986,
      "step": 109710
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.8841282725334167,
      "learning_rate": 0.00032243825818621143,
      "loss": 1.9998,
      "step": 109711
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.9622456431388855,
      "learning_rate": 0.0003224341790796534,
      "loss": 2.0167,
      "step": 109712
    },
    {
      "epoch": 1.43,
      "grad_norm": 1.1884703636169434,
      "learning_rate": 0.0003224300999689246,
      "loss": 2.194,
      "step": 109713
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.9190611839294434,
      "learning_rate": 0.0003224260208540255,
      "loss": 2.0666,
      "step": 109714
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.8755582571029663,
      "learning_rate": 0.00032242194173495703,
      "loss": 2.1549,
      "step": 109715
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.8378131985664368,
      "learning_rate": 0.0003224178626117199,
      "loss": 2.1104,
      "step": 109716
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.9351481795310974,
      "learning_rate": 0.0003224137834843148,
      "loss": 1.9886,
      "step": 109717
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.8528518080711365,
      "learning_rate": 0.0003224097043527427,
      "loss": 2.061,
      "step": 109718
    },
    {
      "epoch": 1.43,
      "grad_norm": 1.0976969003677368,
      "learning_rate": 0.00032240562521700404,
      "loss": 2.3327,
      "step": 109719
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.930842936038971,
      "learning_rate": 0.00032240154607709987,
      "loss": 2.1843,
      "step": 109720
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.8303395509719849,
      "learning_rate": 0.0003223974669330308,
      "loss": 2.1564,
      "step": 109721
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.8772109746932983,
      "learning_rate": 0.0003223933877847976,
      "loss": 2.4033,
      "step": 109722
    },
    {
      "epoch": 1.43,
      "grad_norm": 1.0671236515045166,
      "learning_rate": 0.00032238930863240107,
      "loss": 2.1544,
      "step": 109723
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.9415106177330017,
      "learning_rate": 0.00032238522947584193,
      "loss": 2.3139,
      "step": 109724
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.9270569682121277,
      "learning_rate": 0.0003223811503151209,
      "loss": 2.3387,
      "step": 109725
    },
    {
      "epoch": 1.43,
      "grad_norm": 1.022836685180664,
      "learning_rate": 0.00032237707115023885,
      "loss": 2.1786,
      "step": 109726
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.9541391730308533,
      "learning_rate": 0.0003223729919811966,
      "loss": 2.0531,
      "step": 109727
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.8366339206695557,
      "learning_rate": 0.0003223689128079946,
      "loss": 2.0833,
      "step": 109728
    },
    {
      "epoch": 1.43,
      "grad_norm": 1.1063218116760254,
      "learning_rate": 0.0003223648336306339,
      "loss": 2.0343,
      "step": 109729
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.969607949256897,
      "learning_rate": 0.0003223607544491151,
      "loss": 2.1651,
      "step": 109730
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.9469513297080994,
      "learning_rate": 0.00032235667526343907,
      "loss": 2.2405,
      "step": 109731
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.9754931330680847,
      "learning_rate": 0.0003223525960736064,
      "loss": 2.3562,
      "step": 109732
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.820555567741394,
      "learning_rate": 0.000322348516879618,
      "loss": 2.0859,
      "step": 109733
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.9362626671791077,
      "learning_rate": 0.0003223444376814747,
      "loss": 2.3457,
      "step": 109734
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.8628576993942261,
      "learning_rate": 0.00032234035847917707,
      "loss": 2.1956,
      "step": 109735
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.8314915299415588,
      "learning_rate": 0.0003223362792727259,
      "loss": 2.2239,
      "step": 109736
    },
    {
      "epoch": 1.43,
      "grad_norm": 1.0054124593734741,
      "learning_rate": 0.0003223322000621221,
      "loss": 1.9967,
      "step": 109737
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.846580445766449,
      "learning_rate": 0.0003223281208473662,
      "loss": 2.11,
      "step": 109738
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.9572007656097412,
      "learning_rate": 0.00032232404162845917,
      "loss": 2.1711,
      "step": 109739
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.8620874285697937,
      "learning_rate": 0.0003223199624054017,
      "loss": 1.9943,
      "step": 109740
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.9528442621231079,
      "learning_rate": 0.00032231588317819447,
      "loss": 2.1621,
      "step": 109741
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.9827153086662292,
      "learning_rate": 0.0003223118039468383,
      "loss": 2.0881,
      "step": 109742
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.9072232246398926,
      "learning_rate": 0.000322307724711334,
      "loss": 2.1108,
      "step": 109743
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.9091821908950806,
      "learning_rate": 0.00032230364547168215,
      "loss": 1.8853,
      "step": 109744
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.9602832198143005,
      "learning_rate": 0.0003222995662278837,
      "loss": 1.9802,
      "step": 109745
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.999136745929718,
      "learning_rate": 0.0003222954869799394,
      "loss": 1.9782,
      "step": 109746
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.9032147526741028,
      "learning_rate": 0.00032229140772784985,
      "loss": 2.2575,
      "step": 109747
    },
    {
      "epoch": 1.43,
      "grad_norm": 1.0695044994354248,
      "learning_rate": 0.00032228732847161594,
      "loss": 2.0957,
      "step": 109748
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.9287453293800354,
      "learning_rate": 0.00032228324921123843,
      "loss": 2.2378,
      "step": 109749
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.9356290698051453,
      "learning_rate": 0.000322279169946718,
      "loss": 2.1294,
      "step": 109750
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.8883097171783447,
      "learning_rate": 0.00032227509067805546,
      "loss": 2.2998,
      "step": 109751
    },
    {
      "epoch": 1.43,
      "grad_norm": 1.1310029029846191,
      "learning_rate": 0.0003222710114052516,
      "loss": 2.1119,
      "step": 109752
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.897055447101593,
      "learning_rate": 0.0003222669321283071,
      "loss": 2.3394,
      "step": 109753
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.8809677958488464,
      "learning_rate": 0.0003222628528472228,
      "loss": 2.2479,
      "step": 109754
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.9900639653205872,
      "learning_rate": 0.00032225877356199936,
      "loss": 2.246,
      "step": 109755
    },
    {
      "epoch": 1.43,
      "grad_norm": 1.188435435295105,
      "learning_rate": 0.00032225469427263765,
      "loss": 1.9366,
      "step": 109756
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.8688429594039917,
      "learning_rate": 0.00032225061497913825,
      "loss": 2.4839,
      "step": 109757
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.7979639172554016,
      "learning_rate": 0.0003222465356815022,
      "loss": 2.1178,
      "step": 109758
    },
    {
      "epoch": 1.43,
      "grad_norm": 1.0139741897583008,
      "learning_rate": 0.00032224245637973,
      "loss": 2.1578,
      "step": 109759
    },
    {
      "epoch": 1.43,
      "grad_norm": 1.1753052473068237,
      "learning_rate": 0.00032223837707382255,
      "loss": 2.0408,
      "step": 109760
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.8240079283714294,
      "learning_rate": 0.00032223429776378056,
      "loss": 2.191,
      "step": 109761
    },
    {
      "epoch": 1.43,
      "grad_norm": 1.0501306056976318,
      "learning_rate": 0.00032223021844960477,
      "loss": 2.0904,
      "step": 109762
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.9879240393638611,
      "learning_rate": 0.000322226139131296,
      "loss": 2.0482,
      "step": 109763
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.9217020869255066,
      "learning_rate": 0.000322222059808855,
      "loss": 2.1767,
      "step": 109764
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.8892016410827637,
      "learning_rate": 0.0003222179804822824,
      "loss": 2.1978,
      "step": 109765
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.8537159562110901,
      "learning_rate": 0.0003222139011515791,
      "loss": 1.9479,
      "step": 109766
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.9123920202255249,
      "learning_rate": 0.00032220982181674584,
      "loss": 2.0601,
      "step": 109767
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.884421169757843,
      "learning_rate": 0.00032220574247778335,
      "loss": 2.081,
      "step": 109768
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.9323121905326843,
      "learning_rate": 0.0003222016631346924,
      "loss": 2.1738,
      "step": 109769
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.8813332915306091,
      "learning_rate": 0.00032219758378747374,
      "loss": 2.1165,
      "step": 109770
    },
    {
      "epoch": 1.43,
      "grad_norm": 1.2010657787322998,
      "learning_rate": 0.0003221935044361281,
      "loss": 2.1208,
      "step": 109771
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.9526689052581787,
      "learning_rate": 0.0003221894250806562,
      "loss": 1.9077,
      "step": 109772
    },
    {
      "epoch": 1.43,
      "grad_norm": 1.0437934398651123,
      "learning_rate": 0.000322185345721059,
      "loss": 2.161,
      "step": 109773
    },
    {
      "epoch": 1.43,
      "grad_norm": 1.0926905870437622,
      "learning_rate": 0.00032218126635733706,
      "loss": 2.2909,
      "step": 109774
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.8303017616271973,
      "learning_rate": 0.0003221771869894912,
      "loss": 1.8135,
      "step": 109775
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.8550114035606384,
      "learning_rate": 0.00032217310761752224,
      "loss": 2.014,
      "step": 109776
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.9589833617210388,
      "learning_rate": 0.00032216902824143087,
      "loss": 2.1404,
      "step": 109777
    },
    {
      "epoch": 1.43,
      "grad_norm": 1.0389302968978882,
      "learning_rate": 0.00032216494886121773,
      "loss": 2.111,
      "step": 109778
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.9822752475738525,
      "learning_rate": 0.0003221608694768839,
      "loss": 2.1533,
      "step": 109779
    },
    {
      "epoch": 1.43,
      "grad_norm": 1.0138369798660278,
      "learning_rate": 0.00032215679008842983,
      "loss": 2.2023,
      "step": 109780
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.9327496290206909,
      "learning_rate": 0.00032215271069585636,
      "loss": 2.0922,
      "step": 109781
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.9630519151687622,
      "learning_rate": 0.00032214863129916444,
      "loss": 2.0207,
      "step": 109782
    },
    {
      "epoch": 1.43,
      "grad_norm": 1.4647759199142456,
      "learning_rate": 0.00032214455189835456,
      "loss": 2.3756,
      "step": 109783
    },
    {
      "epoch": 1.43,
      "grad_norm": 1.1205980777740479,
      "learning_rate": 0.0003221404724934276,
      "loss": 2.3883,
      "step": 109784
    },
    {
      "epoch": 1.43,
      "grad_norm": 1.0247634649276733,
      "learning_rate": 0.00032213639308438437,
      "loss": 2.2691,
      "step": 109785
    },
    {
      "epoch": 1.43,
      "grad_norm": 1.0437086820602417,
      "learning_rate": 0.0003221323136712254,
      "loss": 2.097,
      "step": 109786
    },
    {
      "epoch": 1.43,
      "grad_norm": 1.1002494096755981,
      "learning_rate": 0.00032212823425395177,
      "loss": 2.0835,
      "step": 109787
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.9477939009666443,
      "learning_rate": 0.00032212415483256405,
      "loss": 2.306,
      "step": 109788
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.8908786177635193,
      "learning_rate": 0.000322120075407063,
      "loss": 2.1977,
      "step": 109789
    },
    {
      "epoch": 1.43,
      "grad_norm": 1.0970063209533691,
      "learning_rate": 0.00032211599597744943,
      "loss": 2.0664,
      "step": 109790
    },
    {
      "epoch": 1.43,
      "grad_norm": 1.0438706874847412,
      "learning_rate": 0.00032211191654372403,
      "loss": 1.9773,
      "step": 109791
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.9892465472221375,
      "learning_rate": 0.0003221078371058877,
      "loss": 2.137,
      "step": 109792
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.9013712406158447,
      "learning_rate": 0.0003221037576639411,
      "loss": 2.2861,
      "step": 109793
    },
    {
      "epoch": 1.43,
      "grad_norm": 1.135372281074524,
      "learning_rate": 0.00032209967821788486,
      "loss": 2.0941,
      "step": 109794
    },
    {
      "epoch": 1.43,
      "grad_norm": 1.3656998872756958,
      "learning_rate": 0.00032209559876772,
      "loss": 2.1845,
      "step": 109795
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.9259212613105774,
      "learning_rate": 0.0003220915193134471,
      "loss": 2.0705,
      "step": 109796
    },
    {
      "epoch": 1.43,
      "grad_norm": 1.0594546794891357,
      "learning_rate": 0.000322087439855067,
      "loss": 2.1263,
      "step": 109797
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.9812197089195251,
      "learning_rate": 0.0003220833603925804,
      "loss": 2.1788,
      "step": 109798
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.9606047868728638,
      "learning_rate": 0.0003220792809259881,
      "loss": 1.9805,
      "step": 109799
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.8608390092849731,
      "learning_rate": 0.0003220752014552909,
      "loss": 1.9507,
      "step": 109800
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.9782217144966125,
      "learning_rate": 0.0003220711219804894,
      "loss": 2.3576,
      "step": 109801
    },
    {
      "epoch": 1.43,
      "grad_norm": 1.0615065097808838,
      "learning_rate": 0.0003220670425015845,
      "loss": 2.1305,
      "step": 109802
    },
    {
      "epoch": 1.43,
      "grad_norm": 1.0114645957946777,
      "learning_rate": 0.00032206296301857696,
      "loss": 2.0406,
      "step": 109803
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.9458909034729004,
      "learning_rate": 0.00032205888353146746,
      "loss": 2.3378,
      "step": 109804
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.9841157793998718,
      "learning_rate": 0.0003220548040402568,
      "loss": 2.0317,
      "step": 109805
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.9386553168296814,
      "learning_rate": 0.0003220507245449458,
      "loss": 2.2408,
      "step": 109806
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.9633336663246155,
      "learning_rate": 0.00032204664504553504,
      "loss": 2.0977,
      "step": 109807
    },
    {
      "epoch": 1.43,
      "grad_norm": 1.073874592781067,
      "learning_rate": 0.00032204256554202546,
      "loss": 2.1603,
      "step": 109808
    },
    {
      "epoch": 1.43,
      "grad_norm": 1.3560212850570679,
      "learning_rate": 0.0003220384860344177,
      "loss": 2.1959,
      "step": 109809
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.903674304485321,
      "learning_rate": 0.00032203440652271264,
      "loss": 2.1439,
      "step": 109810
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.9037156105041504,
      "learning_rate": 0.000322030327006911,
      "loss": 2.3137,
      "step": 109811
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.955091118812561,
      "learning_rate": 0.0003220262474870134,
      "loss": 1.9324,
      "step": 109812
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.9413673877716064,
      "learning_rate": 0.00032202216796302076,
      "loss": 2.1171,
      "step": 109813
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.9340925216674805,
      "learning_rate": 0.0003220180884349338,
      "loss": 2.1801,
      "step": 109814
    },
    {
      "epoch": 1.43,
      "grad_norm": 1.0211272239685059,
      "learning_rate": 0.0003220140089027532,
      "loss": 2.0886,
      "step": 109815
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.927823007106781,
      "learning_rate": 0.0003220099293664798,
      "loss": 2.0307,
      "step": 109816
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.9588504433631897,
      "learning_rate": 0.0003220058498261144,
      "loss": 2.1267,
      "step": 109817
    },
    {
      "epoch": 1.43,
      "grad_norm": 1.0325355529785156,
      "learning_rate": 0.0003220017702816576,
      "loss": 2.0531,
      "step": 109818
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.8400628566741943,
      "learning_rate": 0.0003219976907331104,
      "loss": 2.2176,
      "step": 109819
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.9142181277275085,
      "learning_rate": 0.0003219936111804733,
      "loss": 2.2908,
      "step": 109820
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.8736866116523743,
      "learning_rate": 0.0003219895316237471,
      "loss": 1.988,
      "step": 109821
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.976536750793457,
      "learning_rate": 0.0003219854520629328,
      "loss": 2.261,
      "step": 109822
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.8773028254508972,
      "learning_rate": 0.0003219813724980309,
      "loss": 2.1935,
      "step": 109823
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.8971531391143799,
      "learning_rate": 0.0003219772929290422,
      "loss": 2.1253,
      "step": 109824
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.9307925701141357,
      "learning_rate": 0.0003219732133559677,
      "loss": 2.06,
      "step": 109825
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.870627760887146,
      "learning_rate": 0.00032196913377880777,
      "loss": 2.1864,
      "step": 109826
    },
    {
      "epoch": 1.43,
      "grad_norm": 1.0455827713012695,
      "learning_rate": 0.00032196505419756344,
      "loss": 2.206,
      "step": 109827
    },
    {
      "epoch": 1.43,
      "grad_norm": 1.120302438735962,
      "learning_rate": 0.0003219609746122354,
      "loss": 2.0864,
      "step": 109828
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.9681135416030884,
      "learning_rate": 0.00032195689502282433,
      "loss": 2.0633,
      "step": 109829
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.9631884098052979,
      "learning_rate": 0.00032195281542933107,
      "loss": 2.039,
      "step": 109830
    },
    {
      "epoch": 1.43,
      "grad_norm": 1.0667039155960083,
      "learning_rate": 0.0003219487358317565,
      "loss": 2.013,
      "step": 109831
    },
    {
      "epoch": 1.43,
      "grad_norm": 1.1215543746948242,
      "learning_rate": 0.0003219446562301011,
      "loss": 2.123,
      "step": 109832
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.8825501203536987,
      "learning_rate": 0.0003219405766243658,
      "loss": 2.0519,
      "step": 109833
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.8967332243919373,
      "learning_rate": 0.00032193649701455147,
      "loss": 2.1499,
      "step": 109834
    },
    {
      "epoch": 1.43,
      "grad_norm": 1.0042030811309814,
      "learning_rate": 0.00032193241740065856,
      "loss": 2.024,
      "step": 109835
    },
    {
      "epoch": 1.43,
      "grad_norm": 1.0552316904067993,
      "learning_rate": 0.000321928337782688,
      "loss": 2.1728,
      "step": 109836
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.7756507396697998,
      "learning_rate": 0.00032192425816064067,
      "loss": 1.9851,
      "step": 109837
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.905388593673706,
      "learning_rate": 0.00032192017853451715,
      "loss": 2.142,
      "step": 109838
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.9335557818412781,
      "learning_rate": 0.0003219160989043182,
      "loss": 2.0514,
      "step": 109839
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.9008907079696655,
      "learning_rate": 0.00032191201927004475,
      "loss": 2.232,
      "step": 109840
    },
    {
      "epoch": 1.43,
      "grad_norm": 1.0910412073135376,
      "learning_rate": 0.00032190793963169734,
      "loss": 2.1086,
      "step": 109841
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.9532305598258972,
      "learning_rate": 0.00032190385998927685,
      "loss": 2.0002,
      "step": 109842
    },
    {
      "epoch": 1.43,
      "grad_norm": 1.0373969078063965,
      "learning_rate": 0.0003218997803427841,
      "loss": 2.132,
      "step": 109843
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.885572612285614,
      "learning_rate": 0.0003218957006922196,
      "loss": 2.2235,
      "step": 109844
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.9119901657104492,
      "learning_rate": 0.0003218916210375844,
      "loss": 2.2549,
      "step": 109845
    },
    {
      "epoch": 1.43,
      "grad_norm": 1.1045578718185425,
      "learning_rate": 0.00032188754137887915,
      "loss": 1.9845,
      "step": 109846
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.9562901258468628,
      "learning_rate": 0.00032188346171610455,
      "loss": 2.2375,
      "step": 109847
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.9257857799530029,
      "learning_rate": 0.0003218793820492614,
      "loss": 2.0577,
      "step": 109848
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.9660360217094421,
      "learning_rate": 0.0003218753023783505,
      "loss": 2.1826,
      "step": 109849
    },
    {
      "epoch": 1.43,
      "grad_norm": 1.1283131837844849,
      "learning_rate": 0.0003218712227033725,
      "loss": 2.147,
      "step": 109850
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.8864763379096985,
      "learning_rate": 0.00032186714302432826,
      "loss": 1.9368,
      "step": 109851
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.9125743508338928,
      "learning_rate": 0.00032186306334121853,
      "loss": 2.0295,
      "step": 109852
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.8606469035148621,
      "learning_rate": 0.00032185898365404397,
      "loss": 1.9551,
      "step": 109853
    },
    {
      "epoch": 1.43,
      "grad_norm": 1.0844924449920654,
      "learning_rate": 0.0003218549039628055,
      "loss": 1.815,
      "step": 109854
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.9214646816253662,
      "learning_rate": 0.00032185082426750374,
      "loss": 2.1647,
      "step": 109855
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.9005693197250366,
      "learning_rate": 0.0003218467445681395,
      "loss": 2.1507,
      "step": 109856
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.8332704901695251,
      "learning_rate": 0.00032184266486471355,
      "loss": 2.0637,
      "step": 109857
    },
    {
      "epoch": 1.43,
      "grad_norm": 1.069701910018921,
      "learning_rate": 0.0003218385851572266,
      "loss": 1.9371,
      "step": 109858
    },
    {
      "epoch": 1.43,
      "grad_norm": 1.0644909143447876,
      "learning_rate": 0.0003218345054456795,
      "loss": 2.0066,
      "step": 109859
    },
    {
      "epoch": 1.43,
      "grad_norm": 1.0877737998962402,
      "learning_rate": 0.00032183042573007294,
      "loss": 2.0261,
      "step": 109860
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.9166379570960999,
      "learning_rate": 0.0003218263460104077,
      "loss": 2.0555,
      "step": 109861
    },
    {
      "epoch": 1.43,
      "grad_norm": 1.2028827667236328,
      "learning_rate": 0.0003218222662866845,
      "loss": 2.1531,
      "step": 109862
    },
    {
      "epoch": 1.43,
      "grad_norm": 1.2691253423690796,
      "learning_rate": 0.00032181818655890415,
      "loss": 2.0961,
      "step": 109863
    },
    {
      "epoch": 1.43,
      "grad_norm": 1.0504881143569946,
      "learning_rate": 0.00032181410682706734,
      "loss": 2.0418,
      "step": 109864
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.9559975862503052,
      "learning_rate": 0.000321810027091175,
      "loss": 2.0775,
      "step": 109865
    },
    {
      "epoch": 1.43,
      "grad_norm": 1.2841166257858276,
      "learning_rate": 0.0003218059473512276,
      "loss": 2.3152,
      "step": 109866
    },
    {
      "epoch": 1.43,
      "grad_norm": 1.3735542297363281,
      "learning_rate": 0.00032180186760722617,
      "loss": 2.174,
      "step": 109867
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.8912679553031921,
      "learning_rate": 0.00032179778785917137,
      "loss": 2.0679,
      "step": 109868
    },
    {
      "epoch": 1.43,
      "grad_norm": 1.0054315328598022,
      "learning_rate": 0.0003217937081070639,
      "loss": 2.1204,
      "step": 109869
    },
    {
      "epoch": 1.43,
      "grad_norm": 1.3310407400131226,
      "learning_rate": 0.0003217896283509045,
      "loss": 2.0819,
      "step": 109870
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.9764772057533264,
      "learning_rate": 0.0003217855485906942,
      "loss": 2.2191,
      "step": 109871
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.9301242828369141,
      "learning_rate": 0.0003217814688264334,
      "loss": 2.076,
      "step": 109872
    },
    {
      "epoch": 1.43,
      "grad_norm": 1.0841140747070312,
      "learning_rate": 0.000321777389058123,
      "loss": 2.3958,
      "step": 109873
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.9492329359054565,
      "learning_rate": 0.0003217733092857639,
      "loss": 2.021,
      "step": 109874
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.9781321287155151,
      "learning_rate": 0.00032176922950935663,
      "loss": 1.993,
      "step": 109875
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.987766683101654,
      "learning_rate": 0.000321765149728902,
      "loss": 2.2056,
      "step": 109876
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.8803786039352417,
      "learning_rate": 0.00032176106994440096,
      "loss": 2.0207,
      "step": 109877
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.8868644833564758,
      "learning_rate": 0.00032175699015585404,
      "loss": 2.1093,
      "step": 109878
    },
    {
      "epoch": 1.43,
      "grad_norm": 1.0481268167495728,
      "learning_rate": 0.0003217529103632621,
      "loss": 2.117,
      "step": 109879
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.9729171991348267,
      "learning_rate": 0.0003217488305666259,
      "loss": 2.2516,
      "step": 109880
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.8782806396484375,
      "learning_rate": 0.00032174475076594616,
      "loss": 2.0539,
      "step": 109881
    },
    {
      "epoch": 1.43,
      "grad_norm": 1.1389827728271484,
      "learning_rate": 0.0003217406709612237,
      "loss": 2.1644,
      "step": 109882
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.9131982326507568,
      "learning_rate": 0.0003217365911524591,
      "loss": 2.0292,
      "step": 109883
    },
    {
      "epoch": 1.43,
      "grad_norm": 1.0087867975234985,
      "learning_rate": 0.0003217325113396535,
      "loss": 2.2213,
      "step": 109884
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.8824610114097595,
      "learning_rate": 0.0003217284315228072,
      "loss": 2.1226,
      "step": 109885
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.8611668944358826,
      "learning_rate": 0.00032172435170192123,
      "loss": 2.1826,
      "step": 109886
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.8631089925765991,
      "learning_rate": 0.00032172027187699634,
      "loss": 2.1847,
      "step": 109887
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.9195642471313477,
      "learning_rate": 0.00032171619204803323,
      "loss": 2.2854,
      "step": 109888
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.9845003485679626,
      "learning_rate": 0.00032171211221503266,
      "loss": 2.1776,
      "step": 109889
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.9611623883247375,
      "learning_rate": 0.0003217080323779954,
      "loss": 1.9894,
      "step": 109890
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.9171397686004639,
      "learning_rate": 0.00032170395253692223,
      "loss": 2.2172,
      "step": 109891
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.8737227320671082,
      "learning_rate": 0.0003216998726918138,
      "loss": 1.9558,
      "step": 109892
    },
    {
      "epoch": 1.43,
      "grad_norm": 1.0877888202667236,
      "learning_rate": 0.000321695792842671,
      "loss": 2.0753,
      "step": 109893
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.8997809886932373,
      "learning_rate": 0.0003216917129894946,
      "loss": 2.0881,
      "step": 109894
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.8760338425636292,
      "learning_rate": 0.00032168763313228523,
      "loss": 2.1647,
      "step": 109895
    },
    {
      "epoch": 1.43,
      "grad_norm": 1.0409923791885376,
      "learning_rate": 0.0003216835532710437,
      "loss": 2.2321,
      "step": 109896
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.8803106546401978,
      "learning_rate": 0.0003216794734057708,
      "loss": 2.3389,
      "step": 109897
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.9704411029815674,
      "learning_rate": 0.0003216753935364674,
      "loss": 2.249,
      "step": 109898
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.9224653244018555,
      "learning_rate": 0.00032167131366313403,
      "loss": 2.0868,
      "step": 109899
    },
    {
      "epoch": 1.43,
      "grad_norm": 1.001214623451233,
      "learning_rate": 0.0003216672337857716,
      "loss": 2.1247,
      "step": 109900
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.8690403699874878,
      "learning_rate": 0.0003216631539043808,
      "loss": 2.0689,
      "step": 109901
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.8716908097267151,
      "learning_rate": 0.0003216590740189623,
      "loss": 1.9867,
      "step": 109902
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.9345563650131226,
      "learning_rate": 0.00032165499412951714,
      "loss": 2.2488,
      "step": 109903
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.9466003179550171,
      "learning_rate": 0.00032165091423604585,
      "loss": 2.1412,
      "step": 109904
    },
    {
      "epoch": 1.43,
      "grad_norm": 1.192540168762207,
      "learning_rate": 0.00032164683433854923,
      "loss": 2.0653,
      "step": 109905
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.9587779641151428,
      "learning_rate": 0.00032164275443702806,
      "loss": 2.2655,
      "step": 109906
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.9058939218521118,
      "learning_rate": 0.0003216386745314831,
      "loss": 2.0813,
      "step": 109907
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.96672123670578,
      "learning_rate": 0.00032163459462191513,
      "loss": 2.1936,
      "step": 109908
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.9693360924720764,
      "learning_rate": 0.00032163051470832484,
      "loss": 2.1687,
      "step": 109909
    },
    {
      "epoch": 1.43,
      "grad_norm": 1.123134970664978,
      "learning_rate": 0.0003216264347907131,
      "loss": 2.1282,
      "step": 109910
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.905918300151825,
      "learning_rate": 0.00032162235486908044,
      "loss": 2.3908,
      "step": 109911
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.9017960429191589,
      "learning_rate": 0.00032161827494342786,
      "loss": 2.3846,
      "step": 109912
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.9595584869384766,
      "learning_rate": 0.0003216141950137561,
      "loss": 2.1713,
      "step": 109913
    },
    {
      "epoch": 1.43,
      "grad_norm": 1.0305839776992798,
      "learning_rate": 0.0003216101150800657,
      "loss": 2.1716,
      "step": 109914
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.9036725163459778,
      "learning_rate": 0.0003216060351423577,
      "loss": 2.119,
      "step": 109915
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.9192799925804138,
      "learning_rate": 0.0003216019552006327,
      "loss": 1.9343,
      "step": 109916
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.848063588142395,
      "learning_rate": 0.00032159787525489153,
      "loss": 2.1515,
      "step": 109917
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.9862127900123596,
      "learning_rate": 0.0003215937953051348,
      "loss": 2.0898,
      "step": 109918
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.9691669344902039,
      "learning_rate": 0.00032158971535136354,
      "loss": 2.1451,
      "step": 109919
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.8773990869522095,
      "learning_rate": 0.0003215856353935782,
      "loss": 1.8822,
      "step": 109920
    },
    {
      "epoch": 1.43,
      "grad_norm": 1.0175244808197021,
      "learning_rate": 0.0003215815554317797,
      "loss": 2.1624,
      "step": 109921
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.9775725603103638,
      "learning_rate": 0.0003215774754659688,
      "loss": 2.0251,
      "step": 109922
    },
    {
      "epoch": 1.43,
      "grad_norm": 1.010433554649353,
      "learning_rate": 0.0003215733954961463,
      "loss": 2.1723,
      "step": 109923
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.9108237624168396,
      "learning_rate": 0.0003215693155223128,
      "loss": 2.1072,
      "step": 109924
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.8756005764007568,
      "learning_rate": 0.0003215652355444693,
      "loss": 2.0097,
      "step": 109925
    },
    {
      "epoch": 1.43,
      "grad_norm": 1.2908990383148193,
      "learning_rate": 0.0003215611555626163,
      "loss": 2.1236,
      "step": 109926
    },
    {
      "epoch": 1.43,
      "grad_norm": 1.068202257156372,
      "learning_rate": 0.00032155707557675464,
      "loss": 2.2359,
      "step": 109927
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.9346698522567749,
      "learning_rate": 0.0003215529955868852,
      "loss": 2.3248,
      "step": 109928
    },
    {
      "epoch": 1.43,
      "grad_norm": 1.0014591217041016,
      "learning_rate": 0.0003215489155930086,
      "loss": 2.1773,
      "step": 109929
    },
    {
      "epoch": 1.43,
      "grad_norm": 1.1422998905181885,
      "learning_rate": 0.00032154483559512564,
      "loss": 2.1088,
      "step": 109930
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.9714936017990112,
      "learning_rate": 0.0003215407555932371,
      "loss": 1.9838,
      "step": 109931
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.9153434634208679,
      "learning_rate": 0.00032153667558734386,
      "loss": 1.9845,
      "step": 109932
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.9944180250167847,
      "learning_rate": 0.0003215325955774464,
      "loss": 2.1164,
      "step": 109933
    },
    {
      "epoch": 1.43,
      "grad_norm": 1.0326718091964722,
      "learning_rate": 0.0003215285155635456,
      "loss": 2.276,
      "step": 109934
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.9465439319610596,
      "learning_rate": 0.0003215244355456423,
      "loss": 2.1781,
      "step": 109935
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.8710770010948181,
      "learning_rate": 0.0003215203555237372,
      "loss": 2.2209,
      "step": 109936
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.8788470029830933,
      "learning_rate": 0.0003215162754978311,
      "loss": 2.0872,
      "step": 109937
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.9138880968093872,
      "learning_rate": 0.00032151219546792465,
      "loss": 2.182,
      "step": 109938
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.9823797941207886,
      "learning_rate": 0.00032150811543401865,
      "loss": 2.0534,
      "step": 109939
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.8939152359962463,
      "learning_rate": 0.00032150403539611395,
      "loss": 2.1304,
      "step": 109940
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.9347206354141235,
      "learning_rate": 0.00032149995535421126,
      "loss": 2.372,
      "step": 109941
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.8910932540893555,
      "learning_rate": 0.00032149587530831133,
      "loss": 2.0888,
      "step": 109942
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.8956723213195801,
      "learning_rate": 0.0003214917952584148,
      "loss": 2.1419,
      "step": 109943
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.9207831621170044,
      "learning_rate": 0.00032148771520452274,
      "loss": 2.0772,
      "step": 109944
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.8877472281455994,
      "learning_rate": 0.0003214836351466355,
      "loss": 2.1928,
      "step": 109945
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.9432819485664368,
      "learning_rate": 0.0003214795550847541,
      "loss": 1.904,
      "step": 109946
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.9492190480232239,
      "learning_rate": 0.0003214754750188793,
      "loss": 2.0093,
      "step": 109947
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.9259412288665771,
      "learning_rate": 0.0003214713949490118,
      "loss": 1.983,
      "step": 109948
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.8498772978782654,
      "learning_rate": 0.00032146731487515235,
      "loss": 2.1359,
      "step": 109949
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.916130006313324,
      "learning_rate": 0.0003214632347973017,
      "loss": 2.055,
      "step": 109950
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.9984790682792664,
      "learning_rate": 0.0003214591547154606,
      "loss": 2.2664,
      "step": 109951
    },
    {
      "epoch": 1.43,
      "grad_norm": 1.040546178817749,
      "learning_rate": 0.0003214550746296299,
      "loss": 2.1469,
      "step": 109952
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.9706639647483826,
      "learning_rate": 0.0003214509945398103,
      "loss": 2.2816,
      "step": 109953
    },
    {
      "epoch": 1.43,
      "grad_norm": 1.0980589389801025,
      "learning_rate": 0.00032144691444600256,
      "loss": 2.2246,
      "step": 109954
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.9377866387367249,
      "learning_rate": 0.00032144283434820735,
      "loss": 2.1109,
      "step": 109955
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.9134921431541443,
      "learning_rate": 0.00032143875424642563,
      "loss": 2.0829,
      "step": 109956
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.906252384185791,
      "learning_rate": 0.00032143467414065796,
      "loss": 2.1936,
      "step": 109957
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.8956409096717834,
      "learning_rate": 0.00032143059403090513,
      "loss": 2.0339,
      "step": 109958
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.9069560766220093,
      "learning_rate": 0.0003214265139171681,
      "loss": 2.1825,
      "step": 109959
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.9252932071685791,
      "learning_rate": 0.0003214224337994474,
      "loss": 2.1492,
      "step": 109960
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.8894202709197998,
      "learning_rate": 0.00032141835367774385,
      "loss": 2.2055,
      "step": 109961
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.9507783055305481,
      "learning_rate": 0.00032141427355205824,
      "loss": 2.2352,
      "step": 109962
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.8355730175971985,
      "learning_rate": 0.0003214101934223912,
      "loss": 2.219,
      "step": 109963
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.9777824878692627,
      "learning_rate": 0.0003214061132887438,
      "loss": 2.1296,
      "step": 109964
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.844088613986969,
      "learning_rate": 0.0003214020331511165,
      "loss": 2.0284,
      "step": 109965
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.9231684803962708,
      "learning_rate": 0.0003213979530095102,
      "loss": 2.2136,
      "step": 109966
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.9454681873321533,
      "learning_rate": 0.0003213938728639255,
      "loss": 2.0547,
      "step": 109967
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.9486477971076965,
      "learning_rate": 0.00032138979271436344,
      "loss": 2.0232,
      "step": 109968
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.979360044002533,
      "learning_rate": 0.00032138571256082446,
      "loss": 2.0561,
      "step": 109969
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.8801517486572266,
      "learning_rate": 0.00032138163240330955,
      "loss": 2.1883,
      "step": 109970
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.9300007224082947,
      "learning_rate": 0.00032137755224181935,
      "loss": 2.2125,
      "step": 109971
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.8946744203567505,
      "learning_rate": 0.0003213734720763547,
      "loss": 2.3226,
      "step": 109972
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.8258441090583801,
      "learning_rate": 0.0003213693919069163,
      "loss": 2.0676,
      "step": 109973
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.9156146049499512,
      "learning_rate": 0.00032136531173350495,
      "loss": 2.2839,
      "step": 109974
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.8281065225601196,
      "learning_rate": 0.0003213612315561213,
      "loss": 2.0554,
      "step": 109975
    },
    {
      "epoch": 1.43,
      "grad_norm": 1.1291005611419678,
      "learning_rate": 0.0003213571513747663,
      "loss": 2.0332,
      "step": 109976
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.9483742713928223,
      "learning_rate": 0.0003213530711894406,
      "loss": 2.1914,
      "step": 109977
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.914284348487854,
      "learning_rate": 0.0003213489910001449,
      "loss": 2.0187,
      "step": 109978
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.9078112840652466,
      "learning_rate": 0.0003213449108068801,
      "loss": 2.0005,
      "step": 109979
    },
    {
      "epoch": 1.43,
      "grad_norm": 1.64510977268219,
      "learning_rate": 0.0003213408306096468,
      "loss": 2.1546,
      "step": 109980
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.8716646432876587,
      "learning_rate": 0.00032133675040844586,
      "loss": 2.1419,
      "step": 109981
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.8493413329124451,
      "learning_rate": 0.000321332670203278,
      "loss": 2.0419,
      "step": 109982
    },
    {
      "epoch": 1.43,
      "grad_norm": 1.0434402227401733,
      "learning_rate": 0.000321328589994144,
      "loss": 2.3666,
      "step": 109983
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.9088649749755859,
      "learning_rate": 0.00032132450978104464,
      "loss": 2.2264,
      "step": 109984
    },
    {
      "epoch": 1.43,
      "grad_norm": 1.2173473834991455,
      "learning_rate": 0.0003213204295639806,
      "loss": 2.3935,
      "step": 109985
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.9114411473274231,
      "learning_rate": 0.00032131634934295277,
      "loss": 2.0332,
      "step": 109986
    },
    {
      "epoch": 1.43,
      "grad_norm": 1.1144624948501587,
      "learning_rate": 0.0003213122691179618,
      "loss": 1.9409,
      "step": 109987
    },
    {
      "epoch": 1.43,
      "grad_norm": 1.39235520362854,
      "learning_rate": 0.00032130818888900835,
      "loss": 2.0733,
      "step": 109988
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.941206693649292,
      "learning_rate": 0.0003213041086560935,
      "loss": 2.2315,
      "step": 109989
    },
    {
      "epoch": 1.43,
      "grad_norm": 1.086578369140625,
      "learning_rate": 0.0003213000284192177,
      "loss": 2.1725,
      "step": 109990
    },
    {
      "epoch": 1.43,
      "grad_norm": 1.15755295753479,
      "learning_rate": 0.00032129594817838187,
      "loss": 2.2376,
      "step": 109991
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.9014251232147217,
      "learning_rate": 0.0003212918679335867,
      "loss": 2.0334,
      "step": 109992
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.8876051306724548,
      "learning_rate": 0.00032128778768483296,
      "loss": 2.1661,
      "step": 109993
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.9257716536521912,
      "learning_rate": 0.00032128370743212146,
      "loss": 2.1265,
      "step": 109994
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.9456195831298828,
      "learning_rate": 0.0003212796271754529,
      "loss": 2.3137,
      "step": 109995
    },
    {
      "epoch": 1.43,
      "grad_norm": 1.0456299781799316,
      "learning_rate": 0.00032127554691482803,
      "loss": 2.1674,
      "step": 109996
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.8818404078483582,
      "learning_rate": 0.0003212714666502476,
      "loss": 2.0578,
      "step": 109997
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.9607064127922058,
      "learning_rate": 0.00032126738638171246,
      "loss": 2.1226,
      "step": 109998
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.9921755790710449,
      "learning_rate": 0.00032126330610922333,
      "loss": 2.1123,
      "step": 109999
    },
    {
      "epoch": 1.43,
      "grad_norm": 1.0249215364456177,
      "learning_rate": 0.0003212592258327809,
      "loss": 2.0202,
      "step": 110000
    },
    {
      "epoch": 1.43,
      "grad_norm": 1.0010440349578857,
      "learning_rate": 0.000321255145552386,
      "loss": 2.2052,
      "step": 110001
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.9346053004264832,
      "learning_rate": 0.0003212510652680394,
      "loss": 2.1401,
      "step": 110002
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.9935625791549683,
      "learning_rate": 0.0003212469849797418,
      "loss": 2.1225,
      "step": 110003
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.9993984699249268,
      "learning_rate": 0.000321242904687494,
      "loss": 2.3429,
      "step": 110004
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.9533715844154358,
      "learning_rate": 0.0003212388243912967,
      "loss": 2.1149,
      "step": 110005
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.9257070422172546,
      "learning_rate": 0.00032123474409115067,
      "loss": 2.1588,
      "step": 110006
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.9359204173088074,
      "learning_rate": 0.0003212306637870568,
      "loss": 2.25,
      "step": 110007
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.9445604681968689,
      "learning_rate": 0.00032122658347901573,
      "loss": 2.4109,
      "step": 110008
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.9253365993499756,
      "learning_rate": 0.00032122250316702814,
      "loss": 2.2058,
      "step": 110009
    },
    {
      "epoch": 1.43,
      "grad_norm": 1.031775712966919,
      "learning_rate": 0.00032121842285109497,
      "loss": 2.2425,
      "step": 110010
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.8594008088111877,
      "learning_rate": 0.00032121434253121694,
      "loss": 1.896,
      "step": 110011
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.9500552415847778,
      "learning_rate": 0.0003212102622073947,
      "loss": 2.1936,
      "step": 110012
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.9961127042770386,
      "learning_rate": 0.000321206181879629,
      "loss": 2.2451,
      "step": 110013
    },
    {
      "epoch": 1.43,
      "grad_norm": 1.0905200242996216,
      "learning_rate": 0.00032120210154792077,
      "loss": 2.1492,
      "step": 110014
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.8353405594825745,
      "learning_rate": 0.0003211980212122707,
      "loss": 1.974,
      "step": 110015
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.9639519453048706,
      "learning_rate": 0.0003211939408726794,
      "loss": 2.1381,
      "step": 110016
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.8946183919906616,
      "learning_rate": 0.0003211898605291479,
      "loss": 2.1343,
      "step": 110017
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.9271572232246399,
      "learning_rate": 0.0003211857801816767,
      "loss": 1.9716,
      "step": 110018
    },
    {
      "epoch": 1.43,
      "grad_norm": 1.06698739528656,
      "learning_rate": 0.0003211816998302666,
      "loss": 1.9112,
      "step": 110019
    },
    {
      "epoch": 1.43,
      "grad_norm": 1.0366220474243164,
      "learning_rate": 0.0003211776194749186,
      "loss": 2.0255,
      "step": 110020
    },
    {
      "epoch": 1.43,
      "grad_norm": 1.0050703287124634,
      "learning_rate": 0.0003211735391156331,
      "loss": 2.0092,
      "step": 110021
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.918407142162323,
      "learning_rate": 0.0003211694587524111,
      "loss": 2.3157,
      "step": 110022
    },
    {
      "epoch": 1.43,
      "grad_norm": 1.0148932933807373,
      "learning_rate": 0.0003211653783852534,
      "loss": 2.2874,
      "step": 110023
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.9561854004859924,
      "learning_rate": 0.00032116129801416053,
      "loss": 2.3407,
      "step": 110024
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.8441557288169861,
      "learning_rate": 0.0003211572176391334,
      "loss": 2.1522,
      "step": 110025
    },
    {
      "epoch": 1.43,
      "grad_norm": 1.0958635807037354,
      "learning_rate": 0.0003211531372601728,
      "loss": 2.1834,
      "step": 110026
    },
    {
      "epoch": 1.43,
      "grad_norm": 1.0776997804641724,
      "learning_rate": 0.0003211490568772794,
      "loss": 2.123,
      "step": 110027
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.924235463142395,
      "learning_rate": 0.00032114497649045397,
      "loss": 2.2495,
      "step": 110028
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.9676390886306763,
      "learning_rate": 0.00032114089609969735,
      "loss": 2.3143,
      "step": 110029
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.9277904629707336,
      "learning_rate": 0.0003211368157050102,
      "loss": 2.2912,
      "step": 110030
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.8982101678848267,
      "learning_rate": 0.0003211327353063932,
      "loss": 2.1952,
      "step": 110031
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.871052086353302,
      "learning_rate": 0.00032112865490384747,
      "loss": 1.9905,
      "step": 110032
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.9608925580978394,
      "learning_rate": 0.00032112457449737334,
      "loss": 2.2513,
      "step": 110033
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.9171196222305298,
      "learning_rate": 0.0003211204940869718,
      "loss": 2.1392,
      "step": 110034
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.9486384987831116,
      "learning_rate": 0.00032111641367264365,
      "loss": 2.0878,
      "step": 110035
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.8952835202217102,
      "learning_rate": 0.0003211123332543894,
      "loss": 2.248,
      "step": 110036
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.897950291633606,
      "learning_rate": 0.00032110825283221,
      "loss": 2.2027,
      "step": 110037
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.9705384373664856,
      "learning_rate": 0.00032110417240610627,
      "loss": 2.1767,
      "step": 110038
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.9913869500160217,
      "learning_rate": 0.0003211000919760789,
      "loss": 1.9667,
      "step": 110039
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.9706301689147949,
      "learning_rate": 0.00032109601154212846,
      "loss": 2.1628,
      "step": 110040
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.9078862071037292,
      "learning_rate": 0.000321091931104256,
      "loss": 2.1751,
      "step": 110041
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.9832099080085754,
      "learning_rate": 0.0003210878506624621,
      "loss": 2.023,
      "step": 110042
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.8559897541999817,
      "learning_rate": 0.0003210837702167476,
      "loss": 2.511,
      "step": 110043
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.9376140236854553,
      "learning_rate": 0.00032107968976711327,
      "loss": 2.1691,
      "step": 110044
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.8578213453292847,
      "learning_rate": 0.00032107560931355977,
      "loss": 2.0922,
      "step": 110045
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.8578543066978455,
      "learning_rate": 0.00032107152885608786,
      "loss": 1.981,
      "step": 110046
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.924712061882019,
      "learning_rate": 0.00032106744839469845,
      "loss": 2.0224,
      "step": 110047
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.9116280674934387,
      "learning_rate": 0.00032106336792939215,
      "loss": 1.903,
      "step": 110048
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.8767450451850891,
      "learning_rate": 0.00032105928746016977,
      "loss": 2.2063,
      "step": 110049
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.9768164753913879,
      "learning_rate": 0.0003210552069870321,
      "loss": 2.1644,
      "step": 110050
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.8936809301376343,
      "learning_rate": 0.0003210511265099799,
      "loss": 2.1436,
      "step": 110051
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.8321089148521423,
      "learning_rate": 0.00032104704602901386,
      "loss": 2.0648,
      "step": 110052
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.8272761702537537,
      "learning_rate": 0.00032104296554413483,
      "loss": 2.0112,
      "step": 110053
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.919666051864624,
      "learning_rate": 0.00032103888505534347,
      "loss": 2.227,
      "step": 110054
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.8935878872871399,
      "learning_rate": 0.0003210348045626405,
      "loss": 2.0344,
      "step": 110055
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.8187216520309448,
      "learning_rate": 0.00032103072406602684,
      "loss": 2.0571,
      "step": 110056
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.9721700549125671,
      "learning_rate": 0.00032102664356550316,
      "loss": 2.1441,
      "step": 110057
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.9423078298568726,
      "learning_rate": 0.0003210225630610702,
      "loss": 2.0839,
      "step": 110058
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.9286200404167175,
      "learning_rate": 0.0003210184825527288,
      "loss": 2.1883,
      "step": 110059
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.9099205732345581,
      "learning_rate": 0.00032101440204047974,
      "loss": 2.1424,
      "step": 110060
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.8915258646011353,
      "learning_rate": 0.0003210103215243236,
      "loss": 2.0608,
      "step": 110061
    },
    {
      "epoch": 1.43,
      "grad_norm": 1.0279523134231567,
      "learning_rate": 0.00032100624100426123,
      "loss": 2.2996,
      "step": 110062
    },
    {
      "epoch": 1.43,
      "grad_norm": 1.0092988014221191,
      "learning_rate": 0.00032100216048029347,
      "loss": 2.3037,
      "step": 110063
    },
    {
      "epoch": 1.43,
      "grad_norm": 1.0486376285552979,
      "learning_rate": 0.000320998079952421,
      "loss": 2.1257,
      "step": 110064
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.8755918145179749,
      "learning_rate": 0.0003209939994206446,
      "loss": 1.8728,
      "step": 110065
    },
    {
      "epoch": 1.43,
      "grad_norm": 1.1688650846481323,
      "learning_rate": 0.0003209899188849649,
      "loss": 2.1376,
      "step": 110066
    },
    {
      "epoch": 1.43,
      "grad_norm": 1.0674097537994385,
      "learning_rate": 0.0003209858383453829,
      "loss": 2.517,
      "step": 110067
    },
    {
      "epoch": 1.43,
      "grad_norm": 1.0948182344436646,
      "learning_rate": 0.0003209817578018992,
      "loss": 2.0944,
      "step": 110068
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.9479371309280396,
      "learning_rate": 0.00032097767725451464,
      "loss": 2.178,
      "step": 110069
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.9504961371421814,
      "learning_rate": 0.00032097359670322985,
      "loss": 2.2955,
      "step": 110070
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.9771863222122192,
      "learning_rate": 0.00032096951614804574,
      "loss": 2.2245,
      "step": 110071
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.9227446913719177,
      "learning_rate": 0.000320965435588963,
      "loss": 2.096,
      "step": 110072
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.8603383302688599,
      "learning_rate": 0.00032096135502598233,
      "loss": 2.3005,
      "step": 110073
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.9007773995399475,
      "learning_rate": 0.0003209572744591046,
      "loss": 1.9865,
      "step": 110074
    },
    {
      "epoch": 1.43,
      "grad_norm": 1.029280424118042,
      "learning_rate": 0.00032095319388833054,
      "loss": 2.4071,
      "step": 110075
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.9241489768028259,
      "learning_rate": 0.00032094911331366087,
      "loss": 2.1864,
      "step": 110076
    },
    {
      "epoch": 1.43,
      "grad_norm": 1.0451253652572632,
      "learning_rate": 0.00032094503273509626,
      "loss": 2.1769,
      "step": 110077
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.9180007576942444,
      "learning_rate": 0.00032094095215263774,
      "loss": 2.2172,
      "step": 110078
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.9990551471710205,
      "learning_rate": 0.0003209368715662858,
      "loss": 2.1056,
      "step": 110079
    },
    {
      "epoch": 1.43,
      "grad_norm": 1.0390822887420654,
      "learning_rate": 0.0003209327909760413,
      "loss": 2.0361,
      "step": 110080
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.8397700190544128,
      "learning_rate": 0.0003209287103819051,
      "loss": 2.1475,
      "step": 110081
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.8959141969680786,
      "learning_rate": 0.00032092462978387776,
      "loss": 2.0747,
      "step": 110082
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.8863016963005066,
      "learning_rate": 0.0003209205491819601,
      "loss": 2.2972,
      "step": 110083
    },
    {
      "epoch": 1.43,
      "grad_norm": 1.109138011932373,
      "learning_rate": 0.000320916468576153,
      "loss": 2.0994,
      "step": 110084
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.8938292264938354,
      "learning_rate": 0.0003209123879664571,
      "loss": 1.9853,
      "step": 110085
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.9105551838874817,
      "learning_rate": 0.00032090830735287314,
      "loss": 2.0923,
      "step": 110086
    },
    {
      "epoch": 1.43,
      "grad_norm": 1.0244405269622803,
      "learning_rate": 0.0003209042267354021,
      "loss": 2.2474,
      "step": 110087
    },
    {
      "epoch": 1.43,
      "grad_norm": 1.101180911064148,
      "learning_rate": 0.0003209001461140444,
      "loss": 1.9263,
      "step": 110088
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.9476514458656311,
      "learning_rate": 0.000320896065488801,
      "loss": 2.1262,
      "step": 110089
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.8958235383033752,
      "learning_rate": 0.00032089198485967267,
      "loss": 2.0706,
      "step": 110090
    },
    {
      "epoch": 1.43,
      "grad_norm": 1.1743507385253906,
      "learning_rate": 0.00032088790422666013,
      "loss": 2.2749,
      "step": 110091
    },
    {
      "epoch": 1.43,
      "grad_norm": 1.1584923267364502,
      "learning_rate": 0.0003208838235897641,
      "loss": 2.1581,
      "step": 110092
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.9662473797798157,
      "learning_rate": 0.0003208797429489854,
      "loss": 2.0913,
      "step": 110093
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.9307142496109009,
      "learning_rate": 0.00032087566230432475,
      "loss": 2.2006,
      "step": 110094
    },
    {
      "epoch": 1.43,
      "grad_norm": 1.1316823959350586,
      "learning_rate": 0.00032087158165578286,
      "loss": 2.1349,
      "step": 110095
    },
    {
      "epoch": 1.43,
      "grad_norm": 1.0184320211410522,
      "learning_rate": 0.00032086750100336067,
      "loss": 2.1004,
      "step": 110096
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.8471418619155884,
      "learning_rate": 0.0003208634203470587,
      "loss": 1.9661,
      "step": 110097
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.9719642400741577,
      "learning_rate": 0.0003208593396868779,
      "loss": 2.2152,
      "step": 110098
    },
    {
      "epoch": 1.43,
      "grad_norm": 1.0436060428619385,
      "learning_rate": 0.00032085525902281897,
      "loss": 2.0501,
      "step": 110099
    },
    {
      "epoch": 1.43,
      "grad_norm": 1.065605878829956,
      "learning_rate": 0.0003208511783548826,
      "loss": 2.1431,
      "step": 110100
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.9300814270973206,
      "learning_rate": 0.0003208470976830696,
      "loss": 2.2646,
      "step": 110101
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.897274374961853,
      "learning_rate": 0.0003208430170073808,
      "loss": 2.2008,
      "step": 110102
    },
    {
      "epoch": 1.43,
      "grad_norm": 1.3680145740509033,
      "learning_rate": 0.0003208389363278168,
      "loss": 2.1323,
      "step": 110103
    },
    {
      "epoch": 1.43,
      "grad_norm": 1.026662826538086,
      "learning_rate": 0.00032083485564437847,
      "loss": 2.2787,
      "step": 110104
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.8692053556442261,
      "learning_rate": 0.0003208307749570666,
      "loss": 2.1734,
      "step": 110105
    },
    {
      "epoch": 1.43,
      "grad_norm": 1.1153348684310913,
      "learning_rate": 0.00032082669426588184,
      "loss": 2.2895,
      "step": 110106
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.9511688947677612,
      "learning_rate": 0.000320822613570825,
      "loss": 2.0228,
      "step": 110107
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.8342832922935486,
      "learning_rate": 0.00032081853287189695,
      "loss": 2.1612,
      "step": 110108
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.8833141922950745,
      "learning_rate": 0.0003208144521690982,
      "loss": 2.1471,
      "step": 110109
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.949682354927063,
      "learning_rate": 0.0003208103714624297,
      "loss": 1.9622,
      "step": 110110
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.9028429388999939,
      "learning_rate": 0.0003208062907518922,
      "loss": 2.0617,
      "step": 110111
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.8663398623466492,
      "learning_rate": 0.00032080221003748645,
      "loss": 2.296,
      "step": 110112
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.8959841132164001,
      "learning_rate": 0.00032079812931921307,
      "loss": 2.1662,
      "step": 110113
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.8913983702659607,
      "learning_rate": 0.000320794048597073,
      "loss": 2.2269,
      "step": 110114
    },
    {
      "epoch": 1.43,
      "grad_norm": 1.0554091930389404,
      "learning_rate": 0.0003207899678710669,
      "loss": 2.2825,
      "step": 110115
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.9353867769241333,
      "learning_rate": 0.0003207858871411955,
      "loss": 2.4885,
      "step": 110116
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.8931587934494019,
      "learning_rate": 0.0003207818064074597,
      "loss": 2.0511,
      "step": 110117
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.9276801943778992,
      "learning_rate": 0.00032077772566986015,
      "loss": 2.1072,
      "step": 110118
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.8652740120887756,
      "learning_rate": 0.0003207736449283975,
      "loss": 2.1638,
      "step": 110119
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.863542914390564,
      "learning_rate": 0.0003207695641830728,
      "loss": 2.1425,
      "step": 110120
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.9441947340965271,
      "learning_rate": 0.00032076548343388665,
      "loss": 1.8361,
      "step": 110121
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.9464351534843445,
      "learning_rate": 0.0003207614026808397,
      "loss": 2.2684,
      "step": 110122
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.8950766921043396,
      "learning_rate": 0.0003207573219239329,
      "loss": 2.1304,
      "step": 110123
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.8973468542098999,
      "learning_rate": 0.00032075324116316687,
      "loss": 2.0949,
      "step": 110124
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.9197172522544861,
      "learning_rate": 0.0003207491603985424,
      "loss": 1.925,
      "step": 110125
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.8905249834060669,
      "learning_rate": 0.00032074507963006035,
      "loss": 1.9114,
      "step": 110126
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.8827567100524902,
      "learning_rate": 0.0003207409988577213,
      "loss": 2.1803,
      "step": 110127
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.8979306817054749,
      "learning_rate": 0.0003207369180815261,
      "loss": 2.2018,
      "step": 110128
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.9476886987686157,
      "learning_rate": 0.00032073283730147566,
      "loss": 2.1814,
      "step": 110129
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.8959802985191345,
      "learning_rate": 0.0003207287565175705,
      "loss": 2.0486,
      "step": 110130
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.8909359574317932,
      "learning_rate": 0.0003207246757298114,
      "loss": 2.0307,
      "step": 110131
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.876281201839447,
      "learning_rate": 0.00032072059493819927,
      "loss": 2.0032,
      "step": 110132
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.8432021141052246,
      "learning_rate": 0.00032071651414273487,
      "loss": 2.1791,
      "step": 110133
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.9064804911613464,
      "learning_rate": 0.0003207124333434188,
      "loss": 2.1157,
      "step": 110134
    },
    {
      "epoch": 1.43,
      "grad_norm": 1.1193726062774658,
      "learning_rate": 0.0003207083525402518,
      "loss": 2.0758,
      "step": 110135
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.9977646470069885,
      "learning_rate": 0.00032070427173323487,
      "loss": 2.3423,
      "step": 110136
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.9910467863082886,
      "learning_rate": 0.0003207001909223686,
      "loss": 2.4381,
      "step": 110137
    },
    {
      "epoch": 1.43,
      "grad_norm": 1.0840625762939453,
      "learning_rate": 0.0003206961101076537,
      "loss": 1.9068,
      "step": 110138
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.9157246351242065,
      "learning_rate": 0.00032069202928909114,
      "loss": 2.2294,
      "step": 110139
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.9691101908683777,
      "learning_rate": 0.0003206879484666814,
      "loss": 2.1977,
      "step": 110140
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.9203477501869202,
      "learning_rate": 0.00032068386764042543,
      "loss": 2.1034,
      "step": 110141
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.954771876335144,
      "learning_rate": 0.00032067978681032396,
      "loss": 2.1031,
      "step": 110142
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.9875076413154602,
      "learning_rate": 0.0003206757059763777,
      "loss": 2.2067,
      "step": 110143
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.9140483140945435,
      "learning_rate": 0.00032067162513858745,
      "loss": 2.2549,
      "step": 110144
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.9824817776679993,
      "learning_rate": 0.000320667544296954,
      "loss": 2.2228,
      "step": 110145
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.9509174227714539,
      "learning_rate": 0.000320663463451478,
      "loss": 2.0254,
      "step": 110146
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.93349289894104,
      "learning_rate": 0.00032065938260216023,
      "loss": 2.1249,
      "step": 110147
    },
    {
      "epoch": 1.43,
      "grad_norm": 1.5647341012954712,
      "learning_rate": 0.00032065530174900153,
      "loss": 2.2106,
      "step": 110148
    },
    {
      "epoch": 1.43,
      "grad_norm": 1.0400550365447998,
      "learning_rate": 0.00032065122089200274,
      "loss": 2.2664,
      "step": 110149
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.9039976000785828,
      "learning_rate": 0.00032064714003116437,
      "loss": 2.2147,
      "step": 110150
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.9199849367141724,
      "learning_rate": 0.0003206430591664873,
      "loss": 2.1193,
      "step": 110151
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.8976725339889526,
      "learning_rate": 0.0003206389782979723,
      "loss": 2.1844,
      "step": 110152
    },
    {
      "epoch": 1.43,
      "grad_norm": 1.034363865852356,
      "learning_rate": 0.00032063489742562023,
      "loss": 2.2408,
      "step": 110153
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.9045562744140625,
      "learning_rate": 0.00032063081654943156,
      "loss": 2.2659,
      "step": 110154
    },
    {
      "epoch": 1.43,
      "grad_norm": 1.097414255142212,
      "learning_rate": 0.00032062673566940735,
      "loss": 2.2062,
      "step": 110155
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.8966950178146362,
      "learning_rate": 0.00032062265478554827,
      "loss": 2.0446,
      "step": 110156
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.888373851776123,
      "learning_rate": 0.00032061857389785496,
      "loss": 2.203,
      "step": 110157
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.8585829138755798,
      "learning_rate": 0.00032061449300632835,
      "loss": 2.1082,
      "step": 110158
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.9115500450134277,
      "learning_rate": 0.00032061041211096914,
      "loss": 2.0273,
      "step": 110159
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.9900961518287659,
      "learning_rate": 0.0003206063312117779,
      "loss": 2.2336,
      "step": 110160
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.9001282453536987,
      "learning_rate": 0.00032060225030875573,
      "loss": 2.2657,
      "step": 110161
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.9096270799636841,
      "learning_rate": 0.00032059816940190305,
      "loss": 2.0784,
      "step": 110162
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.9308511018753052,
      "learning_rate": 0.00032059408849122097,
      "loss": 2.1133,
      "step": 110163
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.9167232513427734,
      "learning_rate": 0.0003205900075767099,
      "loss": 2.0952,
      "step": 110164
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.9746795296669006,
      "learning_rate": 0.0003205859266583708,
      "loss": 2.3113,
      "step": 110165
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.9723402857780457,
      "learning_rate": 0.0003205818457362045,
      "loss": 2.0033,
      "step": 110166
    },
    {
      "epoch": 1.43,
      "grad_norm": 1.0378953218460083,
      "learning_rate": 0.00032057776481021153,
      "loss": 2.277,
      "step": 110167
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.8507966995239258,
      "learning_rate": 0.0003205736838803928,
      "loss": 2.1826,
      "step": 110168
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.9849216341972351,
      "learning_rate": 0.00032056960294674905,
      "loss": 2.067,
      "step": 110169
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.9675889611244202,
      "learning_rate": 0.00032056552200928094,
      "loss": 2.2282,
      "step": 110170
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.9249319434165955,
      "learning_rate": 0.0003205614410679894,
      "loss": 2.0503,
      "step": 110171
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.8698716759681702,
      "learning_rate": 0.00032055736012287506,
      "loss": 2.144,
      "step": 110172
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.9763321876525879,
      "learning_rate": 0.00032055327917393875,
      "loss": 2.0114,
      "step": 110173
    },
    {
      "epoch": 1.43,
      "grad_norm": 1.0334187746047974,
      "learning_rate": 0.00032054919822118116,
      "loss": 1.9705,
      "step": 110174
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.9286283254623413,
      "learning_rate": 0.0003205451172646031,
      "loss": 2.2516,
      "step": 110175
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.9476731419563293,
      "learning_rate": 0.00032054103630420534,
      "loss": 2.1271,
      "step": 110176
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.9944618940353394,
      "learning_rate": 0.00032053695533998855,
      "loss": 2.3352,
      "step": 110177
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.9344805479049683,
      "learning_rate": 0.00032053287437195364,
      "loss": 2.0712,
      "step": 110178
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.9540706872940063,
      "learning_rate": 0.00032052879340010125,
      "loss": 2.1958,
      "step": 110179
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.9890908002853394,
      "learning_rate": 0.0003205247124244321,
      "loss": 2.3875,
      "step": 110180
    },
    {
      "epoch": 1.43,
      "grad_norm": 1.0091928243637085,
      "learning_rate": 0.0003205206314449471,
      "loss": 2.1793,
      "step": 110181
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.9346343278884888,
      "learning_rate": 0.0003205165504616469,
      "loss": 2.0018,
      "step": 110182
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.8773817420005798,
      "learning_rate": 0.00032051246947453223,
      "loss": 2.3035,
      "step": 110183
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.883348286151886,
      "learning_rate": 0.00032050838848360407,
      "loss": 2.1799,
      "step": 110184
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.843768835067749,
      "learning_rate": 0.00032050430748886287,
      "loss": 1.9339,
      "step": 110185
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.9578404426574707,
      "learning_rate": 0.00032050022649030956,
      "loss": 2.3015,
      "step": 110186
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.8431217670440674,
      "learning_rate": 0.0003204961454879449,
      "loss": 2.3286,
      "step": 110187
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.8014075756072998,
      "learning_rate": 0.0003204920644817696,
      "loss": 1.8577,
      "step": 110188
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.841717004776001,
      "learning_rate": 0.00032048798347178437,
      "loss": 2.205,
      "step": 110189
    },
    {
      "epoch": 1.43,
      "grad_norm": 1.0258878469467163,
      "learning_rate": 0.00032048390245799016,
      "loss": 2.2607,
      "step": 110190
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.8622356057167053,
      "learning_rate": 0.00032047982144038753,
      "loss": 2.1105,
      "step": 110191
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.9642521739006042,
      "learning_rate": 0.00032047574041897735,
      "loss": 2.1805,
      "step": 110192
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.9879692792892456,
      "learning_rate": 0.0003204716593937603,
      "loss": 2.0575,
      "step": 110193
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.8562281727790833,
      "learning_rate": 0.0003204675783647372,
      "loss": 2.2013,
      "step": 110194
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.9328218698501587,
      "learning_rate": 0.0003204634973319088,
      "loss": 2.084,
      "step": 110195
    },
    {
      "epoch": 1.43,
      "grad_norm": 1.0029807090759277,
      "learning_rate": 0.0003204594162952759,
      "loss": 2.1272,
      "step": 110196
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.9123334884643555,
      "learning_rate": 0.0003204553352548391,
      "loss": 1.7867,
      "step": 110197
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.8905331492424011,
      "learning_rate": 0.0003204512542105994,
      "loss": 2.1202,
      "step": 110198
    },
    {
      "epoch": 1.43,
      "grad_norm": 1.0505244731903076,
      "learning_rate": 0.0003204471731625573,
      "loss": 2.1858,
      "step": 110199
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.9384745359420776,
      "learning_rate": 0.0003204430921107137,
      "loss": 1.9991,
      "step": 110200
    },
    {
      "epoch": 1.43,
      "grad_norm": 1.1609140634536743,
      "learning_rate": 0.00032043901105506935,
      "loss": 2.1007,
      "step": 110201
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.9230009913444519,
      "learning_rate": 0.0003204349299956251,
      "loss": 2.2514,
      "step": 110202
    },
    {
      "epoch": 1.43,
      "grad_norm": 1.0562766790390015,
      "learning_rate": 0.00032043084893238155,
      "loss": 2.0463,
      "step": 110203
    },
    {
      "epoch": 1.43,
      "grad_norm": 1.3132683038711548,
      "learning_rate": 0.00032042676786533947,
      "loss": 2.0109,
      "step": 110204
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.9247044920921326,
      "learning_rate": 0.00032042268679449975,
      "loss": 2.1625,
      "step": 110205
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.877785325050354,
      "learning_rate": 0.00032041860571986304,
      "loss": 2.0535,
      "step": 110206
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.8619930744171143,
      "learning_rate": 0.0003204145246414301,
      "loss": 1.9888,
      "step": 110207
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.8808256983757019,
      "learning_rate": 0.0003204104435592017,
      "loss": 2.1514,
      "step": 110208
    },
    {
      "epoch": 1.43,
      "grad_norm": 0.9855143427848816,
      "learning_rate": 0.0003204063624731787,
      "loss": 1.9863,
      "step": 110209
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.9829450249671936,
      "learning_rate": 0.00032040228138336167,
      "loss": 2.1508,
      "step": 110210
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.8826343417167664,
      "learning_rate": 0.0003203982002897515,
      "loss": 2.0955,
      "step": 110211
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.9727074503898621,
      "learning_rate": 0.00032039411919234897,
      "loss": 2.1828,
      "step": 110212
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.9308907985687256,
      "learning_rate": 0.0003203900380911547,
      "loss": 2.1802,
      "step": 110213
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.9543036818504333,
      "learning_rate": 0.0003203859569861696,
      "loss": 2.05,
      "step": 110214
    },
    {
      "epoch": 1.44,
      "grad_norm": 1.0382410287857056,
      "learning_rate": 0.00032038187587739444,
      "loss": 2.1083,
      "step": 110215
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.9741453528404236,
      "learning_rate": 0.00032037779476482974,
      "loss": 2.1452,
      "step": 110216
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.937461256980896,
      "learning_rate": 0.00032037371364847646,
      "loss": 2.0652,
      "step": 110217
    },
    {
      "epoch": 1.44,
      "grad_norm": 1.0902578830718994,
      "learning_rate": 0.0003203696325283354,
      "loss": 2.1607,
      "step": 110218
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.8879467844963074,
      "learning_rate": 0.00032036555140440724,
      "loss": 2.058,
      "step": 110219
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.9344196319580078,
      "learning_rate": 0.00032036147027669266,
      "loss": 2.1074,
      "step": 110220
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.9378830790519714,
      "learning_rate": 0.0003203573891451926,
      "loss": 2.1091,
      "step": 110221
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.912982702255249,
      "learning_rate": 0.0003203533080099076,
      "loss": 2.2564,
      "step": 110222
    },
    {
      "epoch": 1.44,
      "grad_norm": 1.0111018419265747,
      "learning_rate": 0.00032034922687083857,
      "loss": 2.0123,
      "step": 110223
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.894524097442627,
      "learning_rate": 0.0003203451457279863,
      "loss": 1.916,
      "step": 110224
    },
    {
      "epoch": 1.44,
      "grad_norm": 1.126692295074463,
      "learning_rate": 0.00032034106458135147,
      "loss": 2.1605,
      "step": 110225
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.8688777089118958,
      "learning_rate": 0.0003203369834309348,
      "loss": 2.0294,
      "step": 110226
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.8871026039123535,
      "learning_rate": 0.00032033290227673715,
      "loss": 2.0785,
      "step": 110227
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.9258796572685242,
      "learning_rate": 0.0003203288211187592,
      "loss": 1.9233,
      "step": 110228
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.9074614644050598,
      "learning_rate": 0.0003203247399570017,
      "loss": 2.1912,
      "step": 110229
    },
    {
      "epoch": 1.44,
      "grad_norm": 1.0246622562408447,
      "learning_rate": 0.0003203206587914655,
      "loss": 1.9718,
      "step": 110230
    },
    {
      "epoch": 1.44,
      "grad_norm": 1.044570803642273,
      "learning_rate": 0.00032031657762215135,
      "loss": 2.2509,
      "step": 110231
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.9208086729049683,
      "learning_rate": 0.00032031249644905987,
      "loss": 1.9948,
      "step": 110232
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.8740227818489075,
      "learning_rate": 0.000320308415272192,
      "loss": 2.2063,
      "step": 110233
    },
    {
      "epoch": 1.44,
      "grad_norm": 1.0098804235458374,
      "learning_rate": 0.0003203043340915483,
      "loss": 2.0844,
      "step": 110234
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.8355504870414734,
      "learning_rate": 0.0003203002529071297,
      "loss": 2.3751,
      "step": 110235
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.866603434085846,
      "learning_rate": 0.0003202961717189369,
      "loss": 2.3118,
      "step": 110236
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.8873279690742493,
      "learning_rate": 0.00032029209052697076,
      "loss": 2.2223,
      "step": 110237
    },
    {
      "epoch": 1.44,
      "grad_norm": 1.1288681030273438,
      "learning_rate": 0.00032028800933123176,
      "loss": 2.281,
      "step": 110238
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.9651355743408203,
      "learning_rate": 0.0003202839281317209,
      "loss": 2.1698,
      "step": 110239
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.9394705891609192,
      "learning_rate": 0.0003202798469284389,
      "loss": 2.2892,
      "step": 110240
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.9096487164497375,
      "learning_rate": 0.00032027576572138654,
      "loss": 2.1508,
      "step": 110241
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.9246125817298889,
      "learning_rate": 0.00032027168451056445,
      "loss": 1.9281,
      "step": 110242
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.8669469952583313,
      "learning_rate": 0.0003202676032959735,
      "loss": 1.9973,
      "step": 110243
    },
    {
      "epoch": 1.44,
      "grad_norm": 1.0080429315567017,
      "learning_rate": 0.0003202635220776144,
      "loss": 2.1345,
      "step": 110244
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.8389986753463745,
      "learning_rate": 0.000320259440855488,
      "loss": 2.3981,
      "step": 110245
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.9656248092651367,
      "learning_rate": 0.00032025535962959485,
      "loss": 2.1388,
      "step": 110246
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.902944803237915,
      "learning_rate": 0.000320251278399936,
      "loss": 2.2468,
      "step": 110247
    },
    {
      "epoch": 1.44,
      "grad_norm": 1.1138285398483276,
      "learning_rate": 0.00032024719716651197,
      "loss": 1.9909,
      "step": 110248
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.9768240451812744,
      "learning_rate": 0.00032024311592932365,
      "loss": 2.2666,
      "step": 110249
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.9311357140541077,
      "learning_rate": 0.0003202390346883717,
      "loss": 1.9245,
      "step": 110250
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.8303659558296204,
      "learning_rate": 0.000320234953443657,
      "loss": 2.1183,
      "step": 110251
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.8651261329650879,
      "learning_rate": 0.00032023087219518014,
      "loss": 2.1045,
      "step": 110252
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.913008987903595,
      "learning_rate": 0.00032022679094294206,
      "loss": 2.1404,
      "step": 110253
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.9152083396911621,
      "learning_rate": 0.00032022270968694343,
      "loss": 2.2362,
      "step": 110254
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.9731010794639587,
      "learning_rate": 0.00032021862842718495,
      "loss": 2.2617,
      "step": 110255
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.8772597312927246,
      "learning_rate": 0.0003202145471636675,
      "loss": 1.9141,
      "step": 110256
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.8895362615585327,
      "learning_rate": 0.0003202104658963918,
      "loss": 2.1585,
      "step": 110257
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.8874733448028564,
      "learning_rate": 0.0003202063846253586,
      "loss": 2.2249,
      "step": 110258
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.8785266280174255,
      "learning_rate": 0.0003202023033505686,
      "loss": 2.307,
      "step": 110259
    },
    {
      "epoch": 1.44,
      "grad_norm": 1.0122982263565063,
      "learning_rate": 0.00032019822207202266,
      "loss": 2.2599,
      "step": 110260
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.9027816653251648,
      "learning_rate": 0.00032019414078972145,
      "loss": 2.0668,
      "step": 110261
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.9163249731063843,
      "learning_rate": 0.00032019005950366576,
      "loss": 2.1484,
      "step": 110262
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.9503639936447144,
      "learning_rate": 0.00032018597821385645,
      "loss": 2.2551,
      "step": 110263
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.9914530515670776,
      "learning_rate": 0.00032018189692029406,
      "loss": 2.2055,
      "step": 110264
    },
    {
      "epoch": 1.44,
      "grad_norm": 1.0234639644622803,
      "learning_rate": 0.0003201778156229796,
      "loss": 2.4745,
      "step": 110265
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.9072904586791992,
      "learning_rate": 0.00032017373432191354,
      "loss": 2.0974,
      "step": 110266
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.8198904991149902,
      "learning_rate": 0.0003201696530170969,
      "loss": 2.1937,
      "step": 110267
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.8798618912696838,
      "learning_rate": 0.00032016557170853037,
      "loss": 2.3458,
      "step": 110268
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.8681039810180664,
      "learning_rate": 0.0003201614903962146,
      "loss": 2.2281,
      "step": 110269
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.8498166799545288,
      "learning_rate": 0.00032015740908015056,
      "loss": 2.1505,
      "step": 110270
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.9347525238990784,
      "learning_rate": 0.0003201533277603387,
      "loss": 1.9868,
      "step": 110271
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.9671807289123535,
      "learning_rate": 0.00032014924643678004,
      "loss": 1.969,
      "step": 110272
    },
    {
      "epoch": 1.44,
      "grad_norm": 1.031009554862976,
      "learning_rate": 0.00032014516510947533,
      "loss": 2.2072,
      "step": 110273
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.952965497970581,
      "learning_rate": 0.00032014108377842514,
      "loss": 2.3902,
      "step": 110274
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.8809810876846313,
      "learning_rate": 0.0003201370024436304,
      "loss": 1.9668,
      "step": 110275
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.9072394371032715,
      "learning_rate": 0.0003201329211050918,
      "loss": 1.9876,
      "step": 110276
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.9208997488021851,
      "learning_rate": 0.0003201288397628101,
      "loss": 2.315,
      "step": 110277
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.8621065020561218,
      "learning_rate": 0.0003201247584167861,
      "loss": 2.1547,
      "step": 110278
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.8895667195320129,
      "learning_rate": 0.0003201206770670205,
      "loss": 2.2272,
      "step": 110279
    },
    {
      "epoch": 1.44,
      "grad_norm": 1.0298573970794678,
      "learning_rate": 0.0003201165957135141,
      "loss": 2.0022,
      "step": 110280
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.9317494034767151,
      "learning_rate": 0.0003201125143562676,
      "loss": 2.2532,
      "step": 110281
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.9557157158851624,
      "learning_rate": 0.00032010843299528193,
      "loss": 2.2707,
      "step": 110282
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.9003555178642273,
      "learning_rate": 0.0003201043516305576,
      "loss": 2.1818,
      "step": 110283
    },
    {
      "epoch": 1.44,
      "grad_norm": 1.0862044095993042,
      "learning_rate": 0.00032010027026209545,
      "loss": 2.2609,
      "step": 110284
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.811797022819519,
      "learning_rate": 0.00032009618888989637,
      "loss": 2.2906,
      "step": 110285
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.87659752368927,
      "learning_rate": 0.000320092107513961,
      "loss": 1.9888,
      "step": 110286
    },
    {
      "epoch": 1.44,
      "grad_norm": 1.2980107069015503,
      "learning_rate": 0.00032008802613429016,
      "loss": 2.0838,
      "step": 110287
    },
    {
      "epoch": 1.44,
      "grad_norm": 1.0958977937698364,
      "learning_rate": 0.00032008394475088465,
      "loss": 2.1393,
      "step": 110288
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.8665558695793152,
      "learning_rate": 0.000320079863363745,
      "loss": 2.1634,
      "step": 110289
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.9900644421577454,
      "learning_rate": 0.00032007578197287216,
      "loss": 2.2858,
      "step": 110290
    },
    {
      "epoch": 1.44,
      "grad_norm": 1.148233413696289,
      "learning_rate": 0.00032007170057826695,
      "loss": 2.039,
      "step": 110291
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.9924670457839966,
      "learning_rate": 0.00032006761917992993,
      "loss": 2.1959,
      "step": 110292
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.9042166471481323,
      "learning_rate": 0.00032006353777786194,
      "loss": 2.2156,
      "step": 110293
    },
    {
      "epoch": 1.44,
      "grad_norm": 1.2330334186553955,
      "learning_rate": 0.00032005945637206387,
      "loss": 2.0921,
      "step": 110294
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.9962117671966553,
      "learning_rate": 0.0003200553749625363,
      "loss": 2.1583,
      "step": 110295
    },
    {
      "epoch": 1.44,
      "grad_norm": 1.1275817155838013,
      "learning_rate": 0.00032005129354928006,
      "loss": 2.1349,
      "step": 110296
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.9375355243682861,
      "learning_rate": 0.00032004721213229595,
      "loss": 2.288,
      "step": 110297
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.9425287842750549,
      "learning_rate": 0.0003200431307115846,
      "loss": 2.1919,
      "step": 110298
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.9494035243988037,
      "learning_rate": 0.00032003904928714694,
      "loss": 2.2984,
      "step": 110299
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.9371391534805298,
      "learning_rate": 0.0003200349678589836,
      "loss": 2.1494,
      "step": 110300
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.8825923204421997,
      "learning_rate": 0.0003200308864270954,
      "loss": 2.1448,
      "step": 110301
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.9210401177406311,
      "learning_rate": 0.00032002680499148303,
      "loss": 2.039,
      "step": 110302
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.9854390621185303,
      "learning_rate": 0.0003200227235521474,
      "loss": 1.9368,
      "step": 110303
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.8998912572860718,
      "learning_rate": 0.00032001864210908903,
      "loss": 2.0979,
      "step": 110304
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.9389166831970215,
      "learning_rate": 0.0003200145606623089,
      "loss": 2.0898,
      "step": 110305
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.8676491975784302,
      "learning_rate": 0.00032001047921180774,
      "loss": 2.186,
      "step": 110306
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.9656858444213867,
      "learning_rate": 0.00032000639775758616,
      "loss": 2.1692,
      "step": 110307
    },
    {
      "epoch": 1.44,
      "grad_norm": 1.0777242183685303,
      "learning_rate": 0.00032000231629964506,
      "loss": 2.0672,
      "step": 110308
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.8268076777458191,
      "learning_rate": 0.00031999823483798515,
      "loss": 1.8453,
      "step": 110309
    },
    {
      "epoch": 1.44,
      "grad_norm": 1.0481053590774536,
      "learning_rate": 0.00031999415337260714,
      "loss": 2.1208,
      "step": 110310
    },
    {
      "epoch": 1.44,
      "grad_norm": 1.0291168689727783,
      "learning_rate": 0.0003199900719035118,
      "loss": 2.0102,
      "step": 110311
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.9014352560043335,
      "learning_rate": 0.00031998599043070013,
      "loss": 2.3365,
      "step": 110312
    },
    {
      "epoch": 1.44,
      "grad_norm": 1.1204715967178345,
      "learning_rate": 0.0003199819089541725,
      "loss": 2.2735,
      "step": 110313
    },
    {
      "epoch": 1.44,
      "grad_norm": 1.0146008729934692,
      "learning_rate": 0.0003199778274739299,
      "loss": 2.4373,
      "step": 110314
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.954633891582489,
      "learning_rate": 0.0003199737459899731,
      "loss": 2.0677,
      "step": 110315
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.8946203589439392,
      "learning_rate": 0.00031996966450230276,
      "loss": 2.2837,
      "step": 110316
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.9957519173622131,
      "learning_rate": 0.00031996558301091967,
      "loss": 2.0964,
      "step": 110317
    },
    {
      "epoch": 1.44,
      "grad_norm": 1.082721471786499,
      "learning_rate": 0.00031996150151582465,
      "loss": 2.1596,
      "step": 110318
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.8786386251449585,
      "learning_rate": 0.0003199574200170184,
      "loss": 2.0415,
      "step": 110319
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.881242573261261,
      "learning_rate": 0.00031995333851450163,
      "loss": 2.1711,
      "step": 110320
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.9558143019676208,
      "learning_rate": 0.00031994925700827516,
      "loss": 2.0343,
      "step": 110321
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.923864483833313,
      "learning_rate": 0.00031994517549833985,
      "loss": 2.1531,
      "step": 110322
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.8476182818412781,
      "learning_rate": 0.0003199410939846963,
      "loss": 2.1836,
      "step": 110323
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.9287298917770386,
      "learning_rate": 0.00031993701246734525,
      "loss": 2.2556,
      "step": 110324
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.8570573329925537,
      "learning_rate": 0.00031993293094628765,
      "loss": 2.1696,
      "step": 110325
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.9126460552215576,
      "learning_rate": 0.0003199288494215241,
      "loss": 2.3235,
      "step": 110326
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.925739586353302,
      "learning_rate": 0.00031992476789305537,
      "loss": 2.3696,
      "step": 110327
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.910773754119873,
      "learning_rate": 0.00031992068636088215,
      "loss": 2.2328,
      "step": 110328
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.8748080730438232,
      "learning_rate": 0.0003199166048250055,
      "loss": 2.0462,
      "step": 110329
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.961741030216217,
      "learning_rate": 0.00031991252328542584,
      "loss": 2.477,
      "step": 110330
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.9431821703910828,
      "learning_rate": 0.00031990844174214415,
      "loss": 2.2917,
      "step": 110331
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.9088668823242188,
      "learning_rate": 0.0003199043601951611,
      "loss": 2.4219,
      "step": 110332
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.9313046336174011,
      "learning_rate": 0.0003199002786444774,
      "loss": 2.4029,
      "step": 110333
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.8674452304840088,
      "learning_rate": 0.0003198961970900939,
      "loss": 2.0814,
      "step": 110334
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.886513888835907,
      "learning_rate": 0.00031989211553201126,
      "loss": 2.2736,
      "step": 110335
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.9095631241798401,
      "learning_rate": 0.0003198880339702304,
      "loss": 1.8774,
      "step": 110336
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.9560219049453735,
      "learning_rate": 0.00031988395240475193,
      "loss": 2.1894,
      "step": 110337
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.940671443939209,
      "learning_rate": 0.00031987987083557663,
      "loss": 2.4509,
      "step": 110338
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.9280558824539185,
      "learning_rate": 0.0003198757892627053,
      "loss": 2.2353,
      "step": 110339
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.8245612382888794,
      "learning_rate": 0.00031987170768613875,
      "loss": 1.9061,
      "step": 110340
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.9337233901023865,
      "learning_rate": 0.0003198676261058776,
      "loss": 2.3388,
      "step": 110341
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.8911437392234802,
      "learning_rate": 0.0003198635445219227,
      "loss": 2.3136,
      "step": 110342
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.8876889944076538,
      "learning_rate": 0.0003198594629342748,
      "loss": 2.0227,
      "step": 110343
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.9373772740364075,
      "learning_rate": 0.0003198553813429346,
      "loss": 2.2576,
      "step": 110344
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.9292964339256287,
      "learning_rate": 0.00031985129974790294,
      "loss": 2.3303,
      "step": 110345
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.8890723586082458,
      "learning_rate": 0.0003198472181491807,
      "loss": 1.9624,
      "step": 110346
    },
    {
      "epoch": 1.44,
      "grad_norm": 1.0571268796920776,
      "learning_rate": 0.0003198431365467683,
      "loss": 2.1379,
      "step": 110347
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.8591718077659607,
      "learning_rate": 0.00031983905494066675,
      "loss": 2.119,
      "step": 110348
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.9470207691192627,
      "learning_rate": 0.0003198349733308768,
      "loss": 2.1083,
      "step": 110349
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.9432038068771362,
      "learning_rate": 0.00031983089171739904,
      "loss": 2.3942,
      "step": 110350
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.8964665532112122,
      "learning_rate": 0.0003198268101002344,
      "loss": 2.1575,
      "step": 110351
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.9107341766357422,
      "learning_rate": 0.0003198227284793836,
      "loss": 1.9732,
      "step": 110352
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.9007441401481628,
      "learning_rate": 0.00031981864685484733,
      "loss": 2.3326,
      "step": 110353
    },
    {
      "epoch": 1.44,
      "grad_norm": 1.0456790924072266,
      "learning_rate": 0.0003198145652266264,
      "loss": 2.1905,
      "step": 110354
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.9782411456108093,
      "learning_rate": 0.00031981048359472163,
      "loss": 2.3227,
      "step": 110355
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.9504125118255615,
      "learning_rate": 0.0003198064019591337,
      "loss": 2.2099,
      "step": 110356
    },
    {
      "epoch": 1.44,
      "grad_norm": 1.1972755193710327,
      "learning_rate": 0.0003198023203198633,
      "loss": 2.1355,
      "step": 110357
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.9139357209205627,
      "learning_rate": 0.0003197982386769114,
      "loss": 2.4136,
      "step": 110358
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.8842049241065979,
      "learning_rate": 0.0003197941570302785,
      "loss": 2.1034,
      "step": 110359
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.9524077773094177,
      "learning_rate": 0.0003197900753799656,
      "loss": 2.1467,
      "step": 110360
    },
    {
      "epoch": 1.44,
      "grad_norm": 1.1490371227264404,
      "learning_rate": 0.0003197859937259734,
      "loss": 2.1472,
      "step": 110361
    },
    {
      "epoch": 1.44,
      "grad_norm": 1.0077431201934814,
      "learning_rate": 0.0003197819120683025,
      "loss": 2.1825,
      "step": 110362
    },
    {
      "epoch": 1.44,
      "grad_norm": 1.1253836154937744,
      "learning_rate": 0.0003197778304069538,
      "loss": 2.1567,
      "step": 110363
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.8831323385238647,
      "learning_rate": 0.0003197737487419281,
      "loss": 2.2654,
      "step": 110364
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.9526957869529724,
      "learning_rate": 0.0003197696670732259,
      "loss": 2.0293,
      "step": 110365
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.8795077204704285,
      "learning_rate": 0.0003197655854008483,
      "loss": 2.2429,
      "step": 110366
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.884567379951477,
      "learning_rate": 0.0003197615037247959,
      "loss": 2.1175,
      "step": 110367
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.944226861000061,
      "learning_rate": 0.0003197574220450694,
      "loss": 2.0842,
      "step": 110368
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.9983950257301331,
      "learning_rate": 0.0003197533403616696,
      "loss": 2.2038,
      "step": 110369
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.9114491939544678,
      "learning_rate": 0.0003197492586745974,
      "loss": 2.2439,
      "step": 110370
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.7978887557983398,
      "learning_rate": 0.0003197451769838533,
      "loss": 2.1534,
      "step": 110371
    },
    {
      "epoch": 1.44,
      "grad_norm": 1.0141979455947876,
      "learning_rate": 0.0003197410952894383,
      "loss": 2.1794,
      "step": 110372
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.9860659837722778,
      "learning_rate": 0.0003197370135913531,
      "loss": 1.9261,
      "step": 110373
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.9009250998497009,
      "learning_rate": 0.0003197329318895983,
      "loss": 2.0785,
      "step": 110374
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.8747456669807434,
      "learning_rate": 0.00031972885018417477,
      "loss": 2.137,
      "step": 110375
    },
    {
      "epoch": 1.44,
      "grad_norm": 1.269411563873291,
      "learning_rate": 0.0003197247684750834,
      "loss": 2.0541,
      "step": 110376
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.9856699705123901,
      "learning_rate": 0.00031972068676232475,
      "loss": 2.042,
      "step": 110377
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.8454340696334839,
      "learning_rate": 0.0003197166050458996,
      "loss": 2.299,
      "step": 110378
    },
    {
      "epoch": 1.44,
      "grad_norm": 1.255908727645874,
      "learning_rate": 0.0003197125233258089,
      "loss": 1.9364,
      "step": 110379
    },
    {
      "epoch": 1.44,
      "grad_norm": 1.1115399599075317,
      "learning_rate": 0.0003197084416020531,
      "loss": 2.2152,
      "step": 110380
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.8519649505615234,
      "learning_rate": 0.00031970435987463327,
      "loss": 2.0835,
      "step": 110381
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.9251327514648438,
      "learning_rate": 0.00031970027814355,
      "loss": 2.1846,
      "step": 110382
    },
    {
      "epoch": 1.44,
      "grad_norm": 1.0472662448883057,
      "learning_rate": 0.000319696196408804,
      "loss": 2.2004,
      "step": 110383
    },
    {
      "epoch": 1.44,
      "grad_norm": 1.065493106842041,
      "learning_rate": 0.0003196921146703961,
      "loss": 1.9201,
      "step": 110384
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.8942445516586304,
      "learning_rate": 0.0003196880329283272,
      "loss": 2.1613,
      "step": 110385
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.9293552041053772,
      "learning_rate": 0.00031968395118259777,
      "loss": 1.969,
      "step": 110386
    },
    {
      "epoch": 1.44,
      "grad_norm": 1.438044548034668,
      "learning_rate": 0.0003196798694332088,
      "loss": 2.1695,
      "step": 110387
    },
    {
      "epoch": 1.44,
      "grad_norm": 1.0181058645248413,
      "learning_rate": 0.000319675787680161,
      "loss": 2.1972,
      "step": 110388
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.915450930595398,
      "learning_rate": 0.00031967170592345505,
      "loss": 2.0266,
      "step": 110389
    },
    {
      "epoch": 1.44,
      "grad_norm": 1.0517418384552002,
      "learning_rate": 0.00031966762416309175,
      "loss": 1.9835,
      "step": 110390
    },
    {
      "epoch": 1.44,
      "grad_norm": 1.1601239442825317,
      "learning_rate": 0.00031966354239907187,
      "loss": 2.0395,
      "step": 110391
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.9876447319984436,
      "learning_rate": 0.0003196594606313962,
      "loss": 2.2634,
      "step": 110392
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.8782359957695007,
      "learning_rate": 0.0003196553788600654,
      "loss": 2.0743,
      "step": 110393
    },
    {
      "epoch": 1.44,
      "grad_norm": 1.0812562704086304,
      "learning_rate": 0.0003196512970850804,
      "loss": 2.3935,
      "step": 110394
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.9797073602676392,
      "learning_rate": 0.0003196472153064418,
      "loss": 1.8903,
      "step": 110395
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.8437913060188293,
      "learning_rate": 0.0003196431335241504,
      "loss": 2.0783,
      "step": 110396
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.8384352922439575,
      "learning_rate": 0.00031963905173820696,
      "loss": 2.0903,
      "step": 110397
    },
    {
      "epoch": 1.44,
      "grad_norm": 1.085666537284851,
      "learning_rate": 0.00031963496994861227,
      "loss": 2.4199,
      "step": 110398
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.8778594136238098,
      "learning_rate": 0.00031963088815536707,
      "loss": 1.9773,
      "step": 110399
    },
    {
      "epoch": 1.44,
      "grad_norm": 1.0297225713729858,
      "learning_rate": 0.00031962680635847207,
      "loss": 2.0507,
      "step": 110400
    },
    {
      "epoch": 1.44,
      "grad_norm": 1.0002355575561523,
      "learning_rate": 0.0003196227245579281,
      "loss": 2.0646,
      "step": 110401
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.9663990139961243,
      "learning_rate": 0.00031961864275373595,
      "loss": 2.1854,
      "step": 110402
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.9155376553535461,
      "learning_rate": 0.0003196145609458963,
      "loss": 2.1063,
      "step": 110403
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.9680101275444031,
      "learning_rate": 0.00031961047913440995,
      "loss": 1.9225,
      "step": 110404
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.8848788142204285,
      "learning_rate": 0.0003196063973192776,
      "loss": 2.294,
      "step": 110405
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.9869193434715271,
      "learning_rate": 0.0003196023155005,
      "loss": 2.1477,
      "step": 110406
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.860209047794342,
      "learning_rate": 0.00031959823367807807,
      "loss": 2.224,
      "step": 110407
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.8187234401702881,
      "learning_rate": 0.0003195941518520124,
      "loss": 2.144,
      "step": 110408
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.9979766011238098,
      "learning_rate": 0.00031959007002230384,
      "loss": 2.3134,
      "step": 110409
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.9097248315811157,
      "learning_rate": 0.00031958598818895313,
      "loss": 2.1364,
      "step": 110410
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.8742826581001282,
      "learning_rate": 0.0003195819063519609,
      "loss": 2.063,
      "step": 110411
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.868343710899353,
      "learning_rate": 0.0003195778245113281,
      "loss": 2.2243,
      "step": 110412
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.9985143542289734,
      "learning_rate": 0.0003195737426670555,
      "loss": 2.104,
      "step": 110413
    },
    {
      "epoch": 1.44,
      "grad_norm": 1.0609383583068848,
      "learning_rate": 0.0003195696608191436,
      "loss": 2.1012,
      "step": 110414
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.8402588963508606,
      "learning_rate": 0.00031956557896759344,
      "loss": 2.1987,
      "step": 110415
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.9213231205940247,
      "learning_rate": 0.0003195614971124056,
      "loss": 2.2341,
      "step": 110416
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.9524912238121033,
      "learning_rate": 0.0003195574152535811,
      "loss": 2.114,
      "step": 110417
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.9563514590263367,
      "learning_rate": 0.0003195533333911203,
      "loss": 2.1641,
      "step": 110418
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.8415654301643372,
      "learning_rate": 0.00031954925152502414,
      "loss": 2.0956,
      "step": 110419
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.8358383774757385,
      "learning_rate": 0.0003195451696552935,
      "loss": 2.2337,
      "step": 110420
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.9105300307273865,
      "learning_rate": 0.0003195410877819291,
      "loss": 2.0446,
      "step": 110421
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.9253018498420715,
      "learning_rate": 0.00031953700590493155,
      "loss": 2.052,
      "step": 110422
    },
    {
      "epoch": 1.44,
      "grad_norm": 1.1564288139343262,
      "learning_rate": 0.0003195329240243017,
      "loss": 2.0265,
      "step": 110423
    },
    {
      "epoch": 1.44,
      "grad_norm": 1.0095134973526,
      "learning_rate": 0.0003195288421400404,
      "loss": 2.0941,
      "step": 110424
    },
    {
      "epoch": 1.44,
      "grad_norm": 1.062138557434082,
      "learning_rate": 0.0003195247602521482,
      "loss": 2.2331,
      "step": 110425
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.8816636800765991,
      "learning_rate": 0.0003195206783606261,
      "loss": 2.1346,
      "step": 110426
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.960399866104126,
      "learning_rate": 0.0003195165964654746,
      "loss": 2.0606,
      "step": 110427
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.91935133934021,
      "learning_rate": 0.00031951251456669474,
      "loss": 2.0185,
      "step": 110428
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.912652850151062,
      "learning_rate": 0.0003195084326642871,
      "loss": 2.343,
      "step": 110429
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.9361125230789185,
      "learning_rate": 0.0003195043507582524,
      "loss": 2.1865,
      "step": 110430
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.9020498394966125,
      "learning_rate": 0.00031950026884859157,
      "loss": 2.2441,
      "step": 110431
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.9333561658859253,
      "learning_rate": 0.0003194961869353052,
      "loss": 2.2337,
      "step": 110432
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.8904590606689453,
      "learning_rate": 0.0003194921050183941,
      "loss": 2.1836,
      "step": 110433
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.9213272929191589,
      "learning_rate": 0.0003194880230978591,
      "loss": 2.3236,
      "step": 110434
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.9392653107643127,
      "learning_rate": 0.0003194839411737009,
      "loss": 2.2638,
      "step": 110435
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.886539876461029,
      "learning_rate": 0.00031947985924592025,
      "loss": 2.1339,
      "step": 110436
    },
    {
      "epoch": 1.44,
      "grad_norm": 1.0131480693817139,
      "learning_rate": 0.00031947577731451804,
      "loss": 2.2536,
      "step": 110437
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.9253475666046143,
      "learning_rate": 0.00031947169537949475,
      "loss": 2.025,
      "step": 110438
    },
    {
      "epoch": 1.44,
      "grad_norm": 1.0071104764938354,
      "learning_rate": 0.0003194676134408513,
      "loss": 2.0899,
      "step": 110439
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.7975706458091736,
      "learning_rate": 0.0003194635314985886,
      "loss": 2.0982,
      "step": 110440
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.8412433862686157,
      "learning_rate": 0.0003194594495527072,
      "loss": 2.3147,
      "step": 110441
    },
    {
      "epoch": 1.44,
      "grad_norm": 1.0126395225524902,
      "learning_rate": 0.00031945536760320787,
      "loss": 2.1888,
      "step": 110442
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.8466006517410278,
      "learning_rate": 0.00031945128565009147,
      "loss": 2.1677,
      "step": 110443
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.8121688365936279,
      "learning_rate": 0.0003194472036933587,
      "loss": 2.0856,
      "step": 110444
    },
    {
      "epoch": 1.44,
      "grad_norm": 1.0406293869018555,
      "learning_rate": 0.0003194431217330103,
      "loss": 2.3978,
      "step": 110445
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.9159572124481201,
      "learning_rate": 0.00031943903976904713,
      "loss": 2.1447,
      "step": 110446
    },
    {
      "epoch": 1.44,
      "grad_norm": 1.0065051317214966,
      "learning_rate": 0.0003194349578014698,
      "loss": 2.3355,
      "step": 110447
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.9681901931762695,
      "learning_rate": 0.0003194308758302792,
      "loss": 2.3828,
      "step": 110448
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.937577486038208,
      "learning_rate": 0.000319426793855476,
      "loss": 2.1553,
      "step": 110449
    },
    {
      "epoch": 1.44,
      "grad_norm": 1.0270512104034424,
      "learning_rate": 0.000319422711877061,
      "loss": 1.9687,
      "step": 110450
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.9311134219169617,
      "learning_rate": 0.00031941862989503497,
      "loss": 2.1943,
      "step": 110451
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.8729012608528137,
      "learning_rate": 0.00031941454790939864,
      "loss": 2.2103,
      "step": 110452
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.9212793111801147,
      "learning_rate": 0.0003194104659201528,
      "loss": 2.001,
      "step": 110453
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.9211863875389099,
      "learning_rate": 0.0003194063839272981,
      "loss": 2.1412,
      "step": 110454
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.7822332382202148,
      "learning_rate": 0.00031940230193083547,
      "loss": 1.9088,
      "step": 110455
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.9843958020210266,
      "learning_rate": 0.00031939821993076553,
      "loss": 2.0632,
      "step": 110456
    },
    {
      "epoch": 1.44,
      "grad_norm": 1.0486762523651123,
      "learning_rate": 0.00031939413792708915,
      "loss": 2.1761,
      "step": 110457
    },
    {
      "epoch": 1.44,
      "grad_norm": 1.0001007318496704,
      "learning_rate": 0.000319390055919807,
      "loss": 2.1886,
      "step": 110458
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.9106096625328064,
      "learning_rate": 0.00031938597390891987,
      "loss": 2.2926,
      "step": 110459
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.9676781892776489,
      "learning_rate": 0.00031938189189442847,
      "loss": 2.2503,
      "step": 110460
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.9168407917022705,
      "learning_rate": 0.0003193778098763337,
      "loss": 2.2205,
      "step": 110461
    },
    {
      "epoch": 1.44,
      "grad_norm": 1.100203275680542,
      "learning_rate": 0.00031937372785463623,
      "loss": 2.3398,
      "step": 110462
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.9979983568191528,
      "learning_rate": 0.0003193696458293368,
      "loss": 2.163,
      "step": 110463
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.9909910559654236,
      "learning_rate": 0.00031936556380043614,
      "loss": 2.0251,
      "step": 110464
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.891686201095581,
      "learning_rate": 0.0003193614817679351,
      "loss": 2.1135,
      "step": 110465
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.8791161179542542,
      "learning_rate": 0.00031935739973183437,
      "loss": 1.9668,
      "step": 110466
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.9385514259338379,
      "learning_rate": 0.0003193533176921348,
      "loss": 2.1303,
      "step": 110467
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.9213405847549438,
      "learning_rate": 0.00031934923564883705,
      "loss": 1.9826,
      "step": 110468
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.8497859239578247,
      "learning_rate": 0.00031934515360194183,
      "loss": 2.1508,
      "step": 110469
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.9962509274482727,
      "learning_rate": 0.0003193410715514501,
      "loss": 2.5342,
      "step": 110470
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.9209761023521423,
      "learning_rate": 0.00031933698949736244,
      "loss": 2.2829,
      "step": 110471
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.9373810291290283,
      "learning_rate": 0.00031933290743967963,
      "loss": 2.1399,
      "step": 110472
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.80565345287323,
      "learning_rate": 0.0003193288253784025,
      "loss": 2.003,
      "step": 110473
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.9209486842155457,
      "learning_rate": 0.0003193247433135319,
      "loss": 2.0437,
      "step": 110474
    },
    {
      "epoch": 1.44,
      "grad_norm": 1.0204747915267944,
      "learning_rate": 0.0003193206612450683,
      "loss": 2.5491,
      "step": 110475
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.8543139696121216,
      "learning_rate": 0.0003193165791730127,
      "loss": 2.0542,
      "step": 110476
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.9695388674736023,
      "learning_rate": 0.0003193124970973658,
      "loss": 2.1024,
      "step": 110477
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.9518042802810669,
      "learning_rate": 0.0003193084150181283,
      "loss": 1.9738,
      "step": 110478
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.8855034708976746,
      "learning_rate": 0.00031930433293530093,
      "loss": 1.9465,
      "step": 110479
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.948538601398468,
      "learning_rate": 0.00031930025084888464,
      "loss": 2.1237,
      "step": 110480
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.8366464972496033,
      "learning_rate": 0.00031929616875888,
      "loss": 2.0596,
      "step": 110481
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.9292606711387634,
      "learning_rate": 0.0003192920866652879,
      "loss": 2.2436,
      "step": 110482
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.8417741656303406,
      "learning_rate": 0.000319288004568109,
      "loss": 2.1718,
      "step": 110483
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.8453789353370667,
      "learning_rate": 0.0003192839224673441,
      "loss": 2.4228,
      "step": 110484
    },
    {
      "epoch": 1.44,
      "grad_norm": 1.0069342851638794,
      "learning_rate": 0.0003192798403629939,
      "loss": 2.0624,
      "step": 110485
    },
    {
      "epoch": 1.44,
      "grad_norm": 1.0315386056900024,
      "learning_rate": 0.0003192757582550594,
      "loss": 1.9667,
      "step": 110486
    },
    {
      "epoch": 1.44,
      "grad_norm": 1.4222038984298706,
      "learning_rate": 0.00031927167614354094,
      "loss": 2.4122,
      "step": 110487
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.9965313076972961,
      "learning_rate": 0.0003192675940284396,
      "loss": 2.2935,
      "step": 110488
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.9755154848098755,
      "learning_rate": 0.00031926351190975614,
      "loss": 2.0824,
      "step": 110489
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.8733617663383484,
      "learning_rate": 0.0003192594297874911,
      "loss": 2.1101,
      "step": 110490
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.9670546650886536,
      "learning_rate": 0.0003192553476616454,
      "loss": 2.0426,
      "step": 110491
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.9260938763618469,
      "learning_rate": 0.0003192512655322198,
      "loss": 1.9722,
      "step": 110492
    },
    {
      "epoch": 1.44,
      "grad_norm": 1.1463266611099243,
      "learning_rate": 0.00031924718339921496,
      "loss": 2.1691,
      "step": 110493
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.9249184727668762,
      "learning_rate": 0.0003192431012626318,
      "loss": 2.0888,
      "step": 110494
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.9234586954116821,
      "learning_rate": 0.000319239019122471,
      "loss": 1.975,
      "step": 110495
    },
    {
      "epoch": 1.44,
      "grad_norm": 1.031390905380249,
      "learning_rate": 0.0003192349369787332,
      "loss": 2.2048,
      "step": 110496
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.8937516808509827,
      "learning_rate": 0.0003192308548314193,
      "loss": 2.1272,
      "step": 110497
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.9580427408218384,
      "learning_rate": 0.00031922677268053006,
      "loss": 2.2676,
      "step": 110498
    },
    {
      "epoch": 1.44,
      "grad_norm": 1.0606764554977417,
      "learning_rate": 0.00031922269052606615,
      "loss": 2.1333,
      "step": 110499
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.9072449803352356,
      "learning_rate": 0.0003192186083680284,
      "loss": 2.0321,
      "step": 110500
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.9783980250358582,
      "learning_rate": 0.00031921452620641753,
      "loss": 2.3453,
      "step": 110501
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.8883679509162903,
      "learning_rate": 0.0003192104440412344,
      "loss": 2.0982,
      "step": 110502
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.8827873468399048,
      "learning_rate": 0.00031920636187247957,
      "loss": 2.168,
      "step": 110503
    },
    {
      "epoch": 1.44,
      "grad_norm": 1.0073211193084717,
      "learning_rate": 0.0003192022797001539,
      "loss": 2.0309,
      "step": 110504
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.872174084186554,
      "learning_rate": 0.00031919819752425826,
      "loss": 2.1685,
      "step": 110505
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.8891651630401611,
      "learning_rate": 0.00031919411534479325,
      "loss": 2.1153,
      "step": 110506
    },
    {
      "epoch": 1.44,
      "grad_norm": 1.1164461374282837,
      "learning_rate": 0.00031919003316175975,
      "loss": 2.2036,
      "step": 110507
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.8661614060401917,
      "learning_rate": 0.00031918595097515836,
      "loss": 2.1131,
      "step": 110508
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.9422265291213989,
      "learning_rate": 0.00031918186878499005,
      "loss": 2.1521,
      "step": 110509
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.9450317025184631,
      "learning_rate": 0.0003191777865912554,
      "loss": 1.9326,
      "step": 110510
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.9119126796722412,
      "learning_rate": 0.0003191737043939553,
      "loss": 2.3283,
      "step": 110511
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.8626591563224792,
      "learning_rate": 0.00031916962219309035,
      "loss": 2.186,
      "step": 110512
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.9235872030258179,
      "learning_rate": 0.00031916553998866153,
      "loss": 2.0432,
      "step": 110513
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.9908620715141296,
      "learning_rate": 0.00031916145778066937,
      "loss": 2.3464,
      "step": 110514
    },
    {
      "epoch": 1.44,
      "grad_norm": 1.042641043663025,
      "learning_rate": 0.0003191573755691148,
      "loss": 2.0445,
      "step": 110515
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.9753212928771973,
      "learning_rate": 0.0003191532933539985,
      "loss": 2.1813,
      "step": 110516
    },
    {
      "epoch": 1.44,
      "grad_norm": 1.2593610286712646,
      "learning_rate": 0.0003191492111353212,
      "loss": 1.8539,
      "step": 110517
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.9029606580734253,
      "learning_rate": 0.0003191451289130837,
      "loss": 2.1926,
      "step": 110518
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.9278255105018616,
      "learning_rate": 0.00031914104668728685,
      "loss": 2.0996,
      "step": 110519
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.8952733278274536,
      "learning_rate": 0.00031913696445793124,
      "loss": 2.1033,
      "step": 110520
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.936937153339386,
      "learning_rate": 0.0003191328822250177,
      "loss": 2.1208,
      "step": 110521
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.928963840007782,
      "learning_rate": 0.00031912879998854704,
      "loss": 2.0463,
      "step": 110522
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.8953481912612915,
      "learning_rate": 0.00031912471774851996,
      "loss": 2.4746,
      "step": 110523
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.9181541204452515,
      "learning_rate": 0.0003191206355049372,
      "loss": 2.1314,
      "step": 110524
    },
    {
      "epoch": 1.44,
      "grad_norm": 1.0944461822509766,
      "learning_rate": 0.0003191165532577997,
      "loss": 2.1391,
      "step": 110525
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.9145243167877197,
      "learning_rate": 0.00031911247100710786,
      "loss": 2.0683,
      "step": 110526
    },
    {
      "epoch": 1.44,
      "grad_norm": 1.0293742418289185,
      "learning_rate": 0.00031910838875286277,
      "loss": 2.2656,
      "step": 110527
    },
    {
      "epoch": 1.44,
      "grad_norm": 1.0670299530029297,
      "learning_rate": 0.00031910430649506507,
      "loss": 2.2181,
      "step": 110528
    },
    {
      "epoch": 1.44,
      "grad_norm": 1.0377569198608398,
      "learning_rate": 0.0003191002242337155,
      "loss": 2.2488,
      "step": 110529
    },
    {
      "epoch": 1.44,
      "grad_norm": 1.1113735437393188,
      "learning_rate": 0.00031909614196881474,
      "loss": 2.3672,
      "step": 110530
    },
    {
      "epoch": 1.44,
      "grad_norm": 1.164475917816162,
      "learning_rate": 0.0003190920597003637,
      "loss": 2.3581,
      "step": 110531
    },
    {
      "epoch": 1.44,
      "grad_norm": 1.1765981912612915,
      "learning_rate": 0.00031908797742836325,
      "loss": 2.2429,
      "step": 110532
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.9561217427253723,
      "learning_rate": 0.0003190838951528138,
      "loss": 2.1252,
      "step": 110533
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.977172315120697,
      "learning_rate": 0.0003190798128737163,
      "loss": 2.1153,
      "step": 110534
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.9796890020370483,
      "learning_rate": 0.0003190757305910716,
      "loss": 2.3493,
      "step": 110535
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.9844648241996765,
      "learning_rate": 0.0003190716483048803,
      "loss": 1.9615,
      "step": 110536
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.9199008941650391,
      "learning_rate": 0.00031906756601514326,
      "loss": 2.2124,
      "step": 110537
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.9924664497375488,
      "learning_rate": 0.0003190634837218612,
      "loss": 2.1751,
      "step": 110538
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.944324254989624,
      "learning_rate": 0.0003190594014250348,
      "loss": 2.0234,
      "step": 110539
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.9426350593566895,
      "learning_rate": 0.000319055319124665,
      "loss": 2.359,
      "step": 110540
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.9732221364974976,
      "learning_rate": 0.0003190512368207524,
      "loss": 1.8575,
      "step": 110541
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.9926856160163879,
      "learning_rate": 0.0003190471545132978,
      "loss": 2.1166,
      "step": 110542
    },
    {
      "epoch": 1.44,
      "grad_norm": 1.2147761583328247,
      "learning_rate": 0.00031904307220230205,
      "loss": 2.0052,
      "step": 110543
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.9380956888198853,
      "learning_rate": 0.00031903898988776573,
      "loss": 2.0911,
      "step": 110544
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.859691321849823,
      "learning_rate": 0.00031903490756968975,
      "loss": 2.1838,
      "step": 110545
    },
    {
      "epoch": 1.44,
      "grad_norm": 1.064579963684082,
      "learning_rate": 0.00031903082524807483,
      "loss": 2.2442,
      "step": 110546
    },
    {
      "epoch": 1.44,
      "grad_norm": 1.1356165409088135,
      "learning_rate": 0.0003190267429229218,
      "loss": 2.1356,
      "step": 110547
    },
    {
      "epoch": 1.44,
      "grad_norm": 1.0191231966018677,
      "learning_rate": 0.0003190226605942312,
      "loss": 2.093,
      "step": 110548
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.9003205895423889,
      "learning_rate": 0.00031901857826200394,
      "loss": 2.0046,
      "step": 110549
    },
    {
      "epoch": 1.44,
      "grad_norm": 1.080960750579834,
      "learning_rate": 0.0003190144959262409,
      "loss": 2.186,
      "step": 110550
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.9942980408668518,
      "learning_rate": 0.0003190104135869426,
      "loss": 2.3385,
      "step": 110551
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.9021071195602417,
      "learning_rate": 0.0003190063312441099,
      "loss": 2.2697,
      "step": 110552
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.9542184472084045,
      "learning_rate": 0.00031900224889774365,
      "loss": 1.8243,
      "step": 110553
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.8797377943992615,
      "learning_rate": 0.0003189981665478444,
      "loss": 2.1653,
      "step": 110554
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.8940079808235168,
      "learning_rate": 0.00031899408419441307,
      "loss": 2.0476,
      "step": 110555
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.9786497354507446,
      "learning_rate": 0.00031899000183745045,
      "loss": 1.9152,
      "step": 110556
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.882637619972229,
      "learning_rate": 0.0003189859194769572,
      "loss": 2.4923,
      "step": 110557
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.9528936147689819,
      "learning_rate": 0.00031898183711293407,
      "loss": 2.0702,
      "step": 110558
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.9244087934494019,
      "learning_rate": 0.00031897775474538193,
      "loss": 2.0218,
      "step": 110559
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.9352685213088989,
      "learning_rate": 0.0003189736723743013,
      "loss": 2.0144,
      "step": 110560
    },
    {
      "epoch": 1.44,
      "grad_norm": 1.1060620546340942,
      "learning_rate": 0.00031896958999969325,
      "loss": 2.0531,
      "step": 110561
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.9001596570014954,
      "learning_rate": 0.00031896550762155844,
      "loss": 2.3261,
      "step": 110562
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.9359049201011658,
      "learning_rate": 0.0003189614252398974,
      "loss": 2.3488,
      "step": 110563
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.8707785606384277,
      "learning_rate": 0.00031895734285471123,
      "loss": 1.9919,
      "step": 110564
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.9114651679992676,
      "learning_rate": 0.0003189532604660005,
      "loss": 2.3711,
      "step": 110565
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.9419676065444946,
      "learning_rate": 0.00031894917807376603,
      "loss": 1.9076,
      "step": 110566
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.8885359168052673,
      "learning_rate": 0.0003189450956780085,
      "loss": 2.2764,
      "step": 110567
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.9240527749061584,
      "learning_rate": 0.00031894101327872875,
      "loss": 2.2027,
      "step": 110568
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.9336230158805847,
      "learning_rate": 0.00031893693087592747,
      "loss": 2.2241,
      "step": 110569
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.8556444048881531,
      "learning_rate": 0.00031893284846960546,
      "loss": 2.0359,
      "step": 110570
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.8591485619544983,
      "learning_rate": 0.0003189287660597635,
      "loss": 2.1809,
      "step": 110571
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.836591899394989,
      "learning_rate": 0.00031892468364640234,
      "loss": 2.0594,
      "step": 110572
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.9684614539146423,
      "learning_rate": 0.00031892060122952265,
      "loss": 2.0587,
      "step": 110573
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.8111385703086853,
      "learning_rate": 0.00031891651880912536,
      "loss": 2.3314,
      "step": 110574
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.8889559507369995,
      "learning_rate": 0.0003189124363852111,
      "loss": 2.0737,
      "step": 110575
    },
    {
      "epoch": 1.44,
      "grad_norm": 1.0186338424682617,
      "learning_rate": 0.00031890835395778064,
      "loss": 1.9855,
      "step": 110576
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.9071885943412781,
      "learning_rate": 0.0003189042715268348,
      "loss": 2.3616,
      "step": 110577
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.9872525334358215,
      "learning_rate": 0.0003189001890923742,
      "loss": 2.1203,
      "step": 110578
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.975640594959259,
      "learning_rate": 0.00031889610665439977,
      "loss": 2.1389,
      "step": 110579
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.9106523990631104,
      "learning_rate": 0.00031889202421291226,
      "loss": 2.1804,
      "step": 110580
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.908186674118042,
      "learning_rate": 0.00031888794176791226,
      "loss": 2.2972,
      "step": 110581
    },
    {
      "epoch": 1.44,
      "grad_norm": 1.0724717378616333,
      "learning_rate": 0.00031888385931940063,
      "loss": 2.0751,
      "step": 110582
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.922300398349762,
      "learning_rate": 0.0003188797768673782,
      "loss": 1.986,
      "step": 110583
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.9473940134048462,
      "learning_rate": 0.0003188756944118456,
      "loss": 2.0852,
      "step": 110584
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.9401952028274536,
      "learning_rate": 0.00031887161195280373,
      "loss": 2.3754,
      "step": 110585
    },
    {
      "epoch": 1.44,
      "grad_norm": 1.2792208194732666,
      "learning_rate": 0.00031886752949025326,
      "loss": 2.2067,
      "step": 110586
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.899278461933136,
      "learning_rate": 0.0003188634470241949,
      "loss": 2.2955,
      "step": 110587
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.8035188317298889,
      "learning_rate": 0.0003188593645546295,
      "loss": 2.149,
      "step": 110588
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.9411364793777466,
      "learning_rate": 0.00031885528208155776,
      "loss": 2.1736,
      "step": 110589
    },
    {
      "epoch": 1.44,
      "grad_norm": 1.1822147369384766,
      "learning_rate": 0.00031885119960498053,
      "loss": 2.2092,
      "step": 110590
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.9276168346405029,
      "learning_rate": 0.00031884711712489846,
      "loss": 2.1601,
      "step": 110591
    },
    {
      "epoch": 1.44,
      "grad_norm": 1.0406992435455322,
      "learning_rate": 0.00031884303464131236,
      "loss": 2.0674,
      "step": 110592
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.9203454852104187,
      "learning_rate": 0.0003188389521542229,
      "loss": 2.349,
      "step": 110593
    },
    {
      "epoch": 1.44,
      "grad_norm": 1.0591658353805542,
      "learning_rate": 0.0003188348696636311,
      "loss": 2.3189,
      "step": 110594
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.859104335308075,
      "learning_rate": 0.00031883078716953743,
      "loss": 1.9432,
      "step": 110595
    },
    {
      "epoch": 1.44,
      "grad_norm": 1.0757122039794922,
      "learning_rate": 0.0003188267046719428,
      "loss": 2.0946,
      "step": 110596
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.9880653023719788,
      "learning_rate": 0.00031882262217084795,
      "loss": 2.0354,
      "step": 110597
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.9050339460372925,
      "learning_rate": 0.0003188185396662536,
      "loss": 2.1455,
      "step": 110598
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.9471287131309509,
      "learning_rate": 0.00031881445715816045,
      "loss": 2.3009,
      "step": 110599
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.932715117931366,
      "learning_rate": 0.0003188103746465694,
      "loss": 2.1919,
      "step": 110600
    },
    {
      "epoch": 1.44,
      "grad_norm": 1.0042328834533691,
      "learning_rate": 0.00031880629213148115,
      "loss": 2.1399,
      "step": 110601
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.9567873477935791,
      "learning_rate": 0.0003188022096128965,
      "loss": 2.1328,
      "step": 110602
    },
    {
      "epoch": 1.44,
      "grad_norm": 1.0297945737838745,
      "learning_rate": 0.000318798127090816,
      "loss": 2.2059,
      "step": 110603
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.8935117721557617,
      "learning_rate": 0.00031879404456524083,
      "loss": 2.3992,
      "step": 110604
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.9858793020248413,
      "learning_rate": 0.0003187899620361713,
      "loss": 2.0537,
      "step": 110605
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.8538328409194946,
      "learning_rate": 0.00031878587950360843,
      "loss": 1.8845,
      "step": 110606
    },
    {
      "epoch": 1.44,
      "grad_norm": 1.004862904548645,
      "learning_rate": 0.00031878179696755285,
      "loss": 1.9214,
      "step": 110607
    },
    {
      "epoch": 1.44,
      "grad_norm": 1.0149035453796387,
      "learning_rate": 0.00031877771442800543,
      "loss": 2.6283,
      "step": 110608
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.850477933883667,
      "learning_rate": 0.0003187736318849669,
      "loss": 2.1302,
      "step": 110609
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.8865593075752258,
      "learning_rate": 0.0003187695493384379,
      "loss": 2.3934,
      "step": 110610
    },
    {
      "epoch": 1.44,
      "grad_norm": 1.096177101135254,
      "learning_rate": 0.00031876546678841943,
      "loss": 2.066,
      "step": 110611
    },
    {
      "epoch": 1.44,
      "grad_norm": 1.1737370491027832,
      "learning_rate": 0.000318761384234912,
      "loss": 2.0395,
      "step": 110612
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.8645176887512207,
      "learning_rate": 0.00031875730167791656,
      "loss": 2.2754,
      "step": 110613
    },
    {
      "epoch": 1.44,
      "grad_norm": 1.1361432075500488,
      "learning_rate": 0.00031875321911743374,
      "loss": 2.1585,
      "step": 110614
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.9586960077285767,
      "learning_rate": 0.0003187491365534643,
      "loss": 2.1282,
      "step": 110615
    },
    {
      "epoch": 1.44,
      "grad_norm": 1.124346375465393,
      "learning_rate": 0.0003187450539860091,
      "loss": 2.0171,
      "step": 110616
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.8832857012748718,
      "learning_rate": 0.0003187409714150688,
      "loss": 2.0705,
      "step": 110617
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.8729051947593689,
      "learning_rate": 0.0003187368888406442,
      "loss": 2.269,
      "step": 110618
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.9189462661743164,
      "learning_rate": 0.00031873280626273606,
      "loss": 2.117,
      "step": 110619
    },
    {
      "epoch": 1.44,
      "grad_norm": 1.0604394674301147,
      "learning_rate": 0.0003187287236813452,
      "loss": 2.3092,
      "step": 110620
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.9934595823287964,
      "learning_rate": 0.00031872464109647223,
      "loss": 2.1307,
      "step": 110621
    },
    {
      "epoch": 1.44,
      "grad_norm": 1.0349044799804688,
      "learning_rate": 0.00031872055850811805,
      "loss": 2.2754,
      "step": 110622
    },
    {
      "epoch": 1.44,
      "grad_norm": 1.0064488649368286,
      "learning_rate": 0.0003187164759162834,
      "loss": 2.173,
      "step": 110623
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.9438447952270508,
      "learning_rate": 0.0003187123933209689,
      "loss": 2.1776,
      "step": 110624
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.866997480392456,
      "learning_rate": 0.0003187083107221755,
      "loss": 2.1849,
      "step": 110625
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.9334389567375183,
      "learning_rate": 0.00031870422811990383,
      "loss": 2.3346,
      "step": 110626
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.8500186204910278,
      "learning_rate": 0.0003187001455141547,
      "loss": 2.4293,
      "step": 110627
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.9400134086608887,
      "learning_rate": 0.0003186960629049288,
      "loss": 2.0603,
      "step": 110628
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.8834065198898315,
      "learning_rate": 0.00031869198029222713,
      "loss": 2.0387,
      "step": 110629
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.8931292295455933,
      "learning_rate": 0.00031868789767605004,
      "loss": 2.0667,
      "step": 110630
    },
    {
      "epoch": 1.44,
      "grad_norm": 1.0685231685638428,
      "learning_rate": 0.0003186838150563986,
      "loss": 2.0642,
      "step": 110631
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.9889134764671326,
      "learning_rate": 0.0003186797324332736,
      "loss": 2.4364,
      "step": 110632
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.9530567526817322,
      "learning_rate": 0.00031867564980667557,
      "loss": 2.3097,
      "step": 110633
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.9406207203865051,
      "learning_rate": 0.0003186715671766054,
      "loss": 2.0046,
      "step": 110634
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.8729618787765503,
      "learning_rate": 0.0003186674845430639,
      "loss": 1.9819,
      "step": 110635
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.878982424736023,
      "learning_rate": 0.00031866340190605163,
      "loss": 2.1101,
      "step": 110636
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.9007799029350281,
      "learning_rate": 0.00031865931926556955,
      "loss": 2.2497,
      "step": 110637
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.9190242886543274,
      "learning_rate": 0.0003186552366216184,
      "loss": 2.0714,
      "step": 110638
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.8609877228736877,
      "learning_rate": 0.0003186511539741988,
      "loss": 2.0993,
      "step": 110639
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.9664031863212585,
      "learning_rate": 0.0003186470713233117,
      "loss": 2.327,
      "step": 110640
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.8504906892776489,
      "learning_rate": 0.0003186429886689577,
      "loss": 1.9658,
      "step": 110641
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.8544808030128479,
      "learning_rate": 0.00031863890601113764,
      "loss": 2.1978,
      "step": 110642
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.9676704406738281,
      "learning_rate": 0.00031863482334985215,
      "loss": 2.236,
      "step": 110643
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.8943679332733154,
      "learning_rate": 0.00031863074068510226,
      "loss": 1.8434,
      "step": 110644
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.9018073678016663,
      "learning_rate": 0.00031862665801688845,
      "loss": 2.1907,
      "step": 110645
    },
    {
      "epoch": 1.44,
      "grad_norm": 1.358325719833374,
      "learning_rate": 0.0003186225753452116,
      "loss": 2.5277,
      "step": 110646
    },
    {
      "epoch": 1.44,
      "grad_norm": 1.0134652853012085,
      "learning_rate": 0.0003186184926700725,
      "loss": 2.1731,
      "step": 110647
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.9044117331504822,
      "learning_rate": 0.00031861440999147187,
      "loss": 2.1628,
      "step": 110648
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.9854736924171448,
      "learning_rate": 0.0003186103273094104,
      "loss": 2.048,
      "step": 110649
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.9438863396644592,
      "learning_rate": 0.00031860624462388905,
      "loss": 1.9606,
      "step": 110650
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.9698541760444641,
      "learning_rate": 0.0003186021619349084,
      "loss": 2.0369,
      "step": 110651
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.9397565126419067,
      "learning_rate": 0.00031859807924246916,
      "loss": 2.1872,
      "step": 110652
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.8808035254478455,
      "learning_rate": 0.00031859399654657225,
      "loss": 2.1432,
      "step": 110653
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.8855518102645874,
      "learning_rate": 0.0003185899138472184,
      "loss": 2.2369,
      "step": 110654
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.9163078665733337,
      "learning_rate": 0.00031858583114440826,
      "loss": 2.1575,
      "step": 110655
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.9960390329360962,
      "learning_rate": 0.0003185817484381428,
      "loss": 2.159,
      "step": 110656
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.8938482999801636,
      "learning_rate": 0.0003185776657284225,
      "loss": 2.0925,
      "step": 110657
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.9624431729316711,
      "learning_rate": 0.0003185735830152483,
      "loss": 2.3117,
      "step": 110658
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.9677615165710449,
      "learning_rate": 0.000318569500298621,
      "loss": 2.3121,
      "step": 110659
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.9814587235450745,
      "learning_rate": 0.0003185654175785412,
      "loss": 2.1731,
      "step": 110660
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.8605827689170837,
      "learning_rate": 0.00031856133485500974,
      "loss": 2.1444,
      "step": 110661
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.9139246344566345,
      "learning_rate": 0.00031855725212802735,
      "loss": 1.9568,
      "step": 110662
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.9611303806304932,
      "learning_rate": 0.0003185531693975949,
      "loss": 2.0207,
      "step": 110663
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.9293677806854248,
      "learning_rate": 0.000318549086663713,
      "loss": 2.1341,
      "step": 110664
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.8721163868904114,
      "learning_rate": 0.00031854500392638243,
      "loss": 1.9192,
      "step": 110665
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.960446298122406,
      "learning_rate": 0.0003185409211856041,
      "loss": 1.7748,
      "step": 110666
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.9130665063858032,
      "learning_rate": 0.0003185368384413786,
      "loss": 2.0766,
      "step": 110667
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.8810608386993408,
      "learning_rate": 0.0003185327556937067,
      "loss": 1.8419,
      "step": 110668
    },
    {
      "epoch": 1.44,
      "grad_norm": 1.01283860206604,
      "learning_rate": 0.00031852867294258934,
      "loss": 2.2561,
      "step": 110669
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.9289442896842957,
      "learning_rate": 0.0003185245901880271,
      "loss": 2.0512,
      "step": 110670
    },
    {
      "epoch": 1.44,
      "grad_norm": 1.0347737073898315,
      "learning_rate": 0.0003185205074300208,
      "loss": 2.2198,
      "step": 110671
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.9936656951904297,
      "learning_rate": 0.0003185164246685712,
      "loss": 2.1001,
      "step": 110672
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.9423344731330872,
      "learning_rate": 0.000318512341903679,
      "loss": 2.1005,
      "step": 110673
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.9506430625915527,
      "learning_rate": 0.000318508259135345,
      "loss": 2.0049,
      "step": 110674
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.823565661907196,
      "learning_rate": 0.00031850417636357,
      "loss": 1.981,
      "step": 110675
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.9742358922958374,
      "learning_rate": 0.0003185000935883547,
      "loss": 2.1434,
      "step": 110676
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.9670246243476868,
      "learning_rate": 0.0003184960108096999,
      "loss": 2.115,
      "step": 110677
    },
    {
      "epoch": 1.44,
      "grad_norm": 1.1756726503372192,
      "learning_rate": 0.0003184919280276064,
      "loss": 2.2395,
      "step": 110678
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.9230495095252991,
      "learning_rate": 0.0003184878452420748,
      "loss": 2.4093,
      "step": 110679
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.9188465476036072,
      "learning_rate": 0.000318483762453106,
      "loss": 2.0065,
      "step": 110680
    },
    {
      "epoch": 1.44,
      "grad_norm": 1.0161969661712646,
      "learning_rate": 0.0003184796796607007,
      "loss": 2.1989,
      "step": 110681
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.9148311018943787,
      "learning_rate": 0.0003184755968648597,
      "loss": 2.0836,
      "step": 110682
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.9140618443489075,
      "learning_rate": 0.0003184715140655838,
      "loss": 2.2165,
      "step": 110683
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.9403654932975769,
      "learning_rate": 0.0003184674312628735,
      "loss": 1.9599,
      "step": 110684
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.8696231245994568,
      "learning_rate": 0.00031846334845672995,
      "loss": 2.1437,
      "step": 110685
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.8667015433311462,
      "learning_rate": 0.0003184592656471537,
      "loss": 2.1276,
      "step": 110686
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.9345396757125854,
      "learning_rate": 0.0003184551828341455,
      "loss": 2.131,
      "step": 110687
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.8888581395149231,
      "learning_rate": 0.00031845110001770606,
      "loss": 2.2254,
      "step": 110688
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.9373266100883484,
      "learning_rate": 0.0003184470171978363,
      "loss": 2.1362,
      "step": 110689
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.9399926662445068,
      "learning_rate": 0.00031844293437453683,
      "loss": 2.1473,
      "step": 110690
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.9484819173812866,
      "learning_rate": 0.00031843885154780844,
      "loss": 2.2428,
      "step": 110691
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.9872553944587708,
      "learning_rate": 0.00031843476871765196,
      "loss": 2.2096,
      "step": 110692
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.8628907799720764,
      "learning_rate": 0.0003184306858840682,
      "loss": 2.1781,
      "step": 110693
    },
    {
      "epoch": 1.44,
      "grad_norm": 1.1648132801055908,
      "learning_rate": 0.00031842660304705773,
      "loss": 2.0339,
      "step": 110694
    },
    {
      "epoch": 1.44,
      "grad_norm": 1.108890175819397,
      "learning_rate": 0.0003184225202066214,
      "loss": 2.1116,
      "step": 110695
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.926597535610199,
      "learning_rate": 0.00031841843736276005,
      "loss": 2.1782,
      "step": 110696
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.8603905439376831,
      "learning_rate": 0.00031841435451547426,
      "loss": 2.0811,
      "step": 110697
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.8715998530387878,
      "learning_rate": 0.00031841027166476494,
      "loss": 2.1373,
      "step": 110698
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.8991683125495911,
      "learning_rate": 0.0003184061888106329,
      "loss": 2.0814,
      "step": 110699
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.9519167542457581,
      "learning_rate": 0.0003184021059530787,
      "loss": 2.1409,
      "step": 110700
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.876064121723175,
      "learning_rate": 0.00031839802309210316,
      "loss": 2.301,
      "step": 110701
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.8438165187835693,
      "learning_rate": 0.0003183939402277072,
      "loss": 2.005,
      "step": 110702
    },
    {
      "epoch": 1.44,
      "grad_norm": 1.0151410102844238,
      "learning_rate": 0.00031838985735989134,
      "loss": 2.282,
      "step": 110703
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.8912612795829773,
      "learning_rate": 0.0003183857744886565,
      "loss": 1.884,
      "step": 110704
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.940387487411499,
      "learning_rate": 0.0003183816916140035,
      "loss": 2.1401,
      "step": 110705
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.940916121006012,
      "learning_rate": 0.00031837760873593287,
      "loss": 2.1311,
      "step": 110706
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.9858530163764954,
      "learning_rate": 0.00031837352585444554,
      "loss": 2.2621,
      "step": 110707
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.8892617225646973,
      "learning_rate": 0.0003183694429695423,
      "loss": 2.3674,
      "step": 110708
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.9702473282814026,
      "learning_rate": 0.00031836536008122366,
      "loss": 2.1936,
      "step": 110709
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.833365261554718,
      "learning_rate": 0.0003183612771894906,
      "loss": 1.8592,
      "step": 110710
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.9957938194274902,
      "learning_rate": 0.00031835719429434403,
      "loss": 2.3074,
      "step": 110711
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.9291899800300598,
      "learning_rate": 0.0003183531113957843,
      "loss": 2.226,
      "step": 110712
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.8412073254585266,
      "learning_rate": 0.0003183490284938124,
      "loss": 1.9588,
      "step": 110713
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.9316092133522034,
      "learning_rate": 0.00031834494558842915,
      "loss": 2.27,
      "step": 110714
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.8577840924263,
      "learning_rate": 0.0003183408626796352,
      "loss": 2.309,
      "step": 110715
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.8859603404998779,
      "learning_rate": 0.0003183367797674313,
      "loss": 2.0747,
      "step": 110716
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.8744801878929138,
      "learning_rate": 0.0003183326968518183,
      "loss": 2.1696,
      "step": 110717
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.9318960309028625,
      "learning_rate": 0.0003183286139327969,
      "loss": 2.2697,
      "step": 110718
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.9326112866401672,
      "learning_rate": 0.00031832453101036777,
      "loss": 1.9085,
      "step": 110719
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.8845282196998596,
      "learning_rate": 0.0003183204480845319,
      "loss": 2.105,
      "step": 110720
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.9214411377906799,
      "learning_rate": 0.0003183163651552898,
      "loss": 2.1768,
      "step": 110721
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.9088009595870972,
      "learning_rate": 0.0003183122822226424,
      "loss": 1.8756,
      "step": 110722
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.8775911927223206,
      "learning_rate": 0.00031830819928659043,
      "loss": 2.1387,
      "step": 110723
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.9375528693199158,
      "learning_rate": 0.00031830411634713456,
      "loss": 2.2098,
      "step": 110724
    },
    {
      "epoch": 1.44,
      "grad_norm": 1.0032111406326294,
      "learning_rate": 0.00031830003340427564,
      "loss": 2.2251,
      "step": 110725
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.905756950378418,
      "learning_rate": 0.0003182959504580144,
      "loss": 2.2446,
      "step": 110726
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.8258225917816162,
      "learning_rate": 0.00031829186750835156,
      "loss": 2.1197,
      "step": 110727
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.8605950474739075,
      "learning_rate": 0.0003182877845552879,
      "loss": 2.0491,
      "step": 110728
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.94488924741745,
      "learning_rate": 0.00031828370159882433,
      "loss": 2.0118,
      "step": 110729
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.9972080588340759,
      "learning_rate": 0.0003182796186389613,
      "loss": 2.0763,
      "step": 110730
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.9638550281524658,
      "learning_rate": 0.00031827553567569984,
      "loss": 2.136,
      "step": 110731
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.9445533156394958,
      "learning_rate": 0.0003182714527090406,
      "loss": 2.184,
      "step": 110732
    },
    {
      "epoch": 1.44,
      "grad_norm": 1.1039509773254395,
      "learning_rate": 0.00031826736973898445,
      "loss": 1.9601,
      "step": 110733
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.9353129863739014,
      "learning_rate": 0.0003182632867655319,
      "loss": 2.1684,
      "step": 110734
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.9401184916496277,
      "learning_rate": 0.0003182592037886839,
      "loss": 2.0998,
      "step": 110735
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.8697434067726135,
      "learning_rate": 0.00031825512080844125,
      "loss": 2.1642,
      "step": 110736
    },
    {
      "epoch": 1.44,
      "grad_norm": 1.0580718517303467,
      "learning_rate": 0.0003182510378248045,
      "loss": 2.0458,
      "step": 110737
    },
    {
      "epoch": 1.44,
      "grad_norm": 1.1686044931411743,
      "learning_rate": 0.0003182469548377746,
      "loss": 2.272,
      "step": 110738
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.9441102147102356,
      "learning_rate": 0.0003182428718473523,
      "loss": 2.046,
      "step": 110739
    },
    {
      "epoch": 1.44,
      "grad_norm": 1.116161823272705,
      "learning_rate": 0.00031823878885353825,
      "loss": 2.0425,
      "step": 110740
    },
    {
      "epoch": 1.44,
      "grad_norm": 1.3591787815093994,
      "learning_rate": 0.00031823470585633327,
      "loss": 2.1144,
      "step": 110741
    },
    {
      "epoch": 1.44,
      "grad_norm": 1.080869197845459,
      "learning_rate": 0.0003182306228557381,
      "loss": 2.1594,
      "step": 110742
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.9535367488861084,
      "learning_rate": 0.0003182265398517535,
      "loss": 2.2339,
      "step": 110743
    },
    {
      "epoch": 1.44,
      "grad_norm": 1.4786291122436523,
      "learning_rate": 0.00031822245684438025,
      "loss": 2.0939,
      "step": 110744
    },
    {
      "epoch": 1.44,
      "grad_norm": 1.1278016567230225,
      "learning_rate": 0.00031821837383361917,
      "loss": 2.1531,
      "step": 110745
    },
    {
      "epoch": 1.44,
      "grad_norm": 1.1311736106872559,
      "learning_rate": 0.0003182142908194709,
      "loss": 2.1199,
      "step": 110746
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.94727623462677,
      "learning_rate": 0.0003182102078019362,
      "loss": 2.2085,
      "step": 110747
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.9609307050704956,
      "learning_rate": 0.000318206124781016,
      "loss": 2.2028,
      "step": 110748
    },
    {
      "epoch": 1.44,
      "grad_norm": 1.0368911027908325,
      "learning_rate": 0.0003182020417567108,
      "loss": 2.1658,
      "step": 110749
    },
    {
      "epoch": 1.44,
      "grad_norm": 1.117850661277771,
      "learning_rate": 0.0003181979587290215,
      "loss": 2.1495,
      "step": 110750
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.8240205645561218,
      "learning_rate": 0.00031819387569794903,
      "loss": 1.9671,
      "step": 110751
    },
    {
      "epoch": 1.44,
      "grad_norm": 1.1382646560668945,
      "learning_rate": 0.0003181897926634938,
      "loss": 2.1077,
      "step": 110752
    },
    {
      "epoch": 1.44,
      "grad_norm": 1.1214689016342163,
      "learning_rate": 0.0003181857096256568,
      "loss": 2.2172,
      "step": 110753
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.998430073261261,
      "learning_rate": 0.0003181816265844388,
      "loss": 2.1208,
      "step": 110754
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.8546985983848572,
      "learning_rate": 0.0003181775435398404,
      "loss": 2.3412,
      "step": 110755
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.9419848918914795,
      "learning_rate": 0.0003181734604918625,
      "loss": 2.2692,
      "step": 110756
    },
    {
      "epoch": 1.44,
      "grad_norm": 1.0437211990356445,
      "learning_rate": 0.0003181693774405058,
      "loss": 2.1543,
      "step": 110757
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.9454172849655151,
      "learning_rate": 0.00031816529438577105,
      "loss": 2.0229,
      "step": 110758
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.9270660877227783,
      "learning_rate": 0.000318161211327659,
      "loss": 2.2185,
      "step": 110759
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.93340003490448,
      "learning_rate": 0.0003181571282661706,
      "loss": 2.1421,
      "step": 110760
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.8308891654014587,
      "learning_rate": 0.00031815304520130624,
      "loss": 2.0685,
      "step": 110761
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.8844109177589417,
      "learning_rate": 0.000318148962133067,
      "loss": 2.1131,
      "step": 110762
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.8332750201225281,
      "learning_rate": 0.0003181448790614535,
      "loss": 2.1083,
      "step": 110763
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.9089399576187134,
      "learning_rate": 0.00031814079598646653,
      "loss": 2.0891,
      "step": 110764
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.8542023301124573,
      "learning_rate": 0.00031813671290810685,
      "loss": 2.0529,
      "step": 110765
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.8886305093765259,
      "learning_rate": 0.00031813262982637513,
      "loss": 2.5617,
      "step": 110766
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.9282388091087341,
      "learning_rate": 0.00031812854674127236,
      "loss": 2.3275,
      "step": 110767
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.8590174317359924,
      "learning_rate": 0.000318124463652799,
      "loss": 2.0323,
      "step": 110768
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.8605696558952332,
      "learning_rate": 0.00031812038056095604,
      "loss": 2.0872,
      "step": 110769
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.8938999772071838,
      "learning_rate": 0.00031811629746574416,
      "loss": 2.0899,
      "step": 110770
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.825379490852356,
      "learning_rate": 0.00031811221436716416,
      "loss": 2.0215,
      "step": 110771
    },
    {
      "epoch": 1.44,
      "grad_norm": 1.0185682773590088,
      "learning_rate": 0.0003181081312652167,
      "loss": 2.073,
      "step": 110772
    },
    {
      "epoch": 1.44,
      "grad_norm": 1.0173301696777344,
      "learning_rate": 0.00031810404815990254,
      "loss": 1.9032,
      "step": 110773
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.900778591632843,
      "learning_rate": 0.00031809996505122255,
      "loss": 2.2424,
      "step": 110774
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.9429308176040649,
      "learning_rate": 0.0003180958819391775,
      "loss": 2.033,
      "step": 110775
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.8332729339599609,
      "learning_rate": 0.000318091798823768,
      "loss": 2.069,
      "step": 110776
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.9540249705314636,
      "learning_rate": 0.00031808771570499494,
      "loss": 2.033,
      "step": 110777
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.8524963855743408,
      "learning_rate": 0.000318083632582859,
      "loss": 1.9623,
      "step": 110778
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.9907901287078857,
      "learning_rate": 0.000318079549457361,
      "loss": 2.2032,
      "step": 110779
    },
    {
      "epoch": 1.44,
      "grad_norm": 1.1471428871154785,
      "learning_rate": 0.0003180754663285016,
      "loss": 2.1643,
      "step": 110780
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.9425076842308044,
      "learning_rate": 0.00031807138319628174,
      "loss": 1.7747,
      "step": 110781
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.9005486369132996,
      "learning_rate": 0.000318067300060702,
      "loss": 2.1848,
      "step": 110782
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.8747594952583313,
      "learning_rate": 0.0003180632169217632,
      "loss": 2.3835,
      "step": 110783
    },
    {
      "epoch": 1.44,
      "grad_norm": 1.090057373046875,
      "learning_rate": 0.00031805913377946617,
      "loss": 2.1454,
      "step": 110784
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.9936159253120422,
      "learning_rate": 0.00031805505063381157,
      "loss": 2.0481,
      "step": 110785
    },
    {
      "epoch": 1.44,
      "grad_norm": 1.0112453699111938,
      "learning_rate": 0.0003180509674848001,
      "loss": 2.1948,
      "step": 110786
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.9540812373161316,
      "learning_rate": 0.00031804688433243275,
      "loss": 2.2401,
      "step": 110787
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.9551066160202026,
      "learning_rate": 0.0003180428011767101,
      "loss": 2.135,
      "step": 110788
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.8874479532241821,
      "learning_rate": 0.0003180387180176329,
      "loss": 2.1617,
      "step": 110789
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.8435279130935669,
      "learning_rate": 0.00031803463485520207,
      "loss": 2.1338,
      "step": 110790
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.9122055172920227,
      "learning_rate": 0.0003180305516894182,
      "loss": 2.2185,
      "step": 110791
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.8981903195381165,
      "learning_rate": 0.00031802646852028215,
      "loss": 2.0069,
      "step": 110792
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.9047772288322449,
      "learning_rate": 0.0003180223853477946,
      "loss": 2.18,
      "step": 110793
    },
    {
      "epoch": 1.44,
      "grad_norm": 1.0089019536972046,
      "learning_rate": 0.00031801830217195634,
      "loss": 2.2163,
      "step": 110794
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.8861874341964722,
      "learning_rate": 0.00031801421899276813,
      "loss": 2.3346,
      "step": 110795
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.9400012493133545,
      "learning_rate": 0.00031801013581023076,
      "loss": 2.1009,
      "step": 110796
    },
    {
      "epoch": 1.44,
      "grad_norm": 1.0056402683258057,
      "learning_rate": 0.00031800605262434495,
      "loss": 2.1774,
      "step": 110797
    },
    {
      "epoch": 1.44,
      "grad_norm": 1.0717618465423584,
      "learning_rate": 0.0003180019694351115,
      "loss": 2.0412,
      "step": 110798
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.9907903075218201,
      "learning_rate": 0.0003179978862425311,
      "loss": 2.1197,
      "step": 110799
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.8719976544380188,
      "learning_rate": 0.0003179938030466046,
      "loss": 2.0015,
      "step": 110800
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.9823824167251587,
      "learning_rate": 0.00031798971984733275,
      "loss": 2.2618,
      "step": 110801
    },
    {
      "epoch": 1.44,
      "grad_norm": 1.1342214345932007,
      "learning_rate": 0.0003179856366447162,
      "loss": 2.0558,
      "step": 110802
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.9729114174842834,
      "learning_rate": 0.0003179815534387558,
      "loss": 2.0907,
      "step": 110803
    },
    {
      "epoch": 1.44,
      "grad_norm": 1.008998155593872,
      "learning_rate": 0.00031797747022945227,
      "loss": 2.1418,
      "step": 110804
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.9167581796646118,
      "learning_rate": 0.0003179733870168064,
      "loss": 2.2356,
      "step": 110805
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.8989579677581787,
      "learning_rate": 0.00031796930380081895,
      "loss": 2.1695,
      "step": 110806
    },
    {
      "epoch": 1.44,
      "grad_norm": 1.0894784927368164,
      "learning_rate": 0.0003179652205814907,
      "loss": 2.1378,
      "step": 110807
    },
    {
      "epoch": 1.44,
      "grad_norm": 1.0892090797424316,
      "learning_rate": 0.0003179611373588223,
      "loss": 2.0041,
      "step": 110808
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.9458787441253662,
      "learning_rate": 0.00031795705413281466,
      "loss": 2.6167,
      "step": 110809
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.8837706446647644,
      "learning_rate": 0.00031795297090346844,
      "loss": 2.2816,
      "step": 110810
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.8925099968910217,
      "learning_rate": 0.0003179488876707844,
      "loss": 2.1735,
      "step": 110811
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.8665803670883179,
      "learning_rate": 0.00031794480443476334,
      "loss": 2.0107,
      "step": 110812
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.9848657846450806,
      "learning_rate": 0.000317940721195406,
      "loss": 1.936,
      "step": 110813
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.9733270406723022,
      "learning_rate": 0.0003179366379527131,
      "loss": 2.0615,
      "step": 110814
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.8655613660812378,
      "learning_rate": 0.0003179325547066855,
      "loss": 2.2033,
      "step": 110815
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.9693217873573303,
      "learning_rate": 0.0003179284714573239,
      "loss": 2.1158,
      "step": 110816
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.9908783435821533,
      "learning_rate": 0.000317924388204629,
      "loss": 2.1429,
      "step": 110817
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.8852655291557312,
      "learning_rate": 0.0003179203049486017,
      "loss": 2.2338,
      "step": 110818
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.9507864713668823,
      "learning_rate": 0.0003179162216892426,
      "loss": 2.0045,
      "step": 110819
    },
    {
      "epoch": 1.44,
      "grad_norm": 1.1797124147415161,
      "learning_rate": 0.00031791213842655256,
      "loss": 2.2823,
      "step": 110820
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.9716650247573853,
      "learning_rate": 0.0003179080551605324,
      "loss": 2.2915,
      "step": 110821
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.8399777412414551,
      "learning_rate": 0.0003179039718911827,
      "loss": 2.2706,
      "step": 110822
    },
    {
      "epoch": 1.44,
      "grad_norm": 1.0239337682724,
      "learning_rate": 0.0003178998886185043,
      "loss": 2.1814,
      "step": 110823
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.9484213590621948,
      "learning_rate": 0.00031789580534249804,
      "loss": 2.3908,
      "step": 110824
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.8425754308700562,
      "learning_rate": 0.0003178917220631645,
      "loss": 2.1987,
      "step": 110825
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.9238442182540894,
      "learning_rate": 0.00031788763878050466,
      "loss": 2.1648,
      "step": 110826
    },
    {
      "epoch": 1.44,
      "grad_norm": 1.0082420110702515,
      "learning_rate": 0.00031788355549451923,
      "loss": 2.133,
      "step": 110827
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.8511948585510254,
      "learning_rate": 0.0003178794722052087,
      "loss": 2.2886,
      "step": 110828
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.9274953007698059,
      "learning_rate": 0.0003178753889125742,
      "loss": 2.1704,
      "step": 110829
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.9757729768753052,
      "learning_rate": 0.00031787130561661633,
      "loss": 2.2619,
      "step": 110830
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.9560565948486328,
      "learning_rate": 0.0003178672223173358,
      "loss": 2.1897,
      "step": 110831
    },
    {
      "epoch": 1.44,
      "grad_norm": 1.1612211465835571,
      "learning_rate": 0.00031786313901473336,
      "loss": 2.0347,
      "step": 110832
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.952462911605835,
      "learning_rate": 0.00031785905570880994,
      "loss": 2.138,
      "step": 110833
    },
    {
      "epoch": 1.44,
      "grad_norm": 1.2289592027664185,
      "learning_rate": 0.0003178549723995661,
      "loss": 2.1418,
      "step": 110834
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.913665235042572,
      "learning_rate": 0.0003178508890870027,
      "loss": 2.0543,
      "step": 110835
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.9052554368972778,
      "learning_rate": 0.00031784680577112055,
      "loss": 2.4428,
      "step": 110836
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.9909877181053162,
      "learning_rate": 0.00031784272245192027,
      "loss": 2.356,
      "step": 110837
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.8547327518463135,
      "learning_rate": 0.00031783863912940264,
      "loss": 2.0996,
      "step": 110838
    },
    {
      "epoch": 1.44,
      "grad_norm": 1.0307947397232056,
      "learning_rate": 0.0003178345558035686,
      "loss": 2.1762,
      "step": 110839
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.8959670066833496,
      "learning_rate": 0.0003178304724744187,
      "loss": 2.2136,
      "step": 110840
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.8936958909034729,
      "learning_rate": 0.0003178263891419538,
      "loss": 2.1864,
      "step": 110841
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.8817051649093628,
      "learning_rate": 0.00031782230580617465,
      "loss": 2.1581,
      "step": 110842
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.8802534937858582,
      "learning_rate": 0.00031781822246708196,
      "loss": 2.157,
      "step": 110843
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.8543192744255066,
      "learning_rate": 0.00031781413912467654,
      "loss": 2.1192,
      "step": 110844
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.859131932258606,
      "learning_rate": 0.0003178100557789592,
      "loss": 1.8832,
      "step": 110845
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.8568171262741089,
      "learning_rate": 0.0003178059724299306,
      "loss": 2.0692,
      "step": 110846
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.9329835176467896,
      "learning_rate": 0.00031780188907759143,
      "loss": 1.9445,
      "step": 110847
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.9775256514549255,
      "learning_rate": 0.0003177978057219427,
      "loss": 2.319,
      "step": 110848
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.9089276194572449,
      "learning_rate": 0.00031779372236298494,
      "loss": 2.1207,
      "step": 110849
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.9280154705047607,
      "learning_rate": 0.00031778963900071895,
      "loss": 1.9472,
      "step": 110850
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.9912939667701721,
      "learning_rate": 0.0003177855556351456,
      "loss": 2.0703,
      "step": 110851
    },
    {
      "epoch": 1.44,
      "grad_norm": 1.0376359224319458,
      "learning_rate": 0.0003177814722662656,
      "loss": 2.0446,
      "step": 110852
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.9183390140533447,
      "learning_rate": 0.0003177773888940796,
      "loss": 2.0419,
      "step": 110853
    },
    {
      "epoch": 1.44,
      "grad_norm": 1.0960220098495483,
      "learning_rate": 0.00031777330551858844,
      "loss": 2.1692,
      "step": 110854
    },
    {
      "epoch": 1.44,
      "grad_norm": 1.0270603895187378,
      "learning_rate": 0.000317769222139793,
      "loss": 2.3617,
      "step": 110855
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.8916935920715332,
      "learning_rate": 0.00031776513875769394,
      "loss": 2.1766,
      "step": 110856
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.9264423847198486,
      "learning_rate": 0.0003177610553722919,
      "loss": 2.1925,
      "step": 110857
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.9086096286773682,
      "learning_rate": 0.0003177569719835878,
      "loss": 1.9785,
      "step": 110858
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.9906435012817383,
      "learning_rate": 0.00031775288859158234,
      "loss": 2.2538,
      "step": 110859
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.8722233176231384,
      "learning_rate": 0.00031774880519627627,
      "loss": 2.1753,
      "step": 110860
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.8953691124916077,
      "learning_rate": 0.00031774472179767026,
      "loss": 2.1071,
      "step": 110861
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.9346702694892883,
      "learning_rate": 0.00031774063839576535,
      "loss": 2.2996,
      "step": 110862
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.8806429505348206,
      "learning_rate": 0.000317736554990562,
      "loss": 2.3307,
      "step": 110863
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.805810809135437,
      "learning_rate": 0.00031773247158206115,
      "loss": 2.0951,
      "step": 110864
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.8851586580276489,
      "learning_rate": 0.0003177283881702635,
      "loss": 2.1735,
      "step": 110865
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.9037695527076721,
      "learning_rate": 0.0003177243047551698,
      "loss": 2.1355,
      "step": 110866
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.9258126020431519,
      "learning_rate": 0.0003177202213367808,
      "loss": 2.2087,
      "step": 110867
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.9512700438499451,
      "learning_rate": 0.00031771613791509726,
      "loss": 2.3425,
      "step": 110868
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.9841731190681458,
      "learning_rate": 0.00031771205449012,
      "loss": 2.2989,
      "step": 110869
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.9785552024841309,
      "learning_rate": 0.0003177079710618497,
      "loss": 2.1841,
      "step": 110870
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.8950825929641724,
      "learning_rate": 0.00031770388763028716,
      "loss": 2.2278,
      "step": 110871
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.954296886920929,
      "learning_rate": 0.00031769980419543313,
      "loss": 2.3974,
      "step": 110872
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.8920453190803528,
      "learning_rate": 0.00031769572075728843,
      "loss": 2.3913,
      "step": 110873
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.9557483792304993,
      "learning_rate": 0.0003176916373158537,
      "loss": 2.3177,
      "step": 110874
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.8695780038833618,
      "learning_rate": 0.00031768755387112977,
      "loss": 2.0246,
      "step": 110875
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.9890012741088867,
      "learning_rate": 0.0003176834704231174,
      "loss": 2.0782,
      "step": 110876
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.7921227216720581,
      "learning_rate": 0.00031767938697181735,
      "loss": 2.2574,
      "step": 110877
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.9419052600860596,
      "learning_rate": 0.0003176753035172303,
      "loss": 1.9397,
      "step": 110878
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.9553704857826233,
      "learning_rate": 0.0003176712200593572,
      "loss": 2.0855,
      "step": 110879
    },
    {
      "epoch": 1.44,
      "grad_norm": 1.0273833274841309,
      "learning_rate": 0.00031766713659819857,
      "loss": 1.8867,
      "step": 110880
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.9893613457679749,
      "learning_rate": 0.0003176630531337553,
      "loss": 2.0107,
      "step": 110881
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.9262973666191101,
      "learning_rate": 0.0003176589696660282,
      "loss": 2.1932,
      "step": 110882
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.9487776756286621,
      "learning_rate": 0.000317654886195018,
      "loss": 2.1209,
      "step": 110883
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.8331988453865051,
      "learning_rate": 0.00031765080272072534,
      "loss": 1.8451,
      "step": 110884
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.9681019186973572,
      "learning_rate": 0.00031764671924315103,
      "loss": 2.1386,
      "step": 110885
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.8918711543083191,
      "learning_rate": 0.0003176426357622959,
      "loss": 2.2321,
      "step": 110886
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.8678891062736511,
      "learning_rate": 0.0003176385522781607,
      "loss": 2.0353,
      "step": 110887
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.8721458315849304,
      "learning_rate": 0.0003176344687907462,
      "loss": 2.2367,
      "step": 110888
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.897457480430603,
      "learning_rate": 0.00031763038530005305,
      "loss": 2.1421,
      "step": 110889
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.993093729019165,
      "learning_rate": 0.0003176263018060821,
      "loss": 2.2009,
      "step": 110890
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.9721562266349792,
      "learning_rate": 0.00031762221830883405,
      "loss": 2.0777,
      "step": 110891
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.9093309044837952,
      "learning_rate": 0.00031761813480830973,
      "loss": 2.2664,
      "step": 110892
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.8402305245399475,
      "learning_rate": 0.0003176140513045098,
      "loss": 2.1293,
      "step": 110893
    },
    {
      "epoch": 1.44,
      "grad_norm": 1.046026349067688,
      "learning_rate": 0.00031760996779743524,
      "loss": 2.3387,
      "step": 110894
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.884764552116394,
      "learning_rate": 0.0003176058842870865,
      "loss": 2.1031,
      "step": 110895
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.856381893157959,
      "learning_rate": 0.00031760180077346456,
      "loss": 2.1658,
      "step": 110896
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.9503915309906006,
      "learning_rate": 0.00031759771725657014,
      "loss": 2.334,
      "step": 110897
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.952929675579071,
      "learning_rate": 0.0003175936337364039,
      "loss": 2.2489,
      "step": 110898
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.9292823672294617,
      "learning_rate": 0.00031758955021296667,
      "loss": 2.175,
      "step": 110899
    },
    {
      "epoch": 1.44,
      "grad_norm": 1.0286858081817627,
      "learning_rate": 0.0003175854666862593,
      "loss": 2.0669,
      "step": 110900
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.8512955904006958,
      "learning_rate": 0.0003175813831562824,
      "loss": 2.0973,
      "step": 110901
    },
    {
      "epoch": 1.44,
      "grad_norm": 1.0458096265792847,
      "learning_rate": 0.0003175772996230368,
      "loss": 2.1634,
      "step": 110902
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.9474092721939087,
      "learning_rate": 0.0003175732160865233,
      "loss": 2.2469,
      "step": 110903
    },
    {
      "epoch": 1.44,
      "grad_norm": 1.0159022808074951,
      "learning_rate": 0.00031756913254674254,
      "loss": 2.1307,
      "step": 110904
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.8509379625320435,
      "learning_rate": 0.0003175650490036953,
      "loss": 2.2076,
      "step": 110905
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.884630024433136,
      "learning_rate": 0.00031756096545738254,
      "loss": 2.3922,
      "step": 110906
    },
    {
      "epoch": 1.44,
      "grad_norm": 1.0856069326400757,
      "learning_rate": 0.00031755688190780474,
      "loss": 2.0806,
      "step": 110907
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.9030611515045166,
      "learning_rate": 0.0003175527983549628,
      "loss": 2.2108,
      "step": 110908
    },
    {
      "epoch": 1.44,
      "grad_norm": 1.0698812007904053,
      "learning_rate": 0.00031754871479885754,
      "loss": 2.1871,
      "step": 110909
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.8908623456954956,
      "learning_rate": 0.00031754463123948956,
      "loss": 2.099,
      "step": 110910
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.8733850717544556,
      "learning_rate": 0.0003175405476768597,
      "loss": 2.0483,
      "step": 110911
    },
    {
      "epoch": 1.44,
      "grad_norm": 1.0018764734268188,
      "learning_rate": 0.0003175364641109688,
      "loss": 2.307,
      "step": 110912
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.9619460701942444,
      "learning_rate": 0.00031753238054181744,
      "loss": 2.2738,
      "step": 110913
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.9994871616363525,
      "learning_rate": 0.0003175282969694065,
      "loss": 2.2326,
      "step": 110914
    },
    {
      "epoch": 1.44,
      "grad_norm": 1.0380460023880005,
      "learning_rate": 0.00031752421339373684,
      "loss": 2.1347,
      "step": 110915
    },
    {
      "epoch": 1.44,
      "grad_norm": 1.0105326175689697,
      "learning_rate": 0.0003175201298148089,
      "loss": 1.9999,
      "step": 110916
    },
    {
      "epoch": 1.44,
      "grad_norm": 1.036208987236023,
      "learning_rate": 0.00031751604623262374,
      "loss": 2.2229,
      "step": 110917
    },
    {
      "epoch": 1.44,
      "grad_norm": 1.2885761260986328,
      "learning_rate": 0.0003175119626471821,
      "loss": 2.2455,
      "step": 110918
    },
    {
      "epoch": 1.44,
      "grad_norm": 1.0934104919433594,
      "learning_rate": 0.00031750787905848455,
      "loss": 2.0756,
      "step": 110919
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.932045042514801,
      "learning_rate": 0.00031750379546653196,
      "loss": 2.1002,
      "step": 110920
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.9389598369598389,
      "learning_rate": 0.00031749971187132513,
      "loss": 2.3521,
      "step": 110921
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.9101011157035828,
      "learning_rate": 0.0003174956282728647,
      "loss": 2.1165,
      "step": 110922
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.9287702441215515,
      "learning_rate": 0.0003174915446711516,
      "loss": 2.2251,
      "step": 110923
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.9755764007568359,
      "learning_rate": 0.00031748746106618646,
      "loss": 2.1931,
      "step": 110924
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.8642876148223877,
      "learning_rate": 0.00031748337745797,
      "loss": 2.1003,
      "step": 110925
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.9363320469856262,
      "learning_rate": 0.0003174792938465031,
      "loss": 1.9909,
      "step": 110926
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.9416579008102417,
      "learning_rate": 0.0003174752102317865,
      "loss": 1.9173,
      "step": 110927
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.896250307559967,
      "learning_rate": 0.0003174711266138209,
      "loss": 2.0846,
      "step": 110928
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.9951298832893372,
      "learning_rate": 0.000317467042992607,
      "loss": 2.0003,
      "step": 110929
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.9051963090896606,
      "learning_rate": 0.00031746295936814576,
      "loss": 2.384,
      "step": 110930
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.9292280077934265,
      "learning_rate": 0.00031745887574043786,
      "loss": 2.1423,
      "step": 110931
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.943260908126831,
      "learning_rate": 0.00031745479210948396,
      "loss": 2.1327,
      "step": 110932
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.9592673182487488,
      "learning_rate": 0.00031745070847528483,
      "loss": 2.2366,
      "step": 110933
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.8919508457183838,
      "learning_rate": 0.00031744662483784134,
      "loss": 2.1228,
      "step": 110934
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.9432028532028198,
      "learning_rate": 0.00031744254119715423,
      "loss": 2.1627,
      "step": 110935
    },
    {
      "epoch": 1.44,
      "grad_norm": 1.0177584886550903,
      "learning_rate": 0.00031743845755322417,
      "loss": 2.4947,
      "step": 110936
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.9161173701286316,
      "learning_rate": 0.000317434373906052,
      "loss": 1.9661,
      "step": 110937
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.8653320670127869,
      "learning_rate": 0.00031743029025563843,
      "loss": 2.0779,
      "step": 110938
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.9830118417739868,
      "learning_rate": 0.0003174262066019842,
      "loss": 2.2178,
      "step": 110939
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.9137586355209351,
      "learning_rate": 0.00031742212294509024,
      "loss": 2.1932,
      "step": 110940
    },
    {
      "epoch": 1.44,
      "grad_norm": 1.042384147644043,
      "learning_rate": 0.00031741803928495704,
      "loss": 2.1406,
      "step": 110941
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.9053210020065308,
      "learning_rate": 0.0003174139556215856,
      "loss": 2.0682,
      "step": 110942
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.8439294695854187,
      "learning_rate": 0.00031740987195497653,
      "loss": 1.9513,
      "step": 110943
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.904228687286377,
      "learning_rate": 0.0003174057882851307,
      "loss": 2.2853,
      "step": 110944
    },
    {
      "epoch": 1.44,
      "grad_norm": 1.162493109703064,
      "learning_rate": 0.00031740170461204865,
      "loss": 1.9824,
      "step": 110945
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.9776889085769653,
      "learning_rate": 0.0003173976209357314,
      "loss": 2.1136,
      "step": 110946
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.9812585115432739,
      "learning_rate": 0.0003173935372561796,
      "loss": 2.0756,
      "step": 110947
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.9755639433860779,
      "learning_rate": 0.000317389453573394,
      "loss": 2.3235,
      "step": 110948
    },
    {
      "epoch": 1.44,
      "grad_norm": 1.06892991065979,
      "learning_rate": 0.00031738536988737534,
      "loss": 2.0223,
      "step": 110949
    },
    {
      "epoch": 1.44,
      "grad_norm": 1.049867868423462,
      "learning_rate": 0.00031738128619812454,
      "loss": 2.2646,
      "step": 110950
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.8314236402511597,
      "learning_rate": 0.0003173772025056421,
      "loss": 2.055,
      "step": 110951
    },
    {
      "epoch": 1.44,
      "grad_norm": 1.002848744392395,
      "learning_rate": 0.0003173731188099289,
      "loss": 2.4146,
      "step": 110952
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.9450833201408386,
      "learning_rate": 0.00031736903511098575,
      "loss": 2.2243,
      "step": 110953
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.8465439081192017,
      "learning_rate": 0.00031736495140881343,
      "loss": 2.26,
      "step": 110954
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.963682234287262,
      "learning_rate": 0.0003173608677034126,
      "loss": 2.1675,
      "step": 110955
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.9290345907211304,
      "learning_rate": 0.000317356783994784,
      "loss": 2.0016,
      "step": 110956
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.8946936726570129,
      "learning_rate": 0.00031735270028292845,
      "loss": 2.1577,
      "step": 110957
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.8592877984046936,
      "learning_rate": 0.0003173486165678468,
      "loss": 2.2889,
      "step": 110958
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.8925988078117371,
      "learning_rate": 0.00031734453284953967,
      "loss": 2.2716,
      "step": 110959
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.9028341770172119,
      "learning_rate": 0.0003173404491280078,
      "loss": 2.0034,
      "step": 110960
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.9084974527359009,
      "learning_rate": 0.0003173363654032521,
      "loss": 2.0645,
      "step": 110961
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.8915216326713562,
      "learning_rate": 0.0003173322816752732,
      "loss": 2.1411,
      "step": 110962
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.8703013062477112,
      "learning_rate": 0.0003173281979440719,
      "loss": 2.2344,
      "step": 110963
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.9781314730644226,
      "learning_rate": 0.000317324114209649,
      "loss": 2.1185,
      "step": 110964
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.82065749168396,
      "learning_rate": 0.0003173200304720052,
      "loss": 2.0722,
      "step": 110965
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.8986250758171082,
      "learning_rate": 0.00031731594673114127,
      "loss": 2.1014,
      "step": 110966
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.981592059135437,
      "learning_rate": 0.0003173118629870579,
      "loss": 2.2347,
      "step": 110967
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.8923149704933167,
      "learning_rate": 0.0003173077792397561,
      "loss": 2.2299,
      "step": 110968
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.9731797575950623,
      "learning_rate": 0.0003173036954892363,
      "loss": 2.0359,
      "step": 110969
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.8991920351982117,
      "learning_rate": 0.0003172996117354995,
      "loss": 2.0841,
      "step": 110970
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.8727243542671204,
      "learning_rate": 0.0003172955279785464,
      "loss": 2.2126,
      "step": 110971
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.9342676401138306,
      "learning_rate": 0.0003172914442183776,
      "loss": 2.1129,
      "step": 110972
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.8820035457611084,
      "learning_rate": 0.00031728736045499417,
      "loss": 2.2448,
      "step": 110973
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.8786059617996216,
      "learning_rate": 0.0003172832766883966,
      "loss": 2.149,
      "step": 110974
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.8322492241859436,
      "learning_rate": 0.0003172791929185857,
      "loss": 2.148,
      "step": 110975
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.8378679156303406,
      "learning_rate": 0.00031727510914556243,
      "loss": 1.9995,
      "step": 110976
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.8613552451133728,
      "learning_rate": 0.00031727102536932724,
      "loss": 2.0869,
      "step": 110977
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.8581364750862122,
      "learning_rate": 0.00031726694158988106,
      "loss": 1.9879,
      "step": 110978
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.9210730195045471,
      "learning_rate": 0.0003172628578072247,
      "loss": 2.1205,
      "step": 110979
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.9409463405609131,
      "learning_rate": 0.0003172587740213588,
      "loss": 2.0205,
      "step": 110980
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.9478791952133179,
      "learning_rate": 0.00031725469023228414,
      "loss": 2.525,
      "step": 110981
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.8857436180114746,
      "learning_rate": 0.00031725060644000156,
      "loss": 2.0875,
      "step": 110982
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.9126039147377014,
      "learning_rate": 0.00031724652264451174,
      "loss": 2.2906,
      "step": 110983
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.9802111387252808,
      "learning_rate": 0.00031724243884581545,
      "loss": 2.1724,
      "step": 110984
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.9291430115699768,
      "learning_rate": 0.00031723835504391355,
      "loss": 1.8085,
      "step": 110985
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.9590582847595215,
      "learning_rate": 0.0003172342712388066,
      "loss": 2.2081,
      "step": 110986
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.9803023338317871,
      "learning_rate": 0.00031723018743049555,
      "loss": 2.1969,
      "step": 110987
    },
    {
      "epoch": 1.45,
      "grad_norm": 1.091732382774353,
      "learning_rate": 0.00031722610361898106,
      "loss": 2.1749,
      "step": 110988
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.8519934415817261,
      "learning_rate": 0.0003172220198042639,
      "loss": 2.1062,
      "step": 110989
    },
    {
      "epoch": 1.45,
      "grad_norm": 1.1469144821166992,
      "learning_rate": 0.0003172179359863448,
      "loss": 2.1932,
      "step": 110990
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.8620125651359558,
      "learning_rate": 0.00031721385216522467,
      "loss": 2.2078,
      "step": 110991
    },
    {
      "epoch": 1.45,
      "grad_norm": 1.0549464225769043,
      "learning_rate": 0.0003172097683409041,
      "loss": 2.1856,
      "step": 110992
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.9090685844421387,
      "learning_rate": 0.0003172056845133839,
      "loss": 1.8792,
      "step": 110993
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.866150438785553,
      "learning_rate": 0.0003172016006826649,
      "loss": 2.2956,
      "step": 110994
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.9571053981781006,
      "learning_rate": 0.00031719751684874774,
      "loss": 1.9171,
      "step": 110995
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.8533946871757507,
      "learning_rate": 0.0003171934330116332,
      "loss": 1.9483,
      "step": 110996
    },
    {
      "epoch": 1.45,
      "grad_norm": 1.0656757354736328,
      "learning_rate": 0.0003171893491713221,
      "loss": 2.2356,
      "step": 110997
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.8747793436050415,
      "learning_rate": 0.0003171852653278152,
      "loss": 2.1514,
      "step": 110998
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.9523835778236389,
      "learning_rate": 0.0003171811814811132,
      "loss": 2.3357,
      "step": 110999
    },
    {
      "epoch": 1.45,
      "grad_norm": 1.035282015800476,
      "learning_rate": 0.000317177097631217,
      "loss": 2.1708,
      "step": 111000
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.9698483347892761,
      "learning_rate": 0.00031717301377812715,
      "loss": 2.0268,
      "step": 111001
    },
    {
      "epoch": 1.45,
      "grad_norm": 1.01708984375,
      "learning_rate": 0.00031716892992184455,
      "loss": 2.2142,
      "step": 111002
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.951137363910675,
      "learning_rate": 0.0003171648460623699,
      "loss": 2.2906,
      "step": 111003
    },
    {
      "epoch": 1.45,
      "grad_norm": 1.2092041969299316,
      "learning_rate": 0.00031716076219970405,
      "loss": 2.0022,
      "step": 111004
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.8831116557121277,
      "learning_rate": 0.00031715667833384756,
      "loss": 2.0429,
      "step": 111005
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.974183976650238,
      "learning_rate": 0.0003171525944648014,
      "loss": 1.8863,
      "step": 111006
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.9886993765830994,
      "learning_rate": 0.0003171485105925663,
      "loss": 2.2934,
      "step": 111007
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.9438132643699646,
      "learning_rate": 0.00031714442671714285,
      "loss": 2.3899,
      "step": 111008
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.9511555433273315,
      "learning_rate": 0.00031714034283853197,
      "loss": 2.29,
      "step": 111009
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.9103472828865051,
      "learning_rate": 0.00031713625895673445,
      "loss": 2.2427,
      "step": 111010
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.9565223455429077,
      "learning_rate": 0.0003171321750717509,
      "loss": 1.9824,
      "step": 111011
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.9901361465454102,
      "learning_rate": 0.00031712809118358213,
      "loss": 2.2813,
      "step": 111012
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.9845616221427917,
      "learning_rate": 0.00031712400729222906,
      "loss": 2.4039,
      "step": 111013
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.9025501608848572,
      "learning_rate": 0.00031711992339769217,
      "loss": 2.0205,
      "step": 111014
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.984392523765564,
      "learning_rate": 0.0003171158394999724,
      "loss": 2.1618,
      "step": 111015
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.8551004528999329,
      "learning_rate": 0.00031711175559907053,
      "loss": 2.1943,
      "step": 111016
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.9110496044158936,
      "learning_rate": 0.00031710767169498724,
      "loss": 2.0782,
      "step": 111017
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.9382720589637756,
      "learning_rate": 0.0003171035877877233,
      "loss": 1.844,
      "step": 111018
    },
    {
      "epoch": 1.45,
      "grad_norm": 1.442624807357788,
      "learning_rate": 0.00031709950387727947,
      "loss": 1.9715,
      "step": 111019
    },
    {
      "epoch": 1.45,
      "grad_norm": 1.1823703050613403,
      "learning_rate": 0.0003170954199636565,
      "loss": 2.3303,
      "step": 111020
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.8587088584899902,
      "learning_rate": 0.0003170913360468552,
      "loss": 2.0728,
      "step": 111021
    },
    {
      "epoch": 1.45,
      "grad_norm": 1.0041515827178955,
      "learning_rate": 0.0003170872521268763,
      "loss": 2.0989,
      "step": 111022
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.9812868237495422,
      "learning_rate": 0.00031708316820372055,
      "loss": 2.1509,
      "step": 111023
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.9391647577285767,
      "learning_rate": 0.0003170790842773887,
      "loss": 2.0808,
      "step": 111024
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.981753945350647,
      "learning_rate": 0.0003170750003478816,
      "loss": 2.0347,
      "step": 111025
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.9514775276184082,
      "learning_rate": 0.00031707091641519983,
      "loss": 2.3412,
      "step": 111026
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.8771358728408813,
      "learning_rate": 0.00031706683247934427,
      "loss": 2.1292,
      "step": 111027
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.8375080227851868,
      "learning_rate": 0.0003170627485403157,
      "loss": 2.1657,
      "step": 111028
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.9890096187591553,
      "learning_rate": 0.0003170586645981148,
      "loss": 2.136,
      "step": 111029
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.8941991329193115,
      "learning_rate": 0.0003170545806527424,
      "loss": 2.0434,
      "step": 111030
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.9215199947357178,
      "learning_rate": 0.00031705049670419935,
      "loss": 2.2237,
      "step": 111031
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.9238269925117493,
      "learning_rate": 0.00031704641275248613,
      "loss": 2.0041,
      "step": 111032
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.8784700036048889,
      "learning_rate": 0.00031704232879760364,
      "loss": 2.2776,
      "step": 111033
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.9437803626060486,
      "learning_rate": 0.0003170382448395527,
      "loss": 1.9816,
      "step": 111034
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.9775627255439758,
      "learning_rate": 0.0003170341608783342,
      "loss": 2.2264,
      "step": 111035
    },
    {
      "epoch": 1.45,
      "grad_norm": 1.0282490253448486,
      "learning_rate": 0.0003170300769139485,
      "loss": 2.3097,
      "step": 111036
    },
    {
      "epoch": 1.45,
      "grad_norm": 1.0385154485702515,
      "learning_rate": 0.0003170259929463966,
      "loss": 2.1879,
      "step": 111037
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.9630094766616821,
      "learning_rate": 0.00031702190897567935,
      "loss": 2.464,
      "step": 111038
    },
    {
      "epoch": 1.45,
      "grad_norm": 1.0986167192459106,
      "learning_rate": 0.0003170178250017974,
      "loss": 2.2008,
      "step": 111039
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.9290039539337158,
      "learning_rate": 0.0003170137410247515,
      "loss": 2.1405,
      "step": 111040
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.895443320274353,
      "learning_rate": 0.0003170096570445424,
      "loss": 2.1829,
      "step": 111041
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.9853501915931702,
      "learning_rate": 0.0003170055730611709,
      "loss": 1.9541,
      "step": 111042
    },
    {
      "epoch": 1.45,
      "grad_norm": 1.0907474756240845,
      "learning_rate": 0.00031700148907463775,
      "loss": 1.9705,
      "step": 111043
    },
    {
      "epoch": 1.45,
      "grad_norm": 1.000391960144043,
      "learning_rate": 0.0003169974050849437,
      "loss": 2.1667,
      "step": 111044
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.956419825553894,
      "learning_rate": 0.00031699332109208945,
      "loss": 2.1032,
      "step": 111045
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.9842554926872253,
      "learning_rate": 0.00031698923709607595,
      "loss": 2.4755,
      "step": 111046
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.8754042387008667,
      "learning_rate": 0.0003169851530969037,
      "loss": 2.0416,
      "step": 111047
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.8868191242218018,
      "learning_rate": 0.00031698106909457363,
      "loss": 2.4225,
      "step": 111048
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.9666005373001099,
      "learning_rate": 0.0003169769850890865,
      "loss": 2.2017,
      "step": 111049
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.9074944257736206,
      "learning_rate": 0.00031697290108044306,
      "loss": 2.1527,
      "step": 111050
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.9834502935409546,
      "learning_rate": 0.0003169688170686439,
      "loss": 2.0126,
      "step": 111051
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.9402645826339722,
      "learning_rate": 0.00031696473305369005,
      "loss": 2.1667,
      "step": 111052
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.8679201602935791,
      "learning_rate": 0.00031696064903558203,
      "loss": 2.1135,
      "step": 111053
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.8781334161758423,
      "learning_rate": 0.0003169565650143207,
      "loss": 1.9825,
      "step": 111054
    },
    {
      "epoch": 1.45,
      "grad_norm": 1.0525834560394287,
      "learning_rate": 0.00031695248098990704,
      "loss": 2.2837,
      "step": 111055
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.8511050939559937,
      "learning_rate": 0.00031694839696234136,
      "loss": 2.0863,
      "step": 111056
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.8895836472511292,
      "learning_rate": 0.0003169443129316247,
      "loss": 1.9024,
      "step": 111057
    },
    {
      "epoch": 1.45,
      "grad_norm": 1.0175690650939941,
      "learning_rate": 0.00031694022889775783,
      "loss": 2.292,
      "step": 111058
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.844943106174469,
      "learning_rate": 0.00031693614486074143,
      "loss": 1.9856,
      "step": 111059
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.9632795453071594,
      "learning_rate": 0.00031693206082057624,
      "loss": 2.197,
      "step": 111060
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.8975458145141602,
      "learning_rate": 0.0003169279767772632,
      "loss": 2.0084,
      "step": 111061
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.939384400844574,
      "learning_rate": 0.00031692389273080274,
      "loss": 2.0685,
      "step": 111062
    },
    {
      "epoch": 1.45,
      "grad_norm": 1.2164928913116455,
      "learning_rate": 0.00031691980868119593,
      "loss": 2.2354,
      "step": 111063
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.8123458027839661,
      "learning_rate": 0.0003169157246284434,
      "loss": 2.1418,
      "step": 111064
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.9065576195716858,
      "learning_rate": 0.00031691164057254587,
      "loss": 2.3227,
      "step": 111065
    },
    {
      "epoch": 1.45,
      "grad_norm": 1.1556847095489502,
      "learning_rate": 0.0003169075565135041,
      "loss": 2.1501,
      "step": 111066
    },
    {
      "epoch": 1.45,
      "grad_norm": 1.105366587638855,
      "learning_rate": 0.00031690347245131895,
      "loss": 2.4094,
      "step": 111067
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.892477810382843,
      "learning_rate": 0.0003168993883859912,
      "loss": 2.1242,
      "step": 111068
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.8968049883842468,
      "learning_rate": 0.00031689530431752147,
      "loss": 1.9639,
      "step": 111069
    },
    {
      "epoch": 1.45,
      "grad_norm": 1.3269307613372803,
      "learning_rate": 0.0003168912202459105,
      "loss": 2.1755,
      "step": 111070
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.8739181756973267,
      "learning_rate": 0.0003168871361711593,
      "loss": 2.0818,
      "step": 111071
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.8251615762710571,
      "learning_rate": 0.00031688305209326835,
      "loss": 2.1319,
      "step": 111072
    },
    {
      "epoch": 1.45,
      "grad_norm": 1.0420918464660645,
      "learning_rate": 0.00031687896801223855,
      "loss": 2.0414,
      "step": 111073
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.9382759928703308,
      "learning_rate": 0.0003168748839280707,
      "loss": 2.151,
      "step": 111074
    },
    {
      "epoch": 1.45,
      "grad_norm": 1.040432333946228,
      "learning_rate": 0.0003168707998407653,
      "loss": 2.1561,
      "step": 111075
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.9982683658599854,
      "learning_rate": 0.0003168667157503235,
      "loss": 2.2871,
      "step": 111076
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.8900589942932129,
      "learning_rate": 0.0003168626316567458,
      "loss": 2.3431,
      "step": 111077
    },
    {
      "epoch": 1.45,
      "grad_norm": 1.197610855102539,
      "learning_rate": 0.00031685854756003295,
      "loss": 2.1213,
      "step": 111078
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.9595121145248413,
      "learning_rate": 0.00031685446346018585,
      "loss": 2.1394,
      "step": 111079
    },
    {
      "epoch": 1.45,
      "grad_norm": 1.1074624061584473,
      "learning_rate": 0.0003168503793572052,
      "loss": 2.3905,
      "step": 111080
    },
    {
      "epoch": 1.45,
      "grad_norm": 1.0032384395599365,
      "learning_rate": 0.00031684629525109166,
      "loss": 2.1137,
      "step": 111081
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.9178956747055054,
      "learning_rate": 0.00031684221114184615,
      "loss": 2.1605,
      "step": 111082
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.9214962720870972,
      "learning_rate": 0.00031683812702946936,
      "loss": 2.05,
      "step": 111083
    },
    {
      "epoch": 1.45,
      "grad_norm": 1.0141147375106812,
      "learning_rate": 0.000316834042913962,
      "loss": 2.0744,
      "step": 111084
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.9464598894119263,
      "learning_rate": 0.00031682995879532484,
      "loss": 2.1891,
      "step": 111085
    },
    {
      "epoch": 1.45,
      "grad_norm": 1.0121947526931763,
      "learning_rate": 0.0003168258746735588,
      "loss": 2.1808,
      "step": 111086
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.9187899827957153,
      "learning_rate": 0.0003168217905486644,
      "loss": 2.1086,
      "step": 111087
    },
    {
      "epoch": 1.45,
      "grad_norm": 1.0576633214950562,
      "learning_rate": 0.0003168177064206425,
      "loss": 2.0728,
      "step": 111088
    },
    {
      "epoch": 1.45,
      "grad_norm": 1.078190565109253,
      "learning_rate": 0.000316813622289494,
      "loss": 2.0617,
      "step": 111089
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.8671197891235352,
      "learning_rate": 0.0003168095381552194,
      "loss": 2.0268,
      "step": 111090
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.8859856724739075,
      "learning_rate": 0.0003168054540178196,
      "loss": 2.156,
      "step": 111091
    },
    {
      "epoch": 1.45,
      "grad_norm": 1.0835013389587402,
      "learning_rate": 0.00031680136987729553,
      "loss": 1.9448,
      "step": 111092
    },
    {
      "epoch": 1.45,
      "grad_norm": 1.049656867980957,
      "learning_rate": 0.0003167972857336476,
      "loss": 2.2089,
      "step": 111093
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.8435112237930298,
      "learning_rate": 0.00031679320158687675,
      "loss": 2.0014,
      "step": 111094
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.9498963356018066,
      "learning_rate": 0.00031678911743698374,
      "loss": 2.0303,
      "step": 111095
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.9930124282836914,
      "learning_rate": 0.00031678503328396926,
      "loss": 2.1105,
      "step": 111096
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.9348366856575012,
      "learning_rate": 0.0003167809491278342,
      "loss": 2.3335,
      "step": 111097
    },
    {
      "epoch": 1.45,
      "grad_norm": 1.0785902738571167,
      "learning_rate": 0.00031677686496857924,
      "loss": 2.2371,
      "step": 111098
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.9373317956924438,
      "learning_rate": 0.00031677278080620517,
      "loss": 2.2747,
      "step": 111099
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.9908007979393005,
      "learning_rate": 0.0003167686966407126,
      "loss": 2.0629,
      "step": 111100
    },
    {
      "epoch": 1.45,
      "grad_norm": 1.0528078079223633,
      "learning_rate": 0.0003167646124721026,
      "loss": 2.1938,
      "step": 111101
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.9816668033599854,
      "learning_rate": 0.0003167605283003756,
      "loss": 2.1166,
      "step": 111102
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.8532875776290894,
      "learning_rate": 0.00031675644412553253,
      "loss": 2.0963,
      "step": 111103
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.9932721853256226,
      "learning_rate": 0.00031675235994757414,
      "loss": 2.0077,
      "step": 111104
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.9106431603431702,
      "learning_rate": 0.00031674827576650116,
      "loss": 2.1763,
      "step": 111105
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.9805828332901001,
      "learning_rate": 0.00031674419158231435,
      "loss": 1.9962,
      "step": 111106
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.9087125062942505,
      "learning_rate": 0.0003167401073950145,
      "loss": 2.2068,
      "step": 111107
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.9306663870811462,
      "learning_rate": 0.0003167360232046023,
      "loss": 1.9846,
      "step": 111108
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.9559906125068665,
      "learning_rate": 0.00031673193901107864,
      "loss": 2.1088,
      "step": 111109
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.9481488466262817,
      "learning_rate": 0.0003167278548144442,
      "loss": 2.1183,
      "step": 111110
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.9713211059570312,
      "learning_rate": 0.0003167237706146997,
      "loss": 2.1046,
      "step": 111111
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.883229672908783,
      "learning_rate": 0.0003167196864118458,
      "loss": 2.0459,
      "step": 111112
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.8702371716499329,
      "learning_rate": 0.0003167156022058836,
      "loss": 2.0293,
      "step": 111113
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.9867849349975586,
      "learning_rate": 0.0003167115179968136,
      "loss": 1.9978,
      "step": 111114
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.9340941309928894,
      "learning_rate": 0.0003167074337846365,
      "loss": 2.2849,
      "step": 111115
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.8715445399284363,
      "learning_rate": 0.00031670334956935333,
      "loss": 2.1754,
      "step": 111116
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.8967848420143127,
      "learning_rate": 0.0003166992653509646,
      "loss": 2.3082,
      "step": 111117
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.9991047978401184,
      "learning_rate": 0.0003166951811294712,
      "loss": 2.2287,
      "step": 111118
    },
    {
      "epoch": 1.45,
      "grad_norm": 1.0119625329971313,
      "learning_rate": 0.0003166910969048738,
      "loss": 2.2204,
      "step": 111119
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.9518929123878479,
      "learning_rate": 0.0003166870126771733,
      "loss": 2.3157,
      "step": 111120
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.9053161144256592,
      "learning_rate": 0.0003166829284463703,
      "loss": 1.9748,
      "step": 111121
    },
    {
      "epoch": 1.45,
      "grad_norm": 1.2169671058654785,
      "learning_rate": 0.0003166788442124656,
      "loss": 2.2981,
      "step": 111122
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.9402598738670349,
      "learning_rate": 0.0003166747599754601,
      "loss": 2.2673,
      "step": 111123
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.8968612551689148,
      "learning_rate": 0.0003166706757353543,
      "loss": 2.0405,
      "step": 111124
    },
    {
      "epoch": 1.45,
      "grad_norm": 1.0914705991744995,
      "learning_rate": 0.00031666659149214916,
      "loss": 2.2624,
      "step": 111125
    },
    {
      "epoch": 1.45,
      "grad_norm": 1.0854345560073853,
      "learning_rate": 0.0003166625072458454,
      "loss": 2.3214,
      "step": 111126
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.8835230469703674,
      "learning_rate": 0.0003166584229964438,
      "loss": 1.9045,
      "step": 111127
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.8814675807952881,
      "learning_rate": 0.00031665433874394504,
      "loss": 2.1932,
      "step": 111128
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.9173821806907654,
      "learning_rate": 0.0003166502544883499,
      "loss": 2.2656,
      "step": 111129
    },
    {
      "epoch": 1.45,
      "grad_norm": 1.0261735916137695,
      "learning_rate": 0.0003166461702296592,
      "loss": 2.0147,
      "step": 111130
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.8910796642303467,
      "learning_rate": 0.0003166420859678737,
      "loss": 2.1758,
      "step": 111131
    },
    {
      "epoch": 1.45,
      "grad_norm": 1.0387791395187378,
      "learning_rate": 0.0003166380017029941,
      "loss": 2.1977,
      "step": 111132
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.8902468085289001,
      "learning_rate": 0.0003166339174350211,
      "loss": 2.1768,
      "step": 111133
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.9854252934455872,
      "learning_rate": 0.0003166298331639557,
      "loss": 2.0909,
      "step": 111134
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.8980318307876587,
      "learning_rate": 0.00031662574888979844,
      "loss": 2.3746,
      "step": 111135
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.9007726907730103,
      "learning_rate": 0.00031662166461255006,
      "loss": 2.2032,
      "step": 111136
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.866155743598938,
      "learning_rate": 0.0003166175803322114,
      "loss": 2.1149,
      "step": 111137
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.9629265666007996,
      "learning_rate": 0.00031661349604878336,
      "loss": 2.2978,
      "step": 111138
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.9473632574081421,
      "learning_rate": 0.0003166094117622664,
      "loss": 2.0608,
      "step": 111139
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.9713342189788818,
      "learning_rate": 0.0003166053274726615,
      "loss": 1.9224,
      "step": 111140
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.8652693629264832,
      "learning_rate": 0.0003166012431799694,
      "loss": 2.179,
      "step": 111141
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.9302540421485901,
      "learning_rate": 0.00031659715888419075,
      "loss": 2.3271,
      "step": 111142
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.8573339581489563,
      "learning_rate": 0.0003165930745853264,
      "loss": 2.1981,
      "step": 111143
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.9069972038269043,
      "learning_rate": 0.0003165889902833771,
      "loss": 1.8619,
      "step": 111144
    },
    {
      "epoch": 1.45,
      "grad_norm": 1.0941181182861328,
      "learning_rate": 0.0003165849059783436,
      "loss": 2.3648,
      "step": 111145
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.8159128427505493,
      "learning_rate": 0.0003165808216702266,
      "loss": 2.1056,
      "step": 111146
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.8993824124336243,
      "learning_rate": 0.00031657673735902697,
      "loss": 2.1727,
      "step": 111147
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.9486181735992432,
      "learning_rate": 0.0003165726530447454,
      "loss": 2.1976,
      "step": 111148
    },
    {
      "epoch": 1.45,
      "grad_norm": 1.0356485843658447,
      "learning_rate": 0.0003165685687273826,
      "loss": 2.2931,
      "step": 111149
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.9781934022903442,
      "learning_rate": 0.00031656448440693945,
      "loss": 2.0128,
      "step": 111150
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.9558332562446594,
      "learning_rate": 0.0003165604000834166,
      "loss": 2.2298,
      "step": 111151
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.9590951204299927,
      "learning_rate": 0.00031655631575681486,
      "loss": 2.1389,
      "step": 111152
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.9383426308631897,
      "learning_rate": 0.0003165522314271351,
      "loss": 2.2572,
      "step": 111153
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.93208247423172,
      "learning_rate": 0.0003165481470943779,
      "loss": 2.2429,
      "step": 111154
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.9316519498825073,
      "learning_rate": 0.00031654406275854406,
      "loss": 2.0843,
      "step": 111155
    },
    {
      "epoch": 1.45,
      "grad_norm": 1.0285168886184692,
      "learning_rate": 0.0003165399784196344,
      "loss": 2.0869,
      "step": 111156
    },
    {
      "epoch": 1.45,
      "grad_norm": 1.146243691444397,
      "learning_rate": 0.00031653589407764965,
      "loss": 1.995,
      "step": 111157
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.8866643905639648,
      "learning_rate": 0.0003165318097325905,
      "loss": 2.0106,
      "step": 111158
    },
    {
      "epoch": 1.45,
      "grad_norm": 1.0637249946594238,
      "learning_rate": 0.00031652772538445784,
      "loss": 1.9601,
      "step": 111159
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.9111526608467102,
      "learning_rate": 0.00031652364103325234,
      "loss": 2.0968,
      "step": 111160
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.8204336762428284,
      "learning_rate": 0.0003165195566789748,
      "loss": 1.948,
      "step": 111161
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.9161819815635681,
      "learning_rate": 0.00031651547232162594,
      "loss": 2.2365,
      "step": 111162
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.8912443518638611,
      "learning_rate": 0.00031651138796120657,
      "loss": 2.1357,
      "step": 111163
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.8702755570411682,
      "learning_rate": 0.0003165073035977174,
      "loss": 2.0341,
      "step": 111164
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.942111611366272,
      "learning_rate": 0.0003165032192311592,
      "loss": 1.7972,
      "step": 111165
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.879250168800354,
      "learning_rate": 0.00031649913486153273,
      "loss": 2.0279,
      "step": 111166
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.9239361882209778,
      "learning_rate": 0.0003164950504888388,
      "loss": 2.1237,
      "step": 111167
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.8981600403785706,
      "learning_rate": 0.00031649096611307816,
      "loss": 2.0537,
      "step": 111168
    },
    {
      "epoch": 1.45,
      "grad_norm": 1.063676357269287,
      "learning_rate": 0.00031648688173425147,
      "loss": 2.1127,
      "step": 111169
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.9110912680625916,
      "learning_rate": 0.0003164827973523595,
      "loss": 2.1466,
      "step": 111170
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.8952189087867737,
      "learning_rate": 0.00031647871296740324,
      "loss": 2.0485,
      "step": 111171
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.8831249475479126,
      "learning_rate": 0.0003164746285793831,
      "loss": 2.238,
      "step": 111172
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.9439833760261536,
      "learning_rate": 0.0003164705441883001,
      "loss": 2.322,
      "step": 111173
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.9020674228668213,
      "learning_rate": 0.00031646645979415495,
      "loss": 2.1002,
      "step": 111174
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.8992718458175659,
      "learning_rate": 0.0003164623753969484,
      "loss": 2.0284,
      "step": 111175
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.9409640431404114,
      "learning_rate": 0.00031645829099668105,
      "loss": 2.1302,
      "step": 111176
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.8713986277580261,
      "learning_rate": 0.0003164542065933539,
      "loss": 2.1359,
      "step": 111177
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.8660886883735657,
      "learning_rate": 0.0003164501221869676,
      "loss": 2.0459,
      "step": 111178
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.9342007637023926,
      "learning_rate": 0.00031644603777752277,
      "loss": 2.1976,
      "step": 111179
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.952579140663147,
      "learning_rate": 0.0003164419533650205,
      "loss": 2.3565,
      "step": 111180
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.9965764284133911,
      "learning_rate": 0.00031643786894946125,
      "loss": 2.0971,
      "step": 111181
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.9642772078514099,
      "learning_rate": 0.00031643378453084586,
      "loss": 2.3217,
      "step": 111182
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.958707332611084,
      "learning_rate": 0.00031642970010917524,
      "loss": 2.3577,
      "step": 111183
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.8536570072174072,
      "learning_rate": 0.00031642561568444994,
      "loss": 2.1504,
      "step": 111184
    },
    {
      "epoch": 1.45,
      "grad_norm": 1.1392065286636353,
      "learning_rate": 0.0003164215312566708,
      "loss": 1.9733,
      "step": 111185
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.9494383931159973,
      "learning_rate": 0.0003164174468258387,
      "loss": 1.9451,
      "step": 111186
    },
    {
      "epoch": 1.45,
      "grad_norm": 1.1238652467727661,
      "learning_rate": 0.00031641336239195415,
      "loss": 2.1712,
      "step": 111187
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.90272057056427,
      "learning_rate": 0.0003164092779550181,
      "loss": 2.0904,
      "step": 111188
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.9280105233192444,
      "learning_rate": 0.00031640519351503133,
      "loss": 2.0916,
      "step": 111189
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.8977152109146118,
      "learning_rate": 0.0003164011090719944,
      "loss": 2.3804,
      "step": 111190
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.9209919571876526,
      "learning_rate": 0.00031639702462590825,
      "loss": 2.1684,
      "step": 111191
    },
    {
      "epoch": 1.45,
      "grad_norm": 1.126485824584961,
      "learning_rate": 0.0003163929401767736,
      "loss": 2.3171,
      "step": 111192
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.9213669896125793,
      "learning_rate": 0.0003163888557245912,
      "loss": 2.1532,
      "step": 111193
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.9141202569007874,
      "learning_rate": 0.0003163847712693617,
      "loss": 2.2029,
      "step": 111194
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.9554714560508728,
      "learning_rate": 0.0003163806868110861,
      "loss": 2.1026,
      "step": 111195
    },
    {
      "epoch": 1.45,
      "grad_norm": 1.1390982866287231,
      "learning_rate": 0.00031637660234976495,
      "loss": 2.2624,
      "step": 111196
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.9695402979850769,
      "learning_rate": 0.0003163725178853991,
      "loss": 2.0487,
      "step": 111197
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.8922370076179504,
      "learning_rate": 0.00031636843341798925,
      "loss": 2.0138,
      "step": 111198
    },
    {
      "epoch": 1.45,
      "grad_norm": 1.2822465896606445,
      "learning_rate": 0.00031636434894753625,
      "loss": 2.2028,
      "step": 111199
    },
    {
      "epoch": 1.45,
      "grad_norm": 1.5483959913253784,
      "learning_rate": 0.00031636026447404077,
      "loss": 1.9272,
      "step": 111200
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.846301257610321,
      "learning_rate": 0.00031635617999750356,
      "loss": 2.0346,
      "step": 111201
    },
    {
      "epoch": 1.45,
      "grad_norm": 1.2119297981262207,
      "learning_rate": 0.00031635209551792554,
      "loss": 1.9057,
      "step": 111202
    },
    {
      "epoch": 1.45,
      "grad_norm": 1.2107059955596924,
      "learning_rate": 0.00031634801103530737,
      "loss": 2.4466,
      "step": 111203
    },
    {
      "epoch": 1.45,
      "grad_norm": 1.029395580291748,
      "learning_rate": 0.0003163439265496497,
      "loss": 2.2289,
      "step": 111204
    },
    {
      "epoch": 1.45,
      "grad_norm": 1.0827369689941406,
      "learning_rate": 0.00031633984206095347,
      "loss": 2.2523,
      "step": 111205
    },
    {
      "epoch": 1.45,
      "grad_norm": 1.0145601034164429,
      "learning_rate": 0.00031633575756921927,
      "loss": 2.1573,
      "step": 111206
    },
    {
      "epoch": 1.45,
      "grad_norm": 1.0056028366088867,
      "learning_rate": 0.00031633167307444795,
      "loss": 2.0858,
      "step": 111207
    },
    {
      "epoch": 1.45,
      "grad_norm": 1.0037428140640259,
      "learning_rate": 0.0003163275885766403,
      "loss": 2.2989,
      "step": 111208
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.9481076002120972,
      "learning_rate": 0.000316323504075797,
      "loss": 1.9624,
      "step": 111209
    },
    {
      "epoch": 1.45,
      "grad_norm": 1.008320927619934,
      "learning_rate": 0.0003163194195719189,
      "loss": 1.8761,
      "step": 111210
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.8384035229682922,
      "learning_rate": 0.0003163153350650067,
      "loss": 2.248,
      "step": 111211
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.8312219977378845,
      "learning_rate": 0.00031631125055506124,
      "loss": 2.0708,
      "step": 111212
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.8806796073913574,
      "learning_rate": 0.0003163071660420831,
      "loss": 2.0141,
      "step": 111213
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.9386330246925354,
      "learning_rate": 0.00031630308152607315,
      "loss": 2.3419,
      "step": 111214
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.8767482042312622,
      "learning_rate": 0.0003162989970070323,
      "loss": 2.1907,
      "step": 111215
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.9984108209609985,
      "learning_rate": 0.00031629491248496097,
      "loss": 2.1626,
      "step": 111216
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.895504355430603,
      "learning_rate": 0.0003162908279598602,
      "loss": 2.0807,
      "step": 111217
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.8857603073120117,
      "learning_rate": 0.00031628674343173066,
      "loss": 2.1846,
      "step": 111218
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.8521682024002075,
      "learning_rate": 0.0003162826589005731,
      "loss": 2.15,
      "step": 111219
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.9009522795677185,
      "learning_rate": 0.0003162785743663883,
      "loss": 2.2238,
      "step": 111220
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.8926287889480591,
      "learning_rate": 0.00031627448982917696,
      "loss": 2.3521,
      "step": 111221
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.9445241689682007,
      "learning_rate": 0.00031627040528893995,
      "loss": 2.3345,
      "step": 111222
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.8539566397666931,
      "learning_rate": 0.0003162663207456779,
      "loss": 2.1249,
      "step": 111223
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.8251006007194519,
      "learning_rate": 0.00031626223619939163,
      "loss": 2.2541,
      "step": 111224
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.9375784397125244,
      "learning_rate": 0.0003162581516500819,
      "loss": 2.3387,
      "step": 111225
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.8560569882392883,
      "learning_rate": 0.0003162540670977496,
      "loss": 1.8373,
      "step": 111226
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.9023231863975525,
      "learning_rate": 0.0003162499825423952,
      "loss": 2.0083,
      "step": 111227
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.8385772109031677,
      "learning_rate": 0.0003162458979840197,
      "loss": 2.0595,
      "step": 111228
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.8551115393638611,
      "learning_rate": 0.0003162418134226238,
      "loss": 2.0323,
      "step": 111229
    },
    {
      "epoch": 1.45,
      "grad_norm": 1.0710710287094116,
      "learning_rate": 0.0003162377288582082,
      "loss": 2.0976,
      "step": 111230
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.8086647987365723,
      "learning_rate": 0.0003162336442907737,
      "loss": 2.0502,
      "step": 111231
    },
    {
      "epoch": 1.45,
      "grad_norm": 1.0252443552017212,
      "learning_rate": 0.00031622955972032105,
      "loss": 1.9914,
      "step": 111232
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.9184457659721375,
      "learning_rate": 0.00031622547514685107,
      "loss": 2.1838,
      "step": 111233
    },
    {
      "epoch": 1.45,
      "grad_norm": 1.0468201637268066,
      "learning_rate": 0.0003162213905703644,
      "loss": 2.1444,
      "step": 111234
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.919187605381012,
      "learning_rate": 0.000316217305990862,
      "loss": 1.8857,
      "step": 111235
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.957755446434021,
      "learning_rate": 0.0003162132214083443,
      "loss": 2.2637,
      "step": 111236
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.9242505431175232,
      "learning_rate": 0.0003162091368228124,
      "loss": 2.0899,
      "step": 111237
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.9174478054046631,
      "learning_rate": 0.00031620505223426684,
      "loss": 1.9927,
      "step": 111238
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.9781162738800049,
      "learning_rate": 0.00031620096764270846,
      "loss": 2.1103,
      "step": 111239
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.9451305866241455,
      "learning_rate": 0.0003161968830481381,
      "loss": 2.0722,
      "step": 111240
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.9744089841842651,
      "learning_rate": 0.0003161927984505564,
      "loss": 2.2266,
      "step": 111241
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.8174107670783997,
      "learning_rate": 0.00031618871384996413,
      "loss": 1.9361,
      "step": 111242
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.9599011540412903,
      "learning_rate": 0.00031618462924636205,
      "loss": 1.9628,
      "step": 111243
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.8777651190757751,
      "learning_rate": 0.00031618054463975094,
      "loss": 2.1297,
      "step": 111244
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.8994905948638916,
      "learning_rate": 0.0003161764600301316,
      "loss": 2.0316,
      "step": 111245
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.8828978538513184,
      "learning_rate": 0.0003161723754175047,
      "loss": 2.1314,
      "step": 111246
    },
    {
      "epoch": 1.45,
      "grad_norm": 1.0043636560440063,
      "learning_rate": 0.00031616829080187116,
      "loss": 2.1438,
      "step": 111247
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.9573346972465515,
      "learning_rate": 0.00031616420618323153,
      "loss": 2.0829,
      "step": 111248
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.9301740527153015,
      "learning_rate": 0.00031616012156158665,
      "loss": 1.9332,
      "step": 111249
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.9237706661224365,
      "learning_rate": 0.0003161560369369374,
      "loss": 1.901,
      "step": 111250
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.8428284525871277,
      "learning_rate": 0.00031615195230928436,
      "loss": 2.1416,
      "step": 111251
    },
    {
      "epoch": 1.45,
      "grad_norm": 1.0477194786071777,
      "learning_rate": 0.0003161478676786284,
      "loss": 2.1775,
      "step": 111252
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.9822746515274048,
      "learning_rate": 0.0003161437830449702,
      "loss": 2.1802,
      "step": 111253
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.9999057054519653,
      "learning_rate": 0.0003161396984083106,
      "loss": 2.0279,
      "step": 111254
    },
    {
      "epoch": 1.45,
      "grad_norm": 1.0060895681381226,
      "learning_rate": 0.0003161356137686503,
      "loss": 2.3077,
      "step": 111255
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.8677015900611877,
      "learning_rate": 0.00031613152912599013,
      "loss": 2.2028,
      "step": 111256
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.9248037934303284,
      "learning_rate": 0.0003161274444803308,
      "loss": 2.1562,
      "step": 111257
    },
    {
      "epoch": 1.45,
      "grad_norm": 1.1270902156829834,
      "learning_rate": 0.00031612335983167305,
      "loss": 2.1196,
      "step": 111258
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.8707237839698792,
      "learning_rate": 0.0003161192751800177,
      "loss": 2.0681,
      "step": 111259
    },
    {
      "epoch": 1.45,
      "grad_norm": 1.0019330978393555,
      "learning_rate": 0.00031611519052536545,
      "loss": 2.2246,
      "step": 111260
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.9199368357658386,
      "learning_rate": 0.00031611110586771704,
      "loss": 2.1642,
      "step": 111261
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.8487342000007629,
      "learning_rate": 0.00031610702120707336,
      "loss": 2.1251,
      "step": 111262
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.8512749075889587,
      "learning_rate": 0.000316102936543435,
      "loss": 2.1931,
      "step": 111263
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.9802168607711792,
      "learning_rate": 0.00031609885187680283,
      "loss": 2.096,
      "step": 111264
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.9690728187561035,
      "learning_rate": 0.00031609476720717756,
      "loss": 2.1526,
      "step": 111265
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.8919156193733215,
      "learning_rate": 0.00031609068253456,
      "loss": 1.9896,
      "step": 111266
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.8006271719932556,
      "learning_rate": 0.00031608659785895086,
      "loss": 2.181,
      "step": 111267
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.93039470911026,
      "learning_rate": 0.0003160825131803509,
      "loss": 2.0851,
      "step": 111268
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.9884331822395325,
      "learning_rate": 0.0003160784284987609,
      "loss": 2.0756,
      "step": 111269
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.9602314829826355,
      "learning_rate": 0.00031607434381418167,
      "loss": 1.955,
      "step": 111270
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.95656418800354,
      "learning_rate": 0.00031607025912661383,
      "loss": 2.233,
      "step": 111271
    },
    {
      "epoch": 1.45,
      "grad_norm": 1.2018324136734009,
      "learning_rate": 0.0003160661744360583,
      "loss": 2.0201,
      "step": 111272
    },
    {
      "epoch": 1.45,
      "grad_norm": 1.1654927730560303,
      "learning_rate": 0.00031606208974251576,
      "loss": 1.9339,
      "step": 111273
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.9046441316604614,
      "learning_rate": 0.00031605800504598693,
      "loss": 2.061,
      "step": 111274
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.8452993035316467,
      "learning_rate": 0.00031605392034647265,
      "loss": 1.9895,
      "step": 111275
    },
    {
      "epoch": 1.45,
      "grad_norm": 1.0470061302185059,
      "learning_rate": 0.0003160498356439736,
      "loss": 2.5151,
      "step": 111276
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.9589792490005493,
      "learning_rate": 0.00031604575093849063,
      "loss": 2.0618,
      "step": 111277
    },
    {
      "epoch": 1.45,
      "grad_norm": 1.1185849905014038,
      "learning_rate": 0.00031604166623002447,
      "loss": 2.1615,
      "step": 111278
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.9277415871620178,
      "learning_rate": 0.00031603758151857583,
      "loss": 2.229,
      "step": 111279
    },
    {
      "epoch": 1.45,
      "grad_norm": 1.0477244853973389,
      "learning_rate": 0.00031603349680414547,
      "loss": 2.2857,
      "step": 111280
    },
    {
      "epoch": 1.45,
      "grad_norm": 1.1442036628723145,
      "learning_rate": 0.00031602941208673426,
      "loss": 2.2725,
      "step": 111281
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.9497742056846619,
      "learning_rate": 0.0003160253273663428,
      "loss": 2.0813,
      "step": 111282
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.8503533601760864,
      "learning_rate": 0.00031602124264297195,
      "loss": 2.3711,
      "step": 111283
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.9776294231414795,
      "learning_rate": 0.0003160171579166225,
      "loss": 2.1298,
      "step": 111284
    },
    {
      "epoch": 1.45,
      "grad_norm": 1.1298153400421143,
      "learning_rate": 0.0003160130731872951,
      "loss": 1.8937,
      "step": 111285
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.8931503891944885,
      "learning_rate": 0.0003160089884549906,
      "loss": 2.3121,
      "step": 111286
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.9113215804100037,
      "learning_rate": 0.0003160049037197097,
      "loss": 2.3461,
      "step": 111287
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.9707584381103516,
      "learning_rate": 0.0003160008189814532,
      "loss": 1.9708,
      "step": 111288
    },
    {
      "epoch": 1.45,
      "grad_norm": 1.1468205451965332,
      "learning_rate": 0.0003159967342402218,
      "loss": 2.1778,
      "step": 111289
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.867376983165741,
      "learning_rate": 0.0003159926494960164,
      "loss": 1.9722,
      "step": 111290
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.9111201763153076,
      "learning_rate": 0.0003159885647488376,
      "loss": 2.1393,
      "step": 111291
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.958124041557312,
      "learning_rate": 0.00031598447999868625,
      "loss": 1.8811,
      "step": 111292
    },
    {
      "epoch": 1.45,
      "grad_norm": 1.0898311138153076,
      "learning_rate": 0.0003159803952455631,
      "loss": 2.3576,
      "step": 111293
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.9673749804496765,
      "learning_rate": 0.0003159763104894688,
      "loss": 2.0613,
      "step": 111294
    },
    {
      "epoch": 1.45,
      "grad_norm": 1.1086605787277222,
      "learning_rate": 0.00031597222573040423,
      "loss": 2.1841,
      "step": 111295
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.9681200981140137,
      "learning_rate": 0.00031596814096837024,
      "loss": 2.2853,
      "step": 111296
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.9190379977226257,
      "learning_rate": 0.00031596405620336733,
      "loss": 2.1503,
      "step": 111297
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.9708513021469116,
      "learning_rate": 0.00031595997143539647,
      "loss": 2.4384,
      "step": 111298
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.929620623588562,
      "learning_rate": 0.0003159558866644583,
      "loss": 2.1492,
      "step": 111299
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.98508220911026,
      "learning_rate": 0.0003159518018905537,
      "loss": 2.0899,
      "step": 111300
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.9264208674430847,
      "learning_rate": 0.0003159477171136833,
      "loss": 2.0403,
      "step": 111301
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.941921055316925,
      "learning_rate": 0.00031594363233384794,
      "loss": 2.1729,
      "step": 111302
    },
    {
      "epoch": 1.45,
      "grad_norm": 1.041419267654419,
      "learning_rate": 0.00031593954755104845,
      "loss": 2.2906,
      "step": 111303
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.9637575745582581,
      "learning_rate": 0.00031593546276528534,
      "loss": 1.8529,
      "step": 111304
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.912272036075592,
      "learning_rate": 0.0003159313779765596,
      "loss": 2.1236,
      "step": 111305
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.9982343316078186,
      "learning_rate": 0.0003159272931848719,
      "loss": 2.1403,
      "step": 111306
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.9597416520118713,
      "learning_rate": 0.00031592320839022306,
      "loss": 2.0801,
      "step": 111307
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.8902034759521484,
      "learning_rate": 0.00031591912359261375,
      "loss": 2.113,
      "step": 111308
    },
    {
      "epoch": 1.45,
      "grad_norm": 1.0011906623840332,
      "learning_rate": 0.00031591503879204477,
      "loss": 2.217,
      "step": 111309
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.9774160385131836,
      "learning_rate": 0.00031591095398851685,
      "loss": 2.1915,
      "step": 111310
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.9157276749610901,
      "learning_rate": 0.0003159068691820309,
      "loss": 2.0638,
      "step": 111311
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.9063490629196167,
      "learning_rate": 0.00031590278437258746,
      "loss": 2.1522,
      "step": 111312
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.8796424269676208,
      "learning_rate": 0.00031589869956018745,
      "loss": 2.2605,
      "step": 111313
    },
    {
      "epoch": 1.45,
      "grad_norm": 1.121232271194458,
      "learning_rate": 0.00031589461474483153,
      "loss": 2.2884,
      "step": 111314
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.8888704776763916,
      "learning_rate": 0.00031589052992652057,
      "loss": 2.1442,
      "step": 111315
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.9510928392410278,
      "learning_rate": 0.0003158864451052552,
      "loss": 2.2133,
      "step": 111316
    },
    {
      "epoch": 1.45,
      "grad_norm": 1.0861191749572754,
      "learning_rate": 0.00031588236028103626,
      "loss": 2.2032,
      "step": 111317
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.930065393447876,
      "learning_rate": 0.0003158782754538645,
      "loss": 2.3517,
      "step": 111318
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.9941267967224121,
      "learning_rate": 0.0003158741906237406,
      "loss": 2.24,
      "step": 111319
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.8637501001358032,
      "learning_rate": 0.00031587010579066545,
      "loss": 2.0964,
      "step": 111320
    },
    {
      "epoch": 1.45,
      "grad_norm": 1.1610430479049683,
      "learning_rate": 0.00031586602095463976,
      "loss": 2.0342,
      "step": 111321
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.9297792911529541,
      "learning_rate": 0.0003158619361156642,
      "loss": 2.0524,
      "step": 111322
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.9341621994972229,
      "learning_rate": 0.0003158578512737397,
      "loss": 2.2434,
      "step": 111323
    },
    {
      "epoch": 1.45,
      "grad_norm": 1.0637190341949463,
      "learning_rate": 0.0003158537664288669,
      "loss": 2.26,
      "step": 111324
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.9582525491714478,
      "learning_rate": 0.00031584968158104653,
      "loss": 2.2759,
      "step": 111325
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.9778408408164978,
      "learning_rate": 0.00031584559673027954,
      "loss": 2.1897,
      "step": 111326
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.8419556617736816,
      "learning_rate": 0.0003158415118765664,
      "loss": 2.101,
      "step": 111327
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.8871350884437561,
      "learning_rate": 0.00031583742701990806,
      "loss": 2.1402,
      "step": 111328
    },
    {
      "epoch": 1.45,
      "grad_norm": 1.000580906867981,
      "learning_rate": 0.00031583334216030536,
      "loss": 2.332,
      "step": 111329
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.9237248301506042,
      "learning_rate": 0.0003158292572977588,
      "loss": 1.9756,
      "step": 111330
    },
    {
      "epoch": 1.45,
      "grad_norm": 1.0369740724563599,
      "learning_rate": 0.0003158251724322693,
      "loss": 2.1449,
      "step": 111331
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.8887403607368469,
      "learning_rate": 0.0003158210875638376,
      "loss": 2.1823,
      "step": 111332
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.9523707628250122,
      "learning_rate": 0.00031581700269246455,
      "loss": 2.1718,
      "step": 111333
    },
    {
      "epoch": 1.45,
      "grad_norm": 1.0274049043655396,
      "learning_rate": 0.00031581291781815075,
      "loss": 2.112,
      "step": 111334
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.9083557724952698,
      "learning_rate": 0.000315808832940897,
      "loss": 2.1586,
      "step": 111335
    },
    {
      "epoch": 1.45,
      "grad_norm": 1.1010301113128662,
      "learning_rate": 0.00031580474806070424,
      "loss": 2.1381,
      "step": 111336
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.9141865968704224,
      "learning_rate": 0.00031580066317757295,
      "loss": 2.36,
      "step": 111337
    },
    {
      "epoch": 1.45,
      "grad_norm": 1.137463927268982,
      "learning_rate": 0.000315796578291504,
      "loss": 2.0748,
      "step": 111338
    },
    {
      "epoch": 1.45,
      "grad_norm": 1.1586955785751343,
      "learning_rate": 0.00031579249340249824,
      "loss": 1.9404,
      "step": 111339
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.9033957719802856,
      "learning_rate": 0.00031578840851055633,
      "loss": 2.0817,
      "step": 111340
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.8963087797164917,
      "learning_rate": 0.00031578432361567907,
      "loss": 2.1749,
      "step": 111341
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.8844832181930542,
      "learning_rate": 0.0003157802387178672,
      "loss": 2.2245,
      "step": 111342
    },
    {
      "epoch": 1.45,
      "grad_norm": 1.3306978940963745,
      "learning_rate": 0.00031577615381712145,
      "loss": 2.0402,
      "step": 111343
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.8670375347137451,
      "learning_rate": 0.0003157720689134427,
      "loss": 2.1762,
      "step": 111344
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.9735192060470581,
      "learning_rate": 0.0003157679840068316,
      "loss": 2.1994,
      "step": 111345
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.9783715605735779,
      "learning_rate": 0.0003157638990972888,
      "loss": 2.1015,
      "step": 111346
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.9700031876564026,
      "learning_rate": 0.0003157598141848153,
      "loss": 2.2901,
      "step": 111347
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.8218898773193359,
      "learning_rate": 0.0003157557292694118,
      "loss": 2.0653,
      "step": 111348
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.8280076384544373,
      "learning_rate": 0.0003157516443510789,
      "loss": 2.1671,
      "step": 111349
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.8903840780258179,
      "learning_rate": 0.00031574755942981753,
      "loss": 2.4688,
      "step": 111350
    },
    {
      "epoch": 1.45,
      "grad_norm": 1.028605341911316,
      "learning_rate": 0.0003157434745056284,
      "loss": 2.0718,
      "step": 111351
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.9023390412330627,
      "learning_rate": 0.0003157393895785123,
      "loss": 1.7442,
      "step": 111352
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.947679340839386,
      "learning_rate": 0.00031573530464846985,
      "loss": 2.0899,
      "step": 111353
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.950299859046936,
      "learning_rate": 0.00031573121971550206,
      "loss": 2.0381,
      "step": 111354
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.8953344225883484,
      "learning_rate": 0.0003157271347796094,
      "loss": 2.1537,
      "step": 111355
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.9626321196556091,
      "learning_rate": 0.0003157230498407928,
      "loss": 2.1238,
      "step": 111356
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.8973997831344604,
      "learning_rate": 0.00031571896489905306,
      "loss": 2.2146,
      "step": 111357
    },
    {
      "epoch": 1.45,
      "grad_norm": 1.1402238607406616,
      "learning_rate": 0.0003157148799543908,
      "loss": 2.1979,
      "step": 111358
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.8974353075027466,
      "learning_rate": 0.00031571079500680683,
      "loss": 1.9791,
      "step": 111359
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.9389840960502625,
      "learning_rate": 0.00031570671005630197,
      "loss": 1.9345,
      "step": 111360
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.9435986876487732,
      "learning_rate": 0.0003157026251028769,
      "loss": 2.1576,
      "step": 111361
    },
    {
      "epoch": 1.45,
      "grad_norm": 1.202831745147705,
      "learning_rate": 0.0003156985401465325,
      "loss": 1.9173,
      "step": 111362
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.8796541690826416,
      "learning_rate": 0.0003156944551872693,
      "loss": 2.082,
      "step": 111363
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.9501625299453735,
      "learning_rate": 0.00031569037022508836,
      "loss": 2.0031,
      "step": 111364
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.9284424781799316,
      "learning_rate": 0.0003156862852599902,
      "loss": 2.4531,
      "step": 111365
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.942193329334259,
      "learning_rate": 0.0003156822002919757,
      "loss": 2.0312,
      "step": 111366
    },
    {
      "epoch": 1.45,
      "grad_norm": 1.0183472633361816,
      "learning_rate": 0.00031567811532104553,
      "loss": 2.2363,
      "step": 111367
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.9013624787330627,
      "learning_rate": 0.0003156740303472005,
      "loss": 2.0838,
      "step": 111368
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.9898428916931152,
      "learning_rate": 0.0003156699453704414,
      "loss": 2.1542,
      "step": 111369
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.9520903825759888,
      "learning_rate": 0.00031566586039076896,
      "loss": 2.1059,
      "step": 111370
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.941084086894989,
      "learning_rate": 0.0003156617754081839,
      "loss": 2.1134,
      "step": 111371
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.9684179425239563,
      "learning_rate": 0.0003156576904226871,
      "loss": 2.1494,
      "step": 111372
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.9110983610153198,
      "learning_rate": 0.00031565360543427926,
      "loss": 2.1063,
      "step": 111373
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.8103781342506409,
      "learning_rate": 0.000315649520442961,
      "loss": 2.1566,
      "step": 111374
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.9257374405860901,
      "learning_rate": 0.0003156454354487333,
      "loss": 2.2077,
      "step": 111375
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.8670517802238464,
      "learning_rate": 0.0003156413504515968,
      "loss": 1.9048,
      "step": 111376
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.948671281337738,
      "learning_rate": 0.00031563726545155225,
      "loss": 2.0769,
      "step": 111377
    },
    {
      "epoch": 1.45,
      "grad_norm": 1.1787712574005127,
      "learning_rate": 0.00031563318044860046,
      "loss": 2.1815,
      "step": 111378
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.8096175193786621,
      "learning_rate": 0.00031562909544274214,
      "loss": 2.0997,
      "step": 111379
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.8319898843765259,
      "learning_rate": 0.00031562501043397805,
      "loss": 2.1053,
      "step": 111380
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.8638943433761597,
      "learning_rate": 0.00031562092542230906,
      "loss": 2.3647,
      "step": 111381
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.9287612438201904,
      "learning_rate": 0.00031561684040773576,
      "loss": 2.0919,
      "step": 111382
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.922260582447052,
      "learning_rate": 0.000315612755390259,
      "loss": 2.0641,
      "step": 111383
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.9028552174568176,
      "learning_rate": 0.00031560867036987966,
      "loss": 1.9744,
      "step": 111384
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.8769697546958923,
      "learning_rate": 0.0003156045853465982,
      "loss": 2.3123,
      "step": 111385
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.897632360458374,
      "learning_rate": 0.0003156005003204156,
      "loss": 2.2908,
      "step": 111386
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.940211296081543,
      "learning_rate": 0.0003155964152913326,
      "loss": 2.4038,
      "step": 111387
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.8792669177055359,
      "learning_rate": 0.00031559233025935,
      "loss": 1.9957,
      "step": 111388
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.8943977952003479,
      "learning_rate": 0.00031558824522446845,
      "loss": 2.0411,
      "step": 111389
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.947323203086853,
      "learning_rate": 0.00031558416018668867,
      "loss": 2.0801,
      "step": 111390
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.9263688325881958,
      "learning_rate": 0.0003155800751460115,
      "loss": 2.2123,
      "step": 111391
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.9420682191848755,
      "learning_rate": 0.0003155759901024379,
      "loss": 2.275,
      "step": 111392
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.865276038646698,
      "learning_rate": 0.0003155719050559682,
      "loss": 2.1066,
      "step": 111393
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.9628171324729919,
      "learning_rate": 0.00031556782000660347,
      "loss": 2.0287,
      "step": 111394
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.8641740083694458,
      "learning_rate": 0.00031556373495434444,
      "loss": 2.2584,
      "step": 111395
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.9079535603523254,
      "learning_rate": 0.0003155596498991918,
      "loss": 1.8464,
      "step": 111396
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.8763231039047241,
      "learning_rate": 0.0003155555648411462,
      "loss": 2.0073,
      "step": 111397
    },
    {
      "epoch": 1.45,
      "grad_norm": 1.3038326501846313,
      "learning_rate": 0.00031555147978020863,
      "loss": 2.1386,
      "step": 111398
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.9012831449508667,
      "learning_rate": 0.0003155473947163798,
      "loss": 2.0678,
      "step": 111399
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.9219605326652527,
      "learning_rate": 0.00031554330964966035,
      "loss": 1.9749,
      "step": 111400
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.8770858645439148,
      "learning_rate": 0.00031553922458005105,
      "loss": 2.3427,
      "step": 111401
    },
    {
      "epoch": 1.45,
      "grad_norm": 1.0030996799468994,
      "learning_rate": 0.0003155351395075528,
      "loss": 1.8818,
      "step": 111402
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.9345568418502808,
      "learning_rate": 0.0003155310544321663,
      "loss": 2.0631,
      "step": 111403
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.9076440334320068,
      "learning_rate": 0.00031552696935389224,
      "loss": 2.2196,
      "step": 111404
    },
    {
      "epoch": 1.45,
      "grad_norm": 1.071352481842041,
      "learning_rate": 0.00031552288427273137,
      "loss": 2.1012,
      "step": 111405
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.9874890446662903,
      "learning_rate": 0.0003155187991886846,
      "loss": 2.3427,
      "step": 111406
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.9003674983978271,
      "learning_rate": 0.00031551471410175253,
      "loss": 2.0031,
      "step": 111407
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.9455901384353638,
      "learning_rate": 0.000315510629011936,
      "loss": 1.9933,
      "step": 111408
    },
    {
      "epoch": 1.45,
      "grad_norm": 1.0546804666519165,
      "learning_rate": 0.00031550654391923583,
      "loss": 2.3461,
      "step": 111409
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.8926981091499329,
      "learning_rate": 0.0003155024588236525,
      "loss": 2.0787,
      "step": 111410
    },
    {
      "epoch": 1.45,
      "grad_norm": 1.0129295587539673,
      "learning_rate": 0.00031549837372518713,
      "loss": 2.2941,
      "step": 111411
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.8736268281936646,
      "learning_rate": 0.0003154942886238403,
      "loss": 2.2146,
      "step": 111412
    },
    {
      "epoch": 1.45,
      "grad_norm": 1.031374454498291,
      "learning_rate": 0.00031549020351961276,
      "loss": 2.1546,
      "step": 111413
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.9242792725563049,
      "learning_rate": 0.0003154861184125053,
      "loss": 2.0031,
      "step": 111414
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.9637078642845154,
      "learning_rate": 0.00031548203330251874,
      "loss": 2.1389,
      "step": 111415
    },
    {
      "epoch": 1.45,
      "grad_norm": 1.0398234128952026,
      "learning_rate": 0.0003154779481896537,
      "loss": 2.0335,
      "step": 111416
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.9628514051437378,
      "learning_rate": 0.00031547386307391104,
      "loss": 2.1787,
      "step": 111417
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.9638664126396179,
      "learning_rate": 0.0003154697779552915,
      "loss": 2.1547,
      "step": 111418
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.9071090817451477,
      "learning_rate": 0.00031546569283379583,
      "loss": 2.0888,
      "step": 111419
    },
    {
      "epoch": 1.45,
      "grad_norm": 1.2794536352157593,
      "learning_rate": 0.0003154616077094248,
      "loss": 2.25,
      "step": 111420
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.8678548336029053,
      "learning_rate": 0.0003154575225821792,
      "loss": 1.9712,
      "step": 111421
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.8953484296798706,
      "learning_rate": 0.00031545343745205975,
      "loss": 2.2885,
      "step": 111422
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.999735414981842,
      "learning_rate": 0.00031544935231906713,
      "loss": 2.0575,
      "step": 111423
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.999064028263092,
      "learning_rate": 0.0003154452671832023,
      "loss": 2.2131,
      "step": 111424
    },
    {
      "epoch": 1.45,
      "grad_norm": 1.0553306341171265,
      "learning_rate": 0.0003154411820444658,
      "loss": 2.2178,
      "step": 111425
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.9155409336090088,
      "learning_rate": 0.0003154370969028585,
      "loss": 2.1434,
      "step": 111426
    },
    {
      "epoch": 1.45,
      "grad_norm": 1.063202142715454,
      "learning_rate": 0.0003154330117583812,
      "loss": 2.2059,
      "step": 111427
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.8617140054702759,
      "learning_rate": 0.0003154289266110346,
      "loss": 2.1189,
      "step": 111428
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.8797702789306641,
      "learning_rate": 0.00031542484146081945,
      "loss": 2.1737,
      "step": 111429
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.9809051752090454,
      "learning_rate": 0.00031542075630773667,
      "loss": 2.3728,
      "step": 111430
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.8765156865119934,
      "learning_rate": 0.0003154166711517867,
      "loss": 1.9543,
      "step": 111431
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.9069404602050781,
      "learning_rate": 0.0003154125859929705,
      "loss": 2.1803,
      "step": 111432
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.9096230268478394,
      "learning_rate": 0.0003154085008312889,
      "loss": 2.2618,
      "step": 111433
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.9959293007850647,
      "learning_rate": 0.00031540441566674254,
      "loss": 2.121,
      "step": 111434
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.8883020877838135,
      "learning_rate": 0.00031540033049933213,
      "loss": 1.9355,
      "step": 111435
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.9326203465461731,
      "learning_rate": 0.0003153962453290587,
      "loss": 2.2699,
      "step": 111436
    },
    {
      "epoch": 1.45,
      "grad_norm": 1.0970427989959717,
      "learning_rate": 0.0003153921601559226,
      "loss": 2.3551,
      "step": 111437
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.9243825078010559,
      "learning_rate": 0.0003153880749799249,
      "loss": 2.1978,
      "step": 111438
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.8805016279220581,
      "learning_rate": 0.0003153839898010663,
      "loss": 1.9852,
      "step": 111439
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.9634482860565186,
      "learning_rate": 0.00031537990461934744,
      "loss": 2.0861,
      "step": 111440
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.9072376489639282,
      "learning_rate": 0.0003153758194347691,
      "loss": 2.0964,
      "step": 111441
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.8930628299713135,
      "learning_rate": 0.00031537173424733237,
      "loss": 2.0344,
      "step": 111442
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.9329237937927246,
      "learning_rate": 0.0003153676490570375,
      "loss": 1.9814,
      "step": 111443
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.8905317187309265,
      "learning_rate": 0.00031536356386388563,
      "loss": 1.8495,
      "step": 111444
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.9637062549591064,
      "learning_rate": 0.00031535947866787734,
      "loss": 2.2117,
      "step": 111445
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.9047354459762573,
      "learning_rate": 0.0003153553934690134,
      "loss": 2.0932,
      "step": 111446
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.9247407913208008,
      "learning_rate": 0.0003153513082672946,
      "loss": 2.245,
      "step": 111447
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.9524429440498352,
      "learning_rate": 0.00031534722306272183,
      "loss": 2.1551,
      "step": 111448
    },
    {
      "epoch": 1.45,
      "grad_norm": 1.1092407703399658,
      "learning_rate": 0.00031534313785529553,
      "loss": 2.0942,
      "step": 111449
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.9269474744796753,
      "learning_rate": 0.00031533905264501675,
      "loss": 2.2713,
      "step": 111450
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.9980252385139465,
      "learning_rate": 0.0003153349674318862,
      "loss": 1.9893,
      "step": 111451
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.9737797975540161,
      "learning_rate": 0.00031533088221590456,
      "loss": 2.0024,
      "step": 111452
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.8716412782669067,
      "learning_rate": 0.0003153267969970725,
      "loss": 1.8951,
      "step": 111453
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.8849130868911743,
      "learning_rate": 0.00031532271177539106,
      "loss": 2.0865,
      "step": 111454
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.936058759689331,
      "learning_rate": 0.0003153186265508607,
      "loss": 2.0752,
      "step": 111455
    },
    {
      "epoch": 1.45,
      "grad_norm": 1.2053278684616089,
      "learning_rate": 0.00031531454132348243,
      "loss": 2.2652,
      "step": 111456
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.9416956305503845,
      "learning_rate": 0.00031531045609325693,
      "loss": 2.0955,
      "step": 111457
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.9850741028785706,
      "learning_rate": 0.0003153063708601848,
      "loss": 2.0689,
      "step": 111458
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.9089646339416504,
      "learning_rate": 0.000315302285624267,
      "loss": 1.9778,
      "step": 111459
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.8954940438270569,
      "learning_rate": 0.0003152982003855042,
      "loss": 2.0637,
      "step": 111460
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.9590900540351868,
      "learning_rate": 0.0003152941151438972,
      "loss": 1.896,
      "step": 111461
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.9305747747421265,
      "learning_rate": 0.00031529002989944666,
      "loss": 2.0909,
      "step": 111462
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.9560929536819458,
      "learning_rate": 0.0003152859446521535,
      "loss": 2.1175,
      "step": 111463
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.914588451385498,
      "learning_rate": 0.0003152818594020184,
      "loss": 2.1698,
      "step": 111464
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.8283369541168213,
      "learning_rate": 0.000315277774149042,
      "loss": 2.0761,
      "step": 111465
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.9530137181282043,
      "learning_rate": 0.00031527368889322526,
      "loss": 2.0409,
      "step": 111466
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.9707341194152832,
      "learning_rate": 0.0003152696036345689,
      "loss": 2.0996,
      "step": 111467
    },
    {
      "epoch": 1.45,
      "grad_norm": 1.0208115577697754,
      "learning_rate": 0.00031526551837307354,
      "loss": 1.8391,
      "step": 111468
    },
    {
      "epoch": 1.45,
      "grad_norm": 1.00144624710083,
      "learning_rate": 0.00031526143310874,
      "loss": 2.1392,
      "step": 111469
    },
    {
      "epoch": 1.45,
      "grad_norm": 1.0254777669906616,
      "learning_rate": 0.00031525734784156924,
      "loss": 2.194,
      "step": 111470
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.9747665524482727,
      "learning_rate": 0.00031525326257156165,
      "loss": 2.0314,
      "step": 111471
    },
    {
      "epoch": 1.45,
      "grad_norm": 1.0116417407989502,
      "learning_rate": 0.0003152491772987183,
      "loss": 2.067,
      "step": 111472
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.9635099172592163,
      "learning_rate": 0.00031524509202303987,
      "loss": 2.1254,
      "step": 111473
    },
    {
      "epoch": 1.45,
      "grad_norm": 1.0865999460220337,
      "learning_rate": 0.00031524100674452704,
      "loss": 2.1716,
      "step": 111474
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.9511296153068542,
      "learning_rate": 0.00031523692146318054,
      "loss": 2.0992,
      "step": 111475
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.8693385124206543,
      "learning_rate": 0.0003152328361790014,
      "loss": 1.9513,
      "step": 111476
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.8469322919845581,
      "learning_rate": 0.00031522875089199005,
      "loss": 2.2829,
      "step": 111477
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.907254159450531,
      "learning_rate": 0.00031522466560214736,
      "loss": 2.3066,
      "step": 111478
    },
    {
      "epoch": 1.45,
      "grad_norm": 1.0132266283035278,
      "learning_rate": 0.00031522058030947415,
      "loss": 2.0066,
      "step": 111479
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.9753130078315735,
      "learning_rate": 0.00031521649501397127,
      "loss": 2.1786,
      "step": 111480
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.9089073538780212,
      "learning_rate": 0.0003152124097156393,
      "loss": 2.0875,
      "step": 111481
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.9316603541374207,
      "learning_rate": 0.000315208324414479,
      "loss": 2.2353,
      "step": 111482
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.895033061504364,
      "learning_rate": 0.00031520423911049115,
      "loss": 2.0763,
      "step": 111483
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.988709032535553,
      "learning_rate": 0.0003152001538036766,
      "loss": 1.993,
      "step": 111484
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.9084439873695374,
      "learning_rate": 0.00031519606849403603,
      "loss": 2.4414,
      "step": 111485
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.8702911734580994,
      "learning_rate": 0.00031519198318157027,
      "loss": 2.1908,
      "step": 111486
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.9790160059928894,
      "learning_rate": 0.00031518789786628,
      "loss": 2.1247,
      "step": 111487
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.9562970399856567,
      "learning_rate": 0.000315183812548166,
      "loss": 2.302,
      "step": 111488
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.9697246551513672,
      "learning_rate": 0.0003151797272272291,
      "loss": 2.1346,
      "step": 111489
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.9358689188957214,
      "learning_rate": 0.0003151756419034699,
      "loss": 2.1865,
      "step": 111490
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.9012494683265686,
      "learning_rate": 0.00031517155657688934,
      "loss": 2.1642,
      "step": 111491
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.8936452269554138,
      "learning_rate": 0.0003151674712474881,
      "loss": 2.1573,
      "step": 111492
    },
    {
      "epoch": 1.45,
      "grad_norm": 1.0084288120269775,
      "learning_rate": 0.00031516338591526695,
      "loss": 2.1257,
      "step": 111493
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.8907811641693115,
      "learning_rate": 0.0003151593005802267,
      "loss": 2.2866,
      "step": 111494
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.9174312949180603,
      "learning_rate": 0.0003151552152423679,
      "loss": 2.1372,
      "step": 111495
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.9245018362998962,
      "learning_rate": 0.0003151511299016915,
      "loss": 2.2619,
      "step": 111496
    },
    {
      "epoch": 1.45,
      "grad_norm": 1.016377329826355,
      "learning_rate": 0.0003151470445581983,
      "loss": 2.1957,
      "step": 111497
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.9783769845962524,
      "learning_rate": 0.0003151429592118889,
      "loss": 2.2534,
      "step": 111498
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.9378978610038757,
      "learning_rate": 0.0003151388738627642,
      "loss": 1.939,
      "step": 111499
    },
    {
      "epoch": 1.45,
      "grad_norm": 1.1867685317993164,
      "learning_rate": 0.00031513478851082484,
      "loss": 2.2224,
      "step": 111500
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.8941094875335693,
      "learning_rate": 0.00031513070315607167,
      "loss": 2.3612,
      "step": 111501
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.931145191192627,
      "learning_rate": 0.0003151266177985054,
      "loss": 2.1719,
      "step": 111502
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.8913682699203491,
      "learning_rate": 0.00031512253243812683,
      "loss": 2.2042,
      "step": 111503
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.9893643856048584,
      "learning_rate": 0.0003151184470749366,
      "loss": 2.0467,
      "step": 111504
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.9128162860870361,
      "learning_rate": 0.00031511436170893566,
      "loss": 2.18,
      "step": 111505
    },
    {
      "epoch": 1.45,
      "grad_norm": 1.0945008993148804,
      "learning_rate": 0.00031511027634012473,
      "loss": 2.2805,
      "step": 111506
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.8682788014411926,
      "learning_rate": 0.0003151061909685044,
      "loss": 2.2385,
      "step": 111507
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.9722241163253784,
      "learning_rate": 0.00031510210559407563,
      "loss": 2.3276,
      "step": 111508
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.8665621876716614,
      "learning_rate": 0.00031509802021683904,
      "loss": 2.0542,
      "step": 111509
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.9451603889465332,
      "learning_rate": 0.00031509393483679544,
      "loss": 2.0133,
      "step": 111510
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.930646538734436,
      "learning_rate": 0.0003150898494539456,
      "loss": 2.2189,
      "step": 111511
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.892189621925354,
      "learning_rate": 0.0003150857640682903,
      "loss": 1.9637,
      "step": 111512
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.9797379374504089,
      "learning_rate": 0.00031508167867983026,
      "loss": 2.1725,
      "step": 111513
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.9842365384101868,
      "learning_rate": 0.00031507759328856626,
      "loss": 2.2132,
      "step": 111514
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.9354358911514282,
      "learning_rate": 0.00031507350789449905,
      "loss": 1.9523,
      "step": 111515
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.9888936281204224,
      "learning_rate": 0.00031506942249762935,
      "loss": 2.3027,
      "step": 111516
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.9232102036476135,
      "learning_rate": 0.00031506533709795795,
      "loss": 2.3073,
      "step": 111517
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.8637987375259399,
      "learning_rate": 0.0003150612516954857,
      "loss": 2.2811,
      "step": 111518
    },
    {
      "epoch": 1.45,
      "grad_norm": 1.1003565788269043,
      "learning_rate": 0.00031505716629021325,
      "loss": 2.0933,
      "step": 111519
    },
    {
      "epoch": 1.45,
      "grad_norm": 1.1762100458145142,
      "learning_rate": 0.0003150530808821413,
      "loss": 2.1909,
      "step": 111520
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.9977282285690308,
      "learning_rate": 0.00031504899547127085,
      "loss": 2.2159,
      "step": 111521
    },
    {
      "epoch": 1.45,
      "grad_norm": 1.1943227052688599,
      "learning_rate": 0.00031504491005760236,
      "loss": 2.115,
      "step": 111522
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.8931459188461304,
      "learning_rate": 0.00031504082464113684,
      "loss": 1.9328,
      "step": 111523
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.9928922057151794,
      "learning_rate": 0.00031503673922187494,
      "loss": 2.3456,
      "step": 111524
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.9544432163238525,
      "learning_rate": 0.0003150326537998174,
      "loss": 2.2359,
      "step": 111525
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.9273034334182739,
      "learning_rate": 0.00031502856837496495,
      "loss": 2.2,
      "step": 111526
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.9146670699119568,
      "learning_rate": 0.0003150244829473185,
      "loss": 2.0447,
      "step": 111527
    },
    {
      "epoch": 1.45,
      "grad_norm": 1.2445930242538452,
      "learning_rate": 0.0003150203975168787,
      "loss": 2.1969,
      "step": 111528
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.8919122219085693,
      "learning_rate": 0.0003150163120836462,
      "loss": 2.0568,
      "step": 111529
    },
    {
      "epoch": 1.45,
      "grad_norm": 1.0935125350952148,
      "learning_rate": 0.00031501222664762204,
      "loss": 2.4929,
      "step": 111530
    },
    {
      "epoch": 1.45,
      "grad_norm": 1.2606216669082642,
      "learning_rate": 0.0003150081412088068,
      "loss": 2.2388,
      "step": 111531
    },
    {
      "epoch": 1.45,
      "grad_norm": 1.0510765314102173,
      "learning_rate": 0.00031500405576720115,
      "loss": 1.9774,
      "step": 111532
    },
    {
      "epoch": 1.45,
      "grad_norm": 1.0116184949874878,
      "learning_rate": 0.00031499997032280604,
      "loss": 2.3323,
      "step": 111533
    },
    {
      "epoch": 1.45,
      "grad_norm": 1.0725131034851074,
      "learning_rate": 0.0003149958848756222,
      "loss": 1.9566,
      "step": 111534
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.962828516960144,
      "learning_rate": 0.0003149917994256503,
      "loss": 2.1402,
      "step": 111535
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.9776078462600708,
      "learning_rate": 0.00031498771397289103,
      "loss": 2.2266,
      "step": 111536
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.8882017731666565,
      "learning_rate": 0.00031498362851734544,
      "loss": 2.1091,
      "step": 111537
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.9773964881896973,
      "learning_rate": 0.000314979543059014,
      "loss": 2.0979,
      "step": 111538
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.9868167042732239,
      "learning_rate": 0.0003149754575978976,
      "loss": 2.0399,
      "step": 111539
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.9228076934814453,
      "learning_rate": 0.00031497137213399706,
      "loss": 2.2117,
      "step": 111540
    },
    {
      "epoch": 1.45,
      "grad_norm": 1.0292059183120728,
      "learning_rate": 0.00031496728666731297,
      "loss": 2.2766,
      "step": 111541
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.8979949355125427,
      "learning_rate": 0.00031496320119784617,
      "loss": 2.3851,
      "step": 111542
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.8595226407051086,
      "learning_rate": 0.0003149591157255976,
      "loss": 2.5864,
      "step": 111543
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.8816246390342712,
      "learning_rate": 0.00031495503025056763,
      "loss": 2.2343,
      "step": 111544
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.9018777012825012,
      "learning_rate": 0.0003149509447727573,
      "loss": 2.0514,
      "step": 111545
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.8643490076065063,
      "learning_rate": 0.0003149468592921674,
      "loss": 1.9918,
      "step": 111546
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.8603947758674622,
      "learning_rate": 0.0003149427738087985,
      "loss": 2.303,
      "step": 111547
    },
    {
      "epoch": 1.45,
      "grad_norm": 1.1270638704299927,
      "learning_rate": 0.00031493868832265144,
      "loss": 2.0568,
      "step": 111548
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.8652954697608948,
      "learning_rate": 0.0003149346028337271,
      "loss": 2.0369,
      "step": 111549
    },
    {
      "epoch": 1.45,
      "grad_norm": 1.0121757984161377,
      "learning_rate": 0.00031493051734202604,
      "loss": 2.2677,
      "step": 111550
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.987856388092041,
      "learning_rate": 0.00031492643184754913,
      "loss": 2.2504,
      "step": 111551
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.9295552372932434,
      "learning_rate": 0.00031492234635029716,
      "loss": 2.1348,
      "step": 111552
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.9719484448432922,
      "learning_rate": 0.0003149182608502708,
      "loss": 2.2523,
      "step": 111553
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.923640251159668,
      "learning_rate": 0.0003149141753474708,
      "loss": 2.1506,
      "step": 111554
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.895405650138855,
      "learning_rate": 0.00031491008984189814,
      "loss": 2.2332,
      "step": 111555
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.8651980757713318,
      "learning_rate": 0.00031490600433355326,
      "loss": 2.062,
      "step": 111556
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.8795452117919922,
      "learning_rate": 0.00031490191882243716,
      "loss": 2.0283,
      "step": 111557
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.8396948575973511,
      "learning_rate": 0.00031489783330855054,
      "loss": 2.1784,
      "step": 111558
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.8806230425834656,
      "learning_rate": 0.000314893747791894,
      "loss": 2.252,
      "step": 111559
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.8170274496078491,
      "learning_rate": 0.00031488966227246853,
      "loss": 2.0211,
      "step": 111560
    },
    {
      "epoch": 1.45,
      "grad_norm": 1.1198655366897583,
      "learning_rate": 0.00031488557675027475,
      "loss": 2.3172,
      "step": 111561
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.9382221102714539,
      "learning_rate": 0.0003148814912253135,
      "loss": 2.0513,
      "step": 111562
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.8599129915237427,
      "learning_rate": 0.0003148774056975854,
      "loss": 2.3431,
      "step": 111563
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.9617060422897339,
      "learning_rate": 0.00031487332016709137,
      "loss": 2.1048,
      "step": 111564
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.9718099236488342,
      "learning_rate": 0.00031486923463383223,
      "loss": 2.2451,
      "step": 111565
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.8408060073852539,
      "learning_rate": 0.0003148651490978085,
      "loss": 2.1046,
      "step": 111566
    },
    {
      "epoch": 1.45,
      "grad_norm": 1.0294913053512573,
      "learning_rate": 0.000314861063559021,
      "loss": 2.2163,
      "step": 111567
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.8670485615730286,
      "learning_rate": 0.00031485697801747063,
      "loss": 2.0533,
      "step": 111568
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.9347875714302063,
      "learning_rate": 0.00031485289247315805,
      "loss": 2.1165,
      "step": 111569
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.8894003629684448,
      "learning_rate": 0.00031484880692608406,
      "loss": 1.9462,
      "step": 111570
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.9821845293045044,
      "learning_rate": 0.00031484472137624937,
      "loss": 2.2192,
      "step": 111571
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.9746280908584595,
      "learning_rate": 0.0003148406358236548,
      "loss": 2.2386,
      "step": 111572
    },
    {
      "epoch": 1.45,
      "grad_norm": 1.0847617387771606,
      "learning_rate": 0.000314836550268301,
      "loss": 2.1414,
      "step": 111573
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.9390725493431091,
      "learning_rate": 0.00031483246471018887,
      "loss": 2.0608,
      "step": 111574
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.8079766035079956,
      "learning_rate": 0.0003148283791493191,
      "loss": 2.1791,
      "step": 111575
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.9315155744552612,
      "learning_rate": 0.00031482429358569244,
      "loss": 2.0393,
      "step": 111576
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.884166955947876,
      "learning_rate": 0.0003148202080193097,
      "loss": 2.132,
      "step": 111577
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.9175891876220703,
      "learning_rate": 0.0003148161224501715,
      "loss": 2.1092,
      "step": 111578
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.921215832233429,
      "learning_rate": 0.00031481203687827884,
      "loss": 2.0748,
      "step": 111579
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.912300705909729,
      "learning_rate": 0.00031480795130363223,
      "loss": 2.2205,
      "step": 111580
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.8792302012443542,
      "learning_rate": 0.0003148038657262326,
      "loss": 2.3125,
      "step": 111581
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.9319317936897278,
      "learning_rate": 0.00031479978014608065,
      "loss": 1.9067,
      "step": 111582
    },
    {
      "epoch": 1.45,
      "grad_norm": 1.039136290550232,
      "learning_rate": 0.00031479569456317714,
      "loss": 2.4046,
      "step": 111583
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.8589432239532471,
      "learning_rate": 0.00031479160897752274,
      "loss": 2.0816,
      "step": 111584
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.9063156247138977,
      "learning_rate": 0.0003147875233891185,
      "loss": 2.0971,
      "step": 111585
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.8720195889472961,
      "learning_rate": 0.00031478343779796483,
      "loss": 2.2765,
      "step": 111586
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.9411596655845642,
      "learning_rate": 0.00031477935220406263,
      "loss": 2.0411,
      "step": 111587
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.8950338959693909,
      "learning_rate": 0.0003147752666074128,
      "loss": 2.0438,
      "step": 111588
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.8569862246513367,
      "learning_rate": 0.00031477118100801585,
      "loss": 2.1981,
      "step": 111589
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.9502788782119751,
      "learning_rate": 0.0003147670954058727,
      "loss": 2.1382,
      "step": 111590
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.9744900465011597,
      "learning_rate": 0.00031476300980098405,
      "loss": 2.2078,
      "step": 111591
    },
    {
      "epoch": 1.45,
      "grad_norm": 1.4000180959701538,
      "learning_rate": 0.0003147589241933507,
      "loss": 2.4104,
      "step": 111592
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.9509218335151672,
      "learning_rate": 0.00031475483858297334,
      "loss": 2.2331,
      "step": 111593
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.9283376932144165,
      "learning_rate": 0.0003147507529698529,
      "loss": 2.0606,
      "step": 111594
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.9073578119277954,
      "learning_rate": 0.00031474666735398986,
      "loss": 2.1347,
      "step": 111595
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.875551164150238,
      "learning_rate": 0.00031474258173538514,
      "loss": 2.0539,
      "step": 111596
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.9087105393409729,
      "learning_rate": 0.00031473849611403967,
      "loss": 2.0848,
      "step": 111597
    },
    {
      "epoch": 1.45,
      "grad_norm": 1.00462806224823,
      "learning_rate": 0.0003147344104899539,
      "loss": 2.057,
      "step": 111598
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.8844181299209595,
      "learning_rate": 0.00031473032486312874,
      "loss": 2.1466,
      "step": 111599
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.9668064713478088,
      "learning_rate": 0.0003147262392335649,
      "loss": 2.4076,
      "step": 111600
    },
    {
      "epoch": 1.45,
      "grad_norm": 1.0056596994400024,
      "learning_rate": 0.00031472215360126325,
      "loss": 2.3761,
      "step": 111601
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.8637809157371521,
      "learning_rate": 0.00031471806796622446,
      "loss": 2.1307,
      "step": 111602
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.9289267659187317,
      "learning_rate": 0.0003147139823284492,
      "loss": 2.1707,
      "step": 111603
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.9448886513710022,
      "learning_rate": 0.0003147098966879385,
      "loss": 2.0987,
      "step": 111604
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.9126192927360535,
      "learning_rate": 0.00031470581104469286,
      "loss": 2.1225,
      "step": 111605
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.8126624226570129,
      "learning_rate": 0.0003147017253987131,
      "loss": 2.0571,
      "step": 111606
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.8329333662986755,
      "learning_rate": 0.0003146976397500001,
      "loss": 1.8678,
      "step": 111607
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.8698003888130188,
      "learning_rate": 0.0003146935540985545,
      "loss": 2.2093,
      "step": 111608
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.852196991443634,
      "learning_rate": 0.00031468946844437703,
      "loss": 2.1537,
      "step": 111609
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.9264763593673706,
      "learning_rate": 0.00031468538278746865,
      "loss": 2.1606,
      "step": 111610
    },
    {
      "epoch": 1.45,
      "grad_norm": 1.1348052024841309,
      "learning_rate": 0.00031468129712782986,
      "loss": 2.0591,
      "step": 111611
    },
    {
      "epoch": 1.45,
      "grad_norm": 1.0970394611358643,
      "learning_rate": 0.00031467721146546154,
      "loss": 1.9726,
      "step": 111612
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.9060561060905457,
      "learning_rate": 0.0003146731258003645,
      "loss": 2.0257,
      "step": 111613
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.9751620888710022,
      "learning_rate": 0.00031466904013253944,
      "loss": 2.0391,
      "step": 111614
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.9696133732795715,
      "learning_rate": 0.0003146649544619871,
      "loss": 2.0004,
      "step": 111615
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.8421714901924133,
      "learning_rate": 0.0003146608687887083,
      "loss": 2.2866,
      "step": 111616
    },
    {
      "epoch": 1.45,
      "grad_norm": 1.0044512748718262,
      "learning_rate": 0.00031465678311270377,
      "loss": 2.2673,
      "step": 111617
    },
    {
      "epoch": 1.45,
      "grad_norm": 1.1328239440917969,
      "learning_rate": 0.0003146526974339742,
      "loss": 2.2962,
      "step": 111618
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.9763361811637878,
      "learning_rate": 0.0003146486117525205,
      "loss": 1.9967,
      "step": 111619
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.9257099628448486,
      "learning_rate": 0.0003146445260683433,
      "loss": 2.2587,
      "step": 111620
    },
    {
      "epoch": 1.45,
      "grad_norm": 1.219738245010376,
      "learning_rate": 0.0003146404403814434,
      "loss": 1.9514,
      "step": 111621
    },
    {
      "epoch": 1.45,
      "grad_norm": 1.1683670282363892,
      "learning_rate": 0.0003146363546918216,
      "loss": 1.9519,
      "step": 111622
    },
    {
      "epoch": 1.45,
      "grad_norm": 1.1762551069259644,
      "learning_rate": 0.0003146322689994786,
      "loss": 2.0728,
      "step": 111623
    },
    {
      "epoch": 1.45,
      "grad_norm": 1.0921707153320312,
      "learning_rate": 0.00031462818330441517,
      "loss": 2.3739,
      "step": 111624
    },
    {
      "epoch": 1.45,
      "grad_norm": 1.3045663833618164,
      "learning_rate": 0.00031462409760663214,
      "loss": 2.2595,
      "step": 111625
    },
    {
      "epoch": 1.45,
      "grad_norm": 1.0178865194320679,
      "learning_rate": 0.00031462001190613015,
      "loss": 2.134,
      "step": 111626
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.8985016345977783,
      "learning_rate": 0.00031461592620291007,
      "loss": 2.1115,
      "step": 111627
    },
    {
      "epoch": 1.45,
      "grad_norm": 1.005797266960144,
      "learning_rate": 0.00031461184049697265,
      "loss": 2.3927,
      "step": 111628
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.9960640668869019,
      "learning_rate": 0.00031460775478831855,
      "loss": 2.2119,
      "step": 111629
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.906475305557251,
      "learning_rate": 0.00031460366907694857,
      "loss": 2.0436,
      "step": 111630
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.8807805776596069,
      "learning_rate": 0.0003145995833628636,
      "loss": 2.1972,
      "step": 111631
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.8623047471046448,
      "learning_rate": 0.0003145954976460642,
      "loss": 1.8811,
      "step": 111632
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.9498386979103088,
      "learning_rate": 0.0003145914119265512,
      "loss": 2.2118,
      "step": 111633
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.8763874769210815,
      "learning_rate": 0.0003145873262043255,
      "loss": 2.2817,
      "step": 111634
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.943304717540741,
      "learning_rate": 0.00031458324047938766,
      "loss": 2.2955,
      "step": 111635
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.8821395635604858,
      "learning_rate": 0.00031457915475173844,
      "loss": 2.1917,
      "step": 111636
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.8765712380409241,
      "learning_rate": 0.0003145750690213788,
      "loss": 2.0461,
      "step": 111637
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.9051389694213867,
      "learning_rate": 0.00031457098328830936,
      "loss": 2.197,
      "step": 111638
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.9125789999961853,
      "learning_rate": 0.00031456689755253084,
      "loss": 2.0702,
      "step": 111639
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.9001258015632629,
      "learning_rate": 0.00031456281181404416,
      "loss": 2.1607,
      "step": 111640
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.9428794980049133,
      "learning_rate": 0.00031455872607284995,
      "loss": 2.1918,
      "step": 111641
    },
    {
      "epoch": 1.45,
      "grad_norm": 1.0288728475570679,
      "learning_rate": 0.0003145546403289489,
      "loss": 1.9826,
      "step": 111642
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.9534596800804138,
      "learning_rate": 0.000314550554582342,
      "loss": 2.2641,
      "step": 111643
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.8951551914215088,
      "learning_rate": 0.0003145464688330298,
      "loss": 2.1761,
      "step": 111644
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.9329284429550171,
      "learning_rate": 0.0003145423830810131,
      "loss": 2.1197,
      "step": 111645
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.8809986114501953,
      "learning_rate": 0.0003145382973262928,
      "loss": 2.0893,
      "step": 111646
    },
    {
      "epoch": 1.45,
      "grad_norm": 1.013946294784546,
      "learning_rate": 0.0003145342115688695,
      "loss": 2.0338,
      "step": 111647
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.9116817116737366,
      "learning_rate": 0.000314530125808744,
      "loss": 2.1479,
      "step": 111648
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.9725853204727173,
      "learning_rate": 0.00031452604004591713,
      "loss": 2.2993,
      "step": 111649
    },
    {
      "epoch": 1.45,
      "grad_norm": 1.0233135223388672,
      "learning_rate": 0.0003145219542803896,
      "loss": 2.0781,
      "step": 111650
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.9721490144729614,
      "learning_rate": 0.00031451786851216203,
      "loss": 2.1423,
      "step": 111651
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.9742774963378906,
      "learning_rate": 0.0003145137827412354,
      "loss": 2.251,
      "step": 111652
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.9884515404701233,
      "learning_rate": 0.00031450969696761045,
      "loss": 2.0778,
      "step": 111653
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.9156497716903687,
      "learning_rate": 0.00031450561119128776,
      "loss": 2.522,
      "step": 111654
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.9240219593048096,
      "learning_rate": 0.0003145015254122683,
      "loss": 2.1929,
      "step": 111655
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.9239632487297058,
      "learning_rate": 0.00031449743963055256,
      "loss": 2.0609,
      "step": 111656
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.9578750729560852,
      "learning_rate": 0.00031449335384614173,
      "loss": 2.502,
      "step": 111657
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.8644732236862183,
      "learning_rate": 0.0003144892680590361,
      "loss": 2.1239,
      "step": 111658
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.9111306667327881,
      "learning_rate": 0.0003144851822692367,
      "loss": 2.0824,
      "step": 111659
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.9073323607444763,
      "learning_rate": 0.0003144810964767443,
      "loss": 2.1529,
      "step": 111660
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.9358804821968079,
      "learning_rate": 0.0003144770106815596,
      "loss": 2.1894,
      "step": 111661
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.9126319289207458,
      "learning_rate": 0.00031447292488368325,
      "loss": 2.1231,
      "step": 111662
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.9389495253562927,
      "learning_rate": 0.0003144688390831162,
      "loss": 2.046,
      "step": 111663
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.9780563116073608,
      "learning_rate": 0.00031446475327985904,
      "loss": 2.2802,
      "step": 111664
    },
    {
      "epoch": 1.45,
      "grad_norm": 1.044912576675415,
      "learning_rate": 0.00031446066747391264,
      "loss": 2.1205,
      "step": 111665
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.9177253842353821,
      "learning_rate": 0.0003144565816652777,
      "loss": 2.1208,
      "step": 111666
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.922946572303772,
      "learning_rate": 0.00031445249585395507,
      "loss": 2.3688,
      "step": 111667
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.8270137906074524,
      "learning_rate": 0.0003144484100399454,
      "loss": 2.0077,
      "step": 111668
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.868638813495636,
      "learning_rate": 0.0003144443242232495,
      "loss": 2.1504,
      "step": 111669
    },
    {
      "epoch": 1.45,
      "grad_norm": 1.0800365209579468,
      "learning_rate": 0.0003144402384038682,
      "loss": 2.1485,
      "step": 111670
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.9203406572341919,
      "learning_rate": 0.00031443615258180214,
      "loss": 2.3257,
      "step": 111671
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.8008150458335876,
      "learning_rate": 0.0003144320667570521,
      "loss": 2.0455,
      "step": 111672
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.9342460632324219,
      "learning_rate": 0.0003144279809296189,
      "loss": 2.0265,
      "step": 111673
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.874507486820221,
      "learning_rate": 0.00031442389509950323,
      "loss": 1.8781,
      "step": 111674
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.9025394916534424,
      "learning_rate": 0.000314419809266706,
      "loss": 2.0608,
      "step": 111675
    },
    {
      "epoch": 1.45,
      "grad_norm": 1.0403887033462524,
      "learning_rate": 0.00031441572343122765,
      "loss": 2.059,
      "step": 111676
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.8642013669013977,
      "learning_rate": 0.00031441163759306933,
      "loss": 2.2712,
      "step": 111677
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.9345007538795471,
      "learning_rate": 0.0003144075517522315,
      "loss": 2.0272,
      "step": 111678
    },
    {
      "epoch": 1.45,
      "grad_norm": 1.0681828260421753,
      "learning_rate": 0.0003144034659087151,
      "loss": 1.9415,
      "step": 111679
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.8980547785758972,
      "learning_rate": 0.0003143993800625208,
      "loss": 2.117,
      "step": 111680
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.9434970021247864,
      "learning_rate": 0.0003143952942136494,
      "loss": 2.152,
      "step": 111681
    },
    {
      "epoch": 1.45,
      "grad_norm": 1.14451265335083,
      "learning_rate": 0.0003143912083621016,
      "loss": 2.1709,
      "step": 111682
    },
    {
      "epoch": 1.45,
      "grad_norm": 1.0975042581558228,
      "learning_rate": 0.00031438712250787826,
      "loss": 2.0184,
      "step": 111683
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.930958092212677,
      "learning_rate": 0.00031438303665098005,
      "loss": 2.1928,
      "step": 111684
    },
    {
      "epoch": 1.45,
      "grad_norm": 1.0373718738555908,
      "learning_rate": 0.00031437895079140777,
      "loss": 2.0849,
      "step": 111685
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.9799695014953613,
      "learning_rate": 0.0003143748649291622,
      "loss": 2.0919,
      "step": 111686
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.9069153070449829,
      "learning_rate": 0.000314370779064244,
      "loss": 2.2477,
      "step": 111687
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.8925794363021851,
      "learning_rate": 0.00031436669319665405,
      "loss": 2.1712,
      "step": 111688
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.8750643730163574,
      "learning_rate": 0.0003143626073263931,
      "loss": 2.2282,
      "step": 111689
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.9605024456977844,
      "learning_rate": 0.0003143585214534618,
      "loss": 2.07,
      "step": 111690
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.9848057627677917,
      "learning_rate": 0.000314354435577861,
      "loss": 2.0893,
      "step": 111691
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.9364513158798218,
      "learning_rate": 0.00031435034969959147,
      "loss": 2.1787,
      "step": 111692
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.9090813398361206,
      "learning_rate": 0.0003143462638186539,
      "loss": 2.21,
      "step": 111693
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.9054052829742432,
      "learning_rate": 0.000314342177935049,
      "loss": 2.2541,
      "step": 111694
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.9885392189025879,
      "learning_rate": 0.00031433809204877783,
      "loss": 2.1584,
      "step": 111695
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.9175471663475037,
      "learning_rate": 0.0003143340061598408,
      "loss": 1.9954,
      "step": 111696
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.8790422081947327,
      "learning_rate": 0.0003143299202682388,
      "loss": 2.0299,
      "step": 111697
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.9956991076469421,
      "learning_rate": 0.00031432583437397265,
      "loss": 2.0657,
      "step": 111698
    },
    {
      "epoch": 1.45,
      "grad_norm": 1.060699224472046,
      "learning_rate": 0.000314321748477043,
      "loss": 2.0986,
      "step": 111699
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.8522549867630005,
      "learning_rate": 0.00031431766257745074,
      "loss": 2.1466,
      "step": 111700
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.9529100060462952,
      "learning_rate": 0.00031431357667519657,
      "loss": 2.1557,
      "step": 111701
    },
    {
      "epoch": 1.45,
      "grad_norm": 1.1800686120986938,
      "learning_rate": 0.00031430949077028116,
      "loss": 2.2635,
      "step": 111702
    },
    {
      "epoch": 1.45,
      "grad_norm": 1.0642330646514893,
      "learning_rate": 0.00031430540486270534,
      "loss": 2.2734,
      "step": 111703
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.906834602355957,
      "learning_rate": 0.00031430131895246997,
      "loss": 2.1787,
      "step": 111704
    },
    {
      "epoch": 1.45,
      "grad_norm": 1.0658007860183716,
      "learning_rate": 0.0003142972330395756,
      "loss": 1.9198,
      "step": 111705
    },
    {
      "epoch": 1.45,
      "grad_norm": 1.0941977500915527,
      "learning_rate": 0.00031429314712402314,
      "loss": 1.9063,
      "step": 111706
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.9758147597312927,
      "learning_rate": 0.0003142890612058134,
      "loss": 1.8337,
      "step": 111707
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.9792047739028931,
      "learning_rate": 0.00031428497528494695,
      "loss": 2.3282,
      "step": 111708
    },
    {
      "epoch": 1.45,
      "grad_norm": 1.0613933801651,
      "learning_rate": 0.0003142808893614246,
      "loss": 2.1281,
      "step": 111709
    },
    {
      "epoch": 1.45,
      "grad_norm": 1.0969510078430176,
      "learning_rate": 0.00031427680343524737,
      "loss": 1.9659,
      "step": 111710
    },
    {
      "epoch": 1.45,
      "grad_norm": 1.068217396736145,
      "learning_rate": 0.00031427271750641565,
      "loss": 2.1327,
      "step": 111711
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.8343271017074585,
      "learning_rate": 0.0003142686315749304,
      "loss": 2.154,
      "step": 111712
    },
    {
      "epoch": 1.45,
      "grad_norm": 1.174817681312561,
      "learning_rate": 0.0003142645456407924,
      "loss": 2.0195,
      "step": 111713
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.9288954734802246,
      "learning_rate": 0.00031426045970400226,
      "loss": 2.0706,
      "step": 111714
    },
    {
      "epoch": 1.45,
      "grad_norm": 1.1013344526290894,
      "learning_rate": 0.00031425637376456087,
      "loss": 2.1198,
      "step": 111715
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.9342306852340698,
      "learning_rate": 0.000314252287822469,
      "loss": 2.3495,
      "step": 111716
    },
    {
      "epoch": 1.45,
      "grad_norm": 1.0507766008377075,
      "learning_rate": 0.00031424820187772725,
      "loss": 1.8387,
      "step": 111717
    },
    {
      "epoch": 1.45,
      "grad_norm": 1.0127476453781128,
      "learning_rate": 0.00031424411593033655,
      "loss": 2.1057,
      "step": 111718
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.8913450241088867,
      "learning_rate": 0.0003142400299802976,
      "loss": 2.0899,
      "step": 111719
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.8919480443000793,
      "learning_rate": 0.0003142359440276112,
      "loss": 2.1808,
      "step": 111720
    },
    {
      "epoch": 1.45,
      "grad_norm": 1.0179158449172974,
      "learning_rate": 0.000314231858072278,
      "loss": 2.0372,
      "step": 111721
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.9712009429931641,
      "learning_rate": 0.00031422777211429886,
      "loss": 2.2871,
      "step": 111722
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.8366283178329468,
      "learning_rate": 0.0003142236861536745,
      "loss": 1.8623,
      "step": 111723
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.9185066819190979,
      "learning_rate": 0.0003142196001904057,
      "loss": 2.2331,
      "step": 111724
    },
    {
      "epoch": 1.45,
      "grad_norm": 1.0026795864105225,
      "learning_rate": 0.00031421551422449324,
      "loss": 2.0528,
      "step": 111725
    },
    {
      "epoch": 1.45,
      "grad_norm": 1.1912904977798462,
      "learning_rate": 0.00031421142825593777,
      "loss": 2.0863,
      "step": 111726
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.9541521072387695,
      "learning_rate": 0.0003142073422847402,
      "loss": 1.9809,
      "step": 111727
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.9630423188209534,
      "learning_rate": 0.0003142032563109012,
      "loss": 2.0534,
      "step": 111728
    },
    {
      "epoch": 1.45,
      "grad_norm": 1.1393126249313354,
      "learning_rate": 0.0003141991703344215,
      "loss": 2.1892,
      "step": 111729
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.995114803314209,
      "learning_rate": 0.0003141950843553019,
      "loss": 2.1447,
      "step": 111730
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.9593377113342285,
      "learning_rate": 0.0003141909983735433,
      "loss": 2.1446,
      "step": 111731
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.9220759868621826,
      "learning_rate": 0.0003141869123891462,
      "loss": 1.9165,
      "step": 111732
    },
    {
      "epoch": 1.45,
      "grad_norm": 1.1516878604888916,
      "learning_rate": 0.0003141828264021115,
      "loss": 2.042,
      "step": 111733
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.9361237287521362,
      "learning_rate": 0.00031417874041243994,
      "loss": 2.0977,
      "step": 111734
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.948248028755188,
      "learning_rate": 0.0003141746544201323,
      "loss": 2.0567,
      "step": 111735
    },
    {
      "epoch": 1.45,
      "grad_norm": 1.0329114198684692,
      "learning_rate": 0.0003141705684251894,
      "loss": 1.9697,
      "step": 111736
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.9162577390670776,
      "learning_rate": 0.0003141664824276118,
      "loss": 2.2976,
      "step": 111737
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.8702071309089661,
      "learning_rate": 0.00031416239642740045,
      "loss": 2.1749,
      "step": 111738
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.9423650503158569,
      "learning_rate": 0.00031415831042455605,
      "loss": 2.2811,
      "step": 111739
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.9105809926986694,
      "learning_rate": 0.00031415422441907927,
      "loss": 1.9633,
      "step": 111740
    },
    {
      "epoch": 1.45,
      "grad_norm": 1.086911916732788,
      "learning_rate": 0.00031415013841097107,
      "loss": 2.0337,
      "step": 111741
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.9183623790740967,
      "learning_rate": 0.000314146052400232,
      "loss": 2.2452,
      "step": 111742
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.8682968616485596,
      "learning_rate": 0.00031414196638686293,
      "loss": 2.0522,
      "step": 111743
    },
    {
      "epoch": 1.45,
      "grad_norm": 1.0045427083969116,
      "learning_rate": 0.0003141378803708646,
      "loss": 1.9405,
      "step": 111744
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.925843358039856,
      "learning_rate": 0.0003141337943522379,
      "loss": 2.2776,
      "step": 111745
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.8955087661743164,
      "learning_rate": 0.0003141297083309833,
      "loss": 2.2372,
      "step": 111746
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.9161674976348877,
      "learning_rate": 0.00031412562230710174,
      "loss": 2.0596,
      "step": 111747
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.9598469734191895,
      "learning_rate": 0.0003141215362805941,
      "loss": 1.9646,
      "step": 111748
    },
    {
      "epoch": 1.46,
      "grad_norm": 1.1773220300674438,
      "learning_rate": 0.00031411745025146083,
      "loss": 2.0831,
      "step": 111749
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.8430455923080444,
      "learning_rate": 0.0003141133642197029,
      "loss": 2.0958,
      "step": 111750
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.8900221586227417,
      "learning_rate": 0.00031410927818532105,
      "loss": 2.1319,
      "step": 111751
    },
    {
      "epoch": 1.46,
      "grad_norm": 1.2014237642288208,
      "learning_rate": 0.00031410519214831605,
      "loss": 2.2449,
      "step": 111752
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.9088776111602783,
      "learning_rate": 0.0003141011061086886,
      "loss": 2.1207,
      "step": 111753
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.9403409361839294,
      "learning_rate": 0.00031409702006643945,
      "loss": 2.3914,
      "step": 111754
    },
    {
      "epoch": 1.46,
      "grad_norm": 1.3969045877456665,
      "learning_rate": 0.00031409293402156937,
      "loss": 1.9425,
      "step": 111755
    },
    {
      "epoch": 1.46,
      "grad_norm": 1.047067403793335,
      "learning_rate": 0.0003140888479740793,
      "loss": 2.1553,
      "step": 111756
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.8964583277702332,
      "learning_rate": 0.0003140847619239698,
      "loss": 2.0799,
      "step": 111757
    },
    {
      "epoch": 1.46,
      "grad_norm": 1.0180976390838623,
      "learning_rate": 0.0003140806758712415,
      "loss": 2.1465,
      "step": 111758
    },
    {
      "epoch": 1.46,
      "grad_norm": 1.008322834968567,
      "learning_rate": 0.0003140765898158956,
      "loss": 2.1417,
      "step": 111759
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.8713087439537048,
      "learning_rate": 0.0003140725037579324,
      "loss": 2.1001,
      "step": 111760
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.8977609872817993,
      "learning_rate": 0.0003140684176973529,
      "loss": 2.3999,
      "step": 111761
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.9429848194122314,
      "learning_rate": 0.0003140643316341579,
      "loss": 2.2497,
      "step": 111762
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.9606290459632874,
      "learning_rate": 0.000314060245568348,
      "loss": 2.3696,
      "step": 111763
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.9296522736549377,
      "learning_rate": 0.00031405615949992407,
      "loss": 2.1559,
      "step": 111764
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.8385616540908813,
      "learning_rate": 0.0003140520734288868,
      "loss": 2.152,
      "step": 111765
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.9471758008003235,
      "learning_rate": 0.000314047987355237,
      "loss": 2.037,
      "step": 111766
    },
    {
      "epoch": 1.46,
      "grad_norm": 1.0846726894378662,
      "learning_rate": 0.0003140439012789754,
      "loss": 2.228,
      "step": 111767
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.9155973792076111,
      "learning_rate": 0.00031403981520010285,
      "loss": 2.0468,
      "step": 111768
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.9352647066116333,
      "learning_rate": 0.0003140357291186199,
      "loss": 1.8132,
      "step": 111769
    },
    {
      "epoch": 1.46,
      "grad_norm": 1.2275549173355103,
      "learning_rate": 0.0003140316430345275,
      "loss": 2.1255,
      "step": 111770
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.9384800791740417,
      "learning_rate": 0.0003140275569478265,
      "loss": 2.2469,
      "step": 111771
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.9623285531997681,
      "learning_rate": 0.0003140234708585173,
      "loss": 2.153,
      "step": 111772
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.8549976348876953,
      "learning_rate": 0.0003140193847666009,
      "loss": 2.0877,
      "step": 111773
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.9776506423950195,
      "learning_rate": 0.0003140152986720781,
      "loss": 2.0393,
      "step": 111774
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.8788241744041443,
      "learning_rate": 0.0003140112125749496,
      "loss": 2.1673,
      "step": 111775
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.9271631836891174,
      "learning_rate": 0.0003140071264752161,
      "loss": 2.4158,
      "step": 111776
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.8956488966941833,
      "learning_rate": 0.00031400304037287847,
      "loss": 2.2991,
      "step": 111777
    },
    {
      "epoch": 1.46,
      "grad_norm": 1.0601171255111694,
      "learning_rate": 0.00031399895426793736,
      "loss": 2.1577,
      "step": 111778
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.8755055665969849,
      "learning_rate": 0.0003139948681603936,
      "loss": 2.1753,
      "step": 111779
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.9546383619308472,
      "learning_rate": 0.000313990782050248,
      "loss": 1.8751,
      "step": 111780
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.9083042740821838,
      "learning_rate": 0.0003139866959375011,
      "loss": 2.3139,
      "step": 111781
    },
    {
      "epoch": 1.46,
      "grad_norm": 1.000003695487976,
      "learning_rate": 0.0003139826098221539,
      "loss": 2.3524,
      "step": 111782
    },
    {
      "epoch": 1.46,
      "grad_norm": 1.0302127599716187,
      "learning_rate": 0.0003139785237042071,
      "loss": 2.2307,
      "step": 111783
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.8747624158859253,
      "learning_rate": 0.0003139744375836614,
      "loss": 2.4324,
      "step": 111784
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.9983958005905151,
      "learning_rate": 0.00031397035146051754,
      "loss": 2.0982,
      "step": 111785
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.8589383363723755,
      "learning_rate": 0.00031396626533477634,
      "loss": 2.1682,
      "step": 111786
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.878463864326477,
      "learning_rate": 0.0003139621792064386,
      "loss": 2.1827,
      "step": 111787
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.9860370755195618,
      "learning_rate": 0.0003139580930755049,
      "loss": 2.049,
      "step": 111788
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.8373212218284607,
      "learning_rate": 0.0003139540069419763,
      "loss": 2.2608,
      "step": 111789
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.9508141875267029,
      "learning_rate": 0.00031394992080585333,
      "loss": 2.4006,
      "step": 111790
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.9637361764907837,
      "learning_rate": 0.0003139458346671367,
      "loss": 2.4986,
      "step": 111791
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.9847629070281982,
      "learning_rate": 0.00031394174852582746,
      "loss": 2.2386,
      "step": 111792
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.9890273213386536,
      "learning_rate": 0.0003139376623819261,
      "loss": 2.0094,
      "step": 111793
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.9592689275741577,
      "learning_rate": 0.00031393357623543336,
      "loss": 2.0925,
      "step": 111794
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.9918267130851746,
      "learning_rate": 0.00031392949008635025,
      "loss": 2.2124,
      "step": 111795
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.9309803247451782,
      "learning_rate": 0.0003139254039346773,
      "loss": 2.0857,
      "step": 111796
    },
    {
      "epoch": 1.46,
      "grad_norm": 1.0195207595825195,
      "learning_rate": 0.00031392131778041536,
      "loss": 1.9286,
      "step": 111797
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.8624858260154724,
      "learning_rate": 0.00031391723162356525,
      "loss": 2.0809,
      "step": 111798
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.9401440024375916,
      "learning_rate": 0.00031391314546412764,
      "loss": 2.3103,
      "step": 111799
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.9543249011039734,
      "learning_rate": 0.0003139090593021033,
      "loss": 2.2301,
      "step": 111800
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.8362281322479248,
      "learning_rate": 0.000313904973137493,
      "loss": 2.1926,
      "step": 111801
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.8842787146568298,
      "learning_rate": 0.00031390088697029754,
      "loss": 2.049,
      "step": 111802
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.8645119071006775,
      "learning_rate": 0.00031389680080051756,
      "loss": 2.0223,
      "step": 111803
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.9530190825462341,
      "learning_rate": 0.00031389271462815393,
      "loss": 2.254,
      "step": 111804
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.9836248159408569,
      "learning_rate": 0.00031388862845320746,
      "loss": 1.9248,
      "step": 111805
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.9285959601402283,
      "learning_rate": 0.00031388454227567874,
      "loss": 2.113,
      "step": 111806
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.9016154408454895,
      "learning_rate": 0.00031388045609556864,
      "loss": 2.0279,
      "step": 111807
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.9486311078071594,
      "learning_rate": 0.00031387636991287793,
      "loss": 2.2907,
      "step": 111808
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.939509928226471,
      "learning_rate": 0.0003138722837276073,
      "loss": 2.1414,
      "step": 111809
    },
    {
      "epoch": 1.46,
      "grad_norm": 1.004836916923523,
      "learning_rate": 0.0003138681975397576,
      "loss": 1.9992,
      "step": 111810
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.8654463887214661,
      "learning_rate": 0.00031386411134932957,
      "loss": 1.9539,
      "step": 111811
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.9855483174324036,
      "learning_rate": 0.00031386002515632386,
      "loss": 2.1883,
      "step": 111812
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.8481440544128418,
      "learning_rate": 0.0003138559389607413,
      "loss": 2.2771,
      "step": 111813
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.9521438479423523,
      "learning_rate": 0.00031385185276258274,
      "loss": 2.1097,
      "step": 111814
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.9255713224411011,
      "learning_rate": 0.00031384776656184885,
      "loss": 2.1781,
      "step": 111815
    },
    {
      "epoch": 1.46,
      "grad_norm": 1.0318999290466309,
      "learning_rate": 0.0003138436803585403,
      "loss": 2.0843,
      "step": 111816
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.9466835260391235,
      "learning_rate": 0.00031383959415265806,
      "loss": 2.4077,
      "step": 111817
    },
    {
      "epoch": 1.46,
      "grad_norm": 1.2509516477584839,
      "learning_rate": 0.0003138355079442027,
      "loss": 2.0295,
      "step": 111818
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.9012722969055176,
      "learning_rate": 0.00031383142173317506,
      "loss": 2.2733,
      "step": 111819
    },
    {
      "epoch": 1.46,
      "grad_norm": 1.155124306678772,
      "learning_rate": 0.00031382733551957596,
      "loss": 2.2091,
      "step": 111820
    },
    {
      "epoch": 1.46,
      "grad_norm": 1.0259051322937012,
      "learning_rate": 0.00031382324930340606,
      "loss": 2.126,
      "step": 111821
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.8136170506477356,
      "learning_rate": 0.00031381916308466617,
      "loss": 2.0331,
      "step": 111822
    },
    {
      "epoch": 1.46,
      "grad_norm": 1.1122775077819824,
      "learning_rate": 0.00031381507686335705,
      "loss": 2.1986,
      "step": 111823
    },
    {
      "epoch": 1.46,
      "grad_norm": 1.1108769178390503,
      "learning_rate": 0.00031381099063947944,
      "loss": 2.2291,
      "step": 111824
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.9285424947738647,
      "learning_rate": 0.000313806904413034,
      "loss": 2.273,
      "step": 111825
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.9191038608551025,
      "learning_rate": 0.0003138028181840218,
      "loss": 2.0947,
      "step": 111826
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.9065505266189575,
      "learning_rate": 0.0003137987319524432,
      "loss": 2.1883,
      "step": 111827
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.9713374376296997,
      "learning_rate": 0.00031379464571829925,
      "loss": 2.1158,
      "step": 111828
    },
    {
      "epoch": 1.46,
      "grad_norm": 1.0244146585464478,
      "learning_rate": 0.00031379055948159066,
      "loss": 2.3795,
      "step": 111829
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.8593937158584595,
      "learning_rate": 0.00031378647324231803,
      "loss": 1.9442,
      "step": 111830
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.9289997816085815,
      "learning_rate": 0.00031378238700048225,
      "loss": 2.0883,
      "step": 111831
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.9621659517288208,
      "learning_rate": 0.00031377830075608417,
      "loss": 2.0923,
      "step": 111832
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.9123408794403076,
      "learning_rate": 0.0003137742145091244,
      "loss": 1.9974,
      "step": 111833
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.9150021076202393,
      "learning_rate": 0.0003137701282596037,
      "loss": 2.1018,
      "step": 111834
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.9373008012771606,
      "learning_rate": 0.0003137660420075229,
      "loss": 2.231,
      "step": 111835
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.9015513062477112,
      "learning_rate": 0.0003137619557528826,
      "loss": 2.1545,
      "step": 111836
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.918756365776062,
      "learning_rate": 0.00031375786949568396,
      "loss": 2.1651,
      "step": 111837
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.9897777438163757,
      "learning_rate": 0.00031375378323592724,
      "loss": 2.3368,
      "step": 111838
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.8909107446670532,
      "learning_rate": 0.00031374969697361353,
      "loss": 2.2927,
      "step": 111839
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.9844528436660767,
      "learning_rate": 0.00031374561070874354,
      "loss": 2.2018,
      "step": 111840
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.9357780814170837,
      "learning_rate": 0.0003137415244413179,
      "loss": 2.2281,
      "step": 111841
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.9276766777038574,
      "learning_rate": 0.00031373743817133745,
      "loss": 2.2535,
      "step": 111842
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.9304050207138062,
      "learning_rate": 0.00031373335189880297,
      "loss": 2.1893,
      "step": 111843
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.8978216052055359,
      "learning_rate": 0.0003137292656237153,
      "loss": 1.977,
      "step": 111844
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.9935089349746704,
      "learning_rate": 0.00031372517934607493,
      "loss": 2.173,
      "step": 111845
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.8974670171737671,
      "learning_rate": 0.00031372109306588284,
      "loss": 2.0142,
      "step": 111846
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.9972617626190186,
      "learning_rate": 0.0003137170067831399,
      "loss": 2.2671,
      "step": 111847
    },
    {
      "epoch": 1.46,
      "grad_norm": 1.209047555923462,
      "learning_rate": 0.0003137129204978466,
      "loss": 2.1925,
      "step": 111848
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.9154343605041504,
      "learning_rate": 0.0003137088342100037,
      "loss": 2.0623,
      "step": 111849
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.9426959156990051,
      "learning_rate": 0.0003137047479196122,
      "loss": 2.1435,
      "step": 111850
    },
    {
      "epoch": 1.46,
      "grad_norm": 1.0302339792251587,
      "learning_rate": 0.0003137006616266727,
      "loss": 1.8367,
      "step": 111851
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.9796910881996155,
      "learning_rate": 0.00031369657533118593,
      "loss": 2.1739,
      "step": 111852
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.898291826248169,
      "learning_rate": 0.0003136924890331528,
      "loss": 2.2845,
      "step": 111853
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.9040860533714294,
      "learning_rate": 0.0003136884027325739,
      "loss": 2.1755,
      "step": 111854
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.8055881857872009,
      "learning_rate": 0.0003136843164294501,
      "loss": 2.1723,
      "step": 111855
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.822385311126709,
      "learning_rate": 0.00031368023012378216,
      "loss": 2.0028,
      "step": 111856
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.8495786190032959,
      "learning_rate": 0.00031367614381557076,
      "loss": 2.0676,
      "step": 111857
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.8593990802764893,
      "learning_rate": 0.00031367205750481675,
      "loss": 2.146,
      "step": 111858
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.9853347539901733,
      "learning_rate": 0.00031366797119152087,
      "loss": 2.1683,
      "step": 111859
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.8595191836357117,
      "learning_rate": 0.00031366388487568374,
      "loss": 2.253,
      "step": 111860
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.9610924124717712,
      "learning_rate": 0.00031365979855730626,
      "loss": 2.118,
      "step": 111861
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.8786740303039551,
      "learning_rate": 0.00031365571223638926,
      "loss": 2.1494,
      "step": 111862
    },
    {
      "epoch": 1.46,
      "grad_norm": 1.0437219142913818,
      "learning_rate": 0.0003136516259129334,
      "loss": 1.9605,
      "step": 111863
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.9136607646942139,
      "learning_rate": 0.0003136475395869393,
      "loss": 2.336,
      "step": 111864
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.9274058938026428,
      "learning_rate": 0.000313643453258408,
      "loss": 2.1243,
      "step": 111865
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.9971098899841309,
      "learning_rate": 0.0003136393669273401,
      "loss": 2.2092,
      "step": 111866
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.9749689102172852,
      "learning_rate": 0.00031363528059373633,
      "loss": 2.1359,
      "step": 111867
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.9529969096183777,
      "learning_rate": 0.00031363119425759756,
      "loss": 2.2007,
      "step": 111868
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.9073338508605957,
      "learning_rate": 0.0003136271079189245,
      "loss": 2.0608,
      "step": 111869
    },
    {
      "epoch": 1.46,
      "grad_norm": 1.1464492082595825,
      "learning_rate": 0.0003136230215777178,
      "loss": 2.2058,
      "step": 111870
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.8978332281112671,
      "learning_rate": 0.00031361893523397847,
      "loss": 2.1651,
      "step": 111871
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.9226914644241333,
      "learning_rate": 0.00031361484888770707,
      "loss": 2.1062,
      "step": 111872
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.8997659683227539,
      "learning_rate": 0.00031361076253890434,
      "loss": 2.1176,
      "step": 111873
    },
    {
      "epoch": 1.46,
      "grad_norm": 1.317344307899475,
      "learning_rate": 0.0003136066761875712,
      "loss": 1.9347,
      "step": 111874
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.9633291959762573,
      "learning_rate": 0.0003136025898337083,
      "loss": 1.9543,
      "step": 111875
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.897759735584259,
      "learning_rate": 0.0003135985034773164,
      "loss": 2.1486,
      "step": 111876
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.9009174704551697,
      "learning_rate": 0.00031359441711839624,
      "loss": 1.9639,
      "step": 111877
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.9115771055221558,
      "learning_rate": 0.0003135903307569488,
      "loss": 2.2442,
      "step": 111878
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.9021275639533997,
      "learning_rate": 0.00031358624439297453,
      "loss": 2.2741,
      "step": 111879
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.9319818615913391,
      "learning_rate": 0.00031358215802647425,
      "loss": 2.2177,
      "step": 111880
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.9230379462242126,
      "learning_rate": 0.0003135780716574489,
      "loss": 2.321,
      "step": 111881
    },
    {
      "epoch": 1.46,
      "grad_norm": 1.1371407508850098,
      "learning_rate": 0.0003135739852858991,
      "loss": 1.8621,
      "step": 111882
    },
    {
      "epoch": 1.46,
      "grad_norm": 1.0634971857070923,
      "learning_rate": 0.00031356989891182565,
      "loss": 1.9929,
      "step": 111883
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.9116109013557434,
      "learning_rate": 0.00031356581253522934,
      "loss": 2.1356,
      "step": 111884
    },
    {
      "epoch": 1.46,
      "grad_norm": 1.0009899139404297,
      "learning_rate": 0.0003135617261561108,
      "loss": 2.047,
      "step": 111885
    },
    {
      "epoch": 1.46,
      "grad_norm": 1.0294156074523926,
      "learning_rate": 0.00031355763977447093,
      "loss": 2.153,
      "step": 111886
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.9160667657852173,
      "learning_rate": 0.0003135535533903104,
      "loss": 2.0772,
      "step": 111887
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.9844284057617188,
      "learning_rate": 0.00031354946700363004,
      "loss": 2.2663,
      "step": 111888
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.9022402167320251,
      "learning_rate": 0.0003135453806144306,
      "loss": 2.2399,
      "step": 111889
    },
    {
      "epoch": 1.46,
      "grad_norm": 1.1894068717956543,
      "learning_rate": 0.0003135412942227128,
      "loss": 2.2484,
      "step": 111890
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.901946485042572,
      "learning_rate": 0.0003135372078284774,
      "loss": 2.1361,
      "step": 111891
    },
    {
      "epoch": 1.46,
      "grad_norm": 1.05708646774292,
      "learning_rate": 0.0003135331214317252,
      "loss": 2.3298,
      "step": 111892
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.9250395894050598,
      "learning_rate": 0.00031352903503245697,
      "loss": 2.3567,
      "step": 111893
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.9437428712844849,
      "learning_rate": 0.00031352494863067334,
      "loss": 2.2135,
      "step": 111894
    },
    {
      "epoch": 1.46,
      "grad_norm": 1.0162097215652466,
      "learning_rate": 0.00031352086222637517,
      "loss": 2.1636,
      "step": 111895
    },
    {
      "epoch": 1.46,
      "grad_norm": 1.0938339233398438,
      "learning_rate": 0.00031351677581956333,
      "loss": 2.2258,
      "step": 111896
    },
    {
      "epoch": 1.46,
      "grad_norm": 1.0038645267486572,
      "learning_rate": 0.00031351268941023837,
      "loss": 2.1783,
      "step": 111897
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.9611344933509827,
      "learning_rate": 0.00031350860299840115,
      "loss": 2.1835,
      "step": 111898
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.8864323496818542,
      "learning_rate": 0.00031350451658405254,
      "loss": 2.1683,
      "step": 111899
    },
    {
      "epoch": 1.46,
      "grad_norm": 1.3654769659042358,
      "learning_rate": 0.0003135004301671931,
      "loss": 2.0762,
      "step": 111900
    },
    {
      "epoch": 1.46,
      "grad_norm": 1.2042489051818848,
      "learning_rate": 0.0003134963437478236,
      "loss": 2.1031,
      "step": 111901
    },
    {
      "epoch": 1.46,
      "grad_norm": 1.0178499221801758,
      "learning_rate": 0.00031349225732594496,
      "loss": 2.2518,
      "step": 111902
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.9338394999504089,
      "learning_rate": 0.00031348817090155783,
      "loss": 2.2297,
      "step": 111903
    },
    {
      "epoch": 1.46,
      "grad_norm": 1.0351812839508057,
      "learning_rate": 0.00031348408447466305,
      "loss": 2.1431,
      "step": 111904
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.9806657433509827,
      "learning_rate": 0.0003134799980452613,
      "loss": 2.2446,
      "step": 111905
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.9866716265678406,
      "learning_rate": 0.00031347591161335334,
      "loss": 2.1607,
      "step": 111906
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.8671635985374451,
      "learning_rate": 0.00031347182517893994,
      "loss": 2.2338,
      "step": 111907
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.9921912550926208,
      "learning_rate": 0.00031346773874202187,
      "loss": 1.9437,
      "step": 111908
    },
    {
      "epoch": 1.46,
      "grad_norm": 1.0426623821258545,
      "learning_rate": 0.0003134636523025999,
      "loss": 2.169,
      "step": 111909
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.8630499243736267,
      "learning_rate": 0.0003134595658606748,
      "loss": 1.8985,
      "step": 111910
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.9023260474205017,
      "learning_rate": 0.00031345547941624734,
      "loss": 1.8156,
      "step": 111911
    },
    {
      "epoch": 1.46,
      "grad_norm": 1.2335988283157349,
      "learning_rate": 0.00031345139296931824,
      "loss": 2.027,
      "step": 111912
    },
    {
      "epoch": 1.46,
      "grad_norm": 1.2104312181472778,
      "learning_rate": 0.0003134473065198882,
      "loss": 2.2521,
      "step": 111913
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.8726971745491028,
      "learning_rate": 0.00031344322006795816,
      "loss": 2.3027,
      "step": 111914
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.8786383867263794,
      "learning_rate": 0.0003134391336135287,
      "loss": 1.9302,
      "step": 111915
    },
    {
      "epoch": 1.46,
      "grad_norm": 1.2735143899917603,
      "learning_rate": 0.00031343504715660066,
      "loss": 2.0527,
      "step": 111916
    },
    {
      "epoch": 1.46,
      "grad_norm": 1.1152338981628418,
      "learning_rate": 0.00031343096069717477,
      "loss": 1.921,
      "step": 111917
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.9281551241874695,
      "learning_rate": 0.0003134268742352519,
      "loss": 2.1333,
      "step": 111918
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.9465569257736206,
      "learning_rate": 0.00031342278777083267,
      "loss": 2.2628,
      "step": 111919
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.9672996401786804,
      "learning_rate": 0.0003134187013039178,
      "loss": 2.2774,
      "step": 111920
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.9856206178665161,
      "learning_rate": 0.0003134146148345083,
      "loss": 2.1302,
      "step": 111921
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.8341097831726074,
      "learning_rate": 0.0003134105283626047,
      "loss": 2.1628,
      "step": 111922
    },
    {
      "epoch": 1.46,
      "grad_norm": 1.0168073177337646,
      "learning_rate": 0.00031340644188820784,
      "loss": 2.0434,
      "step": 111923
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.9156277775764465,
      "learning_rate": 0.00031340235541131844,
      "loss": 1.9461,
      "step": 111924
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.9663193225860596,
      "learning_rate": 0.00031339826893193726,
      "loss": 2.1587,
      "step": 111925
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.9037393927574158,
      "learning_rate": 0.0003133941824500652,
      "loss": 2.1023,
      "step": 111926
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.9444751143455505,
      "learning_rate": 0.00031339009596570276,
      "loss": 2.0508,
      "step": 111927
    },
    {
      "epoch": 1.46,
      "grad_norm": 1.0285707712173462,
      "learning_rate": 0.0003133860094788509,
      "loss": 2.2033,
      "step": 111928
    },
    {
      "epoch": 1.46,
      "grad_norm": 1.0427305698394775,
      "learning_rate": 0.0003133819229895104,
      "loss": 2.1511,
      "step": 111929
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.9115853905677795,
      "learning_rate": 0.0003133778364976819,
      "loss": 1.9486,
      "step": 111930
    },
    {
      "epoch": 1.46,
      "grad_norm": 1.0521714687347412,
      "learning_rate": 0.0003133737500033662,
      "loss": 2.1871,
      "step": 111931
    },
    {
      "epoch": 1.46,
      "grad_norm": 1.0104540586471558,
      "learning_rate": 0.0003133696635065641,
      "loss": 2.2955,
      "step": 111932
    },
    {
      "epoch": 1.46,
      "grad_norm": 1.0395535230636597,
      "learning_rate": 0.0003133655770072763,
      "loss": 2.3172,
      "step": 111933
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.9520878195762634,
      "learning_rate": 0.0003133614905055036,
      "loss": 2.1665,
      "step": 111934
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.9008544683456421,
      "learning_rate": 0.0003133574040012467,
      "loss": 2.0671,
      "step": 111935
    },
    {
      "epoch": 1.46,
      "grad_norm": 1.095818281173706,
      "learning_rate": 0.0003133533174945065,
      "loss": 2.4592,
      "step": 111936
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.9051945209503174,
      "learning_rate": 0.0003133492309852836,
      "loss": 2.1721,
      "step": 111937
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.8557048439979553,
      "learning_rate": 0.00031334514447357875,
      "loss": 2.2547,
      "step": 111938
    },
    {
      "epoch": 1.46,
      "grad_norm": 1.010851263999939,
      "learning_rate": 0.000313341057959393,
      "loss": 2.2373,
      "step": 111939
    },
    {
      "epoch": 1.46,
      "grad_norm": 1.14535653591156,
      "learning_rate": 0.0003133369714427267,
      "loss": 2.1579,
      "step": 111940
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.9915801882743835,
      "learning_rate": 0.00031333288492358085,
      "loss": 2.263,
      "step": 111941
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.9360813498497009,
      "learning_rate": 0.00031332879840195623,
      "loss": 2.4386,
      "step": 111942
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.9144355058670044,
      "learning_rate": 0.00031332471187785346,
      "loss": 2.1994,
      "step": 111943
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.9967785477638245,
      "learning_rate": 0.0003133206253512734,
      "loss": 2.1201,
      "step": 111944
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.9855625629425049,
      "learning_rate": 0.00031331653882221683,
      "loss": 2.1096,
      "step": 111945
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.9554757475852966,
      "learning_rate": 0.00031331245229068445,
      "loss": 2.1274,
      "step": 111946
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.8928216099739075,
      "learning_rate": 0.000313308365756677,
      "loss": 2.3006,
      "step": 111947
    },
    {
      "epoch": 1.46,
      "grad_norm": 1.0565481185913086,
      "learning_rate": 0.0003133042792201953,
      "loss": 2.1461,
      "step": 111948
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.8709705471992493,
      "learning_rate": 0.0003133001926812401,
      "loss": 2.169,
      "step": 111949
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.904539942741394,
      "learning_rate": 0.000313296106139812,
      "loss": 2.358,
      "step": 111950
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.9576181769371033,
      "learning_rate": 0.0003132920195959121,
      "loss": 2.3153,
      "step": 111951
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.9155285954475403,
      "learning_rate": 0.0003132879330495408,
      "loss": 2.342,
      "step": 111952
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.9064698815345764,
      "learning_rate": 0.00031328384650069916,
      "loss": 2.1324,
      "step": 111953
    },
    {
      "epoch": 1.46,
      "grad_norm": 1.0681674480438232,
      "learning_rate": 0.0003132797599493878,
      "loss": 2.132,
      "step": 111954
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.8932009339332581,
      "learning_rate": 0.00031327567339560735,
      "loss": 2.3621,
      "step": 111955
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.8885774612426758,
      "learning_rate": 0.00031327158683935874,
      "loss": 2.2722,
      "step": 111956
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.8828507661819458,
      "learning_rate": 0.0003132675002806428,
      "loss": 2.0436,
      "step": 111957
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.8464640974998474,
      "learning_rate": 0.0003132634137194601,
      "loss": 1.9669,
      "step": 111958
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.8703601360321045,
      "learning_rate": 0.0003132593271558115,
      "loss": 2.0179,
      "step": 111959
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.8908804059028625,
      "learning_rate": 0.0003132552405896977,
      "loss": 2.4031,
      "step": 111960
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.8984492421150208,
      "learning_rate": 0.0003132511540211196,
      "loss": 2.2932,
      "step": 111961
    },
    {
      "epoch": 1.46,
      "grad_norm": 1.0276743173599243,
      "learning_rate": 0.00031324706745007775,
      "loss": 2.1966,
      "step": 111962
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.856810986995697,
      "learning_rate": 0.0003132429808765731,
      "loss": 2.0697,
      "step": 111963
    },
    {
      "epoch": 1.46,
      "grad_norm": 1.057919979095459,
      "learning_rate": 0.00031323889430060625,
      "loss": 2.1553,
      "step": 111964
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.892960250377655,
      "learning_rate": 0.00031323480772217806,
      "loss": 2.0621,
      "step": 111965
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.8332107663154602,
      "learning_rate": 0.0003132307211412893,
      "loss": 2.1542,
      "step": 111966
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.8792455792427063,
      "learning_rate": 0.0003132266345579407,
      "loss": 2.0907,
      "step": 111967
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.8293800950050354,
      "learning_rate": 0.000313222547972133,
      "loss": 2.0291,
      "step": 111968
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.9085581302642822,
      "learning_rate": 0.000313218461383867,
      "loss": 2.13,
      "step": 111969
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.8470827341079712,
      "learning_rate": 0.0003132143747931434,
      "loss": 2.1644,
      "step": 111970
    },
    {
      "epoch": 1.46,
      "grad_norm": 1.0157577991485596,
      "learning_rate": 0.000313210288199963,
      "loss": 1.9984,
      "step": 111971
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.9058868885040283,
      "learning_rate": 0.00031320620160432667,
      "loss": 2.1724,
      "step": 111972
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.8245731592178345,
      "learning_rate": 0.00031320211500623495,
      "loss": 2.0809,
      "step": 111973
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.9292671084403992,
      "learning_rate": 0.0003131980284056887,
      "loss": 2.148,
      "step": 111974
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.9116187691688538,
      "learning_rate": 0.00031319394180268876,
      "loss": 2.2002,
      "step": 111975
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.890105664730072,
      "learning_rate": 0.00031318985519723574,
      "loss": 2.1655,
      "step": 111976
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.893438994884491,
      "learning_rate": 0.0003131857685893305,
      "loss": 2.1381,
      "step": 111977
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.9026111960411072,
      "learning_rate": 0.00031318168197897383,
      "loss": 2.1216,
      "step": 111978
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.9065048098564148,
      "learning_rate": 0.00031317759536616635,
      "loss": 2.0448,
      "step": 111979
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.9465593695640564,
      "learning_rate": 0.00031317350875090895,
      "loss": 2.1096,
      "step": 111980
    },
    {
      "epoch": 1.46,
      "grad_norm": 1.008957028388977,
      "learning_rate": 0.0003131694221332024,
      "loss": 2.2637,
      "step": 111981
    },
    {
      "epoch": 1.46,
      "grad_norm": 1.3816328048706055,
      "learning_rate": 0.0003131653355130473,
      "loss": 2.0449,
      "step": 111982
    },
    {
      "epoch": 1.46,
      "grad_norm": 1.0409250259399414,
      "learning_rate": 0.0003131612488904446,
      "loss": 2.1805,
      "step": 111983
    },
    {
      "epoch": 1.46,
      "grad_norm": 1.049326777458191,
      "learning_rate": 0.00031315716226539497,
      "loss": 2.1836,
      "step": 111984
    },
    {
      "epoch": 1.46,
      "grad_norm": 1.0926930904388428,
      "learning_rate": 0.0003131530756378991,
      "loss": 2.0052,
      "step": 111985
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.8977101445198059,
      "learning_rate": 0.00031314898900795785,
      "loss": 2.0333,
      "step": 111986
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.9285221099853516,
      "learning_rate": 0.000313144902375572,
      "loss": 2.3049,
      "step": 111987
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.9298563599586487,
      "learning_rate": 0.0003131408157407422,
      "loss": 2.239,
      "step": 111988
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.9777763485908508,
      "learning_rate": 0.0003131367291034693,
      "loss": 2.208,
      "step": 111989
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.9944140315055847,
      "learning_rate": 0.0003131326424637541,
      "loss": 2.1357,
      "step": 111990
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.9563460350036621,
      "learning_rate": 0.00031312855582159727,
      "loss": 2.062,
      "step": 111991
    },
    {
      "epoch": 1.46,
      "grad_norm": 1.0538851022720337,
      "learning_rate": 0.0003131244691769995,
      "loss": 2.0744,
      "step": 111992
    },
    {
      "epoch": 1.46,
      "grad_norm": 1.1263035535812378,
      "learning_rate": 0.00031312038252996176,
      "loss": 2.5766,
      "step": 111993
    },
    {
      "epoch": 1.46,
      "grad_norm": 1.0794848203659058,
      "learning_rate": 0.0003131162958804846,
      "loss": 2.0336,
      "step": 111994
    },
    {
      "epoch": 1.46,
      "grad_norm": 1.2033990621566772,
      "learning_rate": 0.0003131122092285689,
      "loss": 2.3816,
      "step": 111995
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.9128081202507019,
      "learning_rate": 0.0003131081225742155,
      "loss": 2.3125,
      "step": 111996
    },
    {
      "epoch": 1.46,
      "grad_norm": 1.1792025566101074,
      "learning_rate": 0.00031310403591742493,
      "loss": 2.1337,
      "step": 111997
    },
    {
      "epoch": 1.46,
      "grad_norm": 1.3538621664047241,
      "learning_rate": 0.0003130999492581981,
      "loss": 2.0218,
      "step": 111998
    },
    {
      "epoch": 1.46,
      "grad_norm": 1.0539226531982422,
      "learning_rate": 0.0003130958625965357,
      "loss": 2.0007,
      "step": 111999
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.9318450093269348,
      "learning_rate": 0.0003130917759324386,
      "loss": 2.257,
      "step": 112000
    },
    {
      "epoch": 1.46,
      "grad_norm": 1.535184621810913,
      "learning_rate": 0.00031308768926590746,
      "loss": 2.1285,
      "step": 112001
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.9497557282447815,
      "learning_rate": 0.00031308360259694307,
      "loss": 2.2642,
      "step": 112002
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.901868999004364,
      "learning_rate": 0.0003130795159255463,
      "loss": 2.128,
      "step": 112003
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.914910078048706,
      "learning_rate": 0.0003130754292517177,
      "loss": 1.9612,
      "step": 112004
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.9851773381233215,
      "learning_rate": 0.00031307134257545807,
      "loss": 2.0222,
      "step": 112005
    },
    {
      "epoch": 1.46,
      "grad_norm": 1.058395266532898,
      "learning_rate": 0.00031306725589676834,
      "loss": 2.118,
      "step": 112006
    },
    {
      "epoch": 1.46,
      "grad_norm": 1.0170960426330566,
      "learning_rate": 0.0003130631692156491,
      "loss": 2.4354,
      "step": 112007
    },
    {
      "epoch": 1.46,
      "grad_norm": 1.1281777620315552,
      "learning_rate": 0.00031305908253210113,
      "loss": 2.0873,
      "step": 112008
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.8484817743301392,
      "learning_rate": 0.00031305499584612524,
      "loss": 2.0526,
      "step": 112009
    },
    {
      "epoch": 1.46,
      "grad_norm": 1.0519123077392578,
      "learning_rate": 0.0003130509091577223,
      "loss": 2.1684,
      "step": 112010
    },
    {
      "epoch": 1.46,
      "grad_norm": 1.3692842721939087,
      "learning_rate": 0.0003130468224668929,
      "loss": 2.1276,
      "step": 112011
    },
    {
      "epoch": 1.46,
      "grad_norm": 1.1161205768585205,
      "learning_rate": 0.00031304273577363774,
      "loss": 2.3189,
      "step": 112012
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.9928598999977112,
      "learning_rate": 0.00031303864907795785,
      "loss": 2.2153,
      "step": 112013
    },
    {
      "epoch": 1.46,
      "grad_norm": 1.0151124000549316,
      "learning_rate": 0.0003130345623798537,
      "loss": 2.2545,
      "step": 112014
    },
    {
      "epoch": 1.46,
      "grad_norm": 1.1786258220672607,
      "learning_rate": 0.0003130304756793263,
      "loss": 2.0851,
      "step": 112015
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.9290446639060974,
      "learning_rate": 0.00031302638897637617,
      "loss": 2.0258,
      "step": 112016
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.9022645354270935,
      "learning_rate": 0.0003130223022710042,
      "loss": 2.2447,
      "step": 112017
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.9024948477745056,
      "learning_rate": 0.0003130182155632112,
      "loss": 2.2146,
      "step": 112018
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.9994432926177979,
      "learning_rate": 0.00031301412885299785,
      "loss": 2.3567,
      "step": 112019
    },
    {
      "epoch": 1.46,
      "grad_norm": 1.1240694522857666,
      "learning_rate": 0.00031301004214036494,
      "loss": 2.3547,
      "step": 112020
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.8754778504371643,
      "learning_rate": 0.0003130059554253132,
      "loss": 2.0671,
      "step": 112021
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.9592156410217285,
      "learning_rate": 0.00031300186870784343,
      "loss": 2.062,
      "step": 112022
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.8397490978240967,
      "learning_rate": 0.00031299778198795624,
      "loss": 2.1145,
      "step": 112023
    },
    {
      "epoch": 1.46,
      "grad_norm": 1.2735915184020996,
      "learning_rate": 0.0003129936952656527,
      "loss": 2.1793,
      "step": 112024
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.9602108597755432,
      "learning_rate": 0.0003129896085409333,
      "loss": 2.0462,
      "step": 112025
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.8769214153289795,
      "learning_rate": 0.0003129855218137989,
      "loss": 2.0879,
      "step": 112026
    },
    {
      "epoch": 1.46,
      "grad_norm": 1.209763765335083,
      "learning_rate": 0.00031298143508425025,
      "loss": 2.167,
      "step": 112027
    },
    {
      "epoch": 1.46,
      "grad_norm": 1.3155778646469116,
      "learning_rate": 0.0003129773483522881,
      "loss": 2.3434,
      "step": 112028
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.9083578586578369,
      "learning_rate": 0.0003129732616179132,
      "loss": 2.3318,
      "step": 112029
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.9464282393455505,
      "learning_rate": 0.00031296917488112636,
      "loss": 2.0594,
      "step": 112030
    },
    {
      "epoch": 1.46,
      "grad_norm": 1.2246263027191162,
      "learning_rate": 0.0003129650881419283,
      "loss": 2.072,
      "step": 112031
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.920192301273346,
      "learning_rate": 0.0003129610014003197,
      "loss": 2.2007,
      "step": 112032
    },
    {
      "epoch": 1.46,
      "grad_norm": 1.0224274396896362,
      "learning_rate": 0.0003129569146563016,
      "loss": 2.3348,
      "step": 112033
    },
    {
      "epoch": 1.46,
      "grad_norm": 1.027937412261963,
      "learning_rate": 0.00031295282790987436,
      "loss": 2.3028,
      "step": 112034
    },
    {
      "epoch": 1.46,
      "grad_norm": 1.0834277868270874,
      "learning_rate": 0.000312948741161039,
      "loss": 2.1552,
      "step": 112035
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.9628793001174927,
      "learning_rate": 0.00031294465440979627,
      "loss": 2.1288,
      "step": 112036
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.9455361366271973,
      "learning_rate": 0.00031294056765614686,
      "loss": 2.4396,
      "step": 112037
    },
    {
      "epoch": 1.46,
      "grad_norm": 1.0421379804611206,
      "learning_rate": 0.00031293648090009155,
      "loss": 2.1293,
      "step": 112038
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.9720768332481384,
      "learning_rate": 0.00031293239414163115,
      "loss": 2.3899,
      "step": 112039
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.8467035293579102,
      "learning_rate": 0.0003129283073807663,
      "loss": 1.9716,
      "step": 112040
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.8707714676856995,
      "learning_rate": 0.00031292422061749786,
      "loss": 2.3063,
      "step": 112041
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.9775310158729553,
      "learning_rate": 0.00031292013385182666,
      "loss": 2.3198,
      "step": 112042
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.942615807056427,
      "learning_rate": 0.00031291604708375323,
      "loss": 2.2281,
      "step": 112043
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.9118410348892212,
      "learning_rate": 0.00031291196031327847,
      "loss": 2.1116,
      "step": 112044
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.8427094221115112,
      "learning_rate": 0.0003129078735404032,
      "loss": 2.0002,
      "step": 112045
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.8640340566635132,
      "learning_rate": 0.0003129037867651281,
      "loss": 1.9916,
      "step": 112046
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.9125492572784424,
      "learning_rate": 0.00031289969998745385,
      "loss": 2.0841,
      "step": 112047
    },
    {
      "epoch": 1.46,
      "grad_norm": 1.0886989831924438,
      "learning_rate": 0.00031289561320738146,
      "loss": 2.1571,
      "step": 112048
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.9183341264724731,
      "learning_rate": 0.0003128915264249114,
      "loss": 2.1151,
      "step": 112049
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.9213659167289734,
      "learning_rate": 0.0003128874396400446,
      "loss": 2.0287,
      "step": 112050
    },
    {
      "epoch": 1.46,
      "grad_norm": 1.0336049795150757,
      "learning_rate": 0.0003128833528527819,
      "loss": 2.2276,
      "step": 112051
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.9267765879631042,
      "learning_rate": 0.0003128792660631238,
      "loss": 2.0355,
      "step": 112052
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.8646042943000793,
      "learning_rate": 0.00031287517927107124,
      "loss": 2.0685,
      "step": 112053
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.8216981291770935,
      "learning_rate": 0.000312871092476625,
      "loss": 1.8653,
      "step": 112054
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.9787119030952454,
      "learning_rate": 0.0003128670056797856,
      "loss": 1.9333,
      "step": 112055
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.9759708046913147,
      "learning_rate": 0.0003128629188805541,
      "loss": 2.3512,
      "step": 112056
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.9314045310020447,
      "learning_rate": 0.0003128588320789312,
      "loss": 1.979,
      "step": 112057
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.9471423029899597,
      "learning_rate": 0.00031285474527491757,
      "loss": 2.1122,
      "step": 112058
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.9957523345947266,
      "learning_rate": 0.0003128506584685139,
      "loss": 2.162,
      "step": 112059
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.9482482671737671,
      "learning_rate": 0.0003128465716597212,
      "loss": 2.2216,
      "step": 112060
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.9563929438591003,
      "learning_rate": 0.00031284248484854,
      "loss": 2.1954,
      "step": 112061
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.9959166646003723,
      "learning_rate": 0.0003128383980349711,
      "loss": 2.1071,
      "step": 112062
    },
    {
      "epoch": 1.46,
      "grad_norm": 1.0073295831680298,
      "learning_rate": 0.0003128343112190154,
      "loss": 2.0036,
      "step": 112063
    },
    {
      "epoch": 1.46,
      "grad_norm": 1.008716106414795,
      "learning_rate": 0.00031283022440067344,
      "loss": 2.1445,
      "step": 112064
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.8605614900588989,
      "learning_rate": 0.0003128261375799462,
      "loss": 2.0367,
      "step": 112065
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.9036020040512085,
      "learning_rate": 0.00031282205075683426,
      "loss": 2.0541,
      "step": 112066
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.9351999759674072,
      "learning_rate": 0.00031281796393133857,
      "loss": 2.2325,
      "step": 112067
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.9498483538627625,
      "learning_rate": 0.00031281387710345963,
      "loss": 2.1195,
      "step": 112068
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.8880034685134888,
      "learning_rate": 0.00031280979027319844,
      "loss": 2.3132,
      "step": 112069
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.8580161929130554,
      "learning_rate": 0.0003128057034405557,
      "loss": 2.1562,
      "step": 112070
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.8274231553077698,
      "learning_rate": 0.00031280161660553207,
      "loss": 2.3022,
      "step": 112071
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.9097392559051514,
      "learning_rate": 0.00031279752976812837,
      "loss": 2.0197,
      "step": 112072
    },
    {
      "epoch": 1.46,
      "grad_norm": 1.0762603282928467,
      "learning_rate": 0.00031279344292834543,
      "loss": 2.1685,
      "step": 112073
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.8410246968269348,
      "learning_rate": 0.00031278935608618387,
      "loss": 2.1133,
      "step": 112074
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.9570212364196777,
      "learning_rate": 0.0003127852692416446,
      "loss": 2.1341,
      "step": 112075
    },
    {
      "epoch": 1.46,
      "grad_norm": 1.0092068910598755,
      "learning_rate": 0.00031278118239472827,
      "loss": 2.2786,
      "step": 112076
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.9261891841888428,
      "learning_rate": 0.0003127770955454357,
      "loss": 1.9733,
      "step": 112077
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.8607282638549805,
      "learning_rate": 0.00031277300869376757,
      "loss": 2.089,
      "step": 112078
    },
    {
      "epoch": 1.46,
      "grad_norm": 1.10112464427948,
      "learning_rate": 0.0003127689218397248,
      "loss": 2.4348,
      "step": 112079
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.9869585633277893,
      "learning_rate": 0.00031276483498330796,
      "loss": 2.1979,
      "step": 112080
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.9415050745010376,
      "learning_rate": 0.0003127607481245179,
      "loss": 2.2232,
      "step": 112081
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.9408600926399231,
      "learning_rate": 0.0003127566612633554,
      "loss": 2.1065,
      "step": 112082
    },
    {
      "epoch": 1.46,
      "grad_norm": 1.0251563787460327,
      "learning_rate": 0.00031275257439982124,
      "loss": 2.2243,
      "step": 112083
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.9679428935050964,
      "learning_rate": 0.0003127484875339161,
      "loss": 2.4355,
      "step": 112084
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.8811214566230774,
      "learning_rate": 0.00031274440066564075,
      "loss": 2.2601,
      "step": 112085
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.861926794052124,
      "learning_rate": 0.00031274031379499595,
      "loss": 2.1871,
      "step": 112086
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.9710558652877808,
      "learning_rate": 0.0003127362269219825,
      "loss": 2.2546,
      "step": 112087
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.9208887219429016,
      "learning_rate": 0.00031273214004660124,
      "loss": 2.1364,
      "step": 112088
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.9551330208778381,
      "learning_rate": 0.00031272805316885275,
      "loss": 2.1909,
      "step": 112089
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.989557147026062,
      "learning_rate": 0.0003127239662887378,
      "loss": 2.4402,
      "step": 112090
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.9559003710746765,
      "learning_rate": 0.0003127198794062574,
      "loss": 2.0475,
      "step": 112091
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.9162596464157104,
      "learning_rate": 0.000312715792521412,
      "loss": 2.2255,
      "step": 112092
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.8913456797599792,
      "learning_rate": 0.00031271170563420257,
      "loss": 2.283,
      "step": 112093
    },
    {
      "epoch": 1.46,
      "grad_norm": 1.008793592453003,
      "learning_rate": 0.0003127076187446298,
      "loss": 2.2067,
      "step": 112094
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.865956723690033,
      "learning_rate": 0.00031270353185269443,
      "loss": 2.1518,
      "step": 112095
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.91490238904953,
      "learning_rate": 0.00031269944495839716,
      "loss": 2.1133,
      "step": 112096
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.8913198113441467,
      "learning_rate": 0.0003126953580617389,
      "loss": 2.1711,
      "step": 112097
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.9877927899360657,
      "learning_rate": 0.00031269127116272035,
      "loss": 2.0301,
      "step": 112098
    },
    {
      "epoch": 1.46,
      "grad_norm": 1.0602837800979614,
      "learning_rate": 0.00031268718426134216,
      "loss": 2.1324,
      "step": 112099
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.9134092926979065,
      "learning_rate": 0.0003126830973576052,
      "loss": 2.169,
      "step": 112100
    },
    {
      "epoch": 1.46,
      "grad_norm": 1.071861982345581,
      "learning_rate": 0.0003126790104515102,
      "loss": 2.2373,
      "step": 112101
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.8707881569862366,
      "learning_rate": 0.00031267492354305804,
      "loss": 2.1701,
      "step": 112102
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.8712792992591858,
      "learning_rate": 0.00031267083663224935,
      "loss": 2.223,
      "step": 112103
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.9125061631202698,
      "learning_rate": 0.00031266674971908485,
      "loss": 2.0761,
      "step": 112104
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.9223272204399109,
      "learning_rate": 0.0003126626628035654,
      "loss": 2.153,
      "step": 112105
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.9117047190666199,
      "learning_rate": 0.0003126585758856917,
      "loss": 2.1707,
      "step": 112106
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.8801403045654297,
      "learning_rate": 0.0003126544889654645,
      "loss": 2.2185,
      "step": 112107
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.9125483632087708,
      "learning_rate": 0.00031265040204288464,
      "loss": 2.0921,
      "step": 112108
    },
    {
      "epoch": 1.46,
      "grad_norm": 1.0425255298614502,
      "learning_rate": 0.0003126463151179528,
      "loss": 2.1468,
      "step": 112109
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.9615695476531982,
      "learning_rate": 0.00031264222819066983,
      "loss": 2.0725,
      "step": 112110
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.9656820893287659,
      "learning_rate": 0.00031263814126103635,
      "loss": 2.0403,
      "step": 112111
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.9206826090812683,
      "learning_rate": 0.0003126340543290533,
      "loss": 2.0402,
      "step": 112112
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.889704167842865,
      "learning_rate": 0.00031262996739472133,
      "loss": 2.1784,
      "step": 112113
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.886381983757019,
      "learning_rate": 0.0003126258804580411,
      "loss": 2.114,
      "step": 112114
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.9069901704788208,
      "learning_rate": 0.00031262179351901364,
      "loss": 2.028,
      "step": 112115
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.9475728869438171,
      "learning_rate": 0.00031261770657763945,
      "loss": 2.1731,
      "step": 112116
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.8807297348976135,
      "learning_rate": 0.0003126136196339193,
      "loss": 2.3227,
      "step": 112117
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.9628507494926453,
      "learning_rate": 0.0003126095326878542,
      "loss": 2.3015,
      "step": 112118
    },
    {
      "epoch": 1.46,
      "grad_norm": 1.1086891889572144,
      "learning_rate": 0.0003126054457394447,
      "loss": 2.2234,
      "step": 112119
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.9296435117721558,
      "learning_rate": 0.00031260135878869154,
      "loss": 2.0014,
      "step": 112120
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.8254774212837219,
      "learning_rate": 0.00031259727183559566,
      "loss": 2.1238,
      "step": 112121
    },
    {
      "epoch": 1.46,
      "grad_norm": 1.0259242057800293,
      "learning_rate": 0.00031259318488015763,
      "loss": 2.198,
      "step": 112122
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.8314313292503357,
      "learning_rate": 0.0003125890979223783,
      "loss": 1.9418,
      "step": 112123
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.8338456153869629,
      "learning_rate": 0.00031258501096225854,
      "loss": 2.1238,
      "step": 112124
    },
    {
      "epoch": 1.46,
      "grad_norm": 1.0396184921264648,
      "learning_rate": 0.00031258092399979884,
      "loss": 2.0787,
      "step": 112125
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.9331403374671936,
      "learning_rate": 0.00031257683703500013,
      "loss": 1.998,
      "step": 112126
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.7966329455375671,
      "learning_rate": 0.00031257275006786324,
      "loss": 2.2731,
      "step": 112127
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.8694881200790405,
      "learning_rate": 0.0003125686630983887,
      "loss": 2.054,
      "step": 112128
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.9937692284584045,
      "learning_rate": 0.0003125645761265775,
      "loss": 1.9792,
      "step": 112129
    },
    {
      "epoch": 1.46,
      "grad_norm": 1.0067908763885498,
      "learning_rate": 0.00031256048915243034,
      "loss": 1.9716,
      "step": 112130
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.9378926753997803,
      "learning_rate": 0.00031255640217594784,
      "loss": 1.8448,
      "step": 112131
    },
    {
      "epoch": 1.46,
      "grad_norm": 1.0477678775787354,
      "learning_rate": 0.00031255231519713094,
      "loss": 2.1311,
      "step": 112132
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.9034093022346497,
      "learning_rate": 0.0003125482282159803,
      "loss": 2.1738,
      "step": 112133
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.9749990105628967,
      "learning_rate": 0.0003125441412324967,
      "loss": 2.106,
      "step": 112134
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.984768271446228,
      "learning_rate": 0.0003125400542466809,
      "loss": 2.4225,
      "step": 112135
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.9936115741729736,
      "learning_rate": 0.0003125359672585337,
      "loss": 2.2661,
      "step": 112136
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.8942602276802063,
      "learning_rate": 0.00031253188026805594,
      "loss": 1.6447,
      "step": 112137
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.9181809425354004,
      "learning_rate": 0.00031252779327524806,
      "loss": 2.1314,
      "step": 112138
    },
    {
      "epoch": 1.46,
      "grad_norm": 1.0258841514587402,
      "learning_rate": 0.0003125237062801111,
      "loss": 2.0254,
      "step": 112139
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.9503103494644165,
      "learning_rate": 0.0003125196192826458,
      "loss": 2.1336,
      "step": 112140
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.9373348355293274,
      "learning_rate": 0.0003125155322828528,
      "loss": 2.0075,
      "step": 112141
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.9897879362106323,
      "learning_rate": 0.00031251144528073293,
      "loss": 2.1611,
      "step": 112142
    },
    {
      "epoch": 1.46,
      "grad_norm": 1.3753538131713867,
      "learning_rate": 0.000312507358276287,
      "loss": 1.9637,
      "step": 112143
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.9362998008728027,
      "learning_rate": 0.00031250327126951575,
      "loss": 2.2927,
      "step": 112144
    },
    {
      "epoch": 1.46,
      "grad_norm": 1.1592665910720825,
      "learning_rate": 0.0003124991842604198,
      "loss": 2.1248,
      "step": 112145
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.9554829001426697,
      "learning_rate": 0.00031249509724900007,
      "loss": 2.2437,
      "step": 112146
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.8379252552986145,
      "learning_rate": 0.0003124910102352572,
      "loss": 2.2278,
      "step": 112147
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.8370186686515808,
      "learning_rate": 0.00031248692321919207,
      "loss": 2.0819,
      "step": 112148
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.9300398826599121,
      "learning_rate": 0.0003124828362008055,
      "loss": 2.1702,
      "step": 112149
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.9310785531997681,
      "learning_rate": 0.00031247874918009794,
      "loss": 2.2051,
      "step": 112150
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.9068288207054138,
      "learning_rate": 0.0003124746621570704,
      "loss": 2.397,
      "step": 112151
    },
    {
      "epoch": 1.46,
      "grad_norm": 1.0219086408615112,
      "learning_rate": 0.0003124705751317236,
      "loss": 2.1287,
      "step": 112152
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.9433868527412415,
      "learning_rate": 0.0003124664881040583,
      "loss": 2.0592,
      "step": 112153
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.976701021194458,
      "learning_rate": 0.0003124624010740752,
      "loss": 2.0733,
      "step": 112154
    },
    {
      "epoch": 1.46,
      "grad_norm": 1.162023901939392,
      "learning_rate": 0.0003124583140417752,
      "loss": 2.1951,
      "step": 112155
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.9053798317909241,
      "learning_rate": 0.000312454227007159,
      "loss": 2.0726,
      "step": 112156
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.9218760132789612,
      "learning_rate": 0.0003124501399702271,
      "loss": 2.1745,
      "step": 112157
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.8673558831214905,
      "learning_rate": 0.0003124460529309807,
      "loss": 2.0917,
      "step": 112158
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.8976852893829346,
      "learning_rate": 0.00031244196588942026,
      "loss": 2.1674,
      "step": 112159
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.9600878953933716,
      "learning_rate": 0.00031243787884554657,
      "loss": 1.966,
      "step": 112160
    },
    {
      "epoch": 1.46,
      "grad_norm": 1.0885083675384521,
      "learning_rate": 0.00031243379179936054,
      "loss": 2.0175,
      "step": 112161
    },
    {
      "epoch": 1.46,
      "grad_norm": 1.114127516746521,
      "learning_rate": 0.0003124297047508628,
      "loss": 2.2746,
      "step": 112162
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.9106482863426208,
      "learning_rate": 0.0003124256177000541,
      "loss": 2.2654,
      "step": 112163
    },
    {
      "epoch": 1.46,
      "grad_norm": 1.0096055269241333,
      "learning_rate": 0.0003124215306469353,
      "loss": 2.1784,
      "step": 112164
    },
    {
      "epoch": 1.46,
      "grad_norm": 1.135475754737854,
      "learning_rate": 0.0003124174435915071,
      "loss": 2.1787,
      "step": 112165
    },
    {
      "epoch": 1.46,
      "grad_norm": 1.0489325523376465,
      "learning_rate": 0.0003124133565337702,
      "loss": 2.0733,
      "step": 112166
    },
    {
      "epoch": 1.46,
      "grad_norm": 3.2542827129364014,
      "learning_rate": 0.00031240926947372554,
      "loss": 2.0836,
      "step": 112167
    },
    {
      "epoch": 1.46,
      "grad_norm": 1.0992244482040405,
      "learning_rate": 0.00031240518241137363,
      "loss": 2.1375,
      "step": 112168
    },
    {
      "epoch": 1.46,
      "grad_norm": 1.1727861166000366,
      "learning_rate": 0.0003124010953467154,
      "loss": 2.2204,
      "step": 112169
    },
    {
      "epoch": 1.46,
      "grad_norm": 1.0854395627975464,
      "learning_rate": 0.0003123970082797517,
      "loss": 2.184,
      "step": 112170
    },
    {
      "epoch": 1.46,
      "grad_norm": 1.0062917470932007,
      "learning_rate": 0.000312392921210483,
      "loss": 1.9151,
      "step": 112171
    },
    {
      "epoch": 1.46,
      "grad_norm": 1.2332086563110352,
      "learning_rate": 0.0003123888341389102,
      "loss": 2.0616,
      "step": 112172
    },
    {
      "epoch": 1.46,
      "grad_norm": 1.059029221534729,
      "learning_rate": 0.0003123847470650342,
      "loss": 2.1456,
      "step": 112173
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.982078492641449,
      "learning_rate": 0.00031238065998885563,
      "loss": 2.1411,
      "step": 112174
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.9694148898124695,
      "learning_rate": 0.0003123765729103752,
      "loss": 2.4234,
      "step": 112175
    },
    {
      "epoch": 1.46,
      "grad_norm": 1.0545642375946045,
      "learning_rate": 0.0003123724858295938,
      "loss": 2.1896,
      "step": 112176
    },
    {
      "epoch": 1.46,
      "grad_norm": 1.051011323928833,
      "learning_rate": 0.0003123683987465121,
      "loss": 1.9345,
      "step": 112177
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.8751069903373718,
      "learning_rate": 0.0003123643116611308,
      "loss": 2.1791,
      "step": 112178
    },
    {
      "epoch": 1.46,
      "grad_norm": 1.1506669521331787,
      "learning_rate": 0.00031236022457345084,
      "loss": 2.1517,
      "step": 112179
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.9702436327934265,
      "learning_rate": 0.00031235613748347286,
      "loss": 2.1972,
      "step": 112180
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.9143296480178833,
      "learning_rate": 0.00031235205039119757,
      "loss": 2.0969,
      "step": 112181
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.9145830273628235,
      "learning_rate": 0.0003123479632966259,
      "loss": 2.1115,
      "step": 112182
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.9359972476959229,
      "learning_rate": 0.00031234387619975845,
      "loss": 2.0724,
      "step": 112183
    },
    {
      "epoch": 1.46,
      "grad_norm": 1.003124475479126,
      "learning_rate": 0.0003123397891005961,
      "loss": 2.1087,
      "step": 112184
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.8938893675804138,
      "learning_rate": 0.00031233570199913944,
      "loss": 2.0023,
      "step": 112185
    },
    {
      "epoch": 1.46,
      "grad_norm": 1.0479416847229004,
      "learning_rate": 0.0003123316148953895,
      "loss": 2.1533,
      "step": 112186
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.998072624206543,
      "learning_rate": 0.0003123275277893468,
      "loss": 2.2835,
      "step": 112187
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.8333489894866943,
      "learning_rate": 0.0003123234406810121,
      "loss": 2.29,
      "step": 112188
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.9568095803260803,
      "learning_rate": 0.00031231935357038627,
      "loss": 2.1013,
      "step": 112189
    },
    {
      "epoch": 1.46,
      "grad_norm": 1.0214678049087524,
      "learning_rate": 0.00031231526645747015,
      "loss": 2.2271,
      "step": 112190
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.8525379300117493,
      "learning_rate": 0.00031231117934226425,
      "loss": 2.0871,
      "step": 112191
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.8422504663467407,
      "learning_rate": 0.00031230709222476955,
      "loss": 2.2259,
      "step": 112192
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.8970518708229065,
      "learning_rate": 0.00031230300510498674,
      "loss": 2.262,
      "step": 112193
    },
    {
      "epoch": 1.46,
      "grad_norm": 1.0284924507141113,
      "learning_rate": 0.00031229891798291654,
      "loss": 2.2992,
      "step": 112194
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.8530644178390503,
      "learning_rate": 0.00031229483085855976,
      "loss": 2.043,
      "step": 112195
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.8677756786346436,
      "learning_rate": 0.0003122907437319171,
      "loss": 2.1928,
      "step": 112196
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.8778573870658875,
      "learning_rate": 0.00031228665660298937,
      "loss": 2.0554,
      "step": 112197
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.8268673419952393,
      "learning_rate": 0.00031228256947177734,
      "loss": 2.2956,
      "step": 112198
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.8784633874893188,
      "learning_rate": 0.0003122784823382817,
      "loss": 2.0282,
      "step": 112199
    },
    {
      "epoch": 1.46,
      "grad_norm": 1.0828319787979126,
      "learning_rate": 0.0003122743952025033,
      "loss": 2.0579,
      "step": 112200
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.9983949065208435,
      "learning_rate": 0.0003122703080644429,
      "loss": 1.9886,
      "step": 112201
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.855469286441803,
      "learning_rate": 0.0003122662209241011,
      "loss": 1.911,
      "step": 112202
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.8770018815994263,
      "learning_rate": 0.00031226213378147884,
      "loss": 2.1834,
      "step": 112203
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.8191547989845276,
      "learning_rate": 0.0003122580466365769,
      "loss": 2.0988,
      "step": 112204
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.7679815888404846,
      "learning_rate": 0.0003122539594893959,
      "loss": 2.1066,
      "step": 112205
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.8467320799827576,
      "learning_rate": 0.0003122498723399366,
      "loss": 2.2549,
      "step": 112206
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.9078274965286255,
      "learning_rate": 0.0003122457851881999,
      "loss": 2.1044,
      "step": 112207
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.9110121130943298,
      "learning_rate": 0.00031224169803418644,
      "loss": 1.9011,
      "step": 112208
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.8794594407081604,
      "learning_rate": 0.000312237610877897,
      "loss": 2.0336,
      "step": 112209
    },
    {
      "epoch": 1.46,
      "grad_norm": 1.0230364799499512,
      "learning_rate": 0.0003122335237193324,
      "loss": 2.2262,
      "step": 112210
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.9262658357620239,
      "learning_rate": 0.0003122294365584933,
      "loss": 1.9851,
      "step": 112211
    },
    {
      "epoch": 1.46,
      "grad_norm": 1.0067431926727295,
      "learning_rate": 0.00031222534939538053,
      "loss": 2.0305,
      "step": 112212
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.8971129655838013,
      "learning_rate": 0.00031222126222999494,
      "loss": 1.9632,
      "step": 112213
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.8793829083442688,
      "learning_rate": 0.00031221717506233707,
      "loss": 2.2687,
      "step": 112214
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.9089478850364685,
      "learning_rate": 0.0003122130878924078,
      "loss": 2.0041,
      "step": 112215
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.9454542994499207,
      "learning_rate": 0.000312209000720208,
      "loss": 2.1141,
      "step": 112216
    },
    {
      "epoch": 1.46,
      "grad_norm": 1.0284757614135742,
      "learning_rate": 0.0003122049135457382,
      "loss": 2.2311,
      "step": 112217
    },
    {
      "epoch": 1.46,
      "grad_norm": 1.098265290260315,
      "learning_rate": 0.00031220082636899927,
      "loss": 2.2259,
      "step": 112218
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.9396682977676392,
      "learning_rate": 0.0003121967391899921,
      "loss": 1.9802,
      "step": 112219
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.9439416527748108,
      "learning_rate": 0.0003121926520087172,
      "loss": 2.1278,
      "step": 112220
    },
    {
      "epoch": 1.46,
      "grad_norm": 1.2115817070007324,
      "learning_rate": 0.00031218856482517543,
      "loss": 2.3373,
      "step": 112221
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.9995531439781189,
      "learning_rate": 0.0003121844776393677,
      "loss": 2.0793,
      "step": 112222
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.924536943435669,
      "learning_rate": 0.00031218039045129457,
      "loss": 2.413,
      "step": 112223
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.9273638129234314,
      "learning_rate": 0.0003121763032609569,
      "loss": 2.3072,
      "step": 112224
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.8995968103408813,
      "learning_rate": 0.0003121722160683555,
      "loss": 2.1295,
      "step": 112225
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.9970572590827942,
      "learning_rate": 0.0003121681288734909,
      "loss": 2.1324,
      "step": 112226
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.9811632037162781,
      "learning_rate": 0.0003121640416763641,
      "loss": 2.1222,
      "step": 112227
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.9324835538864136,
      "learning_rate": 0.00031215995447697576,
      "loss": 2.1285,
      "step": 112228
    },
    {
      "epoch": 1.46,
      "grad_norm": 1.0144010782241821,
      "learning_rate": 0.0003121558672753267,
      "loss": 2.2387,
      "step": 112229
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.9005773663520813,
      "learning_rate": 0.0003121517800714175,
      "loss": 2.3207,
      "step": 112230
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.90627521276474,
      "learning_rate": 0.00031214769286524926,
      "loss": 1.9905,
      "step": 112231
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.9513282775878906,
      "learning_rate": 0.00031214360565682235,
      "loss": 2.0683,
      "step": 112232
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.8903737664222717,
      "learning_rate": 0.00031213951844613776,
      "loss": 2.011,
      "step": 112233
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.8952952027320862,
      "learning_rate": 0.0003121354312331963,
      "loss": 1.9185,
      "step": 112234
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.9716135859489441,
      "learning_rate": 0.00031213134401799855,
      "loss": 1.9968,
      "step": 112235
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.9775514006614685,
      "learning_rate": 0.0003121272568005454,
      "loss": 2.3735,
      "step": 112236
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.9931094646453857,
      "learning_rate": 0.00031212316958083747,
      "loss": 2.2168,
      "step": 112237
    },
    {
      "epoch": 1.46,
      "grad_norm": 1.1215256452560425,
      "learning_rate": 0.00031211908235887566,
      "loss": 2.4661,
      "step": 112238
    },
    {
      "epoch": 1.46,
      "grad_norm": 1.003193974494934,
      "learning_rate": 0.00031211499513466076,
      "loss": 2.2583,
      "step": 112239
    },
    {
      "epoch": 1.46,
      "grad_norm": 1.002852439880371,
      "learning_rate": 0.0003121109079081934,
      "loss": 2.1957,
      "step": 112240
    },
    {
      "epoch": 1.46,
      "grad_norm": 1.1500542163848877,
      "learning_rate": 0.0003121068206794744,
      "loss": 2.0453,
      "step": 112241
    },
    {
      "epoch": 1.46,
      "grad_norm": 1.0029981136322021,
      "learning_rate": 0.00031210273344850443,
      "loss": 2.1656,
      "step": 112242
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.8419345617294312,
      "learning_rate": 0.0003120986462152845,
      "loss": 2.0256,
      "step": 112243
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.9246966242790222,
      "learning_rate": 0.0003120945589798151,
      "loss": 2.0356,
      "step": 112244
    },
    {
      "epoch": 1.46,
      "grad_norm": 1.0428118705749512,
      "learning_rate": 0.00031209047174209705,
      "loss": 2.2025,
      "step": 112245
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.9695461392402649,
      "learning_rate": 0.0003120863845021313,
      "loss": 2.1235,
      "step": 112246
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.9772906303405762,
      "learning_rate": 0.0003120822972599183,
      "loss": 2.0534,
      "step": 112247
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.9725109338760376,
      "learning_rate": 0.00031207821001545905,
      "loss": 2.1669,
      "step": 112248
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.9194541573524475,
      "learning_rate": 0.0003120741227687542,
      "loss": 1.9855,
      "step": 112249
    },
    {
      "epoch": 1.46,
      "grad_norm": 1.0090614557266235,
      "learning_rate": 0.0003120700355198046,
      "loss": 1.9373,
      "step": 112250
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.8803028464317322,
      "learning_rate": 0.0003120659482686109,
      "loss": 2.0822,
      "step": 112251
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.9144062995910645,
      "learning_rate": 0.00031206186101517396,
      "loss": 2.3012,
      "step": 112252
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.9518621563911438,
      "learning_rate": 0.00031205777375949444,
      "loss": 2.0551,
      "step": 112253
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.9002760052680969,
      "learning_rate": 0.00031205368650157316,
      "loss": 2.0414,
      "step": 112254
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.9016004800796509,
      "learning_rate": 0.00031204959924141095,
      "loss": 2.1763,
      "step": 112255
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.8872627019882202,
      "learning_rate": 0.0003120455119790084,
      "loss": 2.2004,
      "step": 112256
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.9525409936904907,
      "learning_rate": 0.0003120414247143664,
      "loss": 2.2003,
      "step": 112257
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.8643580675125122,
      "learning_rate": 0.0003120373374474856,
      "loss": 2.0493,
      "step": 112258
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.9348518252372742,
      "learning_rate": 0.000312033250178367,
      "loss": 2.1576,
      "step": 112259
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.8416397571563721,
      "learning_rate": 0.000312029162907011,
      "loss": 2.0576,
      "step": 112260
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.9081845283508301,
      "learning_rate": 0.0003120250756334187,
      "loss": 2.192,
      "step": 112261
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.8854495882987976,
      "learning_rate": 0.00031202098835759063,
      "loss": 2.0446,
      "step": 112262
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.9386550784111023,
      "learning_rate": 0.0003120169010795276,
      "loss": 2.2875,
      "step": 112263
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.9492664933204651,
      "learning_rate": 0.00031201281379923055,
      "loss": 2.1302,
      "step": 112264
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.9176470637321472,
      "learning_rate": 0.00031200872651669995,
      "loss": 2.0176,
      "step": 112265
    },
    {
      "epoch": 1.46,
      "grad_norm": 1.0029215812683105,
      "learning_rate": 0.0003120046392319367,
      "loss": 2.3507,
      "step": 112266
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.8903951048851013,
      "learning_rate": 0.0003120005519449416,
      "loss": 2.1712,
      "step": 112267
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.8863707780838013,
      "learning_rate": 0.0003119964646557155,
      "loss": 2.1577,
      "step": 112268
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.9324533939361572,
      "learning_rate": 0.00031199237736425885,
      "loss": 2.1786,
      "step": 112269
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.9626159071922302,
      "learning_rate": 0.0003119882900705726,
      "loss": 2.1352,
      "step": 112270
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.9561875462532043,
      "learning_rate": 0.0003119842027746576,
      "loss": 2.1947,
      "step": 112271
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.9004048109054565,
      "learning_rate": 0.0003119801154765145,
      "loss": 2.1106,
      "step": 112272
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.9149456024169922,
      "learning_rate": 0.000311976028176144,
      "loss": 2.1704,
      "step": 112273
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.9277322292327881,
      "learning_rate": 0.00031197194087354696,
      "loss": 2.2115,
      "step": 112274
    },
    {
      "epoch": 1.46,
      "grad_norm": 1.015293836593628,
      "learning_rate": 0.0003119678535687241,
      "loss": 2.0111,
      "step": 112275
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.8915148973464966,
      "learning_rate": 0.0003119637662616762,
      "loss": 2.02,
      "step": 112276
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.8793549537658691,
      "learning_rate": 0.00031195967895240395,
      "loss": 2.1382,
      "step": 112277
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.9736413359642029,
      "learning_rate": 0.0003119555916409083,
      "loss": 2.0831,
      "step": 112278
    },
    {
      "epoch": 1.46,
      "grad_norm": 1.0408000946044922,
      "learning_rate": 0.0003119515043271898,
      "loss": 1.8144,
      "step": 112279
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.916604220867157,
      "learning_rate": 0.00031194741701124924,
      "loss": 2.375,
      "step": 112280
    },
    {
      "epoch": 1.46,
      "grad_norm": 1.1093817949295044,
      "learning_rate": 0.0003119433296930875,
      "loss": 2.086,
      "step": 112281
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.9360291957855225,
      "learning_rate": 0.0003119392423727053,
      "loss": 2.1846,
      "step": 112282
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.9714648723602295,
      "learning_rate": 0.0003119351550501033,
      "loss": 2.232,
      "step": 112283
    },
    {
      "epoch": 1.46,
      "grad_norm": 1.0752969980239868,
      "learning_rate": 0.0003119310677252823,
      "loss": 2.0338,
      "step": 112284
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.9611430168151855,
      "learning_rate": 0.00031192698039824324,
      "loss": 2.0293,
      "step": 112285
    },
    {
      "epoch": 1.46,
      "grad_norm": 1.2108155488967896,
      "learning_rate": 0.0003119228930689866,
      "loss": 1.9981,
      "step": 112286
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.9438397884368896,
      "learning_rate": 0.0003119188057375133,
      "loss": 2.2758,
      "step": 112287
    },
    {
      "epoch": 1.46,
      "grad_norm": 1.0017578601837158,
      "learning_rate": 0.00031191471840382406,
      "loss": 2.3037,
      "step": 112288
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.8500760793685913,
      "learning_rate": 0.0003119106310679197,
      "loss": 2.2851,
      "step": 112289
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.8570589423179626,
      "learning_rate": 0.0003119065437298009,
      "loss": 1.9727,
      "step": 112290
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.9285476207733154,
      "learning_rate": 0.00031190245638946834,
      "loss": 2.2706,
      "step": 112291
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.959287166595459,
      "learning_rate": 0.0003118983690469231,
      "loss": 2.1808,
      "step": 112292
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.9421052932739258,
      "learning_rate": 0.00031189428170216556,
      "loss": 2.0309,
      "step": 112293
    },
    {
      "epoch": 1.46,
      "grad_norm": 1.024222731590271,
      "learning_rate": 0.00031189019435519676,
      "loss": 2.0266,
      "step": 112294
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.9974849224090576,
      "learning_rate": 0.00031188610700601726,
      "loss": 2.3121,
      "step": 112295
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.9664252400398254,
      "learning_rate": 0.00031188201965462795,
      "loss": 1.9675,
      "step": 112296
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.9372965693473816,
      "learning_rate": 0.00031187793230102957,
      "loss": 2.172,
      "step": 112297
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.9012712240219116,
      "learning_rate": 0.0003118738449452228,
      "loss": 2.0535,
      "step": 112298
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.8589677810668945,
      "learning_rate": 0.0003118697575872085,
      "loss": 2.2858,
      "step": 112299
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.9280828237533569,
      "learning_rate": 0.00031186567022698734,
      "loss": 2.181,
      "step": 112300
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.8748907446861267,
      "learning_rate": 0.00031186158286456025,
      "loss": 2.1474,
      "step": 112301
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.8379198908805847,
      "learning_rate": 0.00031185749549992773,
      "loss": 2.1621,
      "step": 112302
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.9492565393447876,
      "learning_rate": 0.00031185340813309076,
      "loss": 2.1874,
      "step": 112303
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.8101814389228821,
      "learning_rate": 0.00031184932076405,
      "loss": 2.3268,
      "step": 112304
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.8656026721000671,
      "learning_rate": 0.00031184523339280616,
      "loss": 2.0869,
      "step": 112305
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.9428291916847229,
      "learning_rate": 0.00031184114601936016,
      "loss": 2.3255,
      "step": 112306
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.8660512566566467,
      "learning_rate": 0.0003118370586437127,
      "loss": 2.2697,
      "step": 112307
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.9751634001731873,
      "learning_rate": 0.00031183297126586434,
      "loss": 2.1838,
      "step": 112308
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.9379531145095825,
      "learning_rate": 0.00031182888388581615,
      "loss": 1.9095,
      "step": 112309
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.9213334321975708,
      "learning_rate": 0.0003118247965035687,
      "loss": 2.3377,
      "step": 112310
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.850933313369751,
      "learning_rate": 0.0003118207091191228,
      "loss": 2.1049,
      "step": 112311
    },
    {
      "epoch": 1.46,
      "grad_norm": 1.3002070188522339,
      "learning_rate": 0.00031181662173247916,
      "loss": 2.0839,
      "step": 112312
    },
    {
      "epoch": 1.46,
      "grad_norm": 1.0468590259552002,
      "learning_rate": 0.0003118125343436387,
      "loss": 2.4439,
      "step": 112313
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.8730732202529907,
      "learning_rate": 0.000311808446952602,
      "loss": 1.9529,
      "step": 112314
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.9153599739074707,
      "learning_rate": 0.00031180435955936985,
      "loss": 2.0249,
      "step": 112315
    },
    {
      "epoch": 1.46,
      "grad_norm": 1.1897879838943481,
      "learning_rate": 0.0003118002721639431,
      "loss": 2.1199,
      "step": 112316
    },
    {
      "epoch": 1.46,
      "grad_norm": 1.0307809114456177,
      "learning_rate": 0.0003117961847663224,
      "loss": 2.0621,
      "step": 112317
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.9157593846321106,
      "learning_rate": 0.0003117920973665086,
      "loss": 2.0462,
      "step": 112318
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.9283409118652344,
      "learning_rate": 0.0003117880099645025,
      "loss": 2.1192,
      "step": 112319
    },
    {
      "epoch": 1.46,
      "grad_norm": 1.089006781578064,
      "learning_rate": 0.00031178392256030473,
      "loss": 2.1357,
      "step": 112320
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.7972033023834229,
      "learning_rate": 0.000311779835153916,
      "loss": 1.9552,
      "step": 112321
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.8859332799911499,
      "learning_rate": 0.00031177574774533736,
      "loss": 2.1766,
      "step": 112322
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.9852004051208496,
      "learning_rate": 0.0003117716603345693,
      "loss": 1.9036,
      "step": 112323
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.8179253339767456,
      "learning_rate": 0.0003117675729216126,
      "loss": 2.2334,
      "step": 112324
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.9182751774787903,
      "learning_rate": 0.0003117634855064682,
      "loss": 2.1285,
      "step": 112325
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.9445570111274719,
      "learning_rate": 0.00031175939808913674,
      "loss": 2.2463,
      "step": 112326
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.96219801902771,
      "learning_rate": 0.00031175531066961887,
      "loss": 2.2745,
      "step": 112327
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.8647465109825134,
      "learning_rate": 0.0003117512232479156,
      "loss": 2.0469,
      "step": 112328
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.8316347002983093,
      "learning_rate": 0.0003117471358240275,
      "loss": 2.2483,
      "step": 112329
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.9162797927856445,
      "learning_rate": 0.00031174304839795536,
      "loss": 2.0929,
      "step": 112330
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.8422313928604126,
      "learning_rate": 0.0003117389609697,
      "loss": 2.3816,
      "step": 112331
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.8699932098388672,
      "learning_rate": 0.0003117348735392621,
      "loss": 2.0799,
      "step": 112332
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.9707221388816833,
      "learning_rate": 0.0003117307861066425,
      "loss": 2.3193,
      "step": 112333
    },
    {
      "epoch": 1.46,
      "grad_norm": 1.050376057624817,
      "learning_rate": 0.0003117266986718419,
      "loss": 2.1277,
      "step": 112334
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.8601037859916687,
      "learning_rate": 0.00031172261123486113,
      "loss": 2.1128,
      "step": 112335
    },
    {
      "epoch": 1.46,
      "grad_norm": 1.0667428970336914,
      "learning_rate": 0.0003117185237957009,
      "loss": 2.1306,
      "step": 112336
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.9630099534988403,
      "learning_rate": 0.00031171443635436187,
      "loss": 2.2123,
      "step": 112337
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.9242914319038391,
      "learning_rate": 0.0003117103489108451,
      "loss": 2.092,
      "step": 112338
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.8523271679878235,
      "learning_rate": 0.000311706261465151,
      "loss": 1.9891,
      "step": 112339
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.9659411311149597,
      "learning_rate": 0.00031170217401728053,
      "loss": 2.2702,
      "step": 112340
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.9226682186126709,
      "learning_rate": 0.00031169808656723446,
      "loss": 2.1266,
      "step": 112341
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.8981333374977112,
      "learning_rate": 0.0003116939991150134,
      "loss": 2.0885,
      "step": 112342
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.9442036747932434,
      "learning_rate": 0.00031168991166061823,
      "loss": 2.527,
      "step": 112343
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.9130398035049438,
      "learning_rate": 0.0003116858242040497,
      "loss": 2.1263,
      "step": 112344
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.9397236704826355,
      "learning_rate": 0.0003116817367453086,
      "loss": 2.1765,
      "step": 112345
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.9662395119667053,
      "learning_rate": 0.00031167764928439557,
      "loss": 2.0503,
      "step": 112346
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.9095920324325562,
      "learning_rate": 0.00031167356182131153,
      "loss": 1.9119,
      "step": 112347
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.8461049199104309,
      "learning_rate": 0.000311669474356057,
      "loss": 2.0323,
      "step": 112348
    },
    {
      "epoch": 1.46,
      "grad_norm": 1.1161268949508667,
      "learning_rate": 0.00031166538688863303,
      "loss": 2.0092,
      "step": 112349
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.9135652184486389,
      "learning_rate": 0.00031166129941904026,
      "loss": 2.1183,
      "step": 112350
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.9756481647491455,
      "learning_rate": 0.0003116572119472794,
      "loss": 2.1341,
      "step": 112351
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.801434338092804,
      "learning_rate": 0.0003116531244733512,
      "loss": 2.1027,
      "step": 112352
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.9254400730133057,
      "learning_rate": 0.00031164903699725655,
      "loss": 2.0069,
      "step": 112353
    },
    {
      "epoch": 1.46,
      "grad_norm": 1.1804816722869873,
      "learning_rate": 0.000311644949518996,
      "loss": 2.2665,
      "step": 112354
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.9816967248916626,
      "learning_rate": 0.0003116408620385705,
      "loss": 2.0399,
      "step": 112355
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.8907611966133118,
      "learning_rate": 0.00031163677455598083,
      "loss": 2.2579,
      "step": 112356
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.9716194868087769,
      "learning_rate": 0.00031163268707122755,
      "loss": 2.0843,
      "step": 112357
    },
    {
      "epoch": 1.46,
      "grad_norm": 1.019705057144165,
      "learning_rate": 0.00031162859958431155,
      "loss": 2.3607,
      "step": 112358
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.9996320605278015,
      "learning_rate": 0.0003116245120952336,
      "loss": 2.1033,
      "step": 112359
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.8932028412818909,
      "learning_rate": 0.0003116204246039944,
      "loss": 2.0878,
      "step": 112360
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.8999335169792175,
      "learning_rate": 0.0003116163371105948,
      "loss": 2.0995,
      "step": 112361
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.8375451564788818,
      "learning_rate": 0.0003116122496150354,
      "loss": 2.0188,
      "step": 112362
    },
    {
      "epoch": 1.46,
      "grad_norm": 1.0755889415740967,
      "learning_rate": 0.0003116081621173172,
      "loss": 2.0862,
      "step": 112363
    },
    {
      "epoch": 1.46,
      "grad_norm": 1.014846920967102,
      "learning_rate": 0.0003116040746174407,
      "loss": 2.0321,
      "step": 112364
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.8741785287857056,
      "learning_rate": 0.00031159998711540685,
      "loss": 2.4173,
      "step": 112365
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.962740421295166,
      "learning_rate": 0.0003115958996112163,
      "loss": 2.1038,
      "step": 112366
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.9256994128227234,
      "learning_rate": 0.0003115918121048699,
      "loss": 2.2684,
      "step": 112367
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.8713523745536804,
      "learning_rate": 0.0003115877245963683,
      "loss": 2.2185,
      "step": 112368
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.9119295477867126,
      "learning_rate": 0.00031158363708571235,
      "loss": 2.1641,
      "step": 112369
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.9584356546401978,
      "learning_rate": 0.0003115795495729029,
      "loss": 2.135,
      "step": 112370
    },
    {
      "epoch": 1.46,
      "grad_norm": 1.0154470205307007,
      "learning_rate": 0.00031157546205794044,
      "loss": 2.245,
      "step": 112371
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.9525653123855591,
      "learning_rate": 0.00031157137454082594,
      "loss": 2.1653,
      "step": 112372
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.8979711532592773,
      "learning_rate": 0.00031156728702156006,
      "loss": 2.303,
      "step": 112373
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.8581758737564087,
      "learning_rate": 0.0003115631995001437,
      "loss": 2.0627,
      "step": 112374
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.9566367864608765,
      "learning_rate": 0.0003115591119765775,
      "loss": 2.1006,
      "step": 112375
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.8889358639717102,
      "learning_rate": 0.00031155502445086216,
      "loss": 1.9586,
      "step": 112376
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.9321042895317078,
      "learning_rate": 0.0003115509369229986,
      "loss": 2.0719,
      "step": 112377
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.8937382698059082,
      "learning_rate": 0.0003115468493929875,
      "loss": 2.2276,
      "step": 112378
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.8810217976570129,
      "learning_rate": 0.0003115427618608295,
      "loss": 2.0414,
      "step": 112379
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.9490734338760376,
      "learning_rate": 0.0003115386743265257,
      "loss": 2.1659,
      "step": 112380
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.9194326400756836,
      "learning_rate": 0.0003115345867900765,
      "loss": 1.9115,
      "step": 112381
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.8698441386222839,
      "learning_rate": 0.0003115304992514828,
      "loss": 2.1136,
      "step": 112382
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.9218051433563232,
      "learning_rate": 0.0003115264117107454,
      "loss": 1.9664,
      "step": 112383
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.8914685845375061,
      "learning_rate": 0.0003115223241678651,
      "loss": 2.2764,
      "step": 112384
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.9143990874290466,
      "learning_rate": 0.0003115182366228424,
      "loss": 2.1693,
      "step": 112385
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.964783251285553,
      "learning_rate": 0.0003115141490756784,
      "loss": 2.2575,
      "step": 112386
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.811997652053833,
      "learning_rate": 0.00031151006152637365,
      "loss": 2.0149,
      "step": 112387
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.8606321215629578,
      "learning_rate": 0.0003115059739749289,
      "loss": 2.1953,
      "step": 112388
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.7992886900901794,
      "learning_rate": 0.0003115018864213451,
      "loss": 1.9691,
      "step": 112389
    },
    {
      "epoch": 1.46,
      "grad_norm": 1.0043361186981201,
      "learning_rate": 0.00031149779886562273,
      "loss": 2.2478,
      "step": 112390
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.8698210120201111,
      "learning_rate": 0.00031149371130776283,
      "loss": 1.8994,
      "step": 112391
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.9932703971862793,
      "learning_rate": 0.00031148962374776603,
      "loss": 2.1737,
      "step": 112392
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.9925189018249512,
      "learning_rate": 0.00031148553618563297,
      "loss": 2.333,
      "step": 112393
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.9190341830253601,
      "learning_rate": 0.0003114814486213646,
      "loss": 2.2863,
      "step": 112394
    },
    {
      "epoch": 1.46,
      "grad_norm": 1.0943810939788818,
      "learning_rate": 0.0003114773610549617,
      "loss": 2.1215,
      "step": 112395
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.8834261894226074,
      "learning_rate": 0.00031147327348642485,
      "loss": 2.0303,
      "step": 112396
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.91243976354599,
      "learning_rate": 0.00031146918591575483,
      "loss": 2.1523,
      "step": 112397
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.9381563663482666,
      "learning_rate": 0.0003114650983429527,
      "loss": 2.2005,
      "step": 112398
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.8904033303260803,
      "learning_rate": 0.0003114610107680187,
      "loss": 2.0127,
      "step": 112399
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.8660594820976257,
      "learning_rate": 0.000311456923190954,
      "loss": 1.9243,
      "step": 112400
    },
    {
      "epoch": 1.46,
      "grad_norm": 1.1441770792007446,
      "learning_rate": 0.0003114528356117593,
      "loss": 2.0684,
      "step": 112401
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.8876981735229492,
      "learning_rate": 0.0003114487480304353,
      "loss": 2.1662,
      "step": 112402
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.9245006442070007,
      "learning_rate": 0.0003114446604469827,
      "loss": 2.0621,
      "step": 112403
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.9107301831245422,
      "learning_rate": 0.00031144057286140237,
      "loss": 2.2306,
      "step": 112404
    },
    {
      "epoch": 1.46,
      "grad_norm": 1.037148118019104,
      "learning_rate": 0.00031143648527369504,
      "loss": 2.3507,
      "step": 112405
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.8650484085083008,
      "learning_rate": 0.0003114323976838614,
      "loss": 2.2592,
      "step": 112406
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.8383638262748718,
      "learning_rate": 0.00031142831009190223,
      "loss": 2.2057,
      "step": 112407
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.9386125206947327,
      "learning_rate": 0.00031142422249781843,
      "loss": 2.0788,
      "step": 112408
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.8656310439109802,
      "learning_rate": 0.0003114201349016106,
      "loss": 2.0335,
      "step": 112409
    },
    {
      "epoch": 1.46,
      "grad_norm": 1.0151351690292358,
      "learning_rate": 0.0003114160473032795,
      "loss": 2.1272,
      "step": 112410
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.9291796684265137,
      "learning_rate": 0.00031141195970282604,
      "loss": 2.3127,
      "step": 112411
    },
    {
      "epoch": 1.46,
      "grad_norm": 1.0275741815567017,
      "learning_rate": 0.00031140787210025083,
      "loss": 2.1164,
      "step": 112412
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.9957408905029297,
      "learning_rate": 0.00031140378449555464,
      "loss": 2.3232,
      "step": 112413
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.8721898794174194,
      "learning_rate": 0.00031139969688873836,
      "loss": 2.0572,
      "step": 112414
    },
    {
      "epoch": 1.46,
      "grad_norm": 1.0063633918762207,
      "learning_rate": 0.00031139560927980256,
      "loss": 2.1327,
      "step": 112415
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.9491778612136841,
      "learning_rate": 0.00031139152166874813,
      "loss": 2.0575,
      "step": 112416
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.8823350667953491,
      "learning_rate": 0.0003113874340555759,
      "loss": 2.0066,
      "step": 112417
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.8988322615623474,
      "learning_rate": 0.0003113833464402864,
      "loss": 1.9917,
      "step": 112418
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.9635670185089111,
      "learning_rate": 0.0003113792588228806,
      "loss": 2.0429,
      "step": 112419
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.9568027853965759,
      "learning_rate": 0.0003113751712033592,
      "loss": 2.363,
      "step": 112420
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.880829393863678,
      "learning_rate": 0.0003113710835817229,
      "loss": 2.0358,
      "step": 112421
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.8847904801368713,
      "learning_rate": 0.00031136699595797244,
      "loss": 1.9659,
      "step": 112422
    },
    {
      "epoch": 1.46,
      "grad_norm": 1.002381443977356,
      "learning_rate": 0.0003113629083321087,
      "loss": 2.2694,
      "step": 112423
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.8703531622886658,
      "learning_rate": 0.00031135882070413234,
      "loss": 2.0302,
      "step": 112424
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.9564862251281738,
      "learning_rate": 0.0003113547330740442,
      "loss": 2.1846,
      "step": 112425
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.9717678427696228,
      "learning_rate": 0.0003113506454418451,
      "loss": 2.1445,
      "step": 112426
    },
    {
      "epoch": 1.46,
      "grad_norm": 1.0046669244766235,
      "learning_rate": 0.0003113465578075355,
      "loss": 2.306,
      "step": 112427
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.9162562489509583,
      "learning_rate": 0.0003113424701711165,
      "loss": 2.0901,
      "step": 112428
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.8911517858505249,
      "learning_rate": 0.0003113383825325887,
      "loss": 1.8443,
      "step": 112429
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.9045059084892273,
      "learning_rate": 0.0003113342948919529,
      "loss": 2.1541,
      "step": 112430
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.8888535499572754,
      "learning_rate": 0.0003113302072492097,
      "loss": 2.3214,
      "step": 112431
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.8936174511909485,
      "learning_rate": 0.00031132611960436017,
      "loss": 2.2166,
      "step": 112432
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.8925762176513672,
      "learning_rate": 0.0003113220319574049,
      "loss": 2.2532,
      "step": 112433
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.9189082384109497,
      "learning_rate": 0.0003113179443083445,
      "loss": 2.3519,
      "step": 112434
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.8976936936378479,
      "learning_rate": 0.00031131385665718,
      "loss": 2.2863,
      "step": 112435
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.8166099190711975,
      "learning_rate": 0.000311309769003912,
      "loss": 2.1123,
      "step": 112436
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.8782830238342285,
      "learning_rate": 0.0003113056813485413,
      "loss": 2.3956,
      "step": 112437
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.8705660700798035,
      "learning_rate": 0.0003113015936910687,
      "loss": 2.141,
      "step": 112438
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.8458154797554016,
      "learning_rate": 0.0003112975060314949,
      "loss": 2.2441,
      "step": 112439
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.8776338696479797,
      "learning_rate": 0.0003112934183698206,
      "loss": 1.9897,
      "step": 112440
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.9004037380218506,
      "learning_rate": 0.0003112893307060467,
      "loss": 2.3536,
      "step": 112441
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.8876702785491943,
      "learning_rate": 0.00031128524304017394,
      "loss": 2.1671,
      "step": 112442
    },
    {
      "epoch": 1.46,
      "grad_norm": 1.0081955194473267,
      "learning_rate": 0.00031128115537220295,
      "loss": 2.0257,
      "step": 112443
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.9186951518058777,
      "learning_rate": 0.00031127706770213475,
      "loss": 2.3284,
      "step": 112444
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.8731798529624939,
      "learning_rate": 0.0003112729800299698,
      "loss": 2.2951,
      "step": 112445
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.9497848749160767,
      "learning_rate": 0.000311268892355709,
      "loss": 2.0706,
      "step": 112446
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.8923618793487549,
      "learning_rate": 0.0003112648046793531,
      "loss": 2.1703,
      "step": 112447
    },
    {
      "epoch": 1.46,
      "grad_norm": 1.1302026510238647,
      "learning_rate": 0.0003112607170009028,
      "loss": 2.0317,
      "step": 112448
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.9565657377243042,
      "learning_rate": 0.00031125662932035905,
      "loss": 2.0696,
      "step": 112449
    },
    {
      "epoch": 1.46,
      "grad_norm": 1.096224308013916,
      "learning_rate": 0.0003112525416377224,
      "loss": 2.147,
      "step": 112450
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.9871995449066162,
      "learning_rate": 0.00031124845395299377,
      "loss": 2.1216,
      "step": 112451
    },
    {
      "epoch": 1.46,
      "grad_norm": 1.1090352535247803,
      "learning_rate": 0.00031124436626617377,
      "loss": 2.0375,
      "step": 112452
    },
    {
      "epoch": 1.46,
      "grad_norm": 1.0224288702011108,
      "learning_rate": 0.0003112402785772632,
      "loss": 2.1156,
      "step": 112453
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.8993604183197021,
      "learning_rate": 0.00031123619088626285,
      "loss": 2.1221,
      "step": 112454
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.9977304935455322,
      "learning_rate": 0.00031123210319317356,
      "loss": 2.1313,
      "step": 112455
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.904276430606842,
      "learning_rate": 0.00031122801549799594,
      "loss": 2.1369,
      "step": 112456
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.8958887457847595,
      "learning_rate": 0.00031122392780073084,
      "loss": 2.2985,
      "step": 112457
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.945452094078064,
      "learning_rate": 0.00031121984010137904,
      "loss": 2.2516,
      "step": 112458
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.8852919936180115,
      "learning_rate": 0.00031121575239994123,
      "loss": 2.1801,
      "step": 112459
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.8732625246047974,
      "learning_rate": 0.00031121166469641817,
      "loss": 2.1524,
      "step": 112460
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.900888979434967,
      "learning_rate": 0.00031120757699081063,
      "loss": 2.2014,
      "step": 112461
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.8999847769737244,
      "learning_rate": 0.0003112034892831195,
      "loss": 2.1282,
      "step": 112462
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.9729211926460266,
      "learning_rate": 0.0003111994015733454,
      "loss": 2.3224,
      "step": 112463
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.8345404267311096,
      "learning_rate": 0.00031119531386148905,
      "loss": 2.2519,
      "step": 112464
    },
    {
      "epoch": 1.46,
      "grad_norm": 1.0123921632766724,
      "learning_rate": 0.00031119122614755136,
      "loss": 2.1447,
      "step": 112465
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.9345925450325012,
      "learning_rate": 0.00031118713843153293,
      "loss": 2.3439,
      "step": 112466
    },
    {
      "epoch": 1.46,
      "grad_norm": 1.0388517379760742,
      "learning_rate": 0.0003111830507134346,
      "loss": 2.1628,
      "step": 112467
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.9064805507659912,
      "learning_rate": 0.0003111789629932572,
      "loss": 2.0155,
      "step": 112468
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.9094015955924988,
      "learning_rate": 0.00031117487527100146,
      "loss": 2.0994,
      "step": 112469
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.9318533539772034,
      "learning_rate": 0.00031117078754666795,
      "loss": 2.163,
      "step": 112470
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.9056282043457031,
      "learning_rate": 0.00031116669982025766,
      "loss": 1.9869,
      "step": 112471
    },
    {
      "epoch": 1.46,
      "grad_norm": 1.049699306488037,
      "learning_rate": 0.0003111626120917713,
      "loss": 2.2167,
      "step": 112472
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.925950825214386,
      "learning_rate": 0.0003111585243612096,
      "loss": 2.0211,
      "step": 112473
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.8980789184570312,
      "learning_rate": 0.0003111544366285732,
      "loss": 2.174,
      "step": 112474
    },
    {
      "epoch": 1.46,
      "grad_norm": 1.0371086597442627,
      "learning_rate": 0.00031115034889386313,
      "loss": 2.2246,
      "step": 112475
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.8633044958114624,
      "learning_rate": 0.00031114626115707995,
      "loss": 2.0477,
      "step": 112476
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.9269150495529175,
      "learning_rate": 0.0003111421734182244,
      "loss": 2.0974,
      "step": 112477
    },
    {
      "epoch": 1.46,
      "grad_norm": 1.1090341806411743,
      "learning_rate": 0.0003111380856772975,
      "loss": 2.1256,
      "step": 112478
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.9739118814468384,
      "learning_rate": 0.00031113399793429954,
      "loss": 2.1802,
      "step": 112479
    },
    {
      "epoch": 1.46,
      "grad_norm": 1.014209508895874,
      "learning_rate": 0.0003111299101892317,
      "loss": 2.1895,
      "step": 112480
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.9489915370941162,
      "learning_rate": 0.00031112582244209467,
      "loss": 2.1779,
      "step": 112481
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.9691271781921387,
      "learning_rate": 0.0003111217346928891,
      "loss": 2.1565,
      "step": 112482
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.8680397272109985,
      "learning_rate": 0.00031111764694161575,
      "loss": 1.9095,
      "step": 112483
    },
    {
      "epoch": 1.46,
      "grad_norm": 1.0742088556289673,
      "learning_rate": 0.00031111355918827545,
      "loss": 2.0111,
      "step": 112484
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.8999168276786804,
      "learning_rate": 0.0003111094714328689,
      "loss": 1.989,
      "step": 112485
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.955122709274292,
      "learning_rate": 0.00031110538367539684,
      "loss": 2.2626,
      "step": 112486
    },
    {
      "epoch": 1.46,
      "grad_norm": 1.0117697715759277,
      "learning_rate": 0.0003111012959158602,
      "loss": 2.3325,
      "step": 112487
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.8276895880699158,
      "learning_rate": 0.00031109720815425954,
      "loss": 2.0352,
      "step": 112488
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.9042321443557739,
      "learning_rate": 0.00031109312039059574,
      "loss": 2.1414,
      "step": 112489
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.9279303550720215,
      "learning_rate": 0.00031108903262486956,
      "loss": 2.1832,
      "step": 112490
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.9061643481254578,
      "learning_rate": 0.0003110849448570816,
      "loss": 2.1507,
      "step": 112491
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.9481519460678101,
      "learning_rate": 0.0003110808570872328,
      "loss": 2.0606,
      "step": 112492
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.9356971979141235,
      "learning_rate": 0.00031107676931532393,
      "loss": 2.1715,
      "step": 112493
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.8670477271080017,
      "learning_rate": 0.00031107268154135554,
      "loss": 2.0701,
      "step": 112494
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.9047734141349792,
      "learning_rate": 0.0003110685937653286,
      "loss": 2.0927,
      "step": 112495
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.8771898150444031,
      "learning_rate": 0.0003110645059872438,
      "loss": 2.3248,
      "step": 112496
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.9182472229003906,
      "learning_rate": 0.0003110604182071019,
      "loss": 1.9997,
      "step": 112497
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.8478869199752808,
      "learning_rate": 0.0003110563304249036,
      "loss": 2.0352,
      "step": 112498
    },
    {
      "epoch": 1.46,
      "grad_norm": 1.1445692777633667,
      "learning_rate": 0.0003110522426406499,
      "loss": 2.2581,
      "step": 112499
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.9414263367652893,
      "learning_rate": 0.0003110481548543412,
      "loss": 2.1102,
      "step": 112500
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.8713000416755676,
      "learning_rate": 0.0003110440670659785,
      "loss": 2.0641,
      "step": 112501
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.9801012873649597,
      "learning_rate": 0.00031103997927556254,
      "loss": 2.1244,
      "step": 112502
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.8930941224098206,
      "learning_rate": 0.0003110358914830939,
      "loss": 2.1347,
      "step": 112503
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.8920331597328186,
      "learning_rate": 0.0003110318036885736,
      "loss": 1.9003,
      "step": 112504
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.8587667346000671,
      "learning_rate": 0.00031102771589200227,
      "loss": 2.049,
      "step": 112505
    },
    {
      "epoch": 1.46,
      "grad_norm": 1.2141450643539429,
      "learning_rate": 0.0003110236280933807,
      "loss": 2.123,
      "step": 112506
    },
    {
      "epoch": 1.46,
      "grad_norm": 1.0562963485717773,
      "learning_rate": 0.00031101954029270955,
      "loss": 2.1811,
      "step": 112507
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.9106214642524719,
      "learning_rate": 0.00031101545248998975,
      "loss": 2.0985,
      "step": 112508
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.899898111820221,
      "learning_rate": 0.00031101136468522183,
      "loss": 2.2836,
      "step": 112509
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.91072678565979,
      "learning_rate": 0.0003110072768784068,
      "loss": 2.1208,
      "step": 112510
    },
    {
      "epoch": 1.46,
      "grad_norm": 1.0528759956359863,
      "learning_rate": 0.00031100318906954536,
      "loss": 2.207,
      "step": 112511
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.9720837473869324,
      "learning_rate": 0.0003109991012586381,
      "loss": 2.0796,
      "step": 112512
    },
    {
      "epoch": 1.46,
      "grad_norm": 0.9660158157348633,
      "learning_rate": 0.00031099501344568597,
      "loss": 2.2016,
      "step": 112513
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.9885163307189941,
      "learning_rate": 0.00031099092563068964,
      "loss": 2.0108,
      "step": 112514
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.9181508421897888,
      "learning_rate": 0.00031098683781364986,
      "loss": 2.2952,
      "step": 112515
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.9231328368186951,
      "learning_rate": 0.00031098274999456744,
      "loss": 2.2318,
      "step": 112516
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.9480867385864258,
      "learning_rate": 0.00031097866217344314,
      "loss": 1.9571,
      "step": 112517
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.9572477340698242,
      "learning_rate": 0.0003109745743502777,
      "loss": 2.2041,
      "step": 112518
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.9437109231948853,
      "learning_rate": 0.00031097048652507184,
      "loss": 2.4813,
      "step": 112519
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.9161948561668396,
      "learning_rate": 0.0003109663986978264,
      "loss": 2.1201,
      "step": 112520
    },
    {
      "epoch": 1.47,
      "grad_norm": 1.0001205205917358,
      "learning_rate": 0.0003109623108685421,
      "loss": 2.1605,
      "step": 112521
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.9980071187019348,
      "learning_rate": 0.0003109582230372196,
      "loss": 1.9447,
      "step": 112522
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.9627948999404907,
      "learning_rate": 0.00031095413520385994,
      "loss": 2.2299,
      "step": 112523
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.9591270089149475,
      "learning_rate": 0.0003109500473684636,
      "loss": 2.0239,
      "step": 112524
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.8940274119377136,
      "learning_rate": 0.0003109459595310314,
      "loss": 2.2108,
      "step": 112525
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.9452671408653259,
      "learning_rate": 0.0003109418716915642,
      "loss": 2.2256,
      "step": 112526
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.8659531474113464,
      "learning_rate": 0.00031093778385006267,
      "loss": 2.0862,
      "step": 112527
    },
    {
      "epoch": 1.47,
      "grad_norm": 1.0760210752487183,
      "learning_rate": 0.0003109336960065276,
      "loss": 2.2079,
      "step": 112528
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.9349290132522583,
      "learning_rate": 0.00031092960816095984,
      "loss": 2.4252,
      "step": 112529
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.8939402103424072,
      "learning_rate": 0.0003109255203133599,
      "loss": 2.1261,
      "step": 112530
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.8925543427467346,
      "learning_rate": 0.0003109214324637288,
      "loss": 2.1592,
      "step": 112531
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.9764334559440613,
      "learning_rate": 0.00031091734461206725,
      "loss": 2.3026,
      "step": 112532
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.8890115022659302,
      "learning_rate": 0.0003109132567583758,
      "loss": 2.1405,
      "step": 112533
    },
    {
      "epoch": 1.47,
      "grad_norm": 1.01268470287323,
      "learning_rate": 0.0003109091689026555,
      "loss": 2.1072,
      "step": 112534
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.9094081521034241,
      "learning_rate": 0.00031090508104490694,
      "loss": 2.3625,
      "step": 112535
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.8862072229385376,
      "learning_rate": 0.0003109009931851309,
      "loss": 2.1001,
      "step": 112536
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.8985626101493835,
      "learning_rate": 0.0003108969053233282,
      "loss": 2.0141,
      "step": 112537
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.9598677754402161,
      "learning_rate": 0.00031089281745949946,
      "loss": 2.3672,
      "step": 112538
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.9757195711135864,
      "learning_rate": 0.00031088872959364575,
      "loss": 1.9525,
      "step": 112539
    },
    {
      "epoch": 1.47,
      "grad_norm": 1.0137865543365479,
      "learning_rate": 0.0003108846417257674,
      "loss": 2.1039,
      "step": 112540
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.9517979621887207,
      "learning_rate": 0.0003108805538558655,
      "loss": 2.294,
      "step": 112541
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.8939801454544067,
      "learning_rate": 0.0003108764659839406,
      "loss": 2.157,
      "step": 112542
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.9430403709411621,
      "learning_rate": 0.0003108723781099937,
      "loss": 2.1458,
      "step": 112543
    },
    {
      "epoch": 1.47,
      "grad_norm": 1.0766501426696777,
      "learning_rate": 0.0003108682902340253,
      "loss": 2.2004,
      "step": 112544
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.987241804599762,
      "learning_rate": 0.00031086420235603633,
      "loss": 2.0326,
      "step": 112545
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.8796682953834534,
      "learning_rate": 0.0003108601144760275,
      "loss": 2.0861,
      "step": 112546
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.9229532480239868,
      "learning_rate": 0.0003108560265939996,
      "loss": 2.0312,
      "step": 112547
    },
    {
      "epoch": 1.47,
      "grad_norm": 1.206494688987732,
      "learning_rate": 0.00031085193870995337,
      "loss": 2.2624,
      "step": 112548
    },
    {
      "epoch": 1.47,
      "grad_norm": 1.0353999137878418,
      "learning_rate": 0.0003108478508238894,
      "loss": 2.1797,
      "step": 112549
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.8947567939758301,
      "learning_rate": 0.00031084376293580887,
      "loss": 2.1857,
      "step": 112550
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.9604637622833252,
      "learning_rate": 0.0003108396750457121,
      "loss": 2.2696,
      "step": 112551
    },
    {
      "epoch": 1.47,
      "grad_norm": 1.1044912338256836,
      "learning_rate": 0.0003108355871536,
      "loss": 2.3456,
      "step": 112552
    },
    {
      "epoch": 1.47,
      "grad_norm": 1.0379703044891357,
      "learning_rate": 0.0003108314992594734,
      "loss": 2.0872,
      "step": 112553
    },
    {
      "epoch": 1.47,
      "grad_norm": 1.2196056842803955,
      "learning_rate": 0.00031082741136333316,
      "loss": 2.0861,
      "step": 112554
    },
    {
      "epoch": 1.47,
      "grad_norm": 1.1540513038635254,
      "learning_rate": 0.0003108233234651798,
      "loss": 2.3358,
      "step": 112555
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.9692935943603516,
      "learning_rate": 0.00031081923556501416,
      "loss": 2.1741,
      "step": 112556
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.9294474720954895,
      "learning_rate": 0.000310815147662837,
      "loss": 1.9696,
      "step": 112557
    },
    {
      "epoch": 1.47,
      "grad_norm": 1.2231688499450684,
      "learning_rate": 0.0003108110597586492,
      "loss": 2.4571,
      "step": 112558
    },
    {
      "epoch": 1.47,
      "grad_norm": 1.0077067613601685,
      "learning_rate": 0.0003108069718524513,
      "loss": 2.2874,
      "step": 112559
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.8849127888679504,
      "learning_rate": 0.00031080288394424424,
      "loss": 2.2594,
      "step": 112560
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.911654531955719,
      "learning_rate": 0.00031079879603402867,
      "loss": 2.1546,
      "step": 112561
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.8919468522071838,
      "learning_rate": 0.0003107947081218055,
      "loss": 2.2062,
      "step": 112562
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.930213987827301,
      "learning_rate": 0.00031079062020757537,
      "loss": 2.2359,
      "step": 112563
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.8476978540420532,
      "learning_rate": 0.00031078653229133895,
      "loss": 2.1313,
      "step": 112564
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.8750764727592468,
      "learning_rate": 0.0003107824443730972,
      "loss": 2.018,
      "step": 112565
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.908748984336853,
      "learning_rate": 0.0003107783564528508,
      "loss": 2.3465,
      "step": 112566
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.826923131942749,
      "learning_rate": 0.0003107742685306004,
      "loss": 2.1486,
      "step": 112567
    },
    {
      "epoch": 1.47,
      "grad_norm": 1.0009255409240723,
      "learning_rate": 0.0003107701806063469,
      "loss": 2.3698,
      "step": 112568
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.9633489847183228,
      "learning_rate": 0.00031076609268009117,
      "loss": 2.0077,
      "step": 112569
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.9821609854698181,
      "learning_rate": 0.00031076200475183366,
      "loss": 2.1934,
      "step": 112570
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.894244909286499,
      "learning_rate": 0.0003107579168215753,
      "loss": 2.0273,
      "step": 112571
    },
    {
      "epoch": 1.47,
      "grad_norm": 1.1842316389083862,
      "learning_rate": 0.0003107538288893169,
      "loss": 2.2449,
      "step": 112572
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.9131550192832947,
      "learning_rate": 0.0003107497409550591,
      "loss": 2.0933,
      "step": 112573
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.9457207918167114,
      "learning_rate": 0.0003107456530188028,
      "loss": 2.0194,
      "step": 112574
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.9221592545509338,
      "learning_rate": 0.0003107415650805486,
      "loss": 2.1829,
      "step": 112575
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.9200632572174072,
      "learning_rate": 0.0003107374771402974,
      "loss": 2.2621,
      "step": 112576
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.9310423135757446,
      "learning_rate": 0.00031073338919804987,
      "loss": 2.2002,
      "step": 112577
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.9165187478065491,
      "learning_rate": 0.0003107293012538068,
      "loss": 2.1708,
      "step": 112578
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.9012946486473083,
      "learning_rate": 0.00031072521330756896,
      "loss": 2.1249,
      "step": 112579
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.9319027066230774,
      "learning_rate": 0.00031072112535933703,
      "loss": 2.0559,
      "step": 112580
    },
    {
      "epoch": 1.47,
      "grad_norm": 1.0385119915008545,
      "learning_rate": 0.00031071703740911195,
      "loss": 2.1942,
      "step": 112581
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.9366471171379089,
      "learning_rate": 0.00031071294945689433,
      "loss": 2.0493,
      "step": 112582
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.9453955888748169,
      "learning_rate": 0.000310708861502685,
      "loss": 2.138,
      "step": 112583
    },
    {
      "epoch": 1.47,
      "grad_norm": 1.0246471166610718,
      "learning_rate": 0.0003107047735464847,
      "loss": 2.3665,
      "step": 112584
    },
    {
      "epoch": 1.47,
      "grad_norm": 1.0520037412643433,
      "learning_rate": 0.00031070068558829405,
      "loss": 2.087,
      "step": 112585
    },
    {
      "epoch": 1.47,
      "grad_norm": 1.0826359987258911,
      "learning_rate": 0.00031069659762811405,
      "loss": 2.2682,
      "step": 112586
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.9072123169898987,
      "learning_rate": 0.0003106925096659453,
      "loss": 2.0897,
      "step": 112587
    },
    {
      "epoch": 1.47,
      "grad_norm": 1.3786029815673828,
      "learning_rate": 0.00031068842170178866,
      "loss": 2.2124,
      "step": 112588
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.9307679533958435,
      "learning_rate": 0.0003106843337356448,
      "loss": 2.2426,
      "step": 112589
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.9318485856056213,
      "learning_rate": 0.0003106802457675146,
      "loss": 2.0545,
      "step": 112590
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.9503588080406189,
      "learning_rate": 0.00031067615779739865,
      "loss": 2.2528,
      "step": 112591
    },
    {
      "epoch": 1.47,
      "grad_norm": 1.040411353111267,
      "learning_rate": 0.00031067206982529784,
      "loss": 2.0284,
      "step": 112592
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.9860014319419861,
      "learning_rate": 0.0003106679818512129,
      "loss": 2.3186,
      "step": 112593
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.9401150345802307,
      "learning_rate": 0.0003106638938751445,
      "loss": 2.2921,
      "step": 112594
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.9648726582527161,
      "learning_rate": 0.0003106598058970936,
      "loss": 2.0929,
      "step": 112595
    },
    {
      "epoch": 1.47,
      "grad_norm": 1.02937650680542,
      "learning_rate": 0.0003106557179170608,
      "loss": 2.2475,
      "step": 112596
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.8809165358543396,
      "learning_rate": 0.00031065162993504684,
      "loss": 2.2486,
      "step": 112597
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.8566297888755798,
      "learning_rate": 0.0003106475419510526,
      "loss": 1.8922,
      "step": 112598
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.9793261885643005,
      "learning_rate": 0.00031064345396507884,
      "loss": 2.1379,
      "step": 112599
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.8240804076194763,
      "learning_rate": 0.00031063936597712613,
      "loss": 2.4026,
      "step": 112600
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.8434752821922302,
      "learning_rate": 0.00031063527798719535,
      "loss": 2.012,
      "step": 112601
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.9544757604598999,
      "learning_rate": 0.00031063118999528737,
      "loss": 2.0737,
      "step": 112602
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.8997395634651184,
      "learning_rate": 0.0003106271020014028,
      "loss": 2.3149,
      "step": 112603
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.9697669744491577,
      "learning_rate": 0.0003106230140055425,
      "loss": 2.0546,
      "step": 112604
    },
    {
      "epoch": 1.47,
      "grad_norm": 1.0436526536941528,
      "learning_rate": 0.00031061892600770717,
      "loss": 2.2422,
      "step": 112605
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.9567030668258667,
      "learning_rate": 0.0003106148380078976,
      "loss": 2.1112,
      "step": 112606
    },
    {
      "epoch": 1.47,
      "grad_norm": 1.008490800857544,
      "learning_rate": 0.00031061075000611446,
      "loss": 2.2828,
      "step": 112607
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.9141491055488586,
      "learning_rate": 0.0003106066620023586,
      "loss": 2.2137,
      "step": 112608
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.9029450416564941,
      "learning_rate": 0.0003106025739966308,
      "loss": 2.0993,
      "step": 112609
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.8667300939559937,
      "learning_rate": 0.00031059848598893175,
      "loss": 1.9877,
      "step": 112610
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.924831748008728,
      "learning_rate": 0.00031059439797926226,
      "loss": 2.2559,
      "step": 112611
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.9767513275146484,
      "learning_rate": 0.0003105903099676231,
      "loss": 2.1596,
      "step": 112612
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.8761412501335144,
      "learning_rate": 0.00031058622195401497,
      "loss": 2.2704,
      "step": 112613
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.8645102381706238,
      "learning_rate": 0.00031058213393843863,
      "loss": 2.2424,
      "step": 112614
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.9355151057243347,
      "learning_rate": 0.0003105780459208949,
      "loss": 2.2479,
      "step": 112615
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.9068267941474915,
      "learning_rate": 0.0003105739579013845,
      "loss": 2.0427,
      "step": 112616
    },
    {
      "epoch": 1.47,
      "grad_norm": 1.0476293563842773,
      "learning_rate": 0.0003105698698799082,
      "loss": 2.1662,
      "step": 112617
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.9212016463279724,
      "learning_rate": 0.00031056578185646684,
      "loss": 2.0368,
      "step": 112618
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.8188623189926147,
      "learning_rate": 0.00031056169383106096,
      "loss": 2.126,
      "step": 112619
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.99989914894104,
      "learning_rate": 0.00031055760580369154,
      "loss": 2.1765,
      "step": 112620
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.8504614233970642,
      "learning_rate": 0.0003105535177743593,
      "loss": 2.2687,
      "step": 112621
    },
    {
      "epoch": 1.47,
      "grad_norm": 1.015257716178894,
      "learning_rate": 0.0003105494297430649,
      "loss": 2.2264,
      "step": 112622
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.9389276504516602,
      "learning_rate": 0.0003105453417098092,
      "loss": 2.3157,
      "step": 112623
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.8673717975616455,
      "learning_rate": 0.0003105412536745929,
      "loss": 2.1409,
      "step": 112624
    },
    {
      "epoch": 1.47,
      "grad_norm": 1.0274919271469116,
      "learning_rate": 0.00031053716563741676,
      "loss": 1.9869,
      "step": 112625
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.8499315977096558,
      "learning_rate": 0.0003105330775982816,
      "loss": 1.9894,
      "step": 112626
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.9165967702865601,
      "learning_rate": 0.00031052898955718806,
      "loss": 1.9993,
      "step": 112627
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.9414778351783752,
      "learning_rate": 0.0003105249015141371,
      "loss": 2.3122,
      "step": 112628
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.8765208721160889,
      "learning_rate": 0.00031052081346912935,
      "loss": 2.2372,
      "step": 112629
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.8918349146842957,
      "learning_rate": 0.0003105167254221655,
      "loss": 1.9376,
      "step": 112630
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.9292271137237549,
      "learning_rate": 0.00031051263737324645,
      "loss": 2.1865,
      "step": 112631
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.8711216449737549,
      "learning_rate": 0.00031050854932237294,
      "loss": 2.2172,
      "step": 112632
    },
    {
      "epoch": 1.47,
      "grad_norm": 1.0383154153823853,
      "learning_rate": 0.0003105044612695456,
      "loss": 2.2171,
      "step": 112633
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.9562004208564758,
      "learning_rate": 0.0003105003732147653,
      "loss": 2.2255,
      "step": 112634
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.8522289395332336,
      "learning_rate": 0.0003104962851580328,
      "loss": 2.2553,
      "step": 112635
    },
    {
      "epoch": 1.47,
      "grad_norm": 1.075050711631775,
      "learning_rate": 0.0003104921970993489,
      "loss": 2.1593,
      "step": 112636
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.9653001427650452,
      "learning_rate": 0.0003104881090387142,
      "loss": 2.0755,
      "step": 112637
    },
    {
      "epoch": 1.47,
      "grad_norm": 1.0897752046585083,
      "learning_rate": 0.00031048402097612966,
      "loss": 2.0529,
      "step": 112638
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.8052396774291992,
      "learning_rate": 0.00031047993291159585,
      "loss": 2.1279,
      "step": 112639
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.9000524878501892,
      "learning_rate": 0.0003104758448451137,
      "loss": 2.196,
      "step": 112640
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.9740508794784546,
      "learning_rate": 0.0003104717567766838,
      "loss": 2.2593,
      "step": 112641
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.9202233552932739,
      "learning_rate": 0.0003104676687063071,
      "loss": 2.1566,
      "step": 112642
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.9631937742233276,
      "learning_rate": 0.0003104635806339842,
      "loss": 2.188,
      "step": 112643
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.8857473134994507,
      "learning_rate": 0.00031045949255971596,
      "loss": 2.2418,
      "step": 112644
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.946183979511261,
      "learning_rate": 0.0003104554044835031,
      "loss": 2.2096,
      "step": 112645
    },
    {
      "epoch": 1.47,
      "grad_norm": 1.1289069652557373,
      "learning_rate": 0.0003104513164053464,
      "loss": 2.2969,
      "step": 112646
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.8872700333595276,
      "learning_rate": 0.00031044722832524656,
      "loss": 2.1229,
      "step": 112647
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.89182049036026,
      "learning_rate": 0.0003104431402432044,
      "loss": 2.215,
      "step": 112648
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.8777572512626648,
      "learning_rate": 0.0003104390521592207,
      "loss": 2.2436,
      "step": 112649
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.9042055010795593,
      "learning_rate": 0.00031043496407329613,
      "loss": 2.2962,
      "step": 112650
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.8491028547286987,
      "learning_rate": 0.00031043087598543154,
      "loss": 2.091,
      "step": 112651
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.8650761842727661,
      "learning_rate": 0.0003104267878956276,
      "loss": 2.1539,
      "step": 112652
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.8655890822410583,
      "learning_rate": 0.0003104226998038851,
      "loss": 2.0967,
      "step": 112653
    },
    {
      "epoch": 1.47,
      "grad_norm": 1.0198498964309692,
      "learning_rate": 0.00031041861171020497,
      "loss": 1.954,
      "step": 112654
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.9982622861862183,
      "learning_rate": 0.00031041452361458774,
      "loss": 2.3101,
      "step": 112655
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.8513236045837402,
      "learning_rate": 0.0003104104355170343,
      "loss": 1.9907,
      "step": 112656
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.8458524346351624,
      "learning_rate": 0.00031040634741754525,
      "loss": 2.1879,
      "step": 112657
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.813023567199707,
      "learning_rate": 0.00031040225931612157,
      "loss": 2.0536,
      "step": 112658
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.8997966647148132,
      "learning_rate": 0.00031039817121276383,
      "loss": 2.2893,
      "step": 112659
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.974557638168335,
      "learning_rate": 0.00031039408310747297,
      "loss": 2.0287,
      "step": 112660
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.9978553652763367,
      "learning_rate": 0.00031038999500024956,
      "loss": 2.1918,
      "step": 112661
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.9300063252449036,
      "learning_rate": 0.0003103859068910945,
      "loss": 2.1356,
      "step": 112662
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.8496986031532288,
      "learning_rate": 0.0003103818187800085,
      "loss": 2.0945,
      "step": 112663
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.9080454111099243,
      "learning_rate": 0.0003103777306669923,
      "loss": 2.2743,
      "step": 112664
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.9773096442222595,
      "learning_rate": 0.0003103736425520467,
      "loss": 2.4026,
      "step": 112665
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.9832976460456848,
      "learning_rate": 0.00031036955443517237,
      "loss": 2.2154,
      "step": 112666
    },
    {
      "epoch": 1.47,
      "grad_norm": 1.0680850744247437,
      "learning_rate": 0.00031036546631637034,
      "loss": 1.9582,
      "step": 112667
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.9395350217819214,
      "learning_rate": 0.000310361378195641,
      "loss": 2.1912,
      "step": 112668
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.9149674773216248,
      "learning_rate": 0.0003103572900729853,
      "loss": 2.062,
      "step": 112669
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.9467132687568665,
      "learning_rate": 0.0003103532019484041,
      "loss": 2.0538,
      "step": 112670
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.9156161546707153,
      "learning_rate": 0.00031034911382189794,
      "loss": 2.0532,
      "step": 112671
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.867559552192688,
      "learning_rate": 0.0003103450256934677,
      "loss": 2.2267,
      "step": 112672
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.9269401431083679,
      "learning_rate": 0.00031034093756311415,
      "loss": 2.1453,
      "step": 112673
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.8489634394645691,
      "learning_rate": 0.000310336849430838,
      "loss": 1.9077,
      "step": 112674
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.9430640339851379,
      "learning_rate": 0.00031033276129664006,
      "loss": 2.303,
      "step": 112675
    },
    {
      "epoch": 1.47,
      "grad_norm": 1.005017638206482,
      "learning_rate": 0.0003103286731605211,
      "loss": 1.967,
      "step": 112676
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.8773042559623718,
      "learning_rate": 0.00031032458502248174,
      "loss": 2.1202,
      "step": 112677
    },
    {
      "epoch": 1.47,
      "grad_norm": 1.0026345252990723,
      "learning_rate": 0.0003103204968825229,
      "loss": 2.0098,
      "step": 112678
    },
    {
      "epoch": 1.47,
      "grad_norm": 1.147726058959961,
      "learning_rate": 0.0003103164087406453,
      "loss": 2.1399,
      "step": 112679
    },
    {
      "epoch": 1.47,
      "grad_norm": 1.0060456991195679,
      "learning_rate": 0.00031031232059684967,
      "loss": 2.0574,
      "step": 112680
    },
    {
      "epoch": 1.47,
      "grad_norm": 1.2195379734039307,
      "learning_rate": 0.00031030823245113676,
      "loss": 2.1579,
      "step": 112681
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.9454050064086914,
      "learning_rate": 0.0003103041443035074,
      "loss": 2.1017,
      "step": 112682
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.9101002812385559,
      "learning_rate": 0.0003103000561539623,
      "loss": 2.0096,
      "step": 112683
    },
    {
      "epoch": 1.47,
      "grad_norm": 1.0608956813812256,
      "learning_rate": 0.00031029596800250214,
      "loss": 2.1757,
      "step": 112684
    },
    {
      "epoch": 1.47,
      "grad_norm": 1.2161978483200073,
      "learning_rate": 0.00031029187984912794,
      "loss": 2.1426,
      "step": 112685
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.9954721927642822,
      "learning_rate": 0.00031028779169384017,
      "loss": 1.9837,
      "step": 112686
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.9096905589103699,
      "learning_rate": 0.00031028370353663966,
      "loss": 2.2836,
      "step": 112687
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.9992436766624451,
      "learning_rate": 0.0003102796153775273,
      "loss": 1.8412,
      "step": 112688
    },
    {
      "epoch": 1.47,
      "grad_norm": 1.1694941520690918,
      "learning_rate": 0.00031027552721650374,
      "loss": 1.9627,
      "step": 112689
    },
    {
      "epoch": 1.47,
      "grad_norm": 1.0524331331253052,
      "learning_rate": 0.00031027143905356974,
      "loss": 2.1669,
      "step": 112690
    },
    {
      "epoch": 1.47,
      "grad_norm": 1.0469759702682495,
      "learning_rate": 0.00031026735088872615,
      "loss": 2.3305,
      "step": 112691
    },
    {
      "epoch": 1.47,
      "grad_norm": 1.1909799575805664,
      "learning_rate": 0.00031026326272197365,
      "loss": 2.2449,
      "step": 112692
    },
    {
      "epoch": 1.47,
      "grad_norm": 1.0557620525360107,
      "learning_rate": 0.0003102591745533129,
      "loss": 2.1743,
      "step": 112693
    },
    {
      "epoch": 1.47,
      "grad_norm": 1.0979712009429932,
      "learning_rate": 0.000310255086382745,
      "loss": 2.0985,
      "step": 112694
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.9191492199897766,
      "learning_rate": 0.0003102509982102703,
      "loss": 2.2829,
      "step": 112695
    },
    {
      "epoch": 1.47,
      "grad_norm": 1.299363374710083,
      "learning_rate": 0.0003102469100358897,
      "loss": 2.2,
      "step": 112696
    },
    {
      "epoch": 1.47,
      "grad_norm": 1.1954684257507324,
      "learning_rate": 0.00031024282185960415,
      "loss": 1.9957,
      "step": 112697
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.8677427768707275,
      "learning_rate": 0.00031023873368141424,
      "loss": 2.2788,
      "step": 112698
    },
    {
      "epoch": 1.47,
      "grad_norm": 1.1899138689041138,
      "learning_rate": 0.00031023464550132074,
      "loss": 2.0685,
      "step": 112699
    },
    {
      "epoch": 1.47,
      "grad_norm": 1.0745748281478882,
      "learning_rate": 0.00031023055731932447,
      "loss": 2.2318,
      "step": 112700
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.8856920003890991,
      "learning_rate": 0.00031022646913542607,
      "loss": 2.1194,
      "step": 112701
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.8865239024162292,
      "learning_rate": 0.00031022238094962636,
      "loss": 2.1028,
      "step": 112702
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.9292750358581543,
      "learning_rate": 0.0003102182927619262,
      "loss": 2.1998,
      "step": 112703
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.9609313011169434,
      "learning_rate": 0.0003102142045723262,
      "loss": 2.1201,
      "step": 112704
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.9450617432594299,
      "learning_rate": 0.00031021011638082723,
      "loss": 2.0595,
      "step": 112705
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.9704471230506897,
      "learning_rate": 0.00031020602818743003,
      "loss": 2.2373,
      "step": 112706
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.8952661752700806,
      "learning_rate": 0.00031020193999213526,
      "loss": 2.036,
      "step": 112707
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.9046981334686279,
      "learning_rate": 0.00031019785179494373,
      "loss": 2.2162,
      "step": 112708
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.8592858910560608,
      "learning_rate": 0.00031019376359585636,
      "loss": 2.0352,
      "step": 112709
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.9698823094367981,
      "learning_rate": 0.00031018967539487365,
      "loss": 2.151,
      "step": 112710
    },
    {
      "epoch": 1.47,
      "grad_norm": 1.0189824104309082,
      "learning_rate": 0.00031018558719199656,
      "loss": 2.2529,
      "step": 112711
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.8698756694793701,
      "learning_rate": 0.0003101814989872258,
      "loss": 1.9506,
      "step": 112712
    },
    {
      "epoch": 1.47,
      "grad_norm": 1.0358604192733765,
      "learning_rate": 0.00031017741078056204,
      "loss": 2.368,
      "step": 112713
    },
    {
      "epoch": 1.47,
      "grad_norm": 1.0707446336746216,
      "learning_rate": 0.0003101733225720061,
      "loss": 2.1044,
      "step": 112714
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.8382986783981323,
      "learning_rate": 0.0003101692343615588,
      "loss": 2.0793,
      "step": 112715
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.8570956587791443,
      "learning_rate": 0.00031016514614922087,
      "loss": 2.219,
      "step": 112716
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.8789724111557007,
      "learning_rate": 0.000310161057934993,
      "loss": 2.0005,
      "step": 112717
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.9144538044929504,
      "learning_rate": 0.00031015696971887596,
      "loss": 2.0893,
      "step": 112718
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.8297698497772217,
      "learning_rate": 0.0003101528815008706,
      "loss": 2.089,
      "step": 112719
    },
    {
      "epoch": 1.47,
      "grad_norm": 1.0242098569869995,
      "learning_rate": 0.0003101487932809777,
      "loss": 2.1654,
      "step": 112720
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.8310755491256714,
      "learning_rate": 0.00031014470505919776,
      "loss": 1.993,
      "step": 112721
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.8670856356620789,
      "learning_rate": 0.00031014061683553185,
      "loss": 2.1511,
      "step": 112722
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.9668546915054321,
      "learning_rate": 0.00031013652860998063,
      "loss": 1.7238,
      "step": 112723
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.9277878403663635,
      "learning_rate": 0.0003101324403825448,
      "loss": 2.1327,
      "step": 112724
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.9643619060516357,
      "learning_rate": 0.00031012835215322516,
      "loss": 2.1299,
      "step": 112725
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.8629467487335205,
      "learning_rate": 0.0003101242639220225,
      "loss": 1.8958,
      "step": 112726
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.9081279635429382,
      "learning_rate": 0.00031012017568893746,
      "loss": 1.9796,
      "step": 112727
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.9402876496315002,
      "learning_rate": 0.00031011608745397097,
      "loss": 2.1296,
      "step": 112728
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.9305182695388794,
      "learning_rate": 0.00031011199921712366,
      "loss": 2.2831,
      "step": 112729
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.9635355472564697,
      "learning_rate": 0.0003101079109783964,
      "loss": 2.0524,
      "step": 112730
    },
    {
      "epoch": 1.47,
      "grad_norm": 1.0689557790756226,
      "learning_rate": 0.00031010382273778984,
      "loss": 1.9513,
      "step": 112731
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.8591950535774231,
      "learning_rate": 0.0003100997344953048,
      "loss": 2.0747,
      "step": 112732
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.9097533226013184,
      "learning_rate": 0.000310095646250942,
      "loss": 2.0542,
      "step": 112733
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.9238080382347107,
      "learning_rate": 0.00031009155800470226,
      "loss": 2.1566,
      "step": 112734
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.9400377869606018,
      "learning_rate": 0.0003100874697565863,
      "loss": 2.1842,
      "step": 112735
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.8965766429901123,
      "learning_rate": 0.0003100833815065949,
      "loss": 2.0252,
      "step": 112736
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.9419206976890564,
      "learning_rate": 0.0003100792932547288,
      "loss": 2.1877,
      "step": 112737
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.9583519697189331,
      "learning_rate": 0.0003100752050009887,
      "loss": 1.9782,
      "step": 112738
    },
    {
      "epoch": 1.47,
      "grad_norm": 1.1749117374420166,
      "learning_rate": 0.00031007111674537555,
      "loss": 2.0969,
      "step": 112739
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.9271628856658936,
      "learning_rate": 0.00031006702848788997,
      "loss": 2.3283,
      "step": 112740
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.8997688889503479,
      "learning_rate": 0.0003100629402285327,
      "loss": 2.2307,
      "step": 112741
    },
    {
      "epoch": 1.47,
      "grad_norm": 1.2277605533599854,
      "learning_rate": 0.00031005885196730453,
      "loss": 2.2808,
      "step": 112742
    },
    {
      "epoch": 1.47,
      "grad_norm": 1.1319135427474976,
      "learning_rate": 0.0003100547637042063,
      "loss": 2.1499,
      "step": 112743
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.9280848503112793,
      "learning_rate": 0.0003100506754392386,
      "loss": 2.4216,
      "step": 112744
    },
    {
      "epoch": 1.47,
      "grad_norm": 1.0967168807983398,
      "learning_rate": 0.00031004658717240235,
      "loss": 2.1256,
      "step": 112745
    },
    {
      "epoch": 1.47,
      "grad_norm": 1.4785102605819702,
      "learning_rate": 0.0003100424989036983,
      "loss": 2.1416,
      "step": 112746
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.8942234516143799,
      "learning_rate": 0.000310038410633127,
      "loss": 1.8861,
      "step": 112747
    },
    {
      "epoch": 1.47,
      "grad_norm": 1.035825490951538,
      "learning_rate": 0.0003100343223606895,
      "loss": 2.3155,
      "step": 112748
    },
    {
      "epoch": 1.47,
      "grad_norm": 1.4010125398635864,
      "learning_rate": 0.00031003023408638644,
      "loss": 2.2548,
      "step": 112749
    },
    {
      "epoch": 1.47,
      "grad_norm": 1.1318824291229248,
      "learning_rate": 0.0003100261458102185,
      "loss": 2.1216,
      "step": 112750
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.9731065034866333,
      "learning_rate": 0.0003100220575321865,
      "loss": 2.1166,
      "step": 112751
    },
    {
      "epoch": 1.47,
      "grad_norm": 1.2782536745071411,
      "learning_rate": 0.0003100179692522913,
      "loss": 2.2662,
      "step": 112752
    },
    {
      "epoch": 1.47,
      "grad_norm": 1.054490327835083,
      "learning_rate": 0.00031001388097053354,
      "loss": 2.1401,
      "step": 112753
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.966763436794281,
      "learning_rate": 0.00031000979268691397,
      "loss": 2.0344,
      "step": 112754
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.9547420740127563,
      "learning_rate": 0.0003100057044014334,
      "loss": 2.135,
      "step": 112755
    },
    {
      "epoch": 1.47,
      "grad_norm": 1.7741403579711914,
      "learning_rate": 0.00031000161611409265,
      "loss": 2.2716,
      "step": 112756
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.9467616081237793,
      "learning_rate": 0.0003099975278248923,
      "loss": 2.1664,
      "step": 112757
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.8908282518386841,
      "learning_rate": 0.00030999343953383326,
      "loss": 2.1757,
      "step": 112758
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.8942708969116211,
      "learning_rate": 0.0003099893512409163,
      "loss": 2.1616,
      "step": 112759
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.8988403677940369,
      "learning_rate": 0.00030998526294614205,
      "loss": 2.0979,
      "step": 112760
    },
    {
      "epoch": 1.47,
      "grad_norm": 1.0857086181640625,
      "learning_rate": 0.00030998117464951146,
      "loss": 2.1428,
      "step": 112761
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.996419370174408,
      "learning_rate": 0.00030997708635102503,
      "loss": 1.9374,
      "step": 112762
    },
    {
      "epoch": 1.47,
      "grad_norm": 1.0954787731170654,
      "learning_rate": 0.0003099729980506838,
      "loss": 2.1117,
      "step": 112763
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.8898552060127258,
      "learning_rate": 0.0003099689097484884,
      "loss": 2.048,
      "step": 112764
    },
    {
      "epoch": 1.47,
      "grad_norm": 1.0748271942138672,
      "learning_rate": 0.0003099648214444395,
      "loss": 2.1959,
      "step": 112765
    },
    {
      "epoch": 1.47,
      "grad_norm": 1.0938493013381958,
      "learning_rate": 0.000309960733138538,
      "loss": 2.0509,
      "step": 112766
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.8940030336380005,
      "learning_rate": 0.0003099566448307847,
      "loss": 1.9743,
      "step": 112767
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.9127424955368042,
      "learning_rate": 0.00030995255652118016,
      "loss": 2.2003,
      "step": 112768
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.8913534283638,
      "learning_rate": 0.00030994846820972527,
      "loss": 2.0927,
      "step": 112769
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.8354601860046387,
      "learning_rate": 0.0003099443798964208,
      "loss": 1.9531,
      "step": 112770
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.919417679309845,
      "learning_rate": 0.0003099402915812674,
      "loss": 2.225,
      "step": 112771
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.8568242192268372,
      "learning_rate": 0.000309936203264266,
      "loss": 2.0291,
      "step": 112772
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.9098027348518372,
      "learning_rate": 0.0003099321149454173,
      "loss": 2.1195,
      "step": 112773
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.9158704280853271,
      "learning_rate": 0.00030992802662472196,
      "loss": 2.0715,
      "step": 112774
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.9034525752067566,
      "learning_rate": 0.0003099239383021808,
      "loss": 2.2122,
      "step": 112775
    },
    {
      "epoch": 1.47,
      "grad_norm": 1.0069639682769775,
      "learning_rate": 0.0003099198499777946,
      "loss": 2.2049,
      "step": 112776
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.8839358687400818,
      "learning_rate": 0.0003099157616515642,
      "loss": 2.2291,
      "step": 112777
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.9244740009307861,
      "learning_rate": 0.0003099116733234902,
      "loss": 2.2139,
      "step": 112778
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.9408257007598877,
      "learning_rate": 0.00030990758499357344,
      "loss": 2.2368,
      "step": 112779
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.9347588419914246,
      "learning_rate": 0.00030990349666181464,
      "loss": 2.3514,
      "step": 112780
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.937977135181427,
      "learning_rate": 0.00030989940832821465,
      "loss": 2.0613,
      "step": 112781
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.9190055727958679,
      "learning_rate": 0.00030989531999277417,
      "loss": 2.0626,
      "step": 112782
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.9586805701255798,
      "learning_rate": 0.00030989123165549396,
      "loss": 2.0943,
      "step": 112783
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.8772472143173218,
      "learning_rate": 0.0003098871433163747,
      "loss": 2.1483,
      "step": 112784
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.9948129057884216,
      "learning_rate": 0.0003098830549754174,
      "loss": 2.134,
      "step": 112785
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.9109917879104614,
      "learning_rate": 0.0003098789666326225,
      "loss": 1.9226,
      "step": 112786
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.9038200974464417,
      "learning_rate": 0.0003098748782879909,
      "loss": 2.0613,
      "step": 112787
    },
    {
      "epoch": 1.47,
      "grad_norm": 1.0211114883422852,
      "learning_rate": 0.00030987078994152356,
      "loss": 1.988,
      "step": 112788
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.9852253198623657,
      "learning_rate": 0.0003098667015932209,
      "loss": 1.9978,
      "step": 112789
    },
    {
      "epoch": 1.47,
      "grad_norm": 1.0311261415481567,
      "learning_rate": 0.0003098626132430838,
      "loss": 2.1385,
      "step": 112790
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.99863201379776,
      "learning_rate": 0.0003098585248911132,
      "loss": 2.189,
      "step": 112791
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.9609931111335754,
      "learning_rate": 0.0003098544365373096,
      "loss": 2.1633,
      "step": 112792
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.9088261127471924,
      "learning_rate": 0.0003098503481816739,
      "loss": 2.1609,
      "step": 112793
    },
    {
      "epoch": 1.47,
      "grad_norm": 1.0137850046157837,
      "learning_rate": 0.00030984625982420687,
      "loss": 2.2579,
      "step": 112794
    },
    {
      "epoch": 1.47,
      "grad_norm": 1.0018913745880127,
      "learning_rate": 0.0003098421714649092,
      "loss": 2.2583,
      "step": 112795
    },
    {
      "epoch": 1.47,
      "grad_norm": 1.092732548713684,
      "learning_rate": 0.00030983808310378167,
      "loss": 2.2661,
      "step": 112796
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.9342421293258667,
      "learning_rate": 0.00030983399474082517,
      "loss": 2.1186,
      "step": 112797
    },
    {
      "epoch": 1.47,
      "grad_norm": 1.0070555210113525,
      "learning_rate": 0.0003098299063760402,
      "loss": 2.2354,
      "step": 112798
    },
    {
      "epoch": 1.47,
      "grad_norm": 1.0119117498397827,
      "learning_rate": 0.00030982581800942773,
      "loss": 2.2134,
      "step": 112799
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.9618287086486816,
      "learning_rate": 0.0003098217296409884,
      "loss": 2.3027,
      "step": 112800
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.908382773399353,
      "learning_rate": 0.00030981764127072315,
      "loss": 2.0265,
      "step": 112801
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.8871266841888428,
      "learning_rate": 0.0003098135528986325,
      "loss": 2.0967,
      "step": 112802
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.8957606554031372,
      "learning_rate": 0.00030980946452471733,
      "loss": 2.3464,
      "step": 112803
    },
    {
      "epoch": 1.47,
      "grad_norm": 1.149284839630127,
      "learning_rate": 0.00030980537614897847,
      "loss": 1.9832,
      "step": 112804
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.8757627010345459,
      "learning_rate": 0.0003098012877714165,
      "loss": 2.071,
      "step": 112805
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.8990955948829651,
      "learning_rate": 0.00030979719939203235,
      "loss": 2.1334,
      "step": 112806
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.9652180075645447,
      "learning_rate": 0.0003097931110108267,
      "loss": 2.2666,
      "step": 112807
    },
    {
      "epoch": 1.47,
      "grad_norm": 1.113197922706604,
      "learning_rate": 0.0003097890226278004,
      "loss": 2.2252,
      "step": 112808
    },
    {
      "epoch": 1.47,
      "grad_norm": 1.030920386314392,
      "learning_rate": 0.00030978493424295406,
      "loss": 2.2452,
      "step": 112809
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.9860137701034546,
      "learning_rate": 0.0003097808458562885,
      "loss": 2.1228,
      "step": 112810
    },
    {
      "epoch": 1.47,
      "grad_norm": 1.069473385810852,
      "learning_rate": 0.00030977675746780457,
      "loss": 2.2168,
      "step": 112811
    },
    {
      "epoch": 1.47,
      "grad_norm": 1.0054389238357544,
      "learning_rate": 0.0003097726690775029,
      "loss": 2.1722,
      "step": 112812
    },
    {
      "epoch": 1.47,
      "grad_norm": 1.094843864440918,
      "learning_rate": 0.00030976858068538425,
      "loss": 2.3,
      "step": 112813
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.9461420774459839,
      "learning_rate": 0.00030976449229144955,
      "loss": 2.1715,
      "step": 112814
    },
    {
      "epoch": 1.47,
      "grad_norm": 1.002505898475647,
      "learning_rate": 0.0003097604038956994,
      "loss": 2.2804,
      "step": 112815
    },
    {
      "epoch": 1.47,
      "grad_norm": 1.2331140041351318,
      "learning_rate": 0.0003097563154981346,
      "loss": 2.1661,
      "step": 112816
    },
    {
      "epoch": 1.47,
      "grad_norm": 1.0256305932998657,
      "learning_rate": 0.0003097522270987559,
      "loss": 2.3271,
      "step": 112817
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.8818193674087524,
      "learning_rate": 0.0003097481386975641,
      "loss": 2.2075,
      "step": 112818
    },
    {
      "epoch": 1.47,
      "grad_norm": 1.269140601158142,
      "learning_rate": 0.00030974405029456,
      "loss": 2.2469,
      "step": 112819
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.990407407283783,
      "learning_rate": 0.0003097399618897442,
      "loss": 2.3503,
      "step": 112820
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.9489749670028687,
      "learning_rate": 0.00030973587348311754,
      "loss": 2.1474,
      "step": 112821
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.9348942637443542,
      "learning_rate": 0.00030973178507468093,
      "loss": 2.0264,
      "step": 112822
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.9738097190856934,
      "learning_rate": 0.0003097276966644349,
      "loss": 2.1461,
      "step": 112823
    },
    {
      "epoch": 1.47,
      "grad_norm": 1.1946539878845215,
      "learning_rate": 0.0003097236082523803,
      "loss": 2.0409,
      "step": 112824
    },
    {
      "epoch": 1.47,
      "grad_norm": 1.1503376960754395,
      "learning_rate": 0.0003097195198385179,
      "loss": 2.2103,
      "step": 112825
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.9875860214233398,
      "learning_rate": 0.0003097154314228485,
      "loss": 2.1925,
      "step": 112826
    },
    {
      "epoch": 1.47,
      "grad_norm": 1.0452920198440552,
      "learning_rate": 0.0003097113430053728,
      "loss": 2.028,
      "step": 112827
    },
    {
      "epoch": 1.47,
      "grad_norm": 1.2633999586105347,
      "learning_rate": 0.0003097072545860916,
      "loss": 2.4321,
      "step": 112828
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.8598372936248779,
      "learning_rate": 0.0003097031661650056,
      "loss": 2.0847,
      "step": 112829
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.8929094672203064,
      "learning_rate": 0.0003096990777421156,
      "loss": 2.312,
      "step": 112830
    },
    {
      "epoch": 1.47,
      "grad_norm": 1.0666519403457642,
      "learning_rate": 0.0003096949893174224,
      "loss": 2.2143,
      "step": 112831
    },
    {
      "epoch": 1.47,
      "grad_norm": 1.1458089351654053,
      "learning_rate": 0.0003096909008909267,
      "loss": 2.188,
      "step": 112832
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.83705073595047,
      "learning_rate": 0.0003096868124626292,
      "loss": 2.0258,
      "step": 112833
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.886040985584259,
      "learning_rate": 0.0003096827240325309,
      "loss": 2.0166,
      "step": 112834
    },
    {
      "epoch": 1.47,
      "grad_norm": 1.044468641281128,
      "learning_rate": 0.00030967863560063226,
      "loss": 2.1701,
      "step": 112835
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.8755154609680176,
      "learning_rate": 0.00030967454716693423,
      "loss": 2.3406,
      "step": 112836
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.9138383269309998,
      "learning_rate": 0.00030967045873143756,
      "loss": 2.0748,
      "step": 112837
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.9858465194702148,
      "learning_rate": 0.0003096663702941429,
      "loss": 2.0027,
      "step": 112838
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.987221896648407,
      "learning_rate": 0.00030966228185505106,
      "loss": 2.2927,
      "step": 112839
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.917424738407135,
      "learning_rate": 0.0003096581934141629,
      "loss": 2.1424,
      "step": 112840
    },
    {
      "epoch": 1.47,
      "grad_norm": 1.0376118421554565,
      "learning_rate": 0.000309654104971479,
      "loss": 2.2223,
      "step": 112841
    },
    {
      "epoch": 1.47,
      "grad_norm": 1.041051983833313,
      "learning_rate": 0.00030965001652700033,
      "loss": 2.0093,
      "step": 112842
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.9717346429824829,
      "learning_rate": 0.00030964592808072755,
      "loss": 2.3363,
      "step": 112843
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.8718845248222351,
      "learning_rate": 0.0003096418396326613,
      "loss": 2.1704,
      "step": 112844
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.8742749691009521,
      "learning_rate": 0.0003096377511828025,
      "loss": 1.9706,
      "step": 112845
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.8487178683280945,
      "learning_rate": 0.0003096336627311519,
      "loss": 2.0415,
      "step": 112846
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.9523836374282837,
      "learning_rate": 0.00030962957427771014,
      "loss": 2.1808,
      "step": 112847
    },
    {
      "epoch": 1.47,
      "grad_norm": 1.0176351070404053,
      "learning_rate": 0.0003096254858224781,
      "loss": 1.9924,
      "step": 112848
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.9955006837844849,
      "learning_rate": 0.0003096213973654565,
      "loss": 2.1589,
      "step": 112849
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.9754354357719421,
      "learning_rate": 0.0003096173089066462,
      "loss": 2.1557,
      "step": 112850
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.8834550380706787,
      "learning_rate": 0.0003096132204460477,
      "loss": 2.0486,
      "step": 112851
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.932609498500824,
      "learning_rate": 0.00030960913198366203,
      "loss": 2.3662,
      "step": 112852
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.8813879489898682,
      "learning_rate": 0.0003096050435194898,
      "loss": 2.1756,
      "step": 112853
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.9208500385284424,
      "learning_rate": 0.0003096009550535318,
      "loss": 2.2713,
      "step": 112854
    },
    {
      "epoch": 1.47,
      "grad_norm": 1.0296566486358643,
      "learning_rate": 0.00030959686658578884,
      "loss": 2.2974,
      "step": 112855
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.9625287055969238,
      "learning_rate": 0.00030959277811626155,
      "loss": 2.0563,
      "step": 112856
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.8976179361343384,
      "learning_rate": 0.00030958868964495076,
      "loss": 2.1273,
      "step": 112857
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.9528452157974243,
      "learning_rate": 0.0003095846011718574,
      "loss": 2.2154,
      "step": 112858
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.8996462821960449,
      "learning_rate": 0.00030958051269698204,
      "loss": 2.1827,
      "step": 112859
    },
    {
      "epoch": 1.47,
      "grad_norm": 1.023989200592041,
      "learning_rate": 0.0003095764242203254,
      "loss": 1.99,
      "step": 112860
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.8839545249938965,
      "learning_rate": 0.00030957233574188843,
      "loss": 2.1365,
      "step": 112861
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.9398645758628845,
      "learning_rate": 0.0003095682472616717,
      "loss": 2.3404,
      "step": 112862
    },
    {
      "epoch": 1.47,
      "grad_norm": 1.0063799619674683,
      "learning_rate": 0.00030956415877967606,
      "loss": 2.1469,
      "step": 112863
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.9862843155860901,
      "learning_rate": 0.00030956007029590236,
      "loss": 2.3034,
      "step": 112864
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.846407949924469,
      "learning_rate": 0.00030955598181035116,
      "loss": 1.9434,
      "step": 112865
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.9598526954650879,
      "learning_rate": 0.00030955189332302333,
      "loss": 2.3663,
      "step": 112866
    },
    {
      "epoch": 1.47,
      "grad_norm": 1.0244613885879517,
      "learning_rate": 0.0003095478048339196,
      "loss": 2.2587,
      "step": 112867
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.8702866435050964,
      "learning_rate": 0.0003095437163430409,
      "loss": 2.1349,
      "step": 112868
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.8384965062141418,
      "learning_rate": 0.0003095396278503877,
      "loss": 2.2457,
      "step": 112869
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.9601234197616577,
      "learning_rate": 0.00030953553935596093,
      "loss": 2.1116,
      "step": 112870
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.9651694297790527,
      "learning_rate": 0.0003095314508597613,
      "loss": 2.1644,
      "step": 112871
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.9973169565200806,
      "learning_rate": 0.0003095273623617897,
      "loss": 2.1567,
      "step": 112872
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.9776278734207153,
      "learning_rate": 0.00030952327386204667,
      "loss": 2.2972,
      "step": 112873
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.8348185420036316,
      "learning_rate": 0.0003095191853605332,
      "loss": 2.305,
      "step": 112874
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.9448744654655457,
      "learning_rate": 0.00030951509685724977,
      "loss": 2.0997,
      "step": 112875
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.9156457781791687,
      "learning_rate": 0.0003095110083521974,
      "loss": 2.1654,
      "step": 112876
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.9832189083099365,
      "learning_rate": 0.00030950691984537676,
      "loss": 2.1271,
      "step": 112877
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.9415836334228516,
      "learning_rate": 0.0003095028313367886,
      "loss": 2.1098,
      "step": 112878
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.9699952006340027,
      "learning_rate": 0.0003094987428264337,
      "loss": 2.3033,
      "step": 112879
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.925895631313324,
      "learning_rate": 0.00030949465431431275,
      "loss": 2.27,
      "step": 112880
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.9240096807479858,
      "learning_rate": 0.0003094905658004266,
      "loss": 2.0582,
      "step": 112881
    },
    {
      "epoch": 1.47,
      "grad_norm": 1.403473973274231,
      "learning_rate": 0.0003094864772847759,
      "loss": 1.9658,
      "step": 112882
    },
    {
      "epoch": 1.47,
      "grad_norm": 1.2766438722610474,
      "learning_rate": 0.00030948238876736165,
      "loss": 2.3927,
      "step": 112883
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.8903120160102844,
      "learning_rate": 0.0003094783002481843,
      "loss": 2.3815,
      "step": 112884
    },
    {
      "epoch": 1.47,
      "grad_norm": 1.1286596059799194,
      "learning_rate": 0.0003094742117272448,
      "loss": 2.1293,
      "step": 112885
    },
    {
      "epoch": 1.47,
      "grad_norm": 1.197053074836731,
      "learning_rate": 0.0003094701232045439,
      "loss": 2.2638,
      "step": 112886
    },
    {
      "epoch": 1.47,
      "grad_norm": 1.3874176740646362,
      "learning_rate": 0.0003094660346800823,
      "loss": 2.1396,
      "step": 112887
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.9045766592025757,
      "learning_rate": 0.00030946194615386074,
      "loss": 2.2311,
      "step": 112888
    },
    {
      "epoch": 1.47,
      "grad_norm": 1.0559108257293701,
      "learning_rate": 0.0003094578576258801,
      "loss": 2.1606,
      "step": 112889
    },
    {
      "epoch": 1.47,
      "grad_norm": 1.089558720588684,
      "learning_rate": 0.00030945376909614106,
      "loss": 2.1826,
      "step": 112890
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.9476606249809265,
      "learning_rate": 0.0003094496805646443,
      "loss": 2.3115,
      "step": 112891
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.8053348660469055,
      "learning_rate": 0.00030944559203139075,
      "loss": 1.9819,
      "step": 112892
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.8773341774940491,
      "learning_rate": 0.00030944150349638107,
      "loss": 2.3006,
      "step": 112893
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.923007071018219,
      "learning_rate": 0.00030943741495961596,
      "loss": 2.4074,
      "step": 112894
    },
    {
      "epoch": 1.47,
      "grad_norm": 1.1814945936203003,
      "learning_rate": 0.0003094333264210964,
      "loss": 2.2159,
      "step": 112895
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.8992747664451599,
      "learning_rate": 0.0003094292378808229,
      "loss": 2.2395,
      "step": 112896
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.9376285076141357,
      "learning_rate": 0.0003094251493387964,
      "loss": 2.194,
      "step": 112897
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.9537972211837769,
      "learning_rate": 0.0003094210607950175,
      "loss": 2.2093,
      "step": 112898
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.9723436236381531,
      "learning_rate": 0.00030941697224948705,
      "loss": 2.1073,
      "step": 112899
    },
    {
      "epoch": 1.47,
      "grad_norm": 1.0329958200454712,
      "learning_rate": 0.0003094128837022059,
      "loss": 2.396,
      "step": 112900
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.9622696042060852,
      "learning_rate": 0.00030940879515317464,
      "loss": 2.1633,
      "step": 112901
    },
    {
      "epoch": 1.47,
      "grad_norm": 1.3185063600540161,
      "learning_rate": 0.0003094047066023941,
      "loss": 2.3887,
      "step": 112902
    },
    {
      "epoch": 1.47,
      "grad_norm": 1.1791527271270752,
      "learning_rate": 0.00030940061804986507,
      "loss": 2.044,
      "step": 112903
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.9051012992858887,
      "learning_rate": 0.0003093965294955883,
      "loss": 2.1165,
      "step": 112904
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.9061294198036194,
      "learning_rate": 0.00030939244093956456,
      "loss": 2.1526,
      "step": 112905
    },
    {
      "epoch": 1.47,
      "grad_norm": 1.0533032417297363,
      "learning_rate": 0.00030938835238179455,
      "loss": 2.0811,
      "step": 112906
    },
    {
      "epoch": 1.47,
      "grad_norm": 1.1524745225906372,
      "learning_rate": 0.0003093842638222791,
      "loss": 2.2496,
      "step": 112907
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.9175103902816772,
      "learning_rate": 0.0003093801752610189,
      "loss": 2.1739,
      "step": 112908
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.9816027283668518,
      "learning_rate": 0.00030937608669801475,
      "loss": 2.0883,
      "step": 112909
    },
    {
      "epoch": 1.47,
      "grad_norm": 1.1026016473770142,
      "learning_rate": 0.00030937199813326746,
      "loss": 2.1188,
      "step": 112910
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.8869240880012512,
      "learning_rate": 0.0003093679095667777,
      "loss": 2.1269,
      "step": 112911
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.9396331906318665,
      "learning_rate": 0.0003093638209985462,
      "loss": 1.988,
      "step": 112912
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.9166551232337952,
      "learning_rate": 0.00030935973242857394,
      "loss": 2.0975,
      "step": 112913
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.9157217741012573,
      "learning_rate": 0.0003093556438568614,
      "loss": 2.1072,
      "step": 112914
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.93937087059021,
      "learning_rate": 0.00030935155528340956,
      "loss": 2.2819,
      "step": 112915
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.9197066426277161,
      "learning_rate": 0.0003093474667082191,
      "loss": 2.0117,
      "step": 112916
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.8919340968132019,
      "learning_rate": 0.00030934337813129067,
      "loss": 2.2895,
      "step": 112917
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.87619549036026,
      "learning_rate": 0.00030933928955262515,
      "loss": 2.1676,
      "step": 112918
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.9141746759414673,
      "learning_rate": 0.0003093352009722233,
      "loss": 2.2813,
      "step": 112919
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.9554963111877441,
      "learning_rate": 0.0003093311123900859,
      "loss": 2.009,
      "step": 112920
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.9761675596237183,
      "learning_rate": 0.0003093270238062136,
      "loss": 2.1433,
      "step": 112921
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.9016351699829102,
      "learning_rate": 0.00030932293522060734,
      "loss": 2.3271,
      "step": 112922
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.897196888923645,
      "learning_rate": 0.0003093188466332676,
      "loss": 1.9708,
      "step": 112923
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.9288721084594727,
      "learning_rate": 0.00030931475804419544,
      "loss": 2.2213,
      "step": 112924
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.9550402164459229,
      "learning_rate": 0.00030931066945339146,
      "loss": 2.2706,
      "step": 112925
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.8382636308670044,
      "learning_rate": 0.00030930658086085645,
      "loss": 1.9687,
      "step": 112926
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.9560075998306274,
      "learning_rate": 0.00030930249226659117,
      "loss": 2.0268,
      "step": 112927
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.9873979091644287,
      "learning_rate": 0.0003092984036705964,
      "loss": 2.0698,
      "step": 112928
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.9738755822181702,
      "learning_rate": 0.0003092943150728729,
      "loss": 2.3127,
      "step": 112929
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.9311257004737854,
      "learning_rate": 0.0003092902264734213,
      "loss": 2.0823,
      "step": 112930
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.8711177706718445,
      "learning_rate": 0.00030928613787224263,
      "loss": 1.9957,
      "step": 112931
    },
    {
      "epoch": 1.47,
      "grad_norm": 1.004028558731079,
      "learning_rate": 0.0003092820492693374,
      "loss": 2.0925,
      "step": 112932
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.9536240696907043,
      "learning_rate": 0.0003092779606647064,
      "loss": 2.0928,
      "step": 112933
    },
    {
      "epoch": 1.47,
      "grad_norm": 1.1841557025909424,
      "learning_rate": 0.0003092738720583505,
      "loss": 2.0767,
      "step": 112934
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.9152097105979919,
      "learning_rate": 0.00030926978345027046,
      "loss": 2.0175,
      "step": 112935
    },
    {
      "epoch": 1.47,
      "grad_norm": 1.01422119140625,
      "learning_rate": 0.00030926569484046704,
      "loss": 2.2084,
      "step": 112936
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.9339852333068848,
      "learning_rate": 0.0003092616062289408,
      "loss": 2.1767,
      "step": 112937
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.9351192712783813,
      "learning_rate": 0.0003092575176156928,
      "loss": 2.3696,
      "step": 112938
    },
    {
      "epoch": 1.47,
      "grad_norm": 1.004764437675476,
      "learning_rate": 0.0003092534290007235,
      "loss": 2.2449,
      "step": 112939
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.9660791158676147,
      "learning_rate": 0.0003092493403840339,
      "loss": 2.268,
      "step": 112940
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.827560305595398,
      "learning_rate": 0.0003092452517656247,
      "loss": 2.3353,
      "step": 112941
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.9218130707740784,
      "learning_rate": 0.0003092411631454966,
      "loss": 2.1157,
      "step": 112942
    },
    {
      "epoch": 1.47,
      "grad_norm": 1.0659230947494507,
      "learning_rate": 0.0003092370745236504,
      "loss": 2.0717,
      "step": 112943
    },
    {
      "epoch": 1.47,
      "grad_norm": 1.0218256711959839,
      "learning_rate": 0.0003092329859000869,
      "loss": 2.2026,
      "step": 112944
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.9286023378372192,
      "learning_rate": 0.00030922889727480674,
      "loss": 2.1851,
      "step": 112945
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.9616540670394897,
      "learning_rate": 0.00030922480864781075,
      "loss": 2.2548,
      "step": 112946
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.8957963585853577,
      "learning_rate": 0.00030922072001909976,
      "loss": 2.0615,
      "step": 112947
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.9559576511383057,
      "learning_rate": 0.00030921663138867443,
      "loss": 2.1361,
      "step": 112948
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.9136697053909302,
      "learning_rate": 0.00030921254275653547,
      "loss": 2.2015,
      "step": 112949
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.9353340268135071,
      "learning_rate": 0.00030920845412268385,
      "loss": 2.0503,
      "step": 112950
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.8758553862571716,
      "learning_rate": 0.00030920436548712016,
      "loss": 2.345,
      "step": 112951
    },
    {
      "epoch": 1.47,
      "grad_norm": 1.1675701141357422,
      "learning_rate": 0.0003092002768498452,
      "loss": 2.0682,
      "step": 112952
    },
    {
      "epoch": 1.47,
      "grad_norm": 1.003401279449463,
      "learning_rate": 0.00030919618821085974,
      "loss": 2.1484,
      "step": 112953
    },
    {
      "epoch": 1.47,
      "grad_norm": 1.0154128074645996,
      "learning_rate": 0.0003091920995701646,
      "loss": 2.2612,
      "step": 112954
    },
    {
      "epoch": 1.47,
      "grad_norm": 1.1127572059631348,
      "learning_rate": 0.00030918801092776033,
      "loss": 2.119,
      "step": 112955
    },
    {
      "epoch": 1.47,
      "grad_norm": 1.0280473232269287,
      "learning_rate": 0.000309183922283648,
      "loss": 2.1715,
      "step": 112956
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.8759208917617798,
      "learning_rate": 0.00030917983363782807,
      "loss": 2.0856,
      "step": 112957
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.9818626046180725,
      "learning_rate": 0.00030917574499030146,
      "loss": 1.8327,
      "step": 112958
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.9145938754081726,
      "learning_rate": 0.000309171656341069,
      "loss": 2.1082,
      "step": 112959
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.8360936641693115,
      "learning_rate": 0.00030916756769013123,
      "loss": 2.2264,
      "step": 112960
    },
    {
      "epoch": 1.47,
      "grad_norm": 1.010894775390625,
      "learning_rate": 0.00030916347903748903,
      "loss": 1.8438,
      "step": 112961
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.9058308005332947,
      "learning_rate": 0.0003091593903831433,
      "loss": 2.2445,
      "step": 112962
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.9223878979682922,
      "learning_rate": 0.00030915530172709457,
      "loss": 2.0983,
      "step": 112963
    },
    {
      "epoch": 1.47,
      "grad_norm": 1.0686872005462646,
      "learning_rate": 0.00030915121306934377,
      "loss": 1.9993,
      "step": 112964
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.8983522057533264,
      "learning_rate": 0.00030914712440989153,
      "loss": 2.1121,
      "step": 112965
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.8495740294456482,
      "learning_rate": 0.00030914303574873866,
      "loss": 2.1028,
      "step": 112966
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.8731775879859924,
      "learning_rate": 0.00030913894708588593,
      "loss": 2.4318,
      "step": 112967
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.9582659602165222,
      "learning_rate": 0.00030913485842133415,
      "loss": 2.2504,
      "step": 112968
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.8790915012359619,
      "learning_rate": 0.00030913076975508396,
      "loss": 2.2084,
      "step": 112969
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.8892726302146912,
      "learning_rate": 0.00030912668108713624,
      "loss": 2.2623,
      "step": 112970
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.9313997030258179,
      "learning_rate": 0.00030912259241749164,
      "loss": 2.0506,
      "step": 112971
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.9369808435440063,
      "learning_rate": 0.000309118503746151,
      "loss": 1.9496,
      "step": 112972
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.9668492078781128,
      "learning_rate": 0.0003091144150731151,
      "loss": 1.9134,
      "step": 112973
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.8934824466705322,
      "learning_rate": 0.00030911032639838464,
      "loss": 1.9123,
      "step": 112974
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.9685381054878235,
      "learning_rate": 0.0003091062377219604,
      "loss": 2.1225,
      "step": 112975
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.9294587969779968,
      "learning_rate": 0.00030910214904384303,
      "loss": 1.9892,
      "step": 112976
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.9712272882461548,
      "learning_rate": 0.0003090980603640336,
      "loss": 2.108,
      "step": 112977
    },
    {
      "epoch": 1.47,
      "grad_norm": 1.01565682888031,
      "learning_rate": 0.0003090939716825325,
      "loss": 2.3355,
      "step": 112978
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.9196712374687195,
      "learning_rate": 0.00030908988299934077,
      "loss": 2.2137,
      "step": 112979
    },
    {
      "epoch": 1.47,
      "grad_norm": 1.141790747642517,
      "learning_rate": 0.00030908579431445895,
      "loss": 2.0914,
      "step": 112980
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.9134012460708618,
      "learning_rate": 0.00030908170562788794,
      "loss": 2.1924,
      "step": 112981
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.863766074180603,
      "learning_rate": 0.00030907761693962856,
      "loss": 2.0567,
      "step": 112982
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.92227703332901,
      "learning_rate": 0.0003090735282496814,
      "loss": 2.2552,
      "step": 112983
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.9321898221969604,
      "learning_rate": 0.00030906943955804733,
      "loss": 2.223,
      "step": 112984
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.8182195425033569,
      "learning_rate": 0.00030906535086472705,
      "loss": 1.8375,
      "step": 112985
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.9052846431732178,
      "learning_rate": 0.00030906126216972133,
      "loss": 2.3513,
      "step": 112986
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.8715739846229553,
      "learning_rate": 0.000309057173473031,
      "loss": 2.0381,
      "step": 112987
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.9918263554573059,
      "learning_rate": 0.00030905308477465675,
      "loss": 2.122,
      "step": 112988
    },
    {
      "epoch": 1.47,
      "grad_norm": 1.3816131353378296,
      "learning_rate": 0.00030904899607459935,
      "loss": 1.9443,
      "step": 112989
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.9190788865089417,
      "learning_rate": 0.0003090449073728596,
      "loss": 2.0018,
      "step": 112990
    },
    {
      "epoch": 1.47,
      "grad_norm": 1.0324697494506836,
      "learning_rate": 0.00030904081866943824,
      "loss": 2.2921,
      "step": 112991
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.9428195953369141,
      "learning_rate": 0.000309036729964336,
      "loss": 2.2648,
      "step": 112992
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.8963196873664856,
      "learning_rate": 0.00030903264125755367,
      "loss": 2.1818,
      "step": 112993
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.9301084876060486,
      "learning_rate": 0.00030902855254909197,
      "loss": 2.0375,
      "step": 112994
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.8807952404022217,
      "learning_rate": 0.0003090244638389517,
      "loss": 2.1399,
      "step": 112995
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.9651781916618347,
      "learning_rate": 0.0003090203751271337,
      "loss": 2.101,
      "step": 112996
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.9387566447257996,
      "learning_rate": 0.0003090162864136385,
      "loss": 2.315,
      "step": 112997
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.8527021408081055,
      "learning_rate": 0.00030901219769846714,
      "loss": 1.932,
      "step": 112998
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.9296391606330872,
      "learning_rate": 0.00030900810898162016,
      "loss": 2.2565,
      "step": 112999
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.9528404474258423,
      "learning_rate": 0.00030900402026309837,
      "loss": 2.091,
      "step": 113000
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.9013792872428894,
      "learning_rate": 0.0003089999315429026,
      "loss": 2.191,
      "step": 113001
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.9057667851448059,
      "learning_rate": 0.00030899584282103367,
      "loss": 2.1315,
      "step": 113002
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.8772202730178833,
      "learning_rate": 0.00030899175409749213,
      "loss": 2.1041,
      "step": 113003
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.9808611273765564,
      "learning_rate": 0.00030898766537227885,
      "loss": 2.1387,
      "step": 113004
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.8974536657333374,
      "learning_rate": 0.00030898357664539466,
      "loss": 2.0363,
      "step": 113005
    },
    {
      "epoch": 1.47,
      "grad_norm": 1.108697772026062,
      "learning_rate": 0.00030897948791684015,
      "loss": 2.3403,
      "step": 113006
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.9090535044670105,
      "learning_rate": 0.00030897539918661623,
      "loss": 2.0861,
      "step": 113007
    },
    {
      "epoch": 1.47,
      "grad_norm": 1.0102205276489258,
      "learning_rate": 0.00030897131045472373,
      "loss": 2.2209,
      "step": 113008
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.9581335783004761,
      "learning_rate": 0.00030896722172116313,
      "loss": 1.9855,
      "step": 113009
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.9727845788002014,
      "learning_rate": 0.0003089631329859354,
      "loss": 2.0491,
      "step": 113010
    },
    {
      "epoch": 1.47,
      "grad_norm": 1.0236563682556152,
      "learning_rate": 0.00030895904424904137,
      "loss": 2.0919,
      "step": 113011
    },
    {
      "epoch": 1.47,
      "grad_norm": 1.0357177257537842,
      "learning_rate": 0.00030895495551048157,
      "loss": 2.0865,
      "step": 113012
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.9521721005439758,
      "learning_rate": 0.0003089508667702568,
      "loss": 2.0474,
      "step": 113013
    },
    {
      "epoch": 1.47,
      "grad_norm": 1.3419779539108276,
      "learning_rate": 0.00030894677802836807,
      "loss": 2.1019,
      "step": 113014
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.9113278388977051,
      "learning_rate": 0.00030894268928481584,
      "loss": 1.9874,
      "step": 113015
    },
    {
      "epoch": 1.47,
      "grad_norm": 1.0261601209640503,
      "learning_rate": 0.00030893860053960104,
      "loss": 2.048,
      "step": 113016
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.9789390563964844,
      "learning_rate": 0.00030893451179272443,
      "loss": 2.1874,
      "step": 113017
    },
    {
      "epoch": 1.47,
      "grad_norm": 1.1465827226638794,
      "learning_rate": 0.00030893042304418666,
      "loss": 2.0199,
      "step": 113018
    },
    {
      "epoch": 1.47,
      "grad_norm": 1.0079715251922607,
      "learning_rate": 0.0003089263342939886,
      "loss": 2.2238,
      "step": 113019
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.9183331727981567,
      "learning_rate": 0.00030892224554213095,
      "loss": 2.235,
      "step": 113020
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.9613009095191956,
      "learning_rate": 0.00030891815678861447,
      "loss": 2.0005,
      "step": 113021
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.9824131727218628,
      "learning_rate": 0.0003089140680334399,
      "loss": 2.0442,
      "step": 113022
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.9994247555732727,
      "learning_rate": 0.0003089099792766081,
      "loss": 2.0291,
      "step": 113023
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.900547444820404,
      "learning_rate": 0.00030890589051811975,
      "loss": 2.2681,
      "step": 113024
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.868691623210907,
      "learning_rate": 0.0003089018017579756,
      "loss": 2.0607,
      "step": 113025
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.8569832444190979,
      "learning_rate": 0.00030889771299617657,
      "loss": 2.0866,
      "step": 113026
    },
    {
      "epoch": 1.47,
      "grad_norm": 1.07453453540802,
      "learning_rate": 0.0003088936242327231,
      "loss": 2.1931,
      "step": 113027
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.8410316109657288,
      "learning_rate": 0.00030888953546761623,
      "loss": 2.3007,
      "step": 113028
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.936103105545044,
      "learning_rate": 0.00030888544670085665,
      "loss": 2.1183,
      "step": 113029
    },
    {
      "epoch": 1.47,
      "grad_norm": 1.0720109939575195,
      "learning_rate": 0.000308881357932445,
      "loss": 1.8748,
      "step": 113030
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.9387760758399963,
      "learning_rate": 0.0003088772691623822,
      "loss": 2.5004,
      "step": 113031
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.9007430672645569,
      "learning_rate": 0.000308873180390669,
      "loss": 2.2653,
      "step": 113032
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.9177051186561584,
      "learning_rate": 0.00030886909161730605,
      "loss": 1.9621,
      "step": 113033
    },
    {
      "epoch": 1.47,
      "grad_norm": 1.018397331237793,
      "learning_rate": 0.0003088650028422942,
      "loss": 2.2856,
      "step": 113034
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.832512617111206,
      "learning_rate": 0.0003088609140656341,
      "loss": 2.1014,
      "step": 113035
    },
    {
      "epoch": 1.47,
      "grad_norm": 1.0196079015731812,
      "learning_rate": 0.00030885682528732664,
      "loss": 2.0976,
      "step": 113036
    },
    {
      "epoch": 1.47,
      "grad_norm": 2.071617364883423,
      "learning_rate": 0.0003088527365073726,
      "loss": 2.1874,
      "step": 113037
    },
    {
      "epoch": 1.47,
      "grad_norm": 1.009892463684082,
      "learning_rate": 0.0003088486477257726,
      "loss": 2.2107,
      "step": 113038
    },
    {
      "epoch": 1.47,
      "grad_norm": 1.0121779441833496,
      "learning_rate": 0.0003088445589425274,
      "loss": 2.2749,
      "step": 113039
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.8901588320732117,
      "learning_rate": 0.00030884047015763787,
      "loss": 2.3877,
      "step": 113040
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.909492552280426,
      "learning_rate": 0.0003088363813711049,
      "loss": 2.1798,
      "step": 113041
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.8714290261268616,
      "learning_rate": 0.00030883229258292885,
      "loss": 2.0423,
      "step": 113042
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.9887617826461792,
      "learning_rate": 0.0003088282037931108,
      "loss": 2.2776,
      "step": 113043
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.9700497388839722,
      "learning_rate": 0.00030882411500165145,
      "loss": 2.262,
      "step": 113044
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.8894742727279663,
      "learning_rate": 0.00030882002620855146,
      "loss": 2.0944,
      "step": 113045
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.8891518712043762,
      "learning_rate": 0.00030881593741381166,
      "loss": 1.9982,
      "step": 113046
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.872093141078949,
      "learning_rate": 0.0003088118486174329,
      "loss": 2.1514,
      "step": 113047
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.8796091079711914,
      "learning_rate": 0.00030880775981941585,
      "loss": 2.1071,
      "step": 113048
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.9063603281974792,
      "learning_rate": 0.0003088036710197612,
      "loss": 2.0192,
      "step": 113049
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.9366709589958191,
      "learning_rate": 0.00030879958221846984,
      "loss": 2.3033,
      "step": 113050
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.9122424125671387,
      "learning_rate": 0.0003087954934155424,
      "loss": 2.1125,
      "step": 113051
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.9122607111930847,
      "learning_rate": 0.00030879140461097975,
      "loss": 2.0958,
      "step": 113052
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.9199602007865906,
      "learning_rate": 0.00030878731580478264,
      "loss": 2.3255,
      "step": 113053
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.9128482341766357,
      "learning_rate": 0.0003087832269969517,
      "loss": 2.2355,
      "step": 113054
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.9277916550636292,
      "learning_rate": 0.0003087791381874879,
      "loss": 2.0977,
      "step": 113055
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.8958948850631714,
      "learning_rate": 0.0003087750493763919,
      "loss": 2.0932,
      "step": 113056
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.8746770620346069,
      "learning_rate": 0.00030877096056366435,
      "loss": 2.2015,
      "step": 113057
    },
    {
      "epoch": 1.47,
      "grad_norm": 1.129318118095398,
      "learning_rate": 0.0003087668717493062,
      "loss": 2.2847,
      "step": 113058
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.9242250323295593,
      "learning_rate": 0.00030876278293331814,
      "loss": 2.0581,
      "step": 113059
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.9481918811798096,
      "learning_rate": 0.00030875869411570086,
      "loss": 2.224,
      "step": 113060
    },
    {
      "epoch": 1.47,
      "grad_norm": 1.0522334575653076,
      "learning_rate": 0.0003087546052964551,
      "loss": 2.2328,
      "step": 113061
    },
    {
      "epoch": 1.47,
      "grad_norm": 1.21428382396698,
      "learning_rate": 0.00030875051647558186,
      "loss": 2.1902,
      "step": 113062
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.9497345089912415,
      "learning_rate": 0.0003087464276530816,
      "loss": 2.0604,
      "step": 113063
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.9862563610076904,
      "learning_rate": 0.00030874233882895527,
      "loss": 2.1461,
      "step": 113064
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.9911181330680847,
      "learning_rate": 0.0003087382500032036,
      "loss": 2.1328,
      "step": 113065
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.8824079632759094,
      "learning_rate": 0.00030873416117582726,
      "loss": 2.1525,
      "step": 113066
    },
    {
      "epoch": 1.47,
      "grad_norm": 1.0810911655426025,
      "learning_rate": 0.00030873007234682716,
      "loss": 2.1919,
      "step": 113067
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.9741555452346802,
      "learning_rate": 0.00030872598351620385,
      "loss": 2.0824,
      "step": 113068
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.9559638500213623,
      "learning_rate": 0.00030872189468395835,
      "loss": 2.1781,
      "step": 113069
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.9158901572227478,
      "learning_rate": 0.00030871780585009116,
      "loss": 2.2733,
      "step": 113070
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.8791887164115906,
      "learning_rate": 0.00030871371701460325,
      "loss": 2.1742,
      "step": 113071
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.8929564356803894,
      "learning_rate": 0.0003087096281774952,
      "loss": 2.0697,
      "step": 113072
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.8927314877510071,
      "learning_rate": 0.00030870553933876803,
      "loss": 2.1467,
      "step": 113073
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.9094114303588867,
      "learning_rate": 0.0003087014504984222,
      "loss": 2.1774,
      "step": 113074
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.8921862244606018,
      "learning_rate": 0.00030869736165645866,
      "loss": 2.0853,
      "step": 113075
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.8254357576370239,
      "learning_rate": 0.0003086932728128781,
      "loss": 2.3136,
      "step": 113076
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.8836190104484558,
      "learning_rate": 0.0003086891839676813,
      "loss": 1.9412,
      "step": 113077
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.8758053779602051,
      "learning_rate": 0.00030868509512086903,
      "loss": 2.3214,
      "step": 113078
    },
    {
      "epoch": 1.47,
      "grad_norm": 1.0038933753967285,
      "learning_rate": 0.000308681006272442,
      "loss": 2.038,
      "step": 113079
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.9639304280281067,
      "learning_rate": 0.00030867691742240104,
      "loss": 2.1109,
      "step": 113080
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.9382052421569824,
      "learning_rate": 0.00030867282857074686,
      "loss": 2.1703,
      "step": 113081
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.9319473505020142,
      "learning_rate": 0.00030866873971748026,
      "loss": 2.2142,
      "step": 113082
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.9304749965667725,
      "learning_rate": 0.000308664650862602,
      "loss": 1.9827,
      "step": 113083
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.9273444414138794,
      "learning_rate": 0.00030866056200611275,
      "loss": 2.1605,
      "step": 113084
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.9853515028953552,
      "learning_rate": 0.00030865647314801337,
      "loss": 2.3094,
      "step": 113085
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.9241533875465393,
      "learning_rate": 0.00030865238428830455,
      "loss": 2.0336,
      "step": 113086
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.9803757667541504,
      "learning_rate": 0.0003086482954269872,
      "loss": 2.165,
      "step": 113087
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.9070064425468445,
      "learning_rate": 0.0003086442065640618,
      "loss": 2.2365,
      "step": 113088
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.9576765298843384,
      "learning_rate": 0.00030864011769952943,
      "loss": 2.2906,
      "step": 113089
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.8824260830879211,
      "learning_rate": 0.0003086360288333907,
      "loss": 2.0827,
      "step": 113090
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.9656122922897339,
      "learning_rate": 0.0003086319399656463,
      "loss": 2.3271,
      "step": 113091
    },
    {
      "epoch": 1.47,
      "grad_norm": 1.0248419046401978,
      "learning_rate": 0.000308627851096297,
      "loss": 2.0904,
      "step": 113092
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.8960063457489014,
      "learning_rate": 0.00030862376222534377,
      "loss": 2.2418,
      "step": 113093
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.8269273042678833,
      "learning_rate": 0.00030861967335278715,
      "loss": 2.0257,
      "step": 113094
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.8784646987915039,
      "learning_rate": 0.00030861558447862786,
      "loss": 2.0949,
      "step": 113095
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.9009355902671814,
      "learning_rate": 0.000308611495602867,
      "loss": 1.8072,
      "step": 113096
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.9153671264648438,
      "learning_rate": 0.00030860740672550497,
      "loss": 1.9273,
      "step": 113097
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.8854258060455322,
      "learning_rate": 0.0003086033178465426,
      "loss": 2.197,
      "step": 113098
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.9458150267601013,
      "learning_rate": 0.0003085992289659808,
      "loss": 2.1718,
      "step": 113099
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.9284683465957642,
      "learning_rate": 0.0003085951400838202,
      "loss": 2.1745,
      "step": 113100
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.9341436624526978,
      "learning_rate": 0.00030859105120006165,
      "loss": 2.2058,
      "step": 113101
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.9066395163536072,
      "learning_rate": 0.00030858696231470586,
      "loss": 2.2232,
      "step": 113102
    },
    {
      "epoch": 1.47,
      "grad_norm": 1.0276732444763184,
      "learning_rate": 0.00030858287342775347,
      "loss": 2.3171,
      "step": 113103
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.9939603805541992,
      "learning_rate": 0.00030857878453920546,
      "loss": 1.9396,
      "step": 113104
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.926215648651123,
      "learning_rate": 0.00030857469564906254,
      "loss": 2.1787,
      "step": 113105
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.9668364524841309,
      "learning_rate": 0.00030857060675732535,
      "loss": 2.1883,
      "step": 113106
    },
    {
      "epoch": 1.47,
      "grad_norm": 1.1969038248062134,
      "learning_rate": 0.0003085665178639947,
      "loss": 2.1257,
      "step": 113107
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.9021274447441101,
      "learning_rate": 0.0003085624289690715,
      "loss": 1.9612,
      "step": 113108
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.8550645709037781,
      "learning_rate": 0.0003085583400725563,
      "loss": 2.19,
      "step": 113109
    },
    {
      "epoch": 1.47,
      "grad_norm": 1.0202957391738892,
      "learning_rate": 0.0003085542511744499,
      "loss": 2.138,
      "step": 113110
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.9512560963630676,
      "learning_rate": 0.00030855016227475314,
      "loss": 2.1783,
      "step": 113111
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.8582198023796082,
      "learning_rate": 0.0003085460733734667,
      "loss": 2.0144,
      "step": 113112
    },
    {
      "epoch": 1.47,
      "grad_norm": 1.0432509183883667,
      "learning_rate": 0.00030854198447059143,
      "loss": 2.4096,
      "step": 113113
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.9824069142341614,
      "learning_rate": 0.00030853789556612804,
      "loss": 2.2806,
      "step": 113114
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.9522160291671753,
      "learning_rate": 0.0003085338066600773,
      "loss": 2.3002,
      "step": 113115
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.954558789730072,
      "learning_rate": 0.00030852971775244,
      "loss": 2.1812,
      "step": 113116
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.9301915168762207,
      "learning_rate": 0.00030852562884321684,
      "loss": 2.3525,
      "step": 113117
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.8513377904891968,
      "learning_rate": 0.00030852153993240854,
      "loss": 2.3325,
      "step": 113118
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.990351676940918,
      "learning_rate": 0.000308517451020016,
      "loss": 2.2193,
      "step": 113119
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.9223989248275757,
      "learning_rate": 0.00030851336210603993,
      "loss": 1.9545,
      "step": 113120
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.8212443590164185,
      "learning_rate": 0.000308509273190481,
      "loss": 1.9917,
      "step": 113121
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.8572792410850525,
      "learning_rate": 0.00030850518427334,
      "loss": 2.1337,
      "step": 113122
    },
    {
      "epoch": 1.47,
      "grad_norm": 1.0923882722854614,
      "learning_rate": 0.00030850109535461784,
      "loss": 2.2425,
      "step": 113123
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.9084864258766174,
      "learning_rate": 0.000308497006434315,
      "loss": 2.1066,
      "step": 113124
    },
    {
      "epoch": 1.47,
      "grad_norm": 1.0775436162948608,
      "learning_rate": 0.00030849291751243257,
      "loss": 2.2024,
      "step": 113125
    },
    {
      "epoch": 1.47,
      "grad_norm": 1.0731406211853027,
      "learning_rate": 0.00030848882858897114,
      "loss": 2.3859,
      "step": 113126
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.997944712638855,
      "learning_rate": 0.0003084847396639314,
      "loss": 2.2714,
      "step": 113127
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.8540942072868347,
      "learning_rate": 0.00030848065073731416,
      "loss": 2.2905,
      "step": 113128
    },
    {
      "epoch": 1.47,
      "grad_norm": 1.0466821193695068,
      "learning_rate": 0.0003084765618091203,
      "loss": 2.2043,
      "step": 113129
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.9711620807647705,
      "learning_rate": 0.0003084724728793504,
      "loss": 2.0817,
      "step": 113130
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.9580978155136108,
      "learning_rate": 0.00030846838394800534,
      "loss": 2.1374,
      "step": 113131
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.9419540166854858,
      "learning_rate": 0.00030846429501508587,
      "loss": 2.2188,
      "step": 113132
    },
    {
      "epoch": 1.47,
      "grad_norm": 1.0129530429840088,
      "learning_rate": 0.0003084602060805927,
      "loss": 2.2311,
      "step": 113133
    },
    {
      "epoch": 1.47,
      "grad_norm": 1.0372159481048584,
      "learning_rate": 0.00030845611714452664,
      "loss": 2.2886,
      "step": 113134
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.8463169932365417,
      "learning_rate": 0.0003084520282068884,
      "loss": 2.3264,
      "step": 113135
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.8915860652923584,
      "learning_rate": 0.0003084479392676789,
      "loss": 2.1427,
      "step": 113136
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.9611428380012512,
      "learning_rate": 0.00030844385032689856,
      "loss": 2.4013,
      "step": 113137
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.934440553188324,
      "learning_rate": 0.00030843976138454843,
      "loss": 2.2355,
      "step": 113138
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.8082972168922424,
      "learning_rate": 0.00030843567244062925,
      "loss": 2.0396,
      "step": 113139
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.9056079983711243,
      "learning_rate": 0.00030843158349514165,
      "loss": 2.0041,
      "step": 113140
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.8692792057991028,
      "learning_rate": 0.00030842749454808646,
      "loss": 2.182,
      "step": 113141
    },
    {
      "epoch": 1.47,
      "grad_norm": 1.0715579986572266,
      "learning_rate": 0.00030842340559946455,
      "loss": 2.3001,
      "step": 113142
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.983879029750824,
      "learning_rate": 0.00030841931664927645,
      "loss": 2.2743,
      "step": 113143
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.965584397315979,
      "learning_rate": 0.00030841522769752303,
      "loss": 2.1079,
      "step": 113144
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.9490917325019836,
      "learning_rate": 0.00030841113874420517,
      "loss": 2.1398,
      "step": 113145
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.9020359516143799,
      "learning_rate": 0.0003084070497893234,
      "loss": 1.9742,
      "step": 113146
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.937366783618927,
      "learning_rate": 0.00030840296083287864,
      "loss": 2.1898,
      "step": 113147
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.8767877817153931,
      "learning_rate": 0.00030839887187487165,
      "loss": 2.1737,
      "step": 113148
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.986501157283783,
      "learning_rate": 0.00030839478291530315,
      "loss": 2.286,
      "step": 113149
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.9224976897239685,
      "learning_rate": 0.0003083906939541738,
      "loss": 2.1513,
      "step": 113150
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.8741743564605713,
      "learning_rate": 0.00030838660499148466,
      "loss": 2.0979,
      "step": 113151
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.9469369649887085,
      "learning_rate": 0.00030838251602723614,
      "loss": 2.2556,
      "step": 113152
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.9264469742774963,
      "learning_rate": 0.00030837842706142913,
      "loss": 2.3647,
      "step": 113153
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.9896238446235657,
      "learning_rate": 0.00030837433809406456,
      "loss": 2.1993,
      "step": 113154
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.9419480562210083,
      "learning_rate": 0.0003083702491251429,
      "loss": 2.1634,
      "step": 113155
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.8906750679016113,
      "learning_rate": 0.0003083661601546651,
      "loss": 2.1923,
      "step": 113156
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.9080997109413147,
      "learning_rate": 0.0003083620711826319,
      "loss": 1.9856,
      "step": 113157
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.9202333092689514,
      "learning_rate": 0.000308357982209044,
      "loss": 2.5704,
      "step": 113158
    },
    {
      "epoch": 1.47,
      "grad_norm": 1.0536081790924072,
      "learning_rate": 0.0003083538932339022,
      "loss": 1.9069,
      "step": 113159
    },
    {
      "epoch": 1.47,
      "grad_norm": 1.0711461305618286,
      "learning_rate": 0.0003083498042572072,
      "loss": 2.0986,
      "step": 113160
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.9789736270904541,
      "learning_rate": 0.0003083457152789599,
      "loss": 2.2096,
      "step": 113161
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.8751001954078674,
      "learning_rate": 0.0003083416262991609,
      "loss": 2.0356,
      "step": 113162
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.9643834233283997,
      "learning_rate": 0.00030833753731781107,
      "loss": 2.091,
      "step": 113163
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.8627825379371643,
      "learning_rate": 0.0003083334483349112,
      "loss": 2.1502,
      "step": 113164
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.9853464961051941,
      "learning_rate": 0.0003083293593504619,
      "loss": 2.2566,
      "step": 113165
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.9053523540496826,
      "learning_rate": 0.000308325270364464,
      "loss": 2.0897,
      "step": 113166
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.9825003147125244,
      "learning_rate": 0.00030832118137691833,
      "loss": 2.0497,
      "step": 113167
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.9511057734489441,
      "learning_rate": 0.00030831709238782564,
      "loss": 2.0986,
      "step": 113168
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.9460774064064026,
      "learning_rate": 0.00030831300339718656,
      "loss": 2.2513,
      "step": 113169
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.8879102468490601,
      "learning_rate": 0.0003083089144050019,
      "loss": 2.1819,
      "step": 113170
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.8584403991699219,
      "learning_rate": 0.00030830482541127256,
      "loss": 2.2295,
      "step": 113171
    },
    {
      "epoch": 1.47,
      "grad_norm": 1.002812385559082,
      "learning_rate": 0.00030830073641599916,
      "loss": 2.2484,
      "step": 113172
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.9530351161956787,
      "learning_rate": 0.0003082966474191825,
      "loss": 2.1176,
      "step": 113173
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.9753277897834778,
      "learning_rate": 0.00030829255842082325,
      "loss": 2.0858,
      "step": 113174
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.9561878442764282,
      "learning_rate": 0.00030828846942092237,
      "loss": 2.1833,
      "step": 113175
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.9735193252563477,
      "learning_rate": 0.00030828438041948045,
      "loss": 2.1528,
      "step": 113176
    },
    {
      "epoch": 1.47,
      "grad_norm": 1.1320503950119019,
      "learning_rate": 0.0003082802914164983,
      "loss": 2.151,
      "step": 113177
    },
    {
      "epoch": 1.47,
      "grad_norm": 1.0142064094543457,
      "learning_rate": 0.0003082762024119768,
      "loss": 2.3537,
      "step": 113178
    },
    {
      "epoch": 1.47,
      "grad_norm": 1.0062377452850342,
      "learning_rate": 0.00030827211340591655,
      "loss": 2.1426,
      "step": 113179
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.8970158696174622,
      "learning_rate": 0.00030826802439831827,
      "loss": 2.2088,
      "step": 113180
    },
    {
      "epoch": 1.47,
      "grad_norm": 1.0200202465057373,
      "learning_rate": 0.00030826393538918286,
      "loss": 2.0776,
      "step": 113181
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.894740641117096,
      "learning_rate": 0.0003082598463785111,
      "loss": 2.0276,
      "step": 113182
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.896569013595581,
      "learning_rate": 0.0003082557573663035,
      "loss": 1.9952,
      "step": 113183
    },
    {
      "epoch": 1.47,
      "grad_norm": 1.0383524894714355,
      "learning_rate": 0.0003082516683525612,
      "loss": 2.0652,
      "step": 113184
    },
    {
      "epoch": 1.47,
      "grad_norm": 1.0703074932098389,
      "learning_rate": 0.00030824757933728457,
      "loss": 2.2703,
      "step": 113185
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.9321029186248779,
      "learning_rate": 0.0003082434903204747,
      "loss": 2.0202,
      "step": 113186
    },
    {
      "epoch": 1.47,
      "grad_norm": 1.0261988639831543,
      "learning_rate": 0.0003082394013021322,
      "loss": 2.1194,
      "step": 113187
    },
    {
      "epoch": 1.47,
      "grad_norm": 1.116504430770874,
      "learning_rate": 0.00030823531228225774,
      "loss": 2.0622,
      "step": 113188
    },
    {
      "epoch": 1.47,
      "grad_norm": 1.0193476676940918,
      "learning_rate": 0.00030823122326085225,
      "loss": 2.025,
      "step": 113189
    },
    {
      "epoch": 1.47,
      "grad_norm": 1.0413025617599487,
      "learning_rate": 0.0003082271342379165,
      "loss": 2.0042,
      "step": 113190
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.935308039188385,
      "learning_rate": 0.000308223045213451,
      "loss": 2.4197,
      "step": 113191
    },
    {
      "epoch": 1.47,
      "grad_norm": 1.3317002058029175,
      "learning_rate": 0.0003082189561874568,
      "loss": 1.9594,
      "step": 113192
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.9448999166488647,
      "learning_rate": 0.00030821486715993453,
      "loss": 2.1577,
      "step": 113193
    },
    {
      "epoch": 1.47,
      "grad_norm": 1.0933374166488647,
      "learning_rate": 0.0003082107781308849,
      "loss": 2.2774,
      "step": 113194
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.9429996013641357,
      "learning_rate": 0.0003082066891003087,
      "loss": 2.1391,
      "step": 113195
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.911173939704895,
      "learning_rate": 0.00030820260006820686,
      "loss": 2.1073,
      "step": 113196
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.9840654730796814,
      "learning_rate": 0.00030819851103458,
      "loss": 2.2776,
      "step": 113197
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.8834023475646973,
      "learning_rate": 0.00030819442199942875,
      "loss": 2.2921,
      "step": 113198
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.8313775062561035,
      "learning_rate": 0.0003081903329627541,
      "loss": 1.8088,
      "step": 113199
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.8660650253295898,
      "learning_rate": 0.00030818624392455667,
      "loss": 1.9923,
      "step": 113200
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.8878933787345886,
      "learning_rate": 0.00030818215488483727,
      "loss": 2.012,
      "step": 113201
    },
    {
      "epoch": 1.47,
      "grad_norm": 1.0077931880950928,
      "learning_rate": 0.00030817806584359657,
      "loss": 2.3232,
      "step": 113202
    },
    {
      "epoch": 1.47,
      "grad_norm": 1.0204468965530396,
      "learning_rate": 0.0003081739768008356,
      "loss": 2.1397,
      "step": 113203
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.9684227705001831,
      "learning_rate": 0.0003081698877565548,
      "loss": 2.2596,
      "step": 113204
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.9365401864051819,
      "learning_rate": 0.00030816579871075503,
      "loss": 1.9717,
      "step": 113205
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.948701798915863,
      "learning_rate": 0.0003081617096634372,
      "loss": 2.2668,
      "step": 113206
    },
    {
      "epoch": 1.47,
      "grad_norm": 1.124915361404419,
      "learning_rate": 0.0003081576206146019,
      "loss": 2.1969,
      "step": 113207
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.926449179649353,
      "learning_rate": 0.0003081535315642499,
      "loss": 2.141,
      "step": 113208
    },
    {
      "epoch": 1.47,
      "grad_norm": 1.2127306461334229,
      "learning_rate": 0.0003081494425123821,
      "loss": 2.2512,
      "step": 113209
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.8270591497421265,
      "learning_rate": 0.0003081453534589991,
      "loss": 2.0341,
      "step": 113210
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.92691570520401,
      "learning_rate": 0.00030814126440410175,
      "loss": 2.3219,
      "step": 113211
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.9313664436340332,
      "learning_rate": 0.00030813717534769084,
      "loss": 1.9063,
      "step": 113212
    },
    {
      "epoch": 1.47,
      "grad_norm": 1.0236717462539673,
      "learning_rate": 0.000308133086289767,
      "loss": 1.9848,
      "step": 113213
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.8895362019538879,
      "learning_rate": 0.00030812899723033104,
      "loss": 2.0136,
      "step": 113214
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.9201312065124512,
      "learning_rate": 0.0003081249081693838,
      "loss": 2.031,
      "step": 113215
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.987257182598114,
      "learning_rate": 0.00030812081910692595,
      "loss": 2.1963,
      "step": 113216
    },
    {
      "epoch": 1.47,
      "grad_norm": 1.0334197282791138,
      "learning_rate": 0.00030811673004295826,
      "loss": 2.216,
      "step": 113217
    },
    {
      "epoch": 1.47,
      "grad_norm": 1.0397286415100098,
      "learning_rate": 0.00030811264097748163,
      "loss": 2.1644,
      "step": 113218
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.9408084750175476,
      "learning_rate": 0.00030810855191049666,
      "loss": 2.1339,
      "step": 113219
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.9738655090332031,
      "learning_rate": 0.0003081044628420041,
      "loss": 2.2051,
      "step": 113220
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.9522165656089783,
      "learning_rate": 0.00030810037377200484,
      "loss": 2.2913,
      "step": 113221
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.8664286136627197,
      "learning_rate": 0.0003080962847004995,
      "loss": 2.2883,
      "step": 113222
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.915848433971405,
      "learning_rate": 0.0003080921956274889,
      "loss": 2.0909,
      "step": 113223
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.9313275814056396,
      "learning_rate": 0.0003080881065529739,
      "loss": 2.3207,
      "step": 113224
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.9848204851150513,
      "learning_rate": 0.00030808401747695513,
      "loss": 2.1303,
      "step": 113225
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.9537283778190613,
      "learning_rate": 0.00030807992839943334,
      "loss": 2.2135,
      "step": 113226
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.9265283346176147,
      "learning_rate": 0.00030807583932040945,
      "loss": 2.1917,
      "step": 113227
    },
    {
      "epoch": 1.47,
      "grad_norm": 1.0483816862106323,
      "learning_rate": 0.000308071750239884,
      "loss": 2.0299,
      "step": 113228
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.9112474918365479,
      "learning_rate": 0.0003080676611578579,
      "loss": 2.1525,
      "step": 113229
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.9866667985916138,
      "learning_rate": 0.0003080635720743319,
      "loss": 2.3271,
      "step": 113230
    },
    {
      "epoch": 1.47,
      "grad_norm": 1.46812105178833,
      "learning_rate": 0.0003080594829893066,
      "loss": 2.0385,
      "step": 113231
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.9929998517036438,
      "learning_rate": 0.00030805539390278295,
      "loss": 2.181,
      "step": 113232
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.9311212301254272,
      "learning_rate": 0.0003080513048147618,
      "loss": 2.1881,
      "step": 113233
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.9325361251831055,
      "learning_rate": 0.0003080472157252436,
      "loss": 2.1244,
      "step": 113234
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.9973688125610352,
      "learning_rate": 0.0003080431266342293,
      "loss": 2.0463,
      "step": 113235
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.9795584082603455,
      "learning_rate": 0.0003080390375417197,
      "loss": 2.2038,
      "step": 113236
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.9165408611297607,
      "learning_rate": 0.00030803494844771533,
      "loss": 1.9824,
      "step": 113237
    },
    {
      "epoch": 1.47,
      "grad_norm": 1.3703036308288574,
      "learning_rate": 0.00030803085935221725,
      "loss": 2.0361,
      "step": 113238
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.9156553149223328,
      "learning_rate": 0.00030802677025522606,
      "loss": 2.3103,
      "step": 113239
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.9401321411132812,
      "learning_rate": 0.00030802268115674253,
      "loss": 2.2536,
      "step": 113240
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.9512748122215271,
      "learning_rate": 0.0003080185920567674,
      "loss": 2.1564,
      "step": 113241
    },
    {
      "epoch": 1.47,
      "grad_norm": 1.1547902822494507,
      "learning_rate": 0.00030801450295530155,
      "loss": 2.1077,
      "step": 113242
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.9050657749176025,
      "learning_rate": 0.00030801041385234557,
      "loss": 2.127,
      "step": 113243
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.8814738988876343,
      "learning_rate": 0.0003080063247479003,
      "loss": 2.0546,
      "step": 113244
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.9318860173225403,
      "learning_rate": 0.0003080022356419666,
      "loss": 2.1883,
      "step": 113245
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.9433618187904358,
      "learning_rate": 0.00030799814653454505,
      "loss": 2.3794,
      "step": 113246
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.9188924431800842,
      "learning_rate": 0.0003079940574256365,
      "loss": 2.1581,
      "step": 113247
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.9374459385871887,
      "learning_rate": 0.0003079899683152417,
      "loss": 2.2334,
      "step": 113248
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.8757889270782471,
      "learning_rate": 0.0003079858792033614,
      "loss": 2.0798,
      "step": 113249
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.9375302791595459,
      "learning_rate": 0.0003079817900899964,
      "loss": 2.0943,
      "step": 113250
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.926719069480896,
      "learning_rate": 0.00030797770097514747,
      "loss": 1.986,
      "step": 113251
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.8850085139274597,
      "learning_rate": 0.00030797361185881526,
      "loss": 1.9662,
      "step": 113252
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.8897066116333008,
      "learning_rate": 0.0003079695227410007,
      "loss": 1.9731,
      "step": 113253
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.9585806131362915,
      "learning_rate": 0.00030796543362170435,
      "loss": 2.1427,
      "step": 113254
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.8991863131523132,
      "learning_rate": 0.00030796134450092707,
      "loss": 2.1598,
      "step": 113255
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.839596688747406,
      "learning_rate": 0.00030795725537866974,
      "loss": 1.7218,
      "step": 113256
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.9614853858947754,
      "learning_rate": 0.0003079531662549329,
      "loss": 1.8652,
      "step": 113257
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.8849650621414185,
      "learning_rate": 0.0003079490771297175,
      "loss": 2.1102,
      "step": 113258
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.918175220489502,
      "learning_rate": 0.00030794498800302417,
      "loss": 2.065,
      "step": 113259
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.8991642594337463,
      "learning_rate": 0.00030794089887485376,
      "loss": 2.0938,
      "step": 113260
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.9559497833251953,
      "learning_rate": 0.00030793680974520695,
      "loss": 1.7997,
      "step": 113261
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.9396059513092041,
      "learning_rate": 0.00030793272061408454,
      "loss": 2.1591,
      "step": 113262
    },
    {
      "epoch": 1.47,
      "grad_norm": 1.1086844205856323,
      "learning_rate": 0.00030792863148148734,
      "loss": 2.0005,
      "step": 113263
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.9838016629219055,
      "learning_rate": 0.00030792454234741595,
      "loss": 2.1542,
      "step": 113264
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.9326660633087158,
      "learning_rate": 0.0003079204532118713,
      "loss": 2.1862,
      "step": 113265
    },
    {
      "epoch": 1.47,
      "grad_norm": 1.1000754833221436,
      "learning_rate": 0.0003079163640748541,
      "loss": 2.2683,
      "step": 113266
    },
    {
      "epoch": 1.47,
      "grad_norm": 1.1722097396850586,
      "learning_rate": 0.00030791227493636505,
      "loss": 2.1743,
      "step": 113267
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.8760850429534912,
      "learning_rate": 0.00030790818579640503,
      "loss": 2.0311,
      "step": 113268
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.9859498143196106,
      "learning_rate": 0.0003079040966549746,
      "loss": 2.455,
      "step": 113269
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.880090057849884,
      "learning_rate": 0.0003079000075120748,
      "loss": 2.368,
      "step": 113270
    },
    {
      "epoch": 1.47,
      "grad_norm": 1.015254020690918,
      "learning_rate": 0.00030789591836770613,
      "loss": 2.2493,
      "step": 113271
    },
    {
      "epoch": 1.47,
      "grad_norm": 1.2736701965332031,
      "learning_rate": 0.0003078918292218695,
      "loss": 2.106,
      "step": 113272
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.857164740562439,
      "learning_rate": 0.0003078877400745657,
      "loss": 2.0598,
      "step": 113273
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.9772576093673706,
      "learning_rate": 0.00030788365092579537,
      "loss": 2.1821,
      "step": 113274
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.876521110534668,
      "learning_rate": 0.0003078795617755593,
      "loss": 1.9014,
      "step": 113275
    },
    {
      "epoch": 1.47,
      "grad_norm": 1.0413753986358643,
      "learning_rate": 0.00030787547262385833,
      "loss": 2.294,
      "step": 113276
    },
    {
      "epoch": 1.47,
      "grad_norm": 1.0840215682983398,
      "learning_rate": 0.00030787138347069306,
      "loss": 1.968,
      "step": 113277
    },
    {
      "epoch": 1.47,
      "grad_norm": 1.0643432140350342,
      "learning_rate": 0.0003078672943160644,
      "loss": 2.0251,
      "step": 113278
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.9584325551986694,
      "learning_rate": 0.00030786320515997307,
      "loss": 2.2207,
      "step": 113279
    },
    {
      "epoch": 1.47,
      "grad_norm": 1.0107917785644531,
      "learning_rate": 0.00030785911600241986,
      "loss": 2.0011,
      "step": 113280
    },
    {
      "epoch": 1.47,
      "grad_norm": 0.9329304695129395,
      "learning_rate": 0.0003078550268434054,
      "loss": 2.1719,
      "step": 113281
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.9755309224128723,
      "learning_rate": 0.0003078509376829306,
      "loss": 2.0651,
      "step": 113282
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.982552707195282,
      "learning_rate": 0.00030784684852099616,
      "loss": 2.117,
      "step": 113283
    },
    {
      "epoch": 1.48,
      "grad_norm": 1.2066930532455444,
      "learning_rate": 0.0003078427593576028,
      "loss": 2.1663,
      "step": 113284
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.9727907180786133,
      "learning_rate": 0.0003078386701927514,
      "loss": 1.8681,
      "step": 113285
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.9280011057853699,
      "learning_rate": 0.00030783458102644256,
      "loss": 2.1326,
      "step": 113286
    },
    {
      "epoch": 1.48,
      "grad_norm": 1.169560194015503,
      "learning_rate": 0.00030783049185867714,
      "loss": 2.0011,
      "step": 113287
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.9360527992248535,
      "learning_rate": 0.000307826402689456,
      "loss": 2.0827,
      "step": 113288
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.9704736471176147,
      "learning_rate": 0.00030782231351877964,
      "loss": 2.0546,
      "step": 113289
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.9787138104438782,
      "learning_rate": 0.000307818224346649,
      "loss": 2.1744,
      "step": 113290
    },
    {
      "epoch": 1.48,
      "grad_norm": 1.1531431674957275,
      "learning_rate": 0.00030781413517306486,
      "loss": 1.9443,
      "step": 113291
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.9284986257553101,
      "learning_rate": 0.0003078100459980279,
      "loss": 2.0597,
      "step": 113292
    },
    {
      "epoch": 1.48,
      "grad_norm": 1.154484510421753,
      "learning_rate": 0.00030780595682153885,
      "loss": 2.063,
      "step": 113293
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.8437947630882263,
      "learning_rate": 0.00030780186764359864,
      "loss": 2.2585,
      "step": 113294
    },
    {
      "epoch": 1.48,
      "grad_norm": 1.2419300079345703,
      "learning_rate": 0.00030779777846420776,
      "loss": 2.1303,
      "step": 113295
    },
    {
      "epoch": 1.48,
      "grad_norm": 1.0763601064682007,
      "learning_rate": 0.00030779368928336725,
      "loss": 2.0566,
      "step": 113296
    },
    {
      "epoch": 1.48,
      "grad_norm": 1.015328288078308,
      "learning_rate": 0.00030778960010107776,
      "loss": 2.3343,
      "step": 113297
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.8665938377380371,
      "learning_rate": 0.00030778551091733994,
      "loss": 2.4103,
      "step": 113298
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.9867259860038757,
      "learning_rate": 0.00030778142173215465,
      "loss": 2.2265,
      "step": 113299
    },
    {
      "epoch": 1.48,
      "grad_norm": 1.0924564599990845,
      "learning_rate": 0.00030777733254552277,
      "loss": 2.1772,
      "step": 113300
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.9425532817840576,
      "learning_rate": 0.0003077732433574448,
      "loss": 2.2551,
      "step": 113301
    },
    {
      "epoch": 1.48,
      "grad_norm": 1.0237618684768677,
      "learning_rate": 0.00030776915416792165,
      "loss": 2.3022,
      "step": 113302
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.9457772374153137,
      "learning_rate": 0.00030776506497695415,
      "loss": 2.3311,
      "step": 113303
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.9422967433929443,
      "learning_rate": 0.00030776097578454287,
      "loss": 2.2264,
      "step": 113304
    },
    {
      "epoch": 1.48,
      "grad_norm": 1.122178077697754,
      "learning_rate": 0.00030775688659068874,
      "loss": 2.0399,
      "step": 113305
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.8451980948448181,
      "learning_rate": 0.0003077527973953925,
      "loss": 2.0741,
      "step": 113306
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.9784507751464844,
      "learning_rate": 0.0003077487081986548,
      "loss": 2.2439,
      "step": 113307
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.8904034495353699,
      "learning_rate": 0.0003077446190004765,
      "loss": 2.1559,
      "step": 113308
    },
    {
      "epoch": 1.48,
      "grad_norm": 1.568872094154358,
      "learning_rate": 0.0003077405298008584,
      "loss": 2.1419,
      "step": 113309
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.843542754650116,
      "learning_rate": 0.0003077364405998011,
      "loss": 2.0482,
      "step": 113310
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.9684678316116333,
      "learning_rate": 0.0003077323513973055,
      "loss": 2.3319,
      "step": 113311
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.9314445853233337,
      "learning_rate": 0.00030772826219337227,
      "loss": 2.3626,
      "step": 113312
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.8974325656890869,
      "learning_rate": 0.0003077241729880021,
      "loss": 2.3523,
      "step": 113313
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.9873945116996765,
      "learning_rate": 0.00030772008378119605,
      "loss": 2.0164,
      "step": 113314
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.933178186416626,
      "learning_rate": 0.0003077159945729546,
      "loss": 2.1533,
      "step": 113315
    },
    {
      "epoch": 1.48,
      "grad_norm": 1.0307040214538574,
      "learning_rate": 0.0003077119053632786,
      "loss": 1.8506,
      "step": 113316
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.9910544753074646,
      "learning_rate": 0.0003077078161521688,
      "loss": 2.1399,
      "step": 113317
    },
    {
      "epoch": 1.48,
      "grad_norm": 1.0214446783065796,
      "learning_rate": 0.0003077037269396261,
      "loss": 2.1606,
      "step": 113318
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.8660714030265808,
      "learning_rate": 0.000307699637725651,
      "loss": 2.1517,
      "step": 113319
    },
    {
      "epoch": 1.48,
      "grad_norm": 1.1834220886230469,
      "learning_rate": 0.00030769554851024436,
      "loss": 2.0777,
      "step": 113320
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.9694968461990356,
      "learning_rate": 0.0003076914592934071,
      "loss": 2.2492,
      "step": 113321
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.8636780381202698,
      "learning_rate": 0.00030768737007513976,
      "loss": 2.0958,
      "step": 113322
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.9052260518074036,
      "learning_rate": 0.00030768328085544323,
      "loss": 1.9973,
      "step": 113323
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.8973473906517029,
      "learning_rate": 0.00030767919163431827,
      "loss": 2.2556,
      "step": 113324
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.8932880163192749,
      "learning_rate": 0.0003076751024117656,
      "loss": 2.2651,
      "step": 113325
    },
    {
      "epoch": 1.48,
      "grad_norm": 1.0213032960891724,
      "learning_rate": 0.00030767101318778585,
      "loss": 2.3466,
      "step": 113326
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.8706877827644348,
      "learning_rate": 0.00030766692396238,
      "loss": 2.1925,
      "step": 113327
    },
    {
      "epoch": 1.48,
      "grad_norm": 1.000754952430725,
      "learning_rate": 0.00030766283473554875,
      "loss": 2.0823,
      "step": 113328
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.9677481055259705,
      "learning_rate": 0.0003076587455072928,
      "loss": 2.1945,
      "step": 113329
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.9585303664207458,
      "learning_rate": 0.000307654656277613,
      "loss": 2.1886,
      "step": 113330
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.8805420994758606,
      "learning_rate": 0.00030765056704651003,
      "loss": 2.1286,
      "step": 113331
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.8357776403427124,
      "learning_rate": 0.00030764647781398463,
      "loss": 1.9538,
      "step": 113332
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.9311240315437317,
      "learning_rate": 0.0003076423885800376,
      "loss": 2.0539,
      "step": 113333
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.9127575755119324,
      "learning_rate": 0.00030763829934466976,
      "loss": 2.0068,
      "step": 113334
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.8697004318237305,
      "learning_rate": 0.00030763421010788177,
      "loss": 2.0599,
      "step": 113335
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.9403048753738403,
      "learning_rate": 0.0003076301208696745,
      "loss": 2.1503,
      "step": 113336
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.8863911032676697,
      "learning_rate": 0.0003076260316300486,
      "loss": 2.1307,
      "step": 113337
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.9948405027389526,
      "learning_rate": 0.0003076219423890049,
      "loss": 2.2546,
      "step": 113338
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.9516998529434204,
      "learning_rate": 0.00030761785314654407,
      "loss": 2.1995,
      "step": 113339
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.8907915353775024,
      "learning_rate": 0.000307613763902667,
      "loss": 2.124,
      "step": 113340
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.8778069019317627,
      "learning_rate": 0.00030760967465737443,
      "loss": 1.7948,
      "step": 113341
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.9467166066169739,
      "learning_rate": 0.00030760558541066696,
      "loss": 2.2996,
      "step": 113342
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.8892274498939514,
      "learning_rate": 0.0003076014961625455,
      "loss": 1.9624,
      "step": 113343
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.8414570093154907,
      "learning_rate": 0.0003075974069130108,
      "loss": 1.9364,
      "step": 113344
    },
    {
      "epoch": 1.48,
      "grad_norm": 1.035225749015808,
      "learning_rate": 0.0003075933176620637,
      "loss": 2.0583,
      "step": 113345
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.9747022390365601,
      "learning_rate": 0.00030758922840970476,
      "loss": 2.1223,
      "step": 113346
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.9243444800376892,
      "learning_rate": 0.0003075851391559348,
      "loss": 2.0523,
      "step": 113347
    },
    {
      "epoch": 1.48,
      "grad_norm": 1.4717270135879517,
      "learning_rate": 0.0003075810499007547,
      "loss": 2.0887,
      "step": 113348
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.9691561460494995,
      "learning_rate": 0.00030757696064416507,
      "loss": 1.9435,
      "step": 113349
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.868569016456604,
      "learning_rate": 0.00030757287138616677,
      "loss": 2.1893,
      "step": 113350
    },
    {
      "epoch": 1.48,
      "grad_norm": 1.1753828525543213,
      "learning_rate": 0.0003075687821267605,
      "loss": 2.1775,
      "step": 113351
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.9920552968978882,
      "learning_rate": 0.0003075646928659472,
      "loss": 2.3108,
      "step": 113352
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.8990501761436462,
      "learning_rate": 0.00030756060360372735,
      "loss": 2.3884,
      "step": 113353
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.8768869042396545,
      "learning_rate": 0.0003075565143401018,
      "loss": 2.1405,
      "step": 113354
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.9515559077262878,
      "learning_rate": 0.00030755242507507144,
      "loss": 1.972,
      "step": 113355
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.9883653521537781,
      "learning_rate": 0.0003075483358086369,
      "loss": 2.1217,
      "step": 113356
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.8415176868438721,
      "learning_rate": 0.00030754424654079893,
      "loss": 2.0057,
      "step": 113357
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.8679394721984863,
      "learning_rate": 0.00030754015727155854,
      "loss": 2.1949,
      "step": 113358
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.8878650665283203,
      "learning_rate": 0.00030753606800091606,
      "loss": 2.0783,
      "step": 113359
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.8913446664810181,
      "learning_rate": 0.0003075319787288726,
      "loss": 2.2436,
      "step": 113360
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.9142211675643921,
      "learning_rate": 0.00030752788945542884,
      "loss": 2.2732,
      "step": 113361
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.9156135320663452,
      "learning_rate": 0.0003075238001805854,
      "loss": 2.1743,
      "step": 113362
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.8851101994514465,
      "learning_rate": 0.00030751971090434317,
      "loss": 2.3539,
      "step": 113363
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.9649059176445007,
      "learning_rate": 0.000307515621626703,
      "loss": 2.0574,
      "step": 113364
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.926211953163147,
      "learning_rate": 0.00030751153234766536,
      "loss": 2.1392,
      "step": 113365
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.9936069250106812,
      "learning_rate": 0.0003075074430672313,
      "loss": 2.0184,
      "step": 113366
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.8798961043357849,
      "learning_rate": 0.00030750335378540146,
      "loss": 2.2415,
      "step": 113367
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.9918165802955627,
      "learning_rate": 0.00030749926450217654,
      "loss": 2.0583,
      "step": 113368
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.9755709767341614,
      "learning_rate": 0.0003074951752175574,
      "loss": 1.7908,
      "step": 113369
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.8611701130867004,
      "learning_rate": 0.00030749108593154477,
      "loss": 2.1594,
      "step": 113370
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.9961293339729309,
      "learning_rate": 0.0003074869966441394,
      "loss": 2.0188,
      "step": 113371
    },
    {
      "epoch": 1.48,
      "grad_norm": 1.4025089740753174,
      "learning_rate": 0.00030748290735534205,
      "loss": 2.2707,
      "step": 113372
    },
    {
      "epoch": 1.48,
      "grad_norm": 1.0073505640029907,
      "learning_rate": 0.0003074788180651535,
      "loss": 2.1996,
      "step": 113373
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.8850976228713989,
      "learning_rate": 0.00030747472877357444,
      "loss": 2.4069,
      "step": 113374
    },
    {
      "epoch": 1.48,
      "grad_norm": 1.0895543098449707,
      "learning_rate": 0.00030747063948060574,
      "loss": 2.0913,
      "step": 113375
    },
    {
      "epoch": 1.48,
      "grad_norm": 1.2876139879226685,
      "learning_rate": 0.0003074665501862482,
      "loss": 2.004,
      "step": 113376
    },
    {
      "epoch": 1.48,
      "grad_norm": 1.0033220052719116,
      "learning_rate": 0.00030746246089050235,
      "loss": 2.3614,
      "step": 113377
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.958450436592102,
      "learning_rate": 0.00030745837159336907,
      "loss": 2.1837,
      "step": 113378
    },
    {
      "epoch": 1.48,
      "grad_norm": 1.0266711711883545,
      "learning_rate": 0.00030745428229484925,
      "loss": 2.0978,
      "step": 113379
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.9978634715080261,
      "learning_rate": 0.00030745019299494345,
      "loss": 2.3188,
      "step": 113380
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.9842133522033691,
      "learning_rate": 0.0003074461036936526,
      "loss": 2.2536,
      "step": 113381
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.8906786441802979,
      "learning_rate": 0.0003074420143909773,
      "loss": 2.0642,
      "step": 113382
    },
    {
      "epoch": 1.48,
      "grad_norm": 1.0746994018554688,
      "learning_rate": 0.0003074379250869184,
      "loss": 2.0401,
      "step": 113383
    },
    {
      "epoch": 1.48,
      "grad_norm": 1.0555377006530762,
      "learning_rate": 0.00030743383578147666,
      "loss": 2.1378,
      "step": 113384
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.886531412601471,
      "learning_rate": 0.0003074297464746529,
      "loss": 2.0579,
      "step": 113385
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.8505309820175171,
      "learning_rate": 0.00030742565716644773,
      "loss": 2.1201,
      "step": 113386
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.8722074031829834,
      "learning_rate": 0.00030742156785686195,
      "loss": 2.1555,
      "step": 113387
    },
    {
      "epoch": 1.48,
      "grad_norm": 1.0516777038574219,
      "learning_rate": 0.00030741747854589653,
      "loss": 2.1258,
      "step": 113388
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.9006138443946838,
      "learning_rate": 0.0003074133892335519,
      "loss": 2.1254,
      "step": 113389
    },
    {
      "epoch": 1.48,
      "grad_norm": 1.0453943014144897,
      "learning_rate": 0.000307409299919829,
      "loss": 2.1476,
      "step": 113390
    },
    {
      "epoch": 1.48,
      "grad_norm": 1.03550124168396,
      "learning_rate": 0.00030740521060472866,
      "loss": 2.225,
      "step": 113391
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.9385326504707336,
      "learning_rate": 0.0003074011212882515,
      "loss": 2.0982,
      "step": 113392
    },
    {
      "epoch": 1.48,
      "grad_norm": 1.044429063796997,
      "learning_rate": 0.0003073970319703983,
      "loss": 2.2904,
      "step": 113393
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.9714918732643127,
      "learning_rate": 0.0003073929426511699,
      "loss": 2.0565,
      "step": 113394
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.950758159160614,
      "learning_rate": 0.000307388853330567,
      "loss": 2.164,
      "step": 113395
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.9122273921966553,
      "learning_rate": 0.0003073847640085903,
      "loss": 2.0915,
      "step": 113396
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.9947473406791687,
      "learning_rate": 0.0003073806746852408,
      "loss": 2.137,
      "step": 113397
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.8566886782646179,
      "learning_rate": 0.00030737658536051895,
      "loss": 1.9796,
      "step": 113398
    },
    {
      "epoch": 1.48,
      "grad_norm": 1.0359073877334595,
      "learning_rate": 0.00030737249603442565,
      "loss": 2.1079,
      "step": 113399
    },
    {
      "epoch": 1.48,
      "grad_norm": 1.0463262796401978,
      "learning_rate": 0.00030736840670696165,
      "loss": 2.0184,
      "step": 113400
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.8749154806137085,
      "learning_rate": 0.0003073643173781278,
      "loss": 2.066,
      "step": 113401
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.9207082986831665,
      "learning_rate": 0.0003073602280479248,
      "loss": 2.1173,
      "step": 113402
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.9728010892868042,
      "learning_rate": 0.00030735613871635335,
      "loss": 2.0741,
      "step": 113403
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.8911005854606628,
      "learning_rate": 0.00030735204938341434,
      "loss": 2.0308,
      "step": 113404
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.9109827876091003,
      "learning_rate": 0.00030734796004910827,
      "loss": 1.8944,
      "step": 113405
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.9036489725112915,
      "learning_rate": 0.0003073438707134362,
      "loss": 2.129,
      "step": 113406
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.967162013053894,
      "learning_rate": 0.0003073397813763988,
      "loss": 2.3351,
      "step": 113407
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.9535468816757202,
      "learning_rate": 0.00030733569203799673,
      "loss": 2.2441,
      "step": 113408
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.9877943396568298,
      "learning_rate": 0.0003073316026982308,
      "loss": 2.3323,
      "step": 113409
    },
    {
      "epoch": 1.48,
      "grad_norm": 1.0389810800552368,
      "learning_rate": 0.0003073275133571018,
      "loss": 2.2036,
      "step": 113410
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.8738793730735779,
      "learning_rate": 0.00030732342401461045,
      "loss": 2.043,
      "step": 113411
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.9198296666145325,
      "learning_rate": 0.0003073193346707576,
      "loss": 2.1699,
      "step": 113412
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.9570754170417786,
      "learning_rate": 0.00030731524532554396,
      "loss": 2.2794,
      "step": 113413
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.8852483630180359,
      "learning_rate": 0.00030731115597897024,
      "loss": 1.9923,
      "step": 113414
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.989670991897583,
      "learning_rate": 0.0003073070666310372,
      "loss": 1.9278,
      "step": 113415
    },
    {
      "epoch": 1.48,
      "grad_norm": 1.0339717864990234,
      "learning_rate": 0.00030730297728174573,
      "loss": 2.1138,
      "step": 113416
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.9459620118141174,
      "learning_rate": 0.0003072988879310964,
      "loss": 2.1156,
      "step": 113417
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.952077329158783,
      "learning_rate": 0.0003072947985790901,
      "loss": 2.0661,
      "step": 113418
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.9619995951652527,
      "learning_rate": 0.0003072907092257277,
      "loss": 2.2113,
      "step": 113419
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.8432030081748962,
      "learning_rate": 0.0003072866198710096,
      "loss": 2.2408,
      "step": 113420
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.8858156800270081,
      "learning_rate": 0.0003072825305149368,
      "loss": 2.0714,
      "step": 113421
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.9501639008522034,
      "learning_rate": 0.00030727844115751023,
      "loss": 2.1645,
      "step": 113422
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.9603659510612488,
      "learning_rate": 0.0003072743517987303,
      "loss": 2.1792,
      "step": 113423
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.9894474744796753,
      "learning_rate": 0.000307270262438598,
      "loss": 1.9392,
      "step": 113424
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.915321946144104,
      "learning_rate": 0.00030726617307711394,
      "loss": 2.0952,
      "step": 113425
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.8924164772033691,
      "learning_rate": 0.00030726208371427904,
      "loss": 2.2663,
      "step": 113426
    },
    {
      "epoch": 1.48,
      "grad_norm": 1.0050630569458008,
      "learning_rate": 0.0003072579943500939,
      "loss": 2.0348,
      "step": 113427
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.9303164482116699,
      "learning_rate": 0.00030725390498455947,
      "loss": 2.1179,
      "step": 113428
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.9142875075340271,
      "learning_rate": 0.00030724981561767637,
      "loss": 2.1992,
      "step": 113429
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.8595033288002014,
      "learning_rate": 0.0003072457262494454,
      "loss": 2.2017,
      "step": 113430
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.9009090662002563,
      "learning_rate": 0.00030724163687986724,
      "loss": 2.0972,
      "step": 113431
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.8984069228172302,
      "learning_rate": 0.0003072375475089427,
      "loss": 2.2631,
      "step": 113432
    },
    {
      "epoch": 1.48,
      "grad_norm": 1.030129075050354,
      "learning_rate": 0.0003072334581366727,
      "loss": 2.2105,
      "step": 113433
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.8580452799797058,
      "learning_rate": 0.0003072293687630578,
      "loss": 1.8288,
      "step": 113434
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.9596258997917175,
      "learning_rate": 0.0003072252793880988,
      "loss": 2.3192,
      "step": 113435
    },
    {
      "epoch": 1.48,
      "grad_norm": 1.1392455101013184,
      "learning_rate": 0.0003072211900117965,
      "loss": 2.064,
      "step": 113436
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.8690572381019592,
      "learning_rate": 0.00030721710063415166,
      "loss": 1.9621,
      "step": 113437
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.9767611026763916,
      "learning_rate": 0.000307213011255165,
      "loss": 2.2221,
      "step": 113438
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.9151192903518677,
      "learning_rate": 0.0003072089218748373,
      "loss": 2.242,
      "step": 113439
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.9151168465614319,
      "learning_rate": 0.0003072048324931694,
      "loss": 2.0741,
      "step": 113440
    },
    {
      "epoch": 1.48,
      "grad_norm": 1.038837194442749,
      "learning_rate": 0.00030720074311016193,
      "loss": 2.0115,
      "step": 113441
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.8972587585449219,
      "learning_rate": 0.00030719665372581565,
      "loss": 2.0748,
      "step": 113442
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.8935765027999878,
      "learning_rate": 0.0003071925643401315,
      "loss": 2.0222,
      "step": 113443
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.8822295069694519,
      "learning_rate": 0.00030718847495311,
      "loss": 2.1715,
      "step": 113444
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.9144765734672546,
      "learning_rate": 0.00030718438556475206,
      "loss": 1.918,
      "step": 113445
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.972296416759491,
      "learning_rate": 0.00030718029617505843,
      "loss": 2.2421,
      "step": 113446
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.875188410282135,
      "learning_rate": 0.00030717620678402983,
      "loss": 2.1436,
      "step": 113447
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.9921144247055054,
      "learning_rate": 0.00030717211739166706,
      "loss": 2.2911,
      "step": 113448
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.9105960726737976,
      "learning_rate": 0.00030716802799797094,
      "loss": 2.3082,
      "step": 113449
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.9098156094551086,
      "learning_rate": 0.00030716393860294195,
      "loss": 2.2604,
      "step": 113450
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.9494021534919739,
      "learning_rate": 0.0003071598492065812,
      "loss": 2.1644,
      "step": 113451
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.8683708906173706,
      "learning_rate": 0.0003071557598088893,
      "loss": 2.0597,
      "step": 113452
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.9945055842399597,
      "learning_rate": 0.00030715167040986686,
      "loss": 2.0275,
      "step": 113453
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.9148163795471191,
      "learning_rate": 0.00030714758100951493,
      "loss": 2.0309,
      "step": 113454
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.9941552877426147,
      "learning_rate": 0.0003071434916078341,
      "loss": 1.9784,
      "step": 113455
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.9786021709442139,
      "learning_rate": 0.00030713940220482516,
      "loss": 2.0064,
      "step": 113456
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.9405345916748047,
      "learning_rate": 0.0003071353128004888,
      "loss": 2.1276,
      "step": 113457
    },
    {
      "epoch": 1.48,
      "grad_norm": 1.0037387609481812,
      "learning_rate": 0.00030713122339482595,
      "loss": 2.3418,
      "step": 113458
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.9156951308250427,
      "learning_rate": 0.00030712713398783724,
      "loss": 1.9902,
      "step": 113459
    },
    {
      "epoch": 1.48,
      "grad_norm": 1.0305286645889282,
      "learning_rate": 0.0003071230445795234,
      "loss": 2.1436,
      "step": 113460
    },
    {
      "epoch": 1.48,
      "grad_norm": 1.037041187286377,
      "learning_rate": 0.00030711895516988535,
      "loss": 2.1137,
      "step": 113461
    },
    {
      "epoch": 1.48,
      "grad_norm": 1.1195037364959717,
      "learning_rate": 0.0003071148657589237,
      "loss": 2.2075,
      "step": 113462
    },
    {
      "epoch": 1.48,
      "grad_norm": 1.0253387689590454,
      "learning_rate": 0.0003071107763466393,
      "loss": 2.0294,
      "step": 113463
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.8396984338760376,
      "learning_rate": 0.0003071066869330329,
      "loss": 2.2446,
      "step": 113464
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.918382465839386,
      "learning_rate": 0.0003071025975181051,
      "loss": 2.0248,
      "step": 113465
    },
    {
      "epoch": 1.48,
      "grad_norm": 1.1699903011322021,
      "learning_rate": 0.0003070985081018568,
      "loss": 2.2691,
      "step": 113466
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.9281124472618103,
      "learning_rate": 0.00030709441868428884,
      "loss": 2.2336,
      "step": 113467
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.8966413736343384,
      "learning_rate": 0.00030709032926540193,
      "loss": 2.0585,
      "step": 113468
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.8957323431968689,
      "learning_rate": 0.00030708623984519674,
      "loss": 2.1963,
      "step": 113469
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.97017902135849,
      "learning_rate": 0.00030708215042367403,
      "loss": 2.0948,
      "step": 113470
    },
    {
      "epoch": 1.48,
      "grad_norm": 1.1476936340332031,
      "learning_rate": 0.0003070780610008347,
      "loss": 2.2174,
      "step": 113471
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.8651292324066162,
      "learning_rate": 0.0003070739715766794,
      "loss": 2.1767,
      "step": 113472
    },
    {
      "epoch": 1.48,
      "grad_norm": 1.0397491455078125,
      "learning_rate": 0.0003070698821512089,
      "loss": 2.2643,
      "step": 113473
    },
    {
      "epoch": 1.48,
      "grad_norm": 1.1131091117858887,
      "learning_rate": 0.000307065792724424,
      "loss": 2.2372,
      "step": 113474
    },
    {
      "epoch": 1.48,
      "grad_norm": 1.0368112325668335,
      "learning_rate": 0.0003070617032963254,
      "loss": 1.9853,
      "step": 113475
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.9849116802215576,
      "learning_rate": 0.00030705761386691386,
      "loss": 2.0573,
      "step": 113476
    },
    {
      "epoch": 1.48,
      "grad_norm": 1.0570292472839355,
      "learning_rate": 0.0003070535244361903,
      "loss": 2.0656,
      "step": 113477
    },
    {
      "epoch": 1.48,
      "grad_norm": 1.0029624700546265,
      "learning_rate": 0.0003070494350041553,
      "loss": 2.1983,
      "step": 113478
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.9289631247520447,
      "learning_rate": 0.0003070453455708096,
      "loss": 2.2797,
      "step": 113479
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.8880773186683655,
      "learning_rate": 0.00030704125613615415,
      "loss": 2.2077,
      "step": 113480
    },
    {
      "epoch": 1.48,
      "grad_norm": 1.087570071220398,
      "learning_rate": 0.0003070371667001896,
      "loss": 2.2782,
      "step": 113481
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.9237616062164307,
      "learning_rate": 0.0003070330772629166,
      "loss": 2.2333,
      "step": 113482
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.8554540276527405,
      "learning_rate": 0.0003070289878243361,
      "loss": 1.9653,
      "step": 113483
    },
    {
      "epoch": 1.48,
      "grad_norm": 1.061201810836792,
      "learning_rate": 0.00030702489838444874,
      "loss": 2.0168,
      "step": 113484
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.9646334648132324,
      "learning_rate": 0.0003070208089432553,
      "loss": 2.0937,
      "step": 113485
    },
    {
      "epoch": 1.48,
      "grad_norm": 1.0007389783859253,
      "learning_rate": 0.0003070167195007566,
      "loss": 2.2675,
      "step": 113486
    },
    {
      "epoch": 1.48,
      "grad_norm": 1.0579525232315063,
      "learning_rate": 0.00030701263005695334,
      "loss": 2.1417,
      "step": 113487
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.956151008605957,
      "learning_rate": 0.00030700854061184636,
      "loss": 2.08,
      "step": 113488
    },
    {
      "epoch": 1.48,
      "grad_norm": 1.0182026624679565,
      "learning_rate": 0.0003070044511654363,
      "loss": 2.3741,
      "step": 113489
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.9640573263168335,
      "learning_rate": 0.00030700036171772395,
      "loss": 2.1759,
      "step": 113490
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.849754273891449,
      "learning_rate": 0.00030699627226871015,
      "loss": 2.1399,
      "step": 113491
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.990021288394928,
      "learning_rate": 0.00030699218281839564,
      "loss": 2.0063,
      "step": 113492
    },
    {
      "epoch": 1.48,
      "grad_norm": 1.1045554876327515,
      "learning_rate": 0.0003069880933667811,
      "loss": 2.1248,
      "step": 113493
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.9048200845718384,
      "learning_rate": 0.00030698400391386733,
      "loss": 2.0936,
      "step": 113494
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.9203184843063354,
      "learning_rate": 0.0003069799144596552,
      "loss": 2.3003,
      "step": 113495
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.8888938426971436,
      "learning_rate": 0.00030697582500414526,
      "loss": 2.1021,
      "step": 113496
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.8528760671615601,
      "learning_rate": 0.0003069717355473384,
      "loss": 2.3768,
      "step": 113497
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.9562544226646423,
      "learning_rate": 0.00030696764608923545,
      "loss": 2.1031,
      "step": 113498
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.9178994297981262,
      "learning_rate": 0.000306963556629837,
      "loss": 2.1777,
      "step": 113499
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.9197922348976135,
      "learning_rate": 0.0003069594671691439,
      "loss": 2.5362,
      "step": 113500
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.8382686376571655,
      "learning_rate": 0.0003069553777071569,
      "loss": 2.0208,
      "step": 113501
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.886870265007019,
      "learning_rate": 0.00030695128824387676,
      "loss": 2.1183,
      "step": 113502
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.9190245270729065,
      "learning_rate": 0.0003069471987793043,
      "loss": 2.1837,
      "step": 113503
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.9843270778656006,
      "learning_rate": 0.0003069431093134402,
      "loss": 1.9357,
      "step": 113504
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.9370266795158386,
      "learning_rate": 0.0003069390198462852,
      "loss": 2.288,
      "step": 113505
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.9853132367134094,
      "learning_rate": 0.0003069349303778401,
      "loss": 2.2259,
      "step": 113506
    },
    {
      "epoch": 1.48,
      "grad_norm": 1.0753897428512573,
      "learning_rate": 0.00030693084090810575,
      "loss": 2.1989,
      "step": 113507
    },
    {
      "epoch": 1.48,
      "grad_norm": 1.0006020069122314,
      "learning_rate": 0.0003069267514370828,
      "loss": 2.3314,
      "step": 113508
    },
    {
      "epoch": 1.48,
      "grad_norm": 1.024276614189148,
      "learning_rate": 0.00030692266196477195,
      "loss": 2.2685,
      "step": 113509
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.9422636032104492,
      "learning_rate": 0.0003069185724911742,
      "loss": 2.0533,
      "step": 113510
    },
    {
      "epoch": 1.48,
      "grad_norm": 1.0289318561553955,
      "learning_rate": 0.00030691448301629,
      "loss": 2.2608,
      "step": 113511
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.8645623922348022,
      "learning_rate": 0.0003069103935401204,
      "loss": 2.1162,
      "step": 113512
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.9507094621658325,
      "learning_rate": 0.0003069063040626659,
      "loss": 2.1871,
      "step": 113513
    },
    {
      "epoch": 1.48,
      "grad_norm": 1.0369194746017456,
      "learning_rate": 0.00030690221458392746,
      "loss": 2.393,
      "step": 113514
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.9492326378822327,
      "learning_rate": 0.00030689812510390575,
      "loss": 2.1867,
      "step": 113515
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.9057322144508362,
      "learning_rate": 0.0003068940356226016,
      "loss": 1.9718,
      "step": 113516
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.8728265762329102,
      "learning_rate": 0.0003068899461400156,
      "loss": 2.1265,
      "step": 113517
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.9733418226242065,
      "learning_rate": 0.00030688585665614875,
      "loss": 2.1653,
      "step": 113518
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.9600777626037598,
      "learning_rate": 0.0003068817671710016,
      "loss": 2.0081,
      "step": 113519
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.9060373902320862,
      "learning_rate": 0.00030687767768457507,
      "loss": 2.0615,
      "step": 113520
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.8312416672706604,
      "learning_rate": 0.00030687358819686983,
      "loss": 2.0252,
      "step": 113521
    },
    {
      "epoch": 1.48,
      "grad_norm": 1.0418564081192017,
      "learning_rate": 0.00030686949870788665,
      "loss": 2.1497,
      "step": 113522
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.9976580739021301,
      "learning_rate": 0.0003068654092176263,
      "loss": 2.2186,
      "step": 113523
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.9306021332740784,
      "learning_rate": 0.0003068613197260895,
      "loss": 2.1691,
      "step": 113524
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.976134717464447,
      "learning_rate": 0.00030685723023327717,
      "loss": 2.3256,
      "step": 113525
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.9771803021430969,
      "learning_rate": 0.00030685314073918983,
      "loss": 2.2019,
      "step": 113526
    },
    {
      "epoch": 1.48,
      "grad_norm": 1.0013717412948608,
      "learning_rate": 0.0003068490512438284,
      "loss": 2.1091,
      "step": 113527
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.8954657316207886,
      "learning_rate": 0.0003068449617471937,
      "loss": 1.9739,
      "step": 113528
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.8933643102645874,
      "learning_rate": 0.0003068408722492862,
      "loss": 2.2203,
      "step": 113529
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.9152101278305054,
      "learning_rate": 0.000306836782750107,
      "loss": 2.1045,
      "step": 113530
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.8273274302482605,
      "learning_rate": 0.0003068326932496566,
      "loss": 2.0578,
      "step": 113531
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.94319748878479,
      "learning_rate": 0.00030682860374793596,
      "loss": 2.1339,
      "step": 113532
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.9116577506065369,
      "learning_rate": 0.00030682451424494575,
      "loss": 2.1078,
      "step": 113533
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.9165437817573547,
      "learning_rate": 0.0003068204247406867,
      "loss": 1.863,
      "step": 113534
    },
    {
      "epoch": 1.48,
      "grad_norm": 1.0242414474487305,
      "learning_rate": 0.0003068163352351597,
      "loss": 2.0806,
      "step": 113535
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.8451217412948608,
      "learning_rate": 0.0003068122457283653,
      "loss": 2.1651,
      "step": 113536
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.9579242467880249,
      "learning_rate": 0.0003068081562203044,
      "loss": 2.1733,
      "step": 113537
    },
    {
      "epoch": 1.48,
      "grad_norm": 1.0668871402740479,
      "learning_rate": 0.0003068040667109778,
      "loss": 2.1494,
      "step": 113538
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.9159581661224365,
      "learning_rate": 0.0003067999772003861,
      "loss": 2.0018,
      "step": 113539
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.9603475332260132,
      "learning_rate": 0.0003067958876885302,
      "loss": 2.3596,
      "step": 113540
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.9108549356460571,
      "learning_rate": 0.00030679179817541085,
      "loss": 2.3164,
      "step": 113541
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.9493983387947083,
      "learning_rate": 0.0003067877086610287,
      "loss": 2.1714,
      "step": 113542
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.8592245578765869,
      "learning_rate": 0.0003067836191453846,
      "loss": 2.1075,
      "step": 113543
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.8207433223724365,
      "learning_rate": 0.00030677952962847944,
      "loss": 2.1533,
      "step": 113544
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.9551612734794617,
      "learning_rate": 0.00030677544011031366,
      "loss": 2.2602,
      "step": 113545
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.8869116902351379,
      "learning_rate": 0.0003067713505908882,
      "loss": 1.971,
      "step": 113546
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.9960906505584717,
      "learning_rate": 0.000306767261070204,
      "loss": 2.1808,
      "step": 113547
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.9116349816322327,
      "learning_rate": 0.0003067631715482614,
      "loss": 2.0272,
      "step": 113548
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.8432486653327942,
      "learning_rate": 0.0003067590820250616,
      "loss": 2.3741,
      "step": 113549
    },
    {
      "epoch": 1.48,
      "grad_norm": 1.0181761980056763,
      "learning_rate": 0.0003067549925006051,
      "loss": 2.1202,
      "step": 113550
    },
    {
      "epoch": 1.48,
      "grad_norm": 1.2134833335876465,
      "learning_rate": 0.00030675090297489267,
      "loss": 2.0078,
      "step": 113551
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.8998677730560303,
      "learning_rate": 0.0003067468134479251,
      "loss": 2.1005,
      "step": 113552
    },
    {
      "epoch": 1.48,
      "grad_norm": 1.6591562032699585,
      "learning_rate": 0.0003067427239197033,
      "loss": 2.0938,
      "step": 113553
    },
    {
      "epoch": 1.48,
      "grad_norm": 1.0483589172363281,
      "learning_rate": 0.00030673863439022777,
      "loss": 2.1211,
      "step": 113554
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.8862305283546448,
      "learning_rate": 0.00030673454485949945,
      "loss": 2.1107,
      "step": 113555
    },
    {
      "epoch": 1.48,
      "grad_norm": 1.0980287790298462,
      "learning_rate": 0.00030673045532751903,
      "loss": 2.2858,
      "step": 113556
    },
    {
      "epoch": 1.48,
      "grad_norm": 1.314251184463501,
      "learning_rate": 0.00030672636579428733,
      "loss": 1.9347,
      "step": 113557
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.9704973101615906,
      "learning_rate": 0.000306722276259805,
      "loss": 2.0323,
      "step": 113558
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.9833519458770752,
      "learning_rate": 0.00030671818672407304,
      "loss": 2.1785,
      "step": 113559
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.9911882281303406,
      "learning_rate": 0.0003067140971870919,
      "loss": 2.1783,
      "step": 113560
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.9564257264137268,
      "learning_rate": 0.00030671000764886247,
      "loss": 2.2626,
      "step": 113561
    },
    {
      "epoch": 1.48,
      "grad_norm": 1.0905994176864624,
      "learning_rate": 0.0003067059181093856,
      "loss": 2.2312,
      "step": 113562
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.8512719869613647,
      "learning_rate": 0.0003067018285686619,
      "loss": 2.1723,
      "step": 113563
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.8592871427536011,
      "learning_rate": 0.0003066977390266922,
      "loss": 1.8861,
      "step": 113564
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.9751335382461548,
      "learning_rate": 0.0003066936494834774,
      "loss": 1.9295,
      "step": 113565
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.8885462284088135,
      "learning_rate": 0.00030668955993901797,
      "loss": 2.2636,
      "step": 113566
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.9584610462188721,
      "learning_rate": 0.00030668547039331486,
      "loss": 2.1504,
      "step": 113567
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.8720060586929321,
      "learning_rate": 0.0003066813808463689,
      "loss": 2.0532,
      "step": 113568
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.9993047714233398,
      "learning_rate": 0.0003066772912981806,
      "loss": 2.0995,
      "step": 113569
    },
    {
      "epoch": 1.48,
      "grad_norm": 1.1565237045288086,
      "learning_rate": 0.0003066732017487509,
      "loss": 2.2316,
      "step": 113570
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.9178962111473083,
      "learning_rate": 0.0003066691121980806,
      "loss": 2.3549,
      "step": 113571
    },
    {
      "epoch": 1.48,
      "grad_norm": 1.136718988418579,
      "learning_rate": 0.00030666502264617033,
      "loss": 2.1035,
      "step": 113572
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.9333348870277405,
      "learning_rate": 0.00030666093309302087,
      "loss": 2.0362,
      "step": 113573
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.8654988408088684,
      "learning_rate": 0.0003066568435386331,
      "loss": 2.2484,
      "step": 113574
    },
    {
      "epoch": 1.48,
      "grad_norm": 1.1478205919265747,
      "learning_rate": 0.0003066527539830077,
      "loss": 2.4537,
      "step": 113575
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.9365850687026978,
      "learning_rate": 0.0003066486644261453,
      "loss": 2.2501,
      "step": 113576
    },
    {
      "epoch": 1.48,
      "grad_norm": 1.0149494409561157,
      "learning_rate": 0.00030664457486804695,
      "loss": 2.2977,
      "step": 113577
    },
    {
      "epoch": 1.48,
      "grad_norm": 1.0106428861618042,
      "learning_rate": 0.00030664048530871314,
      "loss": 2.0622,
      "step": 113578
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.8840740919113159,
      "learning_rate": 0.00030663639574814475,
      "loss": 2.1988,
      "step": 113579
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.8325671553611755,
      "learning_rate": 0.0003066323061863426,
      "loss": 2.2495,
      "step": 113580
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.8979701399803162,
      "learning_rate": 0.0003066282166233073,
      "loss": 2.1661,
      "step": 113581
    },
    {
      "epoch": 1.48,
      "grad_norm": 1.0089904069900513,
      "learning_rate": 0.0003066241270590397,
      "loss": 1.8926,
      "step": 113582
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.9752750396728516,
      "learning_rate": 0.00030662003749354057,
      "loss": 1.8166,
      "step": 113583
    },
    {
      "epoch": 1.48,
      "grad_norm": 1.0157585144042969,
      "learning_rate": 0.00030661594792681066,
      "loss": 2.1244,
      "step": 113584
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.8961666226387024,
      "learning_rate": 0.0003066118583588507,
      "loss": 1.7773,
      "step": 113585
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.9870156645774841,
      "learning_rate": 0.0003066077687896615,
      "loss": 2.0667,
      "step": 113586
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.9343002438545227,
      "learning_rate": 0.0003066036792192438,
      "loss": 1.9451,
      "step": 113587
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.9279614686965942,
      "learning_rate": 0.00030659958964759823,
      "loss": 2.1693,
      "step": 113588
    },
    {
      "epoch": 1.48,
      "grad_norm": 1.1585239171981812,
      "learning_rate": 0.0003065955000747258,
      "loss": 2.0861,
      "step": 113589
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.9323320984840393,
      "learning_rate": 0.00030659141050062714,
      "loss": 2.1857,
      "step": 113590
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.9274145364761353,
      "learning_rate": 0.00030658732092530297,
      "loss": 2.2576,
      "step": 113591
    },
    {
      "epoch": 1.48,
      "grad_norm": 1.1931437253952026,
      "learning_rate": 0.0003065832313487541,
      "loss": 2.1231,
      "step": 113592
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.8844775557518005,
      "learning_rate": 0.00030657914177098126,
      "loss": 2.2542,
      "step": 113593
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.9056155681610107,
      "learning_rate": 0.0003065750521919853,
      "loss": 2.2236,
      "step": 113594
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.9866431355476379,
      "learning_rate": 0.00030657096261176687,
      "loss": 2.362,
      "step": 113595
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.9721384048461914,
      "learning_rate": 0.0003065668730303267,
      "loss": 2.031,
      "step": 113596
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.9488517045974731,
      "learning_rate": 0.00030656278344766576,
      "loss": 2.2001,
      "step": 113597
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.8402281999588013,
      "learning_rate": 0.00030655869386378453,
      "loss": 2.1544,
      "step": 113598
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.9782745242118835,
      "learning_rate": 0.0003065546042786841,
      "loss": 2.1663,
      "step": 113599
    },
    {
      "epoch": 1.48,
      "grad_norm": 1.008744478225708,
      "learning_rate": 0.0003065505146923649,
      "loss": 2.107,
      "step": 113600
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.9497613310813904,
      "learning_rate": 0.00030654642510482785,
      "loss": 2.1875,
      "step": 113601
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.8851916790008545,
      "learning_rate": 0.0003065423355160738,
      "loss": 2.2223,
      "step": 113602
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.9884501099586487,
      "learning_rate": 0.00030653824592610324,
      "loss": 2.0278,
      "step": 113603
    },
    {
      "epoch": 1.48,
      "grad_norm": 1.0009422302246094,
      "learning_rate": 0.00030653415633491725,
      "loss": 2.221,
      "step": 113604
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.9183985590934753,
      "learning_rate": 0.0003065300667425163,
      "loss": 2.297,
      "step": 113605
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.9427083134651184,
      "learning_rate": 0.00030652597714890144,
      "loss": 2.3141,
      "step": 113606
    },
    {
      "epoch": 1.48,
      "grad_norm": 1.0475497245788574,
      "learning_rate": 0.0003065218875540732,
      "loss": 2.1716,
      "step": 113607
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.9622966051101685,
      "learning_rate": 0.00030651779795803235,
      "loss": 2.084,
      "step": 113608
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.9028198719024658,
      "learning_rate": 0.0003065137083607798,
      "loss": 2.2155,
      "step": 113609
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.8810485005378723,
      "learning_rate": 0.00030650961876231624,
      "loss": 2.1873,
      "step": 113610
    },
    {
      "epoch": 1.48,
      "grad_norm": 1.092055320739746,
      "learning_rate": 0.0003065055291626423,
      "loss": 2.1517,
      "step": 113611
    },
    {
      "epoch": 1.48,
      "grad_norm": 1.313451886177063,
      "learning_rate": 0.000306501439561759,
      "loss": 2.1643,
      "step": 113612
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.9092296361923218,
      "learning_rate": 0.00030649734995966696,
      "loss": 2.4057,
      "step": 113613
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.9510975480079651,
      "learning_rate": 0.00030649326035636686,
      "loss": 2.0134,
      "step": 113614
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.9861380457878113,
      "learning_rate": 0.00030648917075185955,
      "loss": 2.0981,
      "step": 113615
    },
    {
      "epoch": 1.48,
      "grad_norm": 1.0065029859542847,
      "learning_rate": 0.00030648508114614574,
      "loss": 2.2524,
      "step": 113616
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.8696300983428955,
      "learning_rate": 0.0003064809915392264,
      "loss": 2.1536,
      "step": 113617
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.9024463891983032,
      "learning_rate": 0.000306476901931102,
      "loss": 2.2058,
      "step": 113618
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.9429427981376648,
      "learning_rate": 0.0003064728123217734,
      "loss": 2.0368,
      "step": 113619
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.9791876673698425,
      "learning_rate": 0.00030646872271124145,
      "loss": 1.9472,
      "step": 113620
    },
    {
      "epoch": 1.48,
      "grad_norm": 1.103067398071289,
      "learning_rate": 0.0003064646330995068,
      "loss": 2.0468,
      "step": 113621
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.8813495635986328,
      "learning_rate": 0.0003064605434865702,
      "loss": 2.0852,
      "step": 113622
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.8771939277648926,
      "learning_rate": 0.0003064564538724326,
      "loss": 2.1474,
      "step": 113623
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.9238117933273315,
      "learning_rate": 0.00030645236425709457,
      "loss": 2.2444,
      "step": 113624
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.9032951593399048,
      "learning_rate": 0.0003064482746405569,
      "loss": 2.3995,
      "step": 113625
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.8662120699882507,
      "learning_rate": 0.00030644418502282037,
      "loss": 2.1813,
      "step": 113626
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.9147171974182129,
      "learning_rate": 0.00030644009540388575,
      "loss": 2.4092,
      "step": 113627
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.9182253479957581,
      "learning_rate": 0.0003064360057837538,
      "loss": 2.0816,
      "step": 113628
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.8974371552467346,
      "learning_rate": 0.00030643191616242524,
      "loss": 2.2599,
      "step": 113629
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.8996497988700867,
      "learning_rate": 0.0003064278265399009,
      "loss": 2.2821,
      "step": 113630
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.9348300695419312,
      "learning_rate": 0.0003064237369161815,
      "loss": 2.2316,
      "step": 113631
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.9441325664520264,
      "learning_rate": 0.00030641964729126784,
      "loss": 2.0675,
      "step": 113632
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.9050593376159668,
      "learning_rate": 0.0003064155576651606,
      "loss": 2.0912,
      "step": 113633
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.8866119384765625,
      "learning_rate": 0.0003064114680378606,
      "loss": 2.0509,
      "step": 113634
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.9182999134063721,
      "learning_rate": 0.00030640737840936864,
      "loss": 2.1547,
      "step": 113635
    },
    {
      "epoch": 1.48,
      "grad_norm": 1.1694937944412231,
      "learning_rate": 0.00030640328877968535,
      "loss": 2.0468,
      "step": 113636
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.9637701511383057,
      "learning_rate": 0.0003063991991488116,
      "loss": 2.1582,
      "step": 113637
    },
    {
      "epoch": 1.48,
      "grad_norm": 1.0396597385406494,
      "learning_rate": 0.0003063951095167481,
      "loss": 2.2025,
      "step": 113638
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.9978132247924805,
      "learning_rate": 0.00030639101988349566,
      "loss": 2.1695,
      "step": 113639
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.92448890209198,
      "learning_rate": 0.00030638693024905494,
      "loss": 2.1874,
      "step": 113640
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.8786538243293762,
      "learning_rate": 0.00030638284061342686,
      "loss": 2.0511,
      "step": 113641
    },
    {
      "epoch": 1.48,
      "grad_norm": 1.0423957109451294,
      "learning_rate": 0.0003063787509766121,
      "loss": 2.2122,
      "step": 113642
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.9137816429138184,
      "learning_rate": 0.0003063746613386113,
      "loss": 2.1086,
      "step": 113643
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.9166789650917053,
      "learning_rate": 0.0003063705716994254,
      "loss": 2.1959,
      "step": 113644
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.8651161193847656,
      "learning_rate": 0.00030636648205905514,
      "loss": 2.1564,
      "step": 113645
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.9068326354026794,
      "learning_rate": 0.00030636239241750113,
      "loss": 2.2866,
      "step": 113646
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.9089158773422241,
      "learning_rate": 0.00030635830277476425,
      "loss": 2.1481,
      "step": 113647
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.9463871717453003,
      "learning_rate": 0.0003063542131308453,
      "loss": 2.0482,
      "step": 113648
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.8618935346603394,
      "learning_rate": 0.0003063501234857449,
      "loss": 2.1636,
      "step": 113649
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.9028151035308838,
      "learning_rate": 0.00030634603383946404,
      "loss": 2.2563,
      "step": 113650
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.9315352439880371,
      "learning_rate": 0.00030634194419200313,
      "loss": 2.3044,
      "step": 113651
    },
    {
      "epoch": 1.48,
      "grad_norm": 1.036154866218567,
      "learning_rate": 0.00030633785454336327,
      "loss": 2.1475,
      "step": 113652
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.8470961451530457,
      "learning_rate": 0.0003063337648935451,
      "loss": 2.0056,
      "step": 113653
    },
    {
      "epoch": 1.48,
      "grad_norm": 1.0315929651260376,
      "learning_rate": 0.0003063296752425492,
      "loss": 2.2205,
      "step": 113654
    },
    {
      "epoch": 1.48,
      "grad_norm": 1.1471307277679443,
      "learning_rate": 0.0003063255855903766,
      "loss": 2.0635,
      "step": 113655
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.8427866101264954,
      "learning_rate": 0.000306321495937028,
      "loss": 2.1388,
      "step": 113656
    },
    {
      "epoch": 1.48,
      "grad_norm": 1.011106014251709,
      "learning_rate": 0.0003063174062825041,
      "loss": 2.1927,
      "step": 113657
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.9457045793533325,
      "learning_rate": 0.0003063133166268056,
      "loss": 2.1012,
      "step": 113658
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.9748605489730835,
      "learning_rate": 0.00030630922696993346,
      "loss": 2.1095,
      "step": 113659
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.8891709446907043,
      "learning_rate": 0.0003063051373118882,
      "loss": 2.2511,
      "step": 113660
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.8594663143157959,
      "learning_rate": 0.0003063010476526707,
      "loss": 1.9613,
      "step": 113661
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.9474273324012756,
      "learning_rate": 0.0003062969579922818,
      "loss": 2.2342,
      "step": 113662
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.9228438138961792,
      "learning_rate": 0.0003062928683307221,
      "loss": 2.064,
      "step": 113663
    },
    {
      "epoch": 1.48,
      "grad_norm": 1.003734827041626,
      "learning_rate": 0.00030628877866799244,
      "loss": 2.2634,
      "step": 113664
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.9072622060775757,
      "learning_rate": 0.0003062846890040936,
      "loss": 2.1681,
      "step": 113665
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.9990629553794861,
      "learning_rate": 0.00030628059933902635,
      "loss": 2.1977,
      "step": 113666
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.9193337559700012,
      "learning_rate": 0.00030627650967279135,
      "loss": 2.3097,
      "step": 113667
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.8532553315162659,
      "learning_rate": 0.0003062724200053894,
      "loss": 2.068,
      "step": 113668
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.9467693567276001,
      "learning_rate": 0.00030626833033682143,
      "loss": 2.2255,
      "step": 113669
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.9402328729629517,
      "learning_rate": 0.000306264240667088,
      "loss": 2.2055,
      "step": 113670
    },
    {
      "epoch": 1.48,
      "grad_norm": 1.0480332374572754,
      "learning_rate": 0.0003062601509961899,
      "loss": 2.3504,
      "step": 113671
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.8896910548210144,
      "learning_rate": 0.000306256061324128,
      "loss": 2.2012,
      "step": 113672
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.9437573552131653,
      "learning_rate": 0.0003062519716509028,
      "loss": 2.0074,
      "step": 113673
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.979716420173645,
      "learning_rate": 0.00030624788197651536,
      "loss": 2.256,
      "step": 113674
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.9532042741775513,
      "learning_rate": 0.00030624379230096636,
      "loss": 2.1108,
      "step": 113675
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.9507200121879578,
      "learning_rate": 0.00030623970262425645,
      "loss": 2.2081,
      "step": 113676
    },
    {
      "epoch": 1.48,
      "grad_norm": 1.0731154680252075,
      "learning_rate": 0.0003062356129463865,
      "loss": 2.2047,
      "step": 113677
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.8713734149932861,
      "learning_rate": 0.00030623152326735725,
      "loss": 2.2665,
      "step": 113678
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.9820783138275146,
      "learning_rate": 0.0003062274335871694,
      "loss": 2.114,
      "step": 113679
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.893008828163147,
      "learning_rate": 0.00030622334390582374,
      "loss": 1.8518,
      "step": 113680
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.8530466556549072,
      "learning_rate": 0.0003062192542233211,
      "loss": 2.0721,
      "step": 113681
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.8852878212928772,
      "learning_rate": 0.0003062151645396621,
      "loss": 2.2799,
      "step": 113682
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.9484024047851562,
      "learning_rate": 0.0003062110748548476,
      "loss": 2.2099,
      "step": 113683
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.9592838287353516,
      "learning_rate": 0.0003062069851688785,
      "loss": 2.2032,
      "step": 113684
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.9211328625679016,
      "learning_rate": 0.0003062028954817552,
      "loss": 2.3225,
      "step": 113685
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.8649215698242188,
      "learning_rate": 0.00030619880579347875,
      "loss": 2.1919,
      "step": 113686
    },
    {
      "epoch": 1.48,
      "grad_norm": 1.0394381284713745,
      "learning_rate": 0.00030619471610404984,
      "loss": 1.9766,
      "step": 113687
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.8727893233299255,
      "learning_rate": 0.0003061906264134692,
      "loss": 2.0417,
      "step": 113688
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.9379622936248779,
      "learning_rate": 0.0003061865367217376,
      "loss": 2.3625,
      "step": 113689
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.865129292011261,
      "learning_rate": 0.00030618244702885583,
      "loss": 2.2087,
      "step": 113690
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.9765331149101257,
      "learning_rate": 0.00030617835733482454,
      "loss": 2.2702,
      "step": 113691
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.8552746176719666,
      "learning_rate": 0.00030617426763964464,
      "loss": 1.8888,
      "step": 113692
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.9983862042427063,
      "learning_rate": 0.00030617017794331685,
      "loss": 2.2677,
      "step": 113693
    },
    {
      "epoch": 1.48,
      "grad_norm": 1.041049599647522,
      "learning_rate": 0.0003061660882458419,
      "loss": 2.077,
      "step": 113694
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.8927402496337891,
      "learning_rate": 0.0003061619985472205,
      "loss": 2.0919,
      "step": 113695
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.818427324295044,
      "learning_rate": 0.0003061579088474535,
      "loss": 2.0659,
      "step": 113696
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.8735796213150024,
      "learning_rate": 0.00030615381914654165,
      "loss": 1.9921,
      "step": 113697
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.8812315464019775,
      "learning_rate": 0.0003061497294444857,
      "loss": 2.1037,
      "step": 113698
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.8757189512252808,
      "learning_rate": 0.00030614563974128633,
      "loss": 2.2077,
      "step": 113699
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.9766337871551514,
      "learning_rate": 0.00030614155003694444,
      "loss": 2.4598,
      "step": 113700
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.9562499523162842,
      "learning_rate": 0.00030613746033146067,
      "loss": 2.2388,
      "step": 113701
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.9106866717338562,
      "learning_rate": 0.0003061333706248359,
      "loss": 2.143,
      "step": 113702
    },
    {
      "epoch": 1.48,
      "grad_norm": 1.088417887687683,
      "learning_rate": 0.0003061292809170708,
      "loss": 2.2239,
      "step": 113703
    },
    {
      "epoch": 1.48,
      "grad_norm": 1.0155820846557617,
      "learning_rate": 0.00030612519120816603,
      "loss": 2.2295,
      "step": 113704
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.9610918164253235,
      "learning_rate": 0.0003061211014981226,
      "loss": 2.031,
      "step": 113705
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.9458864331245422,
      "learning_rate": 0.0003061170117869411,
      "loss": 1.9305,
      "step": 113706
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.8710533976554871,
      "learning_rate": 0.00030611292207462237,
      "loss": 2.1029,
      "step": 113707
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.959191620349884,
      "learning_rate": 0.00030610883236116716,
      "loss": 2.0413,
      "step": 113708
    },
    {
      "epoch": 1.48,
      "grad_norm": 1.0035005807876587,
      "learning_rate": 0.0003061047426465761,
      "loss": 2.1992,
      "step": 113709
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.9552941918373108,
      "learning_rate": 0.0003061006529308501,
      "loss": 2.1583,
      "step": 113710
    },
    {
      "epoch": 1.48,
      "grad_norm": 1.0035274028778076,
      "learning_rate": 0.00030609656321398994,
      "loss": 1.9925,
      "step": 113711
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.8820178508758545,
      "learning_rate": 0.0003060924734959962,
      "loss": 2.0082,
      "step": 113712
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.981419026851654,
      "learning_rate": 0.00030608838377686975,
      "loss": 2.0669,
      "step": 113713
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.9359241127967834,
      "learning_rate": 0.00030608429405661145,
      "loss": 2.2062,
      "step": 113714
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.9118381142616272,
      "learning_rate": 0.0003060802043352219,
      "loss": 2.0412,
      "step": 113715
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.9752638339996338,
      "learning_rate": 0.00030607611461270194,
      "loss": 2.1829,
      "step": 113716
    },
    {
      "epoch": 1.48,
      "grad_norm": 1.0177682638168335,
      "learning_rate": 0.0003060720248890524,
      "loss": 2.1622,
      "step": 113717
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.8414238095283508,
      "learning_rate": 0.00030606793516427385,
      "loss": 2.172,
      "step": 113718
    },
    {
      "epoch": 1.48,
      "grad_norm": 1.1275309324264526,
      "learning_rate": 0.0003060638454383672,
      "loss": 2.1302,
      "step": 113719
    },
    {
      "epoch": 1.48,
      "grad_norm": 1.1835476160049438,
      "learning_rate": 0.00030605975571133316,
      "loss": 2.2271,
      "step": 113720
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.906397819519043,
      "learning_rate": 0.0003060556659831725,
      "loss": 2.4875,
      "step": 113721
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.9555383324623108,
      "learning_rate": 0.0003060515762538859,
      "loss": 1.9575,
      "step": 113722
    },
    {
      "epoch": 1.48,
      "grad_norm": 1.1348137855529785,
      "learning_rate": 0.00030604748652347433,
      "loss": 2.0005,
      "step": 113723
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.9985975027084351,
      "learning_rate": 0.0003060433967919383,
      "loss": 2.349,
      "step": 113724
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.9930987358093262,
      "learning_rate": 0.0003060393070592787,
      "loss": 2.2213,
      "step": 113725
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.8889708518981934,
      "learning_rate": 0.00030603521732549643,
      "loss": 1.9397,
      "step": 113726
    },
    {
      "epoch": 1.48,
      "grad_norm": 1.025356650352478,
      "learning_rate": 0.00030603112759059197,
      "loss": 2.1491,
      "step": 113727
    },
    {
      "epoch": 1.48,
      "grad_norm": 1.1046034097671509,
      "learning_rate": 0.00030602703785456616,
      "loss": 2.1067,
      "step": 113728
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.9921576380729675,
      "learning_rate": 0.00030602294811741995,
      "loss": 2.1323,
      "step": 113729
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.970512866973877,
      "learning_rate": 0.00030601885837915386,
      "loss": 2.332,
      "step": 113730
    },
    {
      "epoch": 1.48,
      "grad_norm": 1.1213250160217285,
      "learning_rate": 0.00030601476863976876,
      "loss": 2.097,
      "step": 113731
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.9119865298271179,
      "learning_rate": 0.0003060106788992654,
      "loss": 2.0382,
      "step": 113732
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.8809887766838074,
      "learning_rate": 0.0003060065891576446,
      "loss": 2.2172,
      "step": 113733
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.9141674637794495,
      "learning_rate": 0.00030600249941490704,
      "loss": 2.1409,
      "step": 113734
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.9685389995574951,
      "learning_rate": 0.00030599840967105346,
      "loss": 1.97,
      "step": 113735
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.9305070638656616,
      "learning_rate": 0.00030599431992608467,
      "loss": 2.2336,
      "step": 113736
    },
    {
      "epoch": 1.48,
      "grad_norm": 1.1186211109161377,
      "learning_rate": 0.00030599023018000143,
      "loss": 2.3024,
      "step": 113737
    },
    {
      "epoch": 1.48,
      "grad_norm": 1.0497252941131592,
      "learning_rate": 0.0003059861404328045,
      "loss": 2.1567,
      "step": 113738
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.9204362630844116,
      "learning_rate": 0.0003059820506844947,
      "loss": 2.1112,
      "step": 113739
    },
    {
      "epoch": 1.48,
      "grad_norm": 1.4310357570648193,
      "learning_rate": 0.0003059779609350726,
      "loss": 2.0058,
      "step": 113740
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.975049614906311,
      "learning_rate": 0.0003059738711845392,
      "loss": 2.201,
      "step": 113741
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.9656639099121094,
      "learning_rate": 0.0003059697814328951,
      "loss": 1.9484,
      "step": 113742
    },
    {
      "epoch": 1.48,
      "grad_norm": 1.5080883502960205,
      "learning_rate": 0.00030596569168014106,
      "loss": 2.3248,
      "step": 113743
    },
    {
      "epoch": 1.48,
      "grad_norm": 1.767634630203247,
      "learning_rate": 0.00030596160192627786,
      "loss": 2.0458,
      "step": 113744
    },
    {
      "epoch": 1.48,
      "grad_norm": 1.1161344051361084,
      "learning_rate": 0.0003059575121713064,
      "loss": 2.1731,
      "step": 113745
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.90279221534729,
      "learning_rate": 0.0003059534224152273,
      "loss": 2.3168,
      "step": 113746
    },
    {
      "epoch": 1.48,
      "grad_norm": 1.199538230895996,
      "learning_rate": 0.00030594933265804127,
      "loss": 2.0917,
      "step": 113747
    },
    {
      "epoch": 1.48,
      "grad_norm": 1.5956069231033325,
      "learning_rate": 0.00030594524289974926,
      "loss": 2.1153,
      "step": 113748
    },
    {
      "epoch": 1.48,
      "grad_norm": 1.0215189456939697,
      "learning_rate": 0.0003059411531403518,
      "loss": 2.3754,
      "step": 113749
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.9650644659996033,
      "learning_rate": 0.0003059370633798498,
      "loss": 2.2896,
      "step": 113750
    },
    {
      "epoch": 1.48,
      "grad_norm": 1.2867847681045532,
      "learning_rate": 0.0003059329736182441,
      "loss": 2.2105,
      "step": 113751
    },
    {
      "epoch": 1.48,
      "grad_norm": 1.0234450101852417,
      "learning_rate": 0.0003059288838555352,
      "loss": 2.0234,
      "step": 113752
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.9466812014579773,
      "learning_rate": 0.0003059247940917241,
      "loss": 2.2191,
      "step": 113753
    },
    {
      "epoch": 1.48,
      "grad_norm": 1.0394459962844849,
      "learning_rate": 0.0003059207043268114,
      "loss": 2.0698,
      "step": 113754
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.9393327236175537,
      "learning_rate": 0.00030591661456079797,
      "loss": 2.0657,
      "step": 113755
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.9827410578727722,
      "learning_rate": 0.0003059125247936845,
      "loss": 2.0933,
      "step": 113756
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.9947040677070618,
      "learning_rate": 0.00030590843502547183,
      "loss": 2.2644,
      "step": 113757
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.873931884765625,
      "learning_rate": 0.0003059043452561606,
      "loss": 1.92,
      "step": 113758
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.8382183313369751,
      "learning_rate": 0.0003059002554857517,
      "loss": 2.0814,
      "step": 113759
    },
    {
      "epoch": 1.48,
      "grad_norm": 1.0506417751312256,
      "learning_rate": 0.0003058961657142459,
      "loss": 2.1374,
      "step": 113760
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.933125376701355,
      "learning_rate": 0.00030589207594164375,
      "loss": 2.2049,
      "step": 113761
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.8543726205825806,
      "learning_rate": 0.0003058879861679462,
      "loss": 2.1423,
      "step": 113762
    },
    {
      "epoch": 1.48,
      "grad_norm": 1.0871163606643677,
      "learning_rate": 0.000305883896393154,
      "loss": 2.0862,
      "step": 113763
    },
    {
      "epoch": 1.48,
      "grad_norm": 1.2559301853179932,
      "learning_rate": 0.00030587980661726783,
      "loss": 2.1934,
      "step": 113764
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.9420669674873352,
      "learning_rate": 0.0003058757168402885,
      "loss": 2.2414,
      "step": 113765
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.9377025365829468,
      "learning_rate": 0.0003058716270622168,
      "loss": 2.1786,
      "step": 113766
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.9259384870529175,
      "learning_rate": 0.0003058675372830534,
      "loss": 2.0341,
      "step": 113767
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.9144764542579651,
      "learning_rate": 0.0003058634475027991,
      "loss": 2.3333,
      "step": 113768
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.8897450566291809,
      "learning_rate": 0.0003058593577214548,
      "loss": 2.0088,
      "step": 113769
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.8903772234916687,
      "learning_rate": 0.000305855267939021,
      "loss": 2.1856,
      "step": 113770
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.9964696168899536,
      "learning_rate": 0.0003058511781554986,
      "loss": 2.1836,
      "step": 113771
    },
    {
      "epoch": 1.48,
      "grad_norm": 1.0347998142242432,
      "learning_rate": 0.00030584708837088845,
      "loss": 2.2391,
      "step": 113772
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.8642290234565735,
      "learning_rate": 0.00030584299858519117,
      "loss": 2.0752,
      "step": 113773
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.9253087043762207,
      "learning_rate": 0.0003058389087984075,
      "loss": 2.3616,
      "step": 113774
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.905376672744751,
      "learning_rate": 0.0003058348190105384,
      "loss": 2.2708,
      "step": 113775
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.9436947107315063,
      "learning_rate": 0.00030583072922158437,
      "loss": 2.1526,
      "step": 113776
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.9484081268310547,
      "learning_rate": 0.0003058266394315464,
      "loss": 2.0553,
      "step": 113777
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.9492650628089905,
      "learning_rate": 0.000305822549640425,
      "loss": 2.003,
      "step": 113778
    },
    {
      "epoch": 1.48,
      "grad_norm": 1.0299371480941772,
      "learning_rate": 0.0003058184598482212,
      "loss": 2.164,
      "step": 113779
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.9829022288322449,
      "learning_rate": 0.00030581437005493565,
      "loss": 1.9452,
      "step": 113780
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.8912943005561829,
      "learning_rate": 0.00030581028026056904,
      "loss": 2.1101,
      "step": 113781
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.8409240245819092,
      "learning_rate": 0.00030580619046512224,
      "loss": 1.8246,
      "step": 113782
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.9158145189285278,
      "learning_rate": 0.00030580210066859595,
      "loss": 2.0687,
      "step": 113783
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.9453058242797852,
      "learning_rate": 0.0003057980108709909,
      "loss": 2.2064,
      "step": 113784
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.9313753247261047,
      "learning_rate": 0.00030579392107230784,
      "loss": 2.4072,
      "step": 113785
    },
    {
      "epoch": 1.48,
      "grad_norm": 1.0065863132476807,
      "learning_rate": 0.0003057898312725477,
      "loss": 2.2357,
      "step": 113786
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.8971735835075378,
      "learning_rate": 0.00030578574147171106,
      "loss": 2.355,
      "step": 113787
    },
    {
      "epoch": 1.48,
      "grad_norm": 1.0170332193374634,
      "learning_rate": 0.0003057816516697987,
      "loss": 2.1593,
      "step": 113788
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.9613015055656433,
      "learning_rate": 0.0003057775618668115,
      "loss": 2.1021,
      "step": 113789
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.8562191128730774,
      "learning_rate": 0.0003057734720627502,
      "loss": 2.0076,
      "step": 113790
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.9256146550178528,
      "learning_rate": 0.0003057693822576154,
      "loss": 2.0265,
      "step": 113791
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.9253628849983215,
      "learning_rate": 0.0003057652924514079,
      "loss": 2.2578,
      "step": 113792
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.836615264415741,
      "learning_rate": 0.0003057612026441287,
      "loss": 2.0129,
      "step": 113793
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.8952586054801941,
      "learning_rate": 0.00030575711283577824,
      "loss": 2.1128,
      "step": 113794
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.8963815569877625,
      "learning_rate": 0.00030575302302635743,
      "loss": 2.2613,
      "step": 113795
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.9071106314659119,
      "learning_rate": 0.00030574893321586716,
      "loss": 2.2338,
      "step": 113796
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.870128870010376,
      "learning_rate": 0.000305744843404308,
      "loss": 1.8631,
      "step": 113797
    },
    {
      "epoch": 1.48,
      "grad_norm": 1.1394014358520508,
      "learning_rate": 0.0003057407535916807,
      "loss": 2.0291,
      "step": 113798
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.8918716311454773,
      "learning_rate": 0.0003057366637779861,
      "loss": 1.9616,
      "step": 113799
    },
    {
      "epoch": 1.48,
      "grad_norm": 1.0826973915100098,
      "learning_rate": 0.00030573257396322495,
      "loss": 2.2905,
      "step": 113800
    },
    {
      "epoch": 1.48,
      "grad_norm": 1.0600389242172241,
      "learning_rate": 0.00030572848414739796,
      "loss": 2.089,
      "step": 113801
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.9111661314964294,
      "learning_rate": 0.000305724394330506,
      "loss": 2.1037,
      "step": 113802
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.9314976930618286,
      "learning_rate": 0.00030572030451254977,
      "loss": 2.1145,
      "step": 113803
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.8411508202552795,
      "learning_rate": 0.0003057162146935301,
      "loss": 1.9732,
      "step": 113804
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.9675775766372681,
      "learning_rate": 0.00030571212487344755,
      "loss": 2.0623,
      "step": 113805
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.9394626617431641,
      "learning_rate": 0.00030570803505230304,
      "loss": 2.0134,
      "step": 113806
    },
    {
      "epoch": 1.48,
      "grad_norm": 1.0206687450408936,
      "learning_rate": 0.0003057039452300973,
      "loss": 1.9839,
      "step": 113807
    },
    {
      "epoch": 1.48,
      "grad_norm": 1.1186119318008423,
      "learning_rate": 0.00030569985540683113,
      "loss": 2.1307,
      "step": 113808
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.8837680220603943,
      "learning_rate": 0.00030569576558250525,
      "loss": 2.1314,
      "step": 113809
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.9327953457832336,
      "learning_rate": 0.00030569167575712037,
      "loss": 2.0416,
      "step": 113810
    },
    {
      "epoch": 1.48,
      "grad_norm": 1.1050982475280762,
      "learning_rate": 0.0003056875859306773,
      "loss": 2.0373,
      "step": 113811
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.9309135675430298,
      "learning_rate": 0.00030568349610317683,
      "loss": 2.2584,
      "step": 113812
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.9050255417823792,
      "learning_rate": 0.0003056794062746197,
      "loss": 2.1564,
      "step": 113813
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.9025008678436279,
      "learning_rate": 0.0003056753164450066,
      "loss": 2.1495,
      "step": 113814
    },
    {
      "epoch": 1.48,
      "grad_norm": 1.1299723386764526,
      "learning_rate": 0.00030567122661433844,
      "loss": 2.1527,
      "step": 113815
    },
    {
      "epoch": 1.48,
      "grad_norm": 1.0115257501602173,
      "learning_rate": 0.0003056671367826158,
      "loss": 2.1183,
      "step": 113816
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.9102404117584229,
      "learning_rate": 0.0003056630469498396,
      "loss": 2.1477,
      "step": 113817
    },
    {
      "epoch": 1.48,
      "grad_norm": 1.0065348148345947,
      "learning_rate": 0.00030565895711601057,
      "loss": 2.0931,
      "step": 113818
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.8988457918167114,
      "learning_rate": 0.00030565486728112933,
      "loss": 2.018,
      "step": 113819
    },
    {
      "epoch": 1.48,
      "grad_norm": 1.1376831531524658,
      "learning_rate": 0.0003056507774451968,
      "loss": 2.2178,
      "step": 113820
    },
    {
      "epoch": 1.48,
      "grad_norm": 1.12397301197052,
      "learning_rate": 0.0003056466876082137,
      "loss": 2.0311,
      "step": 113821
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.9653456807136536,
      "learning_rate": 0.0003056425977701807,
      "loss": 1.993,
      "step": 113822
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.9644028544425964,
      "learning_rate": 0.00030563850793109866,
      "loss": 1.9407,
      "step": 113823
    },
    {
      "epoch": 1.48,
      "grad_norm": 1.200870156288147,
      "learning_rate": 0.0003056344180909684,
      "loss": 1.9584,
      "step": 113824
    },
    {
      "epoch": 1.48,
      "grad_norm": 1.0003395080566406,
      "learning_rate": 0.0003056303282497905,
      "loss": 2.3283,
      "step": 113825
    },
    {
      "epoch": 1.48,
      "grad_norm": 1.0140999555587769,
      "learning_rate": 0.0003056262384075658,
      "loss": 2.1443,
      "step": 113826
    },
    {
      "epoch": 1.48,
      "grad_norm": 1.0588278770446777,
      "learning_rate": 0.0003056221485642952,
      "loss": 2.0709,
      "step": 113827
    },
    {
      "epoch": 1.48,
      "grad_norm": 1.0187499523162842,
      "learning_rate": 0.0003056180587199792,
      "loss": 2.1784,
      "step": 113828
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.9763185381889343,
      "learning_rate": 0.00030561396887461876,
      "loss": 2.0793,
      "step": 113829
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.9783807992935181,
      "learning_rate": 0.0003056098790282146,
      "loss": 2.226,
      "step": 113830
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.938462495803833,
      "learning_rate": 0.0003056057891807675,
      "loss": 1.9412,
      "step": 113831
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.8291410803794861,
      "learning_rate": 0.00030560169933227803,
      "loss": 2.0755,
      "step": 113832
    },
    {
      "epoch": 1.48,
      "grad_norm": 1.0210309028625488,
      "learning_rate": 0.0003055976094827472,
      "loss": 2.0241,
      "step": 113833
    },
    {
      "epoch": 1.48,
      "grad_norm": 1.033306360244751,
      "learning_rate": 0.00030559351963217567,
      "loss": 2.371,
      "step": 113834
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.9435672163963318,
      "learning_rate": 0.00030558942978056416,
      "loss": 2.1595,
      "step": 113835
    },
    {
      "epoch": 1.48,
      "grad_norm": 1.164981484413147,
      "learning_rate": 0.00030558533992791354,
      "loss": 1.8224,
      "step": 113836
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.8845391273498535,
      "learning_rate": 0.00030558125007422443,
      "loss": 2.2406,
      "step": 113837
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.8902402520179749,
      "learning_rate": 0.00030557716021949763,
      "loss": 1.9814,
      "step": 113838
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.9328576326370239,
      "learning_rate": 0.000305573070363734,
      "loss": 2.0376,
      "step": 113839
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.9299378395080566,
      "learning_rate": 0.0003055689805069342,
      "loss": 2.0439,
      "step": 113840
    },
    {
      "epoch": 1.48,
      "grad_norm": 1.0114108324050903,
      "learning_rate": 0.000305564890649099,
      "loss": 2.3961,
      "step": 113841
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.8814972639083862,
      "learning_rate": 0.0003055608007902293,
      "loss": 1.9425,
      "step": 113842
    },
    {
      "epoch": 1.48,
      "grad_norm": 1.0479475259780884,
      "learning_rate": 0.00030555671093032564,
      "loss": 2.1254,
      "step": 113843
    },
    {
      "epoch": 1.48,
      "grad_norm": 1.082170009613037,
      "learning_rate": 0.00030555262106938885,
      "loss": 1.9188,
      "step": 113844
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.9939032793045044,
      "learning_rate": 0.0003055485312074198,
      "loss": 2.4292,
      "step": 113845
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.8933759331703186,
      "learning_rate": 0.0003055444413444191,
      "loss": 2.1317,
      "step": 113846
    },
    {
      "epoch": 1.48,
      "grad_norm": 1.0031144618988037,
      "learning_rate": 0.00030554035148038763,
      "loss": 2.2384,
      "step": 113847
    },
    {
      "epoch": 1.48,
      "grad_norm": 1.403562307357788,
      "learning_rate": 0.00030553626161532616,
      "loss": 2.0587,
      "step": 113848
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.8845361471176147,
      "learning_rate": 0.0003055321717492353,
      "loss": 1.8575,
      "step": 113849
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.9600049257278442,
      "learning_rate": 0.0003055280818821159,
      "loss": 2.1618,
      "step": 113850
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.9879977107048035,
      "learning_rate": 0.00030552399201396884,
      "loss": 2.0371,
      "step": 113851
    },
    {
      "epoch": 1.48,
      "grad_norm": 1.286242961883545,
      "learning_rate": 0.0003055199021447946,
      "loss": 2.1283,
      "step": 113852
    },
    {
      "epoch": 1.48,
      "grad_norm": 1.1749340295791626,
      "learning_rate": 0.0003055158122745942,
      "loss": 2.1438,
      "step": 113853
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.9631633758544922,
      "learning_rate": 0.00030551172240336836,
      "loss": 1.9871,
      "step": 113854
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.9539546370506287,
      "learning_rate": 0.0003055076325311177,
      "loss": 2.2634,
      "step": 113855
    },
    {
      "epoch": 1.48,
      "grad_norm": 1.1737284660339355,
      "learning_rate": 0.000305503542657843,
      "loss": 1.8437,
      "step": 113856
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.9854830503463745,
      "learning_rate": 0.0003054994527835453,
      "loss": 2.1381,
      "step": 113857
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.9594072103500366,
      "learning_rate": 0.000305495362908225,
      "loss": 2.2217,
      "step": 113858
    },
    {
      "epoch": 1.48,
      "grad_norm": 1.2820944786071777,
      "learning_rate": 0.000305491273031883,
      "loss": 2.1589,
      "step": 113859
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.9404165148735046,
      "learning_rate": 0.00030548718315452015,
      "loss": 2.1812,
      "step": 113860
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.8933039903640747,
      "learning_rate": 0.00030548309327613714,
      "loss": 2.0802,
      "step": 113861
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.915120005607605,
      "learning_rate": 0.0003054790033967346,
      "loss": 2.3304,
      "step": 113862
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.8785720467567444,
      "learning_rate": 0.0003054749135163135,
      "loss": 2.0743,
      "step": 113863
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.9836465120315552,
      "learning_rate": 0.00030547082363487446,
      "loss": 2.166,
      "step": 113864
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.9822999238967896,
      "learning_rate": 0.0003054667337524183,
      "loss": 2.1774,
      "step": 113865
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.8953369855880737,
      "learning_rate": 0.0003054626438689458,
      "loss": 2.033,
      "step": 113866
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.9846935868263245,
      "learning_rate": 0.00030545855398445774,
      "loss": 1.9958,
      "step": 113867
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.9705911874771118,
      "learning_rate": 0.0003054544640989547,
      "loss": 2.2289,
      "step": 113868
    },
    {
      "epoch": 1.48,
      "grad_norm": 1.0259997844696045,
      "learning_rate": 0.0003054503742124377,
      "loss": 2.1906,
      "step": 113869
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.9821314811706543,
      "learning_rate": 0.0003054462843249072,
      "loss": 2.2308,
      "step": 113870
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.9335097074508667,
      "learning_rate": 0.0003054421944363643,
      "loss": 1.969,
      "step": 113871
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.8238276839256287,
      "learning_rate": 0.0003054381045468096,
      "loss": 2.1041,
      "step": 113872
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.8866680264472961,
      "learning_rate": 0.00030543401465624376,
      "loss": 2.1521,
      "step": 113873
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.8920282125473022,
      "learning_rate": 0.0003054299247646677,
      "loss": 2.277,
      "step": 113874
    },
    {
      "epoch": 1.48,
      "grad_norm": 1.2358380556106567,
      "learning_rate": 0.00030542583487208207,
      "loss": 2.1022,
      "step": 113875
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.9822898507118225,
      "learning_rate": 0.0003054217449784877,
      "loss": 2.0157,
      "step": 113876
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.9794326424598694,
      "learning_rate": 0.00030541765508388524,
      "loss": 2.0359,
      "step": 113877
    },
    {
      "epoch": 1.48,
      "grad_norm": 1.0262176990509033,
      "learning_rate": 0.00030541356518827567,
      "loss": 2.255,
      "step": 113878
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.9829005599021912,
      "learning_rate": 0.0003054094752916595,
      "loss": 2.1576,
      "step": 113879
    },
    {
      "epoch": 1.48,
      "grad_norm": 1.0646705627441406,
      "learning_rate": 0.00030540538539403767,
      "loss": 2.127,
      "step": 113880
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.9107639789581299,
      "learning_rate": 0.00030540129549541087,
      "loss": 2.0968,
      "step": 113881
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.9252516031265259,
      "learning_rate": 0.0003053972055957799,
      "loss": 2.1429,
      "step": 113882
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.9336019158363342,
      "learning_rate": 0.0003053931156951454,
      "loss": 2.2046,
      "step": 113883
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.838076651096344,
      "learning_rate": 0.0003053890257935083,
      "loss": 2.2093,
      "step": 113884
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.9862239956855774,
      "learning_rate": 0.00030538493589086927,
      "loss": 2.0919,
      "step": 113885
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.9871594905853271,
      "learning_rate": 0.00030538084598722906,
      "loss": 2.1914,
      "step": 113886
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.8591204285621643,
      "learning_rate": 0.0003053767560825884,
      "loss": 2.111,
      "step": 113887
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.9188243746757507,
      "learning_rate": 0.0003053726661769482,
      "loss": 2.2433,
      "step": 113888
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.9901143312454224,
      "learning_rate": 0.000305368576270309,
      "loss": 2.0716,
      "step": 113889
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.9361944198608398,
      "learning_rate": 0.0003053644863626718,
      "loss": 2.1065,
      "step": 113890
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.9283506274223328,
      "learning_rate": 0.00030536039645403726,
      "loss": 2.1347,
      "step": 113891
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.9254105091094971,
      "learning_rate": 0.000305356306544406,
      "loss": 2.3059,
      "step": 113892
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.8836991190910339,
      "learning_rate": 0.0003053522166337789,
      "loss": 2.0584,
      "step": 113893
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.9296053647994995,
      "learning_rate": 0.0003053481267221568,
      "loss": 2.066,
      "step": 113894
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.9129655957221985,
      "learning_rate": 0.00030534403680954037,
      "loss": 2.112,
      "step": 113895
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.9520440101623535,
      "learning_rate": 0.0003053399468959304,
      "loss": 2.1224,
      "step": 113896
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.8759589791297913,
      "learning_rate": 0.0003053358569813276,
      "loss": 2.4479,
      "step": 113897
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.9316675662994385,
      "learning_rate": 0.0003053317670657327,
      "loss": 2.4194,
      "step": 113898
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.9033334851264954,
      "learning_rate": 0.0003053276771491466,
      "loss": 2.1674,
      "step": 113899
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.8938766121864319,
      "learning_rate": 0.00030532358723157004,
      "loss": 2.0266,
      "step": 113900
    },
    {
      "epoch": 1.48,
      "grad_norm": 1.0187454223632812,
      "learning_rate": 0.0003053194973130036,
      "loss": 2.1575,
      "step": 113901
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.8982380032539368,
      "learning_rate": 0.0003053154073934482,
      "loss": 2.0776,
      "step": 113902
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.9052842855453491,
      "learning_rate": 0.00030531131747290463,
      "loss": 2.0843,
      "step": 113903
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.9240944981575012,
      "learning_rate": 0.0003053072275513736,
      "loss": 2.1819,
      "step": 113904
    },
    {
      "epoch": 1.48,
      "grad_norm": 1.0212482213974,
      "learning_rate": 0.00030530313762885573,
      "loss": 2.1586,
      "step": 113905
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.8969642519950867,
      "learning_rate": 0.00030529904770535207,
      "loss": 2.1038,
      "step": 113906
    },
    {
      "epoch": 1.48,
      "grad_norm": 1.008129358291626,
      "learning_rate": 0.00030529495778086304,
      "loss": 2.4032,
      "step": 113907
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.913214385509491,
      "learning_rate": 0.00030529086785538966,
      "loss": 2.2166,
      "step": 113908
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.8511567115783691,
      "learning_rate": 0.00030528677792893263,
      "loss": 2.1424,
      "step": 113909
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.9266309142112732,
      "learning_rate": 0.00030528268800149267,
      "loss": 2.1539,
      "step": 113910
    },
    {
      "epoch": 1.48,
      "grad_norm": 1.0041295289993286,
      "learning_rate": 0.00030527859807307053,
      "loss": 2.2092,
      "step": 113911
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.912265419960022,
      "learning_rate": 0.0003052745081436671,
      "loss": 2.1531,
      "step": 113912
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.9048230648040771,
      "learning_rate": 0.0003052704182132829,
      "loss": 1.9728,
      "step": 113913
    },
    {
      "epoch": 1.48,
      "grad_norm": 1.0420411825180054,
      "learning_rate": 0.0003052663282819189,
      "loss": 2.299,
      "step": 113914
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.9223647713661194,
      "learning_rate": 0.0003052622383495758,
      "loss": 2.2132,
      "step": 113915
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.8584012985229492,
      "learning_rate": 0.0003052581484162543,
      "loss": 2.0292,
      "step": 113916
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.9930456280708313,
      "learning_rate": 0.00030525405848195526,
      "loss": 2.0139,
      "step": 113917
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.8416188359260559,
      "learning_rate": 0.00030524996854667944,
      "loss": 2.0978,
      "step": 113918
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.8540482521057129,
      "learning_rate": 0.00030524587861042746,
      "loss": 1.8481,
      "step": 113919
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.944812536239624,
      "learning_rate": 0.00030524178867320014,
      "loss": 2.0624,
      "step": 113920
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.9924818277359009,
      "learning_rate": 0.00030523769873499844,
      "loss": 2.138,
      "step": 113921
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.9419780969619751,
      "learning_rate": 0.00030523360879582275,
      "loss": 2.1575,
      "step": 113922
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.9611658453941345,
      "learning_rate": 0.00030522951885567415,
      "loss": 2.2284,
      "step": 113923
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.9225779175758362,
      "learning_rate": 0.0003052254289145533,
      "loss": 2.0552,
      "step": 113924
    },
    {
      "epoch": 1.48,
      "grad_norm": 1.0030537843704224,
      "learning_rate": 0.0003052213389724609,
      "loss": 2.3525,
      "step": 113925
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.9232835173606873,
      "learning_rate": 0.00030521724902939773,
      "loss": 1.9533,
      "step": 113926
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.9218336343765259,
      "learning_rate": 0.00030521315908536465,
      "loss": 2.0263,
      "step": 113927
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.8957207202911377,
      "learning_rate": 0.0003052090691403623,
      "loss": 2.1911,
      "step": 113928
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.9765790700912476,
      "learning_rate": 0.0003052049791943915,
      "loss": 2.3777,
      "step": 113929
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.9072727560997009,
      "learning_rate": 0.000305200889247453,
      "loss": 2.1626,
      "step": 113930
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.8693982362747192,
      "learning_rate": 0.00030519679929954754,
      "loss": 2.0591,
      "step": 113931
    },
    {
      "epoch": 1.48,
      "grad_norm": 1.093746542930603,
      "learning_rate": 0.00030519270935067583,
      "loss": 2.2281,
      "step": 113932
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.8411847352981567,
      "learning_rate": 0.0003051886194008388,
      "loss": 1.9439,
      "step": 113933
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.837181806564331,
      "learning_rate": 0.0003051845294500371,
      "loss": 2.1186,
      "step": 113934
    },
    {
      "epoch": 1.48,
      "grad_norm": 1.0611627101898193,
      "learning_rate": 0.00030518043949827144,
      "loss": 2.2244,
      "step": 113935
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.9214479327201843,
      "learning_rate": 0.0003051763495455426,
      "loss": 2.2437,
      "step": 113936
    },
    {
      "epoch": 1.48,
      "grad_norm": 1.0103912353515625,
      "learning_rate": 0.00030517225959185147,
      "loss": 2.224,
      "step": 113937
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.9758579730987549,
      "learning_rate": 0.00030516816963719874,
      "loss": 2.1452,
      "step": 113938
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.895914614200592,
      "learning_rate": 0.0003051640796815851,
      "loss": 2.078,
      "step": 113939
    },
    {
      "epoch": 1.48,
      "grad_norm": 1.27727472782135,
      "learning_rate": 0.0003051599897250114,
      "loss": 2.1473,
      "step": 113940
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.8860529661178589,
      "learning_rate": 0.0003051558997674783,
      "loss": 2.0936,
      "step": 113941
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.8575699925422668,
      "learning_rate": 0.0003051518098089866,
      "loss": 2.1751,
      "step": 113942
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.9968057870864868,
      "learning_rate": 0.0003051477198495372,
      "loss": 2.3259,
      "step": 113943
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.9861052632331848,
      "learning_rate": 0.00030514362988913067,
      "loss": 2.2068,
      "step": 113944
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.9209367632865906,
      "learning_rate": 0.0003051395399277678,
      "loss": 2.2563,
      "step": 113945
    },
    {
      "epoch": 1.48,
      "grad_norm": 1.0404595136642456,
      "learning_rate": 0.00030513544996544945,
      "loss": 2.1323,
      "step": 113946
    },
    {
      "epoch": 1.48,
      "grad_norm": 1.2775487899780273,
      "learning_rate": 0.00030513136000217633,
      "loss": 2.1153,
      "step": 113947
    },
    {
      "epoch": 1.48,
      "grad_norm": 1.0133633613586426,
      "learning_rate": 0.0003051272700379491,
      "loss": 2.1288,
      "step": 113948
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.894382894039154,
      "learning_rate": 0.0003051231800727688,
      "loss": 2.2038,
      "step": 113949
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.9377112984657288,
      "learning_rate": 0.0003051190901066358,
      "loss": 1.8355,
      "step": 113950
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.8928977847099304,
      "learning_rate": 0.0003051150001395512,
      "loss": 2.1812,
      "step": 113951
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.9480172991752625,
      "learning_rate": 0.0003051109101715156,
      "loss": 2.1282,
      "step": 113952
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.8650422096252441,
      "learning_rate": 0.00030510682020252973,
      "loss": 1.8247,
      "step": 113953
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.9123637676239014,
      "learning_rate": 0.0003051027302325944,
      "loss": 2.1166,
      "step": 113954
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.9985752105712891,
      "learning_rate": 0.0003050986402617105,
      "loss": 2.2377,
      "step": 113955
    },
    {
      "epoch": 1.48,
      "grad_norm": 1.0664106607437134,
      "learning_rate": 0.00030509455028987856,
      "loss": 2.2013,
      "step": 113956
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.9013803601264954,
      "learning_rate": 0.00030509046031709943,
      "loss": 2.1044,
      "step": 113957
    },
    {
      "epoch": 1.48,
      "grad_norm": 1.152652382850647,
      "learning_rate": 0.00030508637034337395,
      "loss": 2.0971,
      "step": 113958
    },
    {
      "epoch": 1.48,
      "grad_norm": 1.0712400674819946,
      "learning_rate": 0.00030508228036870284,
      "loss": 2.1796,
      "step": 113959
    },
    {
      "epoch": 1.48,
      "grad_norm": 1.203283429145813,
      "learning_rate": 0.0003050781903930868,
      "loss": 2.3165,
      "step": 113960
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.9199045896530151,
      "learning_rate": 0.00030507410041652656,
      "loss": 1.7974,
      "step": 113961
    },
    {
      "epoch": 1.48,
      "grad_norm": 1.0391737222671509,
      "learning_rate": 0.00030507001043902297,
      "loss": 2.2086,
      "step": 113962
    },
    {
      "epoch": 1.48,
      "grad_norm": 1.1026060581207275,
      "learning_rate": 0.0003050659204605769,
      "loss": 2.1549,
      "step": 113963
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.9926925897598267,
      "learning_rate": 0.00030506183048118884,
      "loss": 2.2815,
      "step": 113964
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.8934105038642883,
      "learning_rate": 0.00030505774050085976,
      "loss": 2.0516,
      "step": 113965
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.9283166527748108,
      "learning_rate": 0.00030505365051959035,
      "loss": 2.0611,
      "step": 113966
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.9519971013069153,
      "learning_rate": 0.0003050495605373813,
      "loss": 2.2185,
      "step": 113967
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.9296537041664124,
      "learning_rate": 0.0003050454705542335,
      "loss": 2.252,
      "step": 113968
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.920955240726471,
      "learning_rate": 0.00030504138057014764,
      "loss": 2.1295,
      "step": 113969
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.9361080527305603,
      "learning_rate": 0.0003050372905851245,
      "loss": 2.1843,
      "step": 113970
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.9085496664047241,
      "learning_rate": 0.00030503320059916485,
      "loss": 2.0927,
      "step": 113971
    },
    {
      "epoch": 1.48,
      "grad_norm": 1.0109089612960815,
      "learning_rate": 0.00030502911061226935,
      "loss": 2.1882,
      "step": 113972
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.8596262335777283,
      "learning_rate": 0.000305025020624439,
      "loss": 2.228,
      "step": 113973
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.8491637706756592,
      "learning_rate": 0.0003050209306356743,
      "loss": 2.0607,
      "step": 113974
    },
    {
      "epoch": 1.48,
      "grad_norm": 1.0120238065719604,
      "learning_rate": 0.0003050168406459761,
      "loss": 1.8386,
      "step": 113975
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.9431220293045044,
      "learning_rate": 0.0003050127506553452,
      "loss": 2.2263,
      "step": 113976
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.8382654190063477,
      "learning_rate": 0.00030500866066378237,
      "loss": 2.0969,
      "step": 113977
    },
    {
      "epoch": 1.48,
      "grad_norm": 1.0151273012161255,
      "learning_rate": 0.0003050045706712882,
      "loss": 2.1556,
      "step": 113978
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.857018768787384,
      "learning_rate": 0.0003050004806778637,
      "loss": 2.2946,
      "step": 113979
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.9857510924339294,
      "learning_rate": 0.00030499639068350956,
      "loss": 2.2819,
      "step": 113980
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.9413257837295532,
      "learning_rate": 0.0003049923006882264,
      "loss": 2.1515,
      "step": 113981
    },
    {
      "epoch": 1.48,
      "grad_norm": 1.0118530988693237,
      "learning_rate": 0.0003049882106920152,
      "loss": 2.1736,
      "step": 113982
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.937270998954773,
      "learning_rate": 0.00030498412069487646,
      "loss": 2.3735,
      "step": 113983
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.8701781034469604,
      "learning_rate": 0.00030498003069681114,
      "loss": 1.8653,
      "step": 113984
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.9223722219467163,
      "learning_rate": 0.00030497594069781996,
      "loss": 2.3024,
      "step": 113985
    },
    {
      "epoch": 1.48,
      "grad_norm": 1.0416125059127808,
      "learning_rate": 0.00030497185069790363,
      "loss": 2.0877,
      "step": 113986
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.8832888603210449,
      "learning_rate": 0.00030496776069706286,
      "loss": 2.2222,
      "step": 113987
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.8679103851318359,
      "learning_rate": 0.0003049636706952986,
      "loss": 2.3226,
      "step": 113988
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.8971477150917053,
      "learning_rate": 0.0003049595806926115,
      "loss": 2.0953,
      "step": 113989
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.8994961380958557,
      "learning_rate": 0.0003049554906890022,
      "loss": 1.8969,
      "step": 113990
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.9114030599594116,
      "learning_rate": 0.0003049514006844718,
      "loss": 2.2177,
      "step": 113991
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.8935571312904358,
      "learning_rate": 0.00030494731067902067,
      "loss": 2.1247,
      "step": 113992
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.9443609714508057,
      "learning_rate": 0.00030494322067264976,
      "loss": 2.199,
      "step": 113993
    },
    {
      "epoch": 1.48,
      "grad_norm": 1.0181807279586792,
      "learning_rate": 0.00030493913066535983,
      "loss": 2.1151,
      "step": 113994
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.9564663767814636,
      "learning_rate": 0.00030493504065715163,
      "loss": 2.2786,
      "step": 113995
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.8399249315261841,
      "learning_rate": 0.00030493095064802587,
      "loss": 2.088,
      "step": 113996
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.946286678314209,
      "learning_rate": 0.0003049268606379834,
      "loss": 2.0623,
      "step": 113997
    },
    {
      "epoch": 1.48,
      "grad_norm": 1.046543002128601,
      "learning_rate": 0.000304922770627025,
      "loss": 2.2488,
      "step": 113998
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.8741168975830078,
      "learning_rate": 0.0003049186806151512,
      "loss": 2.0957,
      "step": 113999
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.882032573223114,
      "learning_rate": 0.00030491459060236306,
      "loss": 2.0942,
      "step": 114000
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.9509886503219604,
      "learning_rate": 0.00030491050058866116,
      "loss": 2.0108,
      "step": 114001
    },
    {
      "epoch": 1.48,
      "grad_norm": 1.0105036497116089,
      "learning_rate": 0.00030490641057404625,
      "loss": 2.1863,
      "step": 114002
    },
    {
      "epoch": 1.48,
      "grad_norm": 1.0066598653793335,
      "learning_rate": 0.0003049023205585192,
      "loss": 2.1071,
      "step": 114003
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.9703829288482666,
      "learning_rate": 0.00030489823054208077,
      "loss": 2.1134,
      "step": 114004
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.9540302753448486,
      "learning_rate": 0.00030489414052473155,
      "loss": 2.1398,
      "step": 114005
    },
    {
      "epoch": 1.48,
      "grad_norm": 1.1686017513275146,
      "learning_rate": 0.00030489005050647253,
      "loss": 2.1509,
      "step": 114006
    },
    {
      "epoch": 1.48,
      "grad_norm": 1.0819153785705566,
      "learning_rate": 0.0003048859604873043,
      "loss": 2.0698,
      "step": 114007
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.9082951545715332,
      "learning_rate": 0.00030488187046722766,
      "loss": 2.2302,
      "step": 114008
    },
    {
      "epoch": 1.48,
      "grad_norm": 1.0325201749801636,
      "learning_rate": 0.0003048777804462434,
      "loss": 1.9869,
      "step": 114009
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.9805470108985901,
      "learning_rate": 0.00030487369042435234,
      "loss": 2.2817,
      "step": 114010
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.9757806658744812,
      "learning_rate": 0.00030486960040155506,
      "loss": 2.2523,
      "step": 114011
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.9245845079421997,
      "learning_rate": 0.0003048655103778525,
      "loss": 2.1428,
      "step": 114012
    },
    {
      "epoch": 1.48,
      "grad_norm": 1.1419122219085693,
      "learning_rate": 0.00030486142035324534,
      "loss": 2.1607,
      "step": 114013
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.9776204228401184,
      "learning_rate": 0.0003048573303277343,
      "loss": 2.1915,
      "step": 114014
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.8703770041465759,
      "learning_rate": 0.00030485324030132024,
      "loss": 2.2336,
      "step": 114015
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.9595360159873962,
      "learning_rate": 0.00030484915027400385,
      "loss": 2.1013,
      "step": 114016
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.8670990467071533,
      "learning_rate": 0.0003048450602457859,
      "loss": 2.1199,
      "step": 114017
    },
    {
      "epoch": 1.48,
      "grad_norm": 1.087714433670044,
      "learning_rate": 0.00030484097021666717,
      "loss": 2.2754,
      "step": 114018
    },
    {
      "epoch": 1.48,
      "grad_norm": 1.044136643409729,
      "learning_rate": 0.0003048368801866485,
      "loss": 1.9485,
      "step": 114019
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.8634030222892761,
      "learning_rate": 0.00030483279015573045,
      "loss": 2.1415,
      "step": 114020
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.8965394496917725,
      "learning_rate": 0.00030482870012391397,
      "loss": 2.0413,
      "step": 114021
    },
    {
      "epoch": 1.48,
      "grad_norm": 1.2629168033599854,
      "learning_rate": 0.0003048246100911997,
      "loss": 2.3808,
      "step": 114022
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.9355530142784119,
      "learning_rate": 0.00030482052005758844,
      "loss": 2.0727,
      "step": 114023
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.8584596514701843,
      "learning_rate": 0.0003048164300230809,
      "loss": 1.9759,
      "step": 114024
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.9908151626586914,
      "learning_rate": 0.000304812339987678,
      "loss": 2.2996,
      "step": 114025
    },
    {
      "epoch": 1.48,
      "grad_norm": 1.2261877059936523,
      "learning_rate": 0.00030480824995138034,
      "loss": 2.2686,
      "step": 114026
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.9274884462356567,
      "learning_rate": 0.00030480415991418876,
      "loss": 2.0747,
      "step": 114027
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.8499030470848083,
      "learning_rate": 0.000304800069876104,
      "loss": 2.0108,
      "step": 114028
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.995611846446991,
      "learning_rate": 0.00030479597983712676,
      "loss": 2.0474,
      "step": 114029
    },
    {
      "epoch": 1.48,
      "grad_norm": 1.3715972900390625,
      "learning_rate": 0.0003047918897972579,
      "loss": 2.1563,
      "step": 114030
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.9615687131881714,
      "learning_rate": 0.00030478779975649816,
      "loss": 1.9251,
      "step": 114031
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.9182709455490112,
      "learning_rate": 0.0003047837097148483,
      "loss": 2.1177,
      "step": 114032
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.9983193874359131,
      "learning_rate": 0.000304779619672309,
      "loss": 2.0654,
      "step": 114033
    },
    {
      "epoch": 1.48,
      "grad_norm": 1.1685690879821777,
      "learning_rate": 0.0003047755296288811,
      "loss": 1.9747,
      "step": 114034
    },
    {
      "epoch": 1.48,
      "grad_norm": 1.0099890232086182,
      "learning_rate": 0.00030477143958456527,
      "loss": 2.1795,
      "step": 114035
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.9199828505516052,
      "learning_rate": 0.0003047673495393624,
      "loss": 2.2009,
      "step": 114036
    },
    {
      "epoch": 1.48,
      "grad_norm": 1.1835037469863892,
      "learning_rate": 0.0003047632594932732,
      "loss": 2.1887,
      "step": 114037
    },
    {
      "epoch": 1.48,
      "grad_norm": 1.7719067335128784,
      "learning_rate": 0.00030475916944629845,
      "loss": 2.0358,
      "step": 114038
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.9535983204841614,
      "learning_rate": 0.00030475507939843876,
      "loss": 2.0185,
      "step": 114039
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.8317434787750244,
      "learning_rate": 0.00030475098934969517,
      "loss": 2.0964,
      "step": 114040
    },
    {
      "epoch": 1.48,
      "grad_norm": 1.0806727409362793,
      "learning_rate": 0.00030474689930006817,
      "loss": 2.2158,
      "step": 114041
    },
    {
      "epoch": 1.48,
      "grad_norm": 1.1453417539596558,
      "learning_rate": 0.00030474280924955863,
      "loss": 2.3275,
      "step": 114042
    },
    {
      "epoch": 1.48,
      "grad_norm": 1.0645278692245483,
      "learning_rate": 0.0003047387191981673,
      "loss": 2.0797,
      "step": 114043
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.9406613707542419,
      "learning_rate": 0.0003047346291458951,
      "loss": 2.1726,
      "step": 114044
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.9836204648017883,
      "learning_rate": 0.0003047305390927425,
      "loss": 2.2962,
      "step": 114045
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.9456101655960083,
      "learning_rate": 0.00030472644903871036,
      "loss": 2.1591,
      "step": 114046
    },
    {
      "epoch": 1.48,
      "grad_norm": 1.0220924615859985,
      "learning_rate": 0.00030472235898379966,
      "loss": 2.1563,
      "step": 114047
    },
    {
      "epoch": 1.48,
      "grad_norm": 1.0576093196868896,
      "learning_rate": 0.0003047182689280108,
      "loss": 2.2308,
      "step": 114048
    },
    {
      "epoch": 1.48,
      "grad_norm": 1.0123512744903564,
      "learning_rate": 0.00030471417887134486,
      "loss": 2.1221,
      "step": 114049
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.8222783803939819,
      "learning_rate": 0.00030471008881380237,
      "loss": 2.2245,
      "step": 114050
    },
    {
      "epoch": 1.49,
      "grad_norm": 1.088317632675171,
      "learning_rate": 0.0003047059987553843,
      "loss": 2.1097,
      "step": 114051
    },
    {
      "epoch": 1.49,
      "grad_norm": 1.1886602640151978,
      "learning_rate": 0.00030470190869609124,
      "loss": 2.0665,
      "step": 114052
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.9118693470954895,
      "learning_rate": 0.00030469781863592396,
      "loss": 2.2159,
      "step": 114053
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.9218026995658875,
      "learning_rate": 0.00030469372857488334,
      "loss": 2.1824,
      "step": 114054
    },
    {
      "epoch": 1.49,
      "grad_norm": 1.046033501625061,
      "learning_rate": 0.00030468963851297,
      "loss": 2.0293,
      "step": 114055
    },
    {
      "epoch": 1.49,
      "grad_norm": 1.166822075843811,
      "learning_rate": 0.0003046855484501848,
      "loss": 2.2225,
      "step": 114056
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.9449536800384521,
      "learning_rate": 0.00030468145838652845,
      "loss": 2.0835,
      "step": 114057
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.9305241703987122,
      "learning_rate": 0.0003046773683220018,
      "loss": 2.0677,
      "step": 114058
    },
    {
      "epoch": 1.49,
      "grad_norm": 1.3332968950271606,
      "learning_rate": 0.00030467327825660546,
      "loss": 1.9846,
      "step": 114059
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.9865524768829346,
      "learning_rate": 0.0003046691881903403,
      "loss": 2.1164,
      "step": 114060
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.9725647568702698,
      "learning_rate": 0.00030466509812320707,
      "loss": 2.2667,
      "step": 114061
    },
    {
      "epoch": 1.49,
      "grad_norm": 1.0607504844665527,
      "learning_rate": 0.0003046610080552064,
      "loss": 2.0155,
      "step": 114062
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.9178417921066284,
      "learning_rate": 0.00030465691798633933,
      "loss": 1.9245,
      "step": 114063
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.8487539291381836,
      "learning_rate": 0.00030465282791660635,
      "loss": 1.9648,
      "step": 114064
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.9743326902389526,
      "learning_rate": 0.00030464873784600833,
      "loss": 2.2661,
      "step": 114065
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.9095456004142761,
      "learning_rate": 0.00030464464777454605,
      "loss": 2.0829,
      "step": 114066
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.9078909754753113,
      "learning_rate": 0.0003046405577022202,
      "loss": 2.2795,
      "step": 114067
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.9009470343589783,
      "learning_rate": 0.0003046364676290316,
      "loss": 2.0045,
      "step": 114068
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.9049898386001587,
      "learning_rate": 0.00030463237755498103,
      "loss": 2.0818,
      "step": 114069
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.9191551804542542,
      "learning_rate": 0.00030462828748006914,
      "loss": 1.9371,
      "step": 114070
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.9168073534965515,
      "learning_rate": 0.0003046241974042969,
      "loss": 2.4468,
      "step": 114071
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.9544690847396851,
      "learning_rate": 0.00030462010732766473,
      "loss": 1.9502,
      "step": 114072
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.8706346750259399,
      "learning_rate": 0.00030461601725017374,
      "loss": 2.205,
      "step": 114073
    },
    {
      "epoch": 1.49,
      "grad_norm": 1.0654993057250977,
      "learning_rate": 0.00030461192717182455,
      "loss": 1.9336,
      "step": 114074
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.9146601557731628,
      "learning_rate": 0.0003046078370926178,
      "loss": 2.2494,
      "step": 114075
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.8838974237442017,
      "learning_rate": 0.00030460374701255443,
      "loss": 2.3437,
      "step": 114076
    },
    {
      "epoch": 1.49,
      "grad_norm": 1.1674641370773315,
      "learning_rate": 0.0003045996569316352,
      "loss": 2.0962,
      "step": 114077
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.8972315192222595,
      "learning_rate": 0.00030459556684986073,
      "loss": 2.0893,
      "step": 114078
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.9278081059455872,
      "learning_rate": 0.0003045914767672319,
      "loss": 2.2608,
      "step": 114079
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.9675448536872864,
      "learning_rate": 0.00030458738668374946,
      "loss": 2.3131,
      "step": 114080
    },
    {
      "epoch": 1.49,
      "grad_norm": 1.06374192237854,
      "learning_rate": 0.00030458329659941406,
      "loss": 2.0694,
      "step": 114081
    },
    {
      "epoch": 1.49,
      "grad_norm": 1.1170049905776978,
      "learning_rate": 0.00030457920651422654,
      "loss": 2.1476,
      "step": 114082
    },
    {
      "epoch": 1.49,
      "grad_norm": 1.328867793083191,
      "learning_rate": 0.00030457511642818775,
      "loss": 2.2264,
      "step": 114083
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.9235478043556213,
      "learning_rate": 0.0003045710263412983,
      "loss": 2.1736,
      "step": 114084
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.8634849190711975,
      "learning_rate": 0.00030456693625355893,
      "loss": 2.1011,
      "step": 114085
    },
    {
      "epoch": 1.49,
      "grad_norm": 1.314687728881836,
      "learning_rate": 0.00030456284616497064,
      "loss": 1.9938,
      "step": 114086
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.9604209661483765,
      "learning_rate": 0.0003045587560755339,
      "loss": 2.2015,
      "step": 114087
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.9034088253974915,
      "learning_rate": 0.00030455466598524965,
      "loss": 2.2702,
      "step": 114088
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.9995726346969604,
      "learning_rate": 0.0003045505758941186,
      "loss": 2.1194,
      "step": 114089
    },
    {
      "epoch": 1.49,
      "grad_norm": 1.0602976083755493,
      "learning_rate": 0.0003045464858021415,
      "loss": 2.2707,
      "step": 114090
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.9001803398132324,
      "learning_rate": 0.00030454239570931914,
      "loss": 2.2331,
      "step": 114091
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.9823718667030334,
      "learning_rate": 0.00030453830561565234,
      "loss": 2.243,
      "step": 114092
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.9740762114524841,
      "learning_rate": 0.00030453421552114165,
      "loss": 2.3554,
      "step": 114093
    },
    {
      "epoch": 1.49,
      "grad_norm": 1.1385400295257568,
      "learning_rate": 0.000304530125425788,
      "loss": 2.3081,
      "step": 114094
    },
    {
      "epoch": 1.49,
      "grad_norm": 1.412394642829895,
      "learning_rate": 0.0003045260353295922,
      "loss": 2.1438,
      "step": 114095
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.8916704058647156,
      "learning_rate": 0.00030452194523255475,
      "loss": 2.0868,
      "step": 114096
    },
    {
      "epoch": 1.49,
      "grad_norm": 1.1373707056045532,
      "learning_rate": 0.00030451785513467675,
      "loss": 2.3264,
      "step": 114097
    },
    {
      "epoch": 1.49,
      "grad_norm": 1.505911111831665,
      "learning_rate": 0.00030451376503595875,
      "loss": 2.1567,
      "step": 114098
    },
    {
      "epoch": 1.49,
      "grad_norm": 1.1587245464324951,
      "learning_rate": 0.0003045096749364015,
      "loss": 2.049,
      "step": 114099
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.9141885042190552,
      "learning_rate": 0.0003045055848360058,
      "loss": 2.1467,
      "step": 114100
    },
    {
      "epoch": 1.49,
      "grad_norm": 1.0755352973937988,
      "learning_rate": 0.00030450149473477256,
      "loss": 2.1116,
      "step": 114101
    },
    {
      "epoch": 1.49,
      "grad_norm": 1.2710647583007812,
      "learning_rate": 0.0003044974046327023,
      "loss": 2.1588,
      "step": 114102
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.9269725680351257,
      "learning_rate": 0.0003044933145297959,
      "loss": 2.0515,
      "step": 114103
    },
    {
      "epoch": 1.49,
      "grad_norm": 1.0037342309951782,
      "learning_rate": 0.00030448922442605415,
      "loss": 2.1957,
      "step": 114104
    },
    {
      "epoch": 1.49,
      "grad_norm": 1.2936265468597412,
      "learning_rate": 0.0003044851343214777,
      "loss": 2.1878,
      "step": 114105
    },
    {
      "epoch": 1.49,
      "grad_norm": 1.073032021522522,
      "learning_rate": 0.00030448104421606744,
      "loss": 1.9555,
      "step": 114106
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.9308806657791138,
      "learning_rate": 0.0003044769541098241,
      "loss": 2.1941,
      "step": 114107
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.8707970976829529,
      "learning_rate": 0.0003044728640027484,
      "loss": 1.8906,
      "step": 114108
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.9415178894996643,
      "learning_rate": 0.000304468773894841,
      "loss": 2.0196,
      "step": 114109
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.9159192442893982,
      "learning_rate": 0.00030446468378610294,
      "loss": 2.0285,
      "step": 114110
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.9885181188583374,
      "learning_rate": 0.00030446059367653463,
      "loss": 2.2376,
      "step": 114111
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.9683250784873962,
      "learning_rate": 0.0003044565035661371,
      "loss": 2.0804,
      "step": 114112
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.8931113481521606,
      "learning_rate": 0.0003044524134549111,
      "loss": 2.1416,
      "step": 114113
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.9209571480751038,
      "learning_rate": 0.0003044483233428572,
      "loss": 2.3616,
      "step": 114114
    },
    {
      "epoch": 1.49,
      "grad_norm": 1.0273246765136719,
      "learning_rate": 0.00030444423322997625,
      "loss": 2.3228,
      "step": 114115
    },
    {
      "epoch": 1.49,
      "grad_norm": 1.0909833908081055,
      "learning_rate": 0.0003044401431162691,
      "loss": 2.0992,
      "step": 114116
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.9146533608436584,
      "learning_rate": 0.00030443605300173646,
      "loss": 2.1965,
      "step": 114117
    },
    {
      "epoch": 1.49,
      "grad_norm": 1.0885945558547974,
      "learning_rate": 0.000304431962886379,
      "loss": 2.0085,
      "step": 114118
    },
    {
      "epoch": 1.49,
      "grad_norm": 1.0854144096374512,
      "learning_rate": 0.00030442787277019766,
      "loss": 1.8179,
      "step": 114119
    },
    {
      "epoch": 1.49,
      "grad_norm": 1.0126534700393677,
      "learning_rate": 0.00030442378265319297,
      "loss": 2.2374,
      "step": 114120
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.8684874176979065,
      "learning_rate": 0.00030441969253536586,
      "loss": 2.2298,
      "step": 114121
    },
    {
      "epoch": 1.49,
      "grad_norm": 1.076390027999878,
      "learning_rate": 0.0003044156024167171,
      "loss": 2.2009,
      "step": 114122
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.9916722178459167,
      "learning_rate": 0.00030441151229724736,
      "loss": 2.089,
      "step": 114123
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.949259340763092,
      "learning_rate": 0.0003044074221769574,
      "loss": 2.2598,
      "step": 114124
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.9192626476287842,
      "learning_rate": 0.0003044033320558481,
      "loss": 2.1028,
      "step": 114125
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.8703674077987671,
      "learning_rate": 0.00030439924193392007,
      "loss": 2.0401,
      "step": 114126
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.8502851128578186,
      "learning_rate": 0.00030439515181117415,
      "loss": 2.2078,
      "step": 114127
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.9908626675605774,
      "learning_rate": 0.0003043910616876111,
      "loss": 2.3371,
      "step": 114128
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.8589363098144531,
      "learning_rate": 0.00030438697156323164,
      "loss": 2.1669,
      "step": 114129
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.9399445056915283,
      "learning_rate": 0.0003043828814380366,
      "loss": 2.3231,
      "step": 114130
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.9158208966255188,
      "learning_rate": 0.00030437879131202665,
      "loss": 2.0932,
      "step": 114131
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.8707208633422852,
      "learning_rate": 0.00030437470118520267,
      "loss": 2.2193,
      "step": 114132
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.9025073647499084,
      "learning_rate": 0.0003043706110575653,
      "loss": 1.8662,
      "step": 114133
    },
    {
      "epoch": 1.49,
      "grad_norm": 1.0147221088409424,
      "learning_rate": 0.00030436652092911536,
      "loss": 2.13,
      "step": 114134
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.8778513073921204,
      "learning_rate": 0.0003043624307998536,
      "loss": 2.133,
      "step": 114135
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.9040623307228088,
      "learning_rate": 0.0003043583406697808,
      "loss": 2.0363,
      "step": 114136
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.9136861562728882,
      "learning_rate": 0.0003043542505388977,
      "loss": 2.2898,
      "step": 114137
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.8134459257125854,
      "learning_rate": 0.000304350160407205,
      "loss": 1.983,
      "step": 114138
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.9041599035263062,
      "learning_rate": 0.00030434607027470363,
      "loss": 2.1309,
      "step": 114139
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.9871183633804321,
      "learning_rate": 0.0003043419801413942,
      "loss": 2.1657,
      "step": 114140
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.905158519744873,
      "learning_rate": 0.00030433789000727747,
      "loss": 2.0251,
      "step": 114141
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.9472715854644775,
      "learning_rate": 0.0003043337998723543,
      "loss": 2.2861,
      "step": 114142
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.9094650149345398,
      "learning_rate": 0.00030432970973662537,
      "loss": 2.1311,
      "step": 114143
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.8749241232872009,
      "learning_rate": 0.00030432561960009143,
      "loss": 2.0858,
      "step": 114144
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.8373526334762573,
      "learning_rate": 0.00030432152946275335,
      "loss": 1.9494,
      "step": 114145
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.9492054581642151,
      "learning_rate": 0.00030431743932461176,
      "loss": 2.1949,
      "step": 114146
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.9509454965591431,
      "learning_rate": 0.0003043133491856675,
      "loss": 2.1584,
      "step": 114147
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.9673641920089722,
      "learning_rate": 0.00030430925904592127,
      "loss": 2.1339,
      "step": 114148
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.9036353230476379,
      "learning_rate": 0.0003043051689053739,
      "loss": 2.1632,
      "step": 114149
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.9709171652793884,
      "learning_rate": 0.00030430107876402616,
      "loss": 2.0601,
      "step": 114150
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.8804844617843628,
      "learning_rate": 0.0003042969886218787,
      "loss": 2.093,
      "step": 114151
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.987542986869812,
      "learning_rate": 0.0003042928984789324,
      "loss": 2.0152,
      "step": 114152
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.8361608386039734,
      "learning_rate": 0.00030428880833518795,
      "loss": 2.0435,
      "step": 114153
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.9572727680206299,
      "learning_rate": 0.0003042847181906461,
      "loss": 2.1338,
      "step": 114154
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.8504485487937927,
      "learning_rate": 0.00030428062804530763,
      "loss": 2.3356,
      "step": 114155
    },
    {
      "epoch": 1.49,
      "grad_norm": 1.0220340490341187,
      "learning_rate": 0.0003042765378991734,
      "loss": 2.1095,
      "step": 114156
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.9441689848899841,
      "learning_rate": 0.000304272447752244,
      "loss": 1.9372,
      "step": 114157
    },
    {
      "epoch": 1.49,
      "grad_norm": 1.0768728256225586,
      "learning_rate": 0.0003042683576045203,
      "loss": 2.1624,
      "step": 114158
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.8745664358139038,
      "learning_rate": 0.0003042642674560031,
      "loss": 2.0644,
      "step": 114159
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.9084093570709229,
      "learning_rate": 0.00030426017730669297,
      "loss": 2.1404,
      "step": 114160
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.9354917407035828,
      "learning_rate": 0.00030425608715659085,
      "loss": 2.2125,
      "step": 114161
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.9698843359947205,
      "learning_rate": 0.00030425199700569747,
      "loss": 2.072,
      "step": 114162
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.9956391453742981,
      "learning_rate": 0.0003042479068540135,
      "loss": 2.1626,
      "step": 114163
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.8758039474487305,
      "learning_rate": 0.0003042438167015398,
      "loss": 1.8718,
      "step": 114164
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.8919821381568909,
      "learning_rate": 0.0003042397265482771,
      "loss": 2.2609,
      "step": 114165
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.9572589993476868,
      "learning_rate": 0.00030423563639422614,
      "loss": 2.2746,
      "step": 114166
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.9116535782814026,
      "learning_rate": 0.0003042315462393876,
      "loss": 2.1763,
      "step": 114167
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.9166091084480286,
      "learning_rate": 0.0003042274560837625,
      "loss": 1.9568,
      "step": 114168
    },
    {
      "epoch": 1.49,
      "grad_norm": 1.0039591789245605,
      "learning_rate": 0.0003042233659273514,
      "loss": 2.2326,
      "step": 114169
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.9084675312042236,
      "learning_rate": 0.000304219275770155,
      "loss": 2.1978,
      "step": 114170
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.9282261729240417,
      "learning_rate": 0.0003042151856121743,
      "loss": 2.1461,
      "step": 114171
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.8915240168571472,
      "learning_rate": 0.0003042110954534098,
      "loss": 2.1587,
      "step": 114172
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.9046372771263123,
      "learning_rate": 0.00030420700529386243,
      "loss": 2.0045,
      "step": 114173
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.8455068469047546,
      "learning_rate": 0.0003042029151335329,
      "loss": 1.9401,
      "step": 114174
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.954451858997345,
      "learning_rate": 0.00030419882497242196,
      "loss": 2.1798,
      "step": 114175
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.8649150729179382,
      "learning_rate": 0.0003041947348105304,
      "loss": 2.0515,
      "step": 114176
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.9762033224105835,
      "learning_rate": 0.00030419064464785894,
      "loss": 2.1103,
      "step": 114177
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.9272269606590271,
      "learning_rate": 0.00030418655448440833,
      "loss": 2.4949,
      "step": 114178
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.9051512479782104,
      "learning_rate": 0.0003041824643201794,
      "loss": 1.832,
      "step": 114179
    },
    {
      "epoch": 1.49,
      "grad_norm": 1.0773828029632568,
      "learning_rate": 0.0003041783741551729,
      "loss": 2.0771,
      "step": 114180
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.9398339986801147,
      "learning_rate": 0.0003041742839893895,
      "loss": 2.1124,
      "step": 114181
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.9239652156829834,
      "learning_rate": 0.00030417019382282995,
      "loss": 2.058,
      "step": 114182
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.9624780416488647,
      "learning_rate": 0.0003041661036554953,
      "loss": 2.0368,
      "step": 114183
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.940889298915863,
      "learning_rate": 0.00030416201348738587,
      "loss": 2.0738,
      "step": 114184
    },
    {
      "epoch": 1.49,
      "grad_norm": 1.0517725944519043,
      "learning_rate": 0.00030415792331850276,
      "loss": 2.3375,
      "step": 114185
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.9469842910766602,
      "learning_rate": 0.00030415383314884667,
      "loss": 1.9389,
      "step": 114186
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.8649744391441345,
      "learning_rate": 0.0003041497429784182,
      "loss": 2.1125,
      "step": 114187
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.9277761578559875,
      "learning_rate": 0.0003041456528072182,
      "loss": 2.0702,
      "step": 114188
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.8883371353149414,
      "learning_rate": 0.00030414156263524754,
      "loss": 2.1379,
      "step": 114189
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.8911049365997314,
      "learning_rate": 0.00030413747246250673,
      "loss": 2.1063,
      "step": 114190
    },
    {
      "epoch": 1.49,
      "grad_norm": 1.0613969564437866,
      "learning_rate": 0.0003041333822889968,
      "loss": 1.8513,
      "step": 114191
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.9376765489578247,
      "learning_rate": 0.00030412929211471847,
      "loss": 2.3452,
      "step": 114192
    },
    {
      "epoch": 1.49,
      "grad_norm": 1.0883561372756958,
      "learning_rate": 0.00030412520193967227,
      "loss": 2.1018,
      "step": 114193
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.994377613067627,
      "learning_rate": 0.00030412111176385915,
      "loss": 2.1052,
      "step": 114194
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.9534657597541809,
      "learning_rate": 0.00030411702158727994,
      "loss": 2.0027,
      "step": 114195
    },
    {
      "epoch": 1.49,
      "grad_norm": 1.062424898147583,
      "learning_rate": 0.0003041129314099352,
      "loss": 2.1414,
      "step": 114196
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.9701566696166992,
      "learning_rate": 0.0003041088412318258,
      "loss": 1.9869,
      "step": 114197
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.9381287097930908,
      "learning_rate": 0.00030410475105295255,
      "loss": 2.3624,
      "step": 114198
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.8504788279533386,
      "learning_rate": 0.0003041006608733161,
      "loss": 2.0422,
      "step": 114199
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.9605176448822021,
      "learning_rate": 0.0003040965706929172,
      "loss": 2.1084,
      "step": 114200
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.9139751195907593,
      "learning_rate": 0.0003040924805117568,
      "loss": 2.1879,
      "step": 114201
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.8931896686553955,
      "learning_rate": 0.0003040883903298355,
      "loss": 2.202,
      "step": 114202
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.8749547600746155,
      "learning_rate": 0.000304084300147154,
      "loss": 2.2048,
      "step": 114203
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.8939717411994934,
      "learning_rate": 0.0003040802099637132,
      "loss": 1.9906,
      "step": 114204
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.9832513928413391,
      "learning_rate": 0.00030407611977951386,
      "loss": 2.2326,
      "step": 114205
    },
    {
      "epoch": 1.49,
      "grad_norm": 1.0526762008666992,
      "learning_rate": 0.0003040720295945567,
      "loss": 2.275,
      "step": 114206
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.9258854985237122,
      "learning_rate": 0.00030406793940884237,
      "loss": 2.1702,
      "step": 114207
    },
    {
      "epoch": 1.49,
      "grad_norm": 1.0220452547073364,
      "learning_rate": 0.0003040638492223718,
      "loss": 2.3083,
      "step": 114208
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.9550107717514038,
      "learning_rate": 0.00030405975903514563,
      "loss": 2.178,
      "step": 114209
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.9470842480659485,
      "learning_rate": 0.00030405566884716473,
      "loss": 2.0776,
      "step": 114210
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.895413875579834,
      "learning_rate": 0.00030405157865842983,
      "loss": 2.1457,
      "step": 114211
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.9602386355400085,
      "learning_rate": 0.00030404748846894167,
      "loss": 2.1518,
      "step": 114212
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.913550615310669,
      "learning_rate": 0.0003040433982787009,
      "loss": 1.9464,
      "step": 114213
    },
    {
      "epoch": 1.49,
      "grad_norm": 1.0086688995361328,
      "learning_rate": 0.00030403930808770854,
      "loss": 2.3446,
      "step": 114214
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.8601423501968384,
      "learning_rate": 0.000304035217895965,
      "loss": 2.1174,
      "step": 114215
    },
    {
      "epoch": 1.49,
      "grad_norm": 1.0591340065002441,
      "learning_rate": 0.0003040311277034714,
      "loss": 2.0641,
      "step": 114216
    },
    {
      "epoch": 1.49,
      "grad_norm": 1.1126903295516968,
      "learning_rate": 0.0003040270375102283,
      "loss": 2.0755,
      "step": 114217
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.9214277863502502,
      "learning_rate": 0.00030402294731623645,
      "loss": 1.9765,
      "step": 114218
    },
    {
      "epoch": 1.49,
      "grad_norm": 1.188425064086914,
      "learning_rate": 0.0003040188571214967,
      "loss": 2.272,
      "step": 114219
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.9678444862365723,
      "learning_rate": 0.0003040147669260098,
      "loss": 2.3471,
      "step": 114220
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.8939673900604248,
      "learning_rate": 0.0003040106767297764,
      "loss": 2.3183,
      "step": 114221
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.9342067837715149,
      "learning_rate": 0.0003040065865327974,
      "loss": 2.1328,
      "step": 114222
    },
    {
      "epoch": 1.49,
      "grad_norm": 1.0034595727920532,
      "learning_rate": 0.0003040024963350734,
      "loss": 2.022,
      "step": 114223
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.8768996596336365,
      "learning_rate": 0.0003039984061366054,
      "loss": 2.2387,
      "step": 114224
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.84452885389328,
      "learning_rate": 0.00030399431593739395,
      "loss": 2.2722,
      "step": 114225
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.9397461414337158,
      "learning_rate": 0.00030399022573743983,
      "loss": 1.9952,
      "step": 114226
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.9806702733039856,
      "learning_rate": 0.00030398613553674395,
      "loss": 2.3801,
      "step": 114227
    },
    {
      "epoch": 1.49,
      "grad_norm": 1.1381629705429077,
      "learning_rate": 0.0003039820453353069,
      "loss": 2.2717,
      "step": 114228
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.9739619493484497,
      "learning_rate": 0.0003039779551331295,
      "loss": 2.1136,
      "step": 114229
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.8810866475105286,
      "learning_rate": 0.00030397386493021255,
      "loss": 2.2301,
      "step": 114230
    },
    {
      "epoch": 1.49,
      "grad_norm": 1.0061619281768799,
      "learning_rate": 0.00030396977472655685,
      "loss": 1.9511,
      "step": 114231
    },
    {
      "epoch": 1.49,
      "grad_norm": 1.0162955522537231,
      "learning_rate": 0.0003039656845221631,
      "loss": 1.9731,
      "step": 114232
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.968953549861908,
      "learning_rate": 0.00030396159431703195,
      "loss": 2.2236,
      "step": 114233
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.9677563905715942,
      "learning_rate": 0.00030395750411116426,
      "loss": 2.2263,
      "step": 114234
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.9707204103469849,
      "learning_rate": 0.00030395341390456094,
      "loss": 2.1177,
      "step": 114235
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.9961003661155701,
      "learning_rate": 0.00030394932369722243,
      "loss": 1.8364,
      "step": 114236
    },
    {
      "epoch": 1.49,
      "grad_norm": 1.0258814096450806,
      "learning_rate": 0.00030394523348914975,
      "loss": 2.2247,
      "step": 114237
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.882215678691864,
      "learning_rate": 0.0003039411432803436,
      "loss": 2.1195,
      "step": 114238
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.9320019483566284,
      "learning_rate": 0.0003039370530708047,
      "loss": 2.0321,
      "step": 114239
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.893653154373169,
      "learning_rate": 0.0003039329628605338,
      "loss": 2.1695,
      "step": 114240
    },
    {
      "epoch": 1.49,
      "grad_norm": 1.0987155437469482,
      "learning_rate": 0.00030392887264953176,
      "loss": 2.0927,
      "step": 114241
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.9318378567695618,
      "learning_rate": 0.0003039247824377992,
      "loss": 1.94,
      "step": 114242
    },
    {
      "epoch": 1.49,
      "grad_norm": 1.0518519878387451,
      "learning_rate": 0.00030392069222533695,
      "loss": 2.1482,
      "step": 114243
    },
    {
      "epoch": 1.49,
      "grad_norm": 1.0798745155334473,
      "learning_rate": 0.0003039166020121458,
      "loss": 2.2439,
      "step": 114244
    },
    {
      "epoch": 1.49,
      "grad_norm": 1.0170891284942627,
      "learning_rate": 0.0003039125117982264,
      "loss": 2.0599,
      "step": 114245
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.9991511106491089,
      "learning_rate": 0.0003039084215835797,
      "loss": 2.3151,
      "step": 114246
    },
    {
      "epoch": 1.49,
      "grad_norm": 1.062242865562439,
      "learning_rate": 0.00030390433136820633,
      "loss": 2.1158,
      "step": 114247
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.934548556804657,
      "learning_rate": 0.00030390024115210703,
      "loss": 2.2782,
      "step": 114248
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.9139818549156189,
      "learning_rate": 0.0003038961509352826,
      "loss": 2.1219,
      "step": 114249
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.968936026096344,
      "learning_rate": 0.00030389206071773384,
      "loss": 2.1407,
      "step": 114250
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.9406562447547913,
      "learning_rate": 0.0003038879704994615,
      "loss": 2.1492,
      "step": 114251
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.9790871143341064,
      "learning_rate": 0.0003038838802804662,
      "loss": 1.9987,
      "step": 114252
    },
    {
      "epoch": 1.49,
      "grad_norm": 1.0412667989730835,
      "learning_rate": 0.0003038797900607489,
      "loss": 1.9497,
      "step": 114253
    },
    {
      "epoch": 1.49,
      "grad_norm": 1.1899528503417969,
      "learning_rate": 0.00030387569984031027,
      "loss": 2.0661,
      "step": 114254
    },
    {
      "epoch": 1.49,
      "grad_norm": 1.4125851392745972,
      "learning_rate": 0.00030387160961915105,
      "loss": 2.3171,
      "step": 114255
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.9154195189476013,
      "learning_rate": 0.00030386751939727203,
      "loss": 2.3083,
      "step": 114256
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.9710167050361633,
      "learning_rate": 0.00030386342917467396,
      "loss": 1.9979,
      "step": 114257
    },
    {
      "epoch": 1.49,
      "grad_norm": 1.0686638355255127,
      "learning_rate": 0.0003038593389513576,
      "loss": 2.1052,
      "step": 114258
    },
    {
      "epoch": 1.49,
      "grad_norm": 1.2344093322753906,
      "learning_rate": 0.0003038552487273238,
      "loss": 2.3053,
      "step": 114259
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.9124535918235779,
      "learning_rate": 0.0003038511585025731,
      "loss": 2.311,
      "step": 114260
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.943162739276886,
      "learning_rate": 0.00030384706827710646,
      "loss": 1.9225,
      "step": 114261
    },
    {
      "epoch": 1.49,
      "grad_norm": 1.2370495796203613,
      "learning_rate": 0.0003038429780509246,
      "loss": 2.0077,
      "step": 114262
    },
    {
      "epoch": 1.49,
      "grad_norm": 1.446677565574646,
      "learning_rate": 0.00030383888782402823,
      "loss": 2.3885,
      "step": 114263
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.9098042845726013,
      "learning_rate": 0.00030383479759641817,
      "loss": 2.0743,
      "step": 114264
    },
    {
      "epoch": 1.49,
      "grad_norm": 1.0150728225708008,
      "learning_rate": 0.00030383070736809514,
      "loss": 2.2075,
      "step": 114265
    },
    {
      "epoch": 1.49,
      "grad_norm": 1.4168169498443604,
      "learning_rate": 0.00030382661713905984,
      "loss": 2.0328,
      "step": 114266
    },
    {
      "epoch": 1.49,
      "grad_norm": 1.1912931203842163,
      "learning_rate": 0.00030382252690931314,
      "loss": 2.086,
      "step": 114267
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.9410918354988098,
      "learning_rate": 0.00030381843667885575,
      "loss": 2.1123,
      "step": 114268
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.8806294202804565,
      "learning_rate": 0.00030381434644768853,
      "loss": 2.2246,
      "step": 114269
    },
    {
      "epoch": 1.49,
      "grad_norm": 1.2426165342330933,
      "learning_rate": 0.00030381025621581203,
      "loss": 2.0308,
      "step": 114270
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.9853672981262207,
      "learning_rate": 0.0003038061659832272,
      "loss": 2.111,
      "step": 114271
    },
    {
      "epoch": 1.49,
      "grad_norm": 1.0664361715316772,
      "learning_rate": 0.00030380207574993474,
      "loss": 2.1367,
      "step": 114272
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.9843327403068542,
      "learning_rate": 0.00030379798551593537,
      "loss": 2.1138,
      "step": 114273
    },
    {
      "epoch": 1.49,
      "grad_norm": 1.2311323881149292,
      "learning_rate": 0.00030379389528122986,
      "loss": 2.3251,
      "step": 114274
    },
    {
      "epoch": 1.49,
      "grad_norm": 1.0084640979766846,
      "learning_rate": 0.00030378980504581907,
      "loss": 2.5105,
      "step": 114275
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.8713772296905518,
      "learning_rate": 0.0003037857148097036,
      "loss": 2.0886,
      "step": 114276
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.9725189805030823,
      "learning_rate": 0.0003037816245728844,
      "loss": 2.0672,
      "step": 114277
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.9424376487731934,
      "learning_rate": 0.0003037775343353621,
      "loss": 2.4038,
      "step": 114278
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.8586776256561279,
      "learning_rate": 0.0003037734440971374,
      "loss": 2.4016,
      "step": 114279
    },
    {
      "epoch": 1.49,
      "grad_norm": 1.103519082069397,
      "learning_rate": 0.00030376935385821114,
      "loss": 2.1695,
      "step": 114280
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.906157374382019,
      "learning_rate": 0.0003037652636185842,
      "loss": 2.0185,
      "step": 114281
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.8594653606414795,
      "learning_rate": 0.0003037611733782571,
      "loss": 2.1257,
      "step": 114282
    },
    {
      "epoch": 1.49,
      "grad_norm": 1.3071783781051636,
      "learning_rate": 0.00030375708313723075,
      "loss": 2.034,
      "step": 114283
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.8748167157173157,
      "learning_rate": 0.00030375299289550606,
      "loss": 1.9256,
      "step": 114284
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.9165425300598145,
      "learning_rate": 0.0003037489026530834,
      "loss": 2.2159,
      "step": 114285
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.9018877744674683,
      "learning_rate": 0.00030374481240996384,
      "loss": 2.2986,
      "step": 114286
    },
    {
      "epoch": 1.49,
      "grad_norm": 1.1056110858917236,
      "learning_rate": 0.0003037407221661481,
      "loss": 2.2545,
      "step": 114287
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.8870611190795898,
      "learning_rate": 0.0003037366319216368,
      "loss": 2.3072,
      "step": 114288
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.8829175233840942,
      "learning_rate": 0.00030373254167643084,
      "loss": 2.2511,
      "step": 114289
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.8495705723762512,
      "learning_rate": 0.0003037284514305309,
      "loss": 2.2038,
      "step": 114290
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.9938356280326843,
      "learning_rate": 0.00030372436118393775,
      "loss": 2.1226,
      "step": 114291
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.9298509359359741,
      "learning_rate": 0.00030372027093665224,
      "loss": 2.221,
      "step": 114292
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.958785891532898,
      "learning_rate": 0.000303716180688675,
      "loss": 1.8856,
      "step": 114293
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.9711166620254517,
      "learning_rate": 0.00030371209044000686,
      "loss": 2.0878,
      "step": 114294
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.8747390508651733,
      "learning_rate": 0.0003037080001906486,
      "loss": 1.9575,
      "step": 114295
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.9474685192108154,
      "learning_rate": 0.000303703909940601,
      "loss": 2.127,
      "step": 114296
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.9475576281547546,
      "learning_rate": 0.0003036998196898647,
      "loss": 2.4888,
      "step": 114297
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.855695903301239,
      "learning_rate": 0.0003036957294384405,
      "loss": 2.2758,
      "step": 114298
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.8825560212135315,
      "learning_rate": 0.00030369163918632933,
      "loss": 2.1625,
      "step": 114299
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.9631966948509216,
      "learning_rate": 0.0003036875489335316,
      "loss": 2.0716,
      "step": 114300
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.9272016286849976,
      "learning_rate": 0.00030368345868004844,
      "loss": 1.9471,
      "step": 114301
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.80599445104599,
      "learning_rate": 0.0003036793684258805,
      "loss": 1.7272,
      "step": 114302
    },
    {
      "epoch": 1.49,
      "grad_norm": 1.0301228761672974,
      "learning_rate": 0.00030367527817102836,
      "loss": 2.1354,
      "step": 114303
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.9223204851150513,
      "learning_rate": 0.00030367118791549294,
      "loss": 2.1474,
      "step": 114304
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.9469899535179138,
      "learning_rate": 0.00030366709765927507,
      "loss": 2.1423,
      "step": 114305
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.9375292062759399,
      "learning_rate": 0.0003036630074023754,
      "loss": 2.0515,
      "step": 114306
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.8743751645088196,
      "learning_rate": 0.0003036589171447946,
      "loss": 1.9063,
      "step": 114307
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.9221516251564026,
      "learning_rate": 0.0003036548268865336,
      "loss": 2.0437,
      "step": 114308
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.8945791721343994,
      "learning_rate": 0.00030365073662759316,
      "loss": 2.2478,
      "step": 114309
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.8061894774436951,
      "learning_rate": 0.0003036466463679739,
      "loss": 1.9112,
      "step": 114310
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.8963020443916321,
      "learning_rate": 0.00030364255610767664,
      "loss": 2.1082,
      "step": 114311
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.8714427351951599,
      "learning_rate": 0.00030363846584670226,
      "loss": 2.0669,
      "step": 114312
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.9993671178817749,
      "learning_rate": 0.00030363437558505134,
      "loss": 1.9569,
      "step": 114313
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.9792172312736511,
      "learning_rate": 0.00030363028532272475,
      "loss": 2.1706,
      "step": 114314
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.9750823974609375,
      "learning_rate": 0.0003036261950597232,
      "loss": 2.154,
      "step": 114315
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.9103623032569885,
      "learning_rate": 0.0003036221047960475,
      "loss": 2.2288,
      "step": 114316
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.8791521787643433,
      "learning_rate": 0.0003036180145316984,
      "loss": 2.0671,
      "step": 114317
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.9226094484329224,
      "learning_rate": 0.0003036139242666766,
      "loss": 2.3644,
      "step": 114318
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.9120739102363586,
      "learning_rate": 0.0003036098340009829,
      "loss": 2.0674,
      "step": 114319
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.86540687084198,
      "learning_rate": 0.0003036057437346181,
      "loss": 2.2146,
      "step": 114320
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.9142727851867676,
      "learning_rate": 0.0003036016534675829,
      "loss": 2.1701,
      "step": 114321
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.9818079471588135,
      "learning_rate": 0.000303597563199878,
      "loss": 2.181,
      "step": 114322
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.9420956373214722,
      "learning_rate": 0.0003035934729315044,
      "loss": 2.1269,
      "step": 114323
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.9562471508979797,
      "learning_rate": 0.0003035893826624627,
      "loss": 2.1835,
      "step": 114324
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.9181797504425049,
      "learning_rate": 0.00030358529239275355,
      "loss": 2.0881,
      "step": 114325
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.8013893365859985,
      "learning_rate": 0.00030358120212237795,
      "loss": 2.0805,
      "step": 114326
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.9059063196182251,
      "learning_rate": 0.00030357711185133643,
      "loss": 1.9345,
      "step": 114327
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.899905800819397,
      "learning_rate": 0.00030357302157962986,
      "loss": 2.2158,
      "step": 114328
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.9427613019943237,
      "learning_rate": 0.00030356893130725906,
      "loss": 2.2422,
      "step": 114329
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.9931714534759521,
      "learning_rate": 0.0003035648410342247,
      "loss": 2.0359,
      "step": 114330
    },
    {
      "epoch": 1.49,
      "grad_norm": 1.014392375946045,
      "learning_rate": 0.0003035607507605276,
      "loss": 2.2879,
      "step": 114331
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.9198265671730042,
      "learning_rate": 0.00030355666048616844,
      "loss": 2.0644,
      "step": 114332
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.965552806854248,
      "learning_rate": 0.0003035525702111481,
      "loss": 2.2328,
      "step": 114333
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.9038069844245911,
      "learning_rate": 0.0003035484799354672,
      "loss": 2.2337,
      "step": 114334
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.9655491709709167,
      "learning_rate": 0.00030354438965912657,
      "loss": 2.0702,
      "step": 114335
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.93916255235672,
      "learning_rate": 0.000303540299382127,
      "loss": 2.0841,
      "step": 114336
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.8986937403678894,
      "learning_rate": 0.00030353620910446923,
      "loss": 2.0242,
      "step": 114337
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.8895989656448364,
      "learning_rate": 0.00030353211882615396,
      "loss": 2.2153,
      "step": 114338
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.96932452917099,
      "learning_rate": 0.00030352802854718205,
      "loss": 2.0193,
      "step": 114339
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.9849840998649597,
      "learning_rate": 0.0003035239382675542,
      "loss": 2.1494,
      "step": 114340
    },
    {
      "epoch": 1.49,
      "grad_norm": 1.029004693031311,
      "learning_rate": 0.0003035198479872712,
      "loss": 2.1836,
      "step": 114341
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.9073965549468994,
      "learning_rate": 0.00030351575770633384,
      "loss": 1.9729,
      "step": 114342
    },
    {
      "epoch": 1.49,
      "grad_norm": 1.1783822774887085,
      "learning_rate": 0.0003035116674247427,
      "loss": 2.2404,
      "step": 114343
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.9863276481628418,
      "learning_rate": 0.0003035075771424988,
      "loss": 2.0754,
      "step": 114344
    },
    {
      "epoch": 1.49,
      "grad_norm": 1.025006651878357,
      "learning_rate": 0.0003035034868596027,
      "loss": 1.9016,
      "step": 114345
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.8364955186843872,
      "learning_rate": 0.00030349939657605524,
      "loss": 2.2849,
      "step": 114346
    },
    {
      "epoch": 1.49,
      "grad_norm": 1.0158010721206665,
      "learning_rate": 0.00030349530629185715,
      "loss": 2.133,
      "step": 114347
    },
    {
      "epoch": 1.49,
      "grad_norm": 1.1906192302703857,
      "learning_rate": 0.00030349121600700934,
      "loss": 2.0927,
      "step": 114348
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.9475676417350769,
      "learning_rate": 0.0003034871257215124,
      "loss": 2.0411,
      "step": 114349
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.9303637146949768,
      "learning_rate": 0.00030348303543536705,
      "loss": 2.079,
      "step": 114350
    },
    {
      "epoch": 1.49,
      "grad_norm": 1.1073598861694336,
      "learning_rate": 0.00030347894514857424,
      "loss": 2.127,
      "step": 114351
    },
    {
      "epoch": 1.49,
      "grad_norm": 1.0736217498779297,
      "learning_rate": 0.00030347485486113455,
      "loss": 2.3566,
      "step": 114352
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.9088636636734009,
      "learning_rate": 0.0003034707645730489,
      "loss": 2.1837,
      "step": 114353
    },
    {
      "epoch": 1.49,
      "grad_norm": 1.0888605117797852,
      "learning_rate": 0.00030346667428431795,
      "loss": 2.0075,
      "step": 114354
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.9392399191856384,
      "learning_rate": 0.0003034625839949424,
      "loss": 2.0322,
      "step": 114355
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.9327462315559387,
      "learning_rate": 0.00030345849370492313,
      "loss": 2.1437,
      "step": 114356
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.8738654255867004,
      "learning_rate": 0.00030345440341426086,
      "loss": 2.0243,
      "step": 114357
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.9651410579681396,
      "learning_rate": 0.00030345031312295636,
      "loss": 1.9766,
      "step": 114358
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.935787558555603,
      "learning_rate": 0.0003034462228310104,
      "loss": 2.3426,
      "step": 114359
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.9248979687690735,
      "learning_rate": 0.00030344213253842377,
      "loss": 2.35,
      "step": 114360
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.9578490853309631,
      "learning_rate": 0.00030343804224519704,
      "loss": 2.0299,
      "step": 114361
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.872413158416748,
      "learning_rate": 0.00030343395195133116,
      "loss": 1.9552,
      "step": 114362
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.9377991557121277,
      "learning_rate": 0.00030342986165682697,
      "loss": 2.3838,
      "step": 114363
    },
    {
      "epoch": 1.49,
      "grad_norm": 1.2675888538360596,
      "learning_rate": 0.00030342577136168505,
      "loss": 2.3105,
      "step": 114364
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.961279034614563,
      "learning_rate": 0.0003034216810659061,
      "loss": 2.1533,
      "step": 114365
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.9488533735275269,
      "learning_rate": 0.0003034175907694911,
      "loss": 2.2316,
      "step": 114366
    },
    {
      "epoch": 1.49,
      "grad_norm": 1.2105462551116943,
      "learning_rate": 0.00030341350047244066,
      "loss": 2.4613,
      "step": 114367
    },
    {
      "epoch": 1.49,
      "grad_norm": 1.6046820878982544,
      "learning_rate": 0.00030340941017475563,
      "loss": 2.1239,
      "step": 114368
    },
    {
      "epoch": 1.49,
      "grad_norm": 1.4114807844161987,
      "learning_rate": 0.0003034053198764367,
      "loss": 1.9797,
      "step": 114369
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.9155821204185486,
      "learning_rate": 0.00030340122957748464,
      "loss": 2.1717,
      "step": 114370
    },
    {
      "epoch": 1.49,
      "grad_norm": 1.0725620985031128,
      "learning_rate": 0.0003033971392779002,
      "loss": 2.2238,
      "step": 114371
    },
    {
      "epoch": 1.49,
      "grad_norm": 1.6904594898223877,
      "learning_rate": 0.00030339304897768427,
      "loss": 2.2488,
      "step": 114372
    },
    {
      "epoch": 1.49,
      "grad_norm": 1.1297919750213623,
      "learning_rate": 0.00030338895867683743,
      "loss": 2.2458,
      "step": 114373
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.965028703212738,
      "learning_rate": 0.0003033848683753605,
      "loss": 1.9823,
      "step": 114374
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.9964167475700378,
      "learning_rate": 0.0003033807780732544,
      "loss": 2.0684,
      "step": 114375
    },
    {
      "epoch": 1.49,
      "grad_norm": 1.5042085647583008,
      "learning_rate": 0.0003033766877705196,
      "loss": 2.2495,
      "step": 114376
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.9845662713050842,
      "learning_rate": 0.000303372597467157,
      "loss": 2.2638,
      "step": 114377
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.885631799697876,
      "learning_rate": 0.00030336850716316746,
      "loss": 2.0436,
      "step": 114378
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.9171478748321533,
      "learning_rate": 0.0003033644168585516,
      "loss": 2.0832,
      "step": 114379
    },
    {
      "epoch": 1.49,
      "grad_norm": 1.2347707748413086,
      "learning_rate": 0.00030336032655331024,
      "loss": 2.1347,
      "step": 114380
    },
    {
      "epoch": 1.49,
      "grad_norm": 1.0627784729003906,
      "learning_rate": 0.0003033562362474442,
      "loss": 1.999,
      "step": 114381
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.8888914585113525,
      "learning_rate": 0.0003033521459409541,
      "loss": 2.1115,
      "step": 114382
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.8486905694007874,
      "learning_rate": 0.00030334805563384074,
      "loss": 2.0775,
      "step": 114383
    },
    {
      "epoch": 1.49,
      "grad_norm": 1.1468087434768677,
      "learning_rate": 0.000303343965326105,
      "loss": 2.0957,
      "step": 114384
    },
    {
      "epoch": 1.49,
      "grad_norm": 1.3676302433013916,
      "learning_rate": 0.0003033398750177475,
      "loss": 2.0366,
      "step": 114385
    },
    {
      "epoch": 1.49,
      "grad_norm": 1.264931559562683,
      "learning_rate": 0.00030333578470876906,
      "loss": 1.9932,
      "step": 114386
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.9435750246047974,
      "learning_rate": 0.00030333169439917043,
      "loss": 2.2022,
      "step": 114387
    },
    {
      "epoch": 1.49,
      "grad_norm": 1.1377476453781128,
      "learning_rate": 0.00030332760408895235,
      "loss": 2.1263,
      "step": 114388
    },
    {
      "epoch": 1.49,
      "grad_norm": 1.132383108139038,
      "learning_rate": 0.0003033235137781157,
      "loss": 2.2824,
      "step": 114389
    },
    {
      "epoch": 1.49,
      "grad_norm": 1.244755744934082,
      "learning_rate": 0.0003033194234666611,
      "loss": 2.1281,
      "step": 114390
    },
    {
      "epoch": 1.49,
      "grad_norm": 1.029738187789917,
      "learning_rate": 0.0003033153331545894,
      "loss": 2.2752,
      "step": 114391
    },
    {
      "epoch": 1.49,
      "grad_norm": 1.0665857791900635,
      "learning_rate": 0.0003033112428419012,
      "loss": 2.2555,
      "step": 114392
    },
    {
      "epoch": 1.49,
      "grad_norm": 1.0943728685379028,
      "learning_rate": 0.0003033071525285975,
      "loss": 2.1019,
      "step": 114393
    },
    {
      "epoch": 1.49,
      "grad_norm": 1.332047700881958,
      "learning_rate": 0.0003033030622146788,
      "loss": 2.2362,
      "step": 114394
    },
    {
      "epoch": 1.49,
      "grad_norm": 1.0667492151260376,
      "learning_rate": 0.0003032989719001461,
      "loss": 2.1984,
      "step": 114395
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.9868505597114563,
      "learning_rate": 0.00030329488158499995,
      "loss": 2.3106,
      "step": 114396
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.9964035153388977,
      "learning_rate": 0.00030329079126924143,
      "loss": 1.9573,
      "step": 114397
    },
    {
      "epoch": 1.49,
      "grad_norm": 1.7350468635559082,
      "learning_rate": 0.0003032867009528709,
      "loss": 2.1503,
      "step": 114398
    },
    {
      "epoch": 1.49,
      "grad_norm": 1.048638105392456,
      "learning_rate": 0.00030328261063588933,
      "loss": 2.2392,
      "step": 114399
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.9771386384963989,
      "learning_rate": 0.0003032785203182976,
      "loss": 2.1201,
      "step": 114400
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.8909544944763184,
      "learning_rate": 0.00030327443000009616,
      "loss": 2.0471,
      "step": 114401
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.9371434450149536,
      "learning_rate": 0.000303270339681286,
      "loss": 2.147,
      "step": 114402
    },
    {
      "epoch": 1.49,
      "grad_norm": 1.2143093347549438,
      "learning_rate": 0.0003032662493618678,
      "loss": 2.0931,
      "step": 114403
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.9918391108512878,
      "learning_rate": 0.00030326215904184246,
      "loss": 2.299,
      "step": 114404
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.9623761177062988,
      "learning_rate": 0.0003032580687212105,
      "loss": 2.0795,
      "step": 114405
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.9478158950805664,
      "learning_rate": 0.0003032539783999729,
      "loss": 2.2552,
      "step": 114406
    },
    {
      "epoch": 1.49,
      "grad_norm": 1.066705584526062,
      "learning_rate": 0.0003032498880781302,
      "loss": 2.1352,
      "step": 114407
    },
    {
      "epoch": 1.49,
      "grad_norm": 1.0604394674301147,
      "learning_rate": 0.00030324579775568335,
      "loss": 2.0304,
      "step": 114408
    },
    {
      "epoch": 1.49,
      "grad_norm": 1.0871747732162476,
      "learning_rate": 0.0003032417074326331,
      "loss": 2.1229,
      "step": 114409
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.9382296800613403,
      "learning_rate": 0.00030323761710898006,
      "loss": 2.3798,
      "step": 114410
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.9855620861053467,
      "learning_rate": 0.00030323352678472514,
      "loss": 2.0536,
      "step": 114411
    },
    {
      "epoch": 1.49,
      "grad_norm": 1.4245651960372925,
      "learning_rate": 0.0003032294364598691,
      "loss": 2.2115,
      "step": 114412
    },
    {
      "epoch": 1.49,
      "grad_norm": 1.1126739978790283,
      "learning_rate": 0.00030322534613441254,
      "loss": 2.1123,
      "step": 114413
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.985886812210083,
      "learning_rate": 0.0003032212558083563,
      "loss": 2.2531,
      "step": 114414
    },
    {
      "epoch": 1.49,
      "grad_norm": 1.2150990962982178,
      "learning_rate": 0.00030321716548170136,
      "loss": 2.0376,
      "step": 114415
    },
    {
      "epoch": 1.49,
      "grad_norm": 1.0734590291976929,
      "learning_rate": 0.0003032130751544481,
      "loss": 2.0765,
      "step": 114416
    },
    {
      "epoch": 1.49,
      "grad_norm": 1.1302640438079834,
      "learning_rate": 0.00030320898482659757,
      "loss": 2.1122,
      "step": 114417
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.9553811550140381,
      "learning_rate": 0.0003032048944981504,
      "loss": 2.1371,
      "step": 114418
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.9061369895935059,
      "learning_rate": 0.00030320080416910733,
      "loss": 2.1318,
      "step": 114419
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.9679338932037354,
      "learning_rate": 0.0003031967138394692,
      "loss": 2.0308,
      "step": 114420
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.8864864706993103,
      "learning_rate": 0.0003031926235092368,
      "loss": 2.4133,
      "step": 114421
    },
    {
      "epoch": 1.49,
      "grad_norm": 1.0665993690490723,
      "learning_rate": 0.0003031885331784108,
      "loss": 2.2906,
      "step": 114422
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.9487300515174866,
      "learning_rate": 0.00030318444284699195,
      "loss": 2.1747,
      "step": 114423
    },
    {
      "epoch": 1.49,
      "grad_norm": 1.1817511320114136,
      "learning_rate": 0.0003031803525149811,
      "loss": 1.9142,
      "step": 114424
    },
    {
      "epoch": 1.49,
      "grad_norm": 1.0520744323730469,
      "learning_rate": 0.00030317626218237894,
      "loss": 2.3222,
      "step": 114425
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.9510815143585205,
      "learning_rate": 0.00030317217184918624,
      "loss": 2.3185,
      "step": 114426
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.9409047961235046,
      "learning_rate": 0.00030316808151540384,
      "loss": 2.1592,
      "step": 114427
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.9698630571365356,
      "learning_rate": 0.00030316399118103234,
      "loss": 2.163,
      "step": 114428
    },
    {
      "epoch": 1.49,
      "grad_norm": 1.039817452430725,
      "learning_rate": 0.0003031599008460726,
      "loss": 1.946,
      "step": 114429
    },
    {
      "epoch": 1.49,
      "grad_norm": 1.3594151735305786,
      "learning_rate": 0.0003031558105105255,
      "loss": 2.0543,
      "step": 114430
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.9010004997253418,
      "learning_rate": 0.00030315172017439157,
      "loss": 2.0936,
      "step": 114431
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.9497700929641724,
      "learning_rate": 0.0003031476298376717,
      "loss": 1.9568,
      "step": 114432
    },
    {
      "epoch": 1.49,
      "grad_norm": 1.1087157726287842,
      "learning_rate": 0.0003031435395003666,
      "loss": 2.1459,
      "step": 114433
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.9016054272651672,
      "learning_rate": 0.0003031394491624771,
      "loss": 1.9235,
      "step": 114434
    },
    {
      "epoch": 1.49,
      "grad_norm": 1.0048896074295044,
      "learning_rate": 0.0003031353588240038,
      "loss": 2.323,
      "step": 114435
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.8919529914855957,
      "learning_rate": 0.00030313126848494776,
      "loss": 2.2462,
      "step": 114436
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.8567195534706116,
      "learning_rate": 0.00030312717814530947,
      "loss": 2.044,
      "step": 114437
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.947974443435669,
      "learning_rate": 0.0003031230878050898,
      "loss": 2.2051,
      "step": 114438
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.9218394160270691,
      "learning_rate": 0.00030311899746428945,
      "loss": 2.3042,
      "step": 114439
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.8715696334838867,
      "learning_rate": 0.00030311490712290923,
      "loss": 1.952,
      "step": 114440
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.9980319738388062,
      "learning_rate": 0.0003031108167809499,
      "loss": 2.2916,
      "step": 114441
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.9008603096008301,
      "learning_rate": 0.0003031067264384122,
      "loss": 2.1204,
      "step": 114442
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.9366194605827332,
      "learning_rate": 0.0003031026360952969,
      "loss": 2.2164,
      "step": 114443
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.937044084072113,
      "learning_rate": 0.0003030985457516048,
      "loss": 2.1156,
      "step": 114444
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.9755533933639526,
      "learning_rate": 0.0003030944554073366,
      "loss": 2.3636,
      "step": 114445
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.9651578664779663,
      "learning_rate": 0.0003030903650624931,
      "loss": 2.0947,
      "step": 114446
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.8921233415603638,
      "learning_rate": 0.000303086274717075,
      "loss": 2.0628,
      "step": 114447
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.9175252914428711,
      "learning_rate": 0.0003030821843710831,
      "loss": 2.1148,
      "step": 114448
    },
    {
      "epoch": 1.49,
      "grad_norm": 1.1149022579193115,
      "learning_rate": 0.0003030780940245182,
      "loss": 2.1196,
      "step": 114449
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.976966142654419,
      "learning_rate": 0.0003030740036773809,
      "loss": 2.216,
      "step": 114450
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.8891909718513489,
      "learning_rate": 0.0003030699133296723,
      "loss": 1.9642,
      "step": 114451
    },
    {
      "epoch": 1.49,
      "grad_norm": 1.0515583753585815,
      "learning_rate": 0.0003030658229813928,
      "loss": 2.2082,
      "step": 114452
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.9115814566612244,
      "learning_rate": 0.00030306173263254337,
      "loss": 2.1476,
      "step": 114453
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.9961644411087036,
      "learning_rate": 0.0003030576422831247,
      "loss": 2.1466,
      "step": 114454
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.9704306721687317,
      "learning_rate": 0.0003030535519331375,
      "loss": 2.1937,
      "step": 114455
    },
    {
      "epoch": 1.49,
      "grad_norm": 1.0184913873672485,
      "learning_rate": 0.0003030494615825826,
      "loss": 2.2017,
      "step": 114456
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.9592286944389343,
      "learning_rate": 0.0003030453712314608,
      "loss": 1.8902,
      "step": 114457
    },
    {
      "epoch": 1.49,
      "grad_norm": 1.2011018991470337,
      "learning_rate": 0.00030304128087977277,
      "loss": 2.0633,
      "step": 114458
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.9020569324493408,
      "learning_rate": 0.0003030371905275193,
      "loss": 2.1996,
      "step": 114459
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.993445634841919,
      "learning_rate": 0.00030303310017470124,
      "loss": 2.091,
      "step": 114460
    },
    {
      "epoch": 1.49,
      "grad_norm": 1.2395440340042114,
      "learning_rate": 0.0003030290098213192,
      "loss": 2.0002,
      "step": 114461
    },
    {
      "epoch": 1.49,
      "grad_norm": 1.0020370483398438,
      "learning_rate": 0.000303024919467374,
      "loss": 2.0712,
      "step": 114462
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.8916625380516052,
      "learning_rate": 0.00030302082911286645,
      "loss": 2.322,
      "step": 114463
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.9251944422721863,
      "learning_rate": 0.0003030167387577972,
      "loss": 2.2991,
      "step": 114464
    },
    {
      "epoch": 1.49,
      "grad_norm": 1.1142277717590332,
      "learning_rate": 0.00030301264840216715,
      "loss": 2.227,
      "step": 114465
    },
    {
      "epoch": 1.49,
      "grad_norm": 1.2334914207458496,
      "learning_rate": 0.000303008558045977,
      "loss": 1.9722,
      "step": 114466
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.9219930171966553,
      "learning_rate": 0.0003030044676892274,
      "loss": 2.0987,
      "step": 114467
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.9537620544433594,
      "learning_rate": 0.0003030003773319193,
      "loss": 2.2377,
      "step": 114468
    },
    {
      "epoch": 1.49,
      "grad_norm": 1.1396678686141968,
      "learning_rate": 0.00030299628697405334,
      "loss": 2.0122,
      "step": 114469
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.9724523425102234,
      "learning_rate": 0.0003029921966156304,
      "loss": 2.0526,
      "step": 114470
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.8625677824020386,
      "learning_rate": 0.00030298810625665107,
      "loss": 2.1859,
      "step": 114471
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.9485688209533691,
      "learning_rate": 0.00030298401589711615,
      "loss": 1.952,
      "step": 114472
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.9787748456001282,
      "learning_rate": 0.0003029799255370265,
      "loss": 2.1833,
      "step": 114473
    },
    {
      "epoch": 1.49,
      "grad_norm": 1.0840660333633423,
      "learning_rate": 0.00030297583517638284,
      "loss": 2.0764,
      "step": 114474
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.8952726125717163,
      "learning_rate": 0.0003029717448151859,
      "loss": 2.2039,
      "step": 114475
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.8890434503555298,
      "learning_rate": 0.00030296765445343655,
      "loss": 2.1721,
      "step": 114476
    },
    {
      "epoch": 1.49,
      "grad_norm": 1.0855107307434082,
      "learning_rate": 0.00030296356409113535,
      "loss": 2.006,
      "step": 114477
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.9081493020057678,
      "learning_rate": 0.0003029594737282831,
      "loss": 1.7715,
      "step": 114478
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.9151971936225891,
      "learning_rate": 0.0003029553833648808,
      "loss": 2.2423,
      "step": 114479
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.8593620657920837,
      "learning_rate": 0.0003029512930009289,
      "loss": 2.1702,
      "step": 114480
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.912149965763092,
      "learning_rate": 0.00030294720263642833,
      "loss": 2.3348,
      "step": 114481
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.9992585182189941,
      "learning_rate": 0.0003029431122713799,
      "loss": 2.0018,
      "step": 114482
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.9639028906822205,
      "learning_rate": 0.0003029390219057842,
      "loss": 2.2788,
      "step": 114483
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.8444043397903442,
      "learning_rate": 0.0003029349315396421,
      "loss": 2.2521,
      "step": 114484
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.9078467488288879,
      "learning_rate": 0.0003029308411729544,
      "loss": 2.0147,
      "step": 114485
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.927690327167511,
      "learning_rate": 0.00030292675080572176,
      "loss": 2.1329,
      "step": 114486
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.8832613229751587,
      "learning_rate": 0.0003029226604379449,
      "loss": 2.2806,
      "step": 114487
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.9166162014007568,
      "learning_rate": 0.00030291857006962475,
      "loss": 2.2979,
      "step": 114488
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.9547159671783447,
      "learning_rate": 0.00030291447970076207,
      "loss": 2.3125,
      "step": 114489
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.9214640259742737,
      "learning_rate": 0.0003029103893313574,
      "loss": 2.0155,
      "step": 114490
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.8802372217178345,
      "learning_rate": 0.00030290629896141165,
      "loss": 2.2438,
      "step": 114491
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.9570574164390564,
      "learning_rate": 0.0003029022085909256,
      "loss": 2.2625,
      "step": 114492
    },
    {
      "epoch": 1.49,
      "grad_norm": 1.0864410400390625,
      "learning_rate": 0.0003028981182199,
      "loss": 2.3175,
      "step": 114493
    },
    {
      "epoch": 1.49,
      "grad_norm": 1.0024464130401611,
      "learning_rate": 0.00030289402784833546,
      "loss": 2.0725,
      "step": 114494
    },
    {
      "epoch": 1.49,
      "grad_norm": 1.172741174697876,
      "learning_rate": 0.00030288993747623296,
      "loss": 2.2558,
      "step": 114495
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.9246134757995605,
      "learning_rate": 0.00030288584710359313,
      "loss": 2.2047,
      "step": 114496
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.9184611439704895,
      "learning_rate": 0.0003028817567304168,
      "loss": 2.0485,
      "step": 114497
    },
    {
      "epoch": 1.49,
      "grad_norm": 1.006898283958435,
      "learning_rate": 0.00030287766635670463,
      "loss": 2.0924,
      "step": 114498
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.8763327598571777,
      "learning_rate": 0.0003028735759824576,
      "loss": 2.3306,
      "step": 114499
    },
    {
      "epoch": 1.49,
      "grad_norm": 1.0300860404968262,
      "learning_rate": 0.0003028694856076762,
      "loss": 2.2007,
      "step": 114500
    },
    {
      "epoch": 1.49,
      "grad_norm": 1.0233811140060425,
      "learning_rate": 0.0003028653952323613,
      "loss": 2.1783,
      "step": 114501
    },
    {
      "epoch": 1.49,
      "grad_norm": 1.0195529460906982,
      "learning_rate": 0.0003028613048565137,
      "loss": 2.1986,
      "step": 114502
    },
    {
      "epoch": 1.49,
      "grad_norm": 1.0981744527816772,
      "learning_rate": 0.00030285721448013414,
      "loss": 2.0022,
      "step": 114503
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.868715226650238,
      "learning_rate": 0.00030285312410322337,
      "loss": 2.0442,
      "step": 114504
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.9389673471450806,
      "learning_rate": 0.00030284903372578215,
      "loss": 2.1373,
      "step": 114505
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.9877180457115173,
      "learning_rate": 0.00030284494334781125,
      "loss": 2.1453,
      "step": 114506
    },
    {
      "epoch": 1.49,
      "grad_norm": 1.5041522979736328,
      "learning_rate": 0.00030284085296931136,
      "loss": 2.3093,
      "step": 114507
    },
    {
      "epoch": 1.49,
      "grad_norm": 1.0357756614685059,
      "learning_rate": 0.00030283676259028336,
      "loss": 2.18,
      "step": 114508
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.8951950669288635,
      "learning_rate": 0.00030283267221072795,
      "loss": 2.017,
      "step": 114509
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.9632498025894165,
      "learning_rate": 0.00030282858183064583,
      "loss": 2.0572,
      "step": 114510
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.8753649592399597,
      "learning_rate": 0.0003028244914500379,
      "loss": 1.9513,
      "step": 114511
    },
    {
      "epoch": 1.49,
      "grad_norm": 1.0205284357070923,
      "learning_rate": 0.0003028204010689049,
      "loss": 2.4301,
      "step": 114512
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.9556796550750732,
      "learning_rate": 0.0003028163106872474,
      "loss": 2.184,
      "step": 114513
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.9478359818458557,
      "learning_rate": 0.00030281222030506636,
      "loss": 2.0724,
      "step": 114514
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.8313506245613098,
      "learning_rate": 0.0003028081299223625,
      "loss": 1.8011,
      "step": 114515
    },
    {
      "epoch": 1.49,
      "grad_norm": 1.2517063617706299,
      "learning_rate": 0.00030280403953913645,
      "loss": 2.0876,
      "step": 114516
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.8919593691825867,
      "learning_rate": 0.00030279994915538913,
      "loss": 2.0599,
      "step": 114517
    },
    {
      "epoch": 1.49,
      "grad_norm": 1.0022153854370117,
      "learning_rate": 0.0003027958587711213,
      "loss": 2.288,
      "step": 114518
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.941227912902832,
      "learning_rate": 0.00030279176838633363,
      "loss": 2.1227,
      "step": 114519
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.9350950121879578,
      "learning_rate": 0.0003027876780010269,
      "loss": 2.3712,
      "step": 114520
    },
    {
      "epoch": 1.49,
      "grad_norm": 1.157033920288086,
      "learning_rate": 0.00030278358761520193,
      "loss": 2.203,
      "step": 114521
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.9386400580406189,
      "learning_rate": 0.00030277949722885936,
      "loss": 2.1298,
      "step": 114522
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.9287115931510925,
      "learning_rate": 0.00030277540684200006,
      "loss": 1.8777,
      "step": 114523
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.8835870027542114,
      "learning_rate": 0.00030277131645462484,
      "loss": 1.9829,
      "step": 114524
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.9285120964050293,
      "learning_rate": 0.0003027672260667343,
      "loss": 2.2997,
      "step": 114525
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.9989203810691833,
      "learning_rate": 0.00030276313567832925,
      "loss": 2.1706,
      "step": 114526
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.9984504580497742,
      "learning_rate": 0.0003027590452894106,
      "loss": 2.027,
      "step": 114527
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.992789626121521,
      "learning_rate": 0.00030275495489997894,
      "loss": 2.501,
      "step": 114528
    },
    {
      "epoch": 1.49,
      "grad_norm": 1.0570863485336304,
      "learning_rate": 0.000302750864510035,
      "loss": 2.1972,
      "step": 114529
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.901562511920929,
      "learning_rate": 0.00030274677411957975,
      "loss": 1.9692,
      "step": 114530
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.8728730082511902,
      "learning_rate": 0.00030274268372861376,
      "loss": 2.0788,
      "step": 114531
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.9387785792350769,
      "learning_rate": 0.0003027385933371378,
      "loss": 2.1764,
      "step": 114532
    },
    {
      "epoch": 1.49,
      "grad_norm": 1.0357612371444702,
      "learning_rate": 0.00030273450294515273,
      "loss": 2.1853,
      "step": 114533
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.8576553463935852,
      "learning_rate": 0.0003027304125526593,
      "loss": 2.2067,
      "step": 114534
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.9091190695762634,
      "learning_rate": 0.0003027263221596582,
      "loss": 2.2836,
      "step": 114535
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.928898274898529,
      "learning_rate": 0.00030272223176615017,
      "loss": 2.115,
      "step": 114536
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.9235538840293884,
      "learning_rate": 0.0003027181413721361,
      "loss": 2.2148,
      "step": 114537
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.8999842405319214,
      "learning_rate": 0.0003027140509776167,
      "loss": 2.0357,
      "step": 114538
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.9922347664833069,
      "learning_rate": 0.0003027099605825926,
      "loss": 2.2183,
      "step": 114539
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.9580342769622803,
      "learning_rate": 0.0003027058701870648,
      "loss": 2.1002,
      "step": 114540
    },
    {
      "epoch": 1.49,
      "grad_norm": 1.2192490100860596,
      "learning_rate": 0.00030270177979103383,
      "loss": 2.1496,
      "step": 114541
    },
    {
      "epoch": 1.49,
      "grad_norm": 1.0465123653411865,
      "learning_rate": 0.0003026976893945006,
      "loss": 2.1054,
      "step": 114542
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.8909615874290466,
      "learning_rate": 0.00030269359899746587,
      "loss": 2.2522,
      "step": 114543
    },
    {
      "epoch": 1.49,
      "grad_norm": 1.0216879844665527,
      "learning_rate": 0.00030268950859993027,
      "loss": 2.005,
      "step": 114544
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.9846693873405457,
      "learning_rate": 0.0003026854182018946,
      "loss": 2.2114,
      "step": 114545
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.8654123544692993,
      "learning_rate": 0.0003026813278033597,
      "loss": 2.2454,
      "step": 114546
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.9977385401725769,
      "learning_rate": 0.00030267723740432634,
      "loss": 2.35,
      "step": 114547
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.8591299653053284,
      "learning_rate": 0.00030267314700479516,
      "loss": 1.9177,
      "step": 114548
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.9554347991943359,
      "learning_rate": 0.00030266905660476703,
      "loss": 2.1694,
      "step": 114549
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.8945866823196411,
      "learning_rate": 0.00030266496620424266,
      "loss": 2.2735,
      "step": 114550
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.9884565472602844,
      "learning_rate": 0.0003026608758032228,
      "loss": 2.1478,
      "step": 114551
    },
    {
      "epoch": 1.49,
      "grad_norm": 1.0030121803283691,
      "learning_rate": 0.00030265678540170835,
      "loss": 2.2574,
      "step": 114552
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.9303198456764221,
      "learning_rate": 0.0003026526949996998,
      "loss": 2.0344,
      "step": 114553
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.9043264389038086,
      "learning_rate": 0.00030264860459719813,
      "loss": 1.9505,
      "step": 114554
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.8669911026954651,
      "learning_rate": 0.00030264451419420406,
      "loss": 2.0938,
      "step": 114555
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.8878737092018127,
      "learning_rate": 0.00030264042379071825,
      "loss": 1.9993,
      "step": 114556
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.950258731842041,
      "learning_rate": 0.0003026363333867416,
      "loss": 2.2696,
      "step": 114557
    },
    {
      "epoch": 1.49,
      "grad_norm": 1.031087040901184,
      "learning_rate": 0.0003026322429822748,
      "loss": 2.3191,
      "step": 114558
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.9348403215408325,
      "learning_rate": 0.0003026281525773186,
      "loss": 2.0869,
      "step": 114559
    },
    {
      "epoch": 1.49,
      "grad_norm": 1.0427464246749878,
      "learning_rate": 0.00030262406217187375,
      "loss": 2.2515,
      "step": 114560
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.9184241890907288,
      "learning_rate": 0.0003026199717659411,
      "loss": 2.2239,
      "step": 114561
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.9990363121032715,
      "learning_rate": 0.0003026158813595213,
      "loss": 2.0391,
      "step": 114562
    },
    {
      "epoch": 1.49,
      "grad_norm": 1.0069247484207153,
      "learning_rate": 0.00030261179095261514,
      "loss": 2.3116,
      "step": 114563
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.8870548605918884,
      "learning_rate": 0.0003026077005452235,
      "loss": 2.1155,
      "step": 114564
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.8674653172492981,
      "learning_rate": 0.0003026036101373469,
      "loss": 2.1984,
      "step": 114565
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.9006378054618835,
      "learning_rate": 0.0003025995197289863,
      "loss": 2.0122,
      "step": 114566
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.9129522442817688,
      "learning_rate": 0.0003025954293201425,
      "loss": 2.1068,
      "step": 114567
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.9188675284385681,
      "learning_rate": 0.00030259133891081603,
      "loss": 2.1714,
      "step": 114568
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.892741858959198,
      "learning_rate": 0.00030258724850100785,
      "loss": 2.1586,
      "step": 114569
    },
    {
      "epoch": 1.49,
      "grad_norm": 1.2623497247695923,
      "learning_rate": 0.0003025831580907186,
      "loss": 2.1879,
      "step": 114570
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.8464720249176025,
      "learning_rate": 0.00030257906767994913,
      "loss": 2.0824,
      "step": 114571
    },
    {
      "epoch": 1.49,
      "grad_norm": 1.0175341367721558,
      "learning_rate": 0.00030257497726870016,
      "loss": 1.9041,
      "step": 114572
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.9202210307121277,
      "learning_rate": 0.00030257088685697246,
      "loss": 2.3583,
      "step": 114573
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.9791882634162903,
      "learning_rate": 0.00030256679644476673,
      "loss": 2.2031,
      "step": 114574
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.8586040139198303,
      "learning_rate": 0.00030256270603208385,
      "loss": 2.1284,
      "step": 114575
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.903446614742279,
      "learning_rate": 0.00030255861561892446,
      "loss": 2.2965,
      "step": 114576
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.9372923970222473,
      "learning_rate": 0.00030255452520528943,
      "loss": 2.1251,
      "step": 114577
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.9813227653503418,
      "learning_rate": 0.0003025504347911794,
      "loss": 2.1921,
      "step": 114578
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.8654890656471252,
      "learning_rate": 0.0003025463443765952,
      "loss": 1.9642,
      "step": 114579
    },
    {
      "epoch": 1.49,
      "grad_norm": 1.0535274744033813,
      "learning_rate": 0.0003025422539615377,
      "loss": 2.0117,
      "step": 114580
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.9789362549781799,
      "learning_rate": 0.0003025381635460074,
      "loss": 2.2197,
      "step": 114581
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.8948004245758057,
      "learning_rate": 0.0003025340731300053,
      "loss": 2.137,
      "step": 114582
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.8976302742958069,
      "learning_rate": 0.00030252998271353197,
      "loss": 2.243,
      "step": 114583
    },
    {
      "epoch": 1.49,
      "grad_norm": 1.0049344301223755,
      "learning_rate": 0.0003025258922965884,
      "loss": 2.0482,
      "step": 114584
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.8517959117889404,
      "learning_rate": 0.0003025218018791752,
      "loss": 2.1008,
      "step": 114585
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.8602052330970764,
      "learning_rate": 0.000302517711461293,
      "loss": 2.1968,
      "step": 114586
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.975145161151886,
      "learning_rate": 0.00030251362104294284,
      "loss": 2.2812,
      "step": 114587
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.9233893752098083,
      "learning_rate": 0.0003025095306241254,
      "loss": 2.1245,
      "step": 114588
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.8696410655975342,
      "learning_rate": 0.0003025054402048412,
      "loss": 1.9994,
      "step": 114589
    },
    {
      "epoch": 1.49,
      "grad_norm": 1.0039490461349487,
      "learning_rate": 0.00030250134978509135,
      "loss": 2.4849,
      "step": 114590
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.9133179783821106,
      "learning_rate": 0.00030249725936487646,
      "loss": 2.0532,
      "step": 114591
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.9286450743675232,
      "learning_rate": 0.0003024931689441972,
      "loss": 2.2994,
      "step": 114592
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.9052679538726807,
      "learning_rate": 0.0003024890785230544,
      "loss": 2.0395,
      "step": 114593
    },
    {
      "epoch": 1.49,
      "grad_norm": 1.1871124505996704,
      "learning_rate": 0.00030248498810144897,
      "loss": 2.2155,
      "step": 114594
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.9001741409301758,
      "learning_rate": 0.0003024808976793814,
      "loss": 2.0978,
      "step": 114595
    },
    {
      "epoch": 1.49,
      "grad_norm": 1.013670802116394,
      "learning_rate": 0.00030247680725685265,
      "loss": 1.8922,
      "step": 114596
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.9490973353385925,
      "learning_rate": 0.00030247271683386345,
      "loss": 2.0861,
      "step": 114597
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.9081099033355713,
      "learning_rate": 0.00030246862641041444,
      "loss": 2.2789,
      "step": 114598
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.9455366134643555,
      "learning_rate": 0.00030246453598650644,
      "loss": 2.2216,
      "step": 114599
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.927072286605835,
      "learning_rate": 0.00030246044556214035,
      "loss": 2.4333,
      "step": 114600
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.929092526435852,
      "learning_rate": 0.0003024563551373168,
      "loss": 2.2855,
      "step": 114601
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.9636548161506653,
      "learning_rate": 0.0003024522647120364,
      "loss": 1.9333,
      "step": 114602
    },
    {
      "epoch": 1.49,
      "grad_norm": 1.0278019905090332,
      "learning_rate": 0.0003024481742863002,
      "loss": 2.3092,
      "step": 114603
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.9243804812431335,
      "learning_rate": 0.0003024440838601089,
      "loss": 2.3968,
      "step": 114604
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.8848754167556763,
      "learning_rate": 0.00030243999343346315,
      "loss": 2.1578,
      "step": 114605
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.9337431788444519,
      "learning_rate": 0.0003024359030063637,
      "loss": 2.1792,
      "step": 114606
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.9763138890266418,
      "learning_rate": 0.0003024318125788115,
      "loss": 2.4132,
      "step": 114607
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.9670261740684509,
      "learning_rate": 0.00030242772215080706,
      "loss": 2.0545,
      "step": 114608
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.9443645477294922,
      "learning_rate": 0.0003024236317223513,
      "loss": 2.1649,
      "step": 114609
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.9195567965507507,
      "learning_rate": 0.00030241954129344497,
      "loss": 2.1937,
      "step": 114610
    },
    {
      "epoch": 1.49,
      "grad_norm": 1.0774924755096436,
      "learning_rate": 0.00030241545086408874,
      "loss": 2.309,
      "step": 114611
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.8881456255912781,
      "learning_rate": 0.0003024113604342834,
      "loss": 2.3081,
      "step": 114612
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.8971952199935913,
      "learning_rate": 0.0003024072700040299,
      "loss": 2.1498,
      "step": 114613
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.8364984393119812,
      "learning_rate": 0.0003024031795733287,
      "loss": 2.1851,
      "step": 114614
    },
    {
      "epoch": 1.49,
      "grad_norm": 1.0038033723831177,
      "learning_rate": 0.00030239908914218074,
      "loss": 2.2411,
      "step": 114615
    },
    {
      "epoch": 1.49,
      "grad_norm": 1.0900686979293823,
      "learning_rate": 0.0003023949987105868,
      "loss": 1.9929,
      "step": 114616
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.9634168744087219,
      "learning_rate": 0.00030239090827854757,
      "loss": 2.1111,
      "step": 114617
    },
    {
      "epoch": 1.49,
      "grad_norm": 1.4469712972640991,
      "learning_rate": 0.00030238681784606374,
      "loss": 2.1727,
      "step": 114618
    },
    {
      "epoch": 1.49,
      "grad_norm": 1.0413681268692017,
      "learning_rate": 0.00030238272741313626,
      "loss": 2.077,
      "step": 114619
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.9081717729568481,
      "learning_rate": 0.0003023786369797657,
      "loss": 2.1469,
      "step": 114620
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.9805376529693604,
      "learning_rate": 0.0003023745465459529,
      "loss": 2.3165,
      "step": 114621
    },
    {
      "epoch": 1.49,
      "grad_norm": 1.108976125717163,
      "learning_rate": 0.0003023704561116987,
      "loss": 2.0629,
      "step": 114622
    },
    {
      "epoch": 1.49,
      "grad_norm": 1.1056276559829712,
      "learning_rate": 0.00030236636567700374,
      "loss": 2.295,
      "step": 114623
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.8913140296936035,
      "learning_rate": 0.0003023622752418689,
      "loss": 2.2516,
      "step": 114624
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.9388232827186584,
      "learning_rate": 0.00030235818480629485,
      "loss": 2.1105,
      "step": 114625
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.9373959302902222,
      "learning_rate": 0.00030235409437028227,
      "loss": 2.1271,
      "step": 114626
    },
    {
      "epoch": 1.49,
      "grad_norm": 1.0836447477340698,
      "learning_rate": 0.0003023500039338321,
      "loss": 2.1809,
      "step": 114627
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.9499460458755493,
      "learning_rate": 0.00030234591349694506,
      "loss": 2.0516,
      "step": 114628
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.9201574325561523,
      "learning_rate": 0.0003023418230596218,
      "loss": 2.0538,
      "step": 114629
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.9012755155563354,
      "learning_rate": 0.00030233773262186315,
      "loss": 2.1317,
      "step": 114630
    },
    {
      "epoch": 1.49,
      "grad_norm": 1.1085894107818604,
      "learning_rate": 0.0003023336421836699,
      "loss": 2.1429,
      "step": 114631
    },
    {
      "epoch": 1.49,
      "grad_norm": 1.0199254751205444,
      "learning_rate": 0.00030232955174504277,
      "loss": 2.3926,
      "step": 114632
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.9265341758728027,
      "learning_rate": 0.0003023254613059825,
      "loss": 2.1578,
      "step": 114633
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.9796034693717957,
      "learning_rate": 0.0003023213708664899,
      "loss": 2.1756,
      "step": 114634
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.913025438785553,
      "learning_rate": 0.00030231728042656573,
      "loss": 2.1256,
      "step": 114635
    },
    {
      "epoch": 1.49,
      "grad_norm": 1.021978497505188,
      "learning_rate": 0.0003023131899862107,
      "loss": 2.0309,
      "step": 114636
    },
    {
      "epoch": 1.49,
      "grad_norm": 1.1127599477767944,
      "learning_rate": 0.0003023090995454257,
      "loss": 2.2395,
      "step": 114637
    },
    {
      "epoch": 1.49,
      "grad_norm": 1.076095700263977,
      "learning_rate": 0.00030230500910421136,
      "loss": 2.0998,
      "step": 114638
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.8529894948005676,
      "learning_rate": 0.0003023009186625684,
      "loss": 1.9596,
      "step": 114639
    },
    {
      "epoch": 1.49,
      "grad_norm": 1.0438047647476196,
      "learning_rate": 0.00030229682822049774,
      "loss": 2.2704,
      "step": 114640
    },
    {
      "epoch": 1.49,
      "grad_norm": 1.1418620347976685,
      "learning_rate": 0.00030229273777799995,
      "loss": 2.2535,
      "step": 114641
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.9724769592285156,
      "learning_rate": 0.000302288647335076,
      "loss": 2.014,
      "step": 114642
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.9792981147766113,
      "learning_rate": 0.0003022845568917266,
      "loss": 2.2921,
      "step": 114643
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.9554977416992188,
      "learning_rate": 0.00030228046644795224,
      "loss": 1.9829,
      "step": 114644
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.9450157284736633,
      "learning_rate": 0.00030227637600375404,
      "loss": 2.0215,
      "step": 114645
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.9122876524925232,
      "learning_rate": 0.0003022722855591327,
      "loss": 2.0035,
      "step": 114646
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.9546072483062744,
      "learning_rate": 0.00030226819511408874,
      "loss": 2.3705,
      "step": 114647
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.981776773929596,
      "learning_rate": 0.0003022641046686231,
      "loss": 2.1644,
      "step": 114648
    },
    {
      "epoch": 1.49,
      "grad_norm": 1.056251049041748,
      "learning_rate": 0.00030226001422273665,
      "loss": 1.961,
      "step": 114649
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.8569244146347046,
      "learning_rate": 0.0003022559237764299,
      "loss": 1.9164,
      "step": 114650
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.9069896936416626,
      "learning_rate": 0.00030225183332970376,
      "loss": 2.1996,
      "step": 114651
    },
    {
      "epoch": 1.49,
      "grad_norm": 1.0100922584533691,
      "learning_rate": 0.00030224774288255897,
      "loss": 2.106,
      "step": 114652
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.8999058604240417,
      "learning_rate": 0.00030224365243499633,
      "loss": 1.8618,
      "step": 114653
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.9513300061225891,
      "learning_rate": 0.00030223956198701643,
      "loss": 2.1366,
      "step": 114654
    },
    {
      "epoch": 1.49,
      "grad_norm": 1.0053616762161255,
      "learning_rate": 0.0003022354715386203,
      "loss": 2.2446,
      "step": 114655
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.9862312078475952,
      "learning_rate": 0.00030223138108980845,
      "loss": 2.0634,
      "step": 114656
    },
    {
      "epoch": 1.49,
      "grad_norm": 1.1023423671722412,
      "learning_rate": 0.0003022272906405817,
      "loss": 2.0408,
      "step": 114657
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.8357378840446472,
      "learning_rate": 0.00030222320019094104,
      "loss": 1.9978,
      "step": 114658
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.9254685044288635,
      "learning_rate": 0.0003022191097408869,
      "loss": 2.0846,
      "step": 114659
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.9800447225570679,
      "learning_rate": 0.00030221501929042017,
      "loss": 1.987,
      "step": 114660
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.9986166954040527,
      "learning_rate": 0.00030221092883954176,
      "loss": 2.3412,
      "step": 114661
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.9375085234642029,
      "learning_rate": 0.0003022068383882522,
      "loss": 2.2264,
      "step": 114662
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.9198312163352966,
      "learning_rate": 0.0003022027479365523,
      "loss": 2.0068,
      "step": 114663
    },
    {
      "epoch": 1.49,
      "grad_norm": 1.05466890335083,
      "learning_rate": 0.0003021986574844429,
      "loss": 2.2785,
      "step": 114664
    },
    {
      "epoch": 1.49,
      "grad_norm": 1.077404499053955,
      "learning_rate": 0.0003021945670319248,
      "loss": 2.4787,
      "step": 114665
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.9136008620262146,
      "learning_rate": 0.0003021904765789987,
      "loss": 2.357,
      "step": 114666
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.8829580545425415,
      "learning_rate": 0.0003021863861256652,
      "loss": 2.2531,
      "step": 114667
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.8942221403121948,
      "learning_rate": 0.00030218229567192527,
      "loss": 2.1302,
      "step": 114668
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.9369621872901917,
      "learning_rate": 0.0003021782052177797,
      "loss": 2.1988,
      "step": 114669
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.902634859085083,
      "learning_rate": 0.0003021741147632291,
      "loss": 2.0536,
      "step": 114670
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.9147995710372925,
      "learning_rate": 0.00030217002430827424,
      "loss": 2.2306,
      "step": 114671
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.8528440594673157,
      "learning_rate": 0.000302165933852916,
      "loss": 2.0723,
      "step": 114672
    },
    {
      "epoch": 1.49,
      "grad_norm": 1.2807507514953613,
      "learning_rate": 0.0003021618433971551,
      "loss": 2.1655,
      "step": 114673
    },
    {
      "epoch": 1.49,
      "grad_norm": 1.1210061311721802,
      "learning_rate": 0.0003021577529409922,
      "loss": 2.1563,
      "step": 114674
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.9583926200866699,
      "learning_rate": 0.00030215366248442815,
      "loss": 2.102,
      "step": 114675
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.9801188707351685,
      "learning_rate": 0.00030214957202746374,
      "loss": 2.2891,
      "step": 114676
    },
    {
      "epoch": 1.49,
      "grad_norm": 1.3050554990768433,
      "learning_rate": 0.00030214548157009964,
      "loss": 2.2044,
      "step": 114677
    },
    {
      "epoch": 1.49,
      "grad_norm": 1.0881831645965576,
      "learning_rate": 0.00030214139111233666,
      "loss": 2.2902,
      "step": 114678
    },
    {
      "epoch": 1.49,
      "grad_norm": 1.0531564950942993,
      "learning_rate": 0.00030213730065417556,
      "loss": 2.0887,
      "step": 114679
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.9225501418113708,
      "learning_rate": 0.00030213321019561704,
      "loss": 2.235,
      "step": 114680
    },
    {
      "epoch": 1.49,
      "grad_norm": 1.2190967798233032,
      "learning_rate": 0.000302129119736662,
      "loss": 2.035,
      "step": 114681
    },
    {
      "epoch": 1.49,
      "grad_norm": 1.0324101448059082,
      "learning_rate": 0.0003021250292773111,
      "loss": 2.193,
      "step": 114682
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.9652764201164246,
      "learning_rate": 0.0003021209388175651,
      "loss": 2.0359,
      "step": 114683
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.9189231395721436,
      "learning_rate": 0.00030211684835742473,
      "loss": 2.0458,
      "step": 114684
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.94534832239151,
      "learning_rate": 0.00030211275789689086,
      "loss": 1.7154,
      "step": 114685
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.9690901637077332,
      "learning_rate": 0.00030210866743596423,
      "loss": 2.3225,
      "step": 114686
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.9836858510971069,
      "learning_rate": 0.0003021045769746455,
      "loss": 2.2077,
      "step": 114687
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.9375689029693604,
      "learning_rate": 0.0003021004865129355,
      "loss": 2.3078,
      "step": 114688
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.9411876201629639,
      "learning_rate": 0.000302096396050835,
      "loss": 2.2336,
      "step": 114689
    },
    {
      "epoch": 1.49,
      "grad_norm": 1.2760941982269287,
      "learning_rate": 0.00030209230558834473,
      "loss": 2.2149,
      "step": 114690
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.909419059753418,
      "learning_rate": 0.00030208821512546535,
      "loss": 2.1267,
      "step": 114691
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.9091651439666748,
      "learning_rate": 0.0003020841246621979,
      "loss": 2.2399,
      "step": 114692
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.9920586347579956,
      "learning_rate": 0.00030208003419854285,
      "loss": 2.3085,
      "step": 114693
    },
    {
      "epoch": 1.49,
      "grad_norm": 1.0061012506484985,
      "learning_rate": 0.0003020759437345012,
      "loss": 2.2386,
      "step": 114694
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.8538260459899902,
      "learning_rate": 0.0003020718532700735,
      "loss": 2.1767,
      "step": 114695
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.8933423161506653,
      "learning_rate": 0.00030206776280526063,
      "loss": 2.2364,
      "step": 114696
    },
    {
      "epoch": 1.49,
      "grad_norm": 1.270911455154419,
      "learning_rate": 0.00030206367234006333,
      "loss": 1.9971,
      "step": 114697
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.9570777416229248,
      "learning_rate": 0.0003020595818744824,
      "loss": 2.2422,
      "step": 114698
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.8827462792396545,
      "learning_rate": 0.00030205549140851844,
      "loss": 2.0333,
      "step": 114699
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.933701753616333,
      "learning_rate": 0.00030205140094217243,
      "loss": 2.0671,
      "step": 114700
    },
    {
      "epoch": 1.49,
      "grad_norm": 1.2759617567062378,
      "learning_rate": 0.000302047310475445,
      "loss": 2.2492,
      "step": 114701
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.898610532283783,
      "learning_rate": 0.0003020432200083369,
      "loss": 2.0298,
      "step": 114702
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.9219159483909607,
      "learning_rate": 0.0003020391295408489,
      "loss": 1.8367,
      "step": 114703
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.9871102571487427,
      "learning_rate": 0.0003020350390729819,
      "loss": 2.0549,
      "step": 114704
    },
    {
      "epoch": 1.49,
      "grad_norm": 1.1066420078277588,
      "learning_rate": 0.00030203094860473644,
      "loss": 2.1437,
      "step": 114705
    },
    {
      "epoch": 1.49,
      "grad_norm": 1.1665499210357666,
      "learning_rate": 0.00030202685813611344,
      "loss": 2.1542,
      "step": 114706
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.7985915541648865,
      "learning_rate": 0.0003020227676671136,
      "loss": 2.0412,
      "step": 114707
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.9835879802703857,
      "learning_rate": 0.0003020186771977377,
      "loss": 2.1038,
      "step": 114708
    },
    {
      "epoch": 1.49,
      "grad_norm": 1.1751699447631836,
      "learning_rate": 0.0003020145867279864,
      "loss": 2.0241,
      "step": 114709
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.8992425799369812,
      "learning_rate": 0.0003020104962578608,
      "loss": 1.9741,
      "step": 114710
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.9498574733734131,
      "learning_rate": 0.0003020064057873612,
      "loss": 2.09,
      "step": 114711
    },
    {
      "epoch": 1.49,
      "grad_norm": 1.0535985231399536,
      "learning_rate": 0.0003020023153164886,
      "loss": 2.3988,
      "step": 114712
    },
    {
      "epoch": 1.49,
      "grad_norm": 1.0323858261108398,
      "learning_rate": 0.0003019982248452437,
      "loss": 2.0989,
      "step": 114713
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.9664367437362671,
      "learning_rate": 0.00030199413437362733,
      "loss": 2.015,
      "step": 114714
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.9075171947479248,
      "learning_rate": 0.0003019900439016402,
      "loss": 2.0987,
      "step": 114715
    },
    {
      "epoch": 1.49,
      "grad_norm": 1.0665040016174316,
      "learning_rate": 0.00030198595342928317,
      "loss": 2.1559,
      "step": 114716
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.9595471620559692,
      "learning_rate": 0.0003019818629565568,
      "loss": 2.1238,
      "step": 114717
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.8612207770347595,
      "learning_rate": 0.000301977772483462,
      "loss": 2.2643,
      "step": 114718
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.9515774250030518,
      "learning_rate": 0.0003019736820099995,
      "loss": 2.1542,
      "step": 114719
    },
    {
      "epoch": 1.49,
      "grad_norm": 1.0794923305511475,
      "learning_rate": 0.00030196959153617006,
      "loss": 2.2797,
      "step": 114720
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.896716296672821,
      "learning_rate": 0.0003019655010619744,
      "loss": 2.0595,
      "step": 114721
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.93654865026474,
      "learning_rate": 0.0003019614105874134,
      "loss": 2.0772,
      "step": 114722
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.9822242856025696,
      "learning_rate": 0.00030195732011248773,
      "loss": 2.1087,
      "step": 114723
    },
    {
      "epoch": 1.49,
      "grad_norm": 1.1812877655029297,
      "learning_rate": 0.000301953229637198,
      "loss": 2.2072,
      "step": 114724
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.959587574005127,
      "learning_rate": 0.0003019491391615454,
      "loss": 1.9581,
      "step": 114725
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.9622572064399719,
      "learning_rate": 0.0003019450486855302,
      "loss": 2.1312,
      "step": 114726
    },
    {
      "epoch": 1.49,
      "grad_norm": 1.0579959154129028,
      "learning_rate": 0.0003019409582091534,
      "loss": 2.0731,
      "step": 114727
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.8618935942649841,
      "learning_rate": 0.00030193686773241586,
      "loss": 2.0203,
      "step": 114728
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.8754889369010925,
      "learning_rate": 0.0003019327772553181,
      "loss": 2.1743,
      "step": 114729
    },
    {
      "epoch": 1.49,
      "grad_norm": 1.02760910987854,
      "learning_rate": 0.000301928686777861,
      "loss": 2.244,
      "step": 114730
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.8836427330970764,
      "learning_rate": 0.0003019245963000454,
      "loss": 2.1327,
      "step": 114731
    },
    {
      "epoch": 1.49,
      "grad_norm": 1.0875495672225952,
      "learning_rate": 0.000301920505821872,
      "loss": 2.1873,
      "step": 114732
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.8463454246520996,
      "learning_rate": 0.00030191641534334137,
      "loss": 2.0028,
      "step": 114733
    },
    {
      "epoch": 1.49,
      "grad_norm": 1.0062354803085327,
      "learning_rate": 0.00030191232486445467,
      "loss": 2.0738,
      "step": 114734
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.9551856517791748,
      "learning_rate": 0.00030190823438521223,
      "loss": 2.2516,
      "step": 114735
    },
    {
      "epoch": 1.49,
      "grad_norm": 1.01817786693573,
      "learning_rate": 0.0003019041439056151,
      "loss": 2.1532,
      "step": 114736
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.9604505896568298,
      "learning_rate": 0.0003019000534256639,
      "loss": 2.4425,
      "step": 114737
    },
    {
      "epoch": 1.49,
      "grad_norm": 1.1324071884155273,
      "learning_rate": 0.00030189596294535957,
      "loss": 2.4086,
      "step": 114738
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.9376509189605713,
      "learning_rate": 0.00030189187246470263,
      "loss": 2.084,
      "step": 114739
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.8462785482406616,
      "learning_rate": 0.00030188778198369403,
      "loss": 2.1618,
      "step": 114740
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.8869890570640564,
      "learning_rate": 0.0003018836915023344,
      "loss": 2.106,
      "step": 114741
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.9159530401229858,
      "learning_rate": 0.0003018796010206246,
      "loss": 2.252,
      "step": 114742
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.9139440059661865,
      "learning_rate": 0.0003018755105385653,
      "loss": 2.0538,
      "step": 114743
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.9461187124252319,
      "learning_rate": 0.0003018714200561574,
      "loss": 1.9375,
      "step": 114744
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.8866101503372192,
      "learning_rate": 0.00030186732957340144,
      "loss": 1.9833,
      "step": 114745
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.9258934259414673,
      "learning_rate": 0.00030186323909029843,
      "loss": 2.28,
      "step": 114746
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.99373859167099,
      "learning_rate": 0.0003018591486068489,
      "loss": 2.0494,
      "step": 114747
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.9640823006629944,
      "learning_rate": 0.00030185505812305375,
      "loss": 2.1137,
      "step": 114748
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.8711393475532532,
      "learning_rate": 0.00030185096763891376,
      "loss": 2.1406,
      "step": 114749
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.8837145566940308,
      "learning_rate": 0.00030184687715442964,
      "loss": 1.9372,
      "step": 114750
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.9227904081344604,
      "learning_rate": 0.0003018427866696021,
      "loss": 2.327,
      "step": 114751
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.9217538833618164,
      "learning_rate": 0.00030183869618443195,
      "loss": 1.9467,
      "step": 114752
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.9341856837272644,
      "learning_rate": 0.00030183460569892,
      "loss": 1.891,
      "step": 114753
    },
    {
      "epoch": 1.49,
      "grad_norm": 1.002967357635498,
      "learning_rate": 0.00030183051521306693,
      "loss": 2.1466,
      "step": 114754
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.9035923480987549,
      "learning_rate": 0.00030182642472687356,
      "loss": 2.3234,
      "step": 114755
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.8763911128044128,
      "learning_rate": 0.0003018223342403405,
      "loss": 2.063,
      "step": 114756
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.9534018039703369,
      "learning_rate": 0.00030181824375346883,
      "loss": 2.0129,
      "step": 114757
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.9843866229057312,
      "learning_rate": 0.00030181415326625897,
      "loss": 2.0651,
      "step": 114758
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.9512571096420288,
      "learning_rate": 0.0003018100627787119,
      "loss": 1.9903,
      "step": 114759
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.9202980399131775,
      "learning_rate": 0.0003018059722908282,
      "loss": 2.0295,
      "step": 114760
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.9955427646636963,
      "learning_rate": 0.00030180188180260883,
      "loss": 2.1724,
      "step": 114761
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.8907438516616821,
      "learning_rate": 0.00030179779131405445,
      "loss": 1.958,
      "step": 114762
    },
    {
      "epoch": 1.49,
      "grad_norm": 1.3510479927062988,
      "learning_rate": 0.0003017937008251658,
      "loss": 2.2067,
      "step": 114763
    },
    {
      "epoch": 1.49,
      "grad_norm": 1.0200119018554688,
      "learning_rate": 0.00030178961033594374,
      "loss": 2.0675,
      "step": 114764
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.8676044940948486,
      "learning_rate": 0.00030178551984638886,
      "loss": 2.0691,
      "step": 114765
    },
    {
      "epoch": 1.49,
      "grad_norm": 1.0063374042510986,
      "learning_rate": 0.00030178142935650206,
      "loss": 2.0721,
      "step": 114766
    },
    {
      "epoch": 1.49,
      "grad_norm": 1.0124154090881348,
      "learning_rate": 0.0003017773388662841,
      "loss": 1.8889,
      "step": 114767
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.8965994119644165,
      "learning_rate": 0.00030177324837573564,
      "loss": 2.124,
      "step": 114768
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.9428210854530334,
      "learning_rate": 0.0003017691578848575,
      "loss": 2.0544,
      "step": 114769
    },
    {
      "epoch": 1.49,
      "grad_norm": 1.2071030139923096,
      "learning_rate": 0.0003017650673936504,
      "loss": 2.4208,
      "step": 114770
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.9281139373779297,
      "learning_rate": 0.00030176097690211523,
      "loss": 2.18,
      "step": 114771
    },
    {
      "epoch": 1.49,
      "grad_norm": 1.030271291732788,
      "learning_rate": 0.00030175688641025263,
      "loss": 2.1604,
      "step": 114772
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.9309697151184082,
      "learning_rate": 0.00030175279591806336,
      "loss": 2.1899,
      "step": 114773
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.9851584434509277,
      "learning_rate": 0.0003017487054255483,
      "loss": 2.214,
      "step": 114774
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.9209199547767639,
      "learning_rate": 0.00030174461493270806,
      "loss": 1.9941,
      "step": 114775
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.8381190299987793,
      "learning_rate": 0.00030174052443954344,
      "loss": 2.2763,
      "step": 114776
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.9206399917602539,
      "learning_rate": 0.0003017364339460553,
      "loss": 2.0319,
      "step": 114777
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.9653092622756958,
      "learning_rate": 0.0003017323434522442,
      "loss": 2.219,
      "step": 114778
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.8984776735305786,
      "learning_rate": 0.0003017282529581112,
      "loss": 2.1149,
      "step": 114779
    },
    {
      "epoch": 1.49,
      "grad_norm": 1.0799065828323364,
      "learning_rate": 0.00030172416246365674,
      "loss": 2.1311,
      "step": 114780
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.9568418860435486,
      "learning_rate": 0.00030172007196888176,
      "loss": 2.0005,
      "step": 114781
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.8856550455093384,
      "learning_rate": 0.000301715981473787,
      "loss": 1.9568,
      "step": 114782
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.9890266060829163,
      "learning_rate": 0.0003017118909783732,
      "loss": 2.1714,
      "step": 114783
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.9782620668411255,
      "learning_rate": 0.0003017078004826411,
      "loss": 2.0974,
      "step": 114784
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.9692344665527344,
      "learning_rate": 0.0003017037099865915,
      "loss": 2.1791,
      "step": 114785
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.9453879594802856,
      "learning_rate": 0.00030169961949022523,
      "loss": 2.0905,
      "step": 114786
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.9427620768547058,
      "learning_rate": 0.0003016955289935429,
      "loss": 2.3056,
      "step": 114787
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.959417462348938,
      "learning_rate": 0.0003016914384965453,
      "loss": 2.1637,
      "step": 114788
    },
    {
      "epoch": 1.49,
      "grad_norm": 1.0009536743164062,
      "learning_rate": 0.00030168734799923335,
      "loss": 2.3263,
      "step": 114789
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.9710901379585266,
      "learning_rate": 0.00030168325750160757,
      "loss": 1.7555,
      "step": 114790
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.8769686222076416,
      "learning_rate": 0.0003016791670036689,
      "loss": 2.1915,
      "step": 114791
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.9593856930732727,
      "learning_rate": 0.0003016750765054181,
      "loss": 2.2084,
      "step": 114792
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.9203175902366638,
      "learning_rate": 0.00030167098600685573,
      "loss": 2.272,
      "step": 114793
    },
    {
      "epoch": 1.49,
      "grad_norm": 1.04694402217865,
      "learning_rate": 0.00030166689550798273,
      "loss": 2.159,
      "step": 114794
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.8987664580345154,
      "learning_rate": 0.00030166280500879987,
      "loss": 2.193,
      "step": 114795
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.9679075479507446,
      "learning_rate": 0.0003016587145093079,
      "loss": 2.201,
      "step": 114796
    },
    {
      "epoch": 1.49,
      "grad_norm": 1.040020227432251,
      "learning_rate": 0.0003016546240095074,
      "loss": 2.2395,
      "step": 114797
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.9022890329360962,
      "learning_rate": 0.00030165053350939946,
      "loss": 2.2525,
      "step": 114798
    },
    {
      "epoch": 1.49,
      "grad_norm": 1.0063586235046387,
      "learning_rate": 0.0003016464430089845,
      "loss": 2.16,
      "step": 114799
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.9131535887718201,
      "learning_rate": 0.00030164235250826354,
      "loss": 2.0536,
      "step": 114800
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.9114366173744202,
      "learning_rate": 0.0003016382620072372,
      "loss": 2.3216,
      "step": 114801
    },
    {
      "epoch": 1.49,
      "grad_norm": 1.1680192947387695,
      "learning_rate": 0.0003016341715059062,
      "loss": 2.1548,
      "step": 114802
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.8639420866966248,
      "learning_rate": 0.0003016300810042715,
      "loss": 2.3289,
      "step": 114803
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.9042161703109741,
      "learning_rate": 0.00030162599050233355,
      "loss": 1.9788,
      "step": 114804
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.9438378214836121,
      "learning_rate": 0.0003016219000000935,
      "loss": 2.0651,
      "step": 114805
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.8544789552688599,
      "learning_rate": 0.00030161780949755184,
      "loss": 2.2053,
      "step": 114806
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.9198015928268433,
      "learning_rate": 0.00030161371899470937,
      "loss": 2.0459,
      "step": 114807
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.9793235659599304,
      "learning_rate": 0.000301609628491567,
      "loss": 2.1693,
      "step": 114808
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.9126761555671692,
      "learning_rate": 0.0003016055379881252,
      "loss": 2.1785,
      "step": 114809
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.9393420219421387,
      "learning_rate": 0.00030160144748438497,
      "loss": 2.3361,
      "step": 114810
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.9681767225265503,
      "learning_rate": 0.00030159735698034705,
      "loss": 2.0819,
      "step": 114811
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.9033406972885132,
      "learning_rate": 0.00030159326647601206,
      "loss": 2.1666,
      "step": 114812
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.9943363666534424,
      "learning_rate": 0.00030158917597138086,
      "loss": 2.2419,
      "step": 114813
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.8916424512863159,
      "learning_rate": 0.00030158508546645433,
      "loss": 1.9536,
      "step": 114814
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.91972416639328,
      "learning_rate": 0.000301580994961233,
      "loss": 2.0664,
      "step": 114815
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.9473162293434143,
      "learning_rate": 0.0003015769044557177,
      "loss": 2.2922,
      "step": 114816
    },
    {
      "epoch": 1.49,
      "grad_norm": 0.9528825283050537,
      "learning_rate": 0.00030157281394990927,
      "loss": 2.3439,
      "step": 114817
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.8527663350105286,
      "learning_rate": 0.00030156872344380844,
      "loss": 2.112,
      "step": 114818
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.9824361801147461,
      "learning_rate": 0.0003015646329374159,
      "loss": 1.9946,
      "step": 114819
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.9120432138442993,
      "learning_rate": 0.0003015605424307326,
      "loss": 2.0528,
      "step": 114820
    },
    {
      "epoch": 1.5,
      "grad_norm": 1.025107741355896,
      "learning_rate": 0.00030155645192375896,
      "loss": 2.0274,
      "step": 114821
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.8854672312736511,
      "learning_rate": 0.0003015523614164961,
      "loss": 2.132,
      "step": 114822
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.9389398097991943,
      "learning_rate": 0.0003015482709089446,
      "loss": 1.9377,
      "step": 114823
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.9694712162017822,
      "learning_rate": 0.00030154418040110524,
      "loss": 2.003,
      "step": 114824
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.9794395565986633,
      "learning_rate": 0.00030154008989297874,
      "loss": 2.2736,
      "step": 114825
    },
    {
      "epoch": 1.5,
      "grad_norm": 1.1758607625961304,
      "learning_rate": 0.000301535999384566,
      "loss": 2.058,
      "step": 114826
    },
    {
      "epoch": 1.5,
      "grad_norm": 1.0479609966278076,
      "learning_rate": 0.0003015319088758676,
      "loss": 2.2048,
      "step": 114827
    },
    {
      "epoch": 1.5,
      "grad_norm": 1.2244802713394165,
      "learning_rate": 0.00030152781836688435,
      "loss": 2.3466,
      "step": 114828
    },
    {
      "epoch": 1.5,
      "grad_norm": 1.230452060699463,
      "learning_rate": 0.0003015237278576172,
      "loss": 2.0603,
      "step": 114829
    },
    {
      "epoch": 1.5,
      "grad_norm": 1.260624885559082,
      "learning_rate": 0.00030151963734806667,
      "loss": 2.0479,
      "step": 114830
    },
    {
      "epoch": 1.5,
      "grad_norm": 1.0143300294876099,
      "learning_rate": 0.0003015155468382336,
      "loss": 2.1162,
      "step": 114831
    },
    {
      "epoch": 1.5,
      "grad_norm": 1.060621738433838,
      "learning_rate": 0.0003015114563281188,
      "loss": 2.1246,
      "step": 114832
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.9731795191764832,
      "learning_rate": 0.00030150736581772297,
      "loss": 2.1075,
      "step": 114833
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.9420384168624878,
      "learning_rate": 0.0003015032753070469,
      "loss": 2.1056,
      "step": 114834
    },
    {
      "epoch": 1.5,
      "grad_norm": 1.0040262937545776,
      "learning_rate": 0.00030149918479609135,
      "loss": 2.1034,
      "step": 114835
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.9398223161697388,
      "learning_rate": 0.0003014950942848571,
      "loss": 1.9967,
      "step": 114836
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.9568459391593933,
      "learning_rate": 0.0003014910037733448,
      "loss": 1.9386,
      "step": 114837
    },
    {
      "epoch": 1.5,
      "grad_norm": 1.104252815246582,
      "learning_rate": 0.00030148691326155536,
      "loss": 2.1419,
      "step": 114838
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.9223071932792664,
      "learning_rate": 0.0003014828227494895,
      "loss": 2.2637,
      "step": 114839
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.8922308683395386,
      "learning_rate": 0.00030147873223714786,
      "loss": 2.0568,
      "step": 114840
    },
    {
      "epoch": 1.5,
      "grad_norm": 1.0387603044509888,
      "learning_rate": 0.0003014746417245313,
      "loss": 2.2013,
      "step": 114841
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.9953321218490601,
      "learning_rate": 0.0003014705512116407,
      "loss": 2.125,
      "step": 114842
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.9219198822975159,
      "learning_rate": 0.0003014664606984766,
      "loss": 2.247,
      "step": 114843
    },
    {
      "epoch": 1.5,
      "grad_norm": 1.0354305505752563,
      "learning_rate": 0.00030146237018503986,
      "loss": 2.301,
      "step": 114844
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.9730724096298218,
      "learning_rate": 0.0003014582796713313,
      "loss": 2.2843,
      "step": 114845
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.9651996493339539,
      "learning_rate": 0.00030145418915735156,
      "loss": 2.1493,
      "step": 114846
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.9112234115600586,
      "learning_rate": 0.00030145009864310143,
      "loss": 2.083,
      "step": 114847
    },
    {
      "epoch": 1.5,
      "grad_norm": 1.0427327156066895,
      "learning_rate": 0.0003014460081285817,
      "loss": 2.1072,
      "step": 114848
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.9932172894477844,
      "learning_rate": 0.0003014419176137932,
      "loss": 1.9848,
      "step": 114849
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.9065137505531311,
      "learning_rate": 0.0003014378270987366,
      "loss": 2.1011,
      "step": 114850
    },
    {
      "epoch": 1.5,
      "grad_norm": 1.242044448852539,
      "learning_rate": 0.00030143373658341263,
      "loss": 2.1902,
      "step": 114851
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.9575376510620117,
      "learning_rate": 0.0003014296460678222,
      "loss": 2.0122,
      "step": 114852
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.8524041771888733,
      "learning_rate": 0.00030142555555196594,
      "loss": 2.0858,
      "step": 114853
    },
    {
      "epoch": 1.5,
      "grad_norm": 1.1683043241500854,
      "learning_rate": 0.0003014214650358446,
      "loss": 2.3881,
      "step": 114854
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.9602593183517456,
      "learning_rate": 0.00030141737451945896,
      "loss": 2.2915,
      "step": 114855
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.8996516466140747,
      "learning_rate": 0.0003014132840028099,
      "loss": 1.977,
      "step": 114856
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.9046317934989929,
      "learning_rate": 0.00030140919348589797,
      "loss": 2.1675,
      "step": 114857
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.9742465615272522,
      "learning_rate": 0.00030140510296872413,
      "loss": 2.1314,
      "step": 114858
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.943676233291626,
      "learning_rate": 0.00030140101245128905,
      "loss": 2.096,
      "step": 114859
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.8716680407524109,
      "learning_rate": 0.0003013969219335935,
      "loss": 2.1167,
      "step": 114860
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.9884149432182312,
      "learning_rate": 0.0003013928314156381,
      "loss": 2.2262,
      "step": 114861
    },
    {
      "epoch": 1.5,
      "grad_norm": 1.146345615386963,
      "learning_rate": 0.00030138874089742396,
      "loss": 2.4092,
      "step": 114862
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.9348604679107666,
      "learning_rate": 0.00030138465037895146,
      "loss": 1.9945,
      "step": 114863
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.9845744371414185,
      "learning_rate": 0.0003013805598602216,
      "loss": 2.0869,
      "step": 114864
    },
    {
      "epoch": 1.5,
      "grad_norm": 1.0173835754394531,
      "learning_rate": 0.00030137646934123505,
      "loss": 2.0311,
      "step": 114865
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.8939930200576782,
      "learning_rate": 0.0003013723788219926,
      "loss": 2.2142,
      "step": 114866
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.8908735513687134,
      "learning_rate": 0.00030136828830249493,
      "loss": 2.3508,
      "step": 114867
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.9932344555854797,
      "learning_rate": 0.00030136419778274297,
      "loss": 2.2699,
      "step": 114868
    },
    {
      "epoch": 1.5,
      "grad_norm": 1.1496074199676514,
      "learning_rate": 0.0003013601072627374,
      "loss": 2.186,
      "step": 114869
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.9250467419624329,
      "learning_rate": 0.00030135601674247884,
      "loss": 2.2308,
      "step": 114870
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.9704396724700928,
      "learning_rate": 0.0003013519262219682,
      "loss": 2.0781,
      "step": 114871
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.9182001948356628,
      "learning_rate": 0.0003013478357012063,
      "loss": 2.1715,
      "step": 114872
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.945683479309082,
      "learning_rate": 0.0003013437451801938,
      "loss": 1.9244,
      "step": 114873
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.9618855118751526,
      "learning_rate": 0.00030133965465893136,
      "loss": 2.1224,
      "step": 114874
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.890731930732727,
      "learning_rate": 0.00030133556413741995,
      "loss": 2.1461,
      "step": 114875
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.996286153793335,
      "learning_rate": 0.0003013314736156602,
      "loss": 2.216,
      "step": 114876
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.9474579691886902,
      "learning_rate": 0.00030132738309365293,
      "loss": 2.0509,
      "step": 114877
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.8475950956344604,
      "learning_rate": 0.0003013232925713989,
      "loss": 2.1349,
      "step": 114878
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.961080014705658,
      "learning_rate": 0.00030131920204889875,
      "loss": 2.3232,
      "step": 114879
    },
    {
      "epoch": 1.5,
      "grad_norm": 1.0510849952697754,
      "learning_rate": 0.00030131511152615335,
      "loss": 2.1961,
      "step": 114880
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.9477749466896057,
      "learning_rate": 0.0003013110210031636,
      "loss": 2.0754,
      "step": 114881
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.9284480214118958,
      "learning_rate": 0.0003013069304799299,
      "loss": 2.073,
      "step": 114882
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.9848579168319702,
      "learning_rate": 0.0003013028399564533,
      "loss": 2.1148,
      "step": 114883
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.9844013452529907,
      "learning_rate": 0.00030129874943273454,
      "loss": 2.1424,
      "step": 114884
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.9186588525772095,
      "learning_rate": 0.0003012946589087742,
      "loss": 2.0588,
      "step": 114885
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.8824477195739746,
      "learning_rate": 0.00030129056838457325,
      "loss": 2.1866,
      "step": 114886
    },
    {
      "epoch": 1.5,
      "grad_norm": 1.0071228742599487,
      "learning_rate": 0.0003012864778601323,
      "loss": 2.1801,
      "step": 114887
    },
    {
      "epoch": 1.5,
      "grad_norm": 1.0073059797286987,
      "learning_rate": 0.0003012823873354522,
      "loss": 1.9706,
      "step": 114888
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.8734235763549805,
      "learning_rate": 0.0003012782968105337,
      "loss": 2.1464,
      "step": 114889
    },
    {
      "epoch": 1.5,
      "grad_norm": 1.0936232805252075,
      "learning_rate": 0.0003012742062853776,
      "loss": 2.0866,
      "step": 114890
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.9030892848968506,
      "learning_rate": 0.0003012701157599844,
      "loss": 2.0938,
      "step": 114891
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.930999755859375,
      "learning_rate": 0.0003012660252343552,
      "loss": 1.9838,
      "step": 114892
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.9219316840171814,
      "learning_rate": 0.00030126193470849066,
      "loss": 2.1034,
      "step": 114893
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.8935225605964661,
      "learning_rate": 0.0003012578441823914,
      "loss": 2.1215,
      "step": 114894
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.8999153971672058,
      "learning_rate": 0.0003012537536560584,
      "loss": 1.9504,
      "step": 114895
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.965457558631897,
      "learning_rate": 0.0003012496631294922,
      "loss": 2.3192,
      "step": 114896
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.9409674406051636,
      "learning_rate": 0.0003012455726026937,
      "loss": 2.1901,
      "step": 114897
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.9410709738731384,
      "learning_rate": 0.00030124148207566357,
      "loss": 1.9703,
      "step": 114898
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.9146561026573181,
      "learning_rate": 0.00030123739154840274,
      "loss": 1.9445,
      "step": 114899
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.9430860877037048,
      "learning_rate": 0.00030123330102091174,
      "loss": 2.016,
      "step": 114900
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.9938691854476929,
      "learning_rate": 0.00030122921049319153,
      "loss": 2.2617,
      "step": 114901
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.8763149380683899,
      "learning_rate": 0.0003012251199652427,
      "loss": 2.1329,
      "step": 114902
    },
    {
      "epoch": 1.5,
      "grad_norm": 1.0683848857879639,
      "learning_rate": 0.0003012210294370662,
      "loss": 2.2577,
      "step": 114903
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.9059509634971619,
      "learning_rate": 0.0003012169389086626,
      "loss": 2.0149,
      "step": 114904
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.8785931468009949,
      "learning_rate": 0.0003012128483800328,
      "loss": 2.082,
      "step": 114905
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.9499231576919556,
      "learning_rate": 0.0003012087578511775,
      "loss": 2.263,
      "step": 114906
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.9259073734283447,
      "learning_rate": 0.00030120466732209744,
      "loss": 2.2757,
      "step": 114907
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.9591177105903625,
      "learning_rate": 0.0003012005767927935,
      "loss": 2.3839,
      "step": 114908
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.9070026874542236,
      "learning_rate": 0.00030119648626326625,
      "loss": 2.2547,
      "step": 114909
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.8739011883735657,
      "learning_rate": 0.0003011923957335165,
      "loss": 2.1448,
      "step": 114910
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.9898881316184998,
      "learning_rate": 0.00030118830520354525,
      "loss": 2.2413,
      "step": 114911
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.9317692518234253,
      "learning_rate": 0.00030118421467335296,
      "loss": 2.1427,
      "step": 114912
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.948602020740509,
      "learning_rate": 0.0003011801241429405,
      "loss": 2.1802,
      "step": 114913
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.882172167301178,
      "learning_rate": 0.0003011760336123087,
      "loss": 2.2368,
      "step": 114914
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.8982679843902588,
      "learning_rate": 0.0003011719430814581,
      "loss": 2.0391,
      "step": 114915
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.9905283451080322,
      "learning_rate": 0.0003011678525503897,
      "loss": 2.1048,
      "step": 114916
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.8980445265769958,
      "learning_rate": 0.0003011637620191042,
      "loss": 2.4111,
      "step": 114917
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.9671405553817749,
      "learning_rate": 0.00030115967148760235,
      "loss": 2.1723,
      "step": 114918
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.9453816413879395,
      "learning_rate": 0.0003011555809558848,
      "loss": 2.0491,
      "step": 114919
    },
    {
      "epoch": 1.5,
      "grad_norm": 1.007635235786438,
      "learning_rate": 0.0003011514904239526,
      "loss": 1.9156,
      "step": 114920
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.8627616167068481,
      "learning_rate": 0.0003011473998918061,
      "loss": 2.1889,
      "step": 114921
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.8755806088447571,
      "learning_rate": 0.00030114330935944634,
      "loss": 2.1577,
      "step": 114922
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.895085871219635,
      "learning_rate": 0.00030113921882687407,
      "loss": 2.0149,
      "step": 114923
    },
    {
      "epoch": 1.5,
      "grad_norm": 1.0109223127365112,
      "learning_rate": 0.00030113512829409,
      "loss": 2.0306,
      "step": 114924
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.9725669026374817,
      "learning_rate": 0.00030113103776109476,
      "loss": 2.3228,
      "step": 114925
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.8620812892913818,
      "learning_rate": 0.0003011269472278894,
      "loss": 2.1642,
      "step": 114926
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.9413319230079651,
      "learning_rate": 0.00030112285669447444,
      "loss": 1.9833,
      "step": 114927
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.9823824763298035,
      "learning_rate": 0.00030111876616085076,
      "loss": 2.1392,
      "step": 114928
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.971954345703125,
      "learning_rate": 0.000301114675627019,
      "loss": 2.0713,
      "step": 114929
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.9057415723800659,
      "learning_rate": 0.00030111058509298014,
      "loss": 1.8813,
      "step": 114930
    },
    {
      "epoch": 1.5,
      "grad_norm": 1.0406602621078491,
      "learning_rate": 0.00030110649455873466,
      "loss": 2.2392,
      "step": 114931
    },
    {
      "epoch": 1.5,
      "grad_norm": 1.1925288438796997,
      "learning_rate": 0.0003011024040242835,
      "loss": 2.071,
      "step": 114932
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.9465140104293823,
      "learning_rate": 0.00030109831348962735,
      "loss": 2.024,
      "step": 114933
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.8948136568069458,
      "learning_rate": 0.00030109422295476713,
      "loss": 1.9975,
      "step": 114934
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.9434895515441895,
      "learning_rate": 0.0003010901324197033,
      "loss": 2.1311,
      "step": 114935
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.9880699515342712,
      "learning_rate": 0.0003010860418844369,
      "loss": 2.3345,
      "step": 114936
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.9633211493492126,
      "learning_rate": 0.0003010819513489686,
      "loss": 2.1603,
      "step": 114937
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.901611864566803,
      "learning_rate": 0.00030107786081329913,
      "loss": 1.987,
      "step": 114938
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.9193994998931885,
      "learning_rate": 0.00030107377027742925,
      "loss": 2.064,
      "step": 114939
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.8668777942657471,
      "learning_rate": 0.0003010696797413597,
      "loss": 2.0321,
      "step": 114940
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.8628178238868713,
      "learning_rate": 0.00030106558920509134,
      "loss": 1.9543,
      "step": 114941
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.8923507332801819,
      "learning_rate": 0.0003010614986686249,
      "loss": 2.1629,
      "step": 114942
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.9900293946266174,
      "learning_rate": 0.0003010574081319609,
      "loss": 2.0152,
      "step": 114943
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.8721905946731567,
      "learning_rate": 0.0003010533175951006,
      "loss": 2.111,
      "step": 114944
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.9465391635894775,
      "learning_rate": 0.0003010492270580443,
      "loss": 2.2941,
      "step": 114945
    },
    {
      "epoch": 1.5,
      "grad_norm": 1.0255759954452515,
      "learning_rate": 0.0003010451365207929,
      "loss": 2.1907,
      "step": 114946
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.9289298057556152,
      "learning_rate": 0.00030104104598334717,
      "loss": 1.9063,
      "step": 114947
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.922394871711731,
      "learning_rate": 0.00030103695544570803,
      "loss": 2.2222,
      "step": 114948
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.9129411578178406,
      "learning_rate": 0.0003010328649078761,
      "loss": 2.0827,
      "step": 114949
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.8626911640167236,
      "learning_rate": 0.00030102877436985205,
      "loss": 1.968,
      "step": 114950
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.9159062504768372,
      "learning_rate": 0.0003010246838316367,
      "loss": 2.2009,
      "step": 114951
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.9195860028266907,
      "learning_rate": 0.00030102059329323094,
      "loss": 2.0933,
      "step": 114952
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.8913083672523499,
      "learning_rate": 0.0003010165027546354,
      "loss": 2.0604,
      "step": 114953
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.9741479754447937,
      "learning_rate": 0.0003010124122158509,
      "loss": 2.2055,
      "step": 114954
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.8721741437911987,
      "learning_rate": 0.0003010083216768781,
      "loss": 2.165,
      "step": 114955
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.9540786743164062,
      "learning_rate": 0.00030100423113771786,
      "loss": 2.229,
      "step": 114956
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.9152566194534302,
      "learning_rate": 0.00030100014059837094,
      "loss": 2.0389,
      "step": 114957
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.9341561198234558,
      "learning_rate": 0.00030099605005883803,
      "loss": 2.1148,
      "step": 114958
    },
    {
      "epoch": 1.5,
      "grad_norm": 1.0095431804656982,
      "learning_rate": 0.00030099195951911994,
      "loss": 2.0676,
      "step": 114959
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.9378328919410706,
      "learning_rate": 0.0003009878689792175,
      "loss": 1.9681,
      "step": 114960
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.8548330664634705,
      "learning_rate": 0.0003009837784391313,
      "loss": 2.1139,
      "step": 114961
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.9299091100692749,
      "learning_rate": 0.00030097968789886227,
      "loss": 2.1779,
      "step": 114962
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.941646158695221,
      "learning_rate": 0.0003009755973584111,
      "loss": 2.2042,
      "step": 114963
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.9858647584915161,
      "learning_rate": 0.0003009715068177785,
      "loss": 2.086,
      "step": 114964
    },
    {
      "epoch": 1.5,
      "grad_norm": 1.184526801109314,
      "learning_rate": 0.0003009674162769653,
      "loss": 2.2188,
      "step": 114965
    },
    {
      "epoch": 1.5,
      "grad_norm": 1.3085309267044067,
      "learning_rate": 0.0003009633257359723,
      "loss": 2.2872,
      "step": 114966
    },
    {
      "epoch": 1.5,
      "grad_norm": 1.007195234298706,
      "learning_rate": 0.0003009592351948001,
      "loss": 2.2071,
      "step": 114967
    },
    {
      "epoch": 1.5,
      "grad_norm": 1.1038137674331665,
      "learning_rate": 0.0003009551446534496,
      "loss": 2.188,
      "step": 114968
    },
    {
      "epoch": 1.5,
      "grad_norm": 1.0703890323638916,
      "learning_rate": 0.0003009510541119215,
      "loss": 2.1701,
      "step": 114969
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.9288315773010254,
      "learning_rate": 0.0003009469635702166,
      "loss": 1.9828,
      "step": 114970
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.9443056583404541,
      "learning_rate": 0.00030094287302833567,
      "loss": 2.2037,
      "step": 114971
    },
    {
      "epoch": 1.5,
      "grad_norm": 1.0065374374389648,
      "learning_rate": 0.00030093878248627944,
      "loss": 2.0999,
      "step": 114972
    },
    {
      "epoch": 1.5,
      "grad_norm": 1.0772613286972046,
      "learning_rate": 0.0003009346919440486,
      "loss": 2.2673,
      "step": 114973
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.88843834400177,
      "learning_rate": 0.000300930601401644,
      "loss": 2.1231,
      "step": 114974
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.9211437106132507,
      "learning_rate": 0.0003009265108590665,
      "loss": 2.175,
      "step": 114975
    },
    {
      "epoch": 1.5,
      "grad_norm": 1.0309333801269531,
      "learning_rate": 0.00030092242031631654,
      "loss": 2.2405,
      "step": 114976
    },
    {
      "epoch": 1.5,
      "grad_norm": 1.10887610912323,
      "learning_rate": 0.00030091832977339523,
      "loss": 2.1634,
      "step": 114977
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.8400879502296448,
      "learning_rate": 0.0003009142392303032,
      "loss": 2.0983,
      "step": 114978
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.953438937664032,
      "learning_rate": 0.00030091014868704113,
      "loss": 2.0524,
      "step": 114979
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.9486235976219177,
      "learning_rate": 0.00030090605814360986,
      "loss": 2.1791,
      "step": 114980
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.9881986379623413,
      "learning_rate": 0.0003009019676000102,
      "loss": 2.3165,
      "step": 114981
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.891463577747345,
      "learning_rate": 0.00030089787705624274,
      "loss": 2.1748,
      "step": 114982
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.9784747958183289,
      "learning_rate": 0.0003008937865123084,
      "loss": 2.3944,
      "step": 114983
    },
    {
      "epoch": 1.5,
      "grad_norm": 1.3959492444992065,
      "learning_rate": 0.0003008896959682079,
      "loss": 1.9372,
      "step": 114984
    },
    {
      "epoch": 1.5,
      "grad_norm": 1.171139121055603,
      "learning_rate": 0.000300885605423942,
      "loss": 2.0574,
      "step": 114985
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.9092457294464111,
      "learning_rate": 0.0003008815148795114,
      "loss": 2.1447,
      "step": 114986
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.9061804413795471,
      "learning_rate": 0.000300877424334917,
      "loss": 2.0939,
      "step": 114987
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.9436395764350891,
      "learning_rate": 0.00030087333379015936,
      "loss": 2.1067,
      "step": 114988
    },
    {
      "epoch": 1.5,
      "grad_norm": 1.2644020318984985,
      "learning_rate": 0.0003008692432452394,
      "loss": 2.0791,
      "step": 114989
    },
    {
      "epoch": 1.5,
      "grad_norm": 1.4556105136871338,
      "learning_rate": 0.00030086515270015784,
      "loss": 2.1526,
      "step": 114990
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.9564626812934875,
      "learning_rate": 0.0003008610621549154,
      "loss": 2.2689,
      "step": 114991
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.9069240689277649,
      "learning_rate": 0.0003008569716095129,
      "loss": 2.1717,
      "step": 114992
    },
    {
      "epoch": 1.5,
      "grad_norm": 1.1655642986297607,
      "learning_rate": 0.00030085288106395107,
      "loss": 2.1971,
      "step": 114993
    },
    {
      "epoch": 1.5,
      "grad_norm": 1.1065945625305176,
      "learning_rate": 0.00030084879051823067,
      "loss": 2.3014,
      "step": 114994
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.9286707639694214,
      "learning_rate": 0.00030084469997235247,
      "loss": 2.1284,
      "step": 114995
    },
    {
      "epoch": 1.5,
      "grad_norm": 1.0104700326919556,
      "learning_rate": 0.0003008406094263172,
      "loss": 2.1857,
      "step": 114996
    },
    {
      "epoch": 1.5,
      "grad_norm": 1.0416789054870605,
      "learning_rate": 0.0003008365188801257,
      "loss": 2.2284,
      "step": 114997
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.907142698764801,
      "learning_rate": 0.0003008324283337786,
      "loss": 2.096,
      "step": 114998
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.9447965025901794,
      "learning_rate": 0.00030082833778727683,
      "loss": 1.9102,
      "step": 114999
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.9320218563079834,
      "learning_rate": 0.0003008242472406209,
      "loss": 2.0715,
      "step": 115000
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.9399852156639099,
      "learning_rate": 0.0003008201566938119,
      "loss": 1.9784,
      "step": 115001
    },
    {
      "epoch": 1.5,
      "grad_norm": 1.101963758468628,
      "learning_rate": 0.00030081606614685023,
      "loss": 2.065,
      "step": 115002
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.8658503293991089,
      "learning_rate": 0.000300811975599737,
      "loss": 2.2662,
      "step": 115003
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.9160240292549133,
      "learning_rate": 0.00030080788505247277,
      "loss": 2.2343,
      "step": 115004
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.9061889052391052,
      "learning_rate": 0.0003008037945050583,
      "loss": 1.8175,
      "step": 115005
    },
    {
      "epoch": 1.5,
      "grad_norm": 1.0100867748260498,
      "learning_rate": 0.0003007997039574945,
      "loss": 2.2616,
      "step": 115006
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.9458931088447571,
      "learning_rate": 0.00030079561340978187,
      "loss": 2.2631,
      "step": 115007
    },
    {
      "epoch": 1.5,
      "grad_norm": 1.0015678405761719,
      "learning_rate": 0.00030079152286192136,
      "loss": 2.0609,
      "step": 115008
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.9504387974739075,
      "learning_rate": 0.0003007874323139138,
      "loss": 2.2645,
      "step": 115009
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.8695573210716248,
      "learning_rate": 0.00030078334176575973,
      "loss": 2.1516,
      "step": 115010
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.9556170105934143,
      "learning_rate": 0.00030077925121746007,
      "loss": 2.0442,
      "step": 115011
    },
    {
      "epoch": 1.5,
      "grad_norm": 1.0733509063720703,
      "learning_rate": 0.00030077516066901556,
      "loss": 2.2415,
      "step": 115012
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.8494805693626404,
      "learning_rate": 0.00030077107012042685,
      "loss": 2.2053,
      "step": 115013
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.9756481051445007,
      "learning_rate": 0.0003007669795716948,
      "loss": 2.2174,
      "step": 115014
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.918289840221405,
      "learning_rate": 0.0003007628890228202,
      "loss": 2.2793,
      "step": 115015
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.9768126606941223,
      "learning_rate": 0.00030075879847380377,
      "loss": 2.056,
      "step": 115016
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.8926668763160706,
      "learning_rate": 0.0003007547079246462,
      "loss": 2.0826,
      "step": 115017
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.8610455989837646,
      "learning_rate": 0.0003007506173753484,
      "loss": 2.1745,
      "step": 115018
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.9138947129249573,
      "learning_rate": 0.000300746526825911,
      "loss": 2.1657,
      "step": 115019
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.8867126107215881,
      "learning_rate": 0.0003007424362763348,
      "loss": 2.1804,
      "step": 115020
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.938784122467041,
      "learning_rate": 0.0003007383457266205,
      "loss": 2.0543,
      "step": 115021
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.9740110635757446,
      "learning_rate": 0.000300734255176769,
      "loss": 2.3345,
      "step": 115022
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.8683182001113892,
      "learning_rate": 0.000300730164626781,
      "loss": 2.0043,
      "step": 115023
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.9447025060653687,
      "learning_rate": 0.0003007260740766572,
      "loss": 2.124,
      "step": 115024
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.9096827507019043,
      "learning_rate": 0.0003007219835263985,
      "loss": 2.0541,
      "step": 115025
    },
    {
      "epoch": 1.5,
      "grad_norm": 1.0432639122009277,
      "learning_rate": 0.00030071789297600546,
      "loss": 2.1871,
      "step": 115026
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.9290086030960083,
      "learning_rate": 0.00030071380242547907,
      "loss": 2.0598,
      "step": 115027
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.9923526048660278,
      "learning_rate": 0.00030070971187481986,
      "loss": 2.2855,
      "step": 115028
    },
    {
      "epoch": 1.5,
      "grad_norm": 1.0622879266738892,
      "learning_rate": 0.00030070562132402875,
      "loss": 2.2088,
      "step": 115029
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.9260026216506958,
      "learning_rate": 0.00030070153077310646,
      "loss": 2.257,
      "step": 115030
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.8984113335609436,
      "learning_rate": 0.0003006974402220537,
      "loss": 2.0537,
      "step": 115031
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.8843885660171509,
      "learning_rate": 0.00030069334967087124,
      "loss": 1.7996,
      "step": 115032
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.9705485701560974,
      "learning_rate": 0.00030068925911956004,
      "loss": 2.1235,
      "step": 115033
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.9222323298454285,
      "learning_rate": 0.00030068516856812053,
      "loss": 2.1416,
      "step": 115034
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.9072707295417786,
      "learning_rate": 0.00030068107801655363,
      "loss": 2.2165,
      "step": 115035
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.9256377220153809,
      "learning_rate": 0.0003006769874648603,
      "loss": 2.1947,
      "step": 115036
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.9873676896095276,
      "learning_rate": 0.0003006728969130409,
      "loss": 1.903,
      "step": 115037
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.9687700867652893,
      "learning_rate": 0.00030066880636109647,
      "loss": 2.5043,
      "step": 115038
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.9894160628318787,
      "learning_rate": 0.0003006647158090277,
      "loss": 2.0487,
      "step": 115039
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.9574583172798157,
      "learning_rate": 0.00030066062525683536,
      "loss": 2.0699,
      "step": 115040
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.9085373878479004,
      "learning_rate": 0.00030065653470452014,
      "loss": 2.0878,
      "step": 115041
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.9058589339256287,
      "learning_rate": 0.00030065244415208294,
      "loss": 2.1562,
      "step": 115042
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.9265099763870239,
      "learning_rate": 0.00030064835359952443,
      "loss": 2.0575,
      "step": 115043
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.9367458820343018,
      "learning_rate": 0.0003006442630468453,
      "loss": 2.2082,
      "step": 115044
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.8530774712562561,
      "learning_rate": 0.00030064017249404647,
      "loss": 2.1376,
      "step": 115045
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.9456209540367126,
      "learning_rate": 0.00030063608194112854,
      "loss": 2.0235,
      "step": 115046
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.9331833720207214,
      "learning_rate": 0.0003006319913880924,
      "loss": 2.2668,
      "step": 115047
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.9778884649276733,
      "learning_rate": 0.0003006279008349388,
      "loss": 2.2195,
      "step": 115048
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.9892917275428772,
      "learning_rate": 0.00030062381028166834,
      "loss": 2.1945,
      "step": 115049
    },
    {
      "epoch": 1.5,
      "grad_norm": 1.1273572444915771,
      "learning_rate": 0.000300619719728282,
      "loss": 2.2229,
      "step": 115050
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.9395032525062561,
      "learning_rate": 0.0003006156291747805,
      "loss": 2.2492,
      "step": 115051
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.9596287608146667,
      "learning_rate": 0.0003006115386211644,
      "loss": 2.0451,
      "step": 115052
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.9827518463134766,
      "learning_rate": 0.00030060744806743465,
      "loss": 2.2255,
      "step": 115053
    },
    {
      "epoch": 1.5,
      "grad_norm": 1.1544125080108643,
      "learning_rate": 0.00030060335751359206,
      "loss": 2.2304,
      "step": 115054
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.9101030230522156,
      "learning_rate": 0.00030059926695963715,
      "loss": 2.1348,
      "step": 115055
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.8903671503067017,
      "learning_rate": 0.0003005951764055709,
      "loss": 2.2058,
      "step": 115056
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.9405069947242737,
      "learning_rate": 0.00030059108585139404,
      "loss": 2.1901,
      "step": 115057
    },
    {
      "epoch": 1.5,
      "grad_norm": 1.0728379487991333,
      "learning_rate": 0.0003005869952971072,
      "loss": 2.1102,
      "step": 115058
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.9383402466773987,
      "learning_rate": 0.0003005829047427112,
      "loss": 2.2255,
      "step": 115059
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.9633511304855347,
      "learning_rate": 0.00030057881418820695,
      "loss": 2.1331,
      "step": 115060
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.9364238381385803,
      "learning_rate": 0.000300574723633595,
      "loss": 1.9703,
      "step": 115061
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.9797527194023132,
      "learning_rate": 0.0003005706330788762,
      "loss": 2.0458,
      "step": 115062
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.9007101655006409,
      "learning_rate": 0.0003005665425240514,
      "loss": 1.9292,
      "step": 115063
    },
    {
      "epoch": 1.5,
      "grad_norm": 1.078124761581421,
      "learning_rate": 0.00030056245196912113,
      "loss": 2.046,
      "step": 115064
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.9733180999755859,
      "learning_rate": 0.00030055836141408635,
      "loss": 2.1413,
      "step": 115065
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.8728650212287903,
      "learning_rate": 0.0003005542708589478,
      "loss": 2.1142,
      "step": 115066
    },
    {
      "epoch": 1.5,
      "grad_norm": 1.027660608291626,
      "learning_rate": 0.00030055018030370617,
      "loss": 2.1959,
      "step": 115067
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.9478792548179626,
      "learning_rate": 0.00030054608974836223,
      "loss": 2.2392,
      "step": 115068
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.898116409778595,
      "learning_rate": 0.0003005419991929167,
      "loss": 2.1265,
      "step": 115069
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.9535171985626221,
      "learning_rate": 0.00030053790863737056,
      "loss": 2.219,
      "step": 115070
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.9542102217674255,
      "learning_rate": 0.00030053381808172434,
      "loss": 2.1514,
      "step": 115071
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.9188642501831055,
      "learning_rate": 0.0003005297275259788,
      "loss": 1.9667,
      "step": 115072
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.8940027952194214,
      "learning_rate": 0.0003005256369701349,
      "loss": 2.2755,
      "step": 115073
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.9562808275222778,
      "learning_rate": 0.00030052154641419323,
      "loss": 2.3266,
      "step": 115074
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.9109407663345337,
      "learning_rate": 0.0003005174558581546,
      "loss": 2.0934,
      "step": 115075
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.8455527424812317,
      "learning_rate": 0.0003005133653020198,
      "loss": 2.133,
      "step": 115076
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.9943096041679382,
      "learning_rate": 0.00030050927474578945,
      "loss": 2.0376,
      "step": 115077
    },
    {
      "epoch": 1.5,
      "grad_norm": 1.007077693939209,
      "learning_rate": 0.0003005051841894644,
      "loss": 2.0953,
      "step": 115078
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.9366150498390198,
      "learning_rate": 0.0003005010936330456,
      "loss": 2.141,
      "step": 115079
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.9905830025672913,
      "learning_rate": 0.0003004970030765335,
      "loss": 2.1836,
      "step": 115080
    },
    {
      "epoch": 1.5,
      "grad_norm": 1.0355250835418701,
      "learning_rate": 0.000300492912519929,
      "loss": 2.306,
      "step": 115081
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.8809577822685242,
      "learning_rate": 0.0003004888219632329,
      "loss": 2.0572,
      "step": 115082
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.9437179565429688,
      "learning_rate": 0.0003004847314064459,
      "loss": 2.1043,
      "step": 115083
    },
    {
      "epoch": 1.5,
      "grad_norm": 1.0085771083831787,
      "learning_rate": 0.0003004806408495688,
      "loss": 2.2237,
      "step": 115084
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.9488162398338318,
      "learning_rate": 0.0003004765502926024,
      "loss": 2.4413,
      "step": 115085
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.9204079508781433,
      "learning_rate": 0.00030047245973554724,
      "loss": 2.0109,
      "step": 115086
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.963588297367096,
      "learning_rate": 0.00030046836917840434,
      "loss": 2.3132,
      "step": 115087
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.8812952041625977,
      "learning_rate": 0.00030046427862117443,
      "loss": 1.8216,
      "step": 115088
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.9386325478553772,
      "learning_rate": 0.00030046018806385806,
      "loss": 2.1589,
      "step": 115089
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.9575210809707642,
      "learning_rate": 0.00030045609750645614,
      "loss": 2.3314,
      "step": 115090
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.939970850944519,
      "learning_rate": 0.00030045200694896955,
      "loss": 2.0829,
      "step": 115091
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.971969485282898,
      "learning_rate": 0.0003004479163913988,
      "loss": 2.0547,
      "step": 115092
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.9120734930038452,
      "learning_rate": 0.0003004438258337448,
      "loss": 2.1081,
      "step": 115093
    },
    {
      "epoch": 1.5,
      "grad_norm": 1.0851472616195679,
      "learning_rate": 0.0003004397352760084,
      "loss": 2.093,
      "step": 115094
    },
    {
      "epoch": 1.5,
      "grad_norm": 1.0702322721481323,
      "learning_rate": 0.0003004356447181901,
      "loss": 2.1206,
      "step": 115095
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.9814991354942322,
      "learning_rate": 0.0003004315541602909,
      "loss": 1.994,
      "step": 115096
    },
    {
      "epoch": 1.5,
      "grad_norm": 1.0803359746932983,
      "learning_rate": 0.00030042746360231146,
      "loss": 2.0357,
      "step": 115097
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.9926577806472778,
      "learning_rate": 0.0003004233730442524,
      "loss": 2.2144,
      "step": 115098
    },
    {
      "epoch": 1.5,
      "grad_norm": 1.099528193473816,
      "learning_rate": 0.00030041928248611475,
      "loss": 2.2354,
      "step": 115099
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.8948862552642822,
      "learning_rate": 0.0003004151919278992,
      "loss": 2.0919,
      "step": 115100
    },
    {
      "epoch": 1.5,
      "grad_norm": 1.0054914951324463,
      "learning_rate": 0.0003004111013696064,
      "loss": 2.194,
      "step": 115101
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.865742564201355,
      "learning_rate": 0.0003004070108112371,
      "loss": 2.2116,
      "step": 115102
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.8629366159439087,
      "learning_rate": 0.0003004029202527922,
      "loss": 2.1606,
      "step": 115103
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.9435909986495972,
      "learning_rate": 0.0003003988296942724,
      "loss": 1.9538,
      "step": 115104
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.9465128183364868,
      "learning_rate": 0.00030039473913567844,
      "loss": 2.2776,
      "step": 115105
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.8771648406982422,
      "learning_rate": 0.0003003906485770111,
      "loss": 2.0962,
      "step": 115106
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.9349802732467651,
      "learning_rate": 0.0003003865580182711,
      "loss": 2.0734,
      "step": 115107
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.8875788450241089,
      "learning_rate": 0.00030038246745945924,
      "loss": 1.979,
      "step": 115108
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.9232232570648193,
      "learning_rate": 0.0003003783769005762,
      "loss": 2.1284,
      "step": 115109
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.919160783290863,
      "learning_rate": 0.00030037428634162295,
      "loss": 2.1434,
      "step": 115110
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.96847003698349,
      "learning_rate": 0.0003003701957826001,
      "loss": 1.9493,
      "step": 115111
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.9648962616920471,
      "learning_rate": 0.0003003661052235083,
      "loss": 2.0516,
      "step": 115112
    },
    {
      "epoch": 1.5,
      "grad_norm": 1.1409424543380737,
      "learning_rate": 0.0003003620146643485,
      "loss": 2.168,
      "step": 115113
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.9181715250015259,
      "learning_rate": 0.00030035792410512143,
      "loss": 2.3715,
      "step": 115114
    },
    {
      "epoch": 1.5,
      "grad_norm": 1.073742151260376,
      "learning_rate": 0.0003003538335458278,
      "loss": 2.1106,
      "step": 115115
    },
    {
      "epoch": 1.5,
      "grad_norm": 1.1228538751602173,
      "learning_rate": 0.0003003497429864683,
      "loss": 2.2614,
      "step": 115116
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.9485448002815247,
      "learning_rate": 0.0003003456524270439,
      "loss": 2.1326,
      "step": 115117
    },
    {
      "epoch": 1.5,
      "grad_norm": 1.1582231521606445,
      "learning_rate": 0.00030034156186755517,
      "loss": 2.2377,
      "step": 115118
    },
    {
      "epoch": 1.5,
      "grad_norm": 1.2910690307617188,
      "learning_rate": 0.00030033747130800295,
      "loss": 2.1824,
      "step": 115119
    },
    {
      "epoch": 1.5,
      "grad_norm": 1.0131880044937134,
      "learning_rate": 0.00030033338074838796,
      "loss": 2.0093,
      "step": 115120
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.9204328060150146,
      "learning_rate": 0.00030032929018871106,
      "loss": 2.1743,
      "step": 115121
    },
    {
      "epoch": 1.5,
      "grad_norm": 1.1791117191314697,
      "learning_rate": 0.00030032519962897285,
      "loss": 2.0866,
      "step": 115122
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.9813556671142578,
      "learning_rate": 0.00030032110906917425,
      "loss": 2.3357,
      "step": 115123
    },
    {
      "epoch": 1.5,
      "grad_norm": 1.0143191814422607,
      "learning_rate": 0.0003003170185093159,
      "loss": 2.1775,
      "step": 115124
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.9296038150787354,
      "learning_rate": 0.0003003129279493987,
      "loss": 1.9943,
      "step": 115125
    },
    {
      "epoch": 1.5,
      "grad_norm": 1.3022421598434448,
      "learning_rate": 0.0003003088373894232,
      "loss": 2.035,
      "step": 115126
    },
    {
      "epoch": 1.5,
      "grad_norm": 1.0220410823822021,
      "learning_rate": 0.0003003047468293904,
      "loss": 2.145,
      "step": 115127
    },
    {
      "epoch": 1.5,
      "grad_norm": 1.0043646097183228,
      "learning_rate": 0.0003003006562693008,
      "loss": 2.0819,
      "step": 115128
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.955478310585022,
      "learning_rate": 0.00030029656570915544,
      "loss": 2.1054,
      "step": 115129
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.9013214707374573,
      "learning_rate": 0.0003002924751489549,
      "loss": 2.0452,
      "step": 115130
    },
    {
      "epoch": 1.5,
      "grad_norm": 1.428981900215149,
      "learning_rate": 0.0003002883845886999,
      "loss": 1.9587,
      "step": 115131
    },
    {
      "epoch": 1.5,
      "grad_norm": 1.1324657201766968,
      "learning_rate": 0.0003002842940283914,
      "loss": 2.1225,
      "step": 115132
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.9441006779670715,
      "learning_rate": 0.00030028020346803,
      "loss": 2.4744,
      "step": 115133
    },
    {
      "epoch": 1.5,
      "grad_norm": 1.3457871675491333,
      "learning_rate": 0.0003002761129076165,
      "loss": 2.1208,
      "step": 115134
    },
    {
      "epoch": 1.5,
      "grad_norm": 1.0671449899673462,
      "learning_rate": 0.00030027202234715166,
      "loss": 2.0325,
      "step": 115135
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.928873598575592,
      "learning_rate": 0.00030026793178663624,
      "loss": 1.9971,
      "step": 115136
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.9172547459602356,
      "learning_rate": 0.0003002638412260711,
      "loss": 2.078,
      "step": 115137
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.9859578609466553,
      "learning_rate": 0.0003002597506654568,
      "loss": 2.1705,
      "step": 115138
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.9363625645637512,
      "learning_rate": 0.0003002556601047942,
      "loss": 2.2166,
      "step": 115139
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.92899489402771,
      "learning_rate": 0.0003002515695440841,
      "loss": 1.7866,
      "step": 115140
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.8678726553916931,
      "learning_rate": 0.00030024747898332724,
      "loss": 2.2145,
      "step": 115141
    },
    {
      "epoch": 1.5,
      "grad_norm": 1.001717448234558,
      "learning_rate": 0.0003002433884225244,
      "loss": 2.1816,
      "step": 115142
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.9755830764770508,
      "learning_rate": 0.00030023929786167624,
      "loss": 2.0478,
      "step": 115143
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.8874154686927795,
      "learning_rate": 0.0003002352073007837,
      "loss": 2.1322,
      "step": 115144
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.9263421297073364,
      "learning_rate": 0.00030023111673984726,
      "loss": 2.1551,
      "step": 115145
    },
    {
      "epoch": 1.5,
      "grad_norm": 1.041168451309204,
      "learning_rate": 0.000300227026178868,
      "loss": 2.2649,
      "step": 115146
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.9964205026626587,
      "learning_rate": 0.00030022293561784644,
      "loss": 2.3943,
      "step": 115147
    },
    {
      "epoch": 1.5,
      "grad_norm": 1.0676053762435913,
      "learning_rate": 0.00030021884505678344,
      "loss": 2.2731,
      "step": 115148
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.940761387348175,
      "learning_rate": 0.00030021475449567984,
      "loss": 2.1751,
      "step": 115149
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.9109944105148315,
      "learning_rate": 0.00030021066393453623,
      "loss": 2.0446,
      "step": 115150
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.9717761278152466,
      "learning_rate": 0.0003002065733733535,
      "loss": 2.3053,
      "step": 115151
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.9679592847824097,
      "learning_rate": 0.0003002024828121323,
      "loss": 2.2,
      "step": 115152
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.9898471236228943,
      "learning_rate": 0.00030019839225087355,
      "loss": 2.231,
      "step": 115153
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.8937551975250244,
      "learning_rate": 0.00030019430168957777,
      "loss": 1.9179,
      "step": 115154
    },
    {
      "epoch": 1.5,
      "grad_norm": 1.0122801065444946,
      "learning_rate": 0.00030019021112824605,
      "loss": 2.3032,
      "step": 115155
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.9431279897689819,
      "learning_rate": 0.00030018612056687883,
      "loss": 2.1556,
      "step": 115156
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.9129917025566101,
      "learning_rate": 0.000300182030005477,
      "loss": 2.2634,
      "step": 115157
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.9079676866531372,
      "learning_rate": 0.00030017793944404144,
      "loss": 1.9925,
      "step": 115158
    },
    {
      "epoch": 1.5,
      "grad_norm": 1.128796935081482,
      "learning_rate": 0.0003001738488825727,
      "loss": 2.219,
      "step": 115159
    },
    {
      "epoch": 1.5,
      "grad_norm": 1.0248215198516846,
      "learning_rate": 0.00030016975832107166,
      "loss": 2.0426,
      "step": 115160
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.9118326306343079,
      "learning_rate": 0.0003001656677595391,
      "loss": 2.0986,
      "step": 115161
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.9387507438659668,
      "learning_rate": 0.00030016157719797574,
      "loss": 2.0059,
      "step": 115162
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.8835942149162292,
      "learning_rate": 0.00030015748663638224,
      "loss": 2.31,
      "step": 115163
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.8848506212234497,
      "learning_rate": 0.0003001533960747596,
      "loss": 2.1236,
      "step": 115164
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.9696289896965027,
      "learning_rate": 0.0003001493055131083,
      "loss": 2.0655,
      "step": 115165
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.9248190522193909,
      "learning_rate": 0.0003001452149514293,
      "loss": 2.1951,
      "step": 115166
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.8699712157249451,
      "learning_rate": 0.0003001411243897234,
      "loss": 2.1376,
      "step": 115167
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.8872906565666199,
      "learning_rate": 0.00030013703382799115,
      "loss": 2.0066,
      "step": 115168
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.9174789786338806,
      "learning_rate": 0.00030013294326623345,
      "loss": 1.9415,
      "step": 115169
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.9498082399368286,
      "learning_rate": 0.0003001288527044511,
      "loss": 2.2331,
      "step": 115170
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.9308992028236389,
      "learning_rate": 0.00030012476214264465,
      "loss": 2.2013,
      "step": 115171
    },
    {
      "epoch": 1.5,
      "grad_norm": 1.2670631408691406,
      "learning_rate": 0.00030012067158081507,
      "loss": 2.1106,
      "step": 115172
    },
    {
      "epoch": 1.5,
      "grad_norm": 1.0833547115325928,
      "learning_rate": 0.0003001165810189631,
      "loss": 2.4972,
      "step": 115173
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.9279953241348267,
      "learning_rate": 0.0003001124904570894,
      "loss": 2.0504,
      "step": 115174
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.9051656723022461,
      "learning_rate": 0.00030010839989519475,
      "loss": 2.1559,
      "step": 115175
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.9688875079154968,
      "learning_rate": 0.00030010430933328006,
      "loss": 2.2996,
      "step": 115176
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.9016607999801636,
      "learning_rate": 0.00030010021877134585,
      "loss": 2.1962,
      "step": 115177
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.9221364259719849,
      "learning_rate": 0.0003000961282093931,
      "loss": 2.1367,
      "step": 115178
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.8908746838569641,
      "learning_rate": 0.00030009203764742246,
      "loss": 2.0428,
      "step": 115179
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.9322090744972229,
      "learning_rate": 0.00030008794708543475,
      "loss": 2.1033,
      "step": 115180
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.88961261510849,
      "learning_rate": 0.00030008385652343056,
      "loss": 2.1506,
      "step": 115181
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.9976523518562317,
      "learning_rate": 0.0003000797659614109,
      "loss": 2.2335,
      "step": 115182
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.9059234261512756,
      "learning_rate": 0.00030007567539937636,
      "loss": 2.1781,
      "step": 115183
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.8627252578735352,
      "learning_rate": 0.0003000715848373277,
      "loss": 2.0216,
      "step": 115184
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.9182007908821106,
      "learning_rate": 0.00030006749427526595,
      "loss": 2.13,
      "step": 115185
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.9952515363693237,
      "learning_rate": 0.0003000634037131914,
      "loss": 1.9376,
      "step": 115186
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.9143406748771667,
      "learning_rate": 0.0003000593131511051,
      "loss": 2.1556,
      "step": 115187
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.9054222106933594,
      "learning_rate": 0.00030005522258900796,
      "loss": 2.3348,
      "step": 115188
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.9964895248413086,
      "learning_rate": 0.0003000511320269004,
      "loss": 2.1676,
      "step": 115189
    },
    {
      "epoch": 1.5,
      "grad_norm": 1.1897376775741577,
      "learning_rate": 0.0003000470414647833,
      "loss": 2.3751,
      "step": 115190
    },
    {
      "epoch": 1.5,
      "grad_norm": 1.0549126863479614,
      "learning_rate": 0.0003000429509026576,
      "loss": 2.2143,
      "step": 115191
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.9765291810035706,
      "learning_rate": 0.0003000388603405238,
      "loss": 2.162,
      "step": 115192
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.8949524164199829,
      "learning_rate": 0.00030003476977838277,
      "loss": 2.0952,
      "step": 115193
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.9427266716957092,
      "learning_rate": 0.0003000306792162353,
      "loss": 2.309,
      "step": 115194
    },
    {
      "epoch": 1.5,
      "grad_norm": 1.0056849718093872,
      "learning_rate": 0.00030002658865408213,
      "loss": 2.1591,
      "step": 115195
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.9649830460548401,
      "learning_rate": 0.00030002249809192404,
      "loss": 2.1385,
      "step": 115196
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.9739794731140137,
      "learning_rate": 0.0003000184075297617,
      "loss": 2.4162,
      "step": 115197
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.9256700277328491,
      "learning_rate": 0.00030001431696759603,
      "loss": 1.9904,
      "step": 115198
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.9513025283813477,
      "learning_rate": 0.00030001022640542763,
      "loss": 2.0524,
      "step": 115199
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.9511580467224121,
      "learning_rate": 0.00030000613584325734,
      "loss": 2.3275,
      "step": 115200
    },
    {
      "epoch": 1.5,
      "grad_norm": 1.1454750299453735,
      "learning_rate": 0.00030000204528108585,
      "loss": 2.1534,
      "step": 115201
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.9557148218154907,
      "learning_rate": 0.0002999979547189141,
      "loss": 1.9279,
      "step": 115202
    },
    {
      "epoch": 1.5,
      "grad_norm": 1.0217829942703247,
      "learning_rate": 0.00029999386415674266,
      "loss": 2.4297,
      "step": 115203
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.8666059970855713,
      "learning_rate": 0.00029998977359457237,
      "loss": 2.0537,
      "step": 115204
    },
    {
      "epoch": 1.5,
      "grad_norm": 1.0167518854141235,
      "learning_rate": 0.000299985683032404,
      "loss": 2.1844,
      "step": 115205
    },
    {
      "epoch": 1.5,
      "grad_norm": 1.1922870874404907,
      "learning_rate": 0.0002999815924702383,
      "loss": 2.1328,
      "step": 115206
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.9369556307792664,
      "learning_rate": 0.00029997750190807596,
      "loss": 2.129,
      "step": 115207
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.8560112714767456,
      "learning_rate": 0.0002999734113459179,
      "loss": 2.1232,
      "step": 115208
    },
    {
      "epoch": 1.5,
      "grad_norm": 1.1866687536239624,
      "learning_rate": 0.0002999693207837647,
      "loss": 2.3678,
      "step": 115209
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.8383210897445679,
      "learning_rate": 0.00029996523022161723,
      "loss": 1.9766,
      "step": 115210
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.9025932550430298,
      "learning_rate": 0.0002999611396594762,
      "loss": 2.0266,
      "step": 115211
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.988393247127533,
      "learning_rate": 0.0002999570490973424,
      "loss": 2.1574,
      "step": 115212
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.9327038526535034,
      "learning_rate": 0.00029995295853521667,
      "loss": 2.3632,
      "step": 115213
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.912163257598877,
      "learning_rate": 0.00029994886797309967,
      "loss": 2.2164,
      "step": 115214
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.8676900267601013,
      "learning_rate": 0.0002999447774109921,
      "loss": 1.8215,
      "step": 115215
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.9390287399291992,
      "learning_rate": 0.0002999406868488948,
      "loss": 2.226,
      "step": 115216
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.9801586866378784,
      "learning_rate": 0.00029993659628680855,
      "loss": 2.1617,
      "step": 115217
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.955180287361145,
      "learning_rate": 0.0002999325057247341,
      "loss": 2.2378,
      "step": 115218
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.8390130996704102,
      "learning_rate": 0.0002999284151626721,
      "loss": 2.2646,
      "step": 115219
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.9313188195228577,
      "learning_rate": 0.00029992432460062364,
      "loss": 2.4195,
      "step": 115220
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.8985081911087036,
      "learning_rate": 0.00029992023403858914,
      "loss": 2.0603,
      "step": 115221
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.9029953479766846,
      "learning_rate": 0.0002999161434765694,
      "loss": 2.1037,
      "step": 115222
    },
    {
      "epoch": 1.5,
      "grad_norm": 1.0099228620529175,
      "learning_rate": 0.00029991205291456525,
      "loss": 1.955,
      "step": 115223
    },
    {
      "epoch": 1.5,
      "grad_norm": 1.0254311561584473,
      "learning_rate": 0.0002999079623525775,
      "loss": 2.1089,
      "step": 115224
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.9131647348403931,
      "learning_rate": 0.0002999038717906068,
      "loss": 1.9902,
      "step": 115225
    },
    {
      "epoch": 1.5,
      "grad_norm": 1.021350622177124,
      "learning_rate": 0.00029989978122865415,
      "loss": 2.5052,
      "step": 115226
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.9092223048210144,
      "learning_rate": 0.00029989569066672,
      "loss": 2.1467,
      "step": 115227
    },
    {
      "epoch": 1.5,
      "grad_norm": 1.0593538284301758,
      "learning_rate": 0.0002998916001048052,
      "loss": 2.3102,
      "step": 115228
    },
    {
      "epoch": 1.5,
      "grad_norm": 1.2001862525939941,
      "learning_rate": 0.0002998875095429106,
      "loss": 2.2068,
      "step": 115229
    },
    {
      "epoch": 1.5,
      "grad_norm": 1.0729751586914062,
      "learning_rate": 0.0002998834189810369,
      "loss": 1.855,
      "step": 115230
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.9792226552963257,
      "learning_rate": 0.0002998793284191849,
      "loss": 2.0965,
      "step": 115231
    },
    {
      "epoch": 1.5,
      "grad_norm": 1.0035462379455566,
      "learning_rate": 0.00029987523785735546,
      "loss": 2.0949,
      "step": 115232
    },
    {
      "epoch": 1.5,
      "grad_norm": 1.075332522392273,
      "learning_rate": 0.000299871147295549,
      "loss": 2.2495,
      "step": 115233
    },
    {
      "epoch": 1.5,
      "grad_norm": 1.0419862270355225,
      "learning_rate": 0.0002998670567337666,
      "loss": 2.3208,
      "step": 115234
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.9391641020774841,
      "learning_rate": 0.00029986296617200885,
      "loss": 1.9848,
      "step": 115235
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.9823607802391052,
      "learning_rate": 0.00029985887561027663,
      "loss": 2.2885,
      "step": 115236
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.9535461664199829,
      "learning_rate": 0.0002998547850485706,
      "loss": 1.9405,
      "step": 115237
    },
    {
      "epoch": 1.5,
      "grad_norm": 1.2354981899261475,
      "learning_rate": 0.0002998506944868916,
      "loss": 1.8983,
      "step": 115238
    },
    {
      "epoch": 1.5,
      "grad_norm": 1.138304352760315,
      "learning_rate": 0.0002998466039252405,
      "loss": 2.0102,
      "step": 115239
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.9528899788856506,
      "learning_rate": 0.00029984251336361776,
      "loss": 2.0046,
      "step": 115240
    },
    {
      "epoch": 1.5,
      "grad_norm": 1.3931496143341064,
      "learning_rate": 0.0002998384228020243,
      "loss": 2.1645,
      "step": 115241
    },
    {
      "epoch": 1.5,
      "grad_norm": 1.057183861732483,
      "learning_rate": 0.00029983433224046085,
      "loss": 2.4144,
      "step": 115242
    },
    {
      "epoch": 1.5,
      "grad_norm": 1.0855333805084229,
      "learning_rate": 0.0002998302416789283,
      "loss": 2.0339,
      "step": 115243
    },
    {
      "epoch": 1.5,
      "grad_norm": 1.13475501537323,
      "learning_rate": 0.0002998261511174272,
      "loss": 2.4016,
      "step": 115244
    },
    {
      "epoch": 1.5,
      "grad_norm": 1.3388742208480835,
      "learning_rate": 0.0002998220605559586,
      "loss": 2.0691,
      "step": 115245
    },
    {
      "epoch": 1.5,
      "grad_norm": 1.0358880758285522,
      "learning_rate": 0.000299817969994523,
      "loss": 2.0621,
      "step": 115246
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.9327198266983032,
      "learning_rate": 0.00029981387943312117,
      "loss": 2.208,
      "step": 115247
    },
    {
      "epoch": 1.5,
      "grad_norm": 1.1104793548583984,
      "learning_rate": 0.000299809788871754,
      "loss": 2.1234,
      "step": 115248
    },
    {
      "epoch": 1.5,
      "grad_norm": 1.0316294431686401,
      "learning_rate": 0.0002998056983104221,
      "loss": 2.1565,
      "step": 115249
    },
    {
      "epoch": 1.5,
      "grad_norm": 1.161666989326477,
      "learning_rate": 0.0002998016077491264,
      "loss": 2.159,
      "step": 115250
    },
    {
      "epoch": 1.5,
      "grad_norm": 1.0021355152130127,
      "learning_rate": 0.0002997975171878677,
      "loss": 2.0016,
      "step": 115251
    },
    {
      "epoch": 1.5,
      "grad_norm": 1.0790246725082397,
      "learning_rate": 0.0002997934266266465,
      "loss": 1.8786,
      "step": 115252
    },
    {
      "epoch": 1.5,
      "grad_norm": 1.0799158811569214,
      "learning_rate": 0.00029978933606546377,
      "loss": 2.1139,
      "step": 115253
    },
    {
      "epoch": 1.5,
      "grad_norm": 1.356812834739685,
      "learning_rate": 0.00029978524550432016,
      "loss": 2.1929,
      "step": 115254
    },
    {
      "epoch": 1.5,
      "grad_norm": 1.706784963607788,
      "learning_rate": 0.00029978115494321646,
      "loss": 2.2923,
      "step": 115255
    },
    {
      "epoch": 1.5,
      "grad_norm": 1.0030748844146729,
      "learning_rate": 0.0002997770643821535,
      "loss": 2.3266,
      "step": 115256
    },
    {
      "epoch": 1.5,
      "grad_norm": 1.1186914443969727,
      "learning_rate": 0.0002997729738211321,
      "loss": 2.0216,
      "step": 115257
    },
    {
      "epoch": 1.5,
      "grad_norm": 1.6594977378845215,
      "learning_rate": 0.00029976888326015274,
      "loss": 2.2701,
      "step": 115258
    },
    {
      "epoch": 1.5,
      "grad_norm": 1.1982146501541138,
      "learning_rate": 0.0002997647926992164,
      "loss": 2.0729,
      "step": 115259
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.8407056927680969,
      "learning_rate": 0.0002997607021383237,
      "loss": 2.1585,
      "step": 115260
    },
    {
      "epoch": 1.5,
      "grad_norm": 1.127399206161499,
      "learning_rate": 0.00029975661157747555,
      "loss": 2.1637,
      "step": 115261
    },
    {
      "epoch": 1.5,
      "grad_norm": 1.3661456108093262,
      "learning_rate": 0.00029975252101667265,
      "loss": 2.148,
      "step": 115262
    },
    {
      "epoch": 1.5,
      "grad_norm": 1.0740423202514648,
      "learning_rate": 0.0002997484304559159,
      "loss": 1.9828,
      "step": 115263
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.9438630938529968,
      "learning_rate": 0.0002997443398952058,
      "loss": 2.3015,
      "step": 115264
    },
    {
      "epoch": 1.5,
      "grad_norm": 1.1617357730865479,
      "learning_rate": 0.0002997402493345432,
      "loss": 2.078,
      "step": 115265
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.8788539171218872,
      "learning_rate": 0.0002997361587739289,
      "loss": 2.1967,
      "step": 115266
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.9382046461105347,
      "learning_rate": 0.00029973206821336365,
      "loss": 2.1873,
      "step": 115267
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.9234522581100464,
      "learning_rate": 0.0002997279776528483,
      "loss": 2.1349,
      "step": 115268
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.9847819805145264,
      "learning_rate": 0.00029972388709238355,
      "loss": 2.209,
      "step": 115269
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.9211099147796631,
      "learning_rate": 0.00029971979653197,
      "loss": 2.1127,
      "step": 115270
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.9020313620567322,
      "learning_rate": 0.0002997157059716086,
      "loss": 2.1039,
      "step": 115271
    },
    {
      "epoch": 1.5,
      "grad_norm": 1.0170072317123413,
      "learning_rate": 0.0002997116154113,
      "loss": 2.2834,
      "step": 115272
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.8797909617424011,
      "learning_rate": 0.0002997075248510451,
      "loss": 2.2381,
      "step": 115273
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.9450373649597168,
      "learning_rate": 0.00029970343429084446,
      "loss": 2.247,
      "step": 115274
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.926238477230072,
      "learning_rate": 0.00029969934373069913,
      "loss": 2.3272,
      "step": 115275
    },
    {
      "epoch": 1.5,
      "grad_norm": 1.0221045017242432,
      "learning_rate": 0.00029969525317060964,
      "loss": 2.2222,
      "step": 115276
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.9385473132133484,
      "learning_rate": 0.0002996911626105768,
      "loss": 2.2046,
      "step": 115277
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.9146486520767212,
      "learning_rate": 0.0002996870720506013,
      "loss": 2.0946,
      "step": 115278
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.8735153675079346,
      "learning_rate": 0.00029968298149068404,
      "loss": 2.0762,
      "step": 115279
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.9654298424720764,
      "learning_rate": 0.0002996788909308257,
      "loss": 2.1864,
      "step": 115280
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.9677399396896362,
      "learning_rate": 0.0002996748003710271,
      "loss": 2.1724,
      "step": 115281
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.956450343132019,
      "learning_rate": 0.00029967070981128894,
      "loss": 2.0394,
      "step": 115282
    },
    {
      "epoch": 1.5,
      "grad_norm": 1.000571608543396,
      "learning_rate": 0.00029966661925161204,
      "loss": 2.2915,
      "step": 115283
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.9054503440856934,
      "learning_rate": 0.00029966252869199705,
      "loss": 1.9923,
      "step": 115284
    },
    {
      "epoch": 1.5,
      "grad_norm": 1.0556282997131348,
      "learning_rate": 0.00029965843813244483,
      "loss": 2.2884,
      "step": 115285
    },
    {
      "epoch": 1.5,
      "grad_norm": 1.038976788520813,
      "learning_rate": 0.0002996543475729561,
      "loss": 2.2635,
      "step": 115286
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.9016218185424805,
      "learning_rate": 0.00029965025701353165,
      "loss": 1.9951,
      "step": 115287
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.904323399066925,
      "learning_rate": 0.0002996461664541722,
      "loss": 2.1606,
      "step": 115288
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.9562092423439026,
      "learning_rate": 0.00029964207589487857,
      "loss": 2.0392,
      "step": 115289
    },
    {
      "epoch": 1.5,
      "grad_norm": 1.1482501029968262,
      "learning_rate": 0.0002996379853356515,
      "loss": 2.2333,
      "step": 115290
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.946301281452179,
      "learning_rate": 0.0002996338947764917,
      "loss": 2.1386,
      "step": 115291
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.9647886157035828,
      "learning_rate": 0.00029962980421739996,
      "loss": 1.9774,
      "step": 115292
    },
    {
      "epoch": 1.5,
      "grad_norm": 1.4390610456466675,
      "learning_rate": 0.0002996257136583771,
      "loss": 2.0361,
      "step": 115293
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.9818374514579773,
      "learning_rate": 0.0002996216230994238,
      "loss": 2.0245,
      "step": 115294
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.9661641120910645,
      "learning_rate": 0.0002996175325405408,
      "loss": 2.0475,
      "step": 115295
    },
    {
      "epoch": 1.5,
      "grad_norm": 1.0713717937469482,
      "learning_rate": 0.0002996134419817289,
      "loss": 2.3033,
      "step": 115296
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.9011523723602295,
      "learning_rate": 0.0002996093514229889,
      "loss": 1.9877,
      "step": 115297
    },
    {
      "epoch": 1.5,
      "grad_norm": 1.1124756336212158,
      "learning_rate": 0.00029960526086432156,
      "loss": 2.0722,
      "step": 115298
    },
    {
      "epoch": 1.5,
      "grad_norm": 1.1190367937088013,
      "learning_rate": 0.00029960117030572767,
      "loss": 2.2576,
      "step": 115299
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.9846103191375732,
      "learning_rate": 0.0002995970797472078,
      "loss": 2.0857,
      "step": 115300
    },
    {
      "epoch": 1.5,
      "grad_norm": 1.0829057693481445,
      "learning_rate": 0.00029959298918876286,
      "loss": 2.1385,
      "step": 115301
    },
    {
      "epoch": 1.5,
      "grad_norm": 1.081889271736145,
      "learning_rate": 0.0002995888986303936,
      "loss": 2.1597,
      "step": 115302
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.9255036115646362,
      "learning_rate": 0.00029958480807210075,
      "loss": 2.0777,
      "step": 115303
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.955903172492981,
      "learning_rate": 0.00029958071751388514,
      "loss": 2.0731,
      "step": 115304
    },
    {
      "epoch": 1.5,
      "grad_norm": 1.1828092336654663,
      "learning_rate": 0.00029957662695574754,
      "loss": 2.1355,
      "step": 115305
    },
    {
      "epoch": 1.5,
      "grad_norm": 1.1611768007278442,
      "learning_rate": 0.00029957253639768865,
      "loss": 2.3148,
      "step": 115306
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.9220865368843079,
      "learning_rate": 0.0002995684458397091,
      "loss": 2.1031,
      "step": 115307
    },
    {
      "epoch": 1.5,
      "grad_norm": 1.013840913772583,
      "learning_rate": 0.0002995643552818099,
      "loss": 2.4994,
      "step": 115308
    },
    {
      "epoch": 1.5,
      "grad_norm": 1.1100661754608154,
      "learning_rate": 0.00029956026472399163,
      "loss": 2.1327,
      "step": 115309
    },
    {
      "epoch": 1.5,
      "grad_norm": 1.0968047380447388,
      "learning_rate": 0.00029955617416625507,
      "loss": 2.1447,
      "step": 115310
    },
    {
      "epoch": 1.5,
      "grad_norm": 1.0073020458221436,
      "learning_rate": 0.0002995520836086011,
      "loss": 2.0163,
      "step": 115311
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.969977855682373,
      "learning_rate": 0.0002995479930510305,
      "loss": 2.2228,
      "step": 115312
    },
    {
      "epoch": 1.5,
      "grad_norm": 1.0360517501831055,
      "learning_rate": 0.0002995439024935438,
      "loss": 2.1791,
      "step": 115313
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.9530768990516663,
      "learning_rate": 0.00029953981193614194,
      "loss": 1.9566,
      "step": 115314
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.9339377284049988,
      "learning_rate": 0.0002995357213788256,
      "loss": 2.2682,
      "step": 115315
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.9317567348480225,
      "learning_rate": 0.0002995316308215956,
      "loss": 2.2807,
      "step": 115316
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.920359194278717,
      "learning_rate": 0.00029952754026445265,
      "loss": 1.9353,
      "step": 115317
    },
    {
      "epoch": 1.5,
      "grad_norm": 1.106164574623108,
      "learning_rate": 0.0002995234497073977,
      "loss": 1.9333,
      "step": 115318
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.9438756108283997,
      "learning_rate": 0.00029951935915043123,
      "loss": 2.1839,
      "step": 115319
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.9191098809242249,
      "learning_rate": 0.00029951526859355406,
      "loss": 2.0966,
      "step": 115320
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.9552568197250366,
      "learning_rate": 0.00029951117803676705,
      "loss": 2.0221,
      "step": 115321
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.8849392533302307,
      "learning_rate": 0.00029950708748007095,
      "loss": 2.087,
      "step": 115322
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.9893139004707336,
      "learning_rate": 0.0002995029969234664,
      "loss": 2.1774,
      "step": 115323
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.9225935935974121,
      "learning_rate": 0.00029949890636695446,
      "loss": 2.1663,
      "step": 115324
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.9325065612792969,
      "learning_rate": 0.00029949481581053553,
      "loss": 1.9456,
      "step": 115325
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.9452085494995117,
      "learning_rate": 0.00029949072525421055,
      "loss": 2.0522,
      "step": 115326
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.891406774520874,
      "learning_rate": 0.0002994866346979802,
      "loss": 1.9263,
      "step": 115327
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.9851444363594055,
      "learning_rate": 0.00029948254414184536,
      "loss": 2.1817,
      "step": 115328
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.8419589400291443,
      "learning_rate": 0.0002994784535858067,
      "loss": 2.213,
      "step": 115329
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.9601923823356628,
      "learning_rate": 0.0002994743630298651,
      "loss": 2.1463,
      "step": 115330
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.9307589530944824,
      "learning_rate": 0.00029947027247402113,
      "loss": 2.1286,
      "step": 115331
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.9451586008071899,
      "learning_rate": 0.00029946618191827566,
      "loss": 2.0389,
      "step": 115332
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.9777810573577881,
      "learning_rate": 0.0002994620913626294,
      "loss": 2.1294,
      "step": 115333
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.8828592896461487,
      "learning_rate": 0.00029945800080708314,
      "loss": 1.9567,
      "step": 115334
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.8528532385826111,
      "learning_rate": 0.00029945391025163766,
      "loss": 2.0551,
      "step": 115335
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.9739284515380859,
      "learning_rate": 0.0002994498196962939,
      "loss": 2.2055,
      "step": 115336
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.904899537563324,
      "learning_rate": 0.00029944572914105224,
      "loss": 2.0873,
      "step": 115337
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.8988476991653442,
      "learning_rate": 0.00029944163858591365,
      "loss": 2.2865,
      "step": 115338
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.9800247550010681,
      "learning_rate": 0.0002994375480308788,
      "loss": 1.9214,
      "step": 115339
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.9802330136299133,
      "learning_rate": 0.0002994334574759486,
      "loss": 2.1099,
      "step": 115340
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.9438645243644714,
      "learning_rate": 0.00029942936692112373,
      "loss": 2.2159,
      "step": 115341
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.9201015830039978,
      "learning_rate": 0.00029942527636640505,
      "loss": 2.074,
      "step": 115342
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.898076593875885,
      "learning_rate": 0.0002994211858117931,
      "loss": 2.0956,
      "step": 115343
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.9623960256576538,
      "learning_rate": 0.0002994170952572888,
      "loss": 2.1789,
      "step": 115344
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.9047762155532837,
      "learning_rate": 0.0002994130047028928,
      "loss": 2.1384,
      "step": 115345
    },
    {
      "epoch": 1.5,
      "grad_norm": 1.0671982765197754,
      "learning_rate": 0.00029940891414860596,
      "loss": 2.2792,
      "step": 115346
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.9201804399490356,
      "learning_rate": 0.000299404823594429,
      "loss": 2.1125,
      "step": 115347
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.8680339455604553,
      "learning_rate": 0.0002994007330403629,
      "loss": 2.2809,
      "step": 115348
    },
    {
      "epoch": 1.5,
      "grad_norm": 1.095382809638977,
      "learning_rate": 0.00029939664248640805,
      "loss": 2.1403,
      "step": 115349
    },
    {
      "epoch": 1.5,
      "grad_norm": 1.1729745864868164,
      "learning_rate": 0.00029939255193256535,
      "loss": 2.1972,
      "step": 115350
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.9440702795982361,
      "learning_rate": 0.00029938846137883557,
      "loss": 1.994,
      "step": 115351
    },
    {
      "epoch": 1.5,
      "grad_norm": 1.0576938390731812,
      "learning_rate": 0.0002993843708252195,
      "loss": 2.2406,
      "step": 115352
    },
    {
      "epoch": 1.5,
      "grad_norm": 1.0011422634124756,
      "learning_rate": 0.0002993802802717179,
      "loss": 2.0074,
      "step": 115353
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.9527662396430969,
      "learning_rate": 0.00029937618971833166,
      "loss": 1.9905,
      "step": 115354
    },
    {
      "epoch": 1.5,
      "grad_norm": 1.027923822402954,
      "learning_rate": 0.00029937209916506126,
      "loss": 2.0702,
      "step": 115355
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.8879637718200684,
      "learning_rate": 0.00029936800861190757,
      "loss": 2.2345,
      "step": 115356
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.9544082283973694,
      "learning_rate": 0.0002993639180588714,
      "loss": 2.1206,
      "step": 115357
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.8867582678794861,
      "learning_rate": 0.00029935982750595354,
      "loss": 2.0196,
      "step": 115358
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.9178745746612549,
      "learning_rate": 0.00029935573695315465,
      "loss": 2.3113,
      "step": 115359
    },
    {
      "epoch": 1.5,
      "grad_norm": 1.0123846530914307,
      "learning_rate": 0.0002993516464004757,
      "loss": 2.0568,
      "step": 115360
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.9709290266036987,
      "learning_rate": 0.0002993475558479171,
      "loss": 2.2754,
      "step": 115361
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.9271772503852844,
      "learning_rate": 0.00029934346529547986,
      "loss": 2.3544,
      "step": 115362
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.8750124573707581,
      "learning_rate": 0.00029933937474316464,
      "loss": 1.8824,
      "step": 115363
    },
    {
      "epoch": 1.5,
      "grad_norm": 1.0263586044311523,
      "learning_rate": 0.00029933528419097226,
      "loss": 2.2706,
      "step": 115364
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.984054446220398,
      "learning_rate": 0.0002993311936389035,
      "loss": 2.1671,
      "step": 115365
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.8333266377449036,
      "learning_rate": 0.00029932710308695906,
      "loss": 2.0759,
      "step": 115366
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.9400597214698792,
      "learning_rate": 0.00029932301253513975,
      "loss": 2.2403,
      "step": 115367
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.9455261826515198,
      "learning_rate": 0.0002993189219834463,
      "loss": 2.0361,
      "step": 115368
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.9706529378890991,
      "learning_rate": 0.00029931483143187947,
      "loss": 2.0944,
      "step": 115369
    },
    {
      "epoch": 1.5,
      "grad_norm": 1.239138126373291,
      "learning_rate": 0.00029931074088043996,
      "loss": 2.1595,
      "step": 115370
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.9178680777549744,
      "learning_rate": 0.00029930665032912865,
      "loss": 2.3414,
      "step": 115371
    },
    {
      "epoch": 1.5,
      "grad_norm": 1.0083187818527222,
      "learning_rate": 0.0002993025597779462,
      "loss": 2.1431,
      "step": 115372
    },
    {
      "epoch": 1.5,
      "grad_norm": 1.3926441669464111,
      "learning_rate": 0.00029929846922689354,
      "loss": 2.3423,
      "step": 115373
    },
    {
      "epoch": 1.5,
      "grad_norm": 1.1425411701202393,
      "learning_rate": 0.00029929437867597125,
      "loss": 2.3059,
      "step": 115374
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.9879500865936279,
      "learning_rate": 0.00029929028812518014,
      "loss": 2.1828,
      "step": 115375
    },
    {
      "epoch": 1.5,
      "grad_norm": 1.2301764488220215,
      "learning_rate": 0.000299286197574521,
      "loss": 2.2923,
      "step": 115376
    },
    {
      "epoch": 1.5,
      "grad_norm": 1.066841721534729,
      "learning_rate": 0.0002992821070239945,
      "loss": 2.2286,
      "step": 115377
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.8857511281967163,
      "learning_rate": 0.0002992780164736015,
      "loss": 2.0257,
      "step": 115378
    },
    {
      "epoch": 1.5,
      "grad_norm": 1.0265685319900513,
      "learning_rate": 0.0002992739259233428,
      "loss": 2.0797,
      "step": 115379
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.9719879031181335,
      "learning_rate": 0.00029926983537321896,
      "loss": 2.1258,
      "step": 115380
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.9510841369628906,
      "learning_rate": 0.0002992657448232309,
      "loss": 1.9009,
      "step": 115381
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.9594328999519348,
      "learning_rate": 0.0002992616542733795,
      "loss": 2.1661,
      "step": 115382
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.8506733179092407,
      "learning_rate": 0.0002992575637236652,
      "loss": 1.8354,
      "step": 115383
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.9362798929214478,
      "learning_rate": 0.000299253473174089,
      "loss": 2.122,
      "step": 115384
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.8841789960861206,
      "learning_rate": 0.00029924938262465166,
      "loss": 2.0627,
      "step": 115385
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.9934126734733582,
      "learning_rate": 0.00029924529207535377,
      "loss": 2.195,
      "step": 115386
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.9985151290893555,
      "learning_rate": 0.0002992412015261963,
      "loss": 2.5514,
      "step": 115387
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.887042760848999,
      "learning_rate": 0.00029923711097717975,
      "loss": 2.1792,
      "step": 115388
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.9884932637214661,
      "learning_rate": 0.0002992330204283052,
      "loss": 2.0774,
      "step": 115389
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.907501220703125,
      "learning_rate": 0.0002992289298795731,
      "loss": 1.9224,
      "step": 115390
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.9582877159118652,
      "learning_rate": 0.0002992248393309845,
      "loss": 2.1423,
      "step": 115391
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.9422999024391174,
      "learning_rate": 0.00029922074878253993,
      "loss": 2.2126,
      "step": 115392
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.8921624422073364,
      "learning_rate": 0.00029921665823424027,
      "loss": 2.2646,
      "step": 115393
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.952536404132843,
      "learning_rate": 0.00029921256768608616,
      "loss": 2.1934,
      "step": 115394
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.8876084685325623,
      "learning_rate": 0.00029920847713807853,
      "loss": 2.1714,
      "step": 115395
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.8770256638526917,
      "learning_rate": 0.000299204386590218,
      "loss": 2.2351,
      "step": 115396
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.9172796010971069,
      "learning_rate": 0.00029920029604250557,
      "loss": 2.223,
      "step": 115397
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.9759392142295837,
      "learning_rate": 0.00029919620549494165,
      "loss": 1.9486,
      "step": 115398
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.9414121508598328,
      "learning_rate": 0.0002991921149475272,
      "loss": 2.3006,
      "step": 115399
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.9189326167106628,
      "learning_rate": 0.00029918802440026293,
      "loss": 2.235,
      "step": 115400
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.9622571468353271,
      "learning_rate": 0.0002991839338531496,
      "loss": 2.1883,
      "step": 115401
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.9485039710998535,
      "learning_rate": 0.00029917984330618807,
      "loss": 2.0889,
      "step": 115402
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.9802812337875366,
      "learning_rate": 0.0002991757527593791,
      "loss": 2.2529,
      "step": 115403
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.9655768871307373,
      "learning_rate": 0.00029917166221272323,
      "loss": 2.019,
      "step": 115404
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.9758340120315552,
      "learning_rate": 0.0002991675716662214,
      "loss": 2.1492,
      "step": 115405
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.9279466867446899,
      "learning_rate": 0.00029916348111987427,
      "loss": 2.1433,
      "step": 115406
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.8531813621520996,
      "learning_rate": 0.00029915939057368273,
      "loss": 2.364,
      "step": 115407
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.992713212966919,
      "learning_rate": 0.0002991553000276475,
      "loss": 2.2681,
      "step": 115408
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.9022204279899597,
      "learning_rate": 0.0002991512094817694,
      "loss": 2.0472,
      "step": 115409
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.864592969417572,
      "learning_rate": 0.00029914711893604893,
      "loss": 2.0786,
      "step": 115410
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.9699147343635559,
      "learning_rate": 0.0002991430283904871,
      "loss": 2.0192,
      "step": 115411
    },
    {
      "epoch": 1.5,
      "grad_norm": 1.068737268447876,
      "learning_rate": 0.0002991389378450846,
      "loss": 2.2247,
      "step": 115412
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.9125335216522217,
      "learning_rate": 0.0002991348472998421,
      "loss": 2.1605,
      "step": 115413
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.9761672019958496,
      "learning_rate": 0.0002991307567547605,
      "loss": 2.1599,
      "step": 115414
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.9609071612358093,
      "learning_rate": 0.0002991266662098407,
      "loss": 2.3055,
      "step": 115415
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.9248345494270325,
      "learning_rate": 0.00029912257566508307,
      "loss": 2.0541,
      "step": 115416
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.9299289584159851,
      "learning_rate": 0.00029911848512048856,
      "loss": 2.1461,
      "step": 115417
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.9144816398620605,
      "learning_rate": 0.000299114394576058,
      "loss": 2.1275,
      "step": 115418
    },
    {
      "epoch": 1.5,
      "grad_norm": 1.0244112014770508,
      "learning_rate": 0.00029911030403179204,
      "loss": 2.0863,
      "step": 115419
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.9292153120040894,
      "learning_rate": 0.0002991062134876915,
      "loss": 2.1279,
      "step": 115420
    },
    {
      "epoch": 1.5,
      "grad_norm": 1.109337568283081,
      "learning_rate": 0.0002991021229437573,
      "loss": 2.251,
      "step": 115421
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.9126441478729248,
      "learning_rate": 0.00029909803239998984,
      "loss": 2.1145,
      "step": 115422
    },
    {
      "epoch": 1.5,
      "grad_norm": 1.1175001859664917,
      "learning_rate": 0.0002990939418563901,
      "loss": 2.2632,
      "step": 115423
    },
    {
      "epoch": 1.5,
      "grad_norm": 1.0277414321899414,
      "learning_rate": 0.0002990898513129588,
      "loss": 1.9978,
      "step": 115424
    },
    {
      "epoch": 1.5,
      "grad_norm": 1.0239235162734985,
      "learning_rate": 0.0002990857607696968,
      "loss": 2.3436,
      "step": 115425
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.9054884314537048,
      "learning_rate": 0.00029908167022660466,
      "loss": 1.994,
      "step": 115426
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.9362607598304749,
      "learning_rate": 0.00029907757968368346,
      "loss": 2.1511,
      "step": 115427
    },
    {
      "epoch": 1.5,
      "grad_norm": 1.047209620475769,
      "learning_rate": 0.0002990734891409336,
      "loss": 2.1385,
      "step": 115428
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.9759203195571899,
      "learning_rate": 0.000299069398598356,
      "loss": 2.0113,
      "step": 115429
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.9134189486503601,
      "learning_rate": 0.0002990653080559514,
      "loss": 2.1566,
      "step": 115430
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.9351377487182617,
      "learning_rate": 0.00029906121751372057,
      "loss": 2.2855,
      "step": 115431
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.9718625545501709,
      "learning_rate": 0.0002990571269716643,
      "loss": 2.3242,
      "step": 115432
    },
    {
      "epoch": 1.5,
      "grad_norm": 1.1492595672607422,
      "learning_rate": 0.0002990530364297835,
      "loss": 2.0875,
      "step": 115433
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.9343027472496033,
      "learning_rate": 0.00029904894588807853,
      "loss": 2.1132,
      "step": 115434
    },
    {
      "epoch": 1.5,
      "grad_norm": 1.0495821237564087,
      "learning_rate": 0.0002990448553465504,
      "loss": 2.2939,
      "step": 115435
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.8829981088638306,
      "learning_rate": 0.0002990407648051999,
      "loss": 2.2082,
      "step": 115436
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.8987055420875549,
      "learning_rate": 0.0002990366742640277,
      "loss": 2.1967,
      "step": 115437
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.9475966691970825,
      "learning_rate": 0.0002990325837230346,
      "loss": 2.0373,
      "step": 115438
    },
    {
      "epoch": 1.5,
      "grad_norm": 1.011043667793274,
      "learning_rate": 0.0002990284931822214,
      "loss": 2.1628,
      "step": 115439
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.9389532804489136,
      "learning_rate": 0.00029902440264158896,
      "loss": 2.2101,
      "step": 115440
    },
    {
      "epoch": 1.5,
      "grad_norm": 1.1368787288665771,
      "learning_rate": 0.00029902031210113773,
      "loss": 2.1844,
      "step": 115441
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.9657465815544128,
      "learning_rate": 0.00029901622156086865,
      "loss": 2.322,
      "step": 115442
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.9590088725090027,
      "learning_rate": 0.00029901213102078247,
      "loss": 2.3311,
      "step": 115443
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.9722207188606262,
      "learning_rate": 0.00029900804048087995,
      "loss": 2.0426,
      "step": 115444
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.9935095906257629,
      "learning_rate": 0.00029900394994116186,
      "loss": 2.1274,
      "step": 115445
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.904230535030365,
      "learning_rate": 0.0002989998594016291,
      "loss": 2.1524,
      "step": 115446
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.8600566983222961,
      "learning_rate": 0.00029899576886228214,
      "loss": 2.2637,
      "step": 115447
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.8922823667526245,
      "learning_rate": 0.00029899167832312187,
      "loss": 2.1126,
      "step": 115448
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.9379624128341675,
      "learning_rate": 0.0002989875877841491,
      "loss": 1.8784,
      "step": 115449
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.9692365527153015,
      "learning_rate": 0.0002989834972453646,
      "loss": 2.2012,
      "step": 115450
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.9736687541007996,
      "learning_rate": 0.000298979406706769,
      "loss": 2.1543,
      "step": 115451
    },
    {
      "epoch": 1.5,
      "grad_norm": 1.0347063541412354,
      "learning_rate": 0.0002989753161683633,
      "loss": 2.1469,
      "step": 115452
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.9026128649711609,
      "learning_rate": 0.000298971225630148,
      "loss": 2.0847,
      "step": 115453
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.867736279964447,
      "learning_rate": 0.00029896713509212393,
      "loss": 2.1446,
      "step": 115454
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.876457691192627,
      "learning_rate": 0.0002989630445542919,
      "loss": 2.2896,
      "step": 115455
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.8923408389091492,
      "learning_rate": 0.00029895895401665267,
      "loss": 2.1816,
      "step": 115456
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.9993515014648438,
      "learning_rate": 0.000298954863479207,
      "loss": 2.2188,
      "step": 115457
    },
    {
      "epoch": 1.5,
      "grad_norm": 1.1379225254058838,
      "learning_rate": 0.00029895077294195575,
      "loss": 2.0221,
      "step": 115458
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.9082003831863403,
      "learning_rate": 0.00029894668240489943,
      "loss": 2.2362,
      "step": 115459
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.8919399380683899,
      "learning_rate": 0.000298942591868039,
      "loss": 1.9401,
      "step": 115460
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.9843665361404419,
      "learning_rate": 0.0002989385013313752,
      "loss": 2.1425,
      "step": 115461
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.962228000164032,
      "learning_rate": 0.00029893441079490866,
      "loss": 2.1335,
      "step": 115462
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.9485713839530945,
      "learning_rate": 0.00029893032025864023,
      "loss": 1.9889,
      "step": 115463
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.9612329602241516,
      "learning_rate": 0.00029892622972257075,
      "loss": 2.1345,
      "step": 115464
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.9779044985771179,
      "learning_rate": 0.00029892213918670087,
      "loss": 1.9386,
      "step": 115465
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.8921874761581421,
      "learning_rate": 0.00029891804865103135,
      "loss": 2.0457,
      "step": 115466
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.9431983232498169,
      "learning_rate": 0.00029891395811556306,
      "loss": 2.1057,
      "step": 115467
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.9012697339057922,
      "learning_rate": 0.00029890986758029665,
      "loss": 2.0352,
      "step": 115468
    },
    {
      "epoch": 1.5,
      "grad_norm": 1.2254259586334229,
      "learning_rate": 0.00029890577704523287,
      "loss": 2.0938,
      "step": 115469
    },
    {
      "epoch": 1.5,
      "grad_norm": 1.0056021213531494,
      "learning_rate": 0.0002989016865103726,
      "loss": 2.233,
      "step": 115470
    },
    {
      "epoch": 1.5,
      "grad_norm": 1.108273983001709,
      "learning_rate": 0.00029889759597571654,
      "loss": 2.1069,
      "step": 115471
    },
    {
      "epoch": 1.5,
      "grad_norm": 1.0267202854156494,
      "learning_rate": 0.00029889350544126534,
      "loss": 2.0663,
      "step": 115472
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.9260767698287964,
      "learning_rate": 0.0002988894149070199,
      "loss": 2.2724,
      "step": 115473
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.9619148373603821,
      "learning_rate": 0.000298885324372981,
      "loss": 2.1885,
      "step": 115474
    },
    {
      "epoch": 1.5,
      "grad_norm": 1.0767040252685547,
      "learning_rate": 0.00029888123383914924,
      "loss": 2.2487,
      "step": 115475
    },
    {
      "epoch": 1.5,
      "grad_norm": 1.0250645875930786,
      "learning_rate": 0.0002988771433055256,
      "loss": 2.2117,
      "step": 115476
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.9507696032524109,
      "learning_rate": 0.00029887305277211064,
      "loss": 2.1144,
      "step": 115477
    },
    {
      "epoch": 1.5,
      "grad_norm": 1.0146818161010742,
      "learning_rate": 0.0002988689622389052,
      "loss": 2.3617,
      "step": 115478
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.8804341554641724,
      "learning_rate": 0.00029886487170591,
      "loss": 2.1289,
      "step": 115479
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.8828614950180054,
      "learning_rate": 0.0002988607811731259,
      "loss": 2.2018,
      "step": 115480
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.9060455560684204,
      "learning_rate": 0.0002988566906405536,
      "loss": 2.2663,
      "step": 115481
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.9697443246841431,
      "learning_rate": 0.00029885260010819394,
      "loss": 2.3831,
      "step": 115482
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.9052812457084656,
      "learning_rate": 0.00029884850957604753,
      "loss": 2.2074,
      "step": 115483
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.9358345866203308,
      "learning_rate": 0.00029884441904411515,
      "loss": 2.402,
      "step": 115484
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.9683194756507874,
      "learning_rate": 0.00029884032851239766,
      "loss": 2.2238,
      "step": 115485
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.9045935273170471,
      "learning_rate": 0.00029883623798089576,
      "loss": 2.1121,
      "step": 115486
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.9166131615638733,
      "learning_rate": 0.0002988321474496102,
      "loss": 2.3184,
      "step": 115487
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.8443987965583801,
      "learning_rate": 0.0002988280569185419,
      "loss": 1.9453,
      "step": 115488
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.9668343663215637,
      "learning_rate": 0.00029882396638769137,
      "loss": 2.1775,
      "step": 115489
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.9065712690353394,
      "learning_rate": 0.0002988198758570595,
      "loss": 1.9349,
      "step": 115490
    },
    {
      "epoch": 1.5,
      "grad_norm": 1.0874816179275513,
      "learning_rate": 0.00029881578532664705,
      "loss": 2.2782,
      "step": 115491
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.9014192223548889,
      "learning_rate": 0.00029881169479645475,
      "loss": 2.3125,
      "step": 115492
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.9836834073066711,
      "learning_rate": 0.00029880760426648337,
      "loss": 2.2605,
      "step": 115493
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.8370643258094788,
      "learning_rate": 0.0002988035137367338,
      "loss": 2.0223,
      "step": 115494
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.9911805987358093,
      "learning_rate": 0.00029879942320720657,
      "loss": 2.1857,
      "step": 115495
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.9006927609443665,
      "learning_rate": 0.0002987953326779025,
      "loss": 2.239,
      "step": 115496
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.8790410161018372,
      "learning_rate": 0.0002987912421488225,
      "loss": 2.3353,
      "step": 115497
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.9646016955375671,
      "learning_rate": 0.0002987871516199672,
      "loss": 2.2542,
      "step": 115498
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.8454587459564209,
      "learning_rate": 0.00029878306109133734,
      "loss": 2.0237,
      "step": 115499
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.8987978100776672,
      "learning_rate": 0.0002987789705629339,
      "loss": 2.1051,
      "step": 115500
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.9185968041419983,
      "learning_rate": 0.00029877488003475733,
      "loss": 2.228,
      "step": 115501
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.9447733163833618,
      "learning_rate": 0.0002987707895068085,
      "loss": 2.3119,
      "step": 115502
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.9054871797561646,
      "learning_rate": 0.0002987666989790882,
      "loss": 2.0578,
      "step": 115503
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.9107842445373535,
      "learning_rate": 0.00029876260845159726,
      "loss": 2.0705,
      "step": 115504
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.9263501167297363,
      "learning_rate": 0.00029875851792433633,
      "loss": 2.1372,
      "step": 115505
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.8728652000427246,
      "learning_rate": 0.0002987544273973063,
      "loss": 2.0227,
      "step": 115506
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.92305588722229,
      "learning_rate": 0.00029875033687050786,
      "loss": 2.2074,
      "step": 115507
    },
    {
      "epoch": 1.5,
      "grad_norm": 1.0036859512329102,
      "learning_rate": 0.0002987462463439417,
      "loss": 1.9818,
      "step": 115508
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.9149295687675476,
      "learning_rate": 0.00029874215581760855,
      "loss": 2.0621,
      "step": 115509
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.9575750231742859,
      "learning_rate": 0.00029873806529150934,
      "loss": 2.1433,
      "step": 115510
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.9765943884849548,
      "learning_rate": 0.0002987339747656447,
      "loss": 2.1837,
      "step": 115511
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.8835122585296631,
      "learning_rate": 0.0002987298842400155,
      "loss": 2.3022,
      "step": 115512
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.912351131439209,
      "learning_rate": 0.0002987257937146225,
      "loss": 2.0191,
      "step": 115513
    },
    {
      "epoch": 1.5,
      "grad_norm": 1.0506209135055542,
      "learning_rate": 0.0002987217031894663,
      "loss": 2.0231,
      "step": 115514
    },
    {
      "epoch": 1.5,
      "grad_norm": 1.332676887512207,
      "learning_rate": 0.0002987176126645478,
      "loss": 2.3285,
      "step": 115515
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.90384441614151,
      "learning_rate": 0.0002987135221398676,
      "loss": 2.0138,
      "step": 115516
    },
    {
      "epoch": 1.5,
      "grad_norm": 1.0273141860961914,
      "learning_rate": 0.0002987094316154267,
      "loss": 1.9802,
      "step": 115517
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.853477418422699,
      "learning_rate": 0.0002987053410912257,
      "loss": 2.1333,
      "step": 115518
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.8948559165000916,
      "learning_rate": 0.0002987012505672655,
      "loss": 2.489,
      "step": 115519
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.9198435544967651,
      "learning_rate": 0.00029869716004354667,
      "loss": 2.2366,
      "step": 115520
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.8959096074104309,
      "learning_rate": 0.0002986930695200701,
      "loss": 2.1763,
      "step": 115521
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.9133835434913635,
      "learning_rate": 0.0002986889789968364,
      "loss": 2.2044,
      "step": 115522
    },
    {
      "epoch": 1.5,
      "grad_norm": 1.0473462343215942,
      "learning_rate": 0.00029868488847384654,
      "loss": 2.0048,
      "step": 115523
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.9459540247917175,
      "learning_rate": 0.00029868079795110114,
      "loss": 2.1342,
      "step": 115524
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.894183337688446,
      "learning_rate": 0.00029867670742860116,
      "loss": 2.2491,
      "step": 115525
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.9297513961791992,
      "learning_rate": 0.00029867261690634707,
      "loss": 2.1511,
      "step": 115526
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.9651545882225037,
      "learning_rate": 0.00029866852638433974,
      "loss": 2.2177,
      "step": 115527
    },
    {
      "epoch": 1.5,
      "grad_norm": 1.3342236280441284,
      "learning_rate": 0.00029866443586258,
      "loss": 2.1588,
      "step": 115528
    },
    {
      "epoch": 1.5,
      "grad_norm": 1.0245826244354248,
      "learning_rate": 0.00029866034534106853,
      "loss": 2.1325,
      "step": 115529
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.9746413230895996,
      "learning_rate": 0.00029865625481980616,
      "loss": 2.2551,
      "step": 115530
    },
    {
      "epoch": 1.5,
      "grad_norm": 1.0986571311950684,
      "learning_rate": 0.0002986521642987937,
      "loss": 2.1438,
      "step": 115531
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.9347543120384216,
      "learning_rate": 0.00029864807377803174,
      "loss": 1.8417,
      "step": 115532
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.9766679406166077,
      "learning_rate": 0.0002986439832575211,
      "loss": 2.1234,
      "step": 115533
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.9357358813285828,
      "learning_rate": 0.0002986398927372626,
      "loss": 2.2831,
      "step": 115534
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.9273191690444946,
      "learning_rate": 0.00029863580221725693,
      "loss": 1.9962,
      "step": 115535
    },
    {
      "epoch": 1.5,
      "grad_norm": 1.0400482416152954,
      "learning_rate": 0.0002986317116975049,
      "loss": 2.0934,
      "step": 115536
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.9405994415283203,
      "learning_rate": 0.00029862762117800745,
      "loss": 2.0493,
      "step": 115537
    },
    {
      "epoch": 1.5,
      "grad_norm": 1.0607094764709473,
      "learning_rate": 0.00029862353065876495,
      "loss": 2.399,
      "step": 115538
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.9248974323272705,
      "learning_rate": 0.00029861944013977843,
      "loss": 2.2131,
      "step": 115539
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.9004478454589844,
      "learning_rate": 0.0002986153496210485,
      "loss": 2.0651,
      "step": 115540
    },
    {
      "epoch": 1.5,
      "grad_norm": 1.0532965660095215,
      "learning_rate": 0.00029861125910257604,
      "loss": 2.0757,
      "step": 115541
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.8601493835449219,
      "learning_rate": 0.0002986071685843618,
      "loss": 1.8662,
      "step": 115542
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.948620617389679,
      "learning_rate": 0.00029860307806640657,
      "loss": 2.2541,
      "step": 115543
    },
    {
      "epoch": 1.5,
      "grad_norm": 1.062448263168335,
      "learning_rate": 0.000298598987548711,
      "loss": 2.0085,
      "step": 115544
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.9255554676055908,
      "learning_rate": 0.0002985948970312759,
      "loss": 2.3639,
      "step": 115545
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.9893026947975159,
      "learning_rate": 0.000298590806514102,
      "loss": 2.1185,
      "step": 115546
    },
    {
      "epoch": 1.5,
      "grad_norm": 1.030813455581665,
      "learning_rate": 0.00029858671599719013,
      "loss": 2.1347,
      "step": 115547
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.9023159146308899,
      "learning_rate": 0.000298582625480541,
      "loss": 2.1518,
      "step": 115548
    },
    {
      "epoch": 1.5,
      "grad_norm": 1.0342826843261719,
      "learning_rate": 0.0002985785349641554,
      "loss": 2.2115,
      "step": 115549
    },
    {
      "epoch": 1.5,
      "grad_norm": 1.0606671571731567,
      "learning_rate": 0.00029857444444803406,
      "loss": 2.0841,
      "step": 115550
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.8718400001525879,
      "learning_rate": 0.00029857035393217785,
      "loss": 2.1411,
      "step": 115551
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.9483095407485962,
      "learning_rate": 0.0002985662634165873,
      "loss": 2.2523,
      "step": 115552
    },
    {
      "epoch": 1.5,
      "grad_norm": 1.135155200958252,
      "learning_rate": 0.0002985621729012634,
      "loss": 2.1322,
      "step": 115553
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.8506349325180054,
      "learning_rate": 0.0002985580823862067,
      "loss": 2.0038,
      "step": 115554
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.9164266586303711,
      "learning_rate": 0.00029855399187141824,
      "loss": 2.2372,
      "step": 115555
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.9278139472007751,
      "learning_rate": 0.0002985499013568985,
      "loss": 2.1621,
      "step": 115556
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.9616002440452576,
      "learning_rate": 0.00029854581084264844,
      "loss": 2.1426,
      "step": 115557
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.9105631113052368,
      "learning_rate": 0.00029854172032866866,
      "loss": 2.2375,
      "step": 115558
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.9915072917938232,
      "learning_rate": 0.00029853762981496015,
      "loss": 2.0516,
      "step": 115559
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.9490463733673096,
      "learning_rate": 0.0002985335393015234,
      "loss": 2.1656,
      "step": 115560
    },
    {
      "epoch": 1.5,
      "grad_norm": 1.017610788345337,
      "learning_rate": 0.00029852944878835935,
      "loss": 1.8708,
      "step": 115561
    },
    {
      "epoch": 1.5,
      "grad_norm": 1.0219508409500122,
      "learning_rate": 0.00029852535827546865,
      "loss": 2.0766,
      "step": 115562
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.9557982683181763,
      "learning_rate": 0.00029852126776285214,
      "loss": 2.0773,
      "step": 115563
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.9047445058822632,
      "learning_rate": 0.00029851717725051054,
      "loss": 2.0212,
      "step": 115564
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.9683815836906433,
      "learning_rate": 0.0002985130867384446,
      "loss": 2.3144,
      "step": 115565
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.9106740951538086,
      "learning_rate": 0.0002985089962266552,
      "loss": 2.0974,
      "step": 115566
    },
    {
      "epoch": 1.5,
      "grad_norm": 1.0979982614517212,
      "learning_rate": 0.00029850490571514297,
      "loss": 2.0055,
      "step": 115567
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.8954346179962158,
      "learning_rate": 0.0002985008152039087,
      "loss": 2.1425,
      "step": 115568
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.8948819041252136,
      "learning_rate": 0.0002984967246929531,
      "loss": 2.0614,
      "step": 115569
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.9852848052978516,
      "learning_rate": 0.00029849263418227703,
      "loss": 2.119,
      "step": 115570
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.9914103150367737,
      "learning_rate": 0.00029848854367188113,
      "loss": 2.2484,
      "step": 115571
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.9142771363258362,
      "learning_rate": 0.0002984844531617663,
      "loss": 2.3072,
      "step": 115572
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.9770727753639221,
      "learning_rate": 0.00029848036265193333,
      "loss": 2.2296,
      "step": 115573
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.9265663027763367,
      "learning_rate": 0.00029847627214238284,
      "loss": 2.0447,
      "step": 115574
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.8691055178642273,
      "learning_rate": 0.0002984721816331156,
      "loss": 1.9976,
      "step": 115575
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.9028521776199341,
      "learning_rate": 0.0002984680911241324,
      "loss": 2.2257,
      "step": 115576
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.9723031520843506,
      "learning_rate": 0.000298464000615434,
      "loss": 2.0659,
      "step": 115577
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.906894326210022,
      "learning_rate": 0.0002984599101070212,
      "loss": 2.1201,
      "step": 115578
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.9159727096557617,
      "learning_rate": 0.0002984558195988947,
      "loss": 2.1745,
      "step": 115579
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.9282445311546326,
      "learning_rate": 0.00029845172909105547,
      "loss": 2.2877,
      "step": 115580
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.8392802476882935,
      "learning_rate": 0.0002984476385835039,
      "loss": 2.2501,
      "step": 115581
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.9882840514183044,
      "learning_rate": 0.000298443548076241,
      "loss": 2.2196,
      "step": 115582
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.9161057472229004,
      "learning_rate": 0.0002984394575692674,
      "loss": 2.0999,
      "step": 115583
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.9315107464790344,
      "learning_rate": 0.000298435367062584,
      "loss": 2.0735,
      "step": 115584
    },
    {
      "epoch": 1.5,
      "grad_norm": 0.9275715351104736,
      "learning_rate": 0.00029843127655619146,
      "loss": 2.2172,
      "step": 115585
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.9298714399337769,
      "learning_rate": 0.00029842718605009073,
      "loss": 2.2859,
      "step": 115586
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.9188265204429626,
      "learning_rate": 0.0002984230955442823,
      "loss": 2.3047,
      "step": 115587
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.9287516474723816,
      "learning_rate": 0.000298419005038767,
      "loss": 2.1379,
      "step": 115588
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.9652860164642334,
      "learning_rate": 0.0002984149145335457,
      "loss": 2.3302,
      "step": 115589
    },
    {
      "epoch": 1.51,
      "grad_norm": 1.3618416786193848,
      "learning_rate": 0.00029841082402861903,
      "loss": 2.3961,
      "step": 115590
    },
    {
      "epoch": 1.51,
      "grad_norm": 1.012790322303772,
      "learning_rate": 0.00029840673352398784,
      "loss": 2.0693,
      "step": 115591
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.951481282711029,
      "learning_rate": 0.000298402643019653,
      "loss": 2.1721,
      "step": 115592
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.9192454814910889,
      "learning_rate": 0.00029839855251561503,
      "loss": 1.9786,
      "step": 115593
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.898138165473938,
      "learning_rate": 0.0002983944620118748,
      "loss": 2.2717,
      "step": 115594
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.9179250597953796,
      "learning_rate": 0.00029839037150843306,
      "loss": 2.0368,
      "step": 115595
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.8179869055747986,
      "learning_rate": 0.00029838628100529053,
      "loss": 1.8709,
      "step": 115596
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.8787325024604797,
      "learning_rate": 0.0002983821905024481,
      "loss": 2.0672,
      "step": 115597
    },
    {
      "epoch": 1.51,
      "grad_norm": 1.0763498544692993,
      "learning_rate": 0.00029837809999990654,
      "loss": 2.1662,
      "step": 115598
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.9459872245788574,
      "learning_rate": 0.0002983740094976664,
      "loss": 1.8475,
      "step": 115599
    },
    {
      "epoch": 1.51,
      "grad_norm": 1.175594687461853,
      "learning_rate": 0.0002983699189957285,
      "loss": 2.2373,
      "step": 115600
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.9484619498252869,
      "learning_rate": 0.00029836582849409373,
      "loss": 1.9355,
      "step": 115601
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.8757637739181519,
      "learning_rate": 0.0002983617379927628,
      "loss": 2.2117,
      "step": 115602
    },
    {
      "epoch": 1.51,
      "grad_norm": 1.1491607427597046,
      "learning_rate": 0.0002983576474917364,
      "loss": 2.2082,
      "step": 115603
    },
    {
      "epoch": 1.51,
      "grad_norm": 1.0367108583450317,
      "learning_rate": 0.0002983535569910155,
      "loss": 2.1699,
      "step": 115604
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.9381589293479919,
      "learning_rate": 0.0002983494664906006,
      "loss": 1.8178,
      "step": 115605
    },
    {
      "epoch": 1.51,
      "grad_norm": 1.1259044408798218,
      "learning_rate": 0.00029834537599049253,
      "loss": 2.2614,
      "step": 115606
    },
    {
      "epoch": 1.51,
      "grad_norm": 1.0722739696502686,
      "learning_rate": 0.0002983412854906921,
      "loss": 2.0826,
      "step": 115607
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.8963311314582825,
      "learning_rate": 0.0002983371949912001,
      "loss": 2.2914,
      "step": 115608
    },
    {
      "epoch": 1.51,
      "grad_norm": 1.0055537223815918,
      "learning_rate": 0.00029833310449201716,
      "loss": 1.9401,
      "step": 115609
    },
    {
      "epoch": 1.51,
      "grad_norm": 1.0053056478500366,
      "learning_rate": 0.0002983290139931443,
      "loss": 2.2562,
      "step": 115610
    },
    {
      "epoch": 1.51,
      "grad_norm": 1.1908442974090576,
      "learning_rate": 0.000298324923494582,
      "loss": 2.1542,
      "step": 115611
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.8915692567825317,
      "learning_rate": 0.0002983208329963311,
      "loss": 2.1936,
      "step": 115612
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.972769558429718,
      "learning_rate": 0.0002983167424983924,
      "loss": 2.2068,
      "step": 115613
    },
    {
      "epoch": 1.51,
      "grad_norm": 1.0348352193832397,
      "learning_rate": 0.0002983126520007666,
      "loss": 2.1493,
      "step": 115614
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.9346762299537659,
      "learning_rate": 0.0002983085615034546,
      "loss": 2.0443,
      "step": 115615
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.901009738445282,
      "learning_rate": 0.00029830447100645714,
      "loss": 2.3221,
      "step": 115616
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.8878586292266846,
      "learning_rate": 0.00029830038050977477,
      "loss": 1.9607,
      "step": 115617
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.9929257035255432,
      "learning_rate": 0.00029829629001340845,
      "loss": 2.1607,
      "step": 115618
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.9144750237464905,
      "learning_rate": 0.0002982921995173588,
      "loss": 2.4234,
      "step": 115619
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.9524155855178833,
      "learning_rate": 0.00029828810902162676,
      "loss": 2.1566,
      "step": 115620
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.8738571405410767,
      "learning_rate": 0.0002982840185262129,
      "loss": 1.8032,
      "step": 115621
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.8971184492111206,
      "learning_rate": 0.0002982799280311183,
      "loss": 2.0631,
      "step": 115622
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.9877063035964966,
      "learning_rate": 0.00029827583753634326,
      "loss": 2.1764,
      "step": 115623
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.9136051535606384,
      "learning_rate": 0.00029827174704188887,
      "loss": 2.067,
      "step": 115624
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.9404098987579346,
      "learning_rate": 0.00029826765654775573,
      "loss": 2.2001,
      "step": 115625
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.844740629196167,
      "learning_rate": 0.0002982635660539447,
      "loss": 2.1855,
      "step": 115626
    },
    {
      "epoch": 1.51,
      "grad_norm": 1.0184454917907715,
      "learning_rate": 0.00029825947556045646,
      "loss": 2.1706,
      "step": 115627
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.9216556549072266,
      "learning_rate": 0.0002982553850672919,
      "loss": 2.2526,
      "step": 115628
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.927317202091217,
      "learning_rate": 0.0002982512945744517,
      "loss": 2.1069,
      "step": 115629
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.9514394998550415,
      "learning_rate": 0.0002982472040819366,
      "loss": 2.2246,
      "step": 115630
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.9714270234107971,
      "learning_rate": 0.0002982431135897473,
      "loss": 2.0645,
      "step": 115631
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.9565660357475281,
      "learning_rate": 0.0002982390230978847,
      "loss": 2.111,
      "step": 115632
    },
    {
      "epoch": 1.51,
      "grad_norm": 1.0230180025100708,
      "learning_rate": 0.0002982349326063495,
      "loss": 2.1689,
      "step": 115633
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.9018794894218445,
      "learning_rate": 0.0002982308421151425,
      "loss": 2.1681,
      "step": 115634
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.990216851234436,
      "learning_rate": 0.00029822675162426436,
      "loss": 1.9518,
      "step": 115635
    },
    {
      "epoch": 1.51,
      "grad_norm": 1.0457876920700073,
      "learning_rate": 0.000298222661133716,
      "loss": 2.2587,
      "step": 115636
    },
    {
      "epoch": 1.51,
      "grad_norm": 1.0063414573669434,
      "learning_rate": 0.00029821857064349795,
      "loss": 2.1055,
      "step": 115637
    },
    {
      "epoch": 1.51,
      "grad_norm": 1.0067952871322632,
      "learning_rate": 0.0002982144801536111,
      "loss": 2.0772,
      "step": 115638
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.8914965987205505,
      "learning_rate": 0.00029821038966405626,
      "loss": 2.1395,
      "step": 115639
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.9731947779655457,
      "learning_rate": 0.00029820629917483413,
      "loss": 2.078,
      "step": 115640
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.9497334957122803,
      "learning_rate": 0.0002982022086859455,
      "loss": 2.0097,
      "step": 115641
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.9769980311393738,
      "learning_rate": 0.0002981981181973911,
      "loss": 2.21,
      "step": 115642
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.8380321264266968,
      "learning_rate": 0.00029819402770917176,
      "loss": 2.1454,
      "step": 115643
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.9232057332992554,
      "learning_rate": 0.00029818993722128813,
      "loss": 2.072,
      "step": 115644
    },
    {
      "epoch": 1.51,
      "grad_norm": 1.0248173475265503,
      "learning_rate": 0.000298185846733741,
      "loss": 2.2141,
      "step": 115645
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.9781690239906311,
      "learning_rate": 0.0002981817562465312,
      "loss": 1.9578,
      "step": 115646
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.9005306959152222,
      "learning_rate": 0.0002981776657596595,
      "loss": 2.1664,
      "step": 115647
    },
    {
      "epoch": 1.51,
      "grad_norm": 1.004420518875122,
      "learning_rate": 0.00029817357527312644,
      "loss": 2.0813,
      "step": 115648
    },
    {
      "epoch": 1.51,
      "grad_norm": 1.0333569049835205,
      "learning_rate": 0.0002981694847869331,
      "loss": 2.1581,
      "step": 115649
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.9191575050354004,
      "learning_rate": 0.00029816539430108,
      "loss": 2.185,
      "step": 115650
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.9590580463409424,
      "learning_rate": 0.000298161303815568,
      "loss": 2.0598,
      "step": 115651
    },
    {
      "epoch": 1.51,
      "grad_norm": 1.0096935033798218,
      "learning_rate": 0.00029815721333039785,
      "loss": 2.0752,
      "step": 115652
    },
    {
      "epoch": 1.51,
      "grad_norm": 1.0073775053024292,
      "learning_rate": 0.0002981531228455704,
      "loss": 2.0546,
      "step": 115653
    },
    {
      "epoch": 1.51,
      "grad_norm": 1.0178786516189575,
      "learning_rate": 0.0002981490323610862,
      "loss": 2.0588,
      "step": 115654
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.9457507133483887,
      "learning_rate": 0.0002981449418769462,
      "loss": 1.8603,
      "step": 115655
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.9132649302482605,
      "learning_rate": 0.00029814085139315104,
      "loss": 2.0611,
      "step": 115656
    },
    {
      "epoch": 1.51,
      "grad_norm": 1.3367942571640015,
      "learning_rate": 0.0002981367609097015,
      "loss": 1.9206,
      "step": 115657
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.9723870754241943,
      "learning_rate": 0.0002981326704265985,
      "loss": 2.2861,
      "step": 115658
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.9799350500106812,
      "learning_rate": 0.00029812857994384267,
      "loss": 1.9789,
      "step": 115659
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.9323735237121582,
      "learning_rate": 0.00029812448946143463,
      "loss": 2.2277,
      "step": 115660
    },
    {
      "epoch": 1.51,
      "grad_norm": 1.1895204782485962,
      "learning_rate": 0.0002981203989793754,
      "loss": 2.048,
      "step": 115661
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.9871166944503784,
      "learning_rate": 0.0002981163084976655,
      "loss": 2.1994,
      "step": 115662
    },
    {
      "epoch": 1.51,
      "grad_norm": 1.0497148036956787,
      "learning_rate": 0.0002981122180163059,
      "loss": 2.0604,
      "step": 115663
    },
    {
      "epoch": 1.51,
      "grad_norm": 1.070816159248352,
      "learning_rate": 0.00029810812753529726,
      "loss": 1.9781,
      "step": 115664
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.9057830572128296,
      "learning_rate": 0.0002981040370546405,
      "loss": 2.1509,
      "step": 115665
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.9729200601577759,
      "learning_rate": 0.0002980999465743361,
      "loss": 2.2343,
      "step": 115666
    },
    {
      "epoch": 1.51,
      "grad_norm": 1.0314363241195679,
      "learning_rate": 0.00029809585609438485,
      "loss": 2.1323,
      "step": 115667
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.8792814016342163,
      "learning_rate": 0.0002980917656147877,
      "loss": 1.941,
      "step": 115668
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.9666275382041931,
      "learning_rate": 0.0002980876751355454,
      "loss": 2.1474,
      "step": 115669
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.9790098071098328,
      "learning_rate": 0.0002980835846566585,
      "loss": 2.3017,
      "step": 115670
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.9489437937736511,
      "learning_rate": 0.00029807949417812814,
      "loss": 2.1457,
      "step": 115671
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.8262100219726562,
      "learning_rate": 0.00029807540369995467,
      "loss": 1.9889,
      "step": 115672
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.842446506023407,
      "learning_rate": 0.000298071313222139,
      "loss": 1.9708,
      "step": 115673
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.9153900742530823,
      "learning_rate": 0.0002980672227446819,
      "loss": 2.2588,
      "step": 115674
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.9394927620887756,
      "learning_rate": 0.00029806313226758414,
      "loss": 2.3306,
      "step": 115675
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.9066596627235413,
      "learning_rate": 0.0002980590417908465,
      "loss": 2.0579,
      "step": 115676
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.8940379023551941,
      "learning_rate": 0.00029805495131446987,
      "loss": 2.1799,
      "step": 115677
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.9219814538955688,
      "learning_rate": 0.0002980508608384547,
      "loss": 2.0658,
      "step": 115678
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.8914372324943542,
      "learning_rate": 0.0002980467703628019,
      "loss": 2.1668,
      "step": 115679
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.9282069206237793,
      "learning_rate": 0.0002980426798875123,
      "loss": 2.2554,
      "step": 115680
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.9602908492088318,
      "learning_rate": 0.00029803858941258654,
      "loss": 1.8169,
      "step": 115681
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.8963721990585327,
      "learning_rate": 0.00029803449893802547,
      "loss": 1.9589,
      "step": 115682
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.9957287907600403,
      "learning_rate": 0.00029803040846383,
      "loss": 2.2351,
      "step": 115683
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.8946533799171448,
      "learning_rate": 0.0002980263179900005,
      "loss": 2.152,
      "step": 115684
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.9241651892662048,
      "learning_rate": 0.000298022227516538,
      "loss": 2.2288,
      "step": 115685
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.963672399520874,
      "learning_rate": 0.00029801813704344315,
      "loss": 2.1745,
      "step": 115686
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.8827360272407532,
      "learning_rate": 0.00029801404657071683,
      "loss": 2.1087,
      "step": 115687
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.8977856040000916,
      "learning_rate": 0.00029800995609835973,
      "loss": 2.1484,
      "step": 115688
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.948760449886322,
      "learning_rate": 0.0002980058656263727,
      "loss": 2.2127,
      "step": 115689
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.976256787776947,
      "learning_rate": 0.0002980017751547563,
      "loss": 2.1851,
      "step": 115690
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.9571414589881897,
      "learning_rate": 0.0002979976846835115,
      "loss": 2.2385,
      "step": 115691
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.8901517391204834,
      "learning_rate": 0.0002979935942126388,
      "loss": 2.2325,
      "step": 115692
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.9467493295669556,
      "learning_rate": 0.0002979895037421393,
      "loss": 2.062,
      "step": 115693
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.9500829577445984,
      "learning_rate": 0.00029798541327201347,
      "loss": 2.1212,
      "step": 115694
    },
    {
      "epoch": 1.51,
      "grad_norm": 1.0233150720596313,
      "learning_rate": 0.00029798132280226236,
      "loss": 1.9869,
      "step": 115695
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.9448642134666443,
      "learning_rate": 0.00029797723233288643,
      "loss": 2.161,
      "step": 115696
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.9824705123901367,
      "learning_rate": 0.00029797314186388657,
      "loss": 2.3131,
      "step": 115697
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.9699366688728333,
      "learning_rate": 0.0002979690513952635,
      "loss": 2.0281,
      "step": 115698
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.9013103246688843,
      "learning_rate": 0.0002979649609270181,
      "loss": 2.2106,
      "step": 115699
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.9070317149162292,
      "learning_rate": 0.00029796087045915105,
      "loss": 2.3318,
      "step": 115700
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.96271812915802,
      "learning_rate": 0.00029795677999166315,
      "loss": 2.2497,
      "step": 115701
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.8283156752586365,
      "learning_rate": 0.00029795268952455506,
      "loss": 1.9497,
      "step": 115702
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.8922122120857239,
      "learning_rate": 0.0002979485990578276,
      "loss": 1.8838,
      "step": 115703
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.9859720468521118,
      "learning_rate": 0.0002979445085914815,
      "loss": 1.9688,
      "step": 115704
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.946656346321106,
      "learning_rate": 0.0002979404181255176,
      "loss": 2.09,
      "step": 115705
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.9258713126182556,
      "learning_rate": 0.00029793632765993657,
      "loss": 2.055,
      "step": 115706
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.9889528751373291,
      "learning_rate": 0.00029793223719473927,
      "loss": 2.2713,
      "step": 115707
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.9297016859054565,
      "learning_rate": 0.00029792814672992653,
      "loss": 2.0775,
      "step": 115708
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.935155987739563,
      "learning_rate": 0.0002979240562654988,
      "loss": 2.0846,
      "step": 115709
    },
    {
      "epoch": 1.51,
      "grad_norm": 1.0460388660430908,
      "learning_rate": 0.0002979199658014571,
      "loss": 1.9893,
      "step": 115710
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.9182366132736206,
      "learning_rate": 0.0002979158753378021,
      "loss": 2.1653,
      "step": 115711
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.9382639527320862,
      "learning_rate": 0.0002979117848745345,
      "loss": 2.3004,
      "step": 115712
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.9533815383911133,
      "learning_rate": 0.0002979076944116552,
      "loss": 2.3202,
      "step": 115713
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.9271026253700256,
      "learning_rate": 0.00029790360394916506,
      "loss": 2.1728,
      "step": 115714
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.8927546739578247,
      "learning_rate": 0.0002978995134870645,
      "loss": 2.2368,
      "step": 115715
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.9004607200622559,
      "learning_rate": 0.0002978954230253545,
      "loss": 2.1142,
      "step": 115716
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.9168899655342102,
      "learning_rate": 0.0002978913325640357,
      "loss": 1.9131,
      "step": 115717
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.9481428265571594,
      "learning_rate": 0.00029788724210310903,
      "loss": 2.2591,
      "step": 115718
    },
    {
      "epoch": 1.51,
      "grad_norm": 1.0234237909317017,
      "learning_rate": 0.0002978831516425751,
      "loss": 2.0888,
      "step": 115719
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.9190632104873657,
      "learning_rate": 0.00029787906118243485,
      "loss": 2.0911,
      "step": 115720
    },
    {
      "epoch": 1.51,
      "grad_norm": 1.0928969383239746,
      "learning_rate": 0.0002978749707226889,
      "loss": 2.3526,
      "step": 115721
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.8993414640426636,
      "learning_rate": 0.00029787088026333796,
      "loss": 2.2992,
      "step": 115722
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.9320944547653198,
      "learning_rate": 0.0002978667898043829,
      "loss": 2.1248,
      "step": 115723
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.9352245926856995,
      "learning_rate": 0.0002978626993458244,
      "loss": 2.1214,
      "step": 115724
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.9594113230705261,
      "learning_rate": 0.0002978586088876633,
      "loss": 2.1888,
      "step": 115725
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.974777340888977,
      "learning_rate": 0.00029785451842990037,
      "loss": 2.1958,
      "step": 115726
    },
    {
      "epoch": 1.51,
      "grad_norm": 1.0177277326583862,
      "learning_rate": 0.00029785042797253626,
      "loss": 2.1727,
      "step": 115727
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.8579190373420715,
      "learning_rate": 0.00029784633751557185,
      "loss": 1.9822,
      "step": 115728
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.928913950920105,
      "learning_rate": 0.0002978422470590078,
      "loss": 1.9589,
      "step": 115729
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.9851864576339722,
      "learning_rate": 0.00029783815660284487,
      "loss": 1.9558,
      "step": 115730
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.8823597431182861,
      "learning_rate": 0.0002978340661470839,
      "loss": 1.9942,
      "step": 115731
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.9485334753990173,
      "learning_rate": 0.0002978299756917257,
      "loss": 2.3196,
      "step": 115732
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.886911928653717,
      "learning_rate": 0.0002978258852367709,
      "loss": 2.1235,
      "step": 115733
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.8739530444145203,
      "learning_rate": 0.00029782179478222027,
      "loss": 2.2609,
      "step": 115734
    },
    {
      "epoch": 1.51,
      "grad_norm": 1.000515341758728,
      "learning_rate": 0.0002978177043280747,
      "loss": 1.9713,
      "step": 115735
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.8666204214096069,
      "learning_rate": 0.0002978136138743348,
      "loss": 2.1148,
      "step": 115736
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.8590118885040283,
      "learning_rate": 0.0002978095234210014,
      "loss": 2.1798,
      "step": 115737
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.9016433358192444,
      "learning_rate": 0.00029780543296807524,
      "loss": 2.1587,
      "step": 115738
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.9548647403717041,
      "learning_rate": 0.0002978013425155571,
      "loss": 2.0757,
      "step": 115739
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.856855034828186,
      "learning_rate": 0.0002977972520634477,
      "loss": 2.0907,
      "step": 115740
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.9429218173027039,
      "learning_rate": 0.00029779316161174783,
      "loss": 2.2988,
      "step": 115741
    },
    {
      "epoch": 1.51,
      "grad_norm": 1.1109938621520996,
      "learning_rate": 0.00029778907116045824,
      "loss": 2.1897,
      "step": 115742
    },
    {
      "epoch": 1.51,
      "grad_norm": 1.0666663646697998,
      "learning_rate": 0.0002977849807095798,
      "loss": 2.0612,
      "step": 115743
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.8685458302497864,
      "learning_rate": 0.00029778089025911316,
      "loss": 2.0855,
      "step": 115744
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.9546803832054138,
      "learning_rate": 0.000297776799809059,
      "loss": 2.2341,
      "step": 115745
    },
    {
      "epoch": 1.51,
      "grad_norm": 1.1866296529769897,
      "learning_rate": 0.00029777270935941823,
      "loss": 2.0687,
      "step": 115746
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.937465250492096,
      "learning_rate": 0.00029776861891019156,
      "loss": 2.0641,
      "step": 115747
    },
    {
      "epoch": 1.51,
      "grad_norm": 1.0031371116638184,
      "learning_rate": 0.0002977645284613797,
      "loss": 2.1271,
      "step": 115748
    },
    {
      "epoch": 1.51,
      "grad_norm": 1.0410146713256836,
      "learning_rate": 0.00029776043801298346,
      "loss": 1.9565,
      "step": 115749
    },
    {
      "epoch": 1.51,
      "grad_norm": 1.3050144910812378,
      "learning_rate": 0.0002977563475650038,
      "loss": 2.2638,
      "step": 115750
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.8990172147750854,
      "learning_rate": 0.00029775225711744103,
      "loss": 2.2242,
      "step": 115751
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.9980306625366211,
      "learning_rate": 0.00029774816667029624,
      "loss": 2.1216,
      "step": 115752
    },
    {
      "epoch": 1.51,
      "grad_norm": 1.0285162925720215,
      "learning_rate": 0.00029774407622357006,
      "loss": 2.0206,
      "step": 115753
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.9971083402633667,
      "learning_rate": 0.00029773998577726336,
      "loss": 2.1327,
      "step": 115754
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.975434422492981,
      "learning_rate": 0.0002977358953313768,
      "loss": 2.051,
      "step": 115755
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.9106069803237915,
      "learning_rate": 0.0002977318048859113,
      "loss": 2.1281,
      "step": 115756
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.846446692943573,
      "learning_rate": 0.0002977277144408674,
      "loss": 2.3705,
      "step": 115757
    },
    {
      "epoch": 1.51,
      "grad_norm": 1.0957703590393066,
      "learning_rate": 0.00029772362399624596,
      "loss": 1.9798,
      "step": 115758
    },
    {
      "epoch": 1.51,
      "grad_norm": 1.0469894409179688,
      "learning_rate": 0.0002977195335520477,
      "loss": 1.9969,
      "step": 115759
    },
    {
      "epoch": 1.51,
      "grad_norm": 1.078421950340271,
      "learning_rate": 0.0002977154431082735,
      "loss": 2.3185,
      "step": 115760
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.9039744138717651,
      "learning_rate": 0.00029771135266492394,
      "loss": 2.2323,
      "step": 115761
    },
    {
      "epoch": 1.51,
      "grad_norm": 1.1375653743743896,
      "learning_rate": 0.0002977072622220001,
      "loss": 2.1281,
      "step": 115762
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.9929518699645996,
      "learning_rate": 0.0002977031717795023,
      "loss": 2.2188,
      "step": 115763
    },
    {
      "epoch": 1.51,
      "grad_norm": 1.0663477182388306,
      "learning_rate": 0.0002976990813374316,
      "loss": 2.3252,
      "step": 115764
    },
    {
      "epoch": 1.51,
      "grad_norm": 1.3410849571228027,
      "learning_rate": 0.0002976949908957887,
      "loss": 1.9553,
      "step": 115765
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.9109584093093872,
      "learning_rate": 0.00029769090045457427,
      "loss": 2.0395,
      "step": 115766
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.8772171139717102,
      "learning_rate": 0.0002976868100137892,
      "loss": 2.0933,
      "step": 115767
    },
    {
      "epoch": 1.51,
      "grad_norm": 1.1396405696868896,
      "learning_rate": 0.0002976827195734343,
      "loss": 2.4462,
      "step": 115768
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.9744722247123718,
      "learning_rate": 0.0002976786291335101,
      "loss": 2.4268,
      "step": 115769
    },
    {
      "epoch": 1.51,
      "grad_norm": 1.0321979522705078,
      "learning_rate": 0.0002976745386940175,
      "loss": 2.2291,
      "step": 115770
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.973594069480896,
      "learning_rate": 0.00029767044825495723,
      "loss": 2.2574,
      "step": 115771
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.9806104302406311,
      "learning_rate": 0.0002976663578163301,
      "loss": 2.1793,
      "step": 115772
    },
    {
      "epoch": 1.51,
      "grad_norm": 1.059910535812378,
      "learning_rate": 0.0002976622673781368,
      "loss": 2.16,
      "step": 115773
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.9301537871360779,
      "learning_rate": 0.0002976581769403781,
      "loss": 2.0538,
      "step": 115774
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.8774961829185486,
      "learning_rate": 0.000297654086503055,
      "loss": 1.9549,
      "step": 115775
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.9518961310386658,
      "learning_rate": 0.0002976499960661679,
      "loss": 1.9754,
      "step": 115776
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.9768940210342407,
      "learning_rate": 0.0002976459056297177,
      "loss": 1.8324,
      "step": 115777
    },
    {
      "epoch": 1.51,
      "grad_norm": 1.0472095012664795,
      "learning_rate": 0.00029764181519370515,
      "loss": 2.0198,
      "step": 115778
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.9133633971214294,
      "learning_rate": 0.00029763772475813107,
      "loss": 2.2547,
      "step": 115779
    },
    {
      "epoch": 1.51,
      "grad_norm": 1.146960735321045,
      "learning_rate": 0.0002976336343229961,
      "loss": 2.1786,
      "step": 115780
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.9799450635910034,
      "learning_rate": 0.0002976295438883013,
      "loss": 2.3288,
      "step": 115781
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.9342831969261169,
      "learning_rate": 0.00029762545345404703,
      "loss": 2.1285,
      "step": 115782
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.8856329917907715,
      "learning_rate": 0.0002976213630202343,
      "loss": 2.1201,
      "step": 115783
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.9059527516365051,
      "learning_rate": 0.00029761727258686375,
      "loss": 2.1351,
      "step": 115784
    },
    {
      "epoch": 1.51,
      "grad_norm": 1.1637276411056519,
      "learning_rate": 0.0002976131821539362,
      "loss": 2.2163,
      "step": 115785
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.8957696557044983,
      "learning_rate": 0.0002976090917214524,
      "loss": 2.2705,
      "step": 115786
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.9131296277046204,
      "learning_rate": 0.00029760500128941323,
      "loss": 2.0812,
      "step": 115787
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.9676232933998108,
      "learning_rate": 0.00029760091085781926,
      "loss": 2.09,
      "step": 115788
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.9277791976928711,
      "learning_rate": 0.0002975968204266713,
      "loss": 2.2509,
      "step": 115789
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.9843201637268066,
      "learning_rate": 0.0002975927299959701,
      "loss": 2.1273,
      "step": 115790
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.9379290342330933,
      "learning_rate": 0.00029758863956571647,
      "loss": 2.0502,
      "step": 115791
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.9501726031303406,
      "learning_rate": 0.00029758454913591116,
      "loss": 1.9183,
      "step": 115792
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.8911344408988953,
      "learning_rate": 0.0002975804587065551,
      "loss": 1.8198,
      "step": 115793
    },
    {
      "epoch": 1.51,
      "grad_norm": 1.040880799293518,
      "learning_rate": 0.00029757636827764875,
      "loss": 2.0933,
      "step": 115794
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.9318429827690125,
      "learning_rate": 0.00029757227784919294,
      "loss": 2.2031,
      "step": 115795
    },
    {
      "epoch": 1.51,
      "grad_norm": 1.0533684492111206,
      "learning_rate": 0.00029756818742118855,
      "loss": 2.0602,
      "step": 115796
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.9259389638900757,
      "learning_rate": 0.0002975640969936362,
      "loss": 2.0825,
      "step": 115797
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.9051699638366699,
      "learning_rate": 0.0002975600065665368,
      "loss": 2.0955,
      "step": 115798
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.9611115455627441,
      "learning_rate": 0.0002975559161398912,
      "loss": 2.1876,
      "step": 115799
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.906966507434845,
      "learning_rate": 0.0002975518257136998,
      "loss": 2.0925,
      "step": 115800
    },
    {
      "epoch": 1.51,
      "grad_norm": 1.0356727838516235,
      "learning_rate": 0.00029754773528796355,
      "loss": 2.0242,
      "step": 115801
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.8709473013877869,
      "learning_rate": 0.00029754364486268327,
      "loss": 2.0417,
      "step": 115802
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.9190207719802856,
      "learning_rate": 0.00029753955443785965,
      "loss": 2.3243,
      "step": 115803
    },
    {
      "epoch": 1.51,
      "grad_norm": 1.0043160915374756,
      "learning_rate": 0.00029753546401349346,
      "loss": 2.1011,
      "step": 115804
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.955889105796814,
      "learning_rate": 0.0002975313735895856,
      "loss": 2.1368,
      "step": 115805
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.9089835286140442,
      "learning_rate": 0.00029752728316613666,
      "loss": 2.1862,
      "step": 115806
    },
    {
      "epoch": 1.51,
      "grad_norm": 1.0512186288833618,
      "learning_rate": 0.00029752319274314735,
      "loss": 2.2515,
      "step": 115807
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.9978991150856018,
      "learning_rate": 0.00029751910232061855,
      "loss": 2.3371,
      "step": 115808
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.9324721097946167,
      "learning_rate": 0.00029751501189855103,
      "loss": 2.1944,
      "step": 115809
    },
    {
      "epoch": 1.51,
      "grad_norm": 1.1958250999450684,
      "learning_rate": 0.00029751092147694544,
      "loss": 1.9599,
      "step": 115810
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.9858450293540955,
      "learning_rate": 0.00029750683105580274,
      "loss": 2.2625,
      "step": 115811
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.9683105945587158,
      "learning_rate": 0.00029750274063512354,
      "loss": 2.0772,
      "step": 115812
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.8911317586898804,
      "learning_rate": 0.0002974986502149086,
      "loss": 2.3026,
      "step": 115813
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.874658465385437,
      "learning_rate": 0.0002974945597951587,
      "loss": 2.2422,
      "step": 115814
    },
    {
      "epoch": 1.51,
      "grad_norm": 1.0652005672454834,
      "learning_rate": 0.0002974904693758746,
      "loss": 2.0716,
      "step": 115815
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.8484202027320862,
      "learning_rate": 0.0002974863789570571,
      "loss": 2.0774,
      "step": 115816
    },
    {
      "epoch": 1.51,
      "grad_norm": 1.0851510763168335,
      "learning_rate": 0.00029748228853870693,
      "loss": 2.2914,
      "step": 115817
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.9613081216812134,
      "learning_rate": 0.0002974781981208248,
      "loss": 2.0093,
      "step": 115818
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.8574447631835938,
      "learning_rate": 0.00029747410770341156,
      "loss": 2.1601,
      "step": 115819
    },
    {
      "epoch": 1.51,
      "grad_norm": 1.063940405845642,
      "learning_rate": 0.000297470017286468,
      "loss": 2.2489,
      "step": 115820
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.9527962803840637,
      "learning_rate": 0.0002974659268699947,
      "loss": 2.2428,
      "step": 115821
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.859251081943512,
      "learning_rate": 0.00029746183645399255,
      "loss": 2.1955,
      "step": 115822
    },
    {
      "epoch": 1.51,
      "grad_norm": 1.0163185596466064,
      "learning_rate": 0.0002974577460384624,
      "loss": 1.9016,
      "step": 115823
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.9226087331771851,
      "learning_rate": 0.0002974536556234048,
      "loss": 2.2905,
      "step": 115824
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.8761247396469116,
      "learning_rate": 0.0002974495652088206,
      "loss": 2.2518,
      "step": 115825
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.8811081647872925,
      "learning_rate": 0.00029744547479471057,
      "loss": 1.9628,
      "step": 115826
    },
    {
      "epoch": 1.51,
      "grad_norm": 1.0374051332473755,
      "learning_rate": 0.00029744138438107554,
      "loss": 1.8769,
      "step": 115827
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.9150124788284302,
      "learning_rate": 0.0002974372939679162,
      "loss": 1.8347,
      "step": 115828
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.978012204170227,
      "learning_rate": 0.0002974332035552333,
      "loss": 2.3046,
      "step": 115829
    },
    {
      "epoch": 1.51,
      "grad_norm": 1.0674234628677368,
      "learning_rate": 0.0002974291131430276,
      "loss": 2.152,
      "step": 115830
    },
    {
      "epoch": 1.51,
      "grad_norm": 1.051201581954956,
      "learning_rate": 0.00029742502273129984,
      "loss": 1.9854,
      "step": 115831
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.9290345311164856,
      "learning_rate": 0.00029742093232005087,
      "loss": 2.36,
      "step": 115832
    },
    {
      "epoch": 1.51,
      "grad_norm": 1.0048730373382568,
      "learning_rate": 0.00029741684190928133,
      "loss": 2.1098,
      "step": 115833
    },
    {
      "epoch": 1.51,
      "grad_norm": 1.0663701295852661,
      "learning_rate": 0.00029741275149899215,
      "loss": 2.1202,
      "step": 115834
    },
    {
      "epoch": 1.51,
      "grad_norm": 1.2318495512008667,
      "learning_rate": 0.00029740866108918403,
      "loss": 2.1308,
      "step": 115835
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.9561286568641663,
      "learning_rate": 0.00029740457067985756,
      "loss": 2.2247,
      "step": 115836
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.9737958908081055,
      "learning_rate": 0.0002974004802710137,
      "loss": 2.1542,
      "step": 115837
    },
    {
      "epoch": 1.51,
      "grad_norm": 1.0492209196090698,
      "learning_rate": 0.000297396389862653,
      "loss": 2.1961,
      "step": 115838
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.9876444339752197,
      "learning_rate": 0.00029739229945477646,
      "loss": 2.2066,
      "step": 115839
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.9379357695579529,
      "learning_rate": 0.00029738820904738476,
      "loss": 2.1368,
      "step": 115840
    },
    {
      "epoch": 1.51,
      "grad_norm": 1.1311944723129272,
      "learning_rate": 0.0002973841186404786,
      "loss": 2.126,
      "step": 115841
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.9712889194488525,
      "learning_rate": 0.00029738002823405895,
      "loss": 2.3527,
      "step": 115842
    },
    {
      "epoch": 1.51,
      "grad_norm": 1.0927919149398804,
      "learning_rate": 0.0002973759378281262,
      "loss": 2.1418,
      "step": 115843
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.9311581254005432,
      "learning_rate": 0.0002973718474226814,
      "loss": 2.1381,
      "step": 115844
    },
    {
      "epoch": 1.51,
      "grad_norm": 1.0706920623779297,
      "learning_rate": 0.0002973677570177252,
      "loss": 2.3015,
      "step": 115845
    },
    {
      "epoch": 1.51,
      "grad_norm": 1.0268166065216064,
      "learning_rate": 0.00029736366661325833,
      "loss": 1.9716,
      "step": 115846
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.9619171619415283,
      "learning_rate": 0.00029735957620928165,
      "loss": 2.0322,
      "step": 115847
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.9941499829292297,
      "learning_rate": 0.000297355485805796,
      "loss": 2.1565,
      "step": 115848
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.9731006622314453,
      "learning_rate": 0.00029735139540280187,
      "loss": 2.1293,
      "step": 115849
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.9959536194801331,
      "learning_rate": 0.0002973473050003002,
      "loss": 2.0628,
      "step": 115850
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.8881909847259521,
      "learning_rate": 0.0002973432145982917,
      "loss": 2.0656,
      "step": 115851
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.8839717507362366,
      "learning_rate": 0.00029733912419677714,
      "loss": 2.0628,
      "step": 115852
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.9588285684585571,
      "learning_rate": 0.00029733503379575723,
      "loss": 1.8909,
      "step": 115853
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.9287942051887512,
      "learning_rate": 0.00029733094339523303,
      "loss": 2.0343,
      "step": 115854
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.8931461572647095,
      "learning_rate": 0.00029732685299520484,
      "loss": 2.194,
      "step": 115855
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.9903298616409302,
      "learning_rate": 0.0002973227625956737,
      "loss": 2.0092,
      "step": 115856
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.8493505120277405,
      "learning_rate": 0.00029731867219664023,
      "loss": 2.1684,
      "step": 115857
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.9239792227745056,
      "learning_rate": 0.00029731458179810537,
      "loss": 2.2123,
      "step": 115858
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.8912975788116455,
      "learning_rate": 0.0002973104914000697,
      "loss": 2.1463,
      "step": 115859
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.9479776620864868,
      "learning_rate": 0.00029730640100253424,
      "loss": 2.1214,
      "step": 115860
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.9678009748458862,
      "learning_rate": 0.00029730231060549943,
      "loss": 2.114,
      "step": 115861
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.9985741972923279,
      "learning_rate": 0.0002972982202089661,
      "loss": 1.9952,
      "step": 115862
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.9289113283157349,
      "learning_rate": 0.00029729412981293517,
      "loss": 2.3036,
      "step": 115863
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.995864748954773,
      "learning_rate": 0.0002972900394174073,
      "loss": 2.2538,
      "step": 115864
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.9258949160575867,
      "learning_rate": 0.00029728594902238325,
      "loss": 2.102,
      "step": 115865
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.9086061120033264,
      "learning_rate": 0.00029728185862786395,
      "loss": 2.2567,
      "step": 115866
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.8530383110046387,
      "learning_rate": 0.0002972777682338498,
      "loss": 2.1203,
      "step": 115867
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.901573121547699,
      "learning_rate": 0.0002972736778403418,
      "loss": 2.145,
      "step": 115868
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.9258228540420532,
      "learning_rate": 0.00029726958744734073,
      "loss": 2.3897,
      "step": 115869
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.9188003540039062,
      "learning_rate": 0.0002972654970548472,
      "loss": 2.1985,
      "step": 115870
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.9541025757789612,
      "learning_rate": 0.00029726140666286213,
      "loss": 2.1703,
      "step": 115871
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.9471727013587952,
      "learning_rate": 0.00029725731627138635,
      "loss": 2.1854,
      "step": 115872
    },
    {
      "epoch": 1.51,
      "grad_norm": 1.0131316184997559,
      "learning_rate": 0.0002972532258804203,
      "loss": 2.1456,
      "step": 115873
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.9795238375663757,
      "learning_rate": 0.000297249135489965,
      "loss": 2.2292,
      "step": 115874
    },
    {
      "epoch": 1.51,
      "grad_norm": 1.001186490058899,
      "learning_rate": 0.0002972450451000211,
      "loss": 2.1568,
      "step": 115875
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.9532336592674255,
      "learning_rate": 0.0002972409547105894,
      "loss": 2.0461,
      "step": 115876
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.9740975499153137,
      "learning_rate": 0.00029723686432167064,
      "loss": 1.9984,
      "step": 115877
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.9414068460464478,
      "learning_rate": 0.00029723277393326576,
      "loss": 2.1042,
      "step": 115878
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.9542135000228882,
      "learning_rate": 0.0002972286835453752,
      "loss": 2.1499,
      "step": 115879
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.9927197694778442,
      "learning_rate": 0.00029722459315799994,
      "loss": 1.9922,
      "step": 115880
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.9431851506233215,
      "learning_rate": 0.0002972205027711406,
      "loss": 2.2357,
      "step": 115881
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.8942843675613403,
      "learning_rate": 0.00029721641238479807,
      "loss": 2.191,
      "step": 115882
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.9361435770988464,
      "learning_rate": 0.00029721232199897304,
      "loss": 1.9423,
      "step": 115883
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.9323927164077759,
      "learning_rate": 0.00029720823161366643,
      "loss": 2.0108,
      "step": 115884
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.9037558436393738,
      "learning_rate": 0.00029720414122887877,
      "loss": 2.0842,
      "step": 115885
    },
    {
      "epoch": 1.51,
      "grad_norm": 1.401092767715454,
      "learning_rate": 0.00029720005084461087,
      "loss": 2.218,
      "step": 115886
    },
    {
      "epoch": 1.51,
      "grad_norm": 1.0121184587478638,
      "learning_rate": 0.0002971959604608635,
      "loss": 2.2028,
      "step": 115887
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.913396954536438,
      "learning_rate": 0.0002971918700776375,
      "loss": 2.2432,
      "step": 115888
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.936459481716156,
      "learning_rate": 0.0002971877796949336,
      "loss": 2.1771,
      "step": 115889
    },
    {
      "epoch": 1.51,
      "grad_norm": 1.0295395851135254,
      "learning_rate": 0.0002971836893127527,
      "loss": 2.2064,
      "step": 115890
    },
    {
      "epoch": 1.51,
      "grad_norm": 1.243566632270813,
      "learning_rate": 0.0002971795989310952,
      "loss": 2.0676,
      "step": 115891
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.8992999196052551,
      "learning_rate": 0.00029717550854996207,
      "loss": 2.163,
      "step": 115892
    },
    {
      "epoch": 1.51,
      "grad_norm": 1.162833571434021,
      "learning_rate": 0.0002971714181693541,
      "loss": 2.0147,
      "step": 115893
    },
    {
      "epoch": 1.51,
      "grad_norm": 1.0589550733566284,
      "learning_rate": 0.000297167327789272,
      "loss": 2.1153,
      "step": 115894
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.9287649989128113,
      "learning_rate": 0.00029716323740971653,
      "loss": 2.1253,
      "step": 115895
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.8928220272064209,
      "learning_rate": 0.0002971591470306886,
      "loss": 2.0099,
      "step": 115896
    },
    {
      "epoch": 1.51,
      "grad_norm": 1.0240576267242432,
      "learning_rate": 0.0002971550566521888,
      "loss": 2.0257,
      "step": 115897
    },
    {
      "epoch": 1.51,
      "grad_norm": 1.1186281442642212,
      "learning_rate": 0.00029715096627421785,
      "loss": 1.9546,
      "step": 115898
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.9377061128616333,
      "learning_rate": 0.0002971468758967766,
      "loss": 2.4036,
      "step": 115899
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.9941816926002502,
      "learning_rate": 0.0002971427855198658,
      "loss": 2.1654,
      "step": 115900
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.9494842290878296,
      "learning_rate": 0.0002971386951434862,
      "loss": 2.0918,
      "step": 115901
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.8950369954109192,
      "learning_rate": 0.00029713460476763864,
      "loss": 2.083,
      "step": 115902
    },
    {
      "epoch": 1.51,
      "grad_norm": 1.0383055210113525,
      "learning_rate": 0.00029713051439232376,
      "loss": 2.1815,
      "step": 115903
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.9093443751335144,
      "learning_rate": 0.00029712642401754236,
      "loss": 1.9062,
      "step": 115904
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.9121260046958923,
      "learning_rate": 0.0002971223336432953,
      "loss": 2.0883,
      "step": 115905
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.9131925702095032,
      "learning_rate": 0.00029711824326958316,
      "loss": 2.3866,
      "step": 115906
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.9037438631057739,
      "learning_rate": 0.0002971141528964068,
      "loss": 2.0914,
      "step": 115907
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.9416932463645935,
      "learning_rate": 0.000297110062523767,
      "loss": 2.1249,
      "step": 115908
    },
    {
      "epoch": 1.51,
      "grad_norm": 1.0386868715286255,
      "learning_rate": 0.00029710597215166454,
      "loss": 1.9892,
      "step": 115909
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.9447420835494995,
      "learning_rate": 0.00029710188178010003,
      "loss": 2.0888,
      "step": 115910
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.9863002300262451,
      "learning_rate": 0.00029709779140907436,
      "loss": 2.1471,
      "step": 115911
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.8594768643379211,
      "learning_rate": 0.00029709370103858836,
      "loss": 2.1085,
      "step": 115912
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.8945272564888,
      "learning_rate": 0.0002970896106686426,
      "loss": 2.0711,
      "step": 115913
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.8958072066307068,
      "learning_rate": 0.00029708552029923793,
      "loss": 2.1969,
      "step": 115914
    },
    {
      "epoch": 1.51,
      "grad_norm": 1.0400266647338867,
      "learning_rate": 0.0002970814299303752,
      "loss": 2.0779,
      "step": 115915
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.9733620285987854,
      "learning_rate": 0.00029707733956205504,
      "loss": 2.2507,
      "step": 115916
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.8567895293235779,
      "learning_rate": 0.00029707324919427824,
      "loss": 2.1067,
      "step": 115917
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.8886187672615051,
      "learning_rate": 0.00029706915882704555,
      "loss": 2.0759,
      "step": 115918
    },
    {
      "epoch": 1.51,
      "grad_norm": 1.032781720161438,
      "learning_rate": 0.00029706506846035783,
      "loss": 2.2007,
      "step": 115919
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.9252414703369141,
      "learning_rate": 0.00029706097809421573,
      "loss": 2.1229,
      "step": 115920
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.8971239924430847,
      "learning_rate": 0.00029705688772862013,
      "loss": 2.4453,
      "step": 115921
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.958065390586853,
      "learning_rate": 0.0002970527973635716,
      "loss": 2.1557,
      "step": 115922
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.8407676815986633,
      "learning_rate": 0.00029704870699907106,
      "loss": 1.82,
      "step": 115923
    },
    {
      "epoch": 1.51,
      "grad_norm": 1.051431655883789,
      "learning_rate": 0.0002970446166351192,
      "loss": 2.2881,
      "step": 115924
    },
    {
      "epoch": 1.51,
      "grad_norm": 1.0999763011932373,
      "learning_rate": 0.0002970405262717168,
      "loss": 1.9428,
      "step": 115925
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.8867986798286438,
      "learning_rate": 0.0002970364359088647,
      "loss": 2.3284,
      "step": 115926
    },
    {
      "epoch": 1.51,
      "grad_norm": 1.0728684663772583,
      "learning_rate": 0.00029703234554656356,
      "loss": 2.2345,
      "step": 115927
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.9885552525520325,
      "learning_rate": 0.00029702825518481405,
      "loss": 2.1062,
      "step": 115928
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.9659079313278198,
      "learning_rate": 0.0002970241648236171,
      "loss": 2.0418,
      "step": 115929
    },
    {
      "epoch": 1.51,
      "grad_norm": 1.1052923202514648,
      "learning_rate": 0.00029702007446297344,
      "loss": 2.061,
      "step": 115930
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.9119546413421631,
      "learning_rate": 0.00029701598410288374,
      "loss": 2.166,
      "step": 115931
    },
    {
      "epoch": 1.51,
      "grad_norm": 1.0253148078918457,
      "learning_rate": 0.0002970118937433489,
      "loss": 2.1128,
      "step": 115932
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.9183173775672913,
      "learning_rate": 0.0002970078033843697,
      "loss": 2.1917,
      "step": 115933
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.9727186560630798,
      "learning_rate": 0.00029700371302594666,
      "loss": 2.1658,
      "step": 115934
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.9181843400001526,
      "learning_rate": 0.0002969996226680807,
      "loss": 1.8988,
      "step": 115935
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.983223557472229,
      "learning_rate": 0.00029699553231077254,
      "loss": 2.1065,
      "step": 115936
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.9018905162811279,
      "learning_rate": 0.000296991441954023,
      "loss": 2.1203,
      "step": 115937
    },
    {
      "epoch": 1.51,
      "grad_norm": 1.0593332052230835,
      "learning_rate": 0.0002969873515978328,
      "loss": 2.329,
      "step": 115938
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.8383539319038391,
      "learning_rate": 0.00029698326124220285,
      "loss": 2.2939,
      "step": 115939
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.9447163343429565,
      "learning_rate": 0.0002969791708871336,
      "loss": 1.8136,
      "step": 115940
    },
    {
      "epoch": 1.51,
      "grad_norm": 1.0046422481536865,
      "learning_rate": 0.00029697508053262603,
      "loss": 2.1549,
      "step": 115941
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.9070169925689697,
      "learning_rate": 0.0002969709901786808,
      "loss": 2.0702,
      "step": 115942
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.9549937844276428,
      "learning_rate": 0.00029696689982529876,
      "loss": 2.1138,
      "step": 115943
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.8931307792663574,
      "learning_rate": 0.0002969628094724806,
      "loss": 2.2977,
      "step": 115944
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.9029163718223572,
      "learning_rate": 0.0002969587191202273,
      "loss": 2.2618,
      "step": 115945
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.8365235924720764,
      "learning_rate": 0.00029695462876853923,
      "loss": 2.1325,
      "step": 115946
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.8812345862388611,
      "learning_rate": 0.00029695053841741735,
      "loss": 2.2355,
      "step": 115947
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.9660747051239014,
      "learning_rate": 0.00029694644806686245,
      "loss": 2.5625,
      "step": 115948
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.9599767327308655,
      "learning_rate": 0.0002969423577168753,
      "loss": 2.4589,
      "step": 115949
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.8855105042457581,
      "learning_rate": 0.0002969382673674566,
      "loss": 1.9668,
      "step": 115950
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.9231044054031372,
      "learning_rate": 0.00029693417701860724,
      "loss": 2.2482,
      "step": 115951
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.957744836807251,
      "learning_rate": 0.00029693008667032775,
      "loss": 2.1098,
      "step": 115952
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.9055198431015015,
      "learning_rate": 0.00029692599632261905,
      "loss": 2.0845,
      "step": 115953
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.8794133067131042,
      "learning_rate": 0.0002969219059754818,
      "loss": 2.4783,
      "step": 115954
    },
    {
      "epoch": 1.51,
      "grad_norm": 1.1180932521820068,
      "learning_rate": 0.00029691781562891684,
      "loss": 2.1768,
      "step": 115955
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.857104480266571,
      "learning_rate": 0.00029691372528292497,
      "loss": 2.0662,
      "step": 115956
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.9503344893455505,
      "learning_rate": 0.000296909634937507,
      "loss": 1.9267,
      "step": 115957
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.8815311193466187,
      "learning_rate": 0.0002969055445926634,
      "loss": 2.2664,
      "step": 115958
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.971819281578064,
      "learning_rate": 0.0002969014542483952,
      "loss": 2.3512,
      "step": 115959
    },
    {
      "epoch": 1.51,
      "grad_norm": 1.0333259105682373,
      "learning_rate": 0.00029689736390470304,
      "loss": 2.1196,
      "step": 115960
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.8791298270225525,
      "learning_rate": 0.0002968932735615878,
      "loss": 2.1136,
      "step": 115961
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.9317551851272583,
      "learning_rate": 0.00029688918321905006,
      "loss": 2.048,
      "step": 115962
    },
    {
      "epoch": 1.51,
      "grad_norm": 1.173667550086975,
      "learning_rate": 0.0002968850928770908,
      "loss": 2.0474,
      "step": 115963
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.9820033311843872,
      "learning_rate": 0.0002968810025357106,
      "loss": 1.9608,
      "step": 115964
    },
    {
      "epoch": 1.51,
      "grad_norm": 1.106134057044983,
      "learning_rate": 0.0002968769121949102,
      "loss": 2.121,
      "step": 115965
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.8795143961906433,
      "learning_rate": 0.00029687282185469053,
      "loss": 2.3255,
      "step": 115966
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.9595409631729126,
      "learning_rate": 0.0002968687315150522,
      "loss": 2.3043,
      "step": 115967
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.8926742076873779,
      "learning_rate": 0.00029686464117599607,
      "loss": 2.2143,
      "step": 115968
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.955973207950592,
      "learning_rate": 0.00029686055083752287,
      "loss": 2.346,
      "step": 115969
    },
    {
      "epoch": 1.51,
      "grad_norm": 1.0317790508270264,
      "learning_rate": 0.0002968564604996334,
      "loss": 2.0715,
      "step": 115970
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.9404122829437256,
      "learning_rate": 0.0002968523701623283,
      "loss": 2.015,
      "step": 115971
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.9226855039596558,
      "learning_rate": 0.00029684827982560843,
      "loss": 2.1036,
      "step": 115972
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.9055246710777283,
      "learning_rate": 0.0002968441894894745,
      "loss": 2.2867,
      "step": 115973
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.8220526576042175,
      "learning_rate": 0.0002968400991539273,
      "loss": 2.0009,
      "step": 115974
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.8659515380859375,
      "learning_rate": 0.00029683600881896756,
      "loss": 2.0818,
      "step": 115975
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.9572465419769287,
      "learning_rate": 0.0002968319184845962,
      "loss": 2.2496,
      "step": 115976
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.9213741421699524,
      "learning_rate": 0.00029682782815081376,
      "loss": 2.364,
      "step": 115977
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.9344606995582581,
      "learning_rate": 0.00029682373781762106,
      "loss": 2.0734,
      "step": 115978
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.9915259480476379,
      "learning_rate": 0.00029681964748501887,
      "loss": 2.1596,
      "step": 115979
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.9134637117385864,
      "learning_rate": 0.000296815557153008,
      "loss": 2.0293,
      "step": 115980
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.8272541165351868,
      "learning_rate": 0.00029681146682158916,
      "loss": 2.0598,
      "step": 115981
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.9342600703239441,
      "learning_rate": 0.0002968073764907632,
      "loss": 1.9647,
      "step": 115982
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.9266632199287415,
      "learning_rate": 0.00029680328616053076,
      "loss": 2.058,
      "step": 115983
    },
    {
      "epoch": 1.51,
      "grad_norm": 1.3197052478790283,
      "learning_rate": 0.0002967991958308926,
      "loss": 2.0495,
      "step": 115984
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.9512539505958557,
      "learning_rate": 0.0002967951055018496,
      "loss": 2.174,
      "step": 115985
    },
    {
      "epoch": 1.51,
      "grad_norm": 1.0978097915649414,
      "learning_rate": 0.0002967910151734024,
      "loss": 2.3726,
      "step": 115986
    },
    {
      "epoch": 1.51,
      "grad_norm": 1.0500361919403076,
      "learning_rate": 0.0002967869248455518,
      "loss": 2.1654,
      "step": 115987
    },
    {
      "epoch": 1.51,
      "grad_norm": 1.0938916206359863,
      "learning_rate": 0.0002967828345182987,
      "loss": 2.1972,
      "step": 115988
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.9393103122711182,
      "learning_rate": 0.0002967787441916437,
      "loss": 2.1577,
      "step": 115989
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.9380826950073242,
      "learning_rate": 0.00029677465386558746,
      "loss": 1.7055,
      "step": 115990
    },
    {
      "epoch": 1.51,
      "grad_norm": 1.0067400932312012,
      "learning_rate": 0.0002967705635401309,
      "loss": 2.2609,
      "step": 115991
    },
    {
      "epoch": 1.51,
      "grad_norm": 1.1508784294128418,
      "learning_rate": 0.0002967664732152748,
      "loss": 2.1538,
      "step": 115992
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.9596096277236938,
      "learning_rate": 0.0002967623828910199,
      "loss": 2.063,
      "step": 115993
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.9418678879737854,
      "learning_rate": 0.0002967582925673669,
      "loss": 2.4094,
      "step": 115994
    },
    {
      "epoch": 1.51,
      "grad_norm": 1.2461180686950684,
      "learning_rate": 0.0002967542022443166,
      "loss": 1.9048,
      "step": 115995
    },
    {
      "epoch": 1.51,
      "grad_norm": 1.1522561311721802,
      "learning_rate": 0.0002967501119218698,
      "loss": 2.1679,
      "step": 115996
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.9523598551750183,
      "learning_rate": 0.0002967460216000271,
      "loss": 2.2038,
      "step": 115997
    },
    {
      "epoch": 1.51,
      "grad_norm": 1.0681495666503906,
      "learning_rate": 0.00029674193127878943,
      "loss": 2.1338,
      "step": 115998
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.9756332635879517,
      "learning_rate": 0.00029673784095815754,
      "loss": 2.033,
      "step": 115999
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.9783294200897217,
      "learning_rate": 0.0002967337506381322,
      "loss": 2.0344,
      "step": 116000
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.913808286190033,
      "learning_rate": 0.00029672966031871396,
      "loss": 2.3654,
      "step": 116001
    },
    {
      "epoch": 1.51,
      "grad_norm": 1.0148457288742065,
      "learning_rate": 0.0002967255699999038,
      "loss": 1.9012,
      "step": 116002
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.9619152545928955,
      "learning_rate": 0.00029672147968170246,
      "loss": 2.252,
      "step": 116003
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.9588757753372192,
      "learning_rate": 0.0002967173893641106,
      "loss": 2.2573,
      "step": 116004
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.9303293228149414,
      "learning_rate": 0.00029671329904712906,
      "loss": 2.2472,
      "step": 116005
    },
    {
      "epoch": 1.51,
      "grad_norm": 1.0623217821121216,
      "learning_rate": 0.0002967092087307587,
      "loss": 2.219,
      "step": 116006
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.9999979138374329,
      "learning_rate": 0.000296705118415,
      "loss": 2.0524,
      "step": 116007
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.8558494448661804,
      "learning_rate": 0.0002967010280998539,
      "loss": 2.1248,
      "step": 116008
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.9583712816238403,
      "learning_rate": 0.0002966969377853212,
      "loss": 2.3475,
      "step": 116009
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.8464283347129822,
      "learning_rate": 0.00029669284747140255,
      "loss": 2.1258,
      "step": 116010
    },
    {
      "epoch": 1.51,
      "grad_norm": 1.0237897634506226,
      "learning_rate": 0.00029668875715809884,
      "loss": 1.9989,
      "step": 116011
    },
    {
      "epoch": 1.51,
      "grad_norm": 1.0675652027130127,
      "learning_rate": 0.0002966846668454107,
      "loss": 2.1344,
      "step": 116012
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.9137242436408997,
      "learning_rate": 0.00029668057653333893,
      "loss": 1.979,
      "step": 116013
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.9745827317237854,
      "learning_rate": 0.0002966764862218843,
      "loss": 2.3098,
      "step": 116014
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.9685368537902832,
      "learning_rate": 0.0002966723959110476,
      "loss": 2.3871,
      "step": 116015
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.8801172971725464,
      "learning_rate": 0.0002966683056008295,
      "loss": 2.0311,
      "step": 116016
    },
    {
      "epoch": 1.51,
      "grad_norm": 1.1212979555130005,
      "learning_rate": 0.0002966642152912309,
      "loss": 2.3123,
      "step": 116017
    },
    {
      "epoch": 1.51,
      "grad_norm": 1.0097458362579346,
      "learning_rate": 0.00029666012498225254,
      "loss": 2.0938,
      "step": 116018
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.8741416335105896,
      "learning_rate": 0.00029665603467389506,
      "loss": 1.9911,
      "step": 116019
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.8862367868423462,
      "learning_rate": 0.0002966519443661592,
      "loss": 2.2229,
      "step": 116020
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.9017871022224426,
      "learning_rate": 0.0002966478540590459,
      "loss": 2.1963,
      "step": 116021
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.9823712110519409,
      "learning_rate": 0.0002966437637525558,
      "loss": 2.2034,
      "step": 116022
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.9600515961647034,
      "learning_rate": 0.0002966396734466897,
      "loss": 2.0279,
      "step": 116023
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.9254961609840393,
      "learning_rate": 0.00029663558314144844,
      "loss": 2.0506,
      "step": 116024
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.9561853408813477,
      "learning_rate": 0.00029663149283683254,
      "loss": 2.0577,
      "step": 116025
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.9825607538223267,
      "learning_rate": 0.000296627402532843,
      "loss": 2.2248,
      "step": 116026
    },
    {
      "epoch": 1.51,
      "grad_norm": 1.1000406742095947,
      "learning_rate": 0.0002966233122294804,
      "loss": 2.2335,
      "step": 116027
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.862450361251831,
      "learning_rate": 0.0002966192219267456,
      "loss": 2.3205,
      "step": 116028
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.910766065120697,
      "learning_rate": 0.0002966151316246394,
      "loss": 2.3638,
      "step": 116029
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.8790744543075562,
      "learning_rate": 0.0002966110413231626,
      "loss": 1.8602,
      "step": 116030
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.9480071663856506,
      "learning_rate": 0.0002966069510223158,
      "loss": 1.8698,
      "step": 116031
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.9163511395454407,
      "learning_rate": 0.00029660286072209974,
      "loss": 2.0855,
      "step": 116032
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.9163123369216919,
      "learning_rate": 0.00029659877042251536,
      "loss": 2.0291,
      "step": 116033
    },
    {
      "epoch": 1.51,
      "grad_norm": 1.1130595207214355,
      "learning_rate": 0.0002965946801235633,
      "loss": 2.0532,
      "step": 116034
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.9701082110404968,
      "learning_rate": 0.0002965905898252443,
      "loss": 2.0067,
      "step": 116035
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.8499640822410583,
      "learning_rate": 0.00029658649952755923,
      "loss": 2.079,
      "step": 116036
    },
    {
      "epoch": 1.51,
      "grad_norm": 1.037579894065857,
      "learning_rate": 0.00029658240923050895,
      "loss": 2.0991,
      "step": 116037
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.9499582648277283,
      "learning_rate": 0.00029657831893409385,
      "loss": 2.0457,
      "step": 116038
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.985615074634552,
      "learning_rate": 0.000296574228638315,
      "loss": 2.1407,
      "step": 116039
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.8885001540184021,
      "learning_rate": 0.000296570138343173,
      "loss": 2.3555,
      "step": 116040
    },
    {
      "epoch": 1.51,
      "grad_norm": 1.1303706169128418,
      "learning_rate": 0.00029656604804866873,
      "loss": 2.1412,
      "step": 116041
    },
    {
      "epoch": 1.51,
      "grad_norm": 1.0628496408462524,
      "learning_rate": 0.0002965619577548028,
      "loss": 2.2468,
      "step": 116042
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.9114109873771667,
      "learning_rate": 0.0002965578674615763,
      "loss": 2.4907,
      "step": 116043
    },
    {
      "epoch": 1.51,
      "grad_norm": 1.0054285526275635,
      "learning_rate": 0.0002965537771689896,
      "loss": 2.1482,
      "step": 116044
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.9299290180206299,
      "learning_rate": 0.0002965496868770436,
      "loss": 2.1606,
      "step": 116045
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.9479866623878479,
      "learning_rate": 0.0002965455965857391,
      "loss": 2.1941,
      "step": 116046
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.9209719896316528,
      "learning_rate": 0.0002965415062950768,
      "loss": 2.387,
      "step": 116047
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.8798415064811707,
      "learning_rate": 0.00029653741600505753,
      "loss": 2.1995,
      "step": 116048
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.9672124981880188,
      "learning_rate": 0.00029653332571568216,
      "loss": 2.0631,
      "step": 116049
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.9346138834953308,
      "learning_rate": 0.0002965292354269512,
      "loss": 2.0877,
      "step": 116050
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.9515356421470642,
      "learning_rate": 0.0002965251451388654,
      "loss": 2.2648,
      "step": 116051
    },
    {
      "epoch": 1.51,
      "grad_norm": 1.1066508293151855,
      "learning_rate": 0.00029652105485142576,
      "loss": 2.1702,
      "step": 116052
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.9493190050125122,
      "learning_rate": 0.00029651696456463284,
      "loss": 2.3179,
      "step": 116053
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.9439936876296997,
      "learning_rate": 0.0002965128742784875,
      "loss": 2.1257,
      "step": 116054
    },
    {
      "epoch": 1.51,
      "grad_norm": 1.0767782926559448,
      "learning_rate": 0.0002965087839929907,
      "loss": 2.024,
      "step": 116055
    },
    {
      "epoch": 1.51,
      "grad_norm": 1.0162122249603271,
      "learning_rate": 0.0002965046937081428,
      "loss": 2.172,
      "step": 116056
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.908723771572113,
      "learning_rate": 0.00029650060342394476,
      "loss": 2.192,
      "step": 116057
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.9186471104621887,
      "learning_rate": 0.0002964965131403973,
      "loss": 1.9683,
      "step": 116058
    },
    {
      "epoch": 1.51,
      "grad_norm": 1.001466155052185,
      "learning_rate": 0.00029649242285750116,
      "loss": 1.9522,
      "step": 116059
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.9813340306282043,
      "learning_rate": 0.0002964883325752572,
      "loss": 2.1105,
      "step": 116060
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.8811023235321045,
      "learning_rate": 0.00029648424229366627,
      "loss": 1.9194,
      "step": 116061
    },
    {
      "epoch": 1.51,
      "grad_norm": 1.1452968120574951,
      "learning_rate": 0.0002964801520127288,
      "loss": 2.1412,
      "step": 116062
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.9596481919288635,
      "learning_rate": 0.00029647606173244576,
      "loss": 2.1483,
      "step": 116063
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.9881354570388794,
      "learning_rate": 0.0002964719714528179,
      "loss": 2.2475,
      "step": 116064
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.9622073173522949,
      "learning_rate": 0.000296467881173846,
      "loss": 2.0425,
      "step": 116065
    },
    {
      "epoch": 1.51,
      "grad_norm": 1.0257548093795776,
      "learning_rate": 0.0002964637908955307,
      "loss": 2.2329,
      "step": 116066
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.976715087890625,
      "learning_rate": 0.000296459700617873,
      "loss": 1.9572,
      "step": 116067
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.9846445918083191,
      "learning_rate": 0.00029645561034087343,
      "loss": 2.1826,
      "step": 116068
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.9759297370910645,
      "learning_rate": 0.0002964515200645328,
      "loss": 1.9582,
      "step": 116069
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.9059351682662964,
      "learning_rate": 0.00029644742978885194,
      "loss": 2.2163,
      "step": 116070
    },
    {
      "epoch": 1.51,
      "grad_norm": 1.1686666011810303,
      "learning_rate": 0.0002964433395138315,
      "loss": 2.0968,
      "step": 116071
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.9403571486473083,
      "learning_rate": 0.00029643924923947233,
      "loss": 2.3261,
      "step": 116072
    },
    {
      "epoch": 1.51,
      "grad_norm": 1.0055665969848633,
      "learning_rate": 0.0002964351589657753,
      "loss": 2.0792,
      "step": 116073
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.9373034834861755,
      "learning_rate": 0.00029643106869274094,
      "loss": 2.0038,
      "step": 116074
    },
    {
      "epoch": 1.51,
      "grad_norm": 1.053936243057251,
      "learning_rate": 0.0002964269784203701,
      "loss": 2.1613,
      "step": 116075
    },
    {
      "epoch": 1.51,
      "grad_norm": 1.0457154512405396,
      "learning_rate": 0.00029642288814866357,
      "loss": 2.3088,
      "step": 116076
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.9112332463264465,
      "learning_rate": 0.00029641879787762205,
      "loss": 2.0653,
      "step": 116077
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.8907142877578735,
      "learning_rate": 0.00029641470760724634,
      "loss": 1.974,
      "step": 116078
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.9368336200714111,
      "learning_rate": 0.0002964106173375373,
      "loss": 2.2134,
      "step": 116079
    },
    {
      "epoch": 1.51,
      "grad_norm": 1.0124270915985107,
      "learning_rate": 0.00029640652706849555,
      "loss": 1.9121,
      "step": 116080
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.9257943034172058,
      "learning_rate": 0.0002964024368001219,
      "loss": 2.1581,
      "step": 116081
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.9346048831939697,
      "learning_rate": 0.00029639834653241715,
      "loss": 2.1772,
      "step": 116082
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.8831276893615723,
      "learning_rate": 0.00029639425626538186,
      "loss": 2.0166,
      "step": 116083
    },
    {
      "epoch": 1.51,
      "grad_norm": 1.5976639986038208,
      "learning_rate": 0.00029639016599901705,
      "loss": 1.9565,
      "step": 116084
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.8593337535858154,
      "learning_rate": 0.0002963860757333234,
      "loss": 2.1465,
      "step": 116085
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.9553371667861938,
      "learning_rate": 0.0002963819854683016,
      "loss": 2.1436,
      "step": 116086
    },
    {
      "epoch": 1.51,
      "grad_norm": 1.0245028734207153,
      "learning_rate": 0.00029637789520395243,
      "loss": 2.125,
      "step": 116087
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.968208909034729,
      "learning_rate": 0.0002963738049402768,
      "loss": 2.0903,
      "step": 116088
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.9102828502655029,
      "learning_rate": 0.00029636971467727525,
      "loss": 2.0374,
      "step": 116089
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.9215880632400513,
      "learning_rate": 0.0002963656244149486,
      "loss": 2.1682,
      "step": 116090
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.8796347975730896,
      "learning_rate": 0.0002963615341532978,
      "loss": 2.2363,
      "step": 116091
    },
    {
      "epoch": 1.51,
      "grad_norm": 1.190172553062439,
      "learning_rate": 0.0002963574438923233,
      "loss": 1.9997,
      "step": 116092
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.9763033390045166,
      "learning_rate": 0.0002963533536320261,
      "loss": 1.9325,
      "step": 116093
    },
    {
      "epoch": 1.51,
      "grad_norm": 1.058928370475769,
      "learning_rate": 0.00029634926337240684,
      "loss": 2.2026,
      "step": 116094
    },
    {
      "epoch": 1.51,
      "grad_norm": 1.338974952697754,
      "learning_rate": 0.0002963451731134663,
      "loss": 1.7491,
      "step": 116095
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.9128574728965759,
      "learning_rate": 0.0002963410828552054,
      "loss": 2.2548,
      "step": 116096
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.9214067459106445,
      "learning_rate": 0.00029633699259762466,
      "loss": 2.0551,
      "step": 116097
    },
    {
      "epoch": 1.51,
      "grad_norm": 1.0861341953277588,
      "learning_rate": 0.000296332902340725,
      "loss": 2.2087,
      "step": 116098
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.8859596252441406,
      "learning_rate": 0.000296328812084507,
      "loss": 2.096,
      "step": 116099
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.8660561442375183,
      "learning_rate": 0.0002963247218289716,
      "loss": 2.1665,
      "step": 116100
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.9989972710609436,
      "learning_rate": 0.00029632063157411954,
      "loss": 2.1939,
      "step": 116101
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.9307301640510559,
      "learning_rate": 0.0002963165413199515,
      "loss": 2.1546,
      "step": 116102
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.9991516470909119,
      "learning_rate": 0.0002963124510664683,
      "loss": 2.236,
      "step": 116103
    },
    {
      "epoch": 1.51,
      "grad_norm": 1.070028305053711,
      "learning_rate": 0.0002963083608136708,
      "loss": 2.3776,
      "step": 116104
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.924198567867279,
      "learning_rate": 0.0002963042705615595,
      "loss": 2.1683,
      "step": 116105
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.999877393245697,
      "learning_rate": 0.00029630018031013536,
      "loss": 2.2256,
      "step": 116106
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.9748679995536804,
      "learning_rate": 0.00029629609005939905,
      "loss": 2.1154,
      "step": 116107
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.9468793272972107,
      "learning_rate": 0.0002962919998093513,
      "loss": 2.2312,
      "step": 116108
    },
    {
      "epoch": 1.51,
      "grad_norm": 1.0403434038162231,
      "learning_rate": 0.00029628790955999303,
      "loss": 2.0613,
      "step": 116109
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.9078888893127441,
      "learning_rate": 0.000296283819311325,
      "loss": 2.0526,
      "step": 116110
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.9493918418884277,
      "learning_rate": 0.00029627972906334777,
      "loss": 2.1547,
      "step": 116111
    },
    {
      "epoch": 1.51,
      "grad_norm": 1.1040624380111694,
      "learning_rate": 0.00029627563881606226,
      "loss": 2.1406,
      "step": 116112
    },
    {
      "epoch": 1.51,
      "grad_norm": 1.1842409372329712,
      "learning_rate": 0.00029627154856946906,
      "loss": 2.0986,
      "step": 116113
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.9861862659454346,
      "learning_rate": 0.0002962674583235691,
      "loss": 2.2695,
      "step": 116114
    },
    {
      "epoch": 1.51,
      "grad_norm": 1.005558729171753,
      "learning_rate": 0.00029626336807836315,
      "loss": 2.2065,
      "step": 116115
    },
    {
      "epoch": 1.51,
      "grad_norm": 1.329331874847412,
      "learning_rate": 0.00029625927783385195,
      "loss": 2.0842,
      "step": 116116
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.9728603959083557,
      "learning_rate": 0.00029625518759003616,
      "loss": 2.1256,
      "step": 116117
    },
    {
      "epoch": 1.51,
      "grad_norm": 1.0544843673706055,
      "learning_rate": 0.00029625109734691653,
      "loss": 2.1895,
      "step": 116118
    },
    {
      "epoch": 1.51,
      "grad_norm": 1.0982248783111572,
      "learning_rate": 0.00029624700710449394,
      "loss": 2.0399,
      "step": 116119
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.993021547794342,
      "learning_rate": 0.0002962429168627691,
      "loss": 2.1336,
      "step": 116120
    },
    {
      "epoch": 1.51,
      "grad_norm": 1.0290024280548096,
      "learning_rate": 0.0002962388266217428,
      "loss": 2.1207,
      "step": 116121
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.9187003970146179,
      "learning_rate": 0.00029623473638141584,
      "loss": 2.2575,
      "step": 116122
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.954666793346405,
      "learning_rate": 0.0002962306461417888,
      "loss": 2.2808,
      "step": 116123
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.9962711334228516,
      "learning_rate": 0.0002962265559028626,
      "loss": 2.041,
      "step": 116124
    },
    {
      "epoch": 1.51,
      "grad_norm": 1.2874807119369507,
      "learning_rate": 0.00029622246566463794,
      "loss": 2.227,
      "step": 116125
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.9786165952682495,
      "learning_rate": 0.00029621837542711556,
      "loss": 2.1914,
      "step": 116126
    },
    {
      "epoch": 1.51,
      "grad_norm": 1.0139789581298828,
      "learning_rate": 0.0002962142851902963,
      "loss": 1.8984,
      "step": 116127
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.9956519603729248,
      "learning_rate": 0.000296210194954181,
      "loss": 2.3216,
      "step": 116128
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.9376566410064697,
      "learning_rate": 0.0002962061047187701,
      "loss": 2.2094,
      "step": 116129
    },
    {
      "epoch": 1.51,
      "grad_norm": 1.007297158241272,
      "learning_rate": 0.00029620201448406463,
      "loss": 2.4012,
      "step": 116130
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.9523122310638428,
      "learning_rate": 0.0002961979242500652,
      "loss": 2.2584,
      "step": 116131
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.923665463924408,
      "learning_rate": 0.00029619383401677273,
      "loss": 2.1408,
      "step": 116132
    },
    {
      "epoch": 1.51,
      "grad_norm": 1.0394067764282227,
      "learning_rate": 0.00029618974378418786,
      "loss": 2.077,
      "step": 116133
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.9575685262680054,
      "learning_rate": 0.0002961856535523115,
      "loss": 2.3926,
      "step": 116134
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.9514532685279846,
      "learning_rate": 0.00029618156332114425,
      "loss": 1.8799,
      "step": 116135
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.8908361196517944,
      "learning_rate": 0.00029617747309068686,
      "loss": 2.0769,
      "step": 116136
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.987646758556366,
      "learning_rate": 0.0002961733828609401,
      "loss": 2.1609,
      "step": 116137
    },
    {
      "epoch": 1.51,
      "grad_norm": 1.038265347480774,
      "learning_rate": 0.00029616929263190486,
      "loss": 1.9928,
      "step": 116138
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.9685915112495422,
      "learning_rate": 0.0002961652024035818,
      "loss": 2.3437,
      "step": 116139
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.945087194442749,
      "learning_rate": 0.0002961611121759718,
      "loss": 2.4153,
      "step": 116140
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.993816614151001,
      "learning_rate": 0.00029615702194907544,
      "loss": 2.2234,
      "step": 116141
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.881068229675293,
      "learning_rate": 0.0002961529317228935,
      "loss": 2.2394,
      "step": 116142
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.9318035244941711,
      "learning_rate": 0.00029614884149742685,
      "loss": 2.2318,
      "step": 116143
    },
    {
      "epoch": 1.51,
      "grad_norm": 1.0209273099899292,
      "learning_rate": 0.0002961447512726762,
      "loss": 1.9787,
      "step": 116144
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.907547116279602,
      "learning_rate": 0.00029614066104864233,
      "loss": 2.1733,
      "step": 116145
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.9929010272026062,
      "learning_rate": 0.0002961365708253261,
      "loss": 1.983,
      "step": 116146
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.9555352330207825,
      "learning_rate": 0.000296132480602728,
      "loss": 2.1542,
      "step": 116147
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.8705364465713501,
      "learning_rate": 0.00029612839038084895,
      "loss": 2.0529,
      "step": 116148
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.8983756303787231,
      "learning_rate": 0.00029612430015968973,
      "loss": 2.0987,
      "step": 116149
    },
    {
      "epoch": 1.51,
      "grad_norm": 1.0042672157287598,
      "learning_rate": 0.0002961202099392511,
      "loss": 2.1248,
      "step": 116150
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.9953850507736206,
      "learning_rate": 0.00029611611971953373,
      "loss": 1.9254,
      "step": 116151
    },
    {
      "epoch": 1.51,
      "grad_norm": 1.2211508750915527,
      "learning_rate": 0.0002961120295005386,
      "loss": 1.9774,
      "step": 116152
    },
    {
      "epoch": 1.51,
      "grad_norm": 1.0881925821304321,
      "learning_rate": 0.00029610793928226616,
      "loss": 2.2093,
      "step": 116153
    },
    {
      "epoch": 1.51,
      "grad_norm": 1.073668360710144,
      "learning_rate": 0.0002961038490647174,
      "loss": 2.0115,
      "step": 116154
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.9378142356872559,
      "learning_rate": 0.00029609975884789297,
      "loss": 2.0664,
      "step": 116155
    },
    {
      "epoch": 1.51,
      "grad_norm": 1.1555198431015015,
      "learning_rate": 0.00029609566863179367,
      "loss": 2.2753,
      "step": 116156
    },
    {
      "epoch": 1.51,
      "grad_norm": 1.3842910528182983,
      "learning_rate": 0.00029609157841642024,
      "loss": 2.2745,
      "step": 116157
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.9002835154533386,
      "learning_rate": 0.00029608748820177355,
      "loss": 2.2015,
      "step": 116158
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.9957194924354553,
      "learning_rate": 0.00029608339798785425,
      "loss": 2.2466,
      "step": 116159
    },
    {
      "epoch": 1.51,
      "grad_norm": 1.2425060272216797,
      "learning_rate": 0.00029607930777466305,
      "loss": 1.8324,
      "step": 116160
    },
    {
      "epoch": 1.51,
      "grad_norm": 1.0308144092559814,
      "learning_rate": 0.0002960752175622008,
      "loss": 2.2025,
      "step": 116161
    },
    {
      "epoch": 1.51,
      "grad_norm": 1.0540236234664917,
      "learning_rate": 0.00029607112735046824,
      "loss": 2.0934,
      "step": 116162
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.9953891634941101,
      "learning_rate": 0.0002960670371394661,
      "loss": 2.3547,
      "step": 116163
    },
    {
      "epoch": 1.51,
      "grad_norm": 1.1363694667816162,
      "learning_rate": 0.0002960629469291953,
      "loss": 2.0661,
      "step": 116164
    },
    {
      "epoch": 1.51,
      "grad_norm": 1.047702431678772,
      "learning_rate": 0.0002960588567196564,
      "loss": 2.3146,
      "step": 116165
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.9594166874885559,
      "learning_rate": 0.00029605476651085026,
      "loss": 2.0102,
      "step": 116166
    },
    {
      "epoch": 1.51,
      "grad_norm": 1.284717082977295,
      "learning_rate": 0.0002960506763027776,
      "loss": 2.0809,
      "step": 116167
    },
    {
      "epoch": 1.51,
      "grad_norm": 1.1478899717330933,
      "learning_rate": 0.0002960465860954391,
      "loss": 1.9665,
      "step": 116168
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.9366565346717834,
      "learning_rate": 0.00029604249588883563,
      "loss": 2.1143,
      "step": 116169
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.9347547292709351,
      "learning_rate": 0.000296038405682968,
      "loss": 2.3125,
      "step": 116170
    },
    {
      "epoch": 1.51,
      "grad_norm": 1.3245006799697876,
      "learning_rate": 0.000296034315477837,
      "loss": 2.1588,
      "step": 116171
    },
    {
      "epoch": 1.51,
      "grad_norm": 1.0473991632461548,
      "learning_rate": 0.0002960302252734431,
      "loss": 2.2624,
      "step": 116172
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.9722909331321716,
      "learning_rate": 0.0002960261350697874,
      "loss": 2.1444,
      "step": 116173
    },
    {
      "epoch": 1.51,
      "grad_norm": 1.1962480545043945,
      "learning_rate": 0.0002960220448668704,
      "loss": 2.1615,
      "step": 116174
    },
    {
      "epoch": 1.51,
      "grad_norm": 1.0908993482589722,
      "learning_rate": 0.0002960179546646931,
      "loss": 2.2364,
      "step": 116175
    },
    {
      "epoch": 1.51,
      "grad_norm": 1.0050252676010132,
      "learning_rate": 0.00029601386446325605,
      "loss": 2.0238,
      "step": 116176
    },
    {
      "epoch": 1.51,
      "grad_norm": 1.0805656909942627,
      "learning_rate": 0.00029600977426256017,
      "loss": 2.1151,
      "step": 116177
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.9893483519554138,
      "learning_rate": 0.00029600568406260605,
      "loss": 1.9791,
      "step": 116178
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.9928150177001953,
      "learning_rate": 0.0002960015938633946,
      "loss": 2.2398,
      "step": 116179
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.940565288066864,
      "learning_rate": 0.0002959975036649266,
      "loss": 2.0146,
      "step": 116180
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.9633646011352539,
      "learning_rate": 0.0002959934134672026,
      "loss": 2.2842,
      "step": 116181
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.9023289084434509,
      "learning_rate": 0.0002959893232702236,
      "loss": 2.1949,
      "step": 116182
    },
    {
      "epoch": 1.51,
      "grad_norm": 1.1581157445907593,
      "learning_rate": 0.0002959852330739903,
      "loss": 2.1247,
      "step": 116183
    },
    {
      "epoch": 1.51,
      "grad_norm": 1.1098010540008545,
      "learning_rate": 0.0002959811428785033,
      "loss": 2.3227,
      "step": 116184
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.9158878326416016,
      "learning_rate": 0.0002959770526837635,
      "loss": 2.1597,
      "step": 116185
    },
    {
      "epoch": 1.51,
      "grad_norm": 1.152843713760376,
      "learning_rate": 0.00029597296248977167,
      "loss": 2.1778,
      "step": 116186
    },
    {
      "epoch": 1.51,
      "grad_norm": 1.0124458074569702,
      "learning_rate": 0.00029596887229652854,
      "loss": 2.0417,
      "step": 116187
    },
    {
      "epoch": 1.51,
      "grad_norm": 1.0896245241165161,
      "learning_rate": 0.0002959647821040349,
      "loss": 2.0439,
      "step": 116188
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.8567814826965332,
      "learning_rate": 0.0002959606919122916,
      "loss": 2.1333,
      "step": 116189
    },
    {
      "epoch": 1.51,
      "grad_norm": 1.147332787513733,
      "learning_rate": 0.0002959566017212991,
      "loss": 2.1383,
      "step": 116190
    },
    {
      "epoch": 1.51,
      "grad_norm": 1.3205634355545044,
      "learning_rate": 0.00029595251153105833,
      "loss": 2.0754,
      "step": 116191
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.8335767388343811,
      "learning_rate": 0.0002959484213415701,
      "loss": 2.1968,
      "step": 116192
    },
    {
      "epoch": 1.51,
      "grad_norm": 1.0414482355117798,
      "learning_rate": 0.00029594433115283516,
      "loss": 2.0752,
      "step": 116193
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.9141960144042969,
      "learning_rate": 0.0002959402409648542,
      "loss": 2.2362,
      "step": 116194
    },
    {
      "epoch": 1.51,
      "grad_norm": 1.1013792753219604,
      "learning_rate": 0.00029593615077762824,
      "loss": 2.3275,
      "step": 116195
    },
    {
      "epoch": 1.51,
      "grad_norm": 1.05303156375885,
      "learning_rate": 0.00029593206059115763,
      "loss": 2.0823,
      "step": 116196
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.866389811038971,
      "learning_rate": 0.0002959279704054433,
      "loss": 2.3493,
      "step": 116197
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.8630948066711426,
      "learning_rate": 0.0002959238802204861,
      "loss": 2.1375,
      "step": 116198
    },
    {
      "epoch": 1.51,
      "grad_norm": 1.0177521705627441,
      "learning_rate": 0.0002959197900362867,
      "loss": 2.3601,
      "step": 116199
    },
    {
      "epoch": 1.51,
      "grad_norm": 1.162585973739624,
      "learning_rate": 0.0002959156998528459,
      "loss": 2.0644,
      "step": 116200
    },
    {
      "epoch": 1.51,
      "grad_norm": 1.08696711063385,
      "learning_rate": 0.0002959116096701646,
      "loss": 2.3836,
      "step": 116201
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.9097192287445068,
      "learning_rate": 0.0002959075194882432,
      "loss": 2.2728,
      "step": 116202
    },
    {
      "epoch": 1.51,
      "grad_norm": 1.0736050605773926,
      "learning_rate": 0.00029590342930708275,
      "loss": 2.1817,
      "step": 116203
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.9210059642791748,
      "learning_rate": 0.0002958993391266839,
      "loss": 1.9307,
      "step": 116204
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.8424869775772095,
      "learning_rate": 0.0002958952489470474,
      "loss": 2.0408,
      "step": 116205
    },
    {
      "epoch": 1.51,
      "grad_norm": 1.0412312746047974,
      "learning_rate": 0.0002958911587681741,
      "loss": 2.1978,
      "step": 116206
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.9241740107536316,
      "learning_rate": 0.00029588706859006483,
      "loss": 2.0787,
      "step": 116207
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.902141809463501,
      "learning_rate": 0.0002958829784127201,
      "loss": 1.9525,
      "step": 116208
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.884023129940033,
      "learning_rate": 0.0002958788882361408,
      "loss": 2.2742,
      "step": 116209
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.918852686882019,
      "learning_rate": 0.00029587479806032774,
      "loss": 1.999,
      "step": 116210
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.9514351487159729,
      "learning_rate": 0.00029587070788528153,
      "loss": 2.1587,
      "step": 116211
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.8735171556472778,
      "learning_rate": 0.0002958666177110031,
      "loss": 2.2337,
      "step": 116212
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.9397830367088318,
      "learning_rate": 0.00029586252753749327,
      "loss": 2.0736,
      "step": 116213
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.9402210712432861,
      "learning_rate": 0.0002958584373647525,
      "loss": 2.2759,
      "step": 116214
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.9623290300369263,
      "learning_rate": 0.0002958543471927818,
      "loss": 2.2393,
      "step": 116215
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.9429212212562561,
      "learning_rate": 0.0002958502570215818,
      "loss": 2.1355,
      "step": 116216
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.8892624378204346,
      "learning_rate": 0.0002958461668511534,
      "loss": 2.206,
      "step": 116217
    },
    {
      "epoch": 1.51,
      "grad_norm": 1.0585486888885498,
      "learning_rate": 0.0002958420766814972,
      "loss": 2.3041,
      "step": 116218
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.8848546743392944,
      "learning_rate": 0.00029583798651261413,
      "loss": 2.0116,
      "step": 116219
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.8587571978569031,
      "learning_rate": 0.0002958338963445048,
      "loss": 2.1759,
      "step": 116220
    },
    {
      "epoch": 1.51,
      "grad_norm": 1.069689154624939,
      "learning_rate": 0.00029582980617717,
      "loss": 2.3334,
      "step": 116221
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.9819274544715881,
      "learning_rate": 0.0002958257160106105,
      "loss": 2.1345,
      "step": 116222
    },
    {
      "epoch": 1.51,
      "grad_norm": 1.0701326131820679,
      "learning_rate": 0.0002958216258448271,
      "loss": 2.2122,
      "step": 116223
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.9305329322814941,
      "learning_rate": 0.0002958175356798206,
      "loss": 2.166,
      "step": 116224
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.9620425701141357,
      "learning_rate": 0.0002958134455155917,
      "loss": 2.1416,
      "step": 116225
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.8753702044487,
      "learning_rate": 0.0002958093553521411,
      "loss": 2.1888,
      "step": 116226
    },
    {
      "epoch": 1.51,
      "grad_norm": 1.022571086883545,
      "learning_rate": 0.0002958052651894696,
      "loss": 2.2479,
      "step": 116227
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.9501085877418518,
      "learning_rate": 0.000295801175027578,
      "loss": 1.9338,
      "step": 116228
    },
    {
      "epoch": 1.51,
      "grad_norm": 1.075978398323059,
      "learning_rate": 0.0002957970848664671,
      "loss": 2.0909,
      "step": 116229
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.8843246102333069,
      "learning_rate": 0.0002957929947061375,
      "loss": 2.1661,
      "step": 116230
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.8626971244812012,
      "learning_rate": 0.0002957889045465902,
      "loss": 2.1814,
      "step": 116231
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.9714354872703552,
      "learning_rate": 0.00029578481438782573,
      "loss": 2.0968,
      "step": 116232
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.8581069111824036,
      "learning_rate": 0.0002957807242298449,
      "loss": 1.9985,
      "step": 116233
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.8418748378753662,
      "learning_rate": 0.0002957766340726486,
      "loss": 2.004,
      "step": 116234
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.9412686228752136,
      "learning_rate": 0.00029577254391623744,
      "loss": 2.2385,
      "step": 116235
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.9569993019104004,
      "learning_rate": 0.0002957684537606122,
      "loss": 2.2348,
      "step": 116236
    },
    {
      "epoch": 1.51,
      "grad_norm": 1.027530550956726,
      "learning_rate": 0.0002957643636057738,
      "loss": 2.0887,
      "step": 116237
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.9670574069023132,
      "learning_rate": 0.00029576027345172296,
      "loss": 2.244,
      "step": 116238
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.9349450469017029,
      "learning_rate": 0.0002957561832984602,
      "loss": 2.3918,
      "step": 116239
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.9441176056861877,
      "learning_rate": 0.00029575209314598646,
      "loss": 2.0366,
      "step": 116240
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.9851273894309998,
      "learning_rate": 0.00029574800299430253,
      "loss": 2.2518,
      "step": 116241
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.9555061459541321,
      "learning_rate": 0.0002957439128434091,
      "loss": 2.1278,
      "step": 116242
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.9367818832397461,
      "learning_rate": 0.000295739822693307,
      "loss": 2.1861,
      "step": 116243
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.9023168087005615,
      "learning_rate": 0.00029573573254399697,
      "loss": 2.1563,
      "step": 116244
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.9430152773857117,
      "learning_rate": 0.00029573164239547967,
      "loss": 2.121,
      "step": 116245
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.9373946189880371,
      "learning_rate": 0.00029572755224775596,
      "loss": 2.0015,
      "step": 116246
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.926045835018158,
      "learning_rate": 0.0002957234621008266,
      "loss": 2.1518,
      "step": 116247
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.8393352031707764,
      "learning_rate": 0.00029571937195469226,
      "loss": 2.0336,
      "step": 116248
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.8370123505592346,
      "learning_rate": 0.0002957152818093538,
      "loss": 2.4003,
      "step": 116249
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.8407368659973145,
      "learning_rate": 0.0002957111916648121,
      "loss": 2.0346,
      "step": 116250
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.8722707033157349,
      "learning_rate": 0.00029570710152106763,
      "loss": 2.075,
      "step": 116251
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.8912505507469177,
      "learning_rate": 0.0002957030113781213,
      "loss": 2.1484,
      "step": 116252
    },
    {
      "epoch": 1.51,
      "grad_norm": 1.0020321607589722,
      "learning_rate": 0.0002956989212359738,
      "loss": 2.3313,
      "step": 116253
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.8708922266960144,
      "learning_rate": 0.00029569483109462605,
      "loss": 2.1418,
      "step": 116254
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.9865026473999023,
      "learning_rate": 0.0002956907409540786,
      "loss": 2.1664,
      "step": 116255
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.8927789330482483,
      "learning_rate": 0.00029568665081433247,
      "loss": 2.1016,
      "step": 116256
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.9944222569465637,
      "learning_rate": 0.0002956825606753882,
      "loss": 2.0322,
      "step": 116257
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.9280248284339905,
      "learning_rate": 0.00029567847053724666,
      "loss": 1.9246,
      "step": 116258
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.8823007941246033,
      "learning_rate": 0.0002956743803999085,
      "loss": 2.2463,
      "step": 116259
    },
    {
      "epoch": 1.51,
      "grad_norm": 1.045831322669983,
      "learning_rate": 0.00029567029026337464,
      "loss": 2.3209,
      "step": 116260
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.9182262420654297,
      "learning_rate": 0.00029566620012764567,
      "loss": 1.8486,
      "step": 116261
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.9573736190795898,
      "learning_rate": 0.0002956621099927225,
      "loss": 2.2083,
      "step": 116262
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.9794869422912598,
      "learning_rate": 0.0002956580198586058,
      "loss": 2.0932,
      "step": 116263
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.949827253818512,
      "learning_rate": 0.0002956539297252963,
      "loss": 2.264,
      "step": 116264
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.9782040119171143,
      "learning_rate": 0.00029564983959279496,
      "loss": 2.3691,
      "step": 116265
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.9567707180976868,
      "learning_rate": 0.0002956457494611023,
      "loss": 2.1133,
      "step": 116266
    },
    {
      "epoch": 1.51,
      "grad_norm": 1.072916865348816,
      "learning_rate": 0.00029564165933021914,
      "loss": 2.0101,
      "step": 116267
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.8888375163078308,
      "learning_rate": 0.0002956375692001464,
      "loss": 2.3628,
      "step": 116268
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.9213355183601379,
      "learning_rate": 0.0002956334790708846,
      "loss": 1.8555,
      "step": 116269
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.9683030247688293,
      "learning_rate": 0.0002956293889424347,
      "loss": 1.729,
      "step": 116270
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.9157093167304993,
      "learning_rate": 0.0002956252988147973,
      "loss": 2.0591,
      "step": 116271
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.9017696380615234,
      "learning_rate": 0.00029562120868797335,
      "loss": 2.2149,
      "step": 116272
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.9282835125923157,
      "learning_rate": 0.0002956171185619634,
      "loss": 2.2288,
      "step": 116273
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.9350789189338684,
      "learning_rate": 0.00029561302843676837,
      "loss": 2.1477,
      "step": 116274
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.9609336256980896,
      "learning_rate": 0.0002956089383123889,
      "loss": 2.2979,
      "step": 116275
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.9169219732284546,
      "learning_rate": 0.00029560484818882585,
      "loss": 2.1494,
      "step": 116276
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.9409576058387756,
      "learning_rate": 0.0002956007580660799,
      "loss": 2.3627,
      "step": 116277
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.9839555025100708,
      "learning_rate": 0.0002955966679441519,
      "loss": 2.1249,
      "step": 116278
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.9617730975151062,
      "learning_rate": 0.00029559257782304257,
      "loss": 2.1299,
      "step": 116279
    },
    {
      "epoch": 1.51,
      "grad_norm": 1.1091798543930054,
      "learning_rate": 0.0002955884877027527,
      "loss": 2.121,
      "step": 116280
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.902883768081665,
      "learning_rate": 0.0002955843975832829,
      "loss": 2.0709,
      "step": 116281
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.9358676075935364,
      "learning_rate": 0.0002955803074646341,
      "loss": 2.1483,
      "step": 116282
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.9805123209953308,
      "learning_rate": 0.000295576217346807,
      "loss": 2.2683,
      "step": 116283
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.9160963296890259,
      "learning_rate": 0.00029557212722980235,
      "loss": 2.1194,
      "step": 116284
    },
    {
      "epoch": 1.51,
      "grad_norm": 1.1983205080032349,
      "learning_rate": 0.00029556803711362095,
      "loss": 2.3832,
      "step": 116285
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.9507530927658081,
      "learning_rate": 0.00029556394699826365,
      "loss": 2.1172,
      "step": 116286
    },
    {
      "epoch": 1.51,
      "grad_norm": 1.0228652954101562,
      "learning_rate": 0.00029555985688373094,
      "loss": 2.2021,
      "step": 116287
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.9545009732246399,
      "learning_rate": 0.00029555576677002375,
      "loss": 2.0354,
      "step": 116288
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.9430792927742004,
      "learning_rate": 0.0002955516766571428,
      "loss": 2.1901,
      "step": 116289
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.9146637320518494,
      "learning_rate": 0.00029554758654508893,
      "loss": 2.1448,
      "step": 116290
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.9477813839912415,
      "learning_rate": 0.0002955434964338628,
      "loss": 2.0584,
      "step": 116291
    },
    {
      "epoch": 1.51,
      "grad_norm": 1.0688279867172241,
      "learning_rate": 0.0002955394063234654,
      "loss": 2.0425,
      "step": 116292
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.830181360244751,
      "learning_rate": 0.00029553531621389717,
      "loss": 1.8982,
      "step": 116293
    },
    {
      "epoch": 1.51,
      "grad_norm": 1.1507360935211182,
      "learning_rate": 0.000295531226105159,
      "loss": 2.1293,
      "step": 116294
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.8314704298973083,
      "learning_rate": 0.0002955271359972516,
      "loss": 2.1516,
      "step": 116295
    },
    {
      "epoch": 1.51,
      "grad_norm": 1.0548083782196045,
      "learning_rate": 0.0002955230458901759,
      "loss": 2.0045,
      "step": 116296
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.9712043404579163,
      "learning_rate": 0.00029551895578393245,
      "loss": 2.1322,
      "step": 116297
    },
    {
      "epoch": 1.51,
      "grad_norm": 1.1625556945800781,
      "learning_rate": 0.0002955148656785223,
      "loss": 2.1183,
      "step": 116298
    },
    {
      "epoch": 1.51,
      "grad_norm": 1.0677931308746338,
      "learning_rate": 0.00029551077557394585,
      "loss": 2.3591,
      "step": 116299
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.9271562695503235,
      "learning_rate": 0.00029550668547020405,
      "loss": 2.2935,
      "step": 116300
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.9632810950279236,
      "learning_rate": 0.0002955025953672977,
      "loss": 2.0381,
      "step": 116301
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.9522063732147217,
      "learning_rate": 0.00029549850526522744,
      "loss": 2.0465,
      "step": 116302
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.8648367524147034,
      "learning_rate": 0.0002954944151639941,
      "loss": 2.0171,
      "step": 116303
    },
    {
      "epoch": 1.51,
      "grad_norm": 1.5910005569458008,
      "learning_rate": 0.0002954903250635984,
      "loss": 2.056,
      "step": 116304
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.9161781668663025,
      "learning_rate": 0.00029548623496404136,
      "loss": 2.1109,
      "step": 116305
    },
    {
      "epoch": 1.51,
      "grad_norm": 1.196755290031433,
      "learning_rate": 0.0002954821448653233,
      "loss": 2.1167,
      "step": 116306
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.870349645614624,
      "learning_rate": 0.0002954780547674452,
      "loss": 2.1277,
      "step": 116307
    },
    {
      "epoch": 1.51,
      "grad_norm": 1.2815197706222534,
      "learning_rate": 0.0002954739646704078,
      "loss": 2.0812,
      "step": 116308
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.8573604822158813,
      "learning_rate": 0.00029546987457421196,
      "loss": 2.1784,
      "step": 116309
    },
    {
      "epoch": 1.51,
      "grad_norm": 1.0334868431091309,
      "learning_rate": 0.0002954657844788583,
      "loss": 2.0149,
      "step": 116310
    },
    {
      "epoch": 1.51,
      "grad_norm": 1.1526280641555786,
      "learning_rate": 0.0002954616943843478,
      "loss": 1.9899,
      "step": 116311
    },
    {
      "epoch": 1.51,
      "grad_norm": 1.170680046081543,
      "learning_rate": 0.00029545760429068086,
      "loss": 2.1498,
      "step": 116312
    },
    {
      "epoch": 1.51,
      "grad_norm": 1.0445374250411987,
      "learning_rate": 0.0002954535141978585,
      "loss": 2.2225,
      "step": 116313
    },
    {
      "epoch": 1.51,
      "grad_norm": 1.0855454206466675,
      "learning_rate": 0.00029544942410588136,
      "loss": 2.0178,
      "step": 116314
    },
    {
      "epoch": 1.51,
      "grad_norm": 1.2455335855484009,
      "learning_rate": 0.0002954453340147503,
      "loss": 2.1479,
      "step": 116315
    },
    {
      "epoch": 1.51,
      "grad_norm": 1.033944845199585,
      "learning_rate": 0.000295441243924466,
      "loss": 2.3054,
      "step": 116316
    },
    {
      "epoch": 1.51,
      "grad_norm": 1.037286400794983,
      "learning_rate": 0.0002954371538350294,
      "loss": 2.1972,
      "step": 116317
    },
    {
      "epoch": 1.51,
      "grad_norm": 1.0035840272903442,
      "learning_rate": 0.00029543306374644107,
      "loss": 2.109,
      "step": 116318
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.9953452944755554,
      "learning_rate": 0.0002954289736587017,
      "loss": 2.0921,
      "step": 116319
    },
    {
      "epoch": 1.51,
      "grad_norm": 1.054764747619629,
      "learning_rate": 0.00029542488357181226,
      "loss": 1.9464,
      "step": 116320
    },
    {
      "epoch": 1.51,
      "grad_norm": 1.6197105646133423,
      "learning_rate": 0.0002954207934857734,
      "loss": 2.5647,
      "step": 116321
    },
    {
      "epoch": 1.51,
      "grad_norm": 1.143859624862671,
      "learning_rate": 0.0002954167034005859,
      "loss": 2.0296,
      "step": 116322
    },
    {
      "epoch": 1.51,
      "grad_norm": 1.0655268430709839,
      "learning_rate": 0.0002954126133162506,
      "loss": 2.0723,
      "step": 116323
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.9813793301582336,
      "learning_rate": 0.00029540852323276814,
      "loss": 2.2394,
      "step": 116324
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.9961271286010742,
      "learning_rate": 0.0002954044331501392,
      "loss": 1.9129,
      "step": 116325
    },
    {
      "epoch": 1.51,
      "grad_norm": 1.1618272066116333,
      "learning_rate": 0.00029540034306836476,
      "loss": 1.8987,
      "step": 116326
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.890788733959198,
      "learning_rate": 0.00029539625298744546,
      "loss": 1.9645,
      "step": 116327
    },
    {
      "epoch": 1.51,
      "grad_norm": 1.0295275449752808,
      "learning_rate": 0.0002953921629073821,
      "loss": 2.0523,
      "step": 116328
    },
    {
      "epoch": 1.51,
      "grad_norm": 1.127583384513855,
      "learning_rate": 0.0002953880728281755,
      "loss": 1.9996,
      "step": 116329
    },
    {
      "epoch": 1.51,
      "grad_norm": 1.0064626932144165,
      "learning_rate": 0.00029538398274982626,
      "loss": 2.4172,
      "step": 116330
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.9080792665481567,
      "learning_rate": 0.0002953798926723352,
      "loss": 2.135,
      "step": 116331
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.9623849987983704,
      "learning_rate": 0.0002953758025957031,
      "loss": 2.2284,
      "step": 116332
    },
    {
      "epoch": 1.51,
      "grad_norm": 1.0000975131988525,
      "learning_rate": 0.0002953717125199308,
      "loss": 1.9271,
      "step": 116333
    },
    {
      "epoch": 1.51,
      "grad_norm": 1.0097514390945435,
      "learning_rate": 0.0002953676224450189,
      "loss": 2.4172,
      "step": 116334
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.918226420879364,
      "learning_rate": 0.0002953635323709685,
      "loss": 2.1904,
      "step": 116335
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.9132076501846313,
      "learning_rate": 0.0002953594422977798,
      "loss": 2.0854,
      "step": 116336
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.8598905801773071,
      "learning_rate": 0.000295355352225454,
      "loss": 2.1178,
      "step": 116337
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.9112988710403442,
      "learning_rate": 0.0002953512621539916,
      "loss": 2.0817,
      "step": 116338
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.8768934607505798,
      "learning_rate": 0.00029534717208339365,
      "loss": 1.9533,
      "step": 116339
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.9428467154502869,
      "learning_rate": 0.0002953430820136606,
      "loss": 2.1889,
      "step": 116340
    },
    {
      "epoch": 1.51,
      "grad_norm": 1.0022088289260864,
      "learning_rate": 0.00029533899194479353,
      "loss": 2.2392,
      "step": 116341
    },
    {
      "epoch": 1.51,
      "grad_norm": 1.1114158630371094,
      "learning_rate": 0.000295334901876793,
      "loss": 1.8949,
      "step": 116342
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.92353755235672,
      "learning_rate": 0.0002953308118096597,
      "loss": 2.2455,
      "step": 116343
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.8732437491416931,
      "learning_rate": 0.0002953267217433945,
      "loss": 2.1791,
      "step": 116344
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.9547358751296997,
      "learning_rate": 0.0002953226316779982,
      "loss": 2.0458,
      "step": 116345
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.9407404065132141,
      "learning_rate": 0.0002953185416134715,
      "loss": 2.2458,
      "step": 116346
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.8911150693893433,
      "learning_rate": 0.0002953144515498152,
      "loss": 2.36,
      "step": 116347
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.8980788588523865,
      "learning_rate": 0.00029531036148702994,
      "loss": 2.0747,
      "step": 116348
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.9842336177825928,
      "learning_rate": 0.00029530627142511666,
      "loss": 1.9495,
      "step": 116349
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.9791662693023682,
      "learning_rate": 0.00029530218136407605,
      "loss": 2.1121,
      "step": 116350
    },
    {
      "epoch": 1.51,
      "grad_norm": 1.0104954242706299,
      "learning_rate": 0.00029529809130390876,
      "loss": 2.1598,
      "step": 116351
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.9567659497261047,
      "learning_rate": 0.00029529400124461567,
      "loss": 2.353,
      "step": 116352
    },
    {
      "epoch": 1.51,
      "grad_norm": 0.8900439143180847,
      "learning_rate": 0.0002952899111861976,
      "loss": 2.0909,
      "step": 116353
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.9559749364852905,
      "learning_rate": 0.00029528582112865515,
      "loss": 2.1879,
      "step": 116354
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.9839972257614136,
      "learning_rate": 0.00029528173107198913,
      "loss": 2.1522,
      "step": 116355
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.9804788827896118,
      "learning_rate": 0.00029527764101620034,
      "loss": 1.9657,
      "step": 116356
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.9852948188781738,
      "learning_rate": 0.0002952735509612896,
      "loss": 2.0818,
      "step": 116357
    },
    {
      "epoch": 1.52,
      "grad_norm": 1.0555801391601562,
      "learning_rate": 0.00029526946090725753,
      "loss": 2.2454,
      "step": 116358
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.9038208723068237,
      "learning_rate": 0.000295265370854105,
      "loss": 2.238,
      "step": 116359
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.9379340410232544,
      "learning_rate": 0.00029526128080183264,
      "loss": 2.017,
      "step": 116360
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.8918639421463013,
      "learning_rate": 0.00029525719075044137,
      "loss": 2.2371,
      "step": 116361
    },
    {
      "epoch": 1.52,
      "grad_norm": 1.1479402780532837,
      "learning_rate": 0.00029525310069993183,
      "loss": 2.3804,
      "step": 116362
    },
    {
      "epoch": 1.52,
      "grad_norm": 1.0085067749023438,
      "learning_rate": 0.00029524901065030483,
      "loss": 2.2345,
      "step": 116363
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.9966301918029785,
      "learning_rate": 0.0002952449206015612,
      "loss": 2.1698,
      "step": 116364
    },
    {
      "epoch": 1.52,
      "grad_norm": 1.027506947517395,
      "learning_rate": 0.00029524083055370166,
      "loss": 2.0351,
      "step": 116365
    },
    {
      "epoch": 1.52,
      "grad_norm": 1.0242819786071777,
      "learning_rate": 0.00029523674050672684,
      "loss": 2.0004,
      "step": 116366
    },
    {
      "epoch": 1.52,
      "grad_norm": 1.0125995874404907,
      "learning_rate": 0.0002952326504606376,
      "loss": 2.2363,
      "step": 116367
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.9051752686500549,
      "learning_rate": 0.0002952285604154347,
      "loss": 2.2579,
      "step": 116368
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.9879980683326721,
      "learning_rate": 0.0002952244703711189,
      "loss": 2.1961,
      "step": 116369
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.9502215385437012,
      "learning_rate": 0.00029522038032769095,
      "loss": 2.1407,
      "step": 116370
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.954938530921936,
      "learning_rate": 0.0002952162902851517,
      "loss": 2.1414,
      "step": 116371
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.8800554871559143,
      "learning_rate": 0.00029521220024350184,
      "loss": 2.1599,
      "step": 116372
    },
    {
      "epoch": 1.52,
      "grad_norm": 1.083093285560608,
      "learning_rate": 0.0002952081102027421,
      "loss": 2.185,
      "step": 116373
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.8337154388427734,
      "learning_rate": 0.0002952040201628732,
      "loss": 2.129,
      "step": 116374
    },
    {
      "epoch": 1.52,
      "grad_norm": 1.048307180404663,
      "learning_rate": 0.000295199930123896,
      "loss": 2.0856,
      "step": 116375
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.9538565278053284,
      "learning_rate": 0.0002951958400858112,
      "loss": 2.0987,
      "step": 116376
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.934607744216919,
      "learning_rate": 0.00029519175004861955,
      "loss": 1.9591,
      "step": 116377
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.9447647929191589,
      "learning_rate": 0.000295187660012322,
      "loss": 2.0846,
      "step": 116378
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.9461585879325867,
      "learning_rate": 0.00029518356997691904,
      "loss": 2.2414,
      "step": 116379
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.9430215358734131,
      "learning_rate": 0.00029517947994241157,
      "loss": 2.3043,
      "step": 116380
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.8980206251144409,
      "learning_rate": 0.0002951753899088003,
      "loss": 2.1475,
      "step": 116381
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.9157013297080994,
      "learning_rate": 0.000295171299876086,
      "loss": 2.1391,
      "step": 116382
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.931832492351532,
      "learning_rate": 0.00029516720984426944,
      "loss": 2.0912,
      "step": 116383
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.9611985087394714,
      "learning_rate": 0.00029516311981335155,
      "loss": 2.3139,
      "step": 116384
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.9863312244415283,
      "learning_rate": 0.0002951590297833328,
      "loss": 2.1592,
      "step": 116385
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.9505952596664429,
      "learning_rate": 0.00029515493975421406,
      "loss": 2.1369,
      "step": 116386
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.9774886965751648,
      "learning_rate": 0.0002951508497259961,
      "loss": 2.0546,
      "step": 116387
    },
    {
      "epoch": 1.52,
      "grad_norm": 1.0468297004699707,
      "learning_rate": 0.0002951467596986797,
      "loss": 2.1533,
      "step": 116388
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.9196609854698181,
      "learning_rate": 0.0002951426696722656,
      "loss": 2.2277,
      "step": 116389
    },
    {
      "epoch": 1.52,
      "grad_norm": 1.0101463794708252,
      "learning_rate": 0.0002951385796467547,
      "loss": 1.8505,
      "step": 116390
    },
    {
      "epoch": 1.52,
      "grad_norm": 1.01179838180542,
      "learning_rate": 0.0002951344896221475,
      "loss": 2.1005,
      "step": 116391
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.9543001055717468,
      "learning_rate": 0.0002951303995984449,
      "loss": 2.0536,
      "step": 116392
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.9547275900840759,
      "learning_rate": 0.00029512630957564766,
      "loss": 2.296,
      "step": 116393
    },
    {
      "epoch": 1.52,
      "grad_norm": 1.0916430950164795,
      "learning_rate": 0.0002951222195537565,
      "loss": 1.9984,
      "step": 116394
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.9577678442001343,
      "learning_rate": 0.00029511812953277223,
      "loss": 2.1187,
      "step": 116395
    },
    {
      "epoch": 1.52,
      "grad_norm": 1.0295802354812622,
      "learning_rate": 0.00029511403951269576,
      "loss": 2.5321,
      "step": 116396
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.8991451859474182,
      "learning_rate": 0.0002951099494935275,
      "loss": 2.2091,
      "step": 116397
    },
    {
      "epoch": 1.52,
      "grad_norm": 1.217096209526062,
      "learning_rate": 0.0002951058594752684,
      "loss": 2.1682,
      "step": 116398
    },
    {
      "epoch": 1.52,
      "grad_norm": 1.0185428857803345,
      "learning_rate": 0.00029510176945791923,
      "loss": 2.1238,
      "step": 116399
    },
    {
      "epoch": 1.52,
      "grad_norm": 1.1032487154006958,
      "learning_rate": 0.00029509767944148075,
      "loss": 2.045,
      "step": 116400
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.9082208871841431,
      "learning_rate": 0.00029509358942595364,
      "loss": 1.9967,
      "step": 116401
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.9401654005050659,
      "learning_rate": 0.0002950894994113389,
      "loss": 2.1689,
      "step": 116402
    },
    {
      "epoch": 1.52,
      "grad_norm": 1.0354315042495728,
      "learning_rate": 0.000295085409397637,
      "loss": 2.1411,
      "step": 116403
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.9798471331596375,
      "learning_rate": 0.00029508131938484874,
      "loss": 2.0018,
      "step": 116404
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.9581581354141235,
      "learning_rate": 0.000295077229372975,
      "loss": 2.2732,
      "step": 116405
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.892041027545929,
      "learning_rate": 0.00029507313936201653,
      "loss": 2.2472,
      "step": 116406
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.9281412363052368,
      "learning_rate": 0.0002950690493519741,
      "loss": 2.0812,
      "step": 116407
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.9056960940361023,
      "learning_rate": 0.0002950649593428484,
      "loss": 1.9832,
      "step": 116408
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.8769619464874268,
      "learning_rate": 0.0002950608693346402,
      "loss": 2.1542,
      "step": 116409
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.9291848540306091,
      "learning_rate": 0.00029505677932735024,
      "loss": 2.1594,
      "step": 116410
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.9235854148864746,
      "learning_rate": 0.00029505268932097933,
      "loss": 2.3144,
      "step": 116411
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.9303011298179626,
      "learning_rate": 0.00029504859931552826,
      "loss": 2.0003,
      "step": 116412
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.9386909008026123,
      "learning_rate": 0.0002950445093109977,
      "loss": 2.2639,
      "step": 116413
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.8834049105644226,
      "learning_rate": 0.00029504041930738856,
      "loss": 2.3092,
      "step": 116414
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.8997423052787781,
      "learning_rate": 0.00029503632930470144,
      "loss": 2.3129,
      "step": 116415
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.9858048558235168,
      "learning_rate": 0.0002950322393029371,
      "loss": 2.1963,
      "step": 116416
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.9507452249526978,
      "learning_rate": 0.00029502814930209637,
      "loss": 2.2756,
      "step": 116417
    },
    {
      "epoch": 1.52,
      "grad_norm": 1.2776187658309937,
      "learning_rate": 0.00029502405930218004,
      "loss": 2.0581,
      "step": 116418
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.9786542654037476,
      "learning_rate": 0.0002950199693031888,
      "loss": 2.4265,
      "step": 116419
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.9786693453788757,
      "learning_rate": 0.0002950158793051236,
      "loss": 2.269,
      "step": 116420
    },
    {
      "epoch": 1.52,
      "grad_norm": 1.018602967262268,
      "learning_rate": 0.0002950117893079849,
      "loss": 2.01,
      "step": 116421
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.9082192778587341,
      "learning_rate": 0.0002950076993117736,
      "loss": 2.1747,
      "step": 116422
    },
    {
      "epoch": 1.52,
      "grad_norm": 1.0833795070648193,
      "learning_rate": 0.00029500360931649045,
      "loss": 2.3915,
      "step": 116423
    },
    {
      "epoch": 1.52,
      "grad_norm": 1.0622336864471436,
      "learning_rate": 0.00029499951932213624,
      "loss": 2.3202,
      "step": 116424
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.9785394668579102,
      "learning_rate": 0.0002949954293287117,
      "loss": 1.9716,
      "step": 116425
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.8566762208938599,
      "learning_rate": 0.00029499133933621764,
      "loss": 1.9514,
      "step": 116426
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.864608883857727,
      "learning_rate": 0.0002949872493446548,
      "loss": 2.3131,
      "step": 116427
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.9792158007621765,
      "learning_rate": 0.0002949831593540239,
      "loss": 2.0707,
      "step": 116428
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.944500744342804,
      "learning_rate": 0.0002949790693643257,
      "loss": 2.0306,
      "step": 116429
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.93124920129776,
      "learning_rate": 0.000294974979375561,
      "loss": 2.1713,
      "step": 116430
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.9429383873939514,
      "learning_rate": 0.00029497088938773054,
      "loss": 2.1123,
      "step": 116431
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.9802877902984619,
      "learning_rate": 0.00029496679940083515,
      "loss": 2.2046,
      "step": 116432
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.9099082946777344,
      "learning_rate": 0.00029496270941487544,
      "loss": 2.1788,
      "step": 116433
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.8871739506721497,
      "learning_rate": 0.00029495861942985236,
      "loss": 2.1883,
      "step": 116434
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.9311222434043884,
      "learning_rate": 0.00029495452944576647,
      "loss": 2.206,
      "step": 116435
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.976468563079834,
      "learning_rate": 0.0002949504394626187,
      "loss": 2.1199,
      "step": 116436
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.8888314366340637,
      "learning_rate": 0.00029494634948040965,
      "loss": 2.0675,
      "step": 116437
    },
    {
      "epoch": 1.52,
      "grad_norm": 1.033881664276123,
      "learning_rate": 0.0002949422594991402,
      "loss": 1.9495,
      "step": 116438
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.9366906881332397,
      "learning_rate": 0.00029493816951881117,
      "loss": 2.3034,
      "step": 116439
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.9402944445610046,
      "learning_rate": 0.0002949340795394231,
      "loss": 2.0783,
      "step": 116440
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.8426600694656372,
      "learning_rate": 0.000294929989560977,
      "loss": 1.8775,
      "step": 116441
    },
    {
      "epoch": 1.52,
      "grad_norm": 1.3110088109970093,
      "learning_rate": 0.0002949258995834734,
      "loss": 2.2676,
      "step": 116442
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.9577515721321106,
      "learning_rate": 0.0002949218096069132,
      "loss": 2.1438,
      "step": 116443
    },
    {
      "epoch": 1.52,
      "grad_norm": 1.01332426071167,
      "learning_rate": 0.00029491771963129716,
      "loss": 2.3172,
      "step": 116444
    },
    {
      "epoch": 1.52,
      "grad_norm": 1.2257788181304932,
      "learning_rate": 0.00029491362965662605,
      "loss": 2.2942,
      "step": 116445
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.9750650525093079,
      "learning_rate": 0.0002949095396829005,
      "loss": 2.3855,
      "step": 116446
    },
    {
      "epoch": 1.52,
      "grad_norm": 1.3124992847442627,
      "learning_rate": 0.0002949054497101214,
      "loss": 2.0522,
      "step": 116447
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.909004807472229,
      "learning_rate": 0.0002949013597382895,
      "loss": 2.0539,
      "step": 116448
    },
    {
      "epoch": 1.52,
      "grad_norm": 1.1056327819824219,
      "learning_rate": 0.00029489726976740554,
      "loss": 2.3398,
      "step": 116449
    },
    {
      "epoch": 1.52,
      "grad_norm": 1.0124129056930542,
      "learning_rate": 0.0002948931797974702,
      "loss": 2.1435,
      "step": 116450
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.8566812872886658,
      "learning_rate": 0.00029488908982848444,
      "loss": 2.0034,
      "step": 116451
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.8585091233253479,
      "learning_rate": 0.0002948849998604488,
      "loss": 1.9622,
      "step": 116452
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.9308732151985168,
      "learning_rate": 0.00029488090989336413,
      "loss": 2.0161,
      "step": 116453
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.9471458792686462,
      "learning_rate": 0.0002948768199272312,
      "loss": 2.1049,
      "step": 116454
    },
    {
      "epoch": 1.52,
      "grad_norm": 1.0497939586639404,
      "learning_rate": 0.00029487272996205077,
      "loss": 2.2544,
      "step": 116455
    },
    {
      "epoch": 1.52,
      "grad_norm": 1.023798942565918,
      "learning_rate": 0.00029486863999782367,
      "loss": 2.2096,
      "step": 116456
    },
    {
      "epoch": 1.52,
      "grad_norm": 1.1008622646331787,
      "learning_rate": 0.00029486455003455055,
      "loss": 2.1242,
      "step": 116457
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.8930205702781677,
      "learning_rate": 0.0002948604600722322,
      "loss": 1.9736,
      "step": 116458
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.9335357546806335,
      "learning_rate": 0.00029485637011086933,
      "loss": 2.1216,
      "step": 116459
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.9597713947296143,
      "learning_rate": 0.0002948522801504628,
      "loss": 2.0609,
      "step": 116460
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.8972182869911194,
      "learning_rate": 0.0002948481901910133,
      "loss": 2.096,
      "step": 116461
    },
    {
      "epoch": 1.52,
      "grad_norm": 1.0022891759872437,
      "learning_rate": 0.0002948441002325216,
      "loss": 2.1669,
      "step": 116462
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.8936413526535034,
      "learning_rate": 0.0002948400102749887,
      "loss": 2.0017,
      "step": 116463
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.9697272777557373,
      "learning_rate": 0.00029483592031841487,
      "loss": 1.9712,
      "step": 116464
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.967265248298645,
      "learning_rate": 0.00029483183036280126,
      "loss": 2.2024,
      "step": 116465
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.8879724740982056,
      "learning_rate": 0.0002948277404081485,
      "loss": 2.2687,
      "step": 116466
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.9944114089012146,
      "learning_rate": 0.00029482365045445727,
      "loss": 2.1556,
      "step": 116467
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.8553118705749512,
      "learning_rate": 0.0002948195605017285,
      "loss": 2.2424,
      "step": 116468
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.8743905425071716,
      "learning_rate": 0.00029481547054996296,
      "loss": 2.1749,
      "step": 116469
    },
    {
      "epoch": 1.52,
      "grad_norm": 1.1057342290878296,
      "learning_rate": 0.00029481138059916126,
      "loss": 2.1889,
      "step": 116470
    },
    {
      "epoch": 1.52,
      "grad_norm": 1.0738869905471802,
      "learning_rate": 0.0002948072906493241,
      "loss": 2.1188,
      "step": 116471
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.9590252041816711,
      "learning_rate": 0.0002948032007004525,
      "loss": 1.9964,
      "step": 116472
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.9692983627319336,
      "learning_rate": 0.000294799110752547,
      "loss": 2.2101,
      "step": 116473
    },
    {
      "epoch": 1.52,
      "grad_norm": 1.0349836349487305,
      "learning_rate": 0.00029479502080560847,
      "loss": 2.2351,
      "step": 116474
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.9808042645454407,
      "learning_rate": 0.00029479093085963776,
      "loss": 2.0813,
      "step": 116475
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.9457654356956482,
      "learning_rate": 0.0002947868409146354,
      "loss": 1.9855,
      "step": 116476
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.8700748085975647,
      "learning_rate": 0.0002947827509706022,
      "loss": 1.8983,
      "step": 116477
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.9130555391311646,
      "learning_rate": 0.00029477866102753906,
      "loss": 2.0862,
      "step": 116478
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.9066744446754456,
      "learning_rate": 0.00029477457108544665,
      "loss": 2.1653,
      "step": 116479
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.9287306070327759,
      "learning_rate": 0.00029477048114432574,
      "loss": 2.2112,
      "step": 116480
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.8714936971664429,
      "learning_rate": 0.00029476639120417725,
      "loss": 1.8092,
      "step": 116481
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.8699048757553101,
      "learning_rate": 0.00029476230126500167,
      "loss": 2.0699,
      "step": 116482
    },
    {
      "epoch": 1.52,
      "grad_norm": 1.0040580034255981,
      "learning_rate": 0.0002947582113267998,
      "loss": 1.9044,
      "step": 116483
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.9975640177726746,
      "learning_rate": 0.00029475412138957254,
      "loss": 1.8093,
      "step": 116484
    },
    {
      "epoch": 1.52,
      "grad_norm": 1.0056428909301758,
      "learning_rate": 0.00029475003145332056,
      "loss": 2.2014,
      "step": 116485
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.9946718215942383,
      "learning_rate": 0.00029474594151804464,
      "loss": 2.1719,
      "step": 116486
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.9188105463981628,
      "learning_rate": 0.0002947418515837457,
      "loss": 2.0454,
      "step": 116487
    },
    {
      "epoch": 1.52,
      "grad_norm": 1.004375696182251,
      "learning_rate": 0.00029473776165042426,
      "loss": 2.0106,
      "step": 116488
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.9757348895072937,
      "learning_rate": 0.0002947336717180811,
      "loss": 2.1715,
      "step": 116489
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.8433603048324585,
      "learning_rate": 0.0002947295817867171,
      "loss": 2.1328,
      "step": 116490
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.9471949338912964,
      "learning_rate": 0.0002947254918563329,
      "loss": 2.2536,
      "step": 116491
    },
    {
      "epoch": 1.52,
      "grad_norm": 1.0413410663604736,
      "learning_rate": 0.0002947214019269294,
      "loss": 2.0624,
      "step": 116492
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.9364553093910217,
      "learning_rate": 0.0002947173119985074,
      "loss": 2.0535,
      "step": 116493
    },
    {
      "epoch": 1.52,
      "grad_norm": 1.007183313369751,
      "learning_rate": 0.0002947132220710674,
      "loss": 2.2179,
      "step": 116494
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.9324877858161926,
      "learning_rate": 0.00029470913214461034,
      "loss": 2.0484,
      "step": 116495
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.9344531893730164,
      "learning_rate": 0.00029470504221913696,
      "loss": 2.1953,
      "step": 116496
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.9523364305496216,
      "learning_rate": 0.000294700952294648,
      "loss": 2.2553,
      "step": 116497
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.9835970401763916,
      "learning_rate": 0.0002946968623711442,
      "loss": 2.113,
      "step": 116498
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.9094499945640564,
      "learning_rate": 0.0002946927724486265,
      "loss": 1.9959,
      "step": 116499
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.9248884916305542,
      "learning_rate": 0.00029468868252709537,
      "loss": 2.3858,
      "step": 116500
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.9989655613899231,
      "learning_rate": 0.00029468459260655175,
      "loss": 2.1414,
      "step": 116501
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.8883092999458313,
      "learning_rate": 0.0002946805026869964,
      "loss": 1.7892,
      "step": 116502
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.872072696685791,
      "learning_rate": 0.00029467641276842996,
      "loss": 2.1383,
      "step": 116503
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.8448024988174438,
      "learning_rate": 0.0002946723228508533,
      "loss": 2.0757,
      "step": 116504
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.8745735287666321,
      "learning_rate": 0.0002946682329342672,
      "loss": 2.0542,
      "step": 116505
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.8987883925437927,
      "learning_rate": 0.00029466414301867246,
      "loss": 2.1318,
      "step": 116506
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.9824973344802856,
      "learning_rate": 0.0002946600531040696,
      "loss": 2.2788,
      "step": 116507
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.9581096172332764,
      "learning_rate": 0.00029465596319045963,
      "loss": 2.1004,
      "step": 116508
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.7875379920005798,
      "learning_rate": 0.00029465187327784315,
      "loss": 1.9947,
      "step": 116509
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.9797435402870178,
      "learning_rate": 0.000294647783366221,
      "loss": 1.9349,
      "step": 116510
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.909945011138916,
      "learning_rate": 0.0002946436934555939,
      "loss": 2.2313,
      "step": 116511
    },
    {
      "epoch": 1.52,
      "grad_norm": 1.1534576416015625,
      "learning_rate": 0.00029463960354596285,
      "loss": 1.8369,
      "step": 116512
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.89739590883255,
      "learning_rate": 0.0002946355136373282,
      "loss": 2.1722,
      "step": 116513
    },
    {
      "epoch": 1.52,
      "grad_norm": 1.4452106952667236,
      "learning_rate": 0.00029463142372969093,
      "loss": 1.9597,
      "step": 116514
    },
    {
      "epoch": 1.52,
      "grad_norm": 1.033633828163147,
      "learning_rate": 0.00029462733382305177,
      "loss": 1.9807,
      "step": 116515
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.9957656860351562,
      "learning_rate": 0.00029462324391741154,
      "loss": 2.09,
      "step": 116516
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.8870390057563782,
      "learning_rate": 0.0002946191540127709,
      "loss": 1.9937,
      "step": 116517
    },
    {
      "epoch": 1.52,
      "grad_norm": 1.1405284404754639,
      "learning_rate": 0.0002946150641091307,
      "loss": 2.3581,
      "step": 116518
    },
    {
      "epoch": 1.52,
      "grad_norm": 1.170730710029602,
      "learning_rate": 0.00029461097420649174,
      "loss": 2.2582,
      "step": 116519
    },
    {
      "epoch": 1.52,
      "grad_norm": 1.1116185188293457,
      "learning_rate": 0.00029460688430485454,
      "loss": 2.3861,
      "step": 116520
    },
    {
      "epoch": 1.52,
      "grad_norm": 1.0282691717147827,
      "learning_rate": 0.00029460279440422007,
      "loss": 2.1253,
      "step": 116521
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.9732502102851868,
      "learning_rate": 0.0002945987045045891,
      "loss": 2.1293,
      "step": 116522
    },
    {
      "epoch": 1.52,
      "grad_norm": 1.0701748132705688,
      "learning_rate": 0.0002945946146059623,
      "loss": 2.1276,
      "step": 116523
    },
    {
      "epoch": 1.52,
      "grad_norm": 1.0455621480941772,
      "learning_rate": 0.0002945905247083404,
      "loss": 2.019,
      "step": 116524
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.9252549409866333,
      "learning_rate": 0.0002945864348117243,
      "loss": 2.0421,
      "step": 116525
    },
    {
      "epoch": 1.52,
      "grad_norm": 1.0842621326446533,
      "learning_rate": 0.0002945823449161147,
      "loss": 2.1659,
      "step": 116526
    },
    {
      "epoch": 1.52,
      "grad_norm": 1.2625632286071777,
      "learning_rate": 0.00029457825502151236,
      "loss": 2.1606,
      "step": 116527
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.9462625980377197,
      "learning_rate": 0.00029457416512791793,
      "loss": 2.0386,
      "step": 116528
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.9295789003372192,
      "learning_rate": 0.00029457007523533225,
      "loss": 2.2443,
      "step": 116529
    },
    {
      "epoch": 1.52,
      "grad_norm": 1.0173020362854004,
      "learning_rate": 0.00029456598534375624,
      "loss": 2.0829,
      "step": 116530
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.9179147481918335,
      "learning_rate": 0.0002945618954531904,
      "loss": 2.0746,
      "step": 116531
    },
    {
      "epoch": 1.52,
      "grad_norm": 1.005577564239502,
      "learning_rate": 0.00029455780556363563,
      "loss": 2.0384,
      "step": 116532
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.9986205697059631,
      "learning_rate": 0.0002945537156750927,
      "loss": 1.9989,
      "step": 116533
    },
    {
      "epoch": 1.52,
      "grad_norm": 1.083892822265625,
      "learning_rate": 0.00029454962578756236,
      "loss": 2.1087,
      "step": 116534
    },
    {
      "epoch": 1.52,
      "grad_norm": 1.022658348083496,
      "learning_rate": 0.0002945455359010452,
      "loss": 2.3279,
      "step": 116535
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.9473400115966797,
      "learning_rate": 0.0002945414460155423,
      "loss": 2.1222,
      "step": 116536
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.9219878315925598,
      "learning_rate": 0.00029453735613105416,
      "loss": 2.1058,
      "step": 116537
    },
    {
      "epoch": 1.52,
      "grad_norm": 1.164673924446106,
      "learning_rate": 0.00029453326624758163,
      "loss": 2.23,
      "step": 116538
    },
    {
      "epoch": 1.52,
      "grad_norm": 1.0768673419952393,
      "learning_rate": 0.0002945291763651255,
      "loss": 2.2813,
      "step": 116539
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.9654150009155273,
      "learning_rate": 0.00029452508648368644,
      "loss": 2.1053,
      "step": 116540
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.9029721021652222,
      "learning_rate": 0.00029452099660326535,
      "loss": 2.1183,
      "step": 116541
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.8832051753997803,
      "learning_rate": 0.00029451690672386297,
      "loss": 2.172,
      "step": 116542
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.9619386792182922,
      "learning_rate": 0.00029451281684547985,
      "loss": 2.1451,
      "step": 116543
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.8294638991355896,
      "learning_rate": 0.00029450872696811696,
      "loss": 2.1306,
      "step": 116544
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.9593990445137024,
      "learning_rate": 0.00029450463709177496,
      "loss": 2.1213,
      "step": 116545
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.928426206111908,
      "learning_rate": 0.00029450054721645466,
      "loss": 2.1676,
      "step": 116546
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.9197466969490051,
      "learning_rate": 0.0002944964573421568,
      "loss": 2.0059,
      "step": 116547
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.9155609011650085,
      "learning_rate": 0.00029449236746888235,
      "loss": 2.1728,
      "step": 116548
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.9936524033546448,
      "learning_rate": 0.0002944882775966317,
      "loss": 2.2052,
      "step": 116549
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.9497254490852356,
      "learning_rate": 0.00029448418772540577,
      "loss": 2.0762,
      "step": 116550
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.9042466878890991,
      "learning_rate": 0.00029448009785520534,
      "loss": 2.0764,
      "step": 116551
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.9377636313438416,
      "learning_rate": 0.00029447600798603117,
      "loss": 2.1953,
      "step": 116552
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.9277949333190918,
      "learning_rate": 0.000294471918117884,
      "loss": 2.2508,
      "step": 116553
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.8789529204368591,
      "learning_rate": 0.00029446782825076476,
      "loss": 2.0554,
      "step": 116554
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.9554896950721741,
      "learning_rate": 0.0002944637383846739,
      "loss": 2.0766,
      "step": 116555
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.8698830604553223,
      "learning_rate": 0.00029445964851961237,
      "loss": 2.0788,
      "step": 116556
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.847364604473114,
      "learning_rate": 0.0002944555586555809,
      "loss": 2.044,
      "step": 116557
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.901909589767456,
      "learning_rate": 0.00029445146879258016,
      "loss": 2.0329,
      "step": 116558
    },
    {
      "epoch": 1.52,
      "grad_norm": 1.03361976146698,
      "learning_rate": 0.0002944473789306111,
      "loss": 1.934,
      "step": 116559
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.9460379481315613,
      "learning_rate": 0.0002944432890696745,
      "loss": 1.9162,
      "step": 116560
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.944601833820343,
      "learning_rate": 0.00029443919920977077,
      "loss": 2.0924,
      "step": 116561
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.9154930114746094,
      "learning_rate": 0.000294435109350901,
      "loss": 2.0535,
      "step": 116562
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.911026120185852,
      "learning_rate": 0.0002944310194930658,
      "loss": 2.2716,
      "step": 116563
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.9197188019752502,
      "learning_rate": 0.00029442692963626595,
      "loss": 2.0592,
      "step": 116564
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.9005118012428284,
      "learning_rate": 0.0002944228397805023,
      "loss": 2.0418,
      "step": 116565
    },
    {
      "epoch": 1.52,
      "grad_norm": 1.0489565134048462,
      "learning_rate": 0.0002944187499257757,
      "loss": 2.3931,
      "step": 116566
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.9113638997077942,
      "learning_rate": 0.00029441466007208657,
      "loss": 2.0871,
      "step": 116567
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.8949902653694153,
      "learning_rate": 0.00029441057021943585,
      "loss": 2.0596,
      "step": 116568
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.9841355085372925,
      "learning_rate": 0.00029440648036782433,
      "loss": 2.1996,
      "step": 116569
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.9221300482749939,
      "learning_rate": 0.00029440239051725274,
      "loss": 1.9344,
      "step": 116570
    },
    {
      "epoch": 1.52,
      "grad_norm": 1.0992225408554077,
      "learning_rate": 0.00029439830066772186,
      "loss": 2.352,
      "step": 116571
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.9594563245773315,
      "learning_rate": 0.0002943942108192325,
      "loss": 2.1082,
      "step": 116572
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.9328842163085938,
      "learning_rate": 0.00029439012097178544,
      "loss": 2.321,
      "step": 116573
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.9775793552398682,
      "learning_rate": 0.0002943860311253812,
      "loss": 2.0743,
      "step": 116574
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.8979355692863464,
      "learning_rate": 0.00029438194128002075,
      "loss": 2.0162,
      "step": 116575
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.8808692693710327,
      "learning_rate": 0.0002943778514357048,
      "loss": 2.2787,
      "step": 116576
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.9075864553451538,
      "learning_rate": 0.0002943737615924341,
      "loss": 2.2152,
      "step": 116577
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.995468258857727,
      "learning_rate": 0.0002943696717502094,
      "loss": 2.0579,
      "step": 116578
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.9188350439071655,
      "learning_rate": 0.00029436558190903164,
      "loss": 2.0158,
      "step": 116579
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.9805880784988403,
      "learning_rate": 0.0002943614920689013,
      "loss": 2.1096,
      "step": 116580
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.9557709693908691,
      "learning_rate": 0.00029435740222981925,
      "loss": 2.2199,
      "step": 116581
    },
    {
      "epoch": 1.52,
      "grad_norm": 1.0050448179244995,
      "learning_rate": 0.0002943533123917863,
      "loss": 2.3026,
      "step": 116582
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.8418650031089783,
      "learning_rate": 0.00029434922255480316,
      "loss": 2.109,
      "step": 116583
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.8480228781700134,
      "learning_rate": 0.0002943451327188706,
      "loss": 1.9457,
      "step": 116584
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.8983745574951172,
      "learning_rate": 0.0002943410428839895,
      "loss": 2.077,
      "step": 116585
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.9753285646438599,
      "learning_rate": 0.0002943369530501604,
      "loss": 2.15,
      "step": 116586
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.9781830906867981,
      "learning_rate": 0.00029433286321738414,
      "loss": 2.0272,
      "step": 116587
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.9150546789169312,
      "learning_rate": 0.0002943287733856615,
      "loss": 2.1943,
      "step": 116588
    },
    {
      "epoch": 1.52,
      "grad_norm": 1.336527705192566,
      "learning_rate": 0.0002943246835549933,
      "loss": 2.0302,
      "step": 116589
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.9739948511123657,
      "learning_rate": 0.0002943205937253802,
      "loss": 2.0494,
      "step": 116590
    },
    {
      "epoch": 1.52,
      "grad_norm": 1.1297543048858643,
      "learning_rate": 0.0002943165038968232,
      "loss": 2.2223,
      "step": 116591
    },
    {
      "epoch": 1.52,
      "grad_norm": 1.0398149490356445,
      "learning_rate": 0.0002943124140693227,
      "loss": 2.1622,
      "step": 116592
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.9978946447372437,
      "learning_rate": 0.00029430832424287963,
      "loss": 2.2744,
      "step": 116593
    },
    {
      "epoch": 1.52,
      "grad_norm": 1.119641900062561,
      "learning_rate": 0.00029430423441749475,
      "loss": 2.0918,
      "step": 116594
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.9631927013397217,
      "learning_rate": 0.0002943001445931688,
      "loss": 2.2114,
      "step": 116595
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.849342405796051,
      "learning_rate": 0.0002942960547699026,
      "loss": 1.9785,
      "step": 116596
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.8925093412399292,
      "learning_rate": 0.000294291964947697,
      "loss": 2.0469,
      "step": 116597
    },
    {
      "epoch": 1.52,
      "grad_norm": 1.1062581539154053,
      "learning_rate": 0.00029428787512655245,
      "loss": 2.0926,
      "step": 116598
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.9421447515487671,
      "learning_rate": 0.0002942837853064699,
      "loss": 2.3469,
      "step": 116599
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.8983705043792725,
      "learning_rate": 0.0002942796954874502,
      "loss": 2.075,
      "step": 116600
    },
    {
      "epoch": 1.52,
      "grad_norm": 1.0119117498397827,
      "learning_rate": 0.00029427560566949387,
      "loss": 2.2853,
      "step": 116601
    },
    {
      "epoch": 1.52,
      "grad_norm": 1.0528709888458252,
      "learning_rate": 0.00029427151585260193,
      "loss": 1.9789,
      "step": 116602
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.8766975998878479,
      "learning_rate": 0.00029426742603677505,
      "loss": 2.0325,
      "step": 116603
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.9241924285888672,
      "learning_rate": 0.00029426333622201395,
      "loss": 2.2548,
      "step": 116604
    },
    {
      "epoch": 1.52,
      "grad_norm": 1.0117367506027222,
      "learning_rate": 0.00029425924640831937,
      "loss": 1.8401,
      "step": 116605
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.8850721120834351,
      "learning_rate": 0.00029425515659569203,
      "loss": 2.0365,
      "step": 116606
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.9387298822402954,
      "learning_rate": 0.00029425106678413284,
      "loss": 2.0722,
      "step": 116607
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.9110745191574097,
      "learning_rate": 0.00029424697697364246,
      "loss": 2.0189,
      "step": 116608
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.8747302293777466,
      "learning_rate": 0.00029424288716422176,
      "loss": 2.1043,
      "step": 116609
    },
    {
      "epoch": 1.52,
      "grad_norm": 1.01211416721344,
      "learning_rate": 0.00029423879735587127,
      "loss": 2.152,
      "step": 116610
    },
    {
      "epoch": 1.52,
      "grad_norm": 1.0172547101974487,
      "learning_rate": 0.0002942347075485921,
      "loss": 2.0029,
      "step": 116611
    },
    {
      "epoch": 1.52,
      "grad_norm": 1.0255705118179321,
      "learning_rate": 0.0002942306177423846,
      "loss": 2.1559,
      "step": 116612
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.9436785578727722,
      "learning_rate": 0.0002942265279372498,
      "loss": 2.0084,
      "step": 116613
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.9920169711112976,
      "learning_rate": 0.00029422243813318844,
      "loss": 2.1535,
      "step": 116614
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.9239118695259094,
      "learning_rate": 0.0002942183483302012,
      "loss": 2.3156,
      "step": 116615
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.9670642018318176,
      "learning_rate": 0.0002942142585282889,
      "loss": 2.2516,
      "step": 116616
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.9488317370414734,
      "learning_rate": 0.00029421016872745225,
      "loss": 2.2358,
      "step": 116617
    },
    {
      "epoch": 1.52,
      "grad_norm": 1.1506110429763794,
      "learning_rate": 0.0002942060789276921,
      "loss": 1.9067,
      "step": 116618
    },
    {
      "epoch": 1.52,
      "grad_norm": 1.055575966835022,
      "learning_rate": 0.0002942019891290091,
      "loss": 2.1235,
      "step": 116619
    },
    {
      "epoch": 1.52,
      "grad_norm": 1.0435103178024292,
      "learning_rate": 0.00029419789933140405,
      "loss": 2.2331,
      "step": 116620
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.9934868216514587,
      "learning_rate": 0.0002941938095348778,
      "loss": 2.1974,
      "step": 116621
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.9634209871292114,
      "learning_rate": 0.00029418971973943096,
      "loss": 2.057,
      "step": 116622
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.8929946422576904,
      "learning_rate": 0.00029418562994506435,
      "loss": 2.2118,
      "step": 116623
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.960858166217804,
      "learning_rate": 0.00029418154015177874,
      "loss": 2.1199,
      "step": 116624
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.92233806848526,
      "learning_rate": 0.00029417745035957495,
      "loss": 2.1735,
      "step": 116625
    },
    {
      "epoch": 1.52,
      "grad_norm": 1.2075203657150269,
      "learning_rate": 0.0002941733605684536,
      "loss": 2.0891,
      "step": 116626
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.9864917993545532,
      "learning_rate": 0.0002941692707784157,
      "loss": 2.1191,
      "step": 116627
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.9561529755592346,
      "learning_rate": 0.0002941651809894617,
      "loss": 2.2909,
      "step": 116628
    },
    {
      "epoch": 1.52,
      "grad_norm": 1.0037184953689575,
      "learning_rate": 0.0002941610912015925,
      "loss": 2.0429,
      "step": 116629
    },
    {
      "epoch": 1.52,
      "grad_norm": 1.177679181098938,
      "learning_rate": 0.00029415700141480883,
      "loss": 2.1828,
      "step": 116630
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.9592269659042358,
      "learning_rate": 0.00029415291162911155,
      "loss": 2.6171,
      "step": 116631
    },
    {
      "epoch": 1.52,
      "grad_norm": 1.1307634115219116,
      "learning_rate": 0.00029414882184450135,
      "loss": 2.0034,
      "step": 116632
    },
    {
      "epoch": 1.52,
      "grad_norm": 1.0518195629119873,
      "learning_rate": 0.00029414473206097904,
      "loss": 2.2486,
      "step": 116633
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.9023090600967407,
      "learning_rate": 0.0002941406422785453,
      "loss": 1.9151,
      "step": 116634
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.9682831168174744,
      "learning_rate": 0.0002941365524972009,
      "loss": 1.9346,
      "step": 116635
    },
    {
      "epoch": 1.52,
      "grad_norm": 1.098422884941101,
      "learning_rate": 0.0002941324627169466,
      "loss": 2.032,
      "step": 116636
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.885215699672699,
      "learning_rate": 0.0002941283729377832,
      "loss": 2.06,
      "step": 116637
    },
    {
      "epoch": 1.52,
      "grad_norm": 1.0110591650009155,
      "learning_rate": 0.00029412428315971144,
      "loss": 2.2533,
      "step": 116638
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.9116472601890564,
      "learning_rate": 0.00029412019338273207,
      "loss": 1.9411,
      "step": 116639
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.9217468500137329,
      "learning_rate": 0.00029411610360684606,
      "loss": 2.1525,
      "step": 116640
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.9830082058906555,
      "learning_rate": 0.0002941120138320538,
      "loss": 1.9637,
      "step": 116641
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.9853503108024597,
      "learning_rate": 0.00029410792405835625,
      "loss": 2.1258,
      "step": 116642
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.9293778538703918,
      "learning_rate": 0.00029410383428575413,
      "loss": 2.1378,
      "step": 116643
    },
    {
      "epoch": 1.52,
      "grad_norm": 1.0335091352462769,
      "learning_rate": 0.0002940997445142482,
      "loss": 2.1796,
      "step": 116644
    },
    {
      "epoch": 1.52,
      "grad_norm": 1.0344878435134888,
      "learning_rate": 0.00029409565474383925,
      "loss": 2.3229,
      "step": 116645
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.9946545362472534,
      "learning_rate": 0.00029409156497452817,
      "loss": 2.0986,
      "step": 116646
    },
    {
      "epoch": 1.52,
      "grad_norm": 1.0782169103622437,
      "learning_rate": 0.0002940874752063155,
      "loss": 2.1233,
      "step": 116647
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.875238835811615,
      "learning_rate": 0.000294083385439202,
      "loss": 2.1431,
      "step": 116648
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.943318784236908,
      "learning_rate": 0.00029407929567318856,
      "loss": 2.054,
      "step": 116649
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.89119553565979,
      "learning_rate": 0.00029407520590827587,
      "loss": 2.2193,
      "step": 116650
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.8878698348999023,
      "learning_rate": 0.00029407111614446473,
      "loss": 2.1711,
      "step": 116651
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.8894092440605164,
      "learning_rate": 0.000294067026381756,
      "loss": 2.4027,
      "step": 116652
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.9224276542663574,
      "learning_rate": 0.0002940629366201502,
      "loss": 2.2823,
      "step": 116653
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.9668756127357483,
      "learning_rate": 0.00029405884685964814,
      "loss": 2.1314,
      "step": 116654
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.9262579679489136,
      "learning_rate": 0.00029405475710025074,
      "loss": 2.093,
      "step": 116655
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.9048903584480286,
      "learning_rate": 0.0002940506673419587,
      "loss": 2.232,
      "step": 116656
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.9343749284744263,
      "learning_rate": 0.00029404657758477266,
      "loss": 2.2274,
      "step": 116657
    },
    {
      "epoch": 1.52,
      "grad_norm": 1.0139572620391846,
      "learning_rate": 0.0002940424878286936,
      "loss": 2.3214,
      "step": 116658
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.8743236660957336,
      "learning_rate": 0.0002940383980737221,
      "loss": 2.0451,
      "step": 116659
    },
    {
      "epoch": 1.52,
      "grad_norm": 1.0750682353973389,
      "learning_rate": 0.00029403430831985894,
      "loss": 2.3142,
      "step": 116660
    },
    {
      "epoch": 1.52,
      "grad_norm": 1.0026947259902954,
      "learning_rate": 0.00029403021856710493,
      "loss": 2.2998,
      "step": 116661
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.9414398670196533,
      "learning_rate": 0.00029402612881546076,
      "loss": 2.1601,
      "step": 116662
    },
    {
      "epoch": 1.52,
      "grad_norm": 1.027380347251892,
      "learning_rate": 0.0002940220390649273,
      "loss": 1.9985,
      "step": 116663
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.8887150883674622,
      "learning_rate": 0.00029401794931550537,
      "loss": 2.3399,
      "step": 116664
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.957045316696167,
      "learning_rate": 0.00029401385956719545,
      "loss": 2.0419,
      "step": 116665
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.902879536151886,
      "learning_rate": 0.0002940097698199985,
      "loss": 2.1514,
      "step": 116666
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.9412626624107361,
      "learning_rate": 0.0002940056800739153,
      "loss": 1.9499,
      "step": 116667
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.9473825693130493,
      "learning_rate": 0.0002940015903289465,
      "loss": 2.1284,
      "step": 116668
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.9836590886116028,
      "learning_rate": 0.0002939975005850929,
      "loss": 2.1068,
      "step": 116669
    },
    {
      "epoch": 1.52,
      "grad_norm": 1.2017358541488647,
      "learning_rate": 0.00029399341084235546,
      "loss": 2.1881,
      "step": 116670
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.9621903300285339,
      "learning_rate": 0.0002939893211007346,
      "loss": 2.0955,
      "step": 116671
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.9188297390937805,
      "learning_rate": 0.00029398523136023124,
      "loss": 2.2923,
      "step": 116672
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.8918945789337158,
      "learning_rate": 0.00029398114162084614,
      "loss": 2.0592,
      "step": 116673
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.8529883027076721,
      "learning_rate": 0.00029397705188258005,
      "loss": 2.0345,
      "step": 116674
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.9148908853530884,
      "learning_rate": 0.00029397296214543373,
      "loss": 2.1031,
      "step": 116675
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.9328867793083191,
      "learning_rate": 0.0002939688724094081,
      "loss": 2.1247,
      "step": 116676
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.961915910243988,
      "learning_rate": 0.0002939647826745036,
      "loss": 2.0745,
      "step": 116677
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.8506587743759155,
      "learning_rate": 0.00029396069294072125,
      "loss": 2.167,
      "step": 116678
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.944612979888916,
      "learning_rate": 0.0002939566032080617,
      "loss": 2.0269,
      "step": 116679
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.8650357723236084,
      "learning_rate": 0.00029395251347652567,
      "loss": 2.1753,
      "step": 116680
    },
    {
      "epoch": 1.52,
      "grad_norm": 1.0328730344772339,
      "learning_rate": 0.00029394842374611403,
      "loss": 2.3155,
      "step": 116681
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.8978984951972961,
      "learning_rate": 0.0002939443340168276,
      "loss": 2.1798,
      "step": 116682
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.9741012454032898,
      "learning_rate": 0.0002939402442886669,
      "loss": 2.0185,
      "step": 116683
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.8826867938041687,
      "learning_rate": 0.0002939361545616328,
      "loss": 2.1458,
      "step": 116684
    },
    {
      "epoch": 1.52,
      "grad_norm": 1.1049085855484009,
      "learning_rate": 0.00029393206483572615,
      "loss": 2.2223,
      "step": 116685
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.9951679110527039,
      "learning_rate": 0.0002939279751109476,
      "loss": 2.308,
      "step": 116686
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.920625627040863,
      "learning_rate": 0.00029392388538729795,
      "loss": 2.0729,
      "step": 116687
    },
    {
      "epoch": 1.52,
      "grad_norm": 1.0496786832809448,
      "learning_rate": 0.0002939197956647781,
      "loss": 2.1975,
      "step": 116688
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.9839980006217957,
      "learning_rate": 0.0002939157059433886,
      "loss": 1.9711,
      "step": 116689
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.8924230337142944,
      "learning_rate": 0.0002939116162231302,
      "loss": 2.1237,
      "step": 116690
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.9718971848487854,
      "learning_rate": 0.0002939075265040038,
      "loss": 2.1012,
      "step": 116691
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.9225198030471802,
      "learning_rate": 0.0002939034367860101,
      "loss": 2.3072,
      "step": 116692
    },
    {
      "epoch": 1.52,
      "grad_norm": 1.0458581447601318,
      "learning_rate": 0.00029389934706914987,
      "loss": 2.3607,
      "step": 116693
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.8914272785186768,
      "learning_rate": 0.0002938952573534239,
      "loss": 2.2795,
      "step": 116694
    },
    {
      "epoch": 1.52,
      "grad_norm": 1.0065008401870728,
      "learning_rate": 0.00029389116763883295,
      "loss": 1.868,
      "step": 116695
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.889060378074646,
      "learning_rate": 0.00029388707792537763,
      "loss": 2.0712,
      "step": 116696
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.9100605249404907,
      "learning_rate": 0.0002938829882130589,
      "loss": 2.0563,
      "step": 116697
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.8698939085006714,
      "learning_rate": 0.00029387889850187736,
      "loss": 2.1075,
      "step": 116698
    },
    {
      "epoch": 1.52,
      "grad_norm": 1.181450366973877,
      "learning_rate": 0.00029387480879183387,
      "loss": 2.2116,
      "step": 116699
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.9094968438148499,
      "learning_rate": 0.00029387071908292927,
      "loss": 2.1078,
      "step": 116700
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.9942687153816223,
      "learning_rate": 0.0002938666293751641,
      "loss": 2.2651,
      "step": 116701
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.9928027391433716,
      "learning_rate": 0.0002938625396685393,
      "loss": 2.1639,
      "step": 116702
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.8182100057601929,
      "learning_rate": 0.00029385844996305556,
      "loss": 2.0877,
      "step": 116703
    },
    {
      "epoch": 1.52,
      "grad_norm": 1.1975061893463135,
      "learning_rate": 0.0002938543602587136,
      "loss": 2.3129,
      "step": 116704
    },
    {
      "epoch": 1.52,
      "grad_norm": 1.1168581247329712,
      "learning_rate": 0.00029385027055551426,
      "loss": 2.2474,
      "step": 116705
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.9502615332603455,
      "learning_rate": 0.0002938461808534583,
      "loss": 1.9168,
      "step": 116706
    },
    {
      "epoch": 1.52,
      "grad_norm": 1.2733291387557983,
      "learning_rate": 0.0002938420911525465,
      "loss": 1.8901,
      "step": 116707
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.8864870667457581,
      "learning_rate": 0.00029383800145277945,
      "loss": 2.141,
      "step": 116708
    },
    {
      "epoch": 1.52,
      "grad_norm": 1.0004100799560547,
      "learning_rate": 0.0002938339117541581,
      "loss": 2.3287,
      "step": 116709
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.937224805355072,
      "learning_rate": 0.00029382982205668315,
      "loss": 2.4028,
      "step": 116710
    },
    {
      "epoch": 1.52,
      "grad_norm": 1.0731505155563354,
      "learning_rate": 0.00029382573236035536,
      "loss": 2.0977,
      "step": 116711
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.8923910856246948,
      "learning_rate": 0.0002938216426651754,
      "loss": 2.1141,
      "step": 116712
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.8835821747779846,
      "learning_rate": 0.0002938175529711442,
      "loss": 2.167,
      "step": 116713
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.9943547248840332,
      "learning_rate": 0.0002938134632782624,
      "loss": 2.1045,
      "step": 116714
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.9256042838096619,
      "learning_rate": 0.0002938093735865308,
      "loss": 2.065,
      "step": 116715
    },
    {
      "epoch": 1.52,
      "grad_norm": 1.0027589797973633,
      "learning_rate": 0.00029380528389595016,
      "loss": 2.2697,
      "step": 116716
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.9395873546600342,
      "learning_rate": 0.00029380119420652125,
      "loss": 2.1856,
      "step": 116717
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.8614701628684998,
      "learning_rate": 0.00029379710451824475,
      "loss": 1.9685,
      "step": 116718
    },
    {
      "epoch": 1.52,
      "grad_norm": 1.0721434354782104,
      "learning_rate": 0.0002937930148311216,
      "loss": 1.8407,
      "step": 116719
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.9729036092758179,
      "learning_rate": 0.00029378892514515234,
      "loss": 2.2427,
      "step": 116720
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.9220015406608582,
      "learning_rate": 0.00029378483546033785,
      "loss": 1.8197,
      "step": 116721
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.91426020860672,
      "learning_rate": 0.0002937807457766789,
      "loss": 2.1364,
      "step": 116722
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.9020753502845764,
      "learning_rate": 0.0002937766560941762,
      "loss": 1.8883,
      "step": 116723
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.8721117377281189,
      "learning_rate": 0.0002937725664128306,
      "loss": 2.1047,
      "step": 116724
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.9800683856010437,
      "learning_rate": 0.0002937684767326428,
      "loss": 2.172,
      "step": 116725
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.9442843198776245,
      "learning_rate": 0.0002937643870536135,
      "loss": 2.1442,
      "step": 116726
    },
    {
      "epoch": 1.52,
      "grad_norm": 1.05281400680542,
      "learning_rate": 0.0002937602973757435,
      "loss": 2.1197,
      "step": 116727
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.9352182149887085,
      "learning_rate": 0.0002937562076990336,
      "loss": 2.0331,
      "step": 116728
    },
    {
      "epoch": 1.52,
      "grad_norm": 1.0343092679977417,
      "learning_rate": 0.0002937521180234846,
      "loss": 2.1952,
      "step": 116729
    },
    {
      "epoch": 1.52,
      "grad_norm": 1.0185006856918335,
      "learning_rate": 0.0002937480283490971,
      "loss": 2.1425,
      "step": 116730
    },
    {
      "epoch": 1.52,
      "grad_norm": 1.0426074266433716,
      "learning_rate": 0.0002937439386758721,
      "loss": 2.1345,
      "step": 116731
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.9902742505073547,
      "learning_rate": 0.0002937398490038101,
      "loss": 2.2355,
      "step": 116732
    },
    {
      "epoch": 1.52,
      "grad_norm": 1.0504230260849,
      "learning_rate": 0.000293735759332912,
      "loss": 1.9264,
      "step": 116733
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.9721372723579407,
      "learning_rate": 0.00029373166966317857,
      "loss": 2.0941,
      "step": 116734
    },
    {
      "epoch": 1.52,
      "grad_norm": 1.0022920370101929,
      "learning_rate": 0.0002937275799946105,
      "loss": 2.285,
      "step": 116735
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.951972246170044,
      "learning_rate": 0.00029372349032720854,
      "loss": 2.3314,
      "step": 116736
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.9232303500175476,
      "learning_rate": 0.0002937194006609737,
      "loss": 2.1727,
      "step": 116737
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.9312554001808167,
      "learning_rate": 0.0002937153109959064,
      "loss": 2.2038,
      "step": 116738
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.927749752998352,
      "learning_rate": 0.00029371122133200757,
      "loss": 2.2511,
      "step": 116739
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.8795365691184998,
      "learning_rate": 0.0002937071316692779,
      "loss": 2.1835,
      "step": 116740
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.9093538522720337,
      "learning_rate": 0.0002937030420077182,
      "loss": 2.2619,
      "step": 116741
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.8933476209640503,
      "learning_rate": 0.0002936989523473292,
      "loss": 2.1697,
      "step": 116742
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.9056224226951599,
      "learning_rate": 0.00029369486268811186,
      "loss": 2.109,
      "step": 116743
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.9601993560791016,
      "learning_rate": 0.00029369077303006665,
      "loss": 2.1477,
      "step": 116744
    },
    {
      "epoch": 1.52,
      "grad_norm": 1.0321632623672485,
      "learning_rate": 0.0002936866833731944,
      "loss": 2.2066,
      "step": 116745
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.8655925989151001,
      "learning_rate": 0.00029368259371749593,
      "loss": 2.119,
      "step": 116746
    },
    {
      "epoch": 1.52,
      "grad_norm": 1.2278542518615723,
      "learning_rate": 0.00029367850406297193,
      "loss": 2.3677,
      "step": 116747
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.8804319500923157,
      "learning_rate": 0.00029367441440962327,
      "loss": 2.1737,
      "step": 116748
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.8784747123718262,
      "learning_rate": 0.0002936703247574508,
      "loss": 1.9964,
      "step": 116749
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.9831126928329468,
      "learning_rate": 0.000293666235106455,
      "loss": 2.2851,
      "step": 116750
    },
    {
      "epoch": 1.52,
      "grad_norm": 1.0275176763534546,
      "learning_rate": 0.00029366214545663673,
      "loss": 2.1422,
      "step": 116751
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.9506359100341797,
      "learning_rate": 0.00029365805580799687,
      "loss": 2.1084,
      "step": 116752
    },
    {
      "epoch": 1.52,
      "grad_norm": 1.0051368474960327,
      "learning_rate": 0.000293653966160536,
      "loss": 2.2857,
      "step": 116753
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.9940698742866516,
      "learning_rate": 0.000293649876514255,
      "loss": 2.0638,
      "step": 116754
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.997015655040741,
      "learning_rate": 0.0002936457868691548,
      "loss": 2.0463,
      "step": 116755
    },
    {
      "epoch": 1.52,
      "grad_norm": 1.043389081954956,
      "learning_rate": 0.00029364169722523575,
      "loss": 2.0486,
      "step": 116756
    },
    {
      "epoch": 1.52,
      "grad_norm": 1.1548280715942383,
      "learning_rate": 0.00029363760758249893,
      "loss": 2.0687,
      "step": 116757
    },
    {
      "epoch": 1.52,
      "grad_norm": 1.0083035230636597,
      "learning_rate": 0.0002936335179409449,
      "loss": 2.2543,
      "step": 116758
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.9421214461326599,
      "learning_rate": 0.00029362942830057453,
      "loss": 2.3215,
      "step": 116759
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.9761277437210083,
      "learning_rate": 0.0002936253386613886,
      "loss": 2.2898,
      "step": 116760
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.9241862893104553,
      "learning_rate": 0.000293621249023388,
      "loss": 2.2438,
      "step": 116761
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.9470279812812805,
      "learning_rate": 0.00029361715938657314,
      "loss": 2.3588,
      "step": 116762
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.9611905217170715,
      "learning_rate": 0.000293613069750945,
      "loss": 2.2661,
      "step": 116763
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.9677459001541138,
      "learning_rate": 0.00029360898011650434,
      "loss": 2.0825,
      "step": 116764
    },
    {
      "epoch": 1.52,
      "grad_norm": 1.0475648641586304,
      "learning_rate": 0.00029360489048325185,
      "loss": 2.1194,
      "step": 116765
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.9462305903434753,
      "learning_rate": 0.00029360080085118835,
      "loss": 2.0259,
      "step": 116766
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.9380697011947632,
      "learning_rate": 0.0002935967112203147,
      "loss": 2.1181,
      "step": 116767
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.9139915704727173,
      "learning_rate": 0.0002935926215906314,
      "loss": 2.0951,
      "step": 116768
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.9629857540130615,
      "learning_rate": 0.0002935885319621394,
      "loss": 2.1665,
      "step": 116769
    },
    {
      "epoch": 1.52,
      "grad_norm": 1.4851555824279785,
      "learning_rate": 0.0002935844423348394,
      "loss": 2.077,
      "step": 116770
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.9541587829589844,
      "learning_rate": 0.0002935803527087321,
      "loss": 1.8498,
      "step": 116771
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.9093815684318542,
      "learning_rate": 0.00029357626308381844,
      "loss": 2.3824,
      "step": 116772
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.9546211361885071,
      "learning_rate": 0.00029357217346009896,
      "loss": 2.0943,
      "step": 116773
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.9325047731399536,
      "learning_rate": 0.00029356808383757476,
      "loss": 2.155,
      "step": 116774
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.935849130153656,
      "learning_rate": 0.0002935639942162462,
      "loss": 2.1834,
      "step": 116775
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.9989047646522522,
      "learning_rate": 0.0002935599045961142,
      "loss": 2.0666,
      "step": 116776
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.8734363317489624,
      "learning_rate": 0.0002935558149771796,
      "loss": 2.1782,
      "step": 116777
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.9040246605873108,
      "learning_rate": 0.00029355172535944305,
      "loss": 2.1232,
      "step": 116778
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.9232739806175232,
      "learning_rate": 0.0002935476357429054,
      "loss": 2.0679,
      "step": 116779
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.909365713596344,
      "learning_rate": 0.00029354354612756744,
      "loss": 2.1149,
      "step": 116780
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.9271665811538696,
      "learning_rate": 0.00029353945651342976,
      "loss": 2.2792,
      "step": 116781
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.9041491746902466,
      "learning_rate": 0.00029353536690049316,
      "loss": 2.1683,
      "step": 116782
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.9067800045013428,
      "learning_rate": 0.0002935312772887585,
      "loss": 2.0407,
      "step": 116783
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.9853469133377075,
      "learning_rate": 0.00029352718767822655,
      "loss": 2.316,
      "step": 116784
    },
    {
      "epoch": 1.52,
      "grad_norm": 1.0593559741973877,
      "learning_rate": 0.00029352309806889796,
      "loss": 2.0519,
      "step": 116785
    },
    {
      "epoch": 1.52,
      "grad_norm": 1.0057224035263062,
      "learning_rate": 0.0002935190084607736,
      "loss": 2.0172,
      "step": 116786
    },
    {
      "epoch": 1.52,
      "grad_norm": 1.1896206140518188,
      "learning_rate": 0.0002935149188538542,
      "loss": 1.9161,
      "step": 116787
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.9404205083847046,
      "learning_rate": 0.00029351082924814045,
      "loss": 2.1489,
      "step": 116788
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.903662919998169,
      "learning_rate": 0.00029350673964363314,
      "loss": 2.1311,
      "step": 116789
    },
    {
      "epoch": 1.52,
      "grad_norm": 1.0783814191818237,
      "learning_rate": 0.00029350265004033304,
      "loss": 2.0371,
      "step": 116790
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.9190037846565247,
      "learning_rate": 0.0002934985604382409,
      "loss": 2.0064,
      "step": 116791
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.849855899810791,
      "learning_rate": 0.00029349447083735765,
      "loss": 1.9536,
      "step": 116792
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.8821768760681152,
      "learning_rate": 0.00029349038123768377,
      "loss": 2.2061,
      "step": 116793
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.9319401383399963,
      "learning_rate": 0.0002934862916392201,
      "loss": 2.0731,
      "step": 116794
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.8957199454307556,
      "learning_rate": 0.00029348220204196765,
      "loss": 2.1341,
      "step": 116795
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.9343154430389404,
      "learning_rate": 0.0002934781124459268,
      "loss": 2.3528,
      "step": 116796
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.8553250432014465,
      "learning_rate": 0.00029347402285109857,
      "loss": 2.165,
      "step": 116797
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.9184839725494385,
      "learning_rate": 0.0002934699332574837,
      "loss": 2.08,
      "step": 116798
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.8230480551719666,
      "learning_rate": 0.0002934658436650828,
      "loss": 2.0308,
      "step": 116799
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.9833681583404541,
      "learning_rate": 0.0002934617540738967,
      "loss": 1.9664,
      "step": 116800
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.9064148664474487,
      "learning_rate": 0.0002934576644839262,
      "loss": 2.171,
      "step": 116801
    },
    {
      "epoch": 1.52,
      "grad_norm": 1.097909688949585,
      "learning_rate": 0.00029345357489517215,
      "loss": 2.26,
      "step": 116802
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.9172912240028381,
      "learning_rate": 0.0002934494853076351,
      "loss": 2.3214,
      "step": 116803
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.8749344348907471,
      "learning_rate": 0.000293445395721316,
      "loss": 2.0109,
      "step": 116804
    },
    {
      "epoch": 1.52,
      "grad_norm": 1.0393712520599365,
      "learning_rate": 0.0002934413061362154,
      "loss": 2.0748,
      "step": 116805
    },
    {
      "epoch": 1.52,
      "grad_norm": 1.0035665035247803,
      "learning_rate": 0.0002934372165523343,
      "loss": 2.1077,
      "step": 116806
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.884067714214325,
      "learning_rate": 0.0002934331269696732,
      "loss": 2.1839,
      "step": 116807
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.9514397382736206,
      "learning_rate": 0.00029342903738823313,
      "loss": 2.089,
      "step": 116808
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.9919646978378296,
      "learning_rate": 0.0002934249478080147,
      "loss": 2.3383,
      "step": 116809
    },
    {
      "epoch": 1.52,
      "grad_norm": 1.0036407709121704,
      "learning_rate": 0.0002934208582290188,
      "loss": 2.1887,
      "step": 116810
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.8941931128501892,
      "learning_rate": 0.0002934167686512459,
      "loss": 2.22,
      "step": 116811
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.9542767405509949,
      "learning_rate": 0.00029341267907469703,
      "loss": 2.1144,
      "step": 116812
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.9515644907951355,
      "learning_rate": 0.00029340858949937287,
      "loss": 2.1851,
      "step": 116813
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.9390991926193237,
      "learning_rate": 0.0002934044999252741,
      "loss": 2.266,
      "step": 116814
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.9461546540260315,
      "learning_rate": 0.00029340041035240167,
      "loss": 2.1502,
      "step": 116815
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.9086068272590637,
      "learning_rate": 0.0002933963207807563,
      "loss": 2.129,
      "step": 116816
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.8945558667182922,
      "learning_rate": 0.00029339223121033854,
      "loss": 1.9858,
      "step": 116817
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.970527172088623,
      "learning_rate": 0.0002933881416411493,
      "loss": 2.0298,
      "step": 116818
    },
    {
      "epoch": 1.52,
      "grad_norm": 1.002620816230774,
      "learning_rate": 0.00029338405207318934,
      "loss": 2.162,
      "step": 116819
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.9563713073730469,
      "learning_rate": 0.0002933799625064594,
      "loss": 2.074,
      "step": 116820
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.9375866055488586,
      "learning_rate": 0.0002933758729409602,
      "loss": 2.4967,
      "step": 116821
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.8660370111465454,
      "learning_rate": 0.0002933717833766928,
      "loss": 2.2471,
      "step": 116822
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.982887864112854,
      "learning_rate": 0.00029336769381365746,
      "loss": 2.1483,
      "step": 116823
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.9371725916862488,
      "learning_rate": 0.00029336360425185525,
      "loss": 1.8725,
      "step": 116824
    },
    {
      "epoch": 1.52,
      "grad_norm": 1.1778606176376343,
      "learning_rate": 0.00029335951469128686,
      "loss": 2.0145,
      "step": 116825
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.9005905985832214,
      "learning_rate": 0.00029335542513195305,
      "loss": 2.1295,
      "step": 116826
    },
    {
      "epoch": 1.52,
      "grad_norm": 1.2742760181427002,
      "learning_rate": 0.0002933513355738546,
      "loss": 2.087,
      "step": 116827
    },
    {
      "epoch": 1.52,
      "grad_norm": 1.047377109527588,
      "learning_rate": 0.0002933472460169924,
      "loss": 2.2726,
      "step": 116828
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.9699674248695374,
      "learning_rate": 0.0002933431564613669,
      "loss": 2.0995,
      "step": 116829
    },
    {
      "epoch": 1.52,
      "grad_norm": 1.2927762269973755,
      "learning_rate": 0.0002933390669069791,
      "loss": 2.2446,
      "step": 116830
    },
    {
      "epoch": 1.52,
      "grad_norm": 1.2264683246612549,
      "learning_rate": 0.00029333497735382967,
      "loss": 2.2199,
      "step": 116831
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.9813784956932068,
      "learning_rate": 0.0002933308878019194,
      "loss": 2.114,
      "step": 116832
    },
    {
      "epoch": 1.52,
      "grad_norm": 1.040802001953125,
      "learning_rate": 0.000293326798251249,
      "loss": 2.3752,
      "step": 116833
    },
    {
      "epoch": 1.52,
      "grad_norm": 1.1665089130401611,
      "learning_rate": 0.00029332270870181945,
      "loss": 2.1444,
      "step": 116834
    },
    {
      "epoch": 1.52,
      "grad_norm": 1.4025744199752808,
      "learning_rate": 0.0002933186191536312,
      "loss": 2.1659,
      "step": 116835
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.9101978540420532,
      "learning_rate": 0.0002933145296066851,
      "loss": 2.0114,
      "step": 116836
    },
    {
      "epoch": 1.52,
      "grad_norm": 1.1132628917694092,
      "learning_rate": 0.000293310440060982,
      "loss": 2.066,
      "step": 116837
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.9124795198440552,
      "learning_rate": 0.0002933063505165226,
      "loss": 2.2334,
      "step": 116838
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.968072235584259,
      "learning_rate": 0.0002933022609733077,
      "loss": 2.1684,
      "step": 116839
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.9468767046928406,
      "learning_rate": 0.00029329817143133803,
      "loss": 2.3597,
      "step": 116840
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.8713604807853699,
      "learning_rate": 0.00029329408189061445,
      "loss": 2.2563,
      "step": 116841
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.9563228487968445,
      "learning_rate": 0.00029328999235113753,
      "loss": 2.1301,
      "step": 116842
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.9810130596160889,
      "learning_rate": 0.0002932859028129081,
      "loss": 2.0138,
      "step": 116843
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.9249342679977417,
      "learning_rate": 0.000293281813275927,
      "loss": 2.2912,
      "step": 116844
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.8806001543998718,
      "learning_rate": 0.0002932777237401949,
      "loss": 2.0663,
      "step": 116845
    },
    {
      "epoch": 1.52,
      "grad_norm": 1.1324156522750854,
      "learning_rate": 0.00029327363420571256,
      "loss": 1.915,
      "step": 116846
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.9355049133300781,
      "learning_rate": 0.000293269544672481,
      "loss": 2.0234,
      "step": 116847
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.875006377696991,
      "learning_rate": 0.00029326545514050056,
      "loss": 2.1653,
      "step": 116848
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.9105750322341919,
      "learning_rate": 0.00029326136560977223,
      "loss": 2.0801,
      "step": 116849
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.9000210165977478,
      "learning_rate": 0.0002932572760802967,
      "loss": 2.2482,
      "step": 116850
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.9575939774513245,
      "learning_rate": 0.0002932531865520748,
      "loss": 2.0034,
      "step": 116851
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.8606498837471008,
      "learning_rate": 0.0002932490970251073,
      "loss": 1.9755,
      "step": 116852
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.8991075754165649,
      "learning_rate": 0.00029324500749939496,
      "loss": 2.0544,
      "step": 116853
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.9324976205825806,
      "learning_rate": 0.0002932409179749384,
      "loss": 2.1523,
      "step": 116854
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.9197404980659485,
      "learning_rate": 0.0002932368284517385,
      "loss": 2.0847,
      "step": 116855
    },
    {
      "epoch": 1.52,
      "grad_norm": 1.0050526857376099,
      "learning_rate": 0.00029323273892979603,
      "loss": 1.9429,
      "step": 116856
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.9553592205047607,
      "learning_rate": 0.00029322864940911164,
      "loss": 2.3815,
      "step": 116857
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.9417493343353271,
      "learning_rate": 0.0002932245598896863,
      "loss": 2.132,
      "step": 116858
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.9683876037597656,
      "learning_rate": 0.00029322047037152067,
      "loss": 1.8808,
      "step": 116859
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.9649309515953064,
      "learning_rate": 0.0002932163808546153,
      "loss": 2.1884,
      "step": 116860
    },
    {
      "epoch": 1.52,
      "grad_norm": 1.0592926740646362,
      "learning_rate": 0.00029321229133897123,
      "loss": 2.3254,
      "step": 116861
    },
    {
      "epoch": 1.52,
      "grad_norm": 1.207001805305481,
      "learning_rate": 0.00029320820182458915,
      "loss": 1.9181,
      "step": 116862
    },
    {
      "epoch": 1.52,
      "grad_norm": 1.0536326169967651,
      "learning_rate": 0.00029320411231146973,
      "loss": 2.1558,
      "step": 116863
    },
    {
      "epoch": 1.52,
      "grad_norm": 1.0884311199188232,
      "learning_rate": 0.00029320002279961384,
      "loss": 2.0227,
      "step": 116864
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.9388247132301331,
      "learning_rate": 0.00029319593328902224,
      "loss": 2.1737,
      "step": 116865
    },
    {
      "epoch": 1.52,
      "grad_norm": 1.009209394454956,
      "learning_rate": 0.00029319184377969563,
      "loss": 2.3527,
      "step": 116866
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.9894383549690247,
      "learning_rate": 0.00029318775427163466,
      "loss": 2.2888,
      "step": 116867
    },
    {
      "epoch": 1.52,
      "grad_norm": 1.269411563873291,
      "learning_rate": 0.0002931836647648403,
      "loss": 2.0813,
      "step": 116868
    },
    {
      "epoch": 1.52,
      "grad_norm": 1.0265278816223145,
      "learning_rate": 0.0002931795752593132,
      "loss": 1.8138,
      "step": 116869
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.9014496207237244,
      "learning_rate": 0.0002931754857550542,
      "loss": 2.2899,
      "step": 116870
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.8989545702934265,
      "learning_rate": 0.000293171396252064,
      "loss": 1.7927,
      "step": 116871
    },
    {
      "epoch": 1.52,
      "grad_norm": 1.0402915477752686,
      "learning_rate": 0.0002931673067503434,
      "loss": 2.2278,
      "step": 116872
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.8861420750617981,
      "learning_rate": 0.00029316321724989303,
      "loss": 2.1847,
      "step": 116873
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.8838258385658264,
      "learning_rate": 0.00029315912775071374,
      "loss": 2.1106,
      "step": 116874
    },
    {
      "epoch": 1.52,
      "grad_norm": 1.0307986736297607,
      "learning_rate": 0.0002931550382528063,
      "loss": 2.0272,
      "step": 116875
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.9173614382743835,
      "learning_rate": 0.0002931509487561715,
      "loss": 2.2428,
      "step": 116876
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.9621559977531433,
      "learning_rate": 0.0002931468592608101,
      "loss": 2.2569,
      "step": 116877
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.9157767295837402,
      "learning_rate": 0.00029314276976672284,
      "loss": 2.2009,
      "step": 116878
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.978302538394928,
      "learning_rate": 0.00029313868027391047,
      "loss": 2.0799,
      "step": 116879
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.9474238753318787,
      "learning_rate": 0.0002931345907823737,
      "loss": 2.0479,
      "step": 116880
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.9296140670776367,
      "learning_rate": 0.00029313050129211335,
      "loss": 2.1027,
      "step": 116881
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.8827738165855408,
      "learning_rate": 0.0002931264118031301,
      "loss": 2.3358,
      "step": 116882
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.9717536568641663,
      "learning_rate": 0.00029312232231542494,
      "loss": 2.033,
      "step": 116883
    },
    {
      "epoch": 1.52,
      "grad_norm": 1.0246914625167847,
      "learning_rate": 0.00029311823282899836,
      "loss": 2.0497,
      "step": 116884
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.8953993320465088,
      "learning_rate": 0.0002931141433438512,
      "loss": 2.0235,
      "step": 116885
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.9023919105529785,
      "learning_rate": 0.00029311005385998436,
      "loss": 2.066,
      "step": 116886
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.8456873893737793,
      "learning_rate": 0.00029310596437739846,
      "loss": 2.0455,
      "step": 116887
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.8977803587913513,
      "learning_rate": 0.0002931018748960942,
      "loss": 1.8669,
      "step": 116888
    },
    {
      "epoch": 1.52,
      "grad_norm": 1.0340529680252075,
      "learning_rate": 0.00029309778541607254,
      "loss": 2.1072,
      "step": 116889
    },
    {
      "epoch": 1.52,
      "grad_norm": 1.2220560312271118,
      "learning_rate": 0.0002930936959373341,
      "loss": 2.1558,
      "step": 116890
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.942261815071106,
      "learning_rate": 0.00029308960645987967,
      "loss": 2.0201,
      "step": 116891
    },
    {
      "epoch": 1.52,
      "grad_norm": 1.0088481903076172,
      "learning_rate": 0.00029308551698370997,
      "loss": 2.2079,
      "step": 116892
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.9991647005081177,
      "learning_rate": 0.0002930814275088258,
      "loss": 2.059,
      "step": 116893
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.9334944486618042,
      "learning_rate": 0.000293077338035228,
      "loss": 1.9537,
      "step": 116894
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.9492112994194031,
      "learning_rate": 0.0002930732485629173,
      "loss": 2.2543,
      "step": 116895
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.9415882229804993,
      "learning_rate": 0.00029306915909189425,
      "loss": 2.248,
      "step": 116896
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.9270886182785034,
      "learning_rate": 0.00029306506962215983,
      "loss": 2.1376,
      "step": 116897
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.9363896250724792,
      "learning_rate": 0.00029306098015371475,
      "loss": 2.1235,
      "step": 116898
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.9215835928916931,
      "learning_rate": 0.0002930568906865598,
      "loss": 2.2748,
      "step": 116899
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.8940338492393494,
      "learning_rate": 0.00029305280122069566,
      "loss": 2.0763,
      "step": 116900
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.9478877186775208,
      "learning_rate": 0.00029304871175612324,
      "loss": 2.0435,
      "step": 116901
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.8995723724365234,
      "learning_rate": 0.00029304462229284313,
      "loss": 2.1572,
      "step": 116902
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.9735901355743408,
      "learning_rate": 0.0002930405328308561,
      "loss": 2.1526,
      "step": 116903
    },
    {
      "epoch": 1.52,
      "grad_norm": 1.00006103515625,
      "learning_rate": 0.000293036443370163,
      "loss": 2.0695,
      "step": 116904
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.9523420333862305,
      "learning_rate": 0.00029303235391076455,
      "loss": 2.1676,
      "step": 116905
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.9560003876686096,
      "learning_rate": 0.0002930282644526615,
      "loss": 1.9826,
      "step": 116906
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.9114907383918762,
      "learning_rate": 0.00029302417499585463,
      "loss": 1.9143,
      "step": 116907
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.8456131219863892,
      "learning_rate": 0.00029302008554034484,
      "loss": 2.3113,
      "step": 116908
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.9049077033996582,
      "learning_rate": 0.00029301599608613267,
      "loss": 2.2539,
      "step": 116909
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.9165647029876709,
      "learning_rate": 0.0002930119066332189,
      "loss": 2.1984,
      "step": 116910
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.9630593061447144,
      "learning_rate": 0.00029300781718160436,
      "loss": 2.0531,
      "step": 116911
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.8393998146057129,
      "learning_rate": 0.00029300372773128974,
      "loss": 2.1377,
      "step": 116912
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.845281183719635,
      "learning_rate": 0.00029299963828227594,
      "loss": 1.9471,
      "step": 116913
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.9731711745262146,
      "learning_rate": 0.00029299554883456373,
      "loss": 2.2508,
      "step": 116914
    },
    {
      "epoch": 1.52,
      "grad_norm": 1.0876562595367432,
      "learning_rate": 0.00029299145938815364,
      "loss": 2.1078,
      "step": 116915
    },
    {
      "epoch": 1.52,
      "grad_norm": 1.0120468139648438,
      "learning_rate": 0.0002929873699430466,
      "loss": 2.2577,
      "step": 116916
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.8789393901824951,
      "learning_rate": 0.0002929832804992433,
      "loss": 2.0136,
      "step": 116917
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.9693383574485779,
      "learning_rate": 0.0002929791910567446,
      "loss": 2.1415,
      "step": 116918
    },
    {
      "epoch": 1.52,
      "grad_norm": 1.427170753479004,
      "learning_rate": 0.0002929751016155512,
      "loss": 2.0965,
      "step": 116919
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.9200624227523804,
      "learning_rate": 0.00029297101217566396,
      "loss": 2.2582,
      "step": 116920
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.8539358973503113,
      "learning_rate": 0.0002929669227370834,
      "loss": 1.9687,
      "step": 116921
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.9860734939575195,
      "learning_rate": 0.00029296283329981047,
      "loss": 2.2282,
      "step": 116922
    },
    {
      "epoch": 1.52,
      "grad_norm": 1.4410927295684814,
      "learning_rate": 0.0002929587438638458,
      "loss": 2.1794,
      "step": 116923
    },
    {
      "epoch": 1.52,
      "grad_norm": 1.3039019107818604,
      "learning_rate": 0.0002929546544291903,
      "loss": 2.1251,
      "step": 116924
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.968207061290741,
      "learning_rate": 0.00029295056499584464,
      "loss": 2.1033,
      "step": 116925
    },
    {
      "epoch": 1.52,
      "grad_norm": 1.5958608388900757,
      "learning_rate": 0.0002929464755638097,
      "loss": 2.296,
      "step": 116926
    },
    {
      "epoch": 1.52,
      "grad_norm": 1.3751797676086426,
      "learning_rate": 0.0002929423861330861,
      "loss": 2.2784,
      "step": 116927
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.8538353443145752,
      "learning_rate": 0.00029293829670367456,
      "loss": 2.1207,
      "step": 116928
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.9714847803115845,
      "learning_rate": 0.000292934207275576,
      "loss": 2.2746,
      "step": 116929
    },
    {
      "epoch": 1.52,
      "grad_norm": 1.1783313751220703,
      "learning_rate": 0.00029293011784879106,
      "loss": 2.265,
      "step": 116930
    },
    {
      "epoch": 1.52,
      "grad_norm": 1.1983909606933594,
      "learning_rate": 0.00029292602842332055,
      "loss": 1.8056,
      "step": 116931
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.9096925258636475,
      "learning_rate": 0.00029292193899916534,
      "loss": 2.1437,
      "step": 116932
    },
    {
      "epoch": 1.52,
      "grad_norm": 1.0329852104187012,
      "learning_rate": 0.00029291784957632597,
      "loss": 2.1749,
      "step": 116933
    },
    {
      "epoch": 1.52,
      "grad_norm": 1.0679737329483032,
      "learning_rate": 0.00029291376015480326,
      "loss": 2.0925,
      "step": 116934
    },
    {
      "epoch": 1.52,
      "grad_norm": 1.1471099853515625,
      "learning_rate": 0.00029290967073459807,
      "loss": 1.9838,
      "step": 116935
    },
    {
      "epoch": 1.52,
      "grad_norm": 1.0029641389846802,
      "learning_rate": 0.00029290558131571105,
      "loss": 2.2987,
      "step": 116936
    },
    {
      "epoch": 1.52,
      "grad_norm": 1.1895631551742554,
      "learning_rate": 0.0002929014918981431,
      "loss": 2.2177,
      "step": 116937
    },
    {
      "epoch": 1.52,
      "grad_norm": 1.5751776695251465,
      "learning_rate": 0.00029289740248189495,
      "loss": 1.9892,
      "step": 116938
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.9496997594833374,
      "learning_rate": 0.00029289331306696717,
      "loss": 2.2138,
      "step": 116939
    },
    {
      "epoch": 1.52,
      "grad_norm": 1.1216579675674438,
      "learning_rate": 0.0002928892236533607,
      "loss": 2.3053,
      "step": 116940
    },
    {
      "epoch": 1.52,
      "grad_norm": 1.0278483629226685,
      "learning_rate": 0.00029288513424107626,
      "loss": 2.1724,
      "step": 116941
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.985048234462738,
      "learning_rate": 0.0002928810448301146,
      "loss": 2.1097,
      "step": 116942
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.9899569153785706,
      "learning_rate": 0.0002928769554204765,
      "loss": 2.2055,
      "step": 116943
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.8156539797782898,
      "learning_rate": 0.0002928728660121628,
      "loss": 2.2147,
      "step": 116944
    },
    {
      "epoch": 1.52,
      "grad_norm": 1.0575898885726929,
      "learning_rate": 0.0002928687766051741,
      "loss": 2.1653,
      "step": 116945
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.9401573538780212,
      "learning_rate": 0.00029286468719951116,
      "loss": 2.0994,
      "step": 116946
    },
    {
      "epoch": 1.52,
      "grad_norm": 1.014199137687683,
      "learning_rate": 0.00029286059779517484,
      "loss": 2.1662,
      "step": 116947
    },
    {
      "epoch": 1.52,
      "grad_norm": 1.1090455055236816,
      "learning_rate": 0.00029285650839216585,
      "loss": 2.0431,
      "step": 116948
    },
    {
      "epoch": 1.52,
      "grad_norm": 1.0471097230911255,
      "learning_rate": 0.000292852418990485,
      "loss": 2.1912,
      "step": 116949
    },
    {
      "epoch": 1.52,
      "grad_norm": 1.1451884508132935,
      "learning_rate": 0.00029284832959013314,
      "loss": 2.1956,
      "step": 116950
    },
    {
      "epoch": 1.52,
      "grad_norm": 1.3648123741149902,
      "learning_rate": 0.00029284424019111077,
      "loss": 2.1814,
      "step": 116951
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.9831047058105469,
      "learning_rate": 0.0002928401507934188,
      "loss": 2.4319,
      "step": 116952
    },
    {
      "epoch": 1.52,
      "grad_norm": 1.3166688680648804,
      "learning_rate": 0.000292836061397058,
      "loss": 2.2821,
      "step": 116953
    },
    {
      "epoch": 1.52,
      "grad_norm": 1.090084195137024,
      "learning_rate": 0.0002928319720020291,
      "loss": 2.0331,
      "step": 116954
    },
    {
      "epoch": 1.52,
      "grad_norm": 1.0138211250305176,
      "learning_rate": 0.0002928278826083329,
      "loss": 2.0438,
      "step": 116955
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.8697319626808167,
      "learning_rate": 0.0002928237932159701,
      "loss": 2.1893,
      "step": 116956
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.8782186508178711,
      "learning_rate": 0.00029281970382494157,
      "loss": 1.9919,
      "step": 116957
    },
    {
      "epoch": 1.52,
      "grad_norm": 1.2296684980392456,
      "learning_rate": 0.0002928156144352479,
      "loss": 2.0066,
      "step": 116958
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.9952811598777771,
      "learning_rate": 0.00029281152504689,
      "loss": 2.2458,
      "step": 116959
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.9470499157905579,
      "learning_rate": 0.0002928074356598685,
      "loss": 1.9236,
      "step": 116960
    },
    {
      "epoch": 1.52,
      "grad_norm": 1.0441075563430786,
      "learning_rate": 0.00029280334627418425,
      "loss": 2.0946,
      "step": 116961
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.8949535489082336,
      "learning_rate": 0.0002927992568898381,
      "loss": 2.1794,
      "step": 116962
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.9214822053909302,
      "learning_rate": 0.0002927951675068306,
      "loss": 1.9128,
      "step": 116963
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.8288366198539734,
      "learning_rate": 0.0002927910781251627,
      "loss": 1.9433,
      "step": 116964
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.9852588176727295,
      "learning_rate": 0.000292786988744835,
      "loss": 2.1713,
      "step": 116965
    },
    {
      "epoch": 1.52,
      "grad_norm": 1.0846563577651978,
      "learning_rate": 0.0002927828993658483,
      "loss": 2.2951,
      "step": 116966
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.9633880853652954,
      "learning_rate": 0.00029277880998820346,
      "loss": 2.1442,
      "step": 116967
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.9637120366096497,
      "learning_rate": 0.0002927747206119012,
      "loss": 2.033,
      "step": 116968
    },
    {
      "epoch": 1.52,
      "grad_norm": 1.0535610914230347,
      "learning_rate": 0.0002927706312369422,
      "loss": 2.1534,
      "step": 116969
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.888746976852417,
      "learning_rate": 0.00029276654186332723,
      "loss": 2.2768,
      "step": 116970
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.9384168386459351,
      "learning_rate": 0.0002927624524910572,
      "loss": 2.0895,
      "step": 116971
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.805097222328186,
      "learning_rate": 0.00029275836312013276,
      "loss": 1.9831,
      "step": 116972
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.8660808205604553,
      "learning_rate": 0.0002927542737505546,
      "loss": 2.4411,
      "step": 116973
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.9290743470191956,
      "learning_rate": 0.00029275018438232363,
      "loss": 2.199,
      "step": 116974
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.9286995530128479,
      "learning_rate": 0.00029274609501544054,
      "loss": 2.2315,
      "step": 116975
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.923888087272644,
      "learning_rate": 0.000292742005649906,
      "loss": 2.2522,
      "step": 116976
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.9232984185218811,
      "learning_rate": 0.0002927379162857209,
      "loss": 2.1064,
      "step": 116977
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.9782456159591675,
      "learning_rate": 0.00029273382692288606,
      "loss": 2.2488,
      "step": 116978
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.9671031832695007,
      "learning_rate": 0.000292729737561402,
      "loss": 2.0157,
      "step": 116979
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.855995774269104,
      "learning_rate": 0.00029272564820126965,
      "loss": 1.8864,
      "step": 116980
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.993460476398468,
      "learning_rate": 0.0002927215588424898,
      "loss": 2.2523,
      "step": 116981
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.9204860329627991,
      "learning_rate": 0.00029271746948506314,
      "loss": 2.1587,
      "step": 116982
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.9327847957611084,
      "learning_rate": 0.0002927133801289904,
      "loss": 2.0644,
      "step": 116983
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.9408928751945496,
      "learning_rate": 0.0002927092907742724,
      "loss": 2.0619,
      "step": 116984
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.881619393825531,
      "learning_rate": 0.0002927052014209098,
      "loss": 2.2348,
      "step": 116985
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.8864574432373047,
      "learning_rate": 0.00029270111206890356,
      "loss": 1.9265,
      "step": 116986
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.9476341605186462,
      "learning_rate": 0.0002926970227182543,
      "loss": 2.0178,
      "step": 116987
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.8640908002853394,
      "learning_rate": 0.0002926929333689628,
      "loss": 2.0757,
      "step": 116988
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.9963167309761047,
      "learning_rate": 0.00029268884402102976,
      "loss": 2.0669,
      "step": 116989
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.878031313419342,
      "learning_rate": 0.000292684754674456,
      "loss": 2.0319,
      "step": 116990
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.9427253007888794,
      "learning_rate": 0.0002926806653292423,
      "loss": 2.1036,
      "step": 116991
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.9208241105079651,
      "learning_rate": 0.0002926765759853894,
      "loss": 2.2462,
      "step": 116992
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.956807017326355,
      "learning_rate": 0.0002926724866428982,
      "loss": 2.1415,
      "step": 116993
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.8779264092445374,
      "learning_rate": 0.0002926683973017692,
      "loss": 2.0689,
      "step": 116994
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.8536608815193176,
      "learning_rate": 0.00029266430796200327,
      "loss": 2.221,
      "step": 116995
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.9491769075393677,
      "learning_rate": 0.0002926602186236012,
      "loss": 2.2069,
      "step": 116996
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.9176357984542847,
      "learning_rate": 0.00029265612928656374,
      "loss": 2.1186,
      "step": 116997
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.897126317024231,
      "learning_rate": 0.00029265203995089157,
      "loss": 2.037,
      "step": 116998
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.9757121801376343,
      "learning_rate": 0.00029264795061658577,
      "loss": 2.2353,
      "step": 116999
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.855262041091919,
      "learning_rate": 0.00029264386128364665,
      "loss": 2.0799,
      "step": 117000
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.9715033769607544,
      "learning_rate": 0.00029263977195207515,
      "loss": 2.2092,
      "step": 117001
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.8881614208221436,
      "learning_rate": 0.00029263568262187213,
      "loss": 2.037,
      "step": 117002
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.8956041932106018,
      "learning_rate": 0.00029263159329303824,
      "loss": 2.0123,
      "step": 117003
    },
    {
      "epoch": 1.52,
      "grad_norm": 1.0768486261367798,
      "learning_rate": 0.0002926275039655743,
      "loss": 2.1944,
      "step": 117004
    },
    {
      "epoch": 1.52,
      "grad_norm": 1.0121656656265259,
      "learning_rate": 0.0002926234146394811,
      "loss": 2.1069,
      "step": 117005
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.9306108951568604,
      "learning_rate": 0.0002926193253147593,
      "loss": 1.9978,
      "step": 117006
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.9420007467269897,
      "learning_rate": 0.00029261523599140966,
      "loss": 2.1959,
      "step": 117007
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.9772679805755615,
      "learning_rate": 0.00029261114666943297,
      "loss": 2.0032,
      "step": 117008
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.8875048756599426,
      "learning_rate": 0.00029260705734883007,
      "loss": 2.2318,
      "step": 117009
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.9728933572769165,
      "learning_rate": 0.00029260296802960167,
      "loss": 2.2417,
      "step": 117010
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.966256320476532,
      "learning_rate": 0.0002925988787117486,
      "loss": 2.214,
      "step": 117011
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.956753671169281,
      "learning_rate": 0.0002925947893952714,
      "loss": 2.1233,
      "step": 117012
    },
    {
      "epoch": 1.52,
      "grad_norm": 1.0266268253326416,
      "learning_rate": 0.00029259070008017093,
      "loss": 2.1583,
      "step": 117013
    },
    {
      "epoch": 1.52,
      "grad_norm": 1.0409623384475708,
      "learning_rate": 0.00029258661076644806,
      "loss": 2.159,
      "step": 117014
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.9894109964370728,
      "learning_rate": 0.0002925825214541035,
      "loss": 2.1431,
      "step": 117015
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.9433753490447998,
      "learning_rate": 0.00029257843214313794,
      "loss": 2.0068,
      "step": 117016
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.8690469861030579,
      "learning_rate": 0.0002925743428335523,
      "loss": 1.7639,
      "step": 117017
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.9746504426002502,
      "learning_rate": 0.0002925702535253471,
      "loss": 2.1967,
      "step": 117018
    },
    {
      "epoch": 1.52,
      "grad_norm": 1.0015218257904053,
      "learning_rate": 0.0002925661642185233,
      "loss": 2.0122,
      "step": 117019
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.9017809629440308,
      "learning_rate": 0.00029256207491308155,
      "loss": 2.0641,
      "step": 117020
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.910420298576355,
      "learning_rate": 0.00029255798560902266,
      "loss": 2.0906,
      "step": 117021
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.8844488263130188,
      "learning_rate": 0.00029255389630634737,
      "loss": 1.9953,
      "step": 117022
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.8851837515830994,
      "learning_rate": 0.00029254980700505655,
      "loss": 2.2804,
      "step": 117023
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.8682085275650024,
      "learning_rate": 0.00029254571770515075,
      "loss": 2.0716,
      "step": 117024
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.929110050201416,
      "learning_rate": 0.0002925416284066309,
      "loss": 2.2069,
      "step": 117025
    },
    {
      "epoch": 1.52,
      "grad_norm": 1.0055079460144043,
      "learning_rate": 0.00029253753910949765,
      "loss": 2.1249,
      "step": 117026
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.8932390213012695,
      "learning_rate": 0.0002925334498137518,
      "loss": 2.1871,
      "step": 117027
    },
    {
      "epoch": 1.52,
      "grad_norm": 1.0389779806137085,
      "learning_rate": 0.00029252936051939415,
      "loss": 2.0984,
      "step": 117028
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.8749178647994995,
      "learning_rate": 0.00029252527122642556,
      "loss": 2.211,
      "step": 117029
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.9170405268669128,
      "learning_rate": 0.00029252118193484654,
      "loss": 1.9927,
      "step": 117030
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.9668149948120117,
      "learning_rate": 0.00029251709264465795,
      "loss": 2.0749,
      "step": 117031
    },
    {
      "epoch": 1.52,
      "grad_norm": 1.0565584897994995,
      "learning_rate": 0.00029251300335586056,
      "loss": 2.321,
      "step": 117032
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.9488804936408997,
      "learning_rate": 0.00029250891406845523,
      "loss": 2.233,
      "step": 117033
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.9172301292419434,
      "learning_rate": 0.00029250482478244256,
      "loss": 2.2156,
      "step": 117034
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.9626458287239075,
      "learning_rate": 0.0002925007354978235,
      "loss": 2.2191,
      "step": 117035
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.9017062783241272,
      "learning_rate": 0.0002924966462145986,
      "loss": 2.1058,
      "step": 117036
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.9203673601150513,
      "learning_rate": 0.0002924925569327687,
      "loss": 2.0015,
      "step": 117037
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.8743919134140015,
      "learning_rate": 0.0002924884676523346,
      "loss": 2.1016,
      "step": 117038
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.9785355925559998,
      "learning_rate": 0.000292484378373297,
      "loss": 2.0086,
      "step": 117039
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.9341573715209961,
      "learning_rate": 0.0002924802890956567,
      "loss": 2.169,
      "step": 117040
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.9247885942459106,
      "learning_rate": 0.0002924761998194145,
      "loss": 2.0846,
      "step": 117041
    },
    {
      "epoch": 1.52,
      "grad_norm": 1.2644670009613037,
      "learning_rate": 0.0002924721105445712,
      "loss": 2.0864,
      "step": 117042
    },
    {
      "epoch": 1.52,
      "grad_norm": 1.1486955881118774,
      "learning_rate": 0.0002924680212711274,
      "loss": 2.2686,
      "step": 117043
    },
    {
      "epoch": 1.52,
      "grad_norm": 1.1300374269485474,
      "learning_rate": 0.0002924639319990839,
      "loss": 2.2268,
      "step": 117044
    },
    {
      "epoch": 1.52,
      "grad_norm": 1.049883484840393,
      "learning_rate": 0.0002924598427284415,
      "loss": 2.2594,
      "step": 117045
    },
    {
      "epoch": 1.52,
      "grad_norm": 1.702192783355713,
      "learning_rate": 0.00029245575345920096,
      "loss": 2.2065,
      "step": 117046
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.9921406507492065,
      "learning_rate": 0.00029245166419136304,
      "loss": 2.3269,
      "step": 117047
    },
    {
      "epoch": 1.52,
      "grad_norm": 1.2656956911087036,
      "learning_rate": 0.0002924475749249285,
      "loss": 2.2029,
      "step": 117048
    },
    {
      "epoch": 1.52,
      "grad_norm": 1.1720874309539795,
      "learning_rate": 0.0002924434856598982,
      "loss": 2.1548,
      "step": 117049
    },
    {
      "epoch": 1.52,
      "grad_norm": 1.118326187133789,
      "learning_rate": 0.00029243939639627265,
      "loss": 2.1649,
      "step": 117050
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.9800431132316589,
      "learning_rate": 0.00029243530713405285,
      "loss": 2.0503,
      "step": 117051
    },
    {
      "epoch": 1.52,
      "grad_norm": 1.1172266006469727,
      "learning_rate": 0.0002924312178732394,
      "loss": 2.2864,
      "step": 117052
    },
    {
      "epoch": 1.52,
      "grad_norm": 1.3880904912948608,
      "learning_rate": 0.0002924271286138331,
      "loss": 2.3237,
      "step": 117053
    },
    {
      "epoch": 1.52,
      "grad_norm": 1.085957646369934,
      "learning_rate": 0.0002924230393558349,
      "loss": 2.0102,
      "step": 117054
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.8800195455551147,
      "learning_rate": 0.0002924189500992453,
      "loss": 2.0869,
      "step": 117055
    },
    {
      "epoch": 1.52,
      "grad_norm": 1.0687228441238403,
      "learning_rate": 0.0002924148608440652,
      "loss": 1.9922,
      "step": 117056
    },
    {
      "epoch": 1.52,
      "grad_norm": 1.1817878484725952,
      "learning_rate": 0.0002924107715902953,
      "loss": 2.1419,
      "step": 117057
    },
    {
      "epoch": 1.52,
      "grad_norm": 1.1172943115234375,
      "learning_rate": 0.0002924066823379363,
      "loss": 2.199,
      "step": 117058
    },
    {
      "epoch": 1.52,
      "grad_norm": 1.0516537427902222,
      "learning_rate": 0.00029240259308698906,
      "loss": 2.094,
      "step": 117059
    },
    {
      "epoch": 1.52,
      "grad_norm": 1.220413327217102,
      "learning_rate": 0.0002923985038374545,
      "loss": 2.1875,
      "step": 117060
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.9199163317680359,
      "learning_rate": 0.000292394414589333,
      "loss": 2.1853,
      "step": 117061
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.934047520160675,
      "learning_rate": 0.00029239032534262557,
      "loss": 2.1065,
      "step": 117062
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.9212580919265747,
      "learning_rate": 0.000292386236097333,
      "loss": 2.135,
      "step": 117063
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.8840904235839844,
      "learning_rate": 0.0002923821468534559,
      "loss": 2.339,
      "step": 117064
    },
    {
      "epoch": 1.52,
      "grad_norm": 1.0255976915359497,
      "learning_rate": 0.0002923780576109951,
      "loss": 1.9555,
      "step": 117065
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.8969018459320068,
      "learning_rate": 0.0002923739683699514,
      "loss": 2.0356,
      "step": 117066
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.8417717814445496,
      "learning_rate": 0.0002923698791303255,
      "loss": 1.9738,
      "step": 117067
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.965022623538971,
      "learning_rate": 0.0002923657898921182,
      "loss": 2.0134,
      "step": 117068
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.8937378525733948,
      "learning_rate": 0.00029236170065533024,
      "loss": 2.017,
      "step": 117069
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.9340971112251282,
      "learning_rate": 0.0002923576114199624,
      "loss": 2.2588,
      "step": 117070
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.872642993927002,
      "learning_rate": 0.0002923535221860154,
      "loss": 2.2837,
      "step": 117071
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.9481805562973022,
      "learning_rate": 0.00029234943295349,
      "loss": 2.2189,
      "step": 117072
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.9306191802024841,
      "learning_rate": 0.00029234534372238704,
      "loss": 2.2972,
      "step": 117073
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.8942815661430359,
      "learning_rate": 0.0002923412544927072,
      "loss": 1.9456,
      "step": 117074
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.8866382241249084,
      "learning_rate": 0.0002923371652644512,
      "loss": 2.027,
      "step": 117075
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.9026795029640198,
      "learning_rate": 0.0002923330760376199,
      "loss": 2.2496,
      "step": 117076
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.9197945594787598,
      "learning_rate": 0.00029232898681221404,
      "loss": 2.2685,
      "step": 117077
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.9906834363937378,
      "learning_rate": 0.00029232489758823453,
      "loss": 2.1653,
      "step": 117078
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.9418942928314209,
      "learning_rate": 0.0002923208083656818,
      "loss": 1.9283,
      "step": 117079
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.9385956525802612,
      "learning_rate": 0.00029231671914455677,
      "loss": 2.2821,
      "step": 117080
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.91983962059021,
      "learning_rate": 0.0002923126299248602,
      "loss": 2.1625,
      "step": 117081
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.8914971947669983,
      "learning_rate": 0.00029230854070659286,
      "loss": 2.1254,
      "step": 117082
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.9973527789115906,
      "learning_rate": 0.00029230445148975553,
      "loss": 2.2514,
      "step": 117083
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.9579041004180908,
      "learning_rate": 0.00029230036227434903,
      "loss": 2.6744,
      "step": 117084
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.9736394882202148,
      "learning_rate": 0.00029229627306037395,
      "loss": 2.2479,
      "step": 117085
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.9275357127189636,
      "learning_rate": 0.00029229218384783115,
      "loss": 2.0342,
      "step": 117086
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.9436804056167603,
      "learning_rate": 0.00029228809463672134,
      "loss": 2.2432,
      "step": 117087
    },
    {
      "epoch": 1.52,
      "grad_norm": 1.0665810108184814,
      "learning_rate": 0.00029228400542704534,
      "loss": 2.0688,
      "step": 117088
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.9768679141998291,
      "learning_rate": 0.0002922799162188039,
      "loss": 2.1554,
      "step": 117089
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.9585298895835876,
      "learning_rate": 0.0002922758270119979,
      "loss": 1.8997,
      "step": 117090
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.9633844494819641,
      "learning_rate": 0.0002922717378066278,
      "loss": 2.0588,
      "step": 117091
    },
    {
      "epoch": 1.52,
      "grad_norm": 1.011748194694519,
      "learning_rate": 0.0002922676486026946,
      "loss": 2.184,
      "step": 117092
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.8475044965744019,
      "learning_rate": 0.0002922635594001989,
      "loss": 2.1284,
      "step": 117093
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.966971218585968,
      "learning_rate": 0.0002922594701991416,
      "loss": 2.2599,
      "step": 117094
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.9426405429840088,
      "learning_rate": 0.0002922553809995234,
      "loss": 2.2715,
      "step": 117095
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.9928441643714905,
      "learning_rate": 0.00029225129180134523,
      "loss": 2.0015,
      "step": 117096
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.8895741701126099,
      "learning_rate": 0.00029224720260460755,
      "loss": 2.0013,
      "step": 117097
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.9464809894561768,
      "learning_rate": 0.0002922431134093112,
      "loss": 2.1835,
      "step": 117098
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.9320032596588135,
      "learning_rate": 0.0002922390242154571,
      "loss": 2.2358,
      "step": 117099
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.9618322849273682,
      "learning_rate": 0.00029223493502304585,
      "loss": 2.0754,
      "step": 117100
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.8488978147506714,
      "learning_rate": 0.0002922308458320783,
      "loss": 2.1285,
      "step": 117101
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.9292868971824646,
      "learning_rate": 0.0002922267566425553,
      "loss": 2.123,
      "step": 117102
    },
    {
      "epoch": 1.52,
      "grad_norm": 1.123734951019287,
      "learning_rate": 0.00029222266745447734,
      "loss": 2.0785,
      "step": 117103
    },
    {
      "epoch": 1.52,
      "grad_norm": 1.0184091329574585,
      "learning_rate": 0.00029221857826784535,
      "loss": 2.2252,
      "step": 117104
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.99149489402771,
      "learning_rate": 0.00029221448908266006,
      "loss": 2.2246,
      "step": 117105
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.9366758465766907,
      "learning_rate": 0.0002922103998989223,
      "loss": 1.975,
      "step": 117106
    },
    {
      "epoch": 1.52,
      "grad_norm": 1.0523862838745117,
      "learning_rate": 0.0002922063107166327,
      "loss": 2.0941,
      "step": 117107
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.9396045207977295,
      "learning_rate": 0.00029220222153579213,
      "loss": 2.191,
      "step": 117108
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.9363016486167908,
      "learning_rate": 0.0002921981323564014,
      "loss": 1.9136,
      "step": 117109
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.8714787364006042,
      "learning_rate": 0.00029219404317846115,
      "loss": 1.9219,
      "step": 117110
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.88990318775177,
      "learning_rate": 0.00029218995400197214,
      "loss": 2.0107,
      "step": 117111
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.9234166741371155,
      "learning_rate": 0.0002921858648269351,
      "loss": 2.0203,
      "step": 117112
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.9350422024726868,
      "learning_rate": 0.0002921817756533509,
      "loss": 2.1715,
      "step": 117113
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.9017365574836731,
      "learning_rate": 0.00029217768648122025,
      "loss": 2.0414,
      "step": 117114
    },
    {
      "epoch": 1.52,
      "grad_norm": 1.2241119146347046,
      "learning_rate": 0.000292173597310544,
      "loss": 2.2131,
      "step": 117115
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.8891686201095581,
      "learning_rate": 0.0002921695081413228,
      "loss": 2.3145,
      "step": 117116
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.9477603435516357,
      "learning_rate": 0.0002921654189735574,
      "loss": 2.0352,
      "step": 117117
    },
    {
      "epoch": 1.52,
      "grad_norm": 1.0916281938552856,
      "learning_rate": 0.0002921613298072486,
      "loss": 2.1097,
      "step": 117118
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.9721990823745728,
      "learning_rate": 0.00029215724064239715,
      "loss": 1.9052,
      "step": 117119
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.9461231827735901,
      "learning_rate": 0.0002921531514790038,
      "loss": 2.4669,
      "step": 117120
    },
    {
      "epoch": 1.52,
      "grad_norm": 1.0399329662322998,
      "learning_rate": 0.00029214906231706943,
      "loss": 2.1592,
      "step": 117121
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.8568511605262756,
      "learning_rate": 0.00029214497315659463,
      "loss": 2.1619,
      "step": 117122
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.9327700138092041,
      "learning_rate": 0.0002921408839975802,
      "loss": 2.1898,
      "step": 117123
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.9372780919075012,
      "learning_rate": 0.0002921367948400269,
      "loss": 2.155,
      "step": 117124
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.898097038269043,
      "learning_rate": 0.00029213270568393556,
      "loss": 2.0736,
      "step": 117125
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.8460346460342407,
      "learning_rate": 0.0002921286165293069,
      "loss": 2.0605,
      "step": 117126
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.9743594527244568,
      "learning_rate": 0.0002921245273761418,
      "loss": 2.0959,
      "step": 117127
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.903130054473877,
      "learning_rate": 0.0002921204382244407,
      "loss": 2.1282,
      "step": 117128
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.9457905292510986,
      "learning_rate": 0.00029211634907420463,
      "loss": 2.0899,
      "step": 117129
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.9773814678192139,
      "learning_rate": 0.0002921122599254343,
      "loss": 2.2859,
      "step": 117130
    },
    {
      "epoch": 1.53,
      "grad_norm": 1.1115773916244507,
      "learning_rate": 0.00029210817077813037,
      "loss": 2.2089,
      "step": 117131
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.9102383255958557,
      "learning_rate": 0.0002921040816322937,
      "loss": 2.0732,
      "step": 117132
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.9998111724853516,
      "learning_rate": 0.0002920999924879252,
      "loss": 2.1745,
      "step": 117133
    },
    {
      "epoch": 1.53,
      "grad_norm": 1.215836524963379,
      "learning_rate": 0.0002920959033450254,
      "loss": 2.0122,
      "step": 117134
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.9105236530303955,
      "learning_rate": 0.00029209181420359497,
      "loss": 2.1,
      "step": 117135
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.8962849974632263,
      "learning_rate": 0.0002920877250636349,
      "loss": 2.0846,
      "step": 117136
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.9606695771217346,
      "learning_rate": 0.0002920836359251459,
      "loss": 2.4117,
      "step": 117137
    },
    {
      "epoch": 1.53,
      "grad_norm": 1.0820025205612183,
      "learning_rate": 0.0002920795467881286,
      "loss": 2.1778,
      "step": 117138
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.9094958901405334,
      "learning_rate": 0.000292075457652584,
      "loss": 2.2475,
      "step": 117139
    },
    {
      "epoch": 1.53,
      "grad_norm": 1.039539098739624,
      "learning_rate": 0.0002920713685185127,
      "loss": 2.1207,
      "step": 117140
    },
    {
      "epoch": 1.53,
      "grad_norm": 1.0936105251312256,
      "learning_rate": 0.00029206727938591547,
      "loss": 1.9474,
      "step": 117141
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.9737380146980286,
      "learning_rate": 0.000292063190254793,
      "loss": 2.146,
      "step": 117142
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.8733524680137634,
      "learning_rate": 0.0002920591011251462,
      "loss": 2.0186,
      "step": 117143
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.8722814321517944,
      "learning_rate": 0.0002920550119969758,
      "loss": 1.9871,
      "step": 117144
    },
    {
      "epoch": 1.53,
      "grad_norm": 1.228568434715271,
      "learning_rate": 0.0002920509228702825,
      "loss": 2.0441,
      "step": 117145
    },
    {
      "epoch": 1.53,
      "grad_norm": 1.2083587646484375,
      "learning_rate": 0.00029204683374506703,
      "loss": 2.0789,
      "step": 117146
    },
    {
      "epoch": 1.53,
      "grad_norm": 1.0565800666809082,
      "learning_rate": 0.0002920427446213302,
      "loss": 2.1849,
      "step": 117147
    },
    {
      "epoch": 1.53,
      "grad_norm": 1.0869375467300415,
      "learning_rate": 0.0002920386554990729,
      "loss": 2.0431,
      "step": 117148
    },
    {
      "epoch": 1.53,
      "grad_norm": 1.118403673171997,
      "learning_rate": 0.0002920345663782956,
      "loss": 2.2504,
      "step": 117149
    },
    {
      "epoch": 1.53,
      "grad_norm": 1.3191262483596802,
      "learning_rate": 0.0002920304772589993,
      "loss": 2.3668,
      "step": 117150
    },
    {
      "epoch": 1.53,
      "grad_norm": 1.0037692785263062,
      "learning_rate": 0.00029202638814118475,
      "loss": 2.3774,
      "step": 117151
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.9907599091529846,
      "learning_rate": 0.00029202229902485253,
      "loss": 2.0782,
      "step": 117152
    },
    {
      "epoch": 1.53,
      "grad_norm": 1.029292345046997,
      "learning_rate": 0.00029201820991000356,
      "loss": 2.0832,
      "step": 117153
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.9289277791976929,
      "learning_rate": 0.00029201412079663853,
      "loss": 2.0817,
      "step": 117154
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.8582109212875366,
      "learning_rate": 0.0002920100316847583,
      "loss": 2.1306,
      "step": 117155
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.9566038250923157,
      "learning_rate": 0.0002920059425743635,
      "loss": 1.9449,
      "step": 117156
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.9478472471237183,
      "learning_rate": 0.000292001853465455,
      "loss": 2.2775,
      "step": 117157
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.8971065878868103,
      "learning_rate": 0.00029199776435803347,
      "loss": 2.0613,
      "step": 117158
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.9941425919532776,
      "learning_rate": 0.00029199367525209967,
      "loss": 2.0766,
      "step": 117159
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.8765672445297241,
      "learning_rate": 0.0002919895861476544,
      "loss": 2.1522,
      "step": 117160
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.9333071112632751,
      "learning_rate": 0.0002919854970446984,
      "loss": 2.0853,
      "step": 117161
    },
    {
      "epoch": 1.53,
      "grad_norm": 1.2084850072860718,
      "learning_rate": 0.00029198140794323255,
      "loss": 2.1232,
      "step": 117162
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.9796567559242249,
      "learning_rate": 0.00029197731884325753,
      "loss": 2.1336,
      "step": 117163
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.9154494404792786,
      "learning_rate": 0.00029197322974477394,
      "loss": 2.0452,
      "step": 117164
    },
    {
      "epoch": 1.53,
      "grad_norm": 1.009033441543579,
      "learning_rate": 0.00029196914064778275,
      "loss": 2.2318,
      "step": 117165
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.9641046524047852,
      "learning_rate": 0.0002919650515522846,
      "loss": 2.06,
      "step": 117166
    },
    {
      "epoch": 1.53,
      "grad_norm": 1.0005861520767212,
      "learning_rate": 0.00029196096245828035,
      "loss": 2.1795,
      "step": 117167
    },
    {
      "epoch": 1.53,
      "grad_norm": 1.0775021314620972,
      "learning_rate": 0.00029195687336577066,
      "loss": 2.2501,
      "step": 117168
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.9340473413467407,
      "learning_rate": 0.00029195278427475647,
      "loss": 2.0261,
      "step": 117169
    },
    {
      "epoch": 1.53,
      "grad_norm": 1.2580878734588623,
      "learning_rate": 0.0002919486951852383,
      "loss": 2.0047,
      "step": 117170
    },
    {
      "epoch": 1.53,
      "grad_norm": 1.1571398973464966,
      "learning_rate": 0.000291944606097217,
      "loss": 2.2466,
      "step": 117171
    },
    {
      "epoch": 1.53,
      "grad_norm": 1.250896692276001,
      "learning_rate": 0.00029194051701069334,
      "loss": 2.0734,
      "step": 117172
    },
    {
      "epoch": 1.53,
      "grad_norm": 1.0303658246994019,
      "learning_rate": 0.0002919364279256681,
      "loss": 2.0467,
      "step": 117173
    },
    {
      "epoch": 1.53,
      "grad_norm": 1.2169183492660522,
      "learning_rate": 0.00029193233884214207,
      "loss": 2.1017,
      "step": 117174
    },
    {
      "epoch": 1.53,
      "grad_norm": 1.1570583581924438,
      "learning_rate": 0.00029192824976011596,
      "loss": 2.1199,
      "step": 117175
    },
    {
      "epoch": 1.53,
      "grad_norm": 1.0186625719070435,
      "learning_rate": 0.0002919241606795906,
      "loss": 1.8974,
      "step": 117176
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.8937734365463257,
      "learning_rate": 0.00029192007160056666,
      "loss": 2.0555,
      "step": 117177
    },
    {
      "epoch": 1.53,
      "grad_norm": 1.0355242490768433,
      "learning_rate": 0.00029191598252304487,
      "loss": 2.0178,
      "step": 117178
    },
    {
      "epoch": 1.53,
      "grad_norm": 1.1005568504333496,
      "learning_rate": 0.00029191189344702606,
      "loss": 2.2832,
      "step": 117179
    },
    {
      "epoch": 1.53,
      "grad_norm": 1.0201634168624878,
      "learning_rate": 0.000291907804372511,
      "loss": 1.9996,
      "step": 117180
    },
    {
      "epoch": 1.53,
      "grad_norm": 1.0320600271224976,
      "learning_rate": 0.00029190371529950044,
      "loss": 2.1266,
      "step": 117181
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.992837131023407,
      "learning_rate": 0.0002918996262279952,
      "loss": 2.1556,
      "step": 117182
    },
    {
      "epoch": 1.53,
      "grad_norm": 1.0575765371322632,
      "learning_rate": 0.0002918955371579959,
      "loss": 2.0817,
      "step": 117183
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.8535405397415161,
      "learning_rate": 0.00029189144808950334,
      "loss": 2.4673,
      "step": 117184
    },
    {
      "epoch": 1.53,
      "grad_norm": 1.0409622192382812,
      "learning_rate": 0.0002918873590225183,
      "loss": 2.2399,
      "step": 117185
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.9621934294700623,
      "learning_rate": 0.00029188326995704164,
      "loss": 2.1578,
      "step": 117186
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.9983980655670166,
      "learning_rate": 0.00029187918089307395,
      "loss": 2.2255,
      "step": 117187
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.9438080191612244,
      "learning_rate": 0.0002918750918306162,
      "loss": 2.0098,
      "step": 117188
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.9853721857070923,
      "learning_rate": 0.00029187100276966896,
      "loss": 2.1737,
      "step": 117189
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.884628415107727,
      "learning_rate": 0.000291866913710233,
      "loss": 1.9319,
      "step": 117190
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.9398162961006165,
      "learning_rate": 0.0002918628246523092,
      "loss": 2.1654,
      "step": 117191
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.9107600450515747,
      "learning_rate": 0.0002918587355958982,
      "loss": 2.1402,
      "step": 117192
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.9200544953346252,
      "learning_rate": 0.0002918546465410008,
      "loss": 2.0214,
      "step": 117193
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.966552197933197,
      "learning_rate": 0.00029185055748761794,
      "loss": 2.3347,
      "step": 117194
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.9516344666481018,
      "learning_rate": 0.00029184646843575005,
      "loss": 2.1275,
      "step": 117195
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.8846916556358337,
      "learning_rate": 0.0002918423793853981,
      "loss": 2.0943,
      "step": 117196
    },
    {
      "epoch": 1.53,
      "grad_norm": 1.0181372165679932,
      "learning_rate": 0.00029183829033656273,
      "loss": 2.0273,
      "step": 117197
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.9554504752159119,
      "learning_rate": 0.00029183420128924486,
      "loss": 2.2552,
      "step": 117198
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.9148358106613159,
      "learning_rate": 0.00029183011224344515,
      "loss": 2.1459,
      "step": 117199
    },
    {
      "epoch": 1.53,
      "grad_norm": 1.0380715131759644,
      "learning_rate": 0.00029182602319916445,
      "loss": 2.2789,
      "step": 117200
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.9958418011665344,
      "learning_rate": 0.0002918219341564034,
      "loss": 2.1451,
      "step": 117201
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.936533510684967,
      "learning_rate": 0.00029181784511516273,
      "loss": 2.0362,
      "step": 117202
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.9108068346977234,
      "learning_rate": 0.00029181375607544333,
      "loss": 1.991,
      "step": 117203
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.9249438643455505,
      "learning_rate": 0.0002918096670372459,
      "loss": 2.3427,
      "step": 117204
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.9762983918190002,
      "learning_rate": 0.0002918055780005712,
      "loss": 2.2907,
      "step": 117205
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.9730482697486877,
      "learning_rate": 0.0002918014889654201,
      "loss": 1.8372,
      "step": 117206
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.9859979748725891,
      "learning_rate": 0.0002917973999317932,
      "loss": 2.0821,
      "step": 117207
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.9335935711860657,
      "learning_rate": 0.00029179331089969123,
      "loss": 2.0981,
      "step": 117208
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.9538950324058533,
      "learning_rate": 0.0002917892218691151,
      "loss": 2.2595,
      "step": 117209
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.8482257127761841,
      "learning_rate": 0.00029178513284006547,
      "loss": 2.3415,
      "step": 117210
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.8526769876480103,
      "learning_rate": 0.00029178104381254313,
      "loss": 2.2937,
      "step": 117211
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.8961883783340454,
      "learning_rate": 0.000291776954786549,
      "loss": 1.9675,
      "step": 117212
    },
    {
      "epoch": 1.53,
      "grad_norm": 1.056699275970459,
      "learning_rate": 0.0002917728657620836,
      "loss": 2.1495,
      "step": 117213
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.8944573402404785,
      "learning_rate": 0.00029176877673914775,
      "loss": 2.0742,
      "step": 117214
    },
    {
      "epoch": 1.53,
      "grad_norm": 1.0237005949020386,
      "learning_rate": 0.0002917646877177422,
      "loss": 2.3809,
      "step": 117215
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.88130784034729,
      "learning_rate": 0.0002917605986978678,
      "loss": 2.0136,
      "step": 117216
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.9089876413345337,
      "learning_rate": 0.00029175650967952526,
      "loss": 2.0179,
      "step": 117217
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.9855591058731079,
      "learning_rate": 0.00029175242066271543,
      "loss": 2.2566,
      "step": 117218
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.9666329622268677,
      "learning_rate": 0.0002917483316474389,
      "loss": 2.2624,
      "step": 117219
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.9238709807395935,
      "learning_rate": 0.0002917442426336965,
      "loss": 2.1802,
      "step": 117220
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.967015266418457,
      "learning_rate": 0.0002917401536214889,
      "loss": 2.2053,
      "step": 117221
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.9259932041168213,
      "learning_rate": 0.0002917360646108171,
      "loss": 2.1604,
      "step": 117222
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.9299275875091553,
      "learning_rate": 0.0002917319756016817,
      "loss": 2.2042,
      "step": 117223
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.8943292498588562,
      "learning_rate": 0.0002917278865940835,
      "loss": 2.3883,
      "step": 117224
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.9696621894836426,
      "learning_rate": 0.0002917237975880232,
      "loss": 2.326,
      "step": 117225
    },
    {
      "epoch": 1.53,
      "grad_norm": 1.049799919128418,
      "learning_rate": 0.0002917197085835016,
      "loss": 2.1606,
      "step": 117226
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.9472928643226624,
      "learning_rate": 0.0002917156195805195,
      "loss": 1.9759,
      "step": 117227
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.8951773047447205,
      "learning_rate": 0.0002917115305790776,
      "loss": 2.1568,
      "step": 117228
    },
    {
      "epoch": 1.53,
      "grad_norm": 1.0055512189865112,
      "learning_rate": 0.00029170744157917664,
      "loss": 2.0136,
      "step": 117229
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.9908343553543091,
      "learning_rate": 0.0002917033525808175,
      "loss": 2.0723,
      "step": 117230
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.9419898986816406,
      "learning_rate": 0.0002916992635840008,
      "loss": 1.909,
      "step": 117231
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.9172052145004272,
      "learning_rate": 0.00029169517458872745,
      "loss": 2.2816,
      "step": 117232
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.9421393871307373,
      "learning_rate": 0.00029169108559499806,
      "loss": 2.063,
      "step": 117233
    },
    {
      "epoch": 1.53,
      "grad_norm": 1.0035400390625,
      "learning_rate": 0.00029168699660281344,
      "loss": 2.1972,
      "step": 117234
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.9443055987358093,
      "learning_rate": 0.00029168290761217436,
      "loss": 2.0426,
      "step": 117235
    },
    {
      "epoch": 1.53,
      "grad_norm": 1.019644856452942,
      "learning_rate": 0.00029167881862308167,
      "loss": 2.315,
      "step": 117236
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.9233940839767456,
      "learning_rate": 0.0002916747296355359,
      "loss": 2.1513,
      "step": 117237
    },
    {
      "epoch": 1.53,
      "grad_norm": 1.0417628288269043,
      "learning_rate": 0.0002916706406495381,
      "loss": 2.1105,
      "step": 117238
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.9903464317321777,
      "learning_rate": 0.00029166655166508877,
      "loss": 1.8445,
      "step": 117239
    },
    {
      "epoch": 1.53,
      "grad_norm": 1.1670445203781128,
      "learning_rate": 0.0002916624626821889,
      "loss": 2.1262,
      "step": 117240
    },
    {
      "epoch": 1.53,
      "grad_norm": 1.0984910726547241,
      "learning_rate": 0.0002916583737008391,
      "loss": 2.3073,
      "step": 117241
    },
    {
      "epoch": 1.53,
      "grad_norm": 1.0361528396606445,
      "learning_rate": 0.00029165428472104006,
      "loss": 1.9999,
      "step": 117242
    },
    {
      "epoch": 1.53,
      "grad_norm": 1.0938886404037476,
      "learning_rate": 0.0002916501957427928,
      "loss": 2.1232,
      "step": 117243
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.8667283654212952,
      "learning_rate": 0.0002916461067660978,
      "loss": 2.0252,
      "step": 117244
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.9949395060539246,
      "learning_rate": 0.000291642017790956,
      "loss": 2.0225,
      "step": 117245
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.9054332971572876,
      "learning_rate": 0.0002916379288173681,
      "loss": 2.1723,
      "step": 117246
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.9382691383361816,
      "learning_rate": 0.0002916338398453349,
      "loss": 2.1534,
      "step": 117247
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.859653651714325,
      "learning_rate": 0.0002916297508748571,
      "loss": 1.8048,
      "step": 117248
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.9587032198905945,
      "learning_rate": 0.00029162566190593555,
      "loss": 1.9407,
      "step": 117249
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.8987711668014526,
      "learning_rate": 0.0002916215729385708,
      "loss": 2.0968,
      "step": 117250
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.8740328550338745,
      "learning_rate": 0.00029161748397276387,
      "loss": 2.2382,
      "step": 117251
    },
    {
      "epoch": 1.53,
      "grad_norm": 1.2513878345489502,
      "learning_rate": 0.00029161339500851535,
      "loss": 2.0782,
      "step": 117252
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.8718277812004089,
      "learning_rate": 0.00029160930604582607,
      "loss": 1.9332,
      "step": 117253
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.9128834009170532,
      "learning_rate": 0.00029160521708469685,
      "loss": 2.0995,
      "step": 117254
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.8605735301971436,
      "learning_rate": 0.0002916011281251284,
      "loss": 1.9549,
      "step": 117255
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.8861589431762695,
      "learning_rate": 0.0002915970391671213,
      "loss": 2.0451,
      "step": 117256
    },
    {
      "epoch": 1.53,
      "grad_norm": 1.0666141510009766,
      "learning_rate": 0.00029159295021067655,
      "loss": 2.1547,
      "step": 117257
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.8477739095687866,
      "learning_rate": 0.00029158886125579483,
      "loss": 2.137,
      "step": 117258
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.9544152021408081,
      "learning_rate": 0.0002915847723024769,
      "loss": 2.1134,
      "step": 117259
    },
    {
      "epoch": 1.53,
      "grad_norm": 1.0585118532180786,
      "learning_rate": 0.0002915806833507235,
      "loss": 1.947,
      "step": 117260
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.9998103380203247,
      "learning_rate": 0.00029157659440053556,
      "loss": 2.113,
      "step": 117261
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.9530194997787476,
      "learning_rate": 0.0002915725054519135,
      "loss": 2.2401,
      "step": 117262
    },
    {
      "epoch": 1.53,
      "grad_norm": 1.0543540716171265,
      "learning_rate": 0.00029156841650485835,
      "loss": 2.0293,
      "step": 117263
    },
    {
      "epoch": 1.53,
      "grad_norm": 1.03511381149292,
      "learning_rate": 0.0002915643275593707,
      "loss": 2.2407,
      "step": 117264
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.9655483961105347,
      "learning_rate": 0.00029156023861545146,
      "loss": 2.0717,
      "step": 117265
    },
    {
      "epoch": 1.53,
      "grad_norm": 1.1823352575302124,
      "learning_rate": 0.0002915561496731014,
      "loss": 2.2211,
      "step": 117266
    },
    {
      "epoch": 1.53,
      "grad_norm": 1.2091578245162964,
      "learning_rate": 0.00029155206073232123,
      "loss": 2.1302,
      "step": 117267
    },
    {
      "epoch": 1.53,
      "grad_norm": 1.3577083349227905,
      "learning_rate": 0.0002915479717931116,
      "loss": 2.1739,
      "step": 117268
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.949537992477417,
      "learning_rate": 0.00029154388285547336,
      "loss": 2.1708,
      "step": 117269
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.9596301913261414,
      "learning_rate": 0.00029153979391940727,
      "loss": 2.0755,
      "step": 117270
    },
    {
      "epoch": 1.53,
      "grad_norm": 1.1104142665863037,
      "learning_rate": 0.0002915357049849141,
      "loss": 2.1556,
      "step": 117271
    },
    {
      "epoch": 1.53,
      "grad_norm": 1.1009249687194824,
      "learning_rate": 0.00029153161605199456,
      "loss": 2.0523,
      "step": 117272
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.8912953734397888,
      "learning_rate": 0.0002915275271206497,
      "loss": 1.9161,
      "step": 117273
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.8760297894477844,
      "learning_rate": 0.00029152343819087976,
      "loss": 1.9575,
      "step": 117274
    },
    {
      "epoch": 1.53,
      "grad_norm": 1.13758385181427,
      "learning_rate": 0.00029151934926268584,
      "loss": 2.2158,
      "step": 117275
    },
    {
      "epoch": 1.53,
      "grad_norm": 1.1357073783874512,
      "learning_rate": 0.0002915152603360686,
      "loss": 2.0187,
      "step": 117276
    },
    {
      "epoch": 1.53,
      "grad_norm": 1.049581527709961,
      "learning_rate": 0.00029151117141102887,
      "loss": 2.3656,
      "step": 117277
    },
    {
      "epoch": 1.53,
      "grad_norm": 1.004136323928833,
      "learning_rate": 0.0002915070824875674,
      "loss": 2.3036,
      "step": 117278
    },
    {
      "epoch": 1.53,
      "grad_norm": 1.1985849142074585,
      "learning_rate": 0.000291502993565685,
      "loss": 2.203,
      "step": 117279
    },
    {
      "epoch": 1.53,
      "grad_norm": 1.0560553073883057,
      "learning_rate": 0.0002914989046453822,
      "loss": 2.2246,
      "step": 117280
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.8981506824493408,
      "learning_rate": 0.00029149481572666,
      "loss": 2.0903,
      "step": 117281
    },
    {
      "epoch": 1.53,
      "grad_norm": 1.1276174783706665,
      "learning_rate": 0.000291490726809519,
      "loss": 2.1429,
      "step": 117282
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.9840210676193237,
      "learning_rate": 0.0002914866378939601,
      "loss": 2.0887,
      "step": 117283
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.9648956060409546,
      "learning_rate": 0.00029148254897998395,
      "loss": 2.1693,
      "step": 117284
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.9925299882888794,
      "learning_rate": 0.0002914784600675915,
      "loss": 2.1136,
      "step": 117285
    },
    {
      "epoch": 1.53,
      "grad_norm": 1.0076124668121338,
      "learning_rate": 0.0002914743711567832,
      "loss": 2.0543,
      "step": 117286
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.9821251630783081,
      "learning_rate": 0.00029147028224756,
      "loss": 2.238,
      "step": 117287
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.9309051036834717,
      "learning_rate": 0.0002914661933399227,
      "loss": 2.2373,
      "step": 117288
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.9600210189819336,
      "learning_rate": 0.0002914621044338719,
      "loss": 2.3567,
      "step": 117289
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.9034621715545654,
      "learning_rate": 0.00029145801552940846,
      "loss": 2.2267,
      "step": 117290
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.9845962524414062,
      "learning_rate": 0.0002914539266265333,
      "loss": 2.2277,
      "step": 117291
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.9268996715545654,
      "learning_rate": 0.0002914498377252469,
      "loss": 2.0186,
      "step": 117292
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.9399726390838623,
      "learning_rate": 0.0002914457488255501,
      "loss": 2.1282,
      "step": 117293
    },
    {
      "epoch": 1.53,
      "grad_norm": 1.007674217224121,
      "learning_rate": 0.0002914416599274437,
      "loss": 2.1368,
      "step": 117294
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.9712215065956116,
      "learning_rate": 0.0002914375710309285,
      "loss": 2.0893,
      "step": 117295
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.9483523964881897,
      "learning_rate": 0.0002914334821360052,
      "loss": 2.268,
      "step": 117296
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.9781200289726257,
      "learning_rate": 0.0002914293932426747,
      "loss": 2.1352,
      "step": 117297
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.9839006662368774,
      "learning_rate": 0.0002914253043509375,
      "loss": 1.9698,
      "step": 117298
    },
    {
      "epoch": 1.53,
      "grad_norm": 1.088435173034668,
      "learning_rate": 0.0002914212154607945,
      "loss": 2.1925,
      "step": 117299
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.8807916641235352,
      "learning_rate": 0.0002914171265722465,
      "loss": 2.0392,
      "step": 117300
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.9294847846031189,
      "learning_rate": 0.00029141303768529414,
      "loss": 2.199,
      "step": 117301
    },
    {
      "epoch": 1.53,
      "grad_norm": 1.0242137908935547,
      "learning_rate": 0.0002914089487999383,
      "loss": 2.0588,
      "step": 117302
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.9766436815261841,
      "learning_rate": 0.00029140485991617987,
      "loss": 1.9968,
      "step": 117303
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.8889625072479248,
      "learning_rate": 0.0002914007710340192,
      "loss": 2.0287,
      "step": 117304
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.9450153708457947,
      "learning_rate": 0.0002913966821534574,
      "loss": 2.2034,
      "step": 117305
    },
    {
      "epoch": 1.53,
      "grad_norm": 1.0140058994293213,
      "learning_rate": 0.0002913925932744951,
      "loss": 2.2439,
      "step": 117306
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.823215126991272,
      "learning_rate": 0.000291388504397133,
      "loss": 2.1663,
      "step": 117307
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.9079949259757996,
      "learning_rate": 0.000291384415521372,
      "loss": 2.2555,
      "step": 117308
    },
    {
      "epoch": 1.53,
      "grad_norm": 1.0001345872879028,
      "learning_rate": 0.0002913803266472128,
      "loss": 2.2767,
      "step": 117309
    },
    {
      "epoch": 1.53,
      "grad_norm": 1.0728977918624878,
      "learning_rate": 0.0002913762377746563,
      "loss": 2.2885,
      "step": 117310
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.9269622564315796,
      "learning_rate": 0.000291372148903703,
      "loss": 2.1969,
      "step": 117311
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.9586667418479919,
      "learning_rate": 0.0002913680600343537,
      "loss": 2.1517,
      "step": 117312
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.9968774318695068,
      "learning_rate": 0.0002913639711666093,
      "loss": 2.2919,
      "step": 117313
    },
    {
      "epoch": 1.53,
      "grad_norm": 1.2041088342666626,
      "learning_rate": 0.0002913598823004705,
      "loss": 2.1477,
      "step": 117314
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.9041202068328857,
      "learning_rate": 0.000291355793435938,
      "loss": 1.966,
      "step": 117315
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.9252951741218567,
      "learning_rate": 0.0002913517045730128,
      "loss": 2.1142,
      "step": 117316
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.9251282215118408,
      "learning_rate": 0.00029134761571169546,
      "loss": 2.1392,
      "step": 117317
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.9132247567176819,
      "learning_rate": 0.00029134352685198663,
      "loss": 2.0337,
      "step": 117318
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.9572080373764038,
      "learning_rate": 0.0002913394379938872,
      "loss": 2.0246,
      "step": 117319
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.8818448185920715,
      "learning_rate": 0.00029133534913739796,
      "loss": 2.2004,
      "step": 117320
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.9292704463005066,
      "learning_rate": 0.0002913312602825197,
      "loss": 1.848,
      "step": 117321
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.9718096256256104,
      "learning_rate": 0.0002913271714292531,
      "loss": 2.1042,
      "step": 117322
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.8696130514144897,
      "learning_rate": 0.0002913230825775989,
      "loss": 2.1788,
      "step": 117323
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.8625203967094421,
      "learning_rate": 0.00029131899372755797,
      "loss": 2.0364,
      "step": 117324
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.9236244559288025,
      "learning_rate": 0.00029131490487913097,
      "loss": 2.2926,
      "step": 117325
    },
    {
      "epoch": 1.53,
      "grad_norm": 1.0724961757659912,
      "learning_rate": 0.0002913108160323187,
      "loss": 1.9488,
      "step": 117326
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.8664571642875671,
      "learning_rate": 0.00029130672718712184,
      "loss": 2.0131,
      "step": 117327
    },
    {
      "epoch": 1.53,
      "grad_norm": 1.0237318277359009,
      "learning_rate": 0.00029130263834354134,
      "loss": 2.2608,
      "step": 117328
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.856338620185852,
      "learning_rate": 0.00029129854950157776,
      "loss": 1.904,
      "step": 117329
    },
    {
      "epoch": 1.53,
      "grad_norm": 1.0180753469467163,
      "learning_rate": 0.00029129446066123197,
      "loss": 2.2447,
      "step": 117330
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.9900206327438354,
      "learning_rate": 0.00029129037182250474,
      "loss": 2.2377,
      "step": 117331
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.9265586137771606,
      "learning_rate": 0.00029128628298539676,
      "loss": 2.1395,
      "step": 117332
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.8950908184051514,
      "learning_rate": 0.0002912821941499088,
      "loss": 2.1798,
      "step": 117333
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.8053184747695923,
      "learning_rate": 0.0002912781053160417,
      "loss": 1.9142,
      "step": 117334
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.8731195330619812,
      "learning_rate": 0.0002912740164837961,
      "loss": 2.0185,
      "step": 117335
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.8566747903823853,
      "learning_rate": 0.0002912699276531729,
      "loss": 2.1802,
      "step": 117336
    },
    {
      "epoch": 1.53,
      "grad_norm": 1.0507259368896484,
      "learning_rate": 0.0002912658388241727,
      "loss": 2.1289,
      "step": 117337
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.9564558863639832,
      "learning_rate": 0.0002912617499967964,
      "loss": 2.0321,
      "step": 117338
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.9304251670837402,
      "learning_rate": 0.00029125766117104473,
      "loss": 2.1858,
      "step": 117339
    },
    {
      "epoch": 1.53,
      "grad_norm": 1.0078487396240234,
      "learning_rate": 0.0002912535723469184,
      "loss": 2.389,
      "step": 117340
    },
    {
      "epoch": 1.53,
      "grad_norm": 1.0204463005065918,
      "learning_rate": 0.0002912494835244182,
      "loss": 2.1784,
      "step": 117341
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.9574500322341919,
      "learning_rate": 0.00029124539470354484,
      "loss": 2.0866,
      "step": 117342
    },
    {
      "epoch": 1.53,
      "grad_norm": 1.0287367105484009,
      "learning_rate": 0.00029124130588429914,
      "loss": 2.2545,
      "step": 117343
    },
    {
      "epoch": 1.53,
      "grad_norm": 1.1169005632400513,
      "learning_rate": 0.00029123721706668187,
      "loss": 2.0125,
      "step": 117344
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.9215746521949768,
      "learning_rate": 0.0002912331282506937,
      "loss": 2.0031,
      "step": 117345
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.9238904714584351,
      "learning_rate": 0.00029122903943633565,
      "loss": 2.065,
      "step": 117346
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.9771947264671326,
      "learning_rate": 0.0002912249506236082,
      "loss": 2.0663,
      "step": 117347
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.9430078864097595,
      "learning_rate": 0.0002912208618125121,
      "loss": 2.077,
      "step": 117348
    },
    {
      "epoch": 1.53,
      "grad_norm": 1.0284233093261719,
      "learning_rate": 0.00029121677300304825,
      "loss": 1.9334,
      "step": 117349
    },
    {
      "epoch": 1.53,
      "grad_norm": 1.0076384544372559,
      "learning_rate": 0.00029121268419521736,
      "loss": 2.1665,
      "step": 117350
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.9284971356391907,
      "learning_rate": 0.0002912085953890202,
      "loss": 2.1251,
      "step": 117351
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.895565390586853,
      "learning_rate": 0.00029120450658445764,
      "loss": 2.2182,
      "step": 117352
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.8961201310157776,
      "learning_rate": 0.0002912004177815302,
      "loss": 2.0384,
      "step": 117353
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.9157387018203735,
      "learning_rate": 0.0002911963289802388,
      "loss": 2.0702,
      "step": 117354
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.8742039203643799,
      "learning_rate": 0.00029119224018058415,
      "loss": 2.2153,
      "step": 117355
    },
    {
      "epoch": 1.53,
      "grad_norm": 1.0862089395523071,
      "learning_rate": 0.000291188151382567,
      "loss": 2.2306,
      "step": 117356
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.9015249013900757,
      "learning_rate": 0.0002911840625861882,
      "loss": 2.0212,
      "step": 117357
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.9566141366958618,
      "learning_rate": 0.00029117997379144854,
      "loss": 1.8942,
      "step": 117358
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.8690187335014343,
      "learning_rate": 0.0002911758849983486,
      "loss": 2.0668,
      "step": 117359
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.8694508075714111,
      "learning_rate": 0.0002911717962068892,
      "loss": 2.1723,
      "step": 117360
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.9551504850387573,
      "learning_rate": 0.0002911677074170711,
      "loss": 2.3015,
      "step": 117361
    },
    {
      "epoch": 1.53,
      "grad_norm": 1.4496713876724243,
      "learning_rate": 0.00029116361862889517,
      "loss": 2.1132,
      "step": 117362
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.8479276299476624,
      "learning_rate": 0.00029115952984236197,
      "loss": 1.9749,
      "step": 117363
    },
    {
      "epoch": 1.53,
      "grad_norm": 1.3624775409698486,
      "learning_rate": 0.00029115544105747264,
      "loss": 1.8744,
      "step": 117364
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.9180858731269836,
      "learning_rate": 0.0002911513522742275,
      "loss": 2.1952,
      "step": 117365
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.9755839705467224,
      "learning_rate": 0.00029114726349262747,
      "loss": 2.1829,
      "step": 117366
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.8754218816757202,
      "learning_rate": 0.0002911431747126733,
      "loss": 2.065,
      "step": 117367
    },
    {
      "epoch": 1.53,
      "grad_norm": 1.0448817014694214,
      "learning_rate": 0.00029113908593436584,
      "loss": 2.1777,
      "step": 117368
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.8605160713195801,
      "learning_rate": 0.00029113499715770577,
      "loss": 2.0515,
      "step": 117369
    },
    {
      "epoch": 1.53,
      "grad_norm": 1.0400439500808716,
      "learning_rate": 0.000291130908382694,
      "loss": 2.2135,
      "step": 117370
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.9906735420227051,
      "learning_rate": 0.00029112681960933104,
      "loss": 2.1968,
      "step": 117371
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.940945029258728,
      "learning_rate": 0.0002911227308376178,
      "loss": 2.2143,
      "step": 117372
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.9616701602935791,
      "learning_rate": 0.000291118642067555,
      "loss": 2.157,
      "step": 117373
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.9592714905738831,
      "learning_rate": 0.00029111455329914335,
      "loss": 2.1456,
      "step": 117374
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.8708984851837158,
      "learning_rate": 0.0002911104645323837,
      "loss": 2.1919,
      "step": 117375
    },
    {
      "epoch": 1.53,
      "grad_norm": 1.0302083492279053,
      "learning_rate": 0.0002911063757672768,
      "loss": 2.3906,
      "step": 117376
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.9717285633087158,
      "learning_rate": 0.00029110228700382354,
      "loss": 2.0565,
      "step": 117377
    },
    {
      "epoch": 1.53,
      "grad_norm": 1.0266302824020386,
      "learning_rate": 0.00029109819824202436,
      "loss": 2.2224,
      "step": 117378
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.8768059015274048,
      "learning_rate": 0.0002910941094818802,
      "loss": 2.0887,
      "step": 117379
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.9441152215003967,
      "learning_rate": 0.00029109002072339184,
      "loss": 2.054,
      "step": 117380
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.8976278901100159,
      "learning_rate": 0.00029108593196656003,
      "loss": 1.9774,
      "step": 117381
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.9776990413665771,
      "learning_rate": 0.0002910818432113855,
      "loss": 1.9157,
      "step": 117382
    },
    {
      "epoch": 1.53,
      "grad_norm": 1.0578750371932983,
      "learning_rate": 0.0002910777544578691,
      "loss": 2.5267,
      "step": 117383
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.9653261303901672,
      "learning_rate": 0.0002910736657060114,
      "loss": 1.9724,
      "step": 117384
    },
    {
      "epoch": 1.53,
      "grad_norm": 1.1238776445388794,
      "learning_rate": 0.0002910695769558133,
      "loss": 2.1285,
      "step": 117385
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.9728971719741821,
      "learning_rate": 0.00029106548820727557,
      "loss": 2.1542,
      "step": 117386
    },
    {
      "epoch": 1.53,
      "grad_norm": 1.0123281478881836,
      "learning_rate": 0.0002910613994603989,
      "loss": 2.3665,
      "step": 117387
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.9264209866523743,
      "learning_rate": 0.00029105731071518405,
      "loss": 2.1362,
      "step": 117388
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.9695366621017456,
      "learning_rate": 0.000291053221971632,
      "loss": 1.8978,
      "step": 117389
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.9563754796981812,
      "learning_rate": 0.00029104913322974314,
      "loss": 2.425,
      "step": 117390
    },
    {
      "epoch": 1.53,
      "grad_norm": 1.0768418312072754,
      "learning_rate": 0.0002910450444895185,
      "loss": 2.1724,
      "step": 117391
    },
    {
      "epoch": 1.53,
      "grad_norm": 1.045569896697998,
      "learning_rate": 0.0002910409557509587,
      "loss": 1.994,
      "step": 117392
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.9492043852806091,
      "learning_rate": 0.0002910368670140645,
      "loss": 2.099,
      "step": 117393
    },
    {
      "epoch": 1.53,
      "grad_norm": 1.0114318132400513,
      "learning_rate": 0.00029103277827883676,
      "loss": 2.2166,
      "step": 117394
    },
    {
      "epoch": 1.53,
      "grad_norm": 1.0588769912719727,
      "learning_rate": 0.0002910286895452764,
      "loss": 2.1109,
      "step": 117395
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.9359784126281738,
      "learning_rate": 0.0002910246008133837,
      "loss": 2.327,
      "step": 117396
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.9159666299819946,
      "learning_rate": 0.00029102051208315985,
      "loss": 1.9918,
      "step": 117397
    },
    {
      "epoch": 1.53,
      "grad_norm": 1.0689105987548828,
      "learning_rate": 0.00029101642335460534,
      "loss": 2.0408,
      "step": 117398
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.9656040668487549,
      "learning_rate": 0.0002910123346277211,
      "loss": 2.1875,
      "step": 117399
    },
    {
      "epoch": 1.53,
      "grad_norm": 1.02500581741333,
      "learning_rate": 0.0002910082459025078,
      "loss": 2.1804,
      "step": 117400
    },
    {
      "epoch": 1.53,
      "grad_norm": 1.025169849395752,
      "learning_rate": 0.0002910041571789664,
      "loss": 2.3044,
      "step": 117401
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.9498856663703918,
      "learning_rate": 0.0002910000684570974,
      "loss": 2.3126,
      "step": 117402
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.9971778392791748,
      "learning_rate": 0.0002909959797369016,
      "loss": 2.3011,
      "step": 117403
    },
    {
      "epoch": 1.53,
      "grad_norm": 1.0379964113235474,
      "learning_rate": 0.00029099189101837984,
      "loss": 2.1264,
      "step": 117404
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.9948906302452087,
      "learning_rate": 0.00029098780230153286,
      "loss": 2.3416,
      "step": 117405
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.8589892387390137,
      "learning_rate": 0.0002909837135863614,
      "loss": 2.1651,
      "step": 117406
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.9025788307189941,
      "learning_rate": 0.00029097962487286636,
      "loss": 1.9342,
      "step": 117407
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.9840295314788818,
      "learning_rate": 0.00029097553616104824,
      "loss": 2.1225,
      "step": 117408
    },
    {
      "epoch": 1.53,
      "grad_norm": 1.0857036113739014,
      "learning_rate": 0.00029097144745090803,
      "loss": 2.308,
      "step": 117409
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.9488869905471802,
      "learning_rate": 0.00029096735874244633,
      "loss": 2.0365,
      "step": 117410
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.8954497575759888,
      "learning_rate": 0.000290963270035664,
      "loss": 2.0079,
      "step": 117411
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.9335647821426392,
      "learning_rate": 0.0002909591813305617,
      "loss": 2.0388,
      "step": 117412
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.959678590297699,
      "learning_rate": 0.0002909550926271404,
      "loss": 2.0254,
      "step": 117413
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.8754573464393616,
      "learning_rate": 0.0002909510039254006,
      "loss": 2.2183,
      "step": 117414
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.9883123636245728,
      "learning_rate": 0.0002909469152253432,
      "loss": 2.0473,
      "step": 117415
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.9371488094329834,
      "learning_rate": 0.00029094282652696897,
      "loss": 1.9047,
      "step": 117416
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.9801566004753113,
      "learning_rate": 0.0002909387378302786,
      "loss": 2.0227,
      "step": 117417
    },
    {
      "epoch": 1.53,
      "grad_norm": 1.2695947885513306,
      "learning_rate": 0.00029093464913527295,
      "loss": 1.9359,
      "step": 117418
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.9424137473106384,
      "learning_rate": 0.00029093056044195273,
      "loss": 2.3865,
      "step": 117419
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.9328123331069946,
      "learning_rate": 0.0002909264717503186,
      "loss": 2.1839,
      "step": 117420
    },
    {
      "epoch": 1.53,
      "grad_norm": 1.1680593490600586,
      "learning_rate": 0.00029092238306037144,
      "loss": 1.9894,
      "step": 117421
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.933761477470398,
      "learning_rate": 0.000290918294372112,
      "loss": 1.981,
      "step": 117422
    },
    {
      "epoch": 1.53,
      "grad_norm": 1.2693414688110352,
      "learning_rate": 0.00029091420568554105,
      "loss": 2.2332,
      "step": 117423
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.9911113977432251,
      "learning_rate": 0.0002909101170006593,
      "loss": 2.0929,
      "step": 117424
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.9576522707939148,
      "learning_rate": 0.0002909060283174675,
      "loss": 2.3027,
      "step": 117425
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.9992876648902893,
      "learning_rate": 0.00029090193963596646,
      "loss": 2.2769,
      "step": 117426
    },
    {
      "epoch": 1.53,
      "grad_norm": 1.0124715566635132,
      "learning_rate": 0.0002908978509561569,
      "loss": 1.9498,
      "step": 117427
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.892022430896759,
      "learning_rate": 0.00029089376227803965,
      "loss": 1.9745,
      "step": 117428
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.8984308838844299,
      "learning_rate": 0.0002908896736016153,
      "loss": 1.9004,
      "step": 117429
    },
    {
      "epoch": 1.53,
      "grad_norm": 1.0946027040481567,
      "learning_rate": 0.0002908855849268848,
      "loss": 1.946,
      "step": 117430
    },
    {
      "epoch": 1.53,
      "grad_norm": 1.1187700033187866,
      "learning_rate": 0.00029088149625384904,
      "loss": 2.0686,
      "step": 117431
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.9258313179016113,
      "learning_rate": 0.00029087740758250837,
      "loss": 2.1589,
      "step": 117432
    },
    {
      "epoch": 1.53,
      "grad_norm": 1.0950508117675781,
      "learning_rate": 0.00029087331891286376,
      "loss": 2.0586,
      "step": 117433
    },
    {
      "epoch": 1.53,
      "grad_norm": 1.0388693809509277,
      "learning_rate": 0.000290869230244916,
      "loss": 2.1812,
      "step": 117434
    },
    {
      "epoch": 1.53,
      "grad_norm": 1.0192416906356812,
      "learning_rate": 0.00029086514157866585,
      "loss": 2.1823,
      "step": 117435
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.8975797295570374,
      "learning_rate": 0.000290861052914114,
      "loss": 2.1262,
      "step": 117436
    },
    {
      "epoch": 1.53,
      "grad_norm": 1.2029534578323364,
      "learning_rate": 0.0002908569642512613,
      "loss": 2.2197,
      "step": 117437
    },
    {
      "epoch": 1.53,
      "grad_norm": 1.1901546716690063,
      "learning_rate": 0.00029085287559010853,
      "loss": 2.1353,
      "step": 117438
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.9313318133354187,
      "learning_rate": 0.0002908487869306563,
      "loss": 2.2384,
      "step": 117439
    },
    {
      "epoch": 1.53,
      "grad_norm": 1.1782565116882324,
      "learning_rate": 0.0002908446982729054,
      "loss": 2.0642,
      "step": 117440
    },
    {
      "epoch": 1.53,
      "grad_norm": 1.4458262920379639,
      "learning_rate": 0.00029084060961685667,
      "loss": 2.0899,
      "step": 117441
    },
    {
      "epoch": 1.53,
      "grad_norm": 1.039648413658142,
      "learning_rate": 0.00029083652096251086,
      "loss": 2.1367,
      "step": 117442
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.9241423606872559,
      "learning_rate": 0.0002908324323098687,
      "loss": 1.9611,
      "step": 117443
    },
    {
      "epoch": 1.53,
      "grad_norm": 1.051832914352417,
      "learning_rate": 0.0002908283436589311,
      "loss": 2.0906,
      "step": 117444
    },
    {
      "epoch": 1.53,
      "grad_norm": 1.1841371059417725,
      "learning_rate": 0.00029082425500969854,
      "loss": 2.2126,
      "step": 117445
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.9184303879737854,
      "learning_rate": 0.00029082016636217193,
      "loss": 2.0685,
      "step": 117446
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.9928331971168518,
      "learning_rate": 0.000290816077716352,
      "loss": 2.0632,
      "step": 117447
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.9286205172538757,
      "learning_rate": 0.00029081198907223956,
      "loss": 2.3527,
      "step": 117448
    },
    {
      "epoch": 1.53,
      "grad_norm": 1.2761900424957275,
      "learning_rate": 0.00029080790042983537,
      "loss": 2.27,
      "step": 117449
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.9194349050521851,
      "learning_rate": 0.00029080381178914026,
      "loss": 2.1442,
      "step": 117450
    },
    {
      "epoch": 1.53,
      "grad_norm": 1.0569789409637451,
      "learning_rate": 0.0002907997231501548,
      "loss": 2.1397,
      "step": 117451
    },
    {
      "epoch": 1.53,
      "grad_norm": 1.0248032808303833,
      "learning_rate": 0.00029079563451287984,
      "loss": 2.0677,
      "step": 117452
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.8891249895095825,
      "learning_rate": 0.0002907915458773161,
      "loss": 1.834,
      "step": 117453
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.8938131332397461,
      "learning_rate": 0.0002907874572434644,
      "loss": 1.9973,
      "step": 117454
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.9483634829521179,
      "learning_rate": 0.0002907833686113255,
      "loss": 2.0592,
      "step": 117455
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.9101755619049072,
      "learning_rate": 0.0002907792799809003,
      "loss": 2.3723,
      "step": 117456
    },
    {
      "epoch": 1.53,
      "grad_norm": 1.0025951862335205,
      "learning_rate": 0.00029077519135218925,
      "loss": 2.112,
      "step": 117457
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.9899200797080994,
      "learning_rate": 0.00029077110272519326,
      "loss": 2.1116,
      "step": 117458
    },
    {
      "epoch": 1.53,
      "grad_norm": 1.5242741107940674,
      "learning_rate": 0.0002907670140999131,
      "loss": 2.1342,
      "step": 117459
    },
    {
      "epoch": 1.53,
      "grad_norm": 1.2646279335021973,
      "learning_rate": 0.00029076292547634955,
      "loss": 2.18,
      "step": 117460
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.8953059315681458,
      "learning_rate": 0.0002907588368545033,
      "loss": 2.2608,
      "step": 117461
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.9896126389503479,
      "learning_rate": 0.00029075474823437534,
      "loss": 1.9861,
      "step": 117462
    },
    {
      "epoch": 1.53,
      "grad_norm": 1.251541018486023,
      "learning_rate": 0.00029075065961596606,
      "loss": 2.1382,
      "step": 117463
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.9957275986671448,
      "learning_rate": 0.00029074657099927644,
      "loss": 2.1761,
      "step": 117464
    },
    {
      "epoch": 1.53,
      "grad_norm": 1.0865236520767212,
      "learning_rate": 0.0002907424823843072,
      "loss": 1.9836,
      "step": 117465
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.9078882336616516,
      "learning_rate": 0.00029073839377105913,
      "loss": 2.1432,
      "step": 117466
    },
    {
      "epoch": 1.53,
      "grad_norm": 1.2472840547561646,
      "learning_rate": 0.0002907343051595329,
      "loss": 2.1031,
      "step": 117467
    },
    {
      "epoch": 1.53,
      "grad_norm": 1.2365202903747559,
      "learning_rate": 0.00029073021654972954,
      "loss": 2.2635,
      "step": 117468
    },
    {
      "epoch": 1.53,
      "grad_norm": 1.011596441268921,
      "learning_rate": 0.00029072612794164946,
      "loss": 2.2128,
      "step": 117469
    },
    {
      "epoch": 1.53,
      "grad_norm": 1.038288950920105,
      "learning_rate": 0.0002907220393352936,
      "loss": 2.1238,
      "step": 117470
    },
    {
      "epoch": 1.53,
      "grad_norm": 1.15890371799469,
      "learning_rate": 0.0002907179507306626,
      "loss": 1.906,
      "step": 117471
    },
    {
      "epoch": 1.53,
      "grad_norm": 1.0264182090759277,
      "learning_rate": 0.0002907138621277574,
      "loss": 2.278,
      "step": 117472
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.8946894407272339,
      "learning_rate": 0.00029070977352657856,
      "loss": 2.3662,
      "step": 117473
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.9430409073829651,
      "learning_rate": 0.00029070568492712716,
      "loss": 2.3136,
      "step": 117474
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.9850295186042786,
      "learning_rate": 0.00029070159632940364,
      "loss": 2.1959,
      "step": 117475
    },
    {
      "epoch": 1.53,
      "grad_norm": 1.0879522562026978,
      "learning_rate": 0.00029069750773340884,
      "loss": 2.0514,
      "step": 117476
    },
    {
      "epoch": 1.53,
      "grad_norm": 1.0349420309066772,
      "learning_rate": 0.0002906934191391435,
      "loss": 2.0206,
      "step": 117477
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.9504423141479492,
      "learning_rate": 0.0002906893305466085,
      "loss": 1.962,
      "step": 117478
    },
    {
      "epoch": 1.53,
      "grad_norm": 1.183563470840454,
      "learning_rate": 0.0002906852419558045,
      "loss": 2.2275,
      "step": 117479
    },
    {
      "epoch": 1.53,
      "grad_norm": 1.05332350730896,
      "learning_rate": 0.0002906811533667324,
      "loss": 1.9842,
      "step": 117480
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.9051429629325867,
      "learning_rate": 0.0002906770647793927,
      "loss": 2.1398,
      "step": 117481
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.9217292666435242,
      "learning_rate": 0.0002906729761937864,
      "loss": 1.9826,
      "step": 117482
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.9627039432525635,
      "learning_rate": 0.00029066888760991407,
      "loss": 2.479,
      "step": 117483
    },
    {
      "epoch": 1.53,
      "grad_norm": 1.1442480087280273,
      "learning_rate": 0.00029066479902777666,
      "loss": 1.9634,
      "step": 117484
    },
    {
      "epoch": 1.53,
      "grad_norm": 1.12480890750885,
      "learning_rate": 0.0002906607104473748,
      "loss": 2.0202,
      "step": 117485
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.9329622983932495,
      "learning_rate": 0.0002906566218687094,
      "loss": 1.9746,
      "step": 117486
    },
    {
      "epoch": 1.53,
      "grad_norm": 1.0954692363739014,
      "learning_rate": 0.00029065253329178104,
      "loss": 2.165,
      "step": 117487
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.9944697618484497,
      "learning_rate": 0.0002906484447165905,
      "loss": 2.285,
      "step": 117488
    },
    {
      "epoch": 1.53,
      "grad_norm": 1.026741623878479,
      "learning_rate": 0.0002906443561431386,
      "loss": 2.0891,
      "step": 117489
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.9561648964881897,
      "learning_rate": 0.00029064026757142606,
      "loss": 2.1778,
      "step": 117490
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.9318273067474365,
      "learning_rate": 0.00029063617900145373,
      "loss": 2.1215,
      "step": 117491
    },
    {
      "epoch": 1.53,
      "grad_norm": 1.0440348386764526,
      "learning_rate": 0.00029063209043322236,
      "loss": 2.3673,
      "step": 117492
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.9643160104751587,
      "learning_rate": 0.00029062800186673254,
      "loss": 1.9917,
      "step": 117493
    },
    {
      "epoch": 1.53,
      "grad_norm": 1.0293930768966675,
      "learning_rate": 0.00029062391330198525,
      "loss": 2.168,
      "step": 117494
    },
    {
      "epoch": 1.53,
      "grad_norm": 1.007606863975525,
      "learning_rate": 0.0002906198247389811,
      "loss": 2.2296,
      "step": 117495
    },
    {
      "epoch": 1.53,
      "grad_norm": 1.0469551086425781,
      "learning_rate": 0.00029061573617772087,
      "loss": 2.2822,
      "step": 117496
    },
    {
      "epoch": 1.53,
      "grad_norm": 1.0572298765182495,
      "learning_rate": 0.0002906116476182054,
      "loss": 2.0079,
      "step": 117497
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.9261327981948853,
      "learning_rate": 0.00029060755906043544,
      "loss": 1.9448,
      "step": 117498
    },
    {
      "epoch": 1.53,
      "grad_norm": 1.0257632732391357,
      "learning_rate": 0.00029060347050441165,
      "loss": 2.1709,
      "step": 117499
    },
    {
      "epoch": 1.53,
      "grad_norm": 1.267334222793579,
      "learning_rate": 0.0002905993819501349,
      "loss": 2.2046,
      "step": 117500
    },
    {
      "epoch": 1.53,
      "grad_norm": 1.1123534440994263,
      "learning_rate": 0.0002905952933976059,
      "loss": 2.1515,
      "step": 117501
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.9183791875839233,
      "learning_rate": 0.00029059120484682536,
      "loss": 2.1094,
      "step": 117502
    },
    {
      "epoch": 1.53,
      "grad_norm": 1.2663731575012207,
      "learning_rate": 0.0002905871162977941,
      "loss": 2.0163,
      "step": 117503
    },
    {
      "epoch": 1.53,
      "grad_norm": 1.258965253829956,
      "learning_rate": 0.0002905830277505129,
      "loss": 2.0128,
      "step": 117504
    },
    {
      "epoch": 1.53,
      "grad_norm": 1.0178189277648926,
      "learning_rate": 0.0002905789392049825,
      "loss": 1.9967,
      "step": 117505
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.9207172393798828,
      "learning_rate": 0.00029057485066120363,
      "loss": 2.2376,
      "step": 117506
    },
    {
      "epoch": 1.53,
      "grad_norm": 1.2782939672470093,
      "learning_rate": 0.0002905707621191771,
      "loss": 2.1438,
      "step": 117507
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.9652500748634338,
      "learning_rate": 0.00029056667357890367,
      "loss": 1.9852,
      "step": 117508
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.9272680282592773,
      "learning_rate": 0.000290562585040384,
      "loss": 2.3228,
      "step": 117509
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.8830291628837585,
      "learning_rate": 0.00029055849650361893,
      "loss": 2.1116,
      "step": 117510
    },
    {
      "epoch": 1.53,
      "grad_norm": 1.1339200735092163,
      "learning_rate": 0.00029055440796860925,
      "loss": 2.3007,
      "step": 117511
    },
    {
      "epoch": 1.53,
      "grad_norm": 1.2683881521224976,
      "learning_rate": 0.0002905503194353557,
      "loss": 1.8926,
      "step": 117512
    },
    {
      "epoch": 1.53,
      "grad_norm": 1.0216052532196045,
      "learning_rate": 0.00029054623090385895,
      "loss": 2.173,
      "step": 117513
    },
    {
      "epoch": 1.53,
      "grad_norm": 1.063149094581604,
      "learning_rate": 0.00029054214237411984,
      "loss": 1.9409,
      "step": 117514
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.972766101360321,
      "learning_rate": 0.0002905380538461392,
      "loss": 2.162,
      "step": 117515
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.9853547811508179,
      "learning_rate": 0.0002905339653199177,
      "loss": 2.0306,
      "step": 117516
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.9693565964698792,
      "learning_rate": 0.0002905298767954561,
      "loss": 2.5463,
      "step": 117517
    },
    {
      "epoch": 1.53,
      "grad_norm": 1.023400068283081,
      "learning_rate": 0.0002905257882727552,
      "loss": 2.1044,
      "step": 117518
    },
    {
      "epoch": 1.53,
      "grad_norm": 1.0795584917068481,
      "learning_rate": 0.00029052169975181565,
      "loss": 2.4469,
      "step": 117519
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.8641060590744019,
      "learning_rate": 0.00029051761123263835,
      "loss": 2.0481,
      "step": 117520
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.9559603929519653,
      "learning_rate": 0.00029051352271522397,
      "loss": 2.1794,
      "step": 117521
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.9657928347587585,
      "learning_rate": 0.0002905094341995733,
      "loss": 2.1803,
      "step": 117522
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.9790124297142029,
      "learning_rate": 0.00029050534568568725,
      "loss": 1.9718,
      "step": 117523
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.9438086748123169,
      "learning_rate": 0.0002905012571735663,
      "loss": 2.1153,
      "step": 117524
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.972819983959198,
      "learning_rate": 0.0002904971686632114,
      "loss": 2.1846,
      "step": 117525
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.8800039887428284,
      "learning_rate": 0.00029049308015462324,
      "loss": 2.182,
      "step": 117526
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.87373286485672,
      "learning_rate": 0.0002904889916478025,
      "loss": 2.1818,
      "step": 117527
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.9453335404396057,
      "learning_rate": 0.0002904849031427501,
      "loss": 2.392,
      "step": 117528
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.8630750179290771,
      "learning_rate": 0.0002904808146394669,
      "loss": 2.2083,
      "step": 117529
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.9838916063308716,
      "learning_rate": 0.00029047672613795333,
      "loss": 2.1117,
      "step": 117530
    },
    {
      "epoch": 1.53,
      "grad_norm": 1.0342741012573242,
      "learning_rate": 0.00029047263763821036,
      "loss": 2.0625,
      "step": 117531
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.9403603076934814,
      "learning_rate": 0.00029046854914023865,
      "loss": 2.1532,
      "step": 117532
    },
    {
      "epoch": 1.53,
      "grad_norm": 1.116735577583313,
      "learning_rate": 0.000290464460644039,
      "loss": 2.2934,
      "step": 117533
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.9620764255523682,
      "learning_rate": 0.0002904603721496122,
      "loss": 2.0075,
      "step": 117534
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.819556713104248,
      "learning_rate": 0.00029045628365695917,
      "loss": 2.0259,
      "step": 117535
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.9151716232299805,
      "learning_rate": 0.00029045219516608036,
      "loss": 1.997,
      "step": 117536
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.9646050333976746,
      "learning_rate": 0.00029044810667697667,
      "loss": 2.217,
      "step": 117537
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.8960779905319214,
      "learning_rate": 0.0002904440181896488,
      "loss": 2.1357,
      "step": 117538
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.9565435647964478,
      "learning_rate": 0.00029043992970409765,
      "loss": 2.3235,
      "step": 117539
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.8822892308235168,
      "learning_rate": 0.00029043584122032383,
      "loss": 1.9901,
      "step": 117540
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.8934325575828552,
      "learning_rate": 0.00029043175273832833,
      "loss": 2.1113,
      "step": 117541
    },
    {
      "epoch": 1.53,
      "grad_norm": 1.121634840965271,
      "learning_rate": 0.00029042766425811157,
      "loss": 2.3683,
      "step": 117542
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.935427725315094,
      "learning_rate": 0.00029042357577967453,
      "loss": 2.103,
      "step": 117543
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.9242513179779053,
      "learning_rate": 0.00029041948730301796,
      "loss": 1.89,
      "step": 117544
    },
    {
      "epoch": 1.53,
      "grad_norm": 1.1117708683013916,
      "learning_rate": 0.0002904153988281425,
      "loss": 2.1891,
      "step": 117545
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.9209866523742676,
      "learning_rate": 0.0002904113103550491,
      "loss": 1.9301,
      "step": 117546
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.9192860722541809,
      "learning_rate": 0.0002904072218837385,
      "loss": 2.0749,
      "step": 117547
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.9763661026954651,
      "learning_rate": 0.00029040313341421127,
      "loss": 2.2093,
      "step": 117548
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.9340575933456421,
      "learning_rate": 0.00029039904494646826,
      "loss": 2.0191,
      "step": 117549
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.9780591130256653,
      "learning_rate": 0.0002903949564805102,
      "loss": 1.8831,
      "step": 117550
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.9892944693565369,
      "learning_rate": 0.00029039086801633797,
      "loss": 2.2445,
      "step": 117551
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.9932637214660645,
      "learning_rate": 0.0002903867795539522,
      "loss": 2.364,
      "step": 117552
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.9563598036766052,
      "learning_rate": 0.00029038269109335393,
      "loss": 2.126,
      "step": 117553
    },
    {
      "epoch": 1.53,
      "grad_norm": 1.0152535438537598,
      "learning_rate": 0.0002903786026345435,
      "loss": 2.1505,
      "step": 117554
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.9798646569252014,
      "learning_rate": 0.0002903745141775219,
      "loss": 2.2806,
      "step": 117555
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.9272614121437073,
      "learning_rate": 0.0002903704257222898,
      "loss": 1.9286,
      "step": 117556
    },
    {
      "epoch": 1.53,
      "grad_norm": 1.0327072143554688,
      "learning_rate": 0.00029036633726884807,
      "loss": 2.2348,
      "step": 117557
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.9813591837882996,
      "learning_rate": 0.00029036224881719743,
      "loss": 2.074,
      "step": 117558
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.8694512844085693,
      "learning_rate": 0.0002903581603673387,
      "loss": 2.1467,
      "step": 117559
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.9130688309669495,
      "learning_rate": 0.0002903540719192725,
      "loss": 2.0868,
      "step": 117560
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.9315358400344849,
      "learning_rate": 0.0002903499834729996,
      "loss": 2.1171,
      "step": 117561
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.8599368929862976,
      "learning_rate": 0.00029034589502852093,
      "loss": 2.0702,
      "step": 117562
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.9642576575279236,
      "learning_rate": 0.0002903418065858371,
      "loss": 1.9375,
      "step": 117563
    },
    {
      "epoch": 1.53,
      "grad_norm": 1.035352349281311,
      "learning_rate": 0.00029033771814494883,
      "loss": 1.9739,
      "step": 117564
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.9558113813400269,
      "learning_rate": 0.00029033362970585716,
      "loss": 2.1513,
      "step": 117565
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.9065020084381104,
      "learning_rate": 0.0002903295412685625,
      "loss": 2.075,
      "step": 117566
    },
    {
      "epoch": 1.53,
      "grad_norm": 1.241299033164978,
      "learning_rate": 0.0002903254528330658,
      "loss": 2.2402,
      "step": 117567
    },
    {
      "epoch": 1.53,
      "grad_norm": 1.0299508571624756,
      "learning_rate": 0.0002903213643993677,
      "loss": 2.2034,
      "step": 117568
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.8766865134239197,
      "learning_rate": 0.0002903172759674691,
      "loss": 2.1746,
      "step": 117569
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.9168745279312134,
      "learning_rate": 0.00029031318753737067,
      "loss": 2.23,
      "step": 117570
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.9329079985618591,
      "learning_rate": 0.0002903090991090732,
      "loss": 2.2811,
      "step": 117571
    },
    {
      "epoch": 1.53,
      "grad_norm": 1.0118576288223267,
      "learning_rate": 0.00029030501068257763,
      "loss": 2.2641,
      "step": 117572
    },
    {
      "epoch": 1.53,
      "grad_norm": 1.0102002620697021,
      "learning_rate": 0.0002903009222578844,
      "loss": 2.2668,
      "step": 117573
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.8880252242088318,
      "learning_rate": 0.0002902968338349944,
      "loss": 2.0686,
      "step": 117574
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.9278245568275452,
      "learning_rate": 0.0002902927454139084,
      "loss": 2.1381,
      "step": 117575
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.9927792549133301,
      "learning_rate": 0.00029028865699462714,
      "loss": 2.4084,
      "step": 117576
    },
    {
      "epoch": 1.53,
      "grad_norm": 1.015346646308899,
      "learning_rate": 0.0002902845685771514,
      "loss": 2.2728,
      "step": 117577
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.952467679977417,
      "learning_rate": 0.0002902804801614821,
      "loss": 2.1987,
      "step": 117578
    },
    {
      "epoch": 1.53,
      "grad_norm": 1.0058149099349976,
      "learning_rate": 0.0002902763917476197,
      "loss": 2.0353,
      "step": 117579
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.9741294384002686,
      "learning_rate": 0.0002902723033355651,
      "loss": 2.2596,
      "step": 117580
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.8564967513084412,
      "learning_rate": 0.00029026821492531907,
      "loss": 2.1176,
      "step": 117581
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.9409257173538208,
      "learning_rate": 0.00029026412651688235,
      "loss": 2.0628,
      "step": 117582
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.9167870879173279,
      "learning_rate": 0.00029026003811025576,
      "loss": 2.0978,
      "step": 117583
    },
    {
      "epoch": 1.53,
      "grad_norm": 1.07648766040802,
      "learning_rate": 0.00029025594970544,
      "loss": 2.1786,
      "step": 117584
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.994760274887085,
      "learning_rate": 0.0002902518613024359,
      "loss": 2.2404,
      "step": 117585
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.9918149709701538,
      "learning_rate": 0.0002902477729012441,
      "loss": 1.8461,
      "step": 117586
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.9808576107025146,
      "learning_rate": 0.0002902436845018654,
      "loss": 2.3419,
      "step": 117587
    },
    {
      "epoch": 1.53,
      "grad_norm": 1.10256028175354,
      "learning_rate": 0.00029023959610430055,
      "loss": 2.0857,
      "step": 117588
    },
    {
      "epoch": 1.53,
      "grad_norm": 1.103926658630371,
      "learning_rate": 0.0002902355077085504,
      "loss": 2.1311,
      "step": 117589
    },
    {
      "epoch": 1.53,
      "grad_norm": 1.0130691528320312,
      "learning_rate": 0.0002902314193146157,
      "loss": 2.1676,
      "step": 117590
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.944582462310791,
      "learning_rate": 0.0002902273309224971,
      "loss": 2.282,
      "step": 117591
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.9460113048553467,
      "learning_rate": 0.00029022324253219543,
      "loss": 2.2514,
      "step": 117592
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.8753140568733215,
      "learning_rate": 0.0002902191541437115,
      "loss": 1.8664,
      "step": 117593
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.9220619797706604,
      "learning_rate": 0.00029021506575704594,
      "loss": 1.9729,
      "step": 117594
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.9556135535240173,
      "learning_rate": 0.0002902109773721996,
      "loss": 2.117,
      "step": 117595
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.9278067350387573,
      "learning_rate": 0.0002902068889891733,
      "loss": 2.2618,
      "step": 117596
    },
    {
      "epoch": 1.53,
      "grad_norm": 1.0190061330795288,
      "learning_rate": 0.0002902028006079676,
      "loss": 2.0545,
      "step": 117597
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.9244179725646973,
      "learning_rate": 0.00029019871222858343,
      "loss": 2.1856,
      "step": 117598
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.8686878681182861,
      "learning_rate": 0.00029019462385102153,
      "loss": 1.9424,
      "step": 117599
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.9514455199241638,
      "learning_rate": 0.00029019053547528267,
      "loss": 2.1843,
      "step": 117600
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.9265666007995605,
      "learning_rate": 0.0002901864471013675,
      "loss": 2.1064,
      "step": 117601
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.9255975484848022,
      "learning_rate": 0.0002901823587292769,
      "loss": 2.2245,
      "step": 117602
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.886716902256012,
      "learning_rate": 0.0002901782703590116,
      "loss": 2.0886,
      "step": 117603
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.9182994365692139,
      "learning_rate": 0.0002901741819905723,
      "loss": 2.2975,
      "step": 117604
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.9511786699295044,
      "learning_rate": 0.0002901700936239598,
      "loss": 2.0051,
      "step": 117605
    },
    {
      "epoch": 1.53,
      "grad_norm": 1.0440086126327515,
      "learning_rate": 0.00029016600525917484,
      "loss": 2.0585,
      "step": 117606
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.9231773018836975,
      "learning_rate": 0.0002901619168962183,
      "loss": 2.3806,
      "step": 117607
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.9172011613845825,
      "learning_rate": 0.00029015782853509083,
      "loss": 2.2871,
      "step": 117608
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.8782615065574646,
      "learning_rate": 0.00029015374017579313,
      "loss": 2.0456,
      "step": 117609
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.9413632154464722,
      "learning_rate": 0.00029014965181832606,
      "loss": 2.1274,
      "step": 117610
    },
    {
      "epoch": 1.53,
      "grad_norm": 1.0192127227783203,
      "learning_rate": 0.00029014556346269036,
      "loss": 2.1803,
      "step": 117611
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.959884762763977,
      "learning_rate": 0.00029014147510888675,
      "loss": 2.2419,
      "step": 117612
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.9048506617546082,
      "learning_rate": 0.0002901373867569161,
      "loss": 2.2177,
      "step": 117613
    },
    {
      "epoch": 1.53,
      "grad_norm": 1.5467878580093384,
      "learning_rate": 0.0002901332984067792,
      "loss": 2.0656,
      "step": 117614
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.8859875798225403,
      "learning_rate": 0.00029012921005847655,
      "loss": 2.142,
      "step": 117615
    },
    {
      "epoch": 1.53,
      "grad_norm": 1.0465198755264282,
      "learning_rate": 0.00029012512171200903,
      "loss": 2.3003,
      "step": 117616
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.9833816885948181,
      "learning_rate": 0.0002901210333673775,
      "loss": 2.0755,
      "step": 117617
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.9117367267608643,
      "learning_rate": 0.0002901169450245826,
      "loss": 2.1144,
      "step": 117618
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.9139617085456848,
      "learning_rate": 0.0002901128566836252,
      "loss": 2.2459,
      "step": 117619
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.8983425498008728,
      "learning_rate": 0.00029010876834450615,
      "loss": 2.0659,
      "step": 117620
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.934607982635498,
      "learning_rate": 0.0002901046800072259,
      "loss": 2.156,
      "step": 117621
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.9620131850242615,
      "learning_rate": 0.00029010059167178535,
      "loss": 2.0731,
      "step": 117622
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.9312933683395386,
      "learning_rate": 0.00029009650333818536,
      "loss": 2.0917,
      "step": 117623
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.9019465446472168,
      "learning_rate": 0.00029009241500642656,
      "loss": 2.2342,
      "step": 117624
    },
    {
      "epoch": 1.53,
      "grad_norm": 1.1383874416351318,
      "learning_rate": 0.00029008832667650977,
      "loss": 2.1401,
      "step": 117625
    },
    {
      "epoch": 1.53,
      "grad_norm": 1.1355037689208984,
      "learning_rate": 0.0002900842383484359,
      "loss": 2.3041,
      "step": 117626
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.9081991910934448,
      "learning_rate": 0.0002900801500222054,
      "loss": 2.2718,
      "step": 117627
    },
    {
      "epoch": 1.53,
      "grad_norm": 1.1132651567459106,
      "learning_rate": 0.00029007606169781917,
      "loss": 1.9381,
      "step": 117628
    },
    {
      "epoch": 1.53,
      "grad_norm": 1.0614908933639526,
      "learning_rate": 0.00029007197337527804,
      "loss": 2.2287,
      "step": 117629
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.9276248216629028,
      "learning_rate": 0.0002900678850545827,
      "loss": 2.288,
      "step": 117630
    },
    {
      "epoch": 1.53,
      "grad_norm": 1.0123066902160645,
      "learning_rate": 0.00029006379673573394,
      "loss": 2.2792,
      "step": 117631
    },
    {
      "epoch": 1.53,
      "grad_norm": 1.1445534229278564,
      "learning_rate": 0.0002900597084187326,
      "loss": 2.2329,
      "step": 117632
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.8903946876525879,
      "learning_rate": 0.00029005562010357927,
      "loss": 2.1036,
      "step": 117633
    },
    {
      "epoch": 1.53,
      "grad_norm": 1.0380083322525024,
      "learning_rate": 0.0002900515317902748,
      "loss": 2.1251,
      "step": 117634
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.9435300827026367,
      "learning_rate": 0.00029004744347881984,
      "loss": 2.354,
      "step": 117635
    },
    {
      "epoch": 1.53,
      "grad_norm": 1.0703476667404175,
      "learning_rate": 0.00029004335516921536,
      "loss": 2.0336,
      "step": 117636
    },
    {
      "epoch": 1.53,
      "grad_norm": 1.0153558254241943,
      "learning_rate": 0.0002900392668614619,
      "loss": 2.0352,
      "step": 117637
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.9180485606193542,
      "learning_rate": 0.0002900351785555604,
      "loss": 2.0622,
      "step": 117638
    },
    {
      "epoch": 1.53,
      "grad_norm": 1.105238437652588,
      "learning_rate": 0.00029003109025151167,
      "loss": 2.3797,
      "step": 117639
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.9588298201560974,
      "learning_rate": 0.0002900270019493162,
      "loss": 1.8972,
      "step": 117640
    },
    {
      "epoch": 1.53,
      "grad_norm": 1.0262277126312256,
      "learning_rate": 0.0002900229136489749,
      "loss": 2.1267,
      "step": 117641
    },
    {
      "epoch": 1.53,
      "grad_norm": 1.1152225732803345,
      "learning_rate": 0.00029001882535048854,
      "loss": 2.351,
      "step": 117642
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.9141378998756409,
      "learning_rate": 0.00029001473705385784,
      "loss": 2.2671,
      "step": 117643
    },
    {
      "epoch": 1.53,
      "grad_norm": 1.149311900138855,
      "learning_rate": 0.0002900106487590836,
      "loss": 2.0407,
      "step": 117644
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.8608191013336182,
      "learning_rate": 0.00029000656046616674,
      "loss": 2.3452,
      "step": 117645
    },
    {
      "epoch": 1.53,
      "grad_norm": 1.0105305910110474,
      "learning_rate": 0.0002900024721751077,
      "loss": 2.1288,
      "step": 117646
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.9652996063232422,
      "learning_rate": 0.0002899983838859074,
      "loss": 2.2504,
      "step": 117647
    },
    {
      "epoch": 1.53,
      "grad_norm": 1.226508378982544,
      "learning_rate": 0.0002899942955985666,
      "loss": 1.97,
      "step": 117648
    },
    {
      "epoch": 1.53,
      "grad_norm": 1.1582510471343994,
      "learning_rate": 0.000289990207313086,
      "loss": 2.0827,
      "step": 117649
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.9712188243865967,
      "learning_rate": 0.0002899861190294664,
      "loss": 2.1168,
      "step": 117650
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.9793881177902222,
      "learning_rate": 0.00028998203074770873,
      "loss": 2.1055,
      "step": 117651
    },
    {
      "epoch": 1.53,
      "grad_norm": 1.1426596641540527,
      "learning_rate": 0.00028997794246781345,
      "loss": 2.1136,
      "step": 117652
    },
    {
      "epoch": 1.53,
      "grad_norm": 1.4615623950958252,
      "learning_rate": 0.00028997385418978147,
      "loss": 2.0008,
      "step": 117653
    },
    {
      "epoch": 1.53,
      "grad_norm": 1.0381640195846558,
      "learning_rate": 0.0002899697659136135,
      "loss": 2.1448,
      "step": 117654
    },
    {
      "epoch": 1.53,
      "grad_norm": 1.0886962413787842,
      "learning_rate": 0.00028996567763931043,
      "loss": 2.2388,
      "step": 117655
    },
    {
      "epoch": 1.53,
      "grad_norm": 1.1813263893127441,
      "learning_rate": 0.00028996158936687283,
      "loss": 2.2093,
      "step": 117656
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.9062221050262451,
      "learning_rate": 0.0002899575010963018,
      "loss": 2.1656,
      "step": 117657
    },
    {
      "epoch": 1.53,
      "grad_norm": 1.0673532485961914,
      "learning_rate": 0.00028995341282759765,
      "loss": 2.1226,
      "step": 117658
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.9115721583366394,
      "learning_rate": 0.00028994932456076136,
      "loss": 2.2301,
      "step": 117659
    },
    {
      "epoch": 1.53,
      "grad_norm": 1.3431427478790283,
      "learning_rate": 0.0002899452362957937,
      "loss": 2.1215,
      "step": 117660
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.9887726306915283,
      "learning_rate": 0.0002899411480326954,
      "loss": 2.1257,
      "step": 117661
    },
    {
      "epoch": 1.53,
      "grad_norm": 1.005534291267395,
      "learning_rate": 0.0002899370597714672,
      "loss": 2.1124,
      "step": 117662
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.8968374133110046,
      "learning_rate": 0.00028993297151211003,
      "loss": 1.8155,
      "step": 117663
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.9816615581512451,
      "learning_rate": 0.00028992888325462445,
      "loss": 1.8641,
      "step": 117664
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.9131454825401306,
      "learning_rate": 0.0002899247949990112,
      "loss": 2.1161,
      "step": 117665
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.9646140933036804,
      "learning_rate": 0.0002899207067452712,
      "loss": 1.9555,
      "step": 117666
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.8893094062805176,
      "learning_rate": 0.00028991661849340507,
      "loss": 2.1414,
      "step": 117667
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.9775452017784119,
      "learning_rate": 0.00028991253024341366,
      "loss": 2.334,
      "step": 117668
    },
    {
      "epoch": 1.53,
      "grad_norm": 1.1329034566879272,
      "learning_rate": 0.0002899084419952977,
      "loss": 2.1538,
      "step": 117669
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.954555869102478,
      "learning_rate": 0.00028990435374905805,
      "loss": 1.9498,
      "step": 117670
    },
    {
      "epoch": 1.53,
      "grad_norm": 1.0276832580566406,
      "learning_rate": 0.0002899002655046952,
      "loss": 2.0396,
      "step": 117671
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.8339504599571228,
      "learning_rate": 0.00028989617726221016,
      "loss": 1.825,
      "step": 117672
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.9242662787437439,
      "learning_rate": 0.0002898920890216036,
      "loss": 2.0677,
      "step": 117673
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.9519561529159546,
      "learning_rate": 0.0002898880007828763,
      "loss": 2.2536,
      "step": 117674
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.9341529011726379,
      "learning_rate": 0.00028988391254602903,
      "loss": 2.1085,
      "step": 117675
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.904599666595459,
      "learning_rate": 0.0002898798243110625,
      "loss": 2.0979,
      "step": 117676
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.9812193512916565,
      "learning_rate": 0.0002898757360779776,
      "loss": 2.3207,
      "step": 117677
    },
    {
      "epoch": 1.53,
      "grad_norm": 1.0434818267822266,
      "learning_rate": 0.00028987164784677485,
      "loss": 2.1663,
      "step": 117678
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.979828953742981,
      "learning_rate": 0.00028986755961745515,
      "loss": 2.0027,
      "step": 117679
    },
    {
      "epoch": 1.53,
      "grad_norm": 1.125697374343872,
      "learning_rate": 0.0002898634713900193,
      "loss": 2.0641,
      "step": 117680
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.900398850440979,
      "learning_rate": 0.00028985938316446815,
      "loss": 2.2783,
      "step": 117681
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.9787534475326538,
      "learning_rate": 0.0002898552949408022,
      "loss": 2.4429,
      "step": 117682
    },
    {
      "epoch": 1.53,
      "grad_norm": 1.1532472372055054,
      "learning_rate": 0.0002898512067190223,
      "loss": 2.3335,
      "step": 117683
    },
    {
      "epoch": 1.53,
      "grad_norm": 1.0933620929718018,
      "learning_rate": 0.0002898471184991293,
      "loss": 2.4753,
      "step": 117684
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.9294008016586304,
      "learning_rate": 0.00028984303028112404,
      "loss": 2.1607,
      "step": 117685
    },
    {
      "epoch": 1.53,
      "grad_norm": 1.0147860050201416,
      "learning_rate": 0.000289838942065007,
      "loss": 1.9847,
      "step": 117686
    },
    {
      "epoch": 1.53,
      "grad_norm": 1.083421230316162,
      "learning_rate": 0.0002898348538507792,
      "loss": 2.2336,
      "step": 117687
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.9536251425743103,
      "learning_rate": 0.0002898307656384412,
      "loss": 2.294,
      "step": 117688
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.9846324920654297,
      "learning_rate": 0.0002898266774279938,
      "loss": 2.0722,
      "step": 117689
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.9326333999633789,
      "learning_rate": 0.0002898225892194379,
      "loss": 2.0782,
      "step": 117690
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.9870004057884216,
      "learning_rate": 0.00028981850101277414,
      "loss": 2.2211,
      "step": 117691
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.8924468755722046,
      "learning_rate": 0.0002898144128080034,
      "loss": 2.2541,
      "step": 117692
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.9608986973762512,
      "learning_rate": 0.00028981032460512635,
      "loss": 2.0356,
      "step": 117693
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.8770058155059814,
      "learning_rate": 0.00028980623640414364,
      "loss": 2.0321,
      "step": 117694
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.886440634727478,
      "learning_rate": 0.0002898021482050562,
      "loss": 2.1714,
      "step": 117695
    },
    {
      "epoch": 1.53,
      "grad_norm": 1.0795457363128662,
      "learning_rate": 0.00028979806000786474,
      "loss": 2.1334,
      "step": 117696
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.8844987750053406,
      "learning_rate": 0.00028979397181257,
      "loss": 2.3076,
      "step": 117697
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.899537980556488,
      "learning_rate": 0.0002897898836191727,
      "loss": 2.0125,
      "step": 117698
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.8923147320747375,
      "learning_rate": 0.00028978579542767385,
      "loss": 2.23,
      "step": 117699
    },
    {
      "epoch": 1.53,
      "grad_norm": 1.0124261379241943,
      "learning_rate": 0.00028978170723807385,
      "loss": 2.0247,
      "step": 117700
    },
    {
      "epoch": 1.53,
      "grad_norm": 1.071688175201416,
      "learning_rate": 0.00028977761905037364,
      "loss": 2.2513,
      "step": 117701
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.878537118434906,
      "learning_rate": 0.00028977353086457393,
      "loss": 2.1052,
      "step": 117702
    },
    {
      "epoch": 1.53,
      "grad_norm": 1.1250240802764893,
      "learning_rate": 0.00028976944268067553,
      "loss": 1.9533,
      "step": 117703
    },
    {
      "epoch": 1.53,
      "grad_norm": 1.0418426990509033,
      "learning_rate": 0.0002897653544986792,
      "loss": 2.0095,
      "step": 117704
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.9550400972366333,
      "learning_rate": 0.0002897612663185857,
      "loss": 1.8818,
      "step": 117705
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.9618210792541504,
      "learning_rate": 0.00028975717814039585,
      "loss": 2.1357,
      "step": 117706
    },
    {
      "epoch": 1.53,
      "grad_norm": 1.0174168348312378,
      "learning_rate": 0.00028975308996411023,
      "loss": 2.0519,
      "step": 117707
    },
    {
      "epoch": 1.53,
      "grad_norm": 1.206432819366455,
      "learning_rate": 0.0002897490017897297,
      "loss": 2.2085,
      "step": 117708
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.9688854813575745,
      "learning_rate": 0.0002897449136172551,
      "loss": 2.2608,
      "step": 117709
    },
    {
      "epoch": 1.53,
      "grad_norm": 1.0784497261047363,
      "learning_rate": 0.000289740825446687,
      "loss": 2.3913,
      "step": 117710
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.9534687995910645,
      "learning_rate": 0.0002897367372780263,
      "loss": 2.1349,
      "step": 117711
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.9032623171806335,
      "learning_rate": 0.0002897326491112739,
      "loss": 2.0214,
      "step": 117712
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.9611330032348633,
      "learning_rate": 0.00028972856094643026,
      "loss": 2.0133,
      "step": 117713
    },
    {
      "epoch": 1.53,
      "grad_norm": 1.0052776336669922,
      "learning_rate": 0.00028972447278349626,
      "loss": 2.2756,
      "step": 117714
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.9363980293273926,
      "learning_rate": 0.00028972038462247264,
      "loss": 2.0011,
      "step": 117715
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.8830854892730713,
      "learning_rate": 0.00028971629646336024,
      "loss": 2.0553,
      "step": 117716
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.9688203930854797,
      "learning_rate": 0.0002897122083061598,
      "loss": 2.0188,
      "step": 117717
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.9635658860206604,
      "learning_rate": 0.0002897081201508721,
      "loss": 2.2622,
      "step": 117718
    },
    {
      "epoch": 1.53,
      "grad_norm": 1.053064227104187,
      "learning_rate": 0.0002897040319974978,
      "loss": 1.9552,
      "step": 117719
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.9398530125617981,
      "learning_rate": 0.0002896999438460377,
      "loss": 2.1466,
      "step": 117720
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.9547330737113953,
      "learning_rate": 0.00028969585569649253,
      "loss": 2.3384,
      "step": 117721
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.9004135727882385,
      "learning_rate": 0.00028969176754886313,
      "loss": 1.9594,
      "step": 117722
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.9593136310577393,
      "learning_rate": 0.00028968767940315023,
      "loss": 2.1148,
      "step": 117723
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.9119030833244324,
      "learning_rate": 0.00028968359125935473,
      "loss": 2.2379,
      "step": 117724
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.9553315043449402,
      "learning_rate": 0.0002896795031174771,
      "loss": 2.033,
      "step": 117725
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.8822982907295227,
      "learning_rate": 0.0002896754149775182,
      "loss": 2.0848,
      "step": 117726
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.9503462314605713,
      "learning_rate": 0.0002896713268394789,
      "loss": 2.1642,
      "step": 117727
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.917311429977417,
      "learning_rate": 0.0002896672387033599,
      "loss": 2.0887,
      "step": 117728
    },
    {
      "epoch": 1.53,
      "grad_norm": 1.0969401597976685,
      "learning_rate": 0.0002896631505691619,
      "loss": 2.107,
      "step": 117729
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.9000173807144165,
      "learning_rate": 0.00028965906243688585,
      "loss": 2.1288,
      "step": 117730
    },
    {
      "epoch": 1.53,
      "grad_norm": 1.012045979499817,
      "learning_rate": 0.0002896549743065323,
      "loss": 2.4632,
      "step": 117731
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.9598989486694336,
      "learning_rate": 0.00028965088617810206,
      "loss": 1.8818,
      "step": 117732
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.9008728265762329,
      "learning_rate": 0.0002896467980515959,
      "loss": 2.0029,
      "step": 117733
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.9040150046348572,
      "learning_rate": 0.0002896427099270146,
      "loss": 2.0461,
      "step": 117734
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.9499616026878357,
      "learning_rate": 0.0002896386218043589,
      "loss": 2.0004,
      "step": 117735
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.9626784324645996,
      "learning_rate": 0.0002896345336836297,
      "loss": 2.2281,
      "step": 117736
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.9929996728897095,
      "learning_rate": 0.0002896304455648276,
      "loss": 2.1716,
      "step": 117737
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.9669563174247742,
      "learning_rate": 0.0002896263574479533,
      "loss": 2.3887,
      "step": 117738
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.9950084090232849,
      "learning_rate": 0.0002896222693330077,
      "loss": 2.022,
      "step": 117739
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.9232052564620972,
      "learning_rate": 0.0002896181812199915,
      "loss": 2.0853,
      "step": 117740
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.9629387855529785,
      "learning_rate": 0.00028961409310890546,
      "loss": 2.2133,
      "step": 117741
    },
    {
      "epoch": 1.53,
      "grad_norm": 1.1941581964492798,
      "learning_rate": 0.0002896100049997505,
      "loss": 2.2696,
      "step": 117742
    },
    {
      "epoch": 1.53,
      "grad_norm": 1.0487074851989746,
      "learning_rate": 0.0002896059168925271,
      "loss": 2.1979,
      "step": 117743
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.8801009654998779,
      "learning_rate": 0.00028960182878723617,
      "loss": 2.0382,
      "step": 117744
    },
    {
      "epoch": 1.53,
      "grad_norm": 1.0106836557388306,
      "learning_rate": 0.00028959774068387843,
      "loss": 1.9874,
      "step": 117745
    },
    {
      "epoch": 1.53,
      "grad_norm": 1.0363409519195557,
      "learning_rate": 0.0002895936525824547,
      "loss": 2.3004,
      "step": 117746
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.9721844792366028,
      "learning_rate": 0.00028958956448296567,
      "loss": 2.2768,
      "step": 117747
    },
    {
      "epoch": 1.53,
      "grad_norm": 1.113527536392212,
      "learning_rate": 0.0002895854763854123,
      "loss": 2.1561,
      "step": 117748
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.9702431559562683,
      "learning_rate": 0.00028958138828979503,
      "loss": 2.1184,
      "step": 117749
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.8696714043617249,
      "learning_rate": 0.00028957730019611483,
      "loss": 2.1585,
      "step": 117750
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.9650760889053345,
      "learning_rate": 0.00028957321210437237,
      "loss": 2.187,
      "step": 117751
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.906862735748291,
      "learning_rate": 0.0002895691240145684,
      "loss": 2.3679,
      "step": 117752
    },
    {
      "epoch": 1.53,
      "grad_norm": 1.0142252445220947,
      "learning_rate": 0.0002895650359267038,
      "loss": 2.1349,
      "step": 117753
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.8766071200370789,
      "learning_rate": 0.0002895609478407793,
      "loss": 2.1063,
      "step": 117754
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.9173474907875061,
      "learning_rate": 0.0002895568597567956,
      "loss": 2.115,
      "step": 117755
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.9061199426651001,
      "learning_rate": 0.00028955277167475344,
      "loss": 2.1552,
      "step": 117756
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.9831379652023315,
      "learning_rate": 0.0002895486835946536,
      "loss": 2.1202,
      "step": 117757
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.9618110656738281,
      "learning_rate": 0.0002895445955164968,
      "loss": 2.2648,
      "step": 117758
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.948966920375824,
      "learning_rate": 0.00028954050744028393,
      "loss": 2.0506,
      "step": 117759
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.9176048040390015,
      "learning_rate": 0.0002895364193660157,
      "loss": 2.3182,
      "step": 117760
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.9950188994407654,
      "learning_rate": 0.00028953233129369284,
      "loss": 2.4014,
      "step": 117761
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.9896925687789917,
      "learning_rate": 0.00028952824322331616,
      "loss": 2.0086,
      "step": 117762
    },
    {
      "epoch": 1.53,
      "grad_norm": 1.288526177406311,
      "learning_rate": 0.0002895241551548863,
      "loss": 1.9356,
      "step": 117763
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.9457699060440063,
      "learning_rate": 0.00028952006708840415,
      "loss": 2.2161,
      "step": 117764
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.8778261542320251,
      "learning_rate": 0.00028951597902387034,
      "loss": 2.1016,
      "step": 117765
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.948401689529419,
      "learning_rate": 0.00028951189096128576,
      "loss": 2.157,
      "step": 117766
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.9573398232460022,
      "learning_rate": 0.0002895078029006511,
      "loss": 2.3025,
      "step": 117767
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.9835798144340515,
      "learning_rate": 0.0002895037148419671,
      "loss": 2.1,
      "step": 117768
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.9707708358764648,
      "learning_rate": 0.0002894996267852347,
      "loss": 2.0237,
      "step": 117769
    },
    {
      "epoch": 1.53,
      "grad_norm": 1.1733977794647217,
      "learning_rate": 0.0002894955387304544,
      "loss": 2.1946,
      "step": 117770
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.9075498580932617,
      "learning_rate": 0.00028949145067762706,
      "loss": 2.2081,
      "step": 117771
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.9481304883956909,
      "learning_rate": 0.0002894873626267535,
      "loss": 2.2308,
      "step": 117772
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.8989781141281128,
      "learning_rate": 0.0002894832745778345,
      "loss": 2.1916,
      "step": 117773
    },
    {
      "epoch": 1.53,
      "grad_norm": 1.0033419132232666,
      "learning_rate": 0.0002894791865308707,
      "loss": 2.1241,
      "step": 117774
    },
    {
      "epoch": 1.53,
      "grad_norm": 1.006685733795166,
      "learning_rate": 0.0002894750984858629,
      "loss": 2.1102,
      "step": 117775
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.9376763701438904,
      "learning_rate": 0.0002894710104428119,
      "loss": 2.2619,
      "step": 117776
    },
    {
      "epoch": 1.53,
      "grad_norm": 1.275508999824524,
      "learning_rate": 0.0002894669224017184,
      "loss": 1.9471,
      "step": 117777
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.9359796643257141,
      "learning_rate": 0.0002894628343625832,
      "loss": 2.269,
      "step": 117778
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.9579818844795227,
      "learning_rate": 0.0002894587463254071,
      "loss": 2.2346,
      "step": 117779
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.9740646481513977,
      "learning_rate": 0.00028945465829019083,
      "loss": 1.905,
      "step": 117780
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.9668077230453491,
      "learning_rate": 0.00028945057025693506,
      "loss": 2.1602,
      "step": 117781
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.933273434638977,
      "learning_rate": 0.0002894464822256407,
      "loss": 2.3041,
      "step": 117782
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.9503107666969299,
      "learning_rate": 0.0002894423941963084,
      "loss": 2.2011,
      "step": 117783
    },
    {
      "epoch": 1.53,
      "grad_norm": 1.0547711849212646,
      "learning_rate": 0.000289438306168939,
      "loss": 2.1028,
      "step": 117784
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.9769657850265503,
      "learning_rate": 0.0002894342181435332,
      "loss": 2.0477,
      "step": 117785
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.9163011908531189,
      "learning_rate": 0.0002894301301200918,
      "loss": 2.1468,
      "step": 117786
    },
    {
      "epoch": 1.53,
      "grad_norm": 1.0041636228561401,
      "learning_rate": 0.00028942604209861546,
      "loss": 2.1518,
      "step": 117787
    },
    {
      "epoch": 1.53,
      "grad_norm": 1.0157041549682617,
      "learning_rate": 0.00028942195407910505,
      "loss": 2.3251,
      "step": 117788
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.9577701091766357,
      "learning_rate": 0.0002894178660615613,
      "loss": 2.1822,
      "step": 117789
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.961152195930481,
      "learning_rate": 0.000289413778045985,
      "loss": 2.1589,
      "step": 117790
    },
    {
      "epoch": 1.53,
      "grad_norm": 1.0364160537719727,
      "learning_rate": 0.00028940969003237696,
      "loss": 1.9875,
      "step": 117791
    },
    {
      "epoch": 1.53,
      "grad_norm": 1.0923399925231934,
      "learning_rate": 0.00028940560202073774,
      "loss": 2.0849,
      "step": 117792
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.9588646292686462,
      "learning_rate": 0.00028940151401106825,
      "loss": 1.9583,
      "step": 117793
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.8725842833518982,
      "learning_rate": 0.0002893974260033692,
      "loss": 2.173,
      "step": 117794
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.8597429394721985,
      "learning_rate": 0.0002893933379976413,
      "loss": 1.9639,
      "step": 117795
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.8712998032569885,
      "learning_rate": 0.0002893892499938855,
      "loss": 2.3363,
      "step": 117796
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.9741283059120178,
      "learning_rate": 0.0002893851619921025,
      "loss": 2.3083,
      "step": 117797
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.9092571139335632,
      "learning_rate": 0.00028938107399229283,
      "loss": 2.3243,
      "step": 117798
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.8655657768249512,
      "learning_rate": 0.00028937698599445746,
      "loss": 2.1251,
      "step": 117799
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.9871794581413269,
      "learning_rate": 0.0002893728979985971,
      "loss": 2.2124,
      "step": 117800
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.9272499680519104,
      "learning_rate": 0.0002893688100047126,
      "loss": 2.246,
      "step": 117801
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.9083631038665771,
      "learning_rate": 0.00028936472201280454,
      "loss": 2.3009,
      "step": 117802
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.9006779193878174,
      "learning_rate": 0.0002893606340228739,
      "loss": 2.2309,
      "step": 117803
    },
    {
      "epoch": 1.53,
      "grad_norm": 1.0908364057540894,
      "learning_rate": 0.00028935654603492127,
      "loss": 2.0872,
      "step": 117804
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.9731150269508362,
      "learning_rate": 0.0002893524580489474,
      "loss": 2.0886,
      "step": 117805
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.8968086838722229,
      "learning_rate": 0.0002893483700649531,
      "loss": 2.1313,
      "step": 117806
    },
    {
      "epoch": 1.53,
      "grad_norm": 1.063637614250183,
      "learning_rate": 0.0002893442820829392,
      "loss": 2.216,
      "step": 117807
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.8964577317237854,
      "learning_rate": 0.00028934019410290635,
      "loss": 2.1336,
      "step": 117808
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.9456090927124023,
      "learning_rate": 0.0002893361061248555,
      "loss": 2.038,
      "step": 117809
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.9601486325263977,
      "learning_rate": 0.00028933201814878716,
      "loss": 2.1483,
      "step": 117810
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.9828355312347412,
      "learning_rate": 0.00028932793017470216,
      "loss": 1.952,
      "step": 117811
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.9196446537971497,
      "learning_rate": 0.00028932384220260135,
      "loss": 2.2256,
      "step": 117812
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.8966827392578125,
      "learning_rate": 0.0002893197542324854,
      "loss": 2.1314,
      "step": 117813
    },
    {
      "epoch": 1.53,
      "grad_norm": 1.086628794670105,
      "learning_rate": 0.0002893156662643551,
      "loss": 2.1398,
      "step": 117814
    },
    {
      "epoch": 1.53,
      "grad_norm": 1.068017840385437,
      "learning_rate": 0.0002893115782982114,
      "loss": 2.4798,
      "step": 117815
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.8928968906402588,
      "learning_rate": 0.0002893074903340547,
      "loss": 1.9227,
      "step": 117816
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.8712006211280823,
      "learning_rate": 0.00028930340237188595,
      "loss": 2.1118,
      "step": 117817
    },
    {
      "epoch": 1.53,
      "grad_norm": 1.1391253471374512,
      "learning_rate": 0.0002892993144117059,
      "loss": 1.9947,
      "step": 117818
    },
    {
      "epoch": 1.53,
      "grad_norm": 1.0534377098083496,
      "learning_rate": 0.00028929522645351535,
      "loss": 2.0015,
      "step": 117819
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.9500736594200134,
      "learning_rate": 0.00028929113849731497,
      "loss": 2.189,
      "step": 117820
    },
    {
      "epoch": 1.53,
      "grad_norm": 1.097718596458435,
      "learning_rate": 0.0002892870505431057,
      "loss": 2.2215,
      "step": 117821
    },
    {
      "epoch": 1.53,
      "grad_norm": 1.054103970527649,
      "learning_rate": 0.00028928296259088805,
      "loss": 2.0355,
      "step": 117822
    },
    {
      "epoch": 1.53,
      "grad_norm": 1.0019614696502686,
      "learning_rate": 0.0002892788746406629,
      "loss": 2.2161,
      "step": 117823
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.9007226228713989,
      "learning_rate": 0.00028927478669243104,
      "loss": 2.2502,
      "step": 117824
    },
    {
      "epoch": 1.53,
      "grad_norm": 1.1977639198303223,
      "learning_rate": 0.0002892706987461932,
      "loss": 2.0067,
      "step": 117825
    },
    {
      "epoch": 1.53,
      "grad_norm": 1.1045950651168823,
      "learning_rate": 0.0002892666108019501,
      "loss": 2.3233,
      "step": 117826
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.9656922817230225,
      "learning_rate": 0.0002892625228597027,
      "loss": 2.1207,
      "step": 117827
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.9780942797660828,
      "learning_rate": 0.00028925843491945145,
      "loss": 1.9566,
      "step": 117828
    },
    {
      "epoch": 1.53,
      "grad_norm": 1.0902031660079956,
      "learning_rate": 0.0002892543469811972,
      "loss": 2.2886,
      "step": 117829
    },
    {
      "epoch": 1.53,
      "grad_norm": 1.1418241262435913,
      "learning_rate": 0.00028925025904494084,
      "loss": 2.0816,
      "step": 117830
    },
    {
      "epoch": 1.53,
      "grad_norm": 1.069710373878479,
      "learning_rate": 0.0002892461711106831,
      "loss": 1.9769,
      "step": 117831
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.9111249446868896,
      "learning_rate": 0.00028924208317842464,
      "loss": 2.0336,
      "step": 117832
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.9994019865989685,
      "learning_rate": 0.0002892379952481664,
      "loss": 2.0374,
      "step": 117833
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.9493325352668762,
      "learning_rate": 0.00028923390731990894,
      "loss": 2.0773,
      "step": 117834
    },
    {
      "epoch": 1.53,
      "grad_norm": 1.0656167268753052,
      "learning_rate": 0.00028922981939365303,
      "loss": 1.8508,
      "step": 117835
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.8508229851722717,
      "learning_rate": 0.00028922573146939953,
      "loss": 2.1331,
      "step": 117836
    },
    {
      "epoch": 1.53,
      "grad_norm": 1.1669921875,
      "learning_rate": 0.0002892216435471492,
      "loss": 2.0971,
      "step": 117837
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.9694774746894836,
      "learning_rate": 0.00028921755562690276,
      "loss": 2.4153,
      "step": 117838
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.9152500033378601,
      "learning_rate": 0.000289213467708661,
      "loss": 2.3174,
      "step": 117839
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.9021639823913574,
      "learning_rate": 0.00028920937979242474,
      "loss": 2.0362,
      "step": 117840
    },
    {
      "epoch": 1.53,
      "grad_norm": 1.142084002494812,
      "learning_rate": 0.0002892052918781945,
      "loss": 2.0134,
      "step": 117841
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.9752426147460938,
      "learning_rate": 0.0002892012039659713,
      "loss": 2.2822,
      "step": 117842
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.9550935626029968,
      "learning_rate": 0.00028919711605575576,
      "loss": 2.3033,
      "step": 117843
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.8862231969833374,
      "learning_rate": 0.00028919302814754867,
      "loss": 2.1241,
      "step": 117844
    },
    {
      "epoch": 1.53,
      "grad_norm": 1.505537748336792,
      "learning_rate": 0.00028918894024135077,
      "loss": 2.0803,
      "step": 117845
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.9585626721382141,
      "learning_rate": 0.000289184852337163,
      "loss": 2.245,
      "step": 117846
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.9506593346595764,
      "learning_rate": 0.0002891807644349859,
      "loss": 2.1637,
      "step": 117847
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.913287341594696,
      "learning_rate": 0.00028917667653482023,
      "loss": 2.2338,
      "step": 117848
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.9148783683776855,
      "learning_rate": 0.00028917258863666684,
      "loss": 2.2499,
      "step": 117849
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.9080713987350464,
      "learning_rate": 0.0002891685007405265,
      "loss": 2.1645,
      "step": 117850
    },
    {
      "epoch": 1.53,
      "grad_norm": 1.0000383853912354,
      "learning_rate": 0.0002891644128463999,
      "loss": 2.0946,
      "step": 117851
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.9424493312835693,
      "learning_rate": 0.00028916032495428793,
      "loss": 2.1113,
      "step": 117852
    },
    {
      "epoch": 1.53,
      "grad_norm": 1.0465502738952637,
      "learning_rate": 0.00028915623706419113,
      "loss": 2.221,
      "step": 117853
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.9866189956665039,
      "learning_rate": 0.0002891521491761105,
      "loss": 2.0985,
      "step": 117854
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.9442834854125977,
      "learning_rate": 0.00028914806129004664,
      "loss": 2.0372,
      "step": 117855
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.890769898891449,
      "learning_rate": 0.00028914397340600034,
      "loss": 2.1048,
      "step": 117856
    },
    {
      "epoch": 1.53,
      "grad_norm": 1.170501470565796,
      "learning_rate": 0.0002891398855239724,
      "loss": 1.8953,
      "step": 117857
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.9357908964157104,
      "learning_rate": 0.00028913579764396367,
      "loss": 2.0523,
      "step": 117858
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.9145799875259399,
      "learning_rate": 0.00028913170976597465,
      "loss": 2.117,
      "step": 117859
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.9112396836280823,
      "learning_rate": 0.00028912762189000626,
      "loss": 1.9782,
      "step": 117860
    },
    {
      "epoch": 1.53,
      "grad_norm": 1.101732850074768,
      "learning_rate": 0.00028912353401605936,
      "loss": 2.0754,
      "step": 117861
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.8879270553588867,
      "learning_rate": 0.00028911944614413456,
      "loss": 2.1258,
      "step": 117862
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.9601367115974426,
      "learning_rate": 0.0002891153582742326,
      "loss": 2.2713,
      "step": 117863
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.9784584045410156,
      "learning_rate": 0.00028911127040635436,
      "loss": 1.9815,
      "step": 117864
    },
    {
      "epoch": 1.53,
      "grad_norm": 1.0242482423782349,
      "learning_rate": 0.0002891071825405005,
      "loss": 2.2386,
      "step": 117865
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.8646818399429321,
      "learning_rate": 0.0002891030946766718,
      "loss": 2.002,
      "step": 117866
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.9394509792327881,
      "learning_rate": 0.00028909900681486905,
      "loss": 2.3076,
      "step": 117867
    },
    {
      "epoch": 1.53,
      "grad_norm": 1.0010663270950317,
      "learning_rate": 0.0002890949189550931,
      "loss": 2.0791,
      "step": 117868
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.9050945043563843,
      "learning_rate": 0.00028909083109734447,
      "loss": 2.108,
      "step": 117869
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.9747453927993774,
      "learning_rate": 0.0002890867432416242,
      "loss": 2.132,
      "step": 117870
    },
    {
      "epoch": 1.53,
      "grad_norm": 1.0317775011062622,
      "learning_rate": 0.0002890826553879328,
      "loss": 2.2066,
      "step": 117871
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.9571539759635925,
      "learning_rate": 0.0002890785675362712,
      "loss": 1.9191,
      "step": 117872
    },
    {
      "epoch": 1.53,
      "grad_norm": 1.0109256505966187,
      "learning_rate": 0.00028907447968664,
      "loss": 2.1896,
      "step": 117873
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.898690402507782,
      "learning_rate": 0.00028907039183904016,
      "loss": 1.9613,
      "step": 117874
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.9239035248756409,
      "learning_rate": 0.00028906630399347225,
      "loss": 2.0941,
      "step": 117875
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.9376007914543152,
      "learning_rate": 0.00028906221614993734,
      "loss": 2.1952,
      "step": 117876
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.8959106206893921,
      "learning_rate": 0.0002890581283084358,
      "loss": 2.0778,
      "step": 117877
    },
    {
      "epoch": 1.53,
      "grad_norm": 1.039376139640808,
      "learning_rate": 0.0002890540404689686,
      "loss": 2.3519,
      "step": 117878
    },
    {
      "epoch": 1.53,
      "grad_norm": 1.0076439380645752,
      "learning_rate": 0.0002890499526315364,
      "loss": 1.9421,
      "step": 117879
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.8823529481887817,
      "learning_rate": 0.00028904586479614006,
      "loss": 2.0486,
      "step": 117880
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.9220340251922607,
      "learning_rate": 0.00028904177696278027,
      "loss": 2.1969,
      "step": 117881
    },
    {
      "epoch": 1.53,
      "grad_norm": 1.0033808946609497,
      "learning_rate": 0.00028903768913145797,
      "loss": 2.1243,
      "step": 117882
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.9563151001930237,
      "learning_rate": 0.00028903360130217364,
      "loss": 2.2346,
      "step": 117883
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.9757233262062073,
      "learning_rate": 0.00028902951347492816,
      "loss": 2.3312,
      "step": 117884
    },
    {
      "epoch": 1.53,
      "grad_norm": 1.0176748037338257,
      "learning_rate": 0.0002890254256497223,
      "loss": 2.177,
      "step": 117885
    },
    {
      "epoch": 1.53,
      "grad_norm": 1.013262152671814,
      "learning_rate": 0.0002890213378265568,
      "loss": 2.1669,
      "step": 117886
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.9809033274650574,
      "learning_rate": 0.0002890172500054325,
      "loss": 2.0188,
      "step": 117887
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.9113138914108276,
      "learning_rate": 0.0002890131621863502,
      "loss": 2.1421,
      "step": 117888
    },
    {
      "epoch": 1.53,
      "grad_norm": 0.9201117753982544,
      "learning_rate": 0.00028900907436931036,
      "loss": 2.269,
      "step": 117889
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.9906543493270874,
      "learning_rate": 0.00028900498655431403,
      "loss": 2.2602,
      "step": 117890
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.9395939111709595,
      "learning_rate": 0.0002890008987413619,
      "loss": 2.1375,
      "step": 117891
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.9579218029975891,
      "learning_rate": 0.00028899681093045465,
      "loss": 2.0556,
      "step": 117892
    },
    {
      "epoch": 1.54,
      "grad_norm": 1.0615439414978027,
      "learning_rate": 0.0002889927231215931,
      "loss": 2.1815,
      "step": 117893
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.9098852276802063,
      "learning_rate": 0.00028898863531477817,
      "loss": 2.2209,
      "step": 117894
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.928439736366272,
      "learning_rate": 0.0002889845475100103,
      "loss": 2.116,
      "step": 117895
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.9552861452102661,
      "learning_rate": 0.00028898045970729045,
      "loss": 2.1157,
      "step": 117896
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.9252915382385254,
      "learning_rate": 0.0002889763719066193,
      "loss": 2.1112,
      "step": 117897
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.8807538151741028,
      "learning_rate": 0.0002889722841079977,
      "loss": 2.2171,
      "step": 117898
    },
    {
      "epoch": 1.54,
      "grad_norm": 1.0006940364837646,
      "learning_rate": 0.00028896819631142633,
      "loss": 2.1253,
      "step": 117899
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.8721975684165955,
      "learning_rate": 0.00028896410851690613,
      "loss": 2.1867,
      "step": 117900
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.9047293066978455,
      "learning_rate": 0.00028896002072443757,
      "loss": 2.1448,
      "step": 117901
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.8955811858177185,
      "learning_rate": 0.0002889559329340215,
      "loss": 2.0482,
      "step": 117902
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.9952215552330017,
      "learning_rate": 0.00028895184514565877,
      "loss": 2.181,
      "step": 117903
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.933993399143219,
      "learning_rate": 0.00028894775735935016,
      "loss": 2.1346,
      "step": 117904
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.908134937286377,
      "learning_rate": 0.00028894366957509627,
      "loss": 2.1139,
      "step": 117905
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.9268271327018738,
      "learning_rate": 0.00028893958179289803,
      "loss": 2.1325,
      "step": 117906
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.9123769998550415,
      "learning_rate": 0.00028893549401275625,
      "loss": 2.2958,
      "step": 117907
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.9988270998001099,
      "learning_rate": 0.0002889314062346714,
      "loss": 2.1856,
      "step": 117908
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.9969779253005981,
      "learning_rate": 0.0002889273184586444,
      "loss": 2.2139,
      "step": 117909
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.9596033096313477,
      "learning_rate": 0.0002889232306846761,
      "loss": 2.0902,
      "step": 117910
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.91230708360672,
      "learning_rate": 0.0002889191429127671,
      "loss": 2.0315,
      "step": 117911
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.9667572379112244,
      "learning_rate": 0.00028891505514291833,
      "loss": 2.1401,
      "step": 117912
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.934494137763977,
      "learning_rate": 0.00028891096737513055,
      "loss": 2.0812,
      "step": 117913
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.8990235328674316,
      "learning_rate": 0.00028890687960940426,
      "loss": 2.0004,
      "step": 117914
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.8196439146995544,
      "learning_rate": 0.0002889027918457404,
      "loss": 2.1196,
      "step": 117915
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.9050859212875366,
      "learning_rate": 0.0002888987040841398,
      "loss": 2.2462,
      "step": 117916
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.8735950589179993,
      "learning_rate": 0.00028889461632460305,
      "loss": 2.0934,
      "step": 117917
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.9101719260215759,
      "learning_rate": 0.000288890528567131,
      "loss": 2.1311,
      "step": 117918
    },
    {
      "epoch": 1.54,
      "grad_norm": 1.0055370330810547,
      "learning_rate": 0.0002888864408117246,
      "loss": 2.1384,
      "step": 117919
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.9478503465652466,
      "learning_rate": 0.00028888235305838425,
      "loss": 2.3459,
      "step": 117920
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.8526186347007751,
      "learning_rate": 0.0002888782653071109,
      "loss": 2.1752,
      "step": 117921
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.9893916249275208,
      "learning_rate": 0.0002888741775579053,
      "loss": 2.0686,
      "step": 117922
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.9375500679016113,
      "learning_rate": 0.0002888700898107682,
      "loss": 2.098,
      "step": 117923
    },
    {
      "epoch": 1.54,
      "grad_norm": 1.1317678689956665,
      "learning_rate": 0.0002888660020657003,
      "loss": 2.3678,
      "step": 117924
    },
    {
      "epoch": 1.54,
      "grad_norm": 1.0049535036087036,
      "learning_rate": 0.00028886191432270263,
      "loss": 2.0389,
      "step": 117925
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.9819265604019165,
      "learning_rate": 0.0002888578265817756,
      "loss": 2.1307,
      "step": 117926
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.95688796043396,
      "learning_rate": 0.00028885373884292005,
      "loss": 2.2177,
      "step": 117927
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.995090901851654,
      "learning_rate": 0.0002888496511061368,
      "loss": 2.1668,
      "step": 117928
    },
    {
      "epoch": 1.54,
      "grad_norm": 1.0263968706130981,
      "learning_rate": 0.0002888455633714267,
      "loss": 2.0836,
      "step": 117929
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.8557944297790527,
      "learning_rate": 0.00028884147563879037,
      "loss": 1.9719,
      "step": 117930
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.9635202288627625,
      "learning_rate": 0.0002888373879082287,
      "loss": 1.9673,
      "step": 117931
    },
    {
      "epoch": 1.54,
      "grad_norm": 1.0852521657943726,
      "learning_rate": 0.00028883330017974234,
      "loss": 2.1848,
      "step": 117932
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.9985684752464294,
      "learning_rate": 0.000288829212453332,
      "loss": 2.2966,
      "step": 117933
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.9892153739929199,
      "learning_rate": 0.00028882512472899855,
      "loss": 2.1925,
      "step": 117934
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.9773222208023071,
      "learning_rate": 0.0002888210370067427,
      "loss": 1.9667,
      "step": 117935
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.9674146175384521,
      "learning_rate": 0.00028881694928656527,
      "loss": 2.121,
      "step": 117936
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.9574583172798157,
      "learning_rate": 0.000288812861568467,
      "loss": 2.1329,
      "step": 117937
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.9570763111114502,
      "learning_rate": 0.00028880877385244864,
      "loss": 2.1547,
      "step": 117938
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.8881240487098694,
      "learning_rate": 0.00028880468613851095,
      "loss": 2.0043,
      "step": 117939
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.9284123778343201,
      "learning_rate": 0.0002888005984266546,
      "loss": 2.2257,
      "step": 117940
    },
    {
      "epoch": 1.54,
      "grad_norm": 1.0113744735717773,
      "learning_rate": 0.0002887965107168804,
      "loss": 2.2092,
      "step": 117941
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.8907222747802734,
      "learning_rate": 0.0002887924230091892,
      "loss": 1.8699,
      "step": 117942
    },
    {
      "epoch": 1.54,
      "grad_norm": 1.0208455324172974,
      "learning_rate": 0.0002887883353035818,
      "loss": 2.3487,
      "step": 117943
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.9201020002365112,
      "learning_rate": 0.0002887842476000588,
      "loss": 2.1282,
      "step": 117944
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.9551456570625305,
      "learning_rate": 0.0002887801598986209,
      "loss": 2.0683,
      "step": 117945
    },
    {
      "epoch": 1.54,
      "grad_norm": 1.0315005779266357,
      "learning_rate": 0.0002887760721992691,
      "loss": 2.079,
      "step": 117946
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.916361927986145,
      "learning_rate": 0.000288771984502004,
      "loss": 2.2,
      "step": 117947
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.9135628342628479,
      "learning_rate": 0.0002887678968068264,
      "loss": 2.2278,
      "step": 117948
    },
    {
      "epoch": 1.54,
      "grad_norm": 1.0210886001586914,
      "learning_rate": 0.0002887638091137371,
      "loss": 2.2108,
      "step": 117949
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.9914944767951965,
      "learning_rate": 0.0002887597214227368,
      "loss": 2.2394,
      "step": 117950
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.9695029258728027,
      "learning_rate": 0.00028875563373382624,
      "loss": 2.1086,
      "step": 117951
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.8887091875076294,
      "learning_rate": 0.00028875154604700624,
      "loss": 2.1952,
      "step": 117952
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.9529106616973877,
      "learning_rate": 0.0002887474583622776,
      "loss": 2.1925,
      "step": 117953
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.9114023447036743,
      "learning_rate": 0.00028874337067964095,
      "loss": 2.2294,
      "step": 117954
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.9767338633537292,
      "learning_rate": 0.0002887392829990972,
      "loss": 2.1485,
      "step": 117955
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.9252930879592896,
      "learning_rate": 0.0002887351953206469,
      "loss": 2.037,
      "step": 117956
    },
    {
      "epoch": 1.54,
      "grad_norm": 1.003617763519287,
      "learning_rate": 0.000288731107644291,
      "loss": 2.0822,
      "step": 117957
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.9703925251960754,
      "learning_rate": 0.0002887270199700302,
      "loss": 2.3017,
      "step": 117958
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.9132310152053833,
      "learning_rate": 0.00028872293229786526,
      "loss": 2.0502,
      "step": 117959
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.9457920789718628,
      "learning_rate": 0.000288718844627797,
      "loss": 2.2214,
      "step": 117960
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.9887370467185974,
      "learning_rate": 0.0002887147569598261,
      "loss": 1.9137,
      "step": 117961
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.9186567068099976,
      "learning_rate": 0.0002887106692939533,
      "loss": 2.3185,
      "step": 117962
    },
    {
      "epoch": 1.54,
      "grad_norm": 1.0558596849441528,
      "learning_rate": 0.00028870658163017936,
      "loss": 2.0123,
      "step": 117963
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.9206326603889465,
      "learning_rate": 0.00028870249396850515,
      "loss": 2.1741,
      "step": 117964
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.9394171237945557,
      "learning_rate": 0.00028869840630893126,
      "loss": 2.0262,
      "step": 117965
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.9129916429519653,
      "learning_rate": 0.0002886943186514586,
      "loss": 2.1718,
      "step": 117966
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.9627467393875122,
      "learning_rate": 0.00028869023099608806,
      "loss": 2.13,
      "step": 117967
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.9006338715553284,
      "learning_rate": 0.00028868614334282006,
      "loss": 2.117,
      "step": 117968
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.9213574528694153,
      "learning_rate": 0.0002886820556916555,
      "loss": 2.0232,
      "step": 117969
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.9524508714675903,
      "learning_rate": 0.00028867796804259513,
      "loss": 2.3668,
      "step": 117970
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.9406381845474243,
      "learning_rate": 0.0002886738803956398,
      "loss": 2.2453,
      "step": 117971
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.975706934928894,
      "learning_rate": 0.00028866979275079017,
      "loss": 2.1136,
      "step": 117972
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.9305852651596069,
      "learning_rate": 0.00028866570510804705,
      "loss": 2.1882,
      "step": 117973
    },
    {
      "epoch": 1.54,
      "grad_norm": 1.021897315979004,
      "learning_rate": 0.00028866161746741135,
      "loss": 2.0405,
      "step": 117974
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.8910807967185974,
      "learning_rate": 0.0002886575298288835,
      "loss": 2.0004,
      "step": 117975
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.9135884642601013,
      "learning_rate": 0.00028865344219246444,
      "loss": 1.9582,
      "step": 117976
    },
    {
      "epoch": 1.54,
      "grad_norm": 1.0989537239074707,
      "learning_rate": 0.0002886493545581549,
      "loss": 2.31,
      "step": 117977
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.9736872911453247,
      "learning_rate": 0.0002886452669259557,
      "loss": 2.1707,
      "step": 117978
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.9805236458778381,
      "learning_rate": 0.0002886411792958675,
      "loss": 2.1229,
      "step": 117979
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.8731827139854431,
      "learning_rate": 0.00028863709166789133,
      "loss": 2.1497,
      "step": 117980
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.8575621843338013,
      "learning_rate": 0.00028863300404202756,
      "loss": 2.0508,
      "step": 117981
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.9539862871170044,
      "learning_rate": 0.0002886289164182771,
      "loss": 2.0822,
      "step": 117982
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.9624011516571045,
      "learning_rate": 0.0002886248287966408,
      "loss": 2.3855,
      "step": 117983
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.9285945892333984,
      "learning_rate": 0.00028862074117711936,
      "loss": 2.1674,
      "step": 117984
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.9417005777359009,
      "learning_rate": 0.00028861665355971346,
      "loss": 2.0866,
      "step": 117985
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.9012495279312134,
      "learning_rate": 0.0002886125659444242,
      "loss": 2.2151,
      "step": 117986
    },
    {
      "epoch": 1.54,
      "grad_norm": 1.0259901285171509,
      "learning_rate": 0.0002886084783312518,
      "loss": 2.1186,
      "step": 117987
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.9694017767906189,
      "learning_rate": 0.0002886043907201974,
      "loss": 2.2367,
      "step": 117988
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.9703563451766968,
      "learning_rate": 0.00028860030311126164,
      "loss": 2.1227,
      "step": 117989
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.9382787346839905,
      "learning_rate": 0.0002885962155044453,
      "loss": 1.996,
      "step": 117990
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.9368312358856201,
      "learning_rate": 0.0002885921278997491,
      "loss": 2.1411,
      "step": 117991
    },
    {
      "epoch": 1.54,
      "grad_norm": 1.009810447692871,
      "learning_rate": 0.000288588040297174,
      "loss": 1.963,
      "step": 117992
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.9115653038024902,
      "learning_rate": 0.0002885839526967205,
      "loss": 2.2,
      "step": 117993
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.986825704574585,
      "learning_rate": 0.00028857986509838937,
      "loss": 1.9406,
      "step": 117994
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.9981346726417542,
      "learning_rate": 0.0002885757775021815,
      "loss": 2.1856,
      "step": 117995
    },
    {
      "epoch": 1.54,
      "grad_norm": 1.010972023010254,
      "learning_rate": 0.00028857168990809766,
      "loss": 2.1015,
      "step": 117996
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.9133183360099792,
      "learning_rate": 0.0002885676023161385,
      "loss": 2.0269,
      "step": 117997
    },
    {
      "epoch": 1.54,
      "grad_norm": 1.0588456392288208,
      "learning_rate": 0.000288563514726305,
      "loss": 2.1609,
      "step": 117998
    },
    {
      "epoch": 1.54,
      "grad_norm": 1.0750118494033813,
      "learning_rate": 0.00028855942713859763,
      "loss": 2.2298,
      "step": 117999
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.9626694321632385,
      "learning_rate": 0.00028855533955301723,
      "loss": 2.1384,
      "step": 118000
    },
    {
      "epoch": 1.54,
      "grad_norm": 1.030118703842163,
      "learning_rate": 0.00028855125196956463,
      "loss": 2.0367,
      "step": 118001
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.9243606925010681,
      "learning_rate": 0.0002885471643882406,
      "loss": 2.0975,
      "step": 118002
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.9795352220535278,
      "learning_rate": 0.0002885430768090458,
      "loss": 1.9237,
      "step": 118003
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.9002817869186401,
      "learning_rate": 0.0002885389892319813,
      "loss": 1.9841,
      "step": 118004
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.9555577635765076,
      "learning_rate": 0.00028853490165704743,
      "loss": 2.0946,
      "step": 118005
    },
    {
      "epoch": 1.54,
      "grad_norm": 1.1033551692962646,
      "learning_rate": 0.0002885308140842451,
      "loss": 2.1398,
      "step": 118006
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.9737744927406311,
      "learning_rate": 0.00028852672651357515,
      "loss": 2.355,
      "step": 118007
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.9084562659263611,
      "learning_rate": 0.0002885226389450383,
      "loss": 2.3028,
      "step": 118008
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.9882645606994629,
      "learning_rate": 0.0002885185513786353,
      "loss": 2.1332,
      "step": 118009
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.8160616159439087,
      "learning_rate": 0.00028851446381436704,
      "loss": 2.2252,
      "step": 118010
    },
    {
      "epoch": 1.54,
      "grad_norm": 1.0066814422607422,
      "learning_rate": 0.00028851037625223403,
      "loss": 2.2648,
      "step": 118011
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.8864179253578186,
      "learning_rate": 0.0002885062886922372,
      "loss": 1.9765,
      "step": 118012
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.8363642692565918,
      "learning_rate": 0.0002885022011343772,
      "loss": 2.1657,
      "step": 118013
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.9242823719978333,
      "learning_rate": 0.00028849811357865493,
      "loss": 2.1392,
      "step": 118014
    },
    {
      "epoch": 1.54,
      "grad_norm": 1.1234608888626099,
      "learning_rate": 0.000288494026025071,
      "loss": 2.2529,
      "step": 118015
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.9353845715522766,
      "learning_rate": 0.0002884899384736264,
      "loss": 1.9987,
      "step": 118016
    },
    {
      "epoch": 1.54,
      "grad_norm": 1.0566810369491577,
      "learning_rate": 0.0002884858509243217,
      "loss": 2.0295,
      "step": 118017
    },
    {
      "epoch": 1.54,
      "grad_norm": 1.1420491933822632,
      "learning_rate": 0.0002884817633771576,
      "loss": 1.9964,
      "step": 118018
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.9488921165466309,
      "learning_rate": 0.0002884776758321349,
      "loss": 2.1949,
      "step": 118019
    },
    {
      "epoch": 1.54,
      "grad_norm": 1.0121521949768066,
      "learning_rate": 0.0002884735882892545,
      "loss": 2.269,
      "step": 118020
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.9891918897628784,
      "learning_rate": 0.0002884695007485171,
      "loss": 2.1268,
      "step": 118021
    },
    {
      "epoch": 1.54,
      "grad_norm": 1.0491020679473877,
      "learning_rate": 0.0002884654132099235,
      "loss": 2.2462,
      "step": 118022
    },
    {
      "epoch": 1.54,
      "grad_norm": 1.4847148656845093,
      "learning_rate": 0.00028846132567347436,
      "loss": 2.3021,
      "step": 118023
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.9503830671310425,
      "learning_rate": 0.00028845723813917043,
      "loss": 2.3102,
      "step": 118024
    },
    {
      "epoch": 1.54,
      "grad_norm": 1.06302011013031,
      "learning_rate": 0.0002884531506070125,
      "loss": 2.086,
      "step": 118025
    },
    {
      "epoch": 1.54,
      "grad_norm": 1.0882070064544678,
      "learning_rate": 0.0002884490630770014,
      "loss": 2.2798,
      "step": 118026
    },
    {
      "epoch": 1.54,
      "grad_norm": 1.2192407846450806,
      "learning_rate": 0.00028844497554913773,
      "loss": 2.1348,
      "step": 118027
    },
    {
      "epoch": 1.54,
      "grad_norm": 1.2108899354934692,
      "learning_rate": 0.00028844088802342253,
      "loss": 2.0081,
      "step": 118028
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.9689391255378723,
      "learning_rate": 0.0002884368004998563,
      "loss": 1.9764,
      "step": 118029
    },
    {
      "epoch": 1.54,
      "grad_norm": 1.114845633506775,
      "learning_rate": 0.0002884327129784399,
      "loss": 2.1107,
      "step": 118030
    },
    {
      "epoch": 1.54,
      "grad_norm": 1.191372036933899,
      "learning_rate": 0.000288428625459174,
      "loss": 2.0202,
      "step": 118031
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.9766696095466614,
      "learning_rate": 0.0002884245379420595,
      "loss": 2.2876,
      "step": 118032
    },
    {
      "epoch": 1.54,
      "grad_norm": 1.081602931022644,
      "learning_rate": 0.0002884204504270971,
      "loss": 2.201,
      "step": 118033
    },
    {
      "epoch": 1.54,
      "grad_norm": 1.0066637992858887,
      "learning_rate": 0.0002884163629142876,
      "loss": 2.0043,
      "step": 118034
    },
    {
      "epoch": 1.54,
      "grad_norm": 1.0059523582458496,
      "learning_rate": 0.00028841227540363164,
      "loss": 2.3171,
      "step": 118035
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.9685863256454468,
      "learning_rate": 0.00028840818789513005,
      "loss": 2.1148,
      "step": 118036
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.9257397651672363,
      "learning_rate": 0.00028840410038878363,
      "loss": 2.3755,
      "step": 118037
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.937587559223175,
      "learning_rate": 0.0002884000128845932,
      "loss": 2.0453,
      "step": 118038
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.9784867763519287,
      "learning_rate": 0.00028839592538255926,
      "loss": 2.2403,
      "step": 118039
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.9930651187896729,
      "learning_rate": 0.0002883918378826828,
      "loss": 2.0737,
      "step": 118040
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.8976748585700989,
      "learning_rate": 0.0002883877503849646,
      "loss": 1.9606,
      "step": 118041
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.9064561724662781,
      "learning_rate": 0.0002883836628894052,
      "loss": 2.1408,
      "step": 118042
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.8459160327911377,
      "learning_rate": 0.0002883795753960056,
      "loss": 2.1423,
      "step": 118043
    },
    {
      "epoch": 1.54,
      "grad_norm": 1.0231225490570068,
      "learning_rate": 0.0002883754879047664,
      "loss": 2.1768,
      "step": 118044
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.9617785215377808,
      "learning_rate": 0.00028837140041568845,
      "loss": 2.0917,
      "step": 118045
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.8880669474601746,
      "learning_rate": 0.00028836731292877245,
      "loss": 1.9225,
      "step": 118046
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.9139611721038818,
      "learning_rate": 0.0002883632254440193,
      "loss": 1.9769,
      "step": 118047
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.9502915740013123,
      "learning_rate": 0.00028835913796142947,
      "loss": 2.3367,
      "step": 118048
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.9768744111061096,
      "learning_rate": 0.000288355050481004,
      "loss": 2.106,
      "step": 118049
    },
    {
      "epoch": 1.54,
      "grad_norm": 1.0302022695541382,
      "learning_rate": 0.00028835096300274345,
      "loss": 1.9817,
      "step": 118050
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.8907628655433655,
      "learning_rate": 0.0002883468755266488,
      "loss": 1.8343,
      "step": 118051
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.9726751446723938,
      "learning_rate": 0.0002883427880527206,
      "loss": 2.255,
      "step": 118052
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.8894134163856506,
      "learning_rate": 0.00028833870058095974,
      "loss": 2.1903,
      "step": 118053
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.964070737361908,
      "learning_rate": 0.0002883346131113669,
      "loss": 1.9962,
      "step": 118054
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.949288010597229,
      "learning_rate": 0.0002883305256439429,
      "loss": 1.9418,
      "step": 118055
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.9186584949493408,
      "learning_rate": 0.0002883264381786885,
      "loss": 2.1785,
      "step": 118056
    },
    {
      "epoch": 1.54,
      "grad_norm": 1.0682426691055298,
      "learning_rate": 0.0002883223507156043,
      "loss": 1.9529,
      "step": 118057
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.9766527414321899,
      "learning_rate": 0.00028831826325469135,
      "loss": 2.1552,
      "step": 118058
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.9032448530197144,
      "learning_rate": 0.0002883141757959503,
      "loss": 2.1722,
      "step": 118059
    },
    {
      "epoch": 1.54,
      "grad_norm": 1.0147840976715088,
      "learning_rate": 0.0002883100883393817,
      "loss": 2.2304,
      "step": 118060
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.9765450358390808,
      "learning_rate": 0.0002883060008849866,
      "loss": 2.2209,
      "step": 118061
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.9862636923789978,
      "learning_rate": 0.00028830191343276555,
      "loss": 2.162,
      "step": 118062
    },
    {
      "epoch": 1.54,
      "grad_norm": 1.002193808555603,
      "learning_rate": 0.0002882978259827194,
      "loss": 2.0769,
      "step": 118063
    },
    {
      "epoch": 1.54,
      "grad_norm": 1.025274634361267,
      "learning_rate": 0.0002882937385348489,
      "loss": 2.0445,
      "step": 118064
    },
    {
      "epoch": 1.54,
      "grad_norm": 1.0537811517715454,
      "learning_rate": 0.000288289651089155,
      "loss": 2.0766,
      "step": 118065
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.9371894598007202,
      "learning_rate": 0.0002882855636456381,
      "loss": 2.1725,
      "step": 118066
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.9642903804779053,
      "learning_rate": 0.0002882814762042991,
      "loss": 2.3006,
      "step": 118067
    },
    {
      "epoch": 1.54,
      "grad_norm": 1.0496716499328613,
      "learning_rate": 0.0002882773887651388,
      "loss": 2.2456,
      "step": 118068
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.9274550080299377,
      "learning_rate": 0.00028827330132815803,
      "loss": 2.2225,
      "step": 118069
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.9178584218025208,
      "learning_rate": 0.0002882692138933574,
      "loss": 1.7518,
      "step": 118070
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.955493152141571,
      "learning_rate": 0.00028826512646073794,
      "loss": 2.1483,
      "step": 118071
    },
    {
      "epoch": 1.54,
      "grad_norm": 1.0078624486923218,
      "learning_rate": 0.0002882610390303001,
      "loss": 1.994,
      "step": 118072
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.992570698261261,
      "learning_rate": 0.0002882569516020447,
      "loss": 2.2183,
      "step": 118073
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.9400854706764221,
      "learning_rate": 0.00028825286417597255,
      "loss": 2.1955,
      "step": 118074
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.9452162981033325,
      "learning_rate": 0.0002882487767520844,
      "loss": 2.3139,
      "step": 118075
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.9239752292633057,
      "learning_rate": 0.000288244689330381,
      "loss": 2.2697,
      "step": 118076
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.8970984816551208,
      "learning_rate": 0.0002882406019108634,
      "loss": 1.9778,
      "step": 118077
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.9445408582687378,
      "learning_rate": 0.0002882365144935318,
      "loss": 1.9298,
      "step": 118078
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.9251987338066101,
      "learning_rate": 0.0002882324270783874,
      "loss": 2.2167,
      "step": 118079
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.9949750900268555,
      "learning_rate": 0.00028822833966543073,
      "loss": 2.1476,
      "step": 118080
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.8498756289482117,
      "learning_rate": 0.00028822425225466264,
      "loss": 2.0271,
      "step": 118081
    },
    {
      "epoch": 1.54,
      "grad_norm": 1.0710387229919434,
      "learning_rate": 0.0002882201648460839,
      "loss": 2.2902,
      "step": 118082
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.9310219883918762,
      "learning_rate": 0.0002882160774396954,
      "loss": 2.1003,
      "step": 118083
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.9837107062339783,
      "learning_rate": 0.00028821199003549756,
      "loss": 2.0273,
      "step": 118084
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.9619765877723694,
      "learning_rate": 0.00028820790263349135,
      "loss": 2.1809,
      "step": 118085
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.8728516697883606,
      "learning_rate": 0.00028820381523367755,
      "loss": 2.2806,
      "step": 118086
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.9720299243927002,
      "learning_rate": 0.0002881997278360569,
      "loss": 2.165,
      "step": 118087
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.9265251755714417,
      "learning_rate": 0.0002881956404406301,
      "loss": 2.0816,
      "step": 118088
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.9762640595436096,
      "learning_rate": 0.0002881915530473981,
      "loss": 2.389,
      "step": 118089
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.9060706496238708,
      "learning_rate": 0.00028818746565636137,
      "loss": 2.1838,
      "step": 118090
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.9075031876564026,
      "learning_rate": 0.0002881833782675208,
      "loss": 1.9964,
      "step": 118091
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.9144876003265381,
      "learning_rate": 0.00028817929088087717,
      "loss": 2.1556,
      "step": 118092
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.9658071994781494,
      "learning_rate": 0.0002881752034964313,
      "loss": 2.2829,
      "step": 118093
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.9559183120727539,
      "learning_rate": 0.0002881711161141838,
      "loss": 2.2036,
      "step": 118094
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.9193957448005676,
      "learning_rate": 0.0002881670287341357,
      "loss": 2.0716,
      "step": 118095
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.9212455153465271,
      "learning_rate": 0.0002881629413562874,
      "loss": 2.0487,
      "step": 118096
    },
    {
      "epoch": 1.54,
      "grad_norm": 1.04254949092865,
      "learning_rate": 0.0002881588539806399,
      "loss": 2.2434,
      "step": 118097
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.9513470530509949,
      "learning_rate": 0.0002881547666071938,
      "loss": 1.9184,
      "step": 118098
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.9010570049285889,
      "learning_rate": 0.00028815067923595,
      "loss": 1.9715,
      "step": 118099
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.9422113299369812,
      "learning_rate": 0.0002881465918669092,
      "loss": 2.2024,
      "step": 118100
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.937247097492218,
      "learning_rate": 0.0002881425045000723,
      "loss": 2.093,
      "step": 118101
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.9556407928466797,
      "learning_rate": 0.0002881384171354398,
      "loss": 2.1103,
      "step": 118102
    },
    {
      "epoch": 1.54,
      "grad_norm": 1.02810800075531,
      "learning_rate": 0.00028813432977301267,
      "loss": 2.0497,
      "step": 118103
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.9787158370018005,
      "learning_rate": 0.0002881302424127915,
      "loss": 2.1646,
      "step": 118104
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.9534167051315308,
      "learning_rate": 0.00028812615505477716,
      "loss": 2.3574,
      "step": 118105
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.973373532295227,
      "learning_rate": 0.0002881220676989704,
      "loss": 2.0624,
      "step": 118106
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.9007719159126282,
      "learning_rate": 0.000288117980345372,
      "loss": 2.295,
      "step": 118107
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.9207714796066284,
      "learning_rate": 0.00028811389299398274,
      "loss": 2.157,
      "step": 118108
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.9786935448646545,
      "learning_rate": 0.0002881098056448033,
      "loss": 2.0824,
      "step": 118109
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.9689263701438904,
      "learning_rate": 0.0002881057182978344,
      "loss": 2.1349,
      "step": 118110
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.8838450908660889,
      "learning_rate": 0.0002881016309530769,
      "loss": 2.0456,
      "step": 118111
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.9972584843635559,
      "learning_rate": 0.0002880975436105315,
      "loss": 2.1024,
      "step": 118112
    },
    {
      "epoch": 1.54,
      "grad_norm": 1.1584237813949585,
      "learning_rate": 0.00028809345627019904,
      "loss": 2.3139,
      "step": 118113
    },
    {
      "epoch": 1.54,
      "grad_norm": 1.0023770332336426,
      "learning_rate": 0.0002880893689320803,
      "loss": 2.0092,
      "step": 118114
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.9903088212013245,
      "learning_rate": 0.00028808528159617595,
      "loss": 1.9922,
      "step": 118115
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.9784697890281677,
      "learning_rate": 0.00028808119426248667,
      "loss": 2.1942,
      "step": 118116
    },
    {
      "epoch": 1.54,
      "grad_norm": 1.0067851543426514,
      "learning_rate": 0.0002880771069310134,
      "loss": 1.8304,
      "step": 118117
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.923306941986084,
      "learning_rate": 0.00028807301960175676,
      "loss": 2.2215,
      "step": 118118
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.9547733664512634,
      "learning_rate": 0.00028806893227471754,
      "loss": 2.3012,
      "step": 118119
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.8773093223571777,
      "learning_rate": 0.0002880648449498967,
      "loss": 2.1693,
      "step": 118120
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.8896914124488831,
      "learning_rate": 0.0002880607576272947,
      "loss": 2.3232,
      "step": 118121
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.9240485429763794,
      "learning_rate": 0.0002880566703069125,
      "loss": 1.8181,
      "step": 118122
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.9625601172447205,
      "learning_rate": 0.0002880525829887507,
      "loss": 1.9282,
      "step": 118123
    },
    {
      "epoch": 1.54,
      "grad_norm": 1.0601449012756348,
      "learning_rate": 0.0002880484956728102,
      "loss": 2.2307,
      "step": 118124
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.9782155752182007,
      "learning_rate": 0.0002880444083590917,
      "loss": 2.1667,
      "step": 118125
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.9164915680885315,
      "learning_rate": 0.00028804032104759605,
      "loss": 2.0199,
      "step": 118126
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.874039351940155,
      "learning_rate": 0.0002880362337383238,
      "loss": 2.1152,
      "step": 118127
    },
    {
      "epoch": 1.54,
      "grad_norm": 1.1007119417190552,
      "learning_rate": 0.0002880321464312759,
      "loss": 2.2937,
      "step": 118128
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.9940629601478577,
      "learning_rate": 0.000288028059126453,
      "loss": 2.0174,
      "step": 118129
    },
    {
      "epoch": 1.54,
      "grad_norm": 1.0669678449630737,
      "learning_rate": 0.000288023971823856,
      "loss": 2.2148,
      "step": 118130
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.9869844913482666,
      "learning_rate": 0.0002880198845234855,
      "loss": 2.1313,
      "step": 118131
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.8756983280181885,
      "learning_rate": 0.00028801579722534246,
      "loss": 2.1285,
      "step": 118132
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.9555686116218567,
      "learning_rate": 0.0002880117099294274,
      "loss": 2.1415,
      "step": 118133
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.9140327572822571,
      "learning_rate": 0.00028800762263574115,
      "loss": 2.1262,
      "step": 118134
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.9956476092338562,
      "learning_rate": 0.00028800353534428453,
      "loss": 2.3388,
      "step": 118135
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.9601879715919495,
      "learning_rate": 0.00028799944805505827,
      "loss": 2.1688,
      "step": 118136
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.9827389717102051,
      "learning_rate": 0.00028799536076806325,
      "loss": 2.2445,
      "step": 118137
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.9080847501754761,
      "learning_rate": 0.00028799127348330005,
      "loss": 2.163,
      "step": 118138
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.9188392162322998,
      "learning_rate": 0.0002879871862007695,
      "loss": 2.2592,
      "step": 118139
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.9968629479408264,
      "learning_rate": 0.00028798309892047236,
      "loss": 1.9208,
      "step": 118140
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.9381235837936401,
      "learning_rate": 0.00028797901164240937,
      "loss": 2.1286,
      "step": 118141
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.9871414303779602,
      "learning_rate": 0.00028797492436658125,
      "loss": 2.0232,
      "step": 118142
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.974277675151825,
      "learning_rate": 0.0002879708370929889,
      "loss": 2.1828,
      "step": 118143
    },
    {
      "epoch": 1.54,
      "grad_norm": 1.105077862739563,
      "learning_rate": 0.0002879667498216331,
      "loss": 2.0731,
      "step": 118144
    },
    {
      "epoch": 1.54,
      "grad_norm": 1.065584421157837,
      "learning_rate": 0.0002879626625525144,
      "loss": 2.1643,
      "step": 118145
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.9286231994628906,
      "learning_rate": 0.00028795857528563363,
      "loss": 2.3023,
      "step": 118146
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.9190011620521545,
      "learning_rate": 0.0002879544880209916,
      "loss": 2.1294,
      "step": 118147
    },
    {
      "epoch": 1.54,
      "grad_norm": 1.0459544658660889,
      "learning_rate": 0.00028795040075858905,
      "loss": 2.4545,
      "step": 118148
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.9770644903182983,
      "learning_rate": 0.0002879463134984268,
      "loss": 2.0575,
      "step": 118149
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.9711406826972961,
      "learning_rate": 0.0002879422262405056,
      "loss": 1.9448,
      "step": 118150
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.8586111068725586,
      "learning_rate": 0.0002879381389848261,
      "loss": 2.1069,
      "step": 118151
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.9057788252830505,
      "learning_rate": 0.0002879340517313891,
      "loss": 2.1498,
      "step": 118152
    },
    {
      "epoch": 1.54,
      "grad_norm": 1.0482876300811768,
      "learning_rate": 0.0002879299644801954,
      "loss": 2.2124,
      "step": 118153
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.9836305975914001,
      "learning_rate": 0.0002879258772312457,
      "loss": 2.2384,
      "step": 118154
    },
    {
      "epoch": 1.54,
      "grad_norm": 1.0083599090576172,
      "learning_rate": 0.0002879217899845409,
      "loss": 2.1577,
      "step": 118155
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.9063758254051208,
      "learning_rate": 0.00028791770274008173,
      "loss": 2.164,
      "step": 118156
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.9839576482772827,
      "learning_rate": 0.0002879136154978688,
      "loss": 2.1476,
      "step": 118157
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.9470377564430237,
      "learning_rate": 0.0002879095282579029,
      "loss": 2.0751,
      "step": 118158
    },
    {
      "epoch": 1.54,
      "grad_norm": 1.019556999206543,
      "learning_rate": 0.0002879054410201849,
      "loss": 2.1137,
      "step": 118159
    },
    {
      "epoch": 1.54,
      "grad_norm": 1.1158376932144165,
      "learning_rate": 0.0002879013537847155,
      "loss": 2.2685,
      "step": 118160
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.9454668760299683,
      "learning_rate": 0.00028789726655149546,
      "loss": 2.2021,
      "step": 118161
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.9645491242408752,
      "learning_rate": 0.00028789317932052566,
      "loss": 2.2671,
      "step": 118162
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.9325775504112244,
      "learning_rate": 0.00028788909209180667,
      "loss": 2.1601,
      "step": 118163
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.8578421473503113,
      "learning_rate": 0.00028788500486533924,
      "loss": 2.0373,
      "step": 118164
    },
    {
      "epoch": 1.54,
      "grad_norm": 1.1783101558685303,
      "learning_rate": 0.0002878809176411243,
      "loss": 1.9321,
      "step": 118165
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.9294511079788208,
      "learning_rate": 0.0002878768304191625,
      "loss": 2.0641,
      "step": 118166
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.9879274964332581,
      "learning_rate": 0.00028787274319945456,
      "loss": 2.1587,
      "step": 118167
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.9982287287712097,
      "learning_rate": 0.0002878686559820015,
      "loss": 2.4139,
      "step": 118168
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.9910890460014343,
      "learning_rate": 0.00028786456876680376,
      "loss": 2.0686,
      "step": 118169
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.9108865857124329,
      "learning_rate": 0.0002878604815538622,
      "loss": 2.2777,
      "step": 118170
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.8855111598968506,
      "learning_rate": 0.00028785639434317765,
      "loss": 1.9459,
      "step": 118171
    },
    {
      "epoch": 1.54,
      "grad_norm": 1.1184026002883911,
      "learning_rate": 0.00028785230713475075,
      "loss": 2.2864,
      "step": 118172
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.9489737153053284,
      "learning_rate": 0.0002878482199285824,
      "loss": 2.2456,
      "step": 118173
    },
    {
      "epoch": 1.54,
      "grad_norm": 1.016206979751587,
      "learning_rate": 0.00028784413272467326,
      "loss": 2.1954,
      "step": 118174
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.9053696990013123,
      "learning_rate": 0.00028784004552302424,
      "loss": 2.0022,
      "step": 118175
    },
    {
      "epoch": 1.54,
      "grad_norm": 1.0902456045150757,
      "learning_rate": 0.00028783595832363595,
      "loss": 2.2772,
      "step": 118176
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.9335259199142456,
      "learning_rate": 0.0002878318711265091,
      "loss": 2.1784,
      "step": 118177
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.8802735209465027,
      "learning_rate": 0.00028782778393164453,
      "loss": 1.96,
      "step": 118178
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.9724447727203369,
      "learning_rate": 0.000287823696739043,
      "loss": 2.2587,
      "step": 118179
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.8572832942008972,
      "learning_rate": 0.0002878196095487053,
      "loss": 2.1639,
      "step": 118180
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.9239627122879028,
      "learning_rate": 0.0002878155223606323,
      "loss": 2.1685,
      "step": 118181
    },
    {
      "epoch": 1.54,
      "grad_norm": 1.0582619905471802,
      "learning_rate": 0.0002878114351748245,
      "loss": 2.217,
      "step": 118182
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.9312293529510498,
      "learning_rate": 0.00028780734799128277,
      "loss": 1.9358,
      "step": 118183
    },
    {
      "epoch": 1.54,
      "grad_norm": 1.217949628829956,
      "learning_rate": 0.0002878032608100079,
      "loss": 2.255,
      "step": 118184
    },
    {
      "epoch": 1.54,
      "grad_norm": 1.0574419498443604,
      "learning_rate": 0.0002877991736310006,
      "loss": 2.2479,
      "step": 118185
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.9481713175773621,
      "learning_rate": 0.00028779508645426175,
      "loss": 2.2483,
      "step": 118186
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.9441655874252319,
      "learning_rate": 0.00028779099927979206,
      "loss": 1.9931,
      "step": 118187
    },
    {
      "epoch": 1.54,
      "grad_norm": 1.0218902826309204,
      "learning_rate": 0.0002877869121075922,
      "loss": 2.0196,
      "step": 118188
    },
    {
      "epoch": 1.54,
      "grad_norm": 1.0279537439346313,
      "learning_rate": 0.00028778282493766293,
      "loss": 2.0585,
      "step": 118189
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.9630447626113892,
      "learning_rate": 0.00028777873777000506,
      "loss": 2.1968,
      "step": 118190
    },
    {
      "epoch": 1.54,
      "grad_norm": 1.0161895751953125,
      "learning_rate": 0.00028777465060461936,
      "loss": 2.0813,
      "step": 118191
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.9273300170898438,
      "learning_rate": 0.0002877705634415066,
      "loss": 1.9855,
      "step": 118192
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.9791061282157898,
      "learning_rate": 0.00028776647628066764,
      "loss": 2.1647,
      "step": 118193
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.9265055060386658,
      "learning_rate": 0.000287762389122103,
      "loss": 2.127,
      "step": 118194
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.8204984664916992,
      "learning_rate": 0.0002877583019658135,
      "loss": 2.1728,
      "step": 118195
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.9407397508621216,
      "learning_rate": 0.00028775421481180006,
      "loss": 2.2105,
      "step": 118196
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.9875024557113647,
      "learning_rate": 0.0002877501276600633,
      "loss": 1.9768,
      "step": 118197
    },
    {
      "epoch": 1.54,
      "grad_norm": 1.0682801008224487,
      "learning_rate": 0.00028774604051060406,
      "loss": 2.1081,
      "step": 118198
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.908189058303833,
      "learning_rate": 0.0002877419533634231,
      "loss": 2.0431,
      "step": 118199
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.9907836318016052,
      "learning_rate": 0.0002877378662185211,
      "loss": 2.0657,
      "step": 118200
    },
    {
      "epoch": 1.54,
      "grad_norm": 1.0597809553146362,
      "learning_rate": 0.0002877337790758988,
      "loss": 2.1826,
      "step": 118201
    },
    {
      "epoch": 1.54,
      "grad_norm": 1.035254716873169,
      "learning_rate": 0.00028772969193555705,
      "loss": 2.3386,
      "step": 118202
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.8842195868492126,
      "learning_rate": 0.0002877256047974966,
      "loss": 2.1474,
      "step": 118203
    },
    {
      "epoch": 1.54,
      "grad_norm": 1.0165146589279175,
      "learning_rate": 0.0002877215176617182,
      "loss": 2.0073,
      "step": 118204
    },
    {
      "epoch": 1.54,
      "grad_norm": 1.2747703790664673,
      "learning_rate": 0.00028771743052822266,
      "loss": 2.1407,
      "step": 118205
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.9323114156723022,
      "learning_rate": 0.0002877133433970106,
      "loss": 2.1263,
      "step": 118206
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.9457579851150513,
      "learning_rate": 0.0002877092562680829,
      "loss": 2.0488,
      "step": 118207
    },
    {
      "epoch": 1.54,
      "grad_norm": 1.0464147329330444,
      "learning_rate": 0.00028770516914144024,
      "loss": 2.4347,
      "step": 118208
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.9721435904502869,
      "learning_rate": 0.0002877010820170834,
      "loss": 2.1986,
      "step": 118209
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.9414676427841187,
      "learning_rate": 0.0002876969948950132,
      "loss": 2.0123,
      "step": 118210
    },
    {
      "epoch": 1.54,
      "grad_norm": 1.0296918153762817,
      "learning_rate": 0.00028769290777523046,
      "loss": 2.0877,
      "step": 118211
    },
    {
      "epoch": 1.54,
      "grad_norm": 1.130393385887146,
      "learning_rate": 0.0002876888206577357,
      "loss": 2.3342,
      "step": 118212
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.9686211347579956,
      "learning_rate": 0.00028768473354252985,
      "loss": 2.2538,
      "step": 118213
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.9401081800460815,
      "learning_rate": 0.0002876806464296137,
      "loss": 1.9708,
      "step": 118214
    },
    {
      "epoch": 1.54,
      "grad_norm": 1.0329376459121704,
      "learning_rate": 0.0002876765593189879,
      "loss": 1.9994,
      "step": 118215
    },
    {
      "epoch": 1.54,
      "grad_norm": 1.0146424770355225,
      "learning_rate": 0.0002876724722106533,
      "loss": 2.4445,
      "step": 118216
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.9432989358901978,
      "learning_rate": 0.00028766838510461057,
      "loss": 2.0443,
      "step": 118217
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.9170256853103638,
      "learning_rate": 0.0002876642980008605,
      "loss": 2.0184,
      "step": 118218
    },
    {
      "epoch": 1.54,
      "grad_norm": 1.0678048133850098,
      "learning_rate": 0.0002876602108994039,
      "loss": 2.1085,
      "step": 118219
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.9845888614654541,
      "learning_rate": 0.0002876561238002415,
      "loss": 2.1542,
      "step": 118220
    },
    {
      "epoch": 1.54,
      "grad_norm": 1.0361995697021484,
      "learning_rate": 0.0002876520367033741,
      "loss": 2.2843,
      "step": 118221
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.9269155859947205,
      "learning_rate": 0.0002876479496088024,
      "loss": 2.1053,
      "step": 118222
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.8915241956710815,
      "learning_rate": 0.0002876438625165272,
      "loss": 2.1858,
      "step": 118223
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.9278490543365479,
      "learning_rate": 0.0002876397754265492,
      "loss": 2.0716,
      "step": 118224
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.8734754920005798,
      "learning_rate": 0.0002876356883388692,
      "loss": 2.1467,
      "step": 118225
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.8879071474075317,
      "learning_rate": 0.0002876316012534879,
      "loss": 2.0752,
      "step": 118226
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.9426394701004028,
      "learning_rate": 0.00028762751417040616,
      "loss": 1.9771,
      "step": 118227
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.9886695742607117,
      "learning_rate": 0.0002876234270896247,
      "loss": 1.8817,
      "step": 118228
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.9692946076393127,
      "learning_rate": 0.00028761934001114443,
      "loss": 2.2862,
      "step": 118229
    },
    {
      "epoch": 1.54,
      "grad_norm": 1.0609310865402222,
      "learning_rate": 0.0002876152529349658,
      "loss": 2.445,
      "step": 118230
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.8707285523414612,
      "learning_rate": 0.00028761116586108974,
      "loss": 2.2282,
      "step": 118231
    },
    {
      "epoch": 1.54,
      "grad_norm": 1.3371199369430542,
      "learning_rate": 0.000287607078789517,
      "loss": 2.263,
      "step": 118232
    },
    {
      "epoch": 1.54,
      "grad_norm": 1.366369605064392,
      "learning_rate": 0.0002876029917202483,
      "loss": 2.1178,
      "step": 118233
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.93271803855896,
      "learning_rate": 0.0002875989046532845,
      "loss": 2.1771,
      "step": 118234
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.8921040296554565,
      "learning_rate": 0.0002875948175886264,
      "loss": 2.3001,
      "step": 118235
    },
    {
      "epoch": 1.54,
      "grad_norm": 1.0181828737258911,
      "learning_rate": 0.00028759073052627457,
      "loss": 2.3033,
      "step": 118236
    },
    {
      "epoch": 1.54,
      "grad_norm": 1.0104622840881348,
      "learning_rate": 0.0002875866434662298,
      "loss": 2.0173,
      "step": 118237
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.9161467552185059,
      "learning_rate": 0.0002875825564084929,
      "loss": 2.1524,
      "step": 118238
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.968623697757721,
      "learning_rate": 0.0002875784693530647,
      "loss": 2.0411,
      "step": 118239
    },
    {
      "epoch": 1.54,
      "grad_norm": 1.0739072561264038,
      "learning_rate": 0.0002875743822999458,
      "loss": 2.2998,
      "step": 118240
    },
    {
      "epoch": 1.54,
      "grad_norm": 1.0648975372314453,
      "learning_rate": 0.0002875702952491371,
      "loss": 2.284,
      "step": 118241
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.9417524933815002,
      "learning_rate": 0.00028756620820063946,
      "loss": 2.1088,
      "step": 118242
    },
    {
      "epoch": 1.54,
      "grad_norm": 1.1413487195968628,
      "learning_rate": 0.0002875621211544534,
      "loss": 2.2299,
      "step": 118243
    },
    {
      "epoch": 1.54,
      "grad_norm": 1.1233582496643066,
      "learning_rate": 0.00028755803411057974,
      "loss": 2.1444,
      "step": 118244
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.9202128052711487,
      "learning_rate": 0.00028755394706901925,
      "loss": 2.1185,
      "step": 118245
    },
    {
      "epoch": 1.54,
      "grad_norm": 1.019566535949707,
      "learning_rate": 0.0002875498600297728,
      "loss": 2.0693,
      "step": 118246
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.9084599018096924,
      "learning_rate": 0.00028754577299284096,
      "loss": 1.9981,
      "step": 118247
    },
    {
      "epoch": 1.54,
      "grad_norm": 1.0355561971664429,
      "learning_rate": 0.00028754168595822484,
      "loss": 2.0913,
      "step": 118248
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.9667080044746399,
      "learning_rate": 0.00028753759892592473,
      "loss": 2.2274,
      "step": 118249
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.8379716873168945,
      "learning_rate": 0.0002875335118959416,
      "loss": 2.2388,
      "step": 118250
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.9028506875038147,
      "learning_rate": 0.0002875294248682763,
      "loss": 2.0663,
      "step": 118251
    },
    {
      "epoch": 1.54,
      "grad_norm": 1.5431519746780396,
      "learning_rate": 0.00028752533784292953,
      "loss": 1.7639,
      "step": 118252
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.9835575819015503,
      "learning_rate": 0.00028752125081990196,
      "loss": 2.3759,
      "step": 118253
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.8919647932052612,
      "learning_rate": 0.00028751716379919463,
      "loss": 2.2375,
      "step": 118254
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.9403638243675232,
      "learning_rate": 0.0002875130767808079,
      "loss": 2.1415,
      "step": 118255
    },
    {
      "epoch": 1.54,
      "grad_norm": 1.1680738925933838,
      "learning_rate": 0.0002875089897647427,
      "loss": 1.9106,
      "step": 118256
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.9204570651054382,
      "learning_rate": 0.0002875049027509999,
      "loss": 2.2323,
      "step": 118257
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.9775270819664001,
      "learning_rate": 0.00028750081573958015,
      "loss": 2.2592,
      "step": 118258
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.9899745583534241,
      "learning_rate": 0.0002874967287304842,
      "loss": 2.1331,
      "step": 118259
    },
    {
      "epoch": 1.54,
      "grad_norm": 1.1407716274261475,
      "learning_rate": 0.000287492641723713,
      "loss": 1.9231,
      "step": 118260
    },
    {
      "epoch": 1.54,
      "grad_norm": 1.0261884927749634,
      "learning_rate": 0.000287488554719267,
      "loss": 2.408,
      "step": 118261
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.8831388354301453,
      "learning_rate": 0.00028748446771714714,
      "loss": 2.1552,
      "step": 118262
    },
    {
      "epoch": 1.54,
      "grad_norm": 1.025182843208313,
      "learning_rate": 0.0002874803807173542,
      "loss": 1.9889,
      "step": 118263
    },
    {
      "epoch": 1.54,
      "grad_norm": 1.0419044494628906,
      "learning_rate": 0.0002874762937198888,
      "loss": 2.2027,
      "step": 118264
    },
    {
      "epoch": 1.54,
      "grad_norm": 1.0286883115768433,
      "learning_rate": 0.00028747220672475183,
      "loss": 1.8942,
      "step": 118265
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.9019925594329834,
      "learning_rate": 0.0002874681197319442,
      "loss": 2.1625,
      "step": 118266
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.952859103679657,
      "learning_rate": 0.00028746403274146625,
      "loss": 1.8861,
      "step": 118267
    },
    {
      "epoch": 1.54,
      "grad_norm": 1.0465925931930542,
      "learning_rate": 0.00028745994575331903,
      "loss": 2.0895,
      "step": 118268
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.9584780931472778,
      "learning_rate": 0.0002874558587675032,
      "loss": 2.0147,
      "step": 118269
    },
    {
      "epoch": 1.54,
      "grad_norm": 1.1309378147125244,
      "learning_rate": 0.00028745177178401965,
      "loss": 2.2833,
      "step": 118270
    },
    {
      "epoch": 1.54,
      "grad_norm": 1.060149073600769,
      "learning_rate": 0.000287447684802869,
      "loss": 2.0309,
      "step": 118271
    },
    {
      "epoch": 1.54,
      "grad_norm": 1.0053566694259644,
      "learning_rate": 0.0002874435978240522,
      "loss": 2.1769,
      "step": 118272
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.8952658176422119,
      "learning_rate": 0.0002874395108475697,
      "loss": 2.042,
      "step": 118273
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.8842048048973083,
      "learning_rate": 0.0002874354238734225,
      "loss": 2.2178,
      "step": 118274
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.9212412238121033,
      "learning_rate": 0.00028743133690161126,
      "loss": 1.94,
      "step": 118275
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.8904953002929688,
      "learning_rate": 0.00028742724993213677,
      "loss": 1.9762,
      "step": 118276
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.9280710816383362,
      "learning_rate": 0.00028742316296499976,
      "loss": 2.0553,
      "step": 118277
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.96196448802948,
      "learning_rate": 0.0002874190760002012,
      "loss": 1.9762,
      "step": 118278
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.9564528465270996,
      "learning_rate": 0.00028741498903774157,
      "loss": 2.1355,
      "step": 118279
    },
    {
      "epoch": 1.54,
      "grad_norm": 1.031382441520691,
      "learning_rate": 0.0002874109020776217,
      "loss": 2.5424,
      "step": 118280
    },
    {
      "epoch": 1.54,
      "grad_norm": 1.06987464427948,
      "learning_rate": 0.0002874068151198423,
      "loss": 2.1774,
      "step": 118281
    },
    {
      "epoch": 1.54,
      "grad_norm": 1.0284764766693115,
      "learning_rate": 0.00028740272816440434,
      "loss": 2.3123,
      "step": 118282
    },
    {
      "epoch": 1.54,
      "grad_norm": 1.0060763359069824,
      "learning_rate": 0.00028739864121130835,
      "loss": 1.9021,
      "step": 118283
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.9878363013267517,
      "learning_rate": 0.0002873945542605554,
      "loss": 2.218,
      "step": 118284
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.9883930087089539,
      "learning_rate": 0.00028739046731214585,
      "loss": 2.2889,
      "step": 118285
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.9944048523902893,
      "learning_rate": 0.0002873863803660807,
      "loss": 2.2327,
      "step": 118286
    },
    {
      "epoch": 1.54,
      "grad_norm": 1.0609053373336792,
      "learning_rate": 0.0002873822934223606,
      "loss": 2.0569,
      "step": 118287
    },
    {
      "epoch": 1.54,
      "grad_norm": 1.0016067028045654,
      "learning_rate": 0.0002873782064809864,
      "loss": 2.0832,
      "step": 118288
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.9262962937355042,
      "learning_rate": 0.0002873741195419588,
      "loss": 2.0034,
      "step": 118289
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.8874749541282654,
      "learning_rate": 0.00028737003260527873,
      "loss": 2.0357,
      "step": 118290
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.8867087960243225,
      "learning_rate": 0.0002873659456709467,
      "loss": 2.1676,
      "step": 118291
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.8658002614974976,
      "learning_rate": 0.0002873618587389636,
      "loss": 2.1085,
      "step": 118292
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.913032591342926,
      "learning_rate": 0.00028735777180933017,
      "loss": 2.0993,
      "step": 118293
    },
    {
      "epoch": 1.54,
      "grad_norm": 1.0826053619384766,
      "learning_rate": 0.00028735368488204714,
      "loss": 2.2133,
      "step": 118294
    },
    {
      "epoch": 1.54,
      "grad_norm": 1.0675300359725952,
      "learning_rate": 0.0002873495979571153,
      "loss": 1.8758,
      "step": 118295
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.9300210475921631,
      "learning_rate": 0.0002873455110345355,
      "loss": 2.099,
      "step": 118296
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.9603691697120667,
      "learning_rate": 0.00028734142411430833,
      "loss": 2.0677,
      "step": 118297
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.9752103090286255,
      "learning_rate": 0.0002873373371964346,
      "loss": 2.3707,
      "step": 118298
    },
    {
      "epoch": 1.54,
      "grad_norm": 1.1244100332260132,
      "learning_rate": 0.0002873332502809152,
      "loss": 2.1924,
      "step": 118299
    },
    {
      "epoch": 1.54,
      "grad_norm": 1.0212889909744263,
      "learning_rate": 0.0002873291633677507,
      "loss": 2.1148,
      "step": 118300
    },
    {
      "epoch": 1.54,
      "grad_norm": 1.016043782234192,
      "learning_rate": 0.00028732507645694196,
      "loss": 2.2779,
      "step": 118301
    },
    {
      "epoch": 1.54,
      "grad_norm": 1.0974351167678833,
      "learning_rate": 0.00028732098954848973,
      "loss": 2.4199,
      "step": 118302
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.8834677338600159,
      "learning_rate": 0.0002873169026423948,
      "loss": 2.0801,
      "step": 118303
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.9154866933822632,
      "learning_rate": 0.00028731281573865785,
      "loss": 2.2517,
      "step": 118304
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.9265117645263672,
      "learning_rate": 0.0002873087288372797,
      "loss": 2.2059,
      "step": 118305
    },
    {
      "epoch": 1.54,
      "grad_norm": 1.052791714668274,
      "learning_rate": 0.0002873046419382612,
      "loss": 1.944,
      "step": 118306
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.9055915474891663,
      "learning_rate": 0.00028730055504160284,
      "loss": 2.0202,
      "step": 118307
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.9907354712486267,
      "learning_rate": 0.0002872964681473056,
      "loss": 2.1922,
      "step": 118308
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.9841378927230835,
      "learning_rate": 0.00028729238125537024,
      "loss": 2.1746,
      "step": 118309
    },
    {
      "epoch": 1.54,
      "grad_norm": 1.019168496131897,
      "learning_rate": 0.00028728829436579743,
      "loss": 2.2243,
      "step": 118310
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.9319730401039124,
      "learning_rate": 0.0002872842074785879,
      "loss": 2.3746,
      "step": 118311
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.9519126415252686,
      "learning_rate": 0.00028728012059374257,
      "loss": 2.2588,
      "step": 118312
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.9843913316726685,
      "learning_rate": 0.0002872760337112621,
      "loss": 2.3046,
      "step": 118313
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.9339626431465149,
      "learning_rate": 0.00028727194683114725,
      "loss": 2.1324,
      "step": 118314
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.8805989623069763,
      "learning_rate": 0.0002872678599533988,
      "loss": 2.1845,
      "step": 118315
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.9285638332366943,
      "learning_rate": 0.00028726377307801746,
      "loss": 2.2116,
      "step": 118316
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.9462326169013977,
      "learning_rate": 0.000287259686205004,
      "loss": 2.1065,
      "step": 118317
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.9102781414985657,
      "learning_rate": 0.00028725559933435925,
      "loss": 2.1866,
      "step": 118318
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.9477202296257019,
      "learning_rate": 0.00028725151246608387,
      "loss": 1.9788,
      "step": 118319
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.9891822934150696,
      "learning_rate": 0.0002872474256001787,
      "loss": 2.3341,
      "step": 118320
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.9711847901344299,
      "learning_rate": 0.00028724333873664465,
      "loss": 2.2435,
      "step": 118321
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.9429212808609009,
      "learning_rate": 0.0002872392518754821,
      "loss": 2.1637,
      "step": 118322
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.9216672778129578,
      "learning_rate": 0.00028723516501669204,
      "loss": 2.168,
      "step": 118323
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.9432979822158813,
      "learning_rate": 0.0002872310781602752,
      "loss": 2.0838,
      "step": 118324
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.898368775844574,
      "learning_rate": 0.0002872269913062324,
      "loss": 2.4133,
      "step": 118325
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.9681910276412964,
      "learning_rate": 0.00028722290445456426,
      "loss": 1.8794,
      "step": 118326
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.9596271514892578,
      "learning_rate": 0.00028721881760527173,
      "loss": 2.2562,
      "step": 118327
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.9408015012741089,
      "learning_rate": 0.00028721473075835546,
      "loss": 2.1204,
      "step": 118328
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.9562188982963562,
      "learning_rate": 0.00028721064391381613,
      "loss": 2.1458,
      "step": 118329
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.9311678409576416,
      "learning_rate": 0.00028720655707165457,
      "loss": 2.1444,
      "step": 118330
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.9834824800491333,
      "learning_rate": 0.0002872024702318716,
      "loss": 2.1856,
      "step": 118331
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.9002274870872498,
      "learning_rate": 0.0002871983833944679,
      "loss": 1.895,
      "step": 118332
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.9134156703948975,
      "learning_rate": 0.0002871942965594444,
      "loss": 2.1026,
      "step": 118333
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.9785083532333374,
      "learning_rate": 0.00028719020972680157,
      "loss": 2.2046,
      "step": 118334
    },
    {
      "epoch": 1.54,
      "grad_norm": 1.1274569034576416,
      "learning_rate": 0.0002871861228965403,
      "loss": 2.1177,
      "step": 118335
    },
    {
      "epoch": 1.54,
      "grad_norm": 1.048545002937317,
      "learning_rate": 0.00028718203606866143,
      "loss": 1.9908,
      "step": 118336
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.9872000813484192,
      "learning_rate": 0.0002871779492431657,
      "loss": 2.0991,
      "step": 118337
    },
    {
      "epoch": 1.54,
      "grad_norm": 1.0201025009155273,
      "learning_rate": 0.00028717386242005377,
      "loss": 2.3092,
      "step": 118338
    },
    {
      "epoch": 1.54,
      "grad_norm": 1.0040781497955322,
      "learning_rate": 0.00028716977559932656,
      "loss": 2.3111,
      "step": 118339
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.8862898349761963,
      "learning_rate": 0.00028716568878098465,
      "loss": 2.1605,
      "step": 118340
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.8999859690666199,
      "learning_rate": 0.0002871616019650289,
      "loss": 2.0798,
      "step": 118341
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.9832891225814819,
      "learning_rate": 0.00028715751515145997,
      "loss": 2.1933,
      "step": 118342
    },
    {
      "epoch": 1.54,
      "grad_norm": 1.0949416160583496,
      "learning_rate": 0.0002871534283402788,
      "loss": 2.1737,
      "step": 118343
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.9057564735412598,
      "learning_rate": 0.000287149341531486,
      "loss": 2.4325,
      "step": 118344
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.8816041350364685,
      "learning_rate": 0.00028714525472508254,
      "loss": 2.1309,
      "step": 118345
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.905491828918457,
      "learning_rate": 0.0002871411679210688,
      "loss": 2.2147,
      "step": 118346
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.8934684991836548,
      "learning_rate": 0.0002871370811194458,
      "loss": 2.0129,
      "step": 118347
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.9502755403518677,
      "learning_rate": 0.00028713299432021433,
      "loss": 2.1129,
      "step": 118348
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.9591995477676392,
      "learning_rate": 0.000287128907523375,
      "loss": 2.1789,
      "step": 118349
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.983451783657074,
      "learning_rate": 0.0002871248207289287,
      "loss": 1.9209,
      "step": 118350
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.909914493560791,
      "learning_rate": 0.00028712073393687626,
      "loss": 2.1013,
      "step": 118351
    },
    {
      "epoch": 1.54,
      "grad_norm": 1.009799599647522,
      "learning_rate": 0.00028711664714721816,
      "loss": 2.2498,
      "step": 118352
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.8528057932853699,
      "learning_rate": 0.00028711256035995533,
      "loss": 1.9617,
      "step": 118353
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.9748383164405823,
      "learning_rate": 0.00028710847357508854,
      "loss": 2.2811,
      "step": 118354
    },
    {
      "epoch": 1.54,
      "grad_norm": 1.0068079233169556,
      "learning_rate": 0.00028710438679261854,
      "loss": 2.2293,
      "step": 118355
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.9899523258209229,
      "learning_rate": 0.00028710030001254604,
      "loss": 2.1874,
      "step": 118356
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.947172999382019,
      "learning_rate": 0.00028709621323487196,
      "loss": 2.1409,
      "step": 118357
    },
    {
      "epoch": 1.54,
      "grad_norm": 1.1083194017410278,
      "learning_rate": 0.00028709212645959683,
      "loss": 2.1143,
      "step": 118358
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.9091628193855286,
      "learning_rate": 0.0002870880396867215,
      "loss": 2.0694,
      "step": 118359
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.9320225119590759,
      "learning_rate": 0.00028708395291624677,
      "loss": 2.1577,
      "step": 118360
    },
    {
      "epoch": 1.54,
      "grad_norm": 1.03574800491333,
      "learning_rate": 0.0002870798661481734,
      "loss": 2.2518,
      "step": 118361
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.932006299495697,
      "learning_rate": 0.0002870757793825021,
      "loss": 2.1625,
      "step": 118362
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.9844324588775635,
      "learning_rate": 0.00028707169261923374,
      "loss": 1.9501,
      "step": 118363
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.9519510865211487,
      "learning_rate": 0.0002870676058583689,
      "loss": 2.108,
      "step": 118364
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.9946895241737366,
      "learning_rate": 0.00028706351909990846,
      "loss": 2.0748,
      "step": 118365
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.891321063041687,
      "learning_rate": 0.0002870594323438531,
      "loss": 1.9954,
      "step": 118366
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.9660400748252869,
      "learning_rate": 0.0002870553455902037,
      "loss": 2.1516,
      "step": 118367
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.8831123113632202,
      "learning_rate": 0.0002870512588389609,
      "loss": 2.0758,
      "step": 118368
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.8589890599250793,
      "learning_rate": 0.0002870471720901257,
      "loss": 2.3746,
      "step": 118369
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.8889155983924866,
      "learning_rate": 0.00028704308534369853,
      "loss": 2.2193,
      "step": 118370
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.9626953601837158,
      "learning_rate": 0.00028703899859968025,
      "loss": 2.2615,
      "step": 118371
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.9343000650405884,
      "learning_rate": 0.0002870349118580717,
      "loss": 2.1202,
      "step": 118372
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.8340107202529907,
      "learning_rate": 0.00028703082511887364,
      "loss": 2.3835,
      "step": 118373
    },
    {
      "epoch": 1.54,
      "grad_norm": 1.0784627199172974,
      "learning_rate": 0.00028702673838208677,
      "loss": 2.0463,
      "step": 118374
    },
    {
      "epoch": 1.54,
      "grad_norm": 1.014123797416687,
      "learning_rate": 0.00028702265164771187,
      "loss": 2.1781,
      "step": 118375
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.9090935587882996,
      "learning_rate": 0.0002870185649157498,
      "loss": 1.9574,
      "step": 118376
    },
    {
      "epoch": 1.54,
      "grad_norm": 1.007895827293396,
      "learning_rate": 0.0002870144781862011,
      "loss": 2.1385,
      "step": 118377
    },
    {
      "epoch": 1.54,
      "grad_norm": 1.0430470705032349,
      "learning_rate": 0.0002870103914590667,
      "loss": 2.0792,
      "step": 118378
    },
    {
      "epoch": 1.54,
      "grad_norm": 1.48096764087677,
      "learning_rate": 0.00028700630473434726,
      "loss": 2.0706,
      "step": 118379
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.9336510300636292,
      "learning_rate": 0.00028700221801204365,
      "loss": 2.2585,
      "step": 118380
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.9717904329299927,
      "learning_rate": 0.0002869981312921565,
      "loss": 2.1673,
      "step": 118381
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.9273600578308105,
      "learning_rate": 0.00028699404457468677,
      "loss": 2.2064,
      "step": 118382
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.8695375323295593,
      "learning_rate": 0.00028698995785963506,
      "loss": 2.0434,
      "step": 118383
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.9417768716812134,
      "learning_rate": 0.00028698587114700215,
      "loss": 2.1706,
      "step": 118384
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.9122042059898376,
      "learning_rate": 0.00028698178443678876,
      "loss": 1.8235,
      "step": 118385
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.9767046570777893,
      "learning_rate": 0.00028697769772899573,
      "loss": 2.4387,
      "step": 118386
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.8985824584960938,
      "learning_rate": 0.0002869736110236238,
      "loss": 2.1473,
      "step": 118387
    },
    {
      "epoch": 1.54,
      "grad_norm": 1.0488526821136475,
      "learning_rate": 0.0002869695243206738,
      "loss": 2.1757,
      "step": 118388
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.9734393954277039,
      "learning_rate": 0.00028696543762014626,
      "loss": 2.0049,
      "step": 118389
    },
    {
      "epoch": 1.54,
      "grad_norm": 1.0532809495925903,
      "learning_rate": 0.00028696135092204215,
      "loss": 2.0857,
      "step": 118390
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.9820287823677063,
      "learning_rate": 0.0002869572642263622,
      "loss": 2.2192,
      "step": 118391
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.936426043510437,
      "learning_rate": 0.00028695317753310714,
      "loss": 2.2668,
      "step": 118392
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.9859233498573303,
      "learning_rate": 0.00028694909084227765,
      "loss": 2.1868,
      "step": 118393
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.9815517067909241,
      "learning_rate": 0.0002869450041538747,
      "loss": 2.0975,
      "step": 118394
    },
    {
      "epoch": 1.54,
      "grad_norm": 1.0274933576583862,
      "learning_rate": 0.0002869409174678988,
      "loss": 1.955,
      "step": 118395
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.8819032311439514,
      "learning_rate": 0.0002869368307843509,
      "loss": 2.1499,
      "step": 118396
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.9432647824287415,
      "learning_rate": 0.00028693274410323166,
      "loss": 2.0798,
      "step": 118397
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.930787205696106,
      "learning_rate": 0.00028692865742454193,
      "loss": 2.2462,
      "step": 118398
    },
    {
      "epoch": 1.54,
      "grad_norm": 1.0074738264083862,
      "learning_rate": 0.00028692457074828233,
      "loss": 2.2109,
      "step": 118399
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.9880920052528381,
      "learning_rate": 0.00028692048407445383,
      "loss": 2.1729,
      "step": 118400
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.9970561861991882,
      "learning_rate": 0.00028691639740305693,
      "loss": 2.2279,
      "step": 118401
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.8589612245559692,
      "learning_rate": 0.0002869123107340925,
      "loss": 2.0787,
      "step": 118402
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.8549020886421204,
      "learning_rate": 0.0002869082240675613,
      "loss": 2.0175,
      "step": 118403
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.8931155800819397,
      "learning_rate": 0.0002869041374034642,
      "loss": 2.1752,
      "step": 118404
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.9378770589828491,
      "learning_rate": 0.0002869000507418019,
      "loss": 1.9806,
      "step": 118405
    },
    {
      "epoch": 1.54,
      "grad_norm": 1.1076104640960693,
      "learning_rate": 0.0002868959640825751,
      "loss": 1.8007,
      "step": 118406
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.9856853485107422,
      "learning_rate": 0.00028689187742578454,
      "loss": 2.2065,
      "step": 118407
    },
    {
      "epoch": 1.54,
      "grad_norm": 1.0341449975967407,
      "learning_rate": 0.000286887790771431,
      "loss": 2.224,
      "step": 118408
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.9356507062911987,
      "learning_rate": 0.00028688370411951534,
      "loss": 2.3237,
      "step": 118409
    },
    {
      "epoch": 1.54,
      "grad_norm": 1.1519393920898438,
      "learning_rate": 0.00028687961747003824,
      "loss": 2.1331,
      "step": 118410
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.9460141658782959,
      "learning_rate": 0.00028687553082300043,
      "loss": 1.8241,
      "step": 118411
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.9529151916503906,
      "learning_rate": 0.00028687144417840284,
      "loss": 2.1034,
      "step": 118412
    },
    {
      "epoch": 1.54,
      "grad_norm": 1.0166375637054443,
      "learning_rate": 0.0002868673575362459,
      "loss": 2.2575,
      "step": 118413
    },
    {
      "epoch": 1.54,
      "grad_norm": 1.0682692527770996,
      "learning_rate": 0.0002868632708965306,
      "loss": 2.2342,
      "step": 118414
    },
    {
      "epoch": 1.54,
      "grad_norm": 1.1320075988769531,
      "learning_rate": 0.00028685918425925774,
      "loss": 2.1142,
      "step": 118415
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.8882384896278381,
      "learning_rate": 0.00028685509762442794,
      "loss": 1.953,
      "step": 118416
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.9773971438407898,
      "learning_rate": 0.0002868510109920421,
      "loss": 2.2246,
      "step": 118417
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.9803101420402527,
      "learning_rate": 0.00028684692436210097,
      "loss": 2.0131,
      "step": 118418
    },
    {
      "epoch": 1.54,
      "grad_norm": 1.090331792831421,
      "learning_rate": 0.00028684283773460514,
      "loss": 2.1856,
      "step": 118419
    },
    {
      "epoch": 1.54,
      "grad_norm": 1.1127216815948486,
      "learning_rate": 0.00028683875110955545,
      "loss": 1.8904,
      "step": 118420
    },
    {
      "epoch": 1.54,
      "grad_norm": 1.0944674015045166,
      "learning_rate": 0.0002868346644869527,
      "loss": 1.9426,
      "step": 118421
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.988399863243103,
      "learning_rate": 0.0002868305778667976,
      "loss": 2.0984,
      "step": 118422
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.9469767808914185,
      "learning_rate": 0.00028682649124909095,
      "loss": 2.1769,
      "step": 118423
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.8846806287765503,
      "learning_rate": 0.00028682240463383365,
      "loss": 2.3046,
      "step": 118424
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.9714887142181396,
      "learning_rate": 0.0002868183180210262,
      "loss": 2.0124,
      "step": 118425
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.9035000801086426,
      "learning_rate": 0.00028681423141066947,
      "loss": 2.0517,
      "step": 118426
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.8613132834434509,
      "learning_rate": 0.00028681014480276426,
      "loss": 1.9424,
      "step": 118427
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.9973470568656921,
      "learning_rate": 0.00028680605819731125,
      "loss": 2.2872,
      "step": 118428
    },
    {
      "epoch": 1.54,
      "grad_norm": 1.1582982540130615,
      "learning_rate": 0.00028680197159431124,
      "loss": 2.1679,
      "step": 118429
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.9404825568199158,
      "learning_rate": 0.0002867978849937651,
      "loss": 2.0519,
      "step": 118430
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.9235498309135437,
      "learning_rate": 0.0002867937983956734,
      "loss": 2.168,
      "step": 118431
    },
    {
      "epoch": 1.54,
      "grad_norm": 1.316182255744934,
      "learning_rate": 0.00028678971180003696,
      "loss": 2.2234,
      "step": 118432
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.9069471955299377,
      "learning_rate": 0.00028678562520685657,
      "loss": 2.2503,
      "step": 118433
    },
    {
      "epoch": 1.54,
      "grad_norm": 1.1463481187820435,
      "learning_rate": 0.000286781538616133,
      "loss": 2.1073,
      "step": 118434
    },
    {
      "epoch": 1.54,
      "grad_norm": 1.2856003046035767,
      "learning_rate": 0.0002867774520278669,
      "loss": 2.2725,
      "step": 118435
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.9422720074653625,
      "learning_rate": 0.00028677336544205936,
      "loss": 2.1632,
      "step": 118436
    },
    {
      "epoch": 1.54,
      "grad_norm": 1.144150972366333,
      "learning_rate": 0.00028676927885871073,
      "loss": 2.118,
      "step": 118437
    },
    {
      "epoch": 1.54,
      "grad_norm": 1.1008760929107666,
      "learning_rate": 0.00028676519227782194,
      "loss": 2.3891,
      "step": 118438
    },
    {
      "epoch": 1.54,
      "grad_norm": 1.0728163719177246,
      "learning_rate": 0.00028676110569939375,
      "loss": 2.0257,
      "step": 118439
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.9779344797134399,
      "learning_rate": 0.0002867570191234269,
      "loss": 2.1275,
      "step": 118440
    },
    {
      "epoch": 1.54,
      "grad_norm": 1.0032769441604614,
      "learning_rate": 0.0002867529325499222,
      "loss": 2.1788,
      "step": 118441
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.9293192028999329,
      "learning_rate": 0.00028674884597888035,
      "loss": 1.9507,
      "step": 118442
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.9452452063560486,
      "learning_rate": 0.0002867447594103023,
      "loss": 2.1826,
      "step": 118443
    },
    {
      "epoch": 1.54,
      "grad_norm": 1.0067952871322632,
      "learning_rate": 0.0002867406728441885,
      "loss": 2.0663,
      "step": 118444
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.9852485656738281,
      "learning_rate": 0.0002867365862805399,
      "loss": 2.2881,
      "step": 118445
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.9650397300720215,
      "learning_rate": 0.0002867324997193572,
      "loss": 2.1433,
      "step": 118446
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.9437265396118164,
      "learning_rate": 0.0002867284131606411,
      "loss": 2.0898,
      "step": 118447
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.9629803895950317,
      "learning_rate": 0.00028672432660439254,
      "loss": 2.0457,
      "step": 118448
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.8883817195892334,
      "learning_rate": 0.0002867202400506123,
      "loss": 2.1187,
      "step": 118449
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.9406081438064575,
      "learning_rate": 0.00028671615349930085,
      "loss": 2.2417,
      "step": 118450
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.9079596400260925,
      "learning_rate": 0.0002867120669504591,
      "loss": 2.0764,
      "step": 118451
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.9171305298805237,
      "learning_rate": 0.00028670798040408786,
      "loss": 2.389,
      "step": 118452
    },
    {
      "epoch": 1.54,
      "grad_norm": 1.0857617855072021,
      "learning_rate": 0.0002867038938601879,
      "loss": 2.1587,
      "step": 118453
    },
    {
      "epoch": 1.54,
      "grad_norm": 1.1000730991363525,
      "learning_rate": 0.0002866998073187598,
      "loss": 2.0097,
      "step": 118454
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.8975761532783508,
      "learning_rate": 0.00028669572077980477,
      "loss": 2.252,
      "step": 118455
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.866890013217926,
      "learning_rate": 0.000286691634243323,
      "loss": 1.9527,
      "step": 118456
    },
    {
      "epoch": 1.54,
      "grad_norm": 1.3067954778671265,
      "learning_rate": 0.00028668754770931555,
      "loss": 1.8856,
      "step": 118457
    },
    {
      "epoch": 1.54,
      "grad_norm": 1.0425468683242798,
      "learning_rate": 0.0002866834611777831,
      "loss": 1.9147,
      "step": 118458
    },
    {
      "epoch": 1.54,
      "grad_norm": 1.0576598644256592,
      "learning_rate": 0.00028667937464872645,
      "loss": 2.1289,
      "step": 118459
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.8951526284217834,
      "learning_rate": 0.00028667528812214644,
      "loss": 2.25,
      "step": 118460
    },
    {
      "epoch": 1.54,
      "grad_norm": 1.3541914224624634,
      "learning_rate": 0.0002866712015980438,
      "loss": 1.9384,
      "step": 118461
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.9540132880210876,
      "learning_rate": 0.0002866671150764191,
      "loss": 2.2035,
      "step": 118462
    },
    {
      "epoch": 1.54,
      "grad_norm": 1.0080766677856445,
      "learning_rate": 0.0002866630285572733,
      "loss": 2.2468,
      "step": 118463
    },
    {
      "epoch": 1.54,
      "grad_norm": 1.0987414121627808,
      "learning_rate": 0.000286658942040607,
      "loss": 2.1154,
      "step": 118464
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.9920291304588318,
      "learning_rate": 0.0002866548555264211,
      "loss": 2.0914,
      "step": 118465
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.908602774143219,
      "learning_rate": 0.00028665076901471633,
      "loss": 2.0775,
      "step": 118466
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.9713576436042786,
      "learning_rate": 0.0002866466825054935,
      "loss": 2.1447,
      "step": 118467
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.9363119602203369,
      "learning_rate": 0.00028664259599875326,
      "loss": 2.1374,
      "step": 118468
    },
    {
      "epoch": 1.54,
      "grad_norm": 1.1735906600952148,
      "learning_rate": 0.0002866385094944964,
      "loss": 2.1387,
      "step": 118469
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.9861127138137817,
      "learning_rate": 0.0002866344229927237,
      "loss": 2.148,
      "step": 118470
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.875378429889679,
      "learning_rate": 0.0002866303364934359,
      "loss": 2.2812,
      "step": 118471
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.9668111205101013,
      "learning_rate": 0.0002866262499966337,
      "loss": 1.9967,
      "step": 118472
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.939923882484436,
      "learning_rate": 0.00028662216350231807,
      "loss": 2.0383,
      "step": 118473
    },
    {
      "epoch": 1.54,
      "grad_norm": 1.0933876037597656,
      "learning_rate": 0.00028661807701048955,
      "loss": 2.0182,
      "step": 118474
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.9551980495452881,
      "learning_rate": 0.00028661399052114906,
      "loss": 2.2261,
      "step": 118475
    },
    {
      "epoch": 1.54,
      "grad_norm": 1.017989158630371,
      "learning_rate": 0.00028660990403429724,
      "loss": 2.2103,
      "step": 118476
    },
    {
      "epoch": 1.54,
      "grad_norm": 1.0979801416397095,
      "learning_rate": 0.0002866058175499348,
      "loss": 2.1065,
      "step": 118477
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.9157622456550598,
      "learning_rate": 0.0002866017310680627,
      "loss": 2.2936,
      "step": 118478
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.9851186275482178,
      "learning_rate": 0.0002865976445886816,
      "loss": 2.2897,
      "step": 118479
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.9503424763679504,
      "learning_rate": 0.0002865935581117922,
      "loss": 2.2447,
      "step": 118480
    },
    {
      "epoch": 1.54,
      "grad_norm": 1.1574245691299438,
      "learning_rate": 0.0002865894716373953,
      "loss": 2.2472,
      "step": 118481
    },
    {
      "epoch": 1.54,
      "grad_norm": 1.0152308940887451,
      "learning_rate": 0.00028658538516549163,
      "loss": 2.0395,
      "step": 118482
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.9661889672279358,
      "learning_rate": 0.00028658129869608214,
      "loss": 2.0694,
      "step": 118483
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.878807544708252,
      "learning_rate": 0.00028657721222916734,
      "loss": 2.164,
      "step": 118484
    },
    {
      "epoch": 1.54,
      "grad_norm": 1.04689621925354,
      "learning_rate": 0.0002865731257647481,
      "loss": 2.2161,
      "step": 118485
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.957146406173706,
      "learning_rate": 0.0002865690393028252,
      "loss": 2.2846,
      "step": 118486
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.9374030232429504,
      "learning_rate": 0.0002865649528433993,
      "loss": 2.3313,
      "step": 118487
    },
    {
      "epoch": 1.54,
      "grad_norm": 1.1051386594772339,
      "learning_rate": 0.00028656086638647124,
      "loss": 2.1858,
      "step": 118488
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.994792640209198,
      "learning_rate": 0.0002865567799320418,
      "loss": 2.4042,
      "step": 118489
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.9571700692176819,
      "learning_rate": 0.00028655269348011174,
      "loss": 2.4151,
      "step": 118490
    },
    {
      "epoch": 1.54,
      "grad_norm": 1.0162544250488281,
      "learning_rate": 0.0002865486070306818,
      "loss": 2.0463,
      "step": 118491
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.9540030360221863,
      "learning_rate": 0.00028654452058375266,
      "loss": 2.1777,
      "step": 118492
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.9939121007919312,
      "learning_rate": 0.00028654043413932513,
      "loss": 2.103,
      "step": 118493
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.9946669340133667,
      "learning_rate": 0.00028653634769740006,
      "loss": 2.2347,
      "step": 118494
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.9278519153594971,
      "learning_rate": 0.0002865322612579781,
      "loss": 2.0782,
      "step": 118495
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.9288162589073181,
      "learning_rate": 0.00028652817482106,
      "loss": 2.1579,
      "step": 118496
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.9277297854423523,
      "learning_rate": 0.0002865240883866467,
      "loss": 1.7576,
      "step": 118497
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.9137676358222961,
      "learning_rate": 0.00028652000195473874,
      "loss": 1.9175,
      "step": 118498
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.9024541974067688,
      "learning_rate": 0.00028651591552533696,
      "loss": 2.2436,
      "step": 118499
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.8811225295066833,
      "learning_rate": 0.0002865118290984421,
      "loss": 2.0781,
      "step": 118500
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.8947414755821228,
      "learning_rate": 0.000286507742674055,
      "loss": 2.1377,
      "step": 118501
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.9644806385040283,
      "learning_rate": 0.0002865036562521763,
      "loss": 2.2201,
      "step": 118502
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.9072245955467224,
      "learning_rate": 0.000286499569832807,
      "loss": 2.1204,
      "step": 118503
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.9598304033279419,
      "learning_rate": 0.0002864954834159475,
      "loss": 1.9967,
      "step": 118504
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.9510418176651001,
      "learning_rate": 0.00028649139700159885,
      "loss": 2.2625,
      "step": 118505
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.9598342776298523,
      "learning_rate": 0.00028648731058976163,
      "loss": 2.0317,
      "step": 118506
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.8974650502204895,
      "learning_rate": 0.00028648322418043667,
      "loss": 2.0063,
      "step": 118507
    },
    {
      "epoch": 1.54,
      "grad_norm": 1.0991712808609009,
      "learning_rate": 0.0002864791377736247,
      "loss": 2.1109,
      "step": 118508
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.8754262328147888,
      "learning_rate": 0.0002864750513693266,
      "loss": 1.9855,
      "step": 118509
    },
    {
      "epoch": 1.54,
      "grad_norm": 1.032576322555542,
      "learning_rate": 0.00028647096496754314,
      "loss": 2.026,
      "step": 118510
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.9272250533103943,
      "learning_rate": 0.00028646687856827486,
      "loss": 2.1303,
      "step": 118511
    },
    {
      "epoch": 1.54,
      "grad_norm": 1.0554933547973633,
      "learning_rate": 0.0002864627921715226,
      "loss": 2.2007,
      "step": 118512
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.9872072339057922,
      "learning_rate": 0.00028645870577728716,
      "loss": 1.9569,
      "step": 118513
    },
    {
      "epoch": 1.54,
      "grad_norm": 1.038029670715332,
      "learning_rate": 0.00028645461938556937,
      "loss": 2.1208,
      "step": 118514
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.9194290637969971,
      "learning_rate": 0.00028645053299636986,
      "loss": 1.9419,
      "step": 118515
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.9392915964126587,
      "learning_rate": 0.0002864464466096896,
      "loss": 1.9538,
      "step": 118516
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.8759381175041199,
      "learning_rate": 0.00028644236022552907,
      "loss": 2.1231,
      "step": 118517
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.8877946734428406,
      "learning_rate": 0.0002864382738438892,
      "loss": 2.0705,
      "step": 118518
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.9412944912910461,
      "learning_rate": 0.00028643418746477066,
      "loss": 1.9563,
      "step": 118519
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.9282141923904419,
      "learning_rate": 0.0002864301010881743,
      "loss": 2.0377,
      "step": 118520
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.9232560992240906,
      "learning_rate": 0.0002864260147141008,
      "loss": 2.4739,
      "step": 118521
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.8739407062530518,
      "learning_rate": 0.00028642192834255113,
      "loss": 2.0233,
      "step": 118522
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.8915998935699463,
      "learning_rate": 0.0002864178419735257,
      "loss": 2.3221,
      "step": 118523
    },
    {
      "epoch": 1.54,
      "grad_norm": 1.0161324739456177,
      "learning_rate": 0.0002864137556070255,
      "loss": 2.2124,
      "step": 118524
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.9266788959503174,
      "learning_rate": 0.0002864096692430512,
      "loss": 2.1658,
      "step": 118525
    },
    {
      "epoch": 1.54,
      "grad_norm": 1.0395926237106323,
      "learning_rate": 0.0002864055828816036,
      "loss": 2.2546,
      "step": 118526
    },
    {
      "epoch": 1.54,
      "grad_norm": 1.2307982444763184,
      "learning_rate": 0.0002864014965226835,
      "loss": 2.0241,
      "step": 118527
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.9705380201339722,
      "learning_rate": 0.0002863974101662917,
      "loss": 2.3264,
      "step": 118528
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.9199811816215515,
      "learning_rate": 0.0002863933238124288,
      "loss": 2.1769,
      "step": 118529
    },
    {
      "epoch": 1.54,
      "grad_norm": 1.0597480535507202,
      "learning_rate": 0.0002863892374610956,
      "loss": 2.1371,
      "step": 118530
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.9950581789016724,
      "learning_rate": 0.00028638515111229293,
      "loss": 2.2104,
      "step": 118531
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.932953417301178,
      "learning_rate": 0.0002863810647660215,
      "loss": 1.8734,
      "step": 118532
    },
    {
      "epoch": 1.54,
      "grad_norm": 1.0288738012313843,
      "learning_rate": 0.0002863769784222821,
      "loss": 2.3403,
      "step": 118533
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.922230064868927,
      "learning_rate": 0.0002863728920810756,
      "loss": 2.2675,
      "step": 118534
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.8172187805175781,
      "learning_rate": 0.00028636880574240244,
      "loss": 2.1334,
      "step": 118535
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.9771621227264404,
      "learning_rate": 0.00028636471940626367,
      "loss": 1.7935,
      "step": 118536
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.9350075721740723,
      "learning_rate": 0.00028636063307265993,
      "loss": 2.1881,
      "step": 118537
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.980282723903656,
      "learning_rate": 0.000286356546741592,
      "loss": 2.1565,
      "step": 118538
    },
    {
      "epoch": 1.54,
      "grad_norm": 1.0247942209243774,
      "learning_rate": 0.0002863524604130606,
      "loss": 2.2199,
      "step": 118539
    },
    {
      "epoch": 1.54,
      "grad_norm": 1.1437824964523315,
      "learning_rate": 0.00028634837408706673,
      "loss": 2.1772,
      "step": 118540
    },
    {
      "epoch": 1.54,
      "grad_norm": 1.052403211593628,
      "learning_rate": 0.0002863442877636108,
      "loss": 2.1368,
      "step": 118541
    },
    {
      "epoch": 1.54,
      "grad_norm": 1.4011492729187012,
      "learning_rate": 0.0002863402014426937,
      "loss": 2.041,
      "step": 118542
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.984374463558197,
      "learning_rate": 0.00028633611512431627,
      "loss": 2.3477,
      "step": 118543
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.9802259206771851,
      "learning_rate": 0.0002863320288084792,
      "loss": 2.052,
      "step": 118544
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.9006304144859314,
      "learning_rate": 0.0002863279424951832,
      "loss": 2.1258,
      "step": 118545
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.9443951845169067,
      "learning_rate": 0.0002863238561844293,
      "loss": 2.1634,
      "step": 118546
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.8393167853355408,
      "learning_rate": 0.0002863197698762179,
      "loss": 1.9298,
      "step": 118547
    },
    {
      "epoch": 1.54,
      "grad_norm": 1.0240870714187622,
      "learning_rate": 0.00028631568357054987,
      "loss": 2.0951,
      "step": 118548
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.8440638184547424,
      "learning_rate": 0.0002863115972674261,
      "loss": 2.2597,
      "step": 118549
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.9909181594848633,
      "learning_rate": 0.00028630751096684714,
      "loss": 2.118,
      "step": 118550
    },
    {
      "epoch": 1.54,
      "grad_norm": 1.0096691846847534,
      "learning_rate": 0.00028630342466881396,
      "loss": 2.1696,
      "step": 118551
    },
    {
      "epoch": 1.54,
      "grad_norm": 1.1237930059432983,
      "learning_rate": 0.0002862993383733273,
      "loss": 2.1895,
      "step": 118552
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.9737313389778137,
      "learning_rate": 0.00028629525208038786,
      "loss": 1.9893,
      "step": 118553
    },
    {
      "epoch": 1.54,
      "grad_norm": 1.3005801439285278,
      "learning_rate": 0.0002862911657899963,
      "loss": 2.1051,
      "step": 118554
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.913444459438324,
      "learning_rate": 0.0002862870795021535,
      "loss": 1.9518,
      "step": 118555
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.9640863537788391,
      "learning_rate": 0.0002862829932168601,
      "loss": 2.2236,
      "step": 118556
    },
    {
      "epoch": 1.54,
      "grad_norm": 1.0792205333709717,
      "learning_rate": 0.000286278906934117,
      "loss": 2.2262,
      "step": 118557
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.894632875919342,
      "learning_rate": 0.000286274820653925,
      "loss": 2.129,
      "step": 118558
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.9577949047088623,
      "learning_rate": 0.0002862707343762847,
      "loss": 2.0688,
      "step": 118559
    },
    {
      "epoch": 1.54,
      "grad_norm": 1.0041652917861938,
      "learning_rate": 0.00028626664810119703,
      "loss": 2.0506,
      "step": 118560
    },
    {
      "epoch": 1.54,
      "grad_norm": 1.0245352983474731,
      "learning_rate": 0.0002862625618286625,
      "loss": 2.3434,
      "step": 118561
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.9493658542633057,
      "learning_rate": 0.00028625847555868207,
      "loss": 2.3401,
      "step": 118562
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.936881422996521,
      "learning_rate": 0.0002862543892912564,
      "loss": 2.0846,
      "step": 118563
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.8891793489456177,
      "learning_rate": 0.0002862503030263864,
      "loss": 2.1192,
      "step": 118564
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.9537636637687683,
      "learning_rate": 0.0002862462167640727,
      "loss": 1.9586,
      "step": 118565
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.9582133889198303,
      "learning_rate": 0.00028624213050431604,
      "loss": 2.0643,
      "step": 118566
    },
    {
      "epoch": 1.54,
      "grad_norm": 1.0216608047485352,
      "learning_rate": 0.00028623804424711733,
      "loss": 2.129,
      "step": 118567
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.9349350333213806,
      "learning_rate": 0.0002862339579924771,
      "loss": 2.2365,
      "step": 118568
    },
    {
      "epoch": 1.54,
      "grad_norm": 1.0462297201156616,
      "learning_rate": 0.00028622987174039627,
      "loss": 2.0036,
      "step": 118569
    },
    {
      "epoch": 1.54,
      "grad_norm": 1.3957178592681885,
      "learning_rate": 0.00028622578549087566,
      "loss": 2.1391,
      "step": 118570
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.9413850903511047,
      "learning_rate": 0.0002862216992439159,
      "loss": 2.2198,
      "step": 118571
    },
    {
      "epoch": 1.54,
      "grad_norm": 1.3395332098007202,
      "learning_rate": 0.0002862176129995177,
      "loss": 2.187,
      "step": 118572
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.942168116569519,
      "learning_rate": 0.0002862135267576819,
      "loss": 2.026,
      "step": 118573
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.9532875418663025,
      "learning_rate": 0.00028620944051840934,
      "loss": 2.2128,
      "step": 118574
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.8583141565322876,
      "learning_rate": 0.00028620535428170075,
      "loss": 2.1962,
      "step": 118575
    },
    {
      "epoch": 1.54,
      "grad_norm": 1.0502668619155884,
      "learning_rate": 0.00028620126804755673,
      "loss": 2.0634,
      "step": 118576
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.9158275723457336,
      "learning_rate": 0.00028619718181597826,
      "loss": 2.0825,
      "step": 118577
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.8841516971588135,
      "learning_rate": 0.000286193095586966,
      "loss": 2.0549,
      "step": 118578
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.9778123497962952,
      "learning_rate": 0.00028618900936052056,
      "loss": 2.1405,
      "step": 118579
    },
    {
      "epoch": 1.54,
      "grad_norm": 1.1279500722885132,
      "learning_rate": 0.0002861849231366429,
      "loss": 1.9497,
      "step": 118580
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.9813212752342224,
      "learning_rate": 0.0002861808369153338,
      "loss": 2.1909,
      "step": 118581
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.9990743398666382,
      "learning_rate": 0.00028617675069659394,
      "loss": 2.157,
      "step": 118582
    },
    {
      "epoch": 1.54,
      "grad_norm": 1.0247585773468018,
      "learning_rate": 0.00028617266448042404,
      "loss": 2.2354,
      "step": 118583
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.9382991194725037,
      "learning_rate": 0.0002861685782668249,
      "loss": 2.2173,
      "step": 118584
    },
    {
      "epoch": 1.54,
      "grad_norm": 1.0709798336029053,
      "learning_rate": 0.00028616449205579725,
      "loss": 2.2409,
      "step": 118585
    },
    {
      "epoch": 1.54,
      "grad_norm": 1.0651122331619263,
      "learning_rate": 0.00028616040584734194,
      "loss": 2.03,
      "step": 118586
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.9985170364379883,
      "learning_rate": 0.00028615631964145965,
      "loss": 2.212,
      "step": 118587
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.874585747718811,
      "learning_rate": 0.0002861522334381511,
      "loss": 2.0504,
      "step": 118588
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.8623908162117004,
      "learning_rate": 0.0002861481472374173,
      "loss": 2.0098,
      "step": 118589
    },
    {
      "epoch": 1.54,
      "grad_norm": 1.089263677597046,
      "learning_rate": 0.00028614406103925867,
      "loss": 2.1452,
      "step": 118590
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.8378826379776001,
      "learning_rate": 0.00028613997484367614,
      "loss": 2.0496,
      "step": 118591
    },
    {
      "epoch": 1.54,
      "grad_norm": 1.044700264930725,
      "learning_rate": 0.00028613588865067044,
      "loss": 2.1164,
      "step": 118592
    },
    {
      "epoch": 1.54,
      "grad_norm": 1.0506460666656494,
      "learning_rate": 0.0002861318024602423,
      "loss": 2.1627,
      "step": 118593
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.9155508875846863,
      "learning_rate": 0.0002861277162723926,
      "loss": 2.0641,
      "step": 118594
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.9353759288787842,
      "learning_rate": 0.0002861236300871221,
      "loss": 2.1041,
      "step": 118595
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.8859086036682129,
      "learning_rate": 0.00028611954390443136,
      "loss": 2.2679,
      "step": 118596
    },
    {
      "epoch": 1.54,
      "grad_norm": 1.070271372795105,
      "learning_rate": 0.00028611545772432126,
      "loss": 2.4261,
      "step": 118597
    },
    {
      "epoch": 1.54,
      "grad_norm": 1.0656957626342773,
      "learning_rate": 0.00028611137154679254,
      "loss": 1.8996,
      "step": 118598
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.9910061955451965,
      "learning_rate": 0.000286107285371846,
      "loss": 2.1452,
      "step": 118599
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.9526838660240173,
      "learning_rate": 0.0002861031991994824,
      "loss": 1.9249,
      "step": 118600
    },
    {
      "epoch": 1.54,
      "grad_norm": 1.123348355293274,
      "learning_rate": 0.0002860991130297025,
      "loss": 2.1394,
      "step": 118601
    },
    {
      "epoch": 1.54,
      "grad_norm": 1.1299067735671997,
      "learning_rate": 0.000286095026862507,
      "loss": 1.9543,
      "step": 118602
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.9075714945793152,
      "learning_rate": 0.0002860909406978967,
      "loss": 2.2915,
      "step": 118603
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.9394592642784119,
      "learning_rate": 0.00028608685453587236,
      "loss": 2.1056,
      "step": 118604
    },
    {
      "epoch": 1.54,
      "grad_norm": 1.034609079360962,
      "learning_rate": 0.0002860827683764347,
      "loss": 2.0607,
      "step": 118605
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.962388813495636,
      "learning_rate": 0.00028607868221958453,
      "loss": 2.1043,
      "step": 118606
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.9522457718849182,
      "learning_rate": 0.00028607459606532274,
      "loss": 2.2497,
      "step": 118607
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.8754509091377258,
      "learning_rate": 0.0002860705099136498,
      "loss": 1.9389,
      "step": 118608
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.9694110155105591,
      "learning_rate": 0.0002860664237645666,
      "loss": 2.1646,
      "step": 118609
    },
    {
      "epoch": 1.54,
      "grad_norm": 1.0345972776412964,
      "learning_rate": 0.0002860623376180739,
      "loss": 2.1085,
      "step": 118610
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.8561512231826782,
      "learning_rate": 0.00028605825147417254,
      "loss": 2.0542,
      "step": 118611
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.9509490132331848,
      "learning_rate": 0.00028605416533286317,
      "loss": 2.2582,
      "step": 118612
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.8426785469055176,
      "learning_rate": 0.0002860500791941468,
      "loss": 1.9544,
      "step": 118613
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.920833170413971,
      "learning_rate": 0.00028604599305802374,
      "loss": 1.9771,
      "step": 118614
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.9033729434013367,
      "learning_rate": 0.00028604190692449504,
      "loss": 2.1225,
      "step": 118615
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.907351016998291,
      "learning_rate": 0.0002860378207935614,
      "loss": 2.0997,
      "step": 118616
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.9145025014877319,
      "learning_rate": 0.0002860337346652236,
      "loss": 2.0131,
      "step": 118617
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.89630526304245,
      "learning_rate": 0.0002860296485394824,
      "loss": 2.366,
      "step": 118618
    },
    {
      "epoch": 1.54,
      "grad_norm": 1.0454063415527344,
      "learning_rate": 0.0002860255624163387,
      "loss": 2.0647,
      "step": 118619
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.9303333163261414,
      "learning_rate": 0.00028602147629579297,
      "loss": 1.895,
      "step": 118620
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.9330611228942871,
      "learning_rate": 0.0002860173901778461,
      "loss": 2.1582,
      "step": 118621
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.9868395924568176,
      "learning_rate": 0.00028601330406249887,
      "loss": 2.2616,
      "step": 118622
    },
    {
      "epoch": 1.54,
      "grad_norm": 1.235935091972351,
      "learning_rate": 0.00028600921794975204,
      "loss": 2.1146,
      "step": 118623
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.8952297568321228,
      "learning_rate": 0.0002860051318396063,
      "loss": 1.9899,
      "step": 118624
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.9129303097724915,
      "learning_rate": 0.00028600104573206264,
      "loss": 2.2554,
      "step": 118625
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.9494999051094055,
      "learning_rate": 0.00028599695962712153,
      "loss": 2.0717,
      "step": 118626
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.9225987792015076,
      "learning_rate": 0.00028599287352478386,
      "loss": 1.9152,
      "step": 118627
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.967647135257721,
      "learning_rate": 0.0002859887874250504,
      "loss": 2.1479,
      "step": 118628
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.9537146687507629,
      "learning_rate": 0.0002859847013279218,
      "loss": 2.1714,
      "step": 118629
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.9581893682479858,
      "learning_rate": 0.00028598061523339897,
      "loss": 2.1718,
      "step": 118630
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.9455326199531555,
      "learning_rate": 0.00028597652914148276,
      "loss": 1.9745,
      "step": 118631
    },
    {
      "epoch": 1.54,
      "grad_norm": 1.0322540998458862,
      "learning_rate": 0.0002859724430521736,
      "loss": 2.3133,
      "step": 118632
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.98521488904953,
      "learning_rate": 0.0002859683569654725,
      "loss": 2.3278,
      "step": 118633
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.9576252698898315,
      "learning_rate": 0.00028596427088138003,
      "loss": 2.0768,
      "step": 118634
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.8862739205360413,
      "learning_rate": 0.00028596018479989715,
      "loss": 2.1576,
      "step": 118635
    },
    {
      "epoch": 1.54,
      "grad_norm": 1.0416758060455322,
      "learning_rate": 0.00028595609872102454,
      "loss": 2.2027,
      "step": 118636
    },
    {
      "epoch": 1.54,
      "grad_norm": 1.1614288091659546,
      "learning_rate": 0.000285952012644763,
      "loss": 2.0704,
      "step": 118637
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.9302770495414734,
      "learning_rate": 0.00028594792657111323,
      "loss": 2.171,
      "step": 118638
    },
    {
      "epoch": 1.54,
      "grad_norm": 1.1687085628509521,
      "learning_rate": 0.00028594384050007593,
      "loss": 1.9817,
      "step": 118639
    },
    {
      "epoch": 1.54,
      "grad_norm": 1.0397206544876099,
      "learning_rate": 0.00028593975443165194,
      "loss": 2.3169,
      "step": 118640
    },
    {
      "epoch": 1.54,
      "grad_norm": 1.006935477256775,
      "learning_rate": 0.0002859356683658421,
      "loss": 1.945,
      "step": 118641
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.8980159163475037,
      "learning_rate": 0.000285931582302647,
      "loss": 1.8731,
      "step": 118642
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.8802427649497986,
      "learning_rate": 0.0002859274962420675,
      "loss": 2.2604,
      "step": 118643
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.971716582775116,
      "learning_rate": 0.00028592341018410445,
      "loss": 2.2227,
      "step": 118644
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.9619844555854797,
      "learning_rate": 0.00028591932412875843,
      "loss": 2.2264,
      "step": 118645
    },
    {
      "epoch": 1.54,
      "grad_norm": 1.0618667602539062,
      "learning_rate": 0.0002859152380760303,
      "loss": 2.1913,
      "step": 118646
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.930880069732666,
      "learning_rate": 0.0002859111520259207,
      "loss": 2.2933,
      "step": 118647
    },
    {
      "epoch": 1.54,
      "grad_norm": 1.0181454420089722,
      "learning_rate": 0.0002859070659784305,
      "loss": 2.3736,
      "step": 118648
    },
    {
      "epoch": 1.54,
      "grad_norm": 1.0270410776138306,
      "learning_rate": 0.00028590297993356045,
      "loss": 2.2627,
      "step": 118649
    },
    {
      "epoch": 1.54,
      "grad_norm": 1.0434718132019043,
      "learning_rate": 0.00028589889389131143,
      "loss": 2.1867,
      "step": 118650
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.9294764995574951,
      "learning_rate": 0.00028589480785168396,
      "loss": 2.0662,
      "step": 118651
    },
    {
      "epoch": 1.54,
      "grad_norm": 1.0507136583328247,
      "learning_rate": 0.00028589072181467895,
      "loss": 2.0725,
      "step": 118652
    },
    {
      "epoch": 1.54,
      "grad_norm": 1.0070980787277222,
      "learning_rate": 0.00028588663578029707,
      "loss": 2.0845,
      "step": 118653
    },
    {
      "epoch": 1.54,
      "grad_norm": 1.1147290468215942,
      "learning_rate": 0.00028588254974853917,
      "loss": 1.9954,
      "step": 118654
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.880709171295166,
      "learning_rate": 0.0002858784637194059,
      "loss": 2.0303,
      "step": 118655
    },
    {
      "epoch": 1.54,
      "grad_norm": 1.2577662467956543,
      "learning_rate": 0.0002858743776928982,
      "loss": 1.9911,
      "step": 118656
    },
    {
      "epoch": 1.54,
      "grad_norm": 0.908561110496521,
      "learning_rate": 0.00028587029166901666,
      "loss": 2.03,
      "step": 118657
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.8931854367256165,
      "learning_rate": 0.00028586620564776213,
      "loss": 1.9913,
      "step": 118658
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.984018087387085,
      "learning_rate": 0.00028586211962913533,
      "loss": 1.9245,
      "step": 118659
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.8183499574661255,
      "learning_rate": 0.000285858033613137,
      "loss": 2.0134,
      "step": 118660
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.9838147759437561,
      "learning_rate": 0.00028585394759976794,
      "loss": 2.1931,
      "step": 118661
    },
    {
      "epoch": 1.55,
      "grad_norm": 1.0510367155075073,
      "learning_rate": 0.00028584986158902893,
      "loss": 2.2779,
      "step": 118662
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.96217280626297,
      "learning_rate": 0.0002858457755809207,
      "loss": 2.0171,
      "step": 118663
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.905921459197998,
      "learning_rate": 0.00028584168957544395,
      "loss": 2.2579,
      "step": 118664
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.9716992378234863,
      "learning_rate": 0.0002858376035725995,
      "loss": 2.2943,
      "step": 118665
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.9766448736190796,
      "learning_rate": 0.00028583351757238815,
      "loss": 2.256,
      "step": 118666
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.9027928709983826,
      "learning_rate": 0.00028582943157481064,
      "loss": 2.3114,
      "step": 118667
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.9805772304534912,
      "learning_rate": 0.0002858253455798677,
      "loss": 2.0971,
      "step": 118668
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.9986969828605652,
      "learning_rate": 0.00028582125958756,
      "loss": 2.1526,
      "step": 118669
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.9157416820526123,
      "learning_rate": 0.00028581717359788846,
      "loss": 1.928,
      "step": 118670
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.9702489376068115,
      "learning_rate": 0.0002858130876108538,
      "loss": 2.0996,
      "step": 118671
    },
    {
      "epoch": 1.55,
      "grad_norm": 1.0127214193344116,
      "learning_rate": 0.00028580900162645666,
      "loss": 2.0938,
      "step": 118672
    },
    {
      "epoch": 1.55,
      "grad_norm": 1.0283035039901733,
      "learning_rate": 0.00028580491564469794,
      "loss": 2.0459,
      "step": 118673
    },
    {
      "epoch": 1.55,
      "grad_norm": 1.090789794921875,
      "learning_rate": 0.0002858008296655785,
      "loss": 2.1833,
      "step": 118674
    },
    {
      "epoch": 1.55,
      "grad_norm": 1.0720096826553345,
      "learning_rate": 0.0002857967436890988,
      "loss": 2.0476,
      "step": 118675
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.9621385931968689,
      "learning_rate": 0.0002857926577152598,
      "loss": 2.0506,
      "step": 118676
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.9220796823501587,
      "learning_rate": 0.0002857885717440622,
      "loss": 2.0202,
      "step": 118677
    },
    {
      "epoch": 1.55,
      "grad_norm": 1.0274484157562256,
      "learning_rate": 0.0002857844857755067,
      "loss": 2.2543,
      "step": 118678
    },
    {
      "epoch": 1.55,
      "grad_norm": 1.4605021476745605,
      "learning_rate": 0.0002857803998095942,
      "loss": 2.1628,
      "step": 118679
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.9386734366416931,
      "learning_rate": 0.0002857763138463255,
      "loss": 2.2581,
      "step": 118680
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.9216129183769226,
      "learning_rate": 0.00028577222788570114,
      "loss": 2.0015,
      "step": 118681
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.9768185615539551,
      "learning_rate": 0.00028576814192772195,
      "loss": 2.2435,
      "step": 118682
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.9938198924064636,
      "learning_rate": 0.0002857640559723888,
      "loss": 2.1202,
      "step": 118683
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.9053195118904114,
      "learning_rate": 0.0002857599700197023,
      "loss": 2.1189,
      "step": 118684
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.9203853607177734,
      "learning_rate": 0.00028575588406966335,
      "loss": 2.2051,
      "step": 118685
    },
    {
      "epoch": 1.55,
      "grad_norm": 1.0395605564117432,
      "learning_rate": 0.0002857517981222728,
      "loss": 2.181,
      "step": 118686
    },
    {
      "epoch": 1.55,
      "grad_norm": 1.0142866373062134,
      "learning_rate": 0.0002857477121775311,
      "loss": 2.0943,
      "step": 118687
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.9131528735160828,
      "learning_rate": 0.00028574362623543913,
      "loss": 2.0744,
      "step": 118688
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.918372392654419,
      "learning_rate": 0.0002857395402959977,
      "loss": 2.1934,
      "step": 118689
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.9790377020835876,
      "learning_rate": 0.0002857354543592076,
      "loss": 2.0714,
      "step": 118690
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.9886923432350159,
      "learning_rate": 0.00028573136842506954,
      "loss": 2.0581,
      "step": 118691
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.9602985382080078,
      "learning_rate": 0.0002857272824935844,
      "loss": 2.2559,
      "step": 118692
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.9931220412254333,
      "learning_rate": 0.0002857231965647527,
      "loss": 2.0835,
      "step": 118693
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.9329155683517456,
      "learning_rate": 0.0002857191106385753,
      "loss": 2.0472,
      "step": 118694
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.8901615142822266,
      "learning_rate": 0.00028571502471505305,
      "loss": 1.9667,
      "step": 118695
    },
    {
      "epoch": 1.55,
      "grad_norm": 2.189441680908203,
      "learning_rate": 0.0002857109387941866,
      "loss": 1.9792,
      "step": 118696
    },
    {
      "epoch": 1.55,
      "grad_norm": 1.3875502347946167,
      "learning_rate": 0.00028570685287597676,
      "loss": 2.0755,
      "step": 118697
    },
    {
      "epoch": 1.55,
      "grad_norm": 1.1564550399780273,
      "learning_rate": 0.00028570276696042446,
      "loss": 2.1289,
      "step": 118698
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.8759092092514038,
      "learning_rate": 0.0002856986810475301,
      "loss": 1.9707,
      "step": 118699
    },
    {
      "epoch": 1.55,
      "grad_norm": 1.12814462184906,
      "learning_rate": 0.00028569459513729466,
      "loss": 1.8448,
      "step": 118700
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.965837836265564,
      "learning_rate": 0.00028569050922971884,
      "loss": 2.2693,
      "step": 118701
    },
    {
      "epoch": 1.55,
      "grad_norm": 1.0062333345413208,
      "learning_rate": 0.00028568642332480344,
      "loss": 2.082,
      "step": 118702
    },
    {
      "epoch": 1.55,
      "grad_norm": 1.0094342231750488,
      "learning_rate": 0.0002856823374225492,
      "loss": 2.1265,
      "step": 118703
    },
    {
      "epoch": 1.55,
      "grad_norm": 1.0441232919692993,
      "learning_rate": 0.000285678251522957,
      "loss": 2.277,
      "step": 118704
    },
    {
      "epoch": 1.55,
      "grad_norm": 1.0225932598114014,
      "learning_rate": 0.00028567416562602735,
      "loss": 2.4571,
      "step": 118705
    },
    {
      "epoch": 1.55,
      "grad_norm": 1.024364709854126,
      "learning_rate": 0.0002856700797317612,
      "loss": 2.3126,
      "step": 118706
    },
    {
      "epoch": 1.55,
      "grad_norm": 1.0152932405471802,
      "learning_rate": 0.0002856659938401592,
      "loss": 2.4068,
      "step": 118707
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.9424889087677002,
      "learning_rate": 0.00028566190795122217,
      "loss": 2.2382,
      "step": 118708
    },
    {
      "epoch": 1.55,
      "grad_norm": 1.1314386129379272,
      "learning_rate": 0.00028565782206495087,
      "loss": 2.2808,
      "step": 118709
    },
    {
      "epoch": 1.55,
      "grad_norm": 1.3230692148208618,
      "learning_rate": 0.00028565373618134605,
      "loss": 1.9535,
      "step": 118710
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.9651111364364624,
      "learning_rate": 0.0002856496503004086,
      "loss": 1.8975,
      "step": 118711
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.9893662929534912,
      "learning_rate": 0.000285645564422139,
      "loss": 2.2606,
      "step": 118712
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.996699333190918,
      "learning_rate": 0.00028564147854653817,
      "loss": 2.1356,
      "step": 118713
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.8747662305831909,
      "learning_rate": 0.0002856373926736069,
      "loss": 2.1258,
      "step": 118714
    },
    {
      "epoch": 1.55,
      "grad_norm": 1.0149245262145996,
      "learning_rate": 0.0002856333068033459,
      "loss": 2.1704,
      "step": 118715
    },
    {
      "epoch": 1.55,
      "grad_norm": 1.0642858743667603,
      "learning_rate": 0.0002856292209357559,
      "loss": 2.3168,
      "step": 118716
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.9888125658035278,
      "learning_rate": 0.00028562513507083783,
      "loss": 1.9879,
      "step": 118717
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.9470618963241577,
      "learning_rate": 0.00028562104920859223,
      "loss": 2.1095,
      "step": 118718
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.8701436519622803,
      "learning_rate": 0.0002856169633490199,
      "loss": 2.0214,
      "step": 118719
    },
    {
      "epoch": 1.55,
      "grad_norm": 1.038474678993225,
      "learning_rate": 0.0002856128774921217,
      "loss": 2.3865,
      "step": 118720
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.9452081322669983,
      "learning_rate": 0.0002856087916378983,
      "loss": 2.1746,
      "step": 118721
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.9971129894256592,
      "learning_rate": 0.00028560470578635054,
      "loss": 2.1766,
      "step": 118722
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.9653692245483398,
      "learning_rate": 0.00028560061993747923,
      "loss": 2.374,
      "step": 118723
    },
    {
      "epoch": 1.55,
      "grad_norm": 1.0757848024368286,
      "learning_rate": 0.0002855965340912849,
      "loss": 2.1981,
      "step": 118724
    },
    {
      "epoch": 1.55,
      "grad_norm": 1.3272159099578857,
      "learning_rate": 0.0002855924482477684,
      "loss": 2.022,
      "step": 118725
    },
    {
      "epoch": 1.55,
      "grad_norm": 1.0003823041915894,
      "learning_rate": 0.00028558836240693067,
      "loss": 2.1312,
      "step": 118726
    },
    {
      "epoch": 1.55,
      "grad_norm": 1.064329981803894,
      "learning_rate": 0.00028558427656877225,
      "loss": 2.2641,
      "step": 118727
    },
    {
      "epoch": 1.55,
      "grad_norm": 1.0999560356140137,
      "learning_rate": 0.00028558019073329397,
      "loss": 2.0861,
      "step": 118728
    },
    {
      "epoch": 1.55,
      "grad_norm": 1.0131208896636963,
      "learning_rate": 0.00028557610490049677,
      "loss": 2.2281,
      "step": 118729
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.882828950881958,
      "learning_rate": 0.00028557201907038107,
      "loss": 2.1978,
      "step": 118730
    },
    {
      "epoch": 1.55,
      "grad_norm": 1.0803356170654297,
      "learning_rate": 0.00028556793324294784,
      "loss": 1.927,
      "step": 118731
    },
    {
      "epoch": 1.55,
      "grad_norm": 1.0517628192901611,
      "learning_rate": 0.00028556384741819786,
      "loss": 2.1183,
      "step": 118732
    },
    {
      "epoch": 1.55,
      "grad_norm": 1.1492499113082886,
      "learning_rate": 0.00028555976159613176,
      "loss": 2.1656,
      "step": 118733
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.9605758190155029,
      "learning_rate": 0.00028555567577675037,
      "loss": 1.9509,
      "step": 118734
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.9949938654899597,
      "learning_rate": 0.00028555158996005454,
      "loss": 2.472,
      "step": 118735
    },
    {
      "epoch": 1.55,
      "grad_norm": 1.0440832376480103,
      "learning_rate": 0.0002855475041460449,
      "loss": 2.1487,
      "step": 118736
    },
    {
      "epoch": 1.55,
      "grad_norm": 1.0814388990402222,
      "learning_rate": 0.00028554341833472225,
      "loss": 2.0855,
      "step": 118737
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.9241923689842224,
      "learning_rate": 0.0002855393325260873,
      "loss": 2.0908,
      "step": 118738
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.9681293368339539,
      "learning_rate": 0.0002855352467201409,
      "loss": 2.2451,
      "step": 118739
    },
    {
      "epoch": 1.55,
      "grad_norm": 1.2312296628952026,
      "learning_rate": 0.00028553116091688377,
      "loss": 2.2835,
      "step": 118740
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.9686502814292908,
      "learning_rate": 0.00028552707511631675,
      "loss": 2.2577,
      "step": 118741
    },
    {
      "epoch": 1.55,
      "grad_norm": 1.0035067796707153,
      "learning_rate": 0.0002855229893184404,
      "loss": 2.2785,
      "step": 118742
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.9274547100067139,
      "learning_rate": 0.00028551890352325563,
      "loss": 1.9671,
      "step": 118743
    },
    {
      "epoch": 1.55,
      "grad_norm": 1.0050476789474487,
      "learning_rate": 0.00028551481773076326,
      "loss": 2.2806,
      "step": 118744
    },
    {
      "epoch": 1.55,
      "grad_norm": 1.0071009397506714,
      "learning_rate": 0.00028551073194096383,
      "loss": 2.1973,
      "step": 118745
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.9770972728729248,
      "learning_rate": 0.0002855066461538583,
      "loss": 2.064,
      "step": 118746
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.978952944278717,
      "learning_rate": 0.0002855025603694474,
      "loss": 2.0287,
      "step": 118747
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.9733942747116089,
      "learning_rate": 0.00028549847458773177,
      "loss": 2.016,
      "step": 118748
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.9116231203079224,
      "learning_rate": 0.00028549438880871224,
      "loss": 2.1604,
      "step": 118749
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.9748557209968567,
      "learning_rate": 0.00028549030303238955,
      "loss": 1.8836,
      "step": 118750
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.8885114192962646,
      "learning_rate": 0.0002854862172587646,
      "loss": 2.1454,
      "step": 118751
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.9327636361122131,
      "learning_rate": 0.0002854821314878379,
      "loss": 2.1422,
      "step": 118752
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.9472252130508423,
      "learning_rate": 0.00028547804571961053,
      "loss": 2.2896,
      "step": 118753
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.9851807951927185,
      "learning_rate": 0.0002854739599540829,
      "loss": 2.287,
      "step": 118754
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.9566019177436829,
      "learning_rate": 0.000285469874191256,
      "loss": 2.0011,
      "step": 118755
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.951374351978302,
      "learning_rate": 0.0002854657884311305,
      "loss": 2.1099,
      "step": 118756
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.9635343551635742,
      "learning_rate": 0.0002854617026737072,
      "loss": 2.163,
      "step": 118757
    },
    {
      "epoch": 1.55,
      "grad_norm": 1.0282524824142456,
      "learning_rate": 0.00028545761691898684,
      "loss": 2.449,
      "step": 118758
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.9554433226585388,
      "learning_rate": 0.0002854535311669703,
      "loss": 2.1113,
      "step": 118759
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.8812910914421082,
      "learning_rate": 0.00028544944541765806,
      "loss": 2.1459,
      "step": 118760
    },
    {
      "epoch": 1.55,
      "grad_norm": 1.035041093826294,
      "learning_rate": 0.0002854453596710511,
      "loss": 2.0321,
      "step": 118761
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.9653348326683044,
      "learning_rate": 0.0002854412739271501,
      "loss": 2.1404,
      "step": 118762
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.9119349718093872,
      "learning_rate": 0.0002854371881859558,
      "loss": 2.0905,
      "step": 118763
    },
    {
      "epoch": 1.55,
      "grad_norm": 1.0468262434005737,
      "learning_rate": 0.00028543310244746905,
      "loss": 2.0288,
      "step": 118764
    },
    {
      "epoch": 1.55,
      "grad_norm": 1.2599200010299683,
      "learning_rate": 0.00028542901671169065,
      "loss": 2.2126,
      "step": 118765
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.9228278398513794,
      "learning_rate": 0.0002854249309786212,
      "loss": 2.1634,
      "step": 118766
    },
    {
      "epoch": 1.55,
      "grad_norm": 1.1183334589004517,
      "learning_rate": 0.0002854208452482615,
      "loss": 2.1363,
      "step": 118767
    },
    {
      "epoch": 1.55,
      "grad_norm": 1.0510752201080322,
      "learning_rate": 0.00028541675952061234,
      "loss": 2.0646,
      "step": 118768
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.9395473599433899,
      "learning_rate": 0.0002854126737956745,
      "loss": 2.1681,
      "step": 118769
    },
    {
      "epoch": 1.55,
      "grad_norm": 1.0317552089691162,
      "learning_rate": 0.00028540858807344873,
      "loss": 2.2037,
      "step": 118770
    },
    {
      "epoch": 1.55,
      "grad_norm": 1.0935159921646118,
      "learning_rate": 0.00028540450235393585,
      "loss": 2.0211,
      "step": 118771
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.990105390548706,
      "learning_rate": 0.00028540041663713646,
      "loss": 2.1321,
      "step": 118772
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.9204517006874084,
      "learning_rate": 0.0002853963309230514,
      "loss": 2.1982,
      "step": 118773
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.9845380187034607,
      "learning_rate": 0.0002853922452116814,
      "loss": 2.1967,
      "step": 118774
    },
    {
      "epoch": 1.55,
      "grad_norm": 1.2711155414581299,
      "learning_rate": 0.00028538815950302735,
      "loss": 2.1787,
      "step": 118775
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.9052485823631287,
      "learning_rate": 0.0002853840737970898,
      "loss": 2.1937,
      "step": 118776
    },
    {
      "epoch": 1.55,
      "grad_norm": 1.2192907333374023,
      "learning_rate": 0.0002853799880938697,
      "loss": 2.1587,
      "step": 118777
    },
    {
      "epoch": 1.55,
      "grad_norm": 1.2688074111938477,
      "learning_rate": 0.00028537590239336786,
      "loss": 2.4067,
      "step": 118778
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.982820451259613,
      "learning_rate": 0.0002853718166955848,
      "loss": 1.9616,
      "step": 118779
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.9536461234092712,
      "learning_rate": 0.00028536773100052137,
      "loss": 1.9959,
      "step": 118780
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.8931789398193359,
      "learning_rate": 0.0002853636453081784,
      "loss": 2.0815,
      "step": 118781
    },
    {
      "epoch": 1.55,
      "grad_norm": 1.04551100730896,
      "learning_rate": 0.0002853595596185565,
      "loss": 2.0508,
      "step": 118782
    },
    {
      "epoch": 1.55,
      "grad_norm": 1.2147327661514282,
      "learning_rate": 0.0002853554739316566,
      "loss": 2.1748,
      "step": 118783
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.9433260560035706,
      "learning_rate": 0.0002853513882474795,
      "loss": 2.2188,
      "step": 118784
    },
    {
      "epoch": 1.55,
      "grad_norm": 1.0445630550384521,
      "learning_rate": 0.00028534730256602576,
      "loss": 2.1947,
      "step": 118785
    },
    {
      "epoch": 1.55,
      "grad_norm": 1.0048061609268188,
      "learning_rate": 0.00028534321688729623,
      "loss": 2.3156,
      "step": 118786
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.9350830912590027,
      "learning_rate": 0.0002853391312112917,
      "loss": 2.112,
      "step": 118787
    },
    {
      "epoch": 1.55,
      "grad_norm": 1.1213322877883911,
      "learning_rate": 0.00028533504553801287,
      "loss": 2.2032,
      "step": 118788
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.9541725516319275,
      "learning_rate": 0.0002853309598674605,
      "loss": 2.1235,
      "step": 118789
    },
    {
      "epoch": 1.55,
      "grad_norm": 1.085387110710144,
      "learning_rate": 0.00028532687419963556,
      "loss": 2.1694,
      "step": 118790
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.9613523483276367,
      "learning_rate": 0.00028532278853453846,
      "loss": 2.169,
      "step": 118791
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.9673740863800049,
      "learning_rate": 0.00028531870287217014,
      "loss": 2.1089,
      "step": 118792
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.9237280488014221,
      "learning_rate": 0.00028531461721253135,
      "loss": 2.1836,
      "step": 118793
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.9027633666992188,
      "learning_rate": 0.00028531053155562286,
      "loss": 2.1174,
      "step": 118794
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.9545291066169739,
      "learning_rate": 0.0002853064459014454,
      "loss": 2.0693,
      "step": 118795
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.9072659015655518,
      "learning_rate": 0.0002853023602499999,
      "loss": 1.925,
      "step": 118796
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.9470920562744141,
      "learning_rate": 0.0002852982746012869,
      "loss": 2.0425,
      "step": 118797
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.9537298679351807,
      "learning_rate": 0.00028529418895530714,
      "loss": 2.1335,
      "step": 118798
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.8869149684906006,
      "learning_rate": 0.0002852901033120615,
      "loss": 2.0599,
      "step": 118799
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.8366393446922302,
      "learning_rate": 0.0002852860176715507,
      "loss": 2.1017,
      "step": 118800
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.9019717574119568,
      "learning_rate": 0.0002852819320337755,
      "loss": 2.1685,
      "step": 118801
    },
    {
      "epoch": 1.55,
      "grad_norm": 1.0153483152389526,
      "learning_rate": 0.00028527784639873676,
      "loss": 2.1768,
      "step": 118802
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.9016081690788269,
      "learning_rate": 0.00028527376076643504,
      "loss": 2.1438,
      "step": 118803
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.9074676036834717,
      "learning_rate": 0.00028526967513687126,
      "loss": 2.0494,
      "step": 118804
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.9800916910171509,
      "learning_rate": 0.0002852655895100461,
      "loss": 2.2758,
      "step": 118805
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.9194608926773071,
      "learning_rate": 0.00028526150388596033,
      "loss": 2.0764,
      "step": 118806
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.9518643021583557,
      "learning_rate": 0.0002852574182646147,
      "loss": 2.0525,
      "step": 118807
    },
    {
      "epoch": 1.55,
      "grad_norm": 1.166715383529663,
      "learning_rate": 0.0002852533326460102,
      "loss": 2.1313,
      "step": 118808
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.8460449576377869,
      "learning_rate": 0.0002852492470301472,
      "loss": 1.942,
      "step": 118809
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.8816211223602295,
      "learning_rate": 0.00028524516141702666,
      "loss": 2.1005,
      "step": 118810
    },
    {
      "epoch": 1.55,
      "grad_norm": 1.0190086364746094,
      "learning_rate": 0.0002852410758066493,
      "loss": 1.9412,
      "step": 118811
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.9571711421012878,
      "learning_rate": 0.0002852369901990159,
      "loss": 2.1817,
      "step": 118812
    },
    {
      "epoch": 1.55,
      "grad_norm": 1.0670602321624756,
      "learning_rate": 0.00028523290459412726,
      "loss": 2.2952,
      "step": 118813
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.9653784036636353,
      "learning_rate": 0.0002852288189919842,
      "loss": 2.2385,
      "step": 118814
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.9518061280250549,
      "learning_rate": 0.00028522473339258727,
      "loss": 2.0668,
      "step": 118815
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.885839581489563,
      "learning_rate": 0.0002852206477959373,
      "loss": 2.2084,
      "step": 118816
    },
    {
      "epoch": 1.55,
      "grad_norm": 1.0889828205108643,
      "learning_rate": 0.00028521656220203517,
      "loss": 2.168,
      "step": 118817
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.9642952680587769,
      "learning_rate": 0.00028521247661088153,
      "loss": 2.1663,
      "step": 118818
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.8983528017997742,
      "learning_rate": 0.0002852083910224771,
      "loss": 2.0558,
      "step": 118819
    },
    {
      "epoch": 1.55,
      "grad_norm": 1.0531729459762573,
      "learning_rate": 0.00028520430543682286,
      "loss": 2.1153,
      "step": 118820
    },
    {
      "epoch": 1.55,
      "grad_norm": 1.097213864326477,
      "learning_rate": 0.00028520021985391935,
      "loss": 2.2369,
      "step": 118821
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.946530818939209,
      "learning_rate": 0.00028519613427376744,
      "loss": 2.0869,
      "step": 118822
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.9375203847885132,
      "learning_rate": 0.00028519204869636777,
      "loss": 1.9379,
      "step": 118823
    },
    {
      "epoch": 1.55,
      "grad_norm": 1.1359546184539795,
      "learning_rate": 0.00028518796312172116,
      "loss": 2.2833,
      "step": 118824
    },
    {
      "epoch": 1.55,
      "grad_norm": 1.2673461437225342,
      "learning_rate": 0.00028518387754982843,
      "loss": 2.0763,
      "step": 118825
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.9109019637107849,
      "learning_rate": 0.0002851797919806903,
      "loss": 2.0316,
      "step": 118826
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.9289091229438782,
      "learning_rate": 0.00028517570641430756,
      "loss": 2.2275,
      "step": 118827
    },
    {
      "epoch": 1.55,
      "grad_norm": 1.5582953691482544,
      "learning_rate": 0.0002851716208506809,
      "loss": 1.9912,
      "step": 118828
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.9225383400917053,
      "learning_rate": 0.00028516753528981114,
      "loss": 2.1714,
      "step": 118829
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.9163908958435059,
      "learning_rate": 0.000285163449731699,
      "loss": 2.0766,
      "step": 118830
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.9599534869194031,
      "learning_rate": 0.00028515936417634517,
      "loss": 1.9889,
      "step": 118831
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.967354953289032,
      "learning_rate": 0.00028515527862375063,
      "loss": 2.2497,
      "step": 118832
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.9565311074256897,
      "learning_rate": 0.00028515119307391594,
      "loss": 2.1285,
      "step": 118833
    },
    {
      "epoch": 1.55,
      "grad_norm": 1.0797796249389648,
      "learning_rate": 0.00028514710752684195,
      "loss": 2.0998,
      "step": 118834
    },
    {
      "epoch": 1.55,
      "grad_norm": 1.0507771968841553,
      "learning_rate": 0.0002851430219825293,
      "loss": 2.0104,
      "step": 118835
    },
    {
      "epoch": 1.55,
      "grad_norm": 1.0172244310379028,
      "learning_rate": 0.00028513893644097895,
      "loss": 2.1499,
      "step": 118836
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.991545557975769,
      "learning_rate": 0.0002851348509021915,
      "loss": 2.125,
      "step": 118837
    },
    {
      "epoch": 1.55,
      "grad_norm": 1.179160237312317,
      "learning_rate": 0.0002851307653661679,
      "loss": 2.2554,
      "step": 118838
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.9224849939346313,
      "learning_rate": 0.0002851266798329086,
      "loss": 2.2211,
      "step": 118839
    },
    {
      "epoch": 1.55,
      "grad_norm": 1.2501599788665771,
      "learning_rate": 0.00028512259430241453,
      "loss": 2.1458,
      "step": 118840
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.9422816634178162,
      "learning_rate": 0.0002851185087746865,
      "loss": 2.0986,
      "step": 118841
    },
    {
      "epoch": 1.55,
      "grad_norm": 1.0814135074615479,
      "learning_rate": 0.0002851144232497252,
      "loss": 2.3739,
      "step": 118842
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.9264169931411743,
      "learning_rate": 0.0002851103377275315,
      "loss": 1.9712,
      "step": 118843
    },
    {
      "epoch": 1.55,
      "grad_norm": 1.0242159366607666,
      "learning_rate": 0.00028510625220810594,
      "loss": 2.0167,
      "step": 118844
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.9775761365890503,
      "learning_rate": 0.00028510216669144957,
      "loss": 2.3957,
      "step": 118845
    },
    {
      "epoch": 1.55,
      "grad_norm": 1.040371060371399,
      "learning_rate": 0.00028509808117756284,
      "loss": 2.438,
      "step": 118846
    },
    {
      "epoch": 1.55,
      "grad_norm": 1.0309635400772095,
      "learning_rate": 0.0002850939956664467,
      "loss": 2.3081,
      "step": 118847
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.9962234497070312,
      "learning_rate": 0.0002850899101581018,
      "loss": 1.9862,
      "step": 118848
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.8899735808372498,
      "learning_rate": 0.00028508582465252907,
      "loss": 1.9515,
      "step": 118849
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.9032958149909973,
      "learning_rate": 0.0002850817391497292,
      "loss": 1.9157,
      "step": 118850
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.9392222166061401,
      "learning_rate": 0.00028507765364970284,
      "loss": 2.1565,
      "step": 118851
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.8555771708488464,
      "learning_rate": 0.00028507356815245087,
      "loss": 2.1098,
      "step": 118852
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.8765373826026917,
      "learning_rate": 0.0002850694826579739,
      "loss": 2.1188,
      "step": 118853
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.9327361583709717,
      "learning_rate": 0.0002850653971662729,
      "loss": 2.1588,
      "step": 118854
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.9192034006118774,
      "learning_rate": 0.00028506131167734845,
      "loss": 1.9919,
      "step": 118855
    },
    {
      "epoch": 1.55,
      "grad_norm": 1.0010550022125244,
      "learning_rate": 0.0002850572261912014,
      "loss": 2.3471,
      "step": 118856
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.9479917883872986,
      "learning_rate": 0.00028505314070783266,
      "loss": 2.1818,
      "step": 118857
    },
    {
      "epoch": 1.55,
      "grad_norm": 1.0095372200012207,
      "learning_rate": 0.00028504905522724265,
      "loss": 2.4038,
      "step": 118858
    },
    {
      "epoch": 1.55,
      "grad_norm": 1.006962537765503,
      "learning_rate": 0.0002850449697494323,
      "loss": 2.1394,
      "step": 118859
    },
    {
      "epoch": 1.55,
      "grad_norm": 1.040701985359192,
      "learning_rate": 0.0002850408842744024,
      "loss": 2.0395,
      "step": 118860
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.8931713104248047,
      "learning_rate": 0.0002850367988021537,
      "loss": 2.0149,
      "step": 118861
    },
    {
      "epoch": 1.55,
      "grad_norm": 1.18412184715271,
      "learning_rate": 0.0002850327133326869,
      "loss": 2.2836,
      "step": 118862
    },
    {
      "epoch": 1.55,
      "grad_norm": 1.0888912677764893,
      "learning_rate": 0.00028502862786600295,
      "loss": 2.0985,
      "step": 118863
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.8641738295555115,
      "learning_rate": 0.00028502454240210233,
      "loss": 1.9509,
      "step": 118864
    },
    {
      "epoch": 1.55,
      "grad_norm": 1.1760101318359375,
      "learning_rate": 0.00028502045694098595,
      "loss": 2.1324,
      "step": 118865
    },
    {
      "epoch": 1.55,
      "grad_norm": 1.0118002891540527,
      "learning_rate": 0.00028501637148265456,
      "loss": 2.1452,
      "step": 118866
    },
    {
      "epoch": 1.55,
      "grad_norm": 1.1027289628982544,
      "learning_rate": 0.00028501228602710886,
      "loss": 2.1526,
      "step": 118867
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.8901172280311584,
      "learning_rate": 0.00028500820057434967,
      "loss": 1.9893,
      "step": 118868
    },
    {
      "epoch": 1.55,
      "grad_norm": 1.1300674676895142,
      "learning_rate": 0.0002850041151243779,
      "loss": 2.0389,
      "step": 118869
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.9764798283576965,
      "learning_rate": 0.00028500002967719396,
      "loss": 2.5011,
      "step": 118870
    },
    {
      "epoch": 1.55,
      "grad_norm": 1.073698878288269,
      "learning_rate": 0.00028499594423279885,
      "loss": 1.9241,
      "step": 118871
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.9479707479476929,
      "learning_rate": 0.0002849918587911932,
      "loss": 2.2796,
      "step": 118872
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.9645127058029175,
      "learning_rate": 0.00028498777335237796,
      "loss": 2.0593,
      "step": 118873
    },
    {
      "epoch": 1.55,
      "grad_norm": 1.265358805656433,
      "learning_rate": 0.0002849836879163537,
      "loss": 2.2818,
      "step": 118874
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.9662596583366394,
      "learning_rate": 0.0002849796024831214,
      "loss": 2.1936,
      "step": 118875
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.9067395925521851,
      "learning_rate": 0.00028497551705268157,
      "loss": 1.9998,
      "step": 118876
    },
    {
      "epoch": 1.55,
      "grad_norm": 1.0058484077453613,
      "learning_rate": 0.00028497143162503505,
      "loss": 1.9953,
      "step": 118877
    },
    {
      "epoch": 1.55,
      "grad_norm": 1.0590718984603882,
      "learning_rate": 0.00028496734620018263,
      "loss": 2.0789,
      "step": 118878
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.9498881697654724,
      "learning_rate": 0.00028496326077812506,
      "loss": 2.2501,
      "step": 118879
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.9215466380119324,
      "learning_rate": 0.0002849591753588631,
      "loss": 2.1125,
      "step": 118880
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.9635117053985596,
      "learning_rate": 0.00028495508994239764,
      "loss": 2.1214,
      "step": 118881
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.9695529341697693,
      "learning_rate": 0.00028495100452872926,
      "loss": 1.8986,
      "step": 118882
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.9076768755912781,
      "learning_rate": 0.0002849469191178587,
      "loss": 2.2872,
      "step": 118883
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.9533928036689758,
      "learning_rate": 0.0002849428337097868,
      "loss": 2.0721,
      "step": 118884
    },
    {
      "epoch": 1.55,
      "grad_norm": 1.0464303493499756,
      "learning_rate": 0.00028493874830451426,
      "loss": 2.0998,
      "step": 118885
    },
    {
      "epoch": 1.55,
      "grad_norm": 1.203485131263733,
      "learning_rate": 0.000284934662902042,
      "loss": 2.0396,
      "step": 118886
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.9385178089141846,
      "learning_rate": 0.0002849305775023707,
      "loss": 2.3414,
      "step": 118887
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.9444323182106018,
      "learning_rate": 0.000284926492105501,
      "loss": 2.2285,
      "step": 118888
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.927807629108429,
      "learning_rate": 0.0002849224067114338,
      "loss": 2.2016,
      "step": 118889
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.9567853808403015,
      "learning_rate": 0.00028491832132016974,
      "loss": 2.3872,
      "step": 118890
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.8823702335357666,
      "learning_rate": 0.00028491423593170965,
      "loss": 2.1239,
      "step": 118891
    },
    {
      "epoch": 1.55,
      "grad_norm": 1.0129480361938477,
      "learning_rate": 0.00028491015054605434,
      "loss": 2.1554,
      "step": 118892
    },
    {
      "epoch": 1.55,
      "grad_norm": 1.0448116064071655,
      "learning_rate": 0.0002849060651632046,
      "loss": 2.0848,
      "step": 118893
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.9700995087623596,
      "learning_rate": 0.00028490197978316096,
      "loss": 2.3799,
      "step": 118894
    },
    {
      "epoch": 1.55,
      "grad_norm": 1.0469969511032104,
      "learning_rate": 0.0002848978944059244,
      "loss": 2.0588,
      "step": 118895
    },
    {
      "epoch": 1.55,
      "grad_norm": 1.0187071561813354,
      "learning_rate": 0.00028489380903149554,
      "loss": 2.2622,
      "step": 118896
    },
    {
      "epoch": 1.55,
      "grad_norm": 1.0293869972229004,
      "learning_rate": 0.00028488972365987527,
      "loss": 2.3266,
      "step": 118897
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.9798502326011658,
      "learning_rate": 0.00028488563829106423,
      "loss": 2.2496,
      "step": 118898
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.9321582317352295,
      "learning_rate": 0.0002848815529250634,
      "loss": 1.9704,
      "step": 118899
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.9926497936248779,
      "learning_rate": 0.00028487746756187323,
      "loss": 2.1662,
      "step": 118900
    },
    {
      "epoch": 1.55,
      "grad_norm": 1.0339772701263428,
      "learning_rate": 0.0002848733822014946,
      "loss": 2.1672,
      "step": 118901
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.9375032186508179,
      "learning_rate": 0.00028486929684392833,
      "loss": 1.8899,
      "step": 118902
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.9928191900253296,
      "learning_rate": 0.0002848652114891751,
      "loss": 2.1781,
      "step": 118903
    },
    {
      "epoch": 1.55,
      "grad_norm": 1.016886591911316,
      "learning_rate": 0.00028486112613723577,
      "loss": 2.1687,
      "step": 118904
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.9990431666374207,
      "learning_rate": 0.000284857040788111,
      "loss": 2.1418,
      "step": 118905
    },
    {
      "epoch": 1.55,
      "grad_norm": 1.0870654582977295,
      "learning_rate": 0.0002848529554418018,
      "loss": 2.0536,
      "step": 118906
    },
    {
      "epoch": 1.55,
      "grad_norm": 1.0890511274337769,
      "learning_rate": 0.0002848488700983085,
      "loss": 2.0493,
      "step": 118907
    },
    {
      "epoch": 1.55,
      "grad_norm": 1.2006981372833252,
      "learning_rate": 0.00028484478475763206,
      "loss": 2.2828,
      "step": 118908
    },
    {
      "epoch": 1.55,
      "grad_norm": 1.5101094245910645,
      "learning_rate": 0.00028484069941977336,
      "loss": 2.1238,
      "step": 118909
    },
    {
      "epoch": 1.55,
      "grad_norm": 1.1266974210739136,
      "learning_rate": 0.000284836614084733,
      "loss": 2.2423,
      "step": 118910
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.941116988658905,
      "learning_rate": 0.0002848325287525118,
      "loss": 2.27,
      "step": 118911
    },
    {
      "epoch": 1.55,
      "grad_norm": 1.0798907279968262,
      "learning_rate": 0.0002848284434231106,
      "loss": 2.197,
      "step": 118912
    },
    {
      "epoch": 1.55,
      "grad_norm": 1.3932863473892212,
      "learning_rate": 0.00028482435809653006,
      "loss": 1.898,
      "step": 118913
    },
    {
      "epoch": 1.55,
      "grad_norm": 1.0633362531661987,
      "learning_rate": 0.0002848202727727709,
      "loss": 2.1383,
      "step": 118914
    },
    {
      "epoch": 1.55,
      "grad_norm": 1.0034599304199219,
      "learning_rate": 0.00028481618745183397,
      "loss": 2.1495,
      "step": 118915
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.9595065116882324,
      "learning_rate": 0.00028481210213372,
      "loss": 2.0139,
      "step": 118916
    },
    {
      "epoch": 1.55,
      "grad_norm": 1.6421245336532593,
      "learning_rate": 0.0002848080168184297,
      "loss": 1.9849,
      "step": 118917
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.961965799331665,
      "learning_rate": 0.0002848039315059639,
      "loss": 2.0274,
      "step": 118918
    },
    {
      "epoch": 1.55,
      "grad_norm": 1.0043331384658813,
      "learning_rate": 0.0002847998461963234,
      "loss": 2.2794,
      "step": 118919
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.9809086322784424,
      "learning_rate": 0.0002847957608895089,
      "loss": 2.0017,
      "step": 118920
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.9603129625320435,
      "learning_rate": 0.00028479167558552107,
      "loss": 2.2856,
      "step": 118921
    },
    {
      "epoch": 1.55,
      "grad_norm": 1.3995112180709839,
      "learning_rate": 0.0002847875902843608,
      "loss": 1.954,
      "step": 118922
    },
    {
      "epoch": 1.55,
      "grad_norm": 1.0069974660873413,
      "learning_rate": 0.00028478350498602873,
      "loss": 2.0919,
      "step": 118923
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.9129276871681213,
      "learning_rate": 0.0002847794196905258,
      "loss": 2.154,
      "step": 118924
    },
    {
      "epoch": 1.55,
      "grad_norm": 1.3085979223251343,
      "learning_rate": 0.0002847753343978526,
      "loss": 1.9172,
      "step": 118925
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.9377214312553406,
      "learning_rate": 0.00028477124910800995,
      "loss": 2.4038,
      "step": 118926
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.897642970085144,
      "learning_rate": 0.0002847671638209986,
      "loss": 2.1712,
      "step": 118927
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.9175150394439697,
      "learning_rate": 0.0002847630785368194,
      "loss": 2.0336,
      "step": 118928
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.9994964003562927,
      "learning_rate": 0.00028475899325547296,
      "loss": 2.0197,
      "step": 118929
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.9542266726493835,
      "learning_rate": 0.0002847549079769602,
      "loss": 2.0313,
      "step": 118930
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.9126557111740112,
      "learning_rate": 0.0002847508227012817,
      "loss": 2.0784,
      "step": 118931
    },
    {
      "epoch": 1.55,
      "grad_norm": 1.059263825416565,
      "learning_rate": 0.0002847467374284383,
      "loss": 2.031,
      "step": 118932
    },
    {
      "epoch": 1.55,
      "grad_norm": 1.0146148204803467,
      "learning_rate": 0.0002847426521584308,
      "loss": 2.3602,
      "step": 118933
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.9185635447502136,
      "learning_rate": 0.00028473856689125994,
      "loss": 2.2803,
      "step": 118934
    },
    {
      "epoch": 1.55,
      "grad_norm": 1.070723295211792,
      "learning_rate": 0.00028473448162692647,
      "loss": 2.1607,
      "step": 118935
    },
    {
      "epoch": 1.55,
      "grad_norm": 1.05343496799469,
      "learning_rate": 0.0002847303963654312,
      "loss": 2.0299,
      "step": 118936
    },
    {
      "epoch": 1.55,
      "grad_norm": 1.016144037246704,
      "learning_rate": 0.00028472631110677474,
      "loss": 2.0604,
      "step": 118937
    },
    {
      "epoch": 1.55,
      "grad_norm": 1.172204852104187,
      "learning_rate": 0.000284722225850958,
      "loss": 2.1393,
      "step": 118938
    },
    {
      "epoch": 1.55,
      "grad_norm": 1.0229079723358154,
      "learning_rate": 0.00028471814059798166,
      "loss": 2.2288,
      "step": 118939
    },
    {
      "epoch": 1.55,
      "grad_norm": 1.1425138711929321,
      "learning_rate": 0.00028471405534784646,
      "loss": 2.0396,
      "step": 118940
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.9826033115386963,
      "learning_rate": 0.00028470997010055323,
      "loss": 1.9397,
      "step": 118941
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.9740220308303833,
      "learning_rate": 0.0002847058848561029,
      "loss": 2.095,
      "step": 118942
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.8763704895973206,
      "learning_rate": 0.0002847017996144958,
      "loss": 2.0474,
      "step": 118943
    },
    {
      "epoch": 1.55,
      "grad_norm": 1.076711654663086,
      "learning_rate": 0.00028469771437573304,
      "loss": 2.3608,
      "step": 118944
    },
    {
      "epoch": 1.55,
      "grad_norm": 1.5565580129623413,
      "learning_rate": 0.00028469362913981515,
      "loss": 2.2138,
      "step": 118945
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.9431756734848022,
      "learning_rate": 0.0002846895439067431,
      "loss": 2.0291,
      "step": 118946
    },
    {
      "epoch": 1.55,
      "grad_norm": 1.009281039237976,
      "learning_rate": 0.0002846854586765175,
      "loss": 1.993,
      "step": 118947
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.9818921089172363,
      "learning_rate": 0.0002846813734491393,
      "loss": 2.2421,
      "step": 118948
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.9526408910751343,
      "learning_rate": 0.000284677288224609,
      "loss": 2.2677,
      "step": 118949
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.9262160658836365,
      "learning_rate": 0.00028467320300292745,
      "loss": 2.3389,
      "step": 118950
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.9867118000984192,
      "learning_rate": 0.00028466911778409545,
      "loss": 2.3174,
      "step": 118951
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.8883381485939026,
      "learning_rate": 0.00028466503256811376,
      "loss": 2.1078,
      "step": 118952
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.9294863939285278,
      "learning_rate": 0.00028466094735498315,
      "loss": 2.2876,
      "step": 118953
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.9244140982627869,
      "learning_rate": 0.0002846568621447045,
      "loss": 2.1833,
      "step": 118954
    },
    {
      "epoch": 1.55,
      "grad_norm": 1.008223295211792,
      "learning_rate": 0.0002846527769372783,
      "loss": 2.0283,
      "step": 118955
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.9032331705093384,
      "learning_rate": 0.0002846486917327054,
      "loss": 2.033,
      "step": 118956
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.9424234628677368,
      "learning_rate": 0.0002846446065309866,
      "loss": 2.2203,
      "step": 118957
    },
    {
      "epoch": 1.55,
      "grad_norm": 1.0380167961120605,
      "learning_rate": 0.00028464052133212267,
      "loss": 2.2818,
      "step": 118958
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.9548988938331604,
      "learning_rate": 0.0002846364361361143,
      "loss": 2.1091,
      "step": 118959
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.9590800404548645,
      "learning_rate": 0.00028463235094296256,
      "loss": 2.0028,
      "step": 118960
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.9452393054962158,
      "learning_rate": 0.00028462826575266774,
      "loss": 1.9743,
      "step": 118961
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.9531295299530029,
      "learning_rate": 0.0002846241805652308,
      "loss": 2.288,
      "step": 118962
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.9148114919662476,
      "learning_rate": 0.00028462009538065256,
      "loss": 2.2758,
      "step": 118963
    },
    {
      "epoch": 1.55,
      "grad_norm": 1.0334843397140503,
      "learning_rate": 0.0002846160101989337,
      "loss": 2.1304,
      "step": 118964
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.9600352048873901,
      "learning_rate": 0.00028461192502007507,
      "loss": 2.1776,
      "step": 118965
    },
    {
      "epoch": 1.55,
      "grad_norm": 1.0134299993515015,
      "learning_rate": 0.00028460783984407746,
      "loss": 2.2968,
      "step": 118966
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.9834655523300171,
      "learning_rate": 0.00028460375467094143,
      "loss": 2.3572,
      "step": 118967
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.90968257188797,
      "learning_rate": 0.0002845996695006678,
      "loss": 1.8995,
      "step": 118968
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.9017359018325806,
      "learning_rate": 0.00028459558433325746,
      "loss": 2.1659,
      "step": 118969
    },
    {
      "epoch": 1.55,
      "grad_norm": 1.0352760553359985,
      "learning_rate": 0.0002845914991687111,
      "loss": 2.1444,
      "step": 118970
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.9650681018829346,
      "learning_rate": 0.0002845874140070294,
      "loss": 2.2269,
      "step": 118971
    },
    {
      "epoch": 1.55,
      "grad_norm": 1.0034888982772827,
      "learning_rate": 0.00028458332884821324,
      "loss": 2.1471,
      "step": 118972
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.8763158917427063,
      "learning_rate": 0.00028457924369226344,
      "loss": 2.0026,
      "step": 118973
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.9140402674674988,
      "learning_rate": 0.0002845751585391805,
      "loss": 2.0391,
      "step": 118974
    },
    {
      "epoch": 1.55,
      "grad_norm": 1.0558215379714966,
      "learning_rate": 0.0002845710733889654,
      "loss": 2.2206,
      "step": 118975
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.899785041809082,
      "learning_rate": 0.00028456698824161873,
      "loss": 2.0089,
      "step": 118976
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.8662023544311523,
      "learning_rate": 0.00028456290309714137,
      "loss": 2.082,
      "step": 118977
    },
    {
      "epoch": 1.55,
      "grad_norm": 1.0117720365524292,
      "learning_rate": 0.0002845588179555341,
      "loss": 2.3811,
      "step": 118978
    },
    {
      "epoch": 1.55,
      "grad_norm": 1.0001835823059082,
      "learning_rate": 0.00028455473281679776,
      "loss": 2.3135,
      "step": 118979
    },
    {
      "epoch": 1.55,
      "grad_norm": 1.0499746799468994,
      "learning_rate": 0.00028455064768093287,
      "loss": 2.1404,
      "step": 118980
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.8356113433837891,
      "learning_rate": 0.0002845465625479403,
      "loss": 2.2153,
      "step": 118981
    },
    {
      "epoch": 1.55,
      "grad_norm": 1.111100673675537,
      "learning_rate": 0.00028454247741782076,
      "loss": 2.1273,
      "step": 118982
    },
    {
      "epoch": 1.55,
      "grad_norm": 1.1434537172317505,
      "learning_rate": 0.0002845383922905751,
      "loss": 2.1072,
      "step": 118983
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.9272177815437317,
      "learning_rate": 0.00028453430716620406,
      "loss": 2.1718,
      "step": 118984
    },
    {
      "epoch": 1.55,
      "grad_norm": 1.049920678138733,
      "learning_rate": 0.00028453022204470854,
      "loss": 2.2531,
      "step": 118985
    },
    {
      "epoch": 1.55,
      "grad_norm": 1.1217455863952637,
      "learning_rate": 0.00028452613692608896,
      "loss": 2.2082,
      "step": 118986
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.9122632741928101,
      "learning_rate": 0.00028452205181034627,
      "loss": 2.12,
      "step": 118987
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.9746869206428528,
      "learning_rate": 0.00028451796669748126,
      "loss": 2.1747,
      "step": 118988
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.9817302823066711,
      "learning_rate": 0.0002845138815874946,
      "loss": 2.0982,
      "step": 118989
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.8325963616371155,
      "learning_rate": 0.00028450979648038713,
      "loss": 2.0573,
      "step": 118990
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.9837813377380371,
      "learning_rate": 0.00028450571137615975,
      "loss": 2.1826,
      "step": 118991
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.9549129009246826,
      "learning_rate": 0.00028450162627481287,
      "loss": 2.1706,
      "step": 118992
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.9750385284423828,
      "learning_rate": 0.0002844975411763474,
      "loss": 2.2115,
      "step": 118993
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.8992387652397156,
      "learning_rate": 0.00028449345608076417,
      "loss": 2.1605,
      "step": 118994
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.9463902711868286,
      "learning_rate": 0.0002844893709880639,
      "loss": 2.1521,
      "step": 118995
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.8849448561668396,
      "learning_rate": 0.00028448528589824736,
      "loss": 1.9935,
      "step": 118996
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.9141068458557129,
      "learning_rate": 0.0002844812008113154,
      "loss": 2.3923,
      "step": 118997
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.8858771920204163,
      "learning_rate": 0.0002844771157272686,
      "loss": 2.1531,
      "step": 118998
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.9417873024940491,
      "learning_rate": 0.00028447303064610776,
      "loss": 1.9941,
      "step": 118999
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.9663063287734985,
      "learning_rate": 0.0002844689455678337,
      "loss": 2.1691,
      "step": 119000
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.8666564226150513,
      "learning_rate": 0.0002844648604924471,
      "loss": 2.1839,
      "step": 119001
    },
    {
      "epoch": 1.55,
      "grad_norm": 1.0935386419296265,
      "learning_rate": 0.0002844607754199488,
      "loss": 2.2011,
      "step": 119002
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.970096230506897,
      "learning_rate": 0.0002844566903503396,
      "loss": 2.1907,
      "step": 119003
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.984019935131073,
      "learning_rate": 0.00028445260528362013,
      "loss": 2.2191,
      "step": 119004
    },
    {
      "epoch": 1.55,
      "grad_norm": 1.0569477081298828,
      "learning_rate": 0.0002844485202197913,
      "loss": 2.0527,
      "step": 119005
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.9460654854774475,
      "learning_rate": 0.00028444443515885374,
      "loss": 2.0276,
      "step": 119006
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.9611416459083557,
      "learning_rate": 0.0002844403501008082,
      "loss": 1.9526,
      "step": 119007
    },
    {
      "epoch": 1.55,
      "grad_norm": 1.02254319190979,
      "learning_rate": 0.00028443626504565556,
      "loss": 2.3004,
      "step": 119008
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.9622929096221924,
      "learning_rate": 0.0002844321799933965,
      "loss": 2.0815,
      "step": 119009
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.9480481743812561,
      "learning_rate": 0.0002844280949440318,
      "loss": 2.2619,
      "step": 119010
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.9448070526123047,
      "learning_rate": 0.0002844240098975621,
      "loss": 2.2037,
      "step": 119011
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.9707228541374207,
      "learning_rate": 0.0002844199248539884,
      "loss": 2.1711,
      "step": 119012
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.9764587879180908,
      "learning_rate": 0.00028441583981331133,
      "loss": 2.0524,
      "step": 119013
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.9226899743080139,
      "learning_rate": 0.00028441175477553156,
      "loss": 2.0079,
      "step": 119014
    },
    {
      "epoch": 1.55,
      "grad_norm": 1.0302811861038208,
      "learning_rate": 0.00028440766974065007,
      "loss": 2.3454,
      "step": 119015
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.9946189522743225,
      "learning_rate": 0.00028440358470866736,
      "loss": 2.4488,
      "step": 119016
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.9186781048774719,
      "learning_rate": 0.00028439949967958434,
      "loss": 2.1966,
      "step": 119017
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.8882607817649841,
      "learning_rate": 0.00028439541465340173,
      "loss": 2.1711,
      "step": 119018
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.9760891795158386,
      "learning_rate": 0.0002843913296301204,
      "loss": 2.1509,
      "step": 119019
    },
    {
      "epoch": 1.55,
      "grad_norm": 1.0065531730651855,
      "learning_rate": 0.000284387244609741,
      "loss": 2.3508,
      "step": 119020
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.922980785369873,
      "learning_rate": 0.0002843831595922643,
      "loss": 2.0509,
      "step": 119021
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.9076330661773682,
      "learning_rate": 0.000284379074577691,
      "loss": 2.0169,
      "step": 119022
    },
    {
      "epoch": 1.55,
      "grad_norm": 1.0661773681640625,
      "learning_rate": 0.00028437498956602195,
      "loss": 2.0474,
      "step": 119023
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.9735308885574341,
      "learning_rate": 0.00028437090455725786,
      "loss": 2.2093,
      "step": 119024
    },
    {
      "epoch": 1.55,
      "grad_norm": 1.0262516736984253,
      "learning_rate": 0.00028436681955139954,
      "loss": 1.9895,
      "step": 119025
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.9936836361885071,
      "learning_rate": 0.0002843627345484477,
      "loss": 2.322,
      "step": 119026
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.913491427898407,
      "learning_rate": 0.0002843586495484033,
      "loss": 1.9993,
      "step": 119027
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.9351303577423096,
      "learning_rate": 0.00028435456455126675,
      "loss": 2.1215,
      "step": 119028
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.9019762873649597,
      "learning_rate": 0.000284350479557039,
      "loss": 2.1461,
      "step": 119029
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.9766212105751038,
      "learning_rate": 0.00028434639456572074,
      "loss": 2.3644,
      "step": 119030
    },
    {
      "epoch": 1.55,
      "grad_norm": 1.0071572065353394,
      "learning_rate": 0.0002843423095773128,
      "loss": 2.1597,
      "step": 119031
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.9966840147972107,
      "learning_rate": 0.000284338224591816,
      "loss": 2.1889,
      "step": 119032
    },
    {
      "epoch": 1.55,
      "grad_norm": 1.0254288911819458,
      "learning_rate": 0.0002843341396092311,
      "loss": 2.0753,
      "step": 119033
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.9680282473564148,
      "learning_rate": 0.0002843300546295586,
      "loss": 2.0059,
      "step": 119034
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.9360347986221313,
      "learning_rate": 0.0002843259696527995,
      "loss": 2.0482,
      "step": 119035
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.9377730488777161,
      "learning_rate": 0.0002843218846789545,
      "loss": 1.9868,
      "step": 119036
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.905291736125946,
      "learning_rate": 0.0002843177997080243,
      "loss": 2.0589,
      "step": 119037
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.9038078188896179,
      "learning_rate": 0.00028431371474000976,
      "loss": 2.198,
      "step": 119038
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.9947877526283264,
      "learning_rate": 0.0002843096297749116,
      "loss": 2.1239,
      "step": 119039
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.9195584058761597,
      "learning_rate": 0.0002843055448127307,
      "loss": 2.2909,
      "step": 119040
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.9377796053886414,
      "learning_rate": 0.0002843014598534676,
      "loss": 2.0502,
      "step": 119041
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.9299118518829346,
      "learning_rate": 0.0002842973748971231,
      "loss": 2.0697,
      "step": 119042
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.9722134470939636,
      "learning_rate": 0.000284293289943698,
      "loss": 2.2176,
      "step": 119043
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.8991826176643372,
      "learning_rate": 0.00028428920499319306,
      "loss": 1.9093,
      "step": 119044
    },
    {
      "epoch": 1.55,
      "grad_norm": 1.0259255170822144,
      "learning_rate": 0.00028428512004560915,
      "loss": 2.1062,
      "step": 119045
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.9632154107093811,
      "learning_rate": 0.000284281035100947,
      "loss": 2.1451,
      "step": 119046
    },
    {
      "epoch": 1.55,
      "grad_norm": 1.0502028465270996,
      "learning_rate": 0.00028427695015920715,
      "loss": 2.2217,
      "step": 119047
    },
    {
      "epoch": 1.55,
      "grad_norm": 1.0480544567108154,
      "learning_rate": 0.0002842728652203906,
      "loss": 2.112,
      "step": 119048
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.9336990118026733,
      "learning_rate": 0.00028426878028449794,
      "loss": 2.3569,
      "step": 119049
    },
    {
      "epoch": 1.55,
      "grad_norm": 1.0157068967819214,
      "learning_rate": 0.00028426469535153005,
      "loss": 1.8846,
      "step": 119050
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.9455879330635071,
      "learning_rate": 0.0002842606104214876,
      "loss": 1.9286,
      "step": 119051
    },
    {
      "epoch": 1.55,
      "grad_norm": 1.0717730522155762,
      "learning_rate": 0.0002842565254943716,
      "loss": 2.2225,
      "step": 119052
    },
    {
      "epoch": 1.55,
      "grad_norm": 1.0033223628997803,
      "learning_rate": 0.0002842524405701824,
      "loss": 2.3254,
      "step": 119053
    },
    {
      "epoch": 1.55,
      "grad_norm": 1.0172932147979736,
      "learning_rate": 0.00028424835564892103,
      "loss": 2.0188,
      "step": 119054
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.9625692963600159,
      "learning_rate": 0.0002842442707305882,
      "loss": 2.3695,
      "step": 119055
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.9826609492301941,
      "learning_rate": 0.00028424018581518464,
      "loss": 2.2793,
      "step": 119056
    },
    {
      "epoch": 1.55,
      "grad_norm": 1.2799330949783325,
      "learning_rate": 0.0002842361009027111,
      "loss": 2.2433,
      "step": 119057
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.9715662598609924,
      "learning_rate": 0.0002842320159931685,
      "loss": 2.1277,
      "step": 119058
    },
    {
      "epoch": 1.55,
      "grad_norm": 1.032379388809204,
      "learning_rate": 0.00028422793108655737,
      "loss": 2.0745,
      "step": 119059
    },
    {
      "epoch": 1.55,
      "grad_norm": 1.064643383026123,
      "learning_rate": 0.00028422384618287855,
      "loss": 2.1826,
      "step": 119060
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.9589916467666626,
      "learning_rate": 0.00028421976128213276,
      "loss": 2.0178,
      "step": 119061
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.9199519753456116,
      "learning_rate": 0.0002842156763843209,
      "loss": 2.0908,
      "step": 119062
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.9061824679374695,
      "learning_rate": 0.00028421159148944356,
      "loss": 1.8642,
      "step": 119063
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.9027410745620728,
      "learning_rate": 0.00028420750659750177,
      "loss": 2.1705,
      "step": 119064
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.858982503414154,
      "learning_rate": 0.000284203421708496,
      "loss": 2.1041,
      "step": 119065
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.9740191698074341,
      "learning_rate": 0.00028419933682242705,
      "loss": 2.2059,
      "step": 119066
    },
    {
      "epoch": 1.55,
      "grad_norm": 1.029855728149414,
      "learning_rate": 0.00028419525193929576,
      "loss": 2.0403,
      "step": 119067
    },
    {
      "epoch": 1.55,
      "grad_norm": 1.1154059171676636,
      "learning_rate": 0.00028419116705910287,
      "loss": 2.1185,
      "step": 119068
    },
    {
      "epoch": 1.55,
      "grad_norm": 1.0251966714859009,
      "learning_rate": 0.00028418708218184914,
      "loss": 2.0801,
      "step": 119069
    },
    {
      "epoch": 1.55,
      "grad_norm": 1.0277647972106934,
      "learning_rate": 0.0002841829973075355,
      "loss": 2.0751,
      "step": 119070
    },
    {
      "epoch": 1.55,
      "grad_norm": 1.1280090808868408,
      "learning_rate": 0.0002841789124361624,
      "loss": 2.175,
      "step": 119071
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.9524443745613098,
      "learning_rate": 0.0002841748275677307,
      "loss": 2.2188,
      "step": 119072
    },
    {
      "epoch": 1.55,
      "grad_norm": 1.1084136962890625,
      "learning_rate": 0.0002841707427022412,
      "loss": 2.0758,
      "step": 119073
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.9924524426460266,
      "learning_rate": 0.0002841666578396947,
      "loss": 2.1731,
      "step": 119074
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.9750758409500122,
      "learning_rate": 0.00028416257298009183,
      "loss": 2.3444,
      "step": 119075
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.878201425075531,
      "learning_rate": 0.00028415848812343367,
      "loss": 2.053,
      "step": 119076
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.9172276258468628,
      "learning_rate": 0.00028415440326972057,
      "loss": 2.4489,
      "step": 119077
    },
    {
      "epoch": 1.55,
      "grad_norm": 1.159029483795166,
      "learning_rate": 0.00028415031841895347,
      "loss": 2.2466,
      "step": 119078
    },
    {
      "epoch": 1.55,
      "grad_norm": 1.0195997953414917,
      "learning_rate": 0.0002841462335711331,
      "loss": 2.0468,
      "step": 119079
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.8902739882469177,
      "learning_rate": 0.0002841421487262603,
      "loss": 1.9689,
      "step": 119080
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.8903077840805054,
      "learning_rate": 0.0002841380638843357,
      "loss": 1.9237,
      "step": 119081
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.9962171912193298,
      "learning_rate": 0.00028413397904536035,
      "loss": 2.1043,
      "step": 119082
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.9787880182266235,
      "learning_rate": 0.00028412989420933455,
      "loss": 2.1204,
      "step": 119083
    },
    {
      "epoch": 1.55,
      "grad_norm": 1.0052438974380493,
      "learning_rate": 0.0002841258093762594,
      "loss": 2.2169,
      "step": 119084
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.9543350338935852,
      "learning_rate": 0.0002841217245461355,
      "loss": 1.9615,
      "step": 119085
    },
    {
      "epoch": 1.55,
      "grad_norm": 1.09085214138031,
      "learning_rate": 0.00028411763971896375,
      "loss": 2.1047,
      "step": 119086
    },
    {
      "epoch": 1.55,
      "grad_norm": 1.0352745056152344,
      "learning_rate": 0.00028411355489474475,
      "loss": 2.1078,
      "step": 119087
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.9494829773902893,
      "learning_rate": 0.00028410947007347944,
      "loss": 2.1143,
      "step": 119088
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.9113290905952454,
      "learning_rate": 0.0002841053852551684,
      "loss": 2.2172,
      "step": 119089
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.8972402811050415,
      "learning_rate": 0.00028410130043981255,
      "loss": 2.1273,
      "step": 119090
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.9648236036300659,
      "learning_rate": 0.0002840972156274125,
      "loss": 2.3099,
      "step": 119091
    },
    {
      "epoch": 1.55,
      "grad_norm": 1.1323597431182861,
      "learning_rate": 0.0002840931308179691,
      "loss": 2.1902,
      "step": 119092
    },
    {
      "epoch": 1.55,
      "grad_norm": 1.0644608736038208,
      "learning_rate": 0.00028408904601148305,
      "loss": 1.9733,
      "step": 119093
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.9789057970046997,
      "learning_rate": 0.00028408496120795523,
      "loss": 2.1377,
      "step": 119094
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.9182535409927368,
      "learning_rate": 0.00028408087640738626,
      "loss": 2.0834,
      "step": 119095
    },
    {
      "epoch": 1.55,
      "grad_norm": 1.1666849851608276,
      "learning_rate": 0.00028407679160977694,
      "loss": 1.9577,
      "step": 119096
    },
    {
      "epoch": 1.55,
      "grad_norm": 1.2114828824996948,
      "learning_rate": 0.0002840727068151281,
      "loss": 2.0005,
      "step": 119097
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.9011900424957275,
      "learning_rate": 0.0002840686220234404,
      "loss": 2.0885,
      "step": 119098
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.9677206873893738,
      "learning_rate": 0.00028406453723471466,
      "loss": 1.9402,
      "step": 119099
    },
    {
      "epoch": 1.55,
      "grad_norm": 1.2219692468643188,
      "learning_rate": 0.0002840604524489516,
      "loss": 2.058,
      "step": 119100
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.9146994948387146,
      "learning_rate": 0.000284056367666152,
      "loss": 2.1229,
      "step": 119101
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.9660525321960449,
      "learning_rate": 0.00028405228288631666,
      "loss": 2.005,
      "step": 119102
    },
    {
      "epoch": 1.55,
      "grad_norm": 1.1453328132629395,
      "learning_rate": 0.0002840481981094463,
      "loss": 2.054,
      "step": 119103
    },
    {
      "epoch": 1.55,
      "grad_norm": 1.2744333744049072,
      "learning_rate": 0.0002840441133355417,
      "loss": 2.3295,
      "step": 119104
    },
    {
      "epoch": 1.55,
      "grad_norm": 1.0180572271347046,
      "learning_rate": 0.00028404002856460353,
      "loss": 2.3756,
      "step": 119105
    },
    {
      "epoch": 1.55,
      "grad_norm": 1.110865831375122,
      "learning_rate": 0.0002840359437966326,
      "loss": 2.3995,
      "step": 119106
    },
    {
      "epoch": 1.55,
      "grad_norm": 1.235982894897461,
      "learning_rate": 0.0002840318590316298,
      "loss": 2.1442,
      "step": 119107
    },
    {
      "epoch": 1.55,
      "grad_norm": 1.0535109043121338,
      "learning_rate": 0.0002840277742695957,
      "loss": 2.0434,
      "step": 119108
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.9765143990516663,
      "learning_rate": 0.00028402368951053114,
      "loss": 2.1606,
      "step": 119109
    },
    {
      "epoch": 1.55,
      "grad_norm": 1.2356575727462769,
      "learning_rate": 0.00028401960475443695,
      "loss": 2.0638,
      "step": 119110
    },
    {
      "epoch": 1.55,
      "grad_norm": 1.0070593357086182,
      "learning_rate": 0.0002840155200013138,
      "loss": 2.084,
      "step": 119111
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.9461245536804199,
      "learning_rate": 0.0002840114352511624,
      "loss": 2.2533,
      "step": 119112
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.935698926448822,
      "learning_rate": 0.00028400735050398366,
      "loss": 2.2173,
      "step": 119113
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.8593000769615173,
      "learning_rate": 0.0002840032657597782,
      "loss": 1.8055,
      "step": 119114
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.9699970483779907,
      "learning_rate": 0.0002839991810185468,
      "loss": 2.1723,
      "step": 119115
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.8821092247962952,
      "learning_rate": 0.00028399509628029027,
      "loss": 2.1685,
      "step": 119116
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.8647558689117432,
      "learning_rate": 0.00028399101154500935,
      "loss": 2.1059,
      "step": 119117
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.9031293392181396,
      "learning_rate": 0.00028398692681270483,
      "loss": 2.0864,
      "step": 119118
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.9704684615135193,
      "learning_rate": 0.00028398284208337753,
      "loss": 1.968,
      "step": 119119
    },
    {
      "epoch": 1.55,
      "grad_norm": 1.0969772338867188,
      "learning_rate": 0.000283978757357028,
      "loss": 2.0877,
      "step": 119120
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.839572012424469,
      "learning_rate": 0.00028397467263365715,
      "loss": 2.1091,
      "step": 119121
    },
    {
      "epoch": 1.55,
      "grad_norm": 1.07565176486969,
      "learning_rate": 0.00028397058791326574,
      "loss": 2.518,
      "step": 119122
    },
    {
      "epoch": 1.55,
      "grad_norm": 1.003796100616455,
      "learning_rate": 0.0002839665031958544,
      "loss": 2.1183,
      "step": 119123
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.9914668798446655,
      "learning_rate": 0.00028396241848142406,
      "loss": 2.2834,
      "step": 119124
    },
    {
      "epoch": 1.55,
      "grad_norm": 1.0537590980529785,
      "learning_rate": 0.00028395833376997553,
      "loss": 2.1761,
      "step": 119125
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.9185101389884949,
      "learning_rate": 0.00028395424906150937,
      "loss": 2.2134,
      "step": 119126
    },
    {
      "epoch": 1.55,
      "grad_norm": 1.0521680116653442,
      "learning_rate": 0.00028395016435602634,
      "loss": 2.131,
      "step": 119127
    },
    {
      "epoch": 1.55,
      "grad_norm": 1.0111713409423828,
      "learning_rate": 0.0002839460796535273,
      "loss": 2.1092,
      "step": 119128
    },
    {
      "epoch": 1.55,
      "grad_norm": 1.0430703163146973,
      "learning_rate": 0.000283941994954013,
      "loss": 2.0325,
      "step": 119129
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.9343959093093872,
      "learning_rate": 0.00028393791025748414,
      "loss": 2.1058,
      "step": 119130
    },
    {
      "epoch": 1.55,
      "grad_norm": 1.0849905014038086,
      "learning_rate": 0.0002839338255639417,
      "loss": 1.9908,
      "step": 119131
    },
    {
      "epoch": 1.55,
      "grad_norm": 1.1742091178894043,
      "learning_rate": 0.00028392974087338617,
      "loss": 2.1627,
      "step": 119132
    },
    {
      "epoch": 1.55,
      "grad_norm": 1.0266755819320679,
      "learning_rate": 0.00028392565618581833,
      "loss": 2.3272,
      "step": 119133
    },
    {
      "epoch": 1.55,
      "grad_norm": 1.0905001163482666,
      "learning_rate": 0.00028392157150123904,
      "loss": 2.0153,
      "step": 119134
    },
    {
      "epoch": 1.55,
      "grad_norm": 1.035970687866211,
      "learning_rate": 0.00028391748681964906,
      "loss": 2.2714,
      "step": 119135
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.9300143122673035,
      "learning_rate": 0.0002839134021410491,
      "loss": 2.318,
      "step": 119136
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.9733731150627136,
      "learning_rate": 0.00028390931746544005,
      "loss": 2.0536,
      "step": 119137
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.9370360374450684,
      "learning_rate": 0.0002839052327928225,
      "loss": 2.0633,
      "step": 119138
    },
    {
      "epoch": 1.55,
      "grad_norm": 1.011458158493042,
      "learning_rate": 0.00028390114812319717,
      "loss": 2.2243,
      "step": 119139
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.8984826803207397,
      "learning_rate": 0.000283897063456565,
      "loss": 2.1793,
      "step": 119140
    },
    {
      "epoch": 1.55,
      "grad_norm": 1.0003228187561035,
      "learning_rate": 0.00028389297879292664,
      "loss": 2.0423,
      "step": 119141
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.9818565249443054,
      "learning_rate": 0.0002838888941322829,
      "loss": 1.8457,
      "step": 119142
    },
    {
      "epoch": 1.55,
      "grad_norm": 1.0391335487365723,
      "learning_rate": 0.00028388480947463466,
      "loss": 2.0952,
      "step": 119143
    },
    {
      "epoch": 1.55,
      "grad_norm": 1.0542432069778442,
      "learning_rate": 0.0002838807248199823,
      "loss": 2.098,
      "step": 119144
    },
    {
      "epoch": 1.55,
      "grad_norm": 1.0266354084014893,
      "learning_rate": 0.00028387664016832695,
      "loss": 2.0697,
      "step": 119145
    },
    {
      "epoch": 1.55,
      "grad_norm": 1.01449453830719,
      "learning_rate": 0.0002838725555196692,
      "loss": 2.1286,
      "step": 119146
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.9250666499137878,
      "learning_rate": 0.0002838684708740098,
      "loss": 1.9789,
      "step": 119147
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.9840603470802307,
      "learning_rate": 0.0002838643862313496,
      "loss": 1.9469,
      "step": 119148
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.9046909809112549,
      "learning_rate": 0.00028386030159168943,
      "loss": 2.1437,
      "step": 119149
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.934306800365448,
      "learning_rate": 0.0002838562169550298,
      "loss": 2.0723,
      "step": 119150
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.9842398166656494,
      "learning_rate": 0.00028385213232137165,
      "loss": 1.9333,
      "step": 119151
    },
    {
      "epoch": 1.55,
      "grad_norm": 1.080627202987671,
      "learning_rate": 0.00028384804769071564,
      "loss": 2.1513,
      "step": 119152
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.9478377103805542,
      "learning_rate": 0.00028384396306306255,
      "loss": 2.0394,
      "step": 119153
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.9609899520874023,
      "learning_rate": 0.00028383987843841324,
      "loss": 2.1593,
      "step": 119154
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.9318777322769165,
      "learning_rate": 0.0002838357938167685,
      "loss": 2.2573,
      "step": 119155
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.8966701626777649,
      "learning_rate": 0.0002838317091981289,
      "loss": 2.1894,
      "step": 119156
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.9913163185119629,
      "learning_rate": 0.00028382762458249527,
      "loss": 2.3041,
      "step": 119157
    },
    {
      "epoch": 1.55,
      "grad_norm": 1.00291109085083,
      "learning_rate": 0.0002838235399698684,
      "loss": 2.173,
      "step": 119158
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.9639925956726074,
      "learning_rate": 0.000283819455360249,
      "loss": 2.1012,
      "step": 119159
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.9561983346939087,
      "learning_rate": 0.0002838153707536379,
      "loss": 2.1978,
      "step": 119160
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.9160599112510681,
      "learning_rate": 0.000283811286150036,
      "loss": 1.9366,
      "step": 119161
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.9826639890670776,
      "learning_rate": 0.00028380720154944365,
      "loss": 2.3268,
      "step": 119162
    },
    {
      "epoch": 1.55,
      "grad_norm": 1.0087170600891113,
      "learning_rate": 0.0002838031169518619,
      "loss": 2.1386,
      "step": 119163
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.941892683506012,
      "learning_rate": 0.0002837990323572915,
      "loss": 2.2672,
      "step": 119164
    },
    {
      "epoch": 1.55,
      "grad_norm": 1.0078974962234497,
      "learning_rate": 0.0002837949477657331,
      "loss": 2.0562,
      "step": 119165
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.9955587387084961,
      "learning_rate": 0.00028379086317718757,
      "loss": 1.868,
      "step": 119166
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.9071598052978516,
      "learning_rate": 0.00028378677859165573,
      "loss": 2.0575,
      "step": 119167
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.9826322197914124,
      "learning_rate": 0.0002837826940091381,
      "loss": 2.2175,
      "step": 119168
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.9424929618835449,
      "learning_rate": 0.00028377860942963555,
      "loss": 2.1742,
      "step": 119169
    },
    {
      "epoch": 1.55,
      "grad_norm": 1.0714168548583984,
      "learning_rate": 0.00028377452485314894,
      "loss": 2.1812,
      "step": 119170
    },
    {
      "epoch": 1.55,
      "grad_norm": 1.0580720901489258,
      "learning_rate": 0.0002837704402796789,
      "loss": 2.0979,
      "step": 119171
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.8598310947418213,
      "learning_rate": 0.00028376635570922625,
      "loss": 1.9412,
      "step": 119172
    },
    {
      "epoch": 1.55,
      "grad_norm": 1.0570248365402222,
      "learning_rate": 0.00028376227114179176,
      "loss": 2.1238,
      "step": 119173
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.8602997064590454,
      "learning_rate": 0.0002837581865773763,
      "loss": 2.1725,
      "step": 119174
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.8982151746749878,
      "learning_rate": 0.0002837541020159803,
      "loss": 2.2612,
      "step": 119175
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.9436346888542175,
      "learning_rate": 0.00028375001745760474,
      "loss": 2.1455,
      "step": 119176
    },
    {
      "epoch": 1.55,
      "grad_norm": 1.0133415460586548,
      "learning_rate": 0.0002837459329022504,
      "loss": 2.1742,
      "step": 119177
    },
    {
      "epoch": 1.55,
      "grad_norm": 1.0437109470367432,
      "learning_rate": 0.00028374184834991804,
      "loss": 2.1959,
      "step": 119178
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.9040817022323608,
      "learning_rate": 0.00028373776380060826,
      "loss": 2.0868,
      "step": 119179
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.9559151530265808,
      "learning_rate": 0.0002837336792543221,
      "loss": 2.0764,
      "step": 119180
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.94197678565979,
      "learning_rate": 0.00028372959471106005,
      "loss": 1.9737,
      "step": 119181
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.9356025457382202,
      "learning_rate": 0.00028372551017082304,
      "loss": 2.2561,
      "step": 119182
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.8605765700340271,
      "learning_rate": 0.0002837214256336117,
      "loss": 2.1136,
      "step": 119183
    },
    {
      "epoch": 1.55,
      "grad_norm": 1.1455104351043701,
      "learning_rate": 0.00028371734109942687,
      "loss": 2.1203,
      "step": 119184
    },
    {
      "epoch": 1.55,
      "grad_norm": 1.0033462047576904,
      "learning_rate": 0.0002837132565682693,
      "loss": 2.1629,
      "step": 119185
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.8766483664512634,
      "learning_rate": 0.0002837091720401398,
      "loss": 2.1168,
      "step": 119186
    },
    {
      "epoch": 1.55,
      "grad_norm": 1.0084041357040405,
      "learning_rate": 0.000283705087515039,
      "loss": 1.9608,
      "step": 119187
    },
    {
      "epoch": 1.55,
      "grad_norm": 1.0764281749725342,
      "learning_rate": 0.0002837010029929677,
      "loss": 1.9758,
      "step": 119188
    },
    {
      "epoch": 1.55,
      "grad_norm": 1.0900111198425293,
      "learning_rate": 0.0002836969184739268,
      "loss": 2.0868,
      "step": 119189
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.9167469143867493,
      "learning_rate": 0.0002836928339579169,
      "loss": 1.8939,
      "step": 119190
    },
    {
      "epoch": 1.55,
      "grad_norm": 1.1034613847732544,
      "learning_rate": 0.00028368874944493876,
      "loss": 2.2233,
      "step": 119191
    },
    {
      "epoch": 1.55,
      "grad_norm": 1.0204236507415771,
      "learning_rate": 0.0002836846649349933,
      "loss": 2.0571,
      "step": 119192
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.9417355060577393,
      "learning_rate": 0.00028368058042808105,
      "loss": 1.8711,
      "step": 119193
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.8974796533584595,
      "learning_rate": 0.0002836764959242029,
      "loss": 2.4313,
      "step": 119194
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.987038254737854,
      "learning_rate": 0.00028367241142335963,
      "loss": 2.1629,
      "step": 119195
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.9589335918426514,
      "learning_rate": 0.000283668326925552,
      "loss": 2.0739,
      "step": 119196
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.8886386156082153,
      "learning_rate": 0.00028366424243078073,
      "loss": 1.8426,
      "step": 119197
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.8808046579360962,
      "learning_rate": 0.00028366015793904664,
      "loss": 2.1515,
      "step": 119198
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.8999625444412231,
      "learning_rate": 0.0002836560734503503,
      "loss": 2.0993,
      "step": 119199
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.8985041379928589,
      "learning_rate": 0.00028365198896469264,
      "loss": 2.1045,
      "step": 119200
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.9314185380935669,
      "learning_rate": 0.0002836479044820744,
      "loss": 2.2865,
      "step": 119201
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.8738924860954285,
      "learning_rate": 0.0002836438200024963,
      "loss": 2.3023,
      "step": 119202
    },
    {
      "epoch": 1.55,
      "grad_norm": 1.0540690422058105,
      "learning_rate": 0.00028363973552595923,
      "loss": 2.1978,
      "step": 119203
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.885985791683197,
      "learning_rate": 0.0002836356510524638,
      "loss": 2.0318,
      "step": 119204
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.9253600239753723,
      "learning_rate": 0.00028363156658201076,
      "loss": 2.0665,
      "step": 119205
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.9517641067504883,
      "learning_rate": 0.0002836274821146009,
      "loss": 2.0791,
      "step": 119206
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.928557813167572,
      "learning_rate": 0.00028362339765023505,
      "loss": 2.2301,
      "step": 119207
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.8771290183067322,
      "learning_rate": 0.00028361931318891386,
      "loss": 2.1628,
      "step": 119208
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.9459192752838135,
      "learning_rate": 0.0002836152287306382,
      "loss": 1.9745,
      "step": 119209
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.9808624982833862,
      "learning_rate": 0.00028361114427540886,
      "loss": 2.2102,
      "step": 119210
    },
    {
      "epoch": 1.55,
      "grad_norm": 1.0912836790084839,
      "learning_rate": 0.00028360705982322646,
      "loss": 2.1825,
      "step": 119211
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.9841942191123962,
      "learning_rate": 0.00028360297537409176,
      "loss": 2.2064,
      "step": 119212
    },
    {
      "epoch": 1.55,
      "grad_norm": 1.0218310356140137,
      "learning_rate": 0.00028359889092800556,
      "loss": 2.0644,
      "step": 119213
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.9094985723495483,
      "learning_rate": 0.0002835948064849687,
      "loss": 2.1582,
      "step": 119214
    },
    {
      "epoch": 1.55,
      "grad_norm": 1.0743571519851685,
      "learning_rate": 0.0002835907220449818,
      "loss": 1.9755,
      "step": 119215
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.8861563205718994,
      "learning_rate": 0.0002835866376080459,
      "loss": 2.2882,
      "step": 119216
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.9497637748718262,
      "learning_rate": 0.00028358255317416137,
      "loss": 2.1925,
      "step": 119217
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.9370418190956116,
      "learning_rate": 0.00028357846874332913,
      "loss": 2.1147,
      "step": 119218
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.9587051868438721,
      "learning_rate": 0.00028357438431555,
      "loss": 2.2285,
      "step": 119219
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.8867002129554749,
      "learning_rate": 0.0002835702998908247,
      "loss": 2.3176,
      "step": 119220
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.9893280267715454,
      "learning_rate": 0.00028356621546915404,
      "loss": 2.2466,
      "step": 119221
    },
    {
      "epoch": 1.55,
      "grad_norm": 1.0302765369415283,
      "learning_rate": 0.0002835621310505388,
      "loss": 2.1141,
      "step": 119222
    },
    {
      "epoch": 1.55,
      "grad_norm": 1.018701434135437,
      "learning_rate": 0.00028355804663497957,
      "loss": 2.289,
      "step": 119223
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.9028252959251404,
      "learning_rate": 0.0002835539622224772,
      "loss": 2.2442,
      "step": 119224
    },
    {
      "epoch": 1.55,
      "grad_norm": 1.1086666584014893,
      "learning_rate": 0.00028354987781303246,
      "loss": 2.2234,
      "step": 119225
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.945210874080658,
      "learning_rate": 0.00028354579340664606,
      "loss": 2.0412,
      "step": 119226
    },
    {
      "epoch": 1.55,
      "grad_norm": 1.1647865772247314,
      "learning_rate": 0.0002835417090033189,
      "loss": 2.1757,
      "step": 119227
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.9947218894958496,
      "learning_rate": 0.0002835376246030517,
      "loss": 2.2706,
      "step": 119228
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.9768784642219543,
      "learning_rate": 0.00028353354020584505,
      "loss": 2.0927,
      "step": 119229
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.9967820644378662,
      "learning_rate": 0.0002835294558116999,
      "loss": 2.2406,
      "step": 119230
    },
    {
      "epoch": 1.55,
      "grad_norm": 1.006595492362976,
      "learning_rate": 0.00028352537142061683,
      "loss": 2.1114,
      "step": 119231
    },
    {
      "epoch": 1.55,
      "grad_norm": 1.1940118074417114,
      "learning_rate": 0.00028352128703259676,
      "loss": 2.0815,
      "step": 119232
    },
    {
      "epoch": 1.55,
      "grad_norm": 1.190262794494629,
      "learning_rate": 0.00028351720264764037,
      "loss": 2.1096,
      "step": 119233
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.9257284998893738,
      "learning_rate": 0.0002835131182657486,
      "loss": 2.0705,
      "step": 119234
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.9554482698440552,
      "learning_rate": 0.0002835090338869219,
      "loss": 2.0206,
      "step": 119235
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.9748983383178711,
      "learning_rate": 0.00028350494951116117,
      "loss": 2.1007,
      "step": 119236
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.955690860748291,
      "learning_rate": 0.0002835008651384672,
      "loss": 2.1236,
      "step": 119237
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.9901659488677979,
      "learning_rate": 0.00028349678076884075,
      "loss": 1.9871,
      "step": 119238
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.9443225860595703,
      "learning_rate": 0.00028349269640228256,
      "loss": 2.4136,
      "step": 119239
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.9426005482673645,
      "learning_rate": 0.0002834886120387934,
      "loss": 2.1912,
      "step": 119240
    },
    {
      "epoch": 1.55,
      "grad_norm": 1.2267241477966309,
      "learning_rate": 0.0002834845276783741,
      "loss": 2.0609,
      "step": 119241
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.9175817370414734,
      "learning_rate": 0.0002834804433210252,
      "loss": 2.0549,
      "step": 119242
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.9323165416717529,
      "learning_rate": 0.0002834763589667476,
      "loss": 1.9879,
      "step": 119243
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.9362453818321228,
      "learning_rate": 0.0002834722746155421,
      "loss": 2.1797,
      "step": 119244
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.9937335252761841,
      "learning_rate": 0.0002834681902674094,
      "loss": 2.1729,
      "step": 119245
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.9235207438468933,
      "learning_rate": 0.0002834641059223503,
      "loss": 2.1183,
      "step": 119246
    },
    {
      "epoch": 1.55,
      "grad_norm": 1.0806416273117065,
      "learning_rate": 0.0002834600215803656,
      "loss": 2.2864,
      "step": 119247
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.9365121126174927,
      "learning_rate": 0.00028345593724145594,
      "loss": 2.3032,
      "step": 119248
    },
    {
      "epoch": 1.55,
      "grad_norm": 1.1358131170272827,
      "learning_rate": 0.0002834518529056221,
      "loss": 1.9427,
      "step": 119249
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.9652995467185974,
      "learning_rate": 0.00028344776857286484,
      "loss": 2.1985,
      "step": 119250
    },
    {
      "epoch": 1.55,
      "grad_norm": 1.0648021697998047,
      "learning_rate": 0.000283443684243185,
      "loss": 2.1772,
      "step": 119251
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.9966262578964233,
      "learning_rate": 0.0002834395999165833,
      "loss": 2.4391,
      "step": 119252
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.9128237366676331,
      "learning_rate": 0.00028343551559306055,
      "loss": 2.2219,
      "step": 119253
    },
    {
      "epoch": 1.55,
      "grad_norm": 1.1059834957122803,
      "learning_rate": 0.0002834314312726174,
      "loss": 2.1543,
      "step": 119254
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.9375867247581482,
      "learning_rate": 0.0002834273469552546,
      "loss": 2.1099,
      "step": 119255
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.9382813572883606,
      "learning_rate": 0.00028342326264097303,
      "loss": 2.0562,
      "step": 119256
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.8969584703445435,
      "learning_rate": 0.00028341917832977334,
      "loss": 1.9542,
      "step": 119257
    },
    {
      "epoch": 1.55,
      "grad_norm": 1.0550897121429443,
      "learning_rate": 0.00028341509402165637,
      "loss": 2.2306,
      "step": 119258
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.9177666902542114,
      "learning_rate": 0.000283411009716623,
      "loss": 2.2109,
      "step": 119259
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.9441798329353333,
      "learning_rate": 0.0002834069254146736,
      "loss": 2.205,
      "step": 119260
    },
    {
      "epoch": 1.55,
      "grad_norm": 1.165643572807312,
      "learning_rate": 0.0002834028411158092,
      "loss": 2.0308,
      "step": 119261
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.9504329562187195,
      "learning_rate": 0.00028339875682003055,
      "loss": 2.0614,
      "step": 119262
    },
    {
      "epoch": 1.55,
      "grad_norm": 1.0870468616485596,
      "learning_rate": 0.0002833946725273384,
      "loss": 1.8265,
      "step": 119263
    },
    {
      "epoch": 1.55,
      "grad_norm": 1.0643707513809204,
      "learning_rate": 0.0002833905882377335,
      "loss": 2.3213,
      "step": 119264
    },
    {
      "epoch": 1.55,
      "grad_norm": 1.1586415767669678,
      "learning_rate": 0.0002833865039512167,
      "loss": 2.0921,
      "step": 119265
    },
    {
      "epoch": 1.55,
      "grad_norm": 1.0018187761306763,
      "learning_rate": 0.0002833824196677886,
      "loss": 2.3077,
      "step": 119266
    },
    {
      "epoch": 1.55,
      "grad_norm": 1.0065819025039673,
      "learning_rate": 0.00028337833538744995,
      "loss": 2.089,
      "step": 119267
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.9056618213653564,
      "learning_rate": 0.0002833742511102016,
      "loss": 2.1157,
      "step": 119268
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.9886027574539185,
      "learning_rate": 0.0002833701668360443,
      "loss": 2.1449,
      "step": 119269
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.953877866268158,
      "learning_rate": 0.0002833660825649788,
      "loss": 1.9511,
      "step": 119270
    },
    {
      "epoch": 1.55,
      "grad_norm": 1.005928635597229,
      "learning_rate": 0.00028336199829700587,
      "loss": 2.1959,
      "step": 119271
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.9642641544342041,
      "learning_rate": 0.0002833579140321263,
      "loss": 2.2884,
      "step": 119272
    },
    {
      "epoch": 1.55,
      "grad_norm": 1.2442480325698853,
      "learning_rate": 0.0002833538297703408,
      "loss": 2.1149,
      "step": 119273
    },
    {
      "epoch": 1.55,
      "grad_norm": 1.0404878854751587,
      "learning_rate": 0.00028334974551165004,
      "loss": 2.0677,
      "step": 119274
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.927558183670044,
      "learning_rate": 0.00028334566125605496,
      "loss": 1.9734,
      "step": 119275
    },
    {
      "epoch": 1.55,
      "grad_norm": 1.0280121564865112,
      "learning_rate": 0.0002833415770035562,
      "loss": 1.9726,
      "step": 119276
    },
    {
      "epoch": 1.55,
      "grad_norm": 1.0464121103286743,
      "learning_rate": 0.0002833374927541546,
      "loss": 2.339,
      "step": 119277
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.9896325469017029,
      "learning_rate": 0.0002833334085078508,
      "loss": 2.3896,
      "step": 119278
    },
    {
      "epoch": 1.55,
      "grad_norm": 1.0420335531234741,
      "learning_rate": 0.00028332932426464565,
      "loss": 2.0507,
      "step": 119279
    },
    {
      "epoch": 1.55,
      "grad_norm": 1.039697289466858,
      "learning_rate": 0.00028332524002453997,
      "loss": 2.0994,
      "step": 119280
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.9726312756538391,
      "learning_rate": 0.0002833211557875344,
      "loss": 1.9441,
      "step": 119281
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.9629191756248474,
      "learning_rate": 0.00028331707155362977,
      "loss": 2.1303,
      "step": 119282
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.8628922700881958,
      "learning_rate": 0.00028331298732282677,
      "loss": 2.3359,
      "step": 119283
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.8408970832824707,
      "learning_rate": 0.0002833089030951262,
      "loss": 1.9162,
      "step": 119284
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.9472417235374451,
      "learning_rate": 0.00028330481887052885,
      "loss": 2.0792,
      "step": 119285
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.9278658032417297,
      "learning_rate": 0.0002833007346490354,
      "loss": 2.4425,
      "step": 119286
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.9498440623283386,
      "learning_rate": 0.00028329665043064667,
      "loss": 2.2359,
      "step": 119287
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.9682263135910034,
      "learning_rate": 0.0002832925662153635,
      "loss": 2.2833,
      "step": 119288
    },
    {
      "epoch": 1.55,
      "grad_norm": 1.094896912574768,
      "learning_rate": 0.0002832884820031865,
      "loss": 2.1997,
      "step": 119289
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.9833877086639404,
      "learning_rate": 0.0002832843977941164,
      "loss": 2.2127,
      "step": 119290
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.9433977007865906,
      "learning_rate": 0.00028328031358815415,
      "loss": 2.2408,
      "step": 119291
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.8923327922821045,
      "learning_rate": 0.00028327622938530036,
      "loss": 2.3104,
      "step": 119292
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.8258569836616516,
      "learning_rate": 0.0002832721451855558,
      "loss": 2.1365,
      "step": 119293
    },
    {
      "epoch": 1.55,
      "grad_norm": 1.0063419342041016,
      "learning_rate": 0.0002832680609889213,
      "loss": 2.2849,
      "step": 119294
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.9260748624801636,
      "learning_rate": 0.00028326397679539773,
      "loss": 2.1289,
      "step": 119295
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.9526421427726746,
      "learning_rate": 0.0002832598926049855,
      "loss": 2.2496,
      "step": 119296
    },
    {
      "epoch": 1.55,
      "grad_norm": 1.006553292274475,
      "learning_rate": 0.00028325580841768566,
      "loss": 2.3109,
      "step": 119297
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.9901288151741028,
      "learning_rate": 0.0002832517242334988,
      "loss": 2.2244,
      "step": 119298
    },
    {
      "epoch": 1.55,
      "grad_norm": 1.0635356903076172,
      "learning_rate": 0.0002832476400524258,
      "loss": 2.149,
      "step": 119299
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.8519451022148132,
      "learning_rate": 0.0002832435558744674,
      "loss": 2.041,
      "step": 119300
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.9539108872413635,
      "learning_rate": 0.0002832394716996244,
      "loss": 2.0988,
      "step": 119301
    },
    {
      "epoch": 1.55,
      "grad_norm": 1.0619527101516724,
      "learning_rate": 0.0002832353875278975,
      "loss": 2.0441,
      "step": 119302
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.950267493724823,
      "learning_rate": 0.00028323130335928734,
      "loss": 2.1551,
      "step": 119303
    },
    {
      "epoch": 1.55,
      "grad_norm": 1.0752674341201782,
      "learning_rate": 0.00028322721919379483,
      "loss": 2.0887,
      "step": 119304
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.9806268215179443,
      "learning_rate": 0.0002832231350314207,
      "loss": 2.1983,
      "step": 119305
    },
    {
      "epoch": 1.55,
      "grad_norm": 1.0562763214111328,
      "learning_rate": 0.0002832190508721657,
      "loss": 2.041,
      "step": 119306
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.8694740533828735,
      "learning_rate": 0.00028321496671603063,
      "loss": 2.147,
      "step": 119307
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.9795244932174683,
      "learning_rate": 0.0002832108825630163,
      "loss": 2.1489,
      "step": 119308
    },
    {
      "epoch": 1.55,
      "grad_norm": 1.011287808418274,
      "learning_rate": 0.0002832067984131233,
      "loss": 2.0765,
      "step": 119309
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.9280433654785156,
      "learning_rate": 0.0002832027142663524,
      "loss": 2.2337,
      "step": 119310
    },
    {
      "epoch": 1.55,
      "grad_norm": 1.1302404403686523,
      "learning_rate": 0.00028319863012270447,
      "loss": 2.2493,
      "step": 119311
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.9603665471076965,
      "learning_rate": 0.00028319454598218027,
      "loss": 2.1012,
      "step": 119312
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.9509042501449585,
      "learning_rate": 0.00028319046184478047,
      "loss": 2.0312,
      "step": 119313
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.9937143325805664,
      "learning_rate": 0.00028318637771050604,
      "loss": 2.007,
      "step": 119314
    },
    {
      "epoch": 1.55,
      "grad_norm": 1.0958408117294312,
      "learning_rate": 0.0002831822935793574,
      "loss": 2.2596,
      "step": 119315
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.8583233952522278,
      "learning_rate": 0.00028317820945133554,
      "loss": 2.0653,
      "step": 119316
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.9950959086418152,
      "learning_rate": 0.0002831741253264412,
      "loss": 2.0739,
      "step": 119317
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.9315088391304016,
      "learning_rate": 0.00028317004120467505,
      "loss": 2.1197,
      "step": 119318
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.9528975486755371,
      "learning_rate": 0.0002831659570860379,
      "loss": 2.208,
      "step": 119319
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.8976577520370483,
      "learning_rate": 0.0002831618729705307,
      "loss": 2.1267,
      "step": 119320
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.9547011852264404,
      "learning_rate": 0.00028315778885815385,
      "loss": 2.3296,
      "step": 119321
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.9769003987312317,
      "learning_rate": 0.0002831537047489083,
      "loss": 2.1963,
      "step": 119322
    },
    {
      "epoch": 1.55,
      "grad_norm": 1.0303534269332886,
      "learning_rate": 0.00028314962064279476,
      "loss": 2.1189,
      "step": 119323
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.8859909772872925,
      "learning_rate": 0.0002831455365398141,
      "loss": 2.0155,
      "step": 119324
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.9570479989051819,
      "learning_rate": 0.00028314145243996694,
      "loss": 2.167,
      "step": 119325
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.935494601726532,
      "learning_rate": 0.00028313736834325426,
      "loss": 2.184,
      "step": 119326
    },
    {
      "epoch": 1.55,
      "grad_norm": 1.004850149154663,
      "learning_rate": 0.0002831332842496765,
      "loss": 2.1252,
      "step": 119327
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.9210091233253479,
      "learning_rate": 0.00028312920015923464,
      "loss": 2.0724,
      "step": 119328
    },
    {
      "epoch": 1.55,
      "grad_norm": 1.0048019886016846,
      "learning_rate": 0.0002831251160719293,
      "loss": 2.0232,
      "step": 119329
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.9564798474311829,
      "learning_rate": 0.0002831210319877614,
      "loss": 2.2139,
      "step": 119330
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.8491787910461426,
      "learning_rate": 0.0002831169479067316,
      "loss": 1.9543,
      "step": 119331
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.9902552366256714,
      "learning_rate": 0.0002831128638288408,
      "loss": 2.1117,
      "step": 119332
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.9112958312034607,
      "learning_rate": 0.00028310877975408944,
      "loss": 2.2781,
      "step": 119333
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.9586983919143677,
      "learning_rate": 0.00028310469568247854,
      "loss": 2.1533,
      "step": 119334
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.9648321866989136,
      "learning_rate": 0.0002831006116140088,
      "loss": 2.2207,
      "step": 119335
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.9751231074333191,
      "learning_rate": 0.00028309652754868094,
      "loss": 1.9779,
      "step": 119336
    },
    {
      "epoch": 1.55,
      "grad_norm": 1.0179193019866943,
      "learning_rate": 0.0002830924434864958,
      "loss": 2.2857,
      "step": 119337
    },
    {
      "epoch": 1.55,
      "grad_norm": 1.0395840406417847,
      "learning_rate": 0.00028308835942745424,
      "loss": 2.0998,
      "step": 119338
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.9441840052604675,
      "learning_rate": 0.0002830842753715567,
      "loss": 2.1489,
      "step": 119339
    },
    {
      "epoch": 1.55,
      "grad_norm": 1.0433006286621094,
      "learning_rate": 0.0002830801913188041,
      "loss": 2.1775,
      "step": 119340
    },
    {
      "epoch": 1.55,
      "grad_norm": 1.05793297290802,
      "learning_rate": 0.0002830761072691972,
      "loss": 1.9842,
      "step": 119341
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.8851697444915771,
      "learning_rate": 0.0002830720232227368,
      "loss": 1.9576,
      "step": 119342
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.8953981399536133,
      "learning_rate": 0.00028306793917942365,
      "loss": 1.9792,
      "step": 119343
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.8515493869781494,
      "learning_rate": 0.0002830638551392586,
      "loss": 2.1294,
      "step": 119344
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.9859523177146912,
      "learning_rate": 0.0002830597711022422,
      "loss": 2.2102,
      "step": 119345
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.9183744788169861,
      "learning_rate": 0.0002830556870683753,
      "loss": 2.2628,
      "step": 119346
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.9407840967178345,
      "learning_rate": 0.00028305160303765864,
      "loss": 2.0711,
      "step": 119347
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.9948046803474426,
      "learning_rate": 0.000283047519010093,
      "loss": 2.3021,
      "step": 119348
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.9399210214614868,
      "learning_rate": 0.0002830434349856791,
      "loss": 2.2841,
      "step": 119349
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.9663605093955994,
      "learning_rate": 0.0002830393509644179,
      "loss": 2.1739,
      "step": 119350
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.9444116353988647,
      "learning_rate": 0.00028303526694631,
      "loss": 1.957,
      "step": 119351
    },
    {
      "epoch": 1.55,
      "grad_norm": 1.1951237916946411,
      "learning_rate": 0.0002830311829313561,
      "loss": 1.9782,
      "step": 119352
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.9626157283782959,
      "learning_rate": 0.000283027098919557,
      "loss": 1.9739,
      "step": 119353
    },
    {
      "epoch": 1.55,
      "grad_norm": 1.1090943813323975,
      "learning_rate": 0.00028302301491091346,
      "loss": 2.1067,
      "step": 119354
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.9198339581489563,
      "learning_rate": 0.00028301893090542626,
      "loss": 2.2477,
      "step": 119355
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.9568629264831543,
      "learning_rate": 0.00028301484690309626,
      "loss": 2.1176,
      "step": 119356
    },
    {
      "epoch": 1.55,
      "grad_norm": 1.0060272216796875,
      "learning_rate": 0.00028301076290392405,
      "loss": 2.0523,
      "step": 119357
    },
    {
      "epoch": 1.55,
      "grad_norm": 1.0426539182662964,
      "learning_rate": 0.0002830066789079105,
      "loss": 1.8736,
      "step": 119358
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.8974462151527405,
      "learning_rate": 0.0002830025949150563,
      "loss": 2.1471,
      "step": 119359
    },
    {
      "epoch": 1.55,
      "grad_norm": 1.0276358127593994,
      "learning_rate": 0.0002829985109253622,
      "loss": 1.9927,
      "step": 119360
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.8923789262771606,
      "learning_rate": 0.000282994426938829,
      "loss": 2.0422,
      "step": 119361
    },
    {
      "epoch": 1.55,
      "grad_norm": 1.13059401512146,
      "learning_rate": 0.0002829903429554576,
      "loss": 2.0324,
      "step": 119362
    },
    {
      "epoch": 1.55,
      "grad_norm": 1.0332790613174438,
      "learning_rate": 0.00028298625897524846,
      "loss": 2.1619,
      "step": 119363
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.9942321181297302,
      "learning_rate": 0.00028298217499820254,
      "loss": 2.0106,
      "step": 119364
    },
    {
      "epoch": 1.55,
      "grad_norm": 1.0026533603668213,
      "learning_rate": 0.0002829780910243206,
      "loss": 1.985,
      "step": 119365
    },
    {
      "epoch": 1.55,
      "grad_norm": 1.0174332857131958,
      "learning_rate": 0.0002829740070536033,
      "loss": 2.1228,
      "step": 119366
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.90055912733078,
      "learning_rate": 0.0002829699230860515,
      "loss": 2.2616,
      "step": 119367
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.9352598786354065,
      "learning_rate": 0.0002829658391216659,
      "loss": 2.0573,
      "step": 119368
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.9841221570968628,
      "learning_rate": 0.0002829617551604472,
      "loss": 2.2375,
      "step": 119369
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.9482835531234741,
      "learning_rate": 0.0002829576712023963,
      "loss": 2.2058,
      "step": 119370
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.8585397005081177,
      "learning_rate": 0.00028295358724751387,
      "loss": 2.2825,
      "step": 119371
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.9183754920959473,
      "learning_rate": 0.0002829495032958007,
      "loss": 2.0574,
      "step": 119372
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.869113564491272,
      "learning_rate": 0.0002829454193472576,
      "loss": 2.1338,
      "step": 119373
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.9247350096702576,
      "learning_rate": 0.0002829413354018852,
      "loss": 1.8678,
      "step": 119374
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.9315815567970276,
      "learning_rate": 0.0002829372514596843,
      "loss": 2.2094,
      "step": 119375
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.8848309516906738,
      "learning_rate": 0.00028293316752065574,
      "loss": 2.1824,
      "step": 119376
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.8643680214881897,
      "learning_rate": 0.0002829290835848002,
      "loss": 2.2463,
      "step": 119377
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.9972128868103027,
      "learning_rate": 0.00028292499965211845,
      "loss": 2.1356,
      "step": 119378
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.9122510552406311,
      "learning_rate": 0.0002829209157226112,
      "loss": 2.1166,
      "step": 119379
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.9061303734779358,
      "learning_rate": 0.00028291683179627945,
      "loss": 1.9731,
      "step": 119380
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.9939156174659729,
      "learning_rate": 0.00028291274787312375,
      "loss": 1.9968,
      "step": 119381
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.939655065536499,
      "learning_rate": 0.0002829086639531448,
      "loss": 1.9474,
      "step": 119382
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.9587340950965881,
      "learning_rate": 0.0002829045800363435,
      "loss": 2.2506,
      "step": 119383
    },
    {
      "epoch": 1.55,
      "grad_norm": 1.006466031074524,
      "learning_rate": 0.0002829004961227205,
      "loss": 2.2159,
      "step": 119384
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.9754590392112732,
      "learning_rate": 0.00028289641221227667,
      "loss": 2.2338,
      "step": 119385
    },
    {
      "epoch": 1.55,
      "grad_norm": 1.0786837339401245,
      "learning_rate": 0.0002828923283050127,
      "loss": 2.0637,
      "step": 119386
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.9208279252052307,
      "learning_rate": 0.0002828882444009295,
      "loss": 1.8617,
      "step": 119387
    },
    {
      "epoch": 1.55,
      "grad_norm": 1.0572381019592285,
      "learning_rate": 0.00028288416050002756,
      "loss": 2.0429,
      "step": 119388
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.8973754644393921,
      "learning_rate": 0.0002828800766023078,
      "loss": 2.1034,
      "step": 119389
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.9169460535049438,
      "learning_rate": 0.00028287599270777094,
      "loss": 2.0156,
      "step": 119390
    },
    {
      "epoch": 1.55,
      "grad_norm": 1.0226343870162964,
      "learning_rate": 0.00028287190881641776,
      "loss": 2.2924,
      "step": 119391
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.9753789901733398,
      "learning_rate": 0.00028286782492824904,
      "loss": 1.9598,
      "step": 119392
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.9072778820991516,
      "learning_rate": 0.0002828637410432656,
      "loss": 2.3225,
      "step": 119393
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.9484780430793762,
      "learning_rate": 0.000282859657161468,
      "loss": 2.2597,
      "step": 119394
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.8849726319313049,
      "learning_rate": 0.0002828555732828571,
      "loss": 2.0578,
      "step": 119395
    },
    {
      "epoch": 1.55,
      "grad_norm": 1.022983431816101,
      "learning_rate": 0.0002828514894074337,
      "loss": 1.9151,
      "step": 119396
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.8390893936157227,
      "learning_rate": 0.00028284740553519855,
      "loss": 2.1249,
      "step": 119397
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.9450886845588684,
      "learning_rate": 0.00028284332166615233,
      "loss": 2.0441,
      "step": 119398
    },
    {
      "epoch": 1.55,
      "grad_norm": 1.058418869972229,
      "learning_rate": 0.00028283923780029606,
      "loss": 2.1764,
      "step": 119399
    },
    {
      "epoch": 1.55,
      "grad_norm": 1.056801438331604,
      "learning_rate": 0.0002828351539376301,
      "loss": 2.1181,
      "step": 119400
    },
    {
      "epoch": 1.55,
      "grad_norm": 1.011925458908081,
      "learning_rate": 0.00028283107007815545,
      "loss": 2.252,
      "step": 119401
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.9500269889831543,
      "learning_rate": 0.0002828269862218728,
      "loss": 2.1055,
      "step": 119402
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.9994008541107178,
      "learning_rate": 0.00028282290236878295,
      "loss": 2.1264,
      "step": 119403
    },
    {
      "epoch": 1.55,
      "grad_norm": 1.095621943473816,
      "learning_rate": 0.0002828188185188867,
      "loss": 2.1879,
      "step": 119404
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.976909875869751,
      "learning_rate": 0.00028281473467218484,
      "loss": 1.9951,
      "step": 119405
    },
    {
      "epoch": 1.55,
      "grad_norm": 1.3621054887771606,
      "learning_rate": 0.0002828106508286779,
      "loss": 2.2292,
      "step": 119406
    },
    {
      "epoch": 1.55,
      "grad_norm": 1.0042011737823486,
      "learning_rate": 0.0002828065669883668,
      "loss": 2.2925,
      "step": 119407
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.9344760179519653,
      "learning_rate": 0.0002828024831512523,
      "loss": 2.178,
      "step": 119408
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.9478563666343689,
      "learning_rate": 0.0002827983993173351,
      "loss": 2.2569,
      "step": 119409
    },
    {
      "epoch": 1.55,
      "grad_norm": 1.127300500869751,
      "learning_rate": 0.00028279431548661606,
      "loss": 2.6116,
      "step": 119410
    },
    {
      "epoch": 1.55,
      "grad_norm": 1.0012906789779663,
      "learning_rate": 0.0002827902316590959,
      "loss": 2.1646,
      "step": 119411
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.9784107208251953,
      "learning_rate": 0.0002827861478347754,
      "loss": 2.02,
      "step": 119412
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.9613118767738342,
      "learning_rate": 0.00028278206401365514,
      "loss": 2.175,
      "step": 119413
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.9359374642372131,
      "learning_rate": 0.0002827779801957361,
      "loss": 1.9764,
      "step": 119414
    },
    {
      "epoch": 1.55,
      "grad_norm": 1.1050211191177368,
      "learning_rate": 0.0002827738963810189,
      "loss": 1.9639,
      "step": 119415
    },
    {
      "epoch": 1.55,
      "grad_norm": 1.0249762535095215,
      "learning_rate": 0.0002827698125695044,
      "loss": 2.1471,
      "step": 119416
    },
    {
      "epoch": 1.55,
      "grad_norm": 1.0713496208190918,
      "learning_rate": 0.0002827657287611934,
      "loss": 2.0092,
      "step": 119417
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.905381977558136,
      "learning_rate": 0.0002827616449560865,
      "loss": 1.9562,
      "step": 119418
    },
    {
      "epoch": 1.55,
      "grad_norm": 1.0044039487838745,
      "learning_rate": 0.00028275756115418455,
      "loss": 2.1555,
      "step": 119419
    },
    {
      "epoch": 1.55,
      "grad_norm": 1.1310784816741943,
      "learning_rate": 0.0002827534773554882,
      "loss": 2.1084,
      "step": 119420
    },
    {
      "epoch": 1.55,
      "grad_norm": 0.9657569527626038,
      "learning_rate": 0.0002827493935599984,
      "loss": 1.9648,
      "step": 119421
    },
    {
      "epoch": 1.55,
      "grad_norm": 1.0510035753250122,
      "learning_rate": 0.0002827453097677158,
      "loss": 2.1419,
      "step": 119422
    },
    {
      "epoch": 1.55,
      "grad_norm": 1.0221049785614014,
      "learning_rate": 0.00028274122597864127,
      "loss": 2.232,
      "step": 119423
    },
    {
      "epoch": 1.55,
      "grad_norm": 1.0207054615020752,
      "learning_rate": 0.0002827371421927753,
      "loss": 2.0752,
      "step": 119424
    },
    {
      "epoch": 1.55,
      "grad_norm": 1.0463145971298218,
      "learning_rate": 0.00028273305841011894,
      "loss": 2.1778,
      "step": 119425
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.8824294805526733,
      "learning_rate": 0.00028272897463067277,
      "loss": 1.9381,
      "step": 119426
    },
    {
      "epoch": 1.56,
      "grad_norm": 1.0497825145721436,
      "learning_rate": 0.00028272489085443757,
      "loss": 2.1139,
      "step": 119427
    },
    {
      "epoch": 1.56,
      "grad_norm": 1.251182198524475,
      "learning_rate": 0.00028272080708141416,
      "loss": 2.2202,
      "step": 119428
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.9182049632072449,
      "learning_rate": 0.00028271672331160345,
      "loss": 2.0857,
      "step": 119429
    },
    {
      "epoch": 1.56,
      "grad_norm": 1.3303114175796509,
      "learning_rate": 0.0002827126395450059,
      "loss": 2.072,
      "step": 119430
    },
    {
      "epoch": 1.56,
      "grad_norm": 1.0180718898773193,
      "learning_rate": 0.00028270855578162233,
      "loss": 2.1005,
      "step": 119431
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.9863370656967163,
      "learning_rate": 0.0002827044720214536,
      "loss": 2.096,
      "step": 119432
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.8781343698501587,
      "learning_rate": 0.00028270038826450045,
      "loss": 2.0772,
      "step": 119433
    },
    {
      "epoch": 1.56,
      "grad_norm": 1.0018627643585205,
      "learning_rate": 0.0002826963045107636,
      "loss": 2.2847,
      "step": 119434
    },
    {
      "epoch": 1.56,
      "grad_norm": 1.0209392309188843,
      "learning_rate": 0.00028269222076024393,
      "loss": 2.1988,
      "step": 119435
    },
    {
      "epoch": 1.56,
      "grad_norm": 1.012206792831421,
      "learning_rate": 0.0002826881370129421,
      "loss": 2.1353,
      "step": 119436
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.9256621599197388,
      "learning_rate": 0.0002826840532688588,
      "loss": 1.9883,
      "step": 119437
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.9407357573509216,
      "learning_rate": 0.0002826799695279948,
      "loss": 2.1221,
      "step": 119438
    },
    {
      "epoch": 1.56,
      "grad_norm": 1.0480355024337769,
      "learning_rate": 0.000282675885790351,
      "loss": 2.4482,
      "step": 119439
    },
    {
      "epoch": 1.56,
      "grad_norm": 1.006101369857788,
      "learning_rate": 0.00028267180205592804,
      "loss": 2.2023,
      "step": 119440
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.9153993129730225,
      "learning_rate": 0.0002826677183247267,
      "loss": 2.0284,
      "step": 119441
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.9950129985809326,
      "learning_rate": 0.00028266363459674787,
      "loss": 2.1421,
      "step": 119442
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.963295042514801,
      "learning_rate": 0.00028265955087199217,
      "loss": 2.049,
      "step": 119443
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.9495041966438293,
      "learning_rate": 0.00028265546715046033,
      "loss": 2.2324,
      "step": 119444
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.9658874273300171,
      "learning_rate": 0.0002826513834321532,
      "loss": 2.1644,
      "step": 119445
    },
    {
      "epoch": 1.56,
      "grad_norm": 1.031930685043335,
      "learning_rate": 0.00028264729971707144,
      "loss": 2.124,
      "step": 119446
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.9660751819610596,
      "learning_rate": 0.00028264321600521595,
      "loss": 2.1953,
      "step": 119447
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.8887017965316772,
      "learning_rate": 0.0002826391322965874,
      "loss": 2.3248,
      "step": 119448
    },
    {
      "epoch": 1.56,
      "grad_norm": 1.0852701663970947,
      "learning_rate": 0.0002826350485911865,
      "loss": 2.2051,
      "step": 119449
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.8625114560127258,
      "learning_rate": 0.0002826309648890142,
      "loss": 2.1784,
      "step": 119450
    },
    {
      "epoch": 1.56,
      "grad_norm": 1.4171816110610962,
      "learning_rate": 0.00028262688119007107,
      "loss": 1.9133,
      "step": 119451
    },
    {
      "epoch": 1.56,
      "grad_norm": 1.1215273141860962,
      "learning_rate": 0.0002826227974943579,
      "loss": 2.1885,
      "step": 119452
    },
    {
      "epoch": 1.56,
      "grad_norm": 1.0381361246109009,
      "learning_rate": 0.00028261871380187546,
      "loss": 1.9727,
      "step": 119453
    },
    {
      "epoch": 1.56,
      "grad_norm": 1.0119038820266724,
      "learning_rate": 0.0002826146301126246,
      "loss": 1.9987,
      "step": 119454
    },
    {
      "epoch": 1.56,
      "grad_norm": 1.1118186712265015,
      "learning_rate": 0.000282610546426606,
      "loss": 1.9188,
      "step": 119455
    },
    {
      "epoch": 1.56,
      "grad_norm": 1.0929865837097168,
      "learning_rate": 0.00028260646274382034,
      "loss": 2.0741,
      "step": 119456
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.9632867574691772,
      "learning_rate": 0.0002826023790642686,
      "loss": 2.242,
      "step": 119457
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.9563357830047607,
      "learning_rate": 0.0002825982953879513,
      "loss": 2.2218,
      "step": 119458
    },
    {
      "epoch": 1.56,
      "grad_norm": 1.0966532230377197,
      "learning_rate": 0.0002825942117148694,
      "loss": 2.1304,
      "step": 119459
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.8467785120010376,
      "learning_rate": 0.0002825901280450235,
      "loss": 2.111,
      "step": 119460
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.8499298095703125,
      "learning_rate": 0.0002825860443784144,
      "loss": 1.8782,
      "step": 119461
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.9257606267929077,
      "learning_rate": 0.0002825819607150429,
      "loss": 2.0209,
      "step": 119462
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.8767166137695312,
      "learning_rate": 0.00028257787705490977,
      "loss": 1.8438,
      "step": 119463
    },
    {
      "epoch": 1.56,
      "grad_norm": 1.0146600008010864,
      "learning_rate": 0.0002825737933980157,
      "loss": 2.1046,
      "step": 119464
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.9522405862808228,
      "learning_rate": 0.0002825697097443615,
      "loss": 2.2818,
      "step": 119465
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.9997751116752625,
      "learning_rate": 0.00028256562609394804,
      "loss": 2.4684,
      "step": 119466
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.930474042892456,
      "learning_rate": 0.00028256154244677583,
      "loss": 2.1065,
      "step": 119467
    },
    {
      "epoch": 1.56,
      "grad_norm": 1.0576107501983643,
      "learning_rate": 0.00028255745880284577,
      "loss": 2.1958,
      "step": 119468
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.9907059073448181,
      "learning_rate": 0.0002825533751621586,
      "loss": 2.1455,
      "step": 119469
    },
    {
      "epoch": 1.56,
      "grad_norm": 1.0424537658691406,
      "learning_rate": 0.0002825492915247151,
      "loss": 2.1566,
      "step": 119470
    },
    {
      "epoch": 1.56,
      "grad_norm": 1.0253676176071167,
      "learning_rate": 0.000282545207890516,
      "loss": 2.0242,
      "step": 119471
    },
    {
      "epoch": 1.56,
      "grad_norm": 1.020249843597412,
      "learning_rate": 0.0002825411242595622,
      "loss": 2.2764,
      "step": 119472
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.8458431959152222,
      "learning_rate": 0.00028253704063185424,
      "loss": 2.1909,
      "step": 119473
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.9881331920623779,
      "learning_rate": 0.0002825329570073929,
      "loss": 2.1214,
      "step": 119474
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.946571409702301,
      "learning_rate": 0.0002825288733861791,
      "loss": 2.042,
      "step": 119475
    },
    {
      "epoch": 1.56,
      "grad_norm": 1.079888105392456,
      "learning_rate": 0.0002825247897682135,
      "loss": 2.2749,
      "step": 119476
    },
    {
      "epoch": 1.56,
      "grad_norm": 1.009774088859558,
      "learning_rate": 0.00028252070615349687,
      "loss": 2.3885,
      "step": 119477
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.9331946969032288,
      "learning_rate": 0.00028251662254203004,
      "loss": 2.2845,
      "step": 119478
    },
    {
      "epoch": 1.56,
      "grad_norm": 1.0400364398956299,
      "learning_rate": 0.0002825125389338136,
      "loss": 2.0418,
      "step": 119479
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.9599422812461853,
      "learning_rate": 0.0002825084553288484,
      "loss": 2.1866,
      "step": 119480
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.9443202018737793,
      "learning_rate": 0.0002825043717271353,
      "loss": 2.003,
      "step": 119481
    },
    {
      "epoch": 1.56,
      "grad_norm": 1.0032931566238403,
      "learning_rate": 0.00028250028812867487,
      "loss": 2.1813,
      "step": 119482
    },
    {
      "epoch": 1.56,
      "grad_norm": 1.031435489654541,
      "learning_rate": 0.000282496204533468,
      "loss": 2.1876,
      "step": 119483
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.9457229971885681,
      "learning_rate": 0.0002824921209415155,
      "loss": 2.0577,
      "step": 119484
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.8747765421867371,
      "learning_rate": 0.00028248803735281795,
      "loss": 2.1437,
      "step": 119485
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.9059230089187622,
      "learning_rate": 0.0002824839537673762,
      "loss": 2.3779,
      "step": 119486
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.9591666460037231,
      "learning_rate": 0.000282479870185191,
      "loss": 2.1055,
      "step": 119487
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.8802742958068848,
      "learning_rate": 0.00028247578660626317,
      "loss": 2.2175,
      "step": 119488
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.996101438999176,
      "learning_rate": 0.0002824717030305934,
      "loss": 2.0784,
      "step": 119489
    },
    {
      "epoch": 1.56,
      "grad_norm": 1.0640697479248047,
      "learning_rate": 0.0002824676194581826,
      "loss": 2.0983,
      "step": 119490
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.9797611832618713,
      "learning_rate": 0.0002824635358890313,
      "loss": 2.2018,
      "step": 119491
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.9659671187400818,
      "learning_rate": 0.0002824594523231403,
      "loss": 2.181,
      "step": 119492
    },
    {
      "epoch": 1.56,
      "grad_norm": 1.0284463167190552,
      "learning_rate": 0.00028245536876051044,
      "loss": 2.1044,
      "step": 119493
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.8768681883811951,
      "learning_rate": 0.00028245128520114246,
      "loss": 2.1274,
      "step": 119494
    },
    {
      "epoch": 1.56,
      "grad_norm": 1.0492209196090698,
      "learning_rate": 0.0002824472016450371,
      "loss": 2.2386,
      "step": 119495
    },
    {
      "epoch": 1.56,
      "grad_norm": 1.0623236894607544,
      "learning_rate": 0.0002824431180921953,
      "loss": 2.1845,
      "step": 119496
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.9681961536407471,
      "learning_rate": 0.0002824390345426175,
      "loss": 2.084,
      "step": 119497
    },
    {
      "epoch": 1.56,
      "grad_norm": 1.030743956565857,
      "learning_rate": 0.0002824349509963046,
      "loss": 2.2312,
      "step": 119498
    },
    {
      "epoch": 1.56,
      "grad_norm": 1.0180763006210327,
      "learning_rate": 0.00028243086745325746,
      "loss": 2.1683,
      "step": 119499
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.8477338552474976,
      "learning_rate": 0.0002824267839134767,
      "loss": 2.0873,
      "step": 119500
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.8938831090927124,
      "learning_rate": 0.00028242270037696313,
      "loss": 1.9898,
      "step": 119501
    },
    {
      "epoch": 1.56,
      "grad_norm": 1.0624009370803833,
      "learning_rate": 0.00028241861684371764,
      "loss": 2.3698,
      "step": 119502
    },
    {
      "epoch": 1.56,
      "grad_norm": 1.066063404083252,
      "learning_rate": 0.0002824145333137407,
      "loss": 2.2381,
      "step": 119503
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.9034425020217896,
      "learning_rate": 0.0002824104497870333,
      "loss": 2.2339,
      "step": 119504
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.9352882504463196,
      "learning_rate": 0.0002824063662635961,
      "loss": 2.1338,
      "step": 119505
    },
    {
      "epoch": 1.56,
      "grad_norm": 1.1502125263214111,
      "learning_rate": 0.0002824022827434298,
      "loss": 2.3395,
      "step": 119506
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.9673562049865723,
      "learning_rate": 0.0002823981992265354,
      "loss": 2.1308,
      "step": 119507
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.8979051113128662,
      "learning_rate": 0.0002823941157129134,
      "loss": 2.0227,
      "step": 119508
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.9454435706138611,
      "learning_rate": 0.0002823900322025648,
      "loss": 2.0304,
      "step": 119509
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.9488157629966736,
      "learning_rate": 0.00028238594869549013,
      "loss": 2.0582,
      "step": 119510
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.8823529481887817,
      "learning_rate": 0.00028238186519169027,
      "loss": 2.1605,
      "step": 119511
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.903803288936615,
      "learning_rate": 0.00028237778169116595,
      "loss": 2.0137,
      "step": 119512
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.9515352845191956,
      "learning_rate": 0.00028237369819391787,
      "loss": 2.0956,
      "step": 119513
    },
    {
      "epoch": 1.56,
      "grad_norm": 1.0465149879455566,
      "learning_rate": 0.0002823696146999469,
      "loss": 2.0666,
      "step": 119514
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.9039938449859619,
      "learning_rate": 0.00028236553120925386,
      "loss": 1.9762,
      "step": 119515
    },
    {
      "epoch": 1.56,
      "grad_norm": 1.049291968345642,
      "learning_rate": 0.0002823614477218393,
      "loss": 2.1253,
      "step": 119516
    },
    {
      "epoch": 1.56,
      "grad_norm": 1.1313791275024414,
      "learning_rate": 0.0002823573642377041,
      "loss": 2.0239,
      "step": 119517
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.894303023815155,
      "learning_rate": 0.0002823532807568489,
      "loss": 2.1885,
      "step": 119518
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.9855386018753052,
      "learning_rate": 0.0002823491972792746,
      "loss": 2.3488,
      "step": 119519
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.9377725720405579,
      "learning_rate": 0.00028234511380498197,
      "loss": 1.9564,
      "step": 119520
    },
    {
      "epoch": 1.56,
      "grad_norm": 1.006661057472229,
      "learning_rate": 0.00028234103033397185,
      "loss": 2.0614,
      "step": 119521
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.8655240535736084,
      "learning_rate": 0.00028233694686624465,
      "loss": 2.2314,
      "step": 119522
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.9561329483985901,
      "learning_rate": 0.0002823328634018014,
      "loss": 2.2166,
      "step": 119523
    },
    {
      "epoch": 1.56,
      "grad_norm": 1.0125001668930054,
      "learning_rate": 0.0002823287799406428,
      "loss": 2.2256,
      "step": 119524
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.964067280292511,
      "learning_rate": 0.0002823246964827696,
      "loss": 2.1796,
      "step": 119525
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.8425572514533997,
      "learning_rate": 0.0002823206130281826,
      "loss": 2.0924,
      "step": 119526
    },
    {
      "epoch": 1.56,
      "grad_norm": 1.0772225856781006,
      "learning_rate": 0.00028231652957688263,
      "loss": 1.9901,
      "step": 119527
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.9914445877075195,
      "learning_rate": 0.00028231244612887023,
      "loss": 2.3365,
      "step": 119528
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.86355060338974,
      "learning_rate": 0.0002823083626841463,
      "loss": 2.2562,
      "step": 119529
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.9946528673171997,
      "learning_rate": 0.00028230427924271157,
      "loss": 2.3779,
      "step": 119530
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.9970692992210388,
      "learning_rate": 0.0002823001958045668,
      "loss": 2.1368,
      "step": 119531
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.9900494813919067,
      "learning_rate": 0.0002822961123697128,
      "loss": 2.0482,
      "step": 119532
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.949778139591217,
      "learning_rate": 0.0002822920289381503,
      "loss": 2.2012,
      "step": 119533
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.9471365809440613,
      "learning_rate": 0.00028228794550987996,
      "loss": 2.1,
      "step": 119534
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.9059712290763855,
      "learning_rate": 0.00028228386208490274,
      "loss": 2.1892,
      "step": 119535
    },
    {
      "epoch": 1.56,
      "grad_norm": 1.093316674232483,
      "learning_rate": 0.0002822797786632192,
      "loss": 2.2492,
      "step": 119536
    },
    {
      "epoch": 1.56,
      "grad_norm": 1.0622260570526123,
      "learning_rate": 0.0002822756952448302,
      "loss": 2.1996,
      "step": 119537
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.89216548204422,
      "learning_rate": 0.00028227161182973646,
      "loss": 2.2382,
      "step": 119538
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.8456215858459473,
      "learning_rate": 0.00028226752841793885,
      "loss": 2.096,
      "step": 119539
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.9952704310417175,
      "learning_rate": 0.00028226344500943793,
      "loss": 2.1037,
      "step": 119540
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.970954418182373,
      "learning_rate": 0.0002822593616042346,
      "loss": 2.2071,
      "step": 119541
    },
    {
      "epoch": 1.56,
      "grad_norm": 1.006799340248108,
      "learning_rate": 0.0002822552782023297,
      "loss": 2.0966,
      "step": 119542
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.9175474643707275,
      "learning_rate": 0.00028225119480372374,
      "loss": 2.0568,
      "step": 119543
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.8532686829566956,
      "learning_rate": 0.00028224711140841766,
      "loss": 2.2173,
      "step": 119544
    },
    {
      "epoch": 1.56,
      "grad_norm": 1.0176880359649658,
      "learning_rate": 0.0002822430280164122,
      "loss": 2.222,
      "step": 119545
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.9798826575279236,
      "learning_rate": 0.0002822389446277081,
      "loss": 2.1399,
      "step": 119546
    },
    {
      "epoch": 1.56,
      "grad_norm": 1.1950923204421997,
      "learning_rate": 0.0002822348612423061,
      "loss": 2.0907,
      "step": 119547
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.9135346412658691,
      "learning_rate": 0.0002822307778602069,
      "loss": 2.1488,
      "step": 119548
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.8841617107391357,
      "learning_rate": 0.0002822266944814115,
      "loss": 2.2183,
      "step": 119549
    },
    {
      "epoch": 1.56,
      "grad_norm": 1.1460281610488892,
      "learning_rate": 0.00028222261110592037,
      "loss": 2.0659,
      "step": 119550
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.9892009496688843,
      "learning_rate": 0.00028221852773373447,
      "loss": 2.1174,
      "step": 119551
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.9290370941162109,
      "learning_rate": 0.0002822144443648544,
      "loss": 2.162,
      "step": 119552
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.8649468421936035,
      "learning_rate": 0.00028221036099928106,
      "loss": 2.191,
      "step": 119553
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.951776385307312,
      "learning_rate": 0.00028220627763701506,
      "loss": 2.059,
      "step": 119554
    },
    {
      "epoch": 1.56,
      "grad_norm": 1.0281834602355957,
      "learning_rate": 0.00028220219427805735,
      "loss": 2.4441,
      "step": 119555
    },
    {
      "epoch": 1.56,
      "grad_norm": 1.0629037618637085,
      "learning_rate": 0.00028219811092240857,
      "loss": 2.2521,
      "step": 119556
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.9614987969398499,
      "learning_rate": 0.00028219402757006953,
      "loss": 2.3457,
      "step": 119557
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.953264057636261,
      "learning_rate": 0.0002821899442210409,
      "loss": 2.0169,
      "step": 119558
    },
    {
      "epoch": 1.56,
      "grad_norm": 1.029978632926941,
      "learning_rate": 0.00028218586087532346,
      "loss": 2.1676,
      "step": 119559
    },
    {
      "epoch": 1.56,
      "grad_norm": 1.0534356832504272,
      "learning_rate": 0.00028218177753291804,
      "loss": 2.1187,
      "step": 119560
    },
    {
      "epoch": 1.56,
      "grad_norm": 1.1047981977462769,
      "learning_rate": 0.0002821776941938253,
      "loss": 2.2529,
      "step": 119561
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.9860401153564453,
      "learning_rate": 0.00028217361085804615,
      "loss": 2.1729,
      "step": 119562
    },
    {
      "epoch": 1.56,
      "grad_norm": 1.0945851802825928,
      "learning_rate": 0.00028216952752558135,
      "loss": 2.2435,
      "step": 119563
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.9684368968009949,
      "learning_rate": 0.00028216544419643145,
      "loss": 2.257,
      "step": 119564
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.957103431224823,
      "learning_rate": 0.0002821613608705973,
      "loss": 2.1956,
      "step": 119565
    },
    {
      "epoch": 1.56,
      "grad_norm": 1.1603851318359375,
      "learning_rate": 0.00028215727754807974,
      "loss": 2.194,
      "step": 119566
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.9739766716957092,
      "learning_rate": 0.00028215319422887945,
      "loss": 2.3202,
      "step": 119567
    },
    {
      "epoch": 1.56,
      "grad_norm": 1.134310245513916,
      "learning_rate": 0.0002821491109129972,
      "loss": 2.1828,
      "step": 119568
    },
    {
      "epoch": 1.56,
      "grad_norm": 1.115562915802002,
      "learning_rate": 0.00028214502760043396,
      "loss": 2.1193,
      "step": 119569
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.9343336820602417,
      "learning_rate": 0.0002821409442911901,
      "loss": 2.0455,
      "step": 119570
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.9585756659507751,
      "learning_rate": 0.0002821368609852666,
      "loss": 2.1244,
      "step": 119571
    },
    {
      "epoch": 1.56,
      "grad_norm": 1.0813623666763306,
      "learning_rate": 0.00028213277768266424,
      "loss": 2.0559,
      "step": 119572
    },
    {
      "epoch": 1.56,
      "grad_norm": 1.0684807300567627,
      "learning_rate": 0.00028212869438338367,
      "loss": 2.2279,
      "step": 119573
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.9135540723800659,
      "learning_rate": 0.00028212461108742575,
      "loss": 2.1623,
      "step": 119574
    },
    {
      "epoch": 1.56,
      "grad_norm": 1.1419388055801392,
      "learning_rate": 0.0002821205277947912,
      "loss": 2.0894,
      "step": 119575
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.9822698831558228,
      "learning_rate": 0.0002821164445054809,
      "loss": 2.0503,
      "step": 119576
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.8999675512313843,
      "learning_rate": 0.0002821123612194953,
      "loss": 2.0633,
      "step": 119577
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.9650641679763794,
      "learning_rate": 0.00028210827793683543,
      "loss": 2.088,
      "step": 119578
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.9130963087081909,
      "learning_rate": 0.00028210419465750196,
      "loss": 2.1974,
      "step": 119579
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.866736650466919,
      "learning_rate": 0.00028210011138149563,
      "loss": 1.8629,
      "step": 119580
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.9818853735923767,
      "learning_rate": 0.0002820960281088172,
      "loss": 2.3874,
      "step": 119581
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.9626981616020203,
      "learning_rate": 0.00028209194483946767,
      "loss": 2.1164,
      "step": 119582
    },
    {
      "epoch": 1.56,
      "grad_norm": 1.0820825099945068,
      "learning_rate": 0.0002820878615734474,
      "loss": 2.0104,
      "step": 119583
    },
    {
      "epoch": 1.56,
      "grad_norm": 1.0726687908172607,
      "learning_rate": 0.00028208377831075734,
      "loss": 2.0704,
      "step": 119584
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.9547969698905945,
      "learning_rate": 0.0002820796950513983,
      "loss": 1.9862,
      "step": 119585
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.9679813981056213,
      "learning_rate": 0.00028207561179537094,
      "loss": 2.3361,
      "step": 119586
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.8677491545677185,
      "learning_rate": 0.000282071528542676,
      "loss": 2.0598,
      "step": 119587
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.9470974206924438,
      "learning_rate": 0.00028206744529331453,
      "loss": 2.2416,
      "step": 119588
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.936693012714386,
      "learning_rate": 0.0002820633620472869,
      "loss": 2.0619,
      "step": 119589
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.868053674697876,
      "learning_rate": 0.00028205927880459397,
      "loss": 2.0864,
      "step": 119590
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.9045796990394592,
      "learning_rate": 0.0002820551955652366,
      "loss": 2.2424,
      "step": 119591
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.8386865258216858,
      "learning_rate": 0.00028205111232921555,
      "loss": 2.276,
      "step": 119592
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.936089813709259,
      "learning_rate": 0.0002820470290965315,
      "loss": 2.1962,
      "step": 119593
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.9146895408630371,
      "learning_rate": 0.0002820429458671854,
      "loss": 1.9881,
      "step": 119594
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.960647702217102,
      "learning_rate": 0.0002820388626411777,
      "loss": 2.1541,
      "step": 119595
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.9837294220924377,
      "learning_rate": 0.0002820347794185093,
      "loss": 2.1256,
      "step": 119596
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.9294920563697815,
      "learning_rate": 0.000282030696199181,
      "loss": 2.1799,
      "step": 119597
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.8756407499313354,
      "learning_rate": 0.00028202661298319356,
      "loss": 2.1005,
      "step": 119598
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.8991968631744385,
      "learning_rate": 0.0002820225297705476,
      "loss": 2.2287,
      "step": 119599
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.9138907194137573,
      "learning_rate": 0.00028201844656124424,
      "loss": 2.2101,
      "step": 119600
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.8831866383552551,
      "learning_rate": 0.0002820143633552838,
      "loss": 1.9925,
      "step": 119601
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.9201834201812744,
      "learning_rate": 0.00028201028015266725,
      "loss": 2.0252,
      "step": 119602
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.9717152118682861,
      "learning_rate": 0.00028200619695339533,
      "loss": 2.0633,
      "step": 119603
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.8934176564216614,
      "learning_rate": 0.00028200211375746883,
      "loss": 2.1995,
      "step": 119604
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.9833543300628662,
      "learning_rate": 0.0002819980305648884,
      "loss": 2.2607,
      "step": 119605
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.9073004722595215,
      "learning_rate": 0.00028199394737565505,
      "loss": 2.054,
      "step": 119606
    },
    {
      "epoch": 1.56,
      "grad_norm": 1.013362169265747,
      "learning_rate": 0.00028198986418976924,
      "loss": 2.2989,
      "step": 119607
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.9448006749153137,
      "learning_rate": 0.0002819857810072319,
      "loss": 2.1463,
      "step": 119608
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.9565876722335815,
      "learning_rate": 0.00028198169782804366,
      "loss": 2.1912,
      "step": 119609
    },
    {
      "epoch": 1.56,
      "grad_norm": 1.1496295928955078,
      "learning_rate": 0.00028197761465220537,
      "loss": 2.2219,
      "step": 119610
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.9357579350471497,
      "learning_rate": 0.0002819735314797178,
      "loss": 2.1787,
      "step": 119611
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.9761962890625,
      "learning_rate": 0.0002819694483105818,
      "loss": 1.9101,
      "step": 119612
    },
    {
      "epoch": 1.56,
      "grad_norm": 1.123024344444275,
      "learning_rate": 0.00028196536514479793,
      "loss": 2.0719,
      "step": 119613
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.9292457699775696,
      "learning_rate": 0.00028196128198236704,
      "loss": 1.9078,
      "step": 119614
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.9326452016830444,
      "learning_rate": 0.00028195719882328986,
      "loss": 2.1129,
      "step": 119615
    },
    {
      "epoch": 1.56,
      "grad_norm": 1.2209241390228271,
      "learning_rate": 0.0002819531156675672,
      "loss": 2.0246,
      "step": 119616
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.9934414625167847,
      "learning_rate": 0.00028194903251519977,
      "loss": 2.083,
      "step": 119617
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.9517078995704651,
      "learning_rate": 0.00028194494936618843,
      "loss": 2.2005,
      "step": 119618
    },
    {
      "epoch": 1.56,
      "grad_norm": 1.0673038959503174,
      "learning_rate": 0.0002819408662205339,
      "loss": 2.1011,
      "step": 119619
    },
    {
      "epoch": 1.56,
      "grad_norm": 1.0321332216262817,
      "learning_rate": 0.0002819367830782368,
      "loss": 2.0495,
      "step": 119620
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.904484748840332,
      "learning_rate": 0.000281932699939298,
      "loss": 2.0772,
      "step": 119621
    },
    {
      "epoch": 1.56,
      "grad_norm": 1.095809817314148,
      "learning_rate": 0.00028192861680371826,
      "loss": 2.0622,
      "step": 119622
    },
    {
      "epoch": 1.56,
      "grad_norm": 1.2826447486877441,
      "learning_rate": 0.00028192453367149834,
      "loss": 2.1649,
      "step": 119623
    },
    {
      "epoch": 1.56,
      "grad_norm": 1.0233914852142334,
      "learning_rate": 0.000281920450542639,
      "loss": 2.1557,
      "step": 119624
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.9497944116592407,
      "learning_rate": 0.000281916367417141,
      "loss": 2.0906,
      "step": 119625
    },
    {
      "epoch": 1.56,
      "grad_norm": 1.3496845960617065,
      "learning_rate": 0.000281912284295005,
      "loss": 2.0128,
      "step": 119626
    },
    {
      "epoch": 1.56,
      "grad_norm": 1.300102710723877,
      "learning_rate": 0.000281908201176232,
      "loss": 2.0895,
      "step": 119627
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.9040122032165527,
      "learning_rate": 0.0002819041180608225,
      "loss": 2.2418,
      "step": 119628
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.8860180974006653,
      "learning_rate": 0.0002819000349487774,
      "loss": 2.2368,
      "step": 119629
    },
    {
      "epoch": 1.56,
      "grad_norm": 1.093989372253418,
      "learning_rate": 0.00028189595184009746,
      "loss": 2.0016,
      "step": 119630
    },
    {
      "epoch": 1.56,
      "grad_norm": 1.2773916721343994,
      "learning_rate": 0.0002818918687347833,
      "loss": 2.1246,
      "step": 119631
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.9455814361572266,
      "learning_rate": 0.0002818877856328359,
      "loss": 2.1766,
      "step": 119632
    },
    {
      "epoch": 1.56,
      "grad_norm": 1.0223197937011719,
      "learning_rate": 0.0002818837025342558,
      "loss": 2.017,
      "step": 119633
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.9863348007202148,
      "learning_rate": 0.00028187961943904397,
      "loss": 2.2025,
      "step": 119634
    },
    {
      "epoch": 1.56,
      "grad_norm": 1.1875064373016357,
      "learning_rate": 0.000281875536347201,
      "loss": 2.3089,
      "step": 119635
    },
    {
      "epoch": 1.56,
      "grad_norm": 1.0814952850341797,
      "learning_rate": 0.00028187145325872775,
      "loss": 2.3614,
      "step": 119636
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.917456328868866,
      "learning_rate": 0.0002818673701736248,
      "loss": 2.213,
      "step": 119637
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.9989767074584961,
      "learning_rate": 0.0002818632870918932,
      "loss": 2.1961,
      "step": 119638
    },
    {
      "epoch": 1.56,
      "grad_norm": 1.3542755842208862,
      "learning_rate": 0.00028185920401353347,
      "loss": 2.1064,
      "step": 119639
    },
    {
      "epoch": 1.56,
      "grad_norm": 1.06930673122406,
      "learning_rate": 0.00028185512093854647,
      "loss": 2.2629,
      "step": 119640
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.839197039604187,
      "learning_rate": 0.000281851037866933,
      "loss": 2.1934,
      "step": 119641
    },
    {
      "epoch": 1.56,
      "grad_norm": 1.48084557056427,
      "learning_rate": 0.0002818469547986937,
      "loss": 2.113,
      "step": 119642
    },
    {
      "epoch": 1.56,
      "grad_norm": 1.225136637687683,
      "learning_rate": 0.0002818428717338295,
      "loss": 2.0773,
      "step": 119643
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.9569174647331238,
      "learning_rate": 0.00028183878867234094,
      "loss": 2.0791,
      "step": 119644
    },
    {
      "epoch": 1.56,
      "grad_norm": 1.0001494884490967,
      "learning_rate": 0.0002818347056142289,
      "loss": 2.117,
      "step": 119645
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.9686940908432007,
      "learning_rate": 0.0002818306225594942,
      "loss": 2.3086,
      "step": 119646
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.9526504278182983,
      "learning_rate": 0.00028182653950813743,
      "loss": 2.2144,
      "step": 119647
    },
    {
      "epoch": 1.56,
      "grad_norm": 1.1112149953842163,
      "learning_rate": 0.0002818224564601596,
      "loss": 2.2473,
      "step": 119648
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.964993953704834,
      "learning_rate": 0.00028181837341556125,
      "loss": 2.316,
      "step": 119649
    },
    {
      "epoch": 1.56,
      "grad_norm": 1.0006464719772339,
      "learning_rate": 0.0002818142903743432,
      "loss": 2.3497,
      "step": 119650
    },
    {
      "epoch": 1.56,
      "grad_norm": 1.0258846282958984,
      "learning_rate": 0.00028181020733650616,
      "loss": 1.9859,
      "step": 119651
    },
    {
      "epoch": 1.56,
      "grad_norm": 1.3628545999526978,
      "learning_rate": 0.00028180612430205097,
      "loss": 1.85,
      "step": 119652
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.9243756532669067,
      "learning_rate": 0.0002818020412709784,
      "loss": 2.1214,
      "step": 119653
    },
    {
      "epoch": 1.56,
      "grad_norm": 1.0361385345458984,
      "learning_rate": 0.0002817979582432891,
      "loss": 2.0384,
      "step": 119654
    },
    {
      "epoch": 1.56,
      "grad_norm": 1.0752530097961426,
      "learning_rate": 0.0002817938752189841,
      "loss": 2.3134,
      "step": 119655
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.9029817581176758,
      "learning_rate": 0.0002817897921980638,
      "loss": 2.3394,
      "step": 119656
    },
    {
      "epoch": 1.56,
      "grad_norm": 1.092371940612793,
      "learning_rate": 0.00028178570918052913,
      "loss": 2.3304,
      "step": 119657
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.9235181212425232,
      "learning_rate": 0.00028178162616638084,
      "loss": 2.2248,
      "step": 119658
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.931402862071991,
      "learning_rate": 0.0002817775431556197,
      "loss": 2.1759,
      "step": 119659
    },
    {
      "epoch": 1.56,
      "grad_norm": 1.0362887382507324,
      "learning_rate": 0.0002817734601482464,
      "loss": 1.9956,
      "step": 119660
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.8651677966117859,
      "learning_rate": 0.00028176937714426197,
      "loss": 2.0024,
      "step": 119661
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.8943652510643005,
      "learning_rate": 0.0002817652941436668,
      "loss": 2.2178,
      "step": 119662
    },
    {
      "epoch": 1.56,
      "grad_norm": 1.0184136629104614,
      "learning_rate": 0.00028176121114646175,
      "loss": 2.3941,
      "step": 119663
    },
    {
      "epoch": 1.56,
      "grad_norm": 1.1212540864944458,
      "learning_rate": 0.00028175712815264767,
      "loss": 2.3924,
      "step": 119664
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.9323723316192627,
      "learning_rate": 0.00028175304516222534,
      "loss": 1.9523,
      "step": 119665
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.9491364359855652,
      "learning_rate": 0.00028174896217519543,
      "loss": 2.0812,
      "step": 119666
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.9297302961349487,
      "learning_rate": 0.00028174487919155886,
      "loss": 2.0071,
      "step": 119667
    },
    {
      "epoch": 1.56,
      "grad_norm": 1.0430843830108643,
      "learning_rate": 0.0002817407962113161,
      "loss": 1.9317,
      "step": 119668
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.939386248588562,
      "learning_rate": 0.00028173671323446806,
      "loss": 2.1376,
      "step": 119669
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.9352654218673706,
      "learning_rate": 0.00028173263026101556,
      "loss": 1.9938,
      "step": 119670
    },
    {
      "epoch": 1.56,
      "grad_norm": 1.0137939453125,
      "learning_rate": 0.0002817285472909593,
      "loss": 2.0637,
      "step": 119671
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.9511129856109619,
      "learning_rate": 0.00028172446432430005,
      "loss": 1.9958,
      "step": 119672
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.9561079740524292,
      "learning_rate": 0.0002817203813610387,
      "loss": 1.8858,
      "step": 119673
    },
    {
      "epoch": 1.56,
      "grad_norm": 1.1312264204025269,
      "learning_rate": 0.00028171629840117573,
      "loss": 2.0318,
      "step": 119674
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.9923681616783142,
      "learning_rate": 0.000281712215444712,
      "loss": 2.2536,
      "step": 119675
    },
    {
      "epoch": 1.56,
      "grad_norm": 1.0380181074142456,
      "learning_rate": 0.0002817081324916484,
      "loss": 2.3144,
      "step": 119676
    },
    {
      "epoch": 1.56,
      "grad_norm": 1.141299843788147,
      "learning_rate": 0.00028170404954198556,
      "loss": 2.1484,
      "step": 119677
    },
    {
      "epoch": 1.56,
      "grad_norm": 1.1199530363082886,
      "learning_rate": 0.0002816999665957243,
      "loss": 2.1043,
      "step": 119678
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.9391239881515503,
      "learning_rate": 0.0002816958836528655,
      "loss": 2.1093,
      "step": 119679
    },
    {
      "epoch": 1.56,
      "grad_norm": 1.026339054107666,
      "learning_rate": 0.0002816918007134096,
      "loss": 2.3082,
      "step": 119680
    },
    {
      "epoch": 1.56,
      "grad_norm": 1.4451391696929932,
      "learning_rate": 0.0002816877177773576,
      "loss": 1.8835,
      "step": 119681
    },
    {
      "epoch": 1.56,
      "grad_norm": 1.1832170486450195,
      "learning_rate": 0.00028168363484471013,
      "loss": 2.3207,
      "step": 119682
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.9284540414810181,
      "learning_rate": 0.0002816795519154681,
      "loss": 2.2114,
      "step": 119683
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.9811264872550964,
      "learning_rate": 0.0002816754689896321,
      "loss": 1.9856,
      "step": 119684
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.9588531851768494,
      "learning_rate": 0.00028167138606720316,
      "loss": 2.0128,
      "step": 119685
    },
    {
      "epoch": 1.56,
      "grad_norm": 1.08198082447052,
      "learning_rate": 0.0002816673031481817,
      "loss": 1.9019,
      "step": 119686
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.9320688843727112,
      "learning_rate": 0.0002816632202325687,
      "loss": 2.2412,
      "step": 119687
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.9315177798271179,
      "learning_rate": 0.0002816591373203648,
      "loss": 2.0411,
      "step": 119688
    },
    {
      "epoch": 1.56,
      "grad_norm": 1.1146535873413086,
      "learning_rate": 0.0002816550544115708,
      "loss": 1.8373,
      "step": 119689
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.9545869827270508,
      "learning_rate": 0.0002816509715061875,
      "loss": 1.9778,
      "step": 119690
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.9446508884429932,
      "learning_rate": 0.00028164688860421574,
      "loss": 2.2033,
      "step": 119691
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.9754570126533508,
      "learning_rate": 0.0002816428057056561,
      "loss": 2.2316,
      "step": 119692
    },
    {
      "epoch": 1.56,
      "grad_norm": 1.0602525472640991,
      "learning_rate": 0.0002816387228105093,
      "loss": 2.2336,
      "step": 119693
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.9981339573860168,
      "learning_rate": 0.0002816346399187763,
      "loss": 2.236,
      "step": 119694
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.9092590808868408,
      "learning_rate": 0.00028163055703045773,
      "loss": 2.2302,
      "step": 119695
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.9520299434661865,
      "learning_rate": 0.0002816264741455544,
      "loss": 2.1979,
      "step": 119696
    },
    {
      "epoch": 1.56,
      "grad_norm": 1.0133854150772095,
      "learning_rate": 0.00028162239126406713,
      "loss": 2.3469,
      "step": 119697
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.9146593809127808,
      "learning_rate": 0.00028161830838599656,
      "loss": 2.1463,
      "step": 119698
    },
    {
      "epoch": 1.56,
      "grad_norm": 1.0970996618270874,
      "learning_rate": 0.00028161422551134345,
      "loss": 2.1094,
      "step": 119699
    },
    {
      "epoch": 1.56,
      "grad_norm": 1.0279496908187866,
      "learning_rate": 0.0002816101426401086,
      "loss": 2.1817,
      "step": 119700
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.9253728985786438,
      "learning_rate": 0.0002816060597722928,
      "loss": 2.3724,
      "step": 119701
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.9114906191825867,
      "learning_rate": 0.00028160197690789673,
      "loss": 2.0168,
      "step": 119702
    },
    {
      "epoch": 1.56,
      "grad_norm": 1.0823692083358765,
      "learning_rate": 0.0002815978940469213,
      "loss": 2.3349,
      "step": 119703
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.9872097373008728,
      "learning_rate": 0.00028159381118936716,
      "loss": 2.3024,
      "step": 119704
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.9511486887931824,
      "learning_rate": 0.000281589728335235,
      "loss": 2.0156,
      "step": 119705
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.9826439619064331,
      "learning_rate": 0.0002815856454845257,
      "loss": 2.036,
      "step": 119706
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.8952342867851257,
      "learning_rate": 0.00028158156263723995,
      "loss": 2.31,
      "step": 119707
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.9030868411064148,
      "learning_rate": 0.0002815774797933785,
      "loss": 2.0692,
      "step": 119708
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.9303378462791443,
      "learning_rate": 0.0002815733969529422,
      "loss": 2.0424,
      "step": 119709
    },
    {
      "epoch": 1.56,
      "grad_norm": 1.1276472806930542,
      "learning_rate": 0.0002815693141159318,
      "loss": 2.238,
      "step": 119710
    },
    {
      "epoch": 1.56,
      "grad_norm": 1.1690469980239868,
      "learning_rate": 0.000281565231282348,
      "loss": 2.043,
      "step": 119711
    },
    {
      "epoch": 1.56,
      "grad_norm": 1.157530665397644,
      "learning_rate": 0.0002815611484521915,
      "loss": 2.0547,
      "step": 119712
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.8944738507270813,
      "learning_rate": 0.00028155706562546317,
      "loss": 1.9753,
      "step": 119713
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.9340275526046753,
      "learning_rate": 0.0002815529828021637,
      "loss": 2.1695,
      "step": 119714
    },
    {
      "epoch": 1.56,
      "grad_norm": 1.2684470415115356,
      "learning_rate": 0.0002815488999822939,
      "loss": 1.8198,
      "step": 119715
    },
    {
      "epoch": 1.56,
      "grad_norm": 1.117518663406372,
      "learning_rate": 0.00028154481716585457,
      "loss": 2.113,
      "step": 119716
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.9845657348632812,
      "learning_rate": 0.0002815407343528463,
      "loss": 2.1475,
      "step": 119717
    },
    {
      "epoch": 1.56,
      "grad_norm": 1.0326133966445923,
      "learning_rate": 0.00028153665154327006,
      "loss": 2.2767,
      "step": 119718
    },
    {
      "epoch": 1.56,
      "grad_norm": 1.0928540229797363,
      "learning_rate": 0.00028153256873712643,
      "loss": 2.1029,
      "step": 119719
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.9947510361671448,
      "learning_rate": 0.00028152848593441626,
      "loss": 1.9045,
      "step": 119720
    },
    {
      "epoch": 1.56,
      "grad_norm": 1.010926604270935,
      "learning_rate": 0.00028152440313514024,
      "loss": 2.1517,
      "step": 119721
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.9286500811576843,
      "learning_rate": 0.0002815203203392993,
      "loss": 2.0038,
      "step": 119722
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.9779707789421082,
      "learning_rate": 0.000281516237546894,
      "loss": 2.0041,
      "step": 119723
    },
    {
      "epoch": 1.56,
      "grad_norm": 1.0279333591461182,
      "learning_rate": 0.00028151215475792515,
      "loss": 2.0228,
      "step": 119724
    },
    {
      "epoch": 1.56,
      "grad_norm": 1.1233551502227783,
      "learning_rate": 0.00028150807197239357,
      "loss": 2.0691,
      "step": 119725
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.8941583633422852,
      "learning_rate": 0.0002815039891903001,
      "loss": 1.969,
      "step": 119726
    },
    {
      "epoch": 1.56,
      "grad_norm": 1.085758924484253,
      "learning_rate": 0.0002814999064116452,
      "loss": 1.9701,
      "step": 119727
    },
    {
      "epoch": 1.56,
      "grad_norm": 1.0291720628738403,
      "learning_rate": 0.00028149582363643,
      "loss": 2.1792,
      "step": 119728
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.9680368304252625,
      "learning_rate": 0.000281491740864655,
      "loss": 2.0263,
      "step": 119729
    },
    {
      "epoch": 1.56,
      "grad_norm": 1.0132577419281006,
      "learning_rate": 0.00028148765809632095,
      "loss": 2.2222,
      "step": 119730
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.9196929335594177,
      "learning_rate": 0.0002814835753314288,
      "loss": 2.2748,
      "step": 119731
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.9341963529586792,
      "learning_rate": 0.0002814794925699791,
      "loss": 2.1684,
      "step": 119732
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.9134180545806885,
      "learning_rate": 0.0002814754098119729,
      "loss": 2.1635,
      "step": 119733
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.9257879257202148,
      "learning_rate": 0.00028147132705741066,
      "loss": 2.0039,
      "step": 119734
    },
    {
      "epoch": 1.56,
      "grad_norm": 1.0637822151184082,
      "learning_rate": 0.0002814672443062932,
      "loss": 2.1673,
      "step": 119735
    },
    {
      "epoch": 1.56,
      "grad_norm": 1.044024109840393,
      "learning_rate": 0.0002814631615586214,
      "loss": 2.3297,
      "step": 119736
    },
    {
      "epoch": 1.56,
      "grad_norm": 1.0897705554962158,
      "learning_rate": 0.00028145907881439584,
      "loss": 2.2612,
      "step": 119737
    },
    {
      "epoch": 1.56,
      "grad_norm": 1.0330142974853516,
      "learning_rate": 0.00028145499607361746,
      "loss": 2.1887,
      "step": 119738
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.8936783075332642,
      "learning_rate": 0.00028145091333628696,
      "loss": 2.1977,
      "step": 119739
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.9351825714111328,
      "learning_rate": 0.0002814468306024052,
      "loss": 2.0447,
      "step": 119740
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.9227245450019836,
      "learning_rate": 0.00028144274787197265,
      "loss": 2.1253,
      "step": 119741
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.9973228573799133,
      "learning_rate": 0.0002814386651449903,
      "loss": 2.1408,
      "step": 119742
    },
    {
      "epoch": 1.56,
      "grad_norm": 1.2787879705429077,
      "learning_rate": 0.00028143458242145883,
      "loss": 2.0116,
      "step": 119743
    },
    {
      "epoch": 1.56,
      "grad_norm": 1.2813271284103394,
      "learning_rate": 0.000281430499701379,
      "loss": 2.1459,
      "step": 119744
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.9309486746788025,
      "learning_rate": 0.0002814264169847516,
      "loss": 1.9807,
      "step": 119745
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.9729974865913391,
      "learning_rate": 0.00028142233427157757,
      "loss": 2.3067,
      "step": 119746
    },
    {
      "epoch": 1.56,
      "grad_norm": 1.0547947883605957,
      "learning_rate": 0.0002814182515618573,
      "loss": 2.2777,
      "step": 119747
    },
    {
      "epoch": 1.56,
      "grad_norm": 1.2489348649978638,
      "learning_rate": 0.0002814141688555917,
      "loss": 2.2529,
      "step": 119748
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.9348381161689758,
      "learning_rate": 0.0002814100861527816,
      "loss": 2.2775,
      "step": 119749
    },
    {
      "epoch": 1.56,
      "grad_norm": 1.0100008249282837,
      "learning_rate": 0.0002814060034534277,
      "loss": 2.1099,
      "step": 119750
    },
    {
      "epoch": 1.56,
      "grad_norm": 1.0623637437820435,
      "learning_rate": 0.00028140192075753074,
      "loss": 2.0273,
      "step": 119751
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.9527329802513123,
      "learning_rate": 0.0002813978380650917,
      "loss": 2.1877,
      "step": 119752
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.9735941886901855,
      "learning_rate": 0.000281393755376111,
      "loss": 2.0195,
      "step": 119753
    },
    {
      "epoch": 1.56,
      "grad_norm": 1.0128132104873657,
      "learning_rate": 0.00028138967269058954,
      "loss": 2.3836,
      "step": 119754
    },
    {
      "epoch": 1.56,
      "grad_norm": 1.0847804546356201,
      "learning_rate": 0.00028138559000852814,
      "loss": 2.1517,
      "step": 119755
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.9559436440467834,
      "learning_rate": 0.00028138150732992745,
      "loss": 2.275,
      "step": 119756
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.9367109537124634,
      "learning_rate": 0.0002813774246547883,
      "loss": 2.229,
      "step": 119757
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.9433653950691223,
      "learning_rate": 0.0002813733419831116,
      "loss": 2.1075,
      "step": 119758
    },
    {
      "epoch": 1.56,
      "grad_norm": 1.105568289756775,
      "learning_rate": 0.0002813692593148978,
      "loss": 2.3249,
      "step": 119759
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.8700528144836426,
      "learning_rate": 0.0002813651766501478,
      "loss": 2.2169,
      "step": 119760
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.9714381098747253,
      "learning_rate": 0.0002813610939888624,
      "loss": 2.1147,
      "step": 119761
    },
    {
      "epoch": 1.56,
      "grad_norm": 1.0073177814483643,
      "learning_rate": 0.00028135701133104226,
      "loss": 2.3659,
      "step": 119762
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.9987766146659851,
      "learning_rate": 0.00028135292867668824,
      "loss": 2.1915,
      "step": 119763
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.9650261402130127,
      "learning_rate": 0.00028134884602580123,
      "loss": 2.1449,
      "step": 119764
    },
    {
      "epoch": 1.56,
      "grad_norm": 1.0181814432144165,
      "learning_rate": 0.0002813447633783816,
      "loss": 2.3228,
      "step": 119765
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.9695444107055664,
      "learning_rate": 0.0002813406807344304,
      "loss": 1.9499,
      "step": 119766
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.9950366020202637,
      "learning_rate": 0.0002813365980939483,
      "loss": 2.2912,
      "step": 119767
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.9339534044265747,
      "learning_rate": 0.0002813325154569361,
      "loss": 2.2537,
      "step": 119768
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.9020355343818665,
      "learning_rate": 0.00028132843282339454,
      "loss": 2.0349,
      "step": 119769
    },
    {
      "epoch": 1.56,
      "grad_norm": 1.1211493015289307,
      "learning_rate": 0.0002813243501933245,
      "loss": 2.0923,
      "step": 119770
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.9567614793777466,
      "learning_rate": 0.0002813202675667265,
      "loss": 1.8917,
      "step": 119771
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.9367777705192566,
      "learning_rate": 0.00028131618494360133,
      "loss": 2.1855,
      "step": 119772
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.9958431720733643,
      "learning_rate": 0.0002813121023239499,
      "loss": 2.1802,
      "step": 119773
    },
    {
      "epoch": 1.56,
      "grad_norm": 1.0259491205215454,
      "learning_rate": 0.00028130801970777293,
      "loss": 1.9796,
      "step": 119774
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.9457698464393616,
      "learning_rate": 0.0002813039370950711,
      "loss": 2.246,
      "step": 119775
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.9709848165512085,
      "learning_rate": 0.00028129985448584524,
      "loss": 2.245,
      "step": 119776
    },
    {
      "epoch": 1.56,
      "grad_norm": 1.0008888244628906,
      "learning_rate": 0.0002812957718800962,
      "loss": 2.0938,
      "step": 119777
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.9234553575515747,
      "learning_rate": 0.0002812916892778245,
      "loss": 1.9363,
      "step": 119778
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.9349974393844604,
      "learning_rate": 0.0002812876066790311,
      "loss": 2.2642,
      "step": 119779
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.9388425946235657,
      "learning_rate": 0.0002812835240837166,
      "loss": 2.1059,
      "step": 119780
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.9249306917190552,
      "learning_rate": 0.0002812794414918819,
      "loss": 2.1438,
      "step": 119781
    },
    {
      "epoch": 1.56,
      "grad_norm": 1.0388802289962769,
      "learning_rate": 0.00028127535890352766,
      "loss": 2.323,
      "step": 119782
    },
    {
      "epoch": 1.56,
      "grad_norm": 1.000684142112732,
      "learning_rate": 0.00028127127631865484,
      "loss": 2.0758,
      "step": 119783
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.8594157099723816,
      "learning_rate": 0.0002812671937372639,
      "loss": 2.0601,
      "step": 119784
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.9164151549339294,
      "learning_rate": 0.0002812631111593558,
      "loss": 2.4172,
      "step": 119785
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.899236261844635,
      "learning_rate": 0.00028125902858493115,
      "loss": 2.3606,
      "step": 119786
    },
    {
      "epoch": 1.56,
      "grad_norm": 1.1312386989593506,
      "learning_rate": 0.00028125494601399087,
      "loss": 2.1008,
      "step": 119787
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.9082542061805725,
      "learning_rate": 0.0002812508634465356,
      "loss": 2.2811,
      "step": 119788
    },
    {
      "epoch": 1.56,
      "grad_norm": 1.0033706426620483,
      "learning_rate": 0.0002812467808825663,
      "loss": 2.1222,
      "step": 119789
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.9968604445457458,
      "learning_rate": 0.0002812426983220835,
      "loss": 1.9206,
      "step": 119790
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.9266462326049805,
      "learning_rate": 0.00028123861576508796,
      "loss": 2.1367,
      "step": 119791
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.9728717803955078,
      "learning_rate": 0.00028123453321158057,
      "loss": 2.159,
      "step": 119792
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.9760656952857971,
      "learning_rate": 0.000281230450661562,
      "loss": 2.1669,
      "step": 119793
    },
    {
      "epoch": 1.56,
      "grad_norm": 1.061266541481018,
      "learning_rate": 0.00028122636811503307,
      "loss": 2.0387,
      "step": 119794
    },
    {
      "epoch": 1.56,
      "grad_norm": 1.0727217197418213,
      "learning_rate": 0.0002812222855719945,
      "loss": 2.0099,
      "step": 119795
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.9096749424934387,
      "learning_rate": 0.00028121820303244715,
      "loss": 2.0147,
      "step": 119796
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.9534922242164612,
      "learning_rate": 0.00028121412049639163,
      "loss": 2.2337,
      "step": 119797
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.9731283187866211,
      "learning_rate": 0.0002812100379638287,
      "loss": 2.0865,
      "step": 119798
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.9780738949775696,
      "learning_rate": 0.00028120595543475917,
      "loss": 2.2183,
      "step": 119799
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.9170418381690979,
      "learning_rate": 0.00028120187290918387,
      "loss": 2.2863,
      "step": 119800
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.9659188389778137,
      "learning_rate": 0.00028119779038710354,
      "loss": 2.0487,
      "step": 119801
    },
    {
      "epoch": 1.56,
      "grad_norm": 1.0634284019470215,
      "learning_rate": 0.0002811937078685188,
      "loss": 2.0946,
      "step": 119802
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.9069645404815674,
      "learning_rate": 0.0002811896253534306,
      "loss": 2.3086,
      "step": 119803
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.9759739637374878,
      "learning_rate": 0.00028118554284183955,
      "loss": 2.2074,
      "step": 119804
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.9189620018005371,
      "learning_rate": 0.00028118146033374647,
      "loss": 2.1265,
      "step": 119805
    },
    {
      "epoch": 1.56,
      "grad_norm": 1.0029261112213135,
      "learning_rate": 0.00028117737782915205,
      "loss": 2.1731,
      "step": 119806
    },
    {
      "epoch": 1.56,
      "grad_norm": 1.0104660987854004,
      "learning_rate": 0.0002811732953280572,
      "loss": 2.3203,
      "step": 119807
    },
    {
      "epoch": 1.56,
      "grad_norm": 1.0065182447433472,
      "learning_rate": 0.00028116921283046257,
      "loss": 2.0508,
      "step": 119808
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.9775755405426025,
      "learning_rate": 0.00028116513033636886,
      "loss": 1.9399,
      "step": 119809
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.9492990970611572,
      "learning_rate": 0.000281161047845777,
      "loss": 2.1772,
      "step": 119810
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.9252890348434448,
      "learning_rate": 0.00028115696535868764,
      "loss": 2.1316,
      "step": 119811
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.8905481696128845,
      "learning_rate": 0.00028115288287510154,
      "loss": 2.0224,
      "step": 119812
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.9638312458992004,
      "learning_rate": 0.0002811488003950195,
      "loss": 2.2347,
      "step": 119813
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.9148371815681458,
      "learning_rate": 0.00028114471791844224,
      "loss": 2.2413,
      "step": 119814
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.9058324694633484,
      "learning_rate": 0.0002811406354453705,
      "loss": 1.9738,
      "step": 119815
    },
    {
      "epoch": 1.56,
      "grad_norm": 1.0059161186218262,
      "learning_rate": 0.0002811365529758051,
      "loss": 2.1527,
      "step": 119816
    },
    {
      "epoch": 1.56,
      "grad_norm": 1.1725828647613525,
      "learning_rate": 0.00028113247050974674,
      "loss": 2.2164,
      "step": 119817
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.967849612236023,
      "learning_rate": 0.0002811283880471963,
      "loss": 2.0394,
      "step": 119818
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.9537019729614258,
      "learning_rate": 0.00028112430558815445,
      "loss": 2.1881,
      "step": 119819
    },
    {
      "epoch": 1.56,
      "grad_norm": 1.057870626449585,
      "learning_rate": 0.00028112022313262185,
      "loss": 2.108,
      "step": 119820
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.8770801424980164,
      "learning_rate": 0.00028111614068059937,
      "loss": 2.0628,
      "step": 119821
    },
    {
      "epoch": 1.56,
      "grad_norm": 1.0458953380584717,
      "learning_rate": 0.00028111205823208775,
      "loss": 2.1983,
      "step": 119822
    },
    {
      "epoch": 1.56,
      "grad_norm": 1.0261311531066895,
      "learning_rate": 0.00028110797578708775,
      "loss": 2.2457,
      "step": 119823
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.9564394950866699,
      "learning_rate": 0.0002811038933456002,
      "loss": 2.1977,
      "step": 119824
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.9840719103813171,
      "learning_rate": 0.0002810998109076258,
      "loss": 2.2027,
      "step": 119825
    },
    {
      "epoch": 1.56,
      "grad_norm": 1.035948395729065,
      "learning_rate": 0.0002810957284731653,
      "loss": 2.1928,
      "step": 119826
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.992961049079895,
      "learning_rate": 0.00028109164604221936,
      "loss": 2.1478,
      "step": 119827
    },
    {
      "epoch": 1.56,
      "grad_norm": 1.0449484586715698,
      "learning_rate": 0.0002810875636147889,
      "loss": 2.2263,
      "step": 119828
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.9878566861152649,
      "learning_rate": 0.00028108348119087464,
      "loss": 2.2535,
      "step": 119829
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.9701952338218689,
      "learning_rate": 0.00028107939877047725,
      "loss": 2.1329,
      "step": 119830
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.9903674125671387,
      "learning_rate": 0.00028107531635359777,
      "loss": 2.0478,
      "step": 119831
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.929337739944458,
      "learning_rate": 0.0002810712339402365,
      "loss": 2.1235,
      "step": 119832
    },
    {
      "epoch": 1.56,
      "grad_norm": 1.0810364484786987,
      "learning_rate": 0.00028106715153039454,
      "loss": 2.254,
      "step": 119833
    },
    {
      "epoch": 1.56,
      "grad_norm": 1.0535541772842407,
      "learning_rate": 0.00028106306912407253,
      "loss": 2.1576,
      "step": 119834
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.9245268702507019,
      "learning_rate": 0.00028105898672127125,
      "loss": 2.2496,
      "step": 119835
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.990845263004303,
      "learning_rate": 0.00028105490432199146,
      "loss": 2.2768,
      "step": 119836
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.9715631008148193,
      "learning_rate": 0.000281050821926234,
      "loss": 2.3317,
      "step": 119837
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.932617723941803,
      "learning_rate": 0.0002810467395339995,
      "loss": 1.999,
      "step": 119838
    },
    {
      "epoch": 1.56,
      "grad_norm": 1.0101349353790283,
      "learning_rate": 0.0002810426571452887,
      "loss": 2.1953,
      "step": 119839
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.9192568063735962,
      "learning_rate": 0.0002810385747601025,
      "loss": 2.086,
      "step": 119840
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.9471127986907959,
      "learning_rate": 0.00028103449237844156,
      "loss": 2.1429,
      "step": 119841
    },
    {
      "epoch": 1.56,
      "grad_norm": 1.0099730491638184,
      "learning_rate": 0.00028103041000030664,
      "loss": 2.1593,
      "step": 119842
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.9779790043830872,
      "learning_rate": 0.0002810263276256985,
      "loss": 1.954,
      "step": 119843
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.9620871543884277,
      "learning_rate": 0.0002810222452546182,
      "loss": 2.1273,
      "step": 119844
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.9891713261604309,
      "learning_rate": 0.00028101816288706594,
      "loss": 2.1501,
      "step": 119845
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.8716866374015808,
      "learning_rate": 0.0002810140805230428,
      "loss": 2.0274,
      "step": 119846
    },
    {
      "epoch": 1.56,
      "grad_norm": 1.0020718574523926,
      "learning_rate": 0.00028100999816254956,
      "loss": 2.1912,
      "step": 119847
    },
    {
      "epoch": 1.56,
      "grad_norm": 1.0171499252319336,
      "learning_rate": 0.0002810059158055868,
      "loss": 2.2113,
      "step": 119848
    },
    {
      "epoch": 1.56,
      "grad_norm": 1.161828875541687,
      "learning_rate": 0.0002810018334521555,
      "loss": 2.1043,
      "step": 119849
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.9946181774139404,
      "learning_rate": 0.00028099775110225646,
      "loss": 2.3335,
      "step": 119850
    },
    {
      "epoch": 1.56,
      "grad_norm": 1.1831289529800415,
      "learning_rate": 0.0002809936687558901,
      "loss": 2.1086,
      "step": 119851
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.9371004700660706,
      "learning_rate": 0.0002809895864130574,
      "loss": 2.0766,
      "step": 119852
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.9960626363754272,
      "learning_rate": 0.0002809855040737591,
      "loss": 2.1963,
      "step": 119853
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.8974478244781494,
      "learning_rate": 0.00028098142173799595,
      "loss": 1.9788,
      "step": 119854
    },
    {
      "epoch": 1.56,
      "grad_norm": 1.014859676361084,
      "learning_rate": 0.0002809773394057687,
      "loss": 2.2436,
      "step": 119855
    },
    {
      "epoch": 1.56,
      "grad_norm": 1.0505051612854004,
      "learning_rate": 0.0002809732570770783,
      "loss": 2.1928,
      "step": 119856
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.963530421257019,
      "learning_rate": 0.0002809691747519251,
      "loss": 2.0933,
      "step": 119857
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.9058471322059631,
      "learning_rate": 0.0002809650924303102,
      "loss": 2.1349,
      "step": 119858
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.9466278553009033,
      "learning_rate": 0.0002809610101122342,
      "loss": 2.1798,
      "step": 119859
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.9818823933601379,
      "learning_rate": 0.0002809569277976979,
      "loss": 2.1058,
      "step": 119860
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.8959835767745972,
      "learning_rate": 0.0002809528454867021,
      "loss": 2.2629,
      "step": 119861
    },
    {
      "epoch": 1.56,
      "grad_norm": 1.1234618425369263,
      "learning_rate": 0.0002809487631792476,
      "loss": 2.2221,
      "step": 119862
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.950361430644989,
      "learning_rate": 0.000280944680875335,
      "loss": 2.3298,
      "step": 119863
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.8219113349914551,
      "learning_rate": 0.0002809405985749651,
      "loss": 2.2231,
      "step": 119864
    },
    {
      "epoch": 1.56,
      "grad_norm": 1.1165226697921753,
      "learning_rate": 0.0002809365162781388,
      "loss": 2.0258,
      "step": 119865
    },
    {
      "epoch": 1.56,
      "grad_norm": 1.2581322193145752,
      "learning_rate": 0.0002809324339848567,
      "loss": 2.1789,
      "step": 119866
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.9569646120071411,
      "learning_rate": 0.0002809283516951196,
      "loss": 2.2393,
      "step": 119867
    },
    {
      "epoch": 1.56,
      "grad_norm": 1.0723909139633179,
      "learning_rate": 0.00028092426940892843,
      "loss": 2.084,
      "step": 119868
    },
    {
      "epoch": 1.56,
      "grad_norm": 1.1521350145339966,
      "learning_rate": 0.00028092018712628364,
      "loss": 2.1577,
      "step": 119869
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.8546621799468994,
      "learning_rate": 0.0002809161048471862,
      "loss": 2.1176,
      "step": 119870
    },
    {
      "epoch": 1.56,
      "grad_norm": 1.0252392292022705,
      "learning_rate": 0.00028091202257163675,
      "loss": 2.0867,
      "step": 119871
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.9858057498931885,
      "learning_rate": 0.00028090794029963617,
      "loss": 2.1353,
      "step": 119872
    },
    {
      "epoch": 1.56,
      "grad_norm": 1.0428212881088257,
      "learning_rate": 0.00028090385803118516,
      "loss": 2.246,
      "step": 119873
    },
    {
      "epoch": 1.56,
      "grad_norm": 1.0038822889328003,
      "learning_rate": 0.0002808997757662846,
      "loss": 2.234,
      "step": 119874
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.9793132543563843,
      "learning_rate": 0.000280895693504935,
      "loss": 2.1462,
      "step": 119875
    },
    {
      "epoch": 1.56,
      "grad_norm": 1.12264084815979,
      "learning_rate": 0.00028089161124713723,
      "loss": 2.0732,
      "step": 119876
    },
    {
      "epoch": 1.56,
      "grad_norm": 1.3035340309143066,
      "learning_rate": 0.0002808875289928921,
      "loss": 2.1288,
      "step": 119877
    },
    {
      "epoch": 1.56,
      "grad_norm": 1.002399206161499,
      "learning_rate": 0.0002808834467422003,
      "loss": 2.185,
      "step": 119878
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.8531994819641113,
      "learning_rate": 0.0002808793644950627,
      "loss": 2.1651,
      "step": 119879
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.897911787033081,
      "learning_rate": 0.00028087528225148,
      "loss": 2.1861,
      "step": 119880
    },
    {
      "epoch": 1.56,
      "grad_norm": 1.0093677043914795,
      "learning_rate": 0.00028087120001145296,
      "loss": 1.9415,
      "step": 119881
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.9788714647293091,
      "learning_rate": 0.0002808671177749823,
      "loss": 2.1223,
      "step": 119882
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.9571638703346252,
      "learning_rate": 0.0002808630355420687,
      "loss": 2.0328,
      "step": 119883
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.9687179923057556,
      "learning_rate": 0.0002808589533127131,
      "loss": 2.0163,
      "step": 119884
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.857197642326355,
      "learning_rate": 0.00028085487108691617,
      "loss": 1.9104,
      "step": 119885
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.9562440514564514,
      "learning_rate": 0.0002808507888646788,
      "loss": 2.142,
      "step": 119886
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.950796365737915,
      "learning_rate": 0.0002808467066460015,
      "loss": 2.0993,
      "step": 119887
    },
    {
      "epoch": 1.56,
      "grad_norm": 1.0014129877090454,
      "learning_rate": 0.0002808426244308852,
      "loss": 2.1132,
      "step": 119888
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.990664541721344,
      "learning_rate": 0.0002808385422193306,
      "loss": 2.1709,
      "step": 119889
    },
    {
      "epoch": 1.56,
      "grad_norm": 1.0224539041519165,
      "learning_rate": 0.0002808344600113384,
      "loss": 2.1147,
      "step": 119890
    },
    {
      "epoch": 1.56,
      "grad_norm": 1.0351899862289429,
      "learning_rate": 0.00028083037780690954,
      "loss": 2.2217,
      "step": 119891
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.9369682669639587,
      "learning_rate": 0.0002808262956060447,
      "loss": 2.1071,
      "step": 119892
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.9609928131103516,
      "learning_rate": 0.0002808222134087445,
      "loss": 2.169,
      "step": 119893
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.8558789491653442,
      "learning_rate": 0.0002808181312150099,
      "loss": 2.1947,
      "step": 119894
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.9720581769943237,
      "learning_rate": 0.0002808140490248416,
      "loss": 2.0253,
      "step": 119895
    },
    {
      "epoch": 1.56,
      "grad_norm": 1.0039360523223877,
      "learning_rate": 0.00028080996683824025,
      "loss": 2.0219,
      "step": 119896
    },
    {
      "epoch": 1.56,
      "grad_norm": 1.066641926765442,
      "learning_rate": 0.0002808058846552067,
      "loss": 2.2498,
      "step": 119897
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.901791512966156,
      "learning_rate": 0.0002808018024757418,
      "loss": 2.303,
      "step": 119898
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.91112220287323,
      "learning_rate": 0.0002807977202998461,
      "loss": 2.1205,
      "step": 119899
    },
    {
      "epoch": 1.56,
      "grad_norm": 1.1059249639511108,
      "learning_rate": 0.00028079363812752044,
      "loss": 2.4153,
      "step": 119900
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.8961542844772339,
      "learning_rate": 0.00028078955595876566,
      "loss": 2.0093,
      "step": 119901
    },
    {
      "epoch": 1.56,
      "grad_norm": 1.1014511585235596,
      "learning_rate": 0.00028078547379358247,
      "loss": 2.1924,
      "step": 119902
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.8955370187759399,
      "learning_rate": 0.0002807813916319716,
      "loss": 2.1324,
      "step": 119903
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.9253262281417847,
      "learning_rate": 0.0002807773094739339,
      "loss": 1.9492,
      "step": 119904
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.8642206192016602,
      "learning_rate": 0.00028077322731947,
      "loss": 2.058,
      "step": 119905
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.874119758605957,
      "learning_rate": 0.0002807691451685807,
      "loss": 2.0115,
      "step": 119906
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.9601033329963684,
      "learning_rate": 0.0002807650630212668,
      "loss": 2.1798,
      "step": 119907
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.9738238453865051,
      "learning_rate": 0.000280760980877529,
      "loss": 2.1038,
      "step": 119908
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.9930723309516907,
      "learning_rate": 0.0002807568987373681,
      "loss": 2.2163,
      "step": 119909
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.9020670652389526,
      "learning_rate": 0.000280752816600785,
      "loss": 2.0711,
      "step": 119910
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.8994677066802979,
      "learning_rate": 0.00028074873446778026,
      "loss": 1.9587,
      "step": 119911
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.8811948299407959,
      "learning_rate": 0.0002807446523383546,
      "loss": 2.1813,
      "step": 119912
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.9399690628051758,
      "learning_rate": 0.0002807405702125089,
      "loss": 2.1125,
      "step": 119913
    },
    {
      "epoch": 1.56,
      "grad_norm": 1.009276270866394,
      "learning_rate": 0.0002807364880902439,
      "loss": 2.2147,
      "step": 119914
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.9259420037269592,
      "learning_rate": 0.0002807324059715603,
      "loss": 2.1802,
      "step": 119915
    },
    {
      "epoch": 1.56,
      "grad_norm": 1.057108759880066,
      "learning_rate": 0.00028072832385645896,
      "loss": 2.1781,
      "step": 119916
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.9602921605110168,
      "learning_rate": 0.00028072424174494073,
      "loss": 2.1785,
      "step": 119917
    },
    {
      "epoch": 1.56,
      "grad_norm": 1.0007725954055786,
      "learning_rate": 0.00028072015963700607,
      "loss": 2.0913,
      "step": 119918
    },
    {
      "epoch": 1.56,
      "grad_norm": 1.0588865280151367,
      "learning_rate": 0.0002807160775326559,
      "loss": 2.2443,
      "step": 119919
    },
    {
      "epoch": 1.56,
      "grad_norm": 1.0068458318710327,
      "learning_rate": 0.00028071199543189094,
      "loss": 2.2865,
      "step": 119920
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.9169822335243225,
      "learning_rate": 0.00028070791333471205,
      "loss": 2.2423,
      "step": 119921
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.9481079578399658,
      "learning_rate": 0.00028070383124111987,
      "loss": 2.143,
      "step": 119922
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.9426361918449402,
      "learning_rate": 0.00028069974915111536,
      "loss": 2.1184,
      "step": 119923
    },
    {
      "epoch": 1.56,
      "grad_norm": 1.0273507833480835,
      "learning_rate": 0.0002806956670646991,
      "loss": 2.1535,
      "step": 119924
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.9354500770568848,
      "learning_rate": 0.0002806915849818717,
      "loss": 2.0404,
      "step": 119925
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.9347271919250488,
      "learning_rate": 0.0002806875029026342,
      "loss": 2.2054,
      "step": 119926
    },
    {
      "epoch": 1.56,
      "grad_norm": 1.005493402481079,
      "learning_rate": 0.00028068342082698724,
      "loss": 2.177,
      "step": 119927
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.9348340630531311,
      "learning_rate": 0.0002806793387549316,
      "loss": 2.3196,
      "step": 119928
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.9434898495674133,
      "learning_rate": 0.0002806752566864682,
      "loss": 2.2126,
      "step": 119929
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.9060819149017334,
      "learning_rate": 0.0002806711746215974,
      "loss": 2.2624,
      "step": 119930
    },
    {
      "epoch": 1.56,
      "grad_norm": 1.0180304050445557,
      "learning_rate": 0.0002806670925603203,
      "loss": 2.006,
      "step": 119931
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.9843094348907471,
      "learning_rate": 0.0002806630105026375,
      "loss": 2.1267,
      "step": 119932
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.9486585855484009,
      "learning_rate": 0.00028065892844854985,
      "loss": 1.9231,
      "step": 119933
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.8733850717544556,
      "learning_rate": 0.00028065484639805806,
      "loss": 1.9001,
      "step": 119934
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.9196085929870605,
      "learning_rate": 0.000280650764351163,
      "loss": 2.1501,
      "step": 119935
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.9623598456382751,
      "learning_rate": 0.0002806466823078652,
      "loss": 2.4843,
      "step": 119936
    },
    {
      "epoch": 1.56,
      "grad_norm": 1.024906039237976,
      "learning_rate": 0.0002806426002681656,
      "loss": 2.4932,
      "step": 119937
    },
    {
      "epoch": 1.56,
      "grad_norm": 1.026750922203064,
      "learning_rate": 0.00028063851823206484,
      "loss": 2.0691,
      "step": 119938
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.932510256767273,
      "learning_rate": 0.00028063443619956376,
      "loss": 2.0155,
      "step": 119939
    },
    {
      "epoch": 1.56,
      "grad_norm": 1.0298898220062256,
      "learning_rate": 0.00028063035417066315,
      "loss": 2.2341,
      "step": 119940
    },
    {
      "epoch": 1.56,
      "grad_norm": 1.0034123659133911,
      "learning_rate": 0.0002806262721453638,
      "loss": 2.0262,
      "step": 119941
    },
    {
      "epoch": 1.56,
      "grad_norm": 1.0589290857315063,
      "learning_rate": 0.0002806221901236663,
      "loss": 2.2513,
      "step": 119942
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.9465268850326538,
      "learning_rate": 0.0002806181081055715,
      "loss": 1.9684,
      "step": 119943
    },
    {
      "epoch": 1.56,
      "grad_norm": 1.3335212469100952,
      "learning_rate": 0.0002806140260910801,
      "loss": 2.0514,
      "step": 119944
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.9435111284255981,
      "learning_rate": 0.00028060994408019295,
      "loss": 2.159,
      "step": 119945
    },
    {
      "epoch": 1.56,
      "grad_norm": 1.0933395624160767,
      "learning_rate": 0.0002806058620729108,
      "loss": 2.3022,
      "step": 119946
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.9080691933631897,
      "learning_rate": 0.00028060178006923453,
      "loss": 2.0407,
      "step": 119947
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.9326172471046448,
      "learning_rate": 0.0002805976980691646,
      "loss": 2.1399,
      "step": 119948
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.9467549324035645,
      "learning_rate": 0.0002805936160727019,
      "loss": 2.4428,
      "step": 119949
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.9863135814666748,
      "learning_rate": 0.0002805895340798472,
      "loss": 2.0508,
      "step": 119950
    },
    {
      "epoch": 1.56,
      "grad_norm": 1.0505579710006714,
      "learning_rate": 0.00028058545209060136,
      "loss": 2.0929,
      "step": 119951
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.9599350690841675,
      "learning_rate": 0.000280581370104965,
      "loss": 2.0652,
      "step": 119952
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.9729031920433044,
      "learning_rate": 0.00028057728812293903,
      "loss": 1.9419,
      "step": 119953
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.8546549081802368,
      "learning_rate": 0.00028057320614452403,
      "loss": 2.0627,
      "step": 119954
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.9191127419471741,
      "learning_rate": 0.00028056912416972083,
      "loss": 2.3844,
      "step": 119955
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.7904098033905029,
      "learning_rate": 0.00028056504219853013,
      "loss": 1.9717,
      "step": 119956
    },
    {
      "epoch": 1.56,
      "grad_norm": 1.0192933082580566,
      "learning_rate": 0.0002805609602309528,
      "loss": 2.0518,
      "step": 119957
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.9562844634056091,
      "learning_rate": 0.00028055687826698957,
      "loss": 1.8423,
      "step": 119958
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.9430238604545593,
      "learning_rate": 0.00028055279630664133,
      "loss": 2.1959,
      "step": 119959
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.9846982955932617,
      "learning_rate": 0.00028054871434990853,
      "loss": 1.9874,
      "step": 119960
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.9681187868118286,
      "learning_rate": 0.00028054463239679214,
      "loss": 2.1246,
      "step": 119961
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.9427374005317688,
      "learning_rate": 0.0002805405504472928,
      "loss": 2.2748,
      "step": 119962
    },
    {
      "epoch": 1.56,
      "grad_norm": 1.0258440971374512,
      "learning_rate": 0.00028053646850141134,
      "loss": 2.222,
      "step": 119963
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.8744685649871826,
      "learning_rate": 0.00028053238655914857,
      "loss": 2.0057,
      "step": 119964
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.965408444404602,
      "learning_rate": 0.0002805283046205053,
      "loss": 2.2539,
      "step": 119965
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.9364449381828308,
      "learning_rate": 0.0002805242226854821,
      "loss": 2.0102,
      "step": 119966
    },
    {
      "epoch": 1.56,
      "grad_norm": 1.083019495010376,
      "learning_rate": 0.0002805201407540797,
      "loss": 1.9502,
      "step": 119967
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.9203985333442688,
      "learning_rate": 0.00028051605882629905,
      "loss": 2.0535,
      "step": 119968
    },
    {
      "epoch": 1.56,
      "grad_norm": 1.0883601903915405,
      "learning_rate": 0.00028051197690214083,
      "loss": 2.1562,
      "step": 119969
    },
    {
      "epoch": 1.56,
      "grad_norm": 1.0026354789733887,
      "learning_rate": 0.00028050789498160585,
      "loss": 2.252,
      "step": 119970
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.9716161489486694,
      "learning_rate": 0.0002805038130646948,
      "loss": 2.3427,
      "step": 119971
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.9801175594329834,
      "learning_rate": 0.0002804997311514085,
      "loss": 2.0491,
      "step": 119972
    },
    {
      "epoch": 1.56,
      "grad_norm": 1.0620168447494507,
      "learning_rate": 0.00028049564924174757,
      "loss": 2.3367,
      "step": 119973
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.9291974306106567,
      "learning_rate": 0.0002804915673357129,
      "loss": 2.2171,
      "step": 119974
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.8870474696159363,
      "learning_rate": 0.0002804874854333052,
      "loss": 2.2052,
      "step": 119975
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.9463172554969788,
      "learning_rate": 0.0002804834035345253,
      "loss": 1.9372,
      "step": 119976
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.9802181124687195,
      "learning_rate": 0.00028047932163937386,
      "loss": 2.2946,
      "step": 119977
    },
    {
      "epoch": 1.56,
      "grad_norm": 1.0178875923156738,
      "learning_rate": 0.00028047523974785175,
      "loss": 2.2007,
      "step": 119978
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.9617602825164795,
      "learning_rate": 0.00028047115785995955,
      "loss": 2.1337,
      "step": 119979
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.957859218120575,
      "learning_rate": 0.0002804670759756983,
      "loss": 2.2361,
      "step": 119980
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.9961228966712952,
      "learning_rate": 0.00028046299409506846,
      "loss": 2.0507,
      "step": 119981
    },
    {
      "epoch": 1.56,
      "grad_norm": 1.0296530723571777,
      "learning_rate": 0.0002804589122180709,
      "loss": 2.1383,
      "step": 119982
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.9430223703384399,
      "learning_rate": 0.0002804548303447064,
      "loss": 2.1865,
      "step": 119983
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.9558804035186768,
      "learning_rate": 0.0002804507484749758,
      "loss": 2.258,
      "step": 119984
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.8564355969429016,
      "learning_rate": 0.00028044666660887975,
      "loss": 2.3281,
      "step": 119985
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.9644267559051514,
      "learning_rate": 0.000280442584746419,
      "loss": 2.3625,
      "step": 119986
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.9751022458076477,
      "learning_rate": 0.00028043850288759435,
      "loss": 2.2183,
      "step": 119987
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.9700262546539307,
      "learning_rate": 0.00028043442103240656,
      "loss": 2.098,
      "step": 119988
    },
    {
      "epoch": 1.56,
      "grad_norm": 1.0388329029083252,
      "learning_rate": 0.00028043033918085633,
      "loss": 2.1098,
      "step": 119989
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.8803285360336304,
      "learning_rate": 0.0002804262573329446,
      "loss": 2.032,
      "step": 119990
    },
    {
      "epoch": 1.56,
      "grad_norm": 1.0122740268707275,
      "learning_rate": 0.0002804221754886719,
      "loss": 2.0273,
      "step": 119991
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.98578280210495,
      "learning_rate": 0.00028041809364803903,
      "loss": 2.1968,
      "step": 119992
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.9433102607727051,
      "learning_rate": 0.0002804140118110469,
      "loss": 2.1232,
      "step": 119993
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.9433612823486328,
      "learning_rate": 0.00028040992997769616,
      "loss": 2.0684,
      "step": 119994
    },
    {
      "epoch": 1.56,
      "grad_norm": 1.40290105342865,
      "learning_rate": 0.00028040584814798756,
      "loss": 2.1562,
      "step": 119995
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.869199812412262,
      "learning_rate": 0.00028040176632192194,
      "loss": 2.0648,
      "step": 119996
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.928561270236969,
      "learning_rate": 0.0002803976844995,
      "loss": 1.976,
      "step": 119997
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.9310710430145264,
      "learning_rate": 0.0002803936026807224,
      "loss": 2.1073,
      "step": 119998
    },
    {
      "epoch": 1.56,
      "grad_norm": 1.062543511390686,
      "learning_rate": 0.00028038952086559005,
      "loss": 2.0769,
      "step": 119999
    },
    {
      "epoch": 1.56,
      "grad_norm": 1.1045629978179932,
      "learning_rate": 0.00028038543905410364,
      "loss": 2.1926,
      "step": 120000
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.9078657031059265,
      "learning_rate": 0.000280381357246264,
      "loss": 2.1305,
      "step": 120001
    },
    {
      "epoch": 1.56,
      "grad_norm": 1.054956078529358,
      "learning_rate": 0.0002803772754420719,
      "loss": 2.0389,
      "step": 120002
    },
    {
      "epoch": 1.56,
      "grad_norm": 1.1335617303848267,
      "learning_rate": 0.0002803731936415279,
      "loss": 2.2883,
      "step": 120003
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.8544585704803467,
      "learning_rate": 0.00028036911184463293,
      "loss": 2.1408,
      "step": 120004
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.9760571122169495,
      "learning_rate": 0.00028036503005138773,
      "loss": 2.0569,
      "step": 120005
    },
    {
      "epoch": 1.56,
      "grad_norm": 1.095557451248169,
      "learning_rate": 0.000280360948261793,
      "loss": 2.3444,
      "step": 120006
    },
    {
      "epoch": 1.56,
      "grad_norm": 1.0766679048538208,
      "learning_rate": 0.00028035686647584956,
      "loss": 1.9881,
      "step": 120007
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.9600657224655151,
      "learning_rate": 0.0002803527846935583,
      "loss": 2.4509,
      "step": 120008
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.8672859072685242,
      "learning_rate": 0.0002803487029149196,
      "loss": 2.1458,
      "step": 120009
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.9556176662445068,
      "learning_rate": 0.00028034462113993454,
      "loss": 2.2861,
      "step": 120010
    },
    {
      "epoch": 1.56,
      "grad_norm": 1.0115617513656616,
      "learning_rate": 0.0002803405393686038,
      "loss": 2.3744,
      "step": 120011
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.9333113431930542,
      "learning_rate": 0.0002803364576009281,
      "loss": 2.1552,
      "step": 120012
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.9963865280151367,
      "learning_rate": 0.0002803323758369082,
      "loss": 2.153,
      "step": 120013
    },
    {
      "epoch": 1.56,
      "grad_norm": 1.0312037467956543,
      "learning_rate": 0.000280328294076545,
      "loss": 2.1458,
      "step": 120014
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.9244654774665833,
      "learning_rate": 0.00028032421231983905,
      "loss": 2.1803,
      "step": 120015
    },
    {
      "epoch": 1.56,
      "grad_norm": 1.0276994705200195,
      "learning_rate": 0.0002803201305667912,
      "loss": 2.2505,
      "step": 120016
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.8959465622901917,
      "learning_rate": 0.0002803160488174022,
      "loss": 2.0863,
      "step": 120017
    },
    {
      "epoch": 1.56,
      "grad_norm": 1.1068874597549438,
      "learning_rate": 0.0002803119670716728,
      "loss": 2.1251,
      "step": 120018
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.9599209427833557,
      "learning_rate": 0.0002803078853296038,
      "loss": 2.1128,
      "step": 120019
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.9297695159912109,
      "learning_rate": 0.00028030380359119606,
      "loss": 2.012,
      "step": 120020
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.8972075581550598,
      "learning_rate": 0.00028029972185645007,
      "loss": 2.1133,
      "step": 120021
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.9905803203582764,
      "learning_rate": 0.00028029564012536673,
      "loss": 2.1812,
      "step": 120022
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.9670885801315308,
      "learning_rate": 0.00028029155839794686,
      "loss": 2.146,
      "step": 120023
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.8832228779792786,
      "learning_rate": 0.0002802874766741911,
      "loss": 2.095,
      "step": 120024
    },
    {
      "epoch": 1.56,
      "grad_norm": 1.0480213165283203,
      "learning_rate": 0.0002802833949541003,
      "loss": 2.4047,
      "step": 120025
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.8846463561058044,
      "learning_rate": 0.0002802793132376753,
      "loss": 2.2201,
      "step": 120026
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.9182807803153992,
      "learning_rate": 0.00028027523152491667,
      "loss": 1.9188,
      "step": 120027
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.9388259649276733,
      "learning_rate": 0.0002802711498158252,
      "loss": 2.2227,
      "step": 120028
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.9905532002449036,
      "learning_rate": 0.00028026706811040166,
      "loss": 2.0244,
      "step": 120029
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.9545267224311829,
      "learning_rate": 0.0002802629864086469,
      "loss": 2.0392,
      "step": 120030
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.8952404856681824,
      "learning_rate": 0.0002802589047105616,
      "loss": 2.211,
      "step": 120031
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.926887035369873,
      "learning_rate": 0.0002802548230161467,
      "loss": 1.9605,
      "step": 120032
    },
    {
      "epoch": 1.56,
      "grad_norm": 1.0418074131011963,
      "learning_rate": 0.00028025074132540266,
      "loss": 2.2545,
      "step": 120033
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.9121879935264587,
      "learning_rate": 0.00028024665963833034,
      "loss": 1.9509,
      "step": 120034
    },
    {
      "epoch": 1.56,
      "grad_norm": 1.047824501991272,
      "learning_rate": 0.0002802425779549306,
      "loss": 2.0431,
      "step": 120035
    },
    {
      "epoch": 1.56,
      "grad_norm": 1.0070111751556396,
      "learning_rate": 0.00028023849627520413,
      "loss": 2.4527,
      "step": 120036
    },
    {
      "epoch": 1.56,
      "grad_norm": 1.0227010250091553,
      "learning_rate": 0.00028023441459915164,
      "loss": 2.154,
      "step": 120037
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.9727964997291565,
      "learning_rate": 0.0002802303329267741,
      "loss": 2.0954,
      "step": 120038
    },
    {
      "epoch": 1.56,
      "grad_norm": 1.0020263195037842,
      "learning_rate": 0.00028022625125807203,
      "loss": 2.2317,
      "step": 120039
    },
    {
      "epoch": 1.56,
      "grad_norm": 1.0319244861602783,
      "learning_rate": 0.0002802221695930462,
      "loss": 2.038,
      "step": 120040
    },
    {
      "epoch": 1.56,
      "grad_norm": 1.0009289979934692,
      "learning_rate": 0.0002802180879316975,
      "loss": 2.0928,
      "step": 120041
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.9683286547660828,
      "learning_rate": 0.00028021400627402664,
      "loss": 2.1308,
      "step": 120042
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.9201856851577759,
      "learning_rate": 0.0002802099246200343,
      "loss": 2.2309,
      "step": 120043
    },
    {
      "epoch": 1.56,
      "grad_norm": 1.073635220527649,
      "learning_rate": 0.0002802058429697214,
      "loss": 2.0919,
      "step": 120044
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.9380197525024414,
      "learning_rate": 0.00028020176132308866,
      "loss": 2.0479,
      "step": 120045
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.9532285332679749,
      "learning_rate": 0.0002801976796801367,
      "loss": 2.1661,
      "step": 120046
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.9675483703613281,
      "learning_rate": 0.0002801935980408663,
      "loss": 2.1547,
      "step": 120047
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.8932011723518372,
      "learning_rate": 0.00028018951640527837,
      "loss": 2.2238,
      "step": 120048
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.9658597707748413,
      "learning_rate": 0.0002801854347733735,
      "loss": 2.1638,
      "step": 120049
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.8920052647590637,
      "learning_rate": 0.0002801813531451526,
      "loss": 2.291,
      "step": 120050
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.9167979955673218,
      "learning_rate": 0.0002801772715206165,
      "loss": 1.973,
      "step": 120051
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.8788396716117859,
      "learning_rate": 0.0002801731898997656,
      "loss": 2.3487,
      "step": 120052
    },
    {
      "epoch": 1.56,
      "grad_norm": 1.0487804412841797,
      "learning_rate": 0.00028016910828260096,
      "loss": 2.3855,
      "step": 120053
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.9243032336235046,
      "learning_rate": 0.00028016502666912326,
      "loss": 2.1031,
      "step": 120054
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.9295710325241089,
      "learning_rate": 0.0002801609450593332,
      "loss": 2.1267,
      "step": 120055
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.9995700120925903,
      "learning_rate": 0.0002801568634532316,
      "loss": 2.4117,
      "step": 120056
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.9256582856178284,
      "learning_rate": 0.0002801527818508194,
      "loss": 2.0847,
      "step": 120057
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.9232501983642578,
      "learning_rate": 0.000280148700252097,
      "loss": 2.0522,
      "step": 120058
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.9804897308349609,
      "learning_rate": 0.0002801446186570653,
      "loss": 2.0018,
      "step": 120059
    },
    {
      "epoch": 1.56,
      "grad_norm": 1.0217223167419434,
      "learning_rate": 0.0002801405370657252,
      "loss": 2.1713,
      "step": 120060
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.9946359992027283,
      "learning_rate": 0.00028013645547807724,
      "loss": 1.904,
      "step": 120061
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.9689989686012268,
      "learning_rate": 0.00028013237389412236,
      "loss": 2.1819,
      "step": 120062
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.9685442447662354,
      "learning_rate": 0.00028012829231386125,
      "loss": 2.0806,
      "step": 120063
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.907831072807312,
      "learning_rate": 0.0002801242107372947,
      "loss": 1.8347,
      "step": 120064
    },
    {
      "epoch": 1.56,
      "grad_norm": 1.0122705698013306,
      "learning_rate": 0.0002801201291644234,
      "loss": 1.9787,
      "step": 120065
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.9897719025611877,
      "learning_rate": 0.00028011604759524807,
      "loss": 2.2539,
      "step": 120066
    },
    {
      "epoch": 1.56,
      "grad_norm": 1.0035268068313599,
      "learning_rate": 0.00028011196602976957,
      "loss": 2.1548,
      "step": 120067
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.9906973242759705,
      "learning_rate": 0.0002801078844679887,
      "loss": 2.0861,
      "step": 120068
    },
    {
      "epoch": 1.56,
      "grad_norm": 1.0157309770584106,
      "learning_rate": 0.0002801038029099061,
      "loss": 2.2624,
      "step": 120069
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.9194751977920532,
      "learning_rate": 0.0002800997213555226,
      "loss": 2.277,
      "step": 120070
    },
    {
      "epoch": 1.56,
      "grad_norm": 1.0017164945602417,
      "learning_rate": 0.0002800956398048389,
      "loss": 2.1378,
      "step": 120071
    },
    {
      "epoch": 1.56,
      "grad_norm": 1.1391690969467163,
      "learning_rate": 0.00028009155825785585,
      "loss": 2.1025,
      "step": 120072
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.9184580445289612,
      "learning_rate": 0.0002800874767145741,
      "loss": 2.1906,
      "step": 120073
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.9046709537506104,
      "learning_rate": 0.0002800833951749945,
      "loss": 2.1199,
      "step": 120074
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.8952388763427734,
      "learning_rate": 0.0002800793136391178,
      "loss": 1.9896,
      "step": 120075
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.9092884063720703,
      "learning_rate": 0.0002800752321069447,
      "loss": 1.9749,
      "step": 120076
    },
    {
      "epoch": 1.56,
      "grad_norm": 1.074468970298767,
      "learning_rate": 0.0002800711505784759,
      "loss": 2.2457,
      "step": 120077
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.9161821603775024,
      "learning_rate": 0.00028006706905371235,
      "loss": 1.8698,
      "step": 120078
    },
    {
      "epoch": 1.56,
      "grad_norm": 1.1446994543075562,
      "learning_rate": 0.00028006298753265475,
      "loss": 2.39,
      "step": 120079
    },
    {
      "epoch": 1.56,
      "grad_norm": 1.0473589897155762,
      "learning_rate": 0.00028005890601530376,
      "loss": 2.1452,
      "step": 120080
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.9510626196861267,
      "learning_rate": 0.0002800548245016602,
      "loss": 1.8026,
      "step": 120081
    },
    {
      "epoch": 1.56,
      "grad_norm": 1.0771583318710327,
      "learning_rate": 0.00028005074299172484,
      "loss": 2.6173,
      "step": 120082
    },
    {
      "epoch": 1.56,
      "grad_norm": 1.061928153038025,
      "learning_rate": 0.00028004666148549837,
      "loss": 2.0751,
      "step": 120083
    },
    {
      "epoch": 1.56,
      "grad_norm": 1.0673394203186035,
      "learning_rate": 0.0002800425799829816,
      "loss": 2.4465,
      "step": 120084
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.9693037271499634,
      "learning_rate": 0.0002800384984841753,
      "loss": 2.2033,
      "step": 120085
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.9840269088745117,
      "learning_rate": 0.0002800344169890803,
      "loss": 2.0986,
      "step": 120086
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.9728217720985413,
      "learning_rate": 0.0002800303354976973,
      "loss": 2.241,
      "step": 120087
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.9301008582115173,
      "learning_rate": 0.00028002625401002694,
      "loss": 2.1258,
      "step": 120088
    },
    {
      "epoch": 1.56,
      "grad_norm": 1.0003855228424072,
      "learning_rate": 0.00028002217252607003,
      "loss": 1.9809,
      "step": 120089
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.9893296360969543,
      "learning_rate": 0.00028001809104582744,
      "loss": 2.098,
      "step": 120090
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.9731530547142029,
      "learning_rate": 0.00028001400956929987,
      "loss": 2.3026,
      "step": 120091
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.9971716403961182,
      "learning_rate": 0.000280009928096488,
      "loss": 2.2343,
      "step": 120092
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.9691160917282104,
      "learning_rate": 0.0002800058466273929,
      "loss": 2.2295,
      "step": 120093
    },
    {
      "epoch": 1.56,
      "grad_norm": 1.0096590518951416,
      "learning_rate": 0.0002800017651620149,
      "loss": 2.3908,
      "step": 120094
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.9644815921783447,
      "learning_rate": 0.00027999768370035494,
      "loss": 2.1338,
      "step": 120095
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.8809234499931335,
      "learning_rate": 0.0002799936022424138,
      "loss": 1.9855,
      "step": 120096
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.9446083903312683,
      "learning_rate": 0.00027998952078819227,
      "loss": 2.1633,
      "step": 120097
    },
    {
      "epoch": 1.56,
      "grad_norm": 1.0344593524932861,
      "learning_rate": 0.000279985439337691,
      "loss": 1.9526,
      "step": 120098
    },
    {
      "epoch": 1.56,
      "grad_norm": 1.0624920129776,
      "learning_rate": 0.000279981357890911,
      "loss": 2.1088,
      "step": 120099
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.9941631555557251,
      "learning_rate": 0.0002799772764478526,
      "loss": 2.3646,
      "step": 120100
    },
    {
      "epoch": 1.56,
      "grad_norm": 1.1789013147354126,
      "learning_rate": 0.0002799731950085169,
      "loss": 2.1946,
      "step": 120101
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.9755827188491821,
      "learning_rate": 0.0002799691135729046,
      "loss": 2.1983,
      "step": 120102
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.9510272741317749,
      "learning_rate": 0.0002799650321410164,
      "loss": 2.1786,
      "step": 120103
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.9709486365318298,
      "learning_rate": 0.000279960950712853,
      "loss": 2.05,
      "step": 120104
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.9778656959533691,
      "learning_rate": 0.00027995686928841543,
      "loss": 2.0785,
      "step": 120105
    },
    {
      "epoch": 1.56,
      "grad_norm": 1.008855938911438,
      "learning_rate": 0.00027995278786770405,
      "loss": 2.0864,
      "step": 120106
    },
    {
      "epoch": 1.56,
      "grad_norm": 1.0180801153182983,
      "learning_rate": 0.00027994870645071994,
      "loss": 1.9928,
      "step": 120107
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.9381588101387024,
      "learning_rate": 0.00027994462503746364,
      "loss": 1.8487,
      "step": 120108
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.8948217034339905,
      "learning_rate": 0.0002799405436279361,
      "loss": 2.1253,
      "step": 120109
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.9339077472686768,
      "learning_rate": 0.00027993646222213795,
      "loss": 2.0037,
      "step": 120110
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.935128927230835,
      "learning_rate": 0.00027993238082006997,
      "loss": 2.2568,
      "step": 120111
    },
    {
      "epoch": 1.56,
      "grad_norm": 1.063403606414795,
      "learning_rate": 0.0002799282994217331,
      "loss": 2.0067,
      "step": 120112
    },
    {
      "epoch": 1.56,
      "grad_norm": 1.001686453819275,
      "learning_rate": 0.00027992421802712784,
      "loss": 2.2939,
      "step": 120113
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.9043030142784119,
      "learning_rate": 0.000279920136636255,
      "loss": 1.967,
      "step": 120114
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.9979897737503052,
      "learning_rate": 0.0002799160552491154,
      "loss": 2.1254,
      "step": 120115
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.943513810634613,
      "learning_rate": 0.00027991197386570974,
      "loss": 2.2821,
      "step": 120116
    },
    {
      "epoch": 1.56,
      "grad_norm": 1.0269114971160889,
      "learning_rate": 0.0002799078924860389,
      "loss": 2.1555,
      "step": 120117
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.9451582431793213,
      "learning_rate": 0.00027990381111010364,
      "loss": 2.2078,
      "step": 120118
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.9708096385002136,
      "learning_rate": 0.0002798997297379045,
      "loss": 2.1766,
      "step": 120119
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.9137135148048401,
      "learning_rate": 0.00027989564836944245,
      "loss": 2.2595,
      "step": 120120
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.9842253923416138,
      "learning_rate": 0.00027989156700471813,
      "loss": 2.2597,
      "step": 120121
    },
    {
      "epoch": 1.56,
      "grad_norm": 1.0665432214736938,
      "learning_rate": 0.00027988748564373235,
      "loss": 1.9926,
      "step": 120122
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.9094269871711731,
      "learning_rate": 0.00027988340428648583,
      "loss": 2.2331,
      "step": 120123
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.9295357465744019,
      "learning_rate": 0.00027987932293297953,
      "loss": 2.0426,
      "step": 120124
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.9545143246650696,
      "learning_rate": 0.00027987524158321395,
      "loss": 2.1235,
      "step": 120125
    },
    {
      "epoch": 1.56,
      "grad_norm": 1.0687390565872192,
      "learning_rate": 0.0002798711602371899,
      "loss": 2.1462,
      "step": 120126
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.862043023109436,
      "learning_rate": 0.0002798670788949081,
      "loss": 2.2654,
      "step": 120127
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.9640758633613586,
      "learning_rate": 0.00027986299755636957,
      "loss": 2.2902,
      "step": 120128
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.9476861953735352,
      "learning_rate": 0.00027985891622157475,
      "loss": 2.3451,
      "step": 120129
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.88509202003479,
      "learning_rate": 0.0002798548348905247,
      "loss": 1.9763,
      "step": 120130
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.9675496816635132,
      "learning_rate": 0.0002798507535632199,
      "loss": 1.9526,
      "step": 120131
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.8929679989814758,
      "learning_rate": 0.0002798466722396612,
      "loss": 2.2375,
      "step": 120132
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.8999820947647095,
      "learning_rate": 0.00027984259091984944,
      "loss": 2.3535,
      "step": 120133
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.9611527323722839,
      "learning_rate": 0.0002798385096037853,
      "loss": 2.3023,
      "step": 120134
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.9866471886634827,
      "learning_rate": 0.0002798344282914696,
      "loss": 2.1139,
      "step": 120135
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.9844390749931335,
      "learning_rate": 0.0002798303469829031,
      "loss": 2.2429,
      "step": 120136
    },
    {
      "epoch": 1.56,
      "grad_norm": 1.0529441833496094,
      "learning_rate": 0.0002798262656780864,
      "loss": 2.0182,
      "step": 120137
    },
    {
      "epoch": 1.56,
      "grad_norm": 1.0196819305419922,
      "learning_rate": 0.0002798221843770205,
      "loss": 2.2158,
      "step": 120138
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.9486292600631714,
      "learning_rate": 0.00027981810307970594,
      "loss": 1.8453,
      "step": 120139
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.934859573841095,
      "learning_rate": 0.00027981402178614356,
      "loss": 1.962,
      "step": 120140
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.8940964937210083,
      "learning_rate": 0.00027980994049633414,
      "loss": 2.0348,
      "step": 120141
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.9573166370391846,
      "learning_rate": 0.0002798058592102786,
      "loss": 2.0312,
      "step": 120142
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.9577388167381287,
      "learning_rate": 0.0002798017779279774,
      "loss": 2.2095,
      "step": 120143
    },
    {
      "epoch": 1.56,
      "grad_norm": 1.0918710231781006,
      "learning_rate": 0.00027979769664943137,
      "loss": 2.25,
      "step": 120144
    },
    {
      "epoch": 1.56,
      "grad_norm": 1.0061314105987549,
      "learning_rate": 0.0002797936153746414,
      "loss": 2.0371,
      "step": 120145
    },
    {
      "epoch": 1.56,
      "grad_norm": 1.1520297527313232,
      "learning_rate": 0.00027978953410360814,
      "loss": 2.2585,
      "step": 120146
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.9660860300064087,
      "learning_rate": 0.0002797854528363324,
      "loss": 2.0476,
      "step": 120147
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.9410167336463928,
      "learning_rate": 0.000279781371572815,
      "loss": 1.879,
      "step": 120148
    },
    {
      "epoch": 1.56,
      "grad_norm": 1.0299713611602783,
      "learning_rate": 0.0002797772903130566,
      "loss": 1.9149,
      "step": 120149
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.9189203381538391,
      "learning_rate": 0.00027977320905705794,
      "loss": 2.1265,
      "step": 120150
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.9661014080047607,
      "learning_rate": 0.0002797691278048198,
      "loss": 2.2306,
      "step": 120151
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.9795495271682739,
      "learning_rate": 0.000279765046556343,
      "loss": 2.147,
      "step": 120152
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.9346360564231873,
      "learning_rate": 0.00027976096531162826,
      "loss": 2.1254,
      "step": 120153
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.9277589917182922,
      "learning_rate": 0.0002797568840706764,
      "loss": 2.2947,
      "step": 120154
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.9554392695426941,
      "learning_rate": 0.000279752802833488,
      "loss": 1.8081,
      "step": 120155
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.9639025926589966,
      "learning_rate": 0.000279748721600064,
      "loss": 2.0626,
      "step": 120156
    },
    {
      "epoch": 1.56,
      "grad_norm": 1.168616533279419,
      "learning_rate": 0.0002797446403704051,
      "loss": 2.0958,
      "step": 120157
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.9020374417304993,
      "learning_rate": 0.000279740559144512,
      "loss": 2.0696,
      "step": 120158
    },
    {
      "epoch": 1.56,
      "grad_norm": 1.2720197439193726,
      "learning_rate": 0.0002797364779223855,
      "loss": 2.0027,
      "step": 120159
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.9960588216781616,
      "learning_rate": 0.0002797323967040265,
      "loss": 2.2359,
      "step": 120160
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.90679532289505,
      "learning_rate": 0.00027972831548943555,
      "loss": 2.1028,
      "step": 120161
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.979978084564209,
      "learning_rate": 0.00027972423427861346,
      "loss": 2.0987,
      "step": 120162
    },
    {
      "epoch": 1.56,
      "grad_norm": 1.1259256601333618,
      "learning_rate": 0.0002797201530715611,
      "loss": 2.1989,
      "step": 120163
    },
    {
      "epoch": 1.56,
      "grad_norm": 1.0740667581558228,
      "learning_rate": 0.00027971607186827907,
      "loss": 1.9391,
      "step": 120164
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.9063177704811096,
      "learning_rate": 0.0002797119906687682,
      "loss": 2.1028,
      "step": 120165
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.9233872294425964,
      "learning_rate": 0.00027970790947302935,
      "loss": 1.9544,
      "step": 120166
    },
    {
      "epoch": 1.56,
      "grad_norm": 1.2792470455169678,
      "learning_rate": 0.00027970382828106306,
      "loss": 2.1384,
      "step": 120167
    },
    {
      "epoch": 1.56,
      "grad_norm": 1.4072637557983398,
      "learning_rate": 0.00027969974709287027,
      "loss": 2.1924,
      "step": 120168
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.9012608528137207,
      "learning_rate": 0.00027969566590845164,
      "loss": 2.0719,
      "step": 120169
    },
    {
      "epoch": 1.56,
      "grad_norm": 1.0648707151412964,
      "learning_rate": 0.000279691584727808,
      "loss": 2.1818,
      "step": 120170
    },
    {
      "epoch": 1.56,
      "grad_norm": 1.7506029605865479,
      "learning_rate": 0.00027968750355094013,
      "loss": 2.2011,
      "step": 120171
    },
    {
      "epoch": 1.56,
      "grad_norm": 1.1732221841812134,
      "learning_rate": 0.0002796834223778487,
      "loss": 2.0195,
      "step": 120172
    },
    {
      "epoch": 1.56,
      "grad_norm": 1.0010745525360107,
      "learning_rate": 0.0002796793412085345,
      "loss": 2.1137,
      "step": 120173
    },
    {
      "epoch": 1.56,
      "grad_norm": 1.0099533796310425,
      "learning_rate": 0.00027967526004299826,
      "loss": 2.2099,
      "step": 120174
    },
    {
      "epoch": 1.56,
      "grad_norm": 1.0290800333023071,
      "learning_rate": 0.0002796711788812408,
      "loss": 2.2452,
      "step": 120175
    },
    {
      "epoch": 1.56,
      "grad_norm": 1.0159149169921875,
      "learning_rate": 0.00027966709772326285,
      "loss": 1.918,
      "step": 120176
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.9465543031692505,
      "learning_rate": 0.0002796630165690651,
      "loss": 2.1316,
      "step": 120177
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.8572782278060913,
      "learning_rate": 0.00027965893541864853,
      "loss": 1.8745,
      "step": 120178
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.937838613986969,
      "learning_rate": 0.00027965485427201375,
      "loss": 1.8079,
      "step": 120179
    },
    {
      "epoch": 1.56,
      "grad_norm": 1.0580414533615112,
      "learning_rate": 0.0002796507731291614,
      "loss": 2.1548,
      "step": 120180
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.9214098453521729,
      "learning_rate": 0.00027964669199009233,
      "loss": 2.2284,
      "step": 120181
    },
    {
      "epoch": 1.56,
      "grad_norm": 1.004565954208374,
      "learning_rate": 0.0002796426108548074,
      "loss": 2.2017,
      "step": 120182
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.9950354099273682,
      "learning_rate": 0.00027963852972330723,
      "loss": 2.0761,
      "step": 120183
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.934799313545227,
      "learning_rate": 0.0002796344485955927,
      "loss": 2.0576,
      "step": 120184
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.9018486142158508,
      "learning_rate": 0.0002796303674716646,
      "loss": 2.0998,
      "step": 120185
    },
    {
      "epoch": 1.56,
      "grad_norm": 1.015371561050415,
      "learning_rate": 0.0002796262863515235,
      "loss": 2.1814,
      "step": 120186
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.9259093403816223,
      "learning_rate": 0.0002796222052351702,
      "loss": 2.2183,
      "step": 120187
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.8872613310813904,
      "learning_rate": 0.0002796181241226056,
      "loss": 2.1164,
      "step": 120188
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.9524336457252502,
      "learning_rate": 0.00027961404301383037,
      "loss": 2.2652,
      "step": 120189
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.9885769486427307,
      "learning_rate": 0.0002796099619088452,
      "loss": 2.0244,
      "step": 120190
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.8981179594993591,
      "learning_rate": 0.0002796058808076511,
      "loss": 2.218,
      "step": 120191
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.9507036209106445,
      "learning_rate": 0.00027960179971024847,
      "loss": 2.2625,
      "step": 120192
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.8801564574241638,
      "learning_rate": 0.00027959771861663834,
      "loss": 2.0979,
      "step": 120193
    },
    {
      "epoch": 1.57,
      "grad_norm": 1.1263937950134277,
      "learning_rate": 0.00027959363752682134,
      "loss": 2.2829,
      "step": 120194
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.936672031879425,
      "learning_rate": 0.00027958955644079823,
      "loss": 2.0441,
      "step": 120195
    },
    {
      "epoch": 1.57,
      "grad_norm": 1.04258131980896,
      "learning_rate": 0.00027958547535856984,
      "loss": 2.0461,
      "step": 120196
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.8705717325210571,
      "learning_rate": 0.000279581394280137,
      "loss": 1.9568,
      "step": 120197
    },
    {
      "epoch": 1.57,
      "grad_norm": 1.0319170951843262,
      "learning_rate": 0.00027957731320550026,
      "loss": 2.1103,
      "step": 120198
    },
    {
      "epoch": 1.57,
      "grad_norm": 1.0263950824737549,
      "learning_rate": 0.0002795732321346605,
      "loss": 2.2908,
      "step": 120199
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.8887727856636047,
      "learning_rate": 0.00027956915106761845,
      "loss": 2.0445,
      "step": 120200
    },
    {
      "epoch": 1.57,
      "grad_norm": 1.1377569437026978,
      "learning_rate": 0.00027956507000437487,
      "loss": 2.1129,
      "step": 120201
    },
    {
      "epoch": 1.57,
      "grad_norm": 1.3138377666473389,
      "learning_rate": 0.0002795609889449305,
      "loss": 2.2118,
      "step": 120202
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.9836010932922363,
      "learning_rate": 0.0002795569078892863,
      "loss": 2.112,
      "step": 120203
    },
    {
      "epoch": 1.57,
      "grad_norm": 1.0437099933624268,
      "learning_rate": 0.0002795528268374427,
      "loss": 1.9067,
      "step": 120204
    },
    {
      "epoch": 1.57,
      "grad_norm": 1.3741953372955322,
      "learning_rate": 0.00027954874578940063,
      "loss": 2.0678,
      "step": 120205
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.9949125051498413,
      "learning_rate": 0.00027954466474516083,
      "loss": 2.1935,
      "step": 120206
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.9325588345527649,
      "learning_rate": 0.0002795405837047241,
      "loss": 1.9542,
      "step": 120207
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.9820228219032288,
      "learning_rate": 0.0002795365026680911,
      "loss": 2.197,
      "step": 120208
    },
    {
      "epoch": 1.57,
      "grad_norm": 1.0410395860671997,
      "learning_rate": 0.0002795324216352628,
      "loss": 2.0436,
      "step": 120209
    },
    {
      "epoch": 1.57,
      "grad_norm": 1.568385362625122,
      "learning_rate": 0.0002795283406062397,
      "loss": 2.0908,
      "step": 120210
    },
    {
      "epoch": 1.57,
      "grad_norm": 1.0400433540344238,
      "learning_rate": 0.00027952425958102265,
      "loss": 2.2147,
      "step": 120211
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.9687600135803223,
      "learning_rate": 0.00027952017855961247,
      "loss": 2.28,
      "step": 120212
    },
    {
      "epoch": 1.57,
      "grad_norm": 1.0683523416519165,
      "learning_rate": 0.0002795160975420098,
      "loss": 2.2141,
      "step": 120213
    },
    {
      "epoch": 1.57,
      "grad_norm": 1.0858101844787598,
      "learning_rate": 0.0002795120165282155,
      "loss": 2.3527,
      "step": 120214
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.9960414171218872,
      "learning_rate": 0.0002795079355182305,
      "loss": 2.1956,
      "step": 120215
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.9671120047569275,
      "learning_rate": 0.00027950385451205515,
      "loss": 2.1891,
      "step": 120216
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.9471256732940674,
      "learning_rate": 0.00027949977350969044,
      "loss": 1.9028,
      "step": 120217
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.8704490661621094,
      "learning_rate": 0.00027949569251113713,
      "loss": 2.0457,
      "step": 120218
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.8946612477302551,
      "learning_rate": 0.00027949161151639594,
      "loss": 2.2851,
      "step": 120219
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.8761657476425171,
      "learning_rate": 0.00027948753052546766,
      "loss": 2.3076,
      "step": 120220
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.9129531383514404,
      "learning_rate": 0.0002794834495383531,
      "loss": 2.2613,
      "step": 120221
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.9839380383491516,
      "learning_rate": 0.0002794793685550529,
      "loss": 2.2505,
      "step": 120222
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.967449426651001,
      "learning_rate": 0.0002794752875755679,
      "loss": 2.247,
      "step": 120223
    },
    {
      "epoch": 1.57,
      "grad_norm": 1.0764638185501099,
      "learning_rate": 0.00027947120659989875,
      "loss": 2.2266,
      "step": 120224
    },
    {
      "epoch": 1.57,
      "grad_norm": 1.0408929586410522,
      "learning_rate": 0.00027946712562804636,
      "loss": 2.0854,
      "step": 120225
    },
    {
      "epoch": 1.57,
      "grad_norm": 1.00667142868042,
      "learning_rate": 0.00027946304466001134,
      "loss": 2.1613,
      "step": 120226
    },
    {
      "epoch": 1.57,
      "grad_norm": 1.0873298645019531,
      "learning_rate": 0.0002794589636957947,
      "loss": 2.16,
      "step": 120227
    },
    {
      "epoch": 1.57,
      "grad_norm": 1.084233283996582,
      "learning_rate": 0.00027945488273539693,
      "loss": 2.3405,
      "step": 120228
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.9293655753135681,
      "learning_rate": 0.00027945080177881884,
      "loss": 2.3534,
      "step": 120229
    },
    {
      "epoch": 1.57,
      "grad_norm": 1.1300861835479736,
      "learning_rate": 0.00027944672082606126,
      "loss": 2.2226,
      "step": 120230
    },
    {
      "epoch": 1.57,
      "grad_norm": 1.0640854835510254,
      "learning_rate": 0.0002794426398771249,
      "loss": 2.2083,
      "step": 120231
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.940428614616394,
      "learning_rate": 0.00027943855893201056,
      "loss": 2.1459,
      "step": 120232
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.9568925499916077,
      "learning_rate": 0.000279434477990719,
      "loss": 2.1834,
      "step": 120233
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.9217305183410645,
      "learning_rate": 0.000279430397053251,
      "loss": 1.913,
      "step": 120234
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.9660748839378357,
      "learning_rate": 0.00027942631611960726,
      "loss": 2.1533,
      "step": 120235
    },
    {
      "epoch": 1.57,
      "grad_norm": 1.0023486614227295,
      "learning_rate": 0.00027942223518978847,
      "loss": 2.0828,
      "step": 120236
    },
    {
      "epoch": 1.57,
      "grad_norm": 1.0633955001831055,
      "learning_rate": 0.0002794181542637955,
      "loss": 2.1498,
      "step": 120237
    },
    {
      "epoch": 1.57,
      "grad_norm": 1.043422818183899,
      "learning_rate": 0.0002794140733416291,
      "loss": 2.1795,
      "step": 120238
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.9183328747749329,
      "learning_rate": 0.00027940999242329,
      "loss": 1.8344,
      "step": 120239
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.8493395447731018,
      "learning_rate": 0.00027940591150877903,
      "loss": 2.1317,
      "step": 120240
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.9315245151519775,
      "learning_rate": 0.0002794018305980969,
      "loss": 2.0009,
      "step": 120241
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.9580580592155457,
      "learning_rate": 0.00027939774969124433,
      "loss": 2.2643,
      "step": 120242
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.9252626895904541,
      "learning_rate": 0.00027939366878822203,
      "loss": 1.9646,
      "step": 120243
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.8679265379905701,
      "learning_rate": 0.0002793895878890309,
      "loss": 2.1336,
      "step": 120244
    },
    {
      "epoch": 1.57,
      "grad_norm": 1.1385838985443115,
      "learning_rate": 0.0002793855069936716,
      "loss": 2.1638,
      "step": 120245
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.9633471369743347,
      "learning_rate": 0.000279381426102145,
      "loss": 2.0626,
      "step": 120246
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.9012900590896606,
      "learning_rate": 0.00027937734521445173,
      "loss": 2.0792,
      "step": 120247
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.9455227255821228,
      "learning_rate": 0.00027937326433059265,
      "loss": 2.1203,
      "step": 120248
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.9314751029014587,
      "learning_rate": 0.0002793691834505684,
      "loss": 1.872,
      "step": 120249
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.944157063961029,
      "learning_rate": 0.0002793651025743798,
      "loss": 2.2143,
      "step": 120250
    },
    {
      "epoch": 1.57,
      "grad_norm": 1.0587975978851318,
      "learning_rate": 0.00027936102170202765,
      "loss": 2.204,
      "step": 120251
    },
    {
      "epoch": 1.57,
      "grad_norm": 1.089913010597229,
      "learning_rate": 0.00027935694083351274,
      "loss": 2.2923,
      "step": 120252
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.9781005382537842,
      "learning_rate": 0.00027935285996883564,
      "loss": 2.106,
      "step": 120253
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.9041666388511658,
      "learning_rate": 0.0002793487791079973,
      "loss": 2.0372,
      "step": 120254
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.9176180362701416,
      "learning_rate": 0.0002793446982509984,
      "loss": 2.2201,
      "step": 120255
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.9996387958526611,
      "learning_rate": 0.0002793406173978397,
      "loss": 2.1644,
      "step": 120256
    },
    {
      "epoch": 1.57,
      "grad_norm": 1.0066583156585693,
      "learning_rate": 0.000279336536548522,
      "loss": 2.2498,
      "step": 120257
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.9294213056564331,
      "learning_rate": 0.0002793324557030461,
      "loss": 2.268,
      "step": 120258
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.9869166016578674,
      "learning_rate": 0.00027932837486141255,
      "loss": 2.1158,
      "step": 120259
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.9379017949104309,
      "learning_rate": 0.00027932429402362225,
      "loss": 2.1926,
      "step": 120260
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.8676673769950867,
      "learning_rate": 0.000279320213189676,
      "loss": 2.1471,
      "step": 120261
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.8755947947502136,
      "learning_rate": 0.0002793161323595745,
      "loss": 2.2015,
      "step": 120262
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.9278293251991272,
      "learning_rate": 0.00027931205153331856,
      "loss": 2.3209,
      "step": 120263
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.9748730063438416,
      "learning_rate": 0.0002793079707109089,
      "loss": 2.2637,
      "step": 120264
    },
    {
      "epoch": 1.57,
      "grad_norm": 1.0326972007751465,
      "learning_rate": 0.00027930388989234623,
      "loss": 2.1982,
      "step": 120265
    },
    {
      "epoch": 1.57,
      "grad_norm": 1.0331512689590454,
      "learning_rate": 0.0002792998090776314,
      "loss": 2.1367,
      "step": 120266
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.9227837920188904,
      "learning_rate": 0.0002792957282667651,
      "loss": 1.9992,
      "step": 120267
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.9704419374465942,
      "learning_rate": 0.00027929164745974807,
      "loss": 2.1509,
      "step": 120268
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.9499444961547852,
      "learning_rate": 0.00027928756665658117,
      "loss": 2.2868,
      "step": 120269
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.9423490762710571,
      "learning_rate": 0.0002792834858572652,
      "loss": 2.1796,
      "step": 120270
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.9792800545692444,
      "learning_rate": 0.0002792794050618007,
      "loss": 2.1527,
      "step": 120271
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.9147226214408875,
      "learning_rate": 0.00027927532427018855,
      "loss": 2.3465,
      "step": 120272
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.9496100544929504,
      "learning_rate": 0.00027927124348242947,
      "loss": 2.1333,
      "step": 120273
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.9082207083702087,
      "learning_rate": 0.00027926716269852434,
      "loss": 1.9538,
      "step": 120274
    },
    {
      "epoch": 1.57,
      "grad_norm": 1.002482295036316,
      "learning_rate": 0.00027926308191847377,
      "loss": 2.1463,
      "step": 120275
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.9296731948852539,
      "learning_rate": 0.0002792590011422787,
      "loss": 2.2849,
      "step": 120276
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.9195210337638855,
      "learning_rate": 0.0002792549203699397,
      "loss": 2.2291,
      "step": 120277
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.9706940054893494,
      "learning_rate": 0.00027925083960145757,
      "loss": 2.1164,
      "step": 120278
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.9930098652839661,
      "learning_rate": 0.00027924675883683314,
      "loss": 2.1492,
      "step": 120279
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.9117230176925659,
      "learning_rate": 0.0002792426780760671,
      "loss": 1.9146,
      "step": 120280
    },
    {
      "epoch": 1.57,
      "grad_norm": 1.1870596408843994,
      "learning_rate": 0.00027923859731916024,
      "loss": 2.0523,
      "step": 120281
    },
    {
      "epoch": 1.57,
      "grad_norm": 1.2505507469177246,
      "learning_rate": 0.00027923451656611346,
      "loss": 2.2265,
      "step": 120282
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.9560773968696594,
      "learning_rate": 0.0002792304358169272,
      "loss": 2.1178,
      "step": 120283
    },
    {
      "epoch": 1.57,
      "grad_norm": 1.0734748840332031,
      "learning_rate": 0.0002792263550716024,
      "loss": 2.2066,
      "step": 120284
    },
    {
      "epoch": 1.57,
      "grad_norm": 1.588746190071106,
      "learning_rate": 0.0002792222743301399,
      "loss": 2.0717,
      "step": 120285
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.8678233027458191,
      "learning_rate": 0.0002792181935925403,
      "loss": 2.1848,
      "step": 120286
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.9576897621154785,
      "learning_rate": 0.00027921411285880446,
      "loss": 2.0549,
      "step": 120287
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.9136265516281128,
      "learning_rate": 0.0002792100321289332,
      "loss": 2.0781,
      "step": 120288
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.9374662041664124,
      "learning_rate": 0.000279205951402927,
      "loss": 2.1614,
      "step": 120289
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.9963943958282471,
      "learning_rate": 0.00027920187068078693,
      "loss": 2.1869,
      "step": 120290
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.9250897169113159,
      "learning_rate": 0.0002791977899625136,
      "loss": 2.1893,
      "step": 120291
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.9740293622016907,
      "learning_rate": 0.00027919370924810774,
      "loss": 2.2077,
      "step": 120292
    },
    {
      "epoch": 1.57,
      "grad_norm": 1.104243516921997,
      "learning_rate": 0.0002791896285375702,
      "loss": 2.2091,
      "step": 120293
    },
    {
      "epoch": 1.57,
      "grad_norm": 1.1432747840881348,
      "learning_rate": 0.00027918554783090184,
      "loss": 2.137,
      "step": 120294
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.982134997844696,
      "learning_rate": 0.0002791814671281031,
      "loss": 2.255,
      "step": 120295
    },
    {
      "epoch": 1.57,
      "grad_norm": 1.1098697185516357,
      "learning_rate": 0.00027917738642917493,
      "loss": 2.2731,
      "step": 120296
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.9109979271888733,
      "learning_rate": 0.0002791733057341181,
      "loss": 1.8461,
      "step": 120297
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.940065324306488,
      "learning_rate": 0.00027916922504293335,
      "loss": 2.2802,
      "step": 120298
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.9485563635826111,
      "learning_rate": 0.0002791651443556214,
      "loss": 2.3265,
      "step": 120299
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.8777114748954773,
      "learning_rate": 0.00027916106367218324,
      "loss": 2.1187,
      "step": 120300
    },
    {
      "epoch": 1.57,
      "grad_norm": 1.0050649642944336,
      "learning_rate": 0.00027915698299261924,
      "loss": 2.1967,
      "step": 120301
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.9207688570022583,
      "learning_rate": 0.0002791529023169304,
      "loss": 2.1273,
      "step": 120302
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.9070030450820923,
      "learning_rate": 0.0002791488216451174,
      "loss": 1.9136,
      "step": 120303
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.9171714186668396,
      "learning_rate": 0.00027914474097718103,
      "loss": 2.1414,
      "step": 120304
    },
    {
      "epoch": 1.57,
      "grad_norm": 1.1473416090011597,
      "learning_rate": 0.00027914066031312204,
      "loss": 2.1761,
      "step": 120305
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.973452627658844,
      "learning_rate": 0.0002791365796529412,
      "loss": 2.1411,
      "step": 120306
    },
    {
      "epoch": 1.57,
      "grad_norm": 1.06269371509552,
      "learning_rate": 0.0002791324989966394,
      "loss": 2.1924,
      "step": 120307
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.9676080942153931,
      "learning_rate": 0.00027912841834421714,
      "loss": 2.1108,
      "step": 120308
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.9139099717140198,
      "learning_rate": 0.00027912433769567525,
      "loss": 2.0922,
      "step": 120309
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.9521083235740662,
      "learning_rate": 0.0002791202570510146,
      "loss": 2.126,
      "step": 120310
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.8438709378242493,
      "learning_rate": 0.00027911617641023585,
      "loss": 2.0393,
      "step": 120311
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.9452782273292542,
      "learning_rate": 0.0002791120957733398,
      "loss": 2.1642,
      "step": 120312
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.8855183124542236,
      "learning_rate": 0.00027910801514032733,
      "loss": 2.027,
      "step": 120313
    },
    {
      "epoch": 1.57,
      "grad_norm": 1.0275429487228394,
      "learning_rate": 0.000279103934511199,
      "loss": 2.0576,
      "step": 120314
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.9457311034202576,
      "learning_rate": 0.0002790998538859556,
      "loss": 2.137,
      "step": 120315
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.9338716864585876,
      "learning_rate": 0.00027909577326459795,
      "loss": 2.0944,
      "step": 120316
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.8663124442100525,
      "learning_rate": 0.00027909169264712675,
      "loss": 2.1543,
      "step": 120317
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.8558922410011292,
      "learning_rate": 0.00027908761203354284,
      "loss": 1.901,
      "step": 120318
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.9776946902275085,
      "learning_rate": 0.00027908353142384705,
      "loss": 2.3353,
      "step": 120319
    },
    {
      "epoch": 1.57,
      "grad_norm": 1.0833849906921387,
      "learning_rate": 0.0002790794508180399,
      "loss": 2.1115,
      "step": 120320
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.9603427648544312,
      "learning_rate": 0.0002790753702161223,
      "loss": 2.2197,
      "step": 120321
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.9303795695304871,
      "learning_rate": 0.00027907128961809495,
      "loss": 2.2161,
      "step": 120322
    },
    {
      "epoch": 1.57,
      "grad_norm": 1.2363260984420776,
      "learning_rate": 0.00027906720902395865,
      "loss": 1.9723,
      "step": 120323
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.9201053977012634,
      "learning_rate": 0.00027906312843371415,
      "loss": 2.1322,
      "step": 120324
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.9191165566444397,
      "learning_rate": 0.00027905904784736226,
      "loss": 2.1556,
      "step": 120325
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.9552981853485107,
      "learning_rate": 0.0002790549672649037,
      "loss": 2.2448,
      "step": 120326
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.8369309902191162,
      "learning_rate": 0.0002790508866863392,
      "loss": 2.2916,
      "step": 120327
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.9197072386741638,
      "learning_rate": 0.00027904680611166947,
      "loss": 2.097,
      "step": 120328
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.9912410974502563,
      "learning_rate": 0.00027904272554089534,
      "loss": 2.2234,
      "step": 120329
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.955650269985199,
      "learning_rate": 0.00027903864497401756,
      "loss": 1.9728,
      "step": 120330
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.9388877749443054,
      "learning_rate": 0.000279034564411037,
      "loss": 2.0545,
      "step": 120331
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.9147383570671082,
      "learning_rate": 0.0002790304838519542,
      "loss": 2.141,
      "step": 120332
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.8882085084915161,
      "learning_rate": 0.0002790264032967701,
      "loss": 2.0467,
      "step": 120333
    },
    {
      "epoch": 1.57,
      "grad_norm": 1.2171423435211182,
      "learning_rate": 0.00027902232274548537,
      "loss": 2.2683,
      "step": 120334
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.9542964100837708,
      "learning_rate": 0.00027901824219810073,
      "loss": 2.0001,
      "step": 120335
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.9468547701835632,
      "learning_rate": 0.00027901416165461706,
      "loss": 2.0361,
      "step": 120336
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.9934110045433044,
      "learning_rate": 0.00027901008111503504,
      "loss": 2.2756,
      "step": 120337
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.905860424041748,
      "learning_rate": 0.0002790060005793554,
      "loss": 2.1179,
      "step": 120338
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.9763659238815308,
      "learning_rate": 0.000279001920047579,
      "loss": 2.0951,
      "step": 120339
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.9844978451728821,
      "learning_rate": 0.00027899783951970653,
      "loss": 2.1507,
      "step": 120340
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.9443506002426147,
      "learning_rate": 0.0002789937589957387,
      "loss": 2.1465,
      "step": 120341
    },
    {
      "epoch": 1.57,
      "grad_norm": 1.0263768434524536,
      "learning_rate": 0.0002789896784756764,
      "loss": 2.1531,
      "step": 120342
    },
    {
      "epoch": 1.57,
      "grad_norm": 1.0049678087234497,
      "learning_rate": 0.00027898559795952026,
      "loss": 1.8902,
      "step": 120343
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.9982435703277588,
      "learning_rate": 0.00027898151744727115,
      "loss": 2.1389,
      "step": 120344
    },
    {
      "epoch": 1.57,
      "grad_norm": 1.1400483846664429,
      "learning_rate": 0.0002789774369389297,
      "loss": 2.3293,
      "step": 120345
    },
    {
      "epoch": 1.57,
      "grad_norm": 1.0201783180236816,
      "learning_rate": 0.0002789733564344968,
      "loss": 2.0436,
      "step": 120346
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.8829001784324646,
      "learning_rate": 0.00027896927593397316,
      "loss": 2.3207,
      "step": 120347
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.894558310508728,
      "learning_rate": 0.0002789651954373595,
      "loss": 1.9849,
      "step": 120348
    },
    {
      "epoch": 1.57,
      "grad_norm": 1.1993656158447266,
      "learning_rate": 0.00027896111494465665,
      "loss": 2.0813,
      "step": 120349
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.9555044174194336,
      "learning_rate": 0.0002789570344558652,
      "loss": 2.1027,
      "step": 120350
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.9819027781486511,
      "learning_rate": 0.00027895295397098614,
      "loss": 2.1303,
      "step": 120351
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.8383715748786926,
      "learning_rate": 0.00027894887349002006,
      "loss": 2.1636,
      "step": 120352
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.9834023714065552,
      "learning_rate": 0.00027894479301296784,
      "loss": 2.0814,
      "step": 120353
    },
    {
      "epoch": 1.57,
      "grad_norm": 1.2805507183074951,
      "learning_rate": 0.0002789407125398301,
      "loss": 2.2161,
      "step": 120354
    },
    {
      "epoch": 1.57,
      "grad_norm": 1.2471046447753906,
      "learning_rate": 0.0002789366320706079,
      "loss": 2.0079,
      "step": 120355
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.9201034903526306,
      "learning_rate": 0.0002789325516053016,
      "loss": 2.2362,
      "step": 120356
    },
    {
      "epoch": 1.57,
      "grad_norm": 1.0146477222442627,
      "learning_rate": 0.0002789284711439121,
      "loss": 2.1116,
      "step": 120357
    },
    {
      "epoch": 1.57,
      "grad_norm": 1.2633246183395386,
      "learning_rate": 0.00027892439068644023,
      "loss": 1.9909,
      "step": 120358
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.978317141532898,
      "learning_rate": 0.00027892031023288674,
      "loss": 2.1822,
      "step": 120359
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.887622058391571,
      "learning_rate": 0.0002789162297832523,
      "loss": 2.125,
      "step": 120360
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.9585250020027161,
      "learning_rate": 0.0002789121493375379,
      "loss": 2.2726,
      "step": 120361
    },
    {
      "epoch": 1.57,
      "grad_norm": 1.2762978076934814,
      "learning_rate": 0.00027890806889574406,
      "loss": 2.0169,
      "step": 120362
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.9139206409454346,
      "learning_rate": 0.0002789039884578715,
      "loss": 2.175,
      "step": 120363
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.9667105674743652,
      "learning_rate": 0.0002788999080239211,
      "loss": 2.0951,
      "step": 120364
    },
    {
      "epoch": 1.57,
      "grad_norm": 1.0134142637252808,
      "learning_rate": 0.0002788958275938937,
      "loss": 2.0595,
      "step": 120365
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.9359126091003418,
      "learning_rate": 0.0002788917471677899,
      "loss": 1.8733,
      "step": 120366
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.9254142642021179,
      "learning_rate": 0.0002788876667456106,
      "loss": 2.1044,
      "step": 120367
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.901083767414093,
      "learning_rate": 0.00027888358632735646,
      "loss": 1.9439,
      "step": 120368
    },
    {
      "epoch": 1.57,
      "grad_norm": 1.0071070194244385,
      "learning_rate": 0.0002788795059130282,
      "loss": 2.0217,
      "step": 120369
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.9157988429069519,
      "learning_rate": 0.0002788754255026266,
      "loss": 2.1009,
      "step": 120370
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.9538575410842896,
      "learning_rate": 0.00027887134509615253,
      "loss": 2.189,
      "step": 120371
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.9047042727470398,
      "learning_rate": 0.00027886726469360663,
      "loss": 1.9047,
      "step": 120372
    },
    {
      "epoch": 1.57,
      "grad_norm": 1.0838220119476318,
      "learning_rate": 0.00027886318429498977,
      "loss": 2.2884,
      "step": 120373
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.9677094221115112,
      "learning_rate": 0.0002788591039003027,
      "loss": 2.078,
      "step": 120374
    },
    {
      "epoch": 1.57,
      "grad_norm": 1.0375909805297852,
      "learning_rate": 0.00027885502350954603,
      "loss": 2.1616,
      "step": 120375
    },
    {
      "epoch": 1.57,
      "grad_norm": 1.1266025304794312,
      "learning_rate": 0.0002788509431227206,
      "loss": 2.1726,
      "step": 120376
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.9075213074684143,
      "learning_rate": 0.00027884686273982714,
      "loss": 2.0905,
      "step": 120377
    },
    {
      "epoch": 1.57,
      "grad_norm": 1.1108590364456177,
      "learning_rate": 0.00027884278236086653,
      "loss": 2.2744,
      "step": 120378
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.9687672257423401,
      "learning_rate": 0.00027883870198583936,
      "loss": 2.0515,
      "step": 120379
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.9211063385009766,
      "learning_rate": 0.0002788346216147467,
      "loss": 2.0074,
      "step": 120380
    },
    {
      "epoch": 1.57,
      "grad_norm": 1.0031428337097168,
      "learning_rate": 0.00027883054124758883,
      "loss": 2.0935,
      "step": 120381
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.868415117263794,
      "learning_rate": 0.00027882646088436683,
      "loss": 2.078,
      "step": 120382
    },
    {
      "epoch": 1.57,
      "grad_norm": 1.026013970375061,
      "learning_rate": 0.0002788223805250814,
      "loss": 2.1745,
      "step": 120383
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.9994754791259766,
      "learning_rate": 0.0002788183001697333,
      "loss": 1.9974,
      "step": 120384
    },
    {
      "epoch": 1.57,
      "grad_norm": 1.1055784225463867,
      "learning_rate": 0.00027881421981832327,
      "loss": 2.1356,
      "step": 120385
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.8613630533218384,
      "learning_rate": 0.0002788101394708522,
      "loss": 1.9633,
      "step": 120386
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.8915936350822449,
      "learning_rate": 0.00027880605912732054,
      "loss": 2.1078,
      "step": 120387
    },
    {
      "epoch": 1.57,
      "grad_norm": 1.098799705505371,
      "learning_rate": 0.0002788019787877293,
      "loss": 2.0211,
      "step": 120388
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.9894152879714966,
      "learning_rate": 0.0002787978984520792,
      "loss": 2.2611,
      "step": 120389
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.9950690865516663,
      "learning_rate": 0.00027879381812037087,
      "loss": 2.2929,
      "step": 120390
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.8550260663032532,
      "learning_rate": 0.00027878973779260527,
      "loss": 2.0691,
      "step": 120391
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.8937429785728455,
      "learning_rate": 0.0002787856574687831,
      "loss": 1.8169,
      "step": 120392
    },
    {
      "epoch": 1.57,
      "grad_norm": 1.1468223333358765,
      "learning_rate": 0.00027878157714890503,
      "loss": 2.3951,
      "step": 120393
    },
    {
      "epoch": 1.57,
      "grad_norm": 1.05366051197052,
      "learning_rate": 0.0002787774968329718,
      "loss": 2.0044,
      "step": 120394
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.9939082860946655,
      "learning_rate": 0.00027877341652098427,
      "loss": 2.0988,
      "step": 120395
    },
    {
      "epoch": 1.57,
      "grad_norm": 1.0675674676895142,
      "learning_rate": 0.0002787693362129432,
      "loss": 1.9913,
      "step": 120396
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.9842047095298767,
      "learning_rate": 0.00027876525590884917,
      "loss": 2.2375,
      "step": 120397
    },
    {
      "epoch": 1.57,
      "grad_norm": 1.0121772289276123,
      "learning_rate": 0.00027876117560870334,
      "loss": 2.2585,
      "step": 120398
    },
    {
      "epoch": 1.57,
      "grad_norm": 1.00525963306427,
      "learning_rate": 0.00027875709531250607,
      "loss": 2.3439,
      "step": 120399
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.9642131924629211,
      "learning_rate": 0.0002787530150202582,
      "loss": 2.0373,
      "step": 120400
    },
    {
      "epoch": 1.57,
      "grad_norm": 1.0327332019805908,
      "learning_rate": 0.00027874893473196056,
      "loss": 2.2171,
      "step": 120401
    },
    {
      "epoch": 1.57,
      "grad_norm": 1.0028263330459595,
      "learning_rate": 0.00027874485444761394,
      "loss": 2.1722,
      "step": 120402
    },
    {
      "epoch": 1.57,
      "grad_norm": 1.0878710746765137,
      "learning_rate": 0.000278740774167219,
      "loss": 2.1103,
      "step": 120403
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.9406734704971313,
      "learning_rate": 0.0002787366938907767,
      "loss": 2.4249,
      "step": 120404
    },
    {
      "epoch": 1.57,
      "grad_norm": 1.0040603876113892,
      "learning_rate": 0.0002787326136182875,
      "loss": 2.0635,
      "step": 120405
    },
    {
      "epoch": 1.57,
      "grad_norm": 1.2672088146209717,
      "learning_rate": 0.0002787285333497524,
      "loss": 2.111,
      "step": 120406
    },
    {
      "epoch": 1.57,
      "grad_norm": 1.2549660205841064,
      "learning_rate": 0.00027872445308517197,
      "loss": 2.0174,
      "step": 120407
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.9510579109191895,
      "learning_rate": 0.0002787203728245471,
      "loss": 2.1993,
      "step": 120408
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.9391548037528992,
      "learning_rate": 0.0002787162925678785,
      "loss": 2.0139,
      "step": 120409
    },
    {
      "epoch": 1.57,
      "grad_norm": 1.2227802276611328,
      "learning_rate": 0.0002787122123151671,
      "loss": 1.9723,
      "step": 120410
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.9297360777854919,
      "learning_rate": 0.0002787081320664133,
      "loss": 2.1303,
      "step": 120411
    },
    {
      "epoch": 1.57,
      "grad_norm": 1.0337045192718506,
      "learning_rate": 0.00027870405182161813,
      "loss": 2.0971,
      "step": 120412
    },
    {
      "epoch": 1.57,
      "grad_norm": 1.0451993942260742,
      "learning_rate": 0.0002786999715807823,
      "loss": 2.199,
      "step": 120413
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.9187489151954651,
      "learning_rate": 0.0002786958913439065,
      "loss": 2.1942,
      "step": 120414
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.8988787531852722,
      "learning_rate": 0.0002786918111109915,
      "loss": 2.2985,
      "step": 120415
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.9000300765037537,
      "learning_rate": 0.0002786877308820382,
      "loss": 2.1252,
      "step": 120416
    },
    {
      "epoch": 1.57,
      "grad_norm": 1.032138466835022,
      "learning_rate": 0.0002786836506570473,
      "loss": 2.2735,
      "step": 120417
    },
    {
      "epoch": 1.57,
      "grad_norm": 1.0974851846694946,
      "learning_rate": 0.00027867957043601943,
      "loss": 2.2345,
      "step": 120418
    },
    {
      "epoch": 1.57,
      "grad_norm": 1.2347373962402344,
      "learning_rate": 0.00027867549021895537,
      "loss": 1.8243,
      "step": 120419
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.8665550351142883,
      "learning_rate": 0.00027867141000585596,
      "loss": 2.1073,
      "step": 120420
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.9855462312698364,
      "learning_rate": 0.00027866732979672196,
      "loss": 2.1887,
      "step": 120421
    },
    {
      "epoch": 1.57,
      "grad_norm": 1.3897730112075806,
      "learning_rate": 0.0002786632495915542,
      "loss": 2.1881,
      "step": 120422
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.9784659147262573,
      "learning_rate": 0.0002786591693903532,
      "loss": 1.9243,
      "step": 120423
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.9816338419914246,
      "learning_rate": 0.0002786550891931199,
      "loss": 2.1983,
      "step": 120424
    },
    {
      "epoch": 1.57,
      "grad_norm": 1.048418402671814,
      "learning_rate": 0.0002786510089998551,
      "loss": 2.1796,
      "step": 120425
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.9620819687843323,
      "learning_rate": 0.0002786469288105594,
      "loss": 2.063,
      "step": 120426
    },
    {
      "epoch": 1.57,
      "grad_norm": 1.106803297996521,
      "learning_rate": 0.0002786428486252337,
      "loss": 2.1671,
      "step": 120427
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.9169248938560486,
      "learning_rate": 0.0002786387684438786,
      "loss": 2.1886,
      "step": 120428
    },
    {
      "epoch": 1.57,
      "grad_norm": 1.0619064569473267,
      "learning_rate": 0.00027863468826649506,
      "loss": 2.2717,
      "step": 120429
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.9434853196144104,
      "learning_rate": 0.0002786306080930837,
      "loss": 2.062,
      "step": 120430
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.9446966648101807,
      "learning_rate": 0.00027862652792364525,
      "loss": 2.11,
      "step": 120431
    },
    {
      "epoch": 1.57,
      "grad_norm": 1.1629507541656494,
      "learning_rate": 0.00027862244775818065,
      "loss": 2.0813,
      "step": 120432
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.9819698929786682,
      "learning_rate": 0.00027861836759669045,
      "loss": 2.0841,
      "step": 120433
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.9787707924842834,
      "learning_rate": 0.0002786142874391756,
      "loss": 2.1334,
      "step": 120434
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.9718921184539795,
      "learning_rate": 0.00027861020728563667,
      "loss": 2.1233,
      "step": 120435
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.9161977767944336,
      "learning_rate": 0.0002786061271360745,
      "loss": 2.2062,
      "step": 120436
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.9416643977165222,
      "learning_rate": 0.00027860204699048987,
      "loss": 2.3265,
      "step": 120437
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.8740599751472473,
      "learning_rate": 0.00027859796684888346,
      "loss": 2.1414,
      "step": 120438
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.8894813060760498,
      "learning_rate": 0.0002785938867112562,
      "loss": 2.0423,
      "step": 120439
    },
    {
      "epoch": 1.57,
      "grad_norm": 1.004115343093872,
      "learning_rate": 0.0002785898065776087,
      "loss": 1.961,
      "step": 120440
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.9056762456893921,
      "learning_rate": 0.0002785857264479418,
      "loss": 2.1725,
      "step": 120441
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.9760639071464539,
      "learning_rate": 0.0002785816463222562,
      "loss": 2.2031,
      "step": 120442
    },
    {
      "epoch": 1.57,
      "grad_norm": 1.04863703250885,
      "learning_rate": 0.0002785775662005526,
      "loss": 2.098,
      "step": 120443
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.9904354810714722,
      "learning_rate": 0.00027857348608283187,
      "loss": 2.2112,
      "step": 120444
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.9229249358177185,
      "learning_rate": 0.00027856940596909476,
      "loss": 2.1166,
      "step": 120445
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.9900117516517639,
      "learning_rate": 0.00027856532585934205,
      "loss": 2.1249,
      "step": 120446
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.9367552399635315,
      "learning_rate": 0.0002785612457535744,
      "loss": 2.154,
      "step": 120447
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.8825640082359314,
      "learning_rate": 0.00027855716565179265,
      "loss": 2.0255,
      "step": 120448
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.9761510491371155,
      "learning_rate": 0.00027855308555399744,
      "loss": 2.4195,
      "step": 120449
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.9268012046813965,
      "learning_rate": 0.00027854900546018965,
      "loss": 1.8579,
      "step": 120450
    },
    {
      "epoch": 1.57,
      "grad_norm": 1.0787643194198608,
      "learning_rate": 0.00027854492537037005,
      "loss": 2.135,
      "step": 120451
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.9792741537094116,
      "learning_rate": 0.0002785408452845393,
      "loss": 2.0729,
      "step": 120452
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.9367763996124268,
      "learning_rate": 0.0002785367652026983,
      "loss": 2.1374,
      "step": 120453
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.9484889507293701,
      "learning_rate": 0.00027853268512484765,
      "loss": 2.2885,
      "step": 120454
    },
    {
      "epoch": 1.57,
      "grad_norm": 1.004323124885559,
      "learning_rate": 0.00027852860505098815,
      "loss": 2.0362,
      "step": 120455
    },
    {
      "epoch": 1.57,
      "grad_norm": 1.0942068099975586,
      "learning_rate": 0.00027852452498112063,
      "loss": 1.9031,
      "step": 120456
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.9558420777320862,
      "learning_rate": 0.0002785204449152458,
      "loss": 2.1075,
      "step": 120457
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.9957376718521118,
      "learning_rate": 0.0002785163648533644,
      "loss": 2.1393,
      "step": 120458
    },
    {
      "epoch": 1.57,
      "grad_norm": 1.0095546245574951,
      "learning_rate": 0.00027851228479547737,
      "loss": 1.962,
      "step": 120459
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.9229565262794495,
      "learning_rate": 0.00027850820474158513,
      "loss": 1.9734,
      "step": 120460
    },
    {
      "epoch": 1.57,
      "grad_norm": 1.0222837924957275,
      "learning_rate": 0.00027850412469168867,
      "loss": 1.9564,
      "step": 120461
    },
    {
      "epoch": 1.57,
      "grad_norm": 1.1441493034362793,
      "learning_rate": 0.0002785000446457887,
      "loss": 2.3587,
      "step": 120462
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.9478043913841248,
      "learning_rate": 0.00027849596460388594,
      "loss": 2.045,
      "step": 120463
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.9441686868667603,
      "learning_rate": 0.00027849188456598124,
      "loss": 2.003,
      "step": 120464
    },
    {
      "epoch": 1.57,
      "grad_norm": 1.0037747621536255,
      "learning_rate": 0.0002784878045320754,
      "loss": 2.1385,
      "step": 120465
    },
    {
      "epoch": 1.57,
      "grad_norm": 1.0081934928894043,
      "learning_rate": 0.0002784837245021689,
      "loss": 1.9279,
      "step": 120466
    },
    {
      "epoch": 1.57,
      "grad_norm": 1.0498785972595215,
      "learning_rate": 0.00027847964447626276,
      "loss": 2.0772,
      "step": 120467
    },
    {
      "epoch": 1.57,
      "grad_norm": 1.015716791152954,
      "learning_rate": 0.00027847556445435764,
      "loss": 2.0014,
      "step": 120468
    },
    {
      "epoch": 1.57,
      "grad_norm": 1.0628913640975952,
      "learning_rate": 0.00027847148443645435,
      "loss": 2.0429,
      "step": 120469
    },
    {
      "epoch": 1.57,
      "grad_norm": 1.0882201194763184,
      "learning_rate": 0.00027846740442255357,
      "loss": 2.2492,
      "step": 120470
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.9400272369384766,
      "learning_rate": 0.00027846332441265625,
      "loss": 2.2564,
      "step": 120471
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.9536727070808411,
      "learning_rate": 0.0002784592444067629,
      "loss": 2.2099,
      "step": 120472
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.9816932082176208,
      "learning_rate": 0.0002784551644048743,
      "loss": 2.1814,
      "step": 120473
    },
    {
      "epoch": 1.57,
      "grad_norm": 1.10713791847229,
      "learning_rate": 0.00027845108440699136,
      "loss": 2.1093,
      "step": 120474
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.9630302786827087,
      "learning_rate": 0.00027844700441311476,
      "loss": 2.1946,
      "step": 120475
    },
    {
      "epoch": 1.57,
      "grad_norm": 1.0493830442428589,
      "learning_rate": 0.0002784429244232453,
      "loss": 2.208,
      "step": 120476
    },
    {
      "epoch": 1.57,
      "grad_norm": 1.1224092245101929,
      "learning_rate": 0.0002784388444373838,
      "loss": 2.03,
      "step": 120477
    },
    {
      "epoch": 1.57,
      "grad_norm": 1.0458393096923828,
      "learning_rate": 0.0002784347644555308,
      "loss": 2.0749,
      "step": 120478
    },
    {
      "epoch": 1.57,
      "grad_norm": 1.0148121118545532,
      "learning_rate": 0.0002784306844776871,
      "loss": 2.3223,
      "step": 120479
    },
    {
      "epoch": 1.57,
      "grad_norm": 1.1572262048721313,
      "learning_rate": 0.0002784266045038537,
      "loss": 2.1126,
      "step": 120480
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.9325186610221863,
      "learning_rate": 0.00027842252453403113,
      "loss": 2.0869,
      "step": 120481
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.9542617797851562,
      "learning_rate": 0.0002784184445682202,
      "loss": 2.1099,
      "step": 120482
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.9235255122184753,
      "learning_rate": 0.0002784143646064218,
      "loss": 2.082,
      "step": 120483
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.9445145726203918,
      "learning_rate": 0.00027841028464863657,
      "loss": 2.2711,
      "step": 120484
    },
    {
      "epoch": 1.57,
      "grad_norm": 1.0245916843414307,
      "learning_rate": 0.00027840620469486515,
      "loss": 2.1721,
      "step": 120485
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.9040603041648865,
      "learning_rate": 0.00027840212474510847,
      "loss": 2.0068,
      "step": 120486
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.9351667165756226,
      "learning_rate": 0.00027839804479936725,
      "loss": 2.1669,
      "step": 120487
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.9875577092170715,
      "learning_rate": 0.00027839396485764223,
      "loss": 2.2443,
      "step": 120488
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.9971977472305298,
      "learning_rate": 0.0002783898849199343,
      "loss": 2.0994,
      "step": 120489
    },
    {
      "epoch": 1.57,
      "grad_norm": 1.1202009916305542,
      "learning_rate": 0.000278385804986244,
      "loss": 2.2461,
      "step": 120490
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.8952822685241699,
      "learning_rate": 0.00027838172505657214,
      "loss": 2.0282,
      "step": 120491
    },
    {
      "epoch": 1.57,
      "grad_norm": 1.0437512397766113,
      "learning_rate": 0.00027837764513091956,
      "loss": 2.3575,
      "step": 120492
    },
    {
      "epoch": 1.57,
      "grad_norm": 1.1055296659469604,
      "learning_rate": 0.000278373565209287,
      "loss": 2.2615,
      "step": 120493
    },
    {
      "epoch": 1.57,
      "grad_norm": 1.101372480392456,
      "learning_rate": 0.0002783694852916751,
      "loss": 2.2719,
      "step": 120494
    },
    {
      "epoch": 1.57,
      "grad_norm": 1.0298619270324707,
      "learning_rate": 0.000278365405378085,
      "loss": 2.3474,
      "step": 120495
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.9835948348045349,
      "learning_rate": 0.00027836132546851696,
      "loss": 2.1623,
      "step": 120496
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.9529933333396912,
      "learning_rate": 0.00027835724556297194,
      "loss": 2.1003,
      "step": 120497
    },
    {
      "epoch": 1.57,
      "grad_norm": 1.0186067819595337,
      "learning_rate": 0.00027835316566145077,
      "loss": 2.0848,
      "step": 120498
    },
    {
      "epoch": 1.57,
      "grad_norm": 1.0046855211257935,
      "learning_rate": 0.00027834908576395415,
      "loss": 2.2585,
      "step": 120499
    },
    {
      "epoch": 1.57,
      "grad_norm": 1.0241323709487915,
      "learning_rate": 0.0002783450058704828,
      "loss": 2.3314,
      "step": 120500
    },
    {
      "epoch": 1.57,
      "grad_norm": 1.0191214084625244,
      "learning_rate": 0.00027834092598103764,
      "loss": 2.2504,
      "step": 120501
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.9741595387458801,
      "learning_rate": 0.00027833684609561925,
      "loss": 2.1721,
      "step": 120502
    },
    {
      "epoch": 1.57,
      "grad_norm": 1.0130726099014282,
      "learning_rate": 0.00027833276621422846,
      "loss": 2.2227,
      "step": 120503
    },
    {
      "epoch": 1.57,
      "grad_norm": 1.1002168655395508,
      "learning_rate": 0.00027832868633686597,
      "loss": 1.9374,
      "step": 120504
    },
    {
      "epoch": 1.57,
      "grad_norm": 1.096178650856018,
      "learning_rate": 0.0002783246064635326,
      "loss": 2.0639,
      "step": 120505
    },
    {
      "epoch": 1.57,
      "grad_norm": 1.0368462800979614,
      "learning_rate": 0.0002783205265942291,
      "loss": 2.3679,
      "step": 120506
    },
    {
      "epoch": 1.57,
      "grad_norm": 1.0333077907562256,
      "learning_rate": 0.0002783164467289562,
      "loss": 2.168,
      "step": 120507
    },
    {
      "epoch": 1.57,
      "grad_norm": 1.0808086395263672,
      "learning_rate": 0.00027831236686771477,
      "loss": 1.9256,
      "step": 120508
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.9574146866798401,
      "learning_rate": 0.00027830828701050543,
      "loss": 2.0462,
      "step": 120509
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.93048095703125,
      "learning_rate": 0.000278304207157329,
      "loss": 2.0415,
      "step": 120510
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.9322504997253418,
      "learning_rate": 0.00027830012730818617,
      "loss": 2.0338,
      "step": 120511
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.8786373734474182,
      "learning_rate": 0.00027829604746307777,
      "loss": 2.2584,
      "step": 120512
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.9166118502616882,
      "learning_rate": 0.00027829196762200456,
      "loss": 2.1054,
      "step": 120513
    },
    {
      "epoch": 1.57,
      "grad_norm": 1.008893370628357,
      "learning_rate": 0.00027828788778496734,
      "loss": 2.0221,
      "step": 120514
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.9639531970024109,
      "learning_rate": 0.00027828380795196677,
      "loss": 2.1573,
      "step": 120515
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.9326729774475098,
      "learning_rate": 0.0002782797281230036,
      "loss": 2.2969,
      "step": 120516
    },
    {
      "epoch": 1.57,
      "grad_norm": 1.2198219299316406,
      "learning_rate": 0.00027827564829807877,
      "loss": 2.2598,
      "step": 120517
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.9047383666038513,
      "learning_rate": 0.00027827156847719275,
      "loss": 2.0731,
      "step": 120518
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.9066751003265381,
      "learning_rate": 0.0002782674886603465,
      "loss": 2.2012,
      "step": 120519
    },
    {
      "epoch": 1.57,
      "grad_norm": 1.0150851011276245,
      "learning_rate": 0.00027826340884754084,
      "loss": 2.2042,
      "step": 120520
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.8887078166007996,
      "learning_rate": 0.0002782593290387763,
      "loss": 2.2099,
      "step": 120521
    },
    {
      "epoch": 1.57,
      "grad_norm": 1.0642468929290771,
      "learning_rate": 0.0002782552492340538,
      "loss": 2.291,
      "step": 120522
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.8908334970474243,
      "learning_rate": 0.00027825116943337405,
      "loss": 2.1429,
      "step": 120523
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.8671708106994629,
      "learning_rate": 0.0002782470896367379,
      "loss": 1.9553,
      "step": 120524
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.8469325304031372,
      "learning_rate": 0.0002782430098441459,
      "loss": 2.0992,
      "step": 120525
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.9640334248542786,
      "learning_rate": 0.00027823893005559904,
      "loss": 2.317,
      "step": 120526
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.9695920348167419,
      "learning_rate": 0.0002782348502710979,
      "loss": 2.2641,
      "step": 120527
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.9191740155220032,
      "learning_rate": 0.00027823077049064337,
      "loss": 2.1281,
      "step": 120528
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.9277257919311523,
      "learning_rate": 0.0002782266907142361,
      "loss": 2.195,
      "step": 120529
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.9743243455886841,
      "learning_rate": 0.0002782226109418769,
      "loss": 2.3581,
      "step": 120530
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.9650954604148865,
      "learning_rate": 0.0002782185311735666,
      "loss": 2.2149,
      "step": 120531
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.9270180463790894,
      "learning_rate": 0.00027821445140930587,
      "loss": 2.004,
      "step": 120532
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.9049896001815796,
      "learning_rate": 0.00027821037164909543,
      "loss": 2.0807,
      "step": 120533
    },
    {
      "epoch": 1.57,
      "grad_norm": 1.0407986640930176,
      "learning_rate": 0.0002782062918929361,
      "loss": 2.1575,
      "step": 120534
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.988601565361023,
      "learning_rate": 0.00027820221214082863,
      "loss": 2.288,
      "step": 120535
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.9088332056999207,
      "learning_rate": 0.0002781981323927738,
      "loss": 2.1364,
      "step": 120536
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.9731408953666687,
      "learning_rate": 0.0002781940526487723,
      "loss": 2.0309,
      "step": 120537
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.976433515548706,
      "learning_rate": 0.0002781899729088251,
      "loss": 2.289,
      "step": 120538
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.9588816165924072,
      "learning_rate": 0.0002781858931729326,
      "loss": 2.1931,
      "step": 120539
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.9470610618591309,
      "learning_rate": 0.00027818181344109585,
      "loss": 2.3066,
      "step": 120540
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.9100158214569092,
      "learning_rate": 0.0002781777337133155,
      "loss": 2.1335,
      "step": 120541
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.9177016615867615,
      "learning_rate": 0.00027817365398959224,
      "loss": 2.0576,
      "step": 120542
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.9479290246963501,
      "learning_rate": 0.000278169574269927,
      "loss": 1.9303,
      "step": 120543
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.9647261500358582,
      "learning_rate": 0.00027816549455432054,
      "loss": 2.1869,
      "step": 120544
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.8740852475166321,
      "learning_rate": 0.0002781614148427734,
      "loss": 2.1071,
      "step": 120545
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.9620160460472107,
      "learning_rate": 0.00027815733513528645,
      "loss": 1.984,
      "step": 120546
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.970512330532074,
      "learning_rate": 0.00027815325543186046,
      "loss": 2.1082,
      "step": 120547
    },
    {
      "epoch": 1.57,
      "grad_norm": 1.0316981077194214,
      "learning_rate": 0.00027814917573249626,
      "loss": 2.0318,
      "step": 120548
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.9726394414901733,
      "learning_rate": 0.00027814509603719446,
      "loss": 2.1531,
      "step": 120549
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.9712489247322083,
      "learning_rate": 0.0002781410163459561,
      "loss": 2.3205,
      "step": 120550
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.9462188482284546,
      "learning_rate": 0.0002781369366587815,
      "loss": 2.132,
      "step": 120551
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.9806687235832214,
      "learning_rate": 0.00027813285697567174,
      "loss": 2.2968,
      "step": 120552
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.9111563563346863,
      "learning_rate": 0.0002781287772966275,
      "loss": 2.0837,
      "step": 120553
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.9850011467933655,
      "learning_rate": 0.0002781246976216495,
      "loss": 2.0698,
      "step": 120554
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.9520152807235718,
      "learning_rate": 0.00027812061795073857,
      "loss": 2.2448,
      "step": 120555
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.95263671875,
      "learning_rate": 0.0002781165382838955,
      "loss": 1.7841,
      "step": 120556
    },
    {
      "epoch": 1.57,
      "grad_norm": 1.0284968614578247,
      "learning_rate": 0.0002781124586211209,
      "loss": 2.4038,
      "step": 120557
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.9253740906715393,
      "learning_rate": 0.0002781083789624156,
      "loss": 2.2627,
      "step": 120558
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.9158031344413757,
      "learning_rate": 0.0002781042993077803,
      "loss": 2.0041,
      "step": 120559
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.9714059829711914,
      "learning_rate": 0.00027810021965721596,
      "loss": 2.0964,
      "step": 120560
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.9289869070053101,
      "learning_rate": 0.0002780961400107231,
      "loss": 1.9804,
      "step": 120561
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.9069922566413879,
      "learning_rate": 0.0002780920603683027,
      "loss": 2.0257,
      "step": 120562
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.9484973549842834,
      "learning_rate": 0.0002780879807299553,
      "loss": 2.0892,
      "step": 120563
    },
    {
      "epoch": 1.57,
      "grad_norm": 1.0587712526321411,
      "learning_rate": 0.0002780839010956818,
      "loss": 2.1848,
      "step": 120564
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.9627460837364197,
      "learning_rate": 0.00027807982146548285,
      "loss": 1.9827,
      "step": 120565
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.9162907600402832,
      "learning_rate": 0.0002780757418393593,
      "loss": 1.8267,
      "step": 120566
    },
    {
      "epoch": 1.57,
      "grad_norm": 1.051198124885559,
      "learning_rate": 0.0002780716622173119,
      "loss": 2.2633,
      "step": 120567
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.9850853681564331,
      "learning_rate": 0.0002780675825993415,
      "loss": 2.0035,
      "step": 120568
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.9386916160583496,
      "learning_rate": 0.00027806350298544864,
      "loss": 1.9114,
      "step": 120569
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.960705041885376,
      "learning_rate": 0.0002780594233756342,
      "loss": 2.15,
      "step": 120570
    },
    {
      "epoch": 1.57,
      "grad_norm": 1.0972322225570679,
      "learning_rate": 0.00027805534376989887,
      "loss": 2.3356,
      "step": 120571
    },
    {
      "epoch": 1.57,
      "grad_norm": 1.1171371936798096,
      "learning_rate": 0.0002780512641682435,
      "loss": 2.2145,
      "step": 120572
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.9667510390281677,
      "learning_rate": 0.0002780471845706688,
      "loss": 2.0434,
      "step": 120573
    },
    {
      "epoch": 1.57,
      "grad_norm": 1.036718487739563,
      "learning_rate": 0.00027804310497717556,
      "loss": 2.1915,
      "step": 120574
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.9617798924446106,
      "learning_rate": 0.0002780390253877647,
      "loss": 2.2244,
      "step": 120575
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.9719774127006531,
      "learning_rate": 0.0002780349458024366,
      "loss": 2.1617,
      "step": 120576
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.8988080024719238,
      "learning_rate": 0.00027803086622119223,
      "loss": 2.13,
      "step": 120577
    },
    {
      "epoch": 1.57,
      "grad_norm": 1.0249930620193481,
      "learning_rate": 0.00027802678664403237,
      "loss": 2.3881,
      "step": 120578
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.9814130663871765,
      "learning_rate": 0.0002780227070709577,
      "loss": 2.0514,
      "step": 120579
    },
    {
      "epoch": 1.57,
      "grad_norm": 1.0683907270431519,
      "learning_rate": 0.00027801862750196905,
      "loss": 2.2729,
      "step": 120580
    },
    {
      "epoch": 1.57,
      "grad_norm": 1.0293594598770142,
      "learning_rate": 0.00027801454793706723,
      "loss": 2.1104,
      "step": 120581
    },
    {
      "epoch": 1.57,
      "grad_norm": 1.1232749223709106,
      "learning_rate": 0.00027801046837625285,
      "loss": 1.9725,
      "step": 120582
    },
    {
      "epoch": 1.57,
      "grad_norm": 1.0605002641677856,
      "learning_rate": 0.0002780063888195267,
      "loss": 2.1751,
      "step": 120583
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.9959829449653625,
      "learning_rate": 0.00027800230926688967,
      "loss": 2.0829,
      "step": 120584
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.9561598896980286,
      "learning_rate": 0.00027799822971834233,
      "loss": 2.2081,
      "step": 120585
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.8961475491523743,
      "learning_rate": 0.00027799415017388553,
      "loss": 1.9247,
      "step": 120586
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.9013943076133728,
      "learning_rate": 0.0002779900706335202,
      "loss": 2.2947,
      "step": 120587
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.8657897710800171,
      "learning_rate": 0.0002779859910972468,
      "loss": 1.982,
      "step": 120588
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.9888920783996582,
      "learning_rate": 0.0002779819115650662,
      "loss": 2.221,
      "step": 120589
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.9742352366447449,
      "learning_rate": 0.00027797783203697924,
      "loss": 2.4589,
      "step": 120590
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.9402247667312622,
      "learning_rate": 0.00027797375251298654,
      "loss": 2.1658,
      "step": 120591
    },
    {
      "epoch": 1.57,
      "grad_norm": 1.062182903289795,
      "learning_rate": 0.00027796967299308897,
      "loss": 2.4928,
      "step": 120592
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.9843128323554993,
      "learning_rate": 0.0002779655934772873,
      "loss": 1.9886,
      "step": 120593
    },
    {
      "epoch": 1.57,
      "grad_norm": 1.023521065711975,
      "learning_rate": 0.0002779615139655823,
      "loss": 2.3119,
      "step": 120594
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.9046269059181213,
      "learning_rate": 0.00027795743445797454,
      "loss": 2.3273,
      "step": 120595
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.907082736492157,
      "learning_rate": 0.0002779533549544649,
      "loss": 2.1092,
      "step": 120596
    },
    {
      "epoch": 1.57,
      "grad_norm": 1.0383667945861816,
      "learning_rate": 0.0002779492754550542,
      "loss": 2.1583,
      "step": 120597
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.9771523475646973,
      "learning_rate": 0.00027794519595974315,
      "loss": 2.1144,
      "step": 120598
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.9678972363471985,
      "learning_rate": 0.00027794111646853255,
      "loss": 2.2694,
      "step": 120599
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.8640286326408386,
      "learning_rate": 0.000277937036981423,
      "loss": 2.129,
      "step": 120600
    },
    {
      "epoch": 1.57,
      "grad_norm": 1.114932656288147,
      "learning_rate": 0.0002779329574984155,
      "loss": 2.0082,
      "step": 120601
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.9679191708564758,
      "learning_rate": 0.0002779288780195106,
      "loss": 2.2633,
      "step": 120602
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.9131399393081665,
      "learning_rate": 0.0002779247985447091,
      "loss": 2.083,
      "step": 120603
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.8978775143623352,
      "learning_rate": 0.00027792071907401184,
      "loss": 1.992,
      "step": 120604
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.9713897109031677,
      "learning_rate": 0.00027791663960741956,
      "loss": 2.2096,
      "step": 120605
    },
    {
      "epoch": 1.57,
      "grad_norm": 1.1122556924819946,
      "learning_rate": 0.000277912560144933,
      "loss": 2.2012,
      "step": 120606
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.9273660182952881,
      "learning_rate": 0.00027790848068655284,
      "loss": 2.1733,
      "step": 120607
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.937581479549408,
      "learning_rate": 0.00027790440123228,
      "loss": 2.0124,
      "step": 120608
    },
    {
      "epoch": 1.57,
      "grad_norm": 1.0011281967163086,
      "learning_rate": 0.0002779003217821151,
      "loss": 2.0433,
      "step": 120609
    },
    {
      "epoch": 1.57,
      "grad_norm": 1.119410514831543,
      "learning_rate": 0.00027789624233605896,
      "loss": 2.0431,
      "step": 120610
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.9573974013328552,
      "learning_rate": 0.00027789216289411233,
      "loss": 2.0952,
      "step": 120611
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.8456906676292419,
      "learning_rate": 0.00027788808345627597,
      "loss": 1.9239,
      "step": 120612
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.9247059226036072,
      "learning_rate": 0.0002778840040225506,
      "loss": 1.9662,
      "step": 120613
    },
    {
      "epoch": 1.57,
      "grad_norm": 1.2370500564575195,
      "learning_rate": 0.00027787992459293696,
      "loss": 2.0324,
      "step": 120614
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.9694126844406128,
      "learning_rate": 0.0002778758451674359,
      "loss": 2.1262,
      "step": 120615
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.9771971106529236,
      "learning_rate": 0.00027787176574604823,
      "loss": 2.0895,
      "step": 120616
    },
    {
      "epoch": 1.57,
      "grad_norm": 1.033087134361267,
      "learning_rate": 0.00027786768632877463,
      "loss": 2.2832,
      "step": 120617
    },
    {
      "epoch": 1.57,
      "grad_norm": 1.010143518447876,
      "learning_rate": 0.00027786360691561574,
      "loss": 1.7898,
      "step": 120618
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.9068863987922668,
      "learning_rate": 0.0002778595275065724,
      "loss": 2.2679,
      "step": 120619
    },
    {
      "epoch": 1.57,
      "grad_norm": 1.028426170349121,
      "learning_rate": 0.00027785544810164545,
      "loss": 2.0119,
      "step": 120620
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.9897608160972595,
      "learning_rate": 0.00027785136870083556,
      "loss": 2.0936,
      "step": 120621
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.9359160661697388,
      "learning_rate": 0.0002778472893041435,
      "loss": 1.8465,
      "step": 120622
    },
    {
      "epoch": 1.57,
      "grad_norm": 1.0009647607803345,
      "learning_rate": 0.0002778432099115702,
      "loss": 2.2041,
      "step": 120623
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.9555711150169373,
      "learning_rate": 0.0002778391305231162,
      "loss": 1.9791,
      "step": 120624
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.9061813950538635,
      "learning_rate": 0.0002778350511387822,
      "loss": 2.1378,
      "step": 120625
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.9653998613357544,
      "learning_rate": 0.00027783097175856914,
      "loss": 2.267,
      "step": 120626
    },
    {
      "epoch": 1.57,
      "grad_norm": 1.1370612382888794,
      "learning_rate": 0.00027782689238247776,
      "loss": 2.0512,
      "step": 120627
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.8839266300201416,
      "learning_rate": 0.00027782281301050874,
      "loss": 2.3209,
      "step": 120628
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.9324766397476196,
      "learning_rate": 0.00027781873364266294,
      "loss": 2.0415,
      "step": 120629
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.9797958135604858,
      "learning_rate": 0.00027781465427894106,
      "loss": 2.1556,
      "step": 120630
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.980827271938324,
      "learning_rate": 0.00027781057491934377,
      "loss": 2.2397,
      "step": 120631
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.9003173112869263,
      "learning_rate": 0.0002778064955638719,
      "loss": 1.9816,
      "step": 120632
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.9937267303466797,
      "learning_rate": 0.00027780241621252626,
      "loss": 2.2107,
      "step": 120633
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.9909859895706177,
      "learning_rate": 0.00027779833686530757,
      "loss": 2.2302,
      "step": 120634
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.9804261326789856,
      "learning_rate": 0.00027779425752221676,
      "loss": 2.0158,
      "step": 120635
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.8245750665664673,
      "learning_rate": 0.0002777901781832542,
      "loss": 2.1264,
      "step": 120636
    },
    {
      "epoch": 1.57,
      "grad_norm": 1.261164903640747,
      "learning_rate": 0.0002777860988484209,
      "loss": 2.2204,
      "step": 120637
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.9987741112709045,
      "learning_rate": 0.0002777820195177176,
      "loss": 2.3102,
      "step": 120638
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.9261528849601746,
      "learning_rate": 0.000277777940191145,
      "loss": 2.0212,
      "step": 120639
    },
    {
      "epoch": 1.57,
      "grad_norm": 1.0963643789291382,
      "learning_rate": 0.00027777386086870394,
      "loss": 2.0628,
      "step": 120640
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.9710003733634949,
      "learning_rate": 0.0002777697815503951,
      "loss": 2.2156,
      "step": 120641
    },
    {
      "epoch": 1.57,
      "grad_norm": 1.0308904647827148,
      "learning_rate": 0.00027776570223621945,
      "loss": 2.2866,
      "step": 120642
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.9118653535842896,
      "learning_rate": 0.00027776162292617745,
      "loss": 2.1377,
      "step": 120643
    },
    {
      "epoch": 1.57,
      "grad_norm": 1.0474638938903809,
      "learning_rate": 0.00027775754362026995,
      "loss": 2.2122,
      "step": 120644
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.9713125824928284,
      "learning_rate": 0.00027775346431849776,
      "loss": 2.1395,
      "step": 120645
    },
    {
      "epoch": 1.57,
      "grad_norm": 1.0047703981399536,
      "learning_rate": 0.0002777493850208616,
      "loss": 1.9511,
      "step": 120646
    },
    {
      "epoch": 1.57,
      "grad_norm": 1.0501703023910522,
      "learning_rate": 0.0002777453057273623,
      "loss": 2.4483,
      "step": 120647
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.9318467974662781,
      "learning_rate": 0.00027774122643800064,
      "loss": 2.1331,
      "step": 120648
    },
    {
      "epoch": 1.57,
      "grad_norm": 1.3842817544937134,
      "learning_rate": 0.00027773714715277723,
      "loss": 1.9656,
      "step": 120649
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.9563118815422058,
      "learning_rate": 0.00027773306787169286,
      "loss": 2.071,
      "step": 120650
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.9920895099639893,
      "learning_rate": 0.00027772898859474836,
      "loss": 2.0099,
      "step": 120651
    },
    {
      "epoch": 1.57,
      "grad_norm": 1.154720425605774,
      "learning_rate": 0.00027772490932194443,
      "loss": 2.2396,
      "step": 120652
    },
    {
      "epoch": 1.57,
      "grad_norm": 1.4818271398544312,
      "learning_rate": 0.00027772083005328193,
      "loss": 2.0881,
      "step": 120653
    },
    {
      "epoch": 1.57,
      "grad_norm": 1.0150949954986572,
      "learning_rate": 0.0002777167507887616,
      "loss": 2.3199,
      "step": 120654
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.9319706559181213,
      "learning_rate": 0.00027771267152838406,
      "loss": 2.2044,
      "step": 120655
    },
    {
      "epoch": 1.57,
      "grad_norm": 1.092235803604126,
      "learning_rate": 0.0002777085922721501,
      "loss": 2.1889,
      "step": 120656
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.9577284455299377,
      "learning_rate": 0.0002777045130200606,
      "loss": 2.0463,
      "step": 120657
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.9506137371063232,
      "learning_rate": 0.00027770043377211624,
      "loss": 2.0276,
      "step": 120658
    },
    {
      "epoch": 1.57,
      "grad_norm": 1.0009881258010864,
      "learning_rate": 0.00027769635452831775,
      "loss": 1.9509,
      "step": 120659
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.9436419010162354,
      "learning_rate": 0.00027769227528866606,
      "loss": 2.0952,
      "step": 120660
    },
    {
      "epoch": 1.57,
      "grad_norm": 1.127931833267212,
      "learning_rate": 0.0002776881960531617,
      "loss": 2.0931,
      "step": 120661
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.8585121035575867,
      "learning_rate": 0.00027768411682180553,
      "loss": 1.9924,
      "step": 120662
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.931475043296814,
      "learning_rate": 0.0002776800375945983,
      "loss": 2.0285,
      "step": 120663
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.8901709318161011,
      "learning_rate": 0.0002776759583715408,
      "loss": 2.1995,
      "step": 120664
    },
    {
      "epoch": 1.57,
      "grad_norm": 1.0204709768295288,
      "learning_rate": 0.0002776718791526337,
      "loss": 2.1304,
      "step": 120665
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.9651203751564026,
      "learning_rate": 0.000277667799937878,
      "loss": 2.2438,
      "step": 120666
    },
    {
      "epoch": 1.57,
      "grad_norm": 1.022301435470581,
      "learning_rate": 0.0002776637207272741,
      "loss": 2.2795,
      "step": 120667
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.9857595562934875,
      "learning_rate": 0.00027765964152082293,
      "loss": 2.0271,
      "step": 120668
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.9522746801376343,
      "learning_rate": 0.00027765556231852526,
      "loss": 1.9489,
      "step": 120669
    },
    {
      "epoch": 1.57,
      "grad_norm": 1.0490059852600098,
      "learning_rate": 0.0002776514831203819,
      "loss": 2.1884,
      "step": 120670
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.9355697631835938,
      "learning_rate": 0.00027764740392639353,
      "loss": 2.2652,
      "step": 120671
    },
    {
      "epoch": 1.57,
      "grad_norm": 1.0268162488937378,
      "learning_rate": 0.00027764332473656104,
      "loss": 2.0103,
      "step": 120672
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.9210324883460999,
      "learning_rate": 0.0002776392455508849,
      "loss": 2.0086,
      "step": 120673
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.9408453106880188,
      "learning_rate": 0.0002776351663693661,
      "loss": 2.1346,
      "step": 120674
    },
    {
      "epoch": 1.57,
      "grad_norm": 1.4480223655700684,
      "learning_rate": 0.0002776310871920054,
      "loss": 2.1255,
      "step": 120675
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.9722595810890198,
      "learning_rate": 0.00027762700801880347,
      "loss": 2.201,
      "step": 120676
    },
    {
      "epoch": 1.57,
      "grad_norm": 1.02800714969635,
      "learning_rate": 0.00027762292884976104,
      "loss": 2.2889,
      "step": 120677
    },
    {
      "epoch": 1.57,
      "grad_norm": 1.130828857421875,
      "learning_rate": 0.000277618849684879,
      "loss": 2.1416,
      "step": 120678
    },
    {
      "epoch": 1.57,
      "grad_norm": 1.0501518249511719,
      "learning_rate": 0.0002776147705241581,
      "loss": 2.2015,
      "step": 120679
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.9509016871452332,
      "learning_rate": 0.00027761069136759893,
      "loss": 1.9797,
      "step": 120680
    },
    {
      "epoch": 1.57,
      "grad_norm": 1.2192784547805786,
      "learning_rate": 0.0002776066122152024,
      "loss": 2.4072,
      "step": 120681
    },
    {
      "epoch": 1.57,
      "grad_norm": 1.0485600233078003,
      "learning_rate": 0.00027760253306696915,
      "loss": 2.1064,
      "step": 120682
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.9774539470672607,
      "learning_rate": 0.0002775984539229001,
      "loss": 2.0809,
      "step": 120683
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.9876092672348022,
      "learning_rate": 0.0002775943747829959,
      "loss": 2.3318,
      "step": 120684
    },
    {
      "epoch": 1.57,
      "grad_norm": 1.0277584791183472,
      "learning_rate": 0.0002775902956472573,
      "loss": 2.2058,
      "step": 120685
    },
    {
      "epoch": 1.57,
      "grad_norm": 1.0911997556686401,
      "learning_rate": 0.0002775862165156852,
      "loss": 1.8241,
      "step": 120686
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.9719023704528809,
      "learning_rate": 0.00027758213738828013,
      "loss": 1.9792,
      "step": 120687
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.9174168109893799,
      "learning_rate": 0.00027757805826504297,
      "loss": 2.1334,
      "step": 120688
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.9379244446754456,
      "learning_rate": 0.00027757397914597446,
      "loss": 2.16,
      "step": 120689
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.9863144755363464,
      "learning_rate": 0.00027756990003107546,
      "loss": 2.3122,
      "step": 120690
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.9510228633880615,
      "learning_rate": 0.00027756582092034653,
      "loss": 2.3111,
      "step": 120691
    },
    {
      "epoch": 1.57,
      "grad_norm": 1.0182725191116333,
      "learning_rate": 0.00027756174181378857,
      "loss": 1.9228,
      "step": 120692
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.9185720682144165,
      "learning_rate": 0.0002775576627114024,
      "loss": 2.0048,
      "step": 120693
    },
    {
      "epoch": 1.57,
      "grad_norm": 1.0951967239379883,
      "learning_rate": 0.0002775535836131886,
      "loss": 2.0275,
      "step": 120694
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.9383717775344849,
      "learning_rate": 0.00027754950451914796,
      "loss": 2.3509,
      "step": 120695
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.9525898694992065,
      "learning_rate": 0.0002775454254292814,
      "loss": 2.1367,
      "step": 120696
    },
    {
      "epoch": 1.57,
      "grad_norm": 1.0124114751815796,
      "learning_rate": 0.0002775413463435895,
      "loss": 2.0278,
      "step": 120697
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.9244247078895569,
      "learning_rate": 0.0002775372672620731,
      "loss": 2.1619,
      "step": 120698
    },
    {
      "epoch": 1.57,
      "grad_norm": 1.0612397193908691,
      "learning_rate": 0.00027753318818473293,
      "loss": 2.0117,
      "step": 120699
    },
    {
      "epoch": 1.57,
      "grad_norm": 1.1083711385726929,
      "learning_rate": 0.0002775291091115698,
      "loss": 2.1656,
      "step": 120700
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.9025788307189941,
      "learning_rate": 0.00027752503004258445,
      "loss": 2.0989,
      "step": 120701
    },
    {
      "epoch": 1.57,
      "grad_norm": 1.0063632726669312,
      "learning_rate": 0.0002775209509777776,
      "loss": 2.1165,
      "step": 120702
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.9851478934288025,
      "learning_rate": 0.00027751687191715003,
      "loss": 1.9168,
      "step": 120703
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.9857733249664307,
      "learning_rate": 0.00027751279286070243,
      "loss": 1.8341,
      "step": 120704
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.9783036708831787,
      "learning_rate": 0.0002775087138084357,
      "loss": 2.1702,
      "step": 120705
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.9657843708992004,
      "learning_rate": 0.0002775046347603505,
      "loss": 2.1558,
      "step": 120706
    },
    {
      "epoch": 1.57,
      "grad_norm": 1.053122639656067,
      "learning_rate": 0.0002775005557164476,
      "loss": 1.9468,
      "step": 120707
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.9150933027267456,
      "learning_rate": 0.00027749647667672783,
      "loss": 2.2085,
      "step": 120708
    },
    {
      "epoch": 1.57,
      "grad_norm": 1.0657812356948853,
      "learning_rate": 0.0002774923976411919,
      "loss": 2.2224,
      "step": 120709
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.913528323173523,
      "learning_rate": 0.0002774883186098405,
      "loss": 1.9975,
      "step": 120710
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.9701547622680664,
      "learning_rate": 0.00027748423958267443,
      "loss": 2.2147,
      "step": 120711
    },
    {
      "epoch": 1.57,
      "grad_norm": 1.0897092819213867,
      "learning_rate": 0.0002774801605596945,
      "loss": 2.1449,
      "step": 120712
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.9067435264587402,
      "learning_rate": 0.0002774760815409014,
      "loss": 1.862,
      "step": 120713
    },
    {
      "epoch": 1.57,
      "grad_norm": 1.0797673463821411,
      "learning_rate": 0.0002774720025262959,
      "loss": 2.0066,
      "step": 120714
    },
    {
      "epoch": 1.57,
      "grad_norm": 1.0390576124191284,
      "learning_rate": 0.00027746792351587894,
      "loss": 2.3043,
      "step": 120715
    },
    {
      "epoch": 1.57,
      "grad_norm": 1.0011591911315918,
      "learning_rate": 0.00027746384450965105,
      "loss": 2.2518,
      "step": 120716
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.8747965097427368,
      "learning_rate": 0.00027745976550761295,
      "loss": 2.5298,
      "step": 120717
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.861187756061554,
      "learning_rate": 0.00027745568650976556,
      "loss": 2.0296,
      "step": 120718
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.9505311846733093,
      "learning_rate": 0.00027745160751610953,
      "loss": 2.0672,
      "step": 120719
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.9186925292015076,
      "learning_rate": 0.00027744752852664574,
      "loss": 2.1171,
      "step": 120720
    },
    {
      "epoch": 1.57,
      "grad_norm": 1.0115894079208374,
      "learning_rate": 0.000277443449541375,
      "loss": 2.0355,
      "step": 120721
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.9244738817214966,
      "learning_rate": 0.0002774393705602978,
      "loss": 2.2496,
      "step": 120722
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.9584105014801025,
      "learning_rate": 0.00027743529158341505,
      "loss": 2.1544,
      "step": 120723
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.9468117356300354,
      "learning_rate": 0.0002774312126107275,
      "loss": 2.1977,
      "step": 120724
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.9652481079101562,
      "learning_rate": 0.0002774271336422359,
      "loss": 2.1053,
      "step": 120725
    },
    {
      "epoch": 1.57,
      "grad_norm": 1.1310173273086548,
      "learning_rate": 0.00027742305467794105,
      "loss": 2.1732,
      "step": 120726
    },
    {
      "epoch": 1.57,
      "grad_norm": 1.0077168941497803,
      "learning_rate": 0.00027741897571784385,
      "loss": 2.1588,
      "step": 120727
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.8761774301528931,
      "learning_rate": 0.0002774148967619447,
      "loss": 1.9921,
      "step": 120728
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.9605623483657837,
      "learning_rate": 0.0002774108178102445,
      "loss": 2.0558,
      "step": 120729
    },
    {
      "epoch": 1.57,
      "grad_norm": 1.274477481842041,
      "learning_rate": 0.00027740673886274416,
      "loss": 1.985,
      "step": 120730
    },
    {
      "epoch": 1.57,
      "grad_norm": 1.0022457838058472,
      "learning_rate": 0.0002774026599194443,
      "loss": 1.8836,
      "step": 120731
    },
    {
      "epoch": 1.57,
      "grad_norm": 1.1206570863723755,
      "learning_rate": 0.0002773985809803457,
      "loss": 2.1642,
      "step": 120732
    },
    {
      "epoch": 1.57,
      "grad_norm": 1.012352705001831,
      "learning_rate": 0.00027739450204544925,
      "loss": 2.3056,
      "step": 120733
    },
    {
      "epoch": 1.57,
      "grad_norm": 1.0124140977859497,
      "learning_rate": 0.00027739042311475544,
      "loss": 2.2764,
      "step": 120734
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.921904981136322,
      "learning_rate": 0.00027738634418826525,
      "loss": 2.0322,
      "step": 120735
    },
    {
      "epoch": 1.57,
      "grad_norm": 1.035625696182251,
      "learning_rate": 0.0002773822652659793,
      "loss": 2.0645,
      "step": 120736
    },
    {
      "epoch": 1.57,
      "grad_norm": 1.010955810546875,
      "learning_rate": 0.00027737818634789843,
      "loss": 2.1978,
      "step": 120737
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.9097284078598022,
      "learning_rate": 0.00027737410743402337,
      "loss": 2.267,
      "step": 120738
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.9142593145370483,
      "learning_rate": 0.00027737002852435506,
      "loss": 2.2082,
      "step": 120739
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.9793586730957031,
      "learning_rate": 0.00027736594961889395,
      "loss": 2.2213,
      "step": 120740
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.9483867287635803,
      "learning_rate": 0.0002773618707176409,
      "loss": 1.998,
      "step": 120741
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.9393773674964905,
      "learning_rate": 0.0002773577918205967,
      "loss": 2.2122,
      "step": 120742
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.9943299293518066,
      "learning_rate": 0.00027735371292776217,
      "loss": 2.1382,
      "step": 120743
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.9269797205924988,
      "learning_rate": 0.000277349634039138,
      "loss": 1.9625,
      "step": 120744
    },
    {
      "epoch": 1.57,
      "grad_norm": 1.1148587465286255,
      "learning_rate": 0.00027734555515472506,
      "loss": 2.1447,
      "step": 120745
    },
    {
      "epoch": 1.57,
      "grad_norm": 1.0435069799423218,
      "learning_rate": 0.0002773414762745239,
      "loss": 2.2549,
      "step": 120746
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.9406275749206543,
      "learning_rate": 0.00027733739739853536,
      "loss": 2.2023,
      "step": 120747
    },
    {
      "epoch": 1.57,
      "grad_norm": 1.0266714096069336,
      "learning_rate": 0.00027733331852676025,
      "loss": 2.2116,
      "step": 120748
    },
    {
      "epoch": 1.57,
      "grad_norm": 1.0908255577087402,
      "learning_rate": 0.0002773292396591993,
      "loss": 2.2976,
      "step": 120749
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.9944822192192078,
      "learning_rate": 0.0002773251607958533,
      "loss": 1.9276,
      "step": 120750
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.8883735537528992,
      "learning_rate": 0.0002773210819367231,
      "loss": 2.2461,
      "step": 120751
    },
    {
      "epoch": 1.57,
      "grad_norm": 1.0364069938659668,
      "learning_rate": 0.0002773170030818092,
      "loss": 2.1521,
      "step": 120752
    },
    {
      "epoch": 1.57,
      "grad_norm": 1.0436476469039917,
      "learning_rate": 0.0002773129242311125,
      "loss": 2.1044,
      "step": 120753
    },
    {
      "epoch": 1.57,
      "grad_norm": 1.2436425685882568,
      "learning_rate": 0.0002773088453846337,
      "loss": 2.2222,
      "step": 120754
    },
    {
      "epoch": 1.57,
      "grad_norm": 1.0508445501327515,
      "learning_rate": 0.00027730476654237366,
      "loss": 2.3832,
      "step": 120755
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.9594104886054993,
      "learning_rate": 0.0002773006877043331,
      "loss": 2.5095,
      "step": 120756
    },
    {
      "epoch": 1.57,
      "grad_norm": 1.0201870203018188,
      "learning_rate": 0.0002772966088705129,
      "loss": 2.3546,
      "step": 120757
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.8744117021560669,
      "learning_rate": 0.00027729253004091356,
      "loss": 2.1202,
      "step": 120758
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.9613155126571655,
      "learning_rate": 0.00027728845121553596,
      "loss": 2.0734,
      "step": 120759
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.9910075068473816,
      "learning_rate": 0.00027728437239438083,
      "loss": 2.272,
      "step": 120760
    },
    {
      "epoch": 1.57,
      "grad_norm": 1.152645468711853,
      "learning_rate": 0.000277280293577449,
      "loss": 2.0703,
      "step": 120761
    },
    {
      "epoch": 1.57,
      "grad_norm": 1.2344794273376465,
      "learning_rate": 0.0002772762147647412,
      "loss": 2.1566,
      "step": 120762
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.8842512369155884,
      "learning_rate": 0.0002772721359562583,
      "loss": 2.0915,
      "step": 120763
    },
    {
      "epoch": 1.57,
      "grad_norm": 1.2075906991958618,
      "learning_rate": 0.0002772680571520008,
      "loss": 2.1268,
      "step": 120764
    },
    {
      "epoch": 1.57,
      "grad_norm": 1.0995793342590332,
      "learning_rate": 0.0002772639783519696,
      "loss": 2.26,
      "step": 120765
    },
    {
      "epoch": 1.57,
      "grad_norm": 1.0063540935516357,
      "learning_rate": 0.0002772598995561655,
      "loss": 2.2864,
      "step": 120766
    },
    {
      "epoch": 1.57,
      "grad_norm": 1.120941162109375,
      "learning_rate": 0.0002772558207645892,
      "loss": 2.0262,
      "step": 120767
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.9458109736442566,
      "learning_rate": 0.0002772517419772414,
      "loss": 2.3657,
      "step": 120768
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.8719334006309509,
      "learning_rate": 0.00027724766319412304,
      "loss": 2.2469,
      "step": 120769
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.9762340784072876,
      "learning_rate": 0.0002772435844152347,
      "loss": 2.1377,
      "step": 120770
    },
    {
      "epoch": 1.57,
      "grad_norm": 1.1944252252578735,
      "learning_rate": 0.0002772395056405773,
      "loss": 2.2619,
      "step": 120771
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.8756424188613892,
      "learning_rate": 0.00027723542687015145,
      "loss": 2.1654,
      "step": 120772
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.9056962132453918,
      "learning_rate": 0.0002772313481039579,
      "loss": 2.1823,
      "step": 120773
    },
    {
      "epoch": 1.57,
      "grad_norm": 1.0205638408660889,
      "learning_rate": 0.00027722726934199755,
      "loss": 2.0176,
      "step": 120774
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.8681848049163818,
      "learning_rate": 0.000277223190584271,
      "loss": 2.1909,
      "step": 120775
    },
    {
      "epoch": 1.57,
      "grad_norm": 1.0648390054702759,
      "learning_rate": 0.0002772191118307792,
      "loss": 1.9884,
      "step": 120776
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.9654525518417358,
      "learning_rate": 0.00027721503308152275,
      "loss": 2.1171,
      "step": 120777
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.9598265886306763,
      "learning_rate": 0.00027721095433650247,
      "loss": 2.2818,
      "step": 120778
    },
    {
      "epoch": 1.57,
      "grad_norm": 1.0235778093338013,
      "learning_rate": 0.00027720687559571903,
      "loss": 2.3268,
      "step": 120779
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.8933852314949036,
      "learning_rate": 0.00027720279685917337,
      "loss": 2.289,
      "step": 120780
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.941295862197876,
      "learning_rate": 0.00027719871812686606,
      "loss": 2.0411,
      "step": 120781
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.9135758876800537,
      "learning_rate": 0.000277194639398798,
      "loss": 2.1017,
      "step": 120782
    },
    {
      "epoch": 1.57,
      "grad_norm": 1.0294911861419678,
      "learning_rate": 0.0002771905606749698,
      "loss": 2.0095,
      "step": 120783
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.955752432346344,
      "learning_rate": 0.00027718648195538235,
      "loss": 2.1188,
      "step": 120784
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.9340350031852722,
      "learning_rate": 0.0002771824032400364,
      "loss": 2.1237,
      "step": 120785
    },
    {
      "epoch": 1.57,
      "grad_norm": 1.1063895225524902,
      "learning_rate": 0.00027717832452893264,
      "loss": 2.2315,
      "step": 120786
    },
    {
      "epoch": 1.57,
      "grad_norm": 1.0328842401504517,
      "learning_rate": 0.00027717424582207183,
      "loss": 1.9783,
      "step": 120787
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.85980623960495,
      "learning_rate": 0.0002771701671194548,
      "loss": 2.0595,
      "step": 120788
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.9900479912757874,
      "learning_rate": 0.0002771660884210823,
      "loss": 2.1113,
      "step": 120789
    },
    {
      "epoch": 1.57,
      "grad_norm": 1.1493960618972778,
      "learning_rate": 0.000277162009726955,
      "loss": 2.0106,
      "step": 120790
    },
    {
      "epoch": 1.57,
      "grad_norm": 1.014502763748169,
      "learning_rate": 0.00027715793103707365,
      "loss": 2.2271,
      "step": 120791
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.980452835559845,
      "learning_rate": 0.00027715385235143925,
      "loss": 2.0671,
      "step": 120792
    },
    {
      "epoch": 1.57,
      "grad_norm": 1.0341185331344604,
      "learning_rate": 0.00027714977367005226,
      "loss": 2.2531,
      "step": 120793
    },
    {
      "epoch": 1.57,
      "grad_norm": 1.119412899017334,
      "learning_rate": 0.0002771456949929136,
      "loss": 2.0218,
      "step": 120794
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.888816773891449,
      "learning_rate": 0.00027714161632002394,
      "loss": 2.3131,
      "step": 120795
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.9399518966674805,
      "learning_rate": 0.0002771375376513841,
      "loss": 2.4666,
      "step": 120796
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.9596462249755859,
      "learning_rate": 0.0002771334589869948,
      "loss": 2.1255,
      "step": 120797
    },
    {
      "epoch": 1.57,
      "grad_norm": 1.3631895780563354,
      "learning_rate": 0.00027712938032685685,
      "loss": 2.1247,
      "step": 120798
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.9116692543029785,
      "learning_rate": 0.00027712530167097095,
      "loss": 2.0062,
      "step": 120799
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.9961127638816833,
      "learning_rate": 0.00027712122301933804,
      "loss": 2.1933,
      "step": 120800
    },
    {
      "epoch": 1.57,
      "grad_norm": 1.214548945426941,
      "learning_rate": 0.0002771171443719586,
      "loss": 2.192,
      "step": 120801
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.9808528423309326,
      "learning_rate": 0.0002771130657288335,
      "loss": 2.1263,
      "step": 120802
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.9722765684127808,
      "learning_rate": 0.0002771089870899635,
      "loss": 2.2151,
      "step": 120803
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.9213322997093201,
      "learning_rate": 0.0002771049084553494,
      "loss": 2.4163,
      "step": 120804
    },
    {
      "epoch": 1.57,
      "grad_norm": 1.0230834484100342,
      "learning_rate": 0.00027710082982499194,
      "loss": 2.0358,
      "step": 120805
    },
    {
      "epoch": 1.57,
      "grad_norm": 1.1644974946975708,
      "learning_rate": 0.00027709675119889194,
      "loss": 2.3892,
      "step": 120806
    },
    {
      "epoch": 1.57,
      "grad_norm": 1.0085529088974,
      "learning_rate": 0.00027709267257705,
      "loss": 2.0797,
      "step": 120807
    },
    {
      "epoch": 1.57,
      "grad_norm": 1.0166736841201782,
      "learning_rate": 0.00027708859395946694,
      "loss": 2.0824,
      "step": 120808
    },
    {
      "epoch": 1.57,
      "grad_norm": 1.1991747617721558,
      "learning_rate": 0.0002770845153461436,
      "loss": 2.1203,
      "step": 120809
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.9534875750541687,
      "learning_rate": 0.0002770804367370806,
      "loss": 2.1482,
      "step": 120810
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.9679706692695618,
      "learning_rate": 0.00027707635813227884,
      "loss": 1.9588,
      "step": 120811
    },
    {
      "epoch": 1.57,
      "grad_norm": 1.0620611906051636,
      "learning_rate": 0.0002770722795317391,
      "loss": 2.2503,
      "step": 120812
    },
    {
      "epoch": 1.57,
      "grad_norm": 1.0647835731506348,
      "learning_rate": 0.000277068200935462,
      "loss": 1.976,
      "step": 120813
    },
    {
      "epoch": 1.57,
      "grad_norm": 1.141688346862793,
      "learning_rate": 0.0002770641223434483,
      "loss": 2.2339,
      "step": 120814
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.9111195206642151,
      "learning_rate": 0.0002770600437556988,
      "loss": 2.1022,
      "step": 120815
    },
    {
      "epoch": 1.57,
      "grad_norm": 1.0303771495819092,
      "learning_rate": 0.0002770559651722143,
      "loss": 2.0635,
      "step": 120816
    },
    {
      "epoch": 1.57,
      "grad_norm": 1.0441666841506958,
      "learning_rate": 0.0002770518865929955,
      "loss": 2.1634,
      "step": 120817
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.9798623323440552,
      "learning_rate": 0.00027704780801804336,
      "loss": 2.2086,
      "step": 120818
    },
    {
      "epoch": 1.57,
      "grad_norm": 1.1300849914550781,
      "learning_rate": 0.0002770437294473583,
      "loss": 2.1743,
      "step": 120819
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.891648530960083,
      "learning_rate": 0.0002770396508809413,
      "loss": 2.2748,
      "step": 120820
    },
    {
      "epoch": 1.57,
      "grad_norm": 1.0319641828536987,
      "learning_rate": 0.000277035572318793,
      "loss": 2.1361,
      "step": 120821
    },
    {
      "epoch": 1.57,
      "grad_norm": 1.0839266777038574,
      "learning_rate": 0.0002770314937609142,
      "loss": 2.2553,
      "step": 120822
    },
    {
      "epoch": 1.57,
      "grad_norm": 1.1182219982147217,
      "learning_rate": 0.00027702741520730577,
      "loss": 2.0911,
      "step": 120823
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.9955628514289856,
      "learning_rate": 0.0002770233366579684,
      "loss": 2.0823,
      "step": 120824
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.8985311985015869,
      "learning_rate": 0.0002770192581129028,
      "loss": 2.2415,
      "step": 120825
    },
    {
      "epoch": 1.57,
      "grad_norm": 1.0110949277877808,
      "learning_rate": 0.0002770151795721097,
      "loss": 1.9477,
      "step": 120826
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.9679544568061829,
      "learning_rate": 0.0002770111010355899,
      "loss": 2.2542,
      "step": 120827
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.8918417692184448,
      "learning_rate": 0.0002770070225033442,
      "loss": 2.292,
      "step": 120828
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.9315503835678101,
      "learning_rate": 0.00027700294397537333,
      "loss": 2.0115,
      "step": 120829
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.9321288466453552,
      "learning_rate": 0.00027699886545167813,
      "loss": 2.0697,
      "step": 120830
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.9688466787338257,
      "learning_rate": 0.00027699478693225924,
      "loss": 2.2046,
      "step": 120831
    },
    {
      "epoch": 1.57,
      "grad_norm": 1.0413955450057983,
      "learning_rate": 0.00027699070841711736,
      "loss": 2.0385,
      "step": 120832
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.9646406769752502,
      "learning_rate": 0.00027698662990625336,
      "loss": 2.1193,
      "step": 120833
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.8624281287193298,
      "learning_rate": 0.000276982551399668,
      "loss": 2.0376,
      "step": 120834
    },
    {
      "epoch": 1.57,
      "grad_norm": 1.1214721202850342,
      "learning_rate": 0.00027697847289736205,
      "loss": 1.8728,
      "step": 120835
    },
    {
      "epoch": 1.57,
      "grad_norm": 1.0548549890518188,
      "learning_rate": 0.0002769743943993363,
      "loss": 2.2393,
      "step": 120836
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.8977856636047363,
      "learning_rate": 0.0002769703159055913,
      "loss": 2.0502,
      "step": 120837
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.9867134094238281,
      "learning_rate": 0.00027696623741612806,
      "loss": 2.1368,
      "step": 120838
    },
    {
      "epoch": 1.57,
      "grad_norm": 1.0728380680084229,
      "learning_rate": 0.00027696215893094713,
      "loss": 1.9136,
      "step": 120839
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.883541464805603,
      "learning_rate": 0.0002769580804500494,
      "loss": 1.8188,
      "step": 120840
    },
    {
      "epoch": 1.57,
      "grad_norm": 1.0391169786453247,
      "learning_rate": 0.0002769540019734356,
      "loss": 2.1915,
      "step": 120841
    },
    {
      "epoch": 1.57,
      "grad_norm": 1.2643381357192993,
      "learning_rate": 0.00027694992350110655,
      "loss": 2.1346,
      "step": 120842
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.9694792628288269,
      "learning_rate": 0.00027694584503306296,
      "loss": 2.1619,
      "step": 120843
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.9542784094810486,
      "learning_rate": 0.00027694176656930554,
      "loss": 2.0139,
      "step": 120844
    },
    {
      "epoch": 1.57,
      "grad_norm": 1.0503734350204468,
      "learning_rate": 0.000276937688109835,
      "loss": 2.224,
      "step": 120845
    },
    {
      "epoch": 1.57,
      "grad_norm": 1.0476759672164917,
      "learning_rate": 0.0002769336096546522,
      "loss": 2.2064,
      "step": 120846
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.8930910229682922,
      "learning_rate": 0.00027692953120375793,
      "loss": 1.9981,
      "step": 120847
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.9894809126853943,
      "learning_rate": 0.00027692545275715286,
      "loss": 2.4523,
      "step": 120848
    },
    {
      "epoch": 1.57,
      "grad_norm": 1.0152168273925781,
      "learning_rate": 0.00027692137431483793,
      "loss": 2.1523,
      "step": 120849
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.9991841912269592,
      "learning_rate": 0.00027691729587681356,
      "loss": 2.3131,
      "step": 120850
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.9564534425735474,
      "learning_rate": 0.0002769132174430808,
      "loss": 2.2516,
      "step": 120851
    },
    {
      "epoch": 1.57,
      "grad_norm": 1.0521353483200073,
      "learning_rate": 0.00027690913901364027,
      "loss": 2.1347,
      "step": 120852
    },
    {
      "epoch": 1.57,
      "grad_norm": 1.0161076784133911,
      "learning_rate": 0.0002769050605884927,
      "loss": 2.1271,
      "step": 120853
    },
    {
      "epoch": 1.57,
      "grad_norm": 1.0125846862792969,
      "learning_rate": 0.00027690098216763895,
      "loss": 2.0845,
      "step": 120854
    },
    {
      "epoch": 1.57,
      "grad_norm": 1.096834659576416,
      "learning_rate": 0.00027689690375107994,
      "loss": 2.0637,
      "step": 120855
    },
    {
      "epoch": 1.57,
      "grad_norm": 1.0086162090301514,
      "learning_rate": 0.00027689282533881605,
      "loss": 2.245,
      "step": 120856
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.9475809931755066,
      "learning_rate": 0.0002768887469308482,
      "loss": 2.1208,
      "step": 120857
    },
    {
      "epoch": 1.57,
      "grad_norm": 1.0042966604232788,
      "learning_rate": 0.0002768846685271772,
      "loss": 2.3617,
      "step": 120858
    },
    {
      "epoch": 1.57,
      "grad_norm": 1.0395169258117676,
      "learning_rate": 0.0002768805901278038,
      "loss": 2.3502,
      "step": 120859
    },
    {
      "epoch": 1.57,
      "grad_norm": 1.2225548028945923,
      "learning_rate": 0.0002768765117327287,
      "loss": 2.1142,
      "step": 120860
    },
    {
      "epoch": 1.57,
      "grad_norm": 1.0547159910202026,
      "learning_rate": 0.0002768724333419527,
      "loss": 2.2595,
      "step": 120861
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.9342068433761597,
      "learning_rate": 0.0002768683549554767,
      "loss": 2.1285,
      "step": 120862
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.9421586990356445,
      "learning_rate": 0.00027686427657330114,
      "loss": 2.0837,
      "step": 120863
    },
    {
      "epoch": 1.57,
      "grad_norm": 1.0590347051620483,
      "learning_rate": 0.00027686019819542695,
      "loss": 2.2072,
      "step": 120864
    },
    {
      "epoch": 1.57,
      "grad_norm": 1.032965898513794,
      "learning_rate": 0.0002768561198218549,
      "loss": 2.1139,
      "step": 120865
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.9284526705741882,
      "learning_rate": 0.0002768520414525857,
      "loss": 2.3416,
      "step": 120866
    },
    {
      "epoch": 1.57,
      "grad_norm": 1.0076488256454468,
      "learning_rate": 0.0002768479630876203,
      "loss": 1.9248,
      "step": 120867
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.9567434191703796,
      "learning_rate": 0.00027684388472695915,
      "loss": 2.093,
      "step": 120868
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.9116840958595276,
      "learning_rate": 0.0002768398063706032,
      "loss": 2.0668,
      "step": 120869
    },
    {
      "epoch": 1.57,
      "grad_norm": 1.0310657024383545,
      "learning_rate": 0.00027683572801855317,
      "loss": 2.1822,
      "step": 120870
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.9480738043785095,
      "learning_rate": 0.00027683164967080983,
      "loss": 1.9663,
      "step": 120871
    },
    {
      "epoch": 1.57,
      "grad_norm": 1.3083198070526123,
      "learning_rate": 0.00027682757132737386,
      "loss": 2.2793,
      "step": 120872
    },
    {
      "epoch": 1.57,
      "grad_norm": 1.0506489276885986,
      "learning_rate": 0.0002768234929882462,
      "loss": 2.1343,
      "step": 120873
    },
    {
      "epoch": 1.57,
      "grad_norm": 1.050469160079956,
      "learning_rate": 0.0002768194146534274,
      "loss": 2.041,
      "step": 120874
    },
    {
      "epoch": 1.57,
      "grad_norm": 1.0856432914733887,
      "learning_rate": 0.0002768153363229183,
      "loss": 1.9886,
      "step": 120875
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.9753589630126953,
      "learning_rate": 0.0002768112579967197,
      "loss": 2.1426,
      "step": 120876
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.9133800268173218,
      "learning_rate": 0.00027680717967483237,
      "loss": 2.2635,
      "step": 120877
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.9981250762939453,
      "learning_rate": 0.00027680310135725695,
      "loss": 1.929,
      "step": 120878
    },
    {
      "epoch": 1.57,
      "grad_norm": 1.1077455282211304,
      "learning_rate": 0.00027679902304399433,
      "loss": 1.7787,
      "step": 120879
    },
    {
      "epoch": 1.57,
      "grad_norm": 1.0540660619735718,
      "learning_rate": 0.00027679494473504515,
      "loss": 2.2216,
      "step": 120880
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.9367126226425171,
      "learning_rate": 0.00027679086643041023,
      "loss": 2.2675,
      "step": 120881
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.982518196105957,
      "learning_rate": 0.0002767867881300904,
      "loss": 2.1228,
      "step": 120882
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.9623112082481384,
      "learning_rate": 0.0002767827098340862,
      "loss": 2.1583,
      "step": 120883
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.9165482521057129,
      "learning_rate": 0.0002767786315423987,
      "loss": 2.1489,
      "step": 120884
    },
    {
      "epoch": 1.57,
      "grad_norm": 1.0845534801483154,
      "learning_rate": 0.00027677455325502846,
      "loss": 2.2011,
      "step": 120885
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.9907997250556946,
      "learning_rate": 0.0002767704749719762,
      "loss": 2.1957,
      "step": 120886
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.9958593845367432,
      "learning_rate": 0.0002767663966932428,
      "loss": 2.0733,
      "step": 120887
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.880765438079834,
      "learning_rate": 0.0002767623184188289,
      "loss": 2.1597,
      "step": 120888
    },
    {
      "epoch": 1.57,
      "grad_norm": 1.112097144126892,
      "learning_rate": 0.00027675824014873537,
      "loss": 2.1374,
      "step": 120889
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.9034018516540527,
      "learning_rate": 0.00027675416188296286,
      "loss": 1.9505,
      "step": 120890
    },
    {
      "epoch": 1.57,
      "grad_norm": 1.1655080318450928,
      "learning_rate": 0.00027675008362151236,
      "loss": 2.1418,
      "step": 120891
    },
    {
      "epoch": 1.57,
      "grad_norm": 1.0057907104492188,
      "learning_rate": 0.0002767460053643843,
      "loss": 2.0954,
      "step": 120892
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.9500117897987366,
      "learning_rate": 0.00027674192711157967,
      "loss": 2.2434,
      "step": 120893
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.950864315032959,
      "learning_rate": 0.0002767378488630991,
      "loss": 2.2561,
      "step": 120894
    },
    {
      "epoch": 1.57,
      "grad_norm": 1.0487934350967407,
      "learning_rate": 0.00027673377061894337,
      "loss": 2.3053,
      "step": 120895
    },
    {
      "epoch": 1.57,
      "grad_norm": 1.0123335123062134,
      "learning_rate": 0.0002767296923791133,
      "loss": 2.2123,
      "step": 120896
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.9080844521522522,
      "learning_rate": 0.0002767256141436098,
      "loss": 1.9658,
      "step": 120897
    },
    {
      "epoch": 1.57,
      "grad_norm": 1.0097362995147705,
      "learning_rate": 0.00027672153591243325,
      "loss": 2.3328,
      "step": 120898
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.9393630623817444,
      "learning_rate": 0.00027671745768558464,
      "loss": 2.1235,
      "step": 120899
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.8968346118927002,
      "learning_rate": 0.0002767133794630647,
      "loss": 2.1323,
      "step": 120900
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.9920353889465332,
      "learning_rate": 0.0002767093012448742,
      "loss": 2.1836,
      "step": 120901
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.9364585876464844,
      "learning_rate": 0.0002767052230310138,
      "loss": 2.1901,
      "step": 120902
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.9315782189369202,
      "learning_rate": 0.0002767011448214846,
      "loss": 2.1458,
      "step": 120903
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.9291248321533203,
      "learning_rate": 0.0002766970666162868,
      "loss": 2.0241,
      "step": 120904
    },
    {
      "epoch": 1.57,
      "grad_norm": 1.033636450767517,
      "learning_rate": 0.00027669298841542154,
      "loss": 2.2507,
      "step": 120905
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.978774905204773,
      "learning_rate": 0.00027668891021888955,
      "loss": 2.0808,
      "step": 120906
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.9594096541404724,
      "learning_rate": 0.00027668483202669144,
      "loss": 2.1502,
      "step": 120907
    },
    {
      "epoch": 1.57,
      "grad_norm": 1.0135973691940308,
      "learning_rate": 0.0002766807538388281,
      "loss": 2.0282,
      "step": 120908
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.9647312760353088,
      "learning_rate": 0.00027667667565530027,
      "loss": 1.8722,
      "step": 120909
    },
    {
      "epoch": 1.57,
      "grad_norm": 1.1015090942382812,
      "learning_rate": 0.0002766725974761088,
      "loss": 2.2199,
      "step": 120910
    },
    {
      "epoch": 1.57,
      "grad_norm": 1.0399850606918335,
      "learning_rate": 0.00027666851930125417,
      "loss": 2.1268,
      "step": 120911
    },
    {
      "epoch": 1.57,
      "grad_norm": 1.185486078262329,
      "learning_rate": 0.00027666444113073733,
      "loss": 2.2076,
      "step": 120912
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.920511782169342,
      "learning_rate": 0.00027666036296455904,
      "loss": 2.087,
      "step": 120913
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.9612131714820862,
      "learning_rate": 0.00027665628480271994,
      "loss": 2.3491,
      "step": 120914
    },
    {
      "epoch": 1.57,
      "grad_norm": 1.0277599096298218,
      "learning_rate": 0.00027665220664522096,
      "loss": 2.051,
      "step": 120915
    },
    {
      "epoch": 1.57,
      "grad_norm": 1.1374071836471558,
      "learning_rate": 0.0002766481284920629,
      "loss": 2.0376,
      "step": 120916
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.9855856895446777,
      "learning_rate": 0.0002766440503432462,
      "loss": 2.026,
      "step": 120917
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.9314907193183899,
      "learning_rate": 0.0002766399721987719,
      "loss": 2.0854,
      "step": 120918
    },
    {
      "epoch": 1.57,
      "grad_norm": 1.0229772329330444,
      "learning_rate": 0.0002766358940586406,
      "loss": 2.0973,
      "step": 120919
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.9995623826980591,
      "learning_rate": 0.00027663181592285315,
      "loss": 2.1364,
      "step": 120920
    },
    {
      "epoch": 1.57,
      "grad_norm": 1.1305407285690308,
      "learning_rate": 0.0002766277377914103,
      "loss": 2.0004,
      "step": 120921
    },
    {
      "epoch": 1.57,
      "grad_norm": 1.0509140491485596,
      "learning_rate": 0.0002766236596643129,
      "loss": 2.0615,
      "step": 120922
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.9779481887817383,
      "learning_rate": 0.0002766195815415615,
      "loss": 2.1773,
      "step": 120923
    },
    {
      "epoch": 1.57,
      "grad_norm": 1.2923511266708374,
      "learning_rate": 0.00027661550342315697,
      "loss": 2.1995,
      "step": 120924
    },
    {
      "epoch": 1.57,
      "grad_norm": 1.2846828699111938,
      "learning_rate": 0.0002766114253091,
      "loss": 2.1809,
      "step": 120925
    },
    {
      "epoch": 1.57,
      "grad_norm": 1.140153169631958,
      "learning_rate": 0.0002766073471993915,
      "loss": 2.4316,
      "step": 120926
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.9025281667709351,
      "learning_rate": 0.00027660326909403204,
      "loss": 2.1486,
      "step": 120927
    },
    {
      "epoch": 1.57,
      "grad_norm": 1.122133731842041,
      "learning_rate": 0.00027659919099302265,
      "loss": 2.2835,
      "step": 120928
    },
    {
      "epoch": 1.57,
      "grad_norm": 1.0284128189086914,
      "learning_rate": 0.0002765951128963638,
      "loss": 2.1099,
      "step": 120929
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.9877476096153259,
      "learning_rate": 0.0002765910348040563,
      "loss": 2.2584,
      "step": 120930
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.9887682795524597,
      "learning_rate": 0.00027658695671610103,
      "loss": 2.3156,
      "step": 120931
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.9541019797325134,
      "learning_rate": 0.00027658287863249867,
      "loss": 1.9621,
      "step": 120932
    },
    {
      "epoch": 1.57,
      "grad_norm": 1.164659023284912,
      "learning_rate": 0.00027657880055325003,
      "loss": 2.1872,
      "step": 120933
    },
    {
      "epoch": 1.57,
      "grad_norm": 1.0229456424713135,
      "learning_rate": 0.00027657472247835587,
      "loss": 2.0593,
      "step": 120934
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.8993179202079773,
      "learning_rate": 0.0002765706444078169,
      "loss": 2.277,
      "step": 120935
    },
    {
      "epoch": 1.57,
      "grad_norm": 1.040320873260498,
      "learning_rate": 0.0002765665663416338,
      "loss": 2.3652,
      "step": 120936
    },
    {
      "epoch": 1.57,
      "grad_norm": 1.117671251296997,
      "learning_rate": 0.00027656248827980743,
      "loss": 2.1499,
      "step": 120937
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.9897863864898682,
      "learning_rate": 0.00027655841022233856,
      "loss": 1.991,
      "step": 120938
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.9504292607307434,
      "learning_rate": 0.00027655433216922787,
      "loss": 2.0016,
      "step": 120939
    },
    {
      "epoch": 1.57,
      "grad_norm": 1.0704916715621948,
      "learning_rate": 0.00027655025412047637,
      "loss": 1.9959,
      "step": 120940
    },
    {
      "epoch": 1.57,
      "grad_norm": 1.0321792364120483,
      "learning_rate": 0.00027654617607608445,
      "loss": 2.2576,
      "step": 120941
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.9003531336784363,
      "learning_rate": 0.000276542098036053,
      "loss": 1.7463,
      "step": 120942
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.9678099155426025,
      "learning_rate": 0.0002765380200003829,
      "loss": 2.1189,
      "step": 120943
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.924644947052002,
      "learning_rate": 0.0002765339419690748,
      "loss": 2.0486,
      "step": 120944
    },
    {
      "epoch": 1.57,
      "grad_norm": 1.1539323329925537,
      "learning_rate": 0.0002765298639421295,
      "loss": 2.1616,
      "step": 120945
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.9032288789749146,
      "learning_rate": 0.00027652578591954777,
      "loss": 2.0312,
      "step": 120946
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.9368929266929626,
      "learning_rate": 0.00027652170790133033,
      "loss": 2.1493,
      "step": 120947
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.9313325881958008,
      "learning_rate": 0.00027651762988747795,
      "loss": 2.2305,
      "step": 120948
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.9723029732704163,
      "learning_rate": 0.0002765135518779913,
      "loss": 2.1475,
      "step": 120949
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.9470582008361816,
      "learning_rate": 0.0002765094738728713,
      "loss": 2.1442,
      "step": 120950
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.8776332139968872,
      "learning_rate": 0.0002765053958721186,
      "loss": 2.1599,
      "step": 120951
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.9847537279129028,
      "learning_rate": 0.00027650131787573405,
      "loss": 2.1059,
      "step": 120952
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.9580013155937195,
      "learning_rate": 0.00027649723988371826,
      "loss": 1.9839,
      "step": 120953
    },
    {
      "epoch": 1.57,
      "grad_norm": 1.016437292098999,
      "learning_rate": 0.0002764931618960722,
      "loss": 2.2676,
      "step": 120954
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.944119393825531,
      "learning_rate": 0.0002764890839127964,
      "loss": 2.0886,
      "step": 120955
    },
    {
      "epoch": 1.57,
      "grad_norm": 1.1738299131393433,
      "learning_rate": 0.0002764850059338917,
      "loss": 2.0698,
      "step": 120956
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.9284256100654602,
      "learning_rate": 0.00027648092795935894,
      "loss": 2.0455,
      "step": 120957
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.9262940883636475,
      "learning_rate": 0.0002764768499891988,
      "loss": 2.1879,
      "step": 120958
    },
    {
      "epoch": 1.57,
      "grad_norm": 1.095287561416626,
      "learning_rate": 0.0002764727720234121,
      "loss": 2.0512,
      "step": 120959
    },
    {
      "epoch": 1.57,
      "grad_norm": 0.9190536141395569,
      "learning_rate": 0.0002764686940619995,
      "loss": 2.1458,
      "step": 120960
    },
    {
      "epoch": 1.57,
      "grad_norm": 1.183641791343689,
      "learning_rate": 0.00027646461610496184,
      "loss": 2.2381,
      "step": 120961
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.9851272106170654,
      "learning_rate": 0.00027646053815229987,
      "loss": 2.2065,
      "step": 120962
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.9389866590499878,
      "learning_rate": 0.0002764564602040143,
      "loss": 1.8881,
      "step": 120963
    },
    {
      "epoch": 1.58,
      "grad_norm": 1.0778446197509766,
      "learning_rate": 0.000276452382260106,
      "loss": 2.1765,
      "step": 120964
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.9717473983764648,
      "learning_rate": 0.0002764483043205755,
      "loss": 2.0312,
      "step": 120965
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.9887278079986572,
      "learning_rate": 0.0002764442263854238,
      "loss": 1.8492,
      "step": 120966
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.9169424176216125,
      "learning_rate": 0.0002764401484546515,
      "loss": 2.2837,
      "step": 120967
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.976557195186615,
      "learning_rate": 0.00027643607052825947,
      "loss": 2.2773,
      "step": 120968
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.9679881930351257,
      "learning_rate": 0.00027643199260624844,
      "loss": 2.0022,
      "step": 120969
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.9695165157318115,
      "learning_rate": 0.00027642791468861916,
      "loss": 1.953,
      "step": 120970
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.9826377034187317,
      "learning_rate": 0.00027642383677537233,
      "loss": 2.1733,
      "step": 120971
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.9565581679344177,
      "learning_rate": 0.0002764197588665087,
      "loss": 2.1245,
      "step": 120972
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.9425969123840332,
      "learning_rate": 0.0002764156809620292,
      "loss": 2.0723,
      "step": 120973
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.8862930536270142,
      "learning_rate": 0.0002764116030619344,
      "loss": 2.0244,
      "step": 120974
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.9203253388404846,
      "learning_rate": 0.0002764075251662251,
      "loss": 2.0813,
      "step": 120975
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.974498987197876,
      "learning_rate": 0.00027640344727490216,
      "loss": 2.1576,
      "step": 120976
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.9147831797599792,
      "learning_rate": 0.00027639936938796627,
      "loss": 2.3122,
      "step": 120977
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.9886610507965088,
      "learning_rate": 0.00027639529150541814,
      "loss": 2.0449,
      "step": 120978
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.9602107405662537,
      "learning_rate": 0.00027639121362725853,
      "loss": 2.1907,
      "step": 120979
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.9413203001022339,
      "learning_rate": 0.00027638713575348827,
      "loss": 2.0934,
      "step": 120980
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.8870031833648682,
      "learning_rate": 0.0002763830578841081,
      "loss": 2.1154,
      "step": 120981
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.9939366579055786,
      "learning_rate": 0.00027637898001911873,
      "loss": 2.09,
      "step": 120982
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.9085217714309692,
      "learning_rate": 0.00027637490215852114,
      "loss": 2.1624,
      "step": 120983
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.991341233253479,
      "learning_rate": 0.0002763708243023157,
      "loss": 2.2539,
      "step": 120984
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.9655437469482422,
      "learning_rate": 0.0002763667464505034,
      "loss": 2.2035,
      "step": 120985
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.9818493127822876,
      "learning_rate": 0.000276362668603085,
      "loss": 2.3103,
      "step": 120986
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.9846306443214417,
      "learning_rate": 0.0002763585907600612,
      "loss": 2.1613,
      "step": 120987
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.9959346652030945,
      "learning_rate": 0.0002763545129214328,
      "loss": 2.0218,
      "step": 120988
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.9570068717002869,
      "learning_rate": 0.0002763504350872006,
      "loss": 2.0839,
      "step": 120989
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.9411060214042664,
      "learning_rate": 0.00027634635725736525,
      "loss": 2.3274,
      "step": 120990
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.9803011417388916,
      "learning_rate": 0.0002763422794319275,
      "loss": 2.025,
      "step": 120991
    },
    {
      "epoch": 1.58,
      "grad_norm": 1.0409804582595825,
      "learning_rate": 0.00027633820161088824,
      "loss": 2.2601,
      "step": 120992
    },
    {
      "epoch": 1.58,
      "grad_norm": 1.0843651294708252,
      "learning_rate": 0.0002763341237942481,
      "loss": 2.3744,
      "step": 120993
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.9155734181404114,
      "learning_rate": 0.0002763300459820079,
      "loss": 2.0297,
      "step": 120994
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.8978794813156128,
      "learning_rate": 0.0002763259681741685,
      "loss": 1.9959,
      "step": 120995
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.9438555836677551,
      "learning_rate": 0.0002763218903707305,
      "loss": 2.2289,
      "step": 120996
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.8357256054878235,
      "learning_rate": 0.00027631781257169465,
      "loss": 2.1517,
      "step": 120997
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.9454354047775269,
      "learning_rate": 0.00027631373477706177,
      "loss": 2.0884,
      "step": 120998
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.9838237166404724,
      "learning_rate": 0.00027630965698683264,
      "loss": 2.1803,
      "step": 120999
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.945568323135376,
      "learning_rate": 0.0002763055792010079,
      "loss": 2.3272,
      "step": 121000
    },
    {
      "epoch": 1.58,
      "grad_norm": 1.0594251155853271,
      "learning_rate": 0.00027630150141958864,
      "loss": 1.9489,
      "step": 121001
    },
    {
      "epoch": 1.58,
      "grad_norm": 1.0363640785217285,
      "learning_rate": 0.0002762974236425752,
      "loss": 2.2838,
      "step": 121002
    },
    {
      "epoch": 1.58,
      "grad_norm": 1.0056475400924683,
      "learning_rate": 0.00027629334586996856,
      "loss": 1.9231,
      "step": 121003
    },
    {
      "epoch": 1.58,
      "grad_norm": 1.0057731866836548,
      "learning_rate": 0.0002762892681017694,
      "loss": 2.1083,
      "step": 121004
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.9502233266830444,
      "learning_rate": 0.0002762851903379785,
      "loss": 2.164,
      "step": 121005
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.9574469327926636,
      "learning_rate": 0.0002762811125785966,
      "loss": 2.2439,
      "step": 121006
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.960578203201294,
      "learning_rate": 0.0002762770348236247,
      "loss": 1.9711,
      "step": 121007
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.8957474231719971,
      "learning_rate": 0.00027627295707306315,
      "loss": 2.079,
      "step": 121008
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.9372177720069885,
      "learning_rate": 0.0002762688793269129,
      "loss": 1.9175,
      "step": 121009
    },
    {
      "epoch": 1.58,
      "grad_norm": 1.0307241678237915,
      "learning_rate": 0.0002762648015851748,
      "loss": 1.9521,
      "step": 121010
    },
    {
      "epoch": 1.58,
      "grad_norm": 1.012182354927063,
      "learning_rate": 0.00027626072384784946,
      "loss": 2.3486,
      "step": 121011
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.9278178215026855,
      "learning_rate": 0.0002762566461149377,
      "loss": 2.1727,
      "step": 121012
    },
    {
      "epoch": 1.58,
      "grad_norm": 1.0218430757522583,
      "learning_rate": 0.0002762525683864404,
      "loss": 2.0626,
      "step": 121013
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.9754421710968018,
      "learning_rate": 0.00027624849066235806,
      "loss": 1.9344,
      "step": 121014
    },
    {
      "epoch": 1.58,
      "grad_norm": 1.0474588871002197,
      "learning_rate": 0.0002762444129426916,
      "loss": 2.0587,
      "step": 121015
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.9973593950271606,
      "learning_rate": 0.00027624033522744175,
      "loss": 2.0641,
      "step": 121016
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.9009760022163391,
      "learning_rate": 0.0002762362575166093,
      "loss": 2.3193,
      "step": 121017
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.9299883246421814,
      "learning_rate": 0.0002762321798101949,
      "loss": 2.2125,
      "step": 121018
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.8969650864601135,
      "learning_rate": 0.00027622810210819955,
      "loss": 1.926,
      "step": 121019
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.9170755743980408,
      "learning_rate": 0.00027622402441062374,
      "loss": 2.0647,
      "step": 121020
    },
    {
      "epoch": 1.58,
      "grad_norm": 1.0434067249298096,
      "learning_rate": 0.00027621994671746833,
      "loss": 2.3875,
      "step": 121021
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.835723876953125,
      "learning_rate": 0.0002762158690287341,
      "loss": 1.9745,
      "step": 121022
    },
    {
      "epoch": 1.58,
      "grad_norm": 1.0002344846725464,
      "learning_rate": 0.00027621179134442174,
      "loss": 2.1421,
      "step": 121023
    },
    {
      "epoch": 1.58,
      "grad_norm": 1.0111942291259766,
      "learning_rate": 0.00027620771366453203,
      "loss": 1.8708,
      "step": 121024
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.9307406544685364,
      "learning_rate": 0.000276203635989066,
      "loss": 2.0298,
      "step": 121025
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.9295790791511536,
      "learning_rate": 0.0002761995583180239,
      "loss": 2.1168,
      "step": 121026
    },
    {
      "epoch": 1.58,
      "grad_norm": 1.057349681854248,
      "learning_rate": 0.0002761954806514068,
      "loss": 1.9326,
      "step": 121027
    },
    {
      "epoch": 1.58,
      "grad_norm": 1.0006479024887085,
      "learning_rate": 0.0002761914029892154,
      "loss": 2.1732,
      "step": 121028
    },
    {
      "epoch": 1.58,
      "grad_norm": 1.0546914339065552,
      "learning_rate": 0.00027618732533145053,
      "loss": 2.1218,
      "step": 121029
    },
    {
      "epoch": 1.58,
      "grad_norm": 1.1120762825012207,
      "learning_rate": 0.0002761832476781128,
      "loss": 1.9362,
      "step": 121030
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.9337878823280334,
      "learning_rate": 0.0002761791700292032,
      "loss": 2.1476,
      "step": 121031
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.9411716461181641,
      "learning_rate": 0.00027617509238472235,
      "loss": 2.095,
      "step": 121032
    },
    {
      "epoch": 1.58,
      "grad_norm": 1.2474350929260254,
      "learning_rate": 0.00027617101474467084,
      "loss": 2.2898,
      "step": 121033
    },
    {
      "epoch": 1.58,
      "grad_norm": 1.2432032823562622,
      "learning_rate": 0.00027616693710904965,
      "loss": 2.0965,
      "step": 121034
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.9961545467376709,
      "learning_rate": 0.0002761628594778595,
      "loss": 2.2491,
      "step": 121035
    },
    {
      "epoch": 1.58,
      "grad_norm": 1.3431272506713867,
      "learning_rate": 0.0002761587818511011,
      "loss": 2.0333,
      "step": 121036
    },
    {
      "epoch": 1.58,
      "grad_norm": 1.560631513595581,
      "learning_rate": 0.00027615470422877526,
      "loss": 2.1149,
      "step": 121037
    },
    {
      "epoch": 1.58,
      "grad_norm": 1.0829527378082275,
      "learning_rate": 0.00027615062661088266,
      "loss": 2.1085,
      "step": 121038
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.9379690289497375,
      "learning_rate": 0.0002761465489974242,
      "loss": 2.1528,
      "step": 121039
    },
    {
      "epoch": 1.58,
      "grad_norm": 1.053004503250122,
      "learning_rate": 0.0002761424713884005,
      "loss": 2.0511,
      "step": 121040
    },
    {
      "epoch": 1.58,
      "grad_norm": 1.2636250257492065,
      "learning_rate": 0.0002761383937838123,
      "loss": 2.2447,
      "step": 121041
    },
    {
      "epoch": 1.58,
      "grad_norm": 1.0003653764724731,
      "learning_rate": 0.0002761343161836605,
      "loss": 2.229,
      "step": 121042
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.873555064201355,
      "learning_rate": 0.0002761302385879457,
      "loss": 2.2213,
      "step": 121043
    },
    {
      "epoch": 1.58,
      "grad_norm": 1.1295177936553955,
      "learning_rate": 0.00027612616099666887,
      "loss": 2.0029,
      "step": 121044
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.9209751486778259,
      "learning_rate": 0.00027612208340983054,
      "loss": 2.0759,
      "step": 121045
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.9952700734138489,
      "learning_rate": 0.00027611800582743164,
      "loss": 2.1805,
      "step": 121046
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.9232771396636963,
      "learning_rate": 0.00027611392824947276,
      "loss": 2.2121,
      "step": 121047
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.8682316541671753,
      "learning_rate": 0.0002761098506759548,
      "loss": 2.0759,
      "step": 121048
    },
    {
      "epoch": 1.58,
      "grad_norm": 1.063442349433899,
      "learning_rate": 0.00027610577310687844,
      "loss": 2.168,
      "step": 121049
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.9077919125556946,
      "learning_rate": 0.0002761016955422445,
      "loss": 2.2082,
      "step": 121050
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.9890677332878113,
      "learning_rate": 0.0002760976179820537,
      "loss": 2.1998,
      "step": 121051
    },
    {
      "epoch": 1.58,
      "grad_norm": 1.0147730112075806,
      "learning_rate": 0.00027609354042630683,
      "loss": 1.9537,
      "step": 121052
    },
    {
      "epoch": 1.58,
      "grad_norm": 1.371816635131836,
      "learning_rate": 0.0002760894628750046,
      "loss": 2.1525,
      "step": 121053
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.933370053768158,
      "learning_rate": 0.0002760853853281478,
      "loss": 2.0177,
      "step": 121054
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.9096794724464417,
      "learning_rate": 0.0002760813077857371,
      "loss": 2.3591,
      "step": 121055
    },
    {
      "epoch": 1.58,
      "grad_norm": 1.0372220277786255,
      "learning_rate": 0.00027607723024777344,
      "loss": 2.2293,
      "step": 121056
    },
    {
      "epoch": 1.58,
      "grad_norm": 1.0832573175430298,
      "learning_rate": 0.0002760731527142574,
      "loss": 2.0359,
      "step": 121057
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.9513724446296692,
      "learning_rate": 0.0002760690751851898,
      "loss": 2.2766,
      "step": 121058
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.9225418567657471,
      "learning_rate": 0.00027606499766057147,
      "loss": 2.0183,
      "step": 121059
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.9273733496665955,
      "learning_rate": 0.00027606092014040306,
      "loss": 2.2216,
      "step": 121060
    },
    {
      "epoch": 1.58,
      "grad_norm": 1.248477816581726,
      "learning_rate": 0.00027605684262468547,
      "loss": 1.9856,
      "step": 121061
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.9076962471008301,
      "learning_rate": 0.0002760527651134194,
      "loss": 2.014,
      "step": 121062
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.9633525609970093,
      "learning_rate": 0.0002760486876066055,
      "loss": 1.8947,
      "step": 121063
    },
    {
      "epoch": 1.58,
      "grad_norm": 1.0742069482803345,
      "learning_rate": 0.0002760446101042445,
      "loss": 2.1423,
      "step": 121064
    },
    {
      "epoch": 1.58,
      "grad_norm": 1.1352300643920898,
      "learning_rate": 0.0002760405326063373,
      "loss": 2.21,
      "step": 121065
    },
    {
      "epoch": 1.58,
      "grad_norm": 1.0923558473587036,
      "learning_rate": 0.0002760364551128847,
      "loss": 2.0943,
      "step": 121066
    },
    {
      "epoch": 1.58,
      "grad_norm": 1.0093077421188354,
      "learning_rate": 0.0002760323776238873,
      "loss": 2.306,
      "step": 121067
    },
    {
      "epoch": 1.58,
      "grad_norm": 1.0808833837509155,
      "learning_rate": 0.00027602830013934604,
      "loss": 2.3063,
      "step": 121068
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.9909524917602539,
      "learning_rate": 0.0002760242226592615,
      "loss": 1.9461,
      "step": 121069
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.9738292098045349,
      "learning_rate": 0.0002760201451836345,
      "loss": 2.1455,
      "step": 121070
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.9363069534301758,
      "learning_rate": 0.0002760160677124658,
      "loss": 1.7907,
      "step": 121071
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.9382286667823792,
      "learning_rate": 0.00027601199024575616,
      "loss": 2.2767,
      "step": 121072
    },
    {
      "epoch": 1.58,
      "grad_norm": 1.1087534427642822,
      "learning_rate": 0.0002760079127835063,
      "loss": 2.1684,
      "step": 121073
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.9514402747154236,
      "learning_rate": 0.0002760038353257172,
      "loss": 2.3032,
      "step": 121074
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.9301964640617371,
      "learning_rate": 0.00027599975787238933,
      "loss": 2.2713,
      "step": 121075
    },
    {
      "epoch": 1.58,
      "grad_norm": 1.0848373174667358,
      "learning_rate": 0.00027599568042352353,
      "loss": 2.0018,
      "step": 121076
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.9187929630279541,
      "learning_rate": 0.0002759916029791206,
      "loss": 2.228,
      "step": 121077
    },
    {
      "epoch": 1.58,
      "grad_norm": 1.2503702640533447,
      "learning_rate": 0.0002759875255391813,
      "loss": 1.8865,
      "step": 121078
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.9794861078262329,
      "learning_rate": 0.0002759834481037063,
      "loss": 1.8766,
      "step": 121079
    },
    {
      "epoch": 1.58,
      "grad_norm": 1.0953538417816162,
      "learning_rate": 0.0002759793706726967,
      "loss": 2.2603,
      "step": 121080
    },
    {
      "epoch": 1.58,
      "grad_norm": 1.1936544179916382,
      "learning_rate": 0.0002759752932461527,
      "loss": 1.8509,
      "step": 121081
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.9585419297218323,
      "learning_rate": 0.0002759712158240754,
      "loss": 2.1071,
      "step": 121082
    },
    {
      "epoch": 1.58,
      "grad_norm": 1.035520315170288,
      "learning_rate": 0.00027596713840646555,
      "loss": 2.3216,
      "step": 121083
    },
    {
      "epoch": 1.58,
      "grad_norm": 1.2572599649429321,
      "learning_rate": 0.00027596306099332384,
      "loss": 1.9632,
      "step": 121084
    },
    {
      "epoch": 1.58,
      "grad_norm": 1.0066725015640259,
      "learning_rate": 0.000275958983584651,
      "loss": 1.9081,
      "step": 121085
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.9195845127105713,
      "learning_rate": 0.00027595490618044803,
      "loss": 2.1172,
      "step": 121086
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.9387887716293335,
      "learning_rate": 0.00027595082878071535,
      "loss": 2.1912,
      "step": 121087
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.9012555480003357,
      "learning_rate": 0.00027594675138545386,
      "loss": 1.9165,
      "step": 121088
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.8697383999824524,
      "learning_rate": 0.00027594267399466433,
      "loss": 2.0028,
      "step": 121089
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.9593600034713745,
      "learning_rate": 0.0002759385966083475,
      "loss": 2.2848,
      "step": 121090
    },
    {
      "epoch": 1.58,
      "grad_norm": 1.0835949182510376,
      "learning_rate": 0.00027593451922650416,
      "loss": 2.3235,
      "step": 121091
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.9126830697059631,
      "learning_rate": 0.00027593044184913516,
      "loss": 2.3436,
      "step": 121092
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.9039649367332458,
      "learning_rate": 0.000275926364476241,
      "loss": 2.2384,
      "step": 121093
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.9319958090782166,
      "learning_rate": 0.0002759222871078226,
      "loss": 2.2275,
      "step": 121094
    },
    {
      "epoch": 1.58,
      "grad_norm": 1.0209556818008423,
      "learning_rate": 0.00027591820974388073,
      "loss": 1.9636,
      "step": 121095
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.8913139700889587,
      "learning_rate": 0.0002759141323844161,
      "loss": 2.3604,
      "step": 121096
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.9450960755348206,
      "learning_rate": 0.00027591005502942945,
      "loss": 2.0751,
      "step": 121097
    },
    {
      "epoch": 1.58,
      "grad_norm": 1.0570122003555298,
      "learning_rate": 0.00027590597767892177,
      "loss": 2.2086,
      "step": 121098
    },
    {
      "epoch": 1.58,
      "grad_norm": 1.0807236433029175,
      "learning_rate": 0.0002759019003328934,
      "loss": 1.8296,
      "step": 121099
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.9373151659965515,
      "learning_rate": 0.0002758978229913454,
      "loss": 2.0265,
      "step": 121100
    },
    {
      "epoch": 1.58,
      "grad_norm": 1.0151005983352661,
      "learning_rate": 0.00027589374565427843,
      "loss": 2.0098,
      "step": 121101
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.950509786605835,
      "learning_rate": 0.00027588966832169326,
      "loss": 2.3546,
      "step": 121102
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.9896214008331299,
      "learning_rate": 0.0002758855909935907,
      "loss": 2.1787,
      "step": 121103
    },
    {
      "epoch": 1.58,
      "grad_norm": 1.0206258296966553,
      "learning_rate": 0.00027588151366997153,
      "loss": 2.087,
      "step": 121104
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.9089245200157166,
      "learning_rate": 0.0002758774363508363,
      "loss": 2.0382,
      "step": 121105
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.9355217218399048,
      "learning_rate": 0.000275873359036186,
      "loss": 1.9494,
      "step": 121106
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.9748822450637817,
      "learning_rate": 0.00027586928172602124,
      "loss": 2.2924,
      "step": 121107
    },
    {
      "epoch": 1.58,
      "grad_norm": 1.019302248954773,
      "learning_rate": 0.0002758652044203428,
      "loss": 2.2399,
      "step": 121108
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.9796349406242371,
      "learning_rate": 0.00027586112711915154,
      "loss": 2.3074,
      "step": 121109
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.924834132194519,
      "learning_rate": 0.00027585704982244815,
      "loss": 2.0241,
      "step": 121110
    },
    {
      "epoch": 1.58,
      "grad_norm": 1.026499629020691,
      "learning_rate": 0.00027585297253023347,
      "loss": 2.1317,
      "step": 121111
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.9443274140357971,
      "learning_rate": 0.0002758488952425081,
      "loss": 2.0859,
      "step": 121112
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.9111996293067932,
      "learning_rate": 0.0002758448179592728,
      "loss": 2.1317,
      "step": 121113
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.9101355075836182,
      "learning_rate": 0.00027584074068052854,
      "loss": 2.1953,
      "step": 121114
    },
    {
      "epoch": 1.58,
      "grad_norm": 1.0544270277023315,
      "learning_rate": 0.00027583666340627583,
      "loss": 2.1021,
      "step": 121115
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.9193676710128784,
      "learning_rate": 0.00027583258613651557,
      "loss": 2.1905,
      "step": 121116
    },
    {
      "epoch": 1.58,
      "grad_norm": 1.0407060384750366,
      "learning_rate": 0.0002758285088712486,
      "loss": 2.2298,
      "step": 121117
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.9627020955085754,
      "learning_rate": 0.0002758244316104755,
      "loss": 2.0506,
      "step": 121118
    },
    {
      "epoch": 1.58,
      "grad_norm": 1.0919326543807983,
      "learning_rate": 0.00027582035435419705,
      "loss": 2.0999,
      "step": 121119
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.8610050082206726,
      "learning_rate": 0.000275816277102414,
      "loss": 2.1493,
      "step": 121120
    },
    {
      "epoch": 1.58,
      "grad_norm": 1.0518465042114258,
      "learning_rate": 0.00027581219985512725,
      "loss": 2.0587,
      "step": 121121
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.9204607605934143,
      "learning_rate": 0.00027580812261233744,
      "loss": 1.9458,
      "step": 121122
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.8965474963188171,
      "learning_rate": 0.0002758040453740454,
      "loss": 2.0744,
      "step": 121123
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.9475135207176208,
      "learning_rate": 0.00027579996814025185,
      "loss": 2.0595,
      "step": 121124
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.9340869188308716,
      "learning_rate": 0.0002757958909109575,
      "loss": 1.9939,
      "step": 121125
    },
    {
      "epoch": 1.58,
      "grad_norm": 1.053584098815918,
      "learning_rate": 0.00027579181368616316,
      "loss": 2.3527,
      "step": 121126
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.9631457328796387,
      "learning_rate": 0.0002757877364658695,
      "loss": 2.1325,
      "step": 121127
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.9158280491828918,
      "learning_rate": 0.0002757836592500774,
      "loss": 1.878,
      "step": 121128
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.9890400171279907,
      "learning_rate": 0.0002757795820387877,
      "loss": 2.1964,
      "step": 121129
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.9800374507904053,
      "learning_rate": 0.00027577550483200093,
      "loss": 1.9863,
      "step": 121130
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.8614299297332764,
      "learning_rate": 0.000275771427629718,
      "loss": 1.9663,
      "step": 121131
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.985451340675354,
      "learning_rate": 0.00027576735043193955,
      "loss": 2.1585,
      "step": 121132
    },
    {
      "epoch": 1.58,
      "grad_norm": 1.102048635482788,
      "learning_rate": 0.0002757632732386664,
      "loss": 2.2113,
      "step": 121133
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.9467813372612,
      "learning_rate": 0.00027575919604989934,
      "loss": 2.047,
      "step": 121134
    },
    {
      "epoch": 1.58,
      "grad_norm": 1.1081922054290771,
      "learning_rate": 0.00027575511886563915,
      "loss": 1.9828,
      "step": 121135
    },
    {
      "epoch": 1.58,
      "grad_norm": 1.0422987937927246,
      "learning_rate": 0.0002757510416858865,
      "loss": 2.2521,
      "step": 121136
    },
    {
      "epoch": 1.58,
      "grad_norm": 1.29982590675354,
      "learning_rate": 0.0002757469645106422,
      "loss": 2.1992,
      "step": 121137
    },
    {
      "epoch": 1.58,
      "grad_norm": 1.1073130369186401,
      "learning_rate": 0.000275742887339907,
      "loss": 2.1519,
      "step": 121138
    },
    {
      "epoch": 1.58,
      "grad_norm": 1.0812714099884033,
      "learning_rate": 0.00027573881017368165,
      "loss": 2.2003,
      "step": 121139
    },
    {
      "epoch": 1.58,
      "grad_norm": 1.002299427986145,
      "learning_rate": 0.00027573473301196685,
      "loss": 2.1241,
      "step": 121140
    },
    {
      "epoch": 1.58,
      "grad_norm": 1.1341361999511719,
      "learning_rate": 0.00027573065585476353,
      "loss": 1.9762,
      "step": 121141
    },
    {
      "epoch": 1.58,
      "grad_norm": 1.1992084980010986,
      "learning_rate": 0.0002757265787020723,
      "loss": 2.2248,
      "step": 121142
    },
    {
      "epoch": 1.58,
      "grad_norm": 1.0211750268936157,
      "learning_rate": 0.0002757225015538939,
      "loss": 2.1069,
      "step": 121143
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.9880161285400391,
      "learning_rate": 0.0002757184244102292,
      "loss": 1.9548,
      "step": 121144
    },
    {
      "epoch": 1.58,
      "grad_norm": 1.1810667514801025,
      "learning_rate": 0.00027571434727107896,
      "loss": 2.1513,
      "step": 121145
    },
    {
      "epoch": 1.58,
      "grad_norm": 1.1110607385635376,
      "learning_rate": 0.0002757102701364438,
      "loss": 2.1906,
      "step": 121146
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.9087345004081726,
      "learning_rate": 0.0002757061930063246,
      "loss": 2.0954,
      "step": 121147
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.9901875853538513,
      "learning_rate": 0.000275702115880722,
      "loss": 2.0931,
      "step": 121148
    },
    {
      "epoch": 1.58,
      "grad_norm": 1.1159944534301758,
      "learning_rate": 0.0002756980387596369,
      "loss": 2.1476,
      "step": 121149
    },
    {
      "epoch": 1.58,
      "grad_norm": 1.0344533920288086,
      "learning_rate": 0.00027569396164306995,
      "loss": 2.0207,
      "step": 121150
    },
    {
      "epoch": 1.58,
      "grad_norm": 1.0438894033432007,
      "learning_rate": 0.00027568988453102196,
      "loss": 2.0268,
      "step": 121151
    },
    {
      "epoch": 1.58,
      "grad_norm": 1.0828595161437988,
      "learning_rate": 0.0002756858074234937,
      "loss": 2.2266,
      "step": 121152
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.9824777841567993,
      "learning_rate": 0.000275681730320486,
      "loss": 2.005,
      "step": 121153
    },
    {
      "epoch": 1.58,
      "grad_norm": 1.059200644493103,
      "learning_rate": 0.00027567765322199944,
      "loss": 2.0382,
      "step": 121154
    },
    {
      "epoch": 1.58,
      "grad_norm": 1.1478071212768555,
      "learning_rate": 0.0002756735761280348,
      "loss": 2.2261,
      "step": 121155
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.964885950088501,
      "learning_rate": 0.000275669499038593,
      "loss": 2.1579,
      "step": 121156
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.916028618812561,
      "learning_rate": 0.0002756654219536746,
      "loss": 1.9435,
      "step": 121157
    },
    {
      "epoch": 1.58,
      "grad_norm": 1.0552529096603394,
      "learning_rate": 0.0002756613448732805,
      "loss": 2.1928,
      "step": 121158
    },
    {
      "epoch": 1.58,
      "grad_norm": 1.0867841243743896,
      "learning_rate": 0.00027565726779741153,
      "loss": 1.9227,
      "step": 121159
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.9390304088592529,
      "learning_rate": 0.00027565319072606824,
      "loss": 2.1343,
      "step": 121160
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.9996792078018188,
      "learning_rate": 0.0002756491136592515,
      "loss": 2.1502,
      "step": 121161
    },
    {
      "epoch": 1.58,
      "grad_norm": 1.0559216737747192,
      "learning_rate": 0.00027564503659696193,
      "loss": 2.2486,
      "step": 121162
    },
    {
      "epoch": 1.58,
      "grad_norm": 1.1251243352890015,
      "learning_rate": 0.00027564095953920055,
      "loss": 1.893,
      "step": 121163
    },
    {
      "epoch": 1.58,
      "grad_norm": 1.193401575088501,
      "learning_rate": 0.0002756368824859679,
      "loss": 2.0838,
      "step": 121164
    },
    {
      "epoch": 1.58,
      "grad_norm": 1.1816470623016357,
      "learning_rate": 0.00027563280543726494,
      "loss": 2.1785,
      "step": 121165
    },
    {
      "epoch": 1.58,
      "grad_norm": 1.103438138961792,
      "learning_rate": 0.0002756287283930922,
      "loss": 1.9971,
      "step": 121166
    },
    {
      "epoch": 1.58,
      "grad_norm": 1.0500015020370483,
      "learning_rate": 0.0002756246513534505,
      "loss": 2.1187,
      "step": 121167
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.921573281288147,
      "learning_rate": 0.0002756205743183407,
      "loss": 2.0535,
      "step": 121168
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.9052635431289673,
      "learning_rate": 0.00027561649728776344,
      "loss": 2.0388,
      "step": 121169
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.9370849132537842,
      "learning_rate": 0.0002756124202617195,
      "loss": 2.1089,
      "step": 121170
    },
    {
      "epoch": 1.58,
      "grad_norm": 1.0704997777938843,
      "learning_rate": 0.0002756083432402099,
      "loss": 2.172,
      "step": 121171
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.9172676801681519,
      "learning_rate": 0.00027560426622323496,
      "loss": 2.1861,
      "step": 121172
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.9865330457687378,
      "learning_rate": 0.0002756001892107957,
      "loss": 2.1041,
      "step": 121173
    },
    {
      "epoch": 1.58,
      "grad_norm": 1.021705150604248,
      "learning_rate": 0.00027559611220289283,
      "loss": 2.1414,
      "step": 121174
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.9312870502471924,
      "learning_rate": 0.00027559203519952706,
      "loss": 2.1343,
      "step": 121175
    },
    {
      "epoch": 1.58,
      "grad_norm": 1.0918387174606323,
      "learning_rate": 0.0002755879582006992,
      "loss": 1.8318,
      "step": 121176
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.9147450923919678,
      "learning_rate": 0.00027558388120641,
      "loss": 2.1469,
      "step": 121177
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.9842745065689087,
      "learning_rate": 0.00027557980421666036,
      "loss": 2.0355,
      "step": 121178
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.973065197467804,
      "learning_rate": 0.00027557572723145075,
      "loss": 1.9077,
      "step": 121179
    },
    {
      "epoch": 1.58,
      "grad_norm": 1.2327080965042114,
      "learning_rate": 0.0002755716502507821,
      "loss": 2.2244,
      "step": 121180
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.9490468502044678,
      "learning_rate": 0.0002755675732746551,
      "loss": 1.9782,
      "step": 121181
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.9814397692680359,
      "learning_rate": 0.00027556349630307057,
      "loss": 2.258,
      "step": 121182
    },
    {
      "epoch": 1.58,
      "grad_norm": 1.1539661884307861,
      "learning_rate": 0.0002755594193360293,
      "loss": 2.1891,
      "step": 121183
    },
    {
      "epoch": 1.58,
      "grad_norm": 1.080946445465088,
      "learning_rate": 0.00027555534237353206,
      "loss": 2.1011,
      "step": 121184
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.9768550395965576,
      "learning_rate": 0.00027555126541557945,
      "loss": 1.9729,
      "step": 121185
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.9442581534385681,
      "learning_rate": 0.0002755471884621723,
      "loss": 2.1776,
      "step": 121186
    },
    {
      "epoch": 1.58,
      "grad_norm": 1.1596434116363525,
      "learning_rate": 0.0002755431115133114,
      "loss": 2.2043,
      "step": 121187
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.9653837084770203,
      "learning_rate": 0.00027553903456899745,
      "loss": 2.0932,
      "step": 121188
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.9830206036567688,
      "learning_rate": 0.0002755349576292313,
      "loss": 2.1146,
      "step": 121189
    },
    {
      "epoch": 1.58,
      "grad_norm": 1.162585973739624,
      "learning_rate": 0.0002755308806940138,
      "loss": 2.2412,
      "step": 121190
    },
    {
      "epoch": 1.58,
      "grad_norm": 1.0573500394821167,
      "learning_rate": 0.00027552680376334544,
      "loss": 2.1766,
      "step": 121191
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.9901021122932434,
      "learning_rate": 0.00027552272683722707,
      "loss": 2.385,
      "step": 121192
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.9950656890869141,
      "learning_rate": 0.0002755186499156595,
      "loss": 2.0391,
      "step": 121193
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.9349876046180725,
      "learning_rate": 0.0002755145729986435,
      "loss": 2.1544,
      "step": 121194
    },
    {
      "epoch": 1.58,
      "grad_norm": 1.1055556535720825,
      "learning_rate": 0.0002755104960861798,
      "loss": 2.1263,
      "step": 121195
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.9383710026741028,
      "learning_rate": 0.0002755064191782693,
      "loss": 2.1013,
      "step": 121196
    },
    {
      "epoch": 1.58,
      "grad_norm": 1.117414951324463,
      "learning_rate": 0.00027550234227491247,
      "loss": 2.0987,
      "step": 121197
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.936769962310791,
      "learning_rate": 0.0002754982653761102,
      "loss": 1.9759,
      "step": 121198
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.9645493626594543,
      "learning_rate": 0.0002754941884818633,
      "loss": 2.3566,
      "step": 121199
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.9376348853111267,
      "learning_rate": 0.0002754901115921725,
      "loss": 2.1398,
      "step": 121200
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.9587042927742004,
      "learning_rate": 0.00027548603470703854,
      "loss": 2.1289,
      "step": 121201
    },
    {
      "epoch": 1.58,
      "grad_norm": 1.035000205039978,
      "learning_rate": 0.0002754819578264623,
      "loss": 2.4316,
      "step": 121202
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.8835591077804565,
      "learning_rate": 0.0002754778809504443,
      "loss": 2.1369,
      "step": 121203
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.9193483591079712,
      "learning_rate": 0.00027547380407898545,
      "loss": 2.1426,
      "step": 121204
    },
    {
      "epoch": 1.58,
      "grad_norm": 1.028437614440918,
      "learning_rate": 0.0002754697272120865,
      "loss": 2.3322,
      "step": 121205
    },
    {
      "epoch": 1.58,
      "grad_norm": 1.021705150604248,
      "learning_rate": 0.0002754656503497481,
      "loss": 2.0086,
      "step": 121206
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.9631360173225403,
      "learning_rate": 0.00027546157349197115,
      "loss": 2.1178,
      "step": 121207
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.9469364881515503,
      "learning_rate": 0.0002754574966387565,
      "loss": 2.009,
      "step": 121208
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.9892094731330872,
      "learning_rate": 0.00027545341979010464,
      "loss": 2.0428,
      "step": 121209
    },
    {
      "epoch": 1.58,
      "grad_norm": 1.0143420696258545,
      "learning_rate": 0.0002754493429460164,
      "loss": 1.9551,
      "step": 121210
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.8798894286155701,
      "learning_rate": 0.0002754452661064926,
      "loss": 2.0416,
      "step": 121211
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.9322465062141418,
      "learning_rate": 0.000275441189271534,
      "loss": 1.8613,
      "step": 121212
    },
    {
      "epoch": 1.58,
      "grad_norm": 1.0443075895309448,
      "learning_rate": 0.00027543711244114137,
      "loss": 2.1417,
      "step": 121213
    },
    {
      "epoch": 1.58,
      "grad_norm": 1.0278511047363281,
      "learning_rate": 0.00027543303561531555,
      "loss": 2.0951,
      "step": 121214
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.9726586937904358,
      "learning_rate": 0.00027542895879405704,
      "loss": 2.191,
      "step": 121215
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.9882378578186035,
      "learning_rate": 0.00027542488197736687,
      "loss": 2.0734,
      "step": 121216
    },
    {
      "epoch": 1.58,
      "grad_norm": 1.019132137298584,
      "learning_rate": 0.0002754208051652456,
      "loss": 2.0151,
      "step": 121217
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.9426589608192444,
      "learning_rate": 0.0002754167283576941,
      "loss": 2.2478,
      "step": 121218
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.9635043740272522,
      "learning_rate": 0.000275412651554713,
      "loss": 2.2153,
      "step": 121219
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.8838364481925964,
      "learning_rate": 0.0002754085747563033,
      "loss": 2.0308,
      "step": 121220
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.9053271412849426,
      "learning_rate": 0.0002754044979624655,
      "loss": 2.0201,
      "step": 121221
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.9046139121055603,
      "learning_rate": 0.0002754004211732005,
      "loss": 2.1274,
      "step": 121222
    },
    {
      "epoch": 1.58,
      "grad_norm": 1.0141962766647339,
      "learning_rate": 0.0002753963443885091,
      "loss": 2.2833,
      "step": 121223
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.884669303894043,
      "learning_rate": 0.0002753922676083919,
      "loss": 2.125,
      "step": 121224
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.9183580875396729,
      "learning_rate": 0.00027538819083284974,
      "loss": 2.1412,
      "step": 121225
    },
    {
      "epoch": 1.58,
      "grad_norm": 1.029980182647705,
      "learning_rate": 0.00027538411406188346,
      "loss": 2.265,
      "step": 121226
    },
    {
      "epoch": 1.58,
      "grad_norm": 1.0058397054672241,
      "learning_rate": 0.00027538003729549366,
      "loss": 2.2727,
      "step": 121227
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.9845098853111267,
      "learning_rate": 0.00027537596053368114,
      "loss": 1.8825,
      "step": 121228
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.8791021704673767,
      "learning_rate": 0.0002753718837764467,
      "loss": 2.1622,
      "step": 121229
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.8667739033699036,
      "learning_rate": 0.0002753678070237912,
      "loss": 2.2493,
      "step": 121230
    },
    {
      "epoch": 1.58,
      "grad_norm": 1.01139497756958,
      "learning_rate": 0.0002753637302757152,
      "loss": 2.2683,
      "step": 121231
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.9532273411750793,
      "learning_rate": 0.00027535965353221964,
      "loss": 2.0079,
      "step": 121232
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.8766674995422363,
      "learning_rate": 0.0002753555767933051,
      "loss": 2.2625,
      "step": 121233
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.927229106426239,
      "learning_rate": 0.00027535150005897243,
      "loss": 2.0271,
      "step": 121234
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.9406182169914246,
      "learning_rate": 0.0002753474233292224,
      "loss": 2.0284,
      "step": 121235
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.9213380813598633,
      "learning_rate": 0.00027534334660405573,
      "loss": 1.934,
      "step": 121236
    },
    {
      "epoch": 1.58,
      "grad_norm": 1.009497880935669,
      "learning_rate": 0.00027533926988347325,
      "loss": 2.1908,
      "step": 121237
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.9058103561401367,
      "learning_rate": 0.0002753351931674757,
      "loss": 2.0747,
      "step": 121238
    },
    {
      "epoch": 1.58,
      "grad_norm": 1.0365339517593384,
      "learning_rate": 0.0002753311164560637,
      "loss": 2.143,
      "step": 121239
    },
    {
      "epoch": 1.58,
      "grad_norm": 1.129612922668457,
      "learning_rate": 0.0002753270397492381,
      "loss": 2.0455,
      "step": 121240
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.9760376811027527,
      "learning_rate": 0.00027532296304699967,
      "loss": 2.207,
      "step": 121241
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.9349914193153381,
      "learning_rate": 0.0002753188863493492,
      "loss": 2.1917,
      "step": 121242
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.9696623682975769,
      "learning_rate": 0.0002753148096562874,
      "loss": 2.2904,
      "step": 121243
    },
    {
      "epoch": 1.58,
      "grad_norm": 1.052437663078308,
      "learning_rate": 0.0002753107329678151,
      "loss": 2.2974,
      "step": 121244
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.953840434551239,
      "learning_rate": 0.000275306656283933,
      "loss": 2.1137,
      "step": 121245
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.9071817994117737,
      "learning_rate": 0.00027530257960464184,
      "loss": 2.2393,
      "step": 121246
    },
    {
      "epoch": 1.58,
      "grad_norm": 1.02971613407135,
      "learning_rate": 0.00027529850292994237,
      "loss": 2.0756,
      "step": 121247
    },
    {
      "epoch": 1.58,
      "grad_norm": 1.0208799839019775,
      "learning_rate": 0.00027529442625983536,
      "loss": 2.0114,
      "step": 121248
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.8894131779670715,
      "learning_rate": 0.00027529034959432157,
      "loss": 2.1691,
      "step": 121249
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.9274392127990723,
      "learning_rate": 0.0002752862729334018,
      "loss": 2.0971,
      "step": 121250
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.9347901940345764,
      "learning_rate": 0.0002752821962770769,
      "loss": 2.0333,
      "step": 121251
    },
    {
      "epoch": 1.58,
      "grad_norm": 1.0402809381484985,
      "learning_rate": 0.0002752781196253474,
      "loss": 2.1485,
      "step": 121252
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.944754421710968,
      "learning_rate": 0.00027527404297821413,
      "loss": 2.0556,
      "step": 121253
    },
    {
      "epoch": 1.58,
      "grad_norm": 1.0881332159042358,
      "learning_rate": 0.0002752699663356779,
      "loss": 2.0431,
      "step": 121254
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.9878695607185364,
      "learning_rate": 0.00027526588969773943,
      "loss": 2.112,
      "step": 121255
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.9176138639450073,
      "learning_rate": 0.0002752618130643995,
      "loss": 2.2443,
      "step": 121256
    },
    {
      "epoch": 1.58,
      "grad_norm": 1.0820924043655396,
      "learning_rate": 0.000275257736435659,
      "loss": 2.197,
      "step": 121257
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.9375891089439392,
      "learning_rate": 0.00027525365981151845,
      "loss": 2.052,
      "step": 121258
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.9879352450370789,
      "learning_rate": 0.0002752495831919787,
      "loss": 2.0892,
      "step": 121259
    },
    {
      "epoch": 1.58,
      "grad_norm": 1.03811514377594,
      "learning_rate": 0.0002752455065770405,
      "loss": 2.1434,
      "step": 121260
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.9764305353164673,
      "learning_rate": 0.0002752414299667046,
      "loss": 2.1105,
      "step": 121261
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.9378333687782288,
      "learning_rate": 0.0002752373533609719,
      "loss": 2.2239,
      "step": 121262
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.9374097585678101,
      "learning_rate": 0.00027523327675984307,
      "loss": 2.1397,
      "step": 121263
    },
    {
      "epoch": 1.58,
      "grad_norm": 1.043178915977478,
      "learning_rate": 0.00027522920016331874,
      "loss": 2.1588,
      "step": 121264
    },
    {
      "epoch": 1.58,
      "grad_norm": 1.0440322160720825,
      "learning_rate": 0.00027522512357139973,
      "loss": 2.115,
      "step": 121265
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.9025572538375854,
      "learning_rate": 0.0002752210469840869,
      "loss": 2.2175,
      "step": 121266
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.9115965962409973,
      "learning_rate": 0.00027521697040138094,
      "loss": 2.2138,
      "step": 121267
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.9793772101402283,
      "learning_rate": 0.0002752128938232826,
      "loss": 1.9872,
      "step": 121268
    },
    {
      "epoch": 1.58,
      "grad_norm": 1.05935800075531,
      "learning_rate": 0.0002752088172497927,
      "loss": 2.0051,
      "step": 121269
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.9435394406318665,
      "learning_rate": 0.0002752047406809119,
      "loss": 2.1358,
      "step": 121270
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.9349451065063477,
      "learning_rate": 0.00027520066411664104,
      "loss": 2.3287,
      "step": 121271
    },
    {
      "epoch": 1.58,
      "grad_norm": 1.1174713373184204,
      "learning_rate": 0.00027519658755698084,
      "loss": 1.7489,
      "step": 121272
    },
    {
      "epoch": 1.58,
      "grad_norm": 1.0815759897232056,
      "learning_rate": 0.000275192511001932,
      "loss": 2.1986,
      "step": 121273
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.9422692060470581,
      "learning_rate": 0.00027518843445149534,
      "loss": 2.3149,
      "step": 121274
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.8843641877174377,
      "learning_rate": 0.00027518435790567183,
      "loss": 2.1927,
      "step": 121275
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.9624281525611877,
      "learning_rate": 0.0002751802813644618,
      "loss": 2.3467,
      "step": 121276
    },
    {
      "epoch": 1.58,
      "grad_norm": 1.0321338176727295,
      "learning_rate": 0.0002751762048278663,
      "loss": 2.1214,
      "step": 121277
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.9587427973747253,
      "learning_rate": 0.0002751721282958859,
      "loss": 2.1929,
      "step": 121278
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.9128378629684448,
      "learning_rate": 0.00027516805176852155,
      "loss": 2.1195,
      "step": 121279
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.9554420113563538,
      "learning_rate": 0.00027516397524577393,
      "loss": 2.27,
      "step": 121280
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.9932379126548767,
      "learning_rate": 0.0002751598987276439,
      "loss": 2.2445,
      "step": 121281
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.9492295980453491,
      "learning_rate": 0.000275155822214132,
      "loss": 2.2911,
      "step": 121282
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.8673222064971924,
      "learning_rate": 0.0002751517457052391,
      "loss": 2.0064,
      "step": 121283
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.890648603439331,
      "learning_rate": 0.00027514766920096596,
      "loss": 2.0647,
      "step": 121284
    },
    {
      "epoch": 1.58,
      "grad_norm": 1.032379150390625,
      "learning_rate": 0.0002751435927013133,
      "loss": 2.3628,
      "step": 121285
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.9890187382698059,
      "learning_rate": 0.000275139516206282,
      "loss": 2.179,
      "step": 121286
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.9425196647644043,
      "learning_rate": 0.0002751354397158728,
      "loss": 1.9986,
      "step": 121287
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.988664448261261,
      "learning_rate": 0.0002751313632300863,
      "loss": 2.0551,
      "step": 121288
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.9560622572898865,
      "learning_rate": 0.0002751272867489233,
      "loss": 2.1637,
      "step": 121289
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.8800497055053711,
      "learning_rate": 0.00027512321027238463,
      "loss": 2.114,
      "step": 121290
    },
    {
      "epoch": 1.58,
      "grad_norm": 1.2731459140777588,
      "learning_rate": 0.000275119133800471,
      "loss": 1.9592,
      "step": 121291
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.9325686097145081,
      "learning_rate": 0.00027511505733318327,
      "loss": 2.0227,
      "step": 121292
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.991361141204834,
      "learning_rate": 0.0002751109808705222,
      "loss": 2.2911,
      "step": 121293
    },
    {
      "epoch": 1.58,
      "grad_norm": 1.1146283149719238,
      "learning_rate": 0.0002751069044124883,
      "loss": 2.2327,
      "step": 121294
    },
    {
      "epoch": 1.58,
      "grad_norm": 1.0716338157653809,
      "learning_rate": 0.00027510282795908253,
      "loss": 1.8603,
      "step": 121295
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.952349841594696,
      "learning_rate": 0.00027509875151030563,
      "loss": 1.936,
      "step": 121296
    },
    {
      "epoch": 1.58,
      "grad_norm": 1.1569936275482178,
      "learning_rate": 0.00027509467506615833,
      "loss": 2.1164,
      "step": 121297
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.954177737236023,
      "learning_rate": 0.0002750905986266414,
      "loss": 2.264,
      "step": 121298
    },
    {
      "epoch": 1.58,
      "grad_norm": 1.0552514791488647,
      "learning_rate": 0.00027508652219175567,
      "loss": 2.2875,
      "step": 121299
    },
    {
      "epoch": 1.58,
      "grad_norm": 1.028827428817749,
      "learning_rate": 0.00027508244576150183,
      "loss": 2.1108,
      "step": 121300
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.9153143763542175,
      "learning_rate": 0.0002750783693358805,
      "loss": 2.1342,
      "step": 121301
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.9534429311752319,
      "learning_rate": 0.00027507429291489263,
      "loss": 2.1284,
      "step": 121302
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.9631307721138,
      "learning_rate": 0.00027507021649853897,
      "loss": 2.2685,
      "step": 121303
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.9947426915168762,
      "learning_rate": 0.00027506614008682015,
      "loss": 2.1814,
      "step": 121304
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.9269490242004395,
      "learning_rate": 0.00027506206367973706,
      "loss": 2.1344,
      "step": 121305
    },
    {
      "epoch": 1.58,
      "grad_norm": 1.0704022645950317,
      "learning_rate": 0.0002750579872772904,
      "loss": 2.237,
      "step": 121306
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.9453551173210144,
      "learning_rate": 0.000275053910879481,
      "loss": 2.0235,
      "step": 121307
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.9259442090988159,
      "learning_rate": 0.00027504983448630946,
      "loss": 2.2461,
      "step": 121308
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.8857898712158203,
      "learning_rate": 0.00027504575809777663,
      "loss": 2.1589,
      "step": 121309
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.8772478103637695,
      "learning_rate": 0.0002750416817138832,
      "loss": 2.0747,
      "step": 121310
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.9212865233421326,
      "learning_rate": 0.0002750376053346301,
      "loss": 2.0914,
      "step": 121311
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.9706388711929321,
      "learning_rate": 0.000275033528960018,
      "loss": 2.1313,
      "step": 121312
    },
    {
      "epoch": 1.58,
      "grad_norm": 1.149250864982605,
      "learning_rate": 0.0002750294525900475,
      "loss": 2.2025,
      "step": 121313
    },
    {
      "epoch": 1.58,
      "grad_norm": 1.0245147943496704,
      "learning_rate": 0.0002750253762247196,
      "loss": 2.2431,
      "step": 121314
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.9677653908729553,
      "learning_rate": 0.000275021299864035,
      "loss": 2.3069,
      "step": 121315
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.9235014319419861,
      "learning_rate": 0.0002750172235079943,
      "loss": 2.1647,
      "step": 121316
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.933542013168335,
      "learning_rate": 0.0002750131471565984,
      "loss": 2.0857,
      "step": 121317
    },
    {
      "epoch": 1.58,
      "grad_norm": 1.0007010698318481,
      "learning_rate": 0.00027500907080984805,
      "loss": 1.9753,
      "step": 121318
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.993659496307373,
      "learning_rate": 0.000275004994467744,
      "loss": 2.0102,
      "step": 121319
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.9456531405448914,
      "learning_rate": 0.000275000918130287,
      "loss": 2.009,
      "step": 121320
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.9447520971298218,
      "learning_rate": 0.0002749968417974777,
      "loss": 1.888,
      "step": 121321
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.9253956079483032,
      "learning_rate": 0.00027499276546931705,
      "loss": 2.1394,
      "step": 121322
    },
    {
      "epoch": 1.58,
      "grad_norm": 1.027129054069519,
      "learning_rate": 0.00027498868914580573,
      "loss": 2.1264,
      "step": 121323
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.964194118976593,
      "learning_rate": 0.0002749846128269445,
      "loss": 2.0725,
      "step": 121324
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.9983640313148499,
      "learning_rate": 0.00027498053651273406,
      "loss": 2.0643,
      "step": 121325
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.9238784313201904,
      "learning_rate": 0.0002749764602031751,
      "loss": 2.0334,
      "step": 121326
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.9653686881065369,
      "learning_rate": 0.0002749723838982686,
      "loss": 2.0898,
      "step": 121327
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.9594284892082214,
      "learning_rate": 0.0002749683075980152,
      "loss": 2.0998,
      "step": 121328
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.9260252714157104,
      "learning_rate": 0.0002749642313024157,
      "loss": 2.0211,
      "step": 121329
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.9148963689804077,
      "learning_rate": 0.0002749601550114709,
      "loss": 2.3338,
      "step": 121330
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.9788733720779419,
      "learning_rate": 0.0002749560787251813,
      "loss": 2.0283,
      "step": 121331
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.9966418743133545,
      "learning_rate": 0.00027495200244354793,
      "loss": 2.0547,
      "step": 121332
    },
    {
      "epoch": 1.58,
      "grad_norm": 1.0671112537384033,
      "learning_rate": 0.0002749479261665714,
      "loss": 2.3215,
      "step": 121333
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.9393175840377808,
      "learning_rate": 0.0002749438498942525,
      "loss": 2.0813,
      "step": 121334
    },
    {
      "epoch": 1.58,
      "grad_norm": 1.157989740371704,
      "learning_rate": 0.0002749397736265921,
      "loss": 2.1248,
      "step": 121335
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.9714829921722412,
      "learning_rate": 0.0002749356973635909,
      "loss": 2.2829,
      "step": 121336
    },
    {
      "epoch": 1.58,
      "grad_norm": 1.0405484437942505,
      "learning_rate": 0.0002749316211052496,
      "loss": 1.8932,
      "step": 121337
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.991423487663269,
      "learning_rate": 0.00027492754485156887,
      "loss": 2.2156,
      "step": 121338
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.9358254075050354,
      "learning_rate": 0.00027492346860254966,
      "loss": 2.1604,
      "step": 121339
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.9148924946784973,
      "learning_rate": 0.00027491939235819264,
      "loss": 2.1733,
      "step": 121340
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.9563395977020264,
      "learning_rate": 0.00027491531611849854,
      "loss": 2.1375,
      "step": 121341
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.9002472758293152,
      "learning_rate": 0.0002749112398834683,
      "loss": 2.2816,
      "step": 121342
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.8999330401420593,
      "learning_rate": 0.0002749071636531025,
      "loss": 2.0147,
      "step": 121343
    },
    {
      "epoch": 1.58,
      "grad_norm": 1.0249751806259155,
      "learning_rate": 0.0002749030874274018,
      "loss": 2.202,
      "step": 121344
    },
    {
      "epoch": 1.58,
      "grad_norm": 1.0150392055511475,
      "learning_rate": 0.00027489901120636716,
      "loss": 2.3159,
      "step": 121345
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.8602062463760376,
      "learning_rate": 0.00027489493498999924,
      "loss": 2.1512,
      "step": 121346
    },
    {
      "epoch": 1.58,
      "grad_norm": 1.0048270225524902,
      "learning_rate": 0.00027489085877829887,
      "loss": 2.3548,
      "step": 121347
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.9112880229949951,
      "learning_rate": 0.00027488678257126686,
      "loss": 2.144,
      "step": 121348
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.9632523655891418,
      "learning_rate": 0.00027488270636890375,
      "loss": 2.0716,
      "step": 121349
    },
    {
      "epoch": 1.58,
      "grad_norm": 1.061933994293213,
      "learning_rate": 0.0002748786301712104,
      "loss": 2.1472,
      "step": 121350
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.9349681735038757,
      "learning_rate": 0.0002748745539781876,
      "loss": 2.0721,
      "step": 121351
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.9303097128868103,
      "learning_rate": 0.0002748704777898361,
      "loss": 2.219,
      "step": 121352
    },
    {
      "epoch": 1.58,
      "grad_norm": 1.041968584060669,
      "learning_rate": 0.0002748664016061566,
      "loss": 2.0565,
      "step": 121353
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.9685737490653992,
      "learning_rate": 0.0002748623254271501,
      "loss": 2.2603,
      "step": 121354
    },
    {
      "epoch": 1.58,
      "grad_norm": 1.0301834344863892,
      "learning_rate": 0.000274858249252817,
      "loss": 2.259,
      "step": 121355
    },
    {
      "epoch": 1.58,
      "grad_norm": 1.0029610395431519,
      "learning_rate": 0.00027485417308315826,
      "loss": 2.1526,
      "step": 121356
    },
    {
      "epoch": 1.58,
      "grad_norm": 1.038834810256958,
      "learning_rate": 0.0002748500969181746,
      "loss": 2.3314,
      "step": 121357
    },
    {
      "epoch": 1.58,
      "grad_norm": 1.288366675376892,
      "learning_rate": 0.00027484602075786676,
      "loss": 2.1758,
      "step": 121358
    },
    {
      "epoch": 1.58,
      "grad_norm": 1.063480019569397,
      "learning_rate": 0.0002748419446022356,
      "loss": 2.2446,
      "step": 121359
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.948817789554596,
      "learning_rate": 0.0002748378684512818,
      "loss": 2.0717,
      "step": 121360
    },
    {
      "epoch": 1.58,
      "grad_norm": 1.023678183555603,
      "learning_rate": 0.00027483379230500604,
      "loss": 2.2987,
      "step": 121361
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.8765456676483154,
      "learning_rate": 0.00027482971616340916,
      "loss": 2.195,
      "step": 121362
    },
    {
      "epoch": 1.58,
      "grad_norm": 1.132716178894043,
      "learning_rate": 0.0002748256400264919,
      "loss": 2.0212,
      "step": 121363
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.9755210280418396,
      "learning_rate": 0.00027482156389425504,
      "loss": 1.9109,
      "step": 121364
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.9572349190711975,
      "learning_rate": 0.0002748174877666993,
      "loss": 2.2453,
      "step": 121365
    },
    {
      "epoch": 1.58,
      "grad_norm": 1.1428942680358887,
      "learning_rate": 0.0002748134116438257,
      "loss": 2.1546,
      "step": 121366
    },
    {
      "epoch": 1.58,
      "grad_norm": 1.2890127897262573,
      "learning_rate": 0.0002748093355256345,
      "loss": 2.0108,
      "step": 121367
    },
    {
      "epoch": 1.58,
      "grad_norm": 1.0932906866073608,
      "learning_rate": 0.00027480525941212674,
      "loss": 2.2697,
      "step": 121368
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.9319794178009033,
      "learning_rate": 0.00027480118330330323,
      "loss": 2.1531,
      "step": 121369
    },
    {
      "epoch": 1.58,
      "grad_norm": 1.3133177757263184,
      "learning_rate": 0.0002747971071991646,
      "loss": 1.8348,
      "step": 121370
    },
    {
      "epoch": 1.58,
      "grad_norm": 1.2701035737991333,
      "learning_rate": 0.0002747930310997117,
      "loss": 2.3186,
      "step": 121371
    },
    {
      "epoch": 1.58,
      "grad_norm": 1.0397672653198242,
      "learning_rate": 0.0002747889550049454,
      "loss": 2.1036,
      "step": 121372
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.9996691346168518,
      "learning_rate": 0.0002747848789148661,
      "loss": 2.4261,
      "step": 121373
    },
    {
      "epoch": 1.58,
      "grad_norm": 1.0450140237808228,
      "learning_rate": 0.0002747808028294748,
      "loss": 2.1505,
      "step": 121374
    },
    {
      "epoch": 1.58,
      "grad_norm": 1.0107882022857666,
      "learning_rate": 0.00027477672674877223,
      "loss": 2.1546,
      "step": 121375
    },
    {
      "epoch": 1.58,
      "grad_norm": 1.013208270072937,
      "learning_rate": 0.00027477265067275917,
      "loss": 1.8511,
      "step": 121376
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.9256754517555237,
      "learning_rate": 0.0002747685746014363,
      "loss": 2.1805,
      "step": 121377
    },
    {
      "epoch": 1.58,
      "grad_norm": 1.1006405353546143,
      "learning_rate": 0.00027476449853480446,
      "loss": 2.0688,
      "step": 121378
    },
    {
      "epoch": 1.58,
      "grad_norm": 1.0782482624053955,
      "learning_rate": 0.0002747604224728645,
      "loss": 2.1539,
      "step": 121379
    },
    {
      "epoch": 1.58,
      "grad_norm": 1.1134783029556274,
      "learning_rate": 0.0002747563464156169,
      "loss": 1.9874,
      "step": 121380
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.9694639444351196,
      "learning_rate": 0.00027475227036306263,
      "loss": 2.1135,
      "step": 121381
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.9279041886329651,
      "learning_rate": 0.00027474819431520234,
      "loss": 2.1986,
      "step": 121382
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.9278419613838196,
      "learning_rate": 0.00027474411827203686,
      "loss": 2.0746,
      "step": 121383
    },
    {
      "epoch": 1.58,
      "grad_norm": 1.0045253038406372,
      "learning_rate": 0.0002747400422335669,
      "loss": 2.1232,
      "step": 121384
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.8820440769195557,
      "learning_rate": 0.0002747359661997934,
      "loss": 1.9621,
      "step": 121385
    },
    {
      "epoch": 1.58,
      "grad_norm": 1.0010509490966797,
      "learning_rate": 0.0002747318901707168,
      "loss": 2.1637,
      "step": 121386
    },
    {
      "epoch": 1.58,
      "grad_norm": 1.0525799989700317,
      "learning_rate": 0.0002747278141463381,
      "loss": 2.1834,
      "step": 121387
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.9697486758232117,
      "learning_rate": 0.0002747237381266578,
      "loss": 2.1389,
      "step": 121388
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.9412639141082764,
      "learning_rate": 0.000274719662111677,
      "loss": 2.1626,
      "step": 121389
    },
    {
      "epoch": 1.58,
      "grad_norm": 1.04962956905365,
      "learning_rate": 0.00027471558610139623,
      "loss": 2.0811,
      "step": 121390
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.92690509557724,
      "learning_rate": 0.0002747115100958164,
      "loss": 1.9984,
      "step": 121391
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.9809665083885193,
      "learning_rate": 0.0002747074340949382,
      "loss": 2.1944,
      "step": 121392
    },
    {
      "epoch": 1.58,
      "grad_norm": 1.102756142616272,
      "learning_rate": 0.00027470335809876223,
      "loss": 2.2085,
      "step": 121393
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.9927867650985718,
      "learning_rate": 0.0002746992821072894,
      "loss": 2.2539,
      "step": 121394
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.9083983898162842,
      "learning_rate": 0.0002746952061205205,
      "loss": 2.0199,
      "step": 121395
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.9534450769424438,
      "learning_rate": 0.0002746911301384562,
      "loss": 1.959,
      "step": 121396
    },
    {
      "epoch": 1.58,
      "grad_norm": 1.0462080240249634,
      "learning_rate": 0.00027468705416109734,
      "loss": 2.2174,
      "step": 121397
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.9348278045654297,
      "learning_rate": 0.00027468297818844463,
      "loss": 2.2885,
      "step": 121398
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.8856849670410156,
      "learning_rate": 0.0002746789022204989,
      "loss": 2.0982,
      "step": 121399
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.974113404750824,
      "learning_rate": 0.0002746748262572607,
      "loss": 2.1221,
      "step": 121400
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.9682018756866455,
      "learning_rate": 0.000274670750298731,
      "loss": 1.9281,
      "step": 121401
    },
    {
      "epoch": 1.58,
      "grad_norm": 1.0234014987945557,
      "learning_rate": 0.00027466667434491044,
      "loss": 2.2258,
      "step": 121402
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.9947413206100464,
      "learning_rate": 0.0002746625983957999,
      "loss": 2.1112,
      "step": 121403
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.920712411403656,
      "learning_rate": 0.00027465852245139996,
      "loss": 2.232,
      "step": 121404
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.9205363392829895,
      "learning_rate": 0.00027465444651171155,
      "loss": 1.9239,
      "step": 121405
    },
    {
      "epoch": 1.58,
      "grad_norm": 1.049078345298767,
      "learning_rate": 0.0002746503705767353,
      "loss": 1.9491,
      "step": 121406
    },
    {
      "epoch": 1.58,
      "grad_norm": 1.0673983097076416,
      "learning_rate": 0.0002746462946464721,
      "loss": 2.1171,
      "step": 121407
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.9631185531616211,
      "learning_rate": 0.0002746422187209226,
      "loss": 2.0922,
      "step": 121408
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.8760517239570618,
      "learning_rate": 0.00027463814280008763,
      "loss": 2.1687,
      "step": 121409
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.9579447507858276,
      "learning_rate": 0.0002746340668839679,
      "loss": 2.382,
      "step": 121410
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.9182264804840088,
      "learning_rate": 0.0002746299909725641,
      "loss": 2.2296,
      "step": 121411
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.9518115520477295,
      "learning_rate": 0.0002746259150658771,
      "loss": 2.2888,
      "step": 121412
    },
    {
      "epoch": 1.58,
      "grad_norm": 1.0101522207260132,
      "learning_rate": 0.0002746218391639076,
      "loss": 2.1741,
      "step": 121413
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.9786862134933472,
      "learning_rate": 0.0002746177632666564,
      "loss": 2.0276,
      "step": 121414
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.9353376030921936,
      "learning_rate": 0.00027461368737412436,
      "loss": 2.0562,
      "step": 121415
    },
    {
      "epoch": 1.58,
      "grad_norm": 1.037657380104065,
      "learning_rate": 0.00027460961148631193,
      "loss": 2.2896,
      "step": 121416
    },
    {
      "epoch": 1.58,
      "grad_norm": 1.0085725784301758,
      "learning_rate": 0.0002746055356032201,
      "loss": 2.1491,
      "step": 121417
    },
    {
      "epoch": 1.58,
      "grad_norm": 1.00187087059021,
      "learning_rate": 0.00027460145972484963,
      "loss": 2.2769,
      "step": 121418
    },
    {
      "epoch": 1.58,
      "grad_norm": 1.0407626628875732,
      "learning_rate": 0.00027459738385120116,
      "loss": 2.1223,
      "step": 121419
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.9876354932785034,
      "learning_rate": 0.00027459330798227554,
      "loss": 2.054,
      "step": 121420
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.9606956839561462,
      "learning_rate": 0.0002745892321180736,
      "loss": 2.2074,
      "step": 121421
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.9551060199737549,
      "learning_rate": 0.00027458515625859595,
      "loss": 2.1642,
      "step": 121422
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.884160578250885,
      "learning_rate": 0.00027458108040384334,
      "loss": 1.9434,
      "step": 121423
    },
    {
      "epoch": 1.58,
      "grad_norm": 1.018985390663147,
      "learning_rate": 0.0002745770045538166,
      "loss": 2.0311,
      "step": 121424
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.9724069833755493,
      "learning_rate": 0.0002745729287085165,
      "loss": 2.1229,
      "step": 121425
    },
    {
      "epoch": 1.58,
      "grad_norm": 1.0845410823822021,
      "learning_rate": 0.0002745688528679437,
      "loss": 2.2665,
      "step": 121426
    },
    {
      "epoch": 1.58,
      "grad_norm": 1.1088486909866333,
      "learning_rate": 0.00027456477703209924,
      "loss": 2.0696,
      "step": 121427
    },
    {
      "epoch": 1.58,
      "grad_norm": 1.2100496292114258,
      "learning_rate": 0.0002745607012009835,
      "loss": 2.1549,
      "step": 121428
    },
    {
      "epoch": 1.58,
      "grad_norm": 1.032780408859253,
      "learning_rate": 0.0002745566253745974,
      "loss": 2.3186,
      "step": 121429
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.9605019092559814,
      "learning_rate": 0.00027455254955294174,
      "loss": 2.0158,
      "step": 121430
    },
    {
      "epoch": 1.58,
      "grad_norm": 1.147573709487915,
      "learning_rate": 0.00027454847373601716,
      "loss": 1.885,
      "step": 121431
    },
    {
      "epoch": 1.58,
      "grad_norm": 1.0978257656097412,
      "learning_rate": 0.0002745443979238246,
      "loss": 2.2013,
      "step": 121432
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.9382501840591431,
      "learning_rate": 0.0002745403221163648,
      "loss": 1.9587,
      "step": 121433
    },
    {
      "epoch": 1.58,
      "grad_norm": 1.0567889213562012,
      "learning_rate": 0.0002745362463136383,
      "loss": 2.1443,
      "step": 121434
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.9666000008583069,
      "learning_rate": 0.000274532170515646,
      "loss": 2.1572,
      "step": 121435
    },
    {
      "epoch": 1.58,
      "grad_norm": 1.0961837768554688,
      "learning_rate": 0.00027452809472238864,
      "loss": 2.2017,
      "step": 121436
    },
    {
      "epoch": 1.58,
      "grad_norm": 1.120065450668335,
      "learning_rate": 0.000274524018933867,
      "loss": 1.9629,
      "step": 121437
    },
    {
      "epoch": 1.58,
      "grad_norm": 1.125119924545288,
      "learning_rate": 0.00027451994315008183,
      "loss": 2.2859,
      "step": 121438
    },
    {
      "epoch": 1.58,
      "grad_norm": 1.2443324327468872,
      "learning_rate": 0.000274515867371034,
      "loss": 2.295,
      "step": 121439
    },
    {
      "epoch": 1.58,
      "grad_norm": 1.0411806106567383,
      "learning_rate": 0.000274511791596724,
      "loss": 2.046,
      "step": 121440
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.9018520712852478,
      "learning_rate": 0.00027450771582715277,
      "loss": 2.0104,
      "step": 121441
    },
    {
      "epoch": 1.58,
      "grad_norm": 1.0387941598892212,
      "learning_rate": 0.00027450364006232107,
      "loss": 1.9229,
      "step": 121442
    },
    {
      "epoch": 1.58,
      "grad_norm": 1.1229760646820068,
      "learning_rate": 0.0002744995643022296,
      "loss": 1.9927,
      "step": 121443
    },
    {
      "epoch": 1.58,
      "grad_norm": 1.3562288284301758,
      "learning_rate": 0.0002744954885468791,
      "loss": 1.9568,
      "step": 121444
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.9480629563331604,
      "learning_rate": 0.0002744914127962704,
      "loss": 1.9989,
      "step": 121445
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.9656221866607666,
      "learning_rate": 0.00027448733705040437,
      "loss": 2.0444,
      "step": 121446
    },
    {
      "epoch": 1.58,
      "grad_norm": 1.3442615270614624,
      "learning_rate": 0.0002744832613092815,
      "loss": 2.1098,
      "step": 121447
    },
    {
      "epoch": 1.58,
      "grad_norm": 1.1420016288757324,
      "learning_rate": 0.00027447918557290264,
      "loss": 2.1654,
      "step": 121448
    },
    {
      "epoch": 1.58,
      "grad_norm": 1.0799659490585327,
      "learning_rate": 0.0002744751098412686,
      "loss": 2.1984,
      "step": 121449
    },
    {
      "epoch": 1.58,
      "grad_norm": 1.099255084991455,
      "learning_rate": 0.00027447103411438014,
      "loss": 2.0149,
      "step": 121450
    },
    {
      "epoch": 1.58,
      "grad_norm": 1.0520790815353394,
      "learning_rate": 0.0002744669583922379,
      "loss": 2.1835,
      "step": 121451
    },
    {
      "epoch": 1.58,
      "grad_norm": 1.0301780700683594,
      "learning_rate": 0.000274462882674843,
      "loss": 2.1861,
      "step": 121452
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.9341395497322083,
      "learning_rate": 0.00027445880696219574,
      "loss": 2.136,
      "step": 121453
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.8892270922660828,
      "learning_rate": 0.00027445473125429706,
      "loss": 2.0753,
      "step": 121454
    },
    {
      "epoch": 1.58,
      "grad_norm": 1.0158218145370483,
      "learning_rate": 0.00027445065555114774,
      "loss": 1.9634,
      "step": 121455
    },
    {
      "epoch": 1.58,
      "grad_norm": 1.1662882566452026,
      "learning_rate": 0.0002744465798527485,
      "loss": 1.972,
      "step": 121456
    },
    {
      "epoch": 1.58,
      "grad_norm": 1.0254794359207153,
      "learning_rate": 0.00027444250415910017,
      "loss": 2.2261,
      "step": 121457
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.8684621453285217,
      "learning_rate": 0.00027443842847020355,
      "loss": 2.3632,
      "step": 121458
    },
    {
      "epoch": 1.58,
      "grad_norm": 1.2262588739395142,
      "learning_rate": 0.0002744343527860592,
      "loss": 2.0359,
      "step": 121459
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.9701663255691528,
      "learning_rate": 0.000274430277106668,
      "loss": 2.2386,
      "step": 121460
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.8779351115226746,
      "learning_rate": 0.00027442620143203065,
      "loss": 2.0622,
      "step": 121461
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.9132556915283203,
      "learning_rate": 0.000274422125762148,
      "loss": 2.3257,
      "step": 121462
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.9894810318946838,
      "learning_rate": 0.00027441805009702073,
      "loss": 2.0319,
      "step": 121463
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.9972667098045349,
      "learning_rate": 0.0002744139744366498,
      "loss": 1.9961,
      "step": 121464
    },
    {
      "epoch": 1.58,
      "grad_norm": 1.0275850296020508,
      "learning_rate": 0.00027440989878103563,
      "loss": 2.2022,
      "step": 121465
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.9858049154281616,
      "learning_rate": 0.0002744058231301791,
      "loss": 2.3148,
      "step": 121466
    },
    {
      "epoch": 1.58,
      "grad_norm": 1.0282557010650635,
      "learning_rate": 0.00027440174748408105,
      "loss": 2.0027,
      "step": 121467
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.954224705696106,
      "learning_rate": 0.0002743976718427422,
      "loss": 2.4042,
      "step": 121468
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.9635688662528992,
      "learning_rate": 0.0002743935962061633,
      "loss": 2.3919,
      "step": 121469
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.94012051820755,
      "learning_rate": 0.0002743895205743453,
      "loss": 2.3271,
      "step": 121470
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.9216105341911316,
      "learning_rate": 0.0002743854449472886,
      "loss": 2.215,
      "step": 121471
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.8847832679748535,
      "learning_rate": 0.0002743813693249941,
      "loss": 2.1002,
      "step": 121472
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.9561737775802612,
      "learning_rate": 0.0002743772937074626,
      "loss": 2.094,
      "step": 121473
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.9113284349441528,
      "learning_rate": 0.0002743732180946949,
      "loss": 2.0035,
      "step": 121474
    },
    {
      "epoch": 1.58,
      "grad_norm": 1.0693340301513672,
      "learning_rate": 0.00027436914248669166,
      "loss": 1.9738,
      "step": 121475
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.9572826027870178,
      "learning_rate": 0.0002743650668834537,
      "loss": 1.881,
      "step": 121476
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.9956983923912048,
      "learning_rate": 0.00027436099128498186,
      "loss": 2.2389,
      "step": 121477
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.9544494152069092,
      "learning_rate": 0.00027435691569127667,
      "loss": 1.9365,
      "step": 121478
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.9368537664413452,
      "learning_rate": 0.000274352840102339,
      "loss": 1.9447,
      "step": 121479
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.9768250584602356,
      "learning_rate": 0.00027434876451816963,
      "loss": 2.113,
      "step": 121480
    },
    {
      "epoch": 1.58,
      "grad_norm": 1.016007661819458,
      "learning_rate": 0.0002743446889387693,
      "loss": 2.1437,
      "step": 121481
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.8520461916923523,
      "learning_rate": 0.00027434061336413886,
      "loss": 2.0983,
      "step": 121482
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.9463766813278198,
      "learning_rate": 0.0002743365377942789,
      "loss": 2.1385,
      "step": 121483
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.9452739357948303,
      "learning_rate": 0.0002743324622291903,
      "loss": 2.207,
      "step": 121484
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.9542042016983032,
      "learning_rate": 0.0002743283866688738,
      "loss": 2.0143,
      "step": 121485
    },
    {
      "epoch": 1.58,
      "grad_norm": 1.1399550437927246,
      "learning_rate": 0.0002743243111133301,
      "loss": 2.0516,
      "step": 121486
    },
    {
      "epoch": 1.58,
      "grad_norm": 1.0504279136657715,
      "learning_rate": 0.00027432023556256,
      "loss": 2.1256,
      "step": 121487
    },
    {
      "epoch": 1.58,
      "grad_norm": 1.1213626861572266,
      "learning_rate": 0.0002743161600165643,
      "loss": 2.0226,
      "step": 121488
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.9813520312309265,
      "learning_rate": 0.0002743120844753437,
      "loss": 2.0523,
      "step": 121489
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.8568253517150879,
      "learning_rate": 0.0002743080089388989,
      "loss": 2.0702,
      "step": 121490
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.9542057514190674,
      "learning_rate": 0.0002743039334072308,
      "loss": 2.2991,
      "step": 121491
    },
    {
      "epoch": 1.58,
      "grad_norm": 1.1159454584121704,
      "learning_rate": 0.00027429985788034005,
      "loss": 2.2906,
      "step": 121492
    },
    {
      "epoch": 1.58,
      "grad_norm": 1.0617856979370117,
      "learning_rate": 0.0002742957823582274,
      "loss": 2.1453,
      "step": 121493
    },
    {
      "epoch": 1.58,
      "grad_norm": 1.0018717050552368,
      "learning_rate": 0.0002742917068408938,
      "loss": 2.1015,
      "step": 121494
    },
    {
      "epoch": 1.58,
      "grad_norm": 1.057381272315979,
      "learning_rate": 0.00027428763132833966,
      "loss": 1.897,
      "step": 121495
    },
    {
      "epoch": 1.58,
      "grad_norm": 1.0450907945632935,
      "learning_rate": 0.00027428355582056606,
      "loss": 2.0193,
      "step": 121496
    },
    {
      "epoch": 1.58,
      "grad_norm": 1.0484001636505127,
      "learning_rate": 0.0002742794803175736,
      "loss": 2.092,
      "step": 121497
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.9427816271781921,
      "learning_rate": 0.000274275404819363,
      "loss": 2.0047,
      "step": 121498
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.9831680655479431,
      "learning_rate": 0.0002742713293259352,
      "loss": 2.1046,
      "step": 121499
    },
    {
      "epoch": 1.58,
      "grad_norm": 1.1193662881851196,
      "learning_rate": 0.00027426725383729084,
      "loss": 2.0973,
      "step": 121500
    },
    {
      "epoch": 1.58,
      "grad_norm": 1.0136276483535767,
      "learning_rate": 0.0002742631783534306,
      "loss": 2.0792,
      "step": 121501
    },
    {
      "epoch": 1.58,
      "grad_norm": 1.1531546115875244,
      "learning_rate": 0.00027425910287435536,
      "loss": 2.1572,
      "step": 121502
    },
    {
      "epoch": 1.58,
      "grad_norm": 1.0111514329910278,
      "learning_rate": 0.00027425502740006583,
      "loss": 2.122,
      "step": 121503
    },
    {
      "epoch": 1.58,
      "grad_norm": 1.073504090309143,
      "learning_rate": 0.0002742509519305628,
      "loss": 2.2343,
      "step": 121504
    },
    {
      "epoch": 1.58,
      "grad_norm": 1.005867600440979,
      "learning_rate": 0.00027424687646584696,
      "loss": 2.408,
      "step": 121505
    },
    {
      "epoch": 1.58,
      "grad_norm": 1.113185167312622,
      "learning_rate": 0.00027424280100591926,
      "loss": 2.0141,
      "step": 121506
    },
    {
      "epoch": 1.58,
      "grad_norm": 1.0295979976654053,
      "learning_rate": 0.00027423872555078015,
      "loss": 2.0792,
      "step": 121507
    },
    {
      "epoch": 1.58,
      "grad_norm": 1.059125542640686,
      "learning_rate": 0.0002742346501004306,
      "loss": 2.1326,
      "step": 121508
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.8662609457969666,
      "learning_rate": 0.00027423057465487125,
      "loss": 2.1861,
      "step": 121509
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.945999026298523,
      "learning_rate": 0.00027422649921410297,
      "loss": 2.0903,
      "step": 121510
    },
    {
      "epoch": 1.58,
      "grad_norm": 1.0802624225616455,
      "learning_rate": 0.0002742224237781265,
      "loss": 2.0398,
      "step": 121511
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.9731811285018921,
      "learning_rate": 0.0002742183483469425,
      "loss": 2.2237,
      "step": 121512
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.9181215167045593,
      "learning_rate": 0.000274214272920552,
      "loss": 2.0155,
      "step": 121513
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.9327002167701721,
      "learning_rate": 0.00027421019749895533,
      "loss": 1.9893,
      "step": 121514
    },
    {
      "epoch": 1.58,
      "grad_norm": 1.0260993242263794,
      "learning_rate": 0.0002742061220821535,
      "loss": 2.1566,
      "step": 121515
    },
    {
      "epoch": 1.58,
      "grad_norm": 1.0119919776916504,
      "learning_rate": 0.00027420204667014727,
      "loss": 2.1583,
      "step": 121516
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.9446685910224915,
      "learning_rate": 0.0002741979712629374,
      "loss": 2.2068,
      "step": 121517
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.9606345891952515,
      "learning_rate": 0.00027419389586052454,
      "loss": 2.08,
      "step": 121518
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.9823590517044067,
      "learning_rate": 0.0002741898204629097,
      "loss": 2.0739,
      "step": 121519
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.9373116493225098,
      "learning_rate": 0.0002741857450700933,
      "loss": 2.2875,
      "step": 121520
    },
    {
      "epoch": 1.58,
      "grad_norm": 1.181170105934143,
      "learning_rate": 0.00027418166968207626,
      "loss": 1.7875,
      "step": 121521
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.9487715363502502,
      "learning_rate": 0.0002741775942988593,
      "loss": 2.1869,
      "step": 121522
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.9027405381202698,
      "learning_rate": 0.0002741735189204433,
      "loss": 2.0556,
      "step": 121523
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.9600412845611572,
      "learning_rate": 0.0002741694435468289,
      "loss": 2.3073,
      "step": 121524
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.9764917492866516,
      "learning_rate": 0.000274165368178017,
      "loss": 2.0232,
      "step": 121525
    },
    {
      "epoch": 1.58,
      "grad_norm": 1.0724447965621948,
      "learning_rate": 0.0002741612928140081,
      "loss": 2.1608,
      "step": 121526
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.9826530814170837,
      "learning_rate": 0.00027415721745480316,
      "loss": 2.0575,
      "step": 121527
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.9141755104064941,
      "learning_rate": 0.0002741531421004028,
      "loss": 2.2261,
      "step": 121528
    },
    {
      "epoch": 1.58,
      "grad_norm": 1.0449894666671753,
      "learning_rate": 0.0002741490667508079,
      "loss": 2.0875,
      "step": 121529
    },
    {
      "epoch": 1.58,
      "grad_norm": 1.00098717212677,
      "learning_rate": 0.0002741449914060192,
      "loss": 1.9418,
      "step": 121530
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.9274513721466064,
      "learning_rate": 0.0002741409160660375,
      "loss": 2.0231,
      "step": 121531
    },
    {
      "epoch": 1.58,
      "grad_norm": 1.0399929285049438,
      "learning_rate": 0.0002741368407308634,
      "loss": 1.8509,
      "step": 121532
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.9629354476928711,
      "learning_rate": 0.00027413276540049777,
      "loss": 2.2213,
      "step": 121533
    },
    {
      "epoch": 1.58,
      "grad_norm": 1.0258628129959106,
      "learning_rate": 0.0002741286900749413,
      "loss": 1.8581,
      "step": 121534
    },
    {
      "epoch": 1.58,
      "grad_norm": 1.0263185501098633,
      "learning_rate": 0.00027412461475419486,
      "loss": 2.3344,
      "step": 121535
    },
    {
      "epoch": 1.58,
      "grad_norm": 1.0629135370254517,
      "learning_rate": 0.00027412053943825913,
      "loss": 2.0058,
      "step": 121536
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.9374582171440125,
      "learning_rate": 0.000274116464127135,
      "loss": 2.27,
      "step": 121537
    },
    {
      "epoch": 1.58,
      "grad_norm": 1.0247403383255005,
      "learning_rate": 0.00027411238882082294,
      "loss": 2.247,
      "step": 121538
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.9683413505554199,
      "learning_rate": 0.0002741083135193239,
      "loss": 1.9982,
      "step": 121539
    },
    {
      "epoch": 1.58,
      "grad_norm": 1.0578714609146118,
      "learning_rate": 0.0002741042382226386,
      "loss": 1.9869,
      "step": 121540
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.9412413239479065,
      "learning_rate": 0.00027410016293076783,
      "loss": 2.3388,
      "step": 121541
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.9654511213302612,
      "learning_rate": 0.00027409608764371237,
      "loss": 2.3921,
      "step": 121542
    },
    {
      "epoch": 1.58,
      "grad_norm": 1.1808727979660034,
      "learning_rate": 0.00027409201236147304,
      "loss": 2.1637,
      "step": 121543
    },
    {
      "epoch": 1.58,
      "grad_norm": 1.042715311050415,
      "learning_rate": 0.0002740879370840503,
      "loss": 2.2685,
      "step": 121544
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.921789288520813,
      "learning_rate": 0.0002740838618114452,
      "loss": 2.0644,
      "step": 121545
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.9650475382804871,
      "learning_rate": 0.0002740797865436583,
      "loss": 2.1102,
      "step": 121546
    },
    {
      "epoch": 1.58,
      "grad_norm": 1.02381432056427,
      "learning_rate": 0.0002740757112806905,
      "loss": 2.0867,
      "step": 121547
    },
    {
      "epoch": 1.58,
      "grad_norm": 1.1599822044372559,
      "learning_rate": 0.00027407163602254255,
      "loss": 2.1862,
      "step": 121548
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.9705209136009216,
      "learning_rate": 0.00027406756076921526,
      "loss": 2.1323,
      "step": 121549
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.9485352635383606,
      "learning_rate": 0.00027406348552070915,
      "loss": 2.2841,
      "step": 121550
    },
    {
      "epoch": 1.58,
      "grad_norm": 1.4560422897338867,
      "learning_rate": 0.0002740594102770252,
      "loss": 2.2074,
      "step": 121551
    },
    {
      "epoch": 1.58,
      "grad_norm": 1.1220165491104126,
      "learning_rate": 0.00027405533503816403,
      "loss": 2.0462,
      "step": 121552
    },
    {
      "epoch": 1.58,
      "grad_norm": 1.1044297218322754,
      "learning_rate": 0.0002740512598041265,
      "loss": 2.4443,
      "step": 121553
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.9620751142501831,
      "learning_rate": 0.0002740471845749133,
      "loss": 2.4103,
      "step": 121554
    },
    {
      "epoch": 1.58,
      "grad_norm": 1.0284885168075562,
      "learning_rate": 0.00027404310935052534,
      "loss": 2.0238,
      "step": 121555
    },
    {
      "epoch": 1.58,
      "grad_norm": 1.2356092929840088,
      "learning_rate": 0.0002740390341309632,
      "loss": 2.1779,
      "step": 121556
    },
    {
      "epoch": 1.58,
      "grad_norm": 1.2333488464355469,
      "learning_rate": 0.0002740349589162276,
      "loss": 2.443,
      "step": 121557
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.9964497685432434,
      "learning_rate": 0.00027403088370631947,
      "loss": 2.0571,
      "step": 121558
    },
    {
      "epoch": 1.58,
      "grad_norm": 1.1672017574310303,
      "learning_rate": 0.00027402680850123945,
      "loss": 2.1349,
      "step": 121559
    },
    {
      "epoch": 1.58,
      "grad_norm": 1.733132243156433,
      "learning_rate": 0.00027402273330098836,
      "loss": 2.1629,
      "step": 121560
    },
    {
      "epoch": 1.58,
      "grad_norm": 1.062148094177246,
      "learning_rate": 0.000274018658105567,
      "loss": 2.2254,
      "step": 121561
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.9748805165290833,
      "learning_rate": 0.000274014582914976,
      "loss": 2.0707,
      "step": 121562
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.9661787152290344,
      "learning_rate": 0.0002740105077292162,
      "loss": 2.1442,
      "step": 121563
    },
    {
      "epoch": 1.58,
      "grad_norm": 1.1505367755889893,
      "learning_rate": 0.0002740064325482883,
      "loss": 2.2669,
      "step": 121564
    },
    {
      "epoch": 1.58,
      "grad_norm": 1.028133749961853,
      "learning_rate": 0.00027400235737219306,
      "loss": 2.1652,
      "step": 121565
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.9531893134117126,
      "learning_rate": 0.0002739982822009313,
      "loss": 2.0896,
      "step": 121566
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.9948002696037292,
      "learning_rate": 0.0002739942070345038,
      "loss": 2.3142,
      "step": 121567
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.9040100574493408,
      "learning_rate": 0.00027399013187291124,
      "loss": 2.3741,
      "step": 121568
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.9292108416557312,
      "learning_rate": 0.00027398605671615444,
      "loss": 2.2801,
      "step": 121569
    },
    {
      "epoch": 1.58,
      "grad_norm": 1.0182288885116577,
      "learning_rate": 0.0002739819815642341,
      "loss": 2.0817,
      "step": 121570
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.9933024048805237,
      "learning_rate": 0.00027397790641715093,
      "loss": 2.1503,
      "step": 121571
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.8862230777740479,
      "learning_rate": 0.0002739738312749058,
      "loss": 1.9916,
      "step": 121572
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.9393142461776733,
      "learning_rate": 0.0002739697561374995,
      "loss": 1.9194,
      "step": 121573
    },
    {
      "epoch": 1.58,
      "grad_norm": 1.0532077550888062,
      "learning_rate": 0.0002739656810049326,
      "loss": 2.128,
      "step": 121574
    },
    {
      "epoch": 1.58,
      "grad_norm": 1.026750922203064,
      "learning_rate": 0.000273961605877206,
      "loss": 2.0721,
      "step": 121575
    },
    {
      "epoch": 1.58,
      "grad_norm": 1.1789042949676514,
      "learning_rate": 0.00027395753075432056,
      "loss": 2.3633,
      "step": 121576
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.9211587905883789,
      "learning_rate": 0.00027395345563627677,
      "loss": 2.0783,
      "step": 121577
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.9979177713394165,
      "learning_rate": 0.0002739493805230755,
      "loss": 2.1037,
      "step": 121578
    },
    {
      "epoch": 1.58,
      "grad_norm": 1.0271990299224854,
      "learning_rate": 0.0002739453054147176,
      "loss": 2.2527,
      "step": 121579
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.9531494975090027,
      "learning_rate": 0.0002739412303112038,
      "loss": 2.1988,
      "step": 121580
    },
    {
      "epoch": 1.58,
      "grad_norm": 1.1382139921188354,
      "learning_rate": 0.00027393715521253475,
      "loss": 2.5124,
      "step": 121581
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.9644553065299988,
      "learning_rate": 0.0002739330801187113,
      "loss": 2.1254,
      "step": 121582
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.9196967482566833,
      "learning_rate": 0.0002739290050297342,
      "loss": 2.3251,
      "step": 121583
    },
    {
      "epoch": 1.58,
      "grad_norm": 1.1159069538116455,
      "learning_rate": 0.0002739249299456041,
      "loss": 2.0796,
      "step": 121584
    },
    {
      "epoch": 1.58,
      "grad_norm": 1.0808199644088745,
      "learning_rate": 0.0002739208548663219,
      "loss": 2.2912,
      "step": 121585
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.9564164280891418,
      "learning_rate": 0.00027391677979188837,
      "loss": 2.1657,
      "step": 121586
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.9483270049095154,
      "learning_rate": 0.00027391270472230413,
      "loss": 2.1281,
      "step": 121587
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.9025922417640686,
      "learning_rate": 0.00027390862965757,
      "loss": 2.196,
      "step": 121588
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.8839744329452515,
      "learning_rate": 0.0002739045545976868,
      "loss": 2.0082,
      "step": 121589
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.9330312609672546,
      "learning_rate": 0.00027390047954265525,
      "loss": 2.2823,
      "step": 121590
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.98940509557724,
      "learning_rate": 0.00027389640449247606,
      "loss": 2.1393,
      "step": 121591
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.9589996933937073,
      "learning_rate": 0.00027389232944715003,
      "loss": 2.1895,
      "step": 121592
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.9237666130065918,
      "learning_rate": 0.0002738882544066779,
      "loss": 2.454,
      "step": 121593
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.9335085153579712,
      "learning_rate": 0.00027388417937106044,
      "loss": 2.1178,
      "step": 121594
    },
    {
      "epoch": 1.58,
      "grad_norm": 1.0408823490142822,
      "learning_rate": 0.00027388010434029834,
      "loss": 2.2025,
      "step": 121595
    },
    {
      "epoch": 1.58,
      "grad_norm": 1.0988178253173828,
      "learning_rate": 0.0002738760293143925,
      "loss": 2.0278,
      "step": 121596
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.9686614274978638,
      "learning_rate": 0.00027387195429334353,
      "loss": 2.1038,
      "step": 121597
    },
    {
      "epoch": 1.58,
      "grad_norm": 1.1447054147720337,
      "learning_rate": 0.0002738678792771524,
      "loss": 2.1235,
      "step": 121598
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.936328113079071,
      "learning_rate": 0.00027386380426581964,
      "loss": 2.023,
      "step": 121599
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.950768232345581,
      "learning_rate": 0.00027385972925934605,
      "loss": 2.05,
      "step": 121600
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.8502792119979858,
      "learning_rate": 0.00027385565425773245,
      "loss": 2.1951,
      "step": 121601
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.954945981502533,
      "learning_rate": 0.0002738515792609796,
      "loss": 2.1358,
      "step": 121602
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.9404594302177429,
      "learning_rate": 0.00027384750426908815,
      "loss": 2.0513,
      "step": 121603
    },
    {
      "epoch": 1.58,
      "grad_norm": 1.028613805770874,
      "learning_rate": 0.00027384342928205913,
      "loss": 2.0776,
      "step": 121604
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.8766577243804932,
      "learning_rate": 0.00027383935429989303,
      "loss": 2.312,
      "step": 121605
    },
    {
      "epoch": 1.58,
      "grad_norm": 1.001481533050537,
      "learning_rate": 0.0002738352793225906,
      "loss": 2.1863,
      "step": 121606
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.9922721982002258,
      "learning_rate": 0.0002738312043501527,
      "loss": 2.082,
      "step": 121607
    },
    {
      "epoch": 1.58,
      "grad_norm": 1.0443336963653564,
      "learning_rate": 0.00027382712938258015,
      "loss": 2.1755,
      "step": 121608
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.9005640745162964,
      "learning_rate": 0.0002738230544198736,
      "loss": 1.9533,
      "step": 121609
    },
    {
      "epoch": 1.58,
      "grad_norm": 1.0833567380905151,
      "learning_rate": 0.0002738189794620339,
      "loss": 2.0863,
      "step": 121610
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.9024865627288818,
      "learning_rate": 0.00027381490450906166,
      "loss": 2.1403,
      "step": 121611
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.9003439545631409,
      "learning_rate": 0.0002738108295609577,
      "loss": 1.9107,
      "step": 121612
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.9843485951423645,
      "learning_rate": 0.00027380675461772283,
      "loss": 1.9467,
      "step": 121613
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.9582799673080444,
      "learning_rate": 0.0002738026796793578,
      "loss": 1.9831,
      "step": 121614
    },
    {
      "epoch": 1.58,
      "grad_norm": 1.0220872163772583,
      "learning_rate": 0.0002737986047458633,
      "loss": 2.1922,
      "step": 121615
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.9874006509780884,
      "learning_rate": 0.0002737945298172403,
      "loss": 2.0943,
      "step": 121616
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.9786717295646667,
      "learning_rate": 0.00027379045489348924,
      "loss": 2.2982,
      "step": 121617
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.9642105102539062,
      "learning_rate": 0.000273786379974611,
      "loss": 2.0866,
      "step": 121618
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.9946443438529968,
      "learning_rate": 0.0002737823050606064,
      "loss": 2.084,
      "step": 121619
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.8870088458061218,
      "learning_rate": 0.0002737782301514762,
      "loss": 2.155,
      "step": 121620
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.9517516493797302,
      "learning_rate": 0.00027377415524722107,
      "loss": 2.0428,
      "step": 121621
    },
    {
      "epoch": 1.58,
      "grad_norm": 1.011993408203125,
      "learning_rate": 0.000273770080347842,
      "loss": 2.1674,
      "step": 121622
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.991041362285614,
      "learning_rate": 0.0002737660054533394,
      "loss": 2.0647,
      "step": 121623
    },
    {
      "epoch": 1.58,
      "grad_norm": 1.0363577604293823,
      "learning_rate": 0.0002737619305637142,
      "loss": 2.0072,
      "step": 121624
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.9104531407356262,
      "learning_rate": 0.00027375785567896715,
      "loss": 2.1678,
      "step": 121625
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.9334016442298889,
      "learning_rate": 0.000273753780799099,
      "loss": 2.0614,
      "step": 121626
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.9202876687049866,
      "learning_rate": 0.0002737497059241105,
      "loss": 2.0946,
      "step": 121627
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.8797047734260559,
      "learning_rate": 0.0002737456310540026,
      "loss": 2.1694,
      "step": 121628
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.973673403263092,
      "learning_rate": 0.00027374155618877573,
      "loss": 2.2254,
      "step": 121629
    },
    {
      "epoch": 1.58,
      "grad_norm": 1.1144282817840576,
      "learning_rate": 0.0002737374813284308,
      "loss": 2.1621,
      "step": 121630
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.9649077653884888,
      "learning_rate": 0.0002737334064729686,
      "loss": 2.1502,
      "step": 121631
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.8798648118972778,
      "learning_rate": 0.0002737293316223898,
      "loss": 2.006,
      "step": 121632
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.9871565699577332,
      "learning_rate": 0.0002737252567766953,
      "loss": 2.0522,
      "step": 121633
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.9286408424377441,
      "learning_rate": 0.00027372118193588584,
      "loss": 2.1905,
      "step": 121634
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.8763179779052734,
      "learning_rate": 0.00027371710709996196,
      "loss": 2.2983,
      "step": 121635
    },
    {
      "epoch": 1.58,
      "grad_norm": 1.0001745223999023,
      "learning_rate": 0.0002737130322689246,
      "loss": 2.2142,
      "step": 121636
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.9410966038703918,
      "learning_rate": 0.0002737089574427745,
      "loss": 2.2358,
      "step": 121637
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.9578249454498291,
      "learning_rate": 0.00027370488262151233,
      "loss": 2.2423,
      "step": 121638
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.9699535369873047,
      "learning_rate": 0.000273700807805139,
      "loss": 2.2111,
      "step": 121639
    },
    {
      "epoch": 1.58,
      "grad_norm": 1.0533101558685303,
      "learning_rate": 0.0002736967329936553,
      "loss": 1.9423,
      "step": 121640
    },
    {
      "epoch": 1.58,
      "grad_norm": 1.0073652267456055,
      "learning_rate": 0.0002736926581870617,
      "loss": 2.1873,
      "step": 121641
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.877570629119873,
      "learning_rate": 0.00027368858338535913,
      "loss": 1.9596,
      "step": 121642
    },
    {
      "epoch": 1.58,
      "grad_norm": 1.0154790878295898,
      "learning_rate": 0.0002736845085885484,
      "loss": 2.3196,
      "step": 121643
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.9651036858558655,
      "learning_rate": 0.0002736804337966302,
      "loss": 2.3148,
      "step": 121644
    },
    {
      "epoch": 1.58,
      "grad_norm": 1.1453757286071777,
      "learning_rate": 0.0002736763590096053,
      "loss": 2.0661,
      "step": 121645
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.9685177206993103,
      "learning_rate": 0.00027367228422747443,
      "loss": 2.1346,
      "step": 121646
    },
    {
      "epoch": 1.58,
      "grad_norm": 1.090326189994812,
      "learning_rate": 0.00027366820945023853,
      "loss": 2.0754,
      "step": 121647
    },
    {
      "epoch": 1.58,
      "grad_norm": 1.009013056755066,
      "learning_rate": 0.0002736641346778981,
      "loss": 2.2616,
      "step": 121648
    },
    {
      "epoch": 1.58,
      "grad_norm": 1.078676700592041,
      "learning_rate": 0.00027366005991045394,
      "loss": 2.1225,
      "step": 121649
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.93753981590271,
      "learning_rate": 0.0002736559851479069,
      "loss": 2.2249,
      "step": 121650
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.8878932595252991,
      "learning_rate": 0.0002736519103902578,
      "loss": 1.9715,
      "step": 121651
    },
    {
      "epoch": 1.58,
      "grad_norm": 1.0657758712768555,
      "learning_rate": 0.0002736478356375072,
      "loss": 2.2546,
      "step": 121652
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.9086272716522217,
      "learning_rate": 0.00027364376088965614,
      "loss": 2.0739,
      "step": 121653
    },
    {
      "epoch": 1.58,
      "grad_norm": 1.0561907291412354,
      "learning_rate": 0.00027363968614670505,
      "loss": 2.2995,
      "step": 121654
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.9306399822235107,
      "learning_rate": 0.00027363561140865487,
      "loss": 2.1894,
      "step": 121655
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.9509220123291016,
      "learning_rate": 0.00027363153667550626,
      "loss": 1.8827,
      "step": 121656
    },
    {
      "epoch": 1.58,
      "grad_norm": 1.0460008382797241,
      "learning_rate": 0.00027362746194726015,
      "loss": 2.2051,
      "step": 121657
    },
    {
      "epoch": 1.58,
      "grad_norm": 1.1304292678833008,
      "learning_rate": 0.0002736233872239171,
      "loss": 1.7136,
      "step": 121658
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.996921181678772,
      "learning_rate": 0.00027361931250547805,
      "loss": 2.09,
      "step": 121659
    },
    {
      "epoch": 1.58,
      "grad_norm": 1.0031349658966064,
      "learning_rate": 0.00027361523779194364,
      "loss": 2.1665,
      "step": 121660
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.9545778632164001,
      "learning_rate": 0.0002736111630833147,
      "loss": 2.2459,
      "step": 121661
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.9243337512016296,
      "learning_rate": 0.0002736070883795919,
      "loss": 1.9704,
      "step": 121662
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.9713485836982727,
      "learning_rate": 0.000273603013680776,
      "loss": 2.1779,
      "step": 121663
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.9116023182868958,
      "learning_rate": 0.0002735989389868678,
      "loss": 2.0492,
      "step": 121664
    },
    {
      "epoch": 1.58,
      "grad_norm": 1.0023560523986816,
      "learning_rate": 0.00027359486429786813,
      "loss": 2.2596,
      "step": 121665
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.897599995136261,
      "learning_rate": 0.0002735907896137776,
      "loss": 1.968,
      "step": 121666
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.9410766363143921,
      "learning_rate": 0.00027358671493459707,
      "loss": 1.9532,
      "step": 121667
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.958577573299408,
      "learning_rate": 0.0002735826402603273,
      "loss": 2.0755,
      "step": 121668
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.9354096055030823,
      "learning_rate": 0.000273578565590969,
      "loss": 2.0762,
      "step": 121669
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.915075421333313,
      "learning_rate": 0.00027357449092652286,
      "loss": 2.0327,
      "step": 121670
    },
    {
      "epoch": 1.58,
      "grad_norm": 1.0003365278244019,
      "learning_rate": 0.00027357041626698984,
      "loss": 2.1724,
      "step": 121671
    },
    {
      "epoch": 1.58,
      "grad_norm": 1.255444884300232,
      "learning_rate": 0.00027356634161237054,
      "loss": 2.1069,
      "step": 121672
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.9341353178024292,
      "learning_rate": 0.0002735622669626657,
      "loss": 2.2447,
      "step": 121673
    },
    {
      "epoch": 1.58,
      "grad_norm": 1.0166938304901123,
      "learning_rate": 0.00027355819231787617,
      "loss": 2.1548,
      "step": 121674
    },
    {
      "epoch": 1.58,
      "grad_norm": 1.186022162437439,
      "learning_rate": 0.0002735541176780027,
      "loss": 2.3781,
      "step": 121675
    },
    {
      "epoch": 1.58,
      "grad_norm": 1.0864673852920532,
      "learning_rate": 0.000273550043043046,
      "loss": 2.0706,
      "step": 121676
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.9012221693992615,
      "learning_rate": 0.0002735459684130069,
      "loss": 2.0075,
      "step": 121677
    },
    {
      "epoch": 1.58,
      "grad_norm": 1.1986565589904785,
      "learning_rate": 0.000273541893787886,
      "loss": 2.2619,
      "step": 121678
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.996009349822998,
      "learning_rate": 0.0002735378191676842,
      "loss": 2.1982,
      "step": 121679
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.9031418561935425,
      "learning_rate": 0.0002735337445524022,
      "loss": 2.1302,
      "step": 121680
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.9525653123855591,
      "learning_rate": 0.0002735296699420408,
      "loss": 2.1323,
      "step": 121681
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.9489467740058899,
      "learning_rate": 0.0002735255953366008,
      "loss": 2.2068,
      "step": 121682
    },
    {
      "epoch": 1.58,
      "grad_norm": 1.0804288387298584,
      "learning_rate": 0.0002735215207360829,
      "loss": 2.1025,
      "step": 121683
    },
    {
      "epoch": 1.58,
      "grad_norm": 1.0188685655593872,
      "learning_rate": 0.0002735174461404877,
      "loss": 2.073,
      "step": 121684
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.9713050723075867,
      "learning_rate": 0.0002735133715498162,
      "loss": 2.1497,
      "step": 121685
    },
    {
      "epoch": 1.58,
      "grad_norm": 1.0371633768081665,
      "learning_rate": 0.000273509296964069,
      "loss": 2.0543,
      "step": 121686
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.9594012498855591,
      "learning_rate": 0.00027350522238324697,
      "loss": 2.0927,
      "step": 121687
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.9207556843757629,
      "learning_rate": 0.00027350114780735083,
      "loss": 2.0551,
      "step": 121688
    },
    {
      "epoch": 1.58,
      "grad_norm": 1.0887858867645264,
      "learning_rate": 0.0002734970732363814,
      "loss": 2.1504,
      "step": 121689
    },
    {
      "epoch": 1.58,
      "grad_norm": 1.184139370918274,
      "learning_rate": 0.00027349299867033926,
      "loss": 2.0963,
      "step": 121690
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.9084346890449524,
      "learning_rate": 0.0002734889241092253,
      "loss": 2.0389,
      "step": 121691
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.9854284524917603,
      "learning_rate": 0.0002734848495530402,
      "loss": 2.1159,
      "step": 121692
    },
    {
      "epoch": 1.58,
      "grad_norm": 1.1721514463424683,
      "learning_rate": 0.0002734807750017848,
      "loss": 2.23,
      "step": 121693
    },
    {
      "epoch": 1.58,
      "grad_norm": 1.033082365989685,
      "learning_rate": 0.00027347670045545984,
      "loss": 2.0174,
      "step": 121694
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.9425614476203918,
      "learning_rate": 0.00027347262591406616,
      "loss": 2.0865,
      "step": 121695
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.9439403414726257,
      "learning_rate": 0.0002734685513776043,
      "loss": 2.16,
      "step": 121696
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.9522420763969421,
      "learning_rate": 0.00027346447684607517,
      "loss": 2.1794,
      "step": 121697
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.94377201795578,
      "learning_rate": 0.00027346040231947944,
      "loss": 2.0441,
      "step": 121698
    },
    {
      "epoch": 1.58,
      "grad_norm": 1.048627257347107,
      "learning_rate": 0.0002734563277978179,
      "loss": 1.9533,
      "step": 121699
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.9176162481307983,
      "learning_rate": 0.0002734522532810914,
      "loss": 2.0982,
      "step": 121700
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.9196325540542603,
      "learning_rate": 0.0002734481787693007,
      "loss": 2.1182,
      "step": 121701
    },
    {
      "epoch": 1.58,
      "grad_norm": 1.0417174100875854,
      "learning_rate": 0.0002734441042624464,
      "loss": 2.3505,
      "step": 121702
    },
    {
      "epoch": 1.58,
      "grad_norm": 1.093747615814209,
      "learning_rate": 0.00027344002976052934,
      "loss": 2.2342,
      "step": 121703
    },
    {
      "epoch": 1.58,
      "grad_norm": 1.1656299829483032,
      "learning_rate": 0.00027343595526355027,
      "loss": 2.1728,
      "step": 121704
    },
    {
      "epoch": 1.58,
      "grad_norm": 1.0026512145996094,
      "learning_rate": 0.00027343188077150996,
      "loss": 2.1831,
      "step": 121705
    },
    {
      "epoch": 1.58,
      "grad_norm": 1.0193840265274048,
      "learning_rate": 0.0002734278062844091,
      "loss": 2.099,
      "step": 121706
    },
    {
      "epoch": 1.58,
      "grad_norm": 1.0617496967315674,
      "learning_rate": 0.00027342373180224875,
      "loss": 2.2481,
      "step": 121707
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.9164170622825623,
      "learning_rate": 0.00027341965732502927,
      "loss": 2.0882,
      "step": 121708
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.9285656809806824,
      "learning_rate": 0.0002734155828527515,
      "loss": 2.1253,
      "step": 121709
    },
    {
      "epoch": 1.58,
      "grad_norm": 1.0072715282440186,
      "learning_rate": 0.0002734115083854163,
      "loss": 2.1021,
      "step": 121710
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.9197509288787842,
      "learning_rate": 0.0002734074339230245,
      "loss": 2.1936,
      "step": 121711
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.9643558263778687,
      "learning_rate": 0.0002734033594655767,
      "loss": 2.1759,
      "step": 121712
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.9098328351974487,
      "learning_rate": 0.0002733992850130737,
      "loss": 2.348,
      "step": 121713
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.9061795473098755,
      "learning_rate": 0.0002733952105655164,
      "loss": 2.1285,
      "step": 121714
    },
    {
      "epoch": 1.58,
      "grad_norm": 1.0098336935043335,
      "learning_rate": 0.0002733911361229053,
      "loss": 2.097,
      "step": 121715
    },
    {
      "epoch": 1.58,
      "grad_norm": 1.0829862356185913,
      "learning_rate": 0.0002733870616852413,
      "loss": 2.1741,
      "step": 121716
    },
    {
      "epoch": 1.58,
      "grad_norm": 1.0546094179153442,
      "learning_rate": 0.00027338298725252525,
      "loss": 2.1968,
      "step": 121717
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.8682597279548645,
      "learning_rate": 0.0002733789128247577,
      "loss": 2.1101,
      "step": 121718
    },
    {
      "epoch": 1.58,
      "grad_norm": 1.000652551651001,
      "learning_rate": 0.0002733748384019395,
      "loss": 2.3001,
      "step": 121719
    },
    {
      "epoch": 1.58,
      "grad_norm": 1.0236414670944214,
      "learning_rate": 0.0002733707639840716,
      "loss": 2.2666,
      "step": 121720
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.9260107278823853,
      "learning_rate": 0.0002733666895711544,
      "loss": 1.8806,
      "step": 121721
    },
    {
      "epoch": 1.58,
      "grad_norm": 1.0464290380477905,
      "learning_rate": 0.00027336261516318886,
      "loss": 2.0364,
      "step": 121722
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.9763234257698059,
      "learning_rate": 0.0002733585407601757,
      "loss": 2.1042,
      "step": 121723
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.8948968052864075,
      "learning_rate": 0.0002733544663621157,
      "loss": 1.9985,
      "step": 121724
    },
    {
      "epoch": 1.58,
      "grad_norm": 1.007224678993225,
      "learning_rate": 0.0002733503919690096,
      "loss": 2.1754,
      "step": 121725
    },
    {
      "epoch": 1.58,
      "grad_norm": 1.3631516695022583,
      "learning_rate": 0.0002733463175808583,
      "loss": 2.0278,
      "step": 121726
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.9552000164985657,
      "learning_rate": 0.00027334224319766235,
      "loss": 2.1777,
      "step": 121727
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.9443805813789368,
      "learning_rate": 0.0002733381688194225,
      "loss": 2.1598,
      "step": 121728
    },
    {
      "epoch": 1.58,
      "grad_norm": 0.9199776649475098,
      "learning_rate": 0.0002733340944461396,
      "loss": 2.151,
      "step": 121729
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.942966103553772,
      "learning_rate": 0.0002733300200778144,
      "loss": 2.0039,
      "step": 121730
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.0974360704421997,
      "learning_rate": 0.00027332594571444766,
      "loss": 2.1956,
      "step": 121731
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.0324450731277466,
      "learning_rate": 0.00027332187135604025,
      "loss": 2.3974,
      "step": 121732
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.9812600612640381,
      "learning_rate": 0.00027331779700259275,
      "loss": 2.1497,
      "step": 121733
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.2678269147872925,
      "learning_rate": 0.0002733137226541059,
      "loss": 2.0043,
      "step": 121734
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.198428988456726,
      "learning_rate": 0.0002733096483105805,
      "loss": 2.1407,
      "step": 121735
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.1794555187225342,
      "learning_rate": 0.00027330557397201743,
      "loss": 2.0521,
      "step": 121736
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.9672616124153137,
      "learning_rate": 0.0002733014996384173,
      "loss": 2.1055,
      "step": 121737
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.2040491104125977,
      "learning_rate": 0.0002732974253097811,
      "loss": 2.1037,
      "step": 121738
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.3012486696243286,
      "learning_rate": 0.00027329335098610924,
      "loss": 2.2589,
      "step": 121739
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.1082842350006104,
      "learning_rate": 0.00027328927666740266,
      "loss": 2.2381,
      "step": 121740
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.129233479499817,
      "learning_rate": 0.0002732852023536621,
      "loss": 2.1082,
      "step": 121741
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.1508797407150269,
      "learning_rate": 0.00027328112804488833,
      "loss": 2.1639,
      "step": 121742
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.0549547672271729,
      "learning_rate": 0.00027327705374108216,
      "loss": 2.1227,
      "step": 121743
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.9746644496917725,
      "learning_rate": 0.0002732729794422443,
      "loss": 2.3565,
      "step": 121744
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.0448060035705566,
      "learning_rate": 0.0002732689051483755,
      "loss": 2.1704,
      "step": 121745
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.8527871966362,
      "learning_rate": 0.00027326483085947644,
      "loss": 1.8872,
      "step": 121746
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.9843012094497681,
      "learning_rate": 0.000273260756575548,
      "loss": 2.1313,
      "step": 121747
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.063537836074829,
      "learning_rate": 0.00027325668229659085,
      "loss": 1.9218,
      "step": 121748
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.9665142893791199,
      "learning_rate": 0.00027325260802260577,
      "loss": 2.2509,
      "step": 121749
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.0389738082885742,
      "learning_rate": 0.0002732485337535936,
      "loss": 2.2511,
      "step": 121750
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.9716411828994751,
      "learning_rate": 0.00027324445948955504,
      "loss": 2.1666,
      "step": 121751
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.9461790919303894,
      "learning_rate": 0.00027324038523049085,
      "loss": 2.3459,
      "step": 121752
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.2331345081329346,
      "learning_rate": 0.0002732363109764017,
      "loss": 2.4656,
      "step": 121753
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.9266786575317383,
      "learning_rate": 0.0002732322367272884,
      "loss": 2.2031,
      "step": 121754
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.9349508881568909,
      "learning_rate": 0.0002732281624831518,
      "loss": 2.0886,
      "step": 121755
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.2358542680740356,
      "learning_rate": 0.00027322408824399265,
      "loss": 2.2123,
      "step": 121756
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.0773917436599731,
      "learning_rate": 0.00027322001400981156,
      "loss": 2.1146,
      "step": 121757
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.9987926483154297,
      "learning_rate": 0.00027321593978060934,
      "loss": 2.3311,
      "step": 121758
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.9789677262306213,
      "learning_rate": 0.00027321186555638686,
      "loss": 2.359,
      "step": 121759
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.9161016941070557,
      "learning_rate": 0.0002732077913371448,
      "loss": 2.1379,
      "step": 121760
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.903537929058075,
      "learning_rate": 0.0002732037171228839,
      "loss": 2.0116,
      "step": 121761
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.9371258020401001,
      "learning_rate": 0.00027319964291360496,
      "loss": 2.2059,
      "step": 121762
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.9280018210411072,
      "learning_rate": 0.00027319556870930865,
      "loss": 2.1167,
      "step": 121763
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.060660719871521,
      "learning_rate": 0.0002731914945099958,
      "loss": 2.2164,
      "step": 121764
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.9981096386909485,
      "learning_rate": 0.0002731874203156672,
      "loss": 2.2034,
      "step": 121765
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.906273365020752,
      "learning_rate": 0.00027318334612632354,
      "loss": 2.2787,
      "step": 121766
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.8988086581230164,
      "learning_rate": 0.00027317927194196566,
      "loss": 2.2543,
      "step": 121767
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.9180205464363098,
      "learning_rate": 0.00027317519776259424,
      "loss": 2.0577,
      "step": 121768
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.0351473093032837,
      "learning_rate": 0.00027317112358821,
      "loss": 2.0669,
      "step": 121769
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.9557033181190491,
      "learning_rate": 0.00027316704941881376,
      "loss": 2.3245,
      "step": 121770
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.953684389591217,
      "learning_rate": 0.00027316297525440623,
      "loss": 2.2936,
      "step": 121771
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.919123113155365,
      "learning_rate": 0.0002731589010949883,
      "loss": 2.1885,
      "step": 121772
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.1422795057296753,
      "learning_rate": 0.0002731548269405606,
      "loss": 2.0405,
      "step": 121773
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.0420774221420288,
      "learning_rate": 0.0002731507527911239,
      "loss": 2.0121,
      "step": 121774
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.9761829972267151,
      "learning_rate": 0.0002731466786466791,
      "loss": 2.2243,
      "step": 121775
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.9469797611236572,
      "learning_rate": 0.00027314260450722675,
      "loss": 2.1217,
      "step": 121776
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.9650594592094421,
      "learning_rate": 0.00027313853037276765,
      "loss": 1.9626,
      "step": 121777
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.9888787865638733,
      "learning_rate": 0.00027313445624330265,
      "loss": 2.0385,
      "step": 121778
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.9261881113052368,
      "learning_rate": 0.00027313038211883246,
      "loss": 2.1032,
      "step": 121779
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.9196701049804688,
      "learning_rate": 0.00027312630799935784,
      "loss": 2.2125,
      "step": 121780
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.8886488080024719,
      "learning_rate": 0.00027312223388487966,
      "loss": 1.9295,
      "step": 121781
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.9805104732513428,
      "learning_rate": 0.00027311815977539846,
      "loss": 2.169,
      "step": 121782
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.0976171493530273,
      "learning_rate": 0.00027311408567091505,
      "loss": 2.2455,
      "step": 121783
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.0258405208587646,
      "learning_rate": 0.00027311001157143025,
      "loss": 2.1883,
      "step": 121784
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.8989651203155518,
      "learning_rate": 0.0002731059374769448,
      "loss": 2.2076,
      "step": 121785
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.9223766326904297,
      "learning_rate": 0.0002731018633874595,
      "loss": 2.0843,
      "step": 121786
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.0053714513778687,
      "learning_rate": 0.0002730977893029752,
      "loss": 2.1767,
      "step": 121787
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.9786543250083923,
      "learning_rate": 0.0002730937152234924,
      "loss": 2.0543,
      "step": 121788
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.9821454286575317,
      "learning_rate": 0.00027308964114901193,
      "loss": 2.1513,
      "step": 121789
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.0125120878219604,
      "learning_rate": 0.00027308556707953465,
      "loss": 2.1677,
      "step": 121790
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.9486662149429321,
      "learning_rate": 0.0002730814930150613,
      "loss": 2.042,
      "step": 121791
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.9769998788833618,
      "learning_rate": 0.0002730774189555926,
      "loss": 2.0362,
      "step": 121792
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.0250000953674316,
      "learning_rate": 0.0002730733449011294,
      "loss": 2.0629,
      "step": 121793
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.9282940626144409,
      "learning_rate": 0.0002730692708516723,
      "loss": 2.1806,
      "step": 121794
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.1067272424697876,
      "learning_rate": 0.00027306519680722206,
      "loss": 2.2952,
      "step": 121795
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.0090590715408325,
      "learning_rate": 0.00027306112276777956,
      "loss": 2.0414,
      "step": 121796
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.009271264076233,
      "learning_rate": 0.0002730570487333455,
      "loss": 2.2061,
      "step": 121797
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.509272575378418,
      "learning_rate": 0.0002730529747039206,
      "loss": 2.0775,
      "step": 121798
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.9314046502113342,
      "learning_rate": 0.0002730489006795058,
      "loss": 2.119,
      "step": 121799
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.9603415727615356,
      "learning_rate": 0.00027304482666010166,
      "loss": 2.3352,
      "step": 121800
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.3088414669036865,
      "learning_rate": 0.00027304075264570894,
      "loss": 2.1219,
      "step": 121801
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.0736950635910034,
      "learning_rate": 0.00027303667863632844,
      "loss": 2.0496,
      "step": 121802
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.9685205221176147,
      "learning_rate": 0.000273032604631961,
      "loss": 2.1206,
      "step": 121803
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.8918992280960083,
      "learning_rate": 0.0002730285306326073,
      "loss": 2.2386,
      "step": 121804
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.9827110767364502,
      "learning_rate": 0.0002730244566382681,
      "loss": 2.0325,
      "step": 121805
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.0189582109451294,
      "learning_rate": 0.00027302038264894417,
      "loss": 1.9702,
      "step": 121806
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.9594757556915283,
      "learning_rate": 0.0002730163086646362,
      "loss": 2.2114,
      "step": 121807
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.1665184497833252,
      "learning_rate": 0.00027301223468534504,
      "loss": 2.1731,
      "step": 121808
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.9416111707687378,
      "learning_rate": 0.0002730081607110714,
      "loss": 1.913,
      "step": 121809
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.0289361476898193,
      "learning_rate": 0.000273004086741816,
      "loss": 2.1305,
      "step": 121810
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.0332283973693848,
      "learning_rate": 0.0002730000127775799,
      "loss": 2.2106,
      "step": 121811
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.9455485939979553,
      "learning_rate": 0.0002729959388183634,
      "loss": 2.0173,
      "step": 121812
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.9460417032241821,
      "learning_rate": 0.00027299186486416747,
      "loss": 2.139,
      "step": 121813
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.8773214221000671,
      "learning_rate": 0.0002729877909149929,
      "loss": 2.0669,
      "step": 121814
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.9901872873306274,
      "learning_rate": 0.00027298371697084036,
      "loss": 2.0622,
      "step": 121815
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.89932781457901,
      "learning_rate": 0.0002729796430317107,
      "loss": 2.0363,
      "step": 121816
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.8835585117340088,
      "learning_rate": 0.0002729755690976047,
      "loss": 2.137,
      "step": 121817
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.8948951363563538,
      "learning_rate": 0.00027297149516852294,
      "loss": 2.2964,
      "step": 121818
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.971251904964447,
      "learning_rate": 0.00027296742124446633,
      "loss": 2.1232,
      "step": 121819
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.960074782371521,
      "learning_rate": 0.00027296334732543554,
      "loss": 2.0772,
      "step": 121820
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.881994366645813,
      "learning_rate": 0.00027295927341143143,
      "loss": 1.8739,
      "step": 121821
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.048178791999817,
      "learning_rate": 0.00027295519950245467,
      "loss": 2.2493,
      "step": 121822
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.9490470290184021,
      "learning_rate": 0.0002729511255985062,
      "loss": 2.125,
      "step": 121823
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.9870627522468567,
      "learning_rate": 0.00027294705169958645,
      "loss": 1.7886,
      "step": 121824
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.9659073352813721,
      "learning_rate": 0.00027294297780569636,
      "loss": 2.1159,
      "step": 121825
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.050511121749878,
      "learning_rate": 0.00027293890391683675,
      "loss": 2.1169,
      "step": 121826
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.8857726454734802,
      "learning_rate": 0.00027293483003300827,
      "loss": 1.98,
      "step": 121827
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.9797298908233643,
      "learning_rate": 0.0002729307561542117,
      "loss": 2.2792,
      "step": 121828
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.9617401361465454,
      "learning_rate": 0.00027292668228044785,
      "loss": 2.294,
      "step": 121829
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.9521704912185669,
      "learning_rate": 0.0002729226084117175,
      "loss": 1.9075,
      "step": 121830
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.0193979740142822,
      "learning_rate": 0.0002729185345480213,
      "loss": 1.8867,
      "step": 121831
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.9497154355049133,
      "learning_rate": 0.00027291446068935995,
      "loss": 2.1124,
      "step": 121832
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.9133384227752686,
      "learning_rate": 0.0002729103868357344,
      "loss": 2.0478,
      "step": 121833
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.952385425567627,
      "learning_rate": 0.00027290631298714527,
      "loss": 2.169,
      "step": 121834
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.0291389226913452,
      "learning_rate": 0.0002729022391435935,
      "loss": 2.1298,
      "step": 121835
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.98023921251297,
      "learning_rate": 0.00027289816530507956,
      "loss": 2.1919,
      "step": 121836
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.9632616639137268,
      "learning_rate": 0.0002728940914716045,
      "loss": 2.1314,
      "step": 121837
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.8919094800949097,
      "learning_rate": 0.00027289001764316885,
      "loss": 2.0897,
      "step": 121838
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.032975673675537,
      "learning_rate": 0.0002728859438197735,
      "loss": 2.2134,
      "step": 121839
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.0038951635360718,
      "learning_rate": 0.0002728818700014191,
      "loss": 2.1537,
      "step": 121840
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.0672054290771484,
      "learning_rate": 0.0002728777961881065,
      "loss": 2.1169,
      "step": 121841
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.9991616010665894,
      "learning_rate": 0.0002728737223798365,
      "loss": 2.1497,
      "step": 121842
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.0580390691757202,
      "learning_rate": 0.0002728696485766097,
      "loss": 2.2051,
      "step": 121843
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.1140131950378418,
      "learning_rate": 0.000272865574778427,
      "loss": 2.1356,
      "step": 121844
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.9236553907394409,
      "learning_rate": 0.0002728615009852891,
      "loss": 2.2243,
      "step": 121845
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.066834568977356,
      "learning_rate": 0.0002728574271971967,
      "loss": 2.1597,
      "step": 121846
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.9819428324699402,
      "learning_rate": 0.0002728533534141507,
      "loss": 2.2047,
      "step": 121847
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.9706121683120728,
      "learning_rate": 0.0002728492796361517,
      "loss": 2.0133,
      "step": 121848
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.9969104528427124,
      "learning_rate": 0.00027284520586320053,
      "loss": 2.1129,
      "step": 121849
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.9428963661193848,
      "learning_rate": 0.000272841132095298,
      "loss": 2.0992,
      "step": 121850
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.9071187973022461,
      "learning_rate": 0.0002728370583324448,
      "loss": 2.2011,
      "step": 121851
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.0113000869750977,
      "learning_rate": 0.00027283298457464174,
      "loss": 2.2277,
      "step": 121852
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.9983375668525696,
      "learning_rate": 0.00027282891082188945,
      "loss": 2.0159,
      "step": 121853
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.9280844926834106,
      "learning_rate": 0.0002728248370741889,
      "loss": 1.9894,
      "step": 121854
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.0381509065628052,
      "learning_rate": 0.0002728207633315406,
      "loss": 2.1027,
      "step": 121855
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.062493085861206,
      "learning_rate": 0.0002728166895939455,
      "loss": 2.2811,
      "step": 121856
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.0117698907852173,
      "learning_rate": 0.00027281261586140424,
      "loss": 2.1639,
      "step": 121857
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.9415349364280701,
      "learning_rate": 0.00027280854213391765,
      "loss": 1.9554,
      "step": 121858
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.0306260585784912,
      "learning_rate": 0.00027280446841148654,
      "loss": 2.2681,
      "step": 121859
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.0496736764907837,
      "learning_rate": 0.0002728003946941116,
      "loss": 1.9964,
      "step": 121860
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.1018186807632446,
      "learning_rate": 0.0002727963209817935,
      "loss": 2.0464,
      "step": 121861
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.9189615845680237,
      "learning_rate": 0.0002727922472745331,
      "loss": 2.0884,
      "step": 121862
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.9672325253486633,
      "learning_rate": 0.0002727881735723311,
      "loss": 2.1491,
      "step": 121863
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.9705631732940674,
      "learning_rate": 0.00027278409987518823,
      "loss": 2.309,
      "step": 121864
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.0314762592315674,
      "learning_rate": 0.0002727800261831054,
      "loss": 2.1738,
      "step": 121865
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.0719554424285889,
      "learning_rate": 0.0002727759524960834,
      "loss": 2.1471,
      "step": 121866
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.9287354946136475,
      "learning_rate": 0.0002727718788141227,
      "loss": 1.8111,
      "step": 121867
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.2141889333724976,
      "learning_rate": 0.0002727678051372243,
      "loss": 2.1375,
      "step": 121868
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.137811541557312,
      "learning_rate": 0.0002727637314653888,
      "loss": 2.1843,
      "step": 121869
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.9287784695625305,
      "learning_rate": 0.00027275965779861703,
      "loss": 2.1977,
      "step": 121870
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.0018742084503174,
      "learning_rate": 0.0002727555841369098,
      "loss": 1.921,
      "step": 121871
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.248036503791809,
      "learning_rate": 0.0002727515104802679,
      "loss": 2.1708,
      "step": 121872
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.8910112977027893,
      "learning_rate": 0.0002727474368286919,
      "loss": 2.1636,
      "step": 121873
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.032297134399414,
      "learning_rate": 0.0002727433631821827,
      "loss": 2.1643,
      "step": 121874
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.9767084121704102,
      "learning_rate": 0.000272739289540741,
      "loss": 2.334,
      "step": 121875
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.9931603670120239,
      "learning_rate": 0.0002727352159043676,
      "loss": 2.0774,
      "step": 121876
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.99049311876297,
      "learning_rate": 0.00027273114227306325,
      "loss": 2.0089,
      "step": 121877
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.9900479316711426,
      "learning_rate": 0.0002727270686468288,
      "loss": 2.1688,
      "step": 121878
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.1263842582702637,
      "learning_rate": 0.00027272299502566476,
      "loss": 2.0203,
      "step": 121879
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.9137954711914062,
      "learning_rate": 0.000272718921409572,
      "loss": 1.9184,
      "step": 121880
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.9414923191070557,
      "learning_rate": 0.0002727148477985514,
      "loss": 2.1503,
      "step": 121881
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.9714639782905579,
      "learning_rate": 0.00027271077419260354,
      "loss": 2.0757,
      "step": 121882
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.1905004978179932,
      "learning_rate": 0.00027270670059172927,
      "loss": 2.1042,
      "step": 121883
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.0515916347503662,
      "learning_rate": 0.0002727026269959295,
      "loss": 2.0158,
      "step": 121884
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.96782386302948,
      "learning_rate": 0.0002726985534052047,
      "loss": 2.1577,
      "step": 121885
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.9171348810195923,
      "learning_rate": 0.00027269447981955575,
      "loss": 1.9853,
      "step": 121886
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.1010245084762573,
      "learning_rate": 0.0002726904062389834,
      "loss": 1.9821,
      "step": 121887
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.046189308166504,
      "learning_rate": 0.0002726863326634884,
      "loss": 2.1342,
      "step": 121888
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.0279549360275269,
      "learning_rate": 0.0002726822590930716,
      "loss": 1.9109,
      "step": 121889
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.0899978876113892,
      "learning_rate": 0.00027267818552773374,
      "loss": 2.0304,
      "step": 121890
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.9697069525718689,
      "learning_rate": 0.0002726741119674754,
      "loss": 2.2239,
      "step": 121891
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.9153929948806763,
      "learning_rate": 0.0002726700384122975,
      "loss": 2.3529,
      "step": 121892
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.9497650265693665,
      "learning_rate": 0.0002726659648622007,
      "loss": 2.1569,
      "step": 121893
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.9530940651893616,
      "learning_rate": 0.00027266189131718585,
      "loss": 2.1476,
      "step": 121894
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.0371042490005493,
      "learning_rate": 0.00027265781777725367,
      "loss": 2.2613,
      "step": 121895
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.8633288741111755,
      "learning_rate": 0.00027265374424240504,
      "loss": 2.2846,
      "step": 121896
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.9438894391059875,
      "learning_rate": 0.00027264967071264045,
      "loss": 2.166,
      "step": 121897
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.9787455201148987,
      "learning_rate": 0.00027264559718796083,
      "loss": 2.257,
      "step": 121898
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.9501805901527405,
      "learning_rate": 0.0002726415236683669,
      "loss": 2.13,
      "step": 121899
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.9579647183418274,
      "learning_rate": 0.00027263745015385945,
      "loss": 2.1941,
      "step": 121900
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.9917221665382385,
      "learning_rate": 0.0002726333766444392,
      "loss": 2.0745,
      "step": 121901
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.9947903156280518,
      "learning_rate": 0.000272629303140107,
      "loss": 2.1454,
      "step": 121902
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.083474040031433,
      "learning_rate": 0.0002726252296408634,
      "loss": 2.0249,
      "step": 121903
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.0641857385635376,
      "learning_rate": 0.0002726211561467094,
      "loss": 2.3519,
      "step": 121904
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.9227301478385925,
      "learning_rate": 0.00027261708265764555,
      "loss": 2.1452,
      "step": 121905
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.1197737455368042,
      "learning_rate": 0.0002726130091736727,
      "loss": 2.0956,
      "step": 121906
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.924430787563324,
      "learning_rate": 0.00027260893569479164,
      "loss": 2.0544,
      "step": 121907
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.9884803891181946,
      "learning_rate": 0.0002726048622210031,
      "loss": 2.3129,
      "step": 121908
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.9715519547462463,
      "learning_rate": 0.00027260078875230795,
      "loss": 2.1188,
      "step": 121909
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.9653279781341553,
      "learning_rate": 0.0002725967152887067,
      "loss": 2.0615,
      "step": 121910
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.953890323638916,
      "learning_rate": 0.00027259264183020026,
      "loss": 2.2534,
      "step": 121911
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.046569585800171,
      "learning_rate": 0.00027258856837678936,
      "loss": 2.2102,
      "step": 121912
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.9262260794639587,
      "learning_rate": 0.00027258449492847475,
      "loss": 2.1959,
      "step": 121913
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.9708936810493469,
      "learning_rate": 0.00027258042148525715,
      "loss": 2.1814,
      "step": 121914
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.134527325630188,
      "learning_rate": 0.00027257634804713757,
      "loss": 2.2581,
      "step": 121915
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.0431363582611084,
      "learning_rate": 0.0002725722746141164,
      "loss": 2.1072,
      "step": 121916
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.9737104177474976,
      "learning_rate": 0.0002725682011861946,
      "loss": 2.2804,
      "step": 121917
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.0042132139205933,
      "learning_rate": 0.00027256412776337284,
      "loss": 2.0336,
      "step": 121918
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.0576883554458618,
      "learning_rate": 0.0002725600543456519,
      "loss": 2.3026,
      "step": 121919
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.0031366348266602,
      "learning_rate": 0.0002725559809330326,
      "loss": 2.0939,
      "step": 121920
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.9240264296531677,
      "learning_rate": 0.0002725519075255158,
      "loss": 1.9719,
      "step": 121921
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.9296094179153442,
      "learning_rate": 0.000272547834123102,
      "loss": 2.1457,
      "step": 121922
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.9791064858436584,
      "learning_rate": 0.00027254376072579213,
      "loss": 2.27,
      "step": 121923
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.0857317447662354,
      "learning_rate": 0.00027253968733358683,
      "loss": 2.2968,
      "step": 121924
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.9785082340240479,
      "learning_rate": 0.0002725356139464869,
      "loss": 2.1621,
      "step": 121925
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.9188757538795471,
      "learning_rate": 0.00027253154056449313,
      "loss": 2.2043,
      "step": 121926
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.0259217023849487,
      "learning_rate": 0.00027252746718760637,
      "loss": 2.3649,
      "step": 121927
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.0457957983016968,
      "learning_rate": 0.00027252339381582716,
      "loss": 2.0228,
      "step": 121928
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.0110024213790894,
      "learning_rate": 0.00027251932044915645,
      "loss": 2.211,
      "step": 121929
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.937414824962616,
      "learning_rate": 0.0002725152470875948,
      "loss": 2.0646,
      "step": 121930
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.9155212640762329,
      "learning_rate": 0.0002725111737311432,
      "loss": 2.13,
      "step": 121931
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.1058183908462524,
      "learning_rate": 0.0002725071003798022,
      "loss": 2.1066,
      "step": 121932
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.9929919838905334,
      "learning_rate": 0.00027250302703357276,
      "loss": 2.1475,
      "step": 121933
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.9822561740875244,
      "learning_rate": 0.00027249895369245544,
      "loss": 2.1546,
      "step": 121934
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.0453261137008667,
      "learning_rate": 0.00027249488035645107,
      "loss": 2.263,
      "step": 121935
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.1114423274993896,
      "learning_rate": 0.00027249080702556054,
      "loss": 1.9304,
      "step": 121936
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.0212804079055786,
      "learning_rate": 0.00027248673369978437,
      "loss": 2.223,
      "step": 121937
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.858635663986206,
      "learning_rate": 0.00027248266037912345,
      "loss": 2.0861,
      "step": 121938
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.9892714619636536,
      "learning_rate": 0.00027247858706357857,
      "loss": 2.1285,
      "step": 121939
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.0871995687484741,
      "learning_rate": 0.0002724745137531504,
      "loss": 2.2681,
      "step": 121940
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.0742826461791992,
      "learning_rate": 0.00027247044044783974,
      "loss": 1.93,
      "step": 121941
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.9194562435150146,
      "learning_rate": 0.0002724663671476473,
      "loss": 2.324,
      "step": 121942
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.9295468926429749,
      "learning_rate": 0.000272462293852574,
      "loss": 1.9694,
      "step": 121943
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.9739161133766174,
      "learning_rate": 0.0002724582205626204,
      "loss": 2.1505,
      "step": 121944
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.2250388860702515,
      "learning_rate": 0.0002724541472777874,
      "loss": 2.287,
      "step": 121945
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.0385737419128418,
      "learning_rate": 0.00027245007399807566,
      "loss": 2.0548,
      "step": 121946
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.9595888257026672,
      "learning_rate": 0.0002724460007234859,
      "loss": 2.4066,
      "step": 121947
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.122801423072815,
      "learning_rate": 0.000272441927454019,
      "loss": 2.0391,
      "step": 121948
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.1619141101837158,
      "learning_rate": 0.00027243785418967564,
      "loss": 1.9409,
      "step": 121949
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.9886605143547058,
      "learning_rate": 0.0002724337809304566,
      "loss": 2.1871,
      "step": 121950
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.9814507365226746,
      "learning_rate": 0.0002724297076763628,
      "loss": 2.0989,
      "step": 121951
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.6378655433654785,
      "learning_rate": 0.00027242563442739476,
      "loss": 1.9775,
      "step": 121952
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.0873562097549438,
      "learning_rate": 0.0002724215611835532,
      "loss": 2.1449,
      "step": 121953
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.9737197756767273,
      "learning_rate": 0.00027241748794483905,
      "loss": 2.0949,
      "step": 121954
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.9889772534370422,
      "learning_rate": 0.000272413414711253,
      "loss": 1.9574,
      "step": 121955
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.055156946182251,
      "learning_rate": 0.0002724093414827958,
      "loss": 2.4205,
      "step": 121956
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.5996359586715698,
      "learning_rate": 0.0002724052682594684,
      "loss": 2.0264,
      "step": 121957
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.961892306804657,
      "learning_rate": 0.00027240119504127124,
      "loss": 2.1408,
      "step": 121958
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.914657711982727,
      "learning_rate": 0.0002723971218282052,
      "loss": 2.1026,
      "step": 121959
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.1335867643356323,
      "learning_rate": 0.00027239304862027106,
      "loss": 2.1651,
      "step": 121960
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.6644105911254883,
      "learning_rate": 0.00027238897541746954,
      "loss": 1.9563,
      "step": 121961
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.037447452545166,
      "learning_rate": 0.0002723849022198015,
      "loss": 2.0208,
      "step": 121962
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.8668351173400879,
      "learning_rate": 0.0002723808290272677,
      "loss": 2.1845,
      "step": 121963
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.2525123357772827,
      "learning_rate": 0.00027237675583986873,
      "loss": 1.9989,
      "step": 121964
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.2672876119613647,
      "learning_rate": 0.0002723726826576054,
      "loss": 2.1965,
      "step": 121965
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.9151027202606201,
      "learning_rate": 0.00027236860948047856,
      "loss": 2.0793,
      "step": 121966
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.9392252564430237,
      "learning_rate": 0.0002723645363084889,
      "loss": 2.219,
      "step": 121967
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.9170100688934326,
      "learning_rate": 0.0002723604631416372,
      "loss": 1.8705,
      "step": 121968
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.9881141781806946,
      "learning_rate": 0.00027235638997992435,
      "loss": 1.9883,
      "step": 121969
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.986646831035614,
      "learning_rate": 0.00027235231682335077,
      "loss": 2.2471,
      "step": 121970
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.0675568580627441,
      "learning_rate": 0.0002723482436719175,
      "loss": 2.2785,
      "step": 121971
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.993252694606781,
      "learning_rate": 0.0002723441705256252,
      "loss": 1.9845,
      "step": 121972
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.0929431915283203,
      "learning_rate": 0.0002723400973844746,
      "loss": 2.1862,
      "step": 121973
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.2524951696395874,
      "learning_rate": 0.00027233602424846655,
      "loss": 2.2422,
      "step": 121974
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.3991093635559082,
      "learning_rate": 0.00027233195111760173,
      "loss": 2.2943,
      "step": 121975
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.9376326203346252,
      "learning_rate": 0.000272327877991881,
      "loss": 1.9582,
      "step": 121976
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.9750393629074097,
      "learning_rate": 0.000272323804871305,
      "loss": 1.8945,
      "step": 121977
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.179984450340271,
      "learning_rate": 0.00027231973175587446,
      "loss": 2.0627,
      "step": 121978
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.1652523279190063,
      "learning_rate": 0.00027231565864559024,
      "loss": 2.2252,
      "step": 121979
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.9237521290779114,
      "learning_rate": 0.000272311585540453,
      "loss": 1.8971,
      "step": 121980
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.9269091486930847,
      "learning_rate": 0.00027230751244046366,
      "loss": 1.9071,
      "step": 121981
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.4289546012878418,
      "learning_rate": 0.000272303439345623,
      "loss": 1.9544,
      "step": 121982
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.0794997215270996,
      "learning_rate": 0.00027229936625593144,
      "loss": 2.4878,
      "step": 121983
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.0269057750701904,
      "learning_rate": 0.00027229529317139,
      "loss": 2.0627,
      "step": 121984
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.1408530473709106,
      "learning_rate": 0.00027229122009199934,
      "loss": 2.1593,
      "step": 121985
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.174415111541748,
      "learning_rate": 0.0002722871470177603,
      "loss": 2.1813,
      "step": 121986
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.9754590392112732,
      "learning_rate": 0.00027228307394867364,
      "loss": 2.2762,
      "step": 121987
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.0514326095581055,
      "learning_rate": 0.0002722790008847402,
      "loss": 2.1129,
      "step": 121988
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.0482313632965088,
      "learning_rate": 0.0002722749278259604,
      "loss": 2.1209,
      "step": 121989
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.9372321963310242,
      "learning_rate": 0.0002722708547723353,
      "loss": 2.0677,
      "step": 121990
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.271634817123413,
      "learning_rate": 0.00027226678172386556,
      "loss": 2.1992,
      "step": 121991
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.9230942726135254,
      "learning_rate": 0.0002722627086805519,
      "loss": 2.1433,
      "step": 121992
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.0792579650878906,
      "learning_rate": 0.0002722586356423952,
      "loss": 2.1802,
      "step": 121993
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.0464054346084595,
      "learning_rate": 0.00027225456260939623,
      "loss": 2.0146,
      "step": 121994
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.0055330991744995,
      "learning_rate": 0.00027225048958155556,
      "loss": 1.8886,
      "step": 121995
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.8942831158638,
      "learning_rate": 0.000272246416558874,
      "loss": 1.9899,
      "step": 121996
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.9074687957763672,
      "learning_rate": 0.0002722423435413524,
      "loss": 2.2442,
      "step": 121997
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.186381220817566,
      "learning_rate": 0.0002722382705289914,
      "loss": 2.0604,
      "step": 121998
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.9138827919960022,
      "learning_rate": 0.00027223419752179194,
      "loss": 2.4179,
      "step": 121999
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.9196085929870605,
      "learning_rate": 0.00027223012451975474,
      "loss": 2.3221,
      "step": 122000
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.0088781118392944,
      "learning_rate": 0.00027222605152288034,
      "loss": 2.1328,
      "step": 122001
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.9872370958328247,
      "learning_rate": 0.0002722219785311696,
      "loss": 2.245,
      "step": 122002
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.9944792985916138,
      "learning_rate": 0.00027221790554462344,
      "loss": 2.1341,
      "step": 122003
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.5920555591583252,
      "learning_rate": 0.0002722138325632424,
      "loss": 2.037,
      "step": 122004
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.0130152702331543,
      "learning_rate": 0.0002722097595870274,
      "loss": 2.1028,
      "step": 122005
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.1210918426513672,
      "learning_rate": 0.00027220568661597916,
      "loss": 1.8827,
      "step": 122006
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.0805555582046509,
      "learning_rate": 0.00027220161365009836,
      "loss": 1.9846,
      "step": 122007
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.0720818042755127,
      "learning_rate": 0.0002721975406893858,
      "loss": 2.0228,
      "step": 122008
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.9763108491897583,
      "learning_rate": 0.0002721934677338422,
      "loss": 2.1732,
      "step": 122009
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.9574651122093201,
      "learning_rate": 0.00027218939478346844,
      "loss": 2.0061,
      "step": 122010
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.011965274810791,
      "learning_rate": 0.0002721853218382651,
      "loss": 1.9872,
      "step": 122011
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.9421094655990601,
      "learning_rate": 0.00027218124889823316,
      "loss": 2.073,
      "step": 122012
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.9481104016304016,
      "learning_rate": 0.0002721771759633732,
      "loss": 2.1541,
      "step": 122013
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.002726435661316,
      "learning_rate": 0.00027217310303368603,
      "loss": 2.2313,
      "step": 122014
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.9536873698234558,
      "learning_rate": 0.0002721690301091724,
      "loss": 1.9638,
      "step": 122015
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.9887121319770813,
      "learning_rate": 0.000272164957189833,
      "loss": 2.3594,
      "step": 122016
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.9571001529693604,
      "learning_rate": 0.0002721608842756687,
      "loss": 1.8255,
      "step": 122017
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.9114640951156616,
      "learning_rate": 0.0002721568113666803,
      "loss": 1.9157,
      "step": 122018
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.9617008566856384,
      "learning_rate": 0.0002721527384628684,
      "loss": 2.118,
      "step": 122019
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.852344810962677,
      "learning_rate": 0.00027214866556423377,
      "loss": 2.1696,
      "step": 122020
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.9696729183197021,
      "learning_rate": 0.0002721445926707774,
      "loss": 2.1835,
      "step": 122021
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.1065481901168823,
      "learning_rate": 0.00027214051978249974,
      "loss": 2.359,
      "step": 122022
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.9450929164886475,
      "learning_rate": 0.0002721364468994017,
      "loss": 2.1185,
      "step": 122023
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.9174962043762207,
      "learning_rate": 0.0002721323740214841,
      "loss": 2.2921,
      "step": 122024
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.2243247032165527,
      "learning_rate": 0.00027212830114874755,
      "loss": 1.976,
      "step": 122025
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.0529588460922241,
      "learning_rate": 0.0002721242282811929,
      "loss": 2.2219,
      "step": 122026
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.9805734753608704,
      "learning_rate": 0.0002721201554188208,
      "loss": 2.4477,
      "step": 122027
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.9865283370018005,
      "learning_rate": 0.0002721160825616322,
      "loss": 2.2486,
      "step": 122028
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.9559018015861511,
      "learning_rate": 0.0002721120097096277,
      "loss": 2.03,
      "step": 122029
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.9184232950210571,
      "learning_rate": 0.00027210793686280815,
      "loss": 2.0386,
      "step": 122030
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.2000622749328613,
      "learning_rate": 0.0002721038640211742,
      "loss": 1.9927,
      "step": 122031
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.0657328367233276,
      "learning_rate": 0.00027209979118472666,
      "loss": 2.2779,
      "step": 122032
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.0890451669692993,
      "learning_rate": 0.0002720957183534663,
      "loss": 1.9056,
      "step": 122033
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.974550187587738,
      "learning_rate": 0.0002720916455273939,
      "loss": 2.1113,
      "step": 122034
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.9183998703956604,
      "learning_rate": 0.0002720875727065102,
      "loss": 2.1286,
      "step": 122035
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.8730563521385193,
      "learning_rate": 0.00027208349989081603,
      "loss": 1.9232,
      "step": 122036
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.9846239686012268,
      "learning_rate": 0.0002720794270803119,
      "loss": 2.3135,
      "step": 122037
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.9783251285552979,
      "learning_rate": 0.0002720753542749988,
      "loss": 1.9044,
      "step": 122038
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.9201983213424683,
      "learning_rate": 0.0002720712814748774,
      "loss": 1.9545,
      "step": 122039
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.9371310472488403,
      "learning_rate": 0.00027206720867994845,
      "loss": 2.0479,
      "step": 122040
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.9195760488510132,
      "learning_rate": 0.0002720631358902128,
      "loss": 1.9983,
      "step": 122041
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.0966534614562988,
      "learning_rate": 0.0002720590631056711,
      "loss": 1.9665,
      "step": 122042
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.0018296241760254,
      "learning_rate": 0.00027205499032632424,
      "loss": 2.2449,
      "step": 122043
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.9475666880607605,
      "learning_rate": 0.0002720509175521728,
      "loss": 2.0658,
      "step": 122044
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.1034239530563354,
      "learning_rate": 0.0002720468447832176,
      "loss": 2.1704,
      "step": 122045
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.8708630204200745,
      "learning_rate": 0.00027204277201945944,
      "loss": 2.3227,
      "step": 122046
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.9170873165130615,
      "learning_rate": 0.0002720386992608991,
      "loss": 1.761,
      "step": 122047
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.0074983835220337,
      "learning_rate": 0.0002720346265075372,
      "loss": 2.2726,
      "step": 122048
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.034378170967102,
      "learning_rate": 0.00027203055375937476,
      "loss": 2.1784,
      "step": 122049
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.0149686336517334,
      "learning_rate": 0.0002720264810164122,
      "loss": 2.0092,
      "step": 122050
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.017461895942688,
      "learning_rate": 0.00027202240827865054,
      "loss": 2.1479,
      "step": 122051
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.9745951890945435,
      "learning_rate": 0.0002720183355460904,
      "loss": 2.2355,
      "step": 122052
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.9315882325172424,
      "learning_rate": 0.00027201426281873254,
      "loss": 2.0459,
      "step": 122053
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.9603330492973328,
      "learning_rate": 0.0002720101900965778,
      "loss": 2.157,
      "step": 122054
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.9147495627403259,
      "learning_rate": 0.000272006117379627,
      "loss": 2.0581,
      "step": 122055
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.0246278047561646,
      "learning_rate": 0.00027200204466788064,
      "loss": 2.0501,
      "step": 122056
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.9575960040092468,
      "learning_rate": 0.00027199797196133965,
      "loss": 2.3259,
      "step": 122057
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.8773167729377747,
      "learning_rate": 0.00027199389926000484,
      "loss": 2.2346,
      "step": 122058
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.9309148788452148,
      "learning_rate": 0.0002719898265638768,
      "loss": 1.9252,
      "step": 122059
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.9728052616119385,
      "learning_rate": 0.0002719857538729564,
      "loss": 1.9413,
      "step": 122060
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.9772621393203735,
      "learning_rate": 0.0002719816811872445,
      "loss": 2.1539,
      "step": 122061
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.8657099604606628,
      "learning_rate": 0.0002719776085067416,
      "loss": 1.9545,
      "step": 122062
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.966201663017273,
      "learning_rate": 0.0002719735358314486,
      "loss": 2.033,
      "step": 122063
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.988741397857666,
      "learning_rate": 0.0002719694631613663,
      "loss": 2.1502,
      "step": 122064
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.8963585495948792,
      "learning_rate": 0.0002719653904964953,
      "loss": 1.8543,
      "step": 122065
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.9998145699501038,
      "learning_rate": 0.0002719613178368366,
      "loss": 2.2953,
      "step": 122066
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.9955735802650452,
      "learning_rate": 0.00027195724518239085,
      "loss": 2.1088,
      "step": 122067
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.962916910648346,
      "learning_rate": 0.0002719531725331587,
      "loss": 2.1639,
      "step": 122068
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.8948060274124146,
      "learning_rate": 0.00027194909988914097,
      "loss": 2.0062,
      "step": 122069
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.936877429485321,
      "learning_rate": 0.0002719450272503384,
      "loss": 2.4582,
      "step": 122070
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.9294397234916687,
      "learning_rate": 0.0002719409546167518,
      "loss": 2.1716,
      "step": 122071
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.062286376953125,
      "learning_rate": 0.0002719368819883819,
      "loss": 2.168,
      "step": 122072
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.189389944076538,
      "learning_rate": 0.0002719328093652296,
      "loss": 2.2644,
      "step": 122073
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.9643917679786682,
      "learning_rate": 0.0002719287367472954,
      "loss": 2.2253,
      "step": 122074
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.9696998000144958,
      "learning_rate": 0.00027192466413458017,
      "loss": 2.3296,
      "step": 122075
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.9972399473190308,
      "learning_rate": 0.0002719205915270847,
      "loss": 2.3249,
      "step": 122076
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.9908234477043152,
      "learning_rate": 0.0002719165189248096,
      "loss": 2.3408,
      "step": 122077
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.8961782455444336,
      "learning_rate": 0.0002719124463277559,
      "loss": 1.9872,
      "step": 122078
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.9159463047981262,
      "learning_rate": 0.0002719083737359243,
      "loss": 1.9869,
      "step": 122079
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.9010074734687805,
      "learning_rate": 0.0002719043011493153,
      "loss": 1.9979,
      "step": 122080
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.9176223278045654,
      "learning_rate": 0.0002719002285679298,
      "loss": 2.0417,
      "step": 122081
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.0089179277420044,
      "learning_rate": 0.00027189615599176864,
      "loss": 2.032,
      "step": 122082
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.9734036326408386,
      "learning_rate": 0.0002718920834208325,
      "loss": 2.2421,
      "step": 122083
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.060502529144287,
      "learning_rate": 0.0002718880108551221,
      "loss": 2.1447,
      "step": 122084
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.9844249486923218,
      "learning_rate": 0.00027188393829463844,
      "loss": 2.3658,
      "step": 122085
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.0070641040802002,
      "learning_rate": 0.000271879865739382,
      "loss": 2.1774,
      "step": 122086
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.1455944776535034,
      "learning_rate": 0.0002718757931893535,
      "loss": 2.3593,
      "step": 122087
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.0263545513153076,
      "learning_rate": 0.0002718717206445539,
      "loss": 2.0009,
      "step": 122088
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.9293136596679688,
      "learning_rate": 0.0002718676481049838,
      "loss": 2.0409,
      "step": 122089
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.0064946413040161,
      "learning_rate": 0.00027186357557064416,
      "loss": 2.3196,
      "step": 122090
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.9800559282302856,
      "learning_rate": 0.0002718595030415357,
      "loss": 2.1604,
      "step": 122091
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.9170085191726685,
      "learning_rate": 0.0002718554305176589,
      "loss": 2.1637,
      "step": 122092
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.9907684326171875,
      "learning_rate": 0.0002718513579990147,
      "loss": 2.1433,
      "step": 122093
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.8716332912445068,
      "learning_rate": 0.00027184728548560394,
      "loss": 2.1149,
      "step": 122094
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.9715715646743774,
      "learning_rate": 0.00027184321297742727,
      "loss": 1.9512,
      "step": 122095
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.9266486167907715,
      "learning_rate": 0.00027183914047448546,
      "loss": 2.0521,
      "step": 122096
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.9693633913993835,
      "learning_rate": 0.00027183506797677937,
      "loss": 2.0584,
      "step": 122097
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.9503504633903503,
      "learning_rate": 0.00027183099548430965,
      "loss": 2.172,
      "step": 122098
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.9639771580696106,
      "learning_rate": 0.00027182692299707707,
      "loss": 2.0546,
      "step": 122099
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.9124776124954224,
      "learning_rate": 0.00027182285051508237,
      "loss": 1.9266,
      "step": 122100
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.9932190775871277,
      "learning_rate": 0.0002718187780383263,
      "loss": 2.1869,
      "step": 122101
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.9444842338562012,
      "learning_rate": 0.00027181470556680967,
      "loss": 2.2665,
      "step": 122102
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.9613158106803894,
      "learning_rate": 0.0002718106331005333,
      "loss": 2.1933,
      "step": 122103
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.9785680770874023,
      "learning_rate": 0.0002718065606394978,
      "loss": 2.235,
      "step": 122104
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.0074255466461182,
      "learning_rate": 0.000271802488183704,
      "loss": 2.2872,
      "step": 122105
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.0167672634124756,
      "learning_rate": 0.00027179841573315267,
      "loss": 2.2799,
      "step": 122106
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.052363634109497,
      "learning_rate": 0.00027179434328784456,
      "loss": 2.0311,
      "step": 122107
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.997930109500885,
      "learning_rate": 0.00027179027084778035,
      "loss": 2.1992,
      "step": 122108
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.9331618547439575,
      "learning_rate": 0.00027178619841296086,
      "loss": 2.0021,
      "step": 122109
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.9529833793640137,
      "learning_rate": 0.0002717821259833869,
      "loss": 2.1228,
      "step": 122110
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.9769471287727356,
      "learning_rate": 0.0002717780535590591,
      "loss": 2.0446,
      "step": 122111
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.8987509608268738,
      "learning_rate": 0.00027177398113997835,
      "loss": 2.2765,
      "step": 122112
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.9593850374221802,
      "learning_rate": 0.00027176990872614543,
      "loss": 2.2979,
      "step": 122113
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.883077085018158,
      "learning_rate": 0.0002717658363175609,
      "loss": 2.1269,
      "step": 122114
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.950715959072113,
      "learning_rate": 0.00027176176391422563,
      "loss": 2.146,
      "step": 122115
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.107786774635315,
      "learning_rate": 0.0002717576915161405,
      "loss": 2.2784,
      "step": 122116
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.949317455291748,
      "learning_rate": 0.000271753619123306,
      "loss": 2.2836,
      "step": 122117
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.0349503755569458,
      "learning_rate": 0.0002717495467357231,
      "loss": 2.2839,
      "step": 122118
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.9345829486846924,
      "learning_rate": 0.0002717454743533925,
      "loss": 2.0666,
      "step": 122119
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.0268537998199463,
      "learning_rate": 0.000271741401976315,
      "loss": 2.2674,
      "step": 122120
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.9803749918937683,
      "learning_rate": 0.00027173732960449125,
      "loss": 2.2532,
      "step": 122121
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.9742031693458557,
      "learning_rate": 0.0002717332572379221,
      "loss": 2.0428,
      "step": 122122
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.9796637892723083,
      "learning_rate": 0.00027172918487660823,
      "loss": 2.0346,
      "step": 122123
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.0905483961105347,
      "learning_rate": 0.0002717251125205504,
      "loss": 2.3116,
      "step": 122124
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.9070517420768738,
      "learning_rate": 0.00027172104016974947,
      "loss": 2.1908,
      "step": 122125
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.9828560948371887,
      "learning_rate": 0.00027171696782420606,
      "loss": 2.2634,
      "step": 122126
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.9930669665336609,
      "learning_rate": 0.00027171289548392114,
      "loss": 2.094,
      "step": 122127
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.9412779211997986,
      "learning_rate": 0.00027170882314889525,
      "loss": 1.8902,
      "step": 122128
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.9418331980705261,
      "learning_rate": 0.0002717047508191292,
      "loss": 2.217,
      "step": 122129
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.9874287247657776,
      "learning_rate": 0.00027170067849462376,
      "loss": 2.1495,
      "step": 122130
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.9687830805778503,
      "learning_rate": 0.00027169660617537973,
      "loss": 2.1168,
      "step": 122131
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.8975171446800232,
      "learning_rate": 0.0002716925338613978,
      "loss": 2.1743,
      "step": 122132
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.8320863246917725,
      "learning_rate": 0.00027168846155267874,
      "loss": 2.0834,
      "step": 122133
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.9618240594863892,
      "learning_rate": 0.0002716843892492235,
      "loss": 2.099,
      "step": 122134
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.0110139846801758,
      "learning_rate": 0.00027168031695103253,
      "loss": 2.15,
      "step": 122135
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.0500010251998901,
      "learning_rate": 0.00027167624465810675,
      "loss": 2.1471,
      "step": 122136
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.9558516144752502,
      "learning_rate": 0.00027167217237044684,
      "loss": 2.2131,
      "step": 122137
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.9209486246109009,
      "learning_rate": 0.00027166810008805365,
      "loss": 2.1129,
      "step": 122138
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.9972824454307556,
      "learning_rate": 0.00027166402781092784,
      "loss": 1.9962,
      "step": 122139
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.9671196937561035,
      "learning_rate": 0.00027165995553907044,
      "loss": 2.0036,
      "step": 122140
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.0465507507324219,
      "learning_rate": 0.0002716558832724818,
      "loss": 2.0213,
      "step": 122141
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.0127689838409424,
      "learning_rate": 0.0002716518110111629,
      "loss": 2.2161,
      "step": 122142
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.0295097827911377,
      "learning_rate": 0.0002716477387551144,
      "loss": 2.2792,
      "step": 122143
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.0247682332992554,
      "learning_rate": 0.00027164366650433714,
      "loss": 1.9629,
      "step": 122144
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.0074206590652466,
      "learning_rate": 0.0002716395942588319,
      "loss": 1.9866,
      "step": 122145
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.9053282737731934,
      "learning_rate": 0.00027163552201859946,
      "loss": 2.1812,
      "step": 122146
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.1718162298202515,
      "learning_rate": 0.00027163144978364044,
      "loss": 2.0297,
      "step": 122147
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.0955288410186768,
      "learning_rate": 0.00027162737755395564,
      "loss": 2.307,
      "step": 122148
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.9329305291175842,
      "learning_rate": 0.00027162330532954585,
      "loss": 2.1232,
      "step": 122149
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.9484809637069702,
      "learning_rate": 0.00027161923311041185,
      "loss": 1.9601,
      "step": 122150
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.9336641430854797,
      "learning_rate": 0.00027161516089655433,
      "loss": 2.1829,
      "step": 122151
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.082723617553711,
      "learning_rate": 0.0002716110886879742,
      "loss": 2.1254,
      "step": 122152
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.9446582794189453,
      "learning_rate": 0.00027160701648467207,
      "loss": 2.2762,
      "step": 122153
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.9074468016624451,
      "learning_rate": 0.0002716029442866487,
      "loss": 2.0935,
      "step": 122154
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.969782292842865,
      "learning_rate": 0.0002715988720939048,
      "loss": 2.1096,
      "step": 122155
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.9659361839294434,
      "learning_rate": 0.00027159479990644127,
      "loss": 2.1025,
      "step": 122156
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.9964777231216431,
      "learning_rate": 0.0002715907277242588,
      "loss": 1.968,
      "step": 122157
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.0942214727401733,
      "learning_rate": 0.0002715866555473583,
      "loss": 2.1694,
      "step": 122158
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.9588201642036438,
      "learning_rate": 0.00027158258337574016,
      "loss": 1.9795,
      "step": 122159
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.9118636846542358,
      "learning_rate": 0.00027157851120940545,
      "loss": 1.9891,
      "step": 122160
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.0447490215301514,
      "learning_rate": 0.00027157443904835476,
      "loss": 2.2663,
      "step": 122161
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.9630107283592224,
      "learning_rate": 0.00027157036689258896,
      "loss": 2.0607,
      "step": 122162
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.9282117486000061,
      "learning_rate": 0.00027156629474210875,
      "loss": 2.2098,
      "step": 122163
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.9740679860115051,
      "learning_rate": 0.00027156222259691506,
      "loss": 2.3322,
      "step": 122164
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.137708067893982,
      "learning_rate": 0.0002715581504570084,
      "loss": 2.0413,
      "step": 122165
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.0045009851455688,
      "learning_rate": 0.0002715540783223895,
      "loss": 2.3459,
      "step": 122166
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.8786430358886719,
      "learning_rate": 0.00027155000619305926,
      "loss": 2.0375,
      "step": 122167
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.8761338591575623,
      "learning_rate": 0.00027154593406901847,
      "loss": 1.8964,
      "step": 122168
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.9935873746871948,
      "learning_rate": 0.00027154186195026776,
      "loss": 2.143,
      "step": 122169
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.9979045391082764,
      "learning_rate": 0.0002715377898368081,
      "loss": 2.272,
      "step": 122170
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.0328333377838135,
      "learning_rate": 0.00027153371772864005,
      "loss": 2.3193,
      "step": 122171
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.0446791648864746,
      "learning_rate": 0.00027152964562576434,
      "loss": 1.9523,
      "step": 122172
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.0223076343536377,
      "learning_rate": 0.00027152557352818187,
      "loss": 2.192,
      "step": 122173
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.9232363104820251,
      "learning_rate": 0.0002715215014358933,
      "loss": 2.1487,
      "step": 122174
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.0714476108551025,
      "learning_rate": 0.0002715174293488994,
      "loss": 2.042,
      "step": 122175
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.9233419895172119,
      "learning_rate": 0.000271513357267201,
      "loss": 1.8575,
      "step": 122176
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.9842156171798706,
      "learning_rate": 0.00027150928519079884,
      "loss": 2.1497,
      "step": 122177
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.9660013914108276,
      "learning_rate": 0.00027150521311969356,
      "loss": 2.1903,
      "step": 122178
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.8986397981643677,
      "learning_rate": 0.000271501141053886,
      "loss": 2.219,
      "step": 122179
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.9346401691436768,
      "learning_rate": 0.0002714970689933769,
      "loss": 1.9655,
      "step": 122180
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.9049705266952515,
      "learning_rate": 0.00027149299693816713,
      "loss": 2.1555,
      "step": 122181
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.9613175392150879,
      "learning_rate": 0.00027148892488825725,
      "loss": 1.9506,
      "step": 122182
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.0140831470489502,
      "learning_rate": 0.0002714848528436483,
      "loss": 2.2698,
      "step": 122183
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.049389362335205,
      "learning_rate": 0.00027148078080434067,
      "loss": 2.1177,
      "step": 122184
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.9798123240470886,
      "learning_rate": 0.0002714767087703353,
      "loss": 2.2018,
      "step": 122185
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.0672279596328735,
      "learning_rate": 0.00027147263674163303,
      "loss": 2.0248,
      "step": 122186
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.9205804467201233,
      "learning_rate": 0.0002714685647182345,
      "loss": 1.9356,
      "step": 122187
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.1206080913543701,
      "learning_rate": 0.0002714644927001404,
      "loss": 2.2629,
      "step": 122188
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.9625717401504517,
      "learning_rate": 0.00027146042068735177,
      "loss": 2.1297,
      "step": 122189
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.9651038646697998,
      "learning_rate": 0.00027145634867986915,
      "loss": 2.0577,
      "step": 122190
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.9777286648750305,
      "learning_rate": 0.0002714522766776933,
      "loss": 2.1193,
      "step": 122191
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.9745661616325378,
      "learning_rate": 0.000271448204680825,
      "loss": 2.067,
      "step": 122192
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.0097630023956299,
      "learning_rate": 0.000271444132689265,
      "loss": 2.417,
      "step": 122193
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.0429162979125977,
      "learning_rate": 0.0002714400607030141,
      "loss": 2.1843,
      "step": 122194
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.9793990254402161,
      "learning_rate": 0.00027143598872207306,
      "loss": 2.076,
      "step": 122195
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.9300236701965332,
      "learning_rate": 0.0002714319167464425,
      "loss": 2.0607,
      "step": 122196
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.9488978981971741,
      "learning_rate": 0.00027142784477612343,
      "loss": 2.2452,
      "step": 122197
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.0280307531356812,
      "learning_rate": 0.0002714237728111164,
      "loss": 2.1165,
      "step": 122198
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.8902117013931274,
      "learning_rate": 0.0002714197008514222,
      "loss": 2.3076,
      "step": 122199
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.9853471517562866,
      "learning_rate": 0.0002714156288970416,
      "loss": 1.9931,
      "step": 122200
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.2294106483459473,
      "learning_rate": 0.00027141155694797547,
      "loss": 2.2109,
      "step": 122201
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.9178260564804077,
      "learning_rate": 0.00027140748500422443,
      "loss": 2.2789,
      "step": 122202
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.9443539381027222,
      "learning_rate": 0.00027140341306578923,
      "loss": 2.4444,
      "step": 122203
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.0212078094482422,
      "learning_rate": 0.0002713993411326707,
      "loss": 2.0465,
      "step": 122204
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.9409788250923157,
      "learning_rate": 0.0002713952692048696,
      "loss": 2.1573,
      "step": 122205
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.0312108993530273,
      "learning_rate": 0.00027139119728238667,
      "loss": 2.266,
      "step": 122206
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.9361572861671448,
      "learning_rate": 0.00027138712536522265,
      "loss": 1.9483,
      "step": 122207
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.1172654628753662,
      "learning_rate": 0.00027138305345337827,
      "loss": 2.1221,
      "step": 122208
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.9714690446853638,
      "learning_rate": 0.00027137898154685434,
      "loss": 1.8563,
      "step": 122209
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.0067980289459229,
      "learning_rate": 0.00027137490964565156,
      "loss": 2.1346,
      "step": 122210
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.1075574159622192,
      "learning_rate": 0.0002713708377497707,
      "loss": 2.1821,
      "step": 122211
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.9754270911216736,
      "learning_rate": 0.0002713667658592127,
      "loss": 2.1274,
      "step": 122212
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.9993651509284973,
      "learning_rate": 0.00027136269397397804,
      "loss": 2.2799,
      "step": 122213
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.9752679467201233,
      "learning_rate": 0.00027135862209406764,
      "loss": 2.2792,
      "step": 122214
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.9434890151023865,
      "learning_rate": 0.00027135455021948214,
      "loss": 2.3542,
      "step": 122215
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.9523124098777771,
      "learning_rate": 0.0002713504783502224,
      "loss": 2.3183,
      "step": 122216
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.9102458953857422,
      "learning_rate": 0.00027134640648628913,
      "loss": 2.1792,
      "step": 122217
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.1655516624450684,
      "learning_rate": 0.00027134233462768314,
      "loss": 1.9831,
      "step": 122218
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.8974881172180176,
      "learning_rate": 0.00027133826277440525,
      "loss": 1.9513,
      "step": 122219
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.9226018786430359,
      "learning_rate": 0.000271334190926456,
      "loss": 2.1423,
      "step": 122220
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.0328282117843628,
      "learning_rate": 0.00027133011908383624,
      "loss": 2.1524,
      "step": 122221
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.9786896109580994,
      "learning_rate": 0.0002713260472465468,
      "loss": 2.1355,
      "step": 122222
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.9244610071182251,
      "learning_rate": 0.0002713219754145884,
      "loss": 2.0911,
      "step": 122223
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.9225444197654724,
      "learning_rate": 0.0002713179035879617,
      "loss": 1.9679,
      "step": 122224
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.9328587055206299,
      "learning_rate": 0.0002713138317666677,
      "loss": 2.1487,
      "step": 122225
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.9612904191017151,
      "learning_rate": 0.0002713097599507069,
      "loss": 2.0999,
      "step": 122226
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.0305025577545166,
      "learning_rate": 0.00027130568814008013,
      "loss": 2.1423,
      "step": 122227
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.9509448409080505,
      "learning_rate": 0.0002713016163347882,
      "loss": 2.0918,
      "step": 122228
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.080691933631897,
      "learning_rate": 0.00027129754453483187,
      "loss": 2.4048,
      "step": 122229
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.9690598845481873,
      "learning_rate": 0.0002712934727402118,
      "loss": 2.0662,
      "step": 122230
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.9260905385017395,
      "learning_rate": 0.000271289400950929,
      "loss": 2.2309,
      "step": 122231
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.965987503528595,
      "learning_rate": 0.0002712853291669839,
      "loss": 2.073,
      "step": 122232
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.9808993339538574,
      "learning_rate": 0.0002712812573883774,
      "loss": 2.0108,
      "step": 122233
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.0277701616287231,
      "learning_rate": 0.00027127718561511025,
      "loss": 2.0692,
      "step": 122234
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.0650222301483154,
      "learning_rate": 0.0002712731138471832,
      "loss": 2.2101,
      "step": 122235
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.9866177439689636,
      "learning_rate": 0.000271269042084597,
      "loss": 2.071,
      "step": 122236
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.1431384086608887,
      "learning_rate": 0.0002712649703273526,
      "loss": 2.2229,
      "step": 122237
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.0126934051513672,
      "learning_rate": 0.00027126089857545043,
      "loss": 2.2075,
      "step": 122238
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.0623345375061035,
      "learning_rate": 0.0002712568268288914,
      "loss": 2.0269,
      "step": 122239
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.921877384185791,
      "learning_rate": 0.0002712527550876763,
      "loss": 2.1848,
      "step": 122240
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.9743878841400146,
      "learning_rate": 0.00027124868335180583,
      "loss": 1.9391,
      "step": 122241
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.2722376585006714,
      "learning_rate": 0.00027124461162128077,
      "loss": 2.2572,
      "step": 122242
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.9877218008041382,
      "learning_rate": 0.0002712405398961019,
      "loss": 1.9807,
      "step": 122243
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.9792724251747131,
      "learning_rate": 0.0002712364681762701,
      "loss": 2.1113,
      "step": 122244
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.9656291604042053,
      "learning_rate": 0.0002712323964617858,
      "loss": 2.0593,
      "step": 122245
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.0187807083129883,
      "learning_rate": 0.00027122832475265,
      "loss": 2.2072,
      "step": 122246
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.9716576337814331,
      "learning_rate": 0.00027122425304886336,
      "loss": 2.1502,
      "step": 122247
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.1300228834152222,
      "learning_rate": 0.00027122018135042667,
      "loss": 2.044,
      "step": 122248
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.1265864372253418,
      "learning_rate": 0.00027121610965734073,
      "loss": 2.1111,
      "step": 122249
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.9294334650039673,
      "learning_rate": 0.00027121203796960636,
      "loss": 2.1899,
      "step": 122250
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.185849905014038,
      "learning_rate": 0.00027120796628722404,
      "loss": 2.0683,
      "step": 122251
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.9851986169815063,
      "learning_rate": 0.0002712038946101948,
      "loss": 2.0932,
      "step": 122252
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.1822129487991333,
      "learning_rate": 0.00027119982293851927,
      "loss": 2.3724,
      "step": 122253
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.0258101224899292,
      "learning_rate": 0.00027119575127219816,
      "loss": 2.0317,
      "step": 122254
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.9737321734428406,
      "learning_rate": 0.00027119167961123236,
      "loss": 2.143,
      "step": 122255
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.9413591623306274,
      "learning_rate": 0.0002711876079556227,
      "loss": 2.0727,
      "step": 122256
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.2057228088378906,
      "learning_rate": 0.00027118353630536967,
      "loss": 2.1569,
      "step": 122257
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.945016622543335,
      "learning_rate": 0.0002711794646604742,
      "loss": 2.0784,
      "step": 122258
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.8953998684883118,
      "learning_rate": 0.000271175393020937,
      "loss": 2.2315,
      "step": 122259
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.0890288352966309,
      "learning_rate": 0.0002711713213867588,
      "loss": 2.1681,
      "step": 122260
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.9843968749046326,
      "learning_rate": 0.0002711672497579404,
      "loss": 2.0638,
      "step": 122261
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.9684174656867981,
      "learning_rate": 0.00027116317813448274,
      "loss": 2.1339,
      "step": 122262
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.9826326370239258,
      "learning_rate": 0.0002711591065163862,
      "loss": 2.0837,
      "step": 122263
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.0215129852294922,
      "learning_rate": 0.0002711550349036518,
      "loss": 2.1101,
      "step": 122264
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.150742530822754,
      "learning_rate": 0.0002711509632962801,
      "loss": 1.979,
      "step": 122265
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.0823886394500732,
      "learning_rate": 0.00027114689169427207,
      "loss": 2.3831,
      "step": 122266
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.0424402952194214,
      "learning_rate": 0.00027114282009762836,
      "loss": 2.1103,
      "step": 122267
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.9640017151832581,
      "learning_rate": 0.00027113874850634987,
      "loss": 1.9536,
      "step": 122268
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.021273136138916,
      "learning_rate": 0.0002711346769204371,
      "loss": 2.145,
      "step": 122269
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.0757510662078857,
      "learning_rate": 0.0002711306053398909,
      "loss": 1.9712,
      "step": 122270
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.0095683336257935,
      "learning_rate": 0.0002711265337647122,
      "loss": 2.225,
      "step": 122271
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.121044635772705,
      "learning_rate": 0.0002711224621949015,
      "loss": 2.1411,
      "step": 122272
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.1324760913848877,
      "learning_rate": 0.0002711183906304597,
      "loss": 2.0911,
      "step": 122273
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.9628758430480957,
      "learning_rate": 0.0002711143190713876,
      "loss": 1.9796,
      "step": 122274
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.9999744296073914,
      "learning_rate": 0.0002711102475176859,
      "loss": 2.2637,
      "step": 122275
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.0343654155731201,
      "learning_rate": 0.0002711061759693553,
      "loss": 2.0528,
      "step": 122276
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.0580888986587524,
      "learning_rate": 0.0002711021044263966,
      "loss": 2.0893,
      "step": 122277
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.9885962009429932,
      "learning_rate": 0.00027109803288881054,
      "loss": 1.9296,
      "step": 122278
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.0432158708572388,
      "learning_rate": 0.0002710939613565979,
      "loss": 1.8548,
      "step": 122279
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.9819032549858093,
      "learning_rate": 0.0002710898898297596,
      "loss": 1.9244,
      "step": 122280
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.0766783952713013,
      "learning_rate": 0.0002710858183082961,
      "loss": 2.3228,
      "step": 122281
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.0495976209640503,
      "learning_rate": 0.0002710817467922083,
      "loss": 2.2734,
      "step": 122282
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.066138505935669,
      "learning_rate": 0.00027107767528149695,
      "loss": 2.1362,
      "step": 122283
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.9496925473213196,
      "learning_rate": 0.0002710736037761628,
      "loss": 2.0249,
      "step": 122284
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.987409770488739,
      "learning_rate": 0.0002710695322762066,
      "loss": 2.3976,
      "step": 122285
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.0324445962905884,
      "learning_rate": 0.0002710654607816292,
      "loss": 2.0036,
      "step": 122286
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.0567735433578491,
      "learning_rate": 0.0002710613892924312,
      "loss": 2.0536,
      "step": 122287
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.0893200635910034,
      "learning_rate": 0.0002710573178086134,
      "loss": 2.3843,
      "step": 122288
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.147792935371399,
      "learning_rate": 0.0002710532463301767,
      "loss": 2.1975,
      "step": 122289
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.9843934774398804,
      "learning_rate": 0.00027104917485712166,
      "loss": 2.099,
      "step": 122290
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.9005011916160583,
      "learning_rate": 0.00027104510338944916,
      "loss": 2.217,
      "step": 122291
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.063288927078247,
      "learning_rate": 0.00027104103192716,
      "loss": 2.1788,
      "step": 122292
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.9501369595527649,
      "learning_rate": 0.00027103696047025476,
      "loss": 2.2925,
      "step": 122293
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.0201791524887085,
      "learning_rate": 0.00027103288901873427,
      "loss": 1.9101,
      "step": 122294
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.0406885147094727,
      "learning_rate": 0.00027102881757259934,
      "loss": 2.0934,
      "step": 122295
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.8497102856636047,
      "learning_rate": 0.0002710247461318507,
      "loss": 2.2024,
      "step": 122296
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.0445950031280518,
      "learning_rate": 0.0002710206746964892,
      "loss": 2.1484,
      "step": 122297
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.853924572467804,
      "learning_rate": 0.00027101660326651547,
      "loss": 2.0917,
      "step": 122298
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.9557989239692688,
      "learning_rate": 0.0002710125318419302,
      "loss": 2.0786,
      "step": 122299
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.0041143894195557,
      "learning_rate": 0.00027100846042273427,
      "loss": 2.0778,
      "step": 122300
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.9259220361709595,
      "learning_rate": 0.00027100438900892846,
      "loss": 1.9545,
      "step": 122301
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.9295456409454346,
      "learning_rate": 0.00027100031760051346,
      "loss": 2.1727,
      "step": 122302
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.9756056666374207,
      "learning_rate": 0.00027099624619749004,
      "loss": 2.1538,
      "step": 122303
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.1227632761001587,
      "learning_rate": 0.0002709921747998591,
      "loss": 2.1135,
      "step": 122304
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.11542546749115,
      "learning_rate": 0.0002709881034076211,
      "loss": 2.2275,
      "step": 122305
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.9512957334518433,
      "learning_rate": 0.00027098403202077703,
      "loss": 2.1049,
      "step": 122306
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.9095348715782166,
      "learning_rate": 0.0002709799606393275,
      "loss": 1.8137,
      "step": 122307
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.9573608636856079,
      "learning_rate": 0.00027097588926327344,
      "loss": 2.288,
      "step": 122308
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.0061049461364746,
      "learning_rate": 0.0002709718178926154,
      "loss": 2.0561,
      "step": 122309
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.9269834756851196,
      "learning_rate": 0.0002709677465273543,
      "loss": 2.1605,
      "step": 122310
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.9702483415603638,
      "learning_rate": 0.00027096367516749097,
      "loss": 2.2657,
      "step": 122311
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.9963006973266602,
      "learning_rate": 0.0002709596038130259,
      "loss": 1.9934,
      "step": 122312
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.0248054265975952,
      "learning_rate": 0.00027095553246396005,
      "loss": 1.9522,
      "step": 122313
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.8976601362228394,
      "learning_rate": 0.00027095146112029404,
      "loss": 2.0554,
      "step": 122314
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.0657676458358765,
      "learning_rate": 0.0002709473897820287,
      "loss": 2.1833,
      "step": 122315
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.940071702003479,
      "learning_rate": 0.0002709433184491648,
      "loss": 2.0702,
      "step": 122316
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.0094619989395142,
      "learning_rate": 0.0002709392471217033,
      "loss": 2.1279,
      "step": 122317
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.0148173570632935,
      "learning_rate": 0.0002709351757996445,
      "loss": 2.2349,
      "step": 122318
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.9616091847419739,
      "learning_rate": 0.00027093110448298945,
      "loss": 2.0178,
      "step": 122319
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.9564367532730103,
      "learning_rate": 0.00027092703317173886,
      "loss": 2.1242,
      "step": 122320
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.9383518099784851,
      "learning_rate": 0.00027092296186589346,
      "loss": 2.1056,
      "step": 122321
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.9949770569801331,
      "learning_rate": 0.000270918890565454,
      "loss": 1.9165,
      "step": 122322
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.1481605768203735,
      "learning_rate": 0.0002709148192704215,
      "loss": 2.1729,
      "step": 122323
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.0354399681091309,
      "learning_rate": 0.0002709107479807963,
      "loss": 2.1157,
      "step": 122324
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.2525628805160522,
      "learning_rate": 0.0002709066766965793,
      "loss": 2.113,
      "step": 122325
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.064424753189087,
      "learning_rate": 0.0002709026054177713,
      "loss": 2.2861,
      "step": 122326
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.8863261938095093,
      "learning_rate": 0.0002708985341443731,
      "loss": 2.0334,
      "step": 122327
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.0360383987426758,
      "learning_rate": 0.00027089446287638537,
      "loss": 1.8586,
      "step": 122328
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.0748579502105713,
      "learning_rate": 0.0002708903916138091,
      "loss": 2.0686,
      "step": 122329
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.0129215717315674,
      "learning_rate": 0.0002708863203566447,
      "loss": 2.0903,
      "step": 122330
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.8873518705368042,
      "learning_rate": 0.0002708822491048931,
      "loss": 2.0482,
      "step": 122331
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.9762268662452698,
      "learning_rate": 0.000270878177858555,
      "loss": 2.2866,
      "step": 122332
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.9506732821464539,
      "learning_rate": 0.0002708741066176312,
      "loss": 2.2941,
      "step": 122333
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.1985307931900024,
      "learning_rate": 0.00027087003538212246,
      "loss": 2.1091,
      "step": 122334
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.0021406412124634,
      "learning_rate": 0.00027086596415202966,
      "loss": 2.0986,
      "step": 122335
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.9219343066215515,
      "learning_rate": 0.0002708618929273533,
      "loss": 2.1572,
      "step": 122336
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.9143158793449402,
      "learning_rate": 0.00027085782170809425,
      "loss": 2.1229,
      "step": 122337
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.2683502435684204,
      "learning_rate": 0.0002708537504942533,
      "loss": 2.0937,
      "step": 122338
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.1285897493362427,
      "learning_rate": 0.0002708496792858312,
      "loss": 2.0808,
      "step": 122339
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.969241738319397,
      "learning_rate": 0.0002708456080828287,
      "loss": 2.2288,
      "step": 122340
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.238084077835083,
      "learning_rate": 0.0002708415368852467,
      "loss": 2.2015,
      "step": 122341
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.1126573085784912,
      "learning_rate": 0.0002708374656930856,
      "loss": 1.9164,
      "step": 122342
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.1095592975616455,
      "learning_rate": 0.00027083339450634643,
      "loss": 2.2463,
      "step": 122343
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.9467068314552307,
      "learning_rate": 0.00027082932332502983,
      "loss": 1.9929,
      "step": 122344
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.2112573385238647,
      "learning_rate": 0.0002708252521491366,
      "loss": 2.1675,
      "step": 122345
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.2420704364776611,
      "learning_rate": 0.0002708211809786676,
      "loss": 2.1167,
      "step": 122346
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.2273612022399902,
      "learning_rate": 0.00027081710981362355,
      "loss": 2.1446,
      "step": 122347
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.0708134174346924,
      "learning_rate": 0.00027081303865400503,
      "loss": 2.1823,
      "step": 122348
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.1427369117736816,
      "learning_rate": 0.0002708089674998129,
      "loss": 2.2248,
      "step": 122349
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.0279573202133179,
      "learning_rate": 0.000270804896351048,
      "loss": 2.1897,
      "step": 122350
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.4552985429763794,
      "learning_rate": 0.00027080082520771094,
      "loss": 2.177,
      "step": 122351
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.9533904790878296,
      "learning_rate": 0.00027079675406980265,
      "loss": 2.262,
      "step": 122352
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.8685875535011292,
      "learning_rate": 0.00027079268293732383,
      "loss": 2.2202,
      "step": 122353
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.3387572765350342,
      "learning_rate": 0.00027078861181027513,
      "loss": 2.1469,
      "step": 122354
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.0133380889892578,
      "learning_rate": 0.00027078454068865735,
      "loss": 2.2219,
      "step": 122355
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.010603904724121,
      "learning_rate": 0.0002707804695724713,
      "loss": 1.9536,
      "step": 122356
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.9938562512397766,
      "learning_rate": 0.0002707763984617177,
      "loss": 2.4951,
      "step": 122357
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.0606789588928223,
      "learning_rate": 0.0002707723273563973,
      "loss": 2.0445,
      "step": 122358
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.0460269451141357,
      "learning_rate": 0.0002707682562565111,
      "loss": 2.0726,
      "step": 122359
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.9770128726959229,
      "learning_rate": 0.0002707641851620594,
      "loss": 2.1114,
      "step": 122360
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.9694055318832397,
      "learning_rate": 0.00027076011407304316,
      "loss": 2.0264,
      "step": 122361
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.13966703414917,
      "learning_rate": 0.00027075604298946326,
      "loss": 2.1572,
      "step": 122362
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.0864708423614502,
      "learning_rate": 0.0002707519719113203,
      "loss": 2.0596,
      "step": 122363
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.017615556716919,
      "learning_rate": 0.00027074790083861513,
      "loss": 2.3598,
      "step": 122364
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.0734608173370361,
      "learning_rate": 0.0002707438297713486,
      "loss": 2.0977,
      "step": 122365
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.0101388692855835,
      "learning_rate": 0.0002707397587095212,
      "loss": 2.2077,
      "step": 122366
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.9223474264144897,
      "learning_rate": 0.0002707356876531339,
      "loss": 1.923,
      "step": 122367
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.0383950471878052,
      "learning_rate": 0.0002707316166021873,
      "loss": 2.0657,
      "step": 122368
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.9757853746414185,
      "learning_rate": 0.0002707275455566823,
      "loss": 1.964,
      "step": 122369
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.9623063802719116,
      "learning_rate": 0.00027072347451661963,
      "loss": 2.0863,
      "step": 122370
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.0667493343353271,
      "learning_rate": 0.00027071940348200003,
      "loss": 2.15,
      "step": 122371
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.1761139631271362,
      "learning_rate": 0.00027071533245282415,
      "loss": 2.1838,
      "step": 122372
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.0111477375030518,
      "learning_rate": 0.0002707112614290929,
      "loss": 2.1281,
      "step": 122373
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.0513116121292114,
      "learning_rate": 0.00027070719041080704,
      "loss": 2.0621,
      "step": 122374
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.9971461296081543,
      "learning_rate": 0.0002707031193979672,
      "loss": 2.2245,
      "step": 122375
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.0593794584274292,
      "learning_rate": 0.00027069904839057416,
      "loss": 2.062,
      "step": 122376
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.9865798950195312,
      "learning_rate": 0.0002706949773886288,
      "loss": 2.1257,
      "step": 122377
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.0124213695526123,
      "learning_rate": 0.0002706909063921318,
      "loss": 2.1363,
      "step": 122378
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.9990439414978027,
      "learning_rate": 0.00027068683540108383,
      "loss": 2.1822,
      "step": 122379
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.1737568378448486,
      "learning_rate": 0.0002706827644154858,
      "loss": 2.0388,
      "step": 122380
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.9981865286827087,
      "learning_rate": 0.0002706786934353384,
      "loss": 2.0892,
      "step": 122381
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.977418839931488,
      "learning_rate": 0.0002706746224606424,
      "loss": 2.0226,
      "step": 122382
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.1473513841629028,
      "learning_rate": 0.00027067055149139843,
      "loss": 2.157,
      "step": 122383
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.149744987487793,
      "learning_rate": 0.0002706664805276075,
      "loss": 2.1415,
      "step": 122384
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.8548993468284607,
      "learning_rate": 0.0002706624095692701,
      "loss": 1.9759,
      "step": 122385
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.9871096014976501,
      "learning_rate": 0.0002706583386163872,
      "loss": 2.0121,
      "step": 122386
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.9956426024436951,
      "learning_rate": 0.0002706542676689594,
      "loss": 2.1393,
      "step": 122387
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.2502548694610596,
      "learning_rate": 0.0002706501967269876,
      "loss": 2.1068,
      "step": 122388
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.127089262008667,
      "learning_rate": 0.0002706461257904724,
      "loss": 1.8797,
      "step": 122389
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.1070300340652466,
      "learning_rate": 0.0002706420548594148,
      "loss": 2.105,
      "step": 122390
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.2733010053634644,
      "learning_rate": 0.00027063798393381523,
      "loss": 2.1801,
      "step": 122391
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.1522055864334106,
      "learning_rate": 0.00027063391301367464,
      "loss": 1.8386,
      "step": 122392
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.8730883002281189,
      "learning_rate": 0.00027062984209899377,
      "loss": 2.1572,
      "step": 122393
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.0477032661437988,
      "learning_rate": 0.00027062577118977337,
      "loss": 2.3421,
      "step": 122394
    },
    {
      "epoch": 1.59,
      "grad_norm": 2.032346248626709,
      "learning_rate": 0.0002706217002860142,
      "loss": 1.9802,
      "step": 122395
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.0083895921707153,
      "learning_rate": 0.0002706176293877171,
      "loss": 2.221,
      "step": 122396
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.9648445844650269,
      "learning_rate": 0.0002706135584948826,
      "loss": 1.9236,
      "step": 122397
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.9552539587020874,
      "learning_rate": 0.00027060948760751166,
      "loss": 2.0332,
      "step": 122398
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.082959532737732,
      "learning_rate": 0.00027060541672560497,
      "loss": 2.3139,
      "step": 122399
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.2127736806869507,
      "learning_rate": 0.00027060134584916325,
      "loss": 2.1574,
      "step": 122400
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.9649839997291565,
      "learning_rate": 0.00027059727497818727,
      "loss": 2.1938,
      "step": 122401
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.8873776793479919,
      "learning_rate": 0.000270593204112678,
      "loss": 2.106,
      "step": 122402
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.1992782354354858,
      "learning_rate": 0.0002705891332526358,
      "loss": 1.9694,
      "step": 122403
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.062409520149231,
      "learning_rate": 0.00027058506239806176,
      "loss": 1.8534,
      "step": 122404
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.0129879713058472,
      "learning_rate": 0.0002705809915489564,
      "loss": 2.2796,
      "step": 122405
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.8635388016700745,
      "learning_rate": 0.00027057692070532065,
      "loss": 1.9958,
      "step": 122406
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.1066863536834717,
      "learning_rate": 0.0002705728498671552,
      "loss": 2.1137,
      "step": 122407
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.4339838027954102,
      "learning_rate": 0.0002705687790344609,
      "loss": 2.1347,
      "step": 122408
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.1405298709869385,
      "learning_rate": 0.00027056470820723834,
      "loss": 2.0797,
      "step": 122409
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.061659336090088,
      "learning_rate": 0.0002705606373854883,
      "loss": 2.0383,
      "step": 122410
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.1883949041366577,
      "learning_rate": 0.00027055656656921166,
      "loss": 2.3811,
      "step": 122411
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.0882025957107544,
      "learning_rate": 0.00027055249575840906,
      "loss": 2.0649,
      "step": 122412
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.9645904302597046,
      "learning_rate": 0.0002705484249530813,
      "loss": 2.1522,
      "step": 122413
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.063029170036316,
      "learning_rate": 0.0002705443541532293,
      "loss": 2.2235,
      "step": 122414
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.1806060075759888,
      "learning_rate": 0.0002705402833588535,
      "loss": 2.1981,
      "step": 122415
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.004130482673645,
      "learning_rate": 0.0002705362125699549,
      "loss": 2.038,
      "step": 122416
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.9420252442359924,
      "learning_rate": 0.0002705321417865341,
      "loss": 2.1916,
      "step": 122417
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.0568197965621948,
      "learning_rate": 0.00027052807100859193,
      "loss": 2.0356,
      "step": 122418
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.0419270992279053,
      "learning_rate": 0.0002705240002361292,
      "loss": 2.153,
      "step": 122419
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.2154396772384644,
      "learning_rate": 0.0002705199294691467,
      "loss": 2.0032,
      "step": 122420
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.9450543522834778,
      "learning_rate": 0.000270515858707645,
      "loss": 2.1306,
      "step": 122421
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.9642766714096069,
      "learning_rate": 0.0002705117879516249,
      "loss": 2.0723,
      "step": 122422
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.0283317565917969,
      "learning_rate": 0.0002705077172010873,
      "loss": 2.1693,
      "step": 122423
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.8693439364433289,
      "learning_rate": 0.0002705036464560328,
      "loss": 1.9902,
      "step": 122424
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.0558186769485474,
      "learning_rate": 0.0002704995757164623,
      "loss": 2.3579,
      "step": 122425
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.157724142074585,
      "learning_rate": 0.0002704955049823766,
      "loss": 2.116,
      "step": 122426
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.9797645807266235,
      "learning_rate": 0.00027049143425377615,
      "loss": 2.0799,
      "step": 122427
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.0196800231933594,
      "learning_rate": 0.0002704873635306619,
      "loss": 2.0397,
      "step": 122428
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.8788241147994995,
      "learning_rate": 0.0002704832928130347,
      "loss": 2.0574,
      "step": 122429
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.1900426149368286,
      "learning_rate": 0.00027047922210089514,
      "loss": 2.3436,
      "step": 122430
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.9617771506309509,
      "learning_rate": 0.0002704751513942441,
      "loss": 1.7502,
      "step": 122431
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.8558546900749207,
      "learning_rate": 0.00027047108069308236,
      "loss": 2.0071,
      "step": 122432
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.0148202180862427,
      "learning_rate": 0.0002704670099974105,
      "loss": 1.9585,
      "step": 122433
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.0146450996398926,
      "learning_rate": 0.0002704629393072294,
      "loss": 2.3072,
      "step": 122434
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.971825361251831,
      "learning_rate": 0.00027045886862253973,
      "loss": 2.1265,
      "step": 122435
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.9561879634857178,
      "learning_rate": 0.00027045479794334236,
      "loss": 1.9329,
      "step": 122436
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.0136163234710693,
      "learning_rate": 0.000270450727269638,
      "loss": 2.3365,
      "step": 122437
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.9793279767036438,
      "learning_rate": 0.00027044665660142756,
      "loss": 2.1979,
      "step": 122438
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.9715707898139954,
      "learning_rate": 0.0002704425859387115,
      "loss": 2.0731,
      "step": 122439
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.973244309425354,
      "learning_rate": 0.0002704385152814907,
      "loss": 2.0995,
      "step": 122440
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.9477288722991943,
      "learning_rate": 0.00027043444462976595,
      "loss": 1.9915,
      "step": 122441
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.9206424355506897,
      "learning_rate": 0.000270430373983538,
      "loss": 2.0586,
      "step": 122442
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.0209894180297852,
      "learning_rate": 0.00027042630334280757,
      "loss": 2.2534,
      "step": 122443
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.9329423308372498,
      "learning_rate": 0.00027042223270757547,
      "loss": 2.2309,
      "step": 122444
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.9838401675224304,
      "learning_rate": 0.00027041816207784255,
      "loss": 2.0707,
      "step": 122445
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.8623608946800232,
      "learning_rate": 0.0002704140914536094,
      "loss": 2.0849,
      "step": 122446
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.882076621055603,
      "learning_rate": 0.00027041002083487675,
      "loss": 2.2214,
      "step": 122447
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.0022119283676147,
      "learning_rate": 0.00027040595022164544,
      "loss": 1.9456,
      "step": 122448
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.925654411315918,
      "learning_rate": 0.00027040187961391624,
      "loss": 2.1528,
      "step": 122449
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.0068423748016357,
      "learning_rate": 0.0002703978090116899,
      "loss": 2.1793,
      "step": 122450
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.9688037633895874,
      "learning_rate": 0.0002703937384149673,
      "loss": 2.0294,
      "step": 122451
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.0268408060073853,
      "learning_rate": 0.0002703896678237489,
      "loss": 2.0472,
      "step": 122452
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.0068215131759644,
      "learning_rate": 0.00027038559723803566,
      "loss": 2.2474,
      "step": 122453
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.911395788192749,
      "learning_rate": 0.0002703815266578283,
      "loss": 2.295,
      "step": 122454
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.0952321290969849,
      "learning_rate": 0.0002703774560831275,
      "loss": 2.1608,
      "step": 122455
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.0283883810043335,
      "learning_rate": 0.0002703733855139341,
      "loss": 1.9956,
      "step": 122456
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.4546314477920532,
      "learning_rate": 0.000270369314950249,
      "loss": 2.2007,
      "step": 122457
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.9709479212760925,
      "learning_rate": 0.00027036524439207266,
      "loss": 2.1284,
      "step": 122458
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.156506061553955,
      "learning_rate": 0.0002703611738394061,
      "loss": 2.0348,
      "step": 122459
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.0332592725753784,
      "learning_rate": 0.00027035710329224985,
      "loss": 2.0284,
      "step": 122460
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.011785626411438,
      "learning_rate": 0.0002703530327506048,
      "loss": 2.2588,
      "step": 122461
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.0063607692718506,
      "learning_rate": 0.0002703489622144717,
      "loss": 2.2363,
      "step": 122462
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.9144405126571655,
      "learning_rate": 0.0002703448916838513,
      "loss": 2.0639,
      "step": 122463
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.9217696189880371,
      "learning_rate": 0.0002703408211587443,
      "loss": 2.0127,
      "step": 122464
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.0888276100158691,
      "learning_rate": 0.0002703367506391515,
      "loss": 1.9629,
      "step": 122465
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.0427303314208984,
      "learning_rate": 0.00027033268012507375,
      "loss": 2.0644,
      "step": 122466
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.9543267488479614,
      "learning_rate": 0.0002703286096165116,
      "loss": 2.1663,
      "step": 122467
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.9547055959701538,
      "learning_rate": 0.00027032453911346596,
      "loss": 2.274,
      "step": 122468
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.933853805065155,
      "learning_rate": 0.0002703204686159376,
      "loss": 2.188,
      "step": 122469
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.9621960520744324,
      "learning_rate": 0.00027031639812392715,
      "loss": 2.1617,
      "step": 122470
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.957118809223175,
      "learning_rate": 0.00027031232763743545,
      "loss": 2.1067,
      "step": 122471
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.0279510021209717,
      "learning_rate": 0.00027030825715646325,
      "loss": 2.1198,
      "step": 122472
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.9989058375358582,
      "learning_rate": 0.0002703041866810114,
      "loss": 2.1476,
      "step": 122473
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.0028111934661865,
      "learning_rate": 0.00027030011621108043,
      "loss": 2.223,
      "step": 122474
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.4165743589401245,
      "learning_rate": 0.0002702960457466713,
      "loss": 2.0152,
      "step": 122475
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.1700942516326904,
      "learning_rate": 0.0002702919752877847,
      "loss": 2.0782,
      "step": 122476
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.927080512046814,
      "learning_rate": 0.00027028790483442133,
      "loss": 1.936,
      "step": 122477
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.1070873737335205,
      "learning_rate": 0.00027028383438658203,
      "loss": 2.4491,
      "step": 122478
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.118564248085022,
      "learning_rate": 0.00027027976394426755,
      "loss": 2.0879,
      "step": 122479
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.13457190990448,
      "learning_rate": 0.00027027569350747865,
      "loss": 2.303,
      "step": 122480
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.0505566596984863,
      "learning_rate": 0.00027027162307621604,
      "loss": 2.2117,
      "step": 122481
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.0326731204986572,
      "learning_rate": 0.0002702675526504804,
      "loss": 2.184,
      "step": 122482
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.0407005548477173,
      "learning_rate": 0.00027026348223027267,
      "loss": 2.1398,
      "step": 122483
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.083115816116333,
      "learning_rate": 0.0002702594118155935,
      "loss": 2.1695,
      "step": 122484
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.9219854474067688,
      "learning_rate": 0.00027025534140644365,
      "loss": 2.0324,
      "step": 122485
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.103336215019226,
      "learning_rate": 0.0002702512710028239,
      "loss": 2.3368,
      "step": 122486
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.2990714311599731,
      "learning_rate": 0.0002702472006047351,
      "loss": 2.2948,
      "step": 122487
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.9162372350692749,
      "learning_rate": 0.0002702431302121778,
      "loss": 2.1415,
      "step": 122488
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.014726161956787,
      "learning_rate": 0.0002702390598251529,
      "loss": 2.1536,
      "step": 122489
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.2782347202301025,
      "learning_rate": 0.00027023498944366107,
      "loss": 1.8736,
      "step": 122490
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.4229289293289185,
      "learning_rate": 0.0002702309190677031,
      "loss": 2.0521,
      "step": 122491
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.9419057369232178,
      "learning_rate": 0.00027022684869727986,
      "loss": 2.3424,
      "step": 122492
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.8782398700714111,
      "learning_rate": 0.00027022277833239206,
      "loss": 2.1398,
      "step": 122493
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.999499499797821,
      "learning_rate": 0.0002702187079730403,
      "loss": 2.0312,
      "step": 122494
    },
    {
      "epoch": 1.59,
      "grad_norm": 1.011979103088379,
      "learning_rate": 0.00027021463761922544,
      "loss": 2.0502,
      "step": 122495
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.9633892774581909,
      "learning_rate": 0.00027021056727094826,
      "loss": 2.2649,
      "step": 122496
    },
    {
      "epoch": 1.59,
      "grad_norm": 0.9172612428665161,
      "learning_rate": 0.00027020649692820946,
      "loss": 1.9868,
      "step": 122497
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.9112167954444885,
      "learning_rate": 0.00027020242659100984,
      "loss": 2.3759,
      "step": 122498
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.9431187510490417,
      "learning_rate": 0.0002701983562593503,
      "loss": 2.1205,
      "step": 122499
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.1603801250457764,
      "learning_rate": 0.0002701942859332313,
      "loss": 2.2068,
      "step": 122500
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.0429295301437378,
      "learning_rate": 0.0002701902156126538,
      "loss": 2.1988,
      "step": 122501
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.9773720502853394,
      "learning_rate": 0.00027018614529761845,
      "loss": 2.2917,
      "step": 122502
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.080385684967041,
      "learning_rate": 0.00027018207498812606,
      "loss": 1.9901,
      "step": 122503
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.0304179191589355,
      "learning_rate": 0.0002701780046841774,
      "loss": 2.0001,
      "step": 122504
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.0608896017074585,
      "learning_rate": 0.00027017393438577336,
      "loss": 2.2995,
      "step": 122505
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.9771380424499512,
      "learning_rate": 0.00027016986409291436,
      "loss": 2.3309,
      "step": 122506
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.9469597935676575,
      "learning_rate": 0.0002701657938056014,
      "loss": 2.0586,
      "step": 122507
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.1247334480285645,
      "learning_rate": 0.00027016172352383513,
      "loss": 2.1569,
      "step": 122508
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.9636358022689819,
      "learning_rate": 0.0002701576532476164,
      "loss": 2.0786,
      "step": 122509
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.9518246054649353,
      "learning_rate": 0.00027015358297694593,
      "loss": 2.3469,
      "step": 122510
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.290545105934143,
      "learning_rate": 0.00027014951271182446,
      "loss": 2.2058,
      "step": 122511
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.0438796281814575,
      "learning_rate": 0.00027014544245225285,
      "loss": 2.2176,
      "step": 122512
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.0228121280670166,
      "learning_rate": 0.0002701413721982317,
      "loss": 2.5674,
      "step": 122513
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.956747829914093,
      "learning_rate": 0.0002701373019497618,
      "loss": 2.2674,
      "step": 122514
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.0189365148544312,
      "learning_rate": 0.00027013323170684396,
      "loss": 2.3674,
      "step": 122515
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.9829550981521606,
      "learning_rate": 0.0002701291614694789,
      "loss": 2.1255,
      "step": 122516
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.0908524990081787,
      "learning_rate": 0.0002701250912376674,
      "loss": 2.0367,
      "step": 122517
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.9415832161903381,
      "learning_rate": 0.00027012102101141034,
      "loss": 2.0412,
      "step": 122518
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.9917164444923401,
      "learning_rate": 0.0002701169507907082,
      "loss": 2.1705,
      "step": 122519
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.107924461364746,
      "learning_rate": 0.0002701128805755619,
      "loss": 2.0447,
      "step": 122520
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.0166653394699097,
      "learning_rate": 0.00027010881036597216,
      "loss": 2.2896,
      "step": 122521
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.9319902658462524,
      "learning_rate": 0.0002701047401619398,
      "loss": 2.1464,
      "step": 122522
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.9540359377861023,
      "learning_rate": 0.00027010066996346546,
      "loss": 2.0236,
      "step": 122523
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.0865501165390015,
      "learning_rate": 0.0002700965997705501,
      "loss": 2.3298,
      "step": 122524
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.0728387832641602,
      "learning_rate": 0.00027009252958319434,
      "loss": 2.4216,
      "step": 122525
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.8773629665374756,
      "learning_rate": 0.00027008845940139885,
      "loss": 1.9627,
      "step": 122526
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.9124659895896912,
      "learning_rate": 0.00027008438922516444,
      "loss": 2.1285,
      "step": 122527
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.8909047842025757,
      "learning_rate": 0.000270080319054492,
      "loss": 1.9658,
      "step": 122528
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.9291078448295593,
      "learning_rate": 0.00027007624888938215,
      "loss": 2.1469,
      "step": 122529
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.1017835140228271,
      "learning_rate": 0.00027007217872983583,
      "loss": 1.9062,
      "step": 122530
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.1459589004516602,
      "learning_rate": 0.0002700681085758535,
      "loss": 2.133,
      "step": 122531
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.9710782766342163,
      "learning_rate": 0.00027006403842743613,
      "loss": 2.074,
      "step": 122532
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.9598084092140198,
      "learning_rate": 0.00027005996828458436,
      "loss": 2.1306,
      "step": 122533
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.013892412185669,
      "learning_rate": 0.000270055898147299,
      "loss": 2.0851,
      "step": 122534
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.9476802349090576,
      "learning_rate": 0.00027005182801558084,
      "loss": 2.1097,
      "step": 122535
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.9771371483802795,
      "learning_rate": 0.0002700477578894308,
      "loss": 2.0814,
      "step": 122536
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.0024175643920898,
      "learning_rate": 0.0002700436877688492,
      "loss": 2.1261,
      "step": 122537
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.9456407427787781,
      "learning_rate": 0.00027003961765383717,
      "loss": 2.1199,
      "step": 122538
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.8944056630134583,
      "learning_rate": 0.00027003554754439525,
      "loss": 1.9865,
      "step": 122539
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.9229474067687988,
      "learning_rate": 0.0002700314774405243,
      "loss": 2.2052,
      "step": 122540
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.0399166345596313,
      "learning_rate": 0.0002700274073422251,
      "loss": 2.4499,
      "step": 122541
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.933718740940094,
      "learning_rate": 0.0002700233372494984,
      "loss": 2.2005,
      "step": 122542
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.9729260802268982,
      "learning_rate": 0.0002700192671623449,
      "loss": 2.1813,
      "step": 122543
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.9970707893371582,
      "learning_rate": 0.0002700151970807654,
      "loss": 2.1399,
      "step": 122544
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.9421190619468689,
      "learning_rate": 0.0002700111270047606,
      "loss": 2.1821,
      "step": 122545
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.0999908447265625,
      "learning_rate": 0.0002700070569343313,
      "loss": 2.0178,
      "step": 122546
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.0188075304031372,
      "learning_rate": 0.0002700029868694782,
      "loss": 2.217,
      "step": 122547
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.0490309000015259,
      "learning_rate": 0.00026999891681020227,
      "loss": 1.777,
      "step": 122548
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.9317687749862671,
      "learning_rate": 0.000269994846756504,
      "loss": 2.1453,
      "step": 122549
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.10104501247406,
      "learning_rate": 0.00026999077670838425,
      "loss": 2.3844,
      "step": 122550
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.156432867050171,
      "learning_rate": 0.00026998670666584384,
      "loss": 2.1352,
      "step": 122551
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.0778814554214478,
      "learning_rate": 0.00026998263662888344,
      "loss": 2.0155,
      "step": 122552
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.9735353589057922,
      "learning_rate": 0.00026997856659750376,
      "loss": 2.0458,
      "step": 122553
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.9832674264907837,
      "learning_rate": 0.0002699744965717058,
      "loss": 1.9857,
      "step": 122554
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.9959044456481934,
      "learning_rate": 0.00026997042655149,
      "loss": 2.1285,
      "step": 122555
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.9235818386077881,
      "learning_rate": 0.0002699663565368573,
      "loss": 2.2487,
      "step": 122556
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.9203094840049744,
      "learning_rate": 0.0002699622865278084,
      "loss": 2.0992,
      "step": 122557
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.9722339510917664,
      "learning_rate": 0.0002699582165243442,
      "loss": 2.2167,
      "step": 122558
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.9822764992713928,
      "learning_rate": 0.0002699541465264652,
      "loss": 2.1046,
      "step": 122559
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.8852530717849731,
      "learning_rate": 0.00026995007653417237,
      "loss": 1.9245,
      "step": 122560
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.0837815999984741,
      "learning_rate": 0.00026994600654746637,
      "loss": 2.219,
      "step": 122561
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.1103535890579224,
      "learning_rate": 0.0002699419365663479,
      "loss": 2.0005,
      "step": 122562
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.102474570274353,
      "learning_rate": 0.00026993786659081786,
      "loss": 2.2173,
      "step": 122563
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.9864027500152588,
      "learning_rate": 0.0002699337966208769,
      "loss": 2.0038,
      "step": 122564
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.9865323305130005,
      "learning_rate": 0.0002699297266565259,
      "loss": 1.9631,
      "step": 122565
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.1790151596069336,
      "learning_rate": 0.0002699256566977655,
      "loss": 2.0113,
      "step": 122566
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.9713009595870972,
      "learning_rate": 0.00026992158674459644,
      "loss": 2.1888,
      "step": 122567
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.0184502601623535,
      "learning_rate": 0.00026991751679701956,
      "loss": 2.171,
      "step": 122568
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.028821587562561,
      "learning_rate": 0.0002699134468550355,
      "loss": 2.1632,
      "step": 122569
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.9972102642059326,
      "learning_rate": 0.0002699093769186452,
      "loss": 2.0117,
      "step": 122570
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.9667295813560486,
      "learning_rate": 0.00026990530698784923,
      "loss": 2.2335,
      "step": 122571
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.936267614364624,
      "learning_rate": 0.0002699012370626486,
      "loss": 2.2362,
      "step": 122572
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.8944234848022461,
      "learning_rate": 0.00026989716714304377,
      "loss": 1.8654,
      "step": 122573
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.9862462282180786,
      "learning_rate": 0.0002698930972290356,
      "loss": 2.27,
      "step": 122574
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.9600483775138855,
      "learning_rate": 0.0002698890273206249,
      "loss": 2.1596,
      "step": 122575
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.998409628868103,
      "learning_rate": 0.00026988495741781237,
      "loss": 1.9893,
      "step": 122576
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.029418706893921,
      "learning_rate": 0.00026988088752059885,
      "loss": 2.3295,
      "step": 122577
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.9633758664131165,
      "learning_rate": 0.000269876817628985,
      "loss": 2.2198,
      "step": 122578
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.0765060186386108,
      "learning_rate": 0.00026987274774297176,
      "loss": 2.1371,
      "step": 122579
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.9423394799232483,
      "learning_rate": 0.0002698686778625596,
      "loss": 2.1475,
      "step": 122580
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.0359774827957153,
      "learning_rate": 0.0002698646079877494,
      "loss": 2.1096,
      "step": 122581
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.003926396369934,
      "learning_rate": 0.000269860538118542,
      "loss": 2.0195,
      "step": 122582
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.9595322012901306,
      "learning_rate": 0.00026985646825493804,
      "loss": 2.0182,
      "step": 122583
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.043705701828003,
      "learning_rate": 0.00026985239839693833,
      "loss": 1.9903,
      "step": 122584
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.9757322072982788,
      "learning_rate": 0.00026984832854454383,
      "loss": 2.0711,
      "step": 122585
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.9753391742706299,
      "learning_rate": 0.00026984425869775494,
      "loss": 2.283,
      "step": 122586
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.0395498275756836,
      "learning_rate": 0.0002698401888565725,
      "loss": 2.0638,
      "step": 122587
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.9428731203079224,
      "learning_rate": 0.0002698361190209974,
      "loss": 2.1917,
      "step": 122588
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.9048799872398376,
      "learning_rate": 0.0002698320491910303,
      "loss": 2.0686,
      "step": 122589
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.9246034026145935,
      "learning_rate": 0.000269827979366672,
      "loss": 2.1349,
      "step": 122590
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.9282336235046387,
      "learning_rate": 0.00026982390954792345,
      "loss": 2.0644,
      "step": 122591
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.9871606826782227,
      "learning_rate": 0.000269819839734785,
      "loss": 2.1544,
      "step": 122592
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.0032627582550049,
      "learning_rate": 0.0002698157699272576,
      "loss": 2.0966,
      "step": 122593
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.9394667148590088,
      "learning_rate": 0.00026981170012534204,
      "loss": 1.8896,
      "step": 122594
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.9732042551040649,
      "learning_rate": 0.0002698076303290391,
      "loss": 2.0599,
      "step": 122595
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.9674196243286133,
      "learning_rate": 0.00026980356053834945,
      "loss": 2.4197,
      "step": 122596
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.980910062789917,
      "learning_rate": 0.000269799490753274,
      "loss": 2.2802,
      "step": 122597
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.0597913265228271,
      "learning_rate": 0.0002697954209738133,
      "loss": 2.349,
      "step": 122598
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.0158584117889404,
      "learning_rate": 0.00026979135119996816,
      "loss": 2.0847,
      "step": 122599
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.9049630165100098,
      "learning_rate": 0.0002697872814317394,
      "loss": 2.0577,
      "step": 122600
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.1535394191741943,
      "learning_rate": 0.00026978321166912776,
      "loss": 2.3764,
      "step": 122601
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.095380425453186,
      "learning_rate": 0.000269779141912134,
      "loss": 2.2525,
      "step": 122602
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.04047429561615,
      "learning_rate": 0.00026977507216075896,
      "loss": 2.1056,
      "step": 122603
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.1620464324951172,
      "learning_rate": 0.00026977100241500323,
      "loss": 2.1133,
      "step": 122604
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.2340277433395386,
      "learning_rate": 0.00026976693267486763,
      "loss": 2.1593,
      "step": 122605
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.212209701538086,
      "learning_rate": 0.0002697628629403529,
      "loss": 2.2351,
      "step": 122606
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.9651933312416077,
      "learning_rate": 0.00026975879321145983,
      "loss": 2.0006,
      "step": 122607
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.063549518585205,
      "learning_rate": 0.00026975472348818915,
      "loss": 2.1762,
      "step": 122608
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.9754322171211243,
      "learning_rate": 0.0002697506537705418,
      "loss": 2.1851,
      "step": 122609
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.360813856124878,
      "learning_rate": 0.00026974658405851825,
      "loss": 2.0301,
      "step": 122610
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.9379033446311951,
      "learning_rate": 0.0002697425143521194,
      "loss": 2.1514,
      "step": 122611
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.95511794090271,
      "learning_rate": 0.00026973844465134596,
      "loss": 2.3088,
      "step": 122612
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.0233502388000488,
      "learning_rate": 0.0002697343749561987,
      "loss": 2.0173,
      "step": 122613
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.0456651449203491,
      "learning_rate": 0.0002697303052666784,
      "loss": 2.1115,
      "step": 122614
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.9847688674926758,
      "learning_rate": 0.00026972623558278594,
      "loss": 2.4042,
      "step": 122615
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.0777597427368164,
      "learning_rate": 0.0002697221659045218,
      "loss": 2.1381,
      "step": 122616
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.3504999876022339,
      "learning_rate": 0.0002697180962318869,
      "loss": 2.3297,
      "step": 122617
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.0536073446273804,
      "learning_rate": 0.00026971402656488197,
      "loss": 1.7972,
      "step": 122618
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.0102888345718384,
      "learning_rate": 0.0002697099569035078,
      "loss": 2.2023,
      "step": 122619
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.0314079523086548,
      "learning_rate": 0.0002697058872477651,
      "loss": 1.9698,
      "step": 122620
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.8787568211555481,
      "learning_rate": 0.0002697018175976548,
      "loss": 1.7916,
      "step": 122621
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.1106836795806885,
      "learning_rate": 0.0002696977479531773,
      "loss": 2.2661,
      "step": 122622
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.9426717758178711,
      "learning_rate": 0.0002696936783143336,
      "loss": 2.1311,
      "step": 122623
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.2752430438995361,
      "learning_rate": 0.0002696896086811244,
      "loss": 2.1847,
      "step": 122624
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.9117299318313599,
      "learning_rate": 0.0002696855390535505,
      "loss": 1.8433,
      "step": 122625
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.9104068875312805,
      "learning_rate": 0.0002696814694316127,
      "loss": 2.0507,
      "step": 122626
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.231827974319458,
      "learning_rate": 0.0002696773998153117,
      "loss": 1.96,
      "step": 122627
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.9686477780342102,
      "learning_rate": 0.0002696733302046482,
      "loss": 2.014,
      "step": 122628
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.8669477105140686,
      "learning_rate": 0.000269669260599623,
      "loss": 2.089,
      "step": 122629
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.9501388072967529,
      "learning_rate": 0.00026966519100023677,
      "loss": 2.2165,
      "step": 122630
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.4994306564331055,
      "learning_rate": 0.0002696611214064904,
      "loss": 2.1176,
      "step": 122631
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.061226487159729,
      "learning_rate": 0.0002696570518183846,
      "loss": 2.2226,
      "step": 122632
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.92840975522995,
      "learning_rate": 0.0002696529822359202,
      "loss": 2.0686,
      "step": 122633
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.9889472126960754,
      "learning_rate": 0.00026964891265909786,
      "loss": 2.2918,
      "step": 122634
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.9569587707519531,
      "learning_rate": 0.00026964484308791836,
      "loss": 2.0443,
      "step": 122635
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.989635705947876,
      "learning_rate": 0.0002696407735223824,
      "loss": 1.9388,
      "step": 122636
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.8942576050758362,
      "learning_rate": 0.0002696367039624908,
      "loss": 1.8932,
      "step": 122637
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.0692660808563232,
      "learning_rate": 0.0002696326344082443,
      "loss": 2.1741,
      "step": 122638
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.0187371969223022,
      "learning_rate": 0.00026962856485964376,
      "loss": 2.0451,
      "step": 122639
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.9806415438652039,
      "learning_rate": 0.0002696244953166898,
      "loss": 2.2287,
      "step": 122640
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.2721102237701416,
      "learning_rate": 0.00026962042577938314,
      "loss": 2.0198,
      "step": 122641
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.9471891522407532,
      "learning_rate": 0.00026961635624772476,
      "loss": 2.0233,
      "step": 122642
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.9522738456726074,
      "learning_rate": 0.0002696122867217152,
      "loss": 2.176,
      "step": 122643
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.043578863143921,
      "learning_rate": 0.0002696082172013552,
      "loss": 2.072,
      "step": 122644
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.0175220966339111,
      "learning_rate": 0.0002696041476866457,
      "loss": 2.2586,
      "step": 122645
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.9813681244850159,
      "learning_rate": 0.0002696000781775874,
      "loss": 2.0749,
      "step": 122646
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.0140513181686401,
      "learning_rate": 0.00026959600867418093,
      "loss": 2.1581,
      "step": 122647
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.973416268825531,
      "learning_rate": 0.0002695919391764272,
      "loss": 2.0804,
      "step": 122648
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.9930307269096375,
      "learning_rate": 0.00026958786968432684,
      "loss": 2.2008,
      "step": 122649
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.0495803356170654,
      "learning_rate": 0.0002695838001978807,
      "loss": 2.2038,
      "step": 122650
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.917093813419342,
      "learning_rate": 0.0002695797307170895,
      "loss": 2.0672,
      "step": 122651
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.9262679815292358,
      "learning_rate": 0.0002695756612419541,
      "loss": 2.1863,
      "step": 122652
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.934523344039917,
      "learning_rate": 0.00026957159177247507,
      "loss": 2.2264,
      "step": 122653
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.8945619463920593,
      "learning_rate": 0.0002695675223086532,
      "loss": 2.0405,
      "step": 122654
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.9573723077774048,
      "learning_rate": 0.0002695634528504894,
      "loss": 2.0878,
      "step": 122655
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.9680452346801758,
      "learning_rate": 0.0002695593833979842,
      "loss": 1.9516,
      "step": 122656
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.9333076477050781,
      "learning_rate": 0.0002695553139511387,
      "loss": 2.1054,
      "step": 122657
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.995732307434082,
      "learning_rate": 0.00026955124450995334,
      "loss": 2.3105,
      "step": 122658
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.9436236619949341,
      "learning_rate": 0.00026954717507442894,
      "loss": 2.2432,
      "step": 122659
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.9692716598510742,
      "learning_rate": 0.0002695431056445663,
      "loss": 2.3101,
      "step": 122660
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.0478999614715576,
      "learning_rate": 0.0002695390362203661,
      "loss": 2.2069,
      "step": 122661
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.9763672947883606,
      "learning_rate": 0.0002695349668018293,
      "loss": 1.9381,
      "step": 122662
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.1261892318725586,
      "learning_rate": 0.0002695308973889564,
      "loss": 2.0511,
      "step": 122663
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.0030856132507324,
      "learning_rate": 0.0002695268279817485,
      "loss": 1.9358,
      "step": 122664
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.0209699869155884,
      "learning_rate": 0.000269522758580206,
      "loss": 2.3716,
      "step": 122665
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.9742724299430847,
      "learning_rate": 0.0002695186891843297,
      "loss": 2.1434,
      "step": 122666
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.8817744851112366,
      "learning_rate": 0.00026951461979412054,
      "loss": 2.1598,
      "step": 122667
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.9224889874458313,
      "learning_rate": 0.0002695105504095792,
      "loss": 2.3445,
      "step": 122668
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.991699755191803,
      "learning_rate": 0.0002695064810307064,
      "loss": 2.1691,
      "step": 122669
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.9813233613967896,
      "learning_rate": 0.00026950241165750303,
      "loss": 1.9495,
      "step": 122670
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.0616692304611206,
      "learning_rate": 0.0002694983422899696,
      "loss": 2.1024,
      "step": 122671
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.0663098096847534,
      "learning_rate": 0.00026949427292810704,
      "loss": 2.0991,
      "step": 122672
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.0116719007492065,
      "learning_rate": 0.00026949020357191605,
      "loss": 2.1796,
      "step": 122673
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.9463370442390442,
      "learning_rate": 0.0002694861342213974,
      "loss": 2.3038,
      "step": 122674
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.9589999318122864,
      "learning_rate": 0.00026948206487655186,
      "loss": 2.1378,
      "step": 122675
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.0153065919876099,
      "learning_rate": 0.00026947799553738035,
      "loss": 2.1215,
      "step": 122676
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.8649131655693054,
      "learning_rate": 0.0002694739262038833,
      "loss": 1.8922,
      "step": 122677
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.1865445375442505,
      "learning_rate": 0.00026946985687606163,
      "loss": 2.1335,
      "step": 122678
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.0352916717529297,
      "learning_rate": 0.00026946578755391605,
      "loss": 2.165,
      "step": 122679
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.9850540161132812,
      "learning_rate": 0.0002694617182374474,
      "loss": 2.1927,
      "step": 122680
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.9696260690689087,
      "learning_rate": 0.00026945764892665635,
      "loss": 2.1619,
      "step": 122681
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.9984898567199707,
      "learning_rate": 0.00026945357962154386,
      "loss": 1.9554,
      "step": 122682
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.8940048217773438,
      "learning_rate": 0.0002694495103221104,
      "loss": 2.0461,
      "step": 122683
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.8986231088638306,
      "learning_rate": 0.00026944544102835684,
      "loss": 2.256,
      "step": 122684
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.15986967086792,
      "learning_rate": 0.000269441371740284,
      "loss": 2.1641,
      "step": 122685
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.5170148611068726,
      "learning_rate": 0.0002694373024578925,
      "loss": 2.14,
      "step": 122686
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.2016762495040894,
      "learning_rate": 0.0002694332331811832,
      "loss": 2.1497,
      "step": 122687
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.0484868288040161,
      "learning_rate": 0.000269429163910157,
      "loss": 2.2488,
      "step": 122688
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.0148544311523438,
      "learning_rate": 0.00026942509464481434,
      "loss": 2.0506,
      "step": 122689
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.0141384601593018,
      "learning_rate": 0.00026942102538515616,
      "loss": 2.1488,
      "step": 122690
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.9249680042266846,
      "learning_rate": 0.0002694169561311832,
      "loss": 2.2846,
      "step": 122691
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.0201529264450073,
      "learning_rate": 0.0002694128868828962,
      "loss": 2.2435,
      "step": 122692
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.1808538436889648,
      "learning_rate": 0.00026940881764029586,
      "loss": 2.1536,
      "step": 122693
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.2485694885253906,
      "learning_rate": 0.00026940474840338315,
      "loss": 2.1018,
      "step": 122694
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.9198364615440369,
      "learning_rate": 0.0002694006791721586,
      "loss": 2.1706,
      "step": 122695
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.9562869668006897,
      "learning_rate": 0.000269396609946623,
      "loss": 2.0854,
      "step": 122696
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.9690122008323669,
      "learning_rate": 0.0002693925407267772,
      "loss": 2.1241,
      "step": 122697
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.007959008216858,
      "learning_rate": 0.0002693884715126218,
      "loss": 1.9454,
      "step": 122698
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.1171120405197144,
      "learning_rate": 0.00026938440230415777,
      "loss": 2.0778,
      "step": 122699
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.0539162158966064,
      "learning_rate": 0.00026938033310138583,
      "loss": 2.2592,
      "step": 122700
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.971933126449585,
      "learning_rate": 0.00026937626390430655,
      "loss": 2.2287,
      "step": 122701
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.9888935089111328,
      "learning_rate": 0.0002693721947129208,
      "loss": 1.9721,
      "step": 122702
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.0046783685684204,
      "learning_rate": 0.0002693681255272293,
      "loss": 1.951,
      "step": 122703
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.9504693746566772,
      "learning_rate": 0.0002693640563472329,
      "loss": 2.1016,
      "step": 122704
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.1308581829071045,
      "learning_rate": 0.00026935998717293226,
      "loss": 2.1815,
      "step": 122705
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.0504887104034424,
      "learning_rate": 0.00026935591800432836,
      "loss": 2.174,
      "step": 122706
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.0016827583312988,
      "learning_rate": 0.00026935184884142155,
      "loss": 2.1322,
      "step": 122707
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.9749770760536194,
      "learning_rate": 0.0002693477796842129,
      "loss": 1.9076,
      "step": 122708
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.0999408960342407,
      "learning_rate": 0.000269343710532703,
      "loss": 2.1262,
      "step": 122709
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.9465012550354004,
      "learning_rate": 0.00026933964138689277,
      "loss": 1.9433,
      "step": 122710
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.0161619186401367,
      "learning_rate": 0.00026933557224678283,
      "loss": 2.2428,
      "step": 122711
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.3864256143569946,
      "learning_rate": 0.000269331503112374,
      "loss": 2.1774,
      "step": 122712
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.9844504594802856,
      "learning_rate": 0.0002693274339836671,
      "loss": 2.3311,
      "step": 122713
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.9522290825843811,
      "learning_rate": 0.0002693233648606627,
      "loss": 1.9639,
      "step": 122714
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.2642772197723389,
      "learning_rate": 0.0002693192957433617,
      "loss": 2.1618,
      "step": 122715
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.838885486125946,
      "learning_rate": 0.00026931522663176485,
      "loss": 2.0924,
      "step": 122716
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.2102806568145752,
      "learning_rate": 0.0002693111575258728,
      "loss": 2.2833,
      "step": 122717
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.8723992109298706,
      "learning_rate": 0.00026930708842568647,
      "loss": 2.0714,
      "step": 122718
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.9481326937675476,
      "learning_rate": 0.00026930301933120654,
      "loss": 2.0955,
      "step": 122719
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.9170257449150085,
      "learning_rate": 0.00026929895024243374,
      "loss": 1.9049,
      "step": 122720
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.0952212810516357,
      "learning_rate": 0.00026929488115936883,
      "loss": 2.0663,
      "step": 122721
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.9596965909004211,
      "learning_rate": 0.0002692908120820125,
      "loss": 2.0028,
      "step": 122722
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.9375078678131104,
      "learning_rate": 0.00026928674301036573,
      "loss": 2.4788,
      "step": 122723
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.9333667159080505,
      "learning_rate": 0.000269282673944429,
      "loss": 1.9598,
      "step": 122724
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.9049269556999207,
      "learning_rate": 0.00026927860488420335,
      "loss": 2.0093,
      "step": 122725
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.991299033164978,
      "learning_rate": 0.0002692745358296893,
      "loss": 2.0641,
      "step": 122726
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.931582510471344,
      "learning_rate": 0.0002692704667808877,
      "loss": 2.2735,
      "step": 122727
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.9210245013237,
      "learning_rate": 0.0002692663977377993,
      "loss": 1.9989,
      "step": 122728
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.9296154975891113,
      "learning_rate": 0.00026926232870042483,
      "loss": 2.0032,
      "step": 122729
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.8766571283340454,
      "learning_rate": 0.0002692582596687651,
      "loss": 2.1025,
      "step": 122730
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.9642263650894165,
      "learning_rate": 0.00026925419064282083,
      "loss": 2.2915,
      "step": 122731
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.114100694656372,
      "learning_rate": 0.0002692501216225928,
      "loss": 2.0166,
      "step": 122732
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.0127032995224,
      "learning_rate": 0.00026924605260808174,
      "loss": 2.3165,
      "step": 122733
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.9512725472450256,
      "learning_rate": 0.0002692419835992885,
      "loss": 2.1333,
      "step": 122734
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.1220089197158813,
      "learning_rate": 0.0002692379145962136,
      "loss": 1.9744,
      "step": 122735
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.0966553688049316,
      "learning_rate": 0.00026923384559885804,
      "loss": 2.1268,
      "step": 122736
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.9052039980888367,
      "learning_rate": 0.00026922977660722255,
      "loss": 2.0122,
      "step": 122737
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.2607531547546387,
      "learning_rate": 0.00026922570762130774,
      "loss": 2.2101,
      "step": 122738
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.0710337162017822,
      "learning_rate": 0.0002692216386411145,
      "loss": 2.2267,
      "step": 122739
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.9836722612380981,
      "learning_rate": 0.0002692175696666435,
      "loss": 2.1739,
      "step": 122740
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.9347145557403564,
      "learning_rate": 0.0002692135006978955,
      "loss": 2.0637,
      "step": 122741
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.9237506985664368,
      "learning_rate": 0.0002692094317348714,
      "loss": 2.0814,
      "step": 122742
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.0754884481430054,
      "learning_rate": 0.0002692053627775718,
      "loss": 2.008,
      "step": 122743
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.9869471192359924,
      "learning_rate": 0.0002692012938259975,
      "loss": 2.1699,
      "step": 122744
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.8940064907073975,
      "learning_rate": 0.0002691972248801492,
      "loss": 2.0045,
      "step": 122745
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.1456027030944824,
      "learning_rate": 0.00026919315594002775,
      "loss": 2.2512,
      "step": 122746
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.9036468267440796,
      "learning_rate": 0.0002691890870056339,
      "loss": 2.1828,
      "step": 122747
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.0294969081878662,
      "learning_rate": 0.0002691850180769683,
      "loss": 2.0985,
      "step": 122748
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.9385254383087158,
      "learning_rate": 0.000269180949154032,
      "loss": 2.0747,
      "step": 122749
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.050465703010559,
      "learning_rate": 0.0002691768802368254,
      "loss": 2.015,
      "step": 122750
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.0163718461990356,
      "learning_rate": 0.0002691728113253494,
      "loss": 2.3353,
      "step": 122751
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.9621670246124268,
      "learning_rate": 0.0002691687424196047,
      "loss": 2.2713,
      "step": 122752
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.003145456314087,
      "learning_rate": 0.00026916467351959216,
      "loss": 2.2834,
      "step": 122753
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.2058734893798828,
      "learning_rate": 0.0002691606046253125,
      "loss": 1.9543,
      "step": 122754
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.9699937105178833,
      "learning_rate": 0.0002691565357367666,
      "loss": 2.1428,
      "step": 122755
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.9313612580299377,
      "learning_rate": 0.0002691524668539549,
      "loss": 2.0762,
      "step": 122756
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.947451114654541,
      "learning_rate": 0.0002691483979768784,
      "loss": 2.1957,
      "step": 122757
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.9371024370193481,
      "learning_rate": 0.0002691443291055377,
      "loss": 2.2125,
      "step": 122758
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.9970347881317139,
      "learning_rate": 0.00026914026023993373,
      "loss": 2.0908,
      "step": 122759
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.9204145669937134,
      "learning_rate": 0.0002691361913800672,
      "loss": 2.0728,
      "step": 122760
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.0221374034881592,
      "learning_rate": 0.0002691321225259389,
      "loss": 2.3392,
      "step": 122761
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.0952972173690796,
      "learning_rate": 0.00026912805367754943,
      "loss": 2.2756,
      "step": 122762
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.9902197122573853,
      "learning_rate": 0.0002691239848348996,
      "loss": 2.2934,
      "step": 122763
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.8967428803443909,
      "learning_rate": 0.00026911991599799025,
      "loss": 2.0389,
      "step": 122764
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.144627332687378,
      "learning_rate": 0.00026911584716682205,
      "loss": 2.147,
      "step": 122765
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.9897646903991699,
      "learning_rate": 0.0002691117783413958,
      "loss": 2.1596,
      "step": 122766
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.90428227186203,
      "learning_rate": 0.00026910770952171236,
      "loss": 2.0499,
      "step": 122767
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.8783979415893555,
      "learning_rate": 0.0002691036407077723,
      "loss": 2.1638,
      "step": 122768
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.9319818019866943,
      "learning_rate": 0.00026909957189957644,
      "loss": 2.2883,
      "step": 122769
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.916642427444458,
      "learning_rate": 0.00026909550309712555,
      "loss": 2.3328,
      "step": 122770
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.9452788829803467,
      "learning_rate": 0.0002690914343004204,
      "loss": 2.2009,
      "step": 122771
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.8768438100814819,
      "learning_rate": 0.0002690873655094617,
      "loss": 1.9724,
      "step": 122772
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.0442723035812378,
      "learning_rate": 0.0002690832967242504,
      "loss": 2.0161,
      "step": 122773
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.9211859107017517,
      "learning_rate": 0.000269079227944787,
      "loss": 2.2727,
      "step": 122774
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.8811283707618713,
      "learning_rate": 0.0002690751591710723,
      "loss": 2.0071,
      "step": 122775
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.036095142364502,
      "learning_rate": 0.0002690710904031071,
      "loss": 2.2261,
      "step": 122776
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.1361559629440308,
      "learning_rate": 0.00026906702164089226,
      "loss": 2.0564,
      "step": 122777
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.109196424484253,
      "learning_rate": 0.00026906295288442835,
      "loss": 2.2154,
      "step": 122778
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.927272617816925,
      "learning_rate": 0.0002690588841337163,
      "loss": 2.1317,
      "step": 122779
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.0314174890518188,
      "learning_rate": 0.00026905481538875686,
      "loss": 2.178,
      "step": 122780
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.0423505306243896,
      "learning_rate": 0.0002690507466495506,
      "loss": 2.352,
      "step": 122781
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.0695728063583374,
      "learning_rate": 0.00026904667791609836,
      "loss": 1.9852,
      "step": 122782
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.9213456511497498,
      "learning_rate": 0.000269042609188401,
      "loss": 1.9048,
      "step": 122783
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.149280071258545,
      "learning_rate": 0.0002690385404664591,
      "loss": 2.1752,
      "step": 122784
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.2100090980529785,
      "learning_rate": 0.00026903447175027364,
      "loss": 1.9941,
      "step": 122785
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.0030865669250488,
      "learning_rate": 0.00026903040303984533,
      "loss": 2.1686,
      "step": 122786
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.0279850959777832,
      "learning_rate": 0.00026902633433517473,
      "loss": 2.2112,
      "step": 122787
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.9482123255729675,
      "learning_rate": 0.00026902226563626266,
      "loss": 2.1181,
      "step": 122788
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.6592565774917603,
      "learning_rate": 0.00026901819694311,
      "loss": 2.1361,
      "step": 122789
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.0747188329696655,
      "learning_rate": 0.00026901412825571744,
      "loss": 2.356,
      "step": 122790
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.90707927942276,
      "learning_rate": 0.0002690100595740857,
      "loss": 2.1693,
      "step": 122791
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.9889410734176636,
      "learning_rate": 0.00026900599089821575,
      "loss": 2.1448,
      "step": 122792
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.1326417922973633,
      "learning_rate": 0.000269001922228108,
      "loss": 2.1738,
      "step": 122793
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.0869029760360718,
      "learning_rate": 0.0002689978535637634,
      "loss": 2.064,
      "step": 122794
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.9501826763153076,
      "learning_rate": 0.0002689937849051827,
      "loss": 2.0082,
      "step": 122795
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.9404014348983765,
      "learning_rate": 0.00026898971625236663,
      "loss": 2.0496,
      "step": 122796
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.9498875141143799,
      "learning_rate": 0.00026898564760531596,
      "loss": 2.2527,
      "step": 122797
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.0003961324691772,
      "learning_rate": 0.0002689815789640315,
      "loss": 2.3355,
      "step": 122798
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.9294370412826538,
      "learning_rate": 0.00026897751032851393,
      "loss": 1.8677,
      "step": 122799
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.9348942637443542,
      "learning_rate": 0.000268973441698764,
      "loss": 2.1695,
      "step": 122800
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.9140785932540894,
      "learning_rate": 0.00026896937307478247,
      "loss": 2.1061,
      "step": 122801
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.954117476940155,
      "learning_rate": 0.00026896530445657006,
      "loss": 1.9104,
      "step": 122802
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.9126254916191101,
      "learning_rate": 0.0002689612358441277,
      "loss": 2.2275,
      "step": 122803
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.9138352274894714,
      "learning_rate": 0.0002689571672374561,
      "loss": 2.0622,
      "step": 122804
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.0330806970596313,
      "learning_rate": 0.0002689530986365558,
      "loss": 2.1807,
      "step": 122805
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.0413627624511719,
      "learning_rate": 0.0002689490300414277,
      "loss": 2.0677,
      "step": 122806
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.8831827044487,
      "learning_rate": 0.00026894496145207265,
      "loss": 2.1133,
      "step": 122807
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.9273144602775574,
      "learning_rate": 0.0002689408928684913,
      "loss": 2.267,
      "step": 122808
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.899807333946228,
      "learning_rate": 0.0002689368242906844,
      "loss": 2.0156,
      "step": 122809
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.9438652396202087,
      "learning_rate": 0.00026893275571865274,
      "loss": 1.9784,
      "step": 122810
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.9748702049255371,
      "learning_rate": 0.00026892868715239703,
      "loss": 2.1613,
      "step": 122811
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.9577786922454834,
      "learning_rate": 0.00026892461859191817,
      "loss": 2.2383,
      "step": 122812
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.0119574069976807,
      "learning_rate": 0.0002689205500372167,
      "loss": 2.2359,
      "step": 122813
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.9038271307945251,
      "learning_rate": 0.00026891648148829353,
      "loss": 2.1254,
      "step": 122814
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.942108690738678,
      "learning_rate": 0.0002689124129451493,
      "loss": 2.1924,
      "step": 122815
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.2038429975509644,
      "learning_rate": 0.000268908344407785,
      "loss": 2.3074,
      "step": 122816
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.0572096109390259,
      "learning_rate": 0.0002689042758762011,
      "loss": 2.0445,
      "step": 122817
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.893234372138977,
      "learning_rate": 0.0002689002073503985,
      "loss": 2.03,
      "step": 122818
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.083648920059204,
      "learning_rate": 0.000268896138830378,
      "loss": 2.0908,
      "step": 122819
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.9862536787986755,
      "learning_rate": 0.0002688920703161402,
      "loss": 2.2493,
      "step": 122820
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.0489040613174438,
      "learning_rate": 0.000268888001807686,
      "loss": 2.1977,
      "step": 122821
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.012180209159851,
      "learning_rate": 0.00026888393330501615,
      "loss": 2.218,
      "step": 122822
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.9075819253921509,
      "learning_rate": 0.00026887986480813133,
      "loss": 2.0093,
      "step": 122823
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.987544596195221,
      "learning_rate": 0.0002688757963170323,
      "loss": 1.9345,
      "step": 122824
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.0066508054733276,
      "learning_rate": 0.0002688717278317198,
      "loss": 2.3808,
      "step": 122825
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.9730666279792786,
      "learning_rate": 0.00026886765935219475,
      "loss": 1.8687,
      "step": 122826
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.1521145105361938,
      "learning_rate": 0.0002688635908784577,
      "loss": 2.3062,
      "step": 122827
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.023996353149414,
      "learning_rate": 0.00026885952241050964,
      "loss": 2.059,
      "step": 122828
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.0759353637695312,
      "learning_rate": 0.00026885545394835105,
      "loss": 1.9872,
      "step": 122829
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.0105204582214355,
      "learning_rate": 0.0002688513854919828,
      "loss": 2.1222,
      "step": 122830
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.128029704093933,
      "learning_rate": 0.0002688473170414057,
      "loss": 2.3281,
      "step": 122831
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.9769798517227173,
      "learning_rate": 0.0002688432485966205,
      "loss": 2.0634,
      "step": 122832
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.0110262632369995,
      "learning_rate": 0.00026883918015762796,
      "loss": 2.3685,
      "step": 122833
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.1652064323425293,
      "learning_rate": 0.0002688351117244288,
      "loss": 1.9943,
      "step": 122834
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.064927101135254,
      "learning_rate": 0.0002688310432970237,
      "loss": 2.0816,
      "step": 122835
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.9532181620597839,
      "learning_rate": 0.0002688269748754135,
      "loss": 2.1373,
      "step": 122836
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.3973774909973145,
      "learning_rate": 0.000268822906459599,
      "loss": 2.41,
      "step": 122837
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.1277204751968384,
      "learning_rate": 0.0002688188380495809,
      "loss": 2.0441,
      "step": 122838
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.903175413608551,
      "learning_rate": 0.00026881476964535995,
      "loss": 2.0002,
      "step": 122839
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.9582982063293457,
      "learning_rate": 0.000268810701246937,
      "loss": 2.2622,
      "step": 122840
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.0584648847579956,
      "learning_rate": 0.0002688066328543127,
      "loss": 2.0877,
      "step": 122841
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.0237425565719604,
      "learning_rate": 0.0002688025644674878,
      "loss": 1.9737,
      "step": 122842
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.9884588718414307,
      "learning_rate": 0.00026879849608646303,
      "loss": 2.1563,
      "step": 122843
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.948479175567627,
      "learning_rate": 0.0002687944277112393,
      "loss": 1.9741,
      "step": 122844
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.11847984790802,
      "learning_rate": 0.00026879035934181724,
      "loss": 2.054,
      "step": 122845
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.9642878174781799,
      "learning_rate": 0.0002687862909781976,
      "loss": 2.1606,
      "step": 122846
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.0263855457305908,
      "learning_rate": 0.00026878222262038135,
      "loss": 1.933,
      "step": 122847
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.9717679023742676,
      "learning_rate": 0.000268778154268369,
      "loss": 2.2609,
      "step": 122848
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.049991488456726,
      "learning_rate": 0.00026877408592216126,
      "loss": 2.004,
      "step": 122849
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.024190068244934,
      "learning_rate": 0.00026877001758175906,
      "loss": 2.0385,
      "step": 122850
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.16993248462677,
      "learning_rate": 0.00026876594924716315,
      "loss": 2.3119,
      "step": 122851
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.9501126408576965,
      "learning_rate": 0.0002687618809183742,
      "loss": 2.2255,
      "step": 122852
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.979284405708313,
      "learning_rate": 0.0002687578125953931,
      "loss": 2.4637,
      "step": 122853
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.1507458686828613,
      "learning_rate": 0.0002687537442782204,
      "loss": 2.2889,
      "step": 122854
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.247334361076355,
      "learning_rate": 0.000268749675966857,
      "loss": 2.2291,
      "step": 122855
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.9519734978675842,
      "learning_rate": 0.0002687456076613036,
      "loss": 2.1527,
      "step": 122856
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.878044605255127,
      "learning_rate": 0.00026874153936156097,
      "loss": 2.0648,
      "step": 122857
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.9546947479248047,
      "learning_rate": 0.0002687374710676299,
      "loss": 2.0864,
      "step": 122858
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.1330076456069946,
      "learning_rate": 0.0002687334027795113,
      "loss": 2.0299,
      "step": 122859
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.9998750686645508,
      "learning_rate": 0.00026872933449720554,
      "loss": 1.9981,
      "step": 122860
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.8835187554359436,
      "learning_rate": 0.00026872526622071365,
      "loss": 1.8707,
      "step": 122861
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.9668550491333008,
      "learning_rate": 0.0002687211979500363,
      "loss": 2.034,
      "step": 122862
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.056676983833313,
      "learning_rate": 0.00026871712968517424,
      "loss": 2.1708,
      "step": 122863
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.008733868598938,
      "learning_rate": 0.00026871306142612825,
      "loss": 2.2825,
      "step": 122864
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.9638134241104126,
      "learning_rate": 0.00026870899317289927,
      "loss": 2.1553,
      "step": 122865
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.1686450242996216,
      "learning_rate": 0.00026870492492548774,
      "loss": 2.1441,
      "step": 122866
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.053617000579834,
      "learning_rate": 0.00026870085668389453,
      "loss": 2.2545,
      "step": 122867
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.9211552143096924,
      "learning_rate": 0.00026869678844812045,
      "loss": 2.1493,
      "step": 122868
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.9951989054679871,
      "learning_rate": 0.0002686927202181662,
      "loss": 2.0412,
      "step": 122869
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.0785751342773438,
      "learning_rate": 0.0002686886519940326,
      "loss": 2.1913,
      "step": 122870
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.046060562133789,
      "learning_rate": 0.00026868458377572046,
      "loss": 2.0432,
      "step": 122871
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.0209585428237915,
      "learning_rate": 0.00026868051556323033,
      "loss": 2.0484,
      "step": 122872
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.9125143885612488,
      "learning_rate": 0.0002686764473565631,
      "loss": 2.0506,
      "step": 122873
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.9263206720352173,
      "learning_rate": 0.0002686723791557195,
      "loss": 2.1504,
      "step": 122874
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.0655574798583984,
      "learning_rate": 0.0002686683109607003,
      "loss": 2.2548,
      "step": 122875
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.9272663593292236,
      "learning_rate": 0.00026866424277150624,
      "loss": 2.1366,
      "step": 122876
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.0605354309082031,
      "learning_rate": 0.00026866017458813826,
      "loss": 2.2061,
      "step": 122877
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.9593012928962708,
      "learning_rate": 0.0002686561064105968,
      "loss": 2.2562,
      "step": 122878
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.8583312034606934,
      "learning_rate": 0.0002686520382388827,
      "loss": 1.9802,
      "step": 122879
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.9720779657363892,
      "learning_rate": 0.0002686479700729968,
      "loss": 2.2111,
      "step": 122880
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.9968513250350952,
      "learning_rate": 0.0002686439019129399,
      "loss": 1.8905,
      "step": 122881
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.9990285038948059,
      "learning_rate": 0.0002686398337587126,
      "loss": 1.9845,
      "step": 122882
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.9623472690582275,
      "learning_rate": 0.00026863576561031595,
      "loss": 2.1377,
      "step": 122883
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.9083733558654785,
      "learning_rate": 0.00026863169746775035,
      "loss": 2.216,
      "step": 122884
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.1065441370010376,
      "learning_rate": 0.00026862762933101673,
      "loss": 2.109,
      "step": 122885
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.074066162109375,
      "learning_rate": 0.0002686235612001158,
      "loss": 2.1635,
      "step": 122886
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.0578601360321045,
      "learning_rate": 0.00026861949307504835,
      "loss": 1.9892,
      "step": 122887
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.1087313890457153,
      "learning_rate": 0.0002686154249558151,
      "loss": 2.24,
      "step": 122888
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.9978316426277161,
      "learning_rate": 0.000268611356842417,
      "loss": 2.2452,
      "step": 122889
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.9613432884216309,
      "learning_rate": 0.00026860728873485446,
      "loss": 2.1793,
      "step": 122890
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.9129220247268677,
      "learning_rate": 0.0002686032206331285,
      "loss": 2.1792,
      "step": 122891
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.9666746854782104,
      "learning_rate": 0.00026859915253723974,
      "loss": 1.9829,
      "step": 122892
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.5424818992614746,
      "learning_rate": 0.000268595084447189,
      "loss": 2.208,
      "step": 122893
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.3597146272659302,
      "learning_rate": 0.00026859101636297704,
      "loss": 2.041,
      "step": 122894
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.9311628937721252,
      "learning_rate": 0.00026858694828460467,
      "loss": 2.1442,
      "step": 122895
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.0769208669662476,
      "learning_rate": 0.00026858288021207257,
      "loss": 2.2761,
      "step": 122896
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.8964520692825317,
      "learning_rate": 0.0002685788121453814,
      "loss": 2.3922,
      "step": 122897
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.1912919282913208,
      "learning_rate": 0.00026857474408453205,
      "loss": 2.2298,
      "step": 122898
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.0538402795791626,
      "learning_rate": 0.00026857067602952526,
      "loss": 2.1441,
      "step": 122899
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.3642476797103882,
      "learning_rate": 0.0002685666079803618,
      "loss": 2.1214,
      "step": 122900
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.167258620262146,
      "learning_rate": 0.0002685625399370424,
      "loss": 2.1063,
      "step": 122901
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.495819330215454,
      "learning_rate": 0.0002685584718995678,
      "loss": 1.9864,
      "step": 122902
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.9187167882919312,
      "learning_rate": 0.00026855440386793877,
      "loss": 2.099,
      "step": 122903
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.0334855318069458,
      "learning_rate": 0.0002685503358421561,
      "loss": 2.1798,
      "step": 122904
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.9445802569389343,
      "learning_rate": 0.0002685462678222205,
      "loss": 2.0906,
      "step": 122905
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.0900003910064697,
      "learning_rate": 0.00026854219980813274,
      "loss": 2.1669,
      "step": 122906
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.0982952117919922,
      "learning_rate": 0.0002685381317998936,
      "loss": 2.1362,
      "step": 122907
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.9702279567718506,
      "learning_rate": 0.0002685340637975038,
      "loss": 2.1424,
      "step": 122908
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.0265902280807495,
      "learning_rate": 0.0002685299958009641,
      "loss": 2.2304,
      "step": 122909
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.9922154545783997,
      "learning_rate": 0.0002685259278102752,
      "loss": 2.1531,
      "step": 122910
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.0138131380081177,
      "learning_rate": 0.0002685218598254381,
      "loss": 1.9602,
      "step": 122911
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.9709291458129883,
      "learning_rate": 0.0002685177918464532,
      "loss": 2.2927,
      "step": 122912
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.9627485275268555,
      "learning_rate": 0.00026851372387332154,
      "loss": 1.9466,
      "step": 122913
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.0336546897888184,
      "learning_rate": 0.0002685096559060438,
      "loss": 2.1154,
      "step": 122914
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.0804219245910645,
      "learning_rate": 0.0002685055879446206,
      "loss": 1.9581,
      "step": 122915
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.0781434774398804,
      "learning_rate": 0.00026850151998905283,
      "loss": 2.104,
      "step": 122916
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.9551577568054199,
      "learning_rate": 0.00026849745203934126,
      "loss": 2.1468,
      "step": 122917
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.9057771563529968,
      "learning_rate": 0.00026849338409548666,
      "loss": 1.8281,
      "step": 122918
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.0312057733535767,
      "learning_rate": 0.00026848931615748963,
      "loss": 2.0832,
      "step": 122919
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.9088267683982849,
      "learning_rate": 0.00026848524822535114,
      "loss": 2.1074,
      "step": 122920
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.9589796662330627,
      "learning_rate": 0.0002684811802990718,
      "loss": 2.1102,
      "step": 122921
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.9703676104545593,
      "learning_rate": 0.0002684771123786523,
      "loss": 2.3078,
      "step": 122922
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.899754524230957,
      "learning_rate": 0.00026847304446409357,
      "loss": 2.1805,
      "step": 122923
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.001339077949524,
      "learning_rate": 0.0002684689765553963,
      "loss": 2.045,
      "step": 122924
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.0638775825500488,
      "learning_rate": 0.0002684649086525613,
      "loss": 2.2872,
      "step": 122925
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.0119038820266724,
      "learning_rate": 0.00026846084075558927,
      "loss": 2.2089,
      "step": 122926
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.8845939636230469,
      "learning_rate": 0.0002684567728644809,
      "loss": 2.0383,
      "step": 122927
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.0804270505905151,
      "learning_rate": 0.000268452704979237,
      "loss": 1.9451,
      "step": 122928
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.9908099174499512,
      "learning_rate": 0.0002684486370998583,
      "loss": 2.1067,
      "step": 122929
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.028634786605835,
      "learning_rate": 0.00026844456922634564,
      "loss": 2.1211,
      "step": 122930
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.9687643051147461,
      "learning_rate": 0.0002684405013586997,
      "loss": 1.8939,
      "step": 122931
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.9246329665184021,
      "learning_rate": 0.00026843643349692145,
      "loss": 2.1054,
      "step": 122932
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.9653034806251526,
      "learning_rate": 0.0002684323656410113,
      "loss": 2.1304,
      "step": 122933
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.2714588642120361,
      "learning_rate": 0.0002684282977909702,
      "loss": 1.9688,
      "step": 122934
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.9890891313552856,
      "learning_rate": 0.00026842422994679887,
      "loss": 2.3586,
      "step": 122935
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.9699506163597107,
      "learning_rate": 0.000268420162108498,
      "loss": 2.2348,
      "step": 122936
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.0185540914535522,
      "learning_rate": 0.0002684160942760685,
      "loss": 2.16,
      "step": 122937
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.0499094724655151,
      "learning_rate": 0.00026841202644951114,
      "loss": 1.931,
      "step": 122938
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.9667209982872009,
      "learning_rate": 0.00026840795862882646,
      "loss": 2.0941,
      "step": 122939
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.0079362392425537,
      "learning_rate": 0.00026840389081401534,
      "loss": 2.0905,
      "step": 122940
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.9510459899902344,
      "learning_rate": 0.00026839982300507853,
      "loss": 2.3303,
      "step": 122941
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.0073926448822021,
      "learning_rate": 0.00026839575520201675,
      "loss": 2.2589,
      "step": 122942
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.968745768070221,
      "learning_rate": 0.00026839168740483085,
      "loss": 2.1459,
      "step": 122943
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.9696442484855652,
      "learning_rate": 0.00026838761961352165,
      "loss": 1.8981,
      "step": 122944
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.9783512353897095,
      "learning_rate": 0.00026838355182808965,
      "loss": 2.5112,
      "step": 122945
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.8985642790794373,
      "learning_rate": 0.0002683794840485357,
      "loss": 1.9657,
      "step": 122946
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.979945957660675,
      "learning_rate": 0.00026837541627486067,
      "loss": 2.1162,
      "step": 122947
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.0799129009246826,
      "learning_rate": 0.0002683713485070652,
      "loss": 2.0544,
      "step": 122948
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.023320198059082,
      "learning_rate": 0.00026836728074515015,
      "loss": 2.0267,
      "step": 122949
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.0556750297546387,
      "learning_rate": 0.0002683632129891163,
      "loss": 2.0983,
      "step": 122950
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.172547459602356,
      "learning_rate": 0.0002683591452389642,
      "loss": 2.3267,
      "step": 122951
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.9932790994644165,
      "learning_rate": 0.00026835507749469476,
      "loss": 2.1129,
      "step": 122952
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.095053791999817,
      "learning_rate": 0.0002683510097563087,
      "loss": 2.3404,
      "step": 122953
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.219133973121643,
      "learning_rate": 0.00026834694202380676,
      "loss": 1.9588,
      "step": 122954
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.1016627550125122,
      "learning_rate": 0.00026834287429718974,
      "loss": 1.8095,
      "step": 122955
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.0482125282287598,
      "learning_rate": 0.00026833880657645853,
      "loss": 2.3093,
      "step": 122956
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.1278460025787354,
      "learning_rate": 0.00026833473886161356,
      "loss": 2.1496,
      "step": 122957
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.0279415845870972,
      "learning_rate": 0.0002683306711526558,
      "loss": 1.9704,
      "step": 122958
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.0114479064941406,
      "learning_rate": 0.0002683266034495859,
      "loss": 2.2143,
      "step": 122959
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.0924780368804932,
      "learning_rate": 0.0002683225357524047,
      "loss": 1.9425,
      "step": 122960
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.9625899195671082,
      "learning_rate": 0.000268318468061113,
      "loss": 2.2711,
      "step": 122961
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.0433560609817505,
      "learning_rate": 0.00026831440037571155,
      "loss": 2.1927,
      "step": 122962
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.0816303491592407,
      "learning_rate": 0.000268310332696201,
      "loss": 2.2376,
      "step": 122963
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.9592939615249634,
      "learning_rate": 0.0002683062650225821,
      "loss": 2.1842,
      "step": 122964
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.053107738494873,
      "learning_rate": 0.0002683021973548557,
      "loss": 2.0729,
      "step": 122965
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.0145000219345093,
      "learning_rate": 0.00026829812969302253,
      "loss": 2.1053,
      "step": 122966
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.05632746219635,
      "learning_rate": 0.0002682940620370833,
      "loss": 2.1821,
      "step": 122967
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.0051686763763428,
      "learning_rate": 0.0002682899943870389,
      "loss": 2.0322,
      "step": 122968
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.2769700288772583,
      "learning_rate": 0.00026828592674288994,
      "loss": 1.9514,
      "step": 122969
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.8903647065162659,
      "learning_rate": 0.0002682818591046372,
      "loss": 1.8698,
      "step": 122970
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.1732350587844849,
      "learning_rate": 0.0002682777914722814,
      "loss": 2.1203,
      "step": 122971
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.9581547975540161,
      "learning_rate": 0.0002682737238458234,
      "loss": 2.0687,
      "step": 122972
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.9559314250946045,
      "learning_rate": 0.0002682696562252639,
      "loss": 2.1104,
      "step": 122973
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.9663894772529602,
      "learning_rate": 0.00026826558861060384,
      "loss": 2.1473,
      "step": 122974
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.9288101196289062,
      "learning_rate": 0.00026826152100184366,
      "loss": 2.1866,
      "step": 122975
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.936357855796814,
      "learning_rate": 0.00026825745339898425,
      "loss": 2.1744,
      "step": 122976
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.926889181137085,
      "learning_rate": 0.0002682533858020264,
      "loss": 2.0395,
      "step": 122977
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.1262341737747192,
      "learning_rate": 0.0002682493182109708,
      "loss": 1.9489,
      "step": 122978
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.1024783849716187,
      "learning_rate": 0.0002682452506258183,
      "loss": 1.8317,
      "step": 122979
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.9215487241744995,
      "learning_rate": 0.00026824118304656956,
      "loss": 2.2633,
      "step": 122980
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.9678804874420166,
      "learning_rate": 0.00026823711547322555,
      "loss": 2.1349,
      "step": 122981
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.1994341611862183,
      "learning_rate": 0.0002682330479057867,
      "loss": 2.0876,
      "step": 122982
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.9943251609802246,
      "learning_rate": 0.000268228980344254,
      "loss": 2.0065,
      "step": 122983
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.937759518623352,
      "learning_rate": 0.000268224912788628,
      "loss": 2.0816,
      "step": 122984
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.481223702430725,
      "learning_rate": 0.0002682208452389097,
      "loss": 2.0346,
      "step": 122985
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.022308111190796,
      "learning_rate": 0.0002682167776950997,
      "loss": 2.1229,
      "step": 122986
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.0681087970733643,
      "learning_rate": 0.0002682127101571989,
      "loss": 2.1014,
      "step": 122987
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.0434727668762207,
      "learning_rate": 0.0002682086426252079,
      "loss": 1.9836,
      "step": 122988
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.0115184783935547,
      "learning_rate": 0.00026820457509912747,
      "loss": 2.0694,
      "step": 122989
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.1401804685592651,
      "learning_rate": 0.0002682005075789584,
      "loss": 2.1943,
      "step": 122990
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.0589159727096558,
      "learning_rate": 0.00026819644006470154,
      "loss": 2.218,
      "step": 122991
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.9594960808753967,
      "learning_rate": 0.0002681923725563575,
      "loss": 2.2016,
      "step": 122992
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.96323162317276,
      "learning_rate": 0.00026818830505392715,
      "loss": 2.0456,
      "step": 122993
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.0437498092651367,
      "learning_rate": 0.00026818423755741117,
      "loss": 2.0259,
      "step": 122994
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.9119740128517151,
      "learning_rate": 0.00026818017006681027,
      "loss": 2.1467,
      "step": 122995
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.0881839990615845,
      "learning_rate": 0.0002681761025821254,
      "loss": 1.9011,
      "step": 122996
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.0495262145996094,
      "learning_rate": 0.00026817203510335713,
      "loss": 2.0303,
      "step": 122997
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.9571803212165833,
      "learning_rate": 0.00026816796763050625,
      "loss": 2.1774,
      "step": 122998
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.084231972694397,
      "learning_rate": 0.00026816390016357365,
      "loss": 2.0817,
      "step": 122999
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.0144234895706177,
      "learning_rate": 0.00026815983270255986,
      "loss": 2.3504,
      "step": 123000
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.011051893234253,
      "learning_rate": 0.0002681557652474658,
      "loss": 1.9788,
      "step": 123001
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.9369947910308838,
      "learning_rate": 0.0002681516977982922,
      "loss": 1.9948,
      "step": 123002
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.0015052556991577,
      "learning_rate": 0.0002681476303550398,
      "loss": 1.8858,
      "step": 123003
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.9905787706375122,
      "learning_rate": 0.00026814356291770934,
      "loss": 2.2078,
      "step": 123004
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.0174399614334106,
      "learning_rate": 0.00026813949548630165,
      "loss": 2.1817,
      "step": 123005
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.0479693412780762,
      "learning_rate": 0.00026813542806081734,
      "loss": 2.0048,
      "step": 123006
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.9400562644004822,
      "learning_rate": 0.00026813136064125727,
      "loss": 2.3829,
      "step": 123007
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.9188531637191772,
      "learning_rate": 0.0002681272932276222,
      "loss": 1.9818,
      "step": 123008
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.9789927005767822,
      "learning_rate": 0.0002681232258199129,
      "loss": 2.3399,
      "step": 123009
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.0338706970214844,
      "learning_rate": 0.0002681191584181301,
      "loss": 2.1299,
      "step": 123010
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.9957134127616882,
      "learning_rate": 0.0002681150910222746,
      "loss": 2.1935,
      "step": 123011
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.0840387344360352,
      "learning_rate": 0.00026811102363234697,
      "loss": 2.2601,
      "step": 123012
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.0761996507644653,
      "learning_rate": 0.00026810695624834814,
      "loss": 2.2087,
      "step": 123013
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.9061263799667358,
      "learning_rate": 0.0002681028888702788,
      "loss": 2.2582,
      "step": 123014
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.0299328565597534,
      "learning_rate": 0.0002680988214981398,
      "loss": 1.979,
      "step": 123015
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.0135608911514282,
      "learning_rate": 0.0002680947541319318,
      "loss": 2.1641,
      "step": 123016
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.0318505764007568,
      "learning_rate": 0.00026809068677165573,
      "loss": 2.2111,
      "step": 123017
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.2054774761199951,
      "learning_rate": 0.00026808661941731205,
      "loss": 1.9388,
      "step": 123018
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.9420952796936035,
      "learning_rate": 0.00026808255206890166,
      "loss": 2.0893,
      "step": 123019
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.9801543951034546,
      "learning_rate": 0.00026807848472642533,
      "loss": 2.196,
      "step": 123020
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.0326504707336426,
      "learning_rate": 0.00026807441738988387,
      "loss": 2.1813,
      "step": 123021
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.9557318687438965,
      "learning_rate": 0.00026807035005927787,
      "loss": 1.9704,
      "step": 123022
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.9984056353569031,
      "learning_rate": 0.0002680662827346084,
      "loss": 2.2465,
      "step": 123023
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.9913934469223022,
      "learning_rate": 0.0002680622154158759,
      "loss": 2.0598,
      "step": 123024
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.8904922604560852,
      "learning_rate": 0.00026805814810308123,
      "loss": 2.1209,
      "step": 123025
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.9112153649330139,
      "learning_rate": 0.0002680540807962251,
      "loss": 1.9954,
      "step": 123026
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.9468305110931396,
      "learning_rate": 0.0002680500134953083,
      "loss": 2.0076,
      "step": 123027
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.0189497470855713,
      "learning_rate": 0.0002680459462003317,
      "loss": 2.181,
      "step": 123028
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.0807784795761108,
      "learning_rate": 0.00026804187891129607,
      "loss": 2.347,
      "step": 123029
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.9550772309303284,
      "learning_rate": 0.00026803781162820187,
      "loss": 1.9765,
      "step": 123030
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.0070812702178955,
      "learning_rate": 0.0002680337443510501,
      "loss": 2.2128,
      "step": 123031
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.0256239175796509,
      "learning_rate": 0.00026802967707984144,
      "loss": 2.16,
      "step": 123032
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.9218268990516663,
      "learning_rate": 0.0002680256098145767,
      "loss": 2.0848,
      "step": 123033
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.931252121925354,
      "learning_rate": 0.00026802154255525654,
      "loss": 2.1771,
      "step": 123034
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.0662105083465576,
      "learning_rate": 0.00026801747530188195,
      "loss": 1.9824,
      "step": 123035
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.9921488165855408,
      "learning_rate": 0.00026801340805445334,
      "loss": 2.2416,
      "step": 123036
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.9545858502388,
      "learning_rate": 0.0002680093408129717,
      "loss": 2.111,
      "step": 123037
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.2328786849975586,
      "learning_rate": 0.00026800527357743765,
      "loss": 2.116,
      "step": 123038
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.9488669037818909,
      "learning_rate": 0.00026800120634785213,
      "loss": 2.226,
      "step": 123039
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.0221078395843506,
      "learning_rate": 0.0002679971391242157,
      "loss": 2.1613,
      "step": 123040
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.972752034664154,
      "learning_rate": 0.0002679930719065294,
      "loss": 2.0562,
      "step": 123041
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.9375463724136353,
      "learning_rate": 0.0002679890046947936,
      "loss": 2.2694,
      "step": 123042
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.0247368812561035,
      "learning_rate": 0.0002679849374890093,
      "loss": 2.0027,
      "step": 123043
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.8909436464309692,
      "learning_rate": 0.00026798087028917717,
      "loss": 2.1426,
      "step": 123044
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.176927089691162,
      "learning_rate": 0.000267976803095298,
      "loss": 2.1115,
      "step": 123045
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.9243032932281494,
      "learning_rate": 0.0002679727359073726,
      "loss": 2.331,
      "step": 123046
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.972829282283783,
      "learning_rate": 0.0002679686687254016,
      "loss": 2.2808,
      "step": 123047
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.0596421957015991,
      "learning_rate": 0.000267964601549386,
      "loss": 2.2218,
      "step": 123048
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.0459645986557007,
      "learning_rate": 0.00026796053437932625,
      "loss": 2.1938,
      "step": 123049
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.0110881328582764,
      "learning_rate": 0.00026795646721522324,
      "loss": 2.1339,
      "step": 123050
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.9225775599479675,
      "learning_rate": 0.00026795240005707776,
      "loss": 2.0343,
      "step": 123051
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.023237943649292,
      "learning_rate": 0.0002679483329048905,
      "loss": 2.1191,
      "step": 123052
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.0599305629730225,
      "learning_rate": 0.00026794426575866225,
      "loss": 2.2886,
      "step": 123053
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.0548768043518066,
      "learning_rate": 0.00026794019861839385,
      "loss": 2.2563,
      "step": 123054
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.8689814209938049,
      "learning_rate": 0.0002679361314840859,
      "loss": 2.1831,
      "step": 123055
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.9237477779388428,
      "learning_rate": 0.00026793206435573924,
      "loss": 1.9893,
      "step": 123056
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.0545599460601807,
      "learning_rate": 0.0002679279972333546,
      "loss": 2.2129,
      "step": 123057
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.004169225692749,
      "learning_rate": 0.00026792393011693274,
      "loss": 2.0613,
      "step": 123058
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.9119116067886353,
      "learning_rate": 0.0002679198630064744,
      "loss": 2.2593,
      "step": 123059
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.9261108636856079,
      "learning_rate": 0.0002679157959019806,
      "loss": 1.9831,
      "step": 123060
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.0477255582809448,
      "learning_rate": 0.00026791172880345163,
      "loss": 2.2937,
      "step": 123061
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.0085268020629883,
      "learning_rate": 0.00026790766171088846,
      "loss": 2.2826,
      "step": 123062
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.072090983390808,
      "learning_rate": 0.0002679035946242919,
      "loss": 2.1029,
      "step": 123063
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.9163498282432556,
      "learning_rate": 0.0002678995275436627,
      "loss": 2.1399,
      "step": 123064
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.0848660469055176,
      "learning_rate": 0.00026789546046900156,
      "loss": 2.1465,
      "step": 123065
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.1003422737121582,
      "learning_rate": 0.00026789139340030937,
      "loss": 2.2645,
      "step": 123066
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.9006617069244385,
      "learning_rate": 0.00026788732633758664,
      "loss": 2.0871,
      "step": 123067
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.9693412184715271,
      "learning_rate": 0.0002678832592808343,
      "loss": 2.0857,
      "step": 123068
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.0299915075302124,
      "learning_rate": 0.00026787919223005306,
      "loss": 2.2238,
      "step": 123069
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.9318321347236633,
      "learning_rate": 0.0002678751251852437,
      "loss": 2.0944,
      "step": 123070
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.001218318939209,
      "learning_rate": 0.0002678710581464069,
      "loss": 2.158,
      "step": 123071
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.9057509303092957,
      "learning_rate": 0.00026786699111354357,
      "loss": 2.2533,
      "step": 123072
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.9827242493629456,
      "learning_rate": 0.00026786292408665437,
      "loss": 2.2249,
      "step": 123073
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.9734498262405396,
      "learning_rate": 0.00026785885706574004,
      "loss": 2.2136,
      "step": 123074
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.1997168064117432,
      "learning_rate": 0.0002678547900508013,
      "loss": 2.0793,
      "step": 123075
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.9609562754631042,
      "learning_rate": 0.00026785072304183894,
      "loss": 2.2132,
      "step": 123076
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.8808228373527527,
      "learning_rate": 0.0002678466560388538,
      "loss": 1.9236,
      "step": 123077
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.8863645195960999,
      "learning_rate": 0.0002678425890418466,
      "loss": 2.3482,
      "step": 123078
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.0646624565124512,
      "learning_rate": 0.000267838522050818,
      "loss": 2.0923,
      "step": 123079
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.9459385275840759,
      "learning_rate": 0.000267834455065769,
      "loss": 2.0054,
      "step": 123080
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.9898574948310852,
      "learning_rate": 0.00026783038808669997,
      "loss": 2.2454,
      "step": 123081
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.2114753723144531,
      "learning_rate": 0.0002678263211136119,
      "loss": 2.4103,
      "step": 123082
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.002795696258545,
      "learning_rate": 0.00026782225414650557,
      "loss": 1.9969,
      "step": 123083
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.9472159743309021,
      "learning_rate": 0.0002678181871853818,
      "loss": 2.1147,
      "step": 123084
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.9409629702568054,
      "learning_rate": 0.0002678141202302411,
      "loss": 2.1791,
      "step": 123085
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.938178539276123,
      "learning_rate": 0.00026781005328108435,
      "loss": 2.247,
      "step": 123086
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.9858895540237427,
      "learning_rate": 0.0002678059863379124,
      "loss": 2.3794,
      "step": 123087
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.938866138458252,
      "learning_rate": 0.0002678019194007259,
      "loss": 1.7509,
      "step": 123088
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.9654106497764587,
      "learning_rate": 0.0002677978524695256,
      "loss": 2.1908,
      "step": 123089
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.9340547919273376,
      "learning_rate": 0.0002677937855443123,
      "loss": 1.9527,
      "step": 123090
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.0105867385864258,
      "learning_rate": 0.0002677897186250868,
      "loss": 2.0459,
      "step": 123091
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.9726485013961792,
      "learning_rate": 0.00026778565171184973,
      "loss": 2.2507,
      "step": 123092
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.1036993265151978,
      "learning_rate": 0.0002677815848046019,
      "loss": 2.332,
      "step": 123093
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.8871333003044128,
      "learning_rate": 0.0002677775179033441,
      "loss": 2.0697,
      "step": 123094
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.9974956512451172,
      "learning_rate": 0.00026777345100807714,
      "loss": 2.0603,
      "step": 123095
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.004105806350708,
      "learning_rate": 0.0002677693841188017,
      "loss": 2.1122,
      "step": 123096
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.9666306972503662,
      "learning_rate": 0.0002677653172355185,
      "loss": 1.9466,
      "step": 123097
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.9932437539100647,
      "learning_rate": 0.0002677612503582283,
      "loss": 2.0393,
      "step": 123098
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.064861536026001,
      "learning_rate": 0.0002677571834869319,
      "loss": 2.2406,
      "step": 123099
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.96209317445755,
      "learning_rate": 0.00026775311662163,
      "loss": 1.9165,
      "step": 123100
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.0744026899337769,
      "learning_rate": 0.0002677490497623235,
      "loss": 2.1207,
      "step": 123101
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.9017185568809509,
      "learning_rate": 0.0002677449829090131,
      "loss": 2.1621,
      "step": 123102
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.964191734790802,
      "learning_rate": 0.00026774091606169946,
      "loss": 1.9629,
      "step": 123103
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.9541760683059692,
      "learning_rate": 0.00026773684922038333,
      "loss": 2.2518,
      "step": 123104
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.8789827823638916,
      "learning_rate": 0.0002677327823850656,
      "loss": 2.1883,
      "step": 123105
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.0371845960617065,
      "learning_rate": 0.00026772871555574693,
      "loss": 2.0592,
      "step": 123106
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.0426414012908936,
      "learning_rate": 0.0002677246487324281,
      "loss": 2.1274,
      "step": 123107
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.9581799507141113,
      "learning_rate": 0.00026772058191510995,
      "loss": 2.0772,
      "step": 123108
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.8599970936775208,
      "learning_rate": 0.00026771651510379304,
      "loss": 2.2352,
      "step": 123109
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.9497809410095215,
      "learning_rate": 0.00026771244829847823,
      "loss": 1.9399,
      "step": 123110
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.909439206123352,
      "learning_rate": 0.00026770838149916637,
      "loss": 2.2013,
      "step": 123111
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.9452751874923706,
      "learning_rate": 0.000267704314705858,
      "loss": 2.2418,
      "step": 123112
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.0416815280914307,
      "learning_rate": 0.0002677002479185541,
      "loss": 2.0374,
      "step": 123113
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.9789796471595764,
      "learning_rate": 0.00026769618113725536,
      "loss": 2.0908,
      "step": 123114
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.048254132270813,
      "learning_rate": 0.0002676921143619626,
      "loss": 1.97,
      "step": 123115
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.9829185009002686,
      "learning_rate": 0.0002676880475926763,
      "loss": 2.2318,
      "step": 123116
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.9718544483184814,
      "learning_rate": 0.00026768398082939747,
      "loss": 2.0082,
      "step": 123117
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.9283666014671326,
      "learning_rate": 0.0002676799140721268,
      "loss": 2.3251,
      "step": 123118
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.9065619111061096,
      "learning_rate": 0.000267675847320865,
      "loss": 2.0929,
      "step": 123119
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.9935974478721619,
      "learning_rate": 0.0002676717805756129,
      "loss": 1.9907,
      "step": 123120
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.8830931782722473,
      "learning_rate": 0.00026766771383637133,
      "loss": 1.9223,
      "step": 123121
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.8894332051277161,
      "learning_rate": 0.00026766364710314084,
      "loss": 1.8525,
      "step": 123122
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.9430307745933533,
      "learning_rate": 0.0002676595803759223,
      "loss": 2.0447,
      "step": 123123
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.0641292333602905,
      "learning_rate": 0.0002676555136547164,
      "loss": 2.2404,
      "step": 123124
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.9376879930496216,
      "learning_rate": 0.000267651446939524,
      "loss": 2.0153,
      "step": 123125
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.9221282601356506,
      "learning_rate": 0.0002676473802303458,
      "loss": 2.3512,
      "step": 123126
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.0028551816940308,
      "learning_rate": 0.0002676433135271827,
      "loss": 2.1122,
      "step": 123127
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.9107828736305237,
      "learning_rate": 0.00026763924683003516,
      "loss": 2.4187,
      "step": 123128
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.136857271194458,
      "learning_rate": 0.00026763518013890413,
      "loss": 2.106,
      "step": 123129
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.9285087585449219,
      "learning_rate": 0.00026763111345379025,
      "loss": 1.9834,
      "step": 123130
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.0471925735473633,
      "learning_rate": 0.00026762704677469446,
      "loss": 2.382,
      "step": 123131
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.9474270343780518,
      "learning_rate": 0.00026762298010161734,
      "loss": 2.0718,
      "step": 123132
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.8643763065338135,
      "learning_rate": 0.0002676189134345599,
      "loss": 1.9532,
      "step": 123133
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.9088820219039917,
      "learning_rate": 0.00026761484677352256,
      "loss": 2.1763,
      "step": 123134
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.991567075252533,
      "learning_rate": 0.00026761078011850625,
      "loss": 2.2267,
      "step": 123135
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.0211418867111206,
      "learning_rate": 0.0002676067134695117,
      "loss": 1.8707,
      "step": 123136
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.8887181282043457,
      "learning_rate": 0.0002676026468265397,
      "loss": 2.1164,
      "step": 123137
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.8617904186248779,
      "learning_rate": 0.0002675985801895909,
      "loss": 1.8231,
      "step": 123138
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.0022872686386108,
      "learning_rate": 0.0002675945135586663,
      "loss": 1.9347,
      "step": 123139
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.9114458560943604,
      "learning_rate": 0.00026759044693376637,
      "loss": 2.1157,
      "step": 123140
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.0002533197402954,
      "learning_rate": 0.000267586380314892,
      "loss": 2.371,
      "step": 123141
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.9812560677528381,
      "learning_rate": 0.00026758231370204393,
      "loss": 2.2408,
      "step": 123142
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.9518930315971375,
      "learning_rate": 0.0002675782470952229,
      "loss": 2.1328,
      "step": 123143
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.0155342817306519,
      "learning_rate": 0.0002675741804944297,
      "loss": 1.9326,
      "step": 123144
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.9367573857307434,
      "learning_rate": 0.0002675701138996652,
      "loss": 2.1015,
      "step": 123145
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.9537933468818665,
      "learning_rate": 0.0002675660473109299,
      "loss": 2.3159,
      "step": 123146
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.9497426748275757,
      "learning_rate": 0.0002675619807282247,
      "loss": 2.2926,
      "step": 123147
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.0575004816055298,
      "learning_rate": 0.00026755791415155033,
      "loss": 2.0424,
      "step": 123148
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.9129568934440613,
      "learning_rate": 0.00026755384758090755,
      "loss": 2.1387,
      "step": 123149
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.971382737159729,
      "learning_rate": 0.0002675497810162971,
      "loss": 1.9939,
      "step": 123150
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.9465569853782654,
      "learning_rate": 0.00026754571445771993,
      "loss": 2.0027,
      "step": 123151
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.9764940142631531,
      "learning_rate": 0.0002675416479051765,
      "loss": 2.219,
      "step": 123152
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.0468873977661133,
      "learning_rate": 0.0002675375813586676,
      "loss": 1.9222,
      "step": 123153
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.9461231827735901,
      "learning_rate": 0.0002675335148181942,
      "loss": 2.0659,
      "step": 123154
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.9586198925971985,
      "learning_rate": 0.0002675294482837568,
      "loss": 2.1989,
      "step": 123155
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.259089469909668,
      "learning_rate": 0.0002675253817553564,
      "loss": 1.9744,
      "step": 123156
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.1107373237609863,
      "learning_rate": 0.0002675213152329937,
      "loss": 2.1001,
      "step": 123157
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.9480693340301514,
      "learning_rate": 0.0002675172487166694,
      "loss": 2.1343,
      "step": 123158
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.9495393633842468,
      "learning_rate": 0.00026751318220638413,
      "loss": 2.2473,
      "step": 123159
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.9956831336021423,
      "learning_rate": 0.00026750911570213876,
      "loss": 2.0119,
      "step": 123160
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.9881259202957153,
      "learning_rate": 0.00026750504920393413,
      "loss": 2.2513,
      "step": 123161
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.005744218826294,
      "learning_rate": 0.0002675009827117709,
      "loss": 2.0071,
      "step": 123162
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.9266431331634521,
      "learning_rate": 0.0002674969162256499,
      "loss": 2.2448,
      "step": 123163
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.9326281547546387,
      "learning_rate": 0.0002674928497455718,
      "loss": 2.1473,
      "step": 123164
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.9843795895576477,
      "learning_rate": 0.00026748878327153744,
      "loss": 2.1619,
      "step": 123165
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.9410757422447205,
      "learning_rate": 0.00026748471680354747,
      "loss": 2.2402,
      "step": 123166
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.001715064048767,
      "learning_rate": 0.00026748065034160273,
      "loss": 1.8805,
      "step": 123167
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.0429085493087769,
      "learning_rate": 0.00026747658388570395,
      "loss": 2.2294,
      "step": 123168
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.9928285479545593,
      "learning_rate": 0.00026747251743585186,
      "loss": 2.4751,
      "step": 123169
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.9988711476325989,
      "learning_rate": 0.0002674684509920473,
      "loss": 2.2372,
      "step": 123170
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.0263713598251343,
      "learning_rate": 0.0002674643845542909,
      "loss": 1.9713,
      "step": 123171
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.0422080755233765,
      "learning_rate": 0.0002674603181225836,
      "loss": 2.2191,
      "step": 123172
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.9856840968132019,
      "learning_rate": 0.0002674562516969259,
      "loss": 2.0321,
      "step": 123173
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.020153522491455,
      "learning_rate": 0.0002674521852773187,
      "loss": 1.9383,
      "step": 123174
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.1756540536880493,
      "learning_rate": 0.0002674481188637629,
      "loss": 2.1139,
      "step": 123175
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.0903346538543701,
      "learning_rate": 0.000267444052456259,
      "loss": 2.2902,
      "step": 123176
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.032156229019165,
      "learning_rate": 0.00026743998605480786,
      "loss": 2.1674,
      "step": 123177
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.9739201068878174,
      "learning_rate": 0.0002674359196594102,
      "loss": 2.0905,
      "step": 123178
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.0154380798339844,
      "learning_rate": 0.00026743185327006696,
      "loss": 2.2145,
      "step": 123179
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.1309008598327637,
      "learning_rate": 0.0002674277868867787,
      "loss": 2.1983,
      "step": 123180
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.0455279350280762,
      "learning_rate": 0.00026742372050954615,
      "loss": 2.1318,
      "step": 123181
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.0090463161468506,
      "learning_rate": 0.00026741965413837023,
      "loss": 2.1498,
      "step": 123182
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.9190671443939209,
      "learning_rate": 0.0002674155877732516,
      "loss": 1.9976,
      "step": 123183
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.0679212808609009,
      "learning_rate": 0.00026741152141419096,
      "loss": 2.2857,
      "step": 123184
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.9728928804397583,
      "learning_rate": 0.0002674074550611892,
      "loss": 2.2396,
      "step": 123185
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.9568657279014587,
      "learning_rate": 0.0002674033887142469,
      "loss": 2.0789,
      "step": 123186
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.9934651255607605,
      "learning_rate": 0.0002673993223733651,
      "loss": 2.1295,
      "step": 123187
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.9559128284454346,
      "learning_rate": 0.0002673952560385444,
      "loss": 2.127,
      "step": 123188
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.039209246635437,
      "learning_rate": 0.0002673911897097854,
      "loss": 2.3904,
      "step": 123189
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.9224405884742737,
      "learning_rate": 0.00026738712338708896,
      "loss": 1.9643,
      "step": 123190
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.017349362373352,
      "learning_rate": 0.00026738305707045594,
      "loss": 2.0236,
      "step": 123191
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.9433774948120117,
      "learning_rate": 0.00026737899075988703,
      "loss": 2.0533,
      "step": 123192
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.9554739594459534,
      "learning_rate": 0.0002673749244553829,
      "loss": 2.0876,
      "step": 123193
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.9367846846580505,
      "learning_rate": 0.0002673708581569446,
      "loss": 2.3448,
      "step": 123194
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.0715129375457764,
      "learning_rate": 0.0002673667918645725,
      "loss": 2.0138,
      "step": 123195
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.0281009674072266,
      "learning_rate": 0.0002673627255782676,
      "loss": 2.0301,
      "step": 123196
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.9852744340896606,
      "learning_rate": 0.0002673586592980305,
      "loss": 2.1834,
      "step": 123197
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.9286806583404541,
      "learning_rate": 0.0002673545930238621,
      "loss": 2.0733,
      "step": 123198
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.1641943454742432,
      "learning_rate": 0.00026735052675576305,
      "loss": 2.1902,
      "step": 123199
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.0544586181640625,
      "learning_rate": 0.0002673464604937343,
      "loss": 1.921,
      "step": 123200
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.9045801758766174,
      "learning_rate": 0.0002673423942377763,
      "loss": 2.1074,
      "step": 123201
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.9118999242782593,
      "learning_rate": 0.00026733832798789004,
      "loss": 2.3031,
      "step": 123202
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.9504221677780151,
      "learning_rate": 0.00026733426174407616,
      "loss": 2.0965,
      "step": 123203
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.9064980149269104,
      "learning_rate": 0.00026733019550633544,
      "loss": 2.2256,
      "step": 123204
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.9288786053657532,
      "learning_rate": 0.0002673261292746687,
      "loss": 2.2266,
      "step": 123205
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.0049843788146973,
      "learning_rate": 0.00026732206304907677,
      "loss": 2.2349,
      "step": 123206
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.0761367082595825,
      "learning_rate": 0.0002673179968295601,
      "loss": 2.2593,
      "step": 123207
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.012669324874878,
      "learning_rate": 0.0002673139306161197,
      "loss": 2.0866,
      "step": 123208
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.9753344058990479,
      "learning_rate": 0.0002673098644087562,
      "loss": 2.0865,
      "step": 123209
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.025647521018982,
      "learning_rate": 0.00026730579820747046,
      "loss": 1.9386,
      "step": 123210
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.8125468492507935,
      "learning_rate": 0.00026730173201226315,
      "loss": 2.1504,
      "step": 123211
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.951954185962677,
      "learning_rate": 0.0002672976658231352,
      "loss": 2.1333,
      "step": 123212
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.890893280506134,
      "learning_rate": 0.00026729359964008714,
      "loss": 2.1139,
      "step": 123213
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.0338631868362427,
      "learning_rate": 0.00026728953346311974,
      "loss": 2.3304,
      "step": 123214
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.0328730344772339,
      "learning_rate": 0.0002672854672922339,
      "loss": 2.2662,
      "step": 123215
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.0024956464767456,
      "learning_rate": 0.0002672814011274303,
      "loss": 2.1046,
      "step": 123216
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.9529760479927063,
      "learning_rate": 0.0002672773349687097,
      "loss": 2.0355,
      "step": 123217
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.0685778856277466,
      "learning_rate": 0.00026727326881607297,
      "loss": 2.3841,
      "step": 123218
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.9526157975196838,
      "learning_rate": 0.00026726920266952064,
      "loss": 2.0606,
      "step": 123219
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.9705117344856262,
      "learning_rate": 0.0002672651365290536,
      "loss": 2.0146,
      "step": 123220
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.8909071087837219,
      "learning_rate": 0.0002672610703946726,
      "loss": 2.1843,
      "step": 123221
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.0714665651321411,
      "learning_rate": 0.00026725700426637834,
      "loss": 2.0701,
      "step": 123222
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.9439160823822021,
      "learning_rate": 0.00026725293814417166,
      "loss": 2.0086,
      "step": 123223
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.8817958831787109,
      "learning_rate": 0.0002672488720280534,
      "loss": 2.1273,
      "step": 123224
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.0081355571746826,
      "learning_rate": 0.000267244805918024,
      "loss": 2.0751,
      "step": 123225
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.9927100539207458,
      "learning_rate": 0.00026724073981408453,
      "loss": 2.1819,
      "step": 123226
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.9094424843788147,
      "learning_rate": 0.00026723667371623554,
      "loss": 1.9477,
      "step": 123227
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.919119119644165,
      "learning_rate": 0.00026723260762447787,
      "loss": 2.1108,
      "step": 123228
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.0198609828948975,
      "learning_rate": 0.0002672285415388123,
      "loss": 2.2056,
      "step": 123229
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.935123860836029,
      "learning_rate": 0.0002672244754592397,
      "loss": 2.1418,
      "step": 123230
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.9961114525794983,
      "learning_rate": 0.00026722040938576057,
      "loss": 1.9519,
      "step": 123231
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.9598536491394043,
      "learning_rate": 0.0002672163433183757,
      "loss": 2.0645,
      "step": 123232
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.9434253573417664,
      "learning_rate": 0.000267212277257086,
      "loss": 2.2937,
      "step": 123233
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.9613272547721863,
      "learning_rate": 0.00026720821120189214,
      "loss": 1.9862,
      "step": 123234
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.9712271690368652,
      "learning_rate": 0.00026720414515279495,
      "loss": 2.0177,
      "step": 123235
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.9755947589874268,
      "learning_rate": 0.00026720007910979515,
      "loss": 2.0419,
      "step": 123236
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.9724735617637634,
      "learning_rate": 0.00026719601307289344,
      "loss": 2.1267,
      "step": 123237
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.1755801439285278,
      "learning_rate": 0.0002671919470420906,
      "loss": 2.3228,
      "step": 123238
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.0265504121780396,
      "learning_rate": 0.00026718788101738736,
      "loss": 2.0681,
      "step": 123239
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.9764770865440369,
      "learning_rate": 0.0002671838149987845,
      "loss": 2.1961,
      "step": 123240
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.9266708493232727,
      "learning_rate": 0.0002671797489862828,
      "loss": 2.0013,
      "step": 123241
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.1373158693313599,
      "learning_rate": 0.000267175682979883,
      "loss": 2.0252,
      "step": 123242
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.9412537813186646,
      "learning_rate": 0.00026717161697958604,
      "loss": 2.0844,
      "step": 123243
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.952502965927124,
      "learning_rate": 0.00026716755098539236,
      "loss": 2.029,
      "step": 123244
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.9780222177505493,
      "learning_rate": 0.0002671634849973028,
      "loss": 2.1614,
      "step": 123245
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.9675350785255432,
      "learning_rate": 0.0002671594190153182,
      "loss": 2.1338,
      "step": 123246
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.9322434663772583,
      "learning_rate": 0.0002671553530394393,
      "loss": 2.1634,
      "step": 123247
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.9864949584007263,
      "learning_rate": 0.0002671512870696668,
      "loss": 1.9194,
      "step": 123248
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.9604451060295105,
      "learning_rate": 0.00026714722110600164,
      "loss": 1.9928,
      "step": 123249
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.9794963598251343,
      "learning_rate": 0.0002671431551484443,
      "loss": 2.236,
      "step": 123250
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.8895854949951172,
      "learning_rate": 0.00026713908919699566,
      "loss": 2.1951,
      "step": 123251
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.9203284978866577,
      "learning_rate": 0.0002671350232516565,
      "loss": 2.1455,
      "step": 123252
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.9841841459274292,
      "learning_rate": 0.00026713095731242754,
      "loss": 2.0488,
      "step": 123253
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.3968515396118164,
      "learning_rate": 0.0002671268913793096,
      "loss": 2.2273,
      "step": 123254
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.0751382112503052,
      "learning_rate": 0.0002671228254523034,
      "loss": 2.3001,
      "step": 123255
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.8721257448196411,
      "learning_rate": 0.00026711875953140966,
      "loss": 1.8499,
      "step": 123256
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.9970026016235352,
      "learning_rate": 0.0002671146936166292,
      "loss": 2.1886,
      "step": 123257
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.0874322652816772,
      "learning_rate": 0.0002671106277079627,
      "loss": 2.0007,
      "step": 123258
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.1422661542892456,
      "learning_rate": 0.0002671065618054109,
      "loss": 2.1919,
      "step": 123259
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.0121582746505737,
      "learning_rate": 0.0002671024959089747,
      "loss": 2.2464,
      "step": 123260
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.874354898929596,
      "learning_rate": 0.00026709843001865475,
      "loss": 2.1031,
      "step": 123261
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.0258971452713013,
      "learning_rate": 0.0002670943641344518,
      "loss": 2.1397,
      "step": 123262
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.9185806512832642,
      "learning_rate": 0.0002670902982563667,
      "loss": 2.0704,
      "step": 123263
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.930663526058197,
      "learning_rate": 0.0002670862323844001,
      "loss": 2.2808,
      "step": 123264
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.9783392548561096,
      "learning_rate": 0.0002670821665185527,
      "loss": 2.1368,
      "step": 123265
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.1298342943191528,
      "learning_rate": 0.0002670781006588254,
      "loss": 2.2075,
      "step": 123266
    },
    {
      "epoch": 1.61,
      "grad_norm": 0.9778429865837097,
      "learning_rate": 0.000267074034805219,
      "loss": 2.2729,
      "step": 123267
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.032152533531189,
      "learning_rate": 0.000267069968957734,
      "loss": 1.7871,
      "step": 123268
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.0665595531463623,
      "learning_rate": 0.00026706590311637143,
      "loss": 2.048,
      "step": 123269
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.0301727056503296,
      "learning_rate": 0.0002670618372811319,
      "loss": 2.2147,
      "step": 123270
    },
    {
      "epoch": 1.61,
      "grad_norm": 0.9319379925727844,
      "learning_rate": 0.0002670577714520162,
      "loss": 1.9218,
      "step": 123271
    },
    {
      "epoch": 1.61,
      "grad_norm": 0.9583473205566406,
      "learning_rate": 0.00026705370562902505,
      "loss": 2.2561,
      "step": 123272
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.0005513429641724,
      "learning_rate": 0.0002670496398121593,
      "loss": 2.2481,
      "step": 123273
    },
    {
      "epoch": 1.61,
      "grad_norm": 0.9770312309265137,
      "learning_rate": 0.0002670455740014196,
      "loss": 2.1056,
      "step": 123274
    },
    {
      "epoch": 1.61,
      "grad_norm": 0.9900550246238708,
      "learning_rate": 0.0002670415081968067,
      "loss": 2.136,
      "step": 123275
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.0607030391693115,
      "learning_rate": 0.00026703744239832145,
      "loss": 2.25,
      "step": 123276
    },
    {
      "epoch": 1.61,
      "grad_norm": 0.979597806930542,
      "learning_rate": 0.0002670333766059646,
      "loss": 2.1176,
      "step": 123277
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.0391347408294678,
      "learning_rate": 0.00026702931081973687,
      "loss": 2.1642,
      "step": 123278
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.0643608570098877,
      "learning_rate": 0.000267025245039639,
      "loss": 1.9084,
      "step": 123279
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.011458158493042,
      "learning_rate": 0.00026702117926567174,
      "loss": 2.126,
      "step": 123280
    },
    {
      "epoch": 1.61,
      "grad_norm": 0.9968714714050293,
      "learning_rate": 0.00026701711349783584,
      "loss": 2.2516,
      "step": 123281
    },
    {
      "epoch": 1.61,
      "grad_norm": 0.9104875326156616,
      "learning_rate": 0.0002670130477361321,
      "loss": 2.3049,
      "step": 123282
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.0067676305770874,
      "learning_rate": 0.0002670089819805612,
      "loss": 2.1309,
      "step": 123283
    },
    {
      "epoch": 1.61,
      "grad_norm": 0.9137740731239319,
      "learning_rate": 0.00026700491623112406,
      "loss": 2.2628,
      "step": 123284
    },
    {
      "epoch": 1.61,
      "grad_norm": 0.937356173992157,
      "learning_rate": 0.0002670008504878214,
      "loss": 2.0079,
      "step": 123285
    },
    {
      "epoch": 1.61,
      "grad_norm": 0.9671428203582764,
      "learning_rate": 0.00026699678475065376,
      "loss": 2.1299,
      "step": 123286
    },
    {
      "epoch": 1.61,
      "grad_norm": 0.9541056156158447,
      "learning_rate": 0.000266992719019622,
      "loss": 2.0769,
      "step": 123287
    },
    {
      "epoch": 1.61,
      "grad_norm": 0.971100926399231,
      "learning_rate": 0.00026698865329472703,
      "loss": 2.0652,
      "step": 123288
    },
    {
      "epoch": 1.61,
      "grad_norm": 0.9533862471580505,
      "learning_rate": 0.0002669845875759694,
      "loss": 2.0901,
      "step": 123289
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.075109601020813,
      "learning_rate": 0.00026698052186335,
      "loss": 2.1904,
      "step": 123290
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.0566425323486328,
      "learning_rate": 0.00026697645615686964,
      "loss": 2.2848,
      "step": 123291
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.0141092538833618,
      "learning_rate": 0.0002669723904565289,
      "loss": 2.2682,
      "step": 123292
    },
    {
      "epoch": 1.61,
      "grad_norm": 0.9947633147239685,
      "learning_rate": 0.00026696832476232854,
      "loss": 1.9785,
      "step": 123293
    },
    {
      "epoch": 1.61,
      "grad_norm": 0.9551048874855042,
      "learning_rate": 0.0002669642590742695,
      "loss": 2.3003,
      "step": 123294
    },
    {
      "epoch": 1.61,
      "grad_norm": 0.9827767014503479,
      "learning_rate": 0.0002669601933923524,
      "loss": 2.3872,
      "step": 123295
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.117521047592163,
      "learning_rate": 0.0002669561277165779,
      "loss": 2.3478,
      "step": 123296
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.1561994552612305,
      "learning_rate": 0.00026695206204694716,
      "loss": 2.2279,
      "step": 123297
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.1442079544067383,
      "learning_rate": 0.00026694799638346046,
      "loss": 2.1323,
      "step": 123298
    },
    {
      "epoch": 1.61,
      "grad_norm": 0.9290347099304199,
      "learning_rate": 0.00026694393072611874,
      "loss": 2.3046,
      "step": 123299
    },
    {
      "epoch": 1.61,
      "grad_norm": 0.9569382667541504,
      "learning_rate": 0.00026693986507492283,
      "loss": 2.1268,
      "step": 123300
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.025162696838379,
      "learning_rate": 0.00026693579942987336,
      "loss": 2.2191,
      "step": 123301
    },
    {
      "epoch": 1.61,
      "grad_norm": 0.9810936450958252,
      "learning_rate": 0.00026693173379097116,
      "loss": 2.1271,
      "step": 123302
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.0933830738067627,
      "learning_rate": 0.0002669276681582171,
      "loss": 2.0006,
      "step": 123303
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.1375447511672974,
      "learning_rate": 0.0002669236025316117,
      "loss": 2.1132,
      "step": 123304
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.0171170234680176,
      "learning_rate": 0.00026691953691115584,
      "loss": 1.9191,
      "step": 123305
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.0497463941574097,
      "learning_rate": 0.0002669154712968503,
      "loss": 2.3143,
      "step": 123306
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.181536078453064,
      "learning_rate": 0.0002669114056886957,
      "loss": 2.1724,
      "step": 123307
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.137880563735962,
      "learning_rate": 0.00026690734008669295,
      "loss": 2.1485,
      "step": 123308
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.0057992935180664,
      "learning_rate": 0.0002669032744908427,
      "loss": 2.2177,
      "step": 123309
    },
    {
      "epoch": 1.61,
      "grad_norm": 0.9132404327392578,
      "learning_rate": 0.0002668992089011459,
      "loss": 1.9477,
      "step": 123310
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.0795786380767822,
      "learning_rate": 0.00026689514331760313,
      "loss": 2.1984,
      "step": 123311
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.0190702676773071,
      "learning_rate": 0.00026689107774021504,
      "loss": 2.1945,
      "step": 123312
    },
    {
      "epoch": 1.61,
      "grad_norm": 0.9164050221443176,
      "learning_rate": 0.0002668870121689826,
      "loss": 2.3806,
      "step": 123313
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.0785398483276367,
      "learning_rate": 0.0002668829466039065,
      "loss": 2.0579,
      "step": 123314
    },
    {
      "epoch": 1.61,
      "grad_norm": 0.957798957824707,
      "learning_rate": 0.00026687888104498747,
      "loss": 2.0292,
      "step": 123315
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.0636897087097168,
      "learning_rate": 0.0002668748154922264,
      "loss": 1.9829,
      "step": 123316
    },
    {
      "epoch": 1.61,
      "grad_norm": 0.9058950543403625,
      "learning_rate": 0.0002668707499456238,
      "loss": 2.1947,
      "step": 123317
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.0083391666412354,
      "learning_rate": 0.0002668666844051805,
      "loss": 2.1205,
      "step": 123318
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.002632737159729,
      "learning_rate": 0.0002668626188708974,
      "loss": 2.116,
      "step": 123319
    },
    {
      "epoch": 1.61,
      "grad_norm": 0.9854109883308411,
      "learning_rate": 0.0002668585533427751,
      "loss": 1.999,
      "step": 123320
    },
    {
      "epoch": 1.61,
      "grad_norm": 0.9927729964256287,
      "learning_rate": 0.00026685448782081447,
      "loss": 2.0481,
      "step": 123321
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.115442156791687,
      "learning_rate": 0.0002668504223050163,
      "loss": 2.0647,
      "step": 123322
    },
    {
      "epoch": 1.61,
      "grad_norm": 0.9769071936607361,
      "learning_rate": 0.00026684635679538116,
      "loss": 2.1352,
      "step": 123323
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.0643011331558228,
      "learning_rate": 0.00026684229129190996,
      "loss": 2.0589,
      "step": 123324
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.0381025075912476,
      "learning_rate": 0.0002668382257946033,
      "loss": 2.0103,
      "step": 123325
    },
    {
      "epoch": 1.61,
      "grad_norm": 0.9777427911758423,
      "learning_rate": 0.0002668341603034621,
      "loss": 2.131,
      "step": 123326
    },
    {
      "epoch": 1.61,
      "grad_norm": 0.8682481050491333,
      "learning_rate": 0.00026683009481848706,
      "loss": 1.9818,
      "step": 123327
    },
    {
      "epoch": 1.61,
      "grad_norm": 0.967593789100647,
      "learning_rate": 0.000266826029339679,
      "loss": 2.0883,
      "step": 123328
    },
    {
      "epoch": 1.61,
      "grad_norm": 0.9317439198493958,
      "learning_rate": 0.0002668219638670386,
      "loss": 1.8847,
      "step": 123329
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.1299927234649658,
      "learning_rate": 0.0002668178984005665,
      "loss": 2.2336,
      "step": 123330
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.0099235773086548,
      "learning_rate": 0.0002668138329402636,
      "loss": 2.2172,
      "step": 123331
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.0485502481460571,
      "learning_rate": 0.00026680976748613065,
      "loss": 2.1042,
      "step": 123332
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.0372734069824219,
      "learning_rate": 0.00026680570203816843,
      "loss": 1.9752,
      "step": 123333
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.0318734645843506,
      "learning_rate": 0.00026680163659637777,
      "loss": 2.1433,
      "step": 123334
    },
    {
      "epoch": 1.61,
      "grad_norm": 0.8863311409950256,
      "learning_rate": 0.0002667975711607592,
      "loss": 2.1707,
      "step": 123335
    },
    {
      "epoch": 1.61,
      "grad_norm": 0.9527489542961121,
      "learning_rate": 0.0002667935057313135,
      "loss": 2.0334,
      "step": 123336
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.100709319114685,
      "learning_rate": 0.00026678944030804157,
      "loss": 2.3191,
      "step": 123337
    },
    {
      "epoch": 1.61,
      "grad_norm": 0.9750211238861084,
      "learning_rate": 0.00026678537489094415,
      "loss": 2.3967,
      "step": 123338
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.0151859521865845,
      "learning_rate": 0.0002667813094800219,
      "loss": 2.3065,
      "step": 123339
    },
    {
      "epoch": 1.61,
      "grad_norm": 0.888677179813385,
      "learning_rate": 0.0002667772440752757,
      "loss": 2.1242,
      "step": 123340
    },
    {
      "epoch": 1.61,
      "grad_norm": 0.9965661764144897,
      "learning_rate": 0.00026677317867670623,
      "loss": 2.2058,
      "step": 123341
    },
    {
      "epoch": 1.61,
      "grad_norm": 0.8907183408737183,
      "learning_rate": 0.00026676911328431425,
      "loss": 2.1669,
      "step": 123342
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.08344566822052,
      "learning_rate": 0.00026676504789810044,
      "loss": 2.1042,
      "step": 123343
    },
    {
      "epoch": 1.61,
      "grad_norm": 0.962687611579895,
      "learning_rate": 0.0002667609825180657,
      "loss": 2.0427,
      "step": 123344
    },
    {
      "epoch": 1.61,
      "grad_norm": 0.9297160506248474,
      "learning_rate": 0.00026675691714421067,
      "loss": 2.1056,
      "step": 123345
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.0250837802886963,
      "learning_rate": 0.0002667528517765363,
      "loss": 2.3187,
      "step": 123346
    },
    {
      "epoch": 1.61,
      "grad_norm": 0.9521185159683228,
      "learning_rate": 0.00026674878641504303,
      "loss": 2.121,
      "step": 123347
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.1016534566879272,
      "learning_rate": 0.00026674472105973185,
      "loss": 2.2613,
      "step": 123348
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.0013006925582886,
      "learning_rate": 0.0002667406557106035,
      "loss": 2.2098,
      "step": 123349
    },
    {
      "epoch": 1.61,
      "grad_norm": 0.9222647547721863,
      "learning_rate": 0.0002667365903676587,
      "loss": 2.0268,
      "step": 123350
    },
    {
      "epoch": 1.61,
      "grad_norm": 0.9785358905792236,
      "learning_rate": 0.0002667325250308981,
      "loss": 2.135,
      "step": 123351
    },
    {
      "epoch": 1.61,
      "grad_norm": 0.9016190767288208,
      "learning_rate": 0.00026672845970032266,
      "loss": 2.0875,
      "step": 123352
    },
    {
      "epoch": 1.61,
      "grad_norm": 0.9235066175460815,
      "learning_rate": 0.00026672439437593293,
      "loss": 2.2148,
      "step": 123353
    },
    {
      "epoch": 1.61,
      "grad_norm": 0.9722908735275269,
      "learning_rate": 0.00026672032905772976,
      "loss": 2.1045,
      "step": 123354
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.0021036863327026,
      "learning_rate": 0.000266716263745714,
      "loss": 2.2769,
      "step": 123355
    },
    {
      "epoch": 1.61,
      "grad_norm": 0.8910154700279236,
      "learning_rate": 0.00026671219843988627,
      "loss": 1.9305,
      "step": 123356
    },
    {
      "epoch": 1.61,
      "grad_norm": 0.9351691603660583,
      "learning_rate": 0.00026670813314024736,
      "loss": 2.3493,
      "step": 123357
    },
    {
      "epoch": 1.61,
      "grad_norm": 0.9038470387458801,
      "learning_rate": 0.0002667040678467981,
      "loss": 2.1239,
      "step": 123358
    },
    {
      "epoch": 1.61,
      "grad_norm": 0.9720624089241028,
      "learning_rate": 0.0002667000025595391,
      "loss": 2.3155,
      "step": 123359
    },
    {
      "epoch": 1.61,
      "grad_norm": 0.9562878608703613,
      "learning_rate": 0.0002666959372784712,
      "loss": 2.06,
      "step": 123360
    },
    {
      "epoch": 1.61,
      "grad_norm": 0.9915727376937866,
      "learning_rate": 0.00026669187200359514,
      "loss": 2.3321,
      "step": 123361
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.5656028985977173,
      "learning_rate": 0.0002666878067349117,
      "loss": 1.9248,
      "step": 123362
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.0382217168807983,
      "learning_rate": 0.00026668374147242175,
      "loss": 2.1878,
      "step": 123363
    },
    {
      "epoch": 1.61,
      "grad_norm": 0.9581759572029114,
      "learning_rate": 0.0002666796762161258,
      "loss": 2.0617,
      "step": 123364
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.0255937576293945,
      "learning_rate": 0.00026667561096602475,
      "loss": 2.0163,
      "step": 123365
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.1502569913864136,
      "learning_rate": 0.00026667154572211936,
      "loss": 2.1465,
      "step": 123366
    },
    {
      "epoch": 1.61,
      "grad_norm": 0.9228189587593079,
      "learning_rate": 0.0002666674804844103,
      "loss": 2.2183,
      "step": 123367
    },
    {
      "epoch": 1.61,
      "grad_norm": 0.9084939360618591,
      "learning_rate": 0.00026666341525289836,
      "loss": 1.8896,
      "step": 123368
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.1043944358825684,
      "learning_rate": 0.0002666593500275844,
      "loss": 2.1795,
      "step": 123369
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.1824688911437988,
      "learning_rate": 0.00026665528480846915,
      "loss": 1.9253,
      "step": 123370
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.0401605367660522,
      "learning_rate": 0.00026665121959555324,
      "loss": 2.2852,
      "step": 123371
    },
    {
      "epoch": 1.61,
      "grad_norm": 0.9066519737243652,
      "learning_rate": 0.0002666471543888375,
      "loss": 2.1401,
      "step": 123372
    },
    {
      "epoch": 1.61,
      "grad_norm": 0.9822170734405518,
      "learning_rate": 0.00026664308918832264,
      "loss": 2.0887,
      "step": 123373
    },
    {
      "epoch": 1.61,
      "grad_norm": 0.9125945568084717,
      "learning_rate": 0.00026663902399400946,
      "loss": 2.0382,
      "step": 123374
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.3506219387054443,
      "learning_rate": 0.00026663495880589874,
      "loss": 2.2471,
      "step": 123375
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.0002683401107788,
      "learning_rate": 0.00026663089362399117,
      "loss": 2.1706,
      "step": 123376
    },
    {
      "epoch": 1.61,
      "grad_norm": 0.947636067867279,
      "learning_rate": 0.00026662682844828773,
      "loss": 2.1865,
      "step": 123377
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.024539828300476,
      "learning_rate": 0.00026662276327878886,
      "loss": 2.1988,
      "step": 123378
    },
    {
      "epoch": 1.61,
      "grad_norm": 0.881359875202179,
      "learning_rate": 0.0002666186981154954,
      "loss": 2.1904,
      "step": 123379
    },
    {
      "epoch": 1.61,
      "grad_norm": 0.9280999302864075,
      "learning_rate": 0.0002666146329584082,
      "loss": 2.1075,
      "step": 123380
    },
    {
      "epoch": 1.61,
      "grad_norm": 0.9666537642478943,
      "learning_rate": 0.00026661056780752796,
      "loss": 2.1401,
      "step": 123381
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.0842852592468262,
      "learning_rate": 0.0002666065026628555,
      "loss": 2.0994,
      "step": 123382
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.0207102298736572,
      "learning_rate": 0.00026660243752439156,
      "loss": 2.376,
      "step": 123383
    },
    {
      "epoch": 1.61,
      "grad_norm": 0.9084488153457642,
      "learning_rate": 0.00026659837239213676,
      "loss": 2.0163,
      "step": 123384
    },
    {
      "epoch": 1.61,
      "grad_norm": 0.9417034983634949,
      "learning_rate": 0.00026659430726609195,
      "loss": 1.9909,
      "step": 123385
    },
    {
      "epoch": 1.61,
      "grad_norm": 0.9250242114067078,
      "learning_rate": 0.0002665902421462579,
      "loss": 2.0747,
      "step": 123386
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.0308129787445068,
      "learning_rate": 0.0002665861770326353,
      "loss": 2.0745,
      "step": 123387
    },
    {
      "epoch": 1.61,
      "grad_norm": 0.8439181447029114,
      "learning_rate": 0.000266582111925225,
      "loss": 2.0633,
      "step": 123388
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.0199599266052246,
      "learning_rate": 0.0002665780468240279,
      "loss": 2.0904,
      "step": 123389
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.0613927841186523,
      "learning_rate": 0.0002665739817290444,
      "loss": 2.273,
      "step": 123390
    },
    {
      "epoch": 1.61,
      "grad_norm": 0.9261276125907898,
      "learning_rate": 0.00026656991664027543,
      "loss": 2.0289,
      "step": 123391
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.091444969177246,
      "learning_rate": 0.0002665658515577217,
      "loss": 2.202,
      "step": 123392
    },
    {
      "epoch": 1.61,
      "grad_norm": 0.9118191599845886,
      "learning_rate": 0.0002665617864813841,
      "loss": 2.156,
      "step": 123393
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.0467370748519897,
      "learning_rate": 0.0002665577214112632,
      "loss": 2.0973,
      "step": 123394
    },
    {
      "epoch": 1.61,
      "grad_norm": 0.9569925665855408,
      "learning_rate": 0.00026655365634736003,
      "loss": 2.0267,
      "step": 123395
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.1312369108200073,
      "learning_rate": 0.00026654959128967503,
      "loss": 2.0936,
      "step": 123396
    },
    {
      "epoch": 1.61,
      "grad_norm": 0.9758393168449402,
      "learning_rate": 0.0002665455262382091,
      "loss": 2.2354,
      "step": 123397
    },
    {
      "epoch": 1.61,
      "grad_norm": 0.968122124671936,
      "learning_rate": 0.00026654146119296296,
      "loss": 2.0513,
      "step": 123398
    },
    {
      "epoch": 1.61,
      "grad_norm": 0.9314714670181274,
      "learning_rate": 0.0002665373961539374,
      "loss": 2.1134,
      "step": 123399
    },
    {
      "epoch": 1.61,
      "grad_norm": 0.8968801498413086,
      "learning_rate": 0.0002665333311211332,
      "loss": 2.1206,
      "step": 123400
    },
    {
      "epoch": 1.61,
      "grad_norm": 0.934553861618042,
      "learning_rate": 0.00026652926609455123,
      "loss": 2.3043,
      "step": 123401
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.1582399606704712,
      "learning_rate": 0.00026652520107419195,
      "loss": 2.2214,
      "step": 123402
    },
    {
      "epoch": 1.61,
      "grad_norm": 0.8936701416969299,
      "learning_rate": 0.0002665211360600562,
      "loss": 2.0624,
      "step": 123403
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.0044249296188354,
      "learning_rate": 0.0002665170710521449,
      "loss": 1.9974,
      "step": 123404
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.2327120304107666,
      "learning_rate": 0.00026651300605045865,
      "loss": 2.1254,
      "step": 123405
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.273930549621582,
      "learning_rate": 0.0002665089410549983,
      "loss": 1.9843,
      "step": 123406
    },
    {
      "epoch": 1.61,
      "grad_norm": 0.9968268871307373,
      "learning_rate": 0.00026650487606576467,
      "loss": 2.2123,
      "step": 123407
    },
    {
      "epoch": 1.61,
      "grad_norm": 0.9578575491905212,
      "learning_rate": 0.0002665008110827583,
      "loss": 2.1176,
      "step": 123408
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.142521619796753,
      "learning_rate": 0.00026649674610598,
      "loss": 2.02,
      "step": 123409
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.0864683389663696,
      "learning_rate": 0.00026649268113543073,
      "loss": 2.0623,
      "step": 123410
    },
    {
      "epoch": 1.61,
      "grad_norm": 0.9856134057044983,
      "learning_rate": 0.000266488616171111,
      "loss": 2.1312,
      "step": 123411
    },
    {
      "epoch": 1.61,
      "grad_norm": 0.9787420630455017,
      "learning_rate": 0.00026648455121302167,
      "loss": 2.0144,
      "step": 123412
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.22431218624115,
      "learning_rate": 0.0002664804862611637,
      "loss": 2.3104,
      "step": 123413
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.0107853412628174,
      "learning_rate": 0.0002664764213155374,
      "loss": 2.3206,
      "step": 123414
    },
    {
      "epoch": 1.61,
      "grad_norm": 0.9162219762802124,
      "learning_rate": 0.00026647235637614384,
      "loss": 1.926,
      "step": 123415
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.105945348739624,
      "learning_rate": 0.0002664682914429837,
      "loss": 2.1734,
      "step": 123416
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.0354888439178467,
      "learning_rate": 0.0002664642265160577,
      "loss": 2.1671,
      "step": 123417
    },
    {
      "epoch": 1.61,
      "grad_norm": 0.9595284461975098,
      "learning_rate": 0.00026646016159536664,
      "loss": 2.1089,
      "step": 123418
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.073029637336731,
      "learning_rate": 0.0002664560966809114,
      "loss": 2.044,
      "step": 123419
    },
    {
      "epoch": 1.61,
      "grad_norm": 0.9927406311035156,
      "learning_rate": 0.00026645203177269247,
      "loss": 2.2913,
      "step": 123420
    },
    {
      "epoch": 1.61,
      "grad_norm": 0.9556120038032532,
      "learning_rate": 0.0002664479668707108,
      "loss": 2.09,
      "step": 123421
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.0771225690841675,
      "learning_rate": 0.00026644390197496704,
      "loss": 2.0734,
      "step": 123422
    },
    {
      "epoch": 1.61,
      "grad_norm": 0.9782048463821411,
      "learning_rate": 0.00026643983708546197,
      "loss": 2.0485,
      "step": 123423
    },
    {
      "epoch": 1.61,
      "grad_norm": 0.9675590395927429,
      "learning_rate": 0.0002664357722021964,
      "loss": 2.2516,
      "step": 123424
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.1738851070404053,
      "learning_rate": 0.0002664317073251711,
      "loss": 1.8377,
      "step": 123425
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.0667874813079834,
      "learning_rate": 0.0002664276424543868,
      "loss": 2.2173,
      "step": 123426
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.0005499124526978,
      "learning_rate": 0.0002664235775898442,
      "loss": 2.2145,
      "step": 123427
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.0488659143447876,
      "learning_rate": 0.000266419512731544,
      "loss": 2.1415,
      "step": 123428
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.1266661882400513,
      "learning_rate": 0.0002664154478794871,
      "loss": 2.1381,
      "step": 123429
    },
    {
      "epoch": 1.61,
      "grad_norm": 0.8979871869087219,
      "learning_rate": 0.0002664113830336742,
      "loss": 2.2641,
      "step": 123430
    },
    {
      "epoch": 1.61,
      "grad_norm": 0.9230814576148987,
      "learning_rate": 0.0002664073181941062,
      "loss": 2.0856,
      "step": 123431
    },
    {
      "epoch": 1.61,
      "grad_norm": 0.9651448130607605,
      "learning_rate": 0.00026640325336078354,
      "loss": 2.1107,
      "step": 123432
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.0859274864196777,
      "learning_rate": 0.00026639918853370726,
      "loss": 1.9998,
      "step": 123433
    },
    {
      "epoch": 1.61,
      "grad_norm": 0.9029089212417603,
      "learning_rate": 0.0002663951237128779,
      "loss": 2.1543,
      "step": 123434
    },
    {
      "epoch": 1.61,
      "grad_norm": 0.9606338143348694,
      "learning_rate": 0.0002663910588982964,
      "loss": 2.0206,
      "step": 123435
    },
    {
      "epoch": 1.61,
      "grad_norm": 0.9643004536628723,
      "learning_rate": 0.0002663869940899633,
      "loss": 2.2266,
      "step": 123436
    },
    {
      "epoch": 1.61,
      "grad_norm": 0.9595897793769836,
      "learning_rate": 0.0002663829292878797,
      "loss": 2.2053,
      "step": 123437
    },
    {
      "epoch": 1.61,
      "grad_norm": 0.9672313332557678,
      "learning_rate": 0.00026637886449204604,
      "loss": 2.1573,
      "step": 123438
    },
    {
      "epoch": 1.61,
      "grad_norm": 0.9321942925453186,
      "learning_rate": 0.00026637479970246315,
      "loss": 2.1382,
      "step": 123439
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.012387990951538,
      "learning_rate": 0.00026637073491913183,
      "loss": 2.1095,
      "step": 123440
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.0600426197052002,
      "learning_rate": 0.0002663666701420529,
      "loss": 2.1708,
      "step": 123441
    },
    {
      "epoch": 1.61,
      "grad_norm": 0.9578834772109985,
      "learning_rate": 0.00026636260537122697,
      "loss": 1.8127,
      "step": 123442
    },
    {
      "epoch": 1.61,
      "grad_norm": 0.9055032134056091,
      "learning_rate": 0.00026635854060665483,
      "loss": 2.0326,
      "step": 123443
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.1082727909088135,
      "learning_rate": 0.0002663544758483374,
      "loss": 2.3527,
      "step": 123444
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.175135850906372,
      "learning_rate": 0.00026635041109627524,
      "loss": 2.2027,
      "step": 123445
    },
    {
      "epoch": 1.61,
      "grad_norm": 0.9286004900932312,
      "learning_rate": 0.0002663463463504691,
      "loss": 1.8378,
      "step": 123446
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.050342321395874,
      "learning_rate": 0.0002663422816109199,
      "loss": 2.0461,
      "step": 123447
    },
    {
      "epoch": 1.61,
      "grad_norm": 0.867294192314148,
      "learning_rate": 0.00026633821687762835,
      "loss": 2.0728,
      "step": 123448
    },
    {
      "epoch": 1.61,
      "grad_norm": 0.9879111647605896,
      "learning_rate": 0.0002663341521505951,
      "loss": 2.1633,
      "step": 123449
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.065271258354187,
      "learning_rate": 0.000266330087429821,
      "loss": 2.0989,
      "step": 123450
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.1721476316452026,
      "learning_rate": 0.0002663260227153067,
      "loss": 2.221,
      "step": 123451
    },
    {
      "epoch": 1.61,
      "grad_norm": 0.9209553599357605,
      "learning_rate": 0.00026632195800705305,
      "loss": 1.8748,
      "step": 123452
    },
    {
      "epoch": 1.61,
      "grad_norm": 0.9899447560310364,
      "learning_rate": 0.0002663178933050608,
      "loss": 2.1225,
      "step": 123453
    },
    {
      "epoch": 1.61,
      "grad_norm": 0.9421820640563965,
      "learning_rate": 0.0002663138286093307,
      "loss": 2.1138,
      "step": 123454
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.103848934173584,
      "learning_rate": 0.00026630976391986346,
      "loss": 2.2585,
      "step": 123455
    },
    {
      "epoch": 1.61,
      "grad_norm": 0.8393123149871826,
      "learning_rate": 0.00026630569923665994,
      "loss": 2.1145,
      "step": 123456
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.1439815759658813,
      "learning_rate": 0.0002663016345597208,
      "loss": 2.2539,
      "step": 123457
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.0932523012161255,
      "learning_rate": 0.00026629756988904674,
      "loss": 1.9774,
      "step": 123458
    },
    {
      "epoch": 1.61,
      "grad_norm": 0.9733784794807434,
      "learning_rate": 0.0002662935052246387,
      "loss": 2.1625,
      "step": 123459
    },
    {
      "epoch": 1.61,
      "grad_norm": 0.9629445672035217,
      "learning_rate": 0.00026628944056649725,
      "loss": 2.1321,
      "step": 123460
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.1420601606369019,
      "learning_rate": 0.00026628537591462327,
      "loss": 2.0254,
      "step": 123461
    },
    {
      "epoch": 1.61,
      "grad_norm": 0.9904094934463501,
      "learning_rate": 0.00026628131126901755,
      "loss": 2.1263,
      "step": 123462
    },
    {
      "epoch": 1.61,
      "grad_norm": 0.986355185508728,
      "learning_rate": 0.0002662772466296807,
      "loss": 2.1962,
      "step": 123463
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.004297137260437,
      "learning_rate": 0.0002662731819966135,
      "loss": 2.0435,
      "step": 123464
    },
    {
      "epoch": 1.61,
      "grad_norm": 0.9450936913490295,
      "learning_rate": 0.0002662691173698168,
      "loss": 2.1253,
      "step": 123465
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.0373083353042603,
      "learning_rate": 0.0002662650527492913,
      "loss": 2.0606,
      "step": 123466
    },
    {
      "epoch": 1.61,
      "grad_norm": 0.9445154070854187,
      "learning_rate": 0.00026626098813503776,
      "loss": 2.1498,
      "step": 123467
    },
    {
      "epoch": 1.61,
      "grad_norm": 0.9169037938117981,
      "learning_rate": 0.00026625692352705705,
      "loss": 2.1982,
      "step": 123468
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.1108572483062744,
      "learning_rate": 0.0002662528589253497,
      "loss": 2.3559,
      "step": 123469
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.218326210975647,
      "learning_rate": 0.0002662487943299166,
      "loss": 1.9886,
      "step": 123470
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.0484068393707275,
      "learning_rate": 0.00026624472974075844,
      "loss": 2.2227,
      "step": 123471
    },
    {
      "epoch": 1.61,
      "grad_norm": 0.976622998714447,
      "learning_rate": 0.00026624066515787604,
      "loss": 2.1607,
      "step": 123472
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.0498019456863403,
      "learning_rate": 0.00026623660058127014,
      "loss": 1.9184,
      "step": 123473
    },
    {
      "epoch": 1.61,
      "grad_norm": 0.9608570337295532,
      "learning_rate": 0.00026623253601094167,
      "loss": 2.117,
      "step": 123474
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.0303709506988525,
      "learning_rate": 0.000266228471446891,
      "loss": 2.1672,
      "step": 123475
    },
    {
      "epoch": 1.61,
      "grad_norm": 0.930932879447937,
      "learning_rate": 0.00026622440688911915,
      "loss": 2.1191,
      "step": 123476
    },
    {
      "epoch": 1.61,
      "grad_norm": 0.935300350189209,
      "learning_rate": 0.0002662203423376268,
      "loss": 2.0896,
      "step": 123477
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.1453455686569214,
      "learning_rate": 0.00026621627779241467,
      "loss": 2.0599,
      "step": 123478
    },
    {
      "epoch": 1.61,
      "grad_norm": 0.910917341709137,
      "learning_rate": 0.00026621221325348366,
      "loss": 2.0305,
      "step": 123479
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.0510796308517456,
      "learning_rate": 0.0002662081487208345,
      "loss": 1.9814,
      "step": 123480
    },
    {
      "epoch": 1.61,
      "grad_norm": 0.99272620677948,
      "learning_rate": 0.00026620408419446783,
      "loss": 1.9623,
      "step": 123481
    },
    {
      "epoch": 1.61,
      "grad_norm": 0.9678786396980286,
      "learning_rate": 0.0002662000196743844,
      "loss": 2.1722,
      "step": 123482
    },
    {
      "epoch": 1.61,
      "grad_norm": 0.9648885726928711,
      "learning_rate": 0.000266195955160585,
      "loss": 2.1603,
      "step": 123483
    },
    {
      "epoch": 1.61,
      "grad_norm": 0.8893473744392395,
      "learning_rate": 0.0002661918906530705,
      "loss": 2.074,
      "step": 123484
    },
    {
      "epoch": 1.61,
      "grad_norm": 0.95699542760849,
      "learning_rate": 0.00026618782615184147,
      "loss": 1.8855,
      "step": 123485
    },
    {
      "epoch": 1.61,
      "grad_norm": 0.9891278147697449,
      "learning_rate": 0.000266183761656899,
      "loss": 2.1838,
      "step": 123486
    },
    {
      "epoch": 1.61,
      "grad_norm": 0.9836234450340271,
      "learning_rate": 0.0002661796971682434,
      "loss": 2.0899,
      "step": 123487
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.0612502098083496,
      "learning_rate": 0.00026617563268587565,
      "loss": 1.9694,
      "step": 123488
    },
    {
      "epoch": 1.61,
      "grad_norm": 0.9333831667900085,
      "learning_rate": 0.0002661715682097965,
      "loss": 2.2784,
      "step": 123489
    },
    {
      "epoch": 1.61,
      "grad_norm": 0.8912146687507629,
      "learning_rate": 0.00026616750374000667,
      "loss": 2.1175,
      "step": 123490
    },
    {
      "epoch": 1.61,
      "grad_norm": 0.9512474536895752,
      "learning_rate": 0.0002661634392765069,
      "loss": 2.113,
      "step": 123491
    },
    {
      "epoch": 1.61,
      "grad_norm": 0.9705905914306641,
      "learning_rate": 0.0002661593748192982,
      "loss": 2.2885,
      "step": 123492
    },
    {
      "epoch": 1.61,
      "grad_norm": 0.9135208129882812,
      "learning_rate": 0.0002661553103683809,
      "loss": 2.127,
      "step": 123493
    },
    {
      "epoch": 1.61,
      "grad_norm": 0.8790292143821716,
      "learning_rate": 0.00026615124592375603,
      "loss": 2.1749,
      "step": 123494
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.0301884412765503,
      "learning_rate": 0.0002661471814854243,
      "loss": 1.9413,
      "step": 123495
    },
    {
      "epoch": 1.61,
      "grad_norm": 0.9765233993530273,
      "learning_rate": 0.0002661431170533864,
      "loss": 2.2693,
      "step": 123496
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.075195550918579,
      "learning_rate": 0.00026613905262764316,
      "loss": 2.1383,
      "step": 123497
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.012850284576416,
      "learning_rate": 0.0002661349882081954,
      "loss": 2.03,
      "step": 123498
    },
    {
      "epoch": 1.61,
      "grad_norm": 0.9589390754699707,
      "learning_rate": 0.0002661309237950437,
      "loss": 2.0497,
      "step": 123499
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.0372469425201416,
      "learning_rate": 0.00026612685938818886,
      "loss": 2.1819,
      "step": 123500
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.0344228744506836,
      "learning_rate": 0.0002661227949876317,
      "loss": 1.8668,
      "step": 123501
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.0103332996368408,
      "learning_rate": 0.00026611873059337294,
      "loss": 2.3243,
      "step": 123502
    },
    {
      "epoch": 1.61,
      "grad_norm": 0.9712672233581543,
      "learning_rate": 0.0002661146662054134,
      "loss": 2.3315,
      "step": 123503
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.1427186727523804,
      "learning_rate": 0.0002661106018237538,
      "loss": 2.0867,
      "step": 123504
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.0965348482131958,
      "learning_rate": 0.0002661065374483949,
      "loss": 2.3116,
      "step": 123505
    },
    {
      "epoch": 1.61,
      "grad_norm": 0.9267702102661133,
      "learning_rate": 0.0002661024730793373,
      "loss": 2.1299,
      "step": 123506
    },
    {
      "epoch": 1.61,
      "grad_norm": 0.9259757995605469,
      "learning_rate": 0.00026609840871658194,
      "loss": 1.8712,
      "step": 123507
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.0210281610488892,
      "learning_rate": 0.0002660943443601295,
      "loss": 1.9806,
      "step": 123508
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.1081953048706055,
      "learning_rate": 0.00026609028000998083,
      "loss": 2.2731,
      "step": 123509
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.2996025085449219,
      "learning_rate": 0.00026608621566613655,
      "loss": 2.2428,
      "step": 123510
    },
    {
      "epoch": 1.61,
      "grad_norm": 0.9952837824821472,
      "learning_rate": 0.00026608215132859766,
      "loss": 2.1291,
      "step": 123511
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.1746288537979126,
      "learning_rate": 0.0002660780869973646,
      "loss": 2.0342,
      "step": 123512
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.1491979360580444,
      "learning_rate": 0.0002660740226724383,
      "loss": 1.939,
      "step": 123513
    },
    {
      "epoch": 1.61,
      "grad_norm": 0.9880530834197998,
      "learning_rate": 0.0002660699583538194,
      "loss": 2.1527,
      "step": 123514
    },
    {
      "epoch": 1.61,
      "grad_norm": 0.950452983379364,
      "learning_rate": 0.00026606589404150875,
      "loss": 2.0775,
      "step": 123515
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.1052305698394775,
      "learning_rate": 0.0002660618297355071,
      "loss": 2.3173,
      "step": 123516
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.315733551979065,
      "learning_rate": 0.0002660577654358153,
      "loss": 2.1846,
      "step": 123517
    },
    {
      "epoch": 1.61,
      "grad_norm": 0.9332275390625,
      "learning_rate": 0.00026605370114243397,
      "loss": 1.8884,
      "step": 123518
    },
    {
      "epoch": 1.61,
      "grad_norm": 0.9357331991195679,
      "learning_rate": 0.00026604963685536383,
      "loss": 2.0687,
      "step": 123519
    },
    {
      "epoch": 1.61,
      "grad_norm": 0.9843325018882751,
      "learning_rate": 0.0002660455725746057,
      "loss": 2.0941,
      "step": 123520
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.0610653162002563,
      "learning_rate": 0.0002660415083001604,
      "loss": 2.034,
      "step": 123521
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.1387721300125122,
      "learning_rate": 0.0002660374440320285,
      "loss": 2.2298,
      "step": 123522
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.1062278747558594,
      "learning_rate": 0.00026603337977021104,
      "loss": 2.044,
      "step": 123523
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.023242712020874,
      "learning_rate": 0.00026602931551470853,
      "loss": 2.2242,
      "step": 123524
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.0659146308898926,
      "learning_rate": 0.00026602525126552187,
      "loss": 1.9676,
      "step": 123525
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.001522183418274,
      "learning_rate": 0.00026602118702265167,
      "loss": 2.2114,
      "step": 123526
    },
    {
      "epoch": 1.61,
      "grad_norm": 0.9660303592681885,
      "learning_rate": 0.00026601712278609885,
      "loss": 1.976,
      "step": 123527
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.0857325792312622,
      "learning_rate": 0.000266013058555864,
      "loss": 2.0223,
      "step": 123528
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.044049859046936,
      "learning_rate": 0.00026600899433194803,
      "loss": 2.2713,
      "step": 123529
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.1898295879364014,
      "learning_rate": 0.0002660049301143516,
      "loss": 2.0792,
      "step": 123530
    },
    {
      "epoch": 1.61,
      "grad_norm": 0.8737530708312988,
      "learning_rate": 0.00026600086590307555,
      "loss": 2.0223,
      "step": 123531
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.047453761100769,
      "learning_rate": 0.0002659968016981206,
      "loss": 2.0808,
      "step": 123532
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.0386326313018799,
      "learning_rate": 0.00026599273749948737,
      "loss": 2.09,
      "step": 123533
    },
    {
      "epoch": 1.61,
      "grad_norm": 0.9921536445617676,
      "learning_rate": 0.00026598867330717677,
      "loss": 2.1121,
      "step": 123534
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.0176945924758911,
      "learning_rate": 0.0002659846091211896,
      "loss": 1.9843,
      "step": 123535
    },
    {
      "epoch": 1.61,
      "grad_norm": 0.9918643236160278,
      "learning_rate": 0.00026598054494152644,
      "loss": 2.0635,
      "step": 123536
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.0347150564193726,
      "learning_rate": 0.0002659764807681882,
      "loss": 1.9714,
      "step": 123537
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.1039756536483765,
      "learning_rate": 0.0002659724166011755,
      "loss": 2.1054,
      "step": 123538
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.1859608888626099,
      "learning_rate": 0.0002659683524404892,
      "loss": 2.2471,
      "step": 123539
    },
    {
      "epoch": 1.61,
      "grad_norm": 0.9163820147514343,
      "learning_rate": 0.0002659642882861301,
      "loss": 2.2262,
      "step": 123540
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.5907953977584839,
      "learning_rate": 0.00026596022413809885,
      "loss": 2.2914,
      "step": 123541
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.3919373750686646,
      "learning_rate": 0.00026595615999639626,
      "loss": 2.2694,
      "step": 123542
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.0253517627716064,
      "learning_rate": 0.00026595209586102297,
      "loss": 2.0781,
      "step": 123543
    },
    {
      "epoch": 1.61,
      "grad_norm": 0.9758954048156738,
      "learning_rate": 0.00026594803173197985,
      "loss": 1.9711,
      "step": 123544
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.0853592157363892,
      "learning_rate": 0.00026594396760926765,
      "loss": 1.7788,
      "step": 123545
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.0679891109466553,
      "learning_rate": 0.0002659399034928871,
      "loss": 2.0925,
      "step": 123546
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.083680272102356,
      "learning_rate": 0.0002659358393828391,
      "loss": 1.9668,
      "step": 123547
    },
    {
      "epoch": 1.61,
      "grad_norm": 0.9206250905990601,
      "learning_rate": 0.0002659317752791241,
      "loss": 2.1974,
      "step": 123548
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.041841983795166,
      "learning_rate": 0.00026592771118174306,
      "loss": 2.1753,
      "step": 123549
    },
    {
      "epoch": 1.61,
      "grad_norm": 0.9706262350082397,
      "learning_rate": 0.0002659236470906967,
      "loss": 1.9892,
      "step": 123550
    },
    {
      "epoch": 1.61,
      "grad_norm": 0.9844064116477966,
      "learning_rate": 0.00026591958300598576,
      "loss": 2.0815,
      "step": 123551
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.077355980873108,
      "learning_rate": 0.00026591551892761104,
      "loss": 1.9633,
      "step": 123552
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.1575517654418945,
      "learning_rate": 0.00026591145485557337,
      "loss": 1.9313,
      "step": 123553
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.087003469467163,
      "learning_rate": 0.0002659073907898733,
      "loss": 2.0249,
      "step": 123554
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.1133562326431274,
      "learning_rate": 0.0002659033267305117,
      "loss": 2.0109,
      "step": 123555
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.0156290531158447,
      "learning_rate": 0.0002658992626774893,
      "loss": 2.1473,
      "step": 123556
    },
    {
      "epoch": 1.61,
      "grad_norm": 0.9435652494430542,
      "learning_rate": 0.0002658951986308069,
      "loss": 2.31,
      "step": 123557
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.0478427410125732,
      "learning_rate": 0.00026589113459046516,
      "loss": 2.001,
      "step": 123558
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.2925063371658325,
      "learning_rate": 0.00026588707055646505,
      "loss": 2.3578,
      "step": 123559
    },
    {
      "epoch": 1.61,
      "grad_norm": 0.9860467910766602,
      "learning_rate": 0.0002658830065288071,
      "loss": 2.1734,
      "step": 123560
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.0129714012145996,
      "learning_rate": 0.0002658789425074921,
      "loss": 2.026,
      "step": 123561
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.0582038164138794,
      "learning_rate": 0.00026587487849252084,
      "loss": 2.0839,
      "step": 123562
    },
    {
      "epoch": 1.61,
      "grad_norm": 0.9439256191253662,
      "learning_rate": 0.0002658708144838941,
      "loss": 2.0825,
      "step": 123563
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.020869255065918,
      "learning_rate": 0.00026586675048161264,
      "loss": 2.1187,
      "step": 123564
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.0363588333129883,
      "learning_rate": 0.0002658626864856773,
      "loss": 2.0183,
      "step": 123565
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.1563904285430908,
      "learning_rate": 0.0002658586224960886,
      "loss": 1.9222,
      "step": 123566
    },
    {
      "epoch": 1.61,
      "grad_norm": 0.9709557890892029,
      "learning_rate": 0.0002658545585128474,
      "loss": 2.071,
      "step": 123567
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.0776225328445435,
      "learning_rate": 0.00026585049453595454,
      "loss": 2.1388,
      "step": 123568
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.0626676082611084,
      "learning_rate": 0.00026584643056541067,
      "loss": 2.0947,
      "step": 123569
    },
    {
      "epoch": 1.61,
      "grad_norm": 0.9495565891265869,
      "learning_rate": 0.00026584236660121664,
      "loss": 1.9856,
      "step": 123570
    },
    {
      "epoch": 1.61,
      "grad_norm": 0.9936482906341553,
      "learning_rate": 0.00026583830264337326,
      "loss": 2.36,
      "step": 123571
    },
    {
      "epoch": 1.61,
      "grad_norm": 0.9976168274879456,
      "learning_rate": 0.00026583423869188104,
      "loss": 2.1437,
      "step": 123572
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.029069423675537,
      "learning_rate": 0.00026583017474674094,
      "loss": 2.2583,
      "step": 123573
    },
    {
      "epoch": 1.61,
      "grad_norm": 0.9061954617500305,
      "learning_rate": 0.0002658261108079536,
      "loss": 2.1704,
      "step": 123574
    },
    {
      "epoch": 1.61,
      "grad_norm": 0.9961179494857788,
      "learning_rate": 0.0002658220468755198,
      "loss": 2.215,
      "step": 123575
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.0110331773757935,
      "learning_rate": 0.0002658179829494404,
      "loss": 2.1733,
      "step": 123576
    },
    {
      "epoch": 1.61,
      "grad_norm": 0.9460158944129944,
      "learning_rate": 0.00026581391902971604,
      "loss": 2.1768,
      "step": 123577
    },
    {
      "epoch": 1.61,
      "grad_norm": 0.8985175490379333,
      "learning_rate": 0.0002658098551163477,
      "loss": 1.994,
      "step": 123578
    },
    {
      "epoch": 1.61,
      "grad_norm": 0.9035078883171082,
      "learning_rate": 0.00026580579120933577,
      "loss": 1.9706,
      "step": 123579
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.0119342803955078,
      "learning_rate": 0.0002658017273086812,
      "loss": 2.0083,
      "step": 123580
    },
    {
      "epoch": 1.61,
      "grad_norm": 0.8868833780288696,
      "learning_rate": 0.00026579766341438475,
      "loss": 2.0903,
      "step": 123581
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.1196645498275757,
      "learning_rate": 0.0002657935995264471,
      "loss": 2.1995,
      "step": 123582
    },
    {
      "epoch": 1.61,
      "grad_norm": 0.9641110301017761,
      "learning_rate": 0.00026578953564486917,
      "loss": 2.0416,
      "step": 123583
    },
    {
      "epoch": 1.61,
      "grad_norm": 0.9778988361358643,
      "learning_rate": 0.00026578547176965166,
      "loss": 2.0901,
      "step": 123584
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.1898057460784912,
      "learning_rate": 0.0002657814079007952,
      "loss": 2.0511,
      "step": 123585
    },
    {
      "epoch": 1.61,
      "grad_norm": 0.9462112188339233,
      "learning_rate": 0.0002657773440383006,
      "loss": 2.119,
      "step": 123586
    },
    {
      "epoch": 1.61,
      "grad_norm": 0.9082145690917969,
      "learning_rate": 0.0002657732801821686,
      "loss": 1.9748,
      "step": 123587
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.0705251693725586,
      "learning_rate": 0.00026576921633240003,
      "loss": 2.0597,
      "step": 123588
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.0177785158157349,
      "learning_rate": 0.0002657651524889956,
      "loss": 2.2217,
      "step": 123589
    },
    {
      "epoch": 1.61,
      "grad_norm": 0.9539217948913574,
      "learning_rate": 0.0002657610886519562,
      "loss": 2.2341,
      "step": 123590
    },
    {
      "epoch": 1.61,
      "grad_norm": 0.9698200225830078,
      "learning_rate": 0.0002657570248212824,
      "loss": 1.907,
      "step": 123591
    },
    {
      "epoch": 1.61,
      "grad_norm": 0.9082178473472595,
      "learning_rate": 0.0002657529609969749,
      "loss": 1.8494,
      "step": 123592
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.1576989889144897,
      "learning_rate": 0.00026574889717903463,
      "loss": 2.1745,
      "step": 123593
    },
    {
      "epoch": 1.61,
      "grad_norm": 0.9245153665542603,
      "learning_rate": 0.0002657448333674623,
      "loss": 1.995,
      "step": 123594
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.0057836771011353,
      "learning_rate": 0.00026574076956225863,
      "loss": 2.0436,
      "step": 123595
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.0128686428070068,
      "learning_rate": 0.0002657367057634245,
      "loss": 2.299,
      "step": 123596
    },
    {
      "epoch": 1.61,
      "grad_norm": 0.909877598285675,
      "learning_rate": 0.0002657326419709605,
      "loss": 1.9545,
      "step": 123597
    },
    {
      "epoch": 1.61,
      "grad_norm": 0.902260422706604,
      "learning_rate": 0.0002657285781848674,
      "loss": 1.9879,
      "step": 123598
    },
    {
      "epoch": 1.61,
      "grad_norm": 0.8925923109054565,
      "learning_rate": 0.000265724514405146,
      "loss": 2.115,
      "step": 123599
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.0211960077285767,
      "learning_rate": 0.0002657204506317971,
      "loss": 2.2596,
      "step": 123600
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.0612584352493286,
      "learning_rate": 0.00026571638686482136,
      "loss": 2.2742,
      "step": 123601
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.0493218898773193,
      "learning_rate": 0.00026571232310421974,
      "loss": 2.2343,
      "step": 123602
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.0671944618225098,
      "learning_rate": 0.00026570825934999273,
      "loss": 2.1728,
      "step": 123603
    },
    {
      "epoch": 1.61,
      "grad_norm": 0.9454078674316406,
      "learning_rate": 0.0002657041956021412,
      "loss": 2.2321,
      "step": 123604
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.0176584720611572,
      "learning_rate": 0.00026570013186066595,
      "loss": 2.0739,
      "step": 123605
    },
    {
      "epoch": 1.61,
      "grad_norm": 0.9603268504142761,
      "learning_rate": 0.00026569606812556763,
      "loss": 2.0278,
      "step": 123606
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.0250087976455688,
      "learning_rate": 0.0002656920043968471,
      "loss": 2.1378,
      "step": 123607
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.0159000158309937,
      "learning_rate": 0.0002656879406745051,
      "loss": 2.0654,
      "step": 123608
    },
    {
      "epoch": 1.61,
      "grad_norm": 0.9507936239242554,
      "learning_rate": 0.0002656838769585423,
      "loss": 2.1028,
      "step": 123609
    },
    {
      "epoch": 1.61,
      "grad_norm": 0.9640117287635803,
      "learning_rate": 0.0002656798132489596,
      "loss": 2.2511,
      "step": 123610
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.2120360136032104,
      "learning_rate": 0.00026567574954575765,
      "loss": 2.3697,
      "step": 123611
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.00924551486969,
      "learning_rate": 0.00026567168584893714,
      "loss": 2.178,
      "step": 123612
    },
    {
      "epoch": 1.61,
      "grad_norm": 0.9790091514587402,
      "learning_rate": 0.00026566762215849896,
      "loss": 1.9585,
      "step": 123613
    },
    {
      "epoch": 1.61,
      "grad_norm": 0.8688572645187378,
      "learning_rate": 0.00026566355847444385,
      "loss": 2.085,
      "step": 123614
    },
    {
      "epoch": 1.61,
      "grad_norm": 0.8942598700523376,
      "learning_rate": 0.0002656594947967725,
      "loss": 2.2578,
      "step": 123615
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.0051536560058594,
      "learning_rate": 0.0002656554311254857,
      "loss": 2.0299,
      "step": 123616
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.1205302476882935,
      "learning_rate": 0.00026565136746058425,
      "loss": 2.3091,
      "step": 123617
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.0733988285064697,
      "learning_rate": 0.0002656473038020688,
      "loss": 2.0673,
      "step": 123618
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.0703215599060059,
      "learning_rate": 0.00026564324014994017,
      "loss": 2.0434,
      "step": 123619
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.000303030014038,
      "learning_rate": 0.00026563917650419914,
      "loss": 1.9397,
      "step": 123620
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.0327101945877075,
      "learning_rate": 0.0002656351128648464,
      "loss": 2.1797,
      "step": 123621
    },
    {
      "epoch": 1.61,
      "grad_norm": 0.8926435112953186,
      "learning_rate": 0.00026563104923188274,
      "loss": 1.99,
      "step": 123622
    },
    {
      "epoch": 1.61,
      "grad_norm": 0.9080255627632141,
      "learning_rate": 0.0002656269856053089,
      "loss": 2.0876,
      "step": 123623
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.3434877395629883,
      "learning_rate": 0.00026562292198512575,
      "loss": 2.3055,
      "step": 123624
    },
    {
      "epoch": 1.61,
      "grad_norm": 0.8971115946769714,
      "learning_rate": 0.0002656188583713339,
      "loss": 1.8701,
      "step": 123625
    },
    {
      "epoch": 1.61,
      "grad_norm": 0.9814741611480713,
      "learning_rate": 0.00026561479476393414,
      "loss": 2.1621,
      "step": 123626
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.0141048431396484,
      "learning_rate": 0.00026561073116292724,
      "loss": 2.3077,
      "step": 123627
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.0472815036773682,
      "learning_rate": 0.00026560666756831393,
      "loss": 2.0809,
      "step": 123628
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.020247459411621,
      "learning_rate": 0.000265602603980095,
      "loss": 2.1338,
      "step": 123629
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.1808886528015137,
      "learning_rate": 0.0002655985403982712,
      "loss": 2.1517,
      "step": 123630
    },
    {
      "epoch": 1.61,
      "grad_norm": 0.9818706512451172,
      "learning_rate": 0.0002655944768228432,
      "loss": 2.1147,
      "step": 123631
    },
    {
      "epoch": 1.61,
      "grad_norm": 0.9620025157928467,
      "learning_rate": 0.0002655904132538121,
      "loss": 2.1979,
      "step": 123632
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.0327160358428955,
      "learning_rate": 0.00026558634969117814,
      "loss": 2.1997,
      "step": 123633
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.0385633707046509,
      "learning_rate": 0.00026558228613494245,
      "loss": 2.1173,
      "step": 123634
    },
    {
      "epoch": 1.61,
      "grad_norm": 0.9137545228004456,
      "learning_rate": 0.00026557822258510555,
      "loss": 1.98,
      "step": 123635
    },
    {
      "epoch": 1.61,
      "grad_norm": 0.9922466278076172,
      "learning_rate": 0.0002655741590416684,
      "loss": 2.0648,
      "step": 123636
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.0530955791473389,
      "learning_rate": 0.00026557009550463157,
      "loss": 2.1506,
      "step": 123637
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.0380899906158447,
      "learning_rate": 0.0002655660319739961,
      "loss": 2.2068,
      "step": 123638
    },
    {
      "epoch": 1.61,
      "grad_norm": 0.9762035012245178,
      "learning_rate": 0.0002655619684497624,
      "loss": 2.3544,
      "step": 123639
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.0061845779418945,
      "learning_rate": 0.0002655579049319314,
      "loss": 2.1769,
      "step": 123640
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.0247037410736084,
      "learning_rate": 0.0002655538414205038,
      "loss": 1.984,
      "step": 123641
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.033322811126709,
      "learning_rate": 0.0002655497779154804,
      "loss": 1.9875,
      "step": 123642
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.0977580547332764,
      "learning_rate": 0.000265545714416862,
      "loss": 2.0644,
      "step": 123643
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.1160887479782104,
      "learning_rate": 0.0002655416509246493,
      "loss": 2.1721,
      "step": 123644
    },
    {
      "epoch": 1.61,
      "grad_norm": 0.9958276748657227,
      "learning_rate": 0.0002655375874388431,
      "loss": 2.0132,
      "step": 123645
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.0321813821792603,
      "learning_rate": 0.00026553352395944406,
      "loss": 2.1057,
      "step": 123646
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.1833666563034058,
      "learning_rate": 0.000265529460486453,
      "loss": 2.0996,
      "step": 123647
    },
    {
      "epoch": 1.61,
      "grad_norm": 0.9701494574546814,
      "learning_rate": 0.0002655253970198706,
      "loss": 2.1754,
      "step": 123648
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.0730295181274414,
      "learning_rate": 0.0002655213335596977,
      "loss": 2.2052,
      "step": 123649
    },
    {
      "epoch": 1.61,
      "grad_norm": 0.9978880882263184,
      "learning_rate": 0.00026551727010593507,
      "loss": 2.1392,
      "step": 123650
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.1828573942184448,
      "learning_rate": 0.0002655132066585835,
      "loss": 2.193,
      "step": 123651
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.0487353801727295,
      "learning_rate": 0.0002655091432176436,
      "loss": 2.0585,
      "step": 123652
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.1217994689941406,
      "learning_rate": 0.0002655050797831162,
      "loss": 2.2206,
      "step": 123653
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.0471320152282715,
      "learning_rate": 0.00026550101635500206,
      "loss": 2.274,
      "step": 123654
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.2412800788879395,
      "learning_rate": 0.00026549695293330187,
      "loss": 1.9273,
      "step": 123655
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.0090891122817993,
      "learning_rate": 0.0002654928895180165,
      "loss": 2.2582,
      "step": 123656
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.0797419548034668,
      "learning_rate": 0.0002654888261091468,
      "loss": 2.224,
      "step": 123657
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.4088234901428223,
      "learning_rate": 0.0002654847627066932,
      "loss": 2.1276,
      "step": 123658
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.2587066888809204,
      "learning_rate": 0.00026548069931065665,
      "loss": 2.4252,
      "step": 123659
    },
    {
      "epoch": 1.61,
      "grad_norm": 0.9673219919204712,
      "learning_rate": 0.0002654766359210379,
      "loss": 2.0385,
      "step": 123660
    },
    {
      "epoch": 1.61,
      "grad_norm": 0.9802311658859253,
      "learning_rate": 0.00026547257253783766,
      "loss": 2.2168,
      "step": 123661
    },
    {
      "epoch": 1.61,
      "grad_norm": 0.9572069048881531,
      "learning_rate": 0.00026546850916105673,
      "loss": 2.1219,
      "step": 123662
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.0213079452514648,
      "learning_rate": 0.00026546444579069606,
      "loss": 2.0079,
      "step": 123663
    },
    {
      "epoch": 1.61,
      "grad_norm": 0.9632281064987183,
      "learning_rate": 0.00026546038242675603,
      "loss": 2.0019,
      "step": 123664
    },
    {
      "epoch": 1.61,
      "grad_norm": 0.9161848425865173,
      "learning_rate": 0.0002654563190692375,
      "loss": 1.9856,
      "step": 123665
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.1304690837860107,
      "learning_rate": 0.0002654522557181414,
      "loss": 2.309,
      "step": 123666
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.2662543058395386,
      "learning_rate": 0.0002654481923734683,
      "loss": 1.9594,
      "step": 123667
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.0591051578521729,
      "learning_rate": 0.0002654441290352191,
      "loss": 2.0986,
      "step": 123668
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.0616776943206787,
      "learning_rate": 0.00026544006570339453,
      "loss": 2.1055,
      "step": 123669
    },
    {
      "epoch": 1.61,
      "grad_norm": 0.9852327108383179,
      "learning_rate": 0.00026543600237799523,
      "loss": 1.9203,
      "step": 123670
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.6346362829208374,
      "learning_rate": 0.000265431939059022,
      "loss": 2.09,
      "step": 123671
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.1139634847640991,
      "learning_rate": 0.00026542787574647565,
      "loss": 2.0165,
      "step": 123672
    },
    {
      "epoch": 1.61,
      "grad_norm": 0.9474117159843445,
      "learning_rate": 0.00026542381244035694,
      "loss": 2.1001,
      "step": 123673
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.1567678451538086,
      "learning_rate": 0.0002654197491406666,
      "loss": 2.0853,
      "step": 123674
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.6563472747802734,
      "learning_rate": 0.0002654156858474054,
      "loss": 2.0637,
      "step": 123675
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.3095591068267822,
      "learning_rate": 0.000265411622560574,
      "loss": 2.2005,
      "step": 123676
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.037663221359253,
      "learning_rate": 0.0002654075592801733,
      "loss": 2.2669,
      "step": 123677
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.0123684406280518,
      "learning_rate": 0.00026540349600620393,
      "loss": 2.1344,
      "step": 123678
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.3047186136245728,
      "learning_rate": 0.00026539943273866666,
      "loss": 2.0426,
      "step": 123679
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.0427532196044922,
      "learning_rate": 0.00026539536947756234,
      "loss": 2.0398,
      "step": 123680
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.0131546258926392,
      "learning_rate": 0.00026539130622289185,
      "loss": 2.2425,
      "step": 123681
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.0032978057861328,
      "learning_rate": 0.0002653872429746556,
      "loss": 2.0945,
      "step": 123682
    },
    {
      "epoch": 1.61,
      "grad_norm": 0.8858001232147217,
      "learning_rate": 0.0002653831797328545,
      "loss": 2.1448,
      "step": 123683
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.2247271537780762,
      "learning_rate": 0.00026537911649748933,
      "loss": 2.0513,
      "step": 123684
    },
    {
      "epoch": 1.61,
      "grad_norm": 0.9921896457672119,
      "learning_rate": 0.0002653750532685609,
      "loss": 2.2363,
      "step": 123685
    },
    {
      "epoch": 1.61,
      "grad_norm": 0.9764695763587952,
      "learning_rate": 0.0002653709900460698,
      "loss": 2.2153,
      "step": 123686
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.038286566734314,
      "learning_rate": 0.00026536692683001704,
      "loss": 2.2711,
      "step": 123687
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.0347180366516113,
      "learning_rate": 0.0002653628636204031,
      "loss": 2.1446,
      "step": 123688
    },
    {
      "epoch": 1.61,
      "grad_norm": 0.9933621883392334,
      "learning_rate": 0.0002653588004172289,
      "loss": 2.2071,
      "step": 123689
    },
    {
      "epoch": 1.61,
      "grad_norm": 0.9473944306373596,
      "learning_rate": 0.0002653547372204951,
      "loss": 1.9514,
      "step": 123690
    },
    {
      "epoch": 1.61,
      "grad_norm": 0.9713258743286133,
      "learning_rate": 0.0002653506740302025,
      "loss": 2.0146,
      "step": 123691
    },
    {
      "epoch": 1.61,
      "grad_norm": 0.9999726414680481,
      "learning_rate": 0.0002653466108463519,
      "loss": 2.3508,
      "step": 123692
    },
    {
      "epoch": 1.61,
      "grad_norm": 0.9846202731132507,
      "learning_rate": 0.00026534254766894406,
      "loss": 2.1553,
      "step": 123693
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.0360091924667358,
      "learning_rate": 0.0002653384844979797,
      "loss": 2.1344,
      "step": 123694
    },
    {
      "epoch": 1.61,
      "grad_norm": 0.9541034698486328,
      "learning_rate": 0.0002653344213334595,
      "loss": 2.1684,
      "step": 123695
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.117856502532959,
      "learning_rate": 0.0002653303581753843,
      "loss": 2.2907,
      "step": 123696
    },
    {
      "epoch": 1.61,
      "grad_norm": 0.9795454740524292,
      "learning_rate": 0.00026532629502375486,
      "loss": 2.2215,
      "step": 123697
    },
    {
      "epoch": 1.61,
      "grad_norm": 0.934012234210968,
      "learning_rate": 0.00026532223187857186,
      "loss": 2.0804,
      "step": 123698
    },
    {
      "epoch": 1.61,
      "grad_norm": 0.9821769595146179,
      "learning_rate": 0.00026531816873983624,
      "loss": 2.3921,
      "step": 123699
    },
    {
      "epoch": 1.61,
      "grad_norm": 0.9234111905097961,
      "learning_rate": 0.00026531410560754853,
      "loss": 2.0916,
      "step": 123700
    },
    {
      "epoch": 1.61,
      "grad_norm": 0.9660240411758423,
      "learning_rate": 0.0002653100424817096,
      "loss": 2.3011,
      "step": 123701
    },
    {
      "epoch": 1.61,
      "grad_norm": 0.9981375336647034,
      "learning_rate": 0.00026530597936232027,
      "loss": 2.2431,
      "step": 123702
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.0287095308303833,
      "learning_rate": 0.00026530191624938107,
      "loss": 2.1328,
      "step": 123703
    },
    {
      "epoch": 1.61,
      "grad_norm": 0.9582654237747192,
      "learning_rate": 0.0002652978531428929,
      "loss": 2.1678,
      "step": 123704
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.0852701663970947,
      "learning_rate": 0.0002652937900428567,
      "loss": 2.3374,
      "step": 123705
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.041606068611145,
      "learning_rate": 0.0002652897269492729,
      "loss": 2.2066,
      "step": 123706
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.0504844188690186,
      "learning_rate": 0.0002652856638621424,
      "loss": 2.2099,
      "step": 123707
    },
    {
      "epoch": 1.61,
      "grad_norm": 0.9494021534919739,
      "learning_rate": 0.00026528160078146594,
      "loss": 2.1141,
      "step": 123708
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.1484699249267578,
      "learning_rate": 0.00026527753770724436,
      "loss": 2.2011,
      "step": 123709
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.1015490293502808,
      "learning_rate": 0.0002652734746394782,
      "loss": 2.2606,
      "step": 123710
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.3752256631851196,
      "learning_rate": 0.00026526941157816846,
      "loss": 2.0707,
      "step": 123711
    },
    {
      "epoch": 1.61,
      "grad_norm": 0.9672509431838989,
      "learning_rate": 0.00026526534852331583,
      "loss": 1.9145,
      "step": 123712
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.1045358180999756,
      "learning_rate": 0.00026526128547492097,
      "loss": 1.8934,
      "step": 123713
    },
    {
      "epoch": 1.61,
      "grad_norm": 0.9552870392799377,
      "learning_rate": 0.0002652572224329847,
      "loss": 2.0784,
      "step": 123714
    },
    {
      "epoch": 1.61,
      "grad_norm": 0.9724794030189514,
      "learning_rate": 0.00026525315939750774,
      "loss": 2.0997,
      "step": 123715
    },
    {
      "epoch": 1.61,
      "grad_norm": 0.9129488468170166,
      "learning_rate": 0.00026524909636849094,
      "loss": 2.0569,
      "step": 123716
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.0247726440429688,
      "learning_rate": 0.0002652450333459349,
      "loss": 2.2442,
      "step": 123717
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.209612250328064,
      "learning_rate": 0.0002652409703298406,
      "loss": 2.1775,
      "step": 123718
    },
    {
      "epoch": 1.61,
      "grad_norm": 0.999443769454956,
      "learning_rate": 0.00026523690732020857,
      "loss": 2.0208,
      "step": 123719
    },
    {
      "epoch": 1.61,
      "grad_norm": 0.9985995888710022,
      "learning_rate": 0.00026523284431703963,
      "loss": 2.0457,
      "step": 123720
    },
    {
      "epoch": 1.61,
      "grad_norm": 0.9737458825111389,
      "learning_rate": 0.00026522878132033455,
      "loss": 1.8932,
      "step": 123721
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.13812255859375,
      "learning_rate": 0.00026522471833009413,
      "loss": 2.2582,
      "step": 123722
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.026528000831604,
      "learning_rate": 0.00026522065534631913,
      "loss": 2.2679,
      "step": 123723
    },
    {
      "epoch": 1.61,
      "grad_norm": 0.9969226121902466,
      "learning_rate": 0.0002652165923690103,
      "loss": 2.0304,
      "step": 123724
    },
    {
      "epoch": 1.61,
      "grad_norm": 0.9589282870292664,
      "learning_rate": 0.0002652125293981683,
      "loss": 2.228,
      "step": 123725
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.0920405387878418,
      "learning_rate": 0.0002652084664337939,
      "loss": 2.0032,
      "step": 123726
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.0155080556869507,
      "learning_rate": 0.00026520440347588796,
      "loss": 2.2043,
      "step": 123727
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.1720386743545532,
      "learning_rate": 0.00026520034052445113,
      "loss": 2.1417,
      "step": 123728
    },
    {
      "epoch": 1.61,
      "grad_norm": 0.9966194033622742,
      "learning_rate": 0.00026519627757948423,
      "loss": 2.0555,
      "step": 123729
    },
    {
      "epoch": 1.61,
      "grad_norm": 0.9257653951644897,
      "learning_rate": 0.0002651922146409881,
      "loss": 2.199,
      "step": 123730
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.2159545421600342,
      "learning_rate": 0.0002651881517089633,
      "loss": 2.1124,
      "step": 123731
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.087668776512146,
      "learning_rate": 0.00026518408878341065,
      "loss": 2.1555,
      "step": 123732
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.0134295225143433,
      "learning_rate": 0.000265180025864331,
      "loss": 2.1821,
      "step": 123733
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.109399676322937,
      "learning_rate": 0.00026517596295172496,
      "loss": 2.2812,
      "step": 123734
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.2094008922576904,
      "learning_rate": 0.0002651719000455934,
      "loss": 2.1335,
      "step": 123735
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.08954918384552,
      "learning_rate": 0.0002651678371459372,
      "loss": 1.8776,
      "step": 123736
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.086410403251648,
      "learning_rate": 0.0002651637742527568,
      "loss": 2.329,
      "step": 123737
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.0750676393508911,
      "learning_rate": 0.0002651597113660531,
      "loss": 2.0002,
      "step": 123738
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.1239440441131592,
      "learning_rate": 0.0002651556484858269,
      "loss": 2.0966,
      "step": 123739
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.132555603981018,
      "learning_rate": 0.0002651515856120789,
      "loss": 1.8469,
      "step": 123740
    },
    {
      "epoch": 1.61,
      "grad_norm": 0.9553995728492737,
      "learning_rate": 0.00026514752274480987,
      "loss": 2.0855,
      "step": 123741
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.0312154293060303,
      "learning_rate": 0.0002651434598840207,
      "loss": 2.2036,
      "step": 123742
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.0705606937408447,
      "learning_rate": 0.0002651393970297119,
      "loss": 2.3879,
      "step": 123743
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.090582013130188,
      "learning_rate": 0.00026513533418188437,
      "loss": 2.205,
      "step": 123744
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.0630217790603638,
      "learning_rate": 0.00026513127134053884,
      "loss": 2.0275,
      "step": 123745
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.2514222860336304,
      "learning_rate": 0.00026512720850567603,
      "loss": 2.0801,
      "step": 123746
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.0141299962997437,
      "learning_rate": 0.00026512314567729675,
      "loss": 2.1205,
      "step": 123747
    },
    {
      "epoch": 1.61,
      "grad_norm": 0.9234970211982727,
      "learning_rate": 0.00026511908285540187,
      "loss": 2.1144,
      "step": 123748
    },
    {
      "epoch": 1.61,
      "grad_norm": 0.9428660869598389,
      "learning_rate": 0.00026511502003999187,
      "loss": 2.0933,
      "step": 123749
    },
    {
      "epoch": 1.61,
      "grad_norm": 0.955410897731781,
      "learning_rate": 0.0002651109572310676,
      "loss": 2.2205,
      "step": 123750
    },
    {
      "epoch": 1.61,
      "grad_norm": 0.9962111115455627,
      "learning_rate": 0.00026510689442862993,
      "loss": 2.1901,
      "step": 123751
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.0400763750076294,
      "learning_rate": 0.0002651028316326795,
      "loss": 2.2962,
      "step": 123752
    },
    {
      "epoch": 1.61,
      "grad_norm": 0.9716157913208008,
      "learning_rate": 0.0002650987688432172,
      "loss": 2.2063,
      "step": 123753
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.0242031812667847,
      "learning_rate": 0.0002650947060602437,
      "loss": 2.1902,
      "step": 123754
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.026769757270813,
      "learning_rate": 0.0002650906432837597,
      "loss": 2.1909,
      "step": 123755
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.0731093883514404,
      "learning_rate": 0.000265086580513766,
      "loss": 2.3501,
      "step": 123756
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.1634817123413086,
      "learning_rate": 0.0002650825177502634,
      "loss": 2.0798,
      "step": 123757
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.0085636377334595,
      "learning_rate": 0.00026507845499325257,
      "loss": 2.2859,
      "step": 123758
    },
    {
      "epoch": 1.61,
      "grad_norm": 0.9416912198066711,
      "learning_rate": 0.0002650743922427343,
      "loss": 2.363,
      "step": 123759
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.0899955034255981,
      "learning_rate": 0.0002650703294987095,
      "loss": 2.4058,
      "step": 123760
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.1348059177398682,
      "learning_rate": 0.0002650662667611787,
      "loss": 2.0968,
      "step": 123761
    },
    {
      "epoch": 1.61,
      "grad_norm": 0.965810239315033,
      "learning_rate": 0.0002650622040301427,
      "loss": 1.9642,
      "step": 123762
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.0639688968658447,
      "learning_rate": 0.0002650581413056023,
      "loss": 2.0869,
      "step": 123763
    },
    {
      "epoch": 1.61,
      "grad_norm": 0.9619823694229126,
      "learning_rate": 0.00026505407858755825,
      "loss": 2.1329,
      "step": 123764
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.051596999168396,
      "learning_rate": 0.0002650500158760113,
      "loss": 2.4019,
      "step": 123765
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.0193085670471191,
      "learning_rate": 0.00026504595317096235,
      "loss": 2.0707,
      "step": 123766
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.0295281410217285,
      "learning_rate": 0.0002650418904724119,
      "loss": 2.2074,
      "step": 123767
    },
    {
      "epoch": 1.61,
      "grad_norm": 0.9412779211997986,
      "learning_rate": 0.0002650378277803608,
      "loss": 2.2548,
      "step": 123768
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.0428218841552734,
      "learning_rate": 0.00026503376509480985,
      "loss": 2.1489,
      "step": 123769
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.1676501035690308,
      "learning_rate": 0.0002650297024157598,
      "loss": 1.9716,
      "step": 123770
    },
    {
      "epoch": 1.61,
      "grad_norm": 0.9435941576957703,
      "learning_rate": 0.0002650256397432113,
      "loss": 2.0769,
      "step": 123771
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.181463599205017,
      "learning_rate": 0.0002650215770771654,
      "loss": 2.1911,
      "step": 123772
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.1085923910140991,
      "learning_rate": 0.0002650175144176225,
      "loss": 2.0907,
      "step": 123773
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.0309889316558838,
      "learning_rate": 0.0002650134517645835,
      "loss": 2.1215,
      "step": 123774
    },
    {
      "epoch": 1.61,
      "grad_norm": 0.9171042442321777,
      "learning_rate": 0.0002650093891180492,
      "loss": 1.8315,
      "step": 123775
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.2293959856033325,
      "learning_rate": 0.0002650053264780203,
      "loss": 2.1485,
      "step": 123776
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.0623286962509155,
      "learning_rate": 0.0002650012638444975,
      "loss": 2.0496,
      "step": 123777
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.0654664039611816,
      "learning_rate": 0.00026499720121748175,
      "loss": 1.871,
      "step": 123778
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.0579719543457031,
      "learning_rate": 0.0002649931385969737,
      "loss": 2.3053,
      "step": 123779
    },
    {
      "epoch": 1.61,
      "grad_norm": 0.9959107637405396,
      "learning_rate": 0.000264989075982974,
      "loss": 2.0064,
      "step": 123780
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.0719554424285889,
      "learning_rate": 0.00026498501337548347,
      "loss": 1.8609,
      "step": 123781
    },
    {
      "epoch": 1.61,
      "grad_norm": 0.9656383395195007,
      "learning_rate": 0.0002649809507745029,
      "loss": 2.1946,
      "step": 123782
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.0413408279418945,
      "learning_rate": 0.00026497688818003304,
      "loss": 2.0599,
      "step": 123783
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.0629295110702515,
      "learning_rate": 0.00026497282559207467,
      "loss": 2.0552,
      "step": 123784
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.0075740814208984,
      "learning_rate": 0.00026496876301062856,
      "loss": 1.9558,
      "step": 123785
    },
    {
      "epoch": 1.61,
      "grad_norm": 0.9404618740081787,
      "learning_rate": 0.00026496470043569536,
      "loss": 2.1801,
      "step": 123786
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.0424777269363403,
      "learning_rate": 0.0002649606378672759,
      "loss": 2.2046,
      "step": 123787
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.1063247919082642,
      "learning_rate": 0.00026495657530537084,
      "loss": 2.1591,
      "step": 123788
    },
    {
      "epoch": 1.61,
      "grad_norm": 0.9386767148971558,
      "learning_rate": 0.0002649525127499811,
      "loss": 2.1135,
      "step": 123789
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.0500013828277588,
      "learning_rate": 0.0002649484502011073,
      "loss": 2.2753,
      "step": 123790
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.1438387632369995,
      "learning_rate": 0.0002649443876587503,
      "loss": 2.0128,
      "step": 123791
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.0573991537094116,
      "learning_rate": 0.0002649403251229107,
      "loss": 1.9273,
      "step": 123792
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.0145894289016724,
      "learning_rate": 0.0002649362625935894,
      "loss": 2.1532,
      "step": 123793
    },
    {
      "epoch": 1.61,
      "grad_norm": 0.9395294785499573,
      "learning_rate": 0.00026493220007078725,
      "loss": 1.9698,
      "step": 123794
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.0691229104995728,
      "learning_rate": 0.0002649281375545047,
      "loss": 2.1834,
      "step": 123795
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.0535564422607422,
      "learning_rate": 0.0002649240750447427,
      "loss": 2.2538,
      "step": 123796
    },
    {
      "epoch": 1.61,
      "grad_norm": 0.9856210947036743,
      "learning_rate": 0.00026492001254150205,
      "loss": 1.9675,
      "step": 123797
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.1754395961761475,
      "learning_rate": 0.00026491595004478336,
      "loss": 2.0847,
      "step": 123798
    },
    {
      "epoch": 1.61,
      "grad_norm": 0.9419221878051758,
      "learning_rate": 0.00026491188755458744,
      "loss": 2.219,
      "step": 123799
    },
    {
      "epoch": 1.61,
      "grad_norm": 0.8940442800521851,
      "learning_rate": 0.00026490782507091506,
      "loss": 1.9892,
      "step": 123800
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.0208849906921387,
      "learning_rate": 0.0002649037625937671,
      "loss": 2.1138,
      "step": 123801
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.0779528617858887,
      "learning_rate": 0.00026489970012314406,
      "loss": 2.2385,
      "step": 123802
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.0269314050674438,
      "learning_rate": 0.00026489563765904695,
      "loss": 2.3136,
      "step": 123803
    },
    {
      "epoch": 1.61,
      "grad_norm": 0.9483813643455505,
      "learning_rate": 0.0002648915752014763,
      "loss": 2.0994,
      "step": 123804
    },
    {
      "epoch": 1.61,
      "grad_norm": 0.9635505080223083,
      "learning_rate": 0.000264887512750433,
      "loss": 2.1642,
      "step": 123805
    },
    {
      "epoch": 1.61,
      "grad_norm": 0.971778929233551,
      "learning_rate": 0.0002648834503059177,
      "loss": 2.1614,
      "step": 123806
    },
    {
      "epoch": 1.61,
      "grad_norm": 0.9752709865570068,
      "learning_rate": 0.00026487938786793133,
      "loss": 2.2316,
      "step": 123807
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.0680491924285889,
      "learning_rate": 0.00026487532543647456,
      "loss": 2.0041,
      "step": 123808
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.0450743436813354,
      "learning_rate": 0.0002648712630115481,
      "loss": 2.1881,
      "step": 123809
    },
    {
      "epoch": 1.61,
      "grad_norm": 0.9590663909912109,
      "learning_rate": 0.0002648672005931527,
      "loss": 2.3174,
      "step": 123810
    },
    {
      "epoch": 1.61,
      "grad_norm": 0.9906929135322571,
      "learning_rate": 0.00026486313818128917,
      "loss": 2.2402,
      "step": 123811
    },
    {
      "epoch": 1.61,
      "grad_norm": 0.8758707046508789,
      "learning_rate": 0.00026485907577595825,
      "loss": 2.0669,
      "step": 123812
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.1506074666976929,
      "learning_rate": 0.00026485501337716064,
      "loss": 2.3777,
      "step": 123813
    },
    {
      "epoch": 1.61,
      "grad_norm": 0.9553214907646179,
      "learning_rate": 0.00026485095098489723,
      "loss": 2.2529,
      "step": 123814
    },
    {
      "epoch": 1.61,
      "grad_norm": 0.9908257126808167,
      "learning_rate": 0.00026484688859916875,
      "loss": 2.0625,
      "step": 123815
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.2187633514404297,
      "learning_rate": 0.0002648428262199758,
      "loss": 2.0735,
      "step": 123816
    },
    {
      "epoch": 1.61,
      "grad_norm": 0.8733115792274475,
      "learning_rate": 0.0002648387638473192,
      "loss": 1.9415,
      "step": 123817
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.062224268913269,
      "learning_rate": 0.00026483470148119977,
      "loss": 2.199,
      "step": 123818
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.0050268173217773,
      "learning_rate": 0.00026483063912161823,
      "loss": 2.1381,
      "step": 123819
    },
    {
      "epoch": 1.61,
      "grad_norm": 0.8964968919754028,
      "learning_rate": 0.0002648265767685753,
      "loss": 2.0557,
      "step": 123820
    },
    {
      "epoch": 1.61,
      "grad_norm": 0.9273049235343933,
      "learning_rate": 0.00026482251442207195,
      "loss": 2.0018,
      "step": 123821
    },
    {
      "epoch": 1.61,
      "grad_norm": 0.8726680874824524,
      "learning_rate": 0.00026481845208210856,
      "loss": 2.0709,
      "step": 123822
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.167510986328125,
      "learning_rate": 0.00026481438974868617,
      "loss": 2.1456,
      "step": 123823
    },
    {
      "epoch": 1.61,
      "grad_norm": 0.9964441061019897,
      "learning_rate": 0.0002648103274218054,
      "loss": 2.0962,
      "step": 123824
    },
    {
      "epoch": 1.61,
      "grad_norm": 0.9422195553779602,
      "learning_rate": 0.00026480626510146706,
      "loss": 2.1834,
      "step": 123825
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.2289611101150513,
      "learning_rate": 0.0002648022027876719,
      "loss": 2.0674,
      "step": 123826
    },
    {
      "epoch": 1.61,
      "grad_norm": 0.9581356644630432,
      "learning_rate": 0.0002647981404804208,
      "loss": 1.9044,
      "step": 123827
    },
    {
      "epoch": 1.61,
      "grad_norm": 0.8742477297782898,
      "learning_rate": 0.00026479407817971433,
      "loss": 1.9403,
      "step": 123828
    },
    {
      "epoch": 1.61,
      "grad_norm": 0.8730278015136719,
      "learning_rate": 0.0002647900158855532,
      "loss": 2.0295,
      "step": 123829
    },
    {
      "epoch": 1.61,
      "grad_norm": 0.9986369609832764,
      "learning_rate": 0.0002647859535979383,
      "loss": 2.0806,
      "step": 123830
    },
    {
      "epoch": 1.61,
      "grad_norm": 0.901432454586029,
      "learning_rate": 0.0002647818913168704,
      "loss": 2.1248,
      "step": 123831
    },
    {
      "epoch": 1.61,
      "grad_norm": 0.9766387939453125,
      "learning_rate": 0.00026477782904235024,
      "loss": 2.1974,
      "step": 123832
    },
    {
      "epoch": 1.61,
      "grad_norm": 0.9374265074729919,
      "learning_rate": 0.0002647737667743786,
      "loss": 2.0016,
      "step": 123833
    },
    {
      "epoch": 1.61,
      "grad_norm": 0.9076213836669922,
      "learning_rate": 0.00026476970451295604,
      "loss": 2.123,
      "step": 123834
    },
    {
      "epoch": 1.61,
      "grad_norm": 0.9150096774101257,
      "learning_rate": 0.0002647656422580835,
      "loss": 2.0559,
      "step": 123835
    },
    {
      "epoch": 1.61,
      "grad_norm": 0.9771924018859863,
      "learning_rate": 0.0002647615800097617,
      "loss": 2.3221,
      "step": 123836
    },
    {
      "epoch": 1.61,
      "grad_norm": 0.8786138892173767,
      "learning_rate": 0.00026475751776799136,
      "loss": 2.0596,
      "step": 123837
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.1078569889068604,
      "learning_rate": 0.00026475345553277324,
      "loss": 2.2925,
      "step": 123838
    },
    {
      "epoch": 1.61,
      "grad_norm": 0.9431119561195374,
      "learning_rate": 0.0002647493933041083,
      "loss": 2.0657,
      "step": 123839
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.0091133117675781,
      "learning_rate": 0.000264745331081997,
      "loss": 2.2271,
      "step": 123840
    },
    {
      "epoch": 1.61,
      "grad_norm": 0.9439127445220947,
      "learning_rate": 0.00026474126886644016,
      "loss": 1.9937,
      "step": 123841
    },
    {
      "epoch": 1.61,
      "grad_norm": 0.9393836259841919,
      "learning_rate": 0.0002647372066574386,
      "loss": 2.0184,
      "step": 123842
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.0048514604568481,
      "learning_rate": 0.00026473314445499307,
      "loss": 2.0374,
      "step": 123843
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.0588414669036865,
      "learning_rate": 0.00026472908225910427,
      "loss": 2.0406,
      "step": 123844
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.076171875,
      "learning_rate": 0.000264725020069773,
      "loss": 2.0506,
      "step": 123845
    },
    {
      "epoch": 1.61,
      "grad_norm": 0.9330485463142395,
      "learning_rate": 0.00026472095788700016,
      "loss": 1.9653,
      "step": 123846
    },
    {
      "epoch": 1.61,
      "grad_norm": 0.9460737705230713,
      "learning_rate": 0.00026471689571078627,
      "loss": 2.0632,
      "step": 123847
    },
    {
      "epoch": 1.61,
      "grad_norm": 0.939338207244873,
      "learning_rate": 0.0002647128335411322,
      "loss": 2.2025,
      "step": 123848
    },
    {
      "epoch": 1.61,
      "grad_norm": 0.8947303891181946,
      "learning_rate": 0.0002647087713780386,
      "loss": 2.0892,
      "step": 123849
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.059672236442566,
      "learning_rate": 0.00026470470922150633,
      "loss": 2.06,
      "step": 123850
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.0333055257797241,
      "learning_rate": 0.00026470064707153615,
      "loss": 2.2165,
      "step": 123851
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.0163391828536987,
      "learning_rate": 0.00026469658492812886,
      "loss": 1.9777,
      "step": 123852
    },
    {
      "epoch": 1.61,
      "grad_norm": 0.9531511664390564,
      "learning_rate": 0.00026469252279128505,
      "loss": 2.1104,
      "step": 123853
    },
    {
      "epoch": 1.61,
      "grad_norm": 0.9402914047241211,
      "learning_rate": 0.00026468846066100555,
      "loss": 2.0031,
      "step": 123854
    },
    {
      "epoch": 1.61,
      "grad_norm": 0.9498739838600159,
      "learning_rate": 0.0002646843985372911,
      "loss": 2.0658,
      "step": 123855
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.0148009061813354,
      "learning_rate": 0.00026468033642014255,
      "loss": 2.1441,
      "step": 123856
    },
    {
      "epoch": 1.61,
      "grad_norm": 0.9772363901138306,
      "learning_rate": 0.00026467627430956057,
      "loss": 2.274,
      "step": 123857
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.0474226474761963,
      "learning_rate": 0.0002646722122055461,
      "loss": 2.1287,
      "step": 123858
    },
    {
      "epoch": 1.61,
      "grad_norm": 0.9803658723831177,
      "learning_rate": 0.00026466815010809953,
      "loss": 1.9972,
      "step": 123859
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.060287594795227,
      "learning_rate": 0.0002646640880172219,
      "loss": 2.0553,
      "step": 123860
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.0112498998641968,
      "learning_rate": 0.0002646600259329138,
      "loss": 2.0726,
      "step": 123861
    },
    {
      "epoch": 1.61,
      "grad_norm": 0.8887659311294556,
      "learning_rate": 0.0002646559638551761,
      "loss": 1.9115,
      "step": 123862
    },
    {
      "epoch": 1.61,
      "grad_norm": 0.9384112358093262,
      "learning_rate": 0.0002646519017840095,
      "loss": 2.1397,
      "step": 123863
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.0810834169387817,
      "learning_rate": 0.000264647839719415,
      "loss": 2.0771,
      "step": 123864
    },
    {
      "epoch": 1.61,
      "grad_norm": 0.9720816612243652,
      "learning_rate": 0.0002646437776613929,
      "loss": 2.0072,
      "step": 123865
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.1316651105880737,
      "learning_rate": 0.0002646397156099443,
      "loss": 1.901,
      "step": 123866
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.0078054666519165,
      "learning_rate": 0.00026463565356506973,
      "loss": 1.9608,
      "step": 123867
    },
    {
      "epoch": 1.61,
      "grad_norm": 0.9691187739372253,
      "learning_rate": 0.0002646315915267701,
      "loss": 1.8142,
      "step": 123868
    },
    {
      "epoch": 1.61,
      "grad_norm": 0.9605549573898315,
      "learning_rate": 0.00026462752949504614,
      "loss": 1.9396,
      "step": 123869
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.1357080936431885,
      "learning_rate": 0.00026462346746989864,
      "loss": 1.9863,
      "step": 123870
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.0739315748214722,
      "learning_rate": 0.00026461940545132834,
      "loss": 2.1834,
      "step": 123871
    },
    {
      "epoch": 1.61,
      "grad_norm": 0.893957257270813,
      "learning_rate": 0.0002646153434393359,
      "loss": 1.9109,
      "step": 123872
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.0124458074569702,
      "learning_rate": 0.00026461128143392206,
      "loss": 1.9189,
      "step": 123873
    },
    {
      "epoch": 1.61,
      "grad_norm": 0.959455132484436,
      "learning_rate": 0.0002646072194350877,
      "loss": 2.086,
      "step": 123874
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.0724104642868042,
      "learning_rate": 0.00026460315744283354,
      "loss": 2.156,
      "step": 123875
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.0084412097930908,
      "learning_rate": 0.00026459909545716037,
      "loss": 2.3718,
      "step": 123876
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.0493334531784058,
      "learning_rate": 0.0002645950334780688,
      "loss": 2.0517,
      "step": 123877
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.0133479833602905,
      "learning_rate": 0.0002645909715055598,
      "loss": 2.1991,
      "step": 123878
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.0433331727981567,
      "learning_rate": 0.0002645869095396339,
      "loss": 2.0548,
      "step": 123879
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.067192554473877,
      "learning_rate": 0.000264582847580292,
      "loss": 2.1428,
      "step": 123880
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.1484146118164062,
      "learning_rate": 0.0002645787856275348,
      "loss": 2.1173,
      "step": 123881
    },
    {
      "epoch": 1.61,
      "grad_norm": 0.974618136882782,
      "learning_rate": 0.0002645747236813631,
      "loss": 2.3314,
      "step": 123882
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.1291069984436035,
      "learning_rate": 0.0002645706617417776,
      "loss": 2.1791,
      "step": 123883
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.0327785015106201,
      "learning_rate": 0.0002645665998087791,
      "loss": 2.2602,
      "step": 123884
    },
    {
      "epoch": 1.61,
      "grad_norm": 0.9758469462394714,
      "learning_rate": 0.0002645625378823683,
      "loss": 2.0537,
      "step": 123885
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.0281342267990112,
      "learning_rate": 0.00026455847596254604,
      "loss": 2.3053,
      "step": 123886
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.0331941843032837,
      "learning_rate": 0.000264554414049313,
      "loss": 2.1047,
      "step": 123887
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.059800148010254,
      "learning_rate": 0.00026455035214267006,
      "loss": 2.2639,
      "step": 123888
    },
    {
      "epoch": 1.61,
      "grad_norm": 0.9005240797996521,
      "learning_rate": 0.00026454629024261775,
      "loss": 2.051,
      "step": 123889
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.0011622905731201,
      "learning_rate": 0.000264542228349157,
      "loss": 2.114,
      "step": 123890
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.1108142137527466,
      "learning_rate": 0.0002645381664622885,
      "loss": 2.0571,
      "step": 123891
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.0045340061187744,
      "learning_rate": 0.00026453410458201305,
      "loss": 2.094,
      "step": 123892
    },
    {
      "epoch": 1.61,
      "grad_norm": 0.946129322052002,
      "learning_rate": 0.0002645300427083314,
      "loss": 1.9672,
      "step": 123893
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.080321192741394,
      "learning_rate": 0.0002645259808412443,
      "loss": 2.3015,
      "step": 123894
    },
    {
      "epoch": 1.61,
      "grad_norm": 0.9122231602668762,
      "learning_rate": 0.00026452191898075243,
      "loss": 2.087,
      "step": 123895
    },
    {
      "epoch": 1.61,
      "grad_norm": 0.9896184802055359,
      "learning_rate": 0.0002645178571268566,
      "loss": 2.2231,
      "step": 123896
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.0279711484909058,
      "learning_rate": 0.00026451379527955756,
      "loss": 2.2477,
      "step": 123897
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.0484414100646973,
      "learning_rate": 0.00026450973343885616,
      "loss": 2.0054,
      "step": 123898
    },
    {
      "epoch": 1.61,
      "grad_norm": 0.9871233105659485,
      "learning_rate": 0.000264505671604753,
      "loss": 2.047,
      "step": 123899
    },
    {
      "epoch": 1.61,
      "grad_norm": 0.9830417037010193,
      "learning_rate": 0.000264501609777249,
      "loss": 2.1507,
      "step": 123900
    },
    {
      "epoch": 1.61,
      "grad_norm": 0.934497058391571,
      "learning_rate": 0.0002644975479563447,
      "loss": 2.3476,
      "step": 123901
    },
    {
      "epoch": 1.61,
      "grad_norm": 0.9550915956497192,
      "learning_rate": 0.00026449348614204104,
      "loss": 2.0732,
      "step": 123902
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.0092737674713135,
      "learning_rate": 0.00026448942433433863,
      "loss": 2.0645,
      "step": 123903
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.0429761409759521,
      "learning_rate": 0.0002644853625332384,
      "loss": 2.1923,
      "step": 123904
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.0567539930343628,
      "learning_rate": 0.00026448130073874094,
      "loss": 2.1031,
      "step": 123905
    },
    {
      "epoch": 1.61,
      "grad_norm": 0.957580029964447,
      "learning_rate": 0.0002644772389508472,
      "loss": 2.1265,
      "step": 123906
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.050390601158142,
      "learning_rate": 0.0002644731771695577,
      "loss": 2.1387,
      "step": 123907
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.081587553024292,
      "learning_rate": 0.0002644691153948734,
      "loss": 2.2339,
      "step": 123908
    },
    {
      "epoch": 1.61,
      "grad_norm": 0.9770727753639221,
      "learning_rate": 0.00026446505362679487,
      "loss": 2.319,
      "step": 123909
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.0448017120361328,
      "learning_rate": 0.00026446099186532295,
      "loss": 2.316,
      "step": 123910
    },
    {
      "epoch": 1.61,
      "grad_norm": 0.9637664556503296,
      "learning_rate": 0.0002644569301104584,
      "loss": 2.0084,
      "step": 123911
    },
    {
      "epoch": 1.61,
      "grad_norm": 0.9774404764175415,
      "learning_rate": 0.000264452868362202,
      "loss": 2.0015,
      "step": 123912
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.0899120569229126,
      "learning_rate": 0.0002644488066205546,
      "loss": 2.2702,
      "step": 123913
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.001883625984192,
      "learning_rate": 0.0002644447448855168,
      "loss": 2.1596,
      "step": 123914
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.1742444038391113,
      "learning_rate": 0.0002644406831570893,
      "loss": 2.219,
      "step": 123915
    },
    {
      "epoch": 1.61,
      "grad_norm": 0.946036159992218,
      "learning_rate": 0.0002644366214352729,
      "loss": 2.1437,
      "step": 123916
    },
    {
      "epoch": 1.61,
      "grad_norm": 0.8967829942703247,
      "learning_rate": 0.00026443255972006846,
      "loss": 2.1318,
      "step": 123917
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.1104241609573364,
      "learning_rate": 0.0002644284980114767,
      "loss": 2.1331,
      "step": 123918
    },
    {
      "epoch": 1.61,
      "grad_norm": 0.9449875354766846,
      "learning_rate": 0.00026442443630949843,
      "loss": 2.1863,
      "step": 123919
    },
    {
      "epoch": 1.61,
      "grad_norm": 0.9024932980537415,
      "learning_rate": 0.0002644203746141343,
      "loss": 2.1528,
      "step": 123920
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.4376413822174072,
      "learning_rate": 0.000264416312925385,
      "loss": 2.2007,
      "step": 123921
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.0364513397216797,
      "learning_rate": 0.0002644122512432514,
      "loss": 2.1706,
      "step": 123922
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.0596446990966797,
      "learning_rate": 0.0002644081895677343,
      "loss": 2.1378,
      "step": 123923
    },
    {
      "epoch": 1.61,
      "grad_norm": 0.9474664330482483,
      "learning_rate": 0.0002644041278988343,
      "loss": 2.2702,
      "step": 123924
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.10475754737854,
      "learning_rate": 0.00026440006623655235,
      "loss": 1.9943,
      "step": 123925
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.0043952465057373,
      "learning_rate": 0.0002643960045808891,
      "loss": 2.3304,
      "step": 123926
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.1255682706832886,
      "learning_rate": 0.0002643919429318452,
      "loss": 2.296,
      "step": 123927
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.20957350730896,
      "learning_rate": 0.0002643878812894215,
      "loss": 2.1303,
      "step": 123928
    },
    {
      "epoch": 1.61,
      "grad_norm": 0.9925629496574402,
      "learning_rate": 0.00026438381965361883,
      "loss": 2.0141,
      "step": 123929
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.035937786102295,
      "learning_rate": 0.00026437975802443784,
      "loss": 2.1932,
      "step": 123930
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.1308039426803589,
      "learning_rate": 0.0002643756964018795,
      "loss": 2.1411,
      "step": 123931
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.1249570846557617,
      "learning_rate": 0.0002643716347859442,
      "loss": 1.9751,
      "step": 123932
    },
    {
      "epoch": 1.61,
      "grad_norm": 0.9462882876396179,
      "learning_rate": 0.00026436757317663295,
      "loss": 1.967,
      "step": 123933
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.0135908126831055,
      "learning_rate": 0.0002643635115739464,
      "loss": 2.2804,
      "step": 123934
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.0645045042037964,
      "learning_rate": 0.00026435944997788534,
      "loss": 2.1417,
      "step": 123935
    },
    {
      "epoch": 1.61,
      "grad_norm": 0.9384320378303528,
      "learning_rate": 0.00026435538838845055,
      "loss": 1.8661,
      "step": 123936
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.01376473903656,
      "learning_rate": 0.0002643513268056429,
      "loss": 2.1306,
      "step": 123937
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.00503408908844,
      "learning_rate": 0.00026434726522946286,
      "loss": 2.1136,
      "step": 123938
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.0677984952926636,
      "learning_rate": 0.00026434320365991136,
      "loss": 2.0955,
      "step": 123939
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.2151598930358887,
      "learning_rate": 0.0002643391420969891,
      "loss": 1.9349,
      "step": 123940
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.0505632162094116,
      "learning_rate": 0.0002643350805406969,
      "loss": 2.2279,
      "step": 123941
    },
    {
      "epoch": 1.61,
      "grad_norm": 0.9454624652862549,
      "learning_rate": 0.00026433101899103544,
      "loss": 2.0579,
      "step": 123942
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.2456587553024292,
      "learning_rate": 0.0002643269574480057,
      "loss": 2.0652,
      "step": 123943
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.239471197128296,
      "learning_rate": 0.00026432289591160813,
      "loss": 2.1742,
      "step": 123944
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.137353777885437,
      "learning_rate": 0.00026431883438184355,
      "loss": 2.1153,
      "step": 123945
    },
    {
      "epoch": 1.61,
      "grad_norm": 0.945015013217926,
      "learning_rate": 0.0002643147728587128,
      "loss": 2.1881,
      "step": 123946
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.2595876455307007,
      "learning_rate": 0.0002643107113422166,
      "loss": 2.2806,
      "step": 123947
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.130297064781189,
      "learning_rate": 0.0002643066498323557,
      "loss": 1.9675,
      "step": 123948
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.0921382904052734,
      "learning_rate": 0.000264302588329131,
      "loss": 2.1603,
      "step": 123949
    },
    {
      "epoch": 1.61,
      "grad_norm": 0.9169144034385681,
      "learning_rate": 0.0002642985268325431,
      "loss": 2.2006,
      "step": 123950
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.0283987522125244,
      "learning_rate": 0.0002642944653425926,
      "loss": 2.2124,
      "step": 123951
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.0828906297683716,
      "learning_rate": 0.0002642904038592805,
      "loss": 2.0679,
      "step": 123952
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.4824122190475464,
      "learning_rate": 0.00026428634238260753,
      "loss": 1.9096,
      "step": 123953
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.1332041025161743,
      "learning_rate": 0.0002642822809125744,
      "loss": 2.3257,
      "step": 123954
    },
    {
      "epoch": 1.61,
      "grad_norm": 0.9858804941177368,
      "learning_rate": 0.00026427821944918187,
      "loss": 2.0282,
      "step": 123955
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.3400713205337524,
      "learning_rate": 0.00026427415799243077,
      "loss": 2.0893,
      "step": 123956
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.0261458158493042,
      "learning_rate": 0.0002642700965423217,
      "loss": 2.2641,
      "step": 123957
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.4992715120315552,
      "learning_rate": 0.00026426603509885544,
      "loss": 2.1748,
      "step": 123958
    },
    {
      "epoch": 1.61,
      "grad_norm": 0.9906860589981079,
      "learning_rate": 0.00026426197366203285,
      "loss": 2.0461,
      "step": 123959
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.0038106441497803,
      "learning_rate": 0.0002642579122318546,
      "loss": 2.1276,
      "step": 123960
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.2323400974273682,
      "learning_rate": 0.0002642538508083216,
      "loss": 2.0942,
      "step": 123961
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.683856725692749,
      "learning_rate": 0.0002642497893914343,
      "loss": 2.0665,
      "step": 123962
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.0848723649978638,
      "learning_rate": 0.0002642457279811938,
      "loss": 2.2361,
      "step": 123963
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.0826911926269531,
      "learning_rate": 0.00026424166657760065,
      "loss": 2.3237,
      "step": 123964
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.1750478744506836,
      "learning_rate": 0.0002642376051806556,
      "loss": 2.3372,
      "step": 123965
    },
    {
      "epoch": 1.61,
      "grad_norm": 0.947282075881958,
      "learning_rate": 0.0002642335437903594,
      "loss": 2.153,
      "step": 123966
    },
    {
      "epoch": 1.61,
      "grad_norm": 0.8978086113929749,
      "learning_rate": 0.00026422948240671304,
      "loss": 2.2611,
      "step": 123967
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.0171432495117188,
      "learning_rate": 0.000264225421029717,
      "loss": 2.1772,
      "step": 123968
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.131110429763794,
      "learning_rate": 0.0002642213596593721,
      "loss": 2.214,
      "step": 123969
    },
    {
      "epoch": 1.61,
      "grad_norm": 0.9224056601524353,
      "learning_rate": 0.0002642172982956792,
      "loss": 2.0868,
      "step": 123970
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.002516508102417,
      "learning_rate": 0.00026421323693863886,
      "loss": 2.1275,
      "step": 123971
    },
    {
      "epoch": 1.61,
      "grad_norm": 0.8622390031814575,
      "learning_rate": 0.00026420917558825206,
      "loss": 2.0469,
      "step": 123972
    },
    {
      "epoch": 1.61,
      "grad_norm": 0.9558248519897461,
      "learning_rate": 0.00026420511424451944,
      "loss": 1.9499,
      "step": 123973
    },
    {
      "epoch": 1.61,
      "grad_norm": 0.9442727565765381,
      "learning_rate": 0.0002642010529074417,
      "loss": 2.4379,
      "step": 123974
    },
    {
      "epoch": 1.61,
      "grad_norm": 0.9680920839309692,
      "learning_rate": 0.0002641969915770197,
      "loss": 1.8914,
      "step": 123975
    },
    {
      "epoch": 1.61,
      "grad_norm": 0.9804885387420654,
      "learning_rate": 0.00026419293025325416,
      "loss": 2.1364,
      "step": 123976
    },
    {
      "epoch": 1.61,
      "grad_norm": 0.9470292925834656,
      "learning_rate": 0.00026418886893614585,
      "loss": 1.8977,
      "step": 123977
    },
    {
      "epoch": 1.61,
      "grad_norm": 0.8948706984519958,
      "learning_rate": 0.0002641848076256955,
      "loss": 1.9831,
      "step": 123978
    },
    {
      "epoch": 1.61,
      "grad_norm": 0.9806376695632935,
      "learning_rate": 0.0002641807463219038,
      "loss": 2.1036,
      "step": 123979
    },
    {
      "epoch": 1.61,
      "grad_norm": 0.9639816880226135,
      "learning_rate": 0.0002641766850247717,
      "loss": 1.9507,
      "step": 123980
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.0303025245666504,
      "learning_rate": 0.0002641726237342997,
      "loss": 2.1007,
      "step": 123981
    },
    {
      "epoch": 1.61,
      "grad_norm": 0.9895280003547668,
      "learning_rate": 0.00026416856245048875,
      "loss": 2.2916,
      "step": 123982
    },
    {
      "epoch": 1.61,
      "grad_norm": 0.9665994644165039,
      "learning_rate": 0.0002641645011733395,
      "loss": 2.0762,
      "step": 123983
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.1599090099334717,
      "learning_rate": 0.00026416043990285274,
      "loss": 2.1833,
      "step": 123984
    },
    {
      "epoch": 1.61,
      "grad_norm": 0.9544870853424072,
      "learning_rate": 0.00026415637863902935,
      "loss": 2.1878,
      "step": 123985
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.0440113544464111,
      "learning_rate": 0.0002641523173818699,
      "loss": 2.1241,
      "step": 123986
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.0114027261734009,
      "learning_rate": 0.0002641482561313752,
      "loss": 2.1253,
      "step": 123987
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.0194895267486572,
      "learning_rate": 0.00026414419488754594,
      "loss": 2.1666,
      "step": 123988
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.1396987438201904,
      "learning_rate": 0.000264140133650383,
      "loss": 2.0028,
      "step": 123989
    },
    {
      "epoch": 1.61,
      "grad_norm": 0.9931670427322388,
      "learning_rate": 0.00026413607241988714,
      "loss": 2.2662,
      "step": 123990
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.024261236190796,
      "learning_rate": 0.000264132011196059,
      "loss": 2.2133,
      "step": 123991
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.0601516962051392,
      "learning_rate": 0.00026412794997889954,
      "loss": 2.3103,
      "step": 123992
    },
    {
      "epoch": 1.61,
      "grad_norm": 0.9573837518692017,
      "learning_rate": 0.00026412388876840924,
      "loss": 2.1037,
      "step": 123993
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.1055370569229126,
      "learning_rate": 0.00026411982756458896,
      "loss": 2.293,
      "step": 123994
    },
    {
      "epoch": 1.61,
      "grad_norm": 0.889872133731842,
      "learning_rate": 0.0002641157663674395,
      "loss": 2.0577,
      "step": 123995
    },
    {
      "epoch": 1.61,
      "grad_norm": 0.9495548605918884,
      "learning_rate": 0.0002641117051769616,
      "loss": 1.9904,
      "step": 123996
    },
    {
      "epoch": 1.61,
      "grad_norm": 0.9686715602874756,
      "learning_rate": 0.000264107643993156,
      "loss": 2.2908,
      "step": 123997
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.0040956735610962,
      "learning_rate": 0.00026410358281602365,
      "loss": 1.9191,
      "step": 123998
    },
    {
      "epoch": 1.61,
      "grad_norm": 0.9678493142127991,
      "learning_rate": 0.0002640995216455649,
      "loss": 2.0917,
      "step": 123999
    },
    {
      "epoch": 1.61,
      "grad_norm": 0.9497789144515991,
      "learning_rate": 0.0002640954604817808,
      "loss": 2.0528,
      "step": 124000
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.072251558303833,
      "learning_rate": 0.00026409139932467206,
      "loss": 2.0736,
      "step": 124001
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.0693165063858032,
      "learning_rate": 0.00026408733817423935,
      "loss": 2.3542,
      "step": 124002
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.051232933998108,
      "learning_rate": 0.0002640832770304835,
      "loss": 2.0999,
      "step": 124003
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.0320309400558472,
      "learning_rate": 0.00026407921589340533,
      "loss": 2.2539,
      "step": 124004
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.0037559270858765,
      "learning_rate": 0.0002640751547630054,
      "loss": 2.0154,
      "step": 124005
    },
    {
      "epoch": 1.61,
      "grad_norm": 0.9924232363700867,
      "learning_rate": 0.0002640710936392846,
      "loss": 2.0763,
      "step": 124006
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.0025891065597534,
      "learning_rate": 0.0002640670325222437,
      "loss": 2.1484,
      "step": 124007
    },
    {
      "epoch": 1.61,
      "grad_norm": 0.8938438892364502,
      "learning_rate": 0.00026406297141188335,
      "loss": 2.1345,
      "step": 124008
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.044838547706604,
      "learning_rate": 0.00026405891030820437,
      "loss": 1.8344,
      "step": 124009
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.056425929069519,
      "learning_rate": 0.0002640548492112077,
      "loss": 2.3926,
      "step": 124010
    },
    {
      "epoch": 1.61,
      "grad_norm": 0.9890140891075134,
      "learning_rate": 0.00026405078812089374,
      "loss": 2.1341,
      "step": 124011
    },
    {
      "epoch": 1.61,
      "grad_norm": 0.9945623874664307,
      "learning_rate": 0.00026404672703726344,
      "loss": 2.2692,
      "step": 124012
    },
    {
      "epoch": 1.61,
      "grad_norm": 0.9891723394393921,
      "learning_rate": 0.0002640426659603175,
      "loss": 2.1832,
      "step": 124013
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.1625338792800903,
      "learning_rate": 0.00026403860489005674,
      "loss": 1.9607,
      "step": 124014
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.051822543144226,
      "learning_rate": 0.0002640345438264819,
      "loss": 2.0761,
      "step": 124015
    },
    {
      "epoch": 1.61,
      "grad_norm": 0.969965398311615,
      "learning_rate": 0.0002640304827695938,
      "loss": 2.2908,
      "step": 124016
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.0004994869232178,
      "learning_rate": 0.00026402642171939296,
      "loss": 2.1479,
      "step": 124017
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.0134499073028564,
      "learning_rate": 0.00026402236067588037,
      "loss": 2.3368,
      "step": 124018
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.1392903327941895,
      "learning_rate": 0.00026401829963905665,
      "loss": 1.9611,
      "step": 124019
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.0444509983062744,
      "learning_rate": 0.00026401423860892263,
      "loss": 2.0504,
      "step": 124020
    },
    {
      "epoch": 1.61,
      "grad_norm": 0.9633521437644958,
      "learning_rate": 0.00026401017758547905,
      "loss": 2.1591,
      "step": 124021
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.1679214239120483,
      "learning_rate": 0.00026400611656872674,
      "loss": 2.2412,
      "step": 124022
    },
    {
      "epoch": 1.61,
      "grad_norm": 0.9667047262191772,
      "learning_rate": 0.0002640020555586663,
      "loss": 2.187,
      "step": 124023
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.0047708749771118,
      "learning_rate": 0.0002639979945552985,
      "loss": 2.174,
      "step": 124024
    },
    {
      "epoch": 1.61,
      "grad_norm": 0.9483329653739929,
      "learning_rate": 0.0002639939335586242,
      "loss": 2.0385,
      "step": 124025
    },
    {
      "epoch": 1.61,
      "grad_norm": 0.9500004649162292,
      "learning_rate": 0.0002639898725686441,
      "loss": 1.9811,
      "step": 124026
    },
    {
      "epoch": 1.61,
      "grad_norm": 0.9560311436653137,
      "learning_rate": 0.0002639858115853589,
      "loss": 2.133,
      "step": 124027
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.1639870405197144,
      "learning_rate": 0.00026398175060876963,
      "loss": 2.0219,
      "step": 124028
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.033756971359253,
      "learning_rate": 0.00026397768963887664,
      "loss": 2.0268,
      "step": 124029
    },
    {
      "epoch": 1.61,
      "grad_norm": 0.9463180899620056,
      "learning_rate": 0.00026397362867568093,
      "loss": 2.1745,
      "step": 124030
    },
    {
      "epoch": 1.61,
      "grad_norm": 1.0576605796813965,
      "learning_rate": 0.00026396956771918315,
      "loss": 2.0786,
      "step": 124031
    },
    {
      "epoch": 1.61,
      "grad_norm": 0.9646104574203491,
      "learning_rate": 0.00026396550676938417,
      "loss": 1.9631,
      "step": 124032
    },
    {
      "epoch": 1.61,
      "grad_norm": 0.9436600208282471,
      "learning_rate": 0.00026396144582628464,
      "loss": 2.0914,
      "step": 124033
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.9731113910675049,
      "learning_rate": 0.0002639573848898855,
      "loss": 2.179,
      "step": 124034
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.0950921773910522,
      "learning_rate": 0.0002639533239601872,
      "loss": 2.1268,
      "step": 124035
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.0440677404403687,
      "learning_rate": 0.0002639492630371907,
      "loss": 2.1225,
      "step": 124036
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.0710692405700684,
      "learning_rate": 0.00026394520212089676,
      "loss": 2.3115,
      "step": 124037
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.9746182560920715,
      "learning_rate": 0.000263941141211306,
      "loss": 1.9758,
      "step": 124038
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.9012093544006348,
      "learning_rate": 0.00026393708030841926,
      "loss": 1.9486,
      "step": 124039
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.0359375476837158,
      "learning_rate": 0.0002639330194122375,
      "loss": 2.112,
      "step": 124040
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.112929344177246,
      "learning_rate": 0.0002639289585227611,
      "loss": 2.1886,
      "step": 124041
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.0516822338104248,
      "learning_rate": 0.00026392489763999095,
      "loss": 1.9611,
      "step": 124042
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.0124460458755493,
      "learning_rate": 0.0002639208367639279,
      "loss": 2.2982,
      "step": 124043
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.0942225456237793,
      "learning_rate": 0.0002639167758945726,
      "loss": 2.0939,
      "step": 124044
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.9662909507751465,
      "learning_rate": 0.0002639127150319259,
      "loss": 2.4036,
      "step": 124045
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.9855931401252747,
      "learning_rate": 0.0002639086541759885,
      "loss": 2.1757,
      "step": 124046
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.0152509212493896,
      "learning_rate": 0.00026390459332676123,
      "loss": 1.9014,
      "step": 124047
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.114001750946045,
      "learning_rate": 0.0002639005324842447,
      "loss": 1.99,
      "step": 124048
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.0232964754104614,
      "learning_rate": 0.00026389647164843977,
      "loss": 1.9778,
      "step": 124049
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.1275951862335205,
      "learning_rate": 0.00026389241081934713,
      "loss": 2.2016,
      "step": 124050
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.0229471921920776,
      "learning_rate": 0.0002638883499969676,
      "loss": 1.917,
      "step": 124051
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.9566738605499268,
      "learning_rate": 0.0002638842891813019,
      "loss": 2.1659,
      "step": 124052
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.1126375198364258,
      "learning_rate": 0.00026388022837235084,
      "loss": 2.0486,
      "step": 124053
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.2052394151687622,
      "learning_rate": 0.00026387616757011503,
      "loss": 2.0221,
      "step": 124054
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.0099155902862549,
      "learning_rate": 0.0002638721067745954,
      "loss": 2.4293,
      "step": 124055
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.8413160443305969,
      "learning_rate": 0.0002638680459857926,
      "loss": 2.3953,
      "step": 124056
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.9816348552703857,
      "learning_rate": 0.00026386398520370744,
      "loss": 2.1193,
      "step": 124057
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.0642858743667603,
      "learning_rate": 0.0002638599244283406,
      "loss": 2.0934,
      "step": 124058
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.868581235408783,
      "learning_rate": 0.00026385586365969293,
      "loss": 2.1445,
      "step": 124059
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.984294593334198,
      "learning_rate": 0.0002638518028977651,
      "loss": 1.9587,
      "step": 124060
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.1654527187347412,
      "learning_rate": 0.0002638477421425579,
      "loss": 2.3702,
      "step": 124061
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.8773828744888306,
      "learning_rate": 0.0002638436813940721,
      "loss": 2.1429,
      "step": 124062
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.1003074645996094,
      "learning_rate": 0.0002638396206523085,
      "loss": 2.0957,
      "step": 124063
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.1261762380599976,
      "learning_rate": 0.00026383555991726774,
      "loss": 2.1434,
      "step": 124064
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.0170689821243286,
      "learning_rate": 0.00026383149918895066,
      "loss": 2.286,
      "step": 124065
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.0771238803863525,
      "learning_rate": 0.0002638274384673579,
      "loss": 2.2625,
      "step": 124066
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.9429107904434204,
      "learning_rate": 0.00026382337775249036,
      "loss": 1.9837,
      "step": 124067
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.000536561012268,
      "learning_rate": 0.0002638193170443487,
      "loss": 2.2024,
      "step": 124068
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.9486889243125916,
      "learning_rate": 0.00026381525634293384,
      "loss": 2.0846,
      "step": 124069
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.9122260808944702,
      "learning_rate": 0.00026381119564824633,
      "loss": 2.1559,
      "step": 124070
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.9371654987335205,
      "learning_rate": 0.00026380713496028705,
      "loss": 2.1612,
      "step": 124071
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.083143711090088,
      "learning_rate": 0.0002638030742790566,
      "loss": 2.2171,
      "step": 124072
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.925293505191803,
      "learning_rate": 0.00026379901360455585,
      "loss": 2.044,
      "step": 124073
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.0780082941055298,
      "learning_rate": 0.00026379495293678555,
      "loss": 2.2065,
      "step": 124074
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.0191444158554077,
      "learning_rate": 0.0002637908922757465,
      "loss": 2.2063,
      "step": 124075
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.0298651456832886,
      "learning_rate": 0.00026378683162143937,
      "loss": 2.0659,
      "step": 124076
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.9815143346786499,
      "learning_rate": 0.0002637827709738651,
      "loss": 1.9863,
      "step": 124077
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.9670014381408691,
      "learning_rate": 0.00026377871033302417,
      "loss": 1.8897,
      "step": 124078
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.9725973010063171,
      "learning_rate": 0.0002637746496989174,
      "loss": 2.2482,
      "step": 124079
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.8425809741020203,
      "learning_rate": 0.0002637705890715457,
      "loss": 2.0931,
      "step": 124080
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.9815573692321777,
      "learning_rate": 0.0002637665284509097,
      "loss": 2.0954,
      "step": 124081
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.2318966388702393,
      "learning_rate": 0.00026376246783701014,
      "loss": 2.0687,
      "step": 124082
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.9682741761207581,
      "learning_rate": 0.000263758407229848,
      "loss": 2.205,
      "step": 124083
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.073460340499878,
      "learning_rate": 0.0002637543466294237,
      "loss": 2.2386,
      "step": 124084
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.9532096982002258,
      "learning_rate": 0.00026375028603573816,
      "loss": 2.3907,
      "step": 124085
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.9399406909942627,
      "learning_rate": 0.0002637462254487921,
      "loss": 2.3599,
      "step": 124086
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.9871684312820435,
      "learning_rate": 0.00026374216486858636,
      "loss": 2.0525,
      "step": 124087
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.9307203888893127,
      "learning_rate": 0.00026373810429512154,
      "loss": 2.2935,
      "step": 124088
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.9643425941467285,
      "learning_rate": 0.00026373404372839873,
      "loss": 2.2862,
      "step": 124089
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.0087271928787231,
      "learning_rate": 0.00026372998316841824,
      "loss": 2.2133,
      "step": 124090
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.9578002095222473,
      "learning_rate": 0.00026372592261518106,
      "loss": 2.1506,
      "step": 124091
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.067599892616272,
      "learning_rate": 0.00026372186206868794,
      "loss": 1.9085,
      "step": 124092
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.8953917622566223,
      "learning_rate": 0.0002637178015289396,
      "loss": 2.1819,
      "step": 124093
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.9145203232765198,
      "learning_rate": 0.00026371374099593676,
      "loss": 2.3541,
      "step": 124094
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.9541140198707581,
      "learning_rate": 0.0002637096804696804,
      "loss": 2.1236,
      "step": 124095
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.9760487675666809,
      "learning_rate": 0.00026370561995017094,
      "loss": 2.0701,
      "step": 124096
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.0514169931411743,
      "learning_rate": 0.0002637015594374093,
      "loss": 2.2324,
      "step": 124097
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.065477967262268,
      "learning_rate": 0.0002636974989313962,
      "loss": 2.1406,
      "step": 124098
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.0426162481307983,
      "learning_rate": 0.00026369343843213246,
      "loss": 2.171,
      "step": 124099
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.9405046105384827,
      "learning_rate": 0.0002636893779396188,
      "loss": 2.1769,
      "step": 124100
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.1822980642318726,
      "learning_rate": 0.00026368531745385606,
      "loss": 2.0291,
      "step": 124101
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.9520966410636902,
      "learning_rate": 0.0002636812569748448,
      "loss": 2.0565,
      "step": 124102
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.9289117455482483,
      "learning_rate": 0.0002636771965025859,
      "loss": 2.1624,
      "step": 124103
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.3101333379745483,
      "learning_rate": 0.00026367313603708006,
      "loss": 1.9934,
      "step": 124104
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.9559930562973022,
      "learning_rate": 0.00026366907557832806,
      "loss": 2.096,
      "step": 124105
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.0567113161087036,
      "learning_rate": 0.0002636650151263307,
      "loss": 2.1947,
      "step": 124106
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.1526978015899658,
      "learning_rate": 0.0002636609546810888,
      "loss": 2.0408,
      "step": 124107
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.1400001049041748,
      "learning_rate": 0.0002636568942426029,
      "loss": 2.32,
      "step": 124108
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.9786403775215149,
      "learning_rate": 0.0002636528338108738,
      "loss": 2.1935,
      "step": 124109
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.9047743082046509,
      "learning_rate": 0.0002636487733859024,
      "loss": 2.1449,
      "step": 124110
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.8988699316978455,
      "learning_rate": 0.0002636447129676894,
      "loss": 1.8713,
      "step": 124111
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.2744643688201904,
      "learning_rate": 0.0002636406525562355,
      "loss": 1.9836,
      "step": 124112
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.1175510883331299,
      "learning_rate": 0.0002636365921515415,
      "loss": 2.4181,
      "step": 124113
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.9469850063323975,
      "learning_rate": 0.0002636325317536083,
      "loss": 2.1654,
      "step": 124114
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.3151867389678955,
      "learning_rate": 0.0002636284713624363,
      "loss": 2.0468,
      "step": 124115
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.3907698392868042,
      "learning_rate": 0.00026362441097802647,
      "loss": 2.1506,
      "step": 124116
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.9399356245994568,
      "learning_rate": 0.00026362035060037957,
      "loss": 1.8288,
      "step": 124117
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.0662411451339722,
      "learning_rate": 0.00026361629022949635,
      "loss": 1.9856,
      "step": 124118
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.2073299884796143,
      "learning_rate": 0.00026361222986537753,
      "loss": 2.0079,
      "step": 124119
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.1227344274520874,
      "learning_rate": 0.000263608169508024,
      "loss": 2.095,
      "step": 124120
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.9434711337089539,
      "learning_rate": 0.0002636041091574363,
      "loss": 2.055,
      "step": 124121
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.047471046447754,
      "learning_rate": 0.00026360004881361527,
      "loss": 2.0026,
      "step": 124122
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.013992428779602,
      "learning_rate": 0.00026359598847656166,
      "loss": 1.9377,
      "step": 124123
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.9718604683876038,
      "learning_rate": 0.0002635919281462763,
      "loss": 2.1476,
      "step": 124124
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.0104022026062012,
      "learning_rate": 0.0002635878678227598,
      "loss": 2.0482,
      "step": 124125
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.9087852239608765,
      "learning_rate": 0.0002635838075060132,
      "loss": 2.1791,
      "step": 124126
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.1772645711898804,
      "learning_rate": 0.0002635797471960369,
      "loss": 2.0641,
      "step": 124127
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.9877912998199463,
      "learning_rate": 0.0002635756868928318,
      "loss": 2.0838,
      "step": 124128
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.3863307237625122,
      "learning_rate": 0.00026357162659639867,
      "loss": 1.9115,
      "step": 124129
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.9183745980262756,
      "learning_rate": 0.0002635675663067383,
      "loss": 1.8748,
      "step": 124130
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.9453614950180054,
      "learning_rate": 0.00026356350602385137,
      "loss": 2.0159,
      "step": 124131
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.9684022068977356,
      "learning_rate": 0.0002635594457477388,
      "loss": 2.2245,
      "step": 124132
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.9363076686859131,
      "learning_rate": 0.0002635553854784011,
      "loss": 2.1537,
      "step": 124133
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.9726517796516418,
      "learning_rate": 0.0002635513252158392,
      "loss": 2.1088,
      "step": 124134
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.9937934875488281,
      "learning_rate": 0.0002635472649600537,
      "loss": 2.1074,
      "step": 124135
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.8618540167808533,
      "learning_rate": 0.0002635432047110455,
      "loss": 1.7838,
      "step": 124136
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.9003787040710449,
      "learning_rate": 0.0002635391444688153,
      "loss": 2.2018,
      "step": 124137
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.8870110511779785,
      "learning_rate": 0.0002635350842333639,
      "loss": 1.9082,
      "step": 124138
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.972602367401123,
      "learning_rate": 0.000263531024004692,
      "loss": 2.3174,
      "step": 124139
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.9611502289772034,
      "learning_rate": 0.0002635269637828004,
      "loss": 2.2914,
      "step": 124140
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.9440805912017822,
      "learning_rate": 0.0002635229035676898,
      "loss": 2.3385,
      "step": 124141
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.9545348286628723,
      "learning_rate": 0.00026351884335936094,
      "loss": 2.1531,
      "step": 124142
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.006391167640686,
      "learning_rate": 0.0002635147831578146,
      "loss": 2.1363,
      "step": 124143
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.8963447213172913,
      "learning_rate": 0.0002635107229630517,
      "loss": 2.3958,
      "step": 124144
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.111158847808838,
      "learning_rate": 0.00026350666277507267,
      "loss": 2.2105,
      "step": 124145
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.271173119544983,
      "learning_rate": 0.00026350260259387854,
      "loss": 2.0664,
      "step": 124146
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.0563117265701294,
      "learning_rate": 0.00026349854241946994,
      "loss": 2.081,
      "step": 124147
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.1691213846206665,
      "learning_rate": 0.00026349448225184764,
      "loss": 2.0195,
      "step": 124148
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.2405143976211548,
      "learning_rate": 0.0002634904220910124,
      "loss": 2.1001,
      "step": 124149
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.0244563817977905,
      "learning_rate": 0.00026348636193696505,
      "loss": 2.1538,
      "step": 124150
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.9228740334510803,
      "learning_rate": 0.0002634823017897062,
      "loss": 2.217,
      "step": 124151
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.975879967212677,
      "learning_rate": 0.0002634782416492367,
      "loss": 2.4741,
      "step": 124152
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.1394789218902588,
      "learning_rate": 0.00026347418151555723,
      "loss": 2.0223,
      "step": 124153
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.9904438257217407,
      "learning_rate": 0.00026347012138866867,
      "loss": 2.3334,
      "step": 124154
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.9197940230369568,
      "learning_rate": 0.0002634660612685717,
      "loss": 1.8602,
      "step": 124155
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.996694803237915,
      "learning_rate": 0.0002634620011552671,
      "loss": 2.0758,
      "step": 124156
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.9073341488838196,
      "learning_rate": 0.00026345794104875556,
      "loss": 2.3224,
      "step": 124157
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.00579035282135,
      "learning_rate": 0.00026345388094903787,
      "loss": 2.1542,
      "step": 124158
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.030480980873108,
      "learning_rate": 0.00026344982085611476,
      "loss": 1.8835,
      "step": 124159
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.8902440667152405,
      "learning_rate": 0.00026344576076998706,
      "loss": 2.2234,
      "step": 124160
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.9872318506240845,
      "learning_rate": 0.00026344170069065556,
      "loss": 2.2638,
      "step": 124161
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.0309268236160278,
      "learning_rate": 0.0002634376406181209,
      "loss": 2.0613,
      "step": 124162
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.0546698570251465,
      "learning_rate": 0.0002634335805523838,
      "loss": 2.1537,
      "step": 124163
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.9629973769187927,
      "learning_rate": 0.0002634295204934451,
      "loss": 2.2969,
      "step": 124164
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.3969100713729858,
      "learning_rate": 0.0002634254604413056,
      "loss": 2.2376,
      "step": 124165
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.9344180822372437,
      "learning_rate": 0.0002634214003959659,
      "loss": 2.3146,
      "step": 124166
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.0089881420135498,
      "learning_rate": 0.0002634173403574269,
      "loss": 1.9746,
      "step": 124167
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.981666624546051,
      "learning_rate": 0.00026341328032568947,
      "loss": 2.0914,
      "step": 124168
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.047203779220581,
      "learning_rate": 0.000263409220300754,
      "loss": 2.0528,
      "step": 124169
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.9914535284042358,
      "learning_rate": 0.0002634051602826215,
      "loss": 2.1922,
      "step": 124170
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.9842779040336609,
      "learning_rate": 0.0002634011002712927,
      "loss": 2.0367,
      "step": 124171
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.134076476097107,
      "learning_rate": 0.0002633970402667683,
      "loss": 2.1304,
      "step": 124172
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.016594409942627,
      "learning_rate": 0.000263392980269049,
      "loss": 1.9616,
      "step": 124173
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.9435486793518066,
      "learning_rate": 0.00026338892027813585,
      "loss": 2.2766,
      "step": 124174
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.055498719215393,
      "learning_rate": 0.00026338486029402924,
      "loss": 1.9898,
      "step": 124175
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.990485429763794,
      "learning_rate": 0.0002633808003167301,
      "loss": 2.2046,
      "step": 124176
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.175583004951477,
      "learning_rate": 0.00026337674034623913,
      "loss": 1.9868,
      "step": 124177
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.255277156829834,
      "learning_rate": 0.0002633726803825571,
      "loss": 2.164,
      "step": 124178
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.9609309434890747,
      "learning_rate": 0.00026336862042568483,
      "loss": 2.3944,
      "step": 124179
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.0452475547790527,
      "learning_rate": 0.00026336456047562303,
      "loss": 2.268,
      "step": 124180
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.137580394744873,
      "learning_rate": 0.0002633605005323725,
      "loss": 2.0938,
      "step": 124181
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.0479291677474976,
      "learning_rate": 0.0002633564405959339,
      "loss": 2.2152,
      "step": 124182
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.8981943130493164,
      "learning_rate": 0.00026335238066630795,
      "loss": 1.8509,
      "step": 124183
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.0427813529968262,
      "learning_rate": 0.00026334832074349557,
      "loss": 1.9295,
      "step": 124184
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.0611772537231445,
      "learning_rate": 0.0002633442608274974,
      "loss": 2.0329,
      "step": 124185
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.9588990807533264,
      "learning_rate": 0.0002633402009183142,
      "loss": 2.1241,
      "step": 124186
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.9289127588272095,
      "learning_rate": 0.0002633361410159469,
      "loss": 2.0055,
      "step": 124187
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.0803842544555664,
      "learning_rate": 0.00026333208112039596,
      "loss": 1.9813,
      "step": 124188
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.9660174250602722,
      "learning_rate": 0.00026332802123166227,
      "loss": 1.9832,
      "step": 124189
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.9174315333366394,
      "learning_rate": 0.0002633239613497466,
      "loss": 1.982,
      "step": 124190
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.013917088508606,
      "learning_rate": 0.00026331990147464973,
      "loss": 1.9391,
      "step": 124191
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.9333723187446594,
      "learning_rate": 0.0002633158416063723,
      "loss": 2.0111,
      "step": 124192
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.9973140358924866,
      "learning_rate": 0.00026331178174491534,
      "loss": 2.2157,
      "step": 124193
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.9951413869857788,
      "learning_rate": 0.0002633077218902793,
      "loss": 2.0858,
      "step": 124194
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.9655771851539612,
      "learning_rate": 0.000263303662042465,
      "loss": 2.1513,
      "step": 124195
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.0689964294433594,
      "learning_rate": 0.00026329960220147325,
      "loss": 2.0599,
      "step": 124196
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.9476473927497864,
      "learning_rate": 0.0002632955423673048,
      "loss": 2.0781,
      "step": 124197
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.9039055705070496,
      "learning_rate": 0.00026329148253996043,
      "loss": 2.2328,
      "step": 124198
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.9198465347290039,
      "learning_rate": 0.000263287422719441,
      "loss": 2.0697,
      "step": 124199
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.036709189414978,
      "learning_rate": 0.00026328336290574696,
      "loss": 2.0193,
      "step": 124200
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.9887097477912903,
      "learning_rate": 0.00026327930309887927,
      "loss": 2.0142,
      "step": 124201
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.0672357082366943,
      "learning_rate": 0.00026327524329883864,
      "loss": 2.0095,
      "step": 124202
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.9542161822319031,
      "learning_rate": 0.0002632711835056258,
      "loss": 2.2208,
      "step": 124203
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.892600953578949,
      "learning_rate": 0.0002632671237192416,
      "loss": 2.126,
      "step": 124204
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.9371619820594788,
      "learning_rate": 0.00026326306393968684,
      "loss": 1.9844,
      "step": 124205
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.9166087508201599,
      "learning_rate": 0.00026325900416696203,
      "loss": 2.1159,
      "step": 124206
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.8888590931892395,
      "learning_rate": 0.0002632549444010681,
      "loss": 2.1125,
      "step": 124207
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.983363151550293,
      "learning_rate": 0.00026325088464200573,
      "loss": 2.1473,
      "step": 124208
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.0378187894821167,
      "learning_rate": 0.0002632468248897757,
      "loss": 2.1085,
      "step": 124209
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.9767741560935974,
      "learning_rate": 0.00026324276514437886,
      "loss": 2.3983,
      "step": 124210
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.8494842648506165,
      "learning_rate": 0.0002632387054058159,
      "loss": 1.9158,
      "step": 124211
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.9858022928237915,
      "learning_rate": 0.0002632346456740875,
      "loss": 1.918,
      "step": 124212
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.9300514459609985,
      "learning_rate": 0.0002632305859491945,
      "loss": 1.952,
      "step": 124213
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.9524723291397095,
      "learning_rate": 0.0002632265262311375,
      "loss": 2.1278,
      "step": 124214
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.034742832183838,
      "learning_rate": 0.0002632224665199175,
      "loss": 2.1803,
      "step": 124215
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.9480178952217102,
      "learning_rate": 0.0002632184068155351,
      "loss": 2.0732,
      "step": 124216
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.9835008382797241,
      "learning_rate": 0.00026321434711799126,
      "loss": 2.1396,
      "step": 124217
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.0719341039657593,
      "learning_rate": 0.0002632102874272864,
      "loss": 2.0779,
      "step": 124218
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.984372079372406,
      "learning_rate": 0.00026320622774342144,
      "loss": 1.867,
      "step": 124219
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.9932206869125366,
      "learning_rate": 0.0002632021680663972,
      "loss": 2.1995,
      "step": 124220
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.0431318283081055,
      "learning_rate": 0.0002631981083962143,
      "loss": 2.1371,
      "step": 124221
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.0592256784439087,
      "learning_rate": 0.00026319404873287355,
      "loss": 2.1305,
      "step": 124222
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.0009692907333374,
      "learning_rate": 0.0002631899890763758,
      "loss": 2.175,
      "step": 124223
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.9593073129653931,
      "learning_rate": 0.00026318592942672176,
      "loss": 1.9911,
      "step": 124224
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.0277966260910034,
      "learning_rate": 0.0002631818697839121,
      "loss": 2.1249,
      "step": 124225
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.8806419372558594,
      "learning_rate": 0.0002631778101479476,
      "loss": 2.1151,
      "step": 124226
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.0385947227478027,
      "learning_rate": 0.00026317375051882906,
      "loss": 2.1503,
      "step": 124227
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.9622021913528442,
      "learning_rate": 0.00026316969089655714,
      "loss": 1.8511,
      "step": 124228
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.1354196071624756,
      "learning_rate": 0.00026316563128113283,
      "loss": 2.2114,
      "step": 124229
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.2372674942016602,
      "learning_rate": 0.0002631615716725566,
      "loss": 1.9989,
      "step": 124230
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.0062872171401978,
      "learning_rate": 0.0002631575120708294,
      "loss": 2.0519,
      "step": 124231
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.940470278263092,
      "learning_rate": 0.00026315345247595185,
      "loss": 2.1738,
      "step": 124232
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.9945669174194336,
      "learning_rate": 0.0002631493928879248,
      "loss": 2.203,
      "step": 124233
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.2185741662979126,
      "learning_rate": 0.00026314533330674896,
      "loss": 1.929,
      "step": 124234
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.0515215396881104,
      "learning_rate": 0.0002631412737324251,
      "loss": 1.9735,
      "step": 124235
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.9391090869903564,
      "learning_rate": 0.00026313721416495404,
      "loss": 2.1174,
      "step": 124236
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.143114686012268,
      "learning_rate": 0.00026313315460433636,
      "loss": 2.2937,
      "step": 124237
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.1653327941894531,
      "learning_rate": 0.00026312909505057296,
      "loss": 2.0792,
      "step": 124238
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.127931833267212,
      "learning_rate": 0.00026312503550366464,
      "loss": 2.2337,
      "step": 124239
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.0237066745758057,
      "learning_rate": 0.00026312097596361194,
      "loss": 1.9815,
      "step": 124240
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.2979884147644043,
      "learning_rate": 0.00026311691643041585,
      "loss": 2.1077,
      "step": 124241
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.2977159023284912,
      "learning_rate": 0.00026311285690407696,
      "loss": 2.3672,
      "step": 124242
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.1405342817306519,
      "learning_rate": 0.00026310879738459614,
      "loss": 2.2148,
      "step": 124243
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.9413834810256958,
      "learning_rate": 0.000263104737871974,
      "loss": 2.1615,
      "step": 124244
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.064468502998352,
      "learning_rate": 0.00026310067836621146,
      "loss": 2.2189,
      "step": 124245
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.9629813432693481,
      "learning_rate": 0.0002630966188673092,
      "loss": 2.0362,
      "step": 124246
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.3051785230636597,
      "learning_rate": 0.00026309255937526793,
      "loss": 2.1254,
      "step": 124247
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.9201588034629822,
      "learning_rate": 0.0002630884998900885,
      "loss": 2.0243,
      "step": 124248
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.01539146900177,
      "learning_rate": 0.00026308444041177163,
      "loss": 1.925,
      "step": 124249
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.3620492219924927,
      "learning_rate": 0.000263080380940318,
      "loss": 2.2068,
      "step": 124250
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.1025234460830688,
      "learning_rate": 0.0002630763214757284,
      "loss": 2.3304,
      "step": 124251
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.9674279093742371,
      "learning_rate": 0.00026307226201800364,
      "loss": 2.2297,
      "step": 124252
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.8642529249191284,
      "learning_rate": 0.0002630682025671445,
      "loss": 2.2553,
      "step": 124253
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.2321619987487793,
      "learning_rate": 0.0002630641431231517,
      "loss": 2.2045,
      "step": 124254
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.104585886001587,
      "learning_rate": 0.0002630600836860259,
      "loss": 2.2968,
      "step": 124255
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.2120873928070068,
      "learning_rate": 0.0002630560242557679,
      "loss": 2.1334,
      "step": 124256
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.9286442399024963,
      "learning_rate": 0.00026305196483237853,
      "loss": 1.8479,
      "step": 124257
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.0284817218780518,
      "learning_rate": 0.0002630479054158585,
      "loss": 2.2956,
      "step": 124258
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.0788466930389404,
      "learning_rate": 0.00026304384600620847,
      "loss": 2.2343,
      "step": 124259
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.1154112815856934,
      "learning_rate": 0.0002630397866034295,
      "loss": 2.3298,
      "step": 124260
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.9359657168388367,
      "learning_rate": 0.00026303572720752196,
      "loss": 2.2403,
      "step": 124261
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.9459574818611145,
      "learning_rate": 0.0002630316678184868,
      "loss": 2.1559,
      "step": 124262
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.9307027459144592,
      "learning_rate": 0.00026302760843632475,
      "loss": 2.2438,
      "step": 124263
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.9135102033615112,
      "learning_rate": 0.0002630235490610365,
      "loss": 1.9747,
      "step": 124264
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.9386559724807739,
      "learning_rate": 0.000263019489692623,
      "loss": 2.0099,
      "step": 124265
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.9672940969467163,
      "learning_rate": 0.0002630154303310849,
      "loss": 2.022,
      "step": 124266
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.9719998836517334,
      "learning_rate": 0.0002630113709764228,
      "loss": 2.133,
      "step": 124267
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.1142246723175049,
      "learning_rate": 0.00026300731162863766,
      "loss": 2.3425,
      "step": 124268
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.1124298572540283,
      "learning_rate": 0.00026300325228773014,
      "loss": 2.1927,
      "step": 124269
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.9394843578338623,
      "learning_rate": 0.000262999192953701,
      "loss": 2.0746,
      "step": 124270
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.8864338994026184,
      "learning_rate": 0.000262995133626551,
      "loss": 2.1187,
      "step": 124271
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.9297720193862915,
      "learning_rate": 0.00026299107430628103,
      "loss": 2.1373,
      "step": 124272
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.3678730726242065,
      "learning_rate": 0.0002629870149928916,
      "loss": 2.0277,
      "step": 124273
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.8967231512069702,
      "learning_rate": 0.0002629829556863836,
      "loss": 1.9859,
      "step": 124274
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.0447133779525757,
      "learning_rate": 0.0002629788963867577,
      "loss": 2.1039,
      "step": 124275
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.3145633935928345,
      "learning_rate": 0.00026297483709401473,
      "loss": 2.1083,
      "step": 124276
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.057427167892456,
      "learning_rate": 0.0002629707778081555,
      "loss": 1.9359,
      "step": 124277
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.9588988423347473,
      "learning_rate": 0.0002629667185291808,
      "loss": 2.0131,
      "step": 124278
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.0568413734436035,
      "learning_rate": 0.00026296265925709113,
      "loss": 2.1583,
      "step": 124279
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.0366168022155762,
      "learning_rate": 0.0002629585999918875,
      "loss": 2.1187,
      "step": 124280
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.8717286586761475,
      "learning_rate": 0.00026295454073357047,
      "loss": 2.1387,
      "step": 124281
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.9050670862197876,
      "learning_rate": 0.0002629504814821409,
      "loss": 2.3494,
      "step": 124282
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.9988120794296265,
      "learning_rate": 0.00026294642223759955,
      "loss": 2.1431,
      "step": 124283
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.9905871152877808,
      "learning_rate": 0.0002629423629999473,
      "loss": 2.2289,
      "step": 124284
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.9826682806015015,
      "learning_rate": 0.0002629383037691846,
      "loss": 2.1835,
      "step": 124285
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.9723667502403259,
      "learning_rate": 0.0002629342445453124,
      "loss": 2.2413,
      "step": 124286
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.9626964330673218,
      "learning_rate": 0.00026293018532833144,
      "loss": 2.0584,
      "step": 124287
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.0296095609664917,
      "learning_rate": 0.00026292612611824245,
      "loss": 2.3899,
      "step": 124288
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.2022591829299927,
      "learning_rate": 0.00026292206691504616,
      "loss": 2.009,
      "step": 124289
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.9022222757339478,
      "learning_rate": 0.0002629180077187435,
      "loss": 1.9703,
      "step": 124290
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.9642624258995056,
      "learning_rate": 0.000262913948529335,
      "loss": 1.8837,
      "step": 124291
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.0973962545394897,
      "learning_rate": 0.0002629098893468214,
      "loss": 2.1397,
      "step": 124292
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.9235619306564331,
      "learning_rate": 0.0002629058301712036,
      "loss": 2.0642,
      "step": 124293
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.9204591512680054,
      "learning_rate": 0.00026290177100248235,
      "loss": 2.2509,
      "step": 124294
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.918747067451477,
      "learning_rate": 0.0002628977118406583,
      "loss": 2.0089,
      "step": 124295
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.1131476163864136,
      "learning_rate": 0.00026289365268573244,
      "loss": 1.8927,
      "step": 124296
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.0385886430740356,
      "learning_rate": 0.0002628895935377052,
      "loss": 2.2871,
      "step": 124297
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.9282968044281006,
      "learning_rate": 0.0002628855343965775,
      "loss": 2.0288,
      "step": 124298
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.9451037645339966,
      "learning_rate": 0.00026288147526235,
      "loss": 2.1638,
      "step": 124299
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.1080358028411865,
      "learning_rate": 0.00026287741613502364,
      "loss": 2.2947,
      "step": 124300
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.025888442993164,
      "learning_rate": 0.000262873357014599,
      "loss": 2.4465,
      "step": 124301
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.9788074493408203,
      "learning_rate": 0.00026286929790107706,
      "loss": 2.082,
      "step": 124302
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.0114277601242065,
      "learning_rate": 0.0002628652387944583,
      "loss": 2.1961,
      "step": 124303
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.9234777688980103,
      "learning_rate": 0.0002628611796947436,
      "loss": 2.0599,
      "step": 124304
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.9700713753700256,
      "learning_rate": 0.0002628571206019337,
      "loss": 2.1289,
      "step": 124305
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.9892757534980774,
      "learning_rate": 0.0002628530615160293,
      "loss": 2.3343,
      "step": 124306
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.953916609287262,
      "learning_rate": 0.0002628490024370313,
      "loss": 2.0677,
      "step": 124307
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.0719033479690552,
      "learning_rate": 0.0002628449433649404,
      "loss": 2.0878,
      "step": 124308
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.00295889377594,
      "learning_rate": 0.00026284088429975736,
      "loss": 2.048,
      "step": 124309
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.0309151411056519,
      "learning_rate": 0.0002628368252414828,
      "loss": 2.4115,
      "step": 124310
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.9337518811225891,
      "learning_rate": 0.0002628327661901176,
      "loss": 2.0733,
      "step": 124311
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.115714192390442,
      "learning_rate": 0.00026282870714566245,
      "loss": 2.116,
      "step": 124312
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.9806050062179565,
      "learning_rate": 0.00026282464810811816,
      "loss": 2.1059,
      "step": 124313
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.9160363674163818,
      "learning_rate": 0.0002628205890774855,
      "loss": 2.1054,
      "step": 124314
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.9923715591430664,
      "learning_rate": 0.00026281653005376523,
      "loss": 1.8635,
      "step": 124315
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.9405237436294556,
      "learning_rate": 0.0002628124710369581,
      "loss": 1.903,
      "step": 124316
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.1253355741500854,
      "learning_rate": 0.0002628084120270647,
      "loss": 2.2558,
      "step": 124317
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.9848929047584534,
      "learning_rate": 0.000262804353024086,
      "loss": 2.09,
      "step": 124318
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.9477418661117554,
      "learning_rate": 0.0002628002940280226,
      "loss": 1.9445,
      "step": 124319
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.1451332569122314,
      "learning_rate": 0.0002627962350388754,
      "loss": 1.9266,
      "step": 124320
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.9945981502532959,
      "learning_rate": 0.00026279217605664515,
      "loss": 2.0012,
      "step": 124321
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.0422351360321045,
      "learning_rate": 0.0002627881170813324,
      "loss": 2.1053,
      "step": 124322
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.9703759551048279,
      "learning_rate": 0.00026278405811293814,
      "loss": 2.0789,
      "step": 124323
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.063483476638794,
      "learning_rate": 0.000262779999151463,
      "loss": 2.2198,
      "step": 124324
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.0219415426254272,
      "learning_rate": 0.00026277594019690766,
      "loss": 2.1082,
      "step": 124325
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.9015040993690491,
      "learning_rate": 0.00026277188124927307,
      "loss": 2.0472,
      "step": 124326
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.910137414932251,
      "learning_rate": 0.0002627678223085599,
      "loss": 1.8558,
      "step": 124327
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.9725714921951294,
      "learning_rate": 0.00026276376337476884,
      "loss": 2.2849,
      "step": 124328
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.953483521938324,
      "learning_rate": 0.0002627597044479008,
      "loss": 2.2459,
      "step": 124329
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.9052962064743042,
      "learning_rate": 0.0002627556455279563,
      "loss": 2.1733,
      "step": 124330
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.0237481594085693,
      "learning_rate": 0.0002627515866149363,
      "loss": 2.1835,
      "step": 124331
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.0420936346054077,
      "learning_rate": 0.0002627475277088415,
      "loss": 2.2603,
      "step": 124332
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.01209557056427,
      "learning_rate": 0.00026274346880967263,
      "loss": 1.9653,
      "step": 124333
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.0641076564788818,
      "learning_rate": 0.00026273940991743046,
      "loss": 2.1308,
      "step": 124334
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.114227056503296,
      "learning_rate": 0.00026273535103211565,
      "loss": 2.0464,
      "step": 124335
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.3411155939102173,
      "learning_rate": 0.0002627312921537291,
      "loss": 2.1257,
      "step": 124336
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.0072485208511353,
      "learning_rate": 0.0002627272332822715,
      "loss": 1.9706,
      "step": 124337
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.0700613260269165,
      "learning_rate": 0.0002627231744177437,
      "loss": 2.1105,
      "step": 124338
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.207288146018982,
      "learning_rate": 0.0002627191155601463,
      "loss": 1.9406,
      "step": 124339
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.0168471336364746,
      "learning_rate": 0.0002627150567094801,
      "loss": 2.1031,
      "step": 124340
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.9337567687034607,
      "learning_rate": 0.00026271099786574585,
      "loss": 2.1015,
      "step": 124341
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.0900050401687622,
      "learning_rate": 0.00026270693902894437,
      "loss": 2.289,
      "step": 124342
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.252309799194336,
      "learning_rate": 0.00026270288019907635,
      "loss": 2.2181,
      "step": 124343
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.9493221044540405,
      "learning_rate": 0.00026269882137614257,
      "loss": 1.8866,
      "step": 124344
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.083940029144287,
      "learning_rate": 0.0002626947625601439,
      "loss": 2.2341,
      "step": 124345
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.0018573999404907,
      "learning_rate": 0.00026269070375108084,
      "loss": 2.1389,
      "step": 124346
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.0594096183776855,
      "learning_rate": 0.0002626866449489543,
      "loss": 1.9527,
      "step": 124347
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.9387574195861816,
      "learning_rate": 0.00026268258615376504,
      "loss": 1.9792,
      "step": 124348
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.9972794055938721,
      "learning_rate": 0.00026267852736551375,
      "loss": 2.1073,
      "step": 124349
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.0932592153549194,
      "learning_rate": 0.00026267446858420124,
      "loss": 1.9373,
      "step": 124350
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.147909164428711,
      "learning_rate": 0.0002626704098098284,
      "loss": 2.2202,
      "step": 124351
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.9237071871757507,
      "learning_rate": 0.00026266635104239567,
      "loss": 2.3453,
      "step": 124352
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.1519116163253784,
      "learning_rate": 0.000262662292281904,
      "loss": 2.1046,
      "step": 124353
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.9502187371253967,
      "learning_rate": 0.0002626582335283541,
      "loss": 1.8017,
      "step": 124354
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.9129713177680969,
      "learning_rate": 0.00026265417478174677,
      "loss": 1.9868,
      "step": 124355
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.9894469380378723,
      "learning_rate": 0.0002626501160420827,
      "loss": 1.9585,
      "step": 124356
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.0315335988998413,
      "learning_rate": 0.00026264605730936283,
      "loss": 2.2119,
      "step": 124357
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.9740633964538574,
      "learning_rate": 0.0002626419985835876,
      "loss": 2.0993,
      "step": 124358
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.051927924156189,
      "learning_rate": 0.000262637939864758,
      "loss": 2.0023,
      "step": 124359
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.9148013591766357,
      "learning_rate": 0.0002626338811528746,
      "loss": 2.1721,
      "step": 124360
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.3560659885406494,
      "learning_rate": 0.0002626298224479383,
      "loss": 2.0446,
      "step": 124361
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.0510979890823364,
      "learning_rate": 0.0002626257637499499,
      "loss": 2.0337,
      "step": 124362
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.0186388492584229,
      "learning_rate": 0.00026262170505891017,
      "loss": 2.1793,
      "step": 124363
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.9458521604537964,
      "learning_rate": 0.0002626176463748196,
      "loss": 2.0301,
      "step": 124364
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.0196298360824585,
      "learning_rate": 0.0002626135876976791,
      "loss": 2.3063,
      "step": 124365
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.0298197269439697,
      "learning_rate": 0.00026260952902748953,
      "loss": 2.2593,
      "step": 124366
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.945952296257019,
      "learning_rate": 0.00026260547036425147,
      "loss": 2.2315,
      "step": 124367
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.8719901442527771,
      "learning_rate": 0.00026260141170796584,
      "loss": 2.1106,
      "step": 124368
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.176531434059143,
      "learning_rate": 0.0002625973530586334,
      "loss": 2.1546,
      "step": 124369
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.9330295324325562,
      "learning_rate": 0.00026259329441625465,
      "loss": 2.1894,
      "step": 124370
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.1826714277267456,
      "learning_rate": 0.00026258923578083055,
      "loss": 2.1027,
      "step": 124371
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.1403377056121826,
      "learning_rate": 0.0002625851771523618,
      "loss": 2.2289,
      "step": 124372
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.0383003950119019,
      "learning_rate": 0.0002625811185308492,
      "loss": 2.1387,
      "step": 124373
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.0103241205215454,
      "learning_rate": 0.0002625770599162935,
      "loss": 1.9672,
      "step": 124374
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.2796874046325684,
      "learning_rate": 0.0002625730013086955,
      "loss": 2.1171,
      "step": 124375
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.4268224239349365,
      "learning_rate": 0.0002625689427080558,
      "loss": 2.1227,
      "step": 124376
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.0304290056228638,
      "learning_rate": 0.0002625648841143752,
      "loss": 2.0253,
      "step": 124377
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.0696407556533813,
      "learning_rate": 0.0002625608255276545,
      "loss": 2.1822,
      "step": 124378
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.1534743309020996,
      "learning_rate": 0.00026255676694789446,
      "loss": 2.2217,
      "step": 124379
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.108676791191101,
      "learning_rate": 0.0002625527083750958,
      "loss": 2.3454,
      "step": 124380
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.8846290707588196,
      "learning_rate": 0.00026254864980925933,
      "loss": 2.0807,
      "step": 124381
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.0096619129180908,
      "learning_rate": 0.00026254459125038597,
      "loss": 1.9591,
      "step": 124382
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.2837916612625122,
      "learning_rate": 0.00026254053269847606,
      "loss": 1.9857,
      "step": 124383
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.1127994060516357,
      "learning_rate": 0.0002625364741535306,
      "loss": 2.218,
      "step": 124384
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.9165887832641602,
      "learning_rate": 0.0002625324156155503,
      "loss": 2.0273,
      "step": 124385
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.0761100053787231,
      "learning_rate": 0.0002625283570845359,
      "loss": 2.3377,
      "step": 124386
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.1159428358078003,
      "learning_rate": 0.0002625242985604882,
      "loss": 2.0742,
      "step": 124387
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.0132919549942017,
      "learning_rate": 0.00026252024004340814,
      "loss": 2.0096,
      "step": 124388
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.0111247301101685,
      "learning_rate": 0.0002625161815332961,
      "loss": 1.9831,
      "step": 124389
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.008583426475525,
      "learning_rate": 0.00026251212303015297,
      "loss": 1.9711,
      "step": 124390
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.9031507968902588,
      "learning_rate": 0.0002625080645339796,
      "loss": 2.0628,
      "step": 124391
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.0027521848678589,
      "learning_rate": 0.0002625040060447767,
      "loss": 2.1017,
      "step": 124392
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.0176459550857544,
      "learning_rate": 0.00026249994756254494,
      "loss": 2.3506,
      "step": 124393
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.9349730610847473,
      "learning_rate": 0.00026249588908728536,
      "loss": 2.2291,
      "step": 124394
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.896398663520813,
      "learning_rate": 0.0002624918306189983,
      "loss": 1.9761,
      "step": 124395
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.0322051048278809,
      "learning_rate": 0.00026248777215768474,
      "loss": 2.4074,
      "step": 124396
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.9661258459091187,
      "learning_rate": 0.00026248371370334543,
      "loss": 2.1689,
      "step": 124397
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.0783575773239136,
      "learning_rate": 0.0002624796552559811,
      "loss": 2.0871,
      "step": 124398
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.9625961780548096,
      "learning_rate": 0.0002624755968155925,
      "loss": 2.0816,
      "step": 124399
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.9711901545524597,
      "learning_rate": 0.0002624715383821805,
      "loss": 2.2268,
      "step": 124400
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.9712924361228943,
      "learning_rate": 0.0002624674799557457,
      "loss": 2.0635,
      "step": 124401
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.0446512699127197,
      "learning_rate": 0.0002624634215362889,
      "loss": 2.135,
      "step": 124402
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.087143063545227,
      "learning_rate": 0.00026245936312381074,
      "loss": 2.0869,
      "step": 124403
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.001120686531067,
      "learning_rate": 0.0002624553047183122,
      "loss": 2.2004,
      "step": 124404
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.950186550617218,
      "learning_rate": 0.0002624512463197939,
      "loss": 2.1567,
      "step": 124405
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.088759183883667,
      "learning_rate": 0.00026244718792825663,
      "loss": 2.3648,
      "step": 124406
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.088010311126709,
      "learning_rate": 0.0002624431295437012,
      "loss": 2.1307,
      "step": 124407
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.1037060022354126,
      "learning_rate": 0.0002624390711661282,
      "loss": 2.1764,
      "step": 124408
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.986069917678833,
      "learning_rate": 0.0002624350127955386,
      "loss": 2.0571,
      "step": 124409
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.9093963503837585,
      "learning_rate": 0.0002624309544319329,
      "loss": 2.1128,
      "step": 124410
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.2041370868682861,
      "learning_rate": 0.00026242689607531205,
      "loss": 2.037,
      "step": 124411
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.2644195556640625,
      "learning_rate": 0.0002624228377256768,
      "loss": 1.9076,
      "step": 124412
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.9485755562782288,
      "learning_rate": 0.00026241877938302784,
      "loss": 1.9495,
      "step": 124413
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.9869844913482666,
      "learning_rate": 0.00026241472104736584,
      "loss": 2.189,
      "step": 124414
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.028768539428711,
      "learning_rate": 0.0002624106627186918,
      "loss": 1.9821,
      "step": 124415
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.0514538288116455,
      "learning_rate": 0.00026240660439700626,
      "loss": 2.2527,
      "step": 124416
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.069527506828308,
      "learning_rate": 0.00026240254608231,
      "loss": 2.1112,
      "step": 124417
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.080734372138977,
      "learning_rate": 0.00026239848777460386,
      "loss": 2.0992,
      "step": 124418
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.0263556241989136,
      "learning_rate": 0.0002623944294738886,
      "loss": 1.9979,
      "step": 124419
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.1197630167007446,
      "learning_rate": 0.00026239037118016477,
      "loss": 2.0408,
      "step": 124420
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.1141849756240845,
      "learning_rate": 0.0002623863128934334,
      "loss": 2.29,
      "step": 124421
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.931243896484375,
      "learning_rate": 0.0002623822546136951,
      "loss": 1.9692,
      "step": 124422
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.9699374437332153,
      "learning_rate": 0.0002623781963409507,
      "loss": 2.0804,
      "step": 124423
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.9127544164657593,
      "learning_rate": 0.0002623741380752009,
      "loss": 2.1657,
      "step": 124424
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.9209672808647156,
      "learning_rate": 0.00026237007981644636,
      "loss": 2.1791,
      "step": 124425
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.9491453170776367,
      "learning_rate": 0.000262366021564688,
      "loss": 1.9894,
      "step": 124426
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.0232867002487183,
      "learning_rate": 0.00026236196331992645,
      "loss": 2.201,
      "step": 124427
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.0082136392593384,
      "learning_rate": 0.00026235790508216255,
      "loss": 2.1765,
      "step": 124428
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.0545519590377808,
      "learning_rate": 0.000262353846851397,
      "loss": 2.339,
      "step": 124429
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.9354228377342224,
      "learning_rate": 0.0002623497886276308,
      "loss": 2.0608,
      "step": 124430
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.204607367515564,
      "learning_rate": 0.00026234573041086426,
      "loss": 2.0043,
      "step": 124431
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.2331268787384033,
      "learning_rate": 0.00026234167220109836,
      "loss": 2.0568,
      "step": 124432
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.8720791339874268,
      "learning_rate": 0.00026233761399833386,
      "loss": 2.249,
      "step": 124433
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.923893928527832,
      "learning_rate": 0.00026233355580257157,
      "loss": 1.9824,
      "step": 124434
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.9531760811805725,
      "learning_rate": 0.00026232949761381214,
      "loss": 2.1122,
      "step": 124435
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.9848743677139282,
      "learning_rate": 0.0002623254394320565,
      "loss": 2.1103,
      "step": 124436
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.979091227054596,
      "learning_rate": 0.0002623213812573051,
      "loss": 2.1341,
      "step": 124437
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.9987881183624268,
      "learning_rate": 0.00026231732308955893,
      "loss": 2.1577,
      "step": 124438
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.0570307970046997,
      "learning_rate": 0.0002623132649288187,
      "loss": 2.0716,
      "step": 124439
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.9570499658584595,
      "learning_rate": 0.0002623092067750851,
      "loss": 2.0513,
      "step": 124440
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.9745187163352966,
      "learning_rate": 0.0002623051486283589,
      "loss": 2.452,
      "step": 124441
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.8743810653686523,
      "learning_rate": 0.0002623010904886411,
      "loss": 2.1419,
      "step": 124442
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.946478009223938,
      "learning_rate": 0.00026229703235593205,
      "loss": 2.3221,
      "step": 124443
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.0084794759750366,
      "learning_rate": 0.0002622929742302327,
      "loss": 1.9853,
      "step": 124444
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.8866117596626282,
      "learning_rate": 0.0002622889161115438,
      "loss": 2.3833,
      "step": 124445
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.9348027110099792,
      "learning_rate": 0.0002622848579998661,
      "loss": 2.1385,
      "step": 124446
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.9995328187942505,
      "learning_rate": 0.00026228079989520036,
      "loss": 2.2122,
      "step": 124447
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.008762001991272,
      "learning_rate": 0.0002622767417975473,
      "loss": 2.2199,
      "step": 124448
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.0456888675689697,
      "learning_rate": 0.00026227268370690785,
      "loss": 2.1828,
      "step": 124449
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.9208131432533264,
      "learning_rate": 0.0002622686256232825,
      "loss": 2.0436,
      "step": 124450
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.9759409427642822,
      "learning_rate": 0.00026226456754667216,
      "loss": 2.1147,
      "step": 124451
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.9163777232170105,
      "learning_rate": 0.00026226050947707755,
      "loss": 2.1501,
      "step": 124452
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.0336321592330933,
      "learning_rate": 0.00026225645141449936,
      "loss": 2.138,
      "step": 124453
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.128077745437622,
      "learning_rate": 0.00026225239335893845,
      "loss": 2.0298,
      "step": 124454
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.0317599773406982,
      "learning_rate": 0.00026224833531039563,
      "loss": 2.1774,
      "step": 124455
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.110973834991455,
      "learning_rate": 0.0002622442772688715,
      "loss": 2.2889,
      "step": 124456
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.9415277242660522,
      "learning_rate": 0.0002622402192343668,
      "loss": 1.9767,
      "step": 124457
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.0173940658569336,
      "learning_rate": 0.00026223616120688237,
      "loss": 1.9137,
      "step": 124458
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.0426092147827148,
      "learning_rate": 0.000262232103186419,
      "loss": 2.182,
      "step": 124459
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.923288881778717,
      "learning_rate": 0.0002622280451729773,
      "loss": 2.0558,
      "step": 124460
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.1129320859909058,
      "learning_rate": 0.0002622239871665583,
      "loss": 2.1912,
      "step": 124461
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.8693851828575134,
      "learning_rate": 0.0002622199291671625,
      "loss": 2.0194,
      "step": 124462
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.0667027235031128,
      "learning_rate": 0.00026221587117479067,
      "loss": 2.2388,
      "step": 124463
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.166410207748413,
      "learning_rate": 0.0002622118131894436,
      "loss": 2.0645,
      "step": 124464
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.9863132238388062,
      "learning_rate": 0.0002622077552111221,
      "loss": 2.1293,
      "step": 124465
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.0138306617736816,
      "learning_rate": 0.0002622036972398269,
      "loss": 2.0369,
      "step": 124466
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.9598066210746765,
      "learning_rate": 0.0002621996392755589,
      "loss": 1.8975,
      "step": 124467
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.0078529119491577,
      "learning_rate": 0.00026219558131831847,
      "loss": 2.1851,
      "step": 124468
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.9489318132400513,
      "learning_rate": 0.00026219152336810666,
      "loss": 2.1189,
      "step": 124469
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.106862187385559,
      "learning_rate": 0.0002621874654249242,
      "loss": 2.1294,
      "step": 124470
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.9920890927314758,
      "learning_rate": 0.00026218340748877176,
      "loss": 1.958,
      "step": 124471
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.1476439237594604,
      "learning_rate": 0.00026217934955965013,
      "loss": 2.0567,
      "step": 124472
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.0545293092727661,
      "learning_rate": 0.0002621752916375602,
      "loss": 2.0704,
      "step": 124473
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.899819552898407,
      "learning_rate": 0.0002621712337225025,
      "loss": 2.2958,
      "step": 124474
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.8766399025917053,
      "learning_rate": 0.00026216717581447783,
      "loss": 2.0621,
      "step": 124475
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.9517080783843994,
      "learning_rate": 0.000262163117913487,
      "loss": 2.1387,
      "step": 124476
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.947990894317627,
      "learning_rate": 0.00026215906001953083,
      "loss": 2.1687,
      "step": 124477
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.9778886437416077,
      "learning_rate": 0.00026215500213261,
      "loss": 2.1677,
      "step": 124478
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.9911274909973145,
      "learning_rate": 0.0002621509442527254,
      "loss": 2.2746,
      "step": 124479
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.9839557409286499,
      "learning_rate": 0.0002621468863798774,
      "loss": 2.1808,
      "step": 124480
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.949971079826355,
      "learning_rate": 0.0002621428285140671,
      "loss": 2.1844,
      "step": 124481
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.9398064613342285,
      "learning_rate": 0.0002621387706552952,
      "loss": 2.1198,
      "step": 124482
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.933574378490448,
      "learning_rate": 0.00026213471280356234,
      "loss": 2.3619,
      "step": 124483
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.05351984500885,
      "learning_rate": 0.00026213065495886943,
      "loss": 2.118,
      "step": 124484
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.0726070404052734,
      "learning_rate": 0.0002621265971212172,
      "loss": 2.2426,
      "step": 124485
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.9431135654449463,
      "learning_rate": 0.00026212253929060625,
      "loss": 1.9754,
      "step": 124486
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.0694093704223633,
      "learning_rate": 0.00026211848146703745,
      "loss": 2.1561,
      "step": 124487
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.9857656359672546,
      "learning_rate": 0.00026211442365051157,
      "loss": 2.2136,
      "step": 124488
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.967597246170044,
      "learning_rate": 0.00026211036584102926,
      "loss": 2.1148,
      "step": 124489
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.9098215699195862,
      "learning_rate": 0.0002621063080385914,
      "loss": 2.3832,
      "step": 124490
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.9951328039169312,
      "learning_rate": 0.0002621022502431987,
      "loss": 2.0531,
      "step": 124491
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.1879783868789673,
      "learning_rate": 0.0002620981924548519,
      "loss": 2.2776,
      "step": 124492
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.9695440530776978,
      "learning_rate": 0.0002620941346735518,
      "loss": 2.0081,
      "step": 124493
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.0336723327636719,
      "learning_rate": 0.0002620900768992991,
      "loss": 1.8869,
      "step": 124494
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.9315876364707947,
      "learning_rate": 0.0002620860191320945,
      "loss": 2.0484,
      "step": 124495
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.9781931638717651,
      "learning_rate": 0.00026208196137193887,
      "loss": 1.9794,
      "step": 124496
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.9323891401290894,
      "learning_rate": 0.00026207790361883296,
      "loss": 2.0589,
      "step": 124497
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.016648769378662,
      "learning_rate": 0.0002620738458727774,
      "loss": 2.3316,
      "step": 124498
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.0684351921081543,
      "learning_rate": 0.00026206978813377306,
      "loss": 2.0399,
      "step": 124499
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.0239837169647217,
      "learning_rate": 0.00026206573040182075,
      "loss": 2.2038,
      "step": 124500
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.0471529960632324,
      "learning_rate": 0.000262061672676921,
      "loss": 2.0721,
      "step": 124501
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.9608508348464966,
      "learning_rate": 0.00026205761495907477,
      "loss": 1.9798,
      "step": 124502
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.0675246715545654,
      "learning_rate": 0.0002620535572482828,
      "loss": 2.0425,
      "step": 124503
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.0788480043411255,
      "learning_rate": 0.0002620494995445457,
      "loss": 2.0255,
      "step": 124504
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.0078524351119995,
      "learning_rate": 0.0002620454418478643,
      "loss": 2.3157,
      "step": 124505
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.9295164346694946,
      "learning_rate": 0.0002620413841582394,
      "loss": 2.2094,
      "step": 124506
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.0783724784851074,
      "learning_rate": 0.00026203732647567176,
      "loss": 2.2116,
      "step": 124507
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.0005106925964355,
      "learning_rate": 0.0002620332688001621,
      "loss": 2.3138,
      "step": 124508
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.9108143448829651,
      "learning_rate": 0.0002620292111317112,
      "loss": 2.1342,
      "step": 124509
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.9227469563484192,
      "learning_rate": 0.00026202515347031966,
      "loss": 2.2434,
      "step": 124510
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.0334837436676025,
      "learning_rate": 0.0002620210958159884,
      "loss": 1.96,
      "step": 124511
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.0173473358154297,
      "learning_rate": 0.0002620170381687182,
      "loss": 2.091,
      "step": 124512
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.0133079290390015,
      "learning_rate": 0.0002620129805285097,
      "loss": 2.0147,
      "step": 124513
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.0298970937728882,
      "learning_rate": 0.00026200892289536374,
      "loss": 2.2333,
      "step": 124514
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.0185296535491943,
      "learning_rate": 0.000262004865269281,
      "loss": 2.1505,
      "step": 124515
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.9373738765716553,
      "learning_rate": 0.00026200080765026234,
      "loss": 2.4215,
      "step": 124516
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.9634011387825012,
      "learning_rate": 0.0002619967500383084,
      "loss": 2.1446,
      "step": 124517
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.0680646896362305,
      "learning_rate": 0.00026199269243341996,
      "loss": 1.9511,
      "step": 124518
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.0044200420379639,
      "learning_rate": 0.00026198863483559777,
      "loss": 2.2581,
      "step": 124519
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.9595252275466919,
      "learning_rate": 0.00026198457724484263,
      "loss": 2.1024,
      "step": 124520
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.1444307565689087,
      "learning_rate": 0.0002619805196611553,
      "loss": 2.1493,
      "step": 124521
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.9526509642601013,
      "learning_rate": 0.0002619764620845366,
      "loss": 2.0332,
      "step": 124522
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.0027433633804321,
      "learning_rate": 0.0002619724045149871,
      "loss": 2.4143,
      "step": 124523
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.0032321214675903,
      "learning_rate": 0.00026196834695250764,
      "loss": 2.0091,
      "step": 124524
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.9692784547805786,
      "learning_rate": 0.000261964289397099,
      "loss": 2.3859,
      "step": 124525
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.9780094623565674,
      "learning_rate": 0.0002619602318487619,
      "loss": 2.175,
      "step": 124526
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.2617367506027222,
      "learning_rate": 0.0002619561743074971,
      "loss": 2.209,
      "step": 124527
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.1469300985336304,
      "learning_rate": 0.00026195211677330546,
      "loss": 2.2145,
      "step": 124528
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.9921681880950928,
      "learning_rate": 0.00026194805924618757,
      "loss": 2.3782,
      "step": 124529
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.9039160013198853,
      "learning_rate": 0.00026194400172614425,
      "loss": 2.0413,
      "step": 124530
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.3644510507583618,
      "learning_rate": 0.0002619399442131762,
      "loss": 1.9703,
      "step": 124531
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.0371017456054688,
      "learning_rate": 0.0002619358867072843,
      "loss": 2.1402,
      "step": 124532
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.9593201875686646,
      "learning_rate": 0.0002619318292084692,
      "loss": 2.162,
      "step": 124533
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.0486980676651,
      "learning_rate": 0.0002619277717167318,
      "loss": 1.9968,
      "step": 124534
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.9848889112472534,
      "learning_rate": 0.0002619237142320727,
      "loss": 2.1767,
      "step": 124535
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.9401904940605164,
      "learning_rate": 0.00026191965675449264,
      "loss": 2.0909,
      "step": 124536
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.9962771534919739,
      "learning_rate": 0.00026191559928399243,
      "loss": 2.1129,
      "step": 124537
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.9455603361129761,
      "learning_rate": 0.00026191154182057287,
      "loss": 2.0938,
      "step": 124538
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.1801918745040894,
      "learning_rate": 0.0002619074843642346,
      "loss": 2.1618,
      "step": 124539
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.9656120538711548,
      "learning_rate": 0.00026190342691497865,
      "loss": 2.0087,
      "step": 124540
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.9512398838996887,
      "learning_rate": 0.00026189936947280544,
      "loss": 1.88,
      "step": 124541
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.0666289329528809,
      "learning_rate": 0.0002618953120377158,
      "loss": 2.1035,
      "step": 124542
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.133792519569397,
      "learning_rate": 0.0002618912546097106,
      "loss": 2.1365,
      "step": 124543
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.901491641998291,
      "learning_rate": 0.0002618871971887905,
      "loss": 2.3067,
      "step": 124544
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.0572686195373535,
      "learning_rate": 0.0002618831397749563,
      "loss": 2.1657,
      "step": 124545
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.1925253868103027,
      "learning_rate": 0.00026187908236820884,
      "loss": 2.2065,
      "step": 124546
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.0125484466552734,
      "learning_rate": 0.0002618750249685487,
      "loss": 2.5238,
      "step": 124547
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.02638578414917,
      "learning_rate": 0.0002618709675759767,
      "loss": 1.9556,
      "step": 124548
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.9650905132293701,
      "learning_rate": 0.0002618669101904936,
      "loss": 2.1873,
      "step": 124549
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.1547144651412964,
      "learning_rate": 0.00026186285281210015,
      "loss": 1.8323,
      "step": 124550
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.9381431937217712,
      "learning_rate": 0.0002618587954407971,
      "loss": 2.1462,
      "step": 124551
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.0063116550445557,
      "learning_rate": 0.00026185473807658537,
      "loss": 2.1979,
      "step": 124552
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.8946663737297058,
      "learning_rate": 0.00026185068071946544,
      "loss": 2.2079,
      "step": 124553
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.0551605224609375,
      "learning_rate": 0.0002618466233694382,
      "loss": 2.0841,
      "step": 124554
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.9277209043502808,
      "learning_rate": 0.0002618425660265044,
      "loss": 2.1134,
      "step": 124555
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.043797254562378,
      "learning_rate": 0.00026183850869066476,
      "loss": 2.1942,
      "step": 124556
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.9574061632156372,
      "learning_rate": 0.0002618344513619201,
      "loss": 1.9773,
      "step": 124557
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.9694883227348328,
      "learning_rate": 0.0002618303940402712,
      "loss": 2.3168,
      "step": 124558
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.9873447418212891,
      "learning_rate": 0.00026182633672571865,
      "loss": 2.0711,
      "step": 124559
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.9303992390632629,
      "learning_rate": 0.0002618222794182633,
      "loss": 1.9921,
      "step": 124560
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.98297518491745,
      "learning_rate": 0.0002618182221179059,
      "loss": 2.0541,
      "step": 124561
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.9876350164413452,
      "learning_rate": 0.0002618141648246473,
      "loss": 2.1259,
      "step": 124562
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.0504151582717896,
      "learning_rate": 0.00026181010753848806,
      "loss": 2.1121,
      "step": 124563
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.078370451927185,
      "learning_rate": 0.0002618060502594292,
      "loss": 2.056,
      "step": 124564
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.0010954141616821,
      "learning_rate": 0.0002618019929874712,
      "loss": 2.1248,
      "step": 124565
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.995292603969574,
      "learning_rate": 0.000261797935722615,
      "loss": 2.2087,
      "step": 124566
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.1809732913970947,
      "learning_rate": 0.00026179387846486115,
      "loss": 2.1043,
      "step": 124567
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.9668059349060059,
      "learning_rate": 0.00026178982121421063,
      "loss": 2.1394,
      "step": 124568
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.152858018875122,
      "learning_rate": 0.0002617857639706641,
      "loss": 2.2722,
      "step": 124569
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.1087424755096436,
      "learning_rate": 0.00026178170673422235,
      "loss": 2.0057,
      "step": 124570
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.0296545028686523,
      "learning_rate": 0.00026177764950488606,
      "loss": 2.1901,
      "step": 124571
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.0712226629257202,
      "learning_rate": 0.00026177359228265606,
      "loss": 2.0666,
      "step": 124572
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.114591360092163,
      "learning_rate": 0.000261769535067533,
      "loss": 2.3171,
      "step": 124573
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.0736663341522217,
      "learning_rate": 0.0002617654778595177,
      "loss": 2.2538,
      "step": 124574
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.0149633884429932,
      "learning_rate": 0.0002617614206586109,
      "loss": 2.0799,
      "step": 124575
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.9974473714828491,
      "learning_rate": 0.0002617573634648135,
      "loss": 1.9418,
      "step": 124576
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.03617262840271,
      "learning_rate": 0.00026175330627812616,
      "loss": 1.9649,
      "step": 124577
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.9257083535194397,
      "learning_rate": 0.0002617492490985495,
      "loss": 1.8482,
      "step": 124578
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.947201132774353,
      "learning_rate": 0.00026174519192608437,
      "loss": 2.0907,
      "step": 124579
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.0411393642425537,
      "learning_rate": 0.0002617411347607315,
      "loss": 2.0764,
      "step": 124580
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.0303263664245605,
      "learning_rate": 0.00026173707760249177,
      "loss": 2.0267,
      "step": 124581
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.9840952157974243,
      "learning_rate": 0.00026173302045136575,
      "loss": 2.0904,
      "step": 124582
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.029294490814209,
      "learning_rate": 0.0002617289633073543,
      "loss": 2.07,
      "step": 124583
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.9913475513458252,
      "learning_rate": 0.0002617249061704583,
      "loss": 2.3149,
      "step": 124584
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.1382509469985962,
      "learning_rate": 0.0002617208490406782,
      "loss": 2.3353,
      "step": 124585
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.0245417356491089,
      "learning_rate": 0.00026171679191801494,
      "loss": 2.3087,
      "step": 124586
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.9190113544464111,
      "learning_rate": 0.0002617127348024693,
      "loss": 2.1705,
      "step": 124587
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.0036753416061401,
      "learning_rate": 0.0002617086776940419,
      "loss": 2.0496,
      "step": 124588
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.1452642679214478,
      "learning_rate": 0.0002617046205927337,
      "loss": 2.276,
      "step": 124589
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.1095662117004395,
      "learning_rate": 0.00026170056349854526,
      "loss": 2.2727,
      "step": 124590
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.9553382396697998,
      "learning_rate": 0.0002616965064114774,
      "loss": 1.9268,
      "step": 124591
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.1809616088867188,
      "learning_rate": 0.00026169244933153097,
      "loss": 2.1072,
      "step": 124592
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.9854281544685364,
      "learning_rate": 0.0002616883922587065,
      "loss": 2.3229,
      "step": 124593
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.335893988609314,
      "learning_rate": 0.00026168433519300495,
      "loss": 2.1908,
      "step": 124594
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.97078537940979,
      "learning_rate": 0.0002616802781344271,
      "loss": 2.1985,
      "step": 124595
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.2234034538269043,
      "learning_rate": 0.00026167622108297345,
      "loss": 2.2827,
      "step": 124596
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.1575794219970703,
      "learning_rate": 0.000261672164038645,
      "loss": 2.149,
      "step": 124597
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.0947203636169434,
      "learning_rate": 0.0002616681070014424,
      "loss": 1.9662,
      "step": 124598
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.1568673849105835,
      "learning_rate": 0.00026166404997136645,
      "loss": 2.1407,
      "step": 124599
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.376181960105896,
      "learning_rate": 0.0002616599929484178,
      "loss": 2.2565,
      "step": 124600
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.1660364866256714,
      "learning_rate": 0.0002616559359325974,
      "loss": 2.0533,
      "step": 124601
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.6103520393371582,
      "learning_rate": 0.0002616518789239058,
      "loss": 2.1734,
      "step": 124602
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.1560567617416382,
      "learning_rate": 0.0002616478219223438,
      "loss": 2.0813,
      "step": 124603
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.1616764068603516,
      "learning_rate": 0.00026164376492791223,
      "loss": 2.2655,
      "step": 124604
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.3506888151168823,
      "learning_rate": 0.0002616397079406118,
      "loss": 2.02,
      "step": 124605
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.0873253345489502,
      "learning_rate": 0.0002616356509604434,
      "loss": 2.1718,
      "step": 124606
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.914706289768219,
      "learning_rate": 0.00026163159398740756,
      "loss": 2.072,
      "step": 124607
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.0965158939361572,
      "learning_rate": 0.0002616275370215051,
      "loss": 2.1076,
      "step": 124608
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.9742689728736877,
      "learning_rate": 0.00026162348006273685,
      "loss": 2.1529,
      "step": 124609
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.1455440521240234,
      "learning_rate": 0.0002616194231111035,
      "loss": 2.1203,
      "step": 124610
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.9603767395019531,
      "learning_rate": 0.00026161536616660575,
      "loss": 2.0978,
      "step": 124611
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.3324497938156128,
      "learning_rate": 0.0002616113092292445,
      "loss": 2.1617,
      "step": 124612
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.0603861808776855,
      "learning_rate": 0.00026160725229902056,
      "loss": 2.2272,
      "step": 124613
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.1319220066070557,
      "learning_rate": 0.0002616031953759344,
      "loss": 2.0834,
      "step": 124614
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.9906410574913025,
      "learning_rate": 0.00026159913845998697,
      "loss": 2.1236,
      "step": 124615
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.9589422345161438,
      "learning_rate": 0.00026159508155117894,
      "loss": 2.1453,
      "step": 124616
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.1773223876953125,
      "learning_rate": 0.0002615910246495111,
      "loss": 2.0975,
      "step": 124617
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.1312191486358643,
      "learning_rate": 0.0002615869677549843,
      "loss": 1.9598,
      "step": 124618
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.0440346002578735,
      "learning_rate": 0.00026158291086759923,
      "loss": 2.006,
      "step": 124619
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.9950165748596191,
      "learning_rate": 0.0002615788539873566,
      "loss": 2.1871,
      "step": 124620
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.252203345298767,
      "learning_rate": 0.00026157479711425713,
      "loss": 2.1727,
      "step": 124621
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.315346598625183,
      "learning_rate": 0.0002615707402483016,
      "loss": 2.0806,
      "step": 124622
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.0434678792953491,
      "learning_rate": 0.0002615666833894908,
      "loss": 2.0439,
      "step": 124623
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.1500742435455322,
      "learning_rate": 0.00026156262653782554,
      "loss": 2.0955,
      "step": 124624
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.296284556388855,
      "learning_rate": 0.0002615585696933066,
      "loss": 2.0891,
      "step": 124625
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.0770814418792725,
      "learning_rate": 0.0002615545128559345,
      "loss": 2.0532,
      "step": 124626
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.009217381477356,
      "learning_rate": 0.0002615504560257102,
      "loss": 1.9924,
      "step": 124627
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.164360523223877,
      "learning_rate": 0.00026154639920263436,
      "loss": 2.16,
      "step": 124628
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.079575777053833,
      "learning_rate": 0.00026154234238670777,
      "loss": 2.1905,
      "step": 124629
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.122349739074707,
      "learning_rate": 0.00026153828557793116,
      "loss": 2.3525,
      "step": 124630
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.9907011985778809,
      "learning_rate": 0.00026153422877630547,
      "loss": 2.205,
      "step": 124631
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.9111675024032593,
      "learning_rate": 0.00026153017198183117,
      "loss": 2.1088,
      "step": 124632
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.04230535030365,
      "learning_rate": 0.00026152611519450914,
      "loss": 2.1552,
      "step": 124633
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.9045568108558655,
      "learning_rate": 0.0002615220584143401,
      "loss": 2.1334,
      "step": 124634
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.9841641187667847,
      "learning_rate": 0.0002615180016413248,
      "loss": 2.1295,
      "step": 124635
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.9762371778488159,
      "learning_rate": 0.0002615139448754641,
      "loss": 2.3338,
      "step": 124636
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.9650003910064697,
      "learning_rate": 0.00026150988811675884,
      "loss": 2.1493,
      "step": 124637
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.0028700828552246,
      "learning_rate": 0.0002615058313652094,
      "loss": 2.0444,
      "step": 124638
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.0033814907073975,
      "learning_rate": 0.0002615017746208168,
      "loss": 2.1422,
      "step": 124639
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.038564920425415,
      "learning_rate": 0.00026149771788358176,
      "loss": 2.2433,
      "step": 124640
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.9495497345924377,
      "learning_rate": 0.000261493661153505,
      "loss": 2.2212,
      "step": 124641
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.9571796655654907,
      "learning_rate": 0.0002614896044305873,
      "loss": 2.3116,
      "step": 124642
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.9191309809684753,
      "learning_rate": 0.00026148554771482954,
      "loss": 2.3102,
      "step": 124643
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.9931670427322388,
      "learning_rate": 0.0002614814910062322,
      "loss": 2.0699,
      "step": 124644
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.1056135892868042,
      "learning_rate": 0.0002614774343047962,
      "loss": 2.2542,
      "step": 124645
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.9550585150718689,
      "learning_rate": 0.00026147337761052225,
      "loss": 2.0606,
      "step": 124646
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.9792630672454834,
      "learning_rate": 0.0002614693209234112,
      "loss": 1.9843,
      "step": 124647
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.1047676801681519,
      "learning_rate": 0.00026146526424346365,
      "loss": 1.9903,
      "step": 124648
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.0061429738998413,
      "learning_rate": 0.0002614612075706804,
      "loss": 2.0772,
      "step": 124649
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.1050714254379272,
      "learning_rate": 0.00026145715090506247,
      "loss": 2.0104,
      "step": 124650
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.0462238788604736,
      "learning_rate": 0.00026145309424661017,
      "loss": 2.0475,
      "step": 124651
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.384549856185913,
      "learning_rate": 0.00026144903759532456,
      "loss": 2.0981,
      "step": 124652
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.9389033317565918,
      "learning_rate": 0.0002614449809512062,
      "loss": 2.0581,
      "step": 124653
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.0367852449417114,
      "learning_rate": 0.000261440924314256,
      "loss": 1.8907,
      "step": 124654
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.1908570528030396,
      "learning_rate": 0.00026143686768447465,
      "loss": 2.2368,
      "step": 124655
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.9515647888183594,
      "learning_rate": 0.00026143281106186304,
      "loss": 2.3368,
      "step": 124656
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.9918045401573181,
      "learning_rate": 0.0002614287544464217,
      "loss": 2.1134,
      "step": 124657
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.1049737930297852,
      "learning_rate": 0.0002614246978381515,
      "loss": 2.1352,
      "step": 124658
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.1582781076431274,
      "learning_rate": 0.0002614206412370531,
      "loss": 2.2095,
      "step": 124659
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.9044354557991028,
      "learning_rate": 0.00026141658464312736,
      "loss": 1.9053,
      "step": 124660
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.8974543213844299,
      "learning_rate": 0.000261412528056375,
      "loss": 2.0937,
      "step": 124661
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.198323369026184,
      "learning_rate": 0.00026140847147679687,
      "loss": 2.0542,
      "step": 124662
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.9463085532188416,
      "learning_rate": 0.0002614044149043936,
      "loss": 2.0908,
      "step": 124663
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.0732965469360352,
      "learning_rate": 0.0002614003583391659,
      "loss": 2.0478,
      "step": 124664
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.9379222393035889,
      "learning_rate": 0.00026139630178111464,
      "loss": 1.936,
      "step": 124665
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.247172474861145,
      "learning_rate": 0.00026139224523024057,
      "loss": 2.1279,
      "step": 124666
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.5647410154342651,
      "learning_rate": 0.00026138818868654435,
      "loss": 2.1729,
      "step": 124667
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.8994958400726318,
      "learning_rate": 0.0002613841321500269,
      "loss": 2.1423,
      "step": 124668
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.9778055548667908,
      "learning_rate": 0.0002613800756206888,
      "loss": 2.0125,
      "step": 124669
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.2300689220428467,
      "learning_rate": 0.0002613760190985309,
      "loss": 2.2485,
      "step": 124670
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.0420209169387817,
      "learning_rate": 0.0002613719625835538,
      "loss": 2.0563,
      "step": 124671
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.0610898733139038,
      "learning_rate": 0.0002613679060757585,
      "loss": 2.0046,
      "step": 124672
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.1147727966308594,
      "learning_rate": 0.0002613638495751456,
      "loss": 2.0729,
      "step": 124673
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.0097414255142212,
      "learning_rate": 0.000261359793081716,
      "loss": 2.2576,
      "step": 124674
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.131048321723938,
      "learning_rate": 0.00026135573659547023,
      "loss": 2.3083,
      "step": 124675
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.9584039449691772,
      "learning_rate": 0.0002613516801164092,
      "loss": 2.2575,
      "step": 124676
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.0383210182189941,
      "learning_rate": 0.0002613476236445336,
      "loss": 2.3817,
      "step": 124677
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.1700849533081055,
      "learning_rate": 0.0002613435671798442,
      "loss": 2.203,
      "step": 124678
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.9495924711227417,
      "learning_rate": 0.0002613395107223417,
      "loss": 2.2401,
      "step": 124679
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.951324999332428,
      "learning_rate": 0.0002613354542720271,
      "loss": 2.0973,
      "step": 124680
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.8919690251350403,
      "learning_rate": 0.0002613313978289008,
      "loss": 1.9386,
      "step": 124681
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.9757989645004272,
      "learning_rate": 0.00026132734139296376,
      "loss": 2.2412,
      "step": 124682
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.924748957157135,
      "learning_rate": 0.00026132328496421666,
      "loss": 2.3552,
      "step": 124683
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.0671894550323486,
      "learning_rate": 0.0002613192285426604,
      "loss": 2.1396,
      "step": 124684
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.9679393768310547,
      "learning_rate": 0.00026131517212829553,
      "loss": 2.0847,
      "step": 124685
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.9782785773277283,
      "learning_rate": 0.000261311115721123,
      "loss": 1.9668,
      "step": 124686
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.9391111135482788,
      "learning_rate": 0.00026130705932114344,
      "loss": 2.0359,
      "step": 124687
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.9217981100082397,
      "learning_rate": 0.00026130300292835754,
      "loss": 2.0121,
      "step": 124688
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.0738016366958618,
      "learning_rate": 0.0002612989465427662,
      "loss": 2.1552,
      "step": 124689
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.9711163640022278,
      "learning_rate": 0.0002612948901643701,
      "loss": 2.1529,
      "step": 124690
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.9101321697235107,
      "learning_rate": 0.00026129083379317,
      "loss": 2.0583,
      "step": 124691
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.9713279008865356,
      "learning_rate": 0.0002612867774291668,
      "loss": 2.1258,
      "step": 124692
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.4893721342086792,
      "learning_rate": 0.00026128272107236094,
      "loss": 2.0874,
      "step": 124693
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.930501401424408,
      "learning_rate": 0.0002612786647227534,
      "loss": 2.0778,
      "step": 124694
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.9381493330001831,
      "learning_rate": 0.0002612746083803449,
      "loss": 1.9328,
      "step": 124695
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.9187999963760376,
      "learning_rate": 0.0002612705520451362,
      "loss": 2.2432,
      "step": 124696
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.9568983912467957,
      "learning_rate": 0.00026126649571712795,
      "loss": 1.9184,
      "step": 124697
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.0281224250793457,
      "learning_rate": 0.0002612624393963212,
      "loss": 2.1565,
      "step": 124698
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.9650482535362244,
      "learning_rate": 0.0002612583830827163,
      "loss": 2.2286,
      "step": 124699
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.030128002166748,
      "learning_rate": 0.0002612543267763142,
      "loss": 2.2212,
      "step": 124700
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.1684117317199707,
      "learning_rate": 0.00026125027047711573,
      "loss": 2.1433,
      "step": 124701
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.0137341022491455,
      "learning_rate": 0.00026124621418512154,
      "loss": 2.1711,
      "step": 124702
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.012772798538208,
      "learning_rate": 0.0002612421579003324,
      "loss": 2.1774,
      "step": 124703
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.0766054391860962,
      "learning_rate": 0.0002612381016227492,
      "loss": 2.0283,
      "step": 124704
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.004687786102295,
      "learning_rate": 0.0002612340453523724,
      "loss": 2.0024,
      "step": 124705
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.9369941353797913,
      "learning_rate": 0.000261229989089203,
      "loss": 2.1841,
      "step": 124706
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.1786112785339355,
      "learning_rate": 0.0002612259328332416,
      "loss": 2.1196,
      "step": 124707
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.1345030069351196,
      "learning_rate": 0.0002612218765844891,
      "loss": 2.1432,
      "step": 124708
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.0494815111160278,
      "learning_rate": 0.00026121782034294614,
      "loss": 2.0134,
      "step": 124709
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.307209849357605,
      "learning_rate": 0.0002612137641086135,
      "loss": 2.2908,
      "step": 124710
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.013396143913269,
      "learning_rate": 0.0002612097078814921,
      "loss": 2.248,
      "step": 124711
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.0165390968322754,
      "learning_rate": 0.0002612056516615825,
      "loss": 2.0749,
      "step": 124712
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.241191029548645,
      "learning_rate": 0.00026120159544888537,
      "loss": 2.2102,
      "step": 124713
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.154468297958374,
      "learning_rate": 0.0002611975392434017,
      "loss": 2.0889,
      "step": 124714
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.1137449741363525,
      "learning_rate": 0.0002611934830451321,
      "loss": 2.2126,
      "step": 124715
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.9661780595779419,
      "learning_rate": 0.0002611894268540773,
      "loss": 2.1385,
      "step": 124716
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.9735305309295654,
      "learning_rate": 0.00026118537067023836,
      "loss": 2.0481,
      "step": 124717
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.043184757232666,
      "learning_rate": 0.0002611813144936156,
      "loss": 2.2736,
      "step": 124718
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.0748382806777954,
      "learning_rate": 0.00026117725832421,
      "loss": 2.2056,
      "step": 124719
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.0397223234176636,
      "learning_rate": 0.00026117320216202227,
      "loss": 1.9936,
      "step": 124720
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.073988437652588,
      "learning_rate": 0.0002611691460070532,
      "loss": 2.1308,
      "step": 124721
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.9709777235984802,
      "learning_rate": 0.0002611650898593034,
      "loss": 2.1088,
      "step": 124722
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.150585651397705,
      "learning_rate": 0.000261161033718774,
      "loss": 2.1448,
      "step": 124723
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.0548523664474487,
      "learning_rate": 0.0002611569775854653,
      "loss": 2.1851,
      "step": 124724
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.9572361707687378,
      "learning_rate": 0.00026115292145937827,
      "loss": 2.1967,
      "step": 124725
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.9952391982078552,
      "learning_rate": 0.0002611488653405137,
      "loss": 2.2337,
      "step": 124726
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.985874593257904,
      "learning_rate": 0.0002611448092288722,
      "loss": 2.0868,
      "step": 124727
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.0894747972488403,
      "learning_rate": 0.00026114075312445466,
      "loss": 2.1844,
      "step": 124728
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.9937361478805542,
      "learning_rate": 0.0002611366970272619,
      "loss": 2.2588,
      "step": 124729
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.0028778314590454,
      "learning_rate": 0.00026113264093729447,
      "loss": 2.224,
      "step": 124730
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.950532853603363,
      "learning_rate": 0.00026112858485455314,
      "loss": 2.0999,
      "step": 124731
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.0315154790878296,
      "learning_rate": 0.0002611245287790388,
      "loss": 2.3529,
      "step": 124732
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.2212036848068237,
      "learning_rate": 0.00026112047271075215,
      "loss": 2.2499,
      "step": 124733
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.027182698249817,
      "learning_rate": 0.0002611164166496939,
      "loss": 1.8923,
      "step": 124734
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.032588005065918,
      "learning_rate": 0.000261112360595865,
      "loss": 2.0126,
      "step": 124735
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.9933675527572632,
      "learning_rate": 0.00026110830454926593,
      "loss": 2.0386,
      "step": 124736
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.9942272901535034,
      "learning_rate": 0.0002611042485098975,
      "loss": 2.0923,
      "step": 124737
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.953484833240509,
      "learning_rate": 0.00026110019247776057,
      "loss": 2.0756,
      "step": 124738
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.9604048728942871,
      "learning_rate": 0.00026109613645285585,
      "loss": 2.2352,
      "step": 124739
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.9299799799919128,
      "learning_rate": 0.0002610920804351841,
      "loss": 2.0417,
      "step": 124740
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.9805588722229004,
      "learning_rate": 0.00026108802442474617,
      "loss": 2.1388,
      "step": 124741
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.0144155025482178,
      "learning_rate": 0.00026108396842154256,
      "loss": 2.0505,
      "step": 124742
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.0211551189422607,
      "learning_rate": 0.0002610799124255742,
      "loss": 2.2499,
      "step": 124743
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.9738163948059082,
      "learning_rate": 0.0002610758564368419,
      "loss": 2.2106,
      "step": 124744
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.9707915186882019,
      "learning_rate": 0.0002610718004553462,
      "loss": 1.9657,
      "step": 124745
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.0829558372497559,
      "learning_rate": 0.0002610677444810881,
      "loss": 1.9109,
      "step": 124746
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.03935706615448,
      "learning_rate": 0.00026106368851406826,
      "loss": 2.1251,
      "step": 124747
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.9564721584320068,
      "learning_rate": 0.00026105963255428733,
      "loss": 2.0353,
      "step": 124748
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.0171973705291748,
      "learning_rate": 0.00026105557660174617,
      "loss": 2.094,
      "step": 124749
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.0429720878601074,
      "learning_rate": 0.0002610515206564455,
      "loss": 2.2206,
      "step": 124750
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.1092942953109741,
      "learning_rate": 0.0002610474647183861,
      "loss": 2.0673,
      "step": 124751
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.9737994074821472,
      "learning_rate": 0.0002610434087875687,
      "loss": 2.1978,
      "step": 124752
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.928985595703125,
      "learning_rate": 0.0002610393528639942,
      "loss": 2.3194,
      "step": 124753
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.1976871490478516,
      "learning_rate": 0.0002610352969476631,
      "loss": 2.2552,
      "step": 124754
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.9498418569564819,
      "learning_rate": 0.00026103124103857625,
      "loss": 2.0377,
      "step": 124755
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.082523226737976,
      "learning_rate": 0.00026102718513673444,
      "loss": 2.0196,
      "step": 124756
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.979861319065094,
      "learning_rate": 0.0002610231292421384,
      "loss": 2.2115,
      "step": 124757
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.0166031122207642,
      "learning_rate": 0.00026101907335478896,
      "loss": 2.2557,
      "step": 124758
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.9818225502967834,
      "learning_rate": 0.0002610150174746868,
      "loss": 2.1847,
      "step": 124759
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.9598082304000854,
      "learning_rate": 0.00026101096160183266,
      "loss": 2.0172,
      "step": 124760
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.0326478481292725,
      "learning_rate": 0.0002610069057362273,
      "loss": 2.1048,
      "step": 124761
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.0321110486984253,
      "learning_rate": 0.0002610028498778715,
      "loss": 2.0894,
      "step": 124762
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.858869731426239,
      "learning_rate": 0.000260998794026766,
      "loss": 1.7962,
      "step": 124763
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.9563243389129639,
      "learning_rate": 0.0002609947381829116,
      "loss": 1.9269,
      "step": 124764
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.9686465859413147,
      "learning_rate": 0.000260990682346309,
      "loss": 2.1897,
      "step": 124765
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.0367436408996582,
      "learning_rate": 0.00026098662651695897,
      "loss": 2.2706,
      "step": 124766
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.9160920977592468,
      "learning_rate": 0.00026098257069486223,
      "loss": 2.0986,
      "step": 124767
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.8981645703315735,
      "learning_rate": 0.0002609785148800197,
      "loss": 1.9947,
      "step": 124768
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.9565914869308472,
      "learning_rate": 0.00026097445907243184,
      "loss": 2.091,
      "step": 124769
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.9686194062232971,
      "learning_rate": 0.0002609704032720996,
      "loss": 2.3504,
      "step": 124770
    },
    {
      "epoch": 1.62,
      "grad_norm": 2.762279748916626,
      "learning_rate": 0.00026096634747902375,
      "loss": 2.0802,
      "step": 124771
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.982789933681488,
      "learning_rate": 0.00026096229169320496,
      "loss": 1.9928,
      "step": 124772
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.0180977582931519,
      "learning_rate": 0.00026095823591464393,
      "loss": 2.0531,
      "step": 124773
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.9376108050346375,
      "learning_rate": 0.00026095418014334163,
      "loss": 2.2304,
      "step": 124774
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.9021247625350952,
      "learning_rate": 0.0002609501243792986,
      "loss": 2.0418,
      "step": 124775
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.9385480284690857,
      "learning_rate": 0.0002609460686225158,
      "loss": 2.1431,
      "step": 124776
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.993368923664093,
      "learning_rate": 0.00026094201287299375,
      "loss": 2.2107,
      "step": 124777
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.1047879457473755,
      "learning_rate": 0.0002609379571307334,
      "loss": 2.2188,
      "step": 124778
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.038022756576538,
      "learning_rate": 0.0002609339013957353,
      "loss": 2.1565,
      "step": 124779
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.9144202470779419,
      "learning_rate": 0.0002609298456680004,
      "loss": 2.1954,
      "step": 124780
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.9921485185623169,
      "learning_rate": 0.0002609257899475294,
      "loss": 2.0382,
      "step": 124781
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.9944870471954346,
      "learning_rate": 0.000260921734234323,
      "loss": 1.9811,
      "step": 124782
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.1204948425292969,
      "learning_rate": 0.00026091767852838197,
      "loss": 2.0459,
      "step": 124783
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.0183780193328857,
      "learning_rate": 0.0002609136228297072,
      "loss": 2.1466,
      "step": 124784
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.9949983954429626,
      "learning_rate": 0.0002609095671382992,
      "loss": 2.3455,
      "step": 124785
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.9630060195922852,
      "learning_rate": 0.00026090551145415883,
      "loss": 2.1913,
      "step": 124786
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.071095585823059,
      "learning_rate": 0.0002609014557772869,
      "loss": 2.2047,
      "step": 124787
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.0613316297531128,
      "learning_rate": 0.00026089740010768413,
      "loss": 1.9891,
      "step": 124788
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.049088954925537,
      "learning_rate": 0.0002608933444453513,
      "loss": 2.0818,
      "step": 124789
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.9884429574012756,
      "learning_rate": 0.0002608892887902892,
      "loss": 2.0355,
      "step": 124790
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.1682988405227661,
      "learning_rate": 0.00026088523314249845,
      "loss": 1.9398,
      "step": 124791
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.0987578630447388,
      "learning_rate": 0.00026088117750197983,
      "loss": 2.0947,
      "step": 124792
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.9501215815544128,
      "learning_rate": 0.00026087712186873414,
      "loss": 2.0596,
      "step": 124793
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.179313063621521,
      "learning_rate": 0.0002608730662427621,
      "loss": 1.9435,
      "step": 124794
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.0438488721847534,
      "learning_rate": 0.00026086901062406456,
      "loss": 2.012,
      "step": 124795
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.0527713298797607,
      "learning_rate": 0.00026086495501264235,
      "loss": 2.1374,
      "step": 124796
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.9141258597373962,
      "learning_rate": 0.0002608608994084959,
      "loss": 2.1883,
      "step": 124797
    },
    {
      "epoch": 1.62,
      "grad_norm": 0.994439959526062,
      "learning_rate": 0.00026085684381162616,
      "loss": 2.1123,
      "step": 124798
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.1298930644989014,
      "learning_rate": 0.00026085278822203387,
      "loss": 2.1771,
      "step": 124799
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.2477917671203613,
      "learning_rate": 0.0002608487326397198,
      "loss": 2.0375,
      "step": 124800
    },
    {
      "epoch": 1.62,
      "grad_norm": 1.0029306411743164,
      "learning_rate": 0.0002608446770646847,
      "loss": 2.2458,
      "step": 124801
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.0090956687927246,
      "learning_rate": 0.00026084062149692947,
      "loss": 2.0558,
      "step": 124802
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.134069800376892,
      "learning_rate": 0.0002608365659364545,
      "loss": 2.1328,
      "step": 124803
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.0503100156784058,
      "learning_rate": 0.0002608325103832608,
      "loss": 1.9039,
      "step": 124804
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.9642775654792786,
      "learning_rate": 0.00026082845483734905,
      "loss": 2.2362,
      "step": 124805
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.0676488876342773,
      "learning_rate": 0.00026082439929872006,
      "loss": 2.2102,
      "step": 124806
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.0048681497573853,
      "learning_rate": 0.0002608203437673745,
      "loss": 2.1625,
      "step": 124807
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.0313677787780762,
      "learning_rate": 0.00026081628824331336,
      "loss": 2.2378,
      "step": 124808
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.0601000785827637,
      "learning_rate": 0.00026081223272653707,
      "loss": 2.1914,
      "step": 124809
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.9639632105827332,
      "learning_rate": 0.0002608081772170465,
      "loss": 1.9251,
      "step": 124810
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.9387460350990295,
      "learning_rate": 0.0002608041217148425,
      "loss": 2.2008,
      "step": 124811
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.917333722114563,
      "learning_rate": 0.0002608000662199257,
      "loss": 2.1315,
      "step": 124812
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.004581093788147,
      "learning_rate": 0.0002607960107322969,
      "loss": 2.1364,
      "step": 124813
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.0081613063812256,
      "learning_rate": 0.00026079195525195703,
      "loss": 2.1401,
      "step": 124814
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.8900958299636841,
      "learning_rate": 0.0002607878997789065,
      "loss": 2.0475,
      "step": 124815
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.951777994632721,
      "learning_rate": 0.00026078384431314625,
      "loss": 2.1357,
      "step": 124816
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.9691016674041748,
      "learning_rate": 0.00026077978885467705,
      "loss": 2.0395,
      "step": 124817
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.0522313117980957,
      "learning_rate": 0.0002607757334034996,
      "loss": 2.1073,
      "step": 124818
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.8470878005027771,
      "learning_rate": 0.0002607716779596147,
      "loss": 1.9572,
      "step": 124819
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.9604411721229553,
      "learning_rate": 0.0002607676225230232,
      "loss": 2.371,
      "step": 124820
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.046354055404663,
      "learning_rate": 0.0002607635670937256,
      "loss": 1.9721,
      "step": 124821
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.0764920711517334,
      "learning_rate": 0.00026075951167172284,
      "loss": 2.4379,
      "step": 124822
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.0308171510696411,
      "learning_rate": 0.0002607554562570156,
      "loss": 2.0792,
      "step": 124823
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.040395736694336,
      "learning_rate": 0.00026075140084960463,
      "loss": 2.1912,
      "step": 124824
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.0373859405517578,
      "learning_rate": 0.00026074734544949077,
      "loss": 2.2058,
      "step": 124825
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.9577690958976746,
      "learning_rate": 0.0002607432900566748,
      "loss": 2.3741,
      "step": 124826
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.0034823417663574,
      "learning_rate": 0.00026073923467115736,
      "loss": 2.249,
      "step": 124827
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.0232124328613281,
      "learning_rate": 0.0002607351792929391,
      "loss": 1.9516,
      "step": 124828
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.1295208930969238,
      "learning_rate": 0.00026073112392202097,
      "loss": 2.0691,
      "step": 124829
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.0031064748764038,
      "learning_rate": 0.00026072706855840373,
      "loss": 2.1056,
      "step": 124830
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.9236107468605042,
      "learning_rate": 0.000260723013202088,
      "loss": 2.0316,
      "step": 124831
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.0411783456802368,
      "learning_rate": 0.00026071895785307475,
      "loss": 2.2521,
      "step": 124832
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.0117077827453613,
      "learning_rate": 0.00026071490251136447,
      "loss": 2.0787,
      "step": 124833
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.1410990953445435,
      "learning_rate": 0.00026071084717695797,
      "loss": 2.188,
      "step": 124834
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.9950470924377441,
      "learning_rate": 0.0002607067918498561,
      "loss": 2.3764,
      "step": 124835
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.0506298542022705,
      "learning_rate": 0.0002607027365300596,
      "loss": 2.1798,
      "step": 124836
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.9424712657928467,
      "learning_rate": 0.0002606986812175692,
      "loss": 2.2614,
      "step": 124837
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.8586163520812988,
      "learning_rate": 0.0002606946259123858,
      "loss": 2.0018,
      "step": 124838
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.9975892305374146,
      "learning_rate": 0.0002606905706145098,
      "loss": 2.0182,
      "step": 124839
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.0941969156265259,
      "learning_rate": 0.00026068651532394225,
      "loss": 2.1048,
      "step": 124840
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.0268784761428833,
      "learning_rate": 0.0002606824600406838,
      "loss": 2.2436,
      "step": 124841
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.04787278175354,
      "learning_rate": 0.0002606784047647352,
      "loss": 2.0591,
      "step": 124842
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.9687427878379822,
      "learning_rate": 0.0002606743494960973,
      "loss": 2.2031,
      "step": 124843
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.9238841533660889,
      "learning_rate": 0.00026067029423477067,
      "loss": 2.1567,
      "step": 124844
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.1772964000701904,
      "learning_rate": 0.0002606662389807563,
      "loss": 1.9622,
      "step": 124845
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.0296604633331299,
      "learning_rate": 0.0002606621837340548,
      "loss": 2.203,
      "step": 124846
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.0211361646652222,
      "learning_rate": 0.00026065812849466693,
      "loss": 2.1709,
      "step": 124847
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.9310291409492493,
      "learning_rate": 0.0002606540732625934,
      "loss": 1.8066,
      "step": 124848
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.9043650031089783,
      "learning_rate": 0.000260650018037835,
      "loss": 2.172,
      "step": 124849
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.0367751121520996,
      "learning_rate": 0.0002606459628203926,
      "loss": 1.8835,
      "step": 124850
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.0948432683944702,
      "learning_rate": 0.0002606419076102668,
      "loss": 2.1367,
      "step": 124851
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.9869510531425476,
      "learning_rate": 0.00026063785240745845,
      "loss": 2.0833,
      "step": 124852
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.9358049631118774,
      "learning_rate": 0.0002606337972119683,
      "loss": 1.8037,
      "step": 124853
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.1878416538238525,
      "learning_rate": 0.00026062974202379697,
      "loss": 2.0952,
      "step": 124854
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.9387756586074829,
      "learning_rate": 0.00026062568684294537,
      "loss": 2.1838,
      "step": 124855
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.9401630759239197,
      "learning_rate": 0.00026062163166941415,
      "loss": 2.0624,
      "step": 124856
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.0952506065368652,
      "learning_rate": 0.0002606175765032042,
      "loss": 2.1726,
      "step": 124857
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.0135258436203003,
      "learning_rate": 0.0002606135213443161,
      "loss": 2.1461,
      "step": 124858
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.9960721135139465,
      "learning_rate": 0.00026060946619275077,
      "loss": 2.0554,
      "step": 124859
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.9503706693649292,
      "learning_rate": 0.00026060541104850883,
      "loss": 2.0048,
      "step": 124860
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.0883771181106567,
      "learning_rate": 0.0002606013559115911,
      "loss": 2.2412,
      "step": 124861
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.0450496673583984,
      "learning_rate": 0.0002605973007819983,
      "loss": 2.1528,
      "step": 124862
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.1986970901489258,
      "learning_rate": 0.0002605932456597313,
      "loss": 1.9918,
      "step": 124863
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.9407363533973694,
      "learning_rate": 0.00026058919054479065,
      "loss": 2.1101,
      "step": 124864
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.942421555519104,
      "learning_rate": 0.00026058513543717723,
      "loss": 1.9788,
      "step": 124865
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.8771522045135498,
      "learning_rate": 0.0002605810803368918,
      "loss": 2.2025,
      "step": 124866
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.9798285961151123,
      "learning_rate": 0.0002605770252439351,
      "loss": 2.0938,
      "step": 124867
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.0157933235168457,
      "learning_rate": 0.00026057297015830784,
      "loss": 2.0719,
      "step": 124868
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.0092960596084595,
      "learning_rate": 0.0002605689150800109,
      "loss": 2.1139,
      "step": 124869
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.012338399887085,
      "learning_rate": 0.00026056486000904485,
      "loss": 2.0054,
      "step": 124870
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.2143727540969849,
      "learning_rate": 0.0002605608049454105,
      "loss": 2.0497,
      "step": 124871
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.9824573993682861,
      "learning_rate": 0.0002605567498891087,
      "loss": 2.0283,
      "step": 124872
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.9822503328323364,
      "learning_rate": 0.00026055269484014017,
      "loss": 1.9565,
      "step": 124873
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.9282310009002686,
      "learning_rate": 0.00026054863979850556,
      "loss": 2.2396,
      "step": 124874
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.1580805778503418,
      "learning_rate": 0.0002605445847642059,
      "loss": 2.054,
      "step": 124875
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.124235987663269,
      "learning_rate": 0.00026054052973724156,
      "loss": 2.2643,
      "step": 124876
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.1262190341949463,
      "learning_rate": 0.00026053647471761354,
      "loss": 2.1022,
      "step": 124877
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.8974653482437134,
      "learning_rate": 0.0002605324197053225,
      "loss": 2.0078,
      "step": 124878
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.064263105392456,
      "learning_rate": 0.00026052836470036924,
      "loss": 1.9545,
      "step": 124879
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.4438655376434326,
      "learning_rate": 0.0002605243097027545,
      "loss": 2.1459,
      "step": 124880
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.096619963645935,
      "learning_rate": 0.0002605202547124791,
      "loss": 2.2432,
      "step": 124881
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.9643296003341675,
      "learning_rate": 0.0002605161997295437,
      "loss": 2.23,
      "step": 124882
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.0774164199829102,
      "learning_rate": 0.000260512144753949,
      "loss": 1.976,
      "step": 124883
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.9737874865531921,
      "learning_rate": 0.0002605080897856959,
      "loss": 2.0402,
      "step": 124884
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.076688528060913,
      "learning_rate": 0.00026050403482478504,
      "loss": 2.052,
      "step": 124885
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.8675500750541687,
      "learning_rate": 0.0002604999798712173,
      "loss": 2.2558,
      "step": 124886
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.9332187175750732,
      "learning_rate": 0.00026049592492499344,
      "loss": 2.1563,
      "step": 124887
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.9370501637458801,
      "learning_rate": 0.000260491869986114,
      "loss": 1.9433,
      "step": 124888
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.1316192150115967,
      "learning_rate": 0.0002604878150545799,
      "loss": 2.027,
      "step": 124889
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.9841716289520264,
      "learning_rate": 0.0002604837601303918,
      "loss": 2.1416,
      "step": 124890
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.9581972360610962,
      "learning_rate": 0.00026047970521355057,
      "loss": 2.2487,
      "step": 124891
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.0619980096817017,
      "learning_rate": 0.0002604756503040569,
      "loss": 2.3147,
      "step": 124892
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.0369881391525269,
      "learning_rate": 0.00026047159540191166,
      "loss": 2.2705,
      "step": 124893
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.0509458780288696,
      "learning_rate": 0.00026046754050711537,
      "loss": 2.2298,
      "step": 124894
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.9318580627441406,
      "learning_rate": 0.0002604634856196689,
      "loss": 2.1874,
      "step": 124895
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.5121128559112549,
      "learning_rate": 0.000260459430739573,
      "loss": 1.9967,
      "step": 124896
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.1578783988952637,
      "learning_rate": 0.0002604553758668285,
      "loss": 2.0248,
      "step": 124897
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.0461121797561646,
      "learning_rate": 0.0002604513210014361,
      "loss": 2.0498,
      "step": 124898
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.0114649534225464,
      "learning_rate": 0.0002604472661433966,
      "loss": 2.2056,
      "step": 124899
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.1831004619598389,
      "learning_rate": 0.0002604432112927106,
      "loss": 1.942,
      "step": 124900
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.2377197742462158,
      "learning_rate": 0.000260439156449379,
      "loss": 2.4701,
      "step": 124901
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.1863393783569336,
      "learning_rate": 0.0002604351016134025,
      "loss": 2.0265,
      "step": 124902
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.9839810729026794,
      "learning_rate": 0.0002604310467847818,
      "loss": 2.5525,
      "step": 124903
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.0579255819320679,
      "learning_rate": 0.00026042699196351775,
      "loss": 1.9606,
      "step": 124904
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.9932228326797485,
      "learning_rate": 0.0002604229371496112,
      "loss": 2.1812,
      "step": 124905
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.9331566095352173,
      "learning_rate": 0.00026041888234306265,
      "loss": 2.1177,
      "step": 124906
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.052943468093872,
      "learning_rate": 0.00026041482754387294,
      "loss": 2.2213,
      "step": 124907
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.019033670425415,
      "learning_rate": 0.0002604107727520429,
      "loss": 2.0398,
      "step": 124908
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.0658338069915771,
      "learning_rate": 0.0002604067179675732,
      "loss": 2.1008,
      "step": 124909
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.0510215759277344,
      "learning_rate": 0.0002604026631904647,
      "loss": 2.1058,
      "step": 124910
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.0184801816940308,
      "learning_rate": 0.00026039860842071806,
      "loss": 1.9003,
      "step": 124911
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.102283239364624,
      "learning_rate": 0.00026039455365833424,
      "loss": 2.2409,
      "step": 124912
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.9264249205589294,
      "learning_rate": 0.0002603904989033136,
      "loss": 2.1556,
      "step": 124913
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.3011772632598877,
      "learning_rate": 0.00026038644415565717,
      "loss": 1.9022,
      "step": 124914
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.9549372792243958,
      "learning_rate": 0.00026038238941536565,
      "loss": 1.8206,
      "step": 124915
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.9700857996940613,
      "learning_rate": 0.00026037833468243977,
      "loss": 2.1262,
      "step": 124916
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.2390930652618408,
      "learning_rate": 0.00026037427995688033,
      "loss": 2.2804,
      "step": 124917
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.0574389696121216,
      "learning_rate": 0.00026037022523868816,
      "loss": 2.1137,
      "step": 124918
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.9807855486869812,
      "learning_rate": 0.0002603661705278638,
      "loss": 1.9537,
      "step": 124919
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.9586657285690308,
      "learning_rate": 0.00026036211582440815,
      "loss": 2.1881,
      "step": 124920
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.979622483253479,
      "learning_rate": 0.0002603580611283219,
      "loss": 2.2249,
      "step": 124921
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.9560346007347107,
      "learning_rate": 0.0002603540064396058,
      "loss": 2.1052,
      "step": 124922
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.986796498298645,
      "learning_rate": 0.0002603499517582607,
      "loss": 2.0611,
      "step": 124923
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.999655544757843,
      "learning_rate": 0.0002603458970842874,
      "loss": 2.0385,
      "step": 124924
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.995722234249115,
      "learning_rate": 0.0002603418424176864,
      "loss": 2.299,
      "step": 124925
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.9772602319717407,
      "learning_rate": 0.0002603377877584586,
      "loss": 1.9952,
      "step": 124926
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.0012524127960205,
      "learning_rate": 0.0002603337331066048,
      "loss": 2.0891,
      "step": 124927
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.9799079895019531,
      "learning_rate": 0.0002603296784621257,
      "loss": 2.2285,
      "step": 124928
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.9053435325622559,
      "learning_rate": 0.00026032562382502197,
      "loss": 2.0016,
      "step": 124929
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.0850971937179565,
      "learning_rate": 0.0002603215691952947,
      "loss": 2.0853,
      "step": 124930
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.0812647342681885,
      "learning_rate": 0.0002603175145729442,
      "loss": 2.1028,
      "step": 124931
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.055440902709961,
      "learning_rate": 0.0002603134599579714,
      "loss": 2.168,
      "step": 124932
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.9649228453636169,
      "learning_rate": 0.00026030940535037714,
      "loss": 2.2894,
      "step": 124933
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.9505383372306824,
      "learning_rate": 0.0002603053507501621,
      "loss": 1.9972,
      "step": 124934
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.9308295249938965,
      "learning_rate": 0.000260301296157327,
      "loss": 2.1511,
      "step": 124935
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.1246695518493652,
      "learning_rate": 0.0002602972415718728,
      "loss": 2.2452,
      "step": 124936
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.9627590179443359,
      "learning_rate": 0.00026029318699379993,
      "loss": 2.168,
      "step": 124937
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.1903225183486938,
      "learning_rate": 0.0002602891324231094,
      "loss": 2.0882,
      "step": 124938
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.0593940019607544,
      "learning_rate": 0.00026028507785980185,
      "loss": 2.162,
      "step": 124939
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.0398938655853271,
      "learning_rate": 0.000260281023303878,
      "loss": 2.1887,
      "step": 124940
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.9881006479263306,
      "learning_rate": 0.0002602769687553387,
      "loss": 2.1382,
      "step": 124941
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.935805082321167,
      "learning_rate": 0.00026027291421418464,
      "loss": 2.1249,
      "step": 124942
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.0450425148010254,
      "learning_rate": 0.0002602688596804166,
      "loss": 2.1977,
      "step": 124943
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.069782018661499,
      "learning_rate": 0.0002602648051540353,
      "loss": 1.7957,
      "step": 124944
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.8497757911682129,
      "learning_rate": 0.0002602607506350416,
      "loss": 1.8294,
      "step": 124945
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.9551783204078674,
      "learning_rate": 0.0002602566961234362,
      "loss": 2.3673,
      "step": 124946
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.028450846672058,
      "learning_rate": 0.0002602526416192197,
      "loss": 2.2148,
      "step": 124947
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.0044019222259521,
      "learning_rate": 0.0002602485871223931,
      "loss": 2.0729,
      "step": 124948
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.9119505286216736,
      "learning_rate": 0.00026024453263295697,
      "loss": 2.1193,
      "step": 124949
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.9283754229545593,
      "learning_rate": 0.0002602404781509121,
      "loss": 2.1856,
      "step": 124950
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.9698191285133362,
      "learning_rate": 0.00026023642367625934,
      "loss": 1.9792,
      "step": 124951
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.9676677584648132,
      "learning_rate": 0.0002602323692089994,
      "loss": 2.1175,
      "step": 124952
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.9931720495223999,
      "learning_rate": 0.000260228314749133,
      "loss": 2.183,
      "step": 124953
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.9926303625106812,
      "learning_rate": 0.0002602242602966609,
      "loss": 2.0129,
      "step": 124954
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.9252560138702393,
      "learning_rate": 0.0002602202058515838,
      "loss": 2.1906,
      "step": 124955
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.0485602617263794,
      "learning_rate": 0.00026021615141390254,
      "loss": 2.299,
      "step": 124956
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.0128117799758911,
      "learning_rate": 0.00026021209698361787,
      "loss": 2.4767,
      "step": 124957
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.042612075805664,
      "learning_rate": 0.0002602080425607305,
      "loss": 1.974,
      "step": 124958
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.0152760744094849,
      "learning_rate": 0.0002602039881452413,
      "loss": 1.9358,
      "step": 124959
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.9631757736206055,
      "learning_rate": 0.0002601999337371509,
      "loss": 1.8537,
      "step": 124960
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.8747482299804688,
      "learning_rate": 0.00026019587933646,
      "loss": 1.8133,
      "step": 124961
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.9990147352218628,
      "learning_rate": 0.0002601918249431695,
      "loss": 1.9998,
      "step": 124962
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.9609633088111877,
      "learning_rate": 0.00026018777055728003,
      "loss": 2.1805,
      "step": 124963
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.0691325664520264,
      "learning_rate": 0.0002601837161787924,
      "loss": 2.2069,
      "step": 124964
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.9107591509819031,
      "learning_rate": 0.0002601796618077074,
      "loss": 2.0722,
      "step": 124965
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.9703953862190247,
      "learning_rate": 0.0002601756074440259,
      "loss": 1.8811,
      "step": 124966
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.9902045726776123,
      "learning_rate": 0.00026017155308774836,
      "loss": 2.3097,
      "step": 124967
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.0068464279174805,
      "learning_rate": 0.00026016749873887567,
      "loss": 2.0924,
      "step": 124968
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.0799753665924072,
      "learning_rate": 0.00026016344439740857,
      "loss": 2.1918,
      "step": 124969
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.127897024154663,
      "learning_rate": 0.0002601593900633479,
      "loss": 2.1798,
      "step": 124970
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.9737550020217896,
      "learning_rate": 0.00026015533573669434,
      "loss": 2.0639,
      "step": 124971
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.0345966815948486,
      "learning_rate": 0.0002601512814174487,
      "loss": 2.1582,
      "step": 124972
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.0380405187606812,
      "learning_rate": 0.0002601472271056116,
      "loss": 2.1414,
      "step": 124973
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.074818730354309,
      "learning_rate": 0.00026014317280118396,
      "loss": 1.9478,
      "step": 124974
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.41095769405365,
      "learning_rate": 0.0002601391185041664,
      "loss": 2.0017,
      "step": 124975
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.9083593487739563,
      "learning_rate": 0.0002601350642145597,
      "loss": 2.1888,
      "step": 124976
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.9943230152130127,
      "learning_rate": 0.0002601310099323647,
      "loss": 2.4364,
      "step": 124977
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.155053734779358,
      "learning_rate": 0.0002601269556575821,
      "loss": 2.0628,
      "step": 124978
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.1016809940338135,
      "learning_rate": 0.00026012290139021274,
      "loss": 2.0043,
      "step": 124979
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.1713345050811768,
      "learning_rate": 0.00026011884713025716,
      "loss": 2.0926,
      "step": 124980
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.0206215381622314,
      "learning_rate": 0.0002601147928777162,
      "loss": 2.1665,
      "step": 124981
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.482940435409546,
      "learning_rate": 0.0002601107386325907,
      "loss": 1.9944,
      "step": 124982
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.2526832818984985,
      "learning_rate": 0.0002601066843948814,
      "loss": 2.0163,
      "step": 124983
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.0706791877746582,
      "learning_rate": 0.000260102630164589,
      "loss": 2.1779,
      "step": 124984
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.0054774284362793,
      "learning_rate": 0.0002600985759417144,
      "loss": 2.2752,
      "step": 124985
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.999528706073761,
      "learning_rate": 0.00026009452172625806,
      "loss": 2.3758,
      "step": 124986
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.1241536140441895,
      "learning_rate": 0.00026009046751822095,
      "loss": 2.184,
      "step": 124987
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.078582525253296,
      "learning_rate": 0.00026008641331760375,
      "loss": 2.1674,
      "step": 124988
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.1883608102798462,
      "learning_rate": 0.00026008235912440726,
      "loss": 2.1575,
      "step": 124989
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.0151581764221191,
      "learning_rate": 0.0002600783049386322,
      "loss": 2.0982,
      "step": 124990
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.038177728652954,
      "learning_rate": 0.00026007425076027946,
      "loss": 2.1091,
      "step": 124991
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.2059026956558228,
      "learning_rate": 0.00026007019658934956,
      "loss": 2.283,
      "step": 124992
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.9618520736694336,
      "learning_rate": 0.00026006614242584335,
      "loss": 2.2528,
      "step": 124993
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.9741659760475159,
      "learning_rate": 0.0002600620882697616,
      "loss": 2.0067,
      "step": 124994
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.0323809385299683,
      "learning_rate": 0.00026005803412110505,
      "loss": 2.1831,
      "step": 124995
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.9737314581871033,
      "learning_rate": 0.0002600539799798745,
      "loss": 2.0081,
      "step": 124996
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.8939173817634583,
      "learning_rate": 0.00026004992584607075,
      "loss": 2.0619,
      "step": 124997
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.9947583079338074,
      "learning_rate": 0.0002600458717196944,
      "loss": 1.9784,
      "step": 124998
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.0909216403961182,
      "learning_rate": 0.00026004181760074625,
      "loss": 2.0401,
      "step": 124999
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.0844552516937256,
      "learning_rate": 0.00026003776348922704,
      "loss": 2.2156,
      "step": 125000
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.0494283437728882,
      "learning_rate": 0.0002600337093851376,
      "loss": 2.1733,
      "step": 125001
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.0920504331588745,
      "learning_rate": 0.00026002965528847864,
      "loss": 2.3158,
      "step": 125002
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.1851149797439575,
      "learning_rate": 0.00026002560119925107,
      "loss": 2.3408,
      "step": 125003
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.9980299472808838,
      "learning_rate": 0.0002600215471174554,
      "loss": 1.8621,
      "step": 125004
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.9106248021125793,
      "learning_rate": 0.0002600174930430924,
      "loss": 2.1065,
      "step": 125005
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.292128324508667,
      "learning_rate": 0.00026001343897616295,
      "loss": 1.8737,
      "step": 125006
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.9323277473449707,
      "learning_rate": 0.0002600093849166677,
      "loss": 2.0602,
      "step": 125007
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.186018705368042,
      "learning_rate": 0.0002600053308646075,
      "loss": 2.352,
      "step": 125008
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.9696581363677979,
      "learning_rate": 0.0002600012768199832,
      "loss": 1.9228,
      "step": 125009
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.0315223932266235,
      "learning_rate": 0.0002599972227827953,
      "loss": 2.2225,
      "step": 125010
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.0066932439804077,
      "learning_rate": 0.00025999316875304466,
      "loss": 2.1446,
      "step": 125011
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.9481390714645386,
      "learning_rate": 0.00025998911473073204,
      "loss": 1.9375,
      "step": 125012
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.9233559966087341,
      "learning_rate": 0.0002599850607158582,
      "loss": 2.1117,
      "step": 125013
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.0130670070648193,
      "learning_rate": 0.00025998100670842386,
      "loss": 2.2291,
      "step": 125014
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.9796457290649414,
      "learning_rate": 0.00025997695270843003,
      "loss": 1.8659,
      "step": 125015
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.9484891295433044,
      "learning_rate": 0.000259972898715877,
      "loss": 2.122,
      "step": 125016
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.973633348941803,
      "learning_rate": 0.0002599688447307658,
      "loss": 2.0015,
      "step": 125017
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.9518761038780212,
      "learning_rate": 0.0002599647907530972,
      "loss": 2.1709,
      "step": 125018
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.9574440717697144,
      "learning_rate": 0.00025996073678287185,
      "loss": 2.1111,
      "step": 125019
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.0330300331115723,
      "learning_rate": 0.00025995668282009055,
      "loss": 2.2216,
      "step": 125020
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.9454126954078674,
      "learning_rate": 0.00025995262886475414,
      "loss": 2.0755,
      "step": 125021
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.9682589769363403,
      "learning_rate": 0.0002599485749168633,
      "loss": 2.1257,
      "step": 125022
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.019898533821106,
      "learning_rate": 0.0002599445209764187,
      "loss": 2.0087,
      "step": 125023
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.9706348776817322,
      "learning_rate": 0.0002599404670434212,
      "loss": 2.1416,
      "step": 125024
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.9744405746459961,
      "learning_rate": 0.00025993641311787145,
      "loss": 2.1703,
      "step": 125025
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.9410555362701416,
      "learning_rate": 0.00025993235919977034,
      "loss": 2.157,
      "step": 125026
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.041979432106018,
      "learning_rate": 0.0002599283052891186,
      "loss": 1.9471,
      "step": 125027
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.007071852684021,
      "learning_rate": 0.0002599242513859169,
      "loss": 2.2985,
      "step": 125028
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.1328370571136475,
      "learning_rate": 0.00025992019749016606,
      "loss": 2.0954,
      "step": 125029
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.9643887281417847,
      "learning_rate": 0.00025991614360186684,
      "loss": 2.3128,
      "step": 125030
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.99873286485672,
      "learning_rate": 0.0002599120897210199,
      "loss": 2.1481,
      "step": 125031
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.0910139083862305,
      "learning_rate": 0.0002599080358476261,
      "loss": 2.102,
      "step": 125032
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.1374233961105347,
      "learning_rate": 0.0002599039819816861,
      "loss": 2.0966,
      "step": 125033
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.8799238204956055,
      "learning_rate": 0.00025989992812320076,
      "loss": 1.9061,
      "step": 125034
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.94619220495224,
      "learning_rate": 0.0002598958742721707,
      "loss": 2.1868,
      "step": 125035
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.0152534246444702,
      "learning_rate": 0.00025989182042859685,
      "loss": 2.1881,
      "step": 125036
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.9907786250114441,
      "learning_rate": 0.00025988776659247986,
      "loss": 2.0408,
      "step": 125037
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.0247464179992676,
      "learning_rate": 0.00025988371276382044,
      "loss": 2.2627,
      "step": 125038
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.9552541971206665,
      "learning_rate": 0.00025987965894261946,
      "loss": 2.0348,
      "step": 125039
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.9066876769065857,
      "learning_rate": 0.0002598756051288775,
      "loss": 2.0505,
      "step": 125040
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.111247181892395,
      "learning_rate": 0.0002598715513225955,
      "loss": 2.0659,
      "step": 125041
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.0500073432922363,
      "learning_rate": 0.00025986749752377413,
      "loss": 2.0609,
      "step": 125042
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.9524796605110168,
      "learning_rate": 0.00025986344373241414,
      "loss": 2.1229,
      "step": 125043
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.9259679913520813,
      "learning_rate": 0.00025985938994851633,
      "loss": 2.0566,
      "step": 125044
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.9923995733261108,
      "learning_rate": 0.00025985533617208135,
      "loss": 2.1019,
      "step": 125045
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.9658459424972534,
      "learning_rate": 0.0002598512824031101,
      "loss": 2.1456,
      "step": 125046
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.9773784279823303,
      "learning_rate": 0.0002598472286416032,
      "loss": 2.1548,
      "step": 125047
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.9569558501243591,
      "learning_rate": 0.00025984317488756146,
      "loss": 1.9646,
      "step": 125048
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.0351992845535278,
      "learning_rate": 0.00025983912114098563,
      "loss": 2.1851,
      "step": 125049
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.9664618968963623,
      "learning_rate": 0.00025983506740187643,
      "loss": 2.2745,
      "step": 125050
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.9590294361114502,
      "learning_rate": 0.00025983101367023473,
      "loss": 1.9388,
      "step": 125051
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.9479044079780579,
      "learning_rate": 0.00025982695994606123,
      "loss": 2.0922,
      "step": 125052
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.025292992591858,
      "learning_rate": 0.00025982290622935663,
      "loss": 2.296,
      "step": 125053
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.000546932220459,
      "learning_rate": 0.00025981885252012165,
      "loss": 2.1756,
      "step": 125054
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.9330748319625854,
      "learning_rate": 0.0002598147988183571,
      "loss": 2.0777,
      "step": 125055
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.1167434453964233,
      "learning_rate": 0.0002598107451240637,
      "loss": 2.0001,
      "step": 125056
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.0157221555709839,
      "learning_rate": 0.0002598066914372423,
      "loss": 2.2195,
      "step": 125057
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.9876275062561035,
      "learning_rate": 0.00025980263775789375,
      "loss": 2.192,
      "step": 125058
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.9232349991798401,
      "learning_rate": 0.0002597985840860185,
      "loss": 1.9129,
      "step": 125059
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.9860227108001709,
      "learning_rate": 0.0002597945304216174,
      "loss": 1.9957,
      "step": 125060
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.1444000005722046,
      "learning_rate": 0.0002597904767646913,
      "loss": 1.9347,
      "step": 125061
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.015002965927124,
      "learning_rate": 0.0002597864231152409,
      "loss": 2.1582,
      "step": 125062
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.9633775353431702,
      "learning_rate": 0.000259782369473267,
      "loss": 2.2897,
      "step": 125063
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.0160620212554932,
      "learning_rate": 0.00025977831583877034,
      "loss": 2.028,
      "step": 125064
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.9805519580841064,
      "learning_rate": 0.0002597742622117516,
      "loss": 2.0643,
      "step": 125065
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.9216878414154053,
      "learning_rate": 0.00025977020859221163,
      "loss": 2.2071,
      "step": 125066
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.894537091255188,
      "learning_rate": 0.0002597661549801511,
      "loss": 2.1283,
      "step": 125067
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.0493754148483276,
      "learning_rate": 0.0002597621013755707,
      "loss": 2.3429,
      "step": 125068
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.078586220741272,
      "learning_rate": 0.0002597580477784714,
      "loss": 2.0129,
      "step": 125069
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.1187158823013306,
      "learning_rate": 0.00025975399418885395,
      "loss": 2.2137,
      "step": 125070
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.9220367670059204,
      "learning_rate": 0.00025974994060671885,
      "loss": 2.169,
      "step": 125071
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.017600655555725,
      "learning_rate": 0.00025974588703206697,
      "loss": 2.0648,
      "step": 125072
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.9486347436904907,
      "learning_rate": 0.0002597418334648991,
      "loss": 2.1784,
      "step": 125073
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.0738691091537476,
      "learning_rate": 0.00025973777990521605,
      "loss": 2.2878,
      "step": 125074
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.9531606435775757,
      "learning_rate": 0.0002597337263530184,
      "loss": 2.1905,
      "step": 125075
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.2533735036849976,
      "learning_rate": 0.0002597296728083072,
      "loss": 2.1988,
      "step": 125076
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.9245949983596802,
      "learning_rate": 0.0002597256192710829,
      "loss": 1.798,
      "step": 125077
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.0981427431106567,
      "learning_rate": 0.0002597215657413463,
      "loss": 2.0546,
      "step": 125078
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.003862738609314,
      "learning_rate": 0.0002597175122190983,
      "loss": 2.2478,
      "step": 125079
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.4883328676223755,
      "learning_rate": 0.00025971345870433955,
      "loss": 2.1373,
      "step": 125080
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.0930020809173584,
      "learning_rate": 0.00025970940519707083,
      "loss": 1.9704,
      "step": 125081
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.077090859413147,
      "learning_rate": 0.00025970535169729303,
      "loss": 2.0428,
      "step": 125082
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.0438339710235596,
      "learning_rate": 0.00025970129820500657,
      "loss": 2.0708,
      "step": 125083
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.9543412327766418,
      "learning_rate": 0.0002596972447202125,
      "loss": 2.2496,
      "step": 125084
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.0291354656219482,
      "learning_rate": 0.0002596931912429114,
      "loss": 2.1713,
      "step": 125085
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.0012575387954712,
      "learning_rate": 0.0002596891377731041,
      "loss": 1.9349,
      "step": 125086
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.0489420890808105,
      "learning_rate": 0.00025968508431079137,
      "loss": 2.1298,
      "step": 125087
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.9802459478378296,
      "learning_rate": 0.0002596810308559741,
      "loss": 2.1203,
      "step": 125088
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.898516833782196,
      "learning_rate": 0.0002596769774086527,
      "loss": 1.8589,
      "step": 125089
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.9513669013977051,
      "learning_rate": 0.00025967292396882817,
      "loss": 2.1945,
      "step": 125090
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.9611601829528809,
      "learning_rate": 0.0002596688705365012,
      "loss": 2.2059,
      "step": 125091
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.0490642786026,
      "learning_rate": 0.0002596648171116725,
      "loss": 2.1285,
      "step": 125092
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.939727783203125,
      "learning_rate": 0.00025966076369434284,
      "loss": 2.0469,
      "step": 125093
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.0110455751419067,
      "learning_rate": 0.0002596567102845132,
      "loss": 2.2012,
      "step": 125094
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.9361426830291748,
      "learning_rate": 0.00025965265688218404,
      "loss": 2.0815,
      "step": 125095
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.9969578981399536,
      "learning_rate": 0.0002596486034873562,
      "loss": 2.0252,
      "step": 125096
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.047115683555603,
      "learning_rate": 0.00025964455010003037,
      "loss": 2.3107,
      "step": 125097
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.9651046395301819,
      "learning_rate": 0.00025964049672020745,
      "loss": 1.9178,
      "step": 125098
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.9998050332069397,
      "learning_rate": 0.00025963644334788807,
      "loss": 1.9533,
      "step": 125099
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.9547057747840881,
      "learning_rate": 0.0002596323899830732,
      "loss": 2.2111,
      "step": 125100
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.0132906436920166,
      "learning_rate": 0.0002596283366257633,
      "loss": 2.0962,
      "step": 125101
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.9616456627845764,
      "learning_rate": 0.0002596242832759593,
      "loss": 2.0521,
      "step": 125102
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.944815993309021,
      "learning_rate": 0.0002596202299336618,
      "loss": 2.0046,
      "step": 125103
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.9644836187362671,
      "learning_rate": 0.0002596161765988717,
      "loss": 2.1484,
      "step": 125104
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.0452569723129272,
      "learning_rate": 0.00025961212327158976,
      "loss": 2.3958,
      "step": 125105
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.9020310044288635,
      "learning_rate": 0.00025960806995181673,
      "loss": 2.1142,
      "step": 125106
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.9793842434883118,
      "learning_rate": 0.00025960401663955337,
      "loss": 2.2765,
      "step": 125107
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.949840784072876,
      "learning_rate": 0.00025959996333480023,
      "loss": 2.0901,
      "step": 125108
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.9853284358978271,
      "learning_rate": 0.0002595959100375583,
      "loss": 2.2677,
      "step": 125109
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.9896283149719238,
      "learning_rate": 0.00025959185674782816,
      "loss": 2.2151,
      "step": 125110
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.124008059501648,
      "learning_rate": 0.00025958780346561076,
      "loss": 2.094,
      "step": 125111
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.2679922580718994,
      "learning_rate": 0.0002595837501909067,
      "loss": 2.1561,
      "step": 125112
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.9370543360710144,
      "learning_rate": 0.0002595796969237169,
      "loss": 2.267,
      "step": 125113
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.0910714864730835,
      "learning_rate": 0.0002595756436640419,
      "loss": 2.0976,
      "step": 125114
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.2040836811065674,
      "learning_rate": 0.0002595715904118826,
      "loss": 2.0289,
      "step": 125115
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.0908054113388062,
      "learning_rate": 0.00025956753716723955,
      "loss": 2.2817,
      "step": 125116
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.025708556175232,
      "learning_rate": 0.0002595634839301138,
      "loss": 2.1369,
      "step": 125117
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.0374212265014648,
      "learning_rate": 0.0002595594307005059,
      "loss": 2.1079,
      "step": 125118
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.176619529724121,
      "learning_rate": 0.0002595553774784167,
      "loss": 2.1819,
      "step": 125119
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.136290192604065,
      "learning_rate": 0.0002595513242638469,
      "loss": 2.0478,
      "step": 125120
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.0633186101913452,
      "learning_rate": 0.00025954727105679734,
      "loss": 2.0866,
      "step": 125121
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.947350263595581,
      "learning_rate": 0.0002595432178572687,
      "loss": 2.2345,
      "step": 125122
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.0385630130767822,
      "learning_rate": 0.0002595391646652616,
      "loss": 1.9807,
      "step": 125123
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.1835699081420898,
      "learning_rate": 0.000259535111480777,
      "loss": 2.2169,
      "step": 125124
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.0463470220565796,
      "learning_rate": 0.0002595310583038157,
      "loss": 2.0659,
      "step": 125125
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.9626613259315491,
      "learning_rate": 0.00025952700513437823,
      "loss": 2.0936,
      "step": 125126
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.084884524345398,
      "learning_rate": 0.0002595229519724654,
      "loss": 2.3902,
      "step": 125127
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.2462881803512573,
      "learning_rate": 0.0002595188988180781,
      "loss": 2.2923,
      "step": 125128
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.9829659461975098,
      "learning_rate": 0.00025951484567121705,
      "loss": 2.1268,
      "step": 125129
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.211026668548584,
      "learning_rate": 0.0002595107925318829,
      "loss": 1.9954,
      "step": 125130
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.2151527404785156,
      "learning_rate": 0.0002595067394000765,
      "loss": 1.9396,
      "step": 125131
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.264440894126892,
      "learning_rate": 0.0002595026862757985,
      "loss": 1.9771,
      "step": 125132
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.0329523086547852,
      "learning_rate": 0.0002594986331590497,
      "loss": 2.4041,
      "step": 125133
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.0260009765625,
      "learning_rate": 0.00025949458004983085,
      "loss": 1.9347,
      "step": 125134
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.9803746938705444,
      "learning_rate": 0.00025949052694814275,
      "loss": 2.188,
      "step": 125135
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.9868265390396118,
      "learning_rate": 0.0002594864738539862,
      "loss": 2.069,
      "step": 125136
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.9939747452735901,
      "learning_rate": 0.0002594824207673619,
      "loss": 2.1363,
      "step": 125137
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.251028060913086,
      "learning_rate": 0.00025947836768827045,
      "loss": 2.1887,
      "step": 125138
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.0091339349746704,
      "learning_rate": 0.0002594743146167128,
      "loss": 2.1444,
      "step": 125139
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.9643377065658569,
      "learning_rate": 0.00025947026155268964,
      "loss": 2.1829,
      "step": 125140
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.9177997708320618,
      "learning_rate": 0.00025946620849620167,
      "loss": 2.153,
      "step": 125141
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.947713315486908,
      "learning_rate": 0.0002594621554472497,
      "loss": 2.0812,
      "step": 125142
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.9154220223426819,
      "learning_rate": 0.0002594581024058347,
      "loss": 2.1193,
      "step": 125143
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.9348841905593872,
      "learning_rate": 0.000259454049371957,
      "loss": 2.35,
      "step": 125144
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.070957064628601,
      "learning_rate": 0.00025944999634561753,
      "loss": 2.1886,
      "step": 125145
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.9592318534851074,
      "learning_rate": 0.0002594459433268171,
      "loss": 2.2526,
      "step": 125146
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.9638336300849915,
      "learning_rate": 0.0002594418903155565,
      "loss": 1.9308,
      "step": 125147
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.056317925453186,
      "learning_rate": 0.00025943783731183635,
      "loss": 2.1324,
      "step": 125148
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.97181236743927,
      "learning_rate": 0.0002594337843156576,
      "loss": 2.2243,
      "step": 125149
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.9935263991355896,
      "learning_rate": 0.0002594297313270208,
      "loss": 2.0899,
      "step": 125150
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.0055866241455078,
      "learning_rate": 0.0002594256783459267,
      "loss": 2.2108,
      "step": 125151
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.971040666103363,
      "learning_rate": 0.0002594216253723762,
      "loss": 2.2439,
      "step": 125152
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.9454168677330017,
      "learning_rate": 0.00025941757240636993,
      "loss": 1.893,
      "step": 125153
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.9956381916999817,
      "learning_rate": 0.00025941351944790874,
      "loss": 2.1809,
      "step": 125154
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.2125098705291748,
      "learning_rate": 0.0002594094664969935,
      "loss": 2.0898,
      "step": 125155
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.0288718938827515,
      "learning_rate": 0.0002594054135536246,
      "loss": 2.344,
      "step": 125156
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.956462562084198,
      "learning_rate": 0.00025940136061780307,
      "loss": 1.9231,
      "step": 125157
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.9626889228820801,
      "learning_rate": 0.00025939730768952953,
      "loss": 2.0862,
      "step": 125158
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.9359902739524841,
      "learning_rate": 0.0002593932547688048,
      "loss": 2.1213,
      "step": 125159
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.170715570449829,
      "learning_rate": 0.00025938920185562966,
      "loss": 2.0364,
      "step": 125160
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.9378743171691895,
      "learning_rate": 0.000259385148950005,
      "loss": 1.9935,
      "step": 125161
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.9746882915496826,
      "learning_rate": 0.0002593810960519312,
      "loss": 2.2582,
      "step": 125162
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.0895041227340698,
      "learning_rate": 0.00025937704316140927,
      "loss": 2.4141,
      "step": 125163
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.9470779299736023,
      "learning_rate": 0.00025937299027843993,
      "loss": 2.0036,
      "step": 125164
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.9445925951004028,
      "learning_rate": 0.00025936893740302386,
      "loss": 2.041,
      "step": 125165
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.9902328252792358,
      "learning_rate": 0.00025936488453516196,
      "loss": 2.3045,
      "step": 125166
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.910342812538147,
      "learning_rate": 0.0002593608316748549,
      "loss": 2.1267,
      "step": 125167
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.9880003929138184,
      "learning_rate": 0.00025935677882210335,
      "loss": 2.1088,
      "step": 125168
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.9315526485443115,
      "learning_rate": 0.00025935272597690817,
      "loss": 1.9558,
      "step": 125169
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.13627028465271,
      "learning_rate": 0.0002593486731392701,
      "loss": 2.2131,
      "step": 125170
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.0365768671035767,
      "learning_rate": 0.00025934462030918987,
      "loss": 2.3575,
      "step": 125171
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.961896538734436,
      "learning_rate": 0.0002593405674866682,
      "loss": 2.2177,
      "step": 125172
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.0175011157989502,
      "learning_rate": 0.00025933651467170606,
      "loss": 2.0,
      "step": 125173
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.0668401718139648,
      "learning_rate": 0.0002593324618643039,
      "loss": 2.1008,
      "step": 125174
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.9790964126586914,
      "learning_rate": 0.00025932840906446255,
      "loss": 2.1628,
      "step": 125175
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.0346611738204956,
      "learning_rate": 0.00025932435627218283,
      "loss": 2.3055,
      "step": 125176
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.0718482732772827,
      "learning_rate": 0.0002593203034874656,
      "loss": 2.1145,
      "step": 125177
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.1790282726287842,
      "learning_rate": 0.00025931625071031134,
      "loss": 2.2093,
      "step": 125178
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.0199637413024902,
      "learning_rate": 0.000259312197940721,
      "loss": 2.0757,
      "step": 125179
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.9108582139015198,
      "learning_rate": 0.00025930814517869543,
      "loss": 2.1842,
      "step": 125180
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.0392577648162842,
      "learning_rate": 0.0002593040924242351,
      "loss": 2.3612,
      "step": 125181
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.029685378074646,
      "learning_rate": 0.00025930003967734096,
      "loss": 1.9544,
      "step": 125182
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.1560328006744385,
      "learning_rate": 0.00025929598693801367,
      "loss": 1.9953,
      "step": 125183
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.0259768962860107,
      "learning_rate": 0.000259291934206254,
      "loss": 2.2593,
      "step": 125184
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.1090410947799683,
      "learning_rate": 0.00025928788148206277,
      "loss": 2.2623,
      "step": 125185
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.2135676145553589,
      "learning_rate": 0.00025928382876544084,
      "loss": 2.2505,
      "step": 125186
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.0448014736175537,
      "learning_rate": 0.0002592797760563886,
      "loss": 2.3747,
      "step": 125187
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.9745619297027588,
      "learning_rate": 0.00025927572335490706,
      "loss": 2.1423,
      "step": 125188
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.9492961764335632,
      "learning_rate": 0.000259271670660997,
      "loss": 2.1837,
      "step": 125189
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.0487315654754639,
      "learning_rate": 0.000259267617974659,
      "loss": 2.1569,
      "step": 125190
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.1763354539871216,
      "learning_rate": 0.000259263565295894,
      "loss": 1.9404,
      "step": 125191
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.972589373588562,
      "learning_rate": 0.00025925951262470274,
      "loss": 1.9158,
      "step": 125192
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.0389922857284546,
      "learning_rate": 0.00025925545996108583,
      "loss": 2.1847,
      "step": 125193
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.2607793807983398,
      "learning_rate": 0.0002592514073050441,
      "loss": 2.0774,
      "step": 125194
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.2159749269485474,
      "learning_rate": 0.00025924735465657824,
      "loss": 2.0284,
      "step": 125195
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.0793339014053345,
      "learning_rate": 0.0002592433020156891,
      "loss": 2.1559,
      "step": 125196
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.0237904787063599,
      "learning_rate": 0.0002592392493823774,
      "loss": 2.1215,
      "step": 125197
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.0435022115707397,
      "learning_rate": 0.00025923519675664393,
      "loss": 2.041,
      "step": 125198
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.1721723079681396,
      "learning_rate": 0.0002592311441384895,
      "loss": 2.2,
      "step": 125199
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.0503336191177368,
      "learning_rate": 0.0002592270915279146,
      "loss": 2.1449,
      "step": 125200
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.1254092454910278,
      "learning_rate": 0.0002592230389249202,
      "loss": 2.1797,
      "step": 125201
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.2322858572006226,
      "learning_rate": 0.00025921898632950697,
      "loss": 2.1854,
      "step": 125202
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.9543798565864563,
      "learning_rate": 0.00025921493374167574,
      "loss": 2.1313,
      "step": 125203
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.009091854095459,
      "learning_rate": 0.0002592108811614273,
      "loss": 2.2365,
      "step": 125204
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.994667112827301,
      "learning_rate": 0.00025920682858876227,
      "loss": 1.9414,
      "step": 125205
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.0618237257003784,
      "learning_rate": 0.00025920277602368145,
      "loss": 1.9457,
      "step": 125206
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.0610554218292236,
      "learning_rate": 0.0002591987234661856,
      "loss": 2.0611,
      "step": 125207
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.974373996257782,
      "learning_rate": 0.00025919467091627547,
      "loss": 2.1583,
      "step": 125208
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.9078016877174377,
      "learning_rate": 0.00025919061837395183,
      "loss": 2.3466,
      "step": 125209
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.1553977727890015,
      "learning_rate": 0.00025918656583921544,
      "loss": 1.9928,
      "step": 125210
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.9562355279922485,
      "learning_rate": 0.00025918251331206706,
      "loss": 2.057,
      "step": 125211
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.9826083183288574,
      "learning_rate": 0.00025917846079250733,
      "loss": 2.2862,
      "step": 125212
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.024446725845337,
      "learning_rate": 0.00025917440828053724,
      "loss": 2.0392,
      "step": 125213
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.1616450548171997,
      "learning_rate": 0.00025917035577615726,
      "loss": 2.0766,
      "step": 125214
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.1644381284713745,
      "learning_rate": 0.0002591663032793683,
      "loss": 1.8825,
      "step": 125215
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.8952800035476685,
      "learning_rate": 0.0002591622507901712,
      "loss": 2.0315,
      "step": 125216
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.9464945197105408,
      "learning_rate": 0.0002591581983085665,
      "loss": 2.049,
      "step": 125217
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.4985625743865967,
      "learning_rate": 0.0002591541458345551,
      "loss": 2.173,
      "step": 125218
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.2354886531829834,
      "learning_rate": 0.00025915009336813764,
      "loss": 2.2432,
      "step": 125219
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.9950013756752014,
      "learning_rate": 0.00025914604090931503,
      "loss": 2.2917,
      "step": 125220
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.1954385042190552,
      "learning_rate": 0.000259141988458088,
      "loss": 2.0314,
      "step": 125221
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.1381100416183472,
      "learning_rate": 0.0002591379360144572,
      "loss": 2.1574,
      "step": 125222
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.0487688779830933,
      "learning_rate": 0.0002591338835784234,
      "loss": 1.8625,
      "step": 125223
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.9565926790237427,
      "learning_rate": 0.00025912983114998735,
      "loss": 2.1243,
      "step": 125224
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.146727204322815,
      "learning_rate": 0.00025912577872914987,
      "loss": 2.0948,
      "step": 125225
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.2187191247940063,
      "learning_rate": 0.00025912172631591167,
      "loss": 2.0627,
      "step": 125226
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.2685251235961914,
      "learning_rate": 0.0002591176739102735,
      "loss": 2.053,
      "step": 125227
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.0329620838165283,
      "learning_rate": 0.00025911362151223626,
      "loss": 2.2349,
      "step": 125228
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.9699258208274841,
      "learning_rate": 0.00025910956912180046,
      "loss": 2.0458,
      "step": 125229
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.384351372718811,
      "learning_rate": 0.0002591055167389669,
      "loss": 2.2118,
      "step": 125230
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.0302802324295044,
      "learning_rate": 0.0002591014643637365,
      "loss": 2.1269,
      "step": 125231
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.9399517178535461,
      "learning_rate": 0.0002590974119961099,
      "loss": 2.1079,
      "step": 125232
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.1964699029922485,
      "learning_rate": 0.00025909335963608774,
      "loss": 2.0475,
      "step": 125233
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.0675699710845947,
      "learning_rate": 0.0002590893072836711,
      "loss": 2.0835,
      "step": 125234
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.946997344493866,
      "learning_rate": 0.00025908525493886043,
      "loss": 2.1845,
      "step": 125235
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.9610146284103394,
      "learning_rate": 0.00025908120260165657,
      "loss": 1.9414,
      "step": 125236
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.0944095849990845,
      "learning_rate": 0.0002590771502720603,
      "loss": 2.2405,
      "step": 125237
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.1513152122497559,
      "learning_rate": 0.0002590730979500723,
      "loss": 2.2851,
      "step": 125238
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.0051777362823486,
      "learning_rate": 0.0002590690456356934,
      "loss": 1.9304,
      "step": 125239
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.0116569995880127,
      "learning_rate": 0.00025906499332892447,
      "loss": 2.0625,
      "step": 125240
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.9644762873649597,
      "learning_rate": 0.0002590609410297661,
      "loss": 2.3386,
      "step": 125241
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.9859173893928528,
      "learning_rate": 0.00025905688873821894,
      "loss": 2.1341,
      "step": 125242
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.9568606019020081,
      "learning_rate": 0.00025905283645428394,
      "loss": 2.0263,
      "step": 125243
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.986534595489502,
      "learning_rate": 0.0002590487841779618,
      "loss": 2.0413,
      "step": 125244
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.0112223625183105,
      "learning_rate": 0.0002590447319092532,
      "loss": 2.1748,
      "step": 125245
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.9151502251625061,
      "learning_rate": 0.00025904067964815903,
      "loss": 1.8479,
      "step": 125246
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.9463039636611938,
      "learning_rate": 0.00025903662739468,
      "loss": 2.1259,
      "step": 125247
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.0080100297927856,
      "learning_rate": 0.00025903257514881683,
      "loss": 2.0633,
      "step": 125248
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.9172555208206177,
      "learning_rate": 0.00025902852291057016,
      "loss": 2.1913,
      "step": 125249
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.9486998319625854,
      "learning_rate": 0.000259024470679941,
      "loss": 2.0389,
      "step": 125250
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.9554473757743835,
      "learning_rate": 0.00025902041845692986,
      "loss": 1.9484,
      "step": 125251
    },
    {
      "epoch": 1.63,
      "grad_norm": 3.5563621520996094,
      "learning_rate": 0.00025901636624153763,
      "loss": 2.1083,
      "step": 125252
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.060246467590332,
      "learning_rate": 0.00025901231403376517,
      "loss": 1.9205,
      "step": 125253
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.0397307872772217,
      "learning_rate": 0.0002590082618336129,
      "loss": 2.1337,
      "step": 125254
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.0753978490829468,
      "learning_rate": 0.00025900420964108185,
      "loss": 2.2053,
      "step": 125255
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.0779259204864502,
      "learning_rate": 0.0002590001574561727,
      "loss": 2.2861,
      "step": 125256
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.02814519405365,
      "learning_rate": 0.00025899610527888614,
      "loss": 2.0735,
      "step": 125257
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.9792662858963013,
      "learning_rate": 0.00025899205310922296,
      "loss": 2.0466,
      "step": 125258
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.985853374004364,
      "learning_rate": 0.0002589880009471841,
      "loss": 2.0117,
      "step": 125259
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.0231835842132568,
      "learning_rate": 0.00025898394879277,
      "loss": 2.3003,
      "step": 125260
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.0949345827102661,
      "learning_rate": 0.00025897989664598153,
      "loss": 1.9384,
      "step": 125261
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.9031401872634888,
      "learning_rate": 0.00025897584450681956,
      "loss": 2.1114,
      "step": 125262
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.025869369506836,
      "learning_rate": 0.0002589717923752847,
      "loss": 2.0557,
      "step": 125263
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.9481406211853027,
      "learning_rate": 0.00025896774025137775,
      "loss": 2.0063,
      "step": 125264
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.9542813897132874,
      "learning_rate": 0.0002589636881350996,
      "loss": 1.9228,
      "step": 125265
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.9711139798164368,
      "learning_rate": 0.00025895963602645076,
      "loss": 1.9921,
      "step": 125266
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.9556432366371155,
      "learning_rate": 0.0002589555839254321,
      "loss": 2.3242,
      "step": 125267
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.9406368732452393,
      "learning_rate": 0.0002589515318320444,
      "loss": 1.9647,
      "step": 125268
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.9904959201812744,
      "learning_rate": 0.00025894747974628836,
      "loss": 1.8908,
      "step": 125269
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.960154116153717,
      "learning_rate": 0.00025894342766816475,
      "loss": 2.0306,
      "step": 125270
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.981292188167572,
      "learning_rate": 0.00025893937559767446,
      "loss": 2.033,
      "step": 125271
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.038304090499878,
      "learning_rate": 0.000258935323534818,
      "loss": 2.218,
      "step": 125272
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.9689388871192932,
      "learning_rate": 0.0002589312714795962,
      "loss": 1.9521,
      "step": 125273
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.0623939037322998,
      "learning_rate": 0.0002589272194320099,
      "loss": 2.1003,
      "step": 125274
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.8559499979019165,
      "learning_rate": 0.00025892316739205973,
      "loss": 1.8429,
      "step": 125275
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.9536078572273254,
      "learning_rate": 0.0002589191153597466,
      "loss": 2.1717,
      "step": 125276
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.9520350098609924,
      "learning_rate": 0.0002589150633350713,
      "loss": 1.9273,
      "step": 125277
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.0325367450714111,
      "learning_rate": 0.0002589110113180343,
      "loss": 2.2171,
      "step": 125278
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.9947466254234314,
      "learning_rate": 0.00025890695930863653,
      "loss": 2.2586,
      "step": 125279
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.9622930288314819,
      "learning_rate": 0.00025890290730687875,
      "loss": 2.3056,
      "step": 125280
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.9857425093650818,
      "learning_rate": 0.00025889885531276167,
      "loss": 2.1029,
      "step": 125281
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.8838487863540649,
      "learning_rate": 0.0002588948033262861,
      "loss": 1.9692,
      "step": 125282
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.0515636205673218,
      "learning_rate": 0.0002588907513474529,
      "loss": 2.0787,
      "step": 125283
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.958620011806488,
      "learning_rate": 0.0002588866993762625,
      "loss": 2.1282,
      "step": 125284
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.9937999844551086,
      "learning_rate": 0.0002588826474127159,
      "loss": 2.117,
      "step": 125285
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.9931887984275818,
      "learning_rate": 0.0002588785954568138,
      "loss": 2.3547,
      "step": 125286
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.1063668727874756,
      "learning_rate": 0.0002588745435085568,
      "loss": 2.1085,
      "step": 125287
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.0272297859191895,
      "learning_rate": 0.000258870491567946,
      "loss": 2.1897,
      "step": 125288
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.045332908630371,
      "learning_rate": 0.00025886643963498186,
      "loss": 2.1453,
      "step": 125289
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.958985447883606,
      "learning_rate": 0.0002588623877096652,
      "loss": 1.9625,
      "step": 125290
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.0261688232421875,
      "learning_rate": 0.00025885833579199696,
      "loss": 2.1837,
      "step": 125291
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.084019660949707,
      "learning_rate": 0.0002588542838819776,
      "loss": 2.2109,
      "step": 125292
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.922177255153656,
      "learning_rate": 0.00025885023197960797,
      "loss": 2.0908,
      "step": 125293
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.0092554092407227,
      "learning_rate": 0.00025884618008488886,
      "loss": 2.1208,
      "step": 125294
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.258290410041809,
      "learning_rate": 0.00025884212819782113,
      "loss": 2.1277,
      "step": 125295
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.0085349082946777,
      "learning_rate": 0.0002588380763184054,
      "loss": 2.2339,
      "step": 125296
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.9713836312294006,
      "learning_rate": 0.00025883402444664235,
      "loss": 2.188,
      "step": 125297
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.2866798639297485,
      "learning_rate": 0.0002588299725825329,
      "loss": 2.2291,
      "step": 125298
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.0906373262405396,
      "learning_rate": 0.00025882592072607775,
      "loss": 2.0721,
      "step": 125299
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.9653119444847107,
      "learning_rate": 0.0002588218688772776,
      "loss": 2.2603,
      "step": 125300
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.8932366967201233,
      "learning_rate": 0.0002588178170361333,
      "loss": 2.108,
      "step": 125301
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.9189572334289551,
      "learning_rate": 0.0002588137652026455,
      "loss": 2.1818,
      "step": 125302
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.8126634359359741,
      "learning_rate": 0.000258809713376815,
      "loss": 1.9761,
      "step": 125303
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.1996837854385376,
      "learning_rate": 0.00025880566155864253,
      "loss": 1.9897,
      "step": 125304
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.934733510017395,
      "learning_rate": 0.0002588016097481289,
      "loss": 1.9474,
      "step": 125305
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.0197339057922363,
      "learning_rate": 0.0002587975579452748,
      "loss": 2.0333,
      "step": 125306
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.1825443506240845,
      "learning_rate": 0.0002587935061500811,
      "loss": 2.2924,
      "step": 125307
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.0985709428787231,
      "learning_rate": 0.0002587894543625484,
      "loss": 2.2625,
      "step": 125308
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.9982364177703857,
      "learning_rate": 0.00025878540258267746,
      "loss": 2.0434,
      "step": 125309
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.0616517066955566,
      "learning_rate": 0.00025878135081046914,
      "loss": 1.9055,
      "step": 125310
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.9555376172065735,
      "learning_rate": 0.0002587772990459241,
      "loss": 2.1118,
      "step": 125311
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.058943748474121,
      "learning_rate": 0.00025877324728904326,
      "loss": 1.9326,
      "step": 125312
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.9719634652137756,
      "learning_rate": 0.00025876919553982717,
      "loss": 2.0491,
      "step": 125313
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.9793346524238586,
      "learning_rate": 0.0002587651437982767,
      "loss": 2.1566,
      "step": 125314
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.997984766960144,
      "learning_rate": 0.00025876109206439255,
      "loss": 2.009,
      "step": 125315
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.9261672496795654,
      "learning_rate": 0.0002587570403381755,
      "loss": 2.1942,
      "step": 125316
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.9888044595718384,
      "learning_rate": 0.0002587529886196262,
      "loss": 2.0699,
      "step": 125317
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.9660248160362244,
      "learning_rate": 0.00025874893690874557,
      "loss": 2.0578,
      "step": 125318
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.0524135828018188,
      "learning_rate": 0.0002587448852055343,
      "loss": 2.2393,
      "step": 125319
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.006005048751831,
      "learning_rate": 0.00025874083350999326,
      "loss": 2.0761,
      "step": 125320
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.1265290975570679,
      "learning_rate": 0.0002587367818221229,
      "loss": 2.0808,
      "step": 125321
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.1374270915985107,
      "learning_rate": 0.0002587327301419242,
      "loss": 2.0346,
      "step": 125322
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.9267762303352356,
      "learning_rate": 0.00025872867846939786,
      "loss": 2.4185,
      "step": 125323
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.0750635862350464,
      "learning_rate": 0.0002587246268045446,
      "loss": 2.3217,
      "step": 125324
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.0616120100021362,
      "learning_rate": 0.00025872057514736527,
      "loss": 2.0753,
      "step": 125325
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.0568591356277466,
      "learning_rate": 0.00025871652349786066,
      "loss": 2.0249,
      "step": 125326
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.9604434370994568,
      "learning_rate": 0.00025871247185603127,
      "loss": 2.197,
      "step": 125327
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.9894371628761292,
      "learning_rate": 0.00025870842022187804,
      "loss": 2.1371,
      "step": 125328
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.2470144033432007,
      "learning_rate": 0.0002587043685954017,
      "loss": 2.2182,
      "step": 125329
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.9807864427566528,
      "learning_rate": 0.000258700316976603,
      "loss": 2.1615,
      "step": 125330
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.0045912265777588,
      "learning_rate": 0.00025869626536548267,
      "loss": 2.0501,
      "step": 125331
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.082321286201477,
      "learning_rate": 0.00025869221376204163,
      "loss": 1.9992,
      "step": 125332
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.9613932371139526,
      "learning_rate": 0.0002586881621662804,
      "loss": 2.144,
      "step": 125333
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.0835461616516113,
      "learning_rate": 0.0002586841105781997,
      "loss": 2.2831,
      "step": 125334
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.043235182762146,
      "learning_rate": 0.0002586800589978005,
      "loss": 2.2712,
      "step": 125335
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.9427286386489868,
      "learning_rate": 0.00025867600742508344,
      "loss": 2.0415,
      "step": 125336
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.910789966583252,
      "learning_rate": 0.0002586719558600493,
      "loss": 2.4749,
      "step": 125337
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.9802440404891968,
      "learning_rate": 0.00025866790430269893,
      "loss": 2.126,
      "step": 125338
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.8908714652061462,
      "learning_rate": 0.00025866385275303286,
      "loss": 2.2238,
      "step": 125339
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.9798541069030762,
      "learning_rate": 0.000258659801211052,
      "loss": 2.1716,
      "step": 125340
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.006914496421814,
      "learning_rate": 0.000258655749676757,
      "loss": 1.9905,
      "step": 125341
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.940868616104126,
      "learning_rate": 0.0002586516981501487,
      "loss": 1.9485,
      "step": 125342
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.0181035995483398,
      "learning_rate": 0.00025864764663122776,
      "loss": 2.1149,
      "step": 125343
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.9958907961845398,
      "learning_rate": 0.00025864359511999525,
      "loss": 2.2279,
      "step": 125344
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.064159631729126,
      "learning_rate": 0.00025863954361645146,
      "loss": 2.1907,
      "step": 125345
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.9773701429367065,
      "learning_rate": 0.0002586354921205974,
      "loss": 1.924,
      "step": 125346
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.9593796730041504,
      "learning_rate": 0.0002586314406324338,
      "loss": 2.1588,
      "step": 125347
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.048105239868164,
      "learning_rate": 0.00025862738915196135,
      "loss": 2.2234,
      "step": 125348
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.9904072880744934,
      "learning_rate": 0.0002586233376791809,
      "loss": 2.3759,
      "step": 125349
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.9426695704460144,
      "learning_rate": 0.00025861928621409325,
      "loss": 2.1742,
      "step": 125350
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.9937460422515869,
      "learning_rate": 0.0002586152347566989,
      "loss": 2.1742,
      "step": 125351
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.9530757069587708,
      "learning_rate": 0.0002586111833069988,
      "loss": 2.0632,
      "step": 125352
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.9674584269523621,
      "learning_rate": 0.00025860713186499364,
      "loss": 2.1888,
      "step": 125353
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.020678162574768,
      "learning_rate": 0.0002586030804306842,
      "loss": 2.3112,
      "step": 125354
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.0975264310836792,
      "learning_rate": 0.00025859902900407126,
      "loss": 2.0859,
      "step": 125355
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.9617059230804443,
      "learning_rate": 0.0002585949775851557,
      "loss": 1.9918,
      "step": 125356
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.048197627067566,
      "learning_rate": 0.0002585909261739379,
      "loss": 2.0075,
      "step": 125357
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.9619563221931458,
      "learning_rate": 0.00025858687477041886,
      "loss": 2.0298,
      "step": 125358
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.3614610433578491,
      "learning_rate": 0.0002585828233745993,
      "loss": 2.067,
      "step": 125359
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.9680291414260864,
      "learning_rate": 0.00025857877198648,
      "loss": 2.1915,
      "step": 125360
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.0377442836761475,
      "learning_rate": 0.0002585747206060616,
      "loss": 2.3822,
      "step": 125361
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.9660107493400574,
      "learning_rate": 0.00025857066923334513,
      "loss": 2.0811,
      "step": 125362
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.9751628637313843,
      "learning_rate": 0.0002585666178683311,
      "loss": 2.0765,
      "step": 125363
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.2579526901245117,
      "learning_rate": 0.0002585625665110202,
      "loss": 2.3626,
      "step": 125364
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.9469642043113708,
      "learning_rate": 0.0002585585151614134,
      "loss": 2.1258,
      "step": 125365
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.9167484641075134,
      "learning_rate": 0.00025855446381951126,
      "loss": 2.2043,
      "step": 125366
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.9683879017829895,
      "learning_rate": 0.0002585504124853147,
      "loss": 2.1896,
      "step": 125367
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.1864596605300903,
      "learning_rate": 0.0002585463611588245,
      "loss": 2.0567,
      "step": 125368
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.9602787494659424,
      "learning_rate": 0.00025854230984004116,
      "loss": 2.0756,
      "step": 125369
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.0168064832687378,
      "learning_rate": 0.00025853825852896565,
      "loss": 2.0961,
      "step": 125370
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.0086350440979004,
      "learning_rate": 0.00025853420722559865,
      "loss": 1.9375,
      "step": 125371
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.9575941562652588,
      "learning_rate": 0.00025853015592994087,
      "loss": 2.1238,
      "step": 125372
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.9795757532119751,
      "learning_rate": 0.0002585261046419931,
      "loss": 1.9247,
      "step": 125373
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.966010332107544,
      "learning_rate": 0.0002585220533617563,
      "loss": 2.2296,
      "step": 125374
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.1026314496994019,
      "learning_rate": 0.0002585180020892309,
      "loss": 1.897,
      "step": 125375
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.9751400351524353,
      "learning_rate": 0.0002585139508244178,
      "loss": 2.1012,
      "step": 125376
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.0288832187652588,
      "learning_rate": 0.0002585098995673177,
      "loss": 2.1972,
      "step": 125377
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.9456955790519714,
      "learning_rate": 0.00025850584831793144,
      "loss": 2.1008,
      "step": 125378
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.9935477375984192,
      "learning_rate": 0.00025850179707625966,
      "loss": 2.0922,
      "step": 125379
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.9428368210792542,
      "learning_rate": 0.00025849774584230323,
      "loss": 2.2781,
      "step": 125380
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.0878115892410278,
      "learning_rate": 0.0002584936946160629,
      "loss": 2.0004,
      "step": 125381
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.9991617798805237,
      "learning_rate": 0.00025848964339753933,
      "loss": 2.1558,
      "step": 125382
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.9743616580963135,
      "learning_rate": 0.00025848559218673333,
      "loss": 2.3552,
      "step": 125383
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.971200168132782,
      "learning_rate": 0.00025848154098364565,
      "loss": 2.082,
      "step": 125384
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.0016587972640991,
      "learning_rate": 0.00025847748978827695,
      "loss": 2.0544,
      "step": 125385
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.0246204137802124,
      "learning_rate": 0.00025847343860062815,
      "loss": 2.1156,
      "step": 125386
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.9554062485694885,
      "learning_rate": 0.00025846938742069995,
      "loss": 2.0624,
      "step": 125387
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.9889515042304993,
      "learning_rate": 0.000258465336248493,
      "loss": 2.106,
      "step": 125388
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.9846104383468628,
      "learning_rate": 0.0002584612850840082,
      "loss": 2.2269,
      "step": 125389
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.948520302772522,
      "learning_rate": 0.0002584572339272462,
      "loss": 2.0149,
      "step": 125390
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.0945420265197754,
      "learning_rate": 0.00025845318277820773,
      "loss": 2.3474,
      "step": 125391
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.9101805090904236,
      "learning_rate": 0.0002584491316368936,
      "loss": 2.2343,
      "step": 125392
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.0197792053222656,
      "learning_rate": 0.0002584450805033047,
      "loss": 2.1024,
      "step": 125393
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.0686205625534058,
      "learning_rate": 0.0002584410293774415,
      "loss": 2.1082,
      "step": 125394
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.5807592868804932,
      "learning_rate": 0.00025843697825930496,
      "loss": 2.2947,
      "step": 125395
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.9823872447013855,
      "learning_rate": 0.00025843292714889575,
      "loss": 2.127,
      "step": 125396
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.9336767792701721,
      "learning_rate": 0.00025842887604621463,
      "loss": 2.0745,
      "step": 125397
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.0061841011047363,
      "learning_rate": 0.0002584248249512624,
      "loss": 2.1477,
      "step": 125398
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.045178771018982,
      "learning_rate": 0.00025842077386403983,
      "loss": 2.4033,
      "step": 125399
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.0556597709655762,
      "learning_rate": 0.0002584167227845475,
      "loss": 2.2061,
      "step": 125400
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.9792226552963257,
      "learning_rate": 0.0002584126717127864,
      "loss": 2.0126,
      "step": 125401
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.9868208765983582,
      "learning_rate": 0.00025840862064875706,
      "loss": 2.0258,
      "step": 125402
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.0388950109481812,
      "learning_rate": 0.00025840456959246033,
      "loss": 2.2522,
      "step": 125403
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.016594409942627,
      "learning_rate": 0.0002584005185438971,
      "loss": 2.0362,
      "step": 125404
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.1799941062927246,
      "learning_rate": 0.000258396467503068,
      "loss": 2.3716,
      "step": 125405
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.9797805547714233,
      "learning_rate": 0.0002583924164699737,
      "loss": 1.9515,
      "step": 125406
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.8835137486457825,
      "learning_rate": 0.0002583883654446151,
      "loss": 2.1727,
      "step": 125407
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.9921241998672485,
      "learning_rate": 0.00025838431442699275,
      "loss": 2.153,
      "step": 125408
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.9882409572601318,
      "learning_rate": 0.0002583802634171077,
      "loss": 1.9451,
      "step": 125409
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.0800108909606934,
      "learning_rate": 0.00025837621241496045,
      "loss": 2.3449,
      "step": 125410
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.8974587917327881,
      "learning_rate": 0.000258372161420552,
      "loss": 1.9677,
      "step": 125411
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.0349373817443848,
      "learning_rate": 0.00025836811043388277,
      "loss": 1.9571,
      "step": 125412
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.0167591571807861,
      "learning_rate": 0.00025836405945495373,
      "loss": 2.047,
      "step": 125413
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.9610010981559753,
      "learning_rate": 0.00025836000848376557,
      "loss": 2.0133,
      "step": 125414
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.1189545392990112,
      "learning_rate": 0.00025835595752031915,
      "loss": 2.2654,
      "step": 125415
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.0059303045272827,
      "learning_rate": 0.0002583519065646151,
      "loss": 1.89,
      "step": 125416
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.9078125357627869,
      "learning_rate": 0.0002583478556166543,
      "loss": 2.0071,
      "step": 125417
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.8986573815345764,
      "learning_rate": 0.00025834380467643737,
      "loss": 2.2106,
      "step": 125418
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.0376238822937012,
      "learning_rate": 0.0002583397537439651,
      "loss": 2.2609,
      "step": 125419
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.8913201689720154,
      "learning_rate": 0.0002583357028192382,
      "loss": 2.1442,
      "step": 125420
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.9538539052009583,
      "learning_rate": 0.00025833165190225753,
      "loss": 2.1195,
      "step": 125421
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.9946092963218689,
      "learning_rate": 0.0002583276009930238,
      "loss": 2.0025,
      "step": 125422
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.9746044874191284,
      "learning_rate": 0.00025832355009153783,
      "loss": 2.0481,
      "step": 125423
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.9214313626289368,
      "learning_rate": 0.0002583194991978002,
      "loss": 1.8944,
      "step": 125424
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.0116467475891113,
      "learning_rate": 0.00025831544831181173,
      "loss": 2.1688,
      "step": 125425
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.9576301574707031,
      "learning_rate": 0.00025831139743357325,
      "loss": 2.1569,
      "step": 125426
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.0650184154510498,
      "learning_rate": 0.00025830734656308546,
      "loss": 2.0751,
      "step": 125427
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.0201724767684937,
      "learning_rate": 0.0002583032957003491,
      "loss": 1.9406,
      "step": 125428
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.0353434085845947,
      "learning_rate": 0.0002582992448453651,
      "loss": 2.2108,
      "step": 125429
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.162391185760498,
      "learning_rate": 0.00025829519399813393,
      "loss": 1.9531,
      "step": 125430
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.8901515603065491,
      "learning_rate": 0.0002582911431586565,
      "loss": 1.9903,
      "step": 125431
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.0064280033111572,
      "learning_rate": 0.00025828709232693343,
      "loss": 2.2621,
      "step": 125432
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.9871841669082642,
      "learning_rate": 0.00025828304150296566,
      "loss": 2.3211,
      "step": 125433
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.9790782332420349,
      "learning_rate": 0.00025827899068675385,
      "loss": 2.102,
      "step": 125434
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.9327567219734192,
      "learning_rate": 0.0002582749398782989,
      "loss": 1.9824,
      "step": 125435
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.0362430810928345,
      "learning_rate": 0.00025827088907760126,
      "loss": 2.0081,
      "step": 125436
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.4110251665115356,
      "learning_rate": 0.00025826683828466184,
      "loss": 2.1137,
      "step": 125437
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.960253119468689,
      "learning_rate": 0.00025826278749948147,
      "loss": 2.0151,
      "step": 125438
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.940360963344574,
      "learning_rate": 0.0002582587367220608,
      "loss": 2.0764,
      "step": 125439
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.9361796379089355,
      "learning_rate": 0.00025825468595240065,
      "loss": 2.1625,
      "step": 125440
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.9432345032691956,
      "learning_rate": 0.0002582506351905018,
      "loss": 2.1256,
      "step": 125441
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.133404016494751,
      "learning_rate": 0.00025824658443636487,
      "loss": 2.1392,
      "step": 125442
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.9218005537986755,
      "learning_rate": 0.00025824253368999064,
      "loss": 2.0,
      "step": 125443
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.081721544265747,
      "learning_rate": 0.00025823848295137994,
      "loss": 2.1213,
      "step": 125444
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.0107003450393677,
      "learning_rate": 0.0002582344322205335,
      "loss": 2.0669,
      "step": 125445
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.9643890857696533,
      "learning_rate": 0.0002582303814974521,
      "loss": 2.1652,
      "step": 125446
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.008955955505371,
      "learning_rate": 0.00025822633078213635,
      "loss": 2.1756,
      "step": 125447
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.0230193138122559,
      "learning_rate": 0.00025822228007458734,
      "loss": 2.2389,
      "step": 125448
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.1650654077529907,
      "learning_rate": 0.00025821822937480544,
      "loss": 2.3067,
      "step": 125449
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.9553674459457397,
      "learning_rate": 0.0002582141786827916,
      "loss": 2.3739,
      "step": 125450
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.9568867087364197,
      "learning_rate": 0.0002582101279985465,
      "loss": 2.3418,
      "step": 125451
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.9917268753051758,
      "learning_rate": 0.0002582060773220709,
      "loss": 2.1947,
      "step": 125452
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.2325760126113892,
      "learning_rate": 0.0002582020266533656,
      "loss": 2.0733,
      "step": 125453
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.9710657596588135,
      "learning_rate": 0.00025819797599243145,
      "loss": 2.0429,
      "step": 125454
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.9399901032447815,
      "learning_rate": 0.000258193925339269,
      "loss": 2.1631,
      "step": 125455
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.084603190422058,
      "learning_rate": 0.00025818987469387904,
      "loss": 2.1129,
      "step": 125456
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.0190800428390503,
      "learning_rate": 0.0002581858240562624,
      "loss": 2.0811,
      "step": 125457
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.0059744119644165,
      "learning_rate": 0.0002581817734264198,
      "loss": 2.1615,
      "step": 125458
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.9829420447349548,
      "learning_rate": 0.000258177722804352,
      "loss": 1.8192,
      "step": 125459
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.9371495842933655,
      "learning_rate": 0.00025817367219005985,
      "loss": 2.0138,
      "step": 125460
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.6842011213302612,
      "learning_rate": 0.0002581696215835439,
      "loss": 2.1412,
      "step": 125461
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.9945507049560547,
      "learning_rate": 0.000258165570984805,
      "loss": 2.0201,
      "step": 125462
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.9455755949020386,
      "learning_rate": 0.0002581615203938439,
      "loss": 2.2441,
      "step": 125463
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.020218014717102,
      "learning_rate": 0.0002581574698106614,
      "loss": 1.873,
      "step": 125464
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.0395900011062622,
      "learning_rate": 0.00025815341923525817,
      "loss": 2.0312,
      "step": 125465
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.992379367351532,
      "learning_rate": 0.00025814936866763515,
      "loss": 2.1672,
      "step": 125466
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.0014662742614746,
      "learning_rate": 0.00025814531810779286,
      "loss": 2.3086,
      "step": 125467
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.0212249755859375,
      "learning_rate": 0.0002581412675557321,
      "loss": 2.1442,
      "step": 125468
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.1729745864868164,
      "learning_rate": 0.0002581372170114537,
      "loss": 2.1359,
      "step": 125469
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.0682538747787476,
      "learning_rate": 0.0002581331664749584,
      "loss": 2.1844,
      "step": 125470
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.9797093272209167,
      "learning_rate": 0.0002581291159462469,
      "loss": 2.1641,
      "step": 125471
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.9915066361427307,
      "learning_rate": 0.00025812506542532005,
      "loss": 2.3258,
      "step": 125472
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.042630910873413,
      "learning_rate": 0.00025812101491217847,
      "loss": 2.1796,
      "step": 125473
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.9564086198806763,
      "learning_rate": 0.0002581169644068231,
      "loss": 2.0883,
      "step": 125474
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.240158200263977,
      "learning_rate": 0.00025811291390925444,
      "loss": 2.1878,
      "step": 125475
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.9994584321975708,
      "learning_rate": 0.00025810886341947345,
      "loss": 2.3659,
      "step": 125476
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.9671877026557922,
      "learning_rate": 0.0002581048129374808,
      "loss": 1.9878,
      "step": 125477
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.947756290435791,
      "learning_rate": 0.00025810076246327723,
      "loss": 2.042,
      "step": 125478
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.0247529745101929,
      "learning_rate": 0.0002580967119968635,
      "loss": 2.2877,
      "step": 125479
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.0787924528121948,
      "learning_rate": 0.00025809266153824044,
      "loss": 2.1131,
      "step": 125480
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.9433421492576599,
      "learning_rate": 0.0002580886110874087,
      "loss": 2.3757,
      "step": 125481
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.0577349662780762,
      "learning_rate": 0.0002580845606443692,
      "loss": 2.1041,
      "step": 125482
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.9414821863174438,
      "learning_rate": 0.0002580805102091224,
      "loss": 1.9949,
      "step": 125483
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.9495725631713867,
      "learning_rate": 0.0002580764597816694,
      "loss": 2.0692,
      "step": 125484
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.0518888235092163,
      "learning_rate": 0.00025807240936201063,
      "loss": 2.1025,
      "step": 125485
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.0941599607467651,
      "learning_rate": 0.0002580683589501471,
      "loss": 1.9895,
      "step": 125486
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.0087846517562866,
      "learning_rate": 0.0002580643085460793,
      "loss": 1.9228,
      "step": 125487
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.0087401866912842,
      "learning_rate": 0.00025806025814980823,
      "loss": 2.1803,
      "step": 125488
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.9130294919013977,
      "learning_rate": 0.00025805620776133455,
      "loss": 2.0964,
      "step": 125489
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.2296606302261353,
      "learning_rate": 0.0002580521573806591,
      "loss": 2.1785,
      "step": 125490
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.9436391592025757,
      "learning_rate": 0.00025804810700778244,
      "loss": 1.9751,
      "step": 125491
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.9808512926101685,
      "learning_rate": 0.0002580440566427054,
      "loss": 1.9824,
      "step": 125492
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.0866509675979614,
      "learning_rate": 0.00025804000628542885,
      "loss": 2.0998,
      "step": 125493
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.9925652146339417,
      "learning_rate": 0.0002580359559359534,
      "loss": 2.0759,
      "step": 125494
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.8707575798034668,
      "learning_rate": 0.00025803190559427983,
      "loss": 1.9852,
      "step": 125495
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.0828744173049927,
      "learning_rate": 0.00025802785526040907,
      "loss": 2.0753,
      "step": 125496
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.9494670033454895,
      "learning_rate": 0.0002580238049343416,
      "loss": 2.2083,
      "step": 125497
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.9606643319129944,
      "learning_rate": 0.0002580197546160783,
      "loss": 2.0264,
      "step": 125498
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.9623644351959229,
      "learning_rate": 0.0002580157043056199,
      "loss": 2.2327,
      "step": 125499
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.9743397831916809,
      "learning_rate": 0.00025801165400296717,
      "loss": 1.9339,
      "step": 125500
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.9953467845916748,
      "learning_rate": 0.0002580076037081209,
      "loss": 2.0688,
      "step": 125501
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.0545117855072021,
      "learning_rate": 0.00025800355342108194,
      "loss": 2.0403,
      "step": 125502
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.9420473575592041,
      "learning_rate": 0.0002579995031418508,
      "loss": 2.1517,
      "step": 125503
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.0221669673919678,
      "learning_rate": 0.0002579954528704283,
      "loss": 2.1481,
      "step": 125504
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.1973637342453003,
      "learning_rate": 0.00025799140260681524,
      "loss": 2.0063,
      "step": 125505
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.940436601638794,
      "learning_rate": 0.00025798735235101235,
      "loss": 2.0776,
      "step": 125506
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.9145656228065491,
      "learning_rate": 0.0002579833021030205,
      "loss": 2.2185,
      "step": 125507
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.971863329410553,
      "learning_rate": 0.0002579792518628404,
      "loss": 2.0592,
      "step": 125508
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.9959573745727539,
      "learning_rate": 0.0002579752016304726,
      "loss": 2.2885,
      "step": 125509
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.0152183771133423,
      "learning_rate": 0.00025797115140591804,
      "loss": 2.1416,
      "step": 125510
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.020772933959961,
      "learning_rate": 0.00025796710118917743,
      "loss": 2.079,
      "step": 125511
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.1634769439697266,
      "learning_rate": 0.00025796305098025154,
      "loss": 2.0342,
      "step": 125512
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.9417394995689392,
      "learning_rate": 0.0002579590007791411,
      "loss": 1.9964,
      "step": 125513
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.0441738367080688,
      "learning_rate": 0.00025795495058584687,
      "loss": 2.1261,
      "step": 125514
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.0429909229278564,
      "learning_rate": 0.0002579509004003697,
      "loss": 1.9839,
      "step": 125515
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.059340476989746,
      "learning_rate": 0.0002579468502227102,
      "loss": 2.1463,
      "step": 125516
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.1348780393600464,
      "learning_rate": 0.00025794280005286916,
      "loss": 2.0808,
      "step": 125517
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.2551535367965698,
      "learning_rate": 0.00025793874989084737,
      "loss": 2.3895,
      "step": 125518
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.1372551918029785,
      "learning_rate": 0.0002579346997366455,
      "loss": 2.2262,
      "step": 125519
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.9395141005516052,
      "learning_rate": 0.0002579306495902644,
      "loss": 2.113,
      "step": 125520
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.06356680393219,
      "learning_rate": 0.00025792659945170486,
      "loss": 2.0824,
      "step": 125521
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.9345435500144958,
      "learning_rate": 0.00025792254932096746,
      "loss": 2.2048,
      "step": 125522
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.0488207340240479,
      "learning_rate": 0.0002579184991980531,
      "loss": 2.0122,
      "step": 125523
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.9608668684959412,
      "learning_rate": 0.0002579144490829625,
      "loss": 2.3902,
      "step": 125524
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.03977370262146,
      "learning_rate": 0.0002579103989756963,
      "loss": 2.3743,
      "step": 125525
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.2025012969970703,
      "learning_rate": 0.0002579063488762554,
      "loss": 2.0702,
      "step": 125526
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.1933594942092896,
      "learning_rate": 0.00025790229878464065,
      "loss": 2.0606,
      "step": 125527
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.9585586786270142,
      "learning_rate": 0.0002578982487008525,
      "loss": 2.2029,
      "step": 125528
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.1089205741882324,
      "learning_rate": 0.00025789419862489186,
      "loss": 2.2255,
      "step": 125529
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.0040851831436157,
      "learning_rate": 0.0002578901485567595,
      "loss": 1.9606,
      "step": 125530
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.0204139947891235,
      "learning_rate": 0.00025788609849645614,
      "loss": 2.2169,
      "step": 125531
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.0774288177490234,
      "learning_rate": 0.00025788204844398257,
      "loss": 2.2976,
      "step": 125532
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.008800745010376,
      "learning_rate": 0.00025787799839933965,
      "loss": 2.0384,
      "step": 125533
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.0600941181182861,
      "learning_rate": 0.0002578739483625279,
      "loss": 1.9022,
      "step": 125534
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.9603724479675293,
      "learning_rate": 0.0002578698983335481,
      "loss": 2.0995,
      "step": 125535
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.894806981086731,
      "learning_rate": 0.0002578658483124011,
      "loss": 2.228,
      "step": 125536
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.9574642181396484,
      "learning_rate": 0.00025786179829908766,
      "loss": 2.0737,
      "step": 125537
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.041798710823059,
      "learning_rate": 0.0002578577482936085,
      "loss": 2.1704,
      "step": 125538
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.0011000633239746,
      "learning_rate": 0.0002578536982959645,
      "loss": 2.1848,
      "step": 125539
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.9857860207557678,
      "learning_rate": 0.0002578496483061562,
      "loss": 2.0484,
      "step": 125540
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.9369393587112427,
      "learning_rate": 0.00025784559832418445,
      "loss": 2.0244,
      "step": 125541
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.9673075675964355,
      "learning_rate": 0.00025784154835004996,
      "loss": 2.1018,
      "step": 125542
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.9604335427284241,
      "learning_rate": 0.00025783749838375356,
      "loss": 1.9878,
      "step": 125543
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.095771312713623,
      "learning_rate": 0.00025783344842529593,
      "loss": 2.0815,
      "step": 125544
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.0946789979934692,
      "learning_rate": 0.000257829398474678,
      "loss": 2.0451,
      "step": 125545
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.001575231552124,
      "learning_rate": 0.0002578253485319002,
      "loss": 2.3226,
      "step": 125546
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.9854947924613953,
      "learning_rate": 0.00025782129859696354,
      "loss": 2.2447,
      "step": 125547
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.0155879259109497,
      "learning_rate": 0.0002578172486698687,
      "loss": 2.1094,
      "step": 125548
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.9974955320358276,
      "learning_rate": 0.00025781319875061635,
      "loss": 2.0191,
      "step": 125549
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.9368007183074951,
      "learning_rate": 0.0002578091488392074,
      "loss": 1.9894,
      "step": 125550
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.0008755922317505,
      "learning_rate": 0.00025780509893564256,
      "loss": 2.0922,
      "step": 125551
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.0160468816757202,
      "learning_rate": 0.00025780104903992256,
      "loss": 1.9763,
      "step": 125552
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.1275840997695923,
      "learning_rate": 0.0002577969991520481,
      "loss": 2.0597,
      "step": 125553
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.9658894538879395,
      "learning_rate": 0.00025779294927201995,
      "loss": 2.1142,
      "step": 125554
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.9332056045532227,
      "learning_rate": 0.00025778889939983884,
      "loss": 1.946,
      "step": 125555
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.2345845699310303,
      "learning_rate": 0.00025778484953550563,
      "loss": 2.0404,
      "step": 125556
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.1890130043029785,
      "learning_rate": 0.00025778079967902107,
      "loss": 2.2172,
      "step": 125557
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.2056702375411987,
      "learning_rate": 0.0002577767498303858,
      "loss": 2.2097,
      "step": 125558
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.9428224563598633,
      "learning_rate": 0.00025777269998960065,
      "loss": 2.091,
      "step": 125559
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.281050205230713,
      "learning_rate": 0.0002577686501566663,
      "loss": 2.2797,
      "step": 125560
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.9996529817581177,
      "learning_rate": 0.00025776460033158355,
      "loss": 2.2146,
      "step": 125561
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.1425224542617798,
      "learning_rate": 0.0002577605505143532,
      "loss": 2.0983,
      "step": 125562
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.9544326663017273,
      "learning_rate": 0.000257756500704976,
      "loss": 2.3024,
      "step": 125563
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.9750199317932129,
      "learning_rate": 0.00025775245090345263,
      "loss": 2.0143,
      "step": 125564
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.9767283201217651,
      "learning_rate": 0.0002577484011097838,
      "loss": 2.2626,
      "step": 125565
    },
    {
      "epoch": 1.63,
      "grad_norm": 1.129181981086731,
      "learning_rate": 0.0002577443513239705,
      "loss": 2.1861,
      "step": 125566
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.9554760456085205,
      "learning_rate": 0.0002577403015460132,
      "loss": 1.9016,
      "step": 125567
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.9519535899162292,
      "learning_rate": 0.0002577362517759128,
      "loss": 1.9203,
      "step": 125568
    },
    {
      "epoch": 1.63,
      "grad_norm": 0.9327648878097534,
      "learning_rate": 0.00025773220201367005,
      "loss": 2.012,
      "step": 125569
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.1196922063827515,
      "learning_rate": 0.0002577281522592856,
      "loss": 1.982,
      "step": 125570
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.1066820621490479,
      "learning_rate": 0.00025772410251276035,
      "loss": 2.2024,
      "step": 125571
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.0425959825515747,
      "learning_rate": 0.000257720052774095,
      "loss": 2.3607,
      "step": 125572
    },
    {
      "epoch": 1.64,
      "grad_norm": 0.8872422575950623,
      "learning_rate": 0.00025771600304329024,
      "loss": 1.9804,
      "step": 125573
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.1038092374801636,
      "learning_rate": 0.00025771195332034697,
      "loss": 1.891,
      "step": 125574
    },
    {
      "epoch": 1.64,
      "grad_norm": 0.9434775710105896,
      "learning_rate": 0.00025770790360526584,
      "loss": 2.0099,
      "step": 125575
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.0166395902633667,
      "learning_rate": 0.0002577038538980475,
      "loss": 1.9807,
      "step": 125576
    },
    {
      "epoch": 1.64,
      "grad_norm": 0.9565053582191467,
      "learning_rate": 0.0002576998041986929,
      "loss": 2.1242,
      "step": 125577
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.117099642753601,
      "learning_rate": 0.0002576957545072026,
      "loss": 2.4306,
      "step": 125578
    },
    {
      "epoch": 1.64,
      "grad_norm": 0.9461662173271179,
      "learning_rate": 0.00025769170482357754,
      "loss": 1.8016,
      "step": 125579
    },
    {
      "epoch": 1.64,
      "grad_norm": 0.986014723777771,
      "learning_rate": 0.00025768765514781836,
      "loss": 1.9864,
      "step": 125580
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.2028696537017822,
      "learning_rate": 0.00025768360547992594,
      "loss": 2.2036,
      "step": 125581
    },
    {
      "epoch": 1.64,
      "grad_norm": 0.9393407106399536,
      "learning_rate": 0.00025767955581990087,
      "loss": 2.1925,
      "step": 125582
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.0638513565063477,
      "learning_rate": 0.00025767550616774396,
      "loss": 2.0534,
      "step": 125583
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.3973112106323242,
      "learning_rate": 0.000257671456523456,
      "loss": 2.084,
      "step": 125584
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.011749267578125,
      "learning_rate": 0.0002576674068870376,
      "loss": 2.2561,
      "step": 125585
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.0479058027267456,
      "learning_rate": 0.0002576633572584897,
      "loss": 2.1827,
      "step": 125586
    },
    {
      "epoch": 1.64,
      "grad_norm": 0.9739044904708862,
      "learning_rate": 0.000257659307637813,
      "loss": 1.9973,
      "step": 125587
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.3209941387176514,
      "learning_rate": 0.00025765525802500834,
      "loss": 2.0264,
      "step": 125588
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.096524715423584,
      "learning_rate": 0.00025765120842007625,
      "loss": 2.1541,
      "step": 125589
    },
    {
      "epoch": 1.64,
      "grad_norm": 0.8765642046928406,
      "learning_rate": 0.00025764715882301757,
      "loss": 1.9477,
      "step": 125590
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.0933138132095337,
      "learning_rate": 0.00025764310923383313,
      "loss": 2.1545,
      "step": 125591
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.0825663805007935,
      "learning_rate": 0.0002576390596525236,
      "loss": 1.9407,
      "step": 125592
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.084198236465454,
      "learning_rate": 0.0002576350100790897,
      "loss": 2.0468,
      "step": 125593
    },
    {
      "epoch": 1.64,
      "grad_norm": 0.9498018026351929,
      "learning_rate": 0.0002576309605135325,
      "loss": 2.2267,
      "step": 125594
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.1191669702529907,
      "learning_rate": 0.0002576269109558523,
      "loss": 2.0484,
      "step": 125595
    },
    {
      "epoch": 1.64,
      "grad_norm": 0.9403160810470581,
      "learning_rate": 0.0002576228614060501,
      "loss": 2.1947,
      "step": 125596
    },
    {
      "epoch": 1.64,
      "grad_norm": 0.9617764353752136,
      "learning_rate": 0.00025761881186412664,
      "loss": 2.1017,
      "step": 125597
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.0912814140319824,
      "learning_rate": 0.0002576147623300826,
      "loss": 2.178,
      "step": 125598
    },
    {
      "epoch": 1.64,
      "grad_norm": 0.9934394955635071,
      "learning_rate": 0.0002576107128039188,
      "loss": 2.1208,
      "step": 125599
    },
    {
      "epoch": 1.64,
      "grad_norm": 0.9088764190673828,
      "learning_rate": 0.00025760666328563606,
      "loss": 2.3276,
      "step": 125600
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.0159413814544678,
      "learning_rate": 0.0002576026137752349,
      "loss": 2.2052,
      "step": 125601
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.081189513206482,
      "learning_rate": 0.0002575985642727163,
      "loss": 2.0834,
      "step": 125602
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.0725796222686768,
      "learning_rate": 0.0002575945147780808,
      "loss": 1.9949,
      "step": 125603
    },
    {
      "epoch": 1.64,
      "grad_norm": 0.9737394452095032,
      "learning_rate": 0.00025759046529132937,
      "loss": 2.0762,
      "step": 125604
    },
    {
      "epoch": 1.64,
      "grad_norm": 0.9307376742362976,
      "learning_rate": 0.00025758641581246266,
      "loss": 2.195,
      "step": 125605
    },
    {
      "epoch": 1.64,
      "grad_norm": 0.9075160622596741,
      "learning_rate": 0.0002575823663414815,
      "loss": 2.0553,
      "step": 125606
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.0117733478546143,
      "learning_rate": 0.00025757831687838655,
      "loss": 2.2397,
      "step": 125607
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.133981704711914,
      "learning_rate": 0.00025757426742317854,
      "loss": 2.1561,
      "step": 125608
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.0429118871688843,
      "learning_rate": 0.00025757021797585827,
      "loss": 2.0529,
      "step": 125609
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.0372414588928223,
      "learning_rate": 0.00025756616853642655,
      "loss": 2.4456,
      "step": 125610
    },
    {
      "epoch": 1.64,
      "grad_norm": 0.9805473685264587,
      "learning_rate": 0.000257562119104884,
      "loss": 1.9824,
      "step": 125611
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.013729453086853,
      "learning_rate": 0.0002575580696812316,
      "loss": 2.1329,
      "step": 125612
    },
    {
      "epoch": 1.64,
      "grad_norm": 0.9064816236495972,
      "learning_rate": 0.00025755402026546987,
      "loss": 2.1459,
      "step": 125613
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.1482882499694824,
      "learning_rate": 0.00025754997085759965,
      "loss": 2.3879,
      "step": 125614
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.0074354410171509,
      "learning_rate": 0.00025754592145762166,
      "loss": 1.9932,
      "step": 125615
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.0461519956588745,
      "learning_rate": 0.0002575418720655367,
      "loss": 2.0861,
      "step": 125616
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.0069504976272583,
      "learning_rate": 0.00025753782268134553,
      "loss": 2.121,
      "step": 125617
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.0943055152893066,
      "learning_rate": 0.00025753377330504896,
      "loss": 2.1238,
      "step": 125618
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.0158592462539673,
      "learning_rate": 0.00025752972393664755,
      "loss": 1.9924,
      "step": 125619
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.0024033784866333,
      "learning_rate": 0.00025752567457614216,
      "loss": 2.1858,
      "step": 125620
    },
    {
      "epoch": 1.64,
      "grad_norm": 0.8663884401321411,
      "learning_rate": 0.00025752162522353357,
      "loss": 2.1665,
      "step": 125621
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.0551228523254395,
      "learning_rate": 0.0002575175758788225,
      "loss": 2.0963,
      "step": 125622
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.0609441995620728,
      "learning_rate": 0.00025751352654200977,
      "loss": 2.0668,
      "step": 125623
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.009629487991333,
      "learning_rate": 0.00025750947721309614,
      "loss": 2.1446,
      "step": 125624
    },
    {
      "epoch": 1.64,
      "grad_norm": 0.9489302039146423,
      "learning_rate": 0.00025750542789208217,
      "loss": 2.1564,
      "step": 125625
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.0153433084487915,
      "learning_rate": 0.0002575013785789687,
      "loss": 2.0271,
      "step": 125626
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.1931157112121582,
      "learning_rate": 0.0002574973292737566,
      "loss": 2.3327,
      "step": 125627
    },
    {
      "epoch": 1.64,
      "grad_norm": 0.9454083442687988,
      "learning_rate": 0.00025749327997644656,
      "loss": 2.0339,
      "step": 125628
    },
    {
      "epoch": 1.64,
      "grad_norm": 0.9309760928153992,
      "learning_rate": 0.0002574892306870393,
      "loss": 1.9774,
      "step": 125629
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.1283152103424072,
      "learning_rate": 0.0002574851814055357,
      "loss": 2.2118,
      "step": 125630
    },
    {
      "epoch": 1.64,
      "grad_norm": 0.9682580828666687,
      "learning_rate": 0.0002574811321319363,
      "loss": 2.3803,
      "step": 125631
    },
    {
      "epoch": 1.64,
      "grad_norm": 0.9051513671875,
      "learning_rate": 0.00025747708286624196,
      "loss": 2.1362,
      "step": 125632
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.0568991899490356,
      "learning_rate": 0.00025747303360845346,
      "loss": 2.3927,
      "step": 125633
    },
    {
      "epoch": 1.64,
      "grad_norm": 0.9562545418739319,
      "learning_rate": 0.0002574689843585715,
      "loss": 2.2853,
      "step": 125634
    },
    {
      "epoch": 1.64,
      "grad_norm": 0.9297046065330505,
      "learning_rate": 0.00025746493511659684,
      "loss": 1.9834,
      "step": 125635
    },
    {
      "epoch": 1.64,
      "grad_norm": 0.9252467751502991,
      "learning_rate": 0.00025746088588253036,
      "loss": 1.8862,
      "step": 125636
    },
    {
      "epoch": 1.64,
      "grad_norm": 0.9523792862892151,
      "learning_rate": 0.00025745683665637264,
      "loss": 2.1368,
      "step": 125637
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.0172117948532104,
      "learning_rate": 0.00025745278743812444,
      "loss": 2.106,
      "step": 125638
    },
    {
      "epoch": 1.64,
      "grad_norm": 0.9680700898170471,
      "learning_rate": 0.00025744873822778665,
      "loss": 2.0322,
      "step": 125639
    },
    {
      "epoch": 1.64,
      "grad_norm": 0.924674928188324,
      "learning_rate": 0.00025744468902535984,
      "loss": 1.9829,
      "step": 125640
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.0430805683135986,
      "learning_rate": 0.0002574406398308449,
      "loss": 2.1615,
      "step": 125641
    },
    {
      "epoch": 1.64,
      "grad_norm": 0.9414128065109253,
      "learning_rate": 0.00025743659064424266,
      "loss": 1.9761,
      "step": 125642
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.0171860456466675,
      "learning_rate": 0.00025743254146555364,
      "loss": 2.0674,
      "step": 125643
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.0698716640472412,
      "learning_rate": 0.0002574284922947788,
      "loss": 2.2621,
      "step": 125644
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.0916552543640137,
      "learning_rate": 0.00025742444313191877,
      "loss": 2.0952,
      "step": 125645
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.0311548709869385,
      "learning_rate": 0.0002574203939769743,
      "loss": 2.1051,
      "step": 125646
    },
    {
      "epoch": 1.64,
      "grad_norm": 0.9034398198127747,
      "learning_rate": 0.00025741634482994623,
      "loss": 2.0152,
      "step": 125647
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.044206976890564,
      "learning_rate": 0.0002574122956908352,
      "loss": 2.0628,
      "step": 125648
    },
    {
      "epoch": 1.64,
      "grad_norm": 0.9968370199203491,
      "learning_rate": 0.00025740824655964215,
      "loss": 2.0593,
      "step": 125649
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.0575093030929565,
      "learning_rate": 0.00025740419743636763,
      "loss": 2.0784,
      "step": 125650
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.058030128479004,
      "learning_rate": 0.0002574001483210125,
      "loss": 1.992,
      "step": 125651
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.0214390754699707,
      "learning_rate": 0.00025739609921357747,
      "loss": 2.0598,
      "step": 125652
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.382049322128296,
      "learning_rate": 0.0002573920501140633,
      "loss": 2.3554,
      "step": 125653
    },
    {
      "epoch": 1.64,
      "grad_norm": 0.9800295233726501,
      "learning_rate": 0.0002573880010224707,
      "loss": 2.2697,
      "step": 125654
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.0201137065887451,
      "learning_rate": 0.0002573839519388006,
      "loss": 2.0762,
      "step": 125655
    },
    {
      "epoch": 1.64,
      "grad_norm": 0.9362733364105225,
      "learning_rate": 0.00025737990286305354,
      "loss": 2.0355,
      "step": 125656
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.0424447059631348,
      "learning_rate": 0.0002573758537952304,
      "loss": 2.3226,
      "step": 125657
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.0074278116226196,
      "learning_rate": 0.0002573718047353319,
      "loss": 2.2561,
      "step": 125658
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.1035552024841309,
      "learning_rate": 0.00025736775568335876,
      "loss": 1.9415,
      "step": 125659
    },
    {
      "epoch": 1.64,
      "grad_norm": 0.9728197455406189,
      "learning_rate": 0.0002573637066393117,
      "loss": 2.2565,
      "step": 125660
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.0004239082336426,
      "learning_rate": 0.00025735965760319165,
      "loss": 2.1259,
      "step": 125661
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.0503560304641724,
      "learning_rate": 0.0002573556085749992,
      "loss": 2.1745,
      "step": 125662
    },
    {
      "epoch": 1.64,
      "grad_norm": 0.9612938165664673,
      "learning_rate": 0.0002573515595547351,
      "loss": 1.9074,
      "step": 125663
    },
    {
      "epoch": 1.64,
      "grad_norm": 0.9886201024055481,
      "learning_rate": 0.0002573475105424002,
      "loss": 2.0687,
      "step": 125664
    },
    {
      "epoch": 1.64,
      "grad_norm": 0.9973268508911133,
      "learning_rate": 0.00025734346153799515,
      "loss": 2.134,
      "step": 125665
    },
    {
      "epoch": 1.64,
      "grad_norm": 0.9450213313102722,
      "learning_rate": 0.00025733941254152084,
      "loss": 2.0488,
      "step": 125666
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.0027981996536255,
      "learning_rate": 0.00025733536355297795,
      "loss": 2.1384,
      "step": 125667
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.2413177490234375,
      "learning_rate": 0.00025733131457236714,
      "loss": 2.1176,
      "step": 125668
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.0514758825302124,
      "learning_rate": 0.0002573272655996893,
      "loss": 2.1011,
      "step": 125669
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.0099647045135498,
      "learning_rate": 0.00025732321663494507,
      "loss": 2.1511,
      "step": 125670
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.1662899255752563,
      "learning_rate": 0.00025731916767813526,
      "loss": 2.2012,
      "step": 125671
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.1684963703155518,
      "learning_rate": 0.0002573151187292606,
      "loss": 2.2323,
      "step": 125672
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.0595282316207886,
      "learning_rate": 0.00025731106978832207,
      "loss": 2.0378,
      "step": 125673
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.1318073272705078,
      "learning_rate": 0.00025730702085532,
      "loss": 2.0239,
      "step": 125674
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.0838475227355957,
      "learning_rate": 0.0002573029719302554,
      "loss": 2.3479,
      "step": 125675
    },
    {
      "epoch": 1.64,
      "grad_norm": 0.9032445549964905,
      "learning_rate": 0.000257298923013129,
      "loss": 2.0824,
      "step": 125676
    },
    {
      "epoch": 1.64,
      "grad_norm": 0.8802316188812256,
      "learning_rate": 0.00025729487410394155,
      "loss": 1.9125,
      "step": 125677
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.049729347229004,
      "learning_rate": 0.00025729082520269375,
      "loss": 1.9806,
      "step": 125678
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.0169278383255005,
      "learning_rate": 0.00025728677630938653,
      "loss": 1.9569,
      "step": 125679
    },
    {
      "epoch": 1.64,
      "grad_norm": 0.9706427454948425,
      "learning_rate": 0.0002572827274240204,
      "loss": 2.052,
      "step": 125680
    },
    {
      "epoch": 1.64,
      "grad_norm": 0.8510472774505615,
      "learning_rate": 0.0002572786785465962,
      "loss": 2.0579,
      "step": 125681
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.071831226348877,
      "learning_rate": 0.00025727462967711464,
      "loss": 2.17,
      "step": 125682
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.1419380903244019,
      "learning_rate": 0.0002572705808155766,
      "loss": 2.0216,
      "step": 125683
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.0572304725646973,
      "learning_rate": 0.00025726653196198274,
      "loss": 2.2787,
      "step": 125684
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.1948788166046143,
      "learning_rate": 0.000257262483116334,
      "loss": 2.2338,
      "step": 125685
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.007723093032837,
      "learning_rate": 0.0002572584342786308,
      "loss": 2.0909,
      "step": 125686
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.0820097923278809,
      "learning_rate": 0.00025725438544887415,
      "loss": 2.0726,
      "step": 125687
    },
    {
      "epoch": 1.64,
      "grad_norm": 0.9815785884857178,
      "learning_rate": 0.00025725033662706465,
      "loss": 2.1266,
      "step": 125688
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.0813547372817993,
      "learning_rate": 0.0002572462878132031,
      "loss": 2.0604,
      "step": 125689
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.0180140733718872,
      "learning_rate": 0.0002572422390072903,
      "loss": 1.9939,
      "step": 125690
    },
    {
      "epoch": 1.64,
      "grad_norm": 0.9789998531341553,
      "learning_rate": 0.0002572381902093271,
      "loss": 2.197,
      "step": 125691
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.04764723777771,
      "learning_rate": 0.000257234141419314,
      "loss": 1.9259,
      "step": 125692
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.019675374031067,
      "learning_rate": 0.0002572300926372519,
      "loss": 2.2807,
      "step": 125693
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.0272575616836548,
      "learning_rate": 0.00025722604386314147,
      "loss": 2.0274,
      "step": 125694
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.0152804851531982,
      "learning_rate": 0.00025722199509698357,
      "loss": 1.8956,
      "step": 125695
    },
    {
      "epoch": 1.64,
      "grad_norm": 0.9409446716308594,
      "learning_rate": 0.00025721794633877895,
      "loss": 2.0699,
      "step": 125696
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.0334951877593994,
      "learning_rate": 0.00025721389758852836,
      "loss": 2.1502,
      "step": 125697
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.0589776039123535,
      "learning_rate": 0.00025720984884623245,
      "loss": 2.0985,
      "step": 125698
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.0547833442687988,
      "learning_rate": 0.00025720580011189204,
      "loss": 2.1416,
      "step": 125699
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.0064395666122437,
      "learning_rate": 0.00025720175138550783,
      "loss": 2.3847,
      "step": 125700
    },
    {
      "epoch": 1.64,
      "grad_norm": 0.9042452573776245,
      "learning_rate": 0.0002571977026670807,
      "loss": 2.1478,
      "step": 125701
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.0735327005386353,
      "learning_rate": 0.00025719365395661126,
      "loss": 1.9594,
      "step": 125702
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.0007381439208984,
      "learning_rate": 0.0002571896052541005,
      "loss": 2.2529,
      "step": 125703
    },
    {
      "epoch": 1.64,
      "grad_norm": 0.9244572520256042,
      "learning_rate": 0.0002571855565595488,
      "loss": 2.2473,
      "step": 125704
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.1517010927200317,
      "learning_rate": 0.00025718150787295717,
      "loss": 2.2255,
      "step": 125705
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.001526117324829,
      "learning_rate": 0.0002571774591943263,
      "loss": 2.0901,
      "step": 125706
    },
    {
      "epoch": 1.64,
      "grad_norm": 0.980006754398346,
      "learning_rate": 0.0002571734105236569,
      "loss": 1.9759,
      "step": 125707
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.024013638496399,
      "learning_rate": 0.0002571693618609498,
      "loss": 2.1053,
      "step": 125708
    },
    {
      "epoch": 1.64,
      "grad_norm": 0.9241562485694885,
      "learning_rate": 0.0002571653132062059,
      "loss": 2.139,
      "step": 125709
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.1317600011825562,
      "learning_rate": 0.00025716126455942564,
      "loss": 2.1625,
      "step": 125710
    },
    {
      "epoch": 1.64,
      "grad_norm": 0.9108543395996094,
      "learning_rate": 0.0002571572159206099,
      "loss": 1.9072,
      "step": 125711
    },
    {
      "epoch": 1.64,
      "grad_norm": 0.9101061820983887,
      "learning_rate": 0.0002571531672897594,
      "loss": 2.1005,
      "step": 125712
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.08061945438385,
      "learning_rate": 0.00025714911866687497,
      "loss": 2.2658,
      "step": 125713
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.0121650695800781,
      "learning_rate": 0.0002571450700519573,
      "loss": 2.0995,
      "step": 125714
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.1045035123825073,
      "learning_rate": 0.00025714102144500724,
      "loss": 2.16,
      "step": 125715
    },
    {
      "epoch": 1.64,
      "grad_norm": 0.977729856967926,
      "learning_rate": 0.0002571369728460255,
      "loss": 2.2474,
      "step": 125716
    },
    {
      "epoch": 1.64,
      "grad_norm": 0.9881841540336609,
      "learning_rate": 0.0002571329242550127,
      "loss": 2.0385,
      "step": 125717
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.3681193590164185,
      "learning_rate": 0.00025712887567196975,
      "loss": 2.2445,
      "step": 125718
    },
    {
      "epoch": 1.64,
      "grad_norm": 0.9843481779098511,
      "learning_rate": 0.0002571248270968974,
      "loss": 1.9886,
      "step": 125719
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.0081740617752075,
      "learning_rate": 0.00025712077852979624,
      "loss": 2.2029,
      "step": 125720
    },
    {
      "epoch": 1.64,
      "grad_norm": 0.9434864521026611,
      "learning_rate": 0.00025711672997066713,
      "loss": 2.2361,
      "step": 125721
    },
    {
      "epoch": 1.64,
      "grad_norm": 0.9640258550643921,
      "learning_rate": 0.0002571126814195111,
      "loss": 2.2715,
      "step": 125722
    },
    {
      "epoch": 1.64,
      "grad_norm": 0.9935179352760315,
      "learning_rate": 0.00025710863287632836,
      "loss": 2.1125,
      "step": 125723
    },
    {
      "epoch": 1.64,
      "grad_norm": 0.9271507859230042,
      "learning_rate": 0.00025710458434112,
      "loss": 2.1037,
      "step": 125724
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.001277208328247,
      "learning_rate": 0.00025710053581388666,
      "loss": 2.0796,
      "step": 125725
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.002095103263855,
      "learning_rate": 0.0002570964872946292,
      "loss": 2.1278,
      "step": 125726
    },
    {
      "epoch": 1.64,
      "grad_norm": 0.9728073477745056,
      "learning_rate": 0.00025709243878334825,
      "loss": 2.0879,
      "step": 125727
    },
    {
      "epoch": 1.64,
      "grad_norm": 0.9966493248939514,
      "learning_rate": 0.0002570883902800448,
      "loss": 2.0663,
      "step": 125728
    },
    {
      "epoch": 1.64,
      "grad_norm": 0.9263713955879211,
      "learning_rate": 0.0002570843417847193,
      "loss": 1.9429,
      "step": 125729
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.01390540599823,
      "learning_rate": 0.00025708029329737263,
      "loss": 1.895,
      "step": 125730
    },
    {
      "epoch": 1.64,
      "grad_norm": 0.9753784537315369,
      "learning_rate": 0.00025707624481800557,
      "loss": 2.0381,
      "step": 125731
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.091741681098938,
      "learning_rate": 0.00025707219634661877,
      "loss": 1.9606,
      "step": 125732
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.0480281114578247,
      "learning_rate": 0.0002570681478832131,
      "loss": 2.1101,
      "step": 125733
    },
    {
      "epoch": 1.64,
      "grad_norm": 0.9237304925918579,
      "learning_rate": 0.00025706409942778934,
      "loss": 2.2675,
      "step": 125734
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.0250043869018555,
      "learning_rate": 0.0002570600509803481,
      "loss": 2.0488,
      "step": 125735
    },
    {
      "epoch": 1.64,
      "grad_norm": 0.9273784756660461,
      "learning_rate": 0.00025705600254089027,
      "loss": 2.043,
      "step": 125736
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.0753518342971802,
      "learning_rate": 0.0002570519541094165,
      "loss": 2.3059,
      "step": 125737
    },
    {
      "epoch": 1.64,
      "grad_norm": 0.9982229471206665,
      "learning_rate": 0.00025704790568592756,
      "loss": 2.2556,
      "step": 125738
    },
    {
      "epoch": 1.64,
      "grad_norm": 0.9976750016212463,
      "learning_rate": 0.00025704385727042423,
      "loss": 2.0504,
      "step": 125739
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.037943720817566,
      "learning_rate": 0.00025703980886290735,
      "loss": 2.2491,
      "step": 125740
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.0661879777908325,
      "learning_rate": 0.00025703576046337746,
      "loss": 1.8799,
      "step": 125741
    },
    {
      "epoch": 1.64,
      "grad_norm": 0.9671550393104553,
      "learning_rate": 0.0002570317120718355,
      "loss": 2.3223,
      "step": 125742
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.0892423391342163,
      "learning_rate": 0.0002570276636882822,
      "loss": 2.1945,
      "step": 125743
    },
    {
      "epoch": 1.64,
      "grad_norm": 0.878157377243042,
      "learning_rate": 0.00025702361531271814,
      "loss": 2.0583,
      "step": 125744
    },
    {
      "epoch": 1.64,
      "grad_norm": 0.9945936799049377,
      "learning_rate": 0.00025701956694514425,
      "loss": 2.167,
      "step": 125745
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.1894476413726807,
      "learning_rate": 0.0002570155185855613,
      "loss": 2.1247,
      "step": 125746
    },
    {
      "epoch": 1.64,
      "grad_norm": 0.854211151599884,
      "learning_rate": 0.0002570114702339699,
      "loss": 1.8094,
      "step": 125747
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.0213680267333984,
      "learning_rate": 0.00025700742189037085,
      "loss": 2.3275,
      "step": 125748
    },
    {
      "epoch": 1.64,
      "grad_norm": 0.9808833599090576,
      "learning_rate": 0.000257003373554765,
      "loss": 2.2569,
      "step": 125749
    },
    {
      "epoch": 1.64,
      "grad_norm": 0.9672953486442566,
      "learning_rate": 0.00025699932522715304,
      "loss": 2.1401,
      "step": 125750
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.1326442956924438,
      "learning_rate": 0.00025699527690753573,
      "loss": 2.1912,
      "step": 125751
    },
    {
      "epoch": 1.64,
      "grad_norm": 0.9902855157852173,
      "learning_rate": 0.00025699122859591377,
      "loss": 1.9689,
      "step": 125752
    },
    {
      "epoch": 1.64,
      "grad_norm": 0.9754335284233093,
      "learning_rate": 0.000256987180292288,
      "loss": 2.0215,
      "step": 125753
    },
    {
      "epoch": 1.64,
      "grad_norm": 0.9731417298316956,
      "learning_rate": 0.00025698313199665906,
      "loss": 2.1214,
      "step": 125754
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.009666085243225,
      "learning_rate": 0.0002569790837090278,
      "loss": 1.9319,
      "step": 125755
    },
    {
      "epoch": 1.64,
      "grad_norm": 0.9821513295173645,
      "learning_rate": 0.0002569750354293949,
      "loss": 2.0349,
      "step": 125756
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.1550177335739136,
      "learning_rate": 0.0002569709871577612,
      "loss": 2.2103,
      "step": 125757
    },
    {
      "epoch": 1.64,
      "grad_norm": 0.9537741541862488,
      "learning_rate": 0.00025696693889412745,
      "loss": 1.9059,
      "step": 125758
    },
    {
      "epoch": 1.64,
      "grad_norm": 0.8937771916389465,
      "learning_rate": 0.0002569628906384943,
      "loss": 1.9984,
      "step": 125759
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.1713197231292725,
      "learning_rate": 0.0002569588423908625,
      "loss": 2.23,
      "step": 125760
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.0574874877929688,
      "learning_rate": 0.00025695479415123294,
      "loss": 2.2868,
      "step": 125761
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.0431735515594482,
      "learning_rate": 0.0002569507459196063,
      "loss": 2.2444,
      "step": 125762
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.023648738861084,
      "learning_rate": 0.00025694669769598327,
      "loss": 2.1932,
      "step": 125763
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.0686229467391968,
      "learning_rate": 0.0002569426494803648,
      "loss": 2.4137,
      "step": 125764
    },
    {
      "epoch": 1.64,
      "grad_norm": 0.9969260692596436,
      "learning_rate": 0.0002569386012727514,
      "loss": 1.9997,
      "step": 125765
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.0112062692642212,
      "learning_rate": 0.0002569345530731439,
      "loss": 2.0704,
      "step": 125766
    },
    {
      "epoch": 1.64,
      "grad_norm": 0.916110098361969,
      "learning_rate": 0.0002569305048815431,
      "loss": 2.1382,
      "step": 125767
    },
    {
      "epoch": 1.64,
      "grad_norm": 0.9689791202545166,
      "learning_rate": 0.00025692645669794973,
      "loss": 2.1186,
      "step": 125768
    },
    {
      "epoch": 1.64,
      "grad_norm": 0.9564176201820374,
      "learning_rate": 0.00025692240852236457,
      "loss": 2.067,
      "step": 125769
    },
    {
      "epoch": 1.64,
      "grad_norm": 0.8820887804031372,
      "learning_rate": 0.0002569183603547884,
      "loss": 1.9948,
      "step": 125770
    },
    {
      "epoch": 1.64,
      "grad_norm": 0.9555014967918396,
      "learning_rate": 0.00025691431219522184,
      "loss": 2.0155,
      "step": 125771
    },
    {
      "epoch": 1.64,
      "grad_norm": 0.9973827004432678,
      "learning_rate": 0.00025691026404366573,
      "loss": 1.8984,
      "step": 125772
    },
    {
      "epoch": 1.64,
      "grad_norm": 0.9303711652755737,
      "learning_rate": 0.0002569062159001208,
      "loss": 2.1973,
      "step": 125773
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.2856032848358154,
      "learning_rate": 0.00025690216776458785,
      "loss": 2.1437,
      "step": 125774
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.1328527927398682,
      "learning_rate": 0.0002568981196370676,
      "loss": 2.3244,
      "step": 125775
    },
    {
      "epoch": 1.64,
      "grad_norm": 0.9589128494262695,
      "learning_rate": 0.0002568940715175609,
      "loss": 1.9681,
      "step": 125776
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.0099067687988281,
      "learning_rate": 0.0002568900234060682,
      "loss": 2.236,
      "step": 125777
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.0832653045654297,
      "learning_rate": 0.0002568859753025906,
      "loss": 2.2287,
      "step": 125778
    },
    {
      "epoch": 1.64,
      "grad_norm": 0.9670049548149109,
      "learning_rate": 0.00025688192720712866,
      "loss": 2.2767,
      "step": 125779
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.0753810405731201,
      "learning_rate": 0.00025687787911968317,
      "loss": 2.2833,
      "step": 125780
    },
    {
      "epoch": 1.64,
      "grad_norm": 0.9975789189338684,
      "learning_rate": 0.0002568738310402549,
      "loss": 2.169,
      "step": 125781
    },
    {
      "epoch": 1.64,
      "grad_norm": 0.9788947105407715,
      "learning_rate": 0.0002568697829688446,
      "loss": 1.9362,
      "step": 125782
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.1484283208847046,
      "learning_rate": 0.00025686573490545317,
      "loss": 1.919,
      "step": 125783
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.0609928369522095,
      "learning_rate": 0.00025686168685008104,
      "loss": 2.1746,
      "step": 125784
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.021276831626892,
      "learning_rate": 0.0002568576388027292,
      "loss": 1.9517,
      "step": 125785
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.012634038925171,
      "learning_rate": 0.0002568535907633983,
      "loss": 2.1283,
      "step": 125786
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.0874441862106323,
      "learning_rate": 0.0002568495427320891,
      "loss": 2.1243,
      "step": 125787
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.757629156112671,
      "learning_rate": 0.00025684549470880247,
      "loss": 1.9355,
      "step": 125788
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.1244059801101685,
      "learning_rate": 0.00025684144669353915,
      "loss": 1.7413,
      "step": 125789
    },
    {
      "epoch": 1.64,
      "grad_norm": 0.9312057495117188,
      "learning_rate": 0.0002568373986862997,
      "loss": 2.0762,
      "step": 125790
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.0192543268203735,
      "learning_rate": 0.000256833350687085,
      "loss": 2.3705,
      "step": 125791
    },
    {
      "epoch": 1.64,
      "grad_norm": 0.9873842000961304,
      "learning_rate": 0.0002568293026958957,
      "loss": 2.1426,
      "step": 125792
    },
    {
      "epoch": 1.64,
      "grad_norm": 0.9303104877471924,
      "learning_rate": 0.0002568252547127328,
      "loss": 2.2883,
      "step": 125793
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.012650489807129,
      "learning_rate": 0.00025682120673759684,
      "loss": 2.2042,
      "step": 125794
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.0179328918457031,
      "learning_rate": 0.00025681715877048873,
      "loss": 2.1273,
      "step": 125795
    },
    {
      "epoch": 1.64,
      "grad_norm": 0.9550363421440125,
      "learning_rate": 0.00025681311081140904,
      "loss": 1.9899,
      "step": 125796
    },
    {
      "epoch": 1.64,
      "grad_norm": 0.9638373255729675,
      "learning_rate": 0.00025680906286035856,
      "loss": 2.0048,
      "step": 125797
    },
    {
      "epoch": 1.64,
      "grad_norm": 0.9479527473449707,
      "learning_rate": 0.0002568050149173381,
      "loss": 1.9625,
      "step": 125798
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.012866497039795,
      "learning_rate": 0.0002568009669823485,
      "loss": 2.5439,
      "step": 125799
    },
    {
      "epoch": 1.64,
      "grad_norm": 0.9653739929199219,
      "learning_rate": 0.00025679691905539033,
      "loss": 2.215,
      "step": 125800
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.2218835353851318,
      "learning_rate": 0.00025679287113646455,
      "loss": 2.317,
      "step": 125801
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.0507440567016602,
      "learning_rate": 0.0002567888232255717,
      "loss": 2.3731,
      "step": 125802
    },
    {
      "epoch": 1.64,
      "grad_norm": 0.9549388289451599,
      "learning_rate": 0.0002567847753227126,
      "loss": 2.2137,
      "step": 125803
    },
    {
      "epoch": 1.64,
      "grad_norm": 0.9596061706542969,
      "learning_rate": 0.000256780727427888,
      "loss": 2.0616,
      "step": 125804
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.10602605342865,
      "learning_rate": 0.0002567766795410987,
      "loss": 2.4965,
      "step": 125805
    },
    {
      "epoch": 1.64,
      "grad_norm": 0.9923573732376099,
      "learning_rate": 0.0002567726316623455,
      "loss": 2.1692,
      "step": 125806
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.101468801498413,
      "learning_rate": 0.0002567685837916291,
      "loss": 2.2715,
      "step": 125807
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.0578752756118774,
      "learning_rate": 0.0002567645359289502,
      "loss": 2.0575,
      "step": 125808
    },
    {
      "epoch": 1.64,
      "grad_norm": 0.9776834845542908,
      "learning_rate": 0.00025676048807430946,
      "loss": 2.1206,
      "step": 125809
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.0005671977996826,
      "learning_rate": 0.0002567564402277079,
      "loss": 2.0013,
      "step": 125810
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.0782051086425781,
      "learning_rate": 0.0002567523923891461,
      "loss": 2.139,
      "step": 125811
    },
    {
      "epoch": 1.64,
      "grad_norm": 0.9378533363342285,
      "learning_rate": 0.00025674834455862474,
      "loss": 1.9274,
      "step": 125812
    },
    {
      "epoch": 1.64,
      "grad_norm": 0.9679257273674011,
      "learning_rate": 0.000256744296736145,
      "loss": 1.9392,
      "step": 125813
    },
    {
      "epoch": 1.64,
      "grad_norm": 0.922385036945343,
      "learning_rate": 0.000256740248921707,
      "loss": 2.0017,
      "step": 125814
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.004873275756836,
      "learning_rate": 0.0002567362011153119,
      "loss": 2.1561,
      "step": 125815
    },
    {
      "epoch": 1.64,
      "grad_norm": 0.9594953656196594,
      "learning_rate": 0.00025673215331696033,
      "loss": 2.0861,
      "step": 125816
    },
    {
      "epoch": 1.64,
      "grad_norm": 0.9643363356590271,
      "learning_rate": 0.0002567281055266531,
      "loss": 2.1919,
      "step": 125817
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.0563592910766602,
      "learning_rate": 0.0002567240577443909,
      "loss": 2.1208,
      "step": 125818
    },
    {
      "epoch": 1.64,
      "grad_norm": 0.9279717803001404,
      "learning_rate": 0.0002567200099701746,
      "loss": 1.9952,
      "step": 125819
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.1147127151489258,
      "learning_rate": 0.00025671596220400487,
      "loss": 1.9804,
      "step": 125820
    },
    {
      "epoch": 1.64,
      "grad_norm": 0.9522362947463989,
      "learning_rate": 0.00025671191444588237,
      "loss": 2.0585,
      "step": 125821
    },
    {
      "epoch": 1.64,
      "grad_norm": 0.9158809185028076,
      "learning_rate": 0.00025670786669580796,
      "loss": 2.1459,
      "step": 125822
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.0141124725341797,
      "learning_rate": 0.00025670381895378234,
      "loss": 2.0649,
      "step": 125823
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.1360753774642944,
      "learning_rate": 0.0002566997712198064,
      "loss": 2.1798,
      "step": 125824
    },
    {
      "epoch": 1.64,
      "grad_norm": 0.9752551317214966,
      "learning_rate": 0.0002566957234938807,
      "loss": 2.1111,
      "step": 125825
    },
    {
      "epoch": 1.64,
      "grad_norm": 0.9674063920974731,
      "learning_rate": 0.0002566916757760061,
      "loss": 2.2256,
      "step": 125826
    },
    {
      "epoch": 1.64,
      "grad_norm": 0.9422556757926941,
      "learning_rate": 0.00025668762806618333,
      "loss": 2.0946,
      "step": 125827
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.0267349481582642,
      "learning_rate": 0.0002566835803644132,
      "loss": 2.372,
      "step": 125828
    },
    {
      "epoch": 1.64,
      "grad_norm": 0.9588872194290161,
      "learning_rate": 0.0002566795326706964,
      "loss": 2.0692,
      "step": 125829
    },
    {
      "epoch": 1.64,
      "grad_norm": 0.9487835764884949,
      "learning_rate": 0.0002566754849850337,
      "loss": 2.0551,
      "step": 125830
    },
    {
      "epoch": 1.64,
      "grad_norm": 0.996489942073822,
      "learning_rate": 0.0002566714373074258,
      "loss": 2.2328,
      "step": 125831
    },
    {
      "epoch": 1.64,
      "grad_norm": 0.9978175759315491,
      "learning_rate": 0.00025666738963787353,
      "loss": 2.0372,
      "step": 125832
    },
    {
      "epoch": 1.64,
      "grad_norm": 0.949677586555481,
      "learning_rate": 0.0002566633419763776,
      "loss": 2.0146,
      "step": 125833
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.0532004833221436,
      "learning_rate": 0.0002566592943229387,
      "loss": 2.1144,
      "step": 125834
    },
    {
      "epoch": 1.64,
      "grad_norm": 0.9187127947807312,
      "learning_rate": 0.0002566552466775578,
      "loss": 2.1548,
      "step": 125835
    },
    {
      "epoch": 1.64,
      "grad_norm": 0.9472827315330505,
      "learning_rate": 0.0002566511990402354,
      "loss": 2.1537,
      "step": 125836
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.0229581594467163,
      "learning_rate": 0.00025664715141097245,
      "loss": 1.944,
      "step": 125837
    },
    {
      "epoch": 1.64,
      "grad_norm": 0.9943206310272217,
      "learning_rate": 0.0002566431037897695,
      "loss": 2.2427,
      "step": 125838
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.057029128074646,
      "learning_rate": 0.0002566390561766275,
      "loss": 2.0489,
      "step": 125839
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.008726954460144,
      "learning_rate": 0.0002566350085715471,
      "loss": 2.2567,
      "step": 125840
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.1008176803588867,
      "learning_rate": 0.000256630960974529,
      "loss": 2.0771,
      "step": 125841
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.0037357807159424,
      "learning_rate": 0.00025662691338557413,
      "loss": 2.4103,
      "step": 125842
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.0474506616592407,
      "learning_rate": 0.0002566228658046831,
      "loss": 2.0638,
      "step": 125843
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.2199090719223022,
      "learning_rate": 0.0002566188182318567,
      "loss": 2.2598,
      "step": 125844
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.0679742097854614,
      "learning_rate": 0.0002566147706670957,
      "loss": 2.0905,
      "step": 125845
    },
    {
      "epoch": 1.64,
      "grad_norm": 0.9445005655288696,
      "learning_rate": 0.0002566107231104007,
      "loss": 1.9909,
      "step": 125846
    },
    {
      "epoch": 1.64,
      "grad_norm": 0.9455432295799255,
      "learning_rate": 0.0002566066755617727,
      "loss": 2.3407,
      "step": 125847
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.1939343214035034,
      "learning_rate": 0.00025660262802121227,
      "loss": 2.0344,
      "step": 125848
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.0656355619430542,
      "learning_rate": 0.0002565985804887204,
      "loss": 2.3727,
      "step": 125849
    },
    {
      "epoch": 1.64,
      "grad_norm": 0.9082841873168945,
      "learning_rate": 0.00025659453296429755,
      "loss": 2.2631,
      "step": 125850
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.0059949159622192,
      "learning_rate": 0.0002565904854479446,
      "loss": 2.0826,
      "step": 125851
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.1012853384017944,
      "learning_rate": 0.0002565864379396623,
      "loss": 2.3221,
      "step": 125852
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.0953155755996704,
      "learning_rate": 0.0002565823904394513,
      "loss": 2.2544,
      "step": 125853
    },
    {
      "epoch": 1.64,
      "grad_norm": 0.9330976009368896,
      "learning_rate": 0.0002565783429473126,
      "loss": 2.3835,
      "step": 125854
    },
    {
      "epoch": 1.64,
      "grad_norm": 0.987430989742279,
      "learning_rate": 0.0002565742954632467,
      "loss": 1.9911,
      "step": 125855
    },
    {
      "epoch": 1.64,
      "grad_norm": 0.9907817840576172,
      "learning_rate": 0.00025657024798725466,
      "loss": 2.1312,
      "step": 125856
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.0536636114120483,
      "learning_rate": 0.0002565662005193369,
      "loss": 2.2356,
      "step": 125857
    },
    {
      "epoch": 1.64,
      "grad_norm": 0.9309674501419067,
      "learning_rate": 0.0002565621530594943,
      "loss": 2.1021,
      "step": 125858
    },
    {
      "epoch": 1.64,
      "grad_norm": 0.9766379594802856,
      "learning_rate": 0.0002565581056077275,
      "loss": 2.0745,
      "step": 125859
    },
    {
      "epoch": 1.64,
      "grad_norm": 0.9483067989349365,
      "learning_rate": 0.00025655405816403753,
      "loss": 2.4019,
      "step": 125860
    },
    {
      "epoch": 1.64,
      "grad_norm": 0.960849404335022,
      "learning_rate": 0.0002565500107284249,
      "loss": 2.0143,
      "step": 125861
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.0685654878616333,
      "learning_rate": 0.00025654596330089056,
      "loss": 2.2812,
      "step": 125862
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.1994497776031494,
      "learning_rate": 0.00025654191588143503,
      "loss": 2.2359,
      "step": 125863
    },
    {
      "epoch": 1.64,
      "grad_norm": 0.9832522869110107,
      "learning_rate": 0.0002565378684700592,
      "loss": 2.0227,
      "step": 125864
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.0878528356552124,
      "learning_rate": 0.00025653382106676376,
      "loss": 2.1521,
      "step": 125865
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.0259604454040527,
      "learning_rate": 0.00025652977367154954,
      "loss": 2.0811,
      "step": 125866
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.1804609298706055,
      "learning_rate": 0.0002565257262844172,
      "loss": 2.2057,
      "step": 125867
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.010581374168396,
      "learning_rate": 0.00025652167890536775,
      "loss": 2.0848,
      "step": 125868
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.0091664791107178,
      "learning_rate": 0.0002565176315344016,
      "loss": 1.906,
      "step": 125869
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.0272727012634277,
      "learning_rate": 0.00025651358417151963,
      "loss": 2.1789,
      "step": 125870
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.0407007932662964,
      "learning_rate": 0.0002565095368167226,
      "loss": 2.25,
      "step": 125871
    },
    {
      "epoch": 1.64,
      "grad_norm": 0.9904422760009766,
      "learning_rate": 0.00025650548947001126,
      "loss": 2.0257,
      "step": 125872
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.0111993551254272,
      "learning_rate": 0.0002565014421313864,
      "loss": 1.9784,
      "step": 125873
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.0482714176177979,
      "learning_rate": 0.00025649739480084887,
      "loss": 2.0527,
      "step": 125874
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.3080670833587646,
      "learning_rate": 0.00025649334747839913,
      "loss": 2.0142,
      "step": 125875
    },
    {
      "epoch": 1.64,
      "grad_norm": 0.9656082391738892,
      "learning_rate": 0.00025648930016403815,
      "loss": 2.1373,
      "step": 125876
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.0012695789337158,
      "learning_rate": 0.00025648525285776655,
      "loss": 2.1809,
      "step": 125877
    },
    {
      "epoch": 1.64,
      "grad_norm": 0.9354183673858643,
      "learning_rate": 0.0002564812055595852,
      "loss": 2.2731,
      "step": 125878
    },
    {
      "epoch": 1.64,
      "grad_norm": 0.9579216241836548,
      "learning_rate": 0.00025647715826949486,
      "loss": 2.062,
      "step": 125879
    },
    {
      "epoch": 1.64,
      "grad_norm": 0.9322157502174377,
      "learning_rate": 0.00025647311098749633,
      "loss": 2.1697,
      "step": 125880
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.0119960308074951,
      "learning_rate": 0.0002564690637135901,
      "loss": 2.123,
      "step": 125881
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.111645221710205,
      "learning_rate": 0.00025646501644777716,
      "loss": 2.1407,
      "step": 125882
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.0511311292648315,
      "learning_rate": 0.0002564609691900582,
      "loss": 2.1515,
      "step": 125883
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.0321943759918213,
      "learning_rate": 0.00025645692194043395,
      "loss": 2.1951,
      "step": 125884
    },
    {
      "epoch": 1.64,
      "grad_norm": 0.9357374310493469,
      "learning_rate": 0.00025645287469890516,
      "loss": 2.1748,
      "step": 125885
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.0938363075256348,
      "learning_rate": 0.00025644882746547276,
      "loss": 2.0464,
      "step": 125886
    },
    {
      "epoch": 1.64,
      "grad_norm": 0.9916321635246277,
      "learning_rate": 0.0002564447802401372,
      "loss": 2.1678,
      "step": 125887
    },
    {
      "epoch": 1.64,
      "grad_norm": 0.9425390362739563,
      "learning_rate": 0.00025644073302289936,
      "loss": 2.1474,
      "step": 125888
    },
    {
      "epoch": 1.64,
      "grad_norm": 0.907352864742279,
      "learning_rate": 0.00025643668581376,
      "loss": 1.9901,
      "step": 125889
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.0140471458435059,
      "learning_rate": 0.00025643263861271996,
      "loss": 2.0932,
      "step": 125890
    },
    {
      "epoch": 1.64,
      "grad_norm": 0.9892944097518921,
      "learning_rate": 0.0002564285914197798,
      "loss": 2.1427,
      "step": 125891
    },
    {
      "epoch": 1.64,
      "grad_norm": 0.9969463348388672,
      "learning_rate": 0.0002564245442349406,
      "loss": 2.1792,
      "step": 125892
    },
    {
      "epoch": 1.64,
      "grad_norm": 0.9530028104782104,
      "learning_rate": 0.00025642049705820274,
      "loss": 2.0835,
      "step": 125893
    },
    {
      "epoch": 1.64,
      "grad_norm": 0.9709483981132507,
      "learning_rate": 0.00025641644988956715,
      "loss": 2.036,
      "step": 125894
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.0685102939605713,
      "learning_rate": 0.00025641240272903455,
      "loss": 1.9541,
      "step": 125895
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.0836677551269531,
      "learning_rate": 0.0002564083555766057,
      "loss": 2.262,
      "step": 125896
    },
    {
      "epoch": 1.64,
      "grad_norm": 0.9890791177749634,
      "learning_rate": 0.0002564043084322814,
      "loss": 2.0135,
      "step": 125897
    },
    {
      "epoch": 1.64,
      "grad_norm": 0.9968915581703186,
      "learning_rate": 0.0002564002612960625,
      "loss": 2.2211,
      "step": 125898
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.0815950632095337,
      "learning_rate": 0.0002563962141679494,
      "loss": 2.1417,
      "step": 125899
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.069096565246582,
      "learning_rate": 0.00025639216704794317,
      "loss": 2.2025,
      "step": 125900
    },
    {
      "epoch": 1.64,
      "grad_norm": 0.9715240001678467,
      "learning_rate": 0.00025638811993604436,
      "loss": 2.0358,
      "step": 125901
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.0789353847503662,
      "learning_rate": 0.00025638407283225386,
      "loss": 2.1554,
      "step": 125902
    },
    {
      "epoch": 1.64,
      "grad_norm": 0.9140312075614929,
      "learning_rate": 0.0002563800257365724,
      "loss": 2.1924,
      "step": 125903
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.0274410247802734,
      "learning_rate": 0.0002563759786490008,
      "loss": 2.0857,
      "step": 125904
    },
    {
      "epoch": 1.64,
      "grad_norm": 0.9822436571121216,
      "learning_rate": 0.0002563719315695397,
      "loss": 2.3196,
      "step": 125905
    },
    {
      "epoch": 1.64,
      "grad_norm": 0.9805966019630432,
      "learning_rate": 0.00025636788449818977,
      "loss": 2.1175,
      "step": 125906
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.0458039045333862,
      "learning_rate": 0.00025636383743495196,
      "loss": 2.1878,
      "step": 125907
    },
    {
      "epoch": 1.64,
      "grad_norm": 0.9992777109146118,
      "learning_rate": 0.0002563597903798269,
      "loss": 2.2372,
      "step": 125908
    },
    {
      "epoch": 1.64,
      "grad_norm": 0.9744545817375183,
      "learning_rate": 0.0002563557433328154,
      "loss": 1.797,
      "step": 125909
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.0241026878356934,
      "learning_rate": 0.0002563516962939182,
      "loss": 2.0951,
      "step": 125910
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.0474011898040771,
      "learning_rate": 0.000256347649263136,
      "loss": 2.2185,
      "step": 125911
    },
    {
      "epoch": 1.64,
      "grad_norm": 0.8994464874267578,
      "learning_rate": 0.0002563436022404697,
      "loss": 2.0072,
      "step": 125912
    },
    {
      "epoch": 1.64,
      "grad_norm": 0.9536492228507996,
      "learning_rate": 0.00025633955522591983,
      "loss": 1.9649,
      "step": 125913
    },
    {
      "epoch": 1.64,
      "grad_norm": 0.9736594557762146,
      "learning_rate": 0.00025633550821948733,
      "loss": 2.1001,
      "step": 125914
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.0470061302185059,
      "learning_rate": 0.0002563314612211728,
      "loss": 2.0507,
      "step": 125915
    },
    {
      "epoch": 1.64,
      "grad_norm": 0.9333809614181519,
      "learning_rate": 0.00025632741423097714,
      "loss": 2.1868,
      "step": 125916
    },
    {
      "epoch": 1.64,
      "grad_norm": 0.9469568133354187,
      "learning_rate": 0.0002563233672489011,
      "loss": 2.1481,
      "step": 125917
    },
    {
      "epoch": 1.64,
      "grad_norm": 0.9784938097000122,
      "learning_rate": 0.00025631932027494526,
      "loss": 2.0059,
      "step": 125918
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.019871711730957,
      "learning_rate": 0.00025631527330911055,
      "loss": 2.3103,
      "step": 125919
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.0334950685501099,
      "learning_rate": 0.00025631122635139765,
      "loss": 2.318,
      "step": 125920
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.0048387050628662,
      "learning_rate": 0.00025630717940180726,
      "loss": 2.0146,
      "step": 125921
    },
    {
      "epoch": 1.64,
      "grad_norm": 0.9884852766990662,
      "learning_rate": 0.0002563031324603402,
      "loss": 2.0415,
      "step": 125922
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.0048587322235107,
      "learning_rate": 0.0002562990855269973,
      "loss": 2.0247,
      "step": 125923
    },
    {
      "epoch": 1.64,
      "grad_norm": 0.9669787883758545,
      "learning_rate": 0.00025629503860177913,
      "loss": 2.1597,
      "step": 125924
    },
    {
      "epoch": 1.64,
      "grad_norm": 0.9754927754402161,
      "learning_rate": 0.00025629099168468657,
      "loss": 2.0464,
      "step": 125925
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.0908432006835938,
      "learning_rate": 0.0002562869447757203,
      "loss": 1.9382,
      "step": 125926
    },
    {
      "epoch": 1.64,
      "grad_norm": 0.9627520442008972,
      "learning_rate": 0.00025628289787488125,
      "loss": 1.9354,
      "step": 125927
    },
    {
      "epoch": 1.64,
      "grad_norm": 0.9545378684997559,
      "learning_rate": 0.0002562788509821699,
      "loss": 2.137,
      "step": 125928
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.0444403886795044,
      "learning_rate": 0.00025627480409758725,
      "loss": 2.1535,
      "step": 125929
    },
    {
      "epoch": 1.64,
      "grad_norm": 0.887412428855896,
      "learning_rate": 0.00025627075722113384,
      "loss": 2.0896,
      "step": 125930
    },
    {
      "epoch": 1.64,
      "grad_norm": 0.9410398006439209,
      "learning_rate": 0.0002562667103528105,
      "loss": 2.0976,
      "step": 125931
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.0332398414611816,
      "learning_rate": 0.00025626266349261805,
      "loss": 2.1606,
      "step": 125932
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.0363935232162476,
      "learning_rate": 0.0002562586166405572,
      "loss": 2.0688,
      "step": 125933
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.1296806335449219,
      "learning_rate": 0.00025625456979662875,
      "loss": 2.254,
      "step": 125934
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.0178648233413696,
      "learning_rate": 0.00025625052296083335,
      "loss": 2.2305,
      "step": 125935
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.0901283025741577,
      "learning_rate": 0.0002562464761331718,
      "loss": 2.116,
      "step": 125936
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.1752625703811646,
      "learning_rate": 0.00025624242931364486,
      "loss": 2.0479,
      "step": 125937
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.153159737586975,
      "learning_rate": 0.0002562383825022533,
      "loss": 2.2963,
      "step": 125938
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.0154123306274414,
      "learning_rate": 0.0002562343356989978,
      "loss": 1.9788,
      "step": 125939
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.2194424867630005,
      "learning_rate": 0.0002562302889038791,
      "loss": 2.0013,
      "step": 125940
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.3010327816009521,
      "learning_rate": 0.0002562262421168982,
      "loss": 2.0225,
      "step": 125941
    },
    {
      "epoch": 1.64,
      "grad_norm": 0.9963889122009277,
      "learning_rate": 0.0002562221953380556,
      "loss": 2.1883,
      "step": 125942
    },
    {
      "epoch": 1.64,
      "grad_norm": 0.9748380184173584,
      "learning_rate": 0.00025621814856735206,
      "loss": 2.1306,
      "step": 125943
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.0310100317001343,
      "learning_rate": 0.00025621410180478845,
      "loss": 2.1121,
      "step": 125944
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.288581371307373,
      "learning_rate": 0.0002562100550503654,
      "loss": 2.0049,
      "step": 125945
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.044660210609436,
      "learning_rate": 0.00025620600830408375,
      "loss": 2.1683,
      "step": 125946
    },
    {
      "epoch": 1.64,
      "grad_norm": 0.8582308888435364,
      "learning_rate": 0.0002562019615659444,
      "loss": 1.9496,
      "step": 125947
    },
    {
      "epoch": 1.64,
      "grad_norm": 0.9715307354927063,
      "learning_rate": 0.0002561979148359477,
      "loss": 2.3621,
      "step": 125948
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.2711529731750488,
      "learning_rate": 0.00025619386811409476,
      "loss": 2.2663,
      "step": 125949
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.3044350147247314,
      "learning_rate": 0.0002561898214003861,
      "loss": 2.081,
      "step": 125950
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.03128182888031,
      "learning_rate": 0.00025618577469482264,
      "loss": 2.2739,
      "step": 125951
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.1309772729873657,
      "learning_rate": 0.00025618172799740505,
      "loss": 2.1157,
      "step": 125952
    },
    {
      "epoch": 1.64,
      "grad_norm": 0.9784453511238098,
      "learning_rate": 0.0002561776813081342,
      "loss": 1.9541,
      "step": 125953
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.001133918762207,
      "learning_rate": 0.0002561736346270107,
      "loss": 1.8032,
      "step": 125954
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.0254297256469727,
      "learning_rate": 0.0002561695879540353,
      "loss": 2.3875,
      "step": 125955
    },
    {
      "epoch": 1.64,
      "grad_norm": 0.9917744398117065,
      "learning_rate": 0.0002561655412892088,
      "loss": 2.0346,
      "step": 125956
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.0746333599090576,
      "learning_rate": 0.00025616149463253196,
      "loss": 1.9453,
      "step": 125957
    },
    {
      "epoch": 1.64,
      "grad_norm": 0.966492772102356,
      "learning_rate": 0.00025615744798400554,
      "loss": 2.1865,
      "step": 125958
    },
    {
      "epoch": 1.64,
      "grad_norm": 0.935545802116394,
      "learning_rate": 0.0002561534013436304,
      "loss": 1.8675,
      "step": 125959
    },
    {
      "epoch": 1.64,
      "grad_norm": 0.9909840822219849,
      "learning_rate": 0.000256149354711407,
      "loss": 2.2109,
      "step": 125960
    },
    {
      "epoch": 1.64,
      "grad_norm": 0.9553605318069458,
      "learning_rate": 0.00025614530808733634,
      "loss": 2.1338,
      "step": 125961
    },
    {
      "epoch": 1.64,
      "grad_norm": 0.9788898825645447,
      "learning_rate": 0.000256141261471419,
      "loss": 2.0879,
      "step": 125962
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.0933243036270142,
      "learning_rate": 0.00025613721486365587,
      "loss": 2.3137,
      "step": 125963
    },
    {
      "epoch": 1.64,
      "grad_norm": 0.9213735461235046,
      "learning_rate": 0.0002561331682640477,
      "loss": 2.1882,
      "step": 125964
    },
    {
      "epoch": 1.64,
      "grad_norm": 0.9215401411056519,
      "learning_rate": 0.0002561291216725953,
      "loss": 2.1435,
      "step": 125965
    },
    {
      "epoch": 1.64,
      "grad_norm": 0.9739314913749695,
      "learning_rate": 0.0002561250750892992,
      "loss": 1.836,
      "step": 125966
    },
    {
      "epoch": 1.64,
      "grad_norm": 0.9312645792961121,
      "learning_rate": 0.00025612102851416027,
      "loss": 2.1704,
      "step": 125967
    },
    {
      "epoch": 1.64,
      "grad_norm": 0.9629712700843811,
      "learning_rate": 0.00025611698194717925,
      "loss": 2.366,
      "step": 125968
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.0065405368804932,
      "learning_rate": 0.00025611293538835696,
      "loss": 2.2605,
      "step": 125969
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.1513787508010864,
      "learning_rate": 0.00025610888883769405,
      "loss": 2.2745,
      "step": 125970
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.0330772399902344,
      "learning_rate": 0.0002561048422951915,
      "loss": 2.0771,
      "step": 125971
    },
    {
      "epoch": 1.64,
      "grad_norm": 0.9419721364974976,
      "learning_rate": 0.00025610079576084973,
      "loss": 2.121,
      "step": 125972
    },
    {
      "epoch": 1.64,
      "grad_norm": 0.9051543474197388,
      "learning_rate": 0.0002560967492346697,
      "loss": 2.1474,
      "step": 125973
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.282348394393921,
      "learning_rate": 0.00025609270271665204,
      "loss": 1.9561,
      "step": 125974
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.0980576276779175,
      "learning_rate": 0.0002560886562067976,
      "loss": 2.0562,
      "step": 125975
    },
    {
      "epoch": 1.64,
      "grad_norm": 0.9039430618286133,
      "learning_rate": 0.0002560846097051071,
      "loss": 2.4936,
      "step": 125976
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.086069107055664,
      "learning_rate": 0.00025608056321158143,
      "loss": 2.1537,
      "step": 125977
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.3087427616119385,
      "learning_rate": 0.00025607651672622116,
      "loss": 2.1409,
      "step": 125978
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.156136155128479,
      "learning_rate": 0.00025607247024902704,
      "loss": 2.0152,
      "step": 125979
    },
    {
      "epoch": 1.64,
      "grad_norm": 0.9457001686096191,
      "learning_rate": 0.00025606842377999984,
      "loss": 2.0721,
      "step": 125980
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.1006925106048584,
      "learning_rate": 0.00025606437731914036,
      "loss": 2.4266,
      "step": 125981
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.3376317024230957,
      "learning_rate": 0.0002560603308664494,
      "loss": 2.1218,
      "step": 125982
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.1790046691894531,
      "learning_rate": 0.0002560562844219276,
      "loss": 2.2322,
      "step": 125983
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.0517467260360718,
      "learning_rate": 0.0002560522379855759,
      "loss": 2.0498,
      "step": 125984
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.241840124130249,
      "learning_rate": 0.00025604819155739477,
      "loss": 2.0444,
      "step": 125985
    },
    {
      "epoch": 1.64,
      "grad_norm": 0.9630386829376221,
      "learning_rate": 0.00025604414513738516,
      "loss": 2.4235,
      "step": 125986
    },
    {
      "epoch": 1.64,
      "grad_norm": 0.9590988755226135,
      "learning_rate": 0.0002560400987255478,
      "loss": 2.2501,
      "step": 125987
    },
    {
      "epoch": 1.64,
      "grad_norm": 0.9240273237228394,
      "learning_rate": 0.00025603605232188333,
      "loss": 2.351,
      "step": 125988
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.009331226348877,
      "learning_rate": 0.00025603200592639263,
      "loss": 2.1185,
      "step": 125989
    },
    {
      "epoch": 1.64,
      "grad_norm": 0.972886323928833,
      "learning_rate": 0.00025602795953907654,
      "loss": 2.1969,
      "step": 125990
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.0061156749725342,
      "learning_rate": 0.00025602391315993555,
      "loss": 2.0736,
      "step": 125991
    },
    {
      "epoch": 1.64,
      "grad_norm": 0.9901036024093628,
      "learning_rate": 0.00025601986678897053,
      "loss": 2.1968,
      "step": 125992
    },
    {
      "epoch": 1.64,
      "grad_norm": 0.8949464559555054,
      "learning_rate": 0.00025601582042618224,
      "loss": 2.0092,
      "step": 125993
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.0020979642868042,
      "learning_rate": 0.0002560117740715715,
      "loss": 1.9974,
      "step": 125994
    },
    {
      "epoch": 1.64,
      "grad_norm": 0.9908378720283508,
      "learning_rate": 0.000256007727725139,
      "loss": 2.0261,
      "step": 125995
    },
    {
      "epoch": 1.64,
      "grad_norm": 0.9778854846954346,
      "learning_rate": 0.0002560036813868855,
      "loss": 1.9411,
      "step": 125996
    },
    {
      "epoch": 1.64,
      "grad_norm": 0.939868688583374,
      "learning_rate": 0.00025599963505681174,
      "loss": 1.9961,
      "step": 125997
    },
    {
      "epoch": 1.64,
      "grad_norm": 0.9660605192184448,
      "learning_rate": 0.0002559955887349184,
      "loss": 2.0671,
      "step": 125998
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.0922274589538574,
      "learning_rate": 0.0002559915424212064,
      "loss": 1.9824,
      "step": 125999
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.04485023021698,
      "learning_rate": 0.0002559874961156764,
      "loss": 2.1789,
      "step": 126000
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.0202467441558838,
      "learning_rate": 0.00025598344981832906,
      "loss": 2.117,
      "step": 126001
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.0205129384994507,
      "learning_rate": 0.00025597940352916536,
      "loss": 1.9698,
      "step": 126002
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.0431572198867798,
      "learning_rate": 0.0002559753572481858,
      "loss": 2.1329,
      "step": 126003
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.0161911249160767,
      "learning_rate": 0.00025597131097539135,
      "loss": 2.1362,
      "step": 126004
    },
    {
      "epoch": 1.64,
      "grad_norm": 0.9601271748542786,
      "learning_rate": 0.0002559672647107826,
      "loss": 2.3945,
      "step": 126005
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.0784857273101807,
      "learning_rate": 0.0002559632184543604,
      "loss": 2.0778,
      "step": 126006
    },
    {
      "epoch": 1.64,
      "grad_norm": 0.9702132344245911,
      "learning_rate": 0.00025595917220612544,
      "loss": 2.2072,
      "step": 126007
    },
    {
      "epoch": 1.64,
      "grad_norm": 0.970660388469696,
      "learning_rate": 0.00025595512596607854,
      "loss": 2.1811,
      "step": 126008
    },
    {
      "epoch": 1.64,
      "grad_norm": 0.9524794816970825,
      "learning_rate": 0.00025595107973422034,
      "loss": 2.0567,
      "step": 126009
    },
    {
      "epoch": 1.64,
      "grad_norm": 0.9580238461494446,
      "learning_rate": 0.0002559470335105517,
      "loss": 2.092,
      "step": 126010
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.2500157356262207,
      "learning_rate": 0.00025594298729507344,
      "loss": 2.0869,
      "step": 126011
    },
    {
      "epoch": 1.64,
      "grad_norm": 0.9689408540725708,
      "learning_rate": 0.00025593894108778614,
      "loss": 2.1524,
      "step": 126012
    },
    {
      "epoch": 1.64,
      "grad_norm": 0.995223879814148,
      "learning_rate": 0.00025593489488869053,
      "loss": 2.1529,
      "step": 126013
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.0831807851791382,
      "learning_rate": 0.00025593084869778764,
      "loss": 2.1846,
      "step": 126014
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.10594642162323,
      "learning_rate": 0.00025592680251507784,
      "loss": 2.1411,
      "step": 126015
    },
    {
      "epoch": 1.64,
      "grad_norm": 0.9489443302154541,
      "learning_rate": 0.0002559227563405622,
      "loss": 2.2212,
      "step": 126016
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.2149111032485962,
      "learning_rate": 0.0002559187101742413,
      "loss": 2.2215,
      "step": 126017
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.0175931453704834,
      "learning_rate": 0.00025591466401611595,
      "loss": 2.0914,
      "step": 126018
    },
    {
      "epoch": 1.64,
      "grad_norm": 0.8400779366493225,
      "learning_rate": 0.00025591061786618696,
      "loss": 1.987,
      "step": 126019
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.0068416595458984,
      "learning_rate": 0.00025590657172445497,
      "loss": 2.2073,
      "step": 126020
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.1531176567077637,
      "learning_rate": 0.0002559025255909208,
      "loss": 2.2976,
      "step": 126021
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.1542524099349976,
      "learning_rate": 0.00025589847946558514,
      "loss": 1.9951,
      "step": 126022
    },
    {
      "epoch": 1.64,
      "grad_norm": 0.9794403910636902,
      "learning_rate": 0.0002558944333484488,
      "loss": 2.2154,
      "step": 126023
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.003492832183838,
      "learning_rate": 0.0002558903872395125,
      "loss": 2.2352,
      "step": 126024
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.0067241191864014,
      "learning_rate": 0.000255886341138777,
      "loss": 2.2455,
      "step": 126025
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.1436210870742798,
      "learning_rate": 0.0002558822950462432,
      "loss": 1.9786,
      "step": 126026
    },
    {
      "epoch": 1.64,
      "grad_norm": 0.9448622465133667,
      "learning_rate": 0.0002558782489619116,
      "loss": 2.0869,
      "step": 126027
    },
    {
      "epoch": 1.64,
      "grad_norm": 0.9240128993988037,
      "learning_rate": 0.00025587420288578305,
      "loss": 1.8137,
      "step": 126028
    },
    {
      "epoch": 1.64,
      "grad_norm": 0.9740355610847473,
      "learning_rate": 0.00025587015681785834,
      "loss": 2.0852,
      "step": 126029
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.0418342351913452,
      "learning_rate": 0.0002558661107581382,
      "loss": 2.193,
      "step": 126030
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.2954462766647339,
      "learning_rate": 0.0002558620647066234,
      "loss": 2.2264,
      "step": 126031
    },
    {
      "epoch": 1.64,
      "grad_norm": 0.95550936460495,
      "learning_rate": 0.00025585801866331474,
      "loss": 2.2333,
      "step": 126032
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.0518467426300049,
      "learning_rate": 0.00025585397262821284,
      "loss": 2.2364,
      "step": 126033
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.054132103919983,
      "learning_rate": 0.00025584992660131846,
      "loss": 2.0606,
      "step": 126034
    },
    {
      "epoch": 1.64,
      "grad_norm": 0.9589912295341492,
      "learning_rate": 0.00025584588058263246,
      "loss": 1.9745,
      "step": 126035
    },
    {
      "epoch": 1.64,
      "grad_norm": 0.9979461431503296,
      "learning_rate": 0.00025584183457215556,
      "loss": 2.3715,
      "step": 126036
    },
    {
      "epoch": 1.64,
      "grad_norm": 0.9399936199188232,
      "learning_rate": 0.00025583778856988845,
      "loss": 2.1289,
      "step": 126037
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.0686101913452148,
      "learning_rate": 0.0002558337425758321,
      "loss": 2.0929,
      "step": 126038
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.0510663986206055,
      "learning_rate": 0.0002558296965899869,
      "loss": 2.0498,
      "step": 126039
    },
    {
      "epoch": 1.64,
      "grad_norm": 0.9849023222923279,
      "learning_rate": 0.0002558256506123538,
      "loss": 2.0857,
      "step": 126040
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.1411066055297852,
      "learning_rate": 0.0002558216046429336,
      "loss": 2.0858,
      "step": 126041
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.0372121334075928,
      "learning_rate": 0.000255817558681727,
      "loss": 2.1285,
      "step": 126042
    },
    {
      "epoch": 1.64,
      "grad_norm": 0.9656491279602051,
      "learning_rate": 0.0002558135127287347,
      "loss": 2.2498,
      "step": 126043
    },
    {
      "epoch": 1.64,
      "grad_norm": 0.8802595734596252,
      "learning_rate": 0.00025580946678395764,
      "loss": 2.0829,
      "step": 126044
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.0937386751174927,
      "learning_rate": 0.00025580542084739633,
      "loss": 1.8629,
      "step": 126045
    },
    {
      "epoch": 1.64,
      "grad_norm": 0.8689108490943909,
      "learning_rate": 0.00025580137491905164,
      "loss": 1.932,
      "step": 126046
    },
    {
      "epoch": 1.64,
      "grad_norm": 0.930794358253479,
      "learning_rate": 0.0002557973289989243,
      "loss": 1.9175,
      "step": 126047
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.0227116346359253,
      "learning_rate": 0.000255793283087015,
      "loss": 2.1395,
      "step": 126048
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.0051389932632446,
      "learning_rate": 0.00025578923718332466,
      "loss": 2.1797,
      "step": 126049
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.0135809183120728,
      "learning_rate": 0.00025578519128785385,
      "loss": 2.0835,
      "step": 126050
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.0132936239242554,
      "learning_rate": 0.0002557811454006036,
      "loss": 1.9154,
      "step": 126051
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.0876415967941284,
      "learning_rate": 0.00025577709952157436,
      "loss": 2.2368,
      "step": 126052
    },
    {
      "epoch": 1.64,
      "grad_norm": 0.9649786949157715,
      "learning_rate": 0.0002557730536507669,
      "loss": 2.0552,
      "step": 126053
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.001863718032837,
      "learning_rate": 0.00025576900778818214,
      "loss": 2.0979,
      "step": 126054
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.096760869026184,
      "learning_rate": 0.00025576496193382075,
      "loss": 2.1065,
      "step": 126055
    },
    {
      "epoch": 1.64,
      "grad_norm": 0.9764345288276672,
      "learning_rate": 0.00025576091608768347,
      "loss": 2.0947,
      "step": 126056
    },
    {
      "epoch": 1.64,
      "grad_norm": 0.9030618667602539,
      "learning_rate": 0.0002557568702497712,
      "loss": 2.3596,
      "step": 126057
    },
    {
      "epoch": 1.64,
      "grad_norm": 0.9406100511550903,
      "learning_rate": 0.00025575282442008443,
      "loss": 2.0498,
      "step": 126058
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.0127055644989014,
      "learning_rate": 0.00025574877859862403,
      "loss": 2.1028,
      "step": 126059
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.032099962234497,
      "learning_rate": 0.0002557447327853908,
      "loss": 1.9473,
      "step": 126060
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.0252183675765991,
      "learning_rate": 0.00025574068698038546,
      "loss": 1.9695,
      "step": 126061
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.0437397956848145,
      "learning_rate": 0.00025573664118360876,
      "loss": 2.2176,
      "step": 126062
    },
    {
      "epoch": 1.64,
      "grad_norm": 0.9334825873374939,
      "learning_rate": 0.00025573259539506154,
      "loss": 2.0273,
      "step": 126063
    },
    {
      "epoch": 1.64,
      "grad_norm": 0.8949728012084961,
      "learning_rate": 0.0002557285496147444,
      "loss": 1.7655,
      "step": 126064
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.0310711860656738,
      "learning_rate": 0.00025572450384265817,
      "loss": 2.1669,
      "step": 126065
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.0138978958129883,
      "learning_rate": 0.00025572045807880354,
      "loss": 2.2347,
      "step": 126066
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.2055505514144897,
      "learning_rate": 0.0002557164123231813,
      "loss": 2.1751,
      "step": 126067
    },
    {
      "epoch": 1.64,
      "grad_norm": 0.994743287563324,
      "learning_rate": 0.00025571236657579224,
      "loss": 2.2492,
      "step": 126068
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.324344515800476,
      "learning_rate": 0.0002557083208366372,
      "loss": 2.0874,
      "step": 126069
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.2737897634506226,
      "learning_rate": 0.0002557042751057167,
      "loss": 2.1004,
      "step": 126070
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.4895130395889282,
      "learning_rate": 0.00025570022938303165,
      "loss": 2.1301,
      "step": 126071
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.0440187454223633,
      "learning_rate": 0.0002556961836685827,
      "loss": 1.9819,
      "step": 126072
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.1260942220687866,
      "learning_rate": 0.00025569213796237066,
      "loss": 2.0554,
      "step": 126073
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.06814444065094,
      "learning_rate": 0.00025568809226439633,
      "loss": 2.119,
      "step": 126074
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.2209864854812622,
      "learning_rate": 0.0002556840465746605,
      "loss": 2.0266,
      "step": 126075
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.178977131843567,
      "learning_rate": 0.0002556800008931638,
      "loss": 2.1967,
      "step": 126076
    },
    {
      "epoch": 1.64,
      "grad_norm": 0.871139407157898,
      "learning_rate": 0.000255675955219907,
      "loss": 2.2237,
      "step": 126077
    },
    {
      "epoch": 1.64,
      "grad_norm": 0.9584002494812012,
      "learning_rate": 0.00025567190955489084,
      "loss": 1.8765,
      "step": 126078
    },
    {
      "epoch": 1.64,
      "grad_norm": 0.9712583422660828,
      "learning_rate": 0.00025566786389811613,
      "loss": 1.9439,
      "step": 126079
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.107268214225769,
      "learning_rate": 0.0002556638182495836,
      "loss": 2.4137,
      "step": 126080
    },
    {
      "epoch": 1.64,
      "grad_norm": 0.9749321341514587,
      "learning_rate": 0.0002556597726092941,
      "loss": 2.3199,
      "step": 126081
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.1122820377349854,
      "learning_rate": 0.0002556557269772482,
      "loss": 2.1875,
      "step": 126082
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.063591480255127,
      "learning_rate": 0.0002556516813534467,
      "loss": 2.1268,
      "step": 126083
    },
    {
      "epoch": 1.64,
      "grad_norm": 0.9843572974205017,
      "learning_rate": 0.00025564763573789046,
      "loss": 2.0377,
      "step": 126084
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.0751465559005737,
      "learning_rate": 0.0002556435901305801,
      "loss": 2.295,
      "step": 126085
    },
    {
      "epoch": 1.64,
      "grad_norm": 0.8849488496780396,
      "learning_rate": 0.00025563954453151645,
      "loss": 2.1606,
      "step": 126086
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.0268679857254028,
      "learning_rate": 0.00025563549894070027,
      "loss": 2.1872,
      "step": 126087
    },
    {
      "epoch": 1.64,
      "grad_norm": 0.9376797080039978,
      "learning_rate": 0.00025563145335813226,
      "loss": 2.2496,
      "step": 126088
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.0922517776489258,
      "learning_rate": 0.0002556274077838133,
      "loss": 2.1812,
      "step": 126089
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.0236506462097168,
      "learning_rate": 0.00025562336221774393,
      "loss": 2.3963,
      "step": 126090
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.1716407537460327,
      "learning_rate": 0.000255619316659925,
      "loss": 2.0793,
      "step": 126091
    },
    {
      "epoch": 1.64,
      "grad_norm": 0.9885770678520203,
      "learning_rate": 0.00025561527111035736,
      "loss": 2.1017,
      "step": 126092
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.118737816810608,
      "learning_rate": 0.0002556112255690417,
      "loss": 2.0855,
      "step": 126093
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.2201439142227173,
      "learning_rate": 0.0002556071800359786,
      "loss": 2.089,
      "step": 126094
    },
    {
      "epoch": 1.64,
      "grad_norm": 0.9693333506584167,
      "learning_rate": 0.00025560313451116905,
      "loss": 2.2942,
      "step": 126095
    },
    {
      "epoch": 1.64,
      "grad_norm": 0.9968582391738892,
      "learning_rate": 0.0002555990889946138,
      "loss": 2.0092,
      "step": 126096
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.0180071592330933,
      "learning_rate": 0.0002555950434863134,
      "loss": 2.1206,
      "step": 126097
    },
    {
      "epoch": 1.64,
      "grad_norm": 0.9539979100227356,
      "learning_rate": 0.00025559099798626875,
      "loss": 2.1584,
      "step": 126098
    },
    {
      "epoch": 1.64,
      "grad_norm": 0.8750343918800354,
      "learning_rate": 0.0002555869524944806,
      "loss": 2.3676,
      "step": 126099
    },
    {
      "epoch": 1.64,
      "grad_norm": 0.8931108713150024,
      "learning_rate": 0.00025558290701094966,
      "loss": 2.1649,
      "step": 126100
    },
    {
      "epoch": 1.64,
      "grad_norm": 0.9092800617218018,
      "learning_rate": 0.0002555788615356767,
      "loss": 1.9564,
      "step": 126101
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.008908987045288,
      "learning_rate": 0.0002555748160686624,
      "loss": 2.1133,
      "step": 126102
    },
    {
      "epoch": 1.64,
      "grad_norm": 0.9261062741279602,
      "learning_rate": 0.0002555707706099077,
      "loss": 2.225,
      "step": 126103
    },
    {
      "epoch": 1.64,
      "grad_norm": 0.9759464263916016,
      "learning_rate": 0.0002555667251594132,
      "loss": 2.0667,
      "step": 126104
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.072076439857483,
      "learning_rate": 0.00025556267971717965,
      "loss": 2.3017,
      "step": 126105
    },
    {
      "epoch": 1.64,
      "grad_norm": 0.9718927145004272,
      "learning_rate": 0.0002555586342832078,
      "loss": 2.0594,
      "step": 126106
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.0099421739578247,
      "learning_rate": 0.0002555545888574985,
      "loss": 2.0476,
      "step": 126107
    },
    {
      "epoch": 1.64,
      "grad_norm": 0.8752090334892273,
      "learning_rate": 0.00025555054344005235,
      "loss": 2.0218,
      "step": 126108
    },
    {
      "epoch": 1.64,
      "grad_norm": 0.9632082581520081,
      "learning_rate": 0.0002555464980308703,
      "loss": 2.1569,
      "step": 126109
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.020794153213501,
      "learning_rate": 0.00025554245262995293,
      "loss": 2.0953,
      "step": 126110
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.0777323246002197,
      "learning_rate": 0.0002555384072373012,
      "loss": 2.1707,
      "step": 126111
    },
    {
      "epoch": 1.64,
      "grad_norm": 0.9697123765945435,
      "learning_rate": 0.0002555343618529156,
      "loss": 2.2586,
      "step": 126112
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.1230463981628418,
      "learning_rate": 0.000255530316476797,
      "loss": 2.1109,
      "step": 126113
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.0758849382400513,
      "learning_rate": 0.0002555262711089461,
      "loss": 2.2186,
      "step": 126114
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.0688896179199219,
      "learning_rate": 0.00025552222574936375,
      "loss": 1.8554,
      "step": 126115
    },
    {
      "epoch": 1.64,
      "grad_norm": 0.9664026498794556,
      "learning_rate": 0.00025551818039805065,
      "loss": 2.011,
      "step": 126116
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.0700998306274414,
      "learning_rate": 0.0002555141350550076,
      "loss": 2.4184,
      "step": 126117
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.1420615911483765,
      "learning_rate": 0.0002555100897202354,
      "loss": 2.1049,
      "step": 126118
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.2483701705932617,
      "learning_rate": 0.0002555060443937346,
      "loss": 2.0647,
      "step": 126119
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.022010087966919,
      "learning_rate": 0.0002555019990755061,
      "loss": 2.1075,
      "step": 126120
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.0178722143173218,
      "learning_rate": 0.0002554979537655505,
      "loss": 2.0258,
      "step": 126121
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.6033302545547485,
      "learning_rate": 0.0002554939084638688,
      "loss": 1.8882,
      "step": 126122
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.2417997121810913,
      "learning_rate": 0.00025548986317046156,
      "loss": 2.1633,
      "step": 126123
    },
    {
      "epoch": 1.64,
      "grad_norm": 0.9769189953804016,
      "learning_rate": 0.00025548581788532973,
      "loss": 2.213,
      "step": 126124
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.1224815845489502,
      "learning_rate": 0.0002554817726084738,
      "loss": 2.3589,
      "step": 126125
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.150543451309204,
      "learning_rate": 0.00025547772733989463,
      "loss": 2.2193,
      "step": 126126
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.3422335386276245,
      "learning_rate": 0.000255473682079593,
      "loss": 2.0264,
      "step": 126127
    },
    {
      "epoch": 1.64,
      "grad_norm": 0.9935232996940613,
      "learning_rate": 0.0002554696368275697,
      "loss": 1.8589,
      "step": 126128
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.0259028673171997,
      "learning_rate": 0.0002554655915838254,
      "loss": 2.139,
      "step": 126129
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.0948662757873535,
      "learning_rate": 0.000255461546348361,
      "loss": 2.1846,
      "step": 126130
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.1576021909713745,
      "learning_rate": 0.000255457501121177,
      "loss": 2.1605,
      "step": 126131
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.067522406578064,
      "learning_rate": 0.0002554534559022743,
      "loss": 1.8785,
      "step": 126132
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.0574860572814941,
      "learning_rate": 0.0002554494106916537,
      "loss": 2.2261,
      "step": 126133
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.1227976083755493,
      "learning_rate": 0.0002554453654893159,
      "loss": 2.0855,
      "step": 126134
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.166081190109253,
      "learning_rate": 0.0002554413202952616,
      "loss": 2.3221,
      "step": 126135
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.0666273832321167,
      "learning_rate": 0.0002554372751094917,
      "loss": 2.1475,
      "step": 126136
    },
    {
      "epoch": 1.64,
      "grad_norm": 0.9881579279899597,
      "learning_rate": 0.00025543322993200676,
      "loss": 2.1053,
      "step": 126137
    },
    {
      "epoch": 1.64,
      "grad_norm": 0.9505921006202698,
      "learning_rate": 0.0002554291847628076,
      "loss": 2.1653,
      "step": 126138
    },
    {
      "epoch": 1.64,
      "grad_norm": 0.9327917098999023,
      "learning_rate": 0.00025542513960189503,
      "loss": 2.1194,
      "step": 126139
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.2377594709396362,
      "learning_rate": 0.0002554210944492698,
      "loss": 2.0041,
      "step": 126140
    },
    {
      "epoch": 1.64,
      "grad_norm": 0.9015835523605347,
      "learning_rate": 0.00025541704930493256,
      "loss": 2.0206,
      "step": 126141
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.0363359451293945,
      "learning_rate": 0.0002554130041688843,
      "loss": 2.1131,
      "step": 126142
    },
    {
      "epoch": 1.64,
      "grad_norm": 0.9233798384666443,
      "learning_rate": 0.0002554089590411254,
      "loss": 2.0804,
      "step": 126143
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.0768879652023315,
      "learning_rate": 0.00025540491392165687,
      "loss": 2.1179,
      "step": 126144
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.063511848449707,
      "learning_rate": 0.0002554008688104795,
      "loss": 2.2455,
      "step": 126145
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.0120952129364014,
      "learning_rate": 0.0002553968237075938,
      "loss": 2.2372,
      "step": 126146
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.0281552076339722,
      "learning_rate": 0.00025539277861300074,
      "loss": 1.976,
      "step": 126147
    },
    {
      "epoch": 1.64,
      "grad_norm": 0.991911768913269,
      "learning_rate": 0.00025538873352670113,
      "loss": 2.2495,
      "step": 126148
    },
    {
      "epoch": 1.64,
      "grad_norm": 0.8754465579986572,
      "learning_rate": 0.00025538468844869545,
      "loss": 1.9769,
      "step": 126149
    },
    {
      "epoch": 1.64,
      "grad_norm": 0.9688189029693604,
      "learning_rate": 0.0002553806433789847,
      "loss": 2.1387,
      "step": 126150
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.2079123258590698,
      "learning_rate": 0.00025537659831756935,
      "loss": 2.061,
      "step": 126151
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.1055692434310913,
      "learning_rate": 0.00025537255326445044,
      "loss": 2.2308,
      "step": 126152
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.043204426765442,
      "learning_rate": 0.0002553685082196286,
      "loss": 2.1725,
      "step": 126153
    },
    {
      "epoch": 1.64,
      "grad_norm": 0.9011756777763367,
      "learning_rate": 0.0002553644631831048,
      "loss": 2.2117,
      "step": 126154
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.20341157913208,
      "learning_rate": 0.00025536041815487937,
      "loss": 2.0755,
      "step": 126155
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.0594329833984375,
      "learning_rate": 0.00025535637313495326,
      "loss": 2.1782,
      "step": 126156
    },
    {
      "epoch": 1.64,
      "grad_norm": 0.9613282084465027,
      "learning_rate": 0.0002553523281233273,
      "loss": 2.0782,
      "step": 126157
    },
    {
      "epoch": 1.64,
      "grad_norm": 0.9944395422935486,
      "learning_rate": 0.0002553482831200022,
      "loss": 1.9901,
      "step": 126158
    },
    {
      "epoch": 1.64,
      "grad_norm": 0.960654079914093,
      "learning_rate": 0.0002553442381249786,
      "loss": 2.378,
      "step": 126159
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.0726498365402222,
      "learning_rate": 0.00025534019313825756,
      "loss": 2.2586,
      "step": 126160
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.0160096883773804,
      "learning_rate": 0.0002553361481598395,
      "loss": 2.2551,
      "step": 126161
    },
    {
      "epoch": 1.64,
      "grad_norm": 0.9913291335105896,
      "learning_rate": 0.00025533210318972525,
      "loss": 2.2301,
      "step": 126162
    },
    {
      "epoch": 1.64,
      "grad_norm": 0.9635499119758606,
      "learning_rate": 0.00025532805822791563,
      "loss": 2.0451,
      "step": 126163
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.0097181797027588,
      "learning_rate": 0.0002553240132744114,
      "loss": 1.862,
      "step": 126164
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.0173773765563965,
      "learning_rate": 0.0002553199683292132,
      "loss": 2.0377,
      "step": 126165
    },
    {
      "epoch": 1.64,
      "grad_norm": 0.9859676361083984,
      "learning_rate": 0.000255315923392322,
      "loss": 1.9189,
      "step": 126166
    },
    {
      "epoch": 1.64,
      "grad_norm": 0.945158064365387,
      "learning_rate": 0.00025531187846373837,
      "loss": 2.1899,
      "step": 126167
    },
    {
      "epoch": 1.64,
      "grad_norm": 0.8894199728965759,
      "learning_rate": 0.000255307833543463,
      "loss": 2.1326,
      "step": 126168
    },
    {
      "epoch": 1.64,
      "grad_norm": 0.9935476183891296,
      "learning_rate": 0.0002553037886314968,
      "loss": 2.1521,
      "step": 126169
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.0665299892425537,
      "learning_rate": 0.0002552997437278405,
      "loss": 2.0488,
      "step": 126170
    },
    {
      "epoch": 1.64,
      "grad_norm": 0.9415696859359741,
      "learning_rate": 0.0002552956988324948,
      "loss": 2.2372,
      "step": 126171
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.0448377132415771,
      "learning_rate": 0.00025529165394546055,
      "loss": 1.9858,
      "step": 126172
    },
    {
      "epoch": 1.64,
      "grad_norm": 0.9708366394042969,
      "learning_rate": 0.0002552876090667384,
      "loss": 1.9398,
      "step": 126173
    },
    {
      "epoch": 1.64,
      "grad_norm": 0.9341368675231934,
      "learning_rate": 0.00025528356419632906,
      "loss": 2.1432,
      "step": 126174
    },
    {
      "epoch": 1.64,
      "grad_norm": 0.8845100402832031,
      "learning_rate": 0.0002552795193342333,
      "loss": 2.233,
      "step": 126175
    },
    {
      "epoch": 1.64,
      "grad_norm": 0.91623455286026,
      "learning_rate": 0.000255275474480452,
      "loss": 2.0912,
      "step": 126176
    },
    {
      "epoch": 1.64,
      "grad_norm": 0.9393622279167175,
      "learning_rate": 0.00025527142963498585,
      "loss": 1.9578,
      "step": 126177
    },
    {
      "epoch": 1.64,
      "grad_norm": 0.9824514389038086,
      "learning_rate": 0.0002552673847978355,
      "loss": 2.0361,
      "step": 126178
    },
    {
      "epoch": 1.64,
      "grad_norm": 0.9646899700164795,
      "learning_rate": 0.00025526333996900194,
      "loss": 2.4833,
      "step": 126179
    },
    {
      "epoch": 1.64,
      "grad_norm": 0.9557521939277649,
      "learning_rate": 0.0002552592951484856,
      "loss": 2.2051,
      "step": 126180
    },
    {
      "epoch": 1.64,
      "grad_norm": 0.9534639716148376,
      "learning_rate": 0.0002552552503362876,
      "loss": 2.1806,
      "step": 126181
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.0098834037780762,
      "learning_rate": 0.0002552512055324083,
      "loss": 1.9892,
      "step": 126182
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.0092467069625854,
      "learning_rate": 0.00025524716073684875,
      "loss": 2.1589,
      "step": 126183
    },
    {
      "epoch": 1.64,
      "grad_norm": 0.9505136609077454,
      "learning_rate": 0.0002552431159496095,
      "loss": 1.9878,
      "step": 126184
    },
    {
      "epoch": 1.64,
      "grad_norm": 0.9776983857154846,
      "learning_rate": 0.00025523907117069156,
      "loss": 2.1721,
      "step": 126185
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.073431134223938,
      "learning_rate": 0.0002552350264000954,
      "loss": 2.1476,
      "step": 126186
    },
    {
      "epoch": 1.64,
      "grad_norm": 0.9744531512260437,
      "learning_rate": 0.0002552309816378219,
      "loss": 1.9368,
      "step": 126187
    },
    {
      "epoch": 1.64,
      "grad_norm": 0.9446686506271362,
      "learning_rate": 0.00025522693688387186,
      "loss": 2.0468,
      "step": 126188
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.0660507678985596,
      "learning_rate": 0.0002552228921382459,
      "loss": 2.1341,
      "step": 126189
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.116583228111267,
      "learning_rate": 0.0002552188474009449,
      "loss": 2.2006,
      "step": 126190
    },
    {
      "epoch": 1.64,
      "grad_norm": 0.9801785945892334,
      "learning_rate": 0.0002552148026719696,
      "loss": 1.9228,
      "step": 126191
    },
    {
      "epoch": 1.64,
      "grad_norm": 0.9172599911689758,
      "learning_rate": 0.0002552107579513207,
      "loss": 2.1614,
      "step": 126192
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.0334827899932861,
      "learning_rate": 0.0002552067132389989,
      "loss": 2.2355,
      "step": 126193
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.2398109436035156,
      "learning_rate": 0.000255202668535005,
      "loss": 2.1834,
      "step": 126194
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.031889796257019,
      "learning_rate": 0.0002551986238393399,
      "loss": 2.3747,
      "step": 126195
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.0410319566726685,
      "learning_rate": 0.0002551945791520041,
      "loss": 2.1096,
      "step": 126196
    },
    {
      "epoch": 1.64,
      "grad_norm": 0.8966314792633057,
      "learning_rate": 0.0002551905344729986,
      "loss": 2.3388,
      "step": 126197
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.1686502695083618,
      "learning_rate": 0.00025518648980232394,
      "loss": 2.1474,
      "step": 126198
    },
    {
      "epoch": 1.64,
      "grad_norm": 0.9324350357055664,
      "learning_rate": 0.00025518244513998093,
      "loss": 2.2193,
      "step": 126199
    },
    {
      "epoch": 1.64,
      "grad_norm": 0.973031759262085,
      "learning_rate": 0.00025517840048597034,
      "loss": 2.1974,
      "step": 126200
    },
    {
      "epoch": 1.64,
      "grad_norm": 0.9174718260765076,
      "learning_rate": 0.000255174355840293,
      "loss": 2.2024,
      "step": 126201
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.0408293008804321,
      "learning_rate": 0.00025517031120294957,
      "loss": 2.3119,
      "step": 126202
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.0273362398147583,
      "learning_rate": 0.0002551662665739409,
      "loss": 2.211,
      "step": 126203
    },
    {
      "epoch": 1.64,
      "grad_norm": 0.9228630065917969,
      "learning_rate": 0.0002551622219532676,
      "loss": 2.0166,
      "step": 126204
    },
    {
      "epoch": 1.64,
      "grad_norm": 0.9272689819335938,
      "learning_rate": 0.0002551581773409304,
      "loss": 1.8843,
      "step": 126205
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.069598913192749,
      "learning_rate": 0.00025515413273693023,
      "loss": 2.2035,
      "step": 126206
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.2445313930511475,
      "learning_rate": 0.0002551500881412677,
      "loss": 2.3377,
      "step": 126207
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.0487788915634155,
      "learning_rate": 0.0002551460435539437,
      "loss": 2.2401,
      "step": 126208
    },
    {
      "epoch": 1.64,
      "grad_norm": 0.9063974618911743,
      "learning_rate": 0.0002551419989749589,
      "loss": 2.245,
      "step": 126209
    },
    {
      "epoch": 1.64,
      "grad_norm": 0.9987317323684692,
      "learning_rate": 0.000255137954404314,
      "loss": 2.0314,
      "step": 126210
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.029849648475647,
      "learning_rate": 0.0002551339098420098,
      "loss": 2.1947,
      "step": 126211
    },
    {
      "epoch": 1.64,
      "grad_norm": 0.8977258801460266,
      "learning_rate": 0.00025512986528804704,
      "loss": 1.9006,
      "step": 126212
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.0245871543884277,
      "learning_rate": 0.0002551258207424265,
      "loss": 2.1462,
      "step": 126213
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.0053627490997314,
      "learning_rate": 0.0002551217762051489,
      "loss": 1.9898,
      "step": 126214
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.0988843441009521,
      "learning_rate": 0.00025511773167621513,
      "loss": 2.0496,
      "step": 126215
    },
    {
      "epoch": 1.64,
      "grad_norm": 0.9868788123130798,
      "learning_rate": 0.0002551136871556257,
      "loss": 2.3078,
      "step": 126216
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.0082863569259644,
      "learning_rate": 0.0002551096426433815,
      "loss": 2.2087,
      "step": 126217
    },
    {
      "epoch": 1.64,
      "grad_norm": 0.9571571946144104,
      "learning_rate": 0.00025510559813948325,
      "loss": 2.1137,
      "step": 126218
    },
    {
      "epoch": 1.64,
      "grad_norm": 0.9285172820091248,
      "learning_rate": 0.0002551015536439317,
      "loss": 2.0959,
      "step": 126219
    },
    {
      "epoch": 1.64,
      "grad_norm": 0.867420494556427,
      "learning_rate": 0.00025509750915672766,
      "loss": 2.063,
      "step": 126220
    },
    {
      "epoch": 1.64,
      "grad_norm": 0.8870701789855957,
      "learning_rate": 0.00025509346467787194,
      "loss": 2.169,
      "step": 126221
    },
    {
      "epoch": 1.64,
      "grad_norm": 0.9860356450080872,
      "learning_rate": 0.00025508942020736506,
      "loss": 1.9446,
      "step": 126222
    },
    {
      "epoch": 1.64,
      "grad_norm": 0.9275588393211365,
      "learning_rate": 0.0002550853757452079,
      "loss": 2.2337,
      "step": 126223
    },
    {
      "epoch": 1.64,
      "grad_norm": 0.9380103945732117,
      "learning_rate": 0.00025508133129140126,
      "loss": 2.2334,
      "step": 126224
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.0101110935211182,
      "learning_rate": 0.0002550772868459458,
      "loss": 2.24,
      "step": 126225
    },
    {
      "epoch": 1.64,
      "grad_norm": 0.9761517643928528,
      "learning_rate": 0.0002550732424088423,
      "loss": 2.1171,
      "step": 126226
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.0810070037841797,
      "learning_rate": 0.00025506919798009176,
      "loss": 1.9479,
      "step": 126227
    },
    {
      "epoch": 1.64,
      "grad_norm": 0.9797852635383606,
      "learning_rate": 0.0002550651535596945,
      "loss": 1.8823,
      "step": 126228
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.0361524820327759,
      "learning_rate": 0.0002550611091476515,
      "loss": 2.134,
      "step": 126229
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.073209524154663,
      "learning_rate": 0.0002550570647439635,
      "loss": 2.1101,
      "step": 126230
    },
    {
      "epoch": 1.64,
      "grad_norm": 0.9293441772460938,
      "learning_rate": 0.0002550530203486312,
      "loss": 2.2006,
      "step": 126231
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.0829194784164429,
      "learning_rate": 0.0002550489759616554,
      "loss": 2.154,
      "step": 126232
    },
    {
      "epoch": 1.64,
      "grad_norm": 0.9602694511413574,
      "learning_rate": 0.000255044931583037,
      "loss": 2.1046,
      "step": 126233
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.1666874885559082,
      "learning_rate": 0.00025504088721277644,
      "loss": 1.8852,
      "step": 126234
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.0884839296340942,
      "learning_rate": 0.00025503684285087464,
      "loss": 2.166,
      "step": 126235
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.0141433477401733,
      "learning_rate": 0.00025503279849733236,
      "loss": 2.218,
      "step": 126236
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.0396515130996704,
      "learning_rate": 0.00025502875415215026,
      "loss": 2.2666,
      "step": 126237
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.2309982776641846,
      "learning_rate": 0.00025502470981532925,
      "loss": 2.3839,
      "step": 126238
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.0786429643630981,
      "learning_rate": 0.0002550206654868701,
      "loss": 1.8464,
      "step": 126239
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.1659311056137085,
      "learning_rate": 0.0002550166211667733,
      "loss": 2.2801,
      "step": 126240
    },
    {
      "epoch": 1.64,
      "grad_norm": 0.985012948513031,
      "learning_rate": 0.00025501257685503975,
      "loss": 2.1485,
      "step": 126241
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.1753809452056885,
      "learning_rate": 0.0002550085325516703,
      "loss": 1.8305,
      "step": 126242
    },
    {
      "epoch": 1.64,
      "grad_norm": 0.9489991068840027,
      "learning_rate": 0.0002550044882566655,
      "loss": 2.0461,
      "step": 126243
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.0007160902023315,
      "learning_rate": 0.0002550004439700263,
      "loss": 2.0043,
      "step": 126244
    },
    {
      "epoch": 1.64,
      "grad_norm": 0.9207092523574829,
      "learning_rate": 0.0002549963996917533,
      "loss": 2.1789,
      "step": 126245
    },
    {
      "epoch": 1.64,
      "grad_norm": 0.9453385472297668,
      "learning_rate": 0.0002549923554218475,
      "loss": 2.2937,
      "step": 126246
    },
    {
      "epoch": 1.64,
      "grad_norm": 0.9298943877220154,
      "learning_rate": 0.0002549883111603093,
      "loss": 2.3148,
      "step": 126247
    },
    {
      "epoch": 1.64,
      "grad_norm": 0.8887636661529541,
      "learning_rate": 0.00025498426690713966,
      "loss": 2.0981,
      "step": 126248
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.0113584995269775,
      "learning_rate": 0.0002549802226623393,
      "loss": 2.1726,
      "step": 126249
    },
    {
      "epoch": 1.64,
      "grad_norm": 0.9705295562744141,
      "learning_rate": 0.0002549761784259089,
      "loss": 2.2055,
      "step": 126250
    },
    {
      "epoch": 1.64,
      "grad_norm": 0.9258301258087158,
      "learning_rate": 0.0002549721341978493,
      "loss": 2.1702,
      "step": 126251
    },
    {
      "epoch": 1.64,
      "grad_norm": 0.9621323347091675,
      "learning_rate": 0.0002549680899781614,
      "loss": 2.0501,
      "step": 126252
    },
    {
      "epoch": 1.64,
      "grad_norm": 0.9715420603752136,
      "learning_rate": 0.00025496404576684566,
      "loss": 2.1638,
      "step": 126253
    },
    {
      "epoch": 1.64,
      "grad_norm": 0.883954644203186,
      "learning_rate": 0.0002549600015639029,
      "loss": 2.1976,
      "step": 126254
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.0390088558197021,
      "learning_rate": 0.000254955957369334,
      "loss": 2.366,
      "step": 126255
    },
    {
      "epoch": 1.64,
      "grad_norm": 0.9986119270324707,
      "learning_rate": 0.00025495191318313957,
      "loss": 2.2344,
      "step": 126256
    },
    {
      "epoch": 1.64,
      "grad_norm": 0.9718655943870544,
      "learning_rate": 0.00025494786900532045,
      "loss": 2.223,
      "step": 126257
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.056487798690796,
      "learning_rate": 0.0002549438248358775,
      "loss": 2.1711,
      "step": 126258
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.0711817741394043,
      "learning_rate": 0.00025493978067481123,
      "loss": 2.2265,
      "step": 126259
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.043385624885559,
      "learning_rate": 0.0002549357365221225,
      "loss": 2.1161,
      "step": 126260
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.0621187686920166,
      "learning_rate": 0.00025493169237781206,
      "loss": 2.0276,
      "step": 126261
    },
    {
      "epoch": 1.64,
      "grad_norm": 0.8921183943748474,
      "learning_rate": 0.00025492764824188065,
      "loss": 2.1892,
      "step": 126262
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.1535863876342773,
      "learning_rate": 0.00025492360411432904,
      "loss": 2.0548,
      "step": 126263
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.0727782249450684,
      "learning_rate": 0.00025491955999515814,
      "loss": 2.0904,
      "step": 126264
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.0750629901885986,
      "learning_rate": 0.00025491551588436846,
      "loss": 2.1206,
      "step": 126265
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.0801317691802979,
      "learning_rate": 0.00025491147178196075,
      "loss": 2.1058,
      "step": 126266
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.0304803848266602,
      "learning_rate": 0.0002549074276879359,
      "loss": 1.9293,
      "step": 126267
    },
    {
      "epoch": 1.64,
      "grad_norm": 0.9667545557022095,
      "learning_rate": 0.0002549033836022946,
      "loss": 2.0004,
      "step": 126268
    },
    {
      "epoch": 1.64,
      "grad_norm": 0.9882981777191162,
      "learning_rate": 0.0002548993395250376,
      "loss": 2.1463,
      "step": 126269
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.0378966331481934,
      "learning_rate": 0.0002548952954561657,
      "loss": 2.3154,
      "step": 126270
    },
    {
      "epoch": 1.64,
      "grad_norm": 0.9258156418800354,
      "learning_rate": 0.00025489125139567965,
      "loss": 1.9972,
      "step": 126271
    },
    {
      "epoch": 1.64,
      "grad_norm": 0.999257504940033,
      "learning_rate": 0.00025488720734358005,
      "loss": 2.0549,
      "step": 126272
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.2305951118469238,
      "learning_rate": 0.0002548831632998679,
      "loss": 2.3235,
      "step": 126273
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.0453804731369019,
      "learning_rate": 0.0002548791192645437,
      "loss": 2.138,
      "step": 126274
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.0419820547103882,
      "learning_rate": 0.0002548750752376084,
      "loss": 2.1034,
      "step": 126275
    },
    {
      "epoch": 1.64,
      "grad_norm": 0.9775575995445251,
      "learning_rate": 0.0002548710312190627,
      "loss": 2.0603,
      "step": 126276
    },
    {
      "epoch": 1.64,
      "grad_norm": 0.9651768207550049,
      "learning_rate": 0.0002548669872089073,
      "loss": 2.1073,
      "step": 126277
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.0583337545394897,
      "learning_rate": 0.00025486294320714294,
      "loss": 1.8987,
      "step": 126278
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.0632832050323486,
      "learning_rate": 0.00025485889921377045,
      "loss": 2.0286,
      "step": 126279
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.0203208923339844,
      "learning_rate": 0.00025485485522879056,
      "loss": 2.2911,
      "step": 126280
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.050950288772583,
      "learning_rate": 0.00025485081125220396,
      "loss": 2.1024,
      "step": 126281
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.035137414932251,
      "learning_rate": 0.00025484676728401153,
      "loss": 2.0542,
      "step": 126282
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.0825854539871216,
      "learning_rate": 0.0002548427233242138,
      "loss": 2.056,
      "step": 126283
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.009102463722229,
      "learning_rate": 0.00025483867937281177,
      "loss": 2.2039,
      "step": 126284
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.0748220682144165,
      "learning_rate": 0.000254834635429806,
      "loss": 2.1477,
      "step": 126285
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.1389049291610718,
      "learning_rate": 0.00025483059149519733,
      "loss": 2.073,
      "step": 126286
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.0719178915023804,
      "learning_rate": 0.0002548265475689866,
      "loss": 2.1193,
      "step": 126287
    },
    {
      "epoch": 1.64,
      "grad_norm": 0.9383201599121094,
      "learning_rate": 0.00025482250365117446,
      "loss": 2.107,
      "step": 126288
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.1537953615188599,
      "learning_rate": 0.0002548184597417616,
      "loss": 2.2426,
      "step": 126289
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.0041719675064087,
      "learning_rate": 0.00025481441584074884,
      "loss": 2.2309,
      "step": 126290
    },
    {
      "epoch": 1.64,
      "grad_norm": 0.8935158252716064,
      "learning_rate": 0.00025481037194813696,
      "loss": 2.0608,
      "step": 126291
    },
    {
      "epoch": 1.64,
      "grad_norm": 0.9194931983947754,
      "learning_rate": 0.00025480632806392666,
      "loss": 2.1357,
      "step": 126292
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.2341505289077759,
      "learning_rate": 0.0002548022841881187,
      "loss": 2.3076,
      "step": 126293
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.0693045854568481,
      "learning_rate": 0.000254798240320714,
      "loss": 1.8629,
      "step": 126294
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.0035688877105713,
      "learning_rate": 0.00025479419646171303,
      "loss": 1.9888,
      "step": 126295
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.0565088987350464,
      "learning_rate": 0.0002547901526111167,
      "loss": 1.8397,
      "step": 126296
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.0352860689163208,
      "learning_rate": 0.0002547861087689257,
      "loss": 1.8857,
      "step": 126297
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.0844998359680176,
      "learning_rate": 0.0002547820649351408,
      "loss": 2.0697,
      "step": 126298
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.0416280031204224,
      "learning_rate": 0.0002547780211097628,
      "loss": 1.9945,
      "step": 126299
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.0149848461151123,
      "learning_rate": 0.0002547739772927925,
      "loss": 1.9341,
      "step": 126300
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.0015449523925781,
      "learning_rate": 0.0002547699334842305,
      "loss": 2.0291,
      "step": 126301
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.0173944234848022,
      "learning_rate": 0.0002547658896840776,
      "loss": 2.0675,
      "step": 126302
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.0263071060180664,
      "learning_rate": 0.0002547618458923346,
      "loss": 2.2134,
      "step": 126303
    },
    {
      "epoch": 1.64,
      "grad_norm": 0.9254345893859863,
      "learning_rate": 0.00025475780210900214,
      "loss": 2.3429,
      "step": 126304
    },
    {
      "epoch": 1.64,
      "grad_norm": 0.8611230850219727,
      "learning_rate": 0.00025475375833408116,
      "loss": 1.9873,
      "step": 126305
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.089623212814331,
      "learning_rate": 0.0002547497145675724,
      "loss": 2.2124,
      "step": 126306
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.0501227378845215,
      "learning_rate": 0.0002547456708094764,
      "loss": 2.0304,
      "step": 126307
    },
    {
      "epoch": 1.64,
      "grad_norm": 0.9313608407974243,
      "learning_rate": 0.000254741627059794,
      "loss": 2.2284,
      "step": 126308
    },
    {
      "epoch": 1.64,
      "grad_norm": 0.9715550541877747,
      "learning_rate": 0.000254737583318526,
      "loss": 2.1924,
      "step": 126309
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.2297004461288452,
      "learning_rate": 0.0002547335395856732,
      "loss": 2.0526,
      "step": 126310
    },
    {
      "epoch": 1.64,
      "grad_norm": 0.9796100854873657,
      "learning_rate": 0.0002547294958612362,
      "loss": 2.1041,
      "step": 126311
    },
    {
      "epoch": 1.64,
      "grad_norm": 0.912323534488678,
      "learning_rate": 0.0002547254521452159,
      "loss": 2.0407,
      "step": 126312
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.0034881830215454,
      "learning_rate": 0.0002547214084376131,
      "loss": 2.1483,
      "step": 126313
    },
    {
      "epoch": 1.64,
      "grad_norm": 0.9407394528388977,
      "learning_rate": 0.0002547173647384283,
      "loss": 1.9264,
      "step": 126314
    },
    {
      "epoch": 1.64,
      "grad_norm": 0.9931085705757141,
      "learning_rate": 0.00025471332104766247,
      "loss": 2.2488,
      "step": 126315
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.0429672002792358,
      "learning_rate": 0.0002547092773653162,
      "loss": 2.0618,
      "step": 126316
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.0097007751464844,
      "learning_rate": 0.0002547052336913904,
      "loss": 2.1549,
      "step": 126317
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.0532664060592651,
      "learning_rate": 0.0002547011900258857,
      "loss": 2.0635,
      "step": 126318
    },
    {
      "epoch": 1.64,
      "grad_norm": 0.9979774355888367,
      "learning_rate": 0.00025469714636880306,
      "loss": 2.0047,
      "step": 126319
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.0229370594024658,
      "learning_rate": 0.0002546931027201429,
      "loss": 2.3297,
      "step": 126320
    },
    {
      "epoch": 1.64,
      "grad_norm": 0.9581450819969177,
      "learning_rate": 0.0002546890590799062,
      "loss": 2.0817,
      "step": 126321
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.0142004489898682,
      "learning_rate": 0.00025468501544809364,
      "loss": 2.0474,
      "step": 126322
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.0561063289642334,
      "learning_rate": 0.000254680971824706,
      "loss": 2.1957,
      "step": 126323
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.0486705303192139,
      "learning_rate": 0.000254676928209744,
      "loss": 2.4148,
      "step": 126324
    },
    {
      "epoch": 1.64,
      "grad_norm": 0.9710666537284851,
      "learning_rate": 0.00025467288460320856,
      "loss": 2.183,
      "step": 126325
    },
    {
      "epoch": 1.64,
      "grad_norm": 0.9414083361625671,
      "learning_rate": 0.0002546688410051002,
      "loss": 2.1319,
      "step": 126326
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.2091165781021118,
      "learning_rate": 0.0002546647974154197,
      "loss": 1.9814,
      "step": 126327
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.0035561323165894,
      "learning_rate": 0.00025466075383416786,
      "loss": 2.0339,
      "step": 126328
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.037700891494751,
      "learning_rate": 0.0002546567102613455,
      "loss": 1.9192,
      "step": 126329
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.2142539024353027,
      "learning_rate": 0.00025465266669695325,
      "loss": 2.2762,
      "step": 126330
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.0283657312393188,
      "learning_rate": 0.0002546486231409921,
      "loss": 1.8187,
      "step": 126331
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.0917251110076904,
      "learning_rate": 0.00025464457959346247,
      "loss": 2.1483,
      "step": 126332
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.0738914012908936,
      "learning_rate": 0.0002546405360543653,
      "loss": 2.3362,
      "step": 126333
    },
    {
      "epoch": 1.64,
      "grad_norm": 1.0173161029815674,
      "learning_rate": 0.0002546364925237013,
      "loss": 2.2732,
      "step": 126334
    },
    {
      "epoch": 1.64,
      "grad_norm": 0.9816778302192688,
      "learning_rate": 0.00025463244900147123,
      "loss": 1.9937,
      "step": 126335
    },
    {
      "epoch": 1.64,
      "grad_norm": 0.9919973015785217,
      "learning_rate": 0.00025462840548767584,
      "loss": 2.257,
      "step": 126336
    },
    {
      "epoch": 1.64,
      "grad_norm": 0.9923586845397949,
      "learning_rate": 0.000254624361982316,
      "loss": 2.1994,
      "step": 126337
    },
    {
      "epoch": 1.65,
      "grad_norm": 0.9507818222045898,
      "learning_rate": 0.00025462031848539225,
      "loss": 2.1888,
      "step": 126338
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.0849885940551758,
      "learning_rate": 0.0002546162749969054,
      "loss": 2.067,
      "step": 126339
    },
    {
      "epoch": 1.65,
      "grad_norm": 0.9306898713111877,
      "learning_rate": 0.0002546122315168563,
      "loss": 2.1502,
      "step": 126340
    },
    {
      "epoch": 1.65,
      "grad_norm": 0.9833277463912964,
      "learning_rate": 0.0002546081880452456,
      "loss": 2.2768,
      "step": 126341
    },
    {
      "epoch": 1.65,
      "grad_norm": 0.9973998069763184,
      "learning_rate": 0.0002546041445820741,
      "loss": 1.9252,
      "step": 126342
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.0514726638793945,
      "learning_rate": 0.0002546001011273427,
      "loss": 2.0436,
      "step": 126343
    },
    {
      "epoch": 1.65,
      "grad_norm": 0.9364862442016602,
      "learning_rate": 0.0002545960576810518,
      "loss": 2.0647,
      "step": 126344
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.0922222137451172,
      "learning_rate": 0.0002545920142432024,
      "loss": 2.0473,
      "step": 126345
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.0178338289260864,
      "learning_rate": 0.0002545879708137952,
      "loss": 2.1378,
      "step": 126346
    },
    {
      "epoch": 1.65,
      "grad_norm": 0.8966709971427917,
      "learning_rate": 0.00025458392739283094,
      "loss": 2.2742,
      "step": 126347
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.0050363540649414,
      "learning_rate": 0.0002545798839803104,
      "loss": 2.1594,
      "step": 126348
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.0220052003860474,
      "learning_rate": 0.00025457584057623437,
      "loss": 2.1953,
      "step": 126349
    },
    {
      "epoch": 1.65,
      "grad_norm": 0.9146763682365417,
      "learning_rate": 0.0002545717971806036,
      "loss": 2.1293,
      "step": 126350
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.0647777318954468,
      "learning_rate": 0.00025456775379341867,
      "loss": 2.0619,
      "step": 126351
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.0504416227340698,
      "learning_rate": 0.00025456371041468045,
      "loss": 2.0327,
      "step": 126352
    },
    {
      "epoch": 1.65,
      "grad_norm": 0.9078828692436218,
      "learning_rate": 0.0002545596670443897,
      "loss": 2.0875,
      "step": 126353
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.053270697593689,
      "learning_rate": 0.0002545556236825472,
      "loss": 1.9643,
      "step": 126354
    },
    {
      "epoch": 1.65,
      "grad_norm": 0.9440453052520752,
      "learning_rate": 0.00025455158032915365,
      "loss": 2.2645,
      "step": 126355
    },
    {
      "epoch": 1.65,
      "grad_norm": 0.9976332187652588,
      "learning_rate": 0.00025454753698420985,
      "loss": 2.2601,
      "step": 126356
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.038393497467041,
      "learning_rate": 0.00025454349364771654,
      "loss": 2.0457,
      "step": 126357
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.1059694290161133,
      "learning_rate": 0.0002545394503196744,
      "loss": 2.0924,
      "step": 126358
    },
    {
      "epoch": 1.65,
      "grad_norm": 0.9508679509162903,
      "learning_rate": 0.0002545354070000842,
      "loss": 2.3107,
      "step": 126359
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.007481336593628,
      "learning_rate": 0.00025453136368894675,
      "loss": 2.0818,
      "step": 126360
    },
    {
      "epoch": 1.65,
      "grad_norm": 0.9957504272460938,
      "learning_rate": 0.0002545273203862629,
      "loss": 2.0405,
      "step": 126361
    },
    {
      "epoch": 1.65,
      "grad_norm": 0.9811916351318359,
      "learning_rate": 0.0002545232770920331,
      "loss": 2.0079,
      "step": 126362
    },
    {
      "epoch": 1.65,
      "grad_norm": 0.9023503065109253,
      "learning_rate": 0.00025451923380625835,
      "loss": 2.3794,
      "step": 126363
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.200576901435852,
      "learning_rate": 0.00025451519052893944,
      "loss": 2.2491,
      "step": 126364
    },
    {
      "epoch": 1.65,
      "grad_norm": 0.9767888188362122,
      "learning_rate": 0.00025451114726007687,
      "loss": 2.2478,
      "step": 126365
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.0357176065444946,
      "learning_rate": 0.00025450710399967157,
      "loss": 2.341,
      "step": 126366
    },
    {
      "epoch": 1.65,
      "grad_norm": 0.9719451069831848,
      "learning_rate": 0.00025450306074772434,
      "loss": 2.1976,
      "step": 126367
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.1028807163238525,
      "learning_rate": 0.00025449901750423574,
      "loss": 1.8787,
      "step": 126368
    },
    {
      "epoch": 1.65,
      "grad_norm": 0.9437698125839233,
      "learning_rate": 0.00025449497426920667,
      "loss": 2.0578,
      "step": 126369
    },
    {
      "epoch": 1.65,
      "grad_norm": 0.9418835043907166,
      "learning_rate": 0.0002544909310426378,
      "loss": 2.1769,
      "step": 126370
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.0074362754821777,
      "learning_rate": 0.00025448688782453,
      "loss": 1.944,
      "step": 126371
    },
    {
      "epoch": 1.65,
      "grad_norm": 0.93486487865448,
      "learning_rate": 0.00025448284461488394,
      "loss": 2.0539,
      "step": 126372
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.1103249788284302,
      "learning_rate": 0.00025447880141370036,
      "loss": 2.106,
      "step": 126373
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.2651774883270264,
      "learning_rate": 0.00025447475822098006,
      "loss": 2.293,
      "step": 126374
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.0108039379119873,
      "learning_rate": 0.0002544707150367237,
      "loss": 2.0631,
      "step": 126375
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.16398286819458,
      "learning_rate": 0.00025446667186093214,
      "loss": 2.0429,
      "step": 126376
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.046988844871521,
      "learning_rate": 0.000254462628693606,
      "loss": 2.0324,
      "step": 126377
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.2834193706512451,
      "learning_rate": 0.00025445858553474617,
      "loss": 2.1426,
      "step": 126378
    },
    {
      "epoch": 1.65,
      "grad_norm": 0.9559017419815063,
      "learning_rate": 0.0002544545423843534,
      "loss": 2.2578,
      "step": 126379
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.2307710647583008,
      "learning_rate": 0.0002544504992424284,
      "loss": 2.1732,
      "step": 126380
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.0250903367996216,
      "learning_rate": 0.00025444645610897185,
      "loss": 2.1488,
      "step": 126381
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.0043444633483887,
      "learning_rate": 0.0002544424129839846,
      "loss": 2.1262,
      "step": 126382
    },
    {
      "epoch": 1.65,
      "grad_norm": 0.9810274839401245,
      "learning_rate": 0.0002544383698674673,
      "loss": 2.2728,
      "step": 126383
    },
    {
      "epoch": 1.65,
      "grad_norm": 0.9310091733932495,
      "learning_rate": 0.0002544343267594208,
      "loss": 2.1124,
      "step": 126384
    },
    {
      "epoch": 1.65,
      "grad_norm": 0.912815511226654,
      "learning_rate": 0.0002544302836598458,
      "loss": 2.1325,
      "step": 126385
    },
    {
      "epoch": 1.65,
      "grad_norm": 0.983035683631897,
      "learning_rate": 0.00025442624056874324,
      "loss": 2.241,
      "step": 126386
    },
    {
      "epoch": 1.65,
      "grad_norm": 0.9947910904884338,
      "learning_rate": 0.00025442219748611353,
      "loss": 2.0417,
      "step": 126387
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.1547458171844482,
      "learning_rate": 0.00025441815441195763,
      "loss": 2.3718,
      "step": 126388
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.181426763534546,
      "learning_rate": 0.00025441411134627623,
      "loss": 2.1728,
      "step": 126389
    },
    {
      "epoch": 1.65,
      "grad_norm": 0.9477432370185852,
      "learning_rate": 0.0002544100682890701,
      "loss": 1.9441,
      "step": 126390
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.0214669704437256,
      "learning_rate": 0.00025440602524033997,
      "loss": 2.1667,
      "step": 126391
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.0338526964187622,
      "learning_rate": 0.00025440198220008685,
      "loss": 2.1879,
      "step": 126392
    },
    {
      "epoch": 1.65,
      "grad_norm": 0.9482621550559998,
      "learning_rate": 0.00025439793916831106,
      "loss": 2.1448,
      "step": 126393
    },
    {
      "epoch": 1.65,
      "grad_norm": 0.9691778421401978,
      "learning_rate": 0.00025439389614501356,
      "loss": 2.0733,
      "step": 126394
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.0375075340270996,
      "learning_rate": 0.0002543898531301951,
      "loss": 2.0175,
      "step": 126395
    },
    {
      "epoch": 1.65,
      "grad_norm": 0.9154396653175354,
      "learning_rate": 0.00025438581012385644,
      "loss": 2.0537,
      "step": 126396
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.0208925008773804,
      "learning_rate": 0.0002543817671259983,
      "loss": 2.2112,
      "step": 126397
    },
    {
      "epoch": 1.65,
      "grad_norm": 0.909351646900177,
      "learning_rate": 0.00025437772413662156,
      "loss": 1.9593,
      "step": 126398
    },
    {
      "epoch": 1.65,
      "grad_norm": 0.9658040404319763,
      "learning_rate": 0.0002543736811557268,
      "loss": 2.1501,
      "step": 126399
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.0837749242782593,
      "learning_rate": 0.0002543696381833148,
      "loss": 2.2059,
      "step": 126400
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.0047439336776733,
      "learning_rate": 0.00025436559521938636,
      "loss": 2.1469,
      "step": 126401
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.0625251531600952,
      "learning_rate": 0.0002543615522639422,
      "loss": 2.2905,
      "step": 126402
    },
    {
      "epoch": 1.65,
      "grad_norm": 0.9238961338996887,
      "learning_rate": 0.0002543575093169831,
      "loss": 2.2798,
      "step": 126403
    },
    {
      "epoch": 1.65,
      "grad_norm": 0.8500892519950867,
      "learning_rate": 0.0002543534663785099,
      "loss": 2.056,
      "step": 126404
    },
    {
      "epoch": 1.65,
      "grad_norm": 0.9712039232254028,
      "learning_rate": 0.0002543494234485231,
      "loss": 2.2323,
      "step": 126405
    },
    {
      "epoch": 1.65,
      "grad_norm": 0.8868255615234375,
      "learning_rate": 0.0002543453805270237,
      "loss": 1.9908,
      "step": 126406
    },
    {
      "epoch": 1.65,
      "grad_norm": 0.923356294631958,
      "learning_rate": 0.00025434133761401226,
      "loss": 2.1801,
      "step": 126407
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.046370506286621,
      "learning_rate": 0.00025433729470948966,
      "loss": 2.1608,
      "step": 126408
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.0216363668441772,
      "learning_rate": 0.0002543332518134567,
      "loss": 2.1634,
      "step": 126409
    },
    {
      "epoch": 1.65,
      "grad_norm": 0.9442934393882751,
      "learning_rate": 0.00025432920892591403,
      "loss": 1.88,
      "step": 126410
    },
    {
      "epoch": 1.65,
      "grad_norm": 0.9310997128486633,
      "learning_rate": 0.00025432516604686236,
      "loss": 2.1005,
      "step": 126411
    },
    {
      "epoch": 1.65,
      "grad_norm": 0.9236121773719788,
      "learning_rate": 0.00025432112317630255,
      "loss": 1.9862,
      "step": 126412
    },
    {
      "epoch": 1.65,
      "grad_norm": 0.9895135760307312,
      "learning_rate": 0.00025431708031423524,
      "loss": 2.3142,
      "step": 126413
    },
    {
      "epoch": 1.65,
      "grad_norm": 0.9504048824310303,
      "learning_rate": 0.0002543130374606613,
      "loss": 1.9649,
      "step": 126414
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.0329129695892334,
      "learning_rate": 0.0002543089946155814,
      "loss": 2.1991,
      "step": 126415
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.3087049722671509,
      "learning_rate": 0.00025430495177899645,
      "loss": 2.2141,
      "step": 126416
    },
    {
      "epoch": 1.65,
      "grad_norm": 0.9726269245147705,
      "learning_rate": 0.00025430090895090695,
      "loss": 2.1703,
      "step": 126417
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.0698962211608887,
      "learning_rate": 0.0002542968661313138,
      "loss": 2.2784,
      "step": 126418
    },
    {
      "epoch": 1.65,
      "grad_norm": 0.9917749166488647,
      "learning_rate": 0.00025429282332021765,
      "loss": 2.111,
      "step": 126419
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.0158390998840332,
      "learning_rate": 0.0002542887805176194,
      "loss": 2.2092,
      "step": 126420
    },
    {
      "epoch": 1.65,
      "grad_norm": 0.9776486754417419,
      "learning_rate": 0.00025428473772351966,
      "loss": 2.255,
      "step": 126421
    },
    {
      "epoch": 1.65,
      "grad_norm": 0.9803213477134705,
      "learning_rate": 0.00025428069493791945,
      "loss": 2.0621,
      "step": 126422
    },
    {
      "epoch": 1.65,
      "grad_norm": 0.913142740726471,
      "learning_rate": 0.00025427665216081917,
      "loss": 2.3362,
      "step": 126423
    },
    {
      "epoch": 1.65,
      "grad_norm": 0.992977499961853,
      "learning_rate": 0.00025427260939221973,
      "loss": 2.2016,
      "step": 126424
    },
    {
      "epoch": 1.65,
      "grad_norm": 0.9790275692939758,
      "learning_rate": 0.00025426856663212185,
      "loss": 2.0056,
      "step": 126425
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.0298398733139038,
      "learning_rate": 0.0002542645238805263,
      "loss": 2.1478,
      "step": 126426
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.1593596935272217,
      "learning_rate": 0.0002542604811374339,
      "loss": 2.1802,
      "step": 126427
    },
    {
      "epoch": 1.65,
      "grad_norm": 0.8981645107269287,
      "learning_rate": 0.0002542564384028454,
      "loss": 2.4077,
      "step": 126428
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.0887778997421265,
      "learning_rate": 0.0002542523956767614,
      "loss": 1.9247,
      "step": 126429
    },
    {
      "epoch": 1.65,
      "grad_norm": 0.8977687358856201,
      "learning_rate": 0.0002542483529591827,
      "loss": 2.1646,
      "step": 126430
    },
    {
      "epoch": 1.65,
      "grad_norm": 0.9977355599403381,
      "learning_rate": 0.00025424431025011015,
      "loss": 2.3016,
      "step": 126431
    },
    {
      "epoch": 1.65,
      "grad_norm": 0.9652391672134399,
      "learning_rate": 0.0002542402675495444,
      "loss": 2.182,
      "step": 126432
    },
    {
      "epoch": 1.65,
      "grad_norm": 0.9899564385414124,
      "learning_rate": 0.0002542362248574863,
      "loss": 2.2121,
      "step": 126433
    },
    {
      "epoch": 1.65,
      "grad_norm": 0.9207438826560974,
      "learning_rate": 0.0002542321821739366,
      "loss": 2.035,
      "step": 126434
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.024145245552063,
      "learning_rate": 0.00025422813949889595,
      "loss": 2.2913,
      "step": 126435
    },
    {
      "epoch": 1.65,
      "grad_norm": 0.9475383162498474,
      "learning_rate": 0.0002542240968323651,
      "loss": 2.0103,
      "step": 126436
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.0197227001190186,
      "learning_rate": 0.0002542200541743449,
      "loss": 2.0911,
      "step": 126437
    },
    {
      "epoch": 1.65,
      "grad_norm": 0.9744477868080139,
      "learning_rate": 0.000254216011524836,
      "loss": 2.0227,
      "step": 126438
    },
    {
      "epoch": 1.65,
      "grad_norm": 0.9321902394294739,
      "learning_rate": 0.00025421196888383927,
      "loss": 2.1039,
      "step": 126439
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.0085811614990234,
      "learning_rate": 0.00025420792625135546,
      "loss": 2.1984,
      "step": 126440
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.036423921585083,
      "learning_rate": 0.0002542038836273852,
      "loss": 2.0281,
      "step": 126441
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.0710116624832153,
      "learning_rate": 0.0002541998410119293,
      "loss": 2.0416,
      "step": 126442
    },
    {
      "epoch": 1.65,
      "grad_norm": 0.924896240234375,
      "learning_rate": 0.0002541957984049885,
      "loss": 2.1589,
      "step": 126443
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.0188367366790771,
      "learning_rate": 0.00025419175580656357,
      "loss": 2.1426,
      "step": 126444
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.0316365957260132,
      "learning_rate": 0.00025418771321665523,
      "loss": 2.331,
      "step": 126445
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.085370659828186,
      "learning_rate": 0.0002541836706352643,
      "loss": 2.0133,
      "step": 126446
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.2583261728286743,
      "learning_rate": 0.0002541796280623915,
      "loss": 2.0454,
      "step": 126447
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.0514419078826904,
      "learning_rate": 0.0002541755854980375,
      "loss": 2.1604,
      "step": 126448
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.020102858543396,
      "learning_rate": 0.0002541715429422032,
      "loss": 2.1442,
      "step": 126449
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.0353127717971802,
      "learning_rate": 0.0002541675003948893,
      "loss": 2.0931,
      "step": 126450
    },
    {
      "epoch": 1.65,
      "grad_norm": 0.9532780647277832,
      "learning_rate": 0.0002541634578560964,
      "loss": 2.2709,
      "step": 126451
    },
    {
      "epoch": 1.65,
      "grad_norm": 0.9553599953651428,
      "learning_rate": 0.0002541594153258255,
      "loss": 2.1325,
      "step": 126452
    },
    {
      "epoch": 1.65,
      "grad_norm": 0.9376435279846191,
      "learning_rate": 0.0002541553728040772,
      "loss": 2.1059,
      "step": 126453
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.1096612215042114,
      "learning_rate": 0.00025415133029085227,
      "loss": 2.2343,
      "step": 126454
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.0140877962112427,
      "learning_rate": 0.00025414728778615144,
      "loss": 2.1878,
      "step": 126455
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.017359733581543,
      "learning_rate": 0.0002541432452899756,
      "loss": 2.0698,
      "step": 126456
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.0188294649124146,
      "learning_rate": 0.0002541392028023253,
      "loss": 1.9246,
      "step": 126457
    },
    {
      "epoch": 1.65,
      "grad_norm": 0.9714002013206482,
      "learning_rate": 0.0002541351603232014,
      "loss": 2.0646,
      "step": 126458
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.0186740159988403,
      "learning_rate": 0.0002541311178526046,
      "loss": 2.2271,
      "step": 126459
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.0797396898269653,
      "learning_rate": 0.0002541270753905358,
      "loss": 1.9099,
      "step": 126460
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.0815283060073853,
      "learning_rate": 0.00025412303293699556,
      "loss": 2.0153,
      "step": 126461
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.0337305068969727,
      "learning_rate": 0.0002541189904919847,
      "loss": 1.9241,
      "step": 126462
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.043381690979004,
      "learning_rate": 0.00025411494805550406,
      "loss": 2.138,
      "step": 126463
    },
    {
      "epoch": 1.65,
      "grad_norm": 0.9463043808937073,
      "learning_rate": 0.0002541109056275543,
      "loss": 2.0775,
      "step": 126464
    },
    {
      "epoch": 1.65,
      "grad_norm": 0.9803019762039185,
      "learning_rate": 0.0002541068632081362,
      "loss": 2.151,
      "step": 126465
    },
    {
      "epoch": 1.65,
      "grad_norm": 0.9751936197280884,
      "learning_rate": 0.00025410282079725045,
      "loss": 2.0581,
      "step": 126466
    },
    {
      "epoch": 1.65,
      "grad_norm": 0.8930752277374268,
      "learning_rate": 0.0002540987783948979,
      "loss": 2.177,
      "step": 126467
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.0351192951202393,
      "learning_rate": 0.00025409473600107916,
      "loss": 2.1756,
      "step": 126468
    },
    {
      "epoch": 1.65,
      "grad_norm": 0.9388913512229919,
      "learning_rate": 0.0002540906936157951,
      "loss": 2.1281,
      "step": 126469
    },
    {
      "epoch": 1.65,
      "grad_norm": 0.9883289933204651,
      "learning_rate": 0.0002540866512390465,
      "loss": 2.1103,
      "step": 126470
    },
    {
      "epoch": 1.65,
      "grad_norm": 0.9445117115974426,
      "learning_rate": 0.00025408260887083416,
      "loss": 2.1726,
      "step": 126471
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.0544461011886597,
      "learning_rate": 0.0002540785665111586,
      "loss": 2.1056,
      "step": 126472
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.0094693899154663,
      "learning_rate": 0.0002540745241600207,
      "loss": 2.0132,
      "step": 126473
    },
    {
      "epoch": 1.65,
      "grad_norm": 0.9731041193008423,
      "learning_rate": 0.0002540704818174212,
      "loss": 2.4234,
      "step": 126474
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.0121064186096191,
      "learning_rate": 0.00025406643948336084,
      "loss": 2.2434,
      "step": 126475
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.0621955394744873,
      "learning_rate": 0.00025406239715784046,
      "loss": 1.9087,
      "step": 126476
    },
    {
      "epoch": 1.65,
      "grad_norm": 0.9874841570854187,
      "learning_rate": 0.00025405835484086084,
      "loss": 2.1852,
      "step": 126477
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.0443774461746216,
      "learning_rate": 0.00025405431253242255,
      "loss": 1.8844,
      "step": 126478
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.0549302101135254,
      "learning_rate": 0.0002540502702325264,
      "loss": 2.2077,
      "step": 126479
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.1034258604049683,
      "learning_rate": 0.00025404622794117317,
      "loss": 2.0571,
      "step": 126480
    },
    {
      "epoch": 1.65,
      "grad_norm": 0.9519433379173279,
      "learning_rate": 0.0002540421856583636,
      "loss": 2.2164,
      "step": 126481
    },
    {
      "epoch": 1.65,
      "grad_norm": 0.9872221946716309,
      "learning_rate": 0.0002540381433840985,
      "loss": 2.0506,
      "step": 126482
    },
    {
      "epoch": 1.65,
      "grad_norm": 0.954483151435852,
      "learning_rate": 0.00025403410111837863,
      "loss": 2.479,
      "step": 126483
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.1552376747131348,
      "learning_rate": 0.0002540300588612046,
      "loss": 2.0978,
      "step": 126484
    },
    {
      "epoch": 1.65,
      "grad_norm": 0.928433358669281,
      "learning_rate": 0.0002540260166125773,
      "loss": 2.2711,
      "step": 126485
    },
    {
      "epoch": 1.65,
      "grad_norm": 0.9775840640068054,
      "learning_rate": 0.00025402197437249735,
      "loss": 2.1896,
      "step": 126486
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.0706532001495361,
      "learning_rate": 0.0002540179321409656,
      "loss": 2.1792,
      "step": 126487
    },
    {
      "epoch": 1.65,
      "grad_norm": 0.9651429653167725,
      "learning_rate": 0.0002540138899179828,
      "loss": 2.1078,
      "step": 126488
    },
    {
      "epoch": 1.65,
      "grad_norm": 0.9157954454421997,
      "learning_rate": 0.00025400984770354985,
      "loss": 2.1282,
      "step": 126489
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.0110176801681519,
      "learning_rate": 0.0002540058054976672,
      "loss": 2.1459,
      "step": 126490
    },
    {
      "epoch": 1.65,
      "grad_norm": 0.9387082457542419,
      "learning_rate": 0.0002540017633003357,
      "loss": 2.2367,
      "step": 126491
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.0728307962417603,
      "learning_rate": 0.0002539977211115561,
      "loss": 1.9928,
      "step": 126492
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.0509772300720215,
      "learning_rate": 0.0002539936789313293,
      "loss": 1.9833,
      "step": 126493
    },
    {
      "epoch": 1.65,
      "grad_norm": 0.9933574199676514,
      "learning_rate": 0.0002539896367596559,
      "loss": 2.3032,
      "step": 126494
    },
    {
      "epoch": 1.65,
      "grad_norm": 0.9776539206504822,
      "learning_rate": 0.0002539855945965368,
      "loss": 2.1995,
      "step": 126495
    },
    {
      "epoch": 1.65,
      "grad_norm": 0.9633274674415588,
      "learning_rate": 0.0002539815524419725,
      "loss": 2.2374,
      "step": 126496
    },
    {
      "epoch": 1.65,
      "grad_norm": 0.9522407650947571,
      "learning_rate": 0.00025397751029596397,
      "loss": 2.1048,
      "step": 126497
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.0367802381515503,
      "learning_rate": 0.00025397346815851185,
      "loss": 1.9536,
      "step": 126498
    },
    {
      "epoch": 1.65,
      "grad_norm": 0.937722384929657,
      "learning_rate": 0.0002539694260296169,
      "loss": 2.1278,
      "step": 126499
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.024949073791504,
      "learning_rate": 0.00025396538390927995,
      "loss": 2.0354,
      "step": 126500
    },
    {
      "epoch": 1.65,
      "grad_norm": 0.9806972146034241,
      "learning_rate": 0.00025396134179750184,
      "loss": 2.2765,
      "step": 126501
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.09782075881958,
      "learning_rate": 0.00025395729969428303,
      "loss": 2.0334,
      "step": 126502
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.0448487997055054,
      "learning_rate": 0.0002539532575996244,
      "loss": 2.3303,
      "step": 126503
    },
    {
      "epoch": 1.65,
      "grad_norm": 0.9947554469108582,
      "learning_rate": 0.0002539492155135268,
      "loss": 2.2637,
      "step": 126504
    },
    {
      "epoch": 1.65,
      "grad_norm": 0.9724382758140564,
      "learning_rate": 0.0002539451734359909,
      "loss": 1.8637,
      "step": 126505
    },
    {
      "epoch": 1.65,
      "grad_norm": 0.9622994065284729,
      "learning_rate": 0.0002539411313670174,
      "loss": 2.0292,
      "step": 126506
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.0858092308044434,
      "learning_rate": 0.0002539370893066072,
      "loss": 2.2079,
      "step": 126507
    },
    {
      "epoch": 1.65,
      "grad_norm": 0.9912444949150085,
      "learning_rate": 0.000253933047254761,
      "loss": 2.1053,
      "step": 126508
    },
    {
      "epoch": 1.65,
      "grad_norm": 0.9066958427429199,
      "learning_rate": 0.0002539290052114794,
      "loss": 2.0537,
      "step": 126509
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.0635734796524048,
      "learning_rate": 0.00025392496317676327,
      "loss": 2.3681,
      "step": 126510
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.017356514930725,
      "learning_rate": 0.0002539209211506134,
      "loss": 2.1526,
      "step": 126511
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.013170838356018,
      "learning_rate": 0.00025391687913303045,
      "loss": 2.1044,
      "step": 126512
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.002147912979126,
      "learning_rate": 0.00025391283712401525,
      "loss": 2.3388,
      "step": 126513
    },
    {
      "epoch": 1.65,
      "grad_norm": 0.9078319072723389,
      "learning_rate": 0.0002539087951235687,
      "loss": 2.1102,
      "step": 126514
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.0378934144973755,
      "learning_rate": 0.0002539047531316912,
      "loss": 1.8756,
      "step": 126515
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.0199767351150513,
      "learning_rate": 0.00025390071114838366,
      "loss": 2.1579,
      "step": 126516
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.1305012702941895,
      "learning_rate": 0.00025389666917364687,
      "loss": 2.125,
      "step": 126517
    },
    {
      "epoch": 1.65,
      "grad_norm": 0.9715861082077026,
      "learning_rate": 0.00025389262720748156,
      "loss": 1.9463,
      "step": 126518
    },
    {
      "epoch": 1.65,
      "grad_norm": 0.9517166614532471,
      "learning_rate": 0.00025388858524988845,
      "loss": 2.0058,
      "step": 126519
    },
    {
      "epoch": 1.65,
      "grad_norm": 0.9732142686843872,
      "learning_rate": 0.0002538845433008685,
      "loss": 1.9498,
      "step": 126520
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.1169875860214233,
      "learning_rate": 0.00025388050136042215,
      "loss": 2.3596,
      "step": 126521
    },
    {
      "epoch": 1.65,
      "grad_norm": 0.9870112538337708,
      "learning_rate": 0.00025387645942855026,
      "loss": 2.0788,
      "step": 126522
    },
    {
      "epoch": 1.65,
      "grad_norm": 0.9111523628234863,
      "learning_rate": 0.0002538724175052536,
      "loss": 1.9536,
      "step": 126523
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.0103816986083984,
      "learning_rate": 0.00025386837559053293,
      "loss": 2.1364,
      "step": 126524
    },
    {
      "epoch": 1.65,
      "grad_norm": 0.9619134068489075,
      "learning_rate": 0.00025386433368438906,
      "loss": 2.1245,
      "step": 126525
    },
    {
      "epoch": 1.65,
      "grad_norm": 0.9613668918609619,
      "learning_rate": 0.00025386029178682274,
      "loss": 2.2615,
      "step": 126526
    },
    {
      "epoch": 1.65,
      "grad_norm": 0.9061192274093628,
      "learning_rate": 0.0002538562498978346,
      "loss": 2.1293,
      "step": 126527
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.079818606376648,
      "learning_rate": 0.0002538522080174254,
      "loss": 2.304,
      "step": 126528
    },
    {
      "epoch": 1.65,
      "grad_norm": 0.9124727249145508,
      "learning_rate": 0.0002538481661455959,
      "loss": 2.2594,
      "step": 126529
    },
    {
      "epoch": 1.65,
      "grad_norm": 0.9693796038627625,
      "learning_rate": 0.00025384412428234703,
      "loss": 1.9456,
      "step": 126530
    },
    {
      "epoch": 1.65,
      "grad_norm": 0.9725202918052673,
      "learning_rate": 0.0002538400824276793,
      "loss": 2.2383,
      "step": 126531
    },
    {
      "epoch": 1.65,
      "grad_norm": 0.9878870248794556,
      "learning_rate": 0.00025383604058159365,
      "loss": 2.3189,
      "step": 126532
    },
    {
      "epoch": 1.65,
      "grad_norm": 0.9269282221794128,
      "learning_rate": 0.00025383199874409073,
      "loss": 2.1026,
      "step": 126533
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.0911141633987427,
      "learning_rate": 0.0002538279569151713,
      "loss": 2.0483,
      "step": 126534
    },
    {
      "epoch": 1.65,
      "grad_norm": 0.9861074686050415,
      "learning_rate": 0.00025382391509483614,
      "loss": 2.1186,
      "step": 126535
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.0113584995269775,
      "learning_rate": 0.000253819873283086,
      "loss": 2.2742,
      "step": 126536
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.0357328653335571,
      "learning_rate": 0.00025381583147992154,
      "loss": 2.03,
      "step": 126537
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.1716797351837158,
      "learning_rate": 0.0002538117896853437,
      "loss": 1.9637,
      "step": 126538
    },
    {
      "epoch": 1.65,
      "grad_norm": 0.9949354529380798,
      "learning_rate": 0.000253807747899353,
      "loss": 2.0593,
      "step": 126539
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.1321015357971191,
      "learning_rate": 0.0002538037061219504,
      "loss": 2.2727,
      "step": 126540
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.0494743585586548,
      "learning_rate": 0.00025379966435313655,
      "loss": 2.237,
      "step": 126541
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.2665575742721558,
      "learning_rate": 0.00025379562259291225,
      "loss": 1.9691,
      "step": 126542
    },
    {
      "epoch": 1.65,
      "grad_norm": 0.9283987283706665,
      "learning_rate": 0.00025379158084127813,
      "loss": 2.1582,
      "step": 126543
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.0289323329925537,
      "learning_rate": 0.00025378753909823513,
      "loss": 2.1655,
      "step": 126544
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.0806502103805542,
      "learning_rate": 0.0002537834973637838,
      "loss": 2.2211,
      "step": 126545
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.0936787128448486,
      "learning_rate": 0.000253779455637925,
      "loss": 2.0752,
      "step": 126546
    },
    {
      "epoch": 1.65,
      "grad_norm": 0.9520255923271179,
      "learning_rate": 0.00025377541392065944,
      "loss": 1.9522,
      "step": 126547
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.0643094778060913,
      "learning_rate": 0.00025377137221198803,
      "loss": 2.1939,
      "step": 126548
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.0506683588027954,
      "learning_rate": 0.0002537673305119113,
      "loss": 2.1278,
      "step": 126549
    },
    {
      "epoch": 1.65,
      "grad_norm": 0.9667062759399414,
      "learning_rate": 0.0002537632888204302,
      "loss": 2.1234,
      "step": 126550
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.051293969154358,
      "learning_rate": 0.0002537592471375453,
      "loss": 2.3508,
      "step": 126551
    },
    {
      "epoch": 1.65,
      "grad_norm": 0.9523661732673645,
      "learning_rate": 0.0002537552054632574,
      "loss": 1.9397,
      "step": 126552
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.1234506368637085,
      "learning_rate": 0.0002537511637975673,
      "loss": 1.9847,
      "step": 126553
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.0396777391433716,
      "learning_rate": 0.00025374712214047565,
      "loss": 2.0875,
      "step": 126554
    },
    {
      "epoch": 1.65,
      "grad_norm": 0.9936227202415466,
      "learning_rate": 0.00025374308049198337,
      "loss": 2.1122,
      "step": 126555
    },
    {
      "epoch": 1.65,
      "grad_norm": 0.911841869354248,
      "learning_rate": 0.00025373903885209125,
      "loss": 2.2083,
      "step": 126556
    },
    {
      "epoch": 1.65,
      "grad_norm": 0.9170551896095276,
      "learning_rate": 0.0002537349972207998,
      "loss": 2.143,
      "step": 126557
    },
    {
      "epoch": 1.65,
      "grad_norm": 0.9365208745002747,
      "learning_rate": 0.00025373095559810984,
      "loss": 1.8349,
      "step": 126558
    },
    {
      "epoch": 1.65,
      "grad_norm": 0.9824990034103394,
      "learning_rate": 0.00025372691398402216,
      "loss": 2.0755,
      "step": 126559
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.0462592840194702,
      "learning_rate": 0.0002537228723785376,
      "loss": 2.2467,
      "step": 126560
    },
    {
      "epoch": 1.65,
      "grad_norm": 0.955451250076294,
      "learning_rate": 0.00025371883078165674,
      "loss": 2.1963,
      "step": 126561
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.093211054801941,
      "learning_rate": 0.0002537147891933806,
      "loss": 2.1496,
      "step": 126562
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.0015299320220947,
      "learning_rate": 0.00025371074761370964,
      "loss": 2.2154,
      "step": 126563
    },
    {
      "epoch": 1.65,
      "grad_norm": 0.9058078527450562,
      "learning_rate": 0.0002537067060426447,
      "loss": 2.1522,
      "step": 126564
    },
    {
      "epoch": 1.65,
      "grad_norm": 0.9306281208992004,
      "learning_rate": 0.00025370266448018655,
      "loss": 2.181,
      "step": 126565
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.134833812713623,
      "learning_rate": 0.00025369862292633596,
      "loss": 2.2258,
      "step": 126566
    },
    {
      "epoch": 1.65,
      "grad_norm": 0.9824747443199158,
      "learning_rate": 0.00025369458138109366,
      "loss": 2.067,
      "step": 126567
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.0063191652297974,
      "learning_rate": 0.0002536905398444606,
      "loss": 2.3007,
      "step": 126568
    },
    {
      "epoch": 1.65,
      "grad_norm": 0.8704935312271118,
      "learning_rate": 0.0002536864983164371,
      "loss": 1.8593,
      "step": 126569
    },
    {
      "epoch": 1.65,
      "grad_norm": 0.9841236472129822,
      "learning_rate": 0.0002536824567970242,
      "loss": 2.2306,
      "step": 126570
    },
    {
      "epoch": 1.65,
      "grad_norm": 0.9107732772827148,
      "learning_rate": 0.0002536784152862227,
      "loss": 2.3174,
      "step": 126571
    },
    {
      "epoch": 1.65,
      "grad_norm": 0.9728716015815735,
      "learning_rate": 0.00025367437378403314,
      "loss": 2.1879,
      "step": 126572
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.0134694576263428,
      "learning_rate": 0.00025367033229045644,
      "loss": 2.1447,
      "step": 126573
    },
    {
      "epoch": 1.65,
      "grad_norm": 0.9884567856788635,
      "learning_rate": 0.0002536662908054934,
      "loss": 2.1444,
      "step": 126574
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.0222115516662598,
      "learning_rate": 0.00025366224932914455,
      "loss": 2.1088,
      "step": 126575
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.1378076076507568,
      "learning_rate": 0.0002536582078614108,
      "loss": 2.1386,
      "step": 126576
    },
    {
      "epoch": 1.65,
      "grad_norm": 0.9622939825057983,
      "learning_rate": 0.00025365416640229285,
      "loss": 2.2679,
      "step": 126577
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.0569339990615845,
      "learning_rate": 0.00025365012495179144,
      "loss": 2.0797,
      "step": 126578
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.1201201677322388,
      "learning_rate": 0.0002536460835099073,
      "loss": 2.1106,
      "step": 126579
    },
    {
      "epoch": 1.65,
      "grad_norm": 0.9981350302696228,
      "learning_rate": 0.0002536420420766413,
      "loss": 2.0242,
      "step": 126580
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.012665867805481,
      "learning_rate": 0.0002536380006519943,
      "loss": 2.1496,
      "step": 126581
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.1069939136505127,
      "learning_rate": 0.0002536339592359666,
      "loss": 2.2608,
      "step": 126582
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.001386046409607,
      "learning_rate": 0.0002536299178285593,
      "loss": 2.0935,
      "step": 126583
    },
    {
      "epoch": 1.65,
      "grad_norm": 0.922247052192688,
      "learning_rate": 0.0002536258764297731,
      "loss": 2.2083,
      "step": 126584
    },
    {
      "epoch": 1.65,
      "grad_norm": 0.9401560425758362,
      "learning_rate": 0.00025362183503960874,
      "loss": 2.1521,
      "step": 126585
    },
    {
      "epoch": 1.65,
      "grad_norm": 0.9927607774734497,
      "learning_rate": 0.00025361779365806693,
      "loss": 2.1657,
      "step": 126586
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.055259346961975,
      "learning_rate": 0.00025361375228514854,
      "loss": 1.903,
      "step": 126587
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.312325119972229,
      "learning_rate": 0.0002536097109208541,
      "loss": 2.1719,
      "step": 126588
    },
    {
      "epoch": 1.65,
      "grad_norm": 0.9619740843772888,
      "learning_rate": 0.0002536056695651845,
      "loss": 2.1502,
      "step": 126589
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.1367682218551636,
      "learning_rate": 0.0002536016282181405,
      "loss": 2.3747,
      "step": 126590
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.1314183473587036,
      "learning_rate": 0.0002535975868797228,
      "loss": 1.8091,
      "step": 126591
    },
    {
      "epoch": 1.65,
      "grad_norm": 0.9495037198066711,
      "learning_rate": 0.00025359354554993225,
      "loss": 2.1475,
      "step": 126592
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.0715222358703613,
      "learning_rate": 0.0002535895042287696,
      "loss": 2.0212,
      "step": 126593
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.0414979457855225,
      "learning_rate": 0.0002535854629162355,
      "loss": 2.0949,
      "step": 126594
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.0800765752792358,
      "learning_rate": 0.0002535814216123306,
      "loss": 2.2163,
      "step": 126595
    },
    {
      "epoch": 1.65,
      "grad_norm": 0.9045289754867554,
      "learning_rate": 0.0002535773803170559,
      "loss": 2.2571,
      "step": 126596
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.050158977508545,
      "learning_rate": 0.00025357333903041197,
      "loss": 2.0576,
      "step": 126597
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.0539979934692383,
      "learning_rate": 0.0002535692977523997,
      "loss": 2.2417,
      "step": 126598
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.0349525213241577,
      "learning_rate": 0.00025356525648301986,
      "loss": 2.329,
      "step": 126599
    },
    {
      "epoch": 1.65,
      "grad_norm": 0.9519444704055786,
      "learning_rate": 0.000253561215222273,
      "loss": 2.1767,
      "step": 126600
    },
    {
      "epoch": 1.65,
      "grad_norm": 0.9289008378982544,
      "learning_rate": 0.00025355717397016,
      "loss": 2.0126,
      "step": 126601
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.054966926574707,
      "learning_rate": 0.00025355313272668154,
      "loss": 2.1288,
      "step": 126602
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.006742238998413,
      "learning_rate": 0.0002535490914918385,
      "loss": 2.1334,
      "step": 126603
    },
    {
      "epoch": 1.65,
      "grad_norm": 0.986055850982666,
      "learning_rate": 0.00025354505026563154,
      "loss": 1.8577,
      "step": 126604
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.0104628801345825,
      "learning_rate": 0.00025354100904806154,
      "loss": 2.0619,
      "step": 126605
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.1152184009552002,
      "learning_rate": 0.000253536967839129,
      "loss": 2.2094,
      "step": 126606
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.1254926919937134,
      "learning_rate": 0.00025353292663883483,
      "loss": 2.1155,
      "step": 126607
    },
    {
      "epoch": 1.65,
      "grad_norm": 0.9953783750534058,
      "learning_rate": 0.0002535288854471798,
      "loss": 2.0463,
      "step": 126608
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.1809158325195312,
      "learning_rate": 0.0002535248442641646,
      "loss": 2.1486,
      "step": 126609
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.2302826642990112,
      "learning_rate": 0.00025352080308979006,
      "loss": 2.2372,
      "step": 126610
    },
    {
      "epoch": 1.65,
      "grad_norm": 0.9687060117721558,
      "learning_rate": 0.0002535167619240569,
      "loss": 2.1229,
      "step": 126611
    },
    {
      "epoch": 1.65,
      "grad_norm": 0.9967179894447327,
      "learning_rate": 0.0002535127207669658,
      "loss": 2.2023,
      "step": 126612
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.0007916688919067,
      "learning_rate": 0.00025350867961851755,
      "loss": 2.0306,
      "step": 126613
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.2501189708709717,
      "learning_rate": 0.00025350463847871287,
      "loss": 1.9444,
      "step": 126614
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.0993088483810425,
      "learning_rate": 0.00025350059734755264,
      "loss": 2.0679,
      "step": 126615
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.1064767837524414,
      "learning_rate": 0.00025349655622503746,
      "loss": 2.1428,
      "step": 126616
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.5057826042175293,
      "learning_rate": 0.00025349251511116826,
      "loss": 2.2267,
      "step": 126617
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.0162142515182495,
      "learning_rate": 0.00025348847400594567,
      "loss": 2.1794,
      "step": 126618
    },
    {
      "epoch": 1.65,
      "grad_norm": 0.9949150085449219,
      "learning_rate": 0.00025348443290937035,
      "loss": 2.1534,
      "step": 126619
    },
    {
      "epoch": 1.65,
      "grad_norm": 0.9691459536552429,
      "learning_rate": 0.0002534803918214432,
      "loss": 1.8539,
      "step": 126620
    },
    {
      "epoch": 1.65,
      "grad_norm": 0.9674020409584045,
      "learning_rate": 0.00025347635074216484,
      "loss": 2.1632,
      "step": 126621
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.0552185773849487,
      "learning_rate": 0.0002534723096715362,
      "loss": 1.8823,
      "step": 126622
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.0359232425689697,
      "learning_rate": 0.00025346826860955796,
      "loss": 2.3588,
      "step": 126623
    },
    {
      "epoch": 1.65,
      "grad_norm": 0.9991539120674133,
      "learning_rate": 0.0002534642275562308,
      "loss": 2.1505,
      "step": 126624
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.0950316190719604,
      "learning_rate": 0.0002534601865115555,
      "loss": 2.3902,
      "step": 126625
    },
    {
      "epoch": 1.65,
      "grad_norm": 0.9517204165458679,
      "learning_rate": 0.0002534561454755329,
      "loss": 2.2485,
      "step": 126626
    },
    {
      "epoch": 1.65,
      "grad_norm": 0.8920257091522217,
      "learning_rate": 0.0002534521044481636,
      "loss": 2.0381,
      "step": 126627
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.0019564628601074,
      "learning_rate": 0.00025344806342944846,
      "loss": 2.2694,
      "step": 126628
    },
    {
      "epoch": 1.65,
      "grad_norm": 0.9018551707267761,
      "learning_rate": 0.00025344402241938825,
      "loss": 2.1083,
      "step": 126629
    },
    {
      "epoch": 1.65,
      "grad_norm": 0.9409825801849365,
      "learning_rate": 0.0002534399814179836,
      "loss": 1.9634,
      "step": 126630
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.0769894123077393,
      "learning_rate": 0.00025343594042523534,
      "loss": 2.2329,
      "step": 126631
    },
    {
      "epoch": 1.65,
      "grad_norm": 0.9729015827178955,
      "learning_rate": 0.00025343189944114427,
      "loss": 2.0464,
      "step": 126632
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.0620020627975464,
      "learning_rate": 0.0002534278584657111,
      "loss": 2.2461,
      "step": 126633
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.0605792999267578,
      "learning_rate": 0.0002534238174989365,
      "loss": 2.0679,
      "step": 126634
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.009158968925476,
      "learning_rate": 0.0002534197765408214,
      "loss": 2.1703,
      "step": 126635
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.0171887874603271,
      "learning_rate": 0.00025341573559136635,
      "loss": 2.1989,
      "step": 126636
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.1108956336975098,
      "learning_rate": 0.0002534116946505722,
      "loss": 1.9229,
      "step": 126637
    },
    {
      "epoch": 1.65,
      "grad_norm": 0.9264773726463318,
      "learning_rate": 0.00025340765371843966,
      "loss": 2.2065,
      "step": 126638
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.0650614500045776,
      "learning_rate": 0.0002534036127949696,
      "loss": 2.1769,
      "step": 126639
    },
    {
      "epoch": 1.65,
      "grad_norm": 0.9654484987258911,
      "learning_rate": 0.00025339957188016266,
      "loss": 2.1999,
      "step": 126640
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.1749053001403809,
      "learning_rate": 0.00025339553097401964,
      "loss": 2.1975,
      "step": 126641
    },
    {
      "epoch": 1.65,
      "grad_norm": 0.9841113090515137,
      "learning_rate": 0.00025339149007654125,
      "loss": 2.2994,
      "step": 126642
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.0585607290267944,
      "learning_rate": 0.00025338744918772824,
      "loss": 2.1003,
      "step": 126643
    },
    {
      "epoch": 1.65,
      "grad_norm": 0.9023200273513794,
      "learning_rate": 0.00025338340830758136,
      "loss": 2.1796,
      "step": 126644
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.040763020515442,
      "learning_rate": 0.00025337936743610143,
      "loss": 2.0637,
      "step": 126645
    },
    {
      "epoch": 1.65,
      "grad_norm": 0.962070643901825,
      "learning_rate": 0.0002533753265732891,
      "loss": 1.9975,
      "step": 126646
    },
    {
      "epoch": 1.65,
      "grad_norm": 0.9362398982048035,
      "learning_rate": 0.0002533712857191453,
      "loss": 2.0459,
      "step": 126647
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.0957502126693726,
      "learning_rate": 0.00025336724487367066,
      "loss": 2.1048,
      "step": 126648
    },
    {
      "epoch": 1.65,
      "grad_norm": 0.9681354761123657,
      "learning_rate": 0.0002533632040368658,
      "loss": 1.9279,
      "step": 126649
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.0238704681396484,
      "learning_rate": 0.0002533591632087316,
      "loss": 2.2836,
      "step": 126650
    },
    {
      "epoch": 1.65,
      "grad_norm": 0.8940344452857971,
      "learning_rate": 0.0002533551223892689,
      "loss": 1.9034,
      "step": 126651
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.0079882144927979,
      "learning_rate": 0.0002533510815784783,
      "loss": 2.2084,
      "step": 126652
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.1113559007644653,
      "learning_rate": 0.0002533470407763606,
      "loss": 1.9535,
      "step": 126653
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.0085562467575073,
      "learning_rate": 0.0002533429999829167,
      "loss": 2.1183,
      "step": 126654
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.0126690864562988,
      "learning_rate": 0.00025333895919814716,
      "loss": 2.0534,
      "step": 126655
    },
    {
      "epoch": 1.65,
      "grad_norm": 0.9439290761947632,
      "learning_rate": 0.00025333491842205273,
      "loss": 2.0727,
      "step": 126656
    },
    {
      "epoch": 1.65,
      "grad_norm": 0.9728654623031616,
      "learning_rate": 0.0002533308776546342,
      "loss": 2.0501,
      "step": 126657
    },
    {
      "epoch": 1.65,
      "grad_norm": 0.9855261445045471,
      "learning_rate": 0.0002533268368958924,
      "loss": 2.0076,
      "step": 126658
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.0197197198867798,
      "learning_rate": 0.00025332279614582797,
      "loss": 2.0918,
      "step": 126659
    },
    {
      "epoch": 1.65,
      "grad_norm": 0.9092404842376709,
      "learning_rate": 0.0002533187554044419,
      "loss": 2.0702,
      "step": 126660
    },
    {
      "epoch": 1.65,
      "grad_norm": 0.8806653618812561,
      "learning_rate": 0.00025331471467173453,
      "loss": 1.8229,
      "step": 126661
    },
    {
      "epoch": 1.65,
      "grad_norm": 0.9521083831787109,
      "learning_rate": 0.0002533106739477069,
      "loss": 2.3834,
      "step": 126662
    },
    {
      "epoch": 1.65,
      "grad_norm": 0.9620853066444397,
      "learning_rate": 0.0002533066332323597,
      "loss": 1.9046,
      "step": 126663
    },
    {
      "epoch": 1.65,
      "grad_norm": 0.9875996112823486,
      "learning_rate": 0.00025330259252569367,
      "loss": 1.9976,
      "step": 126664
    },
    {
      "epoch": 1.65,
      "grad_norm": 0.9130234718322754,
      "learning_rate": 0.00025329855182770956,
      "loss": 2.1087,
      "step": 126665
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.177099585533142,
      "learning_rate": 0.0002532945111384083,
      "loss": 2.118,
      "step": 126666
    },
    {
      "epoch": 1.65,
      "grad_norm": 0.9752545356750488,
      "learning_rate": 0.00025329047045779033,
      "loss": 2.4041,
      "step": 126667
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.02980375289917,
      "learning_rate": 0.0002532864297858565,
      "loss": 2.1311,
      "step": 126668
    },
    {
      "epoch": 1.65,
      "grad_norm": 0.964604377746582,
      "learning_rate": 0.00025328238912260764,
      "loss": 2.0913,
      "step": 126669
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.0384259223937988,
      "learning_rate": 0.0002532783484680444,
      "loss": 2.133,
      "step": 126670
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.3413585424423218,
      "learning_rate": 0.00025327430782216766,
      "loss": 2.0223,
      "step": 126671
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.0940990447998047,
      "learning_rate": 0.0002532702671849783,
      "loss": 2.222,
      "step": 126672
    },
    {
      "epoch": 1.65,
      "grad_norm": 0.9689902663230896,
      "learning_rate": 0.00025326622655647663,
      "loss": 1.9998,
      "step": 126673
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.2438369989395142,
      "learning_rate": 0.0002532621859366637,
      "loss": 2.2493,
      "step": 126674
    },
    {
      "epoch": 1.65,
      "grad_norm": 0.8487494587898254,
      "learning_rate": 0.0002532581453255402,
      "loss": 2.1099,
      "step": 126675
    },
    {
      "epoch": 1.65,
      "grad_norm": 0.9850863814353943,
      "learning_rate": 0.00025325410472310693,
      "loss": 2.1943,
      "step": 126676
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.0948286056518555,
      "learning_rate": 0.00025325006412936453,
      "loss": 2.1466,
      "step": 126677
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.021012544631958,
      "learning_rate": 0.00025324602354431406,
      "loss": 2.1043,
      "step": 126678
    },
    {
      "epoch": 1.65,
      "grad_norm": 0.9845373034477234,
      "learning_rate": 0.00025324198296795586,
      "loss": 2.1809,
      "step": 126679
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.083911657333374,
      "learning_rate": 0.0002532379424002908,
      "loss": 2.1573,
      "step": 126680
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.0160032510757446,
      "learning_rate": 0.0002532339018413198,
      "loss": 2.0559,
      "step": 126681
    },
    {
      "epoch": 1.65,
      "grad_norm": 0.9816951751708984,
      "learning_rate": 0.0002532298612910434,
      "loss": 2.1092,
      "step": 126682
    },
    {
      "epoch": 1.65,
      "grad_norm": 0.908101499080658,
      "learning_rate": 0.00025322582074946247,
      "loss": 2.3089,
      "step": 126683
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.0471646785736084,
      "learning_rate": 0.0002532217802165779,
      "loss": 2.182,
      "step": 126684
    },
    {
      "epoch": 1.65,
      "grad_norm": 0.9417709708213806,
      "learning_rate": 0.00025321773969239016,
      "loss": 2.1214,
      "step": 126685
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.0709117650985718,
      "learning_rate": 0.00025321369917690015,
      "loss": 1.951,
      "step": 126686
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.0541349649429321,
      "learning_rate": 0.00025320965867010853,
      "loss": 2.2172,
      "step": 126687
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.1343392133712769,
      "learning_rate": 0.0002532056181720161,
      "loss": 2.0904,
      "step": 126688
    },
    {
      "epoch": 1.65,
      "grad_norm": 0.9707205891609192,
      "learning_rate": 0.00025320157768262373,
      "loss": 2.2433,
      "step": 126689
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.302228569984436,
      "learning_rate": 0.0002531975372019321,
      "loss": 2.0856,
      "step": 126690
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.0340766906738281,
      "learning_rate": 0.0002531934967299419,
      "loss": 1.9561,
      "step": 126691
    },
    {
      "epoch": 1.65,
      "grad_norm": 0.9750978350639343,
      "learning_rate": 0.0002531894562666538,
      "loss": 1.9107,
      "step": 126692
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.0273573398590088,
      "learning_rate": 0.00025318541581206876,
      "loss": 2.1011,
      "step": 126693
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.1966933012008667,
      "learning_rate": 0.0002531813753661874,
      "loss": 2.231,
      "step": 126694
    },
    {
      "epoch": 1.65,
      "grad_norm": 0.9819238185882568,
      "learning_rate": 0.0002531773349290105,
      "loss": 2.0279,
      "step": 126695
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.070409893989563,
      "learning_rate": 0.00025317329450053896,
      "loss": 2.2089,
      "step": 126696
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.356505036354065,
      "learning_rate": 0.00025316925408077327,
      "loss": 2.1544,
      "step": 126697
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.183289647102356,
      "learning_rate": 0.00025316521366971427,
      "loss": 2.2213,
      "step": 126698
    },
    {
      "epoch": 1.65,
      "grad_norm": 0.9860815405845642,
      "learning_rate": 0.00025316117326736276,
      "loss": 2.073,
      "step": 126699
    },
    {
      "epoch": 1.65,
      "grad_norm": 0.986649215221405,
      "learning_rate": 0.0002531571328737194,
      "loss": 2.0235,
      "step": 126700
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.099535584449768,
      "learning_rate": 0.0002531530924887851,
      "loss": 2.098,
      "step": 126701
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.0057295560836792,
      "learning_rate": 0.00025314905211256056,
      "loss": 2.0126,
      "step": 126702
    },
    {
      "epoch": 1.65,
      "grad_norm": 0.9330910444259644,
      "learning_rate": 0.00025314501174504653,
      "loss": 2.2187,
      "step": 126703
    },
    {
      "epoch": 1.65,
      "grad_norm": 0.9745497703552246,
      "learning_rate": 0.00025314097138624364,
      "loss": 1.9442,
      "step": 126704
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.1042224168777466,
      "learning_rate": 0.0002531369310361527,
      "loss": 2.1549,
      "step": 126705
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.049713373184204,
      "learning_rate": 0.0002531328906947745,
      "loss": 2.163,
      "step": 126706
    },
    {
      "epoch": 1.65,
      "grad_norm": 0.8939053416252136,
      "learning_rate": 0.0002531288503621098,
      "loss": 2.0689,
      "step": 126707
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.0477248430252075,
      "learning_rate": 0.0002531248100381594,
      "loss": 2.225,
      "step": 126708
    },
    {
      "epoch": 1.65,
      "grad_norm": 0.980114758014679,
      "learning_rate": 0.00025312076972292387,
      "loss": 2.0662,
      "step": 126709
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.0425564050674438,
      "learning_rate": 0.0002531167294164042,
      "loss": 2.0854,
      "step": 126710
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.1210912466049194,
      "learning_rate": 0.0002531126891186009,
      "loss": 2.3524,
      "step": 126711
    },
    {
      "epoch": 1.65,
      "grad_norm": 0.9777209758758545,
      "learning_rate": 0.0002531086488295149,
      "loss": 2.1064,
      "step": 126712
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.0694106817245483,
      "learning_rate": 0.0002531046085491468,
      "loss": 2.401,
      "step": 126713
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.0341923236846924,
      "learning_rate": 0.0002531005682774975,
      "loss": 2.1596,
      "step": 126714
    },
    {
      "epoch": 1.65,
      "grad_norm": 0.9399795532226562,
      "learning_rate": 0.00025309652801456774,
      "loss": 1.9366,
      "step": 126715
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.0089856386184692,
      "learning_rate": 0.00025309248776035815,
      "loss": 2.0649,
      "step": 126716
    },
    {
      "epoch": 1.65,
      "grad_norm": 0.9970284700393677,
      "learning_rate": 0.00025308844751486956,
      "loss": 2.2897,
      "step": 126717
    },
    {
      "epoch": 1.65,
      "grad_norm": 0.9750360250473022,
      "learning_rate": 0.0002530844072781028,
      "loss": 2.0787,
      "step": 126718
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.049206018447876,
      "learning_rate": 0.0002530803670500584,
      "loss": 2.2738,
      "step": 126719
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.2288339138031006,
      "learning_rate": 0.0002530763268307373,
      "loss": 2.2048,
      "step": 126720
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.0130395889282227,
      "learning_rate": 0.00025307228662014024,
      "loss": 2.2001,
      "step": 126721
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.237764596939087,
      "learning_rate": 0.0002530682464182679,
      "loss": 2.1723,
      "step": 126722
    },
    {
      "epoch": 1.65,
      "grad_norm": 0.9053845405578613,
      "learning_rate": 0.000253064206225121,
      "loss": 1.8264,
      "step": 126723
    },
    {
      "epoch": 1.65,
      "grad_norm": 0.9756001234054565,
      "learning_rate": 0.0002530601660407004,
      "loss": 2.2662,
      "step": 126724
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.084132432937622,
      "learning_rate": 0.0002530561258650069,
      "loss": 2.098,
      "step": 126725
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.0459800958633423,
      "learning_rate": 0.000253052085698041,
      "loss": 2.2292,
      "step": 126726
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.0525150299072266,
      "learning_rate": 0.0002530480455398037,
      "loss": 2.0767,
      "step": 126727
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.0987333059310913,
      "learning_rate": 0.0002530440053902956,
      "loss": 2.1608,
      "step": 126728
    },
    {
      "epoch": 1.65,
      "grad_norm": 0.9894068837165833,
      "learning_rate": 0.0002530399652495175,
      "loss": 2.247,
      "step": 126729
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.0368521213531494,
      "learning_rate": 0.00025303592511747013,
      "loss": 2.36,
      "step": 126730
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.0483312606811523,
      "learning_rate": 0.0002530318849941543,
      "loss": 2.0759,
      "step": 126731
    },
    {
      "epoch": 1.65,
      "grad_norm": 0.9756373167037964,
      "learning_rate": 0.0002530278448795708,
      "loss": 2.205,
      "step": 126732
    },
    {
      "epoch": 1.65,
      "grad_norm": 0.9298649430274963,
      "learning_rate": 0.00025302380477372033,
      "loss": 2.2925,
      "step": 126733
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.03931725025177,
      "learning_rate": 0.0002530197646766035,
      "loss": 2.1851,
      "step": 126734
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.1088474988937378,
      "learning_rate": 0.0002530157245882212,
      "loss": 2.4029,
      "step": 126735
    },
    {
      "epoch": 1.65,
      "grad_norm": 0.9552648067474365,
      "learning_rate": 0.0002530116845085742,
      "loss": 1.9233,
      "step": 126736
    },
    {
      "epoch": 1.65,
      "grad_norm": 0.9671697616577148,
      "learning_rate": 0.00025300764443766314,
      "loss": 2.1586,
      "step": 126737
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.011481523513794,
      "learning_rate": 0.0002530036043754889,
      "loss": 1.9847,
      "step": 126738
    },
    {
      "epoch": 1.65,
      "grad_norm": 0.9822016954421997,
      "learning_rate": 0.00025299956432205237,
      "loss": 2.2315,
      "step": 126739
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.3288350105285645,
      "learning_rate": 0.00025299552427735384,
      "loss": 1.9814,
      "step": 126740
    },
    {
      "epoch": 1.65,
      "grad_norm": 0.979607105255127,
      "learning_rate": 0.0002529914842413944,
      "loss": 2.1453,
      "step": 126741
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.0301932096481323,
      "learning_rate": 0.00025298744421417476,
      "loss": 1.9752,
      "step": 126742
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.0737224817276,
      "learning_rate": 0.0002529834041956956,
      "loss": 1.9501,
      "step": 126743
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.0584280490875244,
      "learning_rate": 0.0002529793641859577,
      "loss": 2.3105,
      "step": 126744
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.065636157989502,
      "learning_rate": 0.000252975324184962,
      "loss": 2.1346,
      "step": 126745
    },
    {
      "epoch": 1.65,
      "grad_norm": 0.927847146987915,
      "learning_rate": 0.0002529712841927089,
      "loss": 2.178,
      "step": 126746
    },
    {
      "epoch": 1.65,
      "grad_norm": 0.9213386178016663,
      "learning_rate": 0.00025296724420919935,
      "loss": 2.077,
      "step": 126747
    },
    {
      "epoch": 1.65,
      "grad_norm": 0.9296849370002747,
      "learning_rate": 0.0002529632042344341,
      "loss": 2.0368,
      "step": 126748
    },
    {
      "epoch": 1.65,
      "grad_norm": 0.9858410358428955,
      "learning_rate": 0.00025295916426841384,
      "loss": 2.3329,
      "step": 126749
    },
    {
      "epoch": 1.65,
      "grad_norm": 0.9867480397224426,
      "learning_rate": 0.00025295512431113936,
      "loss": 2.117,
      "step": 126750
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.078238606452942,
      "learning_rate": 0.0002529510843626116,
      "loss": 2.1224,
      "step": 126751
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.044548749923706,
      "learning_rate": 0.0002529470444228309,
      "loss": 2.0431,
      "step": 126752
    },
    {
      "epoch": 1.65,
      "grad_norm": 0.9873486161231995,
      "learning_rate": 0.0002529430044917983,
      "loss": 2.1355,
      "step": 126753
    },
    {
      "epoch": 1.65,
      "grad_norm": 0.9832320809364319,
      "learning_rate": 0.00025293896456951445,
      "loss": 1.9381,
      "step": 126754
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.1372836828231812,
      "learning_rate": 0.00025293492465598016,
      "loss": 2.106,
      "step": 126755
    },
    {
      "epoch": 1.65,
      "grad_norm": 0.9929981827735901,
      "learning_rate": 0.00025293088475119614,
      "loss": 1.8514,
      "step": 126756
    },
    {
      "epoch": 1.65,
      "grad_norm": 0.9507918357849121,
      "learning_rate": 0.0002529268448551633,
      "loss": 2.0343,
      "step": 126757
    },
    {
      "epoch": 1.65,
      "grad_norm": 0.9726232290267944,
      "learning_rate": 0.0002529228049678821,
      "loss": 2.1097,
      "step": 126758
    },
    {
      "epoch": 1.65,
      "grad_norm": 0.9703866243362427,
      "learning_rate": 0.00025291876508935345,
      "loss": 2.1237,
      "step": 126759
    },
    {
      "epoch": 1.65,
      "grad_norm": 0.9963381290435791,
      "learning_rate": 0.0002529147252195781,
      "loss": 1.834,
      "step": 126760
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.0345640182495117,
      "learning_rate": 0.00025291068535855676,
      "loss": 2.2284,
      "step": 126761
    },
    {
      "epoch": 1.65,
      "grad_norm": 0.9892312288284302,
      "learning_rate": 0.00025290664550629025,
      "loss": 2.1018,
      "step": 126762
    },
    {
      "epoch": 1.65,
      "grad_norm": 0.9980854392051697,
      "learning_rate": 0.00025290260566277937,
      "loss": 2.2764,
      "step": 126763
    },
    {
      "epoch": 1.65,
      "grad_norm": 0.9521503448486328,
      "learning_rate": 0.00025289856582802467,
      "loss": 2.2693,
      "step": 126764
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.0196222066879272,
      "learning_rate": 0.000252894526002027,
      "loss": 1.9946,
      "step": 126765
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.2190886735916138,
      "learning_rate": 0.00025289048618478717,
      "loss": 2.1477,
      "step": 126766
    },
    {
      "epoch": 1.65,
      "grad_norm": 0.9013057351112366,
      "learning_rate": 0.00025288644637630584,
      "loss": 2.0901,
      "step": 126767
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.089838981628418,
      "learning_rate": 0.0002528824065765839,
      "loss": 2.0585,
      "step": 126768
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.1120814085006714,
      "learning_rate": 0.00025287836678562206,
      "loss": 2.0956,
      "step": 126769
    },
    {
      "epoch": 1.65,
      "grad_norm": 0.9383482336997986,
      "learning_rate": 0.0002528743270034209,
      "loss": 2.1255,
      "step": 126770
    },
    {
      "epoch": 1.65,
      "grad_norm": 0.9566216468811035,
      "learning_rate": 0.0002528702872299813,
      "loss": 2.04,
      "step": 126771
    },
    {
      "epoch": 1.65,
      "grad_norm": 0.9320852756500244,
      "learning_rate": 0.000252866247465304,
      "loss": 1.992,
      "step": 126772
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.0304638147354126,
      "learning_rate": 0.00025286220770938975,
      "loss": 2.0567,
      "step": 126773
    },
    {
      "epoch": 1.65,
      "grad_norm": 0.9936739802360535,
      "learning_rate": 0.00025285816796223934,
      "loss": 1.9961,
      "step": 126774
    },
    {
      "epoch": 1.65,
      "grad_norm": 0.9537147879600525,
      "learning_rate": 0.0002528541282238536,
      "loss": 2.119,
      "step": 126775
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.1197611093521118,
      "learning_rate": 0.000252850088494233,
      "loss": 2.1749,
      "step": 126776
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.118710994720459,
      "learning_rate": 0.0002528460487733785,
      "loss": 2.1868,
      "step": 126777
    },
    {
      "epoch": 1.65,
      "grad_norm": 0.9866975545883179,
      "learning_rate": 0.00025284200906129076,
      "loss": 2.2073,
      "step": 126778
    },
    {
      "epoch": 1.65,
      "grad_norm": 0.9662932753562927,
      "learning_rate": 0.00025283796935797063,
      "loss": 2.0003,
      "step": 126779
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.0317599773406982,
      "learning_rate": 0.00025283392966341883,
      "loss": 2.2728,
      "step": 126780
    },
    {
      "epoch": 1.65,
      "grad_norm": 0.9479380249977112,
      "learning_rate": 0.000252829889977636,
      "loss": 2.1686,
      "step": 126781
    },
    {
      "epoch": 1.65,
      "grad_norm": 0.9819081425666809,
      "learning_rate": 0.00025282585030062324,
      "loss": 2.2338,
      "step": 126782
    },
    {
      "epoch": 1.65,
      "grad_norm": 0.9546146392822266,
      "learning_rate": 0.0002528218106323808,
      "loss": 1.8057,
      "step": 126783
    },
    {
      "epoch": 1.65,
      "grad_norm": 0.9669780135154724,
      "learning_rate": 0.0002528177709729097,
      "loss": 2.0557,
      "step": 126784
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.0063471794128418,
      "learning_rate": 0.00025281373132221075,
      "loss": 2.0362,
      "step": 126785
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.0899325609207153,
      "learning_rate": 0.00025280969168028455,
      "loss": 2.2211,
      "step": 126786
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.1402804851531982,
      "learning_rate": 0.00025280565204713194,
      "loss": 2.1059,
      "step": 126787
    },
    {
      "epoch": 1.65,
      "grad_norm": 0.9400993585586548,
      "learning_rate": 0.00025280161242275377,
      "loss": 2.0079,
      "step": 126788
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.0806372165679932,
      "learning_rate": 0.00025279757280715053,
      "loss": 2.0521,
      "step": 126789
    },
    {
      "epoch": 1.65,
      "grad_norm": 0.9538013339042664,
      "learning_rate": 0.0002527935332003231,
      "loss": 2.0602,
      "step": 126790
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.037989854812622,
      "learning_rate": 0.0002527894936022723,
      "loss": 2.2909,
      "step": 126791
    },
    {
      "epoch": 1.65,
      "grad_norm": 0.9703261852264404,
      "learning_rate": 0.0002527854540129988,
      "loss": 2.3778,
      "step": 126792
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.058683156967163,
      "learning_rate": 0.0002527814144325034,
      "loss": 2.0717,
      "step": 126793
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.037490963935852,
      "learning_rate": 0.0002527773748607868,
      "loss": 2.064,
      "step": 126794
    },
    {
      "epoch": 1.65,
      "grad_norm": 0.9842760562896729,
      "learning_rate": 0.00025277333529784984,
      "loss": 2.1907,
      "step": 126795
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.0968598127365112,
      "learning_rate": 0.00025276929574369317,
      "loss": 2.2323,
      "step": 126796
    },
    {
      "epoch": 1.65,
      "grad_norm": 0.9425941705703735,
      "learning_rate": 0.0002527652561983175,
      "loss": 2.2264,
      "step": 126797
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.0407838821411133,
      "learning_rate": 0.0002527612166617237,
      "loss": 2.0366,
      "step": 126798
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.0713953971862793,
      "learning_rate": 0.0002527571771339125,
      "loss": 2.0483,
      "step": 126799
    },
    {
      "epoch": 1.65,
      "grad_norm": 0.9713770747184753,
      "learning_rate": 0.0002527531376148847,
      "loss": 2.1674,
      "step": 126800
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.0675814151763916,
      "learning_rate": 0.0002527490981046409,
      "loss": 2.1585,
      "step": 126801
    },
    {
      "epoch": 1.65,
      "grad_norm": 0.9516890645027161,
      "learning_rate": 0.000252745058603182,
      "loss": 2.053,
      "step": 126802
    },
    {
      "epoch": 1.65,
      "grad_norm": 0.9811792373657227,
      "learning_rate": 0.00025274101911050865,
      "loss": 1.9408,
      "step": 126803
    },
    {
      "epoch": 1.65,
      "grad_norm": 0.9787883758544922,
      "learning_rate": 0.0002527369796266216,
      "loss": 2.2991,
      "step": 126804
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.0024698972702026,
      "learning_rate": 0.00025273294015152166,
      "loss": 2.1014,
      "step": 126805
    },
    {
      "epoch": 1.65,
      "grad_norm": 0.9564626216888428,
      "learning_rate": 0.0002527289006852096,
      "loss": 2.1778,
      "step": 126806
    },
    {
      "epoch": 1.65,
      "grad_norm": 0.9836573004722595,
      "learning_rate": 0.0002527248612276861,
      "loss": 1.9678,
      "step": 126807
    },
    {
      "epoch": 1.65,
      "grad_norm": 0.9220284223556519,
      "learning_rate": 0.0002527208217789519,
      "loss": 1.9889,
      "step": 126808
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.0054051876068115,
      "learning_rate": 0.00025271678233900786,
      "loss": 2.0344,
      "step": 126809
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.026658296585083,
      "learning_rate": 0.00025271274290785463,
      "loss": 2.2955,
      "step": 126810
    },
    {
      "epoch": 1.65,
      "grad_norm": 0.9735994935035706,
      "learning_rate": 0.000252708703485493,
      "loss": 2.1405,
      "step": 126811
    },
    {
      "epoch": 1.65,
      "grad_norm": 0.9880525469779968,
      "learning_rate": 0.0002527046640719237,
      "loss": 2.2094,
      "step": 126812
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.0171455144882202,
      "learning_rate": 0.0002527006246671475,
      "loss": 2.197,
      "step": 126813
    },
    {
      "epoch": 1.65,
      "grad_norm": 0.983270525932312,
      "learning_rate": 0.00025269658527116514,
      "loss": 2.1911,
      "step": 126814
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.2583168745040894,
      "learning_rate": 0.0002526925458839774,
      "loss": 2.2263,
      "step": 126815
    },
    {
      "epoch": 1.65,
      "grad_norm": 0.9646631479263306,
      "learning_rate": 0.00025268850650558497,
      "loss": 2.1407,
      "step": 126816
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.0642344951629639,
      "learning_rate": 0.0002526844671359887,
      "loss": 2.0904,
      "step": 126817
    },
    {
      "epoch": 1.65,
      "grad_norm": 0.9618504047393799,
      "learning_rate": 0.0002526804277751893,
      "loss": 2.0801,
      "step": 126818
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.0211591720581055,
      "learning_rate": 0.00025267638842318746,
      "loss": 2.0552,
      "step": 126819
    },
    {
      "epoch": 1.65,
      "grad_norm": 0.9476675391197205,
      "learning_rate": 0.00025267234907998395,
      "loss": 1.9435,
      "step": 126820
    },
    {
      "epoch": 1.65,
      "grad_norm": 0.997430145740509,
      "learning_rate": 0.00025266830974557955,
      "loss": 2.1999,
      "step": 126821
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.0725494623184204,
      "learning_rate": 0.000252664270419975,
      "loss": 2.0271,
      "step": 126822
    },
    {
      "epoch": 1.65,
      "grad_norm": 0.9869824051856995,
      "learning_rate": 0.00025266023110317103,
      "loss": 2.0547,
      "step": 126823
    },
    {
      "epoch": 1.65,
      "grad_norm": 0.9956908226013184,
      "learning_rate": 0.0002526561917951686,
      "loss": 2.2065,
      "step": 126824
    },
    {
      "epoch": 1.65,
      "grad_norm": 0.9466294050216675,
      "learning_rate": 0.0002526521524959681,
      "loss": 2.0884,
      "step": 126825
    },
    {
      "epoch": 1.65,
      "grad_norm": 0.9737887382507324,
      "learning_rate": 0.00025264811320557047,
      "loss": 1.9529,
      "step": 126826
    },
    {
      "epoch": 1.65,
      "grad_norm": 0.9256396293640137,
      "learning_rate": 0.0002526440739239764,
      "loss": 2.0889,
      "step": 126827
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.0836673974990845,
      "learning_rate": 0.0002526400346511868,
      "loss": 1.9774,
      "step": 126828
    },
    {
      "epoch": 1.65,
      "grad_norm": 0.9235042333602905,
      "learning_rate": 0.0002526359953872022,
      "loss": 2.2187,
      "step": 126829
    },
    {
      "epoch": 1.65,
      "grad_norm": 0.9869764447212219,
      "learning_rate": 0.00025263195613202367,
      "loss": 2.1854,
      "step": 126830
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.0726808309555054,
      "learning_rate": 0.0002526279168856516,
      "loss": 2.0037,
      "step": 126831
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.1226364374160767,
      "learning_rate": 0.0002526238776480869,
      "loss": 2.022,
      "step": 126832
    },
    {
      "epoch": 1.65,
      "grad_norm": 0.9097862243652344,
      "learning_rate": 0.0002526198384193303,
      "loss": 2.0191,
      "step": 126833
    },
    {
      "epoch": 1.65,
      "grad_norm": 0.9550034403800964,
      "learning_rate": 0.0002526157991993826,
      "loss": 1.9928,
      "step": 126834
    },
    {
      "epoch": 1.65,
      "grad_norm": 0.9956434369087219,
      "learning_rate": 0.00025261175998824445,
      "loss": 2.1804,
      "step": 126835
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.1632671356201172,
      "learning_rate": 0.00025260772078591686,
      "loss": 2.2671,
      "step": 126836
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.0422532558441162,
      "learning_rate": 0.00025260368159240026,
      "loss": 2.183,
      "step": 126837
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.1969530582427979,
      "learning_rate": 0.0002525996424076955,
      "loss": 2.1112,
      "step": 126838
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.1592037677764893,
      "learning_rate": 0.0002525956032318034,
      "loss": 2.2175,
      "step": 126839
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.0321601629257202,
      "learning_rate": 0.00025259156406472465,
      "loss": 2.1065,
      "step": 126840
    },
    {
      "epoch": 1.65,
      "grad_norm": 0.907474160194397,
      "learning_rate": 0.00025258752490646,
      "loss": 2.0979,
      "step": 126841
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.015271782875061,
      "learning_rate": 0.00025258348575701037,
      "loss": 2.2871,
      "step": 126842
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.0596468448638916,
      "learning_rate": 0.00025257944661637627,
      "loss": 2.2756,
      "step": 126843
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.379701852798462,
      "learning_rate": 0.0002525754074845585,
      "loss": 2.1481,
      "step": 126844
    },
    {
      "epoch": 1.65,
      "grad_norm": 0.9878945350646973,
      "learning_rate": 0.0002525713683615579,
      "loss": 2.1878,
      "step": 126845
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.155346155166626,
      "learning_rate": 0.0002525673292473751,
      "loss": 2.0345,
      "step": 126846
    },
    {
      "epoch": 1.65,
      "grad_norm": 0.9355852007865906,
      "learning_rate": 0.000252563290142011,
      "loss": 2.2506,
      "step": 126847
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.1124095916748047,
      "learning_rate": 0.0002525592510454662,
      "loss": 2.2119,
      "step": 126848
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.1435850858688354,
      "learning_rate": 0.0002525552119577418,
      "loss": 1.9196,
      "step": 126849
    },
    {
      "epoch": 1.65,
      "grad_norm": 0.9909804463386536,
      "learning_rate": 0.00025255117287883804,
      "loss": 2.0873,
      "step": 126850
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.3087680339813232,
      "learning_rate": 0.00025254713380875595,
      "loss": 1.9953,
      "step": 126851
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.0215842723846436,
      "learning_rate": 0.00025254309474749624,
      "loss": 1.9766,
      "step": 126852
    },
    {
      "epoch": 1.65,
      "grad_norm": 0.9161251783370972,
      "learning_rate": 0.00025253905569505965,
      "loss": 2.3072,
      "step": 126853
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.0082175731658936,
      "learning_rate": 0.00025253501665144695,
      "loss": 2.2343,
      "step": 126854
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.1208208799362183,
      "learning_rate": 0.00025253097761665906,
      "loss": 2.0719,
      "step": 126855
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.016281247138977,
      "learning_rate": 0.00025252693859069636,
      "loss": 2.2044,
      "step": 126856
    },
    {
      "epoch": 1.65,
      "grad_norm": 0.9485941529273987,
      "learning_rate": 0.00025252289957355983,
      "loss": 2.0767,
      "step": 126857
    },
    {
      "epoch": 1.65,
      "grad_norm": 0.8957177400588989,
      "learning_rate": 0.0002525188605652502,
      "loss": 2.1244,
      "step": 126858
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.099295973777771,
      "learning_rate": 0.0002525148215657682,
      "loss": 2.2108,
      "step": 126859
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.3801319599151611,
      "learning_rate": 0.0002525107825751146,
      "loss": 2.0272,
      "step": 126860
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.011206030845642,
      "learning_rate": 0.00025250674359329024,
      "loss": 1.9392,
      "step": 126861
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.0373718738555908,
      "learning_rate": 0.0002525027046202957,
      "loss": 2.0866,
      "step": 126862
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.203447937965393,
      "learning_rate": 0.0002524986656561317,
      "loss": 2.2456,
      "step": 126863
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.524404764175415,
      "learning_rate": 0.00025249462670079914,
      "loss": 1.9584,
      "step": 126864
    },
    {
      "epoch": 1.65,
      "grad_norm": 0.9553552269935608,
      "learning_rate": 0.00025249058775429875,
      "loss": 2.0556,
      "step": 126865
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.0145657062530518,
      "learning_rate": 0.0002524865488166313,
      "loss": 2.259,
      "step": 126866
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.2069765329360962,
      "learning_rate": 0.00025248250988779756,
      "loss": 2.4065,
      "step": 126867
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.0275553464889526,
      "learning_rate": 0.0002524784709677981,
      "loss": 2.0648,
      "step": 126868
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.0770974159240723,
      "learning_rate": 0.00025247443205663383,
      "loss": 2.3158,
      "step": 126869
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.0010215044021606,
      "learning_rate": 0.0002524703931543054,
      "loss": 2.1162,
      "step": 126870
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.0463039875030518,
      "learning_rate": 0.0002524663542608137,
      "loss": 2.1348,
      "step": 126871
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.111219882965088,
      "learning_rate": 0.00025246231537615933,
      "loss": 2.0189,
      "step": 126872
    },
    {
      "epoch": 1.65,
      "grad_norm": 0.983951210975647,
      "learning_rate": 0.00025245827650034324,
      "loss": 2.0083,
      "step": 126873
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.047368049621582,
      "learning_rate": 0.00025245423763336594,
      "loss": 2.322,
      "step": 126874
    },
    {
      "epoch": 1.65,
      "grad_norm": 0.9715685248374939,
      "learning_rate": 0.0002524501987752283,
      "loss": 2.221,
      "step": 126875
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.0993437767028809,
      "learning_rate": 0.0002524461599259311,
      "loss": 2.2753,
      "step": 126876
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.2024441957473755,
      "learning_rate": 0.000252442121085475,
      "loss": 2.0533,
      "step": 126877
    },
    {
      "epoch": 1.65,
      "grad_norm": 0.9560361504554749,
      "learning_rate": 0.00025243808225386085,
      "loss": 2.0881,
      "step": 126878
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.0241774320602417,
      "learning_rate": 0.0002524340434310894,
      "loss": 2.0678,
      "step": 126879
    },
    {
      "epoch": 1.65,
      "grad_norm": 0.9431016445159912,
      "learning_rate": 0.00025243000461716136,
      "loss": 2.0271,
      "step": 126880
    },
    {
      "epoch": 1.65,
      "grad_norm": 0.997296154499054,
      "learning_rate": 0.0002524259658120774,
      "loss": 2.2642,
      "step": 126881
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.02946937084198,
      "learning_rate": 0.0002524219270158384,
      "loss": 2.274,
      "step": 126882
    },
    {
      "epoch": 1.65,
      "grad_norm": 0.9189324975013733,
      "learning_rate": 0.00025241788822844507,
      "loss": 2.0543,
      "step": 126883
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.0641378164291382,
      "learning_rate": 0.0002524138494498981,
      "loss": 2.2429,
      "step": 126884
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.043840765953064,
      "learning_rate": 0.00025240981068019836,
      "loss": 2.1116,
      "step": 126885
    },
    {
      "epoch": 1.65,
      "grad_norm": 0.9258421063423157,
      "learning_rate": 0.0002524057719193465,
      "loss": 2.2437,
      "step": 126886
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.038784146308899,
      "learning_rate": 0.0002524017331673433,
      "loss": 2.229,
      "step": 126887
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.228025197982788,
      "learning_rate": 0.00025239769442418955,
      "loss": 2.2373,
      "step": 126888
    },
    {
      "epoch": 1.65,
      "grad_norm": 0.9996718764305115,
      "learning_rate": 0.0002523936556898859,
      "loss": 2.2736,
      "step": 126889
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.168260097503662,
      "learning_rate": 0.0002523896169644332,
      "loss": 2.1276,
      "step": 126890
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.0647400617599487,
      "learning_rate": 0.00025238557824783226,
      "loss": 2.0838,
      "step": 126891
    },
    {
      "epoch": 1.65,
      "grad_norm": 0.9632207155227661,
      "learning_rate": 0.00025238153954008365,
      "loss": 2.1601,
      "step": 126892
    },
    {
      "epoch": 1.65,
      "grad_norm": 0.9344249963760376,
      "learning_rate": 0.0002523775008411882,
      "loss": 2.1007,
      "step": 126893
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.008354663848877,
      "learning_rate": 0.00025237346215114674,
      "loss": 2.0009,
      "step": 126894
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.0323258638381958,
      "learning_rate": 0.00025236942346995985,
      "loss": 2.1025,
      "step": 126895
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.0393321514129639,
      "learning_rate": 0.00025236538479762844,
      "loss": 2.2021,
      "step": 126896
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.0904533863067627,
      "learning_rate": 0.00025236134613415323,
      "loss": 2.1571,
      "step": 126897
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.0432807207107544,
      "learning_rate": 0.0002523573074795349,
      "loss": 1.99,
      "step": 126898
    },
    {
      "epoch": 1.65,
      "grad_norm": 0.994752824306488,
      "learning_rate": 0.0002523532688337743,
      "loss": 1.976,
      "step": 126899
    },
    {
      "epoch": 1.65,
      "grad_norm": 0.9879215359687805,
      "learning_rate": 0.000252349230196872,
      "loss": 2.1436,
      "step": 126900
    },
    {
      "epoch": 1.65,
      "grad_norm": 0.9597719311714172,
      "learning_rate": 0.000252345191568829,
      "loss": 2.3232,
      "step": 126901
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.0048202276229858,
      "learning_rate": 0.0002523411529496459,
      "loss": 2.2465,
      "step": 126902
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.1092712879180908,
      "learning_rate": 0.00025233711433932354,
      "loss": 1.9474,
      "step": 126903
    },
    {
      "epoch": 1.65,
      "grad_norm": 0.9770594835281372,
      "learning_rate": 0.0002523330757378625,
      "loss": 2.0505,
      "step": 126904
    },
    {
      "epoch": 1.65,
      "grad_norm": 0.9681779146194458,
      "learning_rate": 0.0002523290371452637,
      "loss": 2.2934,
      "step": 126905
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.0058521032333374,
      "learning_rate": 0.0002523249985615278,
      "loss": 1.9647,
      "step": 126906
    },
    {
      "epoch": 1.65,
      "grad_norm": 0.928397536277771,
      "learning_rate": 0.0002523209599866556,
      "loss": 2.1147,
      "step": 126907
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.0418376922607422,
      "learning_rate": 0.0002523169214206478,
      "loss": 2.3322,
      "step": 126908
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.0407263040542603,
      "learning_rate": 0.00025231288286350526,
      "loss": 2.3933,
      "step": 126909
    },
    {
      "epoch": 1.65,
      "grad_norm": 0.9759379625320435,
      "learning_rate": 0.0002523088443152286,
      "loss": 1.9109,
      "step": 126910
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.1061087846755981,
      "learning_rate": 0.00025230480577581864,
      "loss": 2.2981,
      "step": 126911
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.0622050762176514,
      "learning_rate": 0.0002523007672452761,
      "loss": 2.1447,
      "step": 126912
    },
    {
      "epoch": 1.65,
      "grad_norm": 0.9246287941932678,
      "learning_rate": 0.00025229672872360175,
      "loss": 1.978,
      "step": 126913
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.0230720043182373,
      "learning_rate": 0.00025229269021079633,
      "loss": 2.0526,
      "step": 126914
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.0418094396591187,
      "learning_rate": 0.00025228865170686066,
      "loss": 2.1235,
      "step": 126915
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.0951858758926392,
      "learning_rate": 0.0002522846132117955,
      "loss": 2.2913,
      "step": 126916
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.0599560737609863,
      "learning_rate": 0.00025228057472560133,
      "loss": 2.13,
      "step": 126917
    },
    {
      "epoch": 1.65,
      "grad_norm": 0.9437960982322693,
      "learning_rate": 0.0002522765362482792,
      "loss": 2.1635,
      "step": 126918
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.0282764434814453,
      "learning_rate": 0.0002522724977798297,
      "loss": 2.1753,
      "step": 126919
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.016500473022461,
      "learning_rate": 0.0002522684593202537,
      "loss": 2.0298,
      "step": 126920
    },
    {
      "epoch": 1.65,
      "grad_norm": 0.9665870070457458,
      "learning_rate": 0.0002522644208695519,
      "loss": 2.106,
      "step": 126921
    },
    {
      "epoch": 1.65,
      "grad_norm": 0.9479079246520996,
      "learning_rate": 0.00025226038242772515,
      "loss": 2.2575,
      "step": 126922
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.0055772066116333,
      "learning_rate": 0.000252256343994774,
      "loss": 2.2665,
      "step": 126923
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.054319143295288,
      "learning_rate": 0.00025225230557069923,
      "loss": 2.2768,
      "step": 126924
    },
    {
      "epoch": 1.65,
      "grad_norm": 0.9453644752502441,
      "learning_rate": 0.0002522482671555017,
      "loss": 2.0578,
      "step": 126925
    },
    {
      "epoch": 1.65,
      "grad_norm": 0.9839304685592651,
      "learning_rate": 0.00025224422874918213,
      "loss": 2.0913,
      "step": 126926
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.0196589231491089,
      "learning_rate": 0.0002522401903517412,
      "loss": 1.9845,
      "step": 126927
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.002787709236145,
      "learning_rate": 0.0002522361519631799,
      "loss": 1.9595,
      "step": 126928
    },
    {
      "epoch": 1.65,
      "grad_norm": 0.9441083073616028,
      "learning_rate": 0.00025223211358349864,
      "loss": 2.1838,
      "step": 126929
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.0277632474899292,
      "learning_rate": 0.0002522280752126984,
      "loss": 2.2429,
      "step": 126930
    },
    {
      "epoch": 1.65,
      "grad_norm": 0.94476318359375,
      "learning_rate": 0.0002522240368507798,
      "loss": 2.0483,
      "step": 126931
    },
    {
      "epoch": 1.65,
      "grad_norm": 0.9240646958351135,
      "learning_rate": 0.00025221999849774367,
      "loss": 2.0588,
      "step": 126932
    },
    {
      "epoch": 1.65,
      "grad_norm": 0.9395429491996765,
      "learning_rate": 0.00025221596015359077,
      "loss": 2.1657,
      "step": 126933
    },
    {
      "epoch": 1.65,
      "grad_norm": 0.9249060750007629,
      "learning_rate": 0.00025221192181832194,
      "loss": 1.9413,
      "step": 126934
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.0885452032089233,
      "learning_rate": 0.00025220788349193766,
      "loss": 2.3181,
      "step": 126935
    },
    {
      "epoch": 1.65,
      "grad_norm": 0.9337298274040222,
      "learning_rate": 0.00025220384517443884,
      "loss": 2.202,
      "step": 126936
    },
    {
      "epoch": 1.65,
      "grad_norm": 0.9478390216827393,
      "learning_rate": 0.0002521998068658262,
      "loss": 2.1273,
      "step": 126937
    },
    {
      "epoch": 1.65,
      "grad_norm": 0.9735973477363586,
      "learning_rate": 0.00025219576856610055,
      "loss": 2.0276,
      "step": 126938
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.162888526916504,
      "learning_rate": 0.00025219173027526265,
      "loss": 2.1741,
      "step": 126939
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.059926152229309,
      "learning_rate": 0.0002521876919933133,
      "loss": 2.1337,
      "step": 126940
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.1078566312789917,
      "learning_rate": 0.00025218365372025307,
      "loss": 2.4158,
      "step": 126941
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.362585186958313,
      "learning_rate": 0.00025217961545608274,
      "loss": 2.0364,
      "step": 126942
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.0623294115066528,
      "learning_rate": 0.00025217557720080315,
      "loss": 2.2782,
      "step": 126943
    },
    {
      "epoch": 1.65,
      "grad_norm": 0.9600574374198914,
      "learning_rate": 0.00025217153895441504,
      "loss": 2.2423,
      "step": 126944
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.0758880376815796,
      "learning_rate": 0.0002521675007169191,
      "loss": 1.9845,
      "step": 126945
    },
    {
      "epoch": 1.65,
      "grad_norm": 0.9858421087265015,
      "learning_rate": 0.0002521634624883163,
      "loss": 1.9631,
      "step": 126946
    },
    {
      "epoch": 1.65,
      "grad_norm": 0.9409645199775696,
      "learning_rate": 0.00025215942426860707,
      "loss": 2.0644,
      "step": 126947
    },
    {
      "epoch": 1.65,
      "grad_norm": 0.983853816986084,
      "learning_rate": 0.0002521553860577923,
      "loss": 1.8752,
      "step": 126948
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.041719675064087,
      "learning_rate": 0.0002521513478558728,
      "loss": 2.2838,
      "step": 126949
    },
    {
      "epoch": 1.65,
      "grad_norm": 0.9759413599967957,
      "learning_rate": 0.00025214730966284917,
      "loss": 2.0472,
      "step": 126950
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.0907187461853027,
      "learning_rate": 0.0002521432714787223,
      "loss": 2.3787,
      "step": 126951
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.0666133165359497,
      "learning_rate": 0.000252139233303493,
      "loss": 2.1037,
      "step": 126952
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.0239123106002808,
      "learning_rate": 0.00025213519513716187,
      "loss": 2.1433,
      "step": 126953
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.1616804599761963,
      "learning_rate": 0.0002521311569797297,
      "loss": 2.3509,
      "step": 126954
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.2827590703964233,
      "learning_rate": 0.0002521271188311972,
      "loss": 2.1766,
      "step": 126955
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.0843218564987183,
      "learning_rate": 0.0002521230806915652,
      "loss": 1.992,
      "step": 126956
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.0157921314239502,
      "learning_rate": 0.0002521190425608344,
      "loss": 1.9716,
      "step": 126957
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.1684331893920898,
      "learning_rate": 0.00025211500443900566,
      "loss": 2.1331,
      "step": 126958
    },
    {
      "epoch": 1.65,
      "grad_norm": 0.9884620308876038,
      "learning_rate": 0.0002521109663260796,
      "loss": 2.0326,
      "step": 126959
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.0000405311584473,
      "learning_rate": 0.000252106928222057,
      "loss": 2.0056,
      "step": 126960
    },
    {
      "epoch": 1.65,
      "grad_norm": 0.9850378632545471,
      "learning_rate": 0.0002521028901269386,
      "loss": 2.2849,
      "step": 126961
    },
    {
      "epoch": 1.65,
      "grad_norm": 0.968532919883728,
      "learning_rate": 0.00025209885204072516,
      "loss": 2.1077,
      "step": 126962
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.6933956146240234,
      "learning_rate": 0.00025209481396341746,
      "loss": 1.9463,
      "step": 126963
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.0397062301635742,
      "learning_rate": 0.00025209077589501643,
      "loss": 2.2841,
      "step": 126964
    },
    {
      "epoch": 1.65,
      "grad_norm": 0.9610018134117126,
      "learning_rate": 0.00025208673783552244,
      "loss": 2.0811,
      "step": 126965
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.0766464471817017,
      "learning_rate": 0.0002520826997849365,
      "loss": 2.1226,
      "step": 126966
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.1183700561523438,
      "learning_rate": 0.0002520786617432592,
      "loss": 2.0869,
      "step": 126967
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.0567415952682495,
      "learning_rate": 0.00025207462371049143,
      "loss": 2.19,
      "step": 126968
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.0038522481918335,
      "learning_rate": 0.00025207058568663387,
      "loss": 2.0678,
      "step": 126969
    },
    {
      "epoch": 1.65,
      "grad_norm": 0.9743505716323853,
      "learning_rate": 0.00025206654767168736,
      "loss": 2.2132,
      "step": 126970
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.0299923419952393,
      "learning_rate": 0.00025206250966565256,
      "loss": 2.3946,
      "step": 126971
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.090183973312378,
      "learning_rate": 0.00025205847166853023,
      "loss": 2.0377,
      "step": 126972
    },
    {
      "epoch": 1.65,
      "grad_norm": 0.9496680498123169,
      "learning_rate": 0.00025205443368032113,
      "loss": 2.4642,
      "step": 126973
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.14396333694458,
      "learning_rate": 0.00025205039570102596,
      "loss": 1.9973,
      "step": 126974
    },
    {
      "epoch": 1.65,
      "grad_norm": 0.9994299411773682,
      "learning_rate": 0.0002520463577306456,
      "loss": 2.1097,
      "step": 126975
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.156472086906433,
      "learning_rate": 0.00025204231976918076,
      "loss": 2.2009,
      "step": 126976
    },
    {
      "epoch": 1.65,
      "grad_norm": 0.9926065802574158,
      "learning_rate": 0.0002520382818166321,
      "loss": 1.9339,
      "step": 126977
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.174264669418335,
      "learning_rate": 0.0002520342438730004,
      "loss": 2.1996,
      "step": 126978
    },
    {
      "epoch": 1.65,
      "grad_norm": 0.9206939935684204,
      "learning_rate": 0.00025203020593828654,
      "loss": 1.9043,
      "step": 126979
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.1426100730895996,
      "learning_rate": 0.00025202616801249105,
      "loss": 2.1423,
      "step": 126980
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.070412039756775,
      "learning_rate": 0.00025202213009561486,
      "loss": 1.8791,
      "step": 126981
    },
    {
      "epoch": 1.65,
      "grad_norm": 0.9524610638618469,
      "learning_rate": 0.0002520180921876586,
      "loss": 2.1248,
      "step": 126982
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.067147970199585,
      "learning_rate": 0.00025201405428862324,
      "loss": 2.1208,
      "step": 126983
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.1091444492340088,
      "learning_rate": 0.0002520100163985092,
      "loss": 2.356,
      "step": 126984
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.0077685117721558,
      "learning_rate": 0.0002520059785173175,
      "loss": 2.2792,
      "step": 126985
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.018847942352295,
      "learning_rate": 0.00025200194064504885,
      "loss": 2.0371,
      "step": 126986
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.0085928440093994,
      "learning_rate": 0.00025199790278170384,
      "loss": 2.2329,
      "step": 126987
    },
    {
      "epoch": 1.65,
      "grad_norm": 0.9250163435935974,
      "learning_rate": 0.00025199386492728333,
      "loss": 2.0508,
      "step": 126988
    },
    {
      "epoch": 1.65,
      "grad_norm": 0.9572675228118896,
      "learning_rate": 0.00025198982708178814,
      "loss": 2.0166,
      "step": 126989
    },
    {
      "epoch": 1.65,
      "grad_norm": 0.9809667468070984,
      "learning_rate": 0.00025198578924521886,
      "loss": 2.13,
      "step": 126990
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.0123357772827148,
      "learning_rate": 0.00025198175141757635,
      "loss": 2.1478,
      "step": 126991
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.0635578632354736,
      "learning_rate": 0.00025197771359886133,
      "loss": 2.032,
      "step": 126992
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.0742521286010742,
      "learning_rate": 0.00025197367578907466,
      "loss": 2.0536,
      "step": 126993
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.0323048830032349,
      "learning_rate": 0.00025196963798821694,
      "loss": 2.1272,
      "step": 126994
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.076835036277771,
      "learning_rate": 0.000251965600196289,
      "loss": 2.2002,
      "step": 126995
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.0132989883422852,
      "learning_rate": 0.00025196156241329143,
      "loss": 2.146,
      "step": 126996
    },
    {
      "epoch": 1.65,
      "grad_norm": 0.9302277565002441,
      "learning_rate": 0.00025195752463922516,
      "loss": 1.9345,
      "step": 126997
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.0910556316375732,
      "learning_rate": 0.0002519534868740909,
      "loss": 2.2911,
      "step": 126998
    },
    {
      "epoch": 1.65,
      "grad_norm": 0.9802746772766113,
      "learning_rate": 0.0002519494491178894,
      "loss": 2.1201,
      "step": 126999
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.2316560745239258,
      "learning_rate": 0.00025194541137062143,
      "loss": 2.35,
      "step": 127000
    },
    {
      "epoch": 1.65,
      "grad_norm": 0.983401894569397,
      "learning_rate": 0.0002519413736322878,
      "loss": 1.9245,
      "step": 127001
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.0263112783432007,
      "learning_rate": 0.0002519373359028891,
      "loss": 2.0188,
      "step": 127002
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.018637776374817,
      "learning_rate": 0.0002519332981824261,
      "loss": 1.9575,
      "step": 127003
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.0115482807159424,
      "learning_rate": 0.00025192926047089964,
      "loss": 2.0433,
      "step": 127004
    },
    {
      "epoch": 1.65,
      "grad_norm": 0.9806915521621704,
      "learning_rate": 0.00025192522276831043,
      "loss": 2.129,
      "step": 127005
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.017935037612915,
      "learning_rate": 0.0002519211850746592,
      "loss": 2.1641,
      "step": 127006
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.1077265739440918,
      "learning_rate": 0.0002519171473899469,
      "loss": 2.2891,
      "step": 127007
    },
    {
      "epoch": 1.65,
      "grad_norm": 0.973015546798706,
      "learning_rate": 0.000251913109714174,
      "loss": 2.3337,
      "step": 127008
    },
    {
      "epoch": 1.65,
      "grad_norm": 0.993959367275238,
      "learning_rate": 0.0002519090720473413,
      "loss": 1.9768,
      "step": 127009
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.0682170391082764,
      "learning_rate": 0.00025190503438944964,
      "loss": 2.1818,
      "step": 127010
    },
    {
      "epoch": 1.65,
      "grad_norm": 0.9409077167510986,
      "learning_rate": 0.00025190099674049976,
      "loss": 2.1433,
      "step": 127011
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.0270626544952393,
      "learning_rate": 0.0002518969591004923,
      "loss": 2.115,
      "step": 127012
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.017073631286621,
      "learning_rate": 0.00025189292146942836,
      "loss": 2.035,
      "step": 127013
    },
    {
      "epoch": 1.65,
      "grad_norm": 0.9176754951477051,
      "learning_rate": 0.00025188888384730825,
      "loss": 2.1805,
      "step": 127014
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.08042573928833,
      "learning_rate": 0.0002518848462341329,
      "loss": 1.9074,
      "step": 127015
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.0207040309906006,
      "learning_rate": 0.00025188080862990306,
      "loss": 1.9678,
      "step": 127016
    },
    {
      "epoch": 1.65,
      "grad_norm": 0.989107608795166,
      "learning_rate": 0.0002518767710346195,
      "loss": 1.9698,
      "step": 127017
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.0555487871170044,
      "learning_rate": 0.00025187273344828295,
      "loss": 2.0123,
      "step": 127018
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.2553324699401855,
      "learning_rate": 0.0002518686958708943,
      "loss": 2.1301,
      "step": 127019
    },
    {
      "epoch": 1.65,
      "grad_norm": 0.9270657300949097,
      "learning_rate": 0.0002518646583024541,
      "loss": 2.0033,
      "step": 127020
    },
    {
      "epoch": 1.65,
      "grad_norm": 0.9434601068496704,
      "learning_rate": 0.00025186062074296307,
      "loss": 1.9693,
      "step": 127021
    },
    {
      "epoch": 1.65,
      "grad_norm": 0.9230366945266724,
      "learning_rate": 0.0002518565831924221,
      "loss": 2.3393,
      "step": 127022
    },
    {
      "epoch": 1.65,
      "grad_norm": 0.952301025390625,
      "learning_rate": 0.00025185254565083186,
      "loss": 2.3166,
      "step": 127023
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.0889776945114136,
      "learning_rate": 0.0002518485081181932,
      "loss": 2.257,
      "step": 127024
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.0986071825027466,
      "learning_rate": 0.0002518444705945069,
      "loss": 2.0417,
      "step": 127025
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.0431104898452759,
      "learning_rate": 0.0002518404330797735,
      "loss": 2.2164,
      "step": 127026
    },
    {
      "epoch": 1.65,
      "grad_norm": 0.9878475069999695,
      "learning_rate": 0.0002518363955739939,
      "loss": 2.195,
      "step": 127027
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.0362437963485718,
      "learning_rate": 0.0002518323580771688,
      "loss": 2.1209,
      "step": 127028
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.061326026916504,
      "learning_rate": 0.000251828320589299,
      "loss": 1.8795,
      "step": 127029
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.1327474117279053,
      "learning_rate": 0.00025182428311038517,
      "loss": 2.0052,
      "step": 127030
    },
    {
      "epoch": 1.65,
      "grad_norm": 0.9312437772750854,
      "learning_rate": 0.0002518202456404283,
      "loss": 2.1523,
      "step": 127031
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.0643612146377563,
      "learning_rate": 0.0002518162081794288,
      "loss": 2.2336,
      "step": 127032
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.0945299863815308,
      "learning_rate": 0.00025181217072738754,
      "loss": 2.0476,
      "step": 127033
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.0406965017318726,
      "learning_rate": 0.00025180813328430537,
      "loss": 2.1384,
      "step": 127034
    },
    {
      "epoch": 1.65,
      "grad_norm": 0.9861131906509399,
      "learning_rate": 0.000251804095850183,
      "loss": 2.3215,
      "step": 127035
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.0118697881698608,
      "learning_rate": 0.00025180005842502106,
      "loss": 2.131,
      "step": 127036
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.1093875169754028,
      "learning_rate": 0.0002517960210088206,
      "loss": 1.999,
      "step": 127037
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.1054168939590454,
      "learning_rate": 0.00025179198360158203,
      "loss": 2.213,
      "step": 127038
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.0097020864486694,
      "learning_rate": 0.0002517879462033062,
      "loss": 2.2502,
      "step": 127039
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.0494951009750366,
      "learning_rate": 0.00025178390881399395,
      "loss": 2.0984,
      "step": 127040
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.2217313051223755,
      "learning_rate": 0.000251779871433646,
      "loss": 2.1394,
      "step": 127041
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.0521509647369385,
      "learning_rate": 0.00025177583406226303,
      "loss": 2.1992,
      "step": 127042
    },
    {
      "epoch": 1.65,
      "grad_norm": 0.9135463237762451,
      "learning_rate": 0.000251771796699846,
      "loss": 1.907,
      "step": 127043
    },
    {
      "epoch": 1.65,
      "grad_norm": 0.9932926297187805,
      "learning_rate": 0.00025176775934639535,
      "loss": 2.0673,
      "step": 127044
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.0694164037704468,
      "learning_rate": 0.000251763722001912,
      "loss": 2.2951,
      "step": 127045
    },
    {
      "epoch": 1.65,
      "grad_norm": 0.988997220993042,
      "learning_rate": 0.00025175968466639666,
      "loss": 2.1068,
      "step": 127046
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.0621728897094727,
      "learning_rate": 0.0002517556473398501,
      "loss": 2.1737,
      "step": 127047
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.01934814453125,
      "learning_rate": 0.00025175161002227313,
      "loss": 2.0821,
      "step": 127048
    },
    {
      "epoch": 1.65,
      "grad_norm": 0.974865734577179,
      "learning_rate": 0.00025174757271366645,
      "loss": 2.186,
      "step": 127049
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.168789029121399,
      "learning_rate": 0.00025174353541403084,
      "loss": 2.1112,
      "step": 127050
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.0609948635101318,
      "learning_rate": 0.000251739498123367,
      "loss": 2.1224,
      "step": 127051
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.1472152471542358,
      "learning_rate": 0.00025173546084167564,
      "loss": 2.3538,
      "step": 127052
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.0406500101089478,
      "learning_rate": 0.00025173142356895757,
      "loss": 2.1697,
      "step": 127053
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.1042321920394897,
      "learning_rate": 0.00025172738630521355,
      "loss": 2.0493,
      "step": 127054
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.060967206954956,
      "learning_rate": 0.00025172334905044434,
      "loss": 2.2354,
      "step": 127055
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.141768455505371,
      "learning_rate": 0.00025171931180465075,
      "loss": 2.0172,
      "step": 127056
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.0328359603881836,
      "learning_rate": 0.0002517152745678334,
      "loss": 2.1139,
      "step": 127057
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.052045464515686,
      "learning_rate": 0.00025171123733999304,
      "loss": 2.0947,
      "step": 127058
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.0880407094955444,
      "learning_rate": 0.00025170720012113044,
      "loss": 2.2235,
      "step": 127059
    },
    {
      "epoch": 1.65,
      "grad_norm": 0.9759948253631592,
      "learning_rate": 0.00025170316291124644,
      "loss": 2.1022,
      "step": 127060
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.0326135158538818,
      "learning_rate": 0.0002516991257103417,
      "loss": 2.2605,
      "step": 127061
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.0132523775100708,
      "learning_rate": 0.0002516950885184171,
      "loss": 2.0248,
      "step": 127062
    },
    {
      "epoch": 1.65,
      "grad_norm": 0.9651266932487488,
      "learning_rate": 0.00025169105133547327,
      "loss": 2.3056,
      "step": 127063
    },
    {
      "epoch": 1.65,
      "grad_norm": 0.9383676648139954,
      "learning_rate": 0.0002516870141615109,
      "loss": 2.1724,
      "step": 127064
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.034699559211731,
      "learning_rate": 0.00025168297699653086,
      "loss": 2.1744,
      "step": 127065
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.0881683826446533,
      "learning_rate": 0.0002516789398405339,
      "loss": 2.1668,
      "step": 127066
    },
    {
      "epoch": 1.65,
      "grad_norm": 0.9522070288658142,
      "learning_rate": 0.00025167490269352064,
      "loss": 2.082,
      "step": 127067
    },
    {
      "epoch": 1.65,
      "grad_norm": 0.9963680505752563,
      "learning_rate": 0.0002516708655554921,
      "loss": 2.2057,
      "step": 127068
    },
    {
      "epoch": 1.65,
      "grad_norm": 0.9521870017051697,
      "learning_rate": 0.0002516668284264487,
      "loss": 2.084,
      "step": 127069
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.019804835319519,
      "learning_rate": 0.00025166279130639144,
      "loss": 2.3229,
      "step": 127070
    },
    {
      "epoch": 1.65,
      "grad_norm": 0.9566962122917175,
      "learning_rate": 0.000251658754195321,
      "loss": 1.9864,
      "step": 127071
    },
    {
      "epoch": 1.65,
      "grad_norm": 0.9868193864822388,
      "learning_rate": 0.000251654717093238,
      "loss": 1.9815,
      "step": 127072
    },
    {
      "epoch": 1.65,
      "grad_norm": 0.990959882736206,
      "learning_rate": 0.00025165068000014334,
      "loss": 2.3427,
      "step": 127073
    },
    {
      "epoch": 1.65,
      "grad_norm": 0.9278482794761658,
      "learning_rate": 0.0002516466429160378,
      "loss": 2.1261,
      "step": 127074
    },
    {
      "epoch": 1.65,
      "grad_norm": 0.9626252055168152,
      "learning_rate": 0.000251642605840922,
      "loss": 1.9087,
      "step": 127075
    },
    {
      "epoch": 1.65,
      "grad_norm": 0.995221734046936,
      "learning_rate": 0.00025163856877479673,
      "loss": 2.1178,
      "step": 127076
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.0368695259094238,
      "learning_rate": 0.00025163453171766283,
      "loss": 1.9492,
      "step": 127077
    },
    {
      "epoch": 1.65,
      "grad_norm": 0.9686946272850037,
      "learning_rate": 0.000251630494669521,
      "loss": 2.0364,
      "step": 127078
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.0508968830108643,
      "learning_rate": 0.00025162645763037193,
      "loss": 2.2615,
      "step": 127079
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.024225115776062,
      "learning_rate": 0.00025162242060021645,
      "loss": 2.316,
      "step": 127080
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.0113046169281006,
      "learning_rate": 0.00025161838357905524,
      "loss": 2.3614,
      "step": 127081
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.0144811868667603,
      "learning_rate": 0.000251614346566889,
      "loss": 1.9339,
      "step": 127082
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.0173548460006714,
      "learning_rate": 0.0002516103095637186,
      "loss": 2.1542,
      "step": 127083
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.0614558458328247,
      "learning_rate": 0.00025160627256954487,
      "loss": 1.9218,
      "step": 127084
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.1070902347564697,
      "learning_rate": 0.0002516022355843684,
      "loss": 2.2424,
      "step": 127085
    },
    {
      "epoch": 1.65,
      "grad_norm": 0.953124463558197,
      "learning_rate": 0.00025159819860819,
      "loss": 2.1506,
      "step": 127086
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.2185837030410767,
      "learning_rate": 0.00025159416164101037,
      "loss": 2.2127,
      "step": 127087
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.086507797241211,
      "learning_rate": 0.0002515901246828303,
      "loss": 2.0686,
      "step": 127088
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.0921462774276733,
      "learning_rate": 0.0002515860877336505,
      "loss": 2.1996,
      "step": 127089
    },
    {
      "epoch": 1.65,
      "grad_norm": 0.940883219242096,
      "learning_rate": 0.00025158205079347184,
      "loss": 2.2268,
      "step": 127090
    },
    {
      "epoch": 1.65,
      "grad_norm": 0.9708134531974792,
      "learning_rate": 0.00025157801386229496,
      "loss": 2.3024,
      "step": 127091
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.1062440872192383,
      "learning_rate": 0.00025157397694012076,
      "loss": 2.2791,
      "step": 127092
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.079251766204834,
      "learning_rate": 0.0002515699400269497,
      "loss": 2.0757,
      "step": 127093
    },
    {
      "epoch": 1.65,
      "grad_norm": 0.9582960605621338,
      "learning_rate": 0.00025156590312278276,
      "loss": 2.1202,
      "step": 127094
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.1503819227218628,
      "learning_rate": 0.0002515618662276206,
      "loss": 2.1873,
      "step": 127095
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.1619367599487305,
      "learning_rate": 0.000251557829341464,
      "loss": 2.0775,
      "step": 127096
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.2466272115707397,
      "learning_rate": 0.0002515537924643137,
      "loss": 2.099,
      "step": 127097
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.0330322980880737,
      "learning_rate": 0.00025154975559617065,
      "loss": 2.2889,
      "step": 127098
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.0821349620819092,
      "learning_rate": 0.00025154571873703527,
      "loss": 2.4011,
      "step": 127099
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.497495412826538,
      "learning_rate": 0.0002515416818869084,
      "loss": 1.9975,
      "step": 127100
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.1212407350540161,
      "learning_rate": 0.0002515376450457909,
      "loss": 2.1744,
      "step": 127101
    },
    {
      "epoch": 1.65,
      "grad_norm": 0.9862462282180786,
      "learning_rate": 0.0002515336082136835,
      "loss": 1.9285,
      "step": 127102
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.1316337585449219,
      "learning_rate": 0.0002515295713905868,
      "loss": 2.2685,
      "step": 127103
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.2130661010742188,
      "learning_rate": 0.00025152553457650194,
      "loss": 1.9372,
      "step": 127104
    },
    {
      "epoch": 1.65,
      "grad_norm": 1.0798300504684448,
      "learning_rate": 0.00025152149777142916,
      "loss": 2.0561,
      "step": 127105
    },
    {
      "epoch": 1.66,
      "grad_norm": 0.8822575211524963,
      "learning_rate": 0.0002515174609753695,
      "loss": 2.0845,
      "step": 127106
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.083823561668396,
      "learning_rate": 0.00025151342418832366,
      "loss": 2.1136,
      "step": 127107
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.0724107027053833,
      "learning_rate": 0.00025150938741029234,
      "loss": 2.2082,
      "step": 127108
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.258700966835022,
      "learning_rate": 0.0002515053506412764,
      "loss": 2.4063,
      "step": 127109
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.165740966796875,
      "learning_rate": 0.00025150131388127666,
      "loss": 1.9812,
      "step": 127110
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.306605339050293,
      "learning_rate": 0.00025149727713029353,
      "loss": 2.2194,
      "step": 127111
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.0216667652130127,
      "learning_rate": 0.00025149324038832804,
      "loss": 2.1719,
      "step": 127112
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.1319911479949951,
      "learning_rate": 0.0002514892036553809,
      "loss": 1.8734,
      "step": 127113
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.0507365465164185,
      "learning_rate": 0.0002514851669314528,
      "loss": 2.2561,
      "step": 127114
    },
    {
      "epoch": 1.66,
      "grad_norm": 0.9785583019256592,
      "learning_rate": 0.0002514811302165445,
      "loss": 2.2597,
      "step": 127115
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.320395588874817,
      "learning_rate": 0.0002514770935106568,
      "loss": 2.2143,
      "step": 127116
    },
    {
      "epoch": 1.66,
      "grad_norm": 0.9457656741142273,
      "learning_rate": 0.00025147305681379054,
      "loss": 2.0665,
      "step": 127117
    },
    {
      "epoch": 1.66,
      "grad_norm": 0.9087268710136414,
      "learning_rate": 0.00025146902012594625,
      "loss": 2.0396,
      "step": 127118
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.0754486322402954,
      "learning_rate": 0.0002514649834471247,
      "loss": 2.3191,
      "step": 127119
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.0514358282089233,
      "learning_rate": 0.00025146094677732685,
      "loss": 2.1422,
      "step": 127120
    },
    {
      "epoch": 1.66,
      "grad_norm": 0.9835528135299683,
      "learning_rate": 0.00025145691011655325,
      "loss": 2.0053,
      "step": 127121
    },
    {
      "epoch": 1.66,
      "grad_norm": 0.8955249786376953,
      "learning_rate": 0.0002514528734648047,
      "loss": 2.1966,
      "step": 127122
    },
    {
      "epoch": 1.66,
      "grad_norm": 0.9055892825126648,
      "learning_rate": 0.0002514488368220822,
      "loss": 2.182,
      "step": 127123
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.048175573348999,
      "learning_rate": 0.0002514448001883861,
      "loss": 2.1011,
      "step": 127124
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.1936379671096802,
      "learning_rate": 0.0002514407635637173,
      "loss": 2.4802,
      "step": 127125
    },
    {
      "epoch": 1.66,
      "grad_norm": 0.9544095993041992,
      "learning_rate": 0.0002514367269480766,
      "loss": 2.0095,
      "step": 127126
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.0077922344207764,
      "learning_rate": 0.00025143269034146477,
      "loss": 2.0059,
      "step": 127127
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.0535496473312378,
      "learning_rate": 0.0002514286537438825,
      "loss": 2.0587,
      "step": 127128
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.0930808782577515,
      "learning_rate": 0.0002514246171553307,
      "loss": 2.099,
      "step": 127129
    },
    {
      "epoch": 1.66,
      "grad_norm": 0.9719513058662415,
      "learning_rate": 0.0002514205805758098,
      "loss": 1.9832,
      "step": 127130
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.0035418272018433,
      "learning_rate": 0.00025141654400532074,
      "loss": 2.1137,
      "step": 127131
    },
    {
      "epoch": 1.66,
      "grad_norm": 0.9848194122314453,
      "learning_rate": 0.0002514125074438643,
      "loss": 2.4349,
      "step": 127132
    },
    {
      "epoch": 1.66,
      "grad_norm": 0.9226686358451843,
      "learning_rate": 0.0002514084708914412,
      "loss": 2.3489,
      "step": 127133
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.0197315216064453,
      "learning_rate": 0.0002514044343480521,
      "loss": 2.1062,
      "step": 127134
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.0640414953231812,
      "learning_rate": 0.00025140039781369804,
      "loss": 2.0928,
      "step": 127135
    },
    {
      "epoch": 1.66,
      "grad_norm": 0.974149227142334,
      "learning_rate": 0.0002513963612883794,
      "loss": 2.1412,
      "step": 127136
    },
    {
      "epoch": 1.66,
      "grad_norm": 0.9579893350601196,
      "learning_rate": 0.0002513923247720971,
      "loss": 2.1785,
      "step": 127137
    },
    {
      "epoch": 1.66,
      "grad_norm": 0.9793598651885986,
      "learning_rate": 0.0002513882882648519,
      "loss": 2.2713,
      "step": 127138
    },
    {
      "epoch": 1.66,
      "grad_norm": 0.9885503649711609,
      "learning_rate": 0.0002513842517666445,
      "loss": 2.3286,
      "step": 127139
    },
    {
      "epoch": 1.66,
      "grad_norm": 0.9624778628349304,
      "learning_rate": 0.0002513802152774757,
      "loss": 2.0609,
      "step": 127140
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.015760064125061,
      "learning_rate": 0.0002513761787973464,
      "loss": 2.0458,
      "step": 127141
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.014297604560852,
      "learning_rate": 0.000251372142326257,
      "loss": 2.1585,
      "step": 127142
    },
    {
      "epoch": 1.66,
      "grad_norm": 0.9221803545951843,
      "learning_rate": 0.0002513681058642085,
      "loss": 2.186,
      "step": 127143
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.0092713832855225,
      "learning_rate": 0.0002513640694112015,
      "loss": 2.0697,
      "step": 127144
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.0372672080993652,
      "learning_rate": 0.0002513600329672369,
      "loss": 2.0626,
      "step": 127145
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.06130850315094,
      "learning_rate": 0.00025135599653231536,
      "loss": 2.1786,
      "step": 127146
    },
    {
      "epoch": 1.66,
      "grad_norm": 0.9019129872322083,
      "learning_rate": 0.0002513519601064378,
      "loss": 1.9478,
      "step": 127147
    },
    {
      "epoch": 1.66,
      "grad_norm": 0.9530673623085022,
      "learning_rate": 0.00025134792368960475,
      "loss": 2.2796,
      "step": 127148
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.1383448839187622,
      "learning_rate": 0.000251343887281817,
      "loss": 2.2743,
      "step": 127149
    },
    {
      "epoch": 1.66,
      "grad_norm": 0.8588722348213196,
      "learning_rate": 0.0002513398508830753,
      "loss": 1.9619,
      "step": 127150
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.0218442678451538,
      "learning_rate": 0.0002513358144933805,
      "loss": 1.8078,
      "step": 127151
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.0218195915222168,
      "learning_rate": 0.00025133177811273325,
      "loss": 1.9118,
      "step": 127152
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.0671916007995605,
      "learning_rate": 0.0002513277417411344,
      "loss": 2.365,
      "step": 127153
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.2336169481277466,
      "learning_rate": 0.00025132370537858466,
      "loss": 2.182,
      "step": 127154
    },
    {
      "epoch": 1.66,
      "grad_norm": 0.9858661890029907,
      "learning_rate": 0.00025131966902508475,
      "loss": 2.0041,
      "step": 127155
    },
    {
      "epoch": 1.66,
      "grad_norm": 0.9724090099334717,
      "learning_rate": 0.00025131563268063537,
      "loss": 2.4104,
      "step": 127156
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.021721363067627,
      "learning_rate": 0.00025131159634523734,
      "loss": 2.086,
      "step": 127157
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.0393178462982178,
      "learning_rate": 0.00025130756001889143,
      "loss": 2.2605,
      "step": 127158
    },
    {
      "epoch": 1.66,
      "grad_norm": 0.9412559270858765,
      "learning_rate": 0.0002513035237015984,
      "loss": 2.1267,
      "step": 127159
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.0153727531433105,
      "learning_rate": 0.00025129948739335893,
      "loss": 1.9789,
      "step": 127160
    },
    {
      "epoch": 1.66,
      "grad_norm": 0.9838315844535828,
      "learning_rate": 0.00025129545109417375,
      "loss": 2.0196,
      "step": 127161
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.0374845266342163,
      "learning_rate": 0.0002512914148040437,
      "loss": 2.3189,
      "step": 127162
    },
    {
      "epoch": 1.66,
      "grad_norm": 0.9904906749725342,
      "learning_rate": 0.0002512873785229696,
      "loss": 2.054,
      "step": 127163
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.0452672243118286,
      "learning_rate": 0.00025128334225095196,
      "loss": 1.993,
      "step": 127164
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.0192513465881348,
      "learning_rate": 0.00025127930598799176,
      "loss": 2.0685,
      "step": 127165
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.0514510869979858,
      "learning_rate": 0.0002512752697340896,
      "loss": 2.0115,
      "step": 127166
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.0362142324447632,
      "learning_rate": 0.00025127123348924623,
      "loss": 2.1029,
      "step": 127167
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.1286191940307617,
      "learning_rate": 0.00025126719725346254,
      "loss": 1.8619,
      "step": 127168
    },
    {
      "epoch": 1.66,
      "grad_norm": 0.8822377920150757,
      "learning_rate": 0.0002512631610267392,
      "loss": 2.1913,
      "step": 127169
    },
    {
      "epoch": 1.66,
      "grad_norm": 0.9759023785591125,
      "learning_rate": 0.00025125912480907696,
      "loss": 2.2171,
      "step": 127170
    },
    {
      "epoch": 1.66,
      "grad_norm": 0.9441455602645874,
      "learning_rate": 0.0002512550886004766,
      "loss": 2.1286,
      "step": 127171
    },
    {
      "epoch": 1.66,
      "grad_norm": 0.989776611328125,
      "learning_rate": 0.00025125105240093876,
      "loss": 2.1326,
      "step": 127172
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.108215093612671,
      "learning_rate": 0.0002512470162104643,
      "loss": 2.057,
      "step": 127173
    },
    {
      "epoch": 1.66,
      "grad_norm": 0.9958747029304504,
      "learning_rate": 0.0002512429800290539,
      "loss": 2.0566,
      "step": 127174
    },
    {
      "epoch": 1.66,
      "grad_norm": 0.9517585635185242,
      "learning_rate": 0.0002512389438567084,
      "loss": 2.1983,
      "step": 127175
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.0139065980911255,
      "learning_rate": 0.0002512349076934284,
      "loss": 1.9987,
      "step": 127176
    },
    {
      "epoch": 1.66,
      "grad_norm": 0.9418671131134033,
      "learning_rate": 0.000251230871539215,
      "loss": 1.859,
      "step": 127177
    },
    {
      "epoch": 1.66,
      "grad_norm": 0.8984357714653015,
      "learning_rate": 0.0002512268353940685,
      "loss": 2.0651,
      "step": 127178
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.0647121667861938,
      "learning_rate": 0.00025122279925798987,
      "loss": 2.0317,
      "step": 127179
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.0099214315414429,
      "learning_rate": 0.0002512187631309799,
      "loss": 2.0036,
      "step": 127180
    },
    {
      "epoch": 1.66,
      "grad_norm": 0.9472309350967407,
      "learning_rate": 0.0002512147270130392,
      "loss": 2.0762,
      "step": 127181
    },
    {
      "epoch": 1.66,
      "grad_norm": 0.9980282783508301,
      "learning_rate": 0.00025121069090416865,
      "loss": 2.1675,
      "step": 127182
    },
    {
      "epoch": 1.66,
      "grad_norm": 0.8988077044487,
      "learning_rate": 0.0002512066548043689,
      "loss": 2.2461,
      "step": 127183
    },
    {
      "epoch": 1.66,
      "grad_norm": 0.9469637274742126,
      "learning_rate": 0.00025120261871364093,
      "loss": 2.103,
      "step": 127184
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.0291765928268433,
      "learning_rate": 0.0002511985826319852,
      "loss": 2.3149,
      "step": 127185
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.0719554424285889,
      "learning_rate": 0.00025119454655940255,
      "loss": 2.0725,
      "step": 127186
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.0211814641952515,
      "learning_rate": 0.00025119051049589375,
      "loss": 2.101,
      "step": 127187
    },
    {
      "epoch": 1.66,
      "grad_norm": 0.9865577220916748,
      "learning_rate": 0.00025118647444145957,
      "loss": 2.1926,
      "step": 127188
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.1954519748687744,
      "learning_rate": 0.00025118243839610075,
      "loss": 1.9688,
      "step": 127189
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.015799880027771,
      "learning_rate": 0.00025117840235981817,
      "loss": 2.1958,
      "step": 127190
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.0856871604919434,
      "learning_rate": 0.00025117436633261225,
      "loss": 2.0654,
      "step": 127191
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.146856427192688,
      "learning_rate": 0.000251170330314484,
      "loss": 2.0691,
      "step": 127192
    },
    {
      "epoch": 1.66,
      "grad_norm": 0.9766455888748169,
      "learning_rate": 0.00025116629430543407,
      "loss": 2.1507,
      "step": 127193
    },
    {
      "epoch": 1.66,
      "grad_norm": 0.9968920946121216,
      "learning_rate": 0.0002511622583054633,
      "loss": 1.8897,
      "step": 127194
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.0719770193099976,
      "learning_rate": 0.0002511582223145724,
      "loss": 2.2063,
      "step": 127195
    },
    {
      "epoch": 1.66,
      "grad_norm": 0.9589375853538513,
      "learning_rate": 0.00025115418633276217,
      "loss": 1.9796,
      "step": 127196
    },
    {
      "epoch": 1.66,
      "grad_norm": 0.9676638245582581,
      "learning_rate": 0.0002511501503600332,
      "loss": 2.0691,
      "step": 127197
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.0446302890777588,
      "learning_rate": 0.0002511461143963863,
      "loss": 2.3056,
      "step": 127198
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.3216794729232788,
      "learning_rate": 0.0002511420784418223,
      "loss": 2.0827,
      "step": 127199
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.1535273790359497,
      "learning_rate": 0.00025113804249634193,
      "loss": 2.1473,
      "step": 127200
    },
    {
      "epoch": 1.66,
      "grad_norm": 0.9412341713905334,
      "learning_rate": 0.00025113400655994586,
      "loss": 2.268,
      "step": 127201
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.107904076576233,
      "learning_rate": 0.0002511299706326351,
      "loss": 2.2005,
      "step": 127202
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.2130649089813232,
      "learning_rate": 0.00025112593471441003,
      "loss": 2.2572,
      "step": 127203
    },
    {
      "epoch": 1.66,
      "grad_norm": 0.9763504266738892,
      "learning_rate": 0.00025112189880527157,
      "loss": 2.1749,
      "step": 127204
    },
    {
      "epoch": 1.66,
      "grad_norm": 0.905501127243042,
      "learning_rate": 0.0002511178629052205,
      "loss": 2.0251,
      "step": 127205
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.039710521697998,
      "learning_rate": 0.0002511138270142575,
      "loss": 2.1982,
      "step": 127206
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.330340027809143,
      "learning_rate": 0.0002511097911323834,
      "loss": 2.1375,
      "step": 127207
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.0434650182724,
      "learning_rate": 0.000251105755259599,
      "loss": 1.9217,
      "step": 127208
    },
    {
      "epoch": 1.66,
      "grad_norm": 0.9981725215911865,
      "learning_rate": 0.0002511017193959049,
      "loss": 2.3685,
      "step": 127209
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.2480595111846924,
      "learning_rate": 0.00025109768354130183,
      "loss": 2.0957,
      "step": 127210
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.5142015218734741,
      "learning_rate": 0.0002510936476957907,
      "loss": 2.2057,
      "step": 127211
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.0000801086425781,
      "learning_rate": 0.0002510896118593721,
      "loss": 2.0915,
      "step": 127212
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.18548583984375,
      "learning_rate": 0.00025108557603204694,
      "loss": 2.2729,
      "step": 127213
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.248338222503662,
      "learning_rate": 0.000251081540213816,
      "loss": 2.2633,
      "step": 127214
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.046721339225769,
      "learning_rate": 0.0002510775044046798,
      "loss": 2.2237,
      "step": 127215
    },
    {
      "epoch": 1.66,
      "grad_norm": 0.9909664988517761,
      "learning_rate": 0.00025107346860463924,
      "loss": 1.9845,
      "step": 127216
    },
    {
      "epoch": 1.66,
      "grad_norm": 0.9768615961074829,
      "learning_rate": 0.00025106943281369496,
      "loss": 2.302,
      "step": 127217
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.6788828372955322,
      "learning_rate": 0.0002510653970318479,
      "loss": 2.0106,
      "step": 127218
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.811147928237915,
      "learning_rate": 0.00025106136125909866,
      "loss": 2.0232,
      "step": 127219
    },
    {
      "epoch": 1.66,
      "grad_norm": 0.9392477869987488,
      "learning_rate": 0.00025105732549544815,
      "loss": 2.1039,
      "step": 127220
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.1489359140396118,
      "learning_rate": 0.0002510532897408969,
      "loss": 2.1951,
      "step": 127221
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.252512812614441,
      "learning_rate": 0.00025104925399544576,
      "loss": 1.9735,
      "step": 127222
    },
    {
      "epoch": 1.66,
      "grad_norm": 0.989300012588501,
      "learning_rate": 0.00025104521825909545,
      "loss": 2.3814,
      "step": 127223
    },
    {
      "epoch": 1.66,
      "grad_norm": 0.9661917686462402,
      "learning_rate": 0.0002510411825318468,
      "loss": 2.3171,
      "step": 127224
    },
    {
      "epoch": 1.66,
      "grad_norm": 0.883262574672699,
      "learning_rate": 0.00025103714681370054,
      "loss": 2.0659,
      "step": 127225
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.1556510925292969,
      "learning_rate": 0.0002510331111046575,
      "loss": 2.3521,
      "step": 127226
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.0509763956069946,
      "learning_rate": 0.0002510290754047182,
      "loss": 2.0451,
      "step": 127227
    },
    {
      "epoch": 1.66,
      "grad_norm": 0.9000186920166016,
      "learning_rate": 0.00025102503971388354,
      "loss": 2.0914,
      "step": 127228
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.0131996870040894,
      "learning_rate": 0.0002510210040321542,
      "loss": 2.0686,
      "step": 127229
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.0600842237472534,
      "learning_rate": 0.000251016968359531,
      "loss": 2.0051,
      "step": 127230
    },
    {
      "epoch": 1.66,
      "grad_norm": 0.8733171224594116,
      "learning_rate": 0.0002510129326960147,
      "loss": 2.0439,
      "step": 127231
    },
    {
      "epoch": 1.66,
      "grad_norm": 0.9533753991127014,
      "learning_rate": 0.0002510088970416061,
      "loss": 2.2518,
      "step": 127232
    },
    {
      "epoch": 1.66,
      "grad_norm": 0.9761272072792053,
      "learning_rate": 0.00025100486139630583,
      "loss": 2.1704,
      "step": 127233
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.2309975624084473,
      "learning_rate": 0.0002510008257601146,
      "loss": 2.0667,
      "step": 127234
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.3221514225006104,
      "learning_rate": 0.0002509967901330333,
      "loss": 2.1837,
      "step": 127235
    },
    {
      "epoch": 1.66,
      "grad_norm": 0.957362711429596,
      "learning_rate": 0.00025099275451506253,
      "loss": 2.1295,
      "step": 127236
    },
    {
      "epoch": 1.66,
      "grad_norm": 0.9615639448165894,
      "learning_rate": 0.00025098871890620323,
      "loss": 2.1714,
      "step": 127237
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.000300407409668,
      "learning_rate": 0.0002509846833064561,
      "loss": 2.179,
      "step": 127238
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.005167841911316,
      "learning_rate": 0.0002509806477158217,
      "loss": 1.9281,
      "step": 127239
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.085281491279602,
      "learning_rate": 0.00025097661213430107,
      "loss": 2.1725,
      "step": 127240
    },
    {
      "epoch": 1.66,
      "grad_norm": 0.9926105737686157,
      "learning_rate": 0.0002509725765618947,
      "loss": 2.061,
      "step": 127241
    },
    {
      "epoch": 1.66,
      "grad_norm": 0.8915013074874878,
      "learning_rate": 0.00025096854099860346,
      "loss": 1.9595,
      "step": 127242
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.0675655603408813,
      "learning_rate": 0.0002509645054444281,
      "loss": 2.2554,
      "step": 127243
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.054995059967041,
      "learning_rate": 0.00025096046989936947,
      "loss": 1.899,
      "step": 127244
    },
    {
      "epoch": 1.66,
      "grad_norm": 0.9508355259895325,
      "learning_rate": 0.00025095643436342813,
      "loss": 2.2284,
      "step": 127245
    },
    {
      "epoch": 1.66,
      "grad_norm": 0.9612363576889038,
      "learning_rate": 0.0002509523988366049,
      "loss": 2.0851,
      "step": 127246
    },
    {
      "epoch": 1.66,
      "grad_norm": 0.9472110271453857,
      "learning_rate": 0.0002509483633189006,
      "loss": 2.0398,
      "step": 127247
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.0793019533157349,
      "learning_rate": 0.00025094432781031585,
      "loss": 2.0629,
      "step": 127248
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.0076442956924438,
      "learning_rate": 0.00025094029231085153,
      "loss": 2.1756,
      "step": 127249
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.024396538734436,
      "learning_rate": 0.00025093625682050827,
      "loss": 2.329,
      "step": 127250
    },
    {
      "epoch": 1.66,
      "grad_norm": 0.9532731175422668,
      "learning_rate": 0.000250932221339287,
      "loss": 2.1769,
      "step": 127251
    },
    {
      "epoch": 1.66,
      "grad_norm": 0.9642764329910278,
      "learning_rate": 0.00025092818586718827,
      "loss": 2.1954,
      "step": 127252
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.0722581148147583,
      "learning_rate": 0.0002509241504042129,
      "loss": 2.0231,
      "step": 127253
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.0255131721496582,
      "learning_rate": 0.00025092011495036175,
      "loss": 1.9981,
      "step": 127254
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.0432286262512207,
      "learning_rate": 0.00025091607950563543,
      "loss": 2.0931,
      "step": 127255
    },
    {
      "epoch": 1.66,
      "grad_norm": 0.9715143442153931,
      "learning_rate": 0.0002509120440700347,
      "loss": 2.2549,
      "step": 127256
    },
    {
      "epoch": 1.66,
      "grad_norm": 0.9818980097770691,
      "learning_rate": 0.00025090800864356047,
      "loss": 2.336,
      "step": 127257
    },
    {
      "epoch": 1.66,
      "grad_norm": 0.9616906642913818,
      "learning_rate": 0.00025090397322621323,
      "loss": 2.2247,
      "step": 127258
    },
    {
      "epoch": 1.66,
      "grad_norm": 0.833970844745636,
      "learning_rate": 0.0002508999378179939,
      "loss": 2.0955,
      "step": 127259
    },
    {
      "epoch": 1.66,
      "grad_norm": 0.9686260223388672,
      "learning_rate": 0.00025089590241890317,
      "loss": 2.0747,
      "step": 127260
    },
    {
      "epoch": 1.66,
      "grad_norm": 0.9934205412864685,
      "learning_rate": 0.00025089186702894185,
      "loss": 2.0704,
      "step": 127261
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.0691295862197876,
      "learning_rate": 0.0002508878316481107,
      "loss": 2.3202,
      "step": 127262
    },
    {
      "epoch": 1.66,
      "grad_norm": 0.9312707185745239,
      "learning_rate": 0.00025088379627641045,
      "loss": 2.0069,
      "step": 127263
    },
    {
      "epoch": 1.66,
      "grad_norm": 0.9692535996437073,
      "learning_rate": 0.0002508797609138418,
      "loss": 2.0982,
      "step": 127264
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.062838077545166,
      "learning_rate": 0.00025087572556040544,
      "loss": 2.0612,
      "step": 127265
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.0715200901031494,
      "learning_rate": 0.00025087169021610227,
      "loss": 1.9244,
      "step": 127266
    },
    {
      "epoch": 1.66,
      "grad_norm": 0.9950087070465088,
      "learning_rate": 0.000250867654880933,
      "loss": 2.0572,
      "step": 127267
    },
    {
      "epoch": 1.66,
      "grad_norm": 0.9343177080154419,
      "learning_rate": 0.0002508636195548983,
      "loss": 2.0651,
      "step": 127268
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.1266372203826904,
      "learning_rate": 0.00025085958423799916,
      "loss": 2.4311,
      "step": 127269
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.0072214603424072,
      "learning_rate": 0.000250855548930236,
      "loss": 2.096,
      "step": 127270
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.0772570371627808,
      "learning_rate": 0.0002508515136316097,
      "loss": 2.0259,
      "step": 127271
    },
    {
      "epoch": 1.66,
      "grad_norm": 0.9714897274971008,
      "learning_rate": 0.00025084747834212107,
      "loss": 2.0528,
      "step": 127272
    },
    {
      "epoch": 1.66,
      "grad_norm": 0.9849013090133667,
      "learning_rate": 0.0002508434430617708,
      "loss": 2.1426,
      "step": 127273
    },
    {
      "epoch": 1.66,
      "grad_norm": 0.9910802245140076,
      "learning_rate": 0.00025083940779055966,
      "loss": 2.0722,
      "step": 127274
    },
    {
      "epoch": 1.66,
      "grad_norm": 0.9506489634513855,
      "learning_rate": 0.00025083537252848856,
      "loss": 2.2447,
      "step": 127275
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.0015186071395874,
      "learning_rate": 0.00025083133727555793,
      "loss": 1.9083,
      "step": 127276
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.2240612506866455,
      "learning_rate": 0.0002508273020317687,
      "loss": 1.7687,
      "step": 127277
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.0944139957427979,
      "learning_rate": 0.00025082326679712157,
      "loss": 2.0743,
      "step": 127278
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.0294009447097778,
      "learning_rate": 0.00025081923157161735,
      "loss": 2.1323,
      "step": 127279
    },
    {
      "epoch": 1.66,
      "grad_norm": 0.995083212852478,
      "learning_rate": 0.0002508151963552568,
      "loss": 2.1679,
      "step": 127280
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.0445064306259155,
      "learning_rate": 0.0002508111611480407,
      "loss": 2.2717,
      "step": 127281
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.01190984249115,
      "learning_rate": 0.0002508071259499696,
      "loss": 1.9205,
      "step": 127282
    },
    {
      "epoch": 1.66,
      "grad_norm": 0.9890876412391663,
      "learning_rate": 0.0002508030907610444,
      "loss": 2.1479,
      "step": 127283
    },
    {
      "epoch": 1.66,
      "grad_norm": 0.9923792481422424,
      "learning_rate": 0.0002507990555812659,
      "loss": 2.4397,
      "step": 127284
    },
    {
      "epoch": 1.66,
      "grad_norm": 0.9450333714485168,
      "learning_rate": 0.0002507950204106347,
      "loss": 2.0936,
      "step": 127285
    },
    {
      "epoch": 1.66,
      "grad_norm": 0.9753754734992981,
      "learning_rate": 0.0002507909852491517,
      "loss": 1.981,
      "step": 127286
    },
    {
      "epoch": 1.66,
      "grad_norm": 0.91289222240448,
      "learning_rate": 0.00025078695009681765,
      "loss": 2.1227,
      "step": 127287
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.010848045349121,
      "learning_rate": 0.00025078291495363316,
      "loss": 2.0087,
      "step": 127288
    },
    {
      "epoch": 1.66,
      "grad_norm": 0.9548730850219727,
      "learning_rate": 0.000250778879819599,
      "loss": 2.237,
      "step": 127289
    },
    {
      "epoch": 1.66,
      "grad_norm": 0.9666903614997864,
      "learning_rate": 0.000250774844694716,
      "loss": 2.2615,
      "step": 127290
    },
    {
      "epoch": 1.66,
      "grad_norm": 0.9319410920143127,
      "learning_rate": 0.00025077080957898495,
      "loss": 2.1752,
      "step": 127291
    },
    {
      "epoch": 1.66,
      "grad_norm": 0.9510554075241089,
      "learning_rate": 0.00025076677447240643,
      "loss": 2.1973,
      "step": 127292
    },
    {
      "epoch": 1.66,
      "grad_norm": 0.9970998764038086,
      "learning_rate": 0.0002507627393749815,
      "loss": 2.0513,
      "step": 127293
    },
    {
      "epoch": 1.66,
      "grad_norm": 0.9657043218612671,
      "learning_rate": 0.00025075870428671056,
      "loss": 2.0605,
      "step": 127294
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.0550376176834106,
      "learning_rate": 0.0002507546692075945,
      "loss": 2.2117,
      "step": 127295
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.01546049118042,
      "learning_rate": 0.00025075063413763406,
      "loss": 2.2921,
      "step": 127296
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.0564823150634766,
      "learning_rate": 0.00025074659907683,
      "loss": 2.2367,
      "step": 127297
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.24857759475708,
      "learning_rate": 0.0002507425640251831,
      "loss": 1.9218,
      "step": 127298
    },
    {
      "epoch": 1.66,
      "grad_norm": 0.9540795087814331,
      "learning_rate": 0.00025073852898269426,
      "loss": 2.0343,
      "step": 127299
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.1561315059661865,
      "learning_rate": 0.00025073449394936385,
      "loss": 2.0375,
      "step": 127300
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.0033495426177979,
      "learning_rate": 0.00025073045892519283,
      "loss": 1.928,
      "step": 127301
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.0983036756515503,
      "learning_rate": 0.000250726423910182,
      "loss": 1.9703,
      "step": 127302
    },
    {
      "epoch": 1.66,
      "grad_norm": 0.9745165109634399,
      "learning_rate": 0.00025072238890433207,
      "loss": 2.1802,
      "step": 127303
    },
    {
      "epoch": 1.66,
      "grad_norm": 0.9715012907981873,
      "learning_rate": 0.00025071835390764373,
      "loss": 2.1525,
      "step": 127304
    },
    {
      "epoch": 1.66,
      "grad_norm": 0.8943135142326355,
      "learning_rate": 0.00025071431892011793,
      "loss": 2.246,
      "step": 127305
    },
    {
      "epoch": 1.66,
      "grad_norm": 0.9138050675392151,
      "learning_rate": 0.0002507102839417551,
      "loss": 2.055,
      "step": 127306
    },
    {
      "epoch": 1.66,
      "grad_norm": 0.9166616797447205,
      "learning_rate": 0.00025070624897255623,
      "loss": 2.2649,
      "step": 127307
    },
    {
      "epoch": 1.66,
      "grad_norm": 0.948676586151123,
      "learning_rate": 0.00025070221401252195,
      "loss": 2.233,
      "step": 127308
    },
    {
      "epoch": 1.66,
      "grad_norm": 0.9740853309631348,
      "learning_rate": 0.00025069817906165303,
      "loss": 2.0169,
      "step": 127309
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.009009838104248,
      "learning_rate": 0.00025069414411995033,
      "loss": 2.181,
      "step": 127310
    },
    {
      "epoch": 1.66,
      "grad_norm": 0.9441967606544495,
      "learning_rate": 0.00025069010918741456,
      "loss": 2.0551,
      "step": 127311
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.0296913385391235,
      "learning_rate": 0.0002506860742640464,
      "loss": 2.0821,
      "step": 127312
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.0524604320526123,
      "learning_rate": 0.0002506820393498465,
      "loss": 1.9617,
      "step": 127313
    },
    {
      "epoch": 1.66,
      "grad_norm": 0.9331548810005188,
      "learning_rate": 0.00025067800444481583,
      "loss": 2.3715,
      "step": 127314
    },
    {
      "epoch": 1.66,
      "grad_norm": 0.9544576406478882,
      "learning_rate": 0.00025067396954895504,
      "loss": 2.2806,
      "step": 127315
    },
    {
      "epoch": 1.66,
      "grad_norm": 0.9286079406738281,
      "learning_rate": 0.00025066993466226487,
      "loss": 2.0595,
      "step": 127316
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.0143802165985107,
      "learning_rate": 0.00025066589978474613,
      "loss": 2.1575,
      "step": 127317
    },
    {
      "epoch": 1.66,
      "grad_norm": 0.8841844797134399,
      "learning_rate": 0.0002506618649163997,
      "loss": 2.1427,
      "step": 127318
    },
    {
      "epoch": 1.66,
      "grad_norm": 0.9884911179542542,
      "learning_rate": 0.00025065783005722586,
      "loss": 1.9395,
      "step": 127319
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.0718663930892944,
      "learning_rate": 0.0002506537952072258,
      "loss": 2.0637,
      "step": 127320
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.0511723756790161,
      "learning_rate": 0.00025064976036640014,
      "loss": 2.1027,
      "step": 127321
    },
    {
      "epoch": 1.66,
      "grad_norm": 0.9905262589454651,
      "learning_rate": 0.00025064572553474954,
      "loss": 2.0514,
      "step": 127322
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.0162291526794434,
      "learning_rate": 0.0002506416907122749,
      "loss": 2.3401,
      "step": 127323
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.008258581161499,
      "learning_rate": 0.0002506376558989769,
      "loss": 1.8808,
      "step": 127324
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.0107935667037964,
      "learning_rate": 0.0002506336210948563,
      "loss": 2.1368,
      "step": 127325
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.0486451387405396,
      "learning_rate": 0.0002506295862999138,
      "loss": 1.9978,
      "step": 127326
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.0575393438339233,
      "learning_rate": 0.00025062555151415025,
      "loss": 2.0959,
      "step": 127327
    },
    {
      "epoch": 1.66,
      "grad_norm": 0.9451627731323242,
      "learning_rate": 0.0002506215167375663,
      "loss": 2.1892,
      "step": 127328
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.2837377786636353,
      "learning_rate": 0.0002506174819701627,
      "loss": 2.0558,
      "step": 127329
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.069328784942627,
      "learning_rate": 0.00025061344721194034,
      "loss": 2.3243,
      "step": 127330
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.0543173551559448,
      "learning_rate": 0.0002506094124628998,
      "loss": 2.5052,
      "step": 127331
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.1416056156158447,
      "learning_rate": 0.00025060537772304196,
      "loss": 2.1675,
      "step": 127332
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.107557773590088,
      "learning_rate": 0.00025060134299236745,
      "loss": 2.2786,
      "step": 127333
    },
    {
      "epoch": 1.66,
      "grad_norm": 0.9746624231338501,
      "learning_rate": 0.0002505973082708771,
      "loss": 2.1477,
      "step": 127334
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.1292647123336792,
      "learning_rate": 0.00025059327355857165,
      "loss": 2.195,
      "step": 127335
    },
    {
      "epoch": 1.66,
      "grad_norm": 0.9890112280845642,
      "learning_rate": 0.0002505892388554519,
      "loss": 2.1589,
      "step": 127336
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.0916653871536255,
      "learning_rate": 0.0002505852041615185,
      "loss": 2.1019,
      "step": 127337
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.0331536531448364,
      "learning_rate": 0.0002505811694767722,
      "loss": 1.9544,
      "step": 127338
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.018781304359436,
      "learning_rate": 0.00025057713480121385,
      "loss": 2.0782,
      "step": 127339
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.0010172128677368,
      "learning_rate": 0.00025057310013484413,
      "loss": 2.1772,
      "step": 127340
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.1766400337219238,
      "learning_rate": 0.0002505690654776638,
      "loss": 2.2207,
      "step": 127341
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.066287875175476,
      "learning_rate": 0.0002505650308296736,
      "loss": 2.0999,
      "step": 127342
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.0620242357254028,
      "learning_rate": 0.0002505609961908743,
      "loss": 2.2727,
      "step": 127343
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.078112006187439,
      "learning_rate": 0.0002505569615612666,
      "loss": 1.995,
      "step": 127344
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.086057186126709,
      "learning_rate": 0.00025055292694085136,
      "loss": 2.2708,
      "step": 127345
    },
    {
      "epoch": 1.66,
      "grad_norm": 0.9913944602012634,
      "learning_rate": 0.00025054889232962926,
      "loss": 2.22,
      "step": 127346
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.0721019506454468,
      "learning_rate": 0.000250544857727601,
      "loss": 2.0239,
      "step": 127347
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.0060287714004517,
      "learning_rate": 0.0002505408231347675,
      "loss": 2.2466,
      "step": 127348
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.1965951919555664,
      "learning_rate": 0.0002505367885511293,
      "loss": 2.0823,
      "step": 127349
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.0329128503799438,
      "learning_rate": 0.00025053275397668727,
      "loss": 1.9615,
      "step": 127350
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.076141357421875,
      "learning_rate": 0.0002505287194114421,
      "loss": 2.3021,
      "step": 127351
    },
    {
      "epoch": 1.66,
      "grad_norm": 0.9296878576278687,
      "learning_rate": 0.0002505246848553946,
      "loss": 1.9399,
      "step": 127352
    },
    {
      "epoch": 1.66,
      "grad_norm": 0.9048646092414856,
      "learning_rate": 0.00025052065030854546,
      "loss": 2.0262,
      "step": 127353
    },
    {
      "epoch": 1.66,
      "grad_norm": 0.9544932842254639,
      "learning_rate": 0.0002505166157708956,
      "loss": 1.868,
      "step": 127354
    },
    {
      "epoch": 1.66,
      "grad_norm": 0.9825001955032349,
      "learning_rate": 0.0002505125812424455,
      "loss": 1.9592,
      "step": 127355
    },
    {
      "epoch": 1.66,
      "grad_norm": 0.9847202897071838,
      "learning_rate": 0.0002505085467231961,
      "loss": 2.1565,
      "step": 127356
    },
    {
      "epoch": 1.66,
      "grad_norm": 0.9597064852714539,
      "learning_rate": 0.00025050451221314806,
      "loss": 2.1668,
      "step": 127357
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.0551010370254517,
      "learning_rate": 0.00025050047771230213,
      "loss": 2.1038,
      "step": 127358
    },
    {
      "epoch": 1.66,
      "grad_norm": 0.9925917983055115,
      "learning_rate": 0.00025049644322065913,
      "loss": 2.159,
      "step": 127359
    },
    {
      "epoch": 1.66,
      "grad_norm": 0.9421687126159668,
      "learning_rate": 0.0002504924087382199,
      "loss": 2.1241,
      "step": 127360
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.0769115686416626,
      "learning_rate": 0.00025048837426498493,
      "loss": 2.2344,
      "step": 127361
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.0478382110595703,
      "learning_rate": 0.00025048433980095515,
      "loss": 2.1129,
      "step": 127362
    },
    {
      "epoch": 1.66,
      "grad_norm": 0.9105990529060364,
      "learning_rate": 0.0002504803053461313,
      "loss": 2.0609,
      "step": 127363
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.004167914390564,
      "learning_rate": 0.00025047627090051395,
      "loss": 2.0942,
      "step": 127364
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.170851707458496,
      "learning_rate": 0.0002504722364641041,
      "loss": 2.1378,
      "step": 127365
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.0712504386901855,
      "learning_rate": 0.0002504682020369025,
      "loss": 2.0683,
      "step": 127366
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.1201717853546143,
      "learning_rate": 0.0002504641676189097,
      "loss": 2.1588,
      "step": 127367
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.0508637428283691,
      "learning_rate": 0.0002504601332101265,
      "loss": 2.1624,
      "step": 127368
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.1450841426849365,
      "learning_rate": 0.0002504560988105537,
      "loss": 1.9643,
      "step": 127369
    },
    {
      "epoch": 1.66,
      "grad_norm": 0.989798367023468,
      "learning_rate": 0.0002504520644201921,
      "loss": 2.0343,
      "step": 127370
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.0833653211593628,
      "learning_rate": 0.00025044803003904234,
      "loss": 2.0595,
      "step": 127371
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.1206649541854858,
      "learning_rate": 0.00025044399566710534,
      "loss": 2.0684,
      "step": 127372
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.0678017139434814,
      "learning_rate": 0.00025043996130438167,
      "loss": 1.9941,
      "step": 127373
    },
    {
      "epoch": 1.66,
      "grad_norm": 0.9285469055175781,
      "learning_rate": 0.00025043592695087204,
      "loss": 1.8567,
      "step": 127374
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.0294609069824219,
      "learning_rate": 0.0002504318926065774,
      "loss": 1.9737,
      "step": 127375
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.0742756128311157,
      "learning_rate": 0.0002504278582714984,
      "loss": 2.2915,
      "step": 127376
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.0030783414840698,
      "learning_rate": 0.0002504238239456358,
      "loss": 2.0841,
      "step": 127377
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.0178414583206177,
      "learning_rate": 0.00025041978962899044,
      "loss": 1.9759,
      "step": 127378
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.0915484428405762,
      "learning_rate": 0.0002504157553215629,
      "loss": 2.2449,
      "step": 127379
    },
    {
      "epoch": 1.66,
      "grad_norm": 0.9510973691940308,
      "learning_rate": 0.00025041172102335395,
      "loss": 2.1345,
      "step": 127380
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.0186086893081665,
      "learning_rate": 0.00025040768673436443,
      "loss": 2.0154,
      "step": 127381
    },
    {
      "epoch": 1.66,
      "grad_norm": 0.997707724571228,
      "learning_rate": 0.000250403652454595,
      "loss": 2.0503,
      "step": 127382
    },
    {
      "epoch": 1.66,
      "grad_norm": 0.9658219814300537,
      "learning_rate": 0.0002503996181840466,
      "loss": 2.0156,
      "step": 127383
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.0004401206970215,
      "learning_rate": 0.00025039558392271974,
      "loss": 2.0818,
      "step": 127384
    },
    {
      "epoch": 1.66,
      "grad_norm": 0.9511722326278687,
      "learning_rate": 0.00025039154967061543,
      "loss": 1.9345,
      "step": 127385
    },
    {
      "epoch": 1.66,
      "grad_norm": 0.9537539482116699,
      "learning_rate": 0.00025038751542773416,
      "loss": 2.1533,
      "step": 127386
    },
    {
      "epoch": 1.66,
      "grad_norm": 0.9458909630775452,
      "learning_rate": 0.00025038348119407674,
      "loss": 2.0829,
      "step": 127387
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.0207242965698242,
      "learning_rate": 0.000250379446969644,
      "loss": 2.2081,
      "step": 127388
    },
    {
      "epoch": 1.66,
      "grad_norm": 3.9230120182037354,
      "learning_rate": 0.00025037541275443664,
      "loss": 2.2756,
      "step": 127389
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.0083558559417725,
      "learning_rate": 0.00025037137854845543,
      "loss": 1.8953,
      "step": 127390
    },
    {
      "epoch": 1.66,
      "grad_norm": 0.9799038171768188,
      "learning_rate": 0.00025036734435170127,
      "loss": 2.1624,
      "step": 127391
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.0052027702331543,
      "learning_rate": 0.0002503633101641746,
      "loss": 2.1616,
      "step": 127392
    },
    {
      "epoch": 1.66,
      "grad_norm": 0.9584972858428955,
      "learning_rate": 0.00025035927598587637,
      "loss": 1.8764,
      "step": 127393
    },
    {
      "epoch": 1.66,
      "grad_norm": 0.9945363402366638,
      "learning_rate": 0.00025035524181680726,
      "loss": 2.1634,
      "step": 127394
    },
    {
      "epoch": 1.66,
      "grad_norm": 0.9844695925712585,
      "learning_rate": 0.00025035120765696807,
      "loss": 1.9186,
      "step": 127395
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.0522408485412598,
      "learning_rate": 0.0002503471735063595,
      "loss": 2.1004,
      "step": 127396
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.0048490762710571,
      "learning_rate": 0.00025034313936498243,
      "loss": 2.0068,
      "step": 127397
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.0829113721847534,
      "learning_rate": 0.00025033910523283745,
      "loss": 2.1403,
      "step": 127398
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.0337899923324585,
      "learning_rate": 0.0002503350711099253,
      "loss": 2.3498,
      "step": 127399
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.071590542793274,
      "learning_rate": 0.00025033103699624684,
      "loss": 2.1833,
      "step": 127400
    },
    {
      "epoch": 1.66,
      "grad_norm": 0.9868535399436951,
      "learning_rate": 0.00025032700289180274,
      "loss": 2.2477,
      "step": 127401
    },
    {
      "epoch": 1.66,
      "grad_norm": 0.974644124507904,
      "learning_rate": 0.0002503229687965938,
      "loss": 2.1429,
      "step": 127402
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.088665246963501,
      "learning_rate": 0.0002503189347106209,
      "loss": 2.1206,
      "step": 127403
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.1468077898025513,
      "learning_rate": 0.00025031490063388454,
      "loss": 1.9866,
      "step": 127404
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.0545408725738525,
      "learning_rate": 0.00025031086656638555,
      "loss": 2.1205,
      "step": 127405
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.0236661434173584,
      "learning_rate": 0.0002503068325081247,
      "loss": 2.1779,
      "step": 127406
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.281167984008789,
      "learning_rate": 0.00025030279845910275,
      "loss": 2.0385,
      "step": 127407
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.1332117319107056,
      "learning_rate": 0.00025029876441932047,
      "loss": 2.1712,
      "step": 127408
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.1374008655548096,
      "learning_rate": 0.0002502947303887787,
      "loss": 2.2066,
      "step": 127409
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.031625509262085,
      "learning_rate": 0.00025029069636747797,
      "loss": 2.2958,
      "step": 127410
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.1139448881149292,
      "learning_rate": 0.0002502866623554191,
      "loss": 2.1266,
      "step": 127411
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.1130849123001099,
      "learning_rate": 0.0002502826283526029,
      "loss": 2.1999,
      "step": 127412
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.0522867441177368,
      "learning_rate": 0.0002502785943590301,
      "loss": 2.0201,
      "step": 127413
    },
    {
      "epoch": 1.66,
      "grad_norm": 0.9404284954071045,
      "learning_rate": 0.0002502745603747015,
      "loss": 1.9359,
      "step": 127414
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.0504151582717896,
      "learning_rate": 0.00025027052639961777,
      "loss": 2.1251,
      "step": 127415
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.0370006561279297,
      "learning_rate": 0.0002502664924337797,
      "loss": 2.137,
      "step": 127416
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.0118472576141357,
      "learning_rate": 0.00025026245847718803,
      "loss": 1.998,
      "step": 127417
    },
    {
      "epoch": 1.66,
      "grad_norm": 0.968870222568512,
      "learning_rate": 0.00025025842452984346,
      "loss": 2.0431,
      "step": 127418
    },
    {
      "epoch": 1.66,
      "grad_norm": 0.995337963104248,
      "learning_rate": 0.0002502543905917468,
      "loss": 2.1097,
      "step": 127419
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.0083695650100708,
      "learning_rate": 0.0002502503566628988,
      "loss": 2.2084,
      "step": 127420
    },
    {
      "epoch": 1.66,
      "grad_norm": 0.9254308342933655,
      "learning_rate": 0.0002502463227433002,
      "loss": 2.202,
      "step": 127421
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.2081912755966187,
      "learning_rate": 0.00025024228883295177,
      "loss": 2.0395,
      "step": 127422
    },
    {
      "epoch": 1.66,
      "grad_norm": 0.9870723485946655,
      "learning_rate": 0.0002502382549318542,
      "loss": 1.9776,
      "step": 127423
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.0287013053894043,
      "learning_rate": 0.0002502342210400083,
      "loss": 2.1686,
      "step": 127424
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.158998966217041,
      "learning_rate": 0.0002502301871574148,
      "loss": 2.0457,
      "step": 127425
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.247431993484497,
      "learning_rate": 0.0002502261532840744,
      "loss": 2.5061,
      "step": 127426
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.0482571125030518,
      "learning_rate": 0.000250222119419988,
      "loss": 2.1791,
      "step": 127427
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.1074941158294678,
      "learning_rate": 0.00025021808556515617,
      "loss": 2.2246,
      "step": 127428
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.0406469106674194,
      "learning_rate": 0.0002502140517195797,
      "loss": 2.2619,
      "step": 127429
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.0186511278152466,
      "learning_rate": 0.0002502100178832594,
      "loss": 2.1474,
      "step": 127430
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.0288364887237549,
      "learning_rate": 0.0002502059840561961,
      "loss": 2.0922,
      "step": 127431
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.069581151008606,
      "learning_rate": 0.00025020195023839033,
      "loss": 2.1563,
      "step": 127432
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.0570759773254395,
      "learning_rate": 0.00025019791642984305,
      "loss": 2.0436,
      "step": 127433
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.3395309448242188,
      "learning_rate": 0.00025019388263055486,
      "loss": 2.2359,
      "step": 127434
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.256289005279541,
      "learning_rate": 0.00025018984884052657,
      "loss": 2.2581,
      "step": 127435
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.0272663831710815,
      "learning_rate": 0.00025018581505975893,
      "loss": 2.0755,
      "step": 127436
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.0523098707199097,
      "learning_rate": 0.00025018178128825266,
      "loss": 2.104,
      "step": 127437
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.0393152236938477,
      "learning_rate": 0.0002501777475260086,
      "loss": 2.2603,
      "step": 127438
    },
    {
      "epoch": 1.66,
      "grad_norm": 0.953541100025177,
      "learning_rate": 0.00025017371377302746,
      "loss": 2.2402,
      "step": 127439
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.0078171491622925,
      "learning_rate": 0.0002501696800293099,
      "loss": 2.0186,
      "step": 127440
    },
    {
      "epoch": 1.66,
      "grad_norm": 0.9582980275154114,
      "learning_rate": 0.0002501656462948567,
      "loss": 2.154,
      "step": 127441
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.031479835510254,
      "learning_rate": 0.0002501616125696687,
      "loss": 2.1805,
      "step": 127442
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.0464928150177002,
      "learning_rate": 0.00025015757885374655,
      "loss": 1.9966,
      "step": 127443
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.0837894678115845,
      "learning_rate": 0.00025015354514709107,
      "loss": 2.1271,
      "step": 127444
    },
    {
      "epoch": 1.66,
      "grad_norm": 0.9810062050819397,
      "learning_rate": 0.00025014951144970314,
      "loss": 1.9863,
      "step": 127445
    },
    {
      "epoch": 1.66,
      "grad_norm": 0.9645041823387146,
      "learning_rate": 0.0002501454777615832,
      "loss": 2.0155,
      "step": 127446
    },
    {
      "epoch": 1.66,
      "grad_norm": 0.9762665629386902,
      "learning_rate": 0.0002501414440827322,
      "loss": 2.1424,
      "step": 127447
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.0039327144622803,
      "learning_rate": 0.0002501374104131508,
      "loss": 2.2057,
      "step": 127448
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.0188461542129517,
      "learning_rate": 0.00025013337675283974,
      "loss": 2.0907,
      "step": 127449
    },
    {
      "epoch": 1.66,
      "grad_norm": 0.9933527112007141,
      "learning_rate": 0.00025012934310179993,
      "loss": 1.9493,
      "step": 127450
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.0934449434280396,
      "learning_rate": 0.00025012530946003195,
      "loss": 2.4087,
      "step": 127451
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.061126947402954,
      "learning_rate": 0.00025012127582753677,
      "loss": 2.0021,
      "step": 127452
    },
    {
      "epoch": 1.66,
      "grad_norm": 0.9819427132606506,
      "learning_rate": 0.0002501172422043149,
      "loss": 2.1113,
      "step": 127453
    },
    {
      "epoch": 1.66,
      "grad_norm": 0.996849536895752,
      "learning_rate": 0.00025011320859036714,
      "loss": 2.1045,
      "step": 127454
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.2735785245895386,
      "learning_rate": 0.00025010917498569427,
      "loss": 2.3458,
      "step": 127455
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.0489381551742554,
      "learning_rate": 0.00025010514139029707,
      "loss": 2.0997,
      "step": 127456
    },
    {
      "epoch": 1.66,
      "grad_norm": 0.9730176329612732,
      "learning_rate": 0.0002501011078041762,
      "loss": 2.1368,
      "step": 127457
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.1767587661743164,
      "learning_rate": 0.0002500970742273327,
      "loss": 2.0505,
      "step": 127458
    },
    {
      "epoch": 1.66,
      "grad_norm": 2.520209550857544,
      "learning_rate": 0.0002500930406597669,
      "loss": 2.3401,
      "step": 127459
    },
    {
      "epoch": 1.66,
      "grad_norm": 0.9806927442550659,
      "learning_rate": 0.0002500890071014797,
      "loss": 2.0854,
      "step": 127460
    },
    {
      "epoch": 1.66,
      "grad_norm": 0.9934827089309692,
      "learning_rate": 0.000250084973552472,
      "loss": 2.1068,
      "step": 127461
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.0449771881103516,
      "learning_rate": 0.00025008094001274437,
      "loss": 2.2566,
      "step": 127462
    },
    {
      "epoch": 1.66,
      "grad_norm": 0.9673107862472534,
      "learning_rate": 0.0002500769064822977,
      "loss": 2.3206,
      "step": 127463
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.204305648803711,
      "learning_rate": 0.0002500728729611328,
      "loss": 2.0368,
      "step": 127464
    },
    {
      "epoch": 1.66,
      "grad_norm": 0.9664444923400879,
      "learning_rate": 0.0002500688394492501,
      "loss": 2.0913,
      "step": 127465
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.051809310913086,
      "learning_rate": 0.0002500648059466506,
      "loss": 2.0974,
      "step": 127466
    },
    {
      "epoch": 1.66,
      "grad_norm": 0.9593704342842102,
      "learning_rate": 0.000250060772453335,
      "loss": 2.0549,
      "step": 127467
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.141425609588623,
      "learning_rate": 0.000250056738969304,
      "loss": 2.2191,
      "step": 127468
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.0429697036743164,
      "learning_rate": 0.00025005270549455845,
      "loss": 2.1529,
      "step": 127469
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.1197714805603027,
      "learning_rate": 0.00025004867202909915,
      "loss": 2.0583,
      "step": 127470
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.1099836826324463,
      "learning_rate": 0.00025004463857292657,
      "loss": 2.2829,
      "step": 127471
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.0250972509384155,
      "learning_rate": 0.0002500406051260417,
      "loss": 2.0063,
      "step": 127472
    },
    {
      "epoch": 1.66,
      "grad_norm": 0.9744088053703308,
      "learning_rate": 0.0002500365716884452,
      "loss": 2.3039,
      "step": 127473
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.274261474609375,
      "learning_rate": 0.00025003253826013787,
      "loss": 2.191,
      "step": 127474
    },
    {
      "epoch": 1.66,
      "grad_norm": 0.9332820773124695,
      "learning_rate": 0.00025002850484112036,
      "loss": 2.252,
      "step": 127475
    },
    {
      "epoch": 1.66,
      "grad_norm": 0.932021975517273,
      "learning_rate": 0.0002500244714313937,
      "loss": 1.9026,
      "step": 127476
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.1315374374389648,
      "learning_rate": 0.0002500204380309583,
      "loss": 2.0682,
      "step": 127477
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.1550898551940918,
      "learning_rate": 0.00025001640463981497,
      "loss": 2.0136,
      "step": 127478
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.1269264221191406,
      "learning_rate": 0.0002500123712579646,
      "loss": 2.1156,
      "step": 127479
    },
    {
      "epoch": 1.66,
      "grad_norm": 0.9746201038360596,
      "learning_rate": 0.0002500083378854079,
      "loss": 1.994,
      "step": 127480
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.0909669399261475,
      "learning_rate": 0.0002500043045221455,
      "loss": 2.133,
      "step": 127481
    },
    {
      "epoch": 1.66,
      "grad_norm": 0.9846354126930237,
      "learning_rate": 0.00025000027116817847,
      "loss": 2.0939,
      "step": 127482
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.0403512716293335,
      "learning_rate": 0.00024999623782350714,
      "loss": 2.1983,
      "step": 127483
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.0633550882339478,
      "learning_rate": 0.00024999220448813247,
      "loss": 2.0564,
      "step": 127484
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.094201922416687,
      "learning_rate": 0.0002499881711620552,
      "loss": 2.1734,
      "step": 127485
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.000657320022583,
      "learning_rate": 0.00024998413784527607,
      "loss": 2.1735,
      "step": 127486
    },
    {
      "epoch": 1.66,
      "grad_norm": 0.9644940495491028,
      "learning_rate": 0.00024998010453779586,
      "loss": 2.1838,
      "step": 127487
    },
    {
      "epoch": 1.66,
      "grad_norm": 0.9684514999389648,
      "learning_rate": 0.0002499760712396154,
      "loss": 2.4405,
      "step": 127488
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.052237868309021,
      "learning_rate": 0.00024997203795073516,
      "loss": 2.2228,
      "step": 127489
    },
    {
      "epoch": 1.66,
      "grad_norm": 0.9422466158866882,
      "learning_rate": 0.0002499680046711561,
      "loss": 2.294,
      "step": 127490
    },
    {
      "epoch": 1.66,
      "grad_norm": 0.9019993543624878,
      "learning_rate": 0.00024996397140087897,
      "loss": 2.2275,
      "step": 127491
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.0359269380569458,
      "learning_rate": 0.0002499599381399044,
      "loss": 2.2971,
      "step": 127492
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.0191808938980103,
      "learning_rate": 0.00024995590488823325,
      "loss": 1.947,
      "step": 127493
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.0326398611068726,
      "learning_rate": 0.0002499518716458664,
      "loss": 2.0993,
      "step": 127494
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.0504366159439087,
      "learning_rate": 0.00024994783841280434,
      "loss": 1.9875,
      "step": 127495
    },
    {
      "epoch": 1.66,
      "grad_norm": 0.9352647066116333,
      "learning_rate": 0.00024994380518904784,
      "loss": 2.0368,
      "step": 127496
    },
    {
      "epoch": 1.66,
      "grad_norm": 0.9711735844612122,
      "learning_rate": 0.0002499397719745978,
      "loss": 2.0837,
      "step": 127497
    },
    {
      "epoch": 1.66,
      "grad_norm": 0.9429259300231934,
      "learning_rate": 0.00024993573876945486,
      "loss": 2.1938,
      "step": 127498
    },
    {
      "epoch": 1.66,
      "grad_norm": 0.8920586705207825,
      "learning_rate": 0.0002499317055736198,
      "loss": 1.9121,
      "step": 127499
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.0301440954208374,
      "learning_rate": 0.00024992767238709353,
      "loss": 1.9992,
      "step": 127500
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.0194697380065918,
      "learning_rate": 0.0002499236392098766,
      "loss": 2.0146,
      "step": 127501
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.1112264394760132,
      "learning_rate": 0.0002499196060419697,
      "loss": 2.3012,
      "step": 127502
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.0006943941116333,
      "learning_rate": 0.00024991557288337374,
      "loss": 2.1404,
      "step": 127503
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.0148533582687378,
      "learning_rate": 0.0002499115397340894,
      "loss": 1.919,
      "step": 127504
    },
    {
      "epoch": 1.66,
      "grad_norm": 0.9542745351791382,
      "learning_rate": 0.00024990750659411746,
      "loss": 2.1165,
      "step": 127505
    },
    {
      "epoch": 1.66,
      "grad_norm": 0.9785341620445251,
      "learning_rate": 0.0002499034734634587,
      "loss": 2.0682,
      "step": 127506
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.1021093130111694,
      "learning_rate": 0.00024989944034211383,
      "loss": 1.9454,
      "step": 127507
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.0398350954055786,
      "learning_rate": 0.0002498954072300836,
      "loss": 2.1428,
      "step": 127508
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.0241385698318481,
      "learning_rate": 0.0002498913741273687,
      "loss": 2.201,
      "step": 127509
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.079343318939209,
      "learning_rate": 0.00024988734103397,
      "loss": 2.0682,
      "step": 127510
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.043861746788025,
      "learning_rate": 0.00024988330794988814,
      "loss": 2.2161,
      "step": 127511
    },
    {
      "epoch": 1.66,
      "grad_norm": 0.9881994724273682,
      "learning_rate": 0.000249879274875124,
      "loss": 2.0028,
      "step": 127512
    },
    {
      "epoch": 1.66,
      "grad_norm": 0.9684028029441833,
      "learning_rate": 0.00024987524180967817,
      "loss": 1.9298,
      "step": 127513
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.0216668844223022,
      "learning_rate": 0.0002498712087535515,
      "loss": 2.1336,
      "step": 127514
    },
    {
      "epoch": 1.66,
      "grad_norm": 0.9738934636116028,
      "learning_rate": 0.0002498671757067447,
      "loss": 2.044,
      "step": 127515
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.050167202949524,
      "learning_rate": 0.0002498631426692586,
      "loss": 2.3722,
      "step": 127516
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.1187684535980225,
      "learning_rate": 0.00024985910964109385,
      "loss": 2.1148,
      "step": 127517
    },
    {
      "epoch": 1.66,
      "grad_norm": 0.9693753719329834,
      "learning_rate": 0.0002498550766222512,
      "loss": 2.3563,
      "step": 127518
    },
    {
      "epoch": 1.66,
      "grad_norm": 0.9379351735115051,
      "learning_rate": 0.0002498510436127315,
      "loss": 1.9462,
      "step": 127519
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.2242511510849,
      "learning_rate": 0.0002498470106125354,
      "loss": 2.1047,
      "step": 127520
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.0210524797439575,
      "learning_rate": 0.0002498429776216637,
      "loss": 1.9663,
      "step": 127521
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.0285160541534424,
      "learning_rate": 0.0002498389446401171,
      "loss": 2.3027,
      "step": 127522
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.231387972831726,
      "learning_rate": 0.0002498349116678965,
      "loss": 1.9298,
      "step": 127523
    },
    {
      "epoch": 1.66,
      "grad_norm": 0.9476797580718994,
      "learning_rate": 0.0002498308787050024,
      "loss": 2.0846,
      "step": 127524
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.0163499116897583,
      "learning_rate": 0.0002498268457514358,
      "loss": 2.1141,
      "step": 127525
    },
    {
      "epoch": 1.66,
      "grad_norm": 0.9071759581565857,
      "learning_rate": 0.0002498228128071973,
      "loss": 1.9577,
      "step": 127526
    },
    {
      "epoch": 1.66,
      "grad_norm": 0.9807984828948975,
      "learning_rate": 0.00024981877987228764,
      "loss": 2.1938,
      "step": 127527
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.1215511560440063,
      "learning_rate": 0.0002498147469467076,
      "loss": 2.1505,
      "step": 127528
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.140302300453186,
      "learning_rate": 0.000249810714030458,
      "loss": 1.968,
      "step": 127529
    },
    {
      "epoch": 1.66,
      "grad_norm": 0.9049529433250427,
      "learning_rate": 0.00024980668112353955,
      "loss": 2.2521,
      "step": 127530
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.0110089778900146,
      "learning_rate": 0.00024980264822595304,
      "loss": 1.8328,
      "step": 127531
    },
    {
      "epoch": 1.66,
      "grad_norm": 0.964968204498291,
      "learning_rate": 0.00024979861533769906,
      "loss": 2.2337,
      "step": 127532
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.0041261911392212,
      "learning_rate": 0.0002497945824587785,
      "loss": 2.1233,
      "step": 127533
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.0279481410980225,
      "learning_rate": 0.000249790549589192,
      "loss": 2.1086,
      "step": 127534
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.010473370552063,
      "learning_rate": 0.00024978651672894045,
      "loss": 1.9472,
      "step": 127535
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.010237693786621,
      "learning_rate": 0.0002497824838780245,
      "loss": 2.1309,
      "step": 127536
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.0202997922897339,
      "learning_rate": 0.0002497784510364451,
      "loss": 2.1043,
      "step": 127537
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.0166423320770264,
      "learning_rate": 0.0002497744182042027,
      "loss": 2.0562,
      "step": 127538
    },
    {
      "epoch": 1.66,
      "grad_norm": 0.9366927742958069,
      "learning_rate": 0.0002497703853812982,
      "loss": 2.0453,
      "step": 127539
    },
    {
      "epoch": 1.66,
      "grad_norm": 0.919684112071991,
      "learning_rate": 0.0002497663525677323,
      "loss": 1.9287,
      "step": 127540
    },
    {
      "epoch": 1.66,
      "grad_norm": 0.9880059361457825,
      "learning_rate": 0.0002497623197635057,
      "loss": 2.2813,
      "step": 127541
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.0378522872924805,
      "learning_rate": 0.0002497582869686194,
      "loss": 2.1297,
      "step": 127542
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.0773606300354004,
      "learning_rate": 0.000249754254183074,
      "loss": 2.26,
      "step": 127543
    },
    {
      "epoch": 1.66,
      "grad_norm": 0.9502119421958923,
      "learning_rate": 0.00024975022140687013,
      "loss": 1.919,
      "step": 127544
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.0491244792938232,
      "learning_rate": 0.00024974618864000863,
      "loss": 2.1874,
      "step": 127545
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.3476675748825073,
      "learning_rate": 0.0002497421558824903,
      "loss": 2.2098,
      "step": 127546
    },
    {
      "epoch": 1.66,
      "grad_norm": 0.9572923183441162,
      "learning_rate": 0.00024973812313431584,
      "loss": 1.9813,
      "step": 127547
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.1267598867416382,
      "learning_rate": 0.000249734090395486,
      "loss": 2.069,
      "step": 127548
    },
    {
      "epoch": 1.66,
      "grad_norm": 0.9747011661529541,
      "learning_rate": 0.0002497300576660017,
      "loss": 2.1205,
      "step": 127549
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.0233232975006104,
      "learning_rate": 0.00024972602494586333,
      "loss": 2.0007,
      "step": 127550
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.0683441162109375,
      "learning_rate": 0.0002497219922350719,
      "loss": 2.2043,
      "step": 127551
    },
    {
      "epoch": 1.66,
      "grad_norm": 0.9533705115318298,
      "learning_rate": 0.0002497179595336281,
      "loss": 1.9788,
      "step": 127552
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.02689790725708,
      "learning_rate": 0.00024971392684153265,
      "loss": 2.2755,
      "step": 127553
    },
    {
      "epoch": 1.66,
      "grad_norm": 0.9957531094551086,
      "learning_rate": 0.0002497098941587864,
      "loss": 2.1009,
      "step": 127554
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.0136778354644775,
      "learning_rate": 0.0002497058614853901,
      "loss": 2.1961,
      "step": 127555
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.080452561378479,
      "learning_rate": 0.0002497018288213443,
      "loss": 2.118,
      "step": 127556
    },
    {
      "epoch": 1.66,
      "grad_norm": 0.8888301849365234,
      "learning_rate": 0.00024969779616664987,
      "loss": 2.1186,
      "step": 127557
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.0291998386383057,
      "learning_rate": 0.0002496937635213076,
      "loss": 2.0562,
      "step": 127558
    },
    {
      "epoch": 1.66,
      "grad_norm": 0.984740674495697,
      "learning_rate": 0.0002496897308853182,
      "loss": 2.2608,
      "step": 127559
    },
    {
      "epoch": 1.66,
      "grad_norm": 0.995785117149353,
      "learning_rate": 0.0002496856982586824,
      "loss": 2.277,
      "step": 127560
    },
    {
      "epoch": 1.66,
      "grad_norm": 0.9661215543746948,
      "learning_rate": 0.0002496816656414011,
      "loss": 2.1712,
      "step": 127561
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.0492113828659058,
      "learning_rate": 0.0002496776330334749,
      "loss": 1.9824,
      "step": 127562
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.0426673889160156,
      "learning_rate": 0.0002496736004349045,
      "loss": 2.0781,
      "step": 127563
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.0146280527114868,
      "learning_rate": 0.0002496695678456907,
      "loss": 2.092,
      "step": 127564
    },
    {
      "epoch": 1.66,
      "grad_norm": 0.9190889596939087,
      "learning_rate": 0.00024966553526583436,
      "loss": 2.0351,
      "step": 127565
    },
    {
      "epoch": 1.66,
      "grad_norm": 0.955427348613739,
      "learning_rate": 0.0002496615026953361,
      "loss": 1.9487,
      "step": 127566
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.0199249982833862,
      "learning_rate": 0.0002496574701341968,
      "loss": 2.0262,
      "step": 127567
    },
    {
      "epoch": 1.66,
      "grad_norm": 0.963684618473053,
      "learning_rate": 0.00024965343758241704,
      "loss": 2.2697,
      "step": 127568
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.0231654644012451,
      "learning_rate": 0.0002496494050399977,
      "loss": 2.0179,
      "step": 127569
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.004451870918274,
      "learning_rate": 0.00024964537250693946,
      "loss": 1.8666,
      "step": 127570
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.2944146394729614,
      "learning_rate": 0.00024964133998324305,
      "loss": 2.0091,
      "step": 127571
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.237478256225586,
      "learning_rate": 0.0002496373074689093,
      "loss": 2.0773,
      "step": 127572
    },
    {
      "epoch": 1.66,
      "grad_norm": 0.9687013626098633,
      "learning_rate": 0.00024963327496393907,
      "loss": 2.1319,
      "step": 127573
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.0736027956008911,
      "learning_rate": 0.0002496292424683328,
      "loss": 2.1162,
      "step": 127574
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.1112271547317505,
      "learning_rate": 0.00024962520998209146,
      "loss": 2.3697,
      "step": 127575
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.056607723236084,
      "learning_rate": 0.00024962117750521566,
      "loss": 2.1893,
      "step": 127576
    },
    {
      "epoch": 1.66,
      "grad_norm": 0.9979932904243469,
      "learning_rate": 0.0002496171450377063,
      "loss": 2.0433,
      "step": 127577
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.0749149322509766,
      "learning_rate": 0.00024961311257956404,
      "loss": 2.0293,
      "step": 127578
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.1214582920074463,
      "learning_rate": 0.0002496090801307898,
      "loss": 2.1638,
      "step": 127579
    },
    {
      "epoch": 1.66,
      "grad_norm": 0.9933168292045593,
      "learning_rate": 0.00024960504769138407,
      "loss": 2.1035,
      "step": 127580
    },
    {
      "epoch": 1.66,
      "grad_norm": 0.9530292749404907,
      "learning_rate": 0.00024960101526134764,
      "loss": 2.2261,
      "step": 127581
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.2000190019607544,
      "learning_rate": 0.00024959698284068144,
      "loss": 2.1101,
      "step": 127582
    },
    {
      "epoch": 1.66,
      "grad_norm": 0.9630203247070312,
      "learning_rate": 0.000249592950429386,
      "loss": 2.1016,
      "step": 127583
    },
    {
      "epoch": 1.66,
      "grad_norm": 0.9553600549697876,
      "learning_rate": 0.00024958891802746223,
      "loss": 1.9997,
      "step": 127584
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.0233312845230103,
      "learning_rate": 0.00024958488563491085,
      "loss": 2.1508,
      "step": 127585
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.0198975801467896,
      "learning_rate": 0.00024958085325173273,
      "loss": 1.8605,
      "step": 127586
    },
    {
      "epoch": 1.66,
      "grad_norm": 0.9583185911178589,
      "learning_rate": 0.0002495768208779283,
      "loss": 1.9885,
      "step": 127587
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.0126135349273682,
      "learning_rate": 0.0002495727885134985,
      "loss": 2.1841,
      "step": 127588
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.2180085182189941,
      "learning_rate": 0.0002495687561584441,
      "loss": 2.1082,
      "step": 127589
    },
    {
      "epoch": 1.66,
      "grad_norm": 0.95814049243927,
      "learning_rate": 0.0002495647238127658,
      "loss": 2.0421,
      "step": 127590
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.0600756406784058,
      "learning_rate": 0.0002495606914764644,
      "loss": 2.0222,
      "step": 127591
    },
    {
      "epoch": 1.66,
      "grad_norm": 0.9928584694862366,
      "learning_rate": 0.0002495566591495407,
      "loss": 2.1078,
      "step": 127592
    },
    {
      "epoch": 1.66,
      "grad_norm": 0.9799762964248657,
      "learning_rate": 0.00024955262683199527,
      "loss": 2.1935,
      "step": 127593
    },
    {
      "epoch": 1.66,
      "grad_norm": 0.9958468675613403,
      "learning_rate": 0.000249548594523829,
      "loss": 2.0214,
      "step": 127594
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.0643351078033447,
      "learning_rate": 0.00024954456222504254,
      "loss": 2.2596,
      "step": 127595
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.004866123199463,
      "learning_rate": 0.0002495405299356367,
      "loss": 2.0612,
      "step": 127596
    },
    {
      "epoch": 1.66,
      "grad_norm": 0.9767473340034485,
      "learning_rate": 0.0002495364976556122,
      "loss": 2.0879,
      "step": 127597
    },
    {
      "epoch": 1.66,
      "grad_norm": 0.9917497634887695,
      "learning_rate": 0.00024953246538496994,
      "loss": 1.9387,
      "step": 127598
    },
    {
      "epoch": 1.66,
      "grad_norm": 0.9142695665359497,
      "learning_rate": 0.0002495284331237105,
      "loss": 2.2694,
      "step": 127599
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.0438176393508911,
      "learning_rate": 0.0002495244008718347,
      "loss": 2.1106,
      "step": 127600
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.077576994895935,
      "learning_rate": 0.0002495203686293432,
      "loss": 2.1025,
      "step": 127601
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.0113029479980469,
      "learning_rate": 0.0002495163363962368,
      "loss": 2.2303,
      "step": 127602
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.084054708480835,
      "learning_rate": 0.00024951230417251635,
      "loss": 2.0288,
      "step": 127603
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.042902946472168,
      "learning_rate": 0.0002495082719581825,
      "loss": 2.1624,
      "step": 127604
    },
    {
      "epoch": 1.66,
      "grad_norm": 0.9300137162208557,
      "learning_rate": 0.000249504239753236,
      "loss": 2.2135,
      "step": 127605
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.0406757593154907,
      "learning_rate": 0.0002495002075576776,
      "loss": 2.2195,
      "step": 127606
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.0297993421554565,
      "learning_rate": 0.000249496175371508,
      "loss": 2.1963,
      "step": 127607
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.1747251749038696,
      "learning_rate": 0.00024949214319472813,
      "loss": 2.303,
      "step": 127608
    },
    {
      "epoch": 1.66,
      "grad_norm": 0.971343457698822,
      "learning_rate": 0.00024948811102733855,
      "loss": 2.2653,
      "step": 127609
    },
    {
      "epoch": 1.66,
      "grad_norm": 0.9389562606811523,
      "learning_rate": 0.00024948407886934016,
      "loss": 2.08,
      "step": 127610
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.0308613777160645,
      "learning_rate": 0.00024948004672073357,
      "loss": 2.102,
      "step": 127611
    },
    {
      "epoch": 1.66,
      "grad_norm": 0.9947636127471924,
      "learning_rate": 0.0002494760145815196,
      "loss": 2.3737,
      "step": 127612
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.048142671585083,
      "learning_rate": 0.00024947198245169896,
      "loss": 2.0415,
      "step": 127613
    },
    {
      "epoch": 1.66,
      "grad_norm": 0.9967978596687317,
      "learning_rate": 0.00024946795033127246,
      "loss": 1.8326,
      "step": 127614
    },
    {
      "epoch": 1.66,
      "grad_norm": 0.9578231573104858,
      "learning_rate": 0.0002494639182202409,
      "loss": 2.0802,
      "step": 127615
    },
    {
      "epoch": 1.66,
      "grad_norm": 0.9808515906333923,
      "learning_rate": 0.0002494598861186049,
      "loss": 2.1559,
      "step": 127616
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.298084020614624,
      "learning_rate": 0.0002494558540263653,
      "loss": 1.902,
      "step": 127617
    },
    {
      "epoch": 1.66,
      "grad_norm": 0.9983001947402954,
      "learning_rate": 0.0002494518219435227,
      "loss": 2.0283,
      "step": 127618
    },
    {
      "epoch": 1.66,
      "grad_norm": 0.9541378021240234,
      "learning_rate": 0.000249447789870078,
      "loss": 2.0832,
      "step": 127619
    },
    {
      "epoch": 1.66,
      "grad_norm": 0.929216742515564,
      "learning_rate": 0.0002494437578060319,
      "loss": 1.9398,
      "step": 127620
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.0270918607711792,
      "learning_rate": 0.0002494397257513852,
      "loss": 1.8732,
      "step": 127621
    },
    {
      "epoch": 1.66,
      "grad_norm": 0.9945829510688782,
      "learning_rate": 0.0002494356937061387,
      "loss": 2.2327,
      "step": 127622
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.149876594543457,
      "learning_rate": 0.00024943166167029293,
      "loss": 2.0687,
      "step": 127623
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.0334620475769043,
      "learning_rate": 0.0002494276296438488,
      "loss": 1.9984,
      "step": 127624
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.0109062194824219,
      "learning_rate": 0.00024942359762680696,
      "loss": 2.1511,
      "step": 127625
    },
    {
      "epoch": 1.66,
      "grad_norm": 0.9320139288902283,
      "learning_rate": 0.00024941956561916833,
      "loss": 2.0726,
      "step": 127626
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.2306934595108032,
      "learning_rate": 0.0002494155336209335,
      "loss": 2.2599,
      "step": 127627
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.0272389650344849,
      "learning_rate": 0.00024941150163210337,
      "loss": 2.1138,
      "step": 127628
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.026861548423767,
      "learning_rate": 0.0002494074696526785,
      "loss": 2.3799,
      "step": 127629
    },
    {
      "epoch": 1.66,
      "grad_norm": 0.9746366143226624,
      "learning_rate": 0.00024940343768265974,
      "loss": 2.1676,
      "step": 127630
    },
    {
      "epoch": 1.66,
      "grad_norm": 0.9382113814353943,
      "learning_rate": 0.0002493994057220479,
      "loss": 2.2947,
      "step": 127631
    },
    {
      "epoch": 1.66,
      "grad_norm": 0.9617322087287903,
      "learning_rate": 0.00024939537377084356,
      "loss": 2.2428,
      "step": 127632
    },
    {
      "epoch": 1.66,
      "grad_norm": 0.9844080209732056,
      "learning_rate": 0.0002493913418290476,
      "loss": 2.1652,
      "step": 127633
    },
    {
      "epoch": 1.66,
      "grad_norm": 0.995093047618866,
      "learning_rate": 0.00024938730989666096,
      "loss": 2.2536,
      "step": 127634
    },
    {
      "epoch": 1.66,
      "grad_norm": 0.9522949457168579,
      "learning_rate": 0.00024938327797368394,
      "loss": 1.9191,
      "step": 127635
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.097220778465271,
      "learning_rate": 0.0002493792460601176,
      "loss": 2.1138,
      "step": 127636
    },
    {
      "epoch": 1.66,
      "grad_norm": 0.9723727703094482,
      "learning_rate": 0.0002493752141559626,
      "loss": 2.049,
      "step": 127637
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.0811048746109009,
      "learning_rate": 0.0002493711822612196,
      "loss": 2.0069,
      "step": 127638
    },
    {
      "epoch": 1.66,
      "grad_norm": 0.9558041095733643,
      "learning_rate": 0.00024936715037588954,
      "loss": 2.0645,
      "step": 127639
    },
    {
      "epoch": 1.66,
      "grad_norm": 0.9300116300582886,
      "learning_rate": 0.00024936311849997324,
      "loss": 2.0593,
      "step": 127640
    },
    {
      "epoch": 1.66,
      "grad_norm": 0.9944270253181458,
      "learning_rate": 0.00024935908663347106,
      "loss": 2.089,
      "step": 127641
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.026397705078125,
      "learning_rate": 0.0002493550547763841,
      "loss": 1.8504,
      "step": 127642
    },
    {
      "epoch": 1.66,
      "grad_norm": 0.9897211790084839,
      "learning_rate": 0.00024935102292871294,
      "loss": 2.1489,
      "step": 127643
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.0546846389770508,
      "learning_rate": 0.00024934699109045835,
      "loss": 2.0336,
      "step": 127644
    },
    {
      "epoch": 1.66,
      "grad_norm": 0.915812075138092,
      "learning_rate": 0.0002493429592616212,
      "loss": 2.1042,
      "step": 127645
    },
    {
      "epoch": 1.66,
      "grad_norm": 0.9910659193992615,
      "learning_rate": 0.0002493389274422021,
      "loss": 2.045,
      "step": 127646
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.073322057723999,
      "learning_rate": 0.0002493348956322019,
      "loss": 2.113,
      "step": 127647
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.1904785633087158,
      "learning_rate": 0.0002493308638316213,
      "loss": 2.1017,
      "step": 127648
    },
    {
      "epoch": 1.66,
      "grad_norm": 0.9990969896316528,
      "learning_rate": 0.00024932683204046096,
      "loss": 2.2884,
      "step": 127649
    },
    {
      "epoch": 1.66,
      "grad_norm": 0.9400461912155151,
      "learning_rate": 0.0002493228002587217,
      "loss": 2.0166,
      "step": 127650
    },
    {
      "epoch": 1.66,
      "grad_norm": 0.9703434109687805,
      "learning_rate": 0.0002493187684864044,
      "loss": 2.155,
      "step": 127651
    },
    {
      "epoch": 1.66,
      "grad_norm": 0.9425574541091919,
      "learning_rate": 0.0002493147367235096,
      "loss": 2.1573,
      "step": 127652
    },
    {
      "epoch": 1.66,
      "grad_norm": 0.9281562566757202,
      "learning_rate": 0.00024931070497003833,
      "loss": 2.1291,
      "step": 127653
    },
    {
      "epoch": 1.66,
      "grad_norm": 0.8955892324447632,
      "learning_rate": 0.000249306673225991,
      "loss": 2.0271,
      "step": 127654
    },
    {
      "epoch": 1.66,
      "grad_norm": 0.9494752883911133,
      "learning_rate": 0.00024930264149136856,
      "loss": 2.2242,
      "step": 127655
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.0519791841506958,
      "learning_rate": 0.00024929860976617165,
      "loss": 2.0932,
      "step": 127656
    },
    {
      "epoch": 1.66,
      "grad_norm": 0.9101178646087646,
      "learning_rate": 0.0002492945780504011,
      "loss": 2.0757,
      "step": 127657
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.1378192901611328,
      "learning_rate": 0.00024929054634405766,
      "loss": 2.0853,
      "step": 127658
    },
    {
      "epoch": 1.66,
      "grad_norm": 0.953707218170166,
      "learning_rate": 0.0002492865146471422,
      "loss": 2.1719,
      "step": 127659
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.0090771913528442,
      "learning_rate": 0.0002492824829596552,
      "loss": 2.1516,
      "step": 127660
    },
    {
      "epoch": 1.66,
      "grad_norm": 0.9701117277145386,
      "learning_rate": 0.00024927845128159756,
      "loss": 2.1639,
      "step": 127661
    },
    {
      "epoch": 1.66,
      "grad_norm": 0.9398564696311951,
      "learning_rate": 0.00024927441961297,
      "loss": 1.9764,
      "step": 127662
    },
    {
      "epoch": 1.66,
      "grad_norm": 0.925872266292572,
      "learning_rate": 0.0002492703879537733,
      "loss": 2.0308,
      "step": 127663
    },
    {
      "epoch": 1.66,
      "grad_norm": 0.9548287987709045,
      "learning_rate": 0.00024926635630400816,
      "loss": 2.0767,
      "step": 127664
    },
    {
      "epoch": 1.66,
      "grad_norm": 0.9589795470237732,
      "learning_rate": 0.0002492623246636755,
      "loss": 2.1314,
      "step": 127665
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.0329813957214355,
      "learning_rate": 0.0002492582930327758,
      "loss": 2.1608,
      "step": 127666
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.0540499687194824,
      "learning_rate": 0.00024925426141130994,
      "loss": 2.1609,
      "step": 127667
    },
    {
      "epoch": 1.66,
      "grad_norm": 0.9800192713737488,
      "learning_rate": 0.0002492502297992787,
      "loss": 2.044,
      "step": 127668
    },
    {
      "epoch": 1.66,
      "grad_norm": 0.9380359649658203,
      "learning_rate": 0.00024924619819668276,
      "loss": 2.29,
      "step": 127669
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.0290253162384033,
      "learning_rate": 0.0002492421666035229,
      "loss": 2.1309,
      "step": 127670
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.1421529054641724,
      "learning_rate": 0.0002492381350198,
      "loss": 2.0752,
      "step": 127671
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.057612657546997,
      "learning_rate": 0.00024923410344551467,
      "loss": 2.3469,
      "step": 127672
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.0503294467926025,
      "learning_rate": 0.00024923007188066757,
      "loss": 2.2285,
      "step": 127673
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.0223989486694336,
      "learning_rate": 0.0002492260403252596,
      "loss": 2.101,
      "step": 127674
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.123365044593811,
      "learning_rate": 0.0002492220087792915,
      "loss": 2.1869,
      "step": 127675
    },
    {
      "epoch": 1.66,
      "grad_norm": 0.9622277617454529,
      "learning_rate": 0.0002492179772427639,
      "loss": 2.0408,
      "step": 127676
    },
    {
      "epoch": 1.66,
      "grad_norm": 0.9137107133865356,
      "learning_rate": 0.0002492139457156778,
      "loss": 2.2506,
      "step": 127677
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.0045197010040283,
      "learning_rate": 0.0002492099141980337,
      "loss": 2.0828,
      "step": 127678
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.09615159034729,
      "learning_rate": 0.0002492058826898324,
      "loss": 2.1937,
      "step": 127679
    },
    {
      "epoch": 1.66,
      "grad_norm": 0.9871110320091248,
      "learning_rate": 0.00024920185119107466,
      "loss": 1.9288,
      "step": 127680
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.0974663496017456,
      "learning_rate": 0.00024919781970176127,
      "loss": 2.0218,
      "step": 127681
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.061687707901001,
      "learning_rate": 0.000249193788221893,
      "loss": 2.3632,
      "step": 127682
    },
    {
      "epoch": 1.66,
      "grad_norm": 0.9770195484161377,
      "learning_rate": 0.00024918975675147065,
      "loss": 2.2019,
      "step": 127683
    },
    {
      "epoch": 1.66,
      "grad_norm": 0.9584320783615112,
      "learning_rate": 0.00024918572529049473,
      "loss": 2.2455,
      "step": 127684
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.0467690229415894,
      "learning_rate": 0.0002491816938389663,
      "loss": 2.2102,
      "step": 127685
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.0156595706939697,
      "learning_rate": 0.0002491776623968858,
      "loss": 2.1102,
      "step": 127686
    },
    {
      "epoch": 1.66,
      "grad_norm": 0.9751831889152527,
      "learning_rate": 0.0002491736309642542,
      "loss": 1.9764,
      "step": 127687
    },
    {
      "epoch": 1.66,
      "grad_norm": 0.9762004613876343,
      "learning_rate": 0.00024916959954107216,
      "loss": 2.0311,
      "step": 127688
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.2373318672180176,
      "learning_rate": 0.0002491655681273405,
      "loss": 2.1299,
      "step": 127689
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.1074743270874023,
      "learning_rate": 0.00024916153672305984,
      "loss": 2.349,
      "step": 127690
    },
    {
      "epoch": 1.66,
      "grad_norm": 0.9732449054718018,
      "learning_rate": 0.00024915750532823104,
      "loss": 2.0293,
      "step": 127691
    },
    {
      "epoch": 1.66,
      "grad_norm": 0.9399436116218567,
      "learning_rate": 0.0002491534739428549,
      "loss": 2.1208,
      "step": 127692
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.18009352684021,
      "learning_rate": 0.000249149442566932,
      "loss": 2.0965,
      "step": 127693
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.1351022720336914,
      "learning_rate": 0.0002491454112004632,
      "loss": 2.1887,
      "step": 127694
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.0141453742980957,
      "learning_rate": 0.00024914137984344926,
      "loss": 2.1442,
      "step": 127695
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.3241169452667236,
      "learning_rate": 0.00024913734849589086,
      "loss": 2.2559,
      "step": 127696
    },
    {
      "epoch": 1.66,
      "grad_norm": 0.9838215708732605,
      "learning_rate": 0.00024913331715778876,
      "loss": 2.3222,
      "step": 127697
    },
    {
      "epoch": 1.66,
      "grad_norm": 0.9460629820823669,
      "learning_rate": 0.0002491292858291438,
      "loss": 2.1291,
      "step": 127698
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.0462857484817505,
      "learning_rate": 0.00024912525450995667,
      "loss": 2.2011,
      "step": 127699
    },
    {
      "epoch": 1.66,
      "grad_norm": 0.9978224635124207,
      "learning_rate": 0.00024912122320022804,
      "loss": 2.0165,
      "step": 127700
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.004334568977356,
      "learning_rate": 0.00024911719189995885,
      "loss": 2.0212,
      "step": 127701
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.1179009675979614,
      "learning_rate": 0.0002491131606091496,
      "loss": 2.0093,
      "step": 127702
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.0768133401870728,
      "learning_rate": 0.00024910912932780125,
      "loss": 1.9833,
      "step": 127703
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.0984628200531006,
      "learning_rate": 0.0002491050980559144,
      "loss": 2.2114,
      "step": 127704
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.083229660987854,
      "learning_rate": 0.00024910106679348995,
      "loss": 1.9794,
      "step": 127705
    },
    {
      "epoch": 1.66,
      "grad_norm": 0.9613410830497742,
      "learning_rate": 0.00024909703554052857,
      "loss": 2.1051,
      "step": 127706
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.2615474462509155,
      "learning_rate": 0.0002490930042970311,
      "loss": 2.2044,
      "step": 127707
    },
    {
      "epoch": 1.66,
      "grad_norm": 0.9512956738471985,
      "learning_rate": 0.00024908897306299803,
      "loss": 2.2209,
      "step": 127708
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.0608941316604614,
      "learning_rate": 0.0002490849418384304,
      "loss": 2.0933,
      "step": 127709
    },
    {
      "epoch": 1.66,
      "grad_norm": 0.9722693562507629,
      "learning_rate": 0.00024908091062332874,
      "loss": 2.2157,
      "step": 127710
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.0287777185440063,
      "learning_rate": 0.00024907687941769387,
      "loss": 2.4352,
      "step": 127711
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.0755507946014404,
      "learning_rate": 0.0002490728482215267,
      "loss": 2.2612,
      "step": 127712
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.1202218532562256,
      "learning_rate": 0.0002490688170348278,
      "loss": 1.9833,
      "step": 127713
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.0474687814712524,
      "learning_rate": 0.00024906478585759803,
      "loss": 2.3553,
      "step": 127714
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.0550801753997803,
      "learning_rate": 0.000249060754689838,
      "loss": 2.1662,
      "step": 127715
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.1747851371765137,
      "learning_rate": 0.0002490567235315485,
      "loss": 2.1378,
      "step": 127716
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.2096521854400635,
      "learning_rate": 0.0002490526923827304,
      "loss": 2.1146,
      "step": 127717
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.0653514862060547,
      "learning_rate": 0.0002490486612433843,
      "loss": 2.0994,
      "step": 127718
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.000563383102417,
      "learning_rate": 0.00024904463011351103,
      "loss": 2.207,
      "step": 127719
    },
    {
      "epoch": 1.66,
      "grad_norm": 0.9714558720588684,
      "learning_rate": 0.0002490405989931114,
      "loss": 1.9586,
      "step": 127720
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.1103801727294922,
      "learning_rate": 0.00024903656788218605,
      "loss": 2.2079,
      "step": 127721
    },
    {
      "epoch": 1.66,
      "grad_norm": 0.9836301207542419,
      "learning_rate": 0.00024903253678073574,
      "loss": 2.0679,
      "step": 127722
    },
    {
      "epoch": 1.66,
      "grad_norm": 0.9912042617797852,
      "learning_rate": 0.00024902850568876125,
      "loss": 2.2296,
      "step": 127723
    },
    {
      "epoch": 1.66,
      "grad_norm": 0.9626169204711914,
      "learning_rate": 0.0002490244746062633,
      "loss": 2.0387,
      "step": 127724
    },
    {
      "epoch": 1.66,
      "grad_norm": 0.9003872871398926,
      "learning_rate": 0.00024902044353324266,
      "loss": 2.1483,
      "step": 127725
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.1811015605926514,
      "learning_rate": 0.0002490164124697002,
      "loss": 2.149,
      "step": 127726
    },
    {
      "epoch": 1.66,
      "grad_norm": 0.9596190452575684,
      "learning_rate": 0.00024901238141563644,
      "loss": 1.7992,
      "step": 127727
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.014012098312378,
      "learning_rate": 0.0002490083503710522,
      "loss": 1.8661,
      "step": 127728
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.0170289278030396,
      "learning_rate": 0.00024900431933594835,
      "loss": 2.1111,
      "step": 127729
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.1618643999099731,
      "learning_rate": 0.0002490002883103255,
      "loss": 2.3227,
      "step": 127730
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.0930923223495483,
      "learning_rate": 0.0002489962572941845,
      "loss": 2.0611,
      "step": 127731
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.0124598741531372,
      "learning_rate": 0.00024899222628752617,
      "loss": 2.2196,
      "step": 127732
    },
    {
      "epoch": 1.66,
      "grad_norm": 0.8854039311408997,
      "learning_rate": 0.000248988195290351,
      "loss": 2.228,
      "step": 127733
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.0621484518051147,
      "learning_rate": 0.0002489841643026599,
      "loss": 2.1457,
      "step": 127734
    },
    {
      "epoch": 1.66,
      "grad_norm": 0.9125187397003174,
      "learning_rate": 0.00024898013332445363,
      "loss": 1.9461,
      "step": 127735
    },
    {
      "epoch": 1.66,
      "grad_norm": 0.9732009172439575,
      "learning_rate": 0.00024897610235573287,
      "loss": 2.1125,
      "step": 127736
    },
    {
      "epoch": 1.66,
      "grad_norm": 0.9763994216918945,
      "learning_rate": 0.00024897207139649844,
      "loss": 2.1057,
      "step": 127737
    },
    {
      "epoch": 1.66,
      "grad_norm": 0.9873765110969543,
      "learning_rate": 0.0002489680404467512,
      "loss": 2.0054,
      "step": 127738
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.0720266103744507,
      "learning_rate": 0.00024896400950649163,
      "loss": 2.0923,
      "step": 127739
    },
    {
      "epoch": 1.66,
      "grad_norm": 0.9303669929504395,
      "learning_rate": 0.0002489599785757206,
      "loss": 2.1096,
      "step": 127740
    },
    {
      "epoch": 1.66,
      "grad_norm": 0.9540020823478699,
      "learning_rate": 0.0002489559476544389,
      "loss": 2.1551,
      "step": 127741
    },
    {
      "epoch": 1.66,
      "grad_norm": 0.9512836337089539,
      "learning_rate": 0.00024895191674264726,
      "loss": 2.0781,
      "step": 127742
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.0877934694290161,
      "learning_rate": 0.0002489478858403464,
      "loss": 2.0356,
      "step": 127743
    },
    {
      "epoch": 1.66,
      "grad_norm": 0.9880963563919067,
      "learning_rate": 0.0002489438549475372,
      "loss": 2.1612,
      "step": 127744
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.0451298952102661,
      "learning_rate": 0.0002489398240642202,
      "loss": 2.0518,
      "step": 127745
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.1221381425857544,
      "learning_rate": 0.0002489357931903962,
      "loss": 2.0025,
      "step": 127746
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.0710954666137695,
      "learning_rate": 0.0002489317623260661,
      "loss": 2.1698,
      "step": 127747
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.007493257522583,
      "learning_rate": 0.00024892773147123045,
      "loss": 2.0534,
      "step": 127748
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.0697340965270996,
      "learning_rate": 0.00024892370062589016,
      "loss": 2.0188,
      "step": 127749
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.0589486360549927,
      "learning_rate": 0.000248919669790046,
      "loss": 2.1409,
      "step": 127750
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.0677298307418823,
      "learning_rate": 0.00024891563896369857,
      "loss": 2.0624,
      "step": 127751
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.1237659454345703,
      "learning_rate": 0.0002489116081468487,
      "loss": 1.9247,
      "step": 127752
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.2827057838439941,
      "learning_rate": 0.00024890757733949705,
      "loss": 2.0612,
      "step": 127753
    },
    {
      "epoch": 1.66,
      "grad_norm": 0.9375758767127991,
      "learning_rate": 0.0002489035465416445,
      "loss": 1.969,
      "step": 127754
    },
    {
      "epoch": 1.66,
      "grad_norm": 0.8995195031166077,
      "learning_rate": 0.0002488995157532917,
      "loss": 2.2518,
      "step": 127755
    },
    {
      "epoch": 1.66,
      "grad_norm": 0.9155216813087463,
      "learning_rate": 0.00024889548497443963,
      "loss": 1.8689,
      "step": 127756
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.0666884183883667,
      "learning_rate": 0.00024889145420508863,
      "loss": 2.3628,
      "step": 127757
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.645311713218689,
      "learning_rate": 0.00024888742344523975,
      "loss": 2.0972,
      "step": 127758
    },
    {
      "epoch": 1.66,
      "grad_norm": 0.9579523801803589,
      "learning_rate": 0.0002488833926948937,
      "loss": 2.1074,
      "step": 127759
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.0283584594726562,
      "learning_rate": 0.0002488793619540511,
      "loss": 2.2026,
      "step": 127760
    },
    {
      "epoch": 1.66,
      "grad_norm": 0.9794191718101501,
      "learning_rate": 0.00024887533122271285,
      "loss": 2.2675,
      "step": 127761
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.0215075016021729,
      "learning_rate": 0.0002488713005008797,
      "loss": 2.0869,
      "step": 127762
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.0130541324615479,
      "learning_rate": 0.00024886726978855234,
      "loss": 2.0613,
      "step": 127763
    },
    {
      "epoch": 1.66,
      "grad_norm": 0.901427149772644,
      "learning_rate": 0.0002488632390857314,
      "loss": 2.3631,
      "step": 127764
    },
    {
      "epoch": 1.66,
      "grad_norm": 0.9431746006011963,
      "learning_rate": 0.0002488592083924178,
      "loss": 2.0297,
      "step": 127765
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.037657618522644,
      "learning_rate": 0.00024885517770861223,
      "loss": 2.0217,
      "step": 127766
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.0348752737045288,
      "learning_rate": 0.00024885114703431546,
      "loss": 2.2351,
      "step": 127767
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.039401888847351,
      "learning_rate": 0.0002488471163695283,
      "loss": 2.2303,
      "step": 127768
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.097480297088623,
      "learning_rate": 0.00024884308571425135,
      "loss": 2.0311,
      "step": 127769
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.017265796661377,
      "learning_rate": 0.00024883905506848546,
      "loss": 2.0463,
      "step": 127770
    },
    {
      "epoch": 1.66,
      "grad_norm": 0.8989935517311096,
      "learning_rate": 0.00024883502443223134,
      "loss": 2.1518,
      "step": 127771
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.1071820259094238,
      "learning_rate": 0.00024883099380548974,
      "loss": 2.1335,
      "step": 127772
    },
    {
      "epoch": 1.66,
      "grad_norm": 0.9630541801452637,
      "learning_rate": 0.0002488269631882614,
      "loss": 2.052,
      "step": 127773
    },
    {
      "epoch": 1.66,
      "grad_norm": 0.9166226387023926,
      "learning_rate": 0.0002488229325805472,
      "loss": 2.0658,
      "step": 127774
    },
    {
      "epoch": 1.66,
      "grad_norm": 0.9676320552825928,
      "learning_rate": 0.0002488189019823477,
      "loss": 2.1362,
      "step": 127775
    },
    {
      "epoch": 1.66,
      "grad_norm": 0.966516375541687,
      "learning_rate": 0.00024881487139366375,
      "loss": 2.1176,
      "step": 127776
    },
    {
      "epoch": 1.66,
      "grad_norm": 0.903668224811554,
      "learning_rate": 0.0002488108408144961,
      "loss": 2.168,
      "step": 127777
    },
    {
      "epoch": 1.66,
      "grad_norm": 0.9165112376213074,
      "learning_rate": 0.00024880681024484546,
      "loss": 2.1328,
      "step": 127778
    },
    {
      "epoch": 1.66,
      "grad_norm": 0.8698633313179016,
      "learning_rate": 0.0002488027796847126,
      "loss": 2.1146,
      "step": 127779
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.0361300706863403,
      "learning_rate": 0.00024879874913409823,
      "loss": 2.0234,
      "step": 127780
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.0192992687225342,
      "learning_rate": 0.00024879471859300323,
      "loss": 2.2241,
      "step": 127781
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.0480241775512695,
      "learning_rate": 0.0002487906880614282,
      "loss": 1.9547,
      "step": 127782
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.0214582681655884,
      "learning_rate": 0.00024878665753937394,
      "loss": 2.1103,
      "step": 127783
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.0108383893966675,
      "learning_rate": 0.0002487826270268413,
      "loss": 2.1598,
      "step": 127784
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.0554702281951904,
      "learning_rate": 0.00024877859652383083,
      "loss": 2.0779,
      "step": 127785
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.0824854373931885,
      "learning_rate": 0.00024877456603034336,
      "loss": 2.156,
      "step": 127786
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.0895450115203857,
      "learning_rate": 0.0002487705355463798,
      "loss": 1.9793,
      "step": 127787
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.107564091682434,
      "learning_rate": 0.0002487665050719407,
      "loss": 2.1462,
      "step": 127788
    },
    {
      "epoch": 1.66,
      "grad_norm": 0.978569746017456,
      "learning_rate": 0.0002487624746070268,
      "loss": 2.1072,
      "step": 127789
    },
    {
      "epoch": 1.66,
      "grad_norm": 0.9747722744941711,
      "learning_rate": 0.00024875844415163894,
      "loss": 2.198,
      "step": 127790
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.073048710823059,
      "learning_rate": 0.000248754413705778,
      "loss": 2.191,
      "step": 127791
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.0497376918792725,
      "learning_rate": 0.0002487503832694444,
      "loss": 2.2141,
      "step": 127792
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.1370776891708374,
      "learning_rate": 0.0002487463528426392,
      "loss": 2.0797,
      "step": 127793
    },
    {
      "epoch": 1.66,
      "grad_norm": 0.9544464945793152,
      "learning_rate": 0.000248742322425363,
      "loss": 1.9853,
      "step": 127794
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.0303391218185425,
      "learning_rate": 0.0002487382920176165,
      "loss": 1.9853,
      "step": 127795
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.1364240646362305,
      "learning_rate": 0.00024873426161940054,
      "loss": 2.3377,
      "step": 127796
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.005639672279358,
      "learning_rate": 0.00024873023123071585,
      "loss": 1.9689,
      "step": 127797
    },
    {
      "epoch": 1.66,
      "grad_norm": 0.9675922989845276,
      "learning_rate": 0.0002487262008515632,
      "loss": 2.0428,
      "step": 127798
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.0238324403762817,
      "learning_rate": 0.0002487221704819434,
      "loss": 1.9936,
      "step": 127799
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.0117335319519043,
      "learning_rate": 0.000248718140121857,
      "loss": 2.2527,
      "step": 127800
    },
    {
      "epoch": 1.66,
      "grad_norm": 0.8866723775863647,
      "learning_rate": 0.00024871410977130487,
      "loss": 2.0909,
      "step": 127801
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.047450304031372,
      "learning_rate": 0.0002487100794302878,
      "loss": 2.1785,
      "step": 127802
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.0266563892364502,
      "learning_rate": 0.0002487060490988064,
      "loss": 2.0519,
      "step": 127803
    },
    {
      "epoch": 1.66,
      "grad_norm": 0.8728979229927063,
      "learning_rate": 0.0002487020187768616,
      "loss": 1.919,
      "step": 127804
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.1531472206115723,
      "learning_rate": 0.00024869798846445415,
      "loss": 1.9725,
      "step": 127805
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.0486592054367065,
      "learning_rate": 0.0002486939581615846,
      "loss": 2.2298,
      "step": 127806
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.0134087800979614,
      "learning_rate": 0.00024868992786825376,
      "loss": 2.2187,
      "step": 127807
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.0226815938949585,
      "learning_rate": 0.00024868589758446255,
      "loss": 2.1335,
      "step": 127808
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.0699725151062012,
      "learning_rate": 0.0002486818673102115,
      "loss": 2.1072,
      "step": 127809
    },
    {
      "epoch": 1.66,
      "grad_norm": 0.9909036159515381,
      "learning_rate": 0.0002486778370455015,
      "loss": 2.1212,
      "step": 127810
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.1523104906082153,
      "learning_rate": 0.00024867380679033336,
      "loss": 2.0847,
      "step": 127811
    },
    {
      "epoch": 1.66,
      "grad_norm": 0.9558950066566467,
      "learning_rate": 0.00024866977654470763,
      "loss": 1.9696,
      "step": 127812
    },
    {
      "epoch": 1.66,
      "grad_norm": 0.9567796587944031,
      "learning_rate": 0.00024866574630862516,
      "loss": 2.0064,
      "step": 127813
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.053596019744873,
      "learning_rate": 0.00024866171608208665,
      "loss": 2.2758,
      "step": 127814
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.3916109800338745,
      "learning_rate": 0.0002486576858650929,
      "loss": 2.2758,
      "step": 127815
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.116577386856079,
      "learning_rate": 0.0002486536556576447,
      "loss": 2.4433,
      "step": 127816
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.0286872386932373,
      "learning_rate": 0.00024864962545974285,
      "loss": 1.9929,
      "step": 127817
    },
    {
      "epoch": 1.66,
      "grad_norm": 0.9241732954978943,
      "learning_rate": 0.00024864559527138793,
      "loss": 2.2036,
      "step": 127818
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.0147466659545898,
      "learning_rate": 0.0002486415650925807,
      "loss": 1.9722,
      "step": 127819
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.3328639268875122,
      "learning_rate": 0.000248637534923322,
      "loss": 2.3417,
      "step": 127820
    },
    {
      "epoch": 1.66,
      "grad_norm": 0.9590587019920349,
      "learning_rate": 0.00024863350476361253,
      "loss": 2.1499,
      "step": 127821
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.023864507675171,
      "learning_rate": 0.0002486294746134531,
      "loss": 2.007,
      "step": 127822
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.1000174283981323,
      "learning_rate": 0.00024862544447284453,
      "loss": 2.0714,
      "step": 127823
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.0149288177490234,
      "learning_rate": 0.0002486214143417873,
      "loss": 2.317,
      "step": 127824
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.591046929359436,
      "learning_rate": 0.00024861738422028235,
      "loss": 2.0766,
      "step": 127825
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.1050612926483154,
      "learning_rate": 0.0002486133541083304,
      "loss": 2.1717,
      "step": 127826
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.0847132205963135,
      "learning_rate": 0.00024860932400593217,
      "loss": 2.3061,
      "step": 127827
    },
    {
      "epoch": 1.66,
      "grad_norm": 0.9986518025398254,
      "learning_rate": 0.00024860529391308847,
      "loss": 1.9705,
      "step": 127828
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.166741132736206,
      "learning_rate": 0.00024860126382980017,
      "loss": 1.9881,
      "step": 127829
    },
    {
      "epoch": 1.66,
      "grad_norm": 0.9699539542198181,
      "learning_rate": 0.00024859723375606765,
      "loss": 2.1012,
      "step": 127830
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.032284140586853,
      "learning_rate": 0.00024859320369189194,
      "loss": 2.181,
      "step": 127831
    },
    {
      "epoch": 1.66,
      "grad_norm": 0.9576193690299988,
      "learning_rate": 0.00024858917363727374,
      "loss": 2.1265,
      "step": 127832
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.0476353168487549,
      "learning_rate": 0.00024858514359221375,
      "loss": 2.1746,
      "step": 127833
    },
    {
      "epoch": 1.66,
      "grad_norm": 0.9507395029067993,
      "learning_rate": 0.00024858111355671275,
      "loss": 2.0158,
      "step": 127834
    },
    {
      "epoch": 1.66,
      "grad_norm": 0.9661847352981567,
      "learning_rate": 0.0002485770835307716,
      "loss": 2.1404,
      "step": 127835
    },
    {
      "epoch": 1.66,
      "grad_norm": 0.9249542355537415,
      "learning_rate": 0.00024857305351439087,
      "loss": 1.8108,
      "step": 127836
    },
    {
      "epoch": 1.66,
      "grad_norm": 0.9354852437973022,
      "learning_rate": 0.00024856902350757135,
      "loss": 1.9899,
      "step": 127837
    },
    {
      "epoch": 1.66,
      "grad_norm": 0.9547032117843628,
      "learning_rate": 0.0002485649935103138,
      "loss": 1.971,
      "step": 127838
    },
    {
      "epoch": 1.66,
      "grad_norm": 0.9718700647354126,
      "learning_rate": 0.000248560963522619,
      "loss": 2.1821,
      "step": 127839
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.0098466873168945,
      "learning_rate": 0.0002485569335444877,
      "loss": 1.9706,
      "step": 127840
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.0250385999679565,
      "learning_rate": 0.00024855290357592077,
      "loss": 2.1447,
      "step": 127841
    },
    {
      "epoch": 1.66,
      "grad_norm": 0.9982016086578369,
      "learning_rate": 0.0002485488736169186,
      "loss": 2.1326,
      "step": 127842
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.0488296747207642,
      "learning_rate": 0.0002485448436674823,
      "loss": 2.1255,
      "step": 127843
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.0579761266708374,
      "learning_rate": 0.0002485408137276124,
      "loss": 2.1292,
      "step": 127844
    },
    {
      "epoch": 1.66,
      "grad_norm": 0.9898473620414734,
      "learning_rate": 0.0002485367837973098,
      "loss": 2.0599,
      "step": 127845
    },
    {
      "epoch": 1.66,
      "grad_norm": 0.9771308898925781,
      "learning_rate": 0.00024853275387657514,
      "loss": 2.2036,
      "step": 127846
    },
    {
      "epoch": 1.66,
      "grad_norm": 0.9905821681022644,
      "learning_rate": 0.00024852872396540924,
      "loss": 2.3159,
      "step": 127847
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.224716305732727,
      "learning_rate": 0.00024852469406381295,
      "loss": 2.2311,
      "step": 127848
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.1329514980316162,
      "learning_rate": 0.00024852066417178675,
      "loss": 2.1108,
      "step": 127849
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.0139321088790894,
      "learning_rate": 0.0002485166342893315,
      "loss": 2.3432,
      "step": 127850
    },
    {
      "epoch": 1.66,
      "grad_norm": 0.9977197647094727,
      "learning_rate": 0.00024851260441644804,
      "loss": 2.3824,
      "step": 127851
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.0329036712646484,
      "learning_rate": 0.00024850857455313704,
      "loss": 2.1633,
      "step": 127852
    },
    {
      "epoch": 1.66,
      "grad_norm": 0.9741498231887817,
      "learning_rate": 0.0002485045446993992,
      "loss": 2.1913,
      "step": 127853
    },
    {
      "epoch": 1.66,
      "grad_norm": 0.9932448267936707,
      "learning_rate": 0.00024850051485523556,
      "loss": 2.1548,
      "step": 127854
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.1327271461486816,
      "learning_rate": 0.0002484964850206465,
      "loss": 2.0643,
      "step": 127855
    },
    {
      "epoch": 1.66,
      "grad_norm": 0.9569797515869141,
      "learning_rate": 0.0002484924551956329,
      "loss": 1.8866,
      "step": 127856
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.0880444049835205,
      "learning_rate": 0.00024848842538019553,
      "loss": 2.2488,
      "step": 127857
    },
    {
      "epoch": 1.66,
      "grad_norm": 0.9445751309394836,
      "learning_rate": 0.0002484843955743351,
      "loss": 2.2835,
      "step": 127858
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.0003584623336792,
      "learning_rate": 0.0002484803657780524,
      "loss": 2.1279,
      "step": 127859
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.040341854095459,
      "learning_rate": 0.00024847633599134827,
      "loss": 2.0557,
      "step": 127860
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.0060495138168335,
      "learning_rate": 0.0002484723062142233,
      "loss": 2.2379,
      "step": 127861
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.1432547569274902,
      "learning_rate": 0.00024846827644667833,
      "loss": 2.0137,
      "step": 127862
    },
    {
      "epoch": 1.66,
      "grad_norm": 0.9010387659072876,
      "learning_rate": 0.000248464246688714,
      "loss": 2.1515,
      "step": 127863
    },
    {
      "epoch": 1.66,
      "grad_norm": 0.9107281565666199,
      "learning_rate": 0.0002484602169403312,
      "loss": 2.0471,
      "step": 127864
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.0707803964614868,
      "learning_rate": 0.00024845618720153057,
      "loss": 2.0396,
      "step": 127865
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.1313539743423462,
      "learning_rate": 0.000248452157472313,
      "loss": 2.3285,
      "step": 127866
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.0118075609207153,
      "learning_rate": 0.00024844812775267906,
      "loss": 2.0778,
      "step": 127867
    },
    {
      "epoch": 1.66,
      "grad_norm": 0.986646831035614,
      "learning_rate": 0.00024844409804262963,
      "loss": 1.9301,
      "step": 127868
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.4060077667236328,
      "learning_rate": 0.0002484400683421655,
      "loss": 1.9698,
      "step": 127869
    },
    {
      "epoch": 1.66,
      "grad_norm": 0.9474387168884277,
      "learning_rate": 0.00024843603865128713,
      "loss": 2.0356,
      "step": 127870
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.0329365730285645,
      "learning_rate": 0.00024843200896999557,
      "loss": 2.0981,
      "step": 127871
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.0648207664489746,
      "learning_rate": 0.00024842797929829156,
      "loss": 2.3878,
      "step": 127872
    },
    {
      "epoch": 1.66,
      "grad_norm": 1.0512007474899292,
      "learning_rate": 0.0002484239496361757,
      "loss": 1.96,
      "step": 127873
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.174181580543518,
      "learning_rate": 0.0002484199199836487,
      "loss": 2.2249,
      "step": 127874
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.1455714702606201,
      "learning_rate": 0.0002484158903407115,
      "loss": 2.3688,
      "step": 127875
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.0292863845825195,
      "learning_rate": 0.00024841186070736484,
      "loss": 2.0823,
      "step": 127876
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.0039626359939575,
      "learning_rate": 0.0002484078310836093,
      "loss": 2.1445,
      "step": 127877
    },
    {
      "epoch": 1.67,
      "grad_norm": 0.961144208908081,
      "learning_rate": 0.0002484038014694458,
      "loss": 2.1326,
      "step": 127878
    },
    {
      "epoch": 1.67,
      "grad_norm": 0.9880547523498535,
      "learning_rate": 0.00024839977186487486,
      "loss": 2.3476,
      "step": 127879
    },
    {
      "epoch": 1.67,
      "grad_norm": 0.9774268865585327,
      "learning_rate": 0.00024839574226989745,
      "loss": 2.1833,
      "step": 127880
    },
    {
      "epoch": 1.67,
      "grad_norm": 0.9233599305152893,
      "learning_rate": 0.0002483917126845142,
      "loss": 2.0694,
      "step": 127881
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.0124248266220093,
      "learning_rate": 0.0002483876831087259,
      "loss": 2.1221,
      "step": 127882
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.0004079341888428,
      "learning_rate": 0.00024838365354253343,
      "loss": 2.1756,
      "step": 127883
    },
    {
      "epoch": 1.67,
      "grad_norm": 0.971765398979187,
      "learning_rate": 0.00024837962398593737,
      "loss": 2.142,
      "step": 127884
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.0080376863479614,
      "learning_rate": 0.0002483755944389385,
      "loss": 1.8791,
      "step": 127885
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.0584936141967773,
      "learning_rate": 0.00024837156490153755,
      "loss": 2.1794,
      "step": 127886
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.103758454322815,
      "learning_rate": 0.00024836753537373524,
      "loss": 2.0311,
      "step": 127887
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.0760031938552856,
      "learning_rate": 0.00024836350585553246,
      "loss": 2.3243,
      "step": 127888
    },
    {
      "epoch": 1.67,
      "grad_norm": 0.9454512000083923,
      "learning_rate": 0.00024835947634692986,
      "loss": 2.0216,
      "step": 127889
    },
    {
      "epoch": 1.67,
      "grad_norm": 0.9066638946533203,
      "learning_rate": 0.0002483554468479283,
      "loss": 2.0846,
      "step": 127890
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.0207405090332031,
      "learning_rate": 0.0002483514173585284,
      "loss": 1.9632,
      "step": 127891
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.037829041481018,
      "learning_rate": 0.00024834738787873087,
      "loss": 2.1962,
      "step": 127892
    },
    {
      "epoch": 1.67,
      "grad_norm": 0.9669731855392456,
      "learning_rate": 0.00024834335840853656,
      "loss": 1.8502,
      "step": 127893
    },
    {
      "epoch": 1.67,
      "grad_norm": 0.9216722249984741,
      "learning_rate": 0.0002483393289479462,
      "loss": 1.8078,
      "step": 127894
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.0322961807250977,
      "learning_rate": 0.00024833529949696046,
      "loss": 2.0048,
      "step": 127895
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.108911156654358,
      "learning_rate": 0.0002483312700555804,
      "loss": 2.1723,
      "step": 127896
    },
    {
      "epoch": 1.67,
      "grad_norm": 0.9534231424331665,
      "learning_rate": 0.0002483272406238063,
      "loss": 2.0455,
      "step": 127897
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.0208461284637451,
      "learning_rate": 0.0002483232112016392,
      "loss": 1.9757,
      "step": 127898
    },
    {
      "epoch": 1.67,
      "grad_norm": 0.9960980415344238,
      "learning_rate": 0.00024831918178907977,
      "loss": 1.9245,
      "step": 127899
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.0207602977752686,
      "learning_rate": 0.0002483151523861288,
      "loss": 1.905,
      "step": 127900
    },
    {
      "epoch": 1.67,
      "grad_norm": 0.971164882183075,
      "learning_rate": 0.00024831112299278704,
      "loss": 1.9907,
      "step": 127901
    },
    {
      "epoch": 1.67,
      "grad_norm": 0.9439325928688049,
      "learning_rate": 0.0002483070936090553,
      "loss": 2.1787,
      "step": 127902
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.1764891147613525,
      "learning_rate": 0.00024830306423493417,
      "loss": 2.1283,
      "step": 127903
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.0344297885894775,
      "learning_rate": 0.0002482990348704245,
      "loss": 2.044,
      "step": 127904
    },
    {
      "epoch": 1.67,
      "grad_norm": 0.9492310881614685,
      "learning_rate": 0.00024829500551552694,
      "loss": 2.0733,
      "step": 127905
    },
    {
      "epoch": 1.67,
      "grad_norm": 0.9330412149429321,
      "learning_rate": 0.0002482909761702423,
      "loss": 2.1292,
      "step": 127906
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.147157907485962,
      "learning_rate": 0.0002482869468345714,
      "loss": 2.0155,
      "step": 127907
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.062515377998352,
      "learning_rate": 0.000248282917508515,
      "loss": 2.1733,
      "step": 127908
    },
    {
      "epoch": 1.67,
      "grad_norm": 0.9835943579673767,
      "learning_rate": 0.0002482788881920737,
      "loss": 2.0811,
      "step": 127909
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.188688039779663,
      "learning_rate": 0.00024827485888524834,
      "loss": 2.0823,
      "step": 127910
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.1369352340698242,
      "learning_rate": 0.00024827082958803963,
      "loss": 1.9389,
      "step": 127911
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.0296655893325806,
      "learning_rate": 0.0002482668003004484,
      "loss": 2.2315,
      "step": 127912
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.0065877437591553,
      "learning_rate": 0.0002482627710224753,
      "loss": 2.3435,
      "step": 127913
    },
    {
      "epoch": 1.67,
      "grad_norm": 0.9449348449707031,
      "learning_rate": 0.00024825874175412113,
      "loss": 2.2336,
      "step": 127914
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.0034033060073853,
      "learning_rate": 0.0002482547124953868,
      "loss": 1.873,
      "step": 127915
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.0257172584533691,
      "learning_rate": 0.0002482506832462727,
      "loss": 1.9982,
      "step": 127916
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.0023703575134277,
      "learning_rate": 0.00024824665400677987,
      "loss": 2.1775,
      "step": 127917
    },
    {
      "epoch": 1.67,
      "grad_norm": 0.9750579595565796,
      "learning_rate": 0.0002482426247769089,
      "loss": 1.8954,
      "step": 127918
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.001261591911316,
      "learning_rate": 0.0002482385955566606,
      "loss": 1.9639,
      "step": 127919
    },
    {
      "epoch": 1.67,
      "grad_norm": 0.9733685851097107,
      "learning_rate": 0.0002482345663460357,
      "loss": 2.1121,
      "step": 127920
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.059920072555542,
      "learning_rate": 0.00024823053714503516,
      "loss": 2.0714,
      "step": 127921
    },
    {
      "epoch": 1.67,
      "grad_norm": 0.9981296062469482,
      "learning_rate": 0.0002482265079536594,
      "loss": 2.2559,
      "step": 127922
    },
    {
      "epoch": 1.67,
      "grad_norm": 0.9625346064567566,
      "learning_rate": 0.0002482224787719093,
      "loss": 1.9934,
      "step": 127923
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.0740615129470825,
      "learning_rate": 0.0002482184495997856,
      "loss": 2.0842,
      "step": 127924
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.1232845783233643,
      "learning_rate": 0.0002482144204372891,
      "loss": 2.073,
      "step": 127925
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.0994504690170288,
      "learning_rate": 0.0002482103912844205,
      "loss": 2.2903,
      "step": 127926
    },
    {
      "epoch": 1.67,
      "grad_norm": 0.9854862093925476,
      "learning_rate": 0.0002482063621411807,
      "loss": 2.0265,
      "step": 127927
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.0007121562957764,
      "learning_rate": 0.00024820233300757013,
      "loss": 2.0102,
      "step": 127928
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.2218025922775269,
      "learning_rate": 0.0002481983038835898,
      "loss": 1.9278,
      "step": 127929
    },
    {
      "epoch": 1.67,
      "grad_norm": 0.9873669743537903,
      "learning_rate": 0.0002481942747692404,
      "loss": 1.9506,
      "step": 127930
    },
    {
      "epoch": 1.67,
      "grad_norm": 0.9847833514213562,
      "learning_rate": 0.0002481902456645226,
      "loss": 1.9363,
      "step": 127931
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.0049629211425781,
      "learning_rate": 0.00024818621656943725,
      "loss": 1.9639,
      "step": 127932
    },
    {
      "epoch": 1.67,
      "grad_norm": 0.9501748085021973,
      "learning_rate": 0.00024818218748398515,
      "loss": 2.2031,
      "step": 127933
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.0418286323547363,
      "learning_rate": 0.00024817815840816684,
      "loss": 2.2182,
      "step": 127934
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.007795810699463,
      "learning_rate": 0.0002481741293419832,
      "loss": 2.1341,
      "step": 127935
    },
    {
      "epoch": 1.67,
      "grad_norm": 0.9098193049430847,
      "learning_rate": 0.000248170100285435,
      "loss": 2.1168,
      "step": 127936
    },
    {
      "epoch": 1.67,
      "grad_norm": 0.9859989285469055,
      "learning_rate": 0.0002481660712385229,
      "loss": 2.0353,
      "step": 127937
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.04473876953125,
      "learning_rate": 0.0002481620422012477,
      "loss": 2.3864,
      "step": 127938
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.1955031156539917,
      "learning_rate": 0.0002481580131736104,
      "loss": 2.0215,
      "step": 127939
    },
    {
      "epoch": 1.67,
      "grad_norm": 0.9830241203308105,
      "learning_rate": 0.00024815398415561126,
      "loss": 2.0967,
      "step": 127940
    },
    {
      "epoch": 1.67,
      "grad_norm": 0.9393448233604431,
      "learning_rate": 0.0002481499551472513,
      "loss": 2.0694,
      "step": 127941
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.023308277130127,
      "learning_rate": 0.0002481459261485312,
      "loss": 2.0892,
      "step": 127942
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.054262399673462,
      "learning_rate": 0.0002481418971594518,
      "loss": 2.2662,
      "step": 127943
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.040035367012024,
      "learning_rate": 0.00024813786818001383,
      "loss": 2.1422,
      "step": 127944
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.0222887992858887,
      "learning_rate": 0.000248133839210218,
      "loss": 1.9906,
      "step": 127945
    },
    {
      "epoch": 1.67,
      "grad_norm": 0.9841532707214355,
      "learning_rate": 0.00024812981025006516,
      "loss": 2.175,
      "step": 127946
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.494075894355774,
      "learning_rate": 0.00024812578129955584,
      "loss": 2.1226,
      "step": 127947
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.6660219430923462,
      "learning_rate": 0.00024812175235869094,
      "loss": 2.1863,
      "step": 127948
    },
    {
      "epoch": 1.67,
      "grad_norm": 0.9668295383453369,
      "learning_rate": 0.00024811772342747115,
      "loss": 2.0837,
      "step": 127949
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.1642403602600098,
      "learning_rate": 0.0002481136945058973,
      "loss": 2.211,
      "step": 127950
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.1388740539550781,
      "learning_rate": 0.0002481096655939701,
      "loss": 2.0879,
      "step": 127951
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.1244251728057861,
      "learning_rate": 0.0002481056366916903,
      "loss": 2.1617,
      "step": 127952
    },
    {
      "epoch": 1.67,
      "grad_norm": 0.9922199249267578,
      "learning_rate": 0.00024810160779905867,
      "loss": 2.1719,
      "step": 127953
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.0295392274856567,
      "learning_rate": 0.0002480975789160758,
      "loss": 2.0115,
      "step": 127954
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.0098639726638794,
      "learning_rate": 0.00024809355004274265,
      "loss": 2.1126,
      "step": 127955
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.0419694185256958,
      "learning_rate": 0.0002480895211790599,
      "loss": 1.9152,
      "step": 127956
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.0364044904708862,
      "learning_rate": 0.00024808549232502834,
      "loss": 2.0244,
      "step": 127957
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.0352354049682617,
      "learning_rate": 0.0002480814634806486,
      "loss": 1.9469,
      "step": 127958
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.1112501621246338,
      "learning_rate": 0.00024807743464592144,
      "loss": 1.9466,
      "step": 127959
    },
    {
      "epoch": 1.67,
      "grad_norm": 0.9395800232887268,
      "learning_rate": 0.0002480734058208477,
      "loss": 2.1117,
      "step": 127960
    },
    {
      "epoch": 1.67,
      "grad_norm": 0.9493386149406433,
      "learning_rate": 0.0002480693770054282,
      "loss": 2.0974,
      "step": 127961
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.0280002355575562,
      "learning_rate": 0.0002480653481996635,
      "loss": 1.8871,
      "step": 127962
    },
    {
      "epoch": 1.67,
      "grad_norm": 0.9630963802337646,
      "learning_rate": 0.0002480613194035545,
      "loss": 2.1058,
      "step": 127963
    },
    {
      "epoch": 1.67,
      "grad_norm": 0.9219418168067932,
      "learning_rate": 0.00024805729061710177,
      "loss": 2.2681,
      "step": 127964
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.2208398580551147,
      "learning_rate": 0.00024805326184030625,
      "loss": 1.9396,
      "step": 127965
    },
    {
      "epoch": 1.67,
      "grad_norm": 0.9654063582420349,
      "learning_rate": 0.00024804923307316854,
      "loss": 2.0517,
      "step": 127966
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.0996479988098145,
      "learning_rate": 0.00024804520431568944,
      "loss": 2.0555,
      "step": 127967
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.0085091590881348,
      "learning_rate": 0.0002480411755678698,
      "loss": 2.0814,
      "step": 127968
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.0056533813476562,
      "learning_rate": 0.00024803714682971033,
      "loss": 1.9883,
      "step": 127969
    },
    {
      "epoch": 1.67,
      "grad_norm": 0.9827340841293335,
      "learning_rate": 0.0002480331181012117,
      "loss": 2.1283,
      "step": 127970
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.0751569271087646,
      "learning_rate": 0.0002480290893823746,
      "loss": 2.0306,
      "step": 127971
    },
    {
      "epoch": 1.67,
      "grad_norm": 0.9747301936149597,
      "learning_rate": 0.00024802506067319994,
      "loss": 2.0546,
      "step": 127972
    },
    {
      "epoch": 1.67,
      "grad_norm": 0.9253837466239929,
      "learning_rate": 0.00024802103197368836,
      "loss": 2.1845,
      "step": 127973
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.0772991180419922,
      "learning_rate": 0.00024801700328384065,
      "loss": 2.1852,
      "step": 127974
    },
    {
      "epoch": 1.67,
      "grad_norm": 0.9757603406906128,
      "learning_rate": 0.00024801297460365774,
      "loss": 2.1537,
      "step": 127975
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.0896456241607666,
      "learning_rate": 0.00024800894593314,
      "loss": 2.0778,
      "step": 127976
    },
    {
      "epoch": 1.67,
      "grad_norm": 0.955049455165863,
      "learning_rate": 0.0002480049172722884,
      "loss": 2.2176,
      "step": 127977
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.1384258270263672,
      "learning_rate": 0.0002480008886211037,
      "loss": 2.248,
      "step": 127978
    },
    {
      "epoch": 1.67,
      "grad_norm": 0.910761296749115,
      "learning_rate": 0.0002479968599795866,
      "loss": 1.93,
      "step": 127979
    },
    {
      "epoch": 1.67,
      "grad_norm": 0.958570659160614,
      "learning_rate": 0.00024799283134773793,
      "loss": 2.2952,
      "step": 127980
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.0304545164108276,
      "learning_rate": 0.00024798880272555826,
      "loss": 2.1615,
      "step": 127981
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.0292993783950806,
      "learning_rate": 0.00024798477411304865,
      "loss": 2.1112,
      "step": 127982
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.0232086181640625,
      "learning_rate": 0.00024798074551020947,
      "loss": 2.2154,
      "step": 127983
    },
    {
      "epoch": 1.67,
      "grad_norm": 0.9149133563041687,
      "learning_rate": 0.0002479767169170417,
      "loss": 1.9822,
      "step": 127984
    },
    {
      "epoch": 1.67,
      "grad_norm": 0.9343137145042419,
      "learning_rate": 0.00024797268833354604,
      "loss": 2.0549,
      "step": 127985
    },
    {
      "epoch": 1.67,
      "grad_norm": 0.939918041229248,
      "learning_rate": 0.0002479686597597232,
      "loss": 2.1695,
      "step": 127986
    },
    {
      "epoch": 1.67,
      "grad_norm": 0.9665600061416626,
      "learning_rate": 0.00024796463119557404,
      "loss": 1.9374,
      "step": 127987
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.0370630025863647,
      "learning_rate": 0.0002479606026410993,
      "loss": 2.3699,
      "step": 127988
    },
    {
      "epoch": 1.67,
      "grad_norm": 0.9576735496520996,
      "learning_rate": 0.00024795657409629955,
      "loss": 1.9575,
      "step": 127989
    },
    {
      "epoch": 1.67,
      "grad_norm": 0.9373168349266052,
      "learning_rate": 0.0002479525455611757,
      "loss": 2.0999,
      "step": 127990
    },
    {
      "epoch": 1.67,
      "grad_norm": 0.9793492555618286,
      "learning_rate": 0.0002479485170357284,
      "loss": 2.1911,
      "step": 127991
    },
    {
      "epoch": 1.67,
      "grad_norm": 0.9622822999954224,
      "learning_rate": 0.0002479444885199585,
      "loss": 1.9777,
      "step": 127992
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.0551666021347046,
      "learning_rate": 0.00024794046001386667,
      "loss": 1.9592,
      "step": 127993
    },
    {
      "epoch": 1.67,
      "grad_norm": 0.9747806191444397,
      "learning_rate": 0.0002479364315174538,
      "loss": 2.223,
      "step": 127994
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.0223276615142822,
      "learning_rate": 0.0002479324030307204,
      "loss": 2.3468,
      "step": 127995
    },
    {
      "epoch": 1.67,
      "grad_norm": 0.9535782933235168,
      "learning_rate": 0.0002479283745536674,
      "loss": 2.1525,
      "step": 127996
    },
    {
      "epoch": 1.67,
      "grad_norm": 0.9585540294647217,
      "learning_rate": 0.0002479243460862955,
      "loss": 2.141,
      "step": 127997
    },
    {
      "epoch": 1.67,
      "grad_norm": 0.8909156918525696,
      "learning_rate": 0.0002479203176286054,
      "loss": 1.9402,
      "step": 127998
    },
    {
      "epoch": 1.67,
      "grad_norm": 0.8719214797019958,
      "learning_rate": 0.0002479162891805979,
      "loss": 1.8991,
      "step": 127999
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.0522515773773193,
      "learning_rate": 0.0002479122607422739,
      "loss": 2.1603,
      "step": 128000
    },
    {
      "epoch": 1.67,
      "grad_norm": 0.9456911087036133,
      "learning_rate": 0.00024790823231363385,
      "loss": 1.9418,
      "step": 128001
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.0090715885162354,
      "learning_rate": 0.00024790420389467866,
      "loss": 2.1563,
      "step": 128002
    },
    {
      "epoch": 1.67,
      "grad_norm": 2.279305934906006,
      "learning_rate": 0.00024790017548540904,
      "loss": 2.0313,
      "step": 128003
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.1421008110046387,
      "learning_rate": 0.0002478961470858257,
      "loss": 2.3885,
      "step": 128004
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.1591404676437378,
      "learning_rate": 0.0002478921186959296,
      "loss": 2.2444,
      "step": 128005
    },
    {
      "epoch": 1.67,
      "grad_norm": 0.983407199382782,
      "learning_rate": 0.00024788809031572136,
      "loss": 1.872,
      "step": 128006
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.2403781414031982,
      "learning_rate": 0.0002478840619452016,
      "loss": 1.9915,
      "step": 128007
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.2064670324325562,
      "learning_rate": 0.0002478800335843712,
      "loss": 2.2717,
      "step": 128008
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.0153717994689941,
      "learning_rate": 0.0002478760052332309,
      "loss": 2.0636,
      "step": 128009
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.256034255027771,
      "learning_rate": 0.0002478719768917814,
      "loss": 2.0206,
      "step": 128010
    },
    {
      "epoch": 1.67,
      "grad_norm": 0.9524993300437927,
      "learning_rate": 0.0002478679485600235,
      "loss": 2.2223,
      "step": 128011
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.1750155687332153,
      "learning_rate": 0.000247863920237958,
      "loss": 2.1247,
      "step": 128012
    },
    {
      "epoch": 1.67,
      "grad_norm": 0.9289270639419556,
      "learning_rate": 0.0002478598919255855,
      "loss": 2.1184,
      "step": 128013
    },
    {
      "epoch": 1.67,
      "grad_norm": 0.98929363489151,
      "learning_rate": 0.00024785586362290686,
      "loss": 2.0817,
      "step": 128014
    },
    {
      "epoch": 1.67,
      "grad_norm": 0.9716688990592957,
      "learning_rate": 0.00024785183532992274,
      "loss": 2.0431,
      "step": 128015
    },
    {
      "epoch": 1.67,
      "grad_norm": 0.9760316014289856,
      "learning_rate": 0.000247847807046634,
      "loss": 2.2665,
      "step": 128016
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.2699562311172485,
      "learning_rate": 0.0002478437787730413,
      "loss": 2.2982,
      "step": 128017
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.1147767305374146,
      "learning_rate": 0.0002478397505091456,
      "loss": 2.0204,
      "step": 128018
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.0686752796173096,
      "learning_rate": 0.00024783572225494727,
      "loss": 1.9558,
      "step": 128019
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.1056150197982788,
      "learning_rate": 0.0002478316940104473,
      "loss": 2.1853,
      "step": 128020
    },
    {
      "epoch": 1.67,
      "grad_norm": 0.9713431596755981,
      "learning_rate": 0.00024782766577564647,
      "loss": 2.0897,
      "step": 128021
    },
    {
      "epoch": 1.67,
      "grad_norm": 0.9175464510917664,
      "learning_rate": 0.0002478236375505454,
      "loss": 1.8494,
      "step": 128022
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.0924359560012817,
      "learning_rate": 0.0002478196093351449,
      "loss": 2.1717,
      "step": 128023
    },
    {
      "epoch": 1.67,
      "grad_norm": 0.9010865688323975,
      "learning_rate": 0.00024781558112944586,
      "loss": 2.1686,
      "step": 128024
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.0627479553222656,
      "learning_rate": 0.0002478115529334487,
      "loss": 2.2467,
      "step": 128025
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.023244857788086,
      "learning_rate": 0.00024780752474715443,
      "loss": 2.1304,
      "step": 128026
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.1142863035202026,
      "learning_rate": 0.0002478034965705637,
      "loss": 2.0567,
      "step": 128027
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.0404757261276245,
      "learning_rate": 0.0002477994684036773,
      "loss": 2.0186,
      "step": 128028
    },
    {
      "epoch": 1.67,
      "grad_norm": 0.9957382082939148,
      "learning_rate": 0.000247795440246496,
      "loss": 2.1523,
      "step": 128029
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.1044914722442627,
      "learning_rate": 0.0002477914120990205,
      "loss": 2.1287,
      "step": 128030
    },
    {
      "epoch": 1.67,
      "grad_norm": 0.9411244988441467,
      "learning_rate": 0.0002477873839612516,
      "loss": 2.1566,
      "step": 128031
    },
    {
      "epoch": 1.67,
      "grad_norm": 0.933235764503479,
      "learning_rate": 0.0002477833558331899,
      "loss": 2.109,
      "step": 128032
    },
    {
      "epoch": 1.67,
      "grad_norm": 0.9822359085083008,
      "learning_rate": 0.00024777932771483633,
      "loss": 2.1652,
      "step": 128033
    },
    {
      "epoch": 1.67,
      "grad_norm": 0.8845738768577576,
      "learning_rate": 0.0002477752996061916,
      "loss": 2.197,
      "step": 128034
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.0031085014343262,
      "learning_rate": 0.00024777127150725633,
      "loss": 2.3222,
      "step": 128035
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.002023458480835,
      "learning_rate": 0.0002477672434180315,
      "loss": 2.1063,
      "step": 128036
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.0347399711608887,
      "learning_rate": 0.0002477632153385176,
      "loss": 2.2002,
      "step": 128037
    },
    {
      "epoch": 1.67,
      "grad_norm": 0.9747058749198914,
      "learning_rate": 0.0002477591872687156,
      "loss": 2.1181,
      "step": 128038
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.0258193016052246,
      "learning_rate": 0.0002477551592086261,
      "loss": 2.0078,
      "step": 128039
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.0425735712051392,
      "learning_rate": 0.0002477511311582499,
      "loss": 2.0328,
      "step": 128040
    },
    {
      "epoch": 1.67,
      "grad_norm": 0.9611179232597351,
      "learning_rate": 0.0002477471031175877,
      "loss": 2.0997,
      "step": 128041
    },
    {
      "epoch": 1.67,
      "grad_norm": 0.941299319267273,
      "learning_rate": 0.00024774307508664044,
      "loss": 2.2309,
      "step": 128042
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.1326006650924683,
      "learning_rate": 0.0002477390470654086,
      "loss": 2.0965,
      "step": 128043
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.0275685787200928,
      "learning_rate": 0.0002477350190538931,
      "loss": 2.2027,
      "step": 128044
    },
    {
      "epoch": 1.67,
      "grad_norm": 0.9157658815383911,
      "learning_rate": 0.0002477309910520947,
      "loss": 2.2202,
      "step": 128045
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.3039162158966064,
      "learning_rate": 0.00024772696306001404,
      "loss": 2.1636,
      "step": 128046
    },
    {
      "epoch": 1.67,
      "grad_norm": 0.9888637065887451,
      "learning_rate": 0.00024772293507765194,
      "loss": 2.2841,
      "step": 128047
    },
    {
      "epoch": 1.67,
      "grad_norm": 0.9448200464248657,
      "learning_rate": 0.00024771890710500907,
      "loss": 2.2493,
      "step": 128048
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.0988390445709229,
      "learning_rate": 0.00024771487914208634,
      "loss": 2.085,
      "step": 128049
    },
    {
      "epoch": 1.67,
      "grad_norm": 0.9450084567070007,
      "learning_rate": 0.00024771085118888435,
      "loss": 2.2235,
      "step": 128050
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.0479322671890259,
      "learning_rate": 0.0002477068232454039,
      "loss": 2.0982,
      "step": 128051
    },
    {
      "epoch": 1.67,
      "grad_norm": 0.9788772463798523,
      "learning_rate": 0.00024770279531164573,
      "loss": 2.0591,
      "step": 128052
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.037190556526184,
      "learning_rate": 0.0002476987673876107,
      "loss": 2.0462,
      "step": 128053
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.0519607067108154,
      "learning_rate": 0.0002476947394732993,
      "loss": 1.9231,
      "step": 128054
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.146141767501831,
      "learning_rate": 0.00024769071156871254,
      "loss": 2.1773,
      "step": 128055
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.051143765449524,
      "learning_rate": 0.00024768668367385104,
      "loss": 2.3448,
      "step": 128056
    },
    {
      "epoch": 1.67,
      "grad_norm": 0.9745684266090393,
      "learning_rate": 0.00024768265578871553,
      "loss": 2.0057,
      "step": 128057
    },
    {
      "epoch": 1.67,
      "grad_norm": 0.9810526967048645,
      "learning_rate": 0.00024767862791330684,
      "loss": 2.0974,
      "step": 128058
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.144936203956604,
      "learning_rate": 0.0002476746000476256,
      "loss": 2.1324,
      "step": 128059
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.2320759296417236,
      "learning_rate": 0.0002476705721916728,
      "loss": 1.7916,
      "step": 128060
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.0121970176696777,
      "learning_rate": 0.000247666544345449,
      "loss": 2.0628,
      "step": 128061
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.2220008373260498,
      "learning_rate": 0.0002476625165089549,
      "loss": 2.2647,
      "step": 128062
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.1093199253082275,
      "learning_rate": 0.00024765848868219133,
      "loss": 1.9715,
      "step": 128063
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.0448051691055298,
      "learning_rate": 0.000247654460865159,
      "loss": 1.9151,
      "step": 128064
    },
    {
      "epoch": 1.67,
      "grad_norm": 0.9158377051353455,
      "learning_rate": 0.00024765043305785875,
      "loss": 1.9997,
      "step": 128065
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.2459813356399536,
      "learning_rate": 0.0002476464052602912,
      "loss": 2.3568,
      "step": 128066
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.1146020889282227,
      "learning_rate": 0.0002476423774724574,
      "loss": 2.2129,
      "step": 128067
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.0691602230072021,
      "learning_rate": 0.00024763834969435773,
      "loss": 2.1015,
      "step": 128068
    },
    {
      "epoch": 1.67,
      "grad_norm": 0.9597960710525513,
      "learning_rate": 0.000247634321925993,
      "loss": 2.3534,
      "step": 128069
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.0036718845367432,
      "learning_rate": 0.00024763029416736406,
      "loss": 2.1898,
      "step": 128070
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.0220483541488647,
      "learning_rate": 0.0002476262664184717,
      "loss": 1.8427,
      "step": 128071
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.2698603868484497,
      "learning_rate": 0.00024762223867931657,
      "loss": 1.9742,
      "step": 128072
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.0254794359207153,
      "learning_rate": 0.00024761821094989954,
      "loss": 2.0153,
      "step": 128073
    },
    {
      "epoch": 1.67,
      "grad_norm": 0.9861479997634888,
      "learning_rate": 0.00024761418323022115,
      "loss": 2.0847,
      "step": 128074
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.0869542360305786,
      "learning_rate": 0.00024761015552028236,
      "loss": 1.89,
      "step": 128075
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.0318787097930908,
      "learning_rate": 0.00024760612782008377,
      "loss": 1.8826,
      "step": 128076
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.045639991760254,
      "learning_rate": 0.0002476021001296262,
      "loss": 2.2621,
      "step": 128077
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.044896125793457,
      "learning_rate": 0.00024759807244891034,
      "loss": 2.1318,
      "step": 128078
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.1183757781982422,
      "learning_rate": 0.0002475940447779372,
      "loss": 2.0126,
      "step": 128079
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.214991569519043,
      "learning_rate": 0.00024759001711670715,
      "loss": 1.9653,
      "step": 128080
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.1204222440719604,
      "learning_rate": 0.00024758598946522106,
      "loss": 2.2956,
      "step": 128081
    },
    {
      "epoch": 1.67,
      "grad_norm": 0.9748441576957703,
      "learning_rate": 0.0002475819618234798,
      "loss": 1.9536,
      "step": 128082
    },
    {
      "epoch": 1.67,
      "grad_norm": 0.9528540968894958,
      "learning_rate": 0.00024757793419148395,
      "loss": 2.0484,
      "step": 128083
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.1753692626953125,
      "learning_rate": 0.00024757390656923446,
      "loss": 2.2136,
      "step": 128084
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.1237249374389648,
      "learning_rate": 0.00024756987895673205,
      "loss": 1.9937,
      "step": 128085
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.01125168800354,
      "learning_rate": 0.0002475658513539772,
      "loss": 1.9601,
      "step": 128086
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.2185674905776978,
      "learning_rate": 0.000247561823760971,
      "loss": 2.1641,
      "step": 128087
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.1501237154006958,
      "learning_rate": 0.0002475577961777139,
      "loss": 2.2124,
      "step": 128088
    },
    {
      "epoch": 1.67,
      "grad_norm": 0.944310188293457,
      "learning_rate": 0.0002475537686042069,
      "loss": 2.1132,
      "step": 128089
    },
    {
      "epoch": 1.67,
      "grad_norm": 0.9582296013832092,
      "learning_rate": 0.00024754974104045053,
      "loss": 2.225,
      "step": 128090
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.0093189477920532,
      "learning_rate": 0.00024754571348644587,
      "loss": 2.075,
      "step": 128091
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.0313929319381714,
      "learning_rate": 0.0002475416859421933,
      "loss": 2.206,
      "step": 128092
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.0294440984725952,
      "learning_rate": 0.0002475376584076937,
      "loss": 1.9883,
      "step": 128093
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.0446230173110962,
      "learning_rate": 0.00024753363088294786,
      "loss": 2.0211,
      "step": 128094
    },
    {
      "epoch": 1.67,
      "grad_norm": 0.9718555808067322,
      "learning_rate": 0.00024752960336795653,
      "loss": 2.2037,
      "step": 128095
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.0865504741668701,
      "learning_rate": 0.0002475255758627204,
      "loss": 2.2214,
      "step": 128096
    },
    {
      "epoch": 1.67,
      "grad_norm": 0.9258677959442139,
      "learning_rate": 0.0002475215483672404,
      "loss": 1.89,
      "step": 128097
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.0022304058074951,
      "learning_rate": 0.000247517520881517,
      "loss": 2.0346,
      "step": 128098
    },
    {
      "epoch": 1.67,
      "grad_norm": 0.926826536655426,
      "learning_rate": 0.0002475134934055511,
      "loss": 2.0138,
      "step": 128099
    },
    {
      "epoch": 1.67,
      "grad_norm": 0.8873429298400879,
      "learning_rate": 0.00024750946593934346,
      "loss": 2.2343,
      "step": 128100
    },
    {
      "epoch": 1.67,
      "grad_norm": 0.965214192867279,
      "learning_rate": 0.0002475054384828947,
      "loss": 2.1605,
      "step": 128101
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.0512268543243408,
      "learning_rate": 0.0002475014110362058,
      "loss": 2.2884,
      "step": 128102
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.0206921100616455,
      "learning_rate": 0.0002474973835992774,
      "loss": 1.9652,
      "step": 128103
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.0180978775024414,
      "learning_rate": 0.00024749335617211014,
      "loss": 2.4604,
      "step": 128104
    },
    {
      "epoch": 1.67,
      "grad_norm": 0.9644250869750977,
      "learning_rate": 0.00024748932875470487,
      "loss": 2.1777,
      "step": 128105
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.0424025058746338,
      "learning_rate": 0.0002474853013470623,
      "loss": 1.9735,
      "step": 128106
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.025011420249939,
      "learning_rate": 0.0002474812739491832,
      "loss": 2.135,
      "step": 128107
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.1327202320098877,
      "learning_rate": 0.00024747724656106834,
      "loss": 2.2954,
      "step": 128108
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.0818926095962524,
      "learning_rate": 0.0002474732191827186,
      "loss": 2.1904,
      "step": 128109
    },
    {
      "epoch": 1.67,
      "grad_norm": 0.9964305758476257,
      "learning_rate": 0.0002474691918141344,
      "loss": 1.9905,
      "step": 128110
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.088898777961731,
      "learning_rate": 0.0002474651644553167,
      "loss": 2.2999,
      "step": 128111
    },
    {
      "epoch": 1.67,
      "grad_norm": 0.9814985394477844,
      "learning_rate": 0.0002474611371062662,
      "loss": 2.1793,
      "step": 128112
    },
    {
      "epoch": 1.67,
      "grad_norm": 0.9825548529624939,
      "learning_rate": 0.00024745710976698367,
      "loss": 2.3186,
      "step": 128113
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.10733163356781,
      "learning_rate": 0.0002474530824374699,
      "loss": 2.2262,
      "step": 128114
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.0076897144317627,
      "learning_rate": 0.00024744905511772555,
      "loss": 2.2953,
      "step": 128115
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.145674705505371,
      "learning_rate": 0.0002474450278077516,
      "loss": 1.8522,
      "step": 128116
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.1381406784057617,
      "learning_rate": 0.00024744100050754837,
      "loss": 1.9887,
      "step": 128117
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.348840594291687,
      "learning_rate": 0.00024743697321711694,
      "loss": 2.1487,
      "step": 128118
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.0483099222183228,
      "learning_rate": 0.000247432945936458,
      "loss": 2.2235,
      "step": 128119
    },
    {
      "epoch": 1.67,
      "grad_norm": 0.9994997382164001,
      "learning_rate": 0.00024742891866557217,
      "loss": 2.2555,
      "step": 128120
    },
    {
      "epoch": 1.67,
      "grad_norm": 0.9272785782814026,
      "learning_rate": 0.00024742489140446033,
      "loss": 1.8376,
      "step": 128121
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.01461660861969,
      "learning_rate": 0.00024742086415312323,
      "loss": 2.1713,
      "step": 128122
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.089341402053833,
      "learning_rate": 0.0002474168369115616,
      "loss": 2.0355,
      "step": 128123
    },
    {
      "epoch": 1.67,
      "grad_norm": 0.9473250508308411,
      "learning_rate": 0.00024741280967977616,
      "loss": 1.9477,
      "step": 128124
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.1174317598342896,
      "learning_rate": 0.00024740878245776765,
      "loss": 1.8653,
      "step": 128125
    },
    {
      "epoch": 1.67,
      "grad_norm": 0.9700695872306824,
      "learning_rate": 0.0002474047552455368,
      "loss": 2.5352,
      "step": 128126
    },
    {
      "epoch": 1.67,
      "grad_norm": 0.978022575378418,
      "learning_rate": 0.0002474007280430845,
      "loss": 2.2225,
      "step": 128127
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.0689729452133179,
      "learning_rate": 0.00024739670085041133,
      "loss": 2.1428,
      "step": 128128
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.068874478340149,
      "learning_rate": 0.00024739267366751807,
      "loss": 2.005,
      "step": 128129
    },
    {
      "epoch": 1.67,
      "grad_norm": 0.981214702129364,
      "learning_rate": 0.0002473886464944056,
      "loss": 2.1159,
      "step": 128130
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.0642368793487549,
      "learning_rate": 0.00024738461933107456,
      "loss": 2.1641,
      "step": 128131
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.056373953819275,
      "learning_rate": 0.00024738059217752566,
      "loss": 1.9531,
      "step": 128132
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.0534309148788452,
      "learning_rate": 0.0002473765650337597,
      "loss": 2.2039,
      "step": 128133
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.0823391675949097,
      "learning_rate": 0.0002473725378997775,
      "loss": 2.3275,
      "step": 128134
    },
    {
      "epoch": 1.67,
      "grad_norm": 0.9534441232681274,
      "learning_rate": 0.0002473685107755797,
      "loss": 2.0357,
      "step": 128135
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.0552189350128174,
      "learning_rate": 0.000247364483661167,
      "loss": 2.1909,
      "step": 128136
    },
    {
      "epoch": 1.67,
      "grad_norm": 0.9054262638092041,
      "learning_rate": 0.00024736045655654037,
      "loss": 2.0974,
      "step": 128137
    },
    {
      "epoch": 1.67,
      "grad_norm": 0.9314796328544617,
      "learning_rate": 0.00024735642946170037,
      "loss": 2.2551,
      "step": 128138
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.0523244142532349,
      "learning_rate": 0.0002473524023766478,
      "loss": 2.3278,
      "step": 128139
    },
    {
      "epoch": 1.67,
      "grad_norm": 0.9951031804084778,
      "learning_rate": 0.00024734837530138344,
      "loss": 2.1571,
      "step": 128140
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.086832880973816,
      "learning_rate": 0.000247344348235908,
      "loss": 2.1007,
      "step": 128141
    },
    {
      "epoch": 1.67,
      "grad_norm": 0.9998260736465454,
      "learning_rate": 0.0002473403211802222,
      "loss": 2.3189,
      "step": 128142
    },
    {
      "epoch": 1.67,
      "grad_norm": 0.9952197074890137,
      "learning_rate": 0.00024733629413432683,
      "loss": 2.0875,
      "step": 128143
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.0833461284637451,
      "learning_rate": 0.0002473322670982227,
      "loss": 2.4793,
      "step": 128144
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.1941936016082764,
      "learning_rate": 0.00024732824007191043,
      "loss": 1.8596,
      "step": 128145
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.0111303329467773,
      "learning_rate": 0.0002473242130553909,
      "loss": 2.203,
      "step": 128146
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.0786019563674927,
      "learning_rate": 0.00024732018604866474,
      "loss": 2.3526,
      "step": 128147
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.0313289165496826,
      "learning_rate": 0.00024731615905173275,
      "loss": 2.1105,
      "step": 128148
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.0758858919143677,
      "learning_rate": 0.0002473121320645957,
      "loss": 2.0845,
      "step": 128149
    },
    {
      "epoch": 1.67,
      "grad_norm": 0.9549549221992493,
      "learning_rate": 0.0002473081050872543,
      "loss": 2.1791,
      "step": 128150
    },
    {
      "epoch": 1.67,
      "grad_norm": 0.8999546766281128,
      "learning_rate": 0.0002473040781197093,
      "loss": 2.1212,
      "step": 128151
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.2042979001998901,
      "learning_rate": 0.0002473000511619616,
      "loss": 2.3748,
      "step": 128152
    },
    {
      "epoch": 1.67,
      "grad_norm": 0.9500583410263062,
      "learning_rate": 0.0002472960242140117,
      "loss": 1.9609,
      "step": 128153
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.0956497192382812,
      "learning_rate": 0.0002472919972758605,
      "loss": 2.2926,
      "step": 128154
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.0076366662979126,
      "learning_rate": 0.0002472879703475086,
      "loss": 2.0612,
      "step": 128155
    },
    {
      "epoch": 1.67,
      "grad_norm": 0.962445080280304,
      "learning_rate": 0.00024728394342895697,
      "loss": 2.1348,
      "step": 128156
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.273999810218811,
      "learning_rate": 0.00024727991652020623,
      "loss": 2.1098,
      "step": 128157
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.0458112955093384,
      "learning_rate": 0.00024727588962125723,
      "loss": 2.0765,
      "step": 128158
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.0220723152160645,
      "learning_rate": 0.00024727186273211055,
      "loss": 1.9826,
      "step": 128159
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.0798702239990234,
      "learning_rate": 0.00024726783585276706,
      "loss": 2.1631,
      "step": 128160
    },
    {
      "epoch": 1.67,
      "grad_norm": 0.9272934198379517,
      "learning_rate": 0.0002472638089832274,
      "loss": 2.1432,
      "step": 128161
    },
    {
      "epoch": 1.67,
      "grad_norm": 0.9982419610023499,
      "learning_rate": 0.00024725978212349243,
      "loss": 2.0038,
      "step": 128162
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.092125654220581,
      "learning_rate": 0.00024725575527356287,
      "loss": 2.2739,
      "step": 128163
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.2363204956054688,
      "learning_rate": 0.0002472517284334396,
      "loss": 2.1057,
      "step": 128164
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.0519797801971436,
      "learning_rate": 0.00024724770160312306,
      "loss": 1.9673,
      "step": 128165
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.06261146068573,
      "learning_rate": 0.0002472436747826142,
      "loss": 2.0592,
      "step": 128166
    },
    {
      "epoch": 1.67,
      "grad_norm": 0.9780079126358032,
      "learning_rate": 0.00024723964797191366,
      "loss": 2.0612,
      "step": 128167
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.1273925304412842,
      "learning_rate": 0.00024723562117102233,
      "loss": 1.8488,
      "step": 128168
    },
    {
      "epoch": 1.67,
      "grad_norm": 0.9610540866851807,
      "learning_rate": 0.00024723159437994093,
      "loss": 2.2133,
      "step": 128169
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.0622972249984741,
      "learning_rate": 0.00024722756759867025,
      "loss": 2.3305,
      "step": 128170
    },
    {
      "epoch": 1.67,
      "grad_norm": 0.9957954287528992,
      "learning_rate": 0.00024722354082721084,
      "loss": 2.1651,
      "step": 128171
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.1010586023330688,
      "learning_rate": 0.0002472195140655636,
      "loss": 2.196,
      "step": 128172
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.0373499393463135,
      "learning_rate": 0.0002472154873137292,
      "loss": 2.2247,
      "step": 128173
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.0325673818588257,
      "learning_rate": 0.0002472114605717085,
      "loss": 2.1988,
      "step": 128174
    },
    {
      "epoch": 1.67,
      "grad_norm": 0.9902001619338989,
      "learning_rate": 0.0002472074338395021,
      "loss": 1.9789,
      "step": 128175
    },
    {
      "epoch": 1.67,
      "grad_norm": 0.9763707518577576,
      "learning_rate": 0.00024720340711711106,
      "loss": 2.0454,
      "step": 128176
    },
    {
      "epoch": 1.67,
      "grad_norm": 0.9112420082092285,
      "learning_rate": 0.0002471993804045357,
      "loss": 2.2891,
      "step": 128177
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.1840927600860596,
      "learning_rate": 0.000247195353701777,
      "loss": 2.2659,
      "step": 128178
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.002392292022705,
      "learning_rate": 0.00024719132700883566,
      "loss": 2.0023,
      "step": 128179
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.0251049995422363,
      "learning_rate": 0.0002471873003257125,
      "loss": 2.3322,
      "step": 128180
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.0983675718307495,
      "learning_rate": 0.00024718327365240813,
      "loss": 1.8398,
      "step": 128181
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.325749397277832,
      "learning_rate": 0.0002471792469889235,
      "loss": 2.2998,
      "step": 128182
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.0282905101776123,
      "learning_rate": 0.0002471752203352593,
      "loss": 2.2488,
      "step": 128183
    },
    {
      "epoch": 1.67,
      "grad_norm": 0.9174670577049255,
      "learning_rate": 0.0002471711936914161,
      "loss": 2.2886,
      "step": 128184
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.0587477684020996,
      "learning_rate": 0.0002471671670573948,
      "loss": 2.0862,
      "step": 128185
    },
    {
      "epoch": 1.67,
      "grad_norm": 0.980353832244873,
      "learning_rate": 0.0002471631404331961,
      "loss": 2.1378,
      "step": 128186
    },
    {
      "epoch": 1.67,
      "grad_norm": 0.9486862421035767,
      "learning_rate": 0.00024715911381882074,
      "loss": 2.1347,
      "step": 128187
    },
    {
      "epoch": 1.67,
      "grad_norm": 0.9730831980705261,
      "learning_rate": 0.00024715508721426954,
      "loss": 2.1861,
      "step": 128188
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.13068425655365,
      "learning_rate": 0.0002471510606195433,
      "loss": 2.1402,
      "step": 128189
    },
    {
      "epoch": 1.67,
      "grad_norm": 0.9532846212387085,
      "learning_rate": 0.00024714703403464264,
      "loss": 2.0389,
      "step": 128190
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.0097213983535767,
      "learning_rate": 0.00024714300745956824,
      "loss": 2.1546,
      "step": 128191
    },
    {
      "epoch": 1.67,
      "grad_norm": 0.9880915284156799,
      "learning_rate": 0.000247138980894321,
      "loss": 1.9722,
      "step": 128192
    },
    {
      "epoch": 1.67,
      "grad_norm": 0.9358701705932617,
      "learning_rate": 0.0002471349543389016,
      "loss": 1.9514,
      "step": 128193
    },
    {
      "epoch": 1.67,
      "grad_norm": 0.943747878074646,
      "learning_rate": 0.0002471309277933108,
      "loss": 1.9544,
      "step": 128194
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.0487810373306274,
      "learning_rate": 0.0002471269012575495,
      "loss": 2.1652,
      "step": 128195
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.0508182048797607,
      "learning_rate": 0.0002471228747316182,
      "loss": 2.042,
      "step": 128196
    },
    {
      "epoch": 1.67,
      "grad_norm": 0.9397883415222168,
      "learning_rate": 0.00024711884821551774,
      "loss": 1.9493,
      "step": 128197
    },
    {
      "epoch": 1.67,
      "grad_norm": 0.9547534584999084,
      "learning_rate": 0.0002471148217092489,
      "loss": 2.1402,
      "step": 128198
    },
    {
      "epoch": 1.67,
      "grad_norm": 0.9219942092895508,
      "learning_rate": 0.0002471107952128124,
      "loss": 2.0748,
      "step": 128199
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.013065218925476,
      "learning_rate": 0.000247106768726209,
      "loss": 2.1246,
      "step": 128200
    },
    {
      "epoch": 1.67,
      "grad_norm": 0.9549044966697693,
      "learning_rate": 0.0002471027422494396,
      "loss": 2.064,
      "step": 128201
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.0424013137817383,
      "learning_rate": 0.0002470987157825046,
      "loss": 2.1042,
      "step": 128202
    },
    {
      "epoch": 1.67,
      "grad_norm": 0.951888918876648,
      "learning_rate": 0.00024709468932540503,
      "loss": 1.896,
      "step": 128203
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.0988346338272095,
      "learning_rate": 0.00024709066287814153,
      "loss": 2.2066,
      "step": 128204
    },
    {
      "epoch": 1.67,
      "grad_norm": 0.9480465054512024,
      "learning_rate": 0.00024708663644071484,
      "loss": 2.1028,
      "step": 128205
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.0715347528457642,
      "learning_rate": 0.0002470826100131258,
      "loss": 2.1951,
      "step": 128206
    },
    {
      "epoch": 1.67,
      "grad_norm": 0.9947040677070618,
      "learning_rate": 0.00024707858359537517,
      "loss": 1.924,
      "step": 128207
    },
    {
      "epoch": 1.67,
      "grad_norm": 0.9711084961891174,
      "learning_rate": 0.00024707455718746355,
      "loss": 2.0918,
      "step": 128208
    },
    {
      "epoch": 1.67,
      "grad_norm": 0.9947466254234314,
      "learning_rate": 0.0002470705307893918,
      "loss": 2.2393,
      "step": 128209
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.064099907875061,
      "learning_rate": 0.00024706650440116054,
      "loss": 2.0539,
      "step": 128210
    },
    {
      "epoch": 1.67,
      "grad_norm": 0.9466675519943237,
      "learning_rate": 0.00024706247802277066,
      "loss": 2.0286,
      "step": 128211
    },
    {
      "epoch": 1.67,
      "grad_norm": 0.9726155996322632,
      "learning_rate": 0.0002470584516542229,
      "loss": 2.164,
      "step": 128212
    },
    {
      "epoch": 1.67,
      "grad_norm": 0.9995304346084595,
      "learning_rate": 0.00024705442529551796,
      "loss": 2.1608,
      "step": 128213
    },
    {
      "epoch": 1.67,
      "grad_norm": 0.9773808717727661,
      "learning_rate": 0.00024705039894665665,
      "loss": 2.0711,
      "step": 128214
    },
    {
      "epoch": 1.67,
      "grad_norm": 0.937041699886322,
      "learning_rate": 0.00024704637260763963,
      "loss": 2.158,
      "step": 128215
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.0195945501327515,
      "learning_rate": 0.00024704234627846765,
      "loss": 2.2611,
      "step": 128216
    },
    {
      "epoch": 1.67,
      "grad_norm": 0.9683763384819031,
      "learning_rate": 0.0002470383199591415,
      "loss": 2.2763,
      "step": 128217
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.01019287109375,
      "learning_rate": 0.00024703429364966197,
      "loss": 2.1829,
      "step": 128218
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.0735434293746948,
      "learning_rate": 0.0002470302673500298,
      "loss": 2.1705,
      "step": 128219
    },
    {
      "epoch": 1.67,
      "grad_norm": 0.9327377080917358,
      "learning_rate": 0.0002470262410602456,
      "loss": 2.0031,
      "step": 128220
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.0144500732421875,
      "learning_rate": 0.0002470222147803103,
      "loss": 2.3756,
      "step": 128221
    },
    {
      "epoch": 1.67,
      "grad_norm": 0.9642617106437683,
      "learning_rate": 0.0002470181885102246,
      "loss": 1.8683,
      "step": 128222
    },
    {
      "epoch": 1.67,
      "grad_norm": 0.9205418229103088,
      "learning_rate": 0.0002470141622499891,
      "loss": 1.9824,
      "step": 128223
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.0262819528579712,
      "learning_rate": 0.00024701013599960473,
      "loss": 2.1696,
      "step": 128224
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.1050269603729248,
      "learning_rate": 0.00024700610975907225,
      "loss": 2.1406,
      "step": 128225
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.030755877494812,
      "learning_rate": 0.00024700208352839223,
      "loss": 2.0703,
      "step": 128226
    },
    {
      "epoch": 1.67,
      "grad_norm": 0.9840949177742004,
      "learning_rate": 0.00024699805730756555,
      "loss": 2.4654,
      "step": 128227
    },
    {
      "epoch": 1.67,
      "grad_norm": 0.900492787361145,
      "learning_rate": 0.0002469940310965929,
      "loss": 2.0219,
      "step": 128228
    },
    {
      "epoch": 1.67,
      "grad_norm": 0.9907957315444946,
      "learning_rate": 0.00024699000489547513,
      "loss": 2.055,
      "step": 128229
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.0077168941497803,
      "learning_rate": 0.0002469859787042129,
      "loss": 1.9638,
      "step": 128230
    },
    {
      "epoch": 1.67,
      "grad_norm": 0.9973687529563904,
      "learning_rate": 0.000246981952522807,
      "loss": 2.0409,
      "step": 128231
    },
    {
      "epoch": 1.67,
      "grad_norm": 0.959320068359375,
      "learning_rate": 0.00024697792635125813,
      "loss": 1.9892,
      "step": 128232
    },
    {
      "epoch": 1.67,
      "grad_norm": 0.9418274164199829,
      "learning_rate": 0.000246973900189567,
      "loss": 2.0589,
      "step": 128233
    },
    {
      "epoch": 1.67,
      "grad_norm": 0.9819052219390869,
      "learning_rate": 0.0002469698740377345,
      "loss": 2.1193,
      "step": 128234
    },
    {
      "epoch": 1.67,
      "grad_norm": 0.9774670600891113,
      "learning_rate": 0.00024696584789576123,
      "loss": 1.9422,
      "step": 128235
    },
    {
      "epoch": 1.67,
      "grad_norm": 0.918741762638092,
      "learning_rate": 0.0002469618217636481,
      "loss": 2.1408,
      "step": 128236
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.0630323886871338,
      "learning_rate": 0.00024695779564139577,
      "loss": 2.0633,
      "step": 128237
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.0077170133590698,
      "learning_rate": 0.0002469537695290049,
      "loss": 1.9691,
      "step": 128238
    },
    {
      "epoch": 1.67,
      "grad_norm": 0.901094377040863,
      "learning_rate": 0.00024694974342647635,
      "loss": 1.9396,
      "step": 128239
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.1114554405212402,
      "learning_rate": 0.0002469457173338109,
      "loss": 2.0206,
      "step": 128240
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.057559609413147,
      "learning_rate": 0.00024694169125100916,
      "loss": 2.1409,
      "step": 128241
    },
    {
      "epoch": 1.67,
      "grad_norm": 0.9908080101013184,
      "learning_rate": 0.000246937665178072,
      "loss": 1.8793,
      "step": 128242
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.1107126474380493,
      "learning_rate": 0.00024693363911500025,
      "loss": 2.2562,
      "step": 128243
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.0986372232437134,
      "learning_rate": 0.00024692961306179437,
      "loss": 2.2235,
      "step": 128244
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.1203985214233398,
      "learning_rate": 0.00024692558701845533,
      "loss": 2.3741,
      "step": 128245
    },
    {
      "epoch": 1.67,
      "grad_norm": 0.9777142405509949,
      "learning_rate": 0.0002469215609849838,
      "loss": 1.9838,
      "step": 128246
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.034957766532898,
      "learning_rate": 0.00024691753496138057,
      "loss": 2.2042,
      "step": 128247
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.0274146795272827,
      "learning_rate": 0.00024691350894764637,
      "loss": 2.2741,
      "step": 128248
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.0846664905548096,
      "learning_rate": 0.00024690948294378193,
      "loss": 2.0988,
      "step": 128249
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.003374457359314,
      "learning_rate": 0.0002469054569497882,
      "loss": 2.3346,
      "step": 128250
    },
    {
      "epoch": 1.67,
      "grad_norm": 0.9813017845153809,
      "learning_rate": 0.00024690143096566555,
      "loss": 2.1118,
      "step": 128251
    },
    {
      "epoch": 1.67,
      "grad_norm": 0.9998457431793213,
      "learning_rate": 0.00024689740499141496,
      "loss": 2.3164,
      "step": 128252
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.1397517919540405,
      "learning_rate": 0.0002468933790270371,
      "loss": 2.1364,
      "step": 128253
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.0190504789352417,
      "learning_rate": 0.0002468893530725328,
      "loss": 2.3585,
      "step": 128254
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.0838377475738525,
      "learning_rate": 0.00024688532712790286,
      "loss": 2.2016,
      "step": 128255
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.17432701587677,
      "learning_rate": 0.00024688130119314796,
      "loss": 2.0278,
      "step": 128256
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.0137773752212524,
      "learning_rate": 0.00024687727526826874,
      "loss": 1.9609,
      "step": 128257
    },
    {
      "epoch": 1.67,
      "grad_norm": 0.9885059595108032,
      "learning_rate": 0.000246873249353266,
      "loss": 1.9499,
      "step": 128258
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.052602767944336,
      "learning_rate": 0.00024686922344814056,
      "loss": 2.3365,
      "step": 128259
    },
    {
      "epoch": 1.67,
      "grad_norm": 0.9187896251678467,
      "learning_rate": 0.00024686519755289313,
      "loss": 1.9554,
      "step": 128260
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.0263714790344238,
      "learning_rate": 0.00024686117166752445,
      "loss": 1.8871,
      "step": 128261
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.0975021123886108,
      "learning_rate": 0.0002468571457920354,
      "loss": 2.2479,
      "step": 128262
    },
    {
      "epoch": 1.67,
      "grad_norm": 0.9579321146011353,
      "learning_rate": 0.0002468531199264265,
      "loss": 2.2568,
      "step": 128263
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.2140988111495972,
      "learning_rate": 0.00024684909407069865,
      "loss": 2.198,
      "step": 128264
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.0268468856811523,
      "learning_rate": 0.0002468450682248525,
      "loss": 2.2125,
      "step": 128265
    },
    {
      "epoch": 1.67,
      "grad_norm": 0.9423505663871765,
      "learning_rate": 0.0002468410423888888,
      "loss": 2.1229,
      "step": 128266
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.0050121545791626,
      "learning_rate": 0.0002468370165628085,
      "loss": 2.0615,
      "step": 128267
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.07364821434021,
      "learning_rate": 0.00024683299074661224,
      "loss": 2.1707,
      "step": 128268
    },
    {
      "epoch": 1.67,
      "grad_norm": 0.9504289031028748,
      "learning_rate": 0.0002468289649403006,
      "loss": 2.302,
      "step": 128269
    },
    {
      "epoch": 1.67,
      "grad_norm": 0.972981333732605,
      "learning_rate": 0.00024682493914387446,
      "loss": 1.9873,
      "step": 128270
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.0405011177062988,
      "learning_rate": 0.0002468209133573346,
      "loss": 2.1076,
      "step": 128271
    },
    {
      "epoch": 1.67,
      "grad_norm": 0.9696385860443115,
      "learning_rate": 0.0002468168875806817,
      "loss": 2.1407,
      "step": 128272
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.0245909690856934,
      "learning_rate": 0.00024681286181391657,
      "loss": 2.1313,
      "step": 128273
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.0542548894882202,
      "learning_rate": 0.0002468088360570401,
      "loss": 2.1228,
      "step": 128274
    },
    {
      "epoch": 1.67,
      "grad_norm": 0.9969446659088135,
      "learning_rate": 0.00024680481031005266,
      "loss": 2.1183,
      "step": 128275
    },
    {
      "epoch": 1.67,
      "grad_norm": 0.8929763436317444,
      "learning_rate": 0.00024680078457295525,
      "loss": 2.1667,
      "step": 128276
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.1093584299087524,
      "learning_rate": 0.0002467967588457486,
      "loss": 2.0991,
      "step": 128277
    },
    {
      "epoch": 1.67,
      "grad_norm": 0.9087000489234924,
      "learning_rate": 0.00024679273312843336,
      "loss": 1.9745,
      "step": 128278
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.006065011024475,
      "learning_rate": 0.00024678870742101045,
      "loss": 1.998,
      "step": 128279
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.0636928081512451,
      "learning_rate": 0.0002467846817234806,
      "loss": 2.0853,
      "step": 128280
    },
    {
      "epoch": 1.67,
      "grad_norm": 0.974894106388092,
      "learning_rate": 0.00024678065603584435,
      "loss": 2.194,
      "step": 128281
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.0897891521453857,
      "learning_rate": 0.0002467766303581026,
      "loss": 2.1908,
      "step": 128282
    },
    {
      "epoch": 1.67,
      "grad_norm": 0.9958805441856384,
      "learning_rate": 0.00024677260469025607,
      "loss": 2.0489,
      "step": 128283
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.0805412530899048,
      "learning_rate": 0.0002467685790323055,
      "loss": 2.1083,
      "step": 128284
    },
    {
      "epoch": 1.67,
      "grad_norm": 0.9873418211936951,
      "learning_rate": 0.0002467645533842517,
      "loss": 2.1595,
      "step": 128285
    },
    {
      "epoch": 1.67,
      "grad_norm": 0.9622355699539185,
      "learning_rate": 0.00024676052774609547,
      "loss": 1.9521,
      "step": 128286
    },
    {
      "epoch": 1.67,
      "grad_norm": 0.989960253238678,
      "learning_rate": 0.0002467565021178374,
      "loss": 2.1164,
      "step": 128287
    },
    {
      "epoch": 1.67,
      "grad_norm": 0.9513818025588989,
      "learning_rate": 0.0002467524764994782,
      "loss": 1.9076,
      "step": 128288
    },
    {
      "epoch": 1.67,
      "grad_norm": 0.9893245100975037,
      "learning_rate": 0.0002467484508910188,
      "loss": 2.2056,
      "step": 128289
    },
    {
      "epoch": 1.67,
      "grad_norm": 0.9573616981506348,
      "learning_rate": 0.0002467444252924598,
      "loss": 2.1072,
      "step": 128290
    },
    {
      "epoch": 1.67,
      "grad_norm": 0.981147825717926,
      "learning_rate": 0.000246740399703802,
      "loss": 2.0068,
      "step": 128291
    },
    {
      "epoch": 1.67,
      "grad_norm": 0.983561635017395,
      "learning_rate": 0.0002467363741250464,
      "loss": 1.9709,
      "step": 128292
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.0747385025024414,
      "learning_rate": 0.00024673234855619334,
      "loss": 2.3669,
      "step": 128293
    },
    {
      "epoch": 1.67,
      "grad_norm": 0.9061299562454224,
      "learning_rate": 0.00024672832299724374,
      "loss": 1.8706,
      "step": 128294
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.0547863245010376,
      "learning_rate": 0.0002467242974481983,
      "loss": 2.3339,
      "step": 128295
    },
    {
      "epoch": 1.67,
      "grad_norm": 0.9992433190345764,
      "learning_rate": 0.00024672027190905785,
      "loss": 2.1969,
      "step": 128296
    },
    {
      "epoch": 1.67,
      "grad_norm": 0.9405419826507568,
      "learning_rate": 0.00024671624637982313,
      "loss": 2.4838,
      "step": 128297
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.1419858932495117,
      "learning_rate": 0.00024671222086049494,
      "loss": 1.9849,
      "step": 128298
    },
    {
      "epoch": 1.67,
      "grad_norm": 0.9755851030349731,
      "learning_rate": 0.0002467081953510739,
      "loss": 2.0812,
      "step": 128299
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.0380998849868774,
      "learning_rate": 0.0002467041698515608,
      "loss": 2.0753,
      "step": 128300
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.0622273683547974,
      "learning_rate": 0.0002467001443619564,
      "loss": 2.1469,
      "step": 128301
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.0183809995651245,
      "learning_rate": 0.0002466961188822614,
      "loss": 2.2331,
      "step": 128302
    },
    {
      "epoch": 1.67,
      "grad_norm": 0.9502498507499695,
      "learning_rate": 0.0002466920934124766,
      "loss": 1.9534,
      "step": 128303
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.020662784576416,
      "learning_rate": 0.00024668806795260286,
      "loss": 2.2755,
      "step": 128304
    },
    {
      "epoch": 1.67,
      "grad_norm": 0.9257737994194031,
      "learning_rate": 0.00024668404250264073,
      "loss": 2.0159,
      "step": 128305
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.2628583908081055,
      "learning_rate": 0.0002466800170625911,
      "loss": 2.1057,
      "step": 128306
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.1428375244140625,
      "learning_rate": 0.0002466759916324546,
      "loss": 2.2304,
      "step": 128307
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.0263354778289795,
      "learning_rate": 0.00024667196621223203,
      "loss": 2.2288,
      "step": 128308
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.085161805152893,
      "learning_rate": 0.00024666794080192416,
      "loss": 2.1818,
      "step": 128309
    },
    {
      "epoch": 1.67,
      "grad_norm": 0.9675776958465576,
      "learning_rate": 0.0002466639154015318,
      "loss": 2.2916,
      "step": 128310
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.0150066614151,
      "learning_rate": 0.00024665989001105556,
      "loss": 2.1311,
      "step": 128311
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.0939360857009888,
      "learning_rate": 0.00024665586463049624,
      "loss": 1.9906,
      "step": 128312
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.0879302024841309,
      "learning_rate": 0.0002466518392598546,
      "loss": 2.0491,
      "step": 128313
    },
    {
      "epoch": 1.67,
      "grad_norm": 0.9261873364448547,
      "learning_rate": 0.0002466478138991315,
      "loss": 2.1335,
      "step": 128314
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.3725676536560059,
      "learning_rate": 0.00024664378854832744,
      "loss": 2.1335,
      "step": 128315
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.30649733543396,
      "learning_rate": 0.0002466397632074433,
      "loss": 2.1185,
      "step": 128316
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.0959774255752563,
      "learning_rate": 0.00024663573787647995,
      "loss": 2.241,
      "step": 128317
    },
    {
      "epoch": 1.67,
      "grad_norm": 0.9813822507858276,
      "learning_rate": 0.00024663171255543797,
      "loss": 2.2978,
      "step": 128318
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.171768069267273,
      "learning_rate": 0.00024662768724431817,
      "loss": 1.9828,
      "step": 128319
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.165514588356018,
      "learning_rate": 0.0002466236619431212,
      "loss": 2.2098,
      "step": 128320
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.0375964641571045,
      "learning_rate": 0.00024661963665184805,
      "loss": 2.2036,
      "step": 128321
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.0462995767593384,
      "learning_rate": 0.00024661561137049925,
      "loss": 2.1739,
      "step": 128322
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.2857164144515991,
      "learning_rate": 0.00024661158609907567,
      "loss": 2.0787,
      "step": 128323
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.0253400802612305,
      "learning_rate": 0.0002466075608375779,
      "loss": 1.9815,
      "step": 128324
    },
    {
      "epoch": 1.67,
      "grad_norm": 0.8803577423095703,
      "learning_rate": 0.00024660353558600683,
      "loss": 2.3807,
      "step": 128325
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.0305029153823853,
      "learning_rate": 0.00024659951034436314,
      "loss": 2.165,
      "step": 128326
    },
    {
      "epoch": 1.67,
      "grad_norm": 0.9424393177032471,
      "learning_rate": 0.00024659548511264765,
      "loss": 1.8878,
      "step": 128327
    },
    {
      "epoch": 1.67,
      "grad_norm": 0.9884268641471863,
      "learning_rate": 0.0002465914598908611,
      "loss": 1.9758,
      "step": 128328
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.0249587297439575,
      "learning_rate": 0.00024658743467900426,
      "loss": 2.018,
      "step": 128329
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.1903302669525146,
      "learning_rate": 0.0002465834094770777,
      "loss": 1.9499,
      "step": 128330
    },
    {
      "epoch": 1.67,
      "grad_norm": 0.9361346960067749,
      "learning_rate": 0.0002465793842850823,
      "loss": 2.2301,
      "step": 128331
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.1292674541473389,
      "learning_rate": 0.0002465753591030188,
      "loss": 2.0899,
      "step": 128332
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.2039854526519775,
      "learning_rate": 0.00024657133393088794,
      "loss": 2.2843,
      "step": 128333
    },
    {
      "epoch": 1.67,
      "grad_norm": 0.9929201006889343,
      "learning_rate": 0.0002465673087686905,
      "loss": 2.1175,
      "step": 128334
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.0120911598205566,
      "learning_rate": 0.00024656328361642737,
      "loss": 2.1042,
      "step": 128335
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.1230610609054565,
      "learning_rate": 0.00024655925847409893,
      "loss": 2.0741,
      "step": 128336
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.4016929864883423,
      "learning_rate": 0.0002465552333417062,
      "loss": 1.986,
      "step": 128337
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.0904490947723389,
      "learning_rate": 0.0002465512082192498,
      "loss": 2.1497,
      "step": 128338
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.0917092561721802,
      "learning_rate": 0.0002465471831067306,
      "loss": 2.0349,
      "step": 128339
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.4282232522964478,
      "learning_rate": 0.00024654315800414926,
      "loss": 2.2685,
      "step": 128340
    },
    {
      "epoch": 1.67,
      "grad_norm": 0.9939609169960022,
      "learning_rate": 0.00024653913291150666,
      "loss": 2.217,
      "step": 128341
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.3655861616134644,
      "learning_rate": 0.0002465351078288034,
      "loss": 2.0414,
      "step": 128342
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.1437331438064575,
      "learning_rate": 0.00024653108275604017,
      "loss": 2.0597,
      "step": 128343
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.1375598907470703,
      "learning_rate": 0.00024652705769321784,
      "loss": 2.3881,
      "step": 128344
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.1113269329071045,
      "learning_rate": 0.00024652303264033717,
      "loss": 1.8666,
      "step": 128345
    },
    {
      "epoch": 1.67,
      "grad_norm": 0.9914587140083313,
      "learning_rate": 0.0002465190075973989,
      "loss": 2.1753,
      "step": 128346
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.0377745628356934,
      "learning_rate": 0.00024651498256440384,
      "loss": 2.4196,
      "step": 128347
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.0679701566696167,
      "learning_rate": 0.00024651095754135255,
      "loss": 1.874,
      "step": 128348
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.2465547323226929,
      "learning_rate": 0.0002465069325282459,
      "loss": 2.0235,
      "step": 128349
    },
    {
      "epoch": 1.67,
      "grad_norm": 0.9934961795806885,
      "learning_rate": 0.00024650290752508455,
      "loss": 2.035,
      "step": 128350
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.0370498895645142,
      "learning_rate": 0.00024649888253186936,
      "loss": 2.16,
      "step": 128351
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.0190367698669434,
      "learning_rate": 0.000246494857548601,
      "loss": 2.1955,
      "step": 128352
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.0247995853424072,
      "learning_rate": 0.00024649083257528045,
      "loss": 2.1855,
      "step": 128353
    },
    {
      "epoch": 1.67,
      "grad_norm": 0.9905549883842468,
      "learning_rate": 0.00024648680761190815,
      "loss": 2.3776,
      "step": 128354
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.0054794549942017,
      "learning_rate": 0.0002464827826584849,
      "loss": 2.1687,
      "step": 128355
    },
    {
      "epoch": 1.67,
      "grad_norm": 0.9986457824707031,
      "learning_rate": 0.0002464787577150115,
      "loss": 2.0189,
      "step": 128356
    },
    {
      "epoch": 1.67,
      "grad_norm": 0.9766221642494202,
      "learning_rate": 0.00024647473278148875,
      "loss": 2.0722,
      "step": 128357
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.0347213745117188,
      "learning_rate": 0.00024647070785791735,
      "loss": 2.013,
      "step": 128358
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.1846410036087036,
      "learning_rate": 0.0002464666829442982,
      "loss": 2.0858,
      "step": 128359
    },
    {
      "epoch": 1.67,
      "grad_norm": 0.9989600777626038,
      "learning_rate": 0.0002464626580406318,
      "loss": 2.0284,
      "step": 128360
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.1028560400009155,
      "learning_rate": 0.0002464586331469189,
      "loss": 2.0939,
      "step": 128361
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.129023790359497,
      "learning_rate": 0.00024645460826316044,
      "loss": 2.0803,
      "step": 128362
    },
    {
      "epoch": 1.67,
      "grad_norm": 0.9876590371131897,
      "learning_rate": 0.00024645058338935706,
      "loss": 2.1022,
      "step": 128363
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.1884374618530273,
      "learning_rate": 0.0002464465585255095,
      "loss": 2.1488,
      "step": 128364
    },
    {
      "epoch": 1.67,
      "grad_norm": 0.976100742816925,
      "learning_rate": 0.0002464425336716187,
      "loss": 2.137,
      "step": 128365
    },
    {
      "epoch": 1.67,
      "grad_norm": 0.9897341132164001,
      "learning_rate": 0.00024643850882768515,
      "loss": 1.9374,
      "step": 128366
    },
    {
      "epoch": 1.67,
      "grad_norm": 0.9617050886154175,
      "learning_rate": 0.0002464344839937096,
      "loss": 2.1507,
      "step": 128367
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.0045361518859863,
      "learning_rate": 0.0002464304591696929,
      "loss": 2.0278,
      "step": 128368
    },
    {
      "epoch": 1.67,
      "grad_norm": 0.9697942733764648,
      "learning_rate": 0.0002464264343556358,
      "loss": 2.0105,
      "step": 128369
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.136726975440979,
      "learning_rate": 0.0002464224095515391,
      "loss": 1.9798,
      "step": 128370
    },
    {
      "epoch": 1.67,
      "grad_norm": 0.9637442827224731,
      "learning_rate": 0.0002464183847574036,
      "loss": 1.8945,
      "step": 128371
    },
    {
      "epoch": 1.67,
      "grad_norm": 0.9611455202102661,
      "learning_rate": 0.00024641435997322975,
      "loss": 2.2191,
      "step": 128372
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.1336064338684082,
      "learning_rate": 0.0002464103351990185,
      "loss": 1.9934,
      "step": 128373
    },
    {
      "epoch": 1.67,
      "grad_norm": 0.9391902089118958,
      "learning_rate": 0.0002464063104347706,
      "loss": 2.0759,
      "step": 128374
    },
    {
      "epoch": 1.67,
      "grad_norm": 0.9339648485183716,
      "learning_rate": 0.00024640228568048676,
      "loss": 1.9664,
      "step": 128375
    },
    {
      "epoch": 1.67,
      "grad_norm": 0.9709765315055847,
      "learning_rate": 0.0002463982609361678,
      "loss": 2.0208,
      "step": 128376
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.0193252563476562,
      "learning_rate": 0.0002463942362018145,
      "loss": 2.0584,
      "step": 128377
    },
    {
      "epoch": 1.67,
      "grad_norm": 0.9101133346557617,
      "learning_rate": 0.00024639021147742734,
      "loss": 2.1507,
      "step": 128378
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.0782232284545898,
      "learning_rate": 0.0002463861867630073,
      "loss": 2.2087,
      "step": 128379
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.020896553993225,
      "learning_rate": 0.0002463821620585551,
      "loss": 2.0309,
      "step": 128380
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.1210975646972656,
      "learning_rate": 0.00024637813736407146,
      "loss": 1.9723,
      "step": 128381
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.0069588422775269,
      "learning_rate": 0.00024637411267955714,
      "loss": 2.3384,
      "step": 128382
    },
    {
      "epoch": 1.67,
      "grad_norm": 0.9278072714805603,
      "learning_rate": 0.0002463700880050129,
      "loss": 2.1781,
      "step": 128383
    },
    {
      "epoch": 1.67,
      "grad_norm": 0.9769336581230164,
      "learning_rate": 0.00024636606334043953,
      "loss": 2.123,
      "step": 128384
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.0928319692611694,
      "learning_rate": 0.00024636203868583765,
      "loss": 2.4609,
      "step": 128385
    },
    {
      "epoch": 1.67,
      "grad_norm": 0.9617059826850891,
      "learning_rate": 0.0002463580140412081,
      "loss": 2.1225,
      "step": 128386
    },
    {
      "epoch": 1.67,
      "grad_norm": 0.9734906554222107,
      "learning_rate": 0.0002463539894065516,
      "loss": 2.2287,
      "step": 128387
    },
    {
      "epoch": 1.67,
      "grad_norm": 0.9799346327781677,
      "learning_rate": 0.00024634996478186885,
      "loss": 2.1893,
      "step": 128388
    },
    {
      "epoch": 1.67,
      "grad_norm": 0.9939112067222595,
      "learning_rate": 0.0002463459401671607,
      "loss": 2.1564,
      "step": 128389
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.0279282331466675,
      "learning_rate": 0.00024634191556242787,
      "loss": 2.2153,
      "step": 128390
    },
    {
      "epoch": 1.67,
      "grad_norm": 0.9424563646316528,
      "learning_rate": 0.0002463378909676711,
      "loss": 1.9286,
      "step": 128391
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.0891903638839722,
      "learning_rate": 0.0002463338663828911,
      "loss": 2.1872,
      "step": 128392
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.1947598457336426,
      "learning_rate": 0.00024632984180808867,
      "loss": 2.1642,
      "step": 128393
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.0272597074508667,
      "learning_rate": 0.0002463258172432645,
      "loss": 1.9297,
      "step": 128394
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.0809476375579834,
      "learning_rate": 0.0002463217926884194,
      "loss": 2.1528,
      "step": 128395
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.1262840032577515,
      "learning_rate": 0.0002463177681435541,
      "loss": 1.9777,
      "step": 128396
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.0628888607025146,
      "learning_rate": 0.0002463137436086693,
      "loss": 2.0439,
      "step": 128397
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.0877166986465454,
      "learning_rate": 0.00024630971908376584,
      "loss": 2.0135,
      "step": 128398
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.041231393814087,
      "learning_rate": 0.0002463056945688444,
      "loss": 2.2941,
      "step": 128399
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.1914057731628418,
      "learning_rate": 0.0002463016700639057,
      "loss": 2.1171,
      "step": 128400
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.1604479551315308,
      "learning_rate": 0.0002462976455689506,
      "loss": 2.4162,
      "step": 128401
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.07749605178833,
      "learning_rate": 0.0002462936210839798,
      "loss": 2.0916,
      "step": 128402
    },
    {
      "epoch": 1.67,
      "grad_norm": 0.9959800839424133,
      "learning_rate": 0.00024628959660899394,
      "loss": 2.2557,
      "step": 128403
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.1128394603729248,
      "learning_rate": 0.00024628557214399386,
      "loss": 2.0771,
      "step": 128404
    },
    {
      "epoch": 1.67,
      "grad_norm": 0.9565653800964355,
      "learning_rate": 0.0002462815476889803,
      "loss": 2.1589,
      "step": 128405
    },
    {
      "epoch": 1.67,
      "grad_norm": 0.943725049495697,
      "learning_rate": 0.0002462775232439541,
      "loss": 2.1179,
      "step": 128406
    },
    {
      "epoch": 1.67,
      "grad_norm": 0.9840222597122192,
      "learning_rate": 0.0002462734988089159,
      "loss": 2.3186,
      "step": 128407
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.006202220916748,
      "learning_rate": 0.00024626947438386647,
      "loss": 2.023,
      "step": 128408
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.0292561054229736,
      "learning_rate": 0.00024626544996880653,
      "loss": 2.1752,
      "step": 128409
    },
    {
      "epoch": 1.67,
      "grad_norm": 0.9436054825782776,
      "learning_rate": 0.00024626142556373687,
      "loss": 2.145,
      "step": 128410
    },
    {
      "epoch": 1.67,
      "grad_norm": 0.9985558390617371,
      "learning_rate": 0.0002462574011686582,
      "loss": 2.1624,
      "step": 128411
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.0708297491073608,
      "learning_rate": 0.0002462533767835713,
      "loss": 2.0484,
      "step": 128412
    },
    {
      "epoch": 1.67,
      "grad_norm": 0.9597570300102234,
      "learning_rate": 0.00024624935240847695,
      "loss": 1.9092,
      "step": 128413
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.6548504829406738,
      "learning_rate": 0.0002462453280433759,
      "loss": 2.1846,
      "step": 128414
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.0355976819992065,
      "learning_rate": 0.0002462413036882688,
      "loss": 2.1916,
      "step": 128415
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.0147567987442017,
      "learning_rate": 0.0002462372793431564,
      "loss": 1.847,
      "step": 128416
    },
    {
      "epoch": 1.67,
      "grad_norm": 0.9838389754295349,
      "learning_rate": 0.00024623325500803953,
      "loss": 2.0873,
      "step": 128417
    },
    {
      "epoch": 1.67,
      "grad_norm": 0.9444048404693604,
      "learning_rate": 0.000246229230682919,
      "loss": 2.0334,
      "step": 128418
    },
    {
      "epoch": 1.67,
      "grad_norm": 0.9630057215690613,
      "learning_rate": 0.00024622520636779534,
      "loss": 2.1118,
      "step": 128419
    },
    {
      "epoch": 1.67,
      "grad_norm": 0.9787923097610474,
      "learning_rate": 0.0002462211820626696,
      "loss": 2.2312,
      "step": 128420
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.0675909519195557,
      "learning_rate": 0.00024621715776754227,
      "loss": 2.1884,
      "step": 128421
    },
    {
      "epoch": 1.67,
      "grad_norm": 0.9465250372886658,
      "learning_rate": 0.00024621313348241415,
      "loss": 2.0367,
      "step": 128422
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.136098861694336,
      "learning_rate": 0.000246209109207286,
      "loss": 2.0194,
      "step": 128423
    },
    {
      "epoch": 1.67,
      "grad_norm": 0.918133556842804,
      "learning_rate": 0.00024620508494215866,
      "loss": 2.0749,
      "step": 128424
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.0223561525344849,
      "learning_rate": 0.00024620106068703275,
      "loss": 2.1532,
      "step": 128425
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.117192268371582,
      "learning_rate": 0.00024619703644190925,
      "loss": 2.2237,
      "step": 128426
    },
    {
      "epoch": 1.67,
      "grad_norm": 0.9856365919113159,
      "learning_rate": 0.00024619301220678855,
      "loss": 2.0104,
      "step": 128427
    },
    {
      "epoch": 1.67,
      "grad_norm": 0.9655449390411377,
      "learning_rate": 0.0002461889879816716,
      "loss": 1.9923,
      "step": 128428
    },
    {
      "epoch": 1.67,
      "grad_norm": 0.9788721203804016,
      "learning_rate": 0.0002461849637665592,
      "loss": 2.1979,
      "step": 128429
    },
    {
      "epoch": 1.67,
      "grad_norm": 0.9089365601539612,
      "learning_rate": 0.000246180939561452,
      "loss": 2.0517,
      "step": 128430
    },
    {
      "epoch": 1.67,
      "grad_norm": 0.9512214064598083,
      "learning_rate": 0.0002461769153663507,
      "loss": 1.9497,
      "step": 128431
    },
    {
      "epoch": 1.67,
      "grad_norm": 0.898424506187439,
      "learning_rate": 0.0002461728911812564,
      "loss": 2.155,
      "step": 128432
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.1686571836471558,
      "learning_rate": 0.00024616886700616934,
      "loss": 2.2312,
      "step": 128433
    },
    {
      "epoch": 1.67,
      "grad_norm": 0.9967302083969116,
      "learning_rate": 0.00024616484284109054,
      "loss": 2.1038,
      "step": 128434
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.1656882762908936,
      "learning_rate": 0.0002461608186860207,
      "loss": 2.0462,
      "step": 128435
    },
    {
      "epoch": 1.67,
      "grad_norm": 0.9992982149124146,
      "learning_rate": 0.00024615679454096056,
      "loss": 2.126,
      "step": 128436
    },
    {
      "epoch": 1.67,
      "grad_norm": 0.963636040687561,
      "learning_rate": 0.0002461527704059109,
      "loss": 1.9211,
      "step": 128437
    },
    {
      "epoch": 1.67,
      "grad_norm": 0.9827312231063843,
      "learning_rate": 0.0002461487462808726,
      "loss": 2.2869,
      "step": 128438
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.0385277271270752,
      "learning_rate": 0.0002461447221658462,
      "loss": 2.3403,
      "step": 128439
    },
    {
      "epoch": 1.67,
      "grad_norm": 0.9672542810440063,
      "learning_rate": 0.0002461406980608324,
      "loss": 1.931,
      "step": 128440
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.0632721185684204,
      "learning_rate": 0.0002461366739658321,
      "loss": 2.2352,
      "step": 128441
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.343355655670166,
      "learning_rate": 0.00024613264988084606,
      "loss": 2.1658,
      "step": 128442
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.0031453371047974,
      "learning_rate": 0.00024612862580587487,
      "loss": 2.139,
      "step": 128443
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.0553187131881714,
      "learning_rate": 0.0002461246017409196,
      "loss": 2.1523,
      "step": 128444
    },
    {
      "epoch": 1.67,
      "grad_norm": 0.9546051025390625,
      "learning_rate": 0.00024612057768598066,
      "loss": 2.3082,
      "step": 128445
    },
    {
      "epoch": 1.67,
      "grad_norm": 0.9943366050720215,
      "learning_rate": 0.00024611655364105883,
      "loss": 1.9487,
      "step": 128446
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.0740243196487427,
      "learning_rate": 0.00024611252960615504,
      "loss": 2.095,
      "step": 128447
    },
    {
      "epoch": 1.67,
      "grad_norm": 0.9817284345626831,
      "learning_rate": 0.0002461085055812699,
      "loss": 2.1545,
      "step": 128448
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.0143375396728516,
      "learning_rate": 0.00024610448156640424,
      "loss": 1.9134,
      "step": 128449
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.0717463493347168,
      "learning_rate": 0.00024610045756155876,
      "loss": 2.0416,
      "step": 128450
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.0509012937545776,
      "learning_rate": 0.0002460964335667344,
      "loss": 2.1027,
      "step": 128451
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.0025476217269897,
      "learning_rate": 0.0002460924095819315,
      "loss": 2.3306,
      "step": 128452
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.1693998575210571,
      "learning_rate": 0.00024608838560715113,
      "loss": 2.3541,
      "step": 128453
    },
    {
      "epoch": 1.67,
      "grad_norm": 0.9969787001609802,
      "learning_rate": 0.0002460843616423939,
      "loss": 2.3831,
      "step": 128454
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.0752851963043213,
      "learning_rate": 0.00024608033768766063,
      "loss": 2.0035,
      "step": 128455
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.176184058189392,
      "learning_rate": 0.000246076313742952,
      "loss": 2.1555,
      "step": 128456
    },
    {
      "epoch": 1.67,
      "grad_norm": 0.9453240036964417,
      "learning_rate": 0.000246072289808269,
      "loss": 2.1383,
      "step": 128457
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.113625407218933,
      "learning_rate": 0.00024606826588361195,
      "loss": 2.14,
      "step": 128458
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.1043769121170044,
      "learning_rate": 0.0002460642419689819,
      "loss": 1.9173,
      "step": 128459
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.017946481704712,
      "learning_rate": 0.00024606021806437953,
      "loss": 1.9729,
      "step": 128460
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.1655009984970093,
      "learning_rate": 0.00024605619416980554,
      "loss": 1.9877,
      "step": 128461
    },
    {
      "epoch": 1.67,
      "grad_norm": 0.9645479917526245,
      "learning_rate": 0.00024605217028526074,
      "loss": 1.9516,
      "step": 128462
    },
    {
      "epoch": 1.67,
      "grad_norm": 0.9647996425628662,
      "learning_rate": 0.000246048146410746,
      "loss": 1.9448,
      "step": 128463
    },
    {
      "epoch": 1.67,
      "grad_norm": 0.9885024428367615,
      "learning_rate": 0.0002460441225462618,
      "loss": 1.9233,
      "step": 128464
    },
    {
      "epoch": 1.67,
      "grad_norm": 0.9555084109306335,
      "learning_rate": 0.000246040098691809,
      "loss": 2.3463,
      "step": 128465
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.0424041748046875,
      "learning_rate": 0.00024603607484738837,
      "loss": 2.3585,
      "step": 128466
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.0762720108032227,
      "learning_rate": 0.0002460320510130007,
      "loss": 2.1896,
      "step": 128467
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.0420347452163696,
      "learning_rate": 0.0002460280271886466,
      "loss": 2.1627,
      "step": 128468
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.0242630243301392,
      "learning_rate": 0.0002460240033743271,
      "loss": 2.156,
      "step": 128469
    },
    {
      "epoch": 1.67,
      "grad_norm": 0.9963563084602356,
      "learning_rate": 0.0002460199795700426,
      "loss": 2.2814,
      "step": 128470
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.0734649896621704,
      "learning_rate": 0.00024601595577579405,
      "loss": 2.1433,
      "step": 128471
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.100779414176941,
      "learning_rate": 0.00024601193199158214,
      "loss": 2.1454,
      "step": 128472
    },
    {
      "epoch": 1.67,
      "grad_norm": 0.9240124225616455,
      "learning_rate": 0.0002460079082174076,
      "loss": 2.2156,
      "step": 128473
    },
    {
      "epoch": 1.67,
      "grad_norm": 0.957166850566864,
      "learning_rate": 0.0002460038844532712,
      "loss": 2.1271,
      "step": 128474
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.1341532468795776,
      "learning_rate": 0.00024599986069917377,
      "loss": 1.9197,
      "step": 128475
    },
    {
      "epoch": 1.67,
      "grad_norm": 0.9505051970481873,
      "learning_rate": 0.000245995836955116,
      "loss": 1.9413,
      "step": 128476
    },
    {
      "epoch": 1.67,
      "grad_norm": 0.9885997176170349,
      "learning_rate": 0.00024599181322109857,
      "loss": 1.9968,
      "step": 128477
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.0023362636566162,
      "learning_rate": 0.0002459877894971223,
      "loss": 2.1104,
      "step": 128478
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.0098228454589844,
      "learning_rate": 0.00024598376578318786,
      "loss": 2.0863,
      "step": 128479
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.0568053722381592,
      "learning_rate": 0.0002459797420792961,
      "loss": 2.2346,
      "step": 128480
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.0267008543014526,
      "learning_rate": 0.00024597571838544776,
      "loss": 2.1763,
      "step": 128481
    },
    {
      "epoch": 1.67,
      "grad_norm": 0.9444447755813599,
      "learning_rate": 0.0002459716947016435,
      "loss": 2.186,
      "step": 128482
    },
    {
      "epoch": 1.67,
      "grad_norm": 0.9582884907722473,
      "learning_rate": 0.0002459676710278842,
      "loss": 2.0915,
      "step": 128483
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.082042932510376,
      "learning_rate": 0.00024596364736417044,
      "loss": 1.9273,
      "step": 128484
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.0039929151535034,
      "learning_rate": 0.0002459596237105031,
      "loss": 2.1328,
      "step": 128485
    },
    {
      "epoch": 1.67,
      "grad_norm": 0.9986509084701538,
      "learning_rate": 0.0002459556000668828,
      "loss": 1.9682,
      "step": 128486
    },
    {
      "epoch": 1.67,
      "grad_norm": 0.9661295413970947,
      "learning_rate": 0.00024595157643331054,
      "loss": 2.0743,
      "step": 128487
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.0180846452713013,
      "learning_rate": 0.0002459475528097868,
      "loss": 2.0751,
      "step": 128488
    },
    {
      "epoch": 1.67,
      "grad_norm": 0.9897419810295105,
      "learning_rate": 0.0002459435291963124,
      "loss": 2.1307,
      "step": 128489
    },
    {
      "epoch": 1.67,
      "grad_norm": 0.9510605931282043,
      "learning_rate": 0.0002459395055928882,
      "loss": 2.2241,
      "step": 128490
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.1568328142166138,
      "learning_rate": 0.0002459354819995148,
      "loss": 2.0231,
      "step": 128491
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.0245543718338013,
      "learning_rate": 0.00024593145841619304,
      "loss": 2.3104,
      "step": 128492
    },
    {
      "epoch": 1.67,
      "grad_norm": 0.9934746026992798,
      "learning_rate": 0.0002459274348429237,
      "loss": 2.1682,
      "step": 128493
    },
    {
      "epoch": 1.67,
      "grad_norm": 0.9322128891944885,
      "learning_rate": 0.0002459234112797074,
      "loss": 2.103,
      "step": 128494
    },
    {
      "epoch": 1.67,
      "grad_norm": 0.8861321806907654,
      "learning_rate": 0.0002459193877265449,
      "loss": 2.1321,
      "step": 128495
    },
    {
      "epoch": 1.67,
      "grad_norm": 0.9807082414627075,
      "learning_rate": 0.00024591536418343713,
      "loss": 2.2025,
      "step": 128496
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.038223385810852,
      "learning_rate": 0.00024591134065038466,
      "loss": 2.1134,
      "step": 128497
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.0463290214538574,
      "learning_rate": 0.00024590731712738833,
      "loss": 2.2975,
      "step": 128498
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.1203663349151611,
      "learning_rate": 0.00024590329361444884,
      "loss": 2.1991,
      "step": 128499
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.0132319927215576,
      "learning_rate": 0.0002458992701115669,
      "loss": 2.2613,
      "step": 128500
    },
    {
      "epoch": 1.67,
      "grad_norm": 0.9488058686256409,
      "learning_rate": 0.0002458952466187433,
      "loss": 1.9013,
      "step": 128501
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.0439435243606567,
      "learning_rate": 0.00024589122313597886,
      "loss": 2.0793,
      "step": 128502
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.0233362913131714,
      "learning_rate": 0.0002458871996632742,
      "loss": 2.1479,
      "step": 128503
    },
    {
      "epoch": 1.67,
      "grad_norm": 0.9326670169830322,
      "learning_rate": 0.00024588317620063017,
      "loss": 2.0752,
      "step": 128504
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.0001497268676758,
      "learning_rate": 0.00024587915274804756,
      "loss": 2.1261,
      "step": 128505
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.0575573444366455,
      "learning_rate": 0.00024587512930552695,
      "loss": 2.1598,
      "step": 128506
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.0096492767333984,
      "learning_rate": 0.00024587110587306915,
      "loss": 1.9687,
      "step": 128507
    },
    {
      "epoch": 1.67,
      "grad_norm": 0.9829898476600647,
      "learning_rate": 0.00024586708245067496,
      "loss": 2.3461,
      "step": 128508
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.0610089302062988,
      "learning_rate": 0.00024586305903834504,
      "loss": 2.0464,
      "step": 128509
    },
    {
      "epoch": 1.67,
      "grad_norm": 0.9277729988098145,
      "learning_rate": 0.00024585903563608026,
      "loss": 1.8971,
      "step": 128510
    },
    {
      "epoch": 1.67,
      "grad_norm": 0.9475437998771667,
      "learning_rate": 0.00024585501224388143,
      "loss": 2.1031,
      "step": 128511
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.109493613243103,
      "learning_rate": 0.00024585098886174904,
      "loss": 2.3006,
      "step": 128512
    },
    {
      "epoch": 1.67,
      "grad_norm": 0.9558185338973999,
      "learning_rate": 0.00024584696548968395,
      "loss": 1.9823,
      "step": 128513
    },
    {
      "epoch": 1.67,
      "grad_norm": 0.9826140999794006,
      "learning_rate": 0.000245842942127687,
      "loss": 2.4106,
      "step": 128514
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.0940898656845093,
      "learning_rate": 0.0002458389187757588,
      "loss": 2.0693,
      "step": 128515
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.02104914188385,
      "learning_rate": 0.00024583489543390016,
      "loss": 2.055,
      "step": 128516
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.0220998525619507,
      "learning_rate": 0.0002458308721021119,
      "loss": 2.2976,
      "step": 128517
    },
    {
      "epoch": 1.67,
      "grad_norm": 0.9746533036231995,
      "learning_rate": 0.00024582684878039475,
      "loss": 2.14,
      "step": 128518
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.2241936922073364,
      "learning_rate": 0.0002458228254687493,
      "loss": 2.0547,
      "step": 128519
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.0041240453720093,
      "learning_rate": 0.00024581880216717646,
      "loss": 2.2702,
      "step": 128520
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.013827919960022,
      "learning_rate": 0.0002458147788756769,
      "loss": 2.292,
      "step": 128521
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.0705828666687012,
      "learning_rate": 0.0002458107555942514,
      "loss": 2.1428,
      "step": 128522
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.074696660041809,
      "learning_rate": 0.0002458067323229007,
      "loss": 2.0793,
      "step": 128523
    },
    {
      "epoch": 1.67,
      "grad_norm": 0.9173550009727478,
      "learning_rate": 0.0002458027090616257,
      "loss": 2.1123,
      "step": 128524
    },
    {
      "epoch": 1.67,
      "grad_norm": 0.9405756592750549,
      "learning_rate": 0.0002457986858104269,
      "loss": 2.1756,
      "step": 128525
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.108770728111267,
      "learning_rate": 0.00024579466256930506,
      "loss": 2.2037,
      "step": 128526
    },
    {
      "epoch": 1.67,
      "grad_norm": 0.947772204875946,
      "learning_rate": 0.000245790639338261,
      "loss": 2.0208,
      "step": 128527
    },
    {
      "epoch": 1.67,
      "grad_norm": 0.9320525527000427,
      "learning_rate": 0.0002457866161172956,
      "loss": 2.0156,
      "step": 128528
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.0176678895950317,
      "learning_rate": 0.0002457825929064094,
      "loss": 2.0115,
      "step": 128529
    },
    {
      "epoch": 1.67,
      "grad_norm": 0.9812596440315247,
      "learning_rate": 0.0002457785697056034,
      "loss": 2.0498,
      "step": 128530
    },
    {
      "epoch": 1.67,
      "grad_norm": 0.8916817307472229,
      "learning_rate": 0.0002457745465148781,
      "loss": 2.0213,
      "step": 128531
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.1444538831710815,
      "learning_rate": 0.0002457705233342342,
      "loss": 2.2434,
      "step": 128532
    },
    {
      "epoch": 1.67,
      "grad_norm": 0.9978646039962769,
      "learning_rate": 0.0002457665001636727,
      "loss": 2.1104,
      "step": 128533
    },
    {
      "epoch": 1.67,
      "grad_norm": 0.9447405934333801,
      "learning_rate": 0.0002457624770031942,
      "loss": 2.2045,
      "step": 128534
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.0212005376815796,
      "learning_rate": 0.00024575845385279953,
      "loss": 2.3546,
      "step": 128535
    },
    {
      "epoch": 1.67,
      "grad_norm": 0.9855066537857056,
      "learning_rate": 0.0002457544307124895,
      "loss": 2.2074,
      "step": 128536
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.126733660697937,
      "learning_rate": 0.0002457504075822646,
      "loss": 2.124,
      "step": 128537
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.060767412185669,
      "learning_rate": 0.0002457463844621257,
      "loss": 2.1109,
      "step": 128538
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.1379820108413696,
      "learning_rate": 0.00024574236135207355,
      "loss": 2.0251,
      "step": 128539
    },
    {
      "epoch": 1.67,
      "grad_norm": 0.9893884658813477,
      "learning_rate": 0.00024573833825210895,
      "loss": 2.3477,
      "step": 128540
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.0028053522109985,
      "learning_rate": 0.0002457343151622326,
      "loss": 2.1761,
      "step": 128541
    },
    {
      "epoch": 1.67,
      "grad_norm": 0.9965389370918274,
      "learning_rate": 0.0002457302920824454,
      "loss": 2.1481,
      "step": 128542
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.00169038772583,
      "learning_rate": 0.0002457262690127479,
      "loss": 2.1022,
      "step": 128543
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.002421498298645,
      "learning_rate": 0.0002457222459531408,
      "loss": 2.0447,
      "step": 128544
    },
    {
      "epoch": 1.67,
      "grad_norm": 0.9173383712768555,
      "learning_rate": 0.00024571822290362505,
      "loss": 2.0187,
      "step": 128545
    },
    {
      "epoch": 1.67,
      "grad_norm": 0.9453832507133484,
      "learning_rate": 0.00024571419986420127,
      "loss": 2.171,
      "step": 128546
    },
    {
      "epoch": 1.67,
      "grad_norm": 0.9244228005409241,
      "learning_rate": 0.00024571017683487026,
      "loss": 2.2331,
      "step": 128547
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.0583083629608154,
      "learning_rate": 0.00024570615381563286,
      "loss": 2.1705,
      "step": 128548
    },
    {
      "epoch": 1.67,
      "grad_norm": 0.9587531089782715,
      "learning_rate": 0.0002457021308064896,
      "loss": 2.3114,
      "step": 128549
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.026972770690918,
      "learning_rate": 0.00024569810780744134,
      "loss": 2.1081,
      "step": 128550
    },
    {
      "epoch": 1.67,
      "grad_norm": 0.9680854678153992,
      "learning_rate": 0.00024569408481848884,
      "loss": 2.1393,
      "step": 128551
    },
    {
      "epoch": 1.67,
      "grad_norm": 0.9915143251419067,
      "learning_rate": 0.00024569006183963277,
      "loss": 2.1172,
      "step": 128552
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.0251433849334717,
      "learning_rate": 0.000245686038870874,
      "loss": 1.8825,
      "step": 128553
    },
    {
      "epoch": 1.67,
      "grad_norm": 0.9757551550865173,
      "learning_rate": 0.00024568201591221333,
      "loss": 2.21,
      "step": 128554
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.0369668006896973,
      "learning_rate": 0.00024567799296365127,
      "loss": 2.1815,
      "step": 128555
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.0057564973831177,
      "learning_rate": 0.0002456739700251887,
      "loss": 2.1701,
      "step": 128556
    },
    {
      "epoch": 1.67,
      "grad_norm": 0.9705771803855896,
      "learning_rate": 0.0002456699470968264,
      "loss": 2.0432,
      "step": 128557
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.067352056503296,
      "learning_rate": 0.00024566592417856506,
      "loss": 2.2307,
      "step": 128558
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.0350244045257568,
      "learning_rate": 0.0002456619012704054,
      "loss": 2.0435,
      "step": 128559
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.030362844467163,
      "learning_rate": 0.0002456578783723484,
      "loss": 2.1963,
      "step": 128560
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.1711714267730713,
      "learning_rate": 0.00024565385548439445,
      "loss": 2.1273,
      "step": 128561
    },
    {
      "epoch": 1.67,
      "grad_norm": 0.9224548935890198,
      "learning_rate": 0.00024564983260654455,
      "loss": 2.1125,
      "step": 128562
    },
    {
      "epoch": 1.67,
      "grad_norm": 0.9922811985015869,
      "learning_rate": 0.0002456458097387993,
      "loss": 2.1838,
      "step": 128563
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.0929557085037231,
      "learning_rate": 0.00024564178688115957,
      "loss": 2.1319,
      "step": 128564
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.071602702140808,
      "learning_rate": 0.00024563776403362607,
      "loss": 2.0577,
      "step": 128565
    },
    {
      "epoch": 1.67,
      "grad_norm": 0.9734020233154297,
      "learning_rate": 0.00024563374119619956,
      "loss": 2.0571,
      "step": 128566
    },
    {
      "epoch": 1.67,
      "grad_norm": 0.9629198312759399,
      "learning_rate": 0.0002456297183688807,
      "loss": 1.8097,
      "step": 128567
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.1212189197540283,
      "learning_rate": 0.00024562569555167037,
      "loss": 2.2987,
      "step": 128568
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.0017796754837036,
      "learning_rate": 0.00024562167274456916,
      "loss": 2.2047,
      "step": 128569
    },
    {
      "epoch": 1.67,
      "grad_norm": 0.8965434432029724,
      "learning_rate": 0.00024561764994757797,
      "loss": 1.9529,
      "step": 128570
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.012906789779663,
      "learning_rate": 0.00024561362716069745,
      "loss": 2.3026,
      "step": 128571
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.0108758211135864,
      "learning_rate": 0.0002456096043839284,
      "loss": 2.1132,
      "step": 128572
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.0356519222259521,
      "learning_rate": 0.0002456055816172716,
      "loss": 1.7233,
      "step": 128573
    },
    {
      "epoch": 1.67,
      "grad_norm": 0.9800735116004944,
      "learning_rate": 0.0002456015588607277,
      "loss": 2.1169,
      "step": 128574
    },
    {
      "epoch": 1.67,
      "grad_norm": 0.9534600377082825,
      "learning_rate": 0.0002455975361142975,
      "loss": 2.0845,
      "step": 128575
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.0863806009292603,
      "learning_rate": 0.00024559351337798173,
      "loss": 2.0431,
      "step": 128576
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.003118634223938,
      "learning_rate": 0.0002455894906517812,
      "loss": 2.0976,
      "step": 128577
    },
    {
      "epoch": 1.67,
      "grad_norm": 0.9720860123634338,
      "learning_rate": 0.0002455854679356966,
      "loss": 2.1853,
      "step": 128578
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.0522289276123047,
      "learning_rate": 0.00024558144522972866,
      "loss": 2.1759,
      "step": 128579
    },
    {
      "epoch": 1.67,
      "grad_norm": 0.9269832372665405,
      "learning_rate": 0.00024557742253387814,
      "loss": 2.1963,
      "step": 128580
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.082422137260437,
      "learning_rate": 0.00024557339984814577,
      "loss": 2.2485,
      "step": 128581
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.0156259536743164,
      "learning_rate": 0.00024556937717253247,
      "loss": 2.1902,
      "step": 128582
    },
    {
      "epoch": 1.67,
      "grad_norm": 0.8635291457176208,
      "learning_rate": 0.00024556535450703874,
      "loss": 1.939,
      "step": 128583
    },
    {
      "epoch": 1.67,
      "grad_norm": 0.9953173995018005,
      "learning_rate": 0.00024556133185166544,
      "loss": 2.2223,
      "step": 128584
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.1389164924621582,
      "learning_rate": 0.0002455573092064134,
      "loss": 2.2258,
      "step": 128585
    },
    {
      "epoch": 1.67,
      "grad_norm": 0.9899349808692932,
      "learning_rate": 0.0002455532865712832,
      "loss": 2.107,
      "step": 128586
    },
    {
      "epoch": 1.67,
      "grad_norm": 0.9104225039482117,
      "learning_rate": 0.0002455492639462757,
      "loss": 2.1781,
      "step": 128587
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.0878770351409912,
      "learning_rate": 0.0002455452413313916,
      "loss": 2.0037,
      "step": 128588
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.1100131273269653,
      "learning_rate": 0.00024554121872663175,
      "loss": 2.1156,
      "step": 128589
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.007271647453308,
      "learning_rate": 0.00024553719613199676,
      "loss": 2.1894,
      "step": 128590
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.0018740892410278,
      "learning_rate": 0.00024553317354748745,
      "loss": 1.9889,
      "step": 128591
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.1457806825637817,
      "learning_rate": 0.0002455291509731045,
      "loss": 2.0715,
      "step": 128592
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.26493501663208,
      "learning_rate": 0.0002455251284088487,
      "loss": 2.1179,
      "step": 128593
    },
    {
      "epoch": 1.67,
      "grad_norm": 0.9738044738769531,
      "learning_rate": 0.00024552110585472083,
      "loss": 2.0093,
      "step": 128594
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.2000991106033325,
      "learning_rate": 0.0002455170833107216,
      "loss": 2.1108,
      "step": 128595
    },
    {
      "epoch": 1.67,
      "grad_norm": 0.9789829850196838,
      "learning_rate": 0.0002455130607768518,
      "loss": 2.0365,
      "step": 128596
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.0839736461639404,
      "learning_rate": 0.00024550903825311225,
      "loss": 2.1112,
      "step": 128597
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.0281811952590942,
      "learning_rate": 0.00024550501573950353,
      "loss": 1.9964,
      "step": 128598
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.0910135507583618,
      "learning_rate": 0.00024550099323602636,
      "loss": 2.3149,
      "step": 128599
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.3212432861328125,
      "learning_rate": 0.00024549697074268166,
      "loss": 2.6199,
      "step": 128600
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.1315397024154663,
      "learning_rate": 0.00024549294825947006,
      "loss": 2.1391,
      "step": 128601
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.058274269104004,
      "learning_rate": 0.0002454889257863924,
      "loss": 2.1011,
      "step": 128602
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.2090502977371216,
      "learning_rate": 0.00024548490332344943,
      "loss": 2.2754,
      "step": 128603
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.1140146255493164,
      "learning_rate": 0.0002454808808706418,
      "loss": 2.1252,
      "step": 128604
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.126234531402588,
      "learning_rate": 0.00024547685842797026,
      "loss": 2.2433,
      "step": 128605
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.0160391330718994,
      "learning_rate": 0.0002454728359954356,
      "loss": 2.285,
      "step": 128606
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.1012042760849,
      "learning_rate": 0.00024546881357303857,
      "loss": 2.1538,
      "step": 128607
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.447977900505066,
      "learning_rate": 0.0002454647911607799,
      "loss": 2.0689,
      "step": 128608
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.0881222486495972,
      "learning_rate": 0.00024546076875866054,
      "loss": 2.212,
      "step": 128609
    },
    {
      "epoch": 1.67,
      "grad_norm": 0.9620148539543152,
      "learning_rate": 0.0002454567463666809,
      "loss": 2.3019,
      "step": 128610
    },
    {
      "epoch": 1.67,
      "grad_norm": 0.990886390209198,
      "learning_rate": 0.0002454527239848418,
      "loss": 2.2558,
      "step": 128611
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.0179502964019775,
      "learning_rate": 0.00024544870161314415,
      "loss": 2.1796,
      "step": 128612
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.0925116539001465,
      "learning_rate": 0.0002454446792515886,
      "loss": 1.9355,
      "step": 128613
    },
    {
      "epoch": 1.67,
      "grad_norm": 0.9737208485603333,
      "learning_rate": 0.0002454406569001759,
      "loss": 1.9888,
      "step": 128614
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.0799003839492798,
      "learning_rate": 0.00024543663455890697,
      "loss": 1.9231,
      "step": 128615
    },
    {
      "epoch": 1.67,
      "grad_norm": 0.9931631088256836,
      "learning_rate": 0.00024543261222778227,
      "loss": 2.1929,
      "step": 128616
    },
    {
      "epoch": 1.67,
      "grad_norm": 0.9864875078201294,
      "learning_rate": 0.00024542858990680267,
      "loss": 2.1506,
      "step": 128617
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.0145844221115112,
      "learning_rate": 0.0002454245675959689,
      "loss": 2.0987,
      "step": 128618
    },
    {
      "epoch": 1.67,
      "grad_norm": 0.9426049590110779,
      "learning_rate": 0.00024542054529528174,
      "loss": 1.9345,
      "step": 128619
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.087153673171997,
      "learning_rate": 0.00024541652300474193,
      "loss": 2.103,
      "step": 128620
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.1292091608047485,
      "learning_rate": 0.0002454125007243504,
      "loss": 2.052,
      "step": 128621
    },
    {
      "epoch": 1.67,
      "grad_norm": 0.986131489276886,
      "learning_rate": 0.00024540847845410757,
      "loss": 2.1181,
      "step": 128622
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.0856090784072876,
      "learning_rate": 0.00024540445619401427,
      "loss": 2.0708,
      "step": 128623
    },
    {
      "epoch": 1.67,
      "grad_norm": 0.9587100148200989,
      "learning_rate": 0.0002454004339440714,
      "loss": 2.0802,
      "step": 128624
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.0248380899429321,
      "learning_rate": 0.00024539641170427955,
      "loss": 2.2616,
      "step": 128625
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.384507417678833,
      "learning_rate": 0.00024539238947463954,
      "loss": 2.3303,
      "step": 128626
    },
    {
      "epoch": 1.67,
      "grad_norm": 0.9733497500419617,
      "learning_rate": 0.00024538836725515226,
      "loss": 1.9497,
      "step": 128627
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.0278692245483398,
      "learning_rate": 0.0002453843450458182,
      "loss": 2.2418,
      "step": 128628
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.1630103588104248,
      "learning_rate": 0.0002453803228466382,
      "loss": 1.8868,
      "step": 128629
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.1614582538604736,
      "learning_rate": 0.00024537630065761307,
      "loss": 1.8728,
      "step": 128630
    },
    {
      "epoch": 1.67,
      "grad_norm": 0.9891358613967896,
      "learning_rate": 0.0002453722784787435,
      "loss": 2.1493,
      "step": 128631
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.0476540327072144,
      "learning_rate": 0.0002453682563100302,
      "loss": 2.0963,
      "step": 128632
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.0193067789077759,
      "learning_rate": 0.0002453642341514742,
      "loss": 2.0321,
      "step": 128633
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.3139336109161377,
      "learning_rate": 0.0002453602120030758,
      "loss": 2.0673,
      "step": 128634
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.0179728269577026,
      "learning_rate": 0.00024535618986483605,
      "loss": 2.1735,
      "step": 128635
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.0636998414993286,
      "learning_rate": 0.00024535216773675553,
      "loss": 2.0276,
      "step": 128636
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.0753488540649414,
      "learning_rate": 0.00024534814561883513,
      "loss": 2.1945,
      "step": 128637
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.0590885877609253,
      "learning_rate": 0.0002453441235110755,
      "loss": 1.9847,
      "step": 128638
    },
    {
      "epoch": 1.67,
      "grad_norm": 0.9771683216094971,
      "learning_rate": 0.0002453401014134776,
      "loss": 2.3416,
      "step": 128639
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.064592957496643,
      "learning_rate": 0.00024533607932604185,
      "loss": 2.2271,
      "step": 128640
    },
    {
      "epoch": 1.67,
      "grad_norm": 1.0610733032226562,
      "learning_rate": 0.00024533205724876916,
      "loss": 2.1844,
      "step": 128641
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.0628364086151123,
      "learning_rate": 0.00024532803518166027,
      "loss": 2.2118,
      "step": 128642
    },
    {
      "epoch": 1.68,
      "grad_norm": 0.9497596025466919,
      "learning_rate": 0.00024532401312471594,
      "loss": 2.197,
      "step": 128643
    },
    {
      "epoch": 1.68,
      "grad_norm": 0.9843348264694214,
      "learning_rate": 0.0002453199910779369,
      "loss": 2.1529,
      "step": 128644
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.0707145929336548,
      "learning_rate": 0.00024531596904132404,
      "loss": 2.0497,
      "step": 128645
    },
    {
      "epoch": 1.68,
      "grad_norm": 0.9344959259033203,
      "learning_rate": 0.0002453119470148779,
      "loss": 2.0309,
      "step": 128646
    },
    {
      "epoch": 1.68,
      "grad_norm": 0.9681717157363892,
      "learning_rate": 0.0002453079249985992,
      "loss": 1.9182,
      "step": 128647
    },
    {
      "epoch": 1.68,
      "grad_norm": 0.9169644713401794,
      "learning_rate": 0.0002453039029924888,
      "loss": 1.8576,
      "step": 128648
    },
    {
      "epoch": 1.68,
      "grad_norm": 0.9797243475914001,
      "learning_rate": 0.0002452998809965475,
      "loss": 2.0993,
      "step": 128649
    },
    {
      "epoch": 1.68,
      "grad_norm": 0.9381810426712036,
      "learning_rate": 0.000245295859010776,
      "loss": 2.0329,
      "step": 128650
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.0213669538497925,
      "learning_rate": 0.00024529183703517497,
      "loss": 2.3386,
      "step": 128651
    },
    {
      "epoch": 1.68,
      "grad_norm": 0.979097843170166,
      "learning_rate": 0.0002452878150697454,
      "loss": 2.3183,
      "step": 128652
    },
    {
      "epoch": 1.68,
      "grad_norm": 0.9502154588699341,
      "learning_rate": 0.00024528379311448766,
      "loss": 2.1286,
      "step": 128653
    },
    {
      "epoch": 1.68,
      "grad_norm": 0.9255488514900208,
      "learning_rate": 0.00024527977116940274,
      "loss": 2.2868,
      "step": 128654
    },
    {
      "epoch": 1.68,
      "grad_norm": 0.9330987334251404,
      "learning_rate": 0.00024527574923449133,
      "loss": 1.944,
      "step": 128655
    },
    {
      "epoch": 1.68,
      "grad_norm": 0.9622242450714111,
      "learning_rate": 0.00024527172730975423,
      "loss": 2.2042,
      "step": 128656
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.1647363901138306,
      "learning_rate": 0.0002452677053951921,
      "loss": 2.3724,
      "step": 128657
    },
    {
      "epoch": 1.68,
      "grad_norm": 0.9351016283035278,
      "learning_rate": 0.0002452636834908058,
      "loss": 2.0322,
      "step": 128658
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.1457353830337524,
      "learning_rate": 0.000245259661596596,
      "loss": 2.0163,
      "step": 128659
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.0684565305709839,
      "learning_rate": 0.00024525563971256346,
      "loss": 2.0674,
      "step": 128660
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.4343806505203247,
      "learning_rate": 0.0002452516178387089,
      "loss": 2.339,
      "step": 128661
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.100049376487732,
      "learning_rate": 0.00024524759597503314,
      "loss": 2.0802,
      "step": 128662
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.176275372505188,
      "learning_rate": 0.0002452435741215368,
      "loss": 2.0884,
      "step": 128663
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.0872975587844849,
      "learning_rate": 0.00024523955227822084,
      "loss": 2.061,
      "step": 128664
    },
    {
      "epoch": 1.68,
      "grad_norm": 0.9862750768661499,
      "learning_rate": 0.0002452355304450858,
      "loss": 2.1728,
      "step": 128665
    },
    {
      "epoch": 1.68,
      "grad_norm": 0.9454951286315918,
      "learning_rate": 0.0002452315086221325,
      "loss": 2.077,
      "step": 128666
    },
    {
      "epoch": 1.68,
      "grad_norm": 0.985321044921875,
      "learning_rate": 0.00024522748680936174,
      "loss": 2.2224,
      "step": 128667
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.0174555778503418,
      "learning_rate": 0.0002452234650067742,
      "loss": 2.0936,
      "step": 128668
    },
    {
      "epoch": 1.68,
      "grad_norm": 0.9724686145782471,
      "learning_rate": 0.00024521944321437067,
      "loss": 2.1364,
      "step": 128669
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.004009485244751,
      "learning_rate": 0.0002452154214321519,
      "loss": 2.1732,
      "step": 128670
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.0047380924224854,
      "learning_rate": 0.00024521139966011854,
      "loss": 2.1618,
      "step": 128671
    },
    {
      "epoch": 1.68,
      "grad_norm": 0.9725817441940308,
      "learning_rate": 0.0002452073778982715,
      "loss": 2.0257,
      "step": 128672
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.0229436159133911,
      "learning_rate": 0.0002452033561466113,
      "loss": 2.261,
      "step": 128673
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.0998691320419312,
      "learning_rate": 0.00024519933440513903,
      "loss": 2.1886,
      "step": 128674
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.0096274614334106,
      "learning_rate": 0.00024519531267385515,
      "loss": 2.2438,
      "step": 128675
    },
    {
      "epoch": 1.68,
      "grad_norm": 0.8843896389007568,
      "learning_rate": 0.0002451912909527605,
      "loss": 2.0393,
      "step": 128676
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.081941843032837,
      "learning_rate": 0.0002451872692418558,
      "loss": 2.0707,
      "step": 128677
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.2227667570114136,
      "learning_rate": 0.00024518324754114183,
      "loss": 2.1267,
      "step": 128678
    },
    {
      "epoch": 1.68,
      "grad_norm": 0.9709631204605103,
      "learning_rate": 0.0002451792258506193,
      "loss": 2.1473,
      "step": 128679
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.061008095741272,
      "learning_rate": 0.000245175204170289,
      "loss": 1.8875,
      "step": 128680
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.124994158744812,
      "learning_rate": 0.00024517118250015174,
      "loss": 2.129,
      "step": 128681
    },
    {
      "epoch": 1.68,
      "grad_norm": 0.9132469296455383,
      "learning_rate": 0.00024516716084020814,
      "loss": 2.0277,
      "step": 128682
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.0277537107467651,
      "learning_rate": 0.00024516313919045903,
      "loss": 2.0136,
      "step": 128683
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.0079776048660278,
      "learning_rate": 0.00024515911755090505,
      "loss": 2.1246,
      "step": 128684
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.0468095541000366,
      "learning_rate": 0.000245155095921547,
      "loss": 2.1584,
      "step": 128685
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.177819013595581,
      "learning_rate": 0.00024515107430238573,
      "loss": 2.1102,
      "step": 128686
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.0707794427871704,
      "learning_rate": 0.0002451470526934219,
      "loss": 2.0767,
      "step": 128687
    },
    {
      "epoch": 1.68,
      "grad_norm": 0.9943435192108154,
      "learning_rate": 0.0002451430310946564,
      "loss": 2.2951,
      "step": 128688
    },
    {
      "epoch": 1.68,
      "grad_norm": 0.9828929901123047,
      "learning_rate": 0.00024513900950608967,
      "loss": 2.1936,
      "step": 128689
    },
    {
      "epoch": 1.68,
      "grad_norm": 0.9626477956771851,
      "learning_rate": 0.00024513498792772264,
      "loss": 2.3056,
      "step": 128690
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.3039920330047607,
      "learning_rate": 0.00024513096635955605,
      "loss": 1.9661,
      "step": 128691
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.1240934133529663,
      "learning_rate": 0.0002451269448015907,
      "loss": 2.193,
      "step": 128692
    },
    {
      "epoch": 1.68,
      "grad_norm": 0.9897727370262146,
      "learning_rate": 0.00024512292325382725,
      "loss": 1.9547,
      "step": 128693
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.1476750373840332,
      "learning_rate": 0.00024511890171626666,
      "loss": 2.009,
      "step": 128694
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.269514799118042,
      "learning_rate": 0.00024511488018890933,
      "loss": 2.0498,
      "step": 128695
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.0058075189590454,
      "learning_rate": 0.00024511085867175614,
      "loss": 2.112,
      "step": 128696
    },
    {
      "epoch": 1.68,
      "grad_norm": 0.957811713218689,
      "learning_rate": 0.00024510683716480794,
      "loss": 2.1708,
      "step": 128697
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.0477759838104248,
      "learning_rate": 0.0002451028156680654,
      "loss": 2.1994,
      "step": 128698
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.0045069456100464,
      "learning_rate": 0.0002450987941815293,
      "loss": 2.1253,
      "step": 128699
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.1136630773544312,
      "learning_rate": 0.00024509477270520045,
      "loss": 2.1364,
      "step": 128700
    },
    {
      "epoch": 1.68,
      "grad_norm": 0.9579139351844788,
      "learning_rate": 0.0002450907512390794,
      "loss": 2.0425,
      "step": 128701
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.3850802183151245,
      "learning_rate": 0.00024508672978316705,
      "loss": 2.073,
      "step": 128702
    },
    {
      "epoch": 1.68,
      "grad_norm": 0.9661440253257751,
      "learning_rate": 0.0002450827083374641,
      "loss": 1.9325,
      "step": 128703
    },
    {
      "epoch": 1.68,
      "grad_norm": 0.9837428331375122,
      "learning_rate": 0.0002450786869019713,
      "loss": 1.9918,
      "step": 128704
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.0484181642532349,
      "learning_rate": 0.0002450746654766894,
      "loss": 2.1177,
      "step": 128705
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.0484910011291504,
      "learning_rate": 0.0002450706440616193,
      "loss": 1.987,
      "step": 128706
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.1173855066299438,
      "learning_rate": 0.0002450666226567615,
      "loss": 2.0708,
      "step": 128707
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.0192574262619019,
      "learning_rate": 0.00024506260126211686,
      "loss": 1.9926,
      "step": 128708
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.0696938037872314,
      "learning_rate": 0.00024505857987768605,
      "loss": 2.1142,
      "step": 128709
    },
    {
      "epoch": 1.68,
      "grad_norm": 0.9378665685653687,
      "learning_rate": 0.00024505455850346995,
      "loss": 2.1223,
      "step": 128710
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.0424563884735107,
      "learning_rate": 0.0002450505371394692,
      "loss": 2.082,
      "step": 128711
    },
    {
      "epoch": 1.68,
      "grad_norm": 0.9591741561889648,
      "learning_rate": 0.0002450465157856848,
      "loss": 1.9614,
      "step": 128712
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.0354371070861816,
      "learning_rate": 0.0002450424944421171,
      "loss": 1.958,
      "step": 128713
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.0182689428329468,
      "learning_rate": 0.000245038473108767,
      "loss": 2.1987,
      "step": 128714
    },
    {
      "epoch": 1.68,
      "grad_norm": 0.9607247710227966,
      "learning_rate": 0.0002450344517856353,
      "loss": 2.0137,
      "step": 128715
    },
    {
      "epoch": 1.68,
      "grad_norm": 0.8467181921005249,
      "learning_rate": 0.00024503043047272277,
      "loss": 2.1534,
      "step": 128716
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.0653167963027954,
      "learning_rate": 0.00024502640917003013,
      "loss": 2.2681,
      "step": 128717
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.0164498090744019,
      "learning_rate": 0.0002450223878775581,
      "loss": 2.0543,
      "step": 128718
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.0520068407058716,
      "learning_rate": 0.00024501836659530755,
      "loss": 2.1179,
      "step": 128719
    },
    {
      "epoch": 1.68,
      "grad_norm": 0.9981280565261841,
      "learning_rate": 0.00024501434532327903,
      "loss": 2.0356,
      "step": 128720
    },
    {
      "epoch": 1.68,
      "grad_norm": 0.9485867023468018,
      "learning_rate": 0.0002450103240614734,
      "loss": 2.0472,
      "step": 128721
    },
    {
      "epoch": 1.68,
      "grad_norm": 0.978986918926239,
      "learning_rate": 0.0002450063028098913,
      "loss": 2.0945,
      "step": 128722
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.0871244668960571,
      "learning_rate": 0.00024500228156853365,
      "loss": 2.2668,
      "step": 128723
    },
    {
      "epoch": 1.68,
      "grad_norm": 0.9525002837181091,
      "learning_rate": 0.00024499826033740107,
      "loss": 2.0178,
      "step": 128724
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.2312588691711426,
      "learning_rate": 0.0002449942391164945,
      "loss": 1.979,
      "step": 128725
    },
    {
      "epoch": 1.68,
      "grad_norm": 0.8878083825111389,
      "learning_rate": 0.00024499021790581443,
      "loss": 2.2147,
      "step": 128726
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.0284574031829834,
      "learning_rate": 0.0002449861967053617,
      "loss": 2.1314,
      "step": 128727
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.1820043325424194,
      "learning_rate": 0.00024498217551513707,
      "loss": 2.2593,
      "step": 128728
    },
    {
      "epoch": 1.68,
      "grad_norm": 0.9724704623222351,
      "learning_rate": 0.0002449781543351413,
      "loss": 2.1893,
      "step": 128729
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.0094295740127563,
      "learning_rate": 0.0002449741331653751,
      "loss": 2.0172,
      "step": 128730
    },
    {
      "epoch": 1.68,
      "grad_norm": 0.8856184482574463,
      "learning_rate": 0.0002449701120058394,
      "loss": 2.132,
      "step": 128731
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.0234929323196411,
      "learning_rate": 0.00024496609085653464,
      "loss": 2.1752,
      "step": 128732
    },
    {
      "epoch": 1.68,
      "grad_norm": 0.9297789335250854,
      "learning_rate": 0.0002449620697174618,
      "loss": 2.1831,
      "step": 128733
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.0072386264801025,
      "learning_rate": 0.00024495804858862144,
      "loss": 1.902,
      "step": 128734
    },
    {
      "epoch": 1.68,
      "grad_norm": 0.9352362155914307,
      "learning_rate": 0.0002449540274700145,
      "loss": 2.1405,
      "step": 128735
    },
    {
      "epoch": 1.68,
      "grad_norm": 0.932400643825531,
      "learning_rate": 0.00024495000636164165,
      "loss": 2.0847,
      "step": 128736
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.1826196908950806,
      "learning_rate": 0.0002449459852635037,
      "loss": 2.0055,
      "step": 128737
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.0276767015457153,
      "learning_rate": 0.00024494196417560127,
      "loss": 1.9365,
      "step": 128738
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.0261212587356567,
      "learning_rate": 0.0002449379430979351,
      "loss": 1.9881,
      "step": 128739
    },
    {
      "epoch": 1.68,
      "grad_norm": 0.993855357170105,
      "learning_rate": 0.000244933922030506,
      "loss": 2.1206,
      "step": 128740
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.0259859561920166,
      "learning_rate": 0.0002449299009733148,
      "loss": 2.1845,
      "step": 128741
    },
    {
      "epoch": 1.68,
      "grad_norm": 0.9654358625411987,
      "learning_rate": 0.00024492587992636214,
      "loss": 2.1498,
      "step": 128742
    },
    {
      "epoch": 1.68,
      "grad_norm": 0.9468899369239807,
      "learning_rate": 0.00024492185888964885,
      "loss": 2.0758,
      "step": 128743
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.01210355758667,
      "learning_rate": 0.0002449178378631756,
      "loss": 2.112,
      "step": 128744
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.0090938806533813,
      "learning_rate": 0.00024491381684694314,
      "loss": 2.2319,
      "step": 128745
    },
    {
      "epoch": 1.68,
      "grad_norm": 0.9901727437973022,
      "learning_rate": 0.00024490979584095224,
      "loss": 2.2084,
      "step": 128746
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.0063279867172241,
      "learning_rate": 0.00024490577484520364,
      "loss": 2.1107,
      "step": 128747
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.0029600858688354,
      "learning_rate": 0.0002449017538596981,
      "loss": 2.1559,
      "step": 128748
    },
    {
      "epoch": 1.68,
      "grad_norm": 0.994697630405426,
      "learning_rate": 0.00024489773288443646,
      "loss": 2.0533,
      "step": 128749
    },
    {
      "epoch": 1.68,
      "grad_norm": 0.9609282612800598,
      "learning_rate": 0.00024489371191941924,
      "loss": 1.964,
      "step": 128750
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.1081888675689697,
      "learning_rate": 0.0002448896909646474,
      "loss": 2.0768,
      "step": 128751
    },
    {
      "epoch": 1.68,
      "grad_norm": 0.9956843852996826,
      "learning_rate": 0.00024488567002012155,
      "loss": 2.0259,
      "step": 128752
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.0048482418060303,
      "learning_rate": 0.00024488164908584254,
      "loss": 1.9734,
      "step": 128753
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.083601713180542,
      "learning_rate": 0.000244877628161811,
      "loss": 2.0164,
      "step": 128754
    },
    {
      "epoch": 1.68,
      "grad_norm": 0.8790310025215149,
      "learning_rate": 0.00024487360724802785,
      "loss": 1.7949,
      "step": 128755
    },
    {
      "epoch": 1.68,
      "grad_norm": 0.9768967032432556,
      "learning_rate": 0.00024486958634449367,
      "loss": 2.1654,
      "step": 128756
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.0716580152511597,
      "learning_rate": 0.0002448655654512093,
      "loss": 2.0818,
      "step": 128757
    },
    {
      "epoch": 1.68,
      "grad_norm": 0.9990395307540894,
      "learning_rate": 0.0002448615445681754,
      "loss": 2.3157,
      "step": 128758
    },
    {
      "epoch": 1.68,
      "grad_norm": 0.9923909306526184,
      "learning_rate": 0.0002448575236953929,
      "loss": 2.1906,
      "step": 128759
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.1942362785339355,
      "learning_rate": 0.0002448535028328624,
      "loss": 2.2437,
      "step": 128760
    },
    {
      "epoch": 1.68,
      "grad_norm": 0.9752339720726013,
      "learning_rate": 0.0002448494819805846,
      "loss": 2.321,
      "step": 128761
    },
    {
      "epoch": 1.68,
      "grad_norm": 0.9580263495445251,
      "learning_rate": 0.0002448454611385604,
      "loss": 1.896,
      "step": 128762
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.0678550004959106,
      "learning_rate": 0.0002448414403067904,
      "loss": 1.8603,
      "step": 128763
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.0993636846542358,
      "learning_rate": 0.00024483741948527543,
      "loss": 2.2316,
      "step": 128764
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.0227158069610596,
      "learning_rate": 0.00024483339867401624,
      "loss": 2.2122,
      "step": 128765
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.0460355281829834,
      "learning_rate": 0.0002448293778730136,
      "loss": 2.3788,
      "step": 128766
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.2300260066986084,
      "learning_rate": 0.0002448253570822682,
      "loss": 2.0993,
      "step": 128767
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.1723419427871704,
      "learning_rate": 0.0002448213363017808,
      "loss": 2.1758,
      "step": 128768
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.1203213930130005,
      "learning_rate": 0.00024481731553155215,
      "loss": 2.2832,
      "step": 128769
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.0145244598388672,
      "learning_rate": 0.00024481329477158293,
      "loss": 2.1593,
      "step": 128770
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.0876610279083252,
      "learning_rate": 0.00024480927402187406,
      "loss": 1.9506,
      "step": 128771
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.1586881875991821,
      "learning_rate": 0.00024480525328242613,
      "loss": 2.2432,
      "step": 128772
    },
    {
      "epoch": 1.68,
      "grad_norm": 0.9874042272567749,
      "learning_rate": 0.00024480123255324007,
      "loss": 2.001,
      "step": 128773
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.0179004669189453,
      "learning_rate": 0.0002447972118343164,
      "loss": 1.9827,
      "step": 128774
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.0379589796066284,
      "learning_rate": 0.0002447931911256559,
      "loss": 2.1613,
      "step": 128775
    },
    {
      "epoch": 1.68,
      "grad_norm": 0.9308146238327026,
      "learning_rate": 0.0002447891704272595,
      "loss": 1.9046,
      "step": 128776
    },
    {
      "epoch": 1.68,
      "grad_norm": 0.9530032873153687,
      "learning_rate": 0.00024478514973912775,
      "loss": 2.132,
      "step": 128777
    },
    {
      "epoch": 1.68,
      "grad_norm": 0.9500352740287781,
      "learning_rate": 0.00024478112906126153,
      "loss": 2.1628,
      "step": 128778
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.262739896774292,
      "learning_rate": 0.00024477710839366167,
      "loss": 2.2794,
      "step": 128779
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.0020020008087158,
      "learning_rate": 0.0002447730877363286,
      "loss": 2.0505,
      "step": 128780
    },
    {
      "epoch": 1.68,
      "grad_norm": 0.9965759515762329,
      "learning_rate": 0.00024476906708926333,
      "loss": 2.163,
      "step": 128781
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.0090538263320923,
      "learning_rate": 0.0002447650464524665,
      "loss": 2.2086,
      "step": 128782
    },
    {
      "epoch": 1.68,
      "grad_norm": 0.9450995326042175,
      "learning_rate": 0.0002447610258259389,
      "loss": 2.0805,
      "step": 128783
    },
    {
      "epoch": 1.68,
      "grad_norm": 0.9447615146636963,
      "learning_rate": 0.0002447570052096813,
      "loss": 2.0755,
      "step": 128784
    },
    {
      "epoch": 1.68,
      "grad_norm": 0.9512488842010498,
      "learning_rate": 0.0002447529846036943,
      "loss": 1.8674,
      "step": 128785
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.214573860168457,
      "learning_rate": 0.000244748964007979,
      "loss": 2.0456,
      "step": 128786
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.1363192796707153,
      "learning_rate": 0.0002447449434225357,
      "loss": 2.1773,
      "step": 128787
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.0149887800216675,
      "learning_rate": 0.00024474092284736544,
      "loss": 2.073,
      "step": 128788
    },
    {
      "epoch": 1.68,
      "grad_norm": 0.9762327075004578,
      "learning_rate": 0.00024473690228246886,
      "loss": 2.4811,
      "step": 128789
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.1046894788742065,
      "learning_rate": 0.0002447328817278467,
      "loss": 1.9419,
      "step": 128790
    },
    {
      "epoch": 1.68,
      "grad_norm": 0.9742581844329834,
      "learning_rate": 0.00024472886118349983,
      "loss": 1.9464,
      "step": 128791
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.0294109582901,
      "learning_rate": 0.00024472484064942896,
      "loss": 2.1385,
      "step": 128792
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.0790762901306152,
      "learning_rate": 0.00024472082012563464,
      "loss": 2.0736,
      "step": 128793
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.1910055875778198,
      "learning_rate": 0.0002447167996121178,
      "loss": 2.2957,
      "step": 128794
    },
    {
      "epoch": 1.68,
      "grad_norm": 0.9979855418205261,
      "learning_rate": 0.00024471277910887915,
      "loss": 2.1989,
      "step": 128795
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.097730040550232,
      "learning_rate": 0.00024470875861591943,
      "loss": 2.399,
      "step": 128796
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.0130444765090942,
      "learning_rate": 0.0002447047381332394,
      "loss": 2.1885,
      "step": 128797
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.0967060327529907,
      "learning_rate": 0.0002447007176608399,
      "loss": 2.2243,
      "step": 128798
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.1634752750396729,
      "learning_rate": 0.00024469669719872153,
      "loss": 2.0616,
      "step": 128799
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.096367597579956,
      "learning_rate": 0.000244692676746885,
      "loss": 1.9875,
      "step": 128800
    },
    {
      "epoch": 1.68,
      "grad_norm": 0.9960343241691589,
      "learning_rate": 0.0002446886563053312,
      "loss": 1.9509,
      "step": 128801
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.3557997941970825,
      "learning_rate": 0.00024468463587406077,
      "loss": 2.1486,
      "step": 128802
    },
    {
      "epoch": 1.68,
      "grad_norm": 0.9912436604499817,
      "learning_rate": 0.00024468061545307457,
      "loss": 2.1505,
      "step": 128803
    },
    {
      "epoch": 1.68,
      "grad_norm": 0.9627319574356079,
      "learning_rate": 0.0002446765950423734,
      "loss": 2.1349,
      "step": 128804
    },
    {
      "epoch": 1.68,
      "grad_norm": 0.9330788254737854,
      "learning_rate": 0.0002446725746419578,
      "loss": 1.8871,
      "step": 128805
    },
    {
      "epoch": 1.68,
      "grad_norm": 0.9529204368591309,
      "learning_rate": 0.00024466855425182854,
      "loss": 1.9854,
      "step": 128806
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.1412714719772339,
      "learning_rate": 0.00024466453387198646,
      "loss": 2.1979,
      "step": 128807
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.0182939767837524,
      "learning_rate": 0.0002446605135024323,
      "loss": 2.0873,
      "step": 128808
    },
    {
      "epoch": 1.68,
      "grad_norm": 0.9932993650436401,
      "learning_rate": 0.00024465649314316676,
      "loss": 2.1468,
      "step": 128809
    },
    {
      "epoch": 1.68,
      "grad_norm": 0.8946935534477234,
      "learning_rate": 0.00024465247279419083,
      "loss": 2.0074,
      "step": 128810
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.0639886856079102,
      "learning_rate": 0.00024464845245550484,
      "loss": 2.0752,
      "step": 128811
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.0114936828613281,
      "learning_rate": 0.0002446444321271098,
      "loss": 2.139,
      "step": 128812
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.0032330751419067,
      "learning_rate": 0.0002446404118090064,
      "loss": 2.1869,
      "step": 128813
    },
    {
      "epoch": 1.68,
      "grad_norm": 0.999882698059082,
      "learning_rate": 0.0002446363915011954,
      "loss": 1.9819,
      "step": 128814
    },
    {
      "epoch": 1.68,
      "grad_norm": 0.9624893069267273,
      "learning_rate": 0.00024463237120367755,
      "loss": 2.0459,
      "step": 128815
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.0109108686447144,
      "learning_rate": 0.00024462835091645365,
      "loss": 1.9685,
      "step": 128816
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.0122557878494263,
      "learning_rate": 0.0002446243306395243,
      "loss": 1.9547,
      "step": 128817
    },
    {
      "epoch": 1.68,
      "grad_norm": 0.9666680693626404,
      "learning_rate": 0.00024462031037289034,
      "loss": 1.8704,
      "step": 128818
    },
    {
      "epoch": 1.68,
      "grad_norm": 0.9580832719802856,
      "learning_rate": 0.0002446162901165525,
      "loss": 2.1682,
      "step": 128819
    },
    {
      "epoch": 1.68,
      "grad_norm": 0.9927055239677429,
      "learning_rate": 0.0002446122698705115,
      "loss": 2.1405,
      "step": 128820
    },
    {
      "epoch": 1.68,
      "grad_norm": 0.8331619501113892,
      "learning_rate": 0.0002446082496347682,
      "loss": 2.0656,
      "step": 128821
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.0483096837997437,
      "learning_rate": 0.00024460422940932334,
      "loss": 2.2604,
      "step": 128822
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.0130231380462646,
      "learning_rate": 0.0002446002091941775,
      "loss": 1.9822,
      "step": 128823
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.0427080392837524,
      "learning_rate": 0.00024459618898933154,
      "loss": 1.9857,
      "step": 128824
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.0766488313674927,
      "learning_rate": 0.00024459216879478616,
      "loss": 2.1555,
      "step": 128825
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.0368921756744385,
      "learning_rate": 0.0002445881486105421,
      "loss": 2.0021,
      "step": 128826
    },
    {
      "epoch": 1.68,
      "grad_norm": 0.9962682723999023,
      "learning_rate": 0.0002445841284366002,
      "loss": 1.9483,
      "step": 128827
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.0510897636413574,
      "learning_rate": 0.00024458010827296123,
      "loss": 2.0777,
      "step": 128828
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.2142906188964844,
      "learning_rate": 0.0002445760881196259,
      "loss": 2.2406,
      "step": 128829
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.071130394935608,
      "learning_rate": 0.0002445720679765948,
      "loss": 1.96,
      "step": 128830
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.0808676481246948,
      "learning_rate": 0.0002445680478438688,
      "loss": 2.2169,
      "step": 128831
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.0489755868911743,
      "learning_rate": 0.00024456402772144863,
      "loss": 2.0668,
      "step": 128832
    },
    {
      "epoch": 1.68,
      "grad_norm": 0.9217413663864136,
      "learning_rate": 0.0002445600076093351,
      "loss": 2.0435,
      "step": 128833
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.1272679567337036,
      "learning_rate": 0.00024455598750752896,
      "loss": 1.9434,
      "step": 128834
    },
    {
      "epoch": 1.68,
      "grad_norm": 0.9612358212471008,
      "learning_rate": 0.0002445519674160308,
      "loss": 2.0798,
      "step": 128835
    },
    {
      "epoch": 1.68,
      "grad_norm": 0.9558107256889343,
      "learning_rate": 0.00024454794733484166,
      "loss": 2.1288,
      "step": 128836
    },
    {
      "epoch": 1.68,
      "grad_norm": 0.9749125242233276,
      "learning_rate": 0.00024454392726396196,
      "loss": 2.1539,
      "step": 128837
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.0885063409805298,
      "learning_rate": 0.00024453990720339254,
      "loss": 1.8415,
      "step": 128838
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.1754168272018433,
      "learning_rate": 0.00024453588715313426,
      "loss": 2.2165,
      "step": 128839
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.0474269390106201,
      "learning_rate": 0.0002445318671131879,
      "loss": 2.1244,
      "step": 128840
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.1345868110656738,
      "learning_rate": 0.000244527847083554,
      "loss": 2.0169,
      "step": 128841
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.0047825574874878,
      "learning_rate": 0.0002445238270642334,
      "loss": 2.1653,
      "step": 128842
    },
    {
      "epoch": 1.68,
      "grad_norm": 0.9381217956542969,
      "learning_rate": 0.000244519807055227,
      "loss": 2.2283,
      "step": 128843
    },
    {
      "epoch": 1.68,
      "grad_norm": 0.9931626319885254,
      "learning_rate": 0.0002445157870565353,
      "loss": 2.0723,
      "step": 128844
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.0769575834274292,
      "learning_rate": 0.00024451176706815914,
      "loss": 2.0757,
      "step": 128845
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.0266920328140259,
      "learning_rate": 0.0002445077470900994,
      "loss": 1.9807,
      "step": 128846
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.0461533069610596,
      "learning_rate": 0.00024450372712235666,
      "loss": 2.1196,
      "step": 128847
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.0474780797958374,
      "learning_rate": 0.0002444997071649317,
      "loss": 2.138,
      "step": 128848
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.1481865644454956,
      "learning_rate": 0.0002444956872178253,
      "loss": 2.0598,
      "step": 128849
    },
    {
      "epoch": 1.68,
      "grad_norm": 0.9800429344177246,
      "learning_rate": 0.0002444916672810382,
      "loss": 2.1128,
      "step": 128850
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.0065319538116455,
      "learning_rate": 0.0002444876473545712,
      "loss": 2.2996,
      "step": 128851
    },
    {
      "epoch": 1.68,
      "grad_norm": 0.9341146945953369,
      "learning_rate": 0.000244483627438425,
      "loss": 2.0674,
      "step": 128852
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.0126218795776367,
      "learning_rate": 0.0002444796075326003,
      "loss": 2.0054,
      "step": 128853
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.0844569206237793,
      "learning_rate": 0.00024447558763709786,
      "loss": 2.0851,
      "step": 128854
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.0301278829574585,
      "learning_rate": 0.00024447156775191846,
      "loss": 2.1358,
      "step": 128855
    },
    {
      "epoch": 1.68,
      "grad_norm": 0.9726428389549255,
      "learning_rate": 0.00024446754787706284,
      "loss": 2.0892,
      "step": 128856
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.098333716392517,
      "learning_rate": 0.0002444635280125318,
      "loss": 2.1853,
      "step": 128857
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.073888897895813,
      "learning_rate": 0.00024445950815832606,
      "loss": 1.7642,
      "step": 128858
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.015771508216858,
      "learning_rate": 0.00024445548831444636,
      "loss": 2.023,
      "step": 128859
    },
    {
      "epoch": 1.68,
      "grad_norm": 0.99891597032547,
      "learning_rate": 0.00024445146848089337,
      "loss": 2.087,
      "step": 128860
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.1122746467590332,
      "learning_rate": 0.0002444474486576679,
      "loss": 1.9972,
      "step": 128861
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.0568856000900269,
      "learning_rate": 0.0002444434288447706,
      "loss": 2.1293,
      "step": 128862
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.0140520334243774,
      "learning_rate": 0.00024443940904220245,
      "loss": 2.2206,
      "step": 128863
    },
    {
      "epoch": 1.68,
      "grad_norm": 0.9978328943252563,
      "learning_rate": 0.000244435389249964,
      "loss": 2.2524,
      "step": 128864
    },
    {
      "epoch": 1.68,
      "grad_norm": 0.9467685222625732,
      "learning_rate": 0.0002444313694680562,
      "loss": 2.0728,
      "step": 128865
    },
    {
      "epoch": 1.68,
      "grad_norm": 0.9580655694007874,
      "learning_rate": 0.00024442734969647953,
      "loss": 1.9606,
      "step": 128866
    },
    {
      "epoch": 1.68,
      "grad_norm": 0.9753015041351318,
      "learning_rate": 0.0002444233299352349,
      "loss": 2.0681,
      "step": 128867
    },
    {
      "epoch": 1.68,
      "grad_norm": 0.8553328514099121,
      "learning_rate": 0.000244419310184323,
      "loss": 2.1114,
      "step": 128868
    },
    {
      "epoch": 1.68,
      "grad_norm": 0.9049539566040039,
      "learning_rate": 0.00024441529044374457,
      "loss": 2.1141,
      "step": 128869
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.012318730354309,
      "learning_rate": 0.0002444112707135004,
      "loss": 2.1835,
      "step": 128870
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.0937349796295166,
      "learning_rate": 0.00024440725099359134,
      "loss": 2.0162,
      "step": 128871
    },
    {
      "epoch": 1.68,
      "grad_norm": 0.9428645968437195,
      "learning_rate": 0.00024440323128401795,
      "loss": 2.0878,
      "step": 128872
    },
    {
      "epoch": 1.68,
      "grad_norm": 0.9591944813728333,
      "learning_rate": 0.000244399211584781,
      "loss": 2.132,
      "step": 128873
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.049686074256897,
      "learning_rate": 0.0002443951918958813,
      "loss": 2.0977,
      "step": 128874
    },
    {
      "epoch": 1.68,
      "grad_norm": 0.9229644536972046,
      "learning_rate": 0.00024439117221731956,
      "loss": 1.9569,
      "step": 128875
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.1669501066207886,
      "learning_rate": 0.00024438715254909655,
      "loss": 2.2568,
      "step": 128876
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.1051501035690308,
      "learning_rate": 0.0002443831328912132,
      "loss": 2.2931,
      "step": 128877
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.2017401456832886,
      "learning_rate": 0.0002443791132436699,
      "loss": 2.137,
      "step": 128878
    },
    {
      "epoch": 1.68,
      "grad_norm": 0.964892566204071,
      "learning_rate": 0.00024437509360646765,
      "loss": 2.0594,
      "step": 128879
    },
    {
      "epoch": 1.68,
      "grad_norm": 0.9903597235679626,
      "learning_rate": 0.000244371073979607,
      "loss": 2.1654,
      "step": 128880
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.056229591369629,
      "learning_rate": 0.0002443670543630889,
      "loss": 2.3106,
      "step": 128881
    },
    {
      "epoch": 1.68,
      "grad_norm": 0.9923532009124756,
      "learning_rate": 0.00024436303475691397,
      "loss": 2.0974,
      "step": 128882
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.0565072298049927,
      "learning_rate": 0.00024435901516108315,
      "loss": 2.2039,
      "step": 128883
    },
    {
      "epoch": 1.68,
      "grad_norm": 0.9876896739006042,
      "learning_rate": 0.0002443549955755969,
      "loss": 2.1883,
      "step": 128884
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.0779191255569458,
      "learning_rate": 0.00024435097600045613,
      "loss": 2.061,
      "step": 128885
    },
    {
      "epoch": 1.68,
      "grad_norm": 0.9782339334487915,
      "learning_rate": 0.00024434695643566157,
      "loss": 2.1287,
      "step": 128886
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.022412896156311,
      "learning_rate": 0.0002443429368812139,
      "loss": 2.0725,
      "step": 128887
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.1228384971618652,
      "learning_rate": 0.000244338917337114,
      "loss": 2.0588,
      "step": 128888
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.0021196603775024,
      "learning_rate": 0.00024433489780336266,
      "loss": 1.9206,
      "step": 128889
    },
    {
      "epoch": 1.68,
      "grad_norm": 0.9877497553825378,
      "learning_rate": 0.00024433087827996033,
      "loss": 2.2585,
      "step": 128890
    },
    {
      "epoch": 1.68,
      "grad_norm": 0.9464728832244873,
      "learning_rate": 0.000244326858766908,
      "loss": 2.0116,
      "step": 128891
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.0141301155090332,
      "learning_rate": 0.0002443228392642063,
      "loss": 2.182,
      "step": 128892
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.0523104667663574,
      "learning_rate": 0.0002443188197718561,
      "loss": 2.082,
      "step": 128893
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.0568691492080688,
      "learning_rate": 0.00024431480028985807,
      "loss": 2.1464,
      "step": 128894
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.041113018989563,
      "learning_rate": 0.0002443107808182131,
      "loss": 2.0489,
      "step": 128895
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.087641954421997,
      "learning_rate": 0.00024430676135692164,
      "loss": 2.2128,
      "step": 128896
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.0795836448669434,
      "learning_rate": 0.0002443027419059846,
      "loss": 2.0851,
      "step": 128897
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.246151328086853,
      "learning_rate": 0.0002442987224654028,
      "loss": 2.1745,
      "step": 128898
    },
    {
      "epoch": 1.68,
      "grad_norm": 0.9532843828201294,
      "learning_rate": 0.0002442947030351769,
      "loss": 2.0385,
      "step": 128899
    },
    {
      "epoch": 1.68,
      "grad_norm": 0.9131288528442383,
      "learning_rate": 0.0002442906836153076,
      "loss": 2.0037,
      "step": 128900
    },
    {
      "epoch": 1.68,
      "grad_norm": 0.8674842119216919,
      "learning_rate": 0.0002442866642057959,
      "loss": 1.7596,
      "step": 128901
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.0623140335083008,
      "learning_rate": 0.00024428264480664223,
      "loss": 1.8165,
      "step": 128902
    },
    {
      "epoch": 1.68,
      "grad_norm": 0.9483792781829834,
      "learning_rate": 0.00024427862541784746,
      "loss": 1.9853,
      "step": 128903
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.0271978378295898,
      "learning_rate": 0.00024427460603941234,
      "loss": 2.3113,
      "step": 128904
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.0416228771209717,
      "learning_rate": 0.00024427058667133764,
      "loss": 2.0818,
      "step": 128905
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.072441577911377,
      "learning_rate": 0.00024426656731362407,
      "loss": 2.2627,
      "step": 128906
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.1627416610717773,
      "learning_rate": 0.00024426254796627254,
      "loss": 1.9971,
      "step": 128907
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.0972840785980225,
      "learning_rate": 0.0002442585286292835,
      "loss": 2.1278,
      "step": 128908
    },
    {
      "epoch": 1.68,
      "grad_norm": 0.9887768030166626,
      "learning_rate": 0.00024425450930265784,
      "loss": 2.0189,
      "step": 128909
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.0277321338653564,
      "learning_rate": 0.0002442504899863964,
      "loss": 2.2295,
      "step": 128910
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.0074584484100342,
      "learning_rate": 0.0002442464706804998,
      "loss": 2.0789,
      "step": 128911
    },
    {
      "epoch": 1.68,
      "grad_norm": 0.9910346269607544,
      "learning_rate": 0.00024424245138496883,
      "loss": 2.259,
      "step": 128912
    },
    {
      "epoch": 1.68,
      "grad_norm": 0.9128152132034302,
      "learning_rate": 0.00024423843209980426,
      "loss": 2.0592,
      "step": 128913
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.0850857496261597,
      "learning_rate": 0.00024423441282500684,
      "loss": 2.0852,
      "step": 128914
    },
    {
      "epoch": 1.68,
      "grad_norm": 0.9837297201156616,
      "learning_rate": 0.00024423039356057726,
      "loss": 2.1486,
      "step": 128915
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.0500956773757935,
      "learning_rate": 0.00024422637430651634,
      "loss": 2.1974,
      "step": 128916
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.091182827949524,
      "learning_rate": 0.00024422235506282473,
      "loss": 2.0072,
      "step": 128917
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.0921615362167358,
      "learning_rate": 0.00024421833582950324,
      "loss": 2.0851,
      "step": 128918
    },
    {
      "epoch": 1.68,
      "grad_norm": 0.9560550451278687,
      "learning_rate": 0.0002442143166065526,
      "loss": 2.0826,
      "step": 128919
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.0999040603637695,
      "learning_rate": 0.00024421029739397367,
      "loss": 1.9851,
      "step": 128920
    },
    {
      "epoch": 1.68,
      "grad_norm": 0.9815146327018738,
      "learning_rate": 0.0002442062781917671,
      "loss": 2.0172,
      "step": 128921
    },
    {
      "epoch": 1.68,
      "grad_norm": 0.9716254472732544,
      "learning_rate": 0.0002442022589999336,
      "loss": 2.2524,
      "step": 128922
    },
    {
      "epoch": 1.68,
      "grad_norm": 0.9871693849563599,
      "learning_rate": 0.0002441982398184739,
      "loss": 1.9114,
      "step": 128923
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.0037405490875244,
      "learning_rate": 0.0002441942206473888,
      "loss": 1.953,
      "step": 128924
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.0004022121429443,
      "learning_rate": 0.00024419020148667904,
      "loss": 2.1195,
      "step": 128925
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.0844043493270874,
      "learning_rate": 0.0002441861823363455,
      "loss": 2.1465,
      "step": 128926
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.0316272974014282,
      "learning_rate": 0.00024418216319638865,
      "loss": 2.2369,
      "step": 128927
    },
    {
      "epoch": 1.68,
      "grad_norm": 0.9274252653121948,
      "learning_rate": 0.00024417814406680954,
      "loss": 2.2127,
      "step": 128928
    },
    {
      "epoch": 1.68,
      "grad_norm": 0.9914078116416931,
      "learning_rate": 0.0002441741249476086,
      "loss": 2.144,
      "step": 128929
    },
    {
      "epoch": 1.68,
      "grad_norm": 0.94697505235672,
      "learning_rate": 0.0002441701058387869,
      "loss": 2.167,
      "step": 128930
    },
    {
      "epoch": 1.68,
      "grad_norm": 0.9839374423027039,
      "learning_rate": 0.0002441660867403449,
      "loss": 2.0001,
      "step": 128931
    },
    {
      "epoch": 1.68,
      "grad_norm": 0.9297699332237244,
      "learning_rate": 0.0002441620676522836,
      "loss": 2.288,
      "step": 128932
    },
    {
      "epoch": 1.68,
      "grad_norm": 0.9725858569145203,
      "learning_rate": 0.0002441580485746035,
      "loss": 1.9817,
      "step": 128933
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.0063717365264893,
      "learning_rate": 0.00024415402950730557,
      "loss": 1.9857,
      "step": 128934
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.0642794370651245,
      "learning_rate": 0.00024415001045039044,
      "loss": 2.1918,
      "step": 128935
    },
    {
      "epoch": 1.68,
      "grad_norm": 0.9621593356132507,
      "learning_rate": 0.00024414599140385883,
      "loss": 1.865,
      "step": 128936
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.0369645357131958,
      "learning_rate": 0.0002441419723677116,
      "loss": 2.0434,
      "step": 128937
    },
    {
      "epoch": 1.68,
      "grad_norm": 0.9445177316665649,
      "learning_rate": 0.00024413795334194942,
      "loss": 2.2168,
      "step": 128938
    },
    {
      "epoch": 1.68,
      "grad_norm": 0.9126758575439453,
      "learning_rate": 0.000244133934326573,
      "loss": 2.2218,
      "step": 128939
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.0899839401245117,
      "learning_rate": 0.00024412991532158317,
      "loss": 2.124,
      "step": 128940
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.0395572185516357,
      "learning_rate": 0.0002441258963269806,
      "loss": 2.0376,
      "step": 128941
    },
    {
      "epoch": 1.68,
      "grad_norm": 0.9383646845817566,
      "learning_rate": 0.00024412187734276612,
      "loss": 2.069,
      "step": 128942
    },
    {
      "epoch": 1.68,
      "grad_norm": 0.9979195594787598,
      "learning_rate": 0.00024411785836894048,
      "loss": 2.1403,
      "step": 128943
    },
    {
      "epoch": 1.68,
      "grad_norm": 0.9758257865905762,
      "learning_rate": 0.0002441138394055044,
      "loss": 2.1234,
      "step": 128944
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.1030277013778687,
      "learning_rate": 0.00024410982045245853,
      "loss": 2.0683,
      "step": 128945
    },
    {
      "epoch": 1.68,
      "grad_norm": 0.9538027048110962,
      "learning_rate": 0.0002441058015098037,
      "loss": 2.0995,
      "step": 128946
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.1074211597442627,
      "learning_rate": 0.00024410178257754065,
      "loss": 2.1234,
      "step": 128947
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.1048492193222046,
      "learning_rate": 0.00024409776365567014,
      "loss": 2.2457,
      "step": 128948
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.0597314834594727,
      "learning_rate": 0.00024409374474419293,
      "loss": 2.124,
      "step": 128949
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.184442162513733,
      "learning_rate": 0.0002440897258431099,
      "loss": 1.9827,
      "step": 128950
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.0067596435546875,
      "learning_rate": 0.00024408570695242144,
      "loss": 2.2868,
      "step": 128951
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.2430086135864258,
      "learning_rate": 0.00024408168807212857,
      "loss": 2.0873,
      "step": 128952
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.4081228971481323,
      "learning_rate": 0.00024407766920223192,
      "loss": 2.2033,
      "step": 128953
    },
    {
      "epoch": 1.68,
      "grad_norm": 0.9580320715904236,
      "learning_rate": 0.0002440736503427323,
      "loss": 1.9122,
      "step": 128954
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.014460802078247,
      "learning_rate": 0.00024406963149363047,
      "loss": 2.1299,
      "step": 128955
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.165645718574524,
      "learning_rate": 0.00024406561265492727,
      "loss": 2.0881,
      "step": 128956
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.0142543315887451,
      "learning_rate": 0.0002440615938266232,
      "loss": 2.279,
      "step": 128957
    },
    {
      "epoch": 1.68,
      "grad_norm": 0.9899419546127319,
      "learning_rate": 0.0002440575750087191,
      "loss": 1.9773,
      "step": 128958
    },
    {
      "epoch": 1.68,
      "grad_norm": 0.9743945002555847,
      "learning_rate": 0.00024405355620121577,
      "loss": 2.0226,
      "step": 128959
    },
    {
      "epoch": 1.68,
      "grad_norm": 0.9502578973770142,
      "learning_rate": 0.00024404953740411397,
      "loss": 2.3244,
      "step": 128960
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.0209840536117554,
      "learning_rate": 0.0002440455186174144,
      "loss": 2.0066,
      "step": 128961
    },
    {
      "epoch": 1.68,
      "grad_norm": 0.9604872465133667,
      "learning_rate": 0.00024404149984111795,
      "loss": 2.189,
      "step": 128962
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.0004628896713257,
      "learning_rate": 0.00024403748107522513,
      "loss": 2.0426,
      "step": 128963
    },
    {
      "epoch": 1.68,
      "grad_norm": 0.9547125101089478,
      "learning_rate": 0.0002440334623197368,
      "loss": 2.2815,
      "step": 128964
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.0593922138214111,
      "learning_rate": 0.00024402944357465367,
      "loss": 2.1705,
      "step": 128965
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.0572766065597534,
      "learning_rate": 0.00024402542483997655,
      "loss": 2.0719,
      "step": 128966
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.0751510858535767,
      "learning_rate": 0.00024402140611570614,
      "loss": 2.3267,
      "step": 128967
    },
    {
      "epoch": 1.68,
      "grad_norm": 0.9253543615341187,
      "learning_rate": 0.00024401738740184334,
      "loss": 2.1637,
      "step": 128968
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.0238507986068726,
      "learning_rate": 0.00024401336869838864,
      "loss": 1.9876,
      "step": 128969
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.007996916770935,
      "learning_rate": 0.0002440093500053429,
      "loss": 1.8513,
      "step": 128970
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.0431015491485596,
      "learning_rate": 0.00024400533132270692,
      "loss": 2.1798,
      "step": 128971
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.0411651134490967,
      "learning_rate": 0.0002440013126504814,
      "loss": 1.9951,
      "step": 128972
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.02035653591156,
      "learning_rate": 0.00024399729398866707,
      "loss": 2.2391,
      "step": 128973
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.0686482191085815,
      "learning_rate": 0.00024399327533726484,
      "loss": 2.0667,
      "step": 128974
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.0541154146194458,
      "learning_rate": 0.00024398925669627517,
      "loss": 2.0262,
      "step": 128975
    },
    {
      "epoch": 1.68,
      "grad_norm": 0.9778273701667786,
      "learning_rate": 0.00024398523806569894,
      "loss": 2.3148,
      "step": 128976
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.0697022676467896,
      "learning_rate": 0.00024398121944553696,
      "loss": 2.038,
      "step": 128977
    },
    {
      "epoch": 1.68,
      "grad_norm": 0.995088517665863,
      "learning_rate": 0.0002439772008357899,
      "loss": 2.2506,
      "step": 128978
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.0353448390960693,
      "learning_rate": 0.00024397318223645854,
      "loss": 2.0688,
      "step": 128979
    },
    {
      "epoch": 1.68,
      "grad_norm": 0.9879262447357178,
      "learning_rate": 0.00024396916364754375,
      "loss": 2.2066,
      "step": 128980
    },
    {
      "epoch": 1.68,
      "grad_norm": 0.9983180165290833,
      "learning_rate": 0.00024396514506904604,
      "loss": 2.0201,
      "step": 128981
    },
    {
      "epoch": 1.68,
      "grad_norm": 0.9655014276504517,
      "learning_rate": 0.00024396112650096623,
      "loss": 2.0045,
      "step": 128982
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.1011652946472168,
      "learning_rate": 0.00024395710794330514,
      "loss": 1.989,
      "step": 128983
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.0447269678115845,
      "learning_rate": 0.00024395308939606346,
      "loss": 1.9409,
      "step": 128984
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.10209321975708,
      "learning_rate": 0.00024394907085924195,
      "loss": 2.1114,
      "step": 128985
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.156752347946167,
      "learning_rate": 0.00024394505233284137,
      "loss": 2.1734,
      "step": 128986
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.0262728929519653,
      "learning_rate": 0.00024394103381686263,
      "loss": 2.2075,
      "step": 128987
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.0371021032333374,
      "learning_rate": 0.00024393701531130616,
      "loss": 1.9179,
      "step": 128988
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.1791186332702637,
      "learning_rate": 0.00024393299681617285,
      "loss": 2.1433,
      "step": 128989
    },
    {
      "epoch": 1.68,
      "grad_norm": 0.9672515392303467,
      "learning_rate": 0.00024392897833146345,
      "loss": 2.008,
      "step": 128990
    },
    {
      "epoch": 1.68,
      "grad_norm": 0.950505793094635,
      "learning_rate": 0.00024392495985717873,
      "loss": 2.0795,
      "step": 128991
    },
    {
      "epoch": 1.68,
      "grad_norm": 0.975689172744751,
      "learning_rate": 0.0002439209413933194,
      "loss": 1.997,
      "step": 128992
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.0398565530776978,
      "learning_rate": 0.00024391692293988638,
      "loss": 2.1257,
      "step": 128993
    },
    {
      "epoch": 1.68,
      "grad_norm": 0.9261075258255005,
      "learning_rate": 0.0002439129044968801,
      "loss": 2.0522,
      "step": 128994
    },
    {
      "epoch": 1.68,
      "grad_norm": 0.9838981032371521,
      "learning_rate": 0.00024390888606430148,
      "loss": 1.9416,
      "step": 128995
    },
    {
      "epoch": 1.68,
      "grad_norm": 0.9304149746894836,
      "learning_rate": 0.00024390486764215124,
      "loss": 2.2355,
      "step": 128996
    },
    {
      "epoch": 1.68,
      "grad_norm": 0.9557098150253296,
      "learning_rate": 0.00024390084923043022,
      "loss": 2.0112,
      "step": 128997
    },
    {
      "epoch": 1.68,
      "grad_norm": 0.9368927478790283,
      "learning_rate": 0.00024389683082913903,
      "loss": 2.165,
      "step": 128998
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.1288102865219116,
      "learning_rate": 0.0002438928124382786,
      "loss": 2.1506,
      "step": 128999
    },
    {
      "epoch": 1.68,
      "grad_norm": 0.9370555281639099,
      "learning_rate": 0.00024388879405784941,
      "loss": 1.8892,
      "step": 129000
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.077987551689148,
      "learning_rate": 0.0002438847756878524,
      "loss": 2.1126,
      "step": 129001
    },
    {
      "epoch": 1.68,
      "grad_norm": 0.9630745053291321,
      "learning_rate": 0.00024388075732828825,
      "loss": 2.2717,
      "step": 129002
    },
    {
      "epoch": 1.68,
      "grad_norm": 0.9315971732139587,
      "learning_rate": 0.00024387673897915773,
      "loss": 2.068,
      "step": 129003
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.046449065208435,
      "learning_rate": 0.0002438727206404616,
      "loss": 2.1308,
      "step": 129004
    },
    {
      "epoch": 1.68,
      "grad_norm": 0.9210516810417175,
      "learning_rate": 0.00024386870231220067,
      "loss": 1.8585,
      "step": 129005
    },
    {
      "epoch": 1.68,
      "grad_norm": 0.9340496063232422,
      "learning_rate": 0.00024386468399437554,
      "loss": 2.1483,
      "step": 129006
    },
    {
      "epoch": 1.68,
      "grad_norm": 0.912514328956604,
      "learning_rate": 0.00024386066568698697,
      "loss": 2.0625,
      "step": 129007
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.01705002784729,
      "learning_rate": 0.0002438566473900358,
      "loss": 2.0134,
      "step": 129008
    },
    {
      "epoch": 1.68,
      "grad_norm": 0.9704145789146423,
      "learning_rate": 0.00024385262910352272,
      "loss": 2.1228,
      "step": 129009
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.1122722625732422,
      "learning_rate": 0.0002438486108274485,
      "loss": 2.0146,
      "step": 129010
    },
    {
      "epoch": 1.68,
      "grad_norm": 0.9488326907157898,
      "learning_rate": 0.00024384459256181394,
      "loss": 1.9538,
      "step": 129011
    },
    {
      "epoch": 1.68,
      "grad_norm": 0.9713537693023682,
      "learning_rate": 0.0002438405743066197,
      "loss": 2.0451,
      "step": 129012
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.0742343664169312,
      "learning_rate": 0.0002438365560618666,
      "loss": 2.2944,
      "step": 129013
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.0325462818145752,
      "learning_rate": 0.00024383253782755528,
      "loss": 2.2127,
      "step": 129014
    },
    {
      "epoch": 1.68,
      "grad_norm": 0.9525749683380127,
      "learning_rate": 0.00024382851960368654,
      "loss": 2.1239,
      "step": 129015
    },
    {
      "epoch": 1.68,
      "grad_norm": 0.9995586276054382,
      "learning_rate": 0.00024382450139026114,
      "loss": 2.1508,
      "step": 129016
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.141616702079773,
      "learning_rate": 0.00024382048318727988,
      "loss": 1.9944,
      "step": 129017
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.0126259326934814,
      "learning_rate": 0.00024381646499474342,
      "loss": 2.2477,
      "step": 129018
    },
    {
      "epoch": 1.68,
      "grad_norm": 0.9513981342315674,
      "learning_rate": 0.00024381244681265254,
      "loss": 2.2288,
      "step": 129019
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.1827541589736938,
      "learning_rate": 0.00024380842864100806,
      "loss": 2.1597,
      "step": 129020
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.039891242980957,
      "learning_rate": 0.00024380441047981054,
      "loss": 2.2091,
      "step": 129021
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.0959551334381104,
      "learning_rate": 0.00024380039232906088,
      "loss": 2.1769,
      "step": 129022
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.0523960590362549,
      "learning_rate": 0.00024379637418875982,
      "loss": 2.1348,
      "step": 129023
    },
    {
      "epoch": 1.68,
      "grad_norm": 0.9847145080566406,
      "learning_rate": 0.00024379235605890802,
      "loss": 2.141,
      "step": 129024
    },
    {
      "epoch": 1.68,
      "grad_norm": 0.9672187566757202,
      "learning_rate": 0.0002437883379395063,
      "loss": 2.0985,
      "step": 129025
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.0267131328582764,
      "learning_rate": 0.0002437843198305554,
      "loss": 1.8719,
      "step": 129026
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.0832878351211548,
      "learning_rate": 0.0002437803017320561,
      "loss": 2.1181,
      "step": 129027
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.2116268873214722,
      "learning_rate": 0.00024377628364400904,
      "loss": 2.2088,
      "step": 129028
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.2278985977172852,
      "learning_rate": 0.0002437722655664151,
      "loss": 2.0274,
      "step": 129029
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.0320391654968262,
      "learning_rate": 0.00024376824749927487,
      "loss": 2.3513,
      "step": 129030
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.1156213283538818,
      "learning_rate": 0.0002437642294425892,
      "loss": 2.1504,
      "step": 129031
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.040887713432312,
      "learning_rate": 0.00024376021139635882,
      "loss": 2.0425,
      "step": 129032
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.1195687055587769,
      "learning_rate": 0.0002437561933605845,
      "loss": 1.9895,
      "step": 129033
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.262152910232544,
      "learning_rate": 0.000243752175335267,
      "loss": 2.2055,
      "step": 129034
    },
    {
      "epoch": 1.68,
      "grad_norm": 0.8552625179290771,
      "learning_rate": 0.00024374815732040704,
      "loss": 2.025,
      "step": 129035
    },
    {
      "epoch": 1.68,
      "grad_norm": 0.9862919449806213,
      "learning_rate": 0.00024374413931600527,
      "loss": 2.1828,
      "step": 129036
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.001387357711792,
      "learning_rate": 0.00024374012132206258,
      "loss": 1.9342,
      "step": 129037
    },
    {
      "epoch": 1.68,
      "grad_norm": 0.9650186896324158,
      "learning_rate": 0.00024373610333857965,
      "loss": 2.2803,
      "step": 129038
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.003435492515564,
      "learning_rate": 0.0002437320853655572,
      "loss": 2.1661,
      "step": 129039
    },
    {
      "epoch": 1.68,
      "grad_norm": 0.984244704246521,
      "learning_rate": 0.00024372806740299605,
      "loss": 2.1144,
      "step": 129040
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.1670665740966797,
      "learning_rate": 0.00024372404945089706,
      "loss": 2.1397,
      "step": 129041
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.0405710935592651,
      "learning_rate": 0.00024372003150926067,
      "loss": 2.0608,
      "step": 129042
    },
    {
      "epoch": 1.68,
      "grad_norm": 0.8879260420799255,
      "learning_rate": 0.00024371601357808781,
      "loss": 1.8789,
      "step": 129043
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.081749677658081,
      "learning_rate": 0.0002437119956573792,
      "loss": 2.2328,
      "step": 129044
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.107640027999878,
      "learning_rate": 0.00024370797774713558,
      "loss": 2.3047,
      "step": 129045
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.03754723072052,
      "learning_rate": 0.00024370395984735774,
      "loss": 2.1795,
      "step": 129046
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.0188422203063965,
      "learning_rate": 0.00024369994195804647,
      "loss": 2.1605,
      "step": 129047
    },
    {
      "epoch": 1.68,
      "grad_norm": 0.9194220900535583,
      "learning_rate": 0.00024369592407920236,
      "loss": 1.7976,
      "step": 129048
    },
    {
      "epoch": 1.68,
      "grad_norm": 0.9366421103477478,
      "learning_rate": 0.00024369190621082625,
      "loss": 2.0186,
      "step": 129049
    },
    {
      "epoch": 1.68,
      "grad_norm": 0.9754623770713806,
      "learning_rate": 0.00024368788835291881,
      "loss": 1.9089,
      "step": 129050
    },
    {
      "epoch": 1.68,
      "grad_norm": 0.9604766964912415,
      "learning_rate": 0.0002436838705054809,
      "loss": 2.0077,
      "step": 129051
    },
    {
      "epoch": 1.68,
      "grad_norm": 0.9077986478805542,
      "learning_rate": 0.00024367985266851325,
      "loss": 2.0672,
      "step": 129052
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.2909437417984009,
      "learning_rate": 0.00024367583484201654,
      "loss": 1.8717,
      "step": 129053
    },
    {
      "epoch": 1.68,
      "grad_norm": 0.9730719923973083,
      "learning_rate": 0.0002436718170259917,
      "loss": 2.0714,
      "step": 129054
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.1064245700836182,
      "learning_rate": 0.0002436677992204392,
      "loss": 2.1556,
      "step": 129055
    },
    {
      "epoch": 1.68,
      "grad_norm": 0.9172207117080688,
      "learning_rate": 0.00024366378142535994,
      "loss": 2.0818,
      "step": 129056
    },
    {
      "epoch": 1.68,
      "grad_norm": 0.9714213013648987,
      "learning_rate": 0.0002436597636407546,
      "loss": 2.2873,
      "step": 129057
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.0967726707458496,
      "learning_rate": 0.00024365574586662398,
      "loss": 2.2527,
      "step": 129058
    },
    {
      "epoch": 1.68,
      "grad_norm": 0.9827016592025757,
      "learning_rate": 0.00024365172810296884,
      "loss": 2.2663,
      "step": 129059
    },
    {
      "epoch": 1.68,
      "grad_norm": 0.9836941361427307,
      "learning_rate": 0.00024364771034979005,
      "loss": 1.959,
      "step": 129060
    },
    {
      "epoch": 1.68,
      "grad_norm": 0.9446232914924622,
      "learning_rate": 0.0002436436926070881,
      "loss": 2.1161,
      "step": 129061
    },
    {
      "epoch": 1.68,
      "grad_norm": 0.9904970526695251,
      "learning_rate": 0.0002436396748748638,
      "loss": 2.1468,
      "step": 129062
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.0822288990020752,
      "learning_rate": 0.00024363565715311794,
      "loss": 2.0087,
      "step": 129063
    },
    {
      "epoch": 1.68,
      "grad_norm": 0.9950870275497437,
      "learning_rate": 0.00024363163944185134,
      "loss": 1.9549,
      "step": 129064
    },
    {
      "epoch": 1.68,
      "grad_norm": 0.9873722195625305,
      "learning_rate": 0.00024362762174106464,
      "loss": 2.0212,
      "step": 129065
    },
    {
      "epoch": 1.68,
      "grad_norm": 0.9726836681365967,
      "learning_rate": 0.00024362360405075876,
      "loss": 2.0626,
      "step": 129066
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.1091562509536743,
      "learning_rate": 0.0002436195863709342,
      "loss": 2.0525,
      "step": 129067
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.1432708501815796,
      "learning_rate": 0.00024361556870159182,
      "loss": 2.1057,
      "step": 129068
    },
    {
      "epoch": 1.68,
      "grad_norm": 0.9792733788490295,
      "learning_rate": 0.00024361155104273238,
      "loss": 2.0765,
      "step": 129069
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.0813920497894287,
      "learning_rate": 0.00024360753339435658,
      "loss": 2.1481,
      "step": 129070
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.010887861251831,
      "learning_rate": 0.0002436035157564652,
      "loss": 2.1882,
      "step": 129071
    },
    {
      "epoch": 1.68,
      "grad_norm": 0.9717054963111877,
      "learning_rate": 0.00024359949812905918,
      "loss": 2.06,
      "step": 129072
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.0817042589187622,
      "learning_rate": 0.0002435954805121389,
      "loss": 1.9783,
      "step": 129073
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.0649574995040894,
      "learning_rate": 0.0002435914629057053,
      "loss": 2.05,
      "step": 129074
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.0789748430252075,
      "learning_rate": 0.00024358744530975912,
      "loss": 2.0352,
      "step": 129075
    },
    {
      "epoch": 1.68,
      "grad_norm": 0.9619330763816833,
      "learning_rate": 0.0002435834277243011,
      "loss": 1.9874,
      "step": 129076
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.3501986265182495,
      "learning_rate": 0.00024357941014933198,
      "loss": 1.9286,
      "step": 129077
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.0698622465133667,
      "learning_rate": 0.00024357539258485262,
      "loss": 2.0977,
      "step": 129078
    },
    {
      "epoch": 1.68,
      "grad_norm": 0.9365466237068176,
      "learning_rate": 0.00024357137503086356,
      "loss": 1.8197,
      "step": 129079
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.1217658519744873,
      "learning_rate": 0.00024356735748736562,
      "loss": 2.2643,
      "step": 129080
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.2364463806152344,
      "learning_rate": 0.0002435633399543596,
      "loss": 2.0611,
      "step": 129081
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.0051742792129517,
      "learning_rate": 0.00024355932243184621,
      "loss": 2.0315,
      "step": 129082
    },
    {
      "epoch": 1.68,
      "grad_norm": 0.9415298700332642,
      "learning_rate": 0.0002435553049198262,
      "loss": 2.2324,
      "step": 129083
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.8202311992645264,
      "learning_rate": 0.0002435512874183005,
      "loss": 1.9032,
      "step": 129084
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.3177112340927124,
      "learning_rate": 0.0002435472699272695,
      "loss": 2.4239,
      "step": 129085
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.105760097503662,
      "learning_rate": 0.00024354325244673413,
      "loss": 2.3004,
      "step": 129086
    },
    {
      "epoch": 1.68,
      "grad_norm": 0.9988242983818054,
      "learning_rate": 0.00024353923497669517,
      "loss": 2.0356,
      "step": 129087
    },
    {
      "epoch": 1.68,
      "grad_norm": 0.9778918623924255,
      "learning_rate": 0.0002435352175171533,
      "loss": 2.2698,
      "step": 129088
    },
    {
      "epoch": 1.68,
      "grad_norm": 0.9848700165748596,
      "learning_rate": 0.00024353120006810928,
      "loss": 1.9454,
      "step": 129089
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.049935221672058,
      "learning_rate": 0.00024352718262956406,
      "loss": 2.1209,
      "step": 129090
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.2080705165863037,
      "learning_rate": 0.000243523165201518,
      "loss": 2.107,
      "step": 129091
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.0705596208572388,
      "learning_rate": 0.00024351914778397214,
      "loss": 2.1879,
      "step": 129092
    },
    {
      "epoch": 1.68,
      "grad_norm": 0.9779819846153259,
      "learning_rate": 0.00024351513037692708,
      "loss": 2.1622,
      "step": 129093
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.0106210708618164,
      "learning_rate": 0.00024351111298038361,
      "loss": 2.0937,
      "step": 129094
    },
    {
      "epoch": 1.68,
      "grad_norm": 0.9749201536178589,
      "learning_rate": 0.00024350709559434257,
      "loss": 2.1446,
      "step": 129095
    },
    {
      "epoch": 1.68,
      "grad_norm": 0.995629072189331,
      "learning_rate": 0.0002435030782188046,
      "loss": 1.9767,
      "step": 129096
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.009095311164856,
      "learning_rate": 0.00024349906085377053,
      "loss": 2.2725,
      "step": 129097
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.0208890438079834,
      "learning_rate": 0.00024349504349924098,
      "loss": 2.0667,
      "step": 129098
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.1169904470443726,
      "learning_rate": 0.00024349102615521675,
      "loss": 2.3336,
      "step": 129099
    },
    {
      "epoch": 1.68,
      "grad_norm": 0.9894130825996399,
      "learning_rate": 0.0002434870088216986,
      "loss": 2.0866,
      "step": 129100
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.0507251024246216,
      "learning_rate": 0.0002434829914986873,
      "loss": 2.0128,
      "step": 129101
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.1182854175567627,
      "learning_rate": 0.00024347897418618364,
      "loss": 2.0146,
      "step": 129102
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.0958096981048584,
      "learning_rate": 0.0002434749568841882,
      "loss": 2.068,
      "step": 129103
    },
    {
      "epoch": 1.68,
      "grad_norm": 0.9622356295585632,
      "learning_rate": 0.00024347093959270193,
      "loss": 2.1403,
      "step": 129104
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.0052634477615356,
      "learning_rate": 0.00024346692231172544,
      "loss": 2.0443,
      "step": 129105
    },
    {
      "epoch": 1.68,
      "grad_norm": 0.9856092929840088,
      "learning_rate": 0.0002434629050412595,
      "loss": 1.9892,
      "step": 129106
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.100563883781433,
      "learning_rate": 0.00024345888778130488,
      "loss": 2.2767,
      "step": 129107
    },
    {
      "epoch": 1.68,
      "grad_norm": 0.9874816536903381,
      "learning_rate": 0.00024345487053186238,
      "loss": 2.1688,
      "step": 129108
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.116801142692566,
      "learning_rate": 0.00024345085329293258,
      "loss": 1.9563,
      "step": 129109
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.1113629341125488,
      "learning_rate": 0.0002434468360645164,
      "loss": 2.2879,
      "step": 129110
    },
    {
      "epoch": 1.68,
      "grad_norm": 0.9317126870155334,
      "learning_rate": 0.00024344281884661447,
      "loss": 2.0368,
      "step": 129111
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.1258018016815186,
      "learning_rate": 0.0002434388016392277,
      "loss": 2.1228,
      "step": 129112
    },
    {
      "epoch": 1.68,
      "grad_norm": 0.9508216381072998,
      "learning_rate": 0.00024343478444235664,
      "loss": 1.9966,
      "step": 129113
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.1240261793136597,
      "learning_rate": 0.0002434307672560021,
      "loss": 2.303,
      "step": 129114
    },
    {
      "epoch": 1.68,
      "grad_norm": 0.9403828382492065,
      "learning_rate": 0.00024342675008016493,
      "loss": 2.1138,
      "step": 129115
    },
    {
      "epoch": 1.68,
      "grad_norm": 0.9434914588928223,
      "learning_rate": 0.0002434227329148457,
      "loss": 2.1953,
      "step": 129116
    },
    {
      "epoch": 1.68,
      "grad_norm": 0.896006166934967,
      "learning_rate": 0.0002434187157600453,
      "loss": 2.0338,
      "step": 129117
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.0709166526794434,
      "learning_rate": 0.00024341469861576442,
      "loss": 1.9692,
      "step": 129118
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.0052103996276855,
      "learning_rate": 0.00024341068148200383,
      "loss": 1.9772,
      "step": 129119
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.0740139484405518,
      "learning_rate": 0.00024340666435876426,
      "loss": 2.1166,
      "step": 129120
    },
    {
      "epoch": 1.68,
      "grad_norm": 0.9594852924346924,
      "learning_rate": 0.0002434026472460465,
      "loss": 2.0314,
      "step": 129121
    },
    {
      "epoch": 1.68,
      "grad_norm": 0.991459310054779,
      "learning_rate": 0.0002433986301438512,
      "loss": 1.9502,
      "step": 129122
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.026763916015625,
      "learning_rate": 0.00024339461305217917,
      "loss": 2.1123,
      "step": 129123
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.163732886314392,
      "learning_rate": 0.0002433905959710311,
      "loss": 1.9535,
      "step": 129124
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.089994192123413,
      "learning_rate": 0.00024338657890040784,
      "loss": 2.2333,
      "step": 129125
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.03280770778656,
      "learning_rate": 0.00024338256184031014,
      "loss": 2.3057,
      "step": 129126
    },
    {
      "epoch": 1.68,
      "grad_norm": 0.9463729858398438,
      "learning_rate": 0.00024337854479073866,
      "loss": 2.3751,
      "step": 129127
    },
    {
      "epoch": 1.68,
      "grad_norm": 0.9840006232261658,
      "learning_rate": 0.00024337452775169413,
      "loss": 2.0094,
      "step": 129128
    },
    {
      "epoch": 1.68,
      "grad_norm": 0.9724982976913452,
      "learning_rate": 0.00024337051072317737,
      "loss": 2.0744,
      "step": 129129
    },
    {
      "epoch": 1.68,
      "grad_norm": 0.9488010406494141,
      "learning_rate": 0.00024336649370518905,
      "loss": 2.0772,
      "step": 129130
    },
    {
      "epoch": 1.68,
      "grad_norm": 0.9442092776298523,
      "learning_rate": 0.00024336247669773001,
      "loss": 1.9886,
      "step": 129131
    },
    {
      "epoch": 1.68,
      "grad_norm": 0.9791611433029175,
      "learning_rate": 0.00024335845970080094,
      "loss": 2.2037,
      "step": 129132
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.1217314004898071,
      "learning_rate": 0.00024335444271440272,
      "loss": 2.0438,
      "step": 129133
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.0324788093566895,
      "learning_rate": 0.00024335042573853588,
      "loss": 2.2027,
      "step": 129134
    },
    {
      "epoch": 1.68,
      "grad_norm": 0.9659591317176819,
      "learning_rate": 0.00024334640877320124,
      "loss": 2.2835,
      "step": 129135
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.0373626947402954,
      "learning_rate": 0.00024334239181839957,
      "loss": 2.0934,
      "step": 129136
    },
    {
      "epoch": 1.68,
      "grad_norm": 0.9734171628952026,
      "learning_rate": 0.00024333837487413164,
      "loss": 1.8814,
      "step": 129137
    },
    {
      "epoch": 1.68,
      "grad_norm": 0.9321110248565674,
      "learning_rate": 0.00024333435794039819,
      "loss": 2.1729,
      "step": 129138
    },
    {
      "epoch": 1.68,
      "grad_norm": 0.934675395488739,
      "learning_rate": 0.00024333034101720004,
      "loss": 2.1519,
      "step": 129139
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.0153985023498535,
      "learning_rate": 0.00024332632410453773,
      "loss": 2.07,
      "step": 129140
    },
    {
      "epoch": 1.68,
      "grad_norm": 0.9869548678398132,
      "learning_rate": 0.0002433223072024121,
      "loss": 2.0562,
      "step": 129141
    },
    {
      "epoch": 1.68,
      "grad_norm": 0.9866814017295837,
      "learning_rate": 0.00024331829031082398,
      "loss": 2.0527,
      "step": 129142
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.0157071352005005,
      "learning_rate": 0.000243314273429774,
      "loss": 2.0273,
      "step": 129143
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.0811697244644165,
      "learning_rate": 0.00024331025655926304,
      "loss": 2.3033,
      "step": 129144
    },
    {
      "epoch": 1.68,
      "grad_norm": 0.9413239359855652,
      "learning_rate": 0.00024330623969929185,
      "loss": 2.0716,
      "step": 129145
    },
    {
      "epoch": 1.68,
      "grad_norm": 0.9238783121109009,
      "learning_rate": 0.00024330222284986097,
      "loss": 1.9722,
      "step": 129146
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.2072014808654785,
      "learning_rate": 0.0002432982060109713,
      "loss": 2.163,
      "step": 129147
    },
    {
      "epoch": 1.68,
      "grad_norm": 0.9818769693374634,
      "learning_rate": 0.00024329418918262355,
      "loss": 2.0551,
      "step": 129148
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.476234793663025,
      "learning_rate": 0.00024329017236481853,
      "loss": 2.2344,
      "step": 129149
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.09071946144104,
      "learning_rate": 0.00024328615555755683,
      "loss": 1.9999,
      "step": 129150
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.0514841079711914,
      "learning_rate": 0.00024328213876083954,
      "loss": 2.1037,
      "step": 129151
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.0933493375778198,
      "learning_rate": 0.00024327812197466698,
      "loss": 2.0342,
      "step": 129152
    },
    {
      "epoch": 1.68,
      "grad_norm": 0.9962890148162842,
      "learning_rate": 0.0002432741051990401,
      "loss": 2.1463,
      "step": 129153
    },
    {
      "epoch": 1.68,
      "grad_norm": 0.9501286149024963,
      "learning_rate": 0.00024327008843395967,
      "loss": 1.9874,
      "step": 129154
    },
    {
      "epoch": 1.68,
      "grad_norm": 0.9751250147819519,
      "learning_rate": 0.00024326607167942632,
      "loss": 2.1779,
      "step": 129155
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.0110195875167847,
      "learning_rate": 0.00024326205493544096,
      "loss": 1.8652,
      "step": 129156
    },
    {
      "epoch": 1.68,
      "grad_norm": 0.9300829172134399,
      "learning_rate": 0.00024325803820200435,
      "loss": 2.1621,
      "step": 129157
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.1060445308685303,
      "learning_rate": 0.000243254021479117,
      "loss": 2.3342,
      "step": 129158
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.0624620914459229,
      "learning_rate": 0.00024325000476677983,
      "loss": 2.1252,
      "step": 129159
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.0293042659759521,
      "learning_rate": 0.00024324598806499355,
      "loss": 1.8462,
      "step": 129160
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.0538653135299683,
      "learning_rate": 0.0002432419713737589,
      "loss": 1.9999,
      "step": 129161
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.045162320137024,
      "learning_rate": 0.00024323795469307664,
      "loss": 2.0309,
      "step": 129162
    },
    {
      "epoch": 1.68,
      "grad_norm": 0.9221396446228027,
      "learning_rate": 0.00024323393802294763,
      "loss": 1.9773,
      "step": 129163
    },
    {
      "epoch": 1.68,
      "grad_norm": 0.9702022075653076,
      "learning_rate": 0.0002432299213633724,
      "loss": 2.1696,
      "step": 129164
    },
    {
      "epoch": 1.68,
      "grad_norm": 0.9706600904464722,
      "learning_rate": 0.0002432259047143518,
      "loss": 2.2411,
      "step": 129165
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.0123001337051392,
      "learning_rate": 0.00024322188807588655,
      "loss": 2.1092,
      "step": 129166
    },
    {
      "epoch": 1.68,
      "grad_norm": 0.9750088453292847,
      "learning_rate": 0.00024321787144797747,
      "loss": 2.2432,
      "step": 129167
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.04893159866333,
      "learning_rate": 0.00024321385483062522,
      "loss": 2.0258,
      "step": 129168
    },
    {
      "epoch": 1.68,
      "grad_norm": 0.9715927243232727,
      "learning_rate": 0.00024320983822383071,
      "loss": 2.1747,
      "step": 129169
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.0544182062149048,
      "learning_rate": 0.00024320582162759444,
      "loss": 2.1717,
      "step": 129170
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.0028530359268188,
      "learning_rate": 0.00024320180504191732,
      "loss": 2.141,
      "step": 129171
    },
    {
      "epoch": 1.68,
      "grad_norm": 0.9871747493743896,
      "learning_rate": 0.0002431977884668,
      "loss": 2.1338,
      "step": 129172
    },
    {
      "epoch": 1.68,
      "grad_norm": 0.9691372513771057,
      "learning_rate": 0.00024319377190224332,
      "loss": 2.0042,
      "step": 129173
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.1432461738586426,
      "learning_rate": 0.00024318975534824794,
      "loss": 2.02,
      "step": 129174
    },
    {
      "epoch": 1.68,
      "grad_norm": 0.949068546295166,
      "learning_rate": 0.00024318573880481484,
      "loss": 2.2195,
      "step": 129175
    },
    {
      "epoch": 1.68,
      "grad_norm": 0.9339422583580017,
      "learning_rate": 0.00024318172227194443,
      "loss": 2.3255,
      "step": 129176
    },
    {
      "epoch": 1.68,
      "grad_norm": 0.9596626162528992,
      "learning_rate": 0.00024317770574963762,
      "loss": 1.9484,
      "step": 129177
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.2059756517410278,
      "learning_rate": 0.00024317368923789515,
      "loss": 2.1399,
      "step": 129178
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.0846562385559082,
      "learning_rate": 0.00024316967273671776,
      "loss": 2.043,
      "step": 129179
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.1289215087890625,
      "learning_rate": 0.0002431656562461062,
      "loss": 2.2232,
      "step": 129180
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.0490779876708984,
      "learning_rate": 0.0002431616397660612,
      "loss": 2.4239,
      "step": 129181
    },
    {
      "epoch": 1.68,
      "grad_norm": 0.9776133894920349,
      "learning_rate": 0.00024315762329658367,
      "loss": 2.1354,
      "step": 129182
    },
    {
      "epoch": 1.68,
      "grad_norm": 0.9573045372962952,
      "learning_rate": 0.0002431536068376741,
      "loss": 2.0253,
      "step": 129183
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.1128299236297607,
      "learning_rate": 0.00024314959038933333,
      "loss": 2.1021,
      "step": 129184
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.193294644355774,
      "learning_rate": 0.00024314557395156212,
      "loss": 2.076,
      "step": 129185
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.001197099685669,
      "learning_rate": 0.0002431415575243612,
      "loss": 1.8111,
      "step": 129186
    },
    {
      "epoch": 1.68,
      "grad_norm": 0.9988583326339722,
      "learning_rate": 0.0002431375411077314,
      "loss": 2.1627,
      "step": 129187
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.0778882503509521,
      "learning_rate": 0.00024313352470167342,
      "loss": 2.3107,
      "step": 129188
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.0607500076293945,
      "learning_rate": 0.000243129508306188,
      "loss": 2.1687,
      "step": 129189
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.0506774187088013,
      "learning_rate": 0.00024312549192127578,
      "loss": 1.9325,
      "step": 129190
    },
    {
      "epoch": 1.68,
      "grad_norm": 0.9544137716293335,
      "learning_rate": 0.0002431214755469377,
      "loss": 2.1617,
      "step": 129191
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.090276837348938,
      "learning_rate": 0.0002431174591831743,
      "loss": 2.1941,
      "step": 129192
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.1343251466751099,
      "learning_rate": 0.0002431134428299865,
      "loss": 2.0537,
      "step": 129193
    },
    {
      "epoch": 1.68,
      "grad_norm": 0.9612189531326294,
      "learning_rate": 0.00024310942648737507,
      "loss": 2.1965,
      "step": 129194
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.0427814722061157,
      "learning_rate": 0.00024310541015534054,
      "loss": 2.0334,
      "step": 129195
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.0477149486541748,
      "learning_rate": 0.0002431013938338839,
      "loss": 1.8602,
      "step": 129196
    },
    {
      "epoch": 1.68,
      "grad_norm": 0.976558268070221,
      "learning_rate": 0.00024309737752300568,
      "loss": 2.0052,
      "step": 129197
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.0519767999649048,
      "learning_rate": 0.00024309336122270677,
      "loss": 2.2439,
      "step": 129198
    },
    {
      "epoch": 1.68,
      "grad_norm": 0.9936559796333313,
      "learning_rate": 0.0002430893449329879,
      "loss": 2.0711,
      "step": 129199
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.066616415977478,
      "learning_rate": 0.00024308532865384983,
      "loss": 2.0819,
      "step": 129200
    },
    {
      "epoch": 1.68,
      "grad_norm": 0.9792123436927795,
      "learning_rate": 0.0002430813123852932,
      "loss": 1.9303,
      "step": 129201
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.0369470119476318,
      "learning_rate": 0.00024307729612731884,
      "loss": 2.3983,
      "step": 129202
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.053253173828125,
      "learning_rate": 0.00024307327987992748,
      "loss": 2.259,
      "step": 129203
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.0277544260025024,
      "learning_rate": 0.00024306926364311992,
      "loss": 2.0375,
      "step": 129204
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.016282558441162,
      "learning_rate": 0.0002430652474168968,
      "loss": 2.1307,
      "step": 129205
    },
    {
      "epoch": 1.68,
      "grad_norm": 0.8920400738716125,
      "learning_rate": 0.000243061231201259,
      "loss": 2.029,
      "step": 129206
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.0142792463302612,
      "learning_rate": 0.00024305721499620714,
      "loss": 2.0259,
      "step": 129207
    },
    {
      "epoch": 1.68,
      "grad_norm": 0.9799248576164246,
      "learning_rate": 0.00024305319880174201,
      "loss": 2.1397,
      "step": 129208
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.0071382522583008,
      "learning_rate": 0.00024304918261786434,
      "loss": 2.0186,
      "step": 129209
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.0345295667648315,
      "learning_rate": 0.00024304516644457495,
      "loss": 2.3029,
      "step": 129210
    },
    {
      "epoch": 1.68,
      "grad_norm": 0.9659146070480347,
      "learning_rate": 0.00024304115028187453,
      "loss": 1.9353,
      "step": 129211
    },
    {
      "epoch": 1.68,
      "grad_norm": 0.9543965458869934,
      "learning_rate": 0.0002430371341297639,
      "loss": 2.0876,
      "step": 129212
    },
    {
      "epoch": 1.68,
      "grad_norm": 0.9640991687774658,
      "learning_rate": 0.00024303311798824366,
      "loss": 2.0397,
      "step": 129213
    },
    {
      "epoch": 1.68,
      "grad_norm": 0.9809623956680298,
      "learning_rate": 0.00024302910185731462,
      "loss": 2.326,
      "step": 129214
    },
    {
      "epoch": 1.68,
      "grad_norm": 0.9539411664009094,
      "learning_rate": 0.00024302508573697762,
      "loss": 1.934,
      "step": 129215
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.0085638761520386,
      "learning_rate": 0.00024302106962723324,
      "loss": 2.259,
      "step": 129216
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.0115666389465332,
      "learning_rate": 0.00024301705352808235,
      "loss": 2.078,
      "step": 129217
    },
    {
      "epoch": 1.68,
      "grad_norm": 0.9848570227622986,
      "learning_rate": 0.00024301303743952583,
      "loss": 2.1867,
      "step": 129218
    },
    {
      "epoch": 1.68,
      "grad_norm": 0.9590315818786621,
      "learning_rate": 0.00024300902136156408,
      "loss": 2.1748,
      "step": 129219
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.1093883514404297,
      "learning_rate": 0.00024300500529419804,
      "loss": 1.9783,
      "step": 129220
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.0289509296417236,
      "learning_rate": 0.00024300098923742846,
      "loss": 1.9262,
      "step": 129221
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.0510303974151611,
      "learning_rate": 0.00024299697319125609,
      "loss": 2.5295,
      "step": 129222
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.0019363164901733,
      "learning_rate": 0.00024299295715568163,
      "loss": 2.1712,
      "step": 129223
    },
    {
      "epoch": 1.68,
      "grad_norm": 0.9837822318077087,
      "learning_rate": 0.00024298894113070597,
      "loss": 1.9414,
      "step": 129224
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.0019868612289429,
      "learning_rate": 0.00024298492511632964,
      "loss": 2.0419,
      "step": 129225
    },
    {
      "epoch": 1.68,
      "grad_norm": 0.9628341197967529,
      "learning_rate": 0.00024298090911255347,
      "loss": 2.322,
      "step": 129226
    },
    {
      "epoch": 1.68,
      "grad_norm": 0.9772926568984985,
      "learning_rate": 0.00024297689311937826,
      "loss": 2.1239,
      "step": 129227
    },
    {
      "epoch": 1.68,
      "grad_norm": 0.9357069730758667,
      "learning_rate": 0.00024297287713680467,
      "loss": 2.1362,
      "step": 129228
    },
    {
      "epoch": 1.68,
      "grad_norm": 0.9655843377113342,
      "learning_rate": 0.00024296886116483357,
      "loss": 1.9544,
      "step": 129229
    },
    {
      "epoch": 1.68,
      "grad_norm": 0.9220267534255981,
      "learning_rate": 0.0002429648452034657,
      "loss": 2.1555,
      "step": 129230
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.070865273475647,
      "learning_rate": 0.00024296082925270163,
      "loss": 1.8715,
      "step": 129231
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.0228058099746704,
      "learning_rate": 0.00024295681331254221,
      "loss": 1.8059,
      "step": 129232
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.0633872747421265,
      "learning_rate": 0.00024295279738298822,
      "loss": 2.0534,
      "step": 129233
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.0710113048553467,
      "learning_rate": 0.00024294878146404036,
      "loss": 2.1618,
      "step": 129234
    },
    {
      "epoch": 1.68,
      "grad_norm": 0.9623807072639465,
      "learning_rate": 0.00024294476555569944,
      "loss": 1.9816,
      "step": 129235
    },
    {
      "epoch": 1.68,
      "grad_norm": 0.9676302075386047,
      "learning_rate": 0.00024294074965796624,
      "loss": 1.9973,
      "step": 129236
    },
    {
      "epoch": 1.68,
      "grad_norm": 0.9738660454750061,
      "learning_rate": 0.00024293673377084134,
      "loss": 2.0416,
      "step": 129237
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.100113868713379,
      "learning_rate": 0.00024293271789432554,
      "loss": 2.1558,
      "step": 129238
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.1614952087402344,
      "learning_rate": 0.00024292870202841967,
      "loss": 2.1328,
      "step": 129239
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.3239994049072266,
      "learning_rate": 0.00024292468617312442,
      "loss": 2.0897,
      "step": 129240
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.0627168416976929,
      "learning_rate": 0.00024292067032844053,
      "loss": 2.2146,
      "step": 129241
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.0353477001190186,
      "learning_rate": 0.0002429166544943689,
      "loss": 1.9721,
      "step": 129242
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.091062307357788,
      "learning_rate": 0.00024291263867091002,
      "loss": 2.0221,
      "step": 129243
    },
    {
      "epoch": 1.68,
      "grad_norm": 0.9713706970214844,
      "learning_rate": 0.00024290862285806475,
      "loss": 2.1148,
      "step": 129244
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.1449081897735596,
      "learning_rate": 0.00024290460705583383,
      "loss": 2.1086,
      "step": 129245
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.07708740234375,
      "learning_rate": 0.00024290059126421806,
      "loss": 2.0739,
      "step": 129246
    },
    {
      "epoch": 1.68,
      "grad_norm": 0.9506388306617737,
      "learning_rate": 0.0002428965754832181,
      "loss": 2.1171,
      "step": 129247
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.080911636352539,
      "learning_rate": 0.0002428925597128348,
      "loss": 2.0976,
      "step": 129248
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.028071403503418,
      "learning_rate": 0.00024288854395306894,
      "loss": 2.0634,
      "step": 129249
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.037215232849121,
      "learning_rate": 0.00024288452820392107,
      "loss": 1.9182,
      "step": 129250
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.0765858888626099,
      "learning_rate": 0.00024288051246539204,
      "loss": 2.0559,
      "step": 129251
    },
    {
      "epoch": 1.68,
      "grad_norm": 0.9581661820411682,
      "learning_rate": 0.0002428764967374826,
      "loss": 2.083,
      "step": 129252
    },
    {
      "epoch": 1.68,
      "grad_norm": 0.9926328659057617,
      "learning_rate": 0.00024287248102019347,
      "loss": 2.0454,
      "step": 129253
    },
    {
      "epoch": 1.68,
      "grad_norm": 0.9799487590789795,
      "learning_rate": 0.00024286846531352543,
      "loss": 2.0321,
      "step": 129254
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.2176316976547241,
      "learning_rate": 0.0002428644496174794,
      "loss": 2.143,
      "step": 129255
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.048227071762085,
      "learning_rate": 0.00024286043393205578,
      "loss": 2.1703,
      "step": 129256
    },
    {
      "epoch": 1.68,
      "grad_norm": 0.9782921671867371,
      "learning_rate": 0.00024285641825725547,
      "loss": 1.9767,
      "step": 129257
    },
    {
      "epoch": 1.68,
      "grad_norm": 0.9193930625915527,
      "learning_rate": 0.00024285240259307926,
      "loss": 2.1188,
      "step": 129258
    },
    {
      "epoch": 1.68,
      "grad_norm": 0.9171305894851685,
      "learning_rate": 0.00024284838693952785,
      "loss": 2.1547,
      "step": 129259
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.2262178659439087,
      "learning_rate": 0.00024284437129660203,
      "loss": 2.0675,
      "step": 129260
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.0822240114212036,
      "learning_rate": 0.00024284035566430261,
      "loss": 2.1345,
      "step": 129261
    },
    {
      "epoch": 1.68,
      "grad_norm": 0.9137521386146545,
      "learning_rate": 0.00024283634004263014,
      "loss": 2.2486,
      "step": 129262
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.1374728679656982,
      "learning_rate": 0.0002428323244315855,
      "loss": 2.1346,
      "step": 129263
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.0418034791946411,
      "learning_rate": 0.00024282830883116938,
      "loss": 2.1989,
      "step": 129264
    },
    {
      "epoch": 1.68,
      "grad_norm": 0.9436930418014526,
      "learning_rate": 0.00024282429324138256,
      "loss": 2.2236,
      "step": 129265
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.0559618473052979,
      "learning_rate": 0.00024282027766222577,
      "loss": 2.0075,
      "step": 129266
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.2328015565872192,
      "learning_rate": 0.0002428162620936999,
      "loss": 2.162,
      "step": 129267
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.1182454824447632,
      "learning_rate": 0.00024281224653580546,
      "loss": 1.9628,
      "step": 129268
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.1076555252075195,
      "learning_rate": 0.0002428082309885433,
      "loss": 2.1585,
      "step": 129269
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.018722653388977,
      "learning_rate": 0.00024280421545191415,
      "loss": 2.3976,
      "step": 129270
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.063522219657898,
      "learning_rate": 0.00024280019992591878,
      "loss": 1.9189,
      "step": 129271
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.0022549629211426,
      "learning_rate": 0.00024279618441055793,
      "loss": 2.2623,
      "step": 129272
    },
    {
      "epoch": 1.68,
      "grad_norm": 0.9859896302223206,
      "learning_rate": 0.0002427921689058324,
      "loss": 2.0511,
      "step": 129273
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.1154308319091797,
      "learning_rate": 0.0002427881534117429,
      "loss": 1.9352,
      "step": 129274
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.0331976413726807,
      "learning_rate": 0.00024278413792829014,
      "loss": 2.1661,
      "step": 129275
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.0166006088256836,
      "learning_rate": 0.00024278012245547485,
      "loss": 2.2476,
      "step": 129276
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.0358436107635498,
      "learning_rate": 0.00024277610699329782,
      "loss": 2.0401,
      "step": 129277
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.060374140739441,
      "learning_rate": 0.0002427720915417598,
      "loss": 2.0606,
      "step": 129278
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.1743090152740479,
      "learning_rate": 0.0002427680761008616,
      "loss": 1.9711,
      "step": 129279
    },
    {
      "epoch": 1.68,
      "grad_norm": 0.9652992486953735,
      "learning_rate": 0.0002427640606706038,
      "loss": 1.9183,
      "step": 129280
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.180466651916504,
      "learning_rate": 0.00024276004525098732,
      "loss": 2.1827,
      "step": 129281
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.0392088890075684,
      "learning_rate": 0.00024275602984201276,
      "loss": 2.2244,
      "step": 129282
    },
    {
      "epoch": 1.68,
      "grad_norm": 0.9586292505264282,
      "learning_rate": 0.00024275201444368095,
      "loss": 2.1209,
      "step": 129283
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.0045944452285767,
      "learning_rate": 0.0002427479990559926,
      "loss": 1.9922,
      "step": 129284
    },
    {
      "epoch": 1.68,
      "grad_norm": 0.937526285648346,
      "learning_rate": 0.00024274398367894853,
      "loss": 2.2757,
      "step": 129285
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.0709973573684692,
      "learning_rate": 0.00024273996831254943,
      "loss": 1.8375,
      "step": 129286
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.0196696519851685,
      "learning_rate": 0.000242735952956796,
      "loss": 2.3844,
      "step": 129287
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.039980411529541,
      "learning_rate": 0.00024273193761168913,
      "loss": 1.9096,
      "step": 129288
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.0444674491882324,
      "learning_rate": 0.00024272792227722937,
      "loss": 1.9341,
      "step": 129289
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.187239408493042,
      "learning_rate": 0.0002427239069534176,
      "loss": 1.9194,
      "step": 129290
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.0107039213180542,
      "learning_rate": 0.0002427198916402546,
      "loss": 2.2117,
      "step": 129291
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.0489552021026611,
      "learning_rate": 0.00024271587633774096,
      "loss": 2.2358,
      "step": 129292
    },
    {
      "epoch": 1.68,
      "grad_norm": 0.9958997368812561,
      "learning_rate": 0.00024271186104587756,
      "loss": 2.2669,
      "step": 129293
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.079463005065918,
      "learning_rate": 0.00024270784576466508,
      "loss": 2.029,
      "step": 129294
    },
    {
      "epoch": 1.68,
      "grad_norm": 0.9921965003013611,
      "learning_rate": 0.0002427038304941043,
      "loss": 2.2133,
      "step": 129295
    },
    {
      "epoch": 1.68,
      "grad_norm": 0.8776415586471558,
      "learning_rate": 0.00024269981523419604,
      "loss": 1.9516,
      "step": 129296
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.0408005714416504,
      "learning_rate": 0.00024269579998494094,
      "loss": 2.1302,
      "step": 129297
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.0274337530136108,
      "learning_rate": 0.00024269178474633975,
      "loss": 2.175,
      "step": 129298
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.256096363067627,
      "learning_rate": 0.00024268776951839318,
      "loss": 2.2134,
      "step": 129299
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.063186526298523,
      "learning_rate": 0.0002426837543011021,
      "loss": 2.403,
      "step": 129300
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.04574716091156,
      "learning_rate": 0.00024267973909446712,
      "loss": 2.2033,
      "step": 129301
    },
    {
      "epoch": 1.68,
      "grad_norm": 0.9824042916297913,
      "learning_rate": 0.00024267572389848912,
      "loss": 2.2888,
      "step": 129302
    },
    {
      "epoch": 1.68,
      "grad_norm": 0.863968014717102,
      "learning_rate": 0.0002426717087131689,
      "loss": 2.0327,
      "step": 129303
    },
    {
      "epoch": 1.68,
      "grad_norm": 0.9936009645462036,
      "learning_rate": 0.00024266769353850694,
      "loss": 2.3344,
      "step": 129304
    },
    {
      "epoch": 1.68,
      "grad_norm": 0.9792426824569702,
      "learning_rate": 0.00024266367837450414,
      "loss": 1.9215,
      "step": 129305
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.1278793811798096,
      "learning_rate": 0.00024265966322116128,
      "loss": 2.2337,
      "step": 129306
    },
    {
      "epoch": 1.68,
      "grad_norm": 0.9163957238197327,
      "learning_rate": 0.00024265564807847902,
      "loss": 2.0823,
      "step": 129307
    },
    {
      "epoch": 1.68,
      "grad_norm": 0.9842844009399414,
      "learning_rate": 0.0002426516329464582,
      "loss": 2.0859,
      "step": 129308
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.1118907928466797,
      "learning_rate": 0.00024264761782509962,
      "loss": 2.0102,
      "step": 129309
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.1474274396896362,
      "learning_rate": 0.00024264360271440385,
      "loss": 2.0046,
      "step": 129310
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.0440897941589355,
      "learning_rate": 0.00024263958761437166,
      "loss": 2.0855,
      "step": 129311
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.0415178537368774,
      "learning_rate": 0.00024263557252500388,
      "loss": 2.1168,
      "step": 129312
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.0247396230697632,
      "learning_rate": 0.00024263155744630124,
      "loss": 1.884,
      "step": 129313
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.0192782878875732,
      "learning_rate": 0.00024262754237826446,
      "loss": 2.1501,
      "step": 129314
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.0392897129058838,
      "learning_rate": 0.0002426235273208943,
      "loss": 2.3623,
      "step": 129315
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.0146244764328003,
      "learning_rate": 0.00024261951227419164,
      "loss": 1.8977,
      "step": 129316
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.0806816816329956,
      "learning_rate": 0.000242615497238157,
      "loss": 1.9864,
      "step": 129317
    },
    {
      "epoch": 1.68,
      "grad_norm": 0.9934613108634949,
      "learning_rate": 0.00024261148221279118,
      "loss": 2.2542,
      "step": 129318
    },
    {
      "epoch": 1.68,
      "grad_norm": 0.9965841174125671,
      "learning_rate": 0.00024260746719809495,
      "loss": 2.302,
      "step": 129319
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.0222374200820923,
      "learning_rate": 0.00024260345219406914,
      "loss": 2.2124,
      "step": 129320
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.3401600122451782,
      "learning_rate": 0.0002425994372007144,
      "loss": 2.3099,
      "step": 129321
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.0460631847381592,
      "learning_rate": 0.00024259542221803163,
      "loss": 2.1307,
      "step": 129322
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.0080581903457642,
      "learning_rate": 0.00024259140724602137,
      "loss": 2.1548,
      "step": 129323
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.0624055862426758,
      "learning_rate": 0.0002425873922846844,
      "loss": 2.1288,
      "step": 129324
    },
    {
      "epoch": 1.68,
      "grad_norm": 0.9542551040649414,
      "learning_rate": 0.00024258337733402152,
      "loss": 1.9631,
      "step": 129325
    },
    {
      "epoch": 1.68,
      "grad_norm": 0.9719880819320679,
      "learning_rate": 0.0002425793623940335,
      "loss": 2.2602,
      "step": 129326
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.000836968421936,
      "learning_rate": 0.00024257534746472103,
      "loss": 2.0597,
      "step": 129327
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.029160499572754,
      "learning_rate": 0.00024257133254608506,
      "loss": 2.262,
      "step": 129328
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.0539096593856812,
      "learning_rate": 0.000242567317638126,
      "loss": 2.0191,
      "step": 129329
    },
    {
      "epoch": 1.68,
      "grad_norm": 0.9609575867652893,
      "learning_rate": 0.00024256330274084476,
      "loss": 1.9768,
      "step": 129330
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.0092780590057373,
      "learning_rate": 0.0002425592878542421,
      "loss": 2.3138,
      "step": 129331
    },
    {
      "epoch": 1.68,
      "grad_norm": 0.9808787107467651,
      "learning_rate": 0.00024255527297831873,
      "loss": 2.2895,
      "step": 129332
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.0198194980621338,
      "learning_rate": 0.00024255125811307543,
      "loss": 2.0127,
      "step": 129333
    },
    {
      "epoch": 1.68,
      "grad_norm": 0.9570065140724182,
      "learning_rate": 0.0002425472432585131,
      "loss": 1.8867,
      "step": 129334
    },
    {
      "epoch": 1.68,
      "grad_norm": 0.8859587907791138,
      "learning_rate": 0.00024254322841463214,
      "loss": 1.9426,
      "step": 129335
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.0615406036376953,
      "learning_rate": 0.00024253921358143352,
      "loss": 2.1782,
      "step": 129336
    },
    {
      "epoch": 1.68,
      "grad_norm": 0.9989951848983765,
      "learning_rate": 0.0002425351987589179,
      "loss": 2.1169,
      "step": 129337
    },
    {
      "epoch": 1.68,
      "grad_norm": 0.8994553089141846,
      "learning_rate": 0.0002425311839470861,
      "loss": 1.9505,
      "step": 129338
    },
    {
      "epoch": 1.68,
      "grad_norm": 0.9704516530036926,
      "learning_rate": 0.00024252716914593882,
      "loss": 2.2106,
      "step": 129339
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.2590707540512085,
      "learning_rate": 0.000242523154355477,
      "loss": 1.9815,
      "step": 129340
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.0940356254577637,
      "learning_rate": 0.00024251913957570103,
      "loss": 2.151,
      "step": 129341
    },
    {
      "epoch": 1.68,
      "grad_norm": 0.9938810467720032,
      "learning_rate": 0.00024251512480661185,
      "loss": 2.0865,
      "step": 129342
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.2806178331375122,
      "learning_rate": 0.0002425111100482102,
      "loss": 2.1854,
      "step": 129343
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.0468688011169434,
      "learning_rate": 0.0002425070953004968,
      "loss": 2.1258,
      "step": 129344
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.0035618543624878,
      "learning_rate": 0.00024250308056347246,
      "loss": 2.0615,
      "step": 129345
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.2805588245391846,
      "learning_rate": 0.00024249906583713798,
      "loss": 2.0512,
      "step": 129346
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.1469428539276123,
      "learning_rate": 0.00024249505112149388,
      "loss": 2.0509,
      "step": 129347
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.0739307403564453,
      "learning_rate": 0.00024249103641654104,
      "loss": 2.0968,
      "step": 129348
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.0674768686294556,
      "learning_rate": 0.0002424870217222802,
      "loss": 2.1578,
      "step": 129349
    },
    {
      "epoch": 1.68,
      "grad_norm": 0.9950128793716431,
      "learning_rate": 0.00024248300703871216,
      "loss": 1.8867,
      "step": 129350
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.3416997194290161,
      "learning_rate": 0.00024247899236583755,
      "loss": 2.136,
      "step": 129351
    },
    {
      "epoch": 1.68,
      "grad_norm": 0.9967358708381653,
      "learning_rate": 0.0002424749777036573,
      "loss": 2.0957,
      "step": 129352
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.0229555368423462,
      "learning_rate": 0.00024247096305217197,
      "loss": 2.3569,
      "step": 129353
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.0580388307571411,
      "learning_rate": 0.00024246694841138233,
      "loss": 2.4947,
      "step": 129354
    },
    {
      "epoch": 1.68,
      "grad_norm": 0.9416811466217041,
      "learning_rate": 0.00024246293378128915,
      "loss": 2.0841,
      "step": 129355
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.1372720003128052,
      "learning_rate": 0.00024245891916189325,
      "loss": 2.1828,
      "step": 129356
    },
    {
      "epoch": 1.68,
      "grad_norm": 0.9820276498794556,
      "learning_rate": 0.0002424549045531953,
      "loss": 1.9648,
      "step": 129357
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.077782392501831,
      "learning_rate": 0.00024245088995519614,
      "loss": 2.1651,
      "step": 129358
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.0259382724761963,
      "learning_rate": 0.00024244687536789644,
      "loss": 2.1115,
      "step": 129359
    },
    {
      "epoch": 1.68,
      "grad_norm": 0.9397900700569153,
      "learning_rate": 0.00024244286079129688,
      "loss": 2.0995,
      "step": 129360
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.0519318580627441,
      "learning_rate": 0.0002424388462253983,
      "loss": 2.0712,
      "step": 129361
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.1097394227981567,
      "learning_rate": 0.00024243483167020138,
      "loss": 2.1097,
      "step": 129362
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.1336982250213623,
      "learning_rate": 0.00024243081712570696,
      "loss": 2.2881,
      "step": 129363
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.070993185043335,
      "learning_rate": 0.0002424268025919158,
      "loss": 2.2124,
      "step": 129364
    },
    {
      "epoch": 1.68,
      "grad_norm": 0.9874458312988281,
      "learning_rate": 0.0002424227880688285,
      "loss": 2.2297,
      "step": 129365
    },
    {
      "epoch": 1.68,
      "grad_norm": 0.9786127209663391,
      "learning_rate": 0.00024241877355644595,
      "loss": 2.0725,
      "step": 129366
    },
    {
      "epoch": 1.68,
      "grad_norm": 0.9560614228248596,
      "learning_rate": 0.0002424147590547688,
      "loss": 2.1666,
      "step": 129367
    },
    {
      "epoch": 1.68,
      "grad_norm": 0.970164954662323,
      "learning_rate": 0.0002424107445637978,
      "loss": 2.0997,
      "step": 129368
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.0539164543151855,
      "learning_rate": 0.00024240673008353377,
      "loss": 2.1265,
      "step": 129369
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.049522876739502,
      "learning_rate": 0.00024240271561397746,
      "loss": 2.1707,
      "step": 129370
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.0107368230819702,
      "learning_rate": 0.00024239870115512946,
      "loss": 2.2304,
      "step": 129371
    },
    {
      "epoch": 1.68,
      "grad_norm": 0.9582156538963318,
      "learning_rate": 0.0002423946867069907,
      "loss": 1.8937,
      "step": 129372
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.0547401905059814,
      "learning_rate": 0.0002423906722695619,
      "loss": 2.1546,
      "step": 129373
    },
    {
      "epoch": 1.68,
      "grad_norm": 0.8617438077926636,
      "learning_rate": 0.0002423866578428437,
      "loss": 2.0015,
      "step": 129374
    },
    {
      "epoch": 1.68,
      "grad_norm": 0.9810014367103577,
      "learning_rate": 0.00024238264342683687,
      "loss": 2.0749,
      "step": 129375
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.0643503665924072,
      "learning_rate": 0.0002423786290215423,
      "loss": 2.1248,
      "step": 129376
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.0111550092697144,
      "learning_rate": 0.00024237461462696055,
      "loss": 2.097,
      "step": 129377
    },
    {
      "epoch": 1.68,
      "grad_norm": 0.9845476150512695,
      "learning_rate": 0.00024237060024309242,
      "loss": 2.2259,
      "step": 129378
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.0049513578414917,
      "learning_rate": 0.00024236658586993873,
      "loss": 2.2604,
      "step": 129379
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.073296070098877,
      "learning_rate": 0.00024236257150750024,
      "loss": 2.2901,
      "step": 129380
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.0602010488510132,
      "learning_rate": 0.00024235855715577757,
      "loss": 1.9996,
      "step": 129381
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.0681328773498535,
      "learning_rate": 0.0002423545428147715,
      "loss": 1.9312,
      "step": 129382
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.0080984830856323,
      "learning_rate": 0.0002423505284844829,
      "loss": 2.0475,
      "step": 129383
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.0550371408462524,
      "learning_rate": 0.00024234651416491236,
      "loss": 2.2174,
      "step": 129384
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.0241725444793701,
      "learning_rate": 0.0002423424998560607,
      "loss": 2.4161,
      "step": 129385
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.0188865661621094,
      "learning_rate": 0.00024233848555792862,
      "loss": 2.0453,
      "step": 129386
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.3306095600128174,
      "learning_rate": 0.00024233447127051692,
      "loss": 2.2432,
      "step": 129387
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.0257067680358887,
      "learning_rate": 0.00024233045699382637,
      "loss": 2.1324,
      "step": 129388
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.0505421161651611,
      "learning_rate": 0.00024232644272785775,
      "loss": 2.3001,
      "step": 129389
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.016878604888916,
      "learning_rate": 0.00024232242847261161,
      "loss": 2.1993,
      "step": 129390
    },
    {
      "epoch": 1.68,
      "grad_norm": 0.9905757904052734,
      "learning_rate": 0.00024231841422808886,
      "loss": 2.3799,
      "step": 129391
    },
    {
      "epoch": 1.68,
      "grad_norm": 0.9833759069442749,
      "learning_rate": 0.00024231439999429014,
      "loss": 2.0011,
      "step": 129392
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.0061591863632202,
      "learning_rate": 0.0002423103857712163,
      "loss": 2.1302,
      "step": 129393
    },
    {
      "epoch": 1.68,
      "grad_norm": 0.936851441860199,
      "learning_rate": 0.00024230637155886809,
      "loss": 2.0559,
      "step": 129394
    },
    {
      "epoch": 1.68,
      "grad_norm": 0.9939318299293518,
      "learning_rate": 0.0002423023573572463,
      "loss": 2.2276,
      "step": 129395
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.0246515274047852,
      "learning_rate": 0.00024229834316635146,
      "loss": 2.2118,
      "step": 129396
    },
    {
      "epoch": 1.68,
      "grad_norm": 0.9456322193145752,
      "learning_rate": 0.00024229432898618445,
      "loss": 2.0517,
      "step": 129397
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.0005189180374146,
      "learning_rate": 0.000242290314816746,
      "loss": 2.1026,
      "step": 129398
    },
    {
      "epoch": 1.68,
      "grad_norm": 0.996244490146637,
      "learning_rate": 0.0002422863006580369,
      "loss": 2.2138,
      "step": 129399
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.1306589841842651,
      "learning_rate": 0.00024228228651005786,
      "loss": 2.1828,
      "step": 129400
    },
    {
      "epoch": 1.68,
      "grad_norm": 0.9675197005271912,
      "learning_rate": 0.00024227827237280976,
      "loss": 2.2779,
      "step": 129401
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.0853554010391235,
      "learning_rate": 0.0002422742582462931,
      "loss": 1.9932,
      "step": 129402
    },
    {
      "epoch": 1.68,
      "grad_norm": 0.9839298129081726,
      "learning_rate": 0.00024227024413050872,
      "loss": 2.0984,
      "step": 129403
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.027451992034912,
      "learning_rate": 0.0002422662300254574,
      "loss": 2.0274,
      "step": 129404
    },
    {
      "epoch": 1.68,
      "grad_norm": 0.968392550945282,
      "learning_rate": 0.00024226221593113986,
      "loss": 2.0048,
      "step": 129405
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.010390281677246,
      "learning_rate": 0.0002422582018475569,
      "loss": 2.1708,
      "step": 129406
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.0033996105194092,
      "learning_rate": 0.00024225418777470934,
      "loss": 2.2765,
      "step": 129407
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.1290223598480225,
      "learning_rate": 0.00024225017371259767,
      "loss": 2.159,
      "step": 129408
    },
    {
      "epoch": 1.68,
      "grad_norm": 1.0163295269012451,
      "learning_rate": 0.00024224615966122283,
      "loss": 2.1895,
      "step": 129409
    },
    {
      "epoch": 1.69,
      "grad_norm": 0.9384570717811584,
      "learning_rate": 0.00024224214562058548,
      "loss": 2.0421,
      "step": 129410
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.1238701343536377,
      "learning_rate": 0.0002422381315906864,
      "loss": 1.9555,
      "step": 129411
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.0565553903579712,
      "learning_rate": 0.00024223411757152638,
      "loss": 2.171,
      "step": 129412
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.0267921686172485,
      "learning_rate": 0.00024223010356310624,
      "loss": 2.2362,
      "step": 129413
    },
    {
      "epoch": 1.69,
      "grad_norm": 0.9467254877090454,
      "learning_rate": 0.00024222608956542645,
      "loss": 1.9667,
      "step": 129414
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.1792150735855103,
      "learning_rate": 0.00024222207557848797,
      "loss": 2.0272,
      "step": 129415
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.0213161706924438,
      "learning_rate": 0.00024221806160229148,
      "loss": 2.1783,
      "step": 129416
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.073614239692688,
      "learning_rate": 0.00024221404763683776,
      "loss": 2.4151,
      "step": 129417
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.358457088470459,
      "learning_rate": 0.0002422100336821275,
      "loss": 2.1179,
      "step": 129418
    },
    {
      "epoch": 1.69,
      "grad_norm": 0.9800297021865845,
      "learning_rate": 0.00024220601973816162,
      "loss": 2.167,
      "step": 129419
    },
    {
      "epoch": 1.69,
      "grad_norm": 0.9936290383338928,
      "learning_rate": 0.00024220200580494063,
      "loss": 2.1963,
      "step": 129420
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.0524640083312988,
      "learning_rate": 0.00024219799188246536,
      "loss": 2.0369,
      "step": 129421
    },
    {
      "epoch": 1.69,
      "grad_norm": 0.9833470582962036,
      "learning_rate": 0.0002421939779707366,
      "loss": 1.9951,
      "step": 129422
    },
    {
      "epoch": 1.69,
      "grad_norm": 0.957926332950592,
      "learning_rate": 0.00024218996406975504,
      "loss": 2.045,
      "step": 129423
    },
    {
      "epoch": 1.69,
      "grad_norm": 0.9231647849082947,
      "learning_rate": 0.00024218595017952148,
      "loss": 2.1423,
      "step": 129424
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.0416432619094849,
      "learning_rate": 0.00024218193630003677,
      "loss": 2.0635,
      "step": 129425
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.1916474103927612,
      "learning_rate": 0.00024217792243130137,
      "loss": 2.1029,
      "step": 129426
    },
    {
      "epoch": 1.69,
      "grad_norm": 0.9971480965614319,
      "learning_rate": 0.00024217390857331623,
      "loss": 2.1125,
      "step": 129427
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.1072797775268555,
      "learning_rate": 0.000242169894726082,
      "loss": 2.0222,
      "step": 129428
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.226015567779541,
      "learning_rate": 0.00024216588088959951,
      "loss": 2.2782,
      "step": 129429
    },
    {
      "epoch": 1.69,
      "grad_norm": 0.9428225159645081,
      "learning_rate": 0.00024216186706386948,
      "loss": 1.9984,
      "step": 129430
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.0769579410552979,
      "learning_rate": 0.00024215785324889277,
      "loss": 2.3807,
      "step": 129431
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.1991496086120605,
      "learning_rate": 0.0002421538394446699,
      "loss": 2.2203,
      "step": 129432
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.1252267360687256,
      "learning_rate": 0.0002421498256512017,
      "loss": 2.1066,
      "step": 129433
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.0978714227676392,
      "learning_rate": 0.00024214581186848892,
      "loss": 2.0153,
      "step": 129434
    },
    {
      "epoch": 1.69,
      "grad_norm": 0.9147141575813293,
      "learning_rate": 0.00024214179809653235,
      "loss": 2.3542,
      "step": 129435
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.204656720161438,
      "learning_rate": 0.00024213778433533273,
      "loss": 2.3404,
      "step": 129436
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.086013913154602,
      "learning_rate": 0.00024213377058489088,
      "loss": 2.2697,
      "step": 129437
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.0262720584869385,
      "learning_rate": 0.00024212975684520733,
      "loss": 2.2351,
      "step": 129438
    },
    {
      "epoch": 1.69,
      "grad_norm": 0.9966897964477539,
      "learning_rate": 0.00024212574311628296,
      "loss": 2.2215,
      "step": 129439
    },
    {
      "epoch": 1.69,
      "grad_norm": 0.9570540189743042,
      "learning_rate": 0.00024212172939811855,
      "loss": 2.1095,
      "step": 129440
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.2255898714065552,
      "learning_rate": 0.00024211771569071473,
      "loss": 2.0677,
      "step": 129441
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.0959267616271973,
      "learning_rate": 0.00024211370199407235,
      "loss": 2.034,
      "step": 129442
    },
    {
      "epoch": 1.69,
      "grad_norm": 0.971932590007782,
      "learning_rate": 0.00024210968830819227,
      "loss": 2.3463,
      "step": 129443
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.0470401048660278,
      "learning_rate": 0.00024210567463307494,
      "loss": 2.002,
      "step": 129444
    },
    {
      "epoch": 1.69,
      "grad_norm": 0.9244969487190247,
      "learning_rate": 0.00024210166096872126,
      "loss": 2.0654,
      "step": 129445
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.3795771598815918,
      "learning_rate": 0.00024209764731513198,
      "loss": 2.1186,
      "step": 129446
    },
    {
      "epoch": 1.69,
      "grad_norm": 0.9792100191116333,
      "learning_rate": 0.00024209363367230785,
      "loss": 1.9142,
      "step": 129447
    },
    {
      "epoch": 1.69,
      "grad_norm": 0.9232221841812134,
      "learning_rate": 0.0002420896200402496,
      "loss": 2.2378,
      "step": 129448
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.140129566192627,
      "learning_rate": 0.00024208560641895797,
      "loss": 2.3006,
      "step": 129449
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.159204125404358,
      "learning_rate": 0.00024208159280843385,
      "loss": 2.121,
      "step": 129450
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.1768360137939453,
      "learning_rate": 0.00024207757920867776,
      "loss": 2.2288,
      "step": 129451
    },
    {
      "epoch": 1.69,
      "grad_norm": 0.9404025673866272,
      "learning_rate": 0.0002420735656196905,
      "loss": 2.1492,
      "step": 129452
    },
    {
      "epoch": 1.69,
      "grad_norm": 0.9781301021575928,
      "learning_rate": 0.0002420695520414729,
      "loss": 2.1498,
      "step": 129453
    },
    {
      "epoch": 1.69,
      "grad_norm": 0.9673189520835876,
      "learning_rate": 0.0002420655384740256,
      "loss": 2.232,
      "step": 129454
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.1856911182403564,
      "learning_rate": 0.00024206152491734945,
      "loss": 2.2056,
      "step": 129455
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.1603292226791382,
      "learning_rate": 0.00024205751137144525,
      "loss": 2.0423,
      "step": 129456
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.0722202062606812,
      "learning_rate": 0.00024205349783631354,
      "loss": 1.9758,
      "step": 129457
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.2240009307861328,
      "learning_rate": 0.00024204948431195524,
      "loss": 2.2421,
      "step": 129458
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.0831161737442017,
      "learning_rate": 0.000242045470798371,
      "loss": 2.0705,
      "step": 129459
    },
    {
      "epoch": 1.69,
      "grad_norm": 0.9822691679000854,
      "learning_rate": 0.0002420414572955616,
      "loss": 2.0977,
      "step": 129460
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.0172733068466187,
      "learning_rate": 0.00024203744380352777,
      "loss": 1.9419,
      "step": 129461
    },
    {
      "epoch": 1.69,
      "grad_norm": 0.8765146136283875,
      "learning_rate": 0.00024203343032227034,
      "loss": 1.9174,
      "step": 129462
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.0360013246536255,
      "learning_rate": 0.00024202941685178993,
      "loss": 2.3767,
      "step": 129463
    },
    {
      "epoch": 1.69,
      "grad_norm": 0.9156076908111572,
      "learning_rate": 0.00024202540339208733,
      "loss": 2.1971,
      "step": 129464
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.0352317094802856,
      "learning_rate": 0.0002420213899431634,
      "loss": 1.8735,
      "step": 129465
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.0441036224365234,
      "learning_rate": 0.0002420173765050187,
      "loss": 2.109,
      "step": 129466
    },
    {
      "epoch": 1.69,
      "grad_norm": 0.9634712338447571,
      "learning_rate": 0.00024201336307765406,
      "loss": 2.1877,
      "step": 129467
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.1223589181900024,
      "learning_rate": 0.00024200934966107034,
      "loss": 1.9392,
      "step": 129468
    },
    {
      "epoch": 1.69,
      "grad_norm": 0.995813250541687,
      "learning_rate": 0.00024200533625526804,
      "loss": 2.1907,
      "step": 129469
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.181188702583313,
      "learning_rate": 0.00024200132286024807,
      "loss": 2.1445,
      "step": 129470
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.0357871055603027,
      "learning_rate": 0.00024199730947601118,
      "loss": 2.1842,
      "step": 129471
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.0606409311294556,
      "learning_rate": 0.00024199329610255813,
      "loss": 2.0502,
      "step": 129472
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.2495417594909668,
      "learning_rate": 0.00024198928273988954,
      "loss": 2.217,
      "step": 129473
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.0821895599365234,
      "learning_rate": 0.00024198526938800635,
      "loss": 2.3479,
      "step": 129474
    },
    {
      "epoch": 1.69,
      "grad_norm": 0.9785723686218262,
      "learning_rate": 0.00024198125604690904,
      "loss": 2.0391,
      "step": 129475
    },
    {
      "epoch": 1.69,
      "grad_norm": 0.9724469780921936,
      "learning_rate": 0.0002419772427165986,
      "loss": 2.0347,
      "step": 129476
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.0281322002410889,
      "learning_rate": 0.00024197322939707567,
      "loss": 2.0812,
      "step": 129477
    },
    {
      "epoch": 1.69,
      "grad_norm": 0.9699280858039856,
      "learning_rate": 0.000241969216088341,
      "loss": 2.3548,
      "step": 129478
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.037437081336975,
      "learning_rate": 0.0002419652027903954,
      "loss": 2.2017,
      "step": 129479
    },
    {
      "epoch": 1.69,
      "grad_norm": 0.9667655825614929,
      "learning_rate": 0.00024196118950323955,
      "loss": 2.1673,
      "step": 129480
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.1300005912780762,
      "learning_rate": 0.00024195717622687419,
      "loss": 2.1716,
      "step": 129481
    },
    {
      "epoch": 1.69,
      "grad_norm": 0.9392702579498291,
      "learning_rate": 0.00024195316296130005,
      "loss": 2.0936,
      "step": 129482
    },
    {
      "epoch": 1.69,
      "grad_norm": 0.9877254962921143,
      "learning_rate": 0.00024194914970651792,
      "loss": 2.0399,
      "step": 129483
    },
    {
      "epoch": 1.69,
      "grad_norm": 0.9335678219795227,
      "learning_rate": 0.00024194513646252856,
      "loss": 2.0675,
      "step": 129484
    },
    {
      "epoch": 1.69,
      "grad_norm": 0.9254547357559204,
      "learning_rate": 0.00024194112322933266,
      "loss": 1.9668,
      "step": 129485
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.0413005352020264,
      "learning_rate": 0.00024193711000693115,
      "loss": 2.187,
      "step": 129486
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.248733639717102,
      "learning_rate": 0.0002419330967953245,
      "loss": 2.1426,
      "step": 129487
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.0417546033859253,
      "learning_rate": 0.00024192908359451358,
      "loss": 2.0849,
      "step": 129488
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.0044087171554565,
      "learning_rate": 0.00024192507040449914,
      "loss": 1.9528,
      "step": 129489
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.1187965869903564,
      "learning_rate": 0.0002419210572252819,
      "loss": 2.1832,
      "step": 129490
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.0274699926376343,
      "learning_rate": 0.00024191704405686267,
      "loss": 2.1096,
      "step": 129491
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.1065062284469604,
      "learning_rate": 0.0002419130308992423,
      "loss": 1.8456,
      "step": 129492
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.0128133296966553,
      "learning_rate": 0.00024190901775242125,
      "loss": 2.2265,
      "step": 129493
    },
    {
      "epoch": 1.69,
      "grad_norm": 0.990280032157898,
      "learning_rate": 0.00024190500461640038,
      "loss": 2.2231,
      "step": 129494
    },
    {
      "epoch": 1.69,
      "grad_norm": 0.9662560224533081,
      "learning_rate": 0.00024190099149118047,
      "loss": 2.2125,
      "step": 129495
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.0048891305923462,
      "learning_rate": 0.0002418969783767623,
      "loss": 2.1719,
      "step": 129496
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.0182442665100098,
      "learning_rate": 0.00024189296527314655,
      "loss": 2.0626,
      "step": 129497
    },
    {
      "epoch": 1.69,
      "grad_norm": 0.9996538758277893,
      "learning_rate": 0.00024188895218033412,
      "loss": 2.4029,
      "step": 129498
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.0006954669952393,
      "learning_rate": 0.00024188493909832551,
      "loss": 2.2278,
      "step": 129499
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.3329914808273315,
      "learning_rate": 0.0002418809260271216,
      "loss": 1.8299,
      "step": 129500
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.0261635780334473,
      "learning_rate": 0.0002418769129667231,
      "loss": 2.3452,
      "step": 129501
    },
    {
      "epoch": 1.69,
      "grad_norm": 0.9304203987121582,
      "learning_rate": 0.00024187289991713082,
      "loss": 2.1637,
      "step": 129502
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.0417118072509766,
      "learning_rate": 0.0002418688868783454,
      "loss": 2.0405,
      "step": 129503
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.1055232286453247,
      "learning_rate": 0.00024186487385036783,
      "loss": 2.121,
      "step": 129504
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.095735788345337,
      "learning_rate": 0.00024186086083319855,
      "loss": 2.2537,
      "step": 129505
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.100789189338684,
      "learning_rate": 0.0002418568478268384,
      "loss": 2.1391,
      "step": 129506
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.0708554983139038,
      "learning_rate": 0.00024185283483128817,
      "loss": 1.941,
      "step": 129507
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.04667067527771,
      "learning_rate": 0.00024184882184654863,
      "loss": 2.0971,
      "step": 129508
    },
    {
      "epoch": 1.69,
      "grad_norm": 0.885621190071106,
      "learning_rate": 0.00024184480887262042,
      "loss": 1.9453,
      "step": 129509
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.1299407482147217,
      "learning_rate": 0.00024184079590950456,
      "loss": 2.1998,
      "step": 129510
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.3856892585754395,
      "learning_rate": 0.00024183678295720143,
      "loss": 1.8989,
      "step": 129511
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.096470594406128,
      "learning_rate": 0.00024183277001571196,
      "loss": 2.0621,
      "step": 129512
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.297886610031128,
      "learning_rate": 0.00024182875708503686,
      "loss": 2.2772,
      "step": 129513
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.2690415382385254,
      "learning_rate": 0.0002418247441651769,
      "loss": 2.2517,
      "step": 129514
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.7345516681671143,
      "learning_rate": 0.0002418207312561328,
      "loss": 1.9642,
      "step": 129515
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.2645140886306763,
      "learning_rate": 0.00024181671835790534,
      "loss": 2.0671,
      "step": 129516
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.0157924890518188,
      "learning_rate": 0.00024181270547049538,
      "loss": 2.0514,
      "step": 129517
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.4256905317306519,
      "learning_rate": 0.0002418086925939034,
      "loss": 2.1414,
      "step": 129518
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.1347075700759888,
      "learning_rate": 0.0002418046797281303,
      "loss": 2.1794,
      "step": 129519
    },
    {
      "epoch": 1.69,
      "grad_norm": 0.9621886610984802,
      "learning_rate": 0.00024180066687317682,
      "loss": 1.8952,
      "step": 129520
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.01167893409729,
      "learning_rate": 0.0002417966540290436,
      "loss": 2.2065,
      "step": 129521
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.3456190824508667,
      "learning_rate": 0.00024179264119573158,
      "loss": 2.09,
      "step": 129522
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.0889183282852173,
      "learning_rate": 0.0002417886283732415,
      "loss": 2.1207,
      "step": 129523
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.1162844896316528,
      "learning_rate": 0.0002417846155615739,
      "loss": 2.0767,
      "step": 129524
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.0297402143478394,
      "learning_rate": 0.00024178060276072963,
      "loss": 2.0352,
      "step": 129525
    },
    {
      "epoch": 1.69,
      "grad_norm": 0.9396093487739563,
      "learning_rate": 0.00024177658997070946,
      "loss": 1.9692,
      "step": 129526
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.070802927017212,
      "learning_rate": 0.00024177257719151408,
      "loss": 2.2271,
      "step": 129527
    },
    {
      "epoch": 1.69,
      "grad_norm": 0.8994960188865662,
      "learning_rate": 0.0002417685644231443,
      "loss": 2.0635,
      "step": 129528
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.0061111450195312,
      "learning_rate": 0.000241764551665601,
      "loss": 2.2327,
      "step": 129529
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.0843931436538696,
      "learning_rate": 0.0002417605389188846,
      "loss": 2.367,
      "step": 129530
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.1519402265548706,
      "learning_rate": 0.00024175652618299605,
      "loss": 2.1404,
      "step": 129531
    },
    {
      "epoch": 1.69,
      "grad_norm": 0.9406995177268982,
      "learning_rate": 0.00024175251345793603,
      "loss": 2.1492,
      "step": 129532
    },
    {
      "epoch": 1.69,
      "grad_norm": 0.9703140258789062,
      "learning_rate": 0.0002417485007437053,
      "loss": 2.1419,
      "step": 129533
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.0524059534072876,
      "learning_rate": 0.00024174448804030466,
      "loss": 2.1237,
      "step": 129534
    },
    {
      "epoch": 1.69,
      "grad_norm": 0.9937947988510132,
      "learning_rate": 0.00024174047534773497,
      "loss": 2.0956,
      "step": 129535
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.0391925573349,
      "learning_rate": 0.00024173646266599665,
      "loss": 1.9668,
      "step": 129536
    },
    {
      "epoch": 1.69,
      "grad_norm": 0.9334884285926819,
      "learning_rate": 0.00024173244999509065,
      "loss": 2.0585,
      "step": 129537
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.0100429058074951,
      "learning_rate": 0.00024172843733501765,
      "loss": 2.2335,
      "step": 129538
    },
    {
      "epoch": 1.69,
      "grad_norm": 0.9900743961334229,
      "learning_rate": 0.00024172442468577848,
      "loss": 2.2124,
      "step": 129539
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.0274025201797485,
      "learning_rate": 0.00024172041204737382,
      "loss": 2.0624,
      "step": 129540
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.0673598051071167,
      "learning_rate": 0.00024171639941980447,
      "loss": 1.9941,
      "step": 129541
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.029695987701416,
      "learning_rate": 0.0002417123868030712,
      "loss": 2.0267,
      "step": 129542
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.0205554962158203,
      "learning_rate": 0.00024170837419717456,
      "loss": 2.0563,
      "step": 129543
    },
    {
      "epoch": 1.69,
      "grad_norm": 0.9308203458786011,
      "learning_rate": 0.0002417043616021155,
      "loss": 2.0335,
      "step": 129544
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.0398368835449219,
      "learning_rate": 0.00024170034901789466,
      "loss": 1.9983,
      "step": 129545
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.2514079809188843,
      "learning_rate": 0.00024169633644451286,
      "loss": 2.0108,
      "step": 129546
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.1461906433105469,
      "learning_rate": 0.00024169232388197086,
      "loss": 2.0814,
      "step": 129547
    },
    {
      "epoch": 1.69,
      "grad_norm": 0.9091262817382812,
      "learning_rate": 0.00024168831133026928,
      "loss": 1.9462,
      "step": 129548
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.408097505569458,
      "learning_rate": 0.00024168429878940894,
      "loss": 2.0249,
      "step": 129549
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.0226949453353882,
      "learning_rate": 0.0002416802862593907,
      "loss": 2.2002,
      "step": 129550
    },
    {
      "epoch": 1.69,
      "grad_norm": 0.9378219246864319,
      "learning_rate": 0.0002416762737402151,
      "loss": 2.2174,
      "step": 129551
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.0345544815063477,
      "learning_rate": 0.00024167226123188295,
      "loss": 2.1139,
      "step": 129552
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.0360280275344849,
      "learning_rate": 0.00024166824873439515,
      "loss": 2.0812,
      "step": 129553
    },
    {
      "epoch": 1.69,
      "grad_norm": 0.9814385771751404,
      "learning_rate": 0.00024166423624775223,
      "loss": 2.113,
      "step": 129554
    },
    {
      "epoch": 1.69,
      "grad_norm": 0.9391029477119446,
      "learning_rate": 0.00024166022377195505,
      "loss": 2.2105,
      "step": 129555
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.0868666172027588,
      "learning_rate": 0.0002416562113070043,
      "loss": 2.2128,
      "step": 129556
    },
    {
      "epoch": 1.69,
      "grad_norm": 0.9759202003479004,
      "learning_rate": 0.00024165219885290083,
      "loss": 2.1076,
      "step": 129557
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.1361865997314453,
      "learning_rate": 0.00024164818640964525,
      "loss": 1.9787,
      "step": 129558
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.2420240640640259,
      "learning_rate": 0.0002416441739772385,
      "loss": 2.289,
      "step": 129559
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.088597297668457,
      "learning_rate": 0.00024164016155568108,
      "loss": 2.2534,
      "step": 129560
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.105722188949585,
      "learning_rate": 0.00024163614914497387,
      "loss": 2.1631,
      "step": 129561
    },
    {
      "epoch": 1.69,
      "grad_norm": 0.9787102341651917,
      "learning_rate": 0.0002416321367451176,
      "loss": 2.1263,
      "step": 129562
    },
    {
      "epoch": 1.69,
      "grad_norm": 0.9696090221405029,
      "learning_rate": 0.000241628124356113,
      "loss": 1.8657,
      "step": 129563
    },
    {
      "epoch": 1.69,
      "grad_norm": 0.9836544394493103,
      "learning_rate": 0.0002416241119779609,
      "loss": 2.023,
      "step": 129564
    },
    {
      "epoch": 1.69,
      "grad_norm": 0.9923864603042603,
      "learning_rate": 0.000241620099610662,
      "loss": 2.1843,
      "step": 129565
    },
    {
      "epoch": 1.69,
      "grad_norm": 0.9780598282814026,
      "learning_rate": 0.00024161608725421693,
      "loss": 2.0433,
      "step": 129566
    },
    {
      "epoch": 1.69,
      "grad_norm": 0.8818677663803101,
      "learning_rate": 0.0002416120749086266,
      "loss": 2.0362,
      "step": 129567
    },
    {
      "epoch": 1.69,
      "grad_norm": 0.9745773077011108,
      "learning_rate": 0.0002416080625738916,
      "loss": 2.0002,
      "step": 129568
    },
    {
      "epoch": 1.69,
      "grad_norm": 0.9767135381698608,
      "learning_rate": 0.00024160405025001284,
      "loss": 1.9905,
      "step": 129569
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.0549777746200562,
      "learning_rate": 0.00024160003793699094,
      "loss": 2.164,
      "step": 129570
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.035869836807251,
      "learning_rate": 0.00024159602563482684,
      "loss": 2.3369,
      "step": 129571
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.1236352920532227,
      "learning_rate": 0.00024159201334352103,
      "loss": 2.2652,
      "step": 129572
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.0706913471221924,
      "learning_rate": 0.00024158800106307434,
      "loss": 2.1393,
      "step": 129573
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.0914618968963623,
      "learning_rate": 0.00024158398879348754,
      "loss": 2.1385,
      "step": 129574
    },
    {
      "epoch": 1.69,
      "grad_norm": 0.9965337514877319,
      "learning_rate": 0.0002415799765347614,
      "loss": 2.0993,
      "step": 129575
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.1710344552993774,
      "learning_rate": 0.00024157596428689662,
      "loss": 2.3744,
      "step": 129576
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.0244287252426147,
      "learning_rate": 0.0002415719520498941,
      "loss": 2.2366,
      "step": 129577
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.2897831201553345,
      "learning_rate": 0.00024156793982375434,
      "loss": 2.1175,
      "step": 129578
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.046762228012085,
      "learning_rate": 0.00024156392760847823,
      "loss": 2.1772,
      "step": 129579
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.079464316368103,
      "learning_rate": 0.00024155991540406645,
      "loss": 2.1035,
      "step": 129580
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.0411303043365479,
      "learning_rate": 0.00024155590321051982,
      "loss": 2.2372,
      "step": 129581
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.028731346130371,
      "learning_rate": 0.00024155189102783903,
      "loss": 2.1122,
      "step": 129582
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.1084026098251343,
      "learning_rate": 0.00024154787885602484,
      "loss": 2.0144,
      "step": 129583
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.1141868829727173,
      "learning_rate": 0.00024154386669507816,
      "loss": 1.7823,
      "step": 129584
    },
    {
      "epoch": 1.69,
      "grad_norm": 0.9861101508140564,
      "learning_rate": 0.0002415398545449994,
      "loss": 2.1314,
      "step": 129585
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.321182370185852,
      "learning_rate": 0.00024153584240578955,
      "loss": 2.2866,
      "step": 129586
    },
    {
      "epoch": 1.69,
      "grad_norm": 0.9565774202346802,
      "learning_rate": 0.0002415318302774492,
      "loss": 1.9707,
      "step": 129587
    },
    {
      "epoch": 1.69,
      "grad_norm": 0.9666930437088013,
      "learning_rate": 0.0002415278181599793,
      "loss": 2.0577,
      "step": 129588
    },
    {
      "epoch": 1.69,
      "grad_norm": 0.9659038186073303,
      "learning_rate": 0.00024152380605338043,
      "loss": 2.0906,
      "step": 129589
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.0572717189788818,
      "learning_rate": 0.00024151979395765352,
      "loss": 2.0743,
      "step": 129590
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.3751100301742554,
      "learning_rate": 0.00024151578187279904,
      "loss": 2.0334,
      "step": 129591
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.04066002368927,
      "learning_rate": 0.00024151176979881784,
      "loss": 1.9248,
      "step": 129592
    },
    {
      "epoch": 1.69,
      "grad_norm": 0.9525853991508484,
      "learning_rate": 0.00024150775773571078,
      "loss": 2.076,
      "step": 129593
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.239662528038025,
      "learning_rate": 0.00024150374568347851,
      "loss": 1.9959,
      "step": 129594
    },
    {
      "epoch": 1.69,
      "grad_norm": 0.921883225440979,
      "learning_rate": 0.00024149973364212177,
      "loss": 2.0433,
      "step": 129595
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.0639452934265137,
      "learning_rate": 0.0002414957216116415,
      "loss": 1.9795,
      "step": 129596
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.0715187788009644,
      "learning_rate": 0.00024149170959203813,
      "loss": 2.1386,
      "step": 129597
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.0090078115463257,
      "learning_rate": 0.00024148769758331257,
      "loss": 2.2201,
      "step": 129598
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.044946551322937,
      "learning_rate": 0.00024148368558546556,
      "loss": 1.9833,
      "step": 129599
    },
    {
      "epoch": 1.69,
      "grad_norm": 0.9986746311187744,
      "learning_rate": 0.0002414796735984978,
      "loss": 2.0125,
      "step": 129600
    },
    {
      "epoch": 1.69,
      "grad_norm": 0.9605387449264526,
      "learning_rate": 0.0002414756616224101,
      "loss": 2.3007,
      "step": 129601
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.1132763624191284,
      "learning_rate": 0.0002414716496572033,
      "loss": 2.0867,
      "step": 129602
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.0103919506072998,
      "learning_rate": 0.00024146763770287787,
      "loss": 2.225,
      "step": 129603
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.150418996810913,
      "learning_rate": 0.00024146362575943475,
      "loss": 2.1641,
      "step": 129604
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.1033822298049927,
      "learning_rate": 0.00024145961382687466,
      "loss": 2.1633,
      "step": 129605
    },
    {
      "epoch": 1.69,
      "grad_norm": 0.9866805076599121,
      "learning_rate": 0.00024145560190519828,
      "loss": 2.0523,
      "step": 129606
    },
    {
      "epoch": 1.69,
      "grad_norm": 0.9982669949531555,
      "learning_rate": 0.00024145158999440644,
      "loss": 2.1464,
      "step": 129607
    },
    {
      "epoch": 1.69,
      "grad_norm": 0.9706432223320007,
      "learning_rate": 0.00024144757809450003,
      "loss": 1.9755,
      "step": 129608
    },
    {
      "epoch": 1.69,
      "grad_norm": 0.9694344997406006,
      "learning_rate": 0.00024144356620547943,
      "loss": 2.3321,
      "step": 129609
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.0322896242141724,
      "learning_rate": 0.00024143955432734559,
      "loss": 1.9607,
      "step": 129610
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.0099470615386963,
      "learning_rate": 0.00024143554246009921,
      "loss": 2.1477,
      "step": 129611
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.047595500946045,
      "learning_rate": 0.0002414315306037411,
      "loss": 2.1954,
      "step": 129612
    },
    {
      "epoch": 1.69,
      "grad_norm": 0.9530847072601318,
      "learning_rate": 0.000241427518758272,
      "loss": 2.0923,
      "step": 129613
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.4529235363006592,
      "learning_rate": 0.00024142350692369272,
      "loss": 2.1069,
      "step": 129614
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.0419563055038452,
      "learning_rate": 0.0002414194951000038,
      "loss": 2.4621,
      "step": 129615
    },
    {
      "epoch": 1.69,
      "grad_norm": 0.9843024611473083,
      "learning_rate": 0.0002414154832872061,
      "loss": 2.2094,
      "step": 129616
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.043188452720642,
      "learning_rate": 0.00024141147148530037,
      "loss": 2.1841,
      "step": 129617
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.0183463096618652,
      "learning_rate": 0.00024140745969428734,
      "loss": 1.9536,
      "step": 129618
    },
    {
      "epoch": 1.69,
      "grad_norm": 0.9691706895828247,
      "learning_rate": 0.00024140344791416778,
      "loss": 1.8139,
      "step": 129619
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.3666846752166748,
      "learning_rate": 0.00024139943614494256,
      "loss": 2.0524,
      "step": 129620
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.1198735237121582,
      "learning_rate": 0.00024139542438661216,
      "loss": 2.0147,
      "step": 129621
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.0444869995117188,
      "learning_rate": 0.00024139141263917745,
      "loss": 2.0025,
      "step": 129622
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.0092791318893433,
      "learning_rate": 0.00024138740090263918,
      "loss": 2.278,
      "step": 129623
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.0578806400299072,
      "learning_rate": 0.00024138338917699813,
      "loss": 2.1218,
      "step": 129624
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.0942121744155884,
      "learning_rate": 0.00024137937746225496,
      "loss": 2.0263,
      "step": 129625
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.129218339920044,
      "learning_rate": 0.00024137536575841056,
      "loss": 1.8563,
      "step": 129626
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.0553914308547974,
      "learning_rate": 0.0002413713540654656,
      "loss": 2.2032,
      "step": 129627
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.1734142303466797,
      "learning_rate": 0.00024136734238342075,
      "loss": 1.9267,
      "step": 129628
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.2128310203552246,
      "learning_rate": 0.0002413633307122768,
      "loss": 2.307,
      "step": 129629
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.0947048664093018,
      "learning_rate": 0.00024135931905203454,
      "loss": 2.179,
      "step": 129630
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.0563229322433472,
      "learning_rate": 0.00024135530740269464,
      "loss": 2.0597,
      "step": 129631
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.073365330696106,
      "learning_rate": 0.000241351295764258,
      "loss": 1.9566,
      "step": 129632
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.2318192720413208,
      "learning_rate": 0.00024134728413672518,
      "loss": 2.2183,
      "step": 129633
    },
    {
      "epoch": 1.69,
      "grad_norm": 0.9522221684455872,
      "learning_rate": 0.00024134327252009707,
      "loss": 2.3203,
      "step": 129634
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.0605862140655518,
      "learning_rate": 0.0002413392609143743,
      "loss": 2.2359,
      "step": 129635
    },
    {
      "epoch": 1.69,
      "grad_norm": 0.9584329128265381,
      "learning_rate": 0.00024133524931955767,
      "loss": 1.921,
      "step": 129636
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.0798968076705933,
      "learning_rate": 0.0002413312377356479,
      "loss": 2.2298,
      "step": 129637
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.0360101461410522,
      "learning_rate": 0.00024132722616264588,
      "loss": 1.862,
      "step": 129638
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.053985357284546,
      "learning_rate": 0.00024132321460055214,
      "loss": 2.1233,
      "step": 129639
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.1462057828903198,
      "learning_rate": 0.0002413192030493675,
      "loss": 2.0062,
      "step": 129640
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.396235466003418,
      "learning_rate": 0.0002413151915090928,
      "loss": 2.1621,
      "step": 129641
    },
    {
      "epoch": 1.69,
      "grad_norm": 0.9564037322998047,
      "learning_rate": 0.00024131117997972868,
      "loss": 2.1774,
      "step": 129642
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.0450338125228882,
      "learning_rate": 0.0002413071684612759,
      "loss": 2.1452,
      "step": 129643
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.093632698059082,
      "learning_rate": 0.00024130315695373528,
      "loss": 2.1352,
      "step": 129644
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.046759009361267,
      "learning_rate": 0.00024129914545710748,
      "loss": 2.275,
      "step": 129645
    },
    {
      "epoch": 1.69,
      "grad_norm": 0.9915357232093811,
      "learning_rate": 0.00024129513397139324,
      "loss": 1.907,
      "step": 129646
    },
    {
      "epoch": 1.69,
      "grad_norm": 0.9986006617546082,
      "learning_rate": 0.00024129112249659333,
      "loss": 2.1295,
      "step": 129647
    },
    {
      "epoch": 1.69,
      "grad_norm": 0.9583407640457153,
      "learning_rate": 0.00024128711103270852,
      "loss": 2.0472,
      "step": 129648
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.0736321210861206,
      "learning_rate": 0.00024128309957973965,
      "loss": 2.0097,
      "step": 129649
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.1824181079864502,
      "learning_rate": 0.00024127908813768726,
      "loss": 2.3428,
      "step": 129650
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.0785491466522217,
      "learning_rate": 0.00024127507670655227,
      "loss": 2.2716,
      "step": 129651
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.0129882097244263,
      "learning_rate": 0.00024127106528633527,
      "loss": 2.3073,
      "step": 129652
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.0455026626586914,
      "learning_rate": 0.00024126705387703714,
      "loss": 2.1416,
      "step": 129653
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.205096960067749,
      "learning_rate": 0.0002412630424786585,
      "loss": 2.0714,
      "step": 129654
    },
    {
      "epoch": 1.69,
      "grad_norm": 0.9540748000144958,
      "learning_rate": 0.0002412590310912002,
      "loss": 2.0682,
      "step": 129655
    },
    {
      "epoch": 1.69,
      "grad_norm": 0.9995133280754089,
      "learning_rate": 0.00024125501971466307,
      "loss": 2.0674,
      "step": 129656
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.096191644668579,
      "learning_rate": 0.00024125100834904766,
      "loss": 2.4864,
      "step": 129657
    },
    {
      "epoch": 1.69,
      "grad_norm": 0.9540245532989502,
      "learning_rate": 0.00024124699699435478,
      "loss": 1.8913,
      "step": 129658
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.0993918180465698,
      "learning_rate": 0.00024124298565058515,
      "loss": 1.9838,
      "step": 129659
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.0667102336883545,
      "learning_rate": 0.0002412389743177396,
      "loss": 2.2273,
      "step": 129660
    },
    {
      "epoch": 1.69,
      "grad_norm": 0.9996322393417358,
      "learning_rate": 0.00024123496299581883,
      "loss": 2.1446,
      "step": 129661
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.0806950330734253,
      "learning_rate": 0.00024123095168482358,
      "loss": 1.892,
      "step": 129662
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.0538021326065063,
      "learning_rate": 0.00024122694038475472,
      "loss": 1.886,
      "step": 129663
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.073632001876831,
      "learning_rate": 0.00024122292909561278,
      "loss": 2.1206,
      "step": 129664
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.0608083009719849,
      "learning_rate": 0.0002412189178173986,
      "loss": 2.1245,
      "step": 129665
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.1269482374191284,
      "learning_rate": 0.00024121490655011293,
      "loss": 2.2406,
      "step": 129666
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.0684053897857666,
      "learning_rate": 0.00024121089529375653,
      "loss": 2.0671,
      "step": 129667
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.1315892934799194,
      "learning_rate": 0.0002412068840483301,
      "loss": 2.0582,
      "step": 129668
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.5250918865203857,
      "learning_rate": 0.0002412028728138346,
      "loss": 2.1702,
      "step": 129669
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.0068570375442505,
      "learning_rate": 0.0002411988615902704,
      "loss": 1.9248,
      "step": 129670
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.004798412322998,
      "learning_rate": 0.0002411948503776385,
      "loss": 2.2578,
      "step": 129671
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.012369990348816,
      "learning_rate": 0.00024119083917593955,
      "loss": 2.1772,
      "step": 129672
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.018851637840271,
      "learning_rate": 0.00024118682798517435,
      "loss": 2.2404,
      "step": 129673
    },
    {
      "epoch": 1.69,
      "grad_norm": 0.9819549322128296,
      "learning_rate": 0.00024118281680534363,
      "loss": 2.1965,
      "step": 129674
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.0746212005615234,
      "learning_rate": 0.00024117880563644827,
      "loss": 2.3211,
      "step": 129675
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.1355786323547363,
      "learning_rate": 0.00024117479447848875,
      "loss": 2.4263,
      "step": 129676
    },
    {
      "epoch": 1.69,
      "grad_norm": 0.9893118143081665,
      "learning_rate": 0.00024117078333146593,
      "loss": 2.0944,
      "step": 129677
    },
    {
      "epoch": 1.69,
      "grad_norm": 0.9422447085380554,
      "learning_rate": 0.00024116677219538057,
      "loss": 2.3397,
      "step": 129678
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.14387047290802,
      "learning_rate": 0.00024116276107023345,
      "loss": 2.1421,
      "step": 129679
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.1333723068237305,
      "learning_rate": 0.00024115874995602523,
      "loss": 2.1426,
      "step": 129680
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.005695104598999,
      "learning_rate": 0.00024115473885275688,
      "loss": 2.1931,
      "step": 129681
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.117706298828125,
      "learning_rate": 0.00024115072776042886,
      "loss": 2.0725,
      "step": 129682
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.1479281187057495,
      "learning_rate": 0.00024114671667904198,
      "loss": 1.9238,
      "step": 129683
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.033088207244873,
      "learning_rate": 0.00024114270560859707,
      "loss": 2.0975,
      "step": 129684
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.0448884963989258,
      "learning_rate": 0.0002411386945490948,
      "loss": 1.8839,
      "step": 129685
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.1094921827316284,
      "learning_rate": 0.00024113468350053603,
      "loss": 1.994,
      "step": 129686
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.1238871812820435,
      "learning_rate": 0.00024113067246292152,
      "loss": 2.051,
      "step": 129687
    },
    {
      "epoch": 1.69,
      "grad_norm": 0.990571141242981,
      "learning_rate": 0.0002411266614362518,
      "loss": 2.0396,
      "step": 129688
    },
    {
      "epoch": 1.69,
      "grad_norm": 0.9708403944969177,
      "learning_rate": 0.00024112265042052774,
      "loss": 2.3099,
      "step": 129689
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.1380523443222046,
      "learning_rate": 0.0002411186394157501,
      "loss": 2.2717,
      "step": 129690
    },
    {
      "epoch": 1.69,
      "grad_norm": 0.9415709376335144,
      "learning_rate": 0.00024111462842191963,
      "loss": 2.0754,
      "step": 129691
    },
    {
      "epoch": 1.69,
      "grad_norm": 0.9729894399642944,
      "learning_rate": 0.00024111061743903705,
      "loss": 2.2633,
      "step": 129692
    },
    {
      "epoch": 1.69,
      "grad_norm": 0.8934153914451599,
      "learning_rate": 0.00024110660646710325,
      "loss": 1.8586,
      "step": 129693
    },
    {
      "epoch": 1.69,
      "grad_norm": 0.9778041243553162,
      "learning_rate": 0.0002411025955061187,
      "loss": 2.0735,
      "step": 129694
    },
    {
      "epoch": 1.69,
      "grad_norm": 0.9560685753822327,
      "learning_rate": 0.00024109858455608428,
      "loss": 2.0872,
      "step": 129695
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.0695219039916992,
      "learning_rate": 0.0002410945736170008,
      "loss": 1.9508,
      "step": 129696
    },
    {
      "epoch": 1.69,
      "grad_norm": 0.9996262788772583,
      "learning_rate": 0.0002410905626888689,
      "loss": 2.095,
      "step": 129697
    },
    {
      "epoch": 1.69,
      "grad_norm": 0.8809024691581726,
      "learning_rate": 0.00024108655177168938,
      "loss": 2.0472,
      "step": 129698
    },
    {
      "epoch": 1.69,
      "grad_norm": 0.9774892330169678,
      "learning_rate": 0.00024108254086546314,
      "loss": 2.1243,
      "step": 129699
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.0043392181396484,
      "learning_rate": 0.00024107852997019063,
      "loss": 1.971,
      "step": 129700
    },
    {
      "epoch": 1.69,
      "grad_norm": 0.937890350818634,
      "learning_rate": 0.00024107451908587273,
      "loss": 2.2173,
      "step": 129701
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.0475504398345947,
      "learning_rate": 0.00024107050821251016,
      "loss": 2.0042,
      "step": 129702
    },
    {
      "epoch": 1.69,
      "grad_norm": 0.959891140460968,
      "learning_rate": 0.0002410664973501037,
      "loss": 2.1765,
      "step": 129703
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.043960452079773,
      "learning_rate": 0.0002410624864986541,
      "loss": 2.0966,
      "step": 129704
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.0297900438308716,
      "learning_rate": 0.00024105847565816227,
      "loss": 2.3496,
      "step": 129705
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.0212472677230835,
      "learning_rate": 0.00024105446482862862,
      "loss": 2.0234,
      "step": 129706
    },
    {
      "epoch": 1.69,
      "grad_norm": 0.9726613759994507,
      "learning_rate": 0.00024105045401005406,
      "loss": 2.3812,
      "step": 129707
    },
    {
      "epoch": 1.69,
      "grad_norm": 0.9956808686256409,
      "learning_rate": 0.00024104644320243934,
      "loss": 1.9755,
      "step": 129708
    },
    {
      "epoch": 1.69,
      "grad_norm": 0.9791694283485413,
      "learning_rate": 0.00024104243240578516,
      "loss": 2.051,
      "step": 129709
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.1947011947631836,
      "learning_rate": 0.00024103842162009233,
      "loss": 2.0935,
      "step": 129710
    },
    {
      "epoch": 1.69,
      "grad_norm": 0.9878056049346924,
      "learning_rate": 0.00024103441084536163,
      "loss": 1.7948,
      "step": 129711
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.06002938747406,
      "learning_rate": 0.0002410304000815938,
      "loss": 2.0273,
      "step": 129712
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.0671610832214355,
      "learning_rate": 0.0002410263893287894,
      "loss": 2.1083,
      "step": 129713
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.0566017627716064,
      "learning_rate": 0.0002410223785869493,
      "loss": 1.9583,
      "step": 129714
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.1666228771209717,
      "learning_rate": 0.0002410183678560743,
      "loss": 2.1127,
      "step": 129715
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.0549466609954834,
      "learning_rate": 0.00024101435713616508,
      "loss": 2.1083,
      "step": 129716
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.3190668821334839,
      "learning_rate": 0.0002410103464272224,
      "loss": 2.272,
      "step": 129717
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.185486078262329,
      "learning_rate": 0.0002410063357292471,
      "loss": 2.301,
      "step": 129718
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.1037172079086304,
      "learning_rate": 0.00024100232504223982,
      "loss": 2.2033,
      "step": 129719
    },
    {
      "epoch": 1.69,
      "grad_norm": 0.9515412449836731,
      "learning_rate": 0.00024099831436620123,
      "loss": 2.0394,
      "step": 129720
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.1595488786697388,
      "learning_rate": 0.0002409943037011322,
      "loss": 1.9114,
      "step": 129721
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.5899837017059326,
      "learning_rate": 0.00024099029304703343,
      "loss": 1.9902,
      "step": 129722
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.0071837902069092,
      "learning_rate": 0.00024098628240390568,
      "loss": 2.2172,
      "step": 129723
    },
    {
      "epoch": 1.69,
      "grad_norm": 0.9399101138114929,
      "learning_rate": 0.00024098227177174976,
      "loss": 2.2595,
      "step": 129724
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.0152645111083984,
      "learning_rate": 0.00024097826115056627,
      "loss": 2.1797,
      "step": 129725
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.041231393814087,
      "learning_rate": 0.00024097425054035615,
      "loss": 2.2574,
      "step": 129726
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.028711199760437,
      "learning_rate": 0.00024097023994111992,
      "loss": 2.0604,
      "step": 129727
    },
    {
      "epoch": 1.69,
      "grad_norm": 0.9753584265708923,
      "learning_rate": 0.00024096622935285844,
      "loss": 2.1178,
      "step": 129728
    },
    {
      "epoch": 1.69,
      "grad_norm": 0.9094184041023254,
      "learning_rate": 0.00024096221877557245,
      "loss": 2.2817,
      "step": 129729
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.025628924369812,
      "learning_rate": 0.00024095820820926276,
      "loss": 2.0108,
      "step": 129730
    },
    {
      "epoch": 1.69,
      "grad_norm": 0.9974060654640198,
      "learning_rate": 0.00024095419765393,
      "loss": 2.1951,
      "step": 129731
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.0489540100097656,
      "learning_rate": 0.000240950187109575,
      "loss": 2.2526,
      "step": 129732
    },
    {
      "epoch": 1.69,
      "grad_norm": 0.9946509003639221,
      "learning_rate": 0.00024094617657619848,
      "loss": 2.0067,
      "step": 129733
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.121043086051941,
      "learning_rate": 0.00024094216605380117,
      "loss": 2.0564,
      "step": 129734
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.1173533201217651,
      "learning_rate": 0.00024093815554238377,
      "loss": 1.992,
      "step": 129735
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.1095343828201294,
      "learning_rate": 0.0002409341450419472,
      "loss": 2.2158,
      "step": 129736
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.100693702697754,
      "learning_rate": 0.00024093013455249202,
      "loss": 2.0419,
      "step": 129737
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.0022504329681396,
      "learning_rate": 0.00024092612407401902,
      "loss": 2.0188,
      "step": 129738
    },
    {
      "epoch": 1.69,
      "grad_norm": 0.9621357321739197,
      "learning_rate": 0.00024092211360652896,
      "loss": 2.069,
      "step": 129739
    },
    {
      "epoch": 1.69,
      "grad_norm": 0.9885489344596863,
      "learning_rate": 0.0002409181031500226,
      "loss": 2.0697,
      "step": 129740
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.1406749486923218,
      "learning_rate": 0.00024091409270450076,
      "loss": 2.0727,
      "step": 129741
    },
    {
      "epoch": 1.69,
      "grad_norm": 0.9895870685577393,
      "learning_rate": 0.0002409100822699641,
      "loss": 2.0901,
      "step": 129742
    },
    {
      "epoch": 1.69,
      "grad_norm": 0.9282057881355286,
      "learning_rate": 0.00024090607184641327,
      "loss": 2.0701,
      "step": 129743
    },
    {
      "epoch": 1.69,
      "grad_norm": 0.9667200446128845,
      "learning_rate": 0.00024090206143384917,
      "loss": 2.1015,
      "step": 129744
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.0303658246994019,
      "learning_rate": 0.00024089805103227248,
      "loss": 2.0025,
      "step": 129745
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.0699149370193481,
      "learning_rate": 0.00024089404064168393,
      "loss": 2.1129,
      "step": 129746
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.0283763408660889,
      "learning_rate": 0.00024089003026208432,
      "loss": 2.0458,
      "step": 129747
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.0075232982635498,
      "learning_rate": 0.00024088601989347453,
      "loss": 2.1575,
      "step": 129748
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.2154793739318848,
      "learning_rate": 0.00024088200953585495,
      "loss": 2.3421,
      "step": 129749
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.2153053283691406,
      "learning_rate": 0.00024087799918922657,
      "loss": 2.2698,
      "step": 129750
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.0947519540786743,
      "learning_rate": 0.00024087398885359006,
      "loss": 1.9996,
      "step": 129751
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.127418875694275,
      "learning_rate": 0.00024086997852894619,
      "loss": 2.102,
      "step": 129752
    },
    {
      "epoch": 1.69,
      "grad_norm": 0.9455866813659668,
      "learning_rate": 0.0002408659682152957,
      "loss": 1.9602,
      "step": 129753
    },
    {
      "epoch": 1.69,
      "grad_norm": 0.950499951839447,
      "learning_rate": 0.00024086195791263952,
      "loss": 1.9934,
      "step": 129754
    },
    {
      "epoch": 1.69,
      "grad_norm": 0.9413514137268066,
      "learning_rate": 0.00024085794762097803,
      "loss": 2.1705,
      "step": 129755
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.0400464534759521,
      "learning_rate": 0.00024085393734031219,
      "loss": 2.2322,
      "step": 129756
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.0976260900497437,
      "learning_rate": 0.0002408499270706427,
      "loss": 1.769,
      "step": 129757
    },
    {
      "epoch": 1.69,
      "grad_norm": 0.9758327007293701,
      "learning_rate": 0.00024084591681197034,
      "loss": 2.0738,
      "step": 129758
    },
    {
      "epoch": 1.69,
      "grad_norm": 0.9590978622436523,
      "learning_rate": 0.00024084190656429588,
      "loss": 2.0449,
      "step": 129759
    },
    {
      "epoch": 1.69,
      "grad_norm": 0.9938681125640869,
      "learning_rate": 0.0002408378963276201,
      "loss": 2.1135,
      "step": 129760
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.0608530044555664,
      "learning_rate": 0.00024083388610194359,
      "loss": 2.1695,
      "step": 129761
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.0120201110839844,
      "learning_rate": 0.00024082987588726714,
      "loss": 1.997,
      "step": 129762
    },
    {
      "epoch": 1.69,
      "grad_norm": 0.9936360120773315,
      "learning_rate": 0.00024082586568359156,
      "loss": 2.009,
      "step": 129763
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.0855367183685303,
      "learning_rate": 0.0002408218554909175,
      "loss": 2.1315,
      "step": 129764
    },
    {
      "epoch": 1.69,
      "grad_norm": 0.9955781102180481,
      "learning_rate": 0.00024081784530924585,
      "loss": 1.9884,
      "step": 129765
    },
    {
      "epoch": 1.69,
      "grad_norm": 0.9515324831008911,
      "learning_rate": 0.0002408138351385774,
      "loss": 2.0181,
      "step": 129766
    },
    {
      "epoch": 1.69,
      "grad_norm": 0.9361107349395752,
      "learning_rate": 0.0002408098249789126,
      "loss": 2.0611,
      "step": 129767
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.2026031017303467,
      "learning_rate": 0.00024080581483025242,
      "loss": 2.1845,
      "step": 129768
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.0034576654434204,
      "learning_rate": 0.00024080180469259754,
      "loss": 2.0462,
      "step": 129769
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.0014533996582031,
      "learning_rate": 0.00024079779456594869,
      "loss": 2.116,
      "step": 129770
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.5455602407455444,
      "learning_rate": 0.0002407937844503067,
      "loss": 2.0057,
      "step": 129771
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.1074097156524658,
      "learning_rate": 0.00024078977434567234,
      "loss": 1.8489,
      "step": 129772
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.0539988279342651,
      "learning_rate": 0.00024078576425204618,
      "loss": 2.1803,
      "step": 129773
    },
    {
      "epoch": 1.69,
      "grad_norm": 0.9540101289749146,
      "learning_rate": 0.00024078175416942908,
      "loss": 2.0923,
      "step": 129774
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.1043918132781982,
      "learning_rate": 0.00024077774409782176,
      "loss": 2.0235,
      "step": 129775
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.1059293746948242,
      "learning_rate": 0.00024077373403722497,
      "loss": 1.9933,
      "step": 129776
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.138197660446167,
      "learning_rate": 0.00024076972398763945,
      "loss": 2.0745,
      "step": 129777
    },
    {
      "epoch": 1.69,
      "grad_norm": 0.9621672630310059,
      "learning_rate": 0.00024076571394906606,
      "loss": 2.0474,
      "step": 129778
    },
    {
      "epoch": 1.69,
      "grad_norm": 0.9670891165733337,
      "learning_rate": 0.00024076170392150538,
      "loss": 2.11,
      "step": 129779
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.093624472618103,
      "learning_rate": 0.00024075769390495818,
      "loss": 1.8117,
      "step": 129780
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.1243199110031128,
      "learning_rate": 0.00024075368389942525,
      "loss": 1.9539,
      "step": 129781
    },
    {
      "epoch": 1.69,
      "grad_norm": 0.9671981334686279,
      "learning_rate": 0.0002407496739049073,
      "loss": 2.1547,
      "step": 129782
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.0338692665100098,
      "learning_rate": 0.00024074566392140513,
      "loss": 2.154,
      "step": 129783
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.0235930681228638,
      "learning_rate": 0.00024074165394891948,
      "loss": 2.2031,
      "step": 129784
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.130266547203064,
      "learning_rate": 0.00024073764398745118,
      "loss": 1.9247,
      "step": 129785
    },
    {
      "epoch": 1.69,
      "grad_norm": 0.9996336698532104,
      "learning_rate": 0.00024073363403700075,
      "loss": 1.9171,
      "step": 129786
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.0170931816101074,
      "learning_rate": 0.00024072962409756905,
      "loss": 2.0848,
      "step": 129787
    },
    {
      "epoch": 1.69,
      "grad_norm": 0.9062988758087158,
      "learning_rate": 0.0002407256141691568,
      "loss": 1.9129,
      "step": 129788
    },
    {
      "epoch": 1.69,
      "grad_norm": 0.9144315123558044,
      "learning_rate": 0.00024072160425176484,
      "loss": 2.246,
      "step": 129789
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.0276522636413574,
      "learning_rate": 0.00024071759434539383,
      "loss": 1.9442,
      "step": 129790
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.1376338005065918,
      "learning_rate": 0.00024071358445004466,
      "loss": 2.1142,
      "step": 129791
    },
    {
      "epoch": 1.69,
      "grad_norm": 0.9997599124908447,
      "learning_rate": 0.00024070957456571781,
      "loss": 2.2093,
      "step": 129792
    },
    {
      "epoch": 1.69,
      "grad_norm": 0.955345630645752,
      "learning_rate": 0.0002407055646924142,
      "loss": 2.0962,
      "step": 129793
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.07835853099823,
      "learning_rate": 0.0002407015548301345,
      "loss": 2.3599,
      "step": 129794
    },
    {
      "epoch": 1.69,
      "grad_norm": 0.9947220683097839,
      "learning_rate": 0.00024069754497887955,
      "loss": 2.1881,
      "step": 129795
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.0542948246002197,
      "learning_rate": 0.00024069353513865002,
      "loss": 2.4038,
      "step": 129796
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.0774730443954468,
      "learning_rate": 0.0002406895253094468,
      "loss": 2.1591,
      "step": 129797
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.0199552774429321,
      "learning_rate": 0.00024068551549127041,
      "loss": 2.0938,
      "step": 129798
    },
    {
      "epoch": 1.69,
      "grad_norm": 0.9793649315834045,
      "learning_rate": 0.0002406815056841217,
      "loss": 2.1333,
      "step": 129799
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.1180230379104614,
      "learning_rate": 0.00024067749588800144,
      "loss": 2.2874,
      "step": 129800
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.007741928100586,
      "learning_rate": 0.0002406734861029103,
      "loss": 2.1032,
      "step": 129801
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.1443978548049927,
      "learning_rate": 0.0002406694763288491,
      "loss": 2.1729,
      "step": 129802
    },
    {
      "epoch": 1.69,
      "grad_norm": 0.9934077262878418,
      "learning_rate": 0.00024066546656581863,
      "loss": 2.0174,
      "step": 129803
    },
    {
      "epoch": 1.69,
      "grad_norm": 0.9898912906646729,
      "learning_rate": 0.00024066145681381955,
      "loss": 2.0791,
      "step": 129804
    },
    {
      "epoch": 1.69,
      "grad_norm": 0.9577246904373169,
      "learning_rate": 0.00024065744707285264,
      "loss": 2.092,
      "step": 129805
    },
    {
      "epoch": 1.69,
      "grad_norm": 0.9163469076156616,
      "learning_rate": 0.00024065343734291854,
      "loss": 2.0522,
      "step": 129806
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.0055196285247803,
      "learning_rate": 0.00024064942762401815,
      "loss": 2.0026,
      "step": 129807
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.1092166900634766,
      "learning_rate": 0.0002406454179161521,
      "loss": 2.1358,
      "step": 129808
    },
    {
      "epoch": 1.69,
      "grad_norm": 0.9982963800430298,
      "learning_rate": 0.00024064140821932126,
      "loss": 2.1541,
      "step": 129809
    },
    {
      "epoch": 1.69,
      "grad_norm": 0.9567794799804688,
      "learning_rate": 0.00024063739853352628,
      "loss": 2.0045,
      "step": 129810
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.0345189571380615,
      "learning_rate": 0.00024063338885876796,
      "loss": 2.1417,
      "step": 129811
    },
    {
      "epoch": 1.69,
      "grad_norm": 0.9595001339912415,
      "learning_rate": 0.00024062937919504693,
      "loss": 2.3091,
      "step": 129812
    },
    {
      "epoch": 1.69,
      "grad_norm": 0.9938135147094727,
      "learning_rate": 0.00024062536954236406,
      "loss": 2.0442,
      "step": 129813
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.0798065662384033,
      "learning_rate": 0.00024062135990072006,
      "loss": 1.9909,
      "step": 129814
    },
    {
      "epoch": 1.69,
      "grad_norm": 0.9727192521095276,
      "learning_rate": 0.0002406173502701157,
      "loss": 1.9006,
      "step": 129815
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.0360103845596313,
      "learning_rate": 0.00024061334065055164,
      "loss": 2.0826,
      "step": 129816
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.0484367609024048,
      "learning_rate": 0.0002406093310420287,
      "loss": 2.0628,
      "step": 129817
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.095314621925354,
      "learning_rate": 0.00024060532144454765,
      "loss": 2.1703,
      "step": 129818
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.057515025138855,
      "learning_rate": 0.00024060131185810914,
      "loss": 2.2788,
      "step": 129819
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.2561510801315308,
      "learning_rate": 0.00024059730228271397,
      "loss": 2.066,
      "step": 129820
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.2142118215560913,
      "learning_rate": 0.00024059329271836294,
      "loss": 2.4656,
      "step": 129821
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.189961552619934,
      "learning_rate": 0.00024058928316505665,
      "loss": 2.1266,
      "step": 129822
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.2769482135772705,
      "learning_rate": 0.00024058527362279594,
      "loss": 1.9934,
      "step": 129823
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.0110822916030884,
      "learning_rate": 0.00024058126409158158,
      "loss": 2.3213,
      "step": 129824
    },
    {
      "epoch": 1.69,
      "grad_norm": 0.9512803554534912,
      "learning_rate": 0.00024057725457141433,
      "loss": 2.0046,
      "step": 129825
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.1359204053878784,
      "learning_rate": 0.00024057324506229484,
      "loss": 2.0284,
      "step": 129826
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.0182080268859863,
      "learning_rate": 0.00024056923556422394,
      "loss": 2.0985,
      "step": 129827
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.014316439628601,
      "learning_rate": 0.00024056522607720233,
      "loss": 2.3178,
      "step": 129828
    },
    {
      "epoch": 1.69,
      "grad_norm": 0.9068283438682556,
      "learning_rate": 0.00024056121660123072,
      "loss": 2.0134,
      "step": 129829
    },
    {
      "epoch": 1.69,
      "grad_norm": 0.9712595343589783,
      "learning_rate": 0.00024055720713630992,
      "loss": 2.1521,
      "step": 129830
    },
    {
      "epoch": 1.69,
      "grad_norm": 0.9667252898216248,
      "learning_rate": 0.00024055319768244063,
      "loss": 2.238,
      "step": 129831
    },
    {
      "epoch": 1.69,
      "grad_norm": 0.9612887501716614,
      "learning_rate": 0.00024054918823962365,
      "loss": 2.1753,
      "step": 129832
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.087666630744934,
      "learning_rate": 0.00024054517880785982,
      "loss": 2.3653,
      "step": 129833
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.1098829507827759,
      "learning_rate": 0.00024054116938714964,
      "loss": 2.2014,
      "step": 129834
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.012448787689209,
      "learning_rate": 0.00024053715997749396,
      "loss": 2.2231,
      "step": 129835
    },
    {
      "epoch": 1.69,
      "grad_norm": 0.9161502718925476,
      "learning_rate": 0.00024053315057889356,
      "loss": 1.6966,
      "step": 129836
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.0970286130905151,
      "learning_rate": 0.00024052914119134918,
      "loss": 1.918,
      "step": 129837
    },
    {
      "epoch": 1.69,
      "grad_norm": 0.9272754788398743,
      "learning_rate": 0.00024052513181486155,
      "loss": 2.0379,
      "step": 129838
    },
    {
      "epoch": 1.69,
      "grad_norm": 0.9597609043121338,
      "learning_rate": 0.00024052112244943153,
      "loss": 1.9664,
      "step": 129839
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.0556690692901611,
      "learning_rate": 0.00024051711309505967,
      "loss": 1.9983,
      "step": 129840
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.1095647811889648,
      "learning_rate": 0.00024051310375174676,
      "loss": 2.0344,
      "step": 129841
    },
    {
      "epoch": 1.69,
      "grad_norm": 0.9146652221679688,
      "learning_rate": 0.0002405090944194936,
      "loss": 2.2747,
      "step": 129842
    },
    {
      "epoch": 1.69,
      "grad_norm": 0.9008116126060486,
      "learning_rate": 0.0002405050850983009,
      "loss": 2.017,
      "step": 129843
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.0871732234954834,
      "learning_rate": 0.00024050107578816948,
      "loss": 2.129,
      "step": 129844
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.135998010635376,
      "learning_rate": 0.0002404970664891001,
      "loss": 2.0494,
      "step": 129845
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.0464493036270142,
      "learning_rate": 0.00024049305720109335,
      "loss": 1.982,
      "step": 129846
    },
    {
      "epoch": 1.69,
      "grad_norm": 0.9522435069084167,
      "learning_rate": 0.00024048904792415003,
      "loss": 2.0737,
      "step": 129847
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.0189601182937622,
      "learning_rate": 0.000240485038658271,
      "loss": 2.1788,
      "step": 129848
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.1681605577468872,
      "learning_rate": 0.00024048102940345685,
      "loss": 2.0705,
      "step": 129849
    },
    {
      "epoch": 1.69,
      "grad_norm": 0.958905816078186,
      "learning_rate": 0.0002404770201597084,
      "loss": 2.2314,
      "step": 129850
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.0150113105773926,
      "learning_rate": 0.00024047301092702645,
      "loss": 1.8645,
      "step": 129851
    },
    {
      "epoch": 1.69,
      "grad_norm": 0.9679205417633057,
      "learning_rate": 0.0002404690017054118,
      "loss": 2.0331,
      "step": 129852
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.3722461462020874,
      "learning_rate": 0.00024046499249486493,
      "loss": 2.1926,
      "step": 129853
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.0285489559173584,
      "learning_rate": 0.00024046098329538672,
      "loss": 2.2562,
      "step": 129854
    },
    {
      "epoch": 1.69,
      "grad_norm": 0.9657289385795593,
      "learning_rate": 0.000240456974106978,
      "loss": 2.0825,
      "step": 129855
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.0533758401870728,
      "learning_rate": 0.0002404529649296394,
      "loss": 2.0152,
      "step": 129856
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.378280758857727,
      "learning_rate": 0.00024044895576337175,
      "loss": 2.1196,
      "step": 129857
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.02163827419281,
      "learning_rate": 0.00024044494660817588,
      "loss": 2.025,
      "step": 129858
    },
    {
      "epoch": 1.69,
      "grad_norm": 0.9767940640449524,
      "learning_rate": 0.0002404409374640523,
      "loss": 2.0363,
      "step": 129859
    },
    {
      "epoch": 1.69,
      "grad_norm": 0.9578477740287781,
      "learning_rate": 0.00024043692833100185,
      "loss": 2.1452,
      "step": 129860
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.0301483869552612,
      "learning_rate": 0.00024043291920902533,
      "loss": 2.1612,
      "step": 129861
    },
    {
      "epoch": 1.69,
      "grad_norm": 0.9812340140342712,
      "learning_rate": 0.00024042891009812346,
      "loss": 1.8783,
      "step": 129862
    },
    {
      "epoch": 1.69,
      "grad_norm": 0.9649237990379333,
      "learning_rate": 0.00024042490099829695,
      "loss": 2.1126,
      "step": 129863
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.1010820865631104,
      "learning_rate": 0.00024042089190954669,
      "loss": 2.2192,
      "step": 129864
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.1387860774993896,
      "learning_rate": 0.00024041688283187322,
      "loss": 2.0721,
      "step": 129865
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.1101142168045044,
      "learning_rate": 0.00024041287376527735,
      "loss": 2.3838,
      "step": 129866
    },
    {
      "epoch": 1.69,
      "grad_norm": 0.8821164965629578,
      "learning_rate": 0.00024040886470975988,
      "loss": 1.8428,
      "step": 129867
    },
    {
      "epoch": 1.69,
      "grad_norm": 0.9390270113945007,
      "learning_rate": 0.0002404048556653215,
      "loss": 2.2228,
      "step": 129868
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.0992201566696167,
      "learning_rate": 0.000240400846631963,
      "loss": 1.9539,
      "step": 129869
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.050162434577942,
      "learning_rate": 0.00024039683760968526,
      "loss": 2.0469,
      "step": 129870
    },
    {
      "epoch": 1.69,
      "grad_norm": 0.9590102434158325,
      "learning_rate": 0.00024039282859848873,
      "loss": 1.9261,
      "step": 129871
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.0337570905685425,
      "learning_rate": 0.00024038881959837424,
      "loss": 2.0517,
      "step": 129872
    },
    {
      "epoch": 1.69,
      "grad_norm": 0.9622744917869568,
      "learning_rate": 0.00024038481060934265,
      "loss": 2.1431,
      "step": 129873
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.053780436515808,
      "learning_rate": 0.00024038080163139465,
      "loss": 2.1311,
      "step": 129874
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.4416757822036743,
      "learning_rate": 0.00024037679266453098,
      "loss": 2.009,
      "step": 129875
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.0512710809707642,
      "learning_rate": 0.0002403727837087525,
      "loss": 2.1032,
      "step": 129876
    },
    {
      "epoch": 1.69,
      "grad_norm": 0.8978365063667297,
      "learning_rate": 0.0002403687747640597,
      "loss": 2.1846,
      "step": 129877
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.0112136602401733,
      "learning_rate": 0.0002403647658304535,
      "loss": 2.2244,
      "step": 129878
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.0971952676773071,
      "learning_rate": 0.0002403607569079346,
      "loss": 2.1048,
      "step": 129879
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.0375592708587646,
      "learning_rate": 0.0002403567479965038,
      "loss": 2.1365,
      "step": 129880
    },
    {
      "epoch": 1.69,
      "grad_norm": 0.9505748152732849,
      "learning_rate": 0.00024035273909616176,
      "loss": 1.9552,
      "step": 129881
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.1230049133300781,
      "learning_rate": 0.00024034873020690943,
      "loss": 2.1268,
      "step": 129882
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.1731493473052979,
      "learning_rate": 0.00024034472132874725,
      "loss": 2.0415,
      "step": 129883
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.0588586330413818,
      "learning_rate": 0.00024034071246167616,
      "loss": 2.1521,
      "step": 129884
    },
    {
      "epoch": 1.69,
      "grad_norm": 0.9228216409683228,
      "learning_rate": 0.0002403367036056968,
      "loss": 2.0649,
      "step": 129885
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.1646838188171387,
      "learning_rate": 0.00024033269476080997,
      "loss": 2.0953,
      "step": 129886
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.1702936887741089,
      "learning_rate": 0.00024032868592701646,
      "loss": 2.0707,
      "step": 129887
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.0431047677993774,
      "learning_rate": 0.00024032467710431707,
      "loss": 1.9672,
      "step": 129888
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.0842598676681519,
      "learning_rate": 0.00024032066829271232,
      "loss": 1.9437,
      "step": 129889
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.1817477941513062,
      "learning_rate": 0.0002403166594922031,
      "loss": 2.0529,
      "step": 129890
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.144407868385315,
      "learning_rate": 0.00024031265070279017,
      "loss": 2.3347,
      "step": 129891
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.001966953277588,
      "learning_rate": 0.0002403086419244742,
      "loss": 2.3921,
      "step": 129892
    },
    {
      "epoch": 1.69,
      "grad_norm": 0.9889843463897705,
      "learning_rate": 0.00024030463315725598,
      "loss": 1.912,
      "step": 129893
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.6524577140808105,
      "learning_rate": 0.00024030062440113633,
      "loss": 2.1349,
      "step": 129894
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.1477227210998535,
      "learning_rate": 0.0002402966156561159,
      "loss": 2.007,
      "step": 129895
    },
    {
      "epoch": 1.69,
      "grad_norm": 0.9820302724838257,
      "learning_rate": 0.00024029260692219544,
      "loss": 2.1334,
      "step": 129896
    },
    {
      "epoch": 1.69,
      "grad_norm": 0.9564131498336792,
      "learning_rate": 0.00024028859819937568,
      "loss": 2.1631,
      "step": 129897
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.0443141460418701,
      "learning_rate": 0.0002402845894876574,
      "loss": 1.9847,
      "step": 129898
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.088665246963501,
      "learning_rate": 0.00024028058078704137,
      "loss": 2.1364,
      "step": 129899
    },
    {
      "epoch": 1.69,
      "grad_norm": 0.9908095002174377,
      "learning_rate": 0.00024027657209752833,
      "loss": 2.3914,
      "step": 129900
    },
    {
      "epoch": 1.69,
      "grad_norm": 0.9720543026924133,
      "learning_rate": 0.00024027256341911893,
      "loss": 1.9422,
      "step": 129901
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.1235888004302979,
      "learning_rate": 0.00024026855475181404,
      "loss": 2.0651,
      "step": 129902
    },
    {
      "epoch": 1.69,
      "grad_norm": 0.9893417358398438,
      "learning_rate": 0.00024026454609561438,
      "loss": 2.2313,
      "step": 129903
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.0664373636245728,
      "learning_rate": 0.00024026053745052062,
      "loss": 2.2416,
      "step": 129904
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.023057222366333,
      "learning_rate": 0.00024025652881653353,
      "loss": 2.1682,
      "step": 129905
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.270051121711731,
      "learning_rate": 0.00024025252019365396,
      "loss": 2.132,
      "step": 129906
    },
    {
      "epoch": 1.69,
      "grad_norm": 0.9486056566238403,
      "learning_rate": 0.0002402485115818825,
      "loss": 2.0003,
      "step": 129907
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.0783506631851196,
      "learning_rate": 0.00024024450298122,
      "loss": 2.0588,
      "step": 129908
    },
    {
      "epoch": 1.69,
      "grad_norm": 0.9672467708587646,
      "learning_rate": 0.00024024049439166716,
      "loss": 2.0583,
      "step": 129909
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.0605982542037964,
      "learning_rate": 0.00024023648581322478,
      "loss": 2.0242,
      "step": 129910
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.111768364906311,
      "learning_rate": 0.00024023247724589352,
      "loss": 2.0535,
      "step": 129911
    },
    {
      "epoch": 1.69,
      "grad_norm": 0.9742724299430847,
      "learning_rate": 0.00024022846868967425,
      "loss": 1.889,
      "step": 129912
    },
    {
      "epoch": 1.69,
      "grad_norm": 0.9545394778251648,
      "learning_rate": 0.00024022446014456753,
      "loss": 2.1272,
      "step": 129913
    },
    {
      "epoch": 1.69,
      "grad_norm": 0.9787342548370361,
      "learning_rate": 0.00024022045161057427,
      "loss": 2.174,
      "step": 129914
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.082267165184021,
      "learning_rate": 0.0002402164430876951,
      "loss": 2.0267,
      "step": 129915
    },
    {
      "epoch": 1.69,
      "grad_norm": 0.9354792833328247,
      "learning_rate": 0.00024021243457593086,
      "loss": 2.1972,
      "step": 129916
    },
    {
      "epoch": 1.69,
      "grad_norm": 0.9646870493888855,
      "learning_rate": 0.00024020842607528232,
      "loss": 1.9088,
      "step": 129917
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.071821689605713,
      "learning_rate": 0.00024020441758575007,
      "loss": 2.2662,
      "step": 129918
    },
    {
      "epoch": 1.69,
      "grad_norm": 0.9793815016746521,
      "learning_rate": 0.00024020040910733504,
      "loss": 2.0746,
      "step": 129919
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.033345103263855,
      "learning_rate": 0.00024019640064003778,
      "loss": 2.4398,
      "step": 129920
    },
    {
      "epoch": 1.69,
      "grad_norm": 0.9450781941413879,
      "learning_rate": 0.00024019239218385917,
      "loss": 1.9977,
      "step": 129921
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.091936707496643,
      "learning_rate": 0.00024018838373879992,
      "loss": 1.9875,
      "step": 129922
    },
    {
      "epoch": 1.69,
      "grad_norm": 0.9420424699783325,
      "learning_rate": 0.0002401843753048608,
      "loss": 2.0182,
      "step": 129923
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.071311116218567,
      "learning_rate": 0.00024018036688204256,
      "loss": 2.1276,
      "step": 129924
    },
    {
      "epoch": 1.69,
      "grad_norm": 0.976236879825592,
      "learning_rate": 0.00024017635847034595,
      "loss": 2.1043,
      "step": 129925
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.076605200767517,
      "learning_rate": 0.0002401723500697716,
      "loss": 2.2669,
      "step": 129926
    },
    {
      "epoch": 1.69,
      "grad_norm": 0.9323815703392029,
      "learning_rate": 0.00024016834168032034,
      "loss": 2.0814,
      "step": 129927
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.0566473007202148,
      "learning_rate": 0.0002401643333019929,
      "loss": 2.3122,
      "step": 129928
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.0328617095947266,
      "learning_rate": 0.00024016032493479004,
      "loss": 2.251,
      "step": 129929
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.0223116874694824,
      "learning_rate": 0.00024015631657871251,
      "loss": 2.2633,
      "step": 129930
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.0647813081741333,
      "learning_rate": 0.00024015230823376123,
      "loss": 2.233,
      "step": 129931
    },
    {
      "epoch": 1.69,
      "grad_norm": 0.9671998023986816,
      "learning_rate": 0.00024014829989993657,
      "loss": 2.1249,
      "step": 129932
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.021693229675293,
      "learning_rate": 0.00024014429157723954,
      "loss": 1.9738,
      "step": 129933
    },
    {
      "epoch": 1.69,
      "grad_norm": 0.9844585061073303,
      "learning_rate": 0.00024014028326567075,
      "loss": 2.2575,
      "step": 129934
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.1066371202468872,
      "learning_rate": 0.00024013627496523104,
      "loss": 2.1002,
      "step": 129935
    },
    {
      "epoch": 1.69,
      "grad_norm": 0.994166910648346,
      "learning_rate": 0.00024013226667592114,
      "loss": 2.1388,
      "step": 129936
    },
    {
      "epoch": 1.69,
      "grad_norm": 0.9365060925483704,
      "learning_rate": 0.00024012825839774186,
      "loss": 1.9965,
      "step": 129937
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.0140607357025146,
      "learning_rate": 0.0002401242501306938,
      "loss": 2.1168,
      "step": 129938
    },
    {
      "epoch": 1.69,
      "grad_norm": 0.9560177326202393,
      "learning_rate": 0.00024012024187477772,
      "loss": 1.8659,
      "step": 129939
    },
    {
      "epoch": 1.69,
      "grad_norm": 0.9692559838294983,
      "learning_rate": 0.00024011623362999444,
      "loss": 2.2576,
      "step": 129940
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.1286736726760864,
      "learning_rate": 0.00024011222539634472,
      "loss": 2.3205,
      "step": 129941
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.2023437023162842,
      "learning_rate": 0.0002401082171738292,
      "loss": 2.2916,
      "step": 129942
    },
    {
      "epoch": 1.69,
      "grad_norm": 0.9811198711395264,
      "learning_rate": 0.00024010420896244884,
      "loss": 2.1222,
      "step": 129943
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.0225989818572998,
      "learning_rate": 0.00024010020076220415,
      "loss": 2.0075,
      "step": 129944
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.0563148260116577,
      "learning_rate": 0.00024009619257309592,
      "loss": 2.082,
      "step": 129945
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.0037931203842163,
      "learning_rate": 0.00024009218439512497,
      "loss": 2.0368,
      "step": 129946
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.150051474571228,
      "learning_rate": 0.000240088176228292,
      "loss": 2.2292,
      "step": 129947
    },
    {
      "epoch": 1.69,
      "grad_norm": 0.94474196434021,
      "learning_rate": 0.00024008416807259775,
      "loss": 2.1283,
      "step": 129948
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.0556329488754272,
      "learning_rate": 0.00024008015992804313,
      "loss": 2.1542,
      "step": 129949
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.0400962829589844,
      "learning_rate": 0.00024007615179462863,
      "loss": 2.1968,
      "step": 129950
    },
    {
      "epoch": 1.69,
      "grad_norm": 0.9653123617172241,
      "learning_rate": 0.00024007214367235507,
      "loss": 2.1382,
      "step": 129951
    },
    {
      "epoch": 1.69,
      "grad_norm": 0.982972264289856,
      "learning_rate": 0.00024006813556122326,
      "loss": 1.8711,
      "step": 129952
    },
    {
      "epoch": 1.69,
      "grad_norm": 0.9421581029891968,
      "learning_rate": 0.00024006412746123388,
      "loss": 2.0558,
      "step": 129953
    },
    {
      "epoch": 1.69,
      "grad_norm": 0.9616117477416992,
      "learning_rate": 0.00024006011937238774,
      "loss": 2.2108,
      "step": 129954
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.0167137384414673,
      "learning_rate": 0.00024005611129468567,
      "loss": 2.1565,
      "step": 129955
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.0163288116455078,
      "learning_rate": 0.00024005210322812818,
      "loss": 2.0692,
      "step": 129956
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.1066981554031372,
      "learning_rate": 0.00024004809517271612,
      "loss": 2.1582,
      "step": 129957
    },
    {
      "epoch": 1.69,
      "grad_norm": 0.9863407611846924,
      "learning_rate": 0.00024004408712845028,
      "loss": 1.9888,
      "step": 129958
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.0450098514556885,
      "learning_rate": 0.00024004007909533134,
      "loss": 2.311,
      "step": 129959
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.1948720216751099,
      "learning_rate": 0.0002400360710733601,
      "loss": 1.9596,
      "step": 129960
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.0534844398498535,
      "learning_rate": 0.00024003206306253742,
      "loss": 1.9829,
      "step": 129961
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.0799602270126343,
      "learning_rate": 0.0002400280550628638,
      "loss": 2.1329,
      "step": 129962
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.0209077596664429,
      "learning_rate": 0.00024002404707434004,
      "loss": 2.0613,
      "step": 129963
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.0120009183883667,
      "learning_rate": 0.00024002003909696698,
      "loss": 1.9704,
      "step": 129964
    },
    {
      "epoch": 1.69,
      "grad_norm": 0.9225879907608032,
      "learning_rate": 0.00024001603113074535,
      "loss": 1.7652,
      "step": 129965
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.0154157876968384,
      "learning_rate": 0.00024001202317567582,
      "loss": 2.1312,
      "step": 129966
    },
    {
      "epoch": 1.69,
      "grad_norm": 0.9920042157173157,
      "learning_rate": 0.00024000801523175935,
      "loss": 2.1317,
      "step": 129967
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.0538849830627441,
      "learning_rate": 0.00024000400729899636,
      "loss": 2.0212,
      "step": 129968
    },
    {
      "epoch": 1.69,
      "grad_norm": 0.9613677859306335,
      "learning_rate": 0.0002399999993773878,
      "loss": 2.026,
      "step": 129969
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.0604525804519653,
      "learning_rate": 0.00023999599146693436,
      "loss": 2.3189,
      "step": 129970
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.0383461713790894,
      "learning_rate": 0.00023999198356763676,
      "loss": 2.1015,
      "step": 129971
    },
    {
      "epoch": 1.69,
      "grad_norm": 0.9457113146781921,
      "learning_rate": 0.00023998797567949585,
      "loss": 2.0812,
      "step": 129972
    },
    {
      "epoch": 1.69,
      "grad_norm": 0.9644067883491516,
      "learning_rate": 0.00023998396780251237,
      "loss": 2.2988,
      "step": 129973
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.0545321702957153,
      "learning_rate": 0.00023997995993668692,
      "loss": 2.0149,
      "step": 129974
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.246456265449524,
      "learning_rate": 0.0002399759520820203,
      "loss": 1.9898,
      "step": 129975
    },
    {
      "epoch": 1.69,
      "grad_norm": 0.9848439693450928,
      "learning_rate": 0.00023997194423851332,
      "loss": 2.1718,
      "step": 129976
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.0858575105667114,
      "learning_rate": 0.00023996793640616664,
      "loss": 2.0286,
      "step": 129977
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.0000313520431519,
      "learning_rate": 0.0002399639285849811,
      "loss": 2.2742,
      "step": 129978
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.245805025100708,
      "learning_rate": 0.00023995992077495743,
      "loss": 2.3034,
      "step": 129979
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.0909985303878784,
      "learning_rate": 0.0002399559129760964,
      "loss": 2.0938,
      "step": 129980
    },
    {
      "epoch": 1.69,
      "grad_norm": 0.9674546122550964,
      "learning_rate": 0.00023995190518839857,
      "loss": 2.0778,
      "step": 129981
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.248622179031372,
      "learning_rate": 0.00023994789741186483,
      "loss": 2.1129,
      "step": 129982
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.324005365371704,
      "learning_rate": 0.00023994388964649593,
      "loss": 2.3037,
      "step": 129983
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.605013132095337,
      "learning_rate": 0.00023993988189229259,
      "loss": 2.05,
      "step": 129984
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.1986587047576904,
      "learning_rate": 0.00023993587414925554,
      "loss": 2.0061,
      "step": 129985
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.1343168020248413,
      "learning_rate": 0.00023993186641738562,
      "loss": 2.1327,
      "step": 129986
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.5659518241882324,
      "learning_rate": 0.00023992785869668347,
      "loss": 2.1981,
      "step": 129987
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.2628759145736694,
      "learning_rate": 0.00023992385098714985,
      "loss": 2.2297,
      "step": 129988
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.1021138429641724,
      "learning_rate": 0.00023991984328878549,
      "loss": 2.0469,
      "step": 129989
    },
    {
      "epoch": 1.69,
      "grad_norm": 0.9456981420516968,
      "learning_rate": 0.0002399158356015912,
      "loss": 1.8723,
      "step": 129990
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.2942523956298828,
      "learning_rate": 0.00023991182792556764,
      "loss": 2.1142,
      "step": 129991
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.1115095615386963,
      "learning_rate": 0.0002399078202607157,
      "loss": 2.2312,
      "step": 129992
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.1537469625473022,
      "learning_rate": 0.000239903812607036,
      "loss": 2.0601,
      "step": 129993
    },
    {
      "epoch": 1.69,
      "grad_norm": 0.9930124878883362,
      "learning_rate": 0.00023989980496452927,
      "loss": 2.0718,
      "step": 129994
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.032865285873413,
      "learning_rate": 0.00023989579733319638,
      "loss": 2.3001,
      "step": 129995
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.1469250917434692,
      "learning_rate": 0.0002398917897130379,
      "loss": 2.3185,
      "step": 129996
    },
    {
      "epoch": 1.69,
      "grad_norm": 0.9228264093399048,
      "learning_rate": 0.0002398877821040547,
      "loss": 2.1807,
      "step": 129997
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.061266541481018,
      "learning_rate": 0.00023988377450624755,
      "loss": 1.9862,
      "step": 129998
    },
    {
      "epoch": 1.69,
      "grad_norm": 0.9804173111915588,
      "learning_rate": 0.00023987976691961708,
      "loss": 2.0373,
      "step": 129999
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.0443079471588135,
      "learning_rate": 0.00023987575934416412,
      "loss": 2.2095,
      "step": 130000
    },
    {
      "epoch": 1.69,
      "grad_norm": 0.9942609071731567,
      "learning_rate": 0.00023987175177988935,
      "loss": 2.22,
      "step": 130001
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.104058027267456,
      "learning_rate": 0.00023986774422679364,
      "loss": 2.2111,
      "step": 130002
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.182202696800232,
      "learning_rate": 0.0002398637366848776,
      "loss": 2.2166,
      "step": 130003
    },
    {
      "epoch": 1.69,
      "grad_norm": 0.9890314936637878,
      "learning_rate": 0.00023985972915414208,
      "loss": 2.1948,
      "step": 130004
    },
    {
      "epoch": 1.69,
      "grad_norm": 0.9827336668968201,
      "learning_rate": 0.0002398557216345877,
      "loss": 2.2104,
      "step": 130005
    },
    {
      "epoch": 1.69,
      "grad_norm": 0.9240493178367615,
      "learning_rate": 0.00023985171412621527,
      "loss": 2.2703,
      "step": 130006
    },
    {
      "epoch": 1.69,
      "grad_norm": 0.9663406610488892,
      "learning_rate": 0.00023984770662902554,
      "loss": 2.2094,
      "step": 130007
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.1044923067092896,
      "learning_rate": 0.0002398436991430193,
      "loss": 2.0621,
      "step": 130008
    },
    {
      "epoch": 1.69,
      "grad_norm": 0.9465324878692627,
      "learning_rate": 0.0002398396916681973,
      "loss": 2.0727,
      "step": 130009
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.1106473207473755,
      "learning_rate": 0.00023983568420456018,
      "loss": 2.0755,
      "step": 130010
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.0579701662063599,
      "learning_rate": 0.00023983167675210873,
      "loss": 2.1675,
      "step": 130011
    },
    {
      "epoch": 1.69,
      "grad_norm": 0.8889029026031494,
      "learning_rate": 0.0002398276693108437,
      "loss": 2.0779,
      "step": 130012
    },
    {
      "epoch": 1.69,
      "grad_norm": 0.9904544949531555,
      "learning_rate": 0.00023982366188076586,
      "loss": 2.1617,
      "step": 130013
    },
    {
      "epoch": 1.69,
      "grad_norm": 0.9293169379234314,
      "learning_rate": 0.0002398196544618759,
      "loss": 2.1511,
      "step": 130014
    },
    {
      "epoch": 1.69,
      "grad_norm": 0.9820516109466553,
      "learning_rate": 0.0002398156470541746,
      "loss": 2.091,
      "step": 130015
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.0005522966384888,
      "learning_rate": 0.0002398116396576629,
      "loss": 2.1441,
      "step": 130016
    },
    {
      "epoch": 1.69,
      "grad_norm": 0.9268176555633545,
      "learning_rate": 0.00023980763227234118,
      "loss": 2.2504,
      "step": 130017
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.0230110883712769,
      "learning_rate": 0.00023980362489821035,
      "loss": 2.4055,
      "step": 130018
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.1448129415512085,
      "learning_rate": 0.00023979961753527117,
      "loss": 2.0237,
      "step": 130019
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.012851595878601,
      "learning_rate": 0.00023979561018352434,
      "loss": 1.9799,
      "step": 130020
    },
    {
      "epoch": 1.69,
      "grad_norm": 0.9894852638244629,
      "learning_rate": 0.0002397916028429707,
      "loss": 2.0072,
      "step": 130021
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.008360743522644,
      "learning_rate": 0.00023978759551361107,
      "loss": 2.066,
      "step": 130022
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.0022953748703003,
      "learning_rate": 0.0002397835881954459,
      "loss": 2.0628,
      "step": 130023
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.1392779350280762,
      "learning_rate": 0.0002397795808884761,
      "loss": 2.223,
      "step": 130024
    },
    {
      "epoch": 1.69,
      "grad_norm": 0.9970415234565735,
      "learning_rate": 0.00023977557359270243,
      "loss": 2.1426,
      "step": 130025
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.0996536016464233,
      "learning_rate": 0.00023977156630812562,
      "loss": 2.2473,
      "step": 130026
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.0094752311706543,
      "learning_rate": 0.0002397675590347464,
      "loss": 2.4693,
      "step": 130027
    },
    {
      "epoch": 1.69,
      "grad_norm": 0.9680350422859192,
      "learning_rate": 0.00023976355177256564,
      "loss": 2.0391,
      "step": 130028
    },
    {
      "epoch": 1.69,
      "grad_norm": 0.9388006329536438,
      "learning_rate": 0.00023975954452158386,
      "loss": 2.1988,
      "step": 130029
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.0676040649414062,
      "learning_rate": 0.00023975553728180195,
      "loss": 2.1608,
      "step": 130030
    },
    {
      "epoch": 1.69,
      "grad_norm": 0.999045729637146,
      "learning_rate": 0.00023975153005322057,
      "loss": 1.8986,
      "step": 130031
    },
    {
      "epoch": 1.69,
      "grad_norm": 0.9948534369468689,
      "learning_rate": 0.00023974752283584052,
      "loss": 2.0732,
      "step": 130032
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.0549299716949463,
      "learning_rate": 0.00023974351562966258,
      "loss": 2.0486,
      "step": 130033
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.0394519567489624,
      "learning_rate": 0.00023973950843468757,
      "loss": 2.3277,
      "step": 130034
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.214483380317688,
      "learning_rate": 0.000239735501250916,
      "loss": 1.9461,
      "step": 130035
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.0854835510253906,
      "learning_rate": 0.00023973149407834874,
      "loss": 2.0862,
      "step": 130036
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.07477867603302,
      "learning_rate": 0.0002397274869169865,
      "loss": 2.2387,
      "step": 130037
    },
    {
      "epoch": 1.69,
      "grad_norm": 0.9832383394241333,
      "learning_rate": 0.0002397234797668301,
      "loss": 2.1875,
      "step": 130038
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.196688175201416,
      "learning_rate": 0.00023971947262788022,
      "loss": 2.1258,
      "step": 130039
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.204952359199524,
      "learning_rate": 0.00023971546550013775,
      "loss": 2.1954,
      "step": 130040
    },
    {
      "epoch": 1.69,
      "grad_norm": 0.9763051867485046,
      "learning_rate": 0.0002397114583836032,
      "loss": 1.8725,
      "step": 130041
    },
    {
      "epoch": 1.69,
      "grad_norm": 0.9612217545509338,
      "learning_rate": 0.00023970745127827742,
      "loss": 2.0494,
      "step": 130042
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.0428497791290283,
      "learning_rate": 0.00023970344418416115,
      "loss": 2.2537,
      "step": 130043
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.0036922693252563,
      "learning_rate": 0.00023969943710125518,
      "loss": 2.1911,
      "step": 130044
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.0883961915969849,
      "learning_rate": 0.00023969543002956022,
      "loss": 2.0618,
      "step": 130045
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.0573877096176147,
      "learning_rate": 0.00023969142296907713,
      "loss": 2.1426,
      "step": 130046
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.0836467742919922,
      "learning_rate": 0.0002396874159198064,
      "loss": 2.2746,
      "step": 130047
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.0752344131469727,
      "learning_rate": 0.00023968340888174894,
      "loss": 2.0711,
      "step": 130048
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.0650367736816406,
      "learning_rate": 0.00023967940185490547,
      "loss": 2.0908,
      "step": 130049
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.0884898900985718,
      "learning_rate": 0.00023967539483927674,
      "loss": 2.1665,
      "step": 130050
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.085118055343628,
      "learning_rate": 0.00023967138783486348,
      "loss": 1.9872,
      "step": 130051
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.2305415868759155,
      "learning_rate": 0.00023966738084166647,
      "loss": 2.0346,
      "step": 130052
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.1719263792037964,
      "learning_rate": 0.00023966337385968655,
      "loss": 2.3423,
      "step": 130053
    },
    {
      "epoch": 1.69,
      "grad_norm": 0.9794095754623413,
      "learning_rate": 0.0002396593668889242,
      "loss": 2.1893,
      "step": 130054
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.050761342048645,
      "learning_rate": 0.00023965535992938038,
      "loss": 2.0445,
      "step": 130055
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.1011512279510498,
      "learning_rate": 0.0002396513529810557,
      "loss": 1.9873,
      "step": 130056
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.335252285003662,
      "learning_rate": 0.00023964734604395096,
      "loss": 2.1874,
      "step": 130057
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.2279270887374878,
      "learning_rate": 0.00023964333911806696,
      "loss": 2.1785,
      "step": 130058
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.5125577449798584,
      "learning_rate": 0.00023963933220340456,
      "loss": 2.1291,
      "step": 130059
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.1293425559997559,
      "learning_rate": 0.00023963532529996414,
      "loss": 2.1387,
      "step": 130060
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.1237033605575562,
      "learning_rate": 0.00023963131840774667,
      "loss": 2.0492,
      "step": 130061
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.2234495878219604,
      "learning_rate": 0.00023962731152675295,
      "loss": 2.1311,
      "step": 130062
    },
    {
      "epoch": 1.69,
      "grad_norm": 0.9477266669273376,
      "learning_rate": 0.00023962330465698357,
      "loss": 2.0991,
      "step": 130063
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.0714515447616577,
      "learning_rate": 0.00023961929779843942,
      "loss": 2.2145,
      "step": 130064
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.0936411619186401,
      "learning_rate": 0.00023961529095112126,
      "loss": 2.2906,
      "step": 130065
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.1049548387527466,
      "learning_rate": 0.00023961128411502963,
      "loss": 2.1674,
      "step": 130066
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.078195333480835,
      "learning_rate": 0.00023960727729016544,
      "loss": 2.0787,
      "step": 130067
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.1405918598175049,
      "learning_rate": 0.00023960327047652937,
      "loss": 2.044,
      "step": 130068
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.1362764835357666,
      "learning_rate": 0.0002395992636741222,
      "loss": 2.3529,
      "step": 130069
    },
    {
      "epoch": 1.69,
      "grad_norm": 0.9833788275718689,
      "learning_rate": 0.00023959525688294466,
      "loss": 2.071,
      "step": 130070
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.1116926670074463,
      "learning_rate": 0.00023959125010299755,
      "loss": 1.9062,
      "step": 130071
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.0396580696105957,
      "learning_rate": 0.0002395872433342816,
      "loss": 2.1663,
      "step": 130072
    },
    {
      "epoch": 1.69,
      "grad_norm": 0.9108296036720276,
      "learning_rate": 0.00023958323657679749,
      "loss": 2.0165,
      "step": 130073
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.0690133571624756,
      "learning_rate": 0.00023957922983054592,
      "loss": 2.0073,
      "step": 130074
    },
    {
      "epoch": 1.69,
      "grad_norm": 0.9846065044403076,
      "learning_rate": 0.00023957522309552777,
      "loss": 2.4222,
      "step": 130075
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.3182671070098877,
      "learning_rate": 0.0002395712163717437,
      "loss": 2.3291,
      "step": 130076
    },
    {
      "epoch": 1.69,
      "grad_norm": 0.9362358450889587,
      "learning_rate": 0.00023956720965919454,
      "loss": 2.1303,
      "step": 130077
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.1654220819473267,
      "learning_rate": 0.0002395632029578809,
      "loss": 2.0871,
      "step": 130078
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.1989450454711914,
      "learning_rate": 0.00023955919626780371,
      "loss": 2.0258,
      "step": 130079
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.005124807357788,
      "learning_rate": 0.0002395551895889635,
      "loss": 2.4126,
      "step": 130080
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.0290230512619019,
      "learning_rate": 0.00023955118292136115,
      "loss": 2.0841,
      "step": 130081
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.0249685049057007,
      "learning_rate": 0.00023954717626499736,
      "loss": 2.2127,
      "step": 130082
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.0965291261672974,
      "learning_rate": 0.00023954316961987294,
      "loss": 1.9497,
      "step": 130083
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.1197723150253296,
      "learning_rate": 0.00023953916298598857,
      "loss": 2.0235,
      "step": 130084
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.2176285982131958,
      "learning_rate": 0.000239535156363345,
      "loss": 2.2074,
      "step": 130085
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.0974154472351074,
      "learning_rate": 0.00023953114975194302,
      "loss": 2.0256,
      "step": 130086
    },
    {
      "epoch": 1.69,
      "grad_norm": 0.9555925130844116,
      "learning_rate": 0.00023952714315178327,
      "loss": 1.8365,
      "step": 130087
    },
    {
      "epoch": 1.69,
      "grad_norm": 0.9895013570785522,
      "learning_rate": 0.00023952313656286662,
      "loss": 2.0064,
      "step": 130088
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.0187934637069702,
      "learning_rate": 0.00023951912998519377,
      "loss": 2.1851,
      "step": 130089
    },
    {
      "epoch": 1.69,
      "grad_norm": 0.979564368724823,
      "learning_rate": 0.00023951512341876543,
      "loss": 1.955,
      "step": 130090
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.0133733749389648,
      "learning_rate": 0.00023951111686358232,
      "loss": 2.213,
      "step": 130091
    },
    {
      "epoch": 1.69,
      "grad_norm": 0.9448419809341431,
      "learning_rate": 0.0002395071103196453,
      "loss": 2.1299,
      "step": 130092
    },
    {
      "epoch": 1.69,
      "grad_norm": 0.9494794011116028,
      "learning_rate": 0.000239503103786955,
      "loss": 2.1038,
      "step": 130093
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.0934491157531738,
      "learning_rate": 0.00023949909726551227,
      "loss": 2.1615,
      "step": 130094
    },
    {
      "epoch": 1.69,
      "grad_norm": 0.9269394278526306,
      "learning_rate": 0.00023949509075531784,
      "loss": 1.9684,
      "step": 130095
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.0666755437850952,
      "learning_rate": 0.00023949108425637235,
      "loss": 2.0584,
      "step": 130096
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.0108327865600586,
      "learning_rate": 0.0002394870777686766,
      "loss": 2.2143,
      "step": 130097
    },
    {
      "epoch": 1.69,
      "grad_norm": 0.9941754341125488,
      "learning_rate": 0.0002394830712922313,
      "loss": 2.2132,
      "step": 130098
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.074277400970459,
      "learning_rate": 0.0002394790648270373,
      "loss": 1.9049,
      "step": 130099
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.0464119911193848,
      "learning_rate": 0.00023947505837309526,
      "loss": 2.0947,
      "step": 130100
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.0775582790374756,
      "learning_rate": 0.0002394710519304061,
      "loss": 2.0776,
      "step": 130101
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.1308375597000122,
      "learning_rate": 0.00023946704549897025,
      "loss": 1.9671,
      "step": 130102
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.1903786659240723,
      "learning_rate": 0.00023946303907878863,
      "loss": 2.1009,
      "step": 130103
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.1167901754379272,
      "learning_rate": 0.00023945903266986199,
      "loss": 2.0727,
      "step": 130104
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.0084233283996582,
      "learning_rate": 0.00023945502627219104,
      "loss": 2.4877,
      "step": 130105
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.3549492359161377,
      "learning_rate": 0.00023945101988577655,
      "loss": 1.8423,
      "step": 130106
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.119248390197754,
      "learning_rate": 0.00023944701351061938,
      "loss": 2.0973,
      "step": 130107
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.031502604484558,
      "learning_rate": 0.00023944300714672006,
      "loss": 2.1577,
      "step": 130108
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.0686684846878052,
      "learning_rate": 0.00023943900079407938,
      "loss": 2.1413,
      "step": 130109
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.1390514373779297,
      "learning_rate": 0.00023943499445269818,
      "loss": 2.3905,
      "step": 130110
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.0205106735229492,
      "learning_rate": 0.00023943098812257715,
      "loss": 1.9173,
      "step": 130111
    },
    {
      "epoch": 1.69,
      "grad_norm": 0.8961402773857117,
      "learning_rate": 0.00023942698180371702,
      "loss": 2.0052,
      "step": 130112
    },
    {
      "epoch": 1.69,
      "grad_norm": 0.9873186945915222,
      "learning_rate": 0.00023942297549611868,
      "loss": 2.0663,
      "step": 130113
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.18128502368927,
      "learning_rate": 0.00023941896919978264,
      "loss": 2.0876,
      "step": 130114
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.1025981903076172,
      "learning_rate": 0.00023941496291470974,
      "loss": 2.1174,
      "step": 130115
    },
    {
      "epoch": 1.69,
      "grad_norm": 0.9155483245849609,
      "learning_rate": 0.00023941095664090076,
      "loss": 2.2114,
      "step": 130116
    },
    {
      "epoch": 1.69,
      "grad_norm": 0.940220057964325,
      "learning_rate": 0.00023940695037835642,
      "loss": 2.003,
      "step": 130117
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.1459627151489258,
      "learning_rate": 0.00023940294412707746,
      "loss": 2.5088,
      "step": 130118
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.0314548015594482,
      "learning_rate": 0.00023939893788706466,
      "loss": 2.1103,
      "step": 130119
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.0005104541778564,
      "learning_rate": 0.00023939493165831886,
      "loss": 2.2007,
      "step": 130120
    },
    {
      "epoch": 1.69,
      "grad_norm": 0.9541640877723694,
      "learning_rate": 0.00023939092544084055,
      "loss": 2.1471,
      "step": 130121
    },
    {
      "epoch": 1.69,
      "grad_norm": 0.9654483199119568,
      "learning_rate": 0.00023938691923463062,
      "loss": 2.0718,
      "step": 130122
    },
    {
      "epoch": 1.69,
      "grad_norm": 0.9589769244194031,
      "learning_rate": 0.00023938291303968979,
      "loss": 2.0693,
      "step": 130123
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.0854262113571167,
      "learning_rate": 0.00023937890685601885,
      "loss": 2.1753,
      "step": 130124
    },
    {
      "epoch": 1.69,
      "grad_norm": 0.9766823649406433,
      "learning_rate": 0.0002393749006836185,
      "loss": 2.0995,
      "step": 130125
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.122999668121338,
      "learning_rate": 0.00023937089452248964,
      "loss": 2.0498,
      "step": 130126
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.2552930116653442,
      "learning_rate": 0.00023936688837263273,
      "loss": 1.9929,
      "step": 130127
    },
    {
      "epoch": 1.69,
      "grad_norm": 0.9465699195861816,
      "learning_rate": 0.00023936288223404866,
      "loss": 2.2237,
      "step": 130128
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.0387998819351196,
      "learning_rate": 0.00023935887610673816,
      "loss": 2.0841,
      "step": 130129
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.0055015087127686,
      "learning_rate": 0.00023935486999070204,
      "loss": 2.1365,
      "step": 130130
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.049761414527893,
      "learning_rate": 0.00023935086388594094,
      "loss": 2.1569,
      "step": 130131
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.1683353185653687,
      "learning_rate": 0.00023934685779245584,
      "loss": 1.9575,
      "step": 130132
    },
    {
      "epoch": 1.69,
      "grad_norm": 0.9274874329566956,
      "learning_rate": 0.00023934285171024714,
      "loss": 2.2073,
      "step": 130133
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.0428602695465088,
      "learning_rate": 0.00023933884563931575,
      "loss": 2.0238,
      "step": 130134
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.5902093648910522,
      "learning_rate": 0.00023933483957966242,
      "loss": 2.1761,
      "step": 130135
    },
    {
      "epoch": 1.69,
      "grad_norm": 0.8976858854293823,
      "learning_rate": 0.00023933083353128788,
      "loss": 2.1504,
      "step": 130136
    },
    {
      "epoch": 1.69,
      "grad_norm": 0.8466507196426392,
      "learning_rate": 0.0002393268274941929,
      "loss": 2.1501,
      "step": 130137
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.2038283348083496,
      "learning_rate": 0.00023932282146837833,
      "loss": 1.8797,
      "step": 130138
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.2657098770141602,
      "learning_rate": 0.00023931881545384467,
      "loss": 1.9864,
      "step": 130139
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.4040063619613647,
      "learning_rate": 0.00023931480945059278,
      "loss": 1.9409,
      "step": 130140
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.051449179649353,
      "learning_rate": 0.0002393108034586234,
      "loss": 2.188,
      "step": 130141
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.0388034582138062,
      "learning_rate": 0.00023930679747793734,
      "loss": 2.2118,
      "step": 130142
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.0835649967193604,
      "learning_rate": 0.00023930279150853522,
      "loss": 2.2098,
      "step": 130143
    },
    {
      "epoch": 1.69,
      "grad_norm": 0.9998072981834412,
      "learning_rate": 0.00023929878555041805,
      "loss": 1.9983,
      "step": 130144
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.0105417966842651,
      "learning_rate": 0.00023929477960358617,
      "loss": 2.3358,
      "step": 130145
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.0047153234481812,
      "learning_rate": 0.0002392907736680406,
      "loss": 2.4354,
      "step": 130146
    },
    {
      "epoch": 1.69,
      "grad_norm": 0.9895645976066589,
      "learning_rate": 0.00023928676774378203,
      "loss": 2.1066,
      "step": 130147
    },
    {
      "epoch": 1.69,
      "grad_norm": 0.983128547668457,
      "learning_rate": 0.00023928276183081117,
      "loss": 2.1744,
      "step": 130148
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.0354394912719727,
      "learning_rate": 0.00023927875592912878,
      "loss": 2.2212,
      "step": 130149
    },
    {
      "epoch": 1.69,
      "grad_norm": 0.9545372724533081,
      "learning_rate": 0.00023927475003873575,
      "loss": 2.115,
      "step": 130150
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.145934820175171,
      "learning_rate": 0.0002392707441596326,
      "loss": 2.2658,
      "step": 130151
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.9423421621322632,
      "learning_rate": 0.00023926673829182012,
      "loss": 2.1505,
      "step": 130152
    },
    {
      "epoch": 1.69,
      "grad_norm": 0.9670588970184326,
      "learning_rate": 0.0002392627324352991,
      "loss": 2.141,
      "step": 130153
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.0837061405181885,
      "learning_rate": 0.00023925872659007032,
      "loss": 2.0378,
      "step": 130154
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.293919324874878,
      "learning_rate": 0.0002392547207561345,
      "loss": 2.0816,
      "step": 130155
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.0981427431106567,
      "learning_rate": 0.00023925071493349239,
      "loss": 1.8499,
      "step": 130156
    },
    {
      "epoch": 1.69,
      "grad_norm": 0.9913696646690369,
      "learning_rate": 0.00023924670912214472,
      "loss": 2.2826,
      "step": 130157
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.0491752624511719,
      "learning_rate": 0.0002392427033220922,
      "loss": 2.0402,
      "step": 130158
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.1412049531936646,
      "learning_rate": 0.00023923869753333558,
      "loss": 1.9604,
      "step": 130159
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.1980714797973633,
      "learning_rate": 0.00023923469175587565,
      "loss": 2.2921,
      "step": 130160
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.0339821577072144,
      "learning_rate": 0.00023923068598971314,
      "loss": 1.9808,
      "step": 130161
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.0254530906677246,
      "learning_rate": 0.00023922668023484883,
      "loss": 2.0502,
      "step": 130162
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.0476713180541992,
      "learning_rate": 0.0002392226744912834,
      "loss": 2.1486,
      "step": 130163
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.0933921337127686,
      "learning_rate": 0.00023921866875901767,
      "loss": 2.1809,
      "step": 130164
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.0978248119354248,
      "learning_rate": 0.00023921466303805227,
      "loss": 2.2159,
      "step": 130165
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.0087754726409912,
      "learning_rate": 0.000239210657328388,
      "loss": 2.1895,
      "step": 130166
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.240251898765564,
      "learning_rate": 0.00023920665163002565,
      "loss": 2.0809,
      "step": 130167
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.040063738822937,
      "learning_rate": 0.00023920264594296596,
      "loss": 2.3108,
      "step": 130168
    },
    {
      "epoch": 1.69,
      "grad_norm": 0.965705931186676,
      "learning_rate": 0.00023919864026720962,
      "loss": 2.176,
      "step": 130169
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.000315546989441,
      "learning_rate": 0.0002391946346027574,
      "loss": 2.0023,
      "step": 130170
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.0696964263916016,
      "learning_rate": 0.00023919062894961008,
      "loss": 2.282,
      "step": 130171
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.028831124305725,
      "learning_rate": 0.00023918662330776828,
      "loss": 2.2903,
      "step": 130172
    },
    {
      "epoch": 1.69,
      "grad_norm": 0.9597891569137573,
      "learning_rate": 0.0002391826176772329,
      "loss": 1.8872,
      "step": 130173
    },
    {
      "epoch": 1.69,
      "grad_norm": 0.9853622317314148,
      "learning_rate": 0.00023917861205800464,
      "loss": 2.0388,
      "step": 130174
    },
    {
      "epoch": 1.69,
      "grad_norm": 0.9565991163253784,
      "learning_rate": 0.00023917460645008417,
      "loss": 2.2722,
      "step": 130175
    },
    {
      "epoch": 1.69,
      "grad_norm": 1.0566816329956055,
      "learning_rate": 0.00023917060085347228,
      "loss": 2.1071,
      "step": 130176
    },
    {
      "epoch": 1.69,
      "grad_norm": 0.9449779987335205,
      "learning_rate": 0.00023916659526816974,
      "loss": 2.2582,
      "step": 130177
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.2116633653640747,
      "learning_rate": 0.00023916258969417734,
      "loss": 2.1672,
      "step": 130178
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.2198258638381958,
      "learning_rate": 0.0002391585841314957,
      "loss": 2.096,
      "step": 130179
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.0104923248291016,
      "learning_rate": 0.0002391545785801257,
      "loss": 2.0456,
      "step": 130180
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.1520289182662964,
      "learning_rate": 0.0002391505730400679,
      "loss": 2.2905,
      "step": 130181
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.17648446559906,
      "learning_rate": 0.00023914656751132323,
      "loss": 2.0741,
      "step": 130182
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.0893876552581787,
      "learning_rate": 0.00023914256199389234,
      "loss": 2.201,
      "step": 130183
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.1247769594192505,
      "learning_rate": 0.00023913855648777593,
      "loss": 2.2931,
      "step": 130184
    },
    {
      "epoch": 1.7,
      "grad_norm": 0.951286256313324,
      "learning_rate": 0.00023913455099297488,
      "loss": 2.0089,
      "step": 130185
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.1312135457992554,
      "learning_rate": 0.00023913054550948992,
      "loss": 2.2234,
      "step": 130186
    },
    {
      "epoch": 1.7,
      "grad_norm": 0.9767636656761169,
      "learning_rate": 0.00023912654003732173,
      "loss": 2.1522,
      "step": 130187
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.0099890232086182,
      "learning_rate": 0.00023912253457647099,
      "loss": 2.1138,
      "step": 130188
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.0006296634674072,
      "learning_rate": 0.00023911852912693855,
      "loss": 2.2194,
      "step": 130189
    },
    {
      "epoch": 1.7,
      "grad_norm": 0.9835348725318909,
      "learning_rate": 0.0002391145236887251,
      "loss": 2.1494,
      "step": 130190
    },
    {
      "epoch": 1.7,
      "grad_norm": 0.9844805002212524,
      "learning_rate": 0.0002391105182618314,
      "loss": 2.1015,
      "step": 130191
    },
    {
      "epoch": 1.7,
      "grad_norm": 0.9786633849143982,
      "learning_rate": 0.00023910651284625827,
      "loss": 2.0051,
      "step": 130192
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.011550784111023,
      "learning_rate": 0.00023910250744200645,
      "loss": 2.0857,
      "step": 130193
    },
    {
      "epoch": 1.7,
      "grad_norm": 0.902633547782898,
      "learning_rate": 0.00023909850204907653,
      "loss": 2.3745,
      "step": 130194
    },
    {
      "epoch": 1.7,
      "grad_norm": 0.9936692118644714,
      "learning_rate": 0.0002390944966674693,
      "loss": 2.0784,
      "step": 130195
    },
    {
      "epoch": 1.7,
      "grad_norm": 0.9797893166542053,
      "learning_rate": 0.0002390904912971856,
      "loss": 2.2059,
      "step": 130196
    },
    {
      "epoch": 1.7,
      "grad_norm": 0.9653200507164001,
      "learning_rate": 0.00023908648593822613,
      "loss": 2.1283,
      "step": 130197
    },
    {
      "epoch": 1.7,
      "grad_norm": 0.9635717272758484,
      "learning_rate": 0.0002390824805905916,
      "loss": 2.1107,
      "step": 130198
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.2146798372268677,
      "learning_rate": 0.00023907847525428294,
      "loss": 2.1982,
      "step": 130199
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.0486931800842285,
      "learning_rate": 0.0002390744699293006,
      "loss": 2.0756,
      "step": 130200
    },
    {
      "epoch": 1.7,
      "grad_norm": 0.9611776471138,
      "learning_rate": 0.00023907046461564546,
      "loss": 2.1308,
      "step": 130201
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.209579348564148,
      "learning_rate": 0.00023906645931331827,
      "loss": 2.1075,
      "step": 130202
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.0435926914215088,
      "learning_rate": 0.00023906245402231978,
      "loss": 2.0559,
      "step": 130203
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.1162222623825073,
      "learning_rate": 0.00023905844874265076,
      "loss": 2.0888,
      "step": 130204
    },
    {
      "epoch": 1.7,
      "grad_norm": 0.9608123302459717,
      "learning_rate": 0.00023905444347431203,
      "loss": 2.1272,
      "step": 130205
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.1317867040634155,
      "learning_rate": 0.00023905043821730406,
      "loss": 2.1322,
      "step": 130206
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.1503695249557495,
      "learning_rate": 0.00023904643297162782,
      "loss": 2.2611,
      "step": 130207
    },
    {
      "epoch": 1.7,
      "grad_norm": 0.9279075264930725,
      "learning_rate": 0.000239042427737284,
      "loss": 1.8896,
      "step": 130208
    },
    {
      "epoch": 1.7,
      "grad_norm": 0.9569937586784363,
      "learning_rate": 0.0002390384225142733,
      "loss": 2.1543,
      "step": 130209
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.1507104635238647,
      "learning_rate": 0.00023903441730259653,
      "loss": 2.1857,
      "step": 130210
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.0315616130828857,
      "learning_rate": 0.00023903041210225455,
      "loss": 2.0248,
      "step": 130211
    },
    {
      "epoch": 1.7,
      "grad_norm": 0.9704416990280151,
      "learning_rate": 0.00023902640691324785,
      "loss": 2.1223,
      "step": 130212
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.1894983053207397,
      "learning_rate": 0.0002390224017355773,
      "loss": 2.2355,
      "step": 130213
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.248239517211914,
      "learning_rate": 0.0002390183965692436,
      "loss": 2.1019,
      "step": 130214
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.0309686660766602,
      "learning_rate": 0.00023901439141424754,
      "loss": 2.1519,
      "step": 130215
    },
    {
      "epoch": 1.7,
      "grad_norm": 0.9754183888435364,
      "learning_rate": 0.00023901038627058984,
      "loss": 2.1192,
      "step": 130216
    },
    {
      "epoch": 1.7,
      "grad_norm": 0.9353978037834167,
      "learning_rate": 0.0002390063811382714,
      "loss": 2.0938,
      "step": 130217
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.0682488679885864,
      "learning_rate": 0.00023900237601729273,
      "loss": 1.9684,
      "step": 130218
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.0448018312454224,
      "learning_rate": 0.00023899837090765464,
      "loss": 1.9712,
      "step": 130219
    },
    {
      "epoch": 1.7,
      "grad_norm": 0.9440944194793701,
      "learning_rate": 0.0002389943658093579,
      "loss": 2.0833,
      "step": 130220
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.0463520288467407,
      "learning_rate": 0.00023899036072240324,
      "loss": 2.1998,
      "step": 130221
    },
    {
      "epoch": 1.7,
      "grad_norm": 0.9841830730438232,
      "learning_rate": 0.00023898635564679148,
      "loss": 2.1664,
      "step": 130222
    },
    {
      "epoch": 1.7,
      "grad_norm": 0.9946655631065369,
      "learning_rate": 0.0002389823505825234,
      "loss": 2.2081,
      "step": 130223
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.0682761669158936,
      "learning_rate": 0.00023897834552959955,
      "loss": 2.0782,
      "step": 130224
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.0280001163482666,
      "learning_rate": 0.00023897434048802073,
      "loss": 2.1708,
      "step": 130225
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.0497682094573975,
      "learning_rate": 0.0002389703354577878,
      "loss": 2.0314,
      "step": 130226
    },
    {
      "epoch": 1.7,
      "grad_norm": 0.9589657187461853,
      "learning_rate": 0.00023896633043890138,
      "loss": 2.0219,
      "step": 130227
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.0495967864990234,
      "learning_rate": 0.00023896232543136228,
      "loss": 2.1787,
      "step": 130228
    },
    {
      "epoch": 1.7,
      "grad_norm": 0.959973156452179,
      "learning_rate": 0.0002389583204351714,
      "loss": 2.0594,
      "step": 130229
    },
    {
      "epoch": 1.7,
      "grad_norm": 0.9598867893218994,
      "learning_rate": 0.00023895431545032915,
      "loss": 2.0679,
      "step": 130230
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.03444242477417,
      "learning_rate": 0.0002389503104768365,
      "loss": 2.2658,
      "step": 130231
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.1020482778549194,
      "learning_rate": 0.00023894630551469405,
      "loss": 2.5066,
      "step": 130232
    },
    {
      "epoch": 1.7,
      "grad_norm": 0.9811159372329712,
      "learning_rate": 0.0002389423005639027,
      "loss": 2.1569,
      "step": 130233
    },
    {
      "epoch": 1.7,
      "grad_norm": 0.9838842749595642,
      "learning_rate": 0.0002389382956244631,
      "loss": 2.2655,
      "step": 130234
    },
    {
      "epoch": 1.7,
      "grad_norm": 0.8952180743217468,
      "learning_rate": 0.00023893429069637618,
      "loss": 2.1738,
      "step": 130235
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.0474791526794434,
      "learning_rate": 0.0002389302857796424,
      "loss": 2.0438,
      "step": 130236
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.2313129901885986,
      "learning_rate": 0.0002389262808742626,
      "loss": 2.0352,
      "step": 130237
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.0525964498519897,
      "learning_rate": 0.00023892227598023758,
      "loss": 2.0245,
      "step": 130238
    },
    {
      "epoch": 1.7,
      "grad_norm": 0.9800266027450562,
      "learning_rate": 0.00023891827109756805,
      "loss": 2.2246,
      "step": 130239
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.070895791053772,
      "learning_rate": 0.00023891426622625478,
      "loss": 2.0265,
      "step": 130240
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.1927906274795532,
      "learning_rate": 0.00023891026136629861,
      "loss": 2.1758,
      "step": 130241
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.146264910697937,
      "learning_rate": 0.00023890625651770006,
      "loss": 2.2784,
      "step": 130242
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.1489245891571045,
      "learning_rate": 0.00023890225168046,
      "loss": 1.8999,
      "step": 130243
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.1256064176559448,
      "learning_rate": 0.00023889824685457915,
      "loss": 1.9589,
      "step": 130244
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.1979544162750244,
      "learning_rate": 0.00023889424204005826,
      "loss": 2.1239,
      "step": 130245
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.0113731622695923,
      "learning_rate": 0.0002388902372368981,
      "loss": 2.2435,
      "step": 130246
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.1035892963409424,
      "learning_rate": 0.00023888623244509947,
      "loss": 2.0759,
      "step": 130247
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.0584386587142944,
      "learning_rate": 0.000238882227664663,
      "loss": 2.2269,
      "step": 130248
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.189620852470398,
      "learning_rate": 0.00023887822289558948,
      "loss": 2.0034,
      "step": 130249
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.324605941772461,
      "learning_rate": 0.00023887421813787964,
      "loss": 2.0965,
      "step": 130250
    },
    {
      "epoch": 1.7,
      "grad_norm": 0.9857586622238159,
      "learning_rate": 0.0002388702133915342,
      "loss": 2.1842,
      "step": 130251
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.036616563796997,
      "learning_rate": 0.00023886620865655396,
      "loss": 1.9574,
      "step": 130252
    },
    {
      "epoch": 1.7,
      "grad_norm": 0.9987624287605286,
      "learning_rate": 0.00023886220393293963,
      "loss": 2.0479,
      "step": 130253
    },
    {
      "epoch": 1.7,
      "grad_norm": 0.9963017106056213,
      "learning_rate": 0.00023885819922069205,
      "loss": 2.3015,
      "step": 130254
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.1706595420837402,
      "learning_rate": 0.00023885419451981184,
      "loss": 2.0608,
      "step": 130255
    },
    {
      "epoch": 1.7,
      "grad_norm": 0.9744559526443481,
      "learning_rate": 0.00023885018983029984,
      "loss": 2.0374,
      "step": 130256
    },
    {
      "epoch": 1.7,
      "grad_norm": 0.9420995712280273,
      "learning_rate": 0.00023884618515215662,
      "loss": 2.1937,
      "step": 130257
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.113020896911621,
      "learning_rate": 0.00023884218048538314,
      "loss": 2.3242,
      "step": 130258
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.1116832494735718,
      "learning_rate": 0.00023883817582998,
      "loss": 2.0703,
      "step": 130259
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.1260952949523926,
      "learning_rate": 0.0002388341711859481,
      "loss": 2.1001,
      "step": 130260
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.0132275819778442,
      "learning_rate": 0.000238830166553288,
      "loss": 2.1546,
      "step": 130261
    },
    {
      "epoch": 1.7,
      "grad_norm": 0.953268826007843,
      "learning_rate": 0.0002388261619320005,
      "loss": 2.093,
      "step": 130262
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.0696752071380615,
      "learning_rate": 0.00023882215732208645,
      "loss": 2.1783,
      "step": 130263
    },
    {
      "epoch": 1.7,
      "grad_norm": 0.9756569862365723,
      "learning_rate": 0.00023881815272354645,
      "loss": 1.9399,
      "step": 130264
    },
    {
      "epoch": 1.7,
      "grad_norm": 0.959923505783081,
      "learning_rate": 0.00023881414813638132,
      "loss": 2.0755,
      "step": 130265
    },
    {
      "epoch": 1.7,
      "grad_norm": 0.9992682337760925,
      "learning_rate": 0.00023881014356059188,
      "loss": 2.2092,
      "step": 130266
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.0530697107315063,
      "learning_rate": 0.00023880613899617865,
      "loss": 2.085,
      "step": 130267
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.0452497005462646,
      "learning_rate": 0.00023880213444314258,
      "loss": 2.2408,
      "step": 130268
    },
    {
      "epoch": 1.7,
      "grad_norm": 0.9912639856338501,
      "learning_rate": 0.00023879812990148434,
      "loss": 2.1159,
      "step": 130269
    },
    {
      "epoch": 1.7,
      "grad_norm": 0.9987345933914185,
      "learning_rate": 0.00023879412537120478,
      "loss": 2.2535,
      "step": 130270
    },
    {
      "epoch": 1.7,
      "grad_norm": 0.989205002784729,
      "learning_rate": 0.0002387901208523044,
      "loss": 2.1517,
      "step": 130271
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.1657477617263794,
      "learning_rate": 0.0002387861163447842,
      "loss": 1.9959,
      "step": 130272
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.0952918529510498,
      "learning_rate": 0.00023878211184864476,
      "loss": 2.1409,
      "step": 130273
    },
    {
      "epoch": 1.7,
      "grad_norm": 0.9778914451599121,
      "learning_rate": 0.0002387781073638869,
      "loss": 2.0539,
      "step": 130274
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.0167659521102905,
      "learning_rate": 0.0002387741028905113,
      "loss": 2.257,
      "step": 130275
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.0602099895477295,
      "learning_rate": 0.00023877009842851877,
      "loss": 2.1965,
      "step": 130276
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.2504732608795166,
      "learning_rate": 0.00023876609397791002,
      "loss": 2.1536,
      "step": 130277
    },
    {
      "epoch": 1.7,
      "grad_norm": 0.9699348211288452,
      "learning_rate": 0.00023876208953868593,
      "loss": 2.2798,
      "step": 130278
    },
    {
      "epoch": 1.7,
      "grad_norm": 0.9198086857795715,
      "learning_rate": 0.00023875808511084705,
      "loss": 2.1079,
      "step": 130279
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.1850950717926025,
      "learning_rate": 0.00023875408069439416,
      "loss": 2.1196,
      "step": 130280
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.1205333471298218,
      "learning_rate": 0.00023875007628932805,
      "loss": 2.2772,
      "step": 130281
    },
    {
      "epoch": 1.7,
      "grad_norm": 0.9453397393226624,
      "learning_rate": 0.00023874607189564945,
      "loss": 2.1191,
      "step": 130282
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.0487574338912964,
      "learning_rate": 0.00023874206751335912,
      "loss": 1.9736,
      "step": 130283
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.1239879131317139,
      "learning_rate": 0.00023873806314245793,
      "loss": 1.7564,
      "step": 130284
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.141714096069336,
      "learning_rate": 0.00023873405878294633,
      "loss": 2.1286,
      "step": 130285
    },
    {
      "epoch": 1.7,
      "grad_norm": 0.9901016354560852,
      "learning_rate": 0.00023873005443482526,
      "loss": 2.3011,
      "step": 130286
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.077941656112671,
      "learning_rate": 0.0002387260500980954,
      "loss": 2.0163,
      "step": 130287
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.1394628286361694,
      "learning_rate": 0.00023872204577275754,
      "loss": 2.2445,
      "step": 130288
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.02080500125885,
      "learning_rate": 0.00023871804145881242,
      "loss": 2.1828,
      "step": 130289
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.117465853691101,
      "learning_rate": 0.0002387140371562609,
      "loss": 1.9773,
      "step": 130290
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.0476950407028198,
      "learning_rate": 0.00023871003286510343,
      "loss": 2.285,
      "step": 130291
    },
    {
      "epoch": 1.7,
      "grad_norm": 0.9907719492912292,
      "learning_rate": 0.00023870602858534098,
      "loss": 2.029,
      "step": 130292
    },
    {
      "epoch": 1.7,
      "grad_norm": 0.9836321473121643,
      "learning_rate": 0.0002387020243169742,
      "loss": 2.1781,
      "step": 130293
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.027090311050415,
      "learning_rate": 0.00023869802006000384,
      "loss": 2.2276,
      "step": 130294
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.0778782367706299,
      "learning_rate": 0.00023869401581443073,
      "loss": 2.1458,
      "step": 130295
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.0350807905197144,
      "learning_rate": 0.0002386900115802557,
      "loss": 2.0628,
      "step": 130296
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.0163547992706299,
      "learning_rate": 0.00023868600735747918,
      "loss": 2.1652,
      "step": 130297
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.0709282159805298,
      "learning_rate": 0.0002386820031461021,
      "loss": 2.2222,
      "step": 130298
    },
    {
      "epoch": 1.7,
      "grad_norm": 0.9970813989639282,
      "learning_rate": 0.0002386779989461252,
      "loss": 2.2429,
      "step": 130299
    },
    {
      "epoch": 1.7,
      "grad_norm": 0.9836374521255493,
      "learning_rate": 0.00023867399475754924,
      "loss": 2.2213,
      "step": 130300
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.0236600637435913,
      "learning_rate": 0.0002386699905803749,
      "loss": 2.2568,
      "step": 130301
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.0153727531433105,
      "learning_rate": 0.0002386659864146031,
      "loss": 2.0904,
      "step": 130302
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.0007939338684082,
      "learning_rate": 0.00023866198226023434,
      "loss": 2.1382,
      "step": 130303
    },
    {
      "epoch": 1.7,
      "grad_norm": 0.9428853988647461,
      "learning_rate": 0.00023865797811726947,
      "loss": 2.0517,
      "step": 130304
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.0431313514709473,
      "learning_rate": 0.00023865397398570924,
      "loss": 2.1842,
      "step": 130305
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.0308420658111572,
      "learning_rate": 0.00023864996986555442,
      "loss": 2.197,
      "step": 130306
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.0208117961883545,
      "learning_rate": 0.00023864596575680567,
      "loss": 1.9629,
      "step": 130307
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.0404736995697021,
      "learning_rate": 0.00023864196165946394,
      "loss": 2.0695,
      "step": 130308
    },
    {
      "epoch": 1.7,
      "grad_norm": 0.9415261745452881,
      "learning_rate": 0.00023863795757352973,
      "loss": 2.1366,
      "step": 130309
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.0571640729904175,
      "learning_rate": 0.00023863395349900388,
      "loss": 2.165,
      "step": 130310
    },
    {
      "epoch": 1.7,
      "grad_norm": 0.9481743574142456,
      "learning_rate": 0.0002386299494358871,
      "loss": 2.0594,
      "step": 130311
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.0257853269577026,
      "learning_rate": 0.00023862594538418021,
      "loss": 2.2974,
      "step": 130312
    },
    {
      "epoch": 1.7,
      "grad_norm": 0.9080226421356201,
      "learning_rate": 0.0002386219413438839,
      "loss": 2.1186,
      "step": 130313
    },
    {
      "epoch": 1.7,
      "grad_norm": 0.9885629415512085,
      "learning_rate": 0.00023861793731499905,
      "loss": 2.0871,
      "step": 130314
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.1224790811538696,
      "learning_rate": 0.00023861393329752616,
      "loss": 2.3592,
      "step": 130315
    },
    {
      "epoch": 1.7,
      "grad_norm": 0.8989797234535217,
      "learning_rate": 0.00023860992929146613,
      "loss": 2.1055,
      "step": 130316
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.0033389329910278,
      "learning_rate": 0.0002386059252968196,
      "loss": 2.15,
      "step": 130317
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.0151976346969604,
      "learning_rate": 0.00023860192131358745,
      "loss": 2.2535,
      "step": 130318
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.2322312593460083,
      "learning_rate": 0.00023859791734177034,
      "loss": 2.2146,
      "step": 130319
    },
    {
      "epoch": 1.7,
      "grad_norm": 0.9679975509643555,
      "learning_rate": 0.00023859391338136904,
      "loss": 2.0245,
      "step": 130320
    },
    {
      "epoch": 1.7,
      "grad_norm": 0.9774382710456848,
      "learning_rate": 0.0002385899094323844,
      "loss": 2.0371,
      "step": 130321
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.1448806524276733,
      "learning_rate": 0.00023858590549481697,
      "loss": 2.2931,
      "step": 130322
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.080894947052002,
      "learning_rate": 0.00023858190156866754,
      "loss": 1.9519,
      "step": 130323
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.0023393630981445,
      "learning_rate": 0.0002385778976539369,
      "loss": 2.0643,
      "step": 130324
    },
    {
      "epoch": 1.7,
      "grad_norm": 0.9688151478767395,
      "learning_rate": 0.00023857389375062582,
      "loss": 1.854,
      "step": 130325
    },
    {
      "epoch": 1.7,
      "grad_norm": 0.9436686635017395,
      "learning_rate": 0.00023856988985873498,
      "loss": 2.2111,
      "step": 130326
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.0079798698425293,
      "learning_rate": 0.0002385658859782653,
      "loss": 2.0067,
      "step": 130327
    },
    {
      "epoch": 1.7,
      "grad_norm": 0.9164019823074341,
      "learning_rate": 0.00023856188210921722,
      "loss": 1.8944,
      "step": 130328
    },
    {
      "epoch": 1.7,
      "grad_norm": 0.9896190166473389,
      "learning_rate": 0.00023855787825159168,
      "loss": 2.0946,
      "step": 130329
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.0439590215682983,
      "learning_rate": 0.0002385538744053894,
      "loss": 2.1185,
      "step": 130330
    },
    {
      "epoch": 1.7,
      "grad_norm": 0.960905134677887,
      "learning_rate": 0.00023854987057061108,
      "loss": 2.0856,
      "step": 130331
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.0293794870376587,
      "learning_rate": 0.00023854586674725752,
      "loss": 1.9032,
      "step": 130332
    },
    {
      "epoch": 1.7,
      "grad_norm": 0.9798585176467896,
      "learning_rate": 0.00023854186293532958,
      "loss": 2.4691,
      "step": 130333
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.018630027770996,
      "learning_rate": 0.00023853785913482774,
      "loss": 2.0158,
      "step": 130334
    },
    {
      "epoch": 1.7,
      "grad_norm": 0.9874868988990784,
      "learning_rate": 0.00023853385534575286,
      "loss": 2.1558,
      "step": 130335
    },
    {
      "epoch": 1.7,
      "grad_norm": 0.998178243637085,
      "learning_rate": 0.00023852985156810568,
      "loss": 2.2001,
      "step": 130336
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.0879848003387451,
      "learning_rate": 0.00023852584780188697,
      "loss": 2.3739,
      "step": 130337
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.0417193174362183,
      "learning_rate": 0.0002385218440470975,
      "loss": 2.0417,
      "step": 130338
    },
    {
      "epoch": 1.7,
      "grad_norm": 0.9558707475662231,
      "learning_rate": 0.000238517840303738,
      "loss": 1.9771,
      "step": 130339
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.042987585067749,
      "learning_rate": 0.0002385138365718092,
      "loss": 1.8847,
      "step": 130340
    },
    {
      "epoch": 1.7,
      "grad_norm": 0.9963648915290833,
      "learning_rate": 0.0002385098328513118,
      "loss": 2.2276,
      "step": 130341
    },
    {
      "epoch": 1.7,
      "grad_norm": 0.9284188151359558,
      "learning_rate": 0.00023850582914224658,
      "loss": 2.107,
      "step": 130342
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.0289537906646729,
      "learning_rate": 0.00023850182544461426,
      "loss": 2.2734,
      "step": 130343
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.0221995115280151,
      "learning_rate": 0.00023849782175841563,
      "loss": 2.1299,
      "step": 130344
    },
    {
      "epoch": 1.7,
      "grad_norm": 0.9800264835357666,
      "learning_rate": 0.00023849381808365146,
      "loss": 2.0889,
      "step": 130345
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.0917925834655762,
      "learning_rate": 0.0002384898144203224,
      "loss": 2.194,
      "step": 130346
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.0845493078231812,
      "learning_rate": 0.00023848581076842926,
      "loss": 1.9535,
      "step": 130347
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.0212808847427368,
      "learning_rate": 0.00023848180712797282,
      "loss": 2.0738,
      "step": 130348
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.079538345336914,
      "learning_rate": 0.00023847780349895372,
      "loss": 2.0139,
      "step": 130349
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.0495439767837524,
      "learning_rate": 0.00023847379988137276,
      "loss": 2.1488,
      "step": 130350
    },
    {
      "epoch": 1.7,
      "grad_norm": 0.9543716907501221,
      "learning_rate": 0.0002384697962752307,
      "loss": 2.4368,
      "step": 130351
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.022086262702942,
      "learning_rate": 0.00023846579268052825,
      "loss": 2.273,
      "step": 130352
    },
    {
      "epoch": 1.7,
      "grad_norm": 0.9296482801437378,
      "learning_rate": 0.00023846178909726616,
      "loss": 1.9906,
      "step": 130353
    },
    {
      "epoch": 1.7,
      "grad_norm": 0.9373010396957397,
      "learning_rate": 0.00023845778552544517,
      "loss": 2.1681,
      "step": 130354
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.0329376459121704,
      "learning_rate": 0.00023845378196506615,
      "loss": 2.1759,
      "step": 130355
    },
    {
      "epoch": 1.7,
      "grad_norm": 0.9358367323875427,
      "learning_rate": 0.0002384497784161296,
      "loss": 1.9666,
      "step": 130356
    },
    {
      "epoch": 1.7,
      "grad_norm": 0.999437689781189,
      "learning_rate": 0.00023844577487863651,
      "loss": 2.1975,
      "step": 130357
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.037500262260437,
      "learning_rate": 0.00023844177135258745,
      "loss": 1.9859,
      "step": 130358
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.0770204067230225,
      "learning_rate": 0.0002384377678379832,
      "loss": 2.2256,
      "step": 130359
    },
    {
      "epoch": 1.7,
      "grad_norm": 0.9822421073913574,
      "learning_rate": 0.00023843376433482454,
      "loss": 2.2031,
      "step": 130360
    },
    {
      "epoch": 1.7,
      "grad_norm": 0.974517285823822,
      "learning_rate": 0.00023842976084311222,
      "loss": 2.1168,
      "step": 130361
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.005745768547058,
      "learning_rate": 0.000238425757362847,
      "loss": 2.0001,
      "step": 130362
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.0352368354797363,
      "learning_rate": 0.00023842175389402964,
      "loss": 2.2143,
      "step": 130363
    },
    {
      "epoch": 1.7,
      "grad_norm": 0.9536420702934265,
      "learning_rate": 0.00023841775043666072,
      "loss": 2.346,
      "step": 130364
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.1855026483535767,
      "learning_rate": 0.00023841374699074116,
      "loss": 1.9151,
      "step": 130365
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.031704306602478,
      "learning_rate": 0.00023840974355627162,
      "loss": 2.0101,
      "step": 130366
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.0269736051559448,
      "learning_rate": 0.00023840574013325285,
      "loss": 2.2568,
      "step": 130367
    },
    {
      "epoch": 1.7,
      "grad_norm": 0.9736940860748291,
      "learning_rate": 0.00023840173672168562,
      "loss": 2.2839,
      "step": 130368
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.0744749307632446,
      "learning_rate": 0.00023839773332157084,
      "loss": 2.3573,
      "step": 130369
    },
    {
      "epoch": 1.7,
      "grad_norm": 0.9980624914169312,
      "learning_rate": 0.00023839372993290895,
      "loss": 2.2287,
      "step": 130370
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.1130011081695557,
      "learning_rate": 0.00023838972655570079,
      "loss": 2.1002,
      "step": 130371
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.0983483791351318,
      "learning_rate": 0.00023838572318994717,
      "loss": 2.0802,
      "step": 130372
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.0639926195144653,
      "learning_rate": 0.0002383817198356488,
      "loss": 1.9672,
      "step": 130373
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.023726224899292,
      "learning_rate": 0.00023837771649280645,
      "loss": 2.3289,
      "step": 130374
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.1090242862701416,
      "learning_rate": 0.0002383737131614209,
      "loss": 2.2583,
      "step": 130375
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.2155429124832153,
      "learning_rate": 0.0002383697098414928,
      "loss": 1.9935,
      "step": 130376
    },
    {
      "epoch": 1.7,
      "grad_norm": 0.912761926651001,
      "learning_rate": 0.00023836570653302288,
      "loss": 1.8703,
      "step": 130377
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.0403236150741577,
      "learning_rate": 0.00023836170323601196,
      "loss": 1.894,
      "step": 130378
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.3695839643478394,
      "learning_rate": 0.0002383576999504608,
      "loss": 2.1158,
      "step": 130379
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.2393845319747925,
      "learning_rate": 0.00023835369667637006,
      "loss": 2.3131,
      "step": 130380
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.292428970336914,
      "learning_rate": 0.00023834969341374066,
      "loss": 2.1613,
      "step": 130381
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.174485206604004,
      "learning_rate": 0.0002383456901625731,
      "loss": 2.1715,
      "step": 130382
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.0422853231430054,
      "learning_rate": 0.00023834168692286824,
      "loss": 1.9261,
      "step": 130383
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.0081722736358643,
      "learning_rate": 0.0002383376836946268,
      "loss": 2.125,
      "step": 130384
    },
    {
      "epoch": 1.7,
      "grad_norm": 0.9739506840705872,
      "learning_rate": 0.00023833368047784957,
      "loss": 2.12,
      "step": 130385
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.2982466220855713,
      "learning_rate": 0.0002383296772725373,
      "loss": 2.1652,
      "step": 130386
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.0101486444473267,
      "learning_rate": 0.00023832567407869069,
      "loss": 2.2469,
      "step": 130387
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.0367509126663208,
      "learning_rate": 0.0002383216708963106,
      "loss": 2.0779,
      "step": 130388
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.020866870880127,
      "learning_rate": 0.00023831766772539756,
      "loss": 2.0638,
      "step": 130389
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.4711164236068726,
      "learning_rate": 0.00023831366456595243,
      "loss": 1.9515,
      "step": 130390
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.0725305080413818,
      "learning_rate": 0.00023830966141797597,
      "loss": 2.3181,
      "step": 130391
    },
    {
      "epoch": 1.7,
      "grad_norm": 0.9657913446426392,
      "learning_rate": 0.00023830565828146887,
      "loss": 2.3474,
      "step": 130392
    },
    {
      "epoch": 1.7,
      "grad_norm": 0.9425497055053711,
      "learning_rate": 0.00023830165515643196,
      "loss": 2.0518,
      "step": 130393
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.563570499420166,
      "learning_rate": 0.00023829765204286607,
      "loss": 2.1307,
      "step": 130394
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.1023337841033936,
      "learning_rate": 0.00023829364894077166,
      "loss": 2.2074,
      "step": 130395
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.025941014289856,
      "learning_rate": 0.0002382896458501496,
      "loss": 2.0918,
      "step": 130396
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.151857614517212,
      "learning_rate": 0.0002382856427710007,
      "loss": 2.1495,
      "step": 130397
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.3734474182128906,
      "learning_rate": 0.00023828163970332562,
      "loss": 2.1194,
      "step": 130398
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.5240328311920166,
      "learning_rate": 0.0002382776366471252,
      "loss": 1.9685,
      "step": 130399
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.1067222356796265,
      "learning_rate": 0.00023827363360240023,
      "loss": 2.1324,
      "step": 130400
    },
    {
      "epoch": 1.7,
      "grad_norm": 0.9620107412338257,
      "learning_rate": 0.00023826963056915125,
      "loss": 1.9436,
      "step": 130401
    },
    {
      "epoch": 1.7,
      "grad_norm": 0.9612191915512085,
      "learning_rate": 0.0002382656275473791,
      "loss": 2.1175,
      "step": 130402
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.111635446548462,
      "learning_rate": 0.0002382616245370845,
      "loss": 2.3002,
      "step": 130403
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.1567078828811646,
      "learning_rate": 0.00023825762153826827,
      "loss": 2.0424,
      "step": 130404
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.092611312866211,
      "learning_rate": 0.0002382536185509311,
      "loss": 2.0,
      "step": 130405
    },
    {
      "epoch": 1.7,
      "grad_norm": 0.9536213874816895,
      "learning_rate": 0.00023824961557507388,
      "loss": 2.1226,
      "step": 130406
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.0652416944503784,
      "learning_rate": 0.00023824561261069714,
      "loss": 2.2725,
      "step": 130407
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.0699392557144165,
      "learning_rate": 0.00023824160965780163,
      "loss": 2.2808,
      "step": 130408
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.0589510202407837,
      "learning_rate": 0.0002382376067163882,
      "loss": 2.0994,
      "step": 130409
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.1273188591003418,
      "learning_rate": 0.00023823360378645757,
      "loss": 1.9886,
      "step": 130410
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.028508186340332,
      "learning_rate": 0.00023822960086801047,
      "loss": 2.024,
      "step": 130411
    },
    {
      "epoch": 1.7,
      "grad_norm": 0.9195565581321716,
      "learning_rate": 0.0002382255979610478,
      "loss": 2.1155,
      "step": 130412
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.160222053527832,
      "learning_rate": 0.00023822159506557,
      "loss": 2.1323,
      "step": 130413
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.2122734785079956,
      "learning_rate": 0.000238217592181578,
      "loss": 2.1114,
      "step": 130414
    },
    {
      "epoch": 1.7,
      "grad_norm": 0.9601315259933472,
      "learning_rate": 0.0002382135893090725,
      "loss": 1.9311,
      "step": 130415
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.0292710065841675,
      "learning_rate": 0.00023820958644805426,
      "loss": 2.0305,
      "step": 130416
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.0822176933288574,
      "learning_rate": 0.00023820558359852404,
      "loss": 2.1913,
      "step": 130417
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.0847573280334473,
      "learning_rate": 0.00023820158076048268,
      "loss": 1.9144,
      "step": 130418
    },
    {
      "epoch": 1.7,
      "grad_norm": 0.9977517127990723,
      "learning_rate": 0.0002381975779339307,
      "loss": 2.1504,
      "step": 130419
    },
    {
      "epoch": 1.7,
      "grad_norm": 0.9495493173599243,
      "learning_rate": 0.00023819357511886894,
      "loss": 1.9712,
      "step": 130420
    },
    {
      "epoch": 1.7,
      "grad_norm": 0.9814050793647766,
      "learning_rate": 0.00023818957231529816,
      "loss": 2.1462,
      "step": 130421
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.0002118349075317,
      "learning_rate": 0.0002381855695232191,
      "loss": 2.1002,
      "step": 130422
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.0104690790176392,
      "learning_rate": 0.00023818156674263255,
      "loss": 2.042,
      "step": 130423
    },
    {
      "epoch": 1.7,
      "grad_norm": 0.9506247639656067,
      "learning_rate": 0.00023817756397353924,
      "loss": 2.0574,
      "step": 130424
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.057436227798462,
      "learning_rate": 0.0002381735612159399,
      "loss": 2.0871,
      "step": 130425
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.1462098360061646,
      "learning_rate": 0.0002381695584698352,
      "loss": 1.9968,
      "step": 130426
    },
    {
      "epoch": 1.7,
      "grad_norm": 0.9046106338500977,
      "learning_rate": 0.00023816555573522594,
      "loss": 1.865,
      "step": 130427
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.0910470485687256,
      "learning_rate": 0.00023816155301211288,
      "loss": 2.2968,
      "step": 130428
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.036879301071167,
      "learning_rate": 0.00023815755030049676,
      "loss": 2.2413,
      "step": 130429
    },
    {
      "epoch": 1.7,
      "grad_norm": 0.9445217251777649,
      "learning_rate": 0.00023815354760037833,
      "loss": 2.0259,
      "step": 130430
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.0305684804916382,
      "learning_rate": 0.0002381495449117583,
      "loss": 2.3193,
      "step": 130431
    },
    {
      "epoch": 1.7,
      "grad_norm": 0.954153299331665,
      "learning_rate": 0.00023814554223463753,
      "loss": 2.0816,
      "step": 130432
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.0103176832199097,
      "learning_rate": 0.00023814153956901654,
      "loss": 2.1657,
      "step": 130433
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.0297164916992188,
      "learning_rate": 0.00023813753691489625,
      "loss": 2.2734,
      "step": 130434
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.0527833700180054,
      "learning_rate": 0.00023813353427227737,
      "loss": 2.27,
      "step": 130435
    },
    {
      "epoch": 1.7,
      "grad_norm": 0.9994033575057983,
      "learning_rate": 0.00023812953164116068,
      "loss": 2.1903,
      "step": 130436
    },
    {
      "epoch": 1.7,
      "grad_norm": 0.9935006499290466,
      "learning_rate": 0.0002381255290215468,
      "loss": 1.878,
      "step": 130437
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.0006680488586426,
      "learning_rate": 0.00023812152641343657,
      "loss": 2.4251,
      "step": 130438
    },
    {
      "epoch": 1.7,
      "grad_norm": 0.9754485487937927,
      "learning_rate": 0.00023811752381683071,
      "loss": 1.9095,
      "step": 130439
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.0800985097885132,
      "learning_rate": 0.00023811352123173003,
      "loss": 1.9049,
      "step": 130440
    },
    {
      "epoch": 1.7,
      "grad_norm": 0.933485746383667,
      "learning_rate": 0.00023810951865813518,
      "loss": 2.1035,
      "step": 130441
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.1361957788467407,
      "learning_rate": 0.00023810551609604696,
      "loss": 2.1339,
      "step": 130442
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.3329771757125854,
      "learning_rate": 0.00023810151354546606,
      "loss": 1.9733,
      "step": 130443
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.0044806003570557,
      "learning_rate": 0.00023809751100639325,
      "loss": 2.3001,
      "step": 130444
    },
    {
      "epoch": 1.7,
      "grad_norm": 0.9751941561698914,
      "learning_rate": 0.00023809350847882926,
      "loss": 2.2728,
      "step": 130445
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.3885982036590576,
      "learning_rate": 0.00023808950596277488,
      "loss": 2.1539,
      "step": 130446
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.0884332656860352,
      "learning_rate": 0.00023808550345823088,
      "loss": 2.165,
      "step": 130447
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.036779761314392,
      "learning_rate": 0.00023808150096519797,
      "loss": 2.15,
      "step": 130448
    },
    {
      "epoch": 1.7,
      "grad_norm": 0.9656258225440979,
      "learning_rate": 0.00023807749848367679,
      "loss": 2.1095,
      "step": 130449
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.0383046865463257,
      "learning_rate": 0.0002380734960136682,
      "loss": 2.1655,
      "step": 130450
    },
    {
      "epoch": 1.7,
      "grad_norm": 0.8872287273406982,
      "learning_rate": 0.0002380694935551729,
      "loss": 2.2402,
      "step": 130451
    },
    {
      "epoch": 1.7,
      "grad_norm": 0.9894545674324036,
      "learning_rate": 0.00023806549110819166,
      "loss": 2.4076,
      "step": 130452
    },
    {
      "epoch": 1.7,
      "grad_norm": 0.9540814757347107,
      "learning_rate": 0.0002380614886727252,
      "loss": 2.0963,
      "step": 130453
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.0421826839447021,
      "learning_rate": 0.00023805748624877436,
      "loss": 2.048,
      "step": 130454
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.158143401145935,
      "learning_rate": 0.0002380534838363398,
      "loss": 2.214,
      "step": 130455
    },
    {
      "epoch": 1.7,
      "grad_norm": 0.9975886940956116,
      "learning_rate": 0.00023804948143542215,
      "loss": 1.9992,
      "step": 130456
    },
    {
      "epoch": 1.7,
      "grad_norm": 0.9689242839813232,
      "learning_rate": 0.00023804547904602232,
      "loss": 2.0013,
      "step": 130457
    },
    {
      "epoch": 1.7,
      "grad_norm": 0.9619513154029846,
      "learning_rate": 0.000238041476668141,
      "loss": 2.1407,
      "step": 130458
    },
    {
      "epoch": 1.7,
      "grad_norm": 0.9824753999710083,
      "learning_rate": 0.00023803747430177895,
      "loss": 2.0387,
      "step": 130459
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.0130356550216675,
      "learning_rate": 0.0002380334719469369,
      "loss": 2.1536,
      "step": 130460
    },
    {
      "epoch": 1.7,
      "grad_norm": 0.8960813283920288,
      "learning_rate": 0.00023802946960361574,
      "loss": 1.8507,
      "step": 130461
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.028017282485962,
      "learning_rate": 0.00023802546727181588,
      "loss": 1.9945,
      "step": 130462
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.1337987184524536,
      "learning_rate": 0.00023802146495153828,
      "loss": 1.9414,
      "step": 130463
    },
    {
      "epoch": 1.7,
      "grad_norm": 0.9687665104866028,
      "learning_rate": 0.0002380174626427837,
      "loss": 2.091,
      "step": 130464
    },
    {
      "epoch": 1.7,
      "grad_norm": 0.9717203974723816,
      "learning_rate": 0.0002380134603455528,
      "loss": 2.0836,
      "step": 130465
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.0923705101013184,
      "learning_rate": 0.0002380094580598464,
      "loss": 2.2844,
      "step": 130466
    },
    {
      "epoch": 1.7,
      "grad_norm": 0.9866470694541931,
      "learning_rate": 0.00023800545578566532,
      "loss": 2.0502,
      "step": 130467
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.0139758586883545,
      "learning_rate": 0.0002380014535230101,
      "loss": 2.1956,
      "step": 130468
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.017326831817627,
      "learning_rate": 0.00023799745127188154,
      "loss": 2.066,
      "step": 130469
    },
    {
      "epoch": 1.7,
      "grad_norm": 0.958094596862793,
      "learning_rate": 0.00023799344903228047,
      "loss": 2.007,
      "step": 130470
    },
    {
      "epoch": 1.7,
      "grad_norm": 0.9882992506027222,
      "learning_rate": 0.00023798944680420754,
      "loss": 2.1618,
      "step": 130471
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.1498507261276245,
      "learning_rate": 0.00023798544458766352,
      "loss": 2.0419,
      "step": 130472
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.0763832330703735,
      "learning_rate": 0.00023798144238264938,
      "loss": 2.0408,
      "step": 130473
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.0514940023422241,
      "learning_rate": 0.00023797744018916547,
      "loss": 2.2082,
      "step": 130474
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.1897914409637451,
      "learning_rate": 0.00023797343800721277,
      "loss": 2.2188,
      "step": 130475
    },
    {
      "epoch": 1.7,
      "grad_norm": 0.9767442345619202,
      "learning_rate": 0.00023796943583679196,
      "loss": 2.0289,
      "step": 130476
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.0043590068817139,
      "learning_rate": 0.0002379654336779038,
      "loss": 2.2665,
      "step": 130477
    },
    {
      "epoch": 1.7,
      "grad_norm": 0.9832340478897095,
      "learning_rate": 0.00023796143153054905,
      "loss": 2.318,
      "step": 130478
    },
    {
      "epoch": 1.7,
      "grad_norm": 0.9603267908096313,
      "learning_rate": 0.00023795742939472858,
      "loss": 2.1055,
      "step": 130479
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.2016310691833496,
      "learning_rate": 0.00023795342727044283,
      "loss": 2.0963,
      "step": 130480
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.6613329648971558,
      "learning_rate": 0.00023794942515769274,
      "loss": 2.0509,
      "step": 130481
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.0286266803741455,
      "learning_rate": 0.00023794542305647897,
      "loss": 2.0066,
      "step": 130482
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.2746721506118774,
      "learning_rate": 0.00023794142096680238,
      "loss": 2.1892,
      "step": 130483
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.150748610496521,
      "learning_rate": 0.00023793741888866363,
      "loss": 2.1806,
      "step": 130484
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.148369312286377,
      "learning_rate": 0.00023793341682206361,
      "loss": 2.0556,
      "step": 130485
    },
    {
      "epoch": 1.7,
      "grad_norm": 0.9205648303031921,
      "learning_rate": 0.00023792941476700278,
      "loss": 2.0895,
      "step": 130486
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.1663086414337158,
      "learning_rate": 0.00023792541272348207,
      "loss": 2.0434,
      "step": 130487
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.2407382726669312,
      "learning_rate": 0.00023792141069150222,
      "loss": 2.3459,
      "step": 130488
    },
    {
      "epoch": 1.7,
      "grad_norm": 0.9941222667694092,
      "learning_rate": 0.00023791740867106393,
      "loss": 2.0015,
      "step": 130489
    },
    {
      "epoch": 1.7,
      "grad_norm": 0.9986267685890198,
      "learning_rate": 0.000237913406662168,
      "loss": 1.9479,
      "step": 130490
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.0992084741592407,
      "learning_rate": 0.0002379094046648152,
      "loss": 2.2677,
      "step": 130491
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.1435961723327637,
      "learning_rate": 0.00023790540267900614,
      "loss": 2.1702,
      "step": 130492
    },
    {
      "epoch": 1.7,
      "grad_norm": 0.9776299595832825,
      "learning_rate": 0.00023790140070474163,
      "loss": 2.0228,
      "step": 130493
    },
    {
      "epoch": 1.7,
      "grad_norm": 0.8992648124694824,
      "learning_rate": 0.0002378973987420224,
      "loss": 2.1796,
      "step": 130494
    },
    {
      "epoch": 1.7,
      "grad_norm": 0.8933658003807068,
      "learning_rate": 0.00023789339679084924,
      "loss": 2.0944,
      "step": 130495
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.0959569215774536,
      "learning_rate": 0.00023788939485122284,
      "loss": 2.2193,
      "step": 130496
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.1522698402404785,
      "learning_rate": 0.00023788539292314413,
      "loss": 2.16,
      "step": 130497
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.0556057691574097,
      "learning_rate": 0.00023788139100661354,
      "loss": 2.1143,
      "step": 130498
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.0212087631225586,
      "learning_rate": 0.000237877389101632,
      "loss": 2.2454,
      "step": 130499
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.1909493207931519,
      "learning_rate": 0.00023787338720820023,
      "loss": 2.1149,
      "step": 130500
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.1491564512252808,
      "learning_rate": 0.00023786938532631893,
      "loss": 1.962,
      "step": 130501
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.0094244480133057,
      "learning_rate": 0.0002378653834559889,
      "loss": 2.2645,
      "step": 130502
    },
    {
      "epoch": 1.7,
      "grad_norm": 0.9604169726371765,
      "learning_rate": 0.00023786138159721107,
      "loss": 2.1587,
      "step": 130503
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.0091779232025146,
      "learning_rate": 0.00023785737974998575,
      "loss": 2.1662,
      "step": 130504
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.0826245546340942,
      "learning_rate": 0.00023785337791431397,
      "loss": 2.321,
      "step": 130505
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.0212050676345825,
      "learning_rate": 0.0002378493760901964,
      "loss": 2.0628,
      "step": 130506
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.0331577062606812,
      "learning_rate": 0.00023784537427763383,
      "loss": 2.0876,
      "step": 130507
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.0386946201324463,
      "learning_rate": 0.00023784137247662694,
      "loss": 2.0574,
      "step": 130508
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.089858055114746,
      "learning_rate": 0.00023783737068717664,
      "loss": 2.0767,
      "step": 130509
    },
    {
      "epoch": 1.7,
      "grad_norm": 0.9359412789344788,
      "learning_rate": 0.00023783336890928348,
      "loss": 1.9556,
      "step": 130510
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.079554796218872,
      "learning_rate": 0.00023782936714294823,
      "loss": 1.9566,
      "step": 130511
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.0968327522277832,
      "learning_rate": 0.00023782536538817164,
      "loss": 2.2054,
      "step": 130512
    },
    {
      "epoch": 1.7,
      "grad_norm": 0.9800997376441956,
      "learning_rate": 0.00023782136364495455,
      "loss": 2.1622,
      "step": 130513
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.0515878200531006,
      "learning_rate": 0.00023781736191329762,
      "loss": 2.1974,
      "step": 130514
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.0140501260757446,
      "learning_rate": 0.00023781336019320162,
      "loss": 2.2377,
      "step": 130515
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.0988414287567139,
      "learning_rate": 0.0002378093584846673,
      "loss": 2.0952,
      "step": 130516
    },
    {
      "epoch": 1.7,
      "grad_norm": 0.9178705811500549,
      "learning_rate": 0.00023780535678769545,
      "loss": 1.9636,
      "step": 130517
    },
    {
      "epoch": 1.7,
      "grad_norm": 0.9715396761894226,
      "learning_rate": 0.0002378013551022867,
      "loss": 2.1928,
      "step": 130518
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.2583955526351929,
      "learning_rate": 0.00023779735342844183,
      "loss": 2.0081,
      "step": 130519
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.0975475311279297,
      "learning_rate": 0.00023779335176616161,
      "loss": 2.5349,
      "step": 130520
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.0232276916503906,
      "learning_rate": 0.0002377893501154468,
      "loss": 2.1717,
      "step": 130521
    },
    {
      "epoch": 1.7,
      "grad_norm": 0.9900763034820557,
      "learning_rate": 0.00023778534847629815,
      "loss": 2.1485,
      "step": 130522
    },
    {
      "epoch": 1.7,
      "grad_norm": 0.9457248449325562,
      "learning_rate": 0.0002377813468487163,
      "loss": 2.2052,
      "step": 130523
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.081419825553894,
      "learning_rate": 0.0002377773452327022,
      "loss": 2.1215,
      "step": 130524
    },
    {
      "epoch": 1.7,
      "grad_norm": 0.9295032620429993,
      "learning_rate": 0.00023777334362825637,
      "loss": 2.3847,
      "step": 130525
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.0878016948699951,
      "learning_rate": 0.00023776934203537965,
      "loss": 1.9766,
      "step": 130526
    },
    {
      "epoch": 1.7,
      "grad_norm": 0.9986550807952881,
      "learning_rate": 0.0002377653404540728,
      "loss": 2.1124,
      "step": 130527
    },
    {
      "epoch": 1.7,
      "grad_norm": 0.9922153353691101,
      "learning_rate": 0.0002377613388843366,
      "loss": 2.0542,
      "step": 130528
    },
    {
      "epoch": 1.7,
      "grad_norm": 0.9996151328086853,
      "learning_rate": 0.00023775733732617167,
      "loss": 2.0483,
      "step": 130529
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.1103376150131226,
      "learning_rate": 0.00023775333577957884,
      "loss": 2.0015,
      "step": 130530
    },
    {
      "epoch": 1.7,
      "grad_norm": 0.9443767070770264,
      "learning_rate": 0.0002377493342445589,
      "loss": 1.9736,
      "step": 130531
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.125299096107483,
      "learning_rate": 0.00023774533272111247,
      "loss": 2.2246,
      "step": 130532
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.2519837617874146,
      "learning_rate": 0.00023774133120924034,
      "loss": 2.1953,
      "step": 130533
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.051976203918457,
      "learning_rate": 0.00023773732970894338,
      "loss": 2.1339,
      "step": 130534
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.0699923038482666,
      "learning_rate": 0.0002377333282202221,
      "loss": 2.2009,
      "step": 130535
    },
    {
      "epoch": 1.7,
      "grad_norm": 0.9786232113838196,
      "learning_rate": 0.0002377293267430774,
      "loss": 2.1118,
      "step": 130536
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.1115621328353882,
      "learning_rate": 0.00023772532527751,
      "loss": 2.0392,
      "step": 130537
    },
    {
      "epoch": 1.7,
      "grad_norm": 0.9522421360015869,
      "learning_rate": 0.00023772132382352066,
      "loss": 2.235,
      "step": 130538
    },
    {
      "epoch": 1.7,
      "grad_norm": 0.903684675693512,
      "learning_rate": 0.0002377173223811101,
      "loss": 1.9482,
      "step": 130539
    },
    {
      "epoch": 1.7,
      "grad_norm": 0.971480667591095,
      "learning_rate": 0.00023771332095027912,
      "loss": 1.8989,
      "step": 130540
    },
    {
      "epoch": 1.7,
      "grad_norm": 0.9401047825813293,
      "learning_rate": 0.0002377093195310283,
      "loss": 2.0351,
      "step": 130541
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.1626607179641724,
      "learning_rate": 0.00023770531812335852,
      "loss": 1.9006,
      "step": 130542
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.0167118310928345,
      "learning_rate": 0.00023770131672727054,
      "loss": 2.0494,
      "step": 130543
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.0423697233200073,
      "learning_rate": 0.000237697315342765,
      "loss": 2.1136,
      "step": 130544
    },
    {
      "epoch": 1.7,
      "grad_norm": 0.9296784996986389,
      "learning_rate": 0.00023769331396984272,
      "loss": 2.1717,
      "step": 130545
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.083032250404358,
      "learning_rate": 0.00023768931260850456,
      "loss": 1.8493,
      "step": 130546
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.1012494564056396,
      "learning_rate": 0.00023768531125875102,
      "loss": 2.0296,
      "step": 130547
    },
    {
      "epoch": 1.7,
      "grad_norm": 0.9542280435562134,
      "learning_rate": 0.0002376813099205829,
      "loss": 1.8975,
      "step": 130548
    },
    {
      "epoch": 1.7,
      "grad_norm": 0.9989879131317139,
      "learning_rate": 0.00023767730859400107,
      "loss": 1.9966,
      "step": 130549
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.0394668579101562,
      "learning_rate": 0.00023767330727900618,
      "loss": 2.0073,
      "step": 130550
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.2218235731124878,
      "learning_rate": 0.000237669305975599,
      "loss": 2.0653,
      "step": 130551
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.089127779006958,
      "learning_rate": 0.00023766530468378043,
      "loss": 2.1457,
      "step": 130552
    },
    {
      "epoch": 1.7,
      "grad_norm": 0.9926643967628479,
      "learning_rate": 0.00023766130340355089,
      "loss": 1.9536,
      "step": 130553
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.2028483152389526,
      "learning_rate": 0.00023765730213491128,
      "loss": 1.9002,
      "step": 130554
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.043407917022705,
      "learning_rate": 0.00023765330087786236,
      "loss": 2.1477,
      "step": 130555
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.0334274768829346,
      "learning_rate": 0.0002376492996324049,
      "loss": 2.0525,
      "step": 130556
    },
    {
      "epoch": 1.7,
      "grad_norm": 0.9788890480995178,
      "learning_rate": 0.0002376452983985396,
      "loss": 2.1571,
      "step": 130557
    },
    {
      "epoch": 1.7,
      "grad_norm": 0.9651594758033752,
      "learning_rate": 0.00023764129717626732,
      "loss": 1.8723,
      "step": 130558
    },
    {
      "epoch": 1.7,
      "grad_norm": 0.953606367111206,
      "learning_rate": 0.00023763729596558861,
      "loss": 1.9621,
      "step": 130559
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.0252373218536377,
      "learning_rate": 0.0002376332947665043,
      "loss": 2.0435,
      "step": 130560
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.100548267364502,
      "learning_rate": 0.00023762929357901507,
      "loss": 2.4723,
      "step": 130561
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.1453349590301514,
      "learning_rate": 0.0002376252924031218,
      "loss": 1.9767,
      "step": 130562
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.0211455821990967,
      "learning_rate": 0.00023762129123882516,
      "loss": 2.0767,
      "step": 130563
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.0029832124710083,
      "learning_rate": 0.000237617290086126,
      "loss": 2.063,
      "step": 130564
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.0093350410461426,
      "learning_rate": 0.00023761328894502485,
      "loss": 2.0323,
      "step": 130565
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.105015754699707,
      "learning_rate": 0.00023760928781552256,
      "loss": 2.3383,
      "step": 130566
    },
    {
      "epoch": 1.7,
      "grad_norm": 0.8900762796401978,
      "learning_rate": 0.0002376052866976199,
      "loss": 1.9817,
      "step": 130567
    },
    {
      "epoch": 1.7,
      "grad_norm": 0.9865502715110779,
      "learning_rate": 0.0002376012855913176,
      "loss": 1.9447,
      "step": 130568
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.006054401397705,
      "learning_rate": 0.00023759728449661638,
      "loss": 2.1105,
      "step": 130569
    },
    {
      "epoch": 1.7,
      "grad_norm": 0.956640899181366,
      "learning_rate": 0.00023759328341351713,
      "loss": 2.0815,
      "step": 130570
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.1183876991271973,
      "learning_rate": 0.0002375892823420203,
      "loss": 2.0094,
      "step": 130571
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.1062394380569458,
      "learning_rate": 0.00023758528128212689,
      "loss": 2.1837,
      "step": 130572
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.0017096996307373,
      "learning_rate": 0.00023758128023383747,
      "loss": 2.2058,
      "step": 130573
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.1592148542404175,
      "learning_rate": 0.0002375772791971529,
      "loss": 2.1614,
      "step": 130574
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.3148106336593628,
      "learning_rate": 0.00023757327817207393,
      "loss": 1.9791,
      "step": 130575
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.0403438806533813,
      "learning_rate": 0.00023756927715860136,
      "loss": 2.2519,
      "step": 130576
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.069107174873352,
      "learning_rate": 0.0002375652761567357,
      "loss": 2.2419,
      "step": 130577
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.2566213607788086,
      "learning_rate": 0.00023756127516647783,
      "loss": 2.1998,
      "step": 130578
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.1074862480163574,
      "learning_rate": 0.00023755727418782856,
      "loss": 1.8967,
      "step": 130579
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.0899182558059692,
      "learning_rate": 0.00023755327322078846,
      "loss": 2.0637,
      "step": 130580
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.140445590019226,
      "learning_rate": 0.0002375492722653585,
      "loss": 2.046,
      "step": 130581
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.0923964977264404,
      "learning_rate": 0.00023754527132153927,
      "loss": 2.2894,
      "step": 130582
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.17546546459198,
      "learning_rate": 0.00023754127038933166,
      "loss": 2.1925,
      "step": 130583
    },
    {
      "epoch": 1.7,
      "grad_norm": 0.9645187258720398,
      "learning_rate": 0.00023753726946873616,
      "loss": 1.9987,
      "step": 130584
    },
    {
      "epoch": 1.7,
      "grad_norm": 0.9851061105728149,
      "learning_rate": 0.00023753326855975373,
      "loss": 2.0475,
      "step": 130585
    },
    {
      "epoch": 1.7,
      "grad_norm": 0.9493728280067444,
      "learning_rate": 0.00023752926766238498,
      "loss": 1.9399,
      "step": 130586
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.0125117301940918,
      "learning_rate": 0.00023752526677663075,
      "loss": 2.1899,
      "step": 130587
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.0135164260864258,
      "learning_rate": 0.00023752126590249173,
      "loss": 2.0527,
      "step": 130588
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.0943440198898315,
      "learning_rate": 0.00023751726503996886,
      "loss": 2.1849,
      "step": 130589
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.0638408660888672,
      "learning_rate": 0.00023751326418906251,
      "loss": 2.0337,
      "step": 130590
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.123634934425354,
      "learning_rate": 0.0002375092633497737,
      "loss": 2.2373,
      "step": 130591
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.0281891822814941,
      "learning_rate": 0.00023750526252210307,
      "loss": 2.2374,
      "step": 130592
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.0553078651428223,
      "learning_rate": 0.00023750126170605138,
      "loss": 1.8389,
      "step": 130593
    },
    {
      "epoch": 1.7,
      "grad_norm": 0.9540854692459106,
      "learning_rate": 0.00023749726090161935,
      "loss": 2.2413,
      "step": 130594
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.0650278329849243,
      "learning_rate": 0.00023749326010880794,
      "loss": 2.1299,
      "step": 130595
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.0194435119628906,
      "learning_rate": 0.00023748925932761758,
      "loss": 2.168,
      "step": 130596
    },
    {
      "epoch": 1.7,
      "grad_norm": 0.9939117431640625,
      "learning_rate": 0.00023748525855804914,
      "loss": 2.0136,
      "step": 130597
    },
    {
      "epoch": 1.7,
      "grad_norm": 0.9365002512931824,
      "learning_rate": 0.00023748125780010337,
      "loss": 1.9481,
      "step": 130598
    },
    {
      "epoch": 1.7,
      "grad_norm": 0.9745163321495056,
      "learning_rate": 0.000237477257053781,
      "loss": 2.0804,
      "step": 130599
    },
    {
      "epoch": 1.7,
      "grad_norm": 0.9146335124969482,
      "learning_rate": 0.00023747325631908283,
      "loss": 2.3623,
      "step": 130600
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.2090692520141602,
      "learning_rate": 0.00023746925559600958,
      "loss": 2.0758,
      "step": 130601
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.004799723625183,
      "learning_rate": 0.000237465254884562,
      "loss": 2.0144,
      "step": 130602
    },
    {
      "epoch": 1.7,
      "grad_norm": 0.9643741846084595,
      "learning_rate": 0.00023746125418474073,
      "loss": 1.9674,
      "step": 130603
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.0709240436553955,
      "learning_rate": 0.00023745725349654662,
      "loss": 2.2986,
      "step": 130604
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.0060138702392578,
      "learning_rate": 0.00023745325281998036,
      "loss": 2.2667,
      "step": 130605
    },
    {
      "epoch": 1.7,
      "grad_norm": 0.8706597685813904,
      "learning_rate": 0.0002374492521550427,
      "loss": 1.972,
      "step": 130606
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.1106207370758057,
      "learning_rate": 0.0002374452515017345,
      "loss": 2.1658,
      "step": 130607
    },
    {
      "epoch": 1.7,
      "grad_norm": 0.9637925624847412,
      "learning_rate": 0.00023744125086005635,
      "loss": 2.0408,
      "step": 130608
    },
    {
      "epoch": 1.7,
      "grad_norm": 0.9498124718666077,
      "learning_rate": 0.00023743725023000911,
      "loss": 2.078,
      "step": 130609
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.073272466659546,
      "learning_rate": 0.00023743324961159341,
      "loss": 2.3548,
      "step": 130610
    },
    {
      "epoch": 1.7,
      "grad_norm": 0.9367886185646057,
      "learning_rate": 0.00023742924900481004,
      "loss": 2.1309,
      "step": 130611
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.0773347616195679,
      "learning_rate": 0.00023742524840965974,
      "loss": 2.1103,
      "step": 130612
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.001628041267395,
      "learning_rate": 0.00023742124782614334,
      "loss": 2.4178,
      "step": 130613
    },
    {
      "epoch": 1.7,
      "grad_norm": 0.9695654511451721,
      "learning_rate": 0.00023741724725426147,
      "loss": 1.9677,
      "step": 130614
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.0016894340515137,
      "learning_rate": 0.00023741324669401492,
      "loss": 2.0509,
      "step": 130615
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.056085467338562,
      "learning_rate": 0.00023740924614540446,
      "loss": 1.9636,
      "step": 130616
    },
    {
      "epoch": 1.7,
      "grad_norm": 0.9035264849662781,
      "learning_rate": 0.00023740524560843075,
      "loss": 2.3599,
      "step": 130617
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.0116698741912842,
      "learning_rate": 0.00023740124508309457,
      "loss": 2.0254,
      "step": 130618
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.026233196258545,
      "learning_rate": 0.0002373972445693968,
      "loss": 1.9493,
      "step": 130619
    },
    {
      "epoch": 1.7,
      "grad_norm": 0.9321078062057495,
      "learning_rate": 0.00023739324406733795,
      "loss": 2.0387,
      "step": 130620
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.0033512115478516,
      "learning_rate": 0.00023738924357691888,
      "loss": 2.2548,
      "step": 130621
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.2474595308303833,
      "learning_rate": 0.00023738524309814034,
      "loss": 2.075,
      "step": 130622
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.0312622785568237,
      "learning_rate": 0.00023738124263100312,
      "loss": 2.1058,
      "step": 130623
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.0688872337341309,
      "learning_rate": 0.00023737724217550785,
      "loss": 2.1929,
      "step": 130624
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.0626933574676514,
      "learning_rate": 0.0002373732417316554,
      "loss": 2.1672,
      "step": 130625
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.1118714809417725,
      "learning_rate": 0.0002373692412994464,
      "loss": 1.9987,
      "step": 130626
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.0068180561065674,
      "learning_rate": 0.00023736524087888163,
      "loss": 2.2585,
      "step": 130627
    },
    {
      "epoch": 1.7,
      "grad_norm": 0.9707061052322388,
      "learning_rate": 0.00023736124046996182,
      "loss": 2.1842,
      "step": 130628
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.0231268405914307,
      "learning_rate": 0.00023735724007268775,
      "loss": 2.1223,
      "step": 130629
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.0618987083435059,
      "learning_rate": 0.00023735323968706017,
      "loss": 2.2118,
      "step": 130630
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.0222811698913574,
      "learning_rate": 0.00023734923931307989,
      "loss": 1.9636,
      "step": 130631
    },
    {
      "epoch": 1.7,
      "grad_norm": 0.986641526222229,
      "learning_rate": 0.0002373452389507475,
      "loss": 2.2179,
      "step": 130632
    },
    {
      "epoch": 1.7,
      "grad_norm": 0.9732574820518494,
      "learning_rate": 0.0002373412386000638,
      "loss": 2.3277,
      "step": 130633
    },
    {
      "epoch": 1.7,
      "grad_norm": 0.993263840675354,
      "learning_rate": 0.0002373372382610295,
      "loss": 2.2702,
      "step": 130634
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.1295908689498901,
      "learning_rate": 0.00023733323793364541,
      "loss": 2.1128,
      "step": 130635
    },
    {
      "epoch": 1.7,
      "grad_norm": 0.8875471353530884,
      "learning_rate": 0.0002373292376179123,
      "loss": 1.9401,
      "step": 130636
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.168104648590088,
      "learning_rate": 0.00023732523731383095,
      "loss": 2.0139,
      "step": 130637
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.0451436042785645,
      "learning_rate": 0.0002373212370214019,
      "loss": 2.2636,
      "step": 130638
    },
    {
      "epoch": 1.7,
      "grad_norm": 0.9704926609992981,
      "learning_rate": 0.00023731723674062604,
      "loss": 2.2294,
      "step": 130639
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.0766793489456177,
      "learning_rate": 0.00023731323647150406,
      "loss": 2.2367,
      "step": 130640
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.3188836574554443,
      "learning_rate": 0.00023730923621403675,
      "loss": 2.0719,
      "step": 130641
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.047407627105713,
      "learning_rate": 0.00023730523596822482,
      "loss": 2.2953,
      "step": 130642
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.137894868850708,
      "learning_rate": 0.00023730123573406916,
      "loss": 2.1707,
      "step": 130643
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.0675517320632935,
      "learning_rate": 0.00023729723551157026,
      "loss": 2.0605,
      "step": 130644
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.1427634954452515,
      "learning_rate": 0.00023729323530072898,
      "loss": 2.0204,
      "step": 130645
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.0234320163726807,
      "learning_rate": 0.00023728923510154607,
      "loss": 2.2248,
      "step": 130646
    },
    {
      "epoch": 1.7,
      "grad_norm": 0.9709259867668152,
      "learning_rate": 0.0002372852349140223,
      "loss": 1.955,
      "step": 130647
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.039795994758606,
      "learning_rate": 0.00023728123473815838,
      "loss": 2.1874,
      "step": 130648
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.0291763544082642,
      "learning_rate": 0.00023727723457395505,
      "loss": 2.2597,
      "step": 130649
    },
    {
      "epoch": 1.7,
      "grad_norm": 0.921368420124054,
      "learning_rate": 0.0002372732344214132,
      "loss": 1.9908,
      "step": 130650
    },
    {
      "epoch": 1.7,
      "grad_norm": 0.9824718832969666,
      "learning_rate": 0.00023726923428053334,
      "loss": 2.3145,
      "step": 130651
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.9249136447906494,
      "learning_rate": 0.00023726523415131627,
      "loss": 1.9639,
      "step": 130652
    },
    {
      "epoch": 1.7,
      "grad_norm": 0.9821953177452087,
      "learning_rate": 0.0002372612340337628,
      "loss": 2.2971,
      "step": 130653
    },
    {
      "epoch": 1.7,
      "grad_norm": 0.9329251050949097,
      "learning_rate": 0.00023725723392787362,
      "loss": 2.1115,
      "step": 130654
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.07908034324646,
      "learning_rate": 0.00023725323383364953,
      "loss": 2.12,
      "step": 130655
    },
    {
      "epoch": 1.7,
      "grad_norm": 0.938553512096405,
      "learning_rate": 0.00023724923375109138,
      "loss": 2.072,
      "step": 130656
    },
    {
      "epoch": 1.7,
      "grad_norm": 0.991736114025116,
      "learning_rate": 0.00023724523368019964,
      "loss": 2.0751,
      "step": 130657
    },
    {
      "epoch": 1.7,
      "grad_norm": 0.9208664894104004,
      "learning_rate": 0.0002372412336209752,
      "loss": 1.9614,
      "step": 130658
    },
    {
      "epoch": 1.7,
      "grad_norm": 0.8896547555923462,
      "learning_rate": 0.0002372372335734188,
      "loss": 1.9555,
      "step": 130659
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.0359808206558228,
      "learning_rate": 0.0002372332335375312,
      "loss": 2.1587,
      "step": 130660
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.0734217166900635,
      "learning_rate": 0.0002372292335133131,
      "loss": 2.0833,
      "step": 130661
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.0586481094360352,
      "learning_rate": 0.00023722523350076541,
      "loss": 2.2126,
      "step": 130662
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.0091257095336914,
      "learning_rate": 0.00023722123349988864,
      "loss": 1.9772,
      "step": 130663
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.1056452989578247,
      "learning_rate": 0.00023721723351068357,
      "loss": 2.0218,
      "step": 130664
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.0468111038208008,
      "learning_rate": 0.00023721323353315105,
      "loss": 2.2296,
      "step": 130665
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.0317087173461914,
      "learning_rate": 0.00023720923356729173,
      "loss": 2.0559,
      "step": 130666
    },
    {
      "epoch": 1.7,
      "grad_norm": 0.9428502917289734,
      "learning_rate": 0.00023720523361310642,
      "loss": 2.1633,
      "step": 130667
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.0491935014724731,
      "learning_rate": 0.00023720123367059598,
      "loss": 2.039,
      "step": 130668
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.1263295412063599,
      "learning_rate": 0.0002371972337397609,
      "loss": 1.8319,
      "step": 130669
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.0818238258361816,
      "learning_rate": 0.000237193233820602,
      "loss": 2.1631,
      "step": 130670
    },
    {
      "epoch": 1.7,
      "grad_norm": 0.9478395581245422,
      "learning_rate": 0.00023718923391312006,
      "loss": 2.0612,
      "step": 130671
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.0243436098098755,
      "learning_rate": 0.00023718523401731586,
      "loss": 2.1082,
      "step": 130672
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.1022967100143433,
      "learning_rate": 0.00023718123413319012,
      "loss": 2.2584,
      "step": 130673
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.0665645599365234,
      "learning_rate": 0.00023717723426074368,
      "loss": 2.2683,
      "step": 130674
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.0000590085983276,
      "learning_rate": 0.00023717323439997705,
      "loss": 2.1027,
      "step": 130675
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.1107743978500366,
      "learning_rate": 0.0002371692345508911,
      "loss": 2.0693,
      "step": 130676
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.0374722480773926,
      "learning_rate": 0.00023716523471348659,
      "loss": 2.0611,
      "step": 130677
    },
    {
      "epoch": 1.7,
      "grad_norm": 0.9329525828361511,
      "learning_rate": 0.00023716123488776422,
      "loss": 2.0205,
      "step": 130678
    },
    {
      "epoch": 1.7,
      "grad_norm": 0.9487002491950989,
      "learning_rate": 0.00023715723507372477,
      "loss": 2.1418,
      "step": 130679
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.0553288459777832,
      "learning_rate": 0.0002371532352713691,
      "loss": 1.9417,
      "step": 130680
    },
    {
      "epoch": 1.7,
      "grad_norm": 0.9349834322929382,
      "learning_rate": 0.00023714923548069774,
      "loss": 2.2377,
      "step": 130681
    },
    {
      "epoch": 1.7,
      "grad_norm": 0.9680298566818237,
      "learning_rate": 0.00023714523570171148,
      "loss": 2.0341,
      "step": 130682
    },
    {
      "epoch": 1.7,
      "grad_norm": 0.9604093432426453,
      "learning_rate": 0.00023714123593441115,
      "loss": 2.0937,
      "step": 130683
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.0849529504776,
      "learning_rate": 0.00023713723617879744,
      "loss": 1.8956,
      "step": 130684
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.1561272144317627,
      "learning_rate": 0.00023713323643487108,
      "loss": 1.8289,
      "step": 130685
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.0683178901672363,
      "learning_rate": 0.00023712923670263298,
      "loss": 2.1774,
      "step": 130686
    },
    {
      "epoch": 1.7,
      "grad_norm": 0.9900320172309875,
      "learning_rate": 0.00023712523698208365,
      "loss": 2.0643,
      "step": 130687
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.0195949077606201,
      "learning_rate": 0.00023712123727322385,
      "loss": 2.0589,
      "step": 130688
    },
    {
      "epoch": 1.7,
      "grad_norm": 0.9370065927505493,
      "learning_rate": 0.00023711723757605446,
      "loss": 1.9826,
      "step": 130689
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.37051260471344,
      "learning_rate": 0.0002371132378905761,
      "loss": 2.1165,
      "step": 130690
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.0476210117340088,
      "learning_rate": 0.00023710923821678963,
      "loss": 2.2262,
      "step": 130691
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.0381085872650146,
      "learning_rate": 0.00023710523855469581,
      "loss": 2.2206,
      "step": 130692
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.0292625427246094,
      "learning_rate": 0.0002371012389042952,
      "loss": 2.1953,
      "step": 130693
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.0615960359573364,
      "learning_rate": 0.00023709723926558876,
      "loss": 1.9593,
      "step": 130694
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.088714838027954,
      "learning_rate": 0.00023709323963857704,
      "loss": 2.1579,
      "step": 130695
    },
    {
      "epoch": 1.7,
      "grad_norm": 0.9877734184265137,
      "learning_rate": 0.0002370892400232609,
      "loss": 1.9972,
      "step": 130696
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.156612753868103,
      "learning_rate": 0.0002370852404196411,
      "loss": 2.1464,
      "step": 130697
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.0656501054763794,
      "learning_rate": 0.00023708124082771832,
      "loss": 2.2771,
      "step": 130698
    },
    {
      "epoch": 1.7,
      "grad_norm": 0.947315514087677,
      "learning_rate": 0.0002370772412474933,
      "loss": 1.9345,
      "step": 130699
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.0174227952957153,
      "learning_rate": 0.00023707324167896677,
      "loss": 2.0914,
      "step": 130700
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.2147387266159058,
      "learning_rate": 0.00023706924212213966,
      "loss": 2.1629,
      "step": 130701
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.0385576486587524,
      "learning_rate": 0.0002370652425770124,
      "loss": 2.1255,
      "step": 130702
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.0867137908935547,
      "learning_rate": 0.000237061243043586,
      "loss": 2.1727,
      "step": 130703
    },
    {
      "epoch": 1.7,
      "grad_norm": 0.9476841688156128,
      "learning_rate": 0.00023705724352186111,
      "loss": 2.4059,
      "step": 130704
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.2544054985046387,
      "learning_rate": 0.0002370532440118384,
      "loss": 2.0221,
      "step": 130705
    },
    {
      "epoch": 1.7,
      "grad_norm": 0.9981504082679749,
      "learning_rate": 0.00023704924451351867,
      "loss": 2.1273,
      "step": 130706
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.109237551689148,
      "learning_rate": 0.00023704524502690273,
      "loss": 2.3493,
      "step": 130707
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.0171741247177124,
      "learning_rate": 0.00023704124555199127,
      "loss": 2.297,
      "step": 130708
    },
    {
      "epoch": 1.7,
      "grad_norm": 0.9764283895492554,
      "learning_rate": 0.00023703724608878498,
      "loss": 1.9351,
      "step": 130709
    },
    {
      "epoch": 1.7,
      "grad_norm": 0.9682469367980957,
      "learning_rate": 0.00023703324663728472,
      "loss": 1.9914,
      "step": 130710
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.0101743936538696,
      "learning_rate": 0.0002370292471974911,
      "loss": 1.9649,
      "step": 130711
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.2236876487731934,
      "learning_rate": 0.00023702524776940498,
      "loss": 1.8926,
      "step": 130712
    },
    {
      "epoch": 1.7,
      "grad_norm": 0.9852972030639648,
      "learning_rate": 0.000237021248353027,
      "loss": 2.2278,
      "step": 130713
    },
    {
      "epoch": 1.7,
      "grad_norm": 0.9920967221260071,
      "learning_rate": 0.000237017248948358,
      "loss": 1.862,
      "step": 130714
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.0675169229507446,
      "learning_rate": 0.00023701324955539873,
      "loss": 2.1307,
      "step": 130715
    },
    {
      "epoch": 1.7,
      "grad_norm": 0.9684197902679443,
      "learning_rate": 0.0002370092501741498,
      "loss": 1.997,
      "step": 130716
    },
    {
      "epoch": 1.7,
      "grad_norm": 0.989931046962738,
      "learning_rate": 0.00023700525080461214,
      "loss": 2.1519,
      "step": 130717
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.0490171909332275,
      "learning_rate": 0.00023700125144678633,
      "loss": 2.1778,
      "step": 130718
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.0587047338485718,
      "learning_rate": 0.00023699725210067315,
      "loss": 2.1349,
      "step": 130719
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.137777328491211,
      "learning_rate": 0.00023699325276627337,
      "loss": 2.0626,
      "step": 130720
    },
    {
      "epoch": 1.7,
      "grad_norm": 0.9802491068840027,
      "learning_rate": 0.00023698925344358774,
      "loss": 1.998,
      "step": 130721
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.011308193206787,
      "learning_rate": 0.00023698525413261702,
      "loss": 2.2785,
      "step": 130722
    },
    {
      "epoch": 1.7,
      "grad_norm": 0.9616671800613403,
      "learning_rate": 0.000236981254833362,
      "loss": 2.2199,
      "step": 130723
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.0109028816223145,
      "learning_rate": 0.00023697725554582328,
      "loss": 1.9475,
      "step": 130724
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.0902234315872192,
      "learning_rate": 0.00023697325627000166,
      "loss": 2.2381,
      "step": 130725
    },
    {
      "epoch": 1.7,
      "grad_norm": 0.9095630645751953,
      "learning_rate": 0.0002369692570058979,
      "loss": 2.0519,
      "step": 130726
    },
    {
      "epoch": 1.7,
      "grad_norm": 0.9797921180725098,
      "learning_rate": 0.00023696525775351277,
      "loss": 2.1233,
      "step": 130727
    },
    {
      "epoch": 1.7,
      "grad_norm": 0.978195071220398,
      "learning_rate": 0.00023696125851284692,
      "loss": 2.1179,
      "step": 130728
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.0041844844818115,
      "learning_rate": 0.00023695725928390136,
      "loss": 2.1748,
      "step": 130729
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.0298113822937012,
      "learning_rate": 0.00023695326006667648,
      "loss": 2.0591,
      "step": 130730
    },
    {
      "epoch": 1.7,
      "grad_norm": 0.9874306917190552,
      "learning_rate": 0.00023694926086117318,
      "loss": 2.1177,
      "step": 130731
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.011959433555603,
      "learning_rate": 0.00023694526166739222,
      "loss": 2.0659,
      "step": 130732
    },
    {
      "epoch": 1.7,
      "grad_norm": 0.942279577255249,
      "learning_rate": 0.0002369412624853343,
      "loss": 1.9299,
      "step": 130733
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.0576049089431763,
      "learning_rate": 0.00023693726331500023,
      "loss": 2.1673,
      "step": 130734
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.0414656400680542,
      "learning_rate": 0.0002369332641563908,
      "loss": 2.2642,
      "step": 130735
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.021636724472046,
      "learning_rate": 0.00023692926500950653,
      "loss": 1.9383,
      "step": 130736
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.0573267936706543,
      "learning_rate": 0.00023692526587434833,
      "loss": 2.1527,
      "step": 130737
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.0034981966018677,
      "learning_rate": 0.0002369212667509169,
      "loss": 1.9982,
      "step": 130738
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.0299861431121826,
      "learning_rate": 0.000236917267639213,
      "loss": 1.9516,
      "step": 130739
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.2754716873168945,
      "learning_rate": 0.0002369132685392374,
      "loss": 2.0994,
      "step": 130740
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.0882121324539185,
      "learning_rate": 0.00023690926945099092,
      "loss": 2.1425,
      "step": 130741
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.0186022520065308,
      "learning_rate": 0.00023690527037447404,
      "loss": 2.1221,
      "step": 130742
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.1527749300003052,
      "learning_rate": 0.0002369012713096877,
      "loss": 2.117,
      "step": 130743
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.0962074995040894,
      "learning_rate": 0.0002368972722566326,
      "loss": 2.103,
      "step": 130744
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.2002874612808228,
      "learning_rate": 0.00023689327321530946,
      "loss": 2.1431,
      "step": 130745
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.0026938915252686,
      "learning_rate": 0.00023688927418571905,
      "loss": 2.0472,
      "step": 130746
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.246179461479187,
      "learning_rate": 0.00023688527516786226,
      "loss": 2.2924,
      "step": 130747
    },
    {
      "epoch": 1.7,
      "grad_norm": 0.9726556539535522,
      "learning_rate": 0.0002368812761617396,
      "loss": 2.0015,
      "step": 130748
    },
    {
      "epoch": 1.7,
      "grad_norm": 0.9263858795166016,
      "learning_rate": 0.00023687727716735184,
      "loss": 2.1208,
      "step": 130749
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.060555100440979,
      "learning_rate": 0.00023687327818469982,
      "loss": 2.3155,
      "step": 130750
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.1289899349212646,
      "learning_rate": 0.0002368692792137842,
      "loss": 2.1862,
      "step": 130751
    },
    {
      "epoch": 1.7,
      "grad_norm": 0.9917191863059998,
      "learning_rate": 0.00023686528025460585,
      "loss": 2.0077,
      "step": 130752
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.196654200553894,
      "learning_rate": 0.00023686128130716554,
      "loss": 2.1542,
      "step": 130753
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.0777156352996826,
      "learning_rate": 0.00023685728237146376,
      "loss": 2.1891,
      "step": 130754
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.0213677883148193,
      "learning_rate": 0.00023685328344750142,
      "loss": 2.1679,
      "step": 130755
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.0443094968795776,
      "learning_rate": 0.00023684928453527923,
      "loss": 2.077,
      "step": 130756
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.0102041959762573,
      "learning_rate": 0.000236845285634798,
      "loss": 2.3093,
      "step": 130757
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.0022307634353638,
      "learning_rate": 0.00023684128674605835,
      "loss": 2.0309,
      "step": 130758
    },
    {
      "epoch": 1.7,
      "grad_norm": 0.9665279388427734,
      "learning_rate": 0.00023683728786906127,
      "loss": 1.9988,
      "step": 130759
    },
    {
      "epoch": 1.7,
      "grad_norm": 0.9412801861763,
      "learning_rate": 0.00023683328900380717,
      "loss": 1.9925,
      "step": 130760
    },
    {
      "epoch": 1.7,
      "grad_norm": 0.9866412281990051,
      "learning_rate": 0.000236829290150297,
      "loss": 2.1759,
      "step": 130761
    },
    {
      "epoch": 1.7,
      "grad_norm": 0.9553504586219788,
      "learning_rate": 0.0002368252913085314,
      "loss": 2.1666,
      "step": 130762
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.0405353307724,
      "learning_rate": 0.00023682129247851123,
      "loss": 2.1078,
      "step": 130763
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.0051155090332031,
      "learning_rate": 0.00023681729366023714,
      "loss": 2.0334,
      "step": 130764
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.0027133226394653,
      "learning_rate": 0.00023681329485371005,
      "loss": 1.8336,
      "step": 130765
    },
    {
      "epoch": 1.7,
      "grad_norm": 0.9514885544776917,
      "learning_rate": 0.00023680929605893045,
      "loss": 2.1914,
      "step": 130766
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.0529981851577759,
      "learning_rate": 0.00023680529727589914,
      "loss": 2.0658,
      "step": 130767
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.1684988737106323,
      "learning_rate": 0.00023680129850461694,
      "loss": 2.0537,
      "step": 130768
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.1415157318115234,
      "learning_rate": 0.00023679729974508462,
      "loss": 1.9,
      "step": 130769
    },
    {
      "epoch": 1.7,
      "grad_norm": 0.9952042102813721,
      "learning_rate": 0.00023679330099730283,
      "loss": 2.1489,
      "step": 130770
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.1141941547393799,
      "learning_rate": 0.0002367893022612725,
      "loss": 2.0419,
      "step": 130771
    },
    {
      "epoch": 1.7,
      "grad_norm": 0.9668042063713074,
      "learning_rate": 0.00023678530353699406,
      "loss": 1.9226,
      "step": 130772
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.078738808631897,
      "learning_rate": 0.0002367813048244685,
      "loss": 2.2242,
      "step": 130773
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.0371514558792114,
      "learning_rate": 0.00023677730612369643,
      "loss": 2.0868,
      "step": 130774
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.0761322975158691,
      "learning_rate": 0.00023677330743467867,
      "loss": 2.2243,
      "step": 130775
    },
    {
      "epoch": 1.7,
      "grad_norm": 0.9591138362884521,
      "learning_rate": 0.00023676930875741594,
      "loss": 2.2464,
      "step": 130776
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.073777675628662,
      "learning_rate": 0.00023676531009190907,
      "loss": 1.9883,
      "step": 130777
    },
    {
      "epoch": 1.7,
      "grad_norm": 0.9401900768280029,
      "learning_rate": 0.00023676131143815873,
      "loss": 2.2268,
      "step": 130778
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.040305256843567,
      "learning_rate": 0.00023675731279616555,
      "loss": 2.2878,
      "step": 130779
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.0941623449325562,
      "learning_rate": 0.00023675331416593041,
      "loss": 2.16,
      "step": 130780
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.0341768264770508,
      "learning_rate": 0.00023674931554745402,
      "loss": 2.51,
      "step": 130781
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.0004980564117432,
      "learning_rate": 0.0002367453169407371,
      "loss": 1.9596,
      "step": 130782
    },
    {
      "epoch": 1.7,
      "grad_norm": 0.9966179728507996,
      "learning_rate": 0.00023674131834578046,
      "loss": 2.1907,
      "step": 130783
    },
    {
      "epoch": 1.7,
      "grad_norm": 0.9878580570220947,
      "learning_rate": 0.00023673731976258484,
      "loss": 2.1255,
      "step": 130784
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.0629546642303467,
      "learning_rate": 0.00023673332119115097,
      "loss": 2.2092,
      "step": 130785
    },
    {
      "epoch": 1.7,
      "grad_norm": 0.8838322758674622,
      "learning_rate": 0.00023672932263147948,
      "loss": 2.1483,
      "step": 130786
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.0264917612075806,
      "learning_rate": 0.00023672532408357123,
      "loss": 2.3572,
      "step": 130787
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.09956693649292,
      "learning_rate": 0.00023672132554742692,
      "loss": 2.1768,
      "step": 130788
    },
    {
      "epoch": 1.7,
      "grad_norm": 0.9531331658363342,
      "learning_rate": 0.00023671732702304732,
      "loss": 2.0409,
      "step": 130789
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.0040580034255981,
      "learning_rate": 0.00023671332851043323,
      "loss": 2.3498,
      "step": 130790
    },
    {
      "epoch": 1.7,
      "grad_norm": 0.9362996220588684,
      "learning_rate": 0.00023670933000958524,
      "loss": 2.0967,
      "step": 130791
    },
    {
      "epoch": 1.7,
      "grad_norm": 0.9454217553138733,
      "learning_rate": 0.00023670533152050417,
      "loss": 1.962,
      "step": 130792
    },
    {
      "epoch": 1.7,
      "grad_norm": 0.9696817398071289,
      "learning_rate": 0.0002367013330431909,
      "loss": 1.9234,
      "step": 130793
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.2333614826202393,
      "learning_rate": 0.00023669733457764593,
      "loss": 1.9983,
      "step": 130794
    },
    {
      "epoch": 1.7,
      "grad_norm": 0.9703671932220459,
      "learning_rate": 0.00023669333612387012,
      "loss": 1.9317,
      "step": 130795
    },
    {
      "epoch": 1.7,
      "grad_norm": 0.9786482453346252,
      "learning_rate": 0.0002366893376818643,
      "loss": 2.0405,
      "step": 130796
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.1365203857421875,
      "learning_rate": 0.00023668533925162906,
      "loss": 2.0906,
      "step": 130797
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.0166712999343872,
      "learning_rate": 0.0002366813408331652,
      "loss": 2.1594,
      "step": 130798
    },
    {
      "epoch": 1.7,
      "grad_norm": 0.9542795419692993,
      "learning_rate": 0.00023667734242647349,
      "loss": 2.0097,
      "step": 130799
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.1250640153884888,
      "learning_rate": 0.0002366733440315547,
      "loss": 2.1069,
      "step": 130800
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.1213561296463013,
      "learning_rate": 0.00023666934564840948,
      "loss": 2.1887,
      "step": 130801
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.0144944190979004,
      "learning_rate": 0.0002366653472770387,
      "loss": 2.0891,
      "step": 130802
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.1051946878433228,
      "learning_rate": 0.00023666134891744294,
      "loss": 2.3203,
      "step": 130803
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.208530068397522,
      "learning_rate": 0.00023665735056962303,
      "loss": 2.2165,
      "step": 130804
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.0594357252120972,
      "learning_rate": 0.00023665335223357976,
      "loss": 2.176,
      "step": 130805
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.0438672304153442,
      "learning_rate": 0.00023664935390931376,
      "loss": 2.2647,
      "step": 130806
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.0158915519714355,
      "learning_rate": 0.00023664535559682595,
      "loss": 2.1812,
      "step": 130807
    },
    {
      "epoch": 1.7,
      "grad_norm": 0.9918815493583679,
      "learning_rate": 0.00023664135729611695,
      "loss": 2.141,
      "step": 130808
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.0697380304336548,
      "learning_rate": 0.00023663735900718747,
      "loss": 1.8927,
      "step": 130809
    },
    {
      "epoch": 1.7,
      "grad_norm": 0.9603420495986938,
      "learning_rate": 0.0002366333607300383,
      "loss": 2.1395,
      "step": 130810
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.0997687578201294,
      "learning_rate": 0.00023662936246467012,
      "loss": 1.9546,
      "step": 130811
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.0123271942138672,
      "learning_rate": 0.0002366253642110838,
      "loss": 2.2282,
      "step": 130812
    },
    {
      "epoch": 1.7,
      "grad_norm": 0.9447373747825623,
      "learning_rate": 0.00023662136596928,
      "loss": 1.7678,
      "step": 130813
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.015659213066101,
      "learning_rate": 0.00023661736773925964,
      "loss": 2.1901,
      "step": 130814
    },
    {
      "epoch": 1.7,
      "grad_norm": 0.9759340882301331,
      "learning_rate": 0.00023661336952102317,
      "loss": 1.9146,
      "step": 130815
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.0027556419372559,
      "learning_rate": 0.00023660937131457144,
      "loss": 2.0327,
      "step": 130816
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.2998524904251099,
      "learning_rate": 0.00023660537311990528,
      "loss": 2.107,
      "step": 130817
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.1038495302200317,
      "learning_rate": 0.00023660137493702532,
      "loss": 2.0904,
      "step": 130818
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.1816489696502686,
      "learning_rate": 0.0002365973767659324,
      "loss": 2.2311,
      "step": 130819
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.1889629364013672,
      "learning_rate": 0.00023659337860662733,
      "loss": 2.0543,
      "step": 130820
    },
    {
      "epoch": 1.7,
      "grad_norm": 0.980657696723938,
      "learning_rate": 0.00023658938045911063,
      "loss": 2.049,
      "step": 130821
    },
    {
      "epoch": 1.7,
      "grad_norm": 0.9464771747589111,
      "learning_rate": 0.00023658538232338314,
      "loss": 2.2428,
      "step": 130822
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.0241390466690063,
      "learning_rate": 0.00023658138419944567,
      "loss": 2.0603,
      "step": 130823
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.1861149072647095,
      "learning_rate": 0.00023657738608729887,
      "loss": 2.1203,
      "step": 130824
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.251957893371582,
      "learning_rate": 0.00023657338798694356,
      "loss": 1.9633,
      "step": 130825
    },
    {
      "epoch": 1.7,
      "grad_norm": 0.9644711017608643,
      "learning_rate": 0.00023656938989838057,
      "loss": 2.2573,
      "step": 130826
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.1447597742080688,
      "learning_rate": 0.00023656539182161042,
      "loss": 1.8762,
      "step": 130827
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.0654032230377197,
      "learning_rate": 0.00023656139375663395,
      "loss": 2.0668,
      "step": 130828
    },
    {
      "epoch": 1.7,
      "grad_norm": 0.9896077513694763,
      "learning_rate": 0.0002365573957034519,
      "loss": 2.0303,
      "step": 130829
    },
    {
      "epoch": 1.7,
      "grad_norm": 0.9909435510635376,
      "learning_rate": 0.00023655339766206506,
      "loss": 2.4156,
      "step": 130830
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.1023467779159546,
      "learning_rate": 0.0002365493996324741,
      "loss": 2.1838,
      "step": 130831
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.0125300884246826,
      "learning_rate": 0.00023654540161467994,
      "loss": 2.1677,
      "step": 130832
    },
    {
      "epoch": 1.7,
      "grad_norm": 0.9653568863868713,
      "learning_rate": 0.0002365414036086831,
      "loss": 2.0896,
      "step": 130833
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.0735249519348145,
      "learning_rate": 0.0002365374056144844,
      "loss": 2.1407,
      "step": 130834
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.0146007537841797,
      "learning_rate": 0.00023653340763208456,
      "loss": 2.0467,
      "step": 130835
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.3595924377441406,
      "learning_rate": 0.00023652940966148437,
      "loss": 2.3911,
      "step": 130836
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.0495576858520508,
      "learning_rate": 0.00023652541170268455,
      "loss": 2.2472,
      "step": 130837
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.0021778345108032,
      "learning_rate": 0.00023652141375568602,
      "loss": 2.2055,
      "step": 130838
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.0323694944381714,
      "learning_rate": 0.0002365174158204892,
      "loss": 2.2683,
      "step": 130839
    },
    {
      "epoch": 1.7,
      "grad_norm": 0.914511501789093,
      "learning_rate": 0.00023651341789709498,
      "loss": 2.2788,
      "step": 130840
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.1027734279632568,
      "learning_rate": 0.00023650941998550415,
      "loss": 2.1275,
      "step": 130841
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.0390064716339111,
      "learning_rate": 0.0002365054220857174,
      "loss": 2.0831,
      "step": 130842
    },
    {
      "epoch": 1.7,
      "grad_norm": 0.9535170793533325,
      "learning_rate": 0.0002365014241977355,
      "loss": 2.0176,
      "step": 130843
    },
    {
      "epoch": 1.7,
      "grad_norm": 0.9597494006156921,
      "learning_rate": 0.00023649742632155928,
      "loss": 2.1959,
      "step": 130844
    },
    {
      "epoch": 1.7,
      "grad_norm": 0.9938012957572937,
      "learning_rate": 0.0002364934284571893,
      "loss": 2.1087,
      "step": 130845
    },
    {
      "epoch": 1.7,
      "grad_norm": 0.8970595002174377,
      "learning_rate": 0.00023648943060462636,
      "loss": 2.2041,
      "step": 130846
    },
    {
      "epoch": 1.7,
      "grad_norm": 0.9225674867630005,
      "learning_rate": 0.00023648543276387122,
      "loss": 2.0846,
      "step": 130847
    },
    {
      "epoch": 1.7,
      "grad_norm": 0.9267868995666504,
      "learning_rate": 0.00023648143493492467,
      "loss": 2.299,
      "step": 130848
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.0463330745697021,
      "learning_rate": 0.00023647743711778743,
      "loss": 1.9528,
      "step": 130849
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.0237507820129395,
      "learning_rate": 0.0002364734393124602,
      "loss": 2.0454,
      "step": 130850
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.07706880569458,
      "learning_rate": 0.0002364694415189439,
      "loss": 1.9825,
      "step": 130851
    },
    {
      "epoch": 1.7,
      "grad_norm": 0.9695380330085754,
      "learning_rate": 0.000236465443737239,
      "loss": 1.9539,
      "step": 130852
    },
    {
      "epoch": 1.7,
      "grad_norm": 0.9647034406661987,
      "learning_rate": 0.00023646144596734637,
      "loss": 1.9874,
      "step": 130853
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.1030528545379639,
      "learning_rate": 0.0002364574482092668,
      "loss": 2.1183,
      "step": 130854
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.0273432731628418,
      "learning_rate": 0.00023645345046300093,
      "loss": 2.1132,
      "step": 130855
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.0375380516052246,
      "learning_rate": 0.0002364494527285496,
      "loss": 2.0911,
      "step": 130856
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.0455611944198608,
      "learning_rate": 0.00023644545500591362,
      "loss": 2.2384,
      "step": 130857
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.0714887380599976,
      "learning_rate": 0.00023644145729509352,
      "loss": 2.1566,
      "step": 130858
    },
    {
      "epoch": 1.7,
      "grad_norm": 0.9791592359542847,
      "learning_rate": 0.00023643745959609012,
      "loss": 2.0102,
      "step": 130859
    },
    {
      "epoch": 1.7,
      "grad_norm": 0.9780356884002686,
      "learning_rate": 0.00023643346190890422,
      "loss": 2.3622,
      "step": 130860
    },
    {
      "epoch": 1.7,
      "grad_norm": 0.9070913791656494,
      "learning_rate": 0.00023642946423353654,
      "loss": 2.2503,
      "step": 130861
    },
    {
      "epoch": 1.7,
      "grad_norm": 0.9953276515007019,
      "learning_rate": 0.00023642546656998782,
      "loss": 1.9057,
      "step": 130862
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.0231654644012451,
      "learning_rate": 0.00023642146891825894,
      "loss": 1.9823,
      "step": 130863
    },
    {
      "epoch": 1.7,
      "grad_norm": 0.9948089718818665,
      "learning_rate": 0.00023641747127835038,
      "loss": 2.1986,
      "step": 130864
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.0337560176849365,
      "learning_rate": 0.00023641347365026301,
      "loss": 2.0373,
      "step": 130865
    },
    {
      "epoch": 1.7,
      "grad_norm": 0.9720239639282227,
      "learning_rate": 0.00023640947603399757,
      "loss": 2.0788,
      "step": 130866
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.0158494710922241,
      "learning_rate": 0.00023640547842955484,
      "loss": 2.1565,
      "step": 130867
    },
    {
      "epoch": 1.7,
      "grad_norm": 0.9917440414428711,
      "learning_rate": 0.00023640148083693547,
      "loss": 2.1499,
      "step": 130868
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.0190887451171875,
      "learning_rate": 0.00023639748325614038,
      "loss": 1.9717,
      "step": 130869
    },
    {
      "epoch": 1.7,
      "grad_norm": 0.9544817805290222,
      "learning_rate": 0.0002363934856871702,
      "loss": 1.8491,
      "step": 130870
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.0558396577835083,
      "learning_rate": 0.0002363894881300256,
      "loss": 2.0645,
      "step": 130871
    },
    {
      "epoch": 1.7,
      "grad_norm": 0.9700393676757812,
      "learning_rate": 0.00023638549058470736,
      "loss": 1.9149,
      "step": 130872
    },
    {
      "epoch": 1.7,
      "grad_norm": 0.924483060836792,
      "learning_rate": 0.00023638149305121628,
      "loss": 2.2566,
      "step": 130873
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.010414958000183,
      "learning_rate": 0.0002363774955295531,
      "loss": 1.9715,
      "step": 130874
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.0614029169082642,
      "learning_rate": 0.00023637349801971857,
      "loss": 1.779,
      "step": 130875
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.0391062498092651,
      "learning_rate": 0.0002363695005217134,
      "loss": 1.9975,
      "step": 130876
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.233836054801941,
      "learning_rate": 0.00023636550303553835,
      "loss": 1.9269,
      "step": 130877
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.0396223068237305,
      "learning_rate": 0.0002363615055611941,
      "loss": 1.9502,
      "step": 130878
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.1217786073684692,
      "learning_rate": 0.00023635750809868146,
      "loss": 2.2356,
      "step": 130879
    },
    {
      "epoch": 1.7,
      "grad_norm": 0.951747715473175,
      "learning_rate": 0.00023635351064800117,
      "loss": 1.9833,
      "step": 130880
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.3442517518997192,
      "learning_rate": 0.00023634951320915403,
      "loss": 2.3669,
      "step": 130881
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.0331296920776367,
      "learning_rate": 0.00023634551578214063,
      "loss": 2.2791,
      "step": 130882
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.1333425045013428,
      "learning_rate": 0.00023634151836696181,
      "loss": 2.2058,
      "step": 130883
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.1902345418930054,
      "learning_rate": 0.00023633752096361828,
      "loss": 2.047,
      "step": 130884
    },
    {
      "epoch": 1.7,
      "grad_norm": 0.9743680357933044,
      "learning_rate": 0.0002363335235721109,
      "loss": 1.9832,
      "step": 130885
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.4123423099517822,
      "learning_rate": 0.00023632952619244028,
      "loss": 1.9767,
      "step": 130886
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.055506706237793,
      "learning_rate": 0.00023632552882460724,
      "loss": 2.1904,
      "step": 130887
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.2500447034835815,
      "learning_rate": 0.0002363215314686124,
      "loss": 2.1713,
      "step": 130888
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.1366487741470337,
      "learning_rate": 0.0002363175341244566,
      "loss": 2.1952,
      "step": 130889
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.8447867631912231,
      "learning_rate": 0.0002363135367921406,
      "loss": 2.1591,
      "step": 130890
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.057296872138977,
      "learning_rate": 0.0002363095394716651,
      "loss": 2.321,
      "step": 130891
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.3847365379333496,
      "learning_rate": 0.0002363055421630309,
      "loss": 2.1737,
      "step": 130892
    },
    {
      "epoch": 1.7,
      "grad_norm": 0.9885444641113281,
      "learning_rate": 0.00023630154486623873,
      "loss": 2.2296,
      "step": 130893
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.083166241645813,
      "learning_rate": 0.00023629754758128924,
      "loss": 2.0855,
      "step": 130894
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.2001910209655762,
      "learning_rate": 0.00023629355030818324,
      "loss": 2.1751,
      "step": 130895
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.0122771263122559,
      "learning_rate": 0.00023628955304692145,
      "loss": 2.072,
      "step": 130896
    },
    {
      "epoch": 1.7,
      "grad_norm": 0.9905148148536682,
      "learning_rate": 0.00023628555579750465,
      "loss": 2.2865,
      "step": 130897
    },
    {
      "epoch": 1.7,
      "grad_norm": 0.9529603123664856,
      "learning_rate": 0.00023628155855993357,
      "loss": 2.0215,
      "step": 130898
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.0173507928848267,
      "learning_rate": 0.0002362775613342091,
      "loss": 1.9654,
      "step": 130899
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.044312596321106,
      "learning_rate": 0.00023627356412033167,
      "loss": 2.1806,
      "step": 130900
    },
    {
      "epoch": 1.7,
      "grad_norm": 0.9136300086975098,
      "learning_rate": 0.0002362695669183022,
      "loss": 1.9327,
      "step": 130901
    },
    {
      "epoch": 1.7,
      "grad_norm": 0.9747952222824097,
      "learning_rate": 0.0002362655697281214,
      "loss": 2.0845,
      "step": 130902
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.012824296951294,
      "learning_rate": 0.00023626157254979006,
      "loss": 1.958,
      "step": 130903
    },
    {
      "epoch": 1.7,
      "grad_norm": 0.9459362626075745,
      "learning_rate": 0.0002362575753833089,
      "loss": 2.279,
      "step": 130904
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.0265341997146606,
      "learning_rate": 0.00023625357822867877,
      "loss": 2.0304,
      "step": 130905
    },
    {
      "epoch": 1.7,
      "grad_norm": 0.934863805770874,
      "learning_rate": 0.0002362495810859002,
      "loss": 2.0642,
      "step": 130906
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.1663328409194946,
      "learning_rate": 0.00023624558395497403,
      "loss": 1.9441,
      "step": 130907
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.0943964719772339,
      "learning_rate": 0.00023624158683590096,
      "loss": 2.3291,
      "step": 130908
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.0720714330673218,
      "learning_rate": 0.00023623758972868183,
      "loss": 2.1894,
      "step": 130909
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.028366208076477,
      "learning_rate": 0.0002362335926333173,
      "loss": 2.3391,
      "step": 130910
    },
    {
      "epoch": 1.7,
      "grad_norm": 0.9987034201622009,
      "learning_rate": 0.00023622959554980833,
      "loss": 1.9426,
      "step": 130911
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.1521941423416138,
      "learning_rate": 0.00023622559847815533,
      "loss": 1.8733,
      "step": 130912
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.0669595003128052,
      "learning_rate": 0.0002362216014183592,
      "loss": 2.164,
      "step": 130913
    },
    {
      "epoch": 1.7,
      "grad_norm": 0.9752747416496277,
      "learning_rate": 0.00023621760437042067,
      "loss": 2.0217,
      "step": 130914
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.0845030546188354,
      "learning_rate": 0.0002362136073343405,
      "loss": 2.1631,
      "step": 130915
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.1364079713821411,
      "learning_rate": 0.00023620961031011942,
      "loss": 1.9686,
      "step": 130916
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.0131069421768188,
      "learning_rate": 0.00023620561329775817,
      "loss": 1.8456,
      "step": 130917
    },
    {
      "epoch": 1.7,
      "grad_norm": 0.9780187606811523,
      "learning_rate": 0.00023620161629725762,
      "loss": 1.9271,
      "step": 130918
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.0726937055587769,
      "learning_rate": 0.00023619761930861828,
      "loss": 2.2287,
      "step": 130919
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.0609551668167114,
      "learning_rate": 0.00023619362233184103,
      "loss": 2.1785,
      "step": 130920
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.0912373065948486,
      "learning_rate": 0.00023618962536692655,
      "loss": 2.1815,
      "step": 130921
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.0942844152450562,
      "learning_rate": 0.00023618562841387565,
      "loss": 2.0827,
      "step": 130922
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.0794907808303833,
      "learning_rate": 0.00023618163147268902,
      "loss": 2.0527,
      "step": 130923
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.0989984273910522,
      "learning_rate": 0.00023617763454336757,
      "loss": 2.3095,
      "step": 130924
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.1301610469818115,
      "learning_rate": 0.00023617363762591176,
      "loss": 2.1642,
      "step": 130925
    },
    {
      "epoch": 1.7,
      "grad_norm": 0.982721209526062,
      "learning_rate": 0.0002361696407203225,
      "loss": 2.1558,
      "step": 130926
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.0558271408081055,
      "learning_rate": 0.00023616564382660053,
      "loss": 1.9837,
      "step": 130927
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.1029226779937744,
      "learning_rate": 0.00023616164694474653,
      "loss": 2.0448,
      "step": 130928
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.126955270767212,
      "learning_rate": 0.0002361576500747613,
      "loss": 2.0827,
      "step": 130929
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.0757091045379639,
      "learning_rate": 0.00023615365321664566,
      "loss": 2.078,
      "step": 130930
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.033767819404602,
      "learning_rate": 0.00023614965637040018,
      "loss": 1.9958,
      "step": 130931
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.023285150527954,
      "learning_rate": 0.00023614565953602567,
      "loss": 2.2579,
      "step": 130932
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.1546422243118286,
      "learning_rate": 0.00023614166271352288,
      "loss": 1.9315,
      "step": 130933
    },
    {
      "epoch": 1.7,
      "grad_norm": 0.98770672082901,
      "learning_rate": 0.0002361376659028926,
      "loss": 2.3121,
      "step": 130934
    },
    {
      "epoch": 1.7,
      "grad_norm": 0.9414249062538147,
      "learning_rate": 0.0002361336691041355,
      "loss": 2.0537,
      "step": 130935
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.0899018049240112,
      "learning_rate": 0.00023612967231725248,
      "loss": 2.2509,
      "step": 130936
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.0490665435791016,
      "learning_rate": 0.00023612567554224403,
      "loss": 2.2066,
      "step": 130937
    },
    {
      "epoch": 1.7,
      "grad_norm": 0.9336898922920227,
      "learning_rate": 0.00023612167877911104,
      "loss": 1.7564,
      "step": 130938
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.011381983757019,
      "learning_rate": 0.00023611768202785424,
      "loss": 2.2299,
      "step": 130939
    },
    {
      "epoch": 1.7,
      "grad_norm": 0.9797970056533813,
      "learning_rate": 0.00023611368528847433,
      "loss": 2.0059,
      "step": 130940
    },
    {
      "epoch": 1.7,
      "grad_norm": 0.9559387564659119,
      "learning_rate": 0.00023610968856097213,
      "loss": 2.1028,
      "step": 130941
    },
    {
      "epoch": 1.7,
      "grad_norm": 0.9425408244132996,
      "learning_rate": 0.00023610569184534847,
      "loss": 2.2077,
      "step": 130942
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.0370726585388184,
      "learning_rate": 0.00023610169514160382,
      "loss": 2.0579,
      "step": 130943
    },
    {
      "epoch": 1.7,
      "grad_norm": 0.9589065909385681,
      "learning_rate": 0.00023609769844973907,
      "loss": 2.199,
      "step": 130944
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.1427252292633057,
      "learning_rate": 0.00023609370176975498,
      "loss": 2.1952,
      "step": 130945
    },
    {
      "epoch": 1.71,
      "grad_norm": 0.9715902805328369,
      "learning_rate": 0.00023608970510165229,
      "loss": 1.8247,
      "step": 130946
    },
    {
      "epoch": 1.71,
      "grad_norm": 0.9988251328468323,
      "learning_rate": 0.0002360857084454317,
      "loss": 2.3251,
      "step": 130947
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.018633484840393,
      "learning_rate": 0.00023608171180109413,
      "loss": 2.0299,
      "step": 130948
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.0763295888900757,
      "learning_rate": 0.00023607771516864006,
      "loss": 2.1657,
      "step": 130949
    },
    {
      "epoch": 1.71,
      "grad_norm": 0.9600459933280945,
      "learning_rate": 0.0002360737185480703,
      "loss": 1.9954,
      "step": 130950
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.0992965698242188,
      "learning_rate": 0.00023606972193938568,
      "loss": 1.8645,
      "step": 130951
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.0237451791763306,
      "learning_rate": 0.00023606572534258692,
      "loss": 2.1206,
      "step": 130952
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.0328444242477417,
      "learning_rate": 0.0002360617287576747,
      "loss": 2.1393,
      "step": 130953
    },
    {
      "epoch": 1.71,
      "grad_norm": 0.9727014899253845,
      "learning_rate": 0.00023605773218464993,
      "loss": 1.9701,
      "step": 130954
    },
    {
      "epoch": 1.71,
      "grad_norm": 0.9569635391235352,
      "learning_rate": 0.00023605373562351318,
      "loss": 2.0857,
      "step": 130955
    },
    {
      "epoch": 1.71,
      "grad_norm": 0.9973533749580383,
      "learning_rate": 0.0002360497390742652,
      "loss": 2.0233,
      "step": 130956
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.0557968616485596,
      "learning_rate": 0.00023604574253690682,
      "loss": 2.0558,
      "step": 130957
    },
    {
      "epoch": 1.71,
      "grad_norm": 0.984972357749939,
      "learning_rate": 0.00023604174601143873,
      "loss": 2.0261,
      "step": 130958
    },
    {
      "epoch": 1.71,
      "grad_norm": 0.9668052196502686,
      "learning_rate": 0.00023603774949786165,
      "loss": 2.2519,
      "step": 130959
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.3712918758392334,
      "learning_rate": 0.00023603375299617647,
      "loss": 1.9723,
      "step": 130960
    },
    {
      "epoch": 1.71,
      "grad_norm": 0.9569602608680725,
      "learning_rate": 0.0002360297565063837,
      "loss": 2.3104,
      "step": 130961
    },
    {
      "epoch": 1.71,
      "grad_norm": 0.9671324491500854,
      "learning_rate": 0.00023602576002848433,
      "loss": 2.0027,
      "step": 130962
    },
    {
      "epoch": 1.71,
      "grad_norm": 0.9987306594848633,
      "learning_rate": 0.0002360217635624789,
      "loss": 2.0206,
      "step": 130963
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.0096174478530884,
      "learning_rate": 0.00023601776710836821,
      "loss": 2.175,
      "step": 130964
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.017783761024475,
      "learning_rate": 0.00023601377066615302,
      "loss": 2.2644,
      "step": 130965
    },
    {
      "epoch": 1.71,
      "grad_norm": 0.9400402903556824,
      "learning_rate": 0.00023600977423583416,
      "loss": 1.9209,
      "step": 130966
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.1429283618927002,
      "learning_rate": 0.00023600577781741222,
      "loss": 2.0467,
      "step": 130967
    },
    {
      "epoch": 1.71,
      "grad_norm": 0.9381863474845886,
      "learning_rate": 0.00023600178141088804,
      "loss": 2.1563,
      "step": 130968
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.1602362394332886,
      "learning_rate": 0.0002359977850162624,
      "loss": 2.1688,
      "step": 130969
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.057525873184204,
      "learning_rate": 0.00023599378863353585,
      "loss": 1.9856,
      "step": 130970
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.0009660720825195,
      "learning_rate": 0.00023598979226270933,
      "loss": 2.0926,
      "step": 130971
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.0637130737304688,
      "learning_rate": 0.00023598579590378355,
      "loss": 2.2208,
      "step": 130972
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.0152488946914673,
      "learning_rate": 0.00023598179955675917,
      "loss": 2.1019,
      "step": 130973
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.0209213495254517,
      "learning_rate": 0.000235977803221637,
      "loss": 1.9605,
      "step": 130974
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.0688949823379517,
      "learning_rate": 0.0002359738068984177,
      "loss": 2.1584,
      "step": 130975
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.1254777908325195,
      "learning_rate": 0.0002359698105871022,
      "loss": 2.2331,
      "step": 130976
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.1005923748016357,
      "learning_rate": 0.00023596581428769105,
      "loss": 2.0584,
      "step": 130977
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.1435327529907227,
      "learning_rate": 0.0002359618180001851,
      "loss": 2.2499,
      "step": 130978
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.0663543939590454,
      "learning_rate": 0.000235957821724585,
      "loss": 2.0036,
      "step": 130979
    },
    {
      "epoch": 1.71,
      "grad_norm": 0.952492892742157,
      "learning_rate": 0.00023595382546089157,
      "loss": 2.2655,
      "step": 130980
    },
    {
      "epoch": 1.71,
      "grad_norm": 0.9770531058311462,
      "learning_rate": 0.00023594982920910554,
      "loss": 2.0448,
      "step": 130981
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.0538804531097412,
      "learning_rate": 0.0002359458329692276,
      "loss": 2.091,
      "step": 130982
    },
    {
      "epoch": 1.71,
      "grad_norm": 0.9404017329216003,
      "learning_rate": 0.0002359418367412586,
      "loss": 2.1263,
      "step": 130983
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.0110059976577759,
      "learning_rate": 0.00023593784052519922,
      "loss": 2.0752,
      "step": 130984
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.08644700050354,
      "learning_rate": 0.00023593384432105027,
      "loss": 2.0361,
      "step": 130985
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.1084250211715698,
      "learning_rate": 0.0002359298481288123,
      "loss": 2.0754,
      "step": 130986
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.00355863571167,
      "learning_rate": 0.00023592585194848623,
      "loss": 2.0845,
      "step": 130987
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.1345839500427246,
      "learning_rate": 0.00023592185578007272,
      "loss": 2.4234,
      "step": 130988
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.0126020908355713,
      "learning_rate": 0.00023591785962357259,
      "loss": 2.2732,
      "step": 130989
    },
    {
      "epoch": 1.71,
      "grad_norm": 0.9148356914520264,
      "learning_rate": 0.0002359138634789865,
      "loss": 2.3309,
      "step": 130990
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.2122173309326172,
      "learning_rate": 0.00023590986734631537,
      "loss": 2.1309,
      "step": 130991
    },
    {
      "epoch": 1.71,
      "grad_norm": 0.9650294780731201,
      "learning_rate": 0.00023590587122555967,
      "loss": 2.2413,
      "step": 130992
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.0137510299682617,
      "learning_rate": 0.0002359018751167203,
      "loss": 2.2241,
      "step": 130993
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.0449775457382202,
      "learning_rate": 0.00023589787901979797,
      "loss": 2.1492,
      "step": 130994
    },
    {
      "epoch": 1.71,
      "grad_norm": 0.9902450442314148,
      "learning_rate": 0.0002358938829347934,
      "loss": 2.0299,
      "step": 130995
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.100356936454773,
      "learning_rate": 0.00023588988686170745,
      "loss": 2.1384,
      "step": 130996
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.1277738809585571,
      "learning_rate": 0.00023588589080054086,
      "loss": 2.0752,
      "step": 130997
    },
    {
      "epoch": 1.71,
      "grad_norm": 0.9076622724533081,
      "learning_rate": 0.00023588189475129415,
      "loss": 2.0129,
      "step": 130998
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.1656546592712402,
      "learning_rate": 0.0002358778987139682,
      "loss": 2.1492,
      "step": 130999
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.0558432340621948,
      "learning_rate": 0.00023587390268856379,
      "loss": 2.0476,
      "step": 131000
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.1514933109283447,
      "learning_rate": 0.00023586990667508163,
      "loss": 2.0487,
      "step": 131001
    },
    {
      "epoch": 1.71,
      "grad_norm": 0.9432054162025452,
      "learning_rate": 0.00023586591067352244,
      "loss": 2.1852,
      "step": 131002
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.0371623039245605,
      "learning_rate": 0.00023586191468388714,
      "loss": 1.9902,
      "step": 131003
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.2247812747955322,
      "learning_rate": 0.0002358579187061762,
      "loss": 1.9459,
      "step": 131004
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.1022452116012573,
      "learning_rate": 0.00023585392274039048,
      "loss": 2.3438,
      "step": 131005
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.03976571559906,
      "learning_rate": 0.0002358499267865307,
      "loss": 2.1006,
      "step": 131006
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.0644888877868652,
      "learning_rate": 0.00023584593084459763,
      "loss": 2.1918,
      "step": 131007
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.0733373165130615,
      "learning_rate": 0.00023584193491459204,
      "loss": 2.0347,
      "step": 131008
    },
    {
      "epoch": 1.71,
      "grad_norm": 0.9924851059913635,
      "learning_rate": 0.00023583793899651477,
      "loss": 2.4199,
      "step": 131009
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.0145630836486816,
      "learning_rate": 0.00023583394309036626,
      "loss": 1.9696,
      "step": 131010
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.1092954874038696,
      "learning_rate": 0.0002358299471961475,
      "loss": 1.9446,
      "step": 131011
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.1289622783660889,
      "learning_rate": 0.00023582595131385915,
      "loss": 2.1999,
      "step": 131012
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.192209243774414,
      "learning_rate": 0.00023582195544350195,
      "loss": 2.0796,
      "step": 131013
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.1854552030563354,
      "learning_rate": 0.00023581795958507664,
      "loss": 2.0703,
      "step": 131014
    },
    {
      "epoch": 1.71,
      "grad_norm": 0.9707993268966675,
      "learning_rate": 0.00023581396373858414,
      "loss": 2.0384,
      "step": 131015
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.0733845233917236,
      "learning_rate": 0.0002358099679040249,
      "loss": 2.1293,
      "step": 131016
    },
    {
      "epoch": 1.71,
      "grad_norm": 0.994705080986023,
      "learning_rate": 0.0002358059720813998,
      "loss": 2.4185,
      "step": 131017
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.0634230375289917,
      "learning_rate": 0.0002358019762707096,
      "loss": 1.8506,
      "step": 131018
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.0044277906417847,
      "learning_rate": 0.000235797980471955,
      "loss": 2.1897,
      "step": 131019
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.0433481931686401,
      "learning_rate": 0.00023579398468513677,
      "loss": 2.0262,
      "step": 131020
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.050629734992981,
      "learning_rate": 0.0002357899889102558,
      "loss": 1.8987,
      "step": 131021
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.1008795499801636,
      "learning_rate": 0.0002357859931473125,
      "loss": 2.0298,
      "step": 131022
    },
    {
      "epoch": 1.71,
      "grad_norm": 0.9505615830421448,
      "learning_rate": 0.00023578199739630784,
      "loss": 2.076,
      "step": 131023
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.010259985923767,
      "learning_rate": 0.00023577800165724253,
      "loss": 2.1061,
      "step": 131024
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.1406937837600708,
      "learning_rate": 0.00023577400593011727,
      "loss": 2.2188,
      "step": 131025
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.1657997369766235,
      "learning_rate": 0.00023577001021493285,
      "loss": 2.4567,
      "step": 131026
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.0066934823989868,
      "learning_rate": 0.00023576601451169015,
      "loss": 2.386,
      "step": 131027
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.093372106552124,
      "learning_rate": 0.0002357620188203896,
      "loss": 2.0325,
      "step": 131028
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.079346776008606,
      "learning_rate": 0.00023575802314103212,
      "loss": 1.8492,
      "step": 131029
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.0826375484466553,
      "learning_rate": 0.0002357540274736184,
      "loss": 2.245,
      "step": 131030
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.2473479509353638,
      "learning_rate": 0.00023575003181814926,
      "loss": 2.2157,
      "step": 131031
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.0694258213043213,
      "learning_rate": 0.00023574603617462536,
      "loss": 1.9686,
      "step": 131032
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.0643967390060425,
      "learning_rate": 0.00023574204054304767,
      "loss": 2.0988,
      "step": 131033
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.05918288230896,
      "learning_rate": 0.00023573804492341656,
      "loss": 2.053,
      "step": 131034
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.0367523431777954,
      "learning_rate": 0.00023573404931573298,
      "loss": 2.1062,
      "step": 131035
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.005839467048645,
      "learning_rate": 0.00023573005371999768,
      "loss": 2.1451,
      "step": 131036
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.046889305114746,
      "learning_rate": 0.00023572605813621135,
      "loss": 2.0906,
      "step": 131037
    },
    {
      "epoch": 1.71,
      "grad_norm": 0.9831898808479309,
      "learning_rate": 0.0002357220625643748,
      "loss": 2.0485,
      "step": 131038
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.0459173917770386,
      "learning_rate": 0.0002357180670044888,
      "loss": 2.1276,
      "step": 131039
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.1164573431015015,
      "learning_rate": 0.00023571407145655393,
      "loss": 2.2918,
      "step": 131040
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.0216718912124634,
      "learning_rate": 0.00023571007592057102,
      "loss": 2.0619,
      "step": 131041
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.1984202861785889,
      "learning_rate": 0.00023570608039654082,
      "loss": 2.3789,
      "step": 131042
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.0168788433074951,
      "learning_rate": 0.00023570208488446408,
      "loss": 2.3305,
      "step": 131043
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.0153837203979492,
      "learning_rate": 0.00023569808938434154,
      "loss": 2.1189,
      "step": 131044
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.0282095670700073,
      "learning_rate": 0.000235694093896174,
      "loss": 2.0779,
      "step": 131045
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.0535660982131958,
      "learning_rate": 0.00023569009841996208,
      "loss": 2.0624,
      "step": 131046
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.128098487854004,
      "learning_rate": 0.00023568610295570663,
      "loss": 2.2646,
      "step": 131047
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.0643787384033203,
      "learning_rate": 0.0002356821075034083,
      "loss": 2.0416,
      "step": 131048
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.1131224632263184,
      "learning_rate": 0.00023567811206306788,
      "loss": 2.1577,
      "step": 131049
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.054049015045166,
      "learning_rate": 0.0002356741166346861,
      "loss": 1.9976,
      "step": 131050
    },
    {
      "epoch": 1.71,
      "grad_norm": 0.9765579104423523,
      "learning_rate": 0.00023567012121826372,
      "loss": 1.9151,
      "step": 131051
    },
    {
      "epoch": 1.71,
      "grad_norm": 0.9982093572616577,
      "learning_rate": 0.00023566612581380156,
      "loss": 2.3978,
      "step": 131052
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.034088373184204,
      "learning_rate": 0.0002356621304213002,
      "loss": 2.1714,
      "step": 131053
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.2808818817138672,
      "learning_rate": 0.00023565813504076053,
      "loss": 2.2068,
      "step": 131054
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.0774266719818115,
      "learning_rate": 0.00023565413967218318,
      "loss": 1.9559,
      "step": 131055
    },
    {
      "epoch": 1.71,
      "grad_norm": 0.9574542045593262,
      "learning_rate": 0.0002356501443155689,
      "loss": 2.0652,
      "step": 131056
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.1162880659103394,
      "learning_rate": 0.0002356461489709185,
      "loss": 2.0285,
      "step": 131057
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.0884273052215576,
      "learning_rate": 0.0002356421536382328,
      "loss": 2.0052,
      "step": 131058
    },
    {
      "epoch": 1.71,
      "grad_norm": 0.995393693447113,
      "learning_rate": 0.0002356381583175123,
      "loss": 2.1138,
      "step": 131059
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.023924469947815,
      "learning_rate": 0.00023563416300875791,
      "loss": 2.1349,
      "step": 131060
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.017890453338623,
      "learning_rate": 0.00023563016771197042,
      "loss": 2.0672,
      "step": 131061
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.0424537658691406,
      "learning_rate": 0.00023562617242715045,
      "loss": 2.0622,
      "step": 131062
    },
    {
      "epoch": 1.71,
      "grad_norm": 0.9611743092536926,
      "learning_rate": 0.00023562217715429875,
      "loss": 2.01,
      "step": 131063
    },
    {
      "epoch": 1.71,
      "grad_norm": 0.9463305473327637,
      "learning_rate": 0.0002356181818934162,
      "loss": 2.0926,
      "step": 131064
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.0330908298492432,
      "learning_rate": 0.00023561418664450336,
      "loss": 2.1157,
      "step": 131065
    },
    {
      "epoch": 1.71,
      "grad_norm": 0.9664115905761719,
      "learning_rate": 0.00023561019140756105,
      "loss": 1.9513,
      "step": 131066
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.2827414274215698,
      "learning_rate": 0.00023560619618259002,
      "loss": 2.0713,
      "step": 131067
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.0551189184188843,
      "learning_rate": 0.0002356022009695911,
      "loss": 2.2211,
      "step": 131068
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.0815255641937256,
      "learning_rate": 0.0002355982057685649,
      "loss": 2.1426,
      "step": 131069
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.0010229349136353,
      "learning_rate": 0.00023559421057951224,
      "loss": 2.1344,
      "step": 131070
    },
    {
      "epoch": 1.71,
      "grad_norm": 0.9998950362205505,
      "learning_rate": 0.00023559021540243378,
      "loss": 2.0968,
      "step": 131071
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.0669784545898438,
      "learning_rate": 0.0002355862202373303,
      "loss": 2.2309,
      "step": 131072
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.098396897315979,
      "learning_rate": 0.00023558222508420257,
      "loss": 2.1397,
      "step": 131073
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.0237462520599365,
      "learning_rate": 0.0002355782299430513,
      "loss": 1.9149,
      "step": 131074
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.0761334896087646,
      "learning_rate": 0.00023557423481387725,
      "loss": 2.1016,
      "step": 131075
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.5428634881973267,
      "learning_rate": 0.0002355702396966813,
      "loss": 2.1304,
      "step": 131076
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.056171178817749,
      "learning_rate": 0.00023556624459146396,
      "loss": 2.3217,
      "step": 131077
    },
    {
      "epoch": 1.71,
      "grad_norm": 0.9650994539260864,
      "learning_rate": 0.00023556224949822607,
      "loss": 2.0007,
      "step": 131078
    },
    {
      "epoch": 1.71,
      "grad_norm": 0.9252890944480896,
      "learning_rate": 0.00023555825441696837,
      "loss": 2.0012,
      "step": 131079
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.3065515756607056,
      "learning_rate": 0.00023555425934769156,
      "loss": 2.0859,
      "step": 131080
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.093480110168457,
      "learning_rate": 0.00023555026429039649,
      "loss": 2.222,
      "step": 131081
    },
    {
      "epoch": 1.71,
      "grad_norm": 0.9456325173377991,
      "learning_rate": 0.00023554626924508396,
      "loss": 2.0608,
      "step": 131082
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.003493070602417,
      "learning_rate": 0.00023554227421175442,
      "loss": 2.1458,
      "step": 131083
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.0158897638320923,
      "learning_rate": 0.00023553827919040886,
      "loss": 2.0336,
      "step": 131084
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.0977752208709717,
      "learning_rate": 0.00023553428418104791,
      "loss": 2.0874,
      "step": 131085
    },
    {
      "epoch": 1.71,
      "grad_norm": 0.9561148881912231,
      "learning_rate": 0.00023553028918367236,
      "loss": 1.9288,
      "step": 131086
    },
    {
      "epoch": 1.71,
      "grad_norm": 0.9905745387077332,
      "learning_rate": 0.00023552629419828293,
      "loss": 2.0182,
      "step": 131087
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.1209911108016968,
      "learning_rate": 0.0002355222992248805,
      "loss": 2.1742,
      "step": 131088
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.106191873550415,
      "learning_rate": 0.0002355183042634656,
      "loss": 2.1998,
      "step": 131089
    },
    {
      "epoch": 1.71,
      "grad_norm": 0.9781343340873718,
      "learning_rate": 0.00023551430931403902,
      "loss": 2.0699,
      "step": 131090
    },
    {
      "epoch": 1.71,
      "grad_norm": 0.8865758776664734,
      "learning_rate": 0.0002355103143766016,
      "loss": 2.0732,
      "step": 131091
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.003831148147583,
      "learning_rate": 0.000235506319451154,
      "loss": 2.272,
      "step": 131092
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.0195149183273315,
      "learning_rate": 0.00023550232453769702,
      "loss": 1.9765,
      "step": 131093
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.1656765937805176,
      "learning_rate": 0.00023549832963623145,
      "loss": 2.0978,
      "step": 131094
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.0737831592559814,
      "learning_rate": 0.0002354943347467579,
      "loss": 1.8741,
      "step": 131095
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.0647141933441162,
      "learning_rate": 0.0002354903398692771,
      "loss": 2.0118,
      "step": 131096
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.0047334432601929,
      "learning_rate": 0.0002354863450037899,
      "loss": 2.2481,
      "step": 131097
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.0283232927322388,
      "learning_rate": 0.000235482350150297,
      "loss": 2.2561,
      "step": 131098
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.2083461284637451,
      "learning_rate": 0.00023547835530879916,
      "loss": 2.1083,
      "step": 131099
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.1102718114852905,
      "learning_rate": 0.00023547436047929724,
      "loss": 2.0502,
      "step": 131100
    },
    {
      "epoch": 1.71,
      "grad_norm": 0.9813631772994995,
      "learning_rate": 0.0002354703656617917,
      "loss": 1.9211,
      "step": 131101
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.2742480039596558,
      "learning_rate": 0.00023546637085628344,
      "loss": 1.9789,
      "step": 131102
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.1362286806106567,
      "learning_rate": 0.00023546237606277324,
      "loss": 2.0886,
      "step": 131103
    },
    {
      "epoch": 1.71,
      "grad_norm": 0.9656556248664856,
      "learning_rate": 0.00023545838128126177,
      "loss": 2.086,
      "step": 131104
    },
    {
      "epoch": 1.71,
      "grad_norm": 0.9971414804458618,
      "learning_rate": 0.00023545438651174978,
      "loss": 2.2806,
      "step": 131105
    },
    {
      "epoch": 1.71,
      "grad_norm": 0.9344306588172913,
      "learning_rate": 0.0002354503917542382,
      "loss": 2.0768,
      "step": 131106
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.0512349605560303,
      "learning_rate": 0.00023544639700872748,
      "loss": 2.119,
      "step": 131107
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.115770697593689,
      "learning_rate": 0.00023544240227521852,
      "loss": 2.1429,
      "step": 131108
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.0432478189468384,
      "learning_rate": 0.000235438407553712,
      "loss": 2.2925,
      "step": 131109
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.0423845052719116,
      "learning_rate": 0.0002354344128442087,
      "loss": 2.1155,
      "step": 131110
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.0302335023880005,
      "learning_rate": 0.00023543041814670938,
      "loss": 2.1865,
      "step": 131111
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.0066038370132446,
      "learning_rate": 0.00023542642346121488,
      "loss": 2.3382,
      "step": 131112
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.1010123491287231,
      "learning_rate": 0.0002354224287877257,
      "loss": 2.2126,
      "step": 131113
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.0505355596542358,
      "learning_rate": 0.0002354184341262427,
      "loss": 2.1265,
      "step": 131114
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.0068151950836182,
      "learning_rate": 0.00023541443947676664,
      "loss": 2.215,
      "step": 131115
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.1170724630355835,
      "learning_rate": 0.00023541044483929826,
      "loss": 2.0976,
      "step": 131116
    },
    {
      "epoch": 1.71,
      "grad_norm": 0.9664531946182251,
      "learning_rate": 0.00023540645021383829,
      "loss": 2.1079,
      "step": 131117
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.0072790384292603,
      "learning_rate": 0.0002354024556003875,
      "loss": 2.0267,
      "step": 131118
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.0393997430801392,
      "learning_rate": 0.0002353984609989467,
      "loss": 1.874,
      "step": 131119
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.0354907512664795,
      "learning_rate": 0.00023539446640951647,
      "loss": 1.968,
      "step": 131120
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.009892225265503,
      "learning_rate": 0.0002353904718320976,
      "loss": 2.1444,
      "step": 131121
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.0059454441070557,
      "learning_rate": 0.00023538647726669085,
      "loss": 2.1833,
      "step": 131122
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.382959246635437,
      "learning_rate": 0.00023538248271329698,
      "loss": 2.1587,
      "step": 131123
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.1941847801208496,
      "learning_rate": 0.00023537848817191674,
      "loss": 2.2102,
      "step": 131124
    },
    {
      "epoch": 1.71,
      "grad_norm": 0.9266253709793091,
      "learning_rate": 0.00023537449364255096,
      "loss": 2.1664,
      "step": 131125
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.4643954038619995,
      "learning_rate": 0.0002353704991252002,
      "loss": 2.072,
      "step": 131126
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.2591456174850464,
      "learning_rate": 0.00023536650461986524,
      "loss": 2.0892,
      "step": 131127
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.0446749925613403,
      "learning_rate": 0.00023536251012654688,
      "loss": 2.3029,
      "step": 131128
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.0575369596481323,
      "learning_rate": 0.00023535851564524587,
      "loss": 1.9796,
      "step": 131129
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.1935124397277832,
      "learning_rate": 0.0002353545211759629,
      "loss": 2.0784,
      "step": 131130
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.4767087697982788,
      "learning_rate": 0.0002353505267186989,
      "loss": 2.0335,
      "step": 131131
    },
    {
      "epoch": 1.71,
      "grad_norm": 0.9832591414451599,
      "learning_rate": 0.00023534653227345434,
      "loss": 2.1548,
      "step": 131132
    },
    {
      "epoch": 1.71,
      "grad_norm": 0.9972568154335022,
      "learning_rate": 0.00023534253784023005,
      "loss": 2.2512,
      "step": 131133
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.293563723564148,
      "learning_rate": 0.0002353385434190268,
      "loss": 2.3278,
      "step": 131134
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.4163826704025269,
      "learning_rate": 0.0002353345490098453,
      "loss": 2.3907,
      "step": 131135
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.148300290107727,
      "learning_rate": 0.0002353305546126864,
      "loss": 2.1234,
      "step": 131136
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.0036497116088867,
      "learning_rate": 0.00023532656022755082,
      "loss": 2.1448,
      "step": 131137
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.1472804546356201,
      "learning_rate": 0.00023532256585443917,
      "loss": 2.1118,
      "step": 131138
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.1028568744659424,
      "learning_rate": 0.00023531857149335233,
      "loss": 2.1536,
      "step": 131139
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.0489921569824219,
      "learning_rate": 0.00023531457714429097,
      "loss": 2.1192,
      "step": 131140
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.1206185817718506,
      "learning_rate": 0.00023531058280725582,
      "loss": 2.3743,
      "step": 131141
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.083570957183838,
      "learning_rate": 0.00023530658848224765,
      "loss": 2.1522,
      "step": 131142
    },
    {
      "epoch": 1.71,
      "grad_norm": 0.9700318574905396,
      "learning_rate": 0.00023530259416926729,
      "loss": 1.8554,
      "step": 131143
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.0006963014602661,
      "learning_rate": 0.0002352985998683153,
      "loss": 2.0141,
      "step": 131144
    },
    {
      "epoch": 1.71,
      "grad_norm": 0.97588050365448,
      "learning_rate": 0.00023529460557939258,
      "loss": 2.1727,
      "step": 131145
    },
    {
      "epoch": 1.71,
      "grad_norm": 0.9967066049575806,
      "learning_rate": 0.0002352906113024998,
      "loss": 2.1959,
      "step": 131146
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.4232025146484375,
      "learning_rate": 0.0002352866170376377,
      "loss": 2.0398,
      "step": 131147
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.0941020250320435,
      "learning_rate": 0.00023528262278480706,
      "loss": 2.052,
      "step": 131148
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.0743508338928223,
      "learning_rate": 0.0002352786285440086,
      "loss": 2.1492,
      "step": 131149
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.0921841859817505,
      "learning_rate": 0.00023527463431524307,
      "loss": 2.1268,
      "step": 131150
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.2153698205947876,
      "learning_rate": 0.00023527064009851118,
      "loss": 2.2402,
      "step": 131151
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.107972502708435,
      "learning_rate": 0.0002352666458938137,
      "loss": 1.8826,
      "step": 131152
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.1141796112060547,
      "learning_rate": 0.00023526265170115143,
      "loss": 1.884,
      "step": 131153
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.1249678134918213,
      "learning_rate": 0.000235258657520525,
      "loss": 2.0028,
      "step": 131154
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.0677189826965332,
      "learning_rate": 0.00023525466335193527,
      "loss": 2.2283,
      "step": 131155
    },
    {
      "epoch": 1.71,
      "grad_norm": 0.8837507963180542,
      "learning_rate": 0.00023525066919538284,
      "loss": 1.9871,
      "step": 131156
    },
    {
      "epoch": 1.71,
      "grad_norm": 0.9560343027114868,
      "learning_rate": 0.00023524667505086857,
      "loss": 2.0208,
      "step": 131157
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.0788697004318237,
      "learning_rate": 0.0002352426809183931,
      "loss": 2.0336,
      "step": 131158
    },
    {
      "epoch": 1.71,
      "grad_norm": 0.9924587607383728,
      "learning_rate": 0.00023523868679795726,
      "loss": 2.3051,
      "step": 131159
    },
    {
      "epoch": 1.71,
      "grad_norm": 0.9401460289955139,
      "learning_rate": 0.0002352346926895619,
      "loss": 1.945,
      "step": 131160
    },
    {
      "epoch": 1.71,
      "grad_norm": 0.993060827255249,
      "learning_rate": 0.00023523069859320757,
      "loss": 1.9986,
      "step": 131161
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.5722994804382324,
      "learning_rate": 0.00023522670450889504,
      "loss": 2.089,
      "step": 131162
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.3583208322525024,
      "learning_rate": 0.00023522271043662505,
      "loss": 2.1213,
      "step": 131163
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.0179861783981323,
      "learning_rate": 0.00023521871637639845,
      "loss": 2.3068,
      "step": 131164
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.059227466583252,
      "learning_rate": 0.00023521472232821585,
      "loss": 2.2005,
      "step": 131165
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.2727552652359009,
      "learning_rate": 0.00023521072829207806,
      "loss": 2.1866,
      "step": 131166
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.1718281507492065,
      "learning_rate": 0.00023520673426798597,
      "loss": 1.9328,
      "step": 131167
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.1355235576629639,
      "learning_rate": 0.00023520274025594004,
      "loss": 2.1606,
      "step": 131168
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.1383470296859741,
      "learning_rate": 0.00023519874625594115,
      "loss": 2.1139,
      "step": 131169
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.393683910369873,
      "learning_rate": 0.00023519475226799002,
      "loss": 2.0544,
      "step": 131170
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.07142174243927,
      "learning_rate": 0.0002351907582920874,
      "loss": 1.9807,
      "step": 131171
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.078977108001709,
      "learning_rate": 0.00023518676432823408,
      "loss": 2.0983,
      "step": 131172
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.534934639930725,
      "learning_rate": 0.00023518277037643085,
      "loss": 1.9776,
      "step": 131173
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.1536296606063843,
      "learning_rate": 0.00023517877643667827,
      "loss": 2.228,
      "step": 131174
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.0081435441970825,
      "learning_rate": 0.00023517478250897718,
      "loss": 1.9971,
      "step": 131175
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.085053563117981,
      "learning_rate": 0.0002351707885933283,
      "loss": 2.06,
      "step": 131176
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.0268555879592896,
      "learning_rate": 0.00023516679468973237,
      "loss": 2.1734,
      "step": 131177
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.1046099662780762,
      "learning_rate": 0.00023516280079819023,
      "loss": 2.0981,
      "step": 131178
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.0053751468658447,
      "learning_rate": 0.00023515880691870263,
      "loss": 2.1568,
      "step": 131179
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.029809594154358,
      "learning_rate": 0.00023515481305127014,
      "loss": 2.2904,
      "step": 131180
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.0762031078338623,
      "learning_rate": 0.00023515081919589354,
      "loss": 2.088,
      "step": 131181
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.0283358097076416,
      "learning_rate": 0.00023514682535257365,
      "loss": 2.1339,
      "step": 131182
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.0696301460266113,
      "learning_rate": 0.00023514283152131118,
      "loss": 2.156,
      "step": 131183
    },
    {
      "epoch": 1.71,
      "grad_norm": 0.9518836736679077,
      "learning_rate": 0.0002351388377021069,
      "loss": 2.1769,
      "step": 131184
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.0220026969909668,
      "learning_rate": 0.00023513484389496152,
      "loss": 1.9797,
      "step": 131185
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.0134966373443604,
      "learning_rate": 0.00023513085009987594,
      "loss": 2.0612,
      "step": 131186
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.056220293045044,
      "learning_rate": 0.00023512685631685062,
      "loss": 1.9736,
      "step": 131187
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.058097243309021,
      "learning_rate": 0.00023512286254588646,
      "loss": 1.9957,
      "step": 131188
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.069887399673462,
      "learning_rate": 0.00023511886878698417,
      "loss": 2.0072,
      "step": 131189
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.0285940170288086,
      "learning_rate": 0.0002351148750401445,
      "loss": 2.1557,
      "step": 131190
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.01146399974823,
      "learning_rate": 0.00023511088130536822,
      "loss": 2.0862,
      "step": 131191
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.0032014846801758,
      "learning_rate": 0.00023510688758265614,
      "loss": 2.1022,
      "step": 131192
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.1930242776870728,
      "learning_rate": 0.00023510289387200883,
      "loss": 1.9451,
      "step": 131193
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.0037565231323242,
      "learning_rate": 0.00023509890017342708,
      "loss": 2.076,
      "step": 131194
    },
    {
      "epoch": 1.71,
      "grad_norm": 0.9731062054634094,
      "learning_rate": 0.0002350949064869117,
      "loss": 2.0867,
      "step": 131195
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.0985761880874634,
      "learning_rate": 0.00023509091281246342,
      "loss": 2.1116,
      "step": 131196
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.2172300815582275,
      "learning_rate": 0.00023508691915008291,
      "loss": 2.0801,
      "step": 131197
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.148480772972107,
      "learning_rate": 0.00023508292549977112,
      "loss": 2.1771,
      "step": 131198
    },
    {
      "epoch": 1.71,
      "grad_norm": 0.965534508228302,
      "learning_rate": 0.00023507893186152854,
      "loss": 2.2288,
      "step": 131199
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.2898708581924438,
      "learning_rate": 0.00023507493823535596,
      "loss": 2.1965,
      "step": 131200
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.1851812601089478,
      "learning_rate": 0.00023507094462125422,
      "loss": 2.1607,
      "step": 131201
    },
    {
      "epoch": 1.71,
      "grad_norm": 0.9442337155342102,
      "learning_rate": 0.00023506695101922399,
      "loss": 2.0008,
      "step": 131202
    },
    {
      "epoch": 1.71,
      "grad_norm": 0.9308156967163086,
      "learning_rate": 0.00023506295742926602,
      "loss": 1.9961,
      "step": 131203
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.002314567565918,
      "learning_rate": 0.00023505896385138123,
      "loss": 1.9644,
      "step": 131204
    },
    {
      "epoch": 1.71,
      "grad_norm": 0.9178110957145691,
      "learning_rate": 0.0002350549702855701,
      "loss": 2.1002,
      "step": 131205
    },
    {
      "epoch": 1.71,
      "grad_norm": 0.9572445154190063,
      "learning_rate": 0.00023505097673183343,
      "loss": 2.0101,
      "step": 131206
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.0148507356643677,
      "learning_rate": 0.00023504698319017206,
      "loss": 2.1532,
      "step": 131207
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.1072934865951538,
      "learning_rate": 0.00023504298966058664,
      "loss": 2.1251,
      "step": 131208
    },
    {
      "epoch": 1.71,
      "grad_norm": 0.9979596138000488,
      "learning_rate": 0.00023503899614307796,
      "loss": 2.0277,
      "step": 131209
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.0119479894638062,
      "learning_rate": 0.00023503500263764688,
      "loss": 1.9525,
      "step": 131210
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.0936369895935059,
      "learning_rate": 0.0002350310091442939,
      "loss": 2.2008,
      "step": 131211
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.1996625661849976,
      "learning_rate": 0.00023502701566301986,
      "loss": 2.3318,
      "step": 131212
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.0748783349990845,
      "learning_rate": 0.00023502302219382558,
      "loss": 2.2256,
      "step": 131213
    },
    {
      "epoch": 1.71,
      "grad_norm": 0.9927871823310852,
      "learning_rate": 0.0002350190287367117,
      "loss": 1.9376,
      "step": 131214
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.1254197359085083,
      "learning_rate": 0.00023501503529167902,
      "loss": 1.9271,
      "step": 131215
    },
    {
      "epoch": 1.71,
      "grad_norm": 0.9768048524856567,
      "learning_rate": 0.00023501104185872843,
      "loss": 2.1107,
      "step": 131216
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.0179237127304077,
      "learning_rate": 0.00023500704843786035,
      "loss": 1.9858,
      "step": 131217
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.0407729148864746,
      "learning_rate": 0.00023500305502907568,
      "loss": 1.8593,
      "step": 131218
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.158728003501892,
      "learning_rate": 0.0002349990616323752,
      "loss": 2.1835,
      "step": 131219
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.1235771179199219,
      "learning_rate": 0.00023499506824775963,
      "loss": 2.2644,
      "step": 131220
    },
    {
      "epoch": 1.71,
      "grad_norm": 0.9726409316062927,
      "learning_rate": 0.00023499107487522965,
      "loss": 1.9662,
      "step": 131221
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.0290491580963135,
      "learning_rate": 0.00023498708151478614,
      "loss": 2.3359,
      "step": 131222
    },
    {
      "epoch": 1.71,
      "grad_norm": 0.9350491762161255,
      "learning_rate": 0.0002349830881664298,
      "loss": 2.0028,
      "step": 131223
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.0353139638900757,
      "learning_rate": 0.00023497909483016124,
      "loss": 2.2825,
      "step": 131224
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.09174382686615,
      "learning_rate": 0.00023497510150598128,
      "loss": 2.033,
      "step": 131225
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.1471725702285767,
      "learning_rate": 0.00023497110819389067,
      "loss": 1.9927,
      "step": 131226
    },
    {
      "epoch": 1.71,
      "grad_norm": 0.9710027575492859,
      "learning_rate": 0.0002349671148938902,
      "loss": 2.2931,
      "step": 131227
    },
    {
      "epoch": 1.71,
      "grad_norm": 0.9893244504928589,
      "learning_rate": 0.00023496312160598058,
      "loss": 1.9341,
      "step": 131228
    },
    {
      "epoch": 1.71,
      "grad_norm": 0.9744608998298645,
      "learning_rate": 0.0002349591283301625,
      "loss": 2.0669,
      "step": 131229
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.004898190498352,
      "learning_rate": 0.0002349551350664368,
      "loss": 2.0634,
      "step": 131230
    },
    {
      "epoch": 1.71,
      "grad_norm": 0.9958885908126831,
      "learning_rate": 0.00023495114181480416,
      "loss": 1.9899,
      "step": 131231
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.023153305053711,
      "learning_rate": 0.0002349471485752653,
      "loss": 2.2222,
      "step": 131232
    },
    {
      "epoch": 1.71,
      "grad_norm": 0.9502928853034973,
      "learning_rate": 0.00023494315534782097,
      "loss": 2.1284,
      "step": 131233
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.0315508842468262,
      "learning_rate": 0.00023493916213247203,
      "loss": 2.1638,
      "step": 131234
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.1815526485443115,
      "learning_rate": 0.00023493516892921906,
      "loss": 1.9696,
      "step": 131235
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.0527743101119995,
      "learning_rate": 0.00023493117573806287,
      "loss": 2.1228,
      "step": 131236
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.257247805595398,
      "learning_rate": 0.0002349271825590042,
      "loss": 2.2508,
      "step": 131237
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.0340025424957275,
      "learning_rate": 0.00023492318939204382,
      "loss": 2.1593,
      "step": 131238
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.0236018896102905,
      "learning_rate": 0.00023491919623718242,
      "loss": 2.1121,
      "step": 131239
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.0373218059539795,
      "learning_rate": 0.00023491520309442082,
      "loss": 2.1242,
      "step": 131240
    },
    {
      "epoch": 1.71,
      "grad_norm": 0.9894125461578369,
      "learning_rate": 0.00023491120996375968,
      "loss": 2.2431,
      "step": 131241
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.0832411050796509,
      "learning_rate": 0.00023490721684519972,
      "loss": 2.1766,
      "step": 131242
    },
    {
      "epoch": 1.71,
      "grad_norm": 0.9646374583244324,
      "learning_rate": 0.00023490322373874176,
      "loss": 2.2008,
      "step": 131243
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.0987951755523682,
      "learning_rate": 0.00023489923064438654,
      "loss": 2.2631,
      "step": 131244
    },
    {
      "epoch": 1.71,
      "grad_norm": 0.9405092597007751,
      "learning_rate": 0.00023489523756213486,
      "loss": 2.0275,
      "step": 131245
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.0088242292404175,
      "learning_rate": 0.00023489124449198735,
      "loss": 2.1471,
      "step": 131246
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.1585800647735596,
      "learning_rate": 0.00023488725143394474,
      "loss": 2.0288,
      "step": 131247
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.0150355100631714,
      "learning_rate": 0.0002348832583880078,
      "loss": 2.1173,
      "step": 131248
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.0255846977233887,
      "learning_rate": 0.0002348792653541773,
      "loss": 2.1261,
      "step": 131249
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.1151012182235718,
      "learning_rate": 0.00023487527233245398,
      "loss": 2.1839,
      "step": 131250
    },
    {
      "epoch": 1.71,
      "grad_norm": 0.9729562401771545,
      "learning_rate": 0.00023487127932283856,
      "loss": 2.2404,
      "step": 131251
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.0231842994689941,
      "learning_rate": 0.0002348672863253319,
      "loss": 2.3146,
      "step": 131252
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.0114041566848755,
      "learning_rate": 0.0002348632933399346,
      "loss": 2.0301,
      "step": 131253
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.0260250568389893,
      "learning_rate": 0.0002348593003666474,
      "loss": 2.1432,
      "step": 131254
    },
    {
      "epoch": 1.71,
      "grad_norm": 0.9776662588119507,
      "learning_rate": 0.00023485530740547107,
      "loss": 2.1829,
      "step": 131255
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.1107661724090576,
      "learning_rate": 0.0002348513144564064,
      "loss": 2.1312,
      "step": 131256
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.050000548362732,
      "learning_rate": 0.00023484732151945407,
      "loss": 2.0031,
      "step": 131257
    },
    {
      "epoch": 1.71,
      "grad_norm": 0.9692866802215576,
      "learning_rate": 0.00023484332859461486,
      "loss": 2.3235,
      "step": 131258
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.0177257061004639,
      "learning_rate": 0.00023483933568188965,
      "loss": 2.3817,
      "step": 131259
    },
    {
      "epoch": 1.71,
      "grad_norm": 0.9567149877548218,
      "learning_rate": 0.0002348353427812789,
      "loss": 2.2078,
      "step": 131260
    },
    {
      "epoch": 1.71,
      "grad_norm": 0.9577694535255432,
      "learning_rate": 0.00023483134989278344,
      "loss": 2.3237,
      "step": 131261
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.048946499824524,
      "learning_rate": 0.0002348273570164041,
      "loss": 2.1648,
      "step": 131262
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.024009346961975,
      "learning_rate": 0.00023482336415214163,
      "loss": 2.177,
      "step": 131263
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.0052433013916016,
      "learning_rate": 0.00023481937129999668,
      "loss": 1.9614,
      "step": 131264
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.1407561302185059,
      "learning_rate": 0.00023481537845997016,
      "loss": 2.1958,
      "step": 131265
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.0558366775512695,
      "learning_rate": 0.00023481138563206258,
      "loss": 2.1884,
      "step": 131266
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.0335994958877563,
      "learning_rate": 0.00023480739281627483,
      "loss": 1.9669,
      "step": 131267
    },
    {
      "epoch": 1.71,
      "grad_norm": 0.9459065794944763,
      "learning_rate": 0.00023480340001260755,
      "loss": 2.2899,
      "step": 131268
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.0838115215301514,
      "learning_rate": 0.00023479940722106159,
      "loss": 2.1975,
      "step": 131269
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.1823023557662964,
      "learning_rate": 0.00023479541444163763,
      "loss": 2.2668,
      "step": 131270
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.0163748264312744,
      "learning_rate": 0.0002347914216743366,
      "loss": 2.1018,
      "step": 131271
    },
    {
      "epoch": 1.71,
      "grad_norm": 0.925880491733551,
      "learning_rate": 0.0002347874289191589,
      "loss": 2.2011,
      "step": 131272
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.0014675855636597,
      "learning_rate": 0.00023478343617610545,
      "loss": 2.248,
      "step": 131273
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.1286144256591797,
      "learning_rate": 0.00023477944344517702,
      "loss": 2.2559,
      "step": 131274
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.1312123537063599,
      "learning_rate": 0.0002347754507263743,
      "loss": 2.1559,
      "step": 131275
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.0097980499267578,
      "learning_rate": 0.0002347714580196981,
      "loss": 2.2232,
      "step": 131276
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.3594108819961548,
      "learning_rate": 0.00023476746532514916,
      "loss": 2.2158,
      "step": 131277
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.4511923789978027,
      "learning_rate": 0.0002347634726427281,
      "loss": 2.0478,
      "step": 131278
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.067405104637146,
      "learning_rate": 0.00023475947997243576,
      "loss": 1.9521,
      "step": 131279
    },
    {
      "epoch": 1.71,
      "grad_norm": 0.9384472370147705,
      "learning_rate": 0.00023475548731427285,
      "loss": 1.9788,
      "step": 131280
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.1910923719406128,
      "learning_rate": 0.0002347514946682401,
      "loss": 2.109,
      "step": 131281
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.0115673542022705,
      "learning_rate": 0.00023474750203433833,
      "loss": 2.2152,
      "step": 131282
    },
    {
      "epoch": 1.71,
      "grad_norm": 0.9965211749076843,
      "learning_rate": 0.00023474350941256832,
      "loss": 2.0977,
      "step": 131283
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.1136243343353271,
      "learning_rate": 0.0002347395168029306,
      "loss": 2.0219,
      "step": 131284
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.23007333278656,
      "learning_rate": 0.00023473552420542605,
      "loss": 2.0894,
      "step": 131285
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.0105434656143188,
      "learning_rate": 0.0002347315316200554,
      "loss": 1.8531,
      "step": 131286
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.2096378803253174,
      "learning_rate": 0.00023472753904681937,
      "loss": 2.1219,
      "step": 131287
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.1610770225524902,
      "learning_rate": 0.00023472354648571873,
      "loss": 2.1045,
      "step": 131288
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.1948319673538208,
      "learning_rate": 0.00023471955393675436,
      "loss": 2.2734,
      "step": 131289
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.3981781005859375,
      "learning_rate": 0.0002347155613999267,
      "loss": 1.8192,
      "step": 131290
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.0761538743972778,
      "learning_rate": 0.00023471156887523665,
      "loss": 2.1001,
      "step": 131291
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.0386615991592407,
      "learning_rate": 0.000234707576362685,
      "loss": 2.102,
      "step": 131292
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.2858370542526245,
      "learning_rate": 0.0002347035838622724,
      "loss": 2.0665,
      "step": 131293
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.3854715824127197,
      "learning_rate": 0.00023469959137399965,
      "loss": 2.3016,
      "step": 131294
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.0016838312149048,
      "learning_rate": 0.00023469559889786762,
      "loss": 1.9959,
      "step": 131295
    },
    {
      "epoch": 1.71,
      "grad_norm": 0.9744511246681213,
      "learning_rate": 0.00023469160643387673,
      "loss": 2.3105,
      "step": 131296
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.256050944328308,
      "learning_rate": 0.00023468761398202795,
      "loss": 2.2888,
      "step": 131297
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.5511242151260376,
      "learning_rate": 0.00023468362154232197,
      "loss": 2.1442,
      "step": 131298
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.1145992279052734,
      "learning_rate": 0.00023467962911475954,
      "loss": 2.3485,
      "step": 131299
    },
    {
      "epoch": 1.71,
      "grad_norm": 0.9656940698623657,
      "learning_rate": 0.0002346756366993414,
      "loss": 2.1925,
      "step": 131300
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.305285930633545,
      "learning_rate": 0.0002346716442960684,
      "loss": 2.0991,
      "step": 131301
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.3683090209960938,
      "learning_rate": 0.00023466765190494106,
      "loss": 2.3052,
      "step": 131302
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.1203166246414185,
      "learning_rate": 0.00023466365952596024,
      "loss": 2.1859,
      "step": 131303
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.0276787281036377,
      "learning_rate": 0.00023465966715912665,
      "loss": 2.2359,
      "step": 131304
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.2307922840118408,
      "learning_rate": 0.0002346556748044411,
      "loss": 2.27,
      "step": 131305
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.1796770095825195,
      "learning_rate": 0.0002346516824619043,
      "loss": 2.5271,
      "step": 131306
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.0093938112258911,
      "learning_rate": 0.000234647690131517,
      "loss": 2.2448,
      "step": 131307
    },
    {
      "epoch": 1.71,
      "grad_norm": 0.9525160789489746,
      "learning_rate": 0.00023464369781327996,
      "loss": 1.9991,
      "step": 131308
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.302087426185608,
      "learning_rate": 0.00023463970550719384,
      "loss": 2.2191,
      "step": 131309
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.0961345434188843,
      "learning_rate": 0.0002346357132132594,
      "loss": 2.0756,
      "step": 131310
    },
    {
      "epoch": 1.71,
      "grad_norm": 0.914334774017334,
      "learning_rate": 0.00023463172093147745,
      "loss": 2.0443,
      "step": 131311
    },
    {
      "epoch": 1.71,
      "grad_norm": 0.9460351467132568,
      "learning_rate": 0.00023462772866184866,
      "loss": 2.0996,
      "step": 131312
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.0551177263259888,
      "learning_rate": 0.0002346237364043739,
      "loss": 2.1872,
      "step": 131313
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.5578278303146362,
      "learning_rate": 0.00023461974415905375,
      "loss": 2.2356,
      "step": 131314
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.2130671739578247,
      "learning_rate": 0.00023461575192588907,
      "loss": 2.2094,
      "step": 131315
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.06082284450531,
      "learning_rate": 0.00023461175970488054,
      "loss": 2.234,
      "step": 131316
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.5346673727035522,
      "learning_rate": 0.00023460776749602887,
      "loss": 2.2695,
      "step": 131317
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.174100399017334,
      "learning_rate": 0.00023460377529933488,
      "loss": 2.1434,
      "step": 131318
    },
    {
      "epoch": 1.71,
      "grad_norm": 0.9646824598312378,
      "learning_rate": 0.00023459978311479927,
      "loss": 1.9619,
      "step": 131319
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.1650817394256592,
      "learning_rate": 0.00023459579094242284,
      "loss": 2.0801,
      "step": 131320
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.093995213508606,
      "learning_rate": 0.00023459179878220624,
      "loss": 2.0465,
      "step": 131321
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.0526182651519775,
      "learning_rate": 0.0002345878066341503,
      "loss": 2.2306,
      "step": 131322
    },
    {
      "epoch": 1.71,
      "grad_norm": 0.9938543438911438,
      "learning_rate": 0.00023458381449825572,
      "loss": 2.0514,
      "step": 131323
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.012755274772644,
      "learning_rate": 0.0002345798223745232,
      "loss": 2.1039,
      "step": 131324
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.0128206014633179,
      "learning_rate": 0.00023457583026295355,
      "loss": 2.0601,
      "step": 131325
    },
    {
      "epoch": 1.71,
      "grad_norm": 0.9523657560348511,
      "learning_rate": 0.0002345718381635475,
      "loss": 1.9786,
      "step": 131326
    },
    {
      "epoch": 1.71,
      "grad_norm": 0.9272923469543457,
      "learning_rate": 0.00023456784607630576,
      "loss": 2.2757,
      "step": 131327
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.1490980386734009,
      "learning_rate": 0.00023456385400122908,
      "loss": 2.131,
      "step": 131328
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.046286940574646,
      "learning_rate": 0.00023455986193831823,
      "loss": 1.844,
      "step": 131329
    },
    {
      "epoch": 1.71,
      "grad_norm": 0.9599178433418274,
      "learning_rate": 0.000234555869887574,
      "loss": 2.3103,
      "step": 131330
    },
    {
      "epoch": 1.71,
      "grad_norm": 0.9729325175285339,
      "learning_rate": 0.00023455187784899697,
      "loss": 2.1903,
      "step": 131331
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.0343891382217407,
      "learning_rate": 0.00023454788582258807,
      "loss": 2.0991,
      "step": 131332
    },
    {
      "epoch": 1.71,
      "grad_norm": 0.9583587646484375,
      "learning_rate": 0.00023454389380834787,
      "loss": 2.1567,
      "step": 131333
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.0478733777999878,
      "learning_rate": 0.00023453990180627725,
      "loss": 2.26,
      "step": 131334
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.0803744792938232,
      "learning_rate": 0.00023453590981637688,
      "loss": 2.0732,
      "step": 131335
    },
    {
      "epoch": 1.71,
      "grad_norm": 0.9924022555351257,
      "learning_rate": 0.0002345319178386475,
      "loss": 2.2489,
      "step": 131336
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.010008454322815,
      "learning_rate": 0.0002345279258730899,
      "loss": 2.2608,
      "step": 131337
    },
    {
      "epoch": 1.71,
      "grad_norm": 0.9609294533729553,
      "learning_rate": 0.00023452393391970485,
      "loss": 2.1236,
      "step": 131338
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.061132788658142,
      "learning_rate": 0.00023451994197849294,
      "loss": 2.139,
      "step": 131339
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.0312670469284058,
      "learning_rate": 0.00023451595004945504,
      "loss": 1.9086,
      "step": 131340
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.000447392463684,
      "learning_rate": 0.00023451195813259184,
      "loss": 2.1572,
      "step": 131341
    },
    {
      "epoch": 1.71,
      "grad_norm": 0.9626176953315735,
      "learning_rate": 0.0002345079662279041,
      "loss": 2.0785,
      "step": 131342
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.054214358329773,
      "learning_rate": 0.0002345039743353926,
      "loss": 2.2185,
      "step": 131343
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.0109169483184814,
      "learning_rate": 0.00023449998245505814,
      "loss": 2.1533,
      "step": 131344
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.1851094961166382,
      "learning_rate": 0.00023449599058690123,
      "loss": 1.943,
      "step": 131345
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.1012086868286133,
      "learning_rate": 0.00023449199873092278,
      "loss": 2.0688,
      "step": 131346
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.1463844776153564,
      "learning_rate": 0.0002344880068871235,
      "loss": 2.1197,
      "step": 131347
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.18247389793396,
      "learning_rate": 0.00023448401505550413,
      "loss": 2.1108,
      "step": 131348
    },
    {
      "epoch": 1.71,
      "grad_norm": 0.9051800966262817,
      "learning_rate": 0.0002344800232360654,
      "loss": 1.8368,
      "step": 131349
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.1077409982681274,
      "learning_rate": 0.00023447603142880827,
      "loss": 2.2642,
      "step": 131350
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.111419439315796,
      "learning_rate": 0.0002344720396337331,
      "loss": 2.2631,
      "step": 131351
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.0048319101333618,
      "learning_rate": 0.00023446804785084082,
      "loss": 2.0658,
      "step": 131352
    },
    {
      "epoch": 1.71,
      "grad_norm": 0.9709125757217407,
      "learning_rate": 0.00023446405608013214,
      "loss": 1.9713,
      "step": 131353
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.1380141973495483,
      "learning_rate": 0.00023446006432160786,
      "loss": 2.0255,
      "step": 131354
    },
    {
      "epoch": 1.71,
      "grad_norm": 0.9843365550041199,
      "learning_rate": 0.0002344560725752687,
      "loss": 2.1115,
      "step": 131355
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.0089027881622314,
      "learning_rate": 0.0002344520808411155,
      "loss": 2.0959,
      "step": 131356
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.008571743965149,
      "learning_rate": 0.00023444808911914878,
      "loss": 2.3097,
      "step": 131357
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.0629843473434448,
      "learning_rate": 0.0002344440974093694,
      "loss": 2.0681,
      "step": 131358
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.0253956317901611,
      "learning_rate": 0.00023444010571177807,
      "loss": 2.0824,
      "step": 131359
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.040317177772522,
      "learning_rate": 0.00023443611402637558,
      "loss": 2.0817,
      "step": 131360
    },
    {
      "epoch": 1.71,
      "grad_norm": 0.934653103351593,
      "learning_rate": 0.0002344321223531627,
      "loss": 2.2077,
      "step": 131361
    },
    {
      "epoch": 1.71,
      "grad_norm": 0.9670472145080566,
      "learning_rate": 0.00023442813069214017,
      "loss": 2.1824,
      "step": 131362
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.0981825590133667,
      "learning_rate": 0.00023442413904330864,
      "loss": 2.2169,
      "step": 131363
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.1398252248764038,
      "learning_rate": 0.00023442014740666884,
      "loss": 2.1907,
      "step": 131364
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.033327579498291,
      "learning_rate": 0.0002344161557822216,
      "loss": 2.2119,
      "step": 131365
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.042809247970581,
      "learning_rate": 0.00023441216416996764,
      "loss": 1.955,
      "step": 131366
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.0190258026123047,
      "learning_rate": 0.0002344081725699077,
      "loss": 2.0959,
      "step": 131367
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.1139276027679443,
      "learning_rate": 0.0002344041809820426,
      "loss": 2.1512,
      "step": 131368
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.025468111038208,
      "learning_rate": 0.00023440018940637294,
      "loss": 2.0552,
      "step": 131369
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.0501989126205444,
      "learning_rate": 0.00023439619784289945,
      "loss": 2.1825,
      "step": 131370
    },
    {
      "epoch": 1.71,
      "grad_norm": 0.952911376953125,
      "learning_rate": 0.000234392206291623,
      "loss": 2.1512,
      "step": 131371
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.0526676177978516,
      "learning_rate": 0.00023438821475254429,
      "loss": 2.1547,
      "step": 131372
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.0055676698684692,
      "learning_rate": 0.000234384223225664,
      "loss": 1.7513,
      "step": 131373
    },
    {
      "epoch": 1.71,
      "grad_norm": 0.964028537273407,
      "learning_rate": 0.00023438023171098307,
      "loss": 2.3205,
      "step": 131374
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.0782862901687622,
      "learning_rate": 0.00023437624020850197,
      "loss": 2.0988,
      "step": 131375
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.0955954790115356,
      "learning_rate": 0.00023437224871822155,
      "loss": 2.3855,
      "step": 131376
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.163608193397522,
      "learning_rate": 0.0002343682572401426,
      "loss": 2.0974,
      "step": 131377
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.0285391807556152,
      "learning_rate": 0.0002343642657742658,
      "loss": 2.2264,
      "step": 131378
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.0071276426315308,
      "learning_rate": 0.00023436027432059193,
      "loss": 2.0226,
      "step": 131379
    },
    {
      "epoch": 1.71,
      "grad_norm": 0.9533605575561523,
      "learning_rate": 0.00023435628287912184,
      "loss": 2.0194,
      "step": 131380
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.0672717094421387,
      "learning_rate": 0.00023435229144985607,
      "loss": 2.1533,
      "step": 131381
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.1483148336410522,
      "learning_rate": 0.0002343483000327954,
      "loss": 2.2686,
      "step": 131382
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.1023142337799072,
      "learning_rate": 0.00023434430862794066,
      "loss": 2.1639,
      "step": 131383
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.0506765842437744,
      "learning_rate": 0.00023434031723529252,
      "loss": 2.0527,
      "step": 131384
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.0101686716079712,
      "learning_rate": 0.00023433632585485176,
      "loss": 2.3122,
      "step": 131385
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.0480475425720215,
      "learning_rate": 0.00023433233448661914,
      "loss": 2.0944,
      "step": 131386
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.1384414434432983,
      "learning_rate": 0.00023432834313059548,
      "loss": 2.2778,
      "step": 131387
    },
    {
      "epoch": 1.71,
      "grad_norm": 0.9739472270011902,
      "learning_rate": 0.00023432435178678132,
      "loss": 2.2598,
      "step": 131388
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.0174415111541748,
      "learning_rate": 0.00023432036045517747,
      "loss": 2.096,
      "step": 131389
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.0887359380722046,
      "learning_rate": 0.00023431636913578472,
      "loss": 2.0528,
      "step": 131390
    },
    {
      "epoch": 1.71,
      "grad_norm": 0.9304754137992859,
      "learning_rate": 0.0002343123778286038,
      "loss": 2.1166,
      "step": 131391
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.053109049797058,
      "learning_rate": 0.00023430838653363544,
      "loss": 2.287,
      "step": 131392
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.055695652961731,
      "learning_rate": 0.00023430439525088054,
      "loss": 2.1461,
      "step": 131393
    },
    {
      "epoch": 1.71,
      "grad_norm": 0.940056562423706,
      "learning_rate": 0.00023430040398033953,
      "loss": 2.211,
      "step": 131394
    },
    {
      "epoch": 1.71,
      "grad_norm": 0.9784838557243347,
      "learning_rate": 0.00023429641272201335,
      "loss": 2.3457,
      "step": 131395
    },
    {
      "epoch": 1.71,
      "grad_norm": 0.9885107278823853,
      "learning_rate": 0.0002342924214759027,
      "loss": 2.0181,
      "step": 131396
    },
    {
      "epoch": 1.71,
      "grad_norm": 0.9623497724533081,
      "learning_rate": 0.00023428843024200834,
      "loss": 1.9675,
      "step": 131397
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.4196069240570068,
      "learning_rate": 0.00023428443902033096,
      "loss": 2.1994,
      "step": 131398
    },
    {
      "epoch": 1.71,
      "grad_norm": 0.9597802758216858,
      "learning_rate": 0.00023428044781087143,
      "loss": 2.2568,
      "step": 131399
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.1287099123001099,
      "learning_rate": 0.0002342764566136304,
      "loss": 2.0675,
      "step": 131400
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.2546130418777466,
      "learning_rate": 0.00023427246542860855,
      "loss": 2.3296,
      "step": 131401
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.0760538578033447,
      "learning_rate": 0.0002342684742558067,
      "loss": 2.1939,
      "step": 131402
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.1262274980545044,
      "learning_rate": 0.0002342644830952256,
      "loss": 1.8914,
      "step": 131403
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.0776853561401367,
      "learning_rate": 0.00023426049194686594,
      "loss": 2.1486,
      "step": 131404
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.1095951795578003,
      "learning_rate": 0.0002342565008107286,
      "loss": 2.0697,
      "step": 131405
    },
    {
      "epoch": 1.71,
      "grad_norm": 0.9782801866531372,
      "learning_rate": 0.00023425250968681414,
      "loss": 2.0556,
      "step": 131406
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.014072060585022,
      "learning_rate": 0.00023424851857512338,
      "loss": 2.0501,
      "step": 131407
    },
    {
      "epoch": 1.71,
      "grad_norm": 0.963742733001709,
      "learning_rate": 0.00023424452747565708,
      "loss": 1.9245,
      "step": 131408
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.0301614999771118,
      "learning_rate": 0.00023424053638841592,
      "loss": 1.9596,
      "step": 131409
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.003955364227295,
      "learning_rate": 0.00023423654531340072,
      "loss": 1.9675,
      "step": 131410
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.009865403175354,
      "learning_rate": 0.00023423255425061222,
      "loss": 2.1854,
      "step": 131411
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.0434589385986328,
      "learning_rate": 0.00023422856320005108,
      "loss": 1.92,
      "step": 131412
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.013691782951355,
      "learning_rate": 0.00023422457216171807,
      "loss": 2.187,
      "step": 131413
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.239408016204834,
      "learning_rate": 0.00023422058113561404,
      "loss": 2.1863,
      "step": 131414
    },
    {
      "epoch": 1.71,
      "grad_norm": 0.9944255948066711,
      "learning_rate": 0.00023421659012173957,
      "loss": 1.956,
      "step": 131415
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.0334099531173706,
      "learning_rate": 0.00023421259912009547,
      "loss": 2.0584,
      "step": 131416
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.1676360368728638,
      "learning_rate": 0.00023420860813068258,
      "loss": 2.096,
      "step": 131417
    },
    {
      "epoch": 1.71,
      "grad_norm": 0.9840588569641113,
      "learning_rate": 0.00023420461715350146,
      "loss": 2.235,
      "step": 131418
    },
    {
      "epoch": 1.71,
      "grad_norm": 0.9376606345176697,
      "learning_rate": 0.00023420062618855297,
      "loss": 2.2389,
      "step": 131419
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.0875087976455688,
      "learning_rate": 0.0002341966352358378,
      "loss": 2.3705,
      "step": 131420
    },
    {
      "epoch": 1.71,
      "grad_norm": 0.995270311832428,
      "learning_rate": 0.00023419264429535682,
      "loss": 2.015,
      "step": 131421
    },
    {
      "epoch": 1.71,
      "grad_norm": 0.9649338722229004,
      "learning_rate": 0.00023418865336711058,
      "loss": 2.1096,
      "step": 131422
    },
    {
      "epoch": 1.71,
      "grad_norm": 0.9540078639984131,
      "learning_rate": 0.00023418466245109996,
      "loss": 2.0797,
      "step": 131423
    },
    {
      "epoch": 1.71,
      "grad_norm": 0.9568403959274292,
      "learning_rate": 0.00023418067154732563,
      "loss": 2.0491,
      "step": 131424
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.0385655164718628,
      "learning_rate": 0.0002341766806557883,
      "loss": 2.1341,
      "step": 131425
    },
    {
      "epoch": 1.71,
      "grad_norm": 0.9940601587295532,
      "learning_rate": 0.0002341726897764888,
      "loss": 2.2035,
      "step": 131426
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.0509775876998901,
      "learning_rate": 0.00023416869890942785,
      "loss": 2.0051,
      "step": 131427
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.0521506071090698,
      "learning_rate": 0.0002341647080546062,
      "loss": 2.163,
      "step": 131428
    },
    {
      "epoch": 1.71,
      "grad_norm": 0.9461464881896973,
      "learning_rate": 0.00023416071721202463,
      "loss": 1.957,
      "step": 131429
    },
    {
      "epoch": 1.71,
      "grad_norm": 0.9775406718254089,
      "learning_rate": 0.00023415672638168374,
      "loss": 2.1278,
      "step": 131430
    },
    {
      "epoch": 1.71,
      "grad_norm": 0.9695388078689575,
      "learning_rate": 0.00023415273556358436,
      "loss": 2.1726,
      "step": 131431
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.0507198572158813,
      "learning_rate": 0.00023414874475772725,
      "loss": 2.0678,
      "step": 131432
    },
    {
      "epoch": 1.71,
      "grad_norm": 0.9622932076454163,
      "learning_rate": 0.00023414475396411308,
      "loss": 2.0353,
      "step": 131433
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.0988526344299316,
      "learning_rate": 0.00023414076318274267,
      "loss": 1.9987,
      "step": 131434
    },
    {
      "epoch": 1.71,
      "grad_norm": 0.9240514636039734,
      "learning_rate": 0.00023413677241361687,
      "loss": 2.0756,
      "step": 131435
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.163527011871338,
      "learning_rate": 0.00023413278165673615,
      "loss": 2.2057,
      "step": 131436
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.103136658668518,
      "learning_rate": 0.00023412879091210141,
      "loss": 2.3404,
      "step": 131437
    },
    {
      "epoch": 1.71,
      "grad_norm": 0.9765742421150208,
      "learning_rate": 0.00023412480017971337,
      "loss": 2.069,
      "step": 131438
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.02506685256958,
      "learning_rate": 0.00023412080945957274,
      "loss": 1.9733,
      "step": 131439
    },
    {
      "epoch": 1.71,
      "grad_norm": 0.9602890014648438,
      "learning_rate": 0.0002341168187516803,
      "loss": 2.1437,
      "step": 131440
    },
    {
      "epoch": 1.71,
      "grad_norm": 0.9423272609710693,
      "learning_rate": 0.00023411282805603693,
      "loss": 1.9011,
      "step": 131441
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.0128793716430664,
      "learning_rate": 0.0002341088373726431,
      "loss": 2.0267,
      "step": 131442
    },
    {
      "epoch": 1.71,
      "grad_norm": 0.9791486263275146,
      "learning_rate": 0.0002341048467014997,
      "loss": 2.1134,
      "step": 131443
    },
    {
      "epoch": 1.71,
      "grad_norm": 0.9949279427528381,
      "learning_rate": 0.0002341008560426074,
      "loss": 2.044,
      "step": 131444
    },
    {
      "epoch": 1.71,
      "grad_norm": 0.906555712223053,
      "learning_rate": 0.00023409686539596703,
      "loss": 2.1244,
      "step": 131445
    },
    {
      "epoch": 1.71,
      "grad_norm": 0.9198043346405029,
      "learning_rate": 0.0002340928747615793,
      "loss": 2.021,
      "step": 131446
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.0154374837875366,
      "learning_rate": 0.0002340888841394451,
      "loss": 2.0209,
      "step": 131447
    },
    {
      "epoch": 1.71,
      "grad_norm": 0.9595025181770325,
      "learning_rate": 0.00023408489352956484,
      "loss": 2.3805,
      "step": 131448
    },
    {
      "epoch": 1.71,
      "grad_norm": 0.9769300818443298,
      "learning_rate": 0.00023408090293193946,
      "loss": 2.0212,
      "step": 131449
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.0212794542312622,
      "learning_rate": 0.0002340769123465697,
      "loss": 2.1046,
      "step": 131450
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.021903395652771,
      "learning_rate": 0.00023407292177345625,
      "loss": 2.1263,
      "step": 131451
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.218052864074707,
      "learning_rate": 0.00023406893121259993,
      "loss": 1.8666,
      "step": 131452
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.0468813180923462,
      "learning_rate": 0.00023406494066400145,
      "loss": 2.3466,
      "step": 131453
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.0335501432418823,
      "learning_rate": 0.00023406095012766163,
      "loss": 2.047,
      "step": 131454
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.0601773262023926,
      "learning_rate": 0.00023405695960358101,
      "loss": 2.3642,
      "step": 131455
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.0739654302597046,
      "learning_rate": 0.00023405296909176043,
      "loss": 2.1284,
      "step": 131456
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.0601261854171753,
      "learning_rate": 0.00023404897859220067,
      "loss": 1.905,
      "step": 131457
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.050377368927002,
      "learning_rate": 0.00023404498810490244,
      "loss": 2.4482,
      "step": 131458
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.0641674995422363,
      "learning_rate": 0.0002340409976298665,
      "loss": 2.0964,
      "step": 131459
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.014019250869751,
      "learning_rate": 0.00023403700716709372,
      "loss": 2.1402,
      "step": 131460
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.1585601568222046,
      "learning_rate": 0.00023403301671658455,
      "loss": 2.4351,
      "step": 131461
    },
    {
      "epoch": 1.71,
      "grad_norm": 0.9109075665473938,
      "learning_rate": 0.0002340290262783399,
      "loss": 2.1922,
      "step": 131462
    },
    {
      "epoch": 1.71,
      "grad_norm": 0.9287638068199158,
      "learning_rate": 0.0002340250358523605,
      "loss": 1.9077,
      "step": 131463
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.0311840772628784,
      "learning_rate": 0.0002340210454386471,
      "loss": 2.2315,
      "step": 131464
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.0029624700546265,
      "learning_rate": 0.00023401705503720043,
      "loss": 2.0891,
      "step": 131465
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.1032694578170776,
      "learning_rate": 0.00023401306464802135,
      "loss": 2.4029,
      "step": 131466
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.005100965499878,
      "learning_rate": 0.00023400907427111037,
      "loss": 1.7068,
      "step": 131467
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.0530251264572144,
      "learning_rate": 0.00023400508390646836,
      "loss": 1.8988,
      "step": 131468
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.0291069746017456,
      "learning_rate": 0.00023400109355409604,
      "loss": 2.0667,
      "step": 131469
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.0544044971466064,
      "learning_rate": 0.00023399710321399416,
      "loss": 2.1508,
      "step": 131470
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.034410834312439,
      "learning_rate": 0.00023399311288616346,
      "loss": 2.0786,
      "step": 131471
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.1131093502044678,
      "learning_rate": 0.0002339891225706048,
      "loss": 1.9522,
      "step": 131472
    },
    {
      "epoch": 1.71,
      "grad_norm": 0.9918280839920044,
      "learning_rate": 0.0002339851322673187,
      "loss": 2.1527,
      "step": 131473
    },
    {
      "epoch": 1.71,
      "grad_norm": 0.9695849418640137,
      "learning_rate": 0.00023398114197630603,
      "loss": 2.1499,
      "step": 131474
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.4305907487869263,
      "learning_rate": 0.00023397715169756747,
      "loss": 1.9145,
      "step": 131475
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.0501766204833984,
      "learning_rate": 0.00023397316143110384,
      "loss": 2.117,
      "step": 131476
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.1030335426330566,
      "learning_rate": 0.00023396917117691582,
      "loss": 2.3145,
      "step": 131477
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.0209423303604126,
      "learning_rate": 0.00023396518093500432,
      "loss": 2.1411,
      "step": 131478
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.1062346696853638,
      "learning_rate": 0.00023396119070536982,
      "loss": 2.2529,
      "step": 131479
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.0567858219146729,
      "learning_rate": 0.00023395720048801315,
      "loss": 1.9596,
      "step": 131480
    },
    {
      "epoch": 1.71,
      "grad_norm": 0.9822980761528015,
      "learning_rate": 0.00023395321028293512,
      "loss": 1.9836,
      "step": 131481
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.1145740747451782,
      "learning_rate": 0.00023394922009013642,
      "loss": 2.0868,
      "step": 131482
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.2490049600601196,
      "learning_rate": 0.00023394522990961782,
      "loss": 2.1299,
      "step": 131483
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.0866022109985352,
      "learning_rate": 0.00023394123974138014,
      "loss": 2.0471,
      "step": 131484
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.0260837078094482,
      "learning_rate": 0.0002339372495854239,
      "loss": 1.9292,
      "step": 131485
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.0345107316970825,
      "learning_rate": 0.00023393325944175004,
      "loss": 2.4581,
      "step": 131486
    },
    {
      "epoch": 1.71,
      "grad_norm": 0.9467102885246277,
      "learning_rate": 0.00023392926931035916,
      "loss": 1.7398,
      "step": 131487
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.14682936668396,
      "learning_rate": 0.0002339252791912521,
      "loss": 2.1144,
      "step": 131488
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.061043620109558,
      "learning_rate": 0.0002339212890844296,
      "loss": 2.1149,
      "step": 131489
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.03324294090271,
      "learning_rate": 0.00023391729898989243,
      "loss": 2.048,
      "step": 131490
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.221741795539856,
      "learning_rate": 0.00023391330890764123,
      "loss": 2.0341,
      "step": 131491
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.0464329719543457,
      "learning_rate": 0.00023390931883767683,
      "loss": 1.9899,
      "step": 131492
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.0708109140396118,
      "learning_rate": 0.00023390532877999985,
      "loss": 2.1081,
      "step": 131493
    },
    {
      "epoch": 1.71,
      "grad_norm": 0.982791006565094,
      "learning_rate": 0.00023390133873461117,
      "loss": 2.1434,
      "step": 131494
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.0863679647445679,
      "learning_rate": 0.00023389734870151143,
      "loss": 1.9371,
      "step": 131495
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.1916749477386475,
      "learning_rate": 0.00023389335868070153,
      "loss": 2.2198,
      "step": 131496
    },
    {
      "epoch": 1.71,
      "grad_norm": 0.9907304644584656,
      "learning_rate": 0.00023388936867218204,
      "loss": 2.4339,
      "step": 131497
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.4654667377471924,
      "learning_rate": 0.00023388537867595374,
      "loss": 2.0072,
      "step": 131498
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.1437989473342896,
      "learning_rate": 0.00023388138869201745,
      "loss": 2.0625,
      "step": 131499
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.201841950416565,
      "learning_rate": 0.00023387739872037384,
      "loss": 2.1163,
      "step": 131500
    },
    {
      "epoch": 1.71,
      "grad_norm": 0.9476701021194458,
      "learning_rate": 0.00023387340876102362,
      "loss": 2.1121,
      "step": 131501
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.246623158454895,
      "learning_rate": 0.00023386941881396766,
      "loss": 2.0952,
      "step": 131502
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.257161021232605,
      "learning_rate": 0.00023386542887920657,
      "loss": 2.036,
      "step": 131503
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.0179363489151,
      "learning_rate": 0.00023386143895674113,
      "loss": 2.0128,
      "step": 131504
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.0396311283111572,
      "learning_rate": 0.0002338574490465721,
      "loss": 2.0977,
      "step": 131505
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.052682876586914,
      "learning_rate": 0.00023385345914870033,
      "loss": 2.0347,
      "step": 131506
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.0337939262390137,
      "learning_rate": 0.00023384946926312633,
      "loss": 1.9334,
      "step": 131507
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.0466086864471436,
      "learning_rate": 0.00023384547938985105,
      "loss": 1.9581,
      "step": 131508
    },
    {
      "epoch": 1.71,
      "grad_norm": 0.9624704718589783,
      "learning_rate": 0.00023384148952887505,
      "loss": 2.2158,
      "step": 131509
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.0485293865203857,
      "learning_rate": 0.0002338374996801992,
      "loss": 2.2198,
      "step": 131510
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.0640199184417725,
      "learning_rate": 0.0002338335098438242,
      "loss": 2.0686,
      "step": 131511
    },
    {
      "epoch": 1.71,
      "grad_norm": 0.9337749481201172,
      "learning_rate": 0.0002338295200197508,
      "loss": 2.2142,
      "step": 131512
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.003641963005066,
      "learning_rate": 0.0002338255302079798,
      "loss": 2.0964,
      "step": 131513
    },
    {
      "epoch": 1.71,
      "grad_norm": 0.94815593957901,
      "learning_rate": 0.00023382154040851188,
      "loss": 2.3404,
      "step": 131514
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.1613733768463135,
      "learning_rate": 0.0002338175506213477,
      "loss": 2.18,
      "step": 131515
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.015397310256958,
      "learning_rate": 0.00023381356084648815,
      "loss": 2.098,
      "step": 131516
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.020129680633545,
      "learning_rate": 0.00023380957108393386,
      "loss": 2.4745,
      "step": 131517
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.130764126777649,
      "learning_rate": 0.00023380558133368563,
      "loss": 1.9784,
      "step": 131518
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.1945422887802124,
      "learning_rate": 0.0002338015915957442,
      "loss": 2.1412,
      "step": 131519
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.0158512592315674,
      "learning_rate": 0.00023379760187011028,
      "loss": 2.1165,
      "step": 131520
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.0126824378967285,
      "learning_rate": 0.00023379361215678475,
      "loss": 1.9663,
      "step": 131521
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.1197623014450073,
      "learning_rate": 0.00023378962245576817,
      "loss": 1.9183,
      "step": 131522
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.0495549440383911,
      "learning_rate": 0.00023378563276706132,
      "loss": 2.3187,
      "step": 131523
    },
    {
      "epoch": 1.71,
      "grad_norm": 0.9619708061218262,
      "learning_rate": 0.00023378164309066497,
      "loss": 1.9945,
      "step": 131524
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.1235610246658325,
      "learning_rate": 0.00023377765342657983,
      "loss": 2.0392,
      "step": 131525
    },
    {
      "epoch": 1.71,
      "grad_norm": 0.9887151122093201,
      "learning_rate": 0.0002337736637748067,
      "loss": 2.0337,
      "step": 131526
    },
    {
      "epoch": 1.71,
      "grad_norm": 0.9840130805969238,
      "learning_rate": 0.00023376967413534644,
      "loss": 2.2354,
      "step": 131527
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.0242911577224731,
      "learning_rate": 0.00023376568450819953,
      "loss": 2.2032,
      "step": 131528
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.0074729919433594,
      "learning_rate": 0.0002337616948933668,
      "loss": 2.1115,
      "step": 131529
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.143013596534729,
      "learning_rate": 0.00023375770529084904,
      "loss": 2.2405,
      "step": 131530
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.3537532091140747,
      "learning_rate": 0.00023375371570064695,
      "loss": 2.3058,
      "step": 131531
    },
    {
      "epoch": 1.71,
      "grad_norm": 0.978481650352478,
      "learning_rate": 0.00023374972612276134,
      "loss": 1.9589,
      "step": 131532
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.1185988187789917,
      "learning_rate": 0.000233745736557193,
      "loss": 2.1025,
      "step": 131533
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.0481371879577637,
      "learning_rate": 0.00023374174700394244,
      "loss": 2.136,
      "step": 131534
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.149835228919983,
      "learning_rate": 0.00023373775746301056,
      "loss": 2.2173,
      "step": 131535
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.1152324676513672,
      "learning_rate": 0.00023373376793439803,
      "loss": 2.0008,
      "step": 131536
    },
    {
      "epoch": 1.71,
      "grad_norm": 0.9745434522628784,
      "learning_rate": 0.00023372977841810573,
      "loss": 2.1543,
      "step": 131537
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.1420085430145264,
      "learning_rate": 0.00023372578891413427,
      "loss": 2.0589,
      "step": 131538
    },
    {
      "epoch": 1.71,
      "grad_norm": 0.9662932753562927,
      "learning_rate": 0.00023372179942248456,
      "loss": 2.3164,
      "step": 131539
    },
    {
      "epoch": 1.71,
      "grad_norm": 0.9526997208595276,
      "learning_rate": 0.00023371780994315707,
      "loss": 2.1132,
      "step": 131540
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.0912858247756958,
      "learning_rate": 0.00023371382047615273,
      "loss": 2.2725,
      "step": 131541
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.0535653829574585,
      "learning_rate": 0.00023370983102147225,
      "loss": 1.9189,
      "step": 131542
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.003623604774475,
      "learning_rate": 0.0002337058415791163,
      "loss": 1.9953,
      "step": 131543
    },
    {
      "epoch": 1.71,
      "grad_norm": 0.9512998461723328,
      "learning_rate": 0.00023370185214908576,
      "loss": 1.9478,
      "step": 131544
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.2694100141525269,
      "learning_rate": 0.0002336978627313814,
      "loss": 2.0444,
      "step": 131545
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.131696105003357,
      "learning_rate": 0.00023369387332600373,
      "loss": 2.173,
      "step": 131546
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.1340625286102295,
      "learning_rate": 0.0002336898839329536,
      "loss": 2.0055,
      "step": 131547
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.0205639600753784,
      "learning_rate": 0.00023368589455223182,
      "loss": 2.0953,
      "step": 131548
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.1581705808639526,
      "learning_rate": 0.000233681905183839,
      "loss": 2.1263,
      "step": 131549
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.042952060699463,
      "learning_rate": 0.00023367791582777606,
      "loss": 2.1744,
      "step": 131550
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.0396560430526733,
      "learning_rate": 0.00023367392648404372,
      "loss": 1.9096,
      "step": 131551
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.0686323642730713,
      "learning_rate": 0.00023366993715264253,
      "loss": 2.1096,
      "step": 131552
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.1441607475280762,
      "learning_rate": 0.00023366594783357335,
      "loss": 2.0532,
      "step": 131553
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.045509934425354,
      "learning_rate": 0.00023366195852683692,
      "loss": 2.2561,
      "step": 131554
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.091907262802124,
      "learning_rate": 0.000233657969232434,
      "loss": 2.1922,
      "step": 131555
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.0269505977630615,
      "learning_rate": 0.0002336539799503653,
      "loss": 2.2016,
      "step": 131556
    },
    {
      "epoch": 1.71,
      "grad_norm": 0.9692974090576172,
      "learning_rate": 0.0002336499906806317,
      "loss": 2.3178,
      "step": 131557
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.400049090385437,
      "learning_rate": 0.00023364600142323374,
      "loss": 2.142,
      "step": 131558
    },
    {
      "epoch": 1.71,
      "grad_norm": 0.9177305102348328,
      "learning_rate": 0.00023364201217817218,
      "loss": 2.2146,
      "step": 131559
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.0215991735458374,
      "learning_rate": 0.00023363802294544786,
      "loss": 2.1349,
      "step": 131560
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.1223362684249878,
      "learning_rate": 0.00023363403372506146,
      "loss": 2.2451,
      "step": 131561
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.0022661685943604,
      "learning_rate": 0.00023363004451701377,
      "loss": 2.2213,
      "step": 131562
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.0922516584396362,
      "learning_rate": 0.0002336260553213056,
      "loss": 2.2158,
      "step": 131563
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.0055882930755615,
      "learning_rate": 0.0002336220661379375,
      "loss": 2.1337,
      "step": 131564
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.002758502960205,
      "learning_rate": 0.00023361807696691033,
      "loss": 2.082,
      "step": 131565
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.033369541168213,
      "learning_rate": 0.0002336140878082248,
      "loss": 2.3242,
      "step": 131566
    },
    {
      "epoch": 1.71,
      "grad_norm": 0.9255306124687195,
      "learning_rate": 0.00023361009866188162,
      "loss": 2.0306,
      "step": 131567
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.1373845338821411,
      "learning_rate": 0.00023360610952788164,
      "loss": 2.1953,
      "step": 131568
    },
    {
      "epoch": 1.71,
      "grad_norm": 0.9739811420440674,
      "learning_rate": 0.00023360212040622564,
      "loss": 2.0095,
      "step": 131569
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.0099549293518066,
      "learning_rate": 0.0002335981312969141,
      "loss": 2.1139,
      "step": 131570
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.123570442199707,
      "learning_rate": 0.00023359414219994794,
      "loss": 2.21,
      "step": 131571
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.090358853340149,
      "learning_rate": 0.0002335901531153279,
      "loss": 2.1192,
      "step": 131572
    },
    {
      "epoch": 1.71,
      "grad_norm": 0.9727075695991516,
      "learning_rate": 0.00023358616404305468,
      "loss": 2.3713,
      "step": 131573
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.4436086416244507,
      "learning_rate": 0.00023358217498312906,
      "loss": 2.2882,
      "step": 131574
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.2139116525650024,
      "learning_rate": 0.00023357818593555183,
      "loss": 2.2759,
      "step": 131575
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.1709775924682617,
      "learning_rate": 0.00023357419690032367,
      "loss": 2.1386,
      "step": 131576
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.0800039768218994,
      "learning_rate": 0.00023357020787744527,
      "loss": 2.2949,
      "step": 131577
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.2612985372543335,
      "learning_rate": 0.0002335662188669174,
      "loss": 2.098,
      "step": 131578
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.3566372394561768,
      "learning_rate": 0.00023356222986874083,
      "loss": 2.1778,
      "step": 131579
    },
    {
      "epoch": 1.71,
      "grad_norm": 0.9787558317184448,
      "learning_rate": 0.0002335582408829163,
      "loss": 2.2463,
      "step": 131580
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.0131163597106934,
      "learning_rate": 0.0002335542519094446,
      "loss": 2.2547,
      "step": 131581
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.0967422723770142,
      "learning_rate": 0.00023355026294832635,
      "loss": 2.2594,
      "step": 131582
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.2656465768814087,
      "learning_rate": 0.00023354627399956238,
      "loss": 2.0687,
      "step": 131583
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.0403116941452026,
      "learning_rate": 0.00023354228506315347,
      "loss": 2.1353,
      "step": 131584
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.0274382829666138,
      "learning_rate": 0.00023353829613910023,
      "loss": 2.2401,
      "step": 131585
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.1620144844055176,
      "learning_rate": 0.0002335343072274035,
      "loss": 2.2207,
      "step": 131586
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.0067119598388672,
      "learning_rate": 0.000233530318328064,
      "loss": 2.1201,
      "step": 131587
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.0119680166244507,
      "learning_rate": 0.0002335263294410825,
      "loss": 2.4544,
      "step": 131588
    },
    {
      "epoch": 1.71,
      "grad_norm": 0.9812886118888855,
      "learning_rate": 0.00023352234056645964,
      "loss": 2.564,
      "step": 131589
    },
    {
      "epoch": 1.71,
      "grad_norm": 0.9847100973129272,
      "learning_rate": 0.00023351835170419625,
      "loss": 2.0341,
      "step": 131590
    },
    {
      "epoch": 1.71,
      "grad_norm": 0.9966972470283508,
      "learning_rate": 0.00023351436285429315,
      "loss": 2.1016,
      "step": 131591
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.0339434146881104,
      "learning_rate": 0.00023351037401675087,
      "loss": 2.127,
      "step": 131592
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.0326186418533325,
      "learning_rate": 0.0002335063851915703,
      "loss": 2.0824,
      "step": 131593
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.0461783409118652,
      "learning_rate": 0.00023350239637875218,
      "loss": 2.1308,
      "step": 131594
    },
    {
      "epoch": 1.71,
      "grad_norm": 0.9724953770637512,
      "learning_rate": 0.00023349840757829718,
      "loss": 2.0571,
      "step": 131595
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.038175344467163,
      "learning_rate": 0.00023349441879020607,
      "loss": 1.9046,
      "step": 131596
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.0174238681793213,
      "learning_rate": 0.0002334904300144796,
      "loss": 2.1301,
      "step": 131597
    },
    {
      "epoch": 1.71,
      "grad_norm": 0.8970708250999451,
      "learning_rate": 0.0002334864412511186,
      "loss": 2.0935,
      "step": 131598
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.0038350820541382,
      "learning_rate": 0.00023348245250012366,
      "loss": 1.9883,
      "step": 131599
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.2203654050827026,
      "learning_rate": 0.00023347846376149564,
      "loss": 2.1826,
      "step": 131600
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.1692031621932983,
      "learning_rate": 0.00023347447503523514,
      "loss": 1.9826,
      "step": 131601
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.0393893718719482,
      "learning_rate": 0.00023347048632134303,
      "loss": 1.937,
      "step": 131602
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.0700607299804688,
      "learning_rate": 0.00023346649761982002,
      "loss": 2.3388,
      "step": 131603
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.0921757221221924,
      "learning_rate": 0.0002334625089306668,
      "loss": 2.0589,
      "step": 131604
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.0285003185272217,
      "learning_rate": 0.00023345852025388424,
      "loss": 2.0662,
      "step": 131605
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.0473229885101318,
      "learning_rate": 0.000233454531589473,
      "loss": 2.1779,
      "step": 131606
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.3607096672058105,
      "learning_rate": 0.00023345054293743376,
      "loss": 2.1172,
      "step": 131607
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.201722502708435,
      "learning_rate": 0.0002334465542977673,
      "loss": 2.2333,
      "step": 131608
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.1861257553100586,
      "learning_rate": 0.00023344256567047443,
      "loss": 2.1629,
      "step": 131609
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.001413106918335,
      "learning_rate": 0.0002334385770555558,
      "loss": 2.2597,
      "step": 131610
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.3144510984420776,
      "learning_rate": 0.0002334345884530122,
      "loss": 2.2177,
      "step": 131611
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.2751376628875732,
      "learning_rate": 0.0002334305998628445,
      "loss": 2.3539,
      "step": 131612
    },
    {
      "epoch": 1.71,
      "grad_norm": 0.9415385127067566,
      "learning_rate": 0.0002334266112850532,
      "loss": 2.0754,
      "step": 131613
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.0432788133621216,
      "learning_rate": 0.00023342262271963912,
      "loss": 2.0506,
      "step": 131614
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.1258147954940796,
      "learning_rate": 0.00023341863416660304,
      "loss": 2.2006,
      "step": 131615
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.0760236978530884,
      "learning_rate": 0.00023341464562594568,
      "loss": 1.9429,
      "step": 131616
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.0382001399993896,
      "learning_rate": 0.00023341065709766784,
      "loss": 2.2456,
      "step": 131617
    },
    {
      "epoch": 1.71,
      "grad_norm": 0.9567494988441467,
      "learning_rate": 0.00023340666858177032,
      "loss": 2.181,
      "step": 131618
    },
    {
      "epoch": 1.71,
      "grad_norm": 0.9981701374053955,
      "learning_rate": 0.00023340268007825357,
      "loss": 2.0776,
      "step": 131619
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.1648247241973877,
      "learning_rate": 0.0002333986915871186,
      "loss": 2.136,
      "step": 131620
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.138580083847046,
      "learning_rate": 0.00023339470310836604,
      "loss": 2.0075,
      "step": 131621
    },
    {
      "epoch": 1.71,
      "grad_norm": 0.9244436025619507,
      "learning_rate": 0.00023339071464199666,
      "loss": 2.0847,
      "step": 131622
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.083519697189331,
      "learning_rate": 0.00023338672618801123,
      "loss": 2.0166,
      "step": 131623
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.3624069690704346,
      "learning_rate": 0.00023338273774641053,
      "loss": 2.0987,
      "step": 131624
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.0935426950454712,
      "learning_rate": 0.00023337874931719517,
      "loss": 2.2012,
      "step": 131625
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.0261014699935913,
      "learning_rate": 0.00023337476090036594,
      "loss": 2.1036,
      "step": 131626
    },
    {
      "epoch": 1.71,
      "grad_norm": 0.9755504131317139,
      "learning_rate": 0.0002333707724959236,
      "loss": 2.1631,
      "step": 131627
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.5214191675186157,
      "learning_rate": 0.00023336678410386887,
      "loss": 2.0465,
      "step": 131628
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.1895663738250732,
      "learning_rate": 0.00023336279572420255,
      "loss": 2.0539,
      "step": 131629
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.1475950479507446,
      "learning_rate": 0.00023335880735692543,
      "loss": 1.9994,
      "step": 131630
    },
    {
      "epoch": 1.71,
      "grad_norm": 0.9996461868286133,
      "learning_rate": 0.00023335481900203808,
      "loss": 2.1667,
      "step": 131631
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.6820048093795776,
      "learning_rate": 0.00023335083065954127,
      "loss": 2.2866,
      "step": 131632
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.1093374490737915,
      "learning_rate": 0.00023334684232943583,
      "loss": 2.4484,
      "step": 131633
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.0110399723052979,
      "learning_rate": 0.00023334285401172247,
      "loss": 2.2359,
      "step": 131634
    },
    {
      "epoch": 1.71,
      "grad_norm": 0.9429113268852234,
      "learning_rate": 0.00023333886570640194,
      "loss": 2.0953,
      "step": 131635
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.2456825971603394,
      "learning_rate": 0.0002333348774134751,
      "loss": 2.2945,
      "step": 131636
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.1487029790878296,
      "learning_rate": 0.00023333088913294247,
      "loss": 2.1039,
      "step": 131637
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.1182013750076294,
      "learning_rate": 0.00023332690086480483,
      "loss": 2.2582,
      "step": 131638
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.0874590873718262,
      "learning_rate": 0.000233322912609063,
      "loss": 2.2074,
      "step": 131639
    },
    {
      "epoch": 1.71,
      "grad_norm": 0.9506800174713135,
      "learning_rate": 0.0002333189243657177,
      "loss": 2.0561,
      "step": 131640
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.1425609588623047,
      "learning_rate": 0.00023331493613476965,
      "loss": 2.087,
      "step": 131641
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.1969956159591675,
      "learning_rate": 0.00023331094791621978,
      "loss": 2.1127,
      "step": 131642
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.0758811235427856,
      "learning_rate": 0.00023330695971006851,
      "loss": 2.2071,
      "step": 131643
    },
    {
      "epoch": 1.71,
      "grad_norm": 0.9858917593955994,
      "learning_rate": 0.00023330297151631673,
      "loss": 2.3206,
      "step": 131644
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.2886193990707397,
      "learning_rate": 0.00023329898333496521,
      "loss": 2.2183,
      "step": 131645
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.1380822658538818,
      "learning_rate": 0.00023329499516601466,
      "loss": 2.0646,
      "step": 131646
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.36659574508667,
      "learning_rate": 0.00023329100700946582,
      "loss": 1.9648,
      "step": 131647
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.0029562711715698,
      "learning_rate": 0.00023328701886531956,
      "loss": 2.0944,
      "step": 131648
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.1560821533203125,
      "learning_rate": 0.00023328303073357636,
      "loss": 2.1097,
      "step": 131649
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.3162153959274292,
      "learning_rate": 0.00023327904261423717,
      "loss": 2.3166,
      "step": 131650
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.0402517318725586,
      "learning_rate": 0.0002332750545073026,
      "loss": 2.198,
      "step": 131651
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.0586419105529785,
      "learning_rate": 0.00023327106641277348,
      "loss": 1.9921,
      "step": 131652
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.1549792289733887,
      "learning_rate": 0.00023326707833065056,
      "loss": 2.266,
      "step": 131653
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.247028112411499,
      "learning_rate": 0.00023326309026093452,
      "loss": 2.2357,
      "step": 131654
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.0651888847351074,
      "learning_rate": 0.00023325910220362627,
      "loss": 2.0995,
      "step": 131655
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.2464148998260498,
      "learning_rate": 0.00023325511415872623,
      "loss": 2.0218,
      "step": 131656
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.2599539756774902,
      "learning_rate": 0.0002332511261262354,
      "loss": 1.8434,
      "step": 131657
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.2218451499938965,
      "learning_rate": 0.00023324713810615443,
      "loss": 2.2147,
      "step": 131658
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.0025882720947266,
      "learning_rate": 0.00023324315009848405,
      "loss": 2.0663,
      "step": 131659
    },
    {
      "epoch": 1.71,
      "grad_norm": 0.9483931064605713,
      "learning_rate": 0.00023323916210322504,
      "loss": 2.0388,
      "step": 131660
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.1355774402618408,
      "learning_rate": 0.00023323517412037826,
      "loss": 2.1073,
      "step": 131661
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.1920535564422607,
      "learning_rate": 0.00023323118614994418,
      "loss": 2.1217,
      "step": 131662
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.1922868490219116,
      "learning_rate": 0.0002332271981919237,
      "loss": 2.1071,
      "step": 131663
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.0890535116195679,
      "learning_rate": 0.00023322321024631755,
      "loss": 1.7273,
      "step": 131664
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.1236547231674194,
      "learning_rate": 0.00023321922231312644,
      "loss": 2.0265,
      "step": 131665
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.0698028802871704,
      "learning_rate": 0.00023321523439235117,
      "loss": 2.2751,
      "step": 131666
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.0025990009307861,
      "learning_rate": 0.0002332112464839925,
      "loss": 2.1433,
      "step": 131667
    },
    {
      "epoch": 1.71,
      "grad_norm": 0.9964052438735962,
      "learning_rate": 0.0002332072585880511,
      "loss": 2.0653,
      "step": 131668
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.0919818878173828,
      "learning_rate": 0.0002332032707045277,
      "loss": 1.8694,
      "step": 131669
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.0870535373687744,
      "learning_rate": 0.000233199282833423,
      "loss": 1.8837,
      "step": 131670
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.0485671758651733,
      "learning_rate": 0.00023319529497473792,
      "loss": 2.2404,
      "step": 131671
    },
    {
      "epoch": 1.71,
      "grad_norm": 0.9901599884033203,
      "learning_rate": 0.00023319130712847301,
      "loss": 2.1953,
      "step": 131672
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.0576239824295044,
      "learning_rate": 0.0002331873192946292,
      "loss": 2.105,
      "step": 131673
    },
    {
      "epoch": 1.71,
      "grad_norm": 0.9976375102996826,
      "learning_rate": 0.00023318333147320707,
      "loss": 1.9999,
      "step": 131674
    },
    {
      "epoch": 1.71,
      "grad_norm": 0.9959952235221863,
      "learning_rate": 0.0002331793436642075,
      "loss": 2.0816,
      "step": 131675
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.0725996494293213,
      "learning_rate": 0.00023317535586763103,
      "loss": 2.0261,
      "step": 131676
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.3309227228164673,
      "learning_rate": 0.00023317136808347858,
      "loss": 2.1872,
      "step": 131677
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.094128131866455,
      "learning_rate": 0.00023316738031175079,
      "loss": 2.2805,
      "step": 131678
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.076368808746338,
      "learning_rate": 0.00023316339255244852,
      "loss": 2.0068,
      "step": 131679
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.0162886381149292,
      "learning_rate": 0.0002331594048055724,
      "loss": 2.3122,
      "step": 131680
    },
    {
      "epoch": 1.71,
      "grad_norm": 0.9574413895606995,
      "learning_rate": 0.0002331554170711232,
      "loss": 2.2194,
      "step": 131681
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.1180319786071777,
      "learning_rate": 0.00023315142934910162,
      "loss": 2.1623,
      "step": 131682
    },
    {
      "epoch": 1.71,
      "grad_norm": 0.9681649208068848,
      "learning_rate": 0.00023314744163950856,
      "loss": 2.3251,
      "step": 131683
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.09786057472229,
      "learning_rate": 0.00023314345394234462,
      "loss": 2.1179,
      "step": 131684
    },
    {
      "epoch": 1.71,
      "grad_norm": 0.9983405470848083,
      "learning_rate": 0.00023313946625761058,
      "loss": 2.0135,
      "step": 131685
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.2319422960281372,
      "learning_rate": 0.00023313547858530714,
      "loss": 2.0976,
      "step": 131686
    },
    {
      "epoch": 1.71,
      "grad_norm": 0.9978843927383423,
      "learning_rate": 0.00023313149092543507,
      "loss": 1.8821,
      "step": 131687
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.062820315361023,
      "learning_rate": 0.00023312750327799512,
      "loss": 2.1034,
      "step": 131688
    },
    {
      "epoch": 1.71,
      "grad_norm": 0.9639859199523926,
      "learning_rate": 0.000233123515642988,
      "loss": 2.337,
      "step": 131689
    },
    {
      "epoch": 1.71,
      "grad_norm": 0.9978565573692322,
      "learning_rate": 0.0002331195280204146,
      "loss": 2.2774,
      "step": 131690
    },
    {
      "epoch": 1.71,
      "grad_norm": 0.975528359413147,
      "learning_rate": 0.00023311554041027552,
      "loss": 2.0084,
      "step": 131691
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.0490238666534424,
      "learning_rate": 0.00023311155281257145,
      "loss": 2.2039,
      "step": 131692
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.0739458799362183,
      "learning_rate": 0.0002331075652273032,
      "loss": 2.4521,
      "step": 131693
    },
    {
      "epoch": 1.71,
      "grad_norm": 0.9679041504859924,
      "learning_rate": 0.0002331035776544715,
      "loss": 2.1132,
      "step": 131694
    },
    {
      "epoch": 1.71,
      "grad_norm": 0.9703369736671448,
      "learning_rate": 0.00023309959009407716,
      "loss": 2.3033,
      "step": 131695
    },
    {
      "epoch": 1.71,
      "grad_norm": 0.9228987693786621,
      "learning_rate": 0.00023309560254612083,
      "loss": 1.982,
      "step": 131696
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.0244948863983154,
      "learning_rate": 0.00023309161501060345,
      "loss": 2.1874,
      "step": 131697
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.1620627641677856,
      "learning_rate": 0.00023308762748752544,
      "loss": 2.2431,
      "step": 131698
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.0168894529342651,
      "learning_rate": 0.00023308363997688773,
      "loss": 2.1894,
      "step": 131699
    },
    {
      "epoch": 1.71,
      "grad_norm": 0.9507639408111572,
      "learning_rate": 0.000233079652478691,
      "loss": 2.1037,
      "step": 131700
    },
    {
      "epoch": 1.71,
      "grad_norm": 0.9193603992462158,
      "learning_rate": 0.00023307566499293603,
      "loss": 2.3187,
      "step": 131701
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.0019837617874146,
      "learning_rate": 0.0002330716775196236,
      "loss": 2.1225,
      "step": 131702
    },
    {
      "epoch": 1.71,
      "grad_norm": 0.9934775829315186,
      "learning_rate": 0.00023306769005875452,
      "loss": 1.842,
      "step": 131703
    },
    {
      "epoch": 1.71,
      "grad_norm": 0.9982231855392456,
      "learning_rate": 0.00023306370261032925,
      "loss": 2.0301,
      "step": 131704
    },
    {
      "epoch": 1.71,
      "grad_norm": 0.9784271121025085,
      "learning_rate": 0.00023305971517434876,
      "loss": 1.9881,
      "step": 131705
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.0299303531646729,
      "learning_rate": 0.0002330557277508137,
      "loss": 2.4586,
      "step": 131706
    },
    {
      "epoch": 1.71,
      "grad_norm": 1.000382900238037,
      "learning_rate": 0.00023305174033972482,
      "loss": 2.0727,
      "step": 131707
    },
    {
      "epoch": 1.71,
      "grad_norm": 0.9909508228302002,
      "learning_rate": 0.00023304775294108294,
      "loss": 2.1445,
      "step": 131708
    },
    {
      "epoch": 1.71,
      "grad_norm": 0.969455897808075,
      "learning_rate": 0.00023304376555488885,
      "loss": 1.9807,
      "step": 131709
    },
    {
      "epoch": 1.71,
      "grad_norm": 0.9722198843955994,
      "learning_rate": 0.00023303977818114308,
      "loss": 1.9096,
      "step": 131710
    },
    {
      "epoch": 1.71,
      "grad_norm": 0.9259265661239624,
      "learning_rate": 0.00023303579081984644,
      "loss": 2.164,
      "step": 131711
    },
    {
      "epoch": 1.71,
      "grad_norm": 0.9707654118537903,
      "learning_rate": 0.00023303180347099975,
      "loss": 1.799,
      "step": 131712
    },
    {
      "epoch": 1.71,
      "grad_norm": 0.9866690039634705,
      "learning_rate": 0.00023302781613460368,
      "loss": 2.392,
      "step": 131713
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.0420784950256348,
      "learning_rate": 0.00023302382881065903,
      "loss": 2.0875,
      "step": 131714
    },
    {
      "epoch": 1.72,
      "grad_norm": 0.9613475799560547,
      "learning_rate": 0.0002330198414991666,
      "loss": 2.1166,
      "step": 131715
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.0891038179397583,
      "learning_rate": 0.00023301585420012696,
      "loss": 2.2732,
      "step": 131716
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.0132278203964233,
      "learning_rate": 0.00023301186691354094,
      "loss": 2.1712,
      "step": 131717
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.012012243270874,
      "learning_rate": 0.00023300787963940922,
      "loss": 2.2713,
      "step": 131718
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.0688531398773193,
      "learning_rate": 0.00023300389237773263,
      "loss": 1.9638,
      "step": 131719
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.2126063108444214,
      "learning_rate": 0.0002329999051285119,
      "loss": 2.1511,
      "step": 131720
    },
    {
      "epoch": 1.72,
      "grad_norm": 0.9622233510017395,
      "learning_rate": 0.00023299591789174774,
      "loss": 1.988,
      "step": 131721
    },
    {
      "epoch": 1.72,
      "grad_norm": 0.9423807859420776,
      "learning_rate": 0.00023299193066744105,
      "loss": 2.0506,
      "step": 131722
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.011722445487976,
      "learning_rate": 0.00023298794345559225,
      "loss": 2.0682,
      "step": 131723
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.0035667419433594,
      "learning_rate": 0.00023298395625620232,
      "loss": 2.0453,
      "step": 131724
    },
    {
      "epoch": 1.72,
      "grad_norm": 0.9719438552856445,
      "learning_rate": 0.00023297996906927192,
      "loss": 2.201,
      "step": 131725
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.126282811164856,
      "learning_rate": 0.00023297598189480178,
      "loss": 2.0653,
      "step": 131726
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.051450490951538,
      "learning_rate": 0.00023297199473279267,
      "loss": 2.051,
      "step": 131727
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.171142339706421,
      "learning_rate": 0.00023296800758324548,
      "loss": 2.2898,
      "step": 131728
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.1564713716506958,
      "learning_rate": 0.0002329640204461607,
      "loss": 2.2733,
      "step": 131729
    },
    {
      "epoch": 1.72,
      "grad_norm": 0.9781832695007324,
      "learning_rate": 0.00023296003332153912,
      "loss": 1.9871,
      "step": 131730
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.1174941062927246,
      "learning_rate": 0.00023295604620938157,
      "loss": 2.2199,
      "step": 131731
    },
    {
      "epoch": 1.72,
      "grad_norm": 0.9318400025367737,
      "learning_rate": 0.00023295205910968872,
      "loss": 1.9806,
      "step": 131732
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.0638864040374756,
      "learning_rate": 0.0002329480720224614,
      "loss": 2.0752,
      "step": 131733
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.0594841241836548,
      "learning_rate": 0.00023294408494770042,
      "loss": 2.0199,
      "step": 131734
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.0290939807891846,
      "learning_rate": 0.00023294009788540622,
      "loss": 2.1141,
      "step": 131735
    },
    {
      "epoch": 1.72,
      "grad_norm": 0.9811779260635376,
      "learning_rate": 0.00023293611083557975,
      "loss": 2.2313,
      "step": 131736
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.0211105346679688,
      "learning_rate": 0.0002329321237982217,
      "loss": 2.2775,
      "step": 131737
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.0293596982955933,
      "learning_rate": 0.0002329281367733329,
      "loss": 2.2373,
      "step": 131738
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.0446739196777344,
      "learning_rate": 0.00023292414976091398,
      "loss": 2.1164,
      "step": 131739
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.1425626277923584,
      "learning_rate": 0.00023292016276096585,
      "loss": 2.1905,
      "step": 131740
    },
    {
      "epoch": 1.72,
      "grad_norm": 0.9980247616767883,
      "learning_rate": 0.000232916175773489,
      "loss": 2.1005,
      "step": 131741
    },
    {
      "epoch": 1.72,
      "grad_norm": 0.9954472184181213,
      "learning_rate": 0.00023291218879848433,
      "loss": 1.8773,
      "step": 131742
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.0368762016296387,
      "learning_rate": 0.0002329082018359525,
      "loss": 1.971,
      "step": 131743
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.0256067514419556,
      "learning_rate": 0.00023290421488589435,
      "loss": 2.1141,
      "step": 131744
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.0429686307907104,
      "learning_rate": 0.00023290022794831056,
      "loss": 2.2975,
      "step": 131745
    },
    {
      "epoch": 1.72,
      "grad_norm": 0.9769486784934998,
      "learning_rate": 0.000232896241023202,
      "loss": 1.8009,
      "step": 131746
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.0151028633117676,
      "learning_rate": 0.00023289225411056918,
      "loss": 2.0526,
      "step": 131747
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.0316669940948486,
      "learning_rate": 0.00023288826721041297,
      "loss": 2.0199,
      "step": 131748
    },
    {
      "epoch": 1.72,
      "grad_norm": 0.9335933923721313,
      "learning_rate": 0.00023288428032273406,
      "loss": 2.0724,
      "step": 131749
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.0468108654022217,
      "learning_rate": 0.00023288029344753326,
      "loss": 2.0858,
      "step": 131750
    },
    {
      "epoch": 1.72,
      "grad_norm": 0.9700280427932739,
      "learning_rate": 0.0002328763065848113,
      "loss": 2.1042,
      "step": 131751
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.0536763668060303,
      "learning_rate": 0.00023287231973456892,
      "loss": 2.2894,
      "step": 131752
    },
    {
      "epoch": 1.72,
      "grad_norm": 0.9116175174713135,
      "learning_rate": 0.00023286833289680686,
      "loss": 1.7089,
      "step": 131753
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.0774706602096558,
      "learning_rate": 0.00023286434607152576,
      "loss": 2.1193,
      "step": 131754
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.0906363725662231,
      "learning_rate": 0.0002328603592587265,
      "loss": 2.3189,
      "step": 131755
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.0079830884933472,
      "learning_rate": 0.0002328563724584097,
      "loss": 2.044,
      "step": 131756
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.0462043285369873,
      "learning_rate": 0.0002328523856705762,
      "loss": 2.2473,
      "step": 131757
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.021708369255066,
      "learning_rate": 0.00023284839889522676,
      "loss": 2.2103,
      "step": 131758
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.051718831062317,
      "learning_rate": 0.00023284441213236202,
      "loss": 2.3534,
      "step": 131759
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.0307406187057495,
      "learning_rate": 0.0002328404253819828,
      "loss": 2.1618,
      "step": 131760
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.17740797996521,
      "learning_rate": 0.00023283643864408982,
      "loss": 2.048,
      "step": 131761
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.0316988229751587,
      "learning_rate": 0.00023283245191868377,
      "loss": 2.2553,
      "step": 131762
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.005617380142212,
      "learning_rate": 0.00023282846520576546,
      "loss": 2.1348,
      "step": 131763
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.0911153554916382,
      "learning_rate": 0.00023282447850533563,
      "loss": 2.1944,
      "step": 131764
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.194428563117981,
      "learning_rate": 0.00023282049181739495,
      "loss": 1.9334,
      "step": 131765
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.000551700592041,
      "learning_rate": 0.00023281650514194423,
      "loss": 2.1323,
      "step": 131766
    },
    {
      "epoch": 1.72,
      "grad_norm": 0.9794946908950806,
      "learning_rate": 0.0002328125184789842,
      "loss": 2.1908,
      "step": 131767
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.0518933534622192,
      "learning_rate": 0.0002328085318285156,
      "loss": 1.8549,
      "step": 131768
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.0676491260528564,
      "learning_rate": 0.00023280454519053912,
      "loss": 2.0543,
      "step": 131769
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.050114631652832,
      "learning_rate": 0.00023280055856505558,
      "loss": 2.3012,
      "step": 131770
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.0511900186538696,
      "learning_rate": 0.00023279657195206566,
      "loss": 2.0835,
      "step": 131771
    },
    {
      "epoch": 1.72,
      "grad_norm": 0.9868552088737488,
      "learning_rate": 0.00023279258535157014,
      "loss": 2.1276,
      "step": 131772
    },
    {
      "epoch": 1.72,
      "grad_norm": 0.9538171291351318,
      "learning_rate": 0.00023278859876356973,
      "loss": 2.1352,
      "step": 131773
    },
    {
      "epoch": 1.72,
      "grad_norm": 0.999422013759613,
      "learning_rate": 0.0002327846121880652,
      "loss": 2.1868,
      "step": 131774
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.030905842781067,
      "learning_rate": 0.00023278062562505733,
      "loss": 2.0492,
      "step": 131775
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.0602953433990479,
      "learning_rate": 0.00023277663907454678,
      "loss": 1.9545,
      "step": 131776
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.0639504194259644,
      "learning_rate": 0.00023277265253653431,
      "loss": 2.249,
      "step": 131777
    },
    {
      "epoch": 1.72,
      "grad_norm": 0.9717675447463989,
      "learning_rate": 0.00023276866601102064,
      "loss": 1.972,
      "step": 131778
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.0337159633636475,
      "learning_rate": 0.00023276467949800655,
      "loss": 1.8565,
      "step": 131779
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.1928904056549072,
      "learning_rate": 0.0002327606929974928,
      "loss": 2.1681,
      "step": 131780
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.052077293395996,
      "learning_rate": 0.00023275670650948008,
      "loss": 2.2135,
      "step": 131781
    },
    {
      "epoch": 1.72,
      "grad_norm": 0.9776877164840698,
      "learning_rate": 0.0002327527200339693,
      "loss": 2.0319,
      "step": 131782
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.1008859872817993,
      "learning_rate": 0.00023274873357096088,
      "loss": 1.9798,
      "step": 131783
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.0125497579574585,
      "learning_rate": 0.00023274474712045579,
      "loss": 2.1126,
      "step": 131784
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.0763349533081055,
      "learning_rate": 0.0002327407606824547,
      "loss": 2.2235,
      "step": 131785
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.0487078428268433,
      "learning_rate": 0.00023273677425695842,
      "loss": 2.0962,
      "step": 131786
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.1551613807678223,
      "learning_rate": 0.00023273278784396762,
      "loss": 1.9015,
      "step": 131787
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.0507138967514038,
      "learning_rate": 0.00023272880144348303,
      "loss": 2.0573,
      "step": 131788
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.0698542594909668,
      "learning_rate": 0.00023272481505550558,
      "loss": 2.2471,
      "step": 131789
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.0703198909759521,
      "learning_rate": 0.00023272082868003573,
      "loss": 2.1087,
      "step": 131790
    },
    {
      "epoch": 1.72,
      "grad_norm": 0.9186506271362305,
      "learning_rate": 0.00023271684231707432,
      "loss": 2.0866,
      "step": 131791
    },
    {
      "epoch": 1.72,
      "grad_norm": 0.9764357209205627,
      "learning_rate": 0.00023271285596662217,
      "loss": 2.3283,
      "step": 131792
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.0183625221252441,
      "learning_rate": 0.00023270886962867992,
      "loss": 2.0535,
      "step": 131793
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.0133440494537354,
      "learning_rate": 0.0002327048833032484,
      "loss": 2.3688,
      "step": 131794
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.0003525018692017,
      "learning_rate": 0.00023270089699032841,
      "loss": 2.2924,
      "step": 131795
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.026291012763977,
      "learning_rate": 0.00023269691068992048,
      "loss": 2.1186,
      "step": 131796
    },
    {
      "epoch": 1.72,
      "grad_norm": 0.9864795207977295,
      "learning_rate": 0.00023269292440202547,
      "loss": 2.041,
      "step": 131797
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.0633704662322998,
      "learning_rate": 0.00023268893812664412,
      "loss": 2.224,
      "step": 131798
    },
    {
      "epoch": 1.72,
      "grad_norm": 0.9784804582595825,
      "learning_rate": 0.00023268495186377715,
      "loss": 2.1715,
      "step": 131799
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.050038456916809,
      "learning_rate": 0.00023268096561342532,
      "loss": 2.0971,
      "step": 131800
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.0256377458572388,
      "learning_rate": 0.0002326769793755895,
      "loss": 2.1786,
      "step": 131801
    },
    {
      "epoch": 1.72,
      "grad_norm": 0.9692030549049377,
      "learning_rate": 0.00023267299315027018,
      "loss": 2.1812,
      "step": 131802
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.1880006790161133,
      "learning_rate": 0.00023266900693746822,
      "loss": 2.1961,
      "step": 131803
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.126379370689392,
      "learning_rate": 0.00023266502073718436,
      "loss": 2.1841,
      "step": 131804
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.0074918270111084,
      "learning_rate": 0.00023266103454941934,
      "loss": 2.2177,
      "step": 131805
    },
    {
      "epoch": 1.72,
      "grad_norm": 0.9457234144210815,
      "learning_rate": 0.0002326570483741739,
      "loss": 2.3043,
      "step": 131806
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.1201444864273071,
      "learning_rate": 0.00023265306221144895,
      "loss": 2.0698,
      "step": 131807
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.083613634109497,
      "learning_rate": 0.0002326490760612449,
      "loss": 2.2049,
      "step": 131808
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.1671067476272583,
      "learning_rate": 0.00023264508992356268,
      "loss": 2.0342,
      "step": 131809
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.0895251035690308,
      "learning_rate": 0.000232641103798403,
      "loss": 2.2492,
      "step": 131810
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.0738123655319214,
      "learning_rate": 0.00023263711768576657,
      "loss": 2.299,
      "step": 131811
    },
    {
      "epoch": 1.72,
      "grad_norm": 0.9729368090629578,
      "learning_rate": 0.00023263313158565423,
      "loss": 2.2167,
      "step": 131812
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.108292818069458,
      "learning_rate": 0.00023262914549806676,
      "loss": 2.071,
      "step": 131813
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.051104187965393,
      "learning_rate": 0.0002326251594230047,
      "loss": 2.1971,
      "step": 131814
    },
    {
      "epoch": 1.72,
      "grad_norm": 0.9254482388496399,
      "learning_rate": 0.00023262117336046885,
      "loss": 2.2098,
      "step": 131815
    },
    {
      "epoch": 1.72,
      "grad_norm": 0.9360504746437073,
      "learning_rate": 0.00023261718731046007,
      "loss": 1.996,
      "step": 131816
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.0099701881408691,
      "learning_rate": 0.00023261320127297897,
      "loss": 2.0436,
      "step": 131817
    },
    {
      "epoch": 1.72,
      "grad_norm": 0.9369488954544067,
      "learning_rate": 0.00023260921524802635,
      "loss": 1.9676,
      "step": 131818
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.0399985313415527,
      "learning_rate": 0.0002326052292356031,
      "loss": 1.9525,
      "step": 131819
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.1003974676132202,
      "learning_rate": 0.00023260124323570965,
      "loss": 2.0777,
      "step": 131820
    },
    {
      "epoch": 1.72,
      "grad_norm": 0.9932062029838562,
      "learning_rate": 0.00023259725724834694,
      "loss": 1.913,
      "step": 131821
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.1088073253631592,
      "learning_rate": 0.00023259327127351565,
      "loss": 1.9912,
      "step": 131822
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.0106382369995117,
      "learning_rate": 0.00023258928531121655,
      "loss": 2.2142,
      "step": 131823
    },
    {
      "epoch": 1.72,
      "grad_norm": 0.9487406611442566,
      "learning_rate": 0.00023258529936145037,
      "loss": 2.1026,
      "step": 131824
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.0357123613357544,
      "learning_rate": 0.000232581313424218,
      "loss": 2.2841,
      "step": 131825
    },
    {
      "epoch": 1.72,
      "grad_norm": 0.950731098651886,
      "learning_rate": 0.00023257732749951986,
      "loss": 2.3134,
      "step": 131826
    },
    {
      "epoch": 1.72,
      "grad_norm": 0.9981765151023865,
      "learning_rate": 0.0002325733415873569,
      "loss": 1.907,
      "step": 131827
    },
    {
      "epoch": 1.72,
      "grad_norm": 0.930042028427124,
      "learning_rate": 0.00023256935568772985,
      "loss": 2.2602,
      "step": 131828
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.0962440967559814,
      "learning_rate": 0.0002325653698006394,
      "loss": 2.1258,
      "step": 131829
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.089539885520935,
      "learning_rate": 0.00023256138392608634,
      "loss": 2.3299,
      "step": 131830
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.0190086364746094,
      "learning_rate": 0.0002325573980640715,
      "loss": 2.2365,
      "step": 131831
    },
    {
      "epoch": 1.72,
      "grad_norm": 0.9941176176071167,
      "learning_rate": 0.0002325534122145954,
      "loss": 2.2047,
      "step": 131832
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.0278806686401367,
      "learning_rate": 0.0002325494263776589,
      "loss": 1.9701,
      "step": 131833
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.0049448013305664,
      "learning_rate": 0.00023254544055326272,
      "loss": 2.1961,
      "step": 131834
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.099914789199829,
      "learning_rate": 0.00023254145474140762,
      "loss": 2.104,
      "step": 131835
    },
    {
      "epoch": 1.72,
      "grad_norm": 0.9997894763946533,
      "learning_rate": 0.00023253746894209433,
      "loss": 1.9841,
      "step": 131836
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.0574358701705933,
      "learning_rate": 0.0002325334831553237,
      "loss": 2.2398,
      "step": 131837
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.0565881729125977,
      "learning_rate": 0.00023252949738109628,
      "loss": 2.2185,
      "step": 131838
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.0264309644699097,
      "learning_rate": 0.00023252551161941288,
      "loss": 2.2251,
      "step": 131839
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.0889561176300049,
      "learning_rate": 0.0002325215258702743,
      "loss": 2.1327,
      "step": 131840
    },
    {
      "epoch": 1.72,
      "grad_norm": 0.9121502041816711,
      "learning_rate": 0.0002325175401336812,
      "loss": 2.0453,
      "step": 131841
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.2003875970840454,
      "learning_rate": 0.00023251355440963436,
      "loss": 2.1648,
      "step": 131842
    },
    {
      "epoch": 1.72,
      "grad_norm": 0.9594862461090088,
      "learning_rate": 0.00023250956869813464,
      "loss": 2.2523,
      "step": 131843
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.0179500579833984,
      "learning_rate": 0.00023250558299918258,
      "loss": 2.1322,
      "step": 131844
    },
    {
      "epoch": 1.72,
      "grad_norm": 0.9847810864448547,
      "learning_rate": 0.00023250159731277902,
      "loss": 2.0956,
      "step": 131845
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.051127552986145,
      "learning_rate": 0.00023249761163892467,
      "loss": 2.026,
      "step": 131846
    },
    {
      "epoch": 1.72,
      "grad_norm": 0.998075544834137,
      "learning_rate": 0.00023249362597762028,
      "loss": 2.2188,
      "step": 131847
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.0136486291885376,
      "learning_rate": 0.00023248964032886662,
      "loss": 1.9852,
      "step": 131848
    },
    {
      "epoch": 1.72,
      "grad_norm": 0.9399881958961487,
      "learning_rate": 0.00023248565469266444,
      "loss": 2.1143,
      "step": 131849
    },
    {
      "epoch": 1.72,
      "grad_norm": 0.9786919951438904,
      "learning_rate": 0.00023248166906901442,
      "loss": 2.1347,
      "step": 131850
    },
    {
      "epoch": 1.72,
      "grad_norm": 0.9549139142036438,
      "learning_rate": 0.0002324776834579173,
      "loss": 2.1682,
      "step": 131851
    },
    {
      "epoch": 1.72,
      "grad_norm": 0.9506949782371521,
      "learning_rate": 0.00023247369785937393,
      "loss": 1.9061,
      "step": 131852
    },
    {
      "epoch": 1.72,
      "grad_norm": 0.9520578384399414,
      "learning_rate": 0.0002324697122733849,
      "loss": 2.0029,
      "step": 131853
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.1095399856567383,
      "learning_rate": 0.00023246572669995103,
      "loss": 2.1356,
      "step": 131854
    },
    {
      "epoch": 1.72,
      "grad_norm": 0.9876486659049988,
      "learning_rate": 0.00023246174113907307,
      "loss": 2.3034,
      "step": 131855
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.0088567733764648,
      "learning_rate": 0.0002324577555907518,
      "loss": 2.0683,
      "step": 131856
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.103800892829895,
      "learning_rate": 0.00023245377005498785,
      "loss": 2.2049,
      "step": 131857
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.2011078596115112,
      "learning_rate": 0.00023244978453178204,
      "loss": 1.9663,
      "step": 131858
    },
    {
      "epoch": 1.72,
      "grad_norm": 0.9556092023849487,
      "learning_rate": 0.0002324457990211351,
      "loss": 1.8772,
      "step": 131859
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.0656107664108276,
      "learning_rate": 0.00023244181352304773,
      "loss": 2.2876,
      "step": 131860
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.0799294710159302,
      "learning_rate": 0.00023243782803752068,
      "loss": 2.3857,
      "step": 131861
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.0290499925613403,
      "learning_rate": 0.00023243384256455482,
      "loss": 2.2034,
      "step": 131862
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.1076858043670654,
      "learning_rate": 0.00023242985710415068,
      "loss": 2.0756,
      "step": 131863
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.1515189409255981,
      "learning_rate": 0.00023242587165630911,
      "loss": 2.0752,
      "step": 131864
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.0284135341644287,
      "learning_rate": 0.00023242188622103085,
      "loss": 2.1092,
      "step": 131865
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.0283337831497192,
      "learning_rate": 0.00023241790079831675,
      "loss": 1.9509,
      "step": 131866
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.1294066905975342,
      "learning_rate": 0.0002324139153881673,
      "loss": 2.1488,
      "step": 131867
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.0180779695510864,
      "learning_rate": 0.0002324099299905835,
      "loss": 2.0834,
      "step": 131868
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.0998132228851318,
      "learning_rate": 0.0002324059446055659,
      "loss": 2.0978,
      "step": 131869
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.27231764793396,
      "learning_rate": 0.00023240195923311527,
      "loss": 2.053,
      "step": 131870
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.3720084428787231,
      "learning_rate": 0.0002323979738732324,
      "loss": 1.9494,
      "step": 131871
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.0542380809783936,
      "learning_rate": 0.00023239398852591803,
      "loss": 2.2605,
      "step": 131872
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.3723212480545044,
      "learning_rate": 0.00023239000319117293,
      "loss": 1.95,
      "step": 131873
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.2799432277679443,
      "learning_rate": 0.00023238601786899787,
      "loss": 2.2281,
      "step": 131874
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.1195614337921143,
      "learning_rate": 0.00023238203255939347,
      "loss": 2.1727,
      "step": 131875
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.0993130207061768,
      "learning_rate": 0.00023237804726236046,
      "loss": 1.9838,
      "step": 131876
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.493595004081726,
      "learning_rate": 0.00023237406197789968,
      "loss": 1.9832,
      "step": 131877
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.0443785190582275,
      "learning_rate": 0.00023237007670601185,
      "loss": 2.0427,
      "step": 131878
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.120708703994751,
      "learning_rate": 0.0002323660914466977,
      "loss": 2.0754,
      "step": 131879
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.1490167379379272,
      "learning_rate": 0.00023236210619995804,
      "loss": 1.9985,
      "step": 131880
    },
    {
      "epoch": 1.72,
      "grad_norm": 0.9674667716026306,
      "learning_rate": 0.00023235812096579348,
      "loss": 2.0841,
      "step": 131881
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.3994687795639038,
      "learning_rate": 0.00023235413574420477,
      "loss": 1.9512,
      "step": 131882
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.0090464353561401,
      "learning_rate": 0.00023235015053519275,
      "loss": 2.159,
      "step": 131883
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.0793806314468384,
      "learning_rate": 0.0002323461653387581,
      "loss": 2.0373,
      "step": 131884
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.0489592552185059,
      "learning_rate": 0.00023234218015490154,
      "loss": 2.1663,
      "step": 131885
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.0331202745437622,
      "learning_rate": 0.000232338194983624,
      "loss": 1.964,
      "step": 131886
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.1733735799789429,
      "learning_rate": 0.00023233420982492596,
      "loss": 2.0804,
      "step": 131887
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.0028483867645264,
      "learning_rate": 0.00023233022467880827,
      "loss": 2.3631,
      "step": 131888
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.0168571472167969,
      "learning_rate": 0.00023232623954527162,
      "loss": 2.0701,
      "step": 131889
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.0792438983917236,
      "learning_rate": 0.00023232225442431682,
      "loss": 2.1655,
      "step": 131890
    },
    {
      "epoch": 1.72,
      "grad_norm": 0.9890764951705933,
      "learning_rate": 0.0002323182693159446,
      "loss": 2.1285,
      "step": 131891
    },
    {
      "epoch": 1.72,
      "grad_norm": 0.923092782497406,
      "learning_rate": 0.0002323142842201558,
      "loss": 2.1111,
      "step": 131892
    },
    {
      "epoch": 1.72,
      "grad_norm": 0.9395527243614197,
      "learning_rate": 0.00023231029913695094,
      "loss": 2.2364,
      "step": 131893
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.2733324766159058,
      "learning_rate": 0.00023230631406633088,
      "loss": 2.1544,
      "step": 131894
    },
    {
      "epoch": 1.72,
      "grad_norm": 0.964533269405365,
      "learning_rate": 0.00023230232900829637,
      "loss": 2.2205,
      "step": 131895
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.0240857601165771,
      "learning_rate": 0.0002322983439628481,
      "loss": 2.0666,
      "step": 131896
    },
    {
      "epoch": 1.72,
      "grad_norm": 0.9384683966636658,
      "learning_rate": 0.00023229435892998686,
      "loss": 2.0706,
      "step": 131897
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.0043905973434448,
      "learning_rate": 0.0002322903739097135,
      "loss": 2.0868,
      "step": 131898
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.0777186155319214,
      "learning_rate": 0.00023228638890202853,
      "loss": 2.0819,
      "step": 131899
    },
    {
      "epoch": 1.72,
      "grad_norm": 0.9371175169944763,
      "learning_rate": 0.00023228240390693274,
      "loss": 2.0699,
      "step": 131900
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.0416944026947021,
      "learning_rate": 0.00023227841892442698,
      "loss": 1.9156,
      "step": 131901
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.0052181482315063,
      "learning_rate": 0.00023227443395451192,
      "loss": 2.3956,
      "step": 131902
    },
    {
      "epoch": 1.72,
      "grad_norm": 0.963929295539856,
      "learning_rate": 0.00023227044899718833,
      "loss": 2.0564,
      "step": 131903
    },
    {
      "epoch": 1.72,
      "grad_norm": 0.9608756303787231,
      "learning_rate": 0.00023226646405245712,
      "loss": 2.068,
      "step": 131904
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.02622652053833,
      "learning_rate": 0.00023226247912031868,
      "loss": 2.1599,
      "step": 131905
    },
    {
      "epoch": 1.72,
      "grad_norm": 0.9216026067733765,
      "learning_rate": 0.00023225849420077392,
      "loss": 1.9614,
      "step": 131906
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.090051293373108,
      "learning_rate": 0.00023225450929382355,
      "loss": 2.1381,
      "step": 131907
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.2229710817337036,
      "learning_rate": 0.0002322505243994684,
      "loss": 2.1872,
      "step": 131908
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.0047180652618408,
      "learning_rate": 0.00023224653951770913,
      "loss": 2.219,
      "step": 131909
    },
    {
      "epoch": 1.72,
      "grad_norm": 0.9699611663818359,
      "learning_rate": 0.00023224255464854664,
      "loss": 2.0925,
      "step": 131910
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.0134159326553345,
      "learning_rate": 0.00023223856979198143,
      "loss": 2.082,
      "step": 131911
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.0833402872085571,
      "learning_rate": 0.00023223458494801431,
      "loss": 2.2442,
      "step": 131912
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.097953200340271,
      "learning_rate": 0.0002322306001166461,
      "loss": 1.9142,
      "step": 131913
    },
    {
      "epoch": 1.72,
      "grad_norm": 0.9959991574287415,
      "learning_rate": 0.00023222661529787746,
      "loss": 2.0479,
      "step": 131914
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.0813978910446167,
      "learning_rate": 0.00023222263049170922,
      "loss": 2.1999,
      "step": 131915
    },
    {
      "epoch": 1.72,
      "grad_norm": 0.9063571095466614,
      "learning_rate": 0.00023221864569814216,
      "loss": 2.1836,
      "step": 131916
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.0418334007263184,
      "learning_rate": 0.00023221466091717682,
      "loss": 1.9873,
      "step": 131917
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.2138694524765015,
      "learning_rate": 0.00023221067614881405,
      "loss": 2.1186,
      "step": 131918
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.186686396598816,
      "learning_rate": 0.0002322066913930546,
      "loss": 2.3291,
      "step": 131919
    },
    {
      "epoch": 1.72,
      "grad_norm": 0.9873037338256836,
      "learning_rate": 0.0002322027066498992,
      "loss": 2.18,
      "step": 131920
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.0897959470748901,
      "learning_rate": 0.0002321987219193486,
      "loss": 2.1722,
      "step": 131921
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.0472480058670044,
      "learning_rate": 0.00023219473720140354,
      "loss": 2.0293,
      "step": 131922
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.0471283197402954,
      "learning_rate": 0.00023219075249606487,
      "loss": 2.1345,
      "step": 131923
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.0304763317108154,
      "learning_rate": 0.00023218676780333315,
      "loss": 1.9312,
      "step": 131924
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.005357027053833,
      "learning_rate": 0.00023218278312320912,
      "loss": 2.0796,
      "step": 131925
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.0223534107208252,
      "learning_rate": 0.0002321787984556936,
      "loss": 2.1241,
      "step": 131926
    },
    {
      "epoch": 1.72,
      "grad_norm": 0.9669048190116882,
      "learning_rate": 0.00023217481380078734,
      "loss": 1.9322,
      "step": 131927
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.0153615474700928,
      "learning_rate": 0.00023217082915849105,
      "loss": 2.0258,
      "step": 131928
    },
    {
      "epoch": 1.72,
      "grad_norm": 0.9620859622955322,
      "learning_rate": 0.00023216684452880565,
      "loss": 2.1238,
      "step": 131929
    },
    {
      "epoch": 1.72,
      "grad_norm": 0.9293130040168762,
      "learning_rate": 0.00023216285991173156,
      "loss": 2.0044,
      "step": 131930
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.0232501029968262,
      "learning_rate": 0.00023215887530726964,
      "loss": 2.189,
      "step": 131931
    },
    {
      "epoch": 1.72,
      "grad_norm": 0.970867931842804,
      "learning_rate": 0.0002321548907154207,
      "loss": 2.2427,
      "step": 131932
    },
    {
      "epoch": 1.72,
      "grad_norm": 0.9407488703727722,
      "learning_rate": 0.00023215090613618549,
      "loss": 2.0413,
      "step": 131933
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.0701303482055664,
      "learning_rate": 0.00023214692156956463,
      "loss": 2.0294,
      "step": 131934
    },
    {
      "epoch": 1.72,
      "grad_norm": 0.9919075965881348,
      "learning_rate": 0.00023214293701555904,
      "loss": 1.9701,
      "step": 131935
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.025011658668518,
      "learning_rate": 0.00023213895247416928,
      "loss": 2.039,
      "step": 131936
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.0355116128921509,
      "learning_rate": 0.00023213496794539625,
      "loss": 2.2025,
      "step": 131937
    },
    {
      "epoch": 1.72,
      "grad_norm": 0.9681450128555298,
      "learning_rate": 0.00023213098342924054,
      "loss": 2.1647,
      "step": 131938
    },
    {
      "epoch": 1.72,
      "grad_norm": 0.9087229371070862,
      "learning_rate": 0.00023212699892570296,
      "loss": 2.1752,
      "step": 131939
    },
    {
      "epoch": 1.72,
      "grad_norm": 0.9331812262535095,
      "learning_rate": 0.00023212301443478425,
      "loss": 1.9576,
      "step": 131940
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.104797124862671,
      "learning_rate": 0.00023211902995648521,
      "loss": 2.1751,
      "step": 131941
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.0291391611099243,
      "learning_rate": 0.00023211504549080648,
      "loss": 2.3483,
      "step": 131942
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.0667953491210938,
      "learning_rate": 0.00023211106103774882,
      "loss": 2.1384,
      "step": 131943
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.0681883096694946,
      "learning_rate": 0.00023210707659731308,
      "loss": 2.0301,
      "step": 131944
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.1218702793121338,
      "learning_rate": 0.00023210309216949984,
      "loss": 2.1345,
      "step": 131945
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.1990292072296143,
      "learning_rate": 0.00023209910775430994,
      "loss": 2.1173,
      "step": 131946
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.0775758028030396,
      "learning_rate": 0.0002320951233517441,
      "loss": 1.9416,
      "step": 131947
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.101776123046875,
      "learning_rate": 0.00023209113896180308,
      "loss": 2.2576,
      "step": 131948
    },
    {
      "epoch": 1.72,
      "grad_norm": 0.9527446627616882,
      "learning_rate": 0.00023208715458448756,
      "loss": 2.1764,
      "step": 131949
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.074280858039856,
      "learning_rate": 0.0002320831702197983,
      "loss": 2.1401,
      "step": 131950
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.1314129829406738,
      "learning_rate": 0.00023207918586773613,
      "loss": 2.1694,
      "step": 131951
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.0230070352554321,
      "learning_rate": 0.00023207520152830169,
      "loss": 2.1122,
      "step": 131952
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.0028735399246216,
      "learning_rate": 0.00023207121720149578,
      "loss": 2.1075,
      "step": 131953
    },
    {
      "epoch": 1.72,
      "grad_norm": 0.9861878156661987,
      "learning_rate": 0.00023206723288731907,
      "loss": 2.0126,
      "step": 131954
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.1024922132492065,
      "learning_rate": 0.00023206324858577232,
      "loss": 2.059,
      "step": 131955
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.001657247543335,
      "learning_rate": 0.00023205926429685633,
      "loss": 2.1985,
      "step": 131956
    },
    {
      "epoch": 1.72,
      "grad_norm": 0.950807511806488,
      "learning_rate": 0.0002320552800205718,
      "loss": 1.9875,
      "step": 131957
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.0159635543823242,
      "learning_rate": 0.00023205129575691954,
      "loss": 2.0857,
      "step": 131958
    },
    {
      "epoch": 1.72,
      "grad_norm": 0.9689075350761414,
      "learning_rate": 0.00023204731150590019,
      "loss": 1.9288,
      "step": 131959
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.0248640775680542,
      "learning_rate": 0.00023204332726751447,
      "loss": 1.991,
      "step": 131960
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.0695297718048096,
      "learning_rate": 0.00023203934304176323,
      "loss": 2.1077,
      "step": 131961
    },
    {
      "epoch": 1.72,
      "grad_norm": 0.9999527931213379,
      "learning_rate": 0.0002320353588286471,
      "loss": 2.2142,
      "step": 131962
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.021633267402649,
      "learning_rate": 0.0002320313746281669,
      "loss": 2.1419,
      "step": 131963
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.0606919527053833,
      "learning_rate": 0.00023202739044032335,
      "loss": 2.2003,
      "step": 131964
    },
    {
      "epoch": 1.72,
      "grad_norm": 0.999031662940979,
      "learning_rate": 0.00023202340626511734,
      "loss": 2.2779,
      "step": 131965
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.0589749813079834,
      "learning_rate": 0.0002320194221025493,
      "loss": 2.1148,
      "step": 131966
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.3389991521835327,
      "learning_rate": 0.00023201543795262018,
      "loss": 2.3551,
      "step": 131967
    },
    {
      "epoch": 1.72,
      "grad_norm": 0.9947914481163025,
      "learning_rate": 0.00023201145381533062,
      "loss": 2.0243,
      "step": 131968
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.0204912424087524,
      "learning_rate": 0.00023200746969068148,
      "loss": 2.2549,
      "step": 131969
    },
    {
      "epoch": 1.72,
      "grad_norm": 0.9727646708488464,
      "learning_rate": 0.00023200348557867337,
      "loss": 2.2185,
      "step": 131970
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.006253957748413,
      "learning_rate": 0.00023199950147930726,
      "loss": 1.9151,
      "step": 131971
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.0174641609191895,
      "learning_rate": 0.00023199551739258356,
      "loss": 2.4246,
      "step": 131972
    },
    {
      "epoch": 1.72,
      "grad_norm": 0.9806106686592102,
      "learning_rate": 0.0002319915333185032,
      "loss": 2.0036,
      "step": 131973
    },
    {
      "epoch": 1.72,
      "grad_norm": 0.9974018335342407,
      "learning_rate": 0.00023198754925706693,
      "loss": 2.1851,
      "step": 131974
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.0008220672607422,
      "learning_rate": 0.00023198356520827542,
      "loss": 1.9686,
      "step": 131975
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.017679214477539,
      "learning_rate": 0.00023197958117212946,
      "loss": 2.0718,
      "step": 131976
    },
    {
      "epoch": 1.72,
      "grad_norm": 0.9688100814819336,
      "learning_rate": 0.00023197559714862987,
      "loss": 2.2107,
      "step": 131977
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.0731947422027588,
      "learning_rate": 0.00023197161313777724,
      "loss": 2.2087,
      "step": 131978
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.0560612678527832,
      "learning_rate": 0.00023196762913957233,
      "loss": 2.0514,
      "step": 131979
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.0387864112854004,
      "learning_rate": 0.00023196364515401592,
      "loss": 2.3242,
      "step": 131980
    },
    {
      "epoch": 1.72,
      "grad_norm": 0.9956958293914795,
      "learning_rate": 0.00023195966118110876,
      "loss": 2.2181,
      "step": 131981
    },
    {
      "epoch": 1.72,
      "grad_norm": 0.941288411617279,
      "learning_rate": 0.0002319556772208516,
      "loss": 2.001,
      "step": 131982
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.0247952938079834,
      "learning_rate": 0.00023195169327324517,
      "loss": 2.0666,
      "step": 131983
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.0143449306488037,
      "learning_rate": 0.0002319477093382903,
      "loss": 2.1411,
      "step": 131984
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.0789598226547241,
      "learning_rate": 0.0002319437254159875,
      "loss": 2.1882,
      "step": 131985
    },
    {
      "epoch": 1.72,
      "grad_norm": 0.9998188018798828,
      "learning_rate": 0.00023193974150633768,
      "loss": 2.3548,
      "step": 131986
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.0982857942581177,
      "learning_rate": 0.00023193575760934153,
      "loss": 1.9277,
      "step": 131987
    },
    {
      "epoch": 1.72,
      "grad_norm": 0.9384368062019348,
      "learning_rate": 0.0002319317737249998,
      "loss": 2.0341,
      "step": 131988
    },
    {
      "epoch": 1.72,
      "grad_norm": 0.9708057045936584,
      "learning_rate": 0.00023192778985331328,
      "loss": 2.2314,
      "step": 131989
    },
    {
      "epoch": 1.72,
      "grad_norm": 0.9824034571647644,
      "learning_rate": 0.00023192380599428274,
      "loss": 2.2047,
      "step": 131990
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.030899167060852,
      "learning_rate": 0.00023191982214790875,
      "loss": 1.8665,
      "step": 131991
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.075850248336792,
      "learning_rate": 0.00023191583831419214,
      "loss": 2.1073,
      "step": 131992
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.0288357734680176,
      "learning_rate": 0.00023191185449313368,
      "loss": 2.0304,
      "step": 131993
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.0856322050094604,
      "learning_rate": 0.0002319078706847341,
      "loss": 2.5109,
      "step": 131994
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.0376248359680176,
      "learning_rate": 0.0002319038868889941,
      "loss": 2.2354,
      "step": 131995
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.1326862573623657,
      "learning_rate": 0.0002318999031059146,
      "loss": 1.9665,
      "step": 131996
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.005228877067566,
      "learning_rate": 0.00023189591933549605,
      "loss": 2.0356,
      "step": 131997
    },
    {
      "epoch": 1.72,
      "grad_norm": 0.979259729385376,
      "learning_rate": 0.00023189193557773935,
      "loss": 2.117,
      "step": 131998
    },
    {
      "epoch": 1.72,
      "grad_norm": 0.9623521566390991,
      "learning_rate": 0.00023188795183264524,
      "loss": 2.028,
      "step": 131999
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.0126023292541504,
      "learning_rate": 0.0002318839681002144,
      "loss": 2.1057,
      "step": 132000
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.1792205572128296,
      "learning_rate": 0.00023187998438044765,
      "loss": 2.0545,
      "step": 132001
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.0801969766616821,
      "learning_rate": 0.00023187600067334582,
      "loss": 2.0723,
      "step": 132002
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.1102380752563477,
      "learning_rate": 0.00023187201697890937,
      "loss": 2.1494,
      "step": 132003
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.136213779449463,
      "learning_rate": 0.00023186803329713923,
      "loss": 1.8085,
      "step": 132004
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.1573989391326904,
      "learning_rate": 0.00023186404962803608,
      "loss": 1.987,
      "step": 132005
    },
    {
      "epoch": 1.72,
      "grad_norm": 0.9510424733161926,
      "learning_rate": 0.00023186006597160073,
      "loss": 2.0959,
      "step": 132006
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.065917730331421,
      "learning_rate": 0.00023185608232783385,
      "loss": 2.2056,
      "step": 132007
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.194014072418213,
      "learning_rate": 0.00023185209869673636,
      "loss": 2.0497,
      "step": 132008
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.3717231750488281,
      "learning_rate": 0.0002318481150783087,
      "loss": 1.9671,
      "step": 132009
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.0828126668930054,
      "learning_rate": 0.00023184413147255177,
      "loss": 2.1396,
      "step": 132010
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.0389673709869385,
      "learning_rate": 0.0002318401478794663,
      "loss": 2.0586,
      "step": 132011
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.0591626167297363,
      "learning_rate": 0.00023183616429905306,
      "loss": 2.2162,
      "step": 132012
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.3487043380737305,
      "learning_rate": 0.00023183218073131277,
      "loss": 2.2202,
      "step": 132013
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.0560033321380615,
      "learning_rate": 0.00023182819717624625,
      "loss": 2.1017,
      "step": 132014
    },
    {
      "epoch": 1.72,
      "grad_norm": 0.9479091167449951,
      "learning_rate": 0.00023182421363385404,
      "loss": 2.2163,
      "step": 132015
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.124348759651184,
      "learning_rate": 0.00023182023010413698,
      "loss": 2.1399,
      "step": 132016
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.237526297569275,
      "learning_rate": 0.00023181624658709585,
      "loss": 2.1828,
      "step": 132017
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.046469807624817,
      "learning_rate": 0.00023181226308273138,
      "loss": 2.0382,
      "step": 132018
    },
    {
      "epoch": 1.72,
      "grad_norm": 0.9273926019668579,
      "learning_rate": 0.00023180827959104426,
      "loss": 2.0879,
      "step": 132019
    },
    {
      "epoch": 1.72,
      "grad_norm": 0.9166645407676697,
      "learning_rate": 0.00023180429611203534,
      "loss": 2.0261,
      "step": 132020
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.314321517944336,
      "learning_rate": 0.0002318003126457053,
      "loss": 2.0294,
      "step": 132021
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.1731702089309692,
      "learning_rate": 0.00023179632919205482,
      "loss": 2.1347,
      "step": 132022
    },
    {
      "epoch": 1.72,
      "grad_norm": 0.9424267411231995,
      "learning_rate": 0.00023179234575108467,
      "loss": 2.248,
      "step": 132023
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.0239195823669434,
      "learning_rate": 0.00023178836232279562,
      "loss": 2.1334,
      "step": 132024
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.0629642009735107,
      "learning_rate": 0.00023178437890718838,
      "loss": 2.0698,
      "step": 132025
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.1370023488998413,
      "learning_rate": 0.0002317803955042638,
      "loss": 2.258,
      "step": 132026
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.0183203220367432,
      "learning_rate": 0.00023177641211402246,
      "loss": 2.05,
      "step": 132027
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.1494083404541016,
      "learning_rate": 0.00023177242873646518,
      "loss": 2.387,
      "step": 132028
    },
    {
      "epoch": 1.72,
      "grad_norm": 0.9857617020606995,
      "learning_rate": 0.00023176844537159275,
      "loss": 1.9947,
      "step": 132029
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.2933349609375,
      "learning_rate": 0.0002317644620194058,
      "loss": 2.0296,
      "step": 132030
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.0488978624343872,
      "learning_rate": 0.00023176047867990508,
      "loss": 2.1721,
      "step": 132031
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.1168047189712524,
      "learning_rate": 0.00023175649535309148,
      "loss": 2.2871,
      "step": 132032
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.2890520095825195,
      "learning_rate": 0.0002317525120389656,
      "loss": 2.2433,
      "step": 132033
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.1561470031738281,
      "learning_rate": 0.00023174852873752814,
      "loss": 2.3922,
      "step": 132034
    },
    {
      "epoch": 1.72,
      "grad_norm": 0.984113335609436,
      "learning_rate": 0.00023174454544877998,
      "loss": 2.3323,
      "step": 132035
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.0377230644226074,
      "learning_rate": 0.0002317405621727218,
      "loss": 2.0127,
      "step": 132036
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.0955218076705933,
      "learning_rate": 0.00023173657890935434,
      "loss": 1.9145,
      "step": 132037
    },
    {
      "epoch": 1.72,
      "grad_norm": 0.9703455567359924,
      "learning_rate": 0.00023173259565867837,
      "loss": 1.9995,
      "step": 132038
    },
    {
      "epoch": 1.72,
      "grad_norm": 0.9684842228889465,
      "learning_rate": 0.00023172861242069456,
      "loss": 2.1032,
      "step": 132039
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.0329588651657104,
      "learning_rate": 0.00023172462919540364,
      "loss": 2.2222,
      "step": 132040
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.0383001565933228,
      "learning_rate": 0.00023172064598280646,
      "loss": 2.1406,
      "step": 132041
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.0427182912826538,
      "learning_rate": 0.00023171666278290368,
      "loss": 2.0584,
      "step": 132042
    },
    {
      "epoch": 1.72,
      "grad_norm": 0.9546343684196472,
      "learning_rate": 0.00023171267959569612,
      "loss": 2.0741,
      "step": 132043
    },
    {
      "epoch": 1.72,
      "grad_norm": 0.9985321164131165,
      "learning_rate": 0.00023170869642118446,
      "loss": 2.0598,
      "step": 132044
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.054970145225525,
      "learning_rate": 0.00023170471325936938,
      "loss": 2.1001,
      "step": 132045
    },
    {
      "epoch": 1.72,
      "grad_norm": 0.9866475462913513,
      "learning_rate": 0.00023170073011025166,
      "loss": 2.1028,
      "step": 132046
    },
    {
      "epoch": 1.72,
      "grad_norm": 0.9997212886810303,
      "learning_rate": 0.00023169674697383206,
      "loss": 2.2281,
      "step": 132047
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.0908178091049194,
      "learning_rate": 0.00023169276385011134,
      "loss": 2.1808,
      "step": 132048
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.0003875494003296,
      "learning_rate": 0.00023168878073909026,
      "loss": 2.0972,
      "step": 132049
    },
    {
      "epoch": 1.72,
      "grad_norm": 0.9560279846191406,
      "learning_rate": 0.00023168479764076958,
      "loss": 1.923,
      "step": 132050
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.067520022392273,
      "learning_rate": 0.00023168081455514998,
      "loss": 2.0888,
      "step": 132051
    },
    {
      "epoch": 1.72,
      "grad_norm": 0.965506374835968,
      "learning_rate": 0.00023167683148223207,
      "loss": 2.2328,
      "step": 132052
    },
    {
      "epoch": 1.72,
      "grad_norm": 0.9695234894752502,
      "learning_rate": 0.00023167284842201682,
      "loss": 2.0041,
      "step": 132053
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.0433317422866821,
      "learning_rate": 0.00023166886537450486,
      "loss": 2.034,
      "step": 132054
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.0397619009017944,
      "learning_rate": 0.0002316648823396969,
      "loss": 2.1725,
      "step": 132055
    },
    {
      "epoch": 1.72,
      "grad_norm": 0.9825720191001892,
      "learning_rate": 0.00023166089931759377,
      "loss": 2.106,
      "step": 132056
    },
    {
      "epoch": 1.72,
      "grad_norm": 0.9684725403785706,
      "learning_rate": 0.00023165691630819626,
      "loss": 1.9051,
      "step": 132057
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.063100814819336,
      "learning_rate": 0.00023165293331150493,
      "loss": 2.1668,
      "step": 132058
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.1207095384597778,
      "learning_rate": 0.0002316489503275206,
      "loss": 2.2566,
      "step": 132059
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.0703167915344238,
      "learning_rate": 0.00023164496735624405,
      "loss": 2.341,
      "step": 132060
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.0976604223251343,
      "learning_rate": 0.00023164098439767595,
      "loss": 2.0175,
      "step": 132061
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.075023889541626,
      "learning_rate": 0.0002316370014518171,
      "loss": 2.2076,
      "step": 132062
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.0256619453430176,
      "learning_rate": 0.00023163301851866834,
      "loss": 2.1787,
      "step": 132063
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.2680412530899048,
      "learning_rate": 0.00023162903559823013,
      "loss": 2.0267,
      "step": 132064
    },
    {
      "epoch": 1.72,
      "grad_norm": 0.9566916823387146,
      "learning_rate": 0.00023162505269050344,
      "loss": 2.1083,
      "step": 132065
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.096643090248108,
      "learning_rate": 0.0002316210697954889,
      "loss": 2.016,
      "step": 132066
    },
    {
      "epoch": 1.72,
      "grad_norm": 0.9741236567497253,
      "learning_rate": 0.0002316170869131873,
      "loss": 2.0267,
      "step": 132067
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.1075031757354736,
      "learning_rate": 0.00023161310404359937,
      "loss": 1.9233,
      "step": 132068
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.1534494161605835,
      "learning_rate": 0.000231609121186726,
      "loss": 1.9459,
      "step": 132069
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.0191701650619507,
      "learning_rate": 0.00023160513834256764,
      "loss": 2.1175,
      "step": 132070
    },
    {
      "epoch": 1.72,
      "grad_norm": 0.9520803093910217,
      "learning_rate": 0.0002316011555111252,
      "loss": 1.8644,
      "step": 132071
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.0511776208877563,
      "learning_rate": 0.0002315971726923994,
      "loss": 2.1445,
      "step": 132072
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.1115596294403076,
      "learning_rate": 0.00023159318988639094,
      "loss": 2.208,
      "step": 132073
    },
    {
      "epoch": 1.72,
      "grad_norm": 0.9723315834999084,
      "learning_rate": 0.00023158920709310062,
      "loss": 2.1796,
      "step": 132074
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.0534896850585938,
      "learning_rate": 0.00023158522431252927,
      "loss": 2.4283,
      "step": 132075
    },
    {
      "epoch": 1.72,
      "grad_norm": 0.9683542251586914,
      "learning_rate": 0.0002315812415446774,
      "loss": 2.2286,
      "step": 132076
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.030885934829712,
      "learning_rate": 0.0002315772587895459,
      "loss": 2.0775,
      "step": 132077
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.4353666305541992,
      "learning_rate": 0.00023157327604713543,
      "loss": 2.0401,
      "step": 132078
    },
    {
      "epoch": 1.72,
      "grad_norm": 0.9112005829811096,
      "learning_rate": 0.0002315692933174468,
      "loss": 2.06,
      "step": 132079
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.0417126417160034,
      "learning_rate": 0.00023156531060048073,
      "loss": 2.0409,
      "step": 132080
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.0133146047592163,
      "learning_rate": 0.0002315613278962381,
      "loss": 2.034,
      "step": 132081
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.0289877653121948,
      "learning_rate": 0.00023155734520471935,
      "loss": 2.2437,
      "step": 132082
    },
    {
      "epoch": 1.72,
      "grad_norm": 0.9661388397216797,
      "learning_rate": 0.00023155336252592541,
      "loss": 2.1125,
      "step": 132083
    },
    {
      "epoch": 1.72,
      "grad_norm": 0.9629696011543274,
      "learning_rate": 0.00023154937985985698,
      "loss": 2.0112,
      "step": 132084
    },
    {
      "epoch": 1.72,
      "grad_norm": 0.959319531917572,
      "learning_rate": 0.0002315453972065148,
      "loss": 1.9998,
      "step": 132085
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.0092695951461792,
      "learning_rate": 0.0002315414145658997,
      "loss": 2.2729,
      "step": 132086
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.0691611766815186,
      "learning_rate": 0.00023153743193801235,
      "loss": 1.995,
      "step": 132087
    },
    {
      "epoch": 1.72,
      "grad_norm": 0.9681949615478516,
      "learning_rate": 0.00023153344932285344,
      "loss": 2.1818,
      "step": 132088
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.0474237203598022,
      "learning_rate": 0.00023152946672042374,
      "loss": 2.106,
      "step": 132089
    },
    {
      "epoch": 1.72,
      "grad_norm": 0.9235805869102478,
      "learning_rate": 0.00023152548413072396,
      "loss": 2.0496,
      "step": 132090
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.0488938093185425,
      "learning_rate": 0.00023152150155375493,
      "loss": 2.2281,
      "step": 132091
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.0000994205474854,
      "learning_rate": 0.00023151751898951734,
      "loss": 2.0279,
      "step": 132092
    },
    {
      "epoch": 1.72,
      "grad_norm": 0.9408000111579895,
      "learning_rate": 0.00023151353643801207,
      "loss": 2.0254,
      "step": 132093
    },
    {
      "epoch": 1.72,
      "grad_norm": 0.9805954694747925,
      "learning_rate": 0.0002315095538992396,
      "loss": 2.0737,
      "step": 132094
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.0424877405166626,
      "learning_rate": 0.00023150557137320077,
      "loss": 2.2555,
      "step": 132095
    },
    {
      "epoch": 1.72,
      "grad_norm": 0.9683307409286499,
      "learning_rate": 0.00023150158885989634,
      "loss": 2.0511,
      "step": 132096
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.0318125486373901,
      "learning_rate": 0.00023149760635932708,
      "loss": 1.9652,
      "step": 132097
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.0599324703216553,
      "learning_rate": 0.0002314936238714937,
      "loss": 2.1089,
      "step": 132098
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.0829440355300903,
      "learning_rate": 0.00023148964139639713,
      "loss": 1.9839,
      "step": 132099
    },
    {
      "epoch": 1.72,
      "grad_norm": 0.9395480155944824,
      "learning_rate": 0.00023148565893403773,
      "loss": 2.2895,
      "step": 132100
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.1351021528244019,
      "learning_rate": 0.0002314816764844165,
      "loss": 2.149,
      "step": 132101
    },
    {
      "epoch": 1.72,
      "grad_norm": 0.9864413142204285,
      "learning_rate": 0.00023147769404753405,
      "loss": 2.1094,
      "step": 132102
    },
    {
      "epoch": 1.72,
      "grad_norm": 0.9704375863075256,
      "learning_rate": 0.00023147371162339123,
      "loss": 2.1681,
      "step": 132103
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.0002570152282715,
      "learning_rate": 0.00023146972921198871,
      "loss": 2.2492,
      "step": 132104
    },
    {
      "epoch": 1.72,
      "grad_norm": 0.9679068922996521,
      "learning_rate": 0.0002314657468133274,
      "loss": 2.3982,
      "step": 132105
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.1031912565231323,
      "learning_rate": 0.00023146176442740782,
      "loss": 2.3992,
      "step": 132106
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.0522692203521729,
      "learning_rate": 0.00023145778205423074,
      "loss": 2.1761,
      "step": 132107
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.0775763988494873,
      "learning_rate": 0.000231453799693797,
      "loss": 2.2413,
      "step": 132108
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.1063605546951294,
      "learning_rate": 0.00023144981734610724,
      "loss": 2.0188,
      "step": 132109
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.0507676601409912,
      "learning_rate": 0.00023144583501116228,
      "loss": 2.2263,
      "step": 132110
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.186423659324646,
      "learning_rate": 0.0002314418526889629,
      "loss": 1.9058,
      "step": 132111
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.064528465270996,
      "learning_rate": 0.0002314378703795097,
      "loss": 1.9394,
      "step": 132112
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.040831446647644,
      "learning_rate": 0.00023143388808280358,
      "loss": 2.1577,
      "step": 132113
    },
    {
      "epoch": 1.72,
      "grad_norm": 0.99862140417099,
      "learning_rate": 0.0002314299057988451,
      "loss": 2.0891,
      "step": 132114
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.0446946620941162,
      "learning_rate": 0.0002314259235276351,
      "loss": 2.176,
      "step": 132115
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.0005568265914917,
      "learning_rate": 0.00023142194126917436,
      "loss": 2.2637,
      "step": 132116
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.116272211074829,
      "learning_rate": 0.0002314179590234635,
      "loss": 2.1517,
      "step": 132117
    },
    {
      "epoch": 1.72,
      "grad_norm": 0.9883462190628052,
      "learning_rate": 0.00023141397679050344,
      "loss": 2.2768,
      "step": 132118
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.1806445121765137,
      "learning_rate": 0.00023140999457029477,
      "loss": 2.104,
      "step": 132119
    },
    {
      "epoch": 1.72,
      "grad_norm": 0.953192949295044,
      "learning_rate": 0.00023140601236283835,
      "loss": 1.8447,
      "step": 132120
    },
    {
      "epoch": 1.72,
      "grad_norm": 0.9596018195152283,
      "learning_rate": 0.00023140203016813472,
      "loss": 2.2396,
      "step": 132121
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.0826338529586792,
      "learning_rate": 0.00023139804798618478,
      "loss": 2.0003,
      "step": 132122
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.1023778915405273,
      "learning_rate": 0.00023139406581698923,
      "loss": 2.1193,
      "step": 132123
    },
    {
      "epoch": 1.72,
      "grad_norm": 0.8750163316726685,
      "learning_rate": 0.00023139008366054889,
      "loss": 2.1588,
      "step": 132124
    },
    {
      "epoch": 1.72,
      "grad_norm": 0.9618548154830933,
      "learning_rate": 0.00023138610151686437,
      "loss": 2.0917,
      "step": 132125
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.1480698585510254,
      "learning_rate": 0.00023138211938593647,
      "loss": 2.0866,
      "step": 132126
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.0181750059127808,
      "learning_rate": 0.0002313781372677659,
      "loss": 2.1335,
      "step": 132127
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.0380651950836182,
      "learning_rate": 0.00023137415516235354,
      "loss": 2.0249,
      "step": 132128
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.0562375783920288,
      "learning_rate": 0.00023137017306969996,
      "loss": 2.159,
      "step": 132129
    },
    {
      "epoch": 1.72,
      "grad_norm": 0.9236564636230469,
      "learning_rate": 0.000231366190989806,
      "loss": 2.2319,
      "step": 132130
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.0414448976516724,
      "learning_rate": 0.00023136220892267225,
      "loss": 2.1316,
      "step": 132131
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.0375958681106567,
      "learning_rate": 0.00023135822686829962,
      "loss": 2.0759,
      "step": 132132
    },
    {
      "epoch": 1.72,
      "grad_norm": 0.956566572189331,
      "learning_rate": 0.00023135424482668878,
      "loss": 2.0797,
      "step": 132133
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.1212003231048584,
      "learning_rate": 0.0002313502627978405,
      "loss": 2.287,
      "step": 132134
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.1183615922927856,
      "learning_rate": 0.00023134628078175554,
      "loss": 2.1917,
      "step": 132135
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.0509371757507324,
      "learning_rate": 0.00023134229877843458,
      "loss": 2.1478,
      "step": 132136
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.096458077430725,
      "learning_rate": 0.00023133831678787836,
      "loss": 2.178,
      "step": 132137
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.0648683309555054,
      "learning_rate": 0.00023133433481008762,
      "loss": 2.3323,
      "step": 132138
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.1953518390655518,
      "learning_rate": 0.00023133035284506314,
      "loss": 2.0139,
      "step": 132139
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.2165015935897827,
      "learning_rate": 0.00023132637089280566,
      "loss": 2.0323,
      "step": 132140
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.0548315048217773,
      "learning_rate": 0.00023132238895331585,
      "loss": 2.0466,
      "step": 132141
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.0429648160934448,
      "learning_rate": 0.00023131840702659466,
      "loss": 2.093,
      "step": 132142
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.101806879043579,
      "learning_rate": 0.00023131442511264256,
      "loss": 2.3205,
      "step": 132143
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.275426983833313,
      "learning_rate": 0.00023131044321146038,
      "loss": 2.0027,
      "step": 132144
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.1438044309616089,
      "learning_rate": 0.00023130646132304893,
      "loss": 2.0813,
      "step": 132145
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.0990192890167236,
      "learning_rate": 0.00023130247944740887,
      "loss": 2.0145,
      "step": 132146
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.0544241666793823,
      "learning_rate": 0.000231298497584541,
      "loss": 2.077,
      "step": 132147
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.0035392045974731,
      "learning_rate": 0.00023129451573444614,
      "loss": 1.9971,
      "step": 132148
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.3785595893859863,
      "learning_rate": 0.0002312905338971248,
      "loss": 2.1777,
      "step": 132149
    },
    {
      "epoch": 1.72,
      "grad_norm": 0.884113609790802,
      "learning_rate": 0.00023128655207257785,
      "loss": 1.9674,
      "step": 132150
    },
    {
      "epoch": 1.72,
      "grad_norm": 0.9817299842834473,
      "learning_rate": 0.00023128257026080602,
      "loss": 1.9483,
      "step": 132151
    },
    {
      "epoch": 1.72,
      "grad_norm": 0.9456648826599121,
      "learning_rate": 0.00023127858846181006,
      "loss": 2.0278,
      "step": 132152
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.1905133724212646,
      "learning_rate": 0.0002312746066755907,
      "loss": 2.1345,
      "step": 132153
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.1572344303131104,
      "learning_rate": 0.00023127062490214883,
      "loss": 2.3179,
      "step": 132154
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.0401427745819092,
      "learning_rate": 0.00023126664314148494,
      "loss": 2.2911,
      "step": 132155
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.043165683746338,
      "learning_rate": 0.00023126266139359984,
      "loss": 2.0834,
      "step": 132156
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.0616739988327026,
      "learning_rate": 0.0002312586796584943,
      "loss": 1.9666,
      "step": 132157
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.0360536575317383,
      "learning_rate": 0.0002312546979361691,
      "loss": 2.3836,
      "step": 132158
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.168744444847107,
      "learning_rate": 0.00023125071622662497,
      "loss": 2.0843,
      "step": 132159
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.0142797231674194,
      "learning_rate": 0.00023124673452986274,
      "loss": 2.0165,
      "step": 132160
    },
    {
      "epoch": 1.72,
      "grad_norm": 0.9999617338180542,
      "learning_rate": 0.00023124275284588288,
      "loss": 1.9828,
      "step": 132161
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.029245376586914,
      "learning_rate": 0.00023123877117468632,
      "loss": 2.0515,
      "step": 132162
    },
    {
      "epoch": 1.72,
      "grad_norm": 0.9637236595153809,
      "learning_rate": 0.00023123478951627373,
      "loss": 2.3538,
      "step": 132163
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.0153447389602661,
      "learning_rate": 0.00023123080787064594,
      "loss": 1.9676,
      "step": 132164
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.2294466495513916,
      "learning_rate": 0.00023122682623780364,
      "loss": 2.176,
      "step": 132165
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.0775481462478638,
      "learning_rate": 0.00023122284461774768,
      "loss": 1.9694,
      "step": 132166
    },
    {
      "epoch": 1.72,
      "grad_norm": 0.9650101661682129,
      "learning_rate": 0.00023121886301047856,
      "loss": 2.2782,
      "step": 132167
    },
    {
      "epoch": 1.72,
      "grad_norm": 0.9640365839004517,
      "learning_rate": 0.00023121488141599717,
      "loss": 1.9603,
      "step": 132168
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.159385323524475,
      "learning_rate": 0.00023121089983430424,
      "loss": 2.1256,
      "step": 132169
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.4059334993362427,
      "learning_rate": 0.00023120691826540045,
      "loss": 2.0481,
      "step": 132170
    },
    {
      "epoch": 1.72,
      "grad_norm": 0.9719269275665283,
      "learning_rate": 0.00023120293670928665,
      "loss": 1.9753,
      "step": 132171
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.0373830795288086,
      "learning_rate": 0.0002311989551659636,
      "loss": 1.9529,
      "step": 132172
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.1240240335464478,
      "learning_rate": 0.00023119497363543188,
      "loss": 2.3911,
      "step": 132173
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.1560205221176147,
      "learning_rate": 0.0002311909921176923,
      "loss": 2.2394,
      "step": 132174
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.060459852218628,
      "learning_rate": 0.0002311870106127456,
      "loss": 2.0626,
      "step": 132175
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.0248277187347412,
      "learning_rate": 0.00023118302912059253,
      "loss": 2.2129,
      "step": 132176
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.0629873275756836,
      "learning_rate": 0.00023117904764123384,
      "loss": 2.2782,
      "step": 132177
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.3026800155639648,
      "learning_rate": 0.00023117506617467042,
      "loss": 1.9972,
      "step": 132178
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.1264315843582153,
      "learning_rate": 0.0002311710847209027,
      "loss": 2.2041,
      "step": 132179
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.0130765438079834,
      "learning_rate": 0.0002311671032799316,
      "loss": 2.1541,
      "step": 132180
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.445881962776184,
      "learning_rate": 0.0002311631218517578,
      "loss": 1.9838,
      "step": 132181
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.386246681213379,
      "learning_rate": 0.00023115914043638207,
      "loss": 2.0788,
      "step": 132182
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.2022202014923096,
      "learning_rate": 0.0002311551590338052,
      "loss": 2.2113,
      "step": 132183
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.0028491020202637,
      "learning_rate": 0.00023115117764402788,
      "loss": 2.2612,
      "step": 132184
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.0158144235610962,
      "learning_rate": 0.00023114719626705094,
      "loss": 1.981,
      "step": 132185
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.0460964441299438,
      "learning_rate": 0.0002311432149028749,
      "loss": 2.1742,
      "step": 132186
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.104776382446289,
      "learning_rate": 0.0002311392335515007,
      "loss": 2.1239,
      "step": 132187
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.089228630065918,
      "learning_rate": 0.00023113525221292897,
      "loss": 2.1081,
      "step": 132188
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.0928516387939453,
      "learning_rate": 0.0002311312708871605,
      "loss": 2.3437,
      "step": 132189
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.1689740419387817,
      "learning_rate": 0.00023112728957419603,
      "loss": 2.1728,
      "step": 132190
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.0938050746917725,
      "learning_rate": 0.00023112330827403647,
      "loss": 1.9631,
      "step": 132191
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.0614792108535767,
      "learning_rate": 0.0002311193269866822,
      "loss": 2.2104,
      "step": 132192
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.0469635725021362,
      "learning_rate": 0.00023111534571213416,
      "loss": 2.3155,
      "step": 132193
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.1804145574569702,
      "learning_rate": 0.0002311113644503931,
      "loss": 1.9599,
      "step": 132194
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.0134055614471436,
      "learning_rate": 0.00023110738320145972,
      "loss": 2.1935,
      "step": 132195
    },
    {
      "epoch": 1.72,
      "grad_norm": 0.9398086667060852,
      "learning_rate": 0.00023110340196533477,
      "loss": 2.0519,
      "step": 132196
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.086728572845459,
      "learning_rate": 0.0002310994207420191,
      "loss": 2.1715,
      "step": 132197
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.2231837511062622,
      "learning_rate": 0.0002310954395315133,
      "loss": 2.0858,
      "step": 132198
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.0633779764175415,
      "learning_rate": 0.0002310914583338181,
      "loss": 2.1418,
      "step": 132199
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.2325496673583984,
      "learning_rate": 0.00023108747714893431,
      "loss": 1.8451,
      "step": 132200
    },
    {
      "epoch": 1.72,
      "grad_norm": 0.9873031973838806,
      "learning_rate": 0.0002310834959768627,
      "loss": 1.9184,
      "step": 132201
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.1406927108764648,
      "learning_rate": 0.0002310795148176039,
      "loss": 2.1131,
      "step": 132202
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.297951579093933,
      "learning_rate": 0.00023107553367115885,
      "loss": 2.3083,
      "step": 132203
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.0691417455673218,
      "learning_rate": 0.00023107155253752806,
      "loss": 2.189,
      "step": 132204
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.0051559209823608,
      "learning_rate": 0.00023106757141671242,
      "loss": 2.0678,
      "step": 132205
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.1706987619400024,
      "learning_rate": 0.00023106359030871258,
      "loss": 2.0637,
      "step": 132206
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.2594655752182007,
      "learning_rate": 0.0002310596092135293,
      "loss": 2.1546,
      "step": 132207
    },
    {
      "epoch": 1.72,
      "grad_norm": 0.9859582185745239,
      "learning_rate": 0.00023105562813116335,
      "loss": 2.1332,
      "step": 132208
    },
    {
      "epoch": 1.72,
      "grad_norm": 0.9358675479888916,
      "learning_rate": 0.00023105164706161555,
      "loss": 2.2174,
      "step": 132209
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.2914801836013794,
      "learning_rate": 0.00023104766600488643,
      "loss": 2.1278,
      "step": 132210
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.1576614379882812,
      "learning_rate": 0.00023104368496097688,
      "loss": 2.0414,
      "step": 132211
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.0422683954238892,
      "learning_rate": 0.00023103970392988768,
      "loss": 2.1565,
      "step": 132212
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.0682905912399292,
      "learning_rate": 0.00023103572291161944,
      "loss": 2.1359,
      "step": 132213
    },
    {
      "epoch": 1.72,
      "grad_norm": 0.9642903804779053,
      "learning_rate": 0.00023103174190617295,
      "loss": 2.2362,
      "step": 132214
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.2745009660720825,
      "learning_rate": 0.00023102776091354904,
      "loss": 1.9344,
      "step": 132215
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.0402053594589233,
      "learning_rate": 0.0002310237799337483,
      "loss": 2.0403,
      "step": 132216
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.039488434791565,
      "learning_rate": 0.00023101979896677152,
      "loss": 2.03,
      "step": 132217
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.0603612661361694,
      "learning_rate": 0.0002310158180126195,
      "loss": 1.9789,
      "step": 132218
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.0425673723220825,
      "learning_rate": 0.0002310118370712929,
      "loss": 2.1968,
      "step": 132219
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.0303435325622559,
      "learning_rate": 0.0002310078561427926,
      "loss": 2.2026,
      "step": 132220
    },
    {
      "epoch": 1.72,
      "grad_norm": 0.9841268062591553,
      "learning_rate": 0.0002310038752271192,
      "loss": 2.0671,
      "step": 132221
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.0523825883865356,
      "learning_rate": 0.00023099989432427348,
      "loss": 2.1485,
      "step": 132222
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.2491134405136108,
      "learning_rate": 0.0002309959134342561,
      "loss": 1.9188,
      "step": 132223
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.024471640586853,
      "learning_rate": 0.00023099193255706797,
      "loss": 2.1198,
      "step": 132224
    },
    {
      "epoch": 1.72,
      "grad_norm": 0.9636040329933167,
      "learning_rate": 0.0002309879516927097,
      "loss": 2.018,
      "step": 132225
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.1246912479400635,
      "learning_rate": 0.00023098397084118206,
      "loss": 2.1363,
      "step": 132226
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.118674397468567,
      "learning_rate": 0.00023097999000248597,
      "loss": 2.139,
      "step": 132227
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.0660037994384766,
      "learning_rate": 0.00023097600917662182,
      "loss": 2.2625,
      "step": 132228
    },
    {
      "epoch": 1.72,
      "grad_norm": 0.9201728701591492,
      "learning_rate": 0.00023097202836359058,
      "loss": 2.1062,
      "step": 132229
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.2636877298355103,
      "learning_rate": 0.00023096804756339294,
      "loss": 1.995,
      "step": 132230
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.0009745359420776,
      "learning_rate": 0.00023096406677602962,
      "loss": 1.9963,
      "step": 132231
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.2413156032562256,
      "learning_rate": 0.00023096008600150138,
      "loss": 2.0745,
      "step": 132232
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.0333566665649414,
      "learning_rate": 0.0002309561052398091,
      "loss": 2.2271,
      "step": 132233
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.1608295440673828,
      "learning_rate": 0.00023095212449095328,
      "loss": 2.1504,
      "step": 132234
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.015315294265747,
      "learning_rate": 0.00023094814375493474,
      "loss": 2.1303,
      "step": 132235
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.1437549591064453,
      "learning_rate": 0.00023094416303175424,
      "loss": 1.9976,
      "step": 132236
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.077499508857727,
      "learning_rate": 0.00023094018232141253,
      "loss": 1.9219,
      "step": 132237
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.115256667137146,
      "learning_rate": 0.00023093620162391033,
      "loss": 2.1357,
      "step": 132238
    },
    {
      "epoch": 1.72,
      "grad_norm": 0.9710819125175476,
      "learning_rate": 0.00023093222093924852,
      "loss": 1.8789,
      "step": 132239
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.2441580295562744,
      "learning_rate": 0.00023092824026742763,
      "loss": 2.1383,
      "step": 132240
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.0168852806091309,
      "learning_rate": 0.00023092425960844845,
      "loss": 2.0882,
      "step": 132241
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.127159595489502,
      "learning_rate": 0.00023092027896231177,
      "loss": 2.1328,
      "step": 132242
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.0729985237121582,
      "learning_rate": 0.0002309162983290183,
      "loss": 2.1453,
      "step": 132243
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.032292366027832,
      "learning_rate": 0.0002309123177085688,
      "loss": 2.0014,
      "step": 132244
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.0403224229812622,
      "learning_rate": 0.00023090833710096413,
      "loss": 2.1471,
      "step": 132245
    },
    {
      "epoch": 1.72,
      "grad_norm": 0.944694459438324,
      "learning_rate": 0.00023090435650620478,
      "loss": 1.9187,
      "step": 132246
    },
    {
      "epoch": 1.72,
      "grad_norm": 0.9627090692520142,
      "learning_rate": 0.0002309003759242916,
      "loss": 2.1162,
      "step": 132247
    },
    {
      "epoch": 1.72,
      "grad_norm": 0.9399558305740356,
      "learning_rate": 0.00023089639535522536,
      "loss": 2.1406,
      "step": 132248
    },
    {
      "epoch": 1.72,
      "grad_norm": 0.9901017546653748,
      "learning_rate": 0.00023089241479900676,
      "loss": 2.0368,
      "step": 132249
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.0822914838790894,
      "learning_rate": 0.0002308884342556366,
      "loss": 2.1827,
      "step": 132250
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.0303497314453125,
      "learning_rate": 0.00023088445372511558,
      "loss": 1.9039,
      "step": 132251
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.0060875415802002,
      "learning_rate": 0.00023088047320744457,
      "loss": 1.8928,
      "step": 132252
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.112650752067566,
      "learning_rate": 0.000230876492702624,
      "loss": 2.1487,
      "step": 132253
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.2006667852401733,
      "learning_rate": 0.00023087251221065488,
      "loss": 2.0539,
      "step": 132254
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.014346718788147,
      "learning_rate": 0.00023086853173153783,
      "loss": 1.9689,
      "step": 132255
    },
    {
      "epoch": 1.72,
      "grad_norm": 0.9874423742294312,
      "learning_rate": 0.0002308645512652736,
      "loss": 2.0157,
      "step": 132256
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.2106777429580688,
      "learning_rate": 0.00023086057081186297,
      "loss": 1.9413,
      "step": 132257
    },
    {
      "epoch": 1.72,
      "grad_norm": 0.9993432760238647,
      "learning_rate": 0.00023085659037130681,
      "loss": 2.1557,
      "step": 132258
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.1275253295898438,
      "learning_rate": 0.00023085260994360558,
      "loss": 2.2625,
      "step": 132259
    },
    {
      "epoch": 1.72,
      "grad_norm": 0.9353761672973633,
      "learning_rate": 0.00023084862952876014,
      "loss": 2.0416,
      "step": 132260
    },
    {
      "epoch": 1.72,
      "grad_norm": 0.9861183762550354,
      "learning_rate": 0.00023084464912677126,
      "loss": 1.9323,
      "step": 132261
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.1626580953598022,
      "learning_rate": 0.00023084066873763968,
      "loss": 2.1549,
      "step": 132262
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.0135645866394043,
      "learning_rate": 0.0002308366883613661,
      "loss": 2.1228,
      "step": 132263
    },
    {
      "epoch": 1.72,
      "grad_norm": 0.9885865449905396,
      "learning_rate": 0.0002308327079979514,
      "loss": 2.1938,
      "step": 132264
    },
    {
      "epoch": 1.72,
      "grad_norm": 0.9707038402557373,
      "learning_rate": 0.00023082872764739612,
      "loss": 1.9218,
      "step": 132265
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.1303216218948364,
      "learning_rate": 0.0002308247473097011,
      "loss": 2.1218,
      "step": 132266
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.1085752248764038,
      "learning_rate": 0.000230820766984867,
      "loss": 2.1681,
      "step": 132267
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.0636816024780273,
      "learning_rate": 0.00023081678667289464,
      "loss": 1.9139,
      "step": 132268
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.0248732566833496,
      "learning_rate": 0.00023081280637378477,
      "loss": 1.9738,
      "step": 132269
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.0362038612365723,
      "learning_rate": 0.00023080882608753824,
      "loss": 1.9988,
      "step": 132270
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.0281343460083008,
      "learning_rate": 0.00023080484581415548,
      "loss": 2.1362,
      "step": 132271
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.0395277738571167,
      "learning_rate": 0.00023080086555363744,
      "loss": 1.8117,
      "step": 132272
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.058725357055664,
      "learning_rate": 0.00023079688530598484,
      "loss": 2.0033,
      "step": 132273
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.1468945741653442,
      "learning_rate": 0.00023079290507119836,
      "loss": 2.0331,
      "step": 132274
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.03093421459198,
      "learning_rate": 0.0002307889248492788,
      "loss": 2.0017,
      "step": 132275
    },
    {
      "epoch": 1.72,
      "grad_norm": 0.9997574090957642,
      "learning_rate": 0.00023078494464022703,
      "loss": 2.0997,
      "step": 132276
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.0392169952392578,
      "learning_rate": 0.00023078096444404354,
      "loss": 2.2234,
      "step": 132277
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.3633514642715454,
      "learning_rate": 0.00023077698426072913,
      "loss": 1.9277,
      "step": 132278
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.324120283126831,
      "learning_rate": 0.0002307730040902846,
      "loss": 2.0909,
      "step": 132279
    },
    {
      "epoch": 1.72,
      "grad_norm": 0.9255955815315247,
      "learning_rate": 0.00023076902393271068,
      "loss": 2.2013,
      "step": 132280
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.045905590057373,
      "learning_rate": 0.00023076504378800812,
      "loss": 1.9059,
      "step": 132281
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.1488919258117676,
      "learning_rate": 0.00023076106365617775,
      "loss": 2.2921,
      "step": 132282
    },
    {
      "epoch": 1.72,
      "grad_norm": 0.9976855516433716,
      "learning_rate": 0.00023075708353722006,
      "loss": 2.1537,
      "step": 132283
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.0576403141021729,
      "learning_rate": 0.000230753103431136,
      "loss": 2.1821,
      "step": 132284
    },
    {
      "epoch": 1.72,
      "grad_norm": 0.9881801009178162,
      "learning_rate": 0.00023074912333792617,
      "loss": 2.0846,
      "step": 132285
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.010437250137329,
      "learning_rate": 0.00023074514325759145,
      "loss": 1.9876,
      "step": 132286
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.106259822845459,
      "learning_rate": 0.00023074116319013246,
      "loss": 2.0937,
      "step": 132287
    },
    {
      "epoch": 1.72,
      "grad_norm": 0.9861741662025452,
      "learning_rate": 0.00023073718313555008,
      "loss": 2.0875,
      "step": 132288
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.0187673568725586,
      "learning_rate": 0.0002307332030938449,
      "loss": 2.0492,
      "step": 132289
    },
    {
      "epoch": 1.72,
      "grad_norm": 0.9335952997207642,
      "learning_rate": 0.0002307292230650178,
      "loss": 2.2849,
      "step": 132290
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.066868543624878,
      "learning_rate": 0.00023072524304906938,
      "loss": 2.1506,
      "step": 132291
    },
    {
      "epoch": 1.72,
      "grad_norm": 0.9632095098495483,
      "learning_rate": 0.00023072126304600048,
      "loss": 1.9799,
      "step": 132292
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.1753085851669312,
      "learning_rate": 0.00023071728305581172,
      "loss": 2.2879,
      "step": 132293
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.0432438850402832,
      "learning_rate": 0.00023071330307850406,
      "loss": 2.1119,
      "step": 132294
    },
    {
      "epoch": 1.72,
      "grad_norm": 0.9709653854370117,
      "learning_rate": 0.000230709323114078,
      "loss": 2.0604,
      "step": 132295
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.0237914323806763,
      "learning_rate": 0.0002307053431625344,
      "loss": 2.2087,
      "step": 132296
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.0277230739593506,
      "learning_rate": 0.00023070136322387406,
      "loss": 2.1049,
      "step": 132297
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.1071174144744873,
      "learning_rate": 0.00023069738329809756,
      "loss": 2.042,
      "step": 132298
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.0002589225769043,
      "learning_rate": 0.00023069340338520574,
      "loss": 2.059,
      "step": 132299
    },
    {
      "epoch": 1.72,
      "grad_norm": 0.9153516292572021,
      "learning_rate": 0.0002306894234851994,
      "loss": 1.9197,
      "step": 132300
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.1820508241653442,
      "learning_rate": 0.0002306854435980791,
      "loss": 2.3241,
      "step": 132301
    },
    {
      "epoch": 1.72,
      "grad_norm": 0.9903241991996765,
      "learning_rate": 0.00023068146372384572,
      "loss": 1.9513,
      "step": 132302
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.0140841007232666,
      "learning_rate": 0.00023067748386249994,
      "loss": 1.9449,
      "step": 132303
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.1827881336212158,
      "learning_rate": 0.0002306735040140426,
      "loss": 2.1547,
      "step": 132304
    },
    {
      "epoch": 1.72,
      "grad_norm": 0.9912883043289185,
      "learning_rate": 0.0002306695241784743,
      "loss": 1.9836,
      "step": 132305
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.2668266296386719,
      "learning_rate": 0.0002306655443557959,
      "loss": 2.1435,
      "step": 132306
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.0579428672790527,
      "learning_rate": 0.00023066156454600804,
      "loss": 1.946,
      "step": 132307
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.0884044170379639,
      "learning_rate": 0.00023065758474911146,
      "loss": 1.9702,
      "step": 132308
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.042768120765686,
      "learning_rate": 0.000230653604965107,
      "loss": 1.9197,
      "step": 132309
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.0036683082580566,
      "learning_rate": 0.00023064962519399532,
      "loss": 2.1397,
      "step": 132310
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.080275058746338,
      "learning_rate": 0.00023064564543577725,
      "loss": 2.1296,
      "step": 132311
    },
    {
      "epoch": 1.72,
      "grad_norm": 0.9707111120223999,
      "learning_rate": 0.00023064166569045346,
      "loss": 2.1098,
      "step": 132312
    },
    {
      "epoch": 1.72,
      "grad_norm": 0.9580884575843811,
      "learning_rate": 0.00023063768595802463,
      "loss": 2.1106,
      "step": 132313
    },
    {
      "epoch": 1.72,
      "grad_norm": 0.9691876769065857,
      "learning_rate": 0.00023063370623849157,
      "loss": 2.2518,
      "step": 132314
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.0727213621139526,
      "learning_rate": 0.00023062972653185499,
      "loss": 1.9801,
      "step": 132315
    },
    {
      "epoch": 1.72,
      "grad_norm": 0.972436249256134,
      "learning_rate": 0.00023062574683811566,
      "loss": 2.1129,
      "step": 132316
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.1090861558914185,
      "learning_rate": 0.00023062176715727434,
      "loss": 2.1762,
      "step": 132317
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.004680871963501,
      "learning_rate": 0.00023061778748933176,
      "loss": 1.9658,
      "step": 132318
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.1130295991897583,
      "learning_rate": 0.0002306138078342887,
      "loss": 2.171,
      "step": 132319
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.1213258504867554,
      "learning_rate": 0.00023060982819214575,
      "loss": 2.3277,
      "step": 132320
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.0133129358291626,
      "learning_rate": 0.00023060584856290377,
      "loss": 2.1259,
      "step": 132321
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.1841456890106201,
      "learning_rate": 0.00023060186894656345,
      "loss": 1.9539,
      "step": 132322
    },
    {
      "epoch": 1.72,
      "grad_norm": 0.953983724117279,
      "learning_rate": 0.00023059788934312557,
      "loss": 2.0404,
      "step": 132323
    },
    {
      "epoch": 1.72,
      "grad_norm": 0.9177192449569702,
      "learning_rate": 0.0002305939097525908,
      "loss": 1.925,
      "step": 132324
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.2855256795883179,
      "learning_rate": 0.00023058993017496013,
      "loss": 2.1533,
      "step": 132325
    },
    {
      "epoch": 1.72,
      "grad_norm": 0.972585141658783,
      "learning_rate": 0.00023058595061023392,
      "loss": 1.8425,
      "step": 132326
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.0282435417175293,
      "learning_rate": 0.00023058197105841312,
      "loss": 2.0794,
      "step": 132327
    },
    {
      "epoch": 1.72,
      "grad_norm": 0.9440656304359436,
      "learning_rate": 0.0002305779915194984,
      "loss": 2.1121,
      "step": 132328
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.0058196783065796,
      "learning_rate": 0.00023057401199349058,
      "loss": 1.9562,
      "step": 132329
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.02314031124115,
      "learning_rate": 0.00023057003248039035,
      "loss": 2.1812,
      "step": 132330
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.0325838327407837,
      "learning_rate": 0.0002305660529801986,
      "loss": 2.211,
      "step": 132331
    },
    {
      "epoch": 1.72,
      "grad_norm": 0.8945141434669495,
      "learning_rate": 0.0002305620734929158,
      "loss": 2.0362,
      "step": 132332
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.011547327041626,
      "learning_rate": 0.00023055809401854283,
      "loss": 2.2347,
      "step": 132333
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.419981837272644,
      "learning_rate": 0.0002305541145570804,
      "loss": 2.0946,
      "step": 132334
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.0309113264083862,
      "learning_rate": 0.0002305501351085293,
      "loss": 2.0869,
      "step": 132335
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.0732088088989258,
      "learning_rate": 0.00023054615567289023,
      "loss": 2.1763,
      "step": 132336
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.2224245071411133,
      "learning_rate": 0.00023054217625016404,
      "loss": 2.0941,
      "step": 132337
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.0922205448150635,
      "learning_rate": 0.00023053819684035128,
      "loss": 2.0393,
      "step": 132338
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.1352485418319702,
      "learning_rate": 0.00023053421744345277,
      "loss": 2.1383,
      "step": 132339
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.0215104818344116,
      "learning_rate": 0.00023053023805946925,
      "loss": 2.037,
      "step": 132340
    },
    {
      "epoch": 1.72,
      "grad_norm": 0.9471047520637512,
      "learning_rate": 0.00023052625868840145,
      "loss": 1.9981,
      "step": 132341
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.019775629043579,
      "learning_rate": 0.0002305222793302502,
      "loss": 2.0427,
      "step": 132342
    },
    {
      "epoch": 1.72,
      "grad_norm": 0.9849345088005066,
      "learning_rate": 0.00023051829998501622,
      "loss": 2.2052,
      "step": 132343
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.0057718753814697,
      "learning_rate": 0.00023051432065270013,
      "loss": 2.0225,
      "step": 132344
    },
    {
      "epoch": 1.72,
      "grad_norm": 0.9919907450675964,
      "learning_rate": 0.0002305103413333027,
      "loss": 2.2608,
      "step": 132345
    },
    {
      "epoch": 1.72,
      "grad_norm": 0.9508578181266785,
      "learning_rate": 0.00023050636202682473,
      "loss": 2.2078,
      "step": 132346
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.0727777481079102,
      "learning_rate": 0.00023050238273326692,
      "loss": 2.1556,
      "step": 132347
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.0219018459320068,
      "learning_rate": 0.00023049840345263004,
      "loss": 2.1162,
      "step": 132348
    },
    {
      "epoch": 1.72,
      "grad_norm": 0.977051854133606,
      "learning_rate": 0.00023049442418491493,
      "loss": 2.0794,
      "step": 132349
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.0319404602050781,
      "learning_rate": 0.00023049044493012212,
      "loss": 2.1538,
      "step": 132350
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.1660224199295044,
      "learning_rate": 0.00023048646568825243,
      "loss": 2.1051,
      "step": 132351
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.0268833637237549,
      "learning_rate": 0.00023048248645930663,
      "loss": 2.2366,
      "step": 132352
    },
    {
      "epoch": 1.72,
      "grad_norm": 0.9964733719825745,
      "learning_rate": 0.00023047850724328543,
      "loss": 2.2722,
      "step": 132353
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.0436707735061646,
      "learning_rate": 0.0002304745280401896,
      "loss": 2.0542,
      "step": 132354
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.030963659286499,
      "learning_rate": 0.00023047054885002002,
      "loss": 2.1053,
      "step": 132355
    },
    {
      "epoch": 1.72,
      "grad_norm": 0.9856655597686768,
      "learning_rate": 0.0002304665696727771,
      "loss": 2.3235,
      "step": 132356
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.0717982053756714,
      "learning_rate": 0.00023046259050846178,
      "loss": 1.9972,
      "step": 132357
    },
    {
      "epoch": 1.72,
      "grad_norm": 0.9946044087409973,
      "learning_rate": 0.0002304586113570748,
      "loss": 2.1064,
      "step": 132358
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.0603505373001099,
      "learning_rate": 0.00023045463221861687,
      "loss": 2.1742,
      "step": 132359
    },
    {
      "epoch": 1.72,
      "grad_norm": 0.9900418519973755,
      "learning_rate": 0.00023045065309308871,
      "loss": 2.1174,
      "step": 132360
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.000657558441162,
      "learning_rate": 0.00023044667398049124,
      "loss": 2.1836,
      "step": 132361
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.2375094890594482,
      "learning_rate": 0.00023044269488082492,
      "loss": 1.9985,
      "step": 132362
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.1261018514633179,
      "learning_rate": 0.00023043871579409058,
      "loss": 2.1361,
      "step": 132363
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.044947624206543,
      "learning_rate": 0.000230434736720289,
      "loss": 2.0543,
      "step": 132364
    },
    {
      "epoch": 1.72,
      "grad_norm": 0.9846084713935852,
      "learning_rate": 0.00023043075765942096,
      "loss": 2.1908,
      "step": 132365
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.0647777318954468,
      "learning_rate": 0.0002304267786114871,
      "loss": 2.0775,
      "step": 132366
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.0441725254058838,
      "learning_rate": 0.00023042279957648842,
      "loss": 2.0522,
      "step": 132367
    },
    {
      "epoch": 1.72,
      "grad_norm": 0.9928150773048401,
      "learning_rate": 0.00023041882055442524,
      "loss": 1.9738,
      "step": 132368
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.035809874534607,
      "learning_rate": 0.00023041484154529858,
      "loss": 2.1556,
      "step": 132369
    },
    {
      "epoch": 1.72,
      "grad_norm": 0.9412217140197754,
      "learning_rate": 0.00023041086254910908,
      "loss": 2.1114,
      "step": 132370
    },
    {
      "epoch": 1.72,
      "grad_norm": 0.9136734008789062,
      "learning_rate": 0.0002304068835658575,
      "loss": 2.1517,
      "step": 132371
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.001431941986084,
      "learning_rate": 0.00023040290459554462,
      "loss": 2.3461,
      "step": 132372
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.09625244140625,
      "learning_rate": 0.0002303989256381712,
      "loss": 2.1964,
      "step": 132373
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.0283845663070679,
      "learning_rate": 0.00023039494669373797,
      "loss": 2.1244,
      "step": 132374
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.1397761106491089,
      "learning_rate": 0.00023039096776224555,
      "loss": 2.1538,
      "step": 132375
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.0593725442886353,
      "learning_rate": 0.00023038698884369477,
      "loss": 1.851,
      "step": 132376
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.2331290245056152,
      "learning_rate": 0.00023038300993808637,
      "loss": 1.7398,
      "step": 132377
    },
    {
      "epoch": 1.72,
      "grad_norm": 0.944960355758667,
      "learning_rate": 0.00023037903104542105,
      "loss": 2.048,
      "step": 132378
    },
    {
      "epoch": 1.72,
      "grad_norm": 0.9578981995582581,
      "learning_rate": 0.00023037505216569966,
      "loss": 2.0001,
      "step": 132379
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.1213634014129639,
      "learning_rate": 0.00023037107329892278,
      "loss": 1.8815,
      "step": 132380
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.4471834897994995,
      "learning_rate": 0.00023036709444509125,
      "loss": 1.9001,
      "step": 132381
    },
    {
      "epoch": 1.72,
      "grad_norm": 0.9877749681472778,
      "learning_rate": 0.00023036311560420587,
      "loss": 2.1181,
      "step": 132382
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.0839923620224,
      "learning_rate": 0.00023035913677626723,
      "loss": 1.985,
      "step": 132383
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.6084091663360596,
      "learning_rate": 0.00023035515796127615,
      "loss": 2.1587,
      "step": 132384
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.0612962245941162,
      "learning_rate": 0.00023035117915923335,
      "loss": 2.2568,
      "step": 132385
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.0865459442138672,
      "learning_rate": 0.0002303472003701396,
      "loss": 1.9425,
      "step": 132386
    },
    {
      "epoch": 1.72,
      "grad_norm": 0.9964405298233032,
      "learning_rate": 0.00023034322159399564,
      "loss": 1.8547,
      "step": 132387
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.0865379571914673,
      "learning_rate": 0.00023033924283080213,
      "loss": 2.1937,
      "step": 132388
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.0524778366088867,
      "learning_rate": 0.00023033526408055993,
      "loss": 1.9655,
      "step": 132389
    },
    {
      "epoch": 1.72,
      "grad_norm": 0.9948480725288391,
      "learning_rate": 0.00023033128534326967,
      "loss": 2.065,
      "step": 132390
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.0141428709030151,
      "learning_rate": 0.00023032730661893218,
      "loss": 1.9233,
      "step": 132391
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.0252301692962646,
      "learning_rate": 0.0002303233279075482,
      "loss": 2.2599,
      "step": 132392
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.1184102296829224,
      "learning_rate": 0.00023031934920911834,
      "loss": 2.1282,
      "step": 132393
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.221988320350647,
      "learning_rate": 0.00023031537052364343,
      "loss": 1.9642,
      "step": 132394
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.1055846214294434,
      "learning_rate": 0.00023031139185112422,
      "loss": 2.2465,
      "step": 132395
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.0544685125350952,
      "learning_rate": 0.0002303074131915615,
      "loss": 2.1109,
      "step": 132396
    },
    {
      "epoch": 1.72,
      "grad_norm": 0.9616641998291016,
      "learning_rate": 0.0002303034345449559,
      "loss": 2.08,
      "step": 132397
    },
    {
      "epoch": 1.72,
      "grad_norm": 0.9981647729873657,
      "learning_rate": 0.00023029945591130822,
      "loss": 2.2226,
      "step": 132398
    },
    {
      "epoch": 1.72,
      "grad_norm": 0.9758670330047607,
      "learning_rate": 0.00023029547729061918,
      "loss": 2.1046,
      "step": 132399
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.022023320198059,
      "learning_rate": 0.00023029149868288953,
      "loss": 2.1165,
      "step": 132400
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.1914114952087402,
      "learning_rate": 0.00023028752008811995,
      "loss": 2.1955,
      "step": 132401
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.0308424234390259,
      "learning_rate": 0.00023028354150631127,
      "loss": 1.9313,
      "step": 132402
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.047707200050354,
      "learning_rate": 0.00023027956293746426,
      "loss": 2.0818,
      "step": 132403
    },
    {
      "epoch": 1.72,
      "grad_norm": 0.9984890818595886,
      "learning_rate": 0.00023027558438157959,
      "loss": 2.0079,
      "step": 132404
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.0138769149780273,
      "learning_rate": 0.00023027160583865796,
      "loss": 2.1591,
      "step": 132405
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.0402026176452637,
      "learning_rate": 0.00023026762730870014,
      "loss": 2.2611,
      "step": 132406
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.005879282951355,
      "learning_rate": 0.00023026364879170688,
      "loss": 2.1096,
      "step": 132407
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.1530356407165527,
      "learning_rate": 0.00023025967028767892,
      "loss": 2.1172,
      "step": 132408
    },
    {
      "epoch": 1.72,
      "grad_norm": 0.994880735874176,
      "learning_rate": 0.00023025569179661702,
      "loss": 2.1242,
      "step": 132409
    },
    {
      "epoch": 1.72,
      "grad_norm": 0.960905134677887,
      "learning_rate": 0.00023025171331852204,
      "loss": 2.2675,
      "step": 132410
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.298866629600525,
      "learning_rate": 0.0002302477348533944,
      "loss": 2.108,
      "step": 132411
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.0598276853561401,
      "learning_rate": 0.00023024375640123506,
      "loss": 2.2094,
      "step": 132412
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.0044479370117188,
      "learning_rate": 0.00023023977796204472,
      "loss": 2.1994,
      "step": 132413
    },
    {
      "epoch": 1.72,
      "grad_norm": 0.982248067855835,
      "learning_rate": 0.00023023579953582412,
      "loss": 2.1418,
      "step": 132414
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.003542423248291,
      "learning_rate": 0.000230231821122574,
      "loss": 2.3604,
      "step": 132415
    },
    {
      "epoch": 1.72,
      "grad_norm": 0.9841529726982117,
      "learning_rate": 0.00023022784272229527,
      "loss": 1.9824,
      "step": 132416
    },
    {
      "epoch": 1.72,
      "grad_norm": 0.9860551953315735,
      "learning_rate": 0.00023022386433498834,
      "loss": 2.0172,
      "step": 132417
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.1575971841812134,
      "learning_rate": 0.0002302198859606541,
      "loss": 2.1823,
      "step": 132418
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.054733395576477,
      "learning_rate": 0.0002302159075992933,
      "loss": 2.1229,
      "step": 132419
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.163964867591858,
      "learning_rate": 0.00023021192925090671,
      "loss": 2.3593,
      "step": 132420
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.0889865159988403,
      "learning_rate": 0.000230207950915495,
      "loss": 2.0171,
      "step": 132421
    },
    {
      "epoch": 1.72,
      "grad_norm": 0.9428768157958984,
      "learning_rate": 0.00023020397259305913,
      "loss": 2.073,
      "step": 132422
    },
    {
      "epoch": 1.72,
      "grad_norm": 0.999857485294342,
      "learning_rate": 0.00023019999428359946,
      "loss": 2.2417,
      "step": 132423
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.029654622077942,
      "learning_rate": 0.000230196015987117,
      "loss": 2.3616,
      "step": 132424
    },
    {
      "epoch": 1.72,
      "grad_norm": 0.9417200684547424,
      "learning_rate": 0.00023019203770361236,
      "loss": 2.1315,
      "step": 132425
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.1244559288024902,
      "learning_rate": 0.00023018805943308634,
      "loss": 1.996,
      "step": 132426
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.1178051233291626,
      "learning_rate": 0.00023018408117553974,
      "loss": 2.0924,
      "step": 132427
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.3163974285125732,
      "learning_rate": 0.00023018010293097328,
      "loss": 2.2111,
      "step": 132428
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.0849542617797852,
      "learning_rate": 0.0002301761246993876,
      "loss": 2.1946,
      "step": 132429
    },
    {
      "epoch": 1.72,
      "grad_norm": 0.9467780590057373,
      "learning_rate": 0.00023017214648078345,
      "loss": 1.9982,
      "step": 132430
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.311526894569397,
      "learning_rate": 0.00023016816827516168,
      "loss": 1.9241,
      "step": 132431
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.1721885204315186,
      "learning_rate": 0.0002301641900825229,
      "loss": 2.0678,
      "step": 132432
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.100260853767395,
      "learning_rate": 0.00023016021190286793,
      "loss": 1.8349,
      "step": 132433
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.0387077331542969,
      "learning_rate": 0.00023015623373619766,
      "loss": 2.1099,
      "step": 132434
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.0618304014205933,
      "learning_rate": 0.00023015225558251247,
      "loss": 2.164,
      "step": 132435
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.1471753120422363,
      "learning_rate": 0.00023014827744181334,
      "loss": 1.9268,
      "step": 132436
    },
    {
      "epoch": 1.72,
      "grad_norm": 0.9449639320373535,
      "learning_rate": 0.00023014429931410095,
      "loss": 2.2438,
      "step": 132437
    },
    {
      "epoch": 1.72,
      "grad_norm": 0.9159871935844421,
      "learning_rate": 0.00023014032119937609,
      "loss": 2.0826,
      "step": 132438
    },
    {
      "epoch": 1.72,
      "grad_norm": 0.9777209758758545,
      "learning_rate": 0.00023013634309763942,
      "loss": 1.9537,
      "step": 132439
    },
    {
      "epoch": 1.72,
      "grad_norm": 0.9694527983665466,
      "learning_rate": 0.00023013236500889184,
      "loss": 1.9451,
      "step": 132440
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.0770515203475952,
      "learning_rate": 0.00023012838693313388,
      "loss": 2.2681,
      "step": 132441
    },
    {
      "epoch": 1.72,
      "grad_norm": 0.9987139105796814,
      "learning_rate": 0.00023012440887036634,
      "loss": 2.241,
      "step": 132442
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.0325521230697632,
      "learning_rate": 0.00023012043082059003,
      "loss": 1.9166,
      "step": 132443
    },
    {
      "epoch": 1.72,
      "grad_norm": 0.9696842432022095,
      "learning_rate": 0.00023011645278380564,
      "loss": 2.1858,
      "step": 132444
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.0478628873825073,
      "learning_rate": 0.0002301124747600139,
      "loss": 2.2472,
      "step": 132445
    },
    {
      "epoch": 1.72,
      "grad_norm": 0.9703658819198608,
      "learning_rate": 0.0002301084967492157,
      "loss": 2.0574,
      "step": 132446
    },
    {
      "epoch": 1.72,
      "grad_norm": 0.9175420999526978,
      "learning_rate": 0.00023010451875141153,
      "loss": 2.0713,
      "step": 132447
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.1512149572372437,
      "learning_rate": 0.00023010054076660224,
      "loss": 2.1264,
      "step": 132448
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.0258972644805908,
      "learning_rate": 0.0002300965627947886,
      "loss": 2.2941,
      "step": 132449
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.209807276725769,
      "learning_rate": 0.0002300925848359713,
      "loss": 2.1244,
      "step": 132450
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.2225286960601807,
      "learning_rate": 0.00023008860689015113,
      "loss": 2.1824,
      "step": 132451
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.0482769012451172,
      "learning_rate": 0.00023008462895732882,
      "loss": 2.3814,
      "step": 132452
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.1319512128829956,
      "learning_rate": 0.0002300806510375052,
      "loss": 2.1251,
      "step": 132453
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.0157110691070557,
      "learning_rate": 0.0002300766731306808,
      "loss": 2.31,
      "step": 132454
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.254183292388916,
      "learning_rate": 0.00023007269523685645,
      "loss": 2.1532,
      "step": 132455
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.18246328830719,
      "learning_rate": 0.0002300687173560329,
      "loss": 2.0841,
      "step": 132456
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.080541968345642,
      "learning_rate": 0.0002300647394882109,
      "loss": 2.1997,
      "step": 132457
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.0966341495513916,
      "learning_rate": 0.0002300607616333912,
      "loss": 2.2076,
      "step": 132458
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.019660234451294,
      "learning_rate": 0.00023005678379157463,
      "loss": 2.1515,
      "step": 132459
    },
    {
      "epoch": 1.72,
      "grad_norm": 0.9932013750076294,
      "learning_rate": 0.00023005280596276174,
      "loss": 2.2049,
      "step": 132460
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.0663419961929321,
      "learning_rate": 0.00023004882814695328,
      "loss": 2.0864,
      "step": 132461
    },
    {
      "epoch": 1.72,
      "grad_norm": 0.9759419560432434,
      "learning_rate": 0.00023004485034415015,
      "loss": 1.9499,
      "step": 132462
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.4092750549316406,
      "learning_rate": 0.00023004087255435292,
      "loss": 2.1392,
      "step": 132463
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.1110262870788574,
      "learning_rate": 0.00023003689477756249,
      "loss": 2.2683,
      "step": 132464
    },
    {
      "epoch": 1.72,
      "grad_norm": 0.948252260684967,
      "learning_rate": 0.00023003291701377955,
      "loss": 2.1686,
      "step": 132465
    },
    {
      "epoch": 1.72,
      "grad_norm": 0.9980029463768005,
      "learning_rate": 0.0002300289392630048,
      "loss": 2.3319,
      "step": 132466
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.005110740661621,
      "learning_rate": 0.0002300249615252389,
      "loss": 2.2802,
      "step": 132467
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.1685657501220703,
      "learning_rate": 0.00023002098380048273,
      "loss": 2.2004,
      "step": 132468
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.3786038160324097,
      "learning_rate": 0.000230017006088737,
      "loss": 2.2559,
      "step": 132469
    },
    {
      "epoch": 1.72,
      "grad_norm": 0.9447077512741089,
      "learning_rate": 0.00023001302839000237,
      "loss": 1.9516,
      "step": 132470
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.044673204421997,
      "learning_rate": 0.00023000905070427974,
      "loss": 2.038,
      "step": 132471
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.1942626237869263,
      "learning_rate": 0.00023000507303156965,
      "loss": 2.1508,
      "step": 132472
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.1220711469650269,
      "learning_rate": 0.00023000109537187297,
      "loss": 2.0823,
      "step": 132473
    },
    {
      "epoch": 1.72,
      "grad_norm": 0.9706985354423523,
      "learning_rate": 0.00022999711772519048,
      "loss": 1.9925,
      "step": 132474
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.0341832637786865,
      "learning_rate": 0.00022999314009152275,
      "loss": 2.2603,
      "step": 132475
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.2649179697036743,
      "learning_rate": 0.00022998916247087066,
      "loss": 2.2787,
      "step": 132476
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.0574687719345093,
      "learning_rate": 0.00022998518486323492,
      "loss": 2.2475,
      "step": 132477
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.1211066246032715,
      "learning_rate": 0.00022998120726861623,
      "loss": 2.061,
      "step": 132478
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.180666208267212,
      "learning_rate": 0.00022997722968701536,
      "loss": 2.2821,
      "step": 132479
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.036047101020813,
      "learning_rate": 0.00022997325211843297,
      "loss": 2.2616,
      "step": 132480
    },
    {
      "epoch": 1.72,
      "grad_norm": 1.0811537504196167,
      "learning_rate": 0.00022996927456287003,
      "loss": 2.2032,
      "step": 132481
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.0157108306884766,
      "learning_rate": 0.000229965297020327,
      "loss": 1.9393,
      "step": 132482
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.0177087783813477,
      "learning_rate": 0.00022996131949080484,
      "loss": 1.8177,
      "step": 132483
    },
    {
      "epoch": 1.73,
      "grad_norm": 0.9589515924453735,
      "learning_rate": 0.00022995734197430413,
      "loss": 1.9798,
      "step": 132484
    },
    {
      "epoch": 1.73,
      "grad_norm": 0.9984920024871826,
      "learning_rate": 0.00022995336447082564,
      "loss": 2.2159,
      "step": 132485
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.0379490852355957,
      "learning_rate": 0.00022994938698037018,
      "loss": 2.1403,
      "step": 132486
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.1414090394973755,
      "learning_rate": 0.00022994540950293843,
      "loss": 2.1887,
      "step": 132487
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.0140107870101929,
      "learning_rate": 0.00022994143203853123,
      "loss": 2.1718,
      "step": 132488
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.170569658279419,
      "learning_rate": 0.00022993745458714925,
      "loss": 2.3689,
      "step": 132489
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.0775954723358154,
      "learning_rate": 0.00022993347714879314,
      "loss": 2.1221,
      "step": 132490
    },
    {
      "epoch": 1.73,
      "grad_norm": 0.9575409293174744,
      "learning_rate": 0.0002299294997234637,
      "loss": 2.2419,
      "step": 132491
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.0535187721252441,
      "learning_rate": 0.00022992552231116168,
      "loss": 2.1618,
      "step": 132492
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.0099629163742065,
      "learning_rate": 0.00022992154491188787,
      "loss": 2.0137,
      "step": 132493
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.0726358890533447,
      "learning_rate": 0.00022991756752564294,
      "loss": 2.1424,
      "step": 132494
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.0833697319030762,
      "learning_rate": 0.0002299135901524278,
      "loss": 2.2251,
      "step": 132495
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.0399155616760254,
      "learning_rate": 0.0002299096127922429,
      "loss": 2.2076,
      "step": 132496
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.027733325958252,
      "learning_rate": 0.00022990563544508915,
      "loss": 2.223,
      "step": 132497
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.1625550985336304,
      "learning_rate": 0.00022990165811096725,
      "loss": 2.2559,
      "step": 132498
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.0870915651321411,
      "learning_rate": 0.00022989768078987796,
      "loss": 2.1749,
      "step": 132499
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.106859803199768,
      "learning_rate": 0.00022989370348182203,
      "loss": 2.0654,
      "step": 132500
    },
    {
      "epoch": 1.73,
      "grad_norm": 0.967276394367218,
      "learning_rate": 0.0002298897261868003,
      "loss": 1.9683,
      "step": 132501
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.238142490386963,
      "learning_rate": 0.00022988574890481323,
      "loss": 1.8845,
      "step": 132502
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.5399373769760132,
      "learning_rate": 0.0002298817716358618,
      "loss": 2.0812,
      "step": 132503
    },
    {
      "epoch": 1.73,
      "grad_norm": 0.9772511720657349,
      "learning_rate": 0.00022987779437994658,
      "loss": 2.1258,
      "step": 132504
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.1155850887298584,
      "learning_rate": 0.00022987381713706847,
      "loss": 2.264,
      "step": 132505
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.504215955734253,
      "learning_rate": 0.0002298698399072281,
      "loss": 2.3041,
      "step": 132506
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.1428308486938477,
      "learning_rate": 0.00022986586269042642,
      "loss": 1.8731,
      "step": 132507
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.0008662939071655,
      "learning_rate": 0.00022986188548666385,
      "loss": 2.2068,
      "step": 132508
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.0872175693511963,
      "learning_rate": 0.00022985790829594129,
      "loss": 2.231,
      "step": 132509
    },
    {
      "epoch": 1.73,
      "grad_norm": 2.1279537677764893,
      "learning_rate": 0.00022985393111825942,
      "loss": 2.1516,
      "step": 132510
    },
    {
      "epoch": 1.73,
      "grad_norm": 2.201972723007202,
      "learning_rate": 0.00022984995395361908,
      "loss": 2.2644,
      "step": 132511
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.3359602689743042,
      "learning_rate": 0.00022984597680202095,
      "loss": 2.0848,
      "step": 132512
    },
    {
      "epoch": 1.73,
      "grad_norm": 0.9449976682662964,
      "learning_rate": 0.00022984199966346588,
      "loss": 1.9923,
      "step": 132513
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.2649281024932861,
      "learning_rate": 0.0002298380225379544,
      "loss": 1.9475,
      "step": 132514
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.0565916299819946,
      "learning_rate": 0.00022983404542548737,
      "loss": 2.059,
      "step": 132515
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.2188565731048584,
      "learning_rate": 0.0002298300683260655,
      "loss": 2.1332,
      "step": 132516
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.3650274276733398,
      "learning_rate": 0.00022982609123968956,
      "loss": 2.0427,
      "step": 132517
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.0208488702774048,
      "learning_rate": 0.00022982211416636023,
      "loss": 1.9732,
      "step": 132518
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.0754175186157227,
      "learning_rate": 0.00022981813710607832,
      "loss": 2.0211,
      "step": 132519
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.7618916034698486,
      "learning_rate": 0.00022981416005884468,
      "loss": 2.0898,
      "step": 132520
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.3929133415222168,
      "learning_rate": 0.00022981018302465977,
      "loss": 2.1961,
      "step": 132521
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.0545612573623657,
      "learning_rate": 0.0002298062060035245,
      "loss": 2.0519,
      "step": 132522
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.0418412685394287,
      "learning_rate": 0.00022980222899543953,
      "loss": 2.2039,
      "step": 132523
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.0552321672439575,
      "learning_rate": 0.0002297982520004057,
      "loss": 2.1084,
      "step": 132524
    },
    {
      "epoch": 1.73,
      "grad_norm": 0.9832080602645874,
      "learning_rate": 0.00022979427501842369,
      "loss": 2.0048,
      "step": 132525
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.096629023551941,
      "learning_rate": 0.00022979029804949438,
      "loss": 2.1838,
      "step": 132526
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.050450086593628,
      "learning_rate": 0.00022978632109361822,
      "loss": 2.1413,
      "step": 132527
    },
    {
      "epoch": 1.73,
      "grad_norm": 0.9833407998085022,
      "learning_rate": 0.00022978234415079612,
      "loss": 2.0109,
      "step": 132528
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.2167550325393677,
      "learning_rate": 0.00022977836722102882,
      "loss": 2.0229,
      "step": 132529
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.0996732711791992,
      "learning_rate": 0.00022977439030431705,
      "loss": 2.1272,
      "step": 132530
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.0918878316879272,
      "learning_rate": 0.00022977041340066157,
      "loss": 2.1608,
      "step": 132531
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.069730281829834,
      "learning_rate": 0.00022976643651006316,
      "loss": 1.9849,
      "step": 132532
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.34456467628479,
      "learning_rate": 0.00022976245963252242,
      "loss": 2.2529,
      "step": 132533
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.295501708984375,
      "learning_rate": 0.0002297584827680401,
      "loss": 2.0486,
      "step": 132534
    },
    {
      "epoch": 1.73,
      "grad_norm": 0.9551419019699097,
      "learning_rate": 0.0002297545059166171,
      "loss": 2.4719,
      "step": 132535
    },
    {
      "epoch": 1.73,
      "grad_norm": 0.9397046566009521,
      "learning_rate": 0.000229750529078254,
      "loss": 1.9742,
      "step": 132536
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.1879780292510986,
      "learning_rate": 0.00022974655225295159,
      "loss": 1.994,
      "step": 132537
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.0703485012054443,
      "learning_rate": 0.0002297425754407108,
      "loss": 2.3564,
      "step": 132538
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.0091434717178345,
      "learning_rate": 0.00022973859864153202,
      "loss": 2.304,
      "step": 132539
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.1033275127410889,
      "learning_rate": 0.00022973462185541617,
      "loss": 2.1011,
      "step": 132540
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.0546239614486694,
      "learning_rate": 0.00022973064508236403,
      "loss": 2.0308,
      "step": 132541
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.0281721353530884,
      "learning_rate": 0.00022972666832237623,
      "loss": 2.1673,
      "step": 132542
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.0619062185287476,
      "learning_rate": 0.00022972269157545358,
      "loss": 1.9716,
      "step": 132543
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.008368968963623,
      "learning_rate": 0.00022971871484159695,
      "loss": 1.9999,
      "step": 132544
    },
    {
      "epoch": 1.73,
      "grad_norm": 0.977898895740509,
      "learning_rate": 0.00022971473812080683,
      "loss": 2.1614,
      "step": 132545
    },
    {
      "epoch": 1.73,
      "grad_norm": 0.9734022617340088,
      "learning_rate": 0.000229710761413084,
      "loss": 1.9862,
      "step": 132546
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.0187618732452393,
      "learning_rate": 0.00022970678471842933,
      "loss": 2.2359,
      "step": 132547
    },
    {
      "epoch": 1.73,
      "grad_norm": 0.9771406650543213,
      "learning_rate": 0.00022970280803684347,
      "loss": 2.1233,
      "step": 132548
    },
    {
      "epoch": 1.73,
      "grad_norm": 0.9958707690238953,
      "learning_rate": 0.0002296988313683272,
      "loss": 2.2051,
      "step": 132549
    },
    {
      "epoch": 1.73,
      "grad_norm": 0.974615216255188,
      "learning_rate": 0.0002296948547128813,
      "loss": 2.1161,
      "step": 132550
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.008518934249878,
      "learning_rate": 0.00022969087807050645,
      "loss": 1.9654,
      "step": 132551
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.0893518924713135,
      "learning_rate": 0.0002296869014412033,
      "loss": 2.0766,
      "step": 132552
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.1968778371810913,
      "learning_rate": 0.00022968292482497272,
      "loss": 1.9545,
      "step": 132553
    },
    {
      "epoch": 1.73,
      "grad_norm": 0.9409911036491394,
      "learning_rate": 0.0002296789482218154,
      "loss": 2.1143,
      "step": 132554
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.0102367401123047,
      "learning_rate": 0.0002296749716317321,
      "loss": 2.2118,
      "step": 132555
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.0132981538772583,
      "learning_rate": 0.0002296709950547236,
      "loss": 2.2069,
      "step": 132556
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.150607705116272,
      "learning_rate": 0.00022966701849079055,
      "loss": 2.4993,
      "step": 132557
    },
    {
      "epoch": 1.73,
      "grad_norm": 0.9966164827346802,
      "learning_rate": 0.00022966304193993374,
      "loss": 2.0977,
      "step": 132558
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.0254145860671997,
      "learning_rate": 0.00022965906540215388,
      "loss": 2.2631,
      "step": 132559
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.2846344709396362,
      "learning_rate": 0.00022965508887745173,
      "loss": 2.2308,
      "step": 132560
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.0599054098129272,
      "learning_rate": 0.00022965111236582803,
      "loss": 2.149,
      "step": 132561
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.0332403182983398,
      "learning_rate": 0.00022964713586728355,
      "loss": 2.155,
      "step": 132562
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.0686302185058594,
      "learning_rate": 0.00022964315938181896,
      "loss": 2.2316,
      "step": 132563
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.0372258424758911,
      "learning_rate": 0.00022963918290943498,
      "loss": 1.9602,
      "step": 132564
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.0996001958847046,
      "learning_rate": 0.00022963520645013255,
      "loss": 2.1666,
      "step": 132565
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.0125224590301514,
      "learning_rate": 0.00022963123000391215,
      "loss": 2.1833,
      "step": 132566
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.0339382886886597,
      "learning_rate": 0.00022962725357077464,
      "loss": 1.9454,
      "step": 132567
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.0873301029205322,
      "learning_rate": 0.0002296232771507208,
      "loss": 2.0821,
      "step": 132568
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.1156952381134033,
      "learning_rate": 0.0002296193007437513,
      "loss": 1.9804,
      "step": 132569
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.0341246128082275,
      "learning_rate": 0.00022961532434986686,
      "loss": 2.0456,
      "step": 132570
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.0047262907028198,
      "learning_rate": 0.00022961134796906828,
      "loss": 2.069,
      "step": 132571
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.0635358095169067,
      "learning_rate": 0.00022960737160135627,
      "loss": 2.0151,
      "step": 132572
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.053937554359436,
      "learning_rate": 0.00022960339524673163,
      "loss": 2.0581,
      "step": 132573
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.0553933382034302,
      "learning_rate": 0.00022959941890519507,
      "loss": 1.9976,
      "step": 132574
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.0620591640472412,
      "learning_rate": 0.00022959544257674725,
      "loss": 2.1347,
      "step": 132575
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.0868839025497437,
      "learning_rate": 0.00022959146626138895,
      "loss": 2.1732,
      "step": 132576
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.0570870637893677,
      "learning_rate": 0.00022958748995912092,
      "loss": 2.1738,
      "step": 132577
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.0444495677947998,
      "learning_rate": 0.00022958351366994393,
      "loss": 2.052,
      "step": 132578
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.0068206787109375,
      "learning_rate": 0.00022957953739385868,
      "loss": 2.0736,
      "step": 132579
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.0230060815811157,
      "learning_rate": 0.00022957556113086604,
      "loss": 2.2683,
      "step": 132580
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.0639134645462036,
      "learning_rate": 0.0002295715848809665,
      "loss": 1.9838,
      "step": 132581
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.034996747970581,
      "learning_rate": 0.00022956760864416097,
      "loss": 1.9125,
      "step": 132582
    },
    {
      "epoch": 1.73,
      "grad_norm": 0.9763895273208618,
      "learning_rate": 0.0002295636324204501,
      "loss": 2.1361,
      "step": 132583
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.0498082637786865,
      "learning_rate": 0.00022955965620983474,
      "loss": 2.1425,
      "step": 132584
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.0147804021835327,
      "learning_rate": 0.0002295556800123155,
      "loss": 1.9819,
      "step": 132585
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.1444607973098755,
      "learning_rate": 0.00022955170382789325,
      "loss": 2.291,
      "step": 132586
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.063564658164978,
      "learning_rate": 0.00022954772765656876,
      "loss": 1.9976,
      "step": 132587
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.001988172531128,
      "learning_rate": 0.00022954375149834255,
      "loss": 2.0545,
      "step": 132588
    },
    {
      "epoch": 1.73,
      "grad_norm": 0.9229787588119507,
      "learning_rate": 0.0002295397753532155,
      "loss": 2.0469,
      "step": 132589
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.0635404586791992,
      "learning_rate": 0.00022953579922118834,
      "loss": 2.1477,
      "step": 132590
    },
    {
      "epoch": 1.73,
      "grad_norm": 0.9871418476104736,
      "learning_rate": 0.0002295318231022618,
      "loss": 2.2363,
      "step": 132591
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.079079031944275,
      "learning_rate": 0.00022952784699643662,
      "loss": 2.1519,
      "step": 132592
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.1164448261260986,
      "learning_rate": 0.00022952387090371368,
      "loss": 1.9157,
      "step": 132593
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.088067889213562,
      "learning_rate": 0.00022951989482409343,
      "loss": 2.0741,
      "step": 132594
    },
    {
      "epoch": 1.73,
      "grad_norm": 0.9902175664901733,
      "learning_rate": 0.00022951591875757679,
      "loss": 1.9518,
      "step": 132595
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.0090479850769043,
      "learning_rate": 0.00022951194270416446,
      "loss": 2.0568,
      "step": 132596
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.2359050512313843,
      "learning_rate": 0.0002295079666638572,
      "loss": 2.187,
      "step": 132597
    },
    {
      "epoch": 1.73,
      "grad_norm": 0.9877941012382507,
      "learning_rate": 0.00022950399063665576,
      "loss": 2.4006,
      "step": 132598
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.0239132642745972,
      "learning_rate": 0.00022950001462256092,
      "loss": 2.2029,
      "step": 132599
    },
    {
      "epoch": 1.73,
      "grad_norm": 0.9642864465713501,
      "learning_rate": 0.00022949603862157326,
      "loss": 1.9095,
      "step": 132600
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.192535400390625,
      "learning_rate": 0.00022949206263369362,
      "loss": 2.0192,
      "step": 132601
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.040126085281372,
      "learning_rate": 0.00022948808665892276,
      "loss": 2.1075,
      "step": 132602
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.045570969581604,
      "learning_rate": 0.00022948411069726138,
      "loss": 2.1597,
      "step": 132603
    },
    {
      "epoch": 1.73,
      "grad_norm": 0.9991509318351746,
      "learning_rate": 0.00022948013474871021,
      "loss": 2.3409,
      "step": 132604
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.082789659500122,
      "learning_rate": 0.0002294761588132702,
      "loss": 2.039,
      "step": 132605
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.121720790863037,
      "learning_rate": 0.00022947218289094172,
      "loss": 2.1657,
      "step": 132606
    },
    {
      "epoch": 1.73,
      "grad_norm": 0.977061927318573,
      "learning_rate": 0.0002294682069817257,
      "loss": 2.1198,
      "step": 132607
    },
    {
      "epoch": 1.73,
      "grad_norm": 0.9933636784553528,
      "learning_rate": 0.00022946423108562288,
      "loss": 1.771,
      "step": 132608
    },
    {
      "epoch": 1.73,
      "grad_norm": 0.9737386107444763,
      "learning_rate": 0.00022946025520263402,
      "loss": 2.1135,
      "step": 132609
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.1087366342544556,
      "learning_rate": 0.00022945627933275978,
      "loss": 2.215,
      "step": 132610
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.0205806493759155,
      "learning_rate": 0.00022945230347600114,
      "loss": 1.8274,
      "step": 132611
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.058064341545105,
      "learning_rate": 0.00022944832763235846,
      "loss": 2.2837,
      "step": 132612
    },
    {
      "epoch": 1.73,
      "grad_norm": 0.9916327595710754,
      "learning_rate": 0.0002294443518018327,
      "loss": 2.029,
      "step": 132613
    },
    {
      "epoch": 1.73,
      "grad_norm": 0.9860502481460571,
      "learning_rate": 0.00022944037598442456,
      "loss": 2.0115,
      "step": 132614
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.1041173934936523,
      "learning_rate": 0.00022943640018013476,
      "loss": 2.0611,
      "step": 132615
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.107153296470642,
      "learning_rate": 0.0002294324243889641,
      "loss": 2.0557,
      "step": 132616
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.0283658504486084,
      "learning_rate": 0.00022942844861091342,
      "loss": 2.1639,
      "step": 132617
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.054024338722229,
      "learning_rate": 0.00022942447284598314,
      "loss": 2.1863,
      "step": 132618
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.058348536491394,
      "learning_rate": 0.00022942049709417424,
      "loss": 2.123,
      "step": 132619
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.0040336847305298,
      "learning_rate": 0.00022941652135548737,
      "loss": 2.1488,
      "step": 132620
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.029978632926941,
      "learning_rate": 0.00022941254562992331,
      "loss": 2.319,
      "step": 132621
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.0665284395217896,
      "learning_rate": 0.00022940856991748282,
      "loss": 2.3755,
      "step": 132622
    },
    {
      "epoch": 1.73,
      "grad_norm": 0.9381243586540222,
      "learning_rate": 0.0002294045942181667,
      "loss": 2.1526,
      "step": 132623
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.1167727708816528,
      "learning_rate": 0.00022940061853197547,
      "loss": 2.0339,
      "step": 132624
    },
    {
      "epoch": 1.73,
      "grad_norm": 0.978614330291748,
      "learning_rate": 0.00022939664285891,
      "loss": 2.0826,
      "step": 132625
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.0795555114746094,
      "learning_rate": 0.00022939266719897104,
      "loss": 2.0268,
      "step": 132626
    },
    {
      "epoch": 1.73,
      "grad_norm": 0.9143746495246887,
      "learning_rate": 0.0002293886915521593,
      "loss": 2.1116,
      "step": 132627
    },
    {
      "epoch": 1.73,
      "grad_norm": 0.9555228352546692,
      "learning_rate": 0.00022938471591847556,
      "loss": 2.1367,
      "step": 132628
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.1004239320755005,
      "learning_rate": 0.00022938074029792065,
      "loss": 2.003,
      "step": 132629
    },
    {
      "epoch": 1.73,
      "grad_norm": 0.9820027351379395,
      "learning_rate": 0.00022937676469049506,
      "loss": 2.1931,
      "step": 132630
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.13218355178833,
      "learning_rate": 0.0002293727890961997,
      "loss": 2.0479,
      "step": 132631
    },
    {
      "epoch": 1.73,
      "grad_norm": 0.9663158059120178,
      "learning_rate": 0.00022936881351503523,
      "loss": 1.9893,
      "step": 132632
    },
    {
      "epoch": 1.73,
      "grad_norm": 0.9332054257392883,
      "learning_rate": 0.00022936483794700244,
      "loss": 2.2113,
      "step": 132633
    },
    {
      "epoch": 1.73,
      "grad_norm": 0.9754804372787476,
      "learning_rate": 0.00022936086239210202,
      "loss": 1.8578,
      "step": 132634
    },
    {
      "epoch": 1.73,
      "grad_norm": 0.9461012482643127,
      "learning_rate": 0.00022935688685033487,
      "loss": 1.868,
      "step": 132635
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.0444064140319824,
      "learning_rate": 0.0002293529113217016,
      "loss": 2.0814,
      "step": 132636
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.0096073150634766,
      "learning_rate": 0.0002293489358062029,
      "loss": 2.2607,
      "step": 132637
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.0792210102081299,
      "learning_rate": 0.00022934496030383952,
      "loss": 2.3389,
      "step": 132638
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.1128969192504883,
      "learning_rate": 0.0002293409848146123,
      "loss": 2.1703,
      "step": 132639
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.2079260349273682,
      "learning_rate": 0.0002293370093385219,
      "loss": 2.0082,
      "step": 132640
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.0437170267105103,
      "learning_rate": 0.00022933303387556915,
      "loss": 2.0775,
      "step": 132641
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.0179775953292847,
      "learning_rate": 0.00022932905842575462,
      "loss": 2.0098,
      "step": 132642
    },
    {
      "epoch": 1.73,
      "grad_norm": 0.9788399338722229,
      "learning_rate": 0.0002293250829890793,
      "loss": 2.1613,
      "step": 132643
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.0492069721221924,
      "learning_rate": 0.00022932110756554365,
      "loss": 2.2958,
      "step": 132644
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.0212781429290771,
      "learning_rate": 0.00022931713215514855,
      "loss": 2.191,
      "step": 132645
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.0139421224594116,
      "learning_rate": 0.00022931315675789476,
      "loss": 1.9954,
      "step": 132646
    },
    {
      "epoch": 1.73,
      "grad_norm": 0.9867880940437317,
      "learning_rate": 0.000229309181373783,
      "loss": 1.9333,
      "step": 132647
    },
    {
      "epoch": 1.73,
      "grad_norm": 0.9847619533538818,
      "learning_rate": 0.00022930520600281397,
      "loss": 2.1444,
      "step": 132648
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.0187368392944336,
      "learning_rate": 0.00022930123064498847,
      "loss": 2.162,
      "step": 132649
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.00986647605896,
      "learning_rate": 0.0002292972553003072,
      "loss": 2.0727,
      "step": 132650
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.0856521129608154,
      "learning_rate": 0.0002292932799687709,
      "loss": 2.1504,
      "step": 132651
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.0631517171859741,
      "learning_rate": 0.00022928930465038026,
      "loss": 2.0343,
      "step": 132652
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.154337763786316,
      "learning_rate": 0.0002292853293451361,
      "loss": 2.1136,
      "step": 132653
    },
    {
      "epoch": 1.73,
      "grad_norm": 0.9657348990440369,
      "learning_rate": 0.0002292813540530392,
      "loss": 2.0315,
      "step": 132654
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.1080925464630127,
      "learning_rate": 0.00022927737877409017,
      "loss": 2.056,
      "step": 132655
    },
    {
      "epoch": 1.73,
      "grad_norm": 0.964145302772522,
      "learning_rate": 0.00022927340350828982,
      "loss": 2.0593,
      "step": 132656
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.0087449550628662,
      "learning_rate": 0.00022926942825563892,
      "loss": 2.112,
      "step": 132657
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.0477477312088013,
      "learning_rate": 0.00022926545301613812,
      "loss": 2.0774,
      "step": 132658
    },
    {
      "epoch": 1.73,
      "grad_norm": 0.9569658637046814,
      "learning_rate": 0.00022926147778978817,
      "loss": 2.2407,
      "step": 132659
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.0499742031097412,
      "learning_rate": 0.00022925750257658998,
      "loss": 2.3571,
      "step": 132660
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.0052709579467773,
      "learning_rate": 0.00022925352737654403,
      "loss": 2.1454,
      "step": 132661
    },
    {
      "epoch": 1.73,
      "grad_norm": 0.9910569190979004,
      "learning_rate": 0.00022924955218965118,
      "loss": 2.0357,
      "step": 132662
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.0621392726898193,
      "learning_rate": 0.0002292455770159122,
      "loss": 1.9741,
      "step": 132663
    },
    {
      "epoch": 1.73,
      "grad_norm": 0.9549537301063538,
      "learning_rate": 0.00022924160185532782,
      "loss": 1.9311,
      "step": 132664
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.1407748460769653,
      "learning_rate": 0.0002292376267078988,
      "loss": 2.1002,
      "step": 132665
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.0855754613876343,
      "learning_rate": 0.00022923365157362586,
      "loss": 1.862,
      "step": 132666
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.0144389867782593,
      "learning_rate": 0.00022922967645250966,
      "loss": 2.0873,
      "step": 132667
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.2778722047805786,
      "learning_rate": 0.00022922570134455099,
      "loss": 1.987,
      "step": 132668
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.162532925605774,
      "learning_rate": 0.00022922172624975057,
      "loss": 2.0651,
      "step": 132669
    },
    {
      "epoch": 1.73,
      "grad_norm": 0.984156608581543,
      "learning_rate": 0.00022921775116810917,
      "loss": 2.07,
      "step": 132670
    },
    {
      "epoch": 1.73,
      "grad_norm": 0.9597532749176025,
      "learning_rate": 0.00022921377609962757,
      "loss": 1.9683,
      "step": 132671
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.1617549657821655,
      "learning_rate": 0.00022920980104430656,
      "loss": 2.2038,
      "step": 132672
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.0995620489120483,
      "learning_rate": 0.00022920582600214666,
      "loss": 2.227,
      "step": 132673
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.0757869482040405,
      "learning_rate": 0.00022920185097314875,
      "loss": 2.1279,
      "step": 132674
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.1396987438201904,
      "learning_rate": 0.00022919787595731354,
      "loss": 2.152,
      "step": 132675
    },
    {
      "epoch": 1.73,
      "grad_norm": 0.9668095111846924,
      "learning_rate": 0.00022919390095464177,
      "loss": 1.8809,
      "step": 132676
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.1329257488250732,
      "learning_rate": 0.00022918992596513422,
      "loss": 2.2537,
      "step": 132677
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.175835132598877,
      "learning_rate": 0.0002291859509887917,
      "loss": 2.1929,
      "step": 132678
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.051279067993164,
      "learning_rate": 0.00022918197602561473,
      "loss": 2.1759,
      "step": 132679
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.0548352003097534,
      "learning_rate": 0.00022917800107560417,
      "loss": 2.2026,
      "step": 132680
    },
    {
      "epoch": 1.73,
      "grad_norm": 0.9172732830047607,
      "learning_rate": 0.00022917402613876076,
      "loss": 2.0174,
      "step": 132681
    },
    {
      "epoch": 1.73,
      "grad_norm": 0.9859389662742615,
      "learning_rate": 0.00022917005121508524,
      "loss": 2.0272,
      "step": 132682
    },
    {
      "epoch": 1.73,
      "grad_norm": 0.9943822026252747,
      "learning_rate": 0.00022916607630457832,
      "loss": 2.1198,
      "step": 132683
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.0069621801376343,
      "learning_rate": 0.0002291621014072409,
      "loss": 2.4302,
      "step": 132684
    },
    {
      "epoch": 1.73,
      "grad_norm": 0.9983388185501099,
      "learning_rate": 0.00022915812652307346,
      "loss": 2.2549,
      "step": 132685
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.229601502418518,
      "learning_rate": 0.00022915415165207688,
      "loss": 2.4503,
      "step": 132686
    },
    {
      "epoch": 1.73,
      "grad_norm": 0.9875012636184692,
      "learning_rate": 0.00022915017679425186,
      "loss": 2.2323,
      "step": 132687
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.024444818496704,
      "learning_rate": 0.00022914620194959914,
      "loss": 2.0731,
      "step": 132688
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.084761142730713,
      "learning_rate": 0.00022914222711811953,
      "loss": 2.1932,
      "step": 132689
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.0784577131271362,
      "learning_rate": 0.0002291382522998138,
      "loss": 2.1643,
      "step": 132690
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.0247411727905273,
      "learning_rate": 0.00022913427749468248,
      "loss": 2.086,
      "step": 132691
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.1094233989715576,
      "learning_rate": 0.0002291303027027265,
      "loss": 1.8807,
      "step": 132692
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.045073390007019,
      "learning_rate": 0.00022912632792394642,
      "loss": 2.0094,
      "step": 132693
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.064384937286377,
      "learning_rate": 0.00022912235315834318,
      "loss": 2.2195,
      "step": 132694
    },
    {
      "epoch": 1.73,
      "grad_norm": 0.9824357628822327,
      "learning_rate": 0.0002291183784059174,
      "loss": 1.9892,
      "step": 132695
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.1137707233428955,
      "learning_rate": 0.00022911440366667,
      "loss": 2.1933,
      "step": 132696
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.0456284284591675,
      "learning_rate": 0.00022911042894060144,
      "loss": 2.051,
      "step": 132697
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.1544135808944702,
      "learning_rate": 0.00022910645422771257,
      "loss": 1.9334,
      "step": 132698
    },
    {
      "epoch": 1.73,
      "grad_norm": 0.945273220539093,
      "learning_rate": 0.00022910247952800418,
      "loss": 2.1421,
      "step": 132699
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.3397042751312256,
      "learning_rate": 0.00022909850484147693,
      "loss": 2.4505,
      "step": 132700
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.107576608657837,
      "learning_rate": 0.00022909453016813164,
      "loss": 2.1104,
      "step": 132701
    },
    {
      "epoch": 1.73,
      "grad_norm": 0.9602046608924866,
      "learning_rate": 0.00022909055550796915,
      "loss": 2.2169,
      "step": 132702
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.0235539674758911,
      "learning_rate": 0.00022908658086098993,
      "loss": 1.9438,
      "step": 132703
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.0230517387390137,
      "learning_rate": 0.00022908260622719485,
      "loss": 2.2012,
      "step": 132704
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.051671028137207,
      "learning_rate": 0.00022907863160658462,
      "loss": 1.9755,
      "step": 132705
    },
    {
      "epoch": 1.73,
      "grad_norm": 0.9735772013664246,
      "learning_rate": 0.00022907465699916005,
      "loss": 2.1811,
      "step": 132706
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.077010154724121,
      "learning_rate": 0.00022907068240492182,
      "loss": 2.2281,
      "step": 132707
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.0369477272033691,
      "learning_rate": 0.00022906670782387085,
      "loss": 2.0801,
      "step": 132708
    },
    {
      "epoch": 1.73,
      "grad_norm": 0.9841381907463074,
      "learning_rate": 0.00022906273325600754,
      "loss": 2.1265,
      "step": 132709
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.0328129529953003,
      "learning_rate": 0.00022905875870133285,
      "loss": 2.0517,
      "step": 132710
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.0411232709884644,
      "learning_rate": 0.00022905478415984748,
      "loss": 2.0758,
      "step": 132711
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.0094373226165771,
      "learning_rate": 0.00022905080963155213,
      "loss": 2.2256,
      "step": 132712
    },
    {
      "epoch": 1.73,
      "grad_norm": 0.9884346723556519,
      "learning_rate": 0.00022904683511644757,
      "loss": 2.2379,
      "step": 132713
    },
    {
      "epoch": 1.73,
      "grad_norm": 0.99594646692276,
      "learning_rate": 0.0002290428606145347,
      "loss": 2.3304,
      "step": 132714
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.0499951839447021,
      "learning_rate": 0.000229038886125814,
      "loss": 2.2759,
      "step": 132715
    },
    {
      "epoch": 1.73,
      "grad_norm": 0.983476459980011,
      "learning_rate": 0.00022903491165028622,
      "loss": 1.993,
      "step": 132716
    },
    {
      "epoch": 1.73,
      "grad_norm": 0.9693266153335571,
      "learning_rate": 0.00022903093718795224,
      "loss": 2.0697,
      "step": 132717
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.0502334833145142,
      "learning_rate": 0.00022902696273881278,
      "loss": 2.0769,
      "step": 132718
    },
    {
      "epoch": 1.73,
      "grad_norm": 0.9059422612190247,
      "learning_rate": 0.00022902298830286853,
      "loss": 1.7325,
      "step": 132719
    },
    {
      "epoch": 1.73,
      "grad_norm": 0.9999813437461853,
      "learning_rate": 0.0002290190138801202,
      "loss": 2.2022,
      "step": 132720
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.085924744606018,
      "learning_rate": 0.00022901503947056874,
      "loss": 2.1927,
      "step": 132721
    },
    {
      "epoch": 1.73,
      "grad_norm": 0.9520219564437866,
      "learning_rate": 0.00022901106507421462,
      "loss": 1.9619,
      "step": 132722
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.0599254369735718,
      "learning_rate": 0.00022900709069105865,
      "loss": 2.2456,
      "step": 132723
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.0222065448760986,
      "learning_rate": 0.0002290031163211016,
      "loss": 2.0739,
      "step": 132724
    },
    {
      "epoch": 1.73,
      "grad_norm": 0.9841436147689819,
      "learning_rate": 0.00022899914196434422,
      "loss": 2.051,
      "step": 132725
    },
    {
      "epoch": 1.73,
      "grad_norm": 0.9065993428230286,
      "learning_rate": 0.00022899516762078724,
      "loss": 2.1004,
      "step": 132726
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.0337352752685547,
      "learning_rate": 0.00022899119329043154,
      "loss": 2.1851,
      "step": 132727
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.0130414962768555,
      "learning_rate": 0.00022898721897327757,
      "loss": 2.2418,
      "step": 132728
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.0366917848587036,
      "learning_rate": 0.00022898324466932623,
      "loss": 2.1319,
      "step": 132729
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.0900410413742065,
      "learning_rate": 0.00022897927037857823,
      "loss": 2.0955,
      "step": 132730
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.1986263990402222,
      "learning_rate": 0.00022897529610103434,
      "loss": 2.2584,
      "step": 132731
    },
    {
      "epoch": 1.73,
      "grad_norm": 0.9592410326004028,
      "learning_rate": 0.00022897132183669527,
      "loss": 1.9902,
      "step": 132732
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.0476198196411133,
      "learning_rate": 0.00022896734758556185,
      "loss": 2.1602,
      "step": 132733
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.151812195777893,
      "learning_rate": 0.00022896337334763465,
      "loss": 2.1348,
      "step": 132734
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.0085771083831787,
      "learning_rate": 0.00022895939912291458,
      "loss": 2.0203,
      "step": 132735
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.0743504762649536,
      "learning_rate": 0.00022895542491140224,
      "loss": 2.0014,
      "step": 132736
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.0150634050369263,
      "learning_rate": 0.00022895145071309843,
      "loss": 2.1952,
      "step": 132737
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.0658069849014282,
      "learning_rate": 0.0002289474765280039,
      "loss": 2.3671,
      "step": 132738
    },
    {
      "epoch": 1.73,
      "grad_norm": 0.9840204119682312,
      "learning_rate": 0.0002289435023561194,
      "loss": 1.8988,
      "step": 132739
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.155704140663147,
      "learning_rate": 0.00022893952819744558,
      "loss": 2.1695,
      "step": 132740
    },
    {
      "epoch": 1.73,
      "grad_norm": 0.9542125463485718,
      "learning_rate": 0.00022893555405198326,
      "loss": 2.2197,
      "step": 132741
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.079885721206665,
      "learning_rate": 0.00022893157991973325,
      "loss": 2.0829,
      "step": 132742
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.381209373474121,
      "learning_rate": 0.00022892760580069615,
      "loss": 2.1955,
      "step": 132743
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.1101559400558472,
      "learning_rate": 0.00022892363169487267,
      "loss": 2.0337,
      "step": 132744
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.00070059299469,
      "learning_rate": 0.00022891965760226375,
      "loss": 2.322,
      "step": 132745
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.1420438289642334,
      "learning_rate": 0.00022891568352286995,
      "loss": 2.0662,
      "step": 132746
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.362308382987976,
      "learning_rate": 0.000228911709456692,
      "loss": 2.1133,
      "step": 132747
    },
    {
      "epoch": 1.73,
      "grad_norm": 0.9874979853630066,
      "learning_rate": 0.00022890773540373079,
      "loss": 2.2476,
      "step": 132748
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.011494517326355,
      "learning_rate": 0.000228903761363987,
      "loss": 2.2009,
      "step": 132749
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.1850826740264893,
      "learning_rate": 0.00022889978733746128,
      "loss": 2.226,
      "step": 132750
    },
    {
      "epoch": 1.73,
      "grad_norm": 0.9733425974845886,
      "learning_rate": 0.00022889581332415448,
      "loss": 1.7167,
      "step": 132751
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.145570993423462,
      "learning_rate": 0.00022889183932406728,
      "loss": 2.1658,
      "step": 132752
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.1015832424163818,
      "learning_rate": 0.00022888786533720035,
      "loss": 2.2414,
      "step": 132753
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.1134213209152222,
      "learning_rate": 0.00022888389136355457,
      "loss": 2.2347,
      "step": 132754
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.1182879209518433,
      "learning_rate": 0.0002288799174031306,
      "loss": 2.3957,
      "step": 132755
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.4920598268508911,
      "learning_rate": 0.00022887594345592923,
      "loss": 1.9211,
      "step": 132756
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.0118136405944824,
      "learning_rate": 0.0002288719695219512,
      "loss": 2.103,
      "step": 132757
    },
    {
      "epoch": 1.73,
      "grad_norm": 0.9926510453224182,
      "learning_rate": 0.00022886799560119716,
      "loss": 1.8654,
      "step": 132758
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.082988977432251,
      "learning_rate": 0.00022886402169366787,
      "loss": 1.9097,
      "step": 132759
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.213708758354187,
      "learning_rate": 0.00022886004779936413,
      "loss": 2.1899,
      "step": 132760
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.0247981548309326,
      "learning_rate": 0.00022885607391828663,
      "loss": 2.0659,
      "step": 132761
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.0058878660202026,
      "learning_rate": 0.0002288521000504361,
      "loss": 2.1526,
      "step": 132762
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.0050957202911377,
      "learning_rate": 0.00022884812619581349,
      "loss": 2.0231,
      "step": 132763
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.2188992500305176,
      "learning_rate": 0.0002288441523544192,
      "loss": 1.984,
      "step": 132764
    },
    {
      "epoch": 1.73,
      "grad_norm": 0.9157253503799438,
      "learning_rate": 0.00022884017852625415,
      "loss": 2.036,
      "step": 132765
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.0659162998199463,
      "learning_rate": 0.00022883620471131903,
      "loss": 2.0389,
      "step": 132766
    },
    {
      "epoch": 1.73,
      "grad_norm": 0.9708033204078674,
      "learning_rate": 0.0002288322309096146,
      "loss": 2.0667,
      "step": 132767
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.1491912603378296,
      "learning_rate": 0.0002288282571211416,
      "loss": 1.936,
      "step": 132768
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.4489656686782837,
      "learning_rate": 0.00022882428334590092,
      "loss": 1.8665,
      "step": 132769
    },
    {
      "epoch": 1.73,
      "grad_norm": 0.9850611090660095,
      "learning_rate": 0.00022882030958389302,
      "loss": 2.2092,
      "step": 132770
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.0176799297332764,
      "learning_rate": 0.00022881633583511876,
      "loss": 2.0707,
      "step": 132771
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.0952990055084229,
      "learning_rate": 0.0002288123620995789,
      "loss": 2.3498,
      "step": 132772
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.1278743743896484,
      "learning_rate": 0.00022880838837727413,
      "loss": 2.3485,
      "step": 132773
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.1855450868606567,
      "learning_rate": 0.00022880441466820522,
      "loss": 2.1341,
      "step": 132774
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.0266613960266113,
      "learning_rate": 0.00022880044097237307,
      "loss": 1.884,
      "step": 132775
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.1056326627731323,
      "learning_rate": 0.00022879646728977813,
      "loss": 2.2342,
      "step": 132776
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.9864978790283203,
      "learning_rate": 0.00022879249362042127,
      "loss": 1.9718,
      "step": 132777
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.067200779914856,
      "learning_rate": 0.00022878851996430325,
      "loss": 1.8668,
      "step": 132778
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.0240061283111572,
      "learning_rate": 0.00022878454632142476,
      "loss": 2.0124,
      "step": 132779
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.0598666667938232,
      "learning_rate": 0.0002287805726917866,
      "loss": 2.0744,
      "step": 132780
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.035137414932251,
      "learning_rate": 0.00022877659907538956,
      "loss": 2.0733,
      "step": 132781
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.118408441543579,
      "learning_rate": 0.00022877262547223417,
      "loss": 1.971,
      "step": 132782
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.0347590446472168,
      "learning_rate": 0.00022876865188232133,
      "loss": 2.0602,
      "step": 132783
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.0309699773788452,
      "learning_rate": 0.00022876467830565173,
      "loss": 2.1609,
      "step": 132784
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.1372928619384766,
      "learning_rate": 0.0002287607047422261,
      "loss": 2.0845,
      "step": 132785
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.365140438079834,
      "learning_rate": 0.00022875673119204521,
      "loss": 1.9212,
      "step": 132786
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.1663391590118408,
      "learning_rate": 0.0002287527576551098,
      "loss": 2.2156,
      "step": 132787
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.144227147102356,
      "learning_rate": 0.0002287487841314207,
      "loss": 1.823,
      "step": 132788
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.3012832403182983,
      "learning_rate": 0.00022874481062097848,
      "loss": 2.1851,
      "step": 132789
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.2898147106170654,
      "learning_rate": 0.0002287408371237839,
      "loss": 2.0498,
      "step": 132790
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.117215871810913,
      "learning_rate": 0.00022873686363983773,
      "loss": 2.1614,
      "step": 132791
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.0104495286941528,
      "learning_rate": 0.0002287328901691407,
      "loss": 1.9534,
      "step": 132792
    },
    {
      "epoch": 1.73,
      "grad_norm": 0.9824489951133728,
      "learning_rate": 0.00022872891671169365,
      "loss": 2.2047,
      "step": 132793
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.088879942893982,
      "learning_rate": 0.00022872494326749731,
      "loss": 2.3613,
      "step": 132794
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.000597596168518,
      "learning_rate": 0.00022872096983655223,
      "loss": 1.9738,
      "step": 132795
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.0090429782867432,
      "learning_rate": 0.0002287169964188593,
      "loss": 2.1923,
      "step": 132796
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.1435662508010864,
      "learning_rate": 0.00022871302301441916,
      "loss": 1.9951,
      "step": 132797
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.152706265449524,
      "learning_rate": 0.0002287090496232327,
      "loss": 2.035,
      "step": 132798
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.093812346458435,
      "learning_rate": 0.00022870507624530046,
      "loss": 2.2371,
      "step": 132799
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.1022495031356812,
      "learning_rate": 0.00022870110288062353,
      "loss": 2.2233,
      "step": 132800
    },
    {
      "epoch": 1.73,
      "grad_norm": 0.9970834851264954,
      "learning_rate": 0.00022869712952920226,
      "loss": 2.0352,
      "step": 132801
    },
    {
      "epoch": 1.73,
      "grad_norm": 0.9102974534034729,
      "learning_rate": 0.0002286931561910375,
      "loss": 2.1456,
      "step": 132802
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.0788620710372925,
      "learning_rate": 0.00022868918286613003,
      "loss": 2.0445,
      "step": 132803
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.0942296981811523,
      "learning_rate": 0.0002286852095544806,
      "loss": 2.2593,
      "step": 132804
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.1466206312179565,
      "learning_rate": 0.00022868123625608994,
      "loss": 2.2422,
      "step": 132805
    },
    {
      "epoch": 1.73,
      "grad_norm": 0.9807042479515076,
      "learning_rate": 0.00022867726297095894,
      "loss": 2.1493,
      "step": 132806
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.0599745512008667,
      "learning_rate": 0.00022867328969908803,
      "loss": 2.0507,
      "step": 132807
    },
    {
      "epoch": 1.73,
      "grad_norm": 0.9671066403388977,
      "learning_rate": 0.0002286693164404781,
      "loss": 2.023,
      "step": 132808
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.0485531091690063,
      "learning_rate": 0.00022866534319512986,
      "loss": 2.2806,
      "step": 132809
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.0473893880844116,
      "learning_rate": 0.00022866136996304413,
      "loss": 2.1094,
      "step": 132810
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.0542463064193726,
      "learning_rate": 0.00022865739674422157,
      "loss": 2.0578,
      "step": 132811
    },
    {
      "epoch": 1.73,
      "grad_norm": 0.9312295913696289,
      "learning_rate": 0.00022865342353866307,
      "loss": 2.101,
      "step": 132812
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.0397543907165527,
      "learning_rate": 0.00022864945034636912,
      "loss": 2.1256,
      "step": 132813
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.0372679233551025,
      "learning_rate": 0.0002286454771673406,
      "loss": 2.1069,
      "step": 132814
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.039312720298767,
      "learning_rate": 0.00022864150400157823,
      "loss": 1.9865,
      "step": 132815
    },
    {
      "epoch": 1.73,
      "grad_norm": 0.9602411389350891,
      "learning_rate": 0.00022863753084908275,
      "loss": 2.0291,
      "step": 132816
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.1281628608703613,
      "learning_rate": 0.0002286335577098549,
      "loss": 2.2429,
      "step": 132817
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.1359206438064575,
      "learning_rate": 0.00022862958458389547,
      "loss": 2.1454,
      "step": 132818
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.163015365600586,
      "learning_rate": 0.00022862561147120515,
      "loss": 2.1935,
      "step": 132819
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.0276554822921753,
      "learning_rate": 0.0002286216383717846,
      "loss": 2.2512,
      "step": 132820
    },
    {
      "epoch": 1.73,
      "grad_norm": 0.9860948920249939,
      "learning_rate": 0.00022861766528563468,
      "loss": 2.0814,
      "step": 132821
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.1858209371566772,
      "learning_rate": 0.00022861369221275604,
      "loss": 1.9513,
      "step": 132822
    },
    {
      "epoch": 1.73,
      "grad_norm": 0.9904860258102417,
      "learning_rate": 0.0002286097191531495,
      "loss": 2.3534,
      "step": 132823
    },
    {
      "epoch": 1.73,
      "grad_norm": 0.9944366812705994,
      "learning_rate": 0.0002286057461068158,
      "loss": 1.931,
      "step": 132824
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.128684639930725,
      "learning_rate": 0.0002286017730737556,
      "loss": 2.2433,
      "step": 132825
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.0169740915298462,
      "learning_rate": 0.00022859780005396964,
      "loss": 2.138,
      "step": 132826
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.062605381011963,
      "learning_rate": 0.00022859382704745878,
      "loss": 2.0768,
      "step": 132827
    },
    {
      "epoch": 1.73,
      "grad_norm": 0.955152690410614,
      "learning_rate": 0.00022858985405422363,
      "loss": 2.0464,
      "step": 132828
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.1006443500518799,
      "learning_rate": 0.00022858588107426493,
      "loss": 2.2735,
      "step": 132829
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.00108802318573,
      "learning_rate": 0.00022858190810758357,
      "loss": 2.0584,
      "step": 132830
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.0859858989715576,
      "learning_rate": 0.0002285779351541801,
      "loss": 2.1675,
      "step": 132831
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.0721229314804077,
      "learning_rate": 0.00022857396221405535,
      "loss": 2.0494,
      "step": 132832
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.0242201089859009,
      "learning_rate": 0.00022856998928720999,
      "loss": 1.8611,
      "step": 132833
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.029154896736145,
      "learning_rate": 0.00022856601637364496,
      "loss": 2.1668,
      "step": 132834
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.1160317659378052,
      "learning_rate": 0.00022856204347336073,
      "loss": 2.2064,
      "step": 132835
    },
    {
      "epoch": 1.73,
      "grad_norm": 0.9372049570083618,
      "learning_rate": 0.00022855807058635825,
      "loss": 1.956,
      "step": 132836
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.128152847290039,
      "learning_rate": 0.00022855409771263817,
      "loss": 1.987,
      "step": 132837
    },
    {
      "epoch": 1.73,
      "grad_norm": 0.989872932434082,
      "learning_rate": 0.00022855012485220117,
      "loss": 1.8548,
      "step": 132838
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.0665004253387451,
      "learning_rate": 0.00022854615200504805,
      "loss": 2.0831,
      "step": 132839
    },
    {
      "epoch": 1.73,
      "grad_norm": 0.9153387546539307,
      "learning_rate": 0.00022854217917117956,
      "loss": 1.9147,
      "step": 132840
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.0683724880218506,
      "learning_rate": 0.0002285382063505965,
      "loss": 2.0802,
      "step": 132841
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.0602360963821411,
      "learning_rate": 0.00022853423354329952,
      "loss": 2.0628,
      "step": 132842
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.1376049518585205,
      "learning_rate": 0.00022853026074928933,
      "loss": 2.1901,
      "step": 132843
    },
    {
      "epoch": 1.73,
      "grad_norm": 0.9603456258773804,
      "learning_rate": 0.0002285262879685667,
      "loss": 2.2098,
      "step": 132844
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.031906247138977,
      "learning_rate": 0.0002285223152011324,
      "loss": 1.8902,
      "step": 132845
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.0418680906295776,
      "learning_rate": 0.00022851834244698712,
      "loss": 2.1227,
      "step": 132846
    },
    {
      "epoch": 1.73,
      "grad_norm": 0.9712125062942505,
      "learning_rate": 0.00022851436970613165,
      "loss": 1.8683,
      "step": 132847
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.1225731372833252,
      "learning_rate": 0.00022851039697856682,
      "loss": 2.0681,
      "step": 132848
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.0231212377548218,
      "learning_rate": 0.00022850642426429314,
      "loss": 2.1229,
      "step": 132849
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.0715397596359253,
      "learning_rate": 0.00022850245156331148,
      "loss": 2.1535,
      "step": 132850
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.0874807834625244,
      "learning_rate": 0.00022849847887562253,
      "loss": 2.0266,
      "step": 132851
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.1926207542419434,
      "learning_rate": 0.00022849450620122708,
      "loss": 2.0454,
      "step": 132852
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.0961005687713623,
      "learning_rate": 0.00022849053354012587,
      "loss": 2.0329,
      "step": 132853
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.2409268617630005,
      "learning_rate": 0.0002284865608923196,
      "loss": 2.4553,
      "step": 132854
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.0545090436935425,
      "learning_rate": 0.00022848258825780915,
      "loss": 2.2332,
      "step": 132855
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.0558537244796753,
      "learning_rate": 0.000228478615636595,
      "loss": 2.2149,
      "step": 132856
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.0768624544143677,
      "learning_rate": 0.00022847464302867807,
      "loss": 2.0407,
      "step": 132857
    },
    {
      "epoch": 1.73,
      "grad_norm": 0.9774876236915588,
      "learning_rate": 0.00022847067043405903,
      "loss": 1.9001,
      "step": 132858
    },
    {
      "epoch": 1.73,
      "grad_norm": 0.9004591107368469,
      "learning_rate": 0.00022846669785273863,
      "loss": 2.0568,
      "step": 132859
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.0974916219711304,
      "learning_rate": 0.00022846272528471764,
      "loss": 2.0737,
      "step": 132860
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.0960136651992798,
      "learning_rate": 0.0002284587527299969,
      "loss": 2.0476,
      "step": 132861
    },
    {
      "epoch": 1.73,
      "grad_norm": 0.9851827621459961,
      "learning_rate": 0.00022845478018857692,
      "loss": 1.9125,
      "step": 132862
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.025694727897644,
      "learning_rate": 0.0002284508076604585,
      "loss": 2.1605,
      "step": 132863
    },
    {
      "epoch": 1.73,
      "grad_norm": 0.9568495154380798,
      "learning_rate": 0.00022844683514564243,
      "loss": 1.9687,
      "step": 132864
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.1619493961334229,
      "learning_rate": 0.00022844286264412947,
      "loss": 2.0855,
      "step": 132865
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.0795239210128784,
      "learning_rate": 0.00022843889015592033,
      "loss": 1.9629,
      "step": 132866
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.0759650468826294,
      "learning_rate": 0.00022843491768101588,
      "loss": 2.1724,
      "step": 132867
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.0138126611709595,
      "learning_rate": 0.0002284309452194166,
      "loss": 2.031,
      "step": 132868
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.1360502243041992,
      "learning_rate": 0.00022842697277112336,
      "loss": 2.0957,
      "step": 132869
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.4109083414077759,
      "learning_rate": 0.0002284230003361369,
      "loss": 2.1676,
      "step": 132870
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.026530146598816,
      "learning_rate": 0.00022841902791445793,
      "loss": 1.9811,
      "step": 132871
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.0872291326522827,
      "learning_rate": 0.00022841505550608724,
      "loss": 1.9878,
      "step": 132872
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.020043969154358,
      "learning_rate": 0.00022841108311102566,
      "loss": 1.9386,
      "step": 132873
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.0086719989776611,
      "learning_rate": 0.0002284071107292737,
      "loss": 2.2116,
      "step": 132874
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.0537478923797607,
      "learning_rate": 0.0002284031383608322,
      "loss": 2.0693,
      "step": 132875
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.056467056274414,
      "learning_rate": 0.00022839916600570192,
      "loss": 2.0819,
      "step": 132876
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.0167033672332764,
      "learning_rate": 0.00022839519366388357,
      "loss": 2.1826,
      "step": 132877
    },
    {
      "epoch": 1.73,
      "grad_norm": 0.9890533685684204,
      "learning_rate": 0.0002283912213353779,
      "loss": 2.2798,
      "step": 132878
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.010074496269226,
      "learning_rate": 0.0002283872490201858,
      "loss": 1.9629,
      "step": 132879
    },
    {
      "epoch": 1.73,
      "grad_norm": 0.9984254240989685,
      "learning_rate": 0.0002283832767183077,
      "loss": 2.254,
      "step": 132880
    },
    {
      "epoch": 1.73,
      "grad_norm": 0.9533018469810486,
      "learning_rate": 0.0002283793044297445,
      "loss": 2.0236,
      "step": 132881
    },
    {
      "epoch": 1.73,
      "grad_norm": 0.9535803198814392,
      "learning_rate": 0.00022837533215449698,
      "loss": 2.0936,
      "step": 132882
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.1219868659973145,
      "learning_rate": 0.00022837135989256583,
      "loss": 2.1028,
      "step": 132883
    },
    {
      "epoch": 1.73,
      "grad_norm": 0.9935619235038757,
      "learning_rate": 0.0002283673876439518,
      "loss": 2.1799,
      "step": 132884
    },
    {
      "epoch": 1.73,
      "grad_norm": 0.9142006635665894,
      "learning_rate": 0.0002283634154086557,
      "loss": 2.1286,
      "step": 132885
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.0368750095367432,
      "learning_rate": 0.00022835944318667814,
      "loss": 2.2262,
      "step": 132886
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.0772442817687988,
      "learning_rate": 0.00022835547097801987,
      "loss": 2.2927,
      "step": 132887
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.0265791416168213,
      "learning_rate": 0.00022835149878268164,
      "loss": 1.9715,
      "step": 132888
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.1017080545425415,
      "learning_rate": 0.00022834752660066425,
      "loss": 2.1541,
      "step": 132889
    },
    {
      "epoch": 1.73,
      "grad_norm": 0.9548481106758118,
      "learning_rate": 0.00022834355443196843,
      "loss": 2.086,
      "step": 132890
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.074995517730713,
      "learning_rate": 0.000228339582276595,
      "loss": 2.0447,
      "step": 132891
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.0743287801742554,
      "learning_rate": 0.00022833561013454448,
      "loss": 2.2285,
      "step": 132892
    },
    {
      "epoch": 1.73,
      "grad_norm": 0.9687629342079163,
      "learning_rate": 0.00022833163800581774,
      "loss": 2.049,
      "step": 132893
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.0406696796417236,
      "learning_rate": 0.00022832766589041546,
      "loss": 2.3245,
      "step": 132894
    },
    {
      "epoch": 1.73,
      "grad_norm": 0.9697263836860657,
      "learning_rate": 0.00022832369378833841,
      "loss": 2.0653,
      "step": 132895
    },
    {
      "epoch": 1.73,
      "grad_norm": 0.9615604281425476,
      "learning_rate": 0.00022831972169958737,
      "loss": 2.0377,
      "step": 132896
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.022344708442688,
      "learning_rate": 0.00022831574962416318,
      "loss": 2.2894,
      "step": 132897
    },
    {
      "epoch": 1.73,
      "grad_norm": 0.9740738868713379,
      "learning_rate": 0.00022831177756206633,
      "loss": 2.1691,
      "step": 132898
    },
    {
      "epoch": 1.73,
      "grad_norm": 0.9405511021614075,
      "learning_rate": 0.00022830780551329763,
      "loss": 1.9685,
      "step": 132899
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.007769227027893,
      "learning_rate": 0.0002283038334778579,
      "loss": 1.9931,
      "step": 132900
    },
    {
      "epoch": 1.73,
      "grad_norm": 0.9443867802619934,
      "learning_rate": 0.0002282998614557478,
      "loss": 2.119,
      "step": 132901
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.1181342601776123,
      "learning_rate": 0.00022829588944696814,
      "loss": 2.1261,
      "step": 132902
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.1419646739959717,
      "learning_rate": 0.0002282919174515197,
      "loss": 2.3454,
      "step": 132903
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.0027108192443848,
      "learning_rate": 0.0002282879454694031,
      "loss": 2.1759,
      "step": 132904
    },
    {
      "epoch": 1.73,
      "grad_norm": 0.9850882291793823,
      "learning_rate": 0.0002282839735006191,
      "loss": 2.2094,
      "step": 132905
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.0787981748580933,
      "learning_rate": 0.00022828000154516848,
      "loss": 2.1569,
      "step": 132906
    },
    {
      "epoch": 1.73,
      "grad_norm": 0.9999209642410278,
      "learning_rate": 0.00022827602960305194,
      "loss": 1.9749,
      "step": 132907
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.022260308265686,
      "learning_rate": 0.00022827205767427018,
      "loss": 1.874,
      "step": 132908
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.0685118436813354,
      "learning_rate": 0.00022826808575882415,
      "loss": 2.2625,
      "step": 132909
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.026741623878479,
      "learning_rate": 0.00022826411385671433,
      "loss": 2.2198,
      "step": 132910
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.0159543752670288,
      "learning_rate": 0.00022826014196794165,
      "loss": 2.164,
      "step": 132911
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.0607582330703735,
      "learning_rate": 0.00022825617009250668,
      "loss": 2.0226,
      "step": 132912
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.1030151844024658,
      "learning_rate": 0.00022825219823041025,
      "loss": 2.2209,
      "step": 132913
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.0402880907058716,
      "learning_rate": 0.0002282482263816531,
      "loss": 2.1835,
      "step": 132914
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.0598777532577515,
      "learning_rate": 0.000228244254546236,
      "loss": 2.1232,
      "step": 132915
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.0249192714691162,
      "learning_rate": 0.00022824028272415965,
      "loss": 2.0121,
      "step": 132916
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.0262885093688965,
      "learning_rate": 0.00022823631091542471,
      "loss": 2.2701,
      "step": 132917
    },
    {
      "epoch": 1.73,
      "grad_norm": 0.9863377809524536,
      "learning_rate": 0.00022823233912003203,
      "loss": 2.0798,
      "step": 132918
    },
    {
      "epoch": 1.73,
      "grad_norm": 0.9273279905319214,
      "learning_rate": 0.00022822836733798238,
      "loss": 2.1966,
      "step": 132919
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.019272804260254,
      "learning_rate": 0.00022822439556927633,
      "loss": 2.2352,
      "step": 132920
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.0231609344482422,
      "learning_rate": 0.00022822042381391474,
      "loss": 1.9724,
      "step": 132921
    },
    {
      "epoch": 1.73,
      "grad_norm": 0.9947380423545837,
      "learning_rate": 0.00022821645207189841,
      "loss": 2.0756,
      "step": 132922
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.0926475524902344,
      "learning_rate": 0.0002282124803432279,
      "loss": 2.0627,
      "step": 132923
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.0985379219055176,
      "learning_rate": 0.00022820850862790407,
      "loss": 2.0554,
      "step": 132924
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.0071327686309814,
      "learning_rate": 0.00022820453692592763,
      "loss": 2.1308,
      "step": 132925
    },
    {
      "epoch": 1.73,
      "grad_norm": 0.9943336248397827,
      "learning_rate": 0.0002282005652372994,
      "loss": 1.9223,
      "step": 132926
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.0393825769424438,
      "learning_rate": 0.00022819659356201994,
      "loss": 2.1356,
      "step": 132927
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.0553289651870728,
      "learning_rate": 0.0002281926219000902,
      "loss": 2.2305,
      "step": 132928
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.086137056350708,
      "learning_rate": 0.0002281886502515107,
      "loss": 2.2765,
      "step": 132929
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.0523760318756104,
      "learning_rate": 0.0002281846786162823,
      "loss": 1.9473,
      "step": 132930
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.0816645622253418,
      "learning_rate": 0.00022818070699440573,
      "loss": 2.1722,
      "step": 132931
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.0061028003692627,
      "learning_rate": 0.00022817673538588174,
      "loss": 1.9412,
      "step": 132932
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.03791344165802,
      "learning_rate": 0.00022817276379071105,
      "loss": 2.0155,
      "step": 132933
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.0145666599273682,
      "learning_rate": 0.00022816879220889443,
      "loss": 2.1291,
      "step": 132934
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.0490853786468506,
      "learning_rate": 0.00022816482064043253,
      "loss": 1.9403,
      "step": 132935
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.0878437757492065,
      "learning_rate": 0.00022816084908532618,
      "loss": 2.2948,
      "step": 132936
    },
    {
      "epoch": 1.73,
      "grad_norm": 0.9820671081542969,
      "learning_rate": 0.00022815687754357606,
      "loss": 2.1073,
      "step": 132937
    },
    {
      "epoch": 1.73,
      "grad_norm": 0.9995513558387756,
      "learning_rate": 0.00022815290601518293,
      "loss": 2.0095,
      "step": 132938
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.0324196815490723,
      "learning_rate": 0.00022814893450014752,
      "loss": 2.2573,
      "step": 132939
    },
    {
      "epoch": 1.73,
      "grad_norm": 0.9762698411941528,
      "learning_rate": 0.00022814496299847068,
      "loss": 2.2617,
      "step": 132940
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.0119181871414185,
      "learning_rate": 0.00022814099151015297,
      "loss": 2.1991,
      "step": 132941
    },
    {
      "epoch": 1.73,
      "grad_norm": 0.973669171333313,
      "learning_rate": 0.00022813702003519518,
      "loss": 2.1501,
      "step": 132942
    },
    {
      "epoch": 1.73,
      "grad_norm": 0.9765135049819946,
      "learning_rate": 0.0002281330485735981,
      "loss": 2.1339,
      "step": 132943
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.0047471523284912,
      "learning_rate": 0.00022812907712536242,
      "loss": 2.0558,
      "step": 132944
    },
    {
      "epoch": 1.73,
      "grad_norm": 0.9540241956710815,
      "learning_rate": 0.00022812510569048894,
      "loss": 2.2387,
      "step": 132945
    },
    {
      "epoch": 1.73,
      "grad_norm": 0.9753679633140564,
      "learning_rate": 0.00022812113426897844,
      "loss": 2.2709,
      "step": 132946
    },
    {
      "epoch": 1.73,
      "grad_norm": 0.9591436386108398,
      "learning_rate": 0.00022811716286083147,
      "loss": 2.2298,
      "step": 132947
    },
    {
      "epoch": 1.73,
      "grad_norm": 0.9286619424819946,
      "learning_rate": 0.00022811319146604888,
      "loss": 2.2542,
      "step": 132948
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.008030652999878,
      "learning_rate": 0.0002281092200846314,
      "loss": 2.2019,
      "step": 132949
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.1114825010299683,
      "learning_rate": 0.00022810524871657976,
      "loss": 2.2699,
      "step": 132950
    },
    {
      "epoch": 1.73,
      "grad_norm": 0.9473473429679871,
      "learning_rate": 0.00022810127736189474,
      "loss": 2.0078,
      "step": 132951
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.0533976554870605,
      "learning_rate": 0.00022809730602057717,
      "loss": 2.1311,
      "step": 132952
    },
    {
      "epoch": 1.73,
      "grad_norm": 0.9592150449752808,
      "learning_rate": 0.0002280933346926275,
      "loss": 2.1207,
      "step": 132953
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.0884203910827637,
      "learning_rate": 0.00022808936337804668,
      "loss": 1.9935,
      "step": 132954
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.0447726249694824,
      "learning_rate": 0.00022808539207683538,
      "loss": 1.9858,
      "step": 132955
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.032734990119934,
      "learning_rate": 0.00022808142078899437,
      "loss": 2.107,
      "step": 132956
    },
    {
      "epoch": 1.73,
      "grad_norm": 0.9729154706001282,
      "learning_rate": 0.0002280774495145244,
      "loss": 2.4233,
      "step": 132957
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.0147510766983032,
      "learning_rate": 0.00022807347825342628,
      "loss": 2.0759,
      "step": 132958
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.0029462575912476,
      "learning_rate": 0.00022806950700570057,
      "loss": 2.0195,
      "step": 132959
    },
    {
      "epoch": 1.73,
      "grad_norm": 0.9918522834777832,
      "learning_rate": 0.00022806553577134808,
      "loss": 2.0289,
      "step": 132960
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.0603431463241577,
      "learning_rate": 0.00022806156455036956,
      "loss": 2.1037,
      "step": 132961
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.078041434288025,
      "learning_rate": 0.00022805759334276575,
      "loss": 1.9801,
      "step": 132962
    },
    {
      "epoch": 1.73,
      "grad_norm": 0.9957859516143799,
      "learning_rate": 0.00022805362214853742,
      "loss": 2.1199,
      "step": 132963
    },
    {
      "epoch": 1.73,
      "grad_norm": 0.9784367084503174,
      "learning_rate": 0.00022804965096768538,
      "loss": 2.0524,
      "step": 132964
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.0815192461013794,
      "learning_rate": 0.00022804567980021015,
      "loss": 2.1057,
      "step": 132965
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.0667004585266113,
      "learning_rate": 0.00022804170864611258,
      "loss": 2.2494,
      "step": 132966
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.0637108087539673,
      "learning_rate": 0.00022803773750539342,
      "loss": 2.2993,
      "step": 132967
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.1164134740829468,
      "learning_rate": 0.00022803376637805343,
      "loss": 2.1666,
      "step": 132968
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.0200680494308472,
      "learning_rate": 0.00022802979526409325,
      "loss": 2.2517,
      "step": 132969
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.0522042512893677,
      "learning_rate": 0.00022802582416351388,
      "loss": 2.0093,
      "step": 132970
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.2177950143814087,
      "learning_rate": 0.00022802185307631571,
      "loss": 2.2078,
      "step": 132971
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.046337366104126,
      "learning_rate": 0.00022801788200249966,
      "loss": 2.2696,
      "step": 132972
    },
    {
      "epoch": 1.73,
      "grad_norm": 0.9858936667442322,
      "learning_rate": 0.0002280139109420664,
      "loss": 2.0405,
      "step": 132973
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.1544424295425415,
      "learning_rate": 0.00022800993989501676,
      "loss": 2.0499,
      "step": 132974
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.105090856552124,
      "learning_rate": 0.00022800596886135143,
      "loss": 2.0506,
      "step": 132975
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.1397781372070312,
      "learning_rate": 0.00022800199784107125,
      "loss": 2.2594,
      "step": 132976
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.0567314624786377,
      "learning_rate": 0.00022799802683417676,
      "loss": 2.2069,
      "step": 132977
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.1402438879013062,
      "learning_rate": 0.00022799405584066876,
      "loss": 2.1502,
      "step": 132978
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.238306999206543,
      "learning_rate": 0.00022799008486054805,
      "loss": 2.2559,
      "step": 132979
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.0963404178619385,
      "learning_rate": 0.00022798611389381534,
      "loss": 2.2492,
      "step": 132980
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.085004210472107,
      "learning_rate": 0.00022798214294047136,
      "loss": 2.1013,
      "step": 132981
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.3284929990768433,
      "learning_rate": 0.000227978172000517,
      "loss": 1.9561,
      "step": 132982
    },
    {
      "epoch": 1.73,
      "grad_norm": 0.9247034788131714,
      "learning_rate": 0.0002279742010739527,
      "loss": 2.0279,
      "step": 132983
    },
    {
      "epoch": 1.73,
      "grad_norm": 0.9842502474784851,
      "learning_rate": 0.00022797023016077938,
      "loss": 2.1325,
      "step": 132984
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.0450900793075562,
      "learning_rate": 0.00022796625926099775,
      "loss": 2.1705,
      "step": 132985
    },
    {
      "epoch": 1.73,
      "grad_norm": 0.9711595773696899,
      "learning_rate": 0.00022796228837460857,
      "loss": 2.1726,
      "step": 132986
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.0671948194503784,
      "learning_rate": 0.00022795831750161254,
      "loss": 2.2175,
      "step": 132987
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.099317193031311,
      "learning_rate": 0.00022795434664201042,
      "loss": 2.2001,
      "step": 132988
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.0429589748382568,
      "learning_rate": 0.00022795037579580307,
      "loss": 1.858,
      "step": 132989
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.0573091506958008,
      "learning_rate": 0.00022794640496299096,
      "loss": 2.0321,
      "step": 132990
    },
    {
      "epoch": 1.73,
      "grad_norm": 0.986438512802124,
      "learning_rate": 0.00022794243414357503,
      "loss": 1.9346,
      "step": 132991
    },
    {
      "epoch": 1.73,
      "grad_norm": 0.9593976140022278,
      "learning_rate": 0.00022793846333755594,
      "loss": 2.1317,
      "step": 132992
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.0809160470962524,
      "learning_rate": 0.00022793449254493442,
      "loss": 2.072,
      "step": 132993
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.0249381065368652,
      "learning_rate": 0.00022793052176571125,
      "loss": 2.0503,
      "step": 132994
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.0664807558059692,
      "learning_rate": 0.00022792655099988725,
      "loss": 2.0745,
      "step": 132995
    },
    {
      "epoch": 1.73,
      "grad_norm": 0.9731449484825134,
      "learning_rate": 0.000227922580247463,
      "loss": 2.0645,
      "step": 132996
    },
    {
      "epoch": 1.73,
      "grad_norm": 0.9725233316421509,
      "learning_rate": 0.0002279186095084393,
      "loss": 1.9574,
      "step": 132997
    },
    {
      "epoch": 1.73,
      "grad_norm": 0.9545137286186218,
      "learning_rate": 0.00022791463878281688,
      "loss": 2.2175,
      "step": 132998
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.077426791191101,
      "learning_rate": 0.00022791066807059652,
      "loss": 2.1041,
      "step": 132999
    },
    {
      "epoch": 1.73,
      "grad_norm": 0.9675751328468323,
      "learning_rate": 0.00022790669737177886,
      "loss": 2.0833,
      "step": 133000
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.0152848958969116,
      "learning_rate": 0.0002279027266863648,
      "loss": 1.8995,
      "step": 133001
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.3091788291931152,
      "learning_rate": 0.00022789875601435494,
      "loss": 2.0399,
      "step": 133002
    },
    {
      "epoch": 1.73,
      "grad_norm": 0.9667308926582336,
      "learning_rate": 0.00022789478535575007,
      "loss": 2.0442,
      "step": 133003
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.042067289352417,
      "learning_rate": 0.0002278908147105509,
      "loss": 2.212,
      "step": 133004
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.253331184387207,
      "learning_rate": 0.0002278868440787582,
      "loss": 2.0708,
      "step": 133005
    },
    {
      "epoch": 1.73,
      "grad_norm": 0.9417887330055237,
      "learning_rate": 0.00022788287346037269,
      "loss": 2.2267,
      "step": 133006
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.1949584484100342,
      "learning_rate": 0.00022787890285539517,
      "loss": 2.0077,
      "step": 133007
    },
    {
      "epoch": 1.73,
      "grad_norm": 0.9451141953468323,
      "learning_rate": 0.00022787493226382623,
      "loss": 2.0968,
      "step": 133008
    },
    {
      "epoch": 1.73,
      "grad_norm": 0.9719168543815613,
      "learning_rate": 0.00022787096168566673,
      "loss": 2.1556,
      "step": 133009
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.0792118310928345,
      "learning_rate": 0.00022786699112091745,
      "loss": 2.0084,
      "step": 133010
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.0040607452392578,
      "learning_rate": 0.000227863020569579,
      "loss": 2.2043,
      "step": 133011
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.0144747495651245,
      "learning_rate": 0.00022785905003165215,
      "loss": 1.9775,
      "step": 133012
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.0002708435058594,
      "learning_rate": 0.00022785507950713775,
      "loss": 2.304,
      "step": 133013
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.0107990503311157,
      "learning_rate": 0.00022785110899603637,
      "loss": 1.9977,
      "step": 133014
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.047271966934204,
      "learning_rate": 0.00022784713849834885,
      "loss": 1.9679,
      "step": 133015
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.0189961194992065,
      "learning_rate": 0.00022784316801407588,
      "loss": 1.991,
      "step": 133016
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.0092829465866089,
      "learning_rate": 0.00022783919754321825,
      "loss": 2.0554,
      "step": 133017
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.0528324842453003,
      "learning_rate": 0.00022783522708577674,
      "loss": 2.082,
      "step": 133018
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.0785003900527954,
      "learning_rate": 0.000227831256641752,
      "loss": 2.0151,
      "step": 133019
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.0163081884384155,
      "learning_rate": 0.00022782728621114476,
      "loss": 2.0261,
      "step": 133020
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.0320543050765991,
      "learning_rate": 0.00022782331579395578,
      "loss": 2.3076,
      "step": 133021
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.0729570388793945,
      "learning_rate": 0.00022781934539018578,
      "loss": 2.1457,
      "step": 133022
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.0476593971252441,
      "learning_rate": 0.00022781537499983555,
      "loss": 2.1045,
      "step": 133023
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.0140738487243652,
      "learning_rate": 0.00022781140462290584,
      "loss": 1.957,
      "step": 133024
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.1478770971298218,
      "learning_rate": 0.0002278074342593974,
      "loss": 2.0232,
      "step": 133025
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.1244192123413086,
      "learning_rate": 0.00022780346390931087,
      "loss": 2.0243,
      "step": 133026
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.256897211074829,
      "learning_rate": 0.00022779949357264702,
      "loss": 2.1582,
      "step": 133027
    },
    {
      "epoch": 1.73,
      "grad_norm": 0.9609810709953308,
      "learning_rate": 0.00022779552324940657,
      "loss": 2.4128,
      "step": 133028
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.105835199356079,
      "learning_rate": 0.00022779155293959033,
      "loss": 2.0967,
      "step": 133029
    },
    {
      "epoch": 1.73,
      "grad_norm": 0.9890455007553101,
      "learning_rate": 0.000227787582643199,
      "loss": 2.0475,
      "step": 133030
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.1333298683166504,
      "learning_rate": 0.00022778361236023341,
      "loss": 1.838,
      "step": 133031
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.1712065935134888,
      "learning_rate": 0.00022777964209069415,
      "loss": 2.1599,
      "step": 133032
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.1197768449783325,
      "learning_rate": 0.00022777567183458197,
      "loss": 2.3695,
      "step": 133033
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.0179409980773926,
      "learning_rate": 0.00022777170159189765,
      "loss": 2.0745,
      "step": 133034
    },
    {
      "epoch": 1.73,
      "grad_norm": 0.9962258338928223,
      "learning_rate": 0.00022776773136264197,
      "loss": 2.0619,
      "step": 133035
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.0679469108581543,
      "learning_rate": 0.0002277637611468156,
      "loss": 2.0293,
      "step": 133036
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.0248584747314453,
      "learning_rate": 0.00022775979094441946,
      "loss": 1.7794,
      "step": 133037
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.005611538887024,
      "learning_rate": 0.000227755820755454,
      "loss": 1.9932,
      "step": 133038
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.0537004470825195,
      "learning_rate": 0.0002277518505799201,
      "loss": 2.0272,
      "step": 133039
    },
    {
      "epoch": 1.73,
      "grad_norm": 0.9981288313865662,
      "learning_rate": 0.0002277478804178185,
      "loss": 2.0212,
      "step": 133040
    },
    {
      "epoch": 1.73,
      "grad_norm": 0.878501296043396,
      "learning_rate": 0.0002277439102691499,
      "loss": 1.98,
      "step": 133041
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.029327392578125,
      "learning_rate": 0.00022773994013391513,
      "loss": 2.1307,
      "step": 133042
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.0486466884613037,
      "learning_rate": 0.00022773597001211494,
      "loss": 2.2109,
      "step": 133043
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.0653055906295776,
      "learning_rate": 0.00022773199990374992,
      "loss": 2.0513,
      "step": 133044
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.0187623500823975,
      "learning_rate": 0.00022772802980882084,
      "loss": 2.0236,
      "step": 133045
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.0519047975540161,
      "learning_rate": 0.00022772405972732853,
      "loss": 1.9212,
      "step": 133046
    },
    {
      "epoch": 1.73,
      "grad_norm": 0.9571017026901245,
      "learning_rate": 0.00022772008965927363,
      "loss": 1.9775,
      "step": 133047
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.0197001695632935,
      "learning_rate": 0.00022771611960465698,
      "loss": 2.0214,
      "step": 133048
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.2942800521850586,
      "learning_rate": 0.00022771214956347936,
      "loss": 2.0554,
      "step": 133049
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.2883678674697876,
      "learning_rate": 0.00022770817953574128,
      "loss": 1.9838,
      "step": 133050
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.3398942947387695,
      "learning_rate": 0.00022770420952144364,
      "loss": 2.0696,
      "step": 133051
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.1021065711975098,
      "learning_rate": 0.00022770023952058715,
      "loss": 1.9841,
      "step": 133052
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.05045747756958,
      "learning_rate": 0.00022769626953317252,
      "loss": 2.299,
      "step": 133053
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.1157951354980469,
      "learning_rate": 0.00022769229955920057,
      "loss": 2.1955,
      "step": 133054
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.062274694442749,
      "learning_rate": 0.00022768832959867196,
      "loss": 1.9633,
      "step": 133055
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.09661066532135,
      "learning_rate": 0.0002276843596515876,
      "loss": 2.0236,
      "step": 133056
    },
    {
      "epoch": 1.73,
      "grad_norm": 0.9878349900245667,
      "learning_rate": 0.00022768038971794792,
      "loss": 2.0872,
      "step": 133057
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.0075300931930542,
      "learning_rate": 0.00022767641979775383,
      "loss": 2.1146,
      "step": 133058
    },
    {
      "epoch": 1.73,
      "grad_norm": 0.9715955853462219,
      "learning_rate": 0.0002276724498910061,
      "loss": 2.0217,
      "step": 133059
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.1568917036056519,
      "learning_rate": 0.00022766847999770536,
      "loss": 2.0437,
      "step": 133060
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.0127573013305664,
      "learning_rate": 0.00022766451011785243,
      "loss": 2.0218,
      "step": 133061
    },
    {
      "epoch": 1.73,
      "grad_norm": 0.9575557708740234,
      "learning_rate": 0.0002276605402514482,
      "loss": 1.9381,
      "step": 133062
    },
    {
      "epoch": 1.73,
      "grad_norm": 0.9586299657821655,
      "learning_rate": 0.0002276565703984931,
      "loss": 1.9886,
      "step": 133063
    },
    {
      "epoch": 1.73,
      "grad_norm": 0.9701096415519714,
      "learning_rate": 0.00022765260055898796,
      "loss": 1.9528,
      "step": 133064
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.0187225341796875,
      "learning_rate": 0.00022764863073293363,
      "loss": 2.2722,
      "step": 133065
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.1142300367355347,
      "learning_rate": 0.00022764466092033077,
      "loss": 2.234,
      "step": 133066
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.0382046699523926,
      "learning_rate": 0.0002276406911211801,
      "loss": 2.1827,
      "step": 133067
    },
    {
      "epoch": 1.73,
      "grad_norm": 0.9729663729667664,
      "learning_rate": 0.00022763672133548258,
      "loss": 2.0373,
      "step": 133068
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.0402809381484985,
      "learning_rate": 0.0002276327515632386,
      "loss": 1.9603,
      "step": 133069
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.013355016708374,
      "learning_rate": 0.00022762878180444904,
      "loss": 2.2031,
      "step": 133070
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.0060101747512817,
      "learning_rate": 0.00022762481205911467,
      "loss": 2.2042,
      "step": 133071
    },
    {
      "epoch": 1.73,
      "grad_norm": 0.9902456998825073,
      "learning_rate": 0.00022762084232723622,
      "loss": 2.1014,
      "step": 133072
    },
    {
      "epoch": 1.73,
      "grad_norm": 0.9631650447845459,
      "learning_rate": 0.0002276168726088144,
      "loss": 1.9497,
      "step": 133073
    },
    {
      "epoch": 1.73,
      "grad_norm": 0.96155846118927,
      "learning_rate": 0.00022761290290385013,
      "loss": 2.1359,
      "step": 133074
    },
    {
      "epoch": 1.73,
      "grad_norm": 0.9666972756385803,
      "learning_rate": 0.00022760893321234385,
      "loss": 1.9591,
      "step": 133075
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.0194813013076782,
      "learning_rate": 0.00022760496353429644,
      "loss": 2.0293,
      "step": 133076
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.0107545852661133,
      "learning_rate": 0.00022760099386970862,
      "loss": 2.0896,
      "step": 133077
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.0766679048538208,
      "learning_rate": 0.00022759702421858114,
      "loss": 2.2568,
      "step": 133078
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.141510248184204,
      "learning_rate": 0.00022759305458091474,
      "loss": 2.1433,
      "step": 133079
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.0347856283187866,
      "learning_rate": 0.00022758908495671022,
      "loss": 2.1086,
      "step": 133080
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.0669233798980713,
      "learning_rate": 0.00022758511534596827,
      "loss": 2.0547,
      "step": 133081
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.1067345142364502,
      "learning_rate": 0.00022758114574868954,
      "loss": 2.2698,
      "step": 133082
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.1066256761550903,
      "learning_rate": 0.00022757717616487484,
      "loss": 2.1831,
      "step": 133083
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.0215566158294678,
      "learning_rate": 0.00022757320659452495,
      "loss": 1.9081,
      "step": 133084
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.022318959236145,
      "learning_rate": 0.0002275692370376405,
      "loss": 2.2153,
      "step": 133085
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.0345327854156494,
      "learning_rate": 0.0002275652674942224,
      "loss": 1.9729,
      "step": 133086
    },
    {
      "epoch": 1.73,
      "grad_norm": 0.9909647703170776,
      "learning_rate": 0.00022756129796427122,
      "loss": 2.2391,
      "step": 133087
    },
    {
      "epoch": 1.73,
      "grad_norm": 0.9947810769081116,
      "learning_rate": 0.00022755732844778782,
      "loss": 2.2663,
      "step": 133088
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.007246971130371,
      "learning_rate": 0.00022755335894477283,
      "loss": 2.062,
      "step": 133089
    },
    {
      "epoch": 1.73,
      "grad_norm": 0.9764518737792969,
      "learning_rate": 0.000227549389455227,
      "loss": 2.1028,
      "step": 133090
    },
    {
      "epoch": 1.73,
      "grad_norm": 0.946646511554718,
      "learning_rate": 0.00022754541997915112,
      "loss": 2.0565,
      "step": 133091
    },
    {
      "epoch": 1.73,
      "grad_norm": 0.946836531162262,
      "learning_rate": 0.00022754145051654602,
      "loss": 2.1497,
      "step": 133092
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.0500744581222534,
      "learning_rate": 0.0002275374810674122,
      "loss": 2.272,
      "step": 133093
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.0666530132293701,
      "learning_rate": 0.0002275335116317506,
      "loss": 2.1989,
      "step": 133094
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.028061032295227,
      "learning_rate": 0.0002275295422095619,
      "loss": 2.0769,
      "step": 133095
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.0157010555267334,
      "learning_rate": 0.00022752557280084678,
      "loss": 2.166,
      "step": 133096
    },
    {
      "epoch": 1.73,
      "grad_norm": 0.9960718154907227,
      "learning_rate": 0.00022752160340560603,
      "loss": 2.2494,
      "step": 133097
    },
    {
      "epoch": 1.73,
      "grad_norm": 0.9747694134712219,
      "learning_rate": 0.00022751763402384045,
      "loss": 2.2451,
      "step": 133098
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.0752427577972412,
      "learning_rate": 0.0002275136646555506,
      "loss": 2.1777,
      "step": 133099
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.1167091131210327,
      "learning_rate": 0.00022750969530073738,
      "loss": 2.0115,
      "step": 133100
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.0717600584030151,
      "learning_rate": 0.00022750572595940145,
      "loss": 2.1204,
      "step": 133101
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.1411795616149902,
      "learning_rate": 0.00022750175663154368,
      "loss": 2.1869,
      "step": 133102
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.119384527206421,
      "learning_rate": 0.00022749778731716457,
      "loss": 2.1202,
      "step": 133103
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.0243606567382812,
      "learning_rate": 0.00022749381801626513,
      "loss": 2.0428,
      "step": 133104
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.0201624631881714,
      "learning_rate": 0.00022748984872884585,
      "loss": 1.8976,
      "step": 133105
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.0751646757125854,
      "learning_rate": 0.00022748587945490756,
      "loss": 1.9872,
      "step": 133106
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.1744370460510254,
      "learning_rate": 0.00022748191019445106,
      "loss": 1.8536,
      "step": 133107
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.1069706678390503,
      "learning_rate": 0.000227477940947477,
      "loss": 2.2231,
      "step": 133108
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.0578533411026,
      "learning_rate": 0.00022747397171398617,
      "loss": 2.1559,
      "step": 133109
    },
    {
      "epoch": 1.73,
      "grad_norm": 0.9535605311393738,
      "learning_rate": 0.00022747000249397943,
      "loss": 2.3226,
      "step": 133110
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.0881296396255493,
      "learning_rate": 0.00022746603328745727,
      "loss": 1.9405,
      "step": 133111
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.013924479484558,
      "learning_rate": 0.00022746206409442052,
      "loss": 2.2194,
      "step": 133112
    },
    {
      "epoch": 1.73,
      "grad_norm": 0.9994447231292725,
      "learning_rate": 0.00022745809491486995,
      "loss": 2.2538,
      "step": 133113
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.0350103378295898,
      "learning_rate": 0.00022745412574880631,
      "loss": 2.1831,
      "step": 133114
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.06820547580719,
      "learning_rate": 0.00022745015659623029,
      "loss": 2.1454,
      "step": 133115
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.3005499839782715,
      "learning_rate": 0.0002274461874571428,
      "loss": 2.2134,
      "step": 133116
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.1769529581069946,
      "learning_rate": 0.0002274422183315443,
      "loss": 2.0347,
      "step": 133117
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.020776629447937,
      "learning_rate": 0.00022743824921943564,
      "loss": 2.1313,
      "step": 133118
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.100759744644165,
      "learning_rate": 0.00022743428012081762,
      "loss": 2.1182,
      "step": 133119
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.1264395713806152,
      "learning_rate": 0.00022743031103569092,
      "loss": 2.118,
      "step": 133120
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.0062175989151,
      "learning_rate": 0.0002274263419640563,
      "loss": 2.1304,
      "step": 133121
    },
    {
      "epoch": 1.73,
      "grad_norm": 0.9700515866279602,
      "learning_rate": 0.00022742237290591447,
      "loss": 2.1271,
      "step": 133122
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.0428005456924438,
      "learning_rate": 0.00022741840386126633,
      "loss": 2.0182,
      "step": 133123
    },
    {
      "epoch": 1.73,
      "grad_norm": 0.9701060652732849,
      "learning_rate": 0.00022741443483011235,
      "loss": 2.1708,
      "step": 133124
    },
    {
      "epoch": 1.73,
      "grad_norm": 0.9755099415779114,
      "learning_rate": 0.00022741046581245338,
      "loss": 2.0613,
      "step": 133125
    },
    {
      "epoch": 1.73,
      "grad_norm": 0.9689168930053711,
      "learning_rate": 0.00022740649680829024,
      "loss": 1.9311,
      "step": 133126
    },
    {
      "epoch": 1.73,
      "grad_norm": 0.9813416600227356,
      "learning_rate": 0.00022740252781762354,
      "loss": 1.9387,
      "step": 133127
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.0756449699401855,
      "learning_rate": 0.0002273985588404541,
      "loss": 1.942,
      "step": 133128
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.061124563217163,
      "learning_rate": 0.00022739458987678277,
      "loss": 2.4585,
      "step": 133129
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.0611926317214966,
      "learning_rate": 0.00022739062092661002,
      "loss": 2.2161,
      "step": 133130
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.040305733680725,
      "learning_rate": 0.00022738665198993672,
      "loss": 2.0553,
      "step": 133131
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.011846899986267,
      "learning_rate": 0.00022738268306676362,
      "loss": 2.1146,
      "step": 133132
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.0461071729660034,
      "learning_rate": 0.00022737871415709146,
      "loss": 2.1886,
      "step": 133133
    },
    {
      "epoch": 1.73,
      "grad_norm": 0.9529227018356323,
      "learning_rate": 0.00022737474526092095,
      "loss": 2.1455,
      "step": 133134
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.068688988685608,
      "learning_rate": 0.00022737077637825301,
      "loss": 1.8738,
      "step": 133135
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.1290572881698608,
      "learning_rate": 0.00022736680750908803,
      "loss": 2.2108,
      "step": 133136
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.1245408058166504,
      "learning_rate": 0.00022736283865342697,
      "loss": 2.3404,
      "step": 133137
    },
    {
      "epoch": 1.73,
      "grad_norm": 0.9838349223136902,
      "learning_rate": 0.00022735886981127052,
      "loss": 1.9856,
      "step": 133138
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.0570030212402344,
      "learning_rate": 0.00022735490098261944,
      "loss": 1.9773,
      "step": 133139
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.0781341791152954,
      "learning_rate": 0.00022735093216747442,
      "loss": 2.0759,
      "step": 133140
    },
    {
      "epoch": 1.73,
      "grad_norm": 0.9905301332473755,
      "learning_rate": 0.0002273469633658364,
      "loss": 2.1995,
      "step": 133141
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.0975617170333862,
      "learning_rate": 0.0002273429945777058,
      "loss": 2.4385,
      "step": 133142
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.0257197618484497,
      "learning_rate": 0.0002273390258030835,
      "loss": 2.1145,
      "step": 133143
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.161656141281128,
      "learning_rate": 0.00022733505704197027,
      "loss": 2.2249,
      "step": 133144
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.080176830291748,
      "learning_rate": 0.00022733108829436682,
      "loss": 2.1892,
      "step": 133145
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.0758613348007202,
      "learning_rate": 0.00022732711956027386,
      "loss": 2.0617,
      "step": 133146
    },
    {
      "epoch": 1.73,
      "grad_norm": 0.981590211391449,
      "learning_rate": 0.00022732315083969236,
      "loss": 2.0481,
      "step": 133147
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.2025651931762695,
      "learning_rate": 0.00022731918213262265,
      "loss": 2.0536,
      "step": 133148
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.015006184577942,
      "learning_rate": 0.00022731521343906573,
      "loss": 1.8826,
      "step": 133149
    },
    {
      "epoch": 1.73,
      "grad_norm": 0.960459291934967,
      "learning_rate": 0.00022731124475902225,
      "loss": 1.9794,
      "step": 133150
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.1080917119979858,
      "learning_rate": 0.000227307276092493,
      "loss": 2.3095,
      "step": 133151
    },
    {
      "epoch": 1.73,
      "grad_norm": 0.9811270833015442,
      "learning_rate": 0.00022730330743947868,
      "loss": 2.1559,
      "step": 133152
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.2775427103042603,
      "learning_rate": 0.00022729933879998018,
      "loss": 2.1372,
      "step": 133153
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.1736687421798706,
      "learning_rate": 0.000227295370173998,
      "loss": 2.1338,
      "step": 133154
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.0588140487670898,
      "learning_rate": 0.00022729140156153298,
      "loss": 2.0801,
      "step": 133155
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.175734519958496,
      "learning_rate": 0.00022728743296258585,
      "loss": 2.088,
      "step": 133156
    },
    {
      "epoch": 1.73,
      "grad_norm": 0.9932524561882019,
      "learning_rate": 0.00022728346437715739,
      "loss": 2.1187,
      "step": 133157
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.0693368911743164,
      "learning_rate": 0.00022727949580524825,
      "loss": 2.1303,
      "step": 133158
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.021154761314392,
      "learning_rate": 0.0002272755272468594,
      "loss": 2.3223,
      "step": 133159
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.0193759202957153,
      "learning_rate": 0.00022727155870199125,
      "loss": 2.2072,
      "step": 133160
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.083123803138733,
      "learning_rate": 0.00022726759017064469,
      "loss": 2.1478,
      "step": 133161
    },
    {
      "epoch": 1.73,
      "grad_norm": 0.9766519069671631,
      "learning_rate": 0.00022726362165282048,
      "loss": 2.1533,
      "step": 133162
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.260263442993164,
      "learning_rate": 0.00022725965314851932,
      "loss": 2.0822,
      "step": 133163
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.013545274734497,
      "learning_rate": 0.00022725568465774192,
      "loss": 2.1416,
      "step": 133164
    },
    {
      "epoch": 1.73,
      "grad_norm": 0.9802374839782715,
      "learning_rate": 0.00022725171618048924,
      "loss": 1.838,
      "step": 133165
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.026659607887268,
      "learning_rate": 0.00022724774771676172,
      "loss": 1.9367,
      "step": 133166
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.1622346639633179,
      "learning_rate": 0.0002272437792665602,
      "loss": 2.2237,
      "step": 133167
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.1869704723358154,
      "learning_rate": 0.00022723981082988542,
      "loss": 2.2014,
      "step": 133168
    },
    {
      "epoch": 1.73,
      "grad_norm": 0.9988446235656738,
      "learning_rate": 0.00022723584240673817,
      "loss": 1.9854,
      "step": 133169
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.0456671714782715,
      "learning_rate": 0.0002272318739971191,
      "loss": 2.0567,
      "step": 133170
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.04836905002594,
      "learning_rate": 0.0002272279056010291,
      "loss": 2.3536,
      "step": 133171
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.1518380641937256,
      "learning_rate": 0.00022722393721846882,
      "loss": 1.9856,
      "step": 133172
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.0009793043136597,
      "learning_rate": 0.00022721996884943888,
      "loss": 2.1372,
      "step": 133173
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.0548675060272217,
      "learning_rate": 0.00022721600049394012,
      "loss": 2.1787,
      "step": 133174
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.0786210298538208,
      "learning_rate": 0.00022721203215197332,
      "loss": 2.4141,
      "step": 133175
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.0355738401412964,
      "learning_rate": 0.00022720806382353918,
      "loss": 2.4433,
      "step": 133176
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.0606412887573242,
      "learning_rate": 0.00022720409550863846,
      "loss": 2.203,
      "step": 133177
    },
    {
      "epoch": 1.73,
      "grad_norm": 0.9520168304443359,
      "learning_rate": 0.0002272001272072718,
      "loss": 1.9535,
      "step": 133178
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.5848358869552612,
      "learning_rate": 0.00022719615891944006,
      "loss": 2.3687,
      "step": 133179
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.0315797328948975,
      "learning_rate": 0.00022719219064514396,
      "loss": 2.0261,
      "step": 133180
    },
    {
      "epoch": 1.73,
      "grad_norm": 0.9669575691223145,
      "learning_rate": 0.00022718822238438412,
      "loss": 2.3279,
      "step": 133181
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.1072142124176025,
      "learning_rate": 0.00022718425413716141,
      "loss": 2.0258,
      "step": 133182
    },
    {
      "epoch": 1.73,
      "grad_norm": 0.9712843894958496,
      "learning_rate": 0.00022718028590347657,
      "loss": 2.0344,
      "step": 133183
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.160967469215393,
      "learning_rate": 0.0002271763176833302,
      "loss": 2.0643,
      "step": 133184
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.0166652202606201,
      "learning_rate": 0.00022717234947672318,
      "loss": 2.0095,
      "step": 133185
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.1099704504013062,
      "learning_rate": 0.00022716838128365618,
      "loss": 2.0279,
      "step": 133186
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.039172649383545,
      "learning_rate": 0.00022716441310412997,
      "loss": 1.9545,
      "step": 133187
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.0241504907608032,
      "learning_rate": 0.00022716044493814523,
      "loss": 2.2803,
      "step": 133188
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.0577645301818848,
      "learning_rate": 0.00022715647678570277,
      "loss": 2.2992,
      "step": 133189
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.0725574493408203,
      "learning_rate": 0.00022715250864680333,
      "loss": 2.3213,
      "step": 133190
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.0375416278839111,
      "learning_rate": 0.00022714854052144758,
      "loss": 2.0471,
      "step": 133191
    },
    {
      "epoch": 1.73,
      "grad_norm": 0.9964499473571777,
      "learning_rate": 0.00022714457240963622,
      "loss": 2.2728,
      "step": 133192
    },
    {
      "epoch": 1.73,
      "grad_norm": 0.940991997718811,
      "learning_rate": 0.00022714060431137012,
      "loss": 2.0405,
      "step": 133193
    },
    {
      "epoch": 1.73,
      "grad_norm": 0.960331916809082,
      "learning_rate": 0.00022713663622665002,
      "loss": 2.0611,
      "step": 133194
    },
    {
      "epoch": 1.73,
      "grad_norm": 0.936095654964447,
      "learning_rate": 0.0002271326681554765,
      "loss": 2.2878,
      "step": 133195
    },
    {
      "epoch": 1.73,
      "grad_norm": 0.9911811351776123,
      "learning_rate": 0.00022712870009785049,
      "loss": 2.2722,
      "step": 133196
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.0172539949417114,
      "learning_rate": 0.00022712473205377254,
      "loss": 2.1019,
      "step": 133197
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.0236438512802124,
      "learning_rate": 0.0002271207640232435,
      "loss": 2.042,
      "step": 133198
    },
    {
      "epoch": 1.73,
      "grad_norm": 0.9898977279663086,
      "learning_rate": 0.00022711679600626405,
      "loss": 2.162,
      "step": 133199
    },
    {
      "epoch": 1.73,
      "grad_norm": 0.9878726005554199,
      "learning_rate": 0.000227112828002835,
      "loss": 2.002,
      "step": 133200
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.0992114543914795,
      "learning_rate": 0.00022710886001295712,
      "loss": 2.1169,
      "step": 133201
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.1476305723190308,
      "learning_rate": 0.00022710489203663106,
      "loss": 2.1771,
      "step": 133202
    },
    {
      "epoch": 1.73,
      "grad_norm": 0.9901612997055054,
      "learning_rate": 0.00022710092407385752,
      "loss": 1.956,
      "step": 133203
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.0384782552719116,
      "learning_rate": 0.0002270969561246373,
      "loss": 1.9903,
      "step": 133204
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.1114158630371094,
      "learning_rate": 0.00022709298818897108,
      "loss": 2.1895,
      "step": 133205
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.0747214555740356,
      "learning_rate": 0.0002270890202668597,
      "loss": 2.1671,
      "step": 133206
    },
    {
      "epoch": 1.73,
      "grad_norm": 0.9996232986450195,
      "learning_rate": 0.00022708505235830383,
      "loss": 1.9455,
      "step": 133207
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.0095300674438477,
      "learning_rate": 0.00022708108446330437,
      "loss": 2.2007,
      "step": 133208
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.2730481624603271,
      "learning_rate": 0.00022707711658186177,
      "loss": 2.1477,
      "step": 133209
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.075442910194397,
      "learning_rate": 0.00022707314871397695,
      "loss": 2.3862,
      "step": 133210
    },
    {
      "epoch": 1.73,
      "grad_norm": 0.9453955292701721,
      "learning_rate": 0.00022706918085965054,
      "loss": 2.1641,
      "step": 133211
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.000855565071106,
      "learning_rate": 0.00022706521301888338,
      "loss": 2.3736,
      "step": 133212
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.0875240564346313,
      "learning_rate": 0.00022706124519167618,
      "loss": 2.1235,
      "step": 133213
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.0564019680023193,
      "learning_rate": 0.00022705727737802982,
      "loss": 2.2971,
      "step": 133214
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.139438509941101,
      "learning_rate": 0.00022705330957794472,
      "loss": 2.1757,
      "step": 133215
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.0761642456054688,
      "learning_rate": 0.00022704934179142177,
      "loss": 1.9424,
      "step": 133216
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.1005889177322388,
      "learning_rate": 0.00022704537401846178,
      "loss": 2.2128,
      "step": 133217
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.0831918716430664,
      "learning_rate": 0.0002270414062590654,
      "loss": 2.2464,
      "step": 133218
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.0826948881149292,
      "learning_rate": 0.00022703743851323343,
      "loss": 2.2005,
      "step": 133219
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.1075565814971924,
      "learning_rate": 0.00022703347078096667,
      "loss": 2.32,
      "step": 133220
    },
    {
      "epoch": 1.73,
      "grad_norm": 0.9884594678878784,
      "learning_rate": 0.00022702950306226566,
      "loss": 2.1804,
      "step": 133221
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.005615234375,
      "learning_rate": 0.00022702553535713121,
      "loss": 2.2974,
      "step": 133222
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.0904370546340942,
      "learning_rate": 0.00022702156766556413,
      "loss": 1.8834,
      "step": 133223
    },
    {
      "epoch": 1.73,
      "grad_norm": 0.9549114108085632,
      "learning_rate": 0.0002270175999875651,
      "loss": 2.1826,
      "step": 133224
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.1143198013305664,
      "learning_rate": 0.00022701363232313486,
      "loss": 2.0438,
      "step": 133225
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.0112138986587524,
      "learning_rate": 0.0002270096646722743,
      "loss": 2.0364,
      "step": 133226
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.196766972541809,
      "learning_rate": 0.00022700569703498392,
      "loss": 2.1775,
      "step": 133227
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.0281469821929932,
      "learning_rate": 0.00022700172941126455,
      "loss": 2.0736,
      "step": 133228
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.0376574993133545,
      "learning_rate": 0.00022699776180111692,
      "loss": 2.1575,
      "step": 133229
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.1177767515182495,
      "learning_rate": 0.00022699379420454177,
      "loss": 1.8543,
      "step": 133230
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.111701488494873,
      "learning_rate": 0.0002269898266215399,
      "loss": 2.1786,
      "step": 133231
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.3770064115524292,
      "learning_rate": 0.0002269858590521121,
      "loss": 1.9649,
      "step": 133232
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.0252525806427002,
      "learning_rate": 0.00022698189149625888,
      "loss": 2.2082,
      "step": 133233
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.2438604831695557,
      "learning_rate": 0.0002269779239539811,
      "loss": 2.071,
      "step": 133234
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.2159698009490967,
      "learning_rate": 0.0002269739564252795,
      "loss": 2.0484,
      "step": 133235
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.0624891519546509,
      "learning_rate": 0.00022696998891015484,
      "loss": 2.1226,
      "step": 133236
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.0611988306045532,
      "learning_rate": 0.00022696602140860782,
      "loss": 2.1323,
      "step": 133237
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.149025797843933,
      "learning_rate": 0.00022696205392063934,
      "loss": 2.0436,
      "step": 133238
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.0236968994140625,
      "learning_rate": 0.00022695808644624985,
      "loss": 1.9823,
      "step": 133239
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.0717840194702148,
      "learning_rate": 0.00022695411898544023,
      "loss": 2.1636,
      "step": 133240
    },
    {
      "epoch": 1.73,
      "grad_norm": 0.9581949710845947,
      "learning_rate": 0.00022695015153821122,
      "loss": 1.992,
      "step": 133241
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.0682320594787598,
      "learning_rate": 0.00022694618410456352,
      "loss": 2.1752,
      "step": 133242
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.0958017110824585,
      "learning_rate": 0.00022694221668449795,
      "loss": 1.9763,
      "step": 133243
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.0688681602478027,
      "learning_rate": 0.00022693824927801535,
      "loss": 2.218,
      "step": 133244
    },
    {
      "epoch": 1.73,
      "grad_norm": 0.9903708100318909,
      "learning_rate": 0.00022693428188511614,
      "loss": 2.2464,
      "step": 133245
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.1755478382110596,
      "learning_rate": 0.00022693031450580123,
      "loss": 2.3772,
      "step": 133246
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.0062496662139893,
      "learning_rate": 0.00022692634714007137,
      "loss": 2.064,
      "step": 133247
    },
    {
      "epoch": 1.73,
      "grad_norm": 1.3623301982879639,
      "learning_rate": 0.00022692237978792724,
      "loss": 2.201,
      "step": 133248
    },
    {
      "epoch": 1.73,
      "grad_norm": 0.9152294397354126,
      "learning_rate": 0.00022691841244936969,
      "loss": 2.0163,
      "step": 133249
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.1299172639846802,
      "learning_rate": 0.00022691444512439946,
      "loss": 2.0277,
      "step": 133250
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.068434476852417,
      "learning_rate": 0.00022691047781301708,
      "loss": 2.1491,
      "step": 133251
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.037858486175537,
      "learning_rate": 0.00022690651051522346,
      "loss": 2.2654,
      "step": 133252
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.0291943550109863,
      "learning_rate": 0.0002269025432310193,
      "loss": 2.2306,
      "step": 133253
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.0091251134872437,
      "learning_rate": 0.0002268985759604053,
      "loss": 2.0845,
      "step": 133254
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.1477956771850586,
      "learning_rate": 0.00022689460870338227,
      "loss": 2.0689,
      "step": 133255
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.038001298904419,
      "learning_rate": 0.0002268906414599509,
      "loss": 2.1045,
      "step": 133256
    },
    {
      "epoch": 1.74,
      "grad_norm": 0.9889510869979858,
      "learning_rate": 0.00022688667423011204,
      "loss": 2.2535,
      "step": 133257
    },
    {
      "epoch": 1.74,
      "grad_norm": 0.9201263785362244,
      "learning_rate": 0.00022688270701386624,
      "loss": 1.889,
      "step": 133258
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.0654793977737427,
      "learning_rate": 0.00022687873981121432,
      "loss": 2.0887,
      "step": 133259
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.0001344680786133,
      "learning_rate": 0.000226874772622157,
      "loss": 2.2146,
      "step": 133260
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.028793215751648,
      "learning_rate": 0.00022687080544669505,
      "loss": 2.1,
      "step": 133261
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.0112098455429077,
      "learning_rate": 0.0002268668382848292,
      "loss": 1.9417,
      "step": 133262
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.0099904537200928,
      "learning_rate": 0.00022686287113656025,
      "loss": 2.2513,
      "step": 133263
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.0530329942703247,
      "learning_rate": 0.00022685890400188884,
      "loss": 2.2379,
      "step": 133264
    },
    {
      "epoch": 1.74,
      "grad_norm": 0.9755508899688721,
      "learning_rate": 0.00022685493688081573,
      "loss": 2.0596,
      "step": 133265
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.1560667753219604,
      "learning_rate": 0.00022685096977334164,
      "loss": 2.1886,
      "step": 133266
    },
    {
      "epoch": 1.74,
      "grad_norm": 0.962945818901062,
      "learning_rate": 0.00022684700267946734,
      "loss": 2.0215,
      "step": 133267
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.0138019323349,
      "learning_rate": 0.00022684303559919355,
      "loss": 1.9869,
      "step": 133268
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.0917448997497559,
      "learning_rate": 0.0002268390685325211,
      "loss": 2.2992,
      "step": 133269
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.2304255962371826,
      "learning_rate": 0.0002268351014794506,
      "loss": 2.0849,
      "step": 133270
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.1166552305221558,
      "learning_rate": 0.00022683113443998278,
      "loss": 2.1904,
      "step": 133271
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.0308011770248413,
      "learning_rate": 0.00022682716741411857,
      "loss": 2.3794,
      "step": 133272
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.0921741724014282,
      "learning_rate": 0.00022682320040185849,
      "loss": 2.1048,
      "step": 133273
    },
    {
      "epoch": 1.74,
      "grad_norm": 0.9815616011619568,
      "learning_rate": 0.00022681923340320332,
      "loss": 2.1364,
      "step": 133274
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.0785715579986572,
      "learning_rate": 0.00022681526641815393,
      "loss": 2.1254,
      "step": 133275
    },
    {
      "epoch": 1.74,
      "grad_norm": 0.9894668459892273,
      "learning_rate": 0.00022681129944671088,
      "loss": 2.1817,
      "step": 133276
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.089321255683899,
      "learning_rate": 0.00022680733248887499,
      "loss": 2.0836,
      "step": 133277
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.0304138660430908,
      "learning_rate": 0.00022680336554464702,
      "loss": 2.1498,
      "step": 133278
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.0906972885131836,
      "learning_rate": 0.00022679939861402777,
      "loss": 2.1055,
      "step": 133279
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.0768275260925293,
      "learning_rate": 0.00022679543169701777,
      "loss": 2.0347,
      "step": 133280
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.0264984369277954,
      "learning_rate": 0.00022679146479361799,
      "loss": 2.1335,
      "step": 133281
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.0117651224136353,
      "learning_rate": 0.00022678749790382897,
      "loss": 2.2352,
      "step": 133282
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.0456805229187012,
      "learning_rate": 0.00022678353102765153,
      "loss": 2.0527,
      "step": 133283
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.1384649276733398,
      "learning_rate": 0.00022677956416508642,
      "loss": 2.0452,
      "step": 133284
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.0352331399917603,
      "learning_rate": 0.0002267755973161344,
      "loss": 2.0996,
      "step": 133285
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.0884754657745361,
      "learning_rate": 0.00022677163048079622,
      "loss": 2.2238,
      "step": 133286
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.0504709482192993,
      "learning_rate": 0.0002267676636590726,
      "loss": 2.1408,
      "step": 133287
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.084614634513855,
      "learning_rate": 0.00022676369685096414,
      "loss": 2.1042,
      "step": 133288
    },
    {
      "epoch": 1.74,
      "grad_norm": 0.9841717481613159,
      "learning_rate": 0.00022675973005647173,
      "loss": 2.2086,
      "step": 133289
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.077268362045288,
      "learning_rate": 0.00022675576327559605,
      "loss": 2.0973,
      "step": 133290
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.0448771715164185,
      "learning_rate": 0.00022675179650833788,
      "loss": 2.0079,
      "step": 133291
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.1292897462844849,
      "learning_rate": 0.0002267478297546979,
      "loss": 2.0963,
      "step": 133292
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.02153480052948,
      "learning_rate": 0.00022674386301467704,
      "loss": 1.9655,
      "step": 133293
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.2278534173965454,
      "learning_rate": 0.00022673989628827575,
      "loss": 1.8498,
      "step": 133294
    },
    {
      "epoch": 1.74,
      "grad_norm": 0.9984342455863953,
      "learning_rate": 0.00022673592957549486,
      "loss": 2.2233,
      "step": 133295
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.0263322591781616,
      "learning_rate": 0.0002267319628763352,
      "loss": 2.0275,
      "step": 133296
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.0425485372543335,
      "learning_rate": 0.0002267279961907974,
      "loss": 2.4215,
      "step": 133297
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.0420019626617432,
      "learning_rate": 0.00022672402951888225,
      "loss": 1.9226,
      "step": 133298
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.0476661920547485,
      "learning_rate": 0.00022672006286059065,
      "loss": 2.0615,
      "step": 133299
    },
    {
      "epoch": 1.74,
      "grad_norm": 0.925613284111023,
      "learning_rate": 0.00022671609621592302,
      "loss": 1.8583,
      "step": 133300
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.0656602382659912,
      "learning_rate": 0.00022671212958488027,
      "loss": 2.1867,
      "step": 133301
    },
    {
      "epoch": 1.74,
      "grad_norm": 0.9996219873428345,
      "learning_rate": 0.0002267081629674631,
      "loss": 2.1916,
      "step": 133302
    },
    {
      "epoch": 1.74,
      "grad_norm": 0.9993304014205933,
      "learning_rate": 0.00022670419636367227,
      "loss": 2.1894,
      "step": 133303
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.0665374994277954,
      "learning_rate": 0.00022670022977350852,
      "loss": 2.0567,
      "step": 133304
    },
    {
      "epoch": 1.74,
      "grad_norm": 0.9495870471000671,
      "learning_rate": 0.0002266962631969727,
      "loss": 2.111,
      "step": 133305
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.0391279458999634,
      "learning_rate": 0.0002266922966340653,
      "loss": 2.03,
      "step": 133306
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.0784913301467896,
      "learning_rate": 0.00022668833008478723,
      "loss": 1.9918,
      "step": 133307
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.0453824996948242,
      "learning_rate": 0.00022668436354913912,
      "loss": 2.1353,
      "step": 133308
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.0767459869384766,
      "learning_rate": 0.00022668039702712178,
      "loss": 2.1686,
      "step": 133309
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.028748869895935,
      "learning_rate": 0.00022667643051873596,
      "loss": 2.0613,
      "step": 133310
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.260095238685608,
      "learning_rate": 0.00022667246402398251,
      "loss": 2.0849,
      "step": 133311
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.1153064966201782,
      "learning_rate": 0.00022666849754286188,
      "loss": 2.0186,
      "step": 133312
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.039965271949768,
      "learning_rate": 0.00022666453107537496,
      "loss": 2.2678,
      "step": 133313
    },
    {
      "epoch": 1.74,
      "grad_norm": 0.9818684458732605,
      "learning_rate": 0.00022666056462152253,
      "loss": 2.1581,
      "step": 133314
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.127995491027832,
      "learning_rate": 0.00022665659818130526,
      "loss": 2.1611,
      "step": 133315
    },
    {
      "epoch": 1.74,
      "grad_norm": 0.9141688942909241,
      "learning_rate": 0.00022665263175472388,
      "loss": 2.1064,
      "step": 133316
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.0191069841384888,
      "learning_rate": 0.00022664866534177931,
      "loss": 1.9773,
      "step": 133317
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.1729520559310913,
      "learning_rate": 0.00022664469894247206,
      "loss": 2.176,
      "step": 133318
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.0754778385162354,
      "learning_rate": 0.00022664073255680288,
      "loss": 2.4314,
      "step": 133319
    },
    {
      "epoch": 1.74,
      "grad_norm": 0.9436277747154236,
      "learning_rate": 0.0002266367661847726,
      "loss": 1.9108,
      "step": 133320
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.1014840602874756,
      "learning_rate": 0.00022663279982638193,
      "loss": 2.2678,
      "step": 133321
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.030838966369629,
      "learning_rate": 0.0002266288334816316,
      "loss": 2.1312,
      "step": 133322
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.059127688407898,
      "learning_rate": 0.00022662486715052234,
      "loss": 2.1127,
      "step": 133323
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.1421334743499756,
      "learning_rate": 0.00022662090083305506,
      "loss": 2.2953,
      "step": 133324
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.136753797531128,
      "learning_rate": 0.00022661693452923022,
      "loss": 2.0169,
      "step": 133325
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.0760738849639893,
      "learning_rate": 0.00022661296823904862,
      "loss": 2.1613,
      "step": 133326
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.038642168045044,
      "learning_rate": 0.0002266090019625111,
      "loss": 2.0598,
      "step": 133327
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.0931941270828247,
      "learning_rate": 0.00022660503569961837,
      "loss": 2.0186,
      "step": 133328
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.0407109260559082,
      "learning_rate": 0.0002266010694503711,
      "loss": 2.1736,
      "step": 133329
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.1673829555511475,
      "learning_rate": 0.00022659710321477024,
      "loss": 2.1577,
      "step": 133330
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.283452033996582,
      "learning_rate": 0.0002265931369928162,
      "loss": 2.0872,
      "step": 133331
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.1809076070785522,
      "learning_rate": 0.00022658917078450993,
      "loss": 2.1676,
      "step": 133332
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.020126223564148,
      "learning_rate": 0.0002265852045898521,
      "loss": 2.0944,
      "step": 133333
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.077819585800171,
      "learning_rate": 0.00022658123840884345,
      "loss": 2.2537,
      "step": 133334
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.4365694522857666,
      "learning_rate": 0.00022657727224148475,
      "loss": 2.1252,
      "step": 133335
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.143799901008606,
      "learning_rate": 0.00022657330608777683,
      "loss": 2.2958,
      "step": 133336
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.0147126913070679,
      "learning_rate": 0.0002265693399477202,
      "loss": 2.1397,
      "step": 133337
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.0624488592147827,
      "learning_rate": 0.00022656537382131572,
      "loss": 2.1923,
      "step": 133338
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.0770069360733032,
      "learning_rate": 0.00022656140770856414,
      "loss": 2.0959,
      "step": 133339
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.0517903566360474,
      "learning_rate": 0.00022655744160946615,
      "loss": 2.3254,
      "step": 133340
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.1804511547088623,
      "learning_rate": 0.00022655347552402253,
      "loss": 2.1117,
      "step": 133341
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.1121224164962769,
      "learning_rate": 0.00022654950945223413,
      "loss": 2.0333,
      "step": 133342
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.0695340633392334,
      "learning_rate": 0.00022654554339410142,
      "loss": 2.0204,
      "step": 133343
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.0126993656158447,
      "learning_rate": 0.00022654157734962533,
      "loss": 1.9645,
      "step": 133344
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.0036154985427856,
      "learning_rate": 0.0002265376113188065,
      "loss": 2.3191,
      "step": 133345
    },
    {
      "epoch": 1.74,
      "grad_norm": 0.9687970876693726,
      "learning_rate": 0.0002265336453016457,
      "loss": 2.0997,
      "step": 133346
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.0630970001220703,
      "learning_rate": 0.00022652967929814375,
      "loss": 1.8845,
      "step": 133347
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.0899161100387573,
      "learning_rate": 0.00022652571330830133,
      "loss": 2.1509,
      "step": 133348
    },
    {
      "epoch": 1.74,
      "grad_norm": 0.9469950199127197,
      "learning_rate": 0.00022652174733211918,
      "loss": 2.0711,
      "step": 133349
    },
    {
      "epoch": 1.74,
      "grad_norm": 0.987408459186554,
      "learning_rate": 0.00022651778136959796,
      "loss": 2.2345,
      "step": 133350
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.0262198448181152,
      "learning_rate": 0.0002265138154207385,
      "loss": 2.2912,
      "step": 133351
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.0758943557739258,
      "learning_rate": 0.00022650984948554146,
      "loss": 2.2087,
      "step": 133352
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.092018723487854,
      "learning_rate": 0.00022650588356400769,
      "loss": 2.1357,
      "step": 133353
    },
    {
      "epoch": 1.74,
      "grad_norm": 0.9487298130989075,
      "learning_rate": 0.00022650191765613786,
      "loss": 2.0659,
      "step": 133354
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.001787781715393,
      "learning_rate": 0.00022649795176193265,
      "loss": 2.2998,
      "step": 133355
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.1926651000976562,
      "learning_rate": 0.00022649398588139296,
      "loss": 2.4314,
      "step": 133356
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.0340571403503418,
      "learning_rate": 0.00022649002001451935,
      "loss": 2.0501,
      "step": 133357
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.2190473079681396,
      "learning_rate": 0.00022648605416131261,
      "loss": 2.1431,
      "step": 133358
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.0940730571746826,
      "learning_rate": 0.00022648208832177352,
      "loss": 1.8251,
      "step": 133359
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.088428258895874,
      "learning_rate": 0.0002264781224959029,
      "loss": 2.3218,
      "step": 133360
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.0191855430603027,
      "learning_rate": 0.00022647415668370125,
      "loss": 2.0654,
      "step": 133361
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.073296070098877,
      "learning_rate": 0.00022647019088516948,
      "loss": 2.1066,
      "step": 133362
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.0772031545639038,
      "learning_rate": 0.00022646622510030833,
      "loss": 2.0622,
      "step": 133363
    },
    {
      "epoch": 1.74,
      "grad_norm": 0.970618724822998,
      "learning_rate": 0.00022646225932911848,
      "loss": 2.1321,
      "step": 133364
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.107847809791565,
      "learning_rate": 0.00022645829357160067,
      "loss": 2.1312,
      "step": 133365
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.0491122007369995,
      "learning_rate": 0.00022645432782775568,
      "loss": 2.2222,
      "step": 133366
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.021791696548462,
      "learning_rate": 0.0002264503620975842,
      "loss": 2.1192,
      "step": 133367
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.0738507509231567,
      "learning_rate": 0.00022644639638108694,
      "loss": 2.2551,
      "step": 133368
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.064901351928711,
      "learning_rate": 0.0002264424306782647,
      "loss": 2.0393,
      "step": 133369
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.0770397186279297,
      "learning_rate": 0.00022643846498911823,
      "loss": 2.0879,
      "step": 133370
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.0281866788864136,
      "learning_rate": 0.00022643449931364828,
      "loss": 2.1021,
      "step": 133371
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.0025473833084106,
      "learning_rate": 0.00022643053365185552,
      "loss": 2.0796,
      "step": 133372
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.0596362352371216,
      "learning_rate": 0.0002264265680037407,
      "loss": 2.2214,
      "step": 133373
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.2031350135803223,
      "learning_rate": 0.00022642260236930455,
      "loss": 2.0417,
      "step": 133374
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.1073216199874878,
      "learning_rate": 0.0002264186367485478,
      "loss": 2.0241,
      "step": 133375
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.1941628456115723,
      "learning_rate": 0.00022641467114147126,
      "loss": 2.0495,
      "step": 133376
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.1102837324142456,
      "learning_rate": 0.00022641070554807555,
      "loss": 1.9113,
      "step": 133377
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.0915963649749756,
      "learning_rate": 0.00022640673996836164,
      "loss": 2.179,
      "step": 133378
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.0368376970291138,
      "learning_rate": 0.00022640277440233003,
      "loss": 2.0433,
      "step": 133379
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.0245556831359863,
      "learning_rate": 0.00022639880884998148,
      "loss": 1.9821,
      "step": 133380
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.0032001733779907,
      "learning_rate": 0.0002263948433113168,
      "loss": 2.0178,
      "step": 133381
    },
    {
      "epoch": 1.74,
      "grad_norm": 0.9784495830535889,
      "learning_rate": 0.0002263908777863367,
      "loss": 2.0842,
      "step": 133382
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.0736581087112427,
      "learning_rate": 0.0002263869122750419,
      "loss": 2.1298,
      "step": 133383
    },
    {
      "epoch": 1.74,
      "grad_norm": 0.9645262956619263,
      "learning_rate": 0.00022638294677743335,
      "loss": 2.0219,
      "step": 133384
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.1377781629562378,
      "learning_rate": 0.00022637898129351143,
      "loss": 2.0246,
      "step": 133385
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.0308661460876465,
      "learning_rate": 0.00022637501582327706,
      "loss": 2.0074,
      "step": 133386
    },
    {
      "epoch": 1.74,
      "grad_norm": 0.937717854976654,
      "learning_rate": 0.00022637105036673095,
      "loss": 2.0837,
      "step": 133387
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.0595438480377197,
      "learning_rate": 0.00022636708492387386,
      "loss": 2.2131,
      "step": 133388
    },
    {
      "epoch": 1.74,
      "grad_norm": 0.9546724557876587,
      "learning_rate": 0.00022636311949470654,
      "loss": 2.2514,
      "step": 133389
    },
    {
      "epoch": 1.74,
      "grad_norm": 0.9785993695259094,
      "learning_rate": 0.00022635915407922966,
      "loss": 2.0351,
      "step": 133390
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.0454424619674683,
      "learning_rate": 0.0002263551886774442,
      "loss": 2.1245,
      "step": 133391
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.1037973165512085,
      "learning_rate": 0.0002263512232893505,
      "loss": 2.2781,
      "step": 133392
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.0020345449447632,
      "learning_rate": 0.00022634725791494955,
      "loss": 1.8707,
      "step": 133393
    },
    {
      "epoch": 1.74,
      "grad_norm": 0.997412919998169,
      "learning_rate": 0.000226343292554242,
      "loss": 2.0396,
      "step": 133394
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.0520600080490112,
      "learning_rate": 0.00022633932720722864,
      "loss": 1.9235,
      "step": 133395
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.1706876754760742,
      "learning_rate": 0.0002263353618739102,
      "loss": 1.9778,
      "step": 133396
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.0222970247268677,
      "learning_rate": 0.0002263313965542875,
      "loss": 2.3277,
      "step": 133397
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.2362432479858398,
      "learning_rate": 0.0002263274312483611,
      "loss": 2.0349,
      "step": 133398
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.2213648557662964,
      "learning_rate": 0.0002263234659561318,
      "loss": 2.1223,
      "step": 133399
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.0041518211364746,
      "learning_rate": 0.00022631950067760034,
      "loss": 2.098,
      "step": 133400
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.1564407348632812,
      "learning_rate": 0.0002263155354127675,
      "loss": 2.0331,
      "step": 133401
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.3985230922698975,
      "learning_rate": 0.00022631157016163398,
      "loss": 2.0297,
      "step": 133402
    },
    {
      "epoch": 1.74,
      "grad_norm": 0.984211802482605,
      "learning_rate": 0.00022630760492420068,
      "loss": 2.2761,
      "step": 133403
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.0431596040725708,
      "learning_rate": 0.00022630363970046802,
      "loss": 2.4208,
      "step": 133404
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.038997769355774,
      "learning_rate": 0.00022629967449043693,
      "loss": 2.1281,
      "step": 133405
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.2827068567276,
      "learning_rate": 0.00022629570929410812,
      "loss": 2.2321,
      "step": 133406
    },
    {
      "epoch": 1.74,
      "grad_norm": 0.9724422097206116,
      "learning_rate": 0.00022629174411148233,
      "loss": 1.9448,
      "step": 133407
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.0671831369400024,
      "learning_rate": 0.0002262877789425603,
      "loss": 2.2507,
      "step": 133408
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.0083584785461426,
      "learning_rate": 0.0002262838137873429,
      "loss": 2.2816,
      "step": 133409
    },
    {
      "epoch": 1.74,
      "grad_norm": 0.9971704483032227,
      "learning_rate": 0.00022627984864583058,
      "loss": 2.0363,
      "step": 133410
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.153477430343628,
      "learning_rate": 0.00022627588351802422,
      "loss": 2.106,
      "step": 133411
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.0602468252182007,
      "learning_rate": 0.00022627191840392458,
      "loss": 2.098,
      "step": 133412
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.0055267810821533,
      "learning_rate": 0.0002262679533035324,
      "loss": 2.0044,
      "step": 133413
    },
    {
      "epoch": 1.74,
      "grad_norm": 0.9938296675682068,
      "learning_rate": 0.00022626398821684837,
      "loss": 1.8742,
      "step": 133414
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.0207533836364746,
      "learning_rate": 0.0002262600231438734,
      "loss": 2.2471,
      "step": 133415
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.0634455680847168,
      "learning_rate": 0.00022625605808460793,
      "loss": 2.1034,
      "step": 133416
    },
    {
      "epoch": 1.74,
      "grad_norm": 0.9548957347869873,
      "learning_rate": 0.0002262520930390529,
      "loss": 2.1144,
      "step": 133417
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.0047696828842163,
      "learning_rate": 0.00022624812800720896,
      "loss": 2.1502,
      "step": 133418
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.0219694375991821,
      "learning_rate": 0.00022624416298907692,
      "loss": 2.017,
      "step": 133419
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.0566205978393555,
      "learning_rate": 0.00022624019798465745,
      "loss": 2.1016,
      "step": 133420
    },
    {
      "epoch": 1.74,
      "grad_norm": 0.9697130918502808,
      "learning_rate": 0.00022623623299395144,
      "loss": 1.9673,
      "step": 133421
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.008225917816162,
      "learning_rate": 0.00022623226801695941,
      "loss": 2.0935,
      "step": 133422
    },
    {
      "epoch": 1.74,
      "grad_norm": 0.9882944822311401,
      "learning_rate": 0.00022622830305368217,
      "loss": 2.1736,
      "step": 133423
    },
    {
      "epoch": 1.74,
      "grad_norm": 0.9887472987174988,
      "learning_rate": 0.00022622433810412051,
      "loss": 2.0838,
      "step": 133424
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.0058706998825073,
      "learning_rate": 0.0002262203731682751,
      "loss": 2.0377,
      "step": 133425
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.0405468940734863,
      "learning_rate": 0.00022621640824614672,
      "loss": 2.1467,
      "step": 133426
    },
    {
      "epoch": 1.74,
      "grad_norm": 0.9817763566970825,
      "learning_rate": 0.00022621244333773625,
      "loss": 2.1739,
      "step": 133427
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.0703493356704712,
      "learning_rate": 0.00022620847844304415,
      "loss": 1.9465,
      "step": 133428
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.3513985872268677,
      "learning_rate": 0.00022620451356207126,
      "loss": 2.087,
      "step": 133429
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.0740612745285034,
      "learning_rate": 0.0002262005486948184,
      "loss": 2.0324,
      "step": 133430
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.0951874256134033,
      "learning_rate": 0.0002261965838412862,
      "loss": 2.1226,
      "step": 133431
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.1881518363952637,
      "learning_rate": 0.00022619261900147546,
      "loss": 1.9344,
      "step": 133432
    },
    {
      "epoch": 1.74,
      "grad_norm": 0.977898359298706,
      "learning_rate": 0.00022618865417538695,
      "loss": 2.0103,
      "step": 133433
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.0334047079086304,
      "learning_rate": 0.0002261846893630214,
      "loss": 1.8203,
      "step": 133434
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.1382057666778564,
      "learning_rate": 0.00022618072456437945,
      "loss": 2.0839,
      "step": 133435
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.0490148067474365,
      "learning_rate": 0.00022617675977946186,
      "loss": 1.7573,
      "step": 133436
    },
    {
      "epoch": 1.74,
      "grad_norm": 0.9969480633735657,
      "learning_rate": 0.00022617279500826944,
      "loss": 2.1295,
      "step": 133437
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.1634029150009155,
      "learning_rate": 0.00022616883025080288,
      "loss": 2.1136,
      "step": 133438
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.065437912940979,
      "learning_rate": 0.00022616486550706296,
      "loss": 2.0278,
      "step": 133439
    },
    {
      "epoch": 1.74,
      "grad_norm": 0.9575338959693909,
      "learning_rate": 0.00022616090077705038,
      "loss": 2.0323,
      "step": 133440
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.0219686031341553,
      "learning_rate": 0.00022615693606076588,
      "loss": 2.087,
      "step": 133441
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.0814204216003418,
      "learning_rate": 0.0002261529713582102,
      "loss": 1.8924,
      "step": 133442
    },
    {
      "epoch": 1.74,
      "grad_norm": 0.9590216279029846,
      "learning_rate": 0.00022614900666938402,
      "loss": 2.1084,
      "step": 133443
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.0093580484390259,
      "learning_rate": 0.00022614504199428817,
      "loss": 2.2694,
      "step": 133444
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.0141929388046265,
      "learning_rate": 0.0002261410773329234,
      "loss": 1.9549,
      "step": 133445
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.027004361152649,
      "learning_rate": 0.00022613711268529033,
      "loss": 2.107,
      "step": 133446
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.1138068437576294,
      "learning_rate": 0.00022613314805138978,
      "loss": 2.5366,
      "step": 133447
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.0584876537322998,
      "learning_rate": 0.00022612918343122254,
      "loss": 2.0054,
      "step": 133448
    },
    {
      "epoch": 1.74,
      "grad_norm": 0.9701457023620605,
      "learning_rate": 0.00022612521882478922,
      "loss": 2.1947,
      "step": 133449
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.0067541599273682,
      "learning_rate": 0.00022612125423209057,
      "loss": 1.9329,
      "step": 133450
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.083756685256958,
      "learning_rate": 0.00022611728965312743,
      "loss": 2.0236,
      "step": 133451
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.1051985025405884,
      "learning_rate": 0.0002261133250879005,
      "loss": 2.1266,
      "step": 133452
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.0273041725158691,
      "learning_rate": 0.00022610936053641043,
      "loss": 1.9713,
      "step": 133453
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.0199652910232544,
      "learning_rate": 0.0002261053959986581,
      "loss": 2.0138,
      "step": 133454
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.067135214805603,
      "learning_rate": 0.00022610143147464414,
      "loss": 1.9476,
      "step": 133455
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.0133072137832642,
      "learning_rate": 0.0002260974669643693,
      "loss": 2.1458,
      "step": 133456
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.0700218677520752,
      "learning_rate": 0.00022609350246783434,
      "loss": 2.1333,
      "step": 133457
    },
    {
      "epoch": 1.74,
      "grad_norm": 0.9541317224502563,
      "learning_rate": 0.00022608953798504007,
      "loss": 2.1773,
      "step": 133458
    },
    {
      "epoch": 1.74,
      "grad_norm": 0.9424155950546265,
      "learning_rate": 0.00022608557351598706,
      "loss": 2.2266,
      "step": 133459
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.028807282447815,
      "learning_rate": 0.00022608160906067615,
      "loss": 2.1828,
      "step": 133460
    },
    {
      "epoch": 1.74,
      "grad_norm": 0.9581558704376221,
      "learning_rate": 0.00022607764461910808,
      "loss": 2.1295,
      "step": 133461
    },
    {
      "epoch": 1.74,
      "grad_norm": 0.9845268130302429,
      "learning_rate": 0.0002260736801912835,
      "loss": 2.0861,
      "step": 133462
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.0223610401153564,
      "learning_rate": 0.00022606971577720337,
      "loss": 2.0994,
      "step": 133463
    },
    {
      "epoch": 1.74,
      "grad_norm": 0.9421042799949646,
      "learning_rate": 0.00022606575137686823,
      "loss": 2.198,
      "step": 133464
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.024701714515686,
      "learning_rate": 0.0002260617869902788,
      "loss": 1.9628,
      "step": 133465
    },
    {
      "epoch": 1.74,
      "grad_norm": 0.965475857257843,
      "learning_rate": 0.0002260578226174359,
      "loss": 2.2323,
      "step": 133466
    },
    {
      "epoch": 1.74,
      "grad_norm": 0.9544395208358765,
      "learning_rate": 0.00022605385825834022,
      "loss": 1.9139,
      "step": 133467
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.067950963973999,
      "learning_rate": 0.00022604989391299251,
      "loss": 2.1297,
      "step": 133468
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.1141854524612427,
      "learning_rate": 0.00022604592958139357,
      "loss": 2.0069,
      "step": 133469
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.2287440299987793,
      "learning_rate": 0.0002260419652635442,
      "loss": 2.1791,
      "step": 133470
    },
    {
      "epoch": 1.74,
      "grad_norm": 0.9592121243476868,
      "learning_rate": 0.00022603800095944486,
      "loss": 2.0376,
      "step": 133471
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.0456936359405518,
      "learning_rate": 0.00022603403666909648,
      "loss": 2.3109,
      "step": 133472
    },
    {
      "epoch": 1.74,
      "grad_norm": 0.992611825466156,
      "learning_rate": 0.0002260300723924998,
      "loss": 2.2472,
      "step": 133473
    },
    {
      "epoch": 1.74,
      "grad_norm": 0.9699193239212036,
      "learning_rate": 0.00022602610812965548,
      "loss": 2.2179,
      "step": 133474
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.011763572692871,
      "learning_rate": 0.00022602214388056432,
      "loss": 2.0104,
      "step": 133475
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.0066725015640259,
      "learning_rate": 0.00022601817964522714,
      "loss": 2.1383,
      "step": 133476
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.0726361274719238,
      "learning_rate": 0.0002260142154236445,
      "loss": 2.1176,
      "step": 133477
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.5838112831115723,
      "learning_rate": 0.00022601025121581718,
      "loss": 2.0215,
      "step": 133478
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.001292109489441,
      "learning_rate": 0.00022600628702174595,
      "loss": 2.1467,
      "step": 133479
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.075914740562439,
      "learning_rate": 0.00022600232284143156,
      "loss": 2.2268,
      "step": 133480
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.2409380674362183,
      "learning_rate": 0.00022599835867487473,
      "loss": 2.1083,
      "step": 133481
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.0975630283355713,
      "learning_rate": 0.00022599439452207634,
      "loss": 2.0596,
      "step": 133482
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.0945298671722412,
      "learning_rate": 0.00022599043038303685,
      "loss": 2.1862,
      "step": 133483
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.0042262077331543,
      "learning_rate": 0.00022598646625775714,
      "loss": 2.1367,
      "step": 133484
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.0218693017959595,
      "learning_rate": 0.00022598250214623795,
      "loss": 2.114,
      "step": 133485
    },
    {
      "epoch": 1.74,
      "grad_norm": 0.9301309585571289,
      "learning_rate": 0.00022597853804848,
      "loss": 2.2732,
      "step": 133486
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.0061453580856323,
      "learning_rate": 0.00022597457396448403,
      "loss": 2.145,
      "step": 133487
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.116195559501648,
      "learning_rate": 0.00022597060989425094,
      "loss": 1.9441,
      "step": 133488
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.0545909404754639,
      "learning_rate": 0.00022596664583778117,
      "loss": 2.2184,
      "step": 133489
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.0243315696716309,
      "learning_rate": 0.00022596268179507557,
      "loss": 2.0543,
      "step": 133490
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.0025696754455566,
      "learning_rate": 0.00022595871776613492,
      "loss": 1.8129,
      "step": 133491
    },
    {
      "epoch": 1.74,
      "grad_norm": 0.9736483693122864,
      "learning_rate": 0.00022595475375095995,
      "loss": 2.1317,
      "step": 133492
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.1456220149993896,
      "learning_rate": 0.0002259507897495514,
      "loss": 2.1694,
      "step": 133493
    },
    {
      "epoch": 1.74,
      "grad_norm": 0.9915715456008911,
      "learning_rate": 0.0002259468257619101,
      "loss": 1.9771,
      "step": 133494
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.061929702758789,
      "learning_rate": 0.00022594286178803658,
      "loss": 2.1489,
      "step": 133495
    },
    {
      "epoch": 1.74,
      "grad_norm": 0.9872284531593323,
      "learning_rate": 0.00022593889782793167,
      "loss": 2.1217,
      "step": 133496
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.0212006568908691,
      "learning_rate": 0.00022593493388159611,
      "loss": 2.194,
      "step": 133497
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.0140299797058105,
      "learning_rate": 0.00022593096994903063,
      "loss": 2.1516,
      "step": 133498
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.0866819620132446,
      "learning_rate": 0.00022592700603023602,
      "loss": 1.9434,
      "step": 133499
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.0450929403305054,
      "learning_rate": 0.0002259230421252131,
      "loss": 1.9226,
      "step": 133500
    },
    {
      "epoch": 1.74,
      "grad_norm": 0.9812188744544983,
      "learning_rate": 0.00022591907823396232,
      "loss": 2.137,
      "step": 133501
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.0599992275238037,
      "learning_rate": 0.0002259151143564846,
      "loss": 2.1216,
      "step": 133502
    },
    {
      "epoch": 1.74,
      "grad_norm": 0.9310222268104553,
      "learning_rate": 0.00022591115049278064,
      "loss": 2.2956,
      "step": 133503
    },
    {
      "epoch": 1.74,
      "grad_norm": 0.9881861805915833,
      "learning_rate": 0.00022590718664285123,
      "loss": 2.0164,
      "step": 133504
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.0567854642868042,
      "learning_rate": 0.00022590322280669706,
      "loss": 2.1319,
      "step": 133505
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.02084219455719,
      "learning_rate": 0.00022589925898431904,
      "loss": 2.1444,
      "step": 133506
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.0566946268081665,
      "learning_rate": 0.00022589529517571757,
      "loss": 2.1806,
      "step": 133507
    },
    {
      "epoch": 1.74,
      "grad_norm": 0.9840008616447449,
      "learning_rate": 0.00022589133138089357,
      "loss": 1.9118,
      "step": 133508
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.0173860788345337,
      "learning_rate": 0.00022588736759984778,
      "loss": 2.122,
      "step": 133509
    },
    {
      "epoch": 1.74,
      "grad_norm": 0.9975063800811768,
      "learning_rate": 0.00022588340383258095,
      "loss": 2.1724,
      "step": 133510
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.0733166933059692,
      "learning_rate": 0.00022587944007909378,
      "loss": 2.0517,
      "step": 133511
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.0641188621520996,
      "learning_rate": 0.00022587547633938713,
      "loss": 2.027,
      "step": 133512
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.0336790084838867,
      "learning_rate": 0.0002258715126134615,
      "loss": 1.9762,
      "step": 133513
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.1458587646484375,
      "learning_rate": 0.00022586754890131777,
      "loss": 2.0253,
      "step": 133514
    },
    {
      "epoch": 1.74,
      "grad_norm": 0.9581544995307922,
      "learning_rate": 0.00022586358520295665,
      "loss": 1.9998,
      "step": 133515
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.107184648513794,
      "learning_rate": 0.0002258596215183789,
      "loss": 1.8093,
      "step": 133516
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.1286897659301758,
      "learning_rate": 0.00022585565784758526,
      "loss": 2.1454,
      "step": 133517
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.0496201515197754,
      "learning_rate": 0.00022585169419057642,
      "loss": 2.2116,
      "step": 133518
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.1334898471832275,
      "learning_rate": 0.0002258477305473533,
      "loss": 2.0781,
      "step": 133519
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.044984221458435,
      "learning_rate": 0.00022584376691791634,
      "loss": 2.125,
      "step": 133520
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.2194722890853882,
      "learning_rate": 0.00022583980330226645,
      "loss": 1.9894,
      "step": 133521
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.0145052671432495,
      "learning_rate": 0.00022583583970040434,
      "loss": 2.1734,
      "step": 133522
    },
    {
      "epoch": 1.74,
      "grad_norm": 0.882535994052887,
      "learning_rate": 0.0002258318761123307,
      "loss": 2.1008,
      "step": 133523
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.2151193618774414,
      "learning_rate": 0.00022582791253804635,
      "loss": 2.252,
      "step": 133524
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.1362460851669312,
      "learning_rate": 0.00022582394897755205,
      "loss": 2.1259,
      "step": 133525
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.1101715564727783,
      "learning_rate": 0.0002258199854308485,
      "loss": 2.1597,
      "step": 133526
    },
    {
      "epoch": 1.74,
      "grad_norm": 0.9478417038917542,
      "learning_rate": 0.00022581602189793634,
      "loss": 2.2648,
      "step": 133527
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.5131945610046387,
      "learning_rate": 0.00022581205837881636,
      "loss": 2.2045,
      "step": 133528
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.1005231142044067,
      "learning_rate": 0.00022580809487348933,
      "loss": 2.1657,
      "step": 133529
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.1196622848510742,
      "learning_rate": 0.00022580413138195596,
      "loss": 2.0064,
      "step": 133530
    },
    {
      "epoch": 1.74,
      "grad_norm": 0.9532880783081055,
      "learning_rate": 0.0002258001679042171,
      "loss": 2.2405,
      "step": 133531
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.054447054862976,
      "learning_rate": 0.0002257962044402733,
      "loss": 2.1121,
      "step": 133532
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.349513053894043,
      "learning_rate": 0.0002257922409901255,
      "loss": 2.0642,
      "step": 133533
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.0299044847488403,
      "learning_rate": 0.00022578827755377422,
      "loss": 1.9838,
      "step": 133534
    },
    {
      "epoch": 1.74,
      "grad_norm": 0.9275791645050049,
      "learning_rate": 0.00022578431413122028,
      "loss": 2.0336,
      "step": 133535
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.1197538375854492,
      "learning_rate": 0.00022578035072246447,
      "loss": 2.5426,
      "step": 133536
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.1668694019317627,
      "learning_rate": 0.00022577638732750754,
      "loss": 2.0244,
      "step": 133537
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.0151184797286987,
      "learning_rate": 0.0002257724239463501,
      "loss": 2.2314,
      "step": 133538
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.1201773881912231,
      "learning_rate": 0.000225768460578993,
      "loss": 2.2005,
      "step": 133539
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.0191336870193481,
      "learning_rate": 0.000225764497225437,
      "loss": 2.1248,
      "step": 133540
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.0898462533950806,
      "learning_rate": 0.00022576053388568269,
      "loss": 2.1505,
      "step": 133541
    },
    {
      "epoch": 1.74,
      "grad_norm": 0.991775393486023,
      "learning_rate": 0.00022575657055973094,
      "loss": 2.0625,
      "step": 133542
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.0754481554031372,
      "learning_rate": 0.0002257526072475825,
      "loss": 1.973,
      "step": 133543
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.067765712738037,
      "learning_rate": 0.000225748643949238,
      "loss": 2.006,
      "step": 133544
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.0523730516433716,
      "learning_rate": 0.0002257446806646982,
      "loss": 2.0436,
      "step": 133545
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.1635642051696777,
      "learning_rate": 0.00022574071739396386,
      "loss": 2.0527,
      "step": 133546
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.0019270181655884,
      "learning_rate": 0.00022573675413703579,
      "loss": 1.9805,
      "step": 133547
    },
    {
      "epoch": 1.74,
      "grad_norm": 0.9770616888999939,
      "learning_rate": 0.0002257327908939146,
      "loss": 2.1169,
      "step": 133548
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.1039831638336182,
      "learning_rate": 0.00022572882766460118,
      "loss": 2.201,
      "step": 133549
    },
    {
      "epoch": 1.74,
      "grad_norm": 0.9578575491905212,
      "learning_rate": 0.00022572486444909605,
      "loss": 1.9399,
      "step": 133550
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.0473536252975464,
      "learning_rate": 0.00022572090124740014,
      "loss": 1.8027,
      "step": 133551
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.3670364618301392,
      "learning_rate": 0.00022571693805951406,
      "loss": 2.0644,
      "step": 133552
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.0400159358978271,
      "learning_rate": 0.00022571297488543863,
      "loss": 2.2182,
      "step": 133553
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.1411819458007812,
      "learning_rate": 0.00022570901172517456,
      "loss": 2.2061,
      "step": 133554
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.1677736043930054,
      "learning_rate": 0.0002257050485787227,
      "loss": 2.0701,
      "step": 133555
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.1441292762756348,
      "learning_rate": 0.00022570108544608356,
      "loss": 2.28,
      "step": 133556
    },
    {
      "epoch": 1.74,
      "grad_norm": 0.9931785464286804,
      "learning_rate": 0.00022569712232725796,
      "loss": 2.092,
      "step": 133557
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.0700688362121582,
      "learning_rate": 0.00022569315922224672,
      "loss": 2.029,
      "step": 133558
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.402941107749939,
      "learning_rate": 0.0002256891961310505,
      "loss": 2.0543,
      "step": 133559
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.3451520204544067,
      "learning_rate": 0.00022568523305367002,
      "loss": 1.9181,
      "step": 133560
    },
    {
      "epoch": 1.74,
      "grad_norm": 0.990117073059082,
      "learning_rate": 0.00022568126999010626,
      "loss": 2.1069,
      "step": 133561
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.0044381618499756,
      "learning_rate": 0.00022567730694035958,
      "loss": 2.0445,
      "step": 133562
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.3643531799316406,
      "learning_rate": 0.0002256733439044309,
      "loss": 1.9377,
      "step": 133563
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.3332749605178833,
      "learning_rate": 0.00022566938088232097,
      "loss": 2.0226,
      "step": 133564
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.1761711835861206,
      "learning_rate": 0.0002256654178740305,
      "loss": 1.9361,
      "step": 133565
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.0006338357925415,
      "learning_rate": 0.0002256614548795602,
      "loss": 2.1714,
      "step": 133566
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.2486560344696045,
      "learning_rate": 0.000225657491898911,
      "loss": 2.0767,
      "step": 133567
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.588171124458313,
      "learning_rate": 0.00022565352893208333,
      "loss": 2.103,
      "step": 133568
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.0229127407073975,
      "learning_rate": 0.00022564956597907808,
      "loss": 2.0458,
      "step": 133569
    },
    {
      "epoch": 1.74,
      "grad_norm": 0.9964993000030518,
      "learning_rate": 0.00022564560303989597,
      "loss": 1.9924,
      "step": 133570
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.012457013130188,
      "learning_rate": 0.00022564164011453777,
      "loss": 1.9527,
      "step": 133571
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.0477814674377441,
      "learning_rate": 0.00022563767720300412,
      "loss": 2.2326,
      "step": 133572
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.245079755783081,
      "learning_rate": 0.00022563371430529605,
      "loss": 2.1448,
      "step": 133573
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.2201926708221436,
      "learning_rate": 0.0002256297514214139,
      "loss": 2.3517,
      "step": 133574
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.1187142133712769,
      "learning_rate": 0.0002256257885513586,
      "loss": 2.0132,
      "step": 133575
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.0125160217285156,
      "learning_rate": 0.00022562182569513085,
      "loss": 2.3142,
      "step": 133576
    },
    {
      "epoch": 1.74,
      "grad_norm": 0.9415320754051208,
      "learning_rate": 0.00022561786285273144,
      "loss": 2.1176,
      "step": 133577
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.1138362884521484,
      "learning_rate": 0.00022561390002416104,
      "loss": 2.0463,
      "step": 133578
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.0120466947555542,
      "learning_rate": 0.00022560993720942055,
      "loss": 2.0341,
      "step": 133579
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.0143036842346191,
      "learning_rate": 0.00022560597440851045,
      "loss": 2.0911,
      "step": 133580
    },
    {
      "epoch": 1.74,
      "grad_norm": 0.9766612648963928,
      "learning_rate": 0.0002256020116214316,
      "loss": 1.956,
      "step": 133581
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.0041027069091797,
      "learning_rate": 0.00022559804884818475,
      "loss": 2.1171,
      "step": 133582
    },
    {
      "epoch": 1.74,
      "grad_norm": 0.9167312979698181,
      "learning_rate": 0.00022559408608877066,
      "loss": 2.0879,
      "step": 133583
    },
    {
      "epoch": 1.74,
      "grad_norm": 0.9464182257652283,
      "learning_rate": 0.00022559012334318998,
      "loss": 2.2073,
      "step": 133584
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.0135657787322998,
      "learning_rate": 0.00022558616061144353,
      "loss": 2.1493,
      "step": 133585
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.3571035861968994,
      "learning_rate": 0.00022558219789353215,
      "loss": 2.2132,
      "step": 133586
    },
    {
      "epoch": 1.74,
      "grad_norm": 0.9717166423797607,
      "learning_rate": 0.0002255782351894563,
      "loss": 2.1564,
      "step": 133587
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.0511343479156494,
      "learning_rate": 0.00022557427249921685,
      "loss": 2.0763,
      "step": 133588
    },
    {
      "epoch": 1.74,
      "grad_norm": 0.9949374198913574,
      "learning_rate": 0.0002255703098228146,
      "loss": 2.4223,
      "step": 133589
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.0078785419464111,
      "learning_rate": 0.00022556634716025016,
      "loss": 1.996,
      "step": 133590
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.0072133541107178,
      "learning_rate": 0.0002255623845115244,
      "loss": 2.0468,
      "step": 133591
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.0537852048873901,
      "learning_rate": 0.00022555842187663814,
      "loss": 1.9938,
      "step": 133592
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.1023569107055664,
      "learning_rate": 0.00022555445925559177,
      "loss": 2.0218,
      "step": 133593
    },
    {
      "epoch": 1.74,
      "grad_norm": 0.9385802149772644,
      "learning_rate": 0.0002255504966483863,
      "loss": 2.1573,
      "step": 133594
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.1018726825714111,
      "learning_rate": 0.00022554653405502237,
      "loss": 2.0605,
      "step": 133595
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.0478720664978027,
      "learning_rate": 0.00022554257147550073,
      "loss": 1.9688,
      "step": 133596
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.0134625434875488,
      "learning_rate": 0.00022553860890982215,
      "loss": 2.0416,
      "step": 133597
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.318402647972107,
      "learning_rate": 0.00022553464635798748,
      "loss": 2.0548,
      "step": 133598
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.0787639617919922,
      "learning_rate": 0.00022553068381999723,
      "loss": 2.3479,
      "step": 133599
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.0327540636062622,
      "learning_rate": 0.00022552672129585213,
      "loss": 2.1153,
      "step": 133600
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.0660979747772217,
      "learning_rate": 0.00022552275878555313,
      "loss": 2.113,
      "step": 133601
    },
    {
      "epoch": 1.74,
      "grad_norm": 0.9711228013038635,
      "learning_rate": 0.00022551879628910076,
      "loss": 1.9038,
      "step": 133602
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.0642598867416382,
      "learning_rate": 0.00022551483380649594,
      "loss": 1.8982,
      "step": 133603
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.1184064149856567,
      "learning_rate": 0.00022551087133773937,
      "loss": 1.9899,
      "step": 133604
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.0060584545135498,
      "learning_rate": 0.00022550690888283167,
      "loss": 1.9745,
      "step": 133605
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.2052807807922363,
      "learning_rate": 0.0002255029464417736,
      "loss": 1.9381,
      "step": 133606
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.1717991828918457,
      "learning_rate": 0.00022549898401456593,
      "loss": 2.2131,
      "step": 133607
    },
    {
      "epoch": 1.74,
      "grad_norm": 0.9558503031730652,
      "learning_rate": 0.00022549502160120944,
      "loss": 1.9512,
      "step": 133608
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.1217325925827026,
      "learning_rate": 0.00022549105920170484,
      "loss": 1.8944,
      "step": 133609
    },
    {
      "epoch": 1.74,
      "grad_norm": 0.9475116729736328,
      "learning_rate": 0.00022548709681605294,
      "loss": 2.0767,
      "step": 133610
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.0004173517227173,
      "learning_rate": 0.00022548313444425432,
      "loss": 1.7473,
      "step": 133611
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.0631067752838135,
      "learning_rate": 0.00022547917208630975,
      "loss": 2.117,
      "step": 133612
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.0773890018463135,
      "learning_rate": 0.00022547520974222003,
      "loss": 2.0982,
      "step": 133613
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.1429624557495117,
      "learning_rate": 0.00022547124741198587,
      "loss": 2.1744,
      "step": 133614
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.1114060878753662,
      "learning_rate": 0.00022546728509560798,
      "loss": 2.1364,
      "step": 133615
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.0190527439117432,
      "learning_rate": 0.00022546332279308726,
      "loss": 2.2613,
      "step": 133616
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.0025274753570557,
      "learning_rate": 0.00022545936050442426,
      "loss": 2.1134,
      "step": 133617
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.0884644985198975,
      "learning_rate": 0.00022545539822961975,
      "loss": 2.2059,
      "step": 133618
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.2020177841186523,
      "learning_rate": 0.00022545143596867443,
      "loss": 1.9943,
      "step": 133619
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.18179190158844,
      "learning_rate": 0.00022544747372158912,
      "loss": 2.1839,
      "step": 133620
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.1212399005889893,
      "learning_rate": 0.00022544351148836458,
      "loss": 2.1249,
      "step": 133621
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.215190052986145,
      "learning_rate": 0.0002254395492690015,
      "loss": 2.1995,
      "step": 133622
    },
    {
      "epoch": 1.74,
      "grad_norm": 0.968973696231842,
      "learning_rate": 0.00022543558706350058,
      "loss": 2.0308,
      "step": 133623
    },
    {
      "epoch": 1.74,
      "grad_norm": 0.9920122623443604,
      "learning_rate": 0.0002254316248718626,
      "loss": 2.0663,
      "step": 133624
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.0751168727874756,
      "learning_rate": 0.00022542766269408832,
      "loss": 2.1015,
      "step": 133625
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.0741971731185913,
      "learning_rate": 0.00022542370053017842,
      "loss": 1.9412,
      "step": 133626
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.0332632064819336,
      "learning_rate": 0.00022541973838013362,
      "loss": 2.1245,
      "step": 133627
    },
    {
      "epoch": 1.74,
      "grad_norm": 0.9768071174621582,
      "learning_rate": 0.00022541577624395483,
      "loss": 2.1746,
      "step": 133628
    },
    {
      "epoch": 1.74,
      "grad_norm": 0.9345806241035461,
      "learning_rate": 0.0002254118141216425,
      "loss": 2.1478,
      "step": 133629
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.0309001207351685,
      "learning_rate": 0.00022540785201319758,
      "loss": 2.1819,
      "step": 133630
    },
    {
      "epoch": 1.74,
      "grad_norm": 0.9501841068267822,
      "learning_rate": 0.00022540388991862075,
      "loss": 2.2956,
      "step": 133631
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.0625348091125488,
      "learning_rate": 0.0002253999278379128,
      "loss": 2.048,
      "step": 133632
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.0270836353302002,
      "learning_rate": 0.00022539596577107434,
      "loss": 2.0636,
      "step": 133633
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.0797505378723145,
      "learning_rate": 0.00022539200371810624,
      "loss": 2.0577,
      "step": 133634
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.0075451135635376,
      "learning_rate": 0.0002253880416790091,
      "loss": 1.892,
      "step": 133635
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.0536502599716187,
      "learning_rate": 0.00022538407965378375,
      "loss": 2.1583,
      "step": 133636
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.1874877214431763,
      "learning_rate": 0.00022538011764243094,
      "loss": 2.0064,
      "step": 133637
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.1685866117477417,
      "learning_rate": 0.0002253761556449513,
      "loss": 2.1864,
      "step": 133638
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.0177737474441528,
      "learning_rate": 0.00022537219366134577,
      "loss": 2.0907,
      "step": 133639
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.050278902053833,
      "learning_rate": 0.00022536823169161496,
      "loss": 1.9855,
      "step": 133640
    },
    {
      "epoch": 1.74,
      "grad_norm": 0.9941638708114624,
      "learning_rate": 0.0002253642697357595,
      "loss": 2.1621,
      "step": 133641
    },
    {
      "epoch": 1.74,
      "grad_norm": 0.9736424684524536,
      "learning_rate": 0.00022536030779378023,
      "loss": 2.074,
      "step": 133642
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.0617103576660156,
      "learning_rate": 0.0002253563458656779,
      "loss": 2.1845,
      "step": 133643
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.0211907625198364,
      "learning_rate": 0.00022535238395145327,
      "loss": 2.0229,
      "step": 133644
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.4204344749450684,
      "learning_rate": 0.00022534842205110698,
      "loss": 1.8958,
      "step": 133645
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.0824439525604248,
      "learning_rate": 0.00022534446016464,
      "loss": 2.0231,
      "step": 133646
    },
    {
      "epoch": 1.74,
      "grad_norm": 0.932535707950592,
      "learning_rate": 0.00022534049829205273,
      "loss": 2.0504,
      "step": 133647
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.0358052253723145,
      "learning_rate": 0.0002253365364333461,
      "loss": 2.2161,
      "step": 133648
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.0417587757110596,
      "learning_rate": 0.0002253325745885208,
      "loss": 2.1047,
      "step": 133649
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.072227954864502,
      "learning_rate": 0.0002253286127575776,
      "loss": 2.3405,
      "step": 133650
    },
    {
      "epoch": 1.74,
      "grad_norm": 0.9473677277565002,
      "learning_rate": 0.0002253246509405172,
      "loss": 2.1416,
      "step": 133651
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.1379488706588745,
      "learning_rate": 0.00022532068913734036,
      "loss": 1.9509,
      "step": 133652
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.0130963325500488,
      "learning_rate": 0.00022531672734804798,
      "loss": 2.2391,
      "step": 133653
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.0473101139068604,
      "learning_rate": 0.00022531276557264047,
      "loss": 1.9002,
      "step": 133654
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.0203698873519897,
      "learning_rate": 0.00022530880381111873,
      "loss": 2.1244,
      "step": 133655
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.0594149827957153,
      "learning_rate": 0.00022530484206348349,
      "loss": 2.0819,
      "step": 133656
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.0728554725646973,
      "learning_rate": 0.0002253008803297355,
      "loss": 1.9161,
      "step": 133657
    },
    {
      "epoch": 1.74,
      "grad_norm": 0.9717970490455627,
      "learning_rate": 0.0002252969186098755,
      "loss": 1.9095,
      "step": 133658
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.0724456310272217,
      "learning_rate": 0.00022529295690390427,
      "loss": 2.1946,
      "step": 133659
    },
    {
      "epoch": 1.74,
      "grad_norm": 0.9706742763519287,
      "learning_rate": 0.00022528899521182243,
      "loss": 2.1886,
      "step": 133660
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.0146920680999756,
      "learning_rate": 0.00022528503353363074,
      "loss": 1.8618,
      "step": 133661
    },
    {
      "epoch": 1.74,
      "grad_norm": 0.9908749461174011,
      "learning_rate": 0.00022528107186933,
      "loss": 2.2083,
      "step": 133662
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.0498452186584473,
      "learning_rate": 0.00022527711021892092,
      "loss": 1.985,
      "step": 133663
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.1237331628799438,
      "learning_rate": 0.00022527314858240423,
      "loss": 2.0446,
      "step": 133664
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.1230303049087524,
      "learning_rate": 0.0002252691869597808,
      "loss": 2.0774,
      "step": 133665
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.1148936748504639,
      "learning_rate": 0.0002252652253510511,
      "loss": 2.2769,
      "step": 133666
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.0145238637924194,
      "learning_rate": 0.00022526126375621602,
      "loss": 1.921,
      "step": 133667
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.011250376701355,
      "learning_rate": 0.00022525730217527626,
      "loss": 2.1654,
      "step": 133668
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.1054998636245728,
      "learning_rate": 0.00022525334060823258,
      "loss": 2.1465,
      "step": 133669
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.050653100013733,
      "learning_rate": 0.00022524937905508573,
      "loss": 2.0739,
      "step": 133670
    },
    {
      "epoch": 1.74,
      "grad_norm": 0.9704014658927917,
      "learning_rate": 0.00022524541751583655,
      "loss": 2.1703,
      "step": 133671
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.054944396018982,
      "learning_rate": 0.00022524145599048554,
      "loss": 2.443,
      "step": 133672
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.2407631874084473,
      "learning_rate": 0.00022523749447903355,
      "loss": 2.2259,
      "step": 133673
    },
    {
      "epoch": 1.74,
      "grad_norm": 0.9123438000679016,
      "learning_rate": 0.00022523353298148133,
      "loss": 2.0442,
      "step": 133674
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.1652929782867432,
      "learning_rate": 0.0002252295714978296,
      "loss": 2.231,
      "step": 133675
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.0450068712234497,
      "learning_rate": 0.00022522561002807912,
      "loss": 1.9858,
      "step": 133676
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.0933116674423218,
      "learning_rate": 0.00022522164857223073,
      "loss": 1.8781,
      "step": 133677
    },
    {
      "epoch": 1.74,
      "grad_norm": 0.9297075867652893,
      "learning_rate": 0.00022521768713028488,
      "loss": 2.2084,
      "step": 133678
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.2047733068466187,
      "learning_rate": 0.0002252137257022425,
      "loss": 2.1497,
      "step": 133679
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.319553017616272,
      "learning_rate": 0.00022520976428810433,
      "loss": 2.0101,
      "step": 133680
    },
    {
      "epoch": 1.74,
      "grad_norm": 0.9814813137054443,
      "learning_rate": 0.00022520580288787104,
      "loss": 2.0217,
      "step": 133681
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.0143671035766602,
      "learning_rate": 0.0002252018415015434,
      "loss": 2.2153,
      "step": 133682
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.3180716037750244,
      "learning_rate": 0.0002251978801291223,
      "loss": 1.9893,
      "step": 133683
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.060762882232666,
      "learning_rate": 0.0002251939187706082,
      "loss": 2.051,
      "step": 133684
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.0593252182006836,
      "learning_rate": 0.00022518995742600197,
      "loss": 2.1404,
      "step": 133685
    },
    {
      "epoch": 1.74,
      "grad_norm": 0.9997661113739014,
      "learning_rate": 0.0002251859960953043,
      "loss": 2.1277,
      "step": 133686
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.005889892578125,
      "learning_rate": 0.00022518203477851596,
      "loss": 1.7713,
      "step": 133687
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.031218409538269,
      "learning_rate": 0.00022517807347563774,
      "loss": 2.1083,
      "step": 133688
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.088094711303711,
      "learning_rate": 0.00022517411218667043,
      "loss": 2.2738,
      "step": 133689
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.0479011535644531,
      "learning_rate": 0.00022517015091161457,
      "loss": 1.8919,
      "step": 133690
    },
    {
      "epoch": 1.74,
      "grad_norm": 0.9941692352294922,
      "learning_rate": 0.00022516618965047092,
      "loss": 2.1224,
      "step": 133691
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.1078591346740723,
      "learning_rate": 0.00022516222840324036,
      "loss": 2.066,
      "step": 133692
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.225253939628601,
      "learning_rate": 0.00022515826716992354,
      "loss": 1.9567,
      "step": 133693
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.058631181716919,
      "learning_rate": 0.00022515430595052116,
      "loss": 2.3411,
      "step": 133694
    },
    {
      "epoch": 1.74,
      "grad_norm": 0.9545338153839111,
      "learning_rate": 0.00022515034474503422,
      "loss": 2.0246,
      "step": 133695
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.052677869796753,
      "learning_rate": 0.00022514638355346307,
      "loss": 2.0609,
      "step": 133696
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.0221185684204102,
      "learning_rate": 0.0002251424223758086,
      "loss": 2.1633,
      "step": 133697
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.1576331853866577,
      "learning_rate": 0.0002251384612120716,
      "loss": 2.3727,
      "step": 133698
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.0190167427062988,
      "learning_rate": 0.00022513450006225274,
      "loss": 2.1417,
      "step": 133699
    },
    {
      "epoch": 1.74,
      "grad_norm": 0.9678346514701843,
      "learning_rate": 0.0002251305389263528,
      "loss": 1.9628,
      "step": 133700
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.0997333526611328,
      "learning_rate": 0.0002251265778043726,
      "loss": 2.1977,
      "step": 133701
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.1765387058258057,
      "learning_rate": 0.00022512261669631274,
      "loss": 2.1548,
      "step": 133702
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.0707744359970093,
      "learning_rate": 0.00022511865560217396,
      "loss": 2.1852,
      "step": 133703
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.0717872381210327,
      "learning_rate": 0.00022511469452195704,
      "loss": 2.1041,
      "step": 133704
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.0666451454162598,
      "learning_rate": 0.0002251107334556627,
      "loss": 2.3488,
      "step": 133705
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.0654512643814087,
      "learning_rate": 0.0002251067724032917,
      "loss": 2.0151,
      "step": 133706
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.246773600578308,
      "learning_rate": 0.00022510281136484483,
      "loss": 2.2208,
      "step": 133707
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.0090605020523071,
      "learning_rate": 0.00022509885034032267,
      "loss": 2.2994,
      "step": 133708
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.0642290115356445,
      "learning_rate": 0.00022509488932972614,
      "loss": 1.9767,
      "step": 133709
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.2664000988006592,
      "learning_rate": 0.0002250909283330558,
      "loss": 2.1664,
      "step": 133710
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.0797911882400513,
      "learning_rate": 0.00022508696735031248,
      "loss": 2.0665,
      "step": 133711
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.0056612491607666,
      "learning_rate": 0.00022508300638149692,
      "loss": 1.9669,
      "step": 133712
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.0411485433578491,
      "learning_rate": 0.00022507904542660987,
      "loss": 2.0046,
      "step": 133713
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.2592411041259766,
      "learning_rate": 0.00022507508448565202,
      "loss": 2.0462,
      "step": 133714
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.3655369281768799,
      "learning_rate": 0.0002250711235586241,
      "loss": 2.0532,
      "step": 133715
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.0549335479736328,
      "learning_rate": 0.00022506716264552687,
      "loss": 2.0941,
      "step": 133716
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.098656415939331,
      "learning_rate": 0.00022506320174636114,
      "loss": 2.1035,
      "step": 133717
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.2925747632980347,
      "learning_rate": 0.0002250592408611275,
      "loss": 1.938,
      "step": 133718
    },
    {
      "epoch": 1.74,
      "grad_norm": 0.8922456502914429,
      "learning_rate": 0.0002250552799898268,
      "loss": 2.173,
      "step": 133719
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.0472967624664307,
      "learning_rate": 0.00022505131913245978,
      "loss": 1.9705,
      "step": 133720
    },
    {
      "epoch": 1.74,
      "grad_norm": 0.9796456694602966,
      "learning_rate": 0.00022504735828902706,
      "loss": 2.1951,
      "step": 133721
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.0396089553833008,
      "learning_rate": 0.00022504339745952945,
      "loss": 2.1316,
      "step": 133722
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.1188396215438843,
      "learning_rate": 0.00022503943664396768,
      "loss": 2.0676,
      "step": 133723
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.148694396018982,
      "learning_rate": 0.0002250354758423426,
      "loss": 2.0406,
      "step": 133724
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.073595643043518,
      "learning_rate": 0.00022503151505465474,
      "loss": 2.3171,
      "step": 133725
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.154646873474121,
      "learning_rate": 0.00022502755428090499,
      "loss": 2.0756,
      "step": 133726
    },
    {
      "epoch": 1.74,
      "grad_norm": 0.9793274402618408,
      "learning_rate": 0.000225023593521094,
      "loss": 2.0058,
      "step": 133727
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.0486856698989868,
      "learning_rate": 0.00022501963277522251,
      "loss": 2.2489,
      "step": 133728
    },
    {
      "epoch": 1.74,
      "grad_norm": 0.9218420386314392,
      "learning_rate": 0.00022501567204329134,
      "loss": 2.2217,
      "step": 133729
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.0134611129760742,
      "learning_rate": 0.0002250117113253011,
      "loss": 2.0987,
      "step": 133730
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.0853790044784546,
      "learning_rate": 0.0002250077506212527,
      "loss": 2.06,
      "step": 133731
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.0422967672348022,
      "learning_rate": 0.0002250037899311468,
      "loss": 2.192,
      "step": 133732
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.1366311311721802,
      "learning_rate": 0.000224999829254984,
      "loss": 2.1363,
      "step": 133733
    },
    {
      "epoch": 1.74,
      "grad_norm": 0.943712055683136,
      "learning_rate": 0.0002249958685927652,
      "loss": 1.879,
      "step": 133734
    },
    {
      "epoch": 1.74,
      "grad_norm": 0.9875593781471252,
      "learning_rate": 0.00022499190794449104,
      "loss": 2.1981,
      "step": 133735
    },
    {
      "epoch": 1.74,
      "grad_norm": 0.915291965007782,
      "learning_rate": 0.00022498794731016229,
      "loss": 2.2944,
      "step": 133736
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.078148365020752,
      "learning_rate": 0.0002249839866897797,
      "loss": 2.0855,
      "step": 133737
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.037850022315979,
      "learning_rate": 0.00022498002608334417,
      "loss": 2.1,
      "step": 133738
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.0254604816436768,
      "learning_rate": 0.00022497606549085611,
      "loss": 2.0387,
      "step": 133739
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.143157958984375,
      "learning_rate": 0.00022497210491231642,
      "loss": 2.08,
      "step": 133740
    },
    {
      "epoch": 1.74,
      "grad_norm": 0.9986736178398132,
      "learning_rate": 0.0002249681443477259,
      "loss": 2.161,
      "step": 133741
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.154496431350708,
      "learning_rate": 0.0002249641837970851,
      "loss": 2.1511,
      "step": 133742
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.0413569211959839,
      "learning_rate": 0.00022496022326039497,
      "loss": 2.2185,
      "step": 133743
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.0299835205078125,
      "learning_rate": 0.00022495626273765618,
      "loss": 2.1054,
      "step": 133744
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.0002658367156982,
      "learning_rate": 0.0002249523022288694,
      "loss": 2.2995,
      "step": 133745
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.1887973546981812,
      "learning_rate": 0.00022494834173403537,
      "loss": 2.0297,
      "step": 133746
    },
    {
      "epoch": 1.74,
      "grad_norm": 0.9493160247802734,
      "learning_rate": 0.00022494438125315487,
      "loss": 2.005,
      "step": 133747
    },
    {
      "epoch": 1.74,
      "grad_norm": 0.9595147371292114,
      "learning_rate": 0.0002249404207862286,
      "loss": 2.121,
      "step": 133748
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.0983877182006836,
      "learning_rate": 0.00022493646033325734,
      "loss": 1.9908,
      "step": 133749
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.0531035661697388,
      "learning_rate": 0.00022493249989424195,
      "loss": 2.179,
      "step": 133750
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.0157846212387085,
      "learning_rate": 0.0002249285394691829,
      "loss": 2.0266,
      "step": 133751
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.0125659704208374,
      "learning_rate": 0.00022492457905808104,
      "loss": 2.1316,
      "step": 133752
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.058824896812439,
      "learning_rate": 0.00022492061866093708,
      "loss": 2.2268,
      "step": 133753
    },
    {
      "epoch": 1.74,
      "grad_norm": 0.9893103241920471,
      "learning_rate": 0.00022491665827775183,
      "loss": 2.1295,
      "step": 133754
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.1383391618728638,
      "learning_rate": 0.00022491269790852602,
      "loss": 2.1341,
      "step": 133755
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.1461944580078125,
      "learning_rate": 0.00022490873755326045,
      "loss": 2.2597,
      "step": 133756
    },
    {
      "epoch": 1.74,
      "grad_norm": 0.9058495759963989,
      "learning_rate": 0.0002249047772119556,
      "loss": 2.0259,
      "step": 133757
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.1892554759979248,
      "learning_rate": 0.0002249008168846124,
      "loss": 2.0922,
      "step": 133758
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.1278127431869507,
      "learning_rate": 0.00022489685657123157,
      "loss": 2.0095,
      "step": 133759
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.0699913501739502,
      "learning_rate": 0.00022489289627181385,
      "loss": 2.1715,
      "step": 133760
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.0206533670425415,
      "learning_rate": 0.00022488893598635994,
      "loss": 2.1218,
      "step": 133761
    },
    {
      "epoch": 1.74,
      "grad_norm": 0.9667497873306274,
      "learning_rate": 0.00022488497571487073,
      "loss": 1.9568,
      "step": 133762
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.0432584285736084,
      "learning_rate": 0.00022488101545734666,
      "loss": 2.2175,
      "step": 133763
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.0894654989242554,
      "learning_rate": 0.00022487705521378866,
      "loss": 2.0993,
      "step": 133764
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.0990757942199707,
      "learning_rate": 0.00022487309498419743,
      "loss": 2.267,
      "step": 133765
    },
    {
      "epoch": 1.74,
      "grad_norm": 0.9928556084632874,
      "learning_rate": 0.00022486913476857368,
      "loss": 2.3367,
      "step": 133766
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.0627789497375488,
      "learning_rate": 0.00022486517456691822,
      "loss": 2.0524,
      "step": 133767
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.1383984088897705,
      "learning_rate": 0.00022486121437923184,
      "loss": 2.1343,
      "step": 133768
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.0376031398773193,
      "learning_rate": 0.00022485725420551506,
      "loss": 1.9831,
      "step": 133769
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.055063009262085,
      "learning_rate": 0.00022485329404576873,
      "loss": 2.0741,
      "step": 133770
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.0056837797164917,
      "learning_rate": 0.00022484933389999357,
      "loss": 2.0849,
      "step": 133771
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.0975029468536377,
      "learning_rate": 0.00022484537376819035,
      "loss": 2.2484,
      "step": 133772
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.1952792406082153,
      "learning_rate": 0.00022484141365035983,
      "loss": 2.0517,
      "step": 133773
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.1518536806106567,
      "learning_rate": 0.0002248374535465028,
      "loss": 2.2308,
      "step": 133774
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.0161490440368652,
      "learning_rate": 0.00022483349345661982,
      "loss": 1.9976,
      "step": 133775
    },
    {
      "epoch": 1.74,
      "grad_norm": 0.9965674877166748,
      "learning_rate": 0.00022482953338071165,
      "loss": 1.9028,
      "step": 133776
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.08940851688385,
      "learning_rate": 0.00022482557331877915,
      "loss": 2.1187,
      "step": 133777
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.4374173879623413,
      "learning_rate": 0.00022482161327082295,
      "loss": 2.0219,
      "step": 133778
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.1290634870529175,
      "learning_rate": 0.00022481765323684386,
      "loss": 2.0456,
      "step": 133779
    },
    {
      "epoch": 1.74,
      "grad_norm": 0.9939621090888977,
      "learning_rate": 0.00022481369321684268,
      "loss": 2.1604,
      "step": 133780
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.0544636249542236,
      "learning_rate": 0.00022480973321082,
      "loss": 1.904,
      "step": 133781
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.0501114130020142,
      "learning_rate": 0.00022480577321877654,
      "loss": 2.2038,
      "step": 133782
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.0462342500686646,
      "learning_rate": 0.0002248018132407131,
      "loss": 2.0315,
      "step": 133783
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.0732232332229614,
      "learning_rate": 0.00022479785327663047,
      "loss": 1.891,
      "step": 133784
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.0912683010101318,
      "learning_rate": 0.0002247938933265293,
      "loss": 2.1763,
      "step": 133785
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.1091029644012451,
      "learning_rate": 0.0002247899333904104,
      "loss": 2.0807,
      "step": 133786
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.412341833114624,
      "learning_rate": 0.00022478597346827457,
      "loss": 2.0255,
      "step": 133787
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.0534874200820923,
      "learning_rate": 0.00022478201356012235,
      "loss": 2.006,
      "step": 133788
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.1792563199996948,
      "learning_rate": 0.00022477805366595454,
      "loss": 1.9885,
      "step": 133789
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.1600087881088257,
      "learning_rate": 0.00022477409378577191,
      "loss": 2.059,
      "step": 133790
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.1377229690551758,
      "learning_rate": 0.00022477013391957522,
      "loss": 2.2053,
      "step": 133791
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.0003185272216797,
      "learning_rate": 0.00022476617406736517,
      "loss": 2.2908,
      "step": 133792
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.0708832740783691,
      "learning_rate": 0.00022476221422914258,
      "loss": 2.0129,
      "step": 133793
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.1186028718948364,
      "learning_rate": 0.0002247582544049081,
      "loss": 2.0946,
      "step": 133794
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.047423243522644,
      "learning_rate": 0.00022475429459466244,
      "loss": 1.889,
      "step": 133795
    },
    {
      "epoch": 1.74,
      "grad_norm": 0.9225717782974243,
      "learning_rate": 0.00022475033479840635,
      "loss": 2.1937,
      "step": 133796
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.2057409286499023,
      "learning_rate": 0.0002247463750161406,
      "loss": 2.1695,
      "step": 133797
    },
    {
      "epoch": 1.74,
      "grad_norm": 0.9919884204864502,
      "learning_rate": 0.0002247424152478659,
      "loss": 2.1074,
      "step": 133798
    },
    {
      "epoch": 1.74,
      "grad_norm": 0.9898582696914673,
      "learning_rate": 0.0002247384554935831,
      "loss": 1.9619,
      "step": 133799
    },
    {
      "epoch": 1.74,
      "grad_norm": 0.9927976727485657,
      "learning_rate": 0.00022473449575329278,
      "loss": 2.1,
      "step": 133800
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.0325217247009277,
      "learning_rate": 0.0002247305360269958,
      "loss": 2.0364,
      "step": 133801
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.3747731447219849,
      "learning_rate": 0.0002247265763146928,
      "loss": 2.1869,
      "step": 133802
    },
    {
      "epoch": 1.74,
      "grad_norm": 0.9955827593803406,
      "learning_rate": 0.0002247226166163845,
      "loss": 2.2382,
      "step": 133803
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.2811866998672485,
      "learning_rate": 0.00022471865693207167,
      "loss": 2.0709,
      "step": 133804
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.05108642578125,
      "learning_rate": 0.00022471469726175516,
      "loss": 2.2941,
      "step": 133805
    },
    {
      "epoch": 1.74,
      "grad_norm": 0.9450722336769104,
      "learning_rate": 0.00022471073760543553,
      "loss": 1.989,
      "step": 133806
    },
    {
      "epoch": 1.74,
      "grad_norm": 0.9520121216773987,
      "learning_rate": 0.00022470677796311366,
      "loss": 2.0737,
      "step": 133807
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.1171423196792603,
      "learning_rate": 0.00022470281833479017,
      "loss": 2.0421,
      "step": 133808
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.0973007678985596,
      "learning_rate": 0.0002246988587204659,
      "loss": 2.1364,
      "step": 133809
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.0638726949691772,
      "learning_rate": 0.00022469489912014149,
      "loss": 2.1592,
      "step": 133810
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.0506014823913574,
      "learning_rate": 0.00022469093953381776,
      "loss": 2.1382,
      "step": 133811
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.1028448343276978,
      "learning_rate": 0.00022468697996149534,
      "loss": 2.2786,
      "step": 133812
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.159250020980835,
      "learning_rate": 0.00022468302040317504,
      "loss": 2.1498,
      "step": 133813
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.2122368812561035,
      "learning_rate": 0.00022467906085885762,
      "loss": 2.0339,
      "step": 133814
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.1283262968063354,
      "learning_rate": 0.00022467510132854373,
      "loss": 1.9899,
      "step": 133815
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.0695708990097046,
      "learning_rate": 0.00022467114181223428,
      "loss": 2.2318,
      "step": 133816
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.1376621723175049,
      "learning_rate": 0.00022466718230992988,
      "loss": 2.0046,
      "step": 133817
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.0763050317764282,
      "learning_rate": 0.00022466322282163118,
      "loss": 2.2317,
      "step": 133818
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.0402522087097168,
      "learning_rate": 0.00022465926334733902,
      "loss": 2.2174,
      "step": 133819
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.0679495334625244,
      "learning_rate": 0.0002246553038870541,
      "loss": 2.2232,
      "step": 133820
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.05946946144104,
      "learning_rate": 0.00022465134444077723,
      "loss": 2.0404,
      "step": 133821
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.0201574563980103,
      "learning_rate": 0.00022464738500850906,
      "loss": 2.2044,
      "step": 133822
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.0667192935943604,
      "learning_rate": 0.00022464342559025052,
      "loss": 2.3181,
      "step": 133823
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.002356767654419,
      "learning_rate": 0.00022463946618600205,
      "loss": 1.8664,
      "step": 133824
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.0865306854248047,
      "learning_rate": 0.0002246355067957645,
      "loss": 1.994,
      "step": 133825
    },
    {
      "epoch": 1.74,
      "grad_norm": 0.990210771560669,
      "learning_rate": 0.00022463154741953867,
      "loss": 2.1538,
      "step": 133826
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.0777775049209595,
      "learning_rate": 0.00022462758805732525,
      "loss": 1.9673,
      "step": 133827
    },
    {
      "epoch": 1.74,
      "grad_norm": 0.9516704082489014,
      "learning_rate": 0.00022462362870912496,
      "loss": 2.0235,
      "step": 133828
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.0016883611679077,
      "learning_rate": 0.00022461966937493868,
      "loss": 2.102,
      "step": 133829
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.0352731943130493,
      "learning_rate": 0.00022461571005476697,
      "loss": 1.8592,
      "step": 133830
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.0446808338165283,
      "learning_rate": 0.00022461175074861057,
      "loss": 2.0424,
      "step": 133831
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.0834981203079224,
      "learning_rate": 0.00022460779145647025,
      "loss": 1.903,
      "step": 133832
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.0088129043579102,
      "learning_rate": 0.0002246038321783468,
      "loss": 2.1901,
      "step": 133833
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.106621503829956,
      "learning_rate": 0.0002245998729142409,
      "loss": 2.1181,
      "step": 133834
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.0589991807937622,
      "learning_rate": 0.00022459591366415346,
      "loss": 2.1815,
      "step": 133835
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.1239418983459473,
      "learning_rate": 0.00022459195442808494,
      "loss": 2.208,
      "step": 133836
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.0901014804840088,
      "learning_rate": 0.00022458799520603615,
      "loss": 2.0154,
      "step": 133837
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.006746768951416,
      "learning_rate": 0.00022458403599800788,
      "loss": 2.2859,
      "step": 133838
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.105133295059204,
      "learning_rate": 0.00022458007680400092,
      "loss": 2.1612,
      "step": 133839
    },
    {
      "epoch": 1.74,
      "grad_norm": 0.9645664095878601,
      "learning_rate": 0.00022457611762401593,
      "loss": 2.0647,
      "step": 133840
    },
    {
      "epoch": 1.74,
      "grad_norm": 0.9609995484352112,
      "learning_rate": 0.00022457215845805376,
      "loss": 2.1677,
      "step": 133841
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.0280112028121948,
      "learning_rate": 0.00022456819930611493,
      "loss": 1.892,
      "step": 133842
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.063706874847412,
      "learning_rate": 0.0002245642401682003,
      "loss": 2.1326,
      "step": 133843
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.5943455696105957,
      "learning_rate": 0.00022456028104431058,
      "loss": 2.4113,
      "step": 133844
    },
    {
      "epoch": 1.74,
      "grad_norm": 0.9874350428581238,
      "learning_rate": 0.00022455632193444657,
      "loss": 1.8875,
      "step": 133845
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.0203262567520142,
      "learning_rate": 0.00022455236283860894,
      "loss": 2.0078,
      "step": 133846
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.1868757009506226,
      "learning_rate": 0.0002245484037567986,
      "loss": 2.1065,
      "step": 133847
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.0728743076324463,
      "learning_rate": 0.00022454444468901597,
      "loss": 2.0733,
      "step": 133848
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.1630892753601074,
      "learning_rate": 0.00022454048563526197,
      "loss": 2.2252,
      "step": 133849
    },
    {
      "epoch": 1.74,
      "grad_norm": 0.9154196381568909,
      "learning_rate": 0.0002245365265955373,
      "loss": 2.1954,
      "step": 133850
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.527360439300537,
      "learning_rate": 0.00022453256756984277,
      "loss": 2.1479,
      "step": 133851
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.1550047397613525,
      "learning_rate": 0.000224528608558179,
      "loss": 2.1022,
      "step": 133852
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.0913670063018799,
      "learning_rate": 0.0002245246495605468,
      "loss": 2.1822,
      "step": 133853
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.2164226770401,
      "learning_rate": 0.00022452069057694705,
      "loss": 2.2111,
      "step": 133854
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.12105131149292,
      "learning_rate": 0.00022451673160738022,
      "loss": 2.1325,
      "step": 133855
    },
    {
      "epoch": 1.74,
      "grad_norm": 0.9799404740333557,
      "learning_rate": 0.00022451277265184708,
      "loss": 1.9698,
      "step": 133856
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.0981979370117188,
      "learning_rate": 0.00022450881371034848,
      "loss": 2.0495,
      "step": 133857
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.0299783945083618,
      "learning_rate": 0.00022450485478288514,
      "loss": 2.2874,
      "step": 133858
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.1568257808685303,
      "learning_rate": 0.00022450089586945773,
      "loss": 2.1514,
      "step": 133859
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.3644167184829712,
      "learning_rate": 0.00022449693697006718,
      "loss": 2.1038,
      "step": 133860
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.0863760709762573,
      "learning_rate": 0.00022449297808471393,
      "loss": 2.0988,
      "step": 133861
    },
    {
      "epoch": 1.74,
      "grad_norm": 0.9498996734619141,
      "learning_rate": 0.00022448901921339886,
      "loss": 2.2162,
      "step": 133862
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.0545815229415894,
      "learning_rate": 0.0002244850603561227,
      "loss": 2.2658,
      "step": 133863
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.1183969974517822,
      "learning_rate": 0.00022448110151288624,
      "loss": 2.0882,
      "step": 133864
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.1879137754440308,
      "learning_rate": 0.00022447714268369014,
      "loss": 2.0027,
      "step": 133865
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.031631588935852,
      "learning_rate": 0.0002244731838685353,
      "loss": 2.0333,
      "step": 133866
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.2510195970535278,
      "learning_rate": 0.00022446922506742217,
      "loss": 2.0725,
      "step": 133867
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.1301774978637695,
      "learning_rate": 0.00022446526628035164,
      "loss": 2.1016,
      "step": 133868
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.2288445234298706,
      "learning_rate": 0.00022446130750732445,
      "loss": 2.0497,
      "step": 133869
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.0023137331008911,
      "learning_rate": 0.00022445734874834136,
      "loss": 2.3251,
      "step": 133870
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.0684397220611572,
      "learning_rate": 0.00022445339000340303,
      "loss": 2.1873,
      "step": 133871
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.1816014051437378,
      "learning_rate": 0.0002244494312725104,
      "loss": 2.0616,
      "step": 133872
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.099445104598999,
      "learning_rate": 0.00022444547255566388,
      "loss": 2.0909,
      "step": 133873
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.1078554391860962,
      "learning_rate": 0.0002244415138528644,
      "loss": 2.2322,
      "step": 133874
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.0467307567596436,
      "learning_rate": 0.00022443755516411267,
      "loss": 2.1888,
      "step": 133875
    },
    {
      "epoch": 1.74,
      "grad_norm": 0.9791780114173889,
      "learning_rate": 0.00022443359648940942,
      "loss": 2.2857,
      "step": 133876
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.0318831205368042,
      "learning_rate": 0.0002244296378287554,
      "loss": 2.1363,
      "step": 133877
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.1486057043075562,
      "learning_rate": 0.0002244256791821514,
      "loss": 2.1762,
      "step": 133878
    },
    {
      "epoch": 1.74,
      "grad_norm": 0.9599966406822205,
      "learning_rate": 0.0002244217205495981,
      "loss": 2.2698,
      "step": 133879
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.0041214227676392,
      "learning_rate": 0.0002244177619310961,
      "loss": 2.0528,
      "step": 133880
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.0331920385360718,
      "learning_rate": 0.00022441380332664633,
      "loss": 2.2822,
      "step": 133881
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.472007393836975,
      "learning_rate": 0.00022440984473624947,
      "loss": 1.909,
      "step": 133882
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.336100697517395,
      "learning_rate": 0.0002244058861599062,
      "loss": 2.2761,
      "step": 133883
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.1931300163269043,
      "learning_rate": 0.0002244019275976174,
      "loss": 2.082,
      "step": 133884
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.1591604948043823,
      "learning_rate": 0.00022439796904938364,
      "loss": 2.2258,
      "step": 133885
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.1476939916610718,
      "learning_rate": 0.00022439401051520578,
      "loss": 2.1863,
      "step": 133886
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.270946979522705,
      "learning_rate": 0.0002243900519950844,
      "loss": 1.9651,
      "step": 133887
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.0120799541473389,
      "learning_rate": 0.0002243860934890204,
      "loss": 2.0999,
      "step": 133888
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.0862910747528076,
      "learning_rate": 0.00022438213499701443,
      "loss": 2.2222,
      "step": 133889
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.055057168006897,
      "learning_rate": 0.00022437817651906733,
      "loss": 1.8963,
      "step": 133890
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.1192561388015747,
      "learning_rate": 0.00022437421805517966,
      "loss": 2.2336,
      "step": 133891
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.0720007419586182,
      "learning_rate": 0.00022437025960535227,
      "loss": 1.9958,
      "step": 133892
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.1601388454437256,
      "learning_rate": 0.00022436630116958597,
      "loss": 2.0504,
      "step": 133893
    },
    {
      "epoch": 1.74,
      "grad_norm": 0.9967312812805176,
      "learning_rate": 0.00022436234274788127,
      "loss": 2.2146,
      "step": 133894
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.099578857421875,
      "learning_rate": 0.00022435838434023907,
      "loss": 2.0048,
      "step": 133895
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.0912460088729858,
      "learning_rate": 0.00022435442594666015,
      "loss": 2.1256,
      "step": 133896
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.0383235216140747,
      "learning_rate": 0.00022435046756714505,
      "loss": 1.9683,
      "step": 133897
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.0042771100997925,
      "learning_rate": 0.0002243465092016947,
      "loss": 1.9598,
      "step": 133898
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.0428482294082642,
      "learning_rate": 0.00022434255085030974,
      "loss": 2.1153,
      "step": 133899
    },
    {
      "epoch": 1.74,
      "grad_norm": 0.9670274257659912,
      "learning_rate": 0.000224338592512991,
      "loss": 2.1157,
      "step": 133900
    },
    {
      "epoch": 1.74,
      "grad_norm": 0.9328801035881042,
      "learning_rate": 0.00022433463418973905,
      "loss": 2.1034,
      "step": 133901
    },
    {
      "epoch": 1.74,
      "grad_norm": 0.9890404939651489,
      "learning_rate": 0.00022433067588055477,
      "loss": 1.9947,
      "step": 133902
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.060705304145813,
      "learning_rate": 0.00022432671758543883,
      "loss": 2.013,
      "step": 133903
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.1005710363388062,
      "learning_rate": 0.00022432275930439196,
      "loss": 2.1948,
      "step": 133904
    },
    {
      "epoch": 1.74,
      "grad_norm": 0.9811242818832397,
      "learning_rate": 0.0002243188010374149,
      "loss": 1.9166,
      "step": 133905
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.0635476112365723,
      "learning_rate": 0.00022431484278450842,
      "loss": 2.157,
      "step": 133906
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.1134566068649292,
      "learning_rate": 0.00022431088454567322,
      "loss": 2.1296,
      "step": 133907
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.1150223016738892,
      "learning_rate": 0.00022430692632091024,
      "loss": 2.1317,
      "step": 133908
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.0414913892745972,
      "learning_rate": 0.00022430296811021987,
      "loss": 2.0585,
      "step": 133909
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.0566872358322144,
      "learning_rate": 0.000224299009913603,
      "loss": 2.0266,
      "step": 133910
    },
    {
      "epoch": 1.74,
      "grad_norm": 0.9674780964851379,
      "learning_rate": 0.0002242950517310604,
      "loss": 2.0572,
      "step": 133911
    },
    {
      "epoch": 1.74,
      "grad_norm": 0.9118971824645996,
      "learning_rate": 0.00022429109356259273,
      "loss": 2.1121,
      "step": 133912
    },
    {
      "epoch": 1.74,
      "grad_norm": 0.976047694683075,
      "learning_rate": 0.00022428713540820084,
      "loss": 1.9726,
      "step": 133913
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.0884838104248047,
      "learning_rate": 0.00022428317726788544,
      "loss": 2.2218,
      "step": 133914
    },
    {
      "epoch": 1.74,
      "grad_norm": 0.9757567048072815,
      "learning_rate": 0.00022427921914164716,
      "loss": 1.8782,
      "step": 133915
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.0454018115997314,
      "learning_rate": 0.00022427526102948678,
      "loss": 1.8774,
      "step": 133916
    },
    {
      "epoch": 1.74,
      "grad_norm": 0.9997714161872864,
      "learning_rate": 0.0002242713029314051,
      "loss": 2.1042,
      "step": 133917
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.1415565013885498,
      "learning_rate": 0.00022426734484740278,
      "loss": 2.1462,
      "step": 133918
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.0448921918869019,
      "learning_rate": 0.00022426338677748059,
      "loss": 2.1385,
      "step": 133919
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.071654200553894,
      "learning_rate": 0.00022425942872163925,
      "loss": 2.0091,
      "step": 133920
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.023478388786316,
      "learning_rate": 0.00022425547067987967,
      "loss": 1.9823,
      "step": 133921
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.0557265281677246,
      "learning_rate": 0.00022425151265220228,
      "loss": 2.1036,
      "step": 133922
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.0353503227233887,
      "learning_rate": 0.00022424755463860797,
      "loss": 2.136,
      "step": 133923
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.016945719718933,
      "learning_rate": 0.00022424359663909748,
      "loss": 2.0566,
      "step": 133924
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.1391271352767944,
      "learning_rate": 0.00022423963865367152,
      "loss": 2.0224,
      "step": 133925
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.0419856309890747,
      "learning_rate": 0.00022423568068233088,
      "loss": 2.1494,
      "step": 133926
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.014404058456421,
      "learning_rate": 0.00022423172272507632,
      "loss": 2.0096,
      "step": 133927
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.1152814626693726,
      "learning_rate": 0.00022422776478190845,
      "loss": 2.3328,
      "step": 133928
    },
    {
      "epoch": 1.74,
      "grad_norm": 0.9603895545005798,
      "learning_rate": 0.00022422380685282802,
      "loss": 2.0154,
      "step": 133929
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.1377655267715454,
      "learning_rate": 0.00022421984893783583,
      "loss": 2.2607,
      "step": 133930
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.166640281677246,
      "learning_rate": 0.0002242158910369326,
      "loss": 2.0625,
      "step": 133931
    },
    {
      "epoch": 1.74,
      "grad_norm": 0.9918932914733887,
      "learning_rate": 0.0002242119331501191,
      "loss": 1.9561,
      "step": 133932
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.0511873960494995,
      "learning_rate": 0.00022420797527739612,
      "loss": 2.0814,
      "step": 133933
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.0130258798599243,
      "learning_rate": 0.0002242040174187642,
      "loss": 2.085,
      "step": 133934
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.0315366983413696,
      "learning_rate": 0.00022420005957422416,
      "loss": 2.1353,
      "step": 133935
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.0861726999282837,
      "learning_rate": 0.00022419610174377678,
      "loss": 2.1051,
      "step": 133936
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.2164943218231201,
      "learning_rate": 0.00022419214392742275,
      "loss": 2.1908,
      "step": 133937
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.102868914604187,
      "learning_rate": 0.00022418818612516286,
      "loss": 2.1407,
      "step": 133938
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.1050819158554077,
      "learning_rate": 0.00022418422833699793,
      "loss": 2.0906,
      "step": 133939
    },
    {
      "epoch": 1.74,
      "grad_norm": 0.908025860786438,
      "learning_rate": 0.00022418027056292846,
      "loss": 2.1842,
      "step": 133940
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.037313461303711,
      "learning_rate": 0.0002241763128029553,
      "loss": 2.1791,
      "step": 133941
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.1227716207504272,
      "learning_rate": 0.00022417235505707922,
      "loss": 2.103,
      "step": 133942
    },
    {
      "epoch": 1.74,
      "grad_norm": 0.9951955080032349,
      "learning_rate": 0.00022416839732530093,
      "loss": 2.1421,
      "step": 133943
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.093035101890564,
      "learning_rate": 0.00022416443960762114,
      "loss": 1.9391,
      "step": 133944
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.0566014051437378,
      "learning_rate": 0.00022416048190404076,
      "loss": 2.218,
      "step": 133945
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.3166216611862183,
      "learning_rate": 0.00022415652421456023,
      "loss": 2.3626,
      "step": 133946
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.1024737358093262,
      "learning_rate": 0.00022415256653918046,
      "loss": 2.2488,
      "step": 133947
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.0615692138671875,
      "learning_rate": 0.00022414860887790217,
      "loss": 2.3509,
      "step": 133948
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.151574730873108,
      "learning_rate": 0.00022414465123072604,
      "loss": 1.8745,
      "step": 133949
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.2228293418884277,
      "learning_rate": 0.00022414069359765284,
      "loss": 2.0916,
      "step": 133950
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.0896258354187012,
      "learning_rate": 0.0002241367359786835,
      "loss": 1.9405,
      "step": 133951
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.1136579513549805,
      "learning_rate": 0.00022413277837381843,
      "loss": 2.1796,
      "step": 133952
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.4318574666976929,
      "learning_rate": 0.00022412882078305852,
      "loss": 2.1321,
      "step": 133953
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.022548794746399,
      "learning_rate": 0.00022412486320640447,
      "loss": 2.1837,
      "step": 133954
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.1156433820724487,
      "learning_rate": 0.00022412090564385704,
      "loss": 1.9681,
      "step": 133955
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.1035946607589722,
      "learning_rate": 0.00022411694809541696,
      "loss": 1.8866,
      "step": 133956
    },
    {
      "epoch": 1.74,
      "grad_norm": 0.997646689414978,
      "learning_rate": 0.00022411299056108511,
      "loss": 2.158,
      "step": 133957
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.1246088743209839,
      "learning_rate": 0.000224109033040862,
      "loss": 2.284,
      "step": 133958
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.1499862670898438,
      "learning_rate": 0.00022410507553474837,
      "loss": 2.0162,
      "step": 133959
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.4955058097839355,
      "learning_rate": 0.00022410111804274505,
      "loss": 2.2672,
      "step": 133960
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.454347848892212,
      "learning_rate": 0.00022409716056485284,
      "loss": 2.2172,
      "step": 133961
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.1207469701766968,
      "learning_rate": 0.00022409320310107236,
      "loss": 2.2844,
      "step": 133962
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.7308299541473389,
      "learning_rate": 0.00022408924565140448,
      "loss": 2.1912,
      "step": 133963
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.0226901769638062,
      "learning_rate": 0.00022408528821584977,
      "loss": 2.0221,
      "step": 133964
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.0367355346679688,
      "learning_rate": 0.000224081330794409,
      "loss": 2.1616,
      "step": 133965
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.0263454914093018,
      "learning_rate": 0.000224077373387083,
      "loss": 2.1587,
      "step": 133966
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.3047113418579102,
      "learning_rate": 0.00022407341599387239,
      "loss": 2.0205,
      "step": 133967
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.4804733991622925,
      "learning_rate": 0.00022406945861477796,
      "loss": 2.1228,
      "step": 133968
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.124207854270935,
      "learning_rate": 0.00022406550124980055,
      "loss": 1.9962,
      "step": 133969
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.2559618949890137,
      "learning_rate": 0.0002240615438989407,
      "loss": 2.0378,
      "step": 133970
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.0627176761627197,
      "learning_rate": 0.00022405758656219934,
      "loss": 2.1714,
      "step": 133971
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.3307843208312988,
      "learning_rate": 0.00022405362923957705,
      "loss": 2.0165,
      "step": 133972
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.196034550666809,
      "learning_rate": 0.0002240496719310746,
      "loss": 1.9315,
      "step": 133973
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.0362725257873535,
      "learning_rate": 0.00022404571463669273,
      "loss": 1.9877,
      "step": 133974
    },
    {
      "epoch": 1.74,
      "grad_norm": 0.9774990081787109,
      "learning_rate": 0.00022404175735643233,
      "loss": 1.8711,
      "step": 133975
    },
    {
      "epoch": 1.74,
      "grad_norm": 0.9888665080070496,
      "learning_rate": 0.0002240378000902939,
      "loss": 2.0995,
      "step": 133976
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.0787217617034912,
      "learning_rate": 0.00022403384283827826,
      "loss": 2.2499,
      "step": 133977
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.2388718128204346,
      "learning_rate": 0.00022402988560038623,
      "loss": 2.2212,
      "step": 133978
    },
    {
      "epoch": 1.74,
      "grad_norm": 0.961834192276001,
      "learning_rate": 0.0002240259283766184,
      "loss": 2.0393,
      "step": 133979
    },
    {
      "epoch": 1.74,
      "grad_norm": 0.9932929873466492,
      "learning_rate": 0.0002240219711669756,
      "loss": 2.0593,
      "step": 133980
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.0293986797332764,
      "learning_rate": 0.00022401801397145863,
      "loss": 2.2332,
      "step": 133981
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.1313824653625488,
      "learning_rate": 0.00022401405679006806,
      "loss": 2.0353,
      "step": 133982
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.1582660675048828,
      "learning_rate": 0.00022401009962280473,
      "loss": 2.271,
      "step": 133983
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.01276433467865,
      "learning_rate": 0.00022400614246966932,
      "loss": 2.0093,
      "step": 133984
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.0121448040008545,
      "learning_rate": 0.0002240021853306627,
      "loss": 2.1225,
      "step": 133985
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.2447540760040283,
      "learning_rate": 0.00022399822820578545,
      "loss": 1.9787,
      "step": 133986
    },
    {
      "epoch": 1.74,
      "grad_norm": 0.952154815196991,
      "learning_rate": 0.00022399427109503833,
      "loss": 2.0176,
      "step": 133987
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.0287749767303467,
      "learning_rate": 0.0002239903139984222,
      "loss": 1.8447,
      "step": 133988
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.0040407180786133,
      "learning_rate": 0.00022398635691593766,
      "loss": 2.3675,
      "step": 133989
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.0728938579559326,
      "learning_rate": 0.00022398239984758547,
      "loss": 2.1884,
      "step": 133990
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.22509765625,
      "learning_rate": 0.00022397844279336637,
      "loss": 1.941,
      "step": 133991
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.1003565788269043,
      "learning_rate": 0.00022397448575328118,
      "loss": 2.039,
      "step": 133992
    },
    {
      "epoch": 1.74,
      "grad_norm": 0.9170858263969421,
      "learning_rate": 0.00022397052872733054,
      "loss": 2.0699,
      "step": 133993
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.2033053636550903,
      "learning_rate": 0.00022396657171551527,
      "loss": 2.2387,
      "step": 133994
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.4355792999267578,
      "learning_rate": 0.00022396261471783595,
      "loss": 2.0377,
      "step": 133995
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.101141095161438,
      "learning_rate": 0.00022395865773429345,
      "loss": 2.1032,
      "step": 133996
    },
    {
      "epoch": 1.74,
      "grad_norm": 0.9878052473068237,
      "learning_rate": 0.00022395470076488845,
      "loss": 2.1635,
      "step": 133997
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.1344841718673706,
      "learning_rate": 0.0002239507438096217,
      "loss": 1.9819,
      "step": 133998
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.4358311891555786,
      "learning_rate": 0.00022394678686849396,
      "loss": 1.9948,
      "step": 133999
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.0898346900939941,
      "learning_rate": 0.00022394282994150607,
      "loss": 2.0898,
      "step": 134000
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.0262421369552612,
      "learning_rate": 0.0002239388730286585,
      "loss": 2.2223,
      "step": 134001
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.1029590368270874,
      "learning_rate": 0.00022393491612995216,
      "loss": 2.2704,
      "step": 134002
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.1554523706436157,
      "learning_rate": 0.00022393095924538774,
      "loss": 2.0669,
      "step": 134003
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.3170342445373535,
      "learning_rate": 0.00022392700237496596,
      "loss": 2.2986,
      "step": 134004
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.0202698707580566,
      "learning_rate": 0.00022392304551868765,
      "loss": 2.0299,
      "step": 134005
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.0603348016738892,
      "learning_rate": 0.00022391908867655356,
      "loss": 2.1586,
      "step": 134006
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.0059770345687866,
      "learning_rate": 0.00022391513184856428,
      "loss": 1.8462,
      "step": 134007
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.2558399438858032,
      "learning_rate": 0.00022391117503472053,
      "loss": 2.0424,
      "step": 134008
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.0619375705718994,
      "learning_rate": 0.00022390721823502317,
      "loss": 2.0732,
      "step": 134009
    },
    {
      "epoch": 1.74,
      "grad_norm": 0.9985530376434326,
      "learning_rate": 0.0002239032614494729,
      "loss": 2.0896,
      "step": 134010
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.049826741218567,
      "learning_rate": 0.00022389930467807043,
      "loss": 2.0516,
      "step": 134011
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.0894783735275269,
      "learning_rate": 0.00022389534792081667,
      "loss": 2.1107,
      "step": 134012
    },
    {
      "epoch": 1.74,
      "grad_norm": 0.9957197904586792,
      "learning_rate": 0.00022389139117771206,
      "loss": 2.2345,
      "step": 134013
    },
    {
      "epoch": 1.74,
      "grad_norm": 0.9558886289596558,
      "learning_rate": 0.00022388743444875744,
      "loss": 2.2855,
      "step": 134014
    },
    {
      "epoch": 1.74,
      "grad_norm": 0.9554316401481628,
      "learning_rate": 0.00022388347773395366,
      "loss": 1.9802,
      "step": 134015
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.0700643062591553,
      "learning_rate": 0.00022387952103330135,
      "loss": 2.393,
      "step": 134016
    },
    {
      "epoch": 1.74,
      "grad_norm": 1.087071180343628,
      "learning_rate": 0.00022387556434680123,
      "loss": 2.1028,
      "step": 134017
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.021223783493042,
      "learning_rate": 0.00022387160767445427,
      "loss": 2.0306,
      "step": 134018
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.0518896579742432,
      "learning_rate": 0.00022386765101626085,
      "loss": 2.1334,
      "step": 134019
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.0866931676864624,
      "learning_rate": 0.00022386369437222184,
      "loss": 2.007,
      "step": 134020
    },
    {
      "epoch": 1.75,
      "grad_norm": 0.9413833022117615,
      "learning_rate": 0.00022385973774233806,
      "loss": 1.9982,
      "step": 134021
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.0685865879058838,
      "learning_rate": 0.0002238557811266102,
      "loss": 2.1864,
      "step": 134022
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.0715067386627197,
      "learning_rate": 0.00022385182452503897,
      "loss": 2.1286,
      "step": 134023
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.0850305557250977,
      "learning_rate": 0.00022384786793762523,
      "loss": 2.0087,
      "step": 134024
    },
    {
      "epoch": 1.75,
      "grad_norm": 0.9797154664993286,
      "learning_rate": 0.0002238439113643695,
      "loss": 1.7515,
      "step": 134025
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.2399725914001465,
      "learning_rate": 0.00022383995480527262,
      "loss": 2.0851,
      "step": 134026
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.1035571098327637,
      "learning_rate": 0.00022383599826033534,
      "loss": 2.2048,
      "step": 134027
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.055717945098877,
      "learning_rate": 0.00022383204172955832,
      "loss": 2.0768,
      "step": 134028
    },
    {
      "epoch": 1.75,
      "grad_norm": 0.9810639023780823,
      "learning_rate": 0.00022382808521294247,
      "loss": 2.1354,
      "step": 134029
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.0606046915054321,
      "learning_rate": 0.00022382412871048847,
      "loss": 2.2247,
      "step": 134030
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.0609464645385742,
      "learning_rate": 0.00022382017222219688,
      "loss": 2.133,
      "step": 134031
    },
    {
      "epoch": 1.75,
      "grad_norm": 0.9484598636627197,
      "learning_rate": 0.00022381621574806855,
      "loss": 2.0293,
      "step": 134032
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.053813099861145,
      "learning_rate": 0.00022381225928810427,
      "loss": 2.0638,
      "step": 134033
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.0731173753738403,
      "learning_rate": 0.00022380830284230474,
      "loss": 2.178,
      "step": 134034
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.1360479593276978,
      "learning_rate": 0.00022380434641067063,
      "loss": 2.1074,
      "step": 134035
    },
    {
      "epoch": 1.75,
      "grad_norm": 0.9694766402244568,
      "learning_rate": 0.00022380038999320287,
      "loss": 2.0229,
      "step": 134036
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.0353775024414062,
      "learning_rate": 0.00022379643358990195,
      "loss": 2.07,
      "step": 134037
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.0015751123428345,
      "learning_rate": 0.0002237924772007687,
      "loss": 2.1885,
      "step": 134038
    },
    {
      "epoch": 1.75,
      "grad_norm": 0.9270858764648438,
      "learning_rate": 0.00022378852082580386,
      "loss": 2.024,
      "step": 134039
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.121821641921997,
      "learning_rate": 0.00022378456446500817,
      "loss": 2.0146,
      "step": 134040
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.0137094259262085,
      "learning_rate": 0.00022378060811838237,
      "loss": 1.8594,
      "step": 134041
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.0192577838897705,
      "learning_rate": 0.00022377665178592736,
      "loss": 2.0568,
      "step": 134042
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.0098941326141357,
      "learning_rate": 0.00022377269546764353,
      "loss": 1.975,
      "step": 134043
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.0582449436187744,
      "learning_rate": 0.00022376873916353182,
      "loss": 2.0982,
      "step": 134044
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.2540018558502197,
      "learning_rate": 0.00022376478287359293,
      "loss": 2.1652,
      "step": 134045
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.112216591835022,
      "learning_rate": 0.00022376082659782763,
      "loss": 2.1575,
      "step": 134046
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.0323131084442139,
      "learning_rate": 0.0002237568703362366,
      "loss": 2.3153,
      "step": 134047
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.1015121936798096,
      "learning_rate": 0.00022375291408882077,
      "loss": 2.1617,
      "step": 134048
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.1043192148208618,
      "learning_rate": 0.00022374895785558056,
      "loss": 2.2377,
      "step": 134049
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.1101865768432617,
      "learning_rate": 0.00022374500163651684,
      "loss": 2.1609,
      "step": 134050
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.0908414125442505,
      "learning_rate": 0.0002237410454316304,
      "loss": 2.103,
      "step": 134051
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.0302284955978394,
      "learning_rate": 0.0002237370892409219,
      "loss": 2.0758,
      "step": 134052
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.0459874868392944,
      "learning_rate": 0.00022373313306439215,
      "loss": 2.2389,
      "step": 134053
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.1061205863952637,
      "learning_rate": 0.00022372917690204183,
      "loss": 1.9064,
      "step": 134054
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.064803957939148,
      "learning_rate": 0.00022372522075387184,
      "loss": 1.8387,
      "step": 134055
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.0491139888763428,
      "learning_rate": 0.00022372126461988261,
      "loss": 2.1231,
      "step": 134056
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.0074537992477417,
      "learning_rate": 0.00022371730850007505,
      "loss": 1.9808,
      "step": 134057
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.1735424995422363,
      "learning_rate": 0.0002237133523944499,
      "loss": 2.2844,
      "step": 134058
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.0214993953704834,
      "learning_rate": 0.0002237093963030079,
      "loss": 1.9901,
      "step": 134059
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.017962098121643,
      "learning_rate": 0.00022370544022574975,
      "loss": 1.9184,
      "step": 134060
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.1454474925994873,
      "learning_rate": 0.0002237014841626762,
      "loss": 2.1311,
      "step": 134061
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.025261640548706,
      "learning_rate": 0.00022369752811378808,
      "loss": 2.0132,
      "step": 134062
    },
    {
      "epoch": 1.75,
      "grad_norm": 0.9785705208778381,
      "learning_rate": 0.00022369357207908591,
      "loss": 2.3184,
      "step": 134063
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.1864678859710693,
      "learning_rate": 0.00022368961605857054,
      "loss": 2.028,
      "step": 134064
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.2475566864013672,
      "learning_rate": 0.00022368566005224273,
      "loss": 1.9593,
      "step": 134065
    },
    {
      "epoch": 1.75,
      "grad_norm": 0.9937849044799805,
      "learning_rate": 0.00022368170406010318,
      "loss": 1.953,
      "step": 134066
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.2910281419754028,
      "learning_rate": 0.00022367774808215274,
      "loss": 2.1751,
      "step": 134067
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.1595189571380615,
      "learning_rate": 0.00022367379211839195,
      "loss": 1.9686,
      "step": 134068
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.1259790658950806,
      "learning_rate": 0.00022366983616882164,
      "loss": 2.1253,
      "step": 134069
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.0558316707611084,
      "learning_rate": 0.00022366588023344262,
      "loss": 2.318,
      "step": 134070
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.0460171699523926,
      "learning_rate": 0.0002236619243122555,
      "loss": 2.0519,
      "step": 134071
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.040751338005066,
      "learning_rate": 0.00022365796840526105,
      "loss": 1.8215,
      "step": 134072
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.120719313621521,
      "learning_rate": 0.00022365401251246009,
      "loss": 2.0984,
      "step": 134073
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.0239295959472656,
      "learning_rate": 0.00022365005663385328,
      "loss": 2.1047,
      "step": 134074
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.0796444416046143,
      "learning_rate": 0.0002236461007694413,
      "loss": 2.0773,
      "step": 134075
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.0482444763183594,
      "learning_rate": 0.00022364214491922497,
      "loss": 2.1187,
      "step": 134076
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.035739779472351,
      "learning_rate": 0.00022363818908320508,
      "loss": 2.0107,
      "step": 134077
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.03506600856781,
      "learning_rate": 0.00022363423326138222,
      "loss": 2.1184,
      "step": 134078
    },
    {
      "epoch": 1.75,
      "grad_norm": 0.9943427443504333,
      "learning_rate": 0.0002236302774537573,
      "loss": 1.9438,
      "step": 134079
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.0623445510864258,
      "learning_rate": 0.00022362632166033082,
      "loss": 2.0187,
      "step": 134080
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.1457271575927734,
      "learning_rate": 0.0002236223658811037,
      "loss": 1.9613,
      "step": 134081
    },
    {
      "epoch": 1.75,
      "grad_norm": 0.9815642237663269,
      "learning_rate": 0.00022361841011607662,
      "loss": 2.1346,
      "step": 134082
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.1688764095306396,
      "learning_rate": 0.00022361445436525028,
      "loss": 2.289,
      "step": 134083
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.278340458869934,
      "learning_rate": 0.00022361049862862557,
      "loss": 1.9554,
      "step": 134084
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.0287315845489502,
      "learning_rate": 0.00022360654290620307,
      "loss": 2.0768,
      "step": 134085
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.0851788520812988,
      "learning_rate": 0.0002236025871979835,
      "loss": 2.1981,
      "step": 134086
    },
    {
      "epoch": 1.75,
      "grad_norm": 0.999269425868988,
      "learning_rate": 0.00022359863150396764,
      "loss": 2.0597,
      "step": 134087
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.977381706237793,
      "learning_rate": 0.00022359467582415627,
      "loss": 2.2215,
      "step": 134088
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.1831705570220947,
      "learning_rate": 0.00022359072015855005,
      "loss": 2.0373,
      "step": 134089
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.0531672239303589,
      "learning_rate": 0.00022358676450714982,
      "loss": 2.0676,
      "step": 134090
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.113167405128479,
      "learning_rate": 0.00022358280886995637,
      "loss": 2.1834,
      "step": 134091
    },
    {
      "epoch": 1.75,
      "grad_norm": 0.9673961997032166,
      "learning_rate": 0.00022357885324697014,
      "loss": 2.1887,
      "step": 134092
    },
    {
      "epoch": 1.75,
      "grad_norm": 0.9861684441566467,
      "learning_rate": 0.0002235748976381921,
      "loss": 2.3101,
      "step": 134093
    },
    {
      "epoch": 1.75,
      "grad_norm": 0.9888747334480286,
      "learning_rate": 0.00022357094204362285,
      "loss": 2.3448,
      "step": 134094
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.0967034101486206,
      "learning_rate": 0.00022356698646326326,
      "loss": 1.995,
      "step": 134095
    },
    {
      "epoch": 1.75,
      "grad_norm": 0.9945986270904541,
      "learning_rate": 0.00022356303089711402,
      "loss": 2.1817,
      "step": 134096
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.1000324487686157,
      "learning_rate": 0.00022355907534517597,
      "loss": 2.2814,
      "step": 134097
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.0396990776062012,
      "learning_rate": 0.0002235551198074496,
      "loss": 2.1413,
      "step": 134098
    },
    {
      "epoch": 1.75,
      "grad_norm": 0.9537009596824646,
      "learning_rate": 0.00022355116428393577,
      "loss": 2.0364,
      "step": 134099
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.0400971174240112,
      "learning_rate": 0.00022354720877463524,
      "loss": 2.0232,
      "step": 134100
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.0667723417282104,
      "learning_rate": 0.00022354325327954869,
      "loss": 2.1529,
      "step": 134101
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.0785902738571167,
      "learning_rate": 0.00022353929779867692,
      "loss": 2.1171,
      "step": 134102
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.0468050241470337,
      "learning_rate": 0.00022353534233202077,
      "loss": 2.0288,
      "step": 134103
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.0698496103286743,
      "learning_rate": 0.0002235313868795807,
      "loss": 2.2866,
      "step": 134104
    },
    {
      "epoch": 1.75,
      "grad_norm": 0.9897513389587402,
      "learning_rate": 0.0002235274314413576,
      "loss": 2.0241,
      "step": 134105
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.173710584640503,
      "learning_rate": 0.0002235234760173522,
      "loss": 2.0798,
      "step": 134106
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.0477073192596436,
      "learning_rate": 0.00022351952060756518,
      "loss": 2.0207,
      "step": 134107
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.0786328315734863,
      "learning_rate": 0.00022351556521199738,
      "loss": 2.1175,
      "step": 134108
    },
    {
      "epoch": 1.75,
      "grad_norm": 0.9545305371284485,
      "learning_rate": 0.0002235116098306496,
      "loss": 2.1097,
      "step": 134109
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.0363036394119263,
      "learning_rate": 0.00022350765446352234,
      "loss": 2.0906,
      "step": 134110
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.1808396577835083,
      "learning_rate": 0.00022350369911061643,
      "loss": 2.5228,
      "step": 134111
    },
    {
      "epoch": 1.75,
      "grad_norm": 0.9867613911628723,
      "learning_rate": 0.0002234997437719326,
      "loss": 1.9145,
      "step": 134112
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.0898783206939697,
      "learning_rate": 0.00022349578844747166,
      "loss": 2.1508,
      "step": 134113
    },
    {
      "epoch": 1.75,
      "grad_norm": 0.9776843190193176,
      "learning_rate": 0.00022349183313723428,
      "loss": 1.9716,
      "step": 134114
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.0429409742355347,
      "learning_rate": 0.00022348787784122134,
      "loss": 1.8632,
      "step": 134115
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.057729959487915,
      "learning_rate": 0.00022348392255943332,
      "loss": 2.282,
      "step": 134116
    },
    {
      "epoch": 1.75,
      "grad_norm": 0.9762057662010193,
      "learning_rate": 0.00022347996729187106,
      "loss": 2.1434,
      "step": 134117
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.002991795539856,
      "learning_rate": 0.00022347601203853536,
      "loss": 2.158,
      "step": 134118
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.07705557346344,
      "learning_rate": 0.00022347205679942685,
      "loss": 2.2082,
      "step": 134119
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.0330157279968262,
      "learning_rate": 0.0002234681015745464,
      "loss": 2.2779,
      "step": 134120
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.0265268087387085,
      "learning_rate": 0.00022346414636389465,
      "loss": 2.2899,
      "step": 134121
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.0216623544692993,
      "learning_rate": 0.00022346019116747248,
      "loss": 1.9902,
      "step": 134122
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.0096434354782104,
      "learning_rate": 0.00022345623598528039,
      "loss": 2.2025,
      "step": 134123
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.0228432416915894,
      "learning_rate": 0.00022345228081731923,
      "loss": 2.206,
      "step": 134124
    },
    {
      "epoch": 1.75,
      "grad_norm": 0.9713631868362427,
      "learning_rate": 0.0002234483256635897,
      "loss": 2.0747,
      "step": 134125
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.014992594718933,
      "learning_rate": 0.0002234443705240926,
      "loss": 2.1973,
      "step": 134126
    },
    {
      "epoch": 1.75,
      "grad_norm": 0.9723671674728394,
      "learning_rate": 0.00022344041539882866,
      "loss": 2.2279,
      "step": 134127
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.0522246360778809,
      "learning_rate": 0.00022343646028779866,
      "loss": 2.2423,
      "step": 134128
    },
    {
      "epoch": 1.75,
      "grad_norm": 0.9801750183105469,
      "learning_rate": 0.0002234325051910032,
      "loss": 2.1658,
      "step": 134129
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.0669888257980347,
      "learning_rate": 0.00022342855010844308,
      "loss": 2.2795,
      "step": 134130
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.0141570568084717,
      "learning_rate": 0.000223424595040119,
      "loss": 2.439,
      "step": 134131
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.0833364725112915,
      "learning_rate": 0.00022342063998603172,
      "loss": 2.0462,
      "step": 134132
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.0582301616668701,
      "learning_rate": 0.00022341668494618204,
      "loss": 2.1993,
      "step": 134133
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.0609444379806519,
      "learning_rate": 0.00022341272992057074,
      "loss": 2.0504,
      "step": 134134
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.0237154960632324,
      "learning_rate": 0.00022340877490919836,
      "loss": 2.0379,
      "step": 134135
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.0593616962432861,
      "learning_rate": 0.0002234048199120657,
      "loss": 2.2645,
      "step": 134136
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.0600848197937012,
      "learning_rate": 0.00022340086492917354,
      "loss": 2.0486,
      "step": 134137
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.070846676826477,
      "learning_rate": 0.0002233969099605226,
      "loss": 2.0946,
      "step": 134138
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.0053255558013916,
      "learning_rate": 0.00022339295500611362,
      "loss": 1.9995,
      "step": 134139
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.0646480321884155,
      "learning_rate": 0.0002233890000659475,
      "loss": 2.3139,
      "step": 134140
    },
    {
      "epoch": 1.75,
      "grad_norm": 0.9653674960136414,
      "learning_rate": 0.00022338504514002466,
      "loss": 2.1802,
      "step": 134141
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.0410254001617432,
      "learning_rate": 0.00022338109022834596,
      "loss": 1.9807,
      "step": 134142
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.910832166671753,
      "learning_rate": 0.00022337713533091222,
      "loss": 2.1669,
      "step": 134143
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.1228824853897095,
      "learning_rate": 0.00022337318044772408,
      "loss": 2.12,
      "step": 134144
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.1160686016082764,
      "learning_rate": 0.0002233692255787823,
      "loss": 2.0155,
      "step": 134145
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.0401321649551392,
      "learning_rate": 0.0002233652707240877,
      "loss": 1.9463,
      "step": 134146
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.0894761085510254,
      "learning_rate": 0.00022336131588364096,
      "loss": 2.1516,
      "step": 134147
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.0651763677597046,
      "learning_rate": 0.00022335736105744272,
      "loss": 2.0364,
      "step": 134148
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.0565924644470215,
      "learning_rate": 0.0002233534062454938,
      "loss": 2.1706,
      "step": 134149
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.0699741840362549,
      "learning_rate": 0.0002233494514477949,
      "loss": 2.0769,
      "step": 134150
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.0279889106750488,
      "learning_rate": 0.00022334549666434684,
      "loss": 1.8705,
      "step": 134151
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.0748093128204346,
      "learning_rate": 0.0002233415418951503,
      "loss": 2.0786,
      "step": 134152
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.0288891792297363,
      "learning_rate": 0.000223337587140206,
      "loss": 2.2183,
      "step": 134153
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.1111046075820923,
      "learning_rate": 0.00022333363239951474,
      "loss": 2.1054,
      "step": 134154
    },
    {
      "epoch": 1.75,
      "grad_norm": 0.9884102940559387,
      "learning_rate": 0.00022332967767307713,
      "loss": 2.2487,
      "step": 134155
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.031696081161499,
      "learning_rate": 0.00022332572296089397,
      "loss": 2.0566,
      "step": 134156
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.0201131105422974,
      "learning_rate": 0.000223321768262966,
      "loss": 2.1698,
      "step": 134157
    },
    {
      "epoch": 1.75,
      "grad_norm": 0.9459729194641113,
      "learning_rate": 0.00022331781357929408,
      "loss": 2.1998,
      "step": 134158
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.0470256805419922,
      "learning_rate": 0.00022331385890987873,
      "loss": 2.2187,
      "step": 134159
    },
    {
      "epoch": 1.75,
      "grad_norm": 0.9936310648918152,
      "learning_rate": 0.00022330990425472078,
      "loss": 2.0229,
      "step": 134160
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.1164844036102295,
      "learning_rate": 0.00022330594961382099,
      "loss": 2.2154,
      "step": 134161
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.2938052415847778,
      "learning_rate": 0.0002233019949871801,
      "loss": 2.2507,
      "step": 134162
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.0257683992385864,
      "learning_rate": 0.00022329804037479875,
      "loss": 2.1942,
      "step": 134163
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.0015534162521362,
      "learning_rate": 0.00022329408577667783,
      "loss": 1.9536,
      "step": 134164
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.1988238096237183,
      "learning_rate": 0.0002232901311928179,
      "loss": 2.2799,
      "step": 134165
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.1542441844940186,
      "learning_rate": 0.0002232861766232198,
      "loss": 1.8835,
      "step": 134166
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.0568844079971313,
      "learning_rate": 0.00022328222206788426,
      "loss": 2.0753,
      "step": 134167
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.0244693756103516,
      "learning_rate": 0.000223278267526812,
      "loss": 2.1054,
      "step": 134168
    },
    {
      "epoch": 1.75,
      "grad_norm": 0.9930199980735779,
      "learning_rate": 0.00022327431300000377,
      "loss": 2.1412,
      "step": 134169
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.2493880987167358,
      "learning_rate": 0.00022327035848746036,
      "loss": 2.1862,
      "step": 134170
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.023099660873413,
      "learning_rate": 0.00022326640398918234,
      "loss": 2.0401,
      "step": 134171
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.0745924711227417,
      "learning_rate": 0.00022326244950517053,
      "loss": 2.3452,
      "step": 134172
    },
    {
      "epoch": 1.75,
      "grad_norm": 0.9944037199020386,
      "learning_rate": 0.00022325849503542573,
      "loss": 2.0327,
      "step": 134173
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.2761123180389404,
      "learning_rate": 0.0002232545405799486,
      "loss": 2.0728,
      "step": 134174
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.007551908493042,
      "learning_rate": 0.00022325058613873987,
      "loss": 2.1025,
      "step": 134175
    },
    {
      "epoch": 1.75,
      "grad_norm": 0.9719732403755188,
      "learning_rate": 0.0002232466317118005,
      "loss": 2.1339,
      "step": 134176
    },
    {
      "epoch": 1.75,
      "grad_norm": 0.9491332769393921,
      "learning_rate": 0.00022324267729913083,
      "loss": 2.0142,
      "step": 134177
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.0365324020385742,
      "learning_rate": 0.00022323872290073185,
      "loss": 2.0709,
      "step": 134178
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.0268354415893555,
      "learning_rate": 0.00022323476851660422,
      "loss": 2.2895,
      "step": 134179
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.101547360420227,
      "learning_rate": 0.00022323081414674865,
      "loss": 2.2789,
      "step": 134180
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.0629491806030273,
      "learning_rate": 0.00022322685979116598,
      "loss": 2.1977,
      "step": 134181
    },
    {
      "epoch": 1.75,
      "grad_norm": 0.9351679682731628,
      "learning_rate": 0.00022322290544985702,
      "loss": 1.9859,
      "step": 134182
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.0554856061935425,
      "learning_rate": 0.0002232189511228222,
      "loss": 1.9841,
      "step": 134183
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.0652356147766113,
      "learning_rate": 0.00022321499681006242,
      "loss": 2.0232,
      "step": 134184
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.0042744874954224,
      "learning_rate": 0.00022321104251157847,
      "loss": 2.2721,
      "step": 134185
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.0245722532272339,
      "learning_rate": 0.00022320708822737103,
      "loss": 1.9784,
      "step": 134186
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.1314315795898438,
      "learning_rate": 0.0002232031339574408,
      "loss": 2.1291,
      "step": 134187
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.1051769256591797,
      "learning_rate": 0.0002231991797017886,
      "loss": 2.1625,
      "step": 134188
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.1380728483200073,
      "learning_rate": 0.0002231952254604152,
      "loss": 2.2544,
      "step": 134189
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.044905185699463,
      "learning_rate": 0.0002231912712333212,
      "loss": 1.9997,
      "step": 134190
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.0807688236236572,
      "learning_rate": 0.00022318731702050733,
      "loss": 2.1828,
      "step": 134191
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.0675591230392456,
      "learning_rate": 0.00022318336282197443,
      "loss": 2.0095,
      "step": 134192
    },
    {
      "epoch": 1.75,
      "grad_norm": 0.9874112606048584,
      "learning_rate": 0.00022317940863772316,
      "loss": 2.0634,
      "step": 134193
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.093550682067871,
      "learning_rate": 0.0002231754544677543,
      "loss": 2.2367,
      "step": 134194
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.0763016939163208,
      "learning_rate": 0.00022317150031206872,
      "loss": 2.0373,
      "step": 134195
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.0083659887313843,
      "learning_rate": 0.00022316754617066685,
      "loss": 1.9388,
      "step": 134196
    },
    {
      "epoch": 1.75,
      "grad_norm": 0.9685947895050049,
      "learning_rate": 0.00022316359204354962,
      "loss": 2.0616,
      "step": 134197
    },
    {
      "epoch": 1.75,
      "grad_norm": 0.9999840259552002,
      "learning_rate": 0.0002231596379307177,
      "loss": 2.0825,
      "step": 134198
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.1951348781585693,
      "learning_rate": 0.00022315568383217188,
      "loss": 2.2011,
      "step": 134199
    },
    {
      "epoch": 1.75,
      "grad_norm": 0.9693160653114319,
      "learning_rate": 0.00022315172974791283,
      "loss": 2.3107,
      "step": 134200
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.229501724243164,
      "learning_rate": 0.00022314777567794148,
      "loss": 2.1475,
      "step": 134201
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.0734807252883911,
      "learning_rate": 0.00022314382162225827,
      "loss": 2.0357,
      "step": 134202
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.0991437435150146,
      "learning_rate": 0.00022313986758086405,
      "loss": 2.1178,
      "step": 134203
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.0477174520492554,
      "learning_rate": 0.00022313591355375967,
      "loss": 1.9488,
      "step": 134204
    },
    {
      "epoch": 1.75,
      "grad_norm": 0.9944434762001038,
      "learning_rate": 0.0002231319595409457,
      "loss": 2.3375,
      "step": 134205
    },
    {
      "epoch": 1.75,
      "grad_norm": 0.9941112995147705,
      "learning_rate": 0.00022312800554242295,
      "loss": 2.2121,
      "step": 134206
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.1910886764526367,
      "learning_rate": 0.00022312405155819226,
      "loss": 2.1264,
      "step": 134207
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.0426732301712036,
      "learning_rate": 0.0002231200975882542,
      "loss": 2.0827,
      "step": 134208
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.0532703399658203,
      "learning_rate": 0.0002231161436326095,
      "loss": 2.255,
      "step": 134209
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.0558147430419922,
      "learning_rate": 0.00022311218969125904,
      "loss": 1.9792,
      "step": 134210
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.1525524854660034,
      "learning_rate": 0.0002231082357642034,
      "loss": 2.1199,
      "step": 134211
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.0739637613296509,
      "learning_rate": 0.0002231042818514434,
      "loss": 2.1544,
      "step": 134212
    },
    {
      "epoch": 1.75,
      "grad_norm": 0.8992601037025452,
      "learning_rate": 0.00022310032795297993,
      "loss": 1.9955,
      "step": 134213
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.1618294715881348,
      "learning_rate": 0.00022309637406881343,
      "loss": 2.0167,
      "step": 134214
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.1804475784301758,
      "learning_rate": 0.00022309242019894472,
      "loss": 2.1191,
      "step": 134215
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.1231727600097656,
      "learning_rate": 0.00022308846634337462,
      "loss": 2.122,
      "step": 134216
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.0876328945159912,
      "learning_rate": 0.00022308451250210382,
      "loss": 2.0734,
      "step": 134217
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.2233413457870483,
      "learning_rate": 0.00022308055867513305,
      "loss": 2.1183,
      "step": 134218
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.4217581748962402,
      "learning_rate": 0.0002230766048624632,
      "loss": 1.977,
      "step": 134219
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.099308729171753,
      "learning_rate": 0.00022307265106409472,
      "loss": 2.2845,
      "step": 134220
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.0419397354125977,
      "learning_rate": 0.0002230686972800285,
      "loss": 2.2842,
      "step": 134221
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.0853853225708008,
      "learning_rate": 0.00022306474351026524,
      "loss": 2.0396,
      "step": 134222
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.2051416635513306,
      "learning_rate": 0.0002230607897548057,
      "loss": 2.1419,
      "step": 134223
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.1219911575317383,
      "learning_rate": 0.0002230568360136506,
      "loss": 2.287,
      "step": 134224
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.0692880153656006,
      "learning_rate": 0.00022305288228680086,
      "loss": 2.3137,
      "step": 134225
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.104649305343628,
      "learning_rate": 0.00022304892857425687,
      "loss": 2.317,
      "step": 134226
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.0066841840744019,
      "learning_rate": 0.0002230449748760196,
      "loss": 2.0869,
      "step": 134227
    },
    {
      "epoch": 1.75,
      "grad_norm": 0.9871699810028076,
      "learning_rate": 0.00022304102119208966,
      "loss": 2.1096,
      "step": 134228
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.049190640449524,
      "learning_rate": 0.00022303706752246785,
      "loss": 2.0188,
      "step": 134229
    },
    {
      "epoch": 1.75,
      "grad_norm": 0.9693763256072998,
      "learning_rate": 0.0002230331138671549,
      "loss": 2.128,
      "step": 134230
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.0711077451705933,
      "learning_rate": 0.00022302916022615168,
      "loss": 2.1573,
      "step": 134231
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.2553776502609253,
      "learning_rate": 0.00022302520659945872,
      "loss": 2.1219,
      "step": 134232
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.0317742824554443,
      "learning_rate": 0.00022302125298707677,
      "loss": 1.9457,
      "step": 134233
    },
    {
      "epoch": 1.75,
      "grad_norm": 0.9136797189712524,
      "learning_rate": 0.00022301729938900664,
      "loss": 2.0963,
      "step": 134234
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.2090221643447876,
      "learning_rate": 0.00022301334580524906,
      "loss": 2.1832,
      "step": 134235
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.0136754512786865,
      "learning_rate": 0.0002230093922358047,
      "loss": 1.8975,
      "step": 134236
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.1189358234405518,
      "learning_rate": 0.00022300543868067447,
      "loss": 1.9399,
      "step": 134237
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.1461011171340942,
      "learning_rate": 0.00022300148513985888,
      "loss": 1.981,
      "step": 134238
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.0722943544387817,
      "learning_rate": 0.00022299753161335884,
      "loss": 1.9911,
      "step": 134239
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.0132801532745361,
      "learning_rate": 0.00022299357810117495,
      "loss": 2.0739,
      "step": 134240
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.0402872562408447,
      "learning_rate": 0.000222989624603308,
      "loss": 2.1429,
      "step": 134241
    },
    {
      "epoch": 1.75,
      "grad_norm": 0.9965912103652954,
      "learning_rate": 0.00022298567111975873,
      "loss": 2.0508,
      "step": 134242
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.0486247539520264,
      "learning_rate": 0.00022298171765052798,
      "loss": 2.3931,
      "step": 134243
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.0082532167434692,
      "learning_rate": 0.0002229777641956163,
      "loss": 2.0297,
      "step": 134244
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.0373872518539429,
      "learning_rate": 0.00022297381075502448,
      "loss": 2.0401,
      "step": 134245
    },
    {
      "epoch": 1.75,
      "grad_norm": 0.9680044054985046,
      "learning_rate": 0.0002229698573287534,
      "loss": 1.9495,
      "step": 134246
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.150803804397583,
      "learning_rate": 0.00022296590391680357,
      "loss": 2.0653,
      "step": 134247
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.0459399223327637,
      "learning_rate": 0.00022296195051917586,
      "loss": 2.2424,
      "step": 134248
    },
    {
      "epoch": 1.75,
      "grad_norm": 0.9869051575660706,
      "learning_rate": 0.000222957997135871,
      "loss": 2.0701,
      "step": 134249
    },
    {
      "epoch": 1.75,
      "grad_norm": 0.9977331757545471,
      "learning_rate": 0.00022295404376688967,
      "loss": 1.9122,
      "step": 134250
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.0512641668319702,
      "learning_rate": 0.00022295009041223263,
      "loss": 2.1938,
      "step": 134251
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.0508575439453125,
      "learning_rate": 0.0002229461370719006,
      "loss": 2.1212,
      "step": 134252
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.025329351425171,
      "learning_rate": 0.00022294218374589438,
      "loss": 2.0336,
      "step": 134253
    },
    {
      "epoch": 1.75,
      "grad_norm": 0.9900878071784973,
      "learning_rate": 0.0002229382304342147,
      "loss": 2.2598,
      "step": 134254
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.012431025505066,
      "learning_rate": 0.0002229342771368622,
      "loss": 2.1299,
      "step": 134255
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.0693379640579224,
      "learning_rate": 0.00022293032385383774,
      "loss": 2.0265,
      "step": 134256
    },
    {
      "epoch": 1.75,
      "grad_norm": 0.9675445556640625,
      "learning_rate": 0.00022292637058514192,
      "loss": 2.0033,
      "step": 134257
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.0775738954544067,
      "learning_rate": 0.00022292241733077556,
      "loss": 1.873,
      "step": 134258
    },
    {
      "epoch": 1.75,
      "grad_norm": 0.99732905626297,
      "learning_rate": 0.00022291846409073934,
      "loss": 2.0619,
      "step": 134259
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.068131923675537,
      "learning_rate": 0.00022291451086503403,
      "loss": 2.0954,
      "step": 134260
    },
    {
      "epoch": 1.75,
      "grad_norm": 0.9856961965560913,
      "learning_rate": 0.00022291055765366046,
      "loss": 2.2706,
      "step": 134261
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.0989668369293213,
      "learning_rate": 0.00022290660445661926,
      "loss": 2.0349,
      "step": 134262
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.0679504871368408,
      "learning_rate": 0.00022290265127391116,
      "loss": 2.312,
      "step": 134263
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.044172763824463,
      "learning_rate": 0.00022289869810553683,
      "loss": 2.0985,
      "step": 134264
    },
    {
      "epoch": 1.75,
      "grad_norm": 0.983597457408905,
      "learning_rate": 0.00022289474495149717,
      "loss": 2.1478,
      "step": 134265
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.0721840858459473,
      "learning_rate": 0.0002228907918117928,
      "loss": 2.0473,
      "step": 134266
    },
    {
      "epoch": 1.75,
      "grad_norm": 0.986789882183075,
      "learning_rate": 0.0002228868386864245,
      "loss": 2.0177,
      "step": 134267
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.0814000368118286,
      "learning_rate": 0.00022288288557539309,
      "loss": 2.251,
      "step": 134268
    },
    {
      "epoch": 1.75,
      "grad_norm": 0.9962129592895508,
      "learning_rate": 0.00022287893247869905,
      "loss": 2.2332,
      "step": 134269
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.1547356843948364,
      "learning_rate": 0.00022287497939634334,
      "loss": 2.1334,
      "step": 134270
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.1237024068832397,
      "learning_rate": 0.00022287102632832662,
      "loss": 2.1412,
      "step": 134271
    },
    {
      "epoch": 1.75,
      "grad_norm": 0.9945606589317322,
      "learning_rate": 0.0002228670732746496,
      "loss": 2.2924,
      "step": 134272
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.3553268909454346,
      "learning_rate": 0.00022286312023531304,
      "loss": 2.0481,
      "step": 134273
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.1067568063735962,
      "learning_rate": 0.00022285916721031783,
      "loss": 2.1119,
      "step": 134274
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.00606107711792,
      "learning_rate": 0.00022285521419966445,
      "loss": 2.0173,
      "step": 134275
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.0322182178497314,
      "learning_rate": 0.00022285126120335372,
      "loss": 2.0628,
      "step": 134276
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.0261884927749634,
      "learning_rate": 0.00022284730822138638,
      "loss": 2.299,
      "step": 134277
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.1408947706222534,
      "learning_rate": 0.00022284335525376322,
      "loss": 2.0176,
      "step": 134278
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.0394744873046875,
      "learning_rate": 0.0002228394023004849,
      "loss": 2.0041,
      "step": 134279
    },
    {
      "epoch": 1.75,
      "grad_norm": 0.9421239495277405,
      "learning_rate": 0.00022283544936155237,
      "loss": 1.9433,
      "step": 134280
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.0090473890304565,
      "learning_rate": 0.000222831496436966,
      "loss": 1.9756,
      "step": 134281
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.1334975957870483,
      "learning_rate": 0.00022282754352672673,
      "loss": 2.1768,
      "step": 134282
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.0585739612579346,
      "learning_rate": 0.0002228235906308353,
      "loss": 2.0257,
      "step": 134283
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.051365613937378,
      "learning_rate": 0.0002228196377492924,
      "loss": 2.1777,
      "step": 134284
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.0606025457382202,
      "learning_rate": 0.00022281568488209875,
      "loss": 2.0347,
      "step": 134285
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.1127341985702515,
      "learning_rate": 0.0002228117320292553,
      "loss": 2.0925,
      "step": 134286
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.0118968486785889,
      "learning_rate": 0.00022280777919076245,
      "loss": 2.0304,
      "step": 134287
    },
    {
      "epoch": 1.75,
      "grad_norm": 0.9801831245422363,
      "learning_rate": 0.00022280382636662112,
      "loss": 2.0837,
      "step": 134288
    },
    {
      "epoch": 1.75,
      "grad_norm": 0.994886040687561,
      "learning_rate": 0.00022279987355683202,
      "loss": 1.9683,
      "step": 134289
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.057732343673706,
      "learning_rate": 0.00022279592076139586,
      "loss": 2.2906,
      "step": 134290
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.02095365524292,
      "learning_rate": 0.0002227919679803134,
      "loss": 2.1095,
      "step": 134291
    },
    {
      "epoch": 1.75,
      "grad_norm": 0.9571731090545654,
      "learning_rate": 0.00022278801521358551,
      "loss": 2.034,
      "step": 134292
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.097692847251892,
      "learning_rate": 0.00022278406246121265,
      "loss": 2.1014,
      "step": 134293
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.018813967704773,
      "learning_rate": 0.00022278010972319568,
      "loss": 1.8895,
      "step": 134294
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.0402129888534546,
      "learning_rate": 0.00022277615699953533,
      "loss": 2.1646,
      "step": 134295
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.0548040866851807,
      "learning_rate": 0.00022277220429023237,
      "loss": 1.9864,
      "step": 134296
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.0769213438034058,
      "learning_rate": 0.00022276825159528755,
      "loss": 2.0461,
      "step": 134297
    },
    {
      "epoch": 1.75,
      "grad_norm": 0.9922546148300171,
      "learning_rate": 0.00022276429891470164,
      "loss": 2.1904,
      "step": 134298
    },
    {
      "epoch": 1.75,
      "grad_norm": 0.9796479940414429,
      "learning_rate": 0.0002227603462484752,
      "loss": 2.1084,
      "step": 134299
    },
    {
      "epoch": 1.75,
      "grad_norm": 0.926559329032898,
      "learning_rate": 0.00022275639359660906,
      "loss": 2.016,
      "step": 134300
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.1729040145874023,
      "learning_rate": 0.000222752440959104,
      "loss": 2.0839,
      "step": 134301
    },
    {
      "epoch": 1.75,
      "grad_norm": 0.9747025966644287,
      "learning_rate": 0.0002227484883359607,
      "loss": 2.0198,
      "step": 134302
    },
    {
      "epoch": 1.75,
      "grad_norm": 0.9930433034896851,
      "learning_rate": 0.00022274453572717988,
      "loss": 2.1629,
      "step": 134303
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.101706862449646,
      "learning_rate": 0.00022274058313276248,
      "loss": 2.1707,
      "step": 134304
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.014634370803833,
      "learning_rate": 0.0002227366305527089,
      "loss": 2.0944,
      "step": 134305
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.1165704727172852,
      "learning_rate": 0.0002227326779870201,
      "loss": 2.1699,
      "step": 134306
    },
    {
      "epoch": 1.75,
      "grad_norm": 0.9508242011070251,
      "learning_rate": 0.0002227287254356967,
      "loss": 2.144,
      "step": 134307
    },
    {
      "epoch": 1.75,
      "grad_norm": 0.9868385791778564,
      "learning_rate": 0.0002227247728987395,
      "loss": 2.0128,
      "step": 134308
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.0516304969787598,
      "learning_rate": 0.00022272082037614924,
      "loss": 2.2016,
      "step": 134309
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.1159411668777466,
      "learning_rate": 0.00022271686786792672,
      "loss": 1.9646,
      "step": 134310
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.1295044422149658,
      "learning_rate": 0.0002227129153740725,
      "loss": 1.9755,
      "step": 134311
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.186026692390442,
      "learning_rate": 0.00022270896289458742,
      "loss": 2.1905,
      "step": 134312
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.0865130424499512,
      "learning_rate": 0.0002227050104294722,
      "loss": 1.9718,
      "step": 134313
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.2237255573272705,
      "learning_rate": 0.00022270105797872757,
      "loss": 1.8597,
      "step": 134314
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.2275053262710571,
      "learning_rate": 0.00022269710554235426,
      "loss": 2.0464,
      "step": 134315
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.0245741605758667,
      "learning_rate": 0.00022269315312035314,
      "loss": 1.7687,
      "step": 134316
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.4710867404937744,
      "learning_rate": 0.00022268920071272474,
      "loss": 2.2612,
      "step": 134317
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.232595682144165,
      "learning_rate": 0.00022268524831946983,
      "loss": 2.1965,
      "step": 134318
    },
    {
      "epoch": 1.75,
      "grad_norm": 0.997200608253479,
      "learning_rate": 0.00022268129594058922,
      "loss": 1.9923,
      "step": 134319
    },
    {
      "epoch": 1.75,
      "grad_norm": 0.9839720129966736,
      "learning_rate": 0.0002226773435760836,
      "loss": 1.998,
      "step": 134320
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.1704374551773071,
      "learning_rate": 0.00022267339122595368,
      "loss": 2.132,
      "step": 134321
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.2018406391143799,
      "learning_rate": 0.00022266943889020027,
      "loss": 2.1303,
      "step": 134322
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.0050331354141235,
      "learning_rate": 0.00022266548656882417,
      "loss": 2.2996,
      "step": 134323
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.0928152799606323,
      "learning_rate": 0.00022266153426182602,
      "loss": 2.014,
      "step": 134324
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.2357066869735718,
      "learning_rate": 0.00022265758196920643,
      "loss": 2.247,
      "step": 134325
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.2457131147384644,
      "learning_rate": 0.0002226536296909663,
      "loss": 2.3031,
      "step": 134326
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.04001784324646,
      "learning_rate": 0.0002226496774271063,
      "loss": 2.1045,
      "step": 134327
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.0362608432769775,
      "learning_rate": 0.0002226457251776272,
      "loss": 2.0869,
      "step": 134328
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.1253453493118286,
      "learning_rate": 0.00022264177294252976,
      "loss": 2.1812,
      "step": 134329
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.0335859060287476,
      "learning_rate": 0.00022263782072181462,
      "loss": 2.1087,
      "step": 134330
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.057348370552063,
      "learning_rate": 0.00022263386851548267,
      "loss": 1.9837,
      "step": 134331
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.060848355293274,
      "learning_rate": 0.00022262991632353442,
      "loss": 2.267,
      "step": 134332
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.0248942375183105,
      "learning_rate": 0.00022262596414597075,
      "loss": 1.9988,
      "step": 134333
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.116747498512268,
      "learning_rate": 0.00022262201198279237,
      "loss": 2.0447,
      "step": 134334
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.007069706916809,
      "learning_rate": 0.00022261805983400012,
      "loss": 2.1641,
      "step": 134335
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.0352882146835327,
      "learning_rate": 0.00022261410769959452,
      "loss": 2.2183,
      "step": 134336
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.0262354612350464,
      "learning_rate": 0.00022261015557957647,
      "loss": 2.3143,
      "step": 134337
    },
    {
      "epoch": 1.75,
      "grad_norm": 0.9467888474464417,
      "learning_rate": 0.00022260620347394668,
      "loss": 2.1346,
      "step": 134338
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.1666936874389648,
      "learning_rate": 0.00022260225138270577,
      "loss": 1.8626,
      "step": 134339
    },
    {
      "epoch": 1.75,
      "grad_norm": 0.9663038849830627,
      "learning_rate": 0.0002225982993058546,
      "loss": 1.9712,
      "step": 134340
    },
    {
      "epoch": 1.75,
      "grad_norm": 0.9537268877029419,
      "learning_rate": 0.00022259434724339395,
      "loss": 2.1942,
      "step": 134341
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.0036505460739136,
      "learning_rate": 0.00022259039519532438,
      "loss": 2.0763,
      "step": 134342
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.165738582611084,
      "learning_rate": 0.00022258644316164674,
      "loss": 2.1863,
      "step": 134343
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.094766616821289,
      "learning_rate": 0.0002225824911423617,
      "loss": 2.19,
      "step": 134344
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.080540657043457,
      "learning_rate": 0.0002225785391374701,
      "loss": 2.0238,
      "step": 134345
    },
    {
      "epoch": 1.75,
      "grad_norm": 0.9831828474998474,
      "learning_rate": 0.00022257458714697258,
      "loss": 2.1254,
      "step": 134346
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.1265689134597778,
      "learning_rate": 0.00022257063517086995,
      "loss": 2.2415,
      "step": 134347
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.0392919778823853,
      "learning_rate": 0.00022256668320916288,
      "loss": 2.0869,
      "step": 134348
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.0972999334335327,
      "learning_rate": 0.00022256273126185206,
      "loss": 2.2048,
      "step": 134349
    },
    {
      "epoch": 1.75,
      "grad_norm": 0.9983106851577759,
      "learning_rate": 0.00022255877932893833,
      "loss": 2.3245,
      "step": 134350
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.0698589086532593,
      "learning_rate": 0.00022255482741042237,
      "loss": 2.0373,
      "step": 134351
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.121391773223877,
      "learning_rate": 0.00022255087550630495,
      "loss": 2.2141,
      "step": 134352
    },
    {
      "epoch": 1.75,
      "grad_norm": 0.931814432144165,
      "learning_rate": 0.0002225469236165869,
      "loss": 2.1071,
      "step": 134353
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.0567713975906372,
      "learning_rate": 0.0002225429717412687,
      "loss": 2.0377,
      "step": 134354
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.0306121110916138,
      "learning_rate": 0.00022253901988035122,
      "loss": 2.2027,
      "step": 134355
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.002657413482666,
      "learning_rate": 0.00022253506803383518,
      "loss": 2.2127,
      "step": 134356
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.0126639604568481,
      "learning_rate": 0.00022253111620172138,
      "loss": 2.1478,
      "step": 134357
    },
    {
      "epoch": 1.75,
      "grad_norm": 0.930719256401062,
      "learning_rate": 0.0002225271643840105,
      "loss": 1.9878,
      "step": 134358
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.0336741209030151,
      "learning_rate": 0.0002225232125807034,
      "loss": 2.1321,
      "step": 134359
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.0386364459991455,
      "learning_rate": 0.00022251926079180058,
      "loss": 1.9919,
      "step": 134360
    },
    {
      "epoch": 1.75,
      "grad_norm": 0.9784460663795471,
      "learning_rate": 0.00022251530901730286,
      "loss": 2.1243,
      "step": 134361
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.0397261381149292,
      "learning_rate": 0.00022251135725721104,
      "loss": 2.1283,
      "step": 134362
    },
    {
      "epoch": 1.75,
      "grad_norm": 0.9977457523345947,
      "learning_rate": 0.0002225074055115258,
      "loss": 1.9659,
      "step": 134363
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.1672849655151367,
      "learning_rate": 0.00022250345378024792,
      "loss": 2.1591,
      "step": 134364
    },
    {
      "epoch": 1.75,
      "grad_norm": 0.9957180619239807,
      "learning_rate": 0.00022249950206337823,
      "loss": 2.2303,
      "step": 134365
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.0906786918640137,
      "learning_rate": 0.0002224955503609172,
      "loss": 2.1926,
      "step": 134366
    },
    {
      "epoch": 1.75,
      "grad_norm": 0.9897679686546326,
      "learning_rate": 0.00022249159867286574,
      "loss": 2.1781,
      "step": 134367
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.0542397499084473,
      "learning_rate": 0.0002224876469992245,
      "loss": 2.3025,
      "step": 134368
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.1040359735488892,
      "learning_rate": 0.00022248369533999434,
      "loss": 1.9696,
      "step": 134369
    },
    {
      "epoch": 1.75,
      "grad_norm": 0.9591059684753418,
      "learning_rate": 0.0002224797436951759,
      "loss": 2.0102,
      "step": 134370
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.074962854385376,
      "learning_rate": 0.00022247579206477005,
      "loss": 2.1053,
      "step": 134371
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.0411065816879272,
      "learning_rate": 0.0002224718404487773,
      "loss": 2.0322,
      "step": 134372
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.0865451097488403,
      "learning_rate": 0.0002224678888471985,
      "loss": 2.3953,
      "step": 134373
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.0271248817443848,
      "learning_rate": 0.00022246393726003434,
      "loss": 2.231,
      "step": 134374
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.0389673709869385,
      "learning_rate": 0.00022245998568728564,
      "loss": 2.1041,
      "step": 134375
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.0306707620620728,
      "learning_rate": 0.0002224560341289531,
      "loss": 2.1265,
      "step": 134376
    },
    {
      "epoch": 1.75,
      "grad_norm": 0.9750356078147888,
      "learning_rate": 0.00022245208258503754,
      "loss": 2.1038,
      "step": 134377
    },
    {
      "epoch": 1.75,
      "grad_norm": 0.9477857351303101,
      "learning_rate": 0.0002224481310555395,
      "loss": 2.1171,
      "step": 134378
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.0763803720474243,
      "learning_rate": 0.0002224441795404598,
      "loss": 2.0503,
      "step": 134379
    },
    {
      "epoch": 1.75,
      "grad_norm": 0.9745498299598694,
      "learning_rate": 0.00022244022803979925,
      "loss": 2.0324,
      "step": 134380
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.0928940773010254,
      "learning_rate": 0.0002224362765535584,
      "loss": 2.0467,
      "step": 134381
    },
    {
      "epoch": 1.75,
      "grad_norm": 0.9939475655555725,
      "learning_rate": 0.00022243232508173825,
      "loss": 2.2824,
      "step": 134382
    },
    {
      "epoch": 1.75,
      "grad_norm": 0.933322012424469,
      "learning_rate": 0.00022242837362433944,
      "loss": 1.9362,
      "step": 134383
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.0355219841003418,
      "learning_rate": 0.00022242442218136255,
      "loss": 2.1849,
      "step": 134384
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.120823621749878,
      "learning_rate": 0.00022242047075280843,
      "loss": 2.3455,
      "step": 134385
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.053638219833374,
      "learning_rate": 0.0002224165193386778,
      "loss": 2.0302,
      "step": 134386
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.021634817123413,
      "learning_rate": 0.0002224125679389714,
      "loss": 2.1409,
      "step": 134387
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.0327869653701782,
      "learning_rate": 0.00022240861655369,
      "loss": 2.1938,
      "step": 134388
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.0388513803482056,
      "learning_rate": 0.00022240466518283426,
      "loss": 2.1078,
      "step": 134389
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.0247138738632202,
      "learning_rate": 0.00022240071382640512,
      "loss": 2.184,
      "step": 134390
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.1213524341583252,
      "learning_rate": 0.00022239676248440298,
      "loss": 2.3015,
      "step": 134391
    },
    {
      "epoch": 1.75,
      "grad_norm": 0.9712890386581421,
      "learning_rate": 0.0002223928111568288,
      "loss": 2.2793,
      "step": 134392
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.0078505277633667,
      "learning_rate": 0.00022238885984368325,
      "loss": 2.244,
      "step": 134393
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.19430673122406,
      "learning_rate": 0.00022238490854496705,
      "loss": 2.0606,
      "step": 134394
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.0022050142288208,
      "learning_rate": 0.00022238095726068098,
      "loss": 1.7788,
      "step": 134395
    },
    {
      "epoch": 1.75,
      "grad_norm": 0.9873535633087158,
      "learning_rate": 0.0002223770059908259,
      "loss": 2.0547,
      "step": 134396
    },
    {
      "epoch": 1.75,
      "grad_norm": 0.9833565950393677,
      "learning_rate": 0.00022237305473540224,
      "loss": 1.9956,
      "step": 134397
    },
    {
      "epoch": 1.75,
      "grad_norm": 0.9674350619316101,
      "learning_rate": 0.00022236910349441092,
      "loss": 2.1339,
      "step": 134398
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.0813802480697632,
      "learning_rate": 0.0002223651522678526,
      "loss": 1.8161,
      "step": 134399
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.0315834283828735,
      "learning_rate": 0.00022236120105572808,
      "loss": 2.3439,
      "step": 134400
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.0051956176757812,
      "learning_rate": 0.0002223572498580381,
      "loss": 2.2829,
      "step": 134401
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.0064970254898071,
      "learning_rate": 0.0002223532986747835,
      "loss": 2.0002,
      "step": 134402
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.039233922958374,
      "learning_rate": 0.00022234934750596473,
      "loss": 2.1887,
      "step": 134403
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.094424843788147,
      "learning_rate": 0.00022234539635158273,
      "loss": 2.1844,
      "step": 134404
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.0767767429351807,
      "learning_rate": 0.00022234144521163818,
      "loss": 2.0714,
      "step": 134405
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.0462762117385864,
      "learning_rate": 0.00022233749408613179,
      "loss": 2.0099,
      "step": 134406
    },
    {
      "epoch": 1.75,
      "grad_norm": 0.9893214106559753,
      "learning_rate": 0.00022233354297506434,
      "loss": 1.9853,
      "step": 134407
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.1398310661315918,
      "learning_rate": 0.00022232959187843668,
      "loss": 1.7857,
      "step": 134408
    },
    {
      "epoch": 1.75,
      "grad_norm": 0.9943724274635315,
      "learning_rate": 0.0002223256407962493,
      "loss": 1.9564,
      "step": 134409
    },
    {
      "epoch": 1.75,
      "grad_norm": 0.9760682582855225,
      "learning_rate": 0.00022232168972850303,
      "loss": 2.0186,
      "step": 134410
    },
    {
      "epoch": 1.75,
      "grad_norm": 0.9749899506568909,
      "learning_rate": 0.00022231773867519862,
      "loss": 2.3602,
      "step": 134411
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.0348548889160156,
      "learning_rate": 0.00022231378763633682,
      "loss": 1.9258,
      "step": 134412
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.0110284090042114,
      "learning_rate": 0.00022230983661191835,
      "loss": 2.1024,
      "step": 134413
    },
    {
      "epoch": 1.75,
      "grad_norm": 0.9370189309120178,
      "learning_rate": 0.00022230588560194403,
      "loss": 2.1569,
      "step": 134414
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.0693620443344116,
      "learning_rate": 0.00022230193460641442,
      "loss": 2.23,
      "step": 134415
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.0830494165420532,
      "learning_rate": 0.00022229798362533045,
      "loss": 2.2607,
      "step": 134416
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.031138300895691,
      "learning_rate": 0.00022229403265869267,
      "loss": 2.2123,
      "step": 134417
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.053128957748413,
      "learning_rate": 0.00022229008170650185,
      "loss": 2.306,
      "step": 134418
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.0448408126831055,
      "learning_rate": 0.0002222861307687588,
      "loss": 2.0934,
      "step": 134419
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.044177770614624,
      "learning_rate": 0.0002222821798454643,
      "loss": 2.1881,
      "step": 134420
    },
    {
      "epoch": 1.75,
      "grad_norm": 0.9992002248764038,
      "learning_rate": 0.00022227822893661892,
      "loss": 2.0355,
      "step": 134421
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.0631364583969116,
      "learning_rate": 0.00022227427804222356,
      "loss": 2.224,
      "step": 134422
    },
    {
      "epoch": 1.75,
      "grad_norm": 0.9703700542449951,
      "learning_rate": 0.00022227032716227887,
      "loss": 2.0971,
      "step": 134423
    },
    {
      "epoch": 1.75,
      "grad_norm": 0.9906730055809021,
      "learning_rate": 0.00022226637629678555,
      "loss": 2.0641,
      "step": 134424
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.1672431230545044,
      "learning_rate": 0.0002222624254457444,
      "loss": 2.2682,
      "step": 134425
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.1778737306594849,
      "learning_rate": 0.00022225847460915615,
      "loss": 2.1803,
      "step": 134426
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.166059970855713,
      "learning_rate": 0.00022225452378702148,
      "loss": 2.08,
      "step": 134427
    },
    {
      "epoch": 1.75,
      "grad_norm": 0.9929381012916565,
      "learning_rate": 0.00022225057297934115,
      "loss": 1.9569,
      "step": 134428
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.1803618669509888,
      "learning_rate": 0.00022224662218611591,
      "loss": 2.2359,
      "step": 134429
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.1406219005584717,
      "learning_rate": 0.00022224267140734654,
      "loss": 2.1438,
      "step": 134430
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.027092456817627,
      "learning_rate": 0.00022223872064303373,
      "loss": 1.9901,
      "step": 134431
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.0055501461029053,
      "learning_rate": 0.00022223476989317818,
      "loss": 2.2256,
      "step": 134432
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.0357221364974976,
      "learning_rate": 0.00022223081915778066,
      "loss": 1.9272,
      "step": 134433
    },
    {
      "epoch": 1.75,
      "grad_norm": 0.9787582755088806,
      "learning_rate": 0.00022222686843684187,
      "loss": 2.0798,
      "step": 134434
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.0427683591842651,
      "learning_rate": 0.00022222291773036257,
      "loss": 2.1212,
      "step": 134435
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.0188857316970825,
      "learning_rate": 0.00022221896703834352,
      "loss": 2.0404,
      "step": 134436
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.122564435005188,
      "learning_rate": 0.00022221501636078545,
      "loss": 2.1609,
      "step": 134437
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.067589521408081,
      "learning_rate": 0.00022221106569768911,
      "loss": 2.1605,
      "step": 134438
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.0700390338897705,
      "learning_rate": 0.00022220711504905514,
      "loss": 2.2616,
      "step": 134439
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.0292185544967651,
      "learning_rate": 0.00022220316441488435,
      "loss": 2.1071,
      "step": 134440
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.1313332319259644,
      "learning_rate": 0.00022219921379517744,
      "loss": 2.0898,
      "step": 134441
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.0645757913589478,
      "learning_rate": 0.00022219526318993516,
      "loss": 2.075,
      "step": 134442
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.1154167652130127,
      "learning_rate": 0.00022219131259915826,
      "loss": 2.1665,
      "step": 134443
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.0357897281646729,
      "learning_rate": 0.00022218736202284756,
      "loss": 2.2638,
      "step": 134444
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.1033380031585693,
      "learning_rate": 0.00022218341146100363,
      "loss": 2.0564,
      "step": 134445
    },
    {
      "epoch": 1.75,
      "grad_norm": 0.992628812789917,
      "learning_rate": 0.00022217946091362723,
      "loss": 2.1849,
      "step": 134446
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.087477207183838,
      "learning_rate": 0.00022217551038071915,
      "loss": 2.1288,
      "step": 134447
    },
    {
      "epoch": 1.75,
      "grad_norm": 0.9971834421157837,
      "learning_rate": 0.0002221715598622801,
      "loss": 1.8732,
      "step": 134448
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.0671876668930054,
      "learning_rate": 0.00022216760935831083,
      "loss": 2.2853,
      "step": 134449
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.2210036516189575,
      "learning_rate": 0.00022216365886881223,
      "loss": 1.9222,
      "step": 134450
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.0838217735290527,
      "learning_rate": 0.00022215970839378472,
      "loss": 2.1137,
      "step": 134451
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.0557920932769775,
      "learning_rate": 0.00022215575793322918,
      "loss": 2.1693,
      "step": 134452
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.0412216186523438,
      "learning_rate": 0.00022215180748714634,
      "loss": 1.9231,
      "step": 134453
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.2170321941375732,
      "learning_rate": 0.00022214785705553699,
      "loss": 2.2791,
      "step": 134454
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.046220064163208,
      "learning_rate": 0.0002221439066384018,
      "loss": 2.0971,
      "step": 134455
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.0772621631622314,
      "learning_rate": 0.00022213995623574154,
      "loss": 2.2255,
      "step": 134456
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.204940915107727,
      "learning_rate": 0.00022213600584755703,
      "loss": 2.1273,
      "step": 134457
    },
    {
      "epoch": 1.75,
      "grad_norm": 0.9800441861152649,
      "learning_rate": 0.00022213205547384882,
      "loss": 1.8496,
      "step": 134458
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.084991216659546,
      "learning_rate": 0.0002221281051146177,
      "loss": 2.296,
      "step": 134459
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.1128970384597778,
      "learning_rate": 0.00022212415476986444,
      "loss": 2.0691,
      "step": 134460
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.1689494848251343,
      "learning_rate": 0.0002221202044395898,
      "loss": 2.0684,
      "step": 134461
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.2299100160598755,
      "learning_rate": 0.00022211625412379445,
      "loss": 2.0953,
      "step": 134462
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.035918951034546,
      "learning_rate": 0.00022211230382247928,
      "loss": 2.1465,
      "step": 134463
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.20521879196167,
      "learning_rate": 0.0002221083535356448,
      "loss": 2.1582,
      "step": 134464
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.0711475610733032,
      "learning_rate": 0.00022210440326329183,
      "loss": 2.0497,
      "step": 134465
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.0831979513168335,
      "learning_rate": 0.00022210045300542113,
      "loss": 2.1933,
      "step": 134466
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.0478595495224,
      "learning_rate": 0.00022209650276203342,
      "loss": 1.9999,
      "step": 134467
    },
    {
      "epoch": 1.75,
      "grad_norm": 0.9907345175743103,
      "learning_rate": 0.00022209255253312945,
      "loss": 2.1256,
      "step": 134468
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.0332856178283691,
      "learning_rate": 0.00022208860231871005,
      "loss": 2.3058,
      "step": 134469
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.1541175842285156,
      "learning_rate": 0.00022208465211877578,
      "loss": 2.1845,
      "step": 134470
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.0668342113494873,
      "learning_rate": 0.0002220807019333274,
      "loss": 2.1596,
      "step": 134471
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.095265507698059,
      "learning_rate": 0.0002220767517623657,
      "loss": 2.0017,
      "step": 134472
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.0265530347824097,
      "learning_rate": 0.0002220728016058914,
      "loss": 2.037,
      "step": 134473
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.0505045652389526,
      "learning_rate": 0.0002220688514639052,
      "loss": 2.3431,
      "step": 134474
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.00925612449646,
      "learning_rate": 0.00022206490133640808,
      "loss": 2.3981,
      "step": 134475
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.0752471685409546,
      "learning_rate": 0.0002220609512234004,
      "loss": 2.0617,
      "step": 134476
    },
    {
      "epoch": 1.75,
      "grad_norm": 0.9481716752052307,
      "learning_rate": 0.00022205700112488304,
      "loss": 1.9811,
      "step": 134477
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.1396359205245972,
      "learning_rate": 0.00022205305104085673,
      "loss": 2.2018,
      "step": 134478
    },
    {
      "epoch": 1.75,
      "grad_norm": 0.932655394077301,
      "learning_rate": 0.0002220491009713223,
      "loss": 1.8798,
      "step": 134479
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.0239404439926147,
      "learning_rate": 0.00022204515091628037,
      "loss": 1.897,
      "step": 134480
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.1128876209259033,
      "learning_rate": 0.00022204120087573187,
      "loss": 2.1944,
      "step": 134481
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.102591633796692,
      "learning_rate": 0.00022203725084967725,
      "loss": 1.9756,
      "step": 134482
    },
    {
      "epoch": 1.75,
      "grad_norm": 0.9538726806640625,
      "learning_rate": 0.00022203330083811734,
      "loss": 2.0702,
      "step": 134483
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.1003950834274292,
      "learning_rate": 0.00022202935084105293,
      "loss": 2.3372,
      "step": 134484
    },
    {
      "epoch": 1.75,
      "grad_norm": 0.8841786980628967,
      "learning_rate": 0.00022202540085848474,
      "loss": 2.0339,
      "step": 134485
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.1233354806900024,
      "learning_rate": 0.00022202145089041353,
      "loss": 2.1126,
      "step": 134486
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.0190564393997192,
      "learning_rate": 0.0002220175009368401,
      "loss": 1.9987,
      "step": 134487
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.040255069732666,
      "learning_rate": 0.00022201355099776496,
      "loss": 2.0814,
      "step": 134488
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.1093268394470215,
      "learning_rate": 0.00022200960107318898,
      "loss": 2.2961,
      "step": 134489
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.0031585693359375,
      "learning_rate": 0.0002220056511631129,
      "loss": 2.2986,
      "step": 134490
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.0027997493743896,
      "learning_rate": 0.0002220017012675374,
      "loss": 2.1849,
      "step": 134491
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.0365015268325806,
      "learning_rate": 0.0002219977513864633,
      "loss": 2.235,
      "step": 134492
    },
    {
      "epoch": 1.75,
      "grad_norm": 0.9536851048469543,
      "learning_rate": 0.0002219938015198914,
      "loss": 2.2731,
      "step": 134493
    },
    {
      "epoch": 1.75,
      "grad_norm": 0.9724196195602417,
      "learning_rate": 0.0002219898516678222,
      "loss": 2.1564,
      "step": 134494
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.0023919343948364,
      "learning_rate": 0.00022198590183025656,
      "loss": 1.9966,
      "step": 134495
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.0810692310333252,
      "learning_rate": 0.0002219819520071952,
      "loss": 2.0973,
      "step": 134496
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.0657637119293213,
      "learning_rate": 0.0002219780021986389,
      "loss": 2.1629,
      "step": 134497
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.29450523853302,
      "learning_rate": 0.00022197405240458835,
      "loss": 2.2192,
      "step": 134498
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.0983576774597168,
      "learning_rate": 0.00022197010262504433,
      "loss": 2.4016,
      "step": 134499
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.0765055418014526,
      "learning_rate": 0.00022196615286000757,
      "loss": 1.9516,
      "step": 134500
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.0117543935775757,
      "learning_rate": 0.0002219622031094787,
      "loss": 2.007,
      "step": 134501
    },
    {
      "epoch": 1.75,
      "grad_norm": 0.9619292616844177,
      "learning_rate": 0.00022195825337345854,
      "loss": 2.2145,
      "step": 134502
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.3123170137405396,
      "learning_rate": 0.0002219543036519478,
      "loss": 2.2075,
      "step": 134503
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.1567825078964233,
      "learning_rate": 0.00022195035394494727,
      "loss": 2.3134,
      "step": 134504
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.1620545387268066,
      "learning_rate": 0.00022194640425245767,
      "loss": 2.135,
      "step": 134505
    },
    {
      "epoch": 1.75,
      "grad_norm": 0.9393638372421265,
      "learning_rate": 0.00022194245457447963,
      "loss": 1.9868,
      "step": 134506
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.1429059505462646,
      "learning_rate": 0.00022193850491101404,
      "loss": 2.3066,
      "step": 134507
    },
    {
      "epoch": 1.75,
      "grad_norm": 0.9455762505531311,
      "learning_rate": 0.00022193455526206148,
      "loss": 2.0315,
      "step": 134508
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.0653865337371826,
      "learning_rate": 0.00022193060562762277,
      "loss": 1.8802,
      "step": 134509
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.157334566116333,
      "learning_rate": 0.00022192665600769866,
      "loss": 2.3977,
      "step": 134510
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.1027770042419434,
      "learning_rate": 0.0002219227064022899,
      "loss": 2.1188,
      "step": 134511
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.101820945739746,
      "learning_rate": 0.00022191875681139712,
      "loss": 1.9891,
      "step": 134512
    },
    {
      "epoch": 1.75,
      "grad_norm": 0.9996576309204102,
      "learning_rate": 0.0002219148072350211,
      "loss": 2.2689,
      "step": 134513
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.1253751516342163,
      "learning_rate": 0.0002219108576731626,
      "loss": 2.167,
      "step": 134514
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.1190719604492188,
      "learning_rate": 0.00022190690812582244,
      "loss": 2.3924,
      "step": 134515
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.1208107471466064,
      "learning_rate": 0.00022190295859300115,
      "loss": 2.1016,
      "step": 134516
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.0343393087387085,
      "learning_rate": 0.00022189900907469963,
      "loss": 2.0834,
      "step": 134517
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.0456806421279907,
      "learning_rate": 0.00022189505957091857,
      "loss": 2.132,
      "step": 134518
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.0033966302871704,
      "learning_rate": 0.00022189111008165865,
      "loss": 2.1668,
      "step": 134519
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.121578574180603,
      "learning_rate": 0.00022188716060692062,
      "loss": 1.8904,
      "step": 134520
    },
    {
      "epoch": 1.75,
      "grad_norm": 0.984127938747406,
      "learning_rate": 0.00022188321114670527,
      "loss": 1.9113,
      "step": 134521
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.028321623802185,
      "learning_rate": 0.00022187926170101337,
      "loss": 2.1045,
      "step": 134522
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.0591766834259033,
      "learning_rate": 0.00022187531226984555,
      "loss": 2.1578,
      "step": 134523
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.0461000204086304,
      "learning_rate": 0.0002218713628532026,
      "loss": 2.0236,
      "step": 134524
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.1171278953552246,
      "learning_rate": 0.00022186741345108519,
      "loss": 1.9817,
      "step": 134525
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.0130856037139893,
      "learning_rate": 0.0002218634640634941,
      "loss": 2.0428,
      "step": 134526
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.1519289016723633,
      "learning_rate": 0.00022185951469043006,
      "loss": 2.0326,
      "step": 134527
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.1303203105926514,
      "learning_rate": 0.00022185556533189382,
      "loss": 2.0314,
      "step": 134528
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.1627944707870483,
      "learning_rate": 0.00022185161598788617,
      "loss": 2.2182,
      "step": 134529
    },
    {
      "epoch": 1.75,
      "grad_norm": 0.9686737060546875,
      "learning_rate": 0.0002218476666584078,
      "loss": 2.2123,
      "step": 134530
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.0363973379135132,
      "learning_rate": 0.00022184371734345934,
      "loss": 2.3887,
      "step": 134531
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.109732985496521,
      "learning_rate": 0.00022183976804304162,
      "loss": 2.1107,
      "step": 134532
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.2583624124526978,
      "learning_rate": 0.00022183581875715537,
      "loss": 2.309,
      "step": 134533
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.1267859935760498,
      "learning_rate": 0.0002218318694858013,
      "loss": 2.0784,
      "step": 134534
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.1348882913589478,
      "learning_rate": 0.00022182792022898014,
      "loss": 2.2117,
      "step": 134535
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.0264739990234375,
      "learning_rate": 0.0002218239709866928,
      "loss": 2.2133,
      "step": 134536
    },
    {
      "epoch": 1.75,
      "grad_norm": 0.9926275014877319,
      "learning_rate": 0.00022182002175893972,
      "loss": 2.0836,
      "step": 134537
    },
    {
      "epoch": 1.75,
      "grad_norm": 0.9559766054153442,
      "learning_rate": 0.0002218160725457218,
      "loss": 1.9626,
      "step": 134538
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.0622498989105225,
      "learning_rate": 0.00022181212334703972,
      "loss": 2.149,
      "step": 134539
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.0949617624282837,
      "learning_rate": 0.00022180817416289422,
      "loss": 2.1139,
      "step": 134540
    },
    {
      "epoch": 1.75,
      "grad_norm": 0.9276134967803955,
      "learning_rate": 0.0002218042249932861,
      "loss": 1.8431,
      "step": 134541
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.0187571048736572,
      "learning_rate": 0.00022180027583821617,
      "loss": 2.1493,
      "step": 134542
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.0955380201339722,
      "learning_rate": 0.00022179632669768493,
      "loss": 2.2346,
      "step": 134543
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.0099464654922485,
      "learning_rate": 0.00022179237757169318,
      "loss": 2.1472,
      "step": 134544
    },
    {
      "epoch": 1.75,
      "grad_norm": 0.9597747325897217,
      "learning_rate": 0.00022178842846024173,
      "loss": 2.1667,
      "step": 134545
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.0790256261825562,
      "learning_rate": 0.00022178447936333133,
      "loss": 2.0963,
      "step": 134546
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.0148706436157227,
      "learning_rate": 0.00022178053028096262,
      "loss": 2.2389,
      "step": 134547
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.029957890510559,
      "learning_rate": 0.00022177658121313654,
      "loss": 2.0087,
      "step": 134548
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.063623070716858,
      "learning_rate": 0.0002217726321598535,
      "loss": 2.1389,
      "step": 134549
    },
    {
      "epoch": 1.75,
      "grad_norm": 0.977768063545227,
      "learning_rate": 0.00022176868312111445,
      "loss": 2.1827,
      "step": 134550
    },
    {
      "epoch": 1.75,
      "grad_norm": 0.9851308465003967,
      "learning_rate": 0.00022176473409692008,
      "loss": 2.0561,
      "step": 134551
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.0469870567321777,
      "learning_rate": 0.00022176078508727106,
      "loss": 2.1259,
      "step": 134552
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.2438037395477295,
      "learning_rate": 0.00022175683609216827,
      "loss": 2.2657,
      "step": 134553
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.050144076347351,
      "learning_rate": 0.00022175288711161244,
      "loss": 2.1401,
      "step": 134554
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.1539050340652466,
      "learning_rate": 0.00022174893814560413,
      "loss": 1.8162,
      "step": 134555
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.1644080877304077,
      "learning_rate": 0.00022174498919414414,
      "loss": 2.0632,
      "step": 134556
    },
    {
      "epoch": 1.75,
      "grad_norm": 0.9881242513656616,
      "learning_rate": 0.00022174104025723322,
      "loss": 2.1105,
      "step": 134557
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.0313467979431152,
      "learning_rate": 0.00022173709133487216,
      "loss": 2.237,
      "step": 134558
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.3809810876846313,
      "learning_rate": 0.00022173314242706163,
      "loss": 2.2327,
      "step": 134559
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.1603236198425293,
      "learning_rate": 0.00022172919353380253,
      "loss": 2.0577,
      "step": 134560
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.017128825187683,
      "learning_rate": 0.0002217252446550953,
      "loss": 2.1156,
      "step": 134561
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.2011178731918335,
      "learning_rate": 0.0002217212957909408,
      "loss": 2.085,
      "step": 134562
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.4718507528305054,
      "learning_rate": 0.00022171734694133985,
      "loss": 1.944,
      "step": 134563
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.0384845733642578,
      "learning_rate": 0.00022171339810629309,
      "loss": 2.0836,
      "step": 134564
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.0653908252716064,
      "learning_rate": 0.0002217094492858013,
      "loss": 1.9564,
      "step": 134565
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.382483959197998,
      "learning_rate": 0.00022170550047986533,
      "loss": 2.2556,
      "step": 134566
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.0526580810546875,
      "learning_rate": 0.00022170155168848565,
      "loss": 2.2281,
      "step": 134567
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.0275682210922241,
      "learning_rate": 0.00022169760291166311,
      "loss": 1.8233,
      "step": 134568
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.0928640365600586,
      "learning_rate": 0.0002216936541493985,
      "loss": 2.0154,
      "step": 134569
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.0764111280441284,
      "learning_rate": 0.00022168970540169247,
      "loss": 1.979,
      "step": 134570
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.1919457912445068,
      "learning_rate": 0.00022168575666854583,
      "loss": 2.224,
      "step": 134571
    },
    {
      "epoch": 1.75,
      "grad_norm": 0.9780120849609375,
      "learning_rate": 0.00022168180794995943,
      "loss": 2.2051,
      "step": 134572
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.0317589044570923,
      "learning_rate": 0.00022167785924593372,
      "loss": 2.2784,
      "step": 134573
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.0040029287338257,
      "learning_rate": 0.00022167391055646956,
      "loss": 2.0386,
      "step": 134574
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.0470820665359497,
      "learning_rate": 0.00022166996188156774,
      "loss": 2.107,
      "step": 134575
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.0345180034637451,
      "learning_rate": 0.0002216660132212289,
      "loss": 2.0315,
      "step": 134576
    },
    {
      "epoch": 1.75,
      "grad_norm": 0.9617114663124084,
      "learning_rate": 0.00022166206457545384,
      "loss": 2.0682,
      "step": 134577
    },
    {
      "epoch": 1.75,
      "grad_norm": 0.969864547252655,
      "learning_rate": 0.00022165811594424342,
      "loss": 2.2146,
      "step": 134578
    },
    {
      "epoch": 1.75,
      "grad_norm": 0.9615187048912048,
      "learning_rate": 0.00022165416732759812,
      "loss": 2.0778,
      "step": 134579
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.100810170173645,
      "learning_rate": 0.00022165021872551878,
      "loss": 1.985,
      "step": 134580
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.1467891931533813,
      "learning_rate": 0.00022164627013800613,
      "loss": 2.4203,
      "step": 134581
    },
    {
      "epoch": 1.75,
      "grad_norm": 0.9632890820503235,
      "learning_rate": 0.00022164232156506093,
      "loss": 2.1631,
      "step": 134582
    },
    {
      "epoch": 1.75,
      "grad_norm": 0.9744455218315125,
      "learning_rate": 0.00022163837300668388,
      "loss": 1.9984,
      "step": 134583
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.0194427967071533,
      "learning_rate": 0.00022163442446287583,
      "loss": 1.8227,
      "step": 134584
    },
    {
      "epoch": 1.75,
      "grad_norm": 0.9585704207420349,
      "learning_rate": 0.0002216304759336374,
      "loss": 2.1995,
      "step": 134585
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.9290663003921509,
      "learning_rate": 0.00022162652741896928,
      "loss": 1.9718,
      "step": 134586
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.392195701599121,
      "learning_rate": 0.00022162257891887227,
      "loss": 2.1406,
      "step": 134587
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.0711640119552612,
      "learning_rate": 0.00022161863043334713,
      "loss": 1.9016,
      "step": 134588
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.026069164276123,
      "learning_rate": 0.00022161468196239454,
      "loss": 2.0829,
      "step": 134589
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.0115996599197388,
      "learning_rate": 0.00022161073350601527,
      "loss": 2.2012,
      "step": 134590
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.4317975044250488,
      "learning_rate": 0.0002216067850642101,
      "loss": 2.1465,
      "step": 134591
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.1081033945083618,
      "learning_rate": 0.0002216028366369797,
      "loss": 2.0902,
      "step": 134592
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.0031076669692993,
      "learning_rate": 0.00022159888822432475,
      "loss": 2.2212,
      "step": 134593
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.138145923614502,
      "learning_rate": 0.00022159493982624608,
      "loss": 2.2563,
      "step": 134594
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.2613111734390259,
      "learning_rate": 0.00022159099144274436,
      "loss": 2.1312,
      "step": 134595
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.000695824623108,
      "learning_rate": 0.0002215870430738204,
      "loss": 2.0878,
      "step": 134596
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.0623605251312256,
      "learning_rate": 0.00022158309471947488,
      "loss": 2.0648,
      "step": 134597
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.027920126914978,
      "learning_rate": 0.0002215791463797085,
      "loss": 1.9266,
      "step": 134598
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.1884737014770508,
      "learning_rate": 0.00022157519805452212,
      "loss": 1.8865,
      "step": 134599
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.1726089715957642,
      "learning_rate": 0.0002215712497439163,
      "loss": 2.1291,
      "step": 134600
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.062872052192688,
      "learning_rate": 0.00022156730144789192,
      "loss": 2.2479,
      "step": 134601
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.0446252822875977,
      "learning_rate": 0.0002215633531664496,
      "loss": 2.004,
      "step": 134602
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.2495626211166382,
      "learning_rate": 0.0002215594048995902,
      "loss": 2.283,
      "step": 134603
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.0582753419876099,
      "learning_rate": 0.00022155545664731435,
      "loss": 1.9786,
      "step": 134604
    },
    {
      "epoch": 1.75,
      "grad_norm": 0.9514442682266235,
      "learning_rate": 0.00022155150840962284,
      "loss": 1.9738,
      "step": 134605
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.0108569860458374,
      "learning_rate": 0.0002215475601865164,
      "loss": 2.0838,
      "step": 134606
    },
    {
      "epoch": 1.75,
      "grad_norm": 0.9884206056594849,
      "learning_rate": 0.00022154361197799572,
      "loss": 2.0082,
      "step": 134607
    },
    {
      "epoch": 1.75,
      "grad_norm": 0.9532360434532166,
      "learning_rate": 0.00022153966378406158,
      "loss": 2.2143,
      "step": 134608
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.0825631618499756,
      "learning_rate": 0.00022153571560471475,
      "loss": 2.3428,
      "step": 134609
    },
    {
      "epoch": 1.75,
      "grad_norm": 0.9819733500480652,
      "learning_rate": 0.0002215317674399558,
      "loss": 1.9987,
      "step": 134610
    },
    {
      "epoch": 1.75,
      "grad_norm": 0.9956238865852356,
      "learning_rate": 0.00022152781928978563,
      "loss": 1.9947,
      "step": 134611
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.2224972248077393,
      "learning_rate": 0.0002215238711542049,
      "loss": 2.204,
      "step": 134612
    },
    {
      "epoch": 1.75,
      "grad_norm": 0.9925976991653442,
      "learning_rate": 0.00022151992303321436,
      "loss": 2.347,
      "step": 134613
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.1061397790908813,
      "learning_rate": 0.0002215159749268148,
      "loss": 2.0761,
      "step": 134614
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.0731970071792603,
      "learning_rate": 0.00022151202683500692,
      "loss": 2.0792,
      "step": 134615
    },
    {
      "epoch": 1.75,
      "grad_norm": 0.9928761124610901,
      "learning_rate": 0.00022150807875779137,
      "loss": 1.7168,
      "step": 134616
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.1289944648742676,
      "learning_rate": 0.00022150413069516893,
      "loss": 1.9133,
      "step": 134617
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.0048640966415405,
      "learning_rate": 0.0002215001826471404,
      "loss": 2.0443,
      "step": 134618
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.0231293439865112,
      "learning_rate": 0.00022149623461370648,
      "loss": 1.8456,
      "step": 134619
    },
    {
      "epoch": 1.75,
      "grad_norm": 0.9777171611785889,
      "learning_rate": 0.00022149228659486782,
      "loss": 2.0767,
      "step": 134620
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.079876184463501,
      "learning_rate": 0.0002214883385906254,
      "loss": 2.0959,
      "step": 134621
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.0588093996047974,
      "learning_rate": 0.00022148439060097963,
      "loss": 1.9027,
      "step": 134622
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.05790376663208,
      "learning_rate": 0.0002214804426259314,
      "loss": 2.1535,
      "step": 134623
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.025302529335022,
      "learning_rate": 0.00022147649466548145,
      "loss": 2.2625,
      "step": 134624
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.405326247215271,
      "learning_rate": 0.0002214725467196305,
      "loss": 2.1504,
      "step": 134625
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.1194794178009033,
      "learning_rate": 0.00022146859878837923,
      "loss": 2.1433,
      "step": 134626
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.387042760848999,
      "learning_rate": 0.00022146465087172863,
      "loss": 2.2981,
      "step": 134627
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.0736032724380493,
      "learning_rate": 0.00022146070296967911,
      "loss": 2.003,
      "step": 134628
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.4333381652832031,
      "learning_rate": 0.0002214567550822315,
      "loss": 2.0026,
      "step": 134629
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.5550504922866821,
      "learning_rate": 0.0002214528072093865,
      "loss": 2.129,
      "step": 134630
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.0035622119903564,
      "learning_rate": 0.000221448859351145,
      "loss": 2.0204,
      "step": 134631
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.0034711360931396,
      "learning_rate": 0.0002214449115075076,
      "loss": 2.0148,
      "step": 134632
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.00240957736969,
      "learning_rate": 0.0002214409636784752,
      "loss": 2.1101,
      "step": 134633
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.5510777235031128,
      "learning_rate": 0.0002214370158640483,
      "loss": 2.2306,
      "step": 134634
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.0041587352752686,
      "learning_rate": 0.00022143306806422772,
      "loss": 2.141,
      "step": 134635
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.107871174812317,
      "learning_rate": 0.00022142912027901424,
      "loss": 1.9948,
      "step": 134636
    },
    {
      "epoch": 1.75,
      "grad_norm": 0.9886553883552551,
      "learning_rate": 0.00022142517250840857,
      "loss": 2.1832,
      "step": 134637
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.0325281620025635,
      "learning_rate": 0.0002214212247524114,
      "loss": 2.1402,
      "step": 134638
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.1695079803466797,
      "learning_rate": 0.00022141727701102367,
      "loss": 2.065,
      "step": 134639
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.0972448587417603,
      "learning_rate": 0.0002214133292842458,
      "loss": 2.2641,
      "step": 134640
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.10811185836792,
      "learning_rate": 0.0002214093815720787,
      "loss": 2.1906,
      "step": 134641
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.0895568132400513,
      "learning_rate": 0.00022140543387452307,
      "loss": 2.2099,
      "step": 134642
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.064604640007019,
      "learning_rate": 0.00022140148619157968,
      "loss": 2.1773,
      "step": 134643
    },
    {
      "epoch": 1.75,
      "grad_norm": 0.9753413200378418,
      "learning_rate": 0.0002213975385232492,
      "loss": 1.983,
      "step": 134644
    },
    {
      "epoch": 1.75,
      "grad_norm": 0.9984485507011414,
      "learning_rate": 0.00022139359086953257,
      "loss": 2.1364,
      "step": 134645
    },
    {
      "epoch": 1.75,
      "grad_norm": 0.9509835243225098,
      "learning_rate": 0.0002213896432304302,
      "loss": 2.1005,
      "step": 134646
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.015708327293396,
      "learning_rate": 0.00022138569560594296,
      "loss": 1.7981,
      "step": 134647
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.1017563343048096,
      "learning_rate": 0.00022138174799607164,
      "loss": 1.9935,
      "step": 134648
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.0670084953308105,
      "learning_rate": 0.00022137780040081692,
      "loss": 2.3262,
      "step": 134649
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.0875380039215088,
      "learning_rate": 0.00022137385282017954,
      "loss": 1.9266,
      "step": 134650
    },
    {
      "epoch": 1.75,
      "grad_norm": 0.9565132260322571,
      "learning_rate": 0.0002213699052541604,
      "loss": 1.837,
      "step": 134651
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.0813783407211304,
      "learning_rate": 0.0002213659577027599,
      "loss": 2.1041,
      "step": 134652
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.1254280805587769,
      "learning_rate": 0.000221362010165979,
      "loss": 2.178,
      "step": 134653
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.0155861377716064,
      "learning_rate": 0.00022135806264381836,
      "loss": 2.0849,
      "step": 134654
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.0161796808242798,
      "learning_rate": 0.0002213541151362788,
      "loss": 1.9502,
      "step": 134655
    },
    {
      "epoch": 1.75,
      "grad_norm": 0.9860970973968506,
      "learning_rate": 0.0002213501676433609,
      "loss": 2.3375,
      "step": 134656
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.137311577796936,
      "learning_rate": 0.00022134622016506557,
      "loss": 2.0245,
      "step": 134657
    },
    {
      "epoch": 1.75,
      "grad_norm": 0.8879144191741943,
      "learning_rate": 0.00022134227270139353,
      "loss": 2.0895,
      "step": 134658
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.0498998165130615,
      "learning_rate": 0.00022133832525234536,
      "loss": 2.0962,
      "step": 134659
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.1759066581726074,
      "learning_rate": 0.0002213343778179219,
      "loss": 1.9389,
      "step": 134660
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.0208619832992554,
      "learning_rate": 0.0002213304303981238,
      "loss": 2.0184,
      "step": 134661
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.0287846326828003,
      "learning_rate": 0.00022132648299295188,
      "loss": 2.0471,
      "step": 134662
    },
    {
      "epoch": 1.75,
      "grad_norm": 0.9454028606414795,
      "learning_rate": 0.0002213225356024069,
      "loss": 1.9561,
      "step": 134663
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.112695336341858,
      "learning_rate": 0.00022131858822648962,
      "loss": 2.2376,
      "step": 134664
    },
    {
      "epoch": 1.75,
      "grad_norm": 0.9251900315284729,
      "learning_rate": 0.00022131464086520058,
      "loss": 1.9659,
      "step": 134665
    },
    {
      "epoch": 1.75,
      "grad_norm": 0.9609667062759399,
      "learning_rate": 0.00022131069351854064,
      "loss": 2.226,
      "step": 134666
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.0403172969818115,
      "learning_rate": 0.00022130674618651053,
      "loss": 2.3407,
      "step": 134667
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.0869027376174927,
      "learning_rate": 0.00022130279886911096,
      "loss": 1.938,
      "step": 134668
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.2164859771728516,
      "learning_rate": 0.0002212988515663427,
      "loss": 2.0687,
      "step": 134669
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.0158851146697998,
      "learning_rate": 0.0002212949042782066,
      "loss": 2.0189,
      "step": 134670
    },
    {
      "epoch": 1.75,
      "grad_norm": 0.926199734210968,
      "learning_rate": 0.00022129095700470315,
      "loss": 2.0374,
      "step": 134671
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.0427364110946655,
      "learning_rate": 0.00022128700974583318,
      "loss": 2.2229,
      "step": 134672
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.0712082386016846,
      "learning_rate": 0.00022128306250159745,
      "loss": 2.1584,
      "step": 134673
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.1687833070755005,
      "learning_rate": 0.00022127911527199665,
      "loss": 2.0707,
      "step": 134674
    },
    {
      "epoch": 1.75,
      "grad_norm": 0.9934539198875427,
      "learning_rate": 0.0002212751680570316,
      "loss": 2.2084,
      "step": 134675
    },
    {
      "epoch": 1.75,
      "grad_norm": 0.9322477579116821,
      "learning_rate": 0.00022127122085670303,
      "loss": 1.8147,
      "step": 134676
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.0318409204483032,
      "learning_rate": 0.00022126727367101161,
      "loss": 1.9545,
      "step": 134677
    },
    {
      "epoch": 1.75,
      "grad_norm": 0.9704926013946533,
      "learning_rate": 0.000221263326499958,
      "loss": 2.0966,
      "step": 134678
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.1034849882125854,
      "learning_rate": 0.00022125937934354309,
      "loss": 2.1473,
      "step": 134679
    },
    {
      "epoch": 1.75,
      "grad_norm": 0.937125027179718,
      "learning_rate": 0.00022125543220176755,
      "loss": 1.9565,
      "step": 134680
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.1847108602523804,
      "learning_rate": 0.00022125148507463207,
      "loss": 2.065,
      "step": 134681
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.1116169691085815,
      "learning_rate": 0.00022124753796213747,
      "loss": 2.0243,
      "step": 134682
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.1030128002166748,
      "learning_rate": 0.0002212435908642844,
      "loss": 2.0236,
      "step": 134683
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.0203737020492554,
      "learning_rate": 0.0002212396437810737,
      "loss": 1.9984,
      "step": 134684
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.0192095041275024,
      "learning_rate": 0.00022123569671250598,
      "loss": 2.269,
      "step": 134685
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.1055666208267212,
      "learning_rate": 0.00022123174965858202,
      "loss": 2.1237,
      "step": 134686
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.0373038053512573,
      "learning_rate": 0.0002212278026193026,
      "loss": 2.1726,
      "step": 134687
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.0499712228775024,
      "learning_rate": 0.00022122385559466843,
      "loss": 2.1529,
      "step": 134688
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.1054610013961792,
      "learning_rate": 0.0002212199085846802,
      "loss": 1.9728,
      "step": 134689
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.0967401266098022,
      "learning_rate": 0.00022121596158933865,
      "loss": 2.0743,
      "step": 134690
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.064395785331726,
      "learning_rate": 0.00022121201460864463,
      "loss": 2.0957,
      "step": 134691
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.0349332094192505,
      "learning_rate": 0.00022120806764259872,
      "loss": 2.167,
      "step": 134692
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.0463868379592896,
      "learning_rate": 0.00022120412069120168,
      "loss": 2.058,
      "step": 134693
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.026932716369629,
      "learning_rate": 0.00022120017375445436,
      "loss": 2.058,
      "step": 134694
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.013820767402649,
      "learning_rate": 0.0002211962268323574,
      "loss": 2.1166,
      "step": 134695
    },
    {
      "epoch": 1.75,
      "grad_norm": 0.9917023777961731,
      "learning_rate": 0.0002211922799249115,
      "loss": 2.0788,
      "step": 134696
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.0494575500488281,
      "learning_rate": 0.00022118833303211744,
      "loss": 2.1446,
      "step": 134697
    },
    {
      "epoch": 1.75,
      "grad_norm": 0.9921441078186035,
      "learning_rate": 0.00022118438615397596,
      "loss": 1.9701,
      "step": 134698
    },
    {
      "epoch": 1.75,
      "grad_norm": 0.9766712188720703,
      "learning_rate": 0.00022118043929048788,
      "loss": 1.9575,
      "step": 134699
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.0627726316452026,
      "learning_rate": 0.00022117649244165382,
      "loss": 2.1678,
      "step": 134700
    },
    {
      "epoch": 1.75,
      "grad_norm": 0.9923608899116516,
      "learning_rate": 0.00022117254560747449,
      "loss": 2.0124,
      "step": 134701
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.0012956857681274,
      "learning_rate": 0.00022116859878795067,
      "loss": 2.1128,
      "step": 134702
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.113287329673767,
      "learning_rate": 0.0002211646519830831,
      "loss": 2.0894,
      "step": 134703
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.0985029935836792,
      "learning_rate": 0.0002211607051928725,
      "loss": 2.0165,
      "step": 134704
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.0245475769042969,
      "learning_rate": 0.00022115675841731958,
      "loss": 2.3282,
      "step": 134705
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.041562795639038,
      "learning_rate": 0.0002211528116564253,
      "loss": 2.0729,
      "step": 134706
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.1656134128570557,
      "learning_rate": 0.00022114886491019002,
      "loss": 2.053,
      "step": 134707
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.011878252029419,
      "learning_rate": 0.0002211449181786147,
      "loss": 1.9934,
      "step": 134708
    },
    {
      "epoch": 1.75,
      "grad_norm": 0.9855649471282959,
      "learning_rate": 0.0002211409714617,
      "loss": 2.1241,
      "step": 134709
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.0876903533935547,
      "learning_rate": 0.00022113702475944668,
      "loss": 2.2963,
      "step": 134710
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.076766014099121,
      "learning_rate": 0.00022113307807185547,
      "loss": 1.9702,
      "step": 134711
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.1172412633895874,
      "learning_rate": 0.00022112913139892728,
      "loss": 2.44,
      "step": 134712
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.0892164707183838,
      "learning_rate": 0.0002211251847406625,
      "loss": 2.0728,
      "step": 134713
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.0373444557189941,
      "learning_rate": 0.00022112123809706204,
      "loss": 2.2346,
      "step": 134714
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.0517523288726807,
      "learning_rate": 0.00022111729146812663,
      "loss": 2.321,
      "step": 134715
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.1809083223342896,
      "learning_rate": 0.00022111334485385703,
      "loss": 2.0083,
      "step": 134716
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.03533136844635,
      "learning_rate": 0.00022110939825425392,
      "loss": 1.8919,
      "step": 134717
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.005624532699585,
      "learning_rate": 0.00022110545166931822,
      "loss": 2.2045,
      "step": 134718
    },
    {
      "epoch": 1.75,
      "grad_norm": 0.9882259964942932,
      "learning_rate": 0.00022110150509905036,
      "loss": 2.1135,
      "step": 134719
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.0268317461013794,
      "learning_rate": 0.00022109755854345122,
      "loss": 2.1303,
      "step": 134720
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.2814613580703735,
      "learning_rate": 0.00022109361200252156,
      "loss": 1.9325,
      "step": 134721
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.0347096920013428,
      "learning_rate": 0.00022108966547626204,
      "loss": 2.0824,
      "step": 134722
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.0634599924087524,
      "learning_rate": 0.00022108571896467343,
      "loss": 2.2324,
      "step": 134723
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.1247141361236572,
      "learning_rate": 0.00022108177246775654,
      "loss": 2.0948,
      "step": 134724
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.03862464427948,
      "learning_rate": 0.0002210778259855121,
      "loss": 2.0932,
      "step": 134725
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.0973252058029175,
      "learning_rate": 0.0002210738795179407,
      "loss": 2.132,
      "step": 134726
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.000364899635315,
      "learning_rate": 0.00022106993306504318,
      "loss": 2.1761,
      "step": 134727
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.1257739067077637,
      "learning_rate": 0.0002210659866268202,
      "loss": 2.1366,
      "step": 134728
    },
    {
      "epoch": 1.75,
      "grad_norm": 0.9764453172683716,
      "learning_rate": 0.00022106204020327255,
      "loss": 2.0277,
      "step": 134729
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.0199309587478638,
      "learning_rate": 0.00022105809379440098,
      "loss": 2.0968,
      "step": 134730
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.0056837797164917,
      "learning_rate": 0.00022105414740020629,
      "loss": 2.096,
      "step": 134731
    },
    {
      "epoch": 1.75,
      "grad_norm": 0.998504638671875,
      "learning_rate": 0.00022105020102068903,
      "loss": 1.9491,
      "step": 134732
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.0666208267211914,
      "learning_rate": 0.00022104625465585,
      "loss": 2.0459,
      "step": 134733
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.051285982131958,
      "learning_rate": 0.00022104230830569002,
      "loss": 1.9932,
      "step": 134734
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.190238118171692,
      "learning_rate": 0.00022103836197020967,
      "loss": 2.1703,
      "step": 134735
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.0126025676727295,
      "learning_rate": 0.00022103441564940988,
      "loss": 2.0456,
      "step": 134736
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.0102248191833496,
      "learning_rate": 0.00022103046934329133,
      "loss": 2.0637,
      "step": 134737
    },
    {
      "epoch": 1.75,
      "grad_norm": 0.9808989763259888,
      "learning_rate": 0.00022102652305185463,
      "loss": 2.0753,
      "step": 134738
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.1405425071716309,
      "learning_rate": 0.00022102257677510055,
      "loss": 1.8829,
      "step": 134739
    },
    {
      "epoch": 1.75,
      "grad_norm": 0.983633279800415,
      "learning_rate": 0.0002210186305130299,
      "loss": 1.9716,
      "step": 134740
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.0182164907455444,
      "learning_rate": 0.00022101468426564331,
      "loss": 2.0436,
      "step": 134741
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.1311447620391846,
      "learning_rate": 0.00022101073803294165,
      "loss": 2.0535,
      "step": 134742
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.0241624116897583,
      "learning_rate": 0.0002210067918149257,
      "loss": 2.0076,
      "step": 134743
    },
    {
      "epoch": 1.75,
      "grad_norm": 0.9578649401664734,
      "learning_rate": 0.00022100284561159597,
      "loss": 1.9966,
      "step": 134744
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.0799529552459717,
      "learning_rate": 0.00022099889942295324,
      "loss": 2.0722,
      "step": 134745
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.0522688627243042,
      "learning_rate": 0.0002209949532489984,
      "loss": 2.1295,
      "step": 134746
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.1768076419830322,
      "learning_rate": 0.00022099100708973204,
      "loss": 2.0354,
      "step": 134747
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.1404722929000854,
      "learning_rate": 0.00022098706094515492,
      "loss": 2.225,
      "step": 134748
    },
    {
      "epoch": 1.75,
      "grad_norm": 0.9411575198173523,
      "learning_rate": 0.00022098311481526792,
      "loss": 1.9041,
      "step": 134749
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.521106243133545,
      "learning_rate": 0.00022097916870007152,
      "loss": 2.0209,
      "step": 134750
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.1942251920700073,
      "learning_rate": 0.00022097522259956665,
      "loss": 2.2482,
      "step": 134751
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.1819390058517456,
      "learning_rate": 0.00022097127651375393,
      "loss": 2.1113,
      "step": 134752
    },
    {
      "epoch": 1.75,
      "grad_norm": 0.9716997146606445,
      "learning_rate": 0.00022096733044263414,
      "loss": 2.0881,
      "step": 134753
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.0595487356185913,
      "learning_rate": 0.00022096338438620805,
      "loss": 2.1689,
      "step": 134754
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.3553847074508667,
      "learning_rate": 0.00022095943834447645,
      "loss": 2.3421,
      "step": 134755
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.4591501951217651,
      "learning_rate": 0.00022095549231743988,
      "loss": 2.0158,
      "step": 134756
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.0288037061691284,
      "learning_rate": 0.00022095154630509913,
      "loss": 1.8821,
      "step": 134757
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.166614294052124,
      "learning_rate": 0.00022094760030745501,
      "loss": 2.1067,
      "step": 134758
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.4054280519485474,
      "learning_rate": 0.00022094365432450823,
      "loss": 2.0676,
      "step": 134759
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.164671778678894,
      "learning_rate": 0.0002209397083562595,
      "loss": 2.1441,
      "step": 134760
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.306640863418579,
      "learning_rate": 0.0002209357624027097,
      "loss": 2.1794,
      "step": 134761
    },
    {
      "epoch": 1.75,
      "grad_norm": 0.9303595423698425,
      "learning_rate": 0.00022093181646385933,
      "loss": 2.0824,
      "step": 134762
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.419413447380066,
      "learning_rate": 0.0002209278705397092,
      "loss": 1.9387,
      "step": 134763
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.1775844097137451,
      "learning_rate": 0.00022092392463026007,
      "loss": 2.2173,
      "step": 134764
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.2405989170074463,
      "learning_rate": 0.00022091997873551272,
      "loss": 2.1201,
      "step": 134765
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.0497398376464844,
      "learning_rate": 0.00022091603285546781,
      "loss": 2.1589,
      "step": 134766
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.0285589694976807,
      "learning_rate": 0.00022091208699012615,
      "loss": 2.0302,
      "step": 134767
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.1545119285583496,
      "learning_rate": 0.00022090814113948838,
      "loss": 2.2169,
      "step": 134768
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.1483052968978882,
      "learning_rate": 0.00022090419530355532,
      "loss": 2.1389,
      "step": 134769
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.0845235586166382,
      "learning_rate": 0.00022090024948232762,
      "loss": 2.1123,
      "step": 134770
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.0761067867279053,
      "learning_rate": 0.00022089630367580607,
      "loss": 2.1142,
      "step": 134771
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.339309573173523,
      "learning_rate": 0.0002208923578839914,
      "loss": 2.1906,
      "step": 134772
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.0555413961410522,
      "learning_rate": 0.00022088841210688437,
      "loss": 1.9508,
      "step": 134773
    },
    {
      "epoch": 1.75,
      "grad_norm": 0.974107027053833,
      "learning_rate": 0.00022088446634448562,
      "loss": 2.0795,
      "step": 134774
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.0562429428100586,
      "learning_rate": 0.0002208805205967959,
      "loss": 2.4453,
      "step": 134775
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.0533982515335083,
      "learning_rate": 0.00022087657486381609,
      "loss": 2.4022,
      "step": 134776
    },
    {
      "epoch": 1.75,
      "grad_norm": 0.981515645980835,
      "learning_rate": 0.00022087262914554677,
      "loss": 2.1613,
      "step": 134777
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.0157381296157837,
      "learning_rate": 0.0002208686834419887,
      "loss": 2.0461,
      "step": 134778
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.0515508651733398,
      "learning_rate": 0.0002208647377531427,
      "loss": 1.966,
      "step": 134779
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.017197608947754,
      "learning_rate": 0.0002208607920790094,
      "loss": 2.0707,
      "step": 134780
    },
    {
      "epoch": 1.75,
      "grad_norm": 0.9732163548469543,
      "learning_rate": 0.00022085684641958952,
      "loss": 2.0893,
      "step": 134781
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.0989456176757812,
      "learning_rate": 0.0002208529007748839,
      "loss": 2.1446,
      "step": 134782
    },
    {
      "epoch": 1.75,
      "grad_norm": 0.9280810356140137,
      "learning_rate": 0.00022084895514489327,
      "loss": 1.8966,
      "step": 134783
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.045739769935608,
      "learning_rate": 0.00022084500952961823,
      "loss": 2.1597,
      "step": 134784
    },
    {
      "epoch": 1.75,
      "grad_norm": 1.026015043258667,
      "learning_rate": 0.00022084106392905966,
      "loss": 2.0406,
      "step": 134785
    },
    {
      "epoch": 1.76,
      "grad_norm": 0.9400380253791809,
      "learning_rate": 0.0002208371183432182,
      "loss": 1.8898,
      "step": 134786
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.1014595031738281,
      "learning_rate": 0.00022083317277209457,
      "loss": 2.2469,
      "step": 134787
    },
    {
      "epoch": 1.76,
      "grad_norm": 0.9826924204826355,
      "learning_rate": 0.00022082922721568956,
      "loss": 2.0291,
      "step": 134788
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.1302874088287354,
      "learning_rate": 0.00022082528167400393,
      "loss": 2.24,
      "step": 134789
    },
    {
      "epoch": 1.76,
      "grad_norm": 0.9879928827285767,
      "learning_rate": 0.00022082133614703838,
      "loss": 1.8453,
      "step": 134790
    },
    {
      "epoch": 1.76,
      "grad_norm": 0.9944538474082947,
      "learning_rate": 0.0002208173906347936,
      "loss": 2.0706,
      "step": 134791
    },
    {
      "epoch": 1.76,
      "grad_norm": 0.9769276976585388,
      "learning_rate": 0.00022081344513727043,
      "loss": 2.0306,
      "step": 134792
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.062484860420227,
      "learning_rate": 0.0002208094996544695,
      "loss": 2.2218,
      "step": 134793
    },
    {
      "epoch": 1.76,
      "grad_norm": 0.9714208245277405,
      "learning_rate": 0.00022080555418639152,
      "loss": 2.2563,
      "step": 134794
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.031427264213562,
      "learning_rate": 0.00022080160873303733,
      "loss": 2.1317,
      "step": 134795
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.0700675249099731,
      "learning_rate": 0.00022079766329440758,
      "loss": 2.1367,
      "step": 134796
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.1077228784561157,
      "learning_rate": 0.00022079371787050308,
      "loss": 2.3377,
      "step": 134797
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.1700780391693115,
      "learning_rate": 0.0002207897724613246,
      "loss": 2.1488,
      "step": 134798
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.0096970796585083,
      "learning_rate": 0.00022078582706687273,
      "loss": 2.0688,
      "step": 134799
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.0006821155548096,
      "learning_rate": 0.0002207818816871482,
      "loss": 1.9873,
      "step": 134800
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.2174879312515259,
      "learning_rate": 0.00022077793632215188,
      "loss": 2.0411,
      "step": 134801
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.0074455738067627,
      "learning_rate": 0.0002207739909718844,
      "loss": 1.8827,
      "step": 134802
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.0531651973724365,
      "learning_rate": 0.00022077004563634655,
      "loss": 2.2201,
      "step": 134803
    },
    {
      "epoch": 1.76,
      "grad_norm": 0.9842411279678345,
      "learning_rate": 0.00022076610031553917,
      "loss": 2.1489,
      "step": 134804
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.339537262916565,
      "learning_rate": 0.0002207621550094627,
      "loss": 2.1511,
      "step": 134805
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.0572497844696045,
      "learning_rate": 0.00022075820971811808,
      "loss": 2.1424,
      "step": 134806
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.0472317934036255,
      "learning_rate": 0.00022075426444150597,
      "loss": 2.0402,
      "step": 134807
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.011853575706482,
      "learning_rate": 0.00022075031917962717,
      "loss": 1.9941,
      "step": 134808
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.1429363489151,
      "learning_rate": 0.0002207463739324824,
      "loss": 2.0831,
      "step": 134809
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.1998234987258911,
      "learning_rate": 0.00022074242870007244,
      "loss": 2.1545,
      "step": 134810
    },
    {
      "epoch": 1.76,
      "grad_norm": 0.959638237953186,
      "learning_rate": 0.0002207384834823979,
      "loss": 2.0311,
      "step": 134811
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.0044540166854858,
      "learning_rate": 0.00022073453827945954,
      "loss": 2.0291,
      "step": 134812
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.065290927886963,
      "learning_rate": 0.00022073059309125808,
      "loss": 2.1839,
      "step": 134813
    },
    {
      "epoch": 1.76,
      "grad_norm": 0.9911468625068665,
      "learning_rate": 0.00022072664791779437,
      "loss": 2.2391,
      "step": 134814
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.3542470932006836,
      "learning_rate": 0.00022072270275906908,
      "loss": 2.1621,
      "step": 134815
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.0098652839660645,
      "learning_rate": 0.000220718757615083,
      "loss": 1.997,
      "step": 134816
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.1284973621368408,
      "learning_rate": 0.0002207148124858367,
      "loss": 2.0297,
      "step": 134817
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.0837560892105103,
      "learning_rate": 0.000220710867371331,
      "loss": 1.618,
      "step": 134818
    },
    {
      "epoch": 1.76,
      "grad_norm": 0.9465940594673157,
      "learning_rate": 0.0002207069222715667,
      "loss": 2.2409,
      "step": 134819
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.0846924781799316,
      "learning_rate": 0.0002207029771865444,
      "loss": 2.2327,
      "step": 134820
    },
    {
      "epoch": 1.76,
      "grad_norm": 0.9527098536491394,
      "learning_rate": 0.00022069903211626495,
      "loss": 2.265,
      "step": 134821
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.1009467840194702,
      "learning_rate": 0.0002206950870607292,
      "loss": 2.1549,
      "step": 134822
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.0176068544387817,
      "learning_rate": 0.0002206911420199376,
      "loss": 2.0681,
      "step": 134823
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.0988820791244507,
      "learning_rate": 0.000220687196993891,
      "loss": 2.0135,
      "step": 134824
    },
    {
      "epoch": 1.76,
      "grad_norm": 0.9064575433731079,
      "learning_rate": 0.00022068325198259015,
      "loss": 2.0833,
      "step": 134825
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.0466840267181396,
      "learning_rate": 0.0002206793069860358,
      "loss": 2.1526,
      "step": 134826
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.162262201309204,
      "learning_rate": 0.0002206753620042286,
      "loss": 2.2695,
      "step": 134827
    },
    {
      "epoch": 1.76,
      "grad_norm": 0.9940540790557861,
      "learning_rate": 0.00022067141703716953,
      "loss": 2.3382,
      "step": 134828
    },
    {
      "epoch": 1.76,
      "grad_norm": 0.9513048529624939,
      "learning_rate": 0.00022066747208485896,
      "loss": 2.0067,
      "step": 134829
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.1476596593856812,
      "learning_rate": 0.00022066352714729788,
      "loss": 1.9423,
      "step": 134830
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.1253515481948853,
      "learning_rate": 0.0002206595822244869,
      "loss": 2.2588,
      "step": 134831
    },
    {
      "epoch": 1.76,
      "grad_norm": 0.9513036608695984,
      "learning_rate": 0.00022065563731642679,
      "loss": 2.0999,
      "step": 134832
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.0629030466079712,
      "learning_rate": 0.0002206516924231183,
      "loss": 1.9653,
      "step": 134833
    },
    {
      "epoch": 1.76,
      "grad_norm": 0.9926595091819763,
      "learning_rate": 0.00022064774754456236,
      "loss": 2.099,
      "step": 134834
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.0396173000335693,
      "learning_rate": 0.00022064380268075928,
      "loss": 1.9525,
      "step": 134835
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.009190559387207,
      "learning_rate": 0.00022063985783171006,
      "loss": 2.06,
      "step": 134836
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.172566294670105,
      "learning_rate": 0.00022063591299741536,
      "loss": 2.2805,
      "step": 134837
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.1249034404754639,
      "learning_rate": 0.000220631968177876,
      "loss": 2.0845,
      "step": 134838
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.0823997259140015,
      "learning_rate": 0.00022062802337309258,
      "loss": 2.1749,
      "step": 134839
    },
    {
      "epoch": 1.76,
      "grad_norm": 0.9687949419021606,
      "learning_rate": 0.00022062407858306607,
      "loss": 2.1446,
      "step": 134840
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.0384221076965332,
      "learning_rate": 0.00022062013380779694,
      "loss": 2.3571,
      "step": 134841
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.138047695159912,
      "learning_rate": 0.000220616189047286,
      "loss": 2.1062,
      "step": 134842
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.1215463876724243,
      "learning_rate": 0.00022061224430153402,
      "loss": 2.0155,
      "step": 134843
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.0129249095916748,
      "learning_rate": 0.0002206082995705417,
      "loss": 2.043,
      "step": 134844
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.0404601097106934,
      "learning_rate": 0.00022060435485430984,
      "loss": 2.1158,
      "step": 134845
    },
    {
      "epoch": 1.76,
      "grad_norm": 0.9951147437095642,
      "learning_rate": 0.00022060041015283923,
      "loss": 2.0605,
      "step": 134846
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.0135366916656494,
      "learning_rate": 0.00022059646546613033,
      "loss": 2.0833,
      "step": 134847
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.0448182821273804,
      "learning_rate": 0.00022059252079418412,
      "loss": 2.2212,
      "step": 134848
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.0806732177734375,
      "learning_rate": 0.00022058857613700122,
      "loss": 2.0793,
      "step": 134849
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.0077773332595825,
      "learning_rate": 0.0002205846314945824,
      "loss": 2.1698,
      "step": 134850
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.0996838808059692,
      "learning_rate": 0.0002205806868669284,
      "loss": 2.1359,
      "step": 134851
    },
    {
      "epoch": 1.76,
      "grad_norm": 0.9998838901519775,
      "learning_rate": 0.00022057674225404,
      "loss": 2.3295,
      "step": 134852
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.1726738214492798,
      "learning_rate": 0.00022057279765591792,
      "loss": 2.0793,
      "step": 134853
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.1231876611709595,
      "learning_rate": 0.0002205688530725628,
      "loss": 2.1996,
      "step": 134854
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.0541054010391235,
      "learning_rate": 0.0002205649085039754,
      "loss": 2.1708,
      "step": 134855
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.0102814435958862,
      "learning_rate": 0.00022056096395015647,
      "loss": 2.1735,
      "step": 134856
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.0390206575393677,
      "learning_rate": 0.00022055701941110678,
      "loss": 2.0336,
      "step": 134857
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.1076339483261108,
      "learning_rate": 0.00022055307488682703,
      "loss": 2.12,
      "step": 134858
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.0942065715789795,
      "learning_rate": 0.000220549130377318,
      "loss": 2.1013,
      "step": 134859
    },
    {
      "epoch": 1.76,
      "grad_norm": 0.9647136926651001,
      "learning_rate": 0.00022054518588258038,
      "loss": 2.0025,
      "step": 134860
    },
    {
      "epoch": 1.76,
      "grad_norm": 0.9713321924209595,
      "learning_rate": 0.00022054124140261495,
      "loss": 2.136,
      "step": 134861
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.0229194164276123,
      "learning_rate": 0.0002205372969374223,
      "loss": 2.2629,
      "step": 134862
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.0160678625106812,
      "learning_rate": 0.00022053335248700334,
      "loss": 2.071,
      "step": 134863
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.1820181608200073,
      "learning_rate": 0.0002205294080513587,
      "loss": 2.1592,
      "step": 134864
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.0214042663574219,
      "learning_rate": 0.00022052546363048918,
      "loss": 2.1179,
      "step": 134865
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.02903151512146,
      "learning_rate": 0.00022052151922439544,
      "loss": 2.2819,
      "step": 134866
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.0377506017684937,
      "learning_rate": 0.00022051757483307824,
      "loss": 2.2112,
      "step": 134867
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.047623872756958,
      "learning_rate": 0.00022051363045653838,
      "loss": 2.0926,
      "step": 134868
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.025814175605774,
      "learning_rate": 0.00022050968609477645,
      "loss": 2.0714,
      "step": 134869
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.0525585412979126,
      "learning_rate": 0.00022050574174779333,
      "loss": 2.1393,
      "step": 134870
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.0037438869476318,
      "learning_rate": 0.0002205017974155897,
      "loss": 2.018,
      "step": 134871
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.3236194849014282,
      "learning_rate": 0.0002204978530981663,
      "loss": 2.3294,
      "step": 134872
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.0431052446365356,
      "learning_rate": 0.0002204939087955238,
      "loss": 2.442,
      "step": 134873
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.0932608842849731,
      "learning_rate": 0.00022048996450766293,
      "loss": 2.1123,
      "step": 134874
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.1358633041381836,
      "learning_rate": 0.0002204860202345846,
      "loss": 2.1452,
      "step": 134875
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.1485933065414429,
      "learning_rate": 0.00022048207597628934,
      "loss": 2.0635,
      "step": 134876
    },
    {
      "epoch": 1.76,
      "grad_norm": 0.9390608072280884,
      "learning_rate": 0.00022047813173277809,
      "loss": 1.8989,
      "step": 134877
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.114936351776123,
      "learning_rate": 0.00022047418750405134,
      "loss": 1.9318,
      "step": 134878
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.0276062488555908,
      "learning_rate": 0.00022047024329010994,
      "loss": 2.4152,
      "step": 134879
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.3240342140197754,
      "learning_rate": 0.00022046629909095459,
      "loss": 1.763,
      "step": 134880
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.028780221939087,
      "learning_rate": 0.00022046235490658607,
      "loss": 2.0178,
      "step": 134881
    },
    {
      "epoch": 1.76,
      "grad_norm": 0.981122612953186,
      "learning_rate": 0.0002204584107370052,
      "loss": 2.181,
      "step": 134882
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.0557254552841187,
      "learning_rate": 0.00022045446658221256,
      "loss": 2.0705,
      "step": 134883
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.2325698137283325,
      "learning_rate": 0.00022045052244220895,
      "loss": 2.1938,
      "step": 134884
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.5891705751419067,
      "learning_rate": 0.00022044657831699505,
      "loss": 1.9867,
      "step": 134885
    },
    {
      "epoch": 1.76,
      "grad_norm": 0.947158932685852,
      "learning_rate": 0.00022044263420657162,
      "loss": 2.1145,
      "step": 134886
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.3766134977340698,
      "learning_rate": 0.0002204386901109394,
      "loss": 2.2041,
      "step": 134887
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.55059814453125,
      "learning_rate": 0.00022043474603009914,
      "loss": 2.0303,
      "step": 134888
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.1121015548706055,
      "learning_rate": 0.00022043080196405173,
      "loss": 2.2637,
      "step": 134889
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.0058006048202515,
      "learning_rate": 0.00022042685791279757,
      "loss": 2.0083,
      "step": 134890
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.330020546913147,
      "learning_rate": 0.00022042291387633757,
      "loss": 2.0162,
      "step": 134891
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.131829023361206,
      "learning_rate": 0.00022041896985467243,
      "loss": 2.1842,
      "step": 134892
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.1845263242721558,
      "learning_rate": 0.0002204150258478029,
      "loss": 2.0747,
      "step": 134893
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.0617903470993042,
      "learning_rate": 0.00022041108185572974,
      "loss": 2.1518,
      "step": 134894
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.0491808652877808,
      "learning_rate": 0.0002204071378784538,
      "loss": 2.0695,
      "step": 134895
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.171850323677063,
      "learning_rate": 0.00022040319391597552,
      "loss": 1.9769,
      "step": 134896
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.2312698364257812,
      "learning_rate": 0.00022039924996829582,
      "loss": 1.867,
      "step": 134897
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.042589783668518,
      "learning_rate": 0.0002203953060354154,
      "loss": 2.3505,
      "step": 134898
    },
    {
      "epoch": 1.76,
      "grad_norm": 0.9928900003433228,
      "learning_rate": 0.000220391362117335,
      "loss": 2.1233,
      "step": 134899
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.0982470512390137,
      "learning_rate": 0.00022038741821405532,
      "loss": 2.2278,
      "step": 134900
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.0749454498291016,
      "learning_rate": 0.00022038347432557726,
      "loss": 2.2292,
      "step": 134901
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.1241453886032104,
      "learning_rate": 0.00022037953045190127,
      "loss": 2.0231,
      "step": 134902
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.0714954137802124,
      "learning_rate": 0.00022037558659302825,
      "loss": 1.9646,
      "step": 134903
    },
    {
      "epoch": 1.76,
      "grad_norm": 0.9838144183158875,
      "learning_rate": 0.0002203716427489589,
      "loss": 1.8728,
      "step": 134904
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.049946665763855,
      "learning_rate": 0.000220367698919694,
      "loss": 2.0366,
      "step": 134905
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.0215591192245483,
      "learning_rate": 0.00022036375510523418,
      "loss": 1.976,
      "step": 134906
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.017327070236206,
      "learning_rate": 0.00022035981130558044,
      "loss": 1.9186,
      "step": 134907
    },
    {
      "epoch": 1.76,
      "grad_norm": 0.9912768006324768,
      "learning_rate": 0.00022035586752073313,
      "loss": 2.2518,
      "step": 134908
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.0293221473693848,
      "learning_rate": 0.0002203519237506932,
      "loss": 2.0125,
      "step": 134909
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.0668836832046509,
      "learning_rate": 0.00022034797999546135,
      "loss": 1.9167,
      "step": 134910
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.0658105611801147,
      "learning_rate": 0.00022034403625503828,
      "loss": 2.0889,
      "step": 134911
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.0449713468551636,
      "learning_rate": 0.00022034009252942482,
      "loss": 2.244,
      "step": 134912
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.0024784803390503,
      "learning_rate": 0.00022033614881862172,
      "loss": 2.0785,
      "step": 134913
    },
    {
      "epoch": 1.76,
      "grad_norm": 0.99868243932724,
      "learning_rate": 0.00022033220512262952,
      "loss": 2.0682,
      "step": 134914
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.001478672027588,
      "learning_rate": 0.00022032826144144908,
      "loss": 1.892,
      "step": 134915
    },
    {
      "epoch": 1.76,
      "grad_norm": 0.9716005325317383,
      "learning_rate": 0.0002203243177750811,
      "loss": 2.1347,
      "step": 134916
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.1122084856033325,
      "learning_rate": 0.00022032037412352633,
      "loss": 1.9296,
      "step": 134917
    },
    {
      "epoch": 1.76,
      "grad_norm": 0.9959254860877991,
      "learning_rate": 0.00022031643048678556,
      "loss": 2.1057,
      "step": 134918
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.0967025756835938,
      "learning_rate": 0.0002203124868648594,
      "loss": 2.0518,
      "step": 134919
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.3060930967330933,
      "learning_rate": 0.0002203085432577488,
      "loss": 2.0581,
      "step": 134920
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.0019886493682861,
      "learning_rate": 0.00022030459966545423,
      "loss": 1.9411,
      "step": 134921
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.0772390365600586,
      "learning_rate": 0.00022030065608797655,
      "loss": 2.5579,
      "step": 134922
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.0850108861923218,
      "learning_rate": 0.00022029671252531645,
      "loss": 1.9202,
      "step": 134923
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.199875831604004,
      "learning_rate": 0.0002202927689774747,
      "loss": 2.105,
      "step": 134924
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.2550466060638428,
      "learning_rate": 0.00022028882544445206,
      "loss": 2.3563,
      "step": 134925
    },
    {
      "epoch": 1.76,
      "grad_norm": 0.9321634769439697,
      "learning_rate": 0.00022028488192624935,
      "loss": 2.169,
      "step": 134926
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.1532658338546753,
      "learning_rate": 0.00022028093842286705,
      "loss": 2.0873,
      "step": 134927
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.0572841167449951,
      "learning_rate": 0.00022027699493430602,
      "loss": 2.0854,
      "step": 134928
    },
    {
      "epoch": 1.76,
      "grad_norm": 0.9472895860671997,
      "learning_rate": 0.000220273051460567,
      "loss": 2.0119,
      "step": 134929
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.09764564037323,
      "learning_rate": 0.00022026910800165071,
      "loss": 2.1553,
      "step": 134930
    },
    {
      "epoch": 1.76,
      "grad_norm": 0.9687442183494568,
      "learning_rate": 0.00022026516455755796,
      "loss": 2.2185,
      "step": 134931
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.0852880477905273,
      "learning_rate": 0.00022026122112828953,
      "loss": 1.8541,
      "step": 134932
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.014590859413147,
      "learning_rate": 0.00022025727771384586,
      "loss": 2.1031,
      "step": 134933
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.0850534439086914,
      "learning_rate": 0.0002202533343142279,
      "loss": 1.9618,
      "step": 134934
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.0876272916793823,
      "learning_rate": 0.0002202493909294364,
      "loss": 2.153,
      "step": 134935
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.06158447265625,
      "learning_rate": 0.00022024544755947195,
      "loss": 2.1916,
      "step": 134936
    },
    {
      "epoch": 1.76,
      "grad_norm": 0.9570552706718445,
      "learning_rate": 0.00022024150420433546,
      "loss": 2.1083,
      "step": 134937
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.112771987915039,
      "learning_rate": 0.00022023756086402764,
      "loss": 2.1774,
      "step": 134938
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.1074126958847046,
      "learning_rate": 0.0002202336175385491,
      "loss": 2.1223,
      "step": 134939
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.0899628400802612,
      "learning_rate": 0.0002202296742279006,
      "loss": 2.2737,
      "step": 134940
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.094040870666504,
      "learning_rate": 0.0002202257309320829,
      "loss": 2.1782,
      "step": 134941
    },
    {
      "epoch": 1.76,
      "grad_norm": 0.9865453243255615,
      "learning_rate": 0.00022022178765109676,
      "loss": 2.1442,
      "step": 134942
    },
    {
      "epoch": 1.76,
      "grad_norm": 0.97450190782547,
      "learning_rate": 0.0002202178443849429,
      "loss": 2.0931,
      "step": 134943
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.0966895818710327,
      "learning_rate": 0.0002202139011336221,
      "loss": 2.0321,
      "step": 134944
    },
    {
      "epoch": 1.76,
      "grad_norm": 0.9986945390701294,
      "learning_rate": 0.00022020995789713506,
      "loss": 2.0543,
      "step": 134945
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.0248249769210815,
      "learning_rate": 0.00022020601467548237,
      "loss": 2.2085,
      "step": 134946
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.2212165594100952,
      "learning_rate": 0.000220202071468665,
      "loss": 2.2799,
      "step": 134947
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.4081625938415527,
      "learning_rate": 0.00022019812827668347,
      "loss": 1.9823,
      "step": 134948
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.0461186170578003,
      "learning_rate": 0.00022019418509953864,
      "loss": 1.8669,
      "step": 134949
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.0227570533752441,
      "learning_rate": 0.00022019024193723128,
      "loss": 1.9999,
      "step": 134950
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.1125596761703491,
      "learning_rate": 0.000220186298789762,
      "loss": 2.2453,
      "step": 134951
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.0340873003005981,
      "learning_rate": 0.00022018235565713168,
      "loss": 2.1275,
      "step": 134952
    },
    {
      "epoch": 1.76,
      "grad_norm": 0.9272511005401611,
      "learning_rate": 0.00022017841253934092,
      "loss": 2.328,
      "step": 134953
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.012705683708191,
      "learning_rate": 0.00022017446943639043,
      "loss": 2.0909,
      "step": 134954
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.0252848863601685,
      "learning_rate": 0.00022017052634828107,
      "loss": 2.432,
      "step": 134955
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.0016357898712158,
      "learning_rate": 0.00022016658327501357,
      "loss": 2.2188,
      "step": 134956
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.1569256782531738,
      "learning_rate": 0.00022016264021658855,
      "loss": 2.1482,
      "step": 134957
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.1658445596694946,
      "learning_rate": 0.00022015869717300677,
      "loss": 2.0781,
      "step": 134958
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.052674412727356,
      "learning_rate": 0.00022015475414426905,
      "loss": 2.172,
      "step": 134959
    },
    {
      "epoch": 1.76,
      "grad_norm": 2.21549391746521,
      "learning_rate": 0.00022015081113037604,
      "loss": 2.1322,
      "step": 134960
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.076229214668274,
      "learning_rate": 0.00022014686813132852,
      "loss": 2.0069,
      "step": 134961
    },
    {
      "epoch": 1.76,
      "grad_norm": 0.9442940950393677,
      "learning_rate": 0.00022014292514712725,
      "loss": 2.0653,
      "step": 134962
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.1208328008651733,
      "learning_rate": 0.00022013898217777283,
      "loss": 2.1706,
      "step": 134963
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.1368770599365234,
      "learning_rate": 0.00022013503922326612,
      "loss": 2.0899,
      "step": 134964
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.0840637683868408,
      "learning_rate": 0.00022013109628360778,
      "loss": 2.163,
      "step": 134965
    },
    {
      "epoch": 1.76,
      "grad_norm": 0.9956768751144409,
      "learning_rate": 0.0002201271533587986,
      "loss": 1.9089,
      "step": 134966
    },
    {
      "epoch": 1.76,
      "grad_norm": 0.9923836588859558,
      "learning_rate": 0.00022012321044883934,
      "loss": 2.1414,
      "step": 134967
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.0337148904800415,
      "learning_rate": 0.00022011926755373067,
      "loss": 2.0417,
      "step": 134968
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.115207552909851,
      "learning_rate": 0.0002201153246734733,
      "loss": 2.167,
      "step": 134969
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.1078490018844604,
      "learning_rate": 0.00022011138180806796,
      "loss": 1.9876,
      "step": 134970
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.0757843255996704,
      "learning_rate": 0.00022010743895751548,
      "loss": 2.0779,
      "step": 134971
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.0542302131652832,
      "learning_rate": 0.0002201034961218165,
      "loss": 1.9455,
      "step": 134972
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.122200846672058,
      "learning_rate": 0.0002200995533009718,
      "loss": 2.0414,
      "step": 134973
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.0373570919036865,
      "learning_rate": 0.0002200956104949822,
      "loss": 1.9048,
      "step": 134974
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.0228307247161865,
      "learning_rate": 0.00022009166770384822,
      "loss": 1.8653,
      "step": 134975
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.0544087886810303,
      "learning_rate": 0.00022008772492757074,
      "loss": 2.1603,
      "step": 134976
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.099513292312622,
      "learning_rate": 0.0002200837821661504,
      "loss": 2.3393,
      "step": 134977
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.02444589138031,
      "learning_rate": 0.00022007983941958802,
      "loss": 2.0735,
      "step": 134978
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.0997347831726074,
      "learning_rate": 0.00022007589668788434,
      "loss": 2.1286,
      "step": 134979
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.1487964391708374,
      "learning_rate": 0.00022007195397104017,
      "loss": 2.1982,
      "step": 134980
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.0173734426498413,
      "learning_rate": 0.000220068011269056,
      "loss": 2.0456,
      "step": 134981
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.0587092638015747,
      "learning_rate": 0.00022006406858193268,
      "loss": 1.9876,
      "step": 134982
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.2848459482192993,
      "learning_rate": 0.00022006012590967099,
      "loss": 1.8436,
      "step": 134983
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.1710205078125,
      "learning_rate": 0.0002200561832522716,
      "loss": 2.2617,
      "step": 134984
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.1594643592834473,
      "learning_rate": 0.00022005224060973527,
      "loss": 2.2538,
      "step": 134985
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.1875001192092896,
      "learning_rate": 0.00022004829798206278,
      "loss": 1.9797,
      "step": 134986
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.3549469709396362,
      "learning_rate": 0.0002200443553692549,
      "loss": 2.2317,
      "step": 134987
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.081404685974121,
      "learning_rate": 0.00022004041277131218,
      "loss": 1.9936,
      "step": 134988
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.0140115022659302,
      "learning_rate": 0.00022003647018823544,
      "loss": 2.0543,
      "step": 134989
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.0334386825561523,
      "learning_rate": 0.00022003252762002547,
      "loss": 1.9943,
      "step": 134990
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.1801738739013672,
      "learning_rate": 0.00022002858506668294,
      "loss": 1.9611,
      "step": 134991
    },
    {
      "epoch": 1.76,
      "grad_norm": 0.9725838303565979,
      "learning_rate": 0.0002200246425282086,
      "loss": 2.1557,
      "step": 134992
    },
    {
      "epoch": 1.76,
      "grad_norm": 0.9852955937385559,
      "learning_rate": 0.00022002070000460335,
      "loss": 2.2914,
      "step": 134993
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.0068995952606201,
      "learning_rate": 0.0002200167574958676,
      "loss": 2.0979,
      "step": 134994
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.0730797052383423,
      "learning_rate": 0.00022001281500200223,
      "loss": 2.1182,
      "step": 134995
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.0420163869857788,
      "learning_rate": 0.000220008872523008,
      "loss": 2.1545,
      "step": 134996
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.0514130592346191,
      "learning_rate": 0.00022000493005888564,
      "loss": 2.0594,
      "step": 134997
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.0380527973175049,
      "learning_rate": 0.00022000098760963587,
      "loss": 2.166,
      "step": 134998
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.0021653175354004,
      "learning_rate": 0.00021999704517525954,
      "loss": 2.3477,
      "step": 134999
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.1866307258605957,
      "learning_rate": 0.0002199931027557572,
      "loss": 1.939,
      "step": 135000
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.1871898174285889,
      "learning_rate": 0.00021998916035112956,
      "loss": 2.1856,
      "step": 135001
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.057238221168518,
      "learning_rate": 0.0002199852179613775,
      "loss": 2.0937,
      "step": 135002
    },
    {
      "epoch": 1.76,
      "grad_norm": 0.963939368724823,
      "learning_rate": 0.0002199812755865017,
      "loss": 1.9775,
      "step": 135003
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.098753809928894,
      "learning_rate": 0.0002199773332265029,
      "loss": 2.1145,
      "step": 135004
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.0104422569274902,
      "learning_rate": 0.0002199733908813819,
      "loss": 2.0539,
      "step": 135005
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.0396678447723389,
      "learning_rate": 0.0002199694485511393,
      "loss": 2.0294,
      "step": 135006
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.0350850820541382,
      "learning_rate": 0.00021996550623577585,
      "loss": 2.071,
      "step": 135007
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.1039857864379883,
      "learning_rate": 0.00021996156393529233,
      "loss": 2.2046,
      "step": 135008
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.2199997901916504,
      "learning_rate": 0.00021995762164968944,
      "loss": 2.1912,
      "step": 135009
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.1005340814590454,
      "learning_rate": 0.00021995367937896799,
      "loss": 2.0829,
      "step": 135010
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.4751482009887695,
      "learning_rate": 0.00021994973712312872,
      "loss": 2.0342,
      "step": 135011
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.2970490455627441,
      "learning_rate": 0.00021994579488217225,
      "loss": 2.0828,
      "step": 135012
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.0141178369522095,
      "learning_rate": 0.00021994185265609935,
      "loss": 2.1177,
      "step": 135013
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.1008776426315308,
      "learning_rate": 0.00021993791044491075,
      "loss": 1.9825,
      "step": 135014
    },
    {
      "epoch": 1.76,
      "grad_norm": 0.9979721903800964,
      "learning_rate": 0.00021993396824860722,
      "loss": 2.0098,
      "step": 135015
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.181746006011963,
      "learning_rate": 0.00021993002606718947,
      "loss": 2.1029,
      "step": 135016
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.1179289817810059,
      "learning_rate": 0.00021992608390065835,
      "loss": 2.2971,
      "step": 135017
    },
    {
      "epoch": 1.76,
      "grad_norm": 0.9680007696151733,
      "learning_rate": 0.00021992214174901438,
      "loss": 2.0994,
      "step": 135018
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.0425033569335938,
      "learning_rate": 0.00021991819961225838,
      "loss": 2.2715,
      "step": 135019
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.2366315126419067,
      "learning_rate": 0.0002199142574903911,
      "loss": 2.043,
      "step": 135020
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.2080827951431274,
      "learning_rate": 0.0002199103153834133,
      "loss": 1.9537,
      "step": 135021
    },
    {
      "epoch": 1.76,
      "grad_norm": 0.9969677329063416,
      "learning_rate": 0.00021990637329132563,
      "loss": 2.013,
      "step": 135022
    },
    {
      "epoch": 1.76,
      "grad_norm": 0.9818477630615234,
      "learning_rate": 0.00021990243121412906,
      "loss": 2.0942,
      "step": 135023
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.0454331636428833,
      "learning_rate": 0.00021989848915182397,
      "loss": 1.9569,
      "step": 135024
    },
    {
      "epoch": 1.76,
      "grad_norm": 0.9585739970207214,
      "learning_rate": 0.00021989454710441127,
      "loss": 1.9997,
      "step": 135025
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.0196256637573242,
      "learning_rate": 0.00021989060507189174,
      "loss": 2.1615,
      "step": 135026
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.0609205961227417,
      "learning_rate": 0.000219886663054266,
      "loss": 2.1595,
      "step": 135027
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.0772067308425903,
      "learning_rate": 0.00021988272105153482,
      "loss": 1.9199,
      "step": 135028
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.148519515991211,
      "learning_rate": 0.0002198787790636991,
      "loss": 2.0951,
      "step": 135029
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.1008766889572144,
      "learning_rate": 0.00021987483709075934,
      "loss": 2.1422,
      "step": 135030
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.1934324502944946,
      "learning_rate": 0.00021987089513271637,
      "loss": 2.144,
      "step": 135031
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.004459023475647,
      "learning_rate": 0.00021986695318957085,
      "loss": 2.0917,
      "step": 135032
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.2258918285369873,
      "learning_rate": 0.00021986301126132365,
      "loss": 2.3164,
      "step": 135033
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.0659277439117432,
      "learning_rate": 0.00021985906934797537,
      "loss": 2.0487,
      "step": 135034
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.0479342937469482,
      "learning_rate": 0.00021985512744952693,
      "loss": 2.0979,
      "step": 135035
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.047229528427124,
      "learning_rate": 0.0002198511855659788,
      "loss": 1.8415,
      "step": 135036
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.0026434659957886,
      "learning_rate": 0.0002198472436973319,
      "loss": 2.2,
      "step": 135037
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.029476523399353,
      "learning_rate": 0.00021984330184358688,
      "loss": 2.2844,
      "step": 135038
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.1029860973358154,
      "learning_rate": 0.00021983936000474451,
      "loss": 2.1034,
      "step": 135039
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.061617136001587,
      "learning_rate": 0.00021983541818080553,
      "loss": 2.0899,
      "step": 135040
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.0261014699935913,
      "learning_rate": 0.00021983147637177073,
      "loss": 1.972,
      "step": 135041
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.1217478513717651,
      "learning_rate": 0.00021982753457764064,
      "loss": 2.1719,
      "step": 135042
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.062050700187683,
      "learning_rate": 0.0002198235927984162,
      "loss": 2.1844,
      "step": 135043
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.0049625635147095,
      "learning_rate": 0.00021981965103409808,
      "loss": 2.1797,
      "step": 135044
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.1919476985931396,
      "learning_rate": 0.00021981570928468697,
      "loss": 2.2132,
      "step": 135045
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.0179433822631836,
      "learning_rate": 0.0002198117675501836,
      "loss": 1.9771,
      "step": 135046
    },
    {
      "epoch": 1.76,
      "grad_norm": 0.940864622592926,
      "learning_rate": 0.00021980782583058881,
      "loss": 2.2469,
      "step": 135047
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.1004000902175903,
      "learning_rate": 0.0002198038841259032,
      "loss": 1.9678,
      "step": 135048
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.0487067699432373,
      "learning_rate": 0.00021979994243612755,
      "loss": 2.0637,
      "step": 135049
    },
    {
      "epoch": 1.76,
      "grad_norm": 0.9685434103012085,
      "learning_rate": 0.00021979600076126266,
      "loss": 2.1718,
      "step": 135050
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.0470598936080933,
      "learning_rate": 0.00021979205910130915,
      "loss": 2.0575,
      "step": 135051
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.0106016397476196,
      "learning_rate": 0.0002197881174562679,
      "loss": 2.1795,
      "step": 135052
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.0482484102249146,
      "learning_rate": 0.00021978417582613945,
      "loss": 1.937,
      "step": 135053
    },
    {
      "epoch": 1.76,
      "grad_norm": 0.9963467717170715,
      "learning_rate": 0.00021978023421092476,
      "loss": 2.138,
      "step": 135054
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.0649100542068481,
      "learning_rate": 0.00021977629261062432,
      "loss": 2.1621,
      "step": 135055
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.1528403759002686,
      "learning_rate": 0.000219772351025239,
      "loss": 2.1445,
      "step": 135056
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.1585813760757446,
      "learning_rate": 0.00021976840945476953,
      "loss": 2.0303,
      "step": 135057
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.020687460899353,
      "learning_rate": 0.0002197644678992166,
      "loss": 2.1751,
      "step": 135058
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.0166465044021606,
      "learning_rate": 0.00021976052635858107,
      "loss": 1.9783,
      "step": 135059
    },
    {
      "epoch": 1.76,
      "grad_norm": 0.8918957710266113,
      "learning_rate": 0.00021975658483286358,
      "loss": 1.9045,
      "step": 135060
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.160067081451416,
      "learning_rate": 0.00021975264332206473,
      "loss": 2.0547,
      "step": 135061
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.012006402015686,
      "learning_rate": 0.00021974870182618544,
      "loss": 2.0421,
      "step": 135062
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.0725452899932861,
      "learning_rate": 0.00021974476034522636,
      "loss": 2.1339,
      "step": 135063
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.004465103149414,
      "learning_rate": 0.0002197408188791882,
      "loss": 2.06,
      "step": 135064
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.1023259162902832,
      "learning_rate": 0.0002197368774280718,
      "loss": 2.0861,
      "step": 135065
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.197511911392212,
      "learning_rate": 0.00021973293599187793,
      "loss": 2.1123,
      "step": 135066
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.056604027748108,
      "learning_rate": 0.00021972899457060709,
      "loss": 2.1011,
      "step": 135067
    },
    {
      "epoch": 1.76,
      "grad_norm": 0.984948456287384,
      "learning_rate": 0.0002197250531642602,
      "loss": 2.1764,
      "step": 135068
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.0045255422592163,
      "learning_rate": 0.00021972111177283784,
      "loss": 2.084,
      "step": 135069
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.1052663326263428,
      "learning_rate": 0.00021971717039634091,
      "loss": 2.12,
      "step": 135070
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.1059964895248413,
      "learning_rate": 0.00021971322903477006,
      "loss": 2.1107,
      "step": 135071
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.0456374883651733,
      "learning_rate": 0.00021970928768812615,
      "loss": 2.031,
      "step": 135072
    },
    {
      "epoch": 1.76,
      "grad_norm": 0.927524745464325,
      "learning_rate": 0.0002197053463564097,
      "loss": 1.9387,
      "step": 135073
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.0862008333206177,
      "learning_rate": 0.00021970140503962152,
      "loss": 1.9927,
      "step": 135074
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.055848240852356,
      "learning_rate": 0.00021969746373776234,
      "loss": 1.9549,
      "step": 135075
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.0324844121932983,
      "learning_rate": 0.00021969352245083297,
      "loss": 2.1293,
      "step": 135076
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.0154529809951782,
      "learning_rate": 0.00021968958117883405,
      "loss": 1.9976,
      "step": 135077
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.0718885660171509,
      "learning_rate": 0.00021968563992176652,
      "loss": 2.1322,
      "step": 135078
    },
    {
      "epoch": 1.76,
      "grad_norm": 0.9596154689788818,
      "learning_rate": 0.00021968169867963078,
      "loss": 2.0492,
      "step": 135079
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.0267246961593628,
      "learning_rate": 0.00021967775745242773,
      "loss": 2.0201,
      "step": 135080
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.0827765464782715,
      "learning_rate": 0.00021967381624015813,
      "loss": 2.0404,
      "step": 135081
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.0997517108917236,
      "learning_rate": 0.00021966987504282263,
      "loss": 1.9039,
      "step": 135082
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.1444393396377563,
      "learning_rate": 0.00021966593386042206,
      "loss": 2.3273,
      "step": 135083
    },
    {
      "epoch": 1.76,
      "grad_norm": 0.986032247543335,
      "learning_rate": 0.00021966199269295725,
      "loss": 2.038,
      "step": 135084
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.0018386840820312,
      "learning_rate": 0.00021965805154042865,
      "loss": 1.9001,
      "step": 135085
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.1575798988342285,
      "learning_rate": 0.00021965411040283714,
      "loss": 2.2488,
      "step": 135086
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.1568412780761719,
      "learning_rate": 0.0002196501692801834,
      "loss": 2.0671,
      "step": 135087
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.0076067447662354,
      "learning_rate": 0.00021964622817246827,
      "loss": 2.2563,
      "step": 135088
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.01542067527771,
      "learning_rate": 0.0002196422870796924,
      "loss": 1.9815,
      "step": 135089
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.1373554468154907,
      "learning_rate": 0.00021963834600185667,
      "loss": 2.0194,
      "step": 135090
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.1002393960952759,
      "learning_rate": 0.00021963440493896156,
      "loss": 2.0107,
      "step": 135091
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.1221157312393188,
      "learning_rate": 0.00021963046389100792,
      "loss": 2.07,
      "step": 135092
    },
    {
      "epoch": 1.76,
      "grad_norm": 0.994730532169342,
      "learning_rate": 0.0002196265228579965,
      "loss": 2.1452,
      "step": 135093
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.2357183694839478,
      "learning_rate": 0.00021962258183992806,
      "loss": 2.005,
      "step": 135094
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.3008705377578735,
      "learning_rate": 0.00021961864083680326,
      "loss": 2.1008,
      "step": 135095
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.096801519393921,
      "learning_rate": 0.000219614699848623,
      "loss": 2.3095,
      "step": 135096
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.044443130493164,
      "learning_rate": 0.00021961075887538782,
      "loss": 2.1835,
      "step": 135097
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.2219451665878296,
      "learning_rate": 0.00021960681791709845,
      "loss": 2.2191,
      "step": 135098
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.0819497108459473,
      "learning_rate": 0.00021960287697375572,
      "loss": 1.9648,
      "step": 135099
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.1733472347259521,
      "learning_rate": 0.00021959893604536033,
      "loss": 2.0955,
      "step": 135100
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.1808832883834839,
      "learning_rate": 0.00021959499513191297,
      "loss": 2.1882,
      "step": 135101
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.2455734014511108,
      "learning_rate": 0.0002195910542334146,
      "loss": 1.8786,
      "step": 135102
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.3118252754211426,
      "learning_rate": 0.0002195871133498656,
      "loss": 2.1166,
      "step": 135103
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.100895643234253,
      "learning_rate": 0.00021958317248126693,
      "loss": 2.0265,
      "step": 135104
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.0872743129730225,
      "learning_rate": 0.00021957923162761923,
      "loss": 2.0371,
      "step": 135105
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.0666680335998535,
      "learning_rate": 0.00021957529078892328,
      "loss": 1.9137,
      "step": 135106
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.1679397821426392,
      "learning_rate": 0.00021957134996517978,
      "loss": 2.0035,
      "step": 135107
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.1826926469802856,
      "learning_rate": 0.00021956740915638962,
      "loss": 2.2362,
      "step": 135108
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.0592005252838135,
      "learning_rate": 0.00021956346836255326,
      "loss": 2.1146,
      "step": 135109
    },
    {
      "epoch": 1.76,
      "grad_norm": 0.9176787734031677,
      "learning_rate": 0.00021955952758367163,
      "loss": 1.9723,
      "step": 135110
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.2619214057922363,
      "learning_rate": 0.00021955558681974535,
      "loss": 2.0657,
      "step": 135111
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.0041747093200684,
      "learning_rate": 0.0002195516460707752,
      "loss": 2.1473,
      "step": 135112
    },
    {
      "epoch": 1.76,
      "grad_norm": 0.997484564781189,
      "learning_rate": 0.00021954770533676196,
      "loss": 2.2007,
      "step": 135113
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.0788521766662598,
      "learning_rate": 0.0002195437646177064,
      "loss": 2.1451,
      "step": 135114
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.0661159753799438,
      "learning_rate": 0.00021953982391360904,
      "loss": 2.1608,
      "step": 135115
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.0471073389053345,
      "learning_rate": 0.00021953588322447078,
      "loss": 2.2512,
      "step": 135116
    },
    {
      "epoch": 1.76,
      "grad_norm": 0.9812217950820923,
      "learning_rate": 0.00021953194255029234,
      "loss": 2.0411,
      "step": 135117
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.0291916131973267,
      "learning_rate": 0.0002195280018910744,
      "loss": 2.0931,
      "step": 135118
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.0159565210342407,
      "learning_rate": 0.0002195240612468177,
      "loss": 2.1192,
      "step": 135119
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.1308834552764893,
      "learning_rate": 0.000219520120617523,
      "loss": 1.9413,
      "step": 135120
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.028255820274353,
      "learning_rate": 0.00021951618000319115,
      "loss": 2.2361,
      "step": 135121
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.0076566934585571,
      "learning_rate": 0.0002195122394038227,
      "loss": 2.0907,
      "step": 135122
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.0172117948532104,
      "learning_rate": 0.00021950829881941841,
      "loss": 2.1506,
      "step": 135123
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.0027343034744263,
      "learning_rate": 0.00021950435824997903,
      "loss": 2.038,
      "step": 135124
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.041939616203308,
      "learning_rate": 0.00021950041769550535,
      "loss": 1.8746,
      "step": 135125
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.0485618114471436,
      "learning_rate": 0.000219496477155998,
      "loss": 1.9888,
      "step": 135126
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.068681001663208,
      "learning_rate": 0.00021949253663145786,
      "loss": 2.0105,
      "step": 135127
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.0877548456192017,
      "learning_rate": 0.00021948859612188557,
      "loss": 2.1087,
      "step": 135128
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.1138145923614502,
      "learning_rate": 0.0002194846556272819,
      "loss": 1.9319,
      "step": 135129
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.0312248468399048,
      "learning_rate": 0.00021948071514764743,
      "loss": 2.1255,
      "step": 135130
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.1196420192718506,
      "learning_rate": 0.00021947677468298307,
      "loss": 2.0221,
      "step": 135131
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.0756419897079468,
      "learning_rate": 0.00021947283423328948,
      "loss": 2.2002,
      "step": 135132
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.0297083854675293,
      "learning_rate": 0.00021946889379856747,
      "loss": 2.0388,
      "step": 135133
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.0891036987304688,
      "learning_rate": 0.00021946495337881768,
      "loss": 2.0252,
      "step": 135134
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.0138466358184814,
      "learning_rate": 0.00021946101297404085,
      "loss": 2.3544,
      "step": 135135
    },
    {
      "epoch": 1.76,
      "grad_norm": 0.9832283854484558,
      "learning_rate": 0.0002194570725842378,
      "loss": 2.0323,
      "step": 135136
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.0483167171478271,
      "learning_rate": 0.00021945313220940918,
      "loss": 2.2953,
      "step": 135137
    },
    {
      "epoch": 1.76,
      "grad_norm": 0.9356244802474976,
      "learning_rate": 0.0002194491918495557,
      "loss": 1.9219,
      "step": 135138
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.04494047164917,
      "learning_rate": 0.00021944525150467816,
      "loss": 2.1005,
      "step": 135139
    },
    {
      "epoch": 1.76,
      "grad_norm": 0.9955441951751709,
      "learning_rate": 0.00021944131117477729,
      "loss": 1.826,
      "step": 135140
    },
    {
      "epoch": 1.76,
      "grad_norm": 0.99833744764328,
      "learning_rate": 0.00021943737085985374,
      "loss": 2.2028,
      "step": 135141
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.0841645002365112,
      "learning_rate": 0.00021943343055990834,
      "loss": 2.2535,
      "step": 135142
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.019667387008667,
      "learning_rate": 0.00021942949027494175,
      "loss": 2.1053,
      "step": 135143
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.0055075883865356,
      "learning_rate": 0.00021942555000495484,
      "loss": 1.9214,
      "step": 135144
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.1087394952774048,
      "learning_rate": 0.00021942160974994823,
      "loss": 2.0679,
      "step": 135145
    },
    {
      "epoch": 1.76,
      "grad_norm": 0.9705765247344971,
      "learning_rate": 0.00021941766950992262,
      "loss": 1.7932,
      "step": 135146
    },
    {
      "epoch": 1.76,
      "grad_norm": 0.9921678304672241,
      "learning_rate": 0.00021941372928487872,
      "loss": 1.9944,
      "step": 135147
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.1347944736480713,
      "learning_rate": 0.00021940978907481742,
      "loss": 2.1123,
      "step": 135148
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.0037492513656616,
      "learning_rate": 0.00021940584887973928,
      "loss": 2.0679,
      "step": 135149
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.1527507305145264,
      "learning_rate": 0.00021940190869964513,
      "loss": 2.1107,
      "step": 135150
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.0170015096664429,
      "learning_rate": 0.00021939796853453586,
      "loss": 1.9563,
      "step": 135151
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.137165904045105,
      "learning_rate": 0.00021939402838441185,
      "loss": 2.156,
      "step": 135152
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.0120415687561035,
      "learning_rate": 0.00021939008824927402,
      "loss": 2.2874,
      "step": 135153
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.0542051792144775,
      "learning_rate": 0.00021938614812912313,
      "loss": 2.175,
      "step": 135154
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.0350890159606934,
      "learning_rate": 0.00021938220802395988,
      "loss": 2.2952,
      "step": 135155
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.0860340595245361,
      "learning_rate": 0.00021937826793378494,
      "loss": 2.1098,
      "step": 135156
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.0217223167419434,
      "learning_rate": 0.0002193743278585993,
      "loss": 2.2391,
      "step": 135157
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.0942517518997192,
      "learning_rate": 0.00021937038779840332,
      "loss": 2.4028,
      "step": 135158
    },
    {
      "epoch": 1.76,
      "grad_norm": 0.9508332014083862,
      "learning_rate": 0.00021936644775319788,
      "loss": 2.1118,
      "step": 135159
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.0830599069595337,
      "learning_rate": 0.0002193625077229838,
      "loss": 2.2227,
      "step": 135160
    },
    {
      "epoch": 1.76,
      "grad_norm": 0.9901601672172546,
      "learning_rate": 0.00021935856770776172,
      "loss": 1.8863,
      "step": 135161
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.0172444581985474,
      "learning_rate": 0.0002193546277075324,
      "loss": 2.2432,
      "step": 135162
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.2736512422561646,
      "learning_rate": 0.00021935068772229675,
      "loss": 2.2652,
      "step": 135163
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.0292489528656006,
      "learning_rate": 0.00021934674775205516,
      "loss": 2.0141,
      "step": 135164
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.1264979839324951,
      "learning_rate": 0.00021934280779680853,
      "loss": 2.1383,
      "step": 135165
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.0055155754089355,
      "learning_rate": 0.0002193388678565576,
      "loss": 1.9254,
      "step": 135166
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.199515461921692,
      "learning_rate": 0.0002193349279313031,
      "loss": 2.1944,
      "step": 135167
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.0908514261245728,
      "learning_rate": 0.00021933098802104579,
      "loss": 2.2645,
      "step": 135168
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.159759283065796,
      "learning_rate": 0.0002193270481257865,
      "loss": 2.118,
      "step": 135169
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.036064863204956,
      "learning_rate": 0.00021932310824552563,
      "loss": 2.0694,
      "step": 135170
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.06350839138031,
      "learning_rate": 0.0002193191683802642,
      "loss": 1.8948,
      "step": 135171
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.0056664943695068,
      "learning_rate": 0.00021931522853000282,
      "loss": 2.1391,
      "step": 135172
    },
    {
      "epoch": 1.76,
      "grad_norm": 0.9864109754562378,
      "learning_rate": 0.00021931128869474227,
      "loss": 2.2524,
      "step": 135173
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.0045316219329834,
      "learning_rate": 0.00021930734887448325,
      "loss": 2.059,
      "step": 135174
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.036664366722107,
      "learning_rate": 0.00021930340906922666,
      "loss": 2.2616,
      "step": 135175
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.0966941118240356,
      "learning_rate": 0.000219299469278973,
      "loss": 2.0874,
      "step": 135176
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.1037242412567139,
      "learning_rate": 0.00021929552950372303,
      "loss": 2.1063,
      "step": 135177
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.0657610893249512,
      "learning_rate": 0.00021929158974347762,
      "loss": 2.007,
      "step": 135178
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.0022233724594116,
      "learning_rate": 0.00021928764999823737,
      "loss": 2.2741,
      "step": 135179
    },
    {
      "epoch": 1.76,
      "grad_norm": 0.9806438684463501,
      "learning_rate": 0.00021928371026800304,
      "loss": 1.9904,
      "step": 135180
    },
    {
      "epoch": 1.76,
      "grad_norm": 0.959467887878418,
      "learning_rate": 0.0002192797705527756,
      "loss": 2.0554,
      "step": 135181
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.0699448585510254,
      "learning_rate": 0.0002192758308525554,
      "loss": 2.0755,
      "step": 135182
    },
    {
      "epoch": 1.76,
      "grad_norm": 0.9933413863182068,
      "learning_rate": 0.00021927189116734334,
      "loss": 2.1257,
      "step": 135183
    },
    {
      "epoch": 1.76,
      "grad_norm": 0.9807724952697754,
      "learning_rate": 0.0002192679514971402,
      "loss": 2.0435,
      "step": 135184
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.0288312435150146,
      "learning_rate": 0.00021926401184194663,
      "loss": 2.01,
      "step": 135185
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.2174428701400757,
      "learning_rate": 0.0002192600722017634,
      "loss": 2.1021,
      "step": 135186
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.3279893398284912,
      "learning_rate": 0.00021925613257659128,
      "loss": 2.1971,
      "step": 135187
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.0021963119506836,
      "learning_rate": 0.0002192521929664311,
      "loss": 2.0557,
      "step": 135188
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.0239847898483276,
      "learning_rate": 0.00021924825337128328,
      "loss": 2.0293,
      "step": 135189
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.1200793981552124,
      "learning_rate": 0.0002192443137911488,
      "loss": 2.1694,
      "step": 135190
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.144666314125061,
      "learning_rate": 0.0002192403742260283,
      "loss": 2.1952,
      "step": 135191
    },
    {
      "epoch": 1.76,
      "grad_norm": 0.9648647308349609,
      "learning_rate": 0.00021923643467592255,
      "loss": 2.0104,
      "step": 135192
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.0625858306884766,
      "learning_rate": 0.00021923249514083223,
      "loss": 2.0081,
      "step": 135193
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.137966513633728,
      "learning_rate": 0.00021922855562075827,
      "loss": 1.9513,
      "step": 135194
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.1996057033538818,
      "learning_rate": 0.00021922461611570114,
      "loss": 1.8589,
      "step": 135195
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.0592153072357178,
      "learning_rate": 0.00021922067662566165,
      "loss": 2.0936,
      "step": 135196
    },
    {
      "epoch": 1.76,
      "grad_norm": 0.9944142699241638,
      "learning_rate": 0.00021921673715064057,
      "loss": 1.9764,
      "step": 135197
    },
    {
      "epoch": 1.76,
      "grad_norm": 0.9844305515289307,
      "learning_rate": 0.00021921279769063863,
      "loss": 2.1704,
      "step": 135198
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.101063847541809,
      "learning_rate": 0.00021920885824565654,
      "loss": 2.0749,
      "step": 135199
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.0969809293746948,
      "learning_rate": 0.00021920491881569518,
      "loss": 2.0073,
      "step": 135200
    },
    {
      "epoch": 1.76,
      "grad_norm": 0.996356189250946,
      "learning_rate": 0.00021920097940075504,
      "loss": 2.2089,
      "step": 135201
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.0699888467788696,
      "learning_rate": 0.00021919704000083695,
      "loss": 2.1834,
      "step": 135202
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.2050946950912476,
      "learning_rate": 0.00021919310061594162,
      "loss": 2.1612,
      "step": 135203
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.0233964920043945,
      "learning_rate": 0.0002191891612460699,
      "loss": 2.0952,
      "step": 135204
    },
    {
      "epoch": 1.76,
      "grad_norm": 0.9570413827896118,
      "learning_rate": 0.00021918522189122238,
      "loss": 2.0577,
      "step": 135205
    },
    {
      "epoch": 1.76,
      "grad_norm": 0.9599152207374573,
      "learning_rate": 0.00021918128255139996,
      "loss": 2.1976,
      "step": 135206
    },
    {
      "epoch": 1.76,
      "grad_norm": 0.9691198468208313,
      "learning_rate": 0.0002191773432266032,
      "loss": 2.1908,
      "step": 135207
    },
    {
      "epoch": 1.76,
      "grad_norm": 0.963988721370697,
      "learning_rate": 0.00021917340391683287,
      "loss": 1.9631,
      "step": 135208
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.0877140760421753,
      "learning_rate": 0.00021916946462208973,
      "loss": 2.276,
      "step": 135209
    },
    {
      "epoch": 1.76,
      "grad_norm": 0.9691182971000671,
      "learning_rate": 0.00021916552534237453,
      "loss": 2.3599,
      "step": 135210
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.0544625520706177,
      "learning_rate": 0.00021916158607768795,
      "loss": 1.9303,
      "step": 135211
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.0634849071502686,
      "learning_rate": 0.0002191576468280308,
      "loss": 1.9675,
      "step": 135212
    },
    {
      "epoch": 1.76,
      "grad_norm": 0.9779930710792542,
      "learning_rate": 0.00021915370759340377,
      "loss": 1.9919,
      "step": 135213
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.05995512008667,
      "learning_rate": 0.00021914976837380764,
      "loss": 2.064,
      "step": 135214
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.0480296611785889,
      "learning_rate": 0.000219145829169243,
      "loss": 2.124,
      "step": 135215
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.1158312559127808,
      "learning_rate": 0.0002191418899797107,
      "loss": 2.0195,
      "step": 135216
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.047351360321045,
      "learning_rate": 0.00021913795080521145,
      "loss": 2.042,
      "step": 135217
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.0625308752059937,
      "learning_rate": 0.00021913401164574603,
      "loss": 2.1211,
      "step": 135218
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.3400980234146118,
      "learning_rate": 0.0002191300725013151,
      "loss": 1.9285,
      "step": 135219
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.1191127300262451,
      "learning_rate": 0.00021912613337191937,
      "loss": 2.1436,
      "step": 135220
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.0375126600265503,
      "learning_rate": 0.00021912219425755967,
      "loss": 2.1564,
      "step": 135221
    },
    {
      "epoch": 1.76,
      "grad_norm": 0.9753044247627258,
      "learning_rate": 0.00021911825515823666,
      "loss": 2.0725,
      "step": 135222
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.0742368698120117,
      "learning_rate": 0.00021911431607395107,
      "loss": 2.0616,
      "step": 135223
    },
    {
      "epoch": 1.76,
      "grad_norm": 0.9665093421936035,
      "learning_rate": 0.00021911037700470375,
      "loss": 2.2416,
      "step": 135224
    },
    {
      "epoch": 1.76,
      "grad_norm": 0.9733477830886841,
      "learning_rate": 0.00021910643795049523,
      "loss": 2.3021,
      "step": 135225
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.0932400226593018,
      "learning_rate": 0.00021910249891132636,
      "loss": 2.1722,
      "step": 135226
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.0183218717575073,
      "learning_rate": 0.0002190985598871979,
      "loss": 2.1376,
      "step": 135227
    },
    {
      "epoch": 1.76,
      "grad_norm": 0.9690365791320801,
      "learning_rate": 0.0002190946208781106,
      "loss": 2.2004,
      "step": 135228
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.1482120752334595,
      "learning_rate": 0.00021909068188406505,
      "loss": 2.0277,
      "step": 135229
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.095950961112976,
      "learning_rate": 0.0002190867429050621,
      "loss": 2.3886,
      "step": 135230
    },
    {
      "epoch": 1.76,
      "grad_norm": 0.9829879403114319,
      "learning_rate": 0.00021908280394110244,
      "loss": 2.0005,
      "step": 135231
    },
    {
      "epoch": 1.76,
      "grad_norm": 0.9569792151451111,
      "learning_rate": 0.0002190788649921868,
      "loss": 2.1323,
      "step": 135232
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.0449354648590088,
      "learning_rate": 0.00021907492605831592,
      "loss": 2.1618,
      "step": 135233
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.0469619035720825,
      "learning_rate": 0.00021907098713949054,
      "loss": 2.1233,
      "step": 135234
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.0365970134735107,
      "learning_rate": 0.00021906704823571147,
      "loss": 2.3786,
      "step": 135235
    },
    {
      "epoch": 1.76,
      "grad_norm": 0.9604159593582153,
      "learning_rate": 0.00021906310934697936,
      "loss": 2.0238,
      "step": 135236
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.068267583847046,
      "learning_rate": 0.00021905917047329487,
      "loss": 1.8392,
      "step": 135237
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.1122796535491943,
      "learning_rate": 0.00021905523161465883,
      "loss": 2.0645,
      "step": 135238
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.0316691398620605,
      "learning_rate": 0.00021905129277107193,
      "loss": 2.1088,
      "step": 135239
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.0225480794906616,
      "learning_rate": 0.00021904735394253494,
      "loss": 2.0012,
      "step": 135240
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.0978044271469116,
      "learning_rate": 0.00021904341512904852,
      "loss": 2.2959,
      "step": 135241
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.103186845779419,
      "learning_rate": 0.00021903947633061365,
      "loss": 2.1606,
      "step": 135242
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.0454996824264526,
      "learning_rate": 0.00021903553754723073,
      "loss": 2.0195,
      "step": 135243
    },
    {
      "epoch": 1.76,
      "grad_norm": 0.9761583805084229,
      "learning_rate": 0.00021903159877890063,
      "loss": 2.1085,
      "step": 135244
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.0681633949279785,
      "learning_rate": 0.00021902766002562403,
      "loss": 2.2081,
      "step": 135245
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.0221617221832275,
      "learning_rate": 0.0002190237212874018,
      "loss": 2.0311,
      "step": 135246
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.0244355201721191,
      "learning_rate": 0.00021901978256423454,
      "loss": 1.9131,
      "step": 135247
    },
    {
      "epoch": 1.76,
      "grad_norm": 0.9637787342071533,
      "learning_rate": 0.0002190158438561232,
      "loss": 2.2271,
      "step": 135248
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.069903016090393,
      "learning_rate": 0.00021901190516306814,
      "loss": 1.9549,
      "step": 135249
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.0201597213745117,
      "learning_rate": 0.00021900796648507037,
      "loss": 2.1737,
      "step": 135250
    },
    {
      "epoch": 1.76,
      "grad_norm": 0.9300563335418701,
      "learning_rate": 0.00021900402782213052,
      "loss": 2.1725,
      "step": 135251
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.2295968532562256,
      "learning_rate": 0.00021900008917424935,
      "loss": 2.2263,
      "step": 135252
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.030194878578186,
      "learning_rate": 0.00021899615054142758,
      "loss": 2.2494,
      "step": 135253
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.079350233078003,
      "learning_rate": 0.00021899221192366597,
      "loss": 2.046,
      "step": 135254
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.141276478767395,
      "learning_rate": 0.00021898827332096535,
      "loss": 2.0123,
      "step": 135255
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.1497776508331299,
      "learning_rate": 0.00021898433473332624,
      "loss": 2.16,
      "step": 135256
    },
    {
      "epoch": 1.76,
      "grad_norm": 0.975073516368866,
      "learning_rate": 0.00021898039616074946,
      "loss": 2.1393,
      "step": 135257
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.0326175689697266,
      "learning_rate": 0.00021897645760323573,
      "loss": 2.3867,
      "step": 135258
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.0131752490997314,
      "learning_rate": 0.0002189725190607858,
      "loss": 2.0564,
      "step": 135259
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.2417857646942139,
      "learning_rate": 0.00021896858053340045,
      "loss": 2.1391,
      "step": 135260
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.068583607673645,
      "learning_rate": 0.00021896464202108052,
      "loss": 2.226,
      "step": 135261
    },
    {
      "epoch": 1.76,
      "grad_norm": 0.9979727268218994,
      "learning_rate": 0.00021896070352382638,
      "loss": 1.9967,
      "step": 135262
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.094730257987976,
      "learning_rate": 0.00021895676504163903,
      "loss": 1.9031,
      "step": 135263
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.108066201210022,
      "learning_rate": 0.00021895282657451916,
      "loss": 2.2066,
      "step": 135264
    },
    {
      "epoch": 1.76,
      "grad_norm": 0.9988281726837158,
      "learning_rate": 0.00021894888812246745,
      "loss": 2.2071,
      "step": 135265
    },
    {
      "epoch": 1.76,
      "grad_norm": 0.981968343257904,
      "learning_rate": 0.00021894494968548465,
      "loss": 2.0588,
      "step": 135266
    },
    {
      "epoch": 1.76,
      "grad_norm": 0.9951614737510681,
      "learning_rate": 0.00021894101126357173,
      "loss": 2.1074,
      "step": 135267
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.0054751634597778,
      "learning_rate": 0.000218937072856729,
      "loss": 2.202,
      "step": 135268
    },
    {
      "epoch": 1.76,
      "grad_norm": 0.9679979085922241,
      "learning_rate": 0.00021893313446495744,
      "loss": 2.0585,
      "step": 135269
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.0822354555130005,
      "learning_rate": 0.00021892919608825775,
      "loss": 1.9604,
      "step": 135270
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.0145936012268066,
      "learning_rate": 0.0002189252577266306,
      "loss": 1.9416,
      "step": 135271
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.5311130285263062,
      "learning_rate": 0.00021892131938007682,
      "loss": 2.1776,
      "step": 135272
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.2060890197753906,
      "learning_rate": 0.00021891738104859717,
      "loss": 2.0932,
      "step": 135273
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.0434949398040771,
      "learning_rate": 0.00021891344273219224,
      "loss": 2.0085,
      "step": 135274
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.1240150928497314,
      "learning_rate": 0.00021890950443086282,
      "loss": 2.144,
      "step": 135275
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.5063079595565796,
      "learning_rate": 0.00021890556614460963,
      "loss": 2.2574,
      "step": 135276
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.158630132675171,
      "learning_rate": 0.00021890162787343345,
      "loss": 2.1687,
      "step": 135277
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.047831654548645,
      "learning_rate": 0.000218897689617335,
      "loss": 1.8806,
      "step": 135278
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.0501089096069336,
      "learning_rate": 0.0002188937513763151,
      "loss": 2.0754,
      "step": 135279
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.1421631574630737,
      "learning_rate": 0.0002188898131503743,
      "loss": 1.8417,
      "step": 135280
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.7239437103271484,
      "learning_rate": 0.00021888587493951333,
      "loss": 2.1291,
      "step": 135281
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.2142367362976074,
      "learning_rate": 0.00021888193674373307,
      "loss": 2.1873,
      "step": 135282
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.0945507287979126,
      "learning_rate": 0.00021887799856303415,
      "loss": 2.1857,
      "step": 135283
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.250138521194458,
      "learning_rate": 0.00021887406039741738,
      "loss": 2.2486,
      "step": 135284
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.6242483854293823,
      "learning_rate": 0.00021887012224688357,
      "loss": 1.9558,
      "step": 135285
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.1723970174789429,
      "learning_rate": 0.0002188661841114332,
      "loss": 2.1474,
      "step": 135286
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.0753213167190552,
      "learning_rate": 0.00021886224599106715,
      "loss": 2.1001,
      "step": 135287
    },
    {
      "epoch": 1.76,
      "grad_norm": 0.9959831833839417,
      "learning_rate": 0.00021885830788578615,
      "loss": 2.1634,
      "step": 135288
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.2958813905715942,
      "learning_rate": 0.0002188543697955909,
      "loss": 2.2396,
      "step": 135289
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.3005415201187134,
      "learning_rate": 0.00021885043172048215,
      "loss": 2.1675,
      "step": 135290
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.111366868019104,
      "learning_rate": 0.0002188464936604608,
      "loss": 1.8822,
      "step": 135291
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.078203797340393,
      "learning_rate": 0.00021884255561552726,
      "loss": 1.9957,
      "step": 135292
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.0877511501312256,
      "learning_rate": 0.00021883861758568242,
      "loss": 2.2174,
      "step": 135293
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.3513094186782837,
      "learning_rate": 0.000218834679570927,
      "loss": 2.2796,
      "step": 135294
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.023733377456665,
      "learning_rate": 0.0002188307415712618,
      "loss": 2.1852,
      "step": 135295
    },
    {
      "epoch": 1.76,
      "grad_norm": 0.9271494150161743,
      "learning_rate": 0.00021882680358668744,
      "loss": 2.1194,
      "step": 135296
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.2244518995285034,
      "learning_rate": 0.0002188228656172048,
      "loss": 1.92,
      "step": 135297
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.428632140159607,
      "learning_rate": 0.0002188189276628145,
      "loss": 2.163,
      "step": 135298
    },
    {
      "epoch": 1.76,
      "grad_norm": 0.9486628770828247,
      "learning_rate": 0.00021881498972351726,
      "loss": 1.9303,
      "step": 135299
    },
    {
      "epoch": 1.76,
      "grad_norm": 0.9571654200553894,
      "learning_rate": 0.00021881105179931385,
      "loss": 2.0339,
      "step": 135300
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.1823843717575073,
      "learning_rate": 0.00021880711389020497,
      "loss": 2.0265,
      "step": 135301
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.0675148963928223,
      "learning_rate": 0.00021880317599619136,
      "loss": 2.0136,
      "step": 135302
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.1486157178878784,
      "learning_rate": 0.00021879923811727388,
      "loss": 2.1104,
      "step": 135303
    },
    {
      "epoch": 1.76,
      "grad_norm": 0.9668206572532654,
      "learning_rate": 0.00021879530025345308,
      "loss": 2.0715,
      "step": 135304
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.0785608291625977,
      "learning_rate": 0.00021879136240472976,
      "loss": 2.1869,
      "step": 135305
    },
    {
      "epoch": 1.76,
      "grad_norm": 0.9499965310096741,
      "learning_rate": 0.00021878742457110472,
      "loss": 1.9773,
      "step": 135306
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.3141345977783203,
      "learning_rate": 0.0002187834867525786,
      "loss": 2.0286,
      "step": 135307
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.0125268697738647,
      "learning_rate": 0.00021877954894915214,
      "loss": 2.1437,
      "step": 135308
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.0501511096954346,
      "learning_rate": 0.00021877561116082615,
      "loss": 2.1536,
      "step": 135309
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.1047570705413818,
      "learning_rate": 0.00021877167338760122,
      "loss": 1.9641,
      "step": 135310
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.0772475004196167,
      "learning_rate": 0.0002187677356294782,
      "loss": 2.1979,
      "step": 135311
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.1173226833343506,
      "learning_rate": 0.0002187637978864578,
      "loss": 2.229,
      "step": 135312
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.0685203075408936,
      "learning_rate": 0.00021875986015854079,
      "loss": 2.2773,
      "step": 135313
    },
    {
      "epoch": 1.76,
      "grad_norm": 0.9998800158500671,
      "learning_rate": 0.00021875592244572785,
      "loss": 2.0602,
      "step": 135314
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.0192722082138062,
      "learning_rate": 0.00021875198474801967,
      "loss": 2.1317,
      "step": 135315
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.0962340831756592,
      "learning_rate": 0.00021874804706541705,
      "loss": 2.478,
      "step": 135316
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.0862295627593994,
      "learning_rate": 0.00021874410939792067,
      "loss": 2.0572,
      "step": 135317
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.1155149936676025,
      "learning_rate": 0.0002187401717455313,
      "loss": 2.0941,
      "step": 135318
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.056394338607788,
      "learning_rate": 0.0002187362341082497,
      "loss": 2.0875,
      "step": 135319
    },
    {
      "epoch": 1.76,
      "grad_norm": 0.967364490032196,
      "learning_rate": 0.00021873229648607657,
      "loss": 2.0038,
      "step": 135320
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.0499577522277832,
      "learning_rate": 0.00021872835887901265,
      "loss": 2.3292,
      "step": 135321
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.3513542413711548,
      "learning_rate": 0.00021872442128705863,
      "loss": 2.1161,
      "step": 135322
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.0406948328018188,
      "learning_rate": 0.0002187204837102153,
      "loss": 2.3992,
      "step": 135323
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.0031709671020508,
      "learning_rate": 0.00021871654614848327,
      "loss": 2.173,
      "step": 135324
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.0664479732513428,
      "learning_rate": 0.00021871260860186346,
      "loss": 2.0779,
      "step": 135325
    },
    {
      "epoch": 1.76,
      "grad_norm": 0.9918095469474792,
      "learning_rate": 0.00021870867107035648,
      "loss": 2.1392,
      "step": 135326
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.0920385122299194,
      "learning_rate": 0.00021870473355396314,
      "loss": 2.1689,
      "step": 135327
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.0824116468429565,
      "learning_rate": 0.00021870079605268414,
      "loss": 2.1397,
      "step": 135328
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.0931910276412964,
      "learning_rate": 0.00021869685856652012,
      "loss": 1.9651,
      "step": 135329
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.2554537057876587,
      "learning_rate": 0.0002186929210954719,
      "loss": 2.2877,
      "step": 135330
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.360309362411499,
      "learning_rate": 0.0002186889836395402,
      "loss": 2.0176,
      "step": 135331
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.0449621677398682,
      "learning_rate": 0.00021868504619872576,
      "loss": 2.0017,
      "step": 135332
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.4361199140548706,
      "learning_rate": 0.0002186811087730293,
      "loss": 1.9865,
      "step": 135333
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.1823495626449585,
      "learning_rate": 0.00021867717136245167,
      "loss": 2.1127,
      "step": 135334
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.3637568950653076,
      "learning_rate": 0.00021867323396699341,
      "loss": 2.0555,
      "step": 135335
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.071480393409729,
      "learning_rate": 0.00021866929658665528,
      "loss": 2.2077,
      "step": 135336
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.1702693700790405,
      "learning_rate": 0.00021866535922143807,
      "loss": 2.0953,
      "step": 135337
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.2693747282028198,
      "learning_rate": 0.0002186614218713425,
      "loss": 2.4367,
      "step": 135338
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.010524034500122,
      "learning_rate": 0.00021865748453636936,
      "loss": 2.1605,
      "step": 135339
    },
    {
      "epoch": 1.76,
      "grad_norm": 0.9953880906105042,
      "learning_rate": 0.0002186535472165194,
      "loss": 1.9822,
      "step": 135340
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.0860027074813843,
      "learning_rate": 0.00021864960991179315,
      "loss": 2.1872,
      "step": 135341
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.3759962320327759,
      "learning_rate": 0.00021864567262219154,
      "loss": 2.0296,
      "step": 135342
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.2580114603042603,
      "learning_rate": 0.00021864173534771514,
      "loss": 2.0468,
      "step": 135343
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.1427909135818481,
      "learning_rate": 0.00021863779808836486,
      "loss": 2.208,
      "step": 135344
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.0461012125015259,
      "learning_rate": 0.00021863386084414128,
      "loss": 2.0697,
      "step": 135345
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.2099889516830444,
      "learning_rate": 0.0002186299236150454,
      "loss": 1.9404,
      "step": 135346
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.0152019262313843,
      "learning_rate": 0.00021862598640107756,
      "loss": 2.055,
      "step": 135347
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.1468546390533447,
      "learning_rate": 0.00021862204920223871,
      "loss": 2.3348,
      "step": 135348
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.0523338317871094,
      "learning_rate": 0.00021861811201852958,
      "loss": 2.0189,
      "step": 135349
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.0273455381393433,
      "learning_rate": 0.00021861417484995082,
      "loss": 2.159,
      "step": 135350
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.0755810737609863,
      "learning_rate": 0.0002186102376965033,
      "loss": 2.2211,
      "step": 135351
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.1311883926391602,
      "learning_rate": 0.00021860630055818775,
      "loss": 1.9785,
      "step": 135352
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.0286149978637695,
      "learning_rate": 0.00021860236343500473,
      "loss": 2.17,
      "step": 135353
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.4432193040847778,
      "learning_rate": 0.00021859842632695502,
      "loss": 2.2265,
      "step": 135354
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.2808122634887695,
      "learning_rate": 0.00021859448923403947,
      "loss": 2.1487,
      "step": 135355
    },
    {
      "epoch": 1.76,
      "grad_norm": 0.9925961494445801,
      "learning_rate": 0.00021859055215625868,
      "loss": 2.1159,
      "step": 135356
    },
    {
      "epoch": 1.76,
      "grad_norm": 0.9991981983184814,
      "learning_rate": 0.00021858661509361344,
      "loss": 2.0761,
      "step": 135357
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.0431530475616455,
      "learning_rate": 0.00021858267804610463,
      "loss": 1.8924,
      "step": 135358
    },
    {
      "epoch": 1.76,
      "grad_norm": 0.9964339137077332,
      "learning_rate": 0.00021857874101373272,
      "loss": 1.9745,
      "step": 135359
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.1286959648132324,
      "learning_rate": 0.00021857480399649855,
      "loss": 2.1779,
      "step": 135360
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.1224149465560913,
      "learning_rate": 0.00021857086699440285,
      "loss": 2.1418,
      "step": 135361
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.0371677875518799,
      "learning_rate": 0.0002185669300074464,
      "loss": 1.9944,
      "step": 135362
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.0334482192993164,
      "learning_rate": 0.00021856299303562985,
      "loss": 2.0291,
      "step": 135363
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.0324972867965698,
      "learning_rate": 0.00021855905607895412,
      "loss": 2.0818,
      "step": 135364
    },
    {
      "epoch": 1.76,
      "grad_norm": 0.9524720311164856,
      "learning_rate": 0.00021855511913741968,
      "loss": 1.7983,
      "step": 135365
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.0465549230575562,
      "learning_rate": 0.00021855118221102737,
      "loss": 2.1475,
      "step": 135366
    },
    {
      "epoch": 1.76,
      "grad_norm": 0.9818018078804016,
      "learning_rate": 0.00021854724529977797,
      "loss": 2.0468,
      "step": 135367
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.1331428289413452,
      "learning_rate": 0.00021854330840367212,
      "loss": 1.9101,
      "step": 135368
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.2170672416687012,
      "learning_rate": 0.00021853937152271064,
      "loss": 2.1212,
      "step": 135369
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.3187042474746704,
      "learning_rate": 0.00021853543465689433,
      "loss": 2.2252,
      "step": 135370
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.1037465333938599,
      "learning_rate": 0.00021853149780622373,
      "loss": 2.1339,
      "step": 135371
    },
    {
      "epoch": 1.76,
      "grad_norm": 0.9988755583763123,
      "learning_rate": 0.00021852756097069962,
      "loss": 2.1391,
      "step": 135372
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.0466992855072021,
      "learning_rate": 0.0002185236241503228,
      "loss": 2.1033,
      "step": 135373
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.054366111755371,
      "learning_rate": 0.000218519687345094,
      "loss": 2.1785,
      "step": 135374
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.01725435256958,
      "learning_rate": 0.00021851575055501386,
      "loss": 1.9388,
      "step": 135375
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.0325639247894287,
      "learning_rate": 0.00021851181378008337,
      "loss": 2.2296,
      "step": 135376
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.0143907070159912,
      "learning_rate": 0.00021850787702030293,
      "loss": 2.0827,
      "step": 135377
    },
    {
      "epoch": 1.76,
      "grad_norm": 0.9204541444778442,
      "learning_rate": 0.0002185039402756734,
      "loss": 2.0214,
      "step": 135378
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.0682698488235474,
      "learning_rate": 0.00021850000354619554,
      "loss": 2.1733,
      "step": 135379
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.1879427433013916,
      "learning_rate": 0.00021849606683187005,
      "loss": 1.948,
      "step": 135380
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.0302915573120117,
      "learning_rate": 0.0002184921301326977,
      "loss": 2.1097,
      "step": 135381
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.035603404045105,
      "learning_rate": 0.00021848819344867925,
      "loss": 2.123,
      "step": 135382
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.0907295942306519,
      "learning_rate": 0.0002184842567798154,
      "loss": 2.269,
      "step": 135383
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.0809857845306396,
      "learning_rate": 0.00021848032012610678,
      "loss": 2.0322,
      "step": 135384
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.1890836954116821,
      "learning_rate": 0.0002184763834875542,
      "loss": 2.2602,
      "step": 135385
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.0367498397827148,
      "learning_rate": 0.00021847244686415846,
      "loss": 1.9235,
      "step": 135386
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.1707415580749512,
      "learning_rate": 0.00021846851025592018,
      "loss": 2.0442,
      "step": 135387
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.0802507400512695,
      "learning_rate": 0.00021846457366284016,
      "loss": 2.203,
      "step": 135388
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.081343173980713,
      "learning_rate": 0.00021846063708491915,
      "loss": 2.1926,
      "step": 135389
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.443359375,
      "learning_rate": 0.00021845670052215787,
      "loss": 2.1048,
      "step": 135390
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.0225937366485596,
      "learning_rate": 0.00021845276397455693,
      "loss": 2.0459,
      "step": 135391
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.090382695198059,
      "learning_rate": 0.0002184488274421172,
      "loss": 2.2795,
      "step": 135392
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.1647262573242188,
      "learning_rate": 0.00021844489092483937,
      "loss": 2.0752,
      "step": 135393
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.0055551528930664,
      "learning_rate": 0.00021844095442272415,
      "loss": 2.0246,
      "step": 135394
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.0682752132415771,
      "learning_rate": 0.00021843701793577242,
      "loss": 1.8321,
      "step": 135395
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.229579210281372,
      "learning_rate": 0.00021843308146398467,
      "loss": 2.0562,
      "step": 135396
    },
    {
      "epoch": 1.76,
      "grad_norm": 0.9643900394439697,
      "learning_rate": 0.00021842914500736182,
      "loss": 1.9897,
      "step": 135397
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.062507152557373,
      "learning_rate": 0.0002184252085659045,
      "loss": 1.9204,
      "step": 135398
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.2601163387298584,
      "learning_rate": 0.00021842127213961342,
      "loss": 2.2511,
      "step": 135399
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.1145877838134766,
      "learning_rate": 0.0002184173357284894,
      "loss": 2.0592,
      "step": 135400
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.054044246673584,
      "learning_rate": 0.0002184133993325332,
      "loss": 1.9342,
      "step": 135401
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.0407512187957764,
      "learning_rate": 0.00021840946295174544,
      "loss": 1.9855,
      "step": 135402
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.1187738180160522,
      "learning_rate": 0.00021840552658612686,
      "loss": 2.0815,
      "step": 135403
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.1657147407531738,
      "learning_rate": 0.00021840159023567824,
      "loss": 2.1318,
      "step": 135404
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.020500659942627,
      "learning_rate": 0.00021839765390040038,
      "loss": 2.2237,
      "step": 135405
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.1367053985595703,
      "learning_rate": 0.00021839371758029387,
      "loss": 2.1101,
      "step": 135406
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.008707880973816,
      "learning_rate": 0.00021838978127535958,
      "loss": 1.9792,
      "step": 135407
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.1095094680786133,
      "learning_rate": 0.0002183858449855981,
      "loss": 2.3022,
      "step": 135408
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.302101969718933,
      "learning_rate": 0.00021838190871101024,
      "loss": 2.0022,
      "step": 135409
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.305481195449829,
      "learning_rate": 0.00021837797245159675,
      "loss": 1.9598,
      "step": 135410
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.0317620038986206,
      "learning_rate": 0.00021837403620735829,
      "loss": 2.1622,
      "step": 135411
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.1352944374084473,
      "learning_rate": 0.0002183700999782957,
      "loss": 2.1226,
      "step": 135412
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.444947361946106,
      "learning_rate": 0.00021836616376440964,
      "loss": 1.9621,
      "step": 135413
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.2085732221603394,
      "learning_rate": 0.00021836222756570084,
      "loss": 2.4243,
      "step": 135414
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.0420243740081787,
      "learning_rate": 0.00021835829138217003,
      "loss": 2.0987,
      "step": 135415
    },
    {
      "epoch": 1.76,
      "grad_norm": 0.9513238072395325,
      "learning_rate": 0.00021835435521381794,
      "loss": 1.9963,
      "step": 135416
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.1527400016784668,
      "learning_rate": 0.0002183504190606453,
      "loss": 2.3041,
      "step": 135417
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.7762813568115234,
      "learning_rate": 0.0002183464829226529,
      "loss": 2.2798,
      "step": 135418
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.111323595046997,
      "learning_rate": 0.00021834254679984152,
      "loss": 2.0327,
      "step": 135419
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.0200369358062744,
      "learning_rate": 0.0002183386106922117,
      "loss": 1.9885,
      "step": 135420
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.1822978258132935,
      "learning_rate": 0.00021833467459976425,
      "loss": 2.0799,
      "step": 135421
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.1766856908798218,
      "learning_rate": 0.00021833073852249994,
      "loss": 2.0155,
      "step": 135422
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.10181725025177,
      "learning_rate": 0.00021832680246041948,
      "loss": 2.1146,
      "step": 135423
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.0445876121520996,
      "learning_rate": 0.0002183228664135236,
      "loss": 1.9301,
      "step": 135424
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.058972716331482,
      "learning_rate": 0.00021831893038181316,
      "loss": 1.9506,
      "step": 135425
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.4526857137680054,
      "learning_rate": 0.00021831499436528867,
      "loss": 1.9868,
      "step": 135426
    },
    {
      "epoch": 1.76,
      "grad_norm": 0.995489776134491,
      "learning_rate": 0.000218311058363951,
      "loss": 2.0852,
      "step": 135427
    },
    {
      "epoch": 1.76,
      "grad_norm": 0.9592591524124146,
      "learning_rate": 0.0002183071223778008,
      "loss": 2.1657,
      "step": 135428
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.0748136043548584,
      "learning_rate": 0.00021830318640683884,
      "loss": 2.01,
      "step": 135429
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.1056116819381714,
      "learning_rate": 0.00021829925045106587,
      "loss": 2.1308,
      "step": 135430
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.1112399101257324,
      "learning_rate": 0.00021829531451048273,
      "loss": 2.1135,
      "step": 135431
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.0286836624145508,
      "learning_rate": 0.00021829137858508994,
      "loss": 2.2375,
      "step": 135432
    },
    {
      "epoch": 1.76,
      "grad_norm": 0.9992730021476746,
      "learning_rate": 0.00021828744267488832,
      "loss": 2.0207,
      "step": 135433
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.1442948579788208,
      "learning_rate": 0.0002182835067798786,
      "loss": 1.7182,
      "step": 135434
    },
    {
      "epoch": 1.76,
      "grad_norm": 0.9964497685432434,
      "learning_rate": 0.00021827957090006148,
      "loss": 2.0332,
      "step": 135435
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.0723527669906616,
      "learning_rate": 0.0002182756350354378,
      "loss": 2.0853,
      "step": 135436
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.091591477394104,
      "learning_rate": 0.0002182716991860083,
      "loss": 2.2881,
      "step": 135437
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.0253897905349731,
      "learning_rate": 0.0002182677633517735,
      "loss": 2.2719,
      "step": 135438
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.074175477027893,
      "learning_rate": 0.0002182638275327343,
      "loss": 2.1941,
      "step": 135439
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.254685640335083,
      "learning_rate": 0.00021825989172889138,
      "loss": 2.0107,
      "step": 135440
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.1251697540283203,
      "learning_rate": 0.00021825595594024546,
      "loss": 2.2305,
      "step": 135441
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.0876632928848267,
      "learning_rate": 0.00021825202016679735,
      "loss": 2.1901,
      "step": 135442
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.1421641111373901,
      "learning_rate": 0.00021824808440854782,
      "loss": 2.1722,
      "step": 135443
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.2503050565719604,
      "learning_rate": 0.0002182441486654974,
      "loss": 1.7322,
      "step": 135444
    },
    {
      "epoch": 1.76,
      "grad_norm": 0.9763807058334351,
      "learning_rate": 0.00021824021293764698,
      "loss": 2.2292,
      "step": 135445
    },
    {
      "epoch": 1.76,
      "grad_norm": 0.9339999556541443,
      "learning_rate": 0.00021823627722499718,
      "loss": 1.9427,
      "step": 135446
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.1096066236495972,
      "learning_rate": 0.00021823234152754884,
      "loss": 2.1337,
      "step": 135447
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.2567081451416016,
      "learning_rate": 0.00021822840584530262,
      "loss": 1.9644,
      "step": 135448
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.01139497756958,
      "learning_rate": 0.00021822447017825943,
      "loss": 1.8879,
      "step": 135449
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.0063693523406982,
      "learning_rate": 0.00021822053452641974,
      "loss": 1.9961,
      "step": 135450
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.2703474760055542,
      "learning_rate": 0.0002182165988897844,
      "loss": 2.0063,
      "step": 135451
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.3040075302124023,
      "learning_rate": 0.0002182126632683541,
      "loss": 2.0224,
      "step": 135452
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.0522419214248657,
      "learning_rate": 0.00021820872766212967,
      "loss": 2.0582,
      "step": 135453
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.051116704940796,
      "learning_rate": 0.00021820479207111172,
      "loss": 2.2235,
      "step": 135454
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.1680628061294556,
      "learning_rate": 0.00021820085649530108,
      "loss": 2.0034,
      "step": 135455
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.5076813697814941,
      "learning_rate": 0.00021819692093469855,
      "loss": 2.1192,
      "step": 135456
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.111035943031311,
      "learning_rate": 0.00021819298538930463,
      "loss": 2.2318,
      "step": 135457
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.0878223180770874,
      "learning_rate": 0.0002181890498591202,
      "loss": 2.0827,
      "step": 135458
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.0321422815322876,
      "learning_rate": 0.00021818511434414594,
      "loss": 2.0182,
      "step": 135459
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.1904774904251099,
      "learning_rate": 0.00021818117884438265,
      "loss": 2.1825,
      "step": 135460
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.0839015245437622,
      "learning_rate": 0.000218177243359831,
      "loss": 2.1996,
      "step": 135461
    },
    {
      "epoch": 1.76,
      "grad_norm": 0.9989756345748901,
      "learning_rate": 0.00021817330789049186,
      "loss": 1.9267,
      "step": 135462
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.019720196723938,
      "learning_rate": 0.00021816937243636573,
      "loss": 2.0379,
      "step": 135463
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.0343332290649414,
      "learning_rate": 0.00021816543699745345,
      "loss": 2.0504,
      "step": 135464
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.1287399530410767,
      "learning_rate": 0.0002181615015737558,
      "loss": 2.1773,
      "step": 135465
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.0487971305847168,
      "learning_rate": 0.00021815756616527346,
      "loss": 2.0187,
      "step": 135466
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.0650979280471802,
      "learning_rate": 0.00021815363077200712,
      "loss": 1.9379,
      "step": 135467
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.1067403554916382,
      "learning_rate": 0.00021814969539395775,
      "loss": 2.1581,
      "step": 135468
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.0430785417556763,
      "learning_rate": 0.00021814576003112573,
      "loss": 2.1491,
      "step": 135469
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.0667401552200317,
      "learning_rate": 0.000218141824683512,
      "loss": 1.9501,
      "step": 135470
    },
    {
      "epoch": 1.76,
      "grad_norm": 0.9857708811759949,
      "learning_rate": 0.0002181378893511172,
      "loss": 2.0733,
      "step": 135471
    },
    {
      "epoch": 1.76,
      "grad_norm": 0.9721569418907166,
      "learning_rate": 0.00021813395403394214,
      "loss": 2.0574,
      "step": 135472
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.1628689765930176,
      "learning_rate": 0.00021813001873198753,
      "loss": 2.2224,
      "step": 135473
    },
    {
      "epoch": 1.76,
      "grad_norm": 0.9994240999221802,
      "learning_rate": 0.00021812608344525414,
      "loss": 2.2294,
      "step": 135474
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.1657463312149048,
      "learning_rate": 0.00021812214817374268,
      "loss": 2.0021,
      "step": 135475
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.1211878061294556,
      "learning_rate": 0.0002181182129174538,
      "loss": 2.3338,
      "step": 135476
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.0951108932495117,
      "learning_rate": 0.00021811427767638825,
      "loss": 2.0982,
      "step": 135477
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.3623493909835815,
      "learning_rate": 0.00021811034245054683,
      "loss": 2.2116,
      "step": 135478
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.0022913217544556,
      "learning_rate": 0.00021810640723993023,
      "loss": 2.2454,
      "step": 135479
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.05509614944458,
      "learning_rate": 0.00021810247204453927,
      "loss": 2.127,
      "step": 135480
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.0215574502944946,
      "learning_rate": 0.00021809853686437456,
      "loss": 2.0255,
      "step": 135481
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.02736496925354,
      "learning_rate": 0.00021809460169943691,
      "loss": 2.1532,
      "step": 135482
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.1790623664855957,
      "learning_rate": 0.00021809066654972695,
      "loss": 1.9101,
      "step": 135483
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.0272701978683472,
      "learning_rate": 0.0002180867314152455,
      "loss": 2.1972,
      "step": 135484
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.1230560541152954,
      "learning_rate": 0.00021808279629599323,
      "loss": 2.0858,
      "step": 135485
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.030738115310669,
      "learning_rate": 0.00021807886119197106,
      "loss": 2.0036,
      "step": 135486
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.0145264863967896,
      "learning_rate": 0.00021807492610317945,
      "loss": 1.9519,
      "step": 135487
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.1525344848632812,
      "learning_rate": 0.00021807099102961927,
      "loss": 2.1008,
      "step": 135488
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.0559864044189453,
      "learning_rate": 0.00021806705597129133,
      "loss": 2.0896,
      "step": 135489
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.126589059829712,
      "learning_rate": 0.00021806312092819614,
      "loss": 2.0101,
      "step": 135490
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.1909376382827759,
      "learning_rate": 0.00021805918590033467,
      "loss": 2.0753,
      "step": 135491
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.12135910987854,
      "learning_rate": 0.0002180552508877075,
      "loss": 2.2059,
      "step": 135492
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.2546923160552979,
      "learning_rate": 0.0002180513158903154,
      "loss": 2.2146,
      "step": 135493
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.1107491254806519,
      "learning_rate": 0.00021804738090815912,
      "loss": 2.4018,
      "step": 135494
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.0515868663787842,
      "learning_rate": 0.00021804344594123932,
      "loss": 2.2006,
      "step": 135495
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.1106452941894531,
      "learning_rate": 0.00021803951098955685,
      "loss": 2.0331,
      "step": 135496
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.1009175777435303,
      "learning_rate": 0.00021803557605311243,
      "loss": 2.1239,
      "step": 135497
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.0450469255447388,
      "learning_rate": 0.00021803164113190672,
      "loss": 2.2407,
      "step": 135498
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.0115095376968384,
      "learning_rate": 0.00021802770622594044,
      "loss": 2.1836,
      "step": 135499
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.0146392583847046,
      "learning_rate": 0.00021802377133521434,
      "loss": 2.0332,
      "step": 135500
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.1663585901260376,
      "learning_rate": 0.00021801983645972916,
      "loss": 2.0498,
      "step": 135501
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.11115562915802,
      "learning_rate": 0.00021801590159948564,
      "loss": 2.1414,
      "step": 135502
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.0251463651657104,
      "learning_rate": 0.00021801196675448457,
      "loss": 2.0188,
      "step": 135503
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.52377188205719,
      "learning_rate": 0.00021800803192472667,
      "loss": 2.324,
      "step": 135504
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.656221628189087,
      "learning_rate": 0.00021800409711021253,
      "loss": 2.016,
      "step": 135505
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.1589964628219604,
      "learning_rate": 0.00021800016231094297,
      "loss": 1.9671,
      "step": 135506
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.020338535308838,
      "learning_rate": 0.00021799622752691878,
      "loss": 2.2337,
      "step": 135507
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.2939354181289673,
      "learning_rate": 0.00021799229275814056,
      "loss": 2.0112,
      "step": 135508
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.0435590744018555,
      "learning_rate": 0.00021798835800460915,
      "loss": 2.3849,
      "step": 135509
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.4650318622589111,
      "learning_rate": 0.00021798442326632537,
      "loss": 1.9205,
      "step": 135510
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.0133651494979858,
      "learning_rate": 0.00021798048854328975,
      "loss": 2.2502,
      "step": 135511
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.0271756649017334,
      "learning_rate": 0.00021797655383550305,
      "loss": 2.0278,
      "step": 135512
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.0567740201950073,
      "learning_rate": 0.0002179726191429661,
      "loss": 2.2913,
      "step": 135513
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.1750867366790771,
      "learning_rate": 0.00021796868446567958,
      "loss": 2.24,
      "step": 135514
    },
    {
      "epoch": 1.76,
      "grad_norm": 0.993307888507843,
      "learning_rate": 0.0002179647498036442,
      "loss": 2.0732,
      "step": 135515
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.059998631477356,
      "learning_rate": 0.0002179608151568609,
      "loss": 2.0649,
      "step": 135516
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.1833728551864624,
      "learning_rate": 0.00021795688052533006,
      "loss": 2.2306,
      "step": 135517
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.0863312482833862,
      "learning_rate": 0.0002179529459090526,
      "loss": 2.0287,
      "step": 135518
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.0319805145263672,
      "learning_rate": 0.00021794901130802927,
      "loss": 2.1812,
      "step": 135519
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.0522702932357788,
      "learning_rate": 0.0002179450767222607,
      "loss": 2.1478,
      "step": 135520
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.4170105457305908,
      "learning_rate": 0.00021794114215174774,
      "loss": 1.9136,
      "step": 135521
    },
    {
      "epoch": 1.76,
      "grad_norm": 0.9878935217857361,
      "learning_rate": 0.00021793720759649107,
      "loss": 2.0972,
      "step": 135522
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.0119558572769165,
      "learning_rate": 0.00021793327305649152,
      "loss": 2.1212,
      "step": 135523
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.16685152053833,
      "learning_rate": 0.0002179293385317496,
      "loss": 2.0296,
      "step": 135524
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.1459712982177734,
      "learning_rate": 0.00021792540402226616,
      "loss": 1.9859,
      "step": 135525
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.0601303577423096,
      "learning_rate": 0.00021792146952804195,
      "loss": 2.1672,
      "step": 135526
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.0430569648742676,
      "learning_rate": 0.00021791753504907772,
      "loss": 2.1673,
      "step": 135527
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.2453211545944214,
      "learning_rate": 0.00021791360058537416,
      "loss": 2.1093,
      "step": 135528
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.1018874645233154,
      "learning_rate": 0.00021790966613693212,
      "loss": 2.0899,
      "step": 135529
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.0115050077438354,
      "learning_rate": 0.00021790573170375204,
      "loss": 2.1879,
      "step": 135530
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.1494745016098022,
      "learning_rate": 0.00021790179728583493,
      "loss": 2.0739,
      "step": 135531
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.0176126956939697,
      "learning_rate": 0.00021789786288318137,
      "loss": 2.0941,
      "step": 135532
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.1411811113357544,
      "learning_rate": 0.00021789392849579217,
      "loss": 2.2267,
      "step": 135533
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.128822922706604,
      "learning_rate": 0.00021788999412366802,
      "loss": 1.9891,
      "step": 135534
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.1537508964538574,
      "learning_rate": 0.0002178860597668098,
      "loss": 2.3622,
      "step": 135535
    },
    {
      "epoch": 1.76,
      "grad_norm": 0.9873295426368713,
      "learning_rate": 0.00021788212542521798,
      "loss": 2.0543,
      "step": 135536
    },
    {
      "epoch": 1.76,
      "grad_norm": 0.9875760674476624,
      "learning_rate": 0.00021787819109889345,
      "loss": 2.0215,
      "step": 135537
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.0213797092437744,
      "learning_rate": 0.00021787425678783688,
      "loss": 2.2503,
      "step": 135538
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.0411115884780884,
      "learning_rate": 0.00021787032249204907,
      "loss": 2.2381,
      "step": 135539
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.0140981674194336,
      "learning_rate": 0.0002178663882115307,
      "loss": 1.8616,
      "step": 135540
    },
    {
      "epoch": 1.76,
      "grad_norm": 0.9681167602539062,
      "learning_rate": 0.00021786245394628265,
      "loss": 2.1482,
      "step": 135541
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.0721261501312256,
      "learning_rate": 0.00021785851969630539,
      "loss": 2.2141,
      "step": 135542
    },
    {
      "epoch": 1.76,
      "grad_norm": 0.9377065896987915,
      "learning_rate": 0.0002178545854615998,
      "loss": 2.1472,
      "step": 135543
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.0203818082809448,
      "learning_rate": 0.00021785065124216657,
      "loss": 2.389,
      "step": 135544
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.0245970487594604,
      "learning_rate": 0.00021784671703800645,
      "loss": 1.9101,
      "step": 135545
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.0808486938476562,
      "learning_rate": 0.00021784278284912018,
      "loss": 2.0927,
      "step": 135546
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.0101860761642456,
      "learning_rate": 0.00021783884867550862,
      "loss": 2.1991,
      "step": 135547
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.0352120399475098,
      "learning_rate": 0.00021783491451717225,
      "loss": 2.2299,
      "step": 135548
    },
    {
      "epoch": 1.76,
      "grad_norm": 0.9650355577468872,
      "learning_rate": 0.00021783098037411194,
      "loss": 2.082,
      "step": 135549
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.1302437782287598,
      "learning_rate": 0.00021782704624632833,
      "loss": 2.1823,
      "step": 135550
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.0341793298721313,
      "learning_rate": 0.0002178231121338223,
      "loss": 2.0161,
      "step": 135551
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.0049372911453247,
      "learning_rate": 0.00021781917803659443,
      "loss": 2.0448,
      "step": 135552
    },
    {
      "epoch": 1.76,
      "grad_norm": 1.0310574769973755,
      "learning_rate": 0.0002178152439546457,
      "loss": 2.1263,
      "step": 135553
    },
    {
      "epoch": 1.77,
      "grad_norm": 0.9839210510253906,
      "learning_rate": 0.00021781130988797657,
      "loss": 2.0408,
      "step": 135554
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.125509262084961,
      "learning_rate": 0.0002178073758365878,
      "loss": 2.2397,
      "step": 135555
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.0136383771896362,
      "learning_rate": 0.00021780344180048026,
      "loss": 2.0822,
      "step": 135556
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.0002940893173218,
      "learning_rate": 0.00021779950777965455,
      "loss": 2.2761,
      "step": 135557
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.0319554805755615,
      "learning_rate": 0.00021779557377411149,
      "loss": 2.0122,
      "step": 135558
    },
    {
      "epoch": 1.77,
      "grad_norm": 0.89377361536026,
      "learning_rate": 0.00021779163978385196,
      "loss": 2.0596,
      "step": 135559
    },
    {
      "epoch": 1.77,
      "grad_norm": 0.9439605474472046,
      "learning_rate": 0.0002177877058088763,
      "loss": 1.9765,
      "step": 135560
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.0198014974594116,
      "learning_rate": 0.0002177837718491855,
      "loss": 2.2198,
      "step": 135561
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.12553870677948,
      "learning_rate": 0.00021777983790478026,
      "loss": 2.2726,
      "step": 135562
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.0200409889221191,
      "learning_rate": 0.00021777590397566127,
      "loss": 1.969,
      "step": 135563
    },
    {
      "epoch": 1.77,
      "grad_norm": 0.9829607009887695,
      "learning_rate": 0.00021777197006182933,
      "loss": 2.0313,
      "step": 135564
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.0765256881713867,
      "learning_rate": 0.00021776803616328517,
      "loss": 2.189,
      "step": 135565
    },
    {
      "epoch": 1.77,
      "grad_norm": 0.977927029132843,
      "learning_rate": 0.00021776410228002943,
      "loss": 1.9062,
      "step": 135566
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.0981515645980835,
      "learning_rate": 0.00021776016841206294,
      "loss": 2.0185,
      "step": 135567
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.1585110425949097,
      "learning_rate": 0.00021775623455938634,
      "loss": 2.1055,
      "step": 135568
    },
    {
      "epoch": 1.77,
      "grad_norm": 0.9901860356330872,
      "learning_rate": 0.00021775230072200043,
      "loss": 2.1105,
      "step": 135569
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.0870059728622437,
      "learning_rate": 0.00021774836689990589,
      "loss": 2.1113,
      "step": 135570
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.1041467189788818,
      "learning_rate": 0.00021774443309310353,
      "loss": 2.1835,
      "step": 135571
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.1766284704208374,
      "learning_rate": 0.00021774049930159398,
      "loss": 2.4377,
      "step": 135572
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.0210424661636353,
      "learning_rate": 0.00021773656552537808,
      "loss": 2.0847,
      "step": 135573
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.1460340023040771,
      "learning_rate": 0.00021773263176445647,
      "loss": 2.0353,
      "step": 135574
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.0097070932388306,
      "learning_rate": 0.00021772869801882995,
      "loss": 1.887,
      "step": 135575
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.2641682624816895,
      "learning_rate": 0.00021772476428849917,
      "loss": 2.1759,
      "step": 135576
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.3269401788711548,
      "learning_rate": 0.00021772083057346498,
      "loss": 2.1962,
      "step": 135577
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.003117322921753,
      "learning_rate": 0.00021771689687372797,
      "loss": 2.1212,
      "step": 135578
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.1306488513946533,
      "learning_rate": 0.0002177129631892889,
      "loss": 2.2094,
      "step": 135579
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.0402758121490479,
      "learning_rate": 0.00021770902952014863,
      "loss": 2.015,
      "step": 135580
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.0647698640823364,
      "learning_rate": 0.00021770509586630784,
      "loss": 1.9256,
      "step": 135581
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.0313098430633545,
      "learning_rate": 0.00021770116222776716,
      "loss": 1.9903,
      "step": 135582
    },
    {
      "epoch": 1.77,
      "grad_norm": 0.9754980802536011,
      "learning_rate": 0.0002176972286045274,
      "loss": 1.9575,
      "step": 135583
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.0656501054763794,
      "learning_rate": 0.0002176932949965893,
      "loss": 2.1254,
      "step": 135584
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.0139998197555542,
      "learning_rate": 0.0002176893614039535,
      "loss": 2.3154,
      "step": 135585
    },
    {
      "epoch": 1.77,
      "grad_norm": 0.9666145443916321,
      "learning_rate": 0.00021768542782662082,
      "loss": 2.1015,
      "step": 135586
    },
    {
      "epoch": 1.77,
      "grad_norm": 0.9931457042694092,
      "learning_rate": 0.000217681494264592,
      "loss": 2.4805,
      "step": 135587
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.060668706893921,
      "learning_rate": 0.0002176775607178677,
      "loss": 2.0498,
      "step": 135588
    },
    {
      "epoch": 1.77,
      "grad_norm": 0.9970547556877136,
      "learning_rate": 0.00021767362718644881,
      "loss": 1.8708,
      "step": 135589
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.0821510553359985,
      "learning_rate": 0.0002176696936703359,
      "loss": 2.3506,
      "step": 135590
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.4687496423721313,
      "learning_rate": 0.00021766576016952975,
      "loss": 2.1512,
      "step": 135591
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.0446592569351196,
      "learning_rate": 0.00021766182668403103,
      "loss": 2.1189,
      "step": 135592
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.0774812698364258,
      "learning_rate": 0.00021765789321384052,
      "loss": 2.0528,
      "step": 135593
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.1038812398910522,
      "learning_rate": 0.00021765395975895902,
      "loss": 2.1333,
      "step": 135594
    },
    {
      "epoch": 1.77,
      "grad_norm": 0.9440119862556458,
      "learning_rate": 0.00021765002631938716,
      "loss": 2.0711,
      "step": 135595
    },
    {
      "epoch": 1.77,
      "grad_norm": 0.9890345335006714,
      "learning_rate": 0.00021764609289512586,
      "loss": 2.2727,
      "step": 135596
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.0200823545455933,
      "learning_rate": 0.00021764215948617558,
      "loss": 2.4545,
      "step": 135597
    },
    {
      "epoch": 1.77,
      "grad_norm": 0.9945034980773926,
      "learning_rate": 0.00021763822609253714,
      "loss": 2.0094,
      "step": 135598
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.2294965982437134,
      "learning_rate": 0.00021763429271421135,
      "loss": 2.2128,
      "step": 135599
    },
    {
      "epoch": 1.77,
      "grad_norm": 0.9731747508049011,
      "learning_rate": 0.00021763035935119888,
      "loss": 1.9595,
      "step": 135600
    },
    {
      "epoch": 1.77,
      "grad_norm": 0.9887920618057251,
      "learning_rate": 0.00021762642600350052,
      "loss": 1.9041,
      "step": 135601
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.1306719779968262,
      "learning_rate": 0.00021762249267111704,
      "loss": 2.1025,
      "step": 135602
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.0899415016174316,
      "learning_rate": 0.000217618559354049,
      "loss": 2.2672,
      "step": 135603
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.0590547323226929,
      "learning_rate": 0.0002176146260522972,
      "loss": 1.7814,
      "step": 135604
    },
    {
      "epoch": 1.77,
      "grad_norm": 0.9979059100151062,
      "learning_rate": 0.00021761069276586242,
      "loss": 1.8899,
      "step": 135605
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.0135990381240845,
      "learning_rate": 0.00021760675949474532,
      "loss": 2.1743,
      "step": 135606
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.087722897529602,
      "learning_rate": 0.00021760282623894673,
      "loss": 2.3466,
      "step": 135607
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.0240180492401123,
      "learning_rate": 0.00021759889299846743,
      "loss": 2.3031,
      "step": 135608
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.0861036777496338,
      "learning_rate": 0.0002175949597733079,
      "loss": 2.1756,
      "step": 135609
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.0035996437072754,
      "learning_rate": 0.0002175910265634691,
      "loss": 2.0798,
      "step": 135610
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.0379502773284912,
      "learning_rate": 0.0002175870933689516,
      "loss": 2.1632,
      "step": 135611
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.0702048540115356,
      "learning_rate": 0.00021758316018975627,
      "loss": 2.195,
      "step": 135612
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.031187891960144,
      "learning_rate": 0.0002175792270258838,
      "loss": 1.9839,
      "step": 135613
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.0771422386169434,
      "learning_rate": 0.000217575293877335,
      "loss": 2.371,
      "step": 135614
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.0391616821289062,
      "learning_rate": 0.0002175713607441104,
      "loss": 1.931,
      "step": 135615
    },
    {
      "epoch": 1.77,
      "grad_norm": 0.9416065216064453,
      "learning_rate": 0.00021756742762621084,
      "loss": 1.9246,
      "step": 135616
    },
    {
      "epoch": 1.77,
      "grad_norm": 0.98027104139328,
      "learning_rate": 0.00021756349452363698,
      "loss": 2.1884,
      "step": 135617
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.0195212364196777,
      "learning_rate": 0.0002175595614363897,
      "loss": 2.2199,
      "step": 135618
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.1385432481765747,
      "learning_rate": 0.00021755562836446961,
      "loss": 2.1486,
      "step": 135619
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.0127657651901245,
      "learning_rate": 0.00021755169530787767,
      "loss": 2.174,
      "step": 135620
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.011351466178894,
      "learning_rate": 0.00021754776226661429,
      "loss": 1.9596,
      "step": 135621
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.0220516920089722,
      "learning_rate": 0.00021754382924068027,
      "loss": 2.0383,
      "step": 135622
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.000426173210144,
      "learning_rate": 0.00021753989623007647,
      "loss": 1.9763,
      "step": 135623
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.0102980136871338,
      "learning_rate": 0.00021753596323480356,
      "loss": 2.186,
      "step": 135624
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.033815860748291,
      "learning_rate": 0.00021753203025486224,
      "loss": 2.1294,
      "step": 135625
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.0057648420333862,
      "learning_rate": 0.0002175280972902534,
      "loss": 1.8763,
      "step": 135626
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.0783863067626953,
      "learning_rate": 0.0002175241643409775,
      "loss": 1.9455,
      "step": 135627
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.0628308057785034,
      "learning_rate": 0.00021752023140703548,
      "loss": 2.1555,
      "step": 135628
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.0066015720367432,
      "learning_rate": 0.00021751629848842797,
      "loss": 2.165,
      "step": 135629
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.0831667184829712,
      "learning_rate": 0.0002175123655851557,
      "loss": 1.8833,
      "step": 135630
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.0237624645233154,
      "learning_rate": 0.00021750843269721952,
      "loss": 2.1592,
      "step": 135631
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.066193699836731,
      "learning_rate": 0.00021750449982462013,
      "loss": 2.1338,
      "step": 135632
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.015114426612854,
      "learning_rate": 0.00021750056696735813,
      "loss": 2.3356,
      "step": 135633
    },
    {
      "epoch": 1.77,
      "grad_norm": 0.9606073498725891,
      "learning_rate": 0.0002174966341254343,
      "loss": 1.9785,
      "step": 135634
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.0200831890106201,
      "learning_rate": 0.0002174927012988494,
      "loss": 2.2071,
      "step": 135635
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.1039767265319824,
      "learning_rate": 0.0002174887684876042,
      "loss": 2.0169,
      "step": 135636
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.0877496004104614,
      "learning_rate": 0.0002174848356916994,
      "loss": 2.1519,
      "step": 135637
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.0509932041168213,
      "learning_rate": 0.0002174809029111358,
      "loss": 2.193,
      "step": 135638
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.0114811658859253,
      "learning_rate": 0.000217476970145914,
      "loss": 2.0558,
      "step": 135639
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.1372729539871216,
      "learning_rate": 0.00021747303739603474,
      "loss": 2.1521,
      "step": 135640
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.0186312198638916,
      "learning_rate": 0.00021746910466149882,
      "loss": 1.9851,
      "step": 135641
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.1029720306396484,
      "learning_rate": 0.00021746517194230693,
      "loss": 2.1581,
      "step": 135642
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.052514672279358,
      "learning_rate": 0.00021746123923845986,
      "loss": 2.0743,
      "step": 135643
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.0540910959243774,
      "learning_rate": 0.0002174573065499584,
      "loss": 2.138,
      "step": 135644
    },
    {
      "epoch": 1.77,
      "grad_norm": 0.9918733239173889,
      "learning_rate": 0.00021745337387680305,
      "loss": 2.0308,
      "step": 135645
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.1046653985977173,
      "learning_rate": 0.00021744944121899467,
      "loss": 2.183,
      "step": 135646
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.0607779026031494,
      "learning_rate": 0.00021744550857653408,
      "loss": 2.0342,
      "step": 135647
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.177730917930603,
      "learning_rate": 0.00021744157594942183,
      "loss": 2.2283,
      "step": 135648
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.1762566566467285,
      "learning_rate": 0.00021743764333765882,
      "loss": 2.1999,
      "step": 135649
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.0961612462997437,
      "learning_rate": 0.0002174337107412457,
      "loss": 2.0742,
      "step": 135650
    },
    {
      "epoch": 1.77,
      "grad_norm": 0.9948736429214478,
      "learning_rate": 0.00021742977816018326,
      "loss": 2.1473,
      "step": 135651
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.0886017084121704,
      "learning_rate": 0.0002174258455944721,
      "loss": 1.979,
      "step": 135652
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.4150164127349854,
      "learning_rate": 0.00021742191304411308,
      "loss": 1.9375,
      "step": 135653
    },
    {
      "epoch": 1.77,
      "grad_norm": 0.972496747970581,
      "learning_rate": 0.00021741798050910685,
      "loss": 1.9895,
      "step": 135654
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.018628478050232,
      "learning_rate": 0.00021741404798945416,
      "loss": 2.0093,
      "step": 135655
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.0722700357437134,
      "learning_rate": 0.00021741011548515575,
      "loss": 1.6863,
      "step": 135656
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.1215051412582397,
      "learning_rate": 0.00021740618299621243,
      "loss": 2.1887,
      "step": 135657
    },
    {
      "epoch": 1.77,
      "grad_norm": 0.9913744926452637,
      "learning_rate": 0.00021740225052262484,
      "loss": 2.0752,
      "step": 135658
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.0551077127456665,
      "learning_rate": 0.00021739831806439377,
      "loss": 2.2101,
      "step": 135659
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.0372995138168335,
      "learning_rate": 0.00021739438562151984,
      "loss": 2.0377,
      "step": 135660
    },
    {
      "epoch": 1.77,
      "grad_norm": 0.9707506895065308,
      "learning_rate": 0.00021739045319400387,
      "loss": 1.9398,
      "step": 135661
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.088301658630371,
      "learning_rate": 0.00021738652078184657,
      "loss": 2.0966,
      "step": 135662
    },
    {
      "epoch": 1.77,
      "grad_norm": 0.9795096516609192,
      "learning_rate": 0.0002173825883850487,
      "loss": 1.8936,
      "step": 135663
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.038788080215454,
      "learning_rate": 0.00021737865600361095,
      "loss": 1.9679,
      "step": 135664
    },
    {
      "epoch": 1.77,
      "grad_norm": 0.9473785161972046,
      "learning_rate": 0.00021737472363753403,
      "loss": 2.0471,
      "step": 135665
    },
    {
      "epoch": 1.77,
      "grad_norm": 0.9145841002464294,
      "learning_rate": 0.00021737079128681883,
      "loss": 2.0092,
      "step": 135666
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.166228175163269,
      "learning_rate": 0.00021736685895146581,
      "loss": 2.2823,
      "step": 135667
    },
    {
      "epoch": 1.77,
      "grad_norm": 0.9193062782287598,
      "learning_rate": 0.00021736292663147592,
      "loss": 1.7299,
      "step": 135668
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.0918415784835815,
      "learning_rate": 0.00021735899432684988,
      "loss": 2.2307,
      "step": 135669
    },
    {
      "epoch": 1.77,
      "grad_norm": 0.9962372183799744,
      "learning_rate": 0.00021735506203758825,
      "loss": 2.0244,
      "step": 135670
    },
    {
      "epoch": 1.77,
      "grad_norm": 0.9456974267959595,
      "learning_rate": 0.00021735112976369197,
      "loss": 2.0181,
      "step": 135671
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.0154786109924316,
      "learning_rate": 0.00021734719750516156,
      "loss": 1.9919,
      "step": 135672
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.0050790309906006,
      "learning_rate": 0.00021734326526199802,
      "loss": 1.9513,
      "step": 135673
    },
    {
      "epoch": 1.77,
      "grad_norm": 0.9936054944992065,
      "learning_rate": 0.00021733933303420185,
      "loss": 2.1481,
      "step": 135674
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.1293954849243164,
      "learning_rate": 0.00021733540082177386,
      "loss": 2.1088,
      "step": 135675
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.0740532875061035,
      "learning_rate": 0.00021733146862471474,
      "loss": 2.3093,
      "step": 135676
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.038700819015503,
      "learning_rate": 0.00021732753644302527,
      "loss": 2.0504,
      "step": 135677
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.3428881168365479,
      "learning_rate": 0.00021732360427670618,
      "loss": 1.9252,
      "step": 135678
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.006835699081421,
      "learning_rate": 0.0002173196721257582,
      "loss": 2.2067,
      "step": 135679
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.0679799318313599,
      "learning_rate": 0.00021731573999018213,
      "loss": 2.0535,
      "step": 135680
    },
    {
      "epoch": 1.77,
      "grad_norm": 0.9566792845726013,
      "learning_rate": 0.00021731180786997858,
      "loss": 2.0443,
      "step": 135681
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.1816728115081787,
      "learning_rate": 0.00021730787576514828,
      "loss": 2.3069,
      "step": 135682
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.1092840433120728,
      "learning_rate": 0.00021730394367569205,
      "loss": 2.1847,
      "step": 135683
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.0512257814407349,
      "learning_rate": 0.0002173000116016105,
      "loss": 1.9183,
      "step": 135684
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.06879460811615,
      "learning_rate": 0.00021729607954290448,
      "loss": 2.2655,
      "step": 135685
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.0604093074798584,
      "learning_rate": 0.00021729214749957465,
      "loss": 2.1587,
      "step": 135686
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.2793798446655273,
      "learning_rate": 0.00021728821547162196,
      "loss": 2.1699,
      "step": 135687
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.0140708684921265,
      "learning_rate": 0.0002172842834590468,
      "loss": 2.0273,
      "step": 135688
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.1187571287155151,
      "learning_rate": 0.00021728035146185003,
      "loss": 2.0978,
      "step": 135689
    },
    {
      "epoch": 1.77,
      "grad_norm": 0.9906126856803894,
      "learning_rate": 0.0002172764194800324,
      "loss": 1.8925,
      "step": 135690
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.0005214214324951,
      "learning_rate": 0.00021727248751359467,
      "loss": 2.2932,
      "step": 135691
    },
    {
      "epoch": 1.77,
      "grad_norm": 0.9714006185531616,
      "learning_rate": 0.00021726855556253752,
      "loss": 2.1314,
      "step": 135692
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.265862226486206,
      "learning_rate": 0.00021726462362686187,
      "loss": 2.2691,
      "step": 135693
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.100672960281372,
      "learning_rate": 0.00021726069170656815,
      "loss": 2.3437,
      "step": 135694
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.2209959030151367,
      "learning_rate": 0.00021725675980165723,
      "loss": 2.1689,
      "step": 135695
    },
    {
      "epoch": 1.77,
      "grad_norm": 0.9928244948387146,
      "learning_rate": 0.00021725282791212983,
      "loss": 2.0267,
      "step": 135696
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.0664316415786743,
      "learning_rate": 0.0002172488960379867,
      "loss": 2.1206,
      "step": 135697
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.0058451890945435,
      "learning_rate": 0.00021724496417922855,
      "loss": 2.0832,
      "step": 135698
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.0804774761199951,
      "learning_rate": 0.00021724103233585626,
      "loss": 2.04,
      "step": 135699
    },
    {
      "epoch": 1.77,
      "grad_norm": 0.9946339726448059,
      "learning_rate": 0.00021723710050787032,
      "loss": 2.0382,
      "step": 135700
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.0765823125839233,
      "learning_rate": 0.0002172331686952715,
      "loss": 2.1396,
      "step": 135701
    },
    {
      "epoch": 1.77,
      "grad_norm": 0.9459877610206604,
      "learning_rate": 0.00021722923689806067,
      "loss": 2.0521,
      "step": 135702
    },
    {
      "epoch": 1.77,
      "grad_norm": 0.9540590047836304,
      "learning_rate": 0.0002172253051162384,
      "loss": 1.9302,
      "step": 135703
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.0334160327911377,
      "learning_rate": 0.00021722137334980558,
      "loss": 2.1348,
      "step": 135704
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.0518102645874023,
      "learning_rate": 0.00021721744159876297,
      "loss": 2.2127,
      "step": 135705
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.0647333860397339,
      "learning_rate": 0.00021721350986311113,
      "loss": 2.0773,
      "step": 135706
    },
    {
      "epoch": 1.77,
      "grad_norm": 0.9482470154762268,
      "learning_rate": 0.0002172095781428508,
      "loss": 2.015,
      "step": 135707
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.0619999170303345,
      "learning_rate": 0.00021720564643798278,
      "loss": 2.1904,
      "step": 135708
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.0421717166900635,
      "learning_rate": 0.00021720171474850779,
      "loss": 2.179,
      "step": 135709
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.061340093612671,
      "learning_rate": 0.0002171977830744266,
      "loss": 1.9116,
      "step": 135710
    },
    {
      "epoch": 1.77,
      "grad_norm": 0.9718552827835083,
      "learning_rate": 0.00021719385141573996,
      "loss": 2.1673,
      "step": 135711
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.1463713645935059,
      "learning_rate": 0.0002171899197724485,
      "loss": 1.73,
      "step": 135712
    },
    {
      "epoch": 1.77,
      "grad_norm": 0.988112211227417,
      "learning_rate": 0.00021718598814455292,
      "loss": 2.0819,
      "step": 135713
    },
    {
      "epoch": 1.77,
      "grad_norm": 0.9374279379844666,
      "learning_rate": 0.00021718205653205408,
      "loss": 1.9696,
      "step": 135714
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.0940065383911133,
      "learning_rate": 0.00021717812493495268,
      "loss": 2.2599,
      "step": 135715
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.0948387384414673,
      "learning_rate": 0.0002171741933532494,
      "loss": 2.0004,
      "step": 135716
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.1124794483184814,
      "learning_rate": 0.0002171702617869451,
      "loss": 2.4862,
      "step": 135717
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.0320749282836914,
      "learning_rate": 0.00021716633023604034,
      "loss": 1.9685,
      "step": 135718
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.110365867614746,
      "learning_rate": 0.00021716239870053585,
      "loss": 2.118,
      "step": 135719
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.1391844749450684,
      "learning_rate": 0.0002171584671804325,
      "loss": 1.9922,
      "step": 135720
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.124731421470642,
      "learning_rate": 0.0002171545356757309,
      "loss": 2.1214,
      "step": 135721
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.1428613662719727,
      "learning_rate": 0.0002171506041864319,
      "loss": 2.0165,
      "step": 135722
    },
    {
      "epoch": 1.77,
      "grad_norm": 0.9406424760818481,
      "learning_rate": 0.00021714667271253613,
      "loss": 2.0609,
      "step": 135723
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.0832557678222656,
      "learning_rate": 0.00021714274125404448,
      "loss": 2.1294,
      "step": 135724
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.0981284379959106,
      "learning_rate": 0.00021713880981095746,
      "loss": 1.9232,
      "step": 135725
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.3030256032943726,
      "learning_rate": 0.00021713487838327585,
      "loss": 2.1857,
      "step": 135726
    },
    {
      "epoch": 1.77,
      "grad_norm": 0.9740003347396851,
      "learning_rate": 0.00021713094697100045,
      "loss": 2.0409,
      "step": 135727
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.0692321062088013,
      "learning_rate": 0.000217127015574132,
      "loss": 2.1562,
      "step": 135728
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.0404930114746094,
      "learning_rate": 0.00021712308419267117,
      "loss": 1.8328,
      "step": 135729
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.0618079900741577,
      "learning_rate": 0.00021711915282661887,
      "loss": 2.1936,
      "step": 135730
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.130470633506775,
      "learning_rate": 0.00021711522147597556,
      "loss": 2.2813,
      "step": 135731
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.0643798112869263,
      "learning_rate": 0.00021711129014074208,
      "loss": 2.2459,
      "step": 135732
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.5660758018493652,
      "learning_rate": 0.0002171073588209192,
      "loss": 1.8783,
      "step": 135733
    },
    {
      "epoch": 1.77,
      "grad_norm": 0.9779777526855469,
      "learning_rate": 0.0002171034275165076,
      "loss": 1.9528,
      "step": 135734
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.0991040468215942,
      "learning_rate": 0.00021709949622750803,
      "loss": 1.9705,
      "step": 135735
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.1048158407211304,
      "learning_rate": 0.0002170955649539214,
      "loss": 1.9986,
      "step": 135736
    },
    {
      "epoch": 1.77,
      "grad_norm": 0.9873899221420288,
      "learning_rate": 0.00021709163369574807,
      "loss": 2.0509,
      "step": 135737
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.092273235321045,
      "learning_rate": 0.00021708770245298904,
      "loss": 2.1137,
      "step": 135738
    },
    {
      "epoch": 1.77,
      "grad_norm": 0.9996558427810669,
      "learning_rate": 0.00021708377122564495,
      "loss": 2.0579,
      "step": 135739
    },
    {
      "epoch": 1.77,
      "grad_norm": 0.9964243173599243,
      "learning_rate": 0.00021707984001371653,
      "loss": 1.9766,
      "step": 135740
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.0019038915634155,
      "learning_rate": 0.00021707590881720455,
      "loss": 2.1446,
      "step": 135741
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.0287988185882568,
      "learning_rate": 0.00021707197763610979,
      "loss": 1.9927,
      "step": 135742
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.032765507698059,
      "learning_rate": 0.00021706804647043292,
      "loss": 2.0716,
      "step": 135743
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.1188876628875732,
      "learning_rate": 0.0002170641153201746,
      "loss": 2.2497,
      "step": 135744
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.0493615865707397,
      "learning_rate": 0.00021706018418533562,
      "loss": 2.1229,
      "step": 135745
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.1133980751037598,
      "learning_rate": 0.00021705625306591672,
      "loss": 1.8869,
      "step": 135746
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.023828387260437,
      "learning_rate": 0.0002170523219619186,
      "loss": 2.0754,
      "step": 135747
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.0317378044128418,
      "learning_rate": 0.00021704839087334214,
      "loss": 1.9319,
      "step": 135748
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.0915265083312988,
      "learning_rate": 0.00021704445980018784,
      "loss": 1.8173,
      "step": 135749
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.0711208581924438,
      "learning_rate": 0.00021704052874245655,
      "loss": 2.0533,
      "step": 135750
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.130291223526001,
      "learning_rate": 0.00021703659770014907,
      "loss": 2.0858,
      "step": 135751
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.006693720817566,
      "learning_rate": 0.00021703266667326597,
      "loss": 1.8854,
      "step": 135752
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.0540003776550293,
      "learning_rate": 0.00021702873566180805,
      "loss": 2.1202,
      "step": 135753
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.083337664604187,
      "learning_rate": 0.00021702480466577616,
      "loss": 2.098,
      "step": 135754
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.1162786483764648,
      "learning_rate": 0.00021702087368517087,
      "loss": 2.1839,
      "step": 135755
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.1163442134857178,
      "learning_rate": 0.00021701694271999291,
      "loss": 2.0972,
      "step": 135756
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.0939092636108398,
      "learning_rate": 0.0002170130117702431,
      "loss": 1.9745,
      "step": 135757
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.1524658203125,
      "learning_rate": 0.0002170090808359222,
      "loss": 2.1343,
      "step": 135758
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.2118968963623047,
      "learning_rate": 0.00021700514991703082,
      "loss": 2.2569,
      "step": 135759
    },
    {
      "epoch": 1.77,
      "grad_norm": 0.9485636949539185,
      "learning_rate": 0.00021700121901356982,
      "loss": 2.2931,
      "step": 135760
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.1266634464263916,
      "learning_rate": 0.00021699728812553973,
      "loss": 2.1053,
      "step": 135761
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.19871985912323,
      "learning_rate": 0.0002169933572529415,
      "loss": 2.1948,
      "step": 135762
    },
    {
      "epoch": 1.77,
      "grad_norm": 0.9685812592506409,
      "learning_rate": 0.00021698942639577574,
      "loss": 1.93,
      "step": 135763
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.0481144189834595,
      "learning_rate": 0.00021698549555404316,
      "loss": 2.0011,
      "step": 135764
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.2146570682525635,
      "learning_rate": 0.00021698156472774467,
      "loss": 2.2674,
      "step": 135765
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.0574476718902588,
      "learning_rate": 0.00021697763391688085,
      "loss": 2.1702,
      "step": 135766
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.0369676351547241,
      "learning_rate": 0.00021697370312145242,
      "loss": 2.2359,
      "step": 135767
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.1509453058242798,
      "learning_rate": 0.00021696977234146012,
      "loss": 2.0632,
      "step": 135768
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.1944791078567505,
      "learning_rate": 0.00021696584157690474,
      "loss": 2.1107,
      "step": 135769
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.095984697341919,
      "learning_rate": 0.00021696191082778694,
      "loss": 2.0895,
      "step": 135770
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.0129486322402954,
      "learning_rate": 0.00021695798009410745,
      "loss": 1.9807,
      "step": 135771
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.0133000612258911,
      "learning_rate": 0.00021695404937586722,
      "loss": 1.8161,
      "step": 135772
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.0744378566741943,
      "learning_rate": 0.0002169501186730667,
      "loss": 2.0168,
      "step": 135773
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.1860990524291992,
      "learning_rate": 0.00021694618798570668,
      "loss": 2.0569,
      "step": 135774
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.0173484086990356,
      "learning_rate": 0.00021694225731378792,
      "loss": 1.8269,
      "step": 135775
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.0856168270111084,
      "learning_rate": 0.0002169383266573112,
      "loss": 2.1233,
      "step": 135776
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.0611330270767212,
      "learning_rate": 0.0002169343960162772,
      "loss": 2.1788,
      "step": 135777
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.232710838317871,
      "learning_rate": 0.0002169304653906868,
      "loss": 2.1985,
      "step": 135778
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.1273475885391235,
      "learning_rate": 0.00021692653478054042,
      "loss": 2.1623,
      "step": 135779
    },
    {
      "epoch": 1.77,
      "grad_norm": 0.9806978106498718,
      "learning_rate": 0.00021692260418583898,
      "loss": 2.2751,
      "step": 135780
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.2796458005905151,
      "learning_rate": 0.00021691867360658324,
      "loss": 2.0516,
      "step": 135781
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.4593422412872314,
      "learning_rate": 0.00021691474304277384,
      "loss": 2.1452,
      "step": 135782
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.1810059547424316,
      "learning_rate": 0.0002169108124944116,
      "loss": 2.0069,
      "step": 135783
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.007555365562439,
      "learning_rate": 0.00021690688196149727,
      "loss": 1.9821,
      "step": 135784
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.237410306930542,
      "learning_rate": 0.00021690295144403143,
      "loss": 2.2836,
      "step": 135785
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.3033989667892456,
      "learning_rate": 0.0002168990209420149,
      "loss": 2.2476,
      "step": 135786
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.23131263256073,
      "learning_rate": 0.00021689509045544843,
      "loss": 1.8819,
      "step": 135787
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.100912094116211,
      "learning_rate": 0.00021689115998433273,
      "loss": 2.0104,
      "step": 135788
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.2132649421691895,
      "learning_rate": 0.00021688722952866853,
      "loss": 2.0422,
      "step": 135789
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.736803650856018,
      "learning_rate": 0.00021688329908845657,
      "loss": 2.0338,
      "step": 135790
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.122305154800415,
      "learning_rate": 0.00021687936866369764,
      "loss": 1.8526,
      "step": 135791
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.1053893566131592,
      "learning_rate": 0.00021687543825439236,
      "loss": 2.0961,
      "step": 135792
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.1059995889663696,
      "learning_rate": 0.00021687150786054143,
      "loss": 2.0697,
      "step": 135793
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.1559655666351318,
      "learning_rate": 0.0002168675774821457,
      "loss": 2.1542,
      "step": 135794
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.5340534448623657,
      "learning_rate": 0.00021686364711920585,
      "loss": 2.085,
      "step": 135795
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.135600209236145,
      "learning_rate": 0.00021685971677172261,
      "loss": 2.0937,
      "step": 135796
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.0629533529281616,
      "learning_rate": 0.00021685578643969688,
      "loss": 2.1074,
      "step": 135797
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.0692566633224487,
      "learning_rate": 0.00021685185612312907,
      "loss": 2.0319,
      "step": 135798
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.2442681789398193,
      "learning_rate": 0.00021684792582202003,
      "loss": 2.1412,
      "step": 135799
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.3383736610412598,
      "learning_rate": 0.00021684399553637058,
      "loss": 2.1339,
      "step": 135800
    },
    {
      "epoch": 1.77,
      "grad_norm": 0.9808258414268494,
      "learning_rate": 0.0002168400652661814,
      "loss": 1.948,
      "step": 135801
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.2289555072784424,
      "learning_rate": 0.0002168361350114532,
      "loss": 2.1909,
      "step": 135802
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.179124116897583,
      "learning_rate": 0.0002168322047721869,
      "loss": 2.0713,
      "step": 135803
    },
    {
      "epoch": 1.77,
      "grad_norm": 0.9647238254547119,
      "learning_rate": 0.0002168282745483829,
      "loss": 1.8144,
      "step": 135804
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.039314866065979,
      "learning_rate": 0.0002168243443400421,
      "loss": 1.9318,
      "step": 135805
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.0230532884597778,
      "learning_rate": 0.00021682041414716524,
      "loss": 1.9961,
      "step": 135806
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.1247119903564453,
      "learning_rate": 0.000216816483969753,
      "loss": 1.956,
      "step": 135807
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.0930140018463135,
      "learning_rate": 0.00021681255380780618,
      "loss": 2.3215,
      "step": 135808
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.0105663537979126,
      "learning_rate": 0.00021680862366132557,
      "loss": 2.3382,
      "step": 135809
    },
    {
      "epoch": 1.77,
      "grad_norm": 0.9046400785446167,
      "learning_rate": 0.00021680469353031173,
      "loss": 1.9614,
      "step": 135810
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.0167478322982788,
      "learning_rate": 0.00021680076341476542,
      "loss": 2.046,
      "step": 135811
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.023941159248352,
      "learning_rate": 0.00021679683331468744,
      "loss": 2.169,
      "step": 135812
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.1036285161972046,
      "learning_rate": 0.0002167929032300785,
      "loss": 2.0432,
      "step": 135813
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.0100524425506592,
      "learning_rate": 0.00021678897316093935,
      "loss": 2.0498,
      "step": 135814
    },
    {
      "epoch": 1.77,
      "grad_norm": 0.9933690428733826,
      "learning_rate": 0.0002167850431072708,
      "loss": 2.3567,
      "step": 135815
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.1085065603256226,
      "learning_rate": 0.00021678111306907337,
      "loss": 1.9364,
      "step": 135816
    },
    {
      "epoch": 1.77,
      "grad_norm": 0.9886847138404846,
      "learning_rate": 0.00021677718304634788,
      "loss": 1.8346,
      "step": 135817
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.049789309501648,
      "learning_rate": 0.00021677325303909512,
      "loss": 2.0537,
      "step": 135818
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.1151736974716187,
      "learning_rate": 0.00021676932304731572,
      "loss": 1.9572,
      "step": 135819
    },
    {
      "epoch": 1.77,
      "grad_norm": 0.9481106400489807,
      "learning_rate": 0.00021676539307101053,
      "loss": 2.2266,
      "step": 135820
    },
    {
      "epoch": 1.77,
      "grad_norm": 0.9867085814476013,
      "learning_rate": 0.00021676146311018032,
      "loss": 1.9726,
      "step": 135821
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.0132302045822144,
      "learning_rate": 0.00021675753316482561,
      "loss": 2.3268,
      "step": 135822
    },
    {
      "epoch": 1.77,
      "grad_norm": 0.9848853945732117,
      "learning_rate": 0.00021675360323494724,
      "loss": 2.1876,
      "step": 135823
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.148033618927002,
      "learning_rate": 0.00021674967332054596,
      "loss": 2.0014,
      "step": 135824
    },
    {
      "epoch": 1.77,
      "grad_norm": 0.9699044227600098,
      "learning_rate": 0.00021674574342162248,
      "loss": 1.8856,
      "step": 135825
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.0930674076080322,
      "learning_rate": 0.00021674181353817756,
      "loss": 2.2053,
      "step": 135826
    },
    {
      "epoch": 1.77,
      "grad_norm": 0.9618011116981506,
      "learning_rate": 0.00021673788367021195,
      "loss": 2.2663,
      "step": 135827
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.0752679109573364,
      "learning_rate": 0.0002167339538177263,
      "loss": 2.1918,
      "step": 135828
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.069198727607727,
      "learning_rate": 0.00021673002398072137,
      "loss": 2.0135,
      "step": 135829
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.0045795440673828,
      "learning_rate": 0.0002167260941591979,
      "loss": 2.338,
      "step": 135830
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.009856939315796,
      "learning_rate": 0.0002167221643531566,
      "loss": 1.8194,
      "step": 135831
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.1274244785308838,
      "learning_rate": 0.00021671823456259823,
      "loss": 2.1104,
      "step": 135832
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.0760773420333862,
      "learning_rate": 0.0002167143047875236,
      "loss": 2.0772,
      "step": 135833
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.0725525617599487,
      "learning_rate": 0.00021671037502793327,
      "loss": 2.1394,
      "step": 135834
    },
    {
      "epoch": 1.77,
      "grad_norm": 0.977462112903595,
      "learning_rate": 0.0002167064452838281,
      "loss": 2.1157,
      "step": 135835
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.1400480270385742,
      "learning_rate": 0.0002167025155552087,
      "loss": 2.1577,
      "step": 135836
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.3206462860107422,
      "learning_rate": 0.00021669858584207595,
      "loss": 2.0332,
      "step": 135837
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.0196442604064941,
      "learning_rate": 0.00021669465614443042,
      "loss": 2.1614,
      "step": 135838
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.0609502792358398,
      "learning_rate": 0.00021669072646227303,
      "loss": 2.0918,
      "step": 135839
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.2860068082809448,
      "learning_rate": 0.0002166867967956043,
      "loss": 2.0823,
      "step": 135840
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.0617731809616089,
      "learning_rate": 0.0002166828671444251,
      "loss": 2.0198,
      "step": 135841
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.0383986234664917,
      "learning_rate": 0.00021667893750873619,
      "loss": 2.0675,
      "step": 135842
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.0443270206451416,
      "learning_rate": 0.00021667500788853823,
      "loss": 2.1241,
      "step": 135843
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.000495195388794,
      "learning_rate": 0.00021667107828383185,
      "loss": 1.954,
      "step": 135844
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.072882890701294,
      "learning_rate": 0.00021666714869461804,
      "loss": 2.0079,
      "step": 135845
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.1307426691055298,
      "learning_rate": 0.00021666321912089729,
      "loss": 2.0613,
      "step": 135846
    },
    {
      "epoch": 1.77,
      "grad_norm": 0.9596788287162781,
      "learning_rate": 0.0002166592895626704,
      "loss": 2.1407,
      "step": 135847
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.0654687881469727,
      "learning_rate": 0.00021665536001993813,
      "loss": 2.2274,
      "step": 135848
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.015281319618225,
      "learning_rate": 0.00021665143049270122,
      "loss": 2.192,
      "step": 135849
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.1368310451507568,
      "learning_rate": 0.00021664750098096045,
      "loss": 2.1881,
      "step": 135850
    },
    {
      "epoch": 1.77,
      "grad_norm": 0.9753738045692444,
      "learning_rate": 0.00021664357148471646,
      "loss": 1.9731,
      "step": 135851
    },
    {
      "epoch": 1.77,
      "grad_norm": 0.9454103112220764,
      "learning_rate": 0.00021663964200396995,
      "loss": 2.1321,
      "step": 135852
    },
    {
      "epoch": 1.77,
      "grad_norm": 0.9671462178230286,
      "learning_rate": 0.0002166357125387217,
      "loss": 2.1427,
      "step": 135853
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.1691443920135498,
      "learning_rate": 0.00021663178308897244,
      "loss": 1.9727,
      "step": 135854
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.148112177848816,
      "learning_rate": 0.00021662785365472292,
      "loss": 2.0246,
      "step": 135855
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.0687143802642822,
      "learning_rate": 0.00021662392423597385,
      "loss": 2.1091,
      "step": 135856
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.0058977603912354,
      "learning_rate": 0.00021661999483272603,
      "loss": 2.1177,
      "step": 135857
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.0425288677215576,
      "learning_rate": 0.00021661606544498012,
      "loss": 2.2278,
      "step": 135858
    },
    {
      "epoch": 1.77,
      "grad_norm": 0.9453145265579224,
      "learning_rate": 0.0002166121360727368,
      "loss": 1.9945,
      "step": 135859
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.1428593397140503,
      "learning_rate": 0.00021660820671599688,
      "loss": 2.0468,
      "step": 135860
    },
    {
      "epoch": 1.77,
      "grad_norm": 0.9514070153236389,
      "learning_rate": 0.00021660427737476103,
      "loss": 1.9862,
      "step": 135861
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.1097323894500732,
      "learning_rate": 0.00021660034804903003,
      "loss": 2.1847,
      "step": 135862
    },
    {
      "epoch": 1.77,
      "grad_norm": 0.9980747699737549,
      "learning_rate": 0.00021659641873880461,
      "loss": 2.1505,
      "step": 135863
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.0824662446975708,
      "learning_rate": 0.00021659248944408562,
      "loss": 2.1073,
      "step": 135864
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.0517911911010742,
      "learning_rate": 0.0002165885601648735,
      "loss": 2.1368,
      "step": 135865
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.0323989391326904,
      "learning_rate": 0.00021658463090116917,
      "loss": 2.0219,
      "step": 135866
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.1458557844161987,
      "learning_rate": 0.00021658070165297334,
      "loss": 2.0789,
      "step": 135867
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.1595368385314941,
      "learning_rate": 0.0002165767724202867,
      "loss": 2.3032,
      "step": 135868
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.0536410808563232,
      "learning_rate": 0.00021657284320311006,
      "loss": 2.1181,
      "step": 135869
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.088119626045227,
      "learning_rate": 0.00021656891400144418,
      "loss": 2.0813,
      "step": 135870
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.0195492506027222,
      "learning_rate": 0.00021656498481528963,
      "loss": 2.146,
      "step": 135871
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.0334986448287964,
      "learning_rate": 0.00021656105564464722,
      "loss": 1.835,
      "step": 135872
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.0943976640701294,
      "learning_rate": 0.00021655712648951765,
      "loss": 2.2775,
      "step": 135873
    },
    {
      "epoch": 1.77,
      "grad_norm": 0.9761488437652588,
      "learning_rate": 0.0002165531973499017,
      "loss": 2.145,
      "step": 135874
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.0683308839797974,
      "learning_rate": 0.00021654926822580007,
      "loss": 1.9889,
      "step": 135875
    },
    {
      "epoch": 1.77,
      "grad_norm": 0.9583242535591125,
      "learning_rate": 0.00021654533911721365,
      "loss": 2.221,
      "step": 135876
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.0329102277755737,
      "learning_rate": 0.00021654141002414292,
      "loss": 2.035,
      "step": 135877
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.0976202487945557,
      "learning_rate": 0.00021653748094658872,
      "loss": 2.0265,
      "step": 135878
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.3881502151489258,
      "learning_rate": 0.00021653355188455174,
      "loss": 2.188,
      "step": 135879
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.0643887519836426,
      "learning_rate": 0.0002165296228380328,
      "loss": 2.157,
      "step": 135880
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.055454134941101,
      "learning_rate": 0.00021652569380703251,
      "loss": 2.0673,
      "step": 135881
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.1300921440124512,
      "learning_rate": 0.00021652176479155185,
      "loss": 2.1205,
      "step": 135882
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.4377150535583496,
      "learning_rate": 0.00021651783579159124,
      "loss": 2.1231,
      "step": 135883
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.0048850774765015,
      "learning_rate": 0.0002165139068071515,
      "loss": 2.0577,
      "step": 135884
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.0848008394241333,
      "learning_rate": 0.00021650997783823344,
      "loss": 2.2768,
      "step": 135885
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.4833831787109375,
      "learning_rate": 0.00021650604888483774,
      "loss": 1.8631,
      "step": 135886
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.0624103546142578,
      "learning_rate": 0.00021650211994696513,
      "loss": 2.0536,
      "step": 135887
    },
    {
      "epoch": 1.77,
      "grad_norm": 0.9528300762176514,
      "learning_rate": 0.0002164981910246165,
      "loss": 1.9578,
      "step": 135888
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.0791878700256348,
      "learning_rate": 0.00021649426211779228,
      "loss": 2.1837,
      "step": 135889
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.664742112159729,
      "learning_rate": 0.00021649033322649336,
      "loss": 2.0835,
      "step": 135890
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.4287174940109253,
      "learning_rate": 0.00021648640435072047,
      "loss": 2.0008,
      "step": 135891
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.051323652267456,
      "learning_rate": 0.00021648247549047432,
      "loss": 2.1239,
      "step": 135892
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.1280510425567627,
      "learning_rate": 0.00021647854664575566,
      "loss": 2.2138,
      "step": 135893
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.1050409078598022,
      "learning_rate": 0.00021647461781656538,
      "loss": 2.0238,
      "step": 135894
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.3583478927612305,
      "learning_rate": 0.00021647068900290385,
      "loss": 2.0631,
      "step": 135895
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.2474191188812256,
      "learning_rate": 0.00021646676020477205,
      "loss": 2.1693,
      "step": 135896
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.020856261253357,
      "learning_rate": 0.00021646283142217064,
      "loss": 2.2005,
      "step": 135897
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.1885132789611816,
      "learning_rate": 0.00021645890265510037,
      "loss": 1.9542,
      "step": 135898
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.0815509557724,
      "learning_rate": 0.00021645497390356195,
      "loss": 2.1411,
      "step": 135899
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.1019028425216675,
      "learning_rate": 0.0002164510451675563,
      "loss": 2.1981,
      "step": 135900
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.0043374300003052,
      "learning_rate": 0.0002164471164470838,
      "loss": 1.9226,
      "step": 135901
    },
    {
      "epoch": 1.77,
      "grad_norm": 0.9906362891197205,
      "learning_rate": 0.00021644318774214537,
      "loss": 1.9288,
      "step": 135902
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.027302861213684,
      "learning_rate": 0.0002164392590527417,
      "loss": 2.1379,
      "step": 135903
    },
    {
      "epoch": 1.77,
      "grad_norm": 0.93511563539505,
      "learning_rate": 0.00021643533037887363,
      "loss": 1.995,
      "step": 135904
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.042693853378296,
      "learning_rate": 0.00021643140172054176,
      "loss": 2.2154,
      "step": 135905
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.2399680614471436,
      "learning_rate": 0.00021642747307774697,
      "loss": 2.1238,
      "step": 135906
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.0691297054290771,
      "learning_rate": 0.0002164235444504898,
      "loss": 2.4419,
      "step": 135907
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.0835481882095337,
      "learning_rate": 0.0002164196158387711,
      "loss": 2.0817,
      "step": 135908
    },
    {
      "epoch": 1.77,
      "grad_norm": 0.9059840440750122,
      "learning_rate": 0.00021641568724259152,
      "loss": 1.9103,
      "step": 135909
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.0240392684936523,
      "learning_rate": 0.00021641175866195185,
      "loss": 2.0583,
      "step": 135910
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.0265474319458008,
      "learning_rate": 0.00021640783009685283,
      "loss": 2.1571,
      "step": 135911
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.0378656387329102,
      "learning_rate": 0.00021640390154729523,
      "loss": 2.1112,
      "step": 135912
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.041661024093628,
      "learning_rate": 0.00021639997301327966,
      "loss": 2.0841,
      "step": 135913
    },
    {
      "epoch": 1.77,
      "grad_norm": 0.9737194776535034,
      "learning_rate": 0.00021639604449480695,
      "loss": 1.7564,
      "step": 135914
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.0441948175430298,
      "learning_rate": 0.00021639211599187774,
      "loss": 1.9657,
      "step": 135915
    },
    {
      "epoch": 1.77,
      "grad_norm": 0.9493628144264221,
      "learning_rate": 0.00021638818750449284,
      "loss": 1.9444,
      "step": 135916
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.1005827188491821,
      "learning_rate": 0.0002163842590326529,
      "loss": 2.2232,
      "step": 135917
    },
    {
      "epoch": 1.77,
      "grad_norm": 0.9701583981513977,
      "learning_rate": 0.0002163803305763588,
      "loss": 1.9224,
      "step": 135918
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.0888882875442505,
      "learning_rate": 0.00021637640213561114,
      "loss": 2.223,
      "step": 135919
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.0334595441818237,
      "learning_rate": 0.0002163724737104107,
      "loss": 2.2552,
      "step": 135920
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.1522302627563477,
      "learning_rate": 0.0002163685453007582,
      "loss": 2.1289,
      "step": 135921
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.1309194564819336,
      "learning_rate": 0.00021636461690665432,
      "loss": 2.1019,
      "step": 135922
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.0312795639038086,
      "learning_rate": 0.00021636068852809985,
      "loss": 2.1643,
      "step": 135923
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.00039541721344,
      "learning_rate": 0.0002163567601650955,
      "loss": 2.0773,
      "step": 135924
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.1164517402648926,
      "learning_rate": 0.00021635283181764206,
      "loss": 2.1773,
      "step": 135925
    },
    {
      "epoch": 1.77,
      "grad_norm": 0.9634823203086853,
      "learning_rate": 0.00021634890348574016,
      "loss": 2.1934,
      "step": 135926
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.1136444807052612,
      "learning_rate": 0.0002163449751693906,
      "loss": 2.3217,
      "step": 135927
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.0297986268997192,
      "learning_rate": 0.00021634104686859407,
      "loss": 1.9725,
      "step": 135928
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.0703026056289673,
      "learning_rate": 0.00021633711858335132,
      "loss": 2.1567,
      "step": 135929
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.066812515258789,
      "learning_rate": 0.00021633319031366305,
      "loss": 2.2384,
      "step": 135930
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.1271271705627441,
      "learning_rate": 0.0002163292620595301,
      "loss": 2.1914,
      "step": 135931
    },
    {
      "epoch": 1.77,
      "grad_norm": 0.9106032252311707,
      "learning_rate": 0.00021632533382095304,
      "loss": 2.0877,
      "step": 135932
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.011900782585144,
      "learning_rate": 0.0002163214055979327,
      "loss": 1.9823,
      "step": 135933
    },
    {
      "epoch": 1.77,
      "grad_norm": 0.9754886031150818,
      "learning_rate": 0.00021631747739046987,
      "loss": 2.1057,
      "step": 135934
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.099348545074463,
      "learning_rate": 0.0002163135491985651,
      "loss": 1.8812,
      "step": 135935
    },
    {
      "epoch": 1.77,
      "grad_norm": 0.993520975112915,
      "learning_rate": 0.00021630962102221925,
      "loss": 2.2335,
      "step": 135936
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.0229966640472412,
      "learning_rate": 0.00021630569286143307,
      "loss": 2.1475,
      "step": 135937
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.103082299232483,
      "learning_rate": 0.00021630176471620718,
      "loss": 2.0514,
      "step": 135938
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.0779765844345093,
      "learning_rate": 0.00021629783658654237,
      "loss": 1.9938,
      "step": 135939
    },
    {
      "epoch": 1.77,
      "grad_norm": 0.9639029502868652,
      "learning_rate": 0.00021629390847243936,
      "loss": 2.3585,
      "step": 135940
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.114383339881897,
      "learning_rate": 0.00021628998037389894,
      "loss": 1.9047,
      "step": 135941
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.572630524635315,
      "learning_rate": 0.0002162860522909218,
      "loss": 2.4659,
      "step": 135942
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.0189839601516724,
      "learning_rate": 0.0002162821242235087,
      "loss": 2.1787,
      "step": 135943
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.03525972366333,
      "learning_rate": 0.00021627819617166025,
      "loss": 2.1955,
      "step": 135944
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.0928378105163574,
      "learning_rate": 0.00021627426813537726,
      "loss": 2.1327,
      "step": 135945
    },
    {
      "epoch": 1.77,
      "grad_norm": 0.9948609471321106,
      "learning_rate": 0.0002162703401146605,
      "loss": 2.2167,
      "step": 135946
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.1104485988616943,
      "learning_rate": 0.00021626641210951064,
      "loss": 2.2493,
      "step": 135947
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.0277878046035767,
      "learning_rate": 0.00021626248411992841,
      "loss": 1.9703,
      "step": 135948
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.0309604406356812,
      "learning_rate": 0.0002162585561459147,
      "loss": 2.2824,
      "step": 135949
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.0522202253341675,
      "learning_rate": 0.00021625462818747,
      "loss": 2.1874,
      "step": 135950
    },
    {
      "epoch": 1.77,
      "grad_norm": 0.9817264676094055,
      "learning_rate": 0.0002162507002445951,
      "loss": 2.2692,
      "step": 135951
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.3083221912384033,
      "learning_rate": 0.00021624677231729084,
      "loss": 1.9381,
      "step": 135952
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.0349496603012085,
      "learning_rate": 0.00021624284440555788,
      "loss": 2.1306,
      "step": 135953
    },
    {
      "epoch": 1.77,
      "grad_norm": 0.9726901650428772,
      "learning_rate": 0.0002162389165093969,
      "loss": 2.0895,
      "step": 135954
    },
    {
      "epoch": 1.77,
      "grad_norm": 0.8898950219154358,
      "learning_rate": 0.00021623498862880885,
      "loss": 2.0917,
      "step": 135955
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.0174981355667114,
      "learning_rate": 0.00021623106076379422,
      "loss": 2.0854,
      "step": 135956
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.094806432723999,
      "learning_rate": 0.00021622713291435375,
      "loss": 2.3055,
      "step": 135957
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.1086111068725586,
      "learning_rate": 0.00021622320508048822,
      "loss": 2.1586,
      "step": 135958
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.0622957944869995,
      "learning_rate": 0.00021621927726219844,
      "loss": 2.1073,
      "step": 135959
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.1087449789047241,
      "learning_rate": 0.00021621534945948507,
      "loss": 2.043,
      "step": 135960
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.0661590099334717,
      "learning_rate": 0.00021621142167234893,
      "loss": 1.9396,
      "step": 135961
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.0188809633255005,
      "learning_rate": 0.00021620749390079057,
      "loss": 1.8714,
      "step": 135962
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.0731654167175293,
      "learning_rate": 0.0002162035661448108,
      "loss": 2.2148,
      "step": 135963
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.0740710496902466,
      "learning_rate": 0.00021619963840441043,
      "loss": 2.264,
      "step": 135964
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.0193790197372437,
      "learning_rate": 0.00021619571067959002,
      "loss": 2.2025,
      "step": 135965
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.0411630868911743,
      "learning_rate": 0.00021619178297035052,
      "loss": 2.1394,
      "step": 135966
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.0225834846496582,
      "learning_rate": 0.0002161878552766926,
      "loss": 2.1395,
      "step": 135967
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.1165883541107178,
      "learning_rate": 0.00021618392759861681,
      "loss": 2.1133,
      "step": 135968
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.0798094272613525,
      "learning_rate": 0.00021617999993612405,
      "loss": 2.1304,
      "step": 135969
    },
    {
      "epoch": 1.77,
      "grad_norm": 0.9760639071464539,
      "learning_rate": 0.000216176072289215,
      "loss": 1.9678,
      "step": 135970
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.0259504318237305,
      "learning_rate": 0.00021617214465789045,
      "loss": 2.1568,
      "step": 135971
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.1029443740844727,
      "learning_rate": 0.000216168217042151,
      "loss": 2.0939,
      "step": 135972
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.149703025817871,
      "learning_rate": 0.00021616428944199763,
      "loss": 1.8983,
      "step": 135973
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.1982351541519165,
      "learning_rate": 0.00021616036185743078,
      "loss": 2.1826,
      "step": 135974
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.2330960035324097,
      "learning_rate": 0.00021615643428845127,
      "loss": 2.1535,
      "step": 135975
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.0778223276138306,
      "learning_rate": 0.0002161525067350599,
      "loss": 1.7955,
      "step": 135976
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.0999369621276855,
      "learning_rate": 0.00021614857919725729,
      "loss": 2.191,
      "step": 135977
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.0511049032211304,
      "learning_rate": 0.00021614465167504431,
      "loss": 2.0795,
      "step": 135978
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.1927087306976318,
      "learning_rate": 0.0002161407241684217,
      "loss": 2.2256,
      "step": 135979
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.0501924753189087,
      "learning_rate": 0.00021613679667739007,
      "loss": 1.9486,
      "step": 135980
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.017093300819397,
      "learning_rate": 0.00021613286920195012,
      "loss": 2.2047,
      "step": 135981
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.0296869277954102,
      "learning_rate": 0.00021612894174210265,
      "loss": 1.9714,
      "step": 135982
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.0513397455215454,
      "learning_rate": 0.00021612501429784842,
      "loss": 1.8068,
      "step": 135983
    },
    {
      "epoch": 1.77,
      "grad_norm": 0.9800319075584412,
      "learning_rate": 0.00021612108686918814,
      "loss": 2.1363,
      "step": 135984
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.0527901649475098,
      "learning_rate": 0.00021611715945612264,
      "loss": 2.0953,
      "step": 135985
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.0928921699523926,
      "learning_rate": 0.00021611323205865241,
      "loss": 1.9528,
      "step": 135986
    },
    {
      "epoch": 1.77,
      "grad_norm": 0.994875431060791,
      "learning_rate": 0.00021610930467677833,
      "loss": 2.2256,
      "step": 135987
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.0323610305786133,
      "learning_rate": 0.0002161053773105011,
      "loss": 2.1654,
      "step": 135988
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.0130105018615723,
      "learning_rate": 0.0002161014499598215,
      "loss": 2.2986,
      "step": 135989
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.0357786417007446,
      "learning_rate": 0.00021609752262474017,
      "loss": 2.0646,
      "step": 135990
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.0380685329437256,
      "learning_rate": 0.00021609359530525794,
      "loss": 2.1182,
      "step": 135991
    },
    {
      "epoch": 1.77,
      "grad_norm": 0.9587172269821167,
      "learning_rate": 0.0002160896680013756,
      "loss": 2.0945,
      "step": 135992
    },
    {
      "epoch": 1.77,
      "grad_norm": 0.9889723062515259,
      "learning_rate": 0.00021608574071309365,
      "loss": 1.9134,
      "step": 135993
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.060359001159668,
      "learning_rate": 0.00021608181344041292,
      "loss": 1.9214,
      "step": 135994
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.0747569799423218,
      "learning_rate": 0.00021607788618333422,
      "loss": 1.9687,
      "step": 135995
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.1806674003601074,
      "learning_rate": 0.00021607395894185816,
      "loss": 2.3243,
      "step": 135996
    },
    {
      "epoch": 1.77,
      "grad_norm": 0.975100040435791,
      "learning_rate": 0.00021607003171598561,
      "loss": 2.2006,
      "step": 135997
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.079131007194519,
      "learning_rate": 0.0002160661045057173,
      "loss": 2.0988,
      "step": 135998
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.1284629106521606,
      "learning_rate": 0.0002160621773110538,
      "loss": 2.0728,
      "step": 135999
    },
    {
      "epoch": 1.77,
      "grad_norm": 0.9955989122390747,
      "learning_rate": 0.00021605825013199586,
      "loss": 1.9571,
      "step": 136000
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.205385684967041,
      "learning_rate": 0.00021605432296854433,
      "loss": 2.3125,
      "step": 136001
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.027960181236267,
      "learning_rate": 0.00021605039582069989,
      "loss": 1.9734,
      "step": 136002
    },
    {
      "epoch": 1.77,
      "grad_norm": 0.9686245322227478,
      "learning_rate": 0.00021604646868846324,
      "loss": 2.0272,
      "step": 136003
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.0274770259857178,
      "learning_rate": 0.00021604254157183526,
      "loss": 2.0289,
      "step": 136004
    },
    {
      "epoch": 1.77,
      "grad_norm": 0.9804195761680603,
      "learning_rate": 0.00021603861447081645,
      "loss": 2.0596,
      "step": 136005
    },
    {
      "epoch": 1.77,
      "grad_norm": 0.9972621202468872,
      "learning_rate": 0.0002160346873854076,
      "loss": 1.9182,
      "step": 136006
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.2117444276809692,
      "learning_rate": 0.00021603076031560957,
      "loss": 2.3618,
      "step": 136007
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.1008740663528442,
      "learning_rate": 0.00021602683326142294,
      "loss": 2.1161,
      "step": 136008
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.059360384941101,
      "learning_rate": 0.00021602290622284854,
      "loss": 1.8857,
      "step": 136009
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.0633652210235596,
      "learning_rate": 0.00021601897919988709,
      "loss": 2.1181,
      "step": 136010
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.1240873336791992,
      "learning_rate": 0.00021601505219253926,
      "loss": 2.3252,
      "step": 136011
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.0556930303573608,
      "learning_rate": 0.0002160111252008059,
      "loss": 2.3299,
      "step": 136012
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.0539889335632324,
      "learning_rate": 0.00021600719822468755,
      "loss": 2.0751,
      "step": 136013
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.0325913429260254,
      "learning_rate": 0.00021600327126418512,
      "loss": 2.1434,
      "step": 136014
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.0246365070343018,
      "learning_rate": 0.00021599934431929918,
      "loss": 2.2189,
      "step": 136015
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.0537304878234863,
      "learning_rate": 0.00021599541739003065,
      "loss": 2.13,
      "step": 136016
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.105334997177124,
      "learning_rate": 0.0002159914904763801,
      "loss": 2.3137,
      "step": 136017
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.0245031118392944,
      "learning_rate": 0.0002159875635783483,
      "loss": 2.2021,
      "step": 136018
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.1025537252426147,
      "learning_rate": 0.0002159836366959361,
      "loss": 2.0064,
      "step": 136019
    },
    {
      "epoch": 1.77,
      "grad_norm": 0.9610938429832458,
      "learning_rate": 0.00021597970982914404,
      "loss": 2.0555,
      "step": 136020
    },
    {
      "epoch": 1.77,
      "grad_norm": 0.9999061226844788,
      "learning_rate": 0.00021597578297797294,
      "loss": 2.1838,
      "step": 136021
    },
    {
      "epoch": 1.77,
      "grad_norm": 0.9642597436904907,
      "learning_rate": 0.00021597185614242353,
      "loss": 2.0265,
      "step": 136022
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.0935540199279785,
      "learning_rate": 0.00021596792932249658,
      "loss": 2.2075,
      "step": 136023
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.1038960218429565,
      "learning_rate": 0.00021596400251819273,
      "loss": 1.9398,
      "step": 136024
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.2038493156433105,
      "learning_rate": 0.00021596007572951273,
      "loss": 2.0501,
      "step": 136025
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.0498639345169067,
      "learning_rate": 0.00021595614895645744,
      "loss": 2.092,
      "step": 136026
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.1008745431900024,
      "learning_rate": 0.0002159522221990274,
      "loss": 2.2677,
      "step": 136027
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.188887596130371,
      "learning_rate": 0.0002159482954572235,
      "loss": 1.8952,
      "step": 136028
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.2333629131317139,
      "learning_rate": 0.00021594436873104638,
      "loss": 1.9729,
      "step": 136029
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.1023982763290405,
      "learning_rate": 0.00021594044202049673,
      "loss": 2.1766,
      "step": 136030
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.0976673364639282,
      "learning_rate": 0.00021593651532557536,
      "loss": 2.046,
      "step": 136031
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.5261861085891724,
      "learning_rate": 0.00021593258864628298,
      "loss": 1.8678,
      "step": 136032
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.3085119724273682,
      "learning_rate": 0.0002159286619826203,
      "loss": 2.1089,
      "step": 136033
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.070326805114746,
      "learning_rate": 0.00021592473533458824,
      "loss": 2.1022,
      "step": 136034
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.0195788145065308,
      "learning_rate": 0.0002159208087021872,
      "loss": 1.9121,
      "step": 136035
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.0346146821975708,
      "learning_rate": 0.00021591688208541805,
      "loss": 2.1758,
      "step": 136036
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.051798939704895,
      "learning_rate": 0.00021591295548428155,
      "loss": 2.0544,
      "step": 136037
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.1874430179595947,
      "learning_rate": 0.00021590902889877844,
      "loss": 2.1948,
      "step": 136038
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.0297484397888184,
      "learning_rate": 0.00021590510232890944,
      "loss": 2.0511,
      "step": 136039
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.1392158269882202,
      "learning_rate": 0.00021590117577467534,
      "loss": 2.2167,
      "step": 136040
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.2926195859909058,
      "learning_rate": 0.00021589724923607673,
      "loss": 2.1652,
      "step": 136041
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.141230821609497,
      "learning_rate": 0.00021589332271311436,
      "loss": 2.2286,
      "step": 136042
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.006859540939331,
      "learning_rate": 0.00021588939620578905,
      "loss": 2.2048,
      "step": 136043
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.1038191318511963,
      "learning_rate": 0.00021588546971410146,
      "loss": 2.2158,
      "step": 136044
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.228201150894165,
      "learning_rate": 0.00021588154323805238,
      "loss": 2.0963,
      "step": 136045
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.0771479606628418,
      "learning_rate": 0.00021587761677764258,
      "loss": 2.2411,
      "step": 136046
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.0277824401855469,
      "learning_rate": 0.00021587369033287268,
      "loss": 2.1316,
      "step": 136047
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.0407758951187134,
      "learning_rate": 0.00021586976390374336,
      "loss": 2.0426,
      "step": 136048
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.1285954713821411,
      "learning_rate": 0.0002158658374902555,
      "loss": 2.0295,
      "step": 136049
    },
    {
      "epoch": 1.77,
      "grad_norm": 0.9698218703269958,
      "learning_rate": 0.00021586191109240974,
      "loss": 1.9329,
      "step": 136050
    },
    {
      "epoch": 1.77,
      "grad_norm": 0.9417705535888672,
      "learning_rate": 0.00021585798471020684,
      "loss": 1.9828,
      "step": 136051
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.3994879722595215,
      "learning_rate": 0.00021585405834364767,
      "loss": 2.0944,
      "step": 136052
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.0571891069412231,
      "learning_rate": 0.00021585013199273268,
      "loss": 2.0859,
      "step": 136053
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.1222825050354004,
      "learning_rate": 0.00021584620565746275,
      "loss": 2.1204,
      "step": 136054
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.0793261528015137,
      "learning_rate": 0.00021584227933783858,
      "loss": 2.0886,
      "step": 136055
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.1359548568725586,
      "learning_rate": 0.00021583835303386092,
      "loss": 2.0103,
      "step": 136056
    },
    {
      "epoch": 1.77,
      "grad_norm": 4.957039833068848,
      "learning_rate": 0.00021583442674553052,
      "loss": 2.0945,
      "step": 136057
    },
    {
      "epoch": 1.77,
      "grad_norm": 0.9507777094841003,
      "learning_rate": 0.00021583050047284809,
      "loss": 2.1794,
      "step": 136058
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.0820297002792358,
      "learning_rate": 0.00021582657421581445,
      "loss": 2.171,
      "step": 136059
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.1480810642242432,
      "learning_rate": 0.00021582264797443014,
      "loss": 2.1111,
      "step": 136060
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.0213391780853271,
      "learning_rate": 0.000215818721748696,
      "loss": 2.0602,
      "step": 136061
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.448556661605835,
      "learning_rate": 0.00021581479553861273,
      "loss": 2.1023,
      "step": 136062
    },
    {
      "epoch": 1.77,
      "grad_norm": 0.9697924256324768,
      "learning_rate": 0.0002158108693441811,
      "loss": 2.1579,
      "step": 136063
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.0924304723739624,
      "learning_rate": 0.0002158069431654018,
      "loss": 2.3259,
      "step": 136064
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.1311664581298828,
      "learning_rate": 0.00021580301700227568,
      "loss": 1.9021,
      "step": 136065
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.0384353399276733,
      "learning_rate": 0.00021579909085480328,
      "loss": 2.1293,
      "step": 136066
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.1293209791183472,
      "learning_rate": 0.0002157951647229854,
      "loss": 2.0273,
      "step": 136067
    },
    {
      "epoch": 1.77,
      "grad_norm": 0.989711582660675,
      "learning_rate": 0.0002157912386068228,
      "loss": 2.2105,
      "step": 136068
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.08115816116333,
      "learning_rate": 0.0002157873125063162,
      "loss": 2.31,
      "step": 136069
    },
    {
      "epoch": 1.77,
      "grad_norm": 0.960281491279602,
      "learning_rate": 0.00021578338642146632,
      "loss": 2.0157,
      "step": 136070
    },
    {
      "epoch": 1.77,
      "grad_norm": 0.9959921836853027,
      "learning_rate": 0.00021577946035227402,
      "loss": 1.9084,
      "step": 136071
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.157167911529541,
      "learning_rate": 0.00021577553429873977,
      "loss": 2.4202,
      "step": 136072
    },
    {
      "epoch": 1.77,
      "grad_norm": 0.9899917840957642,
      "learning_rate": 0.0002157716082608645,
      "loss": 2.1995,
      "step": 136073
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.1007353067398071,
      "learning_rate": 0.00021576768223864883,
      "loss": 2.072,
      "step": 136074
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.0925742387771606,
      "learning_rate": 0.0002157637562320935,
      "loss": 2.1586,
      "step": 136075
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.0646188259124756,
      "learning_rate": 0.00021575983024119932,
      "loss": 2.0402,
      "step": 136076
    },
    {
      "epoch": 1.77,
      "grad_norm": 0.9863684773445129,
      "learning_rate": 0.0002157559042659671,
      "loss": 1.9568,
      "step": 136077
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.0093809366226196,
      "learning_rate": 0.00021575197830639732,
      "loss": 2.1129,
      "step": 136078
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.1797088384628296,
      "learning_rate": 0.00021574805236249087,
      "loss": 1.9982,
      "step": 136079
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.0492407083511353,
      "learning_rate": 0.0002157441264342484,
      "loss": 1.9789,
      "step": 136080
    },
    {
      "epoch": 1.77,
      "grad_norm": 0.9960312843322754,
      "learning_rate": 0.0002157402005216707,
      "loss": 2.0639,
      "step": 136081
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.3275604248046875,
      "learning_rate": 0.00021573627462475852,
      "loss": 2.2124,
      "step": 136082
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.0681071281433105,
      "learning_rate": 0.0002157323487435126,
      "loss": 2.0959,
      "step": 136083
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.0271762609481812,
      "learning_rate": 0.0002157284228779336,
      "loss": 1.8887,
      "step": 136084
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.0418884754180908,
      "learning_rate": 0.00021572449702802224,
      "loss": 2.1652,
      "step": 136085
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.2102205753326416,
      "learning_rate": 0.0002157205711937793,
      "loss": 2.0488,
      "step": 136086
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.076729416847229,
      "learning_rate": 0.00021571664537520545,
      "loss": 2.1248,
      "step": 136087
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.071768879890442,
      "learning_rate": 0.0002157127195723015,
      "loss": 2.176,
      "step": 136088
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.2961753606796265,
      "learning_rate": 0.00021570879378506828,
      "loss": 2.085,
      "step": 136089
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.1040326356887817,
      "learning_rate": 0.00021570486801350623,
      "loss": 2.213,
      "step": 136090
    },
    {
      "epoch": 1.77,
      "grad_norm": 0.9709519147872925,
      "learning_rate": 0.00021570094225761626,
      "loss": 2.3182,
      "step": 136091
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.0742892026901245,
      "learning_rate": 0.00021569701651739907,
      "loss": 2.2535,
      "step": 136092
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.0877805948257446,
      "learning_rate": 0.0002156930907928554,
      "loss": 2.203,
      "step": 136093
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.2063443660736084,
      "learning_rate": 0.00021568916508398595,
      "loss": 2.1411,
      "step": 136094
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.0096452236175537,
      "learning_rate": 0.00021568523939079155,
      "loss": 2.2674,
      "step": 136095
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.0194090604782104,
      "learning_rate": 0.0002156813137132729,
      "loss": 2.0488,
      "step": 136096
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.0790412425994873,
      "learning_rate": 0.0002156773880514306,
      "loss": 1.9965,
      "step": 136097
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.1269696950912476,
      "learning_rate": 0.00021567346240526544,
      "loss": 2.2137,
      "step": 136098
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.0709633827209473,
      "learning_rate": 0.00021566953677477817,
      "loss": 2.2654,
      "step": 136099
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.0164848566055298,
      "learning_rate": 0.00021566561115996953,
      "loss": 2.2002,
      "step": 136100
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.114943027496338,
      "learning_rate": 0.00021566168556084035,
      "loss": 2.2537,
      "step": 136101
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.303544282913208,
      "learning_rate": 0.00021565775997739114,
      "loss": 2.0584,
      "step": 136102
    },
    {
      "epoch": 1.77,
      "grad_norm": 0.9681148529052734,
      "learning_rate": 0.00021565383440962275,
      "loss": 2.0523,
      "step": 136103
    },
    {
      "epoch": 1.77,
      "grad_norm": 0.9934043288230896,
      "learning_rate": 0.00021564990885753605,
      "loss": 2.021,
      "step": 136104
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.1525408029556274,
      "learning_rate": 0.00021564598332113146,
      "loss": 2.155,
      "step": 136105
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.193153977394104,
      "learning_rate": 0.0002156420578004099,
      "loss": 2.1687,
      "step": 136106
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.1304540634155273,
      "learning_rate": 0.00021563813229537215,
      "loss": 1.9608,
      "step": 136107
    },
    {
      "epoch": 1.77,
      "grad_norm": 0.9964272379875183,
      "learning_rate": 0.00021563420680601882,
      "loss": 2.0657,
      "step": 136108
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.1978799104690552,
      "learning_rate": 0.00021563028133235064,
      "loss": 2.0434,
      "step": 136109
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.2094275951385498,
      "learning_rate": 0.00021562635587436842,
      "loss": 2.2045,
      "step": 136110
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.0258569717407227,
      "learning_rate": 0.00021562243043207288,
      "loss": 1.8617,
      "step": 136111
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.344952940940857,
      "learning_rate": 0.00021561850500546471,
      "loss": 2.0536,
      "step": 136112
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.063621997833252,
      "learning_rate": 0.00021561457959454468,
      "loss": 1.9132,
      "step": 136113
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.2700917720794678,
      "learning_rate": 0.0002156106541993134,
      "loss": 1.8484,
      "step": 136114
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.0294991731643677,
      "learning_rate": 0.00021560672881977175,
      "loss": 1.9174,
      "step": 136115
    },
    {
      "epoch": 1.77,
      "grad_norm": 0.9868952631950378,
      "learning_rate": 0.00021560280345592038,
      "loss": 2.0857,
      "step": 136116
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.5470473766326904,
      "learning_rate": 0.00021559887810776004,
      "loss": 2.201,
      "step": 136117
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.0952935218811035,
      "learning_rate": 0.00021559495277529155,
      "loss": 2.2259,
      "step": 136118
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.119421124458313,
      "learning_rate": 0.00021559102745851553,
      "loss": 2.126,
      "step": 136119
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.1618226766586304,
      "learning_rate": 0.00021558710215743265,
      "loss": 1.9234,
      "step": 136120
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.0174530744552612,
      "learning_rate": 0.00021558317687204372,
      "loss": 1.9146,
      "step": 136121
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.3764562606811523,
      "learning_rate": 0.0002155792516023495,
      "loss": 2.274,
      "step": 136122
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.0582906007766724,
      "learning_rate": 0.00021557532634835066,
      "loss": 2.2129,
      "step": 136123
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.0167638063430786,
      "learning_rate": 0.000215571401110048,
      "loss": 2.1868,
      "step": 136124
    },
    {
      "epoch": 1.77,
      "grad_norm": 0.9818055033683777,
      "learning_rate": 0.00021556747588744225,
      "loss": 2.0731,
      "step": 136125
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.2539771795272827,
      "learning_rate": 0.0002155635506805341,
      "loss": 2.1406,
      "step": 136126
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.2088924646377563,
      "learning_rate": 0.00021555962548932422,
      "loss": 2.2198,
      "step": 136127
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.0795100927352905,
      "learning_rate": 0.0002155557003138134,
      "loss": 2.4064,
      "step": 136128
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.1003986597061157,
      "learning_rate": 0.00021555177515400237,
      "loss": 2.1433,
      "step": 136129
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.2003685235977173,
      "learning_rate": 0.0002155478500098919,
      "loss": 2.1456,
      "step": 136130
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.1583170890808105,
      "learning_rate": 0.0002155439248814826,
      "loss": 1.9611,
      "step": 136131
    },
    {
      "epoch": 1.77,
      "grad_norm": 0.9845763444900513,
      "learning_rate": 0.00021553999976877544,
      "loss": 2.1801,
      "step": 136132
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.0309869050979614,
      "learning_rate": 0.0002155360746717709,
      "loss": 2.0927,
      "step": 136133
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.1393204927444458,
      "learning_rate": 0.00021553214959046973,
      "loss": 2.1225,
      "step": 136134
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.1380807161331177,
      "learning_rate": 0.00021552822452487277,
      "loss": 2.0833,
      "step": 136135
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.2005070447921753,
      "learning_rate": 0.0002155242994749807,
      "loss": 2.0354,
      "step": 136136
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.0269731283187866,
      "learning_rate": 0.00021552037444079426,
      "loss": 2.5053,
      "step": 136137
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.0896036624908447,
      "learning_rate": 0.0002155164494223143,
      "loss": 2.0019,
      "step": 136138
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.2298868894577026,
      "learning_rate": 0.0002155125244195413,
      "loss": 1.955,
      "step": 136139
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.2636381387710571,
      "learning_rate": 0.0002155085994324761,
      "loss": 2.2798,
      "step": 136140
    },
    {
      "epoch": 1.77,
      "grad_norm": 0.9705795049667358,
      "learning_rate": 0.00021550467446111948,
      "loss": 2.0658,
      "step": 136141
    },
    {
      "epoch": 1.77,
      "grad_norm": 0.9643788933753967,
      "learning_rate": 0.00021550074950547212,
      "loss": 1.9668,
      "step": 136142
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.114575743675232,
      "learning_rate": 0.00021549682456553472,
      "loss": 2.0634,
      "step": 136143
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.1562901735305786,
      "learning_rate": 0.00021549289964130822,
      "loss": 2.1834,
      "step": 136144
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.1757922172546387,
      "learning_rate": 0.0002154889747327931,
      "loss": 2.0121,
      "step": 136145
    },
    {
      "epoch": 1.77,
      "grad_norm": 0.9872236251831055,
      "learning_rate": 0.00021548504983999014,
      "loss": 2.1636,
      "step": 136146
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.1472514867782593,
      "learning_rate": 0.0002154811249629001,
      "loss": 2.0845,
      "step": 136147
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.1323199272155762,
      "learning_rate": 0.0002154772001015237,
      "loss": 2.228,
      "step": 136148
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.0013481378555298,
      "learning_rate": 0.00021547327525586168,
      "loss": 2.0493,
      "step": 136149
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.0308408737182617,
      "learning_rate": 0.00021546935042591494,
      "loss": 1.9804,
      "step": 136150
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.1449037790298462,
      "learning_rate": 0.0002154654256116839,
      "loss": 2.0833,
      "step": 136151
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.0769875049591064,
      "learning_rate": 0.00021546150081316943,
      "loss": 2.1559,
      "step": 136152
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.0546165704727173,
      "learning_rate": 0.00021545757603037228,
      "loss": 2.2793,
      "step": 136153
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.0190033912658691,
      "learning_rate": 0.00021545365126329316,
      "loss": 2.1793,
      "step": 136154
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.0870219469070435,
      "learning_rate": 0.00021544972651193277,
      "loss": 2.1192,
      "step": 136155
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.0813239812850952,
      "learning_rate": 0.00021544580177629201,
      "loss": 2.1958,
      "step": 136156
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.0892658233642578,
      "learning_rate": 0.00021544187705637138,
      "loss": 2.1292,
      "step": 136157
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.0085628032684326,
      "learning_rate": 0.00021543795235217167,
      "loss": 1.8717,
      "step": 136158
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.0482189655303955,
      "learning_rate": 0.00021543402766369363,
      "loss": 2.1193,
      "step": 136159
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.0937398672103882,
      "learning_rate": 0.000215430102990938,
      "loss": 2.0475,
      "step": 136160
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.2488718032836914,
      "learning_rate": 0.0002154261783339055,
      "loss": 2.1593,
      "step": 136161
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.071266770362854,
      "learning_rate": 0.00021542225369259704,
      "loss": 2.1823,
      "step": 136162
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.2505614757537842,
      "learning_rate": 0.00021541832906701308,
      "loss": 2.119,
      "step": 136163
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.4320831298828125,
      "learning_rate": 0.00021541440445715435,
      "loss": 2.2467,
      "step": 136164
    },
    {
      "epoch": 1.77,
      "grad_norm": 0.9601215720176697,
      "learning_rate": 0.00021541047986302177,
      "loss": 1.921,
      "step": 136165
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.0153640508651733,
      "learning_rate": 0.00021540655528461594,
      "loss": 2.146,
      "step": 136166
    },
    {
      "epoch": 1.77,
      "grad_norm": 0.9280619621276855,
      "learning_rate": 0.00021540263072193763,
      "loss": 2.1436,
      "step": 136167
    },
    {
      "epoch": 1.77,
      "grad_norm": 0.9915780425071716,
      "learning_rate": 0.0002153987061749877,
      "loss": 2.0245,
      "step": 136168
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.0230437517166138,
      "learning_rate": 0.00021539478164376658,
      "loss": 2.0578,
      "step": 136169
    },
    {
      "epoch": 1.77,
      "grad_norm": 0.9519174695014954,
      "learning_rate": 0.00021539085712827524,
      "loss": 2.1839,
      "step": 136170
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.1880844831466675,
      "learning_rate": 0.0002153869326285143,
      "loss": 2.3789,
      "step": 136171
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.1083123683929443,
      "learning_rate": 0.00021538300814448452,
      "loss": 1.8497,
      "step": 136172
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.0016158819198608,
      "learning_rate": 0.0002153790836761866,
      "loss": 2.2645,
      "step": 136173
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.092397928237915,
      "learning_rate": 0.0002153751592236215,
      "loss": 2.1919,
      "step": 136174
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.082176685333252,
      "learning_rate": 0.0002153712347867896,
      "loss": 2.2053,
      "step": 136175
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.0322988033294678,
      "learning_rate": 0.00021536731036569178,
      "loss": 1.9479,
      "step": 136176
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.0290273427963257,
      "learning_rate": 0.0002153633859603288,
      "loss": 2.1938,
      "step": 136177
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.0912630558013916,
      "learning_rate": 0.00021535946157070137,
      "loss": 2.1176,
      "step": 136178
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.2232328653335571,
      "learning_rate": 0.00021535553719681018,
      "loss": 2.2247,
      "step": 136179
    },
    {
      "epoch": 1.77,
      "grad_norm": 0.9586880207061768,
      "learning_rate": 0.00021535161283865603,
      "loss": 1.9991,
      "step": 136180
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.0511564016342163,
      "learning_rate": 0.00021534768849623966,
      "loss": 2.2384,
      "step": 136181
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.2173985242843628,
      "learning_rate": 0.0002153437641695617,
      "loss": 2.0339,
      "step": 136182
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.1846917867660522,
      "learning_rate": 0.00021533983985862288,
      "loss": 2.4273,
      "step": 136183
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.112747311592102,
      "learning_rate": 0.00021533591556342402,
      "loss": 2.1387,
      "step": 136184
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.0689661502838135,
      "learning_rate": 0.00021533199128396583,
      "loss": 2.3962,
      "step": 136185
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.2442858219146729,
      "learning_rate": 0.00021532806702024906,
      "loss": 1.9005,
      "step": 136186
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.2729486227035522,
      "learning_rate": 0.00021532414277227434,
      "loss": 2.108,
      "step": 136187
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.0339679718017578,
      "learning_rate": 0.0002153202185400425,
      "loss": 1.9387,
      "step": 136188
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.131528377532959,
      "learning_rate": 0.00021531629432355417,
      "loss": 1.9835,
      "step": 136189
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.0676168203353882,
      "learning_rate": 0.0002153123701228102,
      "loss": 2.0232,
      "step": 136190
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.0818294286727905,
      "learning_rate": 0.0002153084459378112,
      "loss": 1.8491,
      "step": 136191
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.0338200330734253,
      "learning_rate": 0.00021530452176855798,
      "loss": 2.3336,
      "step": 136192
    },
    {
      "epoch": 1.77,
      "grad_norm": 0.97489994764328,
      "learning_rate": 0.0002153005976150513,
      "loss": 1.9715,
      "step": 136193
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.105150818824768,
      "learning_rate": 0.00021529667347729175,
      "loss": 2.1018,
      "step": 136194
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.2622281312942505,
      "learning_rate": 0.0002152927493552802,
      "loss": 2.2512,
      "step": 136195
    },
    {
      "epoch": 1.77,
      "grad_norm": 0.9552990198135376,
      "learning_rate": 0.00021528882524901737,
      "loss": 2.0571,
      "step": 136196
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.101043939590454,
      "learning_rate": 0.00021528490115850385,
      "loss": 1.8859,
      "step": 136197
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.3450592756271362,
      "learning_rate": 0.0002152809770837405,
      "loss": 2.1362,
      "step": 136198
    },
    {
      "epoch": 1.77,
      "grad_norm": 0.9863818287849426,
      "learning_rate": 0.0002152770530247281,
      "loss": 2.1297,
      "step": 136199
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.109947919845581,
      "learning_rate": 0.00021527312898146717,
      "loss": 2.0915,
      "step": 136200
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.1196703910827637,
      "learning_rate": 0.00021526920495395862,
      "loss": 2.1623,
      "step": 136201
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.0516197681427002,
      "learning_rate": 0.00021526528094220312,
      "loss": 2.035,
      "step": 136202
    },
    {
      "epoch": 1.77,
      "grad_norm": 0.975838303565979,
      "learning_rate": 0.00021526135694620145,
      "loss": 1.9181,
      "step": 136203
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.041003942489624,
      "learning_rate": 0.00021525743296595422,
      "loss": 2.126,
      "step": 136204
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.0536472797393799,
      "learning_rate": 0.0002152535090014623,
      "loss": 2.1198,
      "step": 136205
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.0935065746307373,
      "learning_rate": 0.0002152495850527263,
      "loss": 1.9816,
      "step": 136206
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.2050589323043823,
      "learning_rate": 0.000215245661119747,
      "loss": 2.1255,
      "step": 136207
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.1484001874923706,
      "learning_rate": 0.00021524173720252515,
      "loss": 1.9882,
      "step": 136208
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.0413498878479004,
      "learning_rate": 0.00021523781330106144,
      "loss": 1.9321,
      "step": 136209
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.1932659149169922,
      "learning_rate": 0.00021523388941535666,
      "loss": 1.9884,
      "step": 136210
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.1631032228469849,
      "learning_rate": 0.00021522996554541155,
      "loss": 1.9605,
      "step": 136211
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.20742666721344,
      "learning_rate": 0.0002152260416912267,
      "loss": 2.1974,
      "step": 136212
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.1344817876815796,
      "learning_rate": 0.00021522211785280292,
      "loss": 2.1941,
      "step": 136213
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.1676973104476929,
      "learning_rate": 0.00021521819403014098,
      "loss": 2.1977,
      "step": 136214
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.0826404094696045,
      "learning_rate": 0.00021521427022324155,
      "loss": 2.1808,
      "step": 136215
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.0832582712173462,
      "learning_rate": 0.0002152103464321054,
      "loss": 2.0559,
      "step": 136216
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.0319716930389404,
      "learning_rate": 0.00021520642265673338,
      "loss": 2.1504,
      "step": 136217
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.095348834991455,
      "learning_rate": 0.00021520249889712595,
      "loss": 2.1317,
      "step": 136218
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.187951683998108,
      "learning_rate": 0.00021519857515328394,
      "loss": 1.9026,
      "step": 136219
    },
    {
      "epoch": 1.77,
      "grad_norm": 0.9723703861236572,
      "learning_rate": 0.00021519465142520817,
      "loss": 2.1747,
      "step": 136220
    },
    {
      "epoch": 1.77,
      "grad_norm": 0.9684462547302246,
      "learning_rate": 0.00021519072771289928,
      "loss": 2.1823,
      "step": 136221
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.4442505836486816,
      "learning_rate": 0.00021518680401635807,
      "loss": 2.1323,
      "step": 136222
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.0671640634536743,
      "learning_rate": 0.00021518288033558534,
      "loss": 2.0517,
      "step": 136223
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.0552343130111694,
      "learning_rate": 0.0002151789566705816,
      "loss": 2.1702,
      "step": 136224
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.0257186889648438,
      "learning_rate": 0.00021517503302134769,
      "loss": 2.1871,
      "step": 136225
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.0818666219711304,
      "learning_rate": 0.00021517110938788428,
      "loss": 2.2863,
      "step": 136226
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.0228962898254395,
      "learning_rate": 0.00021516718577019225,
      "loss": 1.9165,
      "step": 136227
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.1301575899124146,
      "learning_rate": 0.00021516326216827219,
      "loss": 2.0561,
      "step": 136228
    },
    {
      "epoch": 1.77,
      "grad_norm": 0.9815588593482971,
      "learning_rate": 0.00021515933858212504,
      "loss": 2.0289,
      "step": 136229
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.027463674545288,
      "learning_rate": 0.00021515541501175122,
      "loss": 2.1669,
      "step": 136230
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.0121287107467651,
      "learning_rate": 0.00021515149145715165,
      "loss": 2.0145,
      "step": 136231
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.3933982849121094,
      "learning_rate": 0.00021514756791832695,
      "loss": 2.0922,
      "step": 136232
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.0471973419189453,
      "learning_rate": 0.00021514364439527797,
      "loss": 2.191,
      "step": 136233
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.254152774810791,
      "learning_rate": 0.00021513972088800534,
      "loss": 2.1903,
      "step": 136234
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.2850103378295898,
      "learning_rate": 0.00021513579739651005,
      "loss": 1.9031,
      "step": 136235
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.062264323234558,
      "learning_rate": 0.00021513187392079242,
      "loss": 1.8303,
      "step": 136236
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.1612340211868286,
      "learning_rate": 0.0002151279504608534,
      "loss": 2.0948,
      "step": 136237
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.0072869062423706,
      "learning_rate": 0.0002151240270166937,
      "loss": 2.1606,
      "step": 136238
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.0359293222427368,
      "learning_rate": 0.00021512010358831402,
      "loss": 2.3484,
      "step": 136239
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.1283444166183472,
      "learning_rate": 0.00021511618017571512,
      "loss": 2.1882,
      "step": 136240
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.0734463930130005,
      "learning_rate": 0.0002151122567788979,
      "loss": 1.9739,
      "step": 136241
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.0036451816558838,
      "learning_rate": 0.00021510833339786274,
      "loss": 2.1597,
      "step": 136242
    },
    {
      "epoch": 1.77,
      "grad_norm": 0.9529796838760376,
      "learning_rate": 0.00021510441003261057,
      "loss": 2.0615,
      "step": 136243
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.0812404155731201,
      "learning_rate": 0.00021510048668314205,
      "loss": 2.054,
      "step": 136244
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.1071847677230835,
      "learning_rate": 0.000215096563349458,
      "loss": 2.2477,
      "step": 136245
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.007987380027771,
      "learning_rate": 0.00021509264003155906,
      "loss": 2.0451,
      "step": 136246
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.0404354333877563,
      "learning_rate": 0.0002150887167294462,
      "loss": 2.1369,
      "step": 136247
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.0726324319839478,
      "learning_rate": 0.00021508479344311974,
      "loss": 2.2391,
      "step": 136248
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.041870355606079,
      "learning_rate": 0.00021508087017258067,
      "loss": 2.1013,
      "step": 136249
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.0660990476608276,
      "learning_rate": 0.00021507694691782963,
      "loss": 2.0986,
      "step": 136250
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.184051513671875,
      "learning_rate": 0.00021507302367886744,
      "loss": 2.3282,
      "step": 136251
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.1174460649490356,
      "learning_rate": 0.00021506910045569473,
      "loss": 2.304,
      "step": 136252
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.0404597520828247,
      "learning_rate": 0.0002150651772483124,
      "loss": 1.8571,
      "step": 136253
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.3662707805633545,
      "learning_rate": 0.00021506125405672098,
      "loss": 2.1147,
      "step": 136254
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.0442684888839722,
      "learning_rate": 0.00021505733088092127,
      "loss": 2.0111,
      "step": 136255
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.1362476348876953,
      "learning_rate": 0.00021505340772091396,
      "loss": 2.084,
      "step": 136256
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.1179956197738647,
      "learning_rate": 0.00021504948457669985,
      "loss": 2.1515,
      "step": 136257
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.0529861450195312,
      "learning_rate": 0.00021504556144827966,
      "loss": 1.9635,
      "step": 136258
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.090136170387268,
      "learning_rate": 0.0002150416383356541,
      "loss": 2.1454,
      "step": 136259
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.0150578022003174,
      "learning_rate": 0.00021503771523882396,
      "loss": 2.1487,
      "step": 136260
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.4000370502471924,
      "learning_rate": 0.00021503379215778986,
      "loss": 2.0595,
      "step": 136261
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.1432081460952759,
      "learning_rate": 0.00021502986909255255,
      "loss": 2.0386,
      "step": 136262
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.10166335105896,
      "learning_rate": 0.00021502594604311284,
      "loss": 2.2574,
      "step": 136263
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.0822765827178955,
      "learning_rate": 0.00021502202300947135,
      "loss": 2.0507,
      "step": 136264
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.4106547832489014,
      "learning_rate": 0.00021501809999162892,
      "loss": 2.1498,
      "step": 136265
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.0282303094863892,
      "learning_rate": 0.00021501417698958632,
      "loss": 1.9999,
      "step": 136266
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.05962336063385,
      "learning_rate": 0.00021501025400334412,
      "loss": 2.0205,
      "step": 136267
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.0321614742279053,
      "learning_rate": 0.000215006331032903,
      "loss": 1.988,
      "step": 136268
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.1050652265548706,
      "learning_rate": 0.00021500240807826391,
      "loss": 2.305,
      "step": 136269
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.0549142360687256,
      "learning_rate": 0.00021499848513942743,
      "loss": 2.1265,
      "step": 136270
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.0006791353225708,
      "learning_rate": 0.00021499456221639438,
      "loss": 1.907,
      "step": 136271
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.103691577911377,
      "learning_rate": 0.0002149906393091655,
      "loss": 2.1816,
      "step": 136272
    },
    {
      "epoch": 1.77,
      "grad_norm": 0.9996963143348694,
      "learning_rate": 0.00021498671641774142,
      "loss": 1.9692,
      "step": 136273
    },
    {
      "epoch": 1.77,
      "grad_norm": 0.9977710843086243,
      "learning_rate": 0.00021498279354212292,
      "loss": 1.9876,
      "step": 136274
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.0697225332260132,
      "learning_rate": 0.00021497887068231065,
      "loss": 2.096,
      "step": 136275
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.1421548128128052,
      "learning_rate": 0.00021497494783830542,
      "loss": 2.0951,
      "step": 136276
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.1234935522079468,
      "learning_rate": 0.00021497102501010802,
      "loss": 2.0986,
      "step": 136277
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.083306074142456,
      "learning_rate": 0.00021496710219771908,
      "loss": 1.8631,
      "step": 136278
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.0248521566390991,
      "learning_rate": 0.00021496317940113938,
      "loss": 2.273,
      "step": 136279
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.1395485401153564,
      "learning_rate": 0.00021495925662036967,
      "loss": 1.9309,
      "step": 136280
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.1672838926315308,
      "learning_rate": 0.00021495533385541057,
      "loss": 2.1196,
      "step": 136281
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.151150107383728,
      "learning_rate": 0.00021495141110626287,
      "loss": 2.0834,
      "step": 136282
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.0647939443588257,
      "learning_rate": 0.0002149474883729273,
      "loss": 2.022,
      "step": 136283
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.1103817224502563,
      "learning_rate": 0.0002149435656554047,
      "loss": 1.8783,
      "step": 136284
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.1278915405273438,
      "learning_rate": 0.0002149396429536956,
      "loss": 2.1295,
      "step": 136285
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.0072691440582275,
      "learning_rate": 0.00021493572026780086,
      "loss": 2.1434,
      "step": 136286
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.1653468608856201,
      "learning_rate": 0.00021493179759772118,
      "loss": 1.9753,
      "step": 136287
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.015349268913269,
      "learning_rate": 0.0002149278749434573,
      "loss": 2.0345,
      "step": 136288
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.0067775249481201,
      "learning_rate": 0.00021492395230500986,
      "loss": 2.3449,
      "step": 136289
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.055546760559082,
      "learning_rate": 0.00021492002968237975,
      "loss": 1.992,
      "step": 136290
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.0413670539855957,
      "learning_rate": 0.00021491610707556754,
      "loss": 2.0224,
      "step": 136291
    },
    {
      "epoch": 1.77,
      "grad_norm": 0.9082266688346863,
      "learning_rate": 0.00021491218448457407,
      "loss": 1.9925,
      "step": 136292
    },
    {
      "epoch": 1.77,
      "grad_norm": 0.944280743598938,
      "learning_rate": 0.0002149082619094,
      "loss": 2.0814,
      "step": 136293
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.0660167932510376,
      "learning_rate": 0.0002149043393500461,
      "loss": 2.1313,
      "step": 136294
    },
    {
      "epoch": 1.77,
      "grad_norm": 0.9825825691223145,
      "learning_rate": 0.00021490041680651312,
      "loss": 2.1138,
      "step": 136295
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.082399845123291,
      "learning_rate": 0.00021489649427880182,
      "loss": 2.2633,
      "step": 136296
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.044291615486145,
      "learning_rate": 0.0002148925717669128,
      "loss": 2.2422,
      "step": 136297
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.0371527671813965,
      "learning_rate": 0.0002148886492708468,
      "loss": 2.2151,
      "step": 136298
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.0490745306015015,
      "learning_rate": 0.00021488472679060465,
      "loss": 2.1922,
      "step": 136299
    },
    {
      "epoch": 1.77,
      "grad_norm": 0.9669976830482483,
      "learning_rate": 0.000214880804326187,
      "loss": 2.1317,
      "step": 136300
    },
    {
      "epoch": 1.77,
      "grad_norm": 0.9890241026878357,
      "learning_rate": 0.00021487688187759468,
      "loss": 1.8908,
      "step": 136301
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.051727294921875,
      "learning_rate": 0.00021487295944482842,
      "loss": 1.852,
      "step": 136302
    },
    {
      "epoch": 1.77,
      "grad_norm": 0.975839376449585,
      "learning_rate": 0.00021486903702788878,
      "loss": 2.1584,
      "step": 136303
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.014967918395996,
      "learning_rate": 0.00021486511462677661,
      "loss": 2.0576,
      "step": 136304
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.0791597366333008,
      "learning_rate": 0.0002148611922414926,
      "loss": 2.0835,
      "step": 136305
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.0842105150222778,
      "learning_rate": 0.0002148572698720375,
      "loss": 2.0732,
      "step": 136306
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.111946702003479,
      "learning_rate": 0.00021485334751841203,
      "loss": 2.3711,
      "step": 136307
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.0125950574874878,
      "learning_rate": 0.00021484942518061706,
      "loss": 2.1822,
      "step": 136308
    },
    {
      "epoch": 1.77,
      "grad_norm": 0.9378119707107544,
      "learning_rate": 0.00021484550285865312,
      "loss": 1.9452,
      "step": 136309
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.0753835439682007,
      "learning_rate": 0.00021484158055252097,
      "loss": 2.2575,
      "step": 136310
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.1616499423980713,
      "learning_rate": 0.00021483765826222135,
      "loss": 2.2468,
      "step": 136311
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.0508596897125244,
      "learning_rate": 0.00021483373598775503,
      "loss": 1.9341,
      "step": 136312
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.1341010332107544,
      "learning_rate": 0.00021482981372912273,
      "loss": 2.0002,
      "step": 136313
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.015509009361267,
      "learning_rate": 0.0002148258914863253,
      "loss": 2.1194,
      "step": 136314
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.019453525543213,
      "learning_rate": 0.00021482196925936322,
      "loss": 2.0777,
      "step": 136315
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.10420560836792,
      "learning_rate": 0.00021481804704823735,
      "loss": 2.293,
      "step": 136316
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.0953302383422852,
      "learning_rate": 0.00021481412485294845,
      "loss": 2.0699,
      "step": 136317
    },
    {
      "epoch": 1.77,
      "grad_norm": 0.9809191823005676,
      "learning_rate": 0.00021481020267349716,
      "loss": 2.0198,
      "step": 136318
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.090545892715454,
      "learning_rate": 0.00021480628050988426,
      "loss": 2.0741,
      "step": 136319
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.136733055114746,
      "learning_rate": 0.00021480235836211065,
      "loss": 2.257,
      "step": 136320
    },
    {
      "epoch": 1.77,
      "grad_norm": 1.1704479455947876,
      "learning_rate": 0.0002147984362301767,
      "loss": 2.0098,
      "step": 136321
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.0319229364395142,
      "learning_rate": 0.00021479451411408337,
      "loss": 2.1792,
      "step": 136322
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.1056920289993286,
      "learning_rate": 0.00021479059201383132,
      "loss": 2.1769,
      "step": 136323
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.142534613609314,
      "learning_rate": 0.00021478666992942136,
      "loss": 1.8477,
      "step": 136324
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.3139811754226685,
      "learning_rate": 0.00021478274786085414,
      "loss": 1.9609,
      "step": 136325
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.1830339431762695,
      "learning_rate": 0.0002147788258081304,
      "loss": 2.1596,
      "step": 136326
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.0709806680679321,
      "learning_rate": 0.00021477490377125105,
      "loss": 1.9554,
      "step": 136327
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.0258548259735107,
      "learning_rate": 0.0002147709817502165,
      "loss": 2.1494,
      "step": 136328
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.2907509803771973,
      "learning_rate": 0.00021476705974502762,
      "loss": 2.1774,
      "step": 136329
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.0546228885650635,
      "learning_rate": 0.00021476313775568514,
      "loss": 2.0707,
      "step": 136330
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.0541054010391235,
      "learning_rate": 0.00021475921578218985,
      "loss": 2.1597,
      "step": 136331
    },
    {
      "epoch": 1.78,
      "grad_norm": 0.995002031326294,
      "learning_rate": 0.0002147552938245424,
      "loss": 2.3537,
      "step": 136332
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.0283381938934326,
      "learning_rate": 0.0002147513718827437,
      "loss": 2.0981,
      "step": 136333
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.0908143520355225,
      "learning_rate": 0.00021474744995679418,
      "loss": 2.0119,
      "step": 136334
    },
    {
      "epoch": 1.78,
      "grad_norm": 0.9937705397605896,
      "learning_rate": 0.00021474352804669474,
      "loss": 1.9136,
      "step": 136335
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.0397096872329712,
      "learning_rate": 0.00021473960615244609,
      "loss": 2.2732,
      "step": 136336
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.0983144044876099,
      "learning_rate": 0.0002147356842740489,
      "loss": 2.0378,
      "step": 136337
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.0642691850662231,
      "learning_rate": 0.000214731762411504,
      "loss": 2.1147,
      "step": 136338
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.1457459926605225,
      "learning_rate": 0.0002147278405648122,
      "loss": 2.3727,
      "step": 136339
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.0474114418029785,
      "learning_rate": 0.00021472391873397396,
      "loss": 2.2117,
      "step": 136340
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.0778745412826538,
      "learning_rate": 0.0002147199969189902,
      "loss": 2.0491,
      "step": 136341
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.1829757690429688,
      "learning_rate": 0.00021471607511986155,
      "loss": 2.1729,
      "step": 136342
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.0269837379455566,
      "learning_rate": 0.00021471215333658883,
      "loss": 2.223,
      "step": 136343
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.0469133853912354,
      "learning_rate": 0.0002147082315691727,
      "loss": 2.0407,
      "step": 136344
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.1529051065444946,
      "learning_rate": 0.00021470430981761408,
      "loss": 1.9664,
      "step": 136345
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.1774523258209229,
      "learning_rate": 0.00021470038808191338,
      "loss": 2.1629,
      "step": 136346
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.0360448360443115,
      "learning_rate": 0.00021469646636207152,
      "loss": 1.9719,
      "step": 136347
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.0102360248565674,
      "learning_rate": 0.00021469254465808916,
      "loss": 1.9312,
      "step": 136348
    },
    {
      "epoch": 1.78,
      "grad_norm": 0.9643909335136414,
      "learning_rate": 0.00021468862296996708,
      "loss": 1.9123,
      "step": 136349
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.1064928770065308,
      "learning_rate": 0.00021468470129770598,
      "loss": 2.2224,
      "step": 136350
    },
    {
      "epoch": 1.78,
      "grad_norm": 0.9890897870063782,
      "learning_rate": 0.00021468077964130675,
      "loss": 1.9461,
      "step": 136351
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.126617193222046,
      "learning_rate": 0.00021467685800076987,
      "loss": 2.0492,
      "step": 136352
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.0569931268692017,
      "learning_rate": 0.00021467293637609616,
      "loss": 2.0845,
      "step": 136353
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.0638262033462524,
      "learning_rate": 0.00021466901476728634,
      "loss": 1.8629,
      "step": 136354
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.049539566040039,
      "learning_rate": 0.00021466509317434119,
      "loss": 2.1131,
      "step": 136355
    },
    {
      "epoch": 1.78,
      "grad_norm": 0.9818323254585266,
      "learning_rate": 0.0002146611715972614,
      "loss": 1.8235,
      "step": 136356
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.1478298902511597,
      "learning_rate": 0.00021465725003604776,
      "loss": 2.2283,
      "step": 136357
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.0106815099716187,
      "learning_rate": 0.00021465332849070093,
      "loss": 2.1027,
      "step": 136358
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.0234599113464355,
      "learning_rate": 0.00021464940696122164,
      "loss": 2.3211,
      "step": 136359
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.2518727779388428,
      "learning_rate": 0.0002146454854476106,
      "loss": 2.2504,
      "step": 136360
    },
    {
      "epoch": 1.78,
      "grad_norm": 0.9934995174407959,
      "learning_rate": 0.0002146415639498686,
      "loss": 2.3198,
      "step": 136361
    },
    {
      "epoch": 1.78,
      "grad_norm": 0.9797136783599854,
      "learning_rate": 0.0002146376424679963,
      "loss": 2.0465,
      "step": 136362
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.0726150274276733,
      "learning_rate": 0.0002146337210019946,
      "loss": 2.0464,
      "step": 136363
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.0069284439086914,
      "learning_rate": 0.000214629799551864,
      "loss": 2.2095,
      "step": 136364
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.164259672164917,
      "learning_rate": 0.00021462587811760538,
      "loss": 2.2831,
      "step": 136365
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.1278303861618042,
      "learning_rate": 0.0002146219566992194,
      "loss": 2.1685,
      "step": 136366
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.1627249717712402,
      "learning_rate": 0.0002146180352967068,
      "loss": 2.1254,
      "step": 136367
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.0463060140609741,
      "learning_rate": 0.00021461411391006832,
      "loss": 2.1524,
      "step": 136368
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.0659748315811157,
      "learning_rate": 0.00021461019253930474,
      "loss": 2.0036,
      "step": 136369
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.0627185106277466,
      "learning_rate": 0.0002146062711844167,
      "loss": 1.9734,
      "step": 136370
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.0321027040481567,
      "learning_rate": 0.00021460234984540491,
      "loss": 2.0326,
      "step": 136371
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.0584381818771362,
      "learning_rate": 0.00021459842852227025,
      "loss": 2.1199,
      "step": 136372
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.1238571405410767,
      "learning_rate": 0.0002145945072150133,
      "loss": 2.2338,
      "step": 136373
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.1632791757583618,
      "learning_rate": 0.0002145905859236348,
      "loss": 2.2517,
      "step": 136374
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.011033058166504,
      "learning_rate": 0.00021458666464813564,
      "loss": 2.0688,
      "step": 136375
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.0551186800003052,
      "learning_rate": 0.00021458274338851635,
      "loss": 1.8276,
      "step": 136376
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.0220500230789185,
      "learning_rate": 0.00021457882214477773,
      "loss": 1.955,
      "step": 136377
    },
    {
      "epoch": 1.78,
      "grad_norm": 0.9289240837097168,
      "learning_rate": 0.00021457490091692056,
      "loss": 2.0931,
      "step": 136378
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.0349334478378296,
      "learning_rate": 0.00021457097970494553,
      "loss": 2.2717,
      "step": 136379
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.1017630100250244,
      "learning_rate": 0.00021456705850885334,
      "loss": 2.1308,
      "step": 136380
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.026646614074707,
      "learning_rate": 0.00021456313732864482,
      "loss": 1.9716,
      "step": 136381
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.0518471002578735,
      "learning_rate": 0.00021455921616432054,
      "loss": 2.1182,
      "step": 136382
    },
    {
      "epoch": 1.78,
      "grad_norm": 0.9234627485275269,
      "learning_rate": 0.00021455529501588133,
      "loss": 2.1852,
      "step": 136383
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.3809728622436523,
      "learning_rate": 0.00021455137388332792,
      "loss": 2.2332,
      "step": 136384
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.064769983291626,
      "learning_rate": 0.00021454745276666093,
      "loss": 2.0821,
      "step": 136385
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.0530040264129639,
      "learning_rate": 0.00021454353166588128,
      "loss": 2.3642,
      "step": 136386
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.0493879318237305,
      "learning_rate": 0.00021453961058098963,
      "loss": 2.2334,
      "step": 136387
    },
    {
      "epoch": 1.78,
      "grad_norm": 0.9871661067008972,
      "learning_rate": 0.00021453568951198667,
      "loss": 2.0587,
      "step": 136388
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.1777554750442505,
      "learning_rate": 0.00021453176845887307,
      "loss": 2.349,
      "step": 136389
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.0148156881332397,
      "learning_rate": 0.00021452784742164967,
      "loss": 1.9283,
      "step": 136390
    },
    {
      "epoch": 1.78,
      "grad_norm": 0.9619788527488708,
      "learning_rate": 0.00021452392640031715,
      "loss": 1.8966,
      "step": 136391
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.1328905820846558,
      "learning_rate": 0.00021452000539487622,
      "loss": 2.205,
      "step": 136392
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.051256775856018,
      "learning_rate": 0.00021451608440532763,
      "loss": 2.1846,
      "step": 136393
    },
    {
      "epoch": 1.78,
      "grad_norm": 0.987700879573822,
      "learning_rate": 0.00021451216343167226,
      "loss": 2.0911,
      "step": 136394
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.0534076690673828,
      "learning_rate": 0.00021450824247391055,
      "loss": 2.2028,
      "step": 136395
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.0093525648117065,
      "learning_rate": 0.00021450432153204336,
      "loss": 1.9721,
      "step": 136396
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.0515683889389038,
      "learning_rate": 0.00021450040060607146,
      "loss": 2.1778,
      "step": 136397
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.045989751815796,
      "learning_rate": 0.00021449647969599555,
      "loss": 2.1962,
      "step": 136398
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.0788803100585938,
      "learning_rate": 0.00021449255880181633,
      "loss": 2.1507,
      "step": 136399
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.017828106880188,
      "learning_rate": 0.00021448863792353473,
      "loss": 2.0771,
      "step": 136400
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.1077823638916016,
      "learning_rate": 0.00021448471706115115,
      "loss": 2.1152,
      "step": 136401
    },
    {
      "epoch": 1.78,
      "grad_norm": 0.9967592358589172,
      "learning_rate": 0.00021448079621466648,
      "loss": 2.3024,
      "step": 136402
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.1212457418441772,
      "learning_rate": 0.0002144768753840814,
      "loss": 2.066,
      "step": 136403
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.074110507965088,
      "learning_rate": 0.00021447295456939676,
      "loss": 2.3112,
      "step": 136404
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.0917017459869385,
      "learning_rate": 0.0002144690337706132,
      "loss": 2.2777,
      "step": 136405
    },
    {
      "epoch": 1.78,
      "grad_norm": 0.9940749406814575,
      "learning_rate": 0.00021446511298773156,
      "loss": 2.2166,
      "step": 136406
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.2451319694519043,
      "learning_rate": 0.00021446119222075233,
      "loss": 1.9559,
      "step": 136407
    },
    {
      "epoch": 1.78,
      "grad_norm": 0.9983391165733337,
      "learning_rate": 0.0002144572714696764,
      "loss": 2.1365,
      "step": 136408
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.0115903615951538,
      "learning_rate": 0.00021445335073450446,
      "loss": 2.1269,
      "step": 136409
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.0662870407104492,
      "learning_rate": 0.00021444943001523728,
      "loss": 2.2465,
      "step": 136410
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.1751304864883423,
      "learning_rate": 0.0002144455093118755,
      "loss": 2.2598,
      "step": 136411
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.1133636236190796,
      "learning_rate": 0.00021444158862442015,
      "loss": 2.2313,
      "step": 136412
    },
    {
      "epoch": 1.78,
      "grad_norm": 0.9411154985427856,
      "learning_rate": 0.00021443766795287153,
      "loss": 2.1917,
      "step": 136413
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.1396335363388062,
      "learning_rate": 0.0002144337472972306,
      "loss": 1.9979,
      "step": 136414
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.0028419494628906,
      "learning_rate": 0.000214429826657498,
      "loss": 2.0993,
      "step": 136415
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.111498475074768,
      "learning_rate": 0.00021442590603367454,
      "loss": 2.3025,
      "step": 136416
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.0464845895767212,
      "learning_rate": 0.00021442198542576094,
      "loss": 2.0995,
      "step": 136417
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.1222460269927979,
      "learning_rate": 0.00021441806483375804,
      "loss": 2.2078,
      "step": 136418
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.1244763135910034,
      "learning_rate": 0.00021441414425766626,
      "loss": 2.1242,
      "step": 136419
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.4724727869033813,
      "learning_rate": 0.00021441022369748656,
      "loss": 2.0248,
      "step": 136420
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.0573291778564453,
      "learning_rate": 0.00021440630315321956,
      "loss": 2.217,
      "step": 136421
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.1154491901397705,
      "learning_rate": 0.00021440238262486607,
      "loss": 2.0016,
      "step": 136422
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.2986607551574707,
      "learning_rate": 0.0002143984621124268,
      "loss": 2.0559,
      "step": 136423
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.012357234954834,
      "learning_rate": 0.00021439454161590257,
      "loss": 2.0185,
      "step": 136424
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.0827698707580566,
      "learning_rate": 0.00021439062113529393,
      "loss": 2.0687,
      "step": 136425
    },
    {
      "epoch": 1.78,
      "grad_norm": 0.9649350643157959,
      "learning_rate": 0.00021438670067060167,
      "loss": 2.0209,
      "step": 136426
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.3368240594863892,
      "learning_rate": 0.0002143827802218265,
      "loss": 2.0436,
      "step": 136427
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.5036766529083252,
      "learning_rate": 0.00021437885978896923,
      "loss": 1.9684,
      "step": 136428
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.122233510017395,
      "learning_rate": 0.00021437493937203055,
      "loss": 2.0964,
      "step": 136429
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.0567086935043335,
      "learning_rate": 0.0002143710189710113,
      "loss": 2.2849,
      "step": 136430
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.4312138557434082,
      "learning_rate": 0.00021436709858591198,
      "loss": 2.1409,
      "step": 136431
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.3305646181106567,
      "learning_rate": 0.0002143631782167334,
      "loss": 2.1481,
      "step": 136432
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.0389008522033691,
      "learning_rate": 0.00021435925786347633,
      "loss": 2.0194,
      "step": 136433
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.012040376663208,
      "learning_rate": 0.00021435533752614148,
      "loss": 2.2266,
      "step": 136434
    },
    {
      "epoch": 1.78,
      "grad_norm": 0.9985736608505249,
      "learning_rate": 0.00021435141720472966,
      "loss": 2.0012,
      "step": 136435
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.0711867809295654,
      "learning_rate": 0.0002143474968992416,
      "loss": 2.1666,
      "step": 136436
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.0703916549682617,
      "learning_rate": 0.00021434357660967782,
      "loss": 2.2579,
      "step": 136437
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.0386220216751099,
      "learning_rate": 0.0002143396563360392,
      "loss": 2.1923,
      "step": 136438
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.051887035369873,
      "learning_rate": 0.00021433573607832648,
      "loss": 1.9941,
      "step": 136439
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.0679404735565186,
      "learning_rate": 0.0002143318158365403,
      "loss": 2.316,
      "step": 136440
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.1062719821929932,
      "learning_rate": 0.00021432789561068148,
      "loss": 2.1019,
      "step": 136441
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.1327179670333862,
      "learning_rate": 0.00021432397540075088,
      "loss": 1.9677,
      "step": 136442
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.0012840032577515,
      "learning_rate": 0.00021432005520674894,
      "loss": 2.2174,
      "step": 136443
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.0278213024139404,
      "learning_rate": 0.0002143161350286765,
      "loss": 2.0241,
      "step": 136444
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.0530805587768555,
      "learning_rate": 0.00021431221486653435,
      "loss": 2.108,
      "step": 136445
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.2104846239089966,
      "learning_rate": 0.0002143082947203231,
      "loss": 2.1387,
      "step": 136446
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.032039999961853,
      "learning_rate": 0.0002143043745900436,
      "loss": 1.9264,
      "step": 136447
    },
    {
      "epoch": 1.78,
      "grad_norm": 0.9763801097869873,
      "learning_rate": 0.00021430045447569658,
      "loss": 1.9655,
      "step": 136448
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.0278187990188599,
      "learning_rate": 0.00021429653437728276,
      "loss": 1.8931,
      "step": 136449
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.1688822507858276,
      "learning_rate": 0.00021429261429480273,
      "loss": 2.185,
      "step": 136450
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.0548135042190552,
      "learning_rate": 0.00021428869422825732,
      "loss": 2.199,
      "step": 136451
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.0543482303619385,
      "learning_rate": 0.00021428477417764728,
      "loss": 2.191,
      "step": 136452
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.0045500993728638,
      "learning_rate": 0.00021428085414297332,
      "loss": 2.1313,
      "step": 136453
    },
    {
      "epoch": 1.78,
      "grad_norm": 0.9101764559745789,
      "learning_rate": 0.00021427693412423617,
      "loss": 2.1066,
      "step": 136454
    },
    {
      "epoch": 1.78,
      "grad_norm": 0.9950029253959656,
      "learning_rate": 0.0002142730141214366,
      "loss": 1.8431,
      "step": 136455
    },
    {
      "epoch": 1.78,
      "grad_norm": 0.9099261164665222,
      "learning_rate": 0.00021426909413457522,
      "loss": 2.1088,
      "step": 136456
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.1458860635757446,
      "learning_rate": 0.00021426517416365292,
      "loss": 2.0021,
      "step": 136457
    },
    {
      "epoch": 1.78,
      "grad_norm": 0.9263100624084473,
      "learning_rate": 0.0002142612542086703,
      "loss": 2.091,
      "step": 136458
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.0650274753570557,
      "learning_rate": 0.00021425733426962812,
      "loss": 2.0806,
      "step": 136459
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.3498584032058716,
      "learning_rate": 0.00021425341434652708,
      "loss": 1.998,
      "step": 136460
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.0092995166778564,
      "learning_rate": 0.00021424949443936807,
      "loss": 2.0771,
      "step": 136461
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.0179626941680908,
      "learning_rate": 0.0002142455745481516,
      "loss": 2.0635,
      "step": 136462
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.4345778226852417,
      "learning_rate": 0.0002142416546728785,
      "loss": 2.0991,
      "step": 136463
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.2011052370071411,
      "learning_rate": 0.00021423773481354954,
      "loss": 2.0251,
      "step": 136464
    },
    {
      "epoch": 1.78,
      "grad_norm": 0.9683561325073242,
      "learning_rate": 0.00021423381497016536,
      "loss": 2.0352,
      "step": 136465
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.0677043199539185,
      "learning_rate": 0.00021422989514272673,
      "loss": 2.1335,
      "step": 136466
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.5126795768737793,
      "learning_rate": 0.00021422597533123444,
      "loss": 2.0234,
      "step": 136467
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.0957684516906738,
      "learning_rate": 0.00021422205553568914,
      "loss": 2.099,
      "step": 136468
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.0027680397033691,
      "learning_rate": 0.00021421813575609152,
      "loss": 2.0035,
      "step": 136469
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.12614905834198,
      "learning_rate": 0.0002142142159924424,
      "loss": 2.3103,
      "step": 136470
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.0880523920059204,
      "learning_rate": 0.00021421029624474255,
      "loss": 2.0569,
      "step": 136471
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.2617172002792358,
      "learning_rate": 0.00021420637651299254,
      "loss": 2.054,
      "step": 136472
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.1061022281646729,
      "learning_rate": 0.00021420245679719326,
      "loss": 2.2924,
      "step": 136473
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.1816835403442383,
      "learning_rate": 0.0002141985370973453,
      "loss": 2.1941,
      "step": 136474
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.1108713150024414,
      "learning_rate": 0.00021419461741344943,
      "loss": 2.1272,
      "step": 136475
    },
    {
      "epoch": 1.78,
      "grad_norm": 0.9784794449806213,
      "learning_rate": 0.0002141906977455064,
      "loss": 2.125,
      "step": 136476
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.1402009725570679,
      "learning_rate": 0.00021418677809351697,
      "loss": 2.0242,
      "step": 136477
    },
    {
      "epoch": 1.78,
      "grad_norm": 0.9889526963233948,
      "learning_rate": 0.00021418285845748182,
      "loss": 2.2799,
      "step": 136478
    },
    {
      "epoch": 1.78,
      "grad_norm": 0.9404022693634033,
      "learning_rate": 0.0002141789388374018,
      "loss": 1.8815,
      "step": 136479
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.1233141422271729,
      "learning_rate": 0.00021417501923327745,
      "loss": 2.0819,
      "step": 136480
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.153082251548767,
      "learning_rate": 0.00021417109964510955,
      "loss": 2.0623,
      "step": 136481
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.0906963348388672,
      "learning_rate": 0.0002141671800728989,
      "loss": 1.991,
      "step": 136482
    },
    {
      "epoch": 1.78,
      "grad_norm": 0.9902524948120117,
      "learning_rate": 0.00021416326051664614,
      "loss": 1.9017,
      "step": 136483
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.0104376077651978,
      "learning_rate": 0.0002141593409763521,
      "loss": 1.8916,
      "step": 136484
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.144284725189209,
      "learning_rate": 0.00021415542145201757,
      "loss": 2.0929,
      "step": 136485
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.044378638267517,
      "learning_rate": 0.00021415150194364303,
      "loss": 2.1274,
      "step": 136486
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.1385527849197388,
      "learning_rate": 0.00021414758245122935,
      "loss": 2.2232,
      "step": 136487
    },
    {
      "epoch": 1.78,
      "grad_norm": 0.9759358167648315,
      "learning_rate": 0.00021414366297477724,
      "loss": 2.1105,
      "step": 136488
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.0614124536514282,
      "learning_rate": 0.00021413974351428745,
      "loss": 2.084,
      "step": 136489
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.0093103647232056,
      "learning_rate": 0.0002141358240697607,
      "loss": 2.0343,
      "step": 136490
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.0552332401275635,
      "learning_rate": 0.00021413190464119786,
      "loss": 2.1019,
      "step": 136491
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.1354800462722778,
      "learning_rate": 0.00021412798522859943,
      "loss": 2.1762,
      "step": 136492
    },
    {
      "epoch": 1.78,
      "grad_norm": 0.9978196620941162,
      "learning_rate": 0.00021412406583196618,
      "loss": 1.8903,
      "step": 136493
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.0923172235488892,
      "learning_rate": 0.00021412014645129894,
      "loss": 2.2238,
      "step": 136494
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.020985722541809,
      "learning_rate": 0.0002141162270865983,
      "loss": 2.1678,
      "step": 136495
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.145004153251648,
      "learning_rate": 0.0002141123077378651,
      "loss": 2.209,
      "step": 136496
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.1170108318328857,
      "learning_rate": 0.00021410838840510025,
      "loss": 2.2002,
      "step": 136497
    },
    {
      "epoch": 1.78,
      "grad_norm": 0.95650315284729,
      "learning_rate": 0.00021410446908830407,
      "loss": 2.1064,
      "step": 136498
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.0898451805114746,
      "learning_rate": 0.00021410054978747753,
      "loss": 2.1401,
      "step": 136499
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.1878693103790283,
      "learning_rate": 0.00021409663050262127,
      "loss": 2.2371,
      "step": 136500
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.0087748765945435,
      "learning_rate": 0.00021409271123373611,
      "loss": 2.145,
      "step": 136501
    },
    {
      "epoch": 1.78,
      "grad_norm": 0.9637881517410278,
      "learning_rate": 0.0002140887919808227,
      "loss": 1.9686,
      "step": 136502
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.1656999588012695,
      "learning_rate": 0.00021408487274388198,
      "loss": 2.2465,
      "step": 136503
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.2361336946487427,
      "learning_rate": 0.00021408095352291434,
      "loss": 2.1897,
      "step": 136504
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.0045475959777832,
      "learning_rate": 0.00021407703431792067,
      "loss": 1.8542,
      "step": 136505
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.137726902961731,
      "learning_rate": 0.00021407311512890173,
      "loss": 1.9853,
      "step": 136506
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.094799518585205,
      "learning_rate": 0.00021406919595585817,
      "loss": 2.1314,
      "step": 136507
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.0895498991012573,
      "learning_rate": 0.0002140652767987908,
      "loss": 2.1616,
      "step": 136508
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.0264688730239868,
      "learning_rate": 0.00021406135765770045,
      "loss": 2.0509,
      "step": 136509
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.129523754119873,
      "learning_rate": 0.0002140574385325876,
      "loss": 2.1114,
      "step": 136510
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.0732402801513672,
      "learning_rate": 0.00021405351942345305,
      "loss": 2.0723,
      "step": 136511
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.0411272048950195,
      "learning_rate": 0.0002140496003302976,
      "loss": 2.0863,
      "step": 136512
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.0876777172088623,
      "learning_rate": 0.000214045681253122,
      "loss": 1.902,
      "step": 136513
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.230560541152954,
      "learning_rate": 0.00021404176219192686,
      "loss": 2.0884,
      "step": 136514
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.1576485633850098,
      "learning_rate": 0.00021403784314671316,
      "loss": 1.9852,
      "step": 136515
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.103825330734253,
      "learning_rate": 0.00021403392411748127,
      "loss": 2.2618,
      "step": 136516
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.0017471313476562,
      "learning_rate": 0.0002140300051042321,
      "loss": 2.1094,
      "step": 136517
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.0988363027572632,
      "learning_rate": 0.0002140260861069664,
      "loss": 2.123,
      "step": 136518
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.033811092376709,
      "learning_rate": 0.0002140221671256849,
      "loss": 2.0291,
      "step": 136519
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.0550181865692139,
      "learning_rate": 0.00021401824816038822,
      "loss": 2.0857,
      "step": 136520
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.0404393672943115,
      "learning_rate": 0.0002140143292110772,
      "loss": 2.0678,
      "step": 136521
    },
    {
      "epoch": 1.78,
      "grad_norm": 0.9887464046478271,
      "learning_rate": 0.00021401041027775274,
      "loss": 2.2061,
      "step": 136522
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.2379791736602783,
      "learning_rate": 0.00021400649136041518,
      "loss": 2.3133,
      "step": 136523
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.0208343267440796,
      "learning_rate": 0.00021400257245906542,
      "loss": 2.2368,
      "step": 136524
    },
    {
      "epoch": 1.78,
      "grad_norm": 0.9957727789878845,
      "learning_rate": 0.00021399865357370424,
      "loss": 2.1041,
      "step": 136525
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.1473522186279297,
      "learning_rate": 0.00021399473470433232,
      "loss": 1.912,
      "step": 136526
    },
    {
      "epoch": 1.78,
      "grad_norm": 0.9677285552024841,
      "learning_rate": 0.00021399081585095039,
      "loss": 2.1029,
      "step": 136527
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.2585512399673462,
      "learning_rate": 0.00021398689701355935,
      "loss": 2.0923,
      "step": 136528
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.0956119298934937,
      "learning_rate": 0.0002139829781921596,
      "loss": 1.9215,
      "step": 136529
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.0676031112670898,
      "learning_rate": 0.0002139790593867521,
      "loss": 2.2274,
      "step": 136530
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.1441712379455566,
      "learning_rate": 0.00021397514059733747,
      "loss": 2.4348,
      "step": 136531
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.1102761030197144,
      "learning_rate": 0.0002139712218239165,
      "loss": 2.1087,
      "step": 136532
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.1891731023788452,
      "learning_rate": 0.0002139673030664899,
      "loss": 2.1721,
      "step": 136533
    },
    {
      "epoch": 1.78,
      "grad_norm": 0.9898141622543335,
      "learning_rate": 0.00021396338432505852,
      "loss": 2.0171,
      "step": 136534
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.1085205078125,
      "learning_rate": 0.00021395946559962285,
      "loss": 2.1981,
      "step": 136535
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.021346092224121,
      "learning_rate": 0.0002139555468901837,
      "loss": 2.2644,
      "step": 136536
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.2049237489700317,
      "learning_rate": 0.0002139516281967419,
      "loss": 2.1101,
      "step": 136537
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.0996577739715576,
      "learning_rate": 0.00021394770951929808,
      "loss": 2.0714,
      "step": 136538
    },
    {
      "epoch": 1.78,
      "grad_norm": 0.9946869015693665,
      "learning_rate": 0.00021394379085785302,
      "loss": 2.002,
      "step": 136539
    },
    {
      "epoch": 1.78,
      "grad_norm": 0.9560784101486206,
      "learning_rate": 0.00021393987221240754,
      "loss": 2.0502,
      "step": 136540
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.0904290676116943,
      "learning_rate": 0.0002139359535829622,
      "loss": 2.0946,
      "step": 136541
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.2529401779174805,
      "learning_rate": 0.00021393203496951777,
      "loss": 2.1321,
      "step": 136542
    },
    {
      "epoch": 1.78,
      "grad_norm": 0.9864333868026733,
      "learning_rate": 0.00021392811637207496,
      "loss": 2.1979,
      "step": 136543
    },
    {
      "epoch": 1.78,
      "grad_norm": 0.9855824708938599,
      "learning_rate": 0.00021392419779063456,
      "loss": 1.9865,
      "step": 136544
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.1359097957611084,
      "learning_rate": 0.00021392027922519723,
      "loss": 2.3346,
      "step": 136545
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.1083753108978271,
      "learning_rate": 0.0002139163606757639,
      "loss": 2.2607,
      "step": 136546
    },
    {
      "epoch": 1.78,
      "grad_norm": 0.9768279194831848,
      "learning_rate": 0.00021391244214233503,
      "loss": 2.1773,
      "step": 136547
    },
    {
      "epoch": 1.78,
      "grad_norm": 0.9552401304244995,
      "learning_rate": 0.00021390852362491143,
      "loss": 1.9697,
      "step": 136548
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.0088201761245728,
      "learning_rate": 0.00021390460512349393,
      "loss": 2.0593,
      "step": 136549
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.036156177520752,
      "learning_rate": 0.0002139006866380832,
      "loss": 1.9753,
      "step": 136550
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.0360735654830933,
      "learning_rate": 0.0002138967681686799,
      "loss": 2.1115,
      "step": 136551
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.1221867799758911,
      "learning_rate": 0.00021389284971528487,
      "loss": 2.0051,
      "step": 136552
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.0292432308197021,
      "learning_rate": 0.00021388893127789873,
      "loss": 2.1295,
      "step": 136553
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.0751739740371704,
      "learning_rate": 0.00021388501285652227,
      "loss": 2.0016,
      "step": 136554
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.0503854751586914,
      "learning_rate": 0.00021388109445115625,
      "loss": 2.1505,
      "step": 136555
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.009209394454956,
      "learning_rate": 0.00021387717606180138,
      "loss": 2.102,
      "step": 136556
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.0417619943618774,
      "learning_rate": 0.0002138732576884583,
      "loss": 2.1075,
      "step": 136557
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.1282832622528076,
      "learning_rate": 0.00021386933933112786,
      "loss": 2.3118,
      "step": 136558
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.045538306236267,
      "learning_rate": 0.00021386542098981073,
      "loss": 2.3145,
      "step": 136559
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.018552541732788,
      "learning_rate": 0.0002138615026645076,
      "loss": 1.9651,
      "step": 136560
    },
    {
      "epoch": 1.78,
      "grad_norm": 0.9534505009651184,
      "learning_rate": 0.00021385758435521927,
      "loss": 2.1226,
      "step": 136561
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.0445863008499146,
      "learning_rate": 0.00021385366606194642,
      "loss": 2.449,
      "step": 136562
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.0749702453613281,
      "learning_rate": 0.00021384974778468983,
      "loss": 1.9918,
      "step": 136563
    },
    {
      "epoch": 1.78,
      "grad_norm": 0.977517306804657,
      "learning_rate": 0.00021384582952345025,
      "loss": 1.9984,
      "step": 136564
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.1274034976959229,
      "learning_rate": 0.0002138419112782283,
      "loss": 2.0766,
      "step": 136565
    },
    {
      "epoch": 1.78,
      "grad_norm": 0.9931370615959167,
      "learning_rate": 0.00021383799304902474,
      "loss": 2.0783,
      "step": 136566
    },
    {
      "epoch": 1.78,
      "grad_norm": 0.9988492131233215,
      "learning_rate": 0.0002138340748358403,
      "loss": 1.9484,
      "step": 136567
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.0096858739852905,
      "learning_rate": 0.00021383015663867578,
      "loss": 2.2947,
      "step": 136568
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.0345324277877808,
      "learning_rate": 0.00021382623845753185,
      "loss": 2.082,
      "step": 136569
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.121628761291504,
      "learning_rate": 0.00021382232029240936,
      "loss": 2.0912,
      "step": 136570
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.1380938291549683,
      "learning_rate": 0.00021381840214330878,
      "loss": 2.0554,
      "step": 136571
    },
    {
      "epoch": 1.78,
      "grad_norm": 0.9563162922859192,
      "learning_rate": 0.000213814484010231,
      "loss": 2.0041,
      "step": 136572
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.0376454591751099,
      "learning_rate": 0.00021381056589317676,
      "loss": 2.2285,
      "step": 136573
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.0974211692810059,
      "learning_rate": 0.00021380664779214675,
      "loss": 2.2513,
      "step": 136574
    },
    {
      "epoch": 1.78,
      "grad_norm": 0.9702708125114441,
      "learning_rate": 0.0002138027297071417,
      "loss": 2.0555,
      "step": 136575
    },
    {
      "epoch": 1.78,
      "grad_norm": 0.9665053486824036,
      "learning_rate": 0.00021379881163816248,
      "loss": 2.1678,
      "step": 136576
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.0527465343475342,
      "learning_rate": 0.0002137948935852096,
      "loss": 1.9831,
      "step": 136577
    },
    {
      "epoch": 1.78,
      "grad_norm": 0.98930823802948,
      "learning_rate": 0.00021379097554828384,
      "loss": 2.0029,
      "step": 136578
    },
    {
      "epoch": 1.78,
      "grad_norm": 0.964553952217102,
      "learning_rate": 0.00021378705752738596,
      "loss": 2.2456,
      "step": 136579
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.1061408519744873,
      "learning_rate": 0.00021378313952251672,
      "loss": 2.0821,
      "step": 136580
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.0397870540618896,
      "learning_rate": 0.00021377922153367678,
      "loss": 2.0854,
      "step": 136581
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.1196763515472412,
      "learning_rate": 0.00021377530356086707,
      "loss": 2.1775,
      "step": 136582
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.0973904132843018,
      "learning_rate": 0.00021377138560408802,
      "loss": 2.1387,
      "step": 136583
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.0337562561035156,
      "learning_rate": 0.00021376746766334052,
      "loss": 1.9268,
      "step": 136584
    },
    {
      "epoch": 1.78,
      "grad_norm": 0.9886370301246643,
      "learning_rate": 0.00021376354973862523,
      "loss": 1.7969,
      "step": 136585
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.04778254032135,
      "learning_rate": 0.00021375963182994294,
      "loss": 2.2169,
      "step": 136586
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.1233489513397217,
      "learning_rate": 0.00021375571393729442,
      "loss": 2.0768,
      "step": 136587
    },
    {
      "epoch": 1.78,
      "grad_norm": 0.9760540127754211,
      "learning_rate": 0.00021375179606068027,
      "loss": 1.9217,
      "step": 136588
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.0806857347488403,
      "learning_rate": 0.00021374787820010145,
      "loss": 2.176,
      "step": 136589
    },
    {
      "epoch": 1.78,
      "grad_norm": 0.9376163482666016,
      "learning_rate": 0.00021374396035555836,
      "loss": 1.9729,
      "step": 136590
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.2961314916610718,
      "learning_rate": 0.0002137400425270519,
      "loss": 2.0435,
      "step": 136591
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.1146717071533203,
      "learning_rate": 0.00021373612471458286,
      "loss": 2.2766,
      "step": 136592
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.0114251375198364,
      "learning_rate": 0.00021373220691815186,
      "loss": 1.9414,
      "step": 136593
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.1259833574295044,
      "learning_rate": 0.00021372828913775968,
      "loss": 2.2534,
      "step": 136594
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.0474801063537598,
      "learning_rate": 0.00021372437137340714,
      "loss": 2.0469,
      "step": 136595
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.0610007047653198,
      "learning_rate": 0.00021372045362509477,
      "loss": 2.138,
      "step": 136596
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.081086277961731,
      "learning_rate": 0.0002137165358928234,
      "loss": 1.9484,
      "step": 136597
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.1024267673492432,
      "learning_rate": 0.00021371261817659373,
      "loss": 2.1521,
      "step": 136598
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.113323450088501,
      "learning_rate": 0.0002137087004764065,
      "loss": 2.2217,
      "step": 136599
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.4913272857666016,
      "learning_rate": 0.0002137047827922625,
      "loss": 1.8887,
      "step": 136600
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.0894023180007935,
      "learning_rate": 0.0002137008651241625,
      "loss": 2.1032,
      "step": 136601
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.1381076574325562,
      "learning_rate": 0.00021369694747210705,
      "loss": 2.2963,
      "step": 136602
    },
    {
      "epoch": 1.78,
      "grad_norm": 0.9721309542655945,
      "learning_rate": 0.0002136930298360969,
      "loss": 2.0039,
      "step": 136603
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.3129665851593018,
      "learning_rate": 0.0002136891122161329,
      "loss": 2.1073,
      "step": 136604
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.1109716892242432,
      "learning_rate": 0.00021368519461221574,
      "loss": 2.0576,
      "step": 136605
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.1910200119018555,
      "learning_rate": 0.0002136812770243461,
      "loss": 2.3553,
      "step": 136606
    },
    {
      "epoch": 1.78,
      "grad_norm": 0.9900075793266296,
      "learning_rate": 0.00021367735945252483,
      "loss": 1.9754,
      "step": 136607
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.010783314704895,
      "learning_rate": 0.0002136734418967525,
      "loss": 2.1351,
      "step": 136608
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.0839098691940308,
      "learning_rate": 0.0002136695243570299,
      "loss": 2.189,
      "step": 136609
    },
    {
      "epoch": 1.78,
      "grad_norm": 0.9806008338928223,
      "learning_rate": 0.00021366560683335771,
      "loss": 2.2754,
      "step": 136610
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.0376975536346436,
      "learning_rate": 0.0002136616893257368,
      "loss": 2.0218,
      "step": 136611
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.012303113937378,
      "learning_rate": 0.00021365777183416772,
      "loss": 2.2135,
      "step": 136612
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.5613110065460205,
      "learning_rate": 0.0002136538543586515,
      "loss": 2.0935,
      "step": 136613
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.244903326034546,
      "learning_rate": 0.00021364993689918847,
      "loss": 2.0755,
      "step": 136614
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.0243723392486572,
      "learning_rate": 0.0002136460194557796,
      "loss": 1.9628,
      "step": 136615
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.225782871246338,
      "learning_rate": 0.00021364210202842547,
      "loss": 2.0905,
      "step": 136616
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.0193045139312744,
      "learning_rate": 0.000213638184617127,
      "loss": 2.3506,
      "step": 136617
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.0522915124893188,
      "learning_rate": 0.00021363426722188482,
      "loss": 1.8595,
      "step": 136618
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.008069396018982,
      "learning_rate": 0.00021363034984269973,
      "loss": 2.0063,
      "step": 136619
    },
    {
      "epoch": 1.78,
      "grad_norm": 0.9901722073554993,
      "learning_rate": 0.0002136264324795723,
      "loss": 2.2571,
      "step": 136620
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.0333609580993652,
      "learning_rate": 0.00021362251513250334,
      "loss": 1.9418,
      "step": 136621
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.1048787832260132,
      "learning_rate": 0.00021361859780149358,
      "loss": 1.9969,
      "step": 136622
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.1228631734848022,
      "learning_rate": 0.00021361468048654373,
      "loss": 2.0708,
      "step": 136623
    },
    {
      "epoch": 1.78,
      "grad_norm": 0.981429398059845,
      "learning_rate": 0.00021361076318765458,
      "loss": 1.8472,
      "step": 136624
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.0437489748001099,
      "learning_rate": 0.00021360684590482687,
      "loss": 2.0787,
      "step": 136625
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.0169334411621094,
      "learning_rate": 0.0002136029286380613,
      "loss": 1.8459,
      "step": 136626
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.0155776739120483,
      "learning_rate": 0.00021359901138735848,
      "loss": 2.0036,
      "step": 136627
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.0104409456253052,
      "learning_rate": 0.00021359509415271922,
      "loss": 2.0431,
      "step": 136628
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.0188472270965576,
      "learning_rate": 0.00021359117693414432,
      "loss": 2.046,
      "step": 136629
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.1345986127853394,
      "learning_rate": 0.00021358725973163442,
      "loss": 2.1634,
      "step": 136630
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.0999675989151,
      "learning_rate": 0.00021358334254519034,
      "loss": 2.2312,
      "step": 136631
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.0882378816604614,
      "learning_rate": 0.00021357942537481267,
      "loss": 2.14,
      "step": 136632
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.1232516765594482,
      "learning_rate": 0.00021357550822050228,
      "loss": 2.3094,
      "step": 136633
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.0153400897979736,
      "learning_rate": 0.0002135715910822598,
      "loss": 2.0477,
      "step": 136634
    },
    {
      "epoch": 1.78,
      "grad_norm": 0.98380446434021,
      "learning_rate": 0.000213567673960086,
      "loss": 1.8152,
      "step": 136635
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.1374375820159912,
      "learning_rate": 0.00021356375685398156,
      "loss": 1.9341,
      "step": 136636
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.1917647123336792,
      "learning_rate": 0.00021355983976394734,
      "loss": 1.9842,
      "step": 136637
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.077205777168274,
      "learning_rate": 0.0002135559226899839,
      "loss": 1.9253,
      "step": 136638
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.0013185739517212,
      "learning_rate": 0.00021355200563209206,
      "loss": 2.188,
      "step": 136639
    },
    {
      "epoch": 1.78,
      "grad_norm": 0.8925473690032959,
      "learning_rate": 0.00021354808859027255,
      "loss": 1.8213,
      "step": 136640
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.031925082206726,
      "learning_rate": 0.00021354417156452612,
      "loss": 2.0935,
      "step": 136641
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.0299170017242432,
      "learning_rate": 0.00021354025455485337,
      "loss": 2.0969,
      "step": 136642
    },
    {
      "epoch": 1.78,
      "grad_norm": 0.9047751426696777,
      "learning_rate": 0.00021353633756125524,
      "loss": 2.0062,
      "step": 136643
    },
    {
      "epoch": 1.78,
      "grad_norm": 0.9483408331871033,
      "learning_rate": 0.00021353242058373223,
      "loss": 2.2211,
      "step": 136644
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.0823817253112793,
      "learning_rate": 0.00021352850362228517,
      "loss": 2.3158,
      "step": 136645
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.079154133796692,
      "learning_rate": 0.00021352458667691485,
      "loss": 1.9506,
      "step": 136646
    },
    {
      "epoch": 1.78,
      "grad_norm": 0.9444893598556519,
      "learning_rate": 0.00021352066974762187,
      "loss": 1.9661,
      "step": 136647
    },
    {
      "epoch": 1.78,
      "grad_norm": 0.9603748917579651,
      "learning_rate": 0.00021351675283440717,
      "loss": 2.1705,
      "step": 136648
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.0160934925079346,
      "learning_rate": 0.00021351283593727127,
      "loss": 2.0332,
      "step": 136649
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.041683316230774,
      "learning_rate": 0.00021350891905621498,
      "loss": 2.2024,
      "step": 136650
    },
    {
      "epoch": 1.78,
      "grad_norm": 0.9754505753517151,
      "learning_rate": 0.00021350500219123893,
      "loss": 2.2249,
      "step": 136651
    },
    {
      "epoch": 1.78,
      "grad_norm": 0.977175235748291,
      "learning_rate": 0.00021350108534234398,
      "loss": 1.9152,
      "step": 136652
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.1104011535644531,
      "learning_rate": 0.00021349716850953078,
      "loss": 2.1076,
      "step": 136653
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.0688755512237549,
      "learning_rate": 0.0002134932516928001,
      "loss": 2.1098,
      "step": 136654
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.006661057472229,
      "learning_rate": 0.00021348933489215272,
      "loss": 2.132,
      "step": 136655
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.0222232341766357,
      "learning_rate": 0.00021348541810758934,
      "loss": 2.2608,
      "step": 136656
    },
    {
      "epoch": 1.78,
      "grad_norm": 0.9670907258987427,
      "learning_rate": 0.00021348150133911054,
      "loss": 2.0321,
      "step": 136657
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.0396816730499268,
      "learning_rate": 0.0002134775845867172,
      "loss": 2.0671,
      "step": 136658
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.1489133834838867,
      "learning_rate": 0.00021347366785041,
      "loss": 1.9338,
      "step": 136659
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.02952241897583,
      "learning_rate": 0.0002134697511301897,
      "loss": 2.0705,
      "step": 136660
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.0775572061538696,
      "learning_rate": 0.00021346583442605698,
      "loss": 2.5014,
      "step": 136661
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.0395305156707764,
      "learning_rate": 0.00021346191773801272,
      "loss": 2.3976,
      "step": 136662
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.0877693891525269,
      "learning_rate": 0.00021345800106605737,
      "loss": 2.0922,
      "step": 136663
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.096643328666687,
      "learning_rate": 0.00021345408441019186,
      "loss": 2.1932,
      "step": 136664
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.0507481098175049,
      "learning_rate": 0.00021345016777041685,
      "loss": 2.0892,
      "step": 136665
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.0512019395828247,
      "learning_rate": 0.0002134462511467331,
      "loss": 2.2639,
      "step": 136666
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.1651755571365356,
      "learning_rate": 0.00021344233453914134,
      "loss": 2.0476,
      "step": 136667
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.0536986589431763,
      "learning_rate": 0.00021343841794764233,
      "loss": 2.116,
      "step": 136668
    },
    {
      "epoch": 1.78,
      "grad_norm": 0.9404696822166443,
      "learning_rate": 0.00021343450137223667,
      "loss": 2.1298,
      "step": 136669
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.0273975133895874,
      "learning_rate": 0.0002134305848129252,
      "loss": 2.1413,
      "step": 136670
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.092658519744873,
      "learning_rate": 0.0002134266682697086,
      "loss": 1.9366,
      "step": 136671
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.0365008115768433,
      "learning_rate": 0.00021342275174258764,
      "loss": 2.1499,
      "step": 136672
    },
    {
      "epoch": 1.78,
      "grad_norm": 0.946478545665741,
      "learning_rate": 0.000213418835231563,
      "loss": 1.9734,
      "step": 136673
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.0783859491348267,
      "learning_rate": 0.00021341491873663556,
      "loss": 2.3152,
      "step": 136674
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.0647962093353271,
      "learning_rate": 0.00021341100225780578,
      "loss": 2.2855,
      "step": 136675
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.1161545515060425,
      "learning_rate": 0.00021340708579507456,
      "loss": 1.9634,
      "step": 136676
    },
    {
      "epoch": 1.78,
      "grad_norm": 0.9416982531547546,
      "learning_rate": 0.00021340316934844255,
      "loss": 2.0552,
      "step": 136677
    },
    {
      "epoch": 1.78,
      "grad_norm": 0.9163469672203064,
      "learning_rate": 0.00021339925291791057,
      "loss": 2.0961,
      "step": 136678
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.087009310722351,
      "learning_rate": 0.0002133953365034793,
      "loss": 1.8481,
      "step": 136679
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.1944504976272583,
      "learning_rate": 0.0002133914201051496,
      "loss": 2.1578,
      "step": 136680
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.1254057884216309,
      "learning_rate": 0.0002133875037229219,
      "loss": 2.2095,
      "step": 136681
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.1145097017288208,
      "learning_rate": 0.00021338358735679714,
      "loss": 2.0871,
      "step": 136682
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.0660306215286255,
      "learning_rate": 0.000213379671006776,
      "loss": 2.1778,
      "step": 136683
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.2515497207641602,
      "learning_rate": 0.0002133757546728592,
      "loss": 2.027,
      "step": 136684
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.011562466621399,
      "learning_rate": 0.0002133718383550475,
      "loss": 1.9587,
      "step": 136685
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.0971863269805908,
      "learning_rate": 0.00021336792205334172,
      "loss": 2.1767,
      "step": 136686
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.1257909536361694,
      "learning_rate": 0.00021336400576774236,
      "loss": 2.0489,
      "step": 136687
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.2153149843215942,
      "learning_rate": 0.00021336008949825029,
      "loss": 2.0352,
      "step": 136688
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.0343858003616333,
      "learning_rate": 0.00021335617324486615,
      "loss": 2.3139,
      "step": 136689
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.0110316276550293,
      "learning_rate": 0.00021335225700759081,
      "loss": 2.0441,
      "step": 136690
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.0629446506500244,
      "learning_rate": 0.00021334834078642485,
      "loss": 1.9784,
      "step": 136691
    },
    {
      "epoch": 1.78,
      "grad_norm": 0.9994521737098694,
      "learning_rate": 0.00021334442458136924,
      "loss": 2.2313,
      "step": 136692
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.1641736030578613,
      "learning_rate": 0.00021334050839242442,
      "loss": 2.1336,
      "step": 136693
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.027609944343567,
      "learning_rate": 0.00021333659221959122,
      "loss": 1.8499,
      "step": 136694
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.0291621685028076,
      "learning_rate": 0.00021333267606287037,
      "loss": 1.8929,
      "step": 136695
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.0615097284317017,
      "learning_rate": 0.00021332875992226264,
      "loss": 2.2349,
      "step": 136696
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.0298479795455933,
      "learning_rate": 0.00021332484379776872,
      "loss": 2.1791,
      "step": 136697
    },
    {
      "epoch": 1.78,
      "grad_norm": 0.9524530172348022,
      "learning_rate": 0.00021332092768938944,
      "loss": 2.011,
      "step": 136698
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.0873242616653442,
      "learning_rate": 0.00021331701159712536,
      "loss": 2.0749,
      "step": 136699
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.0864477157592773,
      "learning_rate": 0.00021331309552097725,
      "loss": 2.0678,
      "step": 136700
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.0841443538665771,
      "learning_rate": 0.0002133091794609459,
      "loss": 2.1775,
      "step": 136701
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.1882498264312744,
      "learning_rate": 0.00021330526341703195,
      "loss": 2.0755,
      "step": 136702
    },
    {
      "epoch": 1.78,
      "grad_norm": 0.910348117351532,
      "learning_rate": 0.00021330134738923624,
      "loss": 2.0946,
      "step": 136703
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.1424634456634521,
      "learning_rate": 0.00021329743137755958,
      "loss": 2.2622,
      "step": 136704
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.326194167137146,
      "learning_rate": 0.00021329351538200243,
      "loss": 2.1762,
      "step": 136705
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.0707210302352905,
      "learning_rate": 0.00021328959940256564,
      "loss": 2.1527,
      "step": 136706
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.0366618633270264,
      "learning_rate": 0.00021328568343924998,
      "loss": 1.92,
      "step": 136707
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.2472968101501465,
      "learning_rate": 0.00021328176749205608,
      "loss": 2.258,
      "step": 136708
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.2820196151733398,
      "learning_rate": 0.00021327785156098478,
      "loss": 1.9994,
      "step": 136709
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.0599576234817505,
      "learning_rate": 0.00021327393564603682,
      "loss": 2.1687,
      "step": 136710
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.3589143753051758,
      "learning_rate": 0.00021327001974721287,
      "loss": 2.0756,
      "step": 136711
    },
    {
      "epoch": 1.78,
      "grad_norm": 0.9853792786598206,
      "learning_rate": 0.00021326610386451363,
      "loss": 2.1673,
      "step": 136712
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.0295374393463135,
      "learning_rate": 0.00021326218799793984,
      "loss": 2.0508,
      "step": 136713
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.014419674873352,
      "learning_rate": 0.0002132582721474922,
      "loss": 2.2567,
      "step": 136714
    },
    {
      "epoch": 1.78,
      "grad_norm": 0.9553385376930237,
      "learning_rate": 0.00021325435631317153,
      "loss": 2.0921,
      "step": 136715
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.0063810348510742,
      "learning_rate": 0.00021325044049497858,
      "loss": 2.041,
      "step": 136716
    },
    {
      "epoch": 1.78,
      "grad_norm": 0.9759901762008667,
      "learning_rate": 0.00021324652469291398,
      "loss": 1.9511,
      "step": 136717
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.1995359659194946,
      "learning_rate": 0.00021324260890697846,
      "loss": 2.0816,
      "step": 136718
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.0532180070877075,
      "learning_rate": 0.0002132386931371728,
      "loss": 2.2878,
      "step": 136719
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.1940163373947144,
      "learning_rate": 0.00021323477738349763,
      "loss": 2.2769,
      "step": 136720
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.076411247253418,
      "learning_rate": 0.0002132308616459538,
      "loss": 1.9641,
      "step": 136721
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.038703441619873,
      "learning_rate": 0.000213226945924542,
      "loss": 2.0795,
      "step": 136722
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.006217360496521,
      "learning_rate": 0.00021322303021926298,
      "loss": 2.2592,
      "step": 136723
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.0548877716064453,
      "learning_rate": 0.00021321911453011738,
      "loss": 2.0416,
      "step": 136724
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.251150369644165,
      "learning_rate": 0.00021321519885710607,
      "loss": 1.9728,
      "step": 136725
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.1558200120925903,
      "learning_rate": 0.0002132112832002296,
      "loss": 2.1532,
      "step": 136726
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.0813299417495728,
      "learning_rate": 0.0002132073675594888,
      "loss": 2.1892,
      "step": 136727
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.0809626579284668,
      "learning_rate": 0.00021320345193488442,
      "loss": 2.2081,
      "step": 136728
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.14500093460083,
      "learning_rate": 0.00021319953632641716,
      "loss": 1.9336,
      "step": 136729
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.0935757160186768,
      "learning_rate": 0.0002131956207340877,
      "loss": 2.1517,
      "step": 136730
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.1018784046173096,
      "learning_rate": 0.00021319170515789686,
      "loss": 2.1874,
      "step": 136731
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.0606704950332642,
      "learning_rate": 0.00021318778959784538,
      "loss": 1.9732,
      "step": 136732
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.1545368432998657,
      "learning_rate": 0.0002131838740539338,
      "loss": 2.1776,
      "step": 136733
    },
    {
      "epoch": 1.78,
      "grad_norm": 0.9968852996826172,
      "learning_rate": 0.000213179958526163,
      "loss": 1.9001,
      "step": 136734
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.5447548627853394,
      "learning_rate": 0.00021317604301453378,
      "loss": 2.0203,
      "step": 136735
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.0282315015792847,
      "learning_rate": 0.0002131721275190467,
      "loss": 2.1106,
      "step": 136736
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.1505014896392822,
      "learning_rate": 0.00021316821203970254,
      "loss": 2.0873,
      "step": 136737
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.1594409942626953,
      "learning_rate": 0.00021316429657650208,
      "loss": 2.0424,
      "step": 136738
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.049047589302063,
      "learning_rate": 0.00021316038112944595,
      "loss": 1.772,
      "step": 136739
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.0515258312225342,
      "learning_rate": 0.00021315646569853506,
      "loss": 2.012,
      "step": 136740
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.0906866788864136,
      "learning_rate": 0.00021315255028377006,
      "loss": 2.193,
      "step": 136741
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.2431879043579102,
      "learning_rate": 0.00021314863488515153,
      "loss": 1.9359,
      "step": 136742
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.072190523147583,
      "learning_rate": 0.00021314471950268027,
      "loss": 2.1488,
      "step": 136743
    },
    {
      "epoch": 1.78,
      "grad_norm": 0.9688288569450378,
      "learning_rate": 0.00021314080413635712,
      "loss": 2.0061,
      "step": 136744
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.2011696100234985,
      "learning_rate": 0.00021313688878618267,
      "loss": 2.0283,
      "step": 136745
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.068919062614441,
      "learning_rate": 0.00021313297345215774,
      "loss": 2.03,
      "step": 136746
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.050047755241394,
      "learning_rate": 0.00021312905813428317,
      "loss": 1.9843,
      "step": 136747
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.109126329421997,
      "learning_rate": 0.00021312514283255944,
      "loss": 2.1094,
      "step": 136748
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.1043487787246704,
      "learning_rate": 0.00021312122754698734,
      "loss": 2.0335,
      "step": 136749
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.241755485534668,
      "learning_rate": 0.00021311731227756768,
      "loss": 2.1276,
      "step": 136750
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.3063757419586182,
      "learning_rate": 0.00021311339702430112,
      "loss": 2.1082,
      "step": 136751
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.0730606317520142,
      "learning_rate": 0.00021310948178718844,
      "loss": 2.1541,
      "step": 136752
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.2392712831497192,
      "learning_rate": 0.00021310556656623045,
      "loss": 1.9202,
      "step": 136753
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.1098542213439941,
      "learning_rate": 0.00021310165136142765,
      "loss": 2.1328,
      "step": 136754
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.160718560218811,
      "learning_rate": 0.00021309773617278095,
      "loss": 2.1782,
      "step": 136755
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.0176975727081299,
      "learning_rate": 0.00021309382100029092,
      "loss": 2.2105,
      "step": 136756
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.0213420391082764,
      "learning_rate": 0.00021308990584395846,
      "loss": 2.0945,
      "step": 136757
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.2471166849136353,
      "learning_rate": 0.00021308599070378424,
      "loss": 2.1734,
      "step": 136758
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.0484052896499634,
      "learning_rate": 0.00021308207557976908,
      "loss": 2.0017,
      "step": 136759
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.0646255016326904,
      "learning_rate": 0.00021307816047191346,
      "loss": 2.1457,
      "step": 136760
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.1149744987487793,
      "learning_rate": 0.00021307424538021826,
      "loss": 2.1891,
      "step": 136761
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.0550861358642578,
      "learning_rate": 0.00021307033030468423,
      "loss": 2.0549,
      "step": 136762
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.0001276731491089,
      "learning_rate": 0.000213066415245312,
      "loss": 2.232,
      "step": 136763
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.2026256322860718,
      "learning_rate": 0.00021306250020210245,
      "loss": 2.2014,
      "step": 136764
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.1308634281158447,
      "learning_rate": 0.0002130585851750563,
      "loss": 2.0628,
      "step": 136765
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.0751008987426758,
      "learning_rate": 0.00021305467016417406,
      "loss": 2.0075,
      "step": 136766
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.3518317937850952,
      "learning_rate": 0.0002130507551694566,
      "loss": 2.0177,
      "step": 136767
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.0660383701324463,
      "learning_rate": 0.00021304684019090466,
      "loss": 1.9417,
      "step": 136768
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.1150132417678833,
      "learning_rate": 0.00021304292522851897,
      "loss": 2.0894,
      "step": 136769
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.1168605089187622,
      "learning_rate": 0.00021303901028230023,
      "loss": 2.1301,
      "step": 136770
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.1945370435714722,
      "learning_rate": 0.0002130350953522493,
      "loss": 2.4476,
      "step": 136771
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.314819097518921,
      "learning_rate": 0.00021303118043836667,
      "loss": 2.1166,
      "step": 136772
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.1385853290557861,
      "learning_rate": 0.00021302726554065315,
      "loss": 2.0796,
      "step": 136773
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.0982260704040527,
      "learning_rate": 0.00021302335065910953,
      "loss": 2.1194,
      "step": 136774
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.3769675493240356,
      "learning_rate": 0.00021301943579373652,
      "loss": 2.1097,
      "step": 136775
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.5607942342758179,
      "learning_rate": 0.00021301552094453484,
      "loss": 1.7939,
      "step": 136776
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.1421955823898315,
      "learning_rate": 0.00021301160611150532,
      "loss": 2.0687,
      "step": 136777
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.0394606590270996,
      "learning_rate": 0.00021300769129464848,
      "loss": 2.2127,
      "step": 136778
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.1731775999069214,
      "learning_rate": 0.00021300377649396512,
      "loss": 1.9556,
      "step": 136779
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.3464469909667969,
      "learning_rate": 0.00021299986170945603,
      "loss": 2.2179,
      "step": 136780
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.5568525791168213,
      "learning_rate": 0.00021299594694112193,
      "loss": 1.8931,
      "step": 136781
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.0747264623641968,
      "learning_rate": 0.00021299203218896347,
      "loss": 1.9167,
      "step": 136782
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.1271904706954956,
      "learning_rate": 0.0002129881174529816,
      "loss": 2.2292,
      "step": 136783
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.1095918416976929,
      "learning_rate": 0.00021298420273317676,
      "loss": 2.1386,
      "step": 136784
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.516506314277649,
      "learning_rate": 0.0002129802880295498,
      "loss": 2.1212,
      "step": 136785
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.3757071495056152,
      "learning_rate": 0.00021297637334210138,
      "loss": 1.9545,
      "step": 136786
    },
    {
      "epoch": 1.78,
      "grad_norm": 0.9665845632553101,
      "learning_rate": 0.00021297245867083237,
      "loss": 2.305,
      "step": 136787
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.1332632303237915,
      "learning_rate": 0.00021296854401574338,
      "loss": 2.3473,
      "step": 136788
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.2033674716949463,
      "learning_rate": 0.00021296462937683518,
      "loss": 1.7991,
      "step": 136789
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.127970576286316,
      "learning_rate": 0.00021296071475410868,
      "loss": 2.139,
      "step": 136790
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.1625479459762573,
      "learning_rate": 0.00021295680014756427,
      "loss": 2.1778,
      "step": 136791
    },
    {
      "epoch": 1.78,
      "grad_norm": 0.9240568280220032,
      "learning_rate": 0.00021295288555720282,
      "loss": 2.0366,
      "step": 136792
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.0850026607513428,
      "learning_rate": 0.00021294897098302507,
      "loss": 1.9496,
      "step": 136793
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.2158191204071045,
      "learning_rate": 0.00021294505642503172,
      "loss": 2.1712,
      "step": 136794
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.014031171798706,
      "learning_rate": 0.0002129411418832236,
      "loss": 2.1191,
      "step": 136795
    },
    {
      "epoch": 1.78,
      "grad_norm": 0.9301000237464905,
      "learning_rate": 0.00021293722735760143,
      "loss": 2.1571,
      "step": 136796
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.404036045074463,
      "learning_rate": 0.00021293331284816578,
      "loss": 2.0283,
      "step": 136797
    },
    {
      "epoch": 1.78,
      "grad_norm": 0.9842972755432129,
      "learning_rate": 0.0002129293983549175,
      "loss": 2.1364,
      "step": 136798
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.1700658798217773,
      "learning_rate": 0.00021292548387785723,
      "loss": 2.2328,
      "step": 136799
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.1405835151672363,
      "learning_rate": 0.00021292156941698576,
      "loss": 2.1348,
      "step": 136800
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.2836644649505615,
      "learning_rate": 0.00021291765497230382,
      "loss": 2.3191,
      "step": 136801
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.2791011333465576,
      "learning_rate": 0.0002129137405438122,
      "loss": 2.2308,
      "step": 136802
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.0045725107192993,
      "learning_rate": 0.00021290982613151157,
      "loss": 1.8539,
      "step": 136803
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.09052312374115,
      "learning_rate": 0.00021290591173540258,
      "loss": 2.1025,
      "step": 136804
    },
    {
      "epoch": 1.78,
      "grad_norm": 0.9903790950775146,
      "learning_rate": 0.00021290199735548605,
      "loss": 2.2302,
      "step": 136805
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.0499958992004395,
      "learning_rate": 0.00021289808299176265,
      "loss": 2.2003,
      "step": 136806
    },
    {
      "epoch": 1.78,
      "grad_norm": 0.9806048274040222,
      "learning_rate": 0.00021289416864423317,
      "loss": 1.936,
      "step": 136807
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.084146499633789,
      "learning_rate": 0.00021289025431289833,
      "loss": 2.1251,
      "step": 136808
    },
    {
      "epoch": 1.78,
      "grad_norm": 0.9929164052009583,
      "learning_rate": 0.0002128863399977588,
      "loss": 2.1018,
      "step": 136809
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.0084874629974365,
      "learning_rate": 0.0002128824256988154,
      "loss": 2.2261,
      "step": 136810
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.0638765096664429,
      "learning_rate": 0.00021287851141606873,
      "loss": 1.8975,
      "step": 136811
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.1416412591934204,
      "learning_rate": 0.0002128745971495196,
      "loss": 2.1704,
      "step": 136812
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.0298038721084595,
      "learning_rate": 0.00021287068289916873,
      "loss": 2.0761,
      "step": 136813
    },
    {
      "epoch": 1.78,
      "grad_norm": 0.9973273277282715,
      "learning_rate": 0.0002128667686650169,
      "loss": 1.997,
      "step": 136814
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.1256693601608276,
      "learning_rate": 0.0002128628544470647,
      "loss": 2.2426,
      "step": 136815
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.018284559249878,
      "learning_rate": 0.000212858940245313,
      "loss": 2.1082,
      "step": 136816
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.030876636505127,
      "learning_rate": 0.00021285502605976247,
      "loss": 2.2227,
      "step": 136817
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.1485739946365356,
      "learning_rate": 0.0002128511118904138,
      "loss": 2.0905,
      "step": 136818
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.0773251056671143,
      "learning_rate": 0.00021284719773726773,
      "loss": 2.1555,
      "step": 136819
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.1358356475830078,
      "learning_rate": 0.0002128432836003251,
      "loss": 2.0984,
      "step": 136820
    },
    {
      "epoch": 1.78,
      "grad_norm": 0.9604291319847107,
      "learning_rate": 0.0002128393694795865,
      "loss": 1.8973,
      "step": 136821
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.0398077964782715,
      "learning_rate": 0.0002128354553750527,
      "loss": 2.2868,
      "step": 136822
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.098781704902649,
      "learning_rate": 0.00021283154128672437,
      "loss": 2.0501,
      "step": 136823
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.4248017072677612,
      "learning_rate": 0.00021282762721460243,
      "loss": 2.0008,
      "step": 136824
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.0801631212234497,
      "learning_rate": 0.0002128237131586874,
      "loss": 2.0431,
      "step": 136825
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.1163058280944824,
      "learning_rate": 0.00021281979911898016,
      "loss": 2.115,
      "step": 136826
    },
    {
      "epoch": 1.78,
      "grad_norm": 0.9925252199172974,
      "learning_rate": 0.0002128158850954813,
      "loss": 1.9584,
      "step": 136827
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.4244171380996704,
      "learning_rate": 0.00021281197108819158,
      "loss": 2.2697,
      "step": 136828
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.0750678777694702,
      "learning_rate": 0.00021280805709711183,
      "loss": 2.1518,
      "step": 136829
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.052877426147461,
      "learning_rate": 0.00021280414312224263,
      "loss": 1.9783,
      "step": 136830
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.1594998836517334,
      "learning_rate": 0.00021280022916358483,
      "loss": 2.0961,
      "step": 136831
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.0103187561035156,
      "learning_rate": 0.00021279631522113922,
      "loss": 2.128,
      "step": 136832
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.0171594619750977,
      "learning_rate": 0.0002127924012949063,
      "loss": 1.9836,
      "step": 136833
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.0750783681869507,
      "learning_rate": 0.00021278848738488692,
      "loss": 2.0214,
      "step": 136834
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.1393369436264038,
      "learning_rate": 0.00021278457349108184,
      "loss": 2.3247,
      "step": 136835
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.0564044713974,
      "learning_rate": 0.00021278065961349168,
      "loss": 2.1101,
      "step": 136836
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.2991187572479248,
      "learning_rate": 0.00021277674575211728,
      "loss": 2.1178,
      "step": 136837
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.276003360748291,
      "learning_rate": 0.00021277283190695949,
      "loss": 1.972,
      "step": 136838
    },
    {
      "epoch": 1.78,
      "grad_norm": 0.96706223487854,
      "learning_rate": 0.00021276891807801873,
      "loss": 2.0538,
      "step": 136839
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.0732824802398682,
      "learning_rate": 0.00021276500426529587,
      "loss": 2.0742,
      "step": 136840
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.2215403318405151,
      "learning_rate": 0.00021276109046879162,
      "loss": 2.0494,
      "step": 136841
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.0650954246520996,
      "learning_rate": 0.00021275717668850676,
      "loss": 2.192,
      "step": 136842
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.2595806121826172,
      "learning_rate": 0.00021275326292444197,
      "loss": 2.3176,
      "step": 136843
    },
    {
      "epoch": 1.78,
      "grad_norm": 0.9694890975952148,
      "learning_rate": 0.00021274934917659815,
      "loss": 2.0953,
      "step": 136844
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.0182565450668335,
      "learning_rate": 0.0002127454354449757,
      "loss": 2.0268,
      "step": 136845
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.0146769285202026,
      "learning_rate": 0.00021274152172957555,
      "loss": 2.4151,
      "step": 136846
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.0183148384094238,
      "learning_rate": 0.00021273760803039842,
      "loss": 2.0385,
      "step": 136847
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.0004432201385498,
      "learning_rate": 0.00021273369434744496,
      "loss": 2.0794,
      "step": 136848
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.0435569286346436,
      "learning_rate": 0.000212729780680716,
      "loss": 1.8434,
      "step": 136849
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.0601149797439575,
      "learning_rate": 0.00021272586703021234,
      "loss": 2.1563,
      "step": 136850
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.1990159749984741,
      "learning_rate": 0.00021272195339593444,
      "loss": 2.1724,
      "step": 136851
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.0612834692001343,
      "learning_rate": 0.00021271803977788317,
      "loss": 2.1849,
      "step": 136852
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.047473430633545,
      "learning_rate": 0.0002127141261760593,
      "loss": 1.9562,
      "step": 136853
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.0147950649261475,
      "learning_rate": 0.0002127102125904635,
      "loss": 2.0467,
      "step": 136854
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.0437077283859253,
      "learning_rate": 0.00021270629902109652,
      "loss": 2.091,
      "step": 136855
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.0298993587493896,
      "learning_rate": 0.0002127023854679591,
      "loss": 2.0787,
      "step": 136856
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.0836302042007446,
      "learning_rate": 0.0002126984719310521,
      "loss": 2.2772,
      "step": 136857
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.170945644378662,
      "learning_rate": 0.00021269455841037592,
      "loss": 2.2847,
      "step": 136858
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.2102556228637695,
      "learning_rate": 0.0002126906449059315,
      "loss": 2.0099,
      "step": 136859
    },
    {
      "epoch": 1.78,
      "grad_norm": 0.9276071190834045,
      "learning_rate": 0.00021268673141771954,
      "loss": 2.1731,
      "step": 136860
    },
    {
      "epoch": 1.78,
      "grad_norm": 0.977035641670227,
      "learning_rate": 0.00021268281794574077,
      "loss": 2.0879,
      "step": 136861
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.0997185707092285,
      "learning_rate": 0.0002126789044899959,
      "loss": 1.9237,
      "step": 136862
    },
    {
      "epoch": 1.78,
      "grad_norm": 0.9818393588066101,
      "learning_rate": 0.00021267499105048578,
      "loss": 1.878,
      "step": 136863
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.069384217262268,
      "learning_rate": 0.00021267107762721094,
      "loss": 2.0243,
      "step": 136864
    },
    {
      "epoch": 1.78,
      "grad_norm": 0.9969497323036194,
      "learning_rate": 0.00021266716422017214,
      "loss": 2.0275,
      "step": 136865
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.060245156288147,
      "learning_rate": 0.00021266325082937025,
      "loss": 2.0052,
      "step": 136866
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.003755807876587,
      "learning_rate": 0.00021265933745480582,
      "loss": 2.0014,
      "step": 136867
    },
    {
      "epoch": 1.78,
      "grad_norm": 0.9763481616973877,
      "learning_rate": 0.0002126554240964797,
      "loss": 2.279,
      "step": 136868
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.10389244556427,
      "learning_rate": 0.00021265151075439273,
      "loss": 2.1027,
      "step": 136869
    },
    {
      "epoch": 1.78,
      "grad_norm": 0.990818977355957,
      "learning_rate": 0.00021264759742854535,
      "loss": 2.1482,
      "step": 136870
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.0470865964889526,
      "learning_rate": 0.00021264368411893844,
      "loss": 2.0907,
      "step": 136871
    },
    {
      "epoch": 1.78,
      "grad_norm": 0.9902692437171936,
      "learning_rate": 0.0002126397708255727,
      "loss": 2.3009,
      "step": 136872
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.0781794786453247,
      "learning_rate": 0.0002126358575484489,
      "loss": 2.3151,
      "step": 136873
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.0548325777053833,
      "learning_rate": 0.00021263194428756773,
      "loss": 1.9594,
      "step": 136874
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.13057541847229,
      "learning_rate": 0.00021262803104293002,
      "loss": 2.1854,
      "step": 136875
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.1158207654953003,
      "learning_rate": 0.00021262411781453635,
      "loss": 2.3244,
      "step": 136876
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.0500216484069824,
      "learning_rate": 0.00021262020460238748,
      "loss": 2.2634,
      "step": 136877
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.062313437461853,
      "learning_rate": 0.00021261629140648412,
      "loss": 2.0568,
      "step": 136878
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.0365861654281616,
      "learning_rate": 0.00021261237822682713,
      "loss": 2.1897,
      "step": 136879
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.0084022283554077,
      "learning_rate": 0.00021260846506341707,
      "loss": 2.0315,
      "step": 136880
    },
    {
      "epoch": 1.78,
      "grad_norm": 0.9979787468910217,
      "learning_rate": 0.0002126045519162549,
      "loss": 2.1362,
      "step": 136881
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.1811200380325317,
      "learning_rate": 0.0002126006387853411,
      "loss": 2.1249,
      "step": 136882
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.0429842472076416,
      "learning_rate": 0.00021259672567067644,
      "loss": 2.0088,
      "step": 136883
    },
    {
      "epoch": 1.78,
      "grad_norm": 0.965944766998291,
      "learning_rate": 0.0002125928125722617,
      "loss": 2.1424,
      "step": 136884
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.0606459379196167,
      "learning_rate": 0.00021258889949009763,
      "loss": 2.0593,
      "step": 136885
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.0751947164535522,
      "learning_rate": 0.00021258498642418494,
      "loss": 2.0737,
      "step": 136886
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.0758428573608398,
      "learning_rate": 0.0002125810733745245,
      "loss": 2.0073,
      "step": 136887
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.023200511932373,
      "learning_rate": 0.00021257716034111672,
      "loss": 2.2572,
      "step": 136888
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.0865026712417603,
      "learning_rate": 0.00021257324732396247,
      "loss": 1.9526,
      "step": 136889
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.1027451753616333,
      "learning_rate": 0.00021256933432306255,
      "loss": 2.2454,
      "step": 136890
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.072226881980896,
      "learning_rate": 0.00021256542133841761,
      "loss": 2.2898,
      "step": 136891
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.1256641149520874,
      "learning_rate": 0.00021256150837002842,
      "loss": 1.8792,
      "step": 136892
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.797347903251648,
      "learning_rate": 0.00021255759541789578,
      "loss": 2.1858,
      "step": 136893
    },
    {
      "epoch": 1.78,
      "grad_norm": 0.9640750885009766,
      "learning_rate": 0.00021255368248202034,
      "loss": 1.8198,
      "step": 136894
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.186097264289856,
      "learning_rate": 0.0002125497695624027,
      "loss": 1.9713,
      "step": 136895
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.1284879446029663,
      "learning_rate": 0.00021254585665904374,
      "loss": 2.0484,
      "step": 136896
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.1079005002975464,
      "learning_rate": 0.00021254194377194418,
      "loss": 2.1589,
      "step": 136897
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.1401101350784302,
      "learning_rate": 0.00021253803090110472,
      "loss": 2.201,
      "step": 136898
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.17747962474823,
      "learning_rate": 0.00021253411804652615,
      "loss": 2.1857,
      "step": 136899
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.1176234483718872,
      "learning_rate": 0.00021253020520820904,
      "loss": 2.4534,
      "step": 136900
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.0409791469573975,
      "learning_rate": 0.00021252629238615426,
      "loss": 2.2094,
      "step": 136901
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.0530104637145996,
      "learning_rate": 0.0002125223795803625,
      "loss": 2.0124,
      "step": 136902
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.1550719738006592,
      "learning_rate": 0.00021251846679083447,
      "loss": 2.1712,
      "step": 136903
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.2217206954956055,
      "learning_rate": 0.0002125145540175709,
      "loss": 2.1826,
      "step": 136904
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.046379804611206,
      "learning_rate": 0.0002125106412605726,
      "loss": 1.8953,
      "step": 136905
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.103172779083252,
      "learning_rate": 0.00021250672851984013,
      "loss": 2.0756,
      "step": 136906
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.0199525356292725,
      "learning_rate": 0.00021250281579537433,
      "loss": 2.1564,
      "step": 136907
    },
    {
      "epoch": 1.78,
      "grad_norm": 0.9669023752212524,
      "learning_rate": 0.0002124989030871759,
      "loss": 2.1149,
      "step": 136908
    },
    {
      "epoch": 1.78,
      "grad_norm": 0.9724538326263428,
      "learning_rate": 0.00021249499039524563,
      "loss": 2.3137,
      "step": 136909
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.115101933479309,
      "learning_rate": 0.00021249107771958414,
      "loss": 2.0577,
      "step": 136910
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.0438097715377808,
      "learning_rate": 0.0002124871650601923,
      "loss": 2.0823,
      "step": 136911
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.1359814405441284,
      "learning_rate": 0.00021248325241707064,
      "loss": 1.9725,
      "step": 136912
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.0280847549438477,
      "learning_rate": 0.00021247933979022,
      "loss": 2.1294,
      "step": 136913
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.0614734888076782,
      "learning_rate": 0.00021247542717964114,
      "loss": 2.0476,
      "step": 136914
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.1008634567260742,
      "learning_rate": 0.0002124715145853347,
      "loss": 2.2165,
      "step": 136915
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.1192117929458618,
      "learning_rate": 0.00021246760200730155,
      "loss": 2.2495,
      "step": 136916
    },
    {
      "epoch": 1.78,
      "grad_norm": 0.94957435131073,
      "learning_rate": 0.00021246368944554235,
      "loss": 1.9795,
      "step": 136917
    },
    {
      "epoch": 1.78,
      "grad_norm": 0.9990159869194031,
      "learning_rate": 0.0002124597769000577,
      "loss": 2.3367,
      "step": 136918
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.0633782148361206,
      "learning_rate": 0.00021245586437084843,
      "loss": 2.2078,
      "step": 136919
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.002643346786499,
      "learning_rate": 0.0002124519518579153,
      "loss": 2.111,
      "step": 136920
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.1694179773330688,
      "learning_rate": 0.000212448039361259,
      "loss": 2.0889,
      "step": 136921
    },
    {
      "epoch": 1.78,
      "grad_norm": 0.991446316242218,
      "learning_rate": 0.00021244412688088023,
      "loss": 2.0294,
      "step": 136922
    },
    {
      "epoch": 1.78,
      "grad_norm": 0.9635736346244812,
      "learning_rate": 0.00021244021441677976,
      "loss": 2.1045,
      "step": 136923
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.003937005996704,
      "learning_rate": 0.00021243630196895845,
      "loss": 1.9632,
      "step": 136924
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.094258427619934,
      "learning_rate": 0.00021243238953741676,
      "loss": 2.1238,
      "step": 136925
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.1101711988449097,
      "learning_rate": 0.00021242847712215553,
      "loss": 2.1953,
      "step": 136926
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.0382168292999268,
      "learning_rate": 0.0002124245647231755,
      "loss": 2.2776,
      "step": 136927
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.0678749084472656,
      "learning_rate": 0.0002124206523404774,
      "loss": 1.8228,
      "step": 136928
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.265432596206665,
      "learning_rate": 0.00021241673997406195,
      "loss": 2.3155,
      "step": 136929
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.2859690189361572,
      "learning_rate": 0.00021241282762392996,
      "loss": 2.1324,
      "step": 136930
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.0404001474380493,
      "learning_rate": 0.000212408915290082,
      "loss": 2.1233,
      "step": 136931
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.0727375745773315,
      "learning_rate": 0.0002124050029725189,
      "loss": 1.8562,
      "step": 136932
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.3864202499389648,
      "learning_rate": 0.00021240109067124135,
      "loss": 2.3824,
      "step": 136933
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.0817492008209229,
      "learning_rate": 0.00021239717838625005,
      "loss": 2.0797,
      "step": 136934
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.1266953945159912,
      "learning_rate": 0.0002123932661175458,
      "loss": 2.0882,
      "step": 136935
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.094751238822937,
      "learning_rate": 0.00021238935386512942,
      "loss": 2.034,
      "step": 136936
    },
    {
      "epoch": 1.78,
      "grad_norm": 0.9848545789718628,
      "learning_rate": 0.0002123854416290014,
      "loss": 2.0832,
      "step": 136937
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.058387279510498,
      "learning_rate": 0.00021238152940916254,
      "loss": 2.2264,
      "step": 136938
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.0580805540084839,
      "learning_rate": 0.00021237761720561363,
      "loss": 2.2899,
      "step": 136939
    },
    {
      "epoch": 1.78,
      "grad_norm": 0.9675925970077515,
      "learning_rate": 0.00021237370501835543,
      "loss": 2.3538,
      "step": 136940
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.3160427808761597,
      "learning_rate": 0.00021236979284738853,
      "loss": 2.1021,
      "step": 136941
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.0573400259017944,
      "learning_rate": 0.0002123658806927139,
      "loss": 2.3851,
      "step": 136942
    },
    {
      "epoch": 1.78,
      "grad_norm": 0.9980397820472717,
      "learning_rate": 0.00021236196855433197,
      "loss": 2.1288,
      "step": 136943
    },
    {
      "epoch": 1.78,
      "grad_norm": 0.9994659423828125,
      "learning_rate": 0.00021235805643224365,
      "loss": 2.1681,
      "step": 136944
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.050731897354126,
      "learning_rate": 0.00021235414432644955,
      "loss": 2.0305,
      "step": 136945
    },
    {
      "epoch": 1.78,
      "grad_norm": 0.995556116104126,
      "learning_rate": 0.00021235023223695052,
      "loss": 2.2272,
      "step": 136946
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.0593420267105103,
      "learning_rate": 0.0002123463201637472,
      "loss": 2.253,
      "step": 136947
    },
    {
      "epoch": 1.78,
      "grad_norm": 0.9925198554992676,
      "learning_rate": 0.00021234240810684052,
      "loss": 1.9549,
      "step": 136948
    },
    {
      "epoch": 1.78,
      "grad_norm": 0.9860315918922424,
      "learning_rate": 0.00021233849606623088,
      "loss": 2.1467,
      "step": 136949
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.1266835927963257,
      "learning_rate": 0.0002123345840419192,
      "loss": 2.1364,
      "step": 136950
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.0732027292251587,
      "learning_rate": 0.00021233067203390617,
      "loss": 1.9966,
      "step": 136951
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.0454455614089966,
      "learning_rate": 0.00021232676004219247,
      "loss": 2.208,
      "step": 136952
    },
    {
      "epoch": 1.78,
      "grad_norm": 0.9539656639099121,
      "learning_rate": 0.00021232284806677892,
      "loss": 1.8467,
      "step": 136953
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.1325854063034058,
      "learning_rate": 0.00021231893610766634,
      "loss": 2.0381,
      "step": 136954
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.0868805646896362,
      "learning_rate": 0.0002123150241648552,
      "loss": 1.9585,
      "step": 136955
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.0011942386627197,
      "learning_rate": 0.00021231111223834632,
      "loss": 2.1352,
      "step": 136956
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.0513925552368164,
      "learning_rate": 0.0002123072003281405,
      "loss": 2.2164,
      "step": 136957
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.0820460319519043,
      "learning_rate": 0.00021230328843423844,
      "loss": 1.9493,
      "step": 136958
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.0697284936904907,
      "learning_rate": 0.00021229937655664082,
      "loss": 2.0038,
      "step": 136959
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.096897840499878,
      "learning_rate": 0.00021229546469534854,
      "loss": 2.1113,
      "step": 136960
    },
    {
      "epoch": 1.78,
      "grad_norm": 0.9903203845024109,
      "learning_rate": 0.00021229155285036204,
      "loss": 1.8944,
      "step": 136961
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.0346344709396362,
      "learning_rate": 0.00021228764102168224,
      "loss": 2.4282,
      "step": 136962
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.0414576530456543,
      "learning_rate": 0.00021228372920930978,
      "loss": 1.9921,
      "step": 136963
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.0573617219924927,
      "learning_rate": 0.00021227981741324544,
      "loss": 1.9894,
      "step": 136964
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.0693764686584473,
      "learning_rate": 0.00021227590563348996,
      "loss": 2.1829,
      "step": 136965
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.1579445600509644,
      "learning_rate": 0.0002122719938700441,
      "loss": 1.8842,
      "step": 136966
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.048478364944458,
      "learning_rate": 0.0002122680821229085,
      "loss": 1.9335,
      "step": 136967
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.0932416915893555,
      "learning_rate": 0.0002122641703920839,
      "loss": 2.1442,
      "step": 136968
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.2461789846420288,
      "learning_rate": 0.000212260258677571,
      "loss": 2.119,
      "step": 136969
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.2144094705581665,
      "learning_rate": 0.0002122563469793706,
      "loss": 2.1227,
      "step": 136970
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.0760177373886108,
      "learning_rate": 0.0002122524352974834,
      "loss": 1.8748,
      "step": 136971
    },
    {
      "epoch": 1.78,
      "grad_norm": 0.9883652329444885,
      "learning_rate": 0.00021224852363191027,
      "loss": 2.0245,
      "step": 136972
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.1957718133926392,
      "learning_rate": 0.00021224461198265164,
      "loss": 2.1788,
      "step": 136973
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.2043575048446655,
      "learning_rate": 0.00021224070034970842,
      "loss": 2.133,
      "step": 136974
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.2302733659744263,
      "learning_rate": 0.00021223678873308132,
      "loss": 1.9215,
      "step": 136975
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.3012568950653076,
      "learning_rate": 0.000212232877132771,
      "loss": 1.9591,
      "step": 136976
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.3319958448410034,
      "learning_rate": 0.00021222896554877832,
      "loss": 2.1372,
      "step": 136977
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.2955708503723145,
      "learning_rate": 0.00021222505398110395,
      "loss": 2.161,
      "step": 136978
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.2081420421600342,
      "learning_rate": 0.0002122211424297486,
      "loss": 1.9172,
      "step": 136979
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.1952896118164062,
      "learning_rate": 0.00021221723089471294,
      "loss": 2.2315,
      "step": 136980
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.031203031539917,
      "learning_rate": 0.00021221331937599776,
      "loss": 2.0859,
      "step": 136981
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.142240285873413,
      "learning_rate": 0.00021220940787360378,
      "loss": 2.1842,
      "step": 136982
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.4884666204452515,
      "learning_rate": 0.0002122054963875317,
      "loss": 2.1455,
      "step": 136983
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.5943340063095093,
      "learning_rate": 0.00021220158491778242,
      "loss": 2.1406,
      "step": 136984
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.0356370210647583,
      "learning_rate": 0.0002121976734643564,
      "loss": 1.9191,
      "step": 136985
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.2338298559188843,
      "learning_rate": 0.0002121937620272545,
      "loss": 2.1786,
      "step": 136986
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.436705470085144,
      "learning_rate": 0.00021218985060647745,
      "loss": 2.3402,
      "step": 136987
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.2049047946929932,
      "learning_rate": 0.00021218593920202595,
      "loss": 1.8884,
      "step": 136988
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.0353702306747437,
      "learning_rate": 0.00021218202781390073,
      "loss": 2.3483,
      "step": 136989
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.0513991117477417,
      "learning_rate": 0.00021217811644210253,
      "loss": 2.1651,
      "step": 136990
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.3206439018249512,
      "learning_rate": 0.00021217420508663217,
      "loss": 2.3574,
      "step": 136991
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.1827657222747803,
      "learning_rate": 0.0002121702937474902,
      "loss": 2.2354,
      "step": 136992
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.006194829940796,
      "learning_rate": 0.00021216638242467742,
      "loss": 2.181,
      "step": 136993
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.0354945659637451,
      "learning_rate": 0.0002121624711181946,
      "loss": 2.1622,
      "step": 136994
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.0237948894500732,
      "learning_rate": 0.00021215855982804243,
      "loss": 1.8623,
      "step": 136995
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.1037647724151611,
      "learning_rate": 0.00021215464855422163,
      "loss": 2.1161,
      "step": 136996
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.1450331211090088,
      "learning_rate": 0.00021215073729673298,
      "loss": 2.2914,
      "step": 136997
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.0470142364501953,
      "learning_rate": 0.0002121468260555771,
      "loss": 2.1486,
      "step": 136998
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.075821876525879,
      "learning_rate": 0.00021214291483075479,
      "loss": 2.013,
      "step": 136999
    },
    {
      "epoch": 1.78,
      "grad_norm": 0.9980539083480835,
      "learning_rate": 0.00021213900362226682,
      "loss": 2.0485,
      "step": 137000
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.1160715818405151,
      "learning_rate": 0.00021213509243011387,
      "loss": 2.0627,
      "step": 137001
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.1835891008377075,
      "learning_rate": 0.0002121311812542966,
      "loss": 2.1339,
      "step": 137002
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.6135386228561401,
      "learning_rate": 0.0002121272700948159,
      "loss": 2.085,
      "step": 137003
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.2408723831176758,
      "learning_rate": 0.0002121233589516723,
      "loss": 2.3295,
      "step": 137004
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.0720207691192627,
      "learning_rate": 0.00021211944782486666,
      "loss": 2.3481,
      "step": 137005
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.0460033416748047,
      "learning_rate": 0.00021211553671439965,
      "loss": 2.2584,
      "step": 137006
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.1814405918121338,
      "learning_rate": 0.00021211162562027208,
      "loss": 2.2513,
      "step": 137007
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.5488264560699463,
      "learning_rate": 0.0002121077145424846,
      "loss": 2.1277,
      "step": 137008
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.0091429948806763,
      "learning_rate": 0.000212103803481038,
      "loss": 2.1776,
      "step": 137009
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.079168438911438,
      "learning_rate": 0.0002120998924359329,
      "loss": 2.134,
      "step": 137010
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.2576920986175537,
      "learning_rate": 0.00021209598140717003,
      "loss": 1.8856,
      "step": 137011
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.0079692602157593,
      "learning_rate": 0.00021209207039475024,
      "loss": 1.9843,
      "step": 137012
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.116741418838501,
      "learning_rate": 0.0002120881593986742,
      "loss": 2.0483,
      "step": 137013
    },
    {
      "epoch": 1.78,
      "grad_norm": 0.9990000128746033,
      "learning_rate": 0.0002120842484189426,
      "loss": 2.1097,
      "step": 137014
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.3497847318649292,
      "learning_rate": 0.0002120803374555563,
      "loss": 2.365,
      "step": 137015
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.0067387819290161,
      "learning_rate": 0.00021207642650851582,
      "loss": 2.0117,
      "step": 137016
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.0383203029632568,
      "learning_rate": 0.00021207251557782202,
      "loss": 2.073,
      "step": 137017
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.0203111171722412,
      "learning_rate": 0.00021206860466347558,
      "loss": 2.2073,
      "step": 137018
    },
    {
      "epoch": 1.78,
      "grad_norm": 0.9705920815467834,
      "learning_rate": 0.00021206469376547723,
      "loss": 1.9538,
      "step": 137019
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.1852811574935913,
      "learning_rate": 0.00021206078288382775,
      "loss": 2.0738,
      "step": 137020
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.1427319049835205,
      "learning_rate": 0.00021205687201852793,
      "loss": 2.01,
      "step": 137021
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.0964360237121582,
      "learning_rate": 0.0002120529611695783,
      "loss": 2.1242,
      "step": 137022
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.2260868549346924,
      "learning_rate": 0.00021204905033697966,
      "loss": 1.9163,
      "step": 137023
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.0892642736434937,
      "learning_rate": 0.00021204513952073276,
      "loss": 2.1507,
      "step": 137024
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.0564589500427246,
      "learning_rate": 0.00021204122872083833,
      "loss": 2.1353,
      "step": 137025
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.073025107383728,
      "learning_rate": 0.00021203731793729708,
      "loss": 2.0203,
      "step": 137026
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.4117307662963867,
      "learning_rate": 0.0002120334071701099,
      "loss": 2.09,
      "step": 137027
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.0333173274993896,
      "learning_rate": 0.0002120294964192772,
      "loss": 2.1111,
      "step": 137028
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.1849931478500366,
      "learning_rate": 0.00021202558568479997,
      "loss": 2.2945,
      "step": 137029
    },
    {
      "epoch": 1.78,
      "grad_norm": 0.9598106145858765,
      "learning_rate": 0.00021202167496667878,
      "loss": 2.2194,
      "step": 137030
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.1072008609771729,
      "learning_rate": 0.0002120177642649144,
      "loss": 2.3513,
      "step": 137031
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.1174226999282837,
      "learning_rate": 0.00021201385357950763,
      "loss": 2.0824,
      "step": 137032
    },
    {
      "epoch": 1.78,
      "grad_norm": 0.9955736994743347,
      "learning_rate": 0.00021200994291045925,
      "loss": 2.1485,
      "step": 137033
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.0665533542633057,
      "learning_rate": 0.00021200603225776977,
      "loss": 2.2057,
      "step": 137034
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.104961633682251,
      "learning_rate": 0.00021200212162144,
      "loss": 1.9627,
      "step": 137035
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.205830693244934,
      "learning_rate": 0.00021199821100147077,
      "loss": 2.1592,
      "step": 137036
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.0239871740341187,
      "learning_rate": 0.00021199430039786266,
      "loss": 1.9663,
      "step": 137037
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.1505931615829468,
      "learning_rate": 0.0002119903898106165,
      "loss": 2.2076,
      "step": 137038
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.2725131511688232,
      "learning_rate": 0.00021198647923973308,
      "loss": 2.1127,
      "step": 137039
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.2431682348251343,
      "learning_rate": 0.00021198256868521295,
      "loss": 2.1531,
      "step": 137040
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.0210309028625488,
      "learning_rate": 0.00021197865814705694,
      "loss": 2.2498,
      "step": 137041
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.128879427909851,
      "learning_rate": 0.0002119747476252657,
      "loss": 2.0571,
      "step": 137042
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.0228376388549805,
      "learning_rate": 0.00021197083711984007,
      "loss": 1.9543,
      "step": 137043
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.1459548473358154,
      "learning_rate": 0.00021196692663078066,
      "loss": 2.1016,
      "step": 137044
    },
    {
      "epoch": 1.78,
      "grad_norm": 0.9566490650177002,
      "learning_rate": 0.00021196301615808845,
      "loss": 2.1251,
      "step": 137045
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.051648736000061,
      "learning_rate": 0.0002119591057017638,
      "loss": 1.8826,
      "step": 137046
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.0325993299484253,
      "learning_rate": 0.00021195519526180764,
      "loss": 2.0144,
      "step": 137047
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.0626779794692993,
      "learning_rate": 0.00021195128483822068,
      "loss": 2.1628,
      "step": 137048
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.0584702491760254,
      "learning_rate": 0.0002119473744310036,
      "loss": 2.0912,
      "step": 137049
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.099029541015625,
      "learning_rate": 0.0002119434640401572,
      "loss": 2.0063,
      "step": 137050
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.028796911239624,
      "learning_rate": 0.00021193955366568228,
      "loss": 2.0777,
      "step": 137051
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.3674654960632324,
      "learning_rate": 0.00021193564330757934,
      "loss": 2.3188,
      "step": 137052
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.3702179193496704,
      "learning_rate": 0.00021193173296584923,
      "loss": 2.1487,
      "step": 137053
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.0774656534194946,
      "learning_rate": 0.00021192782264049266,
      "loss": 1.9271,
      "step": 137054
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.0492750406265259,
      "learning_rate": 0.00021192391233151037,
      "loss": 2.3198,
      "step": 137055
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.3973710536956787,
      "learning_rate": 0.00021192000203890313,
      "loss": 2.3089,
      "step": 137056
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.0716302394866943,
      "learning_rate": 0.00021191609176267158,
      "loss": 1.9956,
      "step": 137057
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.4412362575531006,
      "learning_rate": 0.00021191218150281666,
      "loss": 1.9931,
      "step": 137058
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.1080914735794067,
      "learning_rate": 0.00021190827125933874,
      "loss": 2.1553,
      "step": 137059
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.010848879814148,
      "learning_rate": 0.00021190436103223878,
      "loss": 1.8447,
      "step": 137060
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.3056312799453735,
      "learning_rate": 0.0002119004508215175,
      "loss": 2.0288,
      "step": 137061
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.45401132106781,
      "learning_rate": 0.0002118965406271755,
      "loss": 2.0865,
      "step": 137062
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.096272349357605,
      "learning_rate": 0.0002118926304492136,
      "loss": 2.111,
      "step": 137063
    },
    {
      "epoch": 1.78,
      "grad_norm": 0.9584234952926636,
      "learning_rate": 0.0002118887202876327,
      "loss": 2.0358,
      "step": 137064
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.1850550174713135,
      "learning_rate": 0.00021188481014243322,
      "loss": 2.0365,
      "step": 137065
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.1612108945846558,
      "learning_rate": 0.000211880900013616,
      "loss": 2.2572,
      "step": 137066
    },
    {
      "epoch": 1.78,
      "grad_norm": 0.9606446623802185,
      "learning_rate": 0.0002118769899011818,
      "loss": 1.9781,
      "step": 137067
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.0444873571395874,
      "learning_rate": 0.00021187307980513132,
      "loss": 2.0122,
      "step": 137068
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.2738256454467773,
      "learning_rate": 0.0002118691697254653,
      "loss": 1.9926,
      "step": 137069
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.2482980489730835,
      "learning_rate": 0.00021186525966218454,
      "loss": 1.8978,
      "step": 137070
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.084834337234497,
      "learning_rate": 0.0002118613496152897,
      "loss": 2.0356,
      "step": 137071
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.0355644226074219,
      "learning_rate": 0.0002118574395847814,
      "loss": 2.1744,
      "step": 137072
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.1043697595596313,
      "learning_rate": 0.00021185352957066043,
      "loss": 2.0787,
      "step": 137073
    },
    {
      "epoch": 1.78,
      "grad_norm": 0.9918579459190369,
      "learning_rate": 0.00021184961957292764,
      "loss": 1.9487,
      "step": 137074
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.0687094926834106,
      "learning_rate": 0.00021184570959158363,
      "loss": 1.9941,
      "step": 137075
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.0455247163772583,
      "learning_rate": 0.00021184179962662923,
      "loss": 2.0804,
      "step": 137076
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.2418887615203857,
      "learning_rate": 0.00021183788967806504,
      "loss": 2.2797,
      "step": 137077
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.1494070291519165,
      "learning_rate": 0.00021183397974589192,
      "loss": 2.1185,
      "step": 137078
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.0868974924087524,
      "learning_rate": 0.00021183006983011045,
      "loss": 2.0471,
      "step": 137079
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.0114563703536987,
      "learning_rate": 0.00021182615993072147,
      "loss": 2.2511,
      "step": 137080
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.0645005702972412,
      "learning_rate": 0.00021182225004772563,
      "loss": 2.1004,
      "step": 137081
    },
    {
      "epoch": 1.78,
      "grad_norm": 0.9179568886756897,
      "learning_rate": 0.00021181834018112376,
      "loss": 1.9741,
      "step": 137082
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.043431043624878,
      "learning_rate": 0.0002118144303309165,
      "loss": 2.2414,
      "step": 137083
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.2169067859649658,
      "learning_rate": 0.00021181052049710458,
      "loss": 2.2616,
      "step": 137084
    },
    {
      "epoch": 1.78,
      "grad_norm": 0.9950878620147705,
      "learning_rate": 0.00021180661067968877,
      "loss": 2.1158,
      "step": 137085
    },
    {
      "epoch": 1.78,
      "grad_norm": 0.9969953894615173,
      "learning_rate": 0.00021180270087866983,
      "loss": 2.2254,
      "step": 137086
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.1132211685180664,
      "learning_rate": 0.00021179879109404834,
      "loss": 2.0533,
      "step": 137087
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.2124665975570679,
      "learning_rate": 0.00021179488132582518,
      "loss": 1.9526,
      "step": 137088
    },
    {
      "epoch": 1.78,
      "grad_norm": 1.0862922668457031,
      "learning_rate": 0.000211790971574001,
      "loss": 2.3045,
      "step": 137089
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.2761242389678955,
      "learning_rate": 0.0002117870618385765,
      "loss": 2.2825,
      "step": 137090
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.3974910974502563,
      "learning_rate": 0.00021178315211955247,
      "loss": 2.0961,
      "step": 137091
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.2860666513442993,
      "learning_rate": 0.00021177924241692962,
      "loss": 2.201,
      "step": 137092
    },
    {
      "epoch": 1.79,
      "grad_norm": 0.976851224899292,
      "learning_rate": 0.00021177533273070872,
      "loss": 2.0915,
      "step": 137093
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.0468621253967285,
      "learning_rate": 0.0002117714230608905,
      "loss": 1.9975,
      "step": 137094
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.057848334312439,
      "learning_rate": 0.00021176751340747552,
      "loss": 2.3232,
      "step": 137095
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.0222691297531128,
      "learning_rate": 0.00021176360377046465,
      "loss": 2.1238,
      "step": 137096
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.073167324066162,
      "learning_rate": 0.00021175969414985858,
      "loss": 2.0698,
      "step": 137097
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.0454951524734497,
      "learning_rate": 0.00021175578454565803,
      "loss": 2.232,
      "step": 137098
    },
    {
      "epoch": 1.79,
      "grad_norm": 0.9282516241073608,
      "learning_rate": 0.0002117518749578638,
      "loss": 2.1932,
      "step": 137099
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.2298592329025269,
      "learning_rate": 0.00021174796538647665,
      "loss": 1.929,
      "step": 137100
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.1691254377365112,
      "learning_rate": 0.0002117440558314971,
      "loss": 1.7818,
      "step": 137101
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.0739901065826416,
      "learning_rate": 0.00021174014629292602,
      "loss": 2.0823,
      "step": 137102
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.0055898427963257,
      "learning_rate": 0.00021173623677076407,
      "loss": 2.0719,
      "step": 137103
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.3258264064788818,
      "learning_rate": 0.00021173232726501206,
      "loss": 2.0746,
      "step": 137104
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.1054859161376953,
      "learning_rate": 0.00021172841777567061,
      "loss": 2.3261,
      "step": 137105
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.075460433959961,
      "learning_rate": 0.0002117245083027407,
      "loss": 1.8766,
      "step": 137106
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.0013983249664307,
      "learning_rate": 0.00021172059884622276,
      "loss": 2.0723,
      "step": 137107
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.0946581363677979,
      "learning_rate": 0.00021171668940611762,
      "loss": 1.8797,
      "step": 137108
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.2453304529190063,
      "learning_rate": 0.00021171277998242597,
      "loss": 2.22,
      "step": 137109
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.2527495622634888,
      "learning_rate": 0.0002117088705751486,
      "loss": 2.0197,
      "step": 137110
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.0204951763153076,
      "learning_rate": 0.00021170496118428625,
      "loss": 2.1586,
      "step": 137111
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.110567331314087,
      "learning_rate": 0.00021170105180983966,
      "loss": 1.8379,
      "step": 137112
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.1121042966842651,
      "learning_rate": 0.00021169714245180944,
      "loss": 1.9799,
      "step": 137113
    },
    {
      "epoch": 1.79,
      "grad_norm": 0.9811209440231323,
      "learning_rate": 0.00021169323311019642,
      "loss": 2.2309,
      "step": 137114
    },
    {
      "epoch": 1.79,
      "grad_norm": 0.9337146282196045,
      "learning_rate": 0.00021168932378500126,
      "loss": 2.0602,
      "step": 137115
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.0418802499771118,
      "learning_rate": 0.0002116854144762247,
      "loss": 1.9971,
      "step": 137116
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.0932748317718506,
      "learning_rate": 0.0002116815051838675,
      "loss": 2.1638,
      "step": 137117
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.0666735172271729,
      "learning_rate": 0.0002116775959079305,
      "loss": 2.2018,
      "step": 137118
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.1228457689285278,
      "learning_rate": 0.00021167368664841418,
      "loss": 2.0857,
      "step": 137119
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.0608609914779663,
      "learning_rate": 0.00021166977740531942,
      "loss": 1.9414,
      "step": 137120
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.1812323331832886,
      "learning_rate": 0.00021166586817864685,
      "loss": 2.0897,
      "step": 137121
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.0987142324447632,
      "learning_rate": 0.00021166195896839734,
      "loss": 2.1084,
      "step": 137122
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.0064785480499268,
      "learning_rate": 0.0002116580497745715,
      "loss": 2.065,
      "step": 137123
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.3505582809448242,
      "learning_rate": 0.0002116541405971701,
      "loss": 2.3121,
      "step": 137124
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.1284836530685425,
      "learning_rate": 0.000211650231436194,
      "loss": 2.007,
      "step": 137125
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.1461758613586426,
      "learning_rate": 0.00021164632229164364,
      "loss": 2.262,
      "step": 137126
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.0287991762161255,
      "learning_rate": 0.0002116424131635199,
      "loss": 2.1165,
      "step": 137127
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.0612289905548096,
      "learning_rate": 0.00021163850405182352,
      "loss": 2.1195,
      "step": 137128
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.5863052606582642,
      "learning_rate": 0.00021163459495655517,
      "loss": 2.145,
      "step": 137129
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.2661148309707642,
      "learning_rate": 0.00021163068587771568,
      "loss": 1.9204,
      "step": 137130
    },
    {
      "epoch": 1.79,
      "grad_norm": 0.995680570602417,
      "learning_rate": 0.00021162677681530582,
      "loss": 2.1345,
      "step": 137131
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.0285874605178833,
      "learning_rate": 0.00021162286776932606,
      "loss": 2.3331,
      "step": 137132
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.8593626022338867,
      "learning_rate": 0.00021161895873977727,
      "loss": 2.0308,
      "step": 137133
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.2168112993240356,
      "learning_rate": 0.00021161504972666022,
      "loss": 2.2112,
      "step": 137134
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.0295705795288086,
      "learning_rate": 0.00021161114072997558,
      "loss": 2.118,
      "step": 137135
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.0310736894607544,
      "learning_rate": 0.00021160723174972412,
      "loss": 2.1174,
      "step": 137136
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.3111129999160767,
      "learning_rate": 0.0002116033227859066,
      "loss": 2.3162,
      "step": 137137
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.4902057647705078,
      "learning_rate": 0.00021159941383852362,
      "loss": 2.0538,
      "step": 137138
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.2822331190109253,
      "learning_rate": 0.000211595504907576,
      "loss": 2.1465,
      "step": 137139
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.0037568807601929,
      "learning_rate": 0.00021159159599306443,
      "loss": 2.1711,
      "step": 137140
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.1471010446548462,
      "learning_rate": 0.0002115876870949896,
      "loss": 1.9412,
      "step": 137141
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.112140417098999,
      "learning_rate": 0.00021158377821335237,
      "loss": 2.0547,
      "step": 137142
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.4383718967437744,
      "learning_rate": 0.00021157986934815346,
      "loss": 2.001,
      "step": 137143
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.2512540817260742,
      "learning_rate": 0.00021157596049939338,
      "loss": 1.8794,
      "step": 137144
    },
    {
      "epoch": 1.79,
      "grad_norm": 0.9741683006286621,
      "learning_rate": 0.00021157205166707303,
      "loss": 2.0874,
      "step": 137145
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.4288091659545898,
      "learning_rate": 0.00021156814285119312,
      "loss": 1.8737,
      "step": 137146
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.47492253780365,
      "learning_rate": 0.00021156423405175435,
      "loss": 2.0292,
      "step": 137147
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.0797910690307617,
      "learning_rate": 0.0002115603252687574,
      "loss": 1.9019,
      "step": 137148
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.0131982564926147,
      "learning_rate": 0.00021155641650220323,
      "loss": 2.1065,
      "step": 137149
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.1338999271392822,
      "learning_rate": 0.0002115525077520923,
      "loss": 2.1632,
      "step": 137150
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.2218446731567383,
      "learning_rate": 0.00021154859901842537,
      "loss": 1.894,
      "step": 137151
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.1698907613754272,
      "learning_rate": 0.00021154469030120322,
      "loss": 2.0763,
      "step": 137152
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.177492618560791,
      "learning_rate": 0.00021154078160042666,
      "loss": 2.0931,
      "step": 137153
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.1087969541549683,
      "learning_rate": 0.00021153687291609625,
      "loss": 2.0249,
      "step": 137154
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.0805882215499878,
      "learning_rate": 0.00021153296424821289,
      "loss": 2.1918,
      "step": 137155
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.4364819526672363,
      "learning_rate": 0.00021152905559677725,
      "loss": 2.1178,
      "step": 137156
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.1943612098693848,
      "learning_rate": 0.00021152514696178996,
      "loss": 2.0094,
      "step": 137157
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.0152968168258667,
      "learning_rate": 0.00021152123834325182,
      "loss": 2.1946,
      "step": 137158
    },
    {
      "epoch": 1.79,
      "grad_norm": 0.9916442036628723,
      "learning_rate": 0.00021151732974116353,
      "loss": 2.1473,
      "step": 137159
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.1227097511291504,
      "learning_rate": 0.00021151342115552585,
      "loss": 2.1662,
      "step": 137160
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.0860967636108398,
      "learning_rate": 0.00021150951258633958,
      "loss": 2.0713,
      "step": 137161
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.1242295503616333,
      "learning_rate": 0.00021150560403360524,
      "loss": 2.1087,
      "step": 137162
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.0344597101211548,
      "learning_rate": 0.00021150169549732377,
      "loss": 2.1673,
      "step": 137163
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.1033967733383179,
      "learning_rate": 0.00021149778697749576,
      "loss": 2.0956,
      "step": 137164
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.1442404985427856,
      "learning_rate": 0.00021149387847412197,
      "loss": 2.2721,
      "step": 137165
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.1613726615905762,
      "learning_rate": 0.00021148996998720317,
      "loss": 2.3224,
      "step": 137166
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.0529637336730957,
      "learning_rate": 0.00021148606151674007,
      "loss": 2.2201,
      "step": 137167
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.2664289474487305,
      "learning_rate": 0.0002114821530627333,
      "loss": 2.1258,
      "step": 137168
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.1579515933990479,
      "learning_rate": 0.00021147824462518374,
      "loss": 2.2766,
      "step": 137169
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.0699642896652222,
      "learning_rate": 0.00021147433620409206,
      "loss": 2.1581,
      "step": 137170
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.2440078258514404,
      "learning_rate": 0.00021147042779945892,
      "loss": 2.1262,
      "step": 137171
    },
    {
      "epoch": 1.79,
      "grad_norm": 0.9894998073577881,
      "learning_rate": 0.00021146651941128508,
      "loss": 1.9486,
      "step": 137172
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.0933377742767334,
      "learning_rate": 0.0002114626110395714,
      "loss": 2.4065,
      "step": 137173
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.131372094154358,
      "learning_rate": 0.0002114587026843184,
      "loss": 2.2494,
      "step": 137174
    },
    {
      "epoch": 1.79,
      "grad_norm": 0.9799553751945496,
      "learning_rate": 0.00021145479434552688,
      "loss": 2.246,
      "step": 137175
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.0034819841384888,
      "learning_rate": 0.00021145088602319762,
      "loss": 2.0075,
      "step": 137176
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.0468835830688477,
      "learning_rate": 0.00021144697771733134,
      "loss": 2.0702,
      "step": 137177
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.0752004384994507,
      "learning_rate": 0.0002114430694279287,
      "loss": 2.0086,
      "step": 137178
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.053760290145874,
      "learning_rate": 0.0002114391611549905,
      "loss": 2.1832,
      "step": 137179
    },
    {
      "epoch": 1.79,
      "grad_norm": 0.9394420981407166,
      "learning_rate": 0.00021143525289851745,
      "loss": 1.9656,
      "step": 137180
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.0115668773651123,
      "learning_rate": 0.00021143134465851016,
      "loss": 2.2251,
      "step": 137181
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.0348972082138062,
      "learning_rate": 0.0002114274364349695,
      "loss": 1.994,
      "step": 137182
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.012349247932434,
      "learning_rate": 0.00021142352822789613,
      "loss": 1.969,
      "step": 137183
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.1335583925247192,
      "learning_rate": 0.00021141962003729081,
      "loss": 2.2531,
      "step": 137184
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.3981022834777832,
      "learning_rate": 0.0002114157118631544,
      "loss": 2.0369,
      "step": 137185
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.0543149709701538,
      "learning_rate": 0.0002114118037054873,
      "loss": 1.9574,
      "step": 137186
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.0871381759643555,
      "learning_rate": 0.0002114078955642905,
      "loss": 1.9828,
      "step": 137187
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.4197450876235962,
      "learning_rate": 0.00021140398743956459,
      "loss": 2.1731,
      "step": 137188
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.2462033033370972,
      "learning_rate": 0.00021140007933131034,
      "loss": 2.1999,
      "step": 137189
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.161547064781189,
      "learning_rate": 0.00021139617123952853,
      "loss": 2.0755,
      "step": 137190
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.1195119619369507,
      "learning_rate": 0.00021139226316421985,
      "loss": 2.0758,
      "step": 137191
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.2773418426513672,
      "learning_rate": 0.0002113883551053851,
      "loss": 2.1033,
      "step": 137192
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.2764101028442383,
      "learning_rate": 0.00021138444706302482,
      "loss": 2.2273,
      "step": 137193
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.0145390033721924,
      "learning_rate": 0.00021138053903713985,
      "loss": 2.0017,
      "step": 137194
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.121201992034912,
      "learning_rate": 0.0002113766310277309,
      "loss": 1.885,
      "step": 137195
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.0525065660476685,
      "learning_rate": 0.00021137272303479874,
      "loss": 1.9892,
      "step": 137196
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.1131373643875122,
      "learning_rate": 0.00021136881505834403,
      "loss": 2.1654,
      "step": 137197
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.0671985149383545,
      "learning_rate": 0.00021136490709836766,
      "loss": 2.0455,
      "step": 137198
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.0624446868896484,
      "learning_rate": 0.0002113609991548701,
      "loss": 2.1183,
      "step": 137199
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.04523766040802,
      "learning_rate": 0.0002113570912278522,
      "loss": 1.9052,
      "step": 137200
    },
    {
      "epoch": 1.79,
      "grad_norm": 0.9694861769676208,
      "learning_rate": 0.0002113531833173147,
      "loss": 2.1515,
      "step": 137201
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.1409987211227417,
      "learning_rate": 0.0002113492754232583,
      "loss": 1.9244,
      "step": 137202
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.2352179288864136,
      "learning_rate": 0.00021134536754568374,
      "loss": 1.7952,
      "step": 137203
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.3256195783615112,
      "learning_rate": 0.0002113414596845919,
      "loss": 2.1651,
      "step": 137204
    },
    {
      "epoch": 1.79,
      "grad_norm": 0.992872953414917,
      "learning_rate": 0.00021133755183998324,
      "loss": 2.1612,
      "step": 137205
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.1573280096054077,
      "learning_rate": 0.0002113336440118586,
      "loss": 1.9637,
      "step": 137206
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.0081634521484375,
      "learning_rate": 0.00021132973620021866,
      "loss": 1.897,
      "step": 137207
    },
    {
      "epoch": 1.79,
      "grad_norm": 0.9934765100479126,
      "learning_rate": 0.00021132582840506427,
      "loss": 2.1042,
      "step": 137208
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.0262641906738281,
      "learning_rate": 0.00021132192062639602,
      "loss": 1.9829,
      "step": 137209
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.0062742233276367,
      "learning_rate": 0.00021131801286421485,
      "loss": 1.9862,
      "step": 137210
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.008763074874878,
      "learning_rate": 0.00021131410511852122,
      "loss": 1.7157,
      "step": 137211
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.0971474647521973,
      "learning_rate": 0.00021131019738931595,
      "loss": 2.0069,
      "step": 137212
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.0050954818725586,
      "learning_rate": 0.00021130628967659982,
      "loss": 2.357,
      "step": 137213
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.066314458847046,
      "learning_rate": 0.0002113023819803735,
      "loss": 2.1044,
      "step": 137214
    },
    {
      "epoch": 1.79,
      "grad_norm": 0.9906893372535706,
      "learning_rate": 0.0002112984743006377,
      "loss": 2.1447,
      "step": 137215
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.1239027976989746,
      "learning_rate": 0.0002112945666373934,
      "loss": 2.023,
      "step": 137216
    },
    {
      "epoch": 1.79,
      "grad_norm": 0.9583068490028381,
      "learning_rate": 0.00021129065899064094,
      "loss": 2.1162,
      "step": 137217
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.3296574354171753,
      "learning_rate": 0.00021128675136038126,
      "loss": 2.1759,
      "step": 137218
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.0153555870056152,
      "learning_rate": 0.00021128284374661498,
      "loss": 2.0243,
      "step": 137219
    },
    {
      "epoch": 1.79,
      "grad_norm": 0.9990294575691223,
      "learning_rate": 0.00021127893614934293,
      "loss": 2.107,
      "step": 137220
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.0509122610092163,
      "learning_rate": 0.0002112750285685658,
      "loss": 2.0869,
      "step": 137221
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.0154597759246826,
      "learning_rate": 0.00021127112100428444,
      "loss": 1.9057,
      "step": 137222
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.1072916984558105,
      "learning_rate": 0.00021126721345649934,
      "loss": 2.3409,
      "step": 137223
    },
    {
      "epoch": 1.79,
      "grad_norm": 0.9975478053092957,
      "learning_rate": 0.00021126330592521135,
      "loss": 2.0818,
      "step": 137224
    },
    {
      "epoch": 1.79,
      "grad_norm": 0.9749409556388855,
      "learning_rate": 0.00021125939841042115,
      "loss": 1.9603,
      "step": 137225
    },
    {
      "epoch": 1.79,
      "grad_norm": 0.9885713458061218,
      "learning_rate": 0.00021125549091212953,
      "loss": 2.0318,
      "step": 137226
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.021115779876709,
      "learning_rate": 0.00021125158343033714,
      "loss": 2.0788,
      "step": 137227
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.1022082567214966,
      "learning_rate": 0.00021124767596504494,
      "loss": 2.1209,
      "step": 137228
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.0615407228469849,
      "learning_rate": 0.00021124376851625336,
      "loss": 2.1148,
      "step": 137229
    },
    {
      "epoch": 1.79,
      "grad_norm": 0.9374528527259827,
      "learning_rate": 0.0002112398610839632,
      "loss": 2.0846,
      "step": 137230
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.1104531288146973,
      "learning_rate": 0.00021123595366817524,
      "loss": 1.9171,
      "step": 137231
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.0875147581100464,
      "learning_rate": 0.00021123204626889017,
      "loss": 2.2823,
      "step": 137232
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.0680228471755981,
      "learning_rate": 0.00021122813888610877,
      "loss": 2.0183,
      "step": 137233
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.014072060585022,
      "learning_rate": 0.00021122423151983183,
      "loss": 2.0723,
      "step": 137234
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.1323004961013794,
      "learning_rate": 0.00021122032417005986,
      "loss": 2.2352,
      "step": 137235
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.0029244422912598,
      "learning_rate": 0.0002112164168367937,
      "loss": 2.2099,
      "step": 137236
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.2354247570037842,
      "learning_rate": 0.0002112125095200341,
      "loss": 2.1985,
      "step": 137237
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.1655733585357666,
      "learning_rate": 0.0002112086022197818,
      "loss": 2.1179,
      "step": 137238
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.1378843784332275,
      "learning_rate": 0.00021120469493603745,
      "loss": 2.0396,
      "step": 137239
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.0428353548049927,
      "learning_rate": 0.00021120078766880197,
      "loss": 2.0949,
      "step": 137240
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.1856187582015991,
      "learning_rate": 0.0002111968804180758,
      "loss": 2.1743,
      "step": 137241
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.0651227235794067,
      "learning_rate": 0.00021119297318385983,
      "loss": 2.1174,
      "step": 137242
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.3023884296417236,
      "learning_rate": 0.00021118906596615474,
      "loss": 2.1988,
      "step": 137243
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.0889838933944702,
      "learning_rate": 0.00021118515876496132,
      "loss": 2.0858,
      "step": 137244
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.1139652729034424,
      "learning_rate": 0.0002111812515802802,
      "loss": 2.1386,
      "step": 137245
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.3893963098526,
      "learning_rate": 0.00021117734441211225,
      "loss": 2.1853,
      "step": 137246
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.0793118476867676,
      "learning_rate": 0.00021117343726045803,
      "loss": 2.0801,
      "step": 137247
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.1160894632339478,
      "learning_rate": 0.0002111695301253184,
      "loss": 2.0034,
      "step": 137248
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.0441986322402954,
      "learning_rate": 0.00021116562300669403,
      "loss": 2.044,
      "step": 137249
    },
    {
      "epoch": 1.79,
      "grad_norm": 0.9236513376235962,
      "learning_rate": 0.0002111617159045856,
      "loss": 1.8571,
      "step": 137250
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.1493268013000488,
      "learning_rate": 0.00021115780881899386,
      "loss": 2.0377,
      "step": 137251
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.0543078184127808,
      "learning_rate": 0.00021115390174991966,
      "loss": 2.1352,
      "step": 137252
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.0232932567596436,
      "learning_rate": 0.0002111499946973636,
      "loss": 2.0968,
      "step": 137253
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.0508129596710205,
      "learning_rate": 0.0002111460876613264,
      "loss": 2.2416,
      "step": 137254
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.3562567234039307,
      "learning_rate": 0.00021114218064180882,
      "loss": 2.07,
      "step": 137255
    },
    {
      "epoch": 1.79,
      "grad_norm": 0.9560359120368958,
      "learning_rate": 0.0002111382736388116,
      "loss": 2.0605,
      "step": 137256
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.1452687978744507,
      "learning_rate": 0.00021113436665233542,
      "loss": 2.0082,
      "step": 137257
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.0609294176101685,
      "learning_rate": 0.00021113045968238107,
      "loss": 2.0748,
      "step": 137258
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.105163335800171,
      "learning_rate": 0.0002111265527289493,
      "loss": 2.1475,
      "step": 137259
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.0254356861114502,
      "learning_rate": 0.0002111226457920407,
      "loss": 2.1017,
      "step": 137260
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.255070447921753,
      "learning_rate": 0.00021111873887165605,
      "loss": 2.0819,
      "step": 137261
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.072261095046997,
      "learning_rate": 0.00021111483196779618,
      "loss": 2.1703,
      "step": 137262
    },
    {
      "epoch": 1.79,
      "grad_norm": 0.9787013530731201,
      "learning_rate": 0.00021111092508046175,
      "loss": 2.0257,
      "step": 137263
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.0317586660385132,
      "learning_rate": 0.0002111070182096534,
      "loss": 2.1045,
      "step": 137264
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.0771458148956299,
      "learning_rate": 0.00021110311135537203,
      "loss": 1.9688,
      "step": 137265
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.0454944372177124,
      "learning_rate": 0.00021109920451761822,
      "loss": 2.198,
      "step": 137266
    },
    {
      "epoch": 1.79,
      "grad_norm": 0.9645650386810303,
      "learning_rate": 0.0002110952976963927,
      "loss": 2.1947,
      "step": 137267
    },
    {
      "epoch": 1.79,
      "grad_norm": 0.9868322014808655,
      "learning_rate": 0.00021109139089169627,
      "loss": 2.0402,
      "step": 137268
    },
    {
      "epoch": 1.79,
      "grad_norm": 0.9966064691543579,
      "learning_rate": 0.0002110874841035297,
      "loss": 2.0069,
      "step": 137269
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.01071035861969,
      "learning_rate": 0.00021108357733189357,
      "loss": 2.0331,
      "step": 137270
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.0492899417877197,
      "learning_rate": 0.00021107967057678873,
      "loss": 2.0523,
      "step": 137271
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.0145548582077026,
      "learning_rate": 0.0002110757638382158,
      "loss": 2.0547,
      "step": 137272
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.165182113647461,
      "learning_rate": 0.00021107185711617555,
      "loss": 2.1696,
      "step": 137273
    },
    {
      "epoch": 1.79,
      "grad_norm": 0.9767396450042725,
      "learning_rate": 0.00021106795041066876,
      "loss": 2.0608,
      "step": 137274
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.1412203311920166,
      "learning_rate": 0.00021106404372169611,
      "loss": 2.0186,
      "step": 137275
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.116714596748352,
      "learning_rate": 0.00021106013704925834,
      "loss": 2.1895,
      "step": 137276
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.0796594619750977,
      "learning_rate": 0.00021105623039335624,
      "loss": 2.0429,
      "step": 137277
    },
    {
      "epoch": 1.79,
      "grad_norm": 0.9615845680236816,
      "learning_rate": 0.00021105232375399043,
      "loss": 2.1265,
      "step": 137278
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.1854692697525024,
      "learning_rate": 0.0002110484171311616,
      "loss": 2.1701,
      "step": 137279
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.3243839740753174,
      "learning_rate": 0.00021104451052487055,
      "loss": 1.9122,
      "step": 137280
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.0184091329574585,
      "learning_rate": 0.00021104060393511805,
      "loss": 1.9707,
      "step": 137281
    },
    {
      "epoch": 1.79,
      "grad_norm": 0.9397832155227661,
      "learning_rate": 0.00021103669736190472,
      "loss": 1.8871,
      "step": 137282
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.0612094402313232,
      "learning_rate": 0.0002110327908052315,
      "loss": 2.2196,
      "step": 137283
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.2818485498428345,
      "learning_rate": 0.00021102888426509885,
      "loss": 2.0315,
      "step": 137284
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.0036821365356445,
      "learning_rate": 0.00021102497774150763,
      "loss": 1.9726,
      "step": 137285
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.0987976789474487,
      "learning_rate": 0.00021102107123445853,
      "loss": 2.1429,
      "step": 137286
    },
    {
      "epoch": 1.79,
      "grad_norm": 0.9452219009399414,
      "learning_rate": 0.00021101716474395226,
      "loss": 2.1093,
      "step": 137287
    },
    {
      "epoch": 1.79,
      "grad_norm": 0.9127551913261414,
      "learning_rate": 0.00021101325826998962,
      "loss": 2.0277,
      "step": 137288
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.0759719610214233,
      "learning_rate": 0.00021100935181257138,
      "loss": 2.2022,
      "step": 137289
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.0862020254135132,
      "learning_rate": 0.00021100544537169807,
      "loss": 2.1144,
      "step": 137290
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.1874408721923828,
      "learning_rate": 0.00021100153894737053,
      "loss": 2.1593,
      "step": 137291
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.0301933288574219,
      "learning_rate": 0.0002109976325395895,
      "loss": 2.2538,
      "step": 137292
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.0161398649215698,
      "learning_rate": 0.0002109937261483557,
      "loss": 1.914,
      "step": 137293
    },
    {
      "epoch": 1.79,
      "grad_norm": 0.9747833013534546,
      "learning_rate": 0.00021098981977366979,
      "loss": 2.0565,
      "step": 137294
    },
    {
      "epoch": 1.79,
      "grad_norm": 0.9836922287940979,
      "learning_rate": 0.00021098591341553273,
      "loss": 1.9045,
      "step": 137295
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.0439071655273438,
      "learning_rate": 0.00021098200707394494,
      "loss": 2.312,
      "step": 137296
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.030703067779541,
      "learning_rate": 0.00021097810074890728,
      "loss": 1.9568,
      "step": 137297
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.8060530424118042,
      "learning_rate": 0.00021097419444042045,
      "loss": 2.0344,
      "step": 137298
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.0085644721984863,
      "learning_rate": 0.0002109702881484852,
      "loss": 2.0955,
      "step": 137299
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.0233887434005737,
      "learning_rate": 0.00021096638187310224,
      "loss": 1.9857,
      "step": 137300
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.0891906023025513,
      "learning_rate": 0.00021096247561427244,
      "loss": 2.0847,
      "step": 137301
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.1770801544189453,
      "learning_rate": 0.0002109585693719963,
      "loss": 2.0349,
      "step": 137302
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.069873571395874,
      "learning_rate": 0.00021095466314627464,
      "loss": 1.9591,
      "step": 137303
    },
    {
      "epoch": 1.79,
      "grad_norm": 0.9816282987594604,
      "learning_rate": 0.00021095075693710819,
      "loss": 1.9309,
      "step": 137304
    },
    {
      "epoch": 1.79,
      "grad_norm": 0.9967129826545715,
      "learning_rate": 0.00021094685074449763,
      "loss": 2.1022,
      "step": 137305
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.0706331729888916,
      "learning_rate": 0.00021094294456844377,
      "loss": 2.1108,
      "step": 137306
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.0507971048355103,
      "learning_rate": 0.0002109390384089474,
      "loss": 2.0276,
      "step": 137307
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.0046124458312988,
      "learning_rate": 0.00021093513226600906,
      "loss": 2.1513,
      "step": 137308
    },
    {
      "epoch": 1.79,
      "grad_norm": 0.9109771251678467,
      "learning_rate": 0.00021093122613962954,
      "loss": 2.1302,
      "step": 137309
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.0577430725097656,
      "learning_rate": 0.00021092732002980956,
      "loss": 2.2525,
      "step": 137310
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.1025934219360352,
      "learning_rate": 0.0002109234139365499,
      "loss": 2.2013,
      "step": 137311
    },
    {
      "epoch": 1.79,
      "grad_norm": 0.9471412897109985,
      "learning_rate": 0.00021091950785985124,
      "loss": 2.1164,
      "step": 137312
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.0843093395233154,
      "learning_rate": 0.00021091560179971447,
      "loss": 2.1586,
      "step": 137313
    },
    {
      "epoch": 1.79,
      "grad_norm": 0.9963166117668152,
      "learning_rate": 0.00021091169575614005,
      "loss": 2.0907,
      "step": 137314
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.031283974647522,
      "learning_rate": 0.0002109077897291288,
      "loss": 2.2751,
      "step": 137315
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.0955196619033813,
      "learning_rate": 0.00021090388371868148,
      "loss": 2.1297,
      "step": 137316
    },
    {
      "epoch": 1.79,
      "grad_norm": 0.9496835470199585,
      "learning_rate": 0.00021089997772479882,
      "loss": 2.2977,
      "step": 137317
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.1861976385116577,
      "learning_rate": 0.00021089607174748154,
      "loss": 1.9963,
      "step": 137318
    },
    {
      "epoch": 1.79,
      "grad_norm": 0.9352657794952393,
      "learning_rate": 0.0002108921657867305,
      "loss": 1.7359,
      "step": 137319
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.0268446207046509,
      "learning_rate": 0.00021088825984254614,
      "loss": 2.1961,
      "step": 137320
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.166069507598877,
      "learning_rate": 0.00021088435391492931,
      "loss": 2.2933,
      "step": 137321
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.192757248878479,
      "learning_rate": 0.0002108804480038808,
      "loss": 2.243,
      "step": 137322
    },
    {
      "epoch": 1.79,
      "grad_norm": 0.9690738916397095,
      "learning_rate": 0.0002108765421094013,
      "loss": 2.0217,
      "step": 137323
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.05830717086792,
      "learning_rate": 0.0002108726362314915,
      "loss": 2.0293,
      "step": 137324
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.0247670412063599,
      "learning_rate": 0.00021086873037015223,
      "loss": 2.2554,
      "step": 137325
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.0617259740829468,
      "learning_rate": 0.0002108648245253842,
      "loss": 2.1806,
      "step": 137326
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.0149794816970825,
      "learning_rate": 0.00021086091869718796,
      "loss": 2.0274,
      "step": 137327
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.3979849815368652,
      "learning_rate": 0.0002108570128855644,
      "loss": 1.9962,
      "step": 137328
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.0204230546951294,
      "learning_rate": 0.00021085310709051416,
      "loss": 2.089,
      "step": 137329
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.0825514793395996,
      "learning_rate": 0.00021084920131203804,
      "loss": 2.0722,
      "step": 137330
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.1987758874893188,
      "learning_rate": 0.0002108452955501367,
      "loss": 1.9905,
      "step": 137331
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.085047960281372,
      "learning_rate": 0.00021084138980481104,
      "loss": 2.1193,
      "step": 137332
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.0038139820098877,
      "learning_rate": 0.00021083748407606156,
      "loss": 2.1267,
      "step": 137333
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.1588109731674194,
      "learning_rate": 0.00021083357836388907,
      "loss": 2.2538,
      "step": 137334
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.2552669048309326,
      "learning_rate": 0.00021082967266829427,
      "loss": 2.0112,
      "step": 137335
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.0868901014328003,
      "learning_rate": 0.00021082576698927793,
      "loss": 2.0005,
      "step": 137336
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.1791517734527588,
      "learning_rate": 0.00021082186132684072,
      "loss": 2.1413,
      "step": 137337
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.1465784311294556,
      "learning_rate": 0.00021081795568098354,
      "loss": 2.1145,
      "step": 137338
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.1819616556167603,
      "learning_rate": 0.00021081405005170695,
      "loss": 2.1534,
      "step": 137339
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.0264261960983276,
      "learning_rate": 0.0002108101444390117,
      "loss": 2.1903,
      "step": 137340
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.1706600189208984,
      "learning_rate": 0.0002108062388428985,
      "loss": 2.1604,
      "step": 137341
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.0711982250213623,
      "learning_rate": 0.00021080233326336805,
      "loss": 2.0799,
      "step": 137342
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.0021675825119019,
      "learning_rate": 0.0002107984277004212,
      "loss": 2.2125,
      "step": 137343
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.116952896118164,
      "learning_rate": 0.00021079452215405864,
      "loss": 2.1934,
      "step": 137344
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.1177982091903687,
      "learning_rate": 0.000210790616624281,
      "loss": 2.2366,
      "step": 137345
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.1815122365951538,
      "learning_rate": 0.00021078671111108906,
      "loss": 2.2222,
      "step": 137346
    },
    {
      "epoch": 1.79,
      "grad_norm": 0.9601410627365112,
      "learning_rate": 0.0002107828056144836,
      "loss": 2.0444,
      "step": 137347
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.1117908954620361,
      "learning_rate": 0.00021077890013446526,
      "loss": 1.9764,
      "step": 137348
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.0237610340118408,
      "learning_rate": 0.00021077499467103485,
      "loss": 2.0155,
      "step": 137349
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.0938864946365356,
      "learning_rate": 0.00021077108922419306,
      "loss": 2.0869,
      "step": 137350
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.0871742963790894,
      "learning_rate": 0.00021076718379394054,
      "loss": 2.064,
      "step": 137351
    },
    {
      "epoch": 1.79,
      "grad_norm": 0.9126558303833008,
      "learning_rate": 0.00021076327838027813,
      "loss": 1.9049,
      "step": 137352
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.1008931398391724,
      "learning_rate": 0.00021075937298320647,
      "loss": 2.2289,
      "step": 137353
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.2019990682601929,
      "learning_rate": 0.0002107554676027264,
      "loss": 2.2973,
      "step": 137354
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.0083444118499756,
      "learning_rate": 0.0002107515622388385,
      "loss": 2.0961,
      "step": 137355
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.0963144302368164,
      "learning_rate": 0.0002107476568915437,
      "loss": 2.1131,
      "step": 137356
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.1422122716903687,
      "learning_rate": 0.00021074375156084248,
      "loss": 1.9329,
      "step": 137357
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.0374728441238403,
      "learning_rate": 0.00021073984624673563,
      "loss": 2.1568,
      "step": 137358
    },
    {
      "epoch": 1.79,
      "grad_norm": 0.9951232075691223,
      "learning_rate": 0.00021073594094922398,
      "loss": 2.2209,
      "step": 137359
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.106484055519104,
      "learning_rate": 0.00021073203566830822,
      "loss": 2.2214,
      "step": 137360
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.0452998876571655,
      "learning_rate": 0.00021072813040398913,
      "loss": 1.9478,
      "step": 137361
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.1132694482803345,
      "learning_rate": 0.00021072422515626735,
      "loss": 2.0504,
      "step": 137362
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.0057865381240845,
      "learning_rate": 0.00021072031992514352,
      "loss": 2.1339,
      "step": 137363
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.0145843029022217,
      "learning_rate": 0.0002107164147106185,
      "loss": 2.0084,
      "step": 137364
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.0278732776641846,
      "learning_rate": 0.000210712509512693,
      "loss": 2.0502,
      "step": 137365
    },
    {
      "epoch": 1.79,
      "grad_norm": 0.961108386516571,
      "learning_rate": 0.00021070860433136768,
      "loss": 1.957,
      "step": 137366
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.1770991086959839,
      "learning_rate": 0.00021070469916664337,
      "loss": 2.1511,
      "step": 137367
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.105891227722168,
      "learning_rate": 0.00021070079401852085,
      "loss": 2.2245,
      "step": 137368
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.0338249206542969,
      "learning_rate": 0.00021069688888700062,
      "loss": 2.1175,
      "step": 137369
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.029758334159851,
      "learning_rate": 0.0002106929837720835,
      "loss": 2.17,
      "step": 137370
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.0775015354156494,
      "learning_rate": 0.00021068907867377026,
      "loss": 2.0681,
      "step": 137371
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.0510681867599487,
      "learning_rate": 0.00021068517359206158,
      "loss": 2.0668,
      "step": 137372
    },
    {
      "epoch": 1.79,
      "grad_norm": 0.9017364382743835,
      "learning_rate": 0.00021068126852695824,
      "loss": 1.9341,
      "step": 137373
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.1704168319702148,
      "learning_rate": 0.00021067736347846108,
      "loss": 2.1551,
      "step": 137374
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.0897736549377441,
      "learning_rate": 0.0002106734584465705,
      "loss": 2.1226,
      "step": 137375
    },
    {
      "epoch": 1.79,
      "grad_norm": 0.9978426694869995,
      "learning_rate": 0.00021066955343128745,
      "loss": 2.171,
      "step": 137376
    },
    {
      "epoch": 1.79,
      "grad_norm": 0.9951279759407043,
      "learning_rate": 0.0002106656484326126,
      "loss": 2.0897,
      "step": 137377
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.0879974365234375,
      "learning_rate": 0.00021066174345054668,
      "loss": 2.165,
      "step": 137378
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.0604004859924316,
      "learning_rate": 0.0002106578384850905,
      "loss": 2.1811,
      "step": 137379
    },
    {
      "epoch": 1.79,
      "grad_norm": 0.9713137149810791,
      "learning_rate": 0.00021065393353624474,
      "loss": 2.2599,
      "step": 137380
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.0224522352218628,
      "learning_rate": 0.00021065002860401005,
      "loss": 2.3411,
      "step": 137381
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.0273487567901611,
      "learning_rate": 0.0002106461236883872,
      "loss": 1.9827,
      "step": 137382
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.0210955142974854,
      "learning_rate": 0.0002106422187893769,
      "loss": 2.0871,
      "step": 137383
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.1240850687026978,
      "learning_rate": 0.0002106383139069799,
      "loss": 2.1094,
      "step": 137384
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.1015288829803467,
      "learning_rate": 0.0002106344090411969,
      "loss": 2.0459,
      "step": 137385
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.0339715480804443,
      "learning_rate": 0.00021063050419202882,
      "loss": 2.0393,
      "step": 137386
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.2488406896591187,
      "learning_rate": 0.00021062659935947608,
      "loss": 1.9415,
      "step": 137387
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.1389307975769043,
      "learning_rate": 0.00021062269454353953,
      "loss": 2.2019,
      "step": 137388
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.0509119033813477,
      "learning_rate": 0.0002106187897442199,
      "loss": 2.1885,
      "step": 137389
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.0220463275909424,
      "learning_rate": 0.00021061488496151797,
      "loss": 1.9311,
      "step": 137390
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.057321310043335,
      "learning_rate": 0.00021061098019543439,
      "loss": 2.4566,
      "step": 137391
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.0918092727661133,
      "learning_rate": 0.0002106070754459699,
      "loss": 1.9985,
      "step": 137392
    },
    {
      "epoch": 1.79,
      "grad_norm": 0.9428844451904297,
      "learning_rate": 0.00021060317071312535,
      "loss": 2.0757,
      "step": 137393
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.0357451438903809,
      "learning_rate": 0.0002105992659969013,
      "loss": 2.0675,
      "step": 137394
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.1325145959854126,
      "learning_rate": 0.00021059536129729847,
      "loss": 2.2455,
      "step": 137395
    },
    {
      "epoch": 1.79,
      "grad_norm": 0.9750155210494995,
      "learning_rate": 0.0002105914566143177,
      "loss": 1.9001,
      "step": 137396
    },
    {
      "epoch": 1.79,
      "grad_norm": 0.9808769226074219,
      "learning_rate": 0.00021058755194795962,
      "loss": 2.0085,
      "step": 137397
    },
    {
      "epoch": 1.79,
      "grad_norm": 0.9756973385810852,
      "learning_rate": 0.00021058364729822506,
      "loss": 2.0475,
      "step": 137398
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.0282838344573975,
      "learning_rate": 0.00021057974266511478,
      "loss": 1.9517,
      "step": 137399
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.0878630876541138,
      "learning_rate": 0.0002105758380486293,
      "loss": 1.9948,
      "step": 137400
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.054532527923584,
      "learning_rate": 0.00021057193344876943,
      "loss": 2.0589,
      "step": 137401
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.0021864175796509,
      "learning_rate": 0.00021056802886553592,
      "loss": 2.1857,
      "step": 137402
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.0269711017608643,
      "learning_rate": 0.00021056412429892955,
      "loss": 2.0143,
      "step": 137403
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.0117229223251343,
      "learning_rate": 0.00021056021974895093,
      "loss": 1.9635,
      "step": 137404
    },
    {
      "epoch": 1.79,
      "grad_norm": 0.978709876537323,
      "learning_rate": 0.00021055631521560103,
      "loss": 1.9639,
      "step": 137405
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.0814688205718994,
      "learning_rate": 0.00021055241069888029,
      "loss": 2.077,
      "step": 137406
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.071523666381836,
      "learning_rate": 0.0002105485061987895,
      "loss": 1.9029,
      "step": 137407
    },
    {
      "epoch": 1.79,
      "grad_norm": 0.9898715615272522,
      "learning_rate": 0.00021054460171532945,
      "loss": 1.9658,
      "step": 137408
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.048513412475586,
      "learning_rate": 0.00021054069724850085,
      "loss": 2.1338,
      "step": 137409
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.001903772354126,
      "learning_rate": 0.00021053679279830443,
      "loss": 2.0436,
      "step": 137410
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.1559866666793823,
      "learning_rate": 0.00021053288836474102,
      "loss": 1.869,
      "step": 137411
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.056555151939392,
      "learning_rate": 0.00021052898394781111,
      "loss": 2.0738,
      "step": 137412
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.334673523902893,
      "learning_rate": 0.00021052507954751558,
      "loss": 2.0923,
      "step": 137413
    },
    {
      "epoch": 1.79,
      "grad_norm": 0.9869922399520874,
      "learning_rate": 0.0002105211751638551,
      "loss": 2.091,
      "step": 137414
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.0786670446395874,
      "learning_rate": 0.00021051727079683043,
      "loss": 1.927,
      "step": 137415
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.1878515481948853,
      "learning_rate": 0.00021051336644644226,
      "loss": 2.0937,
      "step": 137416
    },
    {
      "epoch": 1.79,
      "grad_norm": 0.9838895797729492,
      "learning_rate": 0.0002105094621126915,
      "loss": 2.0259,
      "step": 137417
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.0628139972686768,
      "learning_rate": 0.0002105055577955786,
      "loss": 2.1782,
      "step": 137418
    },
    {
      "epoch": 1.79,
      "grad_norm": 0.9504954814910889,
      "learning_rate": 0.00021050165349510442,
      "loss": 2.1023,
      "step": 137419
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.0424314737319946,
      "learning_rate": 0.00021049774921126963,
      "loss": 2.3433,
      "step": 137420
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.0733617544174194,
      "learning_rate": 0.000210493844944075,
      "loss": 2.1348,
      "step": 137421
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.0099714994430542,
      "learning_rate": 0.00021048994069352124,
      "loss": 2.1175,
      "step": 137422
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.159088134765625,
      "learning_rate": 0.00021048603645960922,
      "loss": 2.0769,
      "step": 137423
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.0683454275131226,
      "learning_rate": 0.00021048213224233948,
      "loss": 2.052,
      "step": 137424
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.0519003868103027,
      "learning_rate": 0.00021047822804171274,
      "loss": 2.0885,
      "step": 137425
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.0168216228485107,
      "learning_rate": 0.00021047432385772985,
      "loss": 2.1766,
      "step": 137426
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.009214997291565,
      "learning_rate": 0.00021047041969039142,
      "loss": 2.2671,
      "step": 137427
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.086676001548767,
      "learning_rate": 0.00021046651553969823,
      "loss": 2.033,
      "step": 137428
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.0908633470535278,
      "learning_rate": 0.00021046261140565103,
      "loss": 2.1278,
      "step": 137429
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.1044514179229736,
      "learning_rate": 0.00021045870728825055,
      "loss": 2.1287,
      "step": 137430
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.0465307235717773,
      "learning_rate": 0.00021045480318749747,
      "loss": 2.1208,
      "step": 137431
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.077467918395996,
      "learning_rate": 0.00021045089910339253,
      "loss": 2.0243,
      "step": 137432
    },
    {
      "epoch": 1.79,
      "grad_norm": 0.9854901432991028,
      "learning_rate": 0.00021044699503593642,
      "loss": 2.2708,
      "step": 137433
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.0909377336502075,
      "learning_rate": 0.0002104430909851299,
      "loss": 1.9074,
      "step": 137434
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.0225090980529785,
      "learning_rate": 0.00021043918695097375,
      "loss": 2.0358,
      "step": 137435
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.0074573755264282,
      "learning_rate": 0.00021043528293346862,
      "loss": 1.9338,
      "step": 137436
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.0589133501052856,
      "learning_rate": 0.00021043137893261526,
      "loss": 1.9405,
      "step": 137437
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.0533738136291504,
      "learning_rate": 0.00021042747494841437,
      "loss": 1.7592,
      "step": 137438
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.1306369304656982,
      "learning_rate": 0.00021042357098086678,
      "loss": 2.1233,
      "step": 137439
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.057397723197937,
      "learning_rate": 0.0002104196670299731,
      "loss": 2.073,
      "step": 137440
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.0869942903518677,
      "learning_rate": 0.00021041576309573413,
      "loss": 2.1969,
      "step": 137441
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.0610300302505493,
      "learning_rate": 0.00021041185917815047,
      "loss": 2.0119,
      "step": 137442
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.0314041376113892,
      "learning_rate": 0.00021040795527722299,
      "loss": 2.1612,
      "step": 137443
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.106839656829834,
      "learning_rate": 0.00021040405139295234,
      "loss": 2.1259,
      "step": 137444
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.098195195198059,
      "learning_rate": 0.00021040014752533927,
      "loss": 2.1952,
      "step": 137445
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.1612361669540405,
      "learning_rate": 0.00021039624367438458,
      "loss": 2.2987,
      "step": 137446
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.0393223762512207,
      "learning_rate": 0.0002103923398400889,
      "loss": 2.0513,
      "step": 137447
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.0723265409469604,
      "learning_rate": 0.00021038843602245292,
      "loss": 2.1491,
      "step": 137448
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.2234805822372437,
      "learning_rate": 0.00021038453222147743,
      "loss": 2.0342,
      "step": 137449
    },
    {
      "epoch": 1.79,
      "grad_norm": 0.9758260250091553,
      "learning_rate": 0.00021038062843716312,
      "loss": 2.0299,
      "step": 137450
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.4174437522888184,
      "learning_rate": 0.00021037672466951075,
      "loss": 2.0607,
      "step": 137451
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.0743647813796997,
      "learning_rate": 0.00021037282091852108,
      "loss": 2.1895,
      "step": 137452
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.099557876586914,
      "learning_rate": 0.00021036891718419486,
      "loss": 1.9343,
      "step": 137453
    },
    {
      "epoch": 1.79,
      "grad_norm": 0.926708996295929,
      "learning_rate": 0.00021036501346653264,
      "loss": 1.8675,
      "step": 137454
    },
    {
      "epoch": 1.79,
      "grad_norm": 0.9849438667297363,
      "learning_rate": 0.00021036110976553528,
      "loss": 2.181,
      "step": 137455
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.2881726026535034,
      "learning_rate": 0.00021035720608120347,
      "loss": 2.0741,
      "step": 137456
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.1830790042877197,
      "learning_rate": 0.00021035330241353796,
      "loss": 2.318,
      "step": 137457
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.055050253868103,
      "learning_rate": 0.00021034939876253946,
      "loss": 2.0479,
      "step": 137458
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.1146748065948486,
      "learning_rate": 0.0002103454951282087,
      "loss": 2.5129,
      "step": 137459
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.0407636165618896,
      "learning_rate": 0.0002103415915105465,
      "loss": 2.2532,
      "step": 137460
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.2290403842926025,
      "learning_rate": 0.0002103376879095534,
      "loss": 2.2179,
      "step": 137461
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.1290849447250366,
      "learning_rate": 0.00021033378432523022,
      "loss": 2.2621,
      "step": 137462
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.0163146257400513,
      "learning_rate": 0.00021032988075757767,
      "loss": 2.1085,
      "step": 137463
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.162876009941101,
      "learning_rate": 0.00021032597720659646,
      "loss": 2.3069,
      "step": 137464
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.011889100074768,
      "learning_rate": 0.00021032207367228736,
      "loss": 2.0705,
      "step": 137465
    },
    {
      "epoch": 1.79,
      "grad_norm": 0.9486874938011169,
      "learning_rate": 0.00021031817015465122,
      "loss": 1.8148,
      "step": 137466
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.044432520866394,
      "learning_rate": 0.0002103142666536885,
      "loss": 2.097,
      "step": 137467
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.0070927143096924,
      "learning_rate": 0.00021031036316940003,
      "loss": 2.0418,
      "step": 137468
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.0136299133300781,
      "learning_rate": 0.00021030645970178657,
      "loss": 2.0882,
      "step": 137469
    },
    {
      "epoch": 1.79,
      "grad_norm": 0.9608404040336609,
      "learning_rate": 0.00021030255625084884,
      "loss": 2.1673,
      "step": 137470
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.003739833831787,
      "learning_rate": 0.00021029865281658756,
      "loss": 1.904,
      "step": 137471
    },
    {
      "epoch": 1.79,
      "grad_norm": 0.9995588064193726,
      "learning_rate": 0.00021029474939900355,
      "loss": 2.0057,
      "step": 137472
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.010275959968567,
      "learning_rate": 0.0002102908459980973,
      "loss": 2.0859,
      "step": 137473
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.1386773586273193,
      "learning_rate": 0.00021028694261386973,
      "loss": 2.142,
      "step": 137474
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.0235611200332642,
      "learning_rate": 0.0002102830392463215,
      "loss": 2.0229,
      "step": 137475
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.097373366355896,
      "learning_rate": 0.00021027913589545335,
      "loss": 2.3069,
      "step": 137476
    },
    {
      "epoch": 1.79,
      "grad_norm": 0.9996781945228577,
      "learning_rate": 0.00021027523256126596,
      "loss": 1.9282,
      "step": 137477
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.078127384185791,
      "learning_rate": 0.00021027132924376026,
      "loss": 2.0076,
      "step": 137478
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.1005324125289917,
      "learning_rate": 0.00021026742594293667,
      "loss": 2.1996,
      "step": 137479
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.0699162483215332,
      "learning_rate": 0.00021026352265879607,
      "loss": 2.197,
      "step": 137480
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.0448811054229736,
      "learning_rate": 0.00021025961939133918,
      "loss": 2.1129,
      "step": 137481
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.0760260820388794,
      "learning_rate": 0.0002102557161405667,
      "loss": 2.1944,
      "step": 137482
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.264780879020691,
      "learning_rate": 0.0002102518129064794,
      "loss": 2.0268,
      "step": 137483
    },
    {
      "epoch": 1.79,
      "grad_norm": 0.9588685035705566,
      "learning_rate": 0.00021024790968907813,
      "loss": 2.1804,
      "step": 137484
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.017852544784546,
      "learning_rate": 0.0002102440064883633,
      "loss": 2.1756,
      "step": 137485
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.0822291374206543,
      "learning_rate": 0.00021024010330433578,
      "loss": 2.0107,
      "step": 137486
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.0362714529037476,
      "learning_rate": 0.00021023620013699637,
      "loss": 2.031,
      "step": 137487
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.02830970287323,
      "learning_rate": 0.00021023229698634573,
      "loss": 2.0139,
      "step": 137488
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.1125694513320923,
      "learning_rate": 0.00021022839385238456,
      "loss": 2.0285,
      "step": 137489
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.1120145320892334,
      "learning_rate": 0.00021022449073511383,
      "loss": 2.2247,
      "step": 137490
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.1123974323272705,
      "learning_rate": 0.00021022058763453386,
      "loss": 2.1989,
      "step": 137491
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.049591064453125,
      "learning_rate": 0.00021021668455064563,
      "loss": 2.2713,
      "step": 137492
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.10688054561615,
      "learning_rate": 0.00021021278148344977,
      "loss": 2.0815,
      "step": 137493
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.038387656211853,
      "learning_rate": 0.00021020887843294713,
      "loss": 2.1154,
      "step": 137494
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.411377191543579,
      "learning_rate": 0.00021020497539913827,
      "loss": 2.0011,
      "step": 137495
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.2481170892715454,
      "learning_rate": 0.00021020107238202415,
      "loss": 2.0692,
      "step": 137496
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.234830379486084,
      "learning_rate": 0.00021019716938160522,
      "loss": 2.0906,
      "step": 137497
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.0429306030273438,
      "learning_rate": 0.00021019326639788236,
      "loss": 1.8949,
      "step": 137498
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.0807018280029297,
      "learning_rate": 0.0002101893634308562,
      "loss": 2.1171,
      "step": 137499
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.1230872869491577,
      "learning_rate": 0.00021018546048052762,
      "loss": 2.0657,
      "step": 137500
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.0867993831634521,
      "learning_rate": 0.0002101815575468972,
      "loss": 2.2869,
      "step": 137501
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.0714353322982788,
      "learning_rate": 0.00021017765462996585,
      "loss": 2.183,
      "step": 137502
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.0871837139129639,
      "learning_rate": 0.0002101737517297341,
      "loss": 2.1412,
      "step": 137503
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.0109820365905762,
      "learning_rate": 0.00021016984884620268,
      "loss": 2.1503,
      "step": 137504
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.0221853256225586,
      "learning_rate": 0.00021016594597937242,
      "loss": 1.8564,
      "step": 137505
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.0130609273910522,
      "learning_rate": 0.00021016204312924398,
      "loss": 1.9548,
      "step": 137506
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.0176496505737305,
      "learning_rate": 0.0002101581402958181,
      "loss": 2.0955,
      "step": 137507
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.0188885927200317,
      "learning_rate": 0.00021015423747909563,
      "loss": 1.9474,
      "step": 137508
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.0320101976394653,
      "learning_rate": 0.00021015033467907713,
      "loss": 2.0939,
      "step": 137509
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.0446574687957764,
      "learning_rate": 0.00021014643189576335,
      "loss": 1.8276,
      "step": 137510
    },
    {
      "epoch": 1.79,
      "grad_norm": 0.9904429912567139,
      "learning_rate": 0.00021014252912915506,
      "loss": 1.8855,
      "step": 137511
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.1228829622268677,
      "learning_rate": 0.00021013862637925295,
      "loss": 2.0843,
      "step": 137512
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.0572301149368286,
      "learning_rate": 0.0002101347236460577,
      "loss": 2.0725,
      "step": 137513
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.0374419689178467,
      "learning_rate": 0.00021013082092957028,
      "loss": 2.2298,
      "step": 137514
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.0805532932281494,
      "learning_rate": 0.0002101269182297911,
      "loss": 2.0735,
      "step": 137515
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.1438841819763184,
      "learning_rate": 0.00021012301554672108,
      "loss": 2.1166,
      "step": 137516
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.053317666053772,
      "learning_rate": 0.0002101191128803609,
      "loss": 1.88,
      "step": 137517
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.0504565238952637,
      "learning_rate": 0.0002101152102307112,
      "loss": 2.172,
      "step": 137518
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.2098193168640137,
      "learning_rate": 0.00021011130759777282,
      "loss": 1.9282,
      "step": 137519
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.0696958303451538,
      "learning_rate": 0.00021010740498154647,
      "loss": 2.0557,
      "step": 137520
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.139681339263916,
      "learning_rate": 0.00021010350238203282,
      "loss": 2.1533,
      "step": 137521
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.0999149084091187,
      "learning_rate": 0.0002100995997992326,
      "loss": 2.0863,
      "step": 137522
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.2525343894958496,
      "learning_rate": 0.00021009569723314664,
      "loss": 2.1546,
      "step": 137523
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.1173665523529053,
      "learning_rate": 0.00021009179468377555,
      "loss": 1.956,
      "step": 137524
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.0148751735687256,
      "learning_rate": 0.00021008789215112004,
      "loss": 2.0834,
      "step": 137525
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.169650912284851,
      "learning_rate": 0.00021008398963518092,
      "loss": 1.8556,
      "step": 137526
    },
    {
      "epoch": 1.79,
      "grad_norm": 0.9416636824607849,
      "learning_rate": 0.00021008008713595895,
      "loss": 2.0186,
      "step": 137527
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.003299593925476,
      "learning_rate": 0.0002100761846534547,
      "loss": 2.0821,
      "step": 137528
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.2115638256072998,
      "learning_rate": 0.00021007228218766901,
      "loss": 2.0468,
      "step": 137529
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.2587083578109741,
      "learning_rate": 0.00021006837973860256,
      "loss": 2.1496,
      "step": 137530
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.3031246662139893,
      "learning_rate": 0.00021006447730625616,
      "loss": 2.1711,
      "step": 137531
    },
    {
      "epoch": 1.79,
      "grad_norm": 0.9346675276756287,
      "learning_rate": 0.00021006057489063044,
      "loss": 2.0107,
      "step": 137532
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.1740351915359497,
      "learning_rate": 0.00021005667249172617,
      "loss": 2.1477,
      "step": 137533
    },
    {
      "epoch": 1.79,
      "grad_norm": 0.9660859704017639,
      "learning_rate": 0.00021005277010954405,
      "loss": 2.0266,
      "step": 137534
    },
    {
      "epoch": 1.79,
      "grad_norm": 0.9852338433265686,
      "learning_rate": 0.0002100488677440848,
      "loss": 1.977,
      "step": 137535
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.098186731338501,
      "learning_rate": 0.0002100449653953491,
      "loss": 2.1777,
      "step": 137536
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.0774171352386475,
      "learning_rate": 0.00021004106306333787,
      "loss": 2.042,
      "step": 137537
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.11110520362854,
      "learning_rate": 0.0002100371607480517,
      "loss": 2.1482,
      "step": 137538
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.0207270383834839,
      "learning_rate": 0.00021003325844949126,
      "loss": 1.9476,
      "step": 137539
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.0254002809524536,
      "learning_rate": 0.00021002935616765733,
      "loss": 2.154,
      "step": 137540
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.0592087507247925,
      "learning_rate": 0.00021002545390255063,
      "loss": 2.0984,
      "step": 137541
    },
    {
      "epoch": 1.79,
      "grad_norm": 0.9750319719314575,
      "learning_rate": 0.00021002155165417193,
      "loss": 1.9921,
      "step": 137542
    },
    {
      "epoch": 1.79,
      "grad_norm": 0.9674212336540222,
      "learning_rate": 0.0002100176494225219,
      "loss": 2.3988,
      "step": 137543
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.1420421600341797,
      "learning_rate": 0.0002100137472076013,
      "loss": 2.0037,
      "step": 137544
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.0438547134399414,
      "learning_rate": 0.00021000984500941093,
      "loss": 1.8708,
      "step": 137545
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.06512451171875,
      "learning_rate": 0.00021000594282795132,
      "loss": 2.1275,
      "step": 137546
    },
    {
      "epoch": 1.79,
      "grad_norm": 0.9307476878166199,
      "learning_rate": 0.0002100020406632233,
      "loss": 2.0962,
      "step": 137547
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.183815598487854,
      "learning_rate": 0.00020999813851522764,
      "loss": 2.0299,
      "step": 137548
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.1085408926010132,
      "learning_rate": 0.000209994236383965,
      "loss": 2.2251,
      "step": 137549
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.0959863662719727,
      "learning_rate": 0.0002099903342694361,
      "loss": 2.148,
      "step": 137550
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.0003336668014526,
      "learning_rate": 0.00020998643217164185,
      "loss": 1.9351,
      "step": 137551
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.0564757585525513,
      "learning_rate": 0.00020998253009058267,
      "loss": 2.2017,
      "step": 137552
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.0072388648986816,
      "learning_rate": 0.00020997862802625948,
      "loss": 2.094,
      "step": 137553
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.1007585525512695,
      "learning_rate": 0.00020997472597867294,
      "loss": 2.0582,
      "step": 137554
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.0656660795211792,
      "learning_rate": 0.0002099708239478238,
      "loss": 2.2139,
      "step": 137555
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.0404125452041626,
      "learning_rate": 0.0002099669219337128,
      "loss": 2.019,
      "step": 137556
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.428967833518982,
      "learning_rate": 0.00020996301993634074,
      "loss": 2.057,
      "step": 137557
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.0186632871627808,
      "learning_rate": 0.00020995911795570812,
      "loss": 2.0056,
      "step": 137558
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.0691614151000977,
      "learning_rate": 0.00020995521599181583,
      "loss": 2.1669,
      "step": 137559
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.2462122440338135,
      "learning_rate": 0.0002099513140446646,
      "loss": 1.9493,
      "step": 137560
    },
    {
      "epoch": 1.79,
      "grad_norm": 0.9898229241371155,
      "learning_rate": 0.00020994741211425506,
      "loss": 2.1401,
      "step": 137561
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.119539499282837,
      "learning_rate": 0.000209943510200588,
      "loss": 2.2952,
      "step": 137562
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.068623661994934,
      "learning_rate": 0.00020993960830366427,
      "loss": 2.0783,
      "step": 137563
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.1690768003463745,
      "learning_rate": 0.0002099357064234844,
      "loss": 1.9474,
      "step": 137564
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.0550137758255005,
      "learning_rate": 0.0002099318045600491,
      "loss": 2.0897,
      "step": 137565
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.1426290273666382,
      "learning_rate": 0.00020992790271335922,
      "loss": 2.0057,
      "step": 137566
    },
    {
      "epoch": 1.79,
      "grad_norm": 0.9437685012817383,
      "learning_rate": 0.00020992400088341544,
      "loss": 2.1848,
      "step": 137567
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.0937587022781372,
      "learning_rate": 0.00020992009907021847,
      "loss": 2.3094,
      "step": 137568
    },
    {
      "epoch": 1.79,
      "grad_norm": 0.9957440495491028,
      "learning_rate": 0.0002099161972737692,
      "loss": 2.0224,
      "step": 137569
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.1944196224212646,
      "learning_rate": 0.0002099122954940681,
      "loss": 2.1239,
      "step": 137570
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.1766349077224731,
      "learning_rate": 0.00020990839373111599,
      "loss": 2.0531,
      "step": 137571
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.0371031761169434,
      "learning_rate": 0.00020990449198491363,
      "loss": 1.9652,
      "step": 137572
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.0606096982955933,
      "learning_rate": 0.00020990059025546167,
      "loss": 2.1252,
      "step": 137573
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.0390877723693848,
      "learning_rate": 0.0002098966885427609,
      "loss": 2.1809,
      "step": 137574
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.3520139455795288,
      "learning_rate": 0.0002098927868468122,
      "loss": 2.0923,
      "step": 137575
    },
    {
      "epoch": 1.79,
      "grad_norm": 0.9988396763801575,
      "learning_rate": 0.00020988888516761597,
      "loss": 2.2906,
      "step": 137576
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.062426209449768,
      "learning_rate": 0.00020988498350517313,
      "loss": 2.2108,
      "step": 137577
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.1785608530044556,
      "learning_rate": 0.00020988108185948437,
      "loss": 2.0098,
      "step": 137578
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.2499250173568726,
      "learning_rate": 0.00020987718023055037,
      "loss": 2.2293,
      "step": 137579
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.0362024307250977,
      "learning_rate": 0.000209873278618372,
      "loss": 2.0791,
      "step": 137580
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.0035851001739502,
      "learning_rate": 0.00020986937702294995,
      "loss": 1.848,
      "step": 137581
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.0141469240188599,
      "learning_rate": 0.00020986547544428477,
      "loss": 2.221,
      "step": 137582
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.0004222393035889,
      "learning_rate": 0.00020986157388237726,
      "loss": 1.946,
      "step": 137583
    },
    {
      "epoch": 1.79,
      "grad_norm": 0.9459365010261536,
      "learning_rate": 0.00020985767233722822,
      "loss": 2.0713,
      "step": 137584
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.0012303590774536,
      "learning_rate": 0.00020985377080883836,
      "loss": 2.0141,
      "step": 137585
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.0644875764846802,
      "learning_rate": 0.00020984986929720832,
      "loss": 2.1154,
      "step": 137586
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.0151218175888062,
      "learning_rate": 0.0002098459678023391,
      "loss": 2.3338,
      "step": 137587
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.1503304243087769,
      "learning_rate": 0.00020984206632423103,
      "loss": 2.0822,
      "step": 137588
    },
    {
      "epoch": 1.79,
      "grad_norm": 0.9816102981567383,
      "learning_rate": 0.00020983816486288504,
      "loss": 2.1313,
      "step": 137589
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.0312753915786743,
      "learning_rate": 0.00020983426341830184,
      "loss": 2.1478,
      "step": 137590
    },
    {
      "epoch": 1.79,
      "grad_norm": 0.9991862773895264,
      "learning_rate": 0.00020983036199048217,
      "loss": 2.1315,
      "step": 137591
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.0074700117111206,
      "learning_rate": 0.00020982646057942668,
      "loss": 1.9937,
      "step": 137592
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.0225199460983276,
      "learning_rate": 0.00020982255918513624,
      "loss": 2.1185,
      "step": 137593
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.0374271869659424,
      "learning_rate": 0.00020981865780761157,
      "loss": 2.0187,
      "step": 137594
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.0916091203689575,
      "learning_rate": 0.00020981475644685316,
      "loss": 2.0213,
      "step": 137595
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.0374966859817505,
      "learning_rate": 0.00020981085510286192,
      "loss": 2.381,
      "step": 137596
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.1015896797180176,
      "learning_rate": 0.00020980695377563856,
      "loss": 2.1786,
      "step": 137597
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.0642335414886475,
      "learning_rate": 0.00020980305246518375,
      "loss": 1.9614,
      "step": 137598
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.0073792934417725,
      "learning_rate": 0.00020979915117149827,
      "loss": 2.1511,
      "step": 137599
    },
    {
      "epoch": 1.79,
      "grad_norm": 0.9841189384460449,
      "learning_rate": 0.00020979524989458292,
      "loss": 2.1783,
      "step": 137600
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.1322137117385864,
      "learning_rate": 0.00020979134863443832,
      "loss": 2.255,
      "step": 137601
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.0761898756027222,
      "learning_rate": 0.0002097874473910651,
      "loss": 2.2061,
      "step": 137602
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.019647240638733,
      "learning_rate": 0.00020978354616446417,
      "loss": 1.9659,
      "step": 137603
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.0647414922714233,
      "learning_rate": 0.00020977964495463614,
      "loss": 2.0523,
      "step": 137604
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.0646615028381348,
      "learning_rate": 0.00020977574376158177,
      "loss": 2.206,
      "step": 137605
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.031785011291504,
      "learning_rate": 0.00020977184258530186,
      "loss": 2.1929,
      "step": 137606
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.0231581926345825,
      "learning_rate": 0.00020976794142579702,
      "loss": 2.1346,
      "step": 137607
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.1283609867095947,
      "learning_rate": 0.0002097640402830681,
      "loss": 2.3141,
      "step": 137608
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.0594295263290405,
      "learning_rate": 0.0002097601391571157,
      "loss": 2.0262,
      "step": 137609
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.0645774602890015,
      "learning_rate": 0.00020975623804794053,
      "loss": 1.9759,
      "step": 137610
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.1205742359161377,
      "learning_rate": 0.00020975233695554342,
      "loss": 1.9293,
      "step": 137611
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.1345247030258179,
      "learning_rate": 0.00020974843587992512,
      "loss": 2.1034,
      "step": 137612
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.1313366889953613,
      "learning_rate": 0.0002097445348210862,
      "loss": 2.0514,
      "step": 137613
    },
    {
      "epoch": 1.79,
      "grad_norm": 0.9909804463386536,
      "learning_rate": 0.00020974063377902748,
      "loss": 2.1605,
      "step": 137614
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.0778141021728516,
      "learning_rate": 0.0002097367327537498,
      "loss": 2.0763,
      "step": 137615
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.0464205741882324,
      "learning_rate": 0.0002097328317452536,
      "loss": 2.0437,
      "step": 137616
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.0199341773986816,
      "learning_rate": 0.00020972893075353981,
      "loss": 1.8058,
      "step": 137617
    },
    {
      "epoch": 1.79,
      "grad_norm": 0.9520894289016724,
      "learning_rate": 0.0002097250297786092,
      "loss": 2.1721,
      "step": 137618
    },
    {
      "epoch": 1.79,
      "grad_norm": 0.9242110848426819,
      "learning_rate": 0.00020972112882046236,
      "loss": 1.9901,
      "step": 137619
    },
    {
      "epoch": 1.79,
      "grad_norm": 0.9484882950782776,
      "learning_rate": 0.00020971722787910003,
      "loss": 2.1555,
      "step": 137620
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.038581132888794,
      "learning_rate": 0.00020971332695452302,
      "loss": 2.1902,
      "step": 137621
    },
    {
      "epoch": 1.79,
      "grad_norm": 0.9968944191932678,
      "learning_rate": 0.00020970942604673203,
      "loss": 2.2675,
      "step": 137622
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.0970513820648193,
      "learning_rate": 0.00020970552515572771,
      "loss": 2.2768,
      "step": 137623
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.0228760242462158,
      "learning_rate": 0.0002097016242815109,
      "loss": 2.4398,
      "step": 137624
    },
    {
      "epoch": 1.79,
      "grad_norm": 0.9889030456542969,
      "learning_rate": 0.00020969772342408218,
      "loss": 2.016,
      "step": 137625
    },
    {
      "epoch": 1.79,
      "grad_norm": 0.9749189615249634,
      "learning_rate": 0.0002096938225834424,
      "loss": 2.0826,
      "step": 137626
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.086874008178711,
      "learning_rate": 0.00020968992175959217,
      "loss": 2.3502,
      "step": 137627
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.0406293869018555,
      "learning_rate": 0.00020968602095253235,
      "loss": 2.2777,
      "step": 137628
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.035356879234314,
      "learning_rate": 0.00020968212016226362,
      "loss": 2.2379,
      "step": 137629
    },
    {
      "epoch": 1.79,
      "grad_norm": 0.9959405064582825,
      "learning_rate": 0.0002096782193887868,
      "loss": 2.0974,
      "step": 137630
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.0215274095535278,
      "learning_rate": 0.00020967431863210234,
      "loss": 1.9102,
      "step": 137631
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.006593942642212,
      "learning_rate": 0.00020967041789221117,
      "loss": 2.1561,
      "step": 137632
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.0094389915466309,
      "learning_rate": 0.00020966651716911394,
      "loss": 2.0269,
      "step": 137633
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.0613689422607422,
      "learning_rate": 0.0002096626164628114,
      "loss": 2.0078,
      "step": 137634
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.0293773412704468,
      "learning_rate": 0.00020965871577330436,
      "loss": 2.2381,
      "step": 137635
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.0699723958969116,
      "learning_rate": 0.00020965481510059353,
      "loss": 1.9832,
      "step": 137636
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.2215895652770996,
      "learning_rate": 0.00020965091444467946,
      "loss": 2.0052,
      "step": 137637
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.0938527584075928,
      "learning_rate": 0.000209647013805563,
      "loss": 2.241,
      "step": 137638
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.0743073225021362,
      "learning_rate": 0.00020964311318324487,
      "loss": 2.0583,
      "step": 137639
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.0437098741531372,
      "learning_rate": 0.00020963921257772576,
      "loss": 2.0659,
      "step": 137640
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.0899453163146973,
      "learning_rate": 0.00020963531198900647,
      "loss": 1.9541,
      "step": 137641
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.0378912687301636,
      "learning_rate": 0.0002096314114170878,
      "loss": 2.2055,
      "step": 137642
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.1018413305282593,
      "learning_rate": 0.00020962751086197018,
      "loss": 2.0239,
      "step": 137643
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.1719070672988892,
      "learning_rate": 0.00020962361032365458,
      "loss": 1.9384,
      "step": 137644
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.1391669511795044,
      "learning_rate": 0.0002096197098021416,
      "loss": 2.2195,
      "step": 137645
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.0131889581680298,
      "learning_rate": 0.00020961580929743204,
      "loss": 2.1748,
      "step": 137646
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.0162984132766724,
      "learning_rate": 0.00020961190880952664,
      "loss": 2.2709,
      "step": 137647
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.0586825609207153,
      "learning_rate": 0.0002096080083384262,
      "loss": 2.0585,
      "step": 137648
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.1801714897155762,
      "learning_rate": 0.0002096041078841312,
      "loss": 1.967,
      "step": 137649
    },
    {
      "epoch": 1.79,
      "grad_norm": 0.9872052669525146,
      "learning_rate": 0.00020960020744664252,
      "loss": 1.9887,
      "step": 137650
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.058222770690918,
      "learning_rate": 0.00020959630702596085,
      "loss": 2.1559,
      "step": 137651
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.1326102018356323,
      "learning_rate": 0.00020959240662208692,
      "loss": 2.1724,
      "step": 137652
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.0933983325958252,
      "learning_rate": 0.0002095885062350215,
      "loss": 2.1576,
      "step": 137653
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.038037896156311,
      "learning_rate": 0.00020958460586476538,
      "loss": 2.1631,
      "step": 137654
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.0436630249023438,
      "learning_rate": 0.00020958070551131908,
      "loss": 2.1884,
      "step": 137655
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.2888363599777222,
      "learning_rate": 0.00020957680517468344,
      "loss": 2.1367,
      "step": 137656
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.0270851850509644,
      "learning_rate": 0.00020957290485485916,
      "loss": 1.978,
      "step": 137657
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.2390402555465698,
      "learning_rate": 0.00020956900455184702,
      "loss": 2.2027,
      "step": 137658
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.1361740827560425,
      "learning_rate": 0.00020956510426564767,
      "loss": 2.2519,
      "step": 137659
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.050843358039856,
      "learning_rate": 0.0002095612039962619,
      "loss": 1.9412,
      "step": 137660
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.1860065460205078,
      "learning_rate": 0.00020955730374369052,
      "loss": 2.2032,
      "step": 137661
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.0144689083099365,
      "learning_rate": 0.00020955340350793403,
      "loss": 2.0665,
      "step": 137662
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.0454288721084595,
      "learning_rate": 0.00020954950328899323,
      "loss": 2.4033,
      "step": 137663
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.0462353229522705,
      "learning_rate": 0.00020954560308686894,
      "loss": 2.2362,
      "step": 137664
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.0084764957427979,
      "learning_rate": 0.0002095417029015618,
      "loss": 2.0213,
      "step": 137665
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.0839529037475586,
      "learning_rate": 0.00020953780273307254,
      "loss": 2.1751,
      "step": 137666
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.1034291982650757,
      "learning_rate": 0.0002095339025814021,
      "loss": 2.0679,
      "step": 137667
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.039398431777954,
      "learning_rate": 0.00020953000244655084,
      "loss": 2.1426,
      "step": 137668
    },
    {
      "epoch": 1.79,
      "grad_norm": 0.9953233003616333,
      "learning_rate": 0.00020952610232851968,
      "loss": 1.9176,
      "step": 137669
    },
    {
      "epoch": 1.79,
      "grad_norm": 0.930328905582428,
      "learning_rate": 0.00020952220222730932,
      "loss": 1.9845,
      "step": 137670
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.0816746950149536,
      "learning_rate": 0.0002095183021429205,
      "loss": 2.2368,
      "step": 137671
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.0508556365966797,
      "learning_rate": 0.00020951440207535391,
      "loss": 2.0723,
      "step": 137672
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.0439358949661255,
      "learning_rate": 0.00020951050202461046,
      "loss": 1.9465,
      "step": 137673
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.110244631767273,
      "learning_rate": 0.00020950660199069057,
      "loss": 2.1901,
      "step": 137674
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.0653750896453857,
      "learning_rate": 0.0002095027019735951,
      "loss": 1.947,
      "step": 137675
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.056045651435852,
      "learning_rate": 0.00020949880197332482,
      "loss": 2.0519,
      "step": 137676
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.0272947549819946,
      "learning_rate": 0.0002094949019898804,
      "loss": 2.0316,
      "step": 137677
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.0472735166549683,
      "learning_rate": 0.0002094910020232626,
      "loss": 2.1074,
      "step": 137678
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.0548009872436523,
      "learning_rate": 0.00020948710207347224,
      "loss": 2.0664,
      "step": 137679
    },
    {
      "epoch": 1.79,
      "grad_norm": 0.9356639385223389,
      "learning_rate": 0.00020948320214050984,
      "loss": 2.1719,
      "step": 137680
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.057090163230896,
      "learning_rate": 0.00020947930222437619,
      "loss": 2.0458,
      "step": 137681
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.0041735172271729,
      "learning_rate": 0.00020947540232507206,
      "loss": 2.0155,
      "step": 137682
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.1273889541625977,
      "learning_rate": 0.00020947150244259816,
      "loss": 2.1057,
      "step": 137683
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.0199509859085083,
      "learning_rate": 0.0002094676025769552,
      "loss": 2.2076,
      "step": 137684
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.1371663808822632,
      "learning_rate": 0.00020946370272814405,
      "loss": 1.9795,
      "step": 137685
    },
    {
      "epoch": 1.79,
      "grad_norm": 0.978916585445404,
      "learning_rate": 0.00020945980289616516,
      "loss": 2.1825,
      "step": 137686
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.0305031538009644,
      "learning_rate": 0.00020945590308101946,
      "loss": 2.1058,
      "step": 137687
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.1898387670516968,
      "learning_rate": 0.0002094520032827076,
      "loss": 2.1851,
      "step": 137688
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.0714949369430542,
      "learning_rate": 0.0002094481035012303,
      "loss": 1.9777,
      "step": 137689
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.0873430967330933,
      "learning_rate": 0.00020944420373658832,
      "loss": 2.4148,
      "step": 137690
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.0607569217681885,
      "learning_rate": 0.00020944030398878244,
      "loss": 1.9497,
      "step": 137691
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.0773619413375854,
      "learning_rate": 0.00020943640425781322,
      "loss": 2.2142,
      "step": 137692
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.268520712852478,
      "learning_rate": 0.0002094325045436816,
      "loss": 2.1355,
      "step": 137693
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.10271418094635,
      "learning_rate": 0.0002094286048463881,
      "loss": 1.9155,
      "step": 137694
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.1627877950668335,
      "learning_rate": 0.0002094247051659335,
      "loss": 2.1598,
      "step": 137695
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.0576977729797363,
      "learning_rate": 0.0002094208055023186,
      "loss": 2.2382,
      "step": 137696
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.3882362842559814,
      "learning_rate": 0.00020941690585554411,
      "loss": 2.132,
      "step": 137697
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.3599412441253662,
      "learning_rate": 0.00020941300622561067,
      "loss": 2.2505,
      "step": 137698
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.05526864528656,
      "learning_rate": 0.0002094091066125191,
      "loss": 2.1414,
      "step": 137699
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.0610225200653076,
      "learning_rate": 0.00020940520701627014,
      "loss": 2.1981,
      "step": 137700
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.055874228477478,
      "learning_rate": 0.00020940130743686436,
      "loss": 2.1592,
      "step": 137701
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.1524696350097656,
      "learning_rate": 0.0002093974078743026,
      "loss": 2.0831,
      "step": 137702
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.0107650756835938,
      "learning_rate": 0.00020939350832858568,
      "loss": 1.989,
      "step": 137703
    },
    {
      "epoch": 1.79,
      "grad_norm": 0.9820727109909058,
      "learning_rate": 0.00020938960879971413,
      "loss": 1.9027,
      "step": 137704
    },
    {
      "epoch": 1.79,
      "grad_norm": 0.962254524230957,
      "learning_rate": 0.00020938570928768872,
      "loss": 2.1169,
      "step": 137705
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.0779105424880981,
      "learning_rate": 0.0002093818097925103,
      "loss": 2.0825,
      "step": 137706
    },
    {
      "epoch": 1.79,
      "grad_norm": 0.9291338324546814,
      "learning_rate": 0.00020937791031417952,
      "loss": 2.0347,
      "step": 137707
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.199582815170288,
      "learning_rate": 0.00020937401085269701,
      "loss": 2.072,
      "step": 137708
    },
    {
      "epoch": 1.79,
      "grad_norm": 0.9682714939117432,
      "learning_rate": 0.0002093701114080637,
      "loss": 2.0425,
      "step": 137709
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.0331288576126099,
      "learning_rate": 0.00020936621198028006,
      "loss": 2.0375,
      "step": 137710
    },
    {
      "epoch": 1.79,
      "grad_norm": 0.9679334759712219,
      "learning_rate": 0.00020936231256934702,
      "loss": 2.1484,
      "step": 137711
    },
    {
      "epoch": 1.79,
      "grad_norm": 0.9352083206176758,
      "learning_rate": 0.00020935841317526523,
      "loss": 2.3274,
      "step": 137712
    },
    {
      "epoch": 1.79,
      "grad_norm": 0.9954794645309448,
      "learning_rate": 0.0002093545137980354,
      "loss": 1.9818,
      "step": 137713
    },
    {
      "epoch": 1.79,
      "grad_norm": 0.9867565631866455,
      "learning_rate": 0.00020935061443765835,
      "loss": 2.2895,
      "step": 137714
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.1719672679901123,
      "learning_rate": 0.00020934671509413472,
      "loss": 2.0905,
      "step": 137715
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.03225839138031,
      "learning_rate": 0.00020934281576746525,
      "loss": 2.203,
      "step": 137716
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.0319350957870483,
      "learning_rate": 0.0002093389164576506,
      "loss": 2.1042,
      "step": 137717
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.0607495307922363,
      "learning_rate": 0.00020933501716469154,
      "loss": 2.0983,
      "step": 137718
    },
    {
      "epoch": 1.79,
      "grad_norm": 0.92808598279953,
      "learning_rate": 0.00020933111788858883,
      "loss": 2.0617,
      "step": 137719
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.0695163011550903,
      "learning_rate": 0.00020932721862934318,
      "loss": 2.1099,
      "step": 137720
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.1016428470611572,
      "learning_rate": 0.00020932331938695544,
      "loss": 2.1123,
      "step": 137721
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.1030606031417847,
      "learning_rate": 0.0002093194201614261,
      "loss": 1.9481,
      "step": 137722
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.0703184604644775,
      "learning_rate": 0.00020931552095275593,
      "loss": 2.3328,
      "step": 137723
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.1594899892807007,
      "learning_rate": 0.0002093116217609458,
      "loss": 2.1613,
      "step": 137724
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.127178430557251,
      "learning_rate": 0.0002093077225859963,
      "loss": 2.1196,
      "step": 137725
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.0945515632629395,
      "learning_rate": 0.00020930382342790823,
      "loss": 2.3794,
      "step": 137726
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.0919493436813354,
      "learning_rate": 0.00020929992428668227,
      "loss": 2.1219,
      "step": 137727
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.053719401359558,
      "learning_rate": 0.00020929602516231931,
      "loss": 2.1519,
      "step": 137728
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.149551272392273,
      "learning_rate": 0.00020929212605481977,
      "loss": 2.3181,
      "step": 137729
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.1744141578674316,
      "learning_rate": 0.00020928822696418455,
      "loss": 2.1594,
      "step": 137730
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.0513155460357666,
      "learning_rate": 0.0002092843278904144,
      "loss": 2.1511,
      "step": 137731
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.1215616464614868,
      "learning_rate": 0.00020928042883350994,
      "loss": 2.2327,
      "step": 137732
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.088337779045105,
      "learning_rate": 0.00020927652979347205,
      "loss": 1.9573,
      "step": 137733
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.0997810363769531,
      "learning_rate": 0.00020927263077030143,
      "loss": 2.1507,
      "step": 137734
    },
    {
      "epoch": 1.79,
      "grad_norm": 0.930957019329071,
      "learning_rate": 0.00020926873176399863,
      "loss": 2.0882,
      "step": 137735
    },
    {
      "epoch": 1.79,
      "grad_norm": 0.9728361368179321,
      "learning_rate": 0.00020926483277456444,
      "loss": 2.0025,
      "step": 137736
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.0150845050811768,
      "learning_rate": 0.0002092609338019997,
      "loss": 2.1251,
      "step": 137737
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.0047461986541748,
      "learning_rate": 0.00020925703484630503,
      "loss": 1.9862,
      "step": 137738
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.0241087675094604,
      "learning_rate": 0.00020925313590748122,
      "loss": 2.132,
      "step": 137739
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.3270164728164673,
      "learning_rate": 0.00020924923698552905,
      "loss": 2.0184,
      "step": 137740
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.1295639276504517,
      "learning_rate": 0.00020924533808044905,
      "loss": 2.1473,
      "step": 137741
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.096979022026062,
      "learning_rate": 0.00020924143919224202,
      "loss": 2.0576,
      "step": 137742
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.0795592069625854,
      "learning_rate": 0.00020923754032090879,
      "loss": 2.3511,
      "step": 137743
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.1361802816390991,
      "learning_rate": 0.00020923364146644997,
      "loss": 2.1728,
      "step": 137744
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.0697851181030273,
      "learning_rate": 0.00020922974262886635,
      "loss": 2.0726,
      "step": 137745
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.0621546506881714,
      "learning_rate": 0.00020922584380815877,
      "loss": 2.0212,
      "step": 137746
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.1519641876220703,
      "learning_rate": 0.00020922194500432766,
      "loss": 2.2073,
      "step": 137747
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.0578632354736328,
      "learning_rate": 0.0002092180462173739,
      "loss": 2.2894,
      "step": 137748
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.165834665298462,
      "learning_rate": 0.00020921414744729826,
      "loss": 2.2562,
      "step": 137749
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.1152153015136719,
      "learning_rate": 0.00020921024869410138,
      "loss": 1.9466,
      "step": 137750
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.4861665964126587,
      "learning_rate": 0.00020920634995778405,
      "loss": 2.1325,
      "step": 137751
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.1999799013137817,
      "learning_rate": 0.00020920245123834708,
      "loss": 2.016,
      "step": 137752
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.0673071146011353,
      "learning_rate": 0.000209198552535791,
      "loss": 2.0832,
      "step": 137753
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.0698044300079346,
      "learning_rate": 0.0002091946538501166,
      "loss": 2.2332,
      "step": 137754
    },
    {
      "epoch": 1.79,
      "grad_norm": 0.9996638298034668,
      "learning_rate": 0.00020919075518132464,
      "loss": 1.9797,
      "step": 137755
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.1376363039016724,
      "learning_rate": 0.00020918685652941582,
      "loss": 1.973,
      "step": 137756
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.0399889945983887,
      "learning_rate": 0.00020918295789439087,
      "loss": 2.1742,
      "step": 137757
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.0931060314178467,
      "learning_rate": 0.00020917905927625067,
      "loss": 2.0876,
      "step": 137758
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.1641722917556763,
      "learning_rate": 0.00020917516067499566,
      "loss": 2.1768,
      "step": 137759
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.1200319528579712,
      "learning_rate": 0.0002091712620906267,
      "loss": 2.0072,
      "step": 137760
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.1708251237869263,
      "learning_rate": 0.0002091673635231445,
      "loss": 2.2044,
      "step": 137761
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.112326979637146,
      "learning_rate": 0.0002091634649725498,
      "loss": 2.0242,
      "step": 137762
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.138279914855957,
      "learning_rate": 0.0002091595664388434,
      "loss": 1.9797,
      "step": 137763
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.0709832906723022,
      "learning_rate": 0.00020915566792202602,
      "loss": 2.1592,
      "step": 137764
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.2209854125976562,
      "learning_rate": 0.00020915176942209817,
      "loss": 2.1484,
      "step": 137765
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.0579441785812378,
      "learning_rate": 0.00020914787093906077,
      "loss": 2.3124,
      "step": 137766
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.0745793581008911,
      "learning_rate": 0.0002091439724729145,
      "loss": 2.1957,
      "step": 137767
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.052694320678711,
      "learning_rate": 0.00020914007402366,
      "loss": 2.0028,
      "step": 137768
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.0761713981628418,
      "learning_rate": 0.00020913617559129813,
      "loss": 1.971,
      "step": 137769
    },
    {
      "epoch": 1.79,
      "grad_norm": 0.9753051400184631,
      "learning_rate": 0.0002091322771758297,
      "loss": 2.0158,
      "step": 137770
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.405078411102295,
      "learning_rate": 0.00020912837877725513,
      "loss": 1.9297,
      "step": 137771
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.1740186214447021,
      "learning_rate": 0.00020912448039557536,
      "loss": 1.9865,
      "step": 137772
    },
    {
      "epoch": 1.79,
      "grad_norm": 0.9736607074737549,
      "learning_rate": 0.00020912058203079103,
      "loss": 1.9549,
      "step": 137773
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.0511503219604492,
      "learning_rate": 0.0002091166836829029,
      "loss": 2.1051,
      "step": 137774
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.0557643175125122,
      "learning_rate": 0.00020911278535191172,
      "loss": 2.04,
      "step": 137775
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.2809028625488281,
      "learning_rate": 0.00020910888703781823,
      "loss": 1.9306,
      "step": 137776
    },
    {
      "epoch": 1.79,
      "grad_norm": 0.9887426495552063,
      "learning_rate": 0.0002091049887406231,
      "loss": 1.8299,
      "step": 137777
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.1068382263183594,
      "learning_rate": 0.00020910109046032703,
      "loss": 2.1808,
      "step": 137778
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.0401923656463623,
      "learning_rate": 0.00020909719219693075,
      "loss": 2.0934,
      "step": 137779
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.2812328338623047,
      "learning_rate": 0.00020909329395043508,
      "loss": 2.0916,
      "step": 137780
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.0629565715789795,
      "learning_rate": 0.00020908939572084062,
      "loss": 2.0271,
      "step": 137781
    },
    {
      "epoch": 1.79,
      "grad_norm": 0.9499889016151428,
      "learning_rate": 0.00020908549750814825,
      "loss": 2.0091,
      "step": 137782
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.0733532905578613,
      "learning_rate": 0.00020908159931235854,
      "loss": 2.1275,
      "step": 137783
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.2251410484313965,
      "learning_rate": 0.00020907770113347233,
      "loss": 1.9588,
      "step": 137784
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.296606183052063,
      "learning_rate": 0.00020907380297149027,
      "loss": 1.8307,
      "step": 137785
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.0146262645721436,
      "learning_rate": 0.00020906990482641303,
      "loss": 2.0434,
      "step": 137786
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.0344090461730957,
      "learning_rate": 0.00020906600669824148,
      "loss": 1.9354,
      "step": 137787
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.191456913948059,
      "learning_rate": 0.00020906210858697628,
      "loss": 2.171,
      "step": 137788
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.0972132682800293,
      "learning_rate": 0.00020905821049261813,
      "loss": 2.0224,
      "step": 137789
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.2044341564178467,
      "learning_rate": 0.00020905431241516776,
      "loss": 2.0246,
      "step": 137790
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.1384084224700928,
      "learning_rate": 0.00020905041435462587,
      "loss": 2.1897,
      "step": 137791
    },
    {
      "epoch": 1.79,
      "grad_norm": 0.9940778017044067,
      "learning_rate": 0.00020904651631099337,
      "loss": 2.2036,
      "step": 137792
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.0336921215057373,
      "learning_rate": 0.00020904261828427073,
      "loss": 2.161,
      "step": 137793
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.066940426826477,
      "learning_rate": 0.0002090387202744588,
      "loss": 2.3947,
      "step": 137794
    },
    {
      "epoch": 1.79,
      "grad_norm": 0.9824694991111755,
      "learning_rate": 0.00020903482228155833,
      "loss": 2.0594,
      "step": 137795
    },
    {
      "epoch": 1.79,
      "grad_norm": 0.9915987253189087,
      "learning_rate": 0.00020903092430556996,
      "loss": 1.9344,
      "step": 137796
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.039939045906067,
      "learning_rate": 0.00020902702634649443,
      "loss": 2.022,
      "step": 137797
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.0160000324249268,
      "learning_rate": 0.00020902312840433252,
      "loss": 2.2388,
      "step": 137798
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.0229370594024658,
      "learning_rate": 0.00020901923047908502,
      "loss": 2.2492,
      "step": 137799
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.0817961692810059,
      "learning_rate": 0.00020901533257075243,
      "loss": 2.1984,
      "step": 137800
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.1415389776229858,
      "learning_rate": 0.00020901143467933573,
      "loss": 2.1935,
      "step": 137801
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.017952561378479,
      "learning_rate": 0.00020900753680483543,
      "loss": 2.1469,
      "step": 137802
    },
    {
      "epoch": 1.79,
      "grad_norm": 0.9899069666862488,
      "learning_rate": 0.00020900363894725234,
      "loss": 2.1591,
      "step": 137803
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.1061934232711792,
      "learning_rate": 0.0002089997411065872,
      "loss": 2.1197,
      "step": 137804
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.1960375308990479,
      "learning_rate": 0.00020899584328284073,
      "loss": 2.1953,
      "step": 137805
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.4112834930419922,
      "learning_rate": 0.0002089919454760137,
      "loss": 2.0687,
      "step": 137806
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.0614787340164185,
      "learning_rate": 0.00020898804768610678,
      "loss": 2.0349,
      "step": 137807
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.0624914169311523,
      "learning_rate": 0.00020898414991312068,
      "loss": 2.0729,
      "step": 137808
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.0518372058868408,
      "learning_rate": 0.0002089802521570561,
      "loss": 2.2431,
      "step": 137809
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.0026639699935913,
      "learning_rate": 0.00020897635441791387,
      "loss": 2.2519,
      "step": 137810
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.4773008823394775,
      "learning_rate": 0.00020897245669569459,
      "loss": 2.043,
      "step": 137811
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.0380762815475464,
      "learning_rate": 0.00020896855899039908,
      "loss": 1.9881,
      "step": 137812
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.122410535812378,
      "learning_rate": 0.00020896466130202815,
      "loss": 2.1096,
      "step": 137813
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.0845521688461304,
      "learning_rate": 0.00020896076363058228,
      "loss": 2.0215,
      "step": 137814
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.1382482051849365,
      "learning_rate": 0.0002089568659760623,
      "loss": 2.0158,
      "step": 137815
    },
    {
      "epoch": 1.79,
      "grad_norm": 0.9644845724105835,
      "learning_rate": 0.000208952968338469,
      "loss": 1.9788,
      "step": 137816
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.1051454544067383,
      "learning_rate": 0.00020894907071780303,
      "loss": 2.2661,
      "step": 137817
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.0230382680892944,
      "learning_rate": 0.00020894517311406517,
      "loss": 2.2107,
      "step": 137818
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.2288486957550049,
      "learning_rate": 0.00020894127552725623,
      "loss": 2.1595,
      "step": 137819
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.000750184059143,
      "learning_rate": 0.00020893737795737674,
      "loss": 2.0957,
      "step": 137820
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.1372387409210205,
      "learning_rate": 0.00020893348040442743,
      "loss": 1.984,
      "step": 137821
    },
    {
      "epoch": 1.79,
      "grad_norm": 0.9945563673973083,
      "learning_rate": 0.00020892958286840914,
      "loss": 2.0017,
      "step": 137822
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.288626790046692,
      "learning_rate": 0.00020892568534932262,
      "loss": 2.1563,
      "step": 137823
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.0226354598999023,
      "learning_rate": 0.00020892178784716846,
      "loss": 2.2448,
      "step": 137824
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.0447263717651367,
      "learning_rate": 0.00020891789036194764,
      "loss": 2.2218,
      "step": 137825
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.06210196018219,
      "learning_rate": 0.00020891399289366054,
      "loss": 1.9613,
      "step": 137826
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.034164309501648,
      "learning_rate": 0.00020891009544230803,
      "loss": 1.9374,
      "step": 137827
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.1496431827545166,
      "learning_rate": 0.00020890619800789087,
      "loss": 2.1019,
      "step": 137828
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.1352342367172241,
      "learning_rate": 0.0002089023005904098,
      "loss": 2.0833,
      "step": 137829
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.0714802742004395,
      "learning_rate": 0.0002088984031898655,
      "loss": 2.0589,
      "step": 137830
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.0545077323913574,
      "learning_rate": 0.0002088945058062588,
      "loss": 2.2154,
      "step": 137831
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.0556739568710327,
      "learning_rate": 0.00020889060843959023,
      "loss": 2.1401,
      "step": 137832
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.053009271621704,
      "learning_rate": 0.00020888671108986058,
      "loss": 1.976,
      "step": 137833
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.0853748321533203,
      "learning_rate": 0.00020888281375707068,
      "loss": 1.895,
      "step": 137834
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.0481206178665161,
      "learning_rate": 0.00020887891644122113,
      "loss": 2.3722,
      "step": 137835
    },
    {
      "epoch": 1.79,
      "grad_norm": 0.9818708896636963,
      "learning_rate": 0.00020887501914231273,
      "loss": 2.245,
      "step": 137836
    },
    {
      "epoch": 1.79,
      "grad_norm": 0.9459220170974731,
      "learning_rate": 0.00020887112186034632,
      "loss": 2.0373,
      "step": 137837
    },
    {
      "epoch": 1.79,
      "grad_norm": 0.9821035265922546,
      "learning_rate": 0.00020886722459532235,
      "loss": 2.0055,
      "step": 137838
    },
    {
      "epoch": 1.79,
      "grad_norm": 0.99418044090271,
      "learning_rate": 0.00020886332734724165,
      "loss": 1.9515,
      "step": 137839
    },
    {
      "epoch": 1.79,
      "grad_norm": 0.9808014631271362,
      "learning_rate": 0.00020885943011610502,
      "loss": 2.123,
      "step": 137840
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.07411789894104,
      "learning_rate": 0.00020885553290191311,
      "loss": 2.0603,
      "step": 137841
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.0637034177780151,
      "learning_rate": 0.00020885163570466675,
      "loss": 2.1339,
      "step": 137842
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.0308177471160889,
      "learning_rate": 0.00020884773852436665,
      "loss": 1.9677,
      "step": 137843
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.0352431535720825,
      "learning_rate": 0.00020884384136101336,
      "loss": 2.0498,
      "step": 137844
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.154638648033142,
      "learning_rate": 0.00020883994421460773,
      "loss": 2.1246,
      "step": 137845
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.0793577432632446,
      "learning_rate": 0.00020883604708515045,
      "loss": 2.2663,
      "step": 137846
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.0212041139602661,
      "learning_rate": 0.00020883214997264232,
      "loss": 2.2673,
      "step": 137847
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.1654114723205566,
      "learning_rate": 0.00020882825287708395,
      "loss": 1.8825,
      "step": 137848
    },
    {
      "epoch": 1.79,
      "grad_norm": 0.9790292382240295,
      "learning_rate": 0.0002088243557984763,
      "loss": 2.3686,
      "step": 137849
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.2722365856170654,
      "learning_rate": 0.0002088204587368198,
      "loss": 2.0462,
      "step": 137850
    },
    {
      "epoch": 1.79,
      "grad_norm": 0.9966362714767456,
      "learning_rate": 0.00020881656169211528,
      "loss": 1.9353,
      "step": 137851
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.074784517288208,
      "learning_rate": 0.00020881266466436348,
      "loss": 2.2386,
      "step": 137852
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.0539681911468506,
      "learning_rate": 0.00020880876765356514,
      "loss": 2.315,
      "step": 137853
    },
    {
      "epoch": 1.79,
      "grad_norm": 0.9898139834403992,
      "learning_rate": 0.00020880487065972098,
      "loss": 1.9361,
      "step": 137854
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.1396194696426392,
      "learning_rate": 0.0002088009736828317,
      "loss": 1.8899,
      "step": 137855
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.2613857984542847,
      "learning_rate": 0.00020879707672289818,
      "loss": 2.1308,
      "step": 137856
    },
    {
      "epoch": 1.79,
      "grad_norm": 1.1866546869277954,
      "learning_rate": 0.00020879317977992088,
      "loss": 1.9392,
      "step": 137857
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.1566003561019897,
      "learning_rate": 0.0002087892828539006,
      "loss": 2.0971,
      "step": 137858
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.1171668767929077,
      "learning_rate": 0.00020878538594483817,
      "loss": 2.1459,
      "step": 137859
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.0746709108352661,
      "learning_rate": 0.00020878148905273421,
      "loss": 2.1109,
      "step": 137860
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.0325334072113037,
      "learning_rate": 0.00020877759217758954,
      "loss": 2.0187,
      "step": 137861
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.144605278968811,
      "learning_rate": 0.000208773695319405,
      "loss": 1.9085,
      "step": 137862
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.056998372077942,
      "learning_rate": 0.00020876979847818093,
      "loss": 1.9325,
      "step": 137863
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.055342674255371,
      "learning_rate": 0.00020876590165391833,
      "loss": 2.2412,
      "step": 137864
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.1134639978408813,
      "learning_rate": 0.00020876200484661782,
      "loss": 1.9332,
      "step": 137865
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.1359658241271973,
      "learning_rate": 0.00020875810805628026,
      "loss": 2.215,
      "step": 137866
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.1703754663467407,
      "learning_rate": 0.00020875421128290622,
      "loss": 2.1444,
      "step": 137867
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.1707795858383179,
      "learning_rate": 0.0002087503145264966,
      "loss": 2.2499,
      "step": 137868
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.0417028665542603,
      "learning_rate": 0.000208746417787052,
      "loss": 2.0231,
      "step": 137869
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.0006248950958252,
      "learning_rate": 0.00020874252106457307,
      "loss": 1.9779,
      "step": 137870
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.1398712396621704,
      "learning_rate": 0.0002087386243590607,
      "loss": 2.0372,
      "step": 137871
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.1921324729919434,
      "learning_rate": 0.0002087347276705155,
      "loss": 2.2193,
      "step": 137872
    },
    {
      "epoch": 1.8,
      "grad_norm": 0.9487555027008057,
      "learning_rate": 0.00020873083099893824,
      "loss": 2.1781,
      "step": 137873
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.0215046405792236,
      "learning_rate": 0.00020872693434432972,
      "loss": 2.0394,
      "step": 137874
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.1602283716201782,
      "learning_rate": 0.00020872303770669049,
      "loss": 1.9006,
      "step": 137875
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.0776387453079224,
      "learning_rate": 0.00020871914108602142,
      "loss": 2.0948,
      "step": 137876
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.0318490266799927,
      "learning_rate": 0.0002087152444823232,
      "loss": 1.9696,
      "step": 137877
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.0682153701782227,
      "learning_rate": 0.00020871134789559646,
      "loss": 2.1278,
      "step": 137878
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.1136174201965332,
      "learning_rate": 0.00020870745132584205,
      "loss": 2.0771,
      "step": 137879
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.0097821950912476,
      "learning_rate": 0.00020870355477306069,
      "loss": 1.9537,
      "step": 137880
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.2160775661468506,
      "learning_rate": 0.00020869965823725302,
      "loss": 2.0127,
      "step": 137881
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.1266124248504639,
      "learning_rate": 0.0002086957617184198,
      "loss": 2.0733,
      "step": 137882
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.0876822471618652,
      "learning_rate": 0.00020869186521656172,
      "loss": 2.2564,
      "step": 137883
    },
    {
      "epoch": 1.8,
      "grad_norm": 0.9556868076324463,
      "learning_rate": 0.00020868796873167966,
      "loss": 1.9079,
      "step": 137884
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.2482666969299316,
      "learning_rate": 0.00020868407226377417,
      "loss": 2.1119,
      "step": 137885
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.0425658226013184,
      "learning_rate": 0.00020868017581284607,
      "loss": 2.1334,
      "step": 137886
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.021872878074646,
      "learning_rate": 0.00020867627937889597,
      "loss": 2.0676,
      "step": 137887
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.0569307804107666,
      "learning_rate": 0.0002086723829619247,
      "loss": 2.0649,
      "step": 137888
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.0387375354766846,
      "learning_rate": 0.00020866848656193298,
      "loss": 2.2213,
      "step": 137889
    },
    {
      "epoch": 1.8,
      "grad_norm": 0.9600793123245239,
      "learning_rate": 0.00020866459017892147,
      "loss": 2.0793,
      "step": 137890
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.0420924425125122,
      "learning_rate": 0.00020866069381289103,
      "loss": 2.2898,
      "step": 137891
    },
    {
      "epoch": 1.8,
      "grad_norm": 0.9701268076896667,
      "learning_rate": 0.00020865679746384224,
      "loss": 2.1565,
      "step": 137892
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.082134485244751,
      "learning_rate": 0.00020865290113177587,
      "loss": 1.9361,
      "step": 137893
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.0390793085098267,
      "learning_rate": 0.00020864900481669263,
      "loss": 2.1411,
      "step": 137894
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.0893752574920654,
      "learning_rate": 0.00020864510851859326,
      "loss": 2.1319,
      "step": 137895
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.083480715751648,
      "learning_rate": 0.0002086412122374785,
      "loss": 2.1737,
      "step": 137896
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.0236732959747314,
      "learning_rate": 0.00020863731597334906,
      "loss": 2.1556,
      "step": 137897
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.052964687347412,
      "learning_rate": 0.0002086334197262058,
      "loss": 2.1493,
      "step": 137898
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.0299930572509766,
      "learning_rate": 0.0002086295234960492,
      "loss": 2.2113,
      "step": 137899
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.1415380239486694,
      "learning_rate": 0.00020862562728288006,
      "loss": 2.2365,
      "step": 137900
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.0614111423492432,
      "learning_rate": 0.00020862173108669916,
      "loss": 1.913,
      "step": 137901
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.1490048170089722,
      "learning_rate": 0.0002086178349075072,
      "loss": 2.1921,
      "step": 137902
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.1573349237442017,
      "learning_rate": 0.0002086139387453049,
      "loss": 1.9252,
      "step": 137903
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.0330365896224976,
      "learning_rate": 0.0002086100426000931,
      "loss": 2.0676,
      "step": 137904
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.0759977102279663,
      "learning_rate": 0.00020860614647187232,
      "loss": 2.0501,
      "step": 137905
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.1830929517745972,
      "learning_rate": 0.0002086022503606434,
      "loss": 2.2299,
      "step": 137906
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.2486499547958374,
      "learning_rate": 0.00020859835426640706,
      "loss": 2.1265,
      "step": 137907
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.09920072555542,
      "learning_rate": 0.00020859445818916395,
      "loss": 2.0417,
      "step": 137908
    },
    {
      "epoch": 1.8,
      "grad_norm": 0.9845671653747559,
      "learning_rate": 0.0002085905621289149,
      "loss": 2.1487,
      "step": 137909
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.0602998733520508,
      "learning_rate": 0.00020858666608566072,
      "loss": 2.168,
      "step": 137910
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.0328062772750854,
      "learning_rate": 0.00020858277005940184,
      "loss": 2.0959,
      "step": 137911
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.1883260011672974,
      "learning_rate": 0.00020857887405013919,
      "loss": 2.1459,
      "step": 137912
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.221045970916748,
      "learning_rate": 0.00020857497805787343,
      "loss": 1.9464,
      "step": 137913
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.0186651945114136,
      "learning_rate": 0.0002085710820826053,
      "loss": 1.9265,
      "step": 137914
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.1131516695022583,
      "learning_rate": 0.00020856718612433556,
      "loss": 1.7994,
      "step": 137915
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.1086435317993164,
      "learning_rate": 0.00020856329018306498,
      "loss": 2.1857,
      "step": 137916
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.0544878244400024,
      "learning_rate": 0.00020855939425879413,
      "loss": 2.0939,
      "step": 137917
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.1056127548217773,
      "learning_rate": 0.0002085554983515238,
      "loss": 2.2137,
      "step": 137918
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.0209852457046509,
      "learning_rate": 0.00020855160246125476,
      "loss": 1.9511,
      "step": 137919
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.0865304470062256,
      "learning_rate": 0.00020854770658798767,
      "loss": 1.9924,
      "step": 137920
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.0571271181106567,
      "learning_rate": 0.00020854381073172332,
      "loss": 1.9709,
      "step": 137921
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.0576304197311401,
      "learning_rate": 0.00020853991489246234,
      "loss": 1.8652,
      "step": 137922
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.025542140007019,
      "learning_rate": 0.00020853601907020567,
      "loss": 1.9104,
      "step": 137923
    },
    {
      "epoch": 1.8,
      "grad_norm": 0.952907145023346,
      "learning_rate": 0.0002085321232649538,
      "loss": 2.1187,
      "step": 137924
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.4032483100891113,
      "learning_rate": 0.00020852822747670745,
      "loss": 1.97,
      "step": 137925
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.1247785091400146,
      "learning_rate": 0.0002085243317054675,
      "loss": 1.9363,
      "step": 137926
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.0994632244110107,
      "learning_rate": 0.00020852043595123457,
      "loss": 2.0469,
      "step": 137927
    },
    {
      "epoch": 1.8,
      "grad_norm": 0.9604995846748352,
      "learning_rate": 0.00020851654021400944,
      "loss": 2.2328,
      "step": 137928
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.1826344728469849,
      "learning_rate": 0.00020851264449379291,
      "loss": 1.9931,
      "step": 137929
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.1408660411834717,
      "learning_rate": 0.00020850874879058552,
      "loss": 1.8751,
      "step": 137930
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.0839893817901611,
      "learning_rate": 0.00020850485310438803,
      "loss": 2.1436,
      "step": 137931
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.0767064094543457,
      "learning_rate": 0.00020850095743520124,
      "loss": 2.0179,
      "step": 137932
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.0036664009094238,
      "learning_rate": 0.00020849706178302589,
      "loss": 1.9992,
      "step": 137933
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.1321642398834229,
      "learning_rate": 0.0002084931661478626,
      "loss": 2.0729,
      "step": 137934
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.0295698642730713,
      "learning_rate": 0.00020848927052971228,
      "loss": 1.954,
      "step": 137935
    },
    {
      "epoch": 1.8,
      "grad_norm": 0.9954145550727844,
      "learning_rate": 0.00020848537492857544,
      "loss": 2.0446,
      "step": 137936
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.0444931983947754,
      "learning_rate": 0.0002084814793444529,
      "loss": 2.0869,
      "step": 137937
    },
    {
      "epoch": 1.8,
      "grad_norm": 0.948716402053833,
      "learning_rate": 0.00020847758377734538,
      "loss": 1.776,
      "step": 137938
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.0693788528442383,
      "learning_rate": 0.00020847368822725362,
      "loss": 2.1105,
      "step": 137939
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.1523679494857788,
      "learning_rate": 0.00020846979269417828,
      "loss": 2.1394,
      "step": 137940
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.138426423072815,
      "learning_rate": 0.0002084658971781203,
      "loss": 2.1855,
      "step": 137941
    },
    {
      "epoch": 1.8,
      "grad_norm": 0.984142005443573,
      "learning_rate": 0.0002084620016790801,
      "loss": 2.0133,
      "step": 137942
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.0617293119430542,
      "learning_rate": 0.00020845810619705854,
      "loss": 2.1695,
      "step": 137943
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.139204740524292,
      "learning_rate": 0.0002084542107320563,
      "loss": 1.954,
      "step": 137944
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.0724155902862549,
      "learning_rate": 0.0002084503152840742,
      "loss": 2.0775,
      "step": 137945
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.0031371116638184,
      "learning_rate": 0.00020844641985311293,
      "loss": 2.0474,
      "step": 137946
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.1761161088943481,
      "learning_rate": 0.0002084425244391733,
      "loss": 2.3265,
      "step": 137947
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.0546410083770752,
      "learning_rate": 0.0002084386290422558,
      "loss": 2.0241,
      "step": 137948
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.0668085813522339,
      "learning_rate": 0.0002084347336623613,
      "loss": 2.0215,
      "step": 137949
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.0018233060836792,
      "learning_rate": 0.0002084308382994905,
      "loss": 2.1458,
      "step": 137950
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.007888913154602,
      "learning_rate": 0.00020842694295364415,
      "loss": 2.0596,
      "step": 137951
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.094446063041687,
      "learning_rate": 0.00020842304762482298,
      "loss": 2.192,
      "step": 137952
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.3980921506881714,
      "learning_rate": 0.00020841915231302772,
      "loss": 2.0001,
      "step": 137953
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.1585719585418701,
      "learning_rate": 0.00020841525701825904,
      "loss": 2.0351,
      "step": 137954
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.1095064878463745,
      "learning_rate": 0.0002084113617405177,
      "loss": 1.9439,
      "step": 137955
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.175501823425293,
      "learning_rate": 0.00020840746647980438,
      "loss": 2.0912,
      "step": 137956
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.076143503189087,
      "learning_rate": 0.00020840357123611984,
      "loss": 2.0806,
      "step": 137957
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.1995282173156738,
      "learning_rate": 0.0002083996760094648,
      "loss": 2.135,
      "step": 137958
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.0204434394836426,
      "learning_rate": 0.0002083957807998401,
      "loss": 1.8834,
      "step": 137959
    },
    {
      "epoch": 1.8,
      "grad_norm": 0.9808467030525208,
      "learning_rate": 0.00020839188560724622,
      "loss": 2.3362,
      "step": 137960
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.1315258741378784,
      "learning_rate": 0.00020838799043168413,
      "loss": 2.2395,
      "step": 137961
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.1574890613555908,
      "learning_rate": 0.00020838409527315437,
      "loss": 2.0474,
      "step": 137962
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.09784996509552,
      "learning_rate": 0.00020838020013165773,
      "loss": 2.0092,
      "step": 137963
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.073852777481079,
      "learning_rate": 0.00020837630500719495,
      "loss": 2.0954,
      "step": 137964
    },
    {
      "epoch": 1.8,
      "grad_norm": 0.9877025485038757,
      "learning_rate": 0.0002083724098997668,
      "loss": 2.0525,
      "step": 137965
    },
    {
      "epoch": 1.8,
      "grad_norm": 0.997553825378418,
      "learning_rate": 0.00020836851480937388,
      "loss": 2.0646,
      "step": 137966
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.0428061485290527,
      "learning_rate": 0.00020836461973601698,
      "loss": 2.2112,
      "step": 137967
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.0832651853561401,
      "learning_rate": 0.0002083607246796969,
      "loss": 2.0712,
      "step": 137968
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.1085184812545776,
      "learning_rate": 0.0002083568296404142,
      "loss": 2.0303,
      "step": 137969
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.0324537754058838,
      "learning_rate": 0.00020835293461816975,
      "loss": 2.0455,
      "step": 137970
    },
    {
      "epoch": 1.8,
      "grad_norm": 0.9999242424964905,
      "learning_rate": 0.00020834903961296426,
      "loss": 2.1943,
      "step": 137971
    },
    {
      "epoch": 1.8,
      "grad_norm": 0.9919460415840149,
      "learning_rate": 0.00020834514462479833,
      "loss": 1.9158,
      "step": 137972
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.02796471118927,
      "learning_rate": 0.00020834124965367275,
      "loss": 2.2385,
      "step": 137973
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.1876822710037231,
      "learning_rate": 0.0002083373546995883,
      "loss": 2.0291,
      "step": 137974
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.1566020250320435,
      "learning_rate": 0.00020833345976254566,
      "loss": 2.2065,
      "step": 137975
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.0871163606643677,
      "learning_rate": 0.00020832956484254564,
      "loss": 1.9348,
      "step": 137976
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.1904659271240234,
      "learning_rate": 0.00020832566993958887,
      "loss": 2.1514,
      "step": 137977
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.0298854112625122,
      "learning_rate": 0.000208321775053676,
      "loss": 2.2249,
      "step": 137978
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.0789995193481445,
      "learning_rate": 0.0002083178801848079,
      "loss": 2.2614,
      "step": 137979
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.0923303365707397,
      "learning_rate": 0.0002083139853329852,
      "loss": 2.082,
      "step": 137980
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.0103061199188232,
      "learning_rate": 0.00020831009049820866,
      "loss": 2.0476,
      "step": 137981
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.2461644411087036,
      "learning_rate": 0.000208306195680479,
      "loss": 2.1599,
      "step": 137982
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.2294483184814453,
      "learning_rate": 0.00020830230087979707,
      "loss": 2.043,
      "step": 137983
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.040505051612854,
      "learning_rate": 0.00020829840609616336,
      "loss": 2.0526,
      "step": 137984
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.0267637968063354,
      "learning_rate": 0.00020829451132957873,
      "loss": 2.1048,
      "step": 137985
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.031461477279663,
      "learning_rate": 0.00020829061658004386,
      "loss": 2.1731,
      "step": 137986
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.0902330875396729,
      "learning_rate": 0.00020828672184755945,
      "loss": 2.0252,
      "step": 137987
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.1133315563201904,
      "learning_rate": 0.00020828282713212634,
      "loss": 2.0324,
      "step": 137988
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.0568033456802368,
      "learning_rate": 0.00020827893243374515,
      "loss": 2.1839,
      "step": 137989
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.1177854537963867,
      "learning_rate": 0.0002082750377524168,
      "loss": 1.9919,
      "step": 137990
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.0980764627456665,
      "learning_rate": 0.00020827114308814168,
      "loss": 2.0867,
      "step": 137991
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.0429266691207886,
      "learning_rate": 0.0002082672484409207,
      "loss": 2.1481,
      "step": 137992
    },
    {
      "epoch": 1.8,
      "grad_norm": 0.952069103717804,
      "learning_rate": 0.00020826335381075455,
      "loss": 2.1536,
      "step": 137993
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.1032209396362305,
      "learning_rate": 0.000208259459197644,
      "loss": 2.1993,
      "step": 137994
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.026957392692566,
      "learning_rate": 0.00020825556460158973,
      "loss": 1.9925,
      "step": 137995
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.0528244972229004,
      "learning_rate": 0.00020825167002259265,
      "loss": 2.0738,
      "step": 137996
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.1141986846923828,
      "learning_rate": 0.00020824777546065316,
      "loss": 2.0197,
      "step": 137997
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.0809645652770996,
      "learning_rate": 0.00020824388091577215,
      "loss": 1.9207,
      "step": 137998
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.1423838138580322,
      "learning_rate": 0.0002082399863879503,
      "loss": 2.1776,
      "step": 137999
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.0522476434707642,
      "learning_rate": 0.00020823609187718838,
      "loss": 2.2385,
      "step": 138000
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.1113828420639038,
      "learning_rate": 0.00020823219738348714,
      "loss": 2.1109,
      "step": 138001
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.1276535987854004,
      "learning_rate": 0.0002082283029068474,
      "loss": 1.7887,
      "step": 138002
    },
    {
      "epoch": 1.8,
      "grad_norm": 0.9959025979042053,
      "learning_rate": 0.0002082244084472696,
      "loss": 1.9682,
      "step": 138003
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.1342631578445435,
      "learning_rate": 0.0002082205140047546,
      "loss": 2.1333,
      "step": 138004
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.0436930656433105,
      "learning_rate": 0.00020821661957930315,
      "loss": 2.1364,
      "step": 138005
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.0427697896957397,
      "learning_rate": 0.00020821272517091596,
      "loss": 2.1049,
      "step": 138006
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.0000156164169312,
      "learning_rate": 0.00020820883077959373,
      "loss": 1.9345,
      "step": 138007
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.0893596410751343,
      "learning_rate": 0.00020820493640533739,
      "loss": 2.0932,
      "step": 138008
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.3728597164154053,
      "learning_rate": 0.00020820104204814733,
      "loss": 2.0356,
      "step": 138009
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.054444670677185,
      "learning_rate": 0.00020819714770802447,
      "loss": 2.2403,
      "step": 138010
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.078942060470581,
      "learning_rate": 0.00020819325338496946,
      "loss": 1.8992,
      "step": 138011
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.0885295867919922,
      "learning_rate": 0.00020818935907898305,
      "loss": 2.1898,
      "step": 138012
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.1731492280960083,
      "learning_rate": 0.00020818546479006597,
      "loss": 2.0371,
      "step": 138013
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.1206870079040527,
      "learning_rate": 0.00020818157051821908,
      "loss": 2.0821,
      "step": 138014
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.0597400665283203,
      "learning_rate": 0.00020817767626344285,
      "loss": 1.7834,
      "step": 138015
    },
    {
      "epoch": 1.8,
      "grad_norm": 0.9896588921546936,
      "learning_rate": 0.00020817378202573809,
      "loss": 1.9751,
      "step": 138016
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.264832615852356,
      "learning_rate": 0.0002081698878051056,
      "loss": 2.13,
      "step": 138017
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.1034973859786987,
      "learning_rate": 0.00020816599360154605,
      "loss": 2.2468,
      "step": 138018
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.2117207050323486,
      "learning_rate": 0.00020816209941506016,
      "loss": 2.1416,
      "step": 138019
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.0083839893341064,
      "learning_rate": 0.0002081582052456488,
      "loss": 2.0782,
      "step": 138020
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.1379293203353882,
      "learning_rate": 0.00020815431109331243,
      "loss": 1.9758,
      "step": 138021
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.3784654140472412,
      "learning_rate": 0.00020815041695805192,
      "loss": 2.1093,
      "step": 138022
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.0451191663742065,
      "learning_rate": 0.000208146522839868,
      "loss": 2.1453,
      "step": 138023
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.163166880607605,
      "learning_rate": 0.00020814262873876139,
      "loss": 2.1436,
      "step": 138024
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.2131556272506714,
      "learning_rate": 0.00020813873465473274,
      "loss": 1.9702,
      "step": 138025
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.252606749534607,
      "learning_rate": 0.00020813484058778298,
      "loss": 1.9081,
      "step": 138026
    },
    {
      "epoch": 1.8,
      "grad_norm": 0.9160661101341248,
      "learning_rate": 0.00020813094653791256,
      "loss": 2.2082,
      "step": 138027
    },
    {
      "epoch": 1.8,
      "grad_norm": 0.9665730595588684,
      "learning_rate": 0.00020812705250512233,
      "loss": 2.0428,
      "step": 138028
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.2952698469161987,
      "learning_rate": 0.00020812315848941303,
      "loss": 2.2133,
      "step": 138029
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.2434406280517578,
      "learning_rate": 0.00020811926449078538,
      "loss": 1.9899,
      "step": 138030
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.1433004140853882,
      "learning_rate": 0.00020811537050924002,
      "loss": 2.0144,
      "step": 138031
    },
    {
      "epoch": 1.8,
      "grad_norm": 0.9504826664924622,
      "learning_rate": 0.00020811147654477795,
      "loss": 1.9612,
      "step": 138032
    },
    {
      "epoch": 1.8,
      "grad_norm": 0.980995774269104,
      "learning_rate": 0.00020810758259739954,
      "loss": 2.1195,
      "step": 138033
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.0843894481658936,
      "learning_rate": 0.00020810368866710566,
      "loss": 1.9095,
      "step": 138034
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.0622892379760742,
      "learning_rate": 0.00020809979475389705,
      "loss": 2.1098,
      "step": 138035
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.092576026916504,
      "learning_rate": 0.00020809590085777446,
      "loss": 1.763,
      "step": 138036
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.1340327262878418,
      "learning_rate": 0.00020809200697873852,
      "loss": 2.1599,
      "step": 138037
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.2147893905639648,
      "learning_rate": 0.00020808811311679018,
      "loss": 2.128,
      "step": 138038
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.0156793594360352,
      "learning_rate": 0.00020808421927192983,
      "loss": 2.1678,
      "step": 138039
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.237078309059143,
      "learning_rate": 0.00020808032544415833,
      "loss": 2.0708,
      "step": 138040
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.0239681005477905,
      "learning_rate": 0.0002080764316334765,
      "loss": 2.1368,
      "step": 138041
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.0059070587158203,
      "learning_rate": 0.00020807253783988494,
      "loss": 2.0733,
      "step": 138042
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.220249056816101,
      "learning_rate": 0.00020806864406338447,
      "loss": 2.1815,
      "step": 138043
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.1604313850402832,
      "learning_rate": 0.00020806475030397582,
      "loss": 1.9826,
      "step": 138044
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.1817129850387573,
      "learning_rate": 0.00020806085656165964,
      "loss": 1.9286,
      "step": 138045
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.1689854860305786,
      "learning_rate": 0.0002080569628364367,
      "loss": 2.108,
      "step": 138046
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.6277891397476196,
      "learning_rate": 0.0002080530691283077,
      "loss": 2.2396,
      "step": 138047
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.1473764181137085,
      "learning_rate": 0.0002080491754372733,
      "loss": 1.9217,
      "step": 138048
    },
    {
      "epoch": 1.8,
      "grad_norm": 0.9884033203125,
      "learning_rate": 0.00020804528176333434,
      "loss": 2.3035,
      "step": 138049
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.0604612827301025,
      "learning_rate": 0.00020804138810649153,
      "loss": 2.1122,
      "step": 138050
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.0499026775360107,
      "learning_rate": 0.00020803749446674552,
      "loss": 2.1975,
      "step": 138051
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.1272464990615845,
      "learning_rate": 0.00020803360084409707,
      "loss": 2.1128,
      "step": 138052
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.0525375604629517,
      "learning_rate": 0.00020802970723854698,
      "loss": 2.2518,
      "step": 138053
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.121071219444275,
      "learning_rate": 0.00020802581365009581,
      "loss": 2.0921,
      "step": 138054
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.0616196393966675,
      "learning_rate": 0.00020802192007874441,
      "loss": 2.1905,
      "step": 138055
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.0384142398834229,
      "learning_rate": 0.00020801802652449348,
      "loss": 2.0495,
      "step": 138056
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.1562631130218506,
      "learning_rate": 0.00020801413298734376,
      "loss": 2.0593,
      "step": 138057
    },
    {
      "epoch": 1.8,
      "grad_norm": 0.9742984771728516,
      "learning_rate": 0.0002080102394672959,
      "loss": 2.1195,
      "step": 138058
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.11270010471344,
      "learning_rate": 0.0002080063459643507,
      "loss": 2.2196,
      "step": 138059
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.0523874759674072,
      "learning_rate": 0.0002080024524785089,
      "loss": 2.0089,
      "step": 138060
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.5629181861877441,
      "learning_rate": 0.00020799855900977113,
      "loss": 2.2169,
      "step": 138061
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.1086299419403076,
      "learning_rate": 0.00020799466555813814,
      "loss": 2.0568,
      "step": 138062
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.0415375232696533,
      "learning_rate": 0.00020799077212361073,
      "loss": 2.2053,
      "step": 138063
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.3863073587417603,
      "learning_rate": 0.00020798687870618955,
      "loss": 2.0469,
      "step": 138064
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.1191247701644897,
      "learning_rate": 0.00020798298530587533,
      "loss": 2.0081,
      "step": 138065
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.0492192506790161,
      "learning_rate": 0.00020797909192266879,
      "loss": 2.1333,
      "step": 138066
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.1363822221755981,
      "learning_rate": 0.0002079751985565708,
      "loss": 2.1878,
      "step": 138067
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.0859931707382202,
      "learning_rate": 0.0002079713052075818,
      "loss": 2.0952,
      "step": 138068
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.0995147228240967,
      "learning_rate": 0.00020796741187570277,
      "loss": 2.1174,
      "step": 138069
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.0745129585266113,
      "learning_rate": 0.00020796351856093435,
      "loss": 2.0561,
      "step": 138070
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.1032086610794067,
      "learning_rate": 0.00020795962526327715,
      "loss": 2.1749,
      "step": 138071
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.180331826210022,
      "learning_rate": 0.00020795573198273206,
      "loss": 2.2119,
      "step": 138072
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.1326292753219604,
      "learning_rate": 0.0002079518387192997,
      "loss": 2.0587,
      "step": 138073
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.2564241886138916,
      "learning_rate": 0.00020794794547298082,
      "loss": 2.0872,
      "step": 138074
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.0485646724700928,
      "learning_rate": 0.00020794405224377628,
      "loss": 2.1489,
      "step": 138075
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.4335614442825317,
      "learning_rate": 0.00020794015903168657,
      "loss": 2.1877,
      "step": 138076
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.0394400358200073,
      "learning_rate": 0.00020793626583671254,
      "loss": 2.2385,
      "step": 138077
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.1025725603103638,
      "learning_rate": 0.00020793237265885484,
      "loss": 2.1945,
      "step": 138078
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.3016819953918457,
      "learning_rate": 0.00020792847949811433,
      "loss": 2.1697,
      "step": 138079
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.3898481130599976,
      "learning_rate": 0.00020792458635449158,
      "loss": 1.8924,
      "step": 138080
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.1523975133895874,
      "learning_rate": 0.00020792069322798755,
      "loss": 2.1336,
      "step": 138081
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.0150537490844727,
      "learning_rate": 0.00020791680011860269,
      "loss": 2.2195,
      "step": 138082
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.247052550315857,
      "learning_rate": 0.0002079129070263378,
      "loss": 2.0397,
      "step": 138083
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.1972784996032715,
      "learning_rate": 0.00020790901395119365,
      "loss": 2.049,
      "step": 138084
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.1184297800064087,
      "learning_rate": 0.00020790512089317097,
      "loss": 2.1762,
      "step": 138085
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.0394763946533203,
      "learning_rate": 0.00020790122785227044,
      "loss": 2.2758,
      "step": 138086
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.1084468364715576,
      "learning_rate": 0.00020789733482849297,
      "loss": 2.1722,
      "step": 138087
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.305869460105896,
      "learning_rate": 0.00020789344182183897,
      "loss": 1.9002,
      "step": 138088
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.0860505104064941,
      "learning_rate": 0.00020788954883230933,
      "loss": 2.1002,
      "step": 138089
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.1942871809005737,
      "learning_rate": 0.00020788565585990476,
      "loss": 1.9993,
      "step": 138090
    },
    {
      "epoch": 1.8,
      "grad_norm": 0.9852914214134216,
      "learning_rate": 0.00020788176290462599,
      "loss": 2.2966,
      "step": 138091
    },
    {
      "epoch": 1.8,
      "grad_norm": 0.9488318562507629,
      "learning_rate": 0.0002078778699664737,
      "loss": 1.8507,
      "step": 138092
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.340502142906189,
      "learning_rate": 0.00020787397704544883,
      "loss": 2.1342,
      "step": 138093
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.1128780841827393,
      "learning_rate": 0.0002078700841415518,
      "loss": 2.1392,
      "step": 138094
    },
    {
      "epoch": 1.8,
      "grad_norm": 0.9825331568717957,
      "learning_rate": 0.00020786619125478343,
      "loss": 2.1152,
      "step": 138095
    },
    {
      "epoch": 1.8,
      "grad_norm": 0.9921967387199402,
      "learning_rate": 0.00020786229838514448,
      "loss": 2.083,
      "step": 138096
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.0697981119155884,
      "learning_rate": 0.00020785840553263566,
      "loss": 2.1921,
      "step": 138097
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.1137503385543823,
      "learning_rate": 0.00020785451269725774,
      "loss": 2.0396,
      "step": 138098
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.1425588130950928,
      "learning_rate": 0.0002078506198790115,
      "loss": 2.125,
      "step": 138099
    },
    {
      "epoch": 1.8,
      "grad_norm": 0.9814831018447876,
      "learning_rate": 0.00020784672707789743,
      "loss": 2.1312,
      "step": 138100
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.00946044921875,
      "learning_rate": 0.00020784283429391644,
      "loss": 1.9849,
      "step": 138101
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.0877050161361694,
      "learning_rate": 0.00020783894152706916,
      "loss": 2.0596,
      "step": 138102
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.067562222480774,
      "learning_rate": 0.00020783504877735639,
      "loss": 2.1146,
      "step": 138103
    },
    {
      "epoch": 1.8,
      "grad_norm": 0.9379885196685791,
      "learning_rate": 0.0002078311560447788,
      "loss": 2.0805,
      "step": 138104
    },
    {
      "epoch": 1.8,
      "grad_norm": 0.968113899230957,
      "learning_rate": 0.00020782726332933729,
      "loss": 2.0255,
      "step": 138105
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.0206193923950195,
      "learning_rate": 0.00020782337063103229,
      "loss": 2.1413,
      "step": 138106
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.0390992164611816,
      "learning_rate": 0.0002078194779498647,
      "loss": 1.9596,
      "step": 138107
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.0959900617599487,
      "learning_rate": 0.00020781558528583518,
      "loss": 2.3036,
      "step": 138108
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.0649207830429077,
      "learning_rate": 0.00020781169263894446,
      "loss": 2.1813,
      "step": 138109
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.1110076904296875,
      "learning_rate": 0.0002078078000091933,
      "loss": 2.111,
      "step": 138110
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.1786249876022339,
      "learning_rate": 0.00020780390739658254,
      "loss": 2.2798,
      "step": 138111
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.0814125537872314,
      "learning_rate": 0.00020780001480111266,
      "loss": 1.9964,
      "step": 138112
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.073246717453003,
      "learning_rate": 0.00020779612222278447,
      "loss": 1.9417,
      "step": 138113
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.0425609350204468,
      "learning_rate": 0.00020779222966159876,
      "loss": 2.0702,
      "step": 138114
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.1273447275161743,
      "learning_rate": 0.0002077883371175562,
      "loss": 1.9884,
      "step": 138115
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.223482370376587,
      "learning_rate": 0.00020778444459065753,
      "loss": 2.1463,
      "step": 138116
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.0902374982833862,
      "learning_rate": 0.00020778055208090358,
      "loss": 2.2041,
      "step": 138117
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.0410431623458862,
      "learning_rate": 0.00020777665958829483,
      "loss": 2.4759,
      "step": 138118
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.138630986213684,
      "learning_rate": 0.0002077727671128322,
      "loss": 2.141,
      "step": 138119
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.2547911405563354,
      "learning_rate": 0.0002077688746545163,
      "loss": 2.1045,
      "step": 138120
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.0105922222137451,
      "learning_rate": 0.00020776498221334794,
      "loss": 2.1258,
      "step": 138121
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.0267159938812256,
      "learning_rate": 0.0002077610897893278,
      "loss": 2.1091,
      "step": 138122
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.0525966882705688,
      "learning_rate": 0.00020775719738245658,
      "loss": 2.2915,
      "step": 138123
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.0708884000778198,
      "learning_rate": 0.0002077533049927352,
      "loss": 2.0514,
      "step": 138124
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.351188063621521,
      "learning_rate": 0.0002077494126201641,
      "loss": 1.9319,
      "step": 138125
    },
    {
      "epoch": 1.8,
      "grad_norm": 0.9757946729660034,
      "learning_rate": 0.0002077455202647441,
      "loss": 2.0275,
      "step": 138126
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.2153887748718262,
      "learning_rate": 0.00020774162792647598,
      "loss": 2.1235,
      "step": 138127
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.2902742624282837,
      "learning_rate": 0.00020773773560536042,
      "loss": 2.1056,
      "step": 138128
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.0401129722595215,
      "learning_rate": 0.00020773384330139815,
      "loss": 2.1366,
      "step": 138129
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.0303962230682373,
      "learning_rate": 0.00020772995101459002,
      "loss": 2.1223,
      "step": 138130
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.059324860572815,
      "learning_rate": 0.0002077260587449365,
      "loss": 2.1339,
      "step": 138131
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.2105505466461182,
      "learning_rate": 0.00020772216649243846,
      "loss": 2.2491,
      "step": 138132
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.3392231464385986,
      "learning_rate": 0.00020771827425709665,
      "loss": 2.2014,
      "step": 138133
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.1105661392211914,
      "learning_rate": 0.0002077143820389117,
      "loss": 2.3276,
      "step": 138134
    },
    {
      "epoch": 1.8,
      "grad_norm": 2.19040846824646,
      "learning_rate": 0.00020771048983788442,
      "loss": 2.2406,
      "step": 138135
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.2101178169250488,
      "learning_rate": 0.00020770659765401554,
      "loss": 2.0952,
      "step": 138136
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.0338717699050903,
      "learning_rate": 0.0002077027054873057,
      "loss": 2.1208,
      "step": 138137
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.0066051483154297,
      "learning_rate": 0.00020769881333775576,
      "loss": 2.194,
      "step": 138138
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.0099217891693115,
      "learning_rate": 0.00020769492120536624,
      "loss": 2.1508,
      "step": 138139
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.3706433773040771,
      "learning_rate": 0.000207691029090138,
      "loss": 2.4138,
      "step": 138140
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.0568994283676147,
      "learning_rate": 0.00020768713699207173,
      "loss": 1.9837,
      "step": 138141
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.1116677522659302,
      "learning_rate": 0.00020768324491116823,
      "loss": 2.0908,
      "step": 138142
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.123081088066101,
      "learning_rate": 0.0002076793528474281,
      "loss": 1.929,
      "step": 138143
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.0012991428375244,
      "learning_rate": 0.00020767546080085214,
      "loss": 2.1049,
      "step": 138144
    },
    {
      "epoch": 1.8,
      "grad_norm": 0.9737892746925354,
      "learning_rate": 0.0002076715687714411,
      "loss": 1.9077,
      "step": 138145
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.119924545288086,
      "learning_rate": 0.00020766767675919556,
      "loss": 2.0443,
      "step": 138146
    },
    {
      "epoch": 1.8,
      "grad_norm": 0.9462578892707825,
      "learning_rate": 0.00020766378476411637,
      "loss": 1.6457,
      "step": 138147
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.0830028057098389,
      "learning_rate": 0.00020765989278620432,
      "loss": 2.131,
      "step": 138148
    },
    {
      "epoch": 1.8,
      "grad_norm": 0.9792296290397644,
      "learning_rate": 0.0002076560008254599,
      "loss": 2.13,
      "step": 138149
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.229048728942871,
      "learning_rate": 0.00020765210888188404,
      "loss": 2.2725,
      "step": 138150
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.014917016029358,
      "learning_rate": 0.0002076482169554774,
      "loss": 2.1551,
      "step": 138151
    },
    {
      "epoch": 1.8,
      "grad_norm": 0.9684793949127197,
      "learning_rate": 0.0002076443250462407,
      "loss": 2.1011,
      "step": 138152
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.1899373531341553,
      "learning_rate": 0.00020764043315417465,
      "loss": 1.8327,
      "step": 138153
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.1175683736801147,
      "learning_rate": 0.00020763654127928005,
      "loss": 2.2049,
      "step": 138154
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.0867156982421875,
      "learning_rate": 0.00020763264942155749,
      "loss": 1.9822,
      "step": 138155
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.0575897693634033,
      "learning_rate": 0.00020762875758100773,
      "loss": 2.1462,
      "step": 138156
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.0547876358032227,
      "learning_rate": 0.00020762486575763158,
      "loss": 2.0545,
      "step": 138157
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.0899487733840942,
      "learning_rate": 0.00020762097395142967,
      "loss": 2.2287,
      "step": 138158
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.1932709217071533,
      "learning_rate": 0.00020761708216240286,
      "loss": 2.124,
      "step": 138159
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.017477035522461,
      "learning_rate": 0.0002076131903905518,
      "loss": 1.9894,
      "step": 138160
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.0463835000991821,
      "learning_rate": 0.00020760929863587705,
      "loss": 2.0944,
      "step": 138161
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.0319139957427979,
      "learning_rate": 0.00020760540689837952,
      "loss": 2.1695,
      "step": 138162
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.0665168762207031,
      "learning_rate": 0.00020760151517805993,
      "loss": 2.0267,
      "step": 138163
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.0171642303466797,
      "learning_rate": 0.0002075976234749189,
      "loss": 2.129,
      "step": 138164
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.0444473028182983,
      "learning_rate": 0.0002075937317889572,
      "loss": 2.1121,
      "step": 138165
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.0392621755599976,
      "learning_rate": 0.00020758984012017576,
      "loss": 2.3252,
      "step": 138166
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.1564313173294067,
      "learning_rate": 0.00020758594846857496,
      "loss": 2.0494,
      "step": 138167
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.0283024311065674,
      "learning_rate": 0.0002075820568341557,
      "loss": 2.0882,
      "step": 138168
    },
    {
      "epoch": 1.8,
      "grad_norm": 0.9980134963989258,
      "learning_rate": 0.00020757816521691863,
      "loss": 2.1319,
      "step": 138169
    },
    {
      "epoch": 1.8,
      "grad_norm": 0.9694865942001343,
      "learning_rate": 0.00020757427361686457,
      "loss": 2.0477,
      "step": 138170
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.0898146629333496,
      "learning_rate": 0.00020757038203399414,
      "loss": 2.1668,
      "step": 138171
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.1269681453704834,
      "learning_rate": 0.0002075664904683083,
      "loss": 2.0914,
      "step": 138172
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.0195926427841187,
      "learning_rate": 0.00020756259891980747,
      "loss": 2.0645,
      "step": 138173
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.012343406677246,
      "learning_rate": 0.00020755870738849248,
      "loss": 2.1204,
      "step": 138174
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.1244534254074097,
      "learning_rate": 0.00020755481587436407,
      "loss": 1.8575,
      "step": 138175
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.016948938369751,
      "learning_rate": 0.00020755092437742294,
      "loss": 2.0874,
      "step": 138176
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.0966835021972656,
      "learning_rate": 0.00020754703289766987,
      "loss": 2.1544,
      "step": 138177
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.006874918937683,
      "learning_rate": 0.00020754314143510568,
      "loss": 1.9243,
      "step": 138178
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.00819730758667,
      "learning_rate": 0.0002075392499897308,
      "loss": 2.1353,
      "step": 138179
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.1435930728912354,
      "learning_rate": 0.0002075353585615462,
      "loss": 1.9194,
      "step": 138180
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.0587646961212158,
      "learning_rate": 0.00020753146715055246,
      "loss": 2.1837,
      "step": 138181
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.0444406270980835,
      "learning_rate": 0.00020752757575675038,
      "loss": 2.0416,
      "step": 138182
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.1384822130203247,
      "learning_rate": 0.00020752368438014065,
      "loss": 2.1548,
      "step": 138183
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.042427897453308,
      "learning_rate": 0.00020751979302072415,
      "loss": 2.236,
      "step": 138184
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.2578539848327637,
      "learning_rate": 0.00020751590167850134,
      "loss": 2.0838,
      "step": 138185
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.1007426977157593,
      "learning_rate": 0.00020751201035347307,
      "loss": 2.0936,
      "step": 138186
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.2362107038497925,
      "learning_rate": 0.00020750811904564012,
      "loss": 1.9895,
      "step": 138187
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.2168208360671997,
      "learning_rate": 0.00020750422775500307,
      "loss": 2.1247,
      "step": 138188
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.3397114276885986,
      "learning_rate": 0.0002075003364815628,
      "loss": 2.1291,
      "step": 138189
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.3335869312286377,
      "learning_rate": 0.0002074964452253199,
      "loss": 2.0757,
      "step": 138190
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.1060199737548828,
      "learning_rate": 0.00020749255398627532,
      "loss": 2.0664,
      "step": 138191
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.229546308517456,
      "learning_rate": 0.00020748866276442947,
      "loss": 2.0387,
      "step": 138192
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.0994899272918701,
      "learning_rate": 0.00020748477155978326,
      "loss": 2.0482,
      "step": 138193
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.1165674924850464,
      "learning_rate": 0.00020748088037233735,
      "loss": 1.9035,
      "step": 138194
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.1287955045700073,
      "learning_rate": 0.0002074769892020925,
      "loss": 2.0892,
      "step": 138195
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.2186572551727295,
      "learning_rate": 0.0002074730980490494,
      "loss": 2.0314,
      "step": 138196
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.0560051202774048,
      "learning_rate": 0.00020746920691320895,
      "loss": 2.1396,
      "step": 138197
    },
    {
      "epoch": 1.8,
      "grad_norm": 0.9818944931030273,
      "learning_rate": 0.0002074653157945716,
      "loss": 2.0051,
      "step": 138198
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.0181450843811035,
      "learning_rate": 0.00020746142469313817,
      "loss": 2.138,
      "step": 138199
    },
    {
      "epoch": 1.8,
      "grad_norm": 0.9602856636047363,
      "learning_rate": 0.00020745753360890942,
      "loss": 2.0749,
      "step": 138200
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.1511672735214233,
      "learning_rate": 0.0002074536425418861,
      "loss": 2.1878,
      "step": 138201
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.1943849325180054,
      "learning_rate": 0.00020744975149206886,
      "loss": 2.3623,
      "step": 138202
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.1524579524993896,
      "learning_rate": 0.00020744586045945857,
      "loss": 2.0661,
      "step": 138203
    },
    {
      "epoch": 1.8,
      "grad_norm": 0.9792355298995972,
      "learning_rate": 0.00020744196944405576,
      "loss": 2.0827,
      "step": 138204
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.1449466943740845,
      "learning_rate": 0.0002074380784458612,
      "loss": 2.0833,
      "step": 138205
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.0384963750839233,
      "learning_rate": 0.00020743418746487565,
      "loss": 2.1839,
      "step": 138206
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.0369502305984497,
      "learning_rate": 0.00020743029650109978,
      "loss": 2.0758,
      "step": 138207
    },
    {
      "epoch": 1.8,
      "grad_norm": 0.9863651394844055,
      "learning_rate": 0.00020742640555453446,
      "loss": 1.987,
      "step": 138208
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.1491141319274902,
      "learning_rate": 0.0002074225146251804,
      "loss": 1.9947,
      "step": 138209
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.3653512001037598,
      "learning_rate": 0.00020741862371303812,
      "loss": 2.0903,
      "step": 138210
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.0880283117294312,
      "learning_rate": 0.00020741473281810847,
      "loss": 1.9719,
      "step": 138211
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.011821985244751,
      "learning_rate": 0.00020741084194039215,
      "loss": 2.1213,
      "step": 138212
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.2315441370010376,
      "learning_rate": 0.00020740695107988993,
      "loss": 2.1628,
      "step": 138213
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.1034268140792847,
      "learning_rate": 0.0002074030602366025,
      "loss": 2.1691,
      "step": 138214
    },
    {
      "epoch": 1.8,
      "grad_norm": 0.9585034847259521,
      "learning_rate": 0.00020739916941053073,
      "loss": 2.1817,
      "step": 138215
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.477486491203308,
      "learning_rate": 0.00020739527860167502,
      "loss": 2.1341,
      "step": 138216
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.2820833921432495,
      "learning_rate": 0.0002073913878100363,
      "loss": 1.8502,
      "step": 138217
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.1550145149230957,
      "learning_rate": 0.00020738749703561531,
      "loss": 1.9871,
      "step": 138218
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.0862183570861816,
      "learning_rate": 0.00020738360627841268,
      "loss": 2.266,
      "step": 138219
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.174355149269104,
      "learning_rate": 0.00020737971553842923,
      "loss": 1.8763,
      "step": 138220
    },
    {
      "epoch": 1.8,
      "grad_norm": 0.8911355137825012,
      "learning_rate": 0.00020737582481566565,
      "loss": 1.9958,
      "step": 138221
    },
    {
      "epoch": 1.8,
      "grad_norm": 0.965166449546814,
      "learning_rate": 0.00020737193411012273,
      "loss": 1.9747,
      "step": 138222
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.173107624053955,
      "learning_rate": 0.00020736804342180097,
      "loss": 2.1988,
      "step": 138223
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.0940642356872559,
      "learning_rate": 0.00020736415275070128,
      "loss": 1.9546,
      "step": 138224
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.032470703125,
      "learning_rate": 0.00020736026209682433,
      "loss": 1.8208,
      "step": 138225
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.282749891281128,
      "learning_rate": 0.00020735637146017085,
      "loss": 2.0896,
      "step": 138226
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.057824730873108,
      "learning_rate": 0.00020735248084074166,
      "loss": 1.9747,
      "step": 138227
    },
    {
      "epoch": 1.8,
      "grad_norm": 0.9568272233009338,
      "learning_rate": 0.00020734859023853732,
      "loss": 2.0932,
      "step": 138228
    },
    {
      "epoch": 1.8,
      "grad_norm": 0.9739159345626831,
      "learning_rate": 0.0002073446996535587,
      "loss": 2.029,
      "step": 138229
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.0743755102157593,
      "learning_rate": 0.00020734080908580636,
      "loss": 1.9701,
      "step": 138230
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.0104622840881348,
      "learning_rate": 0.00020733691853528114,
      "loss": 2.2174,
      "step": 138231
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.0828917026519775,
      "learning_rate": 0.00020733302800198368,
      "loss": 2.051,
      "step": 138232
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.0094152688980103,
      "learning_rate": 0.00020732913748591487,
      "loss": 2.1077,
      "step": 138233
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.0365283489227295,
      "learning_rate": 0.00020732524698707525,
      "loss": 2.2462,
      "step": 138234
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.1220731735229492,
      "learning_rate": 0.0002073213565054656,
      "loss": 2.0288,
      "step": 138235
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.148097038269043,
      "learning_rate": 0.0002073174660410867,
      "loss": 2.1969,
      "step": 138236
    },
    {
      "epoch": 1.8,
      "grad_norm": 0.9845048189163208,
      "learning_rate": 0.00020731357559393925,
      "loss": 1.9093,
      "step": 138237
    },
    {
      "epoch": 1.8,
      "grad_norm": 0.9396331310272217,
      "learning_rate": 0.00020730968516402394,
      "loss": 1.862,
      "step": 138238
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.2074161767959595,
      "learning_rate": 0.00020730579475134158,
      "loss": 1.9723,
      "step": 138239
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.0956084728240967,
      "learning_rate": 0.0002073019043558927,
      "loss": 2.0817,
      "step": 138240
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.0790481567382812,
      "learning_rate": 0.00020729801397767817,
      "loss": 1.8517,
      "step": 138241
    },
    {
      "epoch": 1.8,
      "grad_norm": 0.9614539742469788,
      "learning_rate": 0.0002072941236166987,
      "loss": 2.1325,
      "step": 138242
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.0601941347122192,
      "learning_rate": 0.000207290233272955,
      "loss": 2.2027,
      "step": 138243
    },
    {
      "epoch": 1.8,
      "grad_norm": 0.9938151240348816,
      "learning_rate": 0.00020728634294644784,
      "loss": 2.0135,
      "step": 138244
    },
    {
      "epoch": 1.8,
      "grad_norm": 0.9867423176765442,
      "learning_rate": 0.00020728245263717791,
      "loss": 2.2263,
      "step": 138245
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.0771100521087646,
      "learning_rate": 0.00020727856234514588,
      "loss": 1.9395,
      "step": 138246
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.1033848524093628,
      "learning_rate": 0.00020727467207035246,
      "loss": 2.2751,
      "step": 138247
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.0727131366729736,
      "learning_rate": 0.00020727078181279851,
      "loss": 2.1013,
      "step": 138248
    },
    {
      "epoch": 1.8,
      "grad_norm": 0.9983869791030884,
      "learning_rate": 0.00020726689157248463,
      "loss": 2.1266,
      "step": 138249
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.01748526096344,
      "learning_rate": 0.00020726300134941153,
      "loss": 2.0001,
      "step": 138250
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.0078004598617554,
      "learning_rate": 0.0002072591111435802,
      "loss": 1.8595,
      "step": 138251
    },
    {
      "epoch": 1.8,
      "grad_norm": 0.970701277256012,
      "learning_rate": 0.000207255220954991,
      "loss": 2.12,
      "step": 138252
    },
    {
      "epoch": 1.8,
      "grad_norm": 0.961601972579956,
      "learning_rate": 0.0002072513307836448,
      "loss": 2.0897,
      "step": 138253
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.1325159072875977,
      "learning_rate": 0.0002072474406295423,
      "loss": 1.9079,
      "step": 138254
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.1381878852844238,
      "learning_rate": 0.0002072435504926843,
      "loss": 2.0589,
      "step": 138255
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.0855997800827026,
      "learning_rate": 0.00020723966037307142,
      "loss": 2.2768,
      "step": 138256
    },
    {
      "epoch": 1.8,
      "grad_norm": 0.9935555458068848,
      "learning_rate": 0.00020723577027070448,
      "loss": 1.9887,
      "step": 138257
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.0259090662002563,
      "learning_rate": 0.0002072318801855843,
      "loss": 1.8496,
      "step": 138258
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.083531379699707,
      "learning_rate": 0.0002072279901177113,
      "loss": 2.1948,
      "step": 138259
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.0508450269699097,
      "learning_rate": 0.0002072241000670864,
      "loss": 2.2207,
      "step": 138260
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.2725107669830322,
      "learning_rate": 0.00020722021003371027,
      "loss": 2.1211,
      "step": 138261
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.0777404308319092,
      "learning_rate": 0.00020721632001758367,
      "loss": 2.0675,
      "step": 138262
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.0923317670822144,
      "learning_rate": 0.00020721243001870727,
      "loss": 2.0632,
      "step": 138263
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.2247788906097412,
      "learning_rate": 0.000207208540037082,
      "loss": 2.1613,
      "step": 138264
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.143283724784851,
      "learning_rate": 0.00020720465007270831,
      "loss": 1.932,
      "step": 138265
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.2257078886032104,
      "learning_rate": 0.00020720076012558695,
      "loss": 2.2032,
      "step": 138266
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.072221279144287,
      "learning_rate": 0.00020719687019571877,
      "loss": 2.1778,
      "step": 138267
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.098783016204834,
      "learning_rate": 0.00020719298028310445,
      "loss": 2.1409,
      "step": 138268
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.0115629434585571,
      "learning_rate": 0.0002071890903877447,
      "loss": 2.0759,
      "step": 138269
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.0523992776870728,
      "learning_rate": 0.00020718520050964033,
      "loss": 2.2123,
      "step": 138270
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.0652674436569214,
      "learning_rate": 0.00020718131064879192,
      "loss": 2.2657,
      "step": 138271
    },
    {
      "epoch": 1.8,
      "grad_norm": 0.9411019086837769,
      "learning_rate": 0.0002071774208052002,
      "loss": 2.0487,
      "step": 138272
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.145453691482544,
      "learning_rate": 0.00020717353097886597,
      "loss": 2.1754,
      "step": 138273
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.0911625623703003,
      "learning_rate": 0.00020716964116978997,
      "loss": 2.0098,
      "step": 138274
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.1178362369537354,
      "learning_rate": 0.00020716575137797285,
      "loss": 2.1141,
      "step": 138275
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.0090417861938477,
      "learning_rate": 0.0002071618616034155,
      "loss": 2.114,
      "step": 138276
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.4552034139633179,
      "learning_rate": 0.0002071579718461184,
      "loss": 1.9934,
      "step": 138277
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.0153958797454834,
      "learning_rate": 0.00020715408210608238,
      "loss": 2.2112,
      "step": 138278
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.0309394598007202,
      "learning_rate": 0.00020715019238330817,
      "loss": 1.818,
      "step": 138279
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.0449830293655396,
      "learning_rate": 0.0002071463026777965,
      "loss": 2.1322,
      "step": 138280
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.13539719581604,
      "learning_rate": 0.00020714241298954805,
      "loss": 2.1645,
      "step": 138281
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.2643311023712158,
      "learning_rate": 0.00020713852331856372,
      "loss": 2.1325,
      "step": 138282
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.1234941482543945,
      "learning_rate": 0.00020713463366484397,
      "loss": 2.1142,
      "step": 138283
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.1002017259597778,
      "learning_rate": 0.00020713074402838966,
      "loss": 1.8559,
      "step": 138284
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.0091623067855835,
      "learning_rate": 0.00020712685440920148,
      "loss": 2.0286,
      "step": 138285
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.0102956295013428,
      "learning_rate": 0.00020712296480728017,
      "loss": 1.9583,
      "step": 138286
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.1732474565505981,
      "learning_rate": 0.00020711907522262646,
      "loss": 2.2307,
      "step": 138287
    },
    {
      "epoch": 1.8,
      "grad_norm": 0.9900072813034058,
      "learning_rate": 0.0002071151856552412,
      "loss": 1.9607,
      "step": 138288
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.0574510097503662,
      "learning_rate": 0.00020711129610512485,
      "loss": 1.9747,
      "step": 138289
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.0382698774337769,
      "learning_rate": 0.00020710740657227825,
      "loss": 2.0221,
      "step": 138290
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.0474225282669067,
      "learning_rate": 0.00020710351705670215,
      "loss": 2.1495,
      "step": 138291
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.0451774597167969,
      "learning_rate": 0.00020709962755839726,
      "loss": 2.1008,
      "step": 138292
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.055588960647583,
      "learning_rate": 0.00020709573807736428,
      "loss": 2.2973,
      "step": 138293
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.1253553628921509,
      "learning_rate": 0.00020709184861360412,
      "loss": 2.2213,
      "step": 138294
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.1193093061447144,
      "learning_rate": 0.0002070879591671172,
      "loss": 2.2808,
      "step": 138295
    },
    {
      "epoch": 1.8,
      "grad_norm": 0.9681773781776428,
      "learning_rate": 0.00020708406973790437,
      "loss": 2.1203,
      "step": 138296
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.0748600959777832,
      "learning_rate": 0.00020708018032596637,
      "loss": 2.1406,
      "step": 138297
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.120971918106079,
      "learning_rate": 0.00020707629093130392,
      "loss": 2.1695,
      "step": 138298
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.1567714214324951,
      "learning_rate": 0.00020707240155391776,
      "loss": 2.1174,
      "step": 138299
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.0817900896072388,
      "learning_rate": 0.0002070685121938087,
      "loss": 2.2704,
      "step": 138300
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.0426647663116455,
      "learning_rate": 0.00020706462285097722,
      "loss": 2.1368,
      "step": 138301
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.0566928386688232,
      "learning_rate": 0.00020706073352542417,
      "loss": 1.9274,
      "step": 138302
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.1208298206329346,
      "learning_rate": 0.00020705684421715032,
      "loss": 2.172,
      "step": 138303
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.1255333423614502,
      "learning_rate": 0.00020705295492615635,
      "loss": 2.2094,
      "step": 138304
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.031694769859314,
      "learning_rate": 0.00020704906565244296,
      "loss": 2.1363,
      "step": 138305
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.0605430603027344,
      "learning_rate": 0.00020704517639601102,
      "loss": 2.0068,
      "step": 138306
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.0186673402786255,
      "learning_rate": 0.00020704128715686113,
      "loss": 2.0328,
      "step": 138307
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.0582144260406494,
      "learning_rate": 0.0002070373979349939,
      "loss": 2.0023,
      "step": 138308
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.0060287714004517,
      "learning_rate": 0.00020703350873041025,
      "loss": 2.1296,
      "step": 138309
    },
    {
      "epoch": 1.8,
      "grad_norm": 0.9341936111450195,
      "learning_rate": 0.00020702961954311078,
      "loss": 2.0187,
      "step": 138310
    },
    {
      "epoch": 1.8,
      "grad_norm": 0.9980370998382568,
      "learning_rate": 0.00020702573037309624,
      "loss": 2.0084,
      "step": 138311
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.0070983171463013,
      "learning_rate": 0.00020702184122036748,
      "loss": 2.044,
      "step": 138312
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.0278745889663696,
      "learning_rate": 0.00020701795208492502,
      "loss": 1.9539,
      "step": 138313
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.0866552591323853,
      "learning_rate": 0.00020701406296676973,
      "loss": 2.0106,
      "step": 138314
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.1011420488357544,
      "learning_rate": 0.00020701017386590224,
      "loss": 2.1233,
      "step": 138315
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.0660386085510254,
      "learning_rate": 0.00020700628478232332,
      "loss": 2.1278,
      "step": 138316
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.0057153701782227,
      "learning_rate": 0.00020700239571603367,
      "loss": 2.0077,
      "step": 138317
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.196915864944458,
      "learning_rate": 0.00020699850666703408,
      "loss": 2.0271,
      "step": 138318
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.051530361175537,
      "learning_rate": 0.0002069946176353252,
      "loss": 1.928,
      "step": 138319
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.0709967613220215,
      "learning_rate": 0.00020699072862090777,
      "loss": 2.4295,
      "step": 138320
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.0976626873016357,
      "learning_rate": 0.00020698683962378257,
      "loss": 2.2608,
      "step": 138321
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.0564247369766235,
      "learning_rate": 0.0002069829506439502,
      "loss": 1.8781,
      "step": 138322
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.2070965766906738,
      "learning_rate": 0.00020697906168141148,
      "loss": 2.071,
      "step": 138323
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.0187582969665527,
      "learning_rate": 0.00020697517273616712,
      "loss": 2.1726,
      "step": 138324
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.1139881610870361,
      "learning_rate": 0.00020697128380821783,
      "loss": 2.2339,
      "step": 138325
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.1893162727355957,
      "learning_rate": 0.00020696739489756434,
      "loss": 2.1002,
      "step": 138326
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.1739107370376587,
      "learning_rate": 0.00020696350600420732,
      "loss": 2.0123,
      "step": 138327
    },
    {
      "epoch": 1.8,
      "grad_norm": 0.9760886430740356,
      "learning_rate": 0.00020695961712814754,
      "loss": 2.046,
      "step": 138328
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.2767211198806763,
      "learning_rate": 0.0002069557282693858,
      "loss": 2.0528,
      "step": 138329
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.0999093055725098,
      "learning_rate": 0.00020695183942792268,
      "loss": 2.0604,
      "step": 138330
    },
    {
      "epoch": 1.8,
      "grad_norm": 0.9885624647140503,
      "learning_rate": 0.00020694795060375906,
      "loss": 2.0807,
      "step": 138331
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.056834101676941,
      "learning_rate": 0.00020694406179689546,
      "loss": 2.0717,
      "step": 138332
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.1300877332687378,
      "learning_rate": 0.00020694017300733276,
      "loss": 2.1989,
      "step": 138333
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.0909147262573242,
      "learning_rate": 0.00020693628423507164,
      "loss": 2.2038,
      "step": 138334
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.0029174089431763,
      "learning_rate": 0.00020693239548011277,
      "loss": 2.0608,
      "step": 138335
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.0890498161315918,
      "learning_rate": 0.00020692850674245702,
      "loss": 1.9177,
      "step": 138336
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.1462926864624023,
      "learning_rate": 0.00020692461802210502,
      "loss": 1.9697,
      "step": 138337
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.355628252029419,
      "learning_rate": 0.00020692072931905746,
      "loss": 2.01,
      "step": 138338
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.150111436843872,
      "learning_rate": 0.00020691684063331504,
      "loss": 2.016,
      "step": 138339
    },
    {
      "epoch": 1.8,
      "grad_norm": 0.981110692024231,
      "learning_rate": 0.00020691295196487858,
      "loss": 2.0597,
      "step": 138340
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.0307327508926392,
      "learning_rate": 0.00020690906331374871,
      "loss": 2.191,
      "step": 138341
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.0428715944290161,
      "learning_rate": 0.00020690517467992624,
      "loss": 2.2892,
      "step": 138342
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.0355224609375,
      "learning_rate": 0.00020690128606341198,
      "loss": 2.0702,
      "step": 138343
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.1368165016174316,
      "learning_rate": 0.00020689739746420637,
      "loss": 2.1388,
      "step": 138344
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.0349705219268799,
      "learning_rate": 0.00020689350888231033,
      "loss": 2.0296,
      "step": 138345
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.000005841255188,
      "learning_rate": 0.00020688962031772452,
      "loss": 2.093,
      "step": 138346
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.0937154293060303,
      "learning_rate": 0.0002068857317704497,
      "loss": 2.0936,
      "step": 138347
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.0469715595245361,
      "learning_rate": 0.00020688184324048656,
      "loss": 2.0548,
      "step": 138348
    },
    {
      "epoch": 1.8,
      "grad_norm": 0.9353471398353577,
      "learning_rate": 0.00020687795472783602,
      "loss": 2.038,
      "step": 138349
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.078446388244629,
      "learning_rate": 0.00020687406623249842,
      "loss": 2.1334,
      "step": 138350
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.0246880054473877,
      "learning_rate": 0.00020687017775447474,
      "loss": 2.0641,
      "step": 138351
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.023819088935852,
      "learning_rate": 0.00020686628929376564,
      "loss": 2.2429,
      "step": 138352
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.1175353527069092,
      "learning_rate": 0.0002068624008503719,
      "loss": 2.434,
      "step": 138353
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.0591074228286743,
      "learning_rate": 0.00020685851242429414,
      "loss": 2.1282,
      "step": 138354
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.0658690929412842,
      "learning_rate": 0.00020685462401553325,
      "loss": 2.3188,
      "step": 138355
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.1342198848724365,
      "learning_rate": 0.00020685073562408976,
      "loss": 1.9397,
      "step": 138356
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.0100243091583252,
      "learning_rate": 0.00020684684724996447,
      "loss": 2.1654,
      "step": 138357
    },
    {
      "epoch": 1.8,
      "grad_norm": 0.9908640384674072,
      "learning_rate": 0.00020684295889315815,
      "loss": 1.8635,
      "step": 138358
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.0785953998565674,
      "learning_rate": 0.00020683907055367142,
      "loss": 2.0472,
      "step": 138359
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.0920552015304565,
      "learning_rate": 0.00020683518223150508,
      "loss": 2.0158,
      "step": 138360
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.2993111610412598,
      "learning_rate": 0.00020683129392665996,
      "loss": 2.1408,
      "step": 138361
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.2012306451797485,
      "learning_rate": 0.00020682740563913654,
      "loss": 2.1746,
      "step": 138362
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.0639485120773315,
      "learning_rate": 0.0002068235173689357,
      "loss": 2.0401,
      "step": 138363
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.0026347637176514,
      "learning_rate": 0.0002068196291160581,
      "loss": 2.2091,
      "step": 138364
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.2705333232879639,
      "learning_rate": 0.00020681574088050447,
      "loss": 1.9181,
      "step": 138365
    },
    {
      "epoch": 1.8,
      "grad_norm": 0.9815737009048462,
      "learning_rate": 0.00020681185266227558,
      "loss": 1.9396,
      "step": 138366
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.093248963356018,
      "learning_rate": 0.0002068079644613722,
      "loss": 2.0342,
      "step": 138367
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.034784197807312,
      "learning_rate": 0.0002068040762777949,
      "loss": 2.2568,
      "step": 138368
    },
    {
      "epoch": 1.8,
      "grad_norm": 0.9754813313484192,
      "learning_rate": 0.0002068001881115445,
      "loss": 1.955,
      "step": 138369
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.1021690368652344,
      "learning_rate": 0.00020679629996262167,
      "loss": 2.1868,
      "step": 138370
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.045172929763794,
      "learning_rate": 0.00020679241183102715,
      "loss": 1.9728,
      "step": 138371
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.0623587369918823,
      "learning_rate": 0.00020678852371676173,
      "loss": 2.3451,
      "step": 138372
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.0049580335617065,
      "learning_rate": 0.0002067846356198262,
      "loss": 2.1517,
      "step": 138373
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.2231289148330688,
      "learning_rate": 0.00020678074754022102,
      "loss": 2.4219,
      "step": 138374
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.0479862689971924,
      "learning_rate": 0.00020677685947794705,
      "loss": 2.0506,
      "step": 138375
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.5331037044525146,
      "learning_rate": 0.00020677297143300503,
      "loss": 2.1568,
      "step": 138376
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.1315380334854126,
      "learning_rate": 0.00020676908340539572,
      "loss": 1.965,
      "step": 138377
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.1172008514404297,
      "learning_rate": 0.0002067651953951197,
      "loss": 1.9767,
      "step": 138378
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.0868409872055054,
      "learning_rate": 0.00020676130740217798,
      "loss": 2.1937,
      "step": 138379
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.2082328796386719,
      "learning_rate": 0.000206757419426571,
      "loss": 2.2741,
      "step": 138380
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.0968886613845825,
      "learning_rate": 0.00020675353146829952,
      "loss": 2.165,
      "step": 138381
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.0967954397201538,
      "learning_rate": 0.0002067496435273643,
      "loss": 2.0279,
      "step": 138382
    },
    {
      "epoch": 1.8,
      "grad_norm": 0.9997586607933044,
      "learning_rate": 0.00020674575560376612,
      "loss": 2.2856,
      "step": 138383
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.0675976276397705,
      "learning_rate": 0.00020674186769750568,
      "loss": 2.0826,
      "step": 138384
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.0599232912063599,
      "learning_rate": 0.00020673797980858376,
      "loss": 2.1119,
      "step": 138385
    },
    {
      "epoch": 1.8,
      "grad_norm": 0.9798392653465271,
      "learning_rate": 0.00020673409193700096,
      "loss": 1.9501,
      "step": 138386
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.0455453395843506,
      "learning_rate": 0.00020673020408275798,
      "loss": 2.1391,
      "step": 138387
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.1021323204040527,
      "learning_rate": 0.00020672631624585566,
      "loss": 2.1868,
      "step": 138388
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.127237319946289,
      "learning_rate": 0.00020672242842629466,
      "loss": 2.2089,
      "step": 138389
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.1027451753616333,
      "learning_rate": 0.00020671854062407572,
      "loss": 1.919,
      "step": 138390
    },
    {
      "epoch": 1.8,
      "grad_norm": 0.9960047602653503,
      "learning_rate": 0.0002067146528391996,
      "loss": 1.884,
      "step": 138391
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.0453208684921265,
      "learning_rate": 0.00020671076507166708,
      "loss": 2.159,
      "step": 138392
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.0539295673370361,
      "learning_rate": 0.00020670687732147864,
      "loss": 2.2011,
      "step": 138393
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.0888200998306274,
      "learning_rate": 0.00020670298958863517,
      "loss": 2.1828,
      "step": 138394
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.1038391590118408,
      "learning_rate": 0.0002066991018731374,
      "loss": 2.1189,
      "step": 138395
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.0796993970870972,
      "learning_rate": 0.000206695214174986,
      "loss": 2.0485,
      "step": 138396
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.1427135467529297,
      "learning_rate": 0.00020669132649418176,
      "loss": 2.1479,
      "step": 138397
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.081099510192871,
      "learning_rate": 0.00020668743883072541,
      "loss": 2.0472,
      "step": 138398
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.156640648841858,
      "learning_rate": 0.00020668355118461757,
      "loss": 1.9633,
      "step": 138399
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.0660319328308105,
      "learning_rate": 0.00020667966355585903,
      "loss": 2.28,
      "step": 138400
    },
    {
      "epoch": 1.8,
      "grad_norm": 0.9935824275016785,
      "learning_rate": 0.00020667577594445045,
      "loss": 2.0209,
      "step": 138401
    },
    {
      "epoch": 1.8,
      "grad_norm": 0.9724448323249817,
      "learning_rate": 0.00020667188835039266,
      "loss": 2.057,
      "step": 138402
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.0863465070724487,
      "learning_rate": 0.0002066680007736863,
      "loss": 2.1816,
      "step": 138403
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.0089055299758911,
      "learning_rate": 0.00020666411321433218,
      "loss": 2.072,
      "step": 138404
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.212837815284729,
      "learning_rate": 0.0002066602256723309,
      "loss": 2.2957,
      "step": 138405
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.1471022367477417,
      "learning_rate": 0.0002066563381476833,
      "loss": 2.102,
      "step": 138406
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.0302091836929321,
      "learning_rate": 0.00020665245064038998,
      "loss": 2.1797,
      "step": 138407
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.0337835550308228,
      "learning_rate": 0.00020664856315045176,
      "loss": 2.2027,
      "step": 138408
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.1478337049484253,
      "learning_rate": 0.0002066446756778693,
      "loss": 2.0267,
      "step": 138409
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.07166588306427,
      "learning_rate": 0.00020664078822264347,
      "loss": 2.0803,
      "step": 138410
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.0271297693252563,
      "learning_rate": 0.0002066369007847748,
      "loss": 2.0305,
      "step": 138411
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.0735279321670532,
      "learning_rate": 0.00020663301336426405,
      "loss": 2.1719,
      "step": 138412
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.0125455856323242,
      "learning_rate": 0.00020662912596111207,
      "loss": 2.1582,
      "step": 138413
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.0153048038482666,
      "learning_rate": 0.00020662523857531946,
      "loss": 1.9772,
      "step": 138414
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.0173171758651733,
      "learning_rate": 0.000206621351206887,
      "loss": 2.0162,
      "step": 138415
    },
    {
      "epoch": 1.8,
      "grad_norm": 0.9569740295410156,
      "learning_rate": 0.00020661746385581542,
      "loss": 2.117,
      "step": 138416
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.040016531944275,
      "learning_rate": 0.0002066135765221053,
      "loss": 2.0544,
      "step": 138417
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.0220178365707397,
      "learning_rate": 0.00020660968920575753,
      "loss": 2.2004,
      "step": 138418
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.1670539379119873,
      "learning_rate": 0.00020660580190677281,
      "loss": 2.1842,
      "step": 138419
    },
    {
      "epoch": 1.8,
      "grad_norm": 0.9847972989082336,
      "learning_rate": 0.00020660191462515184,
      "loss": 2.1377,
      "step": 138420
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.1308456659317017,
      "learning_rate": 0.00020659802736089534,
      "loss": 2.0535,
      "step": 138421
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.0421732664108276,
      "learning_rate": 0.00020659414011400403,
      "loss": 1.8771,
      "step": 138422
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.2563685178756714,
      "learning_rate": 0.00020659025288447864,
      "loss": 2.1547,
      "step": 138423
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.0760871171951294,
      "learning_rate": 0.00020658636567231982,
      "loss": 1.9743,
      "step": 138424
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.2213212251663208,
      "learning_rate": 0.0002065824784775284,
      "loss": 2.2047,
      "step": 138425
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.0928999185562134,
      "learning_rate": 0.00020657859130010506,
      "loss": 2.1282,
      "step": 138426
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.0607630014419556,
      "learning_rate": 0.0002065747041400505,
      "loss": 2.1221,
      "step": 138427
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.0765105485916138,
      "learning_rate": 0.00020657081699736556,
      "loss": 2.1526,
      "step": 138428
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.051640272140503,
      "learning_rate": 0.00020656692987205082,
      "loss": 2.2178,
      "step": 138429
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.2036324739456177,
      "learning_rate": 0.000206563042764107,
      "loss": 1.935,
      "step": 138430
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.11014986038208,
      "learning_rate": 0.00020655915567353487,
      "loss": 2.1722,
      "step": 138431
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.0783277750015259,
      "learning_rate": 0.0002065552686003352,
      "loss": 2.1941,
      "step": 138432
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.0491690635681152,
      "learning_rate": 0.0002065513815445086,
      "loss": 2.1837,
      "step": 138433
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.1608740091323853,
      "learning_rate": 0.00020654749450605604,
      "loss": 2.0918,
      "step": 138434
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.051232099533081,
      "learning_rate": 0.00020654360748497796,
      "loss": 2.1765,
      "step": 138435
    },
    {
      "epoch": 1.8,
      "grad_norm": 0.9662200808525085,
      "learning_rate": 0.0002065397204812751,
      "loss": 2.0711,
      "step": 138436
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.0258636474609375,
      "learning_rate": 0.00020653583349494837,
      "loss": 2.2214,
      "step": 138437
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.325466513633728,
      "learning_rate": 0.00020653194652599835,
      "loss": 2.0089,
      "step": 138438
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.0127148628234863,
      "learning_rate": 0.0002065280595744258,
      "loss": 2.1275,
      "step": 138439
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.0373485088348389,
      "learning_rate": 0.0002065241726402316,
      "loss": 2.1822,
      "step": 138440
    },
    {
      "epoch": 1.8,
      "grad_norm": 0.9598241448402405,
      "learning_rate": 0.00020652028572341614,
      "loss": 2.0508,
      "step": 138441
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.1280937194824219,
      "learning_rate": 0.00020651639882398036,
      "loss": 2.1138,
      "step": 138442
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.0420500040054321,
      "learning_rate": 0.00020651251194192495,
      "loss": 2.1769,
      "step": 138443
    },
    {
      "epoch": 1.8,
      "grad_norm": 0.9766277074813843,
      "learning_rate": 0.00020650862507725064,
      "loss": 2.1435,
      "step": 138444
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.005030632019043,
      "learning_rate": 0.0002065047382299581,
      "loss": 2.0731,
      "step": 138445
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.0638309717178345,
      "learning_rate": 0.0002065008514000483,
      "loss": 2.0015,
      "step": 138446
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.0690175294876099,
      "learning_rate": 0.0002064969645875215,
      "loss": 2.189,
      "step": 138447
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.0295062065124512,
      "learning_rate": 0.00020649307779237877,
      "loss": 1.9809,
      "step": 138448
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.1202641725540161,
      "learning_rate": 0.00020648919101462073,
      "loss": 2.0889,
      "step": 138449
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.0571938753128052,
      "learning_rate": 0.0002064853042542481,
      "loss": 2.1358,
      "step": 138450
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.068813681602478,
      "learning_rate": 0.00020648141751126165,
      "loss": 2.1197,
      "step": 138451
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.4536969661712646,
      "learning_rate": 0.00020647753078566216,
      "loss": 2.0425,
      "step": 138452
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.0270984172821045,
      "learning_rate": 0.00020647364407745016,
      "loss": 1.8739,
      "step": 138453
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.0086407661437988,
      "learning_rate": 0.00020646975738662646,
      "loss": 1.9587,
      "step": 138454
    },
    {
      "epoch": 1.8,
      "grad_norm": 0.96696937084198,
      "learning_rate": 0.0002064658707131918,
      "loss": 2.0887,
      "step": 138455
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.0562127828598022,
      "learning_rate": 0.00020646198405714693,
      "loss": 2.1043,
      "step": 138456
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.0558273792266846,
      "learning_rate": 0.00020645809741849248,
      "loss": 2.0278,
      "step": 138457
    },
    {
      "epoch": 1.8,
      "grad_norm": 0.9938949942588806,
      "learning_rate": 0.0002064542107972293,
      "loss": 2.207,
      "step": 138458
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.093624472618103,
      "learning_rate": 0.00020645032419335812,
      "loss": 2.0639,
      "step": 138459
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.023725152015686,
      "learning_rate": 0.0002064464376068795,
      "loss": 2.344,
      "step": 138460
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.0913496017456055,
      "learning_rate": 0.00020644255103779424,
      "loss": 2.1213,
      "step": 138461
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.0158652067184448,
      "learning_rate": 0.00020643866448610308,
      "loss": 2.0424,
      "step": 138462
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.12621009349823,
      "learning_rate": 0.00020643477795180676,
      "loss": 1.8478,
      "step": 138463
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.0913622379302979,
      "learning_rate": 0.00020643089143490591,
      "loss": 2.1124,
      "step": 138464
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.0621429681777954,
      "learning_rate": 0.00020642700493540153,
      "loss": 2.1013,
      "step": 138465
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.077795147895813,
      "learning_rate": 0.00020642311845329395,
      "loss": 1.9044,
      "step": 138466
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.0249782800674438,
      "learning_rate": 0.00020641923198858414,
      "loss": 2.0529,
      "step": 138467
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.1558029651641846,
      "learning_rate": 0.00020641534554127273,
      "loss": 2.0331,
      "step": 138468
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.0065432786941528,
      "learning_rate": 0.00020641145911136046,
      "loss": 1.9395,
      "step": 138469
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.0392358303070068,
      "learning_rate": 0.0002064075726988481,
      "loss": 2.1779,
      "step": 138470
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.1399914026260376,
      "learning_rate": 0.00020640368630373644,
      "loss": 2.1183,
      "step": 138471
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.114890456199646,
      "learning_rate": 0.00020639979992602599,
      "loss": 2.2636,
      "step": 138472
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.0263866186141968,
      "learning_rate": 0.00020639591356571755,
      "loss": 2.0236,
      "step": 138473
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.0312979221343994,
      "learning_rate": 0.00020639202722281188,
      "loss": 2.081,
      "step": 138474
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.109418511390686,
      "learning_rate": 0.00020638814089730972,
      "loss": 1.9931,
      "step": 138475
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.062349796295166,
      "learning_rate": 0.00020638425458921177,
      "loss": 2.0532,
      "step": 138476
    },
    {
      "epoch": 1.8,
      "grad_norm": 0.9960459470748901,
      "learning_rate": 0.0002063803682985189,
      "loss": 1.927,
      "step": 138477
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.0867934226989746,
      "learning_rate": 0.00020637648202523153,
      "loss": 2.1268,
      "step": 138478
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.0905177593231201,
      "learning_rate": 0.00020637259576935057,
      "loss": 1.9902,
      "step": 138479
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.052965521812439,
      "learning_rate": 0.00020636870953087667,
      "loss": 2.1228,
      "step": 138480
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.0519129037857056,
      "learning_rate": 0.00020636482330981063,
      "loss": 2.1939,
      "step": 138481
    },
    {
      "epoch": 1.8,
      "grad_norm": 0.9882078766822815,
      "learning_rate": 0.00020636093710615312,
      "loss": 2.1526,
      "step": 138482
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.178515911102295,
      "learning_rate": 0.000206357050919905,
      "loss": 1.8955,
      "step": 138483
    },
    {
      "epoch": 1.8,
      "grad_norm": 0.9666233062744141,
      "learning_rate": 0.00020635316475106673,
      "loss": 2.0178,
      "step": 138484
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.0314724445343018,
      "learning_rate": 0.00020634927859963922,
      "loss": 2.268,
      "step": 138485
    },
    {
      "epoch": 1.8,
      "grad_norm": 0.9964719414710999,
      "learning_rate": 0.00020634539246562313,
      "loss": 2.2128,
      "step": 138486
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.029957890510559,
      "learning_rate": 0.0002063415063490192,
      "loss": 2.2909,
      "step": 138487
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.058275818824768,
      "learning_rate": 0.00020633762024982813,
      "loss": 2.0477,
      "step": 138488
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.0633747577667236,
      "learning_rate": 0.00020633373416805074,
      "loss": 2.1651,
      "step": 138489
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.0756630897521973,
      "learning_rate": 0.00020632984810368764,
      "loss": 2.1443,
      "step": 138490
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.0681664943695068,
      "learning_rate": 0.00020632596205673963,
      "loss": 2.1373,
      "step": 138491
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.151096224784851,
      "learning_rate": 0.0002063220760272073,
      "loss": 2.2218,
      "step": 138492
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.4346256256103516,
      "learning_rate": 0.00020631819001509148,
      "loss": 2.0999,
      "step": 138493
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.0801270008087158,
      "learning_rate": 0.00020631430402039293,
      "loss": 1.8888,
      "step": 138494
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.185219407081604,
      "learning_rate": 0.00020631041804311233,
      "loss": 2.1424,
      "step": 138495
    },
    {
      "epoch": 1.8,
      "grad_norm": 0.9965805411338806,
      "learning_rate": 0.00020630653208325033,
      "loss": 2.0262,
      "step": 138496
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.2150744199752808,
      "learning_rate": 0.00020630264614080772,
      "loss": 2.1709,
      "step": 138497
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.5324015617370605,
      "learning_rate": 0.00020629876021578528,
      "loss": 2.226,
      "step": 138498
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.1599141359329224,
      "learning_rate": 0.00020629487430818365,
      "loss": 2.0307,
      "step": 138499
    },
    {
      "epoch": 1.8,
      "grad_norm": 0.9994481801986694,
      "learning_rate": 0.0002062909884180035,
      "loss": 1.9587,
      "step": 138500
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.2973668575286865,
      "learning_rate": 0.00020628710254524572,
      "loss": 2.1759,
      "step": 138501
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.098453164100647,
      "learning_rate": 0.0002062832166899109,
      "loss": 2.1478,
      "step": 138502
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.2998898029327393,
      "learning_rate": 0.00020627933085199976,
      "loss": 2.2365,
      "step": 138503
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.2388136386871338,
      "learning_rate": 0.0002062754450315131,
      "loss": 2.0484,
      "step": 138504
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.0373953580856323,
      "learning_rate": 0.00020627155922845164,
      "loss": 2.1401,
      "step": 138505
    },
    {
      "epoch": 1.8,
      "grad_norm": 0.9849302768707275,
      "learning_rate": 0.000206267673442816,
      "loss": 2.1374,
      "step": 138506
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.213712453842163,
      "learning_rate": 0.000206263787674607,
      "loss": 2.2207,
      "step": 138507
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.1238749027252197,
      "learning_rate": 0.00020625990192382534,
      "loss": 2.0103,
      "step": 138508
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.007620096206665,
      "learning_rate": 0.0002062560161904717,
      "loss": 1.9905,
      "step": 138509
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.0273185968399048,
      "learning_rate": 0.00020625213047454683,
      "loss": 2.0607,
      "step": 138510
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.1072697639465332,
      "learning_rate": 0.00020624824477605147,
      "loss": 2.1486,
      "step": 138511
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.6245073080062866,
      "learning_rate": 0.00020624435909498634,
      "loss": 2.0424,
      "step": 138512
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.0226161479949951,
      "learning_rate": 0.00020624047343135223,
      "loss": 2.0324,
      "step": 138513
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.291815161705017,
      "learning_rate": 0.0002062365877851497,
      "loss": 2.0455,
      "step": 138514
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.267126441001892,
      "learning_rate": 0.00020623270215637958,
      "loss": 2.0743,
      "step": 138515
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.0198532342910767,
      "learning_rate": 0.00020622881654504252,
      "loss": 2.1552,
      "step": 138516
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.1353201866149902,
      "learning_rate": 0.00020622493095113933,
      "loss": 2.1966,
      "step": 138517
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.0239453315734863,
      "learning_rate": 0.00020622104537467069,
      "loss": 2.3111,
      "step": 138518
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.0833778381347656,
      "learning_rate": 0.00020621715981563742,
      "loss": 1.9651,
      "step": 138519
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.0904841423034668,
      "learning_rate": 0.00020621327427404004,
      "loss": 2.1776,
      "step": 138520
    },
    {
      "epoch": 1.8,
      "grad_norm": 0.9566026926040649,
      "learning_rate": 0.0002062093887498794,
      "loss": 2.1941,
      "step": 138521
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.245418667793274,
      "learning_rate": 0.0002062055032431562,
      "loss": 2.0573,
      "step": 138522
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.041397213935852,
      "learning_rate": 0.00020620161775387118,
      "loss": 2.2115,
      "step": 138523
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.0561541318893433,
      "learning_rate": 0.000206197732282025,
      "loss": 1.9671,
      "step": 138524
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.0466809272766113,
      "learning_rate": 0.0002061938468276185,
      "loss": 1.983,
      "step": 138525
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.145892858505249,
      "learning_rate": 0.0002061899613906524,
      "loss": 2.0077,
      "step": 138526
    },
    {
      "epoch": 1.8,
      "grad_norm": 0.9361758828163147,
      "learning_rate": 0.00020618607597112728,
      "loss": 2.2439,
      "step": 138527
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.087489366531372,
      "learning_rate": 0.0002061821905690439,
      "loss": 2.2188,
      "step": 138528
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.0784460306167603,
      "learning_rate": 0.00020617830518440306,
      "loss": 2.0128,
      "step": 138529
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.0309672355651855,
      "learning_rate": 0.0002061744198172054,
      "loss": 2.111,
      "step": 138530
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.087524175643921,
      "learning_rate": 0.00020617053446745173,
      "loss": 1.8127,
      "step": 138531
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.1322379112243652,
      "learning_rate": 0.0002061666491351428,
      "loss": 2.3877,
      "step": 138532
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.186061143875122,
      "learning_rate": 0.00020616276382027918,
      "loss": 2.0017,
      "step": 138533
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.0136981010437012,
      "learning_rate": 0.0002061588785228617,
      "loss": 1.9904,
      "step": 138534
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.0399272441864014,
      "learning_rate": 0.00020615499324289102,
      "loss": 2.1899,
      "step": 138535
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.1222918033599854,
      "learning_rate": 0.00020615110798036788,
      "loss": 2.1009,
      "step": 138536
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.0182095766067505,
      "learning_rate": 0.00020614722273529305,
      "loss": 2.1643,
      "step": 138537
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.0523340702056885,
      "learning_rate": 0.00020614333750766732,
      "loss": 2.0094,
      "step": 138538
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.096734642982483,
      "learning_rate": 0.00020613945229749122,
      "loss": 2.0991,
      "step": 138539
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.015668272972107,
      "learning_rate": 0.00020613556710476555,
      "loss": 2.1908,
      "step": 138540
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.1838268041610718,
      "learning_rate": 0.00020613168192949103,
      "loss": 2.0378,
      "step": 138541
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.2432833909988403,
      "learning_rate": 0.0002061277967716684,
      "loss": 2.053,
      "step": 138542
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.1657058000564575,
      "learning_rate": 0.00020612391163129848,
      "loss": 2.0033,
      "step": 138543
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.0522830486297607,
      "learning_rate": 0.00020612002650838193,
      "loss": 1.9229,
      "step": 138544
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.0232510566711426,
      "learning_rate": 0.00020611614140291935,
      "loss": 2.0312,
      "step": 138545
    },
    {
      "epoch": 1.8,
      "grad_norm": 0.9851943850517273,
      "learning_rate": 0.0002061122563149115,
      "loss": 1.9798,
      "step": 138546
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.0612163543701172,
      "learning_rate": 0.00020610837124435921,
      "loss": 2.1568,
      "step": 138547
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.0170365571975708,
      "learning_rate": 0.00020610448619126314,
      "loss": 2.038,
      "step": 138548
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.04971444606781,
      "learning_rate": 0.000206100601155624,
      "loss": 2.0836,
      "step": 138549
    },
    {
      "epoch": 1.8,
      "grad_norm": 0.9950829744338989,
      "learning_rate": 0.00020609671613744265,
      "loss": 2.0786,
      "step": 138550
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.0280091762542725,
      "learning_rate": 0.0002060928311367196,
      "loss": 1.9788,
      "step": 138551
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.107316493988037,
      "learning_rate": 0.00020608894615345563,
      "loss": 1.9441,
      "step": 138552
    },
    {
      "epoch": 1.8,
      "grad_norm": 0.9968172311782837,
      "learning_rate": 0.00020608506118765152,
      "loss": 2.037,
      "step": 138553
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.0215498208999634,
      "learning_rate": 0.00020608117623930795,
      "loss": 2.2149,
      "step": 138554
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.0008608102798462,
      "learning_rate": 0.0002060772913084257,
      "loss": 2.1275,
      "step": 138555
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.1643803119659424,
      "learning_rate": 0.00020607340639500554,
      "loss": 2.164,
      "step": 138556
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.1054108142852783,
      "learning_rate": 0.00020606952149904801,
      "loss": 2.0027,
      "step": 138557
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.0517306327819824,
      "learning_rate": 0.00020606563662055392,
      "loss": 1.9457,
      "step": 138558
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.1671956777572632,
      "learning_rate": 0.00020606175175952397,
      "loss": 1.9831,
      "step": 138559
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.2156940698623657,
      "learning_rate": 0.00020605786691595894,
      "loss": 2.2062,
      "step": 138560
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.014232873916626,
      "learning_rate": 0.00020605398208985954,
      "loss": 2.1859,
      "step": 138561
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.1247845888137817,
      "learning_rate": 0.0002060500972812266,
      "loss": 2.2198,
      "step": 138562
    },
    {
      "epoch": 1.8,
      "grad_norm": 0.9978877902030945,
      "learning_rate": 0.00020604621249006058,
      "loss": 2.0578,
      "step": 138563
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.2956682443618774,
      "learning_rate": 0.00020604232771636232,
      "loss": 1.873,
      "step": 138564
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.1051127910614014,
      "learning_rate": 0.00020603844296013263,
      "loss": 2.0323,
      "step": 138565
    },
    {
      "epoch": 1.8,
      "grad_norm": 0.9933516979217529,
      "learning_rate": 0.00020603455822137208,
      "loss": 1.9362,
      "step": 138566
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.1605329513549805,
      "learning_rate": 0.00020603067350008158,
      "loss": 2.2304,
      "step": 138567
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.2278952598571777,
      "learning_rate": 0.0002060267887962618,
      "loss": 2.101,
      "step": 138568
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.138856053352356,
      "learning_rate": 0.00020602290410991333,
      "loss": 2.3159,
      "step": 138569
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.1087777614593506,
      "learning_rate": 0.00020601901944103694,
      "loss": 2.2336,
      "step": 138570
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.073315143585205,
      "learning_rate": 0.00020601513478963343,
      "loss": 2.1257,
      "step": 138571
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.1132005453109741,
      "learning_rate": 0.00020601125015570345,
      "loss": 2.016,
      "step": 138572
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.0574214458465576,
      "learning_rate": 0.00020600736553924776,
      "loss": 2.2125,
      "step": 138573
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.0571637153625488,
      "learning_rate": 0.00020600348094026714,
      "loss": 2.1682,
      "step": 138574
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.1535260677337646,
      "learning_rate": 0.00020599959635876224,
      "loss": 2.0643,
      "step": 138575
    },
    {
      "epoch": 1.8,
      "grad_norm": 0.952433168888092,
      "learning_rate": 0.00020599571179473372,
      "loss": 1.958,
      "step": 138576
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.1410279273986816,
      "learning_rate": 0.0002059918272481824,
      "loss": 2.0356,
      "step": 138577
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.0528905391693115,
      "learning_rate": 0.00020598794271910897,
      "loss": 2.2297,
      "step": 138578
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.1528706550598145,
      "learning_rate": 0.00020598405820751415,
      "loss": 2.196,
      "step": 138579
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.0391839742660522,
      "learning_rate": 0.0002059801737133987,
      "loss": 2.1139,
      "step": 138580
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.1378790140151978,
      "learning_rate": 0.0002059762892367633,
      "loss": 1.8993,
      "step": 138581
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.1362475156784058,
      "learning_rate": 0.00020597240477760866,
      "loss": 2.2079,
      "step": 138582
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.0103800296783447,
      "learning_rate": 0.00020596852033593558,
      "loss": 1.8118,
      "step": 138583
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.0670472383499146,
      "learning_rate": 0.00020596463591174464,
      "loss": 2.338,
      "step": 138584
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.3044602870941162,
      "learning_rate": 0.00020596075150503668,
      "loss": 2.1197,
      "step": 138585
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.0664788484573364,
      "learning_rate": 0.00020595686711581248,
      "loss": 1.9366,
      "step": 138586
    },
    {
      "epoch": 1.8,
      "grad_norm": 0.9269299507141113,
      "learning_rate": 0.00020595298274407256,
      "loss": 1.9248,
      "step": 138587
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.2632840871810913,
      "learning_rate": 0.00020594909838981778,
      "loss": 2.1258,
      "step": 138588
    },
    {
      "epoch": 1.8,
      "grad_norm": 0.9575965404510498,
      "learning_rate": 0.00020594521405304888,
      "loss": 2.0507,
      "step": 138589
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.1182152032852173,
      "learning_rate": 0.0002059413297337665,
      "loss": 1.9041,
      "step": 138590
    },
    {
      "epoch": 1.8,
      "grad_norm": 0.9840338826179504,
      "learning_rate": 0.00020593744543197145,
      "loss": 2.1265,
      "step": 138591
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.0247489213943481,
      "learning_rate": 0.0002059335611476643,
      "loss": 2.1837,
      "step": 138592
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.156916856765747,
      "learning_rate": 0.000205929676880846,
      "loss": 2.1023,
      "step": 138593
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.264999508857727,
      "learning_rate": 0.0002059257926315171,
      "loss": 2.2485,
      "step": 138594
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.057362675666809,
      "learning_rate": 0.00020592190839967836,
      "loss": 2.006,
      "step": 138595
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.167444109916687,
      "learning_rate": 0.00020591802418533047,
      "loss": 2.2675,
      "step": 138596
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.184036135673523,
      "learning_rate": 0.00020591413998847434,
      "loss": 2.0587,
      "step": 138597
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.1649222373962402,
      "learning_rate": 0.00020591025580911042,
      "loss": 2.1519,
      "step": 138598
    },
    {
      "epoch": 1.8,
      "grad_norm": 0.9998751282691956,
      "learning_rate": 0.00020590637164723965,
      "loss": 2.0192,
      "step": 138599
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.1791306734085083,
      "learning_rate": 0.0002059024875028626,
      "loss": 2.1539,
      "step": 138600
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.1747058629989624,
      "learning_rate": 0.00020589860337598,
      "loss": 2.2196,
      "step": 138601
    },
    {
      "epoch": 1.8,
      "grad_norm": 0.899780809879303,
      "learning_rate": 0.00020589471926659266,
      "loss": 2.1022,
      "step": 138602
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.16359281539917,
      "learning_rate": 0.00020589083517470128,
      "loss": 2.0763,
      "step": 138603
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.04237699508667,
      "learning_rate": 0.00020588695110030665,
      "loss": 2.1552,
      "step": 138604
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.100264549255371,
      "learning_rate": 0.00020588306704340936,
      "loss": 1.9695,
      "step": 138605
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.2109594345092773,
      "learning_rate": 0.00020587918300401015,
      "loss": 1.9566,
      "step": 138606
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.1527962684631348,
      "learning_rate": 0.0002058752989821098,
      "loss": 2.0292,
      "step": 138607
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.339019775390625,
      "learning_rate": 0.00020587141497770896,
      "loss": 2.1709,
      "step": 138608
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.3069815635681152,
      "learning_rate": 0.00020586753099080846,
      "loss": 2.0619,
      "step": 138609
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.0670369863510132,
      "learning_rate": 0.0002058636470214089,
      "loss": 2.1833,
      "step": 138610
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.0104652643203735,
      "learning_rate": 0.00020585976306951122,
      "loss": 1.9275,
      "step": 138611
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.0151097774505615,
      "learning_rate": 0.0002058558791351159,
      "loss": 1.9067,
      "step": 138612
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.311342477798462,
      "learning_rate": 0.00020585199521822367,
      "loss": 2.1695,
      "step": 138613
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.0618494749069214,
      "learning_rate": 0.0002058481113188354,
      "loss": 2.2813,
      "step": 138614
    },
    {
      "epoch": 1.8,
      "grad_norm": 0.9901725053787231,
      "learning_rate": 0.00020584422743695173,
      "loss": 2.026,
      "step": 138615
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.0050549507141113,
      "learning_rate": 0.00020584034357257335,
      "loss": 2.2195,
      "step": 138616
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.0352301597595215,
      "learning_rate": 0.00020583645972570118,
      "loss": 2.1234,
      "step": 138617
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.025681972503662,
      "learning_rate": 0.00020583257589633567,
      "loss": 2.261,
      "step": 138618
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.0782314538955688,
      "learning_rate": 0.00020582869208447767,
      "loss": 2.0784,
      "step": 138619
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.0314573049545288,
      "learning_rate": 0.00020582480829012788,
      "loss": 2.0772,
      "step": 138620
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.086722493171692,
      "learning_rate": 0.00020582092451328703,
      "loss": 2.3177,
      "step": 138621
    },
    {
      "epoch": 1.8,
      "grad_norm": 0.9993278980255127,
      "learning_rate": 0.00020581704075395589,
      "loss": 2.0028,
      "step": 138622
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.116876482963562,
      "learning_rate": 0.0002058131570121352,
      "loss": 2.0861,
      "step": 138623
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.1950076818466187,
      "learning_rate": 0.00020580927328782552,
      "loss": 2.1447,
      "step": 138624
    },
    {
      "epoch": 1.8,
      "grad_norm": 1.0722535848617554,
      "learning_rate": 0.00020580538958102771,
      "loss": 1.9955,
      "step": 138625
    },
    {
      "epoch": 1.81,
      "grad_norm": 0.9462834000587463,
      "learning_rate": 0.00020580150589174243,
      "loss": 2.2662,
      "step": 138626
    },
    {
      "epoch": 1.81,
      "grad_norm": 0.9596506953239441,
      "learning_rate": 0.00020579762221997042,
      "loss": 2.0682,
      "step": 138627
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.10152268409729,
      "learning_rate": 0.00020579373856571245,
      "loss": 2.0552,
      "step": 138628
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.0936155319213867,
      "learning_rate": 0.00020578985492896928,
      "loss": 2.0507,
      "step": 138629
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.0430006980895996,
      "learning_rate": 0.00020578597130974143,
      "loss": 2.0419,
      "step": 138630
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.144731879234314,
      "learning_rate": 0.00020578208770802975,
      "loss": 2.0461,
      "step": 138631
    },
    {
      "epoch": 1.81,
      "grad_norm": 0.9521109461784363,
      "learning_rate": 0.000205778204123835,
      "loss": 1.9677,
      "step": 138632
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.075260043144226,
      "learning_rate": 0.00020577432055715782,
      "loss": 2.065,
      "step": 138633
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.1001464128494263,
      "learning_rate": 0.000205770437007999,
      "loss": 2.3412,
      "step": 138634
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.2736632823944092,
      "learning_rate": 0.00020576655347635933,
      "loss": 2.0963,
      "step": 138635
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.044999122619629,
      "learning_rate": 0.00020576266996223934,
      "loss": 2.1003,
      "step": 138636
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.0579044818878174,
      "learning_rate": 0.00020575878646563982,
      "loss": 2.154,
      "step": 138637
    },
    {
      "epoch": 1.81,
      "grad_norm": 0.9766268134117126,
      "learning_rate": 0.00020575490298656152,
      "loss": 1.9142,
      "step": 138638
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.0148591995239258,
      "learning_rate": 0.00020575101952500518,
      "loss": 1.8981,
      "step": 138639
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.1752896308898926,
      "learning_rate": 0.00020574713608097152,
      "loss": 2.1145,
      "step": 138640
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.0570861101150513,
      "learning_rate": 0.00020574325265446136,
      "loss": 2.1075,
      "step": 138641
    },
    {
      "epoch": 1.81,
      "grad_norm": 0.994012713432312,
      "learning_rate": 0.00020573936924547516,
      "loss": 1.9486,
      "step": 138642
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.0565078258514404,
      "learning_rate": 0.00020573548585401378,
      "loss": 2.1691,
      "step": 138643
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.1137648820877075,
      "learning_rate": 0.000205731602480078,
      "loss": 1.9109,
      "step": 138644
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.0136526823043823,
      "learning_rate": 0.00020572771912366847,
      "loss": 2.2785,
      "step": 138645
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.0410977602005005,
      "learning_rate": 0.0002057238357847859,
      "loss": 1.9848,
      "step": 138646
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.003189206123352,
      "learning_rate": 0.00020571995246343122,
      "loss": 2.0593,
      "step": 138647
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.079756259918213,
      "learning_rate": 0.00020571606915960485,
      "loss": 2.289,
      "step": 138648
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.083206295967102,
      "learning_rate": 0.00020571218587330767,
      "loss": 2.0768,
      "step": 138649
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.154146432876587,
      "learning_rate": 0.00020570830260454033,
      "loss": 2.4266,
      "step": 138650
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.138149619102478,
      "learning_rate": 0.00020570441935330362,
      "loss": 1.9979,
      "step": 138651
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.224366307258606,
      "learning_rate": 0.00020570053611959822,
      "loss": 1.9906,
      "step": 138652
    },
    {
      "epoch": 1.81,
      "grad_norm": 0.9904806017875671,
      "learning_rate": 0.000205696652903425,
      "loss": 2.0068,
      "step": 138653
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.0632555484771729,
      "learning_rate": 0.00020569276970478444,
      "loss": 2.0427,
      "step": 138654
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.0434064865112305,
      "learning_rate": 0.00020568888652367737,
      "loss": 2.2195,
      "step": 138655
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.0465856790542603,
      "learning_rate": 0.00020568500336010453,
      "loss": 1.9258,
      "step": 138656
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.1024950742721558,
      "learning_rate": 0.00020568112021406663,
      "loss": 2.0974,
      "step": 138657
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.076361894607544,
      "learning_rate": 0.00020567723708556435,
      "loss": 1.9607,
      "step": 138658
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.24151611328125,
      "learning_rate": 0.00020567335397459846,
      "loss": 1.9908,
      "step": 138659
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.1926157474517822,
      "learning_rate": 0.00020566947088116982,
      "loss": 2.1764,
      "step": 138660
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.2408010959625244,
      "learning_rate": 0.00020566558780527887,
      "loss": 2.0612,
      "step": 138661
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.2716591358184814,
      "learning_rate": 0.00020566170474692647,
      "loss": 2.0978,
      "step": 138662
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.1794353723526,
      "learning_rate": 0.00020565782170611335,
      "loss": 2.1614,
      "step": 138663
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.181209921836853,
      "learning_rate": 0.0002056539386828402,
      "loss": 2.0596,
      "step": 138664
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.0084344148635864,
      "learning_rate": 0.00020565005567710777,
      "loss": 2.1504,
      "step": 138665
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.00628662109375,
      "learning_rate": 0.00020564617268891685,
      "loss": 2.083,
      "step": 138666
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.269126534461975,
      "learning_rate": 0.00020564228971826811,
      "loss": 1.9945,
      "step": 138667
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.1881481409072876,
      "learning_rate": 0.00020563840676516212,
      "loss": 2.1205,
      "step": 138668
    },
    {
      "epoch": 1.81,
      "grad_norm": 0.9673976302146912,
      "learning_rate": 0.00020563452382959976,
      "loss": 1.9797,
      "step": 138669
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.0011656284332275,
      "learning_rate": 0.0002056306409115818,
      "loss": 1.6754,
      "step": 138670
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.1756043434143066,
      "learning_rate": 0.00020562675801110878,
      "loss": 1.9657,
      "step": 138671
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.1196904182434082,
      "learning_rate": 0.00020562287512818165,
      "loss": 2.2414,
      "step": 138672
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.0932930707931519,
      "learning_rate": 0.0002056189922628009,
      "loss": 2.0937,
      "step": 138673
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.0065630674362183,
      "learning_rate": 0.00020561510941496746,
      "loss": 2.0164,
      "step": 138674
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.1452661752700806,
      "learning_rate": 0.00020561122658468186,
      "loss": 2.1913,
      "step": 138675
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.2238181829452515,
      "learning_rate": 0.00020560734377194495,
      "loss": 1.9592,
      "step": 138676
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.0298303365707397,
      "learning_rate": 0.0002056034609767574,
      "loss": 2.2437,
      "step": 138677
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.1428964138031006,
      "learning_rate": 0.00020559957819912003,
      "loss": 2.0944,
      "step": 138678
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.1516309976577759,
      "learning_rate": 0.00020559569543903339,
      "loss": 2.0791,
      "step": 138679
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.0311702489852905,
      "learning_rate": 0.0002055918126964983,
      "loss": 1.9047,
      "step": 138680
    },
    {
      "epoch": 1.81,
      "grad_norm": 0.9257798194885254,
      "learning_rate": 0.0002055879299715155,
      "loss": 2.0875,
      "step": 138681
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.0300109386444092,
      "learning_rate": 0.00020558404726408571,
      "loss": 2.1486,
      "step": 138682
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.0401694774627686,
      "learning_rate": 0.0002055801645742096,
      "loss": 2.1799,
      "step": 138683
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.1210213899612427,
      "learning_rate": 0.00020557628190188794,
      "loss": 2.0837,
      "step": 138684
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.068913459777832,
      "learning_rate": 0.00020557239924712137,
      "loss": 2.4028,
      "step": 138685
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.1291016340255737,
      "learning_rate": 0.0002055685166099107,
      "loss": 2.1018,
      "step": 138686
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.0099211931228638,
      "learning_rate": 0.00020556463399025665,
      "loss": 1.8984,
      "step": 138687
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.048914909362793,
      "learning_rate": 0.0002055607513881599,
      "loss": 2.127,
      "step": 138688
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.1502374410629272,
      "learning_rate": 0.00020555686880362122,
      "loss": 2.1231,
      "step": 138689
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.1788541078567505,
      "learning_rate": 0.0002055529862366413,
      "loss": 2.2467,
      "step": 138690
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.1583034992218018,
      "learning_rate": 0.00020554910368722083,
      "loss": 2.1628,
      "step": 138691
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.0448311567306519,
      "learning_rate": 0.00020554522115536056,
      "loss": 2.1672,
      "step": 138692
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.105906367301941,
      "learning_rate": 0.00020554133864106119,
      "loss": 2.0002,
      "step": 138693
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.0176093578338623,
      "learning_rate": 0.00020553745614432348,
      "loss": 2.1144,
      "step": 138694
    },
    {
      "epoch": 1.81,
      "grad_norm": 0.9330503940582275,
      "learning_rate": 0.00020553357366514817,
      "loss": 1.9116,
      "step": 138695
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.0700340270996094,
      "learning_rate": 0.00020552969120353606,
      "loss": 2.133,
      "step": 138696
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.0725386142730713,
      "learning_rate": 0.0002055258087594876,
      "loss": 2.1253,
      "step": 138697
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.0113413333892822,
      "learning_rate": 0.00020552192633300375,
      "loss": 2.0461,
      "step": 138698
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.0677021741867065,
      "learning_rate": 0.0002055180439240851,
      "loss": 2.1173,
      "step": 138699
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.0218621492385864,
      "learning_rate": 0.00020551416153273246,
      "loss": 2.1191,
      "step": 138700
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.0315145254135132,
      "learning_rate": 0.00020551027915894648,
      "loss": 1.967,
      "step": 138701
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.0146141052246094,
      "learning_rate": 0.00020550639680272812,
      "loss": 2.116,
      "step": 138702
    },
    {
      "epoch": 1.81,
      "grad_norm": 0.9985074996948242,
      "learning_rate": 0.00020550251446407771,
      "loss": 2.1402,
      "step": 138703
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.1326911449432373,
      "learning_rate": 0.0002054986321429962,
      "loss": 2.0137,
      "step": 138704
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.0149892568588257,
      "learning_rate": 0.00020549474983948425,
      "loss": 1.771,
      "step": 138705
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.099509835243225,
      "learning_rate": 0.00020549086755354264,
      "loss": 2.0066,
      "step": 138706
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.0011155605316162,
      "learning_rate": 0.00020548698528517203,
      "loss": 1.9831,
      "step": 138707
    },
    {
      "epoch": 1.81,
      "grad_norm": 0.9996280670166016,
      "learning_rate": 0.0002054831030343733,
      "loss": 2.1912,
      "step": 138708
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.0530331134796143,
      "learning_rate": 0.00020547922080114695,
      "loss": 1.9214,
      "step": 138709
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.16014564037323,
      "learning_rate": 0.0002054753385854938,
      "loss": 2.0812,
      "step": 138710
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.101035475730896,
      "learning_rate": 0.00020547145638741458,
      "loss": 1.9992,
      "step": 138711
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.0059343576431274,
      "learning_rate": 0.00020546757420690997,
      "loss": 1.9911,
      "step": 138712
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.0222110748291016,
      "learning_rate": 0.00020546369204398074,
      "loss": 2.0812,
      "step": 138713
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.1092922687530518,
      "learning_rate": 0.0002054598098986277,
      "loss": 2.0804,
      "step": 138714
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.126468300819397,
      "learning_rate": 0.0002054559277708513,
      "loss": 2.2773,
      "step": 138715
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.0679070949554443,
      "learning_rate": 0.00020545204566065252,
      "loss": 2.0953,
      "step": 138716
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.1639080047607422,
      "learning_rate": 0.00020544816356803195,
      "loss": 2.2469,
      "step": 138717
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.1982958316802979,
      "learning_rate": 0.00020544428149299034,
      "loss": 2.1409,
      "step": 138718
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.0096995830535889,
      "learning_rate": 0.00020544039943552842,
      "loss": 2.388,
      "step": 138719
    },
    {
      "epoch": 1.81,
      "grad_norm": 0.9679723381996155,
      "learning_rate": 0.00020543651739564703,
      "loss": 2.1024,
      "step": 138720
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.0479801893234253,
      "learning_rate": 0.00020543263537334668,
      "loss": 2.198,
      "step": 138721
    },
    {
      "epoch": 1.81,
      "grad_norm": 0.9660943746566772,
      "learning_rate": 0.0002054287533686282,
      "loss": 2.0945,
      "step": 138722
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.0247480869293213,
      "learning_rate": 0.00020542487138149228,
      "loss": 2.0667,
      "step": 138723
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.1298792362213135,
      "learning_rate": 0.00020542098941193966,
      "loss": 2.1349,
      "step": 138724
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.0418715476989746,
      "learning_rate": 0.00020541710745997105,
      "loss": 2.0258,
      "step": 138725
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.1508573293685913,
      "learning_rate": 0.0002054132255255872,
      "loss": 2.2872,
      "step": 138726
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.0446656942367554,
      "learning_rate": 0.00020540934360878895,
      "loss": 2.1004,
      "step": 138727
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.1316999197006226,
      "learning_rate": 0.00020540546170957674,
      "loss": 2.0558,
      "step": 138728
    },
    {
      "epoch": 1.81,
      "grad_norm": 0.9553704857826233,
      "learning_rate": 0.00020540157982795146,
      "loss": 2.1149,
      "step": 138729
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.0555380582809448,
      "learning_rate": 0.0002053976979639138,
      "loss": 1.9224,
      "step": 138730
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.0368940830230713,
      "learning_rate": 0.0002053938161174645,
      "loss": 2.0167,
      "step": 138731
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.1453967094421387,
      "learning_rate": 0.0002053899342886043,
      "loss": 2.151,
      "step": 138732
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.1291459798812866,
      "learning_rate": 0.000205386052477334,
      "loss": 2.3012,
      "step": 138733
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.0040762424468994,
      "learning_rate": 0.0002053821706836541,
      "loss": 2.0162,
      "step": 138734
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.0872889757156372,
      "learning_rate": 0.0002053782889075654,
      "loss": 1.8592,
      "step": 138735
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.189373254776001,
      "learning_rate": 0.0002053744071490687,
      "loss": 2.0386,
      "step": 138736
    },
    {
      "epoch": 1.81,
      "grad_norm": 0.9570783972740173,
      "learning_rate": 0.0002053705254081647,
      "loss": 1.9823,
      "step": 138737
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.0675764083862305,
      "learning_rate": 0.00020536664368485407,
      "loss": 1.9082,
      "step": 138738
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.1082836389541626,
      "learning_rate": 0.00020536276197913772,
      "loss": 1.9508,
      "step": 138739
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.1412746906280518,
      "learning_rate": 0.00020535888029101607,
      "loss": 2.0876,
      "step": 138740
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.1102162599563599,
      "learning_rate": 0.00020535499862049,
      "loss": 2.1366,
      "step": 138741
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.0919541120529175,
      "learning_rate": 0.00020535111696756022,
      "loss": 2.1117,
      "step": 138742
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.0982317924499512,
      "learning_rate": 0.00020534723533222743,
      "loss": 2.041,
      "step": 138743
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.082684874534607,
      "learning_rate": 0.00020534335371449243,
      "loss": 2.2284,
      "step": 138744
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.010299563407898,
      "learning_rate": 0.00020533947211435599,
      "loss": 2.2486,
      "step": 138745
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.0748307704925537,
      "learning_rate": 0.0002053355905318186,
      "loss": 2.0232,
      "step": 138746
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.0901775360107422,
      "learning_rate": 0.00020533170896688108,
      "loss": 2.0086,
      "step": 138747
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.0762594938278198,
      "learning_rate": 0.0002053278274195442,
      "loss": 2.2038,
      "step": 138748
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.0979591608047485,
      "learning_rate": 0.0002053239458898087,
      "loss": 2.2085,
      "step": 138749
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.0670843124389648,
      "learning_rate": 0.00020532006437767528,
      "loss": 1.9432,
      "step": 138750
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.1218485832214355,
      "learning_rate": 0.00020531618288314462,
      "loss": 2.105,
      "step": 138751
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.016502022743225,
      "learning_rate": 0.00020531230140621756,
      "loss": 2.1627,
      "step": 138752
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.2250312566757202,
      "learning_rate": 0.0002053084199468946,
      "loss": 2.1368,
      "step": 138753
    },
    {
      "epoch": 1.81,
      "grad_norm": 0.9992095828056335,
      "learning_rate": 0.00020530453850517663,
      "loss": 1.8168,
      "step": 138754
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.0120549201965332,
      "learning_rate": 0.0002053006570810643,
      "loss": 2.0557,
      "step": 138755
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.009809136390686,
      "learning_rate": 0.0002052967756745584,
      "loss": 2.1463,
      "step": 138756
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.007449984550476,
      "learning_rate": 0.0002052928942856597,
      "loss": 2.0522,
      "step": 138757
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.1540285348892212,
      "learning_rate": 0.00020528901291436873,
      "loss": 2.1483,
      "step": 138758
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.0961207151412964,
      "learning_rate": 0.0002052851315606864,
      "loss": 2.0584,
      "step": 138759
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.1055335998535156,
      "learning_rate": 0.0002052812502246133,
      "loss": 2.2939,
      "step": 138760
    },
    {
      "epoch": 1.81,
      "grad_norm": 0.9887341856956482,
      "learning_rate": 0.00020527736890615017,
      "loss": 1.981,
      "step": 138761
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.070289134979248,
      "learning_rate": 0.00020527348760529778,
      "loss": 2.1241,
      "step": 138762
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.0933822393417358,
      "learning_rate": 0.0002052696063220569,
      "loss": 2.4156,
      "step": 138763
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.1918479204177856,
      "learning_rate": 0.00020526572505642817,
      "loss": 1.9975,
      "step": 138764
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.320871353149414,
      "learning_rate": 0.00020526184380841226,
      "loss": 2.2448,
      "step": 138765
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.052941918373108,
      "learning_rate": 0.00020525796257801006,
      "loss": 2.1175,
      "step": 138766
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.0518285036087036,
      "learning_rate": 0.00020525408136522213,
      "loss": 1.9751,
      "step": 138767
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.0011531114578247,
      "learning_rate": 0.00020525020017004927,
      "loss": 2.0554,
      "step": 138768
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.170706033706665,
      "learning_rate": 0.00020524631899249222,
      "loss": 2.4468,
      "step": 138769
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.1533657312393188,
      "learning_rate": 0.0002052424378325516,
      "loss": 2.0136,
      "step": 138770
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.0616710186004639,
      "learning_rate": 0.00020523855669022823,
      "loss": 2.2664,
      "step": 138771
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.0976307392120361,
      "learning_rate": 0.0002052346755655228,
      "loss": 2.2151,
      "step": 138772
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.146997094154358,
      "learning_rate": 0.000205230794458436,
      "loss": 2.3223,
      "step": 138773
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.031477451324463,
      "learning_rate": 0.00020522691336896865,
      "loss": 2.1152,
      "step": 138774
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.048168659210205,
      "learning_rate": 0.00020522303229712142,
      "loss": 2.1167,
      "step": 138775
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.0892566442489624,
      "learning_rate": 0.00020521915124289497,
      "loss": 1.9245,
      "step": 138776
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.049089789390564,
      "learning_rate": 0.00020521527020629006,
      "loss": 2.0468,
      "step": 138777
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.1113479137420654,
      "learning_rate": 0.00020521138918730742,
      "loss": 2.0556,
      "step": 138778
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.085038423538208,
      "learning_rate": 0.00020520750818594778,
      "loss": 1.989,
      "step": 138779
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.0321799516677856,
      "learning_rate": 0.00020520362720221182,
      "loss": 1.9498,
      "step": 138780
    },
    {
      "epoch": 1.81,
      "grad_norm": 0.9958482980728149,
      "learning_rate": 0.00020519974623610043,
      "loss": 2.0727,
      "step": 138781
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.0309720039367676,
      "learning_rate": 0.00020519586528761408,
      "loss": 2.1945,
      "step": 138782
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.1955151557922363,
      "learning_rate": 0.0002051919843567536,
      "loss": 2.4348,
      "step": 138783
    },
    {
      "epoch": 1.81,
      "grad_norm": 0.9942362904548645,
      "learning_rate": 0.00020518810344351976,
      "loss": 2.0007,
      "step": 138784
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.0383753776550293,
      "learning_rate": 0.00020518422254791322,
      "loss": 2.2234,
      "step": 138785
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.2552050352096558,
      "learning_rate": 0.0002051803416699347,
      "loss": 2.1134,
      "step": 138786
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.252745509147644,
      "learning_rate": 0.0002051764608095851,
      "loss": 1.9261,
      "step": 138787
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.1651040315628052,
      "learning_rate": 0.00020517257996686484,
      "loss": 2.1763,
      "step": 138788
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.1263173818588257,
      "learning_rate": 0.00020516869914177476,
      "loss": 2.3019,
      "step": 138789
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.0977890491485596,
      "learning_rate": 0.00020516481833431565,
      "loss": 2.0102,
      "step": 138790
    },
    {
      "epoch": 1.81,
      "grad_norm": 0.9791253209114075,
      "learning_rate": 0.00020516093754448822,
      "loss": 2.1701,
      "step": 138791
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.054456114768982,
      "learning_rate": 0.00020515705677229309,
      "loss": 2.1237,
      "step": 138792
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.0111253261566162,
      "learning_rate": 0.00020515317601773112,
      "loss": 2.3159,
      "step": 138793
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.0266386270523071,
      "learning_rate": 0.00020514929528080305,
      "loss": 2.2457,
      "step": 138794
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.0960147380828857,
      "learning_rate": 0.0002051454145615094,
      "loss": 2.1762,
      "step": 138795
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.0923259258270264,
      "learning_rate": 0.00020514153385985096,
      "loss": 1.869,
      "step": 138796
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.036158561706543,
      "learning_rate": 0.00020513765317582857,
      "loss": 2.2205,
      "step": 138797
    },
    {
      "epoch": 1.81,
      "grad_norm": 0.9810584187507629,
      "learning_rate": 0.00020513377250944286,
      "loss": 1.9644,
      "step": 138798
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.1185864210128784,
      "learning_rate": 0.00020512989186069456,
      "loss": 1.9807,
      "step": 138799
    },
    {
      "epoch": 1.81,
      "grad_norm": 0.9764237403869629,
      "learning_rate": 0.00020512601122958454,
      "loss": 1.9232,
      "step": 138800
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.1492804288864136,
      "learning_rate": 0.00020512213061611324,
      "loss": 2.0777,
      "step": 138801
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.2199753522872925,
      "learning_rate": 0.00020511825002028156,
      "loss": 2.2146,
      "step": 138802
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.1477631330490112,
      "learning_rate": 0.00020511436944209018,
      "loss": 2.045,
      "step": 138803
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.0256720781326294,
      "learning_rate": 0.00020511048888153982,
      "loss": 1.978,
      "step": 138804
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.0608161687850952,
      "learning_rate": 0.00020510660833863122,
      "loss": 1.9856,
      "step": 138805
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.36599600315094,
      "learning_rate": 0.00020510272781336524,
      "loss": 2.1271,
      "step": 138806
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.6822400093078613,
      "learning_rate": 0.00020509884730574227,
      "loss": 2.1426,
      "step": 138807
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.3130661249160767,
      "learning_rate": 0.00020509496681576326,
      "loss": 1.9261,
      "step": 138808
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.1118658781051636,
      "learning_rate": 0.00020509108634342884,
      "loss": 2.3101,
      "step": 138809
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.0510759353637695,
      "learning_rate": 0.00020508720588873985,
      "loss": 2.2812,
      "step": 138810
    },
    {
      "epoch": 1.81,
      "grad_norm": 0.9806228280067444,
      "learning_rate": 0.0002050833254516969,
      "loss": 1.9202,
      "step": 138811
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.0454856157302856,
      "learning_rate": 0.00020507944503230088,
      "loss": 2.1774,
      "step": 138812
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.2188780307769775,
      "learning_rate": 0.00020507556463055226,
      "loss": 2.0184,
      "step": 138813
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.1071674823760986,
      "learning_rate": 0.00020507168424645187,
      "loss": 2.0707,
      "step": 138814
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.0208300352096558,
      "learning_rate": 0.00020506780388000046,
      "loss": 1.9497,
      "step": 138815
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.0677286386489868,
      "learning_rate": 0.0002050639235311987,
      "loss": 2.1352,
      "step": 138816
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.183127999305725,
      "learning_rate": 0.00020506004320004743,
      "loss": 2.2426,
      "step": 138817
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.1059991121292114,
      "learning_rate": 0.00020505616288654733,
      "loss": 2.1043,
      "step": 138818
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.01155686378479,
      "learning_rate": 0.00020505228259069902,
      "loss": 2.105,
      "step": 138819
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.0547958612442017,
      "learning_rate": 0.0002050484023125033,
      "loss": 2.0709,
      "step": 138820
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.0134303569793701,
      "learning_rate": 0.0002050445220519608,
      "loss": 1.9556,
      "step": 138821
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.1376307010650635,
      "learning_rate": 0.00020504064180907233,
      "loss": 2.131,
      "step": 138822
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.2652682065963745,
      "learning_rate": 0.0002050367615838386,
      "loss": 2.0899,
      "step": 138823
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.035199761390686,
      "learning_rate": 0.0002050328813762605,
      "loss": 2.2496,
      "step": 138824
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.395882487297058,
      "learning_rate": 0.0002050290011863384,
      "loss": 1.9892,
      "step": 138825
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.0206522941589355,
      "learning_rate": 0.00020502512101407326,
      "loss": 2.0987,
      "step": 138826
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.0117448568344116,
      "learning_rate": 0.0002050212408594657,
      "loss": 2.0984,
      "step": 138827
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.027789831161499,
      "learning_rate": 0.0002050173607225165,
      "loss": 2.3335,
      "step": 138828
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.0099650621414185,
      "learning_rate": 0.00020501348060322636,
      "loss": 1.9759,
      "step": 138829
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.0198055505752563,
      "learning_rate": 0.0002050096005015961,
      "loss": 2.0741,
      "step": 138830
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.0529497861862183,
      "learning_rate": 0.00020500572041762626,
      "loss": 2.3113,
      "step": 138831
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.1182693243026733,
      "learning_rate": 0.00020500184035131765,
      "loss": 2.1346,
      "step": 138832
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.1265817880630493,
      "learning_rate": 0.000204997960302671,
      "loss": 1.8744,
      "step": 138833
    },
    {
      "epoch": 1.81,
      "grad_norm": 0.9817794561386108,
      "learning_rate": 0.00020499408027168698,
      "loss": 1.9151,
      "step": 138834
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.1290332078933716,
      "learning_rate": 0.00020499020025836642,
      "loss": 2.1121,
      "step": 138835
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.2963218688964844,
      "learning_rate": 0.00020498632026270998,
      "loss": 2.1464,
      "step": 138836
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.100008487701416,
      "learning_rate": 0.0002049824402847184,
      "loss": 2.2465,
      "step": 138837
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.1295690536499023,
      "learning_rate": 0.0002049785603243923,
      "loss": 2.1464,
      "step": 138838
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.0882741212844849,
      "learning_rate": 0.0002049746803817325,
      "loss": 1.8774,
      "step": 138839
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.207406997680664,
      "learning_rate": 0.0002049708004567397,
      "loss": 2.1482,
      "step": 138840
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.116827368736267,
      "learning_rate": 0.00020496692054941458,
      "loss": 2.1938,
      "step": 138841
    },
    {
      "epoch": 1.81,
      "grad_norm": 0.9323760271072388,
      "learning_rate": 0.000204963040659758,
      "loss": 1.9944,
      "step": 138842
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.2378746271133423,
      "learning_rate": 0.00020495916078777055,
      "loss": 2.0757,
      "step": 138843
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.0018097162246704,
      "learning_rate": 0.00020495528093345298,
      "loss": 2.2283,
      "step": 138844
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.3222039937973022,
      "learning_rate": 0.00020495140109680603,
      "loss": 2.1005,
      "step": 138845
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.254051685333252,
      "learning_rate": 0.00020494752127783034,
      "loss": 2.1043,
      "step": 138846
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.0526001453399658,
      "learning_rate": 0.00020494364147652676,
      "loss": 2.0695,
      "step": 138847
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.7032644748687744,
      "learning_rate": 0.00020493976169289597,
      "loss": 2.0839,
      "step": 138848
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.174299716949463,
      "learning_rate": 0.00020493588192693863,
      "loss": 2.1379,
      "step": 138849
    },
    {
      "epoch": 1.81,
      "grad_norm": 0.9989938735961914,
      "learning_rate": 0.00020493200217865549,
      "loss": 2.1292,
      "step": 138850
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.1631238460540771,
      "learning_rate": 0.00020492812244804736,
      "loss": 1.8828,
      "step": 138851
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.2376009225845337,
      "learning_rate": 0.00020492424273511482,
      "loss": 2.1926,
      "step": 138852
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.1155366897583008,
      "learning_rate": 0.00020492036303985868,
      "loss": 2.3154,
      "step": 138853
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.0058003664016724,
      "learning_rate": 0.00020491648336227967,
      "loss": 2.0764,
      "step": 138854
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.052056074142456,
      "learning_rate": 0.00020491260370237842,
      "loss": 2.1686,
      "step": 138855
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.017289400100708,
      "learning_rate": 0.00020490872406015576,
      "loss": 2.209,
      "step": 138856
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.1169052124023438,
      "learning_rate": 0.00020490484443561234,
      "loss": 1.8259,
      "step": 138857
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.1004430055618286,
      "learning_rate": 0.0002049009648287489,
      "loss": 2.1612,
      "step": 138858
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.0933265686035156,
      "learning_rate": 0.0002048970852395662,
      "loss": 2.0814,
      "step": 138859
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.0250067710876465,
      "learning_rate": 0.00020489320566806487,
      "loss": 1.9516,
      "step": 138860
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.1902689933776855,
      "learning_rate": 0.00020488932611424572,
      "loss": 2.1003,
      "step": 138861
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.2333062887191772,
      "learning_rate": 0.00020488544657810945,
      "loss": 2.0511,
      "step": 138862
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.1824480295181274,
      "learning_rate": 0.00020488156705965674,
      "loss": 2.2063,
      "step": 138863
    },
    {
      "epoch": 1.81,
      "grad_norm": 0.9564096927642822,
      "learning_rate": 0.00020487768755888833,
      "loss": 1.8956,
      "step": 138864
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.2966344356536865,
      "learning_rate": 0.000204873808075805,
      "loss": 1.8729,
      "step": 138865
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.2289644479751587,
      "learning_rate": 0.00020486992861040742,
      "loss": 2.2115,
      "step": 138866
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.094387412071228,
      "learning_rate": 0.00020486604916269635,
      "loss": 1.8656,
      "step": 138867
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.0810257196426392,
      "learning_rate": 0.00020486216973267243,
      "loss": 2.1329,
      "step": 138868
    },
    {
      "epoch": 1.81,
      "grad_norm": 0.9723731875419617,
      "learning_rate": 0.00020485829032033644,
      "loss": 1.9732,
      "step": 138869
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.1485787630081177,
      "learning_rate": 0.00020485441092568904,
      "loss": 2.2548,
      "step": 138870
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.216565489768982,
      "learning_rate": 0.00020485053154873102,
      "loss": 2.2244,
      "step": 138871
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.154945969581604,
      "learning_rate": 0.0002048466521894631,
      "loss": 1.7725,
      "step": 138872
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.221836805343628,
      "learning_rate": 0.0002048427728478861,
      "loss": 2.2835,
      "step": 138873
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.2208861112594604,
      "learning_rate": 0.00020483889352400047,
      "loss": 2.0542,
      "step": 138874
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.0720242261886597,
      "learning_rate": 0.0002048350142178071,
      "loss": 2.169,
      "step": 138875
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.0080012083053589,
      "learning_rate": 0.00020483113492930673,
      "loss": 2.2528,
      "step": 138876
    },
    {
      "epoch": 1.81,
      "grad_norm": 0.9924234747886658,
      "learning_rate": 0.00020482725565850002,
      "loss": 2.0344,
      "step": 138877
    },
    {
      "epoch": 1.81,
      "grad_norm": 0.9727373719215393,
      "learning_rate": 0.00020482337640538776,
      "loss": 2.0438,
      "step": 138878
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.0055618286132812,
      "learning_rate": 0.00020481949716997068,
      "loss": 2.0715,
      "step": 138879
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.0700682401657104,
      "learning_rate": 0.0002048156179522494,
      "loss": 2.0611,
      "step": 138880
    },
    {
      "epoch": 1.81,
      "grad_norm": 0.9917766451835632,
      "learning_rate": 0.00020481173875222468,
      "loss": 2.0447,
      "step": 138881
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.0357513427734375,
      "learning_rate": 0.0002048078595698972,
      "loss": 2.1708,
      "step": 138882
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.0942491292953491,
      "learning_rate": 0.0002048039804052678,
      "loss": 2.3161,
      "step": 138883
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.3180853128433228,
      "learning_rate": 0.0002048001012583371,
      "loss": 2.1102,
      "step": 138884
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.1264946460723877,
      "learning_rate": 0.000204796222129106,
      "loss": 2.1739,
      "step": 138885
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.0549908876419067,
      "learning_rate": 0.00020479234301757493,
      "loss": 2.2422,
      "step": 138886
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.0210984945297241,
      "learning_rate": 0.0002047884639237448,
      "loss": 2.1376,
      "step": 138887
    },
    {
      "epoch": 1.81,
      "grad_norm": 0.9309908151626587,
      "learning_rate": 0.00020478458484761623,
      "loss": 2.0007,
      "step": 138888
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.156144142150879,
      "learning_rate": 0.00020478070578919006,
      "loss": 2.032,
      "step": 138889
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.0808537006378174,
      "learning_rate": 0.00020477682674846693,
      "loss": 1.9855,
      "step": 138890
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.0352396965026855,
      "learning_rate": 0.00020477294772544775,
      "loss": 2.0695,
      "step": 138891
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.0371898412704468,
      "learning_rate": 0.00020476906872013293,
      "loss": 2.0353,
      "step": 138892
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.1211988925933838,
      "learning_rate": 0.0002047651897325233,
      "loss": 2.0487,
      "step": 138893
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.1896443367004395,
      "learning_rate": 0.00020476131076261966,
      "loss": 2.0796,
      "step": 138894
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.1095634698867798,
      "learning_rate": 0.00020475743181042267,
      "loss": 1.914,
      "step": 138895
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.0050996541976929,
      "learning_rate": 0.00020475355287593306,
      "loss": 2.2978,
      "step": 138896
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.0145169496536255,
      "learning_rate": 0.0002047496739591517,
      "loss": 2.1803,
      "step": 138897
    },
    {
      "epoch": 1.81,
      "grad_norm": 0.9784649610519409,
      "learning_rate": 0.0002047457950600791,
      "loss": 1.8727,
      "step": 138898
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.0350821018218994,
      "learning_rate": 0.000204741916178716,
      "loss": 2.0231,
      "step": 138899
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.1499130725860596,
      "learning_rate": 0.00020473803731506317,
      "loss": 2.1214,
      "step": 138900
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.099770188331604,
      "learning_rate": 0.00020473415846912132,
      "loss": 2.2181,
      "step": 138901
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.0224559307098389,
      "learning_rate": 0.00020473027964089125,
      "loss": 2.12,
      "step": 138902
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.2619680166244507,
      "learning_rate": 0.0002047264008303737,
      "loss": 2.0598,
      "step": 138903
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.1099438667297363,
      "learning_rate": 0.00020472252203756922,
      "loss": 2.1364,
      "step": 138904
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.0485620498657227,
      "learning_rate": 0.00020471864326247858,
      "loss": 2.1186,
      "step": 138905
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.07638680934906,
      "learning_rate": 0.00020471476450510258,
      "loss": 1.9579,
      "step": 138906
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.0715428590774536,
      "learning_rate": 0.00020471088576544188,
      "loss": 2.046,
      "step": 138907
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.044701337814331,
      "learning_rate": 0.00020470700704349726,
      "loss": 1.9982,
      "step": 138908
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.1455804109573364,
      "learning_rate": 0.00020470312833926948,
      "loss": 2.0908,
      "step": 138909
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.1023348569869995,
      "learning_rate": 0.00020469924965275908,
      "loss": 2.0052,
      "step": 138910
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.0156444311141968,
      "learning_rate": 0.0002046953709839669,
      "loss": 2.3067,
      "step": 138911
    },
    {
      "epoch": 1.81,
      "grad_norm": 0.9886096715927124,
      "learning_rate": 0.00020469149233289365,
      "loss": 2.1467,
      "step": 138912
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.113762378692627,
      "learning_rate": 0.00020468761369954005,
      "loss": 2.1224,
      "step": 138913
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.1586307287216187,
      "learning_rate": 0.0002046837350839068,
      "loss": 2.3026,
      "step": 138914
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.0761555433273315,
      "learning_rate": 0.00020467985648599482,
      "loss": 2.2035,
      "step": 138915
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.0600972175598145,
      "learning_rate": 0.00020467597790580446,
      "loss": 2.3132,
      "step": 138916
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.053297519683838,
      "learning_rate": 0.00020467209934333665,
      "loss": 1.9852,
      "step": 138917
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.0076738595962524,
      "learning_rate": 0.00020466822079859216,
      "loss": 2.2531,
      "step": 138918
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.1094951629638672,
      "learning_rate": 0.0002046643422715716,
      "loss": 1.9469,
      "step": 138919
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.0067534446716309,
      "learning_rate": 0.00020466046376227574,
      "loss": 2.0705,
      "step": 138920
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.1090378761291504,
      "learning_rate": 0.00020465658527070544,
      "loss": 2.2406,
      "step": 138921
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.041113257408142,
      "learning_rate": 0.0002046527067968611,
      "loss": 1.9226,
      "step": 138922
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.1124508380889893,
      "learning_rate": 0.00020464882834074366,
      "loss": 2.1838,
      "step": 138923
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.273182988166809,
      "learning_rate": 0.00020464494990235382,
      "loss": 2.1268,
      "step": 138924
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.0755257606506348,
      "learning_rate": 0.00020464107148169226,
      "loss": 2.2872,
      "step": 138925
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.0662951469421387,
      "learning_rate": 0.0002046371930787597,
      "loss": 2.0819,
      "step": 138926
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.1002633571624756,
      "learning_rate": 0.00020463331469355696,
      "loss": 2.1878,
      "step": 138927
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.2369550466537476,
      "learning_rate": 0.00020462943632608476,
      "loss": 2.1931,
      "step": 138928
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.0739316940307617,
      "learning_rate": 0.00020462555797634365,
      "loss": 2.1473,
      "step": 138929
    },
    {
      "epoch": 1.81,
      "grad_norm": 0.9430907964706421,
      "learning_rate": 0.00020462167964433446,
      "loss": 1.9112,
      "step": 138930
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.1610124111175537,
      "learning_rate": 0.00020461780133005785,
      "loss": 2.0829,
      "step": 138931
    },
    {
      "epoch": 1.81,
      "grad_norm": 0.9867045879364014,
      "learning_rate": 0.0002046139230335146,
      "loss": 1.8253,
      "step": 138932
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.0885560512542725,
      "learning_rate": 0.00020461004475470543,
      "loss": 2.1888,
      "step": 138933
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.0258944034576416,
      "learning_rate": 0.00020460616649363116,
      "loss": 2.1119,
      "step": 138934
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.084281086921692,
      "learning_rate": 0.00020460228825029228,
      "loss": 1.9651,
      "step": 138935
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.02480947971344,
      "learning_rate": 0.00020459841002468972,
      "loss": 1.994,
      "step": 138936
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.145243763923645,
      "learning_rate": 0.00020459453181682406,
      "loss": 2.2887,
      "step": 138937
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.181128978729248,
      "learning_rate": 0.00020459065362669605,
      "loss": 2.0094,
      "step": 138938
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.12063467502594,
      "learning_rate": 0.00020458677545430652,
      "loss": 2.0949,
      "step": 138939
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.0352578163146973,
      "learning_rate": 0.0002045828972996561,
      "loss": 1.9203,
      "step": 138940
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.0669844150543213,
      "learning_rate": 0.00020457901916274545,
      "loss": 2.2199,
      "step": 138941
    },
    {
      "epoch": 1.81,
      "grad_norm": 2.000195264816284,
      "learning_rate": 0.0002045751410435754,
      "loss": 2.1641,
      "step": 138942
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.1860579252243042,
      "learning_rate": 0.00020457126294214667,
      "loss": 2.1413,
      "step": 138943
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.2464299201965332,
      "learning_rate": 0.00020456738485845992,
      "loss": 2.1723,
      "step": 138944
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.2338535785675049,
      "learning_rate": 0.00020456350679251583,
      "loss": 2.0591,
      "step": 138945
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.2424468994140625,
      "learning_rate": 0.0002045596287443153,
      "loss": 2.1741,
      "step": 138946
    },
    {
      "epoch": 1.81,
      "grad_norm": 0.965246856212616,
      "learning_rate": 0.00020455575071385884,
      "loss": 1.8479,
      "step": 138947
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.067381739616394,
      "learning_rate": 0.0002045518727011473,
      "loss": 2.0576,
      "step": 138948
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.0972543954849243,
      "learning_rate": 0.00020454799470618135,
      "loss": 2.3157,
      "step": 138949
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.0617430210113525,
      "learning_rate": 0.00020454411672896178,
      "loss": 2.1416,
      "step": 138950
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.0112955570220947,
      "learning_rate": 0.00020454023876948922,
      "loss": 2.1587,
      "step": 138951
    },
    {
      "epoch": 1.81,
      "grad_norm": 0.9948828816413879,
      "learning_rate": 0.00020453636082776449,
      "loss": 2.1078,
      "step": 138952
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.034997820854187,
      "learning_rate": 0.00020453248290378822,
      "loss": 1.9712,
      "step": 138953
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.1182451248168945,
      "learning_rate": 0.0002045286049975611,
      "loss": 2.0215,
      "step": 138954
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.0775395631790161,
      "learning_rate": 0.00020452472710908394,
      "loss": 2.1721,
      "step": 138955
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.1046807765960693,
      "learning_rate": 0.00020452084923835745,
      "loss": 2.2742,
      "step": 138956
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.1810836791992188,
      "learning_rate": 0.00020451697138538241,
      "loss": 2.2499,
      "step": 138957
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.0526810884475708,
      "learning_rate": 0.00020451309355015943,
      "loss": 2.1041,
      "step": 138958
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.0118250846862793,
      "learning_rate": 0.00020450921573268922,
      "loss": 1.9711,
      "step": 138959
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.0139479637145996,
      "learning_rate": 0.00020450533793297255,
      "loss": 2.1193,
      "step": 138960
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.0193052291870117,
      "learning_rate": 0.00020450146015101015,
      "loss": 1.9728,
      "step": 138961
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.1084767580032349,
      "learning_rate": 0.00020449758238680273,
      "loss": 2.2148,
      "step": 138962
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.04976487159729,
      "learning_rate": 0.000204493704640351,
      "loss": 2.1735,
      "step": 138963
    },
    {
      "epoch": 1.81,
      "grad_norm": 0.926576554775238,
      "learning_rate": 0.00020448982691165582,
      "loss": 2.1165,
      "step": 138964
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.0979090929031372,
      "learning_rate": 0.00020448594920071766,
      "loss": 1.9809,
      "step": 138965
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.032023310661316,
      "learning_rate": 0.00020448207150753736,
      "loss": 1.9159,
      "step": 138966
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.0464502573013306,
      "learning_rate": 0.00020447819383211566,
      "loss": 2.1281,
      "step": 138967
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.1951916217803955,
      "learning_rate": 0.0002044743161744533,
      "loss": 2.0375,
      "step": 138968
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.0964263677597046,
      "learning_rate": 0.00020447043853455093,
      "loss": 1.9444,
      "step": 138969
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.012399435043335,
      "learning_rate": 0.0002044665609124094,
      "loss": 2.0088,
      "step": 138970
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.0193214416503906,
      "learning_rate": 0.00020446268330802925,
      "loss": 2.1313,
      "step": 138971
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.2011643648147583,
      "learning_rate": 0.0002044588057214113,
      "loss": 2.0985,
      "step": 138972
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.2459967136383057,
      "learning_rate": 0.00020445492815255621,
      "loss": 1.8872,
      "step": 138973
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.0184705257415771,
      "learning_rate": 0.0002044510506014648,
      "loss": 2.0502,
      "step": 138974
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.1096572875976562,
      "learning_rate": 0.00020444717306813775,
      "loss": 1.9843,
      "step": 138975
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.449584722518921,
      "learning_rate": 0.00020444329555257588,
      "loss": 2.0936,
      "step": 138976
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.0119694471359253,
      "learning_rate": 0.0002044394180547797,
      "loss": 2.215,
      "step": 138977
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.0260266065597534,
      "learning_rate": 0.00020443554057475002,
      "loss": 2.2525,
      "step": 138978
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.1122229099273682,
      "learning_rate": 0.00020443166311248756,
      "loss": 2.1381,
      "step": 138979
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.0523326396942139,
      "learning_rate": 0.0002044277856679931,
      "loss": 2.2656,
      "step": 138980
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.0606797933578491,
      "learning_rate": 0.00020442390824126726,
      "loss": 2.057,
      "step": 138981
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.1037554740905762,
      "learning_rate": 0.000204420030832311,
      "loss": 2.0438,
      "step": 138982
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.1470774412155151,
      "learning_rate": 0.0002044161534411247,
      "loss": 1.9495,
      "step": 138983
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.0474447011947632,
      "learning_rate": 0.00020441227606770928,
      "loss": 1.9578,
      "step": 138984
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.1788965463638306,
      "learning_rate": 0.0002044083987120654,
      "loss": 2.2939,
      "step": 138985
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.0954349040985107,
      "learning_rate": 0.00020440452137419375,
      "loss": 2.2438,
      "step": 138986
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.157039999961853,
      "learning_rate": 0.00020440064405409517,
      "loss": 2.0151,
      "step": 138987
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.0531178712844849,
      "learning_rate": 0.00020439676675177045,
      "loss": 2.4176,
      "step": 138988
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.0443936586380005,
      "learning_rate": 0.00020439288946722,
      "loss": 2.1397,
      "step": 138989
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.3280702829360962,
      "learning_rate": 0.00020438901220044478,
      "loss": 2.2415,
      "step": 138990
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.2193185091018677,
      "learning_rate": 0.00020438513495144538,
      "loss": 2.1437,
      "step": 138991
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.1458487510681152,
      "learning_rate": 0.0002043812577202226,
      "loss": 2.0322,
      "step": 138992
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.3472751379013062,
      "learning_rate": 0.00020437738050677718,
      "loss": 1.9165,
      "step": 138993
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.009860634803772,
      "learning_rate": 0.0002043735033111098,
      "loss": 1.9822,
      "step": 138994
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.035300612449646,
      "learning_rate": 0.00020436962613322132,
      "loss": 2.0885,
      "step": 138995
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.242414116859436,
      "learning_rate": 0.0002043657489731122,
      "loss": 2.0957,
      "step": 138996
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.388424277305603,
      "learning_rate": 0.00020436187183078328,
      "loss": 2.1487,
      "step": 138997
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.034570574760437,
      "learning_rate": 0.00020435799470623536,
      "loss": 2.0388,
      "step": 138998
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.1428208351135254,
      "learning_rate": 0.00020435411759946897,
      "loss": 2.3051,
      "step": 138999
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.0746876001358032,
      "learning_rate": 0.00020435024051048504,
      "loss": 2.0759,
      "step": 139000
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.1602872610092163,
      "learning_rate": 0.00020434636343928434,
      "loss": 2.3409,
      "step": 139001
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.0853530168533325,
      "learning_rate": 0.0002043424863858673,
      "loss": 2.173,
      "step": 139002
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.0791770219802856,
      "learning_rate": 0.00020433860935023477,
      "loss": 1.9621,
      "step": 139003
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.0588048696517944,
      "learning_rate": 0.00020433473233238755,
      "loss": 2.0648,
      "step": 139004
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.274906873703003,
      "learning_rate": 0.0002043308553323263,
      "loss": 2.2833,
      "step": 139005
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.0364702939987183,
      "learning_rate": 0.00020432697835005175,
      "loss": 2.1001,
      "step": 139006
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.0686235427856445,
      "learning_rate": 0.00020432310138556475,
      "loss": 2.0617,
      "step": 139007
    },
    {
      "epoch": 1.81,
      "grad_norm": 0.9115118384361267,
      "learning_rate": 0.00020431922443886575,
      "loss": 2.0483,
      "step": 139008
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.0947932004928589,
      "learning_rate": 0.0002043153475099556,
      "loss": 2.1432,
      "step": 139009
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.1014502048492432,
      "learning_rate": 0.00020431147059883504,
      "loss": 2.0433,
      "step": 139010
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.0871208906173706,
      "learning_rate": 0.0002043075937055048,
      "loss": 2.0807,
      "step": 139011
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.4040488004684448,
      "learning_rate": 0.0002043037168299656,
      "loss": 1.786,
      "step": 139012
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.0953487157821655,
      "learning_rate": 0.00020429983997221824,
      "loss": 2.0479,
      "step": 139013
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.1319934129714966,
      "learning_rate": 0.00020429596313226324,
      "loss": 1.9389,
      "step": 139014
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.0011539459228516,
      "learning_rate": 0.00020429208631010142,
      "loss": 1.9811,
      "step": 139015
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.1739293336868286,
      "learning_rate": 0.0002042882095057335,
      "loss": 2.2119,
      "step": 139016
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.1333562135696411,
      "learning_rate": 0.0002042843327191602,
      "loss": 2.1211,
      "step": 139017
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.0521305799484253,
      "learning_rate": 0.00020428045595038226,
      "loss": 2.0219,
      "step": 139018
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.0768814086914062,
      "learning_rate": 0.00020427657919940047,
      "loss": 2.1194,
      "step": 139019
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.0358227491378784,
      "learning_rate": 0.00020427270246621543,
      "loss": 2.0125,
      "step": 139020
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.1525044441223145,
      "learning_rate": 0.0002042688257508279,
      "loss": 2.1822,
      "step": 139021
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.1514331102371216,
      "learning_rate": 0.00020426494905323857,
      "loss": 2.0172,
      "step": 139022
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.056505560874939,
      "learning_rate": 0.00020426107237344818,
      "loss": 2.3529,
      "step": 139023
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.0369634628295898,
      "learning_rate": 0.00020425719571145746,
      "loss": 2.1928,
      "step": 139024
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.1598998308181763,
      "learning_rate": 0.00020425331906726723,
      "loss": 2.065,
      "step": 139025
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.0598571300506592,
      "learning_rate": 0.000204249442440878,
      "loss": 2.2594,
      "step": 139026
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.078594446182251,
      "learning_rate": 0.00020424556583229067,
      "loss": 2.1548,
      "step": 139027
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.2424312829971313,
      "learning_rate": 0.0002042416892415059,
      "loss": 2.1048,
      "step": 139028
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.086898684501648,
      "learning_rate": 0.0002042378126685244,
      "loss": 2.0071,
      "step": 139029
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.1423472166061401,
      "learning_rate": 0.00020423393611334687,
      "loss": 2.2155,
      "step": 139030
    },
    {
      "epoch": 1.81,
      "grad_norm": 0.9862158894538879,
      "learning_rate": 0.00020423005957597408,
      "loss": 2.1073,
      "step": 139031
    },
    {
      "epoch": 1.81,
      "grad_norm": 0.9702250361442566,
      "learning_rate": 0.00020422618305640674,
      "loss": 2.1025,
      "step": 139032
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.106547474861145,
      "learning_rate": 0.00020422230655464554,
      "loss": 2.4487,
      "step": 139033
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.049591064453125,
      "learning_rate": 0.00020421843007069118,
      "loss": 1.873,
      "step": 139034
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.0690771341323853,
      "learning_rate": 0.00020421455360454454,
      "loss": 2.1157,
      "step": 139035
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.1215499639511108,
      "learning_rate": 0.00020421067715620616,
      "loss": 2.1118,
      "step": 139036
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.1144607067108154,
      "learning_rate": 0.00020420680072567683,
      "loss": 2.1211,
      "step": 139037
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.2948004007339478,
      "learning_rate": 0.00020420292431295726,
      "loss": 2.2488,
      "step": 139038
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.1154372692108154,
      "learning_rate": 0.00020419904791804812,
      "loss": 2.1845,
      "step": 139039
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.102918267250061,
      "learning_rate": 0.00020419517154095023,
      "loss": 2.0258,
      "step": 139040
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.198447585105896,
      "learning_rate": 0.00020419129518166426,
      "loss": 1.9468,
      "step": 139041
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.1449248790740967,
      "learning_rate": 0.00020418741884019098,
      "loss": 1.9104,
      "step": 139042
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.083023190498352,
      "learning_rate": 0.00020418354251653105,
      "loss": 2.0783,
      "step": 139043
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.0428578853607178,
      "learning_rate": 0.00020417966621068522,
      "loss": 2.3106,
      "step": 139044
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.0711313486099243,
      "learning_rate": 0.00020417578992265413,
      "loss": 2.3613,
      "step": 139045
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.076035499572754,
      "learning_rate": 0.0002041719136524386,
      "loss": 2.1804,
      "step": 139046
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.0523474216461182,
      "learning_rate": 0.00020416803740003932,
      "loss": 2.2499,
      "step": 139047
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.1196331977844238,
      "learning_rate": 0.000204164161165457,
      "loss": 2.1319,
      "step": 139048
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.0861809253692627,
      "learning_rate": 0.0002041602849486925,
      "loss": 2.1078,
      "step": 139049
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.0904029607772827,
      "learning_rate": 0.0002041564087497463,
      "loss": 2.0939,
      "step": 139050
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.0814764499664307,
      "learning_rate": 0.0002041525325686192,
      "loss": 2.1136,
      "step": 139051
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.1137381792068481,
      "learning_rate": 0.00020414865640531197,
      "loss": 1.9754,
      "step": 139052
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.3102610111236572,
      "learning_rate": 0.00020414478025982533,
      "loss": 1.8515,
      "step": 139053
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.1432069540023804,
      "learning_rate": 0.00020414090413215996,
      "loss": 2.2387,
      "step": 139054
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.0772995948791504,
      "learning_rate": 0.00020413702802231674,
      "loss": 1.9605,
      "step": 139055
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.1287219524383545,
      "learning_rate": 0.00020413315193029612,
      "loss": 2.0062,
      "step": 139056
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.1645452976226807,
      "learning_rate": 0.00020412927585609897,
      "loss": 2.1269,
      "step": 139057
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.249945878982544,
      "learning_rate": 0.000204125399799726,
      "loss": 1.9142,
      "step": 139058
    },
    {
      "epoch": 1.81,
      "grad_norm": 0.9952827095985413,
      "learning_rate": 0.0002041215237611779,
      "loss": 1.878,
      "step": 139059
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.1825882196426392,
      "learning_rate": 0.00020411764774045548,
      "loss": 1.9063,
      "step": 139060
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.2343723773956299,
      "learning_rate": 0.00020411377173755937,
      "loss": 2.1221,
      "step": 139061
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.1444783210754395,
      "learning_rate": 0.00020410989575249042,
      "loss": 2.1163,
      "step": 139062
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.0758780241012573,
      "learning_rate": 0.00020410601978524916,
      "loss": 1.9786,
      "step": 139063
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.1433771848678589,
      "learning_rate": 0.00020410214383583637,
      "loss": 2.1048,
      "step": 139064
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.0936284065246582,
      "learning_rate": 0.00020409826790425281,
      "loss": 2.1527,
      "step": 139065
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.0235531330108643,
      "learning_rate": 0.0002040943919904992,
      "loss": 2.2599,
      "step": 139066
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.0679233074188232,
      "learning_rate": 0.00020409051609457623,
      "loss": 2.1484,
      "step": 139067
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.1894549131393433,
      "learning_rate": 0.00020408664021648483,
      "loss": 1.8847,
      "step": 139068
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.2387933731079102,
      "learning_rate": 0.00020408276435622536,
      "loss": 2.0686,
      "step": 139069
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.0616906881332397,
      "learning_rate": 0.0002040788885137987,
      "loss": 2.1068,
      "step": 139070
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.054901123046875,
      "learning_rate": 0.0002040750126892056,
      "loss": 2.1438,
      "step": 139071
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.3243887424468994,
      "learning_rate": 0.00020407113688244682,
      "loss": 2.0644,
      "step": 139072
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.2146837711334229,
      "learning_rate": 0.00020406726109352294,
      "loss": 2.1254,
      "step": 139073
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.0417684316635132,
      "learning_rate": 0.00020406338532243494,
      "loss": 2.0435,
      "step": 139074
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.6856698989868164,
      "learning_rate": 0.00020405950956918324,
      "loss": 2.3335,
      "step": 139075
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.091148853302002,
      "learning_rate": 0.00020405563383376867,
      "loss": 2.1148,
      "step": 139076
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.323528528213501,
      "learning_rate": 0.000204051758116192,
      "loss": 2.1187,
      "step": 139077
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.0283401012420654,
      "learning_rate": 0.0002040478824164539,
      "loss": 2.151,
      "step": 139078
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.1514936685562134,
      "learning_rate": 0.0002040440067345551,
      "loss": 2.1576,
      "step": 139079
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.0303858518600464,
      "learning_rate": 0.00020404013107049647,
      "loss": 2.1973,
      "step": 139080
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.0196794271469116,
      "learning_rate": 0.00020403625542427846,
      "loss": 1.9028,
      "step": 139081
    },
    {
      "epoch": 1.81,
      "grad_norm": 0.9585750699043274,
      "learning_rate": 0.00020403237979590194,
      "loss": 2.2101,
      "step": 139082
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.154951810836792,
      "learning_rate": 0.00020402850418536761,
      "loss": 2.3364,
      "step": 139083
    },
    {
      "epoch": 1.81,
      "grad_norm": 0.9984984397888184,
      "learning_rate": 0.00020402462859267617,
      "loss": 1.9653,
      "step": 139084
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.1137700080871582,
      "learning_rate": 0.00020402075301782836,
      "loss": 2.0249,
      "step": 139085
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.15305495262146,
      "learning_rate": 0.00020401687746082503,
      "loss": 2.2012,
      "step": 139086
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.1700712442398071,
      "learning_rate": 0.00020401300192166667,
      "loss": 2.207,
      "step": 139087
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.0091960430145264,
      "learning_rate": 0.00020400912640035408,
      "loss": 1.848,
      "step": 139088
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.089181661605835,
      "learning_rate": 0.00020400525089688803,
      "loss": 2.1321,
      "step": 139089
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.005375623703003,
      "learning_rate": 0.0002040013754112692,
      "loss": 2.0695,
      "step": 139090
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.0851062536239624,
      "learning_rate": 0.00020399749994349833,
      "loss": 1.9722,
      "step": 139091
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.0107122659683228,
      "learning_rate": 0.00020399362449357631,
      "loss": 2.1121,
      "step": 139092
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.0381126403808594,
      "learning_rate": 0.00020398974906150349,
      "loss": 1.9282,
      "step": 139093
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.1046792268753052,
      "learning_rate": 0.00020398587364728083,
      "loss": 2.0921,
      "step": 139094
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.0246682167053223,
      "learning_rate": 0.00020398199825090896,
      "loss": 1.8749,
      "step": 139095
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.064957857131958,
      "learning_rate": 0.00020397812287238867,
      "loss": 2.257,
      "step": 139096
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.1161519289016724,
      "learning_rate": 0.00020397424751172072,
      "loss": 2.2424,
      "step": 139097
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.0090211629867554,
      "learning_rate": 0.00020397037216890581,
      "loss": 2.1893,
      "step": 139098
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.0431877374649048,
      "learning_rate": 0.00020396649684394452,
      "loss": 2.2165,
      "step": 139099
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.0734566450119019,
      "learning_rate": 0.0002039626215368377,
      "loss": 2.1712,
      "step": 139100
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.0051183700561523,
      "learning_rate": 0.000203958746247586,
      "loss": 2.3095,
      "step": 139101
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.0132688283920288,
      "learning_rate": 0.00020395487097619024,
      "loss": 1.9644,
      "step": 139102
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.054432988166809,
      "learning_rate": 0.00020395099572265103,
      "loss": 2.1059,
      "step": 139103
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.236767053604126,
      "learning_rate": 0.00020394712048696926,
      "loss": 2.0765,
      "step": 139104
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.050439715385437,
      "learning_rate": 0.00020394324526914548,
      "loss": 1.9941,
      "step": 139105
    },
    {
      "epoch": 1.81,
      "grad_norm": 0.99945068359375,
      "learning_rate": 0.00020393937006918042,
      "loss": 2.2792,
      "step": 139106
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.142153263092041,
      "learning_rate": 0.00020393549488707485,
      "loss": 2.3064,
      "step": 139107
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.0417776107788086,
      "learning_rate": 0.0002039316197228295,
      "loss": 1.991,
      "step": 139108
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.0168566703796387,
      "learning_rate": 0.00020392774457644506,
      "loss": 2.0316,
      "step": 139109
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.0668176412582397,
      "learning_rate": 0.00020392386944792233,
      "loss": 1.9644,
      "step": 139110
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.0816562175750732,
      "learning_rate": 0.00020391999433726186,
      "loss": 2.3257,
      "step": 139111
    },
    {
      "epoch": 1.81,
      "grad_norm": 0.9806763529777527,
      "learning_rate": 0.0002039161192444646,
      "loss": 2.1872,
      "step": 139112
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.0503398180007935,
      "learning_rate": 0.00020391224416953105,
      "loss": 2.039,
      "step": 139113
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.2371318340301514,
      "learning_rate": 0.00020390836911246202,
      "loss": 2.1784,
      "step": 139114
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.5497047901153564,
      "learning_rate": 0.00020390449407325828,
      "loss": 2.3109,
      "step": 139115
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.0144119262695312,
      "learning_rate": 0.00020390061905192053,
      "loss": 1.9176,
      "step": 139116
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.117152452468872,
      "learning_rate": 0.00020389674404844944,
      "loss": 2.249,
      "step": 139117
    },
    {
      "epoch": 1.81,
      "grad_norm": 0.9813492894172668,
      "learning_rate": 0.00020389286906284575,
      "loss": 2.0185,
      "step": 139118
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.200177788734436,
      "learning_rate": 0.00020388899409511024,
      "loss": 2.0398,
      "step": 139119
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.1076741218566895,
      "learning_rate": 0.0002038851191452435,
      "loss": 2.0953,
      "step": 139120
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.0435338020324707,
      "learning_rate": 0.00020388124421324635,
      "loss": 2.0245,
      "step": 139121
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.0152876377105713,
      "learning_rate": 0.00020387736929911956,
      "loss": 1.9636,
      "step": 139122
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.1495977640151978,
      "learning_rate": 0.0002038734944028637,
      "loss": 2.1284,
      "step": 139123
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.086490511894226,
      "learning_rate": 0.00020386961952447962,
      "loss": 1.9885,
      "step": 139124
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.1189757585525513,
      "learning_rate": 0.00020386574466396795,
      "loss": 2.1078,
      "step": 139125
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.012934684753418,
      "learning_rate": 0.00020386186982132947,
      "loss": 2.2368,
      "step": 139126
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.0574169158935547,
      "learning_rate": 0.00020385799499656493,
      "loss": 2.1021,
      "step": 139127
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.2315325736999512,
      "learning_rate": 0.00020385412018967495,
      "loss": 2.1822,
      "step": 139128
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.1471647024154663,
      "learning_rate": 0.00020385024540066037,
      "loss": 1.8658,
      "step": 139129
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.056714415550232,
      "learning_rate": 0.00020384637062952177,
      "loss": 2.1155,
      "step": 139130
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.0813087224960327,
      "learning_rate": 0.00020384249587625998,
      "loss": 2.099,
      "step": 139131
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.2170579433441162,
      "learning_rate": 0.00020383862114087563,
      "loss": 2.1078,
      "step": 139132
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.033940315246582,
      "learning_rate": 0.00020383474642336952,
      "loss": 2.2102,
      "step": 139133
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.0581893920898438,
      "learning_rate": 0.00020383087172374243,
      "loss": 2.0861,
      "step": 139134
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.0077613592147827,
      "learning_rate": 0.000203826997041995,
      "loss": 2.0006,
      "step": 139135
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.0821839570999146,
      "learning_rate": 0.00020382312237812784,
      "loss": 1.9519,
      "step": 139136
    },
    {
      "epoch": 1.81,
      "grad_norm": 0.9634943604469299,
      "learning_rate": 0.0002038192477321418,
      "loss": 2.1308,
      "step": 139137
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.1472162008285522,
      "learning_rate": 0.00020381537310403756,
      "loss": 1.9368,
      "step": 139138
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.2022360563278198,
      "learning_rate": 0.0002038114984938159,
      "loss": 2.2677,
      "step": 139139
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.1300781965255737,
      "learning_rate": 0.00020380762390147743,
      "loss": 2.3555,
      "step": 139140
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.207551121711731,
      "learning_rate": 0.00020380374932702313,
      "loss": 1.9172,
      "step": 139141
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.068688988685608,
      "learning_rate": 0.0002037998747704534,
      "loss": 2.1184,
      "step": 139142
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.0018008947372437,
      "learning_rate": 0.00020379600023176904,
      "loss": 2.1396,
      "step": 139143
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.0347175598144531,
      "learning_rate": 0.00020379212571097088,
      "loss": 2.0951,
      "step": 139144
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.0423049926757812,
      "learning_rate": 0.00020378825120805953,
      "loss": 1.9685,
      "step": 139145
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.069668173789978,
      "learning_rate": 0.00020378437672303578,
      "loss": 2.2517,
      "step": 139146
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.0924664735794067,
      "learning_rate": 0.00020378050225590045,
      "loss": 2.1258,
      "step": 139147
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.145782232284546,
      "learning_rate": 0.00020377662780665402,
      "loss": 1.9544,
      "step": 139148
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.0008618831634521,
      "learning_rate": 0.00020377275337529735,
      "loss": 2.137,
      "step": 139149
    },
    {
      "epoch": 1.81,
      "grad_norm": 0.9746913909912109,
      "learning_rate": 0.00020376887896183109,
      "loss": 2.0155,
      "step": 139150
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.0741575956344604,
      "learning_rate": 0.00020376500456625605,
      "loss": 2.0215,
      "step": 139151
    },
    {
      "epoch": 1.81,
      "grad_norm": 0.939062237739563,
      "learning_rate": 0.00020376113018857292,
      "loss": 2.1043,
      "step": 139152
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.048642873764038,
      "learning_rate": 0.00020375725582878252,
      "loss": 2.1837,
      "step": 139153
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.1129884719848633,
      "learning_rate": 0.00020375338148688536,
      "loss": 2.0903,
      "step": 139154
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.1270873546600342,
      "learning_rate": 0.00020374950716288225,
      "loss": 2.1149,
      "step": 139155
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.1697427034378052,
      "learning_rate": 0.0002037456328567739,
      "loss": 2.2643,
      "step": 139156
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.1929097175598145,
      "learning_rate": 0.0002037417585685611,
      "loss": 2.2043,
      "step": 139157
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.0157067775726318,
      "learning_rate": 0.00020373788429824447,
      "loss": 1.8911,
      "step": 139158
    },
    {
      "epoch": 1.81,
      "grad_norm": 0.9819634556770325,
      "learning_rate": 0.00020373401004582491,
      "loss": 1.9945,
      "step": 139159
    },
    {
      "epoch": 1.81,
      "grad_norm": 0.9952725172042847,
      "learning_rate": 0.00020373013581130293,
      "loss": 2.2448,
      "step": 139160
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.161798357963562,
      "learning_rate": 0.00020372626159467934,
      "loss": 1.9941,
      "step": 139161
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.094942331314087,
      "learning_rate": 0.00020372238739595485,
      "loss": 2.0921,
      "step": 139162
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.0730305910110474,
      "learning_rate": 0.00020371851321513017,
      "loss": 2.3363,
      "step": 139163
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.353074550628662,
      "learning_rate": 0.00020371463905220603,
      "loss": 2.013,
      "step": 139164
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.2776497602462769,
      "learning_rate": 0.0002037107649071833,
      "loss": 1.9809,
      "step": 139165
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.2831158638000488,
      "learning_rate": 0.00020370689078006243,
      "loss": 2.1077,
      "step": 139166
    },
    {
      "epoch": 1.81,
      "grad_norm": 0.9585084915161133,
      "learning_rate": 0.00020370301667084427,
      "loss": 2.2168,
      "step": 139167
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.3564507961273193,
      "learning_rate": 0.0002036991425795295,
      "loss": 2.2589,
      "step": 139168
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.671995997428894,
      "learning_rate": 0.0002036952685061189,
      "loss": 1.9209,
      "step": 139169
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.3816574811935425,
      "learning_rate": 0.0002036913944506132,
      "loss": 2.0626,
      "step": 139170
    },
    {
      "epoch": 1.81,
      "grad_norm": 0.9887215495109558,
      "learning_rate": 0.00020368752041301311,
      "loss": 2.1676,
      "step": 139171
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.092002272605896,
      "learning_rate": 0.0002036836463933193,
      "loss": 2.0378,
      "step": 139172
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.1690233945846558,
      "learning_rate": 0.0002036797723915325,
      "loss": 1.9521,
      "step": 139173
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.7979381084442139,
      "learning_rate": 0.00020367589840765342,
      "loss": 1.8462,
      "step": 139174
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.6887716054916382,
      "learning_rate": 0.00020367202444168285,
      "loss": 2.1374,
      "step": 139175
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.0922082662582397,
      "learning_rate": 0.00020366815049362145,
      "loss": 1.9444,
      "step": 139176
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.0785783529281616,
      "learning_rate": 0.0002036642765634701,
      "loss": 2.2268,
      "step": 139177
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.4688078165054321,
      "learning_rate": 0.00020366040265122924,
      "loss": 1.9563,
      "step": 139178
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.1773109436035156,
      "learning_rate": 0.0002036565287568997,
      "loss": 2.2138,
      "step": 139179
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.1432623863220215,
      "learning_rate": 0.00020365265488048225,
      "loss": 2.0713,
      "step": 139180
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.007080078125,
      "learning_rate": 0.00020364878102197763,
      "loss": 2.0238,
      "step": 139181
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.0589035749435425,
      "learning_rate": 0.00020364490718138647,
      "loss": 2.2091,
      "step": 139182
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.152490496635437,
      "learning_rate": 0.00020364103335870966,
      "loss": 2.272,
      "step": 139183
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.477133870124817,
      "learning_rate": 0.0002036371595539477,
      "loss": 2.2866,
      "step": 139184
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.1035860776901245,
      "learning_rate": 0.00020363328576710144,
      "loss": 2.1784,
      "step": 139185
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.0706539154052734,
      "learning_rate": 0.00020362941199817154,
      "loss": 1.8558,
      "step": 139186
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.0506190061569214,
      "learning_rate": 0.00020362553824715876,
      "loss": 2.2245,
      "step": 139187
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.0631110668182373,
      "learning_rate": 0.00020362166451406378,
      "loss": 1.8685,
      "step": 139188
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.257457971572876,
      "learning_rate": 0.00020361779079888746,
      "loss": 2.1709,
      "step": 139189
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.1397769451141357,
      "learning_rate": 0.0002036139171016304,
      "loss": 2.1235,
      "step": 139190
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.1686956882476807,
      "learning_rate": 0.00020361004342229322,
      "loss": 1.9047,
      "step": 139191
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.3396574258804321,
      "learning_rate": 0.0002036061697608768,
      "loss": 2.1386,
      "step": 139192
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.1042405366897583,
      "learning_rate": 0.00020360229611738183,
      "loss": 2.0665,
      "step": 139193
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.2652138471603394,
      "learning_rate": 0.000203598422491809,
      "loss": 1.955,
      "step": 139194
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.0593293905258179,
      "learning_rate": 0.00020359454888415902,
      "loss": 2.0442,
      "step": 139195
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.1892248392105103,
      "learning_rate": 0.00020359067529443273,
      "loss": 2.0235,
      "step": 139196
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.2988972663879395,
      "learning_rate": 0.0002035868017226307,
      "loss": 2.1999,
      "step": 139197
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.3327035903930664,
      "learning_rate": 0.0002035829281687537,
      "loss": 2.1957,
      "step": 139198
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.1179816722869873,
      "learning_rate": 0.00020357905463280242,
      "loss": 2.2074,
      "step": 139199
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.0169775485992432,
      "learning_rate": 0.0002035751811147776,
      "loss": 2.2118,
      "step": 139200
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.2502845525741577,
      "learning_rate": 0.00020357130761468,
      "loss": 2.046,
      "step": 139201
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.3993772268295288,
      "learning_rate": 0.00020356743413251036,
      "loss": 2.0942,
      "step": 139202
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.0947818756103516,
      "learning_rate": 0.0002035635606682693,
      "loss": 2.0712,
      "step": 139203
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.0264836549758911,
      "learning_rate": 0.0002035596872219577,
      "loss": 2.1814,
      "step": 139204
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.0964893102645874,
      "learning_rate": 0.00020355581379357605,
      "loss": 2.0657,
      "step": 139205
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.1709938049316406,
      "learning_rate": 0.0002035519403831252,
      "loss": 2.0964,
      "step": 139206
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.2102214097976685,
      "learning_rate": 0.00020354806699060586,
      "loss": 2.0725,
      "step": 139207
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.0997555255889893,
      "learning_rate": 0.00020354419361601886,
      "loss": 2.1868,
      "step": 139208
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.0641734600067139,
      "learning_rate": 0.0002035403202593647,
      "loss": 2.031,
      "step": 139209
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.2434043884277344,
      "learning_rate": 0.00020353644692064428,
      "loss": 2.3771,
      "step": 139210
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.2262271642684937,
      "learning_rate": 0.00020353257359985826,
      "loss": 1.9838,
      "step": 139211
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.0436115264892578,
      "learning_rate": 0.0002035287002970073,
      "loss": 2.0005,
      "step": 139212
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.02669358253479,
      "learning_rate": 0.00020352482701209223,
      "loss": 1.9673,
      "step": 139213
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.1693522930145264,
      "learning_rate": 0.00020352095374511373,
      "loss": 2.1198,
      "step": 139214
    },
    {
      "epoch": 1.81,
      "grad_norm": 0.9747768044471741,
      "learning_rate": 0.00020351708049607246,
      "loss": 2.0752,
      "step": 139215
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.0492329597473145,
      "learning_rate": 0.00020351320726496916,
      "loss": 1.9756,
      "step": 139216
    },
    {
      "epoch": 1.81,
      "grad_norm": 0.9850717782974243,
      "learning_rate": 0.00020350933405180463,
      "loss": 2.418,
      "step": 139217
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.0276521444320679,
      "learning_rate": 0.00020350546085657947,
      "loss": 2.0253,
      "step": 139218
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.0802090167999268,
      "learning_rate": 0.00020350158767929458,
      "loss": 1.9623,
      "step": 139219
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.239872694015503,
      "learning_rate": 0.00020349771451995055,
      "loss": 2.2246,
      "step": 139220
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.159833312034607,
      "learning_rate": 0.00020349384137854806,
      "loss": 2.1263,
      "step": 139221
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.37770676612854,
      "learning_rate": 0.00020348996825508785,
      "loss": 2.1819,
      "step": 139222
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.1745290756225586,
      "learning_rate": 0.00020348609514957071,
      "loss": 2.0565,
      "step": 139223
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.0649209022521973,
      "learning_rate": 0.00020348222206199732,
      "loss": 2.1337,
      "step": 139224
    },
    {
      "epoch": 1.81,
      "grad_norm": 0.9895456433296204,
      "learning_rate": 0.00020347834899236842,
      "loss": 2.0874,
      "step": 139225
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.123457908630371,
      "learning_rate": 0.00020347447594068485,
      "loss": 1.8711,
      "step": 139226
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.1795443296432495,
      "learning_rate": 0.00020347060290694705,
      "loss": 2.071,
      "step": 139227
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.104902982711792,
      "learning_rate": 0.0002034667298911559,
      "loss": 2.0769,
      "step": 139228
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.0741738080978394,
      "learning_rate": 0.00020346285689331207,
      "loss": 1.958,
      "step": 139229
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.0190969705581665,
      "learning_rate": 0.00020345898391341635,
      "loss": 1.879,
      "step": 139230
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.1161713600158691,
      "learning_rate": 0.0002034551109514694,
      "loss": 2.1412,
      "step": 139231
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.103161334991455,
      "learning_rate": 0.00020345123800747213,
      "loss": 2.2252,
      "step": 139232
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.073203444480896,
      "learning_rate": 0.00020344736508142496,
      "loss": 2.095,
      "step": 139233
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.1008632183074951,
      "learning_rate": 0.00020344349217332874,
      "loss": 2.148,
      "step": 139234
    },
    {
      "epoch": 1.81,
      "grad_norm": 0.9972255825996399,
      "learning_rate": 0.00020343961928318422,
      "loss": 2.2114,
      "step": 139235
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.0306376218795776,
      "learning_rate": 0.00020343574641099206,
      "loss": 2.1251,
      "step": 139236
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.0157268047332764,
      "learning_rate": 0.00020343187355675304,
      "loss": 2.01,
      "step": 139237
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.1462163925170898,
      "learning_rate": 0.00020342800072046797,
      "loss": 2.1232,
      "step": 139238
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.0715861320495605,
      "learning_rate": 0.00020342412790213732,
      "loss": 1.9396,
      "step": 139239
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.1204134225845337,
      "learning_rate": 0.000203420255101762,
      "loss": 2.2444,
      "step": 139240
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.032536268234253,
      "learning_rate": 0.00020341638231934265,
      "loss": 1.8306,
      "step": 139241
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.1136765480041504,
      "learning_rate": 0.00020341250955488,
      "loss": 2.1072,
      "step": 139242
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.1428717374801636,
      "learning_rate": 0.00020340863680837478,
      "loss": 1.9067,
      "step": 139243
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.0418589115142822,
      "learning_rate": 0.00020340476407982786,
      "loss": 2.1226,
      "step": 139244
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.0907576084136963,
      "learning_rate": 0.00020340089136923972,
      "loss": 2.2085,
      "step": 139245
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.0544697046279907,
      "learning_rate": 0.00020339701867661114,
      "loss": 2.2901,
      "step": 139246
    },
    {
      "epoch": 1.81,
      "grad_norm": 0.9947714805603027,
      "learning_rate": 0.0002033931460019429,
      "loss": 2.1764,
      "step": 139247
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.0869035720825195,
      "learning_rate": 0.00020338927334523566,
      "loss": 1.8589,
      "step": 139248
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.067185401916504,
      "learning_rate": 0.00020338540070649025,
      "loss": 2.188,
      "step": 139249
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.0081318616867065,
      "learning_rate": 0.0002033815280857074,
      "loss": 2.1713,
      "step": 139250
    },
    {
      "epoch": 1.81,
      "grad_norm": 0.9770343899726868,
      "learning_rate": 0.00020337765548288761,
      "loss": 1.7857,
      "step": 139251
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.180552363395691,
      "learning_rate": 0.00020337378289803177,
      "loss": 2.0622,
      "step": 139252
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.1339126825332642,
      "learning_rate": 0.00020336991033114053,
      "loss": 2.2316,
      "step": 139253
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.0780361890792847,
      "learning_rate": 0.00020336603778221466,
      "loss": 2.0408,
      "step": 139254
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.098278522491455,
      "learning_rate": 0.00020336216525125485,
      "loss": 2.2372,
      "step": 139255
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.4128530025482178,
      "learning_rate": 0.00020335829273826202,
      "loss": 2.2972,
      "step": 139256
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.10910165309906,
      "learning_rate": 0.00020335442024323655,
      "loss": 1.9997,
      "step": 139257
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.0254284143447876,
      "learning_rate": 0.00020335054776617932,
      "loss": 2.0649,
      "step": 139258
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.1397356986999512,
      "learning_rate": 0.00020334667530709106,
      "loss": 2.2945,
      "step": 139259
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.0733985900878906,
      "learning_rate": 0.00020334280286597243,
      "loss": 2.2752,
      "step": 139260
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.455424189567566,
      "learning_rate": 0.00020333893044282425,
      "loss": 2.1103,
      "step": 139261
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.3170613050460815,
      "learning_rate": 0.00020333505803764716,
      "loss": 2.1174,
      "step": 139262
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.0934520959854126,
      "learning_rate": 0.00020333118565044202,
      "loss": 2.1873,
      "step": 139263
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.1685571670532227,
      "learning_rate": 0.00020332731328120932,
      "loss": 1.9808,
      "step": 139264
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.3580992221832275,
      "learning_rate": 0.00020332344092994993,
      "loss": 2.151,
      "step": 139265
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.2251601219177246,
      "learning_rate": 0.00020331956859666453,
      "loss": 1.9988,
      "step": 139266
    },
    {
      "epoch": 1.81,
      "grad_norm": 0.9453991651535034,
      "learning_rate": 0.00020331569628135384,
      "loss": 2.0038,
      "step": 139267
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.2123610973358154,
      "learning_rate": 0.00020331182398401857,
      "loss": 2.1237,
      "step": 139268
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.2719982862472534,
      "learning_rate": 0.00020330795170465956,
      "loss": 2.1488,
      "step": 139269
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.1744334697723389,
      "learning_rate": 0.00020330407944327736,
      "loss": 2.0002,
      "step": 139270
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.0494033098220825,
      "learning_rate": 0.00020330020719987275,
      "loss": 2.0395,
      "step": 139271
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.0412421226501465,
      "learning_rate": 0.00020329633497444643,
      "loss": 2.0899,
      "step": 139272
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.3382774591445923,
      "learning_rate": 0.00020329246276699914,
      "loss": 2.1618,
      "step": 139273
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.0317320823669434,
      "learning_rate": 0.00020328859057753163,
      "loss": 2.0977,
      "step": 139274
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.0726746320724487,
      "learning_rate": 0.00020328471840604472,
      "loss": 2.1627,
      "step": 139275
    },
    {
      "epoch": 1.81,
      "grad_norm": 0.9948846101760864,
      "learning_rate": 0.00020328084625253887,
      "loss": 1.9211,
      "step": 139276
    },
    {
      "epoch": 1.81,
      "grad_norm": 0.9862766265869141,
      "learning_rate": 0.00020327697411701493,
      "loss": 2.2078,
      "step": 139277
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.1128438711166382,
      "learning_rate": 0.00020327310199947366,
      "loss": 2.0769,
      "step": 139278
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.0653566122055054,
      "learning_rate": 0.0002032692298999157,
      "loss": 2.2472,
      "step": 139279
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.0209022760391235,
      "learning_rate": 0.00020326535781834182,
      "loss": 1.9269,
      "step": 139280
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.039157509803772,
      "learning_rate": 0.00020326148575475287,
      "loss": 1.913,
      "step": 139281
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.010637879371643,
      "learning_rate": 0.00020325761370914935,
      "loss": 2.3587,
      "step": 139282
    },
    {
      "epoch": 1.81,
      "grad_norm": 0.9725031852722168,
      "learning_rate": 0.00020325374168153205,
      "loss": 1.8349,
      "step": 139283
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.1769027709960938,
      "learning_rate": 0.0002032498696719017,
      "loss": 2.1914,
      "step": 139284
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.0353853702545166,
      "learning_rate": 0.00020324599768025903,
      "loss": 1.9914,
      "step": 139285
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.1985465288162231,
      "learning_rate": 0.00020324212570660473,
      "loss": 2.1862,
      "step": 139286
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.151482105255127,
      "learning_rate": 0.00020323825375093962,
      "loss": 2.0566,
      "step": 139287
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.0760886669158936,
      "learning_rate": 0.00020323438181326427,
      "loss": 2.0437,
      "step": 139288
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.1075605154037476,
      "learning_rate": 0.00020323050989357955,
      "loss": 2.2902,
      "step": 139289
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.0671652555465698,
      "learning_rate": 0.00020322663799188608,
      "loss": 1.909,
      "step": 139290
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.0735487937927246,
      "learning_rate": 0.00020322276610818454,
      "loss": 2.1383,
      "step": 139291
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.2295318841934204,
      "learning_rate": 0.00020321889424247577,
      "loss": 2.1259,
      "step": 139292
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.0957773923873901,
      "learning_rate": 0.00020321502239476047,
      "loss": 2.1829,
      "step": 139293
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.017122745513916,
      "learning_rate": 0.00020321115056503924,
      "loss": 1.9751,
      "step": 139294
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.0422590970993042,
      "learning_rate": 0.00020320727875331292,
      "loss": 2.1006,
      "step": 139295
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.1666280031204224,
      "learning_rate": 0.00020320340695958224,
      "loss": 1.9803,
      "step": 139296
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.2584280967712402,
      "learning_rate": 0.0002031995351838478,
      "loss": 1.9819,
      "step": 139297
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.0326963663101196,
      "learning_rate": 0.0002031956634261104,
      "loss": 2.1173,
      "step": 139298
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.2256739139556885,
      "learning_rate": 0.00020319179168637087,
      "loss": 1.9638,
      "step": 139299
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.1052732467651367,
      "learning_rate": 0.00020318791996462967,
      "loss": 1.9933,
      "step": 139300
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.147873878479004,
      "learning_rate": 0.0002031840482608877,
      "loss": 2.143,
      "step": 139301
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.0755873918533325,
      "learning_rate": 0.00020318017657514563,
      "loss": 2.0337,
      "step": 139302
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.1092442274093628,
      "learning_rate": 0.0002031763049074043,
      "loss": 2.0935,
      "step": 139303
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.0837124586105347,
      "learning_rate": 0.0002031724332576642,
      "loss": 1.9371,
      "step": 139304
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.1221567392349243,
      "learning_rate": 0.00020316856162592628,
      "loss": 2.1851,
      "step": 139305
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.2632386684417725,
      "learning_rate": 0.00020316469001219106,
      "loss": 2.2061,
      "step": 139306
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.1248942613601685,
      "learning_rate": 0.00020316081841645936,
      "loss": 2.1047,
      "step": 139307
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.1294853687286377,
      "learning_rate": 0.00020315694683873192,
      "loss": 2.1736,
      "step": 139308
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.1174678802490234,
      "learning_rate": 0.00020315307527900937,
      "loss": 2.05,
      "step": 139309
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.1158608198165894,
      "learning_rate": 0.00020314920373729254,
      "loss": 2.2983,
      "step": 139310
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.1753482818603516,
      "learning_rate": 0.0002031453322135822,
      "loss": 2.0846,
      "step": 139311
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.0125939846038818,
      "learning_rate": 0.00020314146070787888,
      "loss": 2.1693,
      "step": 139312
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.0349459648132324,
      "learning_rate": 0.00020313758922018335,
      "loss": 2.2323,
      "step": 139313
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.143717885017395,
      "learning_rate": 0.0002031337177504964,
      "loss": 2.1683,
      "step": 139314
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.1571526527404785,
      "learning_rate": 0.00020312984629881876,
      "loss": 2.3602,
      "step": 139315
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.0600764751434326,
      "learning_rate": 0.00020312597486515102,
      "loss": 2.2904,
      "step": 139316
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.081089735031128,
      "learning_rate": 0.00020312210344949417,
      "loss": 2.0114,
      "step": 139317
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.0437694787979126,
      "learning_rate": 0.0002031182320518486,
      "loss": 1.9621,
      "step": 139318
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.153458833694458,
      "learning_rate": 0.0002031143606722152,
      "loss": 2.029,
      "step": 139319
    },
    {
      "epoch": 1.81,
      "grad_norm": 0.998065710067749,
      "learning_rate": 0.00020311048931059465,
      "loss": 1.9584,
      "step": 139320
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.0212997198104858,
      "learning_rate": 0.00020310661796698768,
      "loss": 1.8974,
      "step": 139321
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.0342810153961182,
      "learning_rate": 0.00020310274664139504,
      "loss": 2.1108,
      "step": 139322
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.065748691558838,
      "learning_rate": 0.00020309887533381745,
      "loss": 2.3662,
      "step": 139323
    },
    {
      "epoch": 1.81,
      "grad_norm": 0.932079017162323,
      "learning_rate": 0.00020309500404425569,
      "loss": 1.917,
      "step": 139324
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.0068734884262085,
      "learning_rate": 0.0002030911327727103,
      "loss": 1.9468,
      "step": 139325
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.041911244392395,
      "learning_rate": 0.00020308726151918209,
      "loss": 2.3562,
      "step": 139326
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.0137355327606201,
      "learning_rate": 0.0002030833902836718,
      "loss": 2.023,
      "step": 139327
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.005600094795227,
      "learning_rate": 0.0002030795190661801,
      "loss": 2.0794,
      "step": 139328
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.0501153469085693,
      "learning_rate": 0.00020307564786670778,
      "loss": 2.3312,
      "step": 139329
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.076005458831787,
      "learning_rate": 0.00020307177668525564,
      "loss": 2.2084,
      "step": 139330
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.0216667652130127,
      "learning_rate": 0.00020306790552182416,
      "loss": 2.0356,
      "step": 139331
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.1007053852081299,
      "learning_rate": 0.0002030640343764142,
      "loss": 2.0581,
      "step": 139332
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.0195780992507935,
      "learning_rate": 0.00020306016324902646,
      "loss": 1.9099,
      "step": 139333
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.0706440210342407,
      "learning_rate": 0.00020305629213966168,
      "loss": 2.3075,
      "step": 139334
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.0978094339370728,
      "learning_rate": 0.00020305242104832056,
      "loss": 2.1581,
      "step": 139335
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.1373614072799683,
      "learning_rate": 0.00020304854997500393,
      "loss": 2.1395,
      "step": 139336
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.0494740009307861,
      "learning_rate": 0.00020304467891971227,
      "loss": 2.1583,
      "step": 139337
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.067713737487793,
      "learning_rate": 0.00020304080788244647,
      "loss": 1.945,
      "step": 139338
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.0966235399246216,
      "learning_rate": 0.00020303693686320718,
      "loss": 2.1357,
      "step": 139339
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.0615084171295166,
      "learning_rate": 0.00020303306586199514,
      "loss": 2.0994,
      "step": 139340
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.104249358177185,
      "learning_rate": 0.00020302919487881116,
      "loss": 2.3054,
      "step": 139341
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.091849446296692,
      "learning_rate": 0.00020302532391365592,
      "loss": 1.8081,
      "step": 139342
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.0983810424804688,
      "learning_rate": 0.00020302145296653004,
      "loss": 2.0414,
      "step": 139343
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.0658472776412964,
      "learning_rate": 0.0002030175820374343,
      "loss": 2.0353,
      "step": 139344
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.1099365949630737,
      "learning_rate": 0.0002030137111263694,
      "loss": 2.0659,
      "step": 139345
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.2062047719955444,
      "learning_rate": 0.00020300984023333607,
      "loss": 2.1992,
      "step": 139346
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.0144518613815308,
      "learning_rate": 0.00020300596935833503,
      "loss": 2.0807,
      "step": 139347
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.0677660703659058,
      "learning_rate": 0.00020300209850136722,
      "loss": 2.1073,
      "step": 139348
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.0107718706130981,
      "learning_rate": 0.00020299822766243296,
      "loss": 2.1434,
      "step": 139349
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.112060785293579,
      "learning_rate": 0.00020299435684153314,
      "loss": 2.2199,
      "step": 139350
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.0377696752548218,
      "learning_rate": 0.00020299048603866852,
      "loss": 2.0595,
      "step": 139351
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.1427878141403198,
      "learning_rate": 0.00020298661525383982,
      "loss": 2.0593,
      "step": 139352
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.2038437128067017,
      "learning_rate": 0.00020298274448704774,
      "loss": 2.1922,
      "step": 139353
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.088839054107666,
      "learning_rate": 0.00020297887373829312,
      "loss": 2.1484,
      "step": 139354
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.0888422727584839,
      "learning_rate": 0.00020297500300757643,
      "loss": 1.9071,
      "step": 139355
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.237033486366272,
      "learning_rate": 0.0002029711322948985,
      "loss": 2.1302,
      "step": 139356
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.0930519104003906,
      "learning_rate": 0.00020296726160026011,
      "loss": 1.9411,
      "step": 139357
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.056182861328125,
      "learning_rate": 0.00020296339092366186,
      "loss": 2.1489,
      "step": 139358
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.084140419960022,
      "learning_rate": 0.00020295952026510463,
      "loss": 1.9971,
      "step": 139359
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.2909294366836548,
      "learning_rate": 0.00020295564962458917,
      "loss": 2.0537,
      "step": 139360
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.1338703632354736,
      "learning_rate": 0.0002029517790021159,
      "loss": 2.1011,
      "step": 139361
    },
    {
      "epoch": 1.81,
      "grad_norm": 0.9983099699020386,
      "learning_rate": 0.0002029479083976858,
      "loss": 2.1299,
      "step": 139362
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.2240493297576904,
      "learning_rate": 0.00020294403781129946,
      "loss": 1.9912,
      "step": 139363
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.611098289489746,
      "learning_rate": 0.0002029401672429577,
      "loss": 2.0983,
      "step": 139364
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.0772792100906372,
      "learning_rate": 0.00020293629669266116,
      "loss": 2.1088,
      "step": 139365
    },
    {
      "epoch": 1.81,
      "grad_norm": 0.9637542366981506,
      "learning_rate": 0.0002029324261604107,
      "loss": 1.9498,
      "step": 139366
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.1317838430404663,
      "learning_rate": 0.00020292855564620687,
      "loss": 2.039,
      "step": 139367
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.1738955974578857,
      "learning_rate": 0.00020292468515005043,
      "loss": 2.0439,
      "step": 139368
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.035578727722168,
      "learning_rate": 0.00020292081467194212,
      "loss": 2.1334,
      "step": 139369
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.0633695125579834,
      "learning_rate": 0.00020291694421188268,
      "loss": 2.0542,
      "step": 139370
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.0580006837844849,
      "learning_rate": 0.0002029130737698728,
      "loss": 2.1944,
      "step": 139371
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.056533932685852,
      "learning_rate": 0.00020290920334591325,
      "loss": 1.9918,
      "step": 139372
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.2106304168701172,
      "learning_rate": 0.00020290533294000474,
      "loss": 1.8393,
      "step": 139373
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.1989833116531372,
      "learning_rate": 0.0002029014625521479,
      "loss": 2.0102,
      "step": 139374
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.0749372243881226,
      "learning_rate": 0.0002028975921823435,
      "loss": 2.1423,
      "step": 139375
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.0203951597213745,
      "learning_rate": 0.00020289372183059226,
      "loss": 2.0497,
      "step": 139376
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.2006429433822632,
      "learning_rate": 0.00020288985149689493,
      "loss": 2.1086,
      "step": 139377
    },
    {
      "epoch": 1.81,
      "grad_norm": 0.9439456462860107,
      "learning_rate": 0.0002028859811812523,
      "loss": 2.1194,
      "step": 139378
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.0390477180480957,
      "learning_rate": 0.0002028821108836649,
      "loss": 2.1645,
      "step": 139379
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.0482922792434692,
      "learning_rate": 0.00020287824060413352,
      "loss": 2.0125,
      "step": 139380
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.0558810234069824,
      "learning_rate": 0.00020287437034265901,
      "loss": 2.037,
      "step": 139381
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.1119037866592407,
      "learning_rate": 0.0002028705000992419,
      "loss": 2.3641,
      "step": 139382
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.0914244651794434,
      "learning_rate": 0.00020286662987388304,
      "loss": 2.1599,
      "step": 139383
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.0833595991134644,
      "learning_rate": 0.00020286275966658313,
      "loss": 2.1443,
      "step": 139384
    },
    {
      "epoch": 1.81,
      "grad_norm": 0.9610511064529419,
      "learning_rate": 0.00020285888947734282,
      "loss": 2.1658,
      "step": 139385
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.1433240175247192,
      "learning_rate": 0.0002028550193061629,
      "loss": 2.2467,
      "step": 139386
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.2391812801361084,
      "learning_rate": 0.00020285114915304401,
      "loss": 2.2418,
      "step": 139387
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.734771966934204,
      "learning_rate": 0.00020284727901798705,
      "loss": 2.2191,
      "step": 139388
    },
    {
      "epoch": 1.81,
      "grad_norm": 0.9691565036773682,
      "learning_rate": 0.00020284340890099252,
      "loss": 2.1135,
      "step": 139389
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.188342571258545,
      "learning_rate": 0.00020283953880206122,
      "loss": 2.1114,
      "step": 139390
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.2577755451202393,
      "learning_rate": 0.00020283566872119396,
      "loss": 2.1823,
      "step": 139391
    },
    {
      "epoch": 1.81,
      "grad_norm": 0.9342668652534485,
      "learning_rate": 0.00020283179865839132,
      "loss": 2.1621,
      "step": 139392
    },
    {
      "epoch": 1.81,
      "grad_norm": 1.0374559164047241,
      "learning_rate": 0.0002028279286136541,
      "loss": 2.0486,
      "step": 139393
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.1306384801864624,
      "learning_rate": 0.000202824058586983,
      "loss": 2.2086,
      "step": 139394
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.0490586757659912,
      "learning_rate": 0.00020282018857837878,
      "loss": 1.8932,
      "step": 139395
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.129475712776184,
      "learning_rate": 0.00020281631858784204,
      "loss": 1.7776,
      "step": 139396
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.0502521991729736,
      "learning_rate": 0.00020281244861537369,
      "loss": 2.2573,
      "step": 139397
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.05832040309906,
      "learning_rate": 0.0002028085786609743,
      "loss": 2.3466,
      "step": 139398
    },
    {
      "epoch": 1.82,
      "grad_norm": 0.9717706441879272,
      "learning_rate": 0.00020280470872464457,
      "loss": 2.1903,
      "step": 139399
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.0669169425964355,
      "learning_rate": 0.00020280083880638527,
      "loss": 2.1243,
      "step": 139400
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.022721767425537,
      "learning_rate": 0.00020279696890619718,
      "loss": 2.2387,
      "step": 139401
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.0463985204696655,
      "learning_rate": 0.00020279309902408098,
      "loss": 2.22,
      "step": 139402
    },
    {
      "epoch": 1.82,
      "grad_norm": 0.9364591836929321,
      "learning_rate": 0.00020278922916003745,
      "loss": 1.974,
      "step": 139403
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.160812258720398,
      "learning_rate": 0.0002027853593140671,
      "loss": 2.2206,
      "step": 139404
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.2848875522613525,
      "learning_rate": 0.00020278148948617083,
      "loss": 2.1221,
      "step": 139405
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.1143548488616943,
      "learning_rate": 0.0002027776196763493,
      "loss": 2.1012,
      "step": 139406
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.0988572835922241,
      "learning_rate": 0.00020277374988460326,
      "loss": 1.8937,
      "step": 139407
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.3532061576843262,
      "learning_rate": 0.00020276988011093343,
      "loss": 2.169,
      "step": 139408
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.1532903909683228,
      "learning_rate": 0.00020276601035534058,
      "loss": 2.2196,
      "step": 139409
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.062084674835205,
      "learning_rate": 0.0002027621406178253,
      "loss": 1.9816,
      "step": 139410
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.0668131113052368,
      "learning_rate": 0.00020275827089838834,
      "loss": 1.8309,
      "step": 139411
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.1050214767456055,
      "learning_rate": 0.00020275440119703046,
      "loss": 2.1764,
      "step": 139412
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.1538646221160889,
      "learning_rate": 0.00020275053151375239,
      "loss": 2.1839,
      "step": 139413
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.0311394929885864,
      "learning_rate": 0.00020274666184855482,
      "loss": 2.1633,
      "step": 139414
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.0858094692230225,
      "learning_rate": 0.0002027427922014386,
      "loss": 2.0851,
      "step": 139415
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.0743252038955688,
      "learning_rate": 0.00020273892257240418,
      "loss": 1.9927,
      "step": 139416
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.0638076066970825,
      "learning_rate": 0.0002027350529614525,
      "loss": 2.1438,
      "step": 139417
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.049301266670227,
      "learning_rate": 0.00020273118336858417,
      "loss": 2.07,
      "step": 139418
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.0041673183441162,
      "learning_rate": 0.00020272731379379993,
      "loss": 2.0928,
      "step": 139419
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.010448932647705,
      "learning_rate": 0.00020272344423710053,
      "loss": 2.0666,
      "step": 139420
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.068060040473938,
      "learning_rate": 0.00020271957469848682,
      "loss": 1.9803,
      "step": 139421
    },
    {
      "epoch": 1.82,
      "grad_norm": 0.9563696384429932,
      "learning_rate": 0.00020271570517795922,
      "loss": 1.982,
      "step": 139422
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.103818416595459,
      "learning_rate": 0.00020271183567551864,
      "loss": 2.1202,
      "step": 139423
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.0879195928573608,
      "learning_rate": 0.00020270796619116575,
      "loss": 1.9202,
      "step": 139424
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.2184385061264038,
      "learning_rate": 0.00020270409672490128,
      "loss": 2.1696,
      "step": 139425
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.0462995767593384,
      "learning_rate": 0.00020270022727672597,
      "loss": 2.1367,
      "step": 139426
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.0849580764770508,
      "learning_rate": 0.00020269635784664068,
      "loss": 2.0421,
      "step": 139427
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.0220834016799927,
      "learning_rate": 0.0002026924884346458,
      "loss": 1.9729,
      "step": 139428
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.0420609712600708,
      "learning_rate": 0.00020268861904074222,
      "loss": 2.0463,
      "step": 139429
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.0971405506134033,
      "learning_rate": 0.00020268474966493068,
      "loss": 2.0332,
      "step": 139430
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.0725445747375488,
      "learning_rate": 0.0002026808803072119,
      "loss": 2.1601,
      "step": 139431
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.2272944450378418,
      "learning_rate": 0.00020267701096758652,
      "loss": 2.1933,
      "step": 139432
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.2621891498565674,
      "learning_rate": 0.0002026731416460555,
      "loss": 1.9391,
      "step": 139433
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.1047143936157227,
      "learning_rate": 0.00020266927234261926,
      "loss": 1.9268,
      "step": 139434
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.1769695281982422,
      "learning_rate": 0.00020266540305727862,
      "loss": 1.9224,
      "step": 139435
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.0895578861236572,
      "learning_rate": 0.0002026615337900343,
      "loss": 2.0968,
      "step": 139436
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.2807743549346924,
      "learning_rate": 0.00020265766454088708,
      "loss": 2.1382,
      "step": 139437
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.0818886756896973,
      "learning_rate": 0.00020265379530983764,
      "loss": 2.1656,
      "step": 139438
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.2079098224639893,
      "learning_rate": 0.0002026499260968868,
      "loss": 2.2372,
      "step": 139439
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.202014446258545,
      "learning_rate": 0.00020264605690203504,
      "loss": 1.7165,
      "step": 139440
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.3791035413742065,
      "learning_rate": 0.00020264218772528324,
      "loss": 2.0944,
      "step": 139441
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.1756638288497925,
      "learning_rate": 0.00020263831856663208,
      "loss": 2.0061,
      "step": 139442
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.0547536611557007,
      "learning_rate": 0.0002026344494260823,
      "loss": 1.8577,
      "step": 139443
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.062909722328186,
      "learning_rate": 0.0002026305803036346,
      "loss": 2.0125,
      "step": 139444
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.1652929782867432,
      "learning_rate": 0.00020262671119928986,
      "loss": 2.0553,
      "step": 139445
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.0782800912857056,
      "learning_rate": 0.0002026228421130485,
      "loss": 2.0477,
      "step": 139446
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.177579402923584,
      "learning_rate": 0.0002026189730449114,
      "loss": 1.9164,
      "step": 139447
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.1968096494674683,
      "learning_rate": 0.0002026151039948793,
      "loss": 2.1349,
      "step": 139448
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.0718040466308594,
      "learning_rate": 0.00020261123496295285,
      "loss": 2.0979,
      "step": 139449
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.1232562065124512,
      "learning_rate": 0.00020260736594913283,
      "loss": 2.0685,
      "step": 139450
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.0252994298934937,
      "learning_rate": 0.00020260349695342008,
      "loss": 2.0044,
      "step": 139451
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.0141509771347046,
      "learning_rate": 0.00020259962797581501,
      "loss": 2.0645,
      "step": 139452
    },
    {
      "epoch": 1.82,
      "grad_norm": 0.9922057390213013,
      "learning_rate": 0.00020259575901631854,
      "loss": 2.0082,
      "step": 139453
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.08566415309906,
      "learning_rate": 0.00020259189007493136,
      "loss": 2.0185,
      "step": 139454
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.129273533821106,
      "learning_rate": 0.0002025880211516542,
      "loss": 2.2503,
      "step": 139455
    },
    {
      "epoch": 1.82,
      "grad_norm": 0.9712696075439453,
      "learning_rate": 0.00020258415224648773,
      "loss": 2.1607,
      "step": 139456
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.040378451347351,
      "learning_rate": 0.0002025802833594327,
      "loss": 1.754,
      "step": 139457
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.0617631673812866,
      "learning_rate": 0.00020257641449049,
      "loss": 2.2955,
      "step": 139458
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.1132066249847412,
      "learning_rate": 0.00020257254563966,
      "loss": 2.1563,
      "step": 139459
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.039522647857666,
      "learning_rate": 0.00020256867680694366,
      "loss": 1.9406,
      "step": 139460
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.047993779182434,
      "learning_rate": 0.00020256480799234165,
      "loss": 2.188,
      "step": 139461
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.0665533542633057,
      "learning_rate": 0.00020256093919585462,
      "loss": 2.0275,
      "step": 139462
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.065449595451355,
      "learning_rate": 0.00020255707041748341,
      "loss": 1.9279,
      "step": 139463
    },
    {
      "epoch": 1.82,
      "grad_norm": 0.9584364295005798,
      "learning_rate": 0.00020255320165722869,
      "loss": 1.8843,
      "step": 139464
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.1679027080535889,
      "learning_rate": 0.00020254933291509122,
      "loss": 2.0599,
      "step": 139465
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.0401997566223145,
      "learning_rate": 0.00020254546419107156,
      "loss": 2.1794,
      "step": 139466
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.133715271949768,
      "learning_rate": 0.00020254159548517057,
      "loss": 2.1538,
      "step": 139467
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.3107342720031738,
      "learning_rate": 0.0002025377267973889,
      "loss": 2.447,
      "step": 139468
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.210109829902649,
      "learning_rate": 0.00020253385812772734,
      "loss": 2.1413,
      "step": 139469
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.1148494482040405,
      "learning_rate": 0.00020252998947618663,
      "loss": 2.215,
      "step": 139470
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.0702217817306519,
      "learning_rate": 0.0002025261208427674,
      "loss": 2.0255,
      "step": 139471
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.134800672531128,
      "learning_rate": 0.0002025222522274704,
      "loss": 2.1127,
      "step": 139472
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.3289763927459717,
      "learning_rate": 0.00020251838363029637,
      "loss": 2.1981,
      "step": 139473
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.289470911026001,
      "learning_rate": 0.00020251451505124595,
      "loss": 1.8744,
      "step": 139474
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.0847803354263306,
      "learning_rate": 0.00020251064649031995,
      "loss": 2.1955,
      "step": 139475
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.107176423072815,
      "learning_rate": 0.0002025067779475191,
      "loss": 2.0774,
      "step": 139476
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.2282103300094604,
      "learning_rate": 0.00020250290942284404,
      "loss": 2.1894,
      "step": 139477
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.109415054321289,
      "learning_rate": 0.00020249904091629555,
      "loss": 2.1075,
      "step": 139478
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.0703054666519165,
      "learning_rate": 0.00020249517242787427,
      "loss": 1.9984,
      "step": 139479
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.0531127452850342,
      "learning_rate": 0.00020249130395758112,
      "loss": 2.0869,
      "step": 139480
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.1696799993515015,
      "learning_rate": 0.00020248743550541656,
      "loss": 2.2316,
      "step": 139481
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.0621538162231445,
      "learning_rate": 0.00020248356707138146,
      "loss": 2.1168,
      "step": 139482
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.0646387338638306,
      "learning_rate": 0.00020247969865547653,
      "loss": 1.9764,
      "step": 139483
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.1112453937530518,
      "learning_rate": 0.0002024758302577024,
      "loss": 2.3999,
      "step": 139484
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.123633861541748,
      "learning_rate": 0.00020247196187805986,
      "loss": 1.8956,
      "step": 139485
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.1590452194213867,
      "learning_rate": 0.00020246809351654962,
      "loss": 2.0508,
      "step": 139486
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.0515093803405762,
      "learning_rate": 0.0002024642251731725,
      "loss": 1.9624,
      "step": 139487
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.2161294221878052,
      "learning_rate": 0.00020246035684792912,
      "loss": 2.0476,
      "step": 139488
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.105644702911377,
      "learning_rate": 0.0002024564885408201,
      "loss": 2.0688,
      "step": 139489
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.187058687210083,
      "learning_rate": 0.0002024526202518463,
      "loss": 2.1127,
      "step": 139490
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.0603350400924683,
      "learning_rate": 0.00020244875198100837,
      "loss": 2.1224,
      "step": 139491
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.0775225162506104,
      "learning_rate": 0.00020244488372830706,
      "loss": 2.2468,
      "step": 139492
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.118151068687439,
      "learning_rate": 0.0002024410154937431,
      "loss": 1.9282,
      "step": 139493
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.0605082511901855,
      "learning_rate": 0.00020243714727731732,
      "loss": 2.0205,
      "step": 139494
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.0361722707748413,
      "learning_rate": 0.00020243327907903017,
      "loss": 1.8652,
      "step": 139495
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.0091216564178467,
      "learning_rate": 0.00020242941089888253,
      "loss": 2.2318,
      "step": 139496
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.2708038091659546,
      "learning_rate": 0.0002024255427368751,
      "loss": 2.0756,
      "step": 139497
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.0308586359024048,
      "learning_rate": 0.0002024216745930086,
      "loss": 2.3534,
      "step": 139498
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.075629711151123,
      "learning_rate": 0.0002024178064672838,
      "loss": 2.2494,
      "step": 139499
    },
    {
      "epoch": 1.82,
      "grad_norm": 0.9817773699760437,
      "learning_rate": 0.00020241393835970144,
      "loss": 2.1519,
      "step": 139500
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.079239845275879,
      "learning_rate": 0.00020241007027026208,
      "loss": 2.1221,
      "step": 139501
    },
    {
      "epoch": 1.82,
      "grad_norm": 0.9357197284698486,
      "learning_rate": 0.00020240620219896647,
      "loss": 1.9084,
      "step": 139502
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.0722399950027466,
      "learning_rate": 0.00020240233414581547,
      "loss": 2.2085,
      "step": 139503
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.0869147777557373,
      "learning_rate": 0.00020239846611080968,
      "loss": 2.1892,
      "step": 139504
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.1380257606506348,
      "learning_rate": 0.00020239459809394986,
      "loss": 2.1386,
      "step": 139505
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.1125802993774414,
      "learning_rate": 0.00020239073009523686,
      "loss": 2.1919,
      "step": 139506
    },
    {
      "epoch": 1.82,
      "grad_norm": 0.9947782754898071,
      "learning_rate": 0.00020238686211467114,
      "loss": 1.9256,
      "step": 139507
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.06050705909729,
      "learning_rate": 0.00020238299415225355,
      "loss": 1.9213,
      "step": 139508
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.1719087362289429,
      "learning_rate": 0.00020237912620798477,
      "loss": 2.2972,
      "step": 139509
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.0268834829330444,
      "learning_rate": 0.00020237525828186557,
      "loss": 2.0778,
      "step": 139510
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.0125476121902466,
      "learning_rate": 0.00020237139037389666,
      "loss": 1.9642,
      "step": 139511
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.165987491607666,
      "learning_rate": 0.00020236752248407885,
      "loss": 2.213,
      "step": 139512
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.1058329343795776,
      "learning_rate": 0.0002023636546124127,
      "loss": 2.2331,
      "step": 139513
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.1132797002792358,
      "learning_rate": 0.00020235978675889893,
      "loss": 2.0707,
      "step": 139514
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.1753002405166626,
      "learning_rate": 0.00020235591892353835,
      "loss": 2.168,
      "step": 139515
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.032534122467041,
      "learning_rate": 0.0002023520511063316,
      "loss": 2.3927,
      "step": 139516
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.0865752696990967,
      "learning_rate": 0.0002023481833072795,
      "loss": 2.1418,
      "step": 139517
    },
    {
      "epoch": 1.82,
      "grad_norm": 0.9831040501594543,
      "learning_rate": 0.00020234431552638283,
      "loss": 2.0266,
      "step": 139518
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.161216139793396,
      "learning_rate": 0.00020234044776364205,
      "loss": 2.1808,
      "step": 139519
    },
    {
      "epoch": 1.82,
      "grad_norm": 0.9792501926422119,
      "learning_rate": 0.00020233658001905805,
      "loss": 1.9566,
      "step": 139520
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.0535948276519775,
      "learning_rate": 0.0002023327122926315,
      "loss": 2.2568,
      "step": 139521
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.0568127632141113,
      "learning_rate": 0.00020232884458436319,
      "loss": 2.0392,
      "step": 139522
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.1204440593719482,
      "learning_rate": 0.00020232497689425373,
      "loss": 2.2347,
      "step": 139523
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.0820930004119873,
      "learning_rate": 0.00020232110922230393,
      "loss": 2.1753,
      "step": 139524
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.118147373199463,
      "learning_rate": 0.00020231724156851458,
      "loss": 2.0301,
      "step": 139525
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.1243999004364014,
      "learning_rate": 0.00020231337393288623,
      "loss": 2.0224,
      "step": 139526
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.1401989459991455,
      "learning_rate": 0.0002023095063154196,
      "loss": 2.0953,
      "step": 139527
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.2339612245559692,
      "learning_rate": 0.00020230563871611552,
      "loss": 2.1643,
      "step": 139528
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.046189308166504,
      "learning_rate": 0.00020230177113497464,
      "loss": 2.1568,
      "step": 139529
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.1950651407241821,
      "learning_rate": 0.00020229790357199776,
      "loss": 2.0899,
      "step": 139530
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.1572725772857666,
      "learning_rate": 0.0002022940360271856,
      "loss": 2.2298,
      "step": 139531
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.146597981452942,
      "learning_rate": 0.00020229016850053873,
      "loss": 2.168,
      "step": 139532
    },
    {
      "epoch": 1.82,
      "grad_norm": 0.9860937595367432,
      "learning_rate": 0.00020228630099205795,
      "loss": 2.0249,
      "step": 139533
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.1300042867660522,
      "learning_rate": 0.000202282433501744,
      "loss": 2.1043,
      "step": 139534
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.0536322593688965,
      "learning_rate": 0.00020227856602959759,
      "loss": 1.9496,
      "step": 139535
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.177571177482605,
      "learning_rate": 0.00020227469857561945,
      "loss": 1.9469,
      "step": 139536
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.0002973079681396,
      "learning_rate": 0.0002022708311398104,
      "loss": 2.374,
      "step": 139537
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.0563793182373047,
      "learning_rate": 0.00020226696372217093,
      "loss": 2.2086,
      "step": 139538
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.1616779565811157,
      "learning_rate": 0.00020226309632270185,
      "loss": 2.1791,
      "step": 139539
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.1401903629302979,
      "learning_rate": 0.0002022592289414039,
      "loss": 2.0131,
      "step": 139540
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.0087904930114746,
      "learning_rate": 0.00020225536157827788,
      "loss": 2.1402,
      "step": 139541
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.0727102756500244,
      "learning_rate": 0.00020225149423332435,
      "loss": 2.2533,
      "step": 139542
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.0676960945129395,
      "learning_rate": 0.00020224762690654427,
      "loss": 2.1411,
      "step": 139543
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.266968846321106,
      "learning_rate": 0.00020224375959793813,
      "loss": 1.9671,
      "step": 139544
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.223561406135559,
      "learning_rate": 0.00020223989230750664,
      "loss": 2.0203,
      "step": 139545
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.1139706373214722,
      "learning_rate": 0.00020223602503525063,
      "loss": 2.0383,
      "step": 139546
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.1808737516403198,
      "learning_rate": 0.00020223215778117078,
      "loss": 2.0991,
      "step": 139547
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.075013518333435,
      "learning_rate": 0.00020222829054526782,
      "loss": 2.2547,
      "step": 139548
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.114660382270813,
      "learning_rate": 0.00020222442332754254,
      "loss": 2.1498,
      "step": 139549
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.0162509679794312,
      "learning_rate": 0.00020222055612799557,
      "loss": 1.9485,
      "step": 139550
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.0427963733673096,
      "learning_rate": 0.0002022166889466276,
      "loss": 2.1476,
      "step": 139551
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.2882529497146606,
      "learning_rate": 0.00020221282178343937,
      "loss": 2.2112,
      "step": 139552
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.1823601722717285,
      "learning_rate": 0.00020220895463843165,
      "loss": 2.2485,
      "step": 139553
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.0658761262893677,
      "learning_rate": 0.0002022050875116051,
      "loss": 2.0366,
      "step": 139554
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.2521212100982666,
      "learning_rate": 0.00020220122040296056,
      "loss": 2.2873,
      "step": 139555
    },
    {
      "epoch": 1.82,
      "grad_norm": 0.9416057467460632,
      "learning_rate": 0.0002021973533124986,
      "loss": 1.8155,
      "step": 139556
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.0388820171356201,
      "learning_rate": 0.00020219348624022002,
      "loss": 1.9826,
      "step": 139557
    },
    {
      "epoch": 1.82,
      "grad_norm": 0.990626871585846,
      "learning_rate": 0.0002021896191861255,
      "loss": 2.1277,
      "step": 139558
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.1221603155136108,
      "learning_rate": 0.00020218575215021574,
      "loss": 2.1157,
      "step": 139559
    },
    {
      "epoch": 1.82,
      "grad_norm": 0.9896368980407715,
      "learning_rate": 0.00020218188513249153,
      "loss": 2.199,
      "step": 139560
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.0569796562194824,
      "learning_rate": 0.0002021780181329536,
      "loss": 2.0491,
      "step": 139561
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.0473617315292358,
      "learning_rate": 0.00020217415115160255,
      "loss": 1.939,
      "step": 139562
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.0268900394439697,
      "learning_rate": 0.0002021702841884392,
      "loss": 2.116,
      "step": 139563
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.1801650524139404,
      "learning_rate": 0.00020216641724346424,
      "loss": 2.1025,
      "step": 139564
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.0662435293197632,
      "learning_rate": 0.0002021625503166784,
      "loss": 2.123,
      "step": 139565
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.0403711795806885,
      "learning_rate": 0.00020215868340808235,
      "loss": 2.0186,
      "step": 139566
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.0666834115982056,
      "learning_rate": 0.00020215481651767692,
      "loss": 2.0169,
      "step": 139567
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.0436902046203613,
      "learning_rate": 0.0002021509496454627,
      "loss": 2.1262,
      "step": 139568
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.1976685523986816,
      "learning_rate": 0.00020214708279144044,
      "loss": 2.2052,
      "step": 139569
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.0461660623550415,
      "learning_rate": 0.0002021432159556109,
      "loss": 1.9964,
      "step": 139570
    },
    {
      "epoch": 1.82,
      "grad_norm": 0.9998369812965393,
      "learning_rate": 0.00020213934913797482,
      "loss": 1.9162,
      "step": 139571
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.118751883506775,
      "learning_rate": 0.00020213548233853293,
      "loss": 2.2104,
      "step": 139572
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.295889973640442,
      "learning_rate": 0.00020213161555728584,
      "loss": 1.9541,
      "step": 139573
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.1285455226898193,
      "learning_rate": 0.00020212774879423432,
      "loss": 2.2967,
      "step": 139574
    },
    {
      "epoch": 1.82,
      "grad_norm": 0.99739670753479,
      "learning_rate": 0.00020212388204937908,
      "loss": 2.1895,
      "step": 139575
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.2625477313995361,
      "learning_rate": 0.0002021200153227209,
      "loss": 2.1313,
      "step": 139576
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.0239481925964355,
      "learning_rate": 0.0002021161486142604,
      "loss": 1.9955,
      "step": 139577
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.0976966619491577,
      "learning_rate": 0.0002021122819239984,
      "loss": 2.3474,
      "step": 139578
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.1868547201156616,
      "learning_rate": 0.00020210841525193567,
      "loss": 2.1635,
      "step": 139579
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.0488674640655518,
      "learning_rate": 0.0002021045485980727,
      "loss": 2.1858,
      "step": 139580
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.0555115938186646,
      "learning_rate": 0.00020210068196241038,
      "loss": 2.2799,
      "step": 139581
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.2201520204544067,
      "learning_rate": 0.00020209681534494938,
      "loss": 2.3143,
      "step": 139582
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.0758991241455078,
      "learning_rate": 0.0002020929487456904,
      "loss": 2.187,
      "step": 139583
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.0284531116485596,
      "learning_rate": 0.00020208908216463422,
      "loss": 2.135,
      "step": 139584
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.0174530744552612,
      "learning_rate": 0.00020208521560178164,
      "loss": 1.962,
      "step": 139585
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.018030047416687,
      "learning_rate": 0.0002020813490571332,
      "loss": 1.9039,
      "step": 139586
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.0280994176864624,
      "learning_rate": 0.00020207748253068958,
      "loss": 1.8965,
      "step": 139587
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.10755455493927,
      "learning_rate": 0.00020207361602245165,
      "loss": 1.9797,
      "step": 139588
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.0872809886932373,
      "learning_rate": 0.0002020697495324201,
      "loss": 1.9308,
      "step": 139589
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.0450083017349243,
      "learning_rate": 0.00020206588306059562,
      "loss": 2.0404,
      "step": 139590
    },
    {
      "epoch": 1.82,
      "grad_norm": 0.928658127784729,
      "learning_rate": 0.00020206201660697895,
      "loss": 2.048,
      "step": 139591
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.0233384370803833,
      "learning_rate": 0.00020205815017157092,
      "loss": 2.2308,
      "step": 139592
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.0095953941345215,
      "learning_rate": 0.000202054283754372,
      "loss": 2.0698,
      "step": 139593
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.1208473443984985,
      "learning_rate": 0.00020205041735538303,
      "loss": 2.0269,
      "step": 139594
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.0458855628967285,
      "learning_rate": 0.00020204655097460475,
      "loss": 2.0589,
      "step": 139595
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.026907205581665,
      "learning_rate": 0.00020204268461203786,
      "loss": 2.2254,
      "step": 139596
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.1191953420639038,
      "learning_rate": 0.00020203881826768308,
      "loss": 2.1276,
      "step": 139597
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.0592321157455444,
      "learning_rate": 0.00020203495194154127,
      "loss": 2.2418,
      "step": 139598
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.0084187984466553,
      "learning_rate": 0.00020203108563361286,
      "loss": 1.9398,
      "step": 139599
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.1855441331863403,
      "learning_rate": 0.00020202721934389873,
      "loss": 2.3924,
      "step": 139600
    },
    {
      "epoch": 1.82,
      "grad_norm": 0.9631154537200928,
      "learning_rate": 0.00020202335307239962,
      "loss": 1.9607,
      "step": 139601
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.158506155014038,
      "learning_rate": 0.0002020194868191162,
      "loss": 2.296,
      "step": 139602
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.0260674953460693,
      "learning_rate": 0.0002020156205840492,
      "loss": 2.1953,
      "step": 139603
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.1152476072311401,
      "learning_rate": 0.0002020117543671995,
      "loss": 2.0087,
      "step": 139604
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.0950323343276978,
      "learning_rate": 0.00020200788816856748,
      "loss": 1.9681,
      "step": 139605
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.0228362083435059,
      "learning_rate": 0.0002020040219881541,
      "loss": 2.2387,
      "step": 139606
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.046382188796997,
      "learning_rate": 0.00020200015582596,
      "loss": 1.8752,
      "step": 139607
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.0268011093139648,
      "learning_rate": 0.0002019962896819859,
      "loss": 1.9723,
      "step": 139608
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.1252881288528442,
      "learning_rate": 0.00020199242355623255,
      "loss": 2.2012,
      "step": 139609
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.040718674659729,
      "learning_rate": 0.0002019885574487008,
      "loss": 1.9631,
      "step": 139610
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.1523137092590332,
      "learning_rate": 0.0002019846913593911,
      "loss": 2.0738,
      "step": 139611
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.082677960395813,
      "learning_rate": 0.0002019808252883043,
      "loss": 1.9855,
      "step": 139612
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.0308231115341187,
      "learning_rate": 0.0002019769592354411,
      "loss": 2.2825,
      "step": 139613
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.0046758651733398,
      "learning_rate": 0.0002019730932008022,
      "loss": 1.9632,
      "step": 139614
    },
    {
      "epoch": 1.82,
      "grad_norm": 0.9801113605499268,
      "learning_rate": 0.0002019692271843884,
      "loss": 2.0175,
      "step": 139615
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.0004494190216064,
      "learning_rate": 0.0002019653611862005,
      "loss": 2.1585,
      "step": 139616
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.051832914352417,
      "learning_rate": 0.00020196149520623898,
      "loss": 2.0003,
      "step": 139617
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.1005138158798218,
      "learning_rate": 0.00020195762924450462,
      "loss": 2.0576,
      "step": 139618
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.0878876447677612,
      "learning_rate": 0.00020195376330099822,
      "loss": 2.0231,
      "step": 139619
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.105673909187317,
      "learning_rate": 0.00020194989737572046,
      "loss": 2.0847,
      "step": 139620
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.0743892192840576,
      "learning_rate": 0.00020194603146867206,
      "loss": 2.0042,
      "step": 139621
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.031034231185913,
      "learning_rate": 0.00020194216557985388,
      "loss": 2.1985,
      "step": 139622
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.1900984048843384,
      "learning_rate": 0.00020193829970926636,
      "loss": 2.3071,
      "step": 139623
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.1769146919250488,
      "learning_rate": 0.00020193443385691035,
      "loss": 2.3084,
      "step": 139624
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.0899360179901123,
      "learning_rate": 0.0002019305680227866,
      "loss": 2.3483,
      "step": 139625
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.0938328504562378,
      "learning_rate": 0.0002019267022068958,
      "loss": 2.1315,
      "step": 139626
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.2201894521713257,
      "learning_rate": 0.00020192283640923869,
      "loss": 2.305,
      "step": 139627
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.0704830884933472,
      "learning_rate": 0.0002019189706298161,
      "loss": 2.1576,
      "step": 139628
    },
    {
      "epoch": 1.82,
      "grad_norm": 0.9957531094551086,
      "learning_rate": 0.00020191510486862848,
      "loss": 2.0703,
      "step": 139629
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.0971587896347046,
      "learning_rate": 0.00020191123912567674,
      "loss": 2.1309,
      "step": 139630
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.4645062685012817,
      "learning_rate": 0.00020190737340096152,
      "loss": 1.9465,
      "step": 139631
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.0596203804016113,
      "learning_rate": 0.00020190350769448353,
      "loss": 1.9256,
      "step": 139632
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.0035206079483032,
      "learning_rate": 0.0002018996420062436,
      "loss": 2.0701,
      "step": 139633
    },
    {
      "epoch": 1.82,
      "grad_norm": 0.9811737537384033,
      "learning_rate": 0.0002018957763362424,
      "loss": 2.043,
      "step": 139634
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.0341147184371948,
      "learning_rate": 0.00020189191068448062,
      "loss": 2.0408,
      "step": 139635
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.026018738746643,
      "learning_rate": 0.00020188804505095895,
      "loss": 2.1569,
      "step": 139636
    },
    {
      "epoch": 1.82,
      "grad_norm": 0.9822407364845276,
      "learning_rate": 0.00020188417943567814,
      "loss": 2.158,
      "step": 139637
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.011217713356018,
      "learning_rate": 0.00020188031383863895,
      "loss": 2.0232,
      "step": 139638
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.0094082355499268,
      "learning_rate": 0.000201876448259842,
      "loss": 2.0533,
      "step": 139639
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.0334277153015137,
      "learning_rate": 0.00020187258269928817,
      "loss": 2.1101,
      "step": 139640
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.1944389343261719,
      "learning_rate": 0.000201868717156978,
      "loss": 2.028,
      "step": 139641
    },
    {
      "epoch": 1.82,
      "grad_norm": 0.9455428123474121,
      "learning_rate": 0.00020186485163291236,
      "loss": 1.9983,
      "step": 139642
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.099034309387207,
      "learning_rate": 0.0002018609861270918,
      "loss": 2.1956,
      "step": 139643
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.0549322366714478,
      "learning_rate": 0.00020185712063951715,
      "loss": 2.3906,
      "step": 139644
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.0854600667953491,
      "learning_rate": 0.00020185325517018914,
      "loss": 2.053,
      "step": 139645
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.0851943492889404,
      "learning_rate": 0.0002018493897191085,
      "loss": 1.9069,
      "step": 139646
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.033121109008789,
      "learning_rate": 0.00020184552428627585,
      "loss": 2.044,
      "step": 139647
    },
    {
      "epoch": 1.82,
      "grad_norm": 0.9768929481506348,
      "learning_rate": 0.00020184165887169198,
      "loss": 1.9511,
      "step": 139648
    },
    {
      "epoch": 1.82,
      "grad_norm": 0.999983549118042,
      "learning_rate": 0.00020183779347535767,
      "loss": 2.0229,
      "step": 139649
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.0740281343460083,
      "learning_rate": 0.0002018339280972735,
      "loss": 2.106,
      "step": 139650
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.1156768798828125,
      "learning_rate": 0.00020183006273744025,
      "loss": 2.0902,
      "step": 139651
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.1158957481384277,
      "learning_rate": 0.0002018261973958587,
      "loss": 2.0175,
      "step": 139652
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.279841423034668,
      "learning_rate": 0.00020182233207252943,
      "loss": 2.036,
      "step": 139653
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.0234599113464355,
      "learning_rate": 0.00020181846676745329,
      "loss": 1.9769,
      "step": 139654
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.083006739616394,
      "learning_rate": 0.00020181460148063093,
      "loss": 2.2942,
      "step": 139655
    },
    {
      "epoch": 1.82,
      "grad_norm": 0.9911074638366699,
      "learning_rate": 0.00020181073621206315,
      "loss": 2.0631,
      "step": 139656
    },
    {
      "epoch": 1.82,
      "grad_norm": 0.9169589281082153,
      "learning_rate": 0.00020180687096175054,
      "loss": 1.9807,
      "step": 139657
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.1032369136810303,
      "learning_rate": 0.00020180300572969387,
      "loss": 1.8999,
      "step": 139658
    },
    {
      "epoch": 1.82,
      "grad_norm": 0.9593298435211182,
      "learning_rate": 0.000201799140515894,
      "loss": 2.0511,
      "step": 139659
    },
    {
      "epoch": 1.82,
      "grad_norm": 0.9920545220375061,
      "learning_rate": 0.00020179527532035143,
      "loss": 2.0225,
      "step": 139660
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.1467174291610718,
      "learning_rate": 0.00020179141014306694,
      "loss": 2.161,
      "step": 139661
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.2132325172424316,
      "learning_rate": 0.00020178754498404128,
      "loss": 1.9617,
      "step": 139662
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.05233633518219,
      "learning_rate": 0.0002017836798432752,
      "loss": 2.1464,
      "step": 139663
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.0410091876983643,
      "learning_rate": 0.00020177981472076944,
      "loss": 2.1162,
      "step": 139664
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.1981475353240967,
      "learning_rate": 0.00020177594961652467,
      "loss": 2.0615,
      "step": 139665
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.1413729190826416,
      "learning_rate": 0.00020177208453054155,
      "loss": 2.1884,
      "step": 139666
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.303589105606079,
      "learning_rate": 0.00020176821946282084,
      "loss": 2.0622,
      "step": 139667
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.0536606311798096,
      "learning_rate": 0.0002017643544133633,
      "loss": 1.8713,
      "step": 139668
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.058990240097046,
      "learning_rate": 0.0002017604893821696,
      "loss": 2.0826,
      "step": 139669
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.0387928485870361,
      "learning_rate": 0.00020175662436924046,
      "loss": 2.2696,
      "step": 139670
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.1350566148757935,
      "learning_rate": 0.00020175275937457679,
      "loss": 2.2195,
      "step": 139671
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.1602637767791748,
      "learning_rate": 0.00020174889439817894,
      "loss": 2.1483,
      "step": 139672
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.0240899324417114,
      "learning_rate": 0.00020174502944004787,
      "loss": 2.0818,
      "step": 139673
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.0492371320724487,
      "learning_rate": 0.00020174116450018427,
      "loss": 2.1258,
      "step": 139674
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.220496416091919,
      "learning_rate": 0.0002017372995785888,
      "loss": 2.4807,
      "step": 139675
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.2344040870666504,
      "learning_rate": 0.00020173343467526226,
      "loss": 1.9252,
      "step": 139676
    },
    {
      "epoch": 1.82,
      "grad_norm": 0.9631964564323425,
      "learning_rate": 0.00020172956979020543,
      "loss": 2.0336,
      "step": 139677
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.0525143146514893,
      "learning_rate": 0.00020172570492341883,
      "loss": 2.0864,
      "step": 139678
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.018717646598816,
      "learning_rate": 0.00020172184007490325,
      "loss": 2.1216,
      "step": 139679
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.1382039785385132,
      "learning_rate": 0.00020171797524465944,
      "loss": 2.2666,
      "step": 139680
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.1665772199630737,
      "learning_rate": 0.00020171411043268811,
      "loss": 2.0357,
      "step": 139681
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.0348821878433228,
      "learning_rate": 0.00020171024563899,
      "loss": 2.0386,
      "step": 139682
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.072473168373108,
      "learning_rate": 0.00020170638086356593,
      "loss": 2.09,
      "step": 139683
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.07274329662323,
      "learning_rate": 0.00020170251610641637,
      "loss": 2.2227,
      "step": 139684
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.0720371007919312,
      "learning_rate": 0.00020169865136754217,
      "loss": 2.2662,
      "step": 139685
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.0387563705444336,
      "learning_rate": 0.00020169478664694402,
      "loss": 1.8897,
      "step": 139686
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.1152446269989014,
      "learning_rate": 0.0002016909219446227,
      "loss": 2.2601,
      "step": 139687
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.077616572380066,
      "learning_rate": 0.00020168705726057888,
      "loss": 2.1555,
      "step": 139688
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.0718899965286255,
      "learning_rate": 0.00020168319259481342,
      "loss": 2.2533,
      "step": 139689
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.0819339752197266,
      "learning_rate": 0.00020167932794732678,
      "loss": 1.9358,
      "step": 139690
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.0885857343673706,
      "learning_rate": 0.0002016754633181198,
      "loss": 2.2611,
      "step": 139691
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.041207194328308,
      "learning_rate": 0.00020167159870719325,
      "loss": 2.0922,
      "step": 139692
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.0827629566192627,
      "learning_rate": 0.00020166773411454776,
      "loss": 2.2466,
      "step": 139693
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.3474538326263428,
      "learning_rate": 0.0002016638695401841,
      "loss": 2.0609,
      "step": 139694
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.1327855587005615,
      "learning_rate": 0.00020166000498410313,
      "loss": 2.4531,
      "step": 139695
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.034993052482605,
      "learning_rate": 0.00020165614044630528,
      "loss": 2.0804,
      "step": 139696
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.2106142044067383,
      "learning_rate": 0.00020165227592679143,
      "loss": 2.3897,
      "step": 139697
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.2608226537704468,
      "learning_rate": 0.00020164841142556223,
      "loss": 2.1664,
      "step": 139698
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.036574363708496,
      "learning_rate": 0.0002016445469426185,
      "loss": 2.0665,
      "step": 139699
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.1401159763336182,
      "learning_rate": 0.00020164068247796086,
      "loss": 2.2712,
      "step": 139700
    },
    {
      "epoch": 1.82,
      "grad_norm": 0.9970757365226746,
      "learning_rate": 0.00020163681803159027,
      "loss": 2.007,
      "step": 139701
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.0486680269241333,
      "learning_rate": 0.00020163295360350707,
      "loss": 2.2147,
      "step": 139702
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.1321370601654053,
      "learning_rate": 0.00020162908919371217,
      "loss": 2.1689,
      "step": 139703
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.1127872467041016,
      "learning_rate": 0.00020162522480220624,
      "loss": 2.1855,
      "step": 139704
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.0854623317718506,
      "learning_rate": 0.0002016213604289901,
      "loss": 2.2706,
      "step": 139705
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.184607744216919,
      "learning_rate": 0.00020161749607406436,
      "loss": 2.0703,
      "step": 139706
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.0719584226608276,
      "learning_rate": 0.00020161363173742996,
      "loss": 1.952,
      "step": 139707
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.1004172563552856,
      "learning_rate": 0.0002016097674190873,
      "loss": 2.1323,
      "step": 139708
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.011383295059204,
      "learning_rate": 0.00020160590311903723,
      "loss": 1.9934,
      "step": 139709
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.0650054216384888,
      "learning_rate": 0.0002016020388372805,
      "loss": 1.8496,
      "step": 139710
    },
    {
      "epoch": 1.82,
      "grad_norm": 0.9743003845214844,
      "learning_rate": 0.00020159817457381774,
      "loss": 2.1418,
      "step": 139711
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.0556585788726807,
      "learning_rate": 0.00020159431032864983,
      "loss": 1.8759,
      "step": 139712
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.0413269996643066,
      "learning_rate": 0.00020159044610177748,
      "loss": 2.2118,
      "step": 139713
    },
    {
      "epoch": 1.82,
      "grad_norm": 0.9564794301986694,
      "learning_rate": 0.0002015865818932012,
      "loss": 2.1836,
      "step": 139714
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.0665327310562134,
      "learning_rate": 0.00020158271770292181,
      "loss": 2.1518,
      "step": 139715
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.0997867584228516,
      "learning_rate": 0.00020157885353094007,
      "loss": 2.1141,
      "step": 139716
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.019893765449524,
      "learning_rate": 0.00020157498937725668,
      "loss": 1.9899,
      "step": 139717
    },
    {
      "epoch": 1.82,
      "grad_norm": 0.996406614780426,
      "learning_rate": 0.00020157112524187237,
      "loss": 2.3108,
      "step": 139718
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.0945804119110107,
      "learning_rate": 0.00020156726112478793,
      "loss": 2.0495,
      "step": 139719
    },
    {
      "epoch": 1.82,
      "grad_norm": 0.9889384508132935,
      "learning_rate": 0.00020156339702600393,
      "loss": 2.0854,
      "step": 139720
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.1125288009643555,
      "learning_rate": 0.0002015595329455211,
      "loss": 1.9577,
      "step": 139721
    },
    {
      "epoch": 1.82,
      "grad_norm": 0.9747236371040344,
      "learning_rate": 0.00020155566888334025,
      "loss": 2.0239,
      "step": 139722
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.006731390953064,
      "learning_rate": 0.00020155180483946204,
      "loss": 2.12,
      "step": 139723
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.1739909648895264,
      "learning_rate": 0.00020154794081388722,
      "loss": 1.9803,
      "step": 139724
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.0400432348251343,
      "learning_rate": 0.00020154407680661647,
      "loss": 2.0008,
      "step": 139725
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.0618343353271484,
      "learning_rate": 0.0002015402128176506,
      "loss": 1.9873,
      "step": 139726
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.0609641075134277,
      "learning_rate": 0.00020153634884699027,
      "loss": 1.9166,
      "step": 139727
    },
    {
      "epoch": 1.82,
      "grad_norm": 0.9704602360725403,
      "learning_rate": 0.00020153248489463617,
      "loss": 2.0577,
      "step": 139728
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.4693853855133057,
      "learning_rate": 0.00020152862096058896,
      "loss": 1.9623,
      "step": 139729
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.1616424322128296,
      "learning_rate": 0.00020152475704484946,
      "loss": 2.1702,
      "step": 139730
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.0576790571212769,
      "learning_rate": 0.0002015208931474184,
      "loss": 2.2545,
      "step": 139731
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.2697848081588745,
      "learning_rate": 0.00020151702926829654,
      "loss": 2.0717,
      "step": 139732
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.744189977645874,
      "learning_rate": 0.00020151316540748443,
      "loss": 2.2329,
      "step": 139733
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.2633532285690308,
      "learning_rate": 0.00020150930156498294,
      "loss": 2.0004,
      "step": 139734
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.0586796998977661,
      "learning_rate": 0.00020150543774079269,
      "loss": 2.1019,
      "step": 139735
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.1086909770965576,
      "learning_rate": 0.00020150157393491444,
      "loss": 2.0957,
      "step": 139736
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.196402907371521,
      "learning_rate": 0.00020149771014734888,
      "loss": 2.0837,
      "step": 139737
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.00935959815979,
      "learning_rate": 0.0002014938463780968,
      "loss": 2.0098,
      "step": 139738
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.0295506715774536,
      "learning_rate": 0.00020148998262715888,
      "loss": 2.2115,
      "step": 139739
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.1197395324707031,
      "learning_rate": 0.00020148611889453575,
      "loss": 1.8792,
      "step": 139740
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.3622583150863647,
      "learning_rate": 0.00020148225518022833,
      "loss": 2.117,
      "step": 139741
    },
    {
      "epoch": 1.82,
      "grad_norm": 0.996941328048706,
      "learning_rate": 0.00020147839148423714,
      "loss": 2.0832,
      "step": 139742
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.067429542541504,
      "learning_rate": 0.00020147452780656298,
      "loss": 2.2447,
      "step": 139743
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.645553708076477,
      "learning_rate": 0.00020147066414720664,
      "loss": 2.2413,
      "step": 139744
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.1839812994003296,
      "learning_rate": 0.0002014668005061687,
      "loss": 1.9725,
      "step": 139745
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.0183742046356201,
      "learning_rate": 0.0002014629368834499,
      "loss": 2.1226,
      "step": 139746
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.1745578050613403,
      "learning_rate": 0.00020145907327905104,
      "loss": 2.1574,
      "step": 139747
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.0998704433441162,
      "learning_rate": 0.00020145520969297285,
      "loss": 1.9794,
      "step": 139748
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.0903863906860352,
      "learning_rate": 0.00020145134612521595,
      "loss": 2.1311,
      "step": 139749
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.1057031154632568,
      "learning_rate": 0.00020144748257578112,
      "loss": 2.1186,
      "step": 139750
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.1505398750305176,
      "learning_rate": 0.00020144361904466906,
      "loss": 2.1278,
      "step": 139751
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.1167889833450317,
      "learning_rate": 0.00020143975553188042,
      "loss": 1.9761,
      "step": 139752
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.2469128370285034,
      "learning_rate": 0.00020143589203741607,
      "loss": 2.0191,
      "step": 139753
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.0893306732177734,
      "learning_rate": 0.0002014320285612766,
      "loss": 2.2724,
      "step": 139754
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.1452417373657227,
      "learning_rate": 0.0002014281651034628,
      "loss": 2.0858,
      "step": 139755
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.0850846767425537,
      "learning_rate": 0.00020142430166397547,
      "loss": 2.1267,
      "step": 139756
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.0674803256988525,
      "learning_rate": 0.0002014204382428151,
      "loss": 1.9269,
      "step": 139757
    },
    {
      "epoch": 1.82,
      "grad_norm": 0.9767031669616699,
      "learning_rate": 0.00020141657483998252,
      "loss": 2.0042,
      "step": 139758
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.08773672580719,
      "learning_rate": 0.0002014127114554785,
      "loss": 1.9658,
      "step": 139759
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.0837106704711914,
      "learning_rate": 0.00020140884808930368,
      "loss": 2.1729,
      "step": 139760
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.079254150390625,
      "learning_rate": 0.00020140498474145887,
      "loss": 2.12,
      "step": 139761
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.0786634683609009,
      "learning_rate": 0.0002014011214119448,
      "loss": 2.3614,
      "step": 139762
    },
    {
      "epoch": 1.82,
      "grad_norm": 0.9855511784553528,
      "learning_rate": 0.000201397258100762,
      "loss": 2.2415,
      "step": 139763
    },
    {
      "epoch": 1.82,
      "grad_norm": 0.9704018831253052,
      "learning_rate": 0.00020139339480791134,
      "loss": 1.7623,
      "step": 139764
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.2482874393463135,
      "learning_rate": 0.00020138953153339348,
      "loss": 1.9419,
      "step": 139765
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.3173987865447998,
      "learning_rate": 0.0002013856682772092,
      "loss": 1.9528,
      "step": 139766
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.0905396938323975,
      "learning_rate": 0.00020138180503935917,
      "loss": 2.0086,
      "step": 139767
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.2424646615982056,
      "learning_rate": 0.00020137794181984423,
      "loss": 2.2329,
      "step": 139768
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.088118076324463,
      "learning_rate": 0.00020137407861866492,
      "loss": 2.1033,
      "step": 139769
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.0966633558273315,
      "learning_rate": 0.00020137021543582196,
      "loss": 2.0203,
      "step": 139770
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.0152431726455688,
      "learning_rate": 0.0002013663522713162,
      "loss": 1.9982,
      "step": 139771
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.0642777681350708,
      "learning_rate": 0.00020136248912514828,
      "loss": 1.9596,
      "step": 139772
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.0843596458435059,
      "learning_rate": 0.00020135862599731893,
      "loss": 1.9293,
      "step": 139773
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.1247891187667847,
      "learning_rate": 0.000201354762887829,
      "loss": 2.1527,
      "step": 139774
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.111617088317871,
      "learning_rate": 0.00020135089979667897,
      "loss": 2.0105,
      "step": 139775
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.115357518196106,
      "learning_rate": 0.00020134703672386965,
      "loss": 2.1154,
      "step": 139776
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.0679630041122437,
      "learning_rate": 0.00020134317366940176,
      "loss": 2.1521,
      "step": 139777
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.0674717426300049,
      "learning_rate": 0.00020133931063327608,
      "loss": 2.2013,
      "step": 139778
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.027480959892273,
      "learning_rate": 0.00020133544761549322,
      "loss": 2.309,
      "step": 139779
    },
    {
      "epoch": 1.82,
      "grad_norm": 0.9937663674354553,
      "learning_rate": 0.00020133158461605418,
      "loss": 2.1079,
      "step": 139780
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.0034244060516357,
      "learning_rate": 0.00020132772163495926,
      "loss": 2.057,
      "step": 139781
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.0721012353897095,
      "learning_rate": 0.00020132385867220942,
      "loss": 2.1739,
      "step": 139782
    },
    {
      "epoch": 1.82,
      "grad_norm": 0.9669221043586731,
      "learning_rate": 0.00020131999572780532,
      "loss": 2.0152,
      "step": 139783
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.156991958618164,
      "learning_rate": 0.0002013161328017477,
      "loss": 1.9536,
      "step": 139784
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.0244078636169434,
      "learning_rate": 0.00020131226989403725,
      "loss": 2.1502,
      "step": 139785
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.164722204208374,
      "learning_rate": 0.00020130840700467487,
      "loss": 2.2807,
      "step": 139786
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.0787616968154907,
      "learning_rate": 0.00020130454413366098,
      "loss": 2.0775,
      "step": 139787
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.0164058208465576,
      "learning_rate": 0.00020130068128099645,
      "loss": 2.0299,
      "step": 139788
    },
    {
      "epoch": 1.82,
      "grad_norm": 0.9296966195106506,
      "learning_rate": 0.000201296818446682,
      "loss": 1.9162,
      "step": 139789
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.1290626525878906,
      "learning_rate": 0.0002012929556307183,
      "loss": 1.9208,
      "step": 139790
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.1288297176361084,
      "learning_rate": 0.00020128909283310611,
      "loss": 2.1664,
      "step": 139791
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.063984751701355,
      "learning_rate": 0.00020128523005384614,
      "loss": 2.1683,
      "step": 139792
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.0521727800369263,
      "learning_rate": 0.00020128136729293927,
      "loss": 2.1727,
      "step": 139793
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.2179137468338013,
      "learning_rate": 0.00020127750455038588,
      "loss": 1.9464,
      "step": 139794
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.2155284881591797,
      "learning_rate": 0.0002012736418261869,
      "loss": 2.034,
      "step": 139795
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.0337839126586914,
      "learning_rate": 0.000201269779120343,
      "loss": 2.3016,
      "step": 139796
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.1738132238388062,
      "learning_rate": 0.0002012659164328549,
      "loss": 1.9696,
      "step": 139797
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.127657175064087,
      "learning_rate": 0.00020126205376372336,
      "loss": 2.0937,
      "step": 139798
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.0891549587249756,
      "learning_rate": 0.00020125819111294917,
      "loss": 2.3431,
      "step": 139799
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.0254933834075928,
      "learning_rate": 0.00020125432848053283,
      "loss": 2.2196,
      "step": 139800
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.1668879985809326,
      "learning_rate": 0.00020125046586647516,
      "loss": 2.1673,
      "step": 139801
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.138660192489624,
      "learning_rate": 0.0002012466032707769,
      "loss": 2.1946,
      "step": 139802
    },
    {
      "epoch": 1.82,
      "grad_norm": 0.9799330830574036,
      "learning_rate": 0.0002012427406934388,
      "loss": 2.3357,
      "step": 139803
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.0324541330337524,
      "learning_rate": 0.00020123887813446145,
      "loss": 2.0403,
      "step": 139804
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.0057893991470337,
      "learning_rate": 0.00020123501559384584,
      "loss": 2.0922,
      "step": 139805
    },
    {
      "epoch": 1.82,
      "grad_norm": 0.9480282068252563,
      "learning_rate": 0.00020123115307159241,
      "loss": 2.054,
      "step": 139806
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.0857484340667725,
      "learning_rate": 0.00020122729056770196,
      "loss": 1.8942,
      "step": 139807
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.0173592567443848,
      "learning_rate": 0.0002012234280821752,
      "loss": 2.174,
      "step": 139808
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.098618507385254,
      "learning_rate": 0.00020121956561501282,
      "loss": 2.1931,
      "step": 139809
    },
    {
      "epoch": 1.82,
      "grad_norm": 0.9869318604469299,
      "learning_rate": 0.00020121570316621566,
      "loss": 2.1473,
      "step": 139810
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.1795899868011475,
      "learning_rate": 0.00020121184073578448,
      "loss": 2.3199,
      "step": 139811
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.1305421590805054,
      "learning_rate": 0.00020120797832371977,
      "loss": 2.0863,
      "step": 139812
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.0162582397460938,
      "learning_rate": 0.0002012041159300223,
      "loss": 1.9858,
      "step": 139813
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.0841864347457886,
      "learning_rate": 0.0002012002535546929,
      "loss": 2.141,
      "step": 139814
    },
    {
      "epoch": 1.82,
      "grad_norm": 0.965990424156189,
      "learning_rate": 0.00020119639119773222,
      "loss": 1.8986,
      "step": 139815
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.063152551651001,
      "learning_rate": 0.000201192528859141,
      "loss": 2.1031,
      "step": 139816
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.0382111072540283,
      "learning_rate": 0.00020118866653892002,
      "loss": 2.1466,
      "step": 139817
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.0005676746368408,
      "learning_rate": 0.0002011848042370699,
      "loss": 2.1325,
      "step": 139818
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.0984710454940796,
      "learning_rate": 0.0002011809419535914,
      "loss": 2.0154,
      "step": 139819
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.0831634998321533,
      "learning_rate": 0.0002011770796884852,
      "loss": 2.0038,
      "step": 139820
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.0512396097183228,
      "learning_rate": 0.000201173217441752,
      "loss": 2.0647,
      "step": 139821
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.1063421964645386,
      "learning_rate": 0.0002011693552133926,
      "loss": 2.0841,
      "step": 139822
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.1208473443984985,
      "learning_rate": 0.00020116549300340774,
      "loss": 2.0992,
      "step": 139823
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.190616488456726,
      "learning_rate": 0.000201161630811798,
      "loss": 1.9782,
      "step": 139824
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.0985630750656128,
      "learning_rate": 0.0002011577686385642,
      "loss": 2.3545,
      "step": 139825
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.06973397731781,
      "learning_rate": 0.00020115390648370714,
      "loss": 2.2029,
      "step": 139826
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.1126325130462646,
      "learning_rate": 0.0002011500443472273,
      "loss": 2.0181,
      "step": 139827
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.2825204133987427,
      "learning_rate": 0.00020114618222912556,
      "loss": 2.0847,
      "step": 139828
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.1675219535827637,
      "learning_rate": 0.0002011423201294027,
      "loss": 2.1176,
      "step": 139829
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.0717055797576904,
      "learning_rate": 0.00020113845804805924,
      "loss": 2.1073,
      "step": 139830
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.084499716758728,
      "learning_rate": 0.00020113459598509605,
      "loss": 1.9626,
      "step": 139831
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.0443029403686523,
      "learning_rate": 0.00020113073394051377,
      "loss": 1.7608,
      "step": 139832
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.5305355787277222,
      "learning_rate": 0.0002011268719143132,
      "loss": 2.212,
      "step": 139833
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.0981699228286743,
      "learning_rate": 0.000201123009906495,
      "loss": 1.9635,
      "step": 139834
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.0721927881240845,
      "learning_rate": 0.00020111914791705993,
      "loss": 1.9224,
      "step": 139835
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.1164721250534058,
      "learning_rate": 0.00020111528594600865,
      "loss": 1.9574,
      "step": 139836
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.3275763988494873,
      "learning_rate": 0.00020111142399334187,
      "loss": 2.0654,
      "step": 139837
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.0791478157043457,
      "learning_rate": 0.00020110756205906036,
      "loss": 2.2862,
      "step": 139838
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.0902818441390991,
      "learning_rate": 0.0002011037001431648,
      "loss": 2.1913,
      "step": 139839
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.1143115758895874,
      "learning_rate": 0.00020109983824565599,
      "loss": 2.1038,
      "step": 139840
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.154490351676941,
      "learning_rate": 0.00020109597636653462,
      "loss": 2.0269,
      "step": 139841
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.1839513778686523,
      "learning_rate": 0.00020109211450580126,
      "loss": 2.3598,
      "step": 139842
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.139784574508667,
      "learning_rate": 0.0002010882526634568,
      "loss": 2.2497,
      "step": 139843
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.023025631904602,
      "learning_rate": 0.00020108439083950188,
      "loss": 1.8569,
      "step": 139844
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.0206516981124878,
      "learning_rate": 0.0002010805290339372,
      "loss": 2.043,
      "step": 139845
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.121334195137024,
      "learning_rate": 0.00020107666724676358,
      "loss": 2.1576,
      "step": 139846
    },
    {
      "epoch": 1.82,
      "grad_norm": 0.9399845600128174,
      "learning_rate": 0.0002010728054779818,
      "loss": 2.0253,
      "step": 139847
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.0600638389587402,
      "learning_rate": 0.00020106894372759227,
      "loss": 2.0845,
      "step": 139848
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.2793028354644775,
      "learning_rate": 0.0002010650819955959,
      "loss": 2.2961,
      "step": 139849
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.0456782579421997,
      "learning_rate": 0.00020106122028199343,
      "loss": 2.057,
      "step": 139850
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.0387557744979858,
      "learning_rate": 0.00020105735858678555,
      "loss": 2.1767,
      "step": 139851
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.00527822971344,
      "learning_rate": 0.00020105349690997294,
      "loss": 2.0427,
      "step": 139852
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.2381305694580078,
      "learning_rate": 0.0002010496352515565,
      "loss": 2.1576,
      "step": 139853
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.39241623878479,
      "learning_rate": 0.00020104577361153668,
      "loss": 2.2931,
      "step": 139854
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.077349066734314,
      "learning_rate": 0.00020104191198991436,
      "loss": 1.9327,
      "step": 139855
    },
    {
      "epoch": 1.82,
      "grad_norm": 0.988175630569458,
      "learning_rate": 0.00020103805038669014,
      "loss": 2.0634,
      "step": 139856
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.4062371253967285,
      "learning_rate": 0.00020103418880186485,
      "loss": 2.2231,
      "step": 139857
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.2325645685195923,
      "learning_rate": 0.00020103032723543918,
      "loss": 2.1256,
      "step": 139858
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.237731695175171,
      "learning_rate": 0.00020102646568741382,
      "loss": 2.0682,
      "step": 139859
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.051446795463562,
      "learning_rate": 0.00020102260415778963,
      "loss": 2.1347,
      "step": 139860
    },
    {
      "epoch": 1.82,
      "grad_norm": 0.9998528957366943,
      "learning_rate": 0.0002010187426465671,
      "loss": 2.1325,
      "step": 139861
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.1466755867004395,
      "learning_rate": 0.000201014881153747,
      "loss": 2.1184,
      "step": 139862
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.122180461883545,
      "learning_rate": 0.0002010110196793302,
      "loss": 1.9768,
      "step": 139863
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.1781349182128906,
      "learning_rate": 0.00020100715822331724,
      "loss": 2.1427,
      "step": 139864
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.0763251781463623,
      "learning_rate": 0.00020100329678570896,
      "loss": 1.9201,
      "step": 139865
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.0825583934783936,
      "learning_rate": 0.00020099943536650613,
      "loss": 1.7435,
      "step": 139866
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.041875958442688,
      "learning_rate": 0.00020099557396570928,
      "loss": 2.3353,
      "step": 139867
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.1030170917510986,
      "learning_rate": 0.00020099171258331917,
      "loss": 2.0393,
      "step": 139868
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.041271448135376,
      "learning_rate": 0.00020098785121933663,
      "loss": 2.2272,
      "step": 139869
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.136850118637085,
      "learning_rate": 0.00020098398987376231,
      "loss": 2.5127,
      "step": 139870
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.0486146211624146,
      "learning_rate": 0.00020098012854659694,
      "loss": 2.0374,
      "step": 139871
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.0391067266464233,
      "learning_rate": 0.00020097626723784132,
      "loss": 1.9895,
      "step": 139872
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.1772270202636719,
      "learning_rate": 0.00020097240594749597,
      "loss": 2.1647,
      "step": 139873
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.5223056077957153,
      "learning_rate": 0.00020096854467556175,
      "loss": 2.2337,
      "step": 139874
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.1089266538619995,
      "learning_rate": 0.0002009646834220393,
      "loss": 1.8303,
      "step": 139875
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.1497877836227417,
      "learning_rate": 0.00020096082218692938,
      "loss": 2.2853,
      "step": 139876
    },
    {
      "epoch": 1.82,
      "grad_norm": 0.9608956575393677,
      "learning_rate": 0.00020095696097023271,
      "loss": 2.3487,
      "step": 139877
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.2007852792739868,
      "learning_rate": 0.00020095309977195022,
      "loss": 2.0008,
      "step": 139878
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.0570085048675537,
      "learning_rate": 0.00020094923859208222,
      "loss": 2.0419,
      "step": 139879
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.0236363410949707,
      "learning_rate": 0.00020094537743062962,
      "loss": 1.944,
      "step": 139880
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.0906665325164795,
      "learning_rate": 0.0002009415162875932,
      "loss": 2.0739,
      "step": 139881
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.020787000656128,
      "learning_rate": 0.00020093765516297356,
      "loss": 2.1754,
      "step": 139882
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.0289961099624634,
      "learning_rate": 0.00020093379405677152,
      "loss": 1.929,
      "step": 139883
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.0758455991744995,
      "learning_rate": 0.00020092993296898787,
      "loss": 1.9188,
      "step": 139884
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.0392824411392212,
      "learning_rate": 0.0002009260718996231,
      "loss": 2.2672,
      "step": 139885
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.0282071828842163,
      "learning_rate": 0.00020092221084867803,
      "loss": 1.903,
      "step": 139886
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.0407212972640991,
      "learning_rate": 0.0002009183498161534,
      "loss": 2.1136,
      "step": 139887
    },
    {
      "epoch": 1.82,
      "grad_norm": 0.9404860734939575,
      "learning_rate": 0.0002009144888020499,
      "loss": 2.2471,
      "step": 139888
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.1134012937545776,
      "learning_rate": 0.00020091062780636825,
      "loss": 2.0935,
      "step": 139889
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.104894757270813,
      "learning_rate": 0.00020090676682910934,
      "loss": 2.3179,
      "step": 139890
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.022831678390503,
      "learning_rate": 0.00020090290587027363,
      "loss": 2.1579,
      "step": 139891
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.1009321212768555,
      "learning_rate": 0.00020089904492986193,
      "loss": 1.8763,
      "step": 139892
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.0355241298675537,
      "learning_rate": 0.00020089518400787494,
      "loss": 2.094,
      "step": 139893
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.143141269683838,
      "learning_rate": 0.00020089132310431346,
      "loss": 2.211,
      "step": 139894
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.1405978202819824,
      "learning_rate": 0.0002008874622191781,
      "loss": 2.1121,
      "step": 139895
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.097247838973999,
      "learning_rate": 0.00020088360135246978,
      "loss": 2.1126,
      "step": 139896
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.0279723405838013,
      "learning_rate": 0.00020087974050418895,
      "loss": 2.1577,
      "step": 139897
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.1155099868774414,
      "learning_rate": 0.00020087587967433646,
      "loss": 1.898,
      "step": 139898
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.079725742340088,
      "learning_rate": 0.00020087201886291296,
      "loss": 1.9511,
      "step": 139899
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.1068050861358643,
      "learning_rate": 0.00020086815806991926,
      "loss": 2.0159,
      "step": 139900
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.1132433414459229,
      "learning_rate": 0.00020086429729535604,
      "loss": 2.1896,
      "step": 139901
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.291902780532837,
      "learning_rate": 0.0002008604365392241,
      "loss": 2.1445,
      "step": 139902
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.1115230321884155,
      "learning_rate": 0.00020085657580152402,
      "loss": 2.169,
      "step": 139903
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.041886806488037,
      "learning_rate": 0.00020085271508225655,
      "loss": 1.8522,
      "step": 139904
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.1155004501342773,
      "learning_rate": 0.0002008488543814224,
      "loss": 2.3128,
      "step": 139905
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.0255662202835083,
      "learning_rate": 0.00020084499369902234,
      "loss": 1.9683,
      "step": 139906
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.152221441268921,
      "learning_rate": 0.0002008411330350571,
      "loss": 2.0962,
      "step": 139907
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.0485484600067139,
      "learning_rate": 0.0002008372723895274,
      "loss": 2.0117,
      "step": 139908
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.1095802783966064,
      "learning_rate": 0.00020083341176243386,
      "loss": 2.0835,
      "step": 139909
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.1074868440628052,
      "learning_rate": 0.00020082955115377732,
      "loss": 2.2305,
      "step": 139910
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.3437927961349487,
      "learning_rate": 0.0002008256905635584,
      "loss": 1.9733,
      "step": 139911
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.0638929605484009,
      "learning_rate": 0.00020082182999177781,
      "loss": 2.1776,
      "step": 139912
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.1030986309051514,
      "learning_rate": 0.00020081796943843634,
      "loss": 2.1683,
      "step": 139913
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.149032711982727,
      "learning_rate": 0.00020081410890353474,
      "loss": 2.1632,
      "step": 139914
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.090753197669983,
      "learning_rate": 0.00020081024838707363,
      "loss": 2.0975,
      "step": 139915
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.330937385559082,
      "learning_rate": 0.00020080638788905374,
      "loss": 2.0198,
      "step": 139916
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.109352707862854,
      "learning_rate": 0.00020080252740947583,
      "loss": 2.0464,
      "step": 139917
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.0183686017990112,
      "learning_rate": 0.00020079866694834066,
      "loss": 2.1296,
      "step": 139918
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.095740556716919,
      "learning_rate": 0.00020079480650564882,
      "loss": 2.0401,
      "step": 139919
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.0266276597976685,
      "learning_rate": 0.00020079094608140113,
      "loss": 1.9324,
      "step": 139920
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.0701720714569092,
      "learning_rate": 0.00020078708567559828,
      "loss": 1.9874,
      "step": 139921
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.1996787786483765,
      "learning_rate": 0.00020078322528824092,
      "loss": 2.0664,
      "step": 139922
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.207475185394287,
      "learning_rate": 0.0002007793649193299,
      "loss": 2.1787,
      "step": 139923
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.1106717586517334,
      "learning_rate": 0.00020077550456886584,
      "loss": 2.0942,
      "step": 139924
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.1993865966796875,
      "learning_rate": 0.00020077164423684953,
      "loss": 2.1599,
      "step": 139925
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.1091140508651733,
      "learning_rate": 0.0002007677839232816,
      "loss": 2.1264,
      "step": 139926
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.079318881034851,
      "learning_rate": 0.00020076392362816285,
      "loss": 2.1442,
      "step": 139927
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.0632774829864502,
      "learning_rate": 0.00020076006335149394,
      "loss": 2.2134,
      "step": 139928
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.096746802330017,
      "learning_rate": 0.00020075620309327556,
      "loss": 2.0645,
      "step": 139929
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.1648435592651367,
      "learning_rate": 0.00020075234285350853,
      "loss": 2.1824,
      "step": 139930
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.0250118970870972,
      "learning_rate": 0.00020074848263219346,
      "loss": 2.2891,
      "step": 139931
    },
    {
      "epoch": 1.82,
      "grad_norm": 0.9922440648078918,
      "learning_rate": 0.00020074462242933124,
      "loss": 2.0895,
      "step": 139932
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.0490820407867432,
      "learning_rate": 0.0002007407622449224,
      "loss": 2.0473,
      "step": 139933
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.0678791999816895,
      "learning_rate": 0.0002007369020789677,
      "loss": 2.0391,
      "step": 139934
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.0604523420333862,
      "learning_rate": 0.0002007330419314679,
      "loss": 2.1519,
      "step": 139935
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.0500932931900024,
      "learning_rate": 0.00020072918180242367,
      "loss": 1.9308,
      "step": 139936
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.0606404542922974,
      "learning_rate": 0.00020072532169183578,
      "loss": 2.0604,
      "step": 139937
    },
    {
      "epoch": 1.82,
      "grad_norm": 0.9787089228630066,
      "learning_rate": 0.00020072146159970492,
      "loss": 2.1188,
      "step": 139938
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.0352685451507568,
      "learning_rate": 0.00020071760152603192,
      "loss": 2.2253,
      "step": 139939
    },
    {
      "epoch": 1.82,
      "grad_norm": 0.9558620452880859,
      "learning_rate": 0.0002007137414708173,
      "loss": 2.1561,
      "step": 139940
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.082089900970459,
      "learning_rate": 0.00020070988143406187,
      "loss": 2.2847,
      "step": 139941
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.0048731565475464,
      "learning_rate": 0.00020070602141576634,
      "loss": 1.8611,
      "step": 139942
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.171090841293335,
      "learning_rate": 0.0002007021614159314,
      "loss": 2.0959,
      "step": 139943
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.0605335235595703,
      "learning_rate": 0.00020069830143455784,
      "loss": 2.3046,
      "step": 139944
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.06071937084198,
      "learning_rate": 0.00020069444147164642,
      "loss": 1.8986,
      "step": 139945
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.5597007274627686,
      "learning_rate": 0.00020069058152719768,
      "loss": 2.2354,
      "step": 139946
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.5867222547531128,
      "learning_rate": 0.00020068672160121248,
      "loss": 2.0418,
      "step": 139947
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.0965275764465332,
      "learning_rate": 0.00020068286169369143,
      "loss": 1.9758,
      "step": 139948
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.0839070081710815,
      "learning_rate": 0.00020067900180463537,
      "loss": 2.0032,
      "step": 139949
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.4318221807479858,
      "learning_rate": 0.00020067514193404488,
      "loss": 2.1791,
      "step": 139950
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.2569047212600708,
      "learning_rate": 0.00020067128208192095,
      "loss": 2.0032,
      "step": 139951
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.2755934000015259,
      "learning_rate": 0.00020066742224826394,
      "loss": 2.0525,
      "step": 139952
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.0790914297103882,
      "learning_rate": 0.00020066356243307474,
      "loss": 2.1604,
      "step": 139953
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.1397403478622437,
      "learning_rate": 0.00020065970263635408,
      "loss": 2.3509,
      "step": 139954
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.1753551959991455,
      "learning_rate": 0.00020065584285810264,
      "loss": 2.0316,
      "step": 139955
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.0512418746948242,
      "learning_rate": 0.00020065198309832116,
      "loss": 2.0473,
      "step": 139956
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.0754215717315674,
      "learning_rate": 0.00020064812335701047,
      "loss": 2.0643,
      "step": 139957
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.1785738468170166,
      "learning_rate": 0.00020064426363417107,
      "loss": 2.1264,
      "step": 139958
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.0218784809112549,
      "learning_rate": 0.00020064040392980375,
      "loss": 2.1062,
      "step": 139959
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.042548418045044,
      "learning_rate": 0.00020063654424390924,
      "loss": 2.1701,
      "step": 139960
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.1649940013885498,
      "learning_rate": 0.00020063268457648828,
      "loss": 2.0419,
      "step": 139961
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.072095274925232,
      "learning_rate": 0.0002006288249275416,
      "loss": 2.0213,
      "step": 139962
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.0915658473968506,
      "learning_rate": 0.00020062496529707,
      "loss": 2.1662,
      "step": 139963
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.1904692649841309,
      "learning_rate": 0.00020062110568507398,
      "loss": 2.0816,
      "step": 139964
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.1963191032409668,
      "learning_rate": 0.00020061724609155438,
      "loss": 2.1881,
      "step": 139965
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.3490817546844482,
      "learning_rate": 0.0002006133865165119,
      "loss": 2.2924,
      "step": 139966
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.054884433746338,
      "learning_rate": 0.00020060952695994728,
      "loss": 2.0355,
      "step": 139967
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.0808924436569214,
      "learning_rate": 0.0002006056674218612,
      "loss": 1.9384,
      "step": 139968
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.072237253189087,
      "learning_rate": 0.00020060180790225455,
      "loss": 1.9908,
      "step": 139969
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.180417537689209,
      "learning_rate": 0.00020059794840112776,
      "loss": 2.1261,
      "step": 139970
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.0393211841583252,
      "learning_rate": 0.00020059408891848167,
      "loss": 1.9043,
      "step": 139971
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.0547950267791748,
      "learning_rate": 0.0002005902294543171,
      "loss": 2.1273,
      "step": 139972
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.1515547037124634,
      "learning_rate": 0.00020058637000863456,
      "loss": 2.0039,
      "step": 139973
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.091969609260559,
      "learning_rate": 0.00020058251058143497,
      "loss": 2.1218,
      "step": 139974
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.1301686763763428,
      "learning_rate": 0.0002005786511727191,
      "loss": 2.1979,
      "step": 139975
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.108498454093933,
      "learning_rate": 0.00020057479178248742,
      "loss": 2.1515,
      "step": 139976
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.058786153793335,
      "learning_rate": 0.0002005709324107407,
      "loss": 2.2494,
      "step": 139977
    },
    {
      "epoch": 1.82,
      "grad_norm": 0.9878861308097839,
      "learning_rate": 0.00020056707305747976,
      "loss": 2.279,
      "step": 139978
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.0924206972122192,
      "learning_rate": 0.00020056321372270526,
      "loss": 2.1694,
      "step": 139979
    },
    {
      "epoch": 1.82,
      "grad_norm": 0.9529073238372803,
      "learning_rate": 0.00020055935440641795,
      "loss": 1.9923,
      "step": 139980
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.0102663040161133,
      "learning_rate": 0.00020055549510861865,
      "loss": 2.0881,
      "step": 139981
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.1674798727035522,
      "learning_rate": 0.0002005516358293078,
      "loss": 1.9708,
      "step": 139982
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.0609639883041382,
      "learning_rate": 0.00020054777656848638,
      "loss": 2.3595,
      "step": 139983
    },
    {
      "epoch": 1.82,
      "grad_norm": 0.9655493497848511,
      "learning_rate": 0.00020054391732615494,
      "loss": 2.0775,
      "step": 139984
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.0263309478759766,
      "learning_rate": 0.00020054005810231425,
      "loss": 1.8032,
      "step": 139985
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.017744779586792,
      "learning_rate": 0.00020053619889696505,
      "loss": 2.3399,
      "step": 139986
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.0715798139572144,
      "learning_rate": 0.00020053233971010814,
      "loss": 2.0024,
      "step": 139987
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.1024442911148071,
      "learning_rate": 0.0002005284805417441,
      "loss": 1.977,
      "step": 139988
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.0232998132705688,
      "learning_rate": 0.00020052462139187366,
      "loss": 1.9605,
      "step": 139989
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.193809986114502,
      "learning_rate": 0.00020052076226049754,
      "loss": 2.2103,
      "step": 139990
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.0601587295532227,
      "learning_rate": 0.00020051690314761654,
      "loss": 2.0837,
      "step": 139991
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.146292805671692,
      "learning_rate": 0.0002005130440532313,
      "loss": 2.0882,
      "step": 139992
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.1870664358139038,
      "learning_rate": 0.00020050918497734255,
      "loss": 2.1744,
      "step": 139993
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.1600440740585327,
      "learning_rate": 0.00020050532591995107,
      "loss": 2.1612,
      "step": 139994
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.0161819458007812,
      "learning_rate": 0.00020050146688105756,
      "loss": 1.9674,
      "step": 139995
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.1677650213241577,
      "learning_rate": 0.00020049760786066266,
      "loss": 2.026,
      "step": 139996
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.090405821800232,
      "learning_rate": 0.0002004937488587671,
      "loss": 2.0775,
      "step": 139997
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.050284504890442,
      "learning_rate": 0.00020048988987537164,
      "loss": 2.0574,
      "step": 139998
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.0610321760177612,
      "learning_rate": 0.00020048603091047697,
      "loss": 2.1489,
      "step": 139999
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.3681939840316772,
      "learning_rate": 0.00020048217196408393,
      "loss": 2.1143,
      "step": 140000
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.1243972778320312,
      "learning_rate": 0.00020047831303619305,
      "loss": 2.2946,
      "step": 140001
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.1029433012008667,
      "learning_rate": 0.00020047445412680516,
      "loss": 2.2665,
      "step": 140002
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.1414613723754883,
      "learning_rate": 0.0002004705952359209,
      "loss": 2.032,
      "step": 140003
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.044894814491272,
      "learning_rate": 0.00020046673636354108,
      "loss": 2.0211,
      "step": 140004
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.099385380744934,
      "learning_rate": 0.00020046287750966632,
      "loss": 2.0773,
      "step": 140005
    },
    {
      "epoch": 1.82,
      "grad_norm": 0.9994716644287109,
      "learning_rate": 0.0002004590186742975,
      "loss": 2.1938,
      "step": 140006
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.0482345819473267,
      "learning_rate": 0.00020045515985743514,
      "loss": 1.9097,
      "step": 140007
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.001409888267517,
      "learning_rate": 0.00020045130105908002,
      "loss": 2.0048,
      "step": 140008
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.0323452949523926,
      "learning_rate": 0.00020044744227923299,
      "loss": 2.0246,
      "step": 140009
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.00058913230896,
      "learning_rate": 0.00020044358351789456,
      "loss": 2.2067,
      "step": 140010
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.0081430673599243,
      "learning_rate": 0.00020043972477506558,
      "loss": 2.0998,
      "step": 140011
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.163975715637207,
      "learning_rate": 0.00020043586605074677,
      "loss": 2.2306,
      "step": 140012
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.0819218158721924,
      "learning_rate": 0.00020043200734493877,
      "loss": 2.1938,
      "step": 140013
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.1775177717208862,
      "learning_rate": 0.0002004281486576423,
      "loss": 2.0631,
      "step": 140014
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.0390660762786865,
      "learning_rate": 0.00020042428998885814,
      "loss": 2.4508,
      "step": 140015
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.1848556995391846,
      "learning_rate": 0.000200420431338587,
      "loss": 2.2967,
      "step": 140016
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.1340972185134888,
      "learning_rate": 0.00020041657270682963,
      "loss": 2.084,
      "step": 140017
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.064271092414856,
      "learning_rate": 0.0002004127140935867,
      "loss": 2.1772,
      "step": 140018
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.0900789499282837,
      "learning_rate": 0.00020040885549885883,
      "loss": 2.0049,
      "step": 140019
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.238632082939148,
      "learning_rate": 0.0002004049969226469,
      "loss": 2.3181,
      "step": 140020
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.2391002178192139,
      "learning_rate": 0.00020040113836495153,
      "loss": 2.0631,
      "step": 140021
    },
    {
      "epoch": 1.82,
      "grad_norm": 0.9916965365409851,
      "learning_rate": 0.00020039727982577347,
      "loss": 2.0882,
      "step": 140022
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.2439661026000977,
      "learning_rate": 0.00020039342130511345,
      "loss": 2.175,
      "step": 140023
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.2456363439559937,
      "learning_rate": 0.00020038956280297226,
      "loss": 2.0066,
      "step": 140024
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.0644475221633911,
      "learning_rate": 0.00020038570431935043,
      "loss": 2.1827,
      "step": 140025
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.102650761604309,
      "learning_rate": 0.00020038184585424878,
      "loss": 2.1095,
      "step": 140026
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.0181633234024048,
      "learning_rate": 0.000200377987407668,
      "loss": 1.9707,
      "step": 140027
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.0097649097442627,
      "learning_rate": 0.00020037412897960886,
      "loss": 1.9516,
      "step": 140028
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.0408154726028442,
      "learning_rate": 0.00020037027057007205,
      "loss": 2.015,
      "step": 140029
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.0077590942382812,
      "learning_rate": 0.00020036641217905843,
      "loss": 2.1501,
      "step": 140030
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.035844087600708,
      "learning_rate": 0.00020036255380656838,
      "loss": 2.1921,
      "step": 140031
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.0812184810638428,
      "learning_rate": 0.00020035869545260285,
      "loss": 1.8973,
      "step": 140032
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.2449870109558105,
      "learning_rate": 0.00020035483711716254,
      "loss": 2.0914,
      "step": 140033
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.143836498260498,
      "learning_rate": 0.00020035097880024815,
      "loss": 2.2574,
      "step": 140034
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.0772565603256226,
      "learning_rate": 0.00020034712050186036,
      "loss": 1.9747,
      "step": 140035
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.0734126567840576,
      "learning_rate": 0.0002003432622220001,
      "loss": 1.9503,
      "step": 140036
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.1182984113693237,
      "learning_rate": 0.00020033940396066774,
      "loss": 2.2143,
      "step": 140037
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.1579840183258057,
      "learning_rate": 0.00020033554571786415,
      "loss": 2.0827,
      "step": 140038
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.0922801494598389,
      "learning_rate": 0.0002003316874935901,
      "loss": 1.9932,
      "step": 140039
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.0553914308547974,
      "learning_rate": 0.00020032782928784627,
      "loss": 2.2461,
      "step": 140040
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.1280807256698608,
      "learning_rate": 0.00020032397110063336,
      "loss": 2.2605,
      "step": 140041
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.1177016496658325,
      "learning_rate": 0.00020032011293195223,
      "loss": 2.0638,
      "step": 140042
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.0042762756347656,
      "learning_rate": 0.00020031625478180332,
      "loss": 2.0588,
      "step": 140043
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.073976993560791,
      "learning_rate": 0.00020031239665018754,
      "loss": 2.0255,
      "step": 140044
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.1392133235931396,
      "learning_rate": 0.0002003085385371056,
      "loss": 2.0763,
      "step": 140045
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.2315996885299683,
      "learning_rate": 0.00020030468044255812,
      "loss": 2.2342,
      "step": 140046
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.0708993673324585,
      "learning_rate": 0.0002003008223665459,
      "loss": 2.0642,
      "step": 140047
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.0620394945144653,
      "learning_rate": 0.0002002969643090698,
      "loss": 2.1273,
      "step": 140048
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.327805757522583,
      "learning_rate": 0.00020029310627013022,
      "loss": 2.0958,
      "step": 140049
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.20976722240448,
      "learning_rate": 0.00020028924824972802,
      "loss": 2.1912,
      "step": 140050
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.0772600173950195,
      "learning_rate": 0.00020028539024786397,
      "loss": 1.8417,
      "step": 140051
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.0972566604614258,
      "learning_rate": 0.0002002815322645387,
      "loss": 2.1534,
      "step": 140052
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.2500700950622559,
      "learning_rate": 0.00020027767429975298,
      "loss": 2.1137,
      "step": 140053
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.0749236345291138,
      "learning_rate": 0.00020027381635350771,
      "loss": 1.8829,
      "step": 140054
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.1300197839736938,
      "learning_rate": 0.00020026995842580323,
      "loss": 2.0528,
      "step": 140055
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.0594927072525024,
      "learning_rate": 0.0002002661005166405,
      "loss": 2.1982,
      "step": 140056
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.0675163269042969,
      "learning_rate": 0.00020026224262602012,
      "loss": 2.0575,
      "step": 140057
    },
    {
      "epoch": 1.82,
      "grad_norm": 0.9996054768562317,
      "learning_rate": 0.00020025838475394293,
      "loss": 2.2266,
      "step": 140058
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.1789368391036987,
      "learning_rate": 0.00020025452690040959,
      "loss": 2.1221,
      "step": 140059
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.205004334449768,
      "learning_rate": 0.00020025066906542078,
      "loss": 2.1898,
      "step": 140060
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.0343393087387085,
      "learning_rate": 0.00020024681124897737,
      "loss": 1.9752,
      "step": 140061
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.2757636308670044,
      "learning_rate": 0.00020024295345107989,
      "loss": 1.9376,
      "step": 140062
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.104117512702942,
      "learning_rate": 0.0002002390956717291,
      "loss": 2.2642,
      "step": 140063
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.1291215419769287,
      "learning_rate": 0.0002002352379109257,
      "loss": 2.0083,
      "step": 140064
    },
    {
      "epoch": 1.82,
      "grad_norm": 0.9832372665405273,
      "learning_rate": 0.00020023138016867048,
      "loss": 1.8388,
      "step": 140065
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.238178014755249,
      "learning_rate": 0.0002002275224449642,
      "loss": 1.8803,
      "step": 140066
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.0370287895202637,
      "learning_rate": 0.00020022366473980752,
      "loss": 2.0892,
      "step": 140067
    },
    {
      "epoch": 1.82,
      "grad_norm": 0.9301757216453552,
      "learning_rate": 0.0002002198070532011,
      "loss": 2.1517,
      "step": 140068
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.0401899814605713,
      "learning_rate": 0.00020021594938514565,
      "loss": 2.1149,
      "step": 140069
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.0382678508758545,
      "learning_rate": 0.00020021209173564196,
      "loss": 1.9161,
      "step": 140070
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.0751349925994873,
      "learning_rate": 0.00020020823410469074,
      "loss": 1.9953,
      "step": 140071
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.0768746137619019,
      "learning_rate": 0.0002002043764922927,
      "loss": 1.9051,
      "step": 140072
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.0465368032455444,
      "learning_rate": 0.00020020051889844866,
      "loss": 2.2105,
      "step": 140073
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.1085398197174072,
      "learning_rate": 0.00020019666132315908,
      "loss": 2.0448,
      "step": 140074
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.0785290002822876,
      "learning_rate": 0.00020019280376642486,
      "loss": 2.1323,
      "step": 140075
    },
    {
      "epoch": 1.82,
      "grad_norm": 0.9827190637588501,
      "learning_rate": 0.00020018894622824665,
      "loss": 1.9614,
      "step": 140076
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.0528066158294678,
      "learning_rate": 0.00020018508870862524,
      "loss": 1.9136,
      "step": 140077
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.0189931392669678,
      "learning_rate": 0.00020018123120756128,
      "loss": 2.1955,
      "step": 140078
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.047191858291626,
      "learning_rate": 0.00020017737372505557,
      "loss": 2.0819,
      "step": 140079
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.1395649909973145,
      "learning_rate": 0.00020017351626110876,
      "loss": 1.846,
      "step": 140080
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.3106844425201416,
      "learning_rate": 0.00020016965881572156,
      "loss": 2.0823,
      "step": 140081
    },
    {
      "epoch": 1.82,
      "grad_norm": 0.9881297945976257,
      "learning_rate": 0.00020016580138889468,
      "loss": 2.0628,
      "step": 140082
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.031733512878418,
      "learning_rate": 0.00020016194398062885,
      "loss": 2.1766,
      "step": 140083
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.0412640571594238,
      "learning_rate": 0.0002001580865909248,
      "loss": 2.0185,
      "step": 140084
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.14896821975708,
      "learning_rate": 0.0002001542292197833,
      "loss": 2.0938,
      "step": 140085
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.022681713104248,
      "learning_rate": 0.00020015037186720498,
      "loss": 2.1614,
      "step": 140086
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.0341684818267822,
      "learning_rate": 0.00020014651453319066,
      "loss": 2.2056,
      "step": 140087
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.177248239517212,
      "learning_rate": 0.0002001426572177409,
      "loss": 2.2652,
      "step": 140088
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.0326436758041382,
      "learning_rate": 0.00020013879992085654,
      "loss": 2.2996,
      "step": 140089
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.020513653755188,
      "learning_rate": 0.0002001349426425382,
      "loss": 1.9224,
      "step": 140090
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.120420217514038,
      "learning_rate": 0.00020013108538278676,
      "loss": 2.24,
      "step": 140091
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.087972640991211,
      "learning_rate": 0.00020012722814160278,
      "loss": 2.1911,
      "step": 140092
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.0838992595672607,
      "learning_rate": 0.000200123370918987,
      "loss": 2.1196,
      "step": 140093
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.0866750478744507,
      "learning_rate": 0.00020011951371494028,
      "loss": 1.9199,
      "step": 140094
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.2050117254257202,
      "learning_rate": 0.00020011565652946315,
      "loss": 2.0917,
      "step": 140095
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.0509964227676392,
      "learning_rate": 0.00020011179936255637,
      "loss": 2.1296,
      "step": 140096
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.0145679712295532,
      "learning_rate": 0.00020010794221422078,
      "loss": 2.1146,
      "step": 140097
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.0666948556900024,
      "learning_rate": 0.00020010408508445697,
      "loss": 1.9902,
      "step": 140098
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.2665687799453735,
      "learning_rate": 0.00020010022797326566,
      "loss": 2.1464,
      "step": 140099
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.1263699531555176,
      "learning_rate": 0.00020009637088064762,
      "loss": 2.2424,
      "step": 140100
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.0085699558258057,
      "learning_rate": 0.0002000925138066036,
      "loss": 1.9642,
      "step": 140101
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.1312693357467651,
      "learning_rate": 0.00020008865675113423,
      "loss": 2.1824,
      "step": 140102
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.2147588729858398,
      "learning_rate": 0.0002000847997142403,
      "loss": 2.0983,
      "step": 140103
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.3185625076293945,
      "learning_rate": 0.00020008094269592246,
      "loss": 2.1167,
      "step": 140104
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.067784070968628,
      "learning_rate": 0.00020007708569618144,
      "loss": 1.9729,
      "step": 140105
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.0404571294784546,
      "learning_rate": 0.00020007322871501797,
      "loss": 2.1608,
      "step": 140106
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.042322039604187,
      "learning_rate": 0.00020006937175243275,
      "loss": 2.0921,
      "step": 140107
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.1077396869659424,
      "learning_rate": 0.00020006551480842658,
      "loss": 2.1692,
      "step": 140108
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.0461937189102173,
      "learning_rate": 0.00020006165788300021,
      "loss": 1.979,
      "step": 140109
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.071120023727417,
      "learning_rate": 0.00020005780097615415,
      "loss": 2.5001,
      "step": 140110
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.0602971315383911,
      "learning_rate": 0.00020005394408788922,
      "loss": 2.2205,
      "step": 140111
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.099718451499939,
      "learning_rate": 0.00020005008721820617,
      "loss": 2.025,
      "step": 140112
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.1225781440734863,
      "learning_rate": 0.00020004623036710566,
      "loss": 2.0381,
      "step": 140113
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.0089927911758423,
      "learning_rate": 0.00020004237353458848,
      "loss": 1.9587,
      "step": 140114
    },
    {
      "epoch": 1.82,
      "grad_norm": 0.9727701544761658,
      "learning_rate": 0.0002000385167206554,
      "loss": 2.1487,
      "step": 140115
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.4254522323608398,
      "learning_rate": 0.00020003465992530692,
      "loss": 2.3005,
      "step": 140116
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.4508343935012817,
      "learning_rate": 0.0002000308031485439,
      "loss": 1.968,
      "step": 140117
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.0008364915847778,
      "learning_rate": 0.00020002694639036705,
      "loss": 2.1062,
      "step": 140118
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.241823673248291,
      "learning_rate": 0.00020002308965077708,
      "loss": 1.901,
      "step": 140119
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.3314111232757568,
      "learning_rate": 0.00020001923292977465,
      "loss": 2.0932,
      "step": 140120
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.187595248222351,
      "learning_rate": 0.00020001537622736074,
      "loss": 2.1257,
      "step": 140121
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.2212761640548706,
      "learning_rate": 0.00020001151954353568,
      "loss": 2.2613,
      "step": 140122
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.3392373323440552,
      "learning_rate": 0.00020000766287830036,
      "loss": 1.9216,
      "step": 140123
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.1141530275344849,
      "learning_rate": 0.00020000380623165552,
      "loss": 2.0842,
      "step": 140124
    },
    {
      "epoch": 1.82,
      "grad_norm": 0.9816145896911621,
      "learning_rate": 0.00019999994960360188,
      "loss": 2.2523,
      "step": 140125
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.1584234237670898,
      "learning_rate": 0.00019999609299414014,
      "loss": 2.1289,
      "step": 140126
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.0426788330078125,
      "learning_rate": 0.00019999223640327097,
      "loss": 2.0383,
      "step": 140127
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.077562689781189,
      "learning_rate": 0.0001999883798309953,
      "loss": 2.0283,
      "step": 140128
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.0747148990631104,
      "learning_rate": 0.00019998452327731352,
      "loss": 1.8824,
      "step": 140129
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.0818910598754883,
      "learning_rate": 0.0001999806667422265,
      "loss": 2.403,
      "step": 140130
    },
    {
      "epoch": 1.82,
      "grad_norm": 0.9963242411613464,
      "learning_rate": 0.000199976810225735,
      "loss": 2.0515,
      "step": 140131
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.002653956413269,
      "learning_rate": 0.00019997295372783968,
      "loss": 2.1836,
      "step": 140132
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.0472404956817627,
      "learning_rate": 0.0001999690972485413,
      "loss": 2.0821,
      "step": 140133
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.1548939943313599,
      "learning_rate": 0.00019996524078784065,
      "loss": 2.1907,
      "step": 140134
    },
    {
      "epoch": 1.82,
      "grad_norm": 0.9869696497917175,
      "learning_rate": 0.00019996138434573822,
      "loss": 2.093,
      "step": 140135
    },
    {
      "epoch": 1.82,
      "grad_norm": 0.9869215488433838,
      "learning_rate": 0.00019995752792223485,
      "loss": 2.0624,
      "step": 140136
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.0878629684448242,
      "learning_rate": 0.0001999536715173313,
      "loss": 2.3118,
      "step": 140137
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.0837547779083252,
      "learning_rate": 0.0001999498151310282,
      "loss": 2.1668,
      "step": 140138
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.0909459590911865,
      "learning_rate": 0.00019994595876332636,
      "loss": 2.2365,
      "step": 140139
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.0797576904296875,
      "learning_rate": 0.00019994210241422659,
      "loss": 2.0008,
      "step": 140140
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.0630805492401123,
      "learning_rate": 0.00019993824608372933,
      "loss": 2.11,
      "step": 140141
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.1047611236572266,
      "learning_rate": 0.0001999343897718354,
      "loss": 2.3356,
      "step": 140142
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.0104693174362183,
      "learning_rate": 0.00019993053347854562,
      "loss": 1.9595,
      "step": 140143
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.015995740890503,
      "learning_rate": 0.00019992667720386057,
      "loss": 2.1564,
      "step": 140144
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.0316522121429443,
      "learning_rate": 0.0001999228209477811,
      "loss": 2.2145,
      "step": 140145
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.0101720094680786,
      "learning_rate": 0.00019991896471030797,
      "loss": 1.9228,
      "step": 140146
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.0375076532363892,
      "learning_rate": 0.0001999151084914417,
      "loss": 1.9593,
      "step": 140147
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.075200080871582,
      "learning_rate": 0.00019991125229118303,
      "loss": 2.1399,
      "step": 140148
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.065924882888794,
      "learning_rate": 0.00019990739610953282,
      "loss": 2.0998,
      "step": 140149
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.082108974456787,
      "learning_rate": 0.00019990353994649168,
      "loss": 2.0681,
      "step": 140150
    },
    {
      "epoch": 1.82,
      "grad_norm": 0.999649167060852,
      "learning_rate": 0.00019989968380206038,
      "loss": 2.1982,
      "step": 140151
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.02454674243927,
      "learning_rate": 0.00019989582767623973,
      "loss": 2.1774,
      "step": 140152
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.0193589925765991,
      "learning_rate": 0.00019989197156903017,
      "loss": 2.1284,
      "step": 140153
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.0255045890808105,
      "learning_rate": 0.00019988811548043265,
      "loss": 2.0024,
      "step": 140154
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.0925829410552979,
      "learning_rate": 0.00019988425941044777,
      "loss": 1.9126,
      "step": 140155
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.1389050483703613,
      "learning_rate": 0.00019988040335907635,
      "loss": 2.0566,
      "step": 140156
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.1100534200668335,
      "learning_rate": 0.000199876547326319,
      "loss": 2.1812,
      "step": 140157
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.0602171421051025,
      "learning_rate": 0.00019987269131217664,
      "loss": 2.1079,
      "step": 140158
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.1269981861114502,
      "learning_rate": 0.00019986883531664972,
      "loss": 2.3666,
      "step": 140159
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.029258370399475,
      "learning_rate": 0.00019986497933973906,
      "loss": 2.427,
      "step": 140160
    },
    {
      "epoch": 1.82,
      "grad_norm": 1.0453401803970337,
      "learning_rate": 0.00019986112338144538,
      "loss": 2.0666,
      "step": 140161
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.0929627418518066,
      "learning_rate": 0.0001998572674417694,
      "loss": 2.3398,
      "step": 140162
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.0271086692810059,
      "learning_rate": 0.00019985341152071187,
      "loss": 2.0559,
      "step": 140163
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.1044697761535645,
      "learning_rate": 0.0001998495556182736,
      "loss": 1.9263,
      "step": 140164
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.0244073867797852,
      "learning_rate": 0.00019984569973445507,
      "loss": 2.0015,
      "step": 140165
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.1018024682998657,
      "learning_rate": 0.0001998418438692571,
      "loss": 2.0066,
      "step": 140166
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.1385787725448608,
      "learning_rate": 0.00019983798802268045,
      "loss": 2.1153,
      "step": 140167
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.0826770067214966,
      "learning_rate": 0.00019983413219472579,
      "loss": 2.0876,
      "step": 140168
    },
    {
      "epoch": 1.83,
      "grad_norm": 0.9749101400375366,
      "learning_rate": 0.00019983027638539386,
      "loss": 2.1479,
      "step": 140169
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.0302382707595825,
      "learning_rate": 0.00019982642059468542,
      "loss": 2.1336,
      "step": 140170
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.1202882528305054,
      "learning_rate": 0.00019982256482260107,
      "loss": 1.8604,
      "step": 140171
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.0581393241882324,
      "learning_rate": 0.0001998187090691417,
      "loss": 2.0536,
      "step": 140172
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.1282930374145508,
      "learning_rate": 0.0001998148533343078,
      "loss": 2.191,
      "step": 140173
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.0839800834655762,
      "learning_rate": 0.00019981099761810024,
      "loss": 2.1359,
      "step": 140174
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.0299478769302368,
      "learning_rate": 0.00019980714192051975,
      "loss": 1.9294,
      "step": 140175
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.1622005701065063,
      "learning_rate": 0.00019980328624156704,
      "loss": 2.3764,
      "step": 140176
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.0114052295684814,
      "learning_rate": 0.0001997994305812427,
      "loss": 2.1049,
      "step": 140177
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.0393184423446655,
      "learning_rate": 0.00019979557493954753,
      "loss": 1.8939,
      "step": 140178
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.2182005643844604,
      "learning_rate": 0.00019979171931648235,
      "loss": 2.248,
      "step": 140179
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.1049286127090454,
      "learning_rate": 0.0001997878637120477,
      "loss": 2.1401,
      "step": 140180
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.0457394123077393,
      "learning_rate": 0.0001997840081262444,
      "loss": 2.1535,
      "step": 140181
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.091803789138794,
      "learning_rate": 0.00019978015255907316,
      "loss": 1.9887,
      "step": 140182
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.028928518295288,
      "learning_rate": 0.00019977629701053466,
      "loss": 1.8905,
      "step": 140183
    },
    {
      "epoch": 1.83,
      "grad_norm": 0.9896581768989563,
      "learning_rate": 0.00019977244148062964,
      "loss": 2.0851,
      "step": 140184
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.118327021598816,
      "learning_rate": 0.0001997685859693588,
      "loss": 2.0087,
      "step": 140185
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.160656452178955,
      "learning_rate": 0.00019976473047672293,
      "loss": 2.1478,
      "step": 140186
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.1380919218063354,
      "learning_rate": 0.00019976087500272265,
      "loss": 2.0252,
      "step": 140187
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.0187227725982666,
      "learning_rate": 0.00019975701954735875,
      "loss": 1.9945,
      "step": 140188
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.4123547077178955,
      "learning_rate": 0.00019975316411063187,
      "loss": 2.1035,
      "step": 140189
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.0550075769424438,
      "learning_rate": 0.00019974930869254276,
      "loss": 2.0213,
      "step": 140190
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.1053756475448608,
      "learning_rate": 0.00019974545329309214,
      "loss": 2.0276,
      "step": 140191
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.7567439079284668,
      "learning_rate": 0.00019974159791228072,
      "loss": 1.8486,
      "step": 140192
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.1637376546859741,
      "learning_rate": 0.00019973774255010936,
      "loss": 2.2631,
      "step": 140193
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.1674492359161377,
      "learning_rate": 0.00019973388720657853,
      "loss": 2.1615,
      "step": 140194
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.162945032119751,
      "learning_rate": 0.00019973003188168912,
      "loss": 2.0159,
      "step": 140195
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.2948119640350342,
      "learning_rate": 0.00019972617657544175,
      "loss": 2.0576,
      "step": 140196
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.2532236576080322,
      "learning_rate": 0.00019972232128783715,
      "loss": 1.9821,
      "step": 140197
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.1045433282852173,
      "learning_rate": 0.0001997184660188761,
      "loss": 2.0884,
      "step": 140198
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.1055619716644287,
      "learning_rate": 0.00019971461076855927,
      "loss": 2.3381,
      "step": 140199
    },
    {
      "epoch": 1.83,
      "grad_norm": 0.9915339946746826,
      "learning_rate": 0.00019971075553688736,
      "loss": 1.9357,
      "step": 140200
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.11838960647583,
      "learning_rate": 0.00019970690032386124,
      "loss": 2.2029,
      "step": 140201
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.0629438161849976,
      "learning_rate": 0.00019970304512948142,
      "loss": 2.206,
      "step": 140202
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.0529950857162476,
      "learning_rate": 0.00019969918995374863,
      "loss": 2.2261,
      "step": 140203
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.0324665307998657,
      "learning_rate": 0.0001996953347966637,
      "loss": 2.194,
      "step": 140204
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.5579222440719604,
      "learning_rate": 0.0001996914796582273,
      "loss": 2.0057,
      "step": 140205
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.1921236515045166,
      "learning_rate": 0.00019968762453844013,
      "loss": 2.0517,
      "step": 140206
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.218930721282959,
      "learning_rate": 0.00019968376943730308,
      "loss": 2.1781,
      "step": 140207
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.0982284545898438,
      "learning_rate": 0.00019967991435481655,
      "loss": 2.2164,
      "step": 140208
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.2560110092163086,
      "learning_rate": 0.00019967605929098144,
      "loss": 2.1789,
      "step": 140209
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.2101107835769653,
      "learning_rate": 0.00019967220424579844,
      "loss": 2.1665,
      "step": 140210
    },
    {
      "epoch": 1.83,
      "grad_norm": 0.9884441494941711,
      "learning_rate": 0.00019966834921926826,
      "loss": 1.9638,
      "step": 140211
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.0703644752502441,
      "learning_rate": 0.00019966449421139162,
      "loss": 2.3992,
      "step": 140212
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.0461959838867188,
      "learning_rate": 0.00019966063922216936,
      "loss": 1.8899,
      "step": 140213
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.045232892036438,
      "learning_rate": 0.00019965678425160201,
      "loss": 2.1548,
      "step": 140214
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.0366379022598267,
      "learning_rate": 0.00019965292929969032,
      "loss": 2.2749,
      "step": 140215
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.0173072814941406,
      "learning_rate": 0.00019964907436643506,
      "loss": 1.9201,
      "step": 140216
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.1227651834487915,
      "learning_rate": 0.00019964521945183691,
      "loss": 2.2396,
      "step": 140217
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.1227343082427979,
      "learning_rate": 0.00019964136455589664,
      "loss": 1.9381,
      "step": 140218
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.0610047578811646,
      "learning_rate": 0.00019963750967861508,
      "loss": 2.0475,
      "step": 140219
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.0043847560882568,
      "learning_rate": 0.00019963365481999264,
      "loss": 2.1922,
      "step": 140220
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.3977218866348267,
      "learning_rate": 0.00019962979998003016,
      "loss": 1.932,
      "step": 140221
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.225472092628479,
      "learning_rate": 0.00019962594515872842,
      "loss": 1.9879,
      "step": 140222
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.1442724466323853,
      "learning_rate": 0.00019962209035608814,
      "loss": 2.1822,
      "step": 140223
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.1411150693893433,
      "learning_rate": 0.00019961823557210998,
      "loss": 2.0657,
      "step": 140224
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.0935051441192627,
      "learning_rate": 0.00019961438080679485,
      "loss": 2.1666,
      "step": 140225
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.2216074466705322,
      "learning_rate": 0.00019961052606014314,
      "loss": 2.0438,
      "step": 140226
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.124814510345459,
      "learning_rate": 0.00019960667133215573,
      "loss": 1.9309,
      "step": 140227
    },
    {
      "epoch": 1.83,
      "grad_norm": 0.9944165945053101,
      "learning_rate": 0.00019960281662283335,
      "loss": 1.7429,
      "step": 140228
    },
    {
      "epoch": 1.83,
      "grad_norm": 0.9861517548561096,
      "learning_rate": 0.00019959896193217668,
      "loss": 1.8099,
      "step": 140229
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.2078379392623901,
      "learning_rate": 0.0001995951072601865,
      "loss": 2.1101,
      "step": 140230
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.0921071767807007,
      "learning_rate": 0.00019959125260686356,
      "loss": 2.0238,
      "step": 140231
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.0361305475234985,
      "learning_rate": 0.0001995873979722084,
      "loss": 2.2084,
      "step": 140232
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.396721363067627,
      "learning_rate": 0.0001995835433562218,
      "loss": 1.9181,
      "step": 140233
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.0963386297225952,
      "learning_rate": 0.00019957968875890456,
      "loss": 2.0739,
      "step": 140234
    },
    {
      "epoch": 1.83,
      "grad_norm": 0.9890434741973877,
      "learning_rate": 0.00019957583418025735,
      "loss": 2.0617,
      "step": 140235
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.0441442728042603,
      "learning_rate": 0.0001995719796202809,
      "loss": 2.0942,
      "step": 140236
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.2309844493865967,
      "learning_rate": 0.00019956812507897602,
      "loss": 2.1599,
      "step": 140237
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.0638798475265503,
      "learning_rate": 0.00019956427055634317,
      "loss": 2.099,
      "step": 140238
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.0896458625793457,
      "learning_rate": 0.00019956041605238327,
      "loss": 2.2335,
      "step": 140239
    },
    {
      "epoch": 1.83,
      "grad_norm": 0.9835795760154724,
      "learning_rate": 0.00019955656156709695,
      "loss": 2.1476,
      "step": 140240
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.122799277305603,
      "learning_rate": 0.00019955270710048497,
      "loss": 2.0564,
      "step": 140241
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.092343807220459,
      "learning_rate": 0.000199548852652548,
      "loss": 2.1867,
      "step": 140242
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.0653996467590332,
      "learning_rate": 0.00019954499822328697,
      "loss": 1.9018,
      "step": 140243
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.1367714405059814,
      "learning_rate": 0.00019954114381270227,
      "loss": 2.1114,
      "step": 140244
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.1434844732284546,
      "learning_rate": 0.00019953728942079482,
      "loss": 2.1425,
      "step": 140245
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.0943936109542847,
      "learning_rate": 0.00019953343504756525,
      "loss": 1.9951,
      "step": 140246
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.3069064617156982,
      "learning_rate": 0.00019952958069301428,
      "loss": 2.1678,
      "step": 140247
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.076859712600708,
      "learning_rate": 0.00019952572635714268,
      "loss": 2.0045,
      "step": 140248
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.0618329048156738,
      "learning_rate": 0.0001995218720399513,
      "loss": 2.1821,
      "step": 140249
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.310102939605713,
      "learning_rate": 0.00019951801774144052,
      "loss": 2.0329,
      "step": 140250
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.4160223007202148,
      "learning_rate": 0.00019951416346161127,
      "loss": 2.2595,
      "step": 140251
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.0355802774429321,
      "learning_rate": 0.00019951030920046422,
      "loss": 2.2706,
      "step": 140252
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.083634376525879,
      "learning_rate": 0.0001995064549580001,
      "loss": 2.0405,
      "step": 140253
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.1549241542816162,
      "learning_rate": 0.00019950260073421964,
      "loss": 2.1487,
      "step": 140254
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.1713122129440308,
      "learning_rate": 0.00019949874652912364,
      "loss": 2.1622,
      "step": 140255
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.0139853954315186,
      "learning_rate": 0.00019949489234271267,
      "loss": 2.2899,
      "step": 140256
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.0569881200790405,
      "learning_rate": 0.00019949103817498744,
      "loss": 1.9043,
      "step": 140257
    },
    {
      "epoch": 1.83,
      "grad_norm": 0.9961456656455994,
      "learning_rate": 0.00019948718402594873,
      "loss": 2.1232,
      "step": 140258
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.043271780014038,
      "learning_rate": 0.00019948332989559725,
      "loss": 1.8907,
      "step": 140259
    },
    {
      "epoch": 1.83,
      "grad_norm": 0.9589444994926453,
      "learning_rate": 0.0001994794757839337,
      "loss": 2.2709,
      "step": 140260
    },
    {
      "epoch": 1.83,
      "grad_norm": 0.9539219737052917,
      "learning_rate": 0.00019947562169095887,
      "loss": 2.0756,
      "step": 140261
    },
    {
      "epoch": 1.83,
      "grad_norm": 0.9698629975318909,
      "learning_rate": 0.00019947176761667343,
      "loss": 2.0332,
      "step": 140262
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.041182518005371,
      "learning_rate": 0.0001994679135610781,
      "loss": 2.0417,
      "step": 140263
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.1195579767227173,
      "learning_rate": 0.0001994640595241735,
      "loss": 2.1134,
      "step": 140264
    },
    {
      "epoch": 1.83,
      "grad_norm": 0.9988510608673096,
      "learning_rate": 0.00019946020550596044,
      "loss": 1.9961,
      "step": 140265
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.0592072010040283,
      "learning_rate": 0.00019945635150643966,
      "loss": 2.0032,
      "step": 140266
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.051259994506836,
      "learning_rate": 0.00019945249752561177,
      "loss": 1.9815,
      "step": 140267
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.0270887613296509,
      "learning_rate": 0.0001994486435634777,
      "loss": 2.1583,
      "step": 140268
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.0593715906143188,
      "learning_rate": 0.0001994447896200379,
      "loss": 2.0784,
      "step": 140269
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.0742508172988892,
      "learning_rate": 0.00019944093569529325,
      "loss": 1.9925,
      "step": 140270
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.2178198099136353,
      "learning_rate": 0.0001994370817892445,
      "loss": 2.2879,
      "step": 140271
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.0675864219665527,
      "learning_rate": 0.0001994332279018922,
      "loss": 2.0652,
      "step": 140272
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.0986279249191284,
      "learning_rate": 0.00019942937403323718,
      "loss": 1.9379,
      "step": 140273
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.1576721668243408,
      "learning_rate": 0.0001994255201832802,
      "loss": 2.1376,
      "step": 140274
    },
    {
      "epoch": 1.83,
      "grad_norm": 0.9917062520980835,
      "learning_rate": 0.00019942166635202185,
      "loss": 1.9595,
      "step": 140275
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.1096806526184082,
      "learning_rate": 0.0001994178125394629,
      "loss": 2.2235,
      "step": 140276
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.0491284132003784,
      "learning_rate": 0.00019941395874560408,
      "loss": 2.5005,
      "step": 140277
    },
    {
      "epoch": 1.83,
      "grad_norm": 0.9984053373336792,
      "learning_rate": 0.00019941010497044617,
      "loss": 2.0257,
      "step": 140278
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.0395022630691528,
      "learning_rate": 0.00019940625121398975,
      "loss": 2.0276,
      "step": 140279
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.2705439329147339,
      "learning_rate": 0.00019940239747623568,
      "loss": 1.9894,
      "step": 140280
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.004600167274475,
      "learning_rate": 0.00019939854375718453,
      "loss": 2.0784,
      "step": 140281
    },
    {
      "epoch": 1.83,
      "grad_norm": 0.9932718276977539,
      "learning_rate": 0.0001993946900568371,
      "loss": 2.1759,
      "step": 140282
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.006454586982727,
      "learning_rate": 0.00019939083637519412,
      "loss": 2.036,
      "step": 140283
    },
    {
      "epoch": 1.83,
      "grad_norm": 0.9949490427970886,
      "learning_rate": 0.00019938698271225626,
      "loss": 2.0132,
      "step": 140284
    },
    {
      "epoch": 1.83,
      "grad_norm": 0.9905133247375488,
      "learning_rate": 0.0001993831290680243,
      "loss": 2.0962,
      "step": 140285
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.0508670806884766,
      "learning_rate": 0.00019937927544249893,
      "loss": 1.9232,
      "step": 140286
    },
    {
      "epoch": 1.83,
      "grad_norm": 0.9857302904129028,
      "learning_rate": 0.0001993754218356808,
      "loss": 2.0816,
      "step": 140287
    },
    {
      "epoch": 1.83,
      "grad_norm": 0.9485806226730347,
      "learning_rate": 0.0001993715682475707,
      "loss": 2.0245,
      "step": 140288
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.0789737701416016,
      "learning_rate": 0.00019936771467816926,
      "loss": 2.1318,
      "step": 140289
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.0134871006011963,
      "learning_rate": 0.00019936386112747728,
      "loss": 2.0198,
      "step": 140290
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.184881329536438,
      "learning_rate": 0.0001993600075954955,
      "loss": 2.1815,
      "step": 140291
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.0928728580474854,
      "learning_rate": 0.0001993561540822247,
      "loss": 2.4478,
      "step": 140292
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.16668701171875,
      "learning_rate": 0.00019935230058766538,
      "loss": 2.1226,
      "step": 140293
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.2407549619674683,
      "learning_rate": 0.00019934844711181834,
      "loss": 2.3763,
      "step": 140294
    },
    {
      "epoch": 1.83,
      "grad_norm": 0.9701288938522339,
      "learning_rate": 0.0001993445936546843,
      "loss": 1.8989,
      "step": 140295
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.0257331132888794,
      "learning_rate": 0.00019934074021626407,
      "loss": 2.0263,
      "step": 140296
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.1199331283569336,
      "learning_rate": 0.00019933688679655824,
      "loss": 2.1013,
      "step": 140297
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.1175233125686646,
      "learning_rate": 0.00019933303339556775,
      "loss": 2.0519,
      "step": 140298
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.0434770584106445,
      "learning_rate": 0.00019932918001329297,
      "loss": 2.0574,
      "step": 140299
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.1347407102584839,
      "learning_rate": 0.00019932532664973483,
      "loss": 1.8636,
      "step": 140300
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.0527197122573853,
      "learning_rate": 0.00019932147330489403,
      "loss": 2.0951,
      "step": 140301
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.1146607398986816,
      "learning_rate": 0.00019931761997877125,
      "loss": 2.2563,
      "step": 140302
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.0133312940597534,
      "learning_rate": 0.00019931376667136721,
      "loss": 2.0902,
      "step": 140303
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.1331802606582642,
      "learning_rate": 0.00019930991338268277,
      "loss": 2.0946,
      "step": 140304
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.0347169637680054,
      "learning_rate": 0.00019930606011271838,
      "loss": 1.972,
      "step": 140305
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.0589889287948608,
      "learning_rate": 0.00019930220686147494,
      "loss": 1.9005,
      "step": 140306
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.1039512157440186,
      "learning_rate": 0.00019929835362895312,
      "loss": 2.0134,
      "step": 140307
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.1075825691223145,
      "learning_rate": 0.0001992945004151536,
      "loss": 1.8971,
      "step": 140308
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.5215059518814087,
      "learning_rate": 0.00019929064722007714,
      "loss": 1.9556,
      "step": 140309
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.0524190664291382,
      "learning_rate": 0.00019928679404372456,
      "loss": 2.1909,
      "step": 140310
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.0754961967468262,
      "learning_rate": 0.0001992829408860964,
      "loss": 2.1011,
      "step": 140311
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.0384925603866577,
      "learning_rate": 0.0001992790877471934,
      "loss": 1.8407,
      "step": 140312
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.0690537691116333,
      "learning_rate": 0.00019927523462701633,
      "loss": 1.8791,
      "step": 140313
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.0072988271713257,
      "learning_rate": 0.00019927138152556586,
      "loss": 2.1422,
      "step": 140314
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.4333927631378174,
      "learning_rate": 0.00019926752844284278,
      "loss": 2.1966,
      "step": 140315
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.2430981397628784,
      "learning_rate": 0.00019926367537884788,
      "loss": 1.996,
      "step": 140316
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.1093624830245972,
      "learning_rate": 0.00019925982233358165,
      "loss": 2.0167,
      "step": 140317
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.0533727407455444,
      "learning_rate": 0.0001992559693070449,
      "loss": 2.1924,
      "step": 140318
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.1346139907836914,
      "learning_rate": 0.0001992521162992384,
      "loss": 2.182,
      "step": 140319
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.0982122421264648,
      "learning_rate": 0.0001992482633101628,
      "loss": 2.0401,
      "step": 140320
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.0366560220718384,
      "learning_rate": 0.00019924441033981888,
      "loss": 2.1178,
      "step": 140321
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.052451729774475,
      "learning_rate": 0.00019924055738820742,
      "loss": 2.239,
      "step": 140322
    },
    {
      "epoch": 1.83,
      "grad_norm": 0.9881660342216492,
      "learning_rate": 0.00019923670445532894,
      "loss": 2.1417,
      "step": 140323
    },
    {
      "epoch": 1.83,
      "grad_norm": 0.993109941482544,
      "learning_rate": 0.00019923285154118422,
      "loss": 2.0372,
      "step": 140324
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.0242269039154053,
      "learning_rate": 0.0001992289986457741,
      "loss": 2.0679,
      "step": 140325
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.1060658693313599,
      "learning_rate": 0.0001992251457690991,
      "loss": 2.0606,
      "step": 140326
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.1859259605407715,
      "learning_rate": 0.0001992212929111601,
      "loss": 2.1622,
      "step": 140327
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.026257872581482,
      "learning_rate": 0.0001992174400719578,
      "loss": 2.0787,
      "step": 140328
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.128517508506775,
      "learning_rate": 0.000199213587251493,
      "loss": 1.9034,
      "step": 140329
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.0347899198532104,
      "learning_rate": 0.0001992097344497661,
      "loss": 2.3125,
      "step": 140330
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.1039109230041504,
      "learning_rate": 0.00019920588166677807,
      "loss": 2.0944,
      "step": 140331
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.1141051054000854,
      "learning_rate": 0.00019920202890252953,
      "loss": 1.9915,
      "step": 140332
    },
    {
      "epoch": 1.83,
      "grad_norm": 0.9970365166664124,
      "learning_rate": 0.00019919817615702133,
      "loss": 2.0508,
      "step": 140333
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.1513986587524414,
      "learning_rate": 0.000199194323430254,
      "loss": 2.1626,
      "step": 140334
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.0805741548538208,
      "learning_rate": 0.0001991904707222285,
      "loss": 2.1153,
      "step": 140335
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.0832719802856445,
      "learning_rate": 0.00019918661803294526,
      "loss": 2.1164,
      "step": 140336
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.158699631690979,
      "learning_rate": 0.00019918276536240513,
      "loss": 2.4242,
      "step": 140337
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.014589548110962,
      "learning_rate": 0.00019917891271060883,
      "loss": 2.0787,
      "step": 140338
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.0573612451553345,
      "learning_rate": 0.00019917506007755708,
      "loss": 1.979,
      "step": 140339
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.0901803970336914,
      "learning_rate": 0.00019917120746325062,
      "loss": 2.1427,
      "step": 140340
    },
    {
      "epoch": 1.83,
      "grad_norm": 0.9986729025840759,
      "learning_rate": 0.0001991673548676902,
      "loss": 2.3354,
      "step": 140341
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.1293444633483887,
      "learning_rate": 0.00019916350229087638,
      "loss": 2.177,
      "step": 140342
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.1319535970687866,
      "learning_rate": 0.00019915964973280998,
      "loss": 2.015,
      "step": 140343
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.1245232820510864,
      "learning_rate": 0.00019915579719349168,
      "loss": 2.1244,
      "step": 140344
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.2459709644317627,
      "learning_rate": 0.00019915194467292222,
      "loss": 2.2579,
      "step": 140345
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.0557948350906372,
      "learning_rate": 0.00019914809217110233,
      "loss": 1.964,
      "step": 140346
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.1080657243728638,
      "learning_rate": 0.00019914423968803277,
      "loss": 2.1174,
      "step": 140347
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.0299715995788574,
      "learning_rate": 0.0001991403872237142,
      "loss": 1.9791,
      "step": 140348
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.1717987060546875,
      "learning_rate": 0.00019913653477814731,
      "loss": 2.1281,
      "step": 140349
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.0696154832839966,
      "learning_rate": 0.00019913268235133278,
      "loss": 2.4361,
      "step": 140350
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.02515709400177,
      "learning_rate": 0.00019912882994327138,
      "loss": 2.1065,
      "step": 140351
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.083678960800171,
      "learning_rate": 0.0001991249775539639,
      "loss": 1.9422,
      "step": 140352
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.0268217325210571,
      "learning_rate": 0.00019912112518341102,
      "loss": 2.149,
      "step": 140353
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.0640325546264648,
      "learning_rate": 0.00019911727283161338,
      "loss": 1.9538,
      "step": 140354
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.0669527053833008,
      "learning_rate": 0.0001991134204985718,
      "loss": 2.1257,
      "step": 140355
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.1717838048934937,
      "learning_rate": 0.00019910956818428686,
      "loss": 2.1385,
      "step": 140356
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.0757125616073608,
      "learning_rate": 0.00019910571588875936,
      "loss": 2.2764,
      "step": 140357
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.0630542039871216,
      "learning_rate": 0.00019910186361199003,
      "loss": 2.1282,
      "step": 140358
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.0073306560516357,
      "learning_rate": 0.00019909801135397964,
      "loss": 2.1649,
      "step": 140359
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.0381901264190674,
      "learning_rate": 0.00019909415911472872,
      "loss": 2.1127,
      "step": 140360
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.1644561290740967,
      "learning_rate": 0.00019909030689423812,
      "loss": 2.3081,
      "step": 140361
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.1484829187393188,
      "learning_rate": 0.0001990864546925086,
      "loss": 2.0757,
      "step": 140362
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.0544912815093994,
      "learning_rate": 0.00019908260250954082,
      "loss": 1.8902,
      "step": 140363
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.1870810985565186,
      "learning_rate": 0.00019907875034533542,
      "loss": 2.2373,
      "step": 140364
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.2488503456115723,
      "learning_rate": 0.00019907489819989327,
      "loss": 2.3406,
      "step": 140365
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.0666232109069824,
      "learning_rate": 0.0001990710460732149,
      "loss": 1.8832,
      "step": 140366
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.1112018823623657,
      "learning_rate": 0.0001990671939653012,
      "loss": 2.3708,
      "step": 140367
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.0803617238998413,
      "learning_rate": 0.00019906334187615271,
      "loss": 1.9756,
      "step": 140368
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.108137607574463,
      "learning_rate": 0.00019905948980577034,
      "loss": 1.9819,
      "step": 140369
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.4680614471435547,
      "learning_rate": 0.00019905563775415475,
      "loss": 2.1347,
      "step": 140370
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.1797528266906738,
      "learning_rate": 0.0001990517857213066,
      "loss": 2.0751,
      "step": 140371
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.1481857299804688,
      "learning_rate": 0.00019904793370722663,
      "loss": 2.1133,
      "step": 140372
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.092583179473877,
      "learning_rate": 0.0001990440817119155,
      "loss": 2.1045,
      "step": 140373
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.1602497100830078,
      "learning_rate": 0.00019904022973537397,
      "loss": 2.1526,
      "step": 140374
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.09846830368042,
      "learning_rate": 0.00019903637777760283,
      "loss": 2.372,
      "step": 140375
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.0358190536499023,
      "learning_rate": 0.00019903252583860264,
      "loss": 2.2729,
      "step": 140376
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.147430419921875,
      "learning_rate": 0.0001990286739183744,
      "loss": 2.0549,
      "step": 140377
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.1131256818771362,
      "learning_rate": 0.0001990248220169185,
      "loss": 2.1596,
      "step": 140378
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.5291695594787598,
      "learning_rate": 0.00019902097013423578,
      "loss": 1.9174,
      "step": 140379
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.0720596313476562,
      "learning_rate": 0.000199017118270327,
      "loss": 2.0482,
      "step": 140380
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.0353100299835205,
      "learning_rate": 0.00019901326642519283,
      "loss": 2.1888,
      "step": 140381
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.0478789806365967,
      "learning_rate": 0.00019900941459883396,
      "loss": 2.1252,
      "step": 140382
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.0594353675842285,
      "learning_rate": 0.00019900556279125133,
      "loss": 2.0365,
      "step": 140383
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.16249418258667,
      "learning_rate": 0.00019900171100244532,
      "loss": 2.0794,
      "step": 140384
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.0206929445266724,
      "learning_rate": 0.0001989978592324168,
      "loss": 2.3431,
      "step": 140385
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.163836121559143,
      "learning_rate": 0.00019899400748116648,
      "loss": 2.0467,
      "step": 140386
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.1151094436645508,
      "learning_rate": 0.00019899015574869505,
      "loss": 1.9551,
      "step": 140387
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.0082032680511475,
      "learning_rate": 0.00019898630403500332,
      "loss": 1.9738,
      "step": 140388
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.027571201324463,
      "learning_rate": 0.00019898245234009206,
      "loss": 2.1196,
      "step": 140389
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.0824342966079712,
      "learning_rate": 0.00019897860066396168,
      "loss": 2.0624,
      "step": 140390
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.193928599357605,
      "learning_rate": 0.00019897474900661314,
      "loss": 2.119,
      "step": 140391
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.378368616104126,
      "learning_rate": 0.00019897089736804709,
      "loss": 1.9913,
      "step": 140392
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.0806533098220825,
      "learning_rate": 0.0001989670457482643,
      "loss": 1.8108,
      "step": 140393
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.0779669284820557,
      "learning_rate": 0.00019896319414726535,
      "loss": 1.9005,
      "step": 140394
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.1938706636428833,
      "learning_rate": 0.0001989593425650511,
      "loss": 2.0254,
      "step": 140395
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.0393773317337036,
      "learning_rate": 0.00019895549100162234,
      "loss": 1.8969,
      "step": 140396
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.1848596334457397,
      "learning_rate": 0.00019895163945697955,
      "loss": 2.1847,
      "step": 140397
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.204765796661377,
      "learning_rate": 0.00019894778793112352,
      "loss": 1.9672,
      "step": 140398
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.2194374799728394,
      "learning_rate": 0.00019894393642405503,
      "loss": 2.2933,
      "step": 140399
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.1597001552581787,
      "learning_rate": 0.00019894008493577477,
      "loss": 2.1279,
      "step": 140400
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.0760914087295532,
      "learning_rate": 0.00019893623346628348,
      "loss": 1.9719,
      "step": 140401
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.133863925933838,
      "learning_rate": 0.0001989323820155819,
      "loss": 2.1146,
      "step": 140402
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.1827932596206665,
      "learning_rate": 0.00019892853058367063,
      "loss": 2.1042,
      "step": 140403
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.144728183746338,
      "learning_rate": 0.00019892467917055048,
      "loss": 2.1441,
      "step": 140404
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.0003650188446045,
      "learning_rate": 0.00019892082777622208,
      "loss": 1.9602,
      "step": 140405
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.0645004510879517,
      "learning_rate": 0.00019891697640068623,
      "loss": 1.9777,
      "step": 140406
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.1338454484939575,
      "learning_rate": 0.00019891312504394362,
      "loss": 2.2663,
      "step": 140407
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.060230016708374,
      "learning_rate": 0.0001989092737059951,
      "loss": 2.1942,
      "step": 140408
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.05009925365448,
      "learning_rate": 0.00019890542238684114,
      "loss": 2.0571,
      "step": 140409
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.1767159700393677,
      "learning_rate": 0.00019890157108648257,
      "loss": 2.0908,
      "step": 140410
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.3063100576400757,
      "learning_rate": 0.00019889771980492009,
      "loss": 2.1901,
      "step": 140411
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.0766936540603638,
      "learning_rate": 0.0001988938685421544,
      "loss": 2.1271,
      "step": 140412
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.0828248262405396,
      "learning_rate": 0.0001988900172981863,
      "loss": 2.1859,
      "step": 140413
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.1138497591018677,
      "learning_rate": 0.00019888616607301656,
      "loss": 2.1966,
      "step": 140414
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.1454964876174927,
      "learning_rate": 0.0001988823148666457,
      "loss": 2.1807,
      "step": 140415
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.1741561889648438,
      "learning_rate": 0.00019887846367907448,
      "loss": 2.0715,
      "step": 140416
    },
    {
      "epoch": 1.83,
      "grad_norm": 0.9855483174324036,
      "learning_rate": 0.0001988746125103037,
      "loss": 1.9158,
      "step": 140417
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.0540252923965454,
      "learning_rate": 0.00019887076136033396,
      "loss": 2.1443,
      "step": 140418
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.0447031259536743,
      "learning_rate": 0.00019886691022916615,
      "loss": 2.2143,
      "step": 140419
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.105384349822998,
      "learning_rate": 0.00019886305911680097,
      "loss": 2.2871,
      "step": 140420
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.0198174715042114,
      "learning_rate": 0.00019885920802323895,
      "loss": 2.0386,
      "step": 140421
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.144429326057434,
      "learning_rate": 0.00019885535694848087,
      "loss": 1.9547,
      "step": 140422
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.0501893758773804,
      "learning_rate": 0.00019885150589252755,
      "loss": 2.0072,
      "step": 140423
    },
    {
      "epoch": 1.83,
      "grad_norm": 0.9927269220352173,
      "learning_rate": 0.0001988476548553796,
      "loss": 2.0448,
      "step": 140424
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.0621260404586792,
      "learning_rate": 0.00019884380383703782,
      "loss": 2.1888,
      "step": 140425
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.2450833320617676,
      "learning_rate": 0.00019883995283750296,
      "loss": 2.3548,
      "step": 140426
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.0246483087539673,
      "learning_rate": 0.0001988361018567756,
      "loss": 2.1852,
      "step": 140427
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.0092172622680664,
      "learning_rate": 0.00019883225089485643,
      "loss": 1.9831,
      "step": 140428
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.0146992206573486,
      "learning_rate": 0.0001988283999517463,
      "loss": 2.4526,
      "step": 140429
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.1286624670028687,
      "learning_rate": 0.0001988245490274459,
      "loss": 2.068,
      "step": 140430
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.093135952949524,
      "learning_rate": 0.0001988206981219559,
      "loss": 2.1323,
      "step": 140431
    },
    {
      "epoch": 1.83,
      "grad_norm": 0.9643558263778687,
      "learning_rate": 0.00019881684723527712,
      "loss": 2.0448,
      "step": 140432
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.0001049041748047,
      "learning_rate": 0.0001988129963674102,
      "loss": 2.2349,
      "step": 140433
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.1575504541397095,
      "learning_rate": 0.0001988091455183558,
      "loss": 2.0887,
      "step": 140434
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.1984189748764038,
      "learning_rate": 0.00019880529468811468,
      "loss": 2.0516,
      "step": 140435
    },
    {
      "epoch": 1.83,
      "grad_norm": 0.9630495309829712,
      "learning_rate": 0.00019880144387668755,
      "loss": 2.0803,
      "step": 140436
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.0898756980895996,
      "learning_rate": 0.00019879759308407517,
      "loss": 2.2302,
      "step": 140437
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.1040409803390503,
      "learning_rate": 0.00019879374231027827,
      "loss": 2.2283,
      "step": 140438
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.2833709716796875,
      "learning_rate": 0.00019878989155529744,
      "loss": 2.0081,
      "step": 140439
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.1449705362319946,
      "learning_rate": 0.00019878604081913359,
      "loss": 2.094,
      "step": 140440
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.0549598932266235,
      "learning_rate": 0.0001987821901017872,
      "loss": 2.0281,
      "step": 140441
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.2636092901229858,
      "learning_rate": 0.00019877833940325915,
      "loss": 2.1343,
      "step": 140442
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.1062383651733398,
      "learning_rate": 0.00019877448872355013,
      "loss": 2.1322,
      "step": 140443
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.1028270721435547,
      "learning_rate": 0.00019877063806266092,
      "loss": 2.3028,
      "step": 140444
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.0975435972213745,
      "learning_rate": 0.00019876678742059204,
      "loss": 1.9929,
      "step": 140445
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.2231556177139282,
      "learning_rate": 0.00019876293679734437,
      "loss": 2.0124,
      "step": 140446
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.2273863554000854,
      "learning_rate": 0.00019875908619291865,
      "loss": 2.0549,
      "step": 140447
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.0987907648086548,
      "learning_rate": 0.00019875523560731543,
      "loss": 2.3278,
      "step": 140448
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.1136467456817627,
      "learning_rate": 0.00019875138504053553,
      "loss": 1.8926,
      "step": 140449
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.022255778312683,
      "learning_rate": 0.0001987475344925797,
      "loss": 2.2847,
      "step": 140450
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.026892900466919,
      "learning_rate": 0.0001987436839634486,
      "loss": 2.076,
      "step": 140451
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.1379574537277222,
      "learning_rate": 0.00019873983345314298,
      "loss": 2.2422,
      "step": 140452
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.1603281497955322,
      "learning_rate": 0.00019873598296166344,
      "loss": 2.0837,
      "step": 140453
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.087895393371582,
      "learning_rate": 0.0001987321324890109,
      "loss": 2.1433,
      "step": 140454
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.0318478345870972,
      "learning_rate": 0.00019872828203518594,
      "loss": 2.1576,
      "step": 140455
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.0191541910171509,
      "learning_rate": 0.00019872443160018935,
      "loss": 2.177,
      "step": 140456
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.2350856065750122,
      "learning_rate": 0.00019872058118402172,
      "loss": 2.0047,
      "step": 140457
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.1257177591323853,
      "learning_rate": 0.0001987167307866839,
      "loss": 2.0016,
      "step": 140458
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.1321344375610352,
      "learning_rate": 0.0001987128804081765,
      "loss": 2.1234,
      "step": 140459
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.108756422996521,
      "learning_rate": 0.0001987090300485003,
      "loss": 2.0356,
      "step": 140460
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.2516695261001587,
      "learning_rate": 0.000198705179707656,
      "loss": 2.2392,
      "step": 140461
    },
    {
      "epoch": 1.83,
      "grad_norm": 0.9272753596305847,
      "learning_rate": 0.00019870132938564438,
      "loss": 1.9673,
      "step": 140462
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.1093297004699707,
      "learning_rate": 0.0001986974790824661,
      "loss": 2.2297,
      "step": 140463
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.1152390241622925,
      "learning_rate": 0.0001986936287981218,
      "loss": 2.2631,
      "step": 140464
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.084472894668579,
      "learning_rate": 0.00019868977853261226,
      "loss": 2.1079,
      "step": 140465
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.0999642610549927,
      "learning_rate": 0.00019868592828593823,
      "loss": 1.911,
      "step": 140466
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.051452875137329,
      "learning_rate": 0.00019868207805810034,
      "loss": 2.0741,
      "step": 140467
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.1965460777282715,
      "learning_rate": 0.00019867822784909944,
      "loss": 1.9787,
      "step": 140468
    },
    {
      "epoch": 1.83,
      "grad_norm": 0.9538325071334839,
      "learning_rate": 0.00019867437765893624,
      "loss": 2.0429,
      "step": 140469
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.1004524230957031,
      "learning_rate": 0.00019867052748761128,
      "loss": 2.3248,
      "step": 140470
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.0786381959915161,
      "learning_rate": 0.0001986666773351254,
      "loss": 1.9742,
      "step": 140471
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.0512605905532837,
      "learning_rate": 0.00019866282720147927,
      "loss": 1.9237,
      "step": 140472
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.2727961540222168,
      "learning_rate": 0.00019865897708667364,
      "loss": 1.9909,
      "step": 140473
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.0099276304244995,
      "learning_rate": 0.00019865512699070924,
      "loss": 1.9412,
      "step": 140474
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.0376157760620117,
      "learning_rate": 0.0001986512769135869,
      "loss": 2.2232,
      "step": 140475
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.0018665790557861,
      "learning_rate": 0.00019864742685530703,
      "loss": 2.1465,
      "step": 140476
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.1346291303634644,
      "learning_rate": 0.00019864357681587051,
      "loss": 2.0725,
      "step": 140477
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.0140106678009033,
      "learning_rate": 0.00019863972679527812,
      "loss": 2.1797,
      "step": 140478
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.126218557357788,
      "learning_rate": 0.0001986358767935305,
      "loss": 2.1323,
      "step": 140479
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.1086984872817993,
      "learning_rate": 0.00019863202681062836,
      "loss": 2.3523,
      "step": 140480
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.0918362140655518,
      "learning_rate": 0.00019862817684657255,
      "loss": 2.1304,
      "step": 140481
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.0507196187973022,
      "learning_rate": 0.0001986243269013636,
      "loss": 1.9581,
      "step": 140482
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.0171667337417603,
      "learning_rate": 0.00019862047697500224,
      "loss": 2.0041,
      "step": 140483
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.030471920967102,
      "learning_rate": 0.00019861662706748933,
      "loss": 1.9151,
      "step": 140484
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.0196666717529297,
      "learning_rate": 0.00019861277717882542,
      "loss": 2.074,
      "step": 140485
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.041022777557373,
      "learning_rate": 0.00019860892730901136,
      "loss": 2.0025,
      "step": 140486
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.0142226219177246,
      "learning_rate": 0.00019860507745804792,
      "loss": 2.0566,
      "step": 140487
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.0265283584594727,
      "learning_rate": 0.0001986012276259356,
      "loss": 1.9784,
      "step": 140488
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.0928930044174194,
      "learning_rate": 0.00019859737781267522,
      "loss": 1.9483,
      "step": 140489
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.0966191291809082,
      "learning_rate": 0.00019859352801826748,
      "loss": 2.1316,
      "step": 140490
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.0724090337753296,
      "learning_rate": 0.00019858967824271315,
      "loss": 2.1427,
      "step": 140491
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.0252814292907715,
      "learning_rate": 0.00019858582848601288,
      "loss": 2.0608,
      "step": 140492
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.0203040838241577,
      "learning_rate": 0.00019858197874816758,
      "loss": 2.1036,
      "step": 140493
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.17300546169281,
      "learning_rate": 0.00019857812902917765,
      "loss": 2.056,
      "step": 140494
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.0236170291900635,
      "learning_rate": 0.00019857427932904397,
      "loss": 2.2869,
      "step": 140495
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.1047817468643188,
      "learning_rate": 0.00019857042964776725,
      "loss": 2.3039,
      "step": 140496
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.0111567974090576,
      "learning_rate": 0.00019856657998534822,
      "loss": 1.8163,
      "step": 140497
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.13970148563385,
      "learning_rate": 0.00019856273034178756,
      "loss": 2.2617,
      "step": 140498
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.041201114654541,
      "learning_rate": 0.0001985588807170861,
      "loss": 2.2044,
      "step": 140499
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.0294122695922852,
      "learning_rate": 0.00019855503111124438,
      "loss": 2.144,
      "step": 140500
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.134871006011963,
      "learning_rate": 0.00019855118152426317,
      "loss": 1.835,
      "step": 140501
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.0997328758239746,
      "learning_rate": 0.0001985473319561432,
      "loss": 2.0143,
      "step": 140502
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.010392427444458,
      "learning_rate": 0.00019854348240688522,
      "loss": 1.9363,
      "step": 140503
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.1425676345825195,
      "learning_rate": 0.0001985396328764899,
      "loss": 2.4241,
      "step": 140504
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.0214349031448364,
      "learning_rate": 0.0001985357833649581,
      "loss": 2.1676,
      "step": 140505
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.0908352136611938,
      "learning_rate": 0.00019853193387229034,
      "loss": 2.1658,
      "step": 140506
    },
    {
      "epoch": 1.83,
      "grad_norm": 0.9953041076660156,
      "learning_rate": 0.00019852808439848737,
      "loss": 2.0817,
      "step": 140507
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.0222017765045166,
      "learning_rate": 0.00019852423494354994,
      "loss": 2.0911,
      "step": 140508
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.1761034727096558,
      "learning_rate": 0.00019852038550747877,
      "loss": 2.0681,
      "step": 140509
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.0431077480316162,
      "learning_rate": 0.00019851653609027458,
      "loss": 2.1268,
      "step": 140510
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.1548207998275757,
      "learning_rate": 0.0001985126866919382,
      "loss": 2.0988,
      "step": 140511
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.0585910081863403,
      "learning_rate": 0.0001985088373124701,
      "loss": 2.1085,
      "step": 140512
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.007051706314087,
      "learning_rate": 0.00019850498795187115,
      "loss": 1.8999,
      "step": 140513
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.0596952438354492,
      "learning_rate": 0.00019850113861014202,
      "loss": 2.133,
      "step": 140514
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.1521567106246948,
      "learning_rate": 0.00019849728928728345,
      "loss": 2.0635,
      "step": 140515
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.131335973739624,
      "learning_rate": 0.00019849343998329613,
      "loss": 2.0561,
      "step": 140516
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.0885367393493652,
      "learning_rate": 0.00019848959069818092,
      "loss": 2.2903,
      "step": 140517
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.1000910997390747,
      "learning_rate": 0.00019848574143193831,
      "loss": 2.149,
      "step": 140518
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.0151622295379639,
      "learning_rate": 0.00019848189218456914,
      "loss": 2.0333,
      "step": 140519
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.046862006187439,
      "learning_rate": 0.00019847804295607406,
      "loss": 2.0474,
      "step": 140520
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.0230170488357544,
      "learning_rate": 0.00019847419374645386,
      "loss": 2.0612,
      "step": 140521
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.4256619215011597,
      "learning_rate": 0.0001984703445557092,
      "loss": 2.4061,
      "step": 140522
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.0217549800872803,
      "learning_rate": 0.0001984664953838409,
      "loss": 2.0657,
      "step": 140523
    },
    {
      "epoch": 1.83,
      "grad_norm": 0.9647216200828552,
      "learning_rate": 0.00019846264623084948,
      "loss": 2.1855,
      "step": 140524
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.0379900932312012,
      "learning_rate": 0.0001984587970967359,
      "loss": 1.9569,
      "step": 140525
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.0389125347137451,
      "learning_rate": 0.00019845494798150063,
      "loss": 1.8431,
      "step": 140526
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.058372139930725,
      "learning_rate": 0.00019845109888514453,
      "loss": 2.1921,
      "step": 140527
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.0865455865859985,
      "learning_rate": 0.00019844724980766832,
      "loss": 2.0556,
      "step": 140528
    },
    {
      "epoch": 1.83,
      "grad_norm": 0.9925947189331055,
      "learning_rate": 0.00019844340074907262,
      "loss": 2.047,
      "step": 140529
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.186955213546753,
      "learning_rate": 0.0001984395517093583,
      "loss": 2.1116,
      "step": 140530
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.1683553457260132,
      "learning_rate": 0.00019843570268852592,
      "loss": 2.1234,
      "step": 140531
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.1170296669006348,
      "learning_rate": 0.00019843185368657634,
      "loss": 2.1584,
      "step": 140532
    },
    {
      "epoch": 1.83,
      "grad_norm": 0.964293897151947,
      "learning_rate": 0.0001984280047035101,
      "loss": 2.045,
      "step": 140533
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.0292176008224487,
      "learning_rate": 0.00019842415573932803,
      "loss": 1.9862,
      "step": 140534
    },
    {
      "epoch": 1.83,
      "grad_norm": 0.9323415756225586,
      "learning_rate": 0.00019842030679403081,
      "loss": 2.1239,
      "step": 140535
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.0498605966567993,
      "learning_rate": 0.00019841645786761926,
      "loss": 2.2011,
      "step": 140536
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.0104291439056396,
      "learning_rate": 0.00019841260896009394,
      "loss": 1.893,
      "step": 140537
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.125213384628296,
      "learning_rate": 0.00019840876007145563,
      "loss": 2.4124,
      "step": 140538
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.094355821609497,
      "learning_rate": 0.00019840491120170513,
      "loss": 2.1699,
      "step": 140539
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.006059169769287,
      "learning_rate": 0.000198401062350843,
      "loss": 2.1458,
      "step": 140540
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.1563652753829956,
      "learning_rate": 0.00019839721351887002,
      "loss": 2.2195,
      "step": 140541
    },
    {
      "epoch": 1.83,
      "grad_norm": 0.9845259785652161,
      "learning_rate": 0.00019839336470578697,
      "loss": 2.1841,
      "step": 140542
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.1224207878112793,
      "learning_rate": 0.0001983895159115945,
      "loss": 2.3023,
      "step": 140543
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.0999194383621216,
      "learning_rate": 0.0001983856671362933,
      "loss": 2.3342,
      "step": 140544
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.1159931421279907,
      "learning_rate": 0.00019838181837988412,
      "loss": 2.0639,
      "step": 140545
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.0317728519439697,
      "learning_rate": 0.0001983779696423677,
      "loss": 1.92,
      "step": 140546
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.0351322889328003,
      "learning_rate": 0.00019837412092374473,
      "loss": 2.2192,
      "step": 140547
    },
    {
      "epoch": 1.83,
      "grad_norm": 0.988196074962616,
      "learning_rate": 0.000198370272224016,
      "loss": 2.064,
      "step": 140548
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.050236463546753,
      "learning_rate": 0.00019836642354318203,
      "loss": 2.2544,
      "step": 140549
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.0277249813079834,
      "learning_rate": 0.0001983625748812437,
      "loss": 2.0353,
      "step": 140550
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.0590077638626099,
      "learning_rate": 0.00019835872623820167,
      "loss": 2.1034,
      "step": 140551
    },
    {
      "epoch": 1.83,
      "grad_norm": 0.9698333740234375,
      "learning_rate": 0.0001983548776140567,
      "loss": 1.8499,
      "step": 140552
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.0741190910339355,
      "learning_rate": 0.00019835102900880947,
      "loss": 1.9159,
      "step": 140553
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.0828015804290771,
      "learning_rate": 0.0001983471804224608,
      "loss": 2.1668,
      "step": 140554
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.0957872867584229,
      "learning_rate": 0.0001983433318550112,
      "loss": 1.9473,
      "step": 140555
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.121773600578308,
      "learning_rate": 0.0001983394833064615,
      "loss": 2.1135,
      "step": 140556
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.0522388219833374,
      "learning_rate": 0.0001983356347768124,
      "loss": 1.9558,
      "step": 140557
    },
    {
      "epoch": 1.83,
      "grad_norm": 0.9150916337966919,
      "learning_rate": 0.00019833178626606465,
      "loss": 1.9441,
      "step": 140558
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.1190664768218994,
      "learning_rate": 0.00019832793777421892,
      "loss": 2.1226,
      "step": 140559
    },
    {
      "epoch": 1.83,
      "grad_norm": 0.9593765139579773,
      "learning_rate": 0.00019832408930127606,
      "loss": 2.0819,
      "step": 140560
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.062806248664856,
      "learning_rate": 0.00019832024084723654,
      "loss": 1.9394,
      "step": 140561
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.183217167854309,
      "learning_rate": 0.00019831639241210124,
      "loss": 2.2346,
      "step": 140562
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.03323495388031,
      "learning_rate": 0.00019831254399587083,
      "loss": 2.2763,
      "step": 140563
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.0542441606521606,
      "learning_rate": 0.00019830869559854603,
      "loss": 1.906,
      "step": 140564
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.119639277458191,
      "learning_rate": 0.00019830484722012758,
      "loss": 2.1748,
      "step": 140565
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.075252890586853,
      "learning_rate": 0.00019830099886061628,
      "loss": 2.0135,
      "step": 140566
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.1146636009216309,
      "learning_rate": 0.00019829715052001263,
      "loss": 2.1851,
      "step": 140567
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.0234687328338623,
      "learning_rate": 0.00019829330219831744,
      "loss": 2.0101,
      "step": 140568
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.0554884672164917,
      "learning_rate": 0.00019828945389553147,
      "loss": 2.1608,
      "step": 140569
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.2171188592910767,
      "learning_rate": 0.00019828560561165542,
      "loss": 1.8746,
      "step": 140570
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.0025733709335327,
      "learning_rate": 0.00019828175734668997,
      "loss": 1.8361,
      "step": 140571
    },
    {
      "epoch": 1.83,
      "grad_norm": 0.961043119430542,
      "learning_rate": 0.000198277909100636,
      "loss": 1.9956,
      "step": 140572
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.1704257726669312,
      "learning_rate": 0.00019827406087349393,
      "loss": 2.208,
      "step": 140573
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.1661579608917236,
      "learning_rate": 0.0001982702126652647,
      "loss": 2.3002,
      "step": 140574
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.1243274211883545,
      "learning_rate": 0.00019826636447594894,
      "loss": 1.9944,
      "step": 140575
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.0103906393051147,
      "learning_rate": 0.00019826251630554733,
      "loss": 2.1659,
      "step": 140576
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.2088947296142578,
      "learning_rate": 0.00019825866815406066,
      "loss": 2.2647,
      "step": 140577
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.155648112297058,
      "learning_rate": 0.00019825482002148976,
      "loss": 2.2222,
      "step": 140578
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.0592024326324463,
      "learning_rate": 0.00019825097190783508,
      "loss": 2.0101,
      "step": 140579
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.071297287940979,
      "learning_rate": 0.0001982471238130975,
      "loss": 1.989,
      "step": 140580
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.2074735164642334,
      "learning_rate": 0.00019824327573727762,
      "loss": 2.2422,
      "step": 140581
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.0351628065109253,
      "learning_rate": 0.0001982394276803763,
      "loss": 2.0534,
      "step": 140582
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.0790128707885742,
      "learning_rate": 0.00019823557964239417,
      "loss": 2.0888,
      "step": 140583
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.0742565393447876,
      "learning_rate": 0.00019823173162333211,
      "loss": 2.2068,
      "step": 140584
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.01059889793396,
      "learning_rate": 0.00019822788362319053,
      "loss": 2.1418,
      "step": 140585
    },
    {
      "epoch": 1.83,
      "grad_norm": 0.9369774460792542,
      "learning_rate": 0.00019822403564197032,
      "loss": 2.0831,
      "step": 140586
    },
    {
      "epoch": 1.83,
      "grad_norm": 0.9705827832221985,
      "learning_rate": 0.00019822018767967224,
      "loss": 2.2197,
      "step": 140587
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.1502095460891724,
      "learning_rate": 0.00019821633973629688,
      "loss": 2.0699,
      "step": 140588
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.0854363441467285,
      "learning_rate": 0.00019821249181184507,
      "loss": 2.0944,
      "step": 140589
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.345592975616455,
      "learning_rate": 0.0001982086439063176,
      "loss": 2.2813,
      "step": 140590
    },
    {
      "epoch": 1.83,
      "grad_norm": 0.9662586450576782,
      "learning_rate": 0.0001982047960197149,
      "loss": 2.0623,
      "step": 140591
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.0494846105575562,
      "learning_rate": 0.0001982009481520379,
      "loss": 1.9456,
      "step": 140592
    },
    {
      "epoch": 1.83,
      "grad_norm": 0.9882689118385315,
      "learning_rate": 0.0001981971003032872,
      "loss": 2.1778,
      "step": 140593
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.0168378353118896,
      "learning_rate": 0.00019819325247346362,
      "loss": 2.2758,
      "step": 140594
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.06931471824646,
      "learning_rate": 0.00019818940466256784,
      "loss": 2.1483,
      "step": 140595
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.0043253898620605,
      "learning_rate": 0.00019818555687060055,
      "loss": 2.0248,
      "step": 140596
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.1601614952087402,
      "learning_rate": 0.00019818170909756262,
      "loss": 2.0926,
      "step": 140597
    },
    {
      "epoch": 1.83,
      "grad_norm": 0.9481917023658752,
      "learning_rate": 0.00019817786134345453,
      "loss": 2.0074,
      "step": 140598
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.0905579328536987,
      "learning_rate": 0.00019817401360827707,
      "loss": 2.2193,
      "step": 140599
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.0464564561843872,
      "learning_rate": 0.000198170165892031,
      "loss": 1.9441,
      "step": 140600
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.0350513458251953,
      "learning_rate": 0.00019816631819471702,
      "loss": 1.8084,
      "step": 140601
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.0343676805496216,
      "learning_rate": 0.00019816247051633585,
      "loss": 2.1663,
      "step": 140602
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.153594970703125,
      "learning_rate": 0.00019815862285688825,
      "loss": 2.0309,
      "step": 140603
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.1241062879562378,
      "learning_rate": 0.00019815477521637487,
      "loss": 2.1074,
      "step": 140604
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.0804715156555176,
      "learning_rate": 0.00019815092759479638,
      "loss": 2.1901,
      "step": 140605
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.0977585315704346,
      "learning_rate": 0.00019814707999215355,
      "loss": 1.9796,
      "step": 140606
    },
    {
      "epoch": 1.83,
      "grad_norm": 0.9812350869178772,
      "learning_rate": 0.0001981432324084471,
      "loss": 2.1653,
      "step": 140607
    },
    {
      "epoch": 1.83,
      "grad_norm": 0.9744106531143188,
      "learning_rate": 0.0001981393848436778,
      "loss": 2.1995,
      "step": 140608
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.0649867057800293,
      "learning_rate": 0.0001981355372978464,
      "loss": 1.9711,
      "step": 140609
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.1344205141067505,
      "learning_rate": 0.00019813168977095337,
      "loss": 2.1603,
      "step": 140610
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.0523499250411987,
      "learning_rate": 0.00019812784226299962,
      "loss": 2.123,
      "step": 140611
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.2613959312438965,
      "learning_rate": 0.00019812399477398581,
      "loss": 2.2562,
      "step": 140612
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.0680954456329346,
      "learning_rate": 0.00019812014730391267,
      "loss": 2.1505,
      "step": 140613
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.1645441055297852,
      "learning_rate": 0.00019811629985278092,
      "loss": 2.1294,
      "step": 140614
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.1098743677139282,
      "learning_rate": 0.00019811245242059135,
      "loss": 2.004,
      "step": 140615
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.0575147867202759,
      "learning_rate": 0.00019810860500734455,
      "loss": 2.3243,
      "step": 140616
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.3230668306350708,
      "learning_rate": 0.00019810475761304134,
      "loss": 2.2165,
      "step": 140617
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.1356838941574097,
      "learning_rate": 0.00019810091023768228,
      "loss": 2.2077,
      "step": 140618
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.0400174856185913,
      "learning_rate": 0.0001980970628812682,
      "loss": 2.0257,
      "step": 140619
    },
    {
      "epoch": 1.83,
      "grad_norm": 0.9525293707847595,
      "learning_rate": 0.0001980932155437998,
      "loss": 2.33,
      "step": 140620
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.047433853149414,
      "learning_rate": 0.00019808936822527789,
      "loss": 1.9724,
      "step": 140621
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.0170725584030151,
      "learning_rate": 0.000198085520925703,
      "loss": 2.0853,
      "step": 140622
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.0512288808822632,
      "learning_rate": 0.00019808167364507594,
      "loss": 2.2857,
      "step": 140623
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.1038753986358643,
      "learning_rate": 0.00019807782638339743,
      "loss": 1.9952,
      "step": 140624
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.107365608215332,
      "learning_rate": 0.00019807397914066817,
      "loss": 2.1629,
      "step": 140625
    },
    {
      "epoch": 1.83,
      "grad_norm": 0.9832268357276917,
      "learning_rate": 0.0001980701319168889,
      "loss": 2.0288,
      "step": 140626
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.16273832321167,
      "learning_rate": 0.00019806628471206035,
      "loss": 2.0188,
      "step": 140627
    },
    {
      "epoch": 1.83,
      "grad_norm": 0.9850953817367554,
      "learning_rate": 0.0001980624375261831,
      "loss": 2.122,
      "step": 140628
    },
    {
      "epoch": 1.83,
      "grad_norm": 0.9899588823318481,
      "learning_rate": 0.00019805859035925804,
      "loss": 1.876,
      "step": 140629
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.1361594200134277,
      "learning_rate": 0.0001980547432112858,
      "loss": 2.0956,
      "step": 140630
    },
    {
      "epoch": 1.83,
      "grad_norm": 0.9932699799537659,
      "learning_rate": 0.00019805089608226713,
      "loss": 1.9858,
      "step": 140631
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.0580804347991943,
      "learning_rate": 0.0001980470489722027,
      "loss": 2.1601,
      "step": 140632
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.056820034980774,
      "learning_rate": 0.0001980432018810933,
      "loss": 2.0521,
      "step": 140633
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.133476734161377,
      "learning_rate": 0.0001980393548089395,
      "loss": 2.0175,
      "step": 140634
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.0784860849380493,
      "learning_rate": 0.00019803550775574215,
      "loss": 2.0719,
      "step": 140635
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.2256839275360107,
      "learning_rate": 0.0001980316607215019,
      "loss": 2.0064,
      "step": 140636
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.1098512411117554,
      "learning_rate": 0.00019802781370621946,
      "loss": 1.8405,
      "step": 140637
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.1117933988571167,
      "learning_rate": 0.0001980239667098957,
      "loss": 2.1189,
      "step": 140638
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.0384711027145386,
      "learning_rate": 0.00019802011973253119,
      "loss": 2.0315,
      "step": 140639
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.0770845413208008,
      "learning_rate": 0.00019801627277412658,
      "loss": 2.2267,
      "step": 140640
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.2492268085479736,
      "learning_rate": 0.0001980124258346827,
      "loss": 1.8306,
      "step": 140641
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.1633574962615967,
      "learning_rate": 0.0001980085789142002,
      "loss": 1.9817,
      "step": 140642
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.1314971446990967,
      "learning_rate": 0.00019800473201267985,
      "loss": 2.2115,
      "step": 140643
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.3418123722076416,
      "learning_rate": 0.00019800088513012234,
      "loss": 1.7839,
      "step": 140644
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.223493218421936,
      "learning_rate": 0.00019799703826652856,
      "loss": 1.7889,
      "step": 140645
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.0495306253433228,
      "learning_rate": 0.00019799319142189885,
      "loss": 2.1422,
      "step": 140646
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.1737220287322998,
      "learning_rate": 0.00019798934459623417,
      "loss": 2.2356,
      "step": 140647
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.3058515787124634,
      "learning_rate": 0.00019798549778953523,
      "loss": 1.9821,
      "step": 140648
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.0849498510360718,
      "learning_rate": 0.00019798165100180266,
      "loss": 2.0462,
      "step": 140649
    },
    {
      "epoch": 1.83,
      "grad_norm": 0.9977462291717529,
      "learning_rate": 0.00019797780423303724,
      "loss": 2.117,
      "step": 140650
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.2484047412872314,
      "learning_rate": 0.0001979739574832398,
      "loss": 2.0858,
      "step": 140651
    },
    {
      "epoch": 1.83,
      "grad_norm": 0.9836833477020264,
      "learning_rate": 0.00019797011075241083,
      "loss": 2.0675,
      "step": 140652
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.015087366104126,
      "learning_rate": 0.00019796626404055114,
      "loss": 2.1842,
      "step": 140653
    },
    {
      "epoch": 1.83,
      "grad_norm": 0.9881184101104736,
      "learning_rate": 0.0001979624173476614,
      "loss": 2.0621,
      "step": 140654
    },
    {
      "epoch": 1.83,
      "grad_norm": 0.9898330569267273,
      "learning_rate": 0.00019795857067374242,
      "loss": 1.7932,
      "step": 140655
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.055548071861267,
      "learning_rate": 0.00019795472401879484,
      "loss": 1.8591,
      "step": 140656
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.0852848291397095,
      "learning_rate": 0.00019795087738281951,
      "loss": 2.1501,
      "step": 140657
    },
    {
      "epoch": 1.83,
      "grad_norm": 0.9238020777702332,
      "learning_rate": 0.00019794703076581697,
      "loss": 1.9766,
      "step": 140658
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.1356028318405151,
      "learning_rate": 0.00019794318416778794,
      "loss": 2.3223,
      "step": 140659
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.079285740852356,
      "learning_rate": 0.00019793933758873326,
      "loss": 2.0794,
      "step": 140660
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.006224513053894,
      "learning_rate": 0.00019793549102865352,
      "loss": 1.9075,
      "step": 140661
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.0460858345031738,
      "learning_rate": 0.00019793164448754958,
      "loss": 1.9457,
      "step": 140662
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.1296517848968506,
      "learning_rate": 0.00019792779796542201,
      "loss": 2.2093,
      "step": 140663
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.163081169128418,
      "learning_rate": 0.00019792395146227173,
      "loss": 2.027,
      "step": 140664
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.1392593383789062,
      "learning_rate": 0.0001979201049780992,
      "loss": 2.1163,
      "step": 140665
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.1363612413406372,
      "learning_rate": 0.00019791625851290522,
      "loss": 2.0542,
      "step": 140666
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.0194047689437866,
      "learning_rate": 0.00019791241206669054,
      "loss": 2.3349,
      "step": 140667
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.1959878206253052,
      "learning_rate": 0.0001979085656394559,
      "loss": 1.7883,
      "step": 140668
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.0624184608459473,
      "learning_rate": 0.000197904719231202,
      "loss": 2.2143,
      "step": 140669
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.0622575283050537,
      "learning_rate": 0.0001979008728419296,
      "loss": 2.0328,
      "step": 140670
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.089499592781067,
      "learning_rate": 0.0001978970264716393,
      "loss": 1.9769,
      "step": 140671
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.2909266948699951,
      "learning_rate": 0.0001978931801203318,
      "loss": 2.0094,
      "step": 140672
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.0507006645202637,
      "learning_rate": 0.00019788933378800792,
      "loss": 2.1289,
      "step": 140673
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.0130563974380493,
      "learning_rate": 0.00019788548747466837,
      "loss": 2.0608,
      "step": 140674
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.1052696704864502,
      "learning_rate": 0.0001978816411803138,
      "loss": 1.9767,
      "step": 140675
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.0156631469726562,
      "learning_rate": 0.0001978777949049451,
      "loss": 2.0264,
      "step": 140676
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.1627612113952637,
      "learning_rate": 0.0001978739486485627,
      "loss": 2.3562,
      "step": 140677
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.000091552734375,
      "learning_rate": 0.00019787010241116747,
      "loss": 2.0751,
      "step": 140678
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.1012356281280518,
      "learning_rate": 0.0001978662561927601,
      "loss": 1.8808,
      "step": 140679
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.0825904607772827,
      "learning_rate": 0.00019786240999334134,
      "loss": 1.9783,
      "step": 140680
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.0227831602096558,
      "learning_rate": 0.0001978585638129119,
      "loss": 2.188,
      "step": 140681
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.0523399114608765,
      "learning_rate": 0.00019785471765147258,
      "loss": 1.805,
      "step": 140682
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.1561145782470703,
      "learning_rate": 0.00019785087150902387,
      "loss": 2.0162,
      "step": 140683
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.0596102476119995,
      "learning_rate": 0.00019784702538556663,
      "loss": 2.2352,
      "step": 140684
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.3716627359390259,
      "learning_rate": 0.00019784317928110156,
      "loss": 2.1371,
      "step": 140685
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.1949726343154907,
      "learning_rate": 0.00019783933319562935,
      "loss": 2.2726,
      "step": 140686
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.1470646858215332,
      "learning_rate": 0.00019783548712915076,
      "loss": 2.2683,
      "step": 140687
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.2822774648666382,
      "learning_rate": 0.0001978316410816666,
      "loss": 2.1865,
      "step": 140688
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.0679174661636353,
      "learning_rate": 0.00019782779505317732,
      "loss": 1.8468,
      "step": 140689
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.038546085357666,
      "learning_rate": 0.00019782394904368378,
      "loss": 2.0613,
      "step": 140690
    },
    {
      "epoch": 1.83,
      "grad_norm": 0.9868835210800171,
      "learning_rate": 0.00019782010305318673,
      "loss": 2.0104,
      "step": 140691
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.1162114143371582,
      "learning_rate": 0.00019781625708168682,
      "loss": 2.1454,
      "step": 140692
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.1328821182250977,
      "learning_rate": 0.00019781241112918482,
      "loss": 2.1861,
      "step": 140693
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.0599534511566162,
      "learning_rate": 0.00019780856519568154,
      "loss": 2.0582,
      "step": 140694
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.0418108701705933,
      "learning_rate": 0.00019780471928117742,
      "loss": 2.0599,
      "step": 140695
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.1327273845672607,
      "learning_rate": 0.0001978008733856734,
      "loss": 1.9791,
      "step": 140696
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.1854695081710815,
      "learning_rate": 0.0001977970275091701,
      "loss": 2.3914,
      "step": 140697
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.041959285736084,
      "learning_rate": 0.00019779318165166824,
      "loss": 1.966,
      "step": 140698
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.0088001489639282,
      "learning_rate": 0.00019778933581316856,
      "loss": 2.2884,
      "step": 140699
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.193718433380127,
      "learning_rate": 0.00019778548999367185,
      "loss": 2.1553,
      "step": 140700
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.051145315170288,
      "learning_rate": 0.00019778164419317874,
      "loss": 1.9348,
      "step": 140701
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.1067721843719482,
      "learning_rate": 0.00019777779841168987,
      "loss": 2.1778,
      "step": 140702
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.129772424697876,
      "learning_rate": 0.00019777395264920607,
      "loss": 2.2878,
      "step": 140703
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.074375033378601,
      "learning_rate": 0.00019777010690572798,
      "loss": 2.0327,
      "step": 140704
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.1346875429153442,
      "learning_rate": 0.00019776626118125638,
      "loss": 2.089,
      "step": 140705
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.1039549112319946,
      "learning_rate": 0.00019776241547579204,
      "loss": 1.9386,
      "step": 140706
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.1498169898986816,
      "learning_rate": 0.00019775856978933553,
      "loss": 2.1723,
      "step": 140707
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.2632077932357788,
      "learning_rate": 0.00019775472412188766,
      "loss": 2.1152,
      "step": 140708
    },
    {
      "epoch": 1.83,
      "grad_norm": 0.9767334461212158,
      "learning_rate": 0.0001977508784734491,
      "loss": 2.1735,
      "step": 140709
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.0619499683380127,
      "learning_rate": 0.00019774703284402057,
      "loss": 1.8799,
      "step": 140710
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.0739052295684814,
      "learning_rate": 0.00019774318723360276,
      "loss": 2.0747,
      "step": 140711
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.1438394784927368,
      "learning_rate": 0.00019773934164219652,
      "loss": 2.1656,
      "step": 140712
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.2451339960098267,
      "learning_rate": 0.00019773549606980239,
      "loss": 2.1402,
      "step": 140713
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.1733912229537964,
      "learning_rate": 0.0001977316505164212,
      "loss": 2.1923,
      "step": 140714
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.0481339693069458,
      "learning_rate": 0.0001977278049820536,
      "loss": 1.8793,
      "step": 140715
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.1214300394058228,
      "learning_rate": 0.00019772395946670036,
      "loss": 2.1409,
      "step": 140716
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.0706161260604858,
      "learning_rate": 0.00019772011397036214,
      "loss": 2.2189,
      "step": 140717
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.0880670547485352,
      "learning_rate": 0.00019771626849303977,
      "loss": 2.0201,
      "step": 140718
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.1607825756072998,
      "learning_rate": 0.00019771242303473377,
      "loss": 2.0135,
      "step": 140719
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.1960697174072266,
      "learning_rate": 0.000197708577595445,
      "loss": 2.3056,
      "step": 140720
    },
    {
      "epoch": 1.83,
      "grad_norm": 0.9998931884765625,
      "learning_rate": 0.00019770473217517408,
      "loss": 2.005,
      "step": 140721
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.0939035415649414,
      "learning_rate": 0.00019770088677392183,
      "loss": 2.0084,
      "step": 140722
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.134242296218872,
      "learning_rate": 0.00019769704139168892,
      "loss": 2.0239,
      "step": 140723
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.0611224174499512,
      "learning_rate": 0.00019769319602847606,
      "loss": 2.0583,
      "step": 140724
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.2792739868164062,
      "learning_rate": 0.000197689350684284,
      "loss": 2.2121,
      "step": 140725
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.0937719345092773,
      "learning_rate": 0.00019768550535911338,
      "loss": 1.9733,
      "step": 140726
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.0139464139938354,
      "learning_rate": 0.00019768166005296496,
      "loss": 2.1995,
      "step": 140727
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.0350345373153687,
      "learning_rate": 0.0001976778147658394,
      "loss": 2.0649,
      "step": 140728
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.2537349462509155,
      "learning_rate": 0.0001976739694977375,
      "loss": 2.2606,
      "step": 140729
    },
    {
      "epoch": 1.83,
      "grad_norm": 0.9889194965362549,
      "learning_rate": 0.00019767012424866002,
      "loss": 2.0158,
      "step": 140730
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.096520185470581,
      "learning_rate": 0.00019766627901860755,
      "loss": 2.175,
      "step": 140731
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.1813827753067017,
      "learning_rate": 0.0001976624338075808,
      "loss": 2.1496,
      "step": 140732
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.2458895444869995,
      "learning_rate": 0.0001976585886155806,
      "loss": 2.0343,
      "step": 140733
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.0646253824234009,
      "learning_rate": 0.00019765474344260749,
      "loss": 2.0559,
      "step": 140734
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.0671762228012085,
      "learning_rate": 0.00019765089828866237,
      "loss": 2.1599,
      "step": 140735
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.087123155593872,
      "learning_rate": 0.00019764705315374588,
      "loss": 2.1707,
      "step": 140736
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.0281656980514526,
      "learning_rate": 0.00019764320803785884,
      "loss": 1.9305,
      "step": 140737
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.0411032438278198,
      "learning_rate": 0.00019763936294100174,
      "loss": 2.2968,
      "step": 140738
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.154252052307129,
      "learning_rate": 0.0001976355178631754,
      "loss": 2.1903,
      "step": 140739
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.0643936395645142,
      "learning_rate": 0.00019763167280438057,
      "loss": 2.0656,
      "step": 140740
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.1720541715621948,
      "learning_rate": 0.00019762782776461797,
      "loss": 2.227,
      "step": 140741
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.1874301433563232,
      "learning_rate": 0.0001976239827438882,
      "loss": 1.9423,
      "step": 140742
    },
    {
      "epoch": 1.83,
      "grad_norm": 0.9952124357223511,
      "learning_rate": 0.0001976201377421923,
      "loss": 2.0064,
      "step": 140743
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.025313138961792,
      "learning_rate": 0.00019761629275953055,
      "loss": 2.0178,
      "step": 140744
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.0302722454071045,
      "learning_rate": 0.0001976124477959039,
      "loss": 2.0141,
      "step": 140745
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.17569100856781,
      "learning_rate": 0.000197608602851313,
      "loss": 2.2922,
      "step": 140746
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.101365089416504,
      "learning_rate": 0.00019760475792575857,
      "loss": 1.9865,
      "step": 140747
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.083699107170105,
      "learning_rate": 0.0001976009130192414,
      "loss": 2.0406,
      "step": 140748
    },
    {
      "epoch": 1.83,
      "grad_norm": 0.9278051853179932,
      "learning_rate": 0.0001975970681317623,
      "loss": 2.0846,
      "step": 140749
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.0918515920639038,
      "learning_rate": 0.00019759322326332165,
      "loss": 2.2075,
      "step": 140750
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.2324053049087524,
      "learning_rate": 0.0001975893784139204,
      "loss": 2.1546,
      "step": 140751
    },
    {
      "epoch": 1.83,
      "grad_norm": 0.9889445900917053,
      "learning_rate": 0.00019758553358355919,
      "loss": 2.0819,
      "step": 140752
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.141653060913086,
      "learning_rate": 0.00019758168877223877,
      "loss": 2.4209,
      "step": 140753
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.0347850322723389,
      "learning_rate": 0.00019757784397995984,
      "loss": 2.265,
      "step": 140754
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.0843985080718994,
      "learning_rate": 0.00019757399920672326,
      "loss": 2.0061,
      "step": 140755
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.2916096448898315,
      "learning_rate": 0.00019757015445252948,
      "loss": 2.146,
      "step": 140756
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.108627438545227,
      "learning_rate": 0.00019756630971737933,
      "loss": 2.1539,
      "step": 140757
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.0671151876449585,
      "learning_rate": 0.00019756246500127355,
      "loss": 2.187,
      "step": 140758
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.1986578702926636,
      "learning_rate": 0.00019755862030421283,
      "loss": 2.0619,
      "step": 140759
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.041123628616333,
      "learning_rate": 0.0001975547756261979,
      "loss": 2.0731,
      "step": 140760
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.21193528175354,
      "learning_rate": 0.00019755093096722958,
      "loss": 2.0912,
      "step": 140761
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.0414081811904907,
      "learning_rate": 0.00019754708632730838,
      "loss": 2.1692,
      "step": 140762
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.3654738664627075,
      "learning_rate": 0.0001975432417064351,
      "loss": 2.1632,
      "step": 140763
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.098185658454895,
      "learning_rate": 0.00019753939710461047,
      "loss": 2.1899,
      "step": 140764
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.1752208471298218,
      "learning_rate": 0.00019753555252183518,
      "loss": 2.2218,
      "step": 140765
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.0538933277130127,
      "learning_rate": 0.00019753170795811,
      "loss": 2.132,
      "step": 140766
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.3024730682373047,
      "learning_rate": 0.0001975278634134357,
      "loss": 1.943,
      "step": 140767
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.2786407470703125,
      "learning_rate": 0.0001975240188878128,
      "loss": 1.9863,
      "step": 140768
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.0971709489822388,
      "learning_rate": 0.00019752017438124213,
      "loss": 1.9625,
      "step": 140769
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.2414963245391846,
      "learning_rate": 0.00019751632989372436,
      "loss": 2.3275,
      "step": 140770
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.112164855003357,
      "learning_rate": 0.00019751248542526024,
      "loss": 1.9438,
      "step": 140771
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.0688010454177856,
      "learning_rate": 0.00019750864097585054,
      "loss": 2.0744,
      "step": 140772
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.004603624343872,
      "learning_rate": 0.00019750479654549597,
      "loss": 2.2403,
      "step": 140773
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.1260873079299927,
      "learning_rate": 0.0001975009521341971,
      "loss": 2.0705,
      "step": 140774
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.0424937009811401,
      "learning_rate": 0.00019749710774195475,
      "loss": 2.4151,
      "step": 140775
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.018425464630127,
      "learning_rate": 0.00019749326336876964,
      "loss": 1.9324,
      "step": 140776
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.0961371660232544,
      "learning_rate": 0.0001974894190146424,
      "loss": 1.951,
      "step": 140777
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.1328141689300537,
      "learning_rate": 0.00019748557467957385,
      "loss": 2.1524,
      "step": 140778
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.0044801235198975,
      "learning_rate": 0.0001974817303635648,
      "loss": 2.4561,
      "step": 140779
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.0761538743972778,
      "learning_rate": 0.0001974778860666157,
      "loss": 2.1507,
      "step": 140780
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.0192615985870361,
      "learning_rate": 0.0001974740417887274,
      "loss": 1.9369,
      "step": 140781
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.0993176698684692,
      "learning_rate": 0.00019747019752990058,
      "loss": 1.9757,
      "step": 140782
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.0514867305755615,
      "learning_rate": 0.00019746635329013603,
      "loss": 2.0506,
      "step": 140783
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.09905207157135,
      "learning_rate": 0.0001974625090694344,
      "loss": 2.0457,
      "step": 140784
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.0444056987762451,
      "learning_rate": 0.0001974586648677965,
      "loss": 2.0439,
      "step": 140785
    },
    {
      "epoch": 1.83,
      "grad_norm": 0.9445812106132507,
      "learning_rate": 0.00019745482068522297,
      "loss": 1.8798,
      "step": 140786
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.0987757444381714,
      "learning_rate": 0.00019745097652171443,
      "loss": 2.0334,
      "step": 140787
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.0983630418777466,
      "learning_rate": 0.00019744713237727172,
      "loss": 2.0255,
      "step": 140788
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.1740468740463257,
      "learning_rate": 0.00019744328825189552,
      "loss": 2.1514,
      "step": 140789
    },
    {
      "epoch": 1.83,
      "grad_norm": 0.9484172463417053,
      "learning_rate": 0.0001974394441455865,
      "loss": 1.9246,
      "step": 140790
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.0057282447814941,
      "learning_rate": 0.0001974356000583455,
      "loss": 2.1235,
      "step": 140791
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.1447162628173828,
      "learning_rate": 0.0001974317559901732,
      "loss": 2.0413,
      "step": 140792
    },
    {
      "epoch": 1.83,
      "grad_norm": 0.9861695766448975,
      "learning_rate": 0.00019742791194107026,
      "loss": 2.1955,
      "step": 140793
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.0654927492141724,
      "learning_rate": 0.0001974240679110373,
      "loss": 2.0965,
      "step": 140794
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.065483808517456,
      "learning_rate": 0.00019742022390007523,
      "loss": 2.1959,
      "step": 140795
    },
    {
      "epoch": 1.83,
      "grad_norm": 0.9735463857650757,
      "learning_rate": 0.00019741637990818465,
      "loss": 2.1466,
      "step": 140796
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.0313833951950073,
      "learning_rate": 0.00019741253593536626,
      "loss": 2.2912,
      "step": 140797
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.0920348167419434,
      "learning_rate": 0.0001974086919816209,
      "loss": 2.269,
      "step": 140798
    },
    {
      "epoch": 1.83,
      "grad_norm": 0.9741538763046265,
      "learning_rate": 0.0001974048480469491,
      "loss": 1.9455,
      "step": 140799
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.0526020526885986,
      "learning_rate": 0.00019740100413135183,
      "loss": 1.9385,
      "step": 140800
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.0683741569519043,
      "learning_rate": 0.00019739716023482954,
      "loss": 2.0891,
      "step": 140801
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.0355031490325928,
      "learning_rate": 0.00019739331635738302,
      "loss": 2.089,
      "step": 140802
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.067318081855774,
      "learning_rate": 0.00019738947249901308,
      "loss": 2.1372,
      "step": 140803
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.2709487676620483,
      "learning_rate": 0.0001973856286597204,
      "loss": 1.9912,
      "step": 140804
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.0592503547668457,
      "learning_rate": 0.00019738178483950561,
      "loss": 2.291,
      "step": 140805
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.0920472145080566,
      "learning_rate": 0.0001973779410383695,
      "loss": 2.0755,
      "step": 140806
    },
    {
      "epoch": 1.83,
      "grad_norm": 0.9667856693267822,
      "learning_rate": 0.00019737409725631273,
      "loss": 2.0968,
      "step": 140807
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.0614914894104004,
      "learning_rate": 0.00019737025349333615,
      "loss": 2.1939,
      "step": 140808
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.2760556936264038,
      "learning_rate": 0.0001973664097494403,
      "loss": 2.0598,
      "step": 140809
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.1035257577896118,
      "learning_rate": 0.00019736256602462604,
      "loss": 1.999,
      "step": 140810
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.0746456384658813,
      "learning_rate": 0.00019735872231889398,
      "loss": 2.244,
      "step": 140811
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.2141143083572388,
      "learning_rate": 0.00019735487863224482,
      "loss": 2.1297,
      "step": 140812
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.1636260747909546,
      "learning_rate": 0.00019735103496467936,
      "loss": 2.343,
      "step": 140813
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.1785951852798462,
      "learning_rate": 0.00019734719131619827,
      "loss": 2.2934,
      "step": 140814
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.0028363466262817,
      "learning_rate": 0.00019734334768680233,
      "loss": 1.9115,
      "step": 140815
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.0268086194992065,
      "learning_rate": 0.0001973395040764922,
      "loss": 2.1381,
      "step": 140816
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.0682905912399292,
      "learning_rate": 0.00019733566048526854,
      "loss": 1.9718,
      "step": 140817
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.1808586120605469,
      "learning_rate": 0.0001973318169131321,
      "loss": 2.3783,
      "step": 140818
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.078904628753662,
      "learning_rate": 0.00019732797336008366,
      "loss": 1.9866,
      "step": 140819
    },
    {
      "epoch": 1.83,
      "grad_norm": 0.9626602530479431,
      "learning_rate": 0.00019732412982612383,
      "loss": 1.9594,
      "step": 140820
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.061924695968628,
      "learning_rate": 0.0001973202863112534,
      "loss": 2.1507,
      "step": 140821
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.090596079826355,
      "learning_rate": 0.00019731644281547323,
      "loss": 2.0471,
      "step": 140822
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.0207637548446655,
      "learning_rate": 0.0001973125993387837,
      "loss": 1.8993,
      "step": 140823
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.0206316709518433,
      "learning_rate": 0.00019730875588118573,
      "loss": 2.0153,
      "step": 140824
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.1374444961547852,
      "learning_rate": 0.00019730491244268,
      "loss": 2.0697,
      "step": 140825
    },
    {
      "epoch": 1.83,
      "grad_norm": 0.9835789799690247,
      "learning_rate": 0.0001973010690232672,
      "loss": 2.0272,
      "step": 140826
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.1020275354385376,
      "learning_rate": 0.00019729722562294806,
      "loss": 2.1278,
      "step": 140827
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.1027761697769165,
      "learning_rate": 0.00019729338224172346,
      "loss": 2.1598,
      "step": 140828
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.0594196319580078,
      "learning_rate": 0.00019728953887959385,
      "loss": 2.1311,
      "step": 140829
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.0856142044067383,
      "learning_rate": 0.00019728569553656002,
      "loss": 2.2583,
      "step": 140830
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.0133371353149414,
      "learning_rate": 0.0001972818522126227,
      "loss": 2.2199,
      "step": 140831
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.0014737844467163,
      "learning_rate": 0.00019727800890778266,
      "loss": 2.1527,
      "step": 140832
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.0213452577590942,
      "learning_rate": 0.00019727416562204057,
      "loss": 2.1263,
      "step": 140833
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.0784062147140503,
      "learning_rate": 0.00019727032235539727,
      "loss": 2.2347,
      "step": 140834
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.174147605895996,
      "learning_rate": 0.00019726647910785325,
      "loss": 2.2496,
      "step": 140835
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.0779283046722412,
      "learning_rate": 0.00019726263587940932,
      "loss": 1.8702,
      "step": 140836
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.0505129098892212,
      "learning_rate": 0.00019725879267006618,
      "loss": 1.9776,
      "step": 140837
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.1336733102798462,
      "learning_rate": 0.0001972549494798246,
      "loss": 1.9342,
      "step": 140838
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.2348366975784302,
      "learning_rate": 0.00019725110630868523,
      "loss": 1.9818,
      "step": 140839
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.1663947105407715,
      "learning_rate": 0.00019724726315664896,
      "loss": 2.0439,
      "step": 140840
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.0820051431655884,
      "learning_rate": 0.00019724342002371627,
      "loss": 2.0481,
      "step": 140841
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.1135423183441162,
      "learning_rate": 0.00019723957690988796,
      "loss": 2.1625,
      "step": 140842
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.2298589944839478,
      "learning_rate": 0.0001972357338151647,
      "loss": 2.1312,
      "step": 140843
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.1446607112884521,
      "learning_rate": 0.0001972318907395473,
      "loss": 2.1399,
      "step": 140844
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.085199236869812,
      "learning_rate": 0.00019722804768303642,
      "loss": 2.0787,
      "step": 140845
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.153498888015747,
      "learning_rate": 0.0001972242046456329,
      "loss": 2.0074,
      "step": 140846
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.1743168830871582,
      "learning_rate": 0.00019722036162733725,
      "loss": 2.0784,
      "step": 140847
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.1095682382583618,
      "learning_rate": 0.00019721651862815028,
      "loss": 2.1754,
      "step": 140848
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.081050992012024,
      "learning_rate": 0.0001972126756480727,
      "loss": 2.1895,
      "step": 140849
    },
    {
      "epoch": 1.83,
      "grad_norm": 0.9300405383110046,
      "learning_rate": 0.00019720883268710518,
      "loss": 2.1404,
      "step": 140850
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.2714555263519287,
      "learning_rate": 0.0001972049897452485,
      "loss": 2.2446,
      "step": 140851
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.0096358060836792,
      "learning_rate": 0.00019720114682250347,
      "loss": 1.9695,
      "step": 140852
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.096065878868103,
      "learning_rate": 0.00019719730391887057,
      "loss": 2.2112,
      "step": 140853
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.2184803485870361,
      "learning_rate": 0.00019719346103435064,
      "loss": 2.0095,
      "step": 140854
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.0435281991958618,
      "learning_rate": 0.00019718961816894435,
      "loss": 2.2323,
      "step": 140855
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.1169241666793823,
      "learning_rate": 0.0001971857753226525,
      "loss": 2.3486,
      "step": 140856
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.1992110013961792,
      "learning_rate": 0.0001971819324954757,
      "loss": 2.1501,
      "step": 140857
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.0818202495574951,
      "learning_rate": 0.00019717808968741478,
      "loss": 2.1066,
      "step": 140858
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.0570285320281982,
      "learning_rate": 0.00019717424689847047,
      "loss": 1.9628,
      "step": 140859
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.1641016006469727,
      "learning_rate": 0.0001971704041286433,
      "loss": 2.0532,
      "step": 140860
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.0127758979797363,
      "learning_rate": 0.0001971665613779341,
      "loss": 2.0336,
      "step": 140861
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.073703646659851,
      "learning_rate": 0.00019716271864634357,
      "loss": 2.1131,
      "step": 140862
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.2121167182922363,
      "learning_rate": 0.00019715887593387243,
      "loss": 2.2965,
      "step": 140863
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.1006311178207397,
      "learning_rate": 0.0001971550332405214,
      "loss": 2.3246,
      "step": 140864
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.2407737970352173,
      "learning_rate": 0.00019715119056629133,
      "loss": 2.0051,
      "step": 140865
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.2703572511672974,
      "learning_rate": 0.00019714734791118267,
      "loss": 2.0627,
      "step": 140866
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.0475257635116577,
      "learning_rate": 0.00019714350527519621,
      "loss": 2.1877,
      "step": 140867
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.0097376108169556,
      "learning_rate": 0.00019713966265833279,
      "loss": 1.9947,
      "step": 140868
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.0514190196990967,
      "learning_rate": 0.00019713582006059297,
      "loss": 2.275,
      "step": 140869
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.2675057649612427,
      "learning_rate": 0.00019713197748197762,
      "loss": 1.964,
      "step": 140870
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.0760577917099,
      "learning_rate": 0.00019712813492248746,
      "loss": 2.0189,
      "step": 140871
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.0601129531860352,
      "learning_rate": 0.000197124292382123,
      "loss": 2.14,
      "step": 140872
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.0399943590164185,
      "learning_rate": 0.00019712044986088507,
      "loss": 2.0916,
      "step": 140873
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.0788359642028809,
      "learning_rate": 0.0001971166073587744,
      "loss": 1.97,
      "step": 140874
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.217997670173645,
      "learning_rate": 0.0001971127648757917,
      "loss": 2.0635,
      "step": 140875
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.2582347393035889,
      "learning_rate": 0.00019710892241193767,
      "loss": 2.153,
      "step": 140876
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.0456995964050293,
      "learning_rate": 0.00019710507996721307,
      "loss": 1.8989,
      "step": 140877
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.1725720167160034,
      "learning_rate": 0.00019710123754161863,
      "loss": 2.1958,
      "step": 140878
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.2540194988250732,
      "learning_rate": 0.0001970973951351549,
      "loss": 2.3267,
      "step": 140879
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.1189202070236206,
      "learning_rate": 0.00019709355274782274,
      "loss": 2.1039,
      "step": 140880
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.106927514076233,
      "learning_rate": 0.00019708971037962284,
      "loss": 2.3088,
      "step": 140881
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.270555019378662,
      "learning_rate": 0.00019708586803055586,
      "loss": 2.121,
      "step": 140882
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.1324727535247803,
      "learning_rate": 0.00019708202570062268,
      "loss": 2.2665,
      "step": 140883
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.1901791095733643,
      "learning_rate": 0.00019707818338982378,
      "loss": 2.1724,
      "step": 140884
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.0353344678878784,
      "learning_rate": 0.00019707434109816007,
      "loss": 2.0203,
      "step": 140885
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.095030665397644,
      "learning_rate": 0.00019707049882563212,
      "loss": 2.2276,
      "step": 140886
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.2600374221801758,
      "learning_rate": 0.00019706665657224075,
      "loss": 2.1937,
      "step": 140887
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.2374706268310547,
      "learning_rate": 0.00019706281433798657,
      "loss": 2.0501,
      "step": 140888
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.2893180847167969,
      "learning_rate": 0.00019705897212287044,
      "loss": 1.9505,
      "step": 140889
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.022980809211731,
      "learning_rate": 0.00019705512992689294,
      "loss": 2.1439,
      "step": 140890
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.1705793142318726,
      "learning_rate": 0.00019705128775005483,
      "loss": 1.9523,
      "step": 140891
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.3222121000289917,
      "learning_rate": 0.00019704744559235685,
      "loss": 2.1501,
      "step": 140892
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.0024629831314087,
      "learning_rate": 0.00019704360345379967,
      "loss": 2.0865,
      "step": 140893
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.098017692565918,
      "learning_rate": 0.00019703976133438403,
      "loss": 2.0927,
      "step": 140894
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.0607044696807861,
      "learning_rate": 0.0001970359192341107,
      "loss": 2.1647,
      "step": 140895
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.1611369848251343,
      "learning_rate": 0.00019703207715298027,
      "loss": 2.0656,
      "step": 140896
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.0056383609771729,
      "learning_rate": 0.00019702823509099352,
      "loss": 1.8715,
      "step": 140897
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.0061993598937988,
      "learning_rate": 0.0001970243930481512,
      "loss": 2.0512,
      "step": 140898
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.1526429653167725,
      "learning_rate": 0.000197020551024454,
      "loss": 2.113,
      "step": 140899
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.055802822113037,
      "learning_rate": 0.00019701670901990258,
      "loss": 2.074,
      "step": 140900
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.0717025995254517,
      "learning_rate": 0.00019701286703449774,
      "loss": 2.0628,
      "step": 140901
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.1455475091934204,
      "learning_rate": 0.00019700902506824012,
      "loss": 2.0254,
      "step": 140902
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.05207359790802,
      "learning_rate": 0.00019700518312113042,
      "loss": 2.0969,
      "step": 140903
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.1777513027191162,
      "learning_rate": 0.00019700134119316945,
      "loss": 2.189,
      "step": 140904
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.2160738706588745,
      "learning_rate": 0.00019699749928435786,
      "loss": 2.1029,
      "step": 140905
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.078191876411438,
      "learning_rate": 0.0001969936573946964,
      "loss": 2.1482,
      "step": 140906
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.2865211963653564,
      "learning_rate": 0.00019698981552418584,
      "loss": 2.0701,
      "step": 140907
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.0875916481018066,
      "learning_rate": 0.00019698597367282674,
      "loss": 2.1303,
      "step": 140908
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.2134175300598145,
      "learning_rate": 0.00019698213184061986,
      "loss": 2.2532,
      "step": 140909
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.1486974954605103,
      "learning_rate": 0.00019697829002756595,
      "loss": 2.1619,
      "step": 140910
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.0671095848083496,
      "learning_rate": 0.00019697444823366575,
      "loss": 2.0272,
      "step": 140911
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.3462607860565186,
      "learning_rate": 0.00019697060645891995,
      "loss": 1.9227,
      "step": 140912
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.068618893623352,
      "learning_rate": 0.00019696676470332933,
      "loss": 1.9655,
      "step": 140913
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.0700509548187256,
      "learning_rate": 0.00019696292296689447,
      "loss": 2.0016,
      "step": 140914
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.1126980781555176,
      "learning_rate": 0.0001969590812496161,
      "loss": 2.0638,
      "step": 140915
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.10042142868042,
      "learning_rate": 0.00019695523955149497,
      "loss": 2.3343,
      "step": 140916
    },
    {
      "epoch": 1.83,
      "grad_norm": 0.9814364314079285,
      "learning_rate": 0.00019695139787253186,
      "loss": 2.1556,
      "step": 140917
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.0852795839309692,
      "learning_rate": 0.0001969475562127274,
      "loss": 2.0455,
      "step": 140918
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.2179362773895264,
      "learning_rate": 0.00019694371457208246,
      "loss": 2.0512,
      "step": 140919
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.2912405729293823,
      "learning_rate": 0.0001969398729505975,
      "loss": 2.07,
      "step": 140920
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.0191181898117065,
      "learning_rate": 0.00019693603134827336,
      "loss": 2.1455,
      "step": 140921
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.2592986822128296,
      "learning_rate": 0.00019693218976511077,
      "loss": 2.1295,
      "step": 140922
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.0901211500167847,
      "learning_rate": 0.00019692834820111044,
      "loss": 2.1524,
      "step": 140923
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.096256971359253,
      "learning_rate": 0.00019692450665627305,
      "loss": 2.1493,
      "step": 140924
    },
    {
      "epoch": 1.83,
      "grad_norm": 0.985940158367157,
      "learning_rate": 0.00019692066513059934,
      "loss": 2.1352,
      "step": 140925
    },
    {
      "epoch": 1.83,
      "grad_norm": 0.9774295091629028,
      "learning_rate": 0.00019691682362409018,
      "loss": 2.0318,
      "step": 140926
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.078123688697815,
      "learning_rate": 0.000196912982136746,
      "loss": 2.2809,
      "step": 140927
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.102948546409607,
      "learning_rate": 0.0001969091406685676,
      "loss": 1.9556,
      "step": 140928
    },
    {
      "epoch": 1.83,
      "grad_norm": 1.0830042362213135,
      "learning_rate": 0.00019690529921955578,
      "loss": 2.1005,
      "step": 140929
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.1060051918029785,
      "learning_rate": 0.00019690145778971123,
      "loss": 2.0249,
      "step": 140930
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.106155514717102,
      "learning_rate": 0.0001968976163790346,
      "loss": 1.9709,
      "step": 140931
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.1442344188690186,
      "learning_rate": 0.00019689377498752678,
      "loss": 2.1825,
      "step": 140932
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.120190978050232,
      "learning_rate": 0.00019688993361518823,
      "loss": 2.0788,
      "step": 140933
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.0463918447494507,
      "learning_rate": 0.00019688609226201979,
      "loss": 2.1414,
      "step": 140934
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.03446626663208,
      "learning_rate": 0.00019688225092802218,
      "loss": 1.8976,
      "step": 140935
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.0502275228500366,
      "learning_rate": 0.00019687840961319609,
      "loss": 2.0962,
      "step": 140936
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.1827077865600586,
      "learning_rate": 0.00019687456831754226,
      "loss": 2.2979,
      "step": 140937
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.1785731315612793,
      "learning_rate": 0.0001968707270410615,
      "loss": 1.9631,
      "step": 140938
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.0954034328460693,
      "learning_rate": 0.00019686688578375434,
      "loss": 2.1285,
      "step": 140939
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.0552698373794556,
      "learning_rate": 0.00019686304454562156,
      "loss": 1.9455,
      "step": 140940
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.1282271146774292,
      "learning_rate": 0.00019685920332666385,
      "loss": 2.2232,
      "step": 140941
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.067884087562561,
      "learning_rate": 0.000196855362126882,
      "loss": 2.1158,
      "step": 140942
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.0480008125305176,
      "learning_rate": 0.00019685152094627668,
      "loss": 2.191,
      "step": 140943
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.200666904449463,
      "learning_rate": 0.00019684767978484869,
      "loss": 2.024,
      "step": 140944
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.0352123975753784,
      "learning_rate": 0.00019684383864259858,
      "loss": 1.9811,
      "step": 140945
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.1154451370239258,
      "learning_rate": 0.00019683999751952713,
      "loss": 2.0064,
      "step": 140946
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.111330270767212,
      "learning_rate": 0.0001968361564156351,
      "loss": 2.0548,
      "step": 140947
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.1925221681594849,
      "learning_rate": 0.00019683231533092313,
      "loss": 2.0615,
      "step": 140948
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.0767039060592651,
      "learning_rate": 0.00019682847426539203,
      "loss": 2.1675,
      "step": 140949
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.0255112648010254,
      "learning_rate": 0.0001968246332190426,
      "loss": 2.2677,
      "step": 140950
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.1567460298538208,
      "learning_rate": 0.00019682079219187523,
      "loss": 2.1217,
      "step": 140951
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.0834633111953735,
      "learning_rate": 0.00019681695118389084,
      "loss": 2.0995,
      "step": 140952
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.1482402086257935,
      "learning_rate": 0.00019681311019509017,
      "loss": 2.0693,
      "step": 140953
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.030935525894165,
      "learning_rate": 0.00019680926922547383,
      "loss": 2.181,
      "step": 140954
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.3174333572387695,
      "learning_rate": 0.0001968054282750426,
      "loss": 2.1661,
      "step": 140955
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.010119915008545,
      "learning_rate": 0.00019680158734379735,
      "loss": 2.0579,
      "step": 140956
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.1074576377868652,
      "learning_rate": 0.0001967977464317385,
      "loss": 2.2901,
      "step": 140957
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.314697265625,
      "learning_rate": 0.0001967939055388669,
      "loss": 2.2288,
      "step": 140958
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.138659954071045,
      "learning_rate": 0.00019679006466518325,
      "loss": 2.0134,
      "step": 140959
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.1505823135375977,
      "learning_rate": 0.00019678622381068828,
      "loss": 2.2768,
      "step": 140960
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.1479524374008179,
      "learning_rate": 0.00019678238297538272,
      "loss": 1.9956,
      "step": 140961
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.1540385484695435,
      "learning_rate": 0.00019677854215926737,
      "loss": 2.2624,
      "step": 140962
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.2054188251495361,
      "learning_rate": 0.0001967747013623427,
      "loss": 2.1526,
      "step": 140963
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.0290815830230713,
      "learning_rate": 0.00019677086058460954,
      "loss": 1.804,
      "step": 140964
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.1292656660079956,
      "learning_rate": 0.00019676701982606867,
      "loss": 1.999,
      "step": 140965
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.1264837980270386,
      "learning_rate": 0.00019676317908672077,
      "loss": 2.3804,
      "step": 140966
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.1261574029922485,
      "learning_rate": 0.0001967593383665665,
      "loss": 2.2138,
      "step": 140967
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.1019319295883179,
      "learning_rate": 0.00019675549766560673,
      "loss": 2.019,
      "step": 140968
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.0256962776184082,
      "learning_rate": 0.00019675165698384192,
      "loss": 2.1295,
      "step": 140969
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.0751286745071411,
      "learning_rate": 0.00019674781632127305,
      "loss": 1.9446,
      "step": 140970
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.2096315622329712,
      "learning_rate": 0.00019674397567790065,
      "loss": 2.2068,
      "step": 140971
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.0976059436798096,
      "learning_rate": 0.00019674013505372546,
      "loss": 2.1607,
      "step": 140972
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.0696773529052734,
      "learning_rate": 0.00019673629444874824,
      "loss": 2.0064,
      "step": 140973
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.0552372932434082,
      "learning_rate": 0.00019673245386296978,
      "loss": 2.1164,
      "step": 140974
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.1040233373641968,
      "learning_rate": 0.00019672861329639063,
      "loss": 2.029,
      "step": 140975
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.0441217422485352,
      "learning_rate": 0.00019672477274901157,
      "loss": 2.1264,
      "step": 140976
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.1819746494293213,
      "learning_rate": 0.0001967209322208334,
      "loss": 2.1085,
      "step": 140977
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.1169102191925049,
      "learning_rate": 0.00019671709171185665,
      "loss": 2.1162,
      "step": 140978
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.0590842962265015,
      "learning_rate": 0.0001967132512220822,
      "loss": 1.9568,
      "step": 140979
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.1169027090072632,
      "learning_rate": 0.00019670941075151076,
      "loss": 2.165,
      "step": 140980
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.174651026725769,
      "learning_rate": 0.00019670557030014292,
      "loss": 1.9288,
      "step": 140981
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.1167619228363037,
      "learning_rate": 0.00019670172986797948,
      "loss": 2.044,
      "step": 140982
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.1305266618728638,
      "learning_rate": 0.00019669788945502113,
      "loss": 2.1108,
      "step": 140983
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.0783326625823975,
      "learning_rate": 0.0001966940490612686,
      "loss": 1.9437,
      "step": 140984
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.1602329015731812,
      "learning_rate": 0.0001966902086867226,
      "loss": 2.2935,
      "step": 140985
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.1292805671691895,
      "learning_rate": 0.00019668636833138387,
      "loss": 2.2012,
      "step": 140986
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.0745290517807007,
      "learning_rate": 0.00019668252799525304,
      "loss": 1.859,
      "step": 140987
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.0581419467926025,
      "learning_rate": 0.0001966786876783309,
      "loss": 1.9748,
      "step": 140988
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.1285991668701172,
      "learning_rate": 0.0001966748473806181,
      "loss": 1.9918,
      "step": 140989
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.188382625579834,
      "learning_rate": 0.0001966710071021154,
      "loss": 1.9319,
      "step": 140990
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.1099013090133667,
      "learning_rate": 0.00019666716684282363,
      "loss": 2.0037,
      "step": 140991
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.0552611351013184,
      "learning_rate": 0.00019666332660274327,
      "loss": 2.3069,
      "step": 140992
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.118497610092163,
      "learning_rate": 0.00019665948638187522,
      "loss": 2.0577,
      "step": 140993
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.1215543746948242,
      "learning_rate": 0.00019665564618022007,
      "loss": 2.0748,
      "step": 140994
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.1986421346664429,
      "learning_rate": 0.00019665180599777855,
      "loss": 2.2259,
      "step": 140995
    },
    {
      "epoch": 1.84,
      "grad_norm": 0.9858066439628601,
      "learning_rate": 0.00019664796583455149,
      "loss": 1.9954,
      "step": 140996
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.0513322353363037,
      "learning_rate": 0.00019664412569053943,
      "loss": 2.1243,
      "step": 140997
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.02071213722229,
      "learning_rate": 0.00019664028556574327,
      "loss": 1.9842,
      "step": 140998
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.043363094329834,
      "learning_rate": 0.00019663644546016372,
      "loss": 2.2027,
      "step": 140999
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.0330250263214111,
      "learning_rate": 0.00019663260537380124,
      "loss": 2.1534,
      "step": 141000
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.1410460472106934,
      "learning_rate": 0.00019662876530665674,
      "loss": 2.055,
      "step": 141001
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.1244919300079346,
      "learning_rate": 0.0001966249252587309,
      "loss": 1.9109,
      "step": 141002
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.0783283710479736,
      "learning_rate": 0.00019662108523002447,
      "loss": 2.1334,
      "step": 141003
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.3799042701721191,
      "learning_rate": 0.0001966172452205381,
      "loss": 2.2232,
      "step": 141004
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.0569205284118652,
      "learning_rate": 0.00019661340523027267,
      "loss": 2.2403,
      "step": 141005
    },
    {
      "epoch": 1.84,
      "grad_norm": 0.9755268096923828,
      "learning_rate": 0.0001966095652592286,
      "loss": 2.1476,
      "step": 141006
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.211268663406372,
      "learning_rate": 0.0001966057253074068,
      "loss": 2.1796,
      "step": 141007
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.108604907989502,
      "learning_rate": 0.00019660188537480794,
      "loss": 2.2281,
      "step": 141008
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.063591480255127,
      "learning_rate": 0.00019659804546143275,
      "loss": 1.8579,
      "step": 141009
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.061872959136963,
      "learning_rate": 0.00019659420556728194,
      "loss": 2.1742,
      "step": 141010
    },
    {
      "epoch": 1.84,
      "grad_norm": 0.9999309182167053,
      "learning_rate": 0.00019659036569235632,
      "loss": 1.8534,
      "step": 141011
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.008592963218689,
      "learning_rate": 0.00019658652583665637,
      "loss": 2.1797,
      "step": 141012
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.1135404109954834,
      "learning_rate": 0.00019658268600018298,
      "loss": 1.8035,
      "step": 141013
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.0137248039245605,
      "learning_rate": 0.00019657884618293678,
      "loss": 2.1515,
      "step": 141014
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.1407864093780518,
      "learning_rate": 0.00019657500638491855,
      "loss": 2.076,
      "step": 141015
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.061942458152771,
      "learning_rate": 0.00019657116660612894,
      "loss": 2.1727,
      "step": 141016
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.0219426155090332,
      "learning_rate": 0.0001965673268465689,
      "loss": 2.0023,
      "step": 141017
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.1003485918045044,
      "learning_rate": 0.00019656348710623875,
      "loss": 2.2775,
      "step": 141018
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.0782703161239624,
      "learning_rate": 0.00019655964738513943,
      "loss": 2.1292,
      "step": 141019
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.2889683246612549,
      "learning_rate": 0.00019655580768327161,
      "loss": 2.1011,
      "step": 141020
    },
    {
      "epoch": 1.84,
      "grad_norm": 0.975299596786499,
      "learning_rate": 0.00019655196800063605,
      "loss": 2.1805,
      "step": 141021
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.244465708732605,
      "learning_rate": 0.00019654812833723342,
      "loss": 2.0096,
      "step": 141022
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.0866148471832275,
      "learning_rate": 0.00019654428869306457,
      "loss": 2.1496,
      "step": 141023
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.0704752206802368,
      "learning_rate": 0.00019654044906812994,
      "loss": 2.0929,
      "step": 141024
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.110396146774292,
      "learning_rate": 0.00019653660946243042,
      "loss": 2.0929,
      "step": 141025
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.0559895038604736,
      "learning_rate": 0.0001965327698759667,
      "loss": 2.1481,
      "step": 141026
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.1391122341156006,
      "learning_rate": 0.00019652893030873948,
      "loss": 2.323,
      "step": 141027
    },
    {
      "epoch": 1.84,
      "grad_norm": 0.9688528776168823,
      "learning_rate": 0.00019652509076074947,
      "loss": 1.9436,
      "step": 141028
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.0423272848129272,
      "learning_rate": 0.00019652125123199754,
      "loss": 2.1419,
      "step": 141029
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.073014497756958,
      "learning_rate": 0.00019651741172248415,
      "loss": 2.0887,
      "step": 141030
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.0836570262908936,
      "learning_rate": 0.0001965135722322101,
      "loss": 2.2608,
      "step": 141031
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.6133575439453125,
      "learning_rate": 0.00019650973276117616,
      "loss": 2.0821,
      "step": 141032
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.2613743543624878,
      "learning_rate": 0.00019650589330938298,
      "loss": 2.1834,
      "step": 141033
    },
    {
      "epoch": 1.84,
      "grad_norm": 0.9845802783966064,
      "learning_rate": 0.00019650205387683134,
      "loss": 1.9928,
      "step": 141034
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.0730677843093872,
      "learning_rate": 0.00019649821446352206,
      "loss": 1.9776,
      "step": 141035
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.1018811464309692,
      "learning_rate": 0.00019649437506945552,
      "loss": 2.072,
      "step": 141036
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.0258736610412598,
      "learning_rate": 0.0001964905356946327,
      "loss": 2.1188,
      "step": 141037
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.0842781066894531,
      "learning_rate": 0.00019648669633905418,
      "loss": 2.0595,
      "step": 141038
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.071170687675476,
      "learning_rate": 0.0001964828570027208,
      "loss": 2.3049,
      "step": 141039
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.1040199995040894,
      "learning_rate": 0.00019647901768563318,
      "loss": 2.2567,
      "step": 141040
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.073351263999939,
      "learning_rate": 0.0001964751783877922,
      "loss": 2.0723,
      "step": 141041
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.0234209299087524,
      "learning_rate": 0.00019647133910919833,
      "loss": 2.078,
      "step": 141042
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.2487322092056274,
      "learning_rate": 0.00019646749984985235,
      "loss": 2.1064,
      "step": 141043
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.1795201301574707,
      "learning_rate": 0.00019646366060975504,
      "loss": 2.0746,
      "step": 141044
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.088489055633545,
      "learning_rate": 0.00019645982138890707,
      "loss": 2.1071,
      "step": 141045
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.1624594926834106,
      "learning_rate": 0.0001964559821873092,
      "loss": 1.9585,
      "step": 141046
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.2193753719329834,
      "learning_rate": 0.0001964521430049622,
      "loss": 2.1842,
      "step": 141047
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.1441035270690918,
      "learning_rate": 0.00019644830384186665,
      "loss": 2.1685,
      "step": 141048
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.0413979291915894,
      "learning_rate": 0.0001964444646980233,
      "loss": 1.9432,
      "step": 141049
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.1020609140396118,
      "learning_rate": 0.00019644062557343283,
      "loss": 2.105,
      "step": 141050
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.033996343612671,
      "learning_rate": 0.00019643678646809603,
      "loss": 1.9276,
      "step": 141051
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.054640531539917,
      "learning_rate": 0.0001964329473820136,
      "loss": 2.0622,
      "step": 141052
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.0322299003601074,
      "learning_rate": 0.00019642910831518625,
      "loss": 1.9806,
      "step": 141053
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.0008456707000732,
      "learning_rate": 0.0001964252692676147,
      "loss": 2.1597,
      "step": 141054
    },
    {
      "epoch": 1.84,
      "grad_norm": 0.9982930421829224,
      "learning_rate": 0.0001964214302392996,
      "loss": 2.0668,
      "step": 141055
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.1014702320098877,
      "learning_rate": 0.00019641759123024174,
      "loss": 2.0841,
      "step": 141056
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.2905950546264648,
      "learning_rate": 0.00019641375224044175,
      "loss": 2.2614,
      "step": 141057
    },
    {
      "epoch": 1.84,
      "grad_norm": 0.9895848631858826,
      "learning_rate": 0.00019640991326990044,
      "loss": 2.1176,
      "step": 141058
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.0053589344024658,
      "learning_rate": 0.0001964060743186185,
      "loss": 2.3496,
      "step": 141059
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.1241199970245361,
      "learning_rate": 0.00019640223538659664,
      "loss": 2.0133,
      "step": 141060
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.0553739070892334,
      "learning_rate": 0.0001963983964738355,
      "loss": 2.0196,
      "step": 141061
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.0049922466278076,
      "learning_rate": 0.00019639455758033592,
      "loss": 2.0668,
      "step": 141062
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.0765249729156494,
      "learning_rate": 0.0001963907187060985,
      "loss": 2.1866,
      "step": 141063
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.0529011487960815,
      "learning_rate": 0.000196386879851124,
      "loss": 2.0084,
      "step": 141064
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.0468432903289795,
      "learning_rate": 0.00019638304101541316,
      "loss": 2.125,
      "step": 141065
    },
    {
      "epoch": 1.84,
      "grad_norm": 0.9932323694229126,
      "learning_rate": 0.0001963792021989667,
      "loss": 2.0799,
      "step": 141066
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.243239402770996,
      "learning_rate": 0.00019637536340178524,
      "loss": 2.2111,
      "step": 141067
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.0137940645217896,
      "learning_rate": 0.00019637152462386955,
      "loss": 2.1449,
      "step": 141068
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.1272728443145752,
      "learning_rate": 0.00019636768586522044,
      "loss": 2.2149,
      "step": 141069
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.0990679264068604,
      "learning_rate": 0.00019636384712583843,
      "loss": 2.0259,
      "step": 141070
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.0457358360290527,
      "learning_rate": 0.00019636000840572438,
      "loss": 2.2823,
      "step": 141071
    },
    {
      "epoch": 1.84,
      "grad_norm": 0.984250545501709,
      "learning_rate": 0.000196356169704879,
      "loss": 1.7793,
      "step": 141072
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.124664068222046,
      "learning_rate": 0.00019635233102330293,
      "loss": 1.8845,
      "step": 141073
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.1070371866226196,
      "learning_rate": 0.0001963484923609969,
      "loss": 1.8742,
      "step": 141074
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.0557782649993896,
      "learning_rate": 0.0001963446537179616,
      "loss": 2.0271,
      "step": 141075
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.0823575258255005,
      "learning_rate": 0.00019634081509419787,
      "loss": 2.1122,
      "step": 141076
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.2936660051345825,
      "learning_rate": 0.00019633697648970634,
      "loss": 2.2203,
      "step": 141077
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.124086618423462,
      "learning_rate": 0.00019633313790448772,
      "loss": 2.0342,
      "step": 141078
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.0309791564941406,
      "learning_rate": 0.0001963292993385427,
      "loss": 2.0458,
      "step": 141079
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.1452417373657227,
      "learning_rate": 0.00019632546079187196,
      "loss": 2.0403,
      "step": 141080
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.1532626152038574,
      "learning_rate": 0.00019632162226447634,
      "loss": 2.0312,
      "step": 141081
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.4144057035446167,
      "learning_rate": 0.00019631778375635647,
      "loss": 2.0749,
      "step": 141082
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.0942751169204712,
      "learning_rate": 0.0001963139452675131,
      "loss": 2.2812,
      "step": 141083
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.0772671699523926,
      "learning_rate": 0.00019631010679794697,
      "loss": 2.1487,
      "step": 141084
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.0987282991409302,
      "learning_rate": 0.00019630626834765867,
      "loss": 2.0987,
      "step": 141085
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.1987169981002808,
      "learning_rate": 0.000196302429916649,
      "loss": 2.0417,
      "step": 141086
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.349283218383789,
      "learning_rate": 0.0001962985915049187,
      "loss": 2.2084,
      "step": 141087
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.2394040822982788,
      "learning_rate": 0.0001962947531124684,
      "loss": 1.9075,
      "step": 141088
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.1837679147720337,
      "learning_rate": 0.00019629091473929885,
      "loss": 2.043,
      "step": 141089
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.2903120517730713,
      "learning_rate": 0.00019628707638541092,
      "loss": 2.0187,
      "step": 141090
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.5403741598129272,
      "learning_rate": 0.0001962832380508051,
      "loss": 2.2168,
      "step": 141091
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.068507432937622,
      "learning_rate": 0.00019627939973548212,
      "loss": 2.1431,
      "step": 141092
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.0408787727355957,
      "learning_rate": 0.00019627556143944278,
      "loss": 2.3644,
      "step": 141093
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.1929826736450195,
      "learning_rate": 0.00019627172316268778,
      "loss": 2.095,
      "step": 141094
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.0642527341842651,
      "learning_rate": 0.00019626788490521784,
      "loss": 2.0706,
      "step": 141095
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.0917139053344727,
      "learning_rate": 0.00019626404666703374,
      "loss": 1.8454,
      "step": 141096
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.0843956470489502,
      "learning_rate": 0.00019626020844813603,
      "loss": 2.0721,
      "step": 141097
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.0935907363891602,
      "learning_rate": 0.0001962563702485255,
      "loss": 2.1977,
      "step": 141098
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.1395831108093262,
      "learning_rate": 0.00019625253206820287,
      "loss": 2.1752,
      "step": 141099
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.0345159769058228,
      "learning_rate": 0.0001962486939071688,
      "loss": 2.0929,
      "step": 141100
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.0799925327301025,
      "learning_rate": 0.00019624485576542411,
      "loss": 2.2637,
      "step": 141101
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.3168230056762695,
      "learning_rate": 0.00019624101764296957,
      "loss": 2.2083,
      "step": 141102
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.2801032066345215,
      "learning_rate": 0.00019623717953980567,
      "loss": 2.2637,
      "step": 141103
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.1101194620132446,
      "learning_rate": 0.0001962333414559332,
      "loss": 2.0327,
      "step": 141104
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.0577728748321533,
      "learning_rate": 0.00019622950339135296,
      "loss": 2.1828,
      "step": 141105
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.0690604448318481,
      "learning_rate": 0.00019622566534606555,
      "loss": 1.9376,
      "step": 141106
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.2442504167556763,
      "learning_rate": 0.0001962218273200718,
      "loss": 2.0301,
      "step": 141107
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.0429524183273315,
      "learning_rate": 0.00019621798931337248,
      "loss": 2.1628,
      "step": 141108
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.0092861652374268,
      "learning_rate": 0.00019621415132596806,
      "loss": 2.0117,
      "step": 141109
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.061382532119751,
      "learning_rate": 0.00019621031335785943,
      "loss": 1.9149,
      "step": 141110
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.2543644905090332,
      "learning_rate": 0.00019620647540904724,
      "loss": 2.202,
      "step": 141111
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.0765975713729858,
      "learning_rate": 0.0001962026374795322,
      "loss": 2.0032,
      "step": 141112
    },
    {
      "epoch": 1.84,
      "grad_norm": 0.9784461855888367,
      "learning_rate": 0.00019619879956931508,
      "loss": 1.8404,
      "step": 141113
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.041099190711975,
      "learning_rate": 0.00019619496167839668,
      "loss": 2.3454,
      "step": 141114
    },
    {
      "epoch": 1.84,
      "grad_norm": 0.9980787634849548,
      "learning_rate": 0.00019619112380677746,
      "loss": 1.9165,
      "step": 141115
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.170154094696045,
      "learning_rate": 0.00019618728595445826,
      "loss": 1.932,
      "step": 141116
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.249906063079834,
      "learning_rate": 0.0001961834481214398,
      "loss": 2.2928,
      "step": 141117
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.223523497581482,
      "learning_rate": 0.00019617961030772284,
      "loss": 1.7718,
      "step": 141118
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.170966625213623,
      "learning_rate": 0.00019617577251330805,
      "loss": 2.1387,
      "step": 141119
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.084689974784851,
      "learning_rate": 0.00019617193473819622,
      "loss": 2.0521,
      "step": 141120
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.4763039350509644,
      "learning_rate": 0.00019616809698238787,
      "loss": 2.1936,
      "step": 141121
    },
    {
      "epoch": 1.84,
      "grad_norm": 0.9906159043312073,
      "learning_rate": 0.00019616425924588388,
      "loss": 2.0628,
      "step": 141122
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.4748039245605469,
      "learning_rate": 0.00019616042152868486,
      "loss": 2.3753,
      "step": 141123
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.0843186378479004,
      "learning_rate": 0.0001961565838307916,
      "loss": 1.8843,
      "step": 141124
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.0346418619155884,
      "learning_rate": 0.0001961527461522048,
      "loss": 2.0207,
      "step": 141125
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.0402381420135498,
      "learning_rate": 0.00019614890849292517,
      "loss": 2.1912,
      "step": 141126
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.1371666193008423,
      "learning_rate": 0.0001961450708529535,
      "loss": 2.2021,
      "step": 141127
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.176437258720398,
      "learning_rate": 0.00019614123323229032,
      "loss": 2.1191,
      "step": 141128
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.06385338306427,
      "learning_rate": 0.0001961373956309365,
      "loss": 2.19,
      "step": 141129
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.050518274307251,
      "learning_rate": 0.0001961335580488926,
      "loss": 2.0624,
      "step": 141130
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.1392351388931274,
      "learning_rate": 0.0001961297204861595,
      "loss": 2.123,
      "step": 141131
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.179433822631836,
      "learning_rate": 0.00019612588294273781,
      "loss": 2.0437,
      "step": 141132
    },
    {
      "epoch": 1.84,
      "grad_norm": 0.975473940372467,
      "learning_rate": 0.00019612204541862843,
      "loss": 1.8852,
      "step": 141133
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.063299536705017,
      "learning_rate": 0.0001961182079138318,
      "loss": 2.1631,
      "step": 141134
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.0406841039657593,
      "learning_rate": 0.00019611437042834872,
      "loss": 2.2071,
      "step": 141135
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.072548270225525,
      "learning_rate": 0.00019611053296218,
      "loss": 1.9487,
      "step": 141136
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.1158894300460815,
      "learning_rate": 0.00019610669551532625,
      "loss": 2.0602,
      "step": 141137
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.1877281665802002,
      "learning_rate": 0.00019610285808778822,
      "loss": 1.9899,
      "step": 141138
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.0945286750793457,
      "learning_rate": 0.0001960990206795668,
      "loss": 1.9772,
      "step": 141139
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.408879041671753,
      "learning_rate": 0.00019609518329066235,
      "loss": 1.9074,
      "step": 141140
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.0853191614151,
      "learning_rate": 0.0001960913459210758,
      "loss": 1.9996,
      "step": 141141
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.0771077871322632,
      "learning_rate": 0.00019608750857080782,
      "loss": 2.0069,
      "step": 141142
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.0106970071792603,
      "learning_rate": 0.00019608367123985916,
      "loss": 2.2059,
      "step": 141143
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.1248594522476196,
      "learning_rate": 0.00019607983392823047,
      "loss": 2.2437,
      "step": 141144
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.1184980869293213,
      "learning_rate": 0.0001960759966359226,
      "loss": 2.2048,
      "step": 141145
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.2099159955978394,
      "learning_rate": 0.0001960721593629361,
      "loss": 1.9534,
      "step": 141146
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.1097033023834229,
      "learning_rate": 0.00019606832210927174,
      "loss": 2.1403,
      "step": 141147
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.1087803840637207,
      "learning_rate": 0.00019606448487493024,
      "loss": 2.1705,
      "step": 141148
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.0722105503082275,
      "learning_rate": 0.00019606064765991226,
      "loss": 2.2239,
      "step": 141149
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.0431232452392578,
      "learning_rate": 0.00019605681046421864,
      "loss": 1.9947,
      "step": 141150
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.1304347515106201,
      "learning_rate": 0.00019605297328785003,
      "loss": 2.229,
      "step": 141151
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.0784639120101929,
      "learning_rate": 0.00019604913613080713,
      "loss": 2.1127,
      "step": 141152
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.1069453954696655,
      "learning_rate": 0.0001960452989930907,
      "loss": 2.131,
      "step": 141153
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.1290233135223389,
      "learning_rate": 0.00019604146187470134,
      "loss": 1.9519,
      "step": 141154
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.0382585525512695,
      "learning_rate": 0.0001960376247756398,
      "loss": 1.9583,
      "step": 141155
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.1625440120697021,
      "learning_rate": 0.00019603378769590687,
      "loss": 2.067,
      "step": 141156
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.1116787195205688,
      "learning_rate": 0.00019602995063550325,
      "loss": 1.855,
      "step": 141157
    },
    {
      "epoch": 1.84,
      "grad_norm": 0.9735884070396423,
      "learning_rate": 0.00019602611359442962,
      "loss": 2.3523,
      "step": 141158
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.1044273376464844,
      "learning_rate": 0.00019602227657268662,
      "loss": 1.9897,
      "step": 141159
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.1437983512878418,
      "learning_rate": 0.00019601843957027507,
      "loss": 2.1737,
      "step": 141160
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.2029534578323364,
      "learning_rate": 0.00019601460258719574,
      "loss": 2.2686,
      "step": 141161
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.0342031717300415,
      "learning_rate": 0.00019601076562344923,
      "loss": 1.9593,
      "step": 141162
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.0417187213897705,
      "learning_rate": 0.0001960069286790363,
      "loss": 2.0979,
      "step": 141163
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.3179385662078857,
      "learning_rate": 0.00019600309175395755,
      "loss": 2.0398,
      "step": 141164
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.2492870092391968,
      "learning_rate": 0.00019599925484821383,
      "loss": 2.1353,
      "step": 141165
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.0487581491470337,
      "learning_rate": 0.0001959954179618058,
      "loss": 2.052,
      "step": 141166
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.0642722845077515,
      "learning_rate": 0.0001959915810947342,
      "loss": 2.1582,
      "step": 141167
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.0298153162002563,
      "learning_rate": 0.00019598774424699976,
      "loss": 1.9477,
      "step": 141168
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.079938292503357,
      "learning_rate": 0.0001959839074186032,
      "loss": 2.2815,
      "step": 141169
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.0043768882751465,
      "learning_rate": 0.0001959800706095451,
      "loss": 2.1659,
      "step": 141170
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.153963565826416,
      "learning_rate": 0.00019597623381982626,
      "loss": 2.1087,
      "step": 141171
    },
    {
      "epoch": 1.84,
      "grad_norm": 0.9907351136207581,
      "learning_rate": 0.00019597239704944745,
      "loss": 2.1665,
      "step": 141172
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.0034310817718506,
      "learning_rate": 0.00019596856029840932,
      "loss": 2.0622,
      "step": 141173
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.0240641832351685,
      "learning_rate": 0.00019596472356671256,
      "loss": 2.1099,
      "step": 141174
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.1613945960998535,
      "learning_rate": 0.00019596088685435806,
      "loss": 2.0912,
      "step": 141175
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.319435715675354,
      "learning_rate": 0.00019595705016134632,
      "loss": 1.9688,
      "step": 141176
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.2166509628295898,
      "learning_rate": 0.00019595321348767807,
      "loss": 2.2299,
      "step": 141177
    },
    {
      "epoch": 1.84,
      "grad_norm": 0.943195104598999,
      "learning_rate": 0.0001959493768333541,
      "loss": 2.1735,
      "step": 141178
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.0965263843536377,
      "learning_rate": 0.00019594554019837512,
      "loss": 2.1209,
      "step": 141179
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.2454993724822998,
      "learning_rate": 0.0001959417035827418,
      "loss": 1.9496,
      "step": 141180
    },
    {
      "epoch": 1.84,
      "grad_norm": 0.9862996935844421,
      "learning_rate": 0.00019593786698645502,
      "loss": 1.9624,
      "step": 141181
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.0183429718017578,
      "learning_rate": 0.00019593403040951525,
      "loss": 2.2262,
      "step": 141182
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.0772881507873535,
      "learning_rate": 0.0001959301938519233,
      "loss": 1.8804,
      "step": 141183
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.071609616279602,
      "learning_rate": 0.00019592635731367986,
      "loss": 2.2307,
      "step": 141184
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.0830570459365845,
      "learning_rate": 0.0001959225207947857,
      "loss": 1.9294,
      "step": 141185
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.0693144798278809,
      "learning_rate": 0.00019591868429524151,
      "loss": 2.1325,
      "step": 141186
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.0727132558822632,
      "learning_rate": 0.00019591484781504811,
      "loss": 2.0382,
      "step": 141187
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.0823805332183838,
      "learning_rate": 0.00019591101135420596,
      "loss": 1.9228,
      "step": 141188
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.047383189201355,
      "learning_rate": 0.00019590717491271598,
      "loss": 2.2892,
      "step": 141189
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.0829155445098877,
      "learning_rate": 0.00019590333849057879,
      "loss": 2.0226,
      "step": 141190
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.211248755455017,
      "learning_rate": 0.00019589950208779512,
      "loss": 1.9333,
      "step": 141191
    },
    {
      "epoch": 1.84,
      "grad_norm": 0.996871292591095,
      "learning_rate": 0.00019589566570436573,
      "loss": 2.0387,
      "step": 141192
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.2320514917373657,
      "learning_rate": 0.00019589182934029128,
      "loss": 2.1998,
      "step": 141193
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.0207607746124268,
      "learning_rate": 0.0001958879929955726,
      "loss": 2.0673,
      "step": 141194
    },
    {
      "epoch": 1.84,
      "grad_norm": 0.9994513392448425,
      "learning_rate": 0.00019588415667021024,
      "loss": 2.1723,
      "step": 141195
    },
    {
      "epoch": 1.84,
      "grad_norm": 0.9660261273384094,
      "learning_rate": 0.00019588032036420495,
      "loss": 2.1871,
      "step": 141196
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.0322000980377197,
      "learning_rate": 0.00019587648407755746,
      "loss": 1.969,
      "step": 141197
    },
    {
      "epoch": 1.84,
      "grad_norm": 0.985273540019989,
      "learning_rate": 0.00019587264781026855,
      "loss": 2.1249,
      "step": 141198
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.005340337753296,
      "learning_rate": 0.00019586881156233885,
      "loss": 2.1411,
      "step": 141199
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.0769057273864746,
      "learning_rate": 0.0001958649753337692,
      "loss": 2.1188,
      "step": 141200
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.0685265064239502,
      "learning_rate": 0.00019586113912456015,
      "loss": 2.0722,
      "step": 141201
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.04007887840271,
      "learning_rate": 0.00019585730293471244,
      "loss": 1.9592,
      "step": 141202
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.181837558746338,
      "learning_rate": 0.0001958534667642268,
      "loss": 1.9446,
      "step": 141203
    },
    {
      "epoch": 1.84,
      "grad_norm": 0.9726691842079163,
      "learning_rate": 0.000195849630613104,
      "loss": 2.2023,
      "step": 141204
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.009081244468689,
      "learning_rate": 0.00019584579448134475,
      "loss": 2.0793,
      "step": 141205
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.1292037963867188,
      "learning_rate": 0.0001958419583689498,
      "loss": 2.2561,
      "step": 141206
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.0467698574066162,
      "learning_rate": 0.0001958381222759197,
      "loss": 2.0075,
      "step": 141207
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.045123815536499,
      "learning_rate": 0.00019583428620225526,
      "loss": 2.1837,
      "step": 141208
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.0176868438720703,
      "learning_rate": 0.00019583045014795719,
      "loss": 2.1673,
      "step": 141209
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.0754414796829224,
      "learning_rate": 0.0001958266141130262,
      "loss": 2.0196,
      "step": 141210
    },
    {
      "epoch": 1.84,
      "grad_norm": 0.9511508345603943,
      "learning_rate": 0.00019582277809746302,
      "loss": 2.1478,
      "step": 141211
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.089611291885376,
      "learning_rate": 0.00019581894210126844,
      "loss": 2.1221,
      "step": 141212
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.043871283531189,
      "learning_rate": 0.000195815106124443,
      "loss": 1.9941,
      "step": 141213
    },
    {
      "epoch": 1.84,
      "grad_norm": 0.9830849766731262,
      "learning_rate": 0.00019581127016698747,
      "loss": 2.1335,
      "step": 141214
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.5309311151504517,
      "learning_rate": 0.0001958074342289026,
      "loss": 2.0652,
      "step": 141215
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.0851083993911743,
      "learning_rate": 0.0001958035983101891,
      "loss": 2.1078,
      "step": 141216
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.019923448562622,
      "learning_rate": 0.00019579976241084773,
      "loss": 2.162,
      "step": 141217
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.0066438913345337,
      "learning_rate": 0.0001957959265308792,
      "loss": 2.1328,
      "step": 141218
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.1006710529327393,
      "learning_rate": 0.00019579209067028405,
      "loss": 2.0793,
      "step": 141219
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.086010456085205,
      "learning_rate": 0.00019578825482906317,
      "loss": 2.0029,
      "step": 141220
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.0918587446212769,
      "learning_rate": 0.0001957844190072172,
      "loss": 2.2485,
      "step": 141221
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.1279433965682983,
      "learning_rate": 0.00019578058320474684,
      "loss": 2.0224,
      "step": 141222
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.0624256134033203,
      "learning_rate": 0.00019577674742165288,
      "loss": 2.1048,
      "step": 141223
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.041541337966919,
      "learning_rate": 0.00019577291165793608,
      "loss": 2.2116,
      "step": 141224
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.0219428539276123,
      "learning_rate": 0.00019576907591359698,
      "loss": 2.0248,
      "step": 141225
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.0555912256240845,
      "learning_rate": 0.00019576524018863635,
      "loss": 2.1363,
      "step": 141226
    },
    {
      "epoch": 1.84,
      "grad_norm": 0.9490591883659363,
      "learning_rate": 0.00019576140448305493,
      "loss": 2.2017,
      "step": 141227
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.062473177909851,
      "learning_rate": 0.00019575756879685344,
      "loss": 2.2454,
      "step": 141228
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.1948643922805786,
      "learning_rate": 0.00019575373313003257,
      "loss": 2.5088,
      "step": 141229
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.0489578247070312,
      "learning_rate": 0.0001957498974825931,
      "loss": 2.1443,
      "step": 141230
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.071142554283142,
      "learning_rate": 0.00019574606185453574,
      "loss": 1.8898,
      "step": 141231
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.0746341943740845,
      "learning_rate": 0.00019574222624586107,
      "loss": 1.8861,
      "step": 141232
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.0779390335083008,
      "learning_rate": 0.00019573839065656988,
      "loss": 2.2373,
      "step": 141233
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.0991644859313965,
      "learning_rate": 0.0001957345550866629,
      "loss": 2.0349,
      "step": 141234
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.2594319581985474,
      "learning_rate": 0.00019573071953614085,
      "loss": 2.1056,
      "step": 141235
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.034774661064148,
      "learning_rate": 0.00019572688400500444,
      "loss": 2.0295,
      "step": 141236
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.0838489532470703,
      "learning_rate": 0.00019572304849325438,
      "loss": 2.0928,
      "step": 141237
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.2924190759658813,
      "learning_rate": 0.0001957192130008914,
      "loss": 1.9385,
      "step": 141238
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.2139482498168945,
      "learning_rate": 0.0001957153775279161,
      "loss": 2.0871,
      "step": 141239
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.233573079109192,
      "learning_rate": 0.0001957115420743293,
      "loss": 2.0895,
      "step": 141240
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.092532753944397,
      "learning_rate": 0.00019570770664013168,
      "loss": 2.2732,
      "step": 141241
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.064355731010437,
      "learning_rate": 0.00019570387122532404,
      "loss": 2.1026,
      "step": 141242
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.0514013767242432,
      "learning_rate": 0.00019570003582990698,
      "loss": 2.2248,
      "step": 141243
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.0851951837539673,
      "learning_rate": 0.0001956962004538812,
      "loss": 1.8902,
      "step": 141244
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.1035678386688232,
      "learning_rate": 0.0001956923650972476,
      "loss": 1.9993,
      "step": 141245
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.4758926630020142,
      "learning_rate": 0.00019568852976000663,
      "loss": 2.3375,
      "step": 141246
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.0516433715820312,
      "learning_rate": 0.00019568469444215916,
      "loss": 1.9859,
      "step": 141247
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.0155950784683228,
      "learning_rate": 0.00019568085914370594,
      "loss": 2.2758,
      "step": 141248
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.0932995080947876,
      "learning_rate": 0.0001956770238646475,
      "loss": 2.0378,
      "step": 141249
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.0250496864318848,
      "learning_rate": 0.00019567318860498473,
      "loss": 2.0261,
      "step": 141250
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.196811318397522,
      "learning_rate": 0.0001956693533647183,
      "loss": 2.0537,
      "step": 141251
    },
    {
      "epoch": 1.84,
      "grad_norm": 0.9778933525085449,
      "learning_rate": 0.00019566551814384888,
      "loss": 2.189,
      "step": 141252
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.1107465028762817,
      "learning_rate": 0.00019566168294237726,
      "loss": 2.2764,
      "step": 141253
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.2066755294799805,
      "learning_rate": 0.00019565784776030408,
      "loss": 1.976,
      "step": 141254
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.0350341796875,
      "learning_rate": 0.00019565401259763005,
      "loss": 2.0679,
      "step": 141255
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.1205286979675293,
      "learning_rate": 0.00019565017745435594,
      "loss": 2.0731,
      "step": 141256
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.2682411670684814,
      "learning_rate": 0.00019564634233048235,
      "loss": 2.1436,
      "step": 141257
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.0402843952178955,
      "learning_rate": 0.00019564250722601016,
      "loss": 1.9741,
      "step": 141258
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.1108158826828003,
      "learning_rate": 0.00019563867214093995,
      "loss": 2.2064,
      "step": 141259
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.0313405990600586,
      "learning_rate": 0.00019563483707527255,
      "loss": 2.1643,
      "step": 141260
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.0484081506729126,
      "learning_rate": 0.0001956310020290086,
      "loss": 2.3091,
      "step": 141261
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.209936261177063,
      "learning_rate": 0.00019562716700214877,
      "loss": 2.2438,
      "step": 141262
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.1891603469848633,
      "learning_rate": 0.00019562333199469377,
      "loss": 2.2731,
      "step": 141263
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.1392465829849243,
      "learning_rate": 0.00019561949700664446,
      "loss": 1.9584,
      "step": 141264
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.0312981605529785,
      "learning_rate": 0.00019561566203800138,
      "loss": 2.1653,
      "step": 141265
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.2472302913665771,
      "learning_rate": 0.0001956118270887653,
      "loss": 1.9958,
      "step": 141266
    },
    {
      "epoch": 1.84,
      "grad_norm": 0.9944878816604614,
      "learning_rate": 0.00019560799215893713,
      "loss": 1.9926,
      "step": 141267
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.0462502241134644,
      "learning_rate": 0.00019560415724851726,
      "loss": 2.0217,
      "step": 141268
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.1359502077102661,
      "learning_rate": 0.00019560032235750658,
      "loss": 2.2501,
      "step": 141269
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.0124536752700806,
      "learning_rate": 0.0001955964874859057,
      "loss": 2.0365,
      "step": 141270
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.0305484533309937,
      "learning_rate": 0.00019559265263371548,
      "loss": 2.026,
      "step": 141271
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.256993055343628,
      "learning_rate": 0.00019558881780093652,
      "loss": 2.0213,
      "step": 141272
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.1373409032821655,
      "learning_rate": 0.00019558498298756968,
      "loss": 2.2084,
      "step": 141273
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.0053133964538574,
      "learning_rate": 0.00019558114819361548,
      "loss": 2.1216,
      "step": 141274
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.1794003248214722,
      "learning_rate": 0.0001955773134190747,
      "loss": 2.2056,
      "step": 141275
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.2542353868484497,
      "learning_rate": 0.00019557347866394807,
      "loss": 2.3951,
      "step": 141276
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.480535626411438,
      "learning_rate": 0.0001955696439282363,
      "loss": 2.1349,
      "step": 141277
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.2606197595596313,
      "learning_rate": 0.00019556580921194005,
      "loss": 1.9206,
      "step": 141278
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.1029306650161743,
      "learning_rate": 0.00019556197451506028,
      "loss": 1.9498,
      "step": 141279
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.3121927976608276,
      "learning_rate": 0.00019555813983759734,
      "loss": 2.2932,
      "step": 141280
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.3917250633239746,
      "learning_rate": 0.00019555430517955214,
      "loss": 1.9193,
      "step": 141281
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.0216960906982422,
      "learning_rate": 0.00019555047054092538,
      "loss": 2.025,
      "step": 141282
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.0652050971984863,
      "learning_rate": 0.0001955466359217177,
      "loss": 2.0418,
      "step": 141283
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.283180594444275,
      "learning_rate": 0.00019554280132192992,
      "loss": 1.9563,
      "step": 141284
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.027178168296814,
      "learning_rate": 0.00019553896674156284,
      "loss": 2.0967,
      "step": 141285
    },
    {
      "epoch": 1.84,
      "grad_norm": 0.9803152084350586,
      "learning_rate": 0.00019553513218061692,
      "loss": 1.9837,
      "step": 141286
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.1309471130371094,
      "learning_rate": 0.00019553129763909293,
      "loss": 2.2003,
      "step": 141287
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.2054861783981323,
      "learning_rate": 0.0001955274631169917,
      "loss": 1.9475,
      "step": 141288
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.0577892065048218,
      "learning_rate": 0.00019552362861431386,
      "loss": 2.1446,
      "step": 141289
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.0139076709747314,
      "learning_rate": 0.00019551979413106018,
      "loss": 2.087,
      "step": 141290
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.1990753412246704,
      "learning_rate": 0.00019551595966723142,
      "loss": 2.0817,
      "step": 141291
    },
    {
      "epoch": 1.84,
      "grad_norm": 0.9859792590141296,
      "learning_rate": 0.00019551212522282813,
      "loss": 2.114,
      "step": 141292
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.0883151292800903,
      "learning_rate": 0.00019550829079785108,
      "loss": 1.9733,
      "step": 141293
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.0841898918151855,
      "learning_rate": 0.000195504456392301,
      "loss": 2.1695,
      "step": 141294
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.063429355621338,
      "learning_rate": 0.00019550062200617867,
      "loss": 2.129,
      "step": 141295
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.0487428903579712,
      "learning_rate": 0.00019549678763948472,
      "loss": 2.0894,
      "step": 141296
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.1954114437103271,
      "learning_rate": 0.00019549295329222,
      "loss": 2.0396,
      "step": 141297
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.112597107887268,
      "learning_rate": 0.000195489118964385,
      "loss": 2.0015,
      "step": 141298
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.1507799625396729,
      "learning_rate": 0.00019548528465598055,
      "loss": 1.9074,
      "step": 141299
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.0250808000564575,
      "learning_rate": 0.00019548145036700737,
      "loss": 2.1056,
      "step": 141300
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.297337293624878,
      "learning_rate": 0.00019547761609746612,
      "loss": 2.0518,
      "step": 141301
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.6375856399536133,
      "learning_rate": 0.00019547378184735757,
      "loss": 2.25,
      "step": 141302
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.0126005411148071,
      "learning_rate": 0.00019546994761668256,
      "loss": 2.2256,
      "step": 141303
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.109532356262207,
      "learning_rate": 0.00019546611340544152,
      "loss": 1.9101,
      "step": 141304
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.124576210975647,
      "learning_rate": 0.00019546227921363532,
      "loss": 2.1152,
      "step": 141305
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.819659948348999,
      "learning_rate": 0.00019545844504126466,
      "loss": 2.1889,
      "step": 141306
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.1399871110916138,
      "learning_rate": 0.00019545461088833023,
      "loss": 1.8211,
      "step": 141307
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.1628217697143555,
      "learning_rate": 0.00019545077675483277,
      "loss": 1.951,
      "step": 141308
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.1845812797546387,
      "learning_rate": 0.00019544694264077308,
      "loss": 2.1699,
      "step": 141309
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.5097532272338867,
      "learning_rate": 0.00019544310854615167,
      "loss": 2.0455,
      "step": 141310
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.2610198259353638,
      "learning_rate": 0.0001954392744709694,
      "loss": 2.1974,
      "step": 141311
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.4147274494171143,
      "learning_rate": 0.0001954354404152269,
      "loss": 2.1862,
      "step": 141312
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.1544644832611084,
      "learning_rate": 0.00019543160637892492,
      "loss": 1.9913,
      "step": 141313
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.1684459447860718,
      "learning_rate": 0.0001954277723620642,
      "loss": 2.0952,
      "step": 141314
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.1521724462509155,
      "learning_rate": 0.00019542393836464552,
      "loss": 2.0858,
      "step": 141315
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.0117542743682861,
      "learning_rate": 0.00019542010438666942,
      "loss": 2.1168,
      "step": 141316
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.138994574546814,
      "learning_rate": 0.00019541627042813666,
      "loss": 1.9316,
      "step": 141317
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.4398469924926758,
      "learning_rate": 0.00019541243648904804,
      "loss": 2.2201,
      "step": 141318
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.1335917711257935,
      "learning_rate": 0.00019540860256940417,
      "loss": 2.2383,
      "step": 141319
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.0859636068344116,
      "learning_rate": 0.00019540476866920583,
      "loss": 2.139,
      "step": 141320
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.0972691774368286,
      "learning_rate": 0.00019540093478845382,
      "loss": 2.0642,
      "step": 141321
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.1308012008666992,
      "learning_rate": 0.00019539710092714868,
      "loss": 2.0256,
      "step": 141322
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.0222704410552979,
      "learning_rate": 0.00019539326708529123,
      "loss": 1.8746,
      "step": 141323
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.111351728439331,
      "learning_rate": 0.00019538943326288206,
      "loss": 2.1704,
      "step": 141324
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.077438235282898,
      "learning_rate": 0.00019538559945992198,
      "loss": 2.0639,
      "step": 141325
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.0552767515182495,
      "learning_rate": 0.00019538176567641174,
      "loss": 2.2053,
      "step": 141326
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.2624614238739014,
      "learning_rate": 0.00019537793191235194,
      "loss": 1.973,
      "step": 141327
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.038981556892395,
      "learning_rate": 0.00019537409816774347,
      "loss": 2.2676,
      "step": 141328
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.064499020576477,
      "learning_rate": 0.00019537026444258685,
      "loss": 2.1402,
      "step": 141329
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.230713963508606,
      "learning_rate": 0.00019536643073688291,
      "loss": 2.2545,
      "step": 141330
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.274960994720459,
      "learning_rate": 0.0001953625970506323,
      "loss": 2.2761,
      "step": 141331
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.0662410259246826,
      "learning_rate": 0.00019535876338383574,
      "loss": 2.2053,
      "step": 141332
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.0346053838729858,
      "learning_rate": 0.000195354929736494,
      "loss": 2.1173,
      "step": 141333
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.1973379850387573,
      "learning_rate": 0.00019535109610860777,
      "loss": 2.1952,
      "step": 141334
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.1685370206832886,
      "learning_rate": 0.0001953472625001777,
      "loss": 2.1729,
      "step": 141335
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.0184741020202637,
      "learning_rate": 0.00019534342891120457,
      "loss": 1.8919,
      "step": 141336
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.060774564743042,
      "learning_rate": 0.00019533959534168905,
      "loss": 2.1727,
      "step": 141337
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.1355516910552979,
      "learning_rate": 0.00019533576179163193,
      "loss": 2.0932,
      "step": 141338
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.0440248250961304,
      "learning_rate": 0.00019533192826103377,
      "loss": 2.1661,
      "step": 141339
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.062696099281311,
      "learning_rate": 0.0001953280947498955,
      "loss": 2.1735,
      "step": 141340
    },
    {
      "epoch": 1.84,
      "grad_norm": 0.9982030987739563,
      "learning_rate": 0.00019532426125821764,
      "loss": 2.0567,
      "step": 141341
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.4353009462356567,
      "learning_rate": 0.00019532042778600098,
      "loss": 2.1722,
      "step": 141342
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.1377071142196655,
      "learning_rate": 0.0001953165943332462,
      "loss": 2.2177,
      "step": 141343
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.000491976737976,
      "learning_rate": 0.0001953127608999541,
      "loss": 2.0622,
      "step": 141344
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.0267236232757568,
      "learning_rate": 0.00019530892748612533,
      "loss": 2.1166,
      "step": 141345
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.2335063219070435,
      "learning_rate": 0.00019530509409176064,
      "loss": 2.1882,
      "step": 141346
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.229235291481018,
      "learning_rate": 0.00019530126071686064,
      "loss": 2.2932,
      "step": 141347
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.0602807998657227,
      "learning_rate": 0.0001952974273614261,
      "loss": 2.1503,
      "step": 141348
    },
    {
      "epoch": 1.84,
      "grad_norm": 0.981198787689209,
      "learning_rate": 0.00019529359402545776,
      "loss": 1.9381,
      "step": 141349
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.0341686010360718,
      "learning_rate": 0.00019528976070895635,
      "loss": 2.0661,
      "step": 141350
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.149437427520752,
      "learning_rate": 0.00019528592741192247,
      "loss": 2.0409,
      "step": 141351
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.0328257083892822,
      "learning_rate": 0.00019528209413435712,
      "loss": 2.1188,
      "step": 141352
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.1051678657531738,
      "learning_rate": 0.00019527826087626063,
      "loss": 2.1383,
      "step": 141353
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.347987174987793,
      "learning_rate": 0.00019527442763763388,
      "loss": 2.1163,
      "step": 141354
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.1776683330535889,
      "learning_rate": 0.00019527059441847762,
      "loss": 2.156,
      "step": 141355
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.170209288597107,
      "learning_rate": 0.00019526676121879255,
      "loss": 1.969,
      "step": 141356
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.3211979866027832,
      "learning_rate": 0.00019526292803857934,
      "loss": 2.1732,
      "step": 141357
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.1776865720748901,
      "learning_rate": 0.00019525909487783883,
      "loss": 2.1979,
      "step": 141358
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.305165410041809,
      "learning_rate": 0.00019525526173657154,
      "loss": 2.1875,
      "step": 141359
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.2025307416915894,
      "learning_rate": 0.00019525142861477826,
      "loss": 2.2767,
      "step": 141360
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.183170199394226,
      "learning_rate": 0.00019524759551245972,
      "loss": 2.1961,
      "step": 141361
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.0645126104354858,
      "learning_rate": 0.00019524376242961665,
      "loss": 1.9331,
      "step": 141362
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.3711491823196411,
      "learning_rate": 0.00019523992936624972,
      "loss": 2.2361,
      "step": 141363
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.283866047859192,
      "learning_rate": 0.00019523609632235978,
      "loss": 2.2453,
      "step": 141364
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.2969304323196411,
      "learning_rate": 0.00019523226329794732,
      "loss": 1.958,
      "step": 141365
    },
    {
      "epoch": 1.84,
      "grad_norm": 0.9276759624481201,
      "learning_rate": 0.00019522843029301314,
      "loss": 2.0465,
      "step": 141366
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.1422220468521118,
      "learning_rate": 0.00019522459730755802,
      "loss": 2.1126,
      "step": 141367
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.4159735441207886,
      "learning_rate": 0.00019522076434158256,
      "loss": 1.9182,
      "step": 141368
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.3779062032699585,
      "learning_rate": 0.00019521693139508756,
      "loss": 2.071,
      "step": 141369
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.0497545003890991,
      "learning_rate": 0.00019521309846807385,
      "loss": 2.0947,
      "step": 141370
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.069008469581604,
      "learning_rate": 0.00019520926556054188,
      "loss": 2.0202,
      "step": 141371
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.568782925605774,
      "learning_rate": 0.0001952054326724925,
      "loss": 1.9615,
      "step": 141372
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.1716667413711548,
      "learning_rate": 0.0001952015998039264,
      "loss": 2.1127,
      "step": 141373
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.0722562074661255,
      "learning_rate": 0.0001951977669548443,
      "loss": 1.9646,
      "step": 141374
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.3242721557617188,
      "learning_rate": 0.00019519393412524688,
      "loss": 1.8716,
      "step": 141375
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.215373158454895,
      "learning_rate": 0.00019519010131513503,
      "loss": 2.0537,
      "step": 141376
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.3650968074798584,
      "learning_rate": 0.0001951862685245092,
      "loss": 1.9686,
      "step": 141377
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.0537322759628296,
      "learning_rate": 0.0001951824357533702,
      "loss": 2.053,
      "step": 141378
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.2631709575653076,
      "learning_rate": 0.00019517860300171882,
      "loss": 2.2746,
      "step": 141379
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.3497576713562012,
      "learning_rate": 0.00019517477026955563,
      "loss": 2.2181,
      "step": 141380
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.156015157699585,
      "learning_rate": 0.0001951709375568815,
      "loss": 2.0349,
      "step": 141381
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.1316426992416382,
      "learning_rate": 0.00019516710486369718,
      "loss": 1.9363,
      "step": 141382
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.0125811100006104,
      "learning_rate": 0.00019516327219000314,
      "loss": 2.2364,
      "step": 141383
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.1784255504608154,
      "learning_rate": 0.00019515943953580023,
      "loss": 2.088,
      "step": 141384
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.3258963823318481,
      "learning_rate": 0.00019515560690108917,
      "loss": 2.118,
      "step": 141385
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.0212725400924683,
      "learning_rate": 0.00019515177428587062,
      "loss": 2.0949,
      "step": 141386
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.0806024074554443,
      "learning_rate": 0.00019514794169014538,
      "loss": 2.1361,
      "step": 141387
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.4402748346328735,
      "learning_rate": 0.0001951441091139142,
      "loss": 2.0834,
      "step": 141388
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.2552865743637085,
      "learning_rate": 0.00019514027655717764,
      "loss": 2.0867,
      "step": 141389
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.0379735231399536,
      "learning_rate": 0.00019513644401993646,
      "loss": 1.9719,
      "step": 141390
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.1759215593338013,
      "learning_rate": 0.00019513261150219137,
      "loss": 2.275,
      "step": 141391
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.0586451292037964,
      "learning_rate": 0.00019512877900394315,
      "loss": 1.9077,
      "step": 141392
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.040686011314392,
      "learning_rate": 0.00019512494652519242,
      "loss": 2.1548,
      "step": 141393
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.0188791751861572,
      "learning_rate": 0.00019512111406593997,
      "loss": 2.0641,
      "step": 141394
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.0166432857513428,
      "learning_rate": 0.00019511728162618664,
      "loss": 1.965,
      "step": 141395
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.2297086715698242,
      "learning_rate": 0.00019511344920593283,
      "loss": 2.1823,
      "step": 141396
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.0565268993377686,
      "learning_rate": 0.00019510961680517937,
      "loss": 1.8937,
      "step": 141397
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.1762217283248901,
      "learning_rate": 0.00019510578442392707,
      "loss": 1.8835,
      "step": 141398
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.0576835870742798,
      "learning_rate": 0.00019510195206217656,
      "loss": 2.063,
      "step": 141399
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.0689319372177124,
      "learning_rate": 0.00019509811971992857,
      "loss": 1.8906,
      "step": 141400
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.0618785619735718,
      "learning_rate": 0.00019509428739718394,
      "loss": 2.2384,
      "step": 141401
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.103512167930603,
      "learning_rate": 0.00019509045509394317,
      "loss": 1.9617,
      "step": 141402
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.1716361045837402,
      "learning_rate": 0.00019508662281020705,
      "loss": 1.8789,
      "step": 141403
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.0799504518508911,
      "learning_rate": 0.0001950827905459763,
      "loss": 1.9584,
      "step": 141404
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.1929751634597778,
      "learning_rate": 0.00019507895830125164,
      "loss": 2.0277,
      "step": 141405
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.2069605588912964,
      "learning_rate": 0.0001950751260760338,
      "loss": 2.136,
      "step": 141406
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.22740638256073,
      "learning_rate": 0.00019507129387032356,
      "loss": 2.1186,
      "step": 141407
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.1034445762634277,
      "learning_rate": 0.00019506746168412146,
      "loss": 1.8826,
      "step": 141408
    },
    {
      "epoch": 1.84,
      "grad_norm": 0.9882621169090271,
      "learning_rate": 0.0001950636295174283,
      "loss": 2.126,
      "step": 141409
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.0979593992233276,
      "learning_rate": 0.0001950597973702448,
      "loss": 2.0203,
      "step": 141410
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.3423396348953247,
      "learning_rate": 0.00019505596524257163,
      "loss": 2.2865,
      "step": 141411
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.2111427783966064,
      "learning_rate": 0.00019505213313440953,
      "loss": 1.965,
      "step": 141412
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.0501049757003784,
      "learning_rate": 0.00019504830104575933,
      "loss": 1.995,
      "step": 141413
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.1305081844329834,
      "learning_rate": 0.00019504446897662152,
      "loss": 2.0417,
      "step": 141414
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.108886957168579,
      "learning_rate": 0.00019504063692699704,
      "loss": 2.0783,
      "step": 141415
    },
    {
      "epoch": 1.84,
      "grad_norm": 0.9982953667640686,
      "learning_rate": 0.0001950368048968864,
      "loss": 1.7374,
      "step": 141416
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.1272965669631958,
      "learning_rate": 0.00019503297288629037,
      "loss": 2.0167,
      "step": 141417
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.022150993347168,
      "learning_rate": 0.00019502914089520972,
      "loss": 2.1089,
      "step": 141418
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.1823301315307617,
      "learning_rate": 0.0001950253089236452,
      "loss": 2.0888,
      "step": 141419
    },
    {
      "epoch": 1.84,
      "grad_norm": 0.968123733997345,
      "learning_rate": 0.00019502147697159737,
      "loss": 1.997,
      "step": 141420
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.0116677284240723,
      "learning_rate": 0.00019501764503906705,
      "loss": 2.002,
      "step": 141421
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.043704628944397,
      "learning_rate": 0.000195013813126055,
      "loss": 1.7994,
      "step": 141422
    },
    {
      "epoch": 1.84,
      "grad_norm": 0.9258168935775757,
      "learning_rate": 0.0001950099812325618,
      "loss": 2.2284,
      "step": 141423
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.1434929370880127,
      "learning_rate": 0.00019500614935858822,
      "loss": 2.305,
      "step": 141424
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.1273516416549683,
      "learning_rate": 0.00019500231750413502,
      "loss": 1.9479,
      "step": 141425
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.1038457155227661,
      "learning_rate": 0.00019499848566920287,
      "loss": 1.9456,
      "step": 141426
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.227311134338379,
      "learning_rate": 0.00019499465385379244,
      "loss": 2.163,
      "step": 141427
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.052565574645996,
      "learning_rate": 0.00019499082205790444,
      "loss": 1.9569,
      "step": 141428
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.1133172512054443,
      "learning_rate": 0.00019498699028153976,
      "loss": 2.1635,
      "step": 141429
    },
    {
      "epoch": 1.84,
      "grad_norm": 0.9959226250648499,
      "learning_rate": 0.00019498315852469887,
      "loss": 1.9798,
      "step": 141430
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.123980164527893,
      "learning_rate": 0.0001949793267873827,
      "loss": 1.9678,
      "step": 141431
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.1861615180969238,
      "learning_rate": 0.00019497549506959178,
      "loss": 1.8887,
      "step": 141432
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.0536739826202393,
      "learning_rate": 0.0001949716633713269,
      "loss": 2.1138,
      "step": 141433
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.185594916343689,
      "learning_rate": 0.00019496783169258873,
      "loss": 1.9321,
      "step": 141434
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.130687952041626,
      "learning_rate": 0.00019496400003337808,
      "loss": 2.044,
      "step": 141435
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.0352193117141724,
      "learning_rate": 0.00019496016839369565,
      "loss": 2.1913,
      "step": 141436
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.2186535596847534,
      "learning_rate": 0.0001949563367735421,
      "loss": 2.0064,
      "step": 141437
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.0402886867523193,
      "learning_rate": 0.00019495250517291806,
      "loss": 2.1524,
      "step": 141438
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.1161550283432007,
      "learning_rate": 0.0001949486735918244,
      "loss": 2.1088,
      "step": 141439
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.0950164794921875,
      "learning_rate": 0.00019494484203026171,
      "loss": 1.9831,
      "step": 141440
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.0186216831207275,
      "learning_rate": 0.0001949410104882308,
      "loss": 1.7791,
      "step": 141441
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.0501528978347778,
      "learning_rate": 0.00019493717896573232,
      "loss": 2.0424,
      "step": 141442
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.0635156631469727,
      "learning_rate": 0.00019493334746276708,
      "loss": 2.0548,
      "step": 141443
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.193894386291504,
      "learning_rate": 0.0001949295159793356,
      "loss": 1.9954,
      "step": 141444
    },
    {
      "epoch": 1.84,
      "grad_norm": 0.9929872751235962,
      "learning_rate": 0.0001949256845154388,
      "loss": 1.9711,
      "step": 141445
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.190580129623413,
      "learning_rate": 0.0001949218530710772,
      "loss": 1.9766,
      "step": 141446
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.1773837804794312,
      "learning_rate": 0.00019491802164625164,
      "loss": 2.0922,
      "step": 141447
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.0757588148117065,
      "learning_rate": 0.00019491419024096282,
      "loss": 2.0019,
      "step": 141448
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.402506947517395,
      "learning_rate": 0.00019491035885521154,
      "loss": 2.186,
      "step": 141449
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.1334130764007568,
      "learning_rate": 0.0001949065274889983,
      "loss": 1.9705,
      "step": 141450
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.0212721824645996,
      "learning_rate": 0.0001949026961423239,
      "loss": 1.9932,
      "step": 141451
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.0625454187393188,
      "learning_rate": 0.00019489886481518907,
      "loss": 2.0547,
      "step": 141452
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.0530189275741577,
      "learning_rate": 0.00019489503350759456,
      "loss": 2.1244,
      "step": 141453
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.1008894443511963,
      "learning_rate": 0.000194891202219541,
      "loss": 2.0571,
      "step": 141454
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.1047381162643433,
      "learning_rate": 0.0001948873709510293,
      "loss": 1.8846,
      "step": 141455
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.1155877113342285,
      "learning_rate": 0.0001948835397020599,
      "loss": 2.0658,
      "step": 141456
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.064314365386963,
      "learning_rate": 0.00019487970847263366,
      "loss": 2.0255,
      "step": 141457
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.0140748023986816,
      "learning_rate": 0.00019487587726275125,
      "loss": 2.2196,
      "step": 141458
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.0594853162765503,
      "learning_rate": 0.00019487204607241337,
      "loss": 2.2572,
      "step": 141459
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.2153327465057373,
      "learning_rate": 0.00019486821490162078,
      "loss": 2.1837,
      "step": 141460
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.2265957593917847,
      "learning_rate": 0.00019486438375037414,
      "loss": 2.2065,
      "step": 141461
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.1224724054336548,
      "learning_rate": 0.0001948605526186744,
      "loss": 2.2127,
      "step": 141462
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.7634254693984985,
      "learning_rate": 0.00019485672150652188,
      "loss": 2.1778,
      "step": 141463
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.1249518394470215,
      "learning_rate": 0.00019485289041391748,
      "loss": 2.2268,
      "step": 141464
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.2259734869003296,
      "learning_rate": 0.00019484905934086193,
      "loss": 2.0242,
      "step": 141465
    },
    {
      "epoch": 1.84,
      "grad_norm": 0.9867518544197083,
      "learning_rate": 0.00019484522828735592,
      "loss": 2.1049,
      "step": 141466
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.048590064048767,
      "learning_rate": 0.00019484139725340016,
      "loss": 2.1192,
      "step": 141467
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.2524393796920776,
      "learning_rate": 0.0001948375662389955,
      "loss": 2.0091,
      "step": 141468
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.1241135597229004,
      "learning_rate": 0.00019483373524414242,
      "loss": 2.111,
      "step": 141469
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.0111323595046997,
      "learning_rate": 0.0001948299042688417,
      "loss": 2.3095,
      "step": 141470
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.155248999595642,
      "learning_rate": 0.00019482607331309412,
      "loss": 2.1893,
      "step": 141471
    },
    {
      "epoch": 1.84,
      "grad_norm": 0.9963315725326538,
      "learning_rate": 0.00019482224237690033,
      "loss": 1.9892,
      "step": 141472
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.0851647853851318,
      "learning_rate": 0.0001948184114602611,
      "loss": 2.0588,
      "step": 141473
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.068650722503662,
      "learning_rate": 0.0001948145805631772,
      "loss": 1.8609,
      "step": 141474
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.0156222581863403,
      "learning_rate": 0.00019481074968564913,
      "loss": 2.0734,
      "step": 141475
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.2025712728500366,
      "learning_rate": 0.00019480691882767777,
      "loss": 2.2087,
      "step": 141476
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.1195051670074463,
      "learning_rate": 0.00019480308798926373,
      "loss": 2.1522,
      "step": 141477
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.0266313552856445,
      "learning_rate": 0.00019479925717040783,
      "loss": 2.1008,
      "step": 141478
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.0522812604904175,
      "learning_rate": 0.00019479542637111072,
      "loss": 2.0752,
      "step": 141479
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.0651192665100098,
      "learning_rate": 0.00019479159559137327,
      "loss": 2.203,
      "step": 141480
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.0623934268951416,
      "learning_rate": 0.00019478776483119586,
      "loss": 2.1761,
      "step": 141481
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.0863672494888306,
      "learning_rate": 0.00019478393409057946,
      "loss": 2.0863,
      "step": 141482
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.115243673324585,
      "learning_rate": 0.0001947801033695247,
      "loss": 1.7893,
      "step": 141483
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.0988644361495972,
      "learning_rate": 0.00019477627266803227,
      "loss": 2.3452,
      "step": 141484
    },
    {
      "epoch": 1.84,
      "grad_norm": 0.9624807238578796,
      "learning_rate": 0.00019477244198610295,
      "loss": 1.8535,
      "step": 141485
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.0866403579711914,
      "learning_rate": 0.00019476861132373754,
      "loss": 2.1707,
      "step": 141486
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.0336724519729614,
      "learning_rate": 0.0001947647806809365,
      "loss": 2.0292,
      "step": 141487
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.0942021608352661,
      "learning_rate": 0.00019476095005770065,
      "loss": 2.0754,
      "step": 141488
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.1264033317565918,
      "learning_rate": 0.00019475711945403075,
      "loss": 2.1565,
      "step": 141489
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.175349473953247,
      "learning_rate": 0.00019475328886992753,
      "loss": 2.2924,
      "step": 141490
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.1594094038009644,
      "learning_rate": 0.00019474945830539158,
      "loss": 2.3226,
      "step": 141491
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.0999220609664917,
      "learning_rate": 0.00019474562776042385,
      "loss": 2.052,
      "step": 141492
    },
    {
      "epoch": 1.84,
      "grad_norm": 0.9888232946395874,
      "learning_rate": 0.00019474179723502478,
      "loss": 2.1027,
      "step": 141493
    },
    {
      "epoch": 1.84,
      "grad_norm": 0.9857472777366638,
      "learning_rate": 0.0001947379667291952,
      "loss": 2.105,
      "step": 141494
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.2842457294464111,
      "learning_rate": 0.0001947341362429358,
      "loss": 2.3564,
      "step": 141495
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.123237133026123,
      "learning_rate": 0.00019473030577624731,
      "loss": 2.238,
      "step": 141496
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.1802033185958862,
      "learning_rate": 0.00019472647532913049,
      "loss": 2.0926,
      "step": 141497
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.0868775844573975,
      "learning_rate": 0.000194722644901586,
      "loss": 2.1378,
      "step": 141498
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.1260979175567627,
      "learning_rate": 0.0001947188144936146,
      "loss": 1.9958,
      "step": 141499
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.0310118198394775,
      "learning_rate": 0.0001947149841052169,
      "loss": 2.1609,
      "step": 141500
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.090726375579834,
      "learning_rate": 0.00019471115373639368,
      "loss": 2.0773,
      "step": 141501
    },
    {
      "epoch": 1.84,
      "grad_norm": 0.9466152787208557,
      "learning_rate": 0.00019470732338714558,
      "loss": 2.1284,
      "step": 141502
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.0824626684188843,
      "learning_rate": 0.00019470349305747347,
      "loss": 2.0907,
      "step": 141503
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.153443455696106,
      "learning_rate": 0.00019469966274737795,
      "loss": 2.0409,
      "step": 141504
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.056982398033142,
      "learning_rate": 0.00019469583245685973,
      "loss": 2.0947,
      "step": 141505
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.2073321342468262,
      "learning_rate": 0.00019469200218591952,
      "loss": 2.3254,
      "step": 141506
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.236595869064331,
      "learning_rate": 0.0001946881719345581,
      "loss": 2.0065,
      "step": 141507
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.0768892765045166,
      "learning_rate": 0.00019468434170277614,
      "loss": 2.3154,
      "step": 141508
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.1816574335098267,
      "learning_rate": 0.0001946805114905743,
      "loss": 2.1259,
      "step": 141509
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.300355076789856,
      "learning_rate": 0.00019467668129795343,
      "loss": 2.0396,
      "step": 141510
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.2197532653808594,
      "learning_rate": 0.00019467285112491405,
      "loss": 2.3096,
      "step": 141511
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.2166461944580078,
      "learning_rate": 0.00019466902097145704,
      "loss": 1.8952,
      "step": 141512
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.1524980068206787,
      "learning_rate": 0.00019466519083758297,
      "loss": 2.0993,
      "step": 141513
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.2324020862579346,
      "learning_rate": 0.00019466136072329273,
      "loss": 1.9888,
      "step": 141514
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.3522112369537354,
      "learning_rate": 0.00019465753062858688,
      "loss": 2.0256,
      "step": 141515
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.1491740942001343,
      "learning_rate": 0.0001946537005534662,
      "loss": 2.1551,
      "step": 141516
    },
    {
      "epoch": 1.84,
      "grad_norm": 0.9623180627822876,
      "learning_rate": 0.0001946498704979314,
      "loss": 2.1088,
      "step": 141517
    },
    {
      "epoch": 1.84,
      "grad_norm": 0.9470949769020081,
      "learning_rate": 0.00019464604046198314,
      "loss": 2.1055,
      "step": 141518
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.1206519603729248,
      "learning_rate": 0.00019464221044562213,
      "loss": 2.0382,
      "step": 141519
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.1646682024002075,
      "learning_rate": 0.00019463838044884917,
      "loss": 2.1086,
      "step": 141520
    },
    {
      "epoch": 1.84,
      "grad_norm": 0.9617261290550232,
      "learning_rate": 0.00019463455047166498,
      "loss": 2.1846,
      "step": 141521
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.247161626815796,
      "learning_rate": 0.0001946307205140702,
      "loss": 2.0711,
      "step": 141522
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.2604050636291504,
      "learning_rate": 0.00019462689057606552,
      "loss": 2.039,
      "step": 141523
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.3703174591064453,
      "learning_rate": 0.00019462306065765164,
      "loss": 2.038,
      "step": 141524
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.0001574754714966,
      "learning_rate": 0.00019461923075882938,
      "loss": 2.0193,
      "step": 141525
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.0311700105667114,
      "learning_rate": 0.00019461540087959935,
      "loss": 2.3066,
      "step": 141526
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.0871878862380981,
      "learning_rate": 0.00019461157101996235,
      "loss": 2.0693,
      "step": 141527
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.0694071054458618,
      "learning_rate": 0.00019460774117991908,
      "loss": 2.1699,
      "step": 141528
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.2454179525375366,
      "learning_rate": 0.00019460391135947024,
      "loss": 2.1639,
      "step": 141529
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.0275157690048218,
      "learning_rate": 0.00019460008155861646,
      "loss": 1.9936,
      "step": 141530
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.0489568710327148,
      "learning_rate": 0.00019459625177735846,
      "loss": 2.0326,
      "step": 141531
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.0547306537628174,
      "learning_rate": 0.00019459242201569707,
      "loss": 2.1584,
      "step": 141532
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.006386399269104,
      "learning_rate": 0.00019458859227363293,
      "loss": 2.0418,
      "step": 141533
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.1734505891799927,
      "learning_rate": 0.00019458476255116675,
      "loss": 2.3332,
      "step": 141534
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.0743595361709595,
      "learning_rate": 0.00019458093284829936,
      "loss": 2.1184,
      "step": 141535
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.0793570280075073,
      "learning_rate": 0.00019457710316503126,
      "loss": 2.1106,
      "step": 141536
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.0585558414459229,
      "learning_rate": 0.00019457327350136324,
      "loss": 2.0769,
      "step": 141537
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.4692388772964478,
      "learning_rate": 0.0001945694438572961,
      "loss": 2.0138,
      "step": 141538
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.1026442050933838,
      "learning_rate": 0.00019456561423283043,
      "loss": 2.2686,
      "step": 141539
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.050432562828064,
      "learning_rate": 0.000194561784627967,
      "loss": 2.1011,
      "step": 141540
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.078658103942871,
      "learning_rate": 0.00019455795504270672,
      "loss": 2.1212,
      "step": 141541
    },
    {
      "epoch": 1.84,
      "grad_norm": 0.9746190905570984,
      "learning_rate": 0.00019455412547704991,
      "loss": 2.0311,
      "step": 141542
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.0131551027297974,
      "learning_rate": 0.00019455029593099752,
      "loss": 1.8058,
      "step": 141543
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.1691617965698242,
      "learning_rate": 0.00019454646640455018,
      "loss": 1.7901,
      "step": 141544
    },
    {
      "epoch": 1.84,
      "grad_norm": 0.9771712422370911,
      "learning_rate": 0.00019454263689770867,
      "loss": 1.937,
      "step": 141545
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.0076358318328857,
      "learning_rate": 0.00019453880741047367,
      "loss": 2.1434,
      "step": 141546
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.0168277025222778,
      "learning_rate": 0.00019453497794284604,
      "loss": 2.0802,
      "step": 141547
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.0566974878311157,
      "learning_rate": 0.00019453114849482617,
      "loss": 2.0146,
      "step": 141548
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.044219970703125,
      "learning_rate": 0.000194527319066415,
      "loss": 2.1227,
      "step": 141549
    },
    {
      "epoch": 1.84,
      "grad_norm": 0.9722912311553955,
      "learning_rate": 0.00019452348965761322,
      "loss": 2.0195,
      "step": 141550
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.1815868616104126,
      "learning_rate": 0.00019451966026842147,
      "loss": 2.0707,
      "step": 141551
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.174993872642517,
      "learning_rate": 0.0001945158308988405,
      "loss": 2.0784,
      "step": 141552
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.151543140411377,
      "learning_rate": 0.00019451200154887115,
      "loss": 2.0995,
      "step": 141553
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.0906665325164795,
      "learning_rate": 0.00019450817221851392,
      "loss": 1.9992,
      "step": 141554
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.04366135597229,
      "learning_rate": 0.00019450434290776956,
      "loss": 2.0266,
      "step": 141555
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.1405388116836548,
      "learning_rate": 0.00019450051361663887,
      "loss": 2.2837,
      "step": 141556
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.1556097269058228,
      "learning_rate": 0.0001944966843451225,
      "loss": 2.1506,
      "step": 141557
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.143937587738037,
      "learning_rate": 0.0001944928550932212,
      "loss": 2.1084,
      "step": 141558
    },
    {
      "epoch": 1.84,
      "grad_norm": 0.9988610744476318,
      "learning_rate": 0.00019448902586093584,
      "loss": 1.8943,
      "step": 141559
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.035549283027649,
      "learning_rate": 0.00019448519664826678,
      "loss": 2.0144,
      "step": 141560
    },
    {
      "epoch": 1.84,
      "grad_norm": 0.9836335778236389,
      "learning_rate": 0.00019448136745521495,
      "loss": 2.1348,
      "step": 141561
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.2198477983474731,
      "learning_rate": 0.00019447753828178102,
      "loss": 1.9591,
      "step": 141562
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.0711803436279297,
      "learning_rate": 0.00019447370912796565,
      "loss": 2.1457,
      "step": 141563
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.1067918539047241,
      "learning_rate": 0.00019446987999376965,
      "loss": 2.2821,
      "step": 141564
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.1171534061431885,
      "learning_rate": 0.00019446605087919383,
      "loss": 2.1,
      "step": 141565
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.0413225889205933,
      "learning_rate": 0.00019446222178423866,
      "loss": 2.1074,
      "step": 141566
    },
    {
      "epoch": 1.84,
      "grad_norm": 0.9528573155403137,
      "learning_rate": 0.00019445839270890492,
      "loss": 2.0679,
      "step": 141567
    },
    {
      "epoch": 1.84,
      "grad_norm": 0.9992067813873291,
      "learning_rate": 0.00019445456365319337,
      "loss": 2.0743,
      "step": 141568
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.0747461318969727,
      "learning_rate": 0.0001944507346171047,
      "loss": 2.4304,
      "step": 141569
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.1470855474472046,
      "learning_rate": 0.0001944469056006396,
      "loss": 1.8987,
      "step": 141570
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.0519193410873413,
      "learning_rate": 0.000194443076603799,
      "loss": 2.2189,
      "step": 141571
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.20698082447052,
      "learning_rate": 0.00019443924762658325,
      "loss": 2.0562,
      "step": 141572
    },
    {
      "epoch": 1.84,
      "grad_norm": 0.9902737736701965,
      "learning_rate": 0.00019443541866899326,
      "loss": 2.2777,
      "step": 141573
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.05643892288208,
      "learning_rate": 0.0001944315897310297,
      "loss": 2.0787,
      "step": 141574
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.1829475164413452,
      "learning_rate": 0.0001944277608126933,
      "loss": 2.1857,
      "step": 141575
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.0684130191802979,
      "learning_rate": 0.00019442393191398483,
      "loss": 1.9707,
      "step": 141576
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.0881097316741943,
      "learning_rate": 0.00019442010303490502,
      "loss": 1.9005,
      "step": 141577
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.5903618335723877,
      "learning_rate": 0.0001944162741754544,
      "loss": 1.9213,
      "step": 141578
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.1655972003936768,
      "learning_rate": 0.00019441244533563375,
      "loss": 2.0502,
      "step": 141579
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.1698651313781738,
      "learning_rate": 0.00019440861651544384,
      "loss": 2.1066,
      "step": 141580
    },
    {
      "epoch": 1.84,
      "grad_norm": 0.9598322510719299,
      "learning_rate": 0.00019440478771488536,
      "loss": 1.98,
      "step": 141581
    },
    {
      "epoch": 1.84,
      "grad_norm": 0.9990513324737549,
      "learning_rate": 0.00019440095893395902,
      "loss": 2.0187,
      "step": 141582
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.2184525728225708,
      "learning_rate": 0.0001943971301726656,
      "loss": 1.9305,
      "step": 141583
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.0431429147720337,
      "learning_rate": 0.00019439330143100576,
      "loss": 1.9656,
      "step": 141584
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.0020493268966675,
      "learning_rate": 0.00019438947270898014,
      "loss": 2.0454,
      "step": 141585
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.0528802871704102,
      "learning_rate": 0.0001943856440065895,
      "loss": 2.192,
      "step": 141586
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.0862219333648682,
      "learning_rate": 0.00019438181532383455,
      "loss": 2.1271,
      "step": 141587
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.0485833883285522,
      "learning_rate": 0.00019437798666071605,
      "loss": 2.1061,
      "step": 141588
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.1314263343811035,
      "learning_rate": 0.0001943741580172347,
      "loss": 2.1149,
      "step": 141589
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.0709272623062134,
      "learning_rate": 0.00019437032939339111,
      "loss": 2.2735,
      "step": 141590
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.1281284093856812,
      "learning_rate": 0.00019436650078918618,
      "loss": 1.9805,
      "step": 141591
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.2860338687896729,
      "learning_rate": 0.00019436267220462045,
      "loss": 2.1084,
      "step": 141592
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.0689506530761719,
      "learning_rate": 0.00019435884363969464,
      "loss": 2.1077,
      "step": 141593
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.1419501304626465,
      "learning_rate": 0.0001943550150944096,
      "loss": 2.2291,
      "step": 141594
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.3512258529663086,
      "learning_rate": 0.00019435118656876586,
      "loss": 2.0477,
      "step": 141595
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.153026819229126,
      "learning_rate": 0.00019434735806276438,
      "loss": 2.1853,
      "step": 141596
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.0540508031845093,
      "learning_rate": 0.00019434352957640557,
      "loss": 2.1939,
      "step": 141597
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.0519368648529053,
      "learning_rate": 0.00019433970110969043,
      "loss": 1.8567,
      "step": 141598
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.1754134893417358,
      "learning_rate": 0.00019433587266261943,
      "loss": 2.1545,
      "step": 141599
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.0605623722076416,
      "learning_rate": 0.0001943320442351934,
      "loss": 2.2154,
      "step": 141600
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.226808786392212,
      "learning_rate": 0.00019432821582741303,
      "loss": 2.1199,
      "step": 141601
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.0139338970184326,
      "learning_rate": 0.00019432438743927912,
      "loss": 1.9938,
      "step": 141602
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.1149358749389648,
      "learning_rate": 0.00019432055907079223,
      "loss": 2.2592,
      "step": 141603
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.2548185586929321,
      "learning_rate": 0.00019431673072195315,
      "loss": 2.1741,
      "step": 141604
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.1372874975204468,
      "learning_rate": 0.00019431290239276257,
      "loss": 2.1861,
      "step": 141605
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.0825862884521484,
      "learning_rate": 0.00019430907408322126,
      "loss": 2.1123,
      "step": 141606
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.136855959892273,
      "learning_rate": 0.00019430524579332987,
      "loss": 2.1349,
      "step": 141607
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.1287510395050049,
      "learning_rate": 0.00019430141752308915,
      "loss": 2.152,
      "step": 141608
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.137711763381958,
      "learning_rate": 0.00019429758927249973,
      "loss": 2.1861,
      "step": 141609
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.0782338380813599,
      "learning_rate": 0.00019429376104156243,
      "loss": 2.1721,
      "step": 141610
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.1523151397705078,
      "learning_rate": 0.0001942899328302779,
      "loss": 2.2615,
      "step": 141611
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.073702096939087,
      "learning_rate": 0.00019428610463864682,
      "loss": 1.8906,
      "step": 141612
    },
    {
      "epoch": 1.84,
      "grad_norm": 0.9883857369422913,
      "learning_rate": 0.00019428227646667008,
      "loss": 2.1458,
      "step": 141613
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.0084336996078491,
      "learning_rate": 0.0001942784483143482,
      "loss": 2.0649,
      "step": 141614
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.142733097076416,
      "learning_rate": 0.00019427462018168195,
      "loss": 2.3043,
      "step": 141615
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.0135670900344849,
      "learning_rate": 0.00019427079206867198,
      "loss": 2.1247,
      "step": 141616
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.0728167295455933,
      "learning_rate": 0.00019426696397531912,
      "loss": 2.125,
      "step": 141617
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.149383306503296,
      "learning_rate": 0.00019426313590162399,
      "loss": 2.087,
      "step": 141618
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.0659722089767456,
      "learning_rate": 0.00019425930784758733,
      "loss": 2.0873,
      "step": 141619
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.0714716911315918,
      "learning_rate": 0.00019425547981321003,
      "loss": 1.9288,
      "step": 141620
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.348684549331665,
      "learning_rate": 0.00019425165179849248,
      "loss": 1.9295,
      "step": 141621
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.2674548625946045,
      "learning_rate": 0.00019424782380343556,
      "loss": 2.2268,
      "step": 141622
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.1984301805496216,
      "learning_rate": 0.00019424399582803996,
      "loss": 2.1276,
      "step": 141623
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.031137228012085,
      "learning_rate": 0.00019424016787230637,
      "loss": 1.9701,
      "step": 141624
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.2292006015777588,
      "learning_rate": 0.00019423633993623558,
      "loss": 2.3973,
      "step": 141625
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.7183459997177124,
      "learning_rate": 0.00019423251201982833,
      "loss": 2.0268,
      "step": 141626
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.0420085191726685,
      "learning_rate": 0.00019422868412308515,
      "loss": 2.1123,
      "step": 141627
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.0916645526885986,
      "learning_rate": 0.00019422485624600683,
      "loss": 2.2042,
      "step": 141628
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.0561150312423706,
      "learning_rate": 0.0001942210283885941,
      "loss": 2.094,
      "step": 141629
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.1791220903396606,
      "learning_rate": 0.00019421720055084774,
      "loss": 2.265,
      "step": 141630
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.5533369779586792,
      "learning_rate": 0.00019421337273276833,
      "loss": 2.3158,
      "step": 141631
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.2490540742874146,
      "learning_rate": 0.00019420954493435682,
      "loss": 2.2189,
      "step": 141632
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.1166162490844727,
      "learning_rate": 0.00019420571715561365,
      "loss": 2.2444,
      "step": 141633
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.1613320112228394,
      "learning_rate": 0.00019420188939653956,
      "loss": 2.0995,
      "step": 141634
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.2107062339782715,
      "learning_rate": 0.0001941980616571354,
      "loss": 2.0065,
      "step": 141635
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.028719425201416,
      "learning_rate": 0.00019419423393740178,
      "loss": 2.0686,
      "step": 141636
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.0153783559799194,
      "learning_rate": 0.00019419040623733949,
      "loss": 1.8827,
      "step": 141637
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.1539767980575562,
      "learning_rate": 0.00019418657855694926,
      "loss": 2.2805,
      "step": 141638
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.1571695804595947,
      "learning_rate": 0.00019418275089623166,
      "loss": 2.2689,
      "step": 141639
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.3081692457199097,
      "learning_rate": 0.00019417892325518748,
      "loss": 2.2022,
      "step": 141640
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.344223141670227,
      "learning_rate": 0.00019417509563381745,
      "loss": 2.209,
      "step": 141641
    },
    {
      "epoch": 1.84,
      "grad_norm": 0.9776050448417664,
      "learning_rate": 0.00019417126803212222,
      "loss": 2.0317,
      "step": 141642
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.123705506324768,
      "learning_rate": 0.0001941674404501026,
      "loss": 2.1517,
      "step": 141643
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.3369289636611938,
      "learning_rate": 0.00019416361288775933,
      "loss": 1.9843,
      "step": 141644
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.1636866331100464,
      "learning_rate": 0.00019415978534509298,
      "loss": 1.9684,
      "step": 141645
    },
    {
      "epoch": 1.84,
      "grad_norm": 0.9733678102493286,
      "learning_rate": 0.0001941559578221043,
      "loss": 1.8717,
      "step": 141646
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.1042081117630005,
      "learning_rate": 0.00019415213031879398,
      "loss": 2.2933,
      "step": 141647
    },
    {
      "epoch": 1.84,
      "grad_norm": 0.9936166405677795,
      "learning_rate": 0.00019414830283516282,
      "loss": 2.0988,
      "step": 141648
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.2716542482376099,
      "learning_rate": 0.00019414447537121146,
      "loss": 1.9101,
      "step": 141649
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.0418213605880737,
      "learning_rate": 0.00019414064792694081,
      "loss": 2.1723,
      "step": 141650
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.5093542337417603,
      "learning_rate": 0.00019413682050235126,
      "loss": 1.9718,
      "step": 141651
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.062915325164795,
      "learning_rate": 0.00019413299309744368,
      "loss": 2.1511,
      "step": 141652
    },
    {
      "epoch": 1.84,
      "grad_norm": 0.96822589635849,
      "learning_rate": 0.00019412916571221877,
      "loss": 1.8786,
      "step": 141653
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.1060328483581543,
      "learning_rate": 0.00019412533834667722,
      "loss": 2.4069,
      "step": 141654
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.0684012174606323,
      "learning_rate": 0.0001941215110008198,
      "loss": 1.986,
      "step": 141655
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.0109261274337769,
      "learning_rate": 0.00019411768367464733,
      "loss": 2.0529,
      "step": 141656
    },
    {
      "epoch": 1.84,
      "grad_norm": 0.9818038940429688,
      "learning_rate": 0.0001941138563681602,
      "loss": 2.0335,
      "step": 141657
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.0690115690231323,
      "learning_rate": 0.0001941100290813594,
      "loss": 2.1666,
      "step": 141658
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.1078314781188965,
      "learning_rate": 0.00019410620181424545,
      "loss": 2.0294,
      "step": 141659
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.0175398588180542,
      "learning_rate": 0.00019410237456681916,
      "loss": 2.3108,
      "step": 141660
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.0504133701324463,
      "learning_rate": 0.0001940985473390813,
      "loss": 2.3047,
      "step": 141661
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.0590931177139282,
      "learning_rate": 0.00019409472013103249,
      "loss": 2.2004,
      "step": 141662
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.091944694519043,
      "learning_rate": 0.00019409089294267356,
      "loss": 2.0211,
      "step": 141663
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.0100489854812622,
      "learning_rate": 0.00019408706577400504,
      "loss": 1.9934,
      "step": 141664
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.062195897102356,
      "learning_rate": 0.00019408323862502776,
      "loss": 2.1089,
      "step": 141665
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.1525009870529175,
      "learning_rate": 0.00019407941149574234,
      "loss": 2.1498,
      "step": 141666
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.063984751701355,
      "learning_rate": 0.0001940755843861496,
      "loss": 1.9952,
      "step": 141667
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.0451525449752808,
      "learning_rate": 0.0001940717572962502,
      "loss": 2.1291,
      "step": 141668
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.0571320056915283,
      "learning_rate": 0.00019406793022604495,
      "loss": 2.1605,
      "step": 141669
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.1134305000305176,
      "learning_rate": 0.0001940641031755344,
      "loss": 2.1007,
      "step": 141670
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.0232107639312744,
      "learning_rate": 0.00019406027614471936,
      "loss": 2.047,
      "step": 141671
    },
    {
      "epoch": 1.84,
      "grad_norm": 0.9963330626487732,
      "learning_rate": 0.00019405644913360044,
      "loss": 2.1315,
      "step": 141672
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.055241584777832,
      "learning_rate": 0.00019405262214217844,
      "loss": 2.1407,
      "step": 141673
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.136803150177002,
      "learning_rate": 0.0001940487951704541,
      "loss": 2.0988,
      "step": 141674
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.0800343751907349,
      "learning_rate": 0.0001940449682184281,
      "loss": 2.2083,
      "step": 141675
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.0243260860443115,
      "learning_rate": 0.0001940411412861012,
      "loss": 2.1342,
      "step": 141676
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.135280966758728,
      "learning_rate": 0.00019403731437347396,
      "loss": 2.2122,
      "step": 141677
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.096728801727295,
      "learning_rate": 0.00019403348748054714,
      "loss": 2.2916,
      "step": 141678
    },
    {
      "epoch": 1.84,
      "grad_norm": 0.9837190508842468,
      "learning_rate": 0.00019402966060732158,
      "loss": 2.0989,
      "step": 141679
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.0242652893066406,
      "learning_rate": 0.00019402583375379784,
      "loss": 2.0837,
      "step": 141680
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.0866565704345703,
      "learning_rate": 0.00019402200691997676,
      "loss": 1.9859,
      "step": 141681
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.0200703144073486,
      "learning_rate": 0.00019401818010585895,
      "loss": 2.1173,
      "step": 141682
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.1849324703216553,
      "learning_rate": 0.00019401435331144523,
      "loss": 1.9923,
      "step": 141683
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.1103733777999878,
      "learning_rate": 0.0001940105265367362,
      "loss": 2.1271,
      "step": 141684
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.115473985671997,
      "learning_rate": 0.00019400669978173255,
      "loss": 2.1004,
      "step": 141685
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.1628793478012085,
      "learning_rate": 0.0001940028730464351,
      "loss": 2.1219,
      "step": 141686
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.0383141040802002,
      "learning_rate": 0.00019399904633084455,
      "loss": 2.0834,
      "step": 141687
    },
    {
      "epoch": 1.84,
      "grad_norm": 0.9848306179046631,
      "learning_rate": 0.00019399521963496156,
      "loss": 2.1652,
      "step": 141688
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.0954794883728027,
      "learning_rate": 0.00019399139295878681,
      "loss": 2.1567,
      "step": 141689
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.2767407894134521,
      "learning_rate": 0.00019398756630232116,
      "loss": 2.0667,
      "step": 141690
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.190871238708496,
      "learning_rate": 0.00019398373966556515,
      "loss": 1.9991,
      "step": 141691
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.1237744092941284,
      "learning_rate": 0.00019397991304851954,
      "loss": 2.1589,
      "step": 141692
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.094241976737976,
      "learning_rate": 0.0001939760864511852,
      "loss": 2.0977,
      "step": 141693
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.0130470991134644,
      "learning_rate": 0.00019397225987356255,
      "loss": 2.1016,
      "step": 141694
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.319238305091858,
      "learning_rate": 0.00019396843331565252,
      "loss": 2.1283,
      "step": 141695
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.3611412048339844,
      "learning_rate": 0.0001939646067774557,
      "loss": 1.9871,
      "step": 141696
    },
    {
      "epoch": 1.84,
      "grad_norm": 1.1574902534484863,
      "learning_rate": 0.00019396078025897293,
      "loss": 2.2317,
      "step": 141697
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.214375376701355,
      "learning_rate": 0.00019395695376020489,
      "loss": 2.3382,
      "step": 141698
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.0754852294921875,
      "learning_rate": 0.00019395312728115224,
      "loss": 2.1268,
      "step": 141699
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.0803700685501099,
      "learning_rate": 0.00019394930082181565,
      "loss": 2.1254,
      "step": 141700
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.067448616027832,
      "learning_rate": 0.0001939454743821959,
      "loss": 2.3251,
      "step": 141701
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.0807552337646484,
      "learning_rate": 0.0001939416479622937,
      "loss": 2.0919,
      "step": 141702
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.142870306968689,
      "learning_rate": 0.00019393782156210972,
      "loss": 2.1964,
      "step": 141703
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.0263335704803467,
      "learning_rate": 0.00019393399518164472,
      "loss": 2.0741,
      "step": 141704
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.0683276653289795,
      "learning_rate": 0.0001939301688208995,
      "loss": 2.1171,
      "step": 141705
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.1326100826263428,
      "learning_rate": 0.00019392634247987456,
      "loss": 2.1471,
      "step": 141706
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.2208045721054077,
      "learning_rate": 0.0001939225161585707,
      "loss": 1.9368,
      "step": 141707
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.2636834383010864,
      "learning_rate": 0.00019391868985698865,
      "loss": 1.9394,
      "step": 141708
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.0093220472335815,
      "learning_rate": 0.0001939148635751291,
      "loss": 2.161,
      "step": 141709
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.262144923210144,
      "learning_rate": 0.00019391103731299282,
      "loss": 2.033,
      "step": 141710
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.1038649082183838,
      "learning_rate": 0.00019390721107058058,
      "loss": 1.9164,
      "step": 141711
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.506172776222229,
      "learning_rate": 0.00019390338484789286,
      "loss": 2.0497,
      "step": 141712
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.433019757270813,
      "learning_rate": 0.0001938995586449305,
      "loss": 2.2439,
      "step": 141713
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.0325162410736084,
      "learning_rate": 0.00019389573246169424,
      "loss": 2.2243,
      "step": 141714
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.0987000465393066,
      "learning_rate": 0.00019389190629818477,
      "loss": 2.1467,
      "step": 141715
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.2643446922302246,
      "learning_rate": 0.00019388808015440278,
      "loss": 2.2935,
      "step": 141716
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.6785122156143188,
      "learning_rate": 0.00019388425403034913,
      "loss": 1.9311,
      "step": 141717
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.4713736772537231,
      "learning_rate": 0.00019388042792602424,
      "loss": 2.1406,
      "step": 141718
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.1285061836242676,
      "learning_rate": 0.00019387660184142901,
      "loss": 2.3583,
      "step": 141719
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.1478952169418335,
      "learning_rate": 0.00019387277577656415,
      "loss": 2.0964,
      "step": 141720
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.225039005279541,
      "learning_rate": 0.0001938689497314303,
      "loss": 1.8637,
      "step": 141721
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.3406927585601807,
      "learning_rate": 0.00019386512370602826,
      "loss": 2.0872,
      "step": 141722
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.283957600593567,
      "learning_rate": 0.00019386129770035878,
      "loss": 2.2412,
      "step": 141723
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.0259507894515991,
      "learning_rate": 0.00019385747171442238,
      "loss": 2.1539,
      "step": 141724
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.0539312362670898,
      "learning_rate": 0.00019385364574821986,
      "loss": 1.9081,
      "step": 141725
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.0550669431686401,
      "learning_rate": 0.000193849819801752,
      "loss": 1.9731,
      "step": 141726
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.324787974357605,
      "learning_rate": 0.00019384599387501945,
      "loss": 2.0221,
      "step": 141727
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.3108909130096436,
      "learning_rate": 0.00019384216796802288,
      "loss": 2.1586,
      "step": 141728
    },
    {
      "epoch": 1.85,
      "grad_norm": 0.9871777296066284,
      "learning_rate": 0.00019383834208076313,
      "loss": 2.0426,
      "step": 141729
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.042241096496582,
      "learning_rate": 0.0001938345162132409,
      "loss": 2.1343,
      "step": 141730
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.4651329517364502,
      "learning_rate": 0.00019383069036545676,
      "loss": 2.3798,
      "step": 141731
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.0339093208312988,
      "learning_rate": 0.00019382686453741147,
      "loss": 2.1823,
      "step": 141732
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.0355533361434937,
      "learning_rate": 0.00019382303872910576,
      "loss": 2.0304,
      "step": 141733
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.0666264295578003,
      "learning_rate": 0.00019381921294054036,
      "loss": 1.9881,
      "step": 141734
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.0339876413345337,
      "learning_rate": 0.00019381538717171598,
      "loss": 2.1592,
      "step": 141735
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.095652461051941,
      "learning_rate": 0.00019381156142263343,
      "loss": 2.4013,
      "step": 141736
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.0770115852355957,
      "learning_rate": 0.00019380773569329323,
      "loss": 2.1074,
      "step": 141737
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.1551696062088013,
      "learning_rate": 0.00019380390998369613,
      "loss": 2.1082,
      "step": 141738
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.0062711238861084,
      "learning_rate": 0.00019380008429384292,
      "loss": 2.2236,
      "step": 141739
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.0739516019821167,
      "learning_rate": 0.00019379625862373428,
      "loss": 1.8881,
      "step": 141740
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.0512795448303223,
      "learning_rate": 0.00019379243297337092,
      "loss": 2.092,
      "step": 141741
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.0837092399597168,
      "learning_rate": 0.00019378860734275364,
      "loss": 2.0322,
      "step": 141742
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.2961299419403076,
      "learning_rate": 0.000193784781731883,
      "loss": 2.0971,
      "step": 141743
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.0639294385910034,
      "learning_rate": 0.00019378095614075972,
      "loss": 2.1132,
      "step": 141744
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.1138304471969604,
      "learning_rate": 0.00019377713056938458,
      "loss": 1.9214,
      "step": 141745
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.2292509078979492,
      "learning_rate": 0.0001937733050177583,
      "loss": 2.0368,
      "step": 141746
    },
    {
      "epoch": 1.85,
      "grad_norm": 0.9863492846488953,
      "learning_rate": 0.00019376947948588155,
      "loss": 1.7663,
      "step": 141747
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.0292057991027832,
      "learning_rate": 0.00019376565397375515,
      "loss": 2.0249,
      "step": 141748
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.0530946254730225,
      "learning_rate": 0.00019376182848137964,
      "loss": 1.9557,
      "step": 141749
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.0325926542282104,
      "learning_rate": 0.0001937580030087558,
      "loss": 2.2437,
      "step": 141750
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.0931612253189087,
      "learning_rate": 0.00019375417755588436,
      "loss": 2.1346,
      "step": 141751
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.0639312267303467,
      "learning_rate": 0.000193750352122766,
      "loss": 2.0808,
      "step": 141752
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.1003482341766357,
      "learning_rate": 0.0001937465267094015,
      "loss": 2.0702,
      "step": 141753
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.21333909034729,
      "learning_rate": 0.0001937427013157916,
      "loss": 2.0187,
      "step": 141754
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.0247688293457031,
      "learning_rate": 0.00019373887594193684,
      "loss": 2.2246,
      "step": 141755
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.0580612421035767,
      "learning_rate": 0.000193735050587838,
      "loss": 1.9049,
      "step": 141756
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.1661189794540405,
      "learning_rate": 0.00019373122525349586,
      "loss": 2.1296,
      "step": 141757
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.0794376134872437,
      "learning_rate": 0.00019372739993891107,
      "loss": 1.9663,
      "step": 141758
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.0260725021362305,
      "learning_rate": 0.00019372357464408438,
      "loss": 2.1461,
      "step": 141759
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.012981653213501,
      "learning_rate": 0.0001937197493690166,
      "loss": 1.9777,
      "step": 141760
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.1219167709350586,
      "learning_rate": 0.00019371592411370817,
      "loss": 2.0991,
      "step": 141761
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.153743028640747,
      "learning_rate": 0.00019371209887816,
      "loss": 2.1989,
      "step": 141762
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.080002784729004,
      "learning_rate": 0.00019370827366237272,
      "loss": 1.9724,
      "step": 141763
    },
    {
      "epoch": 1.85,
      "grad_norm": 0.9968612194061279,
      "learning_rate": 0.00019370444846634712,
      "loss": 2.0223,
      "step": 141764
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.08289635181427,
      "learning_rate": 0.00019370062329008384,
      "loss": 2.2185,
      "step": 141765
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.0367131233215332,
      "learning_rate": 0.0001936967981335837,
      "loss": 2.0778,
      "step": 141766
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.029571771621704,
      "learning_rate": 0.00019369297299684727,
      "loss": 1.9311,
      "step": 141767
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.0874426364898682,
      "learning_rate": 0.00019368914787987537,
      "loss": 2.2521,
      "step": 141768
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.0054798126220703,
      "learning_rate": 0.0001936853227826686,
      "loss": 2.0733,
      "step": 141769
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.0313405990600586,
      "learning_rate": 0.00019368149770522774,
      "loss": 2.0708,
      "step": 141770
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.0774950981140137,
      "learning_rate": 0.00019367767264755347,
      "loss": 2.0002,
      "step": 141771
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.013654112815857,
      "learning_rate": 0.00019367384760964667,
      "loss": 1.8728,
      "step": 141772
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.1264233589172363,
      "learning_rate": 0.00019367002259150777,
      "loss": 2.0782,
      "step": 141773
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.3043385744094849,
      "learning_rate": 0.00019366619759313765,
      "loss": 2.0003,
      "step": 141774
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.0974007844924927,
      "learning_rate": 0.00019366237261453703,
      "loss": 1.9966,
      "step": 141775
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.1541547775268555,
      "learning_rate": 0.00019365854765570655,
      "loss": 2.1868,
      "step": 141776
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.0208121538162231,
      "learning_rate": 0.00019365472271664692,
      "loss": 2.1167,
      "step": 141777
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.2887479066848755,
      "learning_rate": 0.00019365089779735896,
      "loss": 1.8558,
      "step": 141778
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.0560181140899658,
      "learning_rate": 0.00019364707289784328,
      "loss": 2.1281,
      "step": 141779
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.026427984237671,
      "learning_rate": 0.0001936432480181006,
      "loss": 2.1873,
      "step": 141780
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.075822114944458,
      "learning_rate": 0.00019363942315813164,
      "loss": 2.1618,
      "step": 141781
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.0252043008804321,
      "learning_rate": 0.00019363559831793713,
      "loss": 2.0334,
      "step": 141782
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.0578988790512085,
      "learning_rate": 0.0001936317734975178,
      "loss": 2.3834,
      "step": 141783
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.030685305595398,
      "learning_rate": 0.00019362794869687435,
      "loss": 2.117,
      "step": 141784
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.0513639450073242,
      "learning_rate": 0.00019362412391600738,
      "loss": 1.9944,
      "step": 141785
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.0726146697998047,
      "learning_rate": 0.00019362029915491775,
      "loss": 2.0755,
      "step": 141786
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.072106122970581,
      "learning_rate": 0.00019361647441360608,
      "loss": 2.1394,
      "step": 141787
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.0605134963989258,
      "learning_rate": 0.00019361264969207313,
      "loss": 2.0931,
      "step": 141788
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.0856176614761353,
      "learning_rate": 0.0001936088249903196,
      "loss": 1.8483,
      "step": 141789
    },
    {
      "epoch": 1.85,
      "grad_norm": 0.9086353182792664,
      "learning_rate": 0.0001936050003083463,
      "loss": 1.9288,
      "step": 141790
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.209546446800232,
      "learning_rate": 0.00019360117564615371,
      "loss": 2.2378,
      "step": 141791
    },
    {
      "epoch": 1.85,
      "grad_norm": 0.9864116907119751,
      "learning_rate": 0.0001935973510037427,
      "loss": 1.8869,
      "step": 141792
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.0702131986618042,
      "learning_rate": 0.0001935935263811139,
      "loss": 2.0932,
      "step": 141793
    },
    {
      "epoch": 1.85,
      "grad_norm": 0.9452314972877502,
      "learning_rate": 0.00019358970177826813,
      "loss": 1.9622,
      "step": 141794
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.207611322402954,
      "learning_rate": 0.000193585877195206,
      "loss": 1.9536,
      "step": 141795
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.1675294637680054,
      "learning_rate": 0.0001935820526319283,
      "loss": 2.2144,
      "step": 141796
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.0273100137710571,
      "learning_rate": 0.00019357822808843583,
      "loss": 2.1431,
      "step": 141797
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.0812559127807617,
      "learning_rate": 0.00019357440356472904,
      "loss": 2.1306,
      "step": 141798
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.0655823945999146,
      "learning_rate": 0.00019357057906080875,
      "loss": 1.9877,
      "step": 141799
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.0897767543792725,
      "learning_rate": 0.00019356675457667574,
      "loss": 2.1136,
      "step": 141800
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.129043459892273,
      "learning_rate": 0.00019356293011233065,
      "loss": 2.1473,
      "step": 141801
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.138736367225647,
      "learning_rate": 0.00019355910566777425,
      "loss": 2.1446,
      "step": 141802
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.2436012029647827,
      "learning_rate": 0.0001935552812430073,
      "loss": 2.0456,
      "step": 141803
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.0633087158203125,
      "learning_rate": 0.00019355145683803033,
      "loss": 2.0068,
      "step": 141804
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.0054694414138794,
      "learning_rate": 0.0001935476324528442,
      "loss": 1.8665,
      "step": 141805
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.084289789199829,
      "learning_rate": 0.00019354380808744948,
      "loss": 2.0674,
      "step": 141806
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.0221718549728394,
      "learning_rate": 0.00019353998374184705,
      "loss": 2.1215,
      "step": 141807
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.1217162609100342,
      "learning_rate": 0.00019353615941603754,
      "loss": 2.0648,
      "step": 141808
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.0398000478744507,
      "learning_rate": 0.0001935323351100218,
      "loss": 2.0318,
      "step": 141809
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.0348368883132935,
      "learning_rate": 0.00019352851082380022,
      "loss": 2.0213,
      "step": 141810
    },
    {
      "epoch": 1.85,
      "grad_norm": 0.9764915704727173,
      "learning_rate": 0.00019352468655737376,
      "loss": 1.8074,
      "step": 141811
    },
    {
      "epoch": 1.85,
      "grad_norm": 0.9956187605857849,
      "learning_rate": 0.00019352086231074305,
      "loss": 1.9701,
      "step": 141812
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.1638051271438599,
      "learning_rate": 0.0001935170380839088,
      "loss": 2.1304,
      "step": 141813
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.1336599588394165,
      "learning_rate": 0.0001935132138768718,
      "loss": 2.0247,
      "step": 141814
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.0943408012390137,
      "learning_rate": 0.0001935093896896328,
      "loss": 2.1592,
      "step": 141815
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.1439954042434692,
      "learning_rate": 0.0001935055655221923,
      "loss": 2.1913,
      "step": 141816
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.1951026916503906,
      "learning_rate": 0.00019350174137455108,
      "loss": 2.2147,
      "step": 141817
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.099654197692871,
      "learning_rate": 0.00019349791724670995,
      "loss": 1.9644,
      "step": 141818
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.085689663887024,
      "learning_rate": 0.00019349409313866952,
      "loss": 1.858,
      "step": 141819
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.0544569492340088,
      "learning_rate": 0.00019349026905043058,
      "loss": 2.129,
      "step": 141820
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.1262673139572144,
      "learning_rate": 0.00019348644498199394,
      "loss": 2.0395,
      "step": 141821
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.1335488557815552,
      "learning_rate": 0.00019348262093336004,
      "loss": 2.0356,
      "step": 141822
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.3575196266174316,
      "learning_rate": 0.00019347879690452974,
      "loss": 2.141,
      "step": 141823
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.1377978324890137,
      "learning_rate": 0.0001934749728955037,
      "loss": 2.152,
      "step": 141824
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.0853972434997559,
      "learning_rate": 0.00019347114890628275,
      "loss": 2.0786,
      "step": 141825
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.268581748008728,
      "learning_rate": 0.0001934673249368675,
      "loss": 2.2057,
      "step": 141826
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.317912220954895,
      "learning_rate": 0.00019346350098725876,
      "loss": 2.3125,
      "step": 141827
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.125643014907837,
      "learning_rate": 0.00019345967705745707,
      "loss": 2.2181,
      "step": 141828
    },
    {
      "epoch": 1.85,
      "grad_norm": 0.9892499446868896,
      "learning_rate": 0.00019345585314746328,
      "loss": 2.1098,
      "step": 141829
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.3297524452209473,
      "learning_rate": 0.00019345202925727801,
      "loss": 2.291,
      "step": 141830
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.0723373889923096,
      "learning_rate": 0.00019344820538690202,
      "loss": 2.1225,
      "step": 141831
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.2035361528396606,
      "learning_rate": 0.00019344438153633603,
      "loss": 2.1016,
      "step": 141832
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.048992395401001,
      "learning_rate": 0.00019344055770558087,
      "loss": 2.0353,
      "step": 141833
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.126989483833313,
      "learning_rate": 0.000193436733894637,
      "loss": 2.147,
      "step": 141834
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.1141529083251953,
      "learning_rate": 0.00019343291010350523,
      "loss": 1.9815,
      "step": 141835
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.1150575876235962,
      "learning_rate": 0.00019342908633218633,
      "loss": 2.1469,
      "step": 141836
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.0099148750305176,
      "learning_rate": 0.00019342526258068097,
      "loss": 1.9744,
      "step": 141837
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.0026391744613647,
      "learning_rate": 0.00019342143884898983,
      "loss": 2.0225,
      "step": 141838
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.0341229438781738,
      "learning_rate": 0.00019341761513711378,
      "loss": 2.0974,
      "step": 141839
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.026881456375122,
      "learning_rate": 0.00019341379144505332,
      "loss": 1.8946,
      "step": 141840
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.0357000827789307,
      "learning_rate": 0.00019340996777280923,
      "loss": 2.1198,
      "step": 141841
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.1867258548736572,
      "learning_rate": 0.00019340614412038227,
      "loss": 2.219,
      "step": 141842
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.0993355512619019,
      "learning_rate": 0.00019340232048777305,
      "loss": 2.1416,
      "step": 141843
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.043899655342102,
      "learning_rate": 0.0001933984968749824,
      "loss": 2.0833,
      "step": 141844
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.1931871175765991,
      "learning_rate": 0.00019339467328201112,
      "loss": 2.004,
      "step": 141845
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.084388017654419,
      "learning_rate": 0.0001933908497088596,
      "loss": 2.0933,
      "step": 141846
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.1495505571365356,
      "learning_rate": 0.0001933870261555288,
      "loss": 1.9657,
      "step": 141847
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.06772780418396,
      "learning_rate": 0.00019338320262201933,
      "loss": 2.0592,
      "step": 141848
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.0704922676086426,
      "learning_rate": 0.00019337937910833192,
      "loss": 2.0911,
      "step": 141849
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.0626420974731445,
      "learning_rate": 0.00019337555561446733,
      "loss": 2.1622,
      "step": 141850
    },
    {
      "epoch": 1.85,
      "grad_norm": 0.9802783727645874,
      "learning_rate": 0.0001933717321404263,
      "loss": 2.1135,
      "step": 141851
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.0648584365844727,
      "learning_rate": 0.00019336790868620946,
      "loss": 2.0223,
      "step": 141852
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.117945671081543,
      "learning_rate": 0.00019336408525181745,
      "loss": 2.1181,
      "step": 141853
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.0713106393814087,
      "learning_rate": 0.00019336026183725108,
      "loss": 1.9304,
      "step": 141854
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.0908126831054688,
      "learning_rate": 0.00019335643844251108,
      "loss": 2.2795,
      "step": 141855
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.0795668363571167,
      "learning_rate": 0.00019335261506759813,
      "loss": 2.308,
      "step": 141856
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.0838334560394287,
      "learning_rate": 0.00019334879171251302,
      "loss": 2.0781,
      "step": 141857
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.0615417957305908,
      "learning_rate": 0.00019334496837725628,
      "loss": 1.9886,
      "step": 141858
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.1393803358078003,
      "learning_rate": 0.00019334114506182874,
      "loss": 2.0021,
      "step": 141859
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.036910057067871,
      "learning_rate": 0.00019333732176623114,
      "loss": 2.1931,
      "step": 141860
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.048921823501587,
      "learning_rate": 0.0001933334984904641,
      "loss": 2.065,
      "step": 141861
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.1527000665664673,
      "learning_rate": 0.00019332967523452837,
      "loss": 2.4119,
      "step": 141862
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.0870628356933594,
      "learning_rate": 0.00019332585199842466,
      "loss": 2.2351,
      "step": 141863
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.1784852743148804,
      "learning_rate": 0.00019332202878215375,
      "loss": 2.0136,
      "step": 141864
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.0050960779190063,
      "learning_rate": 0.00019331820558571625,
      "loss": 2.4122,
      "step": 141865
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.0406757593154907,
      "learning_rate": 0.0001933143824091129,
      "loss": 2.0605,
      "step": 141866
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.0107167959213257,
      "learning_rate": 0.00019331055925234447,
      "loss": 1.8103,
      "step": 141867
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.007507562637329,
      "learning_rate": 0.00019330673611541157,
      "loss": 2.2005,
      "step": 141868
    },
    {
      "epoch": 1.85,
      "grad_norm": 0.9851704239845276,
      "learning_rate": 0.00019330291299831498,
      "loss": 1.8943,
      "step": 141869
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.2188434600830078,
      "learning_rate": 0.0001932990899010554,
      "loss": 2.076,
      "step": 141870
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.1136257648468018,
      "learning_rate": 0.00019329526682363354,
      "loss": 2.0465,
      "step": 141871
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.1505035161972046,
      "learning_rate": 0.00019329144376605006,
      "loss": 2.1905,
      "step": 141872
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.1201870441436768,
      "learning_rate": 0.00019328762072830572,
      "loss": 1.9922,
      "step": 141873
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.0486177206039429,
      "learning_rate": 0.0001932837977104013,
      "loss": 2.0874,
      "step": 141874
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.034056305885315,
      "learning_rate": 0.00019327997471233735,
      "loss": 1.8987,
      "step": 141875
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.0609302520751953,
      "learning_rate": 0.00019327615173411473,
      "loss": 2.0044,
      "step": 141876
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.0847976207733154,
      "learning_rate": 0.00019327232877573407,
      "loss": 2.0717,
      "step": 141877
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.080627202987671,
      "learning_rate": 0.00019326850583719607,
      "loss": 2.0066,
      "step": 141878
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.0418123006820679,
      "learning_rate": 0.00019326468291850146,
      "loss": 2.0628,
      "step": 141879
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.0723731517791748,
      "learning_rate": 0.00019326086001965098,
      "loss": 2.0576,
      "step": 141880
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.2995208501815796,
      "learning_rate": 0.00019325703714064536,
      "loss": 2.0842,
      "step": 141881
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.0960166454315186,
      "learning_rate": 0.0001932532142814853,
      "loss": 1.9605,
      "step": 141882
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.0098332166671753,
      "learning_rate": 0.0001932493914421714,
      "loss": 1.8854,
      "step": 141883
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.165411114692688,
      "learning_rate": 0.00019324556862270443,
      "loss": 2.4774,
      "step": 141884
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.2522979974746704,
      "learning_rate": 0.00019324174582308518,
      "loss": 2.1022,
      "step": 141885
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.0443247556686401,
      "learning_rate": 0.00019323792304331425,
      "loss": 2.2155,
      "step": 141886
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.0988627672195435,
      "learning_rate": 0.00019323410028339243,
      "loss": 2.1736,
      "step": 141887
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.1611379384994507,
      "learning_rate": 0.00019323027754332054,
      "loss": 2.1097,
      "step": 141888
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.3623230457305908,
      "learning_rate": 0.00019322645482309903,
      "loss": 1.9518,
      "step": 141889
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.0111713409423828,
      "learning_rate": 0.00019322263212272873,
      "loss": 2.11,
      "step": 141890
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.1385655403137207,
      "learning_rate": 0.00019321880944221035,
      "loss": 2.1393,
      "step": 141891
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.2535696029663086,
      "learning_rate": 0.00019321498678154464,
      "loss": 2.1532,
      "step": 141892
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.1106013059616089,
      "learning_rate": 0.00019321116414073225,
      "loss": 2.248,
      "step": 141893
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.0300565958023071,
      "learning_rate": 0.00019320734151977406,
      "loss": 2.1351,
      "step": 141894
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.1576159000396729,
      "learning_rate": 0.0001932035189186705,
      "loss": 1.9444,
      "step": 141895
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.1078259944915771,
      "learning_rate": 0.00019319969633742245,
      "loss": 2.3259,
      "step": 141896
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.166780948638916,
      "learning_rate": 0.0001931958737760306,
      "loss": 2.2527,
      "step": 141897
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.067960262298584,
      "learning_rate": 0.0001931920512344956,
      "loss": 1.8907,
      "step": 141898
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.0175230503082275,
      "learning_rate": 0.00019318822871281822,
      "loss": 2.0655,
      "step": 141899
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.049878478050232,
      "learning_rate": 0.00019318440621099933,
      "loss": 1.896,
      "step": 141900
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.2092595100402832,
      "learning_rate": 0.00019318058372903932,
      "loss": 2.1002,
      "step": 141901
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.0084865093231201,
      "learning_rate": 0.0001931767612669391,
      "loss": 1.9427,
      "step": 141902
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.0236778259277344,
      "learning_rate": 0.0001931729388246993,
      "loss": 1.9837,
      "step": 141903
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.1539301872253418,
      "learning_rate": 0.00019316911640232068,
      "loss": 2.1248,
      "step": 141904
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.100601077079773,
      "learning_rate": 0.00019316529399980388,
      "loss": 2.107,
      "step": 141905
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.573718547821045,
      "learning_rate": 0.00019316147161714986,
      "loss": 2.2868,
      "step": 141906
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.1792807579040527,
      "learning_rate": 0.00019315764925435902,
      "loss": 2.0248,
      "step": 141907
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.1000146865844727,
      "learning_rate": 0.00019315382691143215,
      "loss": 2.2748,
      "step": 141908
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.1064479351043701,
      "learning_rate": 0.00019315000458837,
      "loss": 2.1749,
      "step": 141909
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.0900026559829712,
      "learning_rate": 0.00019314618228517328,
      "loss": 2.1739,
      "step": 141910
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.2140415906906128,
      "learning_rate": 0.00019314236000184275,
      "loss": 2.1558,
      "step": 141911
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.0246272087097168,
      "learning_rate": 0.00019313853773837917,
      "loss": 2.2011,
      "step": 141912
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.167375922203064,
      "learning_rate": 0.00019313471549478302,
      "loss": 2.1272,
      "step": 141913
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.3796411752700806,
      "learning_rate": 0.00019313089327105515,
      "loss": 2.1083,
      "step": 141914
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.1635690927505493,
      "learning_rate": 0.00019312707106719626,
      "loss": 1.999,
      "step": 141915
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.066954493522644,
      "learning_rate": 0.00019312324888320705,
      "loss": 1.9462,
      "step": 141916
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.1236391067504883,
      "learning_rate": 0.0001931194267190883,
      "loss": 2.1676,
      "step": 141917
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.1565862894058228,
      "learning_rate": 0.0001931156045748407,
      "loss": 2.0568,
      "step": 141918
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.0151926279067993,
      "learning_rate": 0.0001931117824504649,
      "loss": 2.2398,
      "step": 141919
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.0181078910827637,
      "learning_rate": 0.00019310796034596156,
      "loss": 2.2018,
      "step": 141920
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.081999659538269,
      "learning_rate": 0.00019310413826133151,
      "loss": 1.9194,
      "step": 141921
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.0978375673294067,
      "learning_rate": 0.0001931003161965754,
      "loss": 2.0846,
      "step": 141922
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.0499911308288574,
      "learning_rate": 0.00019309649415169398,
      "loss": 1.9359,
      "step": 141923
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.0507344007492065,
      "learning_rate": 0.0001930926721266881,
      "loss": 2.1152,
      "step": 141924
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.4163308143615723,
      "learning_rate": 0.00019308885012155812,
      "loss": 2.225,
      "step": 141925
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.1672210693359375,
      "learning_rate": 0.00019308502813630496,
      "loss": 1.9268,
      "step": 141926
    },
    {
      "epoch": 1.85,
      "grad_norm": 0.9877490997314453,
      "learning_rate": 0.0001930812061709293,
      "loss": 1.7869,
      "step": 141927
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.0553524494171143,
      "learning_rate": 0.00019307738422543187,
      "loss": 2.2754,
      "step": 141928
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.1092678308486938,
      "learning_rate": 0.00019307356229981336,
      "loss": 2.0585,
      "step": 141929
    },
    {
      "epoch": 1.85,
      "grad_norm": 0.9522138833999634,
      "learning_rate": 0.0001930697403940745,
      "loss": 2.0185,
      "step": 141930
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.0018779039382935,
      "learning_rate": 0.00019306591850821615,
      "loss": 2.068,
      "step": 141931
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.0395786762237549,
      "learning_rate": 0.00019306209664223873,
      "loss": 2.0747,
      "step": 141932
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.0698169469833374,
      "learning_rate": 0.00019305827479614306,
      "loss": 2.0396,
      "step": 141933
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.0087909698486328,
      "learning_rate": 0.0001930544529699299,
      "loss": 2.011,
      "step": 141934
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.059568166732788,
      "learning_rate": 0.00019305063116359988,
      "loss": 2.4308,
      "step": 141935
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.3586912155151367,
      "learning_rate": 0.00019304680937715383,
      "loss": 2.2032,
      "step": 141936
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.1611660718917847,
      "learning_rate": 0.0001930429876105925,
      "loss": 1.9996,
      "step": 141937
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.0689762830734253,
      "learning_rate": 0.00019303916586391637,
      "loss": 2.1848,
      "step": 141938
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.0718661546707153,
      "learning_rate": 0.00019303534413712628,
      "loss": 1.7978,
      "step": 141939
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.0044846534729004,
      "learning_rate": 0.00019303152243022297,
      "loss": 1.9153,
      "step": 141940
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.2066388130187988,
      "learning_rate": 0.00019302770074320708,
      "loss": 2.0966,
      "step": 141941
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.1889543533325195,
      "learning_rate": 0.00019302387907607936,
      "loss": 1.9893,
      "step": 141942
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.056274652481079,
      "learning_rate": 0.0001930200574288406,
      "loss": 2.0423,
      "step": 141943
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.0845650434494019,
      "learning_rate": 0.0001930162358014914,
      "loss": 2.4407,
      "step": 141944
    },
    {
      "epoch": 1.85,
      "grad_norm": 2.888021230697632,
      "learning_rate": 0.00019301241419403245,
      "loss": 2.0075,
      "step": 141945
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.123063087463379,
      "learning_rate": 0.00019300859260646457,
      "loss": 2.2649,
      "step": 141946
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.0624542236328125,
      "learning_rate": 0.00019300477103878836,
      "loss": 2.3711,
      "step": 141947
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.2195038795471191,
      "learning_rate": 0.0001930009494910046,
      "loss": 1.8892,
      "step": 141948
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.0469870567321777,
      "learning_rate": 0.000192997127963114,
      "loss": 2.166,
      "step": 141949
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.3446927070617676,
      "learning_rate": 0.00019299330645511725,
      "loss": 2.1341,
      "step": 141950
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.1603431701660156,
      "learning_rate": 0.00019298948496701505,
      "loss": 2.1658,
      "step": 141951
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.3623517751693726,
      "learning_rate": 0.00019298566349880818,
      "loss": 2.0557,
      "step": 141952
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.0154942274093628,
      "learning_rate": 0.00019298184205049717,
      "loss": 2.0174,
      "step": 141953
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.0638123750686646,
      "learning_rate": 0.00019297802062208293,
      "loss": 1.9933,
      "step": 141954
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.2660155296325684,
      "learning_rate": 0.0001929741992135662,
      "loss": 2.1695,
      "step": 141955
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.1609687805175781,
      "learning_rate": 0.00019297037782494745,
      "loss": 2.145,
      "step": 141956
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.0411326885223389,
      "learning_rate": 0.0001929665564562275,
      "loss": 1.8894,
      "step": 141957
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.0266278982162476,
      "learning_rate": 0.00019296273510740713,
      "loss": 2.0756,
      "step": 141958
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.0406427383422852,
      "learning_rate": 0.0001929589137784871,
      "loss": 2.1686,
      "step": 141959
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.223904013633728,
      "learning_rate": 0.00019295509246946795,
      "loss": 2.2458,
      "step": 141960
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.1305691003799438,
      "learning_rate": 0.0001929512711803505,
      "loss": 2.3721,
      "step": 141961
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.0903716087341309,
      "learning_rate": 0.0001929474499111354,
      "loss": 2.103,
      "step": 141962
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.0919464826583862,
      "learning_rate": 0.0001929436286618234,
      "loss": 1.7582,
      "step": 141963
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.0066356658935547,
      "learning_rate": 0.00019293980743241516,
      "loss": 2.0775,
      "step": 141964
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.0956556797027588,
      "learning_rate": 0.00019293598622291145,
      "loss": 1.9461,
      "step": 141965
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.044413447380066,
      "learning_rate": 0.000192932165033313,
      "loss": 1.9792,
      "step": 141966
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.3157297372817993,
      "learning_rate": 0.00019292834386362053,
      "loss": 2.0176,
      "step": 141967
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.1107792854309082,
      "learning_rate": 0.00019292452271383462,
      "loss": 1.9608,
      "step": 141968
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.0207722187042236,
      "learning_rate": 0.00019292070158395604,
      "loss": 2.1406,
      "step": 141969
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.049707293510437,
      "learning_rate": 0.00019291688047398556,
      "loss": 1.922,
      "step": 141970
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.0180069208145142,
      "learning_rate": 0.00019291305938392382,
      "loss": 2.1212,
      "step": 141971
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.0024540424346924,
      "learning_rate": 0.00019290923831377155,
      "loss": 1.9652,
      "step": 141972
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.083618402481079,
      "learning_rate": 0.00019290541726352966,
      "loss": 2.0985,
      "step": 141973
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.1584982872009277,
      "learning_rate": 0.00019290159623319846,
      "loss": 2.1571,
      "step": 141974
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.2789138555526733,
      "learning_rate": 0.00019289777522277896,
      "loss": 2.171,
      "step": 141975
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.2578505277633667,
      "learning_rate": 0.00019289395423227173,
      "loss": 2.1402,
      "step": 141976
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.3335018157958984,
      "learning_rate": 0.0001928901332616776,
      "loss": 1.9716,
      "step": 141977
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.1685748100280762,
      "learning_rate": 0.00019288631231099716,
      "loss": 2.088,
      "step": 141978
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.1081089973449707,
      "learning_rate": 0.0001928824913802313,
      "loss": 2.0276,
      "step": 141979
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.2412022352218628,
      "learning_rate": 0.00019287867046938051,
      "loss": 2.2392,
      "step": 141980
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.1739875078201294,
      "learning_rate": 0.00019287484957844562,
      "loss": 2.063,
      "step": 141981
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.2246195077896118,
      "learning_rate": 0.00019287102870742725,
      "loss": 2.0441,
      "step": 141982
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.1416542530059814,
      "learning_rate": 0.00019286720785632622,
      "loss": 1.8983,
      "step": 141983
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.0387554168701172,
      "learning_rate": 0.0001928633870251432,
      "loss": 2.0542,
      "step": 141984
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.1019482612609863,
      "learning_rate": 0.00019285956621387904,
      "loss": 2.063,
      "step": 141985
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.078326940536499,
      "learning_rate": 0.00019285574542253413,
      "loss": 2.1875,
      "step": 141986
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.1622406244277954,
      "learning_rate": 0.0001928519246511094,
      "loss": 2.0425,
      "step": 141987
    },
    {
      "epoch": 1.85,
      "grad_norm": 0.9870566725730896,
      "learning_rate": 0.0001928481038996055,
      "loss": 2.0978,
      "step": 141988
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.2271429300308228,
      "learning_rate": 0.0001928442831680232,
      "loss": 2.0889,
      "step": 141989
    },
    {
      "epoch": 1.85,
      "grad_norm": 0.9961974620819092,
      "learning_rate": 0.0001928404624563631,
      "loss": 2.0155,
      "step": 141990
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.2116103172302246,
      "learning_rate": 0.00019283664176462614,
      "loss": 2.2602,
      "step": 141991
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.103977084159851,
      "learning_rate": 0.00019283282109281276,
      "loss": 1.8658,
      "step": 141992
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.0061475038528442,
      "learning_rate": 0.00019282900044092377,
      "loss": 2.2751,
      "step": 141993
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.1712294816970825,
      "learning_rate": 0.0001928251798089599,
      "loss": 1.9522,
      "step": 141994
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.1210564374923706,
      "learning_rate": 0.00019282135919692186,
      "loss": 2.2413,
      "step": 141995
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.0621687173843384,
      "learning_rate": 0.00019281753860481032,
      "loss": 1.8231,
      "step": 141996
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.0452414751052856,
      "learning_rate": 0.00019281371803262604,
      "loss": 2.0188,
      "step": 141997
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.0308889150619507,
      "learning_rate": 0.0001928098974803698,
      "loss": 2.3513,
      "step": 141998
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.1234757900238037,
      "learning_rate": 0.0001928060769480422,
      "loss": 2.0893,
      "step": 141999
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.0043346881866455,
      "learning_rate": 0.00019280225643564385,
      "loss": 2.1116,
      "step": 142000
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.1451773643493652,
      "learning_rate": 0.00019279843594317568,
      "loss": 2.198,
      "step": 142001
    },
    {
      "epoch": 1.85,
      "grad_norm": 0.9629368185997009,
      "learning_rate": 0.00019279461547063828,
      "loss": 1.9907,
      "step": 142002
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.0153148174285889,
      "learning_rate": 0.00019279079501803237,
      "loss": 2.198,
      "step": 142003
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.2904906272888184,
      "learning_rate": 0.00019278697458535878,
      "loss": 2.3663,
      "step": 142004
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.1307376623153687,
      "learning_rate": 0.00019278315417261803,
      "loss": 2.0894,
      "step": 142005
    },
    {
      "epoch": 1.85,
      "grad_norm": 0.9925513863563538,
      "learning_rate": 0.0001927793337798109,
      "loss": 1.9278,
      "step": 142006
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.1057560443878174,
      "learning_rate": 0.00019277551340693812,
      "loss": 2.0414,
      "step": 142007
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.0346955060958862,
      "learning_rate": 0.0001927716930540004,
      "loss": 2.1326,
      "step": 142008
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.0897163152694702,
      "learning_rate": 0.00019276787272099846,
      "loss": 2.1743,
      "step": 142009
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.0003161430358887,
      "learning_rate": 0.00019276405240793308,
      "loss": 2.1106,
      "step": 142010
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.0450105667114258,
      "learning_rate": 0.00019276023211480477,
      "loss": 2.3414,
      "step": 142011
    },
    {
      "epoch": 1.85,
      "grad_norm": 0.997184693813324,
      "learning_rate": 0.0001927564118416144,
      "loss": 1.9237,
      "step": 142012
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.0569902658462524,
      "learning_rate": 0.00019275259158836258,
      "loss": 2.0053,
      "step": 142013
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.0643373727798462,
      "learning_rate": 0.00019274877135505008,
      "loss": 1.9307,
      "step": 142014
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.088753342628479,
      "learning_rate": 0.00019274495114167766,
      "loss": 2.0444,
      "step": 142015
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.0928786993026733,
      "learning_rate": 0.00019274113094824605,
      "loss": 2.2217,
      "step": 142016
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.1115812063217163,
      "learning_rate": 0.0001927373107747558,
      "loss": 2.3165,
      "step": 142017
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.053644061088562,
      "learning_rate": 0.0001927334906212077,
      "loss": 1.9923,
      "step": 142018
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.0430853366851807,
      "learning_rate": 0.00019272967048760247,
      "loss": 2.0456,
      "step": 142019
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.09011971950531,
      "learning_rate": 0.00019272585037394084,
      "loss": 2.1513,
      "step": 142020
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.1022506952285767,
      "learning_rate": 0.00019272203028022344,
      "loss": 2.0791,
      "step": 142021
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.0832477807998657,
      "learning_rate": 0.0001927182102064512,
      "loss": 2.2857,
      "step": 142022
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.2244082689285278,
      "learning_rate": 0.0001927143901526246,
      "loss": 2.0704,
      "step": 142023
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.1799834966659546,
      "learning_rate": 0.00019271057011874433,
      "loss": 2.2214,
      "step": 142024
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.0840399265289307,
      "learning_rate": 0.00019270675010481123,
      "loss": 2.0525,
      "step": 142025
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.0582499504089355,
      "learning_rate": 0.00019270293011082595,
      "loss": 1.9669,
      "step": 142026
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.1870352029800415,
      "learning_rate": 0.00019269911013678923,
      "loss": 2.1463,
      "step": 142027
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.1010531187057495,
      "learning_rate": 0.00019269529018270184,
      "loss": 1.9769,
      "step": 142028
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.0746418237686157,
      "learning_rate": 0.00019269147024856444,
      "loss": 2.1372,
      "step": 142029
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.1081328392028809,
      "learning_rate": 0.0001926876503343776,
      "loss": 2.0931,
      "step": 142030
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.291473627090454,
      "learning_rate": 0.00019268383044014221,
      "loss": 2.1022,
      "step": 142031
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.1627538204193115,
      "learning_rate": 0.0001926800105658589,
      "loss": 1.9445,
      "step": 142032
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.1029679775238037,
      "learning_rate": 0.00019267619071152843,
      "loss": 2.2174,
      "step": 142033
    },
    {
      "epoch": 1.85,
      "grad_norm": 0.9872429966926575,
      "learning_rate": 0.0001926723708771515,
      "loss": 2.2774,
      "step": 142034
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.1499106884002686,
      "learning_rate": 0.00019266855106272874,
      "loss": 2.1895,
      "step": 142035
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.0563303232192993,
      "learning_rate": 0.00019266473126826103,
      "loss": 2.1648,
      "step": 142036
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.0011882781982422,
      "learning_rate": 0.00019266091149374885,
      "loss": 2.0652,
      "step": 142037
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.078902006149292,
      "learning_rate": 0.00019265709173919311,
      "loss": 1.9843,
      "step": 142038
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.0319560766220093,
      "learning_rate": 0.00019265327200459434,
      "loss": 2.0886,
      "step": 142039
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.008305311203003,
      "learning_rate": 0.0001926494522899535,
      "loss": 2.0334,
      "step": 142040
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.0029664039611816,
      "learning_rate": 0.00019264563259527108,
      "loss": 2.1117,
      "step": 142041
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.0396933555603027,
      "learning_rate": 0.00019264181292054784,
      "loss": 2.1773,
      "step": 142042
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.080124020576477,
      "learning_rate": 0.0001926379932657846,
      "loss": 2.0634,
      "step": 142043
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.1075993776321411,
      "learning_rate": 0.0001926341736309819,
      "loss": 2.2375,
      "step": 142044
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.1032257080078125,
      "learning_rate": 0.00019263035401614052,
      "loss": 2.0051,
      "step": 142045
    },
    {
      "epoch": 1.85,
      "grad_norm": 0.9941357374191284,
      "learning_rate": 0.00019262653442126128,
      "loss": 1.8448,
      "step": 142046
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.3738888502120972,
      "learning_rate": 0.00019262271484634473,
      "loss": 2.0903,
      "step": 142047
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.3048287630081177,
      "learning_rate": 0.00019261889529139162,
      "loss": 2.0338,
      "step": 142048
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.4261748790740967,
      "learning_rate": 0.00019261507575640268,
      "loss": 2.1822,
      "step": 142049
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.1321262121200562,
      "learning_rate": 0.00019261125624137865,
      "loss": 2.0592,
      "step": 142050
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.164488673210144,
      "learning_rate": 0.00019260743674632026,
      "loss": 2.1055,
      "step": 142051
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.175981044769287,
      "learning_rate": 0.00019260361727122818,
      "loss": 2.0279,
      "step": 142052
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.1238819360733032,
      "learning_rate": 0.00019259979781610306,
      "loss": 2.2393,
      "step": 142053
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.0173935890197754,
      "learning_rate": 0.00019259597838094564,
      "loss": 1.9428,
      "step": 142054
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.0845383405685425,
      "learning_rate": 0.00019259215896575668,
      "loss": 2.1296,
      "step": 142055
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.24990713596344,
      "learning_rate": 0.0001925883395705369,
      "loss": 2.1226,
      "step": 142056
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.1441655158996582,
      "learning_rate": 0.0001925845201952869,
      "loss": 2.0403,
      "step": 142057
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.1202183961868286,
      "learning_rate": 0.00019258070084000764,
      "loss": 2.2575,
      "step": 142058
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.088294506072998,
      "learning_rate": 0.0001925768815046995,
      "loss": 1.9018,
      "step": 142059
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.185512661933899,
      "learning_rate": 0.00019257306218936336,
      "loss": 2.1115,
      "step": 142060
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.2312718629837036,
      "learning_rate": 0.00019256924289399994,
      "loss": 2.1116,
      "step": 142061
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.0756452083587646,
      "learning_rate": 0.00019256542361860987,
      "loss": 2.1397,
      "step": 142062
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.142051100730896,
      "learning_rate": 0.00019256160436319399,
      "loss": 2.2666,
      "step": 142063
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.0719051361083984,
      "learning_rate": 0.00019255778512775287,
      "loss": 2.1393,
      "step": 142064
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.6753343343734741,
      "learning_rate": 0.00019255396591228742,
      "loss": 2.2938,
      "step": 142065
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.1412590742111206,
      "learning_rate": 0.00019255014671679813,
      "loss": 1.8895,
      "step": 142066
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.043527603149414,
      "learning_rate": 0.00019254632754128577,
      "loss": 2.0518,
      "step": 142067
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.1453872919082642,
      "learning_rate": 0.00019254250838575106,
      "loss": 2.2746,
      "step": 142068
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.0816296339035034,
      "learning_rate": 0.00019253868925019473,
      "loss": 2.0488,
      "step": 142069
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.2760194540023804,
      "learning_rate": 0.00019253487013461755,
      "loss": 2.2259,
      "step": 142070
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.0417159795761108,
      "learning_rate": 0.00019253105103902022,
      "loss": 1.987,
      "step": 142071
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.1878424882888794,
      "learning_rate": 0.0001925272319634033,
      "loss": 2.1751,
      "step": 142072
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.1351395845413208,
      "learning_rate": 0.00019252341290776755,
      "loss": 1.9463,
      "step": 142073
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.1948024034500122,
      "learning_rate": 0.00019251959387211378,
      "loss": 2.0466,
      "step": 142074
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.045831561088562,
      "learning_rate": 0.00019251577485644264,
      "loss": 2.1114,
      "step": 142075
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.1176048517227173,
      "learning_rate": 0.00019251195586075486,
      "loss": 1.971,
      "step": 142076
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.1549588441848755,
      "learning_rate": 0.0001925081368850512,
      "loss": 2.0568,
      "step": 142077
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.16293466091156,
      "learning_rate": 0.00019250431792933225,
      "loss": 2.1043,
      "step": 142078
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.1371780633926392,
      "learning_rate": 0.00019250049899359874,
      "loss": 2.0401,
      "step": 142079
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.4189722537994385,
      "learning_rate": 0.00019249668007785144,
      "loss": 2.1076,
      "step": 142080
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.206244945526123,
      "learning_rate": 0.00019249286118209102,
      "loss": 2.1699,
      "step": 142081
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.1535722017288208,
      "learning_rate": 0.00019248904230631818,
      "loss": 2.0426,
      "step": 142082
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.0180792808532715,
      "learning_rate": 0.00019248522345053382,
      "loss": 2.0651,
      "step": 142083
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.065187931060791,
      "learning_rate": 0.00019248140461473835,
      "loss": 2.3321,
      "step": 142084
    },
    {
      "epoch": 1.85,
      "grad_norm": 0.9719467759132385,
      "learning_rate": 0.00019247758579893262,
      "loss": 2.1569,
      "step": 142085
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.3041104078292847,
      "learning_rate": 0.00019247376700311735,
      "loss": 2.0458,
      "step": 142086
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.2409963607788086,
      "learning_rate": 0.00019246994822729317,
      "loss": 2.1932,
      "step": 142087
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.0753546953201294,
      "learning_rate": 0.00019246612947146092,
      "loss": 2.0547,
      "step": 142088
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.0538513660430908,
      "learning_rate": 0.00019246231073562136,
      "loss": 2.2327,
      "step": 142089
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.1221946477890015,
      "learning_rate": 0.00019245849201977495,
      "loss": 1.9746,
      "step": 142090
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.0454750061035156,
      "learning_rate": 0.00019245467332392255,
      "loss": 2.0321,
      "step": 142091
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.1312944889068604,
      "learning_rate": 0.00019245085464806485,
      "loss": 2.1234,
      "step": 142092
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.05399751663208,
      "learning_rate": 0.00019244703599220257,
      "loss": 1.9473,
      "step": 142093
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.0980517864227295,
      "learning_rate": 0.0001924432173563364,
      "loss": 2.2393,
      "step": 142094
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.1070882081985474,
      "learning_rate": 0.00019243939874046722,
      "loss": 1.9966,
      "step": 142095
    },
    {
      "epoch": 1.85,
      "grad_norm": 0.9934410452842712,
      "learning_rate": 0.00019243558014459545,
      "loss": 1.9766,
      "step": 142096
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.0762555599212646,
      "learning_rate": 0.00019243176156872192,
      "loss": 1.9994,
      "step": 142097
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.1868956089019775,
      "learning_rate": 0.00019242794301284737,
      "loss": 2.0014,
      "step": 142098
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.0101362466812134,
      "learning_rate": 0.0001924241244769725,
      "loss": 2.0053,
      "step": 142099
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.167818307876587,
      "learning_rate": 0.000192420305961098,
      "loss": 2.2041,
      "step": 142100
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.1010602712631226,
      "learning_rate": 0.00019241648746522473,
      "loss": 2.2162,
      "step": 142101
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.0226829051971436,
      "learning_rate": 0.00019241266898935313,
      "loss": 2.089,
      "step": 142102
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.1411937475204468,
      "learning_rate": 0.00019240885053348406,
      "loss": 2.1477,
      "step": 142103
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.0529193878173828,
      "learning_rate": 0.00019240503209761825,
      "loss": 2.118,
      "step": 142104
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.2204666137695312,
      "learning_rate": 0.0001924012136817563,
      "loss": 2.0516,
      "step": 142105
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.0547430515289307,
      "learning_rate": 0.00019239739528589903,
      "loss": 2.1448,
      "step": 142106
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.0408875942230225,
      "learning_rate": 0.00019239357691004725,
      "loss": 2.1684,
      "step": 142107
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.0796279907226562,
      "learning_rate": 0.00019238975855420142,
      "loss": 2.2725,
      "step": 142108
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.1121779680252075,
      "learning_rate": 0.0001923859402183623,
      "loss": 2.0884,
      "step": 142109
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.0898195505142212,
      "learning_rate": 0.00019238212190253072,
      "loss": 2.204,
      "step": 142110
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.088698148727417,
      "learning_rate": 0.00019237830360670734,
      "loss": 2.1145,
      "step": 142111
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.078534722328186,
      "learning_rate": 0.0001923744853308929,
      "loss": 2.0862,
      "step": 142112
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.0911945104599,
      "learning_rate": 0.00019237066707508807,
      "loss": 2.0941,
      "step": 142113
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.0826771259307861,
      "learning_rate": 0.00019236684883929358,
      "loss": 2.0738,
      "step": 142114
    },
    {
      "epoch": 1.85,
      "grad_norm": 0.9177529215812683,
      "learning_rate": 0.00019236303062351005,
      "loss": 1.9863,
      "step": 142115
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.044075608253479,
      "learning_rate": 0.00019235921242773832,
      "loss": 2.0543,
      "step": 142116
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.0159844160079956,
      "learning_rate": 0.00019235539425197897,
      "loss": 2.0316,
      "step": 142117
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.0909420251846313,
      "learning_rate": 0.00019235157609623284,
      "loss": 2.2423,
      "step": 142118
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.1267564296722412,
      "learning_rate": 0.00019234775796050062,
      "loss": 2.0984,
      "step": 142119
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.0457453727722168,
      "learning_rate": 0.00019234393984478293,
      "loss": 1.9803,
      "step": 142120
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.1744787693023682,
      "learning_rate": 0.0001923401217490806,
      "loss": 2.2072,
      "step": 142121
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.0874830484390259,
      "learning_rate": 0.0001923363036733942,
      "loss": 1.9049,
      "step": 142122
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.0901130437850952,
      "learning_rate": 0.0001923324856177245,
      "loss": 2.3363,
      "step": 142123
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.2056821584701538,
      "learning_rate": 0.00019232866758207227,
      "loss": 2.0525,
      "step": 142124
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.234164834022522,
      "learning_rate": 0.0001923248495664382,
      "loss": 2.2158,
      "step": 142125
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.1341403722763062,
      "learning_rate": 0.0001923210315708229,
      "loss": 2.3302,
      "step": 142126
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.2532106637954712,
      "learning_rate": 0.0001923172135952272,
      "loss": 1.9541,
      "step": 142127
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.031462550163269,
      "learning_rate": 0.0001923133956396518,
      "loss": 2.1945,
      "step": 142128
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.0589028596878052,
      "learning_rate": 0.00019230957770409728,
      "loss": 2.3204,
      "step": 142129
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.1208055019378662,
      "learning_rate": 0.00019230575978856448,
      "loss": 2.1931,
      "step": 142130
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.2252576351165771,
      "learning_rate": 0.00019230194189305409,
      "loss": 2.2897,
      "step": 142131
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.1609115600585938,
      "learning_rate": 0.00019229812401756679,
      "loss": 1.8967,
      "step": 142132
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.1018500328063965,
      "learning_rate": 0.0001922943061621033,
      "loss": 2.0524,
      "step": 142133
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.00424325466156,
      "learning_rate": 0.00019229048832666433,
      "loss": 2.1061,
      "step": 142134
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.0076844692230225,
      "learning_rate": 0.00019228667051125062,
      "loss": 2.0359,
      "step": 142135
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.1108574867248535,
      "learning_rate": 0.00019228285271586282,
      "loss": 2.1238,
      "step": 142136
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.0312917232513428,
      "learning_rate": 0.00019227903494050167,
      "loss": 2.0333,
      "step": 142137
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.0932464599609375,
      "learning_rate": 0.00019227521718516793,
      "loss": 2.1154,
      "step": 142138
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.0754570960998535,
      "learning_rate": 0.00019227139944986217,
      "loss": 2.2262,
      "step": 142139
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.0866353511810303,
      "learning_rate": 0.00019226758173458523,
      "loss": 2.1201,
      "step": 142140
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.05125093460083,
      "learning_rate": 0.00019226376403933782,
      "loss": 2.0356,
      "step": 142141
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.108164668083191,
      "learning_rate": 0.00019225994636412052,
      "loss": 2.0655,
      "step": 142142
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.0250349044799805,
      "learning_rate": 0.00019225612870893424,
      "loss": 2.2062,
      "step": 142143
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.097298502922058,
      "learning_rate": 0.00019225231107377957,
      "loss": 2.0341,
      "step": 142144
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.1710476875305176,
      "learning_rate": 0.0001922484934586572,
      "loss": 2.2002,
      "step": 142145
    },
    {
      "epoch": 1.85,
      "grad_norm": 0.9886141419410706,
      "learning_rate": 0.0001922446758635678,
      "loss": 2.2223,
      "step": 142146
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.2032456398010254,
      "learning_rate": 0.0001922408582885122,
      "loss": 2.0749,
      "step": 142147
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.0763753652572632,
      "learning_rate": 0.00019223704073349107,
      "loss": 2.2203,
      "step": 142148
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.0582225322723389,
      "learning_rate": 0.0001922332231985051,
      "loss": 2.1596,
      "step": 142149
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.1419823169708252,
      "learning_rate": 0.00019222940568355506,
      "loss": 2.2392,
      "step": 142150
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.1462551355361938,
      "learning_rate": 0.00019222558818864156,
      "loss": 2.4097,
      "step": 142151
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.064774990081787,
      "learning_rate": 0.00019222177071376534,
      "loss": 1.8312,
      "step": 142152
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.1089266538619995,
      "learning_rate": 0.00019221795325892715,
      "loss": 2.0038,
      "step": 142153
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.0660922527313232,
      "learning_rate": 0.00019221413582412765,
      "loss": 2.143,
      "step": 142154
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.0357980728149414,
      "learning_rate": 0.00019221031840936756,
      "loss": 2.0193,
      "step": 142155
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.114214539527893,
      "learning_rate": 0.00019220650101464773,
      "loss": 2.2278,
      "step": 142156
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.0894185304641724,
      "learning_rate": 0.00019220268363996865,
      "loss": 2.268,
      "step": 142157
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.356762170791626,
      "learning_rate": 0.00019219886628533113,
      "loss": 2.0992,
      "step": 142158
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.134978175163269,
      "learning_rate": 0.0001921950489507359,
      "loss": 2.1356,
      "step": 142159
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.13050377368927,
      "learning_rate": 0.00019219123163618357,
      "loss": 2.2525,
      "step": 142160
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.0121482610702515,
      "learning_rate": 0.00019218741434167498,
      "loss": 2.0819,
      "step": 142161
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.0925747156143188,
      "learning_rate": 0.0001921835970672109,
      "loss": 1.9626,
      "step": 142162
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.1846572160720825,
      "learning_rate": 0.0001921797798127918,
      "loss": 2.0686,
      "step": 142163
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.063144564628601,
      "learning_rate": 0.00019217596257841847,
      "loss": 1.9749,
      "step": 142164
    },
    {
      "epoch": 1.85,
      "grad_norm": 0.9969320893287659,
      "learning_rate": 0.00019217214536409172,
      "loss": 2.1088,
      "step": 142165
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.1497405767440796,
      "learning_rate": 0.00019216832816981217,
      "loss": 1.9642,
      "step": 142166
    },
    {
      "epoch": 1.85,
      "grad_norm": 0.994590699672699,
      "learning_rate": 0.0001921645109955806,
      "loss": 2.0227,
      "step": 142167
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.0226573944091797,
      "learning_rate": 0.0001921606938413978,
      "loss": 1.9279,
      "step": 142168
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.0967075824737549,
      "learning_rate": 0.00019215687670726423,
      "loss": 1.9791,
      "step": 142169
    },
    {
      "epoch": 1.85,
      "grad_norm": 0.9709706902503967,
      "learning_rate": 0.0001921530595931807,
      "loss": 2.1201,
      "step": 142170
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.0417596101760864,
      "learning_rate": 0.000192149242499148,
      "loss": 1.9531,
      "step": 142171
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.1721733808517456,
      "learning_rate": 0.00019214542542516673,
      "loss": 2.1075,
      "step": 142172
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.1280503273010254,
      "learning_rate": 0.00019214160837123772,
      "loss": 2.1998,
      "step": 142173
    },
    {
      "epoch": 1.85,
      "grad_norm": 0.9792914986610413,
      "learning_rate": 0.00019213779133736176,
      "loss": 2.15,
      "step": 142174
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.2936300039291382,
      "learning_rate": 0.00019213397432353923,
      "loss": 2.1593,
      "step": 142175
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.192439317703247,
      "learning_rate": 0.00019213015732977108,
      "loss": 1.9056,
      "step": 142176
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.0627325773239136,
      "learning_rate": 0.00019212634035605796,
      "loss": 2.1092,
      "step": 142177
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.0379512310028076,
      "learning_rate": 0.00019212252340240058,
      "loss": 2.0129,
      "step": 142178
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.0390429496765137,
      "learning_rate": 0.00019211870646879966,
      "loss": 2.3207,
      "step": 142179
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.1464953422546387,
      "learning_rate": 0.00019211488955525602,
      "loss": 2.2424,
      "step": 142180
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.133557677268982,
      "learning_rate": 0.00019211107266177015,
      "loss": 2.1886,
      "step": 142181
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.167876958847046,
      "learning_rate": 0.0001921072557883429,
      "loss": 1.9798,
      "step": 142182
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.079927921295166,
      "learning_rate": 0.0001921034389349749,
      "loss": 1.9812,
      "step": 142183
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.0403186082839966,
      "learning_rate": 0.0001920996221016669,
      "loss": 2.2176,
      "step": 142184
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.2255700826644897,
      "learning_rate": 0.00019209580528841967,
      "loss": 2.1101,
      "step": 142185
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.3818655014038086,
      "learning_rate": 0.00019209198849523394,
      "loss": 2.1173,
      "step": 142186
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.025416374206543,
      "learning_rate": 0.00019208817172211025,
      "loss": 2.1907,
      "step": 142187
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.3290821313858032,
      "learning_rate": 0.00019208435496904938,
      "loss": 2.1592,
      "step": 142188
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.4083149433135986,
      "learning_rate": 0.0001920805382360521,
      "loss": 2.0816,
      "step": 142189
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.0858386754989624,
      "learning_rate": 0.00019207672152311907,
      "loss": 1.8809,
      "step": 142190
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.0995123386383057,
      "learning_rate": 0.000192072904830251,
      "loss": 2.1363,
      "step": 142191
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.0520451068878174,
      "learning_rate": 0.00019206908815744866,
      "loss": 2.1537,
      "step": 142192
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.0773043632507324,
      "learning_rate": 0.00019206527150471278,
      "loss": 2.004,
      "step": 142193
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.0649102926254272,
      "learning_rate": 0.0001920614548720439,
      "loss": 2.1347,
      "step": 142194
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.1369744539260864,
      "learning_rate": 0.00019205763825944286,
      "loss": 2.1237,
      "step": 142195
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.0508463382720947,
      "learning_rate": 0.00019205382166691032,
      "loss": 2.1152,
      "step": 142196
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.163779854774475,
      "learning_rate": 0.000192050005094447,
      "loss": 2.1335,
      "step": 142197
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.112871766090393,
      "learning_rate": 0.00019204618854205363,
      "loss": 2.1807,
      "step": 142198
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.137911081314087,
      "learning_rate": 0.00019204237200973106,
      "loss": 2.1484,
      "step": 142199
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.09650456905365,
      "learning_rate": 0.00019203855549747967,
      "loss": 2.2961,
      "step": 142200
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.0491979122161865,
      "learning_rate": 0.00019203473900530042,
      "loss": 2.0317,
      "step": 142201
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.158687710762024,
      "learning_rate": 0.00019203092253319392,
      "loss": 2.2187,
      "step": 142202
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.1264333724975586,
      "learning_rate": 0.00019202710608116091,
      "loss": 2.1319,
      "step": 142203
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.0840057134628296,
      "learning_rate": 0.0001920232896492021,
      "loss": 2.0083,
      "step": 142204
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.0386677980422974,
      "learning_rate": 0.00019201947323731832,
      "loss": 2.3189,
      "step": 142205
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.2681939601898193,
      "learning_rate": 0.00019201565684551003,
      "loss": 1.8395,
      "step": 142206
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.0918835401535034,
      "learning_rate": 0.00019201184047377807,
      "loss": 2.0611,
      "step": 142207
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.1337776184082031,
      "learning_rate": 0.00019200802412212319,
      "loss": 2.1091,
      "step": 142208
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.0245705842971802,
      "learning_rate": 0.000192004207790546,
      "loss": 2.1939,
      "step": 142209
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.3070920705795288,
      "learning_rate": 0.00019200039147904726,
      "loss": 2.1911,
      "step": 142210
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.591556429862976,
      "learning_rate": 0.00019199657518762777,
      "loss": 2.0597,
      "step": 142211
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.0891170501708984,
      "learning_rate": 0.00019199275891628812,
      "loss": 2.0821,
      "step": 142212
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.1096289157867432,
      "learning_rate": 0.00019198894266502905,
      "loss": 2.1165,
      "step": 142213
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.1333186626434326,
      "learning_rate": 0.00019198512643385125,
      "loss": 2.0933,
      "step": 142214
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.1653276681900024,
      "learning_rate": 0.00019198131022275545,
      "loss": 2.0512,
      "step": 142215
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.1974109411239624,
      "learning_rate": 0.00019197749403174238,
      "loss": 2.0812,
      "step": 142216
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.1299240589141846,
      "learning_rate": 0.0001919736778608128,
      "loss": 2.1634,
      "step": 142217
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.075649619102478,
      "learning_rate": 0.00019196986170996727,
      "loss": 1.9095,
      "step": 142218
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.5184953212738037,
      "learning_rate": 0.0001919660455792066,
      "loss": 2.2393,
      "step": 142219
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.404581069946289,
      "learning_rate": 0.00019196222946853153,
      "loss": 1.8917,
      "step": 142220
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.1108860969543457,
      "learning_rate": 0.00019195841337794266,
      "loss": 2.4135,
      "step": 142221
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.0832856893539429,
      "learning_rate": 0.00019195459730744077,
      "loss": 2.088,
      "step": 142222
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.1758211851119995,
      "learning_rate": 0.0001919507812570266,
      "loss": 2.2039,
      "step": 142223
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.5354819297790527,
      "learning_rate": 0.00019194696522670075,
      "loss": 2.2282,
      "step": 142224
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.1221668720245361,
      "learning_rate": 0.00019194314921646403,
      "loss": 1.9575,
      "step": 142225
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.0792020559310913,
      "learning_rate": 0.00019193933322631708,
      "loss": 2.0505,
      "step": 142226
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.2583495378494263,
      "learning_rate": 0.00019193551725626076,
      "loss": 2.1302,
      "step": 142227
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.3808186054229736,
      "learning_rate": 0.0001919317013062956,
      "loss": 1.9814,
      "step": 142228
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.4269415140151978,
      "learning_rate": 0.0001919278853764224,
      "loss": 2.0998,
      "step": 142229
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.0211223363876343,
      "learning_rate": 0.00019192406946664183,
      "loss": 2.0129,
      "step": 142230
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.3040504455566406,
      "learning_rate": 0.00019192025357695457,
      "loss": 2.0985,
      "step": 142231
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.0926251411437988,
      "learning_rate": 0.0001919164377073614,
      "loss": 1.9609,
      "step": 142232
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.0690988302230835,
      "learning_rate": 0.00019191262185786298,
      "loss": 2.44,
      "step": 142233
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.2626376152038574,
      "learning_rate": 0.00019190880602846012,
      "loss": 2.093,
      "step": 142234
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.0702913999557495,
      "learning_rate": 0.00019190499021915346,
      "loss": 2.0004,
      "step": 142235
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.1292331218719482,
      "learning_rate": 0.00019190117442994365,
      "loss": 2.1743,
      "step": 142236
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.2441259622573853,
      "learning_rate": 0.00019189735866083146,
      "loss": 2.0499,
      "step": 142237
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.1189824342727661,
      "learning_rate": 0.00019189354291181755,
      "loss": 1.9982,
      "step": 142238
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.1252108812332153,
      "learning_rate": 0.00019188972718290273,
      "loss": 2.0979,
      "step": 142239
    },
    {
      "epoch": 1.85,
      "grad_norm": 0.989124059677124,
      "learning_rate": 0.0001918859114740876,
      "loss": 2.1516,
      "step": 142240
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.2408580780029297,
      "learning_rate": 0.00019188209578537306,
      "loss": 2.0174,
      "step": 142241
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.1979142427444458,
      "learning_rate": 0.00019187828011675955,
      "loss": 2.221,
      "step": 142242
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.3341660499572754,
      "learning_rate": 0.00019187446446824793,
      "loss": 2.2324,
      "step": 142243
    },
    {
      "epoch": 1.85,
      "grad_norm": 0.9461930990219116,
      "learning_rate": 0.00019187064883983884,
      "loss": 1.9966,
      "step": 142244
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.0633611679077148,
      "learning_rate": 0.00019186683323153304,
      "loss": 2.2593,
      "step": 142245
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.2011842727661133,
      "learning_rate": 0.00019186301764333128,
      "loss": 2.0846,
      "step": 142246
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.2455201148986816,
      "learning_rate": 0.00019185920207523435,
      "loss": 2.1452,
      "step": 142247
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.1733226776123047,
      "learning_rate": 0.0001918553865272427,
      "loss": 2.155,
      "step": 142248
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.0367234945297241,
      "learning_rate": 0.00019185157099935716,
      "loss": 2.1625,
      "step": 142249
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.1952131986618042,
      "learning_rate": 0.00019184775549157848,
      "loss": 2.053,
      "step": 142250
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.1006827354431152,
      "learning_rate": 0.0001918439400039073,
      "loss": 2.1911,
      "step": 142251
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.0822044610977173,
      "learning_rate": 0.0001918401245363444,
      "loss": 2.3292,
      "step": 142252
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.0285950899124146,
      "learning_rate": 0.00019183630908889063,
      "loss": 2.2049,
      "step": 142253
    },
    {
      "epoch": 1.85,
      "grad_norm": 0.9967040419578552,
      "learning_rate": 0.00019183249366154636,
      "loss": 2.0008,
      "step": 142254
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.1304916143417358,
      "learning_rate": 0.00019182867825431248,
      "loss": 2.0544,
      "step": 142255
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.1437524557113647,
      "learning_rate": 0.0001918248628671897,
      "loss": 1.9016,
      "step": 142256
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.0482712984085083,
      "learning_rate": 0.0001918210475001787,
      "loss": 2.0252,
      "step": 142257
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.065269112586975,
      "learning_rate": 0.0001918172321532802,
      "loss": 1.9786,
      "step": 142258
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.1633541584014893,
      "learning_rate": 0.00019181341682649494,
      "loss": 2.0757,
      "step": 142259
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.2221169471740723,
      "learning_rate": 0.00019180960151982373,
      "loss": 2.2397,
      "step": 142260
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.0646647214889526,
      "learning_rate": 0.00019180578623326705,
      "loss": 2.1179,
      "step": 142261
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.0451765060424805,
      "learning_rate": 0.0001918019709668257,
      "loss": 2.2041,
      "step": 142262
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.2205365896224976,
      "learning_rate": 0.0001917981557205004,
      "loss": 1.9632,
      "step": 142263
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.1228126287460327,
      "learning_rate": 0.00019179434049429187,
      "loss": 2.1236,
      "step": 142264
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.1790324449539185,
      "learning_rate": 0.00019179052528820083,
      "loss": 2.311,
      "step": 142265
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.2650014162063599,
      "learning_rate": 0.00019178671010222807,
      "loss": 2.1063,
      "step": 142266
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.220284342765808,
      "learning_rate": 0.00019178289493637407,
      "loss": 1.9962,
      "step": 142267
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.1899394989013672,
      "learning_rate": 0.00019177907979063972,
      "loss": 1.9843,
      "step": 142268
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.0767430067062378,
      "learning_rate": 0.00019177526466502564,
      "loss": 2.2728,
      "step": 142269
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.0333994626998901,
      "learning_rate": 0.0001917714495595326,
      "loss": 2.1436,
      "step": 142270
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.0882726907730103,
      "learning_rate": 0.0001917676344741613,
      "loss": 2.0216,
      "step": 142271
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.2408829927444458,
      "learning_rate": 0.00019176381940891254,
      "loss": 2.1271,
      "step": 142272
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.0205551385879517,
      "learning_rate": 0.00019176000436378683,
      "loss": 2.0233,
      "step": 142273
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.1248096227645874,
      "learning_rate": 0.00019175618933878495,
      "loss": 2.2416,
      "step": 142274
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.0862960815429688,
      "learning_rate": 0.00019175237433390768,
      "loss": 2.0985,
      "step": 142275
    },
    {
      "epoch": 1.85,
      "grad_norm": 0.9735192656517029,
      "learning_rate": 0.00019174855934915564,
      "loss": 2.0184,
      "step": 142276
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.1982663869857788,
      "learning_rate": 0.00019174474438452961,
      "loss": 2.1613,
      "step": 142277
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.1785633563995361,
      "learning_rate": 0.0001917409294400304,
      "loss": 2.1392,
      "step": 142278
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.0927375555038452,
      "learning_rate": 0.00019173711451565844,
      "loss": 1.9947,
      "step": 142279
    },
    {
      "epoch": 1.85,
      "grad_norm": 0.9737706184387207,
      "learning_rate": 0.00019173329961141465,
      "loss": 2.1554,
      "step": 142280
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.0466506481170654,
      "learning_rate": 0.00019172948472729963,
      "loss": 1.8644,
      "step": 142281
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.0502411127090454,
      "learning_rate": 0.00019172566986331416,
      "loss": 2.3844,
      "step": 142282
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.0698720216751099,
      "learning_rate": 0.00019172185501945896,
      "loss": 2.2066,
      "step": 142283
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.1121870279312134,
      "learning_rate": 0.00019171804019573483,
      "loss": 1.9684,
      "step": 142284
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.034817099571228,
      "learning_rate": 0.00019171422539214223,
      "loss": 2.234,
      "step": 142285
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.1200188398361206,
      "learning_rate": 0.00019171041060868198,
      "loss": 1.8939,
      "step": 142286
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.0802487134933472,
      "learning_rate": 0.0001917065958453548,
      "loss": 2.06,
      "step": 142287
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.2019729614257812,
      "learning_rate": 0.00019170278110216141,
      "loss": 1.7505,
      "step": 142288
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.0409616231918335,
      "learning_rate": 0.00019169896637910257,
      "loss": 2.1612,
      "step": 142289
    },
    {
      "epoch": 1.85,
      "grad_norm": 0.9980539083480835,
      "learning_rate": 0.000191695151676179,
      "loss": 1.9658,
      "step": 142290
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.0515389442443848,
      "learning_rate": 0.00019169133699339125,
      "loss": 2.0415,
      "step": 142291
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.171157717704773,
      "learning_rate": 0.00019168752233074014,
      "loss": 1.9563,
      "step": 142292
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.101455569267273,
      "learning_rate": 0.00019168370768822633,
      "loss": 2.0536,
      "step": 142293
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.2162413597106934,
      "learning_rate": 0.0001916798930658506,
      "loss": 2.0513,
      "step": 142294
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.169083833694458,
      "learning_rate": 0.00019167607846361356,
      "loss": 2.0705,
      "step": 142295
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.1898307800292969,
      "learning_rate": 0.0001916722638815161,
      "loss": 2.1094,
      "step": 142296
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.01272451877594,
      "learning_rate": 0.0001916684493195588,
      "loss": 2.0187,
      "step": 142297
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.3552160263061523,
      "learning_rate": 0.00019166463477774233,
      "loss": 1.8836,
      "step": 142298
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.1864432096481323,
      "learning_rate": 0.00019166082025606743,
      "loss": 1.9363,
      "step": 142299
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.0470991134643555,
      "learning_rate": 0.0001916570057545348,
      "loss": 1.9791,
      "step": 142300
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.0289876461029053,
      "learning_rate": 0.00019165319127314523,
      "loss": 2.0385,
      "step": 142301
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.1516486406326294,
      "learning_rate": 0.0001916493768118994,
      "loss": 1.9037,
      "step": 142302
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.094301700592041,
      "learning_rate": 0.00019164556237079796,
      "loss": 2.1094,
      "step": 142303
    },
    {
      "epoch": 1.85,
      "grad_norm": 0.9959430694580078,
      "learning_rate": 0.00019164174794984167,
      "loss": 2.3516,
      "step": 142304
    },
    {
      "epoch": 1.85,
      "grad_norm": 0.9761258363723755,
      "learning_rate": 0.00019163793354903125,
      "loss": 2.199,
      "step": 142305
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.0789551734924316,
      "learning_rate": 0.00019163411916836733,
      "loss": 2.154,
      "step": 142306
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.0562390089035034,
      "learning_rate": 0.00019163030480785068,
      "loss": 2.2392,
      "step": 142307
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.1069501638412476,
      "learning_rate": 0.00019162649046748206,
      "loss": 2.0359,
      "step": 142308
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.0062226057052612,
      "learning_rate": 0.00019162267614726206,
      "loss": 2.1164,
      "step": 142309
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.066880464553833,
      "learning_rate": 0.00019161886184719147,
      "loss": 2.0799,
      "step": 142310
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.1171387434005737,
      "learning_rate": 0.00019161504756727096,
      "loss": 1.8569,
      "step": 142311
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.0597505569458008,
      "learning_rate": 0.00019161123330750134,
      "loss": 2.0709,
      "step": 142312
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.0393109321594238,
      "learning_rate": 0.00019160741906788315,
      "loss": 1.9223,
      "step": 142313
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.039803147315979,
      "learning_rate": 0.00019160360484841724,
      "loss": 2.09,
      "step": 142314
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.0762381553649902,
      "learning_rate": 0.00019159979064910425,
      "loss": 1.9797,
      "step": 142315
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.1891417503356934,
      "learning_rate": 0.00019159597646994486,
      "loss": 1.9444,
      "step": 142316
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.1434974670410156,
      "learning_rate": 0.00019159216231093984,
      "loss": 2.019,
      "step": 142317
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.1158530712127686,
      "learning_rate": 0.00019158834817208987,
      "loss": 2.0731,
      "step": 142318
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.095450520515442,
      "learning_rate": 0.00019158453405339573,
      "loss": 1.956,
      "step": 142319
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.1847522258758545,
      "learning_rate": 0.00019158071995485813,
      "loss": 2.1676,
      "step": 142320
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.0062353610992432,
      "learning_rate": 0.00019157690587647761,
      "loss": 1.9883,
      "step": 142321
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.0300872325897217,
      "learning_rate": 0.000191573091818255,
      "loss": 1.9882,
      "step": 142322
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.2778894901275635,
      "learning_rate": 0.000191569277780191,
      "loss": 2.0176,
      "step": 142323
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.340659737586975,
      "learning_rate": 0.00019156546376228632,
      "loss": 1.9514,
      "step": 142324
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.1296923160552979,
      "learning_rate": 0.00019156164976454166,
      "loss": 2.1429,
      "step": 142325
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.049828052520752,
      "learning_rate": 0.00019155783578695779,
      "loss": 2.0591,
      "step": 142326
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.5532255172729492,
      "learning_rate": 0.0001915540218295354,
      "loss": 1.9148,
      "step": 142327
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.4757788181304932,
      "learning_rate": 0.00019155020789227506,
      "loss": 1.7885,
      "step": 142328
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.0578747987747192,
      "learning_rate": 0.00019154639397517758,
      "loss": 2.0945,
      "step": 142329
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.0504395961761475,
      "learning_rate": 0.00019154258007824372,
      "loss": 1.9727,
      "step": 142330
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.3020555973052979,
      "learning_rate": 0.00019153876620147412,
      "loss": 2.0321,
      "step": 142331
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.3236017227172852,
      "learning_rate": 0.0001915349523448695,
      "loss": 1.926,
      "step": 142332
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.1350206136703491,
      "learning_rate": 0.0001915311385084307,
      "loss": 1.9797,
      "step": 142333
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.0172151327133179,
      "learning_rate": 0.0001915273246921582,
      "loss": 1.8557,
      "step": 142334
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.309926152229309,
      "learning_rate": 0.00019152351089605281,
      "loss": 2.0885,
      "step": 142335
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.151642918586731,
      "learning_rate": 0.00019151969712011524,
      "loss": 1.9961,
      "step": 142336
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.0029857158660889,
      "learning_rate": 0.00019151588336434626,
      "loss": 2.0232,
      "step": 142337
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.0195109844207764,
      "learning_rate": 0.00019151206962874645,
      "loss": 2.0247,
      "step": 142338
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.0889837741851807,
      "learning_rate": 0.00019150825591331673,
      "loss": 2.1579,
      "step": 142339
    },
    {
      "epoch": 1.85,
      "grad_norm": 0.9597059488296509,
      "learning_rate": 0.0001915044422180576,
      "loss": 1.9567,
      "step": 142340
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.1652153730392456,
      "learning_rate": 0.0001915006285429698,
      "loss": 1.8759,
      "step": 142341
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.1014773845672607,
      "learning_rate": 0.00019149681488805407,
      "loss": 1.9669,
      "step": 142342
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.162956953048706,
      "learning_rate": 0.00019149300125331114,
      "loss": 2.2116,
      "step": 142343
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.154775857925415,
      "learning_rate": 0.00019148918763874174,
      "loss": 2.0078,
      "step": 142344
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.0514147281646729,
      "learning_rate": 0.00019148537404434668,
      "loss": 2.084,
      "step": 142345
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.069784164428711,
      "learning_rate": 0.00019148156047012638,
      "loss": 2.1791,
      "step": 142346
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.0825954675674438,
      "learning_rate": 0.0001914777469160817,
      "loss": 2.1287,
      "step": 142347
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.0524060726165771,
      "learning_rate": 0.00019147393338221337,
      "loss": 2.0385,
      "step": 142348
    },
    {
      "epoch": 1.85,
      "grad_norm": 0.9631765484809875,
      "learning_rate": 0.00019147011986852213,
      "loss": 1.9878,
      "step": 142349
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.0413633584976196,
      "learning_rate": 0.00019146630637500856,
      "loss": 2.1076,
      "step": 142350
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.0780727863311768,
      "learning_rate": 0.00019146249290167362,
      "loss": 2.0921,
      "step": 142351
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.0517627000808716,
      "learning_rate": 0.00019145867944851772,
      "loss": 2.3373,
      "step": 142352
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.11483895778656,
      "learning_rate": 0.00019145486601554172,
      "loss": 2.2388,
      "step": 142353
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.2326527833938599,
      "learning_rate": 0.00019145105260274625,
      "loss": 2.1194,
      "step": 142354
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.1743905544281006,
      "learning_rate": 0.00019144723921013214,
      "loss": 2.1019,
      "step": 142355
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.0691604614257812,
      "learning_rate": 0.0001914434258377,
      "loss": 2.0499,
      "step": 142356
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.078495979309082,
      "learning_rate": 0.00019143961248545074,
      "loss": 1.7969,
      "step": 142357
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.207714557647705,
      "learning_rate": 0.00019143579915338474,
      "loss": 1.9905,
      "step": 142358
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.0360612869262695,
      "learning_rate": 0.00019143198584150293,
      "loss": 2.1736,
      "step": 142359
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.0246928930282593,
      "learning_rate": 0.0001914281725498059,
      "loss": 2.0287,
      "step": 142360
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.051167607307434,
      "learning_rate": 0.00019142435927829441,
      "loss": 2.2765,
      "step": 142361
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.0860397815704346,
      "learning_rate": 0.00019142054602696925,
      "loss": 2.1221,
      "step": 142362
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.0834314823150635,
      "learning_rate": 0.00019141673279583115,
      "loss": 2.2736,
      "step": 142363
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.032712697982788,
      "learning_rate": 0.00019141291958488062,
      "loss": 1.9188,
      "step": 142364
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.0660799741744995,
      "learning_rate": 0.00019140910639411847,
      "loss": 2.2227,
      "step": 142365
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.0706104040145874,
      "learning_rate": 0.0001914052932235454,
      "loss": 2.2239,
      "step": 142366
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.0717896223068237,
      "learning_rate": 0.00019140148007316214,
      "loss": 2.2678,
      "step": 142367
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.0267055034637451,
      "learning_rate": 0.0001913976669429694,
      "loss": 2.0788,
      "step": 142368
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.250252604484558,
      "learning_rate": 0.00019139385383296805,
      "loss": 1.8932,
      "step": 142369
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.0432988405227661,
      "learning_rate": 0.00019139004074315844,
      "loss": 1.8404,
      "step": 142370
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.087255835533142,
      "learning_rate": 0.0001913862276735415,
      "loss": 2.1105,
      "step": 142371
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.1633926630020142,
      "learning_rate": 0.0001913824146241179,
      "loss": 2.0077,
      "step": 142372
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.103399395942688,
      "learning_rate": 0.00019137860159488838,
      "loss": 2.1314,
      "step": 142373
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.0930085182189941,
      "learning_rate": 0.00019137478858585364,
      "loss": 2.106,
      "step": 142374
    },
    {
      "epoch": 1.85,
      "grad_norm": 0.98341965675354,
      "learning_rate": 0.00019137097559701448,
      "loss": 2.2268,
      "step": 142375
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.094287633895874,
      "learning_rate": 0.00019136716262837137,
      "loss": 2.3488,
      "step": 142376
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.0566444396972656,
      "learning_rate": 0.00019136334967992518,
      "loss": 2.147,
      "step": 142377
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.0717661380767822,
      "learning_rate": 0.0001913595367516766,
      "loss": 1.9841,
      "step": 142378
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.0758142471313477,
      "learning_rate": 0.0001913557238436263,
      "loss": 1.998,
      "step": 142379
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.182242512702942,
      "learning_rate": 0.00019135191095577504,
      "loss": 1.9322,
      "step": 142380
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.2297333478927612,
      "learning_rate": 0.00019134809808812357,
      "loss": 2.0695,
      "step": 142381
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.101396918296814,
      "learning_rate": 0.00019134428524067253,
      "loss": 1.9916,
      "step": 142382
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.0785657167434692,
      "learning_rate": 0.00019134047241342258,
      "loss": 1.9608,
      "step": 142383
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.0709409713745117,
      "learning_rate": 0.0001913366596063745,
      "loss": 2.1942,
      "step": 142384
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.0234752893447876,
      "learning_rate": 0.00019133284681952897,
      "loss": 2.1222,
      "step": 142385
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.056045651435852,
      "learning_rate": 0.00019132903405288674,
      "loss": 2.2237,
      "step": 142386
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.1137346029281616,
      "learning_rate": 0.00019132522130644854,
      "loss": 1.9767,
      "step": 142387
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.0802733898162842,
      "learning_rate": 0.00019132140858021496,
      "loss": 2.0781,
      "step": 142388
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.0683419704437256,
      "learning_rate": 0.00019131759587418686,
      "loss": 2.0098,
      "step": 142389
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.0532382726669312,
      "learning_rate": 0.0001913137831883648,
      "loss": 2.1292,
      "step": 142390
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.1948518753051758,
      "learning_rate": 0.00019130997052274956,
      "loss": 2.1329,
      "step": 142391
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.0763859748840332,
      "learning_rate": 0.00019130615787734184,
      "loss": 2.0244,
      "step": 142392
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.2235755920410156,
      "learning_rate": 0.00019130234525214238,
      "loss": 2.1064,
      "step": 142393
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.066975474357605,
      "learning_rate": 0.0001912985326471519,
      "loss": 1.9856,
      "step": 142394
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.1939774751663208,
      "learning_rate": 0.00019129472006237103,
      "loss": 2.1182,
      "step": 142395
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.206985354423523,
      "learning_rate": 0.00019129090749780053,
      "loss": 1.8514,
      "step": 142396
    },
    {
      "epoch": 1.85,
      "grad_norm": 0.9817204475402832,
      "learning_rate": 0.00019128709495344113,
      "loss": 2.1366,
      "step": 142397
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.1442936658859253,
      "learning_rate": 0.00019128328242929349,
      "loss": 2.0415,
      "step": 142398
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.4097892045974731,
      "learning_rate": 0.00019127946992535832,
      "loss": 2.2341,
      "step": 142399
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.3290307521820068,
      "learning_rate": 0.0001912756574416364,
      "loss": 1.9782,
      "step": 142400
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.1019848585128784,
      "learning_rate": 0.00019127184497812837,
      "loss": 2.3026,
      "step": 142401
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.187417984008789,
      "learning_rate": 0.0001912680325348349,
      "loss": 2.0891,
      "step": 142402
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.1999602317810059,
      "learning_rate": 0.00019126422011175683,
      "loss": 2.0518,
      "step": 142403
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.05930757522583,
      "learning_rate": 0.00019126040770889478,
      "loss": 1.9997,
      "step": 142404
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.028077483177185,
      "learning_rate": 0.00019125659532624945,
      "loss": 2.064,
      "step": 142405
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.1224792003631592,
      "learning_rate": 0.00019125278296382162,
      "loss": 2.233,
      "step": 142406
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.224463701248169,
      "learning_rate": 0.00019124897062161188,
      "loss": 2.1049,
      "step": 142407
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.0386815071105957,
      "learning_rate": 0.000191245158299621,
      "loss": 2.2366,
      "step": 142408
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.157585620880127,
      "learning_rate": 0.00019124134599784974,
      "loss": 2.0096,
      "step": 142409
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.0558606386184692,
      "learning_rate": 0.00019123753371629875,
      "loss": 2.1566,
      "step": 142410
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.0730760097503662,
      "learning_rate": 0.00019123372145496884,
      "loss": 1.9873,
      "step": 142411
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.014121651649475,
      "learning_rate": 0.00019122990921386057,
      "loss": 2.0499,
      "step": 142412
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.1001101732254028,
      "learning_rate": 0.0001912260969929747,
      "loss": 2.1687,
      "step": 142413
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.1523247957229614,
      "learning_rate": 0.00019122228479231192,
      "loss": 1.8909,
      "step": 142414
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.0795122385025024,
      "learning_rate": 0.000191218472611873,
      "loss": 2.2077,
      "step": 142415
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.039581298828125,
      "learning_rate": 0.00019121466045165863,
      "loss": 2.1407,
      "step": 142416
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.0513101816177368,
      "learning_rate": 0.00019121084831166948,
      "loss": 2.0937,
      "step": 142417
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.1381754875183105,
      "learning_rate": 0.00019120703619190642,
      "loss": 2.0224,
      "step": 142418
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.056443214416504,
      "learning_rate": 0.00019120322409236988,
      "loss": 2.054,
      "step": 142419
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.1409525871276855,
      "learning_rate": 0.00019119941201306076,
      "loss": 2.1226,
      "step": 142420
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.132054090499878,
      "learning_rate": 0.0001911955999539797,
      "loss": 2.1765,
      "step": 142421
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.1384938955307007,
      "learning_rate": 0.00019119178791512738,
      "loss": 2.0251,
      "step": 142422
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.0604968070983887,
      "learning_rate": 0.00019118797589650464,
      "loss": 1.894,
      "step": 142423
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.0316646099090576,
      "learning_rate": 0.0001911841638981122,
      "loss": 1.9137,
      "step": 142424
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.1912177801132202,
      "learning_rate": 0.00019118035191995056,
      "loss": 2.0256,
      "step": 142425
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.0248571634292603,
      "learning_rate": 0.00019117653996202056,
      "loss": 2.042,
      "step": 142426
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.138644814491272,
      "learning_rate": 0.00019117272802432288,
      "loss": 2.0459,
      "step": 142427
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.0404314994812012,
      "learning_rate": 0.00019116891610685823,
      "loss": 1.9976,
      "step": 142428
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.0278480052947998,
      "learning_rate": 0.00019116510420962734,
      "loss": 2.0474,
      "step": 142429
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.0098189115524292,
      "learning_rate": 0.00019116129233263105,
      "loss": 2.0334,
      "step": 142430
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.0406023263931274,
      "learning_rate": 0.0001911574804758698,
      "loss": 1.9449,
      "step": 142431
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.0312449932098389,
      "learning_rate": 0.00019115366863934442,
      "loss": 1.9757,
      "step": 142432
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.2120064496994019,
      "learning_rate": 0.00019114985682305564,
      "loss": 1.8417,
      "step": 142433
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.1990772485733032,
      "learning_rate": 0.00019114604502700415,
      "loss": 2.2264,
      "step": 142434
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.04090416431427,
      "learning_rate": 0.00019114223325119068,
      "loss": 1.9274,
      "step": 142435
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.1961182355880737,
      "learning_rate": 0.00019113842149561604,
      "loss": 2.0128,
      "step": 142436
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.1415952444076538,
      "learning_rate": 0.00019113460976028068,
      "loss": 2.2778,
      "step": 142437
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.107477068901062,
      "learning_rate": 0.00019113079804518546,
      "loss": 1.888,
      "step": 142438
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.243024468421936,
      "learning_rate": 0.0001911269863503311,
      "loss": 2.062,
      "step": 142439
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.401882290840149,
      "learning_rate": 0.00019112317467571827,
      "loss": 1.9308,
      "step": 142440
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.0088697671890259,
      "learning_rate": 0.0001911193630213477,
      "loss": 1.9291,
      "step": 142441
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.2578904628753662,
      "learning_rate": 0.0001911155513872202,
      "loss": 2.1615,
      "step": 142442
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.049325942993164,
      "learning_rate": 0.00019111173977333628,
      "loss": 1.9902,
      "step": 142443
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.065913438796997,
      "learning_rate": 0.00019110792817969674,
      "loss": 1.8936,
      "step": 142444
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.1353518962860107,
      "learning_rate": 0.00019110411660630227,
      "loss": 2.0396,
      "step": 142445
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.1013269424438477,
      "learning_rate": 0.0001911003050531536,
      "loss": 1.9277,
      "step": 142446
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.1964021921157837,
      "learning_rate": 0.00019109649352025147,
      "loss": 2.2024,
      "step": 142447
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.120521903038025,
      "learning_rate": 0.00019109268200759666,
      "loss": 2.1131,
      "step": 142448
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.0985039472579956,
      "learning_rate": 0.00019108887051518965,
      "loss": 2.1107,
      "step": 142449
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.1350713968276978,
      "learning_rate": 0.00019108505904303128,
      "loss": 2.1864,
      "step": 142450
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.0687596797943115,
      "learning_rate": 0.00019108124759112226,
      "loss": 2.2802,
      "step": 142451
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.1227127313613892,
      "learning_rate": 0.00019107743615946328,
      "loss": 2.1324,
      "step": 142452
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.109065294265747,
      "learning_rate": 0.0001910736247480551,
      "loss": 2.1708,
      "step": 142453
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.088241696357727,
      "learning_rate": 0.00019106981335689844,
      "loss": 2.0052,
      "step": 142454
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.0481793880462646,
      "learning_rate": 0.00019106600198599388,
      "loss": 2.027,
      "step": 142455
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.1846832036972046,
      "learning_rate": 0.0001910621906353422,
      "loss": 2.3037,
      "step": 142456
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.045778751373291,
      "learning_rate": 0.00019105837930494414,
      "loss": 2.0449,
      "step": 142457
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.1299325227737427,
      "learning_rate": 0.00019105456799480035,
      "loss": 1.9193,
      "step": 142458
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.0805946588516235,
      "learning_rate": 0.00019105075670491158,
      "loss": 1.9833,
      "step": 142459
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.1578119993209839,
      "learning_rate": 0.00019104694543527854,
      "loss": 2.105,
      "step": 142460
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.0293501615524292,
      "learning_rate": 0.00019104313418590203,
      "loss": 1.9014,
      "step": 142461
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.0537832975387573,
      "learning_rate": 0.00019103932295678257,
      "loss": 1.9397,
      "step": 142462
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.008180022239685,
      "learning_rate": 0.00019103551174792094,
      "loss": 2.1721,
      "step": 142463
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.0656596422195435,
      "learning_rate": 0.00019103170055931787,
      "loss": 1.9782,
      "step": 142464
    },
    {
      "epoch": 1.85,
      "grad_norm": 1.102799415588379,
      "learning_rate": 0.00019102788939097403,
      "loss": 2.2175,
      "step": 142465
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.1119356155395508,
      "learning_rate": 0.0001910240782428902,
      "loss": 2.2076,
      "step": 142466
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.0226701498031616,
      "learning_rate": 0.00019102026711506715,
      "loss": 2.114,
      "step": 142467
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.0049382448196411,
      "learning_rate": 0.0001910164560075054,
      "loss": 2.0788,
      "step": 142468
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.1124491691589355,
      "learning_rate": 0.0001910126449202057,
      "loss": 2.0371,
      "step": 142469
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.1012845039367676,
      "learning_rate": 0.00019100883385316883,
      "loss": 2.1808,
      "step": 142470
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.0264887809753418,
      "learning_rate": 0.0001910050228063955,
      "loss": 1.9828,
      "step": 142471
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.1687995195388794,
      "learning_rate": 0.00019100121177988638,
      "loss": 1.9942,
      "step": 142472
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.0264310836791992,
      "learning_rate": 0.00019099740077364226,
      "loss": 2.1248,
      "step": 142473
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.1357955932617188,
      "learning_rate": 0.00019099358978766373,
      "loss": 2.0582,
      "step": 142474
    },
    {
      "epoch": 1.86,
      "grad_norm": 0.9975631237030029,
      "learning_rate": 0.00019098977882195153,
      "loss": 1.9818,
      "step": 142475
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.064225196838379,
      "learning_rate": 0.0001909859678765064,
      "loss": 2.1992,
      "step": 142476
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.0707736015319824,
      "learning_rate": 0.00019098215695132902,
      "loss": 2.011,
      "step": 142477
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.1688013076782227,
      "learning_rate": 0.0001909783460464201,
      "loss": 2.1413,
      "step": 142478
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.0938925743103027,
      "learning_rate": 0.00019097453516178046,
      "loss": 2.1368,
      "step": 142479
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.1170399188995361,
      "learning_rate": 0.00019097072429741062,
      "loss": 2.0922,
      "step": 142480
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.1317598819732666,
      "learning_rate": 0.00019096691345331143,
      "loss": 2.1732,
      "step": 142481
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.0220539569854736,
      "learning_rate": 0.00019096310262948355,
      "loss": 1.952,
      "step": 142482
    },
    {
      "epoch": 1.86,
      "grad_norm": 0.9888713955879211,
      "learning_rate": 0.0001909592918259276,
      "loss": 2.0463,
      "step": 142483
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.1681082248687744,
      "learning_rate": 0.00019095548104264442,
      "loss": 2.328,
      "step": 142484
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.1843664646148682,
      "learning_rate": 0.00019095167027963474,
      "loss": 2.2903,
      "step": 142485
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.1600215435028076,
      "learning_rate": 0.00019094785953689916,
      "loss": 2.2265,
      "step": 142486
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.1181303262710571,
      "learning_rate": 0.00019094404881443843,
      "loss": 2.3033,
      "step": 142487
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.1385114192962646,
      "learning_rate": 0.0001909402381122533,
      "loss": 2.0134,
      "step": 142488
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.0218967199325562,
      "learning_rate": 0.00019093642743034438,
      "loss": 2.1398,
      "step": 142489
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.1132773160934448,
      "learning_rate": 0.00019093261676871243,
      "loss": 2.1644,
      "step": 142490
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.12223219871521,
      "learning_rate": 0.00019092880612735824,
      "loss": 1.8672,
      "step": 142491
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.0810507535934448,
      "learning_rate": 0.00019092499550628237,
      "loss": 2.0368,
      "step": 142492
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.078383445739746,
      "learning_rate": 0.0001909211849054856,
      "loss": 1.9737,
      "step": 142493
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.2962578535079956,
      "learning_rate": 0.00019091737432496865,
      "loss": 1.8624,
      "step": 142494
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.1336382627487183,
      "learning_rate": 0.00019091356376473218,
      "loss": 2.2291,
      "step": 142495
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.1258420944213867,
      "learning_rate": 0.00019090975322477708,
      "loss": 1.9603,
      "step": 142496
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.160763144493103,
      "learning_rate": 0.0001909059427051039,
      "loss": 1.9598,
      "step": 142497
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.1548633575439453,
      "learning_rate": 0.00019090213220571326,
      "loss": 2.1468,
      "step": 142498
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.0222365856170654,
      "learning_rate": 0.00019089832172660594,
      "loss": 1.9046,
      "step": 142499
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.006366491317749,
      "learning_rate": 0.00019089451126778274,
      "loss": 2.0069,
      "step": 142500
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.1069506406784058,
      "learning_rate": 0.0001908907008292443,
      "loss": 2.2284,
      "step": 142501
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.0859402418136597,
      "learning_rate": 0.00019088689041099136,
      "loss": 2.3842,
      "step": 142502
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.0439940690994263,
      "learning_rate": 0.0001908830800130247,
      "loss": 2.0433,
      "step": 142503
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.05539071559906,
      "learning_rate": 0.00019087926963534487,
      "loss": 1.9562,
      "step": 142504
    },
    {
      "epoch": 1.86,
      "grad_norm": 0.9699229598045349,
      "learning_rate": 0.00019087545927795255,
      "loss": 1.9417,
      "step": 142505
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.0058141946792603,
      "learning_rate": 0.0001908716489408486,
      "loss": 2.1237,
      "step": 142506
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.1091125011444092,
      "learning_rate": 0.00019086783862403363,
      "loss": 2.088,
      "step": 142507
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.0352586507797241,
      "learning_rate": 0.00019086402832750845,
      "loss": 1.9888,
      "step": 142508
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.2290804386138916,
      "learning_rate": 0.00019086021805127379,
      "loss": 2.0894,
      "step": 142509
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.1158632040023804,
      "learning_rate": 0.00019085640779533015,
      "loss": 2.2065,
      "step": 142510
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.1431325674057007,
      "learning_rate": 0.00019085259755967837,
      "loss": 2.0385,
      "step": 142511
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.4727356433868408,
      "learning_rate": 0.00019084878734431914,
      "loss": 2.0624,
      "step": 142512
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.2182471752166748,
      "learning_rate": 0.0001908449771492532,
      "loss": 2.1852,
      "step": 142513
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.1898679733276367,
      "learning_rate": 0.00019084116697448125,
      "loss": 2.2524,
      "step": 142514
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.0400437116622925,
      "learning_rate": 0.00019083735682000406,
      "loss": 2.1938,
      "step": 142515
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.1806739568710327,
      "learning_rate": 0.00019083354668582222,
      "loss": 1.9692,
      "step": 142516
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.0731072425842285,
      "learning_rate": 0.00019082973657193644,
      "loss": 2.0389,
      "step": 142517
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.019938349723816,
      "learning_rate": 0.00019082592647834748,
      "loss": 2.0129,
      "step": 142518
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.1451853513717651,
      "learning_rate": 0.00019082211640505603,
      "loss": 2.0901,
      "step": 142519
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.0333610773086548,
      "learning_rate": 0.0001908183063520628,
      "loss": 1.9191,
      "step": 142520
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.1403114795684814,
      "learning_rate": 0.00019081449631936864,
      "loss": 2.0912,
      "step": 142521
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.0327210426330566,
      "learning_rate": 0.00019081068630697403,
      "loss": 1.8825,
      "step": 142522
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.1412068605422974,
      "learning_rate": 0.00019080687631487975,
      "loss": 2.1745,
      "step": 142523
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.2197002172470093,
      "learning_rate": 0.00019080306634308654,
      "loss": 2.1295,
      "step": 142524
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.0875424146652222,
      "learning_rate": 0.0001907992563915951,
      "loss": 2.0829,
      "step": 142525
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.0919814109802246,
      "learning_rate": 0.00019079544646040614,
      "loss": 2.048,
      "step": 142526
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.2170637845993042,
      "learning_rate": 0.0001907916365495204,
      "loss": 2.3837,
      "step": 142527
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.258664608001709,
      "learning_rate": 0.0001907878266589386,
      "loss": 2.1082,
      "step": 142528
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.0284926891326904,
      "learning_rate": 0.00019078401678866133,
      "loss": 1.9312,
      "step": 142529
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.055773138999939,
      "learning_rate": 0.00019078020693868937,
      "loss": 2.0213,
      "step": 142530
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.1605664491653442,
      "learning_rate": 0.00019077639710902346,
      "loss": 2.1082,
      "step": 142531
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.1042863130569458,
      "learning_rate": 0.0001907725872996642,
      "loss": 2.1238,
      "step": 142532
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.0729864835739136,
      "learning_rate": 0.00019076877751061244,
      "loss": 2.1418,
      "step": 142533
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.012272596359253,
      "learning_rate": 0.00019076496774186896,
      "loss": 1.9715,
      "step": 142534
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.2418313026428223,
      "learning_rate": 0.00019076115799343416,
      "loss": 1.9629,
      "step": 142535
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.2265746593475342,
      "learning_rate": 0.00019075734826530898,
      "loss": 1.9317,
      "step": 142536
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.0883653163909912,
      "learning_rate": 0.00019075353855749403,
      "loss": 2.1348,
      "step": 142537
    },
    {
      "epoch": 1.86,
      "grad_norm": 0.9954453110694885,
      "learning_rate": 0.0001907497288699901,
      "loss": 1.7553,
      "step": 142538
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.2371463775634766,
      "learning_rate": 0.0001907459192027978,
      "loss": 2.1035,
      "step": 142539
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.0491759777069092,
      "learning_rate": 0.00019074210955591803,
      "loss": 1.9577,
      "step": 142540
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.0650159120559692,
      "learning_rate": 0.0001907382999293513,
      "loss": 2.1377,
      "step": 142541
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.1086955070495605,
      "learning_rate": 0.0001907344903230983,
      "loss": 2.0873,
      "step": 142542
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.104831576347351,
      "learning_rate": 0.0001907306807371599,
      "loss": 2.2569,
      "step": 142543
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.19536554813385,
      "learning_rate": 0.00019072687117153671,
      "loss": 2.0609,
      "step": 142544
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.0380687713623047,
      "learning_rate": 0.00019072306162622944,
      "loss": 1.9305,
      "step": 142545
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.0230318307876587,
      "learning_rate": 0.0001907192521012389,
      "loss": 2.2213,
      "step": 142546
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.0477404594421387,
      "learning_rate": 0.00019071544259656567,
      "loss": 2.0729,
      "step": 142547
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.0249512195587158,
      "learning_rate": 0.00019071163311221043,
      "loss": 2.1992,
      "step": 142548
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.1668206453323364,
      "learning_rate": 0.000190707823648174,
      "loss": 2.1937,
      "step": 142549
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.0909420251846313,
      "learning_rate": 0.00019070401420445704,
      "loss": 2.139,
      "step": 142550
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.1528716087341309,
      "learning_rate": 0.00019070020478106027,
      "loss": 2.3377,
      "step": 142551
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.2000941038131714,
      "learning_rate": 0.0001906963953779845,
      "loss": 1.8964,
      "step": 142552
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.1324456930160522,
      "learning_rate": 0.00019069258599523024,
      "loss": 2.1704,
      "step": 142553
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.0811748504638672,
      "learning_rate": 0.0001906887766327983,
      "loss": 2.0663,
      "step": 142554
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.0925285816192627,
      "learning_rate": 0.00019068496729068935,
      "loss": 2.2592,
      "step": 142555
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.0092698335647583,
      "learning_rate": 0.00019068115796890412,
      "loss": 2.2483,
      "step": 142556
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.142798662185669,
      "learning_rate": 0.00019067734866744332,
      "loss": 1.9681,
      "step": 142557
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.0791603326797485,
      "learning_rate": 0.00019067353938630776,
      "loss": 2.1497,
      "step": 142558
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.1752994060516357,
      "learning_rate": 0.00019066973012549807,
      "loss": 1.9967,
      "step": 142559
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.0213429927825928,
      "learning_rate": 0.00019066592088501485,
      "loss": 2.0082,
      "step": 142560
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.0677244663238525,
      "learning_rate": 0.00019066211166485886,
      "loss": 2.1283,
      "step": 142561
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.020384430885315,
      "learning_rate": 0.0001906583024650309,
      "loss": 2.2486,
      "step": 142562
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.006990671157837,
      "learning_rate": 0.00019065449328553162,
      "loss": 1.892,
      "step": 142563
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.0355725288391113,
      "learning_rate": 0.0001906506841263618,
      "loss": 2.0008,
      "step": 142564
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.164330244064331,
      "learning_rate": 0.000190646874987522,
      "loss": 1.9872,
      "step": 142565
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.1673680543899536,
      "learning_rate": 0.00019064306586901307,
      "loss": 2.3174,
      "step": 142566
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.2589538097381592,
      "learning_rate": 0.00019063925677083562,
      "loss": 1.9274,
      "step": 142567
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.0677282810211182,
      "learning_rate": 0.00019063544769299035,
      "loss": 2.0339,
      "step": 142568
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.1146016120910645,
      "learning_rate": 0.00019063163863547808,
      "loss": 2.2656,
      "step": 142569
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.2258350849151611,
      "learning_rate": 0.00019062782959829947,
      "loss": 1.9679,
      "step": 142570
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.2624924182891846,
      "learning_rate": 0.00019062402058145513,
      "loss": 1.8529,
      "step": 142571
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.0438475608825684,
      "learning_rate": 0.0001906202115849459,
      "loss": 1.9497,
      "step": 142572
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.0624544620513916,
      "learning_rate": 0.00019061640260877247,
      "loss": 2.0004,
      "step": 142573
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.2123278379440308,
      "learning_rate": 0.00019061259365293546,
      "loss": 2.1337,
      "step": 142574
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.345031976699829,
      "learning_rate": 0.00019060878471743566,
      "loss": 2.1064,
      "step": 142575
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.0683960914611816,
      "learning_rate": 0.0001906049758022738,
      "loss": 2.0351,
      "step": 142576
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.0744541883468628,
      "learning_rate": 0.00019060116690745044,
      "loss": 2.1749,
      "step": 142577
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.3280216455459595,
      "learning_rate": 0.00019059735803296642,
      "loss": 2.1889,
      "step": 142578
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.0815255641937256,
      "learning_rate": 0.00019059354917882242,
      "loss": 2.0833,
      "step": 142579
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.0819180011749268,
      "learning_rate": 0.0001905897403450192,
      "loss": 2.3351,
      "step": 142580
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.0493347644805908,
      "learning_rate": 0.00019058593153155735,
      "loss": 2.1765,
      "step": 142581
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.0229294300079346,
      "learning_rate": 0.0001905821227384377,
      "loss": 1.9578,
      "step": 142582
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.0790560245513916,
      "learning_rate": 0.00019057831396566085,
      "loss": 2.1368,
      "step": 142583
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.07548189163208,
      "learning_rate": 0.00019057450521322753,
      "loss": 1.9154,
      "step": 142584
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.0415207147598267,
      "learning_rate": 0.0001905706964811385,
      "loss": 2.2743,
      "step": 142585
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.1824634075164795,
      "learning_rate": 0.0001905668877693944,
      "loss": 2.245,
      "step": 142586
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.0264543294906616,
      "learning_rate": 0.00019056307907799606,
      "loss": 2.0813,
      "step": 142587
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.1034153699874878,
      "learning_rate": 0.00019055927040694418,
      "loss": 2.0564,
      "step": 142588
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.0923504829406738,
      "learning_rate": 0.0001905554617562393,
      "loss": 2.2236,
      "step": 142589
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.0732307434082031,
      "learning_rate": 0.0001905516531258822,
      "loss": 2.185,
      "step": 142590
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.1184909343719482,
      "learning_rate": 0.0001905478445158736,
      "loss": 2.0731,
      "step": 142591
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.145957112312317,
      "learning_rate": 0.00019054403592621427,
      "loss": 2.1621,
      "step": 142592
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.0951849222183228,
      "learning_rate": 0.00019054022735690482,
      "loss": 2.2478,
      "step": 142593
    },
    {
      "epoch": 1.86,
      "grad_norm": 0.9499834179878235,
      "learning_rate": 0.00019053641880794605,
      "loss": 1.9276,
      "step": 142594
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.0808659791946411,
      "learning_rate": 0.00019053261027933874,
      "loss": 1.9994,
      "step": 142595
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.1321102380752563,
      "learning_rate": 0.00019052880177108337,
      "loss": 2.1303,
      "step": 142596
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.124318242073059,
      "learning_rate": 0.00019052499328318072,
      "loss": 2.2243,
      "step": 142597
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.165752649307251,
      "learning_rate": 0.00019052118481563155,
      "loss": 2.04,
      "step": 142598
    },
    {
      "epoch": 1.86,
      "grad_norm": 0.9623938202857971,
      "learning_rate": 0.00019051737636843657,
      "loss": 2.1073,
      "step": 142599
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.1264407634735107,
      "learning_rate": 0.00019051356794159649,
      "loss": 2.3138,
      "step": 142600
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.1800271272659302,
      "learning_rate": 0.0001905097595351121,
      "loss": 2.2553,
      "step": 142601
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.0474534034729004,
      "learning_rate": 0.00019050595114898392,
      "loss": 2.0763,
      "step": 142602
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.0625098943710327,
      "learning_rate": 0.00019050214278321272,
      "loss": 2.2742,
      "step": 142603
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.1520936489105225,
      "learning_rate": 0.0001904983344377993,
      "loss": 1.9231,
      "step": 142604
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.0730034112930298,
      "learning_rate": 0.00019049452611274425,
      "loss": 2.3113,
      "step": 142605
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.0507808923721313,
      "learning_rate": 0.00019049071780804833,
      "loss": 2.1767,
      "step": 142606
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.2557780742645264,
      "learning_rate": 0.00019048690952371236,
      "loss": 1.998,
      "step": 142607
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.2038824558258057,
      "learning_rate": 0.00019048310125973687,
      "loss": 2.11,
      "step": 142608
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.0996733903884888,
      "learning_rate": 0.0001904792930161226,
      "loss": 2.0766,
      "step": 142609
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.1258023977279663,
      "learning_rate": 0.00019047548479287033,
      "loss": 2.0829,
      "step": 142610
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.0655925273895264,
      "learning_rate": 0.00019047167658998074,
      "loss": 2.038,
      "step": 142611
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.0605182647705078,
      "learning_rate": 0.00019046786840745453,
      "loss": 2.1188,
      "step": 142612
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.2066625356674194,
      "learning_rate": 0.0001904640602452925,
      "loss": 2.0048,
      "step": 142613
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.2076430320739746,
      "learning_rate": 0.00019046025210349513,
      "loss": 2.0494,
      "step": 142614
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.1497788429260254,
      "learning_rate": 0.00019045644398206333,
      "loss": 2.0247,
      "step": 142615
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.0938771963119507,
      "learning_rate": 0.00019045263588099773,
      "loss": 2.0744,
      "step": 142616
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.0606529712677002,
      "learning_rate": 0.00019044882780029903,
      "loss": 2.0936,
      "step": 142617
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.6147420406341553,
      "learning_rate": 0.00019044501973996796,
      "loss": 2.2091,
      "step": 142618
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.244789481163025,
      "learning_rate": 0.00019044121170000536,
      "loss": 2.2096,
      "step": 142619
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.016837239265442,
      "learning_rate": 0.00019043740368041167,
      "loss": 2.0184,
      "step": 142620
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.385791540145874,
      "learning_rate": 0.00019043359568118775,
      "loss": 2.1459,
      "step": 142621
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.0756049156188965,
      "learning_rate": 0.00019042978770233428,
      "loss": 1.8824,
      "step": 142622
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.0807456970214844,
      "learning_rate": 0.000190425979743852,
      "loss": 2.1457,
      "step": 142623
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.1172196865081787,
      "learning_rate": 0.0001904221718057416,
      "loss": 2.1126,
      "step": 142624
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.0787345170974731,
      "learning_rate": 0.0001904183638880039,
      "loss": 2.0359,
      "step": 142625
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.1361736059188843,
      "learning_rate": 0.00019041455599063938,
      "loss": 1.8398,
      "step": 142626
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.2154746055603027,
      "learning_rate": 0.00019041074811364887,
      "loss": 2.0457,
      "step": 142627
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.1510568857192993,
      "learning_rate": 0.00019040694025703305,
      "loss": 2.2203,
      "step": 142628
    },
    {
      "epoch": 1.86,
      "grad_norm": 0.9992750883102417,
      "learning_rate": 0.00019040313242079266,
      "loss": 1.949,
      "step": 142629
    },
    {
      "epoch": 1.86,
      "grad_norm": 0.9344233870506287,
      "learning_rate": 0.0001903993246049284,
      "loss": 1.8817,
      "step": 142630
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.2388975620269775,
      "learning_rate": 0.00019039551680944108,
      "loss": 2.1186,
      "step": 142631
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.2860134840011597,
      "learning_rate": 0.00019039170903433118,
      "loss": 2.1814,
      "step": 142632
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.2051321268081665,
      "learning_rate": 0.00019038790127959954,
      "loss": 2.1338,
      "step": 142633
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.1603072881698608,
      "learning_rate": 0.00019038409354524684,
      "loss": 2.07,
      "step": 142634
    },
    {
      "epoch": 1.86,
      "grad_norm": 0.9983216524124146,
      "learning_rate": 0.00019038028583127383,
      "loss": 2.012,
      "step": 142635
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.118168592453003,
      "learning_rate": 0.00019037647813768116,
      "loss": 2.1675,
      "step": 142636
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.0161470174789429,
      "learning_rate": 0.00019037267046446976,
      "loss": 2.013,
      "step": 142637
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.2134448289871216,
      "learning_rate": 0.00019036886281163998,
      "loss": 2.0571,
      "step": 142638
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.2130459547042847,
      "learning_rate": 0.00019036505517919271,
      "loss": 2.429,
      "step": 142639
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.208308219909668,
      "learning_rate": 0.0001903612475671286,
      "loss": 2.127,
      "step": 142640
    },
    {
      "epoch": 1.86,
      "grad_norm": 0.9996507167816162,
      "learning_rate": 0.00019035743997544846,
      "loss": 2.1337,
      "step": 142641
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.145322561264038,
      "learning_rate": 0.00019035363240415293,
      "loss": 2.1894,
      "step": 142642
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.071766972541809,
      "learning_rate": 0.00019034982485324284,
      "loss": 1.9996,
      "step": 142643
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.2511004209518433,
      "learning_rate": 0.0001903460173227187,
      "loss": 2.1694,
      "step": 142644
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.0411940813064575,
      "learning_rate": 0.00019034220981258126,
      "loss": 2.1453,
      "step": 142645
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.0543859004974365,
      "learning_rate": 0.00019033840232283132,
      "loss": 1.9572,
      "step": 142646
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.6318352222442627,
      "learning_rate": 0.00019033459485346953,
      "loss": 2.1505,
      "step": 142647
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.0864081382751465,
      "learning_rate": 0.00019033078740449653,
      "loss": 2.2277,
      "step": 142648
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.1290333271026611,
      "learning_rate": 0.00019032697997591326,
      "loss": 2.2515,
      "step": 142649
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.158118724822998,
      "learning_rate": 0.00019032317256772023,
      "loss": 1.9421,
      "step": 142650
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.1696484088897705,
      "learning_rate": 0.00019031936517991817,
      "loss": 2.1853,
      "step": 142651
    },
    {
      "epoch": 1.86,
      "grad_norm": 0.9879616498947144,
      "learning_rate": 0.00019031555781250776,
      "loss": 2.231,
      "step": 142652
    },
    {
      "epoch": 1.86,
      "grad_norm": 0.9644438624382019,
      "learning_rate": 0.0001903117504654898,
      "loss": 2.2169,
      "step": 142653
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.149502158164978,
      "learning_rate": 0.00019030794313886493,
      "loss": 2.0168,
      "step": 142654
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.0641010999679565,
      "learning_rate": 0.00019030413583263399,
      "loss": 2.0435,
      "step": 142655
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.146638035774231,
      "learning_rate": 0.00019030032854679745,
      "loss": 2.177,
      "step": 142656
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.0163242816925049,
      "learning_rate": 0.0001902965212813562,
      "loss": 1.9407,
      "step": 142657
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.2472859621047974,
      "learning_rate": 0.00019029271403631095,
      "loss": 2.0546,
      "step": 142658
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.1212791204452515,
      "learning_rate": 0.0001902889068116623,
      "loss": 1.989,
      "step": 142659
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.1140118837356567,
      "learning_rate": 0.00019028509960741097,
      "loss": 2.3006,
      "step": 142660
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.1598644256591797,
      "learning_rate": 0.00019028129242355774,
      "loss": 2.1508,
      "step": 142661
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.1837258338928223,
      "learning_rate": 0.00019027748526010335,
      "loss": 2.1148,
      "step": 142662
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.1415220499038696,
      "learning_rate": 0.0001902736781170484,
      "loss": 2.1257,
      "step": 142663
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.0551491975784302,
      "learning_rate": 0.00019026987099439365,
      "loss": 2.2107,
      "step": 142664
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.0891609191894531,
      "learning_rate": 0.00019026606389213984,
      "loss": 2.0272,
      "step": 142665
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.0996241569519043,
      "learning_rate": 0.0001902622568102876,
      "loss": 2.1016,
      "step": 142666
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.226610779762268,
      "learning_rate": 0.00019025844974883763,
      "loss": 2.117,
      "step": 142667
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.1192206144332886,
      "learning_rate": 0.0001902546427077908,
      "loss": 1.889,
      "step": 142668
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.1391459703445435,
      "learning_rate": 0.00019025083568714762,
      "loss": 2.3101,
      "step": 142669
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.1140950918197632,
      "learning_rate": 0.0001902470286869089,
      "loss": 2.133,
      "step": 142670
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.217533826828003,
      "learning_rate": 0.0001902432217070753,
      "loss": 2.1322,
      "step": 142671
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.2489173412322998,
      "learning_rate": 0.00019023941474764764,
      "loss": 2.1771,
      "step": 142672
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.1606404781341553,
      "learning_rate": 0.00019023560780862648,
      "loss": 1.9711,
      "step": 142673
    },
    {
      "epoch": 1.86,
      "grad_norm": 0.9960763454437256,
      "learning_rate": 0.00019023180089001264,
      "loss": 1.9548,
      "step": 142674
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.036067247390747,
      "learning_rate": 0.00019022799399180673,
      "loss": 2.195,
      "step": 142675
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.2947337627410889,
      "learning_rate": 0.0001902241871140095,
      "loss": 2.1726,
      "step": 142676
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.1899592876434326,
      "learning_rate": 0.0001902203802566217,
      "loss": 1.9926,
      "step": 142677
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.0805498361587524,
      "learning_rate": 0.00019021657341964396,
      "loss": 2.043,
      "step": 142678
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.0647352933883667,
      "learning_rate": 0.00019021276660307712,
      "loss": 2.1817,
      "step": 142679
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.0862692594528198,
      "learning_rate": 0.00019020895980692175,
      "loss": 2.099,
      "step": 142680
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.1831964254379272,
      "learning_rate": 0.0001902051530311786,
      "loss": 1.9519,
      "step": 142681
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.1161646842956543,
      "learning_rate": 0.00019020134627584836,
      "loss": 2.2242,
      "step": 142682
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.0221272706985474,
      "learning_rate": 0.00019019753954093176,
      "loss": 2.4477,
      "step": 142683
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.0857452154159546,
      "learning_rate": 0.00019019373282642952,
      "loss": 1.9565,
      "step": 142684
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.1705131530761719,
      "learning_rate": 0.00019018992613234237,
      "loss": 2.0447,
      "step": 142685
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.1351171731948853,
      "learning_rate": 0.00019018611945867105,
      "loss": 2.1287,
      "step": 142686
    },
    {
      "epoch": 1.86,
      "grad_norm": 0.9928929805755615,
      "learning_rate": 0.0001901823128054161,
      "loss": 2.1765,
      "step": 142687
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.0310471057891846,
      "learning_rate": 0.00019017850617257832,
      "loss": 1.9798,
      "step": 142688
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.152643084526062,
      "learning_rate": 0.00019017469956015846,
      "loss": 2.0401,
      "step": 142689
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.1804122924804688,
      "learning_rate": 0.00019017089296815719,
      "loss": 2.1893,
      "step": 142690
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.1052520275115967,
      "learning_rate": 0.00019016708639657523,
      "loss": 2.1719,
      "step": 142691
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.1312626600265503,
      "learning_rate": 0.0001901632798454134,
      "loss": 1.9409,
      "step": 142692
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.068062663078308,
      "learning_rate": 0.00019015947331467214,
      "loss": 2.1241,
      "step": 142693
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.1704156398773193,
      "learning_rate": 0.00019015566680435233,
      "loss": 1.9692,
      "step": 142694
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.244317650794983,
      "learning_rate": 0.00019015186031445463,
      "loss": 2.1413,
      "step": 142695
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.1072584390640259,
      "learning_rate": 0.0001901480538449798,
      "loss": 1.9712,
      "step": 142696
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.2099930047988892,
      "learning_rate": 0.00019014424739592853,
      "loss": 2.0227,
      "step": 142697
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.222221851348877,
      "learning_rate": 0.00019014044096730162,
      "loss": 2.0336,
      "step": 142698
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.0739390850067139,
      "learning_rate": 0.00019013663455909953,
      "loss": 2.219,
      "step": 142699
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.0701825618743896,
      "learning_rate": 0.00019013282817132316,
      "loss": 2.1408,
      "step": 142700
    },
    {
      "epoch": 1.86,
      "grad_norm": 0.9546206593513489,
      "learning_rate": 0.00019012902180397316,
      "loss": 2.1232,
      "step": 142701
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.0315629243850708,
      "learning_rate": 0.00019012521545705023,
      "loss": 1.8995,
      "step": 142702
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.0206375122070312,
      "learning_rate": 0.00019012140913055514,
      "loss": 2.1431,
      "step": 142703
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.2102010250091553,
      "learning_rate": 0.0001901176028244886,
      "loss": 2.0292,
      "step": 142704
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.1907111406326294,
      "learning_rate": 0.00019011379653885118,
      "loss": 2.1131,
      "step": 142705
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.0982352495193481,
      "learning_rate": 0.0001901099902736437,
      "loss": 2.3158,
      "step": 142706
    },
    {
      "epoch": 1.86,
      "grad_norm": 0.9662618041038513,
      "learning_rate": 0.00019010618402886688,
      "loss": 1.9251,
      "step": 142707
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.0806729793548584,
      "learning_rate": 0.0001901023778045213,
      "loss": 2.0026,
      "step": 142708
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.0367258787155151,
      "learning_rate": 0.00019009857160060786,
      "loss": 2.1009,
      "step": 142709
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.1263922452926636,
      "learning_rate": 0.00019009476541712725,
      "loss": 2.0377,
      "step": 142710
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.14603590965271,
      "learning_rate": 0.00019009095925407997,
      "loss": 1.9959,
      "step": 142711
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.221579909324646,
      "learning_rate": 0.00019008715311146686,
      "loss": 2.1616,
      "step": 142712
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.419218897819519,
      "learning_rate": 0.0001900833469892886,
      "loss": 1.913,
      "step": 142713
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.0059014558792114,
      "learning_rate": 0.00019007954088754597,
      "loss": 2.1016,
      "step": 142714
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.327476978302002,
      "learning_rate": 0.0001900757348062396,
      "loss": 2.1763,
      "step": 142715
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.2078027725219727,
      "learning_rate": 0.00019007192874537034,
      "loss": 2.1047,
      "step": 142716
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.0645946264266968,
      "learning_rate": 0.0001900681227049387,
      "loss": 2.0241,
      "step": 142717
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.0579155683517456,
      "learning_rate": 0.00019006431668494543,
      "loss": 2.2455,
      "step": 142718
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.1098493337631226,
      "learning_rate": 0.00019006051068539128,
      "loss": 2.0015,
      "step": 142719
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.1061668395996094,
      "learning_rate": 0.00019005670470627697,
      "loss": 1.9413,
      "step": 142720
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.076378583908081,
      "learning_rate": 0.00019005289874760318,
      "loss": 2.0019,
      "step": 142721
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.0773950815200806,
      "learning_rate": 0.00019004909280937078,
      "loss": 2.2519,
      "step": 142722
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.2193963527679443,
      "learning_rate": 0.0001900452868915802,
      "loss": 2.2127,
      "step": 142723
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.1043890714645386,
      "learning_rate": 0.00019004148099423224,
      "loss": 2.031,
      "step": 142724
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.1728562116622925,
      "learning_rate": 0.00019003767511732768,
      "loss": 2.0889,
      "step": 142725
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.1601064205169678,
      "learning_rate": 0.0001900338692608672,
      "loss": 2.0545,
      "step": 142726
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.2379214763641357,
      "learning_rate": 0.00019003006342485147,
      "loss": 2.2391,
      "step": 142727
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.033709168434143,
      "learning_rate": 0.00019002625760928125,
      "loss": 1.9403,
      "step": 142728
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.1231932640075684,
      "learning_rate": 0.00019002245181415733,
      "loss": 2.2437,
      "step": 142729
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.0961939096450806,
      "learning_rate": 0.0001900186460394802,
      "loss": 2.0556,
      "step": 142730
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.1320841312408447,
      "learning_rate": 0.0001900148402852507,
      "loss": 2.0444,
      "step": 142731
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.0720417499542236,
      "learning_rate": 0.00019001103455146945,
      "loss": 2.0313,
      "step": 142732
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.178560495376587,
      "learning_rate": 0.00019000722883813734,
      "loss": 2.0045,
      "step": 142733
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.0631084442138672,
      "learning_rate": 0.00019000342314525487,
      "loss": 2.0469,
      "step": 142734
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.011756420135498,
      "learning_rate": 0.00018999961747282298,
      "loss": 2.1112,
      "step": 142735
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.1425721645355225,
      "learning_rate": 0.0001899958118208421,
      "loss": 1.855,
      "step": 142736
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.0843833684921265,
      "learning_rate": 0.0001899920061893131,
      "loss": 2.05,
      "step": 142737
    },
    {
      "epoch": 1.86,
      "grad_norm": 0.9922181367874146,
      "learning_rate": 0.00018998820057823665,
      "loss": 2.081,
      "step": 142738
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.107251763343811,
      "learning_rate": 0.00018998439498761347,
      "loss": 2.0,
      "step": 142739
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.1698850393295288,
      "learning_rate": 0.0001899805894174443,
      "loss": 2.1262,
      "step": 142740
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.0596427917480469,
      "learning_rate": 0.00018997678386772988,
      "loss": 2.0762,
      "step": 142741
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.1874810457229614,
      "learning_rate": 0.00018997297833847079,
      "loss": 2.2043,
      "step": 142742
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.0926551818847656,
      "learning_rate": 0.00018996917282966777,
      "loss": 1.9775,
      "step": 142743
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.1144299507141113,
      "learning_rate": 0.00018996536734132156,
      "loss": 2.125,
      "step": 142744
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.0845357179641724,
      "learning_rate": 0.0001899615618734329,
      "loss": 2.0036,
      "step": 142745
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.2156871557235718,
      "learning_rate": 0.00018995775642600239,
      "loss": 2.3087,
      "step": 142746
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.032402753829956,
      "learning_rate": 0.00018995395099903094,
      "loss": 2.286,
      "step": 142747
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.1872308254241943,
      "learning_rate": 0.00018995014559251903,
      "loss": 2.0196,
      "step": 142748
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.0487383604049683,
      "learning_rate": 0.0001899463402064675,
      "loss": 2.1223,
      "step": 142749
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.1130375862121582,
      "learning_rate": 0.00018994253484087704,
      "loss": 1.9515,
      "step": 142750
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.1228729486465454,
      "learning_rate": 0.0001899387294957483,
      "loss": 2.3783,
      "step": 142751
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.1232024431228638,
      "learning_rate": 0.000189934924171082,
      "loss": 2.0199,
      "step": 142752
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.1560838222503662,
      "learning_rate": 0.000189931118866879,
      "loss": 2.3144,
      "step": 142753
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.121321439743042,
      "learning_rate": 0.00018992731358313976,
      "loss": 2.0378,
      "step": 142754
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.2219007015228271,
      "learning_rate": 0.00018992350831986514,
      "loss": 2.3084,
      "step": 142755
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.0685570240020752,
      "learning_rate": 0.0001899197030770558,
      "loss": 2.0912,
      "step": 142756
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.199388861656189,
      "learning_rate": 0.00018991589785471253,
      "loss": 2.234,
      "step": 142757
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.2186564207077026,
      "learning_rate": 0.00018991209265283593,
      "loss": 2.0834,
      "step": 142758
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.06721830368042,
      "learning_rate": 0.00018990828747142676,
      "loss": 1.8351,
      "step": 142759
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.0435115098953247,
      "learning_rate": 0.00018990448231048573,
      "loss": 2.0722,
      "step": 142760
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.0576668977737427,
      "learning_rate": 0.0001899006771700135,
      "loss": 2.1449,
      "step": 142761
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.1902508735656738,
      "learning_rate": 0.00018989687205001082,
      "loss": 2.181,
      "step": 142762
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.0546234846115112,
      "learning_rate": 0.00018989306695047837,
      "loss": 2.2067,
      "step": 142763
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.0671615600585938,
      "learning_rate": 0.00018988926187141694,
      "loss": 2.0619,
      "step": 142764
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.0271910429000854,
      "learning_rate": 0.0001898854568128272,
      "loss": 2.1012,
      "step": 142765
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.1495764255523682,
      "learning_rate": 0.00018988165177470973,
      "loss": 2.0963,
      "step": 142766
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.020368218421936,
      "learning_rate": 0.00018987784675706537,
      "loss": 2.0148,
      "step": 142767
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.0311537981033325,
      "learning_rate": 0.00018987404175989484,
      "loss": 2.0478,
      "step": 142768
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.1508193016052246,
      "learning_rate": 0.00018987023678319873,
      "loss": 2.202,
      "step": 142769
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.0567669868469238,
      "learning_rate": 0.00018986643182697786,
      "loss": 1.8397,
      "step": 142770
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.1870393753051758,
      "learning_rate": 0.000189862626891233,
      "loss": 2.2041,
      "step": 142771
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.139146089553833,
      "learning_rate": 0.00018985882197596467,
      "loss": 2.2139,
      "step": 142772
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.3136706352233887,
      "learning_rate": 0.00018985501708117364,
      "loss": 2.2287,
      "step": 142773
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.206496238708496,
      "learning_rate": 0.0001898512122068606,
      "loss": 1.9433,
      "step": 142774
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.1659088134765625,
      "learning_rate": 0.00018984740735302635,
      "loss": 2.2897,
      "step": 142775
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.0068449974060059,
      "learning_rate": 0.00018984360251967153,
      "loss": 2.2735,
      "step": 142776
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.3506780862808228,
      "learning_rate": 0.000189839797706797,
      "loss": 2.0426,
      "step": 142777
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.541181206703186,
      "learning_rate": 0.0001898359929144032,
      "loss": 1.8632,
      "step": 142778
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.0923075675964355,
      "learning_rate": 0.000189832188142491,
      "loss": 2.0124,
      "step": 142779
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.135616421699524,
      "learning_rate": 0.00018982838339106102,
      "loss": 2.252,
      "step": 142780
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.3966832160949707,
      "learning_rate": 0.00018982457866011405,
      "loss": 2.0717,
      "step": 142781
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.155013918876648,
      "learning_rate": 0.00018982077394965075,
      "loss": 2.0571,
      "step": 142782
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.0537660121917725,
      "learning_rate": 0.00018981696925967202,
      "loss": 2.0729,
      "step": 142783
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.0181002616882324,
      "learning_rate": 0.00018981316459017823,
      "loss": 1.913,
      "step": 142784
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.1634660959243774,
      "learning_rate": 0.00018980935994117025,
      "loss": 2.0451,
      "step": 142785
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.0250605344772339,
      "learning_rate": 0.00018980555531264884,
      "loss": 2.0418,
      "step": 142786
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.141363501548767,
      "learning_rate": 0.0001898017507046146,
      "loss": 2.2049,
      "step": 142787
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.0032938718795776,
      "learning_rate": 0.00018979794611706833,
      "loss": 2.0522,
      "step": 142788
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.440178394317627,
      "learning_rate": 0.0001897941415500108,
      "loss": 2.0864,
      "step": 142789
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.0873662233352661,
      "learning_rate": 0.00018979033700344247,
      "loss": 2.1969,
      "step": 142790
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.1298208236694336,
      "learning_rate": 0.00018978653247736426,
      "loss": 1.9698,
      "step": 142791
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.0690937042236328,
      "learning_rate": 0.0001897827279717768,
      "loss": 2.2456,
      "step": 142792
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.179305911064148,
      "learning_rate": 0.00018977892348668077,
      "loss": 2.1146,
      "step": 142793
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.12406325340271,
      "learning_rate": 0.00018977511902207696,
      "loss": 1.8762,
      "step": 142794
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.1410300731658936,
      "learning_rate": 0.000189771314577966,
      "loss": 2.0309,
      "step": 142795
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.2711646556854248,
      "learning_rate": 0.0001897675101543488,
      "loss": 1.8924,
      "step": 142796
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.2933735847473145,
      "learning_rate": 0.0001897637057512258,
      "loss": 1.9927,
      "step": 142797
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.0229307413101196,
      "learning_rate": 0.00018975990136859775,
      "loss": 2.1004,
      "step": 142798
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.1288961172103882,
      "learning_rate": 0.00018975609700646543,
      "loss": 2.0625,
      "step": 142799
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.1190265417099,
      "learning_rate": 0.00018975229266482953,
      "loss": 2.1892,
      "step": 142800
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.2031465768814087,
      "learning_rate": 0.00018974848834369074,
      "loss": 2.2053,
      "step": 142801
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.1032158136367798,
      "learning_rate": 0.00018974468404304998,
      "loss": 2.1253,
      "step": 142802
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.0855984687805176,
      "learning_rate": 0.0001897408797629076,
      "loss": 2.3257,
      "step": 142803
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.1697523593902588,
      "learning_rate": 0.00018973707550326444,
      "loss": 2.2514,
      "step": 142804
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.01787269115448,
      "learning_rate": 0.00018973327126412125,
      "loss": 2.1158,
      "step": 142805
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.0123580694198608,
      "learning_rate": 0.00018972946704547876,
      "loss": 2.0355,
      "step": 142806
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.1503257751464844,
      "learning_rate": 0.0001897256628473376,
      "loss": 2.2478,
      "step": 142807
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.0786257982254028,
      "learning_rate": 0.0001897218586696987,
      "loss": 2.0204,
      "step": 142808
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.104971170425415,
      "learning_rate": 0.00018971805451256245,
      "loss": 2.2841,
      "step": 142809
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.205737590789795,
      "learning_rate": 0.00018971425037592969,
      "loss": 2.2086,
      "step": 142810
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.0594110488891602,
      "learning_rate": 0.0001897104462598011,
      "loss": 2.3616,
      "step": 142811
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.0517997741699219,
      "learning_rate": 0.00018970664216417745,
      "loss": 2.2611,
      "step": 142812
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.1364275217056274,
      "learning_rate": 0.0001897028380890594,
      "loss": 2.122,
      "step": 142813
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.162210464477539,
      "learning_rate": 0.0001896990340344478,
      "loss": 2.1232,
      "step": 142814
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.0389376878738403,
      "learning_rate": 0.00018969523000034316,
      "loss": 2.0259,
      "step": 142815
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.025119423866272,
      "learning_rate": 0.0001896914259867462,
      "loss": 2.1527,
      "step": 142816
    },
    {
      "epoch": 1.86,
      "grad_norm": 0.9742945432662964,
      "learning_rate": 0.0001896876219936577,
      "loss": 1.7848,
      "step": 142817
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.0636008977890015,
      "learning_rate": 0.00018968381802107834,
      "loss": 2.1144,
      "step": 142818
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.1394104957580566,
      "learning_rate": 0.00018968001406900886,
      "loss": 2.2212,
      "step": 142819
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.1214009523391724,
      "learning_rate": 0.0001896762101374501,
      "loss": 2.0302,
      "step": 142820
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.0595191717147827,
      "learning_rate": 0.00018967240622640242,
      "loss": 2.0058,
      "step": 142821
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.0877712965011597,
      "learning_rate": 0.00018966860233586677,
      "loss": 2.1359,
      "step": 142822
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.06938636302948,
      "learning_rate": 0.00018966479846584382,
      "loss": 1.9951,
      "step": 142823
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.127371907234192,
      "learning_rate": 0.00018966099461633425,
      "loss": 2.0951,
      "step": 142824
    },
    {
      "epoch": 1.86,
      "grad_norm": 0.987994372844696,
      "learning_rate": 0.00018965719078733876,
      "loss": 1.9622,
      "step": 142825
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.1645874977111816,
      "learning_rate": 0.00018965338697885816,
      "loss": 2.3188,
      "step": 142826
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.0710817575454712,
      "learning_rate": 0.00018964958319089308,
      "loss": 2.2729,
      "step": 142827
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.0363872051239014,
      "learning_rate": 0.00018964577942344417,
      "loss": 1.8811,
      "step": 142828
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.063308596611023,
      "learning_rate": 0.0001896419756765122,
      "loss": 2.1863,
      "step": 142829
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.1602586507797241,
      "learning_rate": 0.00018963817195009783,
      "loss": 2.1679,
      "step": 142830
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.0745469331741333,
      "learning_rate": 0.00018963436824420184,
      "loss": 1.9625,
      "step": 142831
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.1642625331878662,
      "learning_rate": 0.00018963056455882497,
      "loss": 2.0313,
      "step": 142832
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.1417884826660156,
      "learning_rate": 0.00018962676089396782,
      "loss": 2.2324,
      "step": 142833
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.0732312202453613,
      "learning_rate": 0.00018962295724963115,
      "loss": 2.1579,
      "step": 142834
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.2094436883926392,
      "learning_rate": 0.0001896191536258156,
      "loss": 2.1331,
      "step": 142835
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.0691397190093994,
      "learning_rate": 0.00018961535002252195,
      "loss": 2.1838,
      "step": 142836
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.02873694896698,
      "learning_rate": 0.0001896115464397509,
      "loss": 2.1209,
      "step": 142837
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.0898654460906982,
      "learning_rate": 0.00018960774287750317,
      "loss": 2.2345,
      "step": 142838
    },
    {
      "epoch": 1.86,
      "grad_norm": 0.9747321009635925,
      "learning_rate": 0.00018960393933577946,
      "loss": 2.0396,
      "step": 142839
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.0989443063735962,
      "learning_rate": 0.0001896001358145804,
      "loss": 2.1038,
      "step": 142840
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.0070127248764038,
      "learning_rate": 0.00018959633231390677,
      "loss": 2.011,
      "step": 142841
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.1222000122070312,
      "learning_rate": 0.00018959252883375932,
      "loss": 1.9337,
      "step": 142842
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.0754563808441162,
      "learning_rate": 0.00018958872537413866,
      "loss": 2.1265,
      "step": 142843
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.1956826448440552,
      "learning_rate": 0.00018958492193504558,
      "loss": 1.9209,
      "step": 142844
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.09519624710083,
      "learning_rate": 0.00018958111851648073,
      "loss": 2.17,
      "step": 142845
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.2102311849594116,
      "learning_rate": 0.00018957731511844478,
      "loss": 1.917,
      "step": 142846
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.1088290214538574,
      "learning_rate": 0.0001895735117409385,
      "loss": 2.0569,
      "step": 142847
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.0878947973251343,
      "learning_rate": 0.0001895697083839626,
      "loss": 1.8936,
      "step": 142848
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.2179701328277588,
      "learning_rate": 0.00018956590504751785,
      "loss": 2.0158,
      "step": 142849
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.2741798162460327,
      "learning_rate": 0.00018956210173160488,
      "loss": 2.0447,
      "step": 142850
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.2291803359985352,
      "learning_rate": 0.00018955829843622434,
      "loss": 2.0186,
      "step": 142851
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.0594135522842407,
      "learning_rate": 0.00018955449516137694,
      "loss": 1.8223,
      "step": 142852
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.2391422986984253,
      "learning_rate": 0.00018955069190706354,
      "loss": 2.4628,
      "step": 142853
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.312264084815979,
      "learning_rate": 0.0001895468886732847,
      "loss": 2.2857,
      "step": 142854
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.0447477102279663,
      "learning_rate": 0.00018954308546004117,
      "loss": 2.1993,
      "step": 142855
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.1882734298706055,
      "learning_rate": 0.0001895392822673338,
      "loss": 1.9608,
      "step": 142856
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.123047947883606,
      "learning_rate": 0.00018953547909516304,
      "loss": 1.9179,
      "step": 142857
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.2734190225601196,
      "learning_rate": 0.00018953167594352972,
      "loss": 1.7864,
      "step": 142858
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.1471818685531616,
      "learning_rate": 0.00018952787281243454,
      "loss": 2.1608,
      "step": 142859
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.110940933227539,
      "learning_rate": 0.0001895240697018782,
      "loss": 1.9831,
      "step": 142860
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.0993605852127075,
      "learning_rate": 0.00018952026661186146,
      "loss": 2.0892,
      "step": 142861
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.079932451248169,
      "learning_rate": 0.00018951646354238496,
      "loss": 2.3615,
      "step": 142862
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.132286548614502,
      "learning_rate": 0.00018951266049344956,
      "loss": 2.0277,
      "step": 142863
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.0036394596099854,
      "learning_rate": 0.00018950885746505573,
      "loss": 2.171,
      "step": 142864
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.1118017435073853,
      "learning_rate": 0.00018950505445720429,
      "loss": 2.0834,
      "step": 142865
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.0418239831924438,
      "learning_rate": 0.00018950125146989595,
      "loss": 2.1323,
      "step": 142866
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.152405023574829,
      "learning_rate": 0.00018949744850313137,
      "loss": 2.2297,
      "step": 142867
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.1233408451080322,
      "learning_rate": 0.0001894936455569114,
      "loss": 2.0609,
      "step": 142868
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.05039644241333,
      "learning_rate": 0.00018948984263123667,
      "loss": 2.0457,
      "step": 142869
    },
    {
      "epoch": 1.86,
      "grad_norm": 0.9962008595466614,
      "learning_rate": 0.00018948603972610773,
      "loss": 2.1933,
      "step": 142870
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.1664342880249023,
      "learning_rate": 0.0001894822368415255,
      "loss": 2.1204,
      "step": 142871
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.0606693029403687,
      "learning_rate": 0.00018947843397749056,
      "loss": 2.0261,
      "step": 142872
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.081468105316162,
      "learning_rate": 0.00018947463113400372,
      "loss": 2.1696,
      "step": 142873
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.2176682949066162,
      "learning_rate": 0.00018947082831106556,
      "loss": 1.9399,
      "step": 142874
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.030519723892212,
      "learning_rate": 0.000189467025508677,
      "loss": 2.0928,
      "step": 142875
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.3509409427642822,
      "learning_rate": 0.00018946322272683853,
      "loss": 1.9936,
      "step": 142876
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.0962848663330078,
      "learning_rate": 0.0001894594199655509,
      "loss": 2.1231,
      "step": 142877
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.0886387825012207,
      "learning_rate": 0.00018945561722481487,
      "loss": 2.046,
      "step": 142878
    },
    {
      "epoch": 1.86,
      "grad_norm": 0.9648102521896362,
      "learning_rate": 0.00018945181450463107,
      "loss": 1.9422,
      "step": 142879
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.0270813703536987,
      "learning_rate": 0.00018944801180500033,
      "loss": 1.9215,
      "step": 142880
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.0658589601516724,
      "learning_rate": 0.00018944420912592338,
      "loss": 2.0634,
      "step": 142881
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.1466251611709595,
      "learning_rate": 0.0001894404064674007,
      "loss": 2.1113,
      "step": 142882
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.237062692642212,
      "learning_rate": 0.0001894366038294332,
      "loss": 2.0895,
      "step": 142883
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.098768949508667,
      "learning_rate": 0.00018943280121202144,
      "loss": 1.9773,
      "step": 142884
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.083208680152893,
      "learning_rate": 0.00018942899861516626,
      "loss": 2.2404,
      "step": 142885
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.1816495656967163,
      "learning_rate": 0.00018942519603886834,
      "loss": 1.8321,
      "step": 142886
    },
    {
      "epoch": 1.86,
      "grad_norm": 0.9900537729263306,
      "learning_rate": 0.00018942139348312846,
      "loss": 2.0123,
      "step": 142887
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.0984370708465576,
      "learning_rate": 0.00018941759094794706,
      "loss": 2.0927,
      "step": 142888
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.1711021661758423,
      "learning_rate": 0.0001894137884333251,
      "loss": 2.2091,
      "step": 142889
    },
    {
      "epoch": 1.86,
      "grad_norm": 0.9820988178253174,
      "learning_rate": 0.00018940998593926312,
      "loss": 1.9638,
      "step": 142890
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.0201711654663086,
      "learning_rate": 0.000189406183465762,
      "loss": 2.0774,
      "step": 142891
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.0250407457351685,
      "learning_rate": 0.00018940238101282229,
      "loss": 2.1626,
      "step": 142892
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.123161792755127,
      "learning_rate": 0.0001893985785804449,
      "loss": 2.0319,
      "step": 142893
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.0693027973175049,
      "learning_rate": 0.0001893947761686303,
      "loss": 2.232,
      "step": 142894
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.2866721153259277,
      "learning_rate": 0.00018939097377737926,
      "loss": 1.996,
      "step": 142895
    },
    {
      "epoch": 1.86,
      "grad_norm": 0.9983169436454773,
      "learning_rate": 0.00018938717140669259,
      "loss": 2.0614,
      "step": 142896
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.1899745464324951,
      "learning_rate": 0.0001893833690565709,
      "loss": 2.1473,
      "step": 142897
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.2316381931304932,
      "learning_rate": 0.0001893795667270149,
      "loss": 2.1525,
      "step": 142898
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.0055904388427734,
      "learning_rate": 0.0001893757644180255,
      "loss": 2.0989,
      "step": 142899
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.161800742149353,
      "learning_rate": 0.00018937196212960307,
      "loss": 2.1082,
      "step": 142900
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.0536713600158691,
      "learning_rate": 0.0001893681598617485,
      "loss": 2.2092,
      "step": 142901
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.2377721071243286,
      "learning_rate": 0.00018936435761446246,
      "loss": 2.1011,
      "step": 142902
    },
    {
      "epoch": 1.86,
      "grad_norm": 0.9404443502426147,
      "learning_rate": 0.0001893605553877457,
      "loss": 1.9828,
      "step": 142903
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.2117750644683838,
      "learning_rate": 0.0001893567531815989,
      "loss": 2.1352,
      "step": 142904
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.2743068933486938,
      "learning_rate": 0.00018935295099602287,
      "loss": 2.0769,
      "step": 142905
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.1010854244232178,
      "learning_rate": 0.00018934914883101807,
      "loss": 1.9229,
      "step": 142906
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.04461669921875,
      "learning_rate": 0.00018934534668658542,
      "loss": 1.9352,
      "step": 142907
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.0833145380020142,
      "learning_rate": 0.0001893415445627255,
      "loss": 2.2138,
      "step": 142908
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.0555890798568726,
      "learning_rate": 0.00018933774245943912,
      "loss": 2.245,
      "step": 142909
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.3976322412490845,
      "learning_rate": 0.00018933394037672693,
      "loss": 2.2035,
      "step": 142910
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.1323727369308472,
      "learning_rate": 0.00018933013831458969,
      "loss": 1.971,
      "step": 142911
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.1772751808166504,
      "learning_rate": 0.0001893263362730281,
      "loss": 2.0087,
      "step": 142912
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.1157656908035278,
      "learning_rate": 0.00018932253425204274,
      "loss": 2.1356,
      "step": 142913
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.3381237983703613,
      "learning_rate": 0.00018931873225163444,
      "loss": 2.0495,
      "step": 142914
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.1010454893112183,
      "learning_rate": 0.00018931493027180384,
      "loss": 1.9479,
      "step": 142915
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.044537901878357,
      "learning_rate": 0.0001893111283125517,
      "loss": 2.1706,
      "step": 142916
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.1760154962539673,
      "learning_rate": 0.00018930732637387876,
      "loss": 1.8856,
      "step": 142917
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.0052932500839233,
      "learning_rate": 0.00018930352445578567,
      "loss": 1.9947,
      "step": 142918
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.1676777601242065,
      "learning_rate": 0.00018929972255827312,
      "loss": 2.2838,
      "step": 142919
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.0835403203964233,
      "learning_rate": 0.00018929592068134184,
      "loss": 1.9695,
      "step": 142920
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.2548813819885254,
      "learning_rate": 0.00018929211882499252,
      "loss": 1.8759,
      "step": 142921
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.1272433996200562,
      "learning_rate": 0.0001892883169892259,
      "loss": 1.9941,
      "step": 142922
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.0040255784988403,
      "learning_rate": 0.0001892845151740427,
      "loss": 2.0082,
      "step": 142923
    },
    {
      "epoch": 1.86,
      "grad_norm": 0.9835289716720581,
      "learning_rate": 0.00018928071337944357,
      "loss": 1.9547,
      "step": 142924
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.012344241142273,
      "learning_rate": 0.00018927691160542924,
      "loss": 1.775,
      "step": 142925
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.1583783626556396,
      "learning_rate": 0.0001892731098520005,
      "loss": 2.1068,
      "step": 142926
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.1363990306854248,
      "learning_rate": 0.0001892693081191579,
      "loss": 2.0643,
      "step": 142927
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.074114441871643,
      "learning_rate": 0.0001892655064069022,
      "loss": 2.1208,
      "step": 142928
    },
    {
      "epoch": 1.86,
      "grad_norm": 0.9851831793785095,
      "learning_rate": 0.00018926170471523417,
      "loss": 2.0816,
      "step": 142929
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.201628565788269,
      "learning_rate": 0.0001892579030441545,
      "loss": 1.9255,
      "step": 142930
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.1422299146652222,
      "learning_rate": 0.00018925410139366384,
      "loss": 1.9657,
      "step": 142931
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.0872018337249756,
      "learning_rate": 0.00018925029976376293,
      "loss": 2.0474,
      "step": 142932
    },
    {
      "epoch": 1.86,
      "grad_norm": 0.9990683197975159,
      "learning_rate": 0.00018924649815445254,
      "loss": 2.1823,
      "step": 142933
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.2068846225738525,
      "learning_rate": 0.0001892426965657333,
      "loss": 1.9453,
      "step": 142934
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.1121468544006348,
      "learning_rate": 0.00018923889499760586,
      "loss": 2.0112,
      "step": 142935
    },
    {
      "epoch": 1.86,
      "grad_norm": 0.991671621799469,
      "learning_rate": 0.0001892350934500711,
      "loss": 2.1095,
      "step": 142936
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.2210527658462524,
      "learning_rate": 0.00018923129192312957,
      "loss": 1.8871,
      "step": 142937
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.0592073202133179,
      "learning_rate": 0.00018922749041678204,
      "loss": 2.0255,
      "step": 142938
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.0622259378433228,
      "learning_rate": 0.00018922368893102917,
      "loss": 2.1586,
      "step": 142939
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.1477770805358887,
      "learning_rate": 0.00018921988746587172,
      "loss": 2.1737,
      "step": 142940
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.132584810256958,
      "learning_rate": 0.00018921608602131046,
      "loss": 2.077,
      "step": 142941
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.5178515911102295,
      "learning_rate": 0.00018921228459734603,
      "loss": 2.1008,
      "step": 142942
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.093941569328308,
      "learning_rate": 0.00018920848319397909,
      "loss": 2.3306,
      "step": 142943
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.0444531440734863,
      "learning_rate": 0.0001892046818112103,
      "loss": 2.1699,
      "step": 142944
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.104827880859375,
      "learning_rate": 0.00018920088044904052,
      "loss": 1.9019,
      "step": 142945
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.124464511871338,
      "learning_rate": 0.00018919707910747036,
      "loss": 1.9757,
      "step": 142946
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.1187676191329956,
      "learning_rate": 0.00018919327778650055,
      "loss": 2.2554,
      "step": 142947
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.0811039209365845,
      "learning_rate": 0.00018918947648613194,
      "loss": 2.1175,
      "step": 142948
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.0974105596542358,
      "learning_rate": 0.000189185675206365,
      "loss": 2.2926,
      "step": 142949
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.1917164325714111,
      "learning_rate": 0.0001891818739472005,
      "loss": 2.0938,
      "step": 142950
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.0664037466049194,
      "learning_rate": 0.00018917807270863916,
      "loss": 2.1846,
      "step": 142951
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.0204339027404785,
      "learning_rate": 0.00018917427149068176,
      "loss": 2.1046,
      "step": 142952
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.132120132446289,
      "learning_rate": 0.00018917047029332892,
      "loss": 2.1,
      "step": 142953
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.1959038972854614,
      "learning_rate": 0.0001891666691165815,
      "loss": 1.9407,
      "step": 142954
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.1228890419006348,
      "learning_rate": 0.00018916286796044,
      "loss": 2.1637,
      "step": 142955
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.1034303903579712,
      "learning_rate": 0.0001891590668249052,
      "loss": 2.1077,
      "step": 142956
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.1592148542404175,
      "learning_rate": 0.0001891552657099778,
      "loss": 2.1771,
      "step": 142957
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.3019605875015259,
      "learning_rate": 0.00018915146461565856,
      "loss": 2.0449,
      "step": 142958
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.1299364566802979,
      "learning_rate": 0.00018914766354194814,
      "loss": 2.1221,
      "step": 142959
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.0831245183944702,
      "learning_rate": 0.00018914386248884737,
      "loss": 2.0996,
      "step": 142960
    },
    {
      "epoch": 1.86,
      "grad_norm": 0.9738743305206299,
      "learning_rate": 0.00018914006145635674,
      "loss": 2.0397,
      "step": 142961
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.097061276435852,
      "learning_rate": 0.0001891362604444771,
      "loss": 2.183,
      "step": 142962
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.0050660371780396,
      "learning_rate": 0.00018913245945320907,
      "loss": 2.2469,
      "step": 142963
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.1423072814941406,
      "learning_rate": 0.00018912865848255342,
      "loss": 2.0029,
      "step": 142964
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.0266486406326294,
      "learning_rate": 0.00018912485753251084,
      "loss": 1.9501,
      "step": 142965
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.013831377029419,
      "learning_rate": 0.00018912105660308216,
      "loss": 1.8122,
      "step": 142966
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.083922266960144,
      "learning_rate": 0.00018911725569426788,
      "loss": 2.1856,
      "step": 142967
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.0040448904037476,
      "learning_rate": 0.00018911345480606876,
      "loss": 2.0595,
      "step": 142968
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.0616453886032104,
      "learning_rate": 0.0001891096539384856,
      "loss": 2.2341,
      "step": 142969
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.0952738523483276,
      "learning_rate": 0.000189105853091519,
      "loss": 2.1247,
      "step": 142970
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.0747992992401123,
      "learning_rate": 0.00018910205226516972,
      "loss": 2.1808,
      "step": 142971
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.0163567066192627,
      "learning_rate": 0.0001890982514594386,
      "loss": 2.0588,
      "step": 142972
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.0646896362304688,
      "learning_rate": 0.00018909445067432607,
      "loss": 2.0652,
      "step": 142973
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.0162553787231445,
      "learning_rate": 0.00018909064990983295,
      "loss": 1.9601,
      "step": 142974
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.1103193759918213,
      "learning_rate": 0.00018908684916596003,
      "loss": 2.0504,
      "step": 142975
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.1712274551391602,
      "learning_rate": 0.00018908304844270795,
      "loss": 2.2697,
      "step": 142976
    },
    {
      "epoch": 1.86,
      "grad_norm": 0.9909586906433105,
      "learning_rate": 0.00018907924774007738,
      "loss": 2.204,
      "step": 142977
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.1637845039367676,
      "learning_rate": 0.0001890754470580692,
      "loss": 2.1716,
      "step": 142978
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.0069410800933838,
      "learning_rate": 0.0001890716463966839,
      "loss": 1.8621,
      "step": 142979
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.1338211297988892,
      "learning_rate": 0.00018906784575592226,
      "loss": 2.2199,
      "step": 142980
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.0771510601043701,
      "learning_rate": 0.00018906404513578502,
      "loss": 1.8826,
      "step": 142981
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.0650885105133057,
      "learning_rate": 0.00018906024453627283,
      "loss": 2.1738,
      "step": 142982
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.0614516735076904,
      "learning_rate": 0.00018905644395738645,
      "loss": 2.1834,
      "step": 142983
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.040065050125122,
      "learning_rate": 0.0001890526433991267,
      "loss": 2.0076,
      "step": 142984
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.0734189748764038,
      "learning_rate": 0.00018904884286149404,
      "loss": 2.106,
      "step": 142985
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.1331887245178223,
      "learning_rate": 0.0001890450423444893,
      "loss": 2.0149,
      "step": 142986
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.0275284051895142,
      "learning_rate": 0.00018904124184811316,
      "loss": 1.8576,
      "step": 142987
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.1014065742492676,
      "learning_rate": 0.0001890374413723663,
      "loss": 2.1925,
      "step": 142988
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.2659015655517578,
      "learning_rate": 0.00018903364091724957,
      "loss": 2.2489,
      "step": 142989
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.1404731273651123,
      "learning_rate": 0.00018902984048276366,
      "loss": 2.2555,
      "step": 142990
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.160434603691101,
      "learning_rate": 0.0001890260400689091,
      "loss": 2.3819,
      "step": 142991
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.027225136756897,
      "learning_rate": 0.00018902223967568667,
      "loss": 2.0545,
      "step": 142992
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.0674794912338257,
      "learning_rate": 0.0001890184393030971,
      "loss": 2.024,
      "step": 142993
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.03134286403656,
      "learning_rate": 0.00018901463895114112,
      "loss": 2.3115,
      "step": 142994
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.4975625276565552,
      "learning_rate": 0.00018901083861981935,
      "loss": 2.1852,
      "step": 142995
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.1425526142120361,
      "learning_rate": 0.0001890070383091326,
      "loss": 2.2141,
      "step": 142996
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.096930980682373,
      "learning_rate": 0.0001890032380190817,
      "loss": 2.1881,
      "step": 142997
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.1615935564041138,
      "learning_rate": 0.000188999437749667,
      "loss": 2.0997,
      "step": 142998
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.0215685367584229,
      "learning_rate": 0.00018899563750088947,
      "loss": 2.0534,
      "step": 142999
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.0930237770080566,
      "learning_rate": 0.0001889918372727497,
      "loss": 2.051,
      "step": 143000
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.154806137084961,
      "learning_rate": 0.0001889880370652484,
      "loss": 2.0779,
      "step": 143001
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.1367322206497192,
      "learning_rate": 0.00018898423687838643,
      "loss": 2.2988,
      "step": 143002
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.1251640319824219,
      "learning_rate": 0.0001889804367121644,
      "loss": 2.1491,
      "step": 143003
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.2306419610977173,
      "learning_rate": 0.00018897663656658296,
      "loss": 1.8229,
      "step": 143004
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.0281628370285034,
      "learning_rate": 0.00018897283644164282,
      "loss": 1.981,
      "step": 143005
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.0890498161315918,
      "learning_rate": 0.00018896903633734474,
      "loss": 2.024,
      "step": 143006
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.1389480829238892,
      "learning_rate": 0.0001889652362536894,
      "loss": 2.1003,
      "step": 143007
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.1070586442947388,
      "learning_rate": 0.00018896143619067754,
      "loss": 2.0102,
      "step": 143008
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.178592562675476,
      "learning_rate": 0.0001889576361483099,
      "loss": 1.9798,
      "step": 143009
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.0088567733764648,
      "learning_rate": 0.00018895383612658705,
      "loss": 1.8754,
      "step": 143010
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.2590699195861816,
      "learning_rate": 0.00018895003612550985,
      "loss": 2.1382,
      "step": 143011
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.0852314233779907,
      "learning_rate": 0.0001889462361450789,
      "loss": 2.1633,
      "step": 143012
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.0626932382583618,
      "learning_rate": 0.00018894243618529488,
      "loss": 2.2495,
      "step": 143013
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.2069628238677979,
      "learning_rate": 0.0001889386362461586,
      "loss": 2.0007,
      "step": 143014
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.1424916982650757,
      "learning_rate": 0.00018893483632767078,
      "loss": 2.2591,
      "step": 143015
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.0465208292007446,
      "learning_rate": 0.00018893103642983198,
      "loss": 2.0695,
      "step": 143016
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.0978145599365234,
      "learning_rate": 0.00018892723655264304,
      "loss": 2.1375,
      "step": 143017
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.1930620670318604,
      "learning_rate": 0.00018892343669610465,
      "loss": 2.0315,
      "step": 143018
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.3181592226028442,
      "learning_rate": 0.0001889196368602175,
      "loss": 2.1366,
      "step": 143019
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.30986487865448,
      "learning_rate": 0.00018891583704498219,
      "loss": 2.0437,
      "step": 143020
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.103171706199646,
      "learning_rate": 0.00018891203725039963,
      "loss": 2.2156,
      "step": 143021
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.013376235961914,
      "learning_rate": 0.00018890823747647037,
      "loss": 2.02,
      "step": 143022
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.1010624170303345,
      "learning_rate": 0.00018890443772319514,
      "loss": 2.3692,
      "step": 143023
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.1517603397369385,
      "learning_rate": 0.0001889006379905747,
      "loss": 2.1899,
      "step": 143024
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.221557855606079,
      "learning_rate": 0.00018889683827860975,
      "loss": 2.0404,
      "step": 143025
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.1231507062911987,
      "learning_rate": 0.00018889303858730093,
      "loss": 2.2653,
      "step": 143026
    },
    {
      "epoch": 1.86,
      "grad_norm": 0.9852997660636902,
      "learning_rate": 0.00018888923891664908,
      "loss": 2.1442,
      "step": 143027
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.038008689880371,
      "learning_rate": 0.00018888543926665475,
      "loss": 1.9933,
      "step": 143028
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.241548776626587,
      "learning_rate": 0.00018888163963731868,
      "loss": 2.2384,
      "step": 143029
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.0343146324157715,
      "learning_rate": 0.00018887784002864165,
      "loss": 2.0169,
      "step": 143030
    },
    {
      "epoch": 1.86,
      "grad_norm": 0.9588002562522888,
      "learning_rate": 0.0001888740404406243,
      "loss": 1.9441,
      "step": 143031
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.1731107234954834,
      "learning_rate": 0.00018887024087326738,
      "loss": 2.0293,
      "step": 143032
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.375718355178833,
      "learning_rate": 0.00018886644132657169,
      "loss": 1.9194,
      "step": 143033
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.077805757522583,
      "learning_rate": 0.0001888626418005377,
      "loss": 2.144,
      "step": 143034
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.034629225730896,
      "learning_rate": 0.00018885884229516626,
      "loss": 1.9727,
      "step": 143035
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.198364019393921,
      "learning_rate": 0.00018885504281045804,
      "loss": 2.0832,
      "step": 143036
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.0819910764694214,
      "learning_rate": 0.0001888512433464138,
      "loss": 1.9997,
      "step": 143037
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.4665489196777344,
      "learning_rate": 0.00018884744390303417,
      "loss": 2.0103,
      "step": 143038
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.0539543628692627,
      "learning_rate": 0.00018884364448032003,
      "loss": 2.1944,
      "step": 143039
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.151017665863037,
      "learning_rate": 0.00018883984507827185,
      "loss": 1.9523,
      "step": 143040
    },
    {
      "epoch": 1.86,
      "grad_norm": 0.998500406742096,
      "learning_rate": 0.00018883604569689042,
      "loss": 2.0645,
      "step": 143041
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.1456825733184814,
      "learning_rate": 0.0001888322463361765,
      "loss": 2.0561,
      "step": 143042
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.0502723455429077,
      "learning_rate": 0.00018882844699613074,
      "loss": 2.1334,
      "step": 143043
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.114833116531372,
      "learning_rate": 0.0001888246476767539,
      "loss": 2.2064,
      "step": 143044
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.4099464416503906,
      "learning_rate": 0.0001888208483780467,
      "loss": 2.1385,
      "step": 143045
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.0397191047668457,
      "learning_rate": 0.00018881704910000973,
      "loss": 2.1825,
      "step": 143046
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.121769905090332,
      "learning_rate": 0.0001888132498426438,
      "loss": 2.0067,
      "step": 143047
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.1084197759628296,
      "learning_rate": 0.00018880945060594954,
      "loss": 2.1693,
      "step": 143048
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.194872260093689,
      "learning_rate": 0.00018880565138992773,
      "loss": 2.0573,
      "step": 143049
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.2292203903198242,
      "learning_rate": 0.00018880185219457906,
      "loss": 2.125,
      "step": 143050
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.1264673471450806,
      "learning_rate": 0.0001887980530199043,
      "loss": 2.0786,
      "step": 143051
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.0649775266647339,
      "learning_rate": 0.000188794253865904,
      "loss": 2.1175,
      "step": 143052
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.1084198951721191,
      "learning_rate": 0.00018879045473257894,
      "loss": 1.9366,
      "step": 143053
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.199476718902588,
      "learning_rate": 0.00018878665561992982,
      "loss": 1.9654,
      "step": 143054
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.1270134449005127,
      "learning_rate": 0.00018878285652795736,
      "loss": 2.1504,
      "step": 143055
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.0728470087051392,
      "learning_rate": 0.00018877905745666226,
      "loss": 2.1428,
      "step": 143056
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.0142234563827515,
      "learning_rate": 0.00018877525840604536,
      "loss": 2.2493,
      "step": 143057
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.3288838863372803,
      "learning_rate": 0.00018877145937610713,
      "loss": 2.1674,
      "step": 143058
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.09699285030365,
      "learning_rate": 0.0001887676603668484,
      "loss": 1.8721,
      "step": 143059
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.2296315431594849,
      "learning_rate": 0.00018876386137826984,
      "loss": 1.9678,
      "step": 143060
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.2084146738052368,
      "learning_rate": 0.00018876006241037217,
      "loss": 2.1316,
      "step": 143061
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.1644105911254883,
      "learning_rate": 0.0001887562634631561,
      "loss": 2.0571,
      "step": 143062
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.1807154417037964,
      "learning_rate": 0.00018875246453662237,
      "loss": 2.1738,
      "step": 143063
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.1386542320251465,
      "learning_rate": 0.00018874866563077178,
      "loss": 2.2072,
      "step": 143064
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.0546154975891113,
      "learning_rate": 0.00018874486674560475,
      "loss": 2.1982,
      "step": 143065
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.0317692756652832,
      "learning_rate": 0.00018874106788112217,
      "loss": 2.0748,
      "step": 143066
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.2176707983016968,
      "learning_rate": 0.00018873726903732472,
      "loss": 1.9195,
      "step": 143067
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.1099889278411865,
      "learning_rate": 0.00018873347021421314,
      "loss": 1.9117,
      "step": 143068
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.120369553565979,
      "learning_rate": 0.00018872967141178807,
      "loss": 2.1526,
      "step": 143069
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.0221807956695557,
      "learning_rate": 0.0001887258726300504,
      "loss": 2.0506,
      "step": 143070
    },
    {
      "epoch": 1.86,
      "grad_norm": 0.9166747331619263,
      "learning_rate": 0.00018872207386900057,
      "loss": 1.8937,
      "step": 143071
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.114636778831482,
      "learning_rate": 0.0001887182751286394,
      "loss": 2.0956,
      "step": 143072
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.0761960744857788,
      "learning_rate": 0.00018871447640896762,
      "loss": 1.986,
      "step": 143073
    },
    {
      "epoch": 1.86,
      "grad_norm": 0.9709038734436035,
      "learning_rate": 0.00018871067770998587,
      "loss": 2.2287,
      "step": 143074
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.1226505041122437,
      "learning_rate": 0.00018870687903169495,
      "loss": 2.2293,
      "step": 143075
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.1028934717178345,
      "learning_rate": 0.00018870308037409564,
      "loss": 2.2286,
      "step": 143076
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.1196616888046265,
      "learning_rate": 0.0001886992817371884,
      "loss": 2.2533,
      "step": 143077
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.0242211818695068,
      "learning_rate": 0.00018869548312097403,
      "loss": 2.0723,
      "step": 143078
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.1136977672576904,
      "learning_rate": 0.0001886916845254533,
      "loss": 2.1072,
      "step": 143079
    },
    {
      "epoch": 1.86,
      "grad_norm": 0.9978536367416382,
      "learning_rate": 0.00018868788595062693,
      "loss": 2.2791,
      "step": 143080
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.0686925649642944,
      "learning_rate": 0.00018868408739649553,
      "loss": 2.1495,
      "step": 143081
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.0877693891525269,
      "learning_rate": 0.00018868028886305998,
      "loss": 2.1539,
      "step": 143082
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.2167301177978516,
      "learning_rate": 0.00018867649035032077,
      "loss": 2.2673,
      "step": 143083
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.0721310377120972,
      "learning_rate": 0.0001886726918582787,
      "loss": 2.1392,
      "step": 143084
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.0918920040130615,
      "learning_rate": 0.00018866889338693446,
      "loss": 1.9925,
      "step": 143085
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.0379390716552734,
      "learning_rate": 0.00018866509493628878,
      "loss": 1.9594,
      "step": 143086
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.1625982522964478,
      "learning_rate": 0.00018866129650634242,
      "loss": 2.1321,
      "step": 143087
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.0200971364974976,
      "learning_rate": 0.00018865749809709607,
      "loss": 2.0799,
      "step": 143088
    },
    {
      "epoch": 1.86,
      "grad_norm": 0.9993144273757935,
      "learning_rate": 0.00018865369970855027,
      "loss": 2.0731,
      "step": 143089
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.056412935256958,
      "learning_rate": 0.0001886499013407059,
      "loss": 2.1411,
      "step": 143090
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.0819836854934692,
      "learning_rate": 0.00018864610299356357,
      "loss": 2.1163,
      "step": 143091
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.162739872932434,
      "learning_rate": 0.00018864230466712406,
      "loss": 2.3214,
      "step": 143092
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.0813710689544678,
      "learning_rate": 0.00018863850636138802,
      "loss": 2.109,
      "step": 143093
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.0425410270690918,
      "learning_rate": 0.00018863470807635628,
      "loss": 2.1983,
      "step": 143094
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.1655546426773071,
      "learning_rate": 0.00018863090981202944,
      "loss": 2.0437,
      "step": 143095
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.1280869245529175,
      "learning_rate": 0.00018862711156840815,
      "loss": 2.1794,
      "step": 143096
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.1392154693603516,
      "learning_rate": 0.00018862331334549317,
      "loss": 2.0653,
      "step": 143097
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.1081284284591675,
      "learning_rate": 0.0001886195151432852,
      "loss": 2.0725,
      "step": 143098
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.3521724939346313,
      "learning_rate": 0.000188615716961785,
      "loss": 1.9833,
      "step": 143099
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.1567379236221313,
      "learning_rate": 0.00018861191880099328,
      "loss": 2.2195,
      "step": 143100
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.2198785543441772,
      "learning_rate": 0.00018860812066091067,
      "loss": 2.1139,
      "step": 143101
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.1503642797470093,
      "learning_rate": 0.0001886043225415379,
      "loss": 2.1256,
      "step": 143102
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.223496913909912,
      "learning_rate": 0.0001886005244428758,
      "loss": 2.0777,
      "step": 143103
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.3954540491104126,
      "learning_rate": 0.00018859672636492485,
      "loss": 2.0583,
      "step": 143104
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.5745306015014648,
      "learning_rate": 0.00018859292830768586,
      "loss": 2.1326,
      "step": 143105
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.1204800605773926,
      "learning_rate": 0.00018858913027115963,
      "loss": 2.1133,
      "step": 143106
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.0775257349014282,
      "learning_rate": 0.00018858533225534673,
      "loss": 2.0652,
      "step": 143107
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.2416985034942627,
      "learning_rate": 0.00018858153426024792,
      "loss": 1.9502,
      "step": 143108
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.0832802057266235,
      "learning_rate": 0.00018857773628586387,
      "loss": 2.0534,
      "step": 143109
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.1112381219863892,
      "learning_rate": 0.00018857393833219543,
      "loss": 2.4974,
      "step": 143110
    },
    {
      "epoch": 1.86,
      "grad_norm": 0.9603438377380371,
      "learning_rate": 0.00018857014039924309,
      "loss": 2.2562,
      "step": 143111
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.0455851554870605,
      "learning_rate": 0.00018856634248700776,
      "loss": 2.0242,
      "step": 143112
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.3802354335784912,
      "learning_rate": 0.00018856254459549,
      "loss": 1.8978,
      "step": 143113
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.3264365196228027,
      "learning_rate": 0.00018855874672469054,
      "loss": 2.1041,
      "step": 143114
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.0651856660842896,
      "learning_rate": 0.0001885549488746101,
      "loss": 1.9989,
      "step": 143115
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.086788296699524,
      "learning_rate": 0.0001885511510452494,
      "loss": 2.0309,
      "step": 143116
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.2134220600128174,
      "learning_rate": 0.00018854735323660924,
      "loss": 1.9558,
      "step": 143117
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.0278024673461914,
      "learning_rate": 0.0001885435554486902,
      "loss": 2.0329,
      "step": 143118
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.0855486392974854,
      "learning_rate": 0.000188539757681493,
      "loss": 2.097,
      "step": 143119
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.0847400426864624,
      "learning_rate": 0.0001885359599350183,
      "loss": 1.8252,
      "step": 143120
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.1713926792144775,
      "learning_rate": 0.00018853216220926692,
      "loss": 2.1464,
      "step": 143121
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.1471409797668457,
      "learning_rate": 0.00018852836450423952,
      "loss": 2.1087,
      "step": 143122
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.0186560153961182,
      "learning_rate": 0.00018852456681993674,
      "loss": 2.2405,
      "step": 143123
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.2551501989364624,
      "learning_rate": 0.00018852076915635955,
      "loss": 2.2253,
      "step": 143124
    },
    {
      "epoch": 1.86,
      "grad_norm": 0.990082323551178,
      "learning_rate": 0.0001885169715135083,
      "loss": 1.9441,
      "step": 143125
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.0933308601379395,
      "learning_rate": 0.0001885131738913838,
      "loss": 2.1994,
      "step": 143126
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.1770421266555786,
      "learning_rate": 0.00018850937628998686,
      "loss": 2.2873,
      "step": 143127
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.0612376928329468,
      "learning_rate": 0.0001885055787093181,
      "loss": 2.0774,
      "step": 143128
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.0555819272994995,
      "learning_rate": 0.0001885017811493783,
      "loss": 2.2503,
      "step": 143129
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.0256675481796265,
      "learning_rate": 0.0001884979836101681,
      "loss": 2.1491,
      "step": 143130
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.1207468509674072,
      "learning_rate": 0.00018849418609168835,
      "loss": 2.2912,
      "step": 143131
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.0760430097579956,
      "learning_rate": 0.0001884903885939395,
      "loss": 1.868,
      "step": 143132
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.0893617868423462,
      "learning_rate": 0.0001884865911169224,
      "loss": 2.0488,
      "step": 143133
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.0404527187347412,
      "learning_rate": 0.00018848279366063777,
      "loss": 2.0233,
      "step": 143134
    },
    {
      "epoch": 1.86,
      "grad_norm": 0.9966337084770203,
      "learning_rate": 0.0001884789962250863,
      "loss": 2.0473,
      "step": 143135
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.0603160858154297,
      "learning_rate": 0.00018847519881026864,
      "loss": 1.9712,
      "step": 143136
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.101283311843872,
      "learning_rate": 0.0001884714014161857,
      "loss": 2.0716,
      "step": 143137
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.002414584159851,
      "learning_rate": 0.00018846760404283788,
      "loss": 2.2524,
      "step": 143138
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.0379444360733032,
      "learning_rate": 0.00018846380669022605,
      "loss": 2.0213,
      "step": 143139
    },
    {
      "epoch": 1.86,
      "grad_norm": 0.9967878460884094,
      "learning_rate": 0.00018846000935835095,
      "loss": 2.0097,
      "step": 143140
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.1643667221069336,
      "learning_rate": 0.00018845621204721316,
      "loss": 2.0747,
      "step": 143141
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.1074841022491455,
      "learning_rate": 0.0001884524147568135,
      "loss": 2.0398,
      "step": 143142
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.0047130584716797,
      "learning_rate": 0.0001884486174871528,
      "loss": 2.0419,
      "step": 143143
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.133100986480713,
      "learning_rate": 0.00018844482023823146,
      "loss": 2.2511,
      "step": 143144
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.0932403802871704,
      "learning_rate": 0.00018844102301005035,
      "loss": 2.2373,
      "step": 143145
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.0781060457229614,
      "learning_rate": 0.00018843722580261013,
      "loss": 1.9187,
      "step": 143146
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.1415938138961792,
      "learning_rate": 0.00018843342861591157,
      "loss": 2.3453,
      "step": 143147
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.1693414449691772,
      "learning_rate": 0.00018842963144995533,
      "loss": 1.9999,
      "step": 143148
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.1990779638290405,
      "learning_rate": 0.0001884258343047422,
      "loss": 2.123,
      "step": 143149
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.2008349895477295,
      "learning_rate": 0.00018842203718027277,
      "loss": 2.0418,
      "step": 143150
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.052551507949829,
      "learning_rate": 0.00018841824007654774,
      "loss": 1.9913,
      "step": 143151
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.5261082649230957,
      "learning_rate": 0.00018841444299356788,
      "loss": 2.1051,
      "step": 143152
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.03030526638031,
      "learning_rate": 0.00018841064593133386,
      "loss": 2.2057,
      "step": 143153
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.0855040550231934,
      "learning_rate": 0.00018840684888984645,
      "loss": 2.0505,
      "step": 143154
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.1108425855636597,
      "learning_rate": 0.0001884030518691064,
      "loss": 2.1574,
      "step": 143155
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.1364930868148804,
      "learning_rate": 0.0001883992548691142,
      "loss": 2.1407,
      "step": 143156
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.2518811225891113,
      "learning_rate": 0.00018839545788987074,
      "loss": 2.0256,
      "step": 143157
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.0504564046859741,
      "learning_rate": 0.0001883916609313766,
      "loss": 2.0302,
      "step": 143158
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.1889458894729614,
      "learning_rate": 0.00018838786399363262,
      "loss": 1.9689,
      "step": 143159
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.0127382278442383,
      "learning_rate": 0.0001883840670766394,
      "loss": 1.9748,
      "step": 143160
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.0089999437332153,
      "learning_rate": 0.0001883802701803978,
      "loss": 1.9354,
      "step": 143161
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.1731295585632324,
      "learning_rate": 0.00018837647330490831,
      "loss": 2.139,
      "step": 143162
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.0696662664413452,
      "learning_rate": 0.00018837267645017177,
      "loss": 2.1107,
      "step": 143163
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.0831313133239746,
      "learning_rate": 0.00018836887961618882,
      "loss": 2.1127,
      "step": 143164
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.2620989084243774,
      "learning_rate": 0.00018836508280296025,
      "loss": 2.02,
      "step": 143165
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.152522325515747,
      "learning_rate": 0.00018836128601048665,
      "loss": 2.2349,
      "step": 143166
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.4914566278457642,
      "learning_rate": 0.00018835748923876896,
      "loss": 2.0889,
      "step": 143167
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.0504381656646729,
      "learning_rate": 0.0001883536924878076,
      "loss": 1.9324,
      "step": 143168
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.0174747705459595,
      "learning_rate": 0.00018834989575760338,
      "loss": 2.185,
      "step": 143169
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.0578521490097046,
      "learning_rate": 0.00018834609904815703,
      "loss": 1.7873,
      "step": 143170
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.0704765319824219,
      "learning_rate": 0.00018834230235946923,
      "loss": 2.1846,
      "step": 143171
    },
    {
      "epoch": 1.86,
      "grad_norm": 0.9566954374313354,
      "learning_rate": 0.00018833850569154072,
      "loss": 2.2204,
      "step": 143172
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.0579215288162231,
      "learning_rate": 0.00018833470904437233,
      "loss": 2.0863,
      "step": 143173
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.2211620807647705,
      "learning_rate": 0.0001883309124179645,
      "loss": 2.0307,
      "step": 143174
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.1543066501617432,
      "learning_rate": 0.00018832711581231803,
      "loss": 1.9923,
      "step": 143175
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.044399380683899,
      "learning_rate": 0.00018832331922743366,
      "loss": 2.1358,
      "step": 143176
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.0797805786132812,
      "learning_rate": 0.00018831952266331212,
      "loss": 2.3239,
      "step": 143177
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.0964850187301636,
      "learning_rate": 0.00018831572611995408,
      "loss": 1.961,
      "step": 143178
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.1710741519927979,
      "learning_rate": 0.00018831192959736033,
      "loss": 2.1555,
      "step": 143179
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.156407117843628,
      "learning_rate": 0.0001883081330955315,
      "loss": 1.999,
      "step": 143180
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.1547127962112427,
      "learning_rate": 0.00018830433661446816,
      "loss": 2.1343,
      "step": 143181
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.233541488647461,
      "learning_rate": 0.00018830054015417122,
      "loss": 1.9558,
      "step": 143182
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.0061534643173218,
      "learning_rate": 0.00018829674371464132,
      "loss": 2.1199,
      "step": 143183
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.1242748498916626,
      "learning_rate": 0.00018829294729587914,
      "loss": 2.0318,
      "step": 143184
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.1743931770324707,
      "learning_rate": 0.00018828915089788548,
      "loss": 2.0337,
      "step": 143185
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.1528576612472534,
      "learning_rate": 0.00018828535452066093,
      "loss": 2.2189,
      "step": 143186
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.0469276905059814,
      "learning_rate": 0.00018828155816420628,
      "loss": 2.0385,
      "step": 143187
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.0502256155014038,
      "learning_rate": 0.00018827776182852216,
      "loss": 2.2282,
      "step": 143188
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.2390329837799072,
      "learning_rate": 0.00018827396551360928,
      "loss": 2.2222,
      "step": 143189
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.338611125946045,
      "learning_rate": 0.0001882701692194684,
      "loss": 2.2056,
      "step": 143190
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.1509472131729126,
      "learning_rate": 0.00018826637294610027,
      "loss": 2.127,
      "step": 143191
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.1511332988739014,
      "learning_rate": 0.00018826257669350547,
      "loss": 2.0613,
      "step": 143192
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.127990961074829,
      "learning_rate": 0.00018825878046168474,
      "loss": 2.027,
      "step": 143193
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.1651742458343506,
      "learning_rate": 0.00018825498425063884,
      "loss": 2.1136,
      "step": 143194
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.1214464902877808,
      "learning_rate": 0.00018825118806036848,
      "loss": 2.0582,
      "step": 143195
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.0590304136276245,
      "learning_rate": 0.0001882473918908743,
      "loss": 1.8856,
      "step": 143196
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.0303807258605957,
      "learning_rate": 0.00018824359574215701,
      "loss": 2.3842,
      "step": 143197
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.0984578132629395,
      "learning_rate": 0.00018823979961421746,
      "loss": 2.317,
      "step": 143198
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.0666207075119019,
      "learning_rate": 0.00018823600350705613,
      "loss": 2.2232,
      "step": 143199
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.047561764717102,
      "learning_rate": 0.00018823220742067385,
      "loss": 1.8554,
      "step": 143200
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.085939884185791,
      "learning_rate": 0.0001882284113550713,
      "loss": 2.07,
      "step": 143201
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.1157805919647217,
      "learning_rate": 0.00018822461531024928,
      "loss": 1.8655,
      "step": 143202
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.094285488128662,
      "learning_rate": 0.00018822081928620838,
      "loss": 1.813,
      "step": 143203
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.1206856966018677,
      "learning_rate": 0.00018821702328294934,
      "loss": 2.1584,
      "step": 143204
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.0767171382904053,
      "learning_rate": 0.00018821322730047283,
      "loss": 2.1885,
      "step": 143205
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.0539214611053467,
      "learning_rate": 0.00018820943133877958,
      "loss": 2.0989,
      "step": 143206
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.1784480810165405,
      "learning_rate": 0.00018820563539787032,
      "loss": 2.0109,
      "step": 143207
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.0586305856704712,
      "learning_rate": 0.00018820183947774576,
      "loss": 2.1607,
      "step": 143208
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.0888688564300537,
      "learning_rate": 0.00018819804357840663,
      "loss": 2.2062,
      "step": 143209
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.350124716758728,
      "learning_rate": 0.00018819424769985364,
      "loss": 2.014,
      "step": 143210
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.1221680641174316,
      "learning_rate": 0.00018819045184208732,
      "loss": 1.8903,
      "step": 143211
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.0493834018707275,
      "learning_rate": 0.00018818665600510851,
      "loss": 2.0413,
      "step": 143212
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.1024259328842163,
      "learning_rate": 0.00018818286018891796,
      "loss": 2.4611,
      "step": 143213
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.2479265928268433,
      "learning_rate": 0.00018817906439351628,
      "loss": 2.1966,
      "step": 143214
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.3908965587615967,
      "learning_rate": 0.00018817526861890425,
      "loss": 2.1299,
      "step": 143215
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.2666352987289429,
      "learning_rate": 0.00018817147286508267,
      "loss": 2.195,
      "step": 143216
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.081250548362732,
      "learning_rate": 0.000188167677132052,
      "loss": 2.0696,
      "step": 143217
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.0312089920043945,
      "learning_rate": 0.00018816388141981307,
      "loss": 2.1923,
      "step": 143218
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.1361844539642334,
      "learning_rate": 0.00018816008572836659,
      "loss": 2.2877,
      "step": 143219
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.4759416580200195,
      "learning_rate": 0.0001881562900577132,
      "loss": 2.3451,
      "step": 143220
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.151326060295105,
      "learning_rate": 0.00018815249440785374,
      "loss": 1.9945,
      "step": 143221
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.1736193895339966,
      "learning_rate": 0.00018814869877878894,
      "loss": 2.1464,
      "step": 143222
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.1753860712051392,
      "learning_rate": 0.0001881449031705193,
      "loss": 2.1356,
      "step": 143223
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.3999661207199097,
      "learning_rate": 0.00018814110758304562,
      "loss": 2.2344,
      "step": 143224
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.1419416666030884,
      "learning_rate": 0.00018813731201636864,
      "loss": 2.335,
      "step": 143225
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.034427523612976,
      "learning_rate": 0.000188133516470489,
      "loss": 2.0623,
      "step": 143226
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.2148085832595825,
      "learning_rate": 0.00018812972094540747,
      "loss": 2.1968,
      "step": 143227
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.2573175430297852,
      "learning_rate": 0.0001881259254411249,
      "loss": 2.1452,
      "step": 143228
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.2308346033096313,
      "learning_rate": 0.00018812212995764164,
      "loss": 2.0714,
      "step": 143229
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.1944628953933716,
      "learning_rate": 0.00018811833449495864,
      "loss": 1.9957,
      "step": 143230
    },
    {
      "epoch": 1.86,
      "grad_norm": 0.9915499091148376,
      "learning_rate": 0.00018811453905307656,
      "loss": 2.0916,
      "step": 143231
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.018809199333191,
      "learning_rate": 0.00018811074363199605,
      "loss": 2.0884,
      "step": 143232
    },
    {
      "epoch": 1.86,
      "grad_norm": 1.6087944507598877,
      "learning_rate": 0.0001881069482317179,
      "loss": 2.0771,
      "step": 143233
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.1025258302688599,
      "learning_rate": 0.0001881031528522429,
      "loss": 1.7204,
      "step": 143234
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.0502815246582031,
      "learning_rate": 0.00018809935749357153,
      "loss": 1.9073,
      "step": 143235
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.1118102073669434,
      "learning_rate": 0.00018809556215570458,
      "loss": 2.1244,
      "step": 143236
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.2041279077529907,
      "learning_rate": 0.00018809176683864278,
      "loss": 1.9881,
      "step": 143237
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.2314376831054688,
      "learning_rate": 0.00018808797154238683,
      "loss": 2.2381,
      "step": 143238
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.1069538593292236,
      "learning_rate": 0.00018808417626693742,
      "loss": 1.9867,
      "step": 143239
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.1702828407287598,
      "learning_rate": 0.00018808038101229545,
      "loss": 2.0402,
      "step": 143240
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.0360749959945679,
      "learning_rate": 0.00018807658577846127,
      "loss": 2.12,
      "step": 143241
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.1737849712371826,
      "learning_rate": 0.00018807279056543577,
      "loss": 1.9776,
      "step": 143242
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.051456332206726,
      "learning_rate": 0.0001880689953732197,
      "loss": 2.0279,
      "step": 143243
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.058903694152832,
      "learning_rate": 0.0001880652002018137,
      "loss": 2.2467,
      "step": 143244
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.0651508569717407,
      "learning_rate": 0.00018806140505121845,
      "loss": 1.9697,
      "step": 143245
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.0248117446899414,
      "learning_rate": 0.00018805760992143486,
      "loss": 2.0367,
      "step": 143246
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.1029366254806519,
      "learning_rate": 0.0001880538148124633,
      "loss": 2.151,
      "step": 143247
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.2704180479049683,
      "learning_rate": 0.0001880500197243047,
      "loss": 2.0402,
      "step": 143248
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.0482648611068726,
      "learning_rate": 0.00018804622465695974,
      "loss": 1.936,
      "step": 143249
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.0045799016952515,
      "learning_rate": 0.00018804242961042905,
      "loss": 1.8946,
      "step": 143250
    },
    {
      "epoch": 1.87,
      "grad_norm": 0.9941148161888123,
      "learning_rate": 0.0001880386345847134,
      "loss": 1.8957,
      "step": 143251
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.2887126207351685,
      "learning_rate": 0.0001880348395798136,
      "loss": 2.2328,
      "step": 143252
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.054259181022644,
      "learning_rate": 0.00018803104459573012,
      "loss": 2.1091,
      "step": 143253
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.1215569972991943,
      "learning_rate": 0.0001880272496324638,
      "loss": 2.1469,
      "step": 143254
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.0823734998703003,
      "learning_rate": 0.00018802345469001532,
      "loss": 2.074,
      "step": 143255
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.0443201065063477,
      "learning_rate": 0.00018801965976838537,
      "loss": 2.2964,
      "step": 143256
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.069673776626587,
      "learning_rate": 0.00018801586486757467,
      "loss": 2.0085,
      "step": 143257
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.3446084260940552,
      "learning_rate": 0.00018801206998758408,
      "loss": 2.138,
      "step": 143258
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.0796265602111816,
      "learning_rate": 0.00018800827512841404,
      "loss": 1.9612,
      "step": 143259
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.0050798654556274,
      "learning_rate": 0.00018800448029006537,
      "loss": 2.1072,
      "step": 143260
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.150282382965088,
      "learning_rate": 0.0001880006854725388,
      "loss": 1.9472,
      "step": 143261
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.1597775220870972,
      "learning_rate": 0.000187996890675835,
      "loss": 2.2921,
      "step": 143262
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.1601587533950806,
      "learning_rate": 0.00018799309589995466,
      "loss": 1.8774,
      "step": 143263
    },
    {
      "epoch": 1.87,
      "grad_norm": 0.9967638254165649,
      "learning_rate": 0.00018798930114489855,
      "loss": 2.3081,
      "step": 143264
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.2694467306137085,
      "learning_rate": 0.0001879855064106675,
      "loss": 2.3362,
      "step": 143265
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.2798837423324585,
      "learning_rate": 0.00018798171169726185,
      "loss": 1.8747,
      "step": 143266
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.0908879041671753,
      "learning_rate": 0.0001879779170046826,
      "loss": 2.0735,
      "step": 143267
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.082300066947937,
      "learning_rate": 0.00018797412233293034,
      "loss": 2.0122,
      "step": 143268
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.2460956573486328,
      "learning_rate": 0.0001879703276820058,
      "loss": 2.0006,
      "step": 143269
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.0752778053283691,
      "learning_rate": 0.00018796653305190968,
      "loss": 2.1494,
      "step": 143270
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.0449742078781128,
      "learning_rate": 0.00018796273844264277,
      "loss": 2.3626,
      "step": 143271
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.1149893999099731,
      "learning_rate": 0.0001879589438542057,
      "loss": 2.1839,
      "step": 143272
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.1116700172424316,
      "learning_rate": 0.00018795514928659913,
      "loss": 1.9685,
      "step": 143273
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.0419731140136719,
      "learning_rate": 0.00018795135473982378,
      "loss": 2.3292,
      "step": 143274
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.1009408235549927,
      "learning_rate": 0.0001879475602138804,
      "loss": 2.2265,
      "step": 143275
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.0539536476135254,
      "learning_rate": 0.0001879437657087697,
      "loss": 2.2165,
      "step": 143276
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.032208800315857,
      "learning_rate": 0.00018793997122449244,
      "loss": 2.1564,
      "step": 143277
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.2223899364471436,
      "learning_rate": 0.00018793617676104917,
      "loss": 2.0558,
      "step": 143278
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.0250247716903687,
      "learning_rate": 0.00018793238231844074,
      "loss": 1.9475,
      "step": 143279
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.032814383506775,
      "learning_rate": 0.00018792858789666773,
      "loss": 2.06,
      "step": 143280
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.0598660707473755,
      "learning_rate": 0.00018792479349573094,
      "loss": 2.2997,
      "step": 143281
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.0771087408065796,
      "learning_rate": 0.00018792099911563102,
      "loss": 2.026,
      "step": 143282
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.0505578517913818,
      "learning_rate": 0.00018791720475636877,
      "loss": 2.0094,
      "step": 143283
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.0936272144317627,
      "learning_rate": 0.00018791341041794478,
      "loss": 1.9501,
      "step": 143284
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.1029211282730103,
      "learning_rate": 0.0001879096161003598,
      "loss": 2.0426,
      "step": 143285
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.052727222442627,
      "learning_rate": 0.00018790582180361454,
      "loss": 2.1969,
      "step": 143286
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.0682882070541382,
      "learning_rate": 0.00018790202752770973,
      "loss": 1.9114,
      "step": 143287
    },
    {
      "epoch": 1.87,
      "grad_norm": 0.9466520547866821,
      "learning_rate": 0.000187898233272646,
      "loss": 2.0964,
      "step": 143288
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.0400036573410034,
      "learning_rate": 0.0001878944390384242,
      "loss": 2.0367,
      "step": 143289
    },
    {
      "epoch": 1.87,
      "grad_norm": 0.9970904588699341,
      "learning_rate": 0.00018789064482504483,
      "loss": 2.0191,
      "step": 143290
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.1665599346160889,
      "learning_rate": 0.00018788685063250875,
      "loss": 2.1288,
      "step": 143291
    },
    {
      "epoch": 1.87,
      "grad_norm": 0.9512306451797485,
      "learning_rate": 0.0001878830564608166,
      "loss": 1.8841,
      "step": 143292
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.212018370628357,
      "learning_rate": 0.0001878792623099691,
      "loss": 2.0058,
      "step": 143293
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.1021661758422852,
      "learning_rate": 0.00018787546817996704,
      "loss": 1.996,
      "step": 143294
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.1447454690933228,
      "learning_rate": 0.00018787167407081104,
      "loss": 2.2386,
      "step": 143295
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.1411199569702148,
      "learning_rate": 0.00018786787998250175,
      "loss": 1.9875,
      "step": 143296
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.1430367231369019,
      "learning_rate": 0.0001878640859150399,
      "loss": 2.0812,
      "step": 143297
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.0186638832092285,
      "learning_rate": 0.0001878602918684263,
      "loss": 1.8859,
      "step": 143298
    },
    {
      "epoch": 1.87,
      "grad_norm": 0.9659832119941711,
      "learning_rate": 0.00018785649784266154,
      "loss": 2.132,
      "step": 143299
    },
    {
      "epoch": 1.87,
      "grad_norm": 0.9972512722015381,
      "learning_rate": 0.00018785270383774643,
      "loss": 1.9069,
      "step": 143300
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.100765347480774,
      "learning_rate": 0.0001878489098536817,
      "loss": 2.1013,
      "step": 143301
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.0856133699417114,
      "learning_rate": 0.00018784511589046786,
      "loss": 1.8749,
      "step": 143302
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.0666803121566772,
      "learning_rate": 0.00018784132194810574,
      "loss": 2.1254,
      "step": 143303
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.122504711151123,
      "learning_rate": 0.000187837528026596,
      "loss": 2.0481,
      "step": 143304
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.0530844926834106,
      "learning_rate": 0.00018783373412593942,
      "loss": 2.027,
      "step": 143305
    },
    {
      "epoch": 1.87,
      "grad_norm": 0.9902398586273193,
      "learning_rate": 0.00018782994024613663,
      "loss": 2.0897,
      "step": 143306
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.1026185750961304,
      "learning_rate": 0.00018782614638718852,
      "loss": 2.1004,
      "step": 143307
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.0443118810653687,
      "learning_rate": 0.0001878223525490955,
      "loss": 2.042,
      "step": 143308
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.1932035684585571,
      "learning_rate": 0.00018781855873185844,
      "loss": 2.2267,
      "step": 143309
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.211755633354187,
      "learning_rate": 0.00018781476493547804,
      "loss": 2.0178,
      "step": 143310
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.2005709409713745,
      "learning_rate": 0.000187810971159955,
      "loss": 2.1215,
      "step": 143311
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.138107180595398,
      "learning_rate": 0.00018780717740529,
      "loss": 2.1662,
      "step": 143312
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.0175734758377075,
      "learning_rate": 0.00018780338367148388,
      "loss": 1.9878,
      "step": 143313
    },
    {
      "epoch": 1.87,
      "grad_norm": 0.971009373664856,
      "learning_rate": 0.0001877995899585371,
      "loss": 2.0462,
      "step": 143314
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.0952842235565186,
      "learning_rate": 0.0001877957962664505,
      "loss": 1.931,
      "step": 143315
    },
    {
      "epoch": 1.87,
      "grad_norm": 0.9898672699928284,
      "learning_rate": 0.0001877920025952248,
      "loss": 1.9377,
      "step": 143316
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.0811729431152344,
      "learning_rate": 0.00018778820894486064,
      "loss": 1.9426,
      "step": 143317
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.053383708000183,
      "learning_rate": 0.00018778441531535882,
      "loss": 1.9332,
      "step": 143318
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.500591516494751,
      "learning_rate": 0.0001877806217067201,
      "loss": 2.1574,
      "step": 143319
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.1735366582870483,
      "learning_rate": 0.00018777682811894493,
      "loss": 2.2162,
      "step": 143320
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.1463725566864014,
      "learning_rate": 0.00018777303455203421,
      "loss": 1.9471,
      "step": 143321
    },
    {
      "epoch": 1.87,
      "grad_norm": 0.9845678806304932,
      "learning_rate": 0.0001877692410059886,
      "loss": 2.1269,
      "step": 143322
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.009962558746338,
      "learning_rate": 0.00018776544748080876,
      "loss": 2.0992,
      "step": 143323
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.0999491214752197,
      "learning_rate": 0.00018776165397649548,
      "loss": 1.8499,
      "step": 143324
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.0440737009048462,
      "learning_rate": 0.00018775786049304955,
      "loss": 2.0741,
      "step": 143325
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.1336380243301392,
      "learning_rate": 0.00018775406703047142,
      "loss": 2.1681,
      "step": 143326
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.0380510091781616,
      "learning_rate": 0.00018775027358876193,
      "loss": 2.1224,
      "step": 143327
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.1058409214019775,
      "learning_rate": 0.00018774648016792176,
      "loss": 2.0709,
      "step": 143328
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.1394779682159424,
      "learning_rate": 0.0001877426867679517,
      "loss": 2.138,
      "step": 143329
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.0343140363693237,
      "learning_rate": 0.0001877388933888523,
      "loss": 2.1209,
      "step": 143330
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.127082347869873,
      "learning_rate": 0.00018773510003062445,
      "loss": 1.8239,
      "step": 143331
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.061315655708313,
      "learning_rate": 0.00018773130669326885,
      "loss": 2.0633,
      "step": 143332
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.0164927244186401,
      "learning_rate": 0.000187727513376786,
      "loss": 2.1471,
      "step": 143333
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.1206163167953491,
      "learning_rate": 0.0001877237200811767,
      "loss": 2.2679,
      "step": 143334
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.1059373617172241,
      "learning_rate": 0.00018771992680644171,
      "loss": 2.193,
      "step": 143335
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.364158272743225,
      "learning_rate": 0.0001877161335525817,
      "loss": 1.8964,
      "step": 143336
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.0914390087127686,
      "learning_rate": 0.00018771234031959735,
      "loss": 2.1019,
      "step": 143337
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.0664914846420288,
      "learning_rate": 0.00018770854710748955,
      "loss": 1.9858,
      "step": 143338
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.0526789426803589,
      "learning_rate": 0.00018770475391625873,
      "loss": 1.947,
      "step": 143339
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.1128393411636353,
      "learning_rate": 0.00018770096074590572,
      "loss": 2.1262,
      "step": 143340
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.1618353128433228,
      "learning_rate": 0.00018769716759643122,
      "loss": 1.8412,
      "step": 143341
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.0353208780288696,
      "learning_rate": 0.00018769337446783592,
      "loss": 1.914,
      "step": 143342
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.016767144203186,
      "learning_rate": 0.00018768958136012052,
      "loss": 1.9858,
      "step": 143343
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.0363298654556274,
      "learning_rate": 0.0001876857882732859,
      "loss": 2.0383,
      "step": 143344
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.0207533836364746,
      "learning_rate": 0.0001876819952073325,
      "loss": 2.0648,
      "step": 143345
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.1308093070983887,
      "learning_rate": 0.00018767820216226112,
      "loss": 1.933,
      "step": 143346
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.0409122705459595,
      "learning_rate": 0.00018767440913807247,
      "loss": 1.9902,
      "step": 143347
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.0972347259521484,
      "learning_rate": 0.0001876706161347673,
      "loss": 1.8558,
      "step": 143348
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.2458754777908325,
      "learning_rate": 0.00018766682315234627,
      "loss": 1.792,
      "step": 143349
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.177239179611206,
      "learning_rate": 0.00018766303019081022,
      "loss": 2.1514,
      "step": 143350
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.1993352174758911,
      "learning_rate": 0.0001876592372501596,
      "loss": 2.0516,
      "step": 143351
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.0606625080108643,
      "learning_rate": 0.00018765544433039524,
      "loss": 2.2881,
      "step": 143352
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.1780738830566406,
      "learning_rate": 0.0001876516514315179,
      "loss": 1.9094,
      "step": 143353
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.1133191585540771,
      "learning_rate": 0.00018764785855352817,
      "loss": 1.9553,
      "step": 143354
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.0342918634414673,
      "learning_rate": 0.0001876440656964269,
      "loss": 2.2168,
      "step": 143355
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.0469948053359985,
      "learning_rate": 0.00018764027286021471,
      "loss": 1.9624,
      "step": 143356
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.4574530124664307,
      "learning_rate": 0.00018763648004489238,
      "loss": 1.6672,
      "step": 143357
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.0419244766235352,
      "learning_rate": 0.00018763268725046045,
      "loss": 2.1627,
      "step": 143358
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.198400616645813,
      "learning_rate": 0.00018762889447691968,
      "loss": 2.0269,
      "step": 143359
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.3115142583847046,
      "learning_rate": 0.00018762510172427087,
      "loss": 2.0755,
      "step": 143360
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.2783608436584473,
      "learning_rate": 0.0001876213089925147,
      "loss": 1.846,
      "step": 143361
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.1204782724380493,
      "learning_rate": 0.00018761751628165193,
      "loss": 2.0189,
      "step": 143362
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.0549393892288208,
      "learning_rate": 0.00018761372359168304,
      "loss": 2.1816,
      "step": 143363
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.2478257417678833,
      "learning_rate": 0.00018760993092260898,
      "loss": 2.0949,
      "step": 143364
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.034096360206604,
      "learning_rate": 0.0001876061382744303,
      "loss": 2.0265,
      "step": 143365
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.0579783916473389,
      "learning_rate": 0.00018760234564714777,
      "loss": 2.0394,
      "step": 143366
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.0974705219268799,
      "learning_rate": 0.00018759855304076203,
      "loss": 2.006,
      "step": 143367
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.227148175239563,
      "learning_rate": 0.00018759476045527394,
      "loss": 2.4565,
      "step": 143368
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.1106624603271484,
      "learning_rate": 0.00018759096789068406,
      "loss": 2.1097,
      "step": 143369
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.119446873664856,
      "learning_rate": 0.00018758717534699316,
      "loss": 1.8396,
      "step": 143370
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.0254143476486206,
      "learning_rate": 0.00018758338282420193,
      "loss": 2.0645,
      "step": 143371
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.0589759349822998,
      "learning_rate": 0.00018757959032231103,
      "loss": 2.0507,
      "step": 143372
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.0646570920944214,
      "learning_rate": 0.0001875757978413212,
      "loss": 1.9915,
      "step": 143373
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.1232666969299316,
      "learning_rate": 0.00018757200538123325,
      "loss": 2.3057,
      "step": 143374
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.1138889789581299,
      "learning_rate": 0.0001875682129420477,
      "loss": 1.9606,
      "step": 143375
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.210196614265442,
      "learning_rate": 0.00018756442052376534,
      "loss": 1.8748,
      "step": 143376
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.0892786979675293,
      "learning_rate": 0.00018756062812638687,
      "loss": 2.0474,
      "step": 143377
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.0778945684432983,
      "learning_rate": 0.0001875568357499131,
      "loss": 2.1882,
      "step": 143378
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.018691062927246,
      "learning_rate": 0.0001875530433943445,
      "loss": 2.0794,
      "step": 143379
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.0190898180007935,
      "learning_rate": 0.000187549251059682,
      "loss": 1.8346,
      "step": 143380
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.3815375566482544,
      "learning_rate": 0.00018754545874592618,
      "loss": 2.0485,
      "step": 143381
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.0672475099563599,
      "learning_rate": 0.0001875416664530778,
      "loss": 2.1441,
      "step": 143382
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.0138078927993774,
      "learning_rate": 0.00018753787418113754,
      "loss": 2.0914,
      "step": 143383
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.2675327062606812,
      "learning_rate": 0.00018753408193010612,
      "loss": 2.019,
      "step": 143384
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.3740164041519165,
      "learning_rate": 0.00018753028969998418,
      "loss": 1.9858,
      "step": 143385
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.2649918794631958,
      "learning_rate": 0.00018752649749077262,
      "loss": 2.1127,
      "step": 143386
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.4008774757385254,
      "learning_rate": 0.00018752270530247191,
      "loss": 2.0509,
      "step": 143387
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.140259861946106,
      "learning_rate": 0.00018751891313508284,
      "loss": 2.1312,
      "step": 143388
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.0708415508270264,
      "learning_rate": 0.00018751512098860615,
      "loss": 2.1969,
      "step": 143389
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.0839595794677734,
      "learning_rate": 0.00018751132886304253,
      "loss": 2.0899,
      "step": 143390
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.0296858549118042,
      "learning_rate": 0.00018750753675839262,
      "loss": 1.8168,
      "step": 143391
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.0487730503082275,
      "learning_rate": 0.00018750374467465733,
      "loss": 1.9608,
      "step": 143392
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.1020267009735107,
      "learning_rate": 0.0001874999526118371,
      "loss": 1.9755,
      "step": 143393
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.0548770427703857,
      "learning_rate": 0.00018749616056993278,
      "loss": 2.2382,
      "step": 143394
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.4016201496124268,
      "learning_rate": 0.00018749236854894504,
      "loss": 2.2,
      "step": 143395
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.107722282409668,
      "learning_rate": 0.00018748857654887454,
      "loss": 2.1499,
      "step": 143396
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.0497244596481323,
      "learning_rate": 0.00018748478456972207,
      "loss": 1.9268,
      "step": 143397
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.2732890844345093,
      "learning_rate": 0.00018748099261148833,
      "loss": 2.0219,
      "step": 143398
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.144836187362671,
      "learning_rate": 0.00018747720067417405,
      "loss": 2.2457,
      "step": 143399
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.1249561309814453,
      "learning_rate": 0.00018747340875777983,
      "loss": 2.346,
      "step": 143400
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.0425509214401245,
      "learning_rate": 0.00018746961686230638,
      "loss": 2.1493,
      "step": 143401
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.1531420946121216,
      "learning_rate": 0.00018746582498775446,
      "loss": 2.2811,
      "step": 143402
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.134026050567627,
      "learning_rate": 0.00018746203313412479,
      "loss": 2.0733,
      "step": 143403
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.0737011432647705,
      "learning_rate": 0.00018745824130141805,
      "loss": 2.1932,
      "step": 143404
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.0996602773666382,
      "learning_rate": 0.00018745444948963506,
      "loss": 2.1302,
      "step": 143405
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.1576306819915771,
      "learning_rate": 0.0001874506576987763,
      "loss": 2.1205,
      "step": 143406
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.0439083576202393,
      "learning_rate": 0.0001874468659288426,
      "loss": 2.1691,
      "step": 143407
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.1147587299346924,
      "learning_rate": 0.0001874430741798346,
      "loss": 1.8087,
      "step": 143408
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.1894677877426147,
      "learning_rate": 0.0001874392824517531,
      "loss": 2.1005,
      "step": 143409
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.0415310859680176,
      "learning_rate": 0.00018743549074459875,
      "loss": 2.2041,
      "step": 143410
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.159423828125,
      "learning_rate": 0.0001874316990583724,
      "loss": 2.1867,
      "step": 143411
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.1523096561431885,
      "learning_rate": 0.0001874279073930745,
      "loss": 2.0656,
      "step": 143412
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.1088333129882812,
      "learning_rate": 0.00018742411574870584,
      "loss": 2.0034,
      "step": 143413
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.1609300374984741,
      "learning_rate": 0.00018742032412526723,
      "loss": 2.1918,
      "step": 143414
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.1304620504379272,
      "learning_rate": 0.00018741653252275924,
      "loss": 2.0588,
      "step": 143415
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.1149417161941528,
      "learning_rate": 0.00018741274094118268,
      "loss": 2.2464,
      "step": 143416
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.1610701084136963,
      "learning_rate": 0.0001874089493805383,
      "loss": 2.1706,
      "step": 143417
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.668236255645752,
      "learning_rate": 0.00018740515784082664,
      "loss": 2.2827,
      "step": 143418
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.1941322088241577,
      "learning_rate": 0.00018740136632204845,
      "loss": 2.0952,
      "step": 143419
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.034207820892334,
      "learning_rate": 0.00018739757482420452,
      "loss": 2.043,
      "step": 143420
    },
    {
      "epoch": 1.87,
      "grad_norm": 0.9725216627120972,
      "learning_rate": 0.00018739378334729548,
      "loss": 2.1436,
      "step": 143421
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.3794859647750854,
      "learning_rate": 0.00018738999189132208,
      "loss": 1.9856,
      "step": 143422
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.0985639095306396,
      "learning_rate": 0.0001873862004562851,
      "loss": 1.9568,
      "step": 143423
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.3043968677520752,
      "learning_rate": 0.00018738240904218504,
      "loss": 1.8594,
      "step": 143424
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.0695255994796753,
      "learning_rate": 0.0001873786176490227,
      "loss": 2.0706,
      "step": 143425
    },
    {
      "epoch": 1.87,
      "grad_norm": 0.9817402958869934,
      "learning_rate": 0.0001873748262767988,
      "loss": 2.0367,
      "step": 143426
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.3137059211730957,
      "learning_rate": 0.0001873710349255141,
      "loss": 2.0317,
      "step": 143427
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.6713640689849854,
      "learning_rate": 0.0001873672435951692,
      "loss": 2.0822,
      "step": 143428
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.0889726877212524,
      "learning_rate": 0.00018736345228576498,
      "loss": 2.0117,
      "step": 143429
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.0686825513839722,
      "learning_rate": 0.00018735966099730188,
      "loss": 2.1935,
      "step": 143430
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.2937986850738525,
      "learning_rate": 0.00018735586972978084,
      "loss": 2.2089,
      "step": 143431
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.3263428211212158,
      "learning_rate": 0.00018735207848320236,
      "loss": 2.1141,
      "step": 143432
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.0706515312194824,
      "learning_rate": 0.0001873482872575673,
      "loss": 2.1989,
      "step": 143433
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.0199580192565918,
      "learning_rate": 0.00018734449605287633,
      "loss": 1.9196,
      "step": 143434
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.2022548913955688,
      "learning_rate": 0.00018734070486913027,
      "loss": 2.0746,
      "step": 143435
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.19290030002594,
      "learning_rate": 0.00018733691370632953,
      "loss": 2.1371,
      "step": 143436
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.024139404296875,
      "learning_rate": 0.000187333122564475,
      "loss": 1.8719,
      "step": 143437
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.1218963861465454,
      "learning_rate": 0.0001873293314435674,
      "loss": 1.8295,
      "step": 143438
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.2353286743164062,
      "learning_rate": 0.00018732554034360744,
      "loss": 2.1831,
      "step": 143439
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.3837213516235352,
      "learning_rate": 0.0001873217492645957,
      "loss": 2.2573,
      "step": 143440
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.0945080518722534,
      "learning_rate": 0.00018731795820653313,
      "loss": 1.9902,
      "step": 143441
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.177115559577942,
      "learning_rate": 0.00018731416716942018,
      "loss": 2.0864,
      "step": 143442
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.1603838205337524,
      "learning_rate": 0.00018731037615325762,
      "loss": 2.1894,
      "step": 143443
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.0661506652832031,
      "learning_rate": 0.0001873065851580462,
      "loss": 1.8745,
      "step": 143444
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.0568487644195557,
      "learning_rate": 0.0001873027941837866,
      "loss": 2.1061,
      "step": 143445
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.1350431442260742,
      "learning_rate": 0.0001872990032304796,
      "loss": 2.1496,
      "step": 143446
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.03828763961792,
      "learning_rate": 0.00018729521229812586,
      "loss": 2.1382,
      "step": 143447
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.0941555500030518,
      "learning_rate": 0.000187291421386726,
      "loss": 2.0452,
      "step": 143448
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.0022201538085938,
      "learning_rate": 0.00018728763049628085,
      "loss": 1.8404,
      "step": 143449
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.0366871356964111,
      "learning_rate": 0.00018728383962679097,
      "loss": 1.8371,
      "step": 143450
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.0559227466583252,
      "learning_rate": 0.00018728004877825724,
      "loss": 2.2187,
      "step": 143451
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.1629133224487305,
      "learning_rate": 0.00018727625795068018,
      "loss": 1.9079,
      "step": 143452
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.1986339092254639,
      "learning_rate": 0.00018727246714406073,
      "loss": 2.2678,
      "step": 143453
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.0723686218261719,
      "learning_rate": 0.00018726867635839934,
      "loss": 2.1367,
      "step": 143454
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.0275967121124268,
      "learning_rate": 0.00018726488559369687,
      "loss": 1.95,
      "step": 143455
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.1267733573913574,
      "learning_rate": 0.00018726109484995404,
      "loss": 2.1084,
      "step": 143456
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.141448974609375,
      "learning_rate": 0.00018725730412717142,
      "loss": 2.2096,
      "step": 143457
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.1305475234985352,
      "learning_rate": 0.0001872535134253498,
      "loss": 1.9613,
      "step": 143458
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.1229352951049805,
      "learning_rate": 0.00018724972274448992,
      "loss": 2.1025,
      "step": 143459
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.1241639852523804,
      "learning_rate": 0.00018724593208459242,
      "loss": 2.1522,
      "step": 143460
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.134265422821045,
      "learning_rate": 0.000187242141445658,
      "loss": 2.1042,
      "step": 143461
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.0385860204696655,
      "learning_rate": 0.00018723835082768743,
      "loss": 1.9851,
      "step": 143462
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.1542965173721313,
      "learning_rate": 0.00018723456023068142,
      "loss": 2.1132,
      "step": 143463
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.0880000591278076,
      "learning_rate": 0.00018723076965464057,
      "loss": 1.9027,
      "step": 143464
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.0867383480072021,
      "learning_rate": 0.00018722697909956566,
      "loss": 2.0257,
      "step": 143465
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.043775200843811,
      "learning_rate": 0.00018722318856545744,
      "loss": 2.1144,
      "step": 143466
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.0406702756881714,
      "learning_rate": 0.00018721939805231653,
      "loss": 2.0217,
      "step": 143467
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.01519775390625,
      "learning_rate": 0.00018721560756014357,
      "loss": 2.4215,
      "step": 143468
    },
    {
      "epoch": 1.87,
      "grad_norm": 0.9722121357917786,
      "learning_rate": 0.00018721181708893946,
      "loss": 2.1546,
      "step": 143469
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.100965976715088,
      "learning_rate": 0.00018720802663870477,
      "loss": 2.1465,
      "step": 143470
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.0524827241897583,
      "learning_rate": 0.00018720423620944025,
      "loss": 2.14,
      "step": 143471
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.2079094648361206,
      "learning_rate": 0.00018720044580114662,
      "loss": 2.3068,
      "step": 143472
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.3960109949111938,
      "learning_rate": 0.00018719665541382455,
      "loss": 2.0585,
      "step": 143473
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.0345993041992188,
      "learning_rate": 0.00018719286504747468,
      "loss": 2.2627,
      "step": 143474
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.0509464740753174,
      "learning_rate": 0.0001871890747020978,
      "loss": 1.7672,
      "step": 143475
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.4555186033248901,
      "learning_rate": 0.0001871852843776946,
      "loss": 1.9704,
      "step": 143476
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.2216771841049194,
      "learning_rate": 0.0001871814940742658,
      "loss": 2.1911,
      "step": 143477
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.1277433633804321,
      "learning_rate": 0.00018717770379181222,
      "loss": 1.9998,
      "step": 143478
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.432962417602539,
      "learning_rate": 0.00018717391353033428,
      "loss": 2.4111,
      "step": 143479
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.6845568418502808,
      "learning_rate": 0.0001871701232898329,
      "loss": 2.0305,
      "step": 143480
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.1707886457443237,
      "learning_rate": 0.00018716633307030867,
      "loss": 2.1678,
      "step": 143481
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.2119392156600952,
      "learning_rate": 0.00018716254287176235,
      "loss": 2.3039,
      "step": 143482
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.1348576545715332,
      "learning_rate": 0.00018715875269419464,
      "loss": 2.1703,
      "step": 143483
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.0836423635482788,
      "learning_rate": 0.0001871549625376064,
      "loss": 2.1648,
      "step": 143484
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.304574966430664,
      "learning_rate": 0.00018715117240199804,
      "loss": 1.9546,
      "step": 143485
    },
    {
      "epoch": 1.87,
      "grad_norm": 0.985731840133667,
      "learning_rate": 0.00018714738228737044,
      "loss": 2.0351,
      "step": 143486
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.1162819862365723,
      "learning_rate": 0.00018714359219372423,
      "loss": 1.9815,
      "step": 143487
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.2049943208694458,
      "learning_rate": 0.00018713980212106015,
      "loss": 1.9948,
      "step": 143488
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.2581403255462646,
      "learning_rate": 0.00018713601206937896,
      "loss": 2.217,
      "step": 143489
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.1726282835006714,
      "learning_rate": 0.00018713222203868142,
      "loss": 1.9371,
      "step": 143490
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.2412688732147217,
      "learning_rate": 0.00018712843202896797,
      "loss": 2.2003,
      "step": 143491
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.0640777349472046,
      "learning_rate": 0.0001871246420402395,
      "loss": 2.2381,
      "step": 143492
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.323776364326477,
      "learning_rate": 0.00018712085207249672,
      "loss": 2.0735,
      "step": 143493
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.4047976732254028,
      "learning_rate": 0.00018711706212574025,
      "loss": 2.2442,
      "step": 143494
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.0125807523727417,
      "learning_rate": 0.00018711327219997088,
      "loss": 1.9592,
      "step": 143495
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.1243407726287842,
      "learning_rate": 0.0001871094822951894,
      "loss": 1.8672,
      "step": 143496
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.2179133892059326,
      "learning_rate": 0.00018710569241139628,
      "loss": 1.9881,
      "step": 143497
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.2493407726287842,
      "learning_rate": 0.00018710190254859235,
      "loss": 1.8033,
      "step": 143498
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.048008680343628,
      "learning_rate": 0.00018709811270677828,
      "loss": 1.8632,
      "step": 143499
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.1064900159835815,
      "learning_rate": 0.00018709432288595482,
      "loss": 2.0715,
      "step": 143500
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.4119994640350342,
      "learning_rate": 0.00018709053308612266,
      "loss": 2.0706,
      "step": 143501
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.137465238571167,
      "learning_rate": 0.00018708674330728263,
      "loss": 1.9454,
      "step": 143502
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.01626455783844,
      "learning_rate": 0.0001870829535494352,
      "loss": 2.1902,
      "step": 143503
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.1289072036743164,
      "learning_rate": 0.0001870791638125811,
      "loss": 2.1045,
      "step": 143504
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.0636705160140991,
      "learning_rate": 0.0001870753740967212,
      "loss": 2.1314,
      "step": 143505
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.170320749282837,
      "learning_rate": 0.0001870715844018561,
      "loss": 2.3764,
      "step": 143506
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.1191210746765137,
      "learning_rate": 0.0001870677947279865,
      "loss": 2.1126,
      "step": 143507
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.0786226987838745,
      "learning_rate": 0.00018706400507511326,
      "loss": 2.1397,
      "step": 143508
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.0561621189117432,
      "learning_rate": 0.00018706021544323686,
      "loss": 2.0187,
      "step": 143509
    },
    {
      "epoch": 1.87,
      "grad_norm": 0.9927716851234436,
      "learning_rate": 0.00018705642583235804,
      "loss": 1.9696,
      "step": 143510
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.033431887626648,
      "learning_rate": 0.00018705263624247762,
      "loss": 1.9927,
      "step": 143511
    },
    {
      "epoch": 1.87,
      "grad_norm": 0.9731229543685913,
      "learning_rate": 0.0001870488466735962,
      "loss": 2.0217,
      "step": 143512
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.0125139951705933,
      "learning_rate": 0.00018704505712571455,
      "loss": 2.055,
      "step": 143513
    },
    {
      "epoch": 1.87,
      "grad_norm": 0.9734697341918945,
      "learning_rate": 0.0001870412675988335,
      "loss": 2.3516,
      "step": 143514
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.047149419784546,
      "learning_rate": 0.00018703747809295347,
      "loss": 1.8746,
      "step": 143515
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.1565449237823486,
      "learning_rate": 0.00018703368860807532,
      "loss": 2.1847,
      "step": 143516
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.100555419921875,
      "learning_rate": 0.00018702989914419973,
      "loss": 2.1302,
      "step": 143517
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.138041615486145,
      "learning_rate": 0.0001870261097013274,
      "loss": 2.0935,
      "step": 143518
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.1012881994247437,
      "learning_rate": 0.00018702232027945908,
      "loss": 1.9955,
      "step": 143519
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.1062792539596558,
      "learning_rate": 0.0001870185308785955,
      "loss": 2.1062,
      "step": 143520
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.2400094270706177,
      "learning_rate": 0.00018701474149873727,
      "loss": 1.9592,
      "step": 143521
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.0275946855545044,
      "learning_rate": 0.00018701095213988507,
      "loss": 1.9875,
      "step": 143522
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.0996297597885132,
      "learning_rate": 0.0001870071628020397,
      "loss": 2.1764,
      "step": 143523
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.2377543449401855,
      "learning_rate": 0.00018700337348520182,
      "loss": 2.0827,
      "step": 143524
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.0670174360275269,
      "learning_rate": 0.00018699958418937215,
      "loss": 2.1047,
      "step": 143525
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.2179737091064453,
      "learning_rate": 0.00018699579491455152,
      "loss": 1.9078,
      "step": 143526
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.147249698638916,
      "learning_rate": 0.00018699200566074032,
      "loss": 1.9494,
      "step": 143527
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.0306204557418823,
      "learning_rate": 0.00018698821642793953,
      "loss": 2.0871,
      "step": 143528
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.2310147285461426,
      "learning_rate": 0.00018698442721614973,
      "loss": 2.1248,
      "step": 143529
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.0299499034881592,
      "learning_rate": 0.00018698063802537163,
      "loss": 2.0292,
      "step": 143530
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.0477992296218872,
      "learning_rate": 0.000186976848855606,
      "loss": 1.8377,
      "step": 143531
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.0564419031143188,
      "learning_rate": 0.0001869730597068535,
      "loss": 2.2938,
      "step": 143532
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.1043671369552612,
      "learning_rate": 0.00018696927057911496,
      "loss": 2.0373,
      "step": 143533
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.5517476797103882,
      "learning_rate": 0.00018696548147239083,
      "loss": 1.8959,
      "step": 143534
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.0813143253326416,
      "learning_rate": 0.00018696169238668198,
      "loss": 2.0737,
      "step": 143535
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.1147143840789795,
      "learning_rate": 0.00018695790332198903,
      "loss": 2.016,
      "step": 143536
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.066331148147583,
      "learning_rate": 0.0001869541142783128,
      "loss": 1.9939,
      "step": 143537
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.5456207990646362,
      "learning_rate": 0.00018695032525565392,
      "loss": 2.0473,
      "step": 143538
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.1419677734375,
      "learning_rate": 0.00018694653625401315,
      "loss": 1.9785,
      "step": 143539
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.0787357091903687,
      "learning_rate": 0.00018694274727339117,
      "loss": 1.953,
      "step": 143540
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.130759358406067,
      "learning_rate": 0.00018693895831378859,
      "loss": 2.1067,
      "step": 143541
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.200754165649414,
      "learning_rate": 0.00018693516937520621,
      "loss": 2.305,
      "step": 143542
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.246029257774353,
      "learning_rate": 0.0001869313804576447,
      "loss": 1.8632,
      "step": 143543
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.0166791677474976,
      "learning_rate": 0.0001869275915611048,
      "loss": 2.1701,
      "step": 143544
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.0245577096939087,
      "learning_rate": 0.00018692380268558724,
      "loss": 2.2064,
      "step": 143545
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.0615026950836182,
      "learning_rate": 0.00018692001383109258,
      "loss": 2.0164,
      "step": 143546
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.2549382448196411,
      "learning_rate": 0.0001869162249976217,
      "loss": 2.2116,
      "step": 143547
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.2885380983352661,
      "learning_rate": 0.00018691243618517523,
      "loss": 2.0518,
      "step": 143548
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.2725400924682617,
      "learning_rate": 0.00018690864739375385,
      "loss": 2.173,
      "step": 143549
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.0934642553329468,
      "learning_rate": 0.00018690485862335828,
      "loss": 2.1781,
      "step": 143550
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.1848868131637573,
      "learning_rate": 0.00018690106987398933,
      "loss": 2.0664,
      "step": 143551
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.1883606910705566,
      "learning_rate": 0.00018689728114564745,
      "loss": 2.0675,
      "step": 143552
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.1001379489898682,
      "learning_rate": 0.0001868934924383336,
      "loss": 2.1118,
      "step": 143553
    },
    {
      "epoch": 1.87,
      "grad_norm": 2.159274101257324,
      "learning_rate": 0.00018688970375204831,
      "loss": 1.9626,
      "step": 143554
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.6406863927841187,
      "learning_rate": 0.00018688591508679242,
      "loss": 2.0589,
      "step": 143555
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.6187851428985596,
      "learning_rate": 0.00018688212644256654,
      "loss": 2.0543,
      "step": 143556
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.2561061382293701,
      "learning_rate": 0.0001868783378193715,
      "loss": 1.872,
      "step": 143557
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.0888041257858276,
      "learning_rate": 0.00018687454921720783,
      "loss": 2.2216,
      "step": 143558
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.316969871520996,
      "learning_rate": 0.00018687076063607632,
      "loss": 2.0828,
      "step": 143559
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.5598989725112915,
      "learning_rate": 0.00018686697207597765,
      "loss": 2.1086,
      "step": 143560
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.098639965057373,
      "learning_rate": 0.00018686318353691253,
      "loss": 2.0296,
      "step": 143561
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.227192759513855,
      "learning_rate": 0.00018685939501888178,
      "loss": 2.0994,
      "step": 143562
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.3369207382202148,
      "learning_rate": 0.000186855606521886,
      "loss": 2.2953,
      "step": 143563
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.2177506685256958,
      "learning_rate": 0.0001868518180459258,
      "loss": 2.0632,
      "step": 143564
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.4611544609069824,
      "learning_rate": 0.00018684802959100202,
      "loss": 2.3624,
      "step": 143565
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.247257947921753,
      "learning_rate": 0.00018684424115711533,
      "loss": 2.0722,
      "step": 143566
    },
    {
      "epoch": 1.87,
      "grad_norm": 0.9711704254150391,
      "learning_rate": 0.0001868404527442664,
      "loss": 1.94,
      "step": 143567
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.0849030017852783,
      "learning_rate": 0.000186836664352456,
      "loss": 2.2466,
      "step": 143568
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.3065918684005737,
      "learning_rate": 0.0001868328759816849,
      "loss": 2.0912,
      "step": 143569
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.0657596588134766,
      "learning_rate": 0.00018682908763195362,
      "loss": 2.1003,
      "step": 143570
    },
    {
      "epoch": 1.87,
      "grad_norm": 0.9851852655410767,
      "learning_rate": 0.00018682529930326287,
      "loss": 1.9926,
      "step": 143571
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.2106655836105347,
      "learning_rate": 0.0001868215109956135,
      "loss": 2.118,
      "step": 143572
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.0030550956726074,
      "learning_rate": 0.00018681772270900615,
      "loss": 2.0159,
      "step": 143573
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.6115798950195312,
      "learning_rate": 0.00018681393444344146,
      "loss": 2.2102,
      "step": 143574
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.093865156173706,
      "learning_rate": 0.00018681014619892036,
      "loss": 2.0104,
      "step": 143575
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.0729284286499023,
      "learning_rate": 0.0001868063579754433,
      "loss": 2.2115,
      "step": 143576
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.2646074295043945,
      "learning_rate": 0.00018680256977301104,
      "loss": 2.3873,
      "step": 143577
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.1187859773635864,
      "learning_rate": 0.0001867987815916243,
      "loss": 1.9705,
      "step": 143578
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.126795768737793,
      "learning_rate": 0.00018679499343128382,
      "loss": 1.9399,
      "step": 143579
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.12322199344635,
      "learning_rate": 0.0001867912052919903,
      "loss": 2.1924,
      "step": 143580
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.1060456037521362,
      "learning_rate": 0.00018678741717374455,
      "loss": 2.21,
      "step": 143581
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.0375044345855713,
      "learning_rate": 0.00018678362907654703,
      "loss": 2.0462,
      "step": 143582
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.127898931503296,
      "learning_rate": 0.0001867798410003986,
      "loss": 2.2618,
      "step": 143583
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.329141616821289,
      "learning_rate": 0.00018677605294529988,
      "loss": 2.2573,
      "step": 143584
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.1011791229248047,
      "learning_rate": 0.00018677226491125163,
      "loss": 2.1893,
      "step": 143585
    },
    {
      "epoch": 1.87,
      "grad_norm": 0.9971826076507568,
      "learning_rate": 0.0001867684768982546,
      "loss": 1.9434,
      "step": 143586
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.21186101436615,
      "learning_rate": 0.00018676468890630956,
      "loss": 2.3928,
      "step": 143587
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.2737603187561035,
      "learning_rate": 0.00018676090093541693,
      "loss": 2.1038,
      "step": 143588
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.0924910306930542,
      "learning_rate": 0.00018675711298557764,
      "loss": 2.0091,
      "step": 143589
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.0111348628997803,
      "learning_rate": 0.00018675332505679233,
      "loss": 2.1574,
      "step": 143590
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.1571160554885864,
      "learning_rate": 0.0001867495371490617,
      "loss": 1.9955,
      "step": 143591
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.2686549425125122,
      "learning_rate": 0.0001867457492623865,
      "loss": 2.209,
      "step": 143592
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.2716526985168457,
      "learning_rate": 0.00018674196139676746,
      "loss": 1.9896,
      "step": 143593
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.1108897924423218,
      "learning_rate": 0.0001867381735522052,
      "loss": 2.1782,
      "step": 143594
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.1139920949935913,
      "learning_rate": 0.00018673438572870037,
      "loss": 1.8294,
      "step": 143595
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.0091453790664673,
      "learning_rate": 0.00018673059792625378,
      "loss": 2.0574,
      "step": 143596
    },
    {
      "epoch": 1.87,
      "grad_norm": 0.9855365753173828,
      "learning_rate": 0.0001867268101448661,
      "loss": 2.0918,
      "step": 143597
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.114342212677002,
      "learning_rate": 0.0001867230223845381,
      "loss": 2.217,
      "step": 143598
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.0688785314559937,
      "learning_rate": 0.00018671923464527035,
      "loss": 2.0304,
      "step": 143599
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.1022485494613647,
      "learning_rate": 0.0001867154469270638,
      "loss": 1.9809,
      "step": 143600
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.049950361251831,
      "learning_rate": 0.00018671165922991886,
      "loss": 2.2003,
      "step": 143601
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.1137820482254028,
      "learning_rate": 0.00018670787155383636,
      "loss": 2.2522,
      "step": 143602
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.0395492315292358,
      "learning_rate": 0.000186704083898817,
      "loss": 1.8738,
      "step": 143603
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.1492646932601929,
      "learning_rate": 0.0001867002962648615,
      "loss": 2.0026,
      "step": 143604
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.021470546722412,
      "learning_rate": 0.00018669650865197054,
      "loss": 2.1763,
      "step": 143605
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.1973061561584473,
      "learning_rate": 0.00018669272106014494,
      "loss": 2.0675,
      "step": 143606
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.2475926876068115,
      "learning_rate": 0.00018668893348938525,
      "loss": 2.0855,
      "step": 143607
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.081862211227417,
      "learning_rate": 0.00018668514593969218,
      "loss": 2.0997,
      "step": 143608
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.0925413370132446,
      "learning_rate": 0.00018668135841106646,
      "loss": 2.2197,
      "step": 143609
    },
    {
      "epoch": 1.87,
      "grad_norm": 0.9858286380767822,
      "learning_rate": 0.00018667757090350885,
      "loss": 2.0481,
      "step": 143610
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.0872989892959595,
      "learning_rate": 0.00018667378341702002,
      "loss": 2.1499,
      "step": 143611
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.293205976486206,
      "learning_rate": 0.00018666999595160077,
      "loss": 2.1749,
      "step": 143612
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.0707831382751465,
      "learning_rate": 0.00018666620850725158,
      "loss": 2.0746,
      "step": 143613
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.0981383323669434,
      "learning_rate": 0.0001866624210839733,
      "loss": 1.8698,
      "step": 143614
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.2394108772277832,
      "learning_rate": 0.0001866586336817666,
      "loss": 2.0957,
      "step": 143615
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.082236886024475,
      "learning_rate": 0.0001866548463006322,
      "loss": 2.0469,
      "step": 143616
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.1127036809921265,
      "learning_rate": 0.00018665105894057085,
      "loss": 2.0572,
      "step": 143617
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.028992772102356,
      "learning_rate": 0.00018664727160158328,
      "loss": 2.095,
      "step": 143618
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.1464787721633911,
      "learning_rate": 0.00018664348428367,
      "loss": 2.1316,
      "step": 143619
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.0738327503204346,
      "learning_rate": 0.00018663969698683186,
      "loss": 2.0857,
      "step": 143620
    },
    {
      "epoch": 1.87,
      "grad_norm": 0.9940458536148071,
      "learning_rate": 0.0001866359097110695,
      "loss": 2.2756,
      "step": 143621
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.1587188243865967,
      "learning_rate": 0.00018663212245638374,
      "loss": 1.9931,
      "step": 143622
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.3209009170532227,
      "learning_rate": 0.00018662833522277515,
      "loss": 1.9551,
      "step": 143623
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.1325324773788452,
      "learning_rate": 0.00018662454801024456,
      "loss": 1.9852,
      "step": 143624
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.1308813095092773,
      "learning_rate": 0.00018662076081879265,
      "loss": 2.1617,
      "step": 143625
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.1766153573989868,
      "learning_rate": 0.00018661697364841994,
      "loss": 1.9678,
      "step": 143626
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.089979648590088,
      "learning_rate": 0.00018661318649912734,
      "loss": 2.0395,
      "step": 143627
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.1304036378860474,
      "learning_rate": 0.00018660939937091548,
      "loss": 2.2358,
      "step": 143628
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.2019262313842773,
      "learning_rate": 0.00018660561226378506,
      "loss": 1.8926,
      "step": 143629
    },
    {
      "epoch": 1.87,
      "grad_norm": 0.9773696660995483,
      "learning_rate": 0.00018660182517773682,
      "loss": 2.0124,
      "step": 143630
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.041174292564392,
      "learning_rate": 0.00018659803811277143,
      "loss": 2.0544,
      "step": 143631
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.2720383405685425,
      "learning_rate": 0.00018659425106888966,
      "loss": 2.2094,
      "step": 143632
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.190832495689392,
      "learning_rate": 0.0001865904640460921,
      "loss": 2.0032,
      "step": 143633
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.072081446647644,
      "learning_rate": 0.00018658667704437952,
      "loss": 2.1563,
      "step": 143634
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.143928050994873,
      "learning_rate": 0.0001865828900637526,
      "loss": 2.1893,
      "step": 143635
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.1368749141693115,
      "learning_rate": 0.0001865791031042121,
      "loss": 2.0035,
      "step": 143636
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.1758936643600464,
      "learning_rate": 0.0001865753161657587,
      "loss": 2.0443,
      "step": 143637
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.0459790229797363,
      "learning_rate": 0.000186571529248393,
      "loss": 2.2355,
      "step": 143638
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.1090835332870483,
      "learning_rate": 0.00018656774235211585,
      "loss": 1.9372,
      "step": 143639
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.207646131515503,
      "learning_rate": 0.00018656395547692796,
      "loss": 1.9753,
      "step": 143640
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.0155282020568848,
      "learning_rate": 0.0001865601686228299,
      "loss": 2.221,
      "step": 143641
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.0547035932540894,
      "learning_rate": 0.0001865563817898225,
      "loss": 2.0033,
      "step": 143642
    },
    {
      "epoch": 1.87,
      "grad_norm": 0.9852619767189026,
      "learning_rate": 0.0001865525949779063,
      "loss": 2.0124,
      "step": 143643
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.1044046878814697,
      "learning_rate": 0.0001865488081870822,
      "loss": 2.0181,
      "step": 143644
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.0481820106506348,
      "learning_rate": 0.00018654502141735079,
      "loss": 2.1465,
      "step": 143645
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.0913575887680054,
      "learning_rate": 0.0001865412346687128,
      "loss": 2.0219,
      "step": 143646
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.052167296409607,
      "learning_rate": 0.00018653744794116896,
      "loss": 2.2871,
      "step": 143647
    },
    {
      "epoch": 1.87,
      "grad_norm": 0.9558382034301758,
      "learning_rate": 0.00018653366123472,
      "loss": 1.9942,
      "step": 143648
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.1311646699905396,
      "learning_rate": 0.00018652987454936652,
      "loss": 2.3322,
      "step": 143649
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.0031092166900635,
      "learning_rate": 0.00018652608788510925,
      "loss": 2.0439,
      "step": 143650
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.036878228187561,
      "learning_rate": 0.0001865223012419489,
      "loss": 1.9221,
      "step": 143651
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.0610538721084595,
      "learning_rate": 0.00018651851461988625,
      "loss": 1.8662,
      "step": 143652
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.158246397972107,
      "learning_rate": 0.00018651472801892193,
      "loss": 2.1185,
      "step": 143653
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.1121957302093506,
      "learning_rate": 0.00018651094143905673,
      "loss": 2.0636,
      "step": 143654
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.088086724281311,
      "learning_rate": 0.00018650715488029127,
      "loss": 1.8472,
      "step": 143655
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.1144318580627441,
      "learning_rate": 0.00018650336834262618,
      "loss": 2.0306,
      "step": 143656
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.0758415460586548,
      "learning_rate": 0.00018649958182606228,
      "loss": 2.1104,
      "step": 143657
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.0761204957962036,
      "learning_rate": 0.0001864957953306003,
      "loss": 2.3275,
      "step": 143658
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.0114420652389526,
      "learning_rate": 0.00018649200885624084,
      "loss": 1.9621,
      "step": 143659
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.045912742614746,
      "learning_rate": 0.0001864882224029847,
      "loss": 2.1199,
      "step": 143660
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.0736546516418457,
      "learning_rate": 0.00018648443597083262,
      "loss": 1.9323,
      "step": 143661
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.1963468790054321,
      "learning_rate": 0.00018648064955978513,
      "loss": 2.1603,
      "step": 143662
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.093411922454834,
      "learning_rate": 0.00018647686316984304,
      "loss": 2.1677,
      "step": 143663
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.1173138618469238,
      "learning_rate": 0.000186473076801007,
      "loss": 2.1069,
      "step": 143664
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.0390193462371826,
      "learning_rate": 0.00018646929045327778,
      "loss": 2.117,
      "step": 143665
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.140731692314148,
      "learning_rate": 0.00018646550412665608,
      "loss": 2.3445,
      "step": 143666
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.0979464054107666,
      "learning_rate": 0.00018646171782114268,
      "loss": 2.2159,
      "step": 143667
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.149212121963501,
      "learning_rate": 0.00018645793153673813,
      "loss": 2.1006,
      "step": 143668
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.0084930658340454,
      "learning_rate": 0.00018645414527344314,
      "loss": 2.1691,
      "step": 143669
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.1061666011810303,
      "learning_rate": 0.00018645035903125848,
      "loss": 2.1176,
      "step": 143670
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.1783472299575806,
      "learning_rate": 0.00018644657281018483,
      "loss": 2.312,
      "step": 143671
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.1665263175964355,
      "learning_rate": 0.00018644278661022294,
      "loss": 2.2881,
      "step": 143672
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.3692817687988281,
      "learning_rate": 0.00018643900043137357,
      "loss": 1.9361,
      "step": 143673
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.2115757465362549,
      "learning_rate": 0.0001864352142736372,
      "loss": 2.2125,
      "step": 143674
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.1605887413024902,
      "learning_rate": 0.00018643142813701467,
      "loss": 2.0068,
      "step": 143675
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.2235071659088135,
      "learning_rate": 0.0001864276420215067,
      "loss": 1.9973,
      "step": 143676
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.2019299268722534,
      "learning_rate": 0.00018642385592711398,
      "loss": 2.0068,
      "step": 143677
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.4041119813919067,
      "learning_rate": 0.00018642006985383723,
      "loss": 2.1709,
      "step": 143678
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.1373292207717896,
      "learning_rate": 0.0001864162838016772,
      "loss": 2.1779,
      "step": 143679
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.1123692989349365,
      "learning_rate": 0.00018641249777063442,
      "loss": 2.1983,
      "step": 143680
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.3223932981491089,
      "learning_rate": 0.00018640871176070972,
      "loss": 2.1222,
      "step": 143681
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.0412441492080688,
      "learning_rate": 0.00018640492577190376,
      "loss": 2.08,
      "step": 143682
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.0698318481445312,
      "learning_rate": 0.00018640113980421732,
      "loss": 2.1211,
      "step": 143683
    },
    {
      "epoch": 1.87,
      "grad_norm": 0.9887681603431702,
      "learning_rate": 0.000186397353857651,
      "loss": 2.2186,
      "step": 143684
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.1441048383712769,
      "learning_rate": 0.0001863935679322057,
      "loss": 2.1131,
      "step": 143685
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.1919785737991333,
      "learning_rate": 0.00018638978202788182,
      "loss": 2.0578,
      "step": 143686
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.2307240962982178,
      "learning_rate": 0.0001863859961446803,
      "loss": 2.2071,
      "step": 143687
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.1032723188400269,
      "learning_rate": 0.0001863822102826017,
      "loss": 2.1213,
      "step": 143688
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.078366994857788,
      "learning_rate": 0.00018637842444164682,
      "loss": 2.1486,
      "step": 143689
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.0140290260314941,
      "learning_rate": 0.00018637463862181632,
      "loss": 2.1371,
      "step": 143690
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.071358561515808,
      "learning_rate": 0.00018637085282311103,
      "loss": 2.0444,
      "step": 143691
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.123874306678772,
      "learning_rate": 0.00018636706704553148,
      "loss": 2.2195,
      "step": 143692
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.0908561944961548,
      "learning_rate": 0.0001863632812890784,
      "loss": 2.054,
      "step": 143693
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.0907738208770752,
      "learning_rate": 0.00018635949555375253,
      "loss": 1.9277,
      "step": 143694
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.0506261587142944,
      "learning_rate": 0.00018635570983955457,
      "loss": 2.2787,
      "step": 143695
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.1089500188827515,
      "learning_rate": 0.00018635192414648527,
      "loss": 2.0424,
      "step": 143696
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.0854002237319946,
      "learning_rate": 0.00018634813847454536,
      "loss": 2.2336,
      "step": 143697
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.10625422000885,
      "learning_rate": 0.00018634435282373536,
      "loss": 1.9109,
      "step": 143698
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.1114177703857422,
      "learning_rate": 0.0001863405671940561,
      "loss": 2.0298,
      "step": 143699
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.1525245904922485,
      "learning_rate": 0.0001863367815855083,
      "loss": 2.1421,
      "step": 143700
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.095726728439331,
      "learning_rate": 0.00018633299599809264,
      "loss": 1.9908,
      "step": 143701
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.1209074258804321,
      "learning_rate": 0.00018632921043180978,
      "loss": 2.332,
      "step": 143702
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.0682172775268555,
      "learning_rate": 0.0001863254248866606,
      "loss": 2.1092,
      "step": 143703
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.1535282135009766,
      "learning_rate": 0.00018632163936264554,
      "loss": 2.0992,
      "step": 143704
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.073669195175171,
      "learning_rate": 0.00018631785385976544,
      "loss": 2.2076,
      "step": 143705
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.0897865295410156,
      "learning_rate": 0.00018631406837802105,
      "loss": 1.8997,
      "step": 143706
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.0081754922866821,
      "learning_rate": 0.00018631028291741294,
      "loss": 1.8994,
      "step": 143707
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.2841382026672363,
      "learning_rate": 0.00018630649747794194,
      "loss": 2.0136,
      "step": 143708
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.0436816215515137,
      "learning_rate": 0.00018630271205960877,
      "loss": 1.8688,
      "step": 143709
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.1519312858581543,
      "learning_rate": 0.00018629892666241404,
      "loss": 2.2235,
      "step": 143710
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.0923281908035278,
      "learning_rate": 0.00018629514128635846,
      "loss": 2.077,
      "step": 143711
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.0269981622695923,
      "learning_rate": 0.0001862913559314428,
      "loss": 2.2482,
      "step": 143712
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.3057469129562378,
      "learning_rate": 0.0001862875705976676,
      "loss": 2.0375,
      "step": 143713
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.1402634382247925,
      "learning_rate": 0.00018628378528503378,
      "loss": 1.9959,
      "step": 143714
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.0439963340759277,
      "learning_rate": 0.000186279999993542,
      "loss": 2.0079,
      "step": 143715
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.1853492259979248,
      "learning_rate": 0.00018627621472319284,
      "loss": 2.0497,
      "step": 143716
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.1181035041809082,
      "learning_rate": 0.0001862724294739871,
      "loss": 2.048,
      "step": 143717
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.1024739742279053,
      "learning_rate": 0.00018626864424592545,
      "loss": 2.0663,
      "step": 143718
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.2914050817489624,
      "learning_rate": 0.0001862648590390086,
      "loss": 2.2574,
      "step": 143719
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.3525291681289673,
      "learning_rate": 0.00018626107385323726,
      "loss": 1.9641,
      "step": 143720
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.248024344444275,
      "learning_rate": 0.00018625728868861222,
      "loss": 1.9687,
      "step": 143721
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.0651335716247559,
      "learning_rate": 0.000186253503545134,
      "loss": 2.0408,
      "step": 143722
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.2252633571624756,
      "learning_rate": 0.00018624971842280344,
      "loss": 2.0933,
      "step": 143723
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.217577338218689,
      "learning_rate": 0.00018624593332162122,
      "loss": 2.0875,
      "step": 143724
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.0301700830459595,
      "learning_rate": 0.00018624214824158793,
      "loss": 1.9572,
      "step": 143725
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.1248162984848022,
      "learning_rate": 0.00018623836318270445,
      "loss": 1.9285,
      "step": 143726
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.0263274908065796,
      "learning_rate": 0.00018623457814497137,
      "loss": 2.2031,
      "step": 143727
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.1572155952453613,
      "learning_rate": 0.00018623079312838953,
      "loss": 2.1124,
      "step": 143728
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.1148203611373901,
      "learning_rate": 0.0001862270081329594,
      "loss": 2.0737,
      "step": 143729
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.079272985458374,
      "learning_rate": 0.0001862232231586819,
      "loss": 2.1289,
      "step": 143730
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.080045223236084,
      "learning_rate": 0.0001862194382055576,
      "loss": 2.2958,
      "step": 143731
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.0362409353256226,
      "learning_rate": 0.00018621565327358732,
      "loss": 1.8275,
      "step": 143732
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.1295886039733887,
      "learning_rate": 0.00018621186836277162,
      "loss": 2.2296,
      "step": 143733
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.0489051342010498,
      "learning_rate": 0.00018620808347311142,
      "loss": 2.0187,
      "step": 143734
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.040482759475708,
      "learning_rate": 0.00018620429860460714,
      "loss": 2.0187,
      "step": 143735
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.0773324966430664,
      "learning_rate": 0.00018620051375725966,
      "loss": 2.0729,
      "step": 143736
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.099302887916565,
      "learning_rate": 0.00018619672893106963,
      "loss": 1.9364,
      "step": 143737
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.032938003540039,
      "learning_rate": 0.00018619294412603783,
      "loss": 1.9448,
      "step": 143738
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.1299370527267456,
      "learning_rate": 0.0001861891593421649,
      "loss": 2.16,
      "step": 143739
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.1136596202850342,
      "learning_rate": 0.0001861853745794516,
      "loss": 2.0618,
      "step": 143740
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.2778652906417847,
      "learning_rate": 0.00018618158983789852,
      "loss": 2.1535,
      "step": 143741
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.0454901456832886,
      "learning_rate": 0.00018617780511750645,
      "loss": 2.0419,
      "step": 143742
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.2957350015640259,
      "learning_rate": 0.00018617402041827603,
      "loss": 2.1154,
      "step": 143743
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.1622225046157837,
      "learning_rate": 0.00018617023574020802,
      "loss": 1.9326,
      "step": 143744
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.0924072265625,
      "learning_rate": 0.00018616645108330315,
      "loss": 1.991,
      "step": 143745
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.1797395944595337,
      "learning_rate": 0.00018616266644756215,
      "loss": 2.13,
      "step": 143746
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.094744086265564,
      "learning_rate": 0.00018615888183298558,
      "loss": 2.1336,
      "step": 143747
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.0777331590652466,
      "learning_rate": 0.00018615509723957422,
      "loss": 2.2438,
      "step": 143748
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.0886260271072388,
      "learning_rate": 0.00018615131266732876,
      "loss": 1.9641,
      "step": 143749
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.1482281684875488,
      "learning_rate": 0.0001861475281162499,
      "loss": 2.1531,
      "step": 143750
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.0594209432601929,
      "learning_rate": 0.0001861437435863384,
      "loss": 2.0885,
      "step": 143751
    },
    {
      "epoch": 1.87,
      "grad_norm": 0.9905786514282227,
      "learning_rate": 0.00018613995907759508,
      "loss": 2.109,
      "step": 143752
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.055707335472107,
      "learning_rate": 0.00018613617459002028,
      "loss": 2.09,
      "step": 143753
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.011741280555725,
      "learning_rate": 0.00018613239012361496,
      "loss": 1.9696,
      "step": 143754
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.096828818321228,
      "learning_rate": 0.0001861286056783798,
      "loss": 2.2598,
      "step": 143755
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.1331169605255127,
      "learning_rate": 0.00018612482125431544,
      "loss": 1.8234,
      "step": 143756
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.0471570491790771,
      "learning_rate": 0.00018612103685142266,
      "loss": 2.0858,
      "step": 143757
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.2748816013336182,
      "learning_rate": 0.00018611725246970226,
      "loss": 1.9451,
      "step": 143758
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.136443018913269,
      "learning_rate": 0.00018611346810915467,
      "loss": 2.0329,
      "step": 143759
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.0866272449493408,
      "learning_rate": 0.00018610968376978074,
      "loss": 1.9618,
      "step": 143760
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.1886552572250366,
      "learning_rate": 0.00018610589945158116,
      "loss": 2.1984,
      "step": 143761
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.1018890142440796,
      "learning_rate": 0.00018610211515455665,
      "loss": 2.1348,
      "step": 143762
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.0284970998764038,
      "learning_rate": 0.0001860983308787079,
      "loss": 2.0311,
      "step": 143763
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.1491018533706665,
      "learning_rate": 0.00018609454662403573,
      "loss": 2.0974,
      "step": 143764
    },
    {
      "epoch": 1.87,
      "grad_norm": 0.9623392820358276,
      "learning_rate": 0.00018609076239054065,
      "loss": 2.2066,
      "step": 143765
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.1722067594528198,
      "learning_rate": 0.0001860869781782234,
      "loss": 2.0137,
      "step": 143766
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.1939928531646729,
      "learning_rate": 0.00018608319398708475,
      "loss": 2.0559,
      "step": 143767
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.4663954973220825,
      "learning_rate": 0.00018607940981712543,
      "loss": 1.9749,
      "step": 143768
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.0839192867279053,
      "learning_rate": 0.00018607562566834603,
      "loss": 2.0832,
      "step": 143769
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.0880200862884521,
      "learning_rate": 0.0001860718415407475,
      "loss": 2.2644,
      "step": 143770
    },
    {
      "epoch": 1.87,
      "grad_norm": 0.9675464630126953,
      "learning_rate": 0.00018606805743433022,
      "loss": 2.0841,
      "step": 143771
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.0861550569534302,
      "learning_rate": 0.00018606427334909504,
      "loss": 2.2037,
      "step": 143772
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.112074851989746,
      "learning_rate": 0.00018606048928504262,
      "loss": 2.3403,
      "step": 143773
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.157947063446045,
      "learning_rate": 0.00018605670524217378,
      "loss": 2.2213,
      "step": 143774
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.0676106214523315,
      "learning_rate": 0.0001860529212204891,
      "loss": 2.2232,
      "step": 143775
    },
    {
      "epoch": 1.87,
      "grad_norm": 0.9714938998222351,
      "learning_rate": 0.00018604913721998947,
      "loss": 2.0684,
      "step": 143776
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.0292679071426392,
      "learning_rate": 0.00018604535324067534,
      "loss": 2.0473,
      "step": 143777
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.0003758668899536,
      "learning_rate": 0.00018604156928254752,
      "loss": 2.2807,
      "step": 143778
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.0637295246124268,
      "learning_rate": 0.00018603778534560675,
      "loss": 2.0779,
      "step": 143779
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.18375825881958,
      "learning_rate": 0.00018603400142985367,
      "loss": 2.0745,
      "step": 143780
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.1652498245239258,
      "learning_rate": 0.000186030217535289,
      "loss": 2.1126,
      "step": 143781
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.1681522130966187,
      "learning_rate": 0.00018602643366191364,
      "loss": 2.1408,
      "step": 143782
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.1285964250564575,
      "learning_rate": 0.000186022649809728,
      "loss": 1.9446,
      "step": 143783
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.0686765909194946,
      "learning_rate": 0.00018601886597873285,
      "loss": 1.9313,
      "step": 143784
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.1590244770050049,
      "learning_rate": 0.00018601508216892898,
      "loss": 1.9707,
      "step": 143785
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.1165515184402466,
      "learning_rate": 0.00018601129838031701,
      "loss": 2.0105,
      "step": 143786
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.275584101676941,
      "learning_rate": 0.00018600751461289776,
      "loss": 2.0808,
      "step": 143787
    },
    {
      "epoch": 1.87,
      "grad_norm": 0.9684080481529236,
      "learning_rate": 0.00018600373086667194,
      "loss": 1.9188,
      "step": 143788
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.0455858707427979,
      "learning_rate": 0.00018599994714164005,
      "loss": 2.1212,
      "step": 143789
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.276123285293579,
      "learning_rate": 0.000185996163437803,
      "loss": 2.0353,
      "step": 143790
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.0515283346176147,
      "learning_rate": 0.00018599237975516135,
      "loss": 2.0908,
      "step": 143791
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.023464322090149,
      "learning_rate": 0.00018598859609371586,
      "loss": 1.9365,
      "step": 143792
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.0241371393203735,
      "learning_rate": 0.00018598481245346728,
      "loss": 2.105,
      "step": 143793
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.1639597415924072,
      "learning_rate": 0.0001859810288344162,
      "loss": 2.0539,
      "step": 143794
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.099574089050293,
      "learning_rate": 0.0001859772452365636,
      "loss": 2.138,
      "step": 143795
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.0708112716674805,
      "learning_rate": 0.00018597346165990988,
      "loss": 2.3254,
      "step": 143796
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.030199408531189,
      "learning_rate": 0.00018596967810445577,
      "loss": 1.9071,
      "step": 143797
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.0481438636779785,
      "learning_rate": 0.0001859658945702021,
      "loss": 2.3408,
      "step": 143798
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.1308584213256836,
      "learning_rate": 0.0001859621110571495,
      "loss": 2.1153,
      "step": 143799
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.0859241485595703,
      "learning_rate": 0.00018595832756529868,
      "loss": 1.9337,
      "step": 143800
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.136411190032959,
      "learning_rate": 0.00018595454409465044,
      "loss": 1.886,
      "step": 143801
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.061158537864685,
      "learning_rate": 0.0001859507606452054,
      "loss": 2.2509,
      "step": 143802
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.1098783016204834,
      "learning_rate": 0.0001859469772169642,
      "loss": 2.0304,
      "step": 143803
    },
    {
      "epoch": 1.87,
      "grad_norm": 0.9958827495574951,
      "learning_rate": 0.00018594319380992765,
      "loss": 1.7423,
      "step": 143804
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.18410325050354,
      "learning_rate": 0.00018593941042409638,
      "loss": 2.2931,
      "step": 143805
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.1293532848358154,
      "learning_rate": 0.00018593562705947113,
      "loss": 2.001,
      "step": 143806
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.1331135034561157,
      "learning_rate": 0.00018593184371605266,
      "loss": 2.2697,
      "step": 143807
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.1309784650802612,
      "learning_rate": 0.00018592806039384154,
      "loss": 2.1635,
      "step": 143808
    },
    {
      "epoch": 1.87,
      "grad_norm": 0.9773526191711426,
      "learning_rate": 0.00018592427709283865,
      "loss": 2.1756,
      "step": 143809
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.1521658897399902,
      "learning_rate": 0.00018592049381304447,
      "loss": 2.0995,
      "step": 143810
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.1106427907943726,
      "learning_rate": 0.00018591671055445986,
      "loss": 2.1214,
      "step": 143811
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.121649980545044,
      "learning_rate": 0.0001859129273170855,
      "loss": 2.0108,
      "step": 143812
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.1053309440612793,
      "learning_rate": 0.0001859091441009221,
      "loss": 2.1416,
      "step": 143813
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.029667615890503,
      "learning_rate": 0.0001859053609059703,
      "loss": 2.0319,
      "step": 143814
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.067642331123352,
      "learning_rate": 0.00018590157773223088,
      "loss": 2.0852,
      "step": 143815
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.0018532276153564,
      "learning_rate": 0.00018589779457970455,
      "loss": 2.0873,
      "step": 143816
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.099560260772705,
      "learning_rate": 0.00018589401144839188,
      "loss": 2.2105,
      "step": 143817
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.1197264194488525,
      "learning_rate": 0.0001858902283382937,
      "loss": 2.0316,
      "step": 143818
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.0879546403884888,
      "learning_rate": 0.00018588644524941073,
      "loss": 2.3395,
      "step": 143819
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.0175174474716187,
      "learning_rate": 0.00018588266218174358,
      "loss": 2.1393,
      "step": 143820
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.0765354633331299,
      "learning_rate": 0.000185878879135293,
      "loss": 2.0422,
      "step": 143821
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.093002200126648,
      "learning_rate": 0.00018587509611005964,
      "loss": 2.3313,
      "step": 143822
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.0304218530654907,
      "learning_rate": 0.00018587131310604437,
      "loss": 2.031,
      "step": 143823
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.0768239498138428,
      "learning_rate": 0.00018586753012324772,
      "loss": 1.8964,
      "step": 143824
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.045350193977356,
      "learning_rate": 0.00018586374716167046,
      "loss": 2.1633,
      "step": 143825
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.0944690704345703,
      "learning_rate": 0.00018585996422131325,
      "loss": 2.1944,
      "step": 143826
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.088582158088684,
      "learning_rate": 0.00018585618130217683,
      "loss": 2.4422,
      "step": 143827
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.1117634773254395,
      "learning_rate": 0.0001858523984042619,
      "loss": 1.9436,
      "step": 143828
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.0565146207809448,
      "learning_rate": 0.00018584861552756917,
      "loss": 2.0403,
      "step": 143829
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.1781976222991943,
      "learning_rate": 0.00018584483267209933,
      "loss": 2.2105,
      "step": 143830
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.0348193645477295,
      "learning_rate": 0.00018584104983785323,
      "loss": 2.2258,
      "step": 143831
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.0425057411193848,
      "learning_rate": 0.00018583726702483126,
      "loss": 1.8931,
      "step": 143832
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.347155213356018,
      "learning_rate": 0.00018583348423303433,
      "loss": 2.2039,
      "step": 143833
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.2125557661056519,
      "learning_rate": 0.00018582970146246313,
      "loss": 2.1411,
      "step": 143834
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.1711643934249878,
      "learning_rate": 0.0001858259187131183,
      "loss": 2.1813,
      "step": 143835
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.019647240638733,
      "learning_rate": 0.00018582213598500064,
      "loss": 2.0597,
      "step": 143836
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.0872913599014282,
      "learning_rate": 0.0001858183532781109,
      "loss": 2.1748,
      "step": 143837
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.0496940612792969,
      "learning_rate": 0.00018581457059244956,
      "loss": 2.1148,
      "step": 143838
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.0392284393310547,
      "learning_rate": 0.00018581078792801744,
      "loss": 2.1631,
      "step": 143839
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.1145384311676025,
      "learning_rate": 0.00018580700528481527,
      "loss": 2.175,
      "step": 143840
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.0687216520309448,
      "learning_rate": 0.00018580322266284368,
      "loss": 2.0052,
      "step": 143841
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.2569091320037842,
      "learning_rate": 0.00018579944006210346,
      "loss": 2.0756,
      "step": 143842
    },
    {
      "epoch": 1.87,
      "grad_norm": 0.979049026966095,
      "learning_rate": 0.00018579565748259543,
      "loss": 2.0095,
      "step": 143843
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.1360905170440674,
      "learning_rate": 0.00018579187492431996,
      "loss": 2.2263,
      "step": 143844
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.2740470170974731,
      "learning_rate": 0.000185788092387278,
      "loss": 2.0965,
      "step": 143845
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.1511362791061401,
      "learning_rate": 0.00018578430987147015,
      "loss": 2.0395,
      "step": 143846
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.2222505807876587,
      "learning_rate": 0.0001857805273768972,
      "loss": 2.2108,
      "step": 143847
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.2512210607528687,
      "learning_rate": 0.00018577674490355975,
      "loss": 2.0375,
      "step": 143848
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.7227189540863037,
      "learning_rate": 0.0001857729624514587,
      "loss": 2.0986,
      "step": 143849
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.263661503791809,
      "learning_rate": 0.00018576918002059448,
      "loss": 2.2348,
      "step": 143850
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.196044683456421,
      "learning_rate": 0.00018576539761096795,
      "loss": 2.0845,
      "step": 143851
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.0624959468841553,
      "learning_rate": 0.00018576161522257978,
      "loss": 2.1975,
      "step": 143852
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.07294762134552,
      "learning_rate": 0.00018575783285543063,
      "loss": 2.2444,
      "step": 143853
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.0459718704223633,
      "learning_rate": 0.00018575405050952134,
      "loss": 2.047,
      "step": 143854
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.0934832096099854,
      "learning_rate": 0.00018575026818485262,
      "loss": 2.203,
      "step": 143855
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.0604093074798584,
      "learning_rate": 0.00018574648588142494,
      "loss": 2.1739,
      "step": 143856
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.0145796537399292,
      "learning_rate": 0.0001857427035992392,
      "loss": 2.0243,
      "step": 143857
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.0136165618896484,
      "learning_rate": 0.00018573892133829596,
      "loss": 2.1205,
      "step": 143858
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.041986107826233,
      "learning_rate": 0.00018573513909859605,
      "loss": 2.173,
      "step": 143859
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.0914312601089478,
      "learning_rate": 0.00018573135688014015,
      "loss": 2.0051,
      "step": 143860
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.0080385208129883,
      "learning_rate": 0.00018572757468292894,
      "loss": 1.8894,
      "step": 143861
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.2989829778671265,
      "learning_rate": 0.00018572379250696326,
      "loss": 2.1131,
      "step": 143862
    },
    {
      "epoch": 1.87,
      "grad_norm": 0.9630675315856934,
      "learning_rate": 0.0001857200103522436,
      "loss": 2.1141,
      "step": 143863
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.209747314453125,
      "learning_rate": 0.00018571622821877067,
      "loss": 2.3205,
      "step": 143864
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.113424301147461,
      "learning_rate": 0.0001857124461065453,
      "loss": 2.2342,
      "step": 143865
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.070092797279358,
      "learning_rate": 0.0001857086640155681,
      "loss": 2.1519,
      "step": 143866
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.0568312406539917,
      "learning_rate": 0.00018570488194583987,
      "loss": 2.2035,
      "step": 143867
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.135047197341919,
      "learning_rate": 0.00018570109989736138,
      "loss": 2.1287,
      "step": 143868
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.0383732318878174,
      "learning_rate": 0.00018569731787013304,
      "loss": 2.2447,
      "step": 143869
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.0054831504821777,
      "learning_rate": 0.00018569353586415574,
      "loss": 2.0098,
      "step": 143870
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.0330852270126343,
      "learning_rate": 0.00018568975387943019,
      "loss": 2.2726,
      "step": 143871
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.0705796480178833,
      "learning_rate": 0.00018568597191595705,
      "loss": 2.1311,
      "step": 143872
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.0605586767196655,
      "learning_rate": 0.00018568218997373707,
      "loss": 2.2032,
      "step": 143873
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.0863615274429321,
      "learning_rate": 0.0001856784080527711,
      "loss": 2.1596,
      "step": 143874
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.027052402496338,
      "learning_rate": 0.00018567462615305945,
      "loss": 2.132,
      "step": 143875
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.1335994005203247,
      "learning_rate": 0.00018567084427460309,
      "loss": 1.9905,
      "step": 143876
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.0591565370559692,
      "learning_rate": 0.0001856670624174027,
      "loss": 2.0064,
      "step": 143877
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.0728577375411987,
      "learning_rate": 0.00018566328058145892,
      "loss": 2.0781,
      "step": 143878
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.1631659269332886,
      "learning_rate": 0.0001856594987667725,
      "loss": 1.9531,
      "step": 143879
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.1176012754440308,
      "learning_rate": 0.00018565571697334426,
      "loss": 1.9998,
      "step": 143880
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.0488256216049194,
      "learning_rate": 0.0001856519352011747,
      "loss": 2.2235,
      "step": 143881
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.1360609531402588,
      "learning_rate": 0.00018564815345026455,
      "loss": 2.2831,
      "step": 143882
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.2127705812454224,
      "learning_rate": 0.00018564437172061458,
      "loss": 1.9486,
      "step": 143883
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.0854820013046265,
      "learning_rate": 0.0001856405900122255,
      "loss": 2.0321,
      "step": 143884
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.1807271242141724,
      "learning_rate": 0.00018563680832509797,
      "loss": 2.1847,
      "step": 143885
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.2029986381530762,
      "learning_rate": 0.00018563302665923286,
      "loss": 2.2262,
      "step": 143886
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.1243164539337158,
      "learning_rate": 0.0001856292450146306,
      "loss": 2.1152,
      "step": 143887
    },
    {
      "epoch": 1.87,
      "grad_norm": 0.9897826313972473,
      "learning_rate": 0.00018562546339129203,
      "loss": 1.7812,
      "step": 143888
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.0508475303649902,
      "learning_rate": 0.00018562168178921782,
      "loss": 2.1103,
      "step": 143889
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.1601457595825195,
      "learning_rate": 0.00018561790020840872,
      "loss": 2.2296,
      "step": 143890
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.1991280317306519,
      "learning_rate": 0.0001856141186488654,
      "loss": 2.2047,
      "step": 143891
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.1042476892471313,
      "learning_rate": 0.00018561033711058864,
      "loss": 2.0403,
      "step": 143892
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.1673935651779175,
      "learning_rate": 0.00018560655559357904,
      "loss": 2.2961,
      "step": 143893
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.1237540245056152,
      "learning_rate": 0.00018560277409783736,
      "loss": 2.3444,
      "step": 143894
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.128976821899414,
      "learning_rate": 0.0001855989926233642,
      "loss": 2.241,
      "step": 143895
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.1522666215896606,
      "learning_rate": 0.00018559521117016045,
      "loss": 2.0766,
      "step": 143896
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.183749794960022,
      "learning_rate": 0.00018559142973822662,
      "loss": 1.9604,
      "step": 143897
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.156692385673523,
      "learning_rate": 0.00018558764832756362,
      "loss": 1.978,
      "step": 143898
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.1966264247894287,
      "learning_rate": 0.00018558386693817193,
      "loss": 2.1183,
      "step": 143899
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.4149225950241089,
      "learning_rate": 0.00018558008557005237,
      "loss": 2.2722,
      "step": 143900
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.112690806388855,
      "learning_rate": 0.0001855763042232057,
      "loss": 2.2189,
      "step": 143901
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.0772149562835693,
      "learning_rate": 0.0001855725228976325,
      "loss": 2.1814,
      "step": 143902
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.0569357872009277,
      "learning_rate": 0.00018556874159333353,
      "loss": 2.1429,
      "step": 143903
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.221445083618164,
      "learning_rate": 0.00018556496031030953,
      "loss": 2.3186,
      "step": 143904
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.1806976795196533,
      "learning_rate": 0.00018556117904856113,
      "loss": 2.1,
      "step": 143905
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.2679020166397095,
      "learning_rate": 0.00018555739780808905,
      "loss": 2.1262,
      "step": 143906
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.0568739175796509,
      "learning_rate": 0.000185553616588894,
      "loss": 2.1633,
      "step": 143907
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.0237747430801392,
      "learning_rate": 0.0001855498353909768,
      "loss": 1.8982,
      "step": 143908
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.0878777503967285,
      "learning_rate": 0.00018554605421433794,
      "loss": 2.0604,
      "step": 143909
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.3299468755722046,
      "learning_rate": 0.00018554227305897828,
      "loss": 2.2215,
      "step": 143910
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.150774359703064,
      "learning_rate": 0.00018553849192489842,
      "loss": 2.1484,
      "step": 143911
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.3202661275863647,
      "learning_rate": 0.00018553471081209915,
      "loss": 2.0324,
      "step": 143912
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.4219279289245605,
      "learning_rate": 0.00018553092972058111,
      "loss": 2.1126,
      "step": 143913
    },
    {
      "epoch": 1.87,
      "grad_norm": 0.996310830116272,
      "learning_rate": 0.00018552714865034502,
      "loss": 2.225,
      "step": 143914
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.249732494354248,
      "learning_rate": 0.00018552336760139167,
      "loss": 2.0473,
      "step": 143915
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.0814588069915771,
      "learning_rate": 0.00018551958657372173,
      "loss": 1.9983,
      "step": 143916
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.1538746356964111,
      "learning_rate": 0.00018551580556733574,
      "loss": 2.2358,
      "step": 143917
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.224554419517517,
      "learning_rate": 0.00018551202458223453,
      "loss": 2.2164,
      "step": 143918
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.0929505825042725,
      "learning_rate": 0.00018550824361841884,
      "loss": 2.1237,
      "step": 143919
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.1232343912124634,
      "learning_rate": 0.00018550446267588926,
      "loss": 1.9939,
      "step": 143920
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.0983617305755615,
      "learning_rate": 0.00018550068175464662,
      "loss": 2.2097,
      "step": 143921
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.0852348804473877,
      "learning_rate": 0.00018549690085469163,
      "loss": 2.3034,
      "step": 143922
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.1650878190994263,
      "learning_rate": 0.00018549311997602486,
      "loss": 1.9004,
      "step": 143923
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.0596548318862915,
      "learning_rate": 0.00018548933911864704,
      "loss": 1.9415,
      "step": 143924
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.2164053916931152,
      "learning_rate": 0.00018548555828255895,
      "loss": 2.2349,
      "step": 143925
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.527598261833191,
      "learning_rate": 0.00018548177746776124,
      "loss": 2.109,
      "step": 143926
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.008093237876892,
      "learning_rate": 0.0001854779966742546,
      "loss": 1.9825,
      "step": 143927
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.0911449193954468,
      "learning_rate": 0.00018547421590203982,
      "loss": 2.0902,
      "step": 143928
    },
    {
      "epoch": 1.87,
      "grad_norm": 0.9961552023887634,
      "learning_rate": 0.0001854704351511176,
      "loss": 1.9722,
      "step": 143929
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.145361304283142,
      "learning_rate": 0.0001854666544214885,
      "loss": 2.2729,
      "step": 143930
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.1630306243896484,
      "learning_rate": 0.00018546287371315335,
      "loss": 1.9039,
      "step": 143931
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.065682291984558,
      "learning_rate": 0.0001854590930261128,
      "loss": 2.0302,
      "step": 143932
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.1588834524154663,
      "learning_rate": 0.0001854553123603675,
      "loss": 2.1574,
      "step": 143933
    },
    {
      "epoch": 1.87,
      "grad_norm": 0.9590806365013123,
      "learning_rate": 0.00018545153171591828,
      "loss": 2.114,
      "step": 143934
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.0872352123260498,
      "learning_rate": 0.0001854477510927659,
      "loss": 2.0589,
      "step": 143935
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.2090320587158203,
      "learning_rate": 0.00018544397049091079,
      "loss": 1.839,
      "step": 143936
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.0725114345550537,
      "learning_rate": 0.00018544018991035384,
      "loss": 2.1377,
      "step": 143937
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.1703338623046875,
      "learning_rate": 0.00018543640935109575,
      "loss": 2.0676,
      "step": 143938
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.2148265838623047,
      "learning_rate": 0.0001854326288131371,
      "loss": 2.2663,
      "step": 143939
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.1115357875823975,
      "learning_rate": 0.00018542884829647876,
      "loss": 2.0201,
      "step": 143940
    },
    {
      "epoch": 1.87,
      "grad_norm": 0.9661020040512085,
      "learning_rate": 0.0001854250678011215,
      "loss": 1.9348,
      "step": 143941
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.0594520568847656,
      "learning_rate": 0.0001854212873270657,
      "loss": 1.9486,
      "step": 143942
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.0689938068389893,
      "learning_rate": 0.00018541750687431225,
      "loss": 1.8806,
      "step": 143943
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.1699120998382568,
      "learning_rate": 0.0001854137264428619,
      "loss": 2.1675,
      "step": 143944
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.1775789260864258,
      "learning_rate": 0.00018540994603271523,
      "loss": 2.1358,
      "step": 143945
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.1650384664535522,
      "learning_rate": 0.00018540616564387307,
      "loss": 2.0317,
      "step": 143946
    },
    {
      "epoch": 1.87,
      "grad_norm": 0.9743818640708923,
      "learning_rate": 0.0001854023852763362,
      "loss": 1.9201,
      "step": 143947
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.1013790369033813,
      "learning_rate": 0.000185398604930105,
      "loss": 1.8826,
      "step": 143948
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.0540821552276611,
      "learning_rate": 0.00018539482460518044,
      "loss": 2.1099,
      "step": 143949
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.1669889688491821,
      "learning_rate": 0.0001853910443015631,
      "loss": 2.1594,
      "step": 143950
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.122967004776001,
      "learning_rate": 0.00018538726401925372,
      "loss": 2.2243,
      "step": 143951
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.1527928113937378,
      "learning_rate": 0.00018538348375825303,
      "loss": 1.9624,
      "step": 143952
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.024011492729187,
      "learning_rate": 0.00018537970351856185,
      "loss": 1.9871,
      "step": 143953
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.0618585348129272,
      "learning_rate": 0.00018537592330018063,
      "loss": 2.0519,
      "step": 143954
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.0825690031051636,
      "learning_rate": 0.00018537214310311017,
      "loss": 2.1928,
      "step": 143955
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.0828227996826172,
      "learning_rate": 0.00018536836292735117,
      "loss": 2.0086,
      "step": 143956
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.0987087488174438,
      "learning_rate": 0.00018536458277290438,
      "loss": 2.267,
      "step": 143957
    },
    {
      "epoch": 1.87,
      "grad_norm": 0.9966015219688416,
      "learning_rate": 0.00018536080263977046,
      "loss": 2.0072,
      "step": 143958
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.2146915197372437,
      "learning_rate": 0.0001853570225279503,
      "loss": 2.3888,
      "step": 143959
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.0433980226516724,
      "learning_rate": 0.00018535324243744425,
      "loss": 2.2009,
      "step": 143960
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.0545566082000732,
      "learning_rate": 0.00018534946236825324,
      "loss": 2.069,
      "step": 143961
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.1823688745498657,
      "learning_rate": 0.00018534568232037791,
      "loss": 2.1003,
      "step": 143962
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.0535796880722046,
      "learning_rate": 0.00018534190229381898,
      "loss": 2.084,
      "step": 143963
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.0168405771255493,
      "learning_rate": 0.00018533812228857717,
      "loss": 1.808,
      "step": 143964
    },
    {
      "epoch": 1.87,
      "grad_norm": 0.9898422956466675,
      "learning_rate": 0.00018533434230465326,
      "loss": 2.0284,
      "step": 143965
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.174702525138855,
      "learning_rate": 0.00018533056234204778,
      "loss": 2.2848,
      "step": 143966
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.0824916362762451,
      "learning_rate": 0.0001853267824007615,
      "loss": 2.1269,
      "step": 143967
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.1723130941390991,
      "learning_rate": 0.00018532300248079512,
      "loss": 2.1203,
      "step": 143968
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.0320799350738525,
      "learning_rate": 0.00018531922258214936,
      "loss": 2.1005,
      "step": 143969
    },
    {
      "epoch": 1.87,
      "grad_norm": 0.940762996673584,
      "learning_rate": 0.00018531544270482492,
      "loss": 1.9686,
      "step": 143970
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.0753635168075562,
      "learning_rate": 0.00018531166284882267,
      "loss": 2.0361,
      "step": 143971
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.0631954669952393,
      "learning_rate": 0.00018530788301414293,
      "loss": 2.3498,
      "step": 143972
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.0427688360214233,
      "learning_rate": 0.00018530410320078672,
      "loss": 2.1284,
      "step": 143973
    },
    {
      "epoch": 1.87,
      "grad_norm": 0.9990182518959045,
      "learning_rate": 0.00018530032340875462,
      "loss": 1.7218,
      "step": 143974
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.25150465965271,
      "learning_rate": 0.0001852965436380473,
      "loss": 1.9922,
      "step": 143975
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.0080901384353638,
      "learning_rate": 0.00018529276388866555,
      "loss": 1.8776,
      "step": 143976
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.013603687286377,
      "learning_rate": 0.00018528898416061008,
      "loss": 1.9999,
      "step": 143977
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.0452579259872437,
      "learning_rate": 0.00018528520445388163,
      "loss": 2.141,
      "step": 143978
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.0658608675003052,
      "learning_rate": 0.0001852814247684807,
      "loss": 2.1296,
      "step": 143979
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.042840600013733,
      "learning_rate": 0.0001852776451044081,
      "loss": 2.1237,
      "step": 143980
    },
    {
      "epoch": 1.87,
      "grad_norm": 0.9598475694656372,
      "learning_rate": 0.0001852738654616646,
      "loss": 2.218,
      "step": 143981
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.0544953346252441,
      "learning_rate": 0.00018527008584025083,
      "loss": 2.0983,
      "step": 143982
    },
    {
      "epoch": 1.87,
      "grad_norm": 0.9748329520225525,
      "learning_rate": 0.0001852663062401676,
      "loss": 2.0231,
      "step": 143983
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.1397587060928345,
      "learning_rate": 0.00018526252666141546,
      "loss": 2.0863,
      "step": 143984
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.0527493953704834,
      "learning_rate": 0.00018525874710399523,
      "loss": 2.1177,
      "step": 143985
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.1618787050247192,
      "learning_rate": 0.0001852549675679075,
      "loss": 2.0659,
      "step": 143986
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.295924186706543,
      "learning_rate": 0.00018525118805315307,
      "loss": 1.9169,
      "step": 143987
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.0543861389160156,
      "learning_rate": 0.00018524740855973255,
      "loss": 1.9869,
      "step": 143988
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.176293969154358,
      "learning_rate": 0.00018524362908764683,
      "loss": 2.0819,
      "step": 143989
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.2875306606292725,
      "learning_rate": 0.0001852398496368964,
      "loss": 1.9453,
      "step": 143990
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.0674707889556885,
      "learning_rate": 0.00018523607020748203,
      "loss": 2.0044,
      "step": 143991
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.1055619716644287,
      "learning_rate": 0.00018523229079940446,
      "loss": 2.3971,
      "step": 143992
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.1547421216964722,
      "learning_rate": 0.00018522851141266445,
      "loss": 2.039,
      "step": 143993
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.0899245738983154,
      "learning_rate": 0.0001852247320472625,
      "loss": 1.8874,
      "step": 143994
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.0458518266677856,
      "learning_rate": 0.0001852209527031995,
      "loss": 2.1567,
      "step": 143995
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.0719467401504517,
      "learning_rate": 0.00018521717338047613,
      "loss": 2.0755,
      "step": 143996
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.1786521673202515,
      "learning_rate": 0.00018521339407909302,
      "loss": 2.2076,
      "step": 143997
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.304970622062683,
      "learning_rate": 0.00018520961479905087,
      "loss": 1.8711,
      "step": 143998
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.0734411478042603,
      "learning_rate": 0.00018520583554035045,
      "loss": 2.0845,
      "step": 143999
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.1858978271484375,
      "learning_rate": 0.00018520205630299244,
      "loss": 2.4445,
      "step": 144000
    },
    {
      "epoch": 1.87,
      "grad_norm": 1.0348867177963257,
      "learning_rate": 0.00018519827708697752,
      "loss": 1.8976,
      "step": 144001
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.1204439401626587,
      "learning_rate": 0.00018519449789230645,
      "loss": 2.1872,
      "step": 144002
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.1078637838363647,
      "learning_rate": 0.00018519071871897985,
      "loss": 1.9176,
      "step": 144003
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.1369516849517822,
      "learning_rate": 0.00018518693956699848,
      "loss": 1.8405,
      "step": 144004
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.0692094564437866,
      "learning_rate": 0.00018518316043636299,
      "loss": 2.0841,
      "step": 144005
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.108892798423767,
      "learning_rate": 0.0001851793813270741,
      "loss": 2.2483,
      "step": 144006
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.0636961460113525,
      "learning_rate": 0.00018517560223913263,
      "loss": 1.8967,
      "step": 144007
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.0550410747528076,
      "learning_rate": 0.00018517182317253917,
      "loss": 2.2288,
      "step": 144008
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.025233268737793,
      "learning_rate": 0.0001851680441272944,
      "loss": 1.9433,
      "step": 144009
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.1504369974136353,
      "learning_rate": 0.000185164265103399,
      "loss": 2.2214,
      "step": 144010
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.2015186548233032,
      "learning_rate": 0.00018516048610085374,
      "loss": 2.0991,
      "step": 144011
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.0283833742141724,
      "learning_rate": 0.00018515670711965936,
      "loss": 2.0525,
      "step": 144012
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.093665361404419,
      "learning_rate": 0.0001851529281598165,
      "loss": 2.0489,
      "step": 144013
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.0843708515167236,
      "learning_rate": 0.000185149149221326,
      "loss": 2.1771,
      "step": 144014
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.1928685903549194,
      "learning_rate": 0.00018514537030418833,
      "loss": 2.0632,
      "step": 144015
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.063708782196045,
      "learning_rate": 0.00018514159140840423,
      "loss": 2.0217,
      "step": 144016
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.1100822687149048,
      "learning_rate": 0.00018513781253397455,
      "loss": 1.9145,
      "step": 144017
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.2071592807769775,
      "learning_rate": 0.0001851340336808999,
      "loss": 2.0113,
      "step": 144018
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.3071492910385132,
      "learning_rate": 0.000185130254849181,
      "loss": 2.3504,
      "step": 144019
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.4070007801055908,
      "learning_rate": 0.0001851264760388187,
      "loss": 1.8262,
      "step": 144020
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.1845080852508545,
      "learning_rate": 0.0001851226972498134,
      "loss": 2.1884,
      "step": 144021
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.2908624410629272,
      "learning_rate": 0.00018511891848216596,
      "loss": 2.0467,
      "step": 144022
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.3485859632492065,
      "learning_rate": 0.00018511513973587708,
      "loss": 2.0092,
      "step": 144023
    },
    {
      "epoch": 1.88,
      "grad_norm": 0.9598994851112366,
      "learning_rate": 0.00018511136101094752,
      "loss": 1.9029,
      "step": 144024
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.0598299503326416,
      "learning_rate": 0.00018510758230737788,
      "loss": 1.9739,
      "step": 144025
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.1797559261322021,
      "learning_rate": 0.000185103803625169,
      "loss": 2.0792,
      "step": 144026
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.5717757940292358,
      "learning_rate": 0.00018510002496432144,
      "loss": 2.2236,
      "step": 144027
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.237241506576538,
      "learning_rate": 0.00018509624632483587,
      "loss": 2.0034,
      "step": 144028
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.1565015316009521,
      "learning_rate": 0.00018509246770671316,
      "loss": 2.0536,
      "step": 144029
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.141282081604004,
      "learning_rate": 0.00018508868910995391,
      "loss": 2.1349,
      "step": 144030
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.3676503896713257,
      "learning_rate": 0.00018508491053455882,
      "loss": 2.2205,
      "step": 144031
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.0196516513824463,
      "learning_rate": 0.00018508113198052877,
      "loss": 1.8943,
      "step": 144032
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.1366467475891113,
      "learning_rate": 0.00018507735344786415,
      "loss": 2.157,
      "step": 144033
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.016965627670288,
      "learning_rate": 0.00018507357493656584,
      "loss": 2.0704,
      "step": 144034
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.0969598293304443,
      "learning_rate": 0.0001850697964466345,
      "loss": 1.923,
      "step": 144035
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.002205729484558,
      "learning_rate": 0.00018506601797807086,
      "loss": 2.2074,
      "step": 144036
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.0895113945007324,
      "learning_rate": 0.0001850622395308756,
      "loss": 1.8855,
      "step": 144037
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.1705242395401,
      "learning_rate": 0.00018505846110504959,
      "loss": 2.122,
      "step": 144038
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.201746940612793,
      "learning_rate": 0.00018505468270059325,
      "loss": 2.2374,
      "step": 144039
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.3006904125213623,
      "learning_rate": 0.0001850509043175074,
      "loss": 2.1104,
      "step": 144040
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.1567445993423462,
      "learning_rate": 0.00018504712595579276,
      "loss": 2.2639,
      "step": 144041
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.0939857959747314,
      "learning_rate": 0.00018504334761545007,
      "loss": 2.2127,
      "step": 144042
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.2725021839141846,
      "learning_rate": 0.00018503956929647993,
      "loss": 2.0073,
      "step": 144043
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.104187250137329,
      "learning_rate": 0.0001850357909988833,
      "loss": 2.015,
      "step": 144044
    },
    {
      "epoch": 1.88,
      "grad_norm": 0.9890622496604919,
      "learning_rate": 0.00018503201272266053,
      "loss": 2.0163,
      "step": 144045
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.1650574207305908,
      "learning_rate": 0.00018502823446781247,
      "loss": 2.0813,
      "step": 144046
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.0378700494766235,
      "learning_rate": 0.00018502445623433983,
      "loss": 1.8562,
      "step": 144047
    },
    {
      "epoch": 1.88,
      "grad_norm": 0.9587523341178894,
      "learning_rate": 0.00018502067802224332,
      "loss": 1.9728,
      "step": 144048
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.0443190336227417,
      "learning_rate": 0.00018501689983152362,
      "loss": 2.0019,
      "step": 144049
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.051957368850708,
      "learning_rate": 0.00018501312166218162,
      "loss": 2.1615,
      "step": 144050
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.1562228202819824,
      "learning_rate": 0.00018500934351421764,
      "loss": 2.2032,
      "step": 144051
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.1092500686645508,
      "learning_rate": 0.00018500556538763268,
      "loss": 2.2159,
      "step": 144052
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.1338857412338257,
      "learning_rate": 0.0001850017872824273,
      "loss": 1.961,
      "step": 144053
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.0228221416473389,
      "learning_rate": 0.0001849980091986023,
      "loss": 2.1022,
      "step": 144054
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.0632343292236328,
      "learning_rate": 0.0001849942311361583,
      "loss": 2.0348,
      "step": 144055
    },
    {
      "epoch": 1.88,
      "grad_norm": 0.9693847894668579,
      "learning_rate": 0.00018499045309509625,
      "loss": 2.3058,
      "step": 144056
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.0759468078613281,
      "learning_rate": 0.00018498667507541643,
      "loss": 2.0207,
      "step": 144057
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.0424333810806274,
      "learning_rate": 0.0001849828970771198,
      "loss": 1.837,
      "step": 144058
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.3843191862106323,
      "learning_rate": 0.000184979119100207,
      "loss": 2.0829,
      "step": 144059
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.3491615056991577,
      "learning_rate": 0.00018497534114467876,
      "loss": 2.1529,
      "step": 144060
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.088827133178711,
      "learning_rate": 0.00018497156321053577,
      "loss": 2.0329,
      "step": 144061
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.3611397743225098,
      "learning_rate": 0.00018496778529777878,
      "loss": 1.8816,
      "step": 144062
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.0312373638153076,
      "learning_rate": 0.00018496400740640852,
      "loss": 2.2358,
      "step": 144063
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.0270187854766846,
      "learning_rate": 0.00018496022953642553,
      "loss": 1.786,
      "step": 144064
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.001224160194397,
      "learning_rate": 0.00018495645168783057,
      "loss": 1.995,
      "step": 144065
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.1200649738311768,
      "learning_rate": 0.00018495267386062445,
      "loss": 1.9551,
      "step": 144066
    },
    {
      "epoch": 1.88,
      "grad_norm": 0.948582649230957,
      "learning_rate": 0.00018494889605480774,
      "loss": 2.1036,
      "step": 144067
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.043452262878418,
      "learning_rate": 0.0001849451182703812,
      "loss": 1.9051,
      "step": 144068
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.1550533771514893,
      "learning_rate": 0.0001849413405073456,
      "loss": 2.2163,
      "step": 144069
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.1005955934524536,
      "learning_rate": 0.00018493756276570161,
      "loss": 1.9046,
      "step": 144070
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.0419559478759766,
      "learning_rate": 0.0001849337850454498,
      "loss": 2.016,
      "step": 144071
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.0552685260772705,
      "learning_rate": 0.000184930007346591,
      "loss": 2.1282,
      "step": 144072
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.0858519077301025,
      "learning_rate": 0.00018492622966912586,
      "loss": 2.1203,
      "step": 144073
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.1007479429244995,
      "learning_rate": 0.00018492245201305512,
      "loss": 2.179,
      "step": 144074
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.090856671333313,
      "learning_rate": 0.00018491867437837954,
      "loss": 2.0347,
      "step": 144075
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.1034634113311768,
      "learning_rate": 0.00018491489676509968,
      "loss": 2.1432,
      "step": 144076
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.0996489524841309,
      "learning_rate": 0.00018491111917321633,
      "loss": 1.9364,
      "step": 144077
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.1212608814239502,
      "learning_rate": 0.00018490734160273015,
      "loss": 2.2524,
      "step": 144078
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.1228418350219727,
      "learning_rate": 0.0001849035640536419,
      "loss": 2.2079,
      "step": 144079
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.1271412372589111,
      "learning_rate": 0.0001848997865259522,
      "loss": 2.0539,
      "step": 144080
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.1167248487472534,
      "learning_rate": 0.0001848960090196619,
      "loss": 1.9934,
      "step": 144081
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.2042961120605469,
      "learning_rate": 0.0001848922315347715,
      "loss": 1.9657,
      "step": 144082
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.1213017702102661,
      "learning_rate": 0.00018488845407128185,
      "loss": 2.0751,
      "step": 144083
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.051428198814392,
      "learning_rate": 0.00018488467662919358,
      "loss": 2.1784,
      "step": 144084
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.0487217903137207,
      "learning_rate": 0.00018488089920850752,
      "loss": 1.9764,
      "step": 144085
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.1142816543579102,
      "learning_rate": 0.00018487712180922415,
      "loss": 2.1512,
      "step": 144086
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.1142301559448242,
      "learning_rate": 0.0001848733444313444,
      "loss": 1.9894,
      "step": 144087
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.0137341022491455,
      "learning_rate": 0.00018486956707486883,
      "loss": 1.9644,
      "step": 144088
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.1956727504730225,
      "learning_rate": 0.00018486578973979817,
      "loss": 2.2225,
      "step": 144089
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.0554381608963013,
      "learning_rate": 0.0001848620124261331,
      "loss": 2.301,
      "step": 144090
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.05950129032135,
      "learning_rate": 0.00018485823513387438,
      "loss": 1.8746,
      "step": 144091
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.2319945096969604,
      "learning_rate": 0.00018485445786302273,
      "loss": 1.9799,
      "step": 144092
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.1251705884933472,
      "learning_rate": 0.00018485068061357884,
      "loss": 2.0722,
      "step": 144093
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.1498966217041016,
      "learning_rate": 0.00018484690338554328,
      "loss": 2.051,
      "step": 144094
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.0824023485183716,
      "learning_rate": 0.0001848431261789169,
      "loss": 2.0859,
      "step": 144095
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.0649460554122925,
      "learning_rate": 0.00018483934899370035,
      "loss": 1.9343,
      "step": 144096
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.126993179321289,
      "learning_rate": 0.00018483557182989428,
      "loss": 1.867,
      "step": 144097
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.1279058456420898,
      "learning_rate": 0.00018483179468749953,
      "loss": 2.2567,
      "step": 144098
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.0487704277038574,
      "learning_rate": 0.00018482801756651676,
      "loss": 2.1842,
      "step": 144099
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.1039838790893555,
      "learning_rate": 0.00018482424046694658,
      "loss": 2.1808,
      "step": 144100
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.0051583051681519,
      "learning_rate": 0.00018482046338878975,
      "loss": 2.0903,
      "step": 144101
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.1043201684951782,
      "learning_rate": 0.00018481668633204694,
      "loss": 2.1275,
      "step": 144102
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.0536775588989258,
      "learning_rate": 0.00018481290929671889,
      "loss": 2.0868,
      "step": 144103
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.1176332235336304,
      "learning_rate": 0.00018480913228280632,
      "loss": 2.2026,
      "step": 144104
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.3948564529418945,
      "learning_rate": 0.00018480535529031002,
      "loss": 2.2686,
      "step": 144105
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.1096794605255127,
      "learning_rate": 0.00018480157831923043,
      "loss": 2.033,
      "step": 144106
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.116184949874878,
      "learning_rate": 0.0001847978013695684,
      "loss": 2.0798,
      "step": 144107
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.098917841911316,
      "learning_rate": 0.00018479402444132468,
      "loss": 2.0286,
      "step": 144108
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.0601317882537842,
      "learning_rate": 0.00018479024753449984,
      "loss": 2.2255,
      "step": 144109
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.1335731744766235,
      "learning_rate": 0.00018478647064909475,
      "loss": 2.1534,
      "step": 144110
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.0662914514541626,
      "learning_rate": 0.00018478269378511013,
      "loss": 2.2439,
      "step": 144111
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.0184322595596313,
      "learning_rate": 0.00018477891694254644,
      "loss": 2.1461,
      "step": 144112
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.4906513690948486,
      "learning_rate": 0.00018477514012140453,
      "loss": 2.2131,
      "step": 144113
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.0502514839172363,
      "learning_rate": 0.0001847713633216851,
      "loss": 1.9899,
      "step": 144114
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.103055715560913,
      "learning_rate": 0.00018476758654338885,
      "loss": 2.033,
      "step": 144115
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.1659789085388184,
      "learning_rate": 0.0001847638097865165,
      "loss": 2.0195,
      "step": 144116
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.2929998636245728,
      "learning_rate": 0.00018476003305106884,
      "loss": 1.8481,
      "step": 144117
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.1889002323150635,
      "learning_rate": 0.00018475625633704632,
      "loss": 2.0004,
      "step": 144118
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.1806293725967407,
      "learning_rate": 0.00018475247964444983,
      "loss": 2.2092,
      "step": 144119
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.0739692449569702,
      "learning_rate": 0.00018474870297328,
      "loss": 2.1017,
      "step": 144120
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.1141571998596191,
      "learning_rate": 0.0001847449263235376,
      "loss": 1.9748,
      "step": 144121
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.1037906408309937,
      "learning_rate": 0.00018474114969522324,
      "loss": 2.1806,
      "step": 144122
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.0517020225524902,
      "learning_rate": 0.00018473737308833782,
      "loss": 2.1068,
      "step": 144123
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.0437699556350708,
      "learning_rate": 0.0001847335965028818,
      "loss": 2.1658,
      "step": 144124
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.0266433954238892,
      "learning_rate": 0.00018472981993885592,
      "loss": 2.1698,
      "step": 144125
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.0854655504226685,
      "learning_rate": 0.00018472604339626097,
      "loss": 1.8771,
      "step": 144126
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.093057632446289,
      "learning_rate": 0.00018472226687509766,
      "loss": 2.1482,
      "step": 144127
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.1700246334075928,
      "learning_rate": 0.00018471849037536662,
      "loss": 2.042,
      "step": 144128
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.1100246906280518,
      "learning_rate": 0.00018471471389706862,
      "loss": 1.977,
      "step": 144129
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.1974949836730957,
      "learning_rate": 0.0001847109374402044,
      "loss": 1.9771,
      "step": 144130
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.1010472774505615,
      "learning_rate": 0.00018470716100477453,
      "loss": 2.0462,
      "step": 144131
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.072076439857483,
      "learning_rate": 0.00018470338459077972,
      "loss": 2.1239,
      "step": 144132
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.0464649200439453,
      "learning_rate": 0.00018469960819822082,
      "loss": 1.9874,
      "step": 144133
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.0670021772384644,
      "learning_rate": 0.00018469583182709832,
      "loss": 1.9808,
      "step": 144134
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.1156013011932373,
      "learning_rate": 0.0001846920554774131,
      "loss": 1.9996,
      "step": 144135
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.0430805683135986,
      "learning_rate": 0.00018468827914916593,
      "loss": 2.0144,
      "step": 144136
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.0890766382217407,
      "learning_rate": 0.00018468450284235728,
      "loss": 2.1848,
      "step": 144137
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.165328025817871,
      "learning_rate": 0.00018468072655698794,
      "loss": 2.0979,
      "step": 144138
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.34393310546875,
      "learning_rate": 0.0001846769502930586,
      "loss": 1.9956,
      "step": 144139
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.1370866298675537,
      "learning_rate": 0.00018467317405057002,
      "loss": 2.282,
      "step": 144140
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.0874921083450317,
      "learning_rate": 0.00018466939782952287,
      "loss": 2.0202,
      "step": 144141
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.061310887336731,
      "learning_rate": 0.00018466562162991796,
      "loss": 1.9799,
      "step": 144142
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.2186882495880127,
      "learning_rate": 0.00018466184545175578,
      "loss": 2.2328,
      "step": 144143
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.2543511390686035,
      "learning_rate": 0.00018465806929503716,
      "loss": 2.0654,
      "step": 144144
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.1342673301696777,
      "learning_rate": 0.00018465429315976277,
      "loss": 2.1214,
      "step": 144145
    },
    {
      "epoch": 1.88,
      "grad_norm": 0.9796308279037476,
      "learning_rate": 0.0001846505170459333,
      "loss": 1.8727,
      "step": 144146
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.2413617372512817,
      "learning_rate": 0.00018464674095354953,
      "loss": 2.1551,
      "step": 144147
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.2043371200561523,
      "learning_rate": 0.00018464296488261218,
      "loss": 2.0784,
      "step": 144148
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.1598255634307861,
      "learning_rate": 0.00018463918883312175,
      "loss": 2.2301,
      "step": 144149
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.2102022171020508,
      "learning_rate": 0.00018463541280507912,
      "loss": 2.2006,
      "step": 144150
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.2403303384780884,
      "learning_rate": 0.00018463163679848492,
      "loss": 2.07,
      "step": 144151
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.0453932285308838,
      "learning_rate": 0.00018462786081333989,
      "loss": 2.183,
      "step": 144152
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.2030764818191528,
      "learning_rate": 0.0001846240848496447,
      "loss": 1.9298,
      "step": 144153
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.1493498086929321,
      "learning_rate": 0.00018462030890740012,
      "loss": 2.2561,
      "step": 144154
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.0670616626739502,
      "learning_rate": 0.00018461653298660683,
      "loss": 1.9818,
      "step": 144155
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.0806853771209717,
      "learning_rate": 0.00018461275708726542,
      "loss": 1.988,
      "step": 144156
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.3402537107467651,
      "learning_rate": 0.0001846089812093767,
      "loss": 1.8962,
      "step": 144157
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.1719932556152344,
      "learning_rate": 0.00018460520535294132,
      "loss": 1.9675,
      "step": 144158
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.0966885089874268,
      "learning_rate": 0.00018460142951796005,
      "loss": 2.0264,
      "step": 144159
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.0868680477142334,
      "learning_rate": 0.0001845976537044336,
      "loss": 1.9901,
      "step": 144160
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.0357835292816162,
      "learning_rate": 0.00018459387791236253,
      "loss": 1.9898,
      "step": 144161
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.0877933502197266,
      "learning_rate": 0.00018459010214174777,
      "loss": 2.1052,
      "step": 144162
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.3658353090286255,
      "learning_rate": 0.00018458632639258978,
      "loss": 2.2221,
      "step": 144163
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.0112076997756958,
      "learning_rate": 0.00018458255066488934,
      "loss": 2.161,
      "step": 144164
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.0103834867477417,
      "learning_rate": 0.00018457877495864721,
      "loss": 2.0186,
      "step": 144165
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.0586750507354736,
      "learning_rate": 0.00018457499927386416,
      "loss": 2.1395,
      "step": 144166
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.1324408054351807,
      "learning_rate": 0.0001845712236105407,
      "loss": 1.9953,
      "step": 144167
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.036070704460144,
      "learning_rate": 0.00018456744796867765,
      "loss": 2.1893,
      "step": 144168
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.049455165863037,
      "learning_rate": 0.00018456367234827572,
      "loss": 1.9755,
      "step": 144169
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.272741436958313,
      "learning_rate": 0.00018455989674933553,
      "loss": 2.0024,
      "step": 144170
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.0120148658752441,
      "learning_rate": 0.00018455612117185784,
      "loss": 2.1925,
      "step": 144171
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.0902769565582275,
      "learning_rate": 0.0001845523456158434,
      "loss": 2.0096,
      "step": 144172
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.43601655960083,
      "learning_rate": 0.0001845485700812928,
      "loss": 2.0198,
      "step": 144173
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.1221460103988647,
      "learning_rate": 0.00018454479456820683,
      "loss": 2.483,
      "step": 144174
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.1542644500732422,
      "learning_rate": 0.00018454101907658613,
      "loss": 2.0879,
      "step": 144175
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.0966356992721558,
      "learning_rate": 0.00018453724360643144,
      "loss": 2.2236,
      "step": 144176
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.095798373222351,
      "learning_rate": 0.0001845334681577435,
      "loss": 1.9622,
      "step": 144177
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.2540924549102783,
      "learning_rate": 0.00018452969273052302,
      "loss": 1.8257,
      "step": 144178
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.101162075996399,
      "learning_rate": 0.00018452591732477056,
      "loss": 2.1309,
      "step": 144179
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.2295485734939575,
      "learning_rate": 0.00018452214194048688,
      "loss": 2.0274,
      "step": 144180
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.0092121362686157,
      "learning_rate": 0.00018451836657767276,
      "loss": 2.0585,
      "step": 144181
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.0770187377929688,
      "learning_rate": 0.00018451459123632885,
      "loss": 1.9765,
      "step": 144182
    },
    {
      "epoch": 1.88,
      "grad_norm": 2.417955160140991,
      "learning_rate": 0.00018451081591645585,
      "loss": 2.149,
      "step": 144183
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.0900373458862305,
      "learning_rate": 0.00018450704061805453,
      "loss": 2.2643,
      "step": 144184
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.0826748609542847,
      "learning_rate": 0.00018450326534112546,
      "loss": 1.9262,
      "step": 144185
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.039095401763916,
      "learning_rate": 0.00018449949008566944,
      "loss": 2.0727,
      "step": 144186
    },
    {
      "epoch": 1.88,
      "grad_norm": 0.9973351359367371,
      "learning_rate": 0.00018449571485168714,
      "loss": 2.1223,
      "step": 144187
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.052017092704773,
      "learning_rate": 0.00018449193963917924,
      "loss": 1.9292,
      "step": 144188
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.0260748863220215,
      "learning_rate": 0.00018448816444814647,
      "loss": 2.0689,
      "step": 144189
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.1311250925064087,
      "learning_rate": 0.00018448438927858967,
      "loss": 2.0037,
      "step": 144190
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.0626438856124878,
      "learning_rate": 0.0001844806141305093,
      "loss": 1.9109,
      "step": 144191
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.217400312423706,
      "learning_rate": 0.0001844768390039061,
      "loss": 2.0883,
      "step": 144192
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.123499870300293,
      "learning_rate": 0.00018447306389878087,
      "loss": 2.0111,
      "step": 144193
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.1665382385253906,
      "learning_rate": 0.00018446928881513428,
      "loss": 2.2783,
      "step": 144194
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.0066368579864502,
      "learning_rate": 0.00018446551375296707,
      "loss": 2.1217,
      "step": 144195
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.1343591213226318,
      "learning_rate": 0.00018446173871227984,
      "loss": 2.304,
      "step": 144196
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.075501799583435,
      "learning_rate": 0.00018445796369307351,
      "loss": 1.8738,
      "step": 144197
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.1520755290985107,
      "learning_rate": 0.00018445418869534848,
      "loss": 2.1412,
      "step": 144198
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.0732909440994263,
      "learning_rate": 0.00018445041371910562,
      "loss": 2.3989,
      "step": 144199
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.0954257249832153,
      "learning_rate": 0.00018444663876434556,
      "loss": 2.2337,
      "step": 144200
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.0934398174285889,
      "learning_rate": 0.00018444286383106911,
      "loss": 2.1523,
      "step": 144201
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.1414883136749268,
      "learning_rate": 0.0001844390889192769,
      "loss": 2.3062,
      "step": 144202
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.0979710817337036,
      "learning_rate": 0.00018443531402896976,
      "loss": 2.245,
      "step": 144203
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.0261452198028564,
      "learning_rate": 0.00018443153916014812,
      "loss": 2.0195,
      "step": 144204
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.175784707069397,
      "learning_rate": 0.00018442776431281288,
      "loss": 2.0946,
      "step": 144205
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.1442656517028809,
      "learning_rate": 0.00018442398948696471,
      "loss": 1.93,
      "step": 144206
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.068036437034607,
      "learning_rate": 0.0001844202146826043,
      "loss": 2.1644,
      "step": 144207
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.0676511526107788,
      "learning_rate": 0.00018441643989973233,
      "loss": 1.9244,
      "step": 144208
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.1476585865020752,
      "learning_rate": 0.00018441266513834968,
      "loss": 2.1751,
      "step": 144209
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.0898345708847046,
      "learning_rate": 0.00018440889039845673,
      "loss": 2.0093,
      "step": 144210
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.1674890518188477,
      "learning_rate": 0.00018440511568005436,
      "loss": 2.0813,
      "step": 144211
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.1299082040786743,
      "learning_rate": 0.00018440134098314325,
      "loss": 2.0684,
      "step": 144212
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.0938469171524048,
      "learning_rate": 0.00018439756630772415,
      "loss": 2.05,
      "step": 144213
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.3172411918640137,
      "learning_rate": 0.00018439379165379773,
      "loss": 2.0973,
      "step": 144214
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.144842267036438,
      "learning_rate": 0.00018439001702136477,
      "loss": 2.0353,
      "step": 144215
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.051007866859436,
      "learning_rate": 0.00018438624241042578,
      "loss": 2.1481,
      "step": 144216
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.4176627397537231,
      "learning_rate": 0.00018438246782098155,
      "loss": 2.1607,
      "step": 144217
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.5548254251480103,
      "learning_rate": 0.00018437869325303285,
      "loss": 2.0747,
      "step": 144218
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.1683557033538818,
      "learning_rate": 0.0001843749187065803,
      "loss": 2.0356,
      "step": 144219
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.0789991617202759,
      "learning_rate": 0.00018437114418162463,
      "loss": 2.3192,
      "step": 144220
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.3309334516525269,
      "learning_rate": 0.00018436736967816672,
      "loss": 2.116,
      "step": 144221
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.2598234415054321,
      "learning_rate": 0.00018436359519620693,
      "loss": 2.0089,
      "step": 144222
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.1770075559616089,
      "learning_rate": 0.00018435982073574614,
      "loss": 2.1428,
      "step": 144223
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.0269064903259277,
      "learning_rate": 0.00018435604629678503,
      "loss": 2.0188,
      "step": 144224
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.1083670854568481,
      "learning_rate": 0.00018435227187932432,
      "loss": 1.984,
      "step": 144225
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.2718650102615356,
      "learning_rate": 0.00018434849748336473,
      "loss": 1.9383,
      "step": 144226
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.0511592626571655,
      "learning_rate": 0.00018434472310890703,
      "loss": 2.0695,
      "step": 144227
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.1277974843978882,
      "learning_rate": 0.00018434094875595175,
      "loss": 2.1249,
      "step": 144228
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.052234411239624,
      "learning_rate": 0.00018433717442449963,
      "loss": 2.0992,
      "step": 144229
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.1110810041427612,
      "learning_rate": 0.00018433340011455142,
      "loss": 2.2091,
      "step": 144230
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.1006673574447632,
      "learning_rate": 0.0001843296258261078,
      "loss": 2.3132,
      "step": 144231
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.1541210412979126,
      "learning_rate": 0.0001843258515591695,
      "loss": 2.1512,
      "step": 144232
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.0513237714767456,
      "learning_rate": 0.00018432207731373736,
      "loss": 2.4101,
      "step": 144233
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.065644383430481,
      "learning_rate": 0.00018431830308981176,
      "loss": 2.1263,
      "step": 144234
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.117189884185791,
      "learning_rate": 0.00018431452888739358,
      "loss": 2.0936,
      "step": 144235
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.1574819087982178,
      "learning_rate": 0.0001843107547064835,
      "loss": 2.1599,
      "step": 144236
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.0256778001785278,
      "learning_rate": 0.0001843069805470823,
      "loss": 2.1782,
      "step": 144237
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.2178571224212646,
      "learning_rate": 0.00018430320640919058,
      "loss": 2.0893,
      "step": 144238
    },
    {
      "epoch": 1.88,
      "grad_norm": 0.966141402721405,
      "learning_rate": 0.00018429943229280918,
      "loss": 1.8309,
      "step": 144239
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.0724371671676636,
      "learning_rate": 0.0001842956581979386,
      "loss": 2.0225,
      "step": 144240
    },
    {
      "epoch": 1.88,
      "grad_norm": 0.9196905493736267,
      "learning_rate": 0.00018429188412457962,
      "loss": 2.04,
      "step": 144241
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.1903455257415771,
      "learning_rate": 0.00018428811007273297,
      "loss": 2.1628,
      "step": 144242
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.1986875534057617,
      "learning_rate": 0.00018428433604239934,
      "loss": 2.141,
      "step": 144243
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.144131064414978,
      "learning_rate": 0.00018428056203357947,
      "loss": 2.3511,
      "step": 144244
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.139445185661316,
      "learning_rate": 0.00018427678804627405,
      "loss": 1.7781,
      "step": 144245
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.1650316715240479,
      "learning_rate": 0.00018427301408048373,
      "loss": 2.2272,
      "step": 144246
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.2546206712722778,
      "learning_rate": 0.00018426924013620924,
      "loss": 2.0008,
      "step": 144247
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.0750302076339722,
      "learning_rate": 0.00018426546621345128,
      "loss": 2.1749,
      "step": 144248
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.234636902809143,
      "learning_rate": 0.00018426169231221056,
      "loss": 2.2186,
      "step": 144249
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.0102351903915405,
      "learning_rate": 0.00018425791843248778,
      "loss": 2.0914,
      "step": 144250
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.0551395416259766,
      "learning_rate": 0.00018425414457428366,
      "loss": 1.8797,
      "step": 144251
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.1222254037857056,
      "learning_rate": 0.00018425037073759883,
      "loss": 1.9717,
      "step": 144252
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.0689587593078613,
      "learning_rate": 0.00018424659692243406,
      "loss": 2.1562,
      "step": 144253
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.031545639038086,
      "learning_rate": 0.00018424282312879008,
      "loss": 2.0037,
      "step": 144254
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.1368534564971924,
      "learning_rate": 0.00018423904935666748,
      "loss": 2.2384,
      "step": 144255
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.0406479835510254,
      "learning_rate": 0.000184235275606067,
      "loss": 2.0691,
      "step": 144256
    },
    {
      "epoch": 1.88,
      "grad_norm": 0.975921094417572,
      "learning_rate": 0.0001842315018769895,
      "loss": 1.9783,
      "step": 144257
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.0206294059753418,
      "learning_rate": 0.00018422772816943542,
      "loss": 2.2975,
      "step": 144258
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.20400869846344,
      "learning_rate": 0.0001842239544834056,
      "loss": 2.4578,
      "step": 144259
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.0410988330841064,
      "learning_rate": 0.00018422018081890075,
      "loss": 1.8196,
      "step": 144260
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.0968842506408691,
      "learning_rate": 0.00018421640717592163,
      "loss": 2.2031,
      "step": 144261
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.068010687828064,
      "learning_rate": 0.0001842126335544688,
      "loss": 1.6813,
      "step": 144262
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.0954577922821045,
      "learning_rate": 0.000184208859954543,
      "loss": 2.1792,
      "step": 144263
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.4674413204193115,
      "learning_rate": 0.00018420508637614506,
      "loss": 2.1489,
      "step": 144264
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.0726871490478516,
      "learning_rate": 0.00018420131281927547,
      "loss": 1.9805,
      "step": 144265
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.0318974256515503,
      "learning_rate": 0.00018419753928393507,
      "loss": 1.8691,
      "step": 144266
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.0828524827957153,
      "learning_rate": 0.00018419376577012453,
      "loss": 2.0213,
      "step": 144267
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.1909314393997192,
      "learning_rate": 0.00018418999227784463,
      "loss": 1.9943,
      "step": 144268
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.112593412399292,
      "learning_rate": 0.0001841862188070959,
      "loss": 2.2688,
      "step": 144269
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.1621148586273193,
      "learning_rate": 0.0001841824453578792,
      "loss": 2.1109,
      "step": 144270
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.285841703414917,
      "learning_rate": 0.00018417867193019514,
      "loss": 2.2274,
      "step": 144271
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.1988993883132935,
      "learning_rate": 0.00018417489852404448,
      "loss": 2.2995,
      "step": 144272
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.1335806846618652,
      "learning_rate": 0.00018417112513942785,
      "loss": 2.347,
      "step": 144273
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.2079391479492188,
      "learning_rate": 0.00018416735177634598,
      "loss": 2.1518,
      "step": 144274
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.21942138671875,
      "learning_rate": 0.0001841635784347996,
      "loss": 2.2245,
      "step": 144275
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.191249966621399,
      "learning_rate": 0.00018415980511478952,
      "loss": 2.2139,
      "step": 144276
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.0826979875564575,
      "learning_rate": 0.00018415603181631617,
      "loss": 2.0121,
      "step": 144277
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.0484598875045776,
      "learning_rate": 0.00018415225853938048,
      "loss": 1.9672,
      "step": 144278
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.328263759613037,
      "learning_rate": 0.000184148485283983,
      "loss": 2.0253,
      "step": 144279
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.3246469497680664,
      "learning_rate": 0.0001841447120501245,
      "loss": 2.1016,
      "step": 144280
    },
    {
      "epoch": 1.88,
      "grad_norm": 0.9978160262107849,
      "learning_rate": 0.0001841409388378057,
      "loss": 2.2273,
      "step": 144281
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.2803939580917358,
      "learning_rate": 0.00018413716564702744,
      "loss": 2.1115,
      "step": 144282
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.2008426189422607,
      "learning_rate": 0.0001841333924777901,
      "loss": 1.885,
      "step": 144283
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.186785101890564,
      "learning_rate": 0.0001841296193300946,
      "loss": 1.9011,
      "step": 144284
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.0464603900909424,
      "learning_rate": 0.00018412584620394157,
      "loss": 2.1515,
      "step": 144285
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.072208046913147,
      "learning_rate": 0.00018412207309933173,
      "loss": 1.8921,
      "step": 144286
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.0850540399551392,
      "learning_rate": 0.00018411830001626578,
      "loss": 2.1348,
      "step": 144287
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.4222313165664673,
      "learning_rate": 0.00018411452695474455,
      "loss": 2.0813,
      "step": 144288
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.1367288827896118,
      "learning_rate": 0.0001841107539147685,
      "loss": 2.0524,
      "step": 144289
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.5516846179962158,
      "learning_rate": 0.00018410698089633843,
      "loss": 2.129,
      "step": 144290
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.1817941665649414,
      "learning_rate": 0.0001841032078994551,
      "loss": 1.8448,
      "step": 144291
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.2275978326797485,
      "learning_rate": 0.00018409943492411914,
      "loss": 2.0917,
      "step": 144292
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.0894169807434082,
      "learning_rate": 0.00018409566197033128,
      "loss": 2.0762,
      "step": 144293
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.3222973346710205,
      "learning_rate": 0.0001840918890380923,
      "loss": 2.1933,
      "step": 144294
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.0748153924942017,
      "learning_rate": 0.00018408811612740278,
      "loss": 2.0994,
      "step": 144295
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.2225368022918701,
      "learning_rate": 0.00018408434323826346,
      "loss": 1.7371,
      "step": 144296
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.048948049545288,
      "learning_rate": 0.00018408057037067502,
      "loss": 2.0691,
      "step": 144297
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.0946956872940063,
      "learning_rate": 0.00018407679752463819,
      "loss": 1.9429,
      "step": 144298
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.1333450078964233,
      "learning_rate": 0.00018407302470015372,
      "loss": 2.0162,
      "step": 144299
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.1583325862884521,
      "learning_rate": 0.00018406925189722235,
      "loss": 2.1428,
      "step": 144300
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.2316306829452515,
      "learning_rate": 0.00018406547911584454,
      "loss": 2.1666,
      "step": 144301
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.1118476390838623,
      "learning_rate": 0.00018406170635602118,
      "loss": 2.0326,
      "step": 144302
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.010481357574463,
      "learning_rate": 0.0001840579336177529,
      "loss": 2.0296,
      "step": 144303
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.112100601196289,
      "learning_rate": 0.00018405416090104052,
      "loss": 2.0534,
      "step": 144304
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.283323049545288,
      "learning_rate": 0.0001840503882058846,
      "loss": 2.2375,
      "step": 144305
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.1632570028305054,
      "learning_rate": 0.00018404661553228605,
      "loss": 2.2761,
      "step": 144306
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.0238471031188965,
      "learning_rate": 0.00018404284288024527,
      "loss": 2.0785,
      "step": 144307
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.1074448823928833,
      "learning_rate": 0.00018403907024976316,
      "loss": 2.2087,
      "step": 144308
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.1395460367202759,
      "learning_rate": 0.00018403529764084031,
      "loss": 2.3697,
      "step": 144309
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.2021710872650146,
      "learning_rate": 0.00018403152505347753,
      "loss": 1.9867,
      "step": 144310
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.0720367431640625,
      "learning_rate": 0.0001840277524876755,
      "loss": 2.2868,
      "step": 144311
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.1471503973007202,
      "learning_rate": 0.00018402397994343497,
      "loss": 2.1952,
      "step": 144312
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.2807724475860596,
      "learning_rate": 0.00018402020742075648,
      "loss": 2.0263,
      "step": 144313
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.3148152828216553,
      "learning_rate": 0.00018401643491964084,
      "loss": 2.2655,
      "step": 144314
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.346948504447937,
      "learning_rate": 0.0001840126624400887,
      "loss": 2.2855,
      "step": 144315
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.0470632314682007,
      "learning_rate": 0.0001840088899821008,
      "loss": 2.0129,
      "step": 144316
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.080672264099121,
      "learning_rate": 0.00018400511754567786,
      "loss": 1.9835,
      "step": 144317
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.078575611114502,
      "learning_rate": 0.00018400134513082064,
      "loss": 1.9976,
      "step": 144318
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.185303807258606,
      "learning_rate": 0.0001839975727375297,
      "loss": 2.2136,
      "step": 144319
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.13068687915802,
      "learning_rate": 0.00018399380036580575,
      "loss": 2.0631,
      "step": 144320
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.0146386623382568,
      "learning_rate": 0.00018399002801564954,
      "loss": 2.0236,
      "step": 144321
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.1307923793792725,
      "learning_rate": 0.0001839862556870618,
      "loss": 2.0872,
      "step": 144322
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.26364004611969,
      "learning_rate": 0.00018398248338004317,
      "loss": 2.0089,
      "step": 144323
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.1232773065567017,
      "learning_rate": 0.00018397871109459456,
      "loss": 2.2249,
      "step": 144324
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.1683584451675415,
      "learning_rate": 0.00018397493883071634,
      "loss": 2.1728,
      "step": 144325
    },
    {
      "epoch": 1.88,
      "grad_norm": 0.9252825379371643,
      "learning_rate": 0.00018397116658840939,
      "loss": 1.8139,
      "step": 144326
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.1887519359588623,
      "learning_rate": 0.00018396739436767434,
      "loss": 2.0918,
      "step": 144327
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.2228306531906128,
      "learning_rate": 0.00018396362216851198,
      "loss": 2.0825,
      "step": 144328
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.0927330255508423,
      "learning_rate": 0.00018395984999092297,
      "loss": 1.9779,
      "step": 144329
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.1106505393981934,
      "learning_rate": 0.00018395607783490805,
      "loss": 2.2003,
      "step": 144330
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.1957416534423828,
      "learning_rate": 0.00018395230570046793,
      "loss": 2.1382,
      "step": 144331
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.1134889125823975,
      "learning_rate": 0.0001839485335876032,
      "loss": 2.2831,
      "step": 144332
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.1299922466278076,
      "learning_rate": 0.0001839447614963146,
      "loss": 1.8707,
      "step": 144333
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.088414192199707,
      "learning_rate": 0.00018394098942660288,
      "loss": 2.0824,
      "step": 144334
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.0028536319732666,
      "learning_rate": 0.00018393721737846872,
      "loss": 2.1194,
      "step": 144335
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.2050715684890747,
      "learning_rate": 0.00018393344535191285,
      "loss": 2.0642,
      "step": 144336
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.0877211093902588,
      "learning_rate": 0.00018392967334693598,
      "loss": 1.9264,
      "step": 144337
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.0962989330291748,
      "learning_rate": 0.0001839259013635387,
      "loss": 2.1071,
      "step": 144338
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.1696974039077759,
      "learning_rate": 0.00018392212940172186,
      "loss": 2.2583,
      "step": 144339
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.0567890405654907,
      "learning_rate": 0.000183918357461486,
      "loss": 2.1431,
      "step": 144340
    },
    {
      "epoch": 1.88,
      "grad_norm": 0.9619958400726318,
      "learning_rate": 0.00018391458554283191,
      "loss": 1.9743,
      "step": 144341
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.1794244050979614,
      "learning_rate": 0.00018391081364576035,
      "loss": 1.9042,
      "step": 144342
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.0442110300064087,
      "learning_rate": 0.00018390704177027197,
      "loss": 2.069,
      "step": 144343
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.2413580417633057,
      "learning_rate": 0.00018390326991636738,
      "loss": 2.1153,
      "step": 144344
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.129339575767517,
      "learning_rate": 0.00018389949808404746,
      "loss": 2.1138,
      "step": 144345
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.0044336318969727,
      "learning_rate": 0.00018389572627331278,
      "loss": 2.2375,
      "step": 144346
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.102866291999817,
      "learning_rate": 0.00018389195448416403,
      "loss": 2.1317,
      "step": 144347
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.0094854831695557,
      "learning_rate": 0.000183888182716602,
      "loss": 2.1877,
      "step": 144348
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.102954387664795,
      "learning_rate": 0.00018388441097062736,
      "loss": 1.9603,
      "step": 144349
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.166585922241211,
      "learning_rate": 0.0001838806392462408,
      "loss": 1.9069,
      "step": 144350
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.0555049180984497,
      "learning_rate": 0.00018387686754344298,
      "loss": 1.948,
      "step": 144351
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.084208607673645,
      "learning_rate": 0.00018387309586223463,
      "loss": 2.0413,
      "step": 144352
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.094326138496399,
      "learning_rate": 0.00018386932420261652,
      "loss": 2.1285,
      "step": 144353
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.0556946992874146,
      "learning_rate": 0.00018386555256458926,
      "loss": 2.0535,
      "step": 144354
    },
    {
      "epoch": 1.88,
      "grad_norm": 0.9965949654579163,
      "learning_rate": 0.00018386178094815367,
      "loss": 2.0178,
      "step": 144355
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.3114699125289917,
      "learning_rate": 0.00018385800935331027,
      "loss": 2.0249,
      "step": 144356
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.159602165222168,
      "learning_rate": 0.00018385423778005986,
      "loss": 1.989,
      "step": 144357
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.0809330940246582,
      "learning_rate": 0.00018385046622840313,
      "loss": 2.1109,
      "step": 144358
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.2988860607147217,
      "learning_rate": 0.0001838466946983408,
      "loss": 2.1827,
      "step": 144359
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.1149705648422241,
      "learning_rate": 0.00018384292318987365,
      "loss": 2.2209,
      "step": 144360
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.2237476110458374,
      "learning_rate": 0.00018383915170300228,
      "loss": 2.2261,
      "step": 144361
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.4039690494537354,
      "learning_rate": 0.00018383538023772732,
      "loss": 2.1764,
      "step": 144362
    },
    {
      "epoch": 1.88,
      "grad_norm": 0.9673509001731873,
      "learning_rate": 0.00018383160879404956,
      "loss": 2.2591,
      "step": 144363
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.2009493112564087,
      "learning_rate": 0.00018382783737196974,
      "loss": 2.0193,
      "step": 144364
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.0847119092941284,
      "learning_rate": 0.0001838240659714885,
      "loss": 1.8764,
      "step": 144365
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.4592243432998657,
      "learning_rate": 0.0001838202945926065,
      "loss": 1.9715,
      "step": 144366
    },
    {
      "epoch": 1.88,
      "grad_norm": 0.9442669153213501,
      "learning_rate": 0.0001838165232353247,
      "loss": 2.1644,
      "step": 144367
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.1846752166748047,
      "learning_rate": 0.00018381275189964343,
      "loss": 2.2226,
      "step": 144368
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.3759311437606812,
      "learning_rate": 0.00018380898058556358,
      "loss": 2.1188,
      "step": 144369
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.088861107826233,
      "learning_rate": 0.0001838052092930858,
      "loss": 2.163,
      "step": 144370
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.274701714515686,
      "learning_rate": 0.0001838014380222109,
      "loss": 2.0446,
      "step": 144371
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.1503885984420776,
      "learning_rate": 0.00018379766677293946,
      "loss": 2.1083,
      "step": 144372
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.0788259506225586,
      "learning_rate": 0.00018379389554527235,
      "loss": 2.1593,
      "step": 144373
    },
    {
      "epoch": 1.88,
      "grad_norm": 0.9284263849258423,
      "learning_rate": 0.00018379012433921004,
      "loss": 1.9464,
      "step": 144374
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.0722932815551758,
      "learning_rate": 0.00018378635315475334,
      "loss": 2.2041,
      "step": 144375
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.4049808979034424,
      "learning_rate": 0.00018378258199190295,
      "loss": 1.9452,
      "step": 144376
    },
    {
      "epoch": 1.88,
      "grad_norm": 0.9820576906204224,
      "learning_rate": 0.00018377881085065954,
      "loss": 1.8516,
      "step": 144377
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.1430176496505737,
      "learning_rate": 0.00018377503973102387,
      "loss": 2.1678,
      "step": 144378
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.095350980758667,
      "learning_rate": 0.0001837712686329967,
      "loss": 1.9949,
      "step": 144379
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.2236034870147705,
      "learning_rate": 0.00018376749755657858,
      "loss": 2.0475,
      "step": 144380
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.0398120880126953,
      "learning_rate": 0.00018376372650177026,
      "loss": 1.8461,
      "step": 144381
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.065521478652954,
      "learning_rate": 0.00018375995546857245,
      "loss": 2.1452,
      "step": 144382
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.081721544265747,
      "learning_rate": 0.00018375618445698585,
      "loss": 1.9764,
      "step": 144383
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.1235498189926147,
      "learning_rate": 0.00018375241346701117,
      "loss": 2.0774,
      "step": 144384
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.1587070226669312,
      "learning_rate": 0.00018374864249864924,
      "loss": 2.1506,
      "step": 144385
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.1293010711669922,
      "learning_rate": 0.0001837448715519005,
      "loss": 2.0351,
      "step": 144386
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.0783698558807373,
      "learning_rate": 0.00018374110062676578,
      "loss": 1.888,
      "step": 144387
    },
    {
      "epoch": 1.88,
      "grad_norm": 0.987196147441864,
      "learning_rate": 0.0001837373297232458,
      "loss": 2.2021,
      "step": 144388
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.135269284248352,
      "learning_rate": 0.00018373355884134126,
      "loss": 2.1745,
      "step": 144389
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.0651394128799438,
      "learning_rate": 0.0001837297879810528,
      "loss": 2.1467,
      "step": 144390
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.0824946165084839,
      "learning_rate": 0.00018372601714238134,
      "loss": 1.9785,
      "step": 144391
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.1004592180252075,
      "learning_rate": 0.00018372224632532728,
      "loss": 2.1624,
      "step": 144392
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.0671366453170776,
      "learning_rate": 0.00018371847552989143,
      "loss": 2.1955,
      "step": 144393
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.077378511428833,
      "learning_rate": 0.00018371470475607449,
      "loss": 1.9502,
      "step": 144394
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.0198090076446533,
      "learning_rate": 0.0001837109340038772,
      "loss": 1.9764,
      "step": 144395
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.084536075592041,
      "learning_rate": 0.00018370716327330026,
      "loss": 2.0162,
      "step": 144396
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.0543055534362793,
      "learning_rate": 0.00018370339256434433,
      "loss": 2.0554,
      "step": 144397
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.08577299118042,
      "learning_rate": 0.00018369962187701029,
      "loss": 1.9163,
      "step": 144398
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.1869090795516968,
      "learning_rate": 0.00018369585121129852,
      "loss": 1.9912,
      "step": 144399
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.184145450592041,
      "learning_rate": 0.00018369208056720992,
      "loss": 2.2109,
      "step": 144400
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.153747797012329,
      "learning_rate": 0.00018368830994474513,
      "loss": 2.2849,
      "step": 144401
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.0270256996154785,
      "learning_rate": 0.0001836845393439049,
      "loss": 2.0004,
      "step": 144402
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.2509644031524658,
      "learning_rate": 0.0001836807687646899,
      "loss": 2.036,
      "step": 144403
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.2064415216445923,
      "learning_rate": 0.00018367699820710096,
      "loss": 2.1372,
      "step": 144404
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.0640588998794556,
      "learning_rate": 0.00018367322767113856,
      "loss": 2.1682,
      "step": 144405
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.1208113431930542,
      "learning_rate": 0.00018366945715680348,
      "loss": 1.9187,
      "step": 144406
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.2015049457550049,
      "learning_rate": 0.00018366568666409642,
      "loss": 2.1456,
      "step": 144407
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.20419180393219,
      "learning_rate": 0.00018366191619301816,
      "loss": 1.9427,
      "step": 144408
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.0069128274917603,
      "learning_rate": 0.00018365814574356935,
      "loss": 2.0477,
      "step": 144409
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.049070954322815,
      "learning_rate": 0.00018365437531575075,
      "loss": 2.0034,
      "step": 144410
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.0490609407424927,
      "learning_rate": 0.00018365060490956292,
      "loss": 2.1358,
      "step": 144411
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.1066601276397705,
      "learning_rate": 0.00018364683452500662,
      "loss": 2.0023,
      "step": 144412
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.533254861831665,
      "learning_rate": 0.00018364306416208256,
      "loss": 2.0409,
      "step": 144413
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.141373634338379,
      "learning_rate": 0.0001836392938207915,
      "loss": 1.8222,
      "step": 144414
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.1126115322113037,
      "learning_rate": 0.00018363552350113402,
      "loss": 1.9878,
      "step": 144415
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.0840692520141602,
      "learning_rate": 0.00018363175320311106,
      "loss": 2.3627,
      "step": 144416
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.1350587606430054,
      "learning_rate": 0.00018362798292672304,
      "loss": 2.3091,
      "step": 144417
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.13180410861969,
      "learning_rate": 0.00018362421267197073,
      "loss": 2.0252,
      "step": 144418
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.1234899759292603,
      "learning_rate": 0.00018362044243885493,
      "loss": 1.9099,
      "step": 144419
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.1819350719451904,
      "learning_rate": 0.00018361667222737627,
      "loss": 2.2229,
      "step": 144420
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.0336993932724,
      "learning_rate": 0.00018361290203753542,
      "loss": 2.1612,
      "step": 144421
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.1309094429016113,
      "learning_rate": 0.00018360913186933327,
      "loss": 1.9158,
      "step": 144422
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.1052525043487549,
      "learning_rate": 0.00018360536172277032,
      "loss": 2.0743,
      "step": 144423
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.0903377532958984,
      "learning_rate": 0.0001836015915978473,
      "loss": 2.1442,
      "step": 144424
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.0750211477279663,
      "learning_rate": 0.0001835978214945649,
      "loss": 1.9195,
      "step": 144425
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.167838454246521,
      "learning_rate": 0.0001835940514129239,
      "loss": 2.2337,
      "step": 144426
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.0743968486785889,
      "learning_rate": 0.00018359028135292498,
      "loss": 2.0403,
      "step": 144427
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.1693322658538818,
      "learning_rate": 0.00018358651131456885,
      "loss": 1.965,
      "step": 144428
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.0402754545211792,
      "learning_rate": 0.00018358274129785614,
      "loss": 1.9645,
      "step": 144429
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.0548253059387207,
      "learning_rate": 0.00018357897130278763,
      "loss": 2.0217,
      "step": 144430
    },
    {
      "epoch": 1.88,
      "grad_norm": 0.9975399971008301,
      "learning_rate": 0.00018357520132936396,
      "loss": 2.0499,
      "step": 144431
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.0264513492584229,
      "learning_rate": 0.00018357143137758585,
      "loss": 1.9421,
      "step": 144432
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.1212433576583862,
      "learning_rate": 0.000183567661447454,
      "loss": 2.0689,
      "step": 144433
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.289506435394287,
      "learning_rate": 0.00018356389153896915,
      "loss": 2.2228,
      "step": 144434
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.270747184753418,
      "learning_rate": 0.00018356012165213198,
      "loss": 2.0235,
      "step": 144435
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.0303364992141724,
      "learning_rate": 0.00018355635178694314,
      "loss": 2.1132,
      "step": 144436
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.188714861869812,
      "learning_rate": 0.00018355258194340337,
      "loss": 2.1273,
      "step": 144437
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.2269649505615234,
      "learning_rate": 0.00018354881212151344,
      "loss": 2.1331,
      "step": 144438
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.0339632034301758,
      "learning_rate": 0.00018354504232127395,
      "loss": 2.001,
      "step": 144439
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.1791880130767822,
      "learning_rate": 0.00018354127254268564,
      "loss": 2.1565,
      "step": 144440
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.1088753938674927,
      "learning_rate": 0.00018353750278574918,
      "loss": 2.1799,
      "step": 144441
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.0458080768585205,
      "learning_rate": 0.00018353373305046527,
      "loss": 1.9446,
      "step": 144442
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.1343692541122437,
      "learning_rate": 0.00018352996333683466,
      "loss": 1.9568,
      "step": 144443
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.0653387308120728,
      "learning_rate": 0.000183526193644858,
      "loss": 2.1024,
      "step": 144444
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.0941050052642822,
      "learning_rate": 0.00018352242397453613,
      "loss": 2.0733,
      "step": 144445
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.180515170097351,
      "learning_rate": 0.00018351865432586964,
      "loss": 2.0302,
      "step": 144446
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.2748628854751587,
      "learning_rate": 0.00018351488469885917,
      "loss": 2.0167,
      "step": 144447
    },
    {
      "epoch": 1.88,
      "grad_norm": 0.9844467639923096,
      "learning_rate": 0.00018351111509350542,
      "loss": 1.8924,
      "step": 144448
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.1548783779144287,
      "learning_rate": 0.00018350734550980918,
      "loss": 2.1299,
      "step": 144449
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.1870191097259521,
      "learning_rate": 0.00018350357594777112,
      "loss": 2.1392,
      "step": 144450
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.0705300569534302,
      "learning_rate": 0.00018349980640739198,
      "loss": 2.0008,
      "step": 144451
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.2368158102035522,
      "learning_rate": 0.00018349603688867252,
      "loss": 2.0454,
      "step": 144452
    },
    {
      "epoch": 1.88,
      "grad_norm": 0.9748827815055847,
      "learning_rate": 0.00018349226739161322,
      "loss": 1.8958,
      "step": 144453
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.0981543064117432,
      "learning_rate": 0.00018348849791621495,
      "loss": 2.103,
      "step": 144454
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.1436289548873901,
      "learning_rate": 0.00018348472846247833,
      "loss": 2.1312,
      "step": 144455
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.058305263519287,
      "learning_rate": 0.00018348095903040408,
      "loss": 2.1633,
      "step": 144456
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.1213836669921875,
      "learning_rate": 0.00018347718961999294,
      "loss": 2.0861,
      "step": 144457
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.081271767616272,
      "learning_rate": 0.0001834734202312457,
      "loss": 2.0475,
      "step": 144458
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.024923324584961,
      "learning_rate": 0.00018346965086416285,
      "loss": 1.9584,
      "step": 144459
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.0212609767913818,
      "learning_rate": 0.00018346588151874519,
      "loss": 2.2038,
      "step": 144460
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.0245916843414307,
      "learning_rate": 0.0001834621121949934,
      "loss": 2.1873,
      "step": 144461
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.026242733001709,
      "learning_rate": 0.00018345834289290824,
      "loss": 1.9392,
      "step": 144462
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.3338881731033325,
      "learning_rate": 0.00018345457361249037,
      "loss": 1.9304,
      "step": 144463
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.1293649673461914,
      "learning_rate": 0.00018345080435374045,
      "loss": 2.0136,
      "step": 144464
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.034125566482544,
      "learning_rate": 0.00018344703511665938,
      "loss": 2.0064,
      "step": 144465
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.118461012840271,
      "learning_rate": 0.0001834432659012476,
      "loss": 2.1755,
      "step": 144466
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.098852276802063,
      "learning_rate": 0.00018343949670750589,
      "loss": 2.1082,
      "step": 144467
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.095930576324463,
      "learning_rate": 0.00018343572753543497,
      "loss": 2.0048,
      "step": 144468
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.0135921239852905,
      "learning_rate": 0.00018343195838503557,
      "loss": 1.9579,
      "step": 144469
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.0214312076568604,
      "learning_rate": 0.00018342818925630837,
      "loss": 2.1067,
      "step": 144470
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.1330801248550415,
      "learning_rate": 0.00018342442014925418,
      "loss": 1.9801,
      "step": 144471
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.0504703521728516,
      "learning_rate": 0.00018342065106387347,
      "loss": 2.1075,
      "step": 144472
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.1458110809326172,
      "learning_rate": 0.00018341688200016707,
      "loss": 2.2334,
      "step": 144473
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.040697693824768,
      "learning_rate": 0.00018341311295813566,
      "loss": 1.9625,
      "step": 144474
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.0172977447509766,
      "learning_rate": 0.00018340934393777994,
      "loss": 2.0412,
      "step": 144475
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.7945307493209839,
      "learning_rate": 0.00018340557493910065,
      "loss": 2.0283,
      "step": 144476
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.238326072692871,
      "learning_rate": 0.0001834018059620986,
      "loss": 2.1278,
      "step": 144477
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.0167324542999268,
      "learning_rate": 0.0001833980370067742,
      "loss": 2.0778,
      "step": 144478
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.4084855318069458,
      "learning_rate": 0.00018339426807312837,
      "loss": 2.1039,
      "step": 144479
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.0946102142333984,
      "learning_rate": 0.00018339049916116164,
      "loss": 2.0646,
      "step": 144480
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.2273372411727905,
      "learning_rate": 0.00018338673027087492,
      "loss": 2.1373,
      "step": 144481
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.137943148612976,
      "learning_rate": 0.00018338296140226875,
      "loss": 1.9677,
      "step": 144482
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.1661232709884644,
      "learning_rate": 0.00018337919255534405,
      "loss": 2.0272,
      "step": 144483
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.0730546712875366,
      "learning_rate": 0.0001833754237301012,
      "loss": 2.0762,
      "step": 144484
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.1330348253250122,
      "learning_rate": 0.00018337165492654105,
      "loss": 2.108,
      "step": 144485
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.1363446712493896,
      "learning_rate": 0.00018336788614466433,
      "loss": 1.8361,
      "step": 144486
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.1948049068450928,
      "learning_rate": 0.00018336411738447173,
      "loss": 2.0157,
      "step": 144487
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.0502263307571411,
      "learning_rate": 0.00018336034864596394,
      "loss": 2.1048,
      "step": 144488
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.4624946117401123,
      "learning_rate": 0.00018335657992914176,
      "loss": 2.1418,
      "step": 144489
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.3578141927719116,
      "learning_rate": 0.0001833528112340057,
      "loss": 1.9734,
      "step": 144490
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.1956651210784912,
      "learning_rate": 0.00018334904256055656,
      "loss": 2.1036,
      "step": 144491
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.0759987831115723,
      "learning_rate": 0.000183345273908795,
      "loss": 2.3138,
      "step": 144492
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.2844656705856323,
      "learning_rate": 0.00018334150527872178,
      "loss": 2.0756,
      "step": 144493
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.2414392232894897,
      "learning_rate": 0.00018333773667033755,
      "loss": 2.0632,
      "step": 144494
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.1658532619476318,
      "learning_rate": 0.00018333396808364318,
      "loss": 2.0824,
      "step": 144495
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.0880473852157593,
      "learning_rate": 0.00018333019951863913,
      "loss": 2.0346,
      "step": 144496
    },
    {
      "epoch": 1.88,
      "grad_norm": 2.133758783340454,
      "learning_rate": 0.00018332643097532619,
      "loss": 2.1044,
      "step": 144497
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.2064769268035889,
      "learning_rate": 0.00018332266245370501,
      "loss": 2.2585,
      "step": 144498
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.197151780128479,
      "learning_rate": 0.00018331889395377642,
      "loss": 1.9628,
      "step": 144499
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.0402072668075562,
      "learning_rate": 0.00018331512547554104,
      "loss": 1.9951,
      "step": 144500
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.144364595413208,
      "learning_rate": 0.00018331135701899967,
      "loss": 2.0245,
      "step": 144501
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.1925323009490967,
      "learning_rate": 0.00018330758858415284,
      "loss": 1.9689,
      "step": 144502
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.094986081123352,
      "learning_rate": 0.00018330382017100127,
      "loss": 2.1567,
      "step": 144503
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.0773298740386963,
      "learning_rate": 0.00018330005177954578,
      "loss": 2.2465,
      "step": 144504
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.0980651378631592,
      "learning_rate": 0.000183296283409787,
      "loss": 2.1293,
      "step": 144505
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.29156494140625,
      "learning_rate": 0.00018329251506172563,
      "loss": 2.362,
      "step": 144506
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.0501551628112793,
      "learning_rate": 0.00018328874673536244,
      "loss": 2.0205,
      "step": 144507
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.2934905290603638,
      "learning_rate": 0.0001832849784306981,
      "loss": 2.1187,
      "step": 144508
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.1366755962371826,
      "learning_rate": 0.0001832812101477332,
      "loss": 2.2174,
      "step": 144509
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.0637738704681396,
      "learning_rate": 0.00018327744188646858,
      "loss": 2.2221,
      "step": 144510
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.056978464126587,
      "learning_rate": 0.00018327367364690482,
      "loss": 2.0796,
      "step": 144511
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.0859882831573486,
      "learning_rate": 0.0001832699054290427,
      "loss": 2.1723,
      "step": 144512
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.2038320302963257,
      "learning_rate": 0.000183266137232883,
      "loss": 2.271,
      "step": 144513
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.1941111087799072,
      "learning_rate": 0.00018326236905842624,
      "loss": 2.1789,
      "step": 144514
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.1037709712982178,
      "learning_rate": 0.00018325860090567328,
      "loss": 2.0013,
      "step": 144515
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.098803162574768,
      "learning_rate": 0.0001832548327746247,
      "loss": 1.9471,
      "step": 144516
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.2113378047943115,
      "learning_rate": 0.00018325106466528127,
      "loss": 2.1599,
      "step": 144517
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.0756685733795166,
      "learning_rate": 0.00018324729657764362,
      "loss": 2.0375,
      "step": 144518
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.0061159133911133,
      "learning_rate": 0.00018324352851171256,
      "loss": 2.0439,
      "step": 144519
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.0889919996261597,
      "learning_rate": 0.0001832397604674887,
      "loss": 2.1885,
      "step": 144520
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.117891550064087,
      "learning_rate": 0.00018323599244497276,
      "loss": 2.0009,
      "step": 144521
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.1079843044281006,
      "learning_rate": 0.00018323222444416553,
      "loss": 1.8437,
      "step": 144522
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.0941519737243652,
      "learning_rate": 0.00018322845646506758,
      "loss": 2.1514,
      "step": 144523
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.0545594692230225,
      "learning_rate": 0.0001832246885076796,
      "loss": 2.0751,
      "step": 144524
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.000003457069397,
      "learning_rate": 0.0001832209205720025,
      "loss": 2.054,
      "step": 144525
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.357191801071167,
      "learning_rate": 0.00018321715265803672,
      "loss": 1.9586,
      "step": 144526
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.1382137537002563,
      "learning_rate": 0.00018321338476578311,
      "loss": 2.0635,
      "step": 144527
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.1632314920425415,
      "learning_rate": 0.00018320961689524234,
      "loss": 2.2687,
      "step": 144528
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.0764552354812622,
      "learning_rate": 0.00018320584904641508,
      "loss": 2.0419,
      "step": 144529
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.0947095155715942,
      "learning_rate": 0.00018320208121930208,
      "loss": 2.2082,
      "step": 144530
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.1342880725860596,
      "learning_rate": 0.00018319831341390402,
      "loss": 2.0394,
      "step": 144531
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.0457810163497925,
      "learning_rate": 0.00018319454563022166,
      "loss": 1.7098,
      "step": 144532
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.0625700950622559,
      "learning_rate": 0.00018319077786825553,
      "loss": 2.1583,
      "step": 144533
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.0213799476623535,
      "learning_rate": 0.00018318701012800647,
      "loss": 2.0905,
      "step": 144534
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.1216943264007568,
      "learning_rate": 0.00018318324240947518,
      "loss": 1.916,
      "step": 144535
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.1093220710754395,
      "learning_rate": 0.0001831794747126623,
      "loss": 2.1301,
      "step": 144536
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.1354951858520508,
      "learning_rate": 0.0001831757070375686,
      "loss": 2.2516,
      "step": 144537
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.0633021593093872,
      "learning_rate": 0.00018317193938419477,
      "loss": 1.9135,
      "step": 144538
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.188538670539856,
      "learning_rate": 0.0001831681717525414,
      "loss": 2.1411,
      "step": 144539
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.0710810422897339,
      "learning_rate": 0.00018316440414260925,
      "loss": 1.9556,
      "step": 144540
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.1947437524795532,
      "learning_rate": 0.00018316063655439912,
      "loss": 2.2545,
      "step": 144541
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.0530925989151,
      "learning_rate": 0.0001831568689879116,
      "loss": 1.9971,
      "step": 144542
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.1338632106781006,
      "learning_rate": 0.0001831531014431474,
      "loss": 2.0703,
      "step": 144543
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.094900131225586,
      "learning_rate": 0.00018314933392010735,
      "loss": 2.2104,
      "step": 144544
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.1316142082214355,
      "learning_rate": 0.00018314556641879198,
      "loss": 2.265,
      "step": 144545
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.0735338926315308,
      "learning_rate": 0.00018314179893920204,
      "loss": 2.0975,
      "step": 144546
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.064636468887329,
      "learning_rate": 0.00018313803148133821,
      "loss": 2.0249,
      "step": 144547
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.1743630170822144,
      "learning_rate": 0.00018313426404520126,
      "loss": 2.1245,
      "step": 144548
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.0153449773788452,
      "learning_rate": 0.00018313049663079182,
      "loss": 1.9145,
      "step": 144549
    },
    {
      "epoch": 1.88,
      "grad_norm": 0.9600694179534912,
      "learning_rate": 0.0001831267292381108,
      "loss": 2.0124,
      "step": 144550
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.1683497428894043,
      "learning_rate": 0.0001831229618671586,
      "loss": 1.935,
      "step": 144551
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.2807847261428833,
      "learning_rate": 0.00018311919451793603,
      "loss": 2.1438,
      "step": 144552
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.116040587425232,
      "learning_rate": 0.0001831154271904438,
      "loss": 2.1166,
      "step": 144553
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.045432209968567,
      "learning_rate": 0.0001831116598846826,
      "loss": 2.0062,
      "step": 144554
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.3017442226409912,
      "learning_rate": 0.00018310789260065317,
      "loss": 2.3719,
      "step": 144555
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.2804442644119263,
      "learning_rate": 0.00018310412533835633,
      "loss": 2.1632,
      "step": 144556
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.0966556072235107,
      "learning_rate": 0.0001831003580977925,
      "loss": 1.9809,
      "step": 144557
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.1817408800125122,
      "learning_rate": 0.0001830965908789625,
      "loss": 2.0567,
      "step": 144558
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.225954294204712,
      "learning_rate": 0.00018309282368186706,
      "loss": 2.0892,
      "step": 144559
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.0372917652130127,
      "learning_rate": 0.0001830890565065069,
      "loss": 1.8947,
      "step": 144560
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.1611675024032593,
      "learning_rate": 0.0001830852893528827,
      "loss": 2.1049,
      "step": 144561
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.0701473951339722,
      "learning_rate": 0.0001830815222209952,
      "loss": 1.9873,
      "step": 144562
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.0552682876586914,
      "learning_rate": 0.000183077755110845,
      "loss": 2.0911,
      "step": 144563
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.1984378099441528,
      "learning_rate": 0.00018307398802243282,
      "loss": 2.316,
      "step": 144564
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.2817368507385254,
      "learning_rate": 0.00018307022095575943,
      "loss": 2.0292,
      "step": 144565
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.1332483291625977,
      "learning_rate": 0.00018306645391082544,
      "loss": 2.0599,
      "step": 144566
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.0686153173446655,
      "learning_rate": 0.00018306268688763166,
      "loss": 2.2294,
      "step": 144567
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.0616638660430908,
      "learning_rate": 0.00018305891988617883,
      "loss": 1.9576,
      "step": 144568
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.025123953819275,
      "learning_rate": 0.00018305515290646738,
      "loss": 2.0305,
      "step": 144569
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.3028969764709473,
      "learning_rate": 0.00018305138594849825,
      "loss": 2.2884,
      "step": 144570
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.236456274986267,
      "learning_rate": 0.00018304761901227206,
      "loss": 2.2572,
      "step": 144571
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.067322850227356,
      "learning_rate": 0.00018304385209778952,
      "loss": 2.1564,
      "step": 144572
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.1345913410186768,
      "learning_rate": 0.00018304008520505136,
      "loss": 1.9562,
      "step": 144573
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.1663144826889038,
      "learning_rate": 0.00018303631833405833,
      "loss": 2.0749,
      "step": 144574
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.1282341480255127,
      "learning_rate": 0.00018303255148481096,
      "loss": 2.0087,
      "step": 144575
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.1241503953933716,
      "learning_rate": 0.00018302878465731006,
      "loss": 2.1185,
      "step": 144576
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.3868732452392578,
      "learning_rate": 0.00018302501785155627,
      "loss": 2.1009,
      "step": 144577
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.1684170961380005,
      "learning_rate": 0.0001830212510675504,
      "loss": 2.1869,
      "step": 144578
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.3017008304595947,
      "learning_rate": 0.00018301748430529305,
      "loss": 2.06,
      "step": 144579
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.3454355001449585,
      "learning_rate": 0.0001830137175647851,
      "loss": 2.0503,
      "step": 144580
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.2022572755813599,
      "learning_rate": 0.00018300995084602698,
      "loss": 2.0424,
      "step": 144581
    },
    {
      "epoch": 1.88,
      "grad_norm": 0.9984604120254517,
      "learning_rate": 0.00018300618414901948,
      "loss": 2.3406,
      "step": 144582
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.3292983770370483,
      "learning_rate": 0.00018300241747376338,
      "loss": 2.3221,
      "step": 144583
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.2421692609786987,
      "learning_rate": 0.00018299865082025936,
      "loss": 1.9635,
      "step": 144584
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.0146323442459106,
      "learning_rate": 0.00018299488418850804,
      "loss": 2.1369,
      "step": 144585
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.1238597631454468,
      "learning_rate": 0.00018299111757851035,
      "loss": 2.1042,
      "step": 144586
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.2339515686035156,
      "learning_rate": 0.00018298735099026665,
      "loss": 2.1632,
      "step": 144587
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.2084344625473022,
      "learning_rate": 0.00018298358442377788,
      "loss": 2.173,
      "step": 144588
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.084496259689331,
      "learning_rate": 0.00018297981787904463,
      "loss": 2.2011,
      "step": 144589
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.126477837562561,
      "learning_rate": 0.00018297605135606762,
      "loss": 1.9371,
      "step": 144590
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.2470800876617432,
      "learning_rate": 0.0001829722848548476,
      "loss": 2.2257,
      "step": 144591
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.194189429283142,
      "learning_rate": 0.00018296851837538536,
      "loss": 2.0453,
      "step": 144592
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.1536129713058472,
      "learning_rate": 0.00018296475191768136,
      "loss": 2.0562,
      "step": 144593
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.066563606262207,
      "learning_rate": 0.0001829609854817364,
      "loss": 1.9698,
      "step": 144594
    },
    {
      "epoch": 1.88,
      "grad_norm": 0.9976534247398376,
      "learning_rate": 0.00018295721906755124,
      "loss": 1.9539,
      "step": 144595
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.0709437131881714,
      "learning_rate": 0.0001829534526751265,
      "loss": 1.7846,
      "step": 144596
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.0735036134719849,
      "learning_rate": 0.00018294968630446293,
      "loss": 1.9793,
      "step": 144597
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.179092288017273,
      "learning_rate": 0.00018294591995556125,
      "loss": 2.1892,
      "step": 144598
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.0793709754943848,
      "learning_rate": 0.0001829421536284222,
      "loss": 2.0173,
      "step": 144599
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.2005380392074585,
      "learning_rate": 0.0001829383873230464,
      "loss": 2.0462,
      "step": 144600
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.1486204862594604,
      "learning_rate": 0.00018293462103943447,
      "loss": 2.1467,
      "step": 144601
    },
    {
      "epoch": 1.88,
      "grad_norm": 0.9804671406745911,
      "learning_rate": 0.00018293085477758724,
      "loss": 2.0841,
      "step": 144602
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.1912213563919067,
      "learning_rate": 0.00018292708853750536,
      "loss": 2.2019,
      "step": 144603
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.1075608730316162,
      "learning_rate": 0.00018292332231918955,
      "loss": 2.1282,
      "step": 144604
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.2708944082260132,
      "learning_rate": 0.00018291955612264056,
      "loss": 2.116,
      "step": 144605
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.0749891996383667,
      "learning_rate": 0.000182915789947859,
      "loss": 1.9484,
      "step": 144606
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.1865649223327637,
      "learning_rate": 0.0001829120237948456,
      "loss": 2.2735,
      "step": 144607
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.391014575958252,
      "learning_rate": 0.00018290825766360106,
      "loss": 1.9349,
      "step": 144608
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.045648217201233,
      "learning_rate": 0.00018290449155412608,
      "loss": 2.1278,
      "step": 144609
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.035707950592041,
      "learning_rate": 0.00018290072546642137,
      "loss": 1.9201,
      "step": 144610
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.0235716104507446,
      "learning_rate": 0.00018289695940048765,
      "loss": 2.2291,
      "step": 144611
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.1590633392333984,
      "learning_rate": 0.0001828931933563256,
      "loss": 2.1419,
      "step": 144612
    },
    {
      "epoch": 1.88,
      "grad_norm": 0.9876846075057983,
      "learning_rate": 0.00018288942733393586,
      "loss": 2.0941,
      "step": 144613
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.0855096578598022,
      "learning_rate": 0.00018288566133331922,
      "loss": 1.8913,
      "step": 144614
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.0684579610824585,
      "learning_rate": 0.00018288189535447635,
      "loss": 2.166,
      "step": 144615
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.1457998752593994,
      "learning_rate": 0.00018287812939740794,
      "loss": 1.958,
      "step": 144616
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.2635623216629028,
      "learning_rate": 0.00018287436346211473,
      "loss": 1.9884,
      "step": 144617
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.0398566722869873,
      "learning_rate": 0.00018287059754859734,
      "loss": 2.2027,
      "step": 144618
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.1556649208068848,
      "learning_rate": 0.0001828668316568565,
      "loss": 2.2038,
      "step": 144619
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.0411266088485718,
      "learning_rate": 0.00018286306578689296,
      "loss": 2.1394,
      "step": 144620
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.192853331565857,
      "learning_rate": 0.00018285929993870737,
      "loss": 1.9277,
      "step": 144621
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.144859790802002,
      "learning_rate": 0.00018285553411230053,
      "loss": 2.2286,
      "step": 144622
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.0693094730377197,
      "learning_rate": 0.00018285176830767298,
      "loss": 2.1588,
      "step": 144623
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.1017484664916992,
      "learning_rate": 0.0001828480025248255,
      "loss": 1.8335,
      "step": 144624
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.1255134344100952,
      "learning_rate": 0.00018284423676375879,
      "loss": 2.3294,
      "step": 144625
    },
    {
      "epoch": 1.88,
      "grad_norm": 0.9714480042457581,
      "learning_rate": 0.00018284047102447353,
      "loss": 1.9518,
      "step": 144626
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.1580021381378174,
      "learning_rate": 0.00018283670530697043,
      "loss": 1.897,
      "step": 144627
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.0168808698654175,
      "learning_rate": 0.00018283293961125026,
      "loss": 1.8163,
      "step": 144628
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.002333402633667,
      "learning_rate": 0.0001828291739373137,
      "loss": 2.1663,
      "step": 144629
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.1101185083389282,
      "learning_rate": 0.00018282540828516133,
      "loss": 2.0415,
      "step": 144630
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.0274590253829956,
      "learning_rate": 0.00018282164265479389,
      "loss": 2.052,
      "step": 144631
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.1759934425354004,
      "learning_rate": 0.00018281787704621218,
      "loss": 2.022,
      "step": 144632
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.0076791048049927,
      "learning_rate": 0.00018281411145941678,
      "loss": 2.34,
      "step": 144633
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.039869785308838,
      "learning_rate": 0.00018281034589440849,
      "loss": 2.0743,
      "step": 144634
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.0669829845428467,
      "learning_rate": 0.00018280658035118804,
      "loss": 2.1336,
      "step": 144635
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.080399513244629,
      "learning_rate": 0.00018280281482975597,
      "loss": 1.8432,
      "step": 144636
    },
    {
      "epoch": 1.88,
      "grad_norm": 0.9350622296333313,
      "learning_rate": 0.00018279904933011308,
      "loss": 1.9434,
      "step": 144637
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.3923567533493042,
      "learning_rate": 0.00018279528385226006,
      "loss": 1.9156,
      "step": 144638
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.1903177499771118,
      "learning_rate": 0.00018279151839619762,
      "loss": 2.0978,
      "step": 144639
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.017987608909607,
      "learning_rate": 0.00018278775296192643,
      "loss": 2.0291,
      "step": 144640
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.0945926904678345,
      "learning_rate": 0.0001827839875494473,
      "loss": 2.025,
      "step": 144641
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.2405116558074951,
      "learning_rate": 0.00018278022215876077,
      "loss": 1.9511,
      "step": 144642
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.4605249166488647,
      "learning_rate": 0.00018277645678986757,
      "loss": 2.3057,
      "step": 144643
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.0672260522842407,
      "learning_rate": 0.00018277269144276846,
      "loss": 2.2463,
      "step": 144644
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.1718312501907349,
      "learning_rate": 0.0001827689261174641,
      "loss": 2.3033,
      "step": 144645
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.2018237113952637,
      "learning_rate": 0.0001827651608139552,
      "loss": 2.058,
      "step": 144646
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.105753779411316,
      "learning_rate": 0.00018276139553224268,
      "loss": 1.9152,
      "step": 144647
    },
    {
      "epoch": 1.88,
      "grad_norm": 0.9799688458442688,
      "learning_rate": 0.00018275763027232681,
      "loss": 2.2287,
      "step": 144648
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.0648021697998047,
      "learning_rate": 0.00018275386503420856,
      "loss": 2.1036,
      "step": 144649
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.168612003326416,
      "learning_rate": 0.00018275009981788852,
      "loss": 1.9407,
      "step": 144650
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.1279820203781128,
      "learning_rate": 0.00018274633462336752,
      "loss": 2.1016,
      "step": 144651
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.0358420610427856,
      "learning_rate": 0.00018274256945064615,
      "loss": 2.1356,
      "step": 144652
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.0630810260772705,
      "learning_rate": 0.00018273880429972532,
      "loss": 1.9317,
      "step": 144653
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.053917407989502,
      "learning_rate": 0.00018273503917060538,
      "loss": 2.0338,
      "step": 144654
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.0353378057479858,
      "learning_rate": 0.00018273127406328725,
      "loss": 2.0805,
      "step": 144655
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.0919972658157349,
      "learning_rate": 0.00018272750897777156,
      "loss": 2.0737,
      "step": 144656
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.0465444326400757,
      "learning_rate": 0.00018272374391405905,
      "loss": 1.957,
      "step": 144657
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.0460245609283447,
      "learning_rate": 0.0001827199788721504,
      "loss": 1.9926,
      "step": 144658
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.0390957593917847,
      "learning_rate": 0.00018271621385204645,
      "loss": 1.9633,
      "step": 144659
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.1604222059249878,
      "learning_rate": 0.00018271244885374765,
      "loss": 2.0388,
      "step": 144660
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.2032326459884644,
      "learning_rate": 0.00018270868387725483,
      "loss": 2.2029,
      "step": 144661
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.1977051496505737,
      "learning_rate": 0.00018270491892256863,
      "loss": 2.1771,
      "step": 144662
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.1332676410675049,
      "learning_rate": 0.00018270115398968985,
      "loss": 2.0288,
      "step": 144663
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.2988094091415405,
      "learning_rate": 0.00018269738907861915,
      "loss": 2.2261,
      "step": 144664
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.2657396793365479,
      "learning_rate": 0.0001826936241893572,
      "loss": 2.2481,
      "step": 144665
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.244660496711731,
      "learning_rate": 0.00018268985932190485,
      "loss": 1.9399,
      "step": 144666
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.0655242204666138,
      "learning_rate": 0.00018268609447626253,
      "loss": 1.9477,
      "step": 144667
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.6135423183441162,
      "learning_rate": 0.00018268232965243107,
      "loss": 1.9232,
      "step": 144668
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.0784542560577393,
      "learning_rate": 0.0001826785648504112,
      "loss": 2.145,
      "step": 144669
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.1328181028366089,
      "learning_rate": 0.00018267480007020363,
      "loss": 2.3428,
      "step": 144670
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.1913204193115234,
      "learning_rate": 0.00018267103531180898,
      "loss": 2.195,
      "step": 144671
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.1826519966125488,
      "learning_rate": 0.00018266727057522812,
      "loss": 2.1328,
      "step": 144672
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.240517258644104,
      "learning_rate": 0.00018266350586046152,
      "loss": 2.1732,
      "step": 144673
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.1143503189086914,
      "learning_rate": 0.00018265974116751002,
      "loss": 2.0363,
      "step": 144674
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.028210997581482,
      "learning_rate": 0.00018265597649637426,
      "loss": 2.1408,
      "step": 144675
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.3767749071121216,
      "learning_rate": 0.00018265221184705495,
      "loss": 1.9767,
      "step": 144676
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.3082729578018188,
      "learning_rate": 0.00018264844721955283,
      "loss": 1.9364,
      "step": 144677
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.1412124633789062,
      "learning_rate": 0.0001826446826138687,
      "loss": 2.1807,
      "step": 144678
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.0640901327133179,
      "learning_rate": 0.00018264091803000304,
      "loss": 2.0713,
      "step": 144679
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.1137384176254272,
      "learning_rate": 0.0001826371534679566,
      "loss": 2.0,
      "step": 144680
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.0294599533081055,
      "learning_rate": 0.00018263338892773016,
      "loss": 1.9399,
      "step": 144681
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.1060528755187988,
      "learning_rate": 0.0001826296244093244,
      "loss": 2.1578,
      "step": 144682
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.1985361576080322,
      "learning_rate": 0.00018262585991273998,
      "loss": 1.9914,
      "step": 144683
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.270862340927124,
      "learning_rate": 0.0001826220954379778,
      "loss": 2.0142,
      "step": 144684
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.0988661050796509,
      "learning_rate": 0.00018261833098503818,
      "loss": 2.1763,
      "step": 144685
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.1350419521331787,
      "learning_rate": 0.0001826145665539221,
      "loss": 1.9392,
      "step": 144686
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.315279483795166,
      "learning_rate": 0.00018261080214463017,
      "loss": 2.2804,
      "step": 144687
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.0922333002090454,
      "learning_rate": 0.0001826070377571631,
      "loss": 1.9635,
      "step": 144688
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.1713191270828247,
      "learning_rate": 0.00018260327339152157,
      "loss": 2.1868,
      "step": 144689
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.1146271228790283,
      "learning_rate": 0.00018259950904770642,
      "loss": 1.9569,
      "step": 144690
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.0093278884887695,
      "learning_rate": 0.00018259574472571816,
      "loss": 2.1674,
      "step": 144691
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.194912075996399,
      "learning_rate": 0.0001825919804255576,
      "loss": 2.0041,
      "step": 144692
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.140379548072815,
      "learning_rate": 0.00018258821614722536,
      "loss": 2.1956,
      "step": 144693
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.1540278196334839,
      "learning_rate": 0.0001825844518907222,
      "loss": 2.1084,
      "step": 144694
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.089980959892273,
      "learning_rate": 0.00018258068765604883,
      "loss": 2.1027,
      "step": 144695
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.0685968399047852,
      "learning_rate": 0.00018257692344320593,
      "loss": 2.0459,
      "step": 144696
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.1767847537994385,
      "learning_rate": 0.00018257315925219417,
      "loss": 2.2165,
      "step": 144697
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.104508638381958,
      "learning_rate": 0.00018256939508301428,
      "loss": 2.2467,
      "step": 144698
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.1398049592971802,
      "learning_rate": 0.000182565630935667,
      "loss": 2.2376,
      "step": 144699
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.0830096006393433,
      "learning_rate": 0.00018256186681015293,
      "loss": 1.902,
      "step": 144700
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.1060357093811035,
      "learning_rate": 0.00018255810270647284,
      "loss": 2.2624,
      "step": 144701
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.1244982481002808,
      "learning_rate": 0.0001825543386246275,
      "loss": 2.2839,
      "step": 144702
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.068769097328186,
      "learning_rate": 0.00018255057456461742,
      "loss": 2.0684,
      "step": 144703
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.0496026277542114,
      "learning_rate": 0.0001825468105264434,
      "loss": 2.1957,
      "step": 144704
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.0268898010253906,
      "learning_rate": 0.00018254304651010614,
      "loss": 2.0215,
      "step": 144705
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.0707123279571533,
      "learning_rate": 0.00018253928251560643,
      "loss": 2.128,
      "step": 144706
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.0741615295410156,
      "learning_rate": 0.0001825355185429448,
      "loss": 1.9816,
      "step": 144707
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.0640281438827515,
      "learning_rate": 0.00018253175459212212,
      "loss": 2.2824,
      "step": 144708
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.0727587938308716,
      "learning_rate": 0.00018252799066313896,
      "loss": 1.8889,
      "step": 144709
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.121965765953064,
      "learning_rate": 0.00018252422675599608,
      "loss": 2.0275,
      "step": 144710
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.0550990104675293,
      "learning_rate": 0.00018252046287069412,
      "loss": 2.1161,
      "step": 144711
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.0224727392196655,
      "learning_rate": 0.0001825166990072338,
      "loss": 2.1583,
      "step": 144712
    },
    {
      "epoch": 1.88,
      "grad_norm": 0.984604001045227,
      "learning_rate": 0.00018251293516561597,
      "loss": 1.9913,
      "step": 144713
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.0530850887298584,
      "learning_rate": 0.00018250917134584113,
      "loss": 2.0008,
      "step": 144714
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.1495813131332397,
      "learning_rate": 0.00018250540754791007,
      "loss": 2.3321,
      "step": 144715
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.108947515487671,
      "learning_rate": 0.0001825016437718234,
      "loss": 1.9877,
      "step": 144716
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.216379165649414,
      "learning_rate": 0.00018249788001758192,
      "loss": 2.1135,
      "step": 144717
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.222976803779602,
      "learning_rate": 0.00018249411628518632,
      "loss": 2.1155,
      "step": 144718
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.0784156322479248,
      "learning_rate": 0.00018249035257463726,
      "loss": 2.1097,
      "step": 144719
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.1103243827819824,
      "learning_rate": 0.00018248658888593558,
      "loss": 2.0953,
      "step": 144720
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.128527045249939,
      "learning_rate": 0.00018248282521908174,
      "loss": 2.0358,
      "step": 144721
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.146857500076294,
      "learning_rate": 0.0001824790615740766,
      "loss": 1.8942,
      "step": 144722
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.273490309715271,
      "learning_rate": 0.00018247529795092078,
      "loss": 1.8517,
      "step": 144723
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.0987348556518555,
      "learning_rate": 0.00018247153434961503,
      "loss": 2.4158,
      "step": 144724
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.2066645622253418,
      "learning_rate": 0.00018246777077016006,
      "loss": 2.0957,
      "step": 144725
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.1969736814498901,
      "learning_rate": 0.00018246400721255666,
      "loss": 2.1011,
      "step": 144726
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.0898139476776123,
      "learning_rate": 0.00018246024367680528,
      "loss": 2.1381,
      "step": 144727
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.1856454610824585,
      "learning_rate": 0.00018245648016290678,
      "loss": 2.0585,
      "step": 144728
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.0384254455566406,
      "learning_rate": 0.0001824527166708618,
      "loss": 2.129,
      "step": 144729
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.1327269077301025,
      "learning_rate": 0.00018244895320067113,
      "loss": 2.1822,
      "step": 144730
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.1471669673919678,
      "learning_rate": 0.00018244518975233542,
      "loss": 2.0504,
      "step": 144731
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.0484241247177124,
      "learning_rate": 0.00018244142632585533,
      "loss": 2.1057,
      "step": 144732
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.0666812658309937,
      "learning_rate": 0.00018243766292123177,
      "loss": 2.231,
      "step": 144733
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.0919345617294312,
      "learning_rate": 0.00018243389953846508,
      "loss": 1.9632,
      "step": 144734
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.1725499629974365,
      "learning_rate": 0.00018243013617755618,
      "loss": 2.159,
      "step": 144735
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.0505683422088623,
      "learning_rate": 0.00018242637283850575,
      "loss": 2.2061,
      "step": 144736
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.1500135660171509,
      "learning_rate": 0.00018242260952131445,
      "loss": 1.8788,
      "step": 144737
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.16196608543396,
      "learning_rate": 0.000182418846225983,
      "loss": 1.9953,
      "step": 144738
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.1331931352615356,
      "learning_rate": 0.0001824150829525123,
      "loss": 1.9759,
      "step": 144739
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.0147814750671387,
      "learning_rate": 0.00018241131970090266,
      "loss": 2.0337,
      "step": 144740
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.1832343339920044,
      "learning_rate": 0.000182407556471155,
      "loss": 2.1501,
      "step": 144741
    },
    {
      "epoch": 1.88,
      "grad_norm": 0.9688636660575867,
      "learning_rate": 0.00018240379326327,
      "loss": 1.8001,
      "step": 144742
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.0783755779266357,
      "learning_rate": 0.00018240003007724838,
      "loss": 2.1523,
      "step": 144743
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.1127737760543823,
      "learning_rate": 0.00018239626691309076,
      "loss": 2.2784,
      "step": 144744
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.0123542547225952,
      "learning_rate": 0.00018239250377079802,
      "loss": 1.9224,
      "step": 144745
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.1403560638427734,
      "learning_rate": 0.00018238874065037064,
      "loss": 2.1464,
      "step": 144746
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.091792106628418,
      "learning_rate": 0.00018238497755180942,
      "loss": 1.7929,
      "step": 144747
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.1474859714508057,
      "learning_rate": 0.000182381214475115,
      "loss": 1.9817,
      "step": 144748
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.1440021991729736,
      "learning_rate": 0.00018237745142028817,
      "loss": 2.1269,
      "step": 144749
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.1778135299682617,
      "learning_rate": 0.0001823736883873296,
      "loss": 1.9853,
      "step": 144750
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.0825444459915161,
      "learning_rate": 0.00018236992537624011,
      "loss": 2.0566,
      "step": 144751
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.1659348011016846,
      "learning_rate": 0.0001823661623870201,
      "loss": 2.2199,
      "step": 144752
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.1615374088287354,
      "learning_rate": 0.0001823623994196705,
      "loss": 1.9951,
      "step": 144753
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.0953221321105957,
      "learning_rate": 0.00018235863647419193,
      "loss": 2.2039,
      "step": 144754
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.0387860536575317,
      "learning_rate": 0.0001823548735505851,
      "loss": 2.0107,
      "step": 144755
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.092401385307312,
      "learning_rate": 0.0001823511106488507,
      "loss": 2.0209,
      "step": 144756
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.0613123178482056,
      "learning_rate": 0.0001823473477689896,
      "loss": 1.9361,
      "step": 144757
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.0332978963851929,
      "learning_rate": 0.00018234358491100224,
      "loss": 1.9854,
      "step": 144758
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.1459470987319946,
      "learning_rate": 0.00018233982207488942,
      "loss": 2.2504,
      "step": 144759
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.0579571723937988,
      "learning_rate": 0.00018233605926065184,
      "loss": 2.0419,
      "step": 144760
    },
    {
      "epoch": 1.88,
      "grad_norm": 0.9816951751708984,
      "learning_rate": 0.00018233229646829019,
      "loss": 2.0465,
      "step": 144761
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.0592259168624878,
      "learning_rate": 0.00018232853369780524,
      "loss": 2.3289,
      "step": 144762
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.1575572490692139,
      "learning_rate": 0.00018232477094919767,
      "loss": 2.1953,
      "step": 144763
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.0972235202789307,
      "learning_rate": 0.0001823210082224681,
      "loss": 1.9139,
      "step": 144764
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.1991767883300781,
      "learning_rate": 0.00018231724551761723,
      "loss": 2.2022,
      "step": 144765
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.1023098230361938,
      "learning_rate": 0.00018231348283464583,
      "loss": 1.8222,
      "step": 144766
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.1232752799987793,
      "learning_rate": 0.0001823097201735546,
      "loss": 2.1851,
      "step": 144767
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.243789553642273,
      "learning_rate": 0.00018230595753434417,
      "loss": 2.3706,
      "step": 144768
    },
    {
      "epoch": 1.88,
      "grad_norm": 1.0160446166992188,
      "learning_rate": 0.00018230219491701543,
      "loss": 2.2041,
      "step": 144769
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.1390244960784912,
      "learning_rate": 0.00018229843232156883,
      "loss": 1.9401,
      "step": 144770
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.133015513420105,
      "learning_rate": 0.0001822946697480051,
      "loss": 2.1577,
      "step": 144771
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.1695808172225952,
      "learning_rate": 0.00018229090719632508,
      "loss": 2.0732,
      "step": 144772
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.0207819938659668,
      "learning_rate": 0.0001822871446665294,
      "loss": 2.0179,
      "step": 144773
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.1304659843444824,
      "learning_rate": 0.00018228338215861873,
      "loss": 2.2545,
      "step": 144774
    },
    {
      "epoch": 1.89,
      "grad_norm": 0.9994906783103943,
      "learning_rate": 0.0001822796196725939,
      "loss": 2.0498,
      "step": 144775
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.0714961290359497,
      "learning_rate": 0.00018227585720845548,
      "loss": 1.7563,
      "step": 144776
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.0939043760299683,
      "learning_rate": 0.0001822720947662041,
      "loss": 1.8781,
      "step": 144777
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.0613349676132202,
      "learning_rate": 0.00018226833234584065,
      "loss": 1.9377,
      "step": 144778
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.0039738416671753,
      "learning_rate": 0.00018226456994736569,
      "loss": 1.9167,
      "step": 144779
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.089508295059204,
      "learning_rate": 0.00018226080757078,
      "loss": 1.9247,
      "step": 144780
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.0375064611434937,
      "learning_rate": 0.00018225704521608428,
      "loss": 1.9632,
      "step": 144781
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.2138506174087524,
      "learning_rate": 0.00018225328288327912,
      "loss": 2.2767,
      "step": 144782
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.2945374250411987,
      "learning_rate": 0.00018224952057236535,
      "loss": 2.0512,
      "step": 144783
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.0331612825393677,
      "learning_rate": 0.00018224575828334366,
      "loss": 1.9395,
      "step": 144784
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.1001288890838623,
      "learning_rate": 0.0001822419960162146,
      "loss": 2.031,
      "step": 144785
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.170188307762146,
      "learning_rate": 0.00018223823377097901,
      "loss": 2.3151,
      "step": 144786
    },
    {
      "epoch": 1.89,
      "grad_norm": 0.9626720547676086,
      "learning_rate": 0.00018223447154763764,
      "loss": 1.956,
      "step": 144787
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.0644800662994385,
      "learning_rate": 0.00018223070934619102,
      "loss": 2.0246,
      "step": 144788
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.1254897117614746,
      "learning_rate": 0.00018222694716663992,
      "loss": 2.215,
      "step": 144789
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.0257264375686646,
      "learning_rate": 0.00018222318500898504,
      "loss": 2.0134,
      "step": 144790
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.0793097019195557,
      "learning_rate": 0.0001822194228732272,
      "loss": 1.9653,
      "step": 144791
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.0679254531860352,
      "learning_rate": 0.00018221566075936693,
      "loss": 2.0197,
      "step": 144792
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.1096844673156738,
      "learning_rate": 0.00018221189866740503,
      "loss": 2.1769,
      "step": 144793
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.1739839315414429,
      "learning_rate": 0.00018220813659734214,
      "loss": 1.9905,
      "step": 144794
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.1634371280670166,
      "learning_rate": 0.0001822043745491789,
      "loss": 1.8891,
      "step": 144795
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.2112703323364258,
      "learning_rate": 0.00018220061252291616,
      "loss": 2.191,
      "step": 144796
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.0974781513214111,
      "learning_rate": 0.00018219685051855454,
      "loss": 2.165,
      "step": 144797
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.047568678855896,
      "learning_rate": 0.00018219308853609478,
      "loss": 2.1161,
      "step": 144798
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.069993019104004,
      "learning_rate": 0.0001821893265755375,
      "loss": 2.0346,
      "step": 144799
    },
    {
      "epoch": 1.89,
      "grad_norm": 0.992871105670929,
      "learning_rate": 0.00018218556463688352,
      "loss": 2.3047,
      "step": 144800
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.2730157375335693,
      "learning_rate": 0.0001821818027201334,
      "loss": 2.1387,
      "step": 144801
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.149743914604187,
      "learning_rate": 0.0001821780408252879,
      "loss": 2.4903,
      "step": 144802
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.1501296758651733,
      "learning_rate": 0.00018217427895234775,
      "loss": 1.8863,
      "step": 144803
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.1159285306930542,
      "learning_rate": 0.00018217051710131362,
      "loss": 2.1706,
      "step": 144804
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.2258731126785278,
      "learning_rate": 0.00018216675527218623,
      "loss": 1.9539,
      "step": 144805
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.0707571506500244,
      "learning_rate": 0.0001821629934649663,
      "loss": 2.0352,
      "step": 144806
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.223678469657898,
      "learning_rate": 0.00018215923167965443,
      "loss": 2.0394,
      "step": 144807
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.1409282684326172,
      "learning_rate": 0.00018215546991625137,
      "loss": 1.8063,
      "step": 144808
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.115936517715454,
      "learning_rate": 0.00018215170817475787,
      "loss": 2.0686,
      "step": 144809
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.0171024799346924,
      "learning_rate": 0.00018214794645517458,
      "loss": 2.1195,
      "step": 144810
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.227014183998108,
      "learning_rate": 0.0001821441847575022,
      "loss": 2.0981,
      "step": 144811
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.1027920246124268,
      "learning_rate": 0.00018214042308174156,
      "loss": 2.1387,
      "step": 144812
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.1043232679367065,
      "learning_rate": 0.00018213666142789315,
      "loss": 2.2055,
      "step": 144813
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.1009577512741089,
      "learning_rate": 0.0001821328997959577,
      "loss": 2.3993,
      "step": 144814
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.0698833465576172,
      "learning_rate": 0.000182129138185936,
      "loss": 2.0449,
      "step": 144815
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.1032907962799072,
      "learning_rate": 0.00018212537659782874,
      "loss": 2.0339,
      "step": 144816
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.08768892288208,
      "learning_rate": 0.00018212161503163658,
      "loss": 2.0062,
      "step": 144817
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.1137118339538574,
      "learning_rate": 0.00018211785348736035,
      "loss": 1.8097,
      "step": 144818
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.0672093629837036,
      "learning_rate": 0.0001821140919650005,
      "loss": 2.1385,
      "step": 144819
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.0878158807754517,
      "learning_rate": 0.0001821103304645579,
      "loss": 2.0946,
      "step": 144820
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.098039984703064,
      "learning_rate": 0.00018210656898603317,
      "loss": 1.9076,
      "step": 144821
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.155548095703125,
      "learning_rate": 0.00018210280752942713,
      "loss": 2.1282,
      "step": 144822
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.021365761756897,
      "learning_rate": 0.00018209904609474032,
      "loss": 2.2372,
      "step": 144823
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.162302851676941,
      "learning_rate": 0.0001820952846819737,
      "loss": 2.1581,
      "step": 144824
    },
    {
      "epoch": 1.89,
      "grad_norm": 0.9757737517356873,
      "learning_rate": 0.00018209152329112766,
      "loss": 1.9447,
      "step": 144825
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.2675238847732544,
      "learning_rate": 0.00018208776192220303,
      "loss": 2.2632,
      "step": 144826
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.0851690769195557,
      "learning_rate": 0.0001820840005752005,
      "loss": 1.9601,
      "step": 144827
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.0856961011886597,
      "learning_rate": 0.00018208023925012076,
      "loss": 2.0566,
      "step": 144828
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.2742635011672974,
      "learning_rate": 0.00018207647794696457,
      "loss": 1.9664,
      "step": 144829
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.054877519607544,
      "learning_rate": 0.0001820727166657327,
      "loss": 1.9669,
      "step": 144830
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.0921810865402222,
      "learning_rate": 0.0001820689554064256,
      "loss": 2.1679,
      "step": 144831
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.2246208190917969,
      "learning_rate": 0.00018206519416904412,
      "loss": 1.922,
      "step": 144832
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.2897270917892456,
      "learning_rate": 0.00018206143295358896,
      "loss": 1.9952,
      "step": 144833
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.152658224105835,
      "learning_rate": 0.0001820576717600608,
      "loss": 1.9083,
      "step": 144834
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.1768383979797363,
      "learning_rate": 0.00018205391058846035,
      "loss": 2.1567,
      "step": 144835
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.1577831506729126,
      "learning_rate": 0.0001820501494387884,
      "loss": 2.0992,
      "step": 144836
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.1238347291946411,
      "learning_rate": 0.00018204638831104545,
      "loss": 1.9199,
      "step": 144837
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.2740095853805542,
      "learning_rate": 0.0001820426272052323,
      "loss": 1.9358,
      "step": 144838
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.090589165687561,
      "learning_rate": 0.00018203886612134964,
      "loss": 2.13,
      "step": 144839
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.2695289850234985,
      "learning_rate": 0.0001820351050593982,
      "loss": 1.9965,
      "step": 144840
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.068068265914917,
      "learning_rate": 0.00018203134401937864,
      "loss": 2.1335,
      "step": 144841
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.1474945545196533,
      "learning_rate": 0.0001820275830012918,
      "loss": 1.9787,
      "step": 144842
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.1834882497787476,
      "learning_rate": 0.00018202382200513815,
      "loss": 2.0084,
      "step": 144843
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.2403080463409424,
      "learning_rate": 0.00018202006103091851,
      "loss": 2.0535,
      "step": 144844
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.3207908868789673,
      "learning_rate": 0.00018201630007863356,
      "loss": 1.8314,
      "step": 144845
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.0293123722076416,
      "learning_rate": 0.00018201253914828402,
      "loss": 2.1486,
      "step": 144846
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.117915153503418,
      "learning_rate": 0.00018200877823987055,
      "loss": 2.0864,
      "step": 144847
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.1281557083129883,
      "learning_rate": 0.000182005017353394,
      "loss": 2.1977,
      "step": 144848
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.194631814956665,
      "learning_rate": 0.0001820012564888548,
      "loss": 2.25,
      "step": 144849
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.0418585538864136,
      "learning_rate": 0.00018199749564625387,
      "loss": 2.1536,
      "step": 144850
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.1039263010025024,
      "learning_rate": 0.0001819937348255918,
      "loss": 2.0503,
      "step": 144851
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.093191385269165,
      "learning_rate": 0.00018198997402686927,
      "loss": 1.8674,
      "step": 144852
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.4491913318634033,
      "learning_rate": 0.00018198621325008708,
      "loss": 2.1394,
      "step": 144853
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.0973517894744873,
      "learning_rate": 0.00018198245249524602,
      "loss": 1.8837,
      "step": 144854
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.1767778396606445,
      "learning_rate": 0.00018197869176234653,
      "loss": 1.9948,
      "step": 144855
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.0287550687789917,
      "learning_rate": 0.00018197493105138936,
      "loss": 1.8419,
      "step": 144856
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.187190294265747,
      "learning_rate": 0.00018197117036237534,
      "loss": 2.0212,
      "step": 144857
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.069734811782837,
      "learning_rate": 0.00018196740969530505,
      "loss": 2.06,
      "step": 144858
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.204819917678833,
      "learning_rate": 0.0001819636490501793,
      "loss": 2.1399,
      "step": 144859
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.3181254863739014,
      "learning_rate": 0.00018195988842699878,
      "loss": 1.9751,
      "step": 144860
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.390485405921936,
      "learning_rate": 0.00018195612782576415,
      "loss": 2.1483,
      "step": 144861
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.1062194108963013,
      "learning_rate": 0.00018195236724647605,
      "loss": 2.1743,
      "step": 144862
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.166237711906433,
      "learning_rate": 0.0001819486066891352,
      "loss": 2.0491,
      "step": 144863
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.7521510124206543,
      "learning_rate": 0.00018194484615374236,
      "loss": 2.0037,
      "step": 144864
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.1881611347198486,
      "learning_rate": 0.00018194108564029818,
      "loss": 2.1007,
      "step": 144865
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.272117018699646,
      "learning_rate": 0.00018193732514880338,
      "loss": 2.0436,
      "step": 144866
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.0522538423538208,
      "learning_rate": 0.00018193356467925876,
      "loss": 2.0094,
      "step": 144867
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.1318964958190918,
      "learning_rate": 0.00018192980423166493,
      "loss": 2.2338,
      "step": 144868
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.3252885341644287,
      "learning_rate": 0.00018192604380602245,
      "loss": 2.172,
      "step": 144869
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.1159871816635132,
      "learning_rate": 0.00018192228340233217,
      "loss": 2.0299,
      "step": 144870
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.0434589385986328,
      "learning_rate": 0.0001819185230205948,
      "loss": 2.0856,
      "step": 144871
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.1030632257461548,
      "learning_rate": 0.000181914762660811,
      "loss": 2.0204,
      "step": 144872
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.1207194328308105,
      "learning_rate": 0.00018191100232298147,
      "loss": 2.0143,
      "step": 144873
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.0703123807907104,
      "learning_rate": 0.00018190724200710693,
      "loss": 1.8858,
      "step": 144874
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.017146110534668,
      "learning_rate": 0.00018190348171318808,
      "loss": 2.3912,
      "step": 144875
    },
    {
      "epoch": 1.89,
      "grad_norm": 0.9749372601509094,
      "learning_rate": 0.00018189972144122557,
      "loss": 2.0799,
      "step": 144876
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.1520283222198486,
      "learning_rate": 0.00018189596119122013,
      "loss": 2.0365,
      "step": 144877
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.1757792234420776,
      "learning_rate": 0.00018189220096317244,
      "loss": 2.1943,
      "step": 144878
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.1777417659759521,
      "learning_rate": 0.0001818884407570833,
      "loss": 2.395,
      "step": 144879
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.1248118877410889,
      "learning_rate": 0.0001818846805729532,
      "loss": 1.8944,
      "step": 144880
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.2569352388381958,
      "learning_rate": 0.00018188092041078306,
      "loss": 2.2537,
      "step": 144881
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.101159691810608,
      "learning_rate": 0.00018187716027057347,
      "loss": 2.1413,
      "step": 144882
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.019902229309082,
      "learning_rate": 0.00018187340015232516,
      "loss": 2.021,
      "step": 144883
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.08963942527771,
      "learning_rate": 0.00018186964005603877,
      "loss": 1.9329,
      "step": 144884
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.0978517532348633,
      "learning_rate": 0.00018186587998171512,
      "loss": 2.2551,
      "step": 144885
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.124799370765686,
      "learning_rate": 0.00018186211992935478,
      "loss": 1.9662,
      "step": 144886
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.1872305870056152,
      "learning_rate": 0.0001818583598989585,
      "loss": 2.2215,
      "step": 144887
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.1322966814041138,
      "learning_rate": 0.00018185459989052694,
      "loss": 2.0668,
      "step": 144888
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.1317927837371826,
      "learning_rate": 0.00018185083990406088,
      "loss": 1.998,
      "step": 144889
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.0047053098678589,
      "learning_rate": 0.00018184707993956103,
      "loss": 2.0134,
      "step": 144890
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.523897647857666,
      "learning_rate": 0.00018184331999702803,
      "loss": 2.1676,
      "step": 144891
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.0448355674743652,
      "learning_rate": 0.00018183956007646255,
      "loss": 2.3508,
      "step": 144892
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.3216118812561035,
      "learning_rate": 0.00018183580017786528,
      "loss": 2.1901,
      "step": 144893
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.1332159042358398,
      "learning_rate": 0.000181832040301237,
      "loss": 2.1801,
      "step": 144894
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.1125874519348145,
      "learning_rate": 0.00018182828044657835,
      "loss": 2.0494,
      "step": 144895
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.0885194540023804,
      "learning_rate": 0.0001818245206138901,
      "loss": 2.0527,
      "step": 144896
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.199621319770813,
      "learning_rate": 0.000181820760803173,
      "loss": 2.0091,
      "step": 144897
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.1793588399887085,
      "learning_rate": 0.00018181700101442753,
      "loss": 2.0367,
      "step": 144898
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.1496739387512207,
      "learning_rate": 0.0001818132412476545,
      "loss": 1.9253,
      "step": 144899
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.1216899156570435,
      "learning_rate": 0.0001818094815028546,
      "loss": 2.002,
      "step": 144900
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.6706652641296387,
      "learning_rate": 0.0001818057217800286,
      "loss": 2.1503,
      "step": 144901
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.090179443359375,
      "learning_rate": 0.0001818019620791771,
      "loss": 2.0403,
      "step": 144902
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.0348566770553589,
      "learning_rate": 0.000181798202400301,
      "loss": 2.1084,
      "step": 144903
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.1118186712265015,
      "learning_rate": 0.0001817944427434007,
      "loss": 1.9815,
      "step": 144904
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.2480480670928955,
      "learning_rate": 0.00018179068310847707,
      "loss": 2.0502,
      "step": 144905
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.1351912021636963,
      "learning_rate": 0.00018178692349553078,
      "loss": 1.9784,
      "step": 144906
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.2636288404464722,
      "learning_rate": 0.00018178316390456253,
      "loss": 2.2295,
      "step": 144907
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.1017472743988037,
      "learning_rate": 0.000181779404335573,
      "loss": 2.0515,
      "step": 144908
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.1114863157272339,
      "learning_rate": 0.00018177564478856305,
      "loss": 2.2084,
      "step": 144909
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.3082116842269897,
      "learning_rate": 0.0001817718852635331,
      "loss": 2.4472,
      "step": 144910
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.0653042793273926,
      "learning_rate": 0.00018176812576048402,
      "loss": 2.0784,
      "step": 144911
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.1645458936691284,
      "learning_rate": 0.00018176436627941644,
      "loss": 2.2166,
      "step": 144912
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.0301671028137207,
      "learning_rate": 0.00018176060682033115,
      "loss": 1.955,
      "step": 144913
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.18589186668396,
      "learning_rate": 0.00018175684738322873,
      "loss": 1.9711,
      "step": 144914
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.2357094287872314,
      "learning_rate": 0.0001817530879681101,
      "loss": 2.0565,
      "step": 144915
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.100718379020691,
      "learning_rate": 0.0001817493285749757,
      "loss": 2.1887,
      "step": 144916
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.0683921575546265,
      "learning_rate": 0.0001817455692038263,
      "loss": 2.1883,
      "step": 144917
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.173561453819275,
      "learning_rate": 0.00018174180985466263,
      "loss": 2.1029,
      "step": 144918
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.078600287437439,
      "learning_rate": 0.0001817380505274854,
      "loss": 2.2409,
      "step": 144919
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.1601026058197021,
      "learning_rate": 0.00018173429122229532,
      "loss": 2.1142,
      "step": 144920
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.1349412202835083,
      "learning_rate": 0.00018173053193909315,
      "loss": 2.2774,
      "step": 144921
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.1054248809814453,
      "learning_rate": 0.00018172677267787942,
      "loss": 2.2776,
      "step": 144922
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.0297883749008179,
      "learning_rate": 0.0001817230134386549,
      "loss": 2.2372,
      "step": 144923
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.191579818725586,
      "learning_rate": 0.0001817192542214203,
      "loss": 2.0368,
      "step": 144924
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.166237235069275,
      "learning_rate": 0.00018171549502617632,
      "loss": 2.1621,
      "step": 144925
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.0436691045761108,
      "learning_rate": 0.0001817117358529237,
      "loss": 2.02,
      "step": 144926
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.2061482667922974,
      "learning_rate": 0.00018170797670166315,
      "loss": 2.3002,
      "step": 144927
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.113416314125061,
      "learning_rate": 0.00018170421757239525,
      "loss": 2.2096,
      "step": 144928
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.1200109720230103,
      "learning_rate": 0.00018170045846512076,
      "loss": 2.0893,
      "step": 144929
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.1442686319351196,
      "learning_rate": 0.00018169669937984038,
      "loss": 2.2602,
      "step": 144930
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.1950340270996094,
      "learning_rate": 0.00018169294031655485,
      "loss": 2.0351,
      "step": 144931
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.2381051778793335,
      "learning_rate": 0.0001816891812752648,
      "loss": 2.1162,
      "step": 144932
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.0043675899505615,
      "learning_rate": 0.00018168542225597097,
      "loss": 1.9335,
      "step": 144933
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.1777945756912231,
      "learning_rate": 0.0001816816632586742,
      "loss": 2.314,
      "step": 144934
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.105666160583496,
      "learning_rate": 0.0001816779042833749,
      "loss": 2.0341,
      "step": 144935
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.198535442352295,
      "learning_rate": 0.0001816741453300739,
      "loss": 2.0584,
      "step": 144936
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.0951145887374878,
      "learning_rate": 0.00018167038639877193,
      "loss": 2.1354,
      "step": 144937
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.0936824083328247,
      "learning_rate": 0.00018166662748946965,
      "loss": 1.9987,
      "step": 144938
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.4362423419952393,
      "learning_rate": 0.00018166286860216775,
      "loss": 2.0125,
      "step": 144939
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.3680864572525024,
      "learning_rate": 0.00018165910973686714,
      "loss": 1.9479,
      "step": 144940
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.252764344215393,
      "learning_rate": 0.00018165535089356818,
      "loss": 1.909,
      "step": 144941
    },
    {
      "epoch": 1.89,
      "grad_norm": 0.9935891628265381,
      "learning_rate": 0.00018165159207227176,
      "loss": 2.0462,
      "step": 144942
    },
    {
      "epoch": 1.89,
      "grad_norm": 0.9700955152511597,
      "learning_rate": 0.00018164783327297848,
      "loss": 2.1211,
      "step": 144943
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.129772663116455,
      "learning_rate": 0.00018164407449568917,
      "loss": 1.8903,
      "step": 144944
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.0845049619674683,
      "learning_rate": 0.00018164031574040442,
      "loss": 2.2461,
      "step": 144945
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.0374013185501099,
      "learning_rate": 0.0001816365570071251,
      "loss": 2.1074,
      "step": 144946
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.0520448684692383,
      "learning_rate": 0.00018163279829585166,
      "loss": 1.8199,
      "step": 144947
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.1368834972381592,
      "learning_rate": 0.0001816290396065849,
      "loss": 2.2311,
      "step": 144948
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.2096271514892578,
      "learning_rate": 0.00018162528093932558,
      "loss": 2.0134,
      "step": 144949
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.188254952430725,
      "learning_rate": 0.00018162152229407428,
      "loss": 2.2509,
      "step": 144950
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.0681136846542358,
      "learning_rate": 0.00018161776367083186,
      "loss": 1.9861,
      "step": 144951
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.0519561767578125,
      "learning_rate": 0.00018161400506959893,
      "loss": 2.0945,
      "step": 144952
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.07768714427948,
      "learning_rate": 0.0001816102464903762,
      "loss": 1.9977,
      "step": 144953
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.2199736833572388,
      "learning_rate": 0.00018160648793316434,
      "loss": 2.1877,
      "step": 144954
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.1115716695785522,
      "learning_rate": 0.00018160272939796404,
      "loss": 2.2379,
      "step": 144955
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.1111961603164673,
      "learning_rate": 0.00018159897088477602,
      "loss": 2.2199,
      "step": 144956
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.0996969938278198,
      "learning_rate": 0.000181595212393601,
      "loss": 1.8287,
      "step": 144957
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.0271786451339722,
      "learning_rate": 0.00018159145392443975,
      "loss": 1.8707,
      "step": 144958
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.0336060523986816,
      "learning_rate": 0.00018158769547729276,
      "loss": 1.8362,
      "step": 144959
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.0472700595855713,
      "learning_rate": 0.00018158393705216096,
      "loss": 2.1316,
      "step": 144960
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.053027868270874,
      "learning_rate": 0.00018158017864904486,
      "loss": 2.127,
      "step": 144961
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.068463683128357,
      "learning_rate": 0.00018157642026794525,
      "loss": 2.0485,
      "step": 144962
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.1866904497146606,
      "learning_rate": 0.00018157266190886283,
      "loss": 2.0597,
      "step": 144963
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.142995834350586,
      "learning_rate": 0.00018156890357179834,
      "loss": 2.0906,
      "step": 144964
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.0333858728408813,
      "learning_rate": 0.00018156514525675236,
      "loss": 2.2239,
      "step": 144965
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.2154020071029663,
      "learning_rate": 0.00018156138696372564,
      "loss": 1.9339,
      "step": 144966
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.103203296661377,
      "learning_rate": 0.00018155762869271898,
      "loss": 2.2007,
      "step": 144967
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.0462512969970703,
      "learning_rate": 0.00018155387044373292,
      "loss": 1.7901,
      "step": 144968
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.0091108083724976,
      "learning_rate": 0.0001815501122167682,
      "loss": 1.874,
      "step": 144969
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.0932207107543945,
      "learning_rate": 0.00018154635401182567,
      "loss": 2.1593,
      "step": 144970
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.0929465293884277,
      "learning_rate": 0.0001815425958289058,
      "loss": 2.1567,
      "step": 144971
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.1323310136795044,
      "learning_rate": 0.00018153883766800944,
      "loss": 2.243,
      "step": 144972
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.0118428468704224,
      "learning_rate": 0.0001815350795291372,
      "loss": 1.9821,
      "step": 144973
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.0545881986618042,
      "learning_rate": 0.0001815313214122899,
      "loss": 2.13,
      "step": 144974
    },
    {
      "epoch": 1.89,
      "grad_norm": 0.9944582581520081,
      "learning_rate": 0.00018152756331746815,
      "loss": 2.0628,
      "step": 144975
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.1901136636734009,
      "learning_rate": 0.00018152380524467268,
      "loss": 2.0145,
      "step": 144976
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.035171627998352,
      "learning_rate": 0.00018152004719390412,
      "loss": 2.301,
      "step": 144977
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.1370964050292969,
      "learning_rate": 0.00018151628916516322,
      "loss": 2.0765,
      "step": 144978
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.0827524662017822,
      "learning_rate": 0.00018151253115845068,
      "loss": 1.9849,
      "step": 144979
    },
    {
      "epoch": 1.89,
      "grad_norm": 0.9467865228652954,
      "learning_rate": 0.00018150877317376718,
      "loss": 2.0232,
      "step": 144980
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.0634511709213257,
      "learning_rate": 0.00018150501521111345,
      "loss": 1.9185,
      "step": 144981
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.3194223642349243,
      "learning_rate": 0.0001815012572704903,
      "loss": 2.2263,
      "step": 144982
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.1063604354858398,
      "learning_rate": 0.00018149749935189818,
      "loss": 2.1176,
      "step": 144983
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.1087357997894287,
      "learning_rate": 0.00018149374145533793,
      "loss": 2.0935,
      "step": 144984
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.0358818769454956,
      "learning_rate": 0.00018148998358081018,
      "loss": 2.3304,
      "step": 144985
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.0813716650009155,
      "learning_rate": 0.00018148622572831572,
      "loss": 2.0831,
      "step": 144986
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.0821499824523926,
      "learning_rate": 0.00018148246789785518,
      "loss": 2.1072,
      "step": 144987
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.0747886896133423,
      "learning_rate": 0.00018147871008942942,
      "loss": 2.0636,
      "step": 144988
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.0888129472732544,
      "learning_rate": 0.0001814749523030389,
      "loss": 2.1506,
      "step": 144989
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.201220989227295,
      "learning_rate": 0.00018147119453868442,
      "loss": 2.2052,
      "step": 144990
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.0394493341445923,
      "learning_rate": 0.00018146743679636667,
      "loss": 2.3222,
      "step": 144991
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.186680793762207,
      "learning_rate": 0.00018146367907608636,
      "loss": 2.1719,
      "step": 144992
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.0472391843795776,
      "learning_rate": 0.0001814599213778442,
      "loss": 2.0323,
      "step": 144993
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.178809642791748,
      "learning_rate": 0.000181456163701641,
      "loss": 2.0659,
      "step": 144994
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.030164122581482,
      "learning_rate": 0.00018145240604747721,
      "loss": 2.095,
      "step": 144995
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.09652841091156,
      "learning_rate": 0.00018144864841535365,
      "loss": 2.1159,
      "step": 144996
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.3644627332687378,
      "learning_rate": 0.00018144489080527103,
      "loss": 2.1869,
      "step": 144997
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.0526634454727173,
      "learning_rate": 0.00018144113321723004,
      "loss": 2.1996,
      "step": 144998
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.1333122253417969,
      "learning_rate": 0.0001814373756512314,
      "loss": 2.142,
      "step": 144999
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.2284820079803467,
      "learning_rate": 0.00018143361810727578,
      "loss": 2.2755,
      "step": 145000
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.0109648704528809,
      "learning_rate": 0.00018142986058536394,
      "loss": 2.1101,
      "step": 145001
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.1423773765563965,
      "learning_rate": 0.00018142610308549649,
      "loss": 2.3301,
      "step": 145002
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.1379430294036865,
      "learning_rate": 0.00018142234560767414,
      "loss": 2.2063,
      "step": 145003
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.0663917064666748,
      "learning_rate": 0.00018141858815189763,
      "loss": 2.1852,
      "step": 145004
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.119110107421875,
      "learning_rate": 0.00018141483071816764,
      "loss": 2.0908,
      "step": 145005
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.102892518043518,
      "learning_rate": 0.00018141107330648488,
      "loss": 2.0553,
      "step": 145006
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.366956114768982,
      "learning_rate": 0.00018140731591685013,
      "loss": 1.8726,
      "step": 145007
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.1465353965759277,
      "learning_rate": 0.00018140355854926387,
      "loss": 2.1081,
      "step": 145008
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.0358761548995972,
      "learning_rate": 0.00018139980120372696,
      "loss": 2.0781,
      "step": 145009
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.1096289157867432,
      "learning_rate": 0.00018139604388024008,
      "loss": 2.1228,
      "step": 145010
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.0662546157836914,
      "learning_rate": 0.00018139228657880385,
      "loss": 1.8901,
      "step": 145011
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.1310304403305054,
      "learning_rate": 0.00018138852929941906,
      "loss": 2.1895,
      "step": 145012
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.0871435403823853,
      "learning_rate": 0.00018138477204208655,
      "loss": 2.0084,
      "step": 145013
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.082197666168213,
      "learning_rate": 0.00018138101480680668,
      "loss": 1.9457,
      "step": 145014
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.2754532098770142,
      "learning_rate": 0.00018137725759358032,
      "loss": 2.0124,
      "step": 145015
    },
    {
      "epoch": 1.89,
      "grad_norm": 0.9840397834777832,
      "learning_rate": 0.00018137350040240818,
      "loss": 2.3708,
      "step": 145016
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.227879524230957,
      "learning_rate": 0.00018136974323329096,
      "loss": 2.1989,
      "step": 145017
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.1370702981948853,
      "learning_rate": 0.0001813659860862293,
      "loss": 2.0108,
      "step": 145018
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.1779069900512695,
      "learning_rate": 0.00018136222896122413,
      "loss": 2.0585,
      "step": 145019
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.0520111322402954,
      "learning_rate": 0.00018135847185827577,
      "loss": 2.0338,
      "step": 145020
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.2191205024719238,
      "learning_rate": 0.00018135471477738517,
      "loss": 2.1677,
      "step": 145021
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.1862356662750244,
      "learning_rate": 0.00018135095771855292,
      "loss": 2.1077,
      "step": 145022
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.1720813512802124,
      "learning_rate": 0.00018134720068177976,
      "loss": 2.0993,
      "step": 145023
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.081627607345581,
      "learning_rate": 0.00018134344366706643,
      "loss": 2.056,
      "step": 145024
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.1793968677520752,
      "learning_rate": 0.0001813396866744137,
      "loss": 1.9521,
      "step": 145025
    },
    {
      "epoch": 1.89,
      "grad_norm": 0.9910581111907959,
      "learning_rate": 0.00018133592970382206,
      "loss": 2.1231,
      "step": 145026
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.0097309350967407,
      "learning_rate": 0.00018133217275529228,
      "loss": 1.8856,
      "step": 145027
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.0159226655960083,
      "learning_rate": 0.00018132841582882512,
      "loss": 1.9165,
      "step": 145028
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.095393180847168,
      "learning_rate": 0.00018132465892442125,
      "loss": 2.1101,
      "step": 145029
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.098962664604187,
      "learning_rate": 0.00018132090204208133,
      "loss": 2.1068,
      "step": 145030
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.1239467859268188,
      "learning_rate": 0.00018131714518180623,
      "loss": 2.3061,
      "step": 145031
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.0721877813339233,
      "learning_rate": 0.00018131338834359643,
      "loss": 2.115,
      "step": 145032
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.0847840309143066,
      "learning_rate": 0.00018130963152745267,
      "loss": 2.0806,
      "step": 145033
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.0638556480407715,
      "learning_rate": 0.0001813058747333757,
      "loss": 1.9964,
      "step": 145034
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.1581425666809082,
      "learning_rate": 0.0001813021179613662,
      "loss": 2.2493,
      "step": 145035
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.077369213104248,
      "learning_rate": 0.0001812983612114249,
      "loss": 2.1018,
      "step": 145036
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.1682255268096924,
      "learning_rate": 0.00018129460448355257,
      "loss": 2.0237,
      "step": 145037
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.1708098649978638,
      "learning_rate": 0.00018129084777774966,
      "loss": 1.9896,
      "step": 145038
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.0568506717681885,
      "learning_rate": 0.00018128709109401712,
      "loss": 2.0751,
      "step": 145039
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.1132601499557495,
      "learning_rate": 0.00018128333443235547,
      "loss": 2.157,
      "step": 145040
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.1582703590393066,
      "learning_rate": 0.0001812795777927655,
      "loss": 2.1022,
      "step": 145041
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.0669982433319092,
      "learning_rate": 0.0001812758211752479,
      "loss": 2.0119,
      "step": 145042
    },
    {
      "epoch": 1.89,
      "grad_norm": 0.9954748153686523,
      "learning_rate": 0.00018127206457980345,
      "loss": 2.028,
      "step": 145043
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.0365716218948364,
      "learning_rate": 0.0001812683080064327,
      "loss": 1.8704,
      "step": 145044
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.0520046949386597,
      "learning_rate": 0.00018126455145513646,
      "loss": 1.9896,
      "step": 145045
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.0871121883392334,
      "learning_rate": 0.00018126079492591533,
      "loss": 2.2652,
      "step": 145046
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.1568289995193481,
      "learning_rate": 0.00018125703841877008,
      "loss": 2.2575,
      "step": 145047
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.1246534585952759,
      "learning_rate": 0.00018125328193370135,
      "loss": 2.0578,
      "step": 145048
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.108477234840393,
      "learning_rate": 0.00018124952547070994,
      "loss": 2.1292,
      "step": 145049
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.0720057487487793,
      "learning_rate": 0.00018124576902979646,
      "loss": 2.1108,
      "step": 145050
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.1256163120269775,
      "learning_rate": 0.0001812420126109616,
      "loss": 2.0334,
      "step": 145051
    },
    {
      "epoch": 1.89,
      "grad_norm": 0.9787282943725586,
      "learning_rate": 0.00018123825621420617,
      "loss": 2.0116,
      "step": 145052
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.0557913780212402,
      "learning_rate": 0.0001812344998395307,
      "loss": 2.3028,
      "step": 145053
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.0428698062896729,
      "learning_rate": 0.00018123074348693602,
      "loss": 2.1124,
      "step": 145054
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.1386626958847046,
      "learning_rate": 0.00018122698715642278,
      "loss": 2.2356,
      "step": 145055
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.084450602531433,
      "learning_rate": 0.00018122323084799168,
      "loss": 1.8862,
      "step": 145056
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.0907195806503296,
      "learning_rate": 0.00018121947456164344,
      "loss": 2.1285,
      "step": 145057
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.0550341606140137,
      "learning_rate": 0.0001812157182973787,
      "loss": 1.9675,
      "step": 145058
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.0992462635040283,
      "learning_rate": 0.00018121196205519828,
      "loss": 2.0872,
      "step": 145059
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.2043179273605347,
      "learning_rate": 0.00018120820583510276,
      "loss": 1.9773,
      "step": 145060
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.0988777875900269,
      "learning_rate": 0.0001812044496370929,
      "loss": 2.0472,
      "step": 145061
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.09588623046875,
      "learning_rate": 0.0001812006934611693,
      "loss": 2.0741,
      "step": 145062
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.0268784761428833,
      "learning_rate": 0.0001811969373073328,
      "loss": 2.1111,
      "step": 145063
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.246988296508789,
      "learning_rate": 0.00018119318117558396,
      "loss": 2.2285,
      "step": 145064
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.085551142692566,
      "learning_rate": 0.0001811894250659236,
      "loss": 2.1318,
      "step": 145065
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.1007646322250366,
      "learning_rate": 0.00018118566897835242,
      "loss": 2.2298,
      "step": 145066
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.0977905988693237,
      "learning_rate": 0.000181181912912871,
      "loss": 1.9525,
      "step": 145067
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.1817803382873535,
      "learning_rate": 0.00018117815686948015,
      "loss": 2.1093,
      "step": 145068
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.0683703422546387,
      "learning_rate": 0.00018117440084818048,
      "loss": 2.0653,
      "step": 145069
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.0763015747070312,
      "learning_rate": 0.0001811706448489727,
      "loss": 2.0394,
      "step": 145070
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.0951601266860962,
      "learning_rate": 0.00018116688887185757,
      "loss": 2.0148,
      "step": 145071
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.048362374305725,
      "learning_rate": 0.00018116313291683577,
      "loss": 2.1379,
      "step": 145072
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.1188230514526367,
      "learning_rate": 0.00018115937698390798,
      "loss": 2.0465,
      "step": 145073
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.1729823350906372,
      "learning_rate": 0.00018115562107307503,
      "loss": 2.0355,
      "step": 145074
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.1050877571105957,
      "learning_rate": 0.00018115186518433734,
      "loss": 2.0413,
      "step": 145075
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.0983675718307495,
      "learning_rate": 0.0001811481093176958,
      "loss": 1.9869,
      "step": 145076
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.0843091011047363,
      "learning_rate": 0.00018114435347315106,
      "loss": 2.0563,
      "step": 145077
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.1377537250518799,
      "learning_rate": 0.00018114059765070382,
      "loss": 2.176,
      "step": 145078
    },
    {
      "epoch": 1.89,
      "grad_norm": 0.985525906085968,
      "learning_rate": 0.0001811368418503548,
      "loss": 1.9791,
      "step": 145079
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.2273974418640137,
      "learning_rate": 0.00018113308607210478,
      "loss": 2.091,
      "step": 145080
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.060912013053894,
      "learning_rate": 0.00018112933031595426,
      "loss": 2.1152,
      "step": 145081
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.0929378271102905,
      "learning_rate": 0.00018112557458190405,
      "loss": 1.9235,
      "step": 145082
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.2051992416381836,
      "learning_rate": 0.0001811218188699548,
      "loss": 2.0686,
      "step": 145083
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.272697925567627,
      "learning_rate": 0.00018111806318010727,
      "loss": 2.0733,
      "step": 145084
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.0069929361343384,
      "learning_rate": 0.00018111430751236216,
      "loss": 2.0192,
      "step": 145085
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.1617392301559448,
      "learning_rate": 0.00018111055186672024,
      "loss": 2.3517,
      "step": 145086
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.2989083528518677,
      "learning_rate": 0.00018110679624318198,
      "loss": 2.2053,
      "step": 145087
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.153487205505371,
      "learning_rate": 0.00018110304064174827,
      "loss": 2.1155,
      "step": 145088
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.1547654867172241,
      "learning_rate": 0.00018109928506241968,
      "loss": 1.9207,
      "step": 145089
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.103855848312378,
      "learning_rate": 0.000181095529505197,
      "loss": 1.9467,
      "step": 145090
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.0787160396575928,
      "learning_rate": 0.00018109177397008093,
      "loss": 1.9137,
      "step": 145091
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.0862016677856445,
      "learning_rate": 0.00018108801845707225,
      "loss": 2.0131,
      "step": 145092
    },
    {
      "epoch": 1.89,
      "grad_norm": 0.9142699241638184,
      "learning_rate": 0.0001810842629661714,
      "loss": 1.8527,
      "step": 145093
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.220145583152771,
      "learning_rate": 0.00018108050749737927,
      "loss": 2.0035,
      "step": 145094
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.1734272241592407,
      "learning_rate": 0.00018107675205069652,
      "loss": 2.0263,
      "step": 145095
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.0053097009658813,
      "learning_rate": 0.00018107299662612384,
      "loss": 2.0998,
      "step": 145096
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.0938937664031982,
      "learning_rate": 0.00018106924122366193,
      "loss": 2.0705,
      "step": 145097
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.0333369970321655,
      "learning_rate": 0.0001810654858433116,
      "loss": 2.1427,
      "step": 145098
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.1141406297683716,
      "learning_rate": 0.00018106173048507333,
      "loss": 1.9871,
      "step": 145099
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.1756322383880615,
      "learning_rate": 0.00018105797514894793,
      "loss": 2.0688,
      "step": 145100
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.0904927253723145,
      "learning_rate": 0.0001810542198349361,
      "loss": 2.0617,
      "step": 145101
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.073022723197937,
      "learning_rate": 0.00018105046454303854,
      "loss": 2.1327,
      "step": 145102
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.091779351234436,
      "learning_rate": 0.00018104670927325593,
      "loss": 2.2579,
      "step": 145103
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.1315175294876099,
      "learning_rate": 0.0001810429540255891,
      "loss": 2.0953,
      "step": 145104
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.1984772682189941,
      "learning_rate": 0.00018103919880003852,
      "loss": 2.1187,
      "step": 145105
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.1350996494293213,
      "learning_rate": 0.00018103544359660502,
      "loss": 2.1346,
      "step": 145106
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.3309341669082642,
      "learning_rate": 0.00018103168841528927,
      "loss": 2.2068,
      "step": 145107
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.0029007196426392,
      "learning_rate": 0.00018102793325609195,
      "loss": 2.2927,
      "step": 145108
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.1698307991027832,
      "learning_rate": 0.0001810241781190138,
      "loss": 2.0716,
      "step": 145109
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.0784962177276611,
      "learning_rate": 0.0001810204230040556,
      "loss": 2.0245,
      "step": 145110
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.1008424758911133,
      "learning_rate": 0.00018101666791121786,
      "loss": 2.224,
      "step": 145111
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.1206767559051514,
      "learning_rate": 0.00018101291284050136,
      "loss": 2.0823,
      "step": 145112
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.08994722366333,
      "learning_rate": 0.00018100915779190678,
      "loss": 2.2266,
      "step": 145113
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.0368547439575195,
      "learning_rate": 0.00018100540276543486,
      "loss": 1.9067,
      "step": 145114
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.09023118019104,
      "learning_rate": 0.00018100164776108628,
      "loss": 2.2682,
      "step": 145115
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.1903444528579712,
      "learning_rate": 0.00018099789277886184,
      "loss": 1.9098,
      "step": 145116
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.1666301488876343,
      "learning_rate": 0.00018099413781876207,
      "loss": 2.3083,
      "step": 145117
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.0989137887954712,
      "learning_rate": 0.00018099038288078772,
      "loss": 1.9431,
      "step": 145118
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.1544448137283325,
      "learning_rate": 0.0001809866279649395,
      "loss": 2.134,
      "step": 145119
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.1627895832061768,
      "learning_rate": 0.0001809828730712181,
      "loss": 2.1286,
      "step": 145120
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.1659458875656128,
      "learning_rate": 0.00018097911819962422,
      "loss": 1.9316,
      "step": 145121
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.1385737657546997,
      "learning_rate": 0.00018097536335015868,
      "loss": 2.1861,
      "step": 145122
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.1059781312942505,
      "learning_rate": 0.000180971608522822,
      "loss": 1.967,
      "step": 145123
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.0954586267471313,
      "learning_rate": 0.0001809678537176149,
      "loss": 2.0734,
      "step": 145124
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.3571786880493164,
      "learning_rate": 0.00018096409893453817,
      "loss": 2.2242,
      "step": 145125
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.113349437713623,
      "learning_rate": 0.0001809603441735924,
      "loss": 2.0609,
      "step": 145126
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.1457710266113281,
      "learning_rate": 0.0001809565894347784,
      "loss": 1.9455,
      "step": 145127
    },
    {
      "epoch": 1.89,
      "grad_norm": 0.9936509728431702,
      "learning_rate": 0.00018095283471809676,
      "loss": 1.8904,
      "step": 145128
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.120478868484497,
      "learning_rate": 0.0001809490800235484,
      "loss": 2.1093,
      "step": 145129
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.120158314704895,
      "learning_rate": 0.00018094532535113374,
      "loss": 2.3084,
      "step": 145130
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.0437216758728027,
      "learning_rate": 0.00018094157070085355,
      "loss": 2.2905,
      "step": 145131
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.3240385055541992,
      "learning_rate": 0.00018093781607270864,
      "loss": 2.1996,
      "step": 145132
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.2312028408050537,
      "learning_rate": 0.00018093406146669957,
      "loss": 2.2024,
      "step": 145133
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.282952070236206,
      "learning_rate": 0.00018093030688282714,
      "loss": 2.2301,
      "step": 145134
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.253067970275879,
      "learning_rate": 0.00018092655232109206,
      "loss": 2.0548,
      "step": 145135
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.0853397846221924,
      "learning_rate": 0.00018092279778149492,
      "loss": 2.3656,
      "step": 145136
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.2133134603500366,
      "learning_rate": 0.0001809190432640365,
      "loss": 2.0335,
      "step": 145137
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.2456347942352295,
      "learning_rate": 0.0001809152887687175,
      "loss": 2.0011,
      "step": 145138
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.152296781539917,
      "learning_rate": 0.00018091153429553852,
      "loss": 2.1642,
      "step": 145139
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.0735174417495728,
      "learning_rate": 0.0001809077798445004,
      "loss": 2.2341,
      "step": 145140
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.3270237445831299,
      "learning_rate": 0.0001809040254156038,
      "loss": 1.9015,
      "step": 145141
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.1190474033355713,
      "learning_rate": 0.00018090027100884935,
      "loss": 1.8908,
      "step": 145142
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.0709521770477295,
      "learning_rate": 0.00018089651662423774,
      "loss": 2.0448,
      "step": 145143
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.1166654825210571,
      "learning_rate": 0.00018089276226176983,
      "loss": 2.1076,
      "step": 145144
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.2674152851104736,
      "learning_rate": 0.00018088900792144607,
      "loss": 2.0682,
      "step": 145145
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.1528581380844116,
      "learning_rate": 0.00018088525360326736,
      "loss": 1.8268,
      "step": 145146
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.2443305253982544,
      "learning_rate": 0.00018088149930723437,
      "loss": 2.2364,
      "step": 145147
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.073264479637146,
      "learning_rate": 0.0001808777450333477,
      "loss": 2.0249,
      "step": 145148
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.357108473777771,
      "learning_rate": 0.00018087399078160808,
      "loss": 2.1417,
      "step": 145149
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.3199362754821777,
      "learning_rate": 0.00018087023655201628,
      "loss": 2.1393,
      "step": 145150
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.2940207719802856,
      "learning_rate": 0.00018086648234457296,
      "loss": 1.9066,
      "step": 145151
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.0694735050201416,
      "learning_rate": 0.0001808627281592788,
      "loss": 2.1998,
      "step": 145152
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.27375328540802,
      "learning_rate": 0.00018085897399613455,
      "loss": 2.2203,
      "step": 145153
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.3468542098999023,
      "learning_rate": 0.00018085521985514082,
      "loss": 2.1681,
      "step": 145154
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.0530834197998047,
      "learning_rate": 0.00018085146573629832,
      "loss": 1.9685,
      "step": 145155
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.0269805192947388,
      "learning_rate": 0.0001808477116396078,
      "loss": 2.0822,
      "step": 145156
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.054026484489441,
      "learning_rate": 0.00018084395756506994,
      "loss": 2.031,
      "step": 145157
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.008312702178955,
      "learning_rate": 0.0001808402035126855,
      "loss": 2.3074,
      "step": 145158
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.039589285850525,
      "learning_rate": 0.00018083644948245508,
      "loss": 1.9563,
      "step": 145159
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.5003315210342407,
      "learning_rate": 0.0001808326954743794,
      "loss": 2.156,
      "step": 145160
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.0000280141830444,
      "learning_rate": 0.00018082894148845917,
      "loss": 1.8803,
      "step": 145161
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.0307962894439697,
      "learning_rate": 0.00018082518752469506,
      "loss": 2.033,
      "step": 145162
    },
    {
      "epoch": 1.89,
      "grad_norm": 0.981147289276123,
      "learning_rate": 0.00018082143358308782,
      "loss": 2.1361,
      "step": 145163
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.1544771194458008,
      "learning_rate": 0.00018081767966363814,
      "loss": 1.9785,
      "step": 145164
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.0938241481781006,
      "learning_rate": 0.0001808139257663468,
      "loss": 1.9857,
      "step": 145165
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.2619694471359253,
      "learning_rate": 0.00018081017189121426,
      "loss": 1.915,
      "step": 145166
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.0719009637832642,
      "learning_rate": 0.00018080641803824138,
      "loss": 2.0805,
      "step": 145167
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.1535801887512207,
      "learning_rate": 0.00018080266420742884,
      "loss": 1.9139,
      "step": 145168
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.222988486289978,
      "learning_rate": 0.00018079891039877733,
      "loss": 2.2329,
      "step": 145169
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.1961017847061157,
      "learning_rate": 0.00018079515661228757,
      "loss": 2.3491,
      "step": 145170
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.1289688348770142,
      "learning_rate": 0.0001807914028479603,
      "loss": 2.1393,
      "step": 145171
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.0498676300048828,
      "learning_rate": 0.0001807876491057961,
      "loss": 1.9265,
      "step": 145172
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.0596542358398438,
      "learning_rate": 0.0001807838953857957,
      "loss": 2.0211,
      "step": 145173
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.8840283155441284,
      "learning_rate": 0.00018078014168795983,
      "loss": 1.7807,
      "step": 145174
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.1170058250427246,
      "learning_rate": 0.0001807763880122892,
      "loss": 2.1273,
      "step": 145175
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.1746948957443237,
      "learning_rate": 0.00018077263435878446,
      "loss": 2.0295,
      "step": 145176
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.22719144821167,
      "learning_rate": 0.00018076888072744647,
      "loss": 1.9942,
      "step": 145177
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.0579345226287842,
      "learning_rate": 0.0001807651271182757,
      "loss": 2.1341,
      "step": 145178
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.0727115869522095,
      "learning_rate": 0.0001807613735312729,
      "loss": 1.9692,
      "step": 145179
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.1199095249176025,
      "learning_rate": 0.00018075761996643887,
      "loss": 1.9251,
      "step": 145180
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.0936493873596191,
      "learning_rate": 0.00018075386642377422,
      "loss": 2.0754,
      "step": 145181
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.3393663167953491,
      "learning_rate": 0.00018075011290327966,
      "loss": 2.1423,
      "step": 145182
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.0605007410049438,
      "learning_rate": 0.00018074635940495604,
      "loss": 2.1224,
      "step": 145183
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.1569150686264038,
      "learning_rate": 0.0001807426059288038,
      "loss": 2.3486,
      "step": 145184
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.9265451431274414,
      "learning_rate": 0.00018073885247482375,
      "loss": 2.0168,
      "step": 145185
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.1575534343719482,
      "learning_rate": 0.0001807350990430166,
      "loss": 2.1321,
      "step": 145186
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.1393409967422485,
      "learning_rate": 0.00018073134563338305,
      "loss": 1.97,
      "step": 145187
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.296172857284546,
      "learning_rate": 0.00018072759224592382,
      "loss": 2.211,
      "step": 145188
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.1005189418792725,
      "learning_rate": 0.00018072383888063966,
      "loss": 2.2422,
      "step": 145189
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.3445240259170532,
      "learning_rate": 0.0001807200855375311,
      "loss": 2.126,
      "step": 145190
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.107182502746582,
      "learning_rate": 0.00018071633221659894,
      "loss": 2.0508,
      "step": 145191
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.0938621759414673,
      "learning_rate": 0.0001807125789178438,
      "loss": 2.1641,
      "step": 145192
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.1774065494537354,
      "learning_rate": 0.0001807088256412665,
      "loss": 1.9698,
      "step": 145193
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.1864663362503052,
      "learning_rate": 0.00018070507238686766,
      "loss": 2.2251,
      "step": 145194
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.1413012742996216,
      "learning_rate": 0.000180701319154648,
      "loss": 2.2299,
      "step": 145195
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.111825704574585,
      "learning_rate": 0.00018069756594460835,
      "loss": 2.1623,
      "step": 145196
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.032320261001587,
      "learning_rate": 0.00018069381275674916,
      "loss": 2.1651,
      "step": 145197
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.1388053894042969,
      "learning_rate": 0.0001806900595910712,
      "loss": 2.2354,
      "step": 145198
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.0974234342575073,
      "learning_rate": 0.0001806863064475753,
      "loss": 2.295,
      "step": 145199
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.170498251914978,
      "learning_rate": 0.00018068255332626197,
      "loss": 1.9515,
      "step": 145200
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.1000847816467285,
      "learning_rate": 0.00018067880022713204,
      "loss": 2.032,
      "step": 145201
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.1201817989349365,
      "learning_rate": 0.00018067504715018633,
      "loss": 2.3423,
      "step": 145202
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.1079130172729492,
      "learning_rate": 0.0001806712940954252,
      "loss": 2.016,
      "step": 145203
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.0988889932632446,
      "learning_rate": 0.00018066754106284958,
      "loss": 2.0386,
      "step": 145204
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.1560825109481812,
      "learning_rate": 0.00018066378805246008,
      "loss": 2.0407,
      "step": 145205
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.1382368803024292,
      "learning_rate": 0.00018066003506425743,
      "loss": 2.2737,
      "step": 145206
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.0356084108352661,
      "learning_rate": 0.00018065628209824235,
      "loss": 2.1505,
      "step": 145207
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.0776811838150024,
      "learning_rate": 0.00018065252915441563,
      "loss": 2.0886,
      "step": 145208
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.002327561378479,
      "learning_rate": 0.00018064877623277778,
      "loss": 1.9987,
      "step": 145209
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.1922911405563354,
      "learning_rate": 0.00018064502333332955,
      "loss": 2.1589,
      "step": 145210
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.0926460027694702,
      "learning_rate": 0.00018064127045607165,
      "loss": 1.9822,
      "step": 145211
    },
    {
      "epoch": 1.89,
      "grad_norm": 0.952496349811554,
      "learning_rate": 0.00018063751760100477,
      "loss": 1.9978,
      "step": 145212
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.2528284788131714,
      "learning_rate": 0.0001806337647681297,
      "loss": 1.954,
      "step": 145213
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.1042746305465698,
      "learning_rate": 0.00018063001195744711,
      "loss": 2.0235,
      "step": 145214
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.0358810424804688,
      "learning_rate": 0.00018062625916895756,
      "loss": 2.0926,
      "step": 145215
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.0138583183288574,
      "learning_rate": 0.0001806225064026619,
      "loss": 2.029,
      "step": 145216
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.2740811109542847,
      "learning_rate": 0.0001806187536585607,
      "loss": 1.9931,
      "step": 145217
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.1301612854003906,
      "learning_rate": 0.00018061500093665474,
      "loss": 2.0352,
      "step": 145218
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.0647562742233276,
      "learning_rate": 0.00018061124823694468,
      "loss": 2.1272,
      "step": 145219
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.0787594318389893,
      "learning_rate": 0.00018060749555943135,
      "loss": 2.0623,
      "step": 145220
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.0549389123916626,
      "learning_rate": 0.00018060374290411533,
      "loss": 2.0169,
      "step": 145221
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.0215108394622803,
      "learning_rate": 0.00018059999027099727,
      "loss": 1.8324,
      "step": 145222
    },
    {
      "epoch": 1.89,
      "grad_norm": 0.9791713356971741,
      "learning_rate": 0.00018059623766007791,
      "loss": 2.0774,
      "step": 145223
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.0976479053497314,
      "learning_rate": 0.00018059248507135798,
      "loss": 2.0512,
      "step": 145224
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.0376352071762085,
      "learning_rate": 0.00018058873250483812,
      "loss": 2.1285,
      "step": 145225
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.1002250909805298,
      "learning_rate": 0.00018058497996051916,
      "loss": 1.7427,
      "step": 145226
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.1807643175125122,
      "learning_rate": 0.00018058122743840166,
      "loss": 2.1126,
      "step": 145227
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.0532636642456055,
      "learning_rate": 0.00018057747493848634,
      "loss": 2.3452,
      "step": 145228
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.0059692859649658,
      "learning_rate": 0.00018057372246077402,
      "loss": 1.9454,
      "step": 145229
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.1218552589416504,
      "learning_rate": 0.00018056997000526518,
      "loss": 2.0433,
      "step": 145230
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.0366170406341553,
      "learning_rate": 0.0001805662175719607,
      "loss": 1.9593,
      "step": 145231
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.1288846731185913,
      "learning_rate": 0.00018056246516086122,
      "loss": 2.1285,
      "step": 145232
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.0590265989303589,
      "learning_rate": 0.00018055871277196739,
      "loss": 2.1053,
      "step": 145233
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.0003135204315186,
      "learning_rate": 0.00018055496040527994,
      "loss": 2.1028,
      "step": 145234
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.2084357738494873,
      "learning_rate": 0.00018055120806079961,
      "loss": 2.4204,
      "step": 145235
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.0803371667861938,
      "learning_rate": 0.0001805474557385271,
      "loss": 2.0546,
      "step": 145236
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.1887714862823486,
      "learning_rate": 0.00018054370343846305,
      "loss": 1.8897,
      "step": 145237
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.312949776649475,
      "learning_rate": 0.00018053995116060822,
      "loss": 2.0936,
      "step": 145238
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.075042724609375,
      "learning_rate": 0.00018053619890496316,
      "loss": 2.0365,
      "step": 145239
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.0529279708862305,
      "learning_rate": 0.00018053244667152874,
      "loss": 2.0704,
      "step": 145240
    },
    {
      "epoch": 1.89,
      "grad_norm": 0.9986661076545715,
      "learning_rate": 0.00018052869446030556,
      "loss": 2.1077,
      "step": 145241
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.1472892761230469,
      "learning_rate": 0.00018052494227129438,
      "loss": 2.1961,
      "step": 145242
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.0810132026672363,
      "learning_rate": 0.0001805211901044959,
      "loss": 1.8375,
      "step": 145243
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.286316156387329,
      "learning_rate": 0.00018051743795991082,
      "loss": 2.2234,
      "step": 145244
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.1344945430755615,
      "learning_rate": 0.00018051368583753966,
      "loss": 2.1128,
      "step": 145245
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.1462442874908447,
      "learning_rate": 0.00018050993373738337,
      "loss": 1.8352,
      "step": 145246
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.2209609746932983,
      "learning_rate": 0.00018050618165944246,
      "loss": 1.9773,
      "step": 145247
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.069727897644043,
      "learning_rate": 0.00018050242960371773,
      "loss": 2.0294,
      "step": 145248
    },
    {
      "epoch": 1.89,
      "grad_norm": 0.9997275471687317,
      "learning_rate": 0.0001804986775702099,
      "loss": 1.9844,
      "step": 145249
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.000857949256897,
      "learning_rate": 0.00018049492555891966,
      "loss": 2.1308,
      "step": 145250
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.1231547594070435,
      "learning_rate": 0.00018049117356984759,
      "loss": 2.0668,
      "step": 145251
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.2687020301818848,
      "learning_rate": 0.00018048742160299444,
      "loss": 2.1881,
      "step": 145252
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.1719530820846558,
      "learning_rate": 0.00018048366965836096,
      "loss": 2.1328,
      "step": 145253
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.1271604299545288,
      "learning_rate": 0.00018047991773594786,
      "loss": 1.8528,
      "step": 145254
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.2895904779434204,
      "learning_rate": 0.00018047616583575572,
      "loss": 2.0643,
      "step": 145255
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.5201234817504883,
      "learning_rate": 0.0001804724139577855,
      "loss": 2.0799,
      "step": 145256
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.119036316871643,
      "learning_rate": 0.00018046866210203756,
      "loss": 2.4083,
      "step": 145257
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.0260921716690063,
      "learning_rate": 0.00018046491026851277,
      "loss": 2.0328,
      "step": 145258
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.1774132251739502,
      "learning_rate": 0.00018046115845721182,
      "loss": 1.9188,
      "step": 145259
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.3520933389663696,
      "learning_rate": 0.00018045740666813536,
      "loss": 1.9118,
      "step": 145260
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.1451352834701538,
      "learning_rate": 0.00018045365490128418,
      "loss": 1.966,
      "step": 145261
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.0213897228240967,
      "learning_rate": 0.00018044990315665887,
      "loss": 2.1123,
      "step": 145262
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.1353671550750732,
      "learning_rate": 0.00018044615143426033,
      "loss": 2.1178,
      "step": 145263
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.0163238048553467,
      "learning_rate": 0.00018044239973408893,
      "loss": 2.0147,
      "step": 145264
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.1159871816635132,
      "learning_rate": 0.00018043864805614564,
      "loss": 1.9547,
      "step": 145265
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.0729385614395142,
      "learning_rate": 0.000180434896400431,
      "loss": 1.9811,
      "step": 145266
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.1090178489685059,
      "learning_rate": 0.00018043114476694577,
      "loss": 1.9448,
      "step": 145267
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.162576675415039,
      "learning_rate": 0.0001804273931556907,
      "loss": 2.1848,
      "step": 145268
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.217328667640686,
      "learning_rate": 0.00018042364156666647,
      "loss": 2.1425,
      "step": 145269
    },
    {
      "epoch": 1.89,
      "grad_norm": 0.9485576748847961,
      "learning_rate": 0.00018041988999987368,
      "loss": 1.9233,
      "step": 145270
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.083241581916809,
      "learning_rate": 0.00018041613845531308,
      "loss": 2.2464,
      "step": 145271
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.2492848634719849,
      "learning_rate": 0.00018041238693298537,
      "loss": 2.0199,
      "step": 145272
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.1028348207473755,
      "learning_rate": 0.00018040863543289127,
      "loss": 2.1649,
      "step": 145273
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.1892675161361694,
      "learning_rate": 0.00018040488395503144,
      "loss": 1.9039,
      "step": 145274
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.4116566181182861,
      "learning_rate": 0.00018040113249940672,
      "loss": 2.1094,
      "step": 145275
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.4186880588531494,
      "learning_rate": 0.00018039738106601755,
      "loss": 2.2379,
      "step": 145276
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.0941922664642334,
      "learning_rate": 0.00018039362965486483,
      "loss": 2.0544,
      "step": 145277
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.0779165029525757,
      "learning_rate": 0.00018038987826594913,
      "loss": 2.1811,
      "step": 145278
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.515911340713501,
      "learning_rate": 0.00018038612689927122,
      "loss": 2.0364,
      "step": 145279
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.349501371383667,
      "learning_rate": 0.00018038237555483182,
      "loss": 1.9876,
      "step": 145280
    },
    {
      "epoch": 1.89,
      "grad_norm": 0.9710591435432434,
      "learning_rate": 0.00018037862423263168,
      "loss": 1.9721,
      "step": 145281
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.2842044830322266,
      "learning_rate": 0.00018037487293267132,
      "loss": 2.0795,
      "step": 145282
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.2368954420089722,
      "learning_rate": 0.0001803711216549515,
      "loss": 1.8102,
      "step": 145283
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.3144437074661255,
      "learning_rate": 0.00018036737039947296,
      "loss": 1.996,
      "step": 145284
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.1182523965835571,
      "learning_rate": 0.0001803636191662364,
      "loss": 2.2267,
      "step": 145285
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.0909074544906616,
      "learning_rate": 0.00018035986795524248,
      "loss": 2.0694,
      "step": 145286
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.0918797254562378,
      "learning_rate": 0.00018035611676649202,
      "loss": 2.1653,
      "step": 145287
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.0816576480865479,
      "learning_rate": 0.00018035236559998553,
      "loss": 2.0474,
      "step": 145288
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.1488252878189087,
      "learning_rate": 0.00018034861445572378,
      "loss": 2.2528,
      "step": 145289
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.1164170503616333,
      "learning_rate": 0.00018034486333370754,
      "loss": 1.9969,
      "step": 145290
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.1286393404006958,
      "learning_rate": 0.00018034111223393735,
      "loss": 1.9682,
      "step": 145291
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.129662036895752,
      "learning_rate": 0.00018033736115641405,
      "loss": 2.1606,
      "step": 145292
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.0503246784210205,
      "learning_rate": 0.00018033361010113844,
      "loss": 1.9493,
      "step": 145293
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.113847017288208,
      "learning_rate": 0.0001803298590681109,
      "loss": 2.1136,
      "step": 145294
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.050001859664917,
      "learning_rate": 0.00018032610805733234,
      "loss": 2.3902,
      "step": 145295
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.0767254829406738,
      "learning_rate": 0.00018032235706880338,
      "loss": 2.1028,
      "step": 145296
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.3107571601867676,
      "learning_rate": 0.0001803186061025248,
      "loss": 1.9788,
      "step": 145297
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.3897311687469482,
      "learning_rate": 0.0001803148551584972,
      "loss": 2.0815,
      "step": 145298
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.1588321924209595,
      "learning_rate": 0.0001803111042367215,
      "loss": 2.0812,
      "step": 145299
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.1266809701919556,
      "learning_rate": 0.00018030735333719806,
      "loss": 2.056,
      "step": 145300
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.1011160612106323,
      "learning_rate": 0.00018030360245992774,
      "loss": 1.9016,
      "step": 145301
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.5005815029144287,
      "learning_rate": 0.0001802998516049113,
      "loss": 2.0549,
      "step": 145302
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.1206982135772705,
      "learning_rate": 0.00018029610077214932,
      "loss": 1.9952,
      "step": 145303
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.2492848634719849,
      "learning_rate": 0.00018029234996164257,
      "loss": 2.3314,
      "step": 145304
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.1516644954681396,
      "learning_rate": 0.00018028859917339178,
      "loss": 2.0913,
      "step": 145305
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.4176846742630005,
      "learning_rate": 0.0001802848484073976,
      "loss": 2.1021,
      "step": 145306
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.541383981704712,
      "learning_rate": 0.00018028109766366068,
      "loss": 2.0191,
      "step": 145307
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.0060960054397583,
      "learning_rate": 0.00018027734694218175,
      "loss": 2.3958,
      "step": 145308
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.03900146484375,
      "learning_rate": 0.00018027359624296152,
      "loss": 1.9211,
      "step": 145309
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.150869369506836,
      "learning_rate": 0.00018026984556600073,
      "loss": 2.1692,
      "step": 145310
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.3160791397094727,
      "learning_rate": 0.00018026609491130004,
      "loss": 2.1941,
      "step": 145311
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.1824711561203003,
      "learning_rate": 0.0001802623442788601,
      "loss": 2.2121,
      "step": 145312
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.0136042833328247,
      "learning_rate": 0.0001802585936686817,
      "loss": 2.3465,
      "step": 145313
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.054003119468689,
      "learning_rate": 0.00018025484308076546,
      "loss": 2.1771,
      "step": 145314
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.083850383758545,
      "learning_rate": 0.0001802510925151121,
      "loss": 1.9818,
      "step": 145315
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.3360956907272339,
      "learning_rate": 0.00018024734197172234,
      "loss": 2.1847,
      "step": 145316
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.1182630062103271,
      "learning_rate": 0.0001802435914505969,
      "loss": 2.1288,
      "step": 145317
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.0545284748077393,
      "learning_rate": 0.00018023984095173637,
      "loss": 1.9994,
      "step": 145318
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.13679039478302,
      "learning_rate": 0.0001802360904751415,
      "loss": 2.0795,
      "step": 145319
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.2486399412155151,
      "learning_rate": 0.00018023234002081308,
      "loss": 1.7989,
      "step": 145320
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.1567453145980835,
      "learning_rate": 0.00018022858958875165,
      "loss": 2.2841,
      "step": 145321
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.0824414491653442,
      "learning_rate": 0.000180224839178958,
      "loss": 2.0654,
      "step": 145322
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.1066924333572388,
      "learning_rate": 0.00018022108879143286,
      "loss": 2.1366,
      "step": 145323
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.1426572799682617,
      "learning_rate": 0.00018021733842617685,
      "loss": 2.2071,
      "step": 145324
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.1058567762374878,
      "learning_rate": 0.00018021358808319068,
      "loss": 2.2157,
      "step": 145325
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.0919359922409058,
      "learning_rate": 0.00018020983776247508,
      "loss": 2.1447,
      "step": 145326
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.1340019702911377,
      "learning_rate": 0.0001802060874640307,
      "loss": 2.3335,
      "step": 145327
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.0761240720748901,
      "learning_rate": 0.00018020233718785836,
      "loss": 1.9744,
      "step": 145328
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.170176386833191,
      "learning_rate": 0.00018019858693395861,
      "loss": 2.0644,
      "step": 145329
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.0740487575531006,
      "learning_rate": 0.00018019483670233223,
      "loss": 1.7191,
      "step": 145330
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.047360897064209,
      "learning_rate": 0.00018019108649297986,
      "loss": 2.2069,
      "step": 145331
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.0919826030731201,
      "learning_rate": 0.00018018733630590224,
      "loss": 2.1248,
      "step": 145332
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.1087511777877808,
      "learning_rate": 0.00018018358614110003,
      "loss": 2.0644,
      "step": 145333
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.1417497396469116,
      "learning_rate": 0.00018017983599857395,
      "loss": 2.0558,
      "step": 145334
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.135335087776184,
      "learning_rate": 0.0001801760858783248,
      "loss": 1.7574,
      "step": 145335
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.0300793647766113,
      "learning_rate": 0.00018017233578035313,
      "loss": 2.244,
      "step": 145336
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.1042674779891968,
      "learning_rate": 0.00018016858570465963,
      "loss": 2.3121,
      "step": 145337
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.4875322580337524,
      "learning_rate": 0.00018016483565124508,
      "loss": 2.2458,
      "step": 145338
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.4084380865097046,
      "learning_rate": 0.00018016108562011015,
      "loss": 2.1377,
      "step": 145339
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.1723499298095703,
      "learning_rate": 0.0001801573356112555,
      "loss": 2.0863,
      "step": 145340
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.0770821571350098,
      "learning_rate": 0.0001801535856246819,
      "loss": 1.956,
      "step": 145341
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.172613263130188,
      "learning_rate": 0.00018014983566039006,
      "loss": 2.2606,
      "step": 145342
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.0801498889923096,
      "learning_rate": 0.0001801460857183806,
      "loss": 2.073,
      "step": 145343
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.4549684524536133,
      "learning_rate": 0.00018014233579865418,
      "loss": 1.8971,
      "step": 145344
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.0121508836746216,
      "learning_rate": 0.0001801385859012116,
      "loss": 1.9151,
      "step": 145345
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.1673731803894043,
      "learning_rate": 0.00018013483602605351,
      "loss": 2.0837,
      "step": 145346
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.2175230979919434,
      "learning_rate": 0.00018013108617318064,
      "loss": 2.1746,
      "step": 145347
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.0675466060638428,
      "learning_rate": 0.00018012733634259376,
      "loss": 1.8685,
      "step": 145348
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.0349057912826538,
      "learning_rate": 0.00018012358653429335,
      "loss": 2.2948,
      "step": 145349
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.096245288848877,
      "learning_rate": 0.0001801198367482802,
      "loss": 1.8883,
      "step": 145350
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.0974960327148438,
      "learning_rate": 0.00018011608698455505,
      "loss": 2.1509,
      "step": 145351
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.2770042419433594,
      "learning_rate": 0.00018011233724311862,
      "loss": 2.1337,
      "step": 145352
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.3303005695343018,
      "learning_rate": 0.00018010858752397154,
      "loss": 2.0982,
      "step": 145353
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.3363986015319824,
      "learning_rate": 0.00018010483782711465,
      "loss": 2.0964,
      "step": 145354
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.1381828784942627,
      "learning_rate": 0.00018010108815254844,
      "loss": 1.9384,
      "step": 145355
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.0365722179412842,
      "learning_rate": 0.0001800973385002737,
      "loss": 2.0694,
      "step": 145356
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.1224948167800903,
      "learning_rate": 0.00018009358887029108,
      "loss": 2.0871,
      "step": 145357
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.1899693012237549,
      "learning_rate": 0.00018008983926260137,
      "loss": 2.1589,
      "step": 145358
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.094211220741272,
      "learning_rate": 0.00018008608967720523,
      "loss": 1.8103,
      "step": 145359
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.2321876287460327,
      "learning_rate": 0.00018008234011410344,
      "loss": 2.1242,
      "step": 145360
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.0482722520828247,
      "learning_rate": 0.00018007859057329653,
      "loss": 2.0664,
      "step": 145361
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.3683688640594482,
      "learning_rate": 0.0001800748410547852,
      "loss": 2.1439,
      "step": 145362
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.1883854866027832,
      "learning_rate": 0.00018007109155857026,
      "loss": 2.05,
      "step": 145363
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.2050578594207764,
      "learning_rate": 0.0001800673420846524,
      "loss": 2.178,
      "step": 145364
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.2813502550125122,
      "learning_rate": 0.00018006359263303225,
      "loss": 2.0201,
      "step": 145365
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.1673266887664795,
      "learning_rate": 0.00018005984320371065,
      "loss": 2.1742,
      "step": 145366
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.190385341644287,
      "learning_rate": 0.0001800560937966881,
      "loss": 2.0147,
      "step": 145367
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.0795072317123413,
      "learning_rate": 0.00018005234441196535,
      "loss": 2.112,
      "step": 145368
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.1285977363586426,
      "learning_rate": 0.00018004859504954318,
      "loss": 1.8899,
      "step": 145369
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.1400781869888306,
      "learning_rate": 0.0001800448457094222,
      "loss": 2.0828,
      "step": 145370
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.1717095375061035,
      "learning_rate": 0.00018004109639160317,
      "loss": 2.197,
      "step": 145371
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.2133498191833496,
      "learning_rate": 0.0001800373470960869,
      "loss": 2.1486,
      "step": 145372
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.3660194873809814,
      "learning_rate": 0.0001800335978228738,
      "loss": 1.9127,
      "step": 145373
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.3987104892730713,
      "learning_rate": 0.00018002984857196474,
      "loss": 2.011,
      "step": 145374
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.4735705852508545,
      "learning_rate": 0.00018002609934336038,
      "loss": 2.1266,
      "step": 145375
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.2684975862503052,
      "learning_rate": 0.00018002235013706144,
      "loss": 2.1525,
      "step": 145376
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.3155455589294434,
      "learning_rate": 0.0001800186009530686,
      "loss": 1.9934,
      "step": 145377
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.956826090812683,
      "learning_rate": 0.0001800148517913827,
      "loss": 2.225,
      "step": 145378
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.332816481590271,
      "learning_rate": 0.0001800111026520042,
      "loss": 1.9233,
      "step": 145379
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.1730780601501465,
      "learning_rate": 0.00018000735353493384,
      "loss": 2.162,
      "step": 145380
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.2486833333969116,
      "learning_rate": 0.00018000360444017242,
      "loss": 2.1738,
      "step": 145381
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.5068532228469849,
      "learning_rate": 0.00017999985536772058,
      "loss": 2.3133,
      "step": 145382
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.235792875289917,
      "learning_rate": 0.00017999610631757906,
      "loss": 2.1959,
      "step": 145383
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.3399449586868286,
      "learning_rate": 0.00017999235728974864,
      "loss": 2.099,
      "step": 145384
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.119446873664856,
      "learning_rate": 0.00017998860828422977,
      "loss": 1.8056,
      "step": 145385
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.5181396007537842,
      "learning_rate": 0.0001799848593010233,
      "loss": 1.9471,
      "step": 145386
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.4275742769241333,
      "learning_rate": 0.00017998111034012992,
      "loss": 2.2932,
      "step": 145387
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.6646908521652222,
      "learning_rate": 0.0001799773614015503,
      "loss": 2.1051,
      "step": 145388
    },
    {
      "epoch": 1.89,
      "grad_norm": 0.9736247062683105,
      "learning_rate": 0.0001799736124852851,
      "loss": 2.0177,
      "step": 145389
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.0937719345092773,
      "learning_rate": 0.00017996986359133525,
      "loss": 2.2475,
      "step": 145390
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.596142053604126,
      "learning_rate": 0.0001799661147197012,
      "loss": 2.1364,
      "step": 145391
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.5716575384140015,
      "learning_rate": 0.00017996236587038364,
      "loss": 2.1304,
      "step": 145392
    },
    {
      "epoch": 1.89,
      "grad_norm": 0.9860959649085999,
      "learning_rate": 0.0001799586170433834,
      "loss": 1.8112,
      "step": 145393
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.103147029876709,
      "learning_rate": 0.00017995486823870103,
      "loss": 2.2327,
      "step": 145394
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.4340776205062866,
      "learning_rate": 0.00017995111945633737,
      "loss": 2.1192,
      "step": 145395
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.2185885906219482,
      "learning_rate": 0.00017994737069629307,
      "loss": 2.1628,
      "step": 145396
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.1409658193588257,
      "learning_rate": 0.00017994362195856888,
      "loss": 2.2919,
      "step": 145397
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.134181022644043,
      "learning_rate": 0.0001799398732431654,
      "loss": 1.9686,
      "step": 145398
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.0771303176879883,
      "learning_rate": 0.00017993612455008336,
      "loss": 2.2625,
      "step": 145399
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.17111074924469,
      "learning_rate": 0.00017993237587932343,
      "loss": 2.1836,
      "step": 145400
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.194667100906372,
      "learning_rate": 0.0001799286272308863,
      "loss": 2.2175,
      "step": 145401
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.2281665802001953,
      "learning_rate": 0.0001799248786047728,
      "loss": 2.1286,
      "step": 145402
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.1324260234832764,
      "learning_rate": 0.00017992113000098352,
      "loss": 2.1808,
      "step": 145403
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.0994309186935425,
      "learning_rate": 0.0001799173814195191,
      "loss": 2.0575,
      "step": 145404
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.082679033279419,
      "learning_rate": 0.00017991363286038042,
      "loss": 2.0377,
      "step": 145405
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.311508059501648,
      "learning_rate": 0.00017990988432356795,
      "loss": 2.1197,
      "step": 145406
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.101088047027588,
      "learning_rate": 0.0001799061358090825,
      "loss": 2.0065,
      "step": 145407
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.0632988214492798,
      "learning_rate": 0.0001799023873169248,
      "loss": 1.9768,
      "step": 145408
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.086761713027954,
      "learning_rate": 0.00017989863884709556,
      "loss": 2.2286,
      "step": 145409
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.4305132627487183,
      "learning_rate": 0.00017989489039959536,
      "loss": 2.2022,
      "step": 145410
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.1629657745361328,
      "learning_rate": 0.00017989114197442498,
      "loss": 1.9837,
      "step": 145411
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.1176559925079346,
      "learning_rate": 0.0001798873935715852,
      "loss": 1.7484,
      "step": 145412
    },
    {
      "epoch": 1.89,
      "grad_norm": 0.9731215238571167,
      "learning_rate": 0.00017988364519107647,
      "loss": 1.9949,
      "step": 145413
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.1722453832626343,
      "learning_rate": 0.0001798798968328997,
      "loss": 2.1661,
      "step": 145414
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.1029610633850098,
      "learning_rate": 0.0001798761484970556,
      "loss": 1.9435,
      "step": 145415
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.0826923847198486,
      "learning_rate": 0.00017987240018354467,
      "loss": 2.0387,
      "step": 145416
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.2258615493774414,
      "learning_rate": 0.0001798686518923678,
      "loss": 1.9652,
      "step": 145417
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.1024366617202759,
      "learning_rate": 0.0001798649036235256,
      "loss": 2.0302,
      "step": 145418
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.0832459926605225,
      "learning_rate": 0.00017986115537701875,
      "loss": 1.9752,
      "step": 145419
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.0795973539352417,
      "learning_rate": 0.00017985740715284804,
      "loss": 2.0177,
      "step": 145420
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.0825480222702026,
      "learning_rate": 0.00017985365895101414,
      "loss": 1.9871,
      "step": 145421
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.037048101425171,
      "learning_rate": 0.00017984991077151763,
      "loss": 2.1788,
      "step": 145422
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.1856763362884521,
      "learning_rate": 0.00017984616261435923,
      "loss": 2.0527,
      "step": 145423
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.0437203645706177,
      "learning_rate": 0.0001798424144795398,
      "loss": 1.8691,
      "step": 145424
    },
    {
      "epoch": 1.89,
      "grad_norm": 0.9952649474143982,
      "learning_rate": 0.00017983866636705987,
      "loss": 2.0411,
      "step": 145425
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.180130958557129,
      "learning_rate": 0.00017983491827692021,
      "loss": 2.0772,
      "step": 145426
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.1299262046813965,
      "learning_rate": 0.00017983117020912162,
      "loss": 2.2158,
      "step": 145427
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.0196160078048706,
      "learning_rate": 0.0001798274221636646,
      "loss": 2.181,
      "step": 145428
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.2787164449691772,
      "learning_rate": 0.0001798236741405499,
      "loss": 1.9859,
      "step": 145429
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.1758917570114136,
      "learning_rate": 0.0001798199261397783,
      "loss": 2.0627,
      "step": 145430
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.0799357891082764,
      "learning_rate": 0.00017981617816135038,
      "loss": 1.9506,
      "step": 145431
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.056240439414978,
      "learning_rate": 0.00017981243020526692,
      "loss": 2.1565,
      "step": 145432
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.1803220510482788,
      "learning_rate": 0.0001798086822715287,
      "loss": 2.1172,
      "step": 145433
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.0928655862808228,
      "learning_rate": 0.00017980493436013626,
      "loss": 2.1564,
      "step": 145434
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.037715196609497,
      "learning_rate": 0.00017980118647109027,
      "loss": 2.1615,
      "step": 145435
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.1086570024490356,
      "learning_rate": 0.00017979743860439156,
      "loss": 2.1049,
      "step": 145436
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.2325439453125,
      "learning_rate": 0.00017979369076004075,
      "loss": 2.2568,
      "step": 145437
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.1104116439819336,
      "learning_rate": 0.0001797899429380386,
      "loss": 2.1127,
      "step": 145438
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.1030288934707642,
      "learning_rate": 0.00017978619513838587,
      "loss": 2.0438,
      "step": 145439
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.2204883098602295,
      "learning_rate": 0.00017978244736108301,
      "loss": 2.1582,
      "step": 145440
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.0176440477371216,
      "learning_rate": 0.00017977869960613088,
      "loss": 2.0651,
      "step": 145441
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.0827298164367676,
      "learning_rate": 0.00017977495187353017,
      "loss": 2.0993,
      "step": 145442
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.1569409370422363,
      "learning_rate": 0.00017977120416328153,
      "loss": 2.1559,
      "step": 145443
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.074792742729187,
      "learning_rate": 0.0001797674564753857,
      "loss": 2.0556,
      "step": 145444
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.1391187906265259,
      "learning_rate": 0.00017976370880984355,
      "loss": 2.0451,
      "step": 145445
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.0159788131713867,
      "learning_rate": 0.00017975996116665542,
      "loss": 2.0179,
      "step": 145446
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.0409120321273804,
      "learning_rate": 0.00017975621354582216,
      "loss": 2.0281,
      "step": 145447
    },
    {
      "epoch": 1.89,
      "grad_norm": 0.9877482652664185,
      "learning_rate": 0.00017975246594734455,
      "loss": 2.2686,
      "step": 145448
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.1125268936157227,
      "learning_rate": 0.0001797487183712232,
      "loss": 2.2234,
      "step": 145449
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.0965838432312012,
      "learning_rate": 0.00017974497081745883,
      "loss": 2.3497,
      "step": 145450
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.0801887512207031,
      "learning_rate": 0.0001797412232860523,
      "loss": 2.1629,
      "step": 145451
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.0486005544662476,
      "learning_rate": 0.000179737475777004,
      "loss": 2.2506,
      "step": 145452
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.0508655309677124,
      "learning_rate": 0.00017973372829031473,
      "loss": 1.8708,
      "step": 145453
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.084757685661316,
      "learning_rate": 0.00017972998082598526,
      "loss": 2.0444,
      "step": 145454
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.1602354049682617,
      "learning_rate": 0.00017972623338401625,
      "loss": 2.2978,
      "step": 145455
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.1408690214157104,
      "learning_rate": 0.00017972248596440844,
      "loss": 2.16,
      "step": 145456
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.1029266119003296,
      "learning_rate": 0.00017971873856716258,
      "loss": 2.3689,
      "step": 145457
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.0985817909240723,
      "learning_rate": 0.0001797149911922792,
      "loss": 2.1629,
      "step": 145458
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.0055707693099976,
      "learning_rate": 0.000179711243839759,
      "loss": 2.0282,
      "step": 145459
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.1717586517333984,
      "learning_rate": 0.00017970749650960278,
      "loss": 2.3267,
      "step": 145460
    },
    {
      "epoch": 1.89,
      "grad_norm": 0.9536157250404358,
      "learning_rate": 0.0001797037492018112,
      "loss": 2.0912,
      "step": 145461
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.036630392074585,
      "learning_rate": 0.000179700001916385,
      "loss": 2.1844,
      "step": 145462
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.1412090063095093,
      "learning_rate": 0.00017969625465332484,
      "loss": 2.1408,
      "step": 145463
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.016531229019165,
      "learning_rate": 0.0001796925074126315,
      "loss": 2.094,
      "step": 145464
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.1054010391235352,
      "learning_rate": 0.00017968876019430546,
      "loss": 1.9692,
      "step": 145465
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.030681848526001,
      "learning_rate": 0.0001796850129983476,
      "loss": 1.9513,
      "step": 145466
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.131381630897522,
      "learning_rate": 0.00017968126582475856,
      "loss": 2.029,
      "step": 145467
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.1292498111724854,
      "learning_rate": 0.000179677518673539,
      "loss": 1.9551,
      "step": 145468
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.1574463844299316,
      "learning_rate": 0.00017967377154468975,
      "loss": 1.9342,
      "step": 145469
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.2560228109359741,
      "learning_rate": 0.00017967002443821143,
      "loss": 2.138,
      "step": 145470
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.058427333831787,
      "learning_rate": 0.00017966627735410467,
      "loss": 2.0357,
      "step": 145471
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.1862043142318726,
      "learning_rate": 0.0001796625302923702,
      "loss": 2.1925,
      "step": 145472
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.107578158378601,
      "learning_rate": 0.00017965878325300875,
      "loss": 2.2507,
      "step": 145473
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.2665354013442993,
      "learning_rate": 0.000179655036236021,
      "loss": 2.1369,
      "step": 145474
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.0606704950332642,
      "learning_rate": 0.00017965128924140762,
      "loss": 2.0214,
      "step": 145475
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.2321027517318726,
      "learning_rate": 0.0001796475422691695,
      "loss": 2.1859,
      "step": 145476
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.1270666122436523,
      "learning_rate": 0.00017964379531930702,
      "loss": 2.1781,
      "step": 145477
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.063402771949768,
      "learning_rate": 0.00017964004839182108,
      "loss": 1.9524,
      "step": 145478
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.050920844078064,
      "learning_rate": 0.00017963630148671228,
      "loss": 1.9244,
      "step": 145479
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.1012736558914185,
      "learning_rate": 0.00017963255460398133,
      "loss": 2.0956,
      "step": 145480
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.1273590326309204,
      "learning_rate": 0.00017962880774362902,
      "loss": 2.0623,
      "step": 145481
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.216394305229187,
      "learning_rate": 0.0001796250609056561,
      "loss": 2.0722,
      "step": 145482
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.0961461067199707,
      "learning_rate": 0.00017962131409006302,
      "loss": 1.9872,
      "step": 145483
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.0580739974975586,
      "learning_rate": 0.00017961756729685063,
      "loss": 2.0417,
      "step": 145484
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.4694509506225586,
      "learning_rate": 0.0001796138205260196,
      "loss": 1.9694,
      "step": 145485
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.4230928421020508,
      "learning_rate": 0.00017961007377757064,
      "loss": 1.9103,
      "step": 145486
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.053295373916626,
      "learning_rate": 0.0001796063270515044,
      "loss": 2.2581,
      "step": 145487
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.1099096536636353,
      "learning_rate": 0.00017960258034782173,
      "loss": 2.1402,
      "step": 145488
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.1911489963531494,
      "learning_rate": 0.00017959883366652313,
      "loss": 2.1973,
      "step": 145489
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.2076765298843384,
      "learning_rate": 0.0001795950870076094,
      "loss": 1.9471,
      "step": 145490
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.1610316038131714,
      "learning_rate": 0.0001795913403710812,
      "loss": 2.2775,
      "step": 145491
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.1455645561218262,
      "learning_rate": 0.00017958759375693925,
      "loss": 1.9305,
      "step": 145492
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.0470983982086182,
      "learning_rate": 0.00017958384716518423,
      "loss": 1.7876,
      "step": 145493
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.6035205125808716,
      "learning_rate": 0.00017958010059581687,
      "loss": 2.1398,
      "step": 145494
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.3396753072738647,
      "learning_rate": 0.0001795763540488378,
      "loss": 2.1409,
      "step": 145495
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.0460948944091797,
      "learning_rate": 0.00017957260752424778,
      "loss": 2.168,
      "step": 145496
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.1747883558273315,
      "learning_rate": 0.00017956886102204756,
      "loss": 1.7382,
      "step": 145497
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.220934271812439,
      "learning_rate": 0.00017956511454223767,
      "loss": 2.3157,
      "step": 145498
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.192295789718628,
      "learning_rate": 0.0001795613680848189,
      "loss": 2.0678,
      "step": 145499
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.0671393871307373,
      "learning_rate": 0.00017955762164979203,
      "loss": 1.9949,
      "step": 145500
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.1409122943878174,
      "learning_rate": 0.00017955387523715756,
      "loss": 2.0054,
      "step": 145501
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.0875247716903687,
      "learning_rate": 0.00017955012884691632,
      "loss": 1.9411,
      "step": 145502
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.196653962135315,
      "learning_rate": 0.00017954638247906905,
      "loss": 1.8943,
      "step": 145503
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.102195382118225,
      "learning_rate": 0.00017954263613361636,
      "loss": 2.0884,
      "step": 145504
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.16864812374115,
      "learning_rate": 0.00017953888981055895,
      "loss": 2.2248,
      "step": 145505
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.1644223928451538,
      "learning_rate": 0.0001795351435098976,
      "loss": 2.1892,
      "step": 145506
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.0466268062591553,
      "learning_rate": 0.0001795313972316328,
      "loss": 1.909,
      "step": 145507
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.0108729600906372,
      "learning_rate": 0.00017952765097576547,
      "loss": 2.0236,
      "step": 145508
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.081068515777588,
      "learning_rate": 0.00017952390474229623,
      "loss": 2.1856,
      "step": 145509
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.0801645517349243,
      "learning_rate": 0.00017952015853122573,
      "loss": 2.0071,
      "step": 145510
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.0136852264404297,
      "learning_rate": 0.00017951641234255477,
      "loss": 2.1415,
      "step": 145511
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.2012859582901,
      "learning_rate": 0.000179512666176284,
      "loss": 2.1284,
      "step": 145512
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.133292317390442,
      "learning_rate": 0.00017950892003241404,
      "loss": 2.1542,
      "step": 145513
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.232148289680481,
      "learning_rate": 0.00017950517391094562,
      "loss": 2.0246,
      "step": 145514
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.1680662631988525,
      "learning_rate": 0.00017950142781187947,
      "loss": 2.1164,
      "step": 145515
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.1427042484283447,
      "learning_rate": 0.0001794976817352163,
      "loss": 2.1193,
      "step": 145516
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.1269371509552002,
      "learning_rate": 0.00017949393568095676,
      "loss": 2.1312,
      "step": 145517
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.138511300086975,
      "learning_rate": 0.0001794901896491017,
      "loss": 2.1614,
      "step": 145518
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.569922924041748,
      "learning_rate": 0.0001794864436396516,
      "loss": 1.9066,
      "step": 145519
    },
    {
      "epoch": 1.89,
      "grad_norm": 0.9774317741394043,
      "learning_rate": 0.00017948269765260722,
      "loss": 1.9481,
      "step": 145520
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.0755367279052734,
      "learning_rate": 0.0001794789516879693,
      "loss": 2.0434,
      "step": 145521
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.2566126585006714,
      "learning_rate": 0.0001794752057457385,
      "loss": 2.0171,
      "step": 145522
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.0891355276107788,
      "learning_rate": 0.00017947145982591552,
      "loss": 2.1497,
      "step": 145523
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.104711651802063,
      "learning_rate": 0.0001794677139285012,
      "loss": 2.2166,
      "step": 145524
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.0611517429351807,
      "learning_rate": 0.000179463968053496,
      "loss": 2.1016,
      "step": 145525
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.2153582572937012,
      "learning_rate": 0.0001794602222009007,
      "loss": 1.8353,
      "step": 145526
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.08064866065979,
      "learning_rate": 0.00017945647637071605,
      "loss": 1.9751,
      "step": 145527
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.1332756280899048,
      "learning_rate": 0.0001794527305629427,
      "loss": 1.8519,
      "step": 145528
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.0618904829025269,
      "learning_rate": 0.00017944898477758136,
      "loss": 2.1101,
      "step": 145529
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.0791611671447754,
      "learning_rate": 0.0001794452390146327,
      "loss": 2.1918,
      "step": 145530
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.0138615369796753,
      "learning_rate": 0.00017944149327409764,
      "loss": 1.9396,
      "step": 145531
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.0510711669921875,
      "learning_rate": 0.0001794377475559765,
      "loss": 1.9746,
      "step": 145532
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.0779634714126587,
      "learning_rate": 0.0001794340018602702,
      "loss": 1.9687,
      "step": 145533
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.4671438932418823,
      "learning_rate": 0.00017943025618697938,
      "loss": 1.8812,
      "step": 145534
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.0319818258285522,
      "learning_rate": 0.00017942651053610474,
      "loss": 2.4002,
      "step": 145535
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.0049922466278076,
      "learning_rate": 0.000179422764907647,
      "loss": 1.8876,
      "step": 145536
    },
    {
      "epoch": 1.89,
      "grad_norm": 1.0507961511611938,
      "learning_rate": 0.00017941901930160697,
      "loss": 2.0863,
      "step": 145537
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.0605851411819458,
      "learning_rate": 0.0001794152737179851,
      "loss": 2.2405,
      "step": 145538
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.2252286672592163,
      "learning_rate": 0.00017941152815678218,
      "loss": 2.2595,
      "step": 145539
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.0437883138656616,
      "learning_rate": 0.00017940778261799894,
      "loss": 1.8948,
      "step": 145540
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.0399634838104248,
      "learning_rate": 0.00017940403710163606,
      "loss": 2.0336,
      "step": 145541
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.079969048500061,
      "learning_rate": 0.00017940029160769426,
      "loss": 1.8613,
      "step": 145542
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.032179594039917,
      "learning_rate": 0.0001793965461361744,
      "loss": 1.9424,
      "step": 145543
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.316882848739624,
      "learning_rate": 0.0001793928006870768,
      "loss": 2.2877,
      "step": 145544
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.1290665864944458,
      "learning_rate": 0.0001793890552604024,
      "loss": 2.1913,
      "step": 145545
    },
    {
      "epoch": 1.9,
      "grad_norm": 0.9741809964179993,
      "learning_rate": 0.0001793853098561518,
      "loss": 1.9875,
      "step": 145546
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.0742371082305908,
      "learning_rate": 0.0001793815644743258,
      "loss": 2.1629,
      "step": 145547
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.161556601524353,
      "learning_rate": 0.00017937781911492502,
      "loss": 1.9889,
      "step": 145548
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.1794568300247192,
      "learning_rate": 0.00017937407377795032,
      "loss": 2.2337,
      "step": 145549
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.0241634845733643,
      "learning_rate": 0.00017937032846340208,
      "loss": 2.0398,
      "step": 145550
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.3071205615997314,
      "learning_rate": 0.00017936658317128126,
      "loss": 2.1168,
      "step": 145551
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.2363704442977905,
      "learning_rate": 0.0001793628379015884,
      "loss": 1.9488,
      "step": 145552
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.173117995262146,
      "learning_rate": 0.0001793590926543243,
      "loss": 1.9654,
      "step": 145553
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.635461688041687,
      "learning_rate": 0.00017935534742948963,
      "loss": 1.8585,
      "step": 145554
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.3966723680496216,
      "learning_rate": 0.00017935160222708516,
      "loss": 1.8607,
      "step": 145555
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.3245482444763184,
      "learning_rate": 0.0001793478570471114,
      "loss": 1.9291,
      "step": 145556
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.1636841297149658,
      "learning_rate": 0.00017934411188956915,
      "loss": 2.079,
      "step": 145557
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.1742593050003052,
      "learning_rate": 0.00017934036675445913,
      "loss": 2.1077,
      "step": 145558
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.19795560836792,
      "learning_rate": 0.000179336621641782,
      "loss": 2.1826,
      "step": 145559
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.1842572689056396,
      "learning_rate": 0.00017933287655153845,
      "loss": 1.9267,
      "step": 145560
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.1360142230987549,
      "learning_rate": 0.0001793291314837293,
      "loss": 2.0043,
      "step": 145561
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.0860745906829834,
      "learning_rate": 0.0001793253864383551,
      "loss": 1.7789,
      "step": 145562
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.2285104990005493,
      "learning_rate": 0.00017932164141541655,
      "loss": 2.0056,
      "step": 145563
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.0951170921325684,
      "learning_rate": 0.00017931789641491438,
      "loss": 2.1907,
      "step": 145564
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.239786148071289,
      "learning_rate": 0.00017931415143684925,
      "loss": 1.9237,
      "step": 145565
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.0475959777832031,
      "learning_rate": 0.00017931040648122193,
      "loss": 1.9805,
      "step": 145566
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.1604101657867432,
      "learning_rate": 0.0001793066615480332,
      "loss": 1.8869,
      "step": 145567
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.0908416509628296,
      "learning_rate": 0.00017930291663728355,
      "loss": 1.9817,
      "step": 145568
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.136009693145752,
      "learning_rate": 0.00017929917174897372,
      "loss": 1.7844,
      "step": 145569
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.0769917964935303,
      "learning_rate": 0.00017929542688310452,
      "loss": 2.3933,
      "step": 145570
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.0944745540618896,
      "learning_rate": 0.0001792916820396765,
      "loss": 2.0402,
      "step": 145571
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.243377923965454,
      "learning_rate": 0.0001792879372186905,
      "loss": 2.2086,
      "step": 145572
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.3798006772994995,
      "learning_rate": 0.00017928419242014716,
      "loss": 2.2329,
      "step": 145573
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.0189707279205322,
      "learning_rate": 0.0001792804476440472,
      "loss": 2.127,
      "step": 145574
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.1567389965057373,
      "learning_rate": 0.00017927670289039123,
      "loss": 2.087,
      "step": 145575
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.3556095361709595,
      "learning_rate": 0.00017927295815917997,
      "loss": 2.0482,
      "step": 145576
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.145155906677246,
      "learning_rate": 0.00017926921345041416,
      "loss": 2.2,
      "step": 145577
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.076092004776001,
      "learning_rate": 0.00017926546876409448,
      "loss": 2.0441,
      "step": 145578
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.1116093397140503,
      "learning_rate": 0.00017926172410022168,
      "loss": 2.2343,
      "step": 145579
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.091536283493042,
      "learning_rate": 0.00017925797945879634,
      "loss": 1.994,
      "step": 145580
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.143875241279602,
      "learning_rate": 0.00017925423483981925,
      "loss": 1.9253,
      "step": 145581
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.1669211387634277,
      "learning_rate": 0.00017925049024329113,
      "loss": 2.1546,
      "step": 145582
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.0849380493164062,
      "learning_rate": 0.00017924674566921253,
      "loss": 1.8762,
      "step": 145583
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.158107042312622,
      "learning_rate": 0.00017924300111758426,
      "loss": 1.9191,
      "step": 145584
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.066082239151001,
      "learning_rate": 0.00017923925658840708,
      "loss": 2.1259,
      "step": 145585
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.1356381177902222,
      "learning_rate": 0.00017923551208168153,
      "loss": 1.9815,
      "step": 145586
    },
    {
      "epoch": 1.9,
      "grad_norm": 0.9926566481590271,
      "learning_rate": 0.00017923176759740834,
      "loss": 2.2801,
      "step": 145587
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.0489158630371094,
      "learning_rate": 0.0001792280231355883,
      "loss": 1.8442,
      "step": 145588
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.0293349027633667,
      "learning_rate": 0.0001792242786962221,
      "loss": 2.0182,
      "step": 145589
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.0544219017028809,
      "learning_rate": 0.00017922053427931028,
      "loss": 2.0549,
      "step": 145590
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.1242114305496216,
      "learning_rate": 0.00017921678988485373,
      "loss": 2.2331,
      "step": 145591
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.1368935108184814,
      "learning_rate": 0.00017921304551285301,
      "loss": 1.9206,
      "step": 145592
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.110435128211975,
      "learning_rate": 0.00017920930116330886,
      "loss": 1.9147,
      "step": 145593
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.2637079954147339,
      "learning_rate": 0.000179205556836222,
      "loss": 2.1025,
      "step": 145594
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.1372076272964478,
      "learning_rate": 0.00017920181253159304,
      "loss": 2.1993,
      "step": 145595
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.0775177478790283,
      "learning_rate": 0.00017919806824942287,
      "loss": 2.0553,
      "step": 145596
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.1243833303451538,
      "learning_rate": 0.000179194323989712,
      "loss": 2.0124,
      "step": 145597
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.1765668392181396,
      "learning_rate": 0.0001791905797524612,
      "loss": 2.0725,
      "step": 145598
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.1706126928329468,
      "learning_rate": 0.00017918683553767113,
      "loss": 1.9905,
      "step": 145599
    },
    {
      "epoch": 1.9,
      "grad_norm": 5.357943058013916,
      "learning_rate": 0.00017918309134534247,
      "loss": 1.9021,
      "step": 145600
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.014289379119873,
      "learning_rate": 0.000179179347175476,
      "loss": 2.047,
      "step": 145601
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.0923962593078613,
      "learning_rate": 0.00017917560302807233,
      "loss": 1.9769,
      "step": 145602
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.1376898288726807,
      "learning_rate": 0.00017917185890313226,
      "loss": 2.0611,
      "step": 145603
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.102048635482788,
      "learning_rate": 0.00017916811480065642,
      "loss": 1.7702,
      "step": 145604
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.0309427976608276,
      "learning_rate": 0.00017916437072064547,
      "loss": 2.0796,
      "step": 145605
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.239288568496704,
      "learning_rate": 0.00017916062666310014,
      "loss": 2.3312,
      "step": 145606
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.112656831741333,
      "learning_rate": 0.00017915688262802116,
      "loss": 2.1278,
      "step": 145607
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.1756649017333984,
      "learning_rate": 0.00017915313861540915,
      "loss": 1.9633,
      "step": 145608
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.1972501277923584,
      "learning_rate": 0.00017914939462526487,
      "loss": 1.9323,
      "step": 145609
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.0932543277740479,
      "learning_rate": 0.0001791456506575891,
      "loss": 2.1047,
      "step": 145610
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.1880269050598145,
      "learning_rate": 0.00017914190671238235,
      "loss": 2.2093,
      "step": 145611
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.242461085319519,
      "learning_rate": 0.0001791381627896454,
      "loss": 2.12,
      "step": 145612
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.078684687614441,
      "learning_rate": 0.00017913441888937892,
      "loss": 1.8878,
      "step": 145613
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.1688495874404907,
      "learning_rate": 0.0001791306750115836,
      "loss": 1.9088,
      "step": 145614
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.3240453004837036,
      "learning_rate": 0.00017912693115626024,
      "loss": 2.1388,
      "step": 145615
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.0809352397918701,
      "learning_rate": 0.00017912318732340955,
      "loss": 2.0225,
      "step": 145616
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.1309442520141602,
      "learning_rate": 0.000179119443513032,
      "loss": 2.1973,
      "step": 145617
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.2417300939559937,
      "learning_rate": 0.0001791156997251285,
      "loss": 2.0068,
      "step": 145618
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.017187476158142,
      "learning_rate": 0.00017911195595969958,
      "loss": 2.1798,
      "step": 145619
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.0044068098068237,
      "learning_rate": 0.0001791082122167461,
      "loss": 2.2092,
      "step": 145620
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.126010775566101,
      "learning_rate": 0.00017910446849626866,
      "loss": 2.1553,
      "step": 145621
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.1267642974853516,
      "learning_rate": 0.0001791007247982681,
      "loss": 2.013,
      "step": 145622
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.1022658348083496,
      "learning_rate": 0.00017909698112274488,
      "loss": 2.2415,
      "step": 145623
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.1341971158981323,
      "learning_rate": 0.00017909323746969984,
      "loss": 2.0674,
      "step": 145624
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.0265065431594849,
      "learning_rate": 0.00017908949383913363,
      "loss": 1.9468,
      "step": 145625
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.0510081052780151,
      "learning_rate": 0.00017908575023104693,
      "loss": 2.1495,
      "step": 145626
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.112528920173645,
      "learning_rate": 0.00017908200664544052,
      "loss": 1.9676,
      "step": 145627
    },
    {
      "epoch": 1.9,
      "grad_norm": 0.9887184500694275,
      "learning_rate": 0.00017907826308231512,
      "loss": 1.9624,
      "step": 145628
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.1278561353683472,
      "learning_rate": 0.00017907451954167128,
      "loss": 1.944,
      "step": 145629
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.0444695949554443,
      "learning_rate": 0.00017907077602350975,
      "loss": 2.1965,
      "step": 145630
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.0829964876174927,
      "learning_rate": 0.00017906703252783124,
      "loss": 2.2082,
      "step": 145631
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.1070228815078735,
      "learning_rate": 0.00017906328905463646,
      "loss": 2.1533,
      "step": 145632
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.228228211402893,
      "learning_rate": 0.00017905954560392607,
      "loss": 2.2134,
      "step": 145633
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.1185518503189087,
      "learning_rate": 0.00017905580217570096,
      "loss": 2.3354,
      "step": 145634
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.0385849475860596,
      "learning_rate": 0.00017905205876996152,
      "loss": 2.3839,
      "step": 145635
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.3336153030395508,
      "learning_rate": 0.0001790483153867086,
      "loss": 2.2104,
      "step": 145636
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.1189855337142944,
      "learning_rate": 0.00017904457202594286,
      "loss": 1.9123,
      "step": 145637
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.1317789554595947,
      "learning_rate": 0.000179040828687665,
      "loss": 2.209,
      "step": 145638
    },
    {
      "epoch": 1.9,
      "grad_norm": 0.985796332359314,
      "learning_rate": 0.00017903708537187577,
      "loss": 2.2986,
      "step": 145639
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.0947033166885376,
      "learning_rate": 0.00017903334207857593,
      "loss": 2.2643,
      "step": 145640
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.1238142251968384,
      "learning_rate": 0.00017902959880776594,
      "loss": 1.9757,
      "step": 145641
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.3242000341415405,
      "learning_rate": 0.00017902585555944667,
      "loss": 2.0027,
      "step": 145642
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.1565362215042114,
      "learning_rate": 0.00017902211233361873,
      "loss": 2.1338,
      "step": 145643
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.1325559616088867,
      "learning_rate": 0.00017901836913028293,
      "loss": 2.0355,
      "step": 145644
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.3126929998397827,
      "learning_rate": 0.00017901462594943981,
      "loss": 2.123,
      "step": 145645
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.5195280313491821,
      "learning_rate": 0.00017901088279109035,
      "loss": 2.0748,
      "step": 145646
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.2353578805923462,
      "learning_rate": 0.00017900713965523496,
      "loss": 2.0893,
      "step": 145647
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.1749790906906128,
      "learning_rate": 0.00017900339654187435,
      "loss": 2.1391,
      "step": 145648
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.8288739919662476,
      "learning_rate": 0.00017899965345100928,
      "loss": 2.1148,
      "step": 145649
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.320151686668396,
      "learning_rate": 0.0001789959103826405,
      "loss": 2.1005,
      "step": 145650
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.181740403175354,
      "learning_rate": 0.00017899216733676865,
      "loss": 2.1781,
      "step": 145651
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.1851173639297485,
      "learning_rate": 0.0001789884243133946,
      "loss": 1.8272,
      "step": 145652
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.132389783859253,
      "learning_rate": 0.00017898468131251872,
      "loss": 2.2676,
      "step": 145653
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.0802996158599854,
      "learning_rate": 0.00017898093833414188,
      "loss": 2.2555,
      "step": 145654
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.1200549602508545,
      "learning_rate": 0.00017897719537826476,
      "loss": 2.0437,
      "step": 145655
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.0028760433197021,
      "learning_rate": 0.0001789734524448881,
      "loss": 2.0321,
      "step": 145656
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.138663649559021,
      "learning_rate": 0.00017896970953401253,
      "loss": 2.2501,
      "step": 145657
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.112793207168579,
      "learning_rate": 0.00017896596664563884,
      "loss": 2.2013,
      "step": 145658
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.1490657329559326,
      "learning_rate": 0.00017896222377976765,
      "loss": 2.0143,
      "step": 145659
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.2264978885650635,
      "learning_rate": 0.00017895848093639962,
      "loss": 2.0318,
      "step": 145660
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.091119647026062,
      "learning_rate": 0.0001789547381155355,
      "loss": 1.9939,
      "step": 145661
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.0061084032058716,
      "learning_rate": 0.000178950995317176,
      "loss": 1.9565,
      "step": 145662
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.0822930335998535,
      "learning_rate": 0.00017894725254132175,
      "loss": 2.0465,
      "step": 145663
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.1155551671981812,
      "learning_rate": 0.0001789435097879735,
      "loss": 2.1337,
      "step": 145664
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.0545883178710938,
      "learning_rate": 0.00017893976705713203,
      "loss": 2.059,
      "step": 145665
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.1288001537322998,
      "learning_rate": 0.00017893602434879792,
      "loss": 1.8046,
      "step": 145666
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.0566368103027344,
      "learning_rate": 0.0001789322816629718,
      "loss": 2.1013,
      "step": 145667
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.0516810417175293,
      "learning_rate": 0.0001789285389996545,
      "loss": 2.1508,
      "step": 145668
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.0599753856658936,
      "learning_rate": 0.00017892479635884666,
      "loss": 1.993,
      "step": 145669
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.2623298168182373,
      "learning_rate": 0.00017892105374054895,
      "loss": 2.1307,
      "step": 145670
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.018244981765747,
      "learning_rate": 0.0001789173111447622,
      "loss": 1.8969,
      "step": 145671
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.3343511819839478,
      "learning_rate": 0.00017891356857148693,
      "loss": 2.1699,
      "step": 145672
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.1058627367019653,
      "learning_rate": 0.00017890982602072395,
      "loss": 2.1196,
      "step": 145673
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.183117151260376,
      "learning_rate": 0.00017890608349247395,
      "loss": 2.1302,
      "step": 145674
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.0708346366882324,
      "learning_rate": 0.00017890234098673753,
      "loss": 1.9905,
      "step": 145675
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.0717705488204956,
      "learning_rate": 0.00017889859850351547,
      "loss": 2.0458,
      "step": 145676
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.17283296585083,
      "learning_rate": 0.00017889485604280847,
      "loss": 2.2827,
      "step": 145677
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.1065410375595093,
      "learning_rate": 0.00017889111360461715,
      "loss": 2.0059,
      "step": 145678
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.145877718925476,
      "learning_rate": 0.0001788873711889423,
      "loss": 1.8645,
      "step": 145679
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.1081397533416748,
      "learning_rate": 0.00017888362879578453,
      "loss": 2.2597,
      "step": 145680
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.1461201906204224,
      "learning_rate": 0.00017887988642514464,
      "loss": 2.0678,
      "step": 145681
    },
    {
      "epoch": 1.9,
      "grad_norm": 2.265845775604248,
      "learning_rate": 0.00017887614407702322,
      "loss": 2.1247,
      "step": 145682
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.4378362894058228,
      "learning_rate": 0.0001788724017514211,
      "loss": 2.2399,
      "step": 145683
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.1247349977493286,
      "learning_rate": 0.00017886865944833879,
      "loss": 2.068,
      "step": 145684
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.0840610265731812,
      "learning_rate": 0.0001788649171677771,
      "loss": 2.1233,
      "step": 145685
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.0490946769714355,
      "learning_rate": 0.0001788611749097367,
      "loss": 2.0843,
      "step": 145686
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.1383202075958252,
      "learning_rate": 0.00017885743267421827,
      "loss": 2.2136,
      "step": 145687
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.1148416996002197,
      "learning_rate": 0.00017885369046122262,
      "loss": 2.2233,
      "step": 145688
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.0984386205673218,
      "learning_rate": 0.00017884994827075036,
      "loss": 2.0033,
      "step": 145689
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.087681770324707,
      "learning_rate": 0.0001788462061028021,
      "loss": 2.1705,
      "step": 145690
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.0878616571426392,
      "learning_rate": 0.00017884246395737864,
      "loss": 1.9385,
      "step": 145691
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.199533462524414,
      "learning_rate": 0.00017883872183448065,
      "loss": 1.982,
      "step": 145692
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.1182960271835327,
      "learning_rate": 0.0001788349797341088,
      "loss": 2.1955,
      "step": 145693
    },
    {
      "epoch": 1.9,
      "grad_norm": 0.9638996124267578,
      "learning_rate": 0.00017883123765626382,
      "loss": 2.0285,
      "step": 145694
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.092812418937683,
      "learning_rate": 0.00017882749560094654,
      "loss": 2.0253,
      "step": 145695
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.0491794347763062,
      "learning_rate": 0.0001788237535681574,
      "loss": 2.0252,
      "step": 145696
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.2818299531936646,
      "learning_rate": 0.0001788200115578972,
      "loss": 1.9077,
      "step": 145697
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.125654697418213,
      "learning_rate": 0.00017881626957016663,
      "loss": 1.8455,
      "step": 145698
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.104163408279419,
      "learning_rate": 0.00017881252760496642,
      "loss": 2.037,
      "step": 145699
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.2159042358398438,
      "learning_rate": 0.0001788087856622973,
      "loss": 2.2012,
      "step": 145700
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.1255613565444946,
      "learning_rate": 0.00017880504374215997,
      "loss": 2.1585,
      "step": 145701
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.1447416543960571,
      "learning_rate": 0.00017880130184455496,
      "loss": 2.0582,
      "step": 145702
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.0437942743301392,
      "learning_rate": 0.00017879755996948307,
      "loss": 2.1036,
      "step": 145703
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.044235110282898,
      "learning_rate": 0.00017879381811694502,
      "loss": 1.8755,
      "step": 145704
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.1525163650512695,
      "learning_rate": 0.0001787900762869415,
      "loss": 2.2581,
      "step": 145705
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.2008708715438843,
      "learning_rate": 0.00017878633447947318,
      "loss": 1.9866,
      "step": 145706
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.103610634803772,
      "learning_rate": 0.00017878259269454088,
      "loss": 2.2259,
      "step": 145707
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.0519156455993652,
      "learning_rate": 0.00017877885093214505,
      "loss": 2.1199,
      "step": 145708
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.2590268850326538,
      "learning_rate": 0.00017877510919228653,
      "loss": 2.1198,
      "step": 145709
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.1789793968200684,
      "learning_rate": 0.000178771367474966,
      "loss": 2.1257,
      "step": 145710
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.2247503995895386,
      "learning_rate": 0.0001787676257801842,
      "loss": 2.1321,
      "step": 145711
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.0325140953063965,
      "learning_rate": 0.00017876388410794175,
      "loss": 1.8794,
      "step": 145712
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.2499768733978271,
      "learning_rate": 0.0001787601424582395,
      "loss": 2.2044,
      "step": 145713
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.0746806859970093,
      "learning_rate": 0.0001787564008310779,
      "loss": 1.9864,
      "step": 145714
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.2182896137237549,
      "learning_rate": 0.0001787526592264578,
      "loss": 2.2389,
      "step": 145715
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.1530925035476685,
      "learning_rate": 0.0001787489176443799,
      "loss": 2.1996,
      "step": 145716
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.1628987789154053,
      "learning_rate": 0.00017874517608484482,
      "loss": 2.0987,
      "step": 145717
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.2063874006271362,
      "learning_rate": 0.0001787414345478533,
      "loss": 2.3478,
      "step": 145718
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.2149097919464111,
      "learning_rate": 0.00017873769303340616,
      "loss": 2.2576,
      "step": 145719
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.037739872932434,
      "learning_rate": 0.00017873395154150386,
      "loss": 2.0291,
      "step": 145720
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.1359961032867432,
      "learning_rate": 0.0001787302100721472,
      "loss": 2.2013,
      "step": 145721
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.0290071964263916,
      "learning_rate": 0.0001787264686253369,
      "loss": 2.1354,
      "step": 145722
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.090835452079773,
      "learning_rate": 0.00017872272720107358,
      "loss": 2.1977,
      "step": 145723
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.1361075639724731,
      "learning_rate": 0.00017871898579935807,
      "loss": 1.9738,
      "step": 145724
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.1407291889190674,
      "learning_rate": 0.00017871524442019106,
      "loss": 2.0688,
      "step": 145725
    },
    {
      "epoch": 1.9,
      "grad_norm": 0.9737786650657654,
      "learning_rate": 0.0001787115030635731,
      "loss": 2.015,
      "step": 145726
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.035805106163025,
      "learning_rate": 0.00017870776172950495,
      "loss": 1.9089,
      "step": 145727
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.293945550918579,
      "learning_rate": 0.00017870402041798728,
      "loss": 2.2217,
      "step": 145728
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.0813672542572021,
      "learning_rate": 0.00017870027912902085,
      "loss": 2.0376,
      "step": 145729
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.2738920450210571,
      "learning_rate": 0.0001786965378626063,
      "loss": 1.9212,
      "step": 145730
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.0797746181488037,
      "learning_rate": 0.0001786927966187444,
      "loss": 2.2562,
      "step": 145731
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.1651077270507812,
      "learning_rate": 0.00017868905539743593,
      "loss": 2.1726,
      "step": 145732
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.1831777095794678,
      "learning_rate": 0.00017868531419868127,
      "loss": 2.0732,
      "step": 145733
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.0380711555480957,
      "learning_rate": 0.00017868157302248133,
      "loss": 2.0052,
      "step": 145734
    },
    {
      "epoch": 1.9,
      "grad_norm": 0.9931790232658386,
      "learning_rate": 0.00017867783186883675,
      "loss": 2.0506,
      "step": 145735
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.1472560167312622,
      "learning_rate": 0.0001786740907377483,
      "loss": 2.0686,
      "step": 145736
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.067512035369873,
      "learning_rate": 0.0001786703496292166,
      "loss": 2.1709,
      "step": 145737
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.0649793148040771,
      "learning_rate": 0.0001786666085432425,
      "loss": 2.2105,
      "step": 145738
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.0526732206344604,
      "learning_rate": 0.00017866286747982645,
      "loss": 2.0763,
      "step": 145739
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.0148109197616577,
      "learning_rate": 0.00017865912643896925,
      "loss": 1.9335,
      "step": 145740
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.070019006729126,
      "learning_rate": 0.00017865538542067165,
      "loss": 2.0619,
      "step": 145741
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.1836650371551514,
      "learning_rate": 0.00017865164442493425,
      "loss": 2.0128,
      "step": 145742
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.2235453128814697,
      "learning_rate": 0.0001786479034517578,
      "loss": 2.2256,
      "step": 145743
    },
    {
      "epoch": 1.9,
      "grad_norm": 0.9760052561759949,
      "learning_rate": 0.00017864416250114312,
      "loss": 1.8942,
      "step": 145744
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.1898789405822754,
      "learning_rate": 0.0001786404215730907,
      "loss": 2.1727,
      "step": 145745
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.1505818367004395,
      "learning_rate": 0.00017863668066760128,
      "loss": 2.2619,
      "step": 145746
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.050636887550354,
      "learning_rate": 0.0001786329397846756,
      "loss": 1.9064,
      "step": 145747
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.1633896827697754,
      "learning_rate": 0.00017862919892431439,
      "loss": 2.1767,
      "step": 145748
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.1298083066940308,
      "learning_rate": 0.00017862545808651825,
      "loss": 2.0695,
      "step": 145749
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.132682204246521,
      "learning_rate": 0.000178621717271288,
      "loss": 2.054,
      "step": 145750
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.201090931892395,
      "learning_rate": 0.00017861797647862428,
      "loss": 2.2282,
      "step": 145751
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.0302352905273438,
      "learning_rate": 0.00017861423570852768,
      "loss": 1.9877,
      "step": 145752
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.1298048496246338,
      "learning_rate": 0.00017861049496099903,
      "loss": 2.223,
      "step": 145753
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.3281681537628174,
      "learning_rate": 0.00017860675423603893,
      "loss": 2.1203,
      "step": 145754
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.104814052581787,
      "learning_rate": 0.00017860301353364816,
      "loss": 2.2198,
      "step": 145755
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.1973885297775269,
      "learning_rate": 0.0001785992728538274,
      "loss": 2.1074,
      "step": 145756
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.0379611253738403,
      "learning_rate": 0.00017859553219657734,
      "loss": 2.0849,
      "step": 145757
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.1175910234451294,
      "learning_rate": 0.00017859179156189865,
      "loss": 2.0669,
      "step": 145758
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.1105142831802368,
      "learning_rate": 0.000178588050949792,
      "loss": 2.2336,
      "step": 145759
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.05098295211792,
      "learning_rate": 0.00017858431036025816,
      "loss": 2.0725,
      "step": 145760
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.0582129955291748,
      "learning_rate": 0.00017858056979329774,
      "loss": 2.0852,
      "step": 145761
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.0346043109893799,
      "learning_rate": 0.00017857682924891154,
      "loss": 1.9848,
      "step": 145762
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.1745036840438843,
      "learning_rate": 0.0001785730887271002,
      "loss": 2.1415,
      "step": 145763
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.0447190999984741,
      "learning_rate": 0.00017856934822786438,
      "loss": 2.0113,
      "step": 145764
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.2286607027053833,
      "learning_rate": 0.00017856560775120483,
      "loss": 2.0179,
      "step": 145765
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.0320650339126587,
      "learning_rate": 0.00017856186729712224,
      "loss": 1.8174,
      "step": 145766
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.1139551401138306,
      "learning_rate": 0.00017855812686561723,
      "loss": 2.0125,
      "step": 145767
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.1891136169433594,
      "learning_rate": 0.00017855438645669064,
      "loss": 2.2375,
      "step": 145768
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.033644676208496,
      "learning_rate": 0.00017855064607034303,
      "loss": 1.9801,
      "step": 145769
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.1178326606750488,
      "learning_rate": 0.00017854690570657516,
      "loss": 2.0419,
      "step": 145770
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.1305750608444214,
      "learning_rate": 0.00017854316536538767,
      "loss": 2.198,
      "step": 145771
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.0867341756820679,
      "learning_rate": 0.00017853942504678135,
      "loss": 2.0721,
      "step": 145772
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.165097951889038,
      "learning_rate": 0.00017853568475075684,
      "loss": 2.2162,
      "step": 145773
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.1423712968826294,
      "learning_rate": 0.00017853194447731488,
      "loss": 2.0572,
      "step": 145774
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.1511955261230469,
      "learning_rate": 0.00017852820422645604,
      "loss": 2.0257,
      "step": 145775
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.0681252479553223,
      "learning_rate": 0.00017852446399818113,
      "loss": 2.2873,
      "step": 145776
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.0681320428848267,
      "learning_rate": 0.0001785207237924908,
      "loss": 1.9207,
      "step": 145777
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.0467787981033325,
      "learning_rate": 0.00017851698360938576,
      "loss": 2.0228,
      "step": 145778
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.025781273841858,
      "learning_rate": 0.00017851324344886671,
      "loss": 2.103,
      "step": 145779
    },
    {
      "epoch": 1.9,
      "grad_norm": 0.9688288569450378,
      "learning_rate": 0.0001785095033109345,
      "loss": 1.9972,
      "step": 145780
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.1790257692337036,
      "learning_rate": 0.00017850576319558946,
      "loss": 2.2765,
      "step": 145781
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.0629819631576538,
      "learning_rate": 0.00017850202310283256,
      "loss": 2.0863,
      "step": 145782
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.1805907487869263,
      "learning_rate": 0.00017849828303266441,
      "loss": 2.0065,
      "step": 145783
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.0001180171966553,
      "learning_rate": 0.00017849454298508573,
      "loss": 2.012,
      "step": 145784
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.1838678121566772,
      "learning_rate": 0.00017849080296009718,
      "loss": 2.1064,
      "step": 145785
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.037247896194458,
      "learning_rate": 0.00017848706295769961,
      "loss": 2.1036,
      "step": 145786
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.0261950492858887,
      "learning_rate": 0.00017848332297789349,
      "loss": 1.8186,
      "step": 145787
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.1475003957748413,
      "learning_rate": 0.00017847958302067964,
      "loss": 2.2319,
      "step": 145788
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.159002661705017,
      "learning_rate": 0.0001784758430860587,
      "loss": 2.0865,
      "step": 145789
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.1454485654830933,
      "learning_rate": 0.00017847210317403138,
      "loss": 2.191,
      "step": 145790
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.1088573932647705,
      "learning_rate": 0.00017846836328459838,
      "loss": 2.0341,
      "step": 145791
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.073660135269165,
      "learning_rate": 0.00017846462341776056,
      "loss": 2.351,
      "step": 145792
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.165062427520752,
      "learning_rate": 0.00017846088357351833,
      "loss": 1.9885,
      "step": 145793
    },
    {
      "epoch": 1.9,
      "grad_norm": 0.9325112104415894,
      "learning_rate": 0.0001784571437518725,
      "loss": 2.0163,
      "step": 145794
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.2401869297027588,
      "learning_rate": 0.0001784534039528238,
      "loss": 2.1036,
      "step": 145795
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.1090912818908691,
      "learning_rate": 0.00017844966417637287,
      "loss": 2.2091,
      "step": 145796
    },
    {
      "epoch": 1.9,
      "grad_norm": 0.9914100766181946,
      "learning_rate": 0.00017844592442252047,
      "loss": 2.0015,
      "step": 145797
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.0309038162231445,
      "learning_rate": 0.00017844218469126728,
      "loss": 2.1973,
      "step": 145798
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.1139284372329712,
      "learning_rate": 0.00017843844498261407,
      "loss": 1.8843,
      "step": 145799
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.0351474285125732,
      "learning_rate": 0.00017843470529656135,
      "loss": 1.9179,
      "step": 145800
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.0997978448867798,
      "learning_rate": 0.00017843096563310992,
      "loss": 2.0413,
      "step": 145801
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.1200892925262451,
      "learning_rate": 0.00017842722599226046,
      "loss": 2.1893,
      "step": 145802
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.1145014762878418,
      "learning_rate": 0.00017842348637401366,
      "loss": 2.1066,
      "step": 145803
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.1681967973709106,
      "learning_rate": 0.00017841974677837026,
      "loss": 2.1316,
      "step": 145804
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.0853575468063354,
      "learning_rate": 0.000178416007205331,
      "loss": 2.1906,
      "step": 145805
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.1069426536560059,
      "learning_rate": 0.00017841226765489645,
      "loss": 2.2021,
      "step": 145806
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.1015475988388062,
      "learning_rate": 0.0001784085281270673,
      "loss": 2.0373,
      "step": 145807
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.088531494140625,
      "learning_rate": 0.0001784047886218443,
      "loss": 2.2774,
      "step": 145808
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.0203763246536255,
      "learning_rate": 0.00017840104913922816,
      "loss": 2.2002,
      "step": 145809
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.126848578453064,
      "learning_rate": 0.00017839730967921953,
      "loss": 2.1183,
      "step": 145810
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.1548327207565308,
      "learning_rate": 0.0001783935702418193,
      "loss": 2.0661,
      "step": 145811
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.1385639905929565,
      "learning_rate": 0.00017838983082702782,
      "loss": 2.1735,
      "step": 145812
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.1694284677505493,
      "learning_rate": 0.00017838609143484605,
      "loss": 2.1669,
      "step": 145813
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.0461617708206177,
      "learning_rate": 0.00017838235206527454,
      "loss": 2.1347,
      "step": 145814
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.3147556781768799,
      "learning_rate": 0.00017837861271831405,
      "loss": 2.0784,
      "step": 145815
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.1747193336486816,
      "learning_rate": 0.0001783748733939653,
      "loss": 1.9035,
      "step": 145816
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.276976466178894,
      "learning_rate": 0.00017837113409222905,
      "loss": 2.1893,
      "step": 145817
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.063336968421936,
      "learning_rate": 0.00017836739481310581,
      "loss": 1.9423,
      "step": 145818
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.0909531116485596,
      "learning_rate": 0.00017836365555659635,
      "loss": 2.2301,
      "step": 145819
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.1596815586090088,
      "learning_rate": 0.00017835991632270137,
      "loss": 2.2462,
      "step": 145820
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.0795395374298096,
      "learning_rate": 0.00017835617711142162,
      "loss": 1.9685,
      "step": 145821
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.0644198656082153,
      "learning_rate": 0.00017835243792275774,
      "loss": 1.9136,
      "step": 145822
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.3493272066116333,
      "learning_rate": 0.00017834869875671053,
      "loss": 2.0506,
      "step": 145823
    },
    {
      "epoch": 1.9,
      "grad_norm": 0.9915485978126526,
      "learning_rate": 0.00017834495961328045,
      "loss": 2.0843,
      "step": 145824
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.083497166633606,
      "learning_rate": 0.0001783412204924684,
      "loss": 2.132,
      "step": 145825
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.0818902254104614,
      "learning_rate": 0.00017833748139427498,
      "loss": 2.1822,
      "step": 145826
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.2503665685653687,
      "learning_rate": 0.00017833374231870092,
      "loss": 2.1069,
      "step": 145827
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.0563812255859375,
      "learning_rate": 0.00017833000326574695,
      "loss": 1.9131,
      "step": 145828
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.2612570524215698,
      "learning_rate": 0.0001783262642354138,
      "loss": 2.2813,
      "step": 145829
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.1142494678497314,
      "learning_rate": 0.000178322525227702,
      "loss": 2.0405,
      "step": 145830
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.070025086402893,
      "learning_rate": 0.00017831878624261234,
      "loss": 1.9503,
      "step": 145831
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.1971596479415894,
      "learning_rate": 0.0001783150472801455,
      "loss": 1.9343,
      "step": 145832
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.1231263875961304,
      "learning_rate": 0.00017831130834030222,
      "loss": 2.2632,
      "step": 145833
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.157679796218872,
      "learning_rate": 0.00017830756942308312,
      "loss": 2.2572,
      "step": 145834
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.0570714473724365,
      "learning_rate": 0.00017830383052848904,
      "loss": 2.0142,
      "step": 145835
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.1748706102371216,
      "learning_rate": 0.00017830009165652056,
      "loss": 2.237,
      "step": 145836
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.1334271430969238,
      "learning_rate": 0.00017829635280717832,
      "loss": 2.1897,
      "step": 145837
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.0717477798461914,
      "learning_rate": 0.00017829261398046308,
      "loss": 2.0102,
      "step": 145838
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.2617114782333374,
      "learning_rate": 0.00017828887517637557,
      "loss": 2.1591,
      "step": 145839
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.1616158485412598,
      "learning_rate": 0.00017828513639491642,
      "loss": 1.9832,
      "step": 145840
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.0893974304199219,
      "learning_rate": 0.00017828139763608646,
      "loss": 2.2244,
      "step": 145841
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.1900588274002075,
      "learning_rate": 0.0001782776588998862,
      "loss": 2.0785,
      "step": 145842
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.4896293878555298,
      "learning_rate": 0.0001782739201863165,
      "loss": 2.1197,
      "step": 145843
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.1466299295425415,
      "learning_rate": 0.0001782701814953779,
      "loss": 2.1939,
      "step": 145844
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.0814740657806396,
      "learning_rate": 0.00017826644282707119,
      "loss": 2.1026,
      "step": 145845
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.1409653425216675,
      "learning_rate": 0.000178262704181397,
      "loss": 2.0589,
      "step": 145846
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.220500111579895,
      "learning_rate": 0.00017825896555835618,
      "loss": 2.1772,
      "step": 145847
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.1603541374206543,
      "learning_rate": 0.0001782552269579492,
      "loss": 1.8663,
      "step": 145848
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.323762059211731,
      "learning_rate": 0.00017825148838017694,
      "loss": 2.1351,
      "step": 145849
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.2289103269577026,
      "learning_rate": 0.00017824774982504003,
      "loss": 2.1498,
      "step": 145850
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.0999351739883423,
      "learning_rate": 0.00017824401129253913,
      "loss": 2.1449,
      "step": 145851
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.111890435218811,
      "learning_rate": 0.00017824027278267498,
      "loss": 2.1078,
      "step": 145852
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.0274462699890137,
      "learning_rate": 0.00017823653429544832,
      "loss": 2.1555,
      "step": 145853
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.0793254375457764,
      "learning_rate": 0.0001782327958308597,
      "loss": 2.1856,
      "step": 145854
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.0029281377792358,
      "learning_rate": 0.0001782290573889099,
      "loss": 1.9314,
      "step": 145855
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.1795943975448608,
      "learning_rate": 0.00017822531896959963,
      "loss": 2.0441,
      "step": 145856
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.2351696491241455,
      "learning_rate": 0.00017822158057292966,
      "loss": 2.0164,
      "step": 145857
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.0943548679351807,
      "learning_rate": 0.0001782178421989005,
      "loss": 1.8873,
      "step": 145858
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.4167444705963135,
      "learning_rate": 0.00017821410384751303,
      "loss": 2.0741,
      "step": 145859
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.146336317062378,
      "learning_rate": 0.00017821036551876775,
      "loss": 2.19,
      "step": 145860
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.1274199485778809,
      "learning_rate": 0.00017820662721266552,
      "loss": 1.9564,
      "step": 145861
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.104016661643982,
      "learning_rate": 0.00017820288892920694,
      "loss": 2.1168,
      "step": 145862
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.069300889968872,
      "learning_rate": 0.00017819915066839273,
      "loss": 2.0368,
      "step": 145863
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.2060636281967163,
      "learning_rate": 0.00017819541243022362,
      "loss": 2.3469,
      "step": 145864
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.0507564544677734,
      "learning_rate": 0.00017819167421470036,
      "loss": 1.9888,
      "step": 145865
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.0473848581314087,
      "learning_rate": 0.00017818793602182353,
      "loss": 2.0983,
      "step": 145866
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.262105941772461,
      "learning_rate": 0.00017818419785159382,
      "loss": 2.3231,
      "step": 145867
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.0714079141616821,
      "learning_rate": 0.00017818045970401198,
      "loss": 1.801,
      "step": 145868
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.2065454721450806,
      "learning_rate": 0.00017817672157907868,
      "loss": 2.1403,
      "step": 145869
    },
    {
      "epoch": 1.9,
      "grad_norm": 0.9991328716278076,
      "learning_rate": 0.00017817298347679462,
      "loss": 2.0183,
      "step": 145870
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.198637843132019,
      "learning_rate": 0.00017816924539716049,
      "loss": 2.0702,
      "step": 145871
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.0802181959152222,
      "learning_rate": 0.00017816550734017713,
      "loss": 1.9753,
      "step": 145872
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.1308176517486572,
      "learning_rate": 0.00017816176930584497,
      "loss": 2.1333,
      "step": 145873
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.1464921236038208,
      "learning_rate": 0.00017815803129416486,
      "loss": 2.0507,
      "step": 145874
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.019948124885559,
      "learning_rate": 0.00017815429330513748,
      "loss": 2.025,
      "step": 145875
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.0612956285476685,
      "learning_rate": 0.00017815055533876352,
      "loss": 2.1682,
      "step": 145876
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.1797443628311157,
      "learning_rate": 0.00017814681739504364,
      "loss": 1.9994,
      "step": 145877
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.0390316247940063,
      "learning_rate": 0.0001781430794739787,
      "loss": 2.052,
      "step": 145878
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.1096702814102173,
      "learning_rate": 0.00017813934157556914,
      "loss": 2.1029,
      "step": 145879
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.1061971187591553,
      "learning_rate": 0.00017813560369981578,
      "loss": 2.0937,
      "step": 145880
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.1808972358703613,
      "learning_rate": 0.00017813186584671927,
      "loss": 2.1833,
      "step": 145881
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.2598754167556763,
      "learning_rate": 0.0001781281280162804,
      "loss": 1.9297,
      "step": 145882
    },
    {
      "epoch": 1.9,
      "grad_norm": 0.9949250817298889,
      "learning_rate": 0.00017812439020849983,
      "loss": 2.1641,
      "step": 145883
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.1589401960372925,
      "learning_rate": 0.0001781206524233783,
      "loss": 1.978,
      "step": 145884
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.1389285326004028,
      "learning_rate": 0.00017811691466091637,
      "loss": 2.0729,
      "step": 145885
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.0016088485717773,
      "learning_rate": 0.00017811317692111476,
      "loss": 2.1368,
      "step": 145886
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.2661532163619995,
      "learning_rate": 0.00017810943920397427,
      "loss": 1.9675,
      "step": 145887
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.1813006401062012,
      "learning_rate": 0.00017810570150949548,
      "loss": 2.1768,
      "step": 145888
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.3095561265945435,
      "learning_rate": 0.00017810196383767916,
      "loss": 2.2325,
      "step": 145889
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.227282166481018,
      "learning_rate": 0.00017809822618852612,
      "loss": 2.0126,
      "step": 145890
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.3171528577804565,
      "learning_rate": 0.00017809448856203678,
      "loss": 1.9681,
      "step": 145891
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.1617995500564575,
      "learning_rate": 0.00017809075095821198,
      "loss": 1.9636,
      "step": 145892
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.1252297163009644,
      "learning_rate": 0.0001780870133770524,
      "loss": 2.0617,
      "step": 145893
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.1302876472473145,
      "learning_rate": 0.00017808327581855877,
      "loss": 2.2013,
      "step": 145894
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.122301459312439,
      "learning_rate": 0.00017807953828273174,
      "loss": 2.0093,
      "step": 145895
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.4068070650100708,
      "learning_rate": 0.00017807580076957218,
      "loss": 2.0991,
      "step": 145896
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.666162133216858,
      "learning_rate": 0.00017807206327908047,
      "loss": 2.0069,
      "step": 145897
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.16312575340271,
      "learning_rate": 0.00017806832581125745,
      "loss": 2.222,
      "step": 145898
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.4122333526611328,
      "learning_rate": 0.00017806458836610388,
      "loss": 2.0617,
      "step": 145899
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.2135322093963623,
      "learning_rate": 0.0001780608509436204,
      "loss": 2.1104,
      "step": 145900
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.1899216175079346,
      "learning_rate": 0.0001780571135438077,
      "loss": 2.0689,
      "step": 145901
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.099070429801941,
      "learning_rate": 0.0001780533761666666,
      "loss": 2.1037,
      "step": 145902
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.1059049367904663,
      "learning_rate": 0.00017804963881219756,
      "loss": 2.1417,
      "step": 145903
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.597931146621704,
      "learning_rate": 0.0001780459014804014,
      "loss": 2.1445,
      "step": 145904
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.0818427801132202,
      "learning_rate": 0.00017804216417127879,
      "loss": 1.9735,
      "step": 145905
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.1131585836410522,
      "learning_rate": 0.00017803842688483043,
      "loss": 2.1165,
      "step": 145906
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.2412164211273193,
      "learning_rate": 0.00017803468962105705,
      "loss": 1.9255,
      "step": 145907
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.0887258052825928,
      "learning_rate": 0.00017803095237995946,
      "loss": 2.0987,
      "step": 145908
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.2421250343322754,
      "learning_rate": 0.00017802721516153806,
      "loss": 2.2491,
      "step": 145909
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.0748462677001953,
      "learning_rate": 0.00017802347796579373,
      "loss": 2.0964,
      "step": 145910
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.1793593168258667,
      "learning_rate": 0.00017801974079272715,
      "loss": 1.8172,
      "step": 145911
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.1506855487823486,
      "learning_rate": 0.000178016003642339,
      "loss": 1.969,
      "step": 145912
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.231900930404663,
      "learning_rate": 0.00017801226651462993,
      "loss": 2.2378,
      "step": 145913
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.1509732007980347,
      "learning_rate": 0.00017800852940960088,
      "loss": 2.2797,
      "step": 145914
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.0578665733337402,
      "learning_rate": 0.0001780047923272522,
      "loss": 2.1258,
      "step": 145915
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.2158524990081787,
      "learning_rate": 0.00017800105526758468,
      "loss": 2.1635,
      "step": 145916
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.2567803859710693,
      "learning_rate": 0.00017799731823059912,
      "loss": 2.3345,
      "step": 145917
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.1112476587295532,
      "learning_rate": 0.00017799358121629614,
      "loss": 2.177,
      "step": 145918
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.022665023803711,
      "learning_rate": 0.00017798984422467645,
      "loss": 1.9243,
      "step": 145919
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.098612904548645,
      "learning_rate": 0.0001779861072557409,
      "loss": 2.1122,
      "step": 145920
    },
    {
      "epoch": 1.9,
      "grad_norm": 0.9947651624679565,
      "learning_rate": 0.00017798237030948992,
      "loss": 1.9359,
      "step": 145921
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.138094425201416,
      "learning_rate": 0.00017797863338592427,
      "loss": 2.0054,
      "step": 145922
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.0634804964065552,
      "learning_rate": 0.00017797489648504471,
      "loss": 2.1811,
      "step": 145923
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.0686614513397217,
      "learning_rate": 0.00017797115960685198,
      "loss": 2.1278,
      "step": 145924
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.0807610750198364,
      "learning_rate": 0.00017796742275134664,
      "loss": 2.3161,
      "step": 145925
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.0846574306488037,
      "learning_rate": 0.00017796368591852955,
      "loss": 2.0443,
      "step": 145926
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.1501528024673462,
      "learning_rate": 0.00017795994910840132,
      "loss": 2.2382,
      "step": 145927
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.0529954433441162,
      "learning_rate": 0.0001779562123209626,
      "loss": 2.0256,
      "step": 145928
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.0746389627456665,
      "learning_rate": 0.00017795247555621406,
      "loss": 2.2827,
      "step": 145929
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.1488003730773926,
      "learning_rate": 0.0001779487388141565,
      "loss": 2.0248,
      "step": 145930
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.0181318521499634,
      "learning_rate": 0.0001779450020947906,
      "loss": 1.9268,
      "step": 145931
    },
    {
      "epoch": 1.9,
      "grad_norm": 0.9766772985458374,
      "learning_rate": 0.00017794126539811699,
      "loss": 2.1966,
      "step": 145932
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.049116611480713,
      "learning_rate": 0.00017793752872413647,
      "loss": 2.2814,
      "step": 145933
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.0618577003479004,
      "learning_rate": 0.0001779337920728496,
      "loss": 2.046,
      "step": 145934
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.0814639329910278,
      "learning_rate": 0.00017793005544425723,
      "loss": 2.1177,
      "step": 145935
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.0865037441253662,
      "learning_rate": 0.00017792631883835987,
      "loss": 2.2506,
      "step": 145936
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.0932178497314453,
      "learning_rate": 0.00017792258225515835,
      "loss": 2.0564,
      "step": 145937
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.2033859491348267,
      "learning_rate": 0.00017791884569465328,
      "loss": 2.2816,
      "step": 145938
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.2382603883743286,
      "learning_rate": 0.00017791510915684552,
      "loss": 2.1843,
      "step": 145939
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.2071926593780518,
      "learning_rate": 0.00017791137264173553,
      "loss": 1.893,
      "step": 145940
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.1721525192260742,
      "learning_rate": 0.00017790763614932415,
      "loss": 2.0998,
      "step": 145941
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.1592098474502563,
      "learning_rate": 0.00017790389967961207,
      "loss": 2.0309,
      "step": 145942
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.3071056604385376,
      "learning_rate": 0.00017790016323259995,
      "loss": 2.0067,
      "step": 145943
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.1624010801315308,
      "learning_rate": 0.00017789642680828845,
      "loss": 2.0276,
      "step": 145944
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.284483551979065,
      "learning_rate": 0.0001778926904066784,
      "loss": 2.01,
      "step": 145945
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.0550533533096313,
      "learning_rate": 0.00017788895402777035,
      "loss": 2.0245,
      "step": 145946
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.109485387802124,
      "learning_rate": 0.000177885217671565,
      "loss": 2.2418,
      "step": 145947
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.075973629951477,
      "learning_rate": 0.00017788148133806312,
      "loss": 2.0857,
      "step": 145948
    },
    {
      "epoch": 1.9,
      "grad_norm": 0.9724957346916199,
      "learning_rate": 0.00017787774502726547,
      "loss": 2.1321,
      "step": 145949
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.0571753978729248,
      "learning_rate": 0.00017787400873917256,
      "loss": 1.7801,
      "step": 145950
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.17848539352417,
      "learning_rate": 0.00017787027247378524,
      "loss": 2.0388,
      "step": 145951
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.0835342407226562,
      "learning_rate": 0.00017786653623110408,
      "loss": 2.0252,
      "step": 145952
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.2578063011169434,
      "learning_rate": 0.00017786280001112986,
      "loss": 2.3465,
      "step": 145953
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.1349568367004395,
      "learning_rate": 0.00017785906381386323,
      "loss": 1.9975,
      "step": 145954
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.220874547958374,
      "learning_rate": 0.0001778553276393049,
      "loss": 2.1437,
      "step": 145955
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.0434919595718384,
      "learning_rate": 0.00017785159148745564,
      "loss": 2.04,
      "step": 145956
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.0972524881362915,
      "learning_rate": 0.00017784785535831607,
      "loss": 2.0981,
      "step": 145957
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.1297006607055664,
      "learning_rate": 0.00017784411925188685,
      "loss": 2.1383,
      "step": 145958
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.1920033693313599,
      "learning_rate": 0.0001778403831681687,
      "loss": 2.0014,
      "step": 145959
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.243374228477478,
      "learning_rate": 0.0001778366471071623,
      "loss": 2.4049,
      "step": 145960
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.0198485851287842,
      "learning_rate": 0.0001778329110688684,
      "loss": 1.9781,
      "step": 145961
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.1757750511169434,
      "learning_rate": 0.00017782917505328768,
      "loss": 2.1634,
      "step": 145962
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.0228919982910156,
      "learning_rate": 0.0001778254390604209,
      "loss": 1.9347,
      "step": 145963
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.1728044748306274,
      "learning_rate": 0.00017782170309026859,
      "loss": 2.0241,
      "step": 145964
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.1952301263809204,
      "learning_rate": 0.00017781796714283154,
      "loss": 2.0722,
      "step": 145965
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.1143165826797485,
      "learning_rate": 0.00017781423121811043,
      "loss": 1.8967,
      "step": 145966
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.040442705154419,
      "learning_rate": 0.00017781049531610598,
      "loss": 2.1986,
      "step": 145967
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.0684336423873901,
      "learning_rate": 0.00017780675943681882,
      "loss": 2.1532,
      "step": 145968
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.0693694353103638,
      "learning_rate": 0.00017780302358024982,
      "loss": 2.2111,
      "step": 145969
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.1009933948516846,
      "learning_rate": 0.00017779928774639946,
      "loss": 1.8691,
      "step": 145970
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.1490205526351929,
      "learning_rate": 0.00017779555193526847,
      "loss": 2.1058,
      "step": 145971
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.1082855463027954,
      "learning_rate": 0.00017779181614685762,
      "loss": 2.0751,
      "step": 145972
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.1438555717468262,
      "learning_rate": 0.00017778808038116756,
      "loss": 2.0064,
      "step": 145973
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.260352373123169,
      "learning_rate": 0.00017778434463819906,
      "loss": 2.1954,
      "step": 145974
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.1594382524490356,
      "learning_rate": 0.0001777806089179528,
      "loss": 2.1165,
      "step": 145975
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.1566805839538574,
      "learning_rate": 0.00017777687322042938,
      "loss": 2.0712,
      "step": 145976
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.1145986318588257,
      "learning_rate": 0.0001777731375456295,
      "loss": 1.992,
      "step": 145977
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.3357325792312622,
      "learning_rate": 0.00017776940189355393,
      "loss": 2.0925,
      "step": 145978
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.2181066274642944,
      "learning_rate": 0.0001777656662642033,
      "loss": 1.9728,
      "step": 145979
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.0330806970596313,
      "learning_rate": 0.00017776193065757837,
      "loss": 2.1105,
      "step": 145980
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.11790132522583,
      "learning_rate": 0.00017775819507367992,
      "loss": 2.1263,
      "step": 145981
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.0752402544021606,
      "learning_rate": 0.00017775445951250839,
      "loss": 1.9541,
      "step": 145982
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.2583131790161133,
      "learning_rate": 0.00017775072397406463,
      "loss": 2.1766,
      "step": 145983
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.1339892148971558,
      "learning_rate": 0.00017774698845834933,
      "loss": 2.0949,
      "step": 145984
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.120106816291809,
      "learning_rate": 0.00017774325296536313,
      "loss": 2.0954,
      "step": 145985
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.1091053485870361,
      "learning_rate": 0.00017773951749510683,
      "loss": 2.2378,
      "step": 145986
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.1637301445007324,
      "learning_rate": 0.00017773578204758112,
      "loss": 2.238,
      "step": 145987
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.0675394535064697,
      "learning_rate": 0.00017773204662278658,
      "loss": 2.2115,
      "step": 145988
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.186950445175171,
      "learning_rate": 0.0001777283112207239,
      "loss": 2.1754,
      "step": 145989
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.0744177103042603,
      "learning_rate": 0.00017772457584139386,
      "loss": 2.371,
      "step": 145990
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.0778756141662598,
      "learning_rate": 0.00017772084048479716,
      "loss": 2.201,
      "step": 145991
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.3530043363571167,
      "learning_rate": 0.0001777171051509344,
      "loss": 2.073,
      "step": 145992
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.0636600255966187,
      "learning_rate": 0.0001777133698398065,
      "loss": 2.0702,
      "step": 145993
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.0793591737747192,
      "learning_rate": 0.00017770963455141387,
      "loss": 2.0191,
      "step": 145994
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.145980954170227,
      "learning_rate": 0.0001777058992857573,
      "loss": 2.0484,
      "step": 145995
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.1227953433990479,
      "learning_rate": 0.00017770216404283751,
      "loss": 1.9881,
      "step": 145996
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.1142892837524414,
      "learning_rate": 0.00017769842882265523,
      "loss": 1.9307,
      "step": 145997
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.1053093671798706,
      "learning_rate": 0.00017769469362521117,
      "loss": 2.1921,
      "step": 145998
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.1544057130813599,
      "learning_rate": 0.0001776909584505059,
      "loss": 2.2679,
      "step": 145999
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.1830086708068848,
      "learning_rate": 0.0001776872232985403,
      "loss": 2.182,
      "step": 146000
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.1851825714111328,
      "learning_rate": 0.00017768348816931485,
      "loss": 2.1386,
      "step": 146001
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.17635178565979,
      "learning_rate": 0.00017767975306283036,
      "loss": 2.2635,
      "step": 146002
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.396342158317566,
      "learning_rate": 0.00017767601797908752,
      "loss": 1.9456,
      "step": 146003
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.2088223695755005,
      "learning_rate": 0.00017767228291808702,
      "loss": 2.1418,
      "step": 146004
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.1463851928710938,
      "learning_rate": 0.00017766854787982952,
      "loss": 2.1197,
      "step": 146005
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.070731520652771,
      "learning_rate": 0.00017766481286431588,
      "loss": 1.9316,
      "step": 146006
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.2817035913467407,
      "learning_rate": 0.00017766107787154655,
      "loss": 1.9836,
      "step": 146007
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.1118178367614746,
      "learning_rate": 0.00017765734290152237,
      "loss": 2.0101,
      "step": 146008
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.1392043828964233,
      "learning_rate": 0.00017765360795424392,
      "loss": 1.9777,
      "step": 146009
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.1347365379333496,
      "learning_rate": 0.00017764987302971202,
      "loss": 2.1984,
      "step": 146010
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.1145989894866943,
      "learning_rate": 0.0001776461381279273,
      "loss": 1.9613,
      "step": 146011
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.1721149682998657,
      "learning_rate": 0.00017764240324889062,
      "loss": 2.2326,
      "step": 146012
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.1222059726715088,
      "learning_rate": 0.0001776386683926024,
      "loss": 2.0858,
      "step": 146013
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.0206071138381958,
      "learning_rate": 0.00017763493355906341,
      "loss": 1.9715,
      "step": 146014
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.0254318714141846,
      "learning_rate": 0.00017763119874827444,
      "loss": 2.0363,
      "step": 146015
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.0828356742858887,
      "learning_rate": 0.00017762746396023614,
      "loss": 2.1633,
      "step": 146016
    },
    {
      "epoch": 1.9,
      "grad_norm": 0.9895143508911133,
      "learning_rate": 0.00017762372919494926,
      "loss": 2.1246,
      "step": 146017
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.1382009983062744,
      "learning_rate": 0.0001776199944524144,
      "loss": 2.0701,
      "step": 146018
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.2557893991470337,
      "learning_rate": 0.00017761625973263236,
      "loss": 2.1576,
      "step": 146019
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.1322808265686035,
      "learning_rate": 0.00017761252503560368,
      "loss": 2.1052,
      "step": 146020
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.083181619644165,
      "learning_rate": 0.00017760879036132912,
      "loss": 2.1927,
      "step": 146021
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.3790687322616577,
      "learning_rate": 0.00017760505570980944,
      "loss": 1.9595,
      "step": 146022
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.1651091575622559,
      "learning_rate": 0.00017760132108104527,
      "loss": 1.8091,
      "step": 146023
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.0274266004562378,
      "learning_rate": 0.00017759758647503739,
      "loss": 2.0538,
      "step": 146024
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.1979084014892578,
      "learning_rate": 0.00017759385189178634,
      "loss": 2.1789,
      "step": 146025
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.3292372226715088,
      "learning_rate": 0.00017759011733129297,
      "loss": 2.0482,
      "step": 146026
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.2989484071731567,
      "learning_rate": 0.0001775863827935579,
      "loss": 1.9534,
      "step": 146027
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.0500952005386353,
      "learning_rate": 0.00017758264827858177,
      "loss": 2.022,
      "step": 146028
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.1721628904342651,
      "learning_rate": 0.00017757891378636538,
      "loss": 2.2643,
      "step": 146029
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.0851778984069824,
      "learning_rate": 0.00017757517931690943,
      "loss": 1.8485,
      "step": 146030
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.1204952001571655,
      "learning_rate": 0.0001775714448702145,
      "loss": 1.8506,
      "step": 146031
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.1925376653671265,
      "learning_rate": 0.00017756771044628137,
      "loss": 2.0149,
      "step": 146032
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.1047379970550537,
      "learning_rate": 0.0001775639760451107,
      "loss": 2.1237,
      "step": 146033
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.1006810665130615,
      "learning_rate": 0.00017756024166670322,
      "loss": 1.9963,
      "step": 146034
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.3883030414581299,
      "learning_rate": 0.00017755650731105958,
      "loss": 1.9063,
      "step": 146035
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.1645762920379639,
      "learning_rate": 0.00017755277297818055,
      "loss": 2.02,
      "step": 146036
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.184691071510315,
      "learning_rate": 0.00017754903866806673,
      "loss": 2.1156,
      "step": 146037
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.11552095413208,
      "learning_rate": 0.0001775453043807188,
      "loss": 2.0132,
      "step": 146038
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.0250558853149414,
      "learning_rate": 0.0001775415701161376,
      "loss": 1.9093,
      "step": 146039
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.1975563764572144,
      "learning_rate": 0.00017753783587432366,
      "loss": 2.0964,
      "step": 146040
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.1325947046279907,
      "learning_rate": 0.00017753410165527783,
      "loss": 1.8884,
      "step": 146041
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.3343676328659058,
      "learning_rate": 0.00017753036745900072,
      "loss": 2.0964,
      "step": 146042
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.532797932624817,
      "learning_rate": 0.00017752663328549298,
      "loss": 2.112,
      "step": 146043
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.1436011791229248,
      "learning_rate": 0.0001775228991347553,
      "loss": 2.159,
      "step": 146044
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.1816117763519287,
      "learning_rate": 0.00017751916500678848,
      "loss": 2.1305,
      "step": 146045
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.0698795318603516,
      "learning_rate": 0.00017751543090159313,
      "loss": 2.167,
      "step": 146046
    },
    {
      "epoch": 1.9,
      "grad_norm": 2.3539810180664062,
      "learning_rate": 0.00017751169681916998,
      "loss": 2.3219,
      "step": 146047
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.2993760108947754,
      "learning_rate": 0.00017750796275951985,
      "loss": 2.1744,
      "step": 146048
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.207742691040039,
      "learning_rate": 0.00017750422872264316,
      "loss": 2.1602,
      "step": 146049
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.1131079196929932,
      "learning_rate": 0.0001775004947085408,
      "loss": 2.054,
      "step": 146050
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.121869444847107,
      "learning_rate": 0.00017749676071721334,
      "loss": 1.8178,
      "step": 146051
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.1952675580978394,
      "learning_rate": 0.00017749302674866158,
      "loss": 1.9251,
      "step": 146052
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.1741780042648315,
      "learning_rate": 0.00017748929280288618,
      "loss": 1.9321,
      "step": 146053
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.2084884643554688,
      "learning_rate": 0.00017748555887988796,
      "loss": 2.2372,
      "step": 146054
    },
    {
      "epoch": 1.9,
      "grad_norm": 0.9968467354774475,
      "learning_rate": 0.00017748182497966735,
      "loss": 1.8954,
      "step": 146055
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.1513481140136719,
      "learning_rate": 0.00017747809110222517,
      "loss": 2.0594,
      "step": 146056
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.194451928138733,
      "learning_rate": 0.00017747435724756216,
      "loss": 2.165,
      "step": 146057
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.2080440521240234,
      "learning_rate": 0.00017747062341567896,
      "loss": 2.0226,
      "step": 146058
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.132143259048462,
      "learning_rate": 0.00017746688960657627,
      "loss": 1.9328,
      "step": 146059
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.2490884065628052,
      "learning_rate": 0.00017746315582025494,
      "loss": 2.0053,
      "step": 146060
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.1253587007522583,
      "learning_rate": 0.0001774594220567154,
      "loss": 2.1201,
      "step": 146061
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.1289676427841187,
      "learning_rate": 0.00017745568831595846,
      "loss": 2.0523,
      "step": 146062
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.194510817527771,
      "learning_rate": 0.00017745195459798483,
      "loss": 2.2913,
      "step": 146063
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.1475478410720825,
      "learning_rate": 0.00017744822090279521,
      "loss": 1.9342,
      "step": 146064
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.2435120344161987,
      "learning_rate": 0.00017744448723039022,
      "loss": 2.1227,
      "step": 146065
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.1348391771316528,
      "learning_rate": 0.00017744075358077068,
      "loss": 1.9566,
      "step": 146066
    },
    {
      "epoch": 1.9,
      "grad_norm": 0.9766319990158081,
      "learning_rate": 0.0001774370199539373,
      "loss": 2.0316,
      "step": 146067
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.085710048675537,
      "learning_rate": 0.00017743328634989058,
      "loss": 2.2199,
      "step": 146068
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.3030732870101929,
      "learning_rate": 0.00017742955276863132,
      "loss": 1.9817,
      "step": 146069
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.1224709749221802,
      "learning_rate": 0.00017742581921016023,
      "loss": 2.1856,
      "step": 146070
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.2500929832458496,
      "learning_rate": 0.00017742208567447798,
      "loss": 2.1133,
      "step": 146071
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.0942223072052002,
      "learning_rate": 0.0001774183521615853,
      "loss": 1.9226,
      "step": 146072
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.1239326000213623,
      "learning_rate": 0.00017741461867148297,
      "loss": 2.1955,
      "step": 146073
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.1786267757415771,
      "learning_rate": 0.00017741088520417146,
      "loss": 1.9282,
      "step": 146074
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.136027216911316,
      "learning_rate": 0.00017740715175965156,
      "loss": 2.1267,
      "step": 146075
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.3029959201812744,
      "learning_rate": 0.00017740341833792403,
      "loss": 2.1867,
      "step": 146076
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.0580427646636963,
      "learning_rate": 0.0001773996849389895,
      "loss": 2.1655,
      "step": 146077
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.10990309715271,
      "learning_rate": 0.00017739595156284865,
      "loss": 1.9288,
      "step": 146078
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.143477201461792,
      "learning_rate": 0.00017739221820950232,
      "loss": 2.1076,
      "step": 146079
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.0339455604553223,
      "learning_rate": 0.000177388484878951,
      "loss": 1.9806,
      "step": 146080
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.1265432834625244,
      "learning_rate": 0.00017738475157119546,
      "loss": 2.0924,
      "step": 146081
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.0665149688720703,
      "learning_rate": 0.00017738101828623643,
      "loss": 1.9296,
      "step": 146082
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.0304254293441772,
      "learning_rate": 0.00017737728502407462,
      "loss": 2.0727,
      "step": 146083
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.0991322994232178,
      "learning_rate": 0.00017737355178471065,
      "loss": 1.9857,
      "step": 146084
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.157477855682373,
      "learning_rate": 0.00017736981856814537,
      "loss": 2.2289,
      "step": 146085
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.1325756311416626,
      "learning_rate": 0.0001773660853743792,
      "loss": 2.1192,
      "step": 146086
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.139223337173462,
      "learning_rate": 0.00017736235220341303,
      "loss": 2.0328,
      "step": 146087
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.0858807563781738,
      "learning_rate": 0.00017735861905524749,
      "loss": 2.0838,
      "step": 146088
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.0677794218063354,
      "learning_rate": 0.0001773548859298833,
      "loss": 1.9835,
      "step": 146089
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.1002923250198364,
      "learning_rate": 0.00017735115282732118,
      "loss": 2.1288,
      "step": 146090
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.058829665184021,
      "learning_rate": 0.0001773474197475619,
      "loss": 2.0188,
      "step": 146091
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.2048661708831787,
      "learning_rate": 0.00017734368669060596,
      "loss": 2.2544,
      "step": 146092
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.0652234554290771,
      "learning_rate": 0.00017733995365645412,
      "loss": 2.0541,
      "step": 146093
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.0375758409500122,
      "learning_rate": 0.00017733622064510705,
      "loss": 2.0179,
      "step": 146094
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.1255170106887817,
      "learning_rate": 0.0001773324876565656,
      "loss": 2.0907,
      "step": 146095
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.2144018411636353,
      "learning_rate": 0.00017732875469083026,
      "loss": 1.8062,
      "step": 146096
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.096417784690857,
      "learning_rate": 0.00017732502174790194,
      "loss": 2.03,
      "step": 146097
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.123836874961853,
      "learning_rate": 0.00017732128882778114,
      "loss": 1.9566,
      "step": 146098
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.1308798789978027,
      "learning_rate": 0.00017731755593046862,
      "loss": 2.1039,
      "step": 146099
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.1695736646652222,
      "learning_rate": 0.00017731382305596509,
      "loss": 2.0638,
      "step": 146100
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.1071618795394897,
      "learning_rate": 0.00017731009020427124,
      "loss": 2.086,
      "step": 146101
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.0812582969665527,
      "learning_rate": 0.0001773063573753877,
      "loss": 1.7655,
      "step": 146102
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.1015992164611816,
      "learning_rate": 0.00017730262456931532,
      "loss": 2.0904,
      "step": 146103
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.0763758420944214,
      "learning_rate": 0.00017729889178605472,
      "loss": 1.9971,
      "step": 146104
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.0250164270401,
      "learning_rate": 0.0001772951590256065,
      "loss": 1.9814,
      "step": 146105
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.0639160871505737,
      "learning_rate": 0.00017729142628797144,
      "loss": 2.2424,
      "step": 146106
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.1838600635528564,
      "learning_rate": 0.0001772876935731502,
      "loss": 2.2175,
      "step": 146107
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.2043172121047974,
      "learning_rate": 0.00017728396088114348,
      "loss": 1.9827,
      "step": 146108
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.0864076614379883,
      "learning_rate": 0.00017728022821195209,
      "loss": 2.0955,
      "step": 146109
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.0497989654541016,
      "learning_rate": 0.00017727649556557653,
      "loss": 1.8787,
      "step": 146110
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.2373074293136597,
      "learning_rate": 0.00017727276294201766,
      "loss": 2.184,
      "step": 146111
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.0849875211715698,
      "learning_rate": 0.000177269030341276,
      "loss": 2.1372,
      "step": 146112
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.231053352355957,
      "learning_rate": 0.00017726529776335238,
      "loss": 2.2123,
      "step": 146113
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.098946452140808,
      "learning_rate": 0.00017726156520824746,
      "loss": 1.9298,
      "step": 146114
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.1468394994735718,
      "learning_rate": 0.000177257832675962,
      "loss": 1.9926,
      "step": 146115
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.026960849761963,
      "learning_rate": 0.00017725410016649658,
      "loss": 2.0429,
      "step": 146116
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.1366337537765503,
      "learning_rate": 0.00017725036767985187,
      "loss": 2.114,
      "step": 146117
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.1718207597732544,
      "learning_rate": 0.0001772466352160287,
      "loss": 2.0268,
      "step": 146118
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.2736198902130127,
      "learning_rate": 0.00017724290277502773,
      "loss": 2.0188,
      "step": 146119
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.2840951681137085,
      "learning_rate": 0.00017723917035684957,
      "loss": 2.3207,
      "step": 146120
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.1159768104553223,
      "learning_rate": 0.000177235437961495,
      "loss": 1.9176,
      "step": 146121
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.2126234769821167,
      "learning_rate": 0.00017723170558896464,
      "loss": 2.1607,
      "step": 146122
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.0280921459197998,
      "learning_rate": 0.00017722797323925925,
      "loss": 1.9705,
      "step": 146123
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.1767935752868652,
      "learning_rate": 0.00017722424091237945,
      "loss": 2.0954,
      "step": 146124
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.6980911493301392,
      "learning_rate": 0.00017722050860832604,
      "loss": 2.203,
      "step": 146125
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.1655186414718628,
      "learning_rate": 0.00017721677632709967,
      "loss": 2.0592,
      "step": 146126
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.1774271726608276,
      "learning_rate": 0.00017721304406870104,
      "loss": 1.9915,
      "step": 146127
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.1428996324539185,
      "learning_rate": 0.00017720931183313072,
      "loss": 2.1738,
      "step": 146128
    },
    {
      "epoch": 1.9,
      "grad_norm": 2.2991623878479004,
      "learning_rate": 0.00017720557962038955,
      "loss": 2.0271,
      "step": 146129
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.1664189100265503,
      "learning_rate": 0.00017720184743047815,
      "loss": 2.0858,
      "step": 146130
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.020001769065857,
      "learning_rate": 0.00017719811526339726,
      "loss": 2.0791,
      "step": 146131
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.1712466478347778,
      "learning_rate": 0.00017719438311914756,
      "loss": 2.0763,
      "step": 146132
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.1084924936294556,
      "learning_rate": 0.00017719065099772982,
      "loss": 2.1236,
      "step": 146133
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.1240628957748413,
      "learning_rate": 0.00017718691889914465,
      "loss": 1.9869,
      "step": 146134
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.1679850816726685,
      "learning_rate": 0.00017718318682339267,
      "loss": 2.1141,
      "step": 146135
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.3094958066940308,
      "learning_rate": 0.00017717945477047466,
      "loss": 2.1952,
      "step": 146136
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.0553267002105713,
      "learning_rate": 0.00017717572274039128,
      "loss": 2.1424,
      "step": 146137
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.0334925651550293,
      "learning_rate": 0.0001771719907331433,
      "loss": 2.0112,
      "step": 146138
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.341404676437378,
      "learning_rate": 0.00017716825874873135,
      "loss": 2.0017,
      "step": 146139
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.2038160562515259,
      "learning_rate": 0.00017716452678715624,
      "loss": 2.1341,
      "step": 146140
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.1444400548934937,
      "learning_rate": 0.00017716079484841847,
      "loss": 2.275,
      "step": 146141
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.111776351928711,
      "learning_rate": 0.00017715706293251884,
      "loss": 2.1262,
      "step": 146142
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.0989583730697632,
      "learning_rate": 0.000177153331039458,
      "loss": 1.9629,
      "step": 146143
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.1282885074615479,
      "learning_rate": 0.00017714959916923667,
      "loss": 2.1086,
      "step": 146144
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.2920255661010742,
      "learning_rate": 0.00017714586732185556,
      "loss": 2.021,
      "step": 146145
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.1209641695022583,
      "learning_rate": 0.0001771421354973155,
      "loss": 2.1903,
      "step": 146146
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.185236930847168,
      "learning_rate": 0.0001771384036956169,
      "loss": 2.1565,
      "step": 146147
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.1445362567901611,
      "learning_rate": 0.00017713467191676055,
      "loss": 2.1946,
      "step": 146148
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.0460885763168335,
      "learning_rate": 0.00017713094016074722,
      "loss": 1.9888,
      "step": 146149
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.2595969438552856,
      "learning_rate": 0.00017712720842757757,
      "loss": 2.1055,
      "step": 146150
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.1851528882980347,
      "learning_rate": 0.0001771234767172523,
      "loss": 2.1437,
      "step": 146151
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.0874265432357788,
      "learning_rate": 0.00017711974502977217,
      "loss": 2.166,
      "step": 146152
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.0346553325653076,
      "learning_rate": 0.0001771160133651377,
      "loss": 1.9554,
      "step": 146153
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.1478514671325684,
      "learning_rate": 0.0001771122817233497,
      "loss": 2.2766,
      "step": 146154
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.038541316986084,
      "learning_rate": 0.00017710855010440882,
      "loss": 1.7695,
      "step": 146155
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.0300896167755127,
      "learning_rate": 0.00017710481850831582,
      "loss": 1.8972,
      "step": 146156
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.1804972887039185,
      "learning_rate": 0.00017710108693507132,
      "loss": 2.1085,
      "step": 146157
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.0871552228927612,
      "learning_rate": 0.0001770973553846762,
      "loss": 1.9404,
      "step": 146158
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.1142982244491577,
      "learning_rate": 0.00017709362385713085,
      "loss": 2.3024,
      "step": 146159
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.1055694818496704,
      "learning_rate": 0.0001770898923524361,
      "loss": 1.9641,
      "step": 146160
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.1797667741775513,
      "learning_rate": 0.0001770861608705927,
      "loss": 1.9754,
      "step": 146161
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.10036039352417,
      "learning_rate": 0.00017708242941160128,
      "loss": 2.0019,
      "step": 146162
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.0778441429138184,
      "learning_rate": 0.00017707869797546256,
      "loss": 1.8669,
      "step": 146163
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.1639189720153809,
      "learning_rate": 0.00017707496656217735,
      "loss": 2.2417,
      "step": 146164
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.016198754310608,
      "learning_rate": 0.00017707123517174612,
      "loss": 2.2484,
      "step": 146165
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.1415361166000366,
      "learning_rate": 0.00017706750380416962,
      "loss": 1.8632,
      "step": 146166
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.1088916063308716,
      "learning_rate": 0.00017706377245944866,
      "loss": 2.0274,
      "step": 146167
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.0600903034210205,
      "learning_rate": 0.0001770600411375838,
      "loss": 2.0591,
      "step": 146168
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.0662628412246704,
      "learning_rate": 0.00017705630983857584,
      "loss": 2.116,
      "step": 146169
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.143462896347046,
      "learning_rate": 0.0001770525785624256,
      "loss": 2.0809,
      "step": 146170
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.235968828201294,
      "learning_rate": 0.00017704884730913338,
      "loss": 2.231,
      "step": 146171
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.0418506860733032,
      "learning_rate": 0.00017704511607870014,
      "loss": 1.9684,
      "step": 146172
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.1252301931381226,
      "learning_rate": 0.00017704138487112657,
      "loss": 2.0309,
      "step": 146173
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.2155380249023438,
      "learning_rate": 0.00017703765368641333,
      "loss": 2.0629,
      "step": 146174
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.1968566179275513,
      "learning_rate": 0.00017703392252456103,
      "loss": 2.1646,
      "step": 146175
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.066075086593628,
      "learning_rate": 0.00017703019138557067,
      "loss": 2.0961,
      "step": 146176
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.043007731437683,
      "learning_rate": 0.00017702646026944254,
      "loss": 2.0911,
      "step": 146177
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.1443030834197998,
      "learning_rate": 0.00017702272917617751,
      "loss": 1.841,
      "step": 146178
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.1545404195785522,
      "learning_rate": 0.00017701899810577628,
      "loss": 2.1279,
      "step": 146179
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.0779107809066772,
      "learning_rate": 0.00017701526705823958,
      "loss": 2.0172,
      "step": 146180
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.2019455432891846,
      "learning_rate": 0.000177011536033568,
      "loss": 2.0004,
      "step": 146181
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.1025617122650146,
      "learning_rate": 0.00017700780503176244,
      "loss": 2.1242,
      "step": 146182
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.0851984024047852,
      "learning_rate": 0.00017700407405282334,
      "loss": 1.9859,
      "step": 146183
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.1449302434921265,
      "learning_rate": 0.0001770003430967515,
      "loss": 2.069,
      "step": 146184
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.1450737714767456,
      "learning_rate": 0.00017699661216354765,
      "loss": 1.9444,
      "step": 146185
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.0576050281524658,
      "learning_rate": 0.0001769928812532124,
      "loss": 2.1845,
      "step": 146186
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.008119821548462,
      "learning_rate": 0.00017698915036574653,
      "loss": 1.8983,
      "step": 146187
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.0085090398788452,
      "learning_rate": 0.00017698541950115072,
      "loss": 1.9617,
      "step": 146188
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.1603529453277588,
      "learning_rate": 0.00017698168865942567,
      "loss": 2.3308,
      "step": 146189
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.1253687143325806,
      "learning_rate": 0.00017697795784057198,
      "loss": 2.184,
      "step": 146190
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.1359026432037354,
      "learning_rate": 0.00017697422704459042,
      "loss": 1.9981,
      "step": 146191
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.1013200283050537,
      "learning_rate": 0.00017697049627148165,
      "loss": 2.2016,
      "step": 146192
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.1328016519546509,
      "learning_rate": 0.0001769667655212464,
      "loss": 1.9944,
      "step": 146193
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.4643068313598633,
      "learning_rate": 0.00017696303479388542,
      "loss": 2.425,
      "step": 146194
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.164188027381897,
      "learning_rate": 0.00017695930408939928,
      "loss": 2.0056,
      "step": 146195
    },
    {
      "epoch": 1.9,
      "grad_norm": 0.9645898342132568,
      "learning_rate": 0.00017695557340778877,
      "loss": 2.2249,
      "step": 146196
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.085598111152649,
      "learning_rate": 0.0001769518427490545,
      "loss": 2.2554,
      "step": 146197
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.451153039932251,
      "learning_rate": 0.00017694811211319719,
      "loss": 2.2011,
      "step": 146198
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.1203032732009888,
      "learning_rate": 0.00017694438150021752,
      "loss": 1.9466,
      "step": 146199
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.126363754272461,
      "learning_rate": 0.00017694065091011628,
      "loss": 2.0979,
      "step": 146200
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.075764775276184,
      "learning_rate": 0.00017693692034289411,
      "loss": 2.2,
      "step": 146201
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.0204999446868896,
      "learning_rate": 0.00017693318979855165,
      "loss": 1.9788,
      "step": 146202
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.2065595388412476,
      "learning_rate": 0.00017692945927708968,
      "loss": 1.917,
      "step": 146203
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.081295371055603,
      "learning_rate": 0.0001769257287785088,
      "loss": 2.0886,
      "step": 146204
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.0560609102249146,
      "learning_rate": 0.00017692199830280972,
      "loss": 2.1988,
      "step": 146205
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.1394699811935425,
      "learning_rate": 0.0001769182678499932,
      "loss": 2.025,
      "step": 146206
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.3867319822311401,
      "learning_rate": 0.00017691453742005995,
      "loss": 2.0959,
      "step": 146207
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.2465126514434814,
      "learning_rate": 0.0001769108070130105,
      "loss": 2.1449,
      "step": 146208
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.0924468040466309,
      "learning_rate": 0.00017690707662884573,
      "loss": 2.1081,
      "step": 146209
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.1746395826339722,
      "learning_rate": 0.00017690334626756625,
      "loss": 2.0783,
      "step": 146210
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.131913423538208,
      "learning_rate": 0.00017689961592917274,
      "loss": 1.9176,
      "step": 146211
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.101755976676941,
      "learning_rate": 0.00017689588561366592,
      "loss": 2.0451,
      "step": 146212
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.102099061012268,
      "learning_rate": 0.00017689215532104652,
      "loss": 2.0725,
      "step": 146213
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.281249761581421,
      "learning_rate": 0.00017688842505131515,
      "loss": 2.0069,
      "step": 146214
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.1742122173309326,
      "learning_rate": 0.00017688469480447254,
      "loss": 2.0078,
      "step": 146215
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.1899508237838745,
      "learning_rate": 0.0001768809645805194,
      "loss": 2.0644,
      "step": 146216
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.1090989112854004,
      "learning_rate": 0.0001768772343794564,
      "loss": 2.0635,
      "step": 146217
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.0931333303451538,
      "learning_rate": 0.0001768735042012843,
      "loss": 2.0969,
      "step": 146218
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.1570183038711548,
      "learning_rate": 0.00017686977404600377,
      "loss": 1.8945,
      "step": 146219
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.0547325611114502,
      "learning_rate": 0.00017686604391361536,
      "loss": 2.0628,
      "step": 146220
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.1944665908813477,
      "learning_rate": 0.00017686231380411994,
      "loss": 2.0956,
      "step": 146221
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.1332601308822632,
      "learning_rate": 0.00017685858371751807,
      "loss": 2.3648,
      "step": 146222
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.0392719507217407,
      "learning_rate": 0.00017685485365381056,
      "loss": 2.0224,
      "step": 146223
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.2730159759521484,
      "learning_rate": 0.00017685112361299807,
      "loss": 2.0603,
      "step": 146224
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.2644727230072021,
      "learning_rate": 0.00017684739359508135,
      "loss": 1.8855,
      "step": 146225
    },
    {
      "epoch": 1.9,
      "grad_norm": 0.996788501739502,
      "learning_rate": 0.00017684366360006096,
      "loss": 2.0763,
      "step": 146226
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.1213676929473877,
      "learning_rate": 0.0001768399336279376,
      "loss": 2.1038,
      "step": 146227
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.4781798124313354,
      "learning_rate": 0.00017683620367871207,
      "loss": 2.0554,
      "step": 146228
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.222912311553955,
      "learning_rate": 0.000176832473752385,
      "loss": 2.1313,
      "step": 146229
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.05460524559021,
      "learning_rate": 0.0001768287438489571,
      "loss": 2.1727,
      "step": 146230
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.0566331148147583,
      "learning_rate": 0.00017682501396842914,
      "loss": 2.0478,
      "step": 146231
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.3069037199020386,
      "learning_rate": 0.00017682128411080166,
      "loss": 2.116,
      "step": 146232
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.113825798034668,
      "learning_rate": 0.00017681755427607543,
      "loss": 2.0142,
      "step": 146233
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.0773723125457764,
      "learning_rate": 0.00017681382446425113,
      "loss": 2.122,
      "step": 146234
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.1665408611297607,
      "learning_rate": 0.00017681009467532946,
      "loss": 2.197,
      "step": 146235
    },
    {
      "epoch": 1.9,
      "grad_norm": 0.9957302808761597,
      "learning_rate": 0.0001768063649093111,
      "loss": 2.0546,
      "step": 146236
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.0996390581130981,
      "learning_rate": 0.00017680263516619695,
      "loss": 2.0173,
      "step": 146237
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.0747363567352295,
      "learning_rate": 0.00017679890544598735,
      "loss": 2.0913,
      "step": 146238
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.036118507385254,
      "learning_rate": 0.00017679517574868315,
      "loss": 2.1492,
      "step": 146239
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.182157039642334,
      "learning_rate": 0.00017679144607428508,
      "loss": 2.0643,
      "step": 146240
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.154718041419983,
      "learning_rate": 0.00017678771642279376,
      "loss": 1.9621,
      "step": 146241
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.0385338068008423,
      "learning_rate": 0.00017678398679421,
      "loss": 2.155,
      "step": 146242
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.0522606372833252,
      "learning_rate": 0.00017678025718853446,
      "loss": 2.007,
      "step": 146243
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.3563092947006226,
      "learning_rate": 0.00017677652760576774,
      "loss": 2.4593,
      "step": 146244
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.1013727188110352,
      "learning_rate": 0.00017677279804591056,
      "loss": 1.8466,
      "step": 146245
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.114366888999939,
      "learning_rate": 0.00017676906850896365,
      "loss": 2.2267,
      "step": 146246
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.2086395025253296,
      "learning_rate": 0.0001767653389949277,
      "loss": 2.0691,
      "step": 146247
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.0939393043518066,
      "learning_rate": 0.0001767616095038034,
      "loss": 2.1447,
      "step": 146248
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.0921287536621094,
      "learning_rate": 0.00017675788003559156,
      "loss": 1.9252,
      "step": 146249
    },
    {
      "epoch": 1.9,
      "grad_norm": 0.9916548132896423,
      "learning_rate": 0.00017675415059029262,
      "loss": 2.0993,
      "step": 146250
    },
    {
      "epoch": 1.9,
      "grad_norm": 0.9963333606719971,
      "learning_rate": 0.00017675042116790745,
      "loss": 2.0525,
      "step": 146251
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.0553902387619019,
      "learning_rate": 0.00017674669176843667,
      "loss": 2.3254,
      "step": 146252
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.0014238357543945,
      "learning_rate": 0.00017674296239188103,
      "loss": 2.0116,
      "step": 146253
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.0889792442321777,
      "learning_rate": 0.00017673923303824116,
      "loss": 2.1017,
      "step": 146254
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.3301795721054077,
      "learning_rate": 0.00017673550370751796,
      "loss": 2.333,
      "step": 146255
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.0893540382385254,
      "learning_rate": 0.0001767317743997118,
      "loss": 2.1755,
      "step": 146256
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.1131798028945923,
      "learning_rate": 0.00017672804511482357,
      "loss": 2.0758,
      "step": 146257
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.1908280849456787,
      "learning_rate": 0.00017672431585285387,
      "loss": 2.1225,
      "step": 146258
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.07335364818573,
      "learning_rate": 0.00017672058661380351,
      "loss": 2.0867,
      "step": 146259
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.169260025024414,
      "learning_rate": 0.00017671685739767306,
      "loss": 2.1,
      "step": 146260
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.0221710205078125,
      "learning_rate": 0.00017671312820446343,
      "loss": 2.0118,
      "step": 146261
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.1588181257247925,
      "learning_rate": 0.000176709399034175,
      "loss": 2.4437,
      "step": 146262
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.1259249448776245,
      "learning_rate": 0.00017670566988680866,
      "loss": 2.3002,
      "step": 146263
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.1358122825622559,
      "learning_rate": 0.00017670194076236506,
      "loss": 2.0139,
      "step": 146264
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.1320830583572388,
      "learning_rate": 0.0001766982116608449,
      "loss": 2.1596,
      "step": 146265
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.0864567756652832,
      "learning_rate": 0.00017669448258224882,
      "loss": 1.9729,
      "step": 146266
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.1351523399353027,
      "learning_rate": 0.0001766907535265776,
      "loss": 1.9622,
      "step": 146267
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.056504249572754,
      "learning_rate": 0.00017668702449383206,
      "loss": 2.2217,
      "step": 146268
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.0834226608276367,
      "learning_rate": 0.00017668329548401252,
      "loss": 2.0146,
      "step": 146269
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.0941247940063477,
      "learning_rate": 0.00017667956649711995,
      "loss": 2.0233,
      "step": 146270
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.1473671197891235,
      "learning_rate": 0.00017667583753315497,
      "loss": 2.002,
      "step": 146271
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.0984065532684326,
      "learning_rate": 0.00017667210859211827,
      "loss": 2.1541,
      "step": 146272
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.2254990339279175,
      "learning_rate": 0.00017666837967401052,
      "loss": 2.2045,
      "step": 146273
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.2694963216781616,
      "learning_rate": 0.00017666465077883258,
      "loss": 1.9479,
      "step": 146274
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.5268713235855103,
      "learning_rate": 0.0001766609219065849,
      "loss": 2.1565,
      "step": 146275
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.0706145763397217,
      "learning_rate": 0.00017665719305726832,
      "loss": 2.2443,
      "step": 146276
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.0516881942749023,
      "learning_rate": 0.00017665346423088348,
      "loss": 2.0389,
      "step": 146277
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.2612040042877197,
      "learning_rate": 0.00017664973542743105,
      "loss": 2.2405,
      "step": 146278
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.1400482654571533,
      "learning_rate": 0.00017664600664691178,
      "loss": 1.9172,
      "step": 146279
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.124809741973877,
      "learning_rate": 0.0001766422778893264,
      "loss": 2.0011,
      "step": 146280
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.0472661256790161,
      "learning_rate": 0.00017663854915467558,
      "loss": 2.1595,
      "step": 146281
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.147470235824585,
      "learning_rate": 0.0001766348204429599,
      "loss": 1.9713,
      "step": 146282
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.0317331552505493,
      "learning_rate": 0.0001766310917541801,
      "loss": 2.4033,
      "step": 146283
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.0370819568634033,
      "learning_rate": 0.00017662736308833696,
      "loss": 1.966,
      "step": 146284
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.157448172569275,
      "learning_rate": 0.00017662363444543113,
      "loss": 2.1375,
      "step": 146285
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.0524977445602417,
      "learning_rate": 0.00017661990582546334,
      "loss": 2.0525,
      "step": 146286
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.1871824264526367,
      "learning_rate": 0.00017661617722843415,
      "loss": 2.2033,
      "step": 146287
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.0080180168151855,
      "learning_rate": 0.00017661244865434445,
      "loss": 2.1231,
      "step": 146288
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.0402761697769165,
      "learning_rate": 0.0001766087201031947,
      "loss": 1.9392,
      "step": 146289
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.0627284049987793,
      "learning_rate": 0.0001766049915749858,
      "loss": 2.0913,
      "step": 146290
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.221784234046936,
      "learning_rate": 0.0001766012630697183,
      "loss": 1.9488,
      "step": 146291
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.1681983470916748,
      "learning_rate": 0.00017659753458739306,
      "loss": 2.2474,
      "step": 146292
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.1182836294174194,
      "learning_rate": 0.00017659380612801054,
      "loss": 2.0254,
      "step": 146293
    },
    {
      "epoch": 1.9,
      "grad_norm": 0.9673587679862976,
      "learning_rate": 0.00017659007769157164,
      "loss": 2.0201,
      "step": 146294
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.238681674003601,
      "learning_rate": 0.00017658634927807695,
      "loss": 1.9454,
      "step": 146295
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.1734514236450195,
      "learning_rate": 0.00017658262088752718,
      "loss": 2.0475,
      "step": 146296
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.108826994895935,
      "learning_rate": 0.00017657889251992306,
      "loss": 2.3087,
      "step": 146297
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.0626815557479858,
      "learning_rate": 0.00017657516417526528,
      "loss": 1.9799,
      "step": 146298
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.1629878282546997,
      "learning_rate": 0.00017657143585355443,
      "loss": 2.1428,
      "step": 146299
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.0689891576766968,
      "learning_rate": 0.00017656770755479127,
      "loss": 2.2135,
      "step": 146300
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.1417559385299683,
      "learning_rate": 0.0001765639792789765,
      "loss": 2.0974,
      "step": 146301
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.2066336870193481,
      "learning_rate": 0.00017656025102611095,
      "loss": 1.9343,
      "step": 146302
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.1306532621383667,
      "learning_rate": 0.00017655652279619506,
      "loss": 2.32,
      "step": 146303
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.0842170715332031,
      "learning_rate": 0.0001765527945892297,
      "loss": 1.9488,
      "step": 146304
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.1163166761398315,
      "learning_rate": 0.00017654906640521547,
      "loss": 2.0731,
      "step": 146305
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.2280117273330688,
      "learning_rate": 0.0001765453382441531,
      "loss": 2.1336,
      "step": 146306
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.0500553846359253,
      "learning_rate": 0.0001765416101060433,
      "loss": 2.135,
      "step": 146307
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.1499450206756592,
      "learning_rate": 0.0001765378819908867,
      "loss": 2.2218,
      "step": 146308
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.1267081499099731,
      "learning_rate": 0.00017653415389868405,
      "loss": 2.3826,
      "step": 146309
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.0800085067749023,
      "learning_rate": 0.00017653042582943618,
      "loss": 1.9405,
      "step": 146310
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.1216769218444824,
      "learning_rate": 0.00017652669778314348,
      "loss": 2.067,
      "step": 146311
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.162533164024353,
      "learning_rate": 0.0001765229697598068,
      "loss": 1.9629,
      "step": 146312
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.0389541387557983,
      "learning_rate": 0.00017651924175942683,
      "loss": 1.8091,
      "step": 146313
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.093559741973877,
      "learning_rate": 0.00017651551378200427,
      "loss": 2.1394,
      "step": 146314
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.4400296211242676,
      "learning_rate": 0.0001765117858275398,
      "loss": 2.0053,
      "step": 146315
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.2167108058929443,
      "learning_rate": 0.0001765080578960343,
      "loss": 1.9688,
      "step": 146316
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.0819734334945679,
      "learning_rate": 0.0001765043299874881,
      "loss": 1.8856,
      "step": 146317
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.137756109237671,
      "learning_rate": 0.00017650060210190206,
      "loss": 2.1569,
      "step": 146318
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.2260768413543701,
      "learning_rate": 0.00017649687423927694,
      "loss": 1.9671,
      "step": 146319
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.1287342309951782,
      "learning_rate": 0.00017649314639961336,
      "loss": 2.1095,
      "step": 146320
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.1990456581115723,
      "learning_rate": 0.00017648941858291204,
      "loss": 2.1333,
      "step": 146321
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.115478277206421,
      "learning_rate": 0.00017648569078917376,
      "loss": 2.1999,
      "step": 146322
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.3005797863006592,
      "learning_rate": 0.00017648196301839903,
      "loss": 2.093,
      "step": 146323
    },
    {
      "epoch": 1.91,
      "grad_norm": 0.9973210692405701,
      "learning_rate": 0.00017647823527058867,
      "loss": 2.0255,
      "step": 146324
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.1100711822509766,
      "learning_rate": 0.00017647450754574326,
      "loss": 2.015,
      "step": 146325
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.0011788606643677,
      "learning_rate": 0.0001764707798438636,
      "loss": 1.9863,
      "step": 146326
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.1268765926361084,
      "learning_rate": 0.0001764670521649504,
      "loss": 2.232,
      "step": 146327
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.1508705615997314,
      "learning_rate": 0.00017646332450900435,
      "loss": 2.0198,
      "step": 146328
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.1370834112167358,
      "learning_rate": 0.00017645959687602607,
      "loss": 2.0688,
      "step": 146329
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.091739535331726,
      "learning_rate": 0.00017645586926601621,
      "loss": 1.9894,
      "step": 146330
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.1133803129196167,
      "learning_rate": 0.00017645214167897556,
      "loss": 1.9948,
      "step": 146331
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.0897561311721802,
      "learning_rate": 0.00017644841411490478,
      "loss": 2.1159,
      "step": 146332
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.1815897226333618,
      "learning_rate": 0.00017644468657380458,
      "loss": 1.9724,
      "step": 146333
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.0924484729766846,
      "learning_rate": 0.00017644095905567565,
      "loss": 2.1939,
      "step": 146334
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.1106516122817993,
      "learning_rate": 0.0001764372315605188,
      "loss": 1.8813,
      "step": 146335
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.268190622329712,
      "learning_rate": 0.00017643350408833445,
      "loss": 2.0337,
      "step": 146336
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.0556261539459229,
      "learning_rate": 0.00017642977663912348,
      "loss": 1.985,
      "step": 146337
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.0534534454345703,
      "learning_rate": 0.00017642604921288654,
      "loss": 2.1885,
      "step": 146338
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.154495358467102,
      "learning_rate": 0.00017642232180962433,
      "loss": 2.1727,
      "step": 146339
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.0633989572525024,
      "learning_rate": 0.00017641859442933756,
      "loss": 2.1022,
      "step": 146340
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.1075658798217773,
      "learning_rate": 0.000176414867072027,
      "loss": 2.0726,
      "step": 146341
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.1361651420593262,
      "learning_rate": 0.00017641113973769309,
      "loss": 2.1771,
      "step": 146342
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.1171071529388428,
      "learning_rate": 0.00017640741242633674,
      "loss": 2.1998,
      "step": 146343
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.0940382480621338,
      "learning_rate": 0.0001764036851379586,
      "loss": 2.017,
      "step": 146344
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.0233787298202515,
      "learning_rate": 0.0001763999578725593,
      "loss": 2.0803,
      "step": 146345
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.1030199527740479,
      "learning_rate": 0.0001763962306301396,
      "loss": 1.9998,
      "step": 146346
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.1450698375701904,
      "learning_rate": 0.0001763925034107003,
      "loss": 2.1944,
      "step": 146347
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.0778394937515259,
      "learning_rate": 0.00017638877621424184,
      "loss": 2.0733,
      "step": 146348
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.097515344619751,
      "learning_rate": 0.00017638504904076503,
      "loss": 1.9733,
      "step": 146349
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.0904418230056763,
      "learning_rate": 0.0001763813218902706,
      "loss": 2.0743,
      "step": 146350
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.03681480884552,
      "learning_rate": 0.0001763775947627592,
      "loss": 1.9699,
      "step": 146351
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.0590397119522095,
      "learning_rate": 0.00017637386765823157,
      "loss": 1.9827,
      "step": 146352
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.1268727779388428,
      "learning_rate": 0.00017637014057668844,
      "loss": 2.1689,
      "step": 146353
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.0486092567443848,
      "learning_rate": 0.0001763664135181303,
      "loss": 2.0899,
      "step": 146354
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.2747454643249512,
      "learning_rate": 0.00017636268648255803,
      "loss": 2.2846,
      "step": 146355
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.0845468044281006,
      "learning_rate": 0.00017635895946997227,
      "loss": 2.2013,
      "step": 146356
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.0368832349777222,
      "learning_rate": 0.00017635523248037366,
      "loss": 2.1815,
      "step": 146357
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.4800270795822144,
      "learning_rate": 0.000176351505513763,
      "loss": 2.1972,
      "step": 146358
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.178808569908142,
      "learning_rate": 0.000176347778570141,
      "loss": 1.937,
      "step": 146359
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.1287915706634521,
      "learning_rate": 0.00017634405164950826,
      "loss": 1.9896,
      "step": 146360
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.1550016403198242,
      "learning_rate": 0.0001763403247518654,
      "loss": 2.1442,
      "step": 146361
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.1483157873153687,
      "learning_rate": 0.00017633659787721323,
      "loss": 1.95,
      "step": 146362
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.064319133758545,
      "learning_rate": 0.00017633287102555246,
      "loss": 2.0751,
      "step": 146363
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.1261029243469238,
      "learning_rate": 0.00017632914419688368,
      "loss": 2.223,
      "step": 146364
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.138737678527832,
      "learning_rate": 0.00017632541739120777,
      "loss": 2.0502,
      "step": 146365
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.1351900100708008,
      "learning_rate": 0.00017632169060852524,
      "loss": 2.0656,
      "step": 146366
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.0377713441848755,
      "learning_rate": 0.00017631796384883682,
      "loss": 2.1841,
      "step": 146367
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.1343528032302856,
      "learning_rate": 0.00017631423711214319,
      "loss": 2.0036,
      "step": 146368
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.092073678970337,
      "learning_rate": 0.00017631051039844513,
      "loss": 2.0162,
      "step": 146369
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.0247381925582886,
      "learning_rate": 0.00017630678370774322,
      "loss": 2.0852,
      "step": 146370
    },
    {
      "epoch": 1.91,
      "grad_norm": 0.9846601486206055,
      "learning_rate": 0.00017630305704003832,
      "loss": 1.9729,
      "step": 146371
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.140138030052185,
      "learning_rate": 0.000176299330395331,
      "loss": 2.1513,
      "step": 146372
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.1042659282684326,
      "learning_rate": 0.00017629560377362192,
      "loss": 2.1716,
      "step": 146373
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.1338144540786743,
      "learning_rate": 0.0001762918771749118,
      "loss": 2.1459,
      "step": 146374
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.0897035598754883,
      "learning_rate": 0.00017628815059920138,
      "loss": 1.9701,
      "step": 146375
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.1322579383850098,
      "learning_rate": 0.0001762844240464913,
      "loss": 2.0465,
      "step": 146376
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.339565634727478,
      "learning_rate": 0.00017628069751678236,
      "loss": 2.1006,
      "step": 146377
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.1289348602294922,
      "learning_rate": 0.00017627697101007514,
      "loss": 2.074,
      "step": 146378
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.0140323638916016,
      "learning_rate": 0.00017627324452637033,
      "loss": 1.973,
      "step": 146379
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.0576300621032715,
      "learning_rate": 0.0001762695180656687,
      "loss": 2.118,
      "step": 146380
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.2421813011169434,
      "learning_rate": 0.0001762657916279709,
      "loss": 2.1552,
      "step": 146381
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.3094013929367065,
      "learning_rate": 0.00017626206521327762,
      "loss": 2.0909,
      "step": 146382
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.0333013534545898,
      "learning_rate": 0.00017625833882158956,
      "loss": 2.1574,
      "step": 146383
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.2458674907684326,
      "learning_rate": 0.00017625461245290738,
      "loss": 2.2278,
      "step": 146384
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.168257713317871,
      "learning_rate": 0.0001762508861072318,
      "loss": 1.8701,
      "step": 146385
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.3498774766921997,
      "learning_rate": 0.00017624715978456355,
      "loss": 2.1015,
      "step": 146386
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.1111100912094116,
      "learning_rate": 0.00017624343348490333,
      "loss": 2.1469,
      "step": 146387
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.1260576248168945,
      "learning_rate": 0.00017623970720825172,
      "loss": 2.007,
      "step": 146388
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.2876547574996948,
      "learning_rate": 0.00017623598095460954,
      "loss": 1.9189,
      "step": 146389
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.258608341217041,
      "learning_rate": 0.0001762322547239774,
      "loss": 2.2408,
      "step": 146390
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.0538649559020996,
      "learning_rate": 0.00017622852851635596,
      "loss": 1.9011,
      "step": 146391
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.099294900894165,
      "learning_rate": 0.00017622480233174602,
      "loss": 2.0623,
      "step": 146392
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.264831781387329,
      "learning_rate": 0.0001762210761701482,
      "loss": 2.1664,
      "step": 146393
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.3804868459701538,
      "learning_rate": 0.00017621735003156336,
      "loss": 2.2074,
      "step": 146394
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.0579230785369873,
      "learning_rate": 0.00017621362391599197,
      "loss": 2.4258,
      "step": 146395
    },
    {
      "epoch": 1.91,
      "grad_norm": 0.9906817078590393,
      "learning_rate": 0.00017620989782343474,
      "loss": 2.079,
      "step": 146396
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.205579400062561,
      "learning_rate": 0.00017620617175389245,
      "loss": 1.969,
      "step": 146397
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.1321669816970825,
      "learning_rate": 0.00017620244570736577,
      "loss": 2.1236,
      "step": 146398
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.2340725660324097,
      "learning_rate": 0.0001761987196838554,
      "loss": 2.1603,
      "step": 146399
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.1084245443344116,
      "learning_rate": 0.00017619499368336198,
      "loss": 2.1319,
      "step": 146400
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.2303316593170166,
      "learning_rate": 0.00017619126770588637,
      "loss": 2.1657,
      "step": 146401
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.0990649461746216,
      "learning_rate": 0.0001761875417514291,
      "loss": 2.1137,
      "step": 146402
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.1393636465072632,
      "learning_rate": 0.00017618381581999086,
      "loss": 1.9972,
      "step": 146403
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.0742018222808838,
      "learning_rate": 0.00017618008991157237,
      "loss": 2.0644,
      "step": 146404
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.0424102544784546,
      "learning_rate": 0.00017617636402617433,
      "loss": 2.0646,
      "step": 146405
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.279211401939392,
      "learning_rate": 0.00017617263816379746,
      "loss": 2.1835,
      "step": 146406
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.1548948287963867,
      "learning_rate": 0.00017616891232444246,
      "loss": 2.0255,
      "step": 146407
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.2114344835281372,
      "learning_rate": 0.00017616518650811008,
      "loss": 2.1089,
      "step": 146408
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.1283446550369263,
      "learning_rate": 0.00017616146071480082,
      "loss": 2.0129,
      "step": 146409
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.1796895265579224,
      "learning_rate": 0.00017615773494451546,
      "loss": 2.1241,
      "step": 146410
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.1885873079299927,
      "learning_rate": 0.00017615400919725475,
      "loss": 2.4788,
      "step": 146411
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.0761557817459106,
      "learning_rate": 0.00017615028347301936,
      "loss": 2.0666,
      "step": 146412
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.1112406253814697,
      "learning_rate": 0.0001761465577718099,
      "loss": 2.2283,
      "step": 146413
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.0356433391571045,
      "learning_rate": 0.00017614283209362733,
      "loss": 2.1211,
      "step": 146414
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.0533685684204102,
      "learning_rate": 0.000176139106438472,
      "loss": 2.2084,
      "step": 146415
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.11048424243927,
      "learning_rate": 0.00017613538080634472,
      "loss": 2.0582,
      "step": 146416
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.054503321647644,
      "learning_rate": 0.00017613165519724623,
      "loss": 2.2232,
      "step": 146417
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.0759179592132568,
      "learning_rate": 0.0001761279296111772,
      "loss": 2.0205,
      "step": 146418
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.1455389261245728,
      "learning_rate": 0.00017612420404813833,
      "loss": 2.1435,
      "step": 146419
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.2120018005371094,
      "learning_rate": 0.00017612047850813043,
      "loss": 2.0834,
      "step": 146420
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.0604867935180664,
      "learning_rate": 0.00017611675299115397,
      "loss": 2.0792,
      "step": 146421
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.04054856300354,
      "learning_rate": 0.0001761130274972097,
      "loss": 2.1161,
      "step": 146422
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.1079375743865967,
      "learning_rate": 0.0001761093020262984,
      "loss": 1.828,
      "step": 146423
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.0769708156585693,
      "learning_rate": 0.0001761055765784207,
      "loss": 2.0184,
      "step": 146424
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.1139020919799805,
      "learning_rate": 0.00017610185115357732,
      "loss": 2.2236,
      "step": 146425
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.0190727710723877,
      "learning_rate": 0.00017609812575176907,
      "loss": 1.8955,
      "step": 146426
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.0849846601486206,
      "learning_rate": 0.00017609440037299635,
      "loss": 2.2071,
      "step": 146427
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.0320429801940918,
      "learning_rate": 0.00017609067501726005,
      "loss": 2.0015,
      "step": 146428
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.3323298692703247,
      "learning_rate": 0.00017608694968456088,
      "loss": 2.0235,
      "step": 146429
    },
    {
      "epoch": 1.91,
      "grad_norm": 0.9983962178230286,
      "learning_rate": 0.0001760832243748994,
      "loss": 2.2271,
      "step": 146430
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.0569897890090942,
      "learning_rate": 0.00017607949908827648,
      "loss": 2.2797,
      "step": 146431
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.1855528354644775,
      "learning_rate": 0.00017607577382469276,
      "loss": 2.1487,
      "step": 146432
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.0783286094665527,
      "learning_rate": 0.00017607204858414882,
      "loss": 1.997,
      "step": 146433
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.2481155395507812,
      "learning_rate": 0.0001760683233666454,
      "loss": 2.1817,
      "step": 146434
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.0512762069702148,
      "learning_rate": 0.00017606459817218323,
      "loss": 1.9966,
      "step": 146435
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.1402465105056763,
      "learning_rate": 0.00017606087300076298,
      "loss": 1.9717,
      "step": 146436
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.1193132400512695,
      "learning_rate": 0.00017605714785238536,
      "loss": 1.9834,
      "step": 146437
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.3189772367477417,
      "learning_rate": 0.00017605342272705116,
      "loss": 2.0263,
      "step": 146438
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.1026465892791748,
      "learning_rate": 0.00017604969762476086,
      "loss": 2.0843,
      "step": 146439
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.0111619234085083,
      "learning_rate": 0.0001760459725455153,
      "loss": 2.1575,
      "step": 146440
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.0175272226333618,
      "learning_rate": 0.0001760422474893151,
      "loss": 2.2761,
      "step": 146441
    },
    {
      "epoch": 1.91,
      "grad_norm": 2.023407459259033,
      "learning_rate": 0.00017603852245616096,
      "loss": 2.1902,
      "step": 146442
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.2852917909622192,
      "learning_rate": 0.00017603479744605368,
      "loss": 2.0425,
      "step": 146443
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.030611515045166,
      "learning_rate": 0.00017603107245899393,
      "loss": 2.0206,
      "step": 146444
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.022792100906372,
      "learning_rate": 0.0001760273474949822,
      "loss": 1.9321,
      "step": 146445
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.2438044548034668,
      "learning_rate": 0.00017602362255401933,
      "loss": 2.0708,
      "step": 146446
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.150639295578003,
      "learning_rate": 0.00017601989763610607,
      "loss": 2.1989,
      "step": 146447
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.1355316638946533,
      "learning_rate": 0.000176016172741243,
      "loss": 2.2904,
      "step": 146448
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.1993838548660278,
      "learning_rate": 0.00017601244786943088,
      "loss": 2.1599,
      "step": 146449
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.1787828207015991,
      "learning_rate": 0.0001760087230206705,
      "loss": 1.7547,
      "step": 146450
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.4008055925369263,
      "learning_rate": 0.00017600499819496232,
      "loss": 2.0463,
      "step": 146451
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.146295189857483,
      "learning_rate": 0.00017600127339230718,
      "loss": 1.9525,
      "step": 146452
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.1033869981765747,
      "learning_rate": 0.0001759975486127057,
      "loss": 1.9968,
      "step": 146453
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.177820086479187,
      "learning_rate": 0.00017599382385615863,
      "loss": 2.0076,
      "step": 146454
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.1334480047225952,
      "learning_rate": 0.00017599009912266667,
      "loss": 2.115,
      "step": 146455
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.2304280996322632,
      "learning_rate": 0.00017598637441223054,
      "loss": 2.1185,
      "step": 146456
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.0198462009429932,
      "learning_rate": 0.0001759826497248509,
      "loss": 2.0833,
      "step": 146457
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.1692297458648682,
      "learning_rate": 0.00017597892506052834,
      "loss": 2.05,
      "step": 146458
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.5129992961883545,
      "learning_rate": 0.00017597520041926363,
      "loss": 2.1593,
      "step": 146459
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.1040725708007812,
      "learning_rate": 0.0001759714758010575,
      "loss": 2.1279,
      "step": 146460
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.2373571395874023,
      "learning_rate": 0.0001759677512059106,
      "loss": 2.1459,
      "step": 146461
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.1482949256896973,
      "learning_rate": 0.0001759640266338237,
      "loss": 1.9397,
      "step": 146462
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.2072051763534546,
      "learning_rate": 0.00017596030208479738,
      "loss": 2.2066,
      "step": 146463
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.002872347831726,
      "learning_rate": 0.00017595657755883242,
      "loss": 2.0522,
      "step": 146464
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.2162779569625854,
      "learning_rate": 0.00017595285305592944,
      "loss": 2.0122,
      "step": 146465
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.0948196649551392,
      "learning_rate": 0.00017594912857608913,
      "loss": 1.9354,
      "step": 146466
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.0603172779083252,
      "learning_rate": 0.00017594540411931223,
      "loss": 1.926,
      "step": 146467
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.1226245164871216,
      "learning_rate": 0.00017594167968559943,
      "loss": 2.1299,
      "step": 146468
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.1008650064468384,
      "learning_rate": 0.00017593795527495146,
      "loss": 2.3728,
      "step": 146469
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.0374289751052856,
      "learning_rate": 0.00017593423088736892,
      "loss": 2.0337,
      "step": 146470
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.1160962581634521,
      "learning_rate": 0.00017593050652285256,
      "loss": 2.1811,
      "step": 146471
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.0667362213134766,
      "learning_rate": 0.00017592678218140308,
      "loss": 2.0592,
      "step": 146472
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.1123203039169312,
      "learning_rate": 0.00017592305786302113,
      "loss": 2.1308,
      "step": 146473
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.1147161722183228,
      "learning_rate": 0.00017591933356770744,
      "loss": 2.1088,
      "step": 146474
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.1015043258666992,
      "learning_rate": 0.0001759156092954627,
      "loss": 2.1093,
      "step": 146475
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.10258948802948,
      "learning_rate": 0.00017591188504628756,
      "loss": 2.1282,
      "step": 146476
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.1786472797393799,
      "learning_rate": 0.00017590816082018276,
      "loss": 2.0104,
      "step": 146477
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.146102786064148,
      "learning_rate": 0.00017590443661714893,
      "loss": 2.0378,
      "step": 146478
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.076059341430664,
      "learning_rate": 0.0001759007124371869,
      "loss": 2.1486,
      "step": 146479
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.2016650438308716,
      "learning_rate": 0.0001758969882802972,
      "loss": 2.0306,
      "step": 146480
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.2755550146102905,
      "learning_rate": 0.00017589326414648064,
      "loss": 2.1091,
      "step": 146481
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.1798349618911743,
      "learning_rate": 0.00017588954003573782,
      "loss": 2.0966,
      "step": 146482
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.0929902791976929,
      "learning_rate": 0.00017588581594806947,
      "loss": 2.0692,
      "step": 146483
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.0963877439498901,
      "learning_rate": 0.0001758820918834763,
      "loss": 2.1863,
      "step": 146484
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.1189781427383423,
      "learning_rate": 0.000175878367841959,
      "loss": 2.107,
      "step": 146485
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.0809762477874756,
      "learning_rate": 0.00017587464382351828,
      "loss": 1.8165,
      "step": 146486
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.2627323865890503,
      "learning_rate": 0.00017587091982815484,
      "loss": 2.2759,
      "step": 146487
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.0316835641860962,
      "learning_rate": 0.00017586719585586927,
      "loss": 1.8497,
      "step": 146488
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.027628779411316,
      "learning_rate": 0.00017586347190666233,
      "loss": 2.0126,
      "step": 146489
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.1081602573394775,
      "learning_rate": 0.0001758597479805347,
      "loss": 2.3274,
      "step": 146490
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.374786376953125,
      "learning_rate": 0.0001758560240774871,
      "loss": 2.1456,
      "step": 146491
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.2266100645065308,
      "learning_rate": 0.00017585230019752019,
      "loss": 2.1669,
      "step": 146492
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.1244434118270874,
      "learning_rate": 0.00017584857634063485,
      "loss": 2.1523,
      "step": 146493
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.1413347721099854,
      "learning_rate": 0.00017584485250683144,
      "loss": 2.0916,
      "step": 146494
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.1129177808761597,
      "learning_rate": 0.00017584112869611086,
      "loss": 2.1595,
      "step": 146495
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.0732438564300537,
      "learning_rate": 0.0001758374049084737,
      "loss": 2.0471,
      "step": 146496
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.1928002834320068,
      "learning_rate": 0.00017583368114392072,
      "loss": 2.105,
      "step": 146497
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.0172241926193237,
      "learning_rate": 0.00017582995740245266,
      "loss": 2.0667,
      "step": 146498
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.0944586992263794,
      "learning_rate": 0.00017582623368407022,
      "loss": 2.0994,
      "step": 146499
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.0510202646255493,
      "learning_rate": 0.0001758225099887739,
      "loss": 2.188,
      "step": 146500
    },
    {
      "epoch": 1.91,
      "grad_norm": 0.932422399520874,
      "learning_rate": 0.00017581878631656452,
      "loss": 1.808,
      "step": 146501
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.1870307922363281,
      "learning_rate": 0.0001758150626674428,
      "loss": 1.9118,
      "step": 146502
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.128878116607666,
      "learning_rate": 0.0001758113390414094,
      "loss": 2.2312,
      "step": 146503
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.1165233850479126,
      "learning_rate": 0.00017580761543846502,
      "loss": 2.0062,
      "step": 146504
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.0090912580490112,
      "learning_rate": 0.00017580389185861047,
      "loss": 1.996,
      "step": 146505
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.0985908508300781,
      "learning_rate": 0.00017580016830184616,
      "loss": 1.9732,
      "step": 146506
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.1218080520629883,
      "learning_rate": 0.000175796444768173,
      "loss": 1.9592,
      "step": 146507
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.10440194606781,
      "learning_rate": 0.00017579272125759157,
      "loss": 1.9712,
      "step": 146508
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.1017404794692993,
      "learning_rate": 0.00017578899777010267,
      "loss": 2.1603,
      "step": 146509
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.1031439304351807,
      "learning_rate": 0.00017578527430570692,
      "loss": 2.004,
      "step": 146510
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.146053671836853,
      "learning_rate": 0.0001757815508644051,
      "loss": 2.1436,
      "step": 146511
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.073184847831726,
      "learning_rate": 0.0001757778274461978,
      "loss": 1.8113,
      "step": 146512
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.0380773544311523,
      "learning_rate": 0.00017577410405108571,
      "loss": 2.0936,
      "step": 146513
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.3934129476547241,
      "learning_rate": 0.00017577038067906954,
      "loss": 2.1586,
      "step": 146514
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.065724492073059,
      "learning_rate": 0.00017576665733015003,
      "loss": 2.2902,
      "step": 146515
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.147964596748352,
      "learning_rate": 0.00017576293400432783,
      "loss": 2.1874,
      "step": 146516
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.0109667778015137,
      "learning_rate": 0.00017575921070160376,
      "loss": 2.2778,
      "step": 146517
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.159985065460205,
      "learning_rate": 0.00017575548742197828,
      "loss": 1.9943,
      "step": 146518
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.2539182901382446,
      "learning_rate": 0.00017575176416545225,
      "loss": 1.9077,
      "step": 146519
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.0345895290374756,
      "learning_rate": 0.00017574804093202623,
      "loss": 2.0707,
      "step": 146520
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.0306198596954346,
      "learning_rate": 0.00017574431772170106,
      "loss": 2.0953,
      "step": 146521
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.2650582790374756,
      "learning_rate": 0.00017574059453447734,
      "loss": 2.0028,
      "step": 146522
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.283653974533081,
      "learning_rate": 0.00017573687137035593,
      "loss": 1.9941,
      "step": 146523
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.1485378742218018,
      "learning_rate": 0.00017573314822933724,
      "loss": 2.1674,
      "step": 146524
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.208848237991333,
      "learning_rate": 0.00017572942511142207,
      "loss": 2.0021,
      "step": 146525
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.3291031122207642,
      "learning_rate": 0.00017572570201661119,
      "loss": 1.9151,
      "step": 146526
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.2958482503890991,
      "learning_rate": 0.00017572197894490522,
      "loss": 1.91,
      "step": 146527
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.4237797260284424,
      "learning_rate": 0.0001757182558963049,
      "loss": 2.224,
      "step": 146528
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.1726763248443604,
      "learning_rate": 0.00017571453287081105,
      "loss": 2.4654,
      "step": 146529
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.2852355241775513,
      "learning_rate": 0.00017571080986842407,
      "loss": 2.0214,
      "step": 146530
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.1952139139175415,
      "learning_rate": 0.00017570708688914477,
      "loss": 2.0189,
      "step": 146531
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.0415793657302856,
      "learning_rate": 0.0001757033639329739,
      "loss": 1.9654,
      "step": 146532
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.1073102951049805,
      "learning_rate": 0.00017569964099991208,
      "loss": 2.0389,
      "step": 146533
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.065205693244934,
      "learning_rate": 0.0001756959180899601,
      "loss": 2.3407,
      "step": 146534
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.152616024017334,
      "learning_rate": 0.00017569219520311856,
      "loss": 2.1083,
      "step": 146535
    },
    {
      "epoch": 1.91,
      "grad_norm": 0.9807770848274231,
      "learning_rate": 0.00017568847233938835,
      "loss": 2.0857,
      "step": 146536
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.0782638788223267,
      "learning_rate": 0.00017568474949876983,
      "loss": 2.2774,
      "step": 146537
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.096252679824829,
      "learning_rate": 0.0001756810266812639,
      "loss": 2.1662,
      "step": 146538
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.2204256057739258,
      "learning_rate": 0.00017567730388687116,
      "loss": 2.2594,
      "step": 146539
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.093626856803894,
      "learning_rate": 0.0001756735811155924,
      "loss": 2.0218,
      "step": 146540
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.202641248703003,
      "learning_rate": 0.00017566985836742826,
      "loss": 2.1462,
      "step": 146541
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.0669666528701782,
      "learning_rate": 0.00017566613564237957,
      "loss": 2.09,
      "step": 146542
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.1701362133026123,
      "learning_rate": 0.00017566241294044674,
      "loss": 2.294,
      "step": 146543
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.0380438566207886,
      "learning_rate": 0.00017565869026163063,
      "loss": 2.1921,
      "step": 146544
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.0962703227996826,
      "learning_rate": 0.00017565496760593192,
      "loss": 2.0195,
      "step": 146545
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.1346487998962402,
      "learning_rate": 0.0001756512449733513,
      "loss": 2.1089,
      "step": 146546
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.1432856321334839,
      "learning_rate": 0.00017564752236388947,
      "loss": 2.1074,
      "step": 146547
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.2216488122940063,
      "learning_rate": 0.00017564379977754718,
      "loss": 2.0175,
      "step": 146548
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.0931044816970825,
      "learning_rate": 0.00017564007721432501,
      "loss": 2.0515,
      "step": 146549
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.057600498199463,
      "learning_rate": 0.00017563635467422365,
      "loss": 1.9854,
      "step": 146550
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.0767130851745605,
      "learning_rate": 0.00017563263215724387,
      "loss": 2.1397,
      "step": 146551
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.2784287929534912,
      "learning_rate": 0.0001756289096633863,
      "loss": 2.0352,
      "step": 146552
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.2322975397109985,
      "learning_rate": 0.0001756251871926517,
      "loss": 2.2647,
      "step": 146553
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.2317055463790894,
      "learning_rate": 0.00017562146474504076,
      "loss": 1.9185,
      "step": 146554
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.1577174663543701,
      "learning_rate": 0.0001756177423205541,
      "loss": 2.2794,
      "step": 146555
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.181214451789856,
      "learning_rate": 0.00017561401991919248,
      "loss": 2.0777,
      "step": 146556
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.2196921110153198,
      "learning_rate": 0.0001756102975409565,
      "loss": 2.0671,
      "step": 146557
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.1079626083374023,
      "learning_rate": 0.00017560657518584695,
      "loss": 2.0078,
      "step": 146558
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.1137603521347046,
      "learning_rate": 0.00017560285285386445,
      "loss": 1.988,
      "step": 146559
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.1373233795166016,
      "learning_rate": 0.0001755991305450098,
      "loss": 2.2212,
      "step": 146560
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.0434930324554443,
      "learning_rate": 0.00017559540825928357,
      "loss": 1.9153,
      "step": 146561
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.153255581855774,
      "learning_rate": 0.0001755916859966865,
      "loss": 1.9962,
      "step": 146562
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.0558408498764038,
      "learning_rate": 0.00017558796375721924,
      "loss": 2.1371,
      "step": 146563
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.1234468221664429,
      "learning_rate": 0.00017558424154088262,
      "loss": 2.1781,
      "step": 146564
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.1192655563354492,
      "learning_rate": 0.0001755805193476772,
      "loss": 1.7758,
      "step": 146565
    },
    {
      "epoch": 1.91,
      "grad_norm": 0.9913225173950195,
      "learning_rate": 0.00017557679717760373,
      "loss": 2.0861,
      "step": 146566
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.3071584701538086,
      "learning_rate": 0.00017557307503066286,
      "loss": 2.2652,
      "step": 146567
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.1022827625274658,
      "learning_rate": 0.0001755693529068553,
      "loss": 2.0165,
      "step": 146568
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.2560242414474487,
      "learning_rate": 0.00017556563080618168,
      "loss": 2.2897,
      "step": 146569
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.1542279720306396,
      "learning_rate": 0.00017556190872864284,
      "loss": 2.2609,
      "step": 146570
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.1295634508132935,
      "learning_rate": 0.0001755581866742394,
      "loss": 2.1926,
      "step": 146571
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.0373320579528809,
      "learning_rate": 0.0001755544646429721,
      "loss": 2.2636,
      "step": 146572
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.132290244102478,
      "learning_rate": 0.00017555074263484146,
      "loss": 2.1552,
      "step": 146573
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.073061227798462,
      "learning_rate": 0.0001755470206498483,
      "loss": 2.1622,
      "step": 146574
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.0706292390823364,
      "learning_rate": 0.00017554329868799332,
      "loss": 1.9378,
      "step": 146575
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.101702332496643,
      "learning_rate": 0.00017553957674927716,
      "loss": 2.3372,
      "step": 146576
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.1396666765213013,
      "learning_rate": 0.00017553585483370056,
      "loss": 2.1549,
      "step": 146577
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.270012378692627,
      "learning_rate": 0.0001755321329412643,
      "loss": 2.1008,
      "step": 146578
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.0338925123214722,
      "learning_rate": 0.00017552841107196885,
      "loss": 2.0194,
      "step": 146579
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.0356664657592773,
      "learning_rate": 0.00017552468922581502,
      "loss": 1.97,
      "step": 146580
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.058332085609436,
      "learning_rate": 0.0001755209674028035,
      "loss": 2.1008,
      "step": 146581
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.1287422180175781,
      "learning_rate": 0.000175517245602935,
      "loss": 2.2014,
      "step": 146582
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.1055235862731934,
      "learning_rate": 0.00017551352382621016,
      "loss": 2.2276,
      "step": 146583
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.0042985677719116,
      "learning_rate": 0.00017550980207262985,
      "loss": 1.8208,
      "step": 146584
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.111899495124817,
      "learning_rate": 0.0001755060803421945,
      "loss": 2.0259,
      "step": 146585
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.177051305770874,
      "learning_rate": 0.00017550235863490488,
      "loss": 2.0411,
      "step": 146586
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.1197710037231445,
      "learning_rate": 0.00017549863695076175,
      "loss": 2.0898,
      "step": 146587
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.173660397529602,
      "learning_rate": 0.00017549491528976577,
      "loss": 2.2041,
      "step": 146588
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.0538849830627441,
      "learning_rate": 0.0001754911936519177,
      "loss": 2.1433,
      "step": 146589
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.1855002641677856,
      "learning_rate": 0.0001754874720372182,
      "loss": 2.093,
      "step": 146590
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.1478698253631592,
      "learning_rate": 0.0001754837504456678,
      "loss": 2.2716,
      "step": 146591
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.221941590309143,
      "learning_rate": 0.00017548002887726738,
      "loss": 2.0635,
      "step": 146592
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.0574251413345337,
      "learning_rate": 0.00017547630733201754,
      "loss": 2.2125,
      "step": 146593
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.069210410118103,
      "learning_rate": 0.00017547258580991902,
      "loss": 2.1821,
      "step": 146594
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.1210412979125977,
      "learning_rate": 0.00017546886431097248,
      "loss": 1.9053,
      "step": 146595
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.036240816116333,
      "learning_rate": 0.00017546514283517866,
      "loss": 1.8086,
      "step": 146596
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.1627827882766724,
      "learning_rate": 0.0001754614213825383,
      "loss": 1.8778,
      "step": 146597
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.2726737260818481,
      "learning_rate": 0.00017545769995305194,
      "loss": 2.2347,
      "step": 146598
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.1329429149627686,
      "learning_rate": 0.00017545397854672033,
      "loss": 2.0426,
      "step": 146599
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.314994215965271,
      "learning_rate": 0.00017545025716354419,
      "loss": 1.9899,
      "step": 146600
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.1195744276046753,
      "learning_rate": 0.00017544653580352416,
      "loss": 2.09,
      "step": 146601
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.4608954191207886,
      "learning_rate": 0.00017544281446666096,
      "loss": 2.1304,
      "step": 146602
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.2713693380355835,
      "learning_rate": 0.00017543909315295545,
      "loss": 2.0162,
      "step": 146603
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.2914232015609741,
      "learning_rate": 0.00017543537186240802,
      "loss": 1.903,
      "step": 146604
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.0978288650512695,
      "learning_rate": 0.00017543165059501952,
      "loss": 2.0796,
      "step": 146605
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.2714029550552368,
      "learning_rate": 0.00017542792935079063,
      "loss": 2.0797,
      "step": 146606
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.12322998046875,
      "learning_rate": 0.00017542420812972204,
      "loss": 2.1154,
      "step": 146607
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.0248281955718994,
      "learning_rate": 0.00017542048693181445,
      "loss": 1.8792,
      "step": 146608
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.2496548891067505,
      "learning_rate": 0.00017541676575706864,
      "loss": 2.2356,
      "step": 146609
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.109654188156128,
      "learning_rate": 0.00017541304460548507,
      "loss": 2.2032,
      "step": 146610
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.0941176414489746,
      "learning_rate": 0.00017540932347706457,
      "loss": 2.2839,
      "step": 146611
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.1332844495773315,
      "learning_rate": 0.00017540560237180788,
      "loss": 2.1806,
      "step": 146612
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.2078770399093628,
      "learning_rate": 0.0001754018812897156,
      "loss": 2.0837,
      "step": 146613
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.0024558305740356,
      "learning_rate": 0.00017539816023078847,
      "loss": 2.0646,
      "step": 146614
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.005022406578064,
      "learning_rate": 0.0001753944391950273,
      "loss": 2.0218,
      "step": 146615
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.1451385021209717,
      "learning_rate": 0.0001753907181824325,
      "loss": 2.1671,
      "step": 146616
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.0633971691131592,
      "learning_rate": 0.000175386997193005,
      "loss": 1.9862,
      "step": 146617
    },
    {
      "epoch": 1.91,
      "grad_norm": 0.9627635478973389,
      "learning_rate": 0.00017538327622674535,
      "loss": 2.176,
      "step": 146618
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.0898007154464722,
      "learning_rate": 0.0001753795552836543,
      "loss": 1.9576,
      "step": 146619
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.2243539094924927,
      "learning_rate": 0.00017537583436373256,
      "loss": 1.9345,
      "step": 146620
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.2636326551437378,
      "learning_rate": 0.0001753721134669809,
      "loss": 2.1092,
      "step": 146621
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.045876383781433,
      "learning_rate": 0.00017536839259339984,
      "loss": 1.9987,
      "step": 146622
    },
    {
      "epoch": 1.91,
      "grad_norm": 0.9828704595565796,
      "learning_rate": 0.00017536467174299013,
      "loss": 2.2705,
      "step": 146623
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.0487278699874878,
      "learning_rate": 0.00017536095091575247,
      "loss": 1.9304,
      "step": 146624
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.2383047342300415,
      "learning_rate": 0.00017535723011168755,
      "loss": 2.1149,
      "step": 146625
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.12815260887146,
      "learning_rate": 0.00017535350933079607,
      "loss": 2.049,
      "step": 146626
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.082475185394287,
      "learning_rate": 0.0001753497885730789,
      "loss": 2.11,
      "step": 146627
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.0356630086898804,
      "learning_rate": 0.00017534606783853635,
      "loss": 1.96,
      "step": 146628
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.0415936708450317,
      "learning_rate": 0.00017534234712716942,
      "loss": 2.1521,
      "step": 146629
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.0192945003509521,
      "learning_rate": 0.0001753386264389786,
      "loss": 2.0704,
      "step": 146630
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.008704423904419,
      "learning_rate": 0.00017533490577396476,
      "loss": 2.0174,
      "step": 146631
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.0956553220748901,
      "learning_rate": 0.0001753311851321285,
      "loss": 2.1659,
      "step": 146632
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.0586882829666138,
      "learning_rate": 0.00017532746451347057,
      "loss": 2.0108,
      "step": 146633
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.3216853141784668,
      "learning_rate": 0.00017532374391799162,
      "loss": 1.9093,
      "step": 146634
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.142439603805542,
      "learning_rate": 0.00017532002334569226,
      "loss": 2.07,
      "step": 146635
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.0818350315093994,
      "learning_rate": 0.00017531630279657328,
      "loss": 2.1426,
      "step": 146636
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.0950688123703003,
      "learning_rate": 0.00017531258227063539,
      "loss": 2.0994,
      "step": 146637
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.1888264417648315,
      "learning_rate": 0.00017530886176787917,
      "loss": 2.2861,
      "step": 146638
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.0844438076019287,
      "learning_rate": 0.0001753051412883055,
      "loss": 2.2067,
      "step": 146639
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.0604534149169922,
      "learning_rate": 0.00017530142083191488,
      "loss": 2.0697,
      "step": 146640
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.2797694206237793,
      "learning_rate": 0.00017529770039870815,
      "loss": 2.1968,
      "step": 146641
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.069704294204712,
      "learning_rate": 0.00017529397998868588,
      "loss": 1.9577,
      "step": 146642
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.1029399633407593,
      "learning_rate": 0.0001752902596018488,
      "loss": 2.098,
      "step": 146643
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.2244806289672852,
      "learning_rate": 0.0001752865392381976,
      "loss": 2.1312,
      "step": 146644
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.0581198930740356,
      "learning_rate": 0.00017528281889773304,
      "loss": 2.0919,
      "step": 146645
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.0567781925201416,
      "learning_rate": 0.00017527909858045573,
      "loss": 2.1863,
      "step": 146646
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.130221962928772,
      "learning_rate": 0.0001752753782863664,
      "loss": 2.1203,
      "step": 146647
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.2927374839782715,
      "learning_rate": 0.00017527165801546572,
      "loss": 1.9918,
      "step": 146648
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.1767159700393677,
      "learning_rate": 0.00017526793776775443,
      "loss": 2.0166,
      "step": 146649
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.2554138898849487,
      "learning_rate": 0.0001752642175432331,
      "loss": 2.1828,
      "step": 146650
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.2199339866638184,
      "learning_rate": 0.0001752604973419026,
      "loss": 2.0485,
      "step": 146651
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.0753368139266968,
      "learning_rate": 0.00017525677716376348,
      "loss": 1.9146,
      "step": 146652
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.187563419342041,
      "learning_rate": 0.0001752530570088165,
      "loss": 2.1521,
      "step": 146653
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.091275930404663,
      "learning_rate": 0.00017524933687706229,
      "loss": 1.9233,
      "step": 146654
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.0649077892303467,
      "learning_rate": 0.00017524561676850165,
      "loss": 1.9962,
      "step": 146655
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.2092416286468506,
      "learning_rate": 0.00017524189668313516,
      "loss": 1.936,
      "step": 146656
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.0497783422470093,
      "learning_rate": 0.00017523817662096357,
      "loss": 1.9544,
      "step": 146657
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.0755739212036133,
      "learning_rate": 0.00017523445658198757,
      "loss": 1.9048,
      "step": 146658
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.1423536539077759,
      "learning_rate": 0.0001752307365662078,
      "loss": 1.9416,
      "step": 146659
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.1190764904022217,
      "learning_rate": 0.00017522701657362498,
      "loss": 2.1248,
      "step": 146660
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.1074731349945068,
      "learning_rate": 0.00017522329660423985,
      "loss": 2.147,
      "step": 146661
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.1859854459762573,
      "learning_rate": 0.00017521957665805305,
      "loss": 2.0508,
      "step": 146662
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.231153130531311,
      "learning_rate": 0.00017521585673506532,
      "loss": 2.1145,
      "step": 146663
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.1221323013305664,
      "learning_rate": 0.00017521213683527732,
      "loss": 2.0782,
      "step": 146664
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.0949087142944336,
      "learning_rate": 0.0001752084169586897,
      "loss": 2.2406,
      "step": 146665
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.0476956367492676,
      "learning_rate": 0.00017520469710530318,
      "loss": 2.202,
      "step": 146666
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.1754251718521118,
      "learning_rate": 0.0001752009772751185,
      "loss": 2.263,
      "step": 146667
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.1523866653442383,
      "learning_rate": 0.00017519725746813626,
      "loss": 2.0425,
      "step": 146668
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.1647790670394897,
      "learning_rate": 0.00017519353768435727,
      "loss": 2.1895,
      "step": 146669
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.0743016004562378,
      "learning_rate": 0.00017518981792378224,
      "loss": 2.0192,
      "step": 146670
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.148878574371338,
      "learning_rate": 0.00017518609818641164,
      "loss": 2.0075,
      "step": 146671
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.224715232849121,
      "learning_rate": 0.0001751823784722463,
      "loss": 2.0102,
      "step": 146672
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.1783924102783203,
      "learning_rate": 0.0001751786587812869,
      "loss": 1.9551,
      "step": 146673
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.0793370008468628,
      "learning_rate": 0.0001751749391135342,
      "loss": 2.0655,
      "step": 146674
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.126222848892212,
      "learning_rate": 0.0001751712194689888,
      "loss": 2.3269,
      "step": 146675
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.075122356414795,
      "learning_rate": 0.00017516749984765155,
      "loss": 2.2346,
      "step": 146676
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.1793932914733887,
      "learning_rate": 0.0001751637802495229,
      "loss": 2.2073,
      "step": 146677
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.18672776222229,
      "learning_rate": 0.00017516006067460368,
      "loss": 1.9062,
      "step": 146678
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.3042380809783936,
      "learning_rate": 0.00017515634112289454,
      "loss": 2.2928,
      "step": 146679
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.015769362449646,
      "learning_rate": 0.0001751526215943962,
      "loss": 1.919,
      "step": 146680
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.00532066822052,
      "learning_rate": 0.0001751489020891094,
      "loss": 2.0752,
      "step": 146681
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.1881282329559326,
      "learning_rate": 0.00017514518260703478,
      "loss": 1.981,
      "step": 146682
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.3736222982406616,
      "learning_rate": 0.00017514146314817302,
      "loss": 2.1203,
      "step": 146683
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.0351669788360596,
      "learning_rate": 0.00017513774371252468,
      "loss": 1.9606,
      "step": 146684
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.147687554359436,
      "learning_rate": 0.00017513402430009074,
      "loss": 2.0464,
      "step": 146685
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.1118240356445312,
      "learning_rate": 0.00017513030491087163,
      "loss": 2.1178,
      "step": 146686
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.023943543434143,
      "learning_rate": 0.00017512658554486824,
      "loss": 1.9866,
      "step": 146687
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.0628910064697266,
      "learning_rate": 0.00017512286620208122,
      "loss": 2.3034,
      "step": 146688
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.3062571287155151,
      "learning_rate": 0.00017511914688251104,
      "loss": 2.0725,
      "step": 146689
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.0514438152313232,
      "learning_rate": 0.00017511542758615872,
      "loss": 2.0659,
      "step": 146690
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.1697617769241333,
      "learning_rate": 0.00017511170831302478,
      "loss": 2.0144,
      "step": 146691
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.0975842475891113,
      "learning_rate": 0.00017510798906310987,
      "loss": 2.008,
      "step": 146692
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.253639817237854,
      "learning_rate": 0.00017510426983641478,
      "loss": 2.1398,
      "step": 146693
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.0134738683700562,
      "learning_rate": 0.00017510055063294023,
      "loss": 2.0638,
      "step": 146694
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.2180516719818115,
      "learning_rate": 0.0001750968314526867,
      "loss": 2.3391,
      "step": 146695
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.0601449012756348,
      "learning_rate": 0.00017509311229565514,
      "loss": 1.9986,
      "step": 146696
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.1068077087402344,
      "learning_rate": 0.00017508939316184603,
      "loss": 2.348,
      "step": 146697
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.101167917251587,
      "learning_rate": 0.00017508567405126026,
      "loss": 2.1688,
      "step": 146698
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.1112534999847412,
      "learning_rate": 0.00017508195496389838,
      "loss": 2.282,
      "step": 146699
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.3081510066986084,
      "learning_rate": 0.0001750782358997612,
      "loss": 2.1508,
      "step": 146700
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.1435706615447998,
      "learning_rate": 0.00017507451685884917,
      "loss": 2.008,
      "step": 146701
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.2705124616622925,
      "learning_rate": 0.0001750707978411633,
      "loss": 2.1454,
      "step": 146702
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.1836730241775513,
      "learning_rate": 0.00017506707884670396,
      "loss": 2.1502,
      "step": 146703
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.0211515426635742,
      "learning_rate": 0.00017506335987547216,
      "loss": 1.9303,
      "step": 146704
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.1783480644226074,
      "learning_rate": 0.00017505964092746844,
      "loss": 1.9968,
      "step": 146705
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.165828824043274,
      "learning_rate": 0.00017505592200269346,
      "loss": 2.125,
      "step": 146706
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.062683343887329,
      "learning_rate": 0.00017505220310114788,
      "loss": 1.9512,
      "step": 146707
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.197776198387146,
      "learning_rate": 0.0001750484842228325,
      "loss": 1.8631,
      "step": 146708
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.1000125408172607,
      "learning_rate": 0.0001750447653677479,
      "loss": 2.1469,
      "step": 146709
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.131677508354187,
      "learning_rate": 0.00017504104653589492,
      "loss": 1.9387,
      "step": 146710
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.116091251373291,
      "learning_rate": 0.0001750373277272741,
      "loss": 1.7557,
      "step": 146711
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.1115548610687256,
      "learning_rate": 0.0001750336089418864,
      "loss": 1.9517,
      "step": 146712
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.085437536239624,
      "learning_rate": 0.00017502989017973209,
      "loss": 2.0991,
      "step": 146713
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.1002249717712402,
      "learning_rate": 0.00017502617144081218,
      "loss": 2.0366,
      "step": 146714
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.162130355834961,
      "learning_rate": 0.00017502245272512715,
      "loss": 1.9722,
      "step": 146715
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.1160695552825928,
      "learning_rate": 0.0001750187340326779,
      "loss": 2.0505,
      "step": 146716
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.1473017930984497,
      "learning_rate": 0.00017501501536346497,
      "loss": 2.3117,
      "step": 146717
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.037674069404602,
      "learning_rate": 0.00017501129671748935,
      "loss": 2.0462,
      "step": 146718
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.1154357194900513,
      "learning_rate": 0.00017500757809475121,
      "loss": 2.0623,
      "step": 146719
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.1429017782211304,
      "learning_rate": 0.00017500385949525163,
      "loss": 2.2819,
      "step": 146720
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.153701663017273,
      "learning_rate": 0.0001750001409189911,
      "loss": 2.0939,
      "step": 146721
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.0903127193450928,
      "learning_rate": 0.0001749964223659705,
      "loss": 2.3012,
      "step": 146722
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.0691479444503784,
      "learning_rate": 0.00017499270383619032,
      "loss": 1.9957,
      "step": 146723
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.0521879196166992,
      "learning_rate": 0.00017498898532965158,
      "loss": 1.9026,
      "step": 146724
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.0690335035324097,
      "learning_rate": 0.00017498526684635454,
      "loss": 2.2429,
      "step": 146725
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.1842570304870605,
      "learning_rate": 0.0001749815483863002,
      "loss": 2.2898,
      "step": 146726
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.123284935951233,
      "learning_rate": 0.000174977829949489,
      "loss": 2.2928,
      "step": 146727
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.145151972770691,
      "learning_rate": 0.00017497411153592196,
      "loss": 2.0048,
      "step": 146728
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.183693528175354,
      "learning_rate": 0.00017497039314559945,
      "loss": 2.172,
      "step": 146729
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.1002452373504639,
      "learning_rate": 0.0001749666747785224,
      "loss": 2.1118,
      "step": 146730
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.1936073303222656,
      "learning_rate": 0.00017496295643469142,
      "loss": 2.0455,
      "step": 146731
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.1432017087936401,
      "learning_rate": 0.00017495923811410718,
      "loss": 1.9926,
      "step": 146732
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.122694730758667,
      "learning_rate": 0.00017495551981677027,
      "loss": 2.2616,
      "step": 146733
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.168497920036316,
      "learning_rate": 0.0001749518015426816,
      "loss": 2.1804,
      "step": 146734
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.1692758798599243,
      "learning_rate": 0.00017494808329184165,
      "loss": 1.8105,
      "step": 146735
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.0631839036941528,
      "learning_rate": 0.0001749443650642513,
      "loss": 2.0216,
      "step": 146736
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.0495274066925049,
      "learning_rate": 0.00017494064685991118,
      "loss": 2.0723,
      "step": 146737
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.1091184616088867,
      "learning_rate": 0.00017493692867882196,
      "loss": 2.08,
      "step": 146738
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.2150908708572388,
      "learning_rate": 0.00017493321052098416,
      "loss": 2.2347,
      "step": 146739
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.0342037677764893,
      "learning_rate": 0.00017492949238639882,
      "loss": 2.1288,
      "step": 146740
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.1163076162338257,
      "learning_rate": 0.00017492577427506627,
      "loss": 2.2,
      "step": 146741
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.0455783605575562,
      "learning_rate": 0.0001749220561869875,
      "loss": 2.1686,
      "step": 146742
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.1635328531265259,
      "learning_rate": 0.00017491833812216317,
      "loss": 2.1251,
      "step": 146743
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.3528282642364502,
      "learning_rate": 0.0001749146200805938,
      "loss": 2.005,
      "step": 146744
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.062928318977356,
      "learning_rate": 0.00017491090206228003,
      "loss": 1.8484,
      "step": 146745
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.0657627582550049,
      "learning_rate": 0.0001749071840672229,
      "loss": 1.9193,
      "step": 146746
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.0564861297607422,
      "learning_rate": 0.00017490346609542266,
      "loss": 2.063,
      "step": 146747
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.190756916999817,
      "learning_rate": 0.00017489974814688042,
      "loss": 2.3603,
      "step": 146748
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.142654538154602,
      "learning_rate": 0.00017489603022159664,
      "loss": 1.8919,
      "step": 146749
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.1768044233322144,
      "learning_rate": 0.00017489231231957207,
      "loss": 2.056,
      "step": 146750
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.2073866128921509,
      "learning_rate": 0.0001748885944408072,
      "loss": 2.0089,
      "step": 146751
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.192306637763977,
      "learning_rate": 0.00017488487658530308,
      "loss": 1.8972,
      "step": 146752
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.0634779930114746,
      "learning_rate": 0.0001748811587530601,
      "loss": 1.9593,
      "step": 146753
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.120445728302002,
      "learning_rate": 0.0001748774409440792,
      "loss": 1.9827,
      "step": 146754
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.1432220935821533,
      "learning_rate": 0.00017487372315836092,
      "loss": 2.1883,
      "step": 146755
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.094711184501648,
      "learning_rate": 0.000174870005395906,
      "loss": 2.0452,
      "step": 146756
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.3252310752868652,
      "learning_rate": 0.00017486628765671497,
      "loss": 1.9118,
      "step": 146757
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.0227231979370117,
      "learning_rate": 0.00017486256994078879,
      "loss": 2.0506,
      "step": 146758
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.090091347694397,
      "learning_rate": 0.0001748588522481279,
      "loss": 2.0632,
      "step": 146759
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.098106026649475,
      "learning_rate": 0.00017485513457873324,
      "loss": 2.196,
      "step": 146760
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.4451937675476074,
      "learning_rate": 0.00017485141693260537,
      "loss": 1.9775,
      "step": 146761
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.0426158905029297,
      "learning_rate": 0.00017484769930974493,
      "loss": 2.0151,
      "step": 146762
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.178934097290039,
      "learning_rate": 0.00017484398171015264,
      "loss": 2.0233,
      "step": 146763
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.1808944940567017,
      "learning_rate": 0.00017484026413382926,
      "loss": 1.9946,
      "step": 146764
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.2263470888137817,
      "learning_rate": 0.00017483654658077531,
      "loss": 2.2578,
      "step": 146765
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.0589052438735962,
      "learning_rate": 0.0001748328290509918,
      "loss": 1.8954,
      "step": 146766
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.1846883296966553,
      "learning_rate": 0.00017482911154447918,
      "loss": 2.085,
      "step": 146767
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.1934828758239746,
      "learning_rate": 0.0001748253940612382,
      "loss": 1.8668,
      "step": 146768
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.1138014793395996,
      "learning_rate": 0.00017482167660126948,
      "loss": 2.0235,
      "step": 146769
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.1003457307815552,
      "learning_rate": 0.0001748179591645738,
      "loss": 2.405,
      "step": 146770
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.1208419799804688,
      "learning_rate": 0.00017481424175115173,
      "loss": 2.0541,
      "step": 146771
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.4335975646972656,
      "learning_rate": 0.0001748105243610042,
      "loss": 2.2317,
      "step": 146772
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.1011452674865723,
      "learning_rate": 0.00017480680699413174,
      "loss": 2.1126,
      "step": 146773
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.2558588981628418,
      "learning_rate": 0.00017480308965053492,
      "loss": 2.1191,
      "step": 146774
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.1093217134475708,
      "learning_rate": 0.00017479937233021476,
      "loss": 2.0894,
      "step": 146775
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.0977625846862793,
      "learning_rate": 0.00017479565503317172,
      "loss": 2.032,
      "step": 146776
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.0979565382003784,
      "learning_rate": 0.00017479193775940638,
      "loss": 1.906,
      "step": 146777
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.0201185941696167,
      "learning_rate": 0.00017478822050891974,
      "loss": 2.0057,
      "step": 146778
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.2263572216033936,
      "learning_rate": 0.00017478450328171232,
      "loss": 1.7303,
      "step": 146779
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.0809333324432373,
      "learning_rate": 0.0001747807860777847,
      "loss": 2.0519,
      "step": 146780
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.345273733139038,
      "learning_rate": 0.00017477706889713782,
      "loss": 2.0249,
      "step": 146781
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.232032060623169,
      "learning_rate": 0.00017477335173977224,
      "loss": 1.8874,
      "step": 146782
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.162401795387268,
      "learning_rate": 0.00017476963460568855,
      "loss": 2.2595,
      "step": 146783
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.2976136207580566,
      "learning_rate": 0.00017476591749488767,
      "loss": 1.9969,
      "step": 146784
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.1924006938934326,
      "learning_rate": 0.00017476220040737016,
      "loss": 2.109,
      "step": 146785
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.3090983629226685,
      "learning_rate": 0.0001747584833431366,
      "loss": 2.0725,
      "step": 146786
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.1800904273986816,
      "learning_rate": 0.0001747547663021879,
      "loss": 2.0265,
      "step": 146787
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.345473051071167,
      "learning_rate": 0.0001747510492845246,
      "loss": 2.0975,
      "step": 146788
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.464402198791504,
      "learning_rate": 0.0001747473322901475,
      "loss": 1.9544,
      "step": 146789
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.11819326877594,
      "learning_rate": 0.00017474361531905725,
      "loss": 1.9258,
      "step": 146790
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.199477195739746,
      "learning_rate": 0.00017473989837125457,
      "loss": 2.018,
      "step": 146791
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.1962050199508667,
      "learning_rate": 0.00017473618144673992,
      "loss": 2.1409,
      "step": 146792
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.4782931804656982,
      "learning_rate": 0.00017473246454551434,
      "loss": 2.001,
      "step": 146793
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.159546971321106,
      "learning_rate": 0.00017472874766757824,
      "loss": 2.0938,
      "step": 146794
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.079384207725525,
      "learning_rate": 0.00017472503081293253,
      "loss": 1.7666,
      "step": 146795
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.2822754383087158,
      "learning_rate": 0.00017472131398157766,
      "loss": 2.204,
      "step": 146796
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.2241359949111938,
      "learning_rate": 0.00017471759717351465,
      "loss": 2.1341,
      "step": 146797
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.0749973058700562,
      "learning_rate": 0.00017471388038874398,
      "loss": 2.1297,
      "step": 146798
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.049139142036438,
      "learning_rate": 0.00017471016362726632,
      "loss": 2.06,
      "step": 146799
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.0581321716308594,
      "learning_rate": 0.00017470644688908233,
      "loss": 2.0698,
      "step": 146800
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.0537155866622925,
      "learning_rate": 0.0001747027301741929,
      "loss": 1.9342,
      "step": 146801
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.028741717338562,
      "learning_rate": 0.00017469901348259845,
      "loss": 2.0145,
      "step": 146802
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.2202593088150024,
      "learning_rate": 0.00017469529681429997,
      "loss": 2.2368,
      "step": 146803
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.2391936779022217,
      "learning_rate": 0.00017469158016929798,
      "loss": 1.9844,
      "step": 146804
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.1877129077911377,
      "learning_rate": 0.00017468786354759316,
      "loss": 1.9182,
      "step": 146805
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.1058825254440308,
      "learning_rate": 0.00017468414694918617,
      "loss": 1.998,
      "step": 146806
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.0743722915649414,
      "learning_rate": 0.00017468043037407785,
      "loss": 1.9399,
      "step": 146807
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.0801689624786377,
      "learning_rate": 0.00017467671382226867,
      "loss": 1.93,
      "step": 146808
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.0956952571868896,
      "learning_rate": 0.00017467299729375958,
      "loss": 2.0521,
      "step": 146809
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.1937350034713745,
      "learning_rate": 0.0001746692807885512,
      "loss": 2.1798,
      "step": 146810
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.0145152807235718,
      "learning_rate": 0.00017466556430664412,
      "loss": 2.1066,
      "step": 146811
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.107866644859314,
      "learning_rate": 0.00017466184784803896,
      "loss": 1.9498,
      "step": 146812
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.1563198566436768,
      "learning_rate": 0.00017465813141273666,
      "loss": 2.2278,
      "step": 146813
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.0329372882843018,
      "learning_rate": 0.00017465441500073763,
      "loss": 2.1539,
      "step": 146814
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.0781381130218506,
      "learning_rate": 0.00017465069861204288,
      "loss": 1.9854,
      "step": 146815
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.2473242282867432,
      "learning_rate": 0.0001746469822466529,
      "loss": 2.2472,
      "step": 146816
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.0468169450759888,
      "learning_rate": 0.0001746432659045684,
      "loss": 1.9859,
      "step": 146817
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.159083366394043,
      "learning_rate": 0.00017463954958578997,
      "loss": 2.0325,
      "step": 146818
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.275057315826416,
      "learning_rate": 0.00017463583329031855,
      "loss": 2.0745,
      "step": 146819
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.109897255897522,
      "learning_rate": 0.00017463211701815458,
      "loss": 2.1637,
      "step": 146820
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.0350680351257324,
      "learning_rate": 0.00017462840076929893,
      "loss": 1.861,
      "step": 146821
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.0925543308258057,
      "learning_rate": 0.00017462468454375232,
      "loss": 1.9876,
      "step": 146822
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.1755820512771606,
      "learning_rate": 0.00017462096834151527,
      "loss": 2.1311,
      "step": 146823
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.2004778385162354,
      "learning_rate": 0.00017461725216258846,
      "loss": 1.976,
      "step": 146824
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.1137276887893677,
      "learning_rate": 0.0001746135360069728,
      "loss": 2.3565,
      "step": 146825
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.0395938158035278,
      "learning_rate": 0.0001746098198746687,
      "loss": 2.1077,
      "step": 146826
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.001755714416504,
      "learning_rate": 0.00017460610376567715,
      "loss": 2.1169,
      "step": 146827
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.2240018844604492,
      "learning_rate": 0.00017460238767999866,
      "loss": 2.0899,
      "step": 146828
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.1867080926895142,
      "learning_rate": 0.00017459867161763397,
      "loss": 2.0572,
      "step": 146829
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.1602861881256104,
      "learning_rate": 0.00017459495557858366,
      "loss": 1.967,
      "step": 146830
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.1405946016311646,
      "learning_rate": 0.0001745912395628486,
      "loss": 2.1986,
      "step": 146831
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.1478157043457031,
      "learning_rate": 0.0001745875235704293,
      "loss": 2.0155,
      "step": 146832
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.0446605682373047,
      "learning_rate": 0.00017458380760132665,
      "loss": 1.9161,
      "step": 146833
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.2020589113235474,
      "learning_rate": 0.00017458009165554123,
      "loss": 2.121,
      "step": 146834
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.1126230955123901,
      "learning_rate": 0.00017457637573307377,
      "loss": 2.0529,
      "step": 146835
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.2129651308059692,
      "learning_rate": 0.00017457265983392478,
      "loss": 2.2088,
      "step": 146836
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.1816431283950806,
      "learning_rate": 0.00017456894395809523,
      "loss": 1.9715,
      "step": 146837
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.153336524963379,
      "learning_rate": 0.00017456522810558556,
      "loss": 2.1838,
      "step": 146838
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.2147862911224365,
      "learning_rate": 0.00017456151227639673,
      "loss": 2.319,
      "step": 146839
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.1305774450302124,
      "learning_rate": 0.00017455779647052925,
      "loss": 1.8772,
      "step": 146840
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.259565830230713,
      "learning_rate": 0.00017455408068798382,
      "loss": 2.1031,
      "step": 146841
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.1306583881378174,
      "learning_rate": 0.00017455036492876108,
      "loss": 2.1574,
      "step": 146842
    },
    {
      "epoch": 1.91,
      "grad_norm": 0.9955323934555054,
      "learning_rate": 0.0001745466491928619,
      "loss": 2.0297,
      "step": 146843
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.1376346349716187,
      "learning_rate": 0.00017454293348028677,
      "loss": 2.0072,
      "step": 146844
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.5179443359375,
      "learning_rate": 0.00017453921779103654,
      "loss": 2.1041,
      "step": 146845
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.1502306461334229,
      "learning_rate": 0.00017453550212511186,
      "loss": 2.3236,
      "step": 146846
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.1215269565582275,
      "learning_rate": 0.00017453178648251343,
      "loss": 2.3124,
      "step": 146847
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.0735293626785278,
      "learning_rate": 0.00017452807086324175,
      "loss": 2.1016,
      "step": 146848
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.137958288192749,
      "learning_rate": 0.00017452435526729782,
      "loss": 1.9043,
      "step": 146849
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.2450026273727417,
      "learning_rate": 0.00017452063969468205,
      "loss": 2.0987,
      "step": 146850
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.1558645963668823,
      "learning_rate": 0.0001745169241453954,
      "loss": 2.1896,
      "step": 146851
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.1666494607925415,
      "learning_rate": 0.0001745132086194384,
      "loss": 2.1162,
      "step": 146852
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.1130399703979492,
      "learning_rate": 0.00017450949311681175,
      "loss": 2.1435,
      "step": 146853
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.1085453033447266,
      "learning_rate": 0.00017450577763751602,
      "loss": 1.9617,
      "step": 146854
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.0329368114471436,
      "learning_rate": 0.00017450206218155222,
      "loss": 2.0342,
      "step": 146855
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.096007227897644,
      "learning_rate": 0.0001744983467489207,
      "loss": 2.3225,
      "step": 146856
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.1857569217681885,
      "learning_rate": 0.00017449463133962243,
      "loss": 1.9869,
      "step": 146857
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.1224756240844727,
      "learning_rate": 0.00017449091595365796,
      "loss": 1.9372,
      "step": 146858
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.1427664756774902,
      "learning_rate": 0.00017448720059102792,
      "loss": 2.335,
      "step": 146859
    },
    {
      "epoch": 1.91,
      "grad_norm": 0.96356201171875,
      "learning_rate": 0.00017448348525173318,
      "loss": 1.8494,
      "step": 146860
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.1029179096221924,
      "learning_rate": 0.00017447976993577427,
      "loss": 2.1355,
      "step": 146861
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.2694482803344727,
      "learning_rate": 0.0001744760546431519,
      "loss": 2.2947,
      "step": 146862
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.1865403652191162,
      "learning_rate": 0.00017447233937386688,
      "loss": 2.0491,
      "step": 146863
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.0751714706420898,
      "learning_rate": 0.00017446862412791973,
      "loss": 2.0289,
      "step": 146864
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.0639228820800781,
      "learning_rate": 0.00017446490890531133,
      "loss": 2.0957,
      "step": 146865
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.1210218667984009,
      "learning_rate": 0.00017446119370604228,
      "loss": 2.1372,
      "step": 146866
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.20892333984375,
      "learning_rate": 0.00017445747853011328,
      "loss": 2.2358,
      "step": 146867
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.169991135597229,
      "learning_rate": 0.00017445376337752483,
      "loss": 2.1301,
      "step": 146868
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.184753179550171,
      "learning_rate": 0.00017445004824827797,
      "loss": 2.2742,
      "step": 146869
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.1293500661849976,
      "learning_rate": 0.0001744463331423731,
      "loss": 1.8997,
      "step": 146870
    },
    {
      "epoch": 1.91,
      "grad_norm": 0.9854446649551392,
      "learning_rate": 0.0001744426180598111,
      "loss": 1.9653,
      "step": 146871
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.3041163682937622,
      "learning_rate": 0.00017443890300059258,
      "loss": 1.9864,
      "step": 146872
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.0671788454055786,
      "learning_rate": 0.0001744351879647182,
      "loss": 2.0015,
      "step": 146873
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.1241050958633423,
      "learning_rate": 0.0001744314729521887,
      "loss": 2.1172,
      "step": 146874
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.0920591354370117,
      "learning_rate": 0.00017442775796300484,
      "loss": 2.1987,
      "step": 146875
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.2089877128601074,
      "learning_rate": 0.0001744240429971671,
      "loss": 2.0375,
      "step": 146876
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.159655213356018,
      "learning_rate": 0.0001744203280546764,
      "loss": 2.111,
      "step": 146877
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.168170690536499,
      "learning_rate": 0.00017441661313553336,
      "loss": 2.1989,
      "step": 146878
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.111009120941162,
      "learning_rate": 0.00017441289823973849,
      "loss": 2.0207,
      "step": 146879
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.1306796073913574,
      "learning_rate": 0.00017440918336729278,
      "loss": 2.231,
      "step": 146880
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.1661207675933838,
      "learning_rate": 0.00017440546851819663,
      "loss": 1.9184,
      "step": 146881
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.1828875541687012,
      "learning_rate": 0.000174401753692451,
      "loss": 2.2163,
      "step": 146882
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.3437466621398926,
      "learning_rate": 0.00017439803889005646,
      "loss": 2.1845,
      "step": 146883
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.182465672492981,
      "learning_rate": 0.0001743943241110137,
      "loss": 2.1689,
      "step": 146884
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.111354112625122,
      "learning_rate": 0.00017439060935532327,
      "loss": 2.0741,
      "step": 146885
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.0420985221862793,
      "learning_rate": 0.00017438689462298613,
      "loss": 2.0872,
      "step": 146886
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.196337342262268,
      "learning_rate": 0.0001743831799140027,
      "loss": 2.0666,
      "step": 146887
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.2235441207885742,
      "learning_rate": 0.00017437946522837395,
      "loss": 1.9622,
      "step": 146888
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.0933561325073242,
      "learning_rate": 0.0001743757505661004,
      "loss": 1.9316,
      "step": 146889
    },
    {
      "epoch": 1.91,
      "grad_norm": 0.9811921715736389,
      "learning_rate": 0.0001743720359271828,
      "loss": 1.8028,
      "step": 146890
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.1940299272537231,
      "learning_rate": 0.0001743683213116217,
      "loss": 2.1599,
      "step": 146891
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.2682543992996216,
      "learning_rate": 0.000174364606719418,
      "loss": 1.9792,
      "step": 146892
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.0216875076293945,
      "learning_rate": 0.00017436089215057217,
      "loss": 2.0909,
      "step": 146893
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.0877901315689087,
      "learning_rate": 0.0001743571776050851,
      "loss": 2.2267,
      "step": 146894
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.1161692142486572,
      "learning_rate": 0.0001743534630829575,
      "loss": 2.1277,
      "step": 146895
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.5266571044921875,
      "learning_rate": 0.0001743497485841899,
      "loss": 2.0725,
      "step": 146896
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.1430805921554565,
      "learning_rate": 0.00017434603410878294,
      "loss": 1.895,
      "step": 146897
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.0152673721313477,
      "learning_rate": 0.00017434231965673754,
      "loss": 2.1478,
      "step": 146898
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.0195157527923584,
      "learning_rate": 0.00017433860522805415,
      "loss": 2.1357,
      "step": 146899
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.1175761222839355,
      "learning_rate": 0.0001743348908227337,
      "loss": 1.9099,
      "step": 146900
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.0973984003067017,
      "learning_rate": 0.00017433117644077682,
      "loss": 2.1765,
      "step": 146901
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.2150766849517822,
      "learning_rate": 0.0001743274620821841,
      "loss": 2.1468,
      "step": 146902
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.186267614364624,
      "learning_rate": 0.00017432374774695614,
      "loss": 2.198,
      "step": 146903
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.0818771123886108,
      "learning_rate": 0.00017432003343509394,
      "loss": 1.9957,
      "step": 146904
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.0910619497299194,
      "learning_rate": 0.0001743163191465979,
      "loss": 2.1686,
      "step": 146905
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.2183377742767334,
      "learning_rate": 0.0001743126048814689,
      "loss": 1.9882,
      "step": 146906
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.1326487064361572,
      "learning_rate": 0.00017430889063970757,
      "loss": 2.2047,
      "step": 146907
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.083040714263916,
      "learning_rate": 0.0001743051764213146,
      "loss": 2.0528,
      "step": 146908
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.074699878692627,
      "learning_rate": 0.00017430146222629055,
      "loss": 2.0978,
      "step": 146909
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.0561538934707642,
      "learning_rate": 0.00017429774805463634,
      "loss": 1.9378,
      "step": 146910
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.173817753791809,
      "learning_rate": 0.00017429403390635247,
      "loss": 1.9865,
      "step": 146911
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.198760747909546,
      "learning_rate": 0.00017429031978143976,
      "loss": 2.1116,
      "step": 146912
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.2000846862792969,
      "learning_rate": 0.00017428660567989893,
      "loss": 1.8724,
      "step": 146913
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.0733953714370728,
      "learning_rate": 0.00017428289160173057,
      "loss": 2.0983,
      "step": 146914
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.196557879447937,
      "learning_rate": 0.00017427917754693525,
      "loss": 2.0339,
      "step": 146915
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.160161018371582,
      "learning_rate": 0.00017427546351551394,
      "loss": 2.0155,
      "step": 146916
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.1378899812698364,
      "learning_rate": 0.00017427174950746706,
      "loss": 2.1731,
      "step": 146917
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.1325366497039795,
      "learning_rate": 0.0001742680355227956,
      "loss": 2.0892,
      "step": 146918
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.2843902111053467,
      "learning_rate": 0.00017426432156150004,
      "loss": 2.0704,
      "step": 146919
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.2254120111465454,
      "learning_rate": 0.00017426060762358112,
      "loss": 2.0781,
      "step": 146920
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.0558329820632935,
      "learning_rate": 0.00017425689370903938,
      "loss": 2.2706,
      "step": 146921
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.1560192108154297,
      "learning_rate": 0.00017425317981787587,
      "loss": 2.217,
      "step": 146922
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.1130152940750122,
      "learning_rate": 0.0001742494659500909,
      "loss": 1.8135,
      "step": 146923
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.0663172006607056,
      "learning_rate": 0.0001742457521056854,
      "loss": 1.9932,
      "step": 146924
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.2521774768829346,
      "learning_rate": 0.00017424203828466005,
      "loss": 2.1798,
      "step": 146925
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.1723215579986572,
      "learning_rate": 0.00017423832448701542,
      "loss": 1.947,
      "step": 146926
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.1049495935440063,
      "learning_rate": 0.00017423461071275218,
      "loss": 2.0612,
      "step": 146927
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.1052430868148804,
      "learning_rate": 0.00017423089696187122,
      "loss": 2.08,
      "step": 146928
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.1017662286758423,
      "learning_rate": 0.00017422718323437296,
      "loss": 2.1364,
      "step": 146929
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.232153296470642,
      "learning_rate": 0.0001742234695302584,
      "loss": 2.1384,
      "step": 146930
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.16170072555542,
      "learning_rate": 0.00017421975584952794,
      "loss": 1.9319,
      "step": 146931
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.0743635892868042,
      "learning_rate": 0.00017421604219218265,
      "loss": 2.0856,
      "step": 146932
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.283633828163147,
      "learning_rate": 0.00017421232855822273,
      "loss": 2.1969,
      "step": 146933
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.3396480083465576,
      "learning_rate": 0.00017420861494764925,
      "loss": 1.9037,
      "step": 146934
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.065874695777893,
      "learning_rate": 0.00017420490136046257,
      "loss": 2.1581,
      "step": 146935
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.1839579343795776,
      "learning_rate": 0.00017420118779666374,
      "loss": 2.1135,
      "step": 146936
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.2052223682403564,
      "learning_rate": 0.00017419747425625318,
      "loss": 2.0608,
      "step": 146937
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.1661708354949951,
      "learning_rate": 0.00017419376073923192,
      "loss": 1.8945,
      "step": 146938
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.2828395366668701,
      "learning_rate": 0.00017419004724560016,
      "loss": 1.9778,
      "step": 146939
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.2080366611480713,
      "learning_rate": 0.00017418633377535897,
      "loss": 2.0878,
      "step": 146940
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.1243418455123901,
      "learning_rate": 0.00017418262032850877,
      "loss": 1.9715,
      "step": 146941
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.1323832273483276,
      "learning_rate": 0.00017417890690505055,
      "loss": 2.0209,
      "step": 146942
    },
    {
      "epoch": 1.91,
      "grad_norm": 0.9830182194709778,
      "learning_rate": 0.00017417519350498473,
      "loss": 1.8805,
      "step": 146943
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.1831809282302856,
      "learning_rate": 0.0001741714801283123,
      "loss": 2.1387,
      "step": 146944
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.051869511604309,
      "learning_rate": 0.00017416776677503358,
      "loss": 2.1262,
      "step": 146945
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.2065857648849487,
      "learning_rate": 0.0001741640534451496,
      "loss": 1.9997,
      "step": 146946
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.1215616464614868,
      "learning_rate": 0.0001741603401386607,
      "loss": 2.0459,
      "step": 146947
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.1183624267578125,
      "learning_rate": 0.00017415662685556796,
      "loss": 2.1179,
      "step": 146948
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.1190388202667236,
      "learning_rate": 0.00017415291359587168,
      "loss": 2.3106,
      "step": 146949
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.1275510787963867,
      "learning_rate": 0.00017414920035957292,
      "loss": 2.0541,
      "step": 146950
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.1688259840011597,
      "learning_rate": 0.00017414548714667218,
      "loss": 2.1554,
      "step": 146951
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.0409252643585205,
      "learning_rate": 0.0001741417739571702,
      "loss": 2.0666,
      "step": 146952
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.2541818618774414,
      "learning_rate": 0.0001741380607910675,
      "loss": 1.8508,
      "step": 146953
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.079067349433899,
      "learning_rate": 0.000174134347648365,
      "loss": 2.3354,
      "step": 146954
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.0303863286972046,
      "learning_rate": 0.0001741306345290632,
      "loss": 2.0899,
      "step": 146955
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.1010817289352417,
      "learning_rate": 0.00017412692143316304,
      "loss": 2.0281,
      "step": 146956
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.1605886220932007,
      "learning_rate": 0.00017412320836066502,
      "loss": 2.3107,
      "step": 146957
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.0878897905349731,
      "learning_rate": 0.0001741194953115698,
      "loss": 2.0248,
      "step": 146958
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.096154808998108,
      "learning_rate": 0.00017411578228587824,
      "loss": 2.1642,
      "step": 146959
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.0990692377090454,
      "learning_rate": 0.00017411206928359094,
      "loss": 2.1531,
      "step": 146960
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.062485694885254,
      "learning_rate": 0.00017410835630470844,
      "loss": 1.9828,
      "step": 146961
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.0974231958389282,
      "learning_rate": 0.0001741046433492317,
      "loss": 1.8687,
      "step": 146962
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.1504472494125366,
      "learning_rate": 0.0001741009304171613,
      "loss": 2.0292,
      "step": 146963
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.2045507431030273,
      "learning_rate": 0.00017409721750849777,
      "loss": 2.2011,
      "step": 146964
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.142073392868042,
      "learning_rate": 0.00017409350462324207,
      "loss": 2.1909,
      "step": 146965
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.1017491817474365,
      "learning_rate": 0.00017408979176139478,
      "loss": 1.9677,
      "step": 146966
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.1089600324630737,
      "learning_rate": 0.0001740860789229564,
      "loss": 2.1493,
      "step": 146967
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.2018028497695923,
      "learning_rate": 0.00017408236610792798,
      "loss": 2.2428,
      "step": 146968
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.0255142450332642,
      "learning_rate": 0.00017407865331631,
      "loss": 2.3697,
      "step": 146969
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.2824721336364746,
      "learning_rate": 0.00017407494054810305,
      "loss": 2.0105,
      "step": 146970
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.372214436531067,
      "learning_rate": 0.0001740712278033081,
      "loss": 1.9002,
      "step": 146971
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.213361144065857,
      "learning_rate": 0.00017406751508192552,
      "loss": 1.9398,
      "step": 146972
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.1844407320022583,
      "learning_rate": 0.0001740638023839563,
      "loss": 2.0322,
      "step": 146973
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.0672531127929688,
      "learning_rate": 0.000174060089709401,
      "loss": 2.1476,
      "step": 146974
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.1460330486297607,
      "learning_rate": 0.0001740563770582603,
      "loss": 2.1072,
      "step": 146975
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.265573263168335,
      "learning_rate": 0.0001740526644305348,
      "loss": 2.1637,
      "step": 146976
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.0425716638565063,
      "learning_rate": 0.0001740489518262254,
      "loss": 2.0347,
      "step": 146977
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.0386021137237549,
      "learning_rate": 0.00017404523924533253,
      "loss": 2.1697,
      "step": 146978
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.0575683116912842,
      "learning_rate": 0.0001740415266878572,
      "loss": 2.1178,
      "step": 146979
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.2546710968017578,
      "learning_rate": 0.0001740378141537999,
      "loss": 1.8846,
      "step": 146980
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.136857509613037,
      "learning_rate": 0.00017403410164316133,
      "loss": 1.9596,
      "step": 146981
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.0951130390167236,
      "learning_rate": 0.00017403038915594208,
      "loss": 2.1824,
      "step": 146982
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.0884307622909546,
      "learning_rate": 0.0001740266766921431,
      "loss": 2.1179,
      "step": 146983
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.083195447921753,
      "learning_rate": 0.00017402296425176477,
      "loss": 1.9911,
      "step": 146984
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.2795233726501465,
      "learning_rate": 0.00017401925183480814,
      "loss": 2.0064,
      "step": 146985
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.351871371269226,
      "learning_rate": 0.00017401553944127368,
      "loss": 1.9159,
      "step": 146986
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.0258394479751587,
      "learning_rate": 0.0001740118270711621,
      "loss": 1.9909,
      "step": 146987
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.1139814853668213,
      "learning_rate": 0.00017400811472447397,
      "loss": 2.1716,
      "step": 146988
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.3001255989074707,
      "learning_rate": 0.00017400440240121026,
      "loss": 2.16,
      "step": 146989
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.1914163827896118,
      "learning_rate": 0.0001740006901013714,
      "loss": 2.1152,
      "step": 146990
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.2191845178604126,
      "learning_rate": 0.00017399697782495826,
      "loss": 2.0701,
      "step": 146991
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.0913002490997314,
      "learning_rate": 0.0001739932655719715,
      "loss": 2.0125,
      "step": 146992
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.1919702291488647,
      "learning_rate": 0.00017398955334241176,
      "loss": 2.1025,
      "step": 146993
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.3190220594406128,
      "learning_rate": 0.00017398584113627961,
      "loss": 2.0943,
      "step": 146994
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.2061101198196411,
      "learning_rate": 0.00017398212895357597,
      "loss": 2.0061,
      "step": 146995
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.1841588020324707,
      "learning_rate": 0.00017397841679430136,
      "loss": 2.0447,
      "step": 146996
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.0912034511566162,
      "learning_rate": 0.0001739747046584566,
      "loss": 2.1566,
      "step": 146997
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.0600600242614746,
      "learning_rate": 0.0001739709925460423,
      "loss": 2.1155,
      "step": 146998
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.2145593166351318,
      "learning_rate": 0.00017396728045705932,
      "loss": 2.1095,
      "step": 146999
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.12510347366333,
      "learning_rate": 0.00017396356839150803,
      "loss": 2.0779,
      "step": 147000
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.136631965637207,
      "learning_rate": 0.00017395985634938933,
      "loss": 2.1601,
      "step": 147001
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.094030737876892,
      "learning_rate": 0.00017395614433070384,
      "loss": 2.2855,
      "step": 147002
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.2287580966949463,
      "learning_rate": 0.00017395243233545233,
      "loss": 2.1148,
      "step": 147003
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.2022795677185059,
      "learning_rate": 0.00017394872036363538,
      "loss": 2.0577,
      "step": 147004
    },
    {
      "epoch": 1.91,
      "grad_norm": 0.9751172065734863,
      "learning_rate": 0.00017394500841525393,
      "loss": 1.7996,
      "step": 147005
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.0839487314224243,
      "learning_rate": 0.00017394129649030826,
      "loss": 1.9996,
      "step": 147006
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.4096699953079224,
      "learning_rate": 0.00017393758458879942,
      "loss": 2.0584,
      "step": 147007
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.1786425113677979,
      "learning_rate": 0.00017393387271072785,
      "loss": 2.0362,
      "step": 147008
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.0220317840576172,
      "learning_rate": 0.00017393016085609446,
      "loss": 1.9788,
      "step": 147009
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.0491610765457153,
      "learning_rate": 0.00017392644902489973,
      "loss": 2.0609,
      "step": 147010
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.2778881788253784,
      "learning_rate": 0.00017392273721714464,
      "loss": 2.0035,
      "step": 147011
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.1263384819030762,
      "learning_rate": 0.00017391902543282948,
      "loss": 1.9515,
      "step": 147012
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.1068439483642578,
      "learning_rate": 0.00017391531367195526,
      "loss": 1.9817,
      "step": 147013
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.2052924633026123,
      "learning_rate": 0.00017391160193452247,
      "loss": 2.2869,
      "step": 147014
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.2577590942382812,
      "learning_rate": 0.00017390789022053203,
      "loss": 2.1258,
      "step": 147015
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.069084882736206,
      "learning_rate": 0.00017390417852998431,
      "loss": 2.0232,
      "step": 147016
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.1765294075012207,
      "learning_rate": 0.00017390046686288047,
      "loss": 2.2593,
      "step": 147017
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.1709450483322144,
      "learning_rate": 0.00017389675521922067,
      "loss": 2.1929,
      "step": 147018
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.179168462753296,
      "learning_rate": 0.00017389304359900597,
      "loss": 2.0573,
      "step": 147019
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.1534804105758667,
      "learning_rate": 0.00017388933200223678,
      "loss": 1.9695,
      "step": 147020
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.2149899005889893,
      "learning_rate": 0.0001738856204289141,
      "loss": 1.9123,
      "step": 147021
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.0894291400909424,
      "learning_rate": 0.00017388190887903833,
      "loss": 2.2163,
      "step": 147022
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.1989713907241821,
      "learning_rate": 0.00017387819735261052,
      "loss": 2.1473,
      "step": 147023
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.4040651321411133,
      "learning_rate": 0.0001738744858496309,
      "loss": 2.1316,
      "step": 147024
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.2213722467422485,
      "learning_rate": 0.00017387077437010053,
      "loss": 2.0307,
      "step": 147025
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.2916285991668701,
      "learning_rate": 0.00017386706291401983,
      "loss": 1.9623,
      "step": 147026
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.0681742429733276,
      "learning_rate": 0.00017386335148138977,
      "loss": 1.9298,
      "step": 147027
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.0871230363845825,
      "learning_rate": 0.0001738596400722107,
      "loss": 1.9504,
      "step": 147028
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.5711380243301392,
      "learning_rate": 0.00017385592868648383,
      "loss": 2.0172,
      "step": 147029
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.4231977462768555,
      "learning_rate": 0.00017385221732420928,
      "loss": 2.0453,
      "step": 147030
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.1982784271240234,
      "learning_rate": 0.00017384850598538808,
      "loss": 2.302,
      "step": 147031
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.0453851222991943,
      "learning_rate": 0.00017384479467002064,
      "loss": 2.0885,
      "step": 147032
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.2009471654891968,
      "learning_rate": 0.00017384108337810804,
      "loss": 2.0964,
      "step": 147033
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.3672643899917603,
      "learning_rate": 0.00017383737210965064,
      "loss": 2.0335,
      "step": 147034
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.2351932525634766,
      "learning_rate": 0.00017383366086464933,
      "loss": 2.0123,
      "step": 147035
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.1161078214645386,
      "learning_rate": 0.00017382994964310473,
      "loss": 2.0571,
      "step": 147036
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.1120142936706543,
      "learning_rate": 0.00017382623844501753,
      "loss": 2.0023,
      "step": 147037
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.147603154182434,
      "learning_rate": 0.0001738225272703883,
      "loss": 1.9623,
      "step": 147038
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.0641430616378784,
      "learning_rate": 0.00017381881611921798,
      "loss": 2.0117,
      "step": 147039
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.2056339979171753,
      "learning_rate": 0.00017381510499150699,
      "loss": 2.1168,
      "step": 147040
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.0368701219558716,
      "learning_rate": 0.0001738113938872563,
      "loss": 2.1656,
      "step": 147041
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.1312134265899658,
      "learning_rate": 0.0001738076828064664,
      "loss": 2.1925,
      "step": 147042
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.1831191778182983,
      "learning_rate": 0.00017380397174913793,
      "loss": 2.0753,
      "step": 147043
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.3284687995910645,
      "learning_rate": 0.00017380026071527182,
      "loss": 2.1001,
      "step": 147044
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.196393370628357,
      "learning_rate": 0.00017379654970486863,
      "loss": 1.9841,
      "step": 147045
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.16623055934906,
      "learning_rate": 0.0001737928387179289,
      "loss": 2.0839,
      "step": 147046
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.2512896060943604,
      "learning_rate": 0.0001737891277544536,
      "loss": 2.1121,
      "step": 147047
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.1318061351776123,
      "learning_rate": 0.0001737854168144433,
      "loss": 2.1679,
      "step": 147048
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.104593276977539,
      "learning_rate": 0.0001737817058978985,
      "loss": 2.3807,
      "step": 147049
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.1865276098251343,
      "learning_rate": 0.00017377799500482022,
      "loss": 1.9915,
      "step": 147050
    },
    {
      "epoch": 1.91,
      "grad_norm": 0.9833849668502808,
      "learning_rate": 0.000173774284135209,
      "loss": 1.9511,
      "step": 147051
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.2264444828033447,
      "learning_rate": 0.00017377057328906537,
      "loss": 2.1036,
      "step": 147052
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.1603174209594727,
      "learning_rate": 0.00017376686246639035,
      "loss": 1.9937,
      "step": 147053
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.0480713844299316,
      "learning_rate": 0.0001737631516671844,
      "loss": 1.9141,
      "step": 147054
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.1758739948272705,
      "learning_rate": 0.0001737594408914482,
      "loss": 2.1953,
      "step": 147055
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.2418687343597412,
      "learning_rate": 0.00017375573013918259,
      "loss": 2.179,
      "step": 147056
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.1344963312149048,
      "learning_rate": 0.000173752019410388,
      "loss": 2.0174,
      "step": 147057
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.1231855154037476,
      "learning_rate": 0.00017374830870506546,
      "loss": 1.9204,
      "step": 147058
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.0304229259490967,
      "learning_rate": 0.00017374459802321554,
      "loss": 1.8389,
      "step": 147059
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.0019681453704834,
      "learning_rate": 0.0001737408873648388,
      "loss": 2.0904,
      "step": 147060
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.1871912479400635,
      "learning_rate": 0.00017373717672993598,
      "loss": 1.9369,
      "step": 147061
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.131898283958435,
      "learning_rate": 0.00017373346611850786,
      "loss": 1.9604,
      "step": 147062
    },
    {
      "epoch": 1.91,
      "grad_norm": 0.9907260537147522,
      "learning_rate": 0.00017372975553055494,
      "loss": 2.0018,
      "step": 147063
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.1935505867004395,
      "learning_rate": 0.0001737260449660782,
      "loss": 2.0893,
      "step": 147064
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.229058027267456,
      "learning_rate": 0.00017372233442507814,
      "loss": 2.0454,
      "step": 147065
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.267770767211914,
      "learning_rate": 0.0001737186239075555,
      "loss": 2.0505,
      "step": 147066
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.0867486000061035,
      "learning_rate": 0.00017371491341351081,
      "loss": 2.1942,
      "step": 147067
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.1597877740859985,
      "learning_rate": 0.00017371120294294506,
      "loss": 2.0546,
      "step": 147068
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.1726162433624268,
      "learning_rate": 0.00017370749249585865,
      "loss": 2.137,
      "step": 147069
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.2004005908966064,
      "learning_rate": 0.00017370378207225249,
      "loss": 2.1227,
      "step": 147070
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.1753382682800293,
      "learning_rate": 0.0001737000716721271,
      "loss": 2.1187,
      "step": 147071
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.0749623775482178,
      "learning_rate": 0.00017369636129548348,
      "loss": 2.0876,
      "step": 147072
    },
    {
      "epoch": 1.91,
      "grad_norm": 1.111946940422058,
      "learning_rate": 0.00017369265094232183,
      "loss": 2.0618,
      "step": 147073
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.1119325160980225,
      "learning_rate": 0.00017368894061264327,
      "loss": 2.1412,
      "step": 147074
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.144294023513794,
      "learning_rate": 0.00017368523030644815,
      "loss": 2.2216,
      "step": 147075
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.095604658126831,
      "learning_rate": 0.00017368152002373744,
      "loss": 2.1071,
      "step": 147076
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.1147788763046265,
      "learning_rate": 0.00017367780976451167,
      "loss": 2.19,
      "step": 147077
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.1077178716659546,
      "learning_rate": 0.00017367409952877177,
      "loss": 2.1348,
      "step": 147078
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.1840020418167114,
      "learning_rate": 0.000173670389316518,
      "loss": 2.1983,
      "step": 147079
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.18411386013031,
      "learning_rate": 0.00017366667912775138,
      "loss": 2.2778,
      "step": 147080
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.19768226146698,
      "learning_rate": 0.00017366296896247245,
      "loss": 1.9238,
      "step": 147081
    },
    {
      "epoch": 1.92,
      "grad_norm": 0.9937465786933899,
      "learning_rate": 0.00017365925882068206,
      "loss": 1.8887,
      "step": 147082
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.0549211502075195,
      "learning_rate": 0.00017365554870238067,
      "loss": 2.3911,
      "step": 147083
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.0572023391723633,
      "learning_rate": 0.00017365183860756933,
      "loss": 2.3322,
      "step": 147084
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.1191991567611694,
      "learning_rate": 0.00017364812853624828,
      "loss": 1.9748,
      "step": 147085
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.0865646600723267,
      "learning_rate": 0.0001736444184884185,
      "loss": 2.0472,
      "step": 147086
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.1099361181259155,
      "learning_rate": 0.00017364070846408056,
      "loss": 2.0081,
      "step": 147087
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.1293967962265015,
      "learning_rate": 0.00017363699846323528,
      "loss": 2.1003,
      "step": 147088
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.1501195430755615,
      "learning_rate": 0.00017363328848588315,
      "loss": 2.0902,
      "step": 147089
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.0932719707489014,
      "learning_rate": 0.0001736295785320252,
      "loss": 2.0389,
      "step": 147090
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.0513899326324463,
      "learning_rate": 0.00017362586860166168,
      "loss": 2.0342,
      "step": 147091
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.168995976448059,
      "learning_rate": 0.00017362215869479365,
      "loss": 2.3034,
      "step": 147092
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.0275360345840454,
      "learning_rate": 0.00017361844881142146,
      "loss": 1.8086,
      "step": 147093
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.180930733680725,
      "learning_rate": 0.00017361473895154613,
      "loss": 1.947,
      "step": 147094
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.07502019405365,
      "learning_rate": 0.00017361102911516807,
      "loss": 2.2188,
      "step": 147095
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.1704199314117432,
      "learning_rate": 0.00017360731930228838,
      "loss": 2.1517,
      "step": 147096
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.2141139507293701,
      "learning_rate": 0.00017360360951290723,
      "loss": 2.2421,
      "step": 147097
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.0597585439682007,
      "learning_rate": 0.00017359989974702567,
      "loss": 2.0976,
      "step": 147098
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.1251258850097656,
      "learning_rate": 0.0001735961900046441,
      "loss": 2.066,
      "step": 147099
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.1273159980773926,
      "learning_rate": 0.00017359248028576355,
      "loss": 1.9881,
      "step": 147100
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.0894596576690674,
      "learning_rate": 0.00017358877059038444,
      "loss": 1.9662,
      "step": 147101
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.1680372953414917,
      "learning_rate": 0.00017358506091850779,
      "loss": 1.9869,
      "step": 147102
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.1992871761322021,
      "learning_rate": 0.00017358135127013385,
      "loss": 2.0209,
      "step": 147103
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.251784086227417,
      "learning_rate": 0.00017357764164526362,
      "loss": 2.0487,
      "step": 147104
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.1719352006912231,
      "learning_rate": 0.00017357393204389755,
      "loss": 2.1515,
      "step": 147105
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.136789083480835,
      "learning_rate": 0.00017357022246603662,
      "loss": 2.2652,
      "step": 147106
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.2182292938232422,
      "learning_rate": 0.00017356651291168122,
      "loss": 2.0638,
      "step": 147107
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.2092288732528687,
      "learning_rate": 0.00017356280338083247,
      "loss": 2.0465,
      "step": 147108
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.1435917615890503,
      "learning_rate": 0.00017355909387349052,
      "loss": 2.0337,
      "step": 147109
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.1217707395553589,
      "learning_rate": 0.00017355538438965646,
      "loss": 2.125,
      "step": 147110
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.0804823637008667,
      "learning_rate": 0.0001735516749293307,
      "loss": 2.0316,
      "step": 147111
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.0019099712371826,
      "learning_rate": 0.0001735479654925142,
      "loss": 2.126,
      "step": 147112
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.2155197858810425,
      "learning_rate": 0.00017354425607920738,
      "loss": 2.1199,
      "step": 147113
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.1169120073318481,
      "learning_rate": 0.00017354054668941132,
      "loss": 2.3548,
      "step": 147114
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.0171186923980713,
      "learning_rate": 0.00017353683732312627,
      "loss": 2.0757,
      "step": 147115
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.1562281847000122,
      "learning_rate": 0.0001735331279803532,
      "loss": 2.0477,
      "step": 147116
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.1517267227172852,
      "learning_rate": 0.00017352941866109257,
      "loss": 2.0894,
      "step": 147117
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.1414833068847656,
      "learning_rate": 0.00017352570936534533,
      "loss": 2.2549,
      "step": 147118
    },
    {
      "epoch": 1.92,
      "grad_norm": 0.9684894680976868,
      "learning_rate": 0.00017352200009311192,
      "loss": 1.8831,
      "step": 147119
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.0766245126724243,
      "learning_rate": 0.00017351829084439328,
      "loss": 1.9666,
      "step": 147120
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.0792908668518066,
      "learning_rate": 0.00017351458161919,
      "loss": 1.9292,
      "step": 147121
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.1601802110671997,
      "learning_rate": 0.00017351087241750273,
      "loss": 2.1973,
      "step": 147122
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.1231122016906738,
      "learning_rate": 0.0001735071632393321,
      "loss": 2.041,
      "step": 147123
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.0511229038238525,
      "learning_rate": 0.00017350345408467894,
      "loss": 1.9697,
      "step": 147124
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.096064805984497,
      "learning_rate": 0.00017349974495354378,
      "loss": 2.1655,
      "step": 147125
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.0520529747009277,
      "learning_rate": 0.0001734960358459275,
      "loss": 2.0737,
      "step": 147126
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.201250433921814,
      "learning_rate": 0.00017349232676183073,
      "loss": 2.1915,
      "step": 147127
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.0607677698135376,
      "learning_rate": 0.0001734886177012541,
      "loss": 1.9759,
      "step": 147128
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.1017824411392212,
      "learning_rate": 0.0001734849086641982,
      "loss": 2.0143,
      "step": 147129
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.1830849647521973,
      "learning_rate": 0.00017348119965066402,
      "loss": 2.0329,
      "step": 147130
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.0401684045791626,
      "learning_rate": 0.0001734774906606519,
      "loss": 2.1386,
      "step": 147131
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.1208692789077759,
      "learning_rate": 0.00017347378169416282,
      "loss": 2.3155,
      "step": 147132
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.1238970756530762,
      "learning_rate": 0.00017347007275119732,
      "loss": 2.1465,
      "step": 147133
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.1450310945510864,
      "learning_rate": 0.0001734663638317561,
      "loss": 2.0439,
      "step": 147134
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.0054250955581665,
      "learning_rate": 0.0001734626549358399,
      "loss": 2.1718,
      "step": 147135
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.1327604055404663,
      "learning_rate": 0.00017345894606344943,
      "loss": 2.1351,
      "step": 147136
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.0583149194717407,
      "learning_rate": 0.0001734552372145852,
      "loss": 2.1207,
      "step": 147137
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.1126019954681396,
      "learning_rate": 0.00017345152838924816,
      "loss": 2.2008,
      "step": 147138
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.202967882156372,
      "learning_rate": 0.00017344781958743882,
      "loss": 1.9979,
      "step": 147139
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.096192479133606,
      "learning_rate": 0.00017344411080915789,
      "loss": 2.2604,
      "step": 147140
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.2439640760421753,
      "learning_rate": 0.00017344040205440615,
      "loss": 2.0955,
      "step": 147141
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.0937398672103882,
      "learning_rate": 0.0001734366933231841,
      "loss": 1.9286,
      "step": 147142
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.143824577331543,
      "learning_rate": 0.00017343298461549268,
      "loss": 2.0683,
      "step": 147143
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.1892752647399902,
      "learning_rate": 0.00017342927593133248,
      "loss": 2.1077,
      "step": 147144
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.2229186296463013,
      "learning_rate": 0.00017342556727070418,
      "loss": 2.1605,
      "step": 147145
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.2222023010253906,
      "learning_rate": 0.0001734218586336083,
      "loss": 1.8872,
      "step": 147146
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.078691005706787,
      "learning_rate": 0.0001734181500200458,
      "loss": 2.2273,
      "step": 147147
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.246765375137329,
      "learning_rate": 0.00017341444143001718,
      "loss": 1.8831,
      "step": 147148
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.120978832244873,
      "learning_rate": 0.00017341073286352327,
      "loss": 2.1563,
      "step": 147149
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.0216702222824097,
      "learning_rate": 0.00017340702432056476,
      "loss": 1.8308,
      "step": 147150
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.1225638389587402,
      "learning_rate": 0.0001734033158011422,
      "loss": 2.2337,
      "step": 147151
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.096736192703247,
      "learning_rate": 0.0001733996073052563,
      "loss": 2.1792,
      "step": 147152
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.113114356994629,
      "learning_rate": 0.00017339589883290789,
      "loss": 2.0862,
      "step": 147153
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.2656731605529785,
      "learning_rate": 0.00017339219038409745,
      "loss": 1.9141,
      "step": 147154
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.1616100072860718,
      "learning_rate": 0.00017338848195882592,
      "loss": 2.1219,
      "step": 147155
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.2861839532852173,
      "learning_rate": 0.00017338477355709374,
      "loss": 2.2948,
      "step": 147156
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.1006765365600586,
      "learning_rate": 0.00017338106517890195,
      "loss": 1.9677,
      "step": 147157
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.143423318862915,
      "learning_rate": 0.00017337735682425081,
      "loss": 2.1969,
      "step": 147158
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.1092166900634766,
      "learning_rate": 0.0001733736484931413,
      "loss": 2.0456,
      "step": 147159
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.1318880319595337,
      "learning_rate": 0.00017336994018557386,
      "loss": 2.0877,
      "step": 147160
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.1489802598953247,
      "learning_rate": 0.00017336623190154947,
      "loss": 2.1375,
      "step": 147161
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.054315447807312,
      "learning_rate": 0.00017336252364106864,
      "loss": 2.1983,
      "step": 147162
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.1565858125686646,
      "learning_rate": 0.00017335881540413227,
      "loss": 1.9864,
      "step": 147163
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.1962482929229736,
      "learning_rate": 0.0001733551071907407,
      "loss": 1.9119,
      "step": 147164
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.1205381155014038,
      "learning_rate": 0.00017335139900089488,
      "loss": 1.9198,
      "step": 147165
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.3054431676864624,
      "learning_rate": 0.00017334769083459536,
      "loss": 1.9348,
      "step": 147166
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.1076126098632812,
      "learning_rate": 0.000173343982691843,
      "loss": 2.1579,
      "step": 147167
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.224902868270874,
      "learning_rate": 0.00017334027457263824,
      "loss": 2.1387,
      "step": 147168
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.0428950786590576,
      "learning_rate": 0.00017333656647698214,
      "loss": 1.9411,
      "step": 147169
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.1095918416976929,
      "learning_rate": 0.00017333285840487493,
      "loss": 2.0255,
      "step": 147170
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.1675801277160645,
      "learning_rate": 0.00017332915035631768,
      "loss": 2.0349,
      "step": 147171
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.3558151721954346,
      "learning_rate": 0.00017332544233131082,
      "loss": 2.5287,
      "step": 147172
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.0806078910827637,
      "learning_rate": 0.00017332173432985524,
      "loss": 2.1292,
      "step": 147173
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.1811058521270752,
      "learning_rate": 0.00017331802635195143,
      "loss": 2.1812,
      "step": 147174
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.1304614543914795,
      "learning_rate": 0.00017331431839760045,
      "loss": 2.0838,
      "step": 147175
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.1094038486480713,
      "learning_rate": 0.00017331061046680247,
      "loss": 2.0269,
      "step": 147176
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.0040804147720337,
      "learning_rate": 0.00017330690255955856,
      "loss": 1.9724,
      "step": 147177
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.0162937641143799,
      "learning_rate": 0.0001733031946758692,
      "loss": 2.2871,
      "step": 147178
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.1017833948135376,
      "learning_rate": 0.00017329948681573523,
      "loss": 2.1509,
      "step": 147179
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.0919783115386963,
      "learning_rate": 0.00017329577897915715,
      "loss": 2.0965,
      "step": 147180
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.1349619626998901,
      "learning_rate": 0.00017329207116613607,
      "loss": 1.9889,
      "step": 147181
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.0368322134017944,
      "learning_rate": 0.00017328836337667216,
      "loss": 2.0774,
      "step": 147182
    },
    {
      "epoch": 1.92,
      "grad_norm": 0.9978813529014587,
      "learning_rate": 0.0001732846556107664,
      "loss": 1.9764,
      "step": 147183
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.0731544494628906,
      "learning_rate": 0.0001732809478684193,
      "loss": 2.0373,
      "step": 147184
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.1763993501663208,
      "learning_rate": 0.00017327724014963184,
      "loss": 2.0597,
      "step": 147185
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.020193338394165,
      "learning_rate": 0.00017327353245440434,
      "loss": 2.1103,
      "step": 147186
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.0560215711593628,
      "learning_rate": 0.000173269824782738,
      "loss": 1.9808,
      "step": 147187
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.1157431602478027,
      "learning_rate": 0.00017326611713463288,
      "loss": 2.1656,
      "step": 147188
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.1244044303894043,
      "learning_rate": 0.00017326240951009013,
      "loss": 2.0003,
      "step": 147189
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.0849708318710327,
      "learning_rate": 0.00017325870190911018,
      "loss": 2.0399,
      "step": 147190
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.0757098197937012,
      "learning_rate": 0.00017325499433169393,
      "loss": 2.3742,
      "step": 147191
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.1478257179260254,
      "learning_rate": 0.00017325128677784187,
      "loss": 2.096,
      "step": 147192
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.0993160009384155,
      "learning_rate": 0.000173247579247555,
      "loss": 2.0248,
      "step": 147193
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.0674188137054443,
      "learning_rate": 0.00017324387174083355,
      "loss": 1.9321,
      "step": 147194
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.0414868593215942,
      "learning_rate": 0.00017324016425767863,
      "loss": 2.1871,
      "step": 147195
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.178873896598816,
      "learning_rate": 0.00017323645679809062,
      "loss": 2.0317,
      "step": 147196
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.1755610704421997,
      "learning_rate": 0.00017323274936207044,
      "loss": 1.8695,
      "step": 147197
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.0344613790512085,
      "learning_rate": 0.0001732290419496186,
      "loss": 1.8991,
      "step": 147198
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.1361000537872314,
      "learning_rate": 0.00017322533456073596,
      "loss": 2.2191,
      "step": 147199
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.222391128540039,
      "learning_rate": 0.00017322162719542313,
      "loss": 2.258,
      "step": 147200
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.097672700881958,
      "learning_rate": 0.0001732179198536808,
      "loss": 2.1352,
      "step": 147201
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.061824083328247,
      "learning_rate": 0.00017321421253550952,
      "loss": 2.0098,
      "step": 147202
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.1501522064208984,
      "learning_rate": 0.00017321050524091026,
      "loss": 2.2449,
      "step": 147203
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.2863668203353882,
      "learning_rate": 0.00017320679796988344,
      "loss": 2.092,
      "step": 147204
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.0928324460983276,
      "learning_rate": 0.00017320309072242995,
      "loss": 2.1938,
      "step": 147205
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.0742137432098389,
      "learning_rate": 0.0001731993834985504,
      "loss": 2.1572,
      "step": 147206
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.0823553800582886,
      "learning_rate": 0.00017319567629824553,
      "loss": 2.1047,
      "step": 147207
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.0993223190307617,
      "learning_rate": 0.00017319196912151581,
      "loss": 2.4173,
      "step": 147208
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.0581670999526978,
      "learning_rate": 0.00017318826196836221,
      "loss": 2.1306,
      "step": 147209
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.1946834325790405,
      "learning_rate": 0.00017318455483878524,
      "loss": 2.0884,
      "step": 147210
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.1266003847122192,
      "learning_rate": 0.00017318084773278574,
      "loss": 2.1594,
      "step": 147211
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.1507141590118408,
      "learning_rate": 0.0001731771406503643,
      "loss": 2.0721,
      "step": 147212
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.0526797771453857,
      "learning_rate": 0.00017317343359152167,
      "loss": 1.9362,
      "step": 147213
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.1728613376617432,
      "learning_rate": 0.0001731697265562583,
      "loss": 1.8758,
      "step": 147214
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.0237151384353638,
      "learning_rate": 0.00017316601954457526,
      "loss": 2.107,
      "step": 147215
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.0477839708328247,
      "learning_rate": 0.00017316231255647292,
      "loss": 2.1033,
      "step": 147216
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.0863667726516724,
      "learning_rate": 0.0001731586055919522,
      "loss": 2.1359,
      "step": 147217
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.3393774032592773,
      "learning_rate": 0.0001731548986510137,
      "loss": 2.1607,
      "step": 147218
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.2855713367462158,
      "learning_rate": 0.000173151191733658,
      "loss": 2.2593,
      "step": 147219
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.0546631813049316,
      "learning_rate": 0.00017314748483988595,
      "loss": 1.9978,
      "step": 147220
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.0411690473556519,
      "learning_rate": 0.00017314377796969822,
      "loss": 2.0533,
      "step": 147221
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.1693800687789917,
      "learning_rate": 0.00017314007112309533,
      "loss": 1.9289,
      "step": 147222
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.1555593013763428,
      "learning_rate": 0.0001731363643000782,
      "loss": 2.0845,
      "step": 147223
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.1008267402648926,
      "learning_rate": 0.00017313265750064746,
      "loss": 2.0936,
      "step": 147224
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.1911181211471558,
      "learning_rate": 0.00017312895072480358,
      "loss": 2.1381,
      "step": 147225
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.2216806411743164,
      "learning_rate": 0.00017312524397254756,
      "loss": 2.2279,
      "step": 147226
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.0970834493637085,
      "learning_rate": 0.00017312153724387996,
      "loss": 2.0178,
      "step": 147227
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.1825809478759766,
      "learning_rate": 0.00017311783053880134,
      "loss": 1.7657,
      "step": 147228
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.073999285697937,
      "learning_rate": 0.0001731141238573126,
      "loss": 1.9633,
      "step": 147229
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.6044480800628662,
      "learning_rate": 0.0001731104171994144,
      "loss": 2.1117,
      "step": 147230
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.0870819091796875,
      "learning_rate": 0.0001731067105651072,
      "loss": 2.0299,
      "step": 147231
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.1619707345962524,
      "learning_rate": 0.00017310300395439202,
      "loss": 2.2553,
      "step": 147232
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.1235147714614868,
      "learning_rate": 0.00017309929736726925,
      "loss": 2.2252,
      "step": 147233
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.0794379711151123,
      "learning_rate": 0.00017309559080373982,
      "loss": 2.0574,
      "step": 147234
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.1086702346801758,
      "learning_rate": 0.0001730918842638043,
      "loss": 2.2547,
      "step": 147235
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.1737637519836426,
      "learning_rate": 0.0001730881777474634,
      "loss": 2.1999,
      "step": 147236
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.1433045864105225,
      "learning_rate": 0.00017308447125471773,
      "loss": 2.2739,
      "step": 147237
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.1238503456115723,
      "learning_rate": 0.00017308076478556812,
      "loss": 2.2323,
      "step": 147238
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.1413828134536743,
      "learning_rate": 0.0001730770583400151,
      "loss": 1.8881,
      "step": 147239
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.235188603401184,
      "learning_rate": 0.00017307335191805957,
      "loss": 2.0824,
      "step": 147240
    },
    {
      "epoch": 1.92,
      "grad_norm": 0.9856376647949219,
      "learning_rate": 0.00017306964551970196,
      "loss": 2.0271,
      "step": 147241
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.1431792974472046,
      "learning_rate": 0.00017306593914494334,
      "loss": 2.2163,
      "step": 147242
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.1008373498916626,
      "learning_rate": 0.00017306223279378398,
      "loss": 2.1425,
      "step": 147243
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.0930919647216797,
      "learning_rate": 0.00017305852646622478,
      "loss": 2.1417,
      "step": 147244
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.1684222221374512,
      "learning_rate": 0.0001730548201622663,
      "loss": 2.2143,
      "step": 147245
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.1432708501815796,
      "learning_rate": 0.00017305111388190948,
      "loss": 1.914,
      "step": 147246
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.0449010133743286,
      "learning_rate": 0.00017304740762515473,
      "loss": 2.0802,
      "step": 147247
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.1725281476974487,
      "learning_rate": 0.00017304370139200307,
      "loss": 2.0164,
      "step": 147248
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.1306567192077637,
      "learning_rate": 0.00017303999518245475,
      "loss": 2.1731,
      "step": 147249
    },
    {
      "epoch": 1.92,
      "grad_norm": 0.9875490069389343,
      "learning_rate": 0.00017303628899651083,
      "loss": 2.1909,
      "step": 147250
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.0139296054840088,
      "learning_rate": 0.0001730325828341717,
      "loss": 1.9285,
      "step": 147251
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.0299028158187866,
      "learning_rate": 0.00017302887669543842,
      "loss": 2.0252,
      "step": 147252
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.1252758502960205,
      "learning_rate": 0.00017302517058031126,
      "loss": 2.0929,
      "step": 147253
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.115443229675293,
      "learning_rate": 0.00017302146448879138,
      "loss": 2.3034,
      "step": 147254
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.0395050048828125,
      "learning_rate": 0.000173017758420879,
      "loss": 2.0155,
      "step": 147255
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.185286521911621,
      "learning_rate": 0.0001730140523765751,
      "loss": 1.9901,
      "step": 147256
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.041435956954956,
      "learning_rate": 0.00017301034635588015,
      "loss": 1.9344,
      "step": 147257
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.2052730321884155,
      "learning_rate": 0.00017300664035879514,
      "loss": 1.9212,
      "step": 147258
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.1146546602249146,
      "learning_rate": 0.00017300293438532045,
      "loss": 2.1042,
      "step": 147259
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.1076149940490723,
      "learning_rate": 0.00017299922843545712,
      "loss": 2.0277,
      "step": 147260
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.1110700368881226,
      "learning_rate": 0.00017299552250920545,
      "loss": 2.1095,
      "step": 147261
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.0657422542572021,
      "learning_rate": 0.00017299181660656644,
      "loss": 2.1519,
      "step": 147262
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.1345337629318237,
      "learning_rate": 0.00017298811072754045,
      "loss": 2.0049,
      "step": 147263
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.1211798191070557,
      "learning_rate": 0.00017298440487212852,
      "loss": 2.1565,
      "step": 147264
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.1409776210784912,
      "learning_rate": 0.00017298069904033106,
      "loss": 2.1123,
      "step": 147265
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.1210283041000366,
      "learning_rate": 0.00017297699323214907,
      "loss": 1.9989,
      "step": 147266
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.0838576555252075,
      "learning_rate": 0.000172973287447583,
      "loss": 2.2384,
      "step": 147267
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.239987850189209,
      "learning_rate": 0.00017296958168663357,
      "loss": 1.9604,
      "step": 147268
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.0823321342468262,
      "learning_rate": 0.00017296587594930145,
      "loss": 2.0857,
      "step": 147269
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.5023224353790283,
      "learning_rate": 0.00017296217023558743,
      "loss": 2.1009,
      "step": 147270
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.3492299318313599,
      "learning_rate": 0.00017295846454549202,
      "loss": 1.8482,
      "step": 147271
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.1878461837768555,
      "learning_rate": 0.00017295475887901617,
      "loss": 2.285,
      "step": 147272
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.1010571718215942,
      "learning_rate": 0.0001729510532361604,
      "loss": 1.9701,
      "step": 147273
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.539007544517517,
      "learning_rate": 0.0001729473476169255,
      "loss": 1.9111,
      "step": 147274
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.2562144994735718,
      "learning_rate": 0.0001729436420213119,
      "loss": 2.0796,
      "step": 147275
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.0384522676467896,
      "learning_rate": 0.0001729399364493206,
      "loss": 2.1214,
      "step": 147276
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.1116470098495483,
      "learning_rate": 0.00017293623090095206,
      "loss": 2.283,
      "step": 147277
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.3431757688522339,
      "learning_rate": 0.0001729325253762072,
      "loss": 2.1781,
      "step": 147278
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.26754891872406,
      "learning_rate": 0.00017292881987508656,
      "loss": 2.1944,
      "step": 147279
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.055351734161377,
      "learning_rate": 0.00017292511439759086,
      "loss": 1.9557,
      "step": 147280
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.141068935394287,
      "learning_rate": 0.0001729214089437207,
      "loss": 2.0843,
      "step": 147281
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.462613582611084,
      "learning_rate": 0.0001729177035134769,
      "loss": 2.136,
      "step": 147282
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.1165581941604614,
      "learning_rate": 0.00017291399810686,
      "loss": 2.4442,
      "step": 147283
    },
    {
      "epoch": 1.92,
      "grad_norm": 0.9487596750259399,
      "learning_rate": 0.00017291029272387094,
      "loss": 2.0465,
      "step": 147284
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.0312013626098633,
      "learning_rate": 0.0001729065873645102,
      "loss": 2.1735,
      "step": 147285
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.4511529207229614,
      "learning_rate": 0.00017290288202877855,
      "loss": 2.0095,
      "step": 147286
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.2791866064071655,
      "learning_rate": 0.00017289917671667654,
      "loss": 2.0995,
      "step": 147287
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.040245771408081,
      "learning_rate": 0.0001728954714282051,
      "loss": 1.9273,
      "step": 147288
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.1835616827011108,
      "learning_rate": 0.00017289176616336468,
      "loss": 2.1304,
      "step": 147289
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.1109265089035034,
      "learning_rate": 0.00017288806092215616,
      "loss": 2.0571,
      "step": 147290
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.2521979808807373,
      "learning_rate": 0.0001728843557045802,
      "loss": 2.0281,
      "step": 147291
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.0688220262527466,
      "learning_rate": 0.00017288065051063742,
      "loss": 2.0795,
      "step": 147292
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.0967422723770142,
      "learning_rate": 0.0001728769453403284,
      "loss": 2.1367,
      "step": 147293
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.2253376245498657,
      "learning_rate": 0.00017287324019365404,
      "loss": 2.0014,
      "step": 147294
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.379784107208252,
      "learning_rate": 0.00017286953507061488,
      "loss": 1.9418,
      "step": 147295
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.1784504652023315,
      "learning_rate": 0.00017286582997121176,
      "loss": 2.1067,
      "step": 147296
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.110236644744873,
      "learning_rate": 0.00017286212489544528,
      "loss": 1.9678,
      "step": 147297
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.0732760429382324,
      "learning_rate": 0.00017285841984331616,
      "loss": 2.0706,
      "step": 147298
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.155860424041748,
      "learning_rate": 0.00017285471481482488,
      "loss": 2.0859,
      "step": 147299
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.188331127166748,
      "learning_rate": 0.0001728510098099725,
      "loss": 2.1122,
      "step": 147300
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.127068042755127,
      "learning_rate": 0.00017284730482875936,
      "loss": 2.2185,
      "step": 147301
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.1360957622528076,
      "learning_rate": 0.00017284359987118644,
      "loss": 2.1299,
      "step": 147302
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.0519599914550781,
      "learning_rate": 0.00017283989493725428,
      "loss": 2.0081,
      "step": 147303
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.1629396677017212,
      "learning_rate": 0.00017283619002696348,
      "loss": 2.1222,
      "step": 147304
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.1982886791229248,
      "learning_rate": 0.00017283248514031493,
      "loss": 2.2418,
      "step": 147305
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.1146042346954346,
      "learning_rate": 0.00017282878027730924,
      "loss": 2.2016,
      "step": 147306
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.141435980796814,
      "learning_rate": 0.000172825075437947,
      "loss": 2.0634,
      "step": 147307
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.1309928894042969,
      "learning_rate": 0.00017282137062222904,
      "loss": 2.2469,
      "step": 147308
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.079072117805481,
      "learning_rate": 0.000172817665830156,
      "loss": 1.8291,
      "step": 147309
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.1112953424453735,
      "learning_rate": 0.00017281396106172842,
      "loss": 1.9597,
      "step": 147310
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.0426409244537354,
      "learning_rate": 0.00017281025631694726,
      "loss": 2.2816,
      "step": 147311
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.193278193473816,
      "learning_rate": 0.0001728065515958131,
      "loss": 2.0262,
      "step": 147312
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.0946460962295532,
      "learning_rate": 0.00017280284689832648,
      "loss": 2.0826,
      "step": 147313
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.1040849685668945,
      "learning_rate": 0.00017279914222448833,
      "loss": 2.1094,
      "step": 147314
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.038779854774475,
      "learning_rate": 0.00017279543757429917,
      "loss": 2.0456,
      "step": 147315
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.0856884717941284,
      "learning_rate": 0.00017279173294775966,
      "loss": 2.1536,
      "step": 147316
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.1448020935058594,
      "learning_rate": 0.00017278802834487072,
      "loss": 2.1603,
      "step": 147317
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.0835891962051392,
      "learning_rate": 0.0001727843237656327,
      "loss": 2.1388,
      "step": 147318
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.0407049655914307,
      "learning_rate": 0.00017278061921004664,
      "loss": 2.0441,
      "step": 147319
    },
    {
      "epoch": 1.92,
      "grad_norm": 0.9655272960662842,
      "learning_rate": 0.00017277691467811304,
      "loss": 1.8558,
      "step": 147320
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.143865704536438,
      "learning_rate": 0.00017277321016983264,
      "loss": 2.0996,
      "step": 147321
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.1109896898269653,
      "learning_rate": 0.00017276950568520595,
      "loss": 1.9442,
      "step": 147322
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.2092647552490234,
      "learning_rate": 0.00017276580122423392,
      "loss": 2.1949,
      "step": 147323
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.077406406402588,
      "learning_rate": 0.00017276209678691705,
      "loss": 2.0044,
      "step": 147324
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.1488205194473267,
      "learning_rate": 0.00017275839237325622,
      "loss": 2.0475,
      "step": 147325
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.036105990409851,
      "learning_rate": 0.00017275468798325186,
      "loss": 2.1047,
      "step": 147326
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.0621426105499268,
      "learning_rate": 0.00017275098361690507,
      "loss": 2.0487,
      "step": 147327
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.0565965175628662,
      "learning_rate": 0.00017274727927421604,
      "loss": 2.284,
      "step": 147328
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.1618086099624634,
      "learning_rate": 0.00017274357495518583,
      "loss": 2.042,
      "step": 147329
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.1479403972625732,
      "learning_rate": 0.00017273987065981482,
      "loss": 1.8426,
      "step": 147330
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.0576038360595703,
      "learning_rate": 0.00017273616638810403,
      "loss": 1.8865,
      "step": 147331
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.2428377866744995,
      "learning_rate": 0.00017273246214005386,
      "loss": 2.1617,
      "step": 147332
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.1781433820724487,
      "learning_rate": 0.00017272875791566524,
      "loss": 2.0154,
      "step": 147333
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.1334370374679565,
      "learning_rate": 0.00017272505371493873,
      "loss": 1.7675,
      "step": 147334
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.1689656972885132,
      "learning_rate": 0.00017272134953787507,
      "loss": 2.2893,
      "step": 147335
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.3385820388793945,
      "learning_rate": 0.00017271764538447475,
      "loss": 2.1156,
      "step": 147336
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.211515188217163,
      "learning_rate": 0.0001727139412547388,
      "loss": 2.096,
      "step": 147337
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.1366839408874512,
      "learning_rate": 0.00017271023714866756,
      "loss": 2.0756,
      "step": 147338
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.089120626449585,
      "learning_rate": 0.000172706533066262,
      "loss": 2.0084,
      "step": 147339
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.1731805801391602,
      "learning_rate": 0.00017270282900752273,
      "loss": 2.1813,
      "step": 147340
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.085593342781067,
      "learning_rate": 0.0001726991249724504,
      "loss": 2.2812,
      "step": 147341
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.1386306285858154,
      "learning_rate": 0.00017269542096104562,
      "loss": 2.0459,
      "step": 147342
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.1488761901855469,
      "learning_rate": 0.00017269171697330924,
      "loss": 2.1388,
      "step": 147343
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.1006672382354736,
      "learning_rate": 0.00017268801300924177,
      "loss": 1.9545,
      "step": 147344
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.0287144184112549,
      "learning_rate": 0.00017268430906884413,
      "loss": 1.9908,
      "step": 147345
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.0558205842971802,
      "learning_rate": 0.00017268060515211687,
      "loss": 2.0107,
      "step": 147346
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.1195405721664429,
      "learning_rate": 0.00017267690125906068,
      "loss": 1.8786,
      "step": 147347
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.079469919204712,
      "learning_rate": 0.00017267319738967614,
      "loss": 2.1915,
      "step": 147348
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.055155873298645,
      "learning_rate": 0.00017266949354396418,
      "loss": 1.9059,
      "step": 147349
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.0541819334030151,
      "learning_rate": 0.00017266578972192524,
      "loss": 2.0852,
      "step": 147350
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.0661650896072388,
      "learning_rate": 0.00017266208592356028,
      "loss": 1.9355,
      "step": 147351
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.1579376459121704,
      "learning_rate": 0.00017265838214886983,
      "loss": 1.7993,
      "step": 147352
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.0529097318649292,
      "learning_rate": 0.00017265467839785455,
      "loss": 1.9515,
      "step": 147353
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.066604495048523,
      "learning_rate": 0.0001726509746705151,
      "loss": 2.1305,
      "step": 147354
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.2214198112487793,
      "learning_rate": 0.00017264727096685237,
      "loss": 2.0112,
      "step": 147355
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.3189586400985718,
      "learning_rate": 0.00017264356728686673,
      "loss": 1.8594,
      "step": 147356
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.4230024814605713,
      "learning_rate": 0.00017263986363055922,
      "loss": 2.0732,
      "step": 147357
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.1168287992477417,
      "learning_rate": 0.00017263615999793034,
      "loss": 2.0706,
      "step": 147358
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.1861330270767212,
      "learning_rate": 0.00017263245638898082,
      "loss": 2.0005,
      "step": 147359
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.237170696258545,
      "learning_rate": 0.0001726287528037112,
      "loss": 2.1885,
      "step": 147360
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.1044100522994995,
      "learning_rate": 0.00017262504924212243,
      "loss": 2.2172,
      "step": 147361
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.071763515472412,
      "learning_rate": 0.00017262134570421496,
      "loss": 1.8737,
      "step": 147362
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.101137638092041,
      "learning_rate": 0.00017261764218998965,
      "loss": 2.0232,
      "step": 147363
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.071174144744873,
      "learning_rate": 0.00017261393869944717,
      "loss": 2.1008,
      "step": 147364
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.1848423480987549,
      "learning_rate": 0.00017261023523258817,
      "loss": 1.9732,
      "step": 147365
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.7011029720306396,
      "learning_rate": 0.0001726065317894132,
      "loss": 2.1225,
      "step": 147366
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.0143321752548218,
      "learning_rate": 0.0001726028283699232,
      "loss": 1.9551,
      "step": 147367
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.1037230491638184,
      "learning_rate": 0.0001725991249741186,
      "loss": 2.1695,
      "step": 147368
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.197923183441162,
      "learning_rate": 0.00017259542160200038,
      "loss": 2.1545,
      "step": 147369
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.1109404563903809,
      "learning_rate": 0.00017259171825356905,
      "loss": 2.2747,
      "step": 147370
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.2258141040802002,
      "learning_rate": 0.00017258801492882537,
      "loss": 2.2762,
      "step": 147371
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.35918390750885,
      "learning_rate": 0.0001725843116277698,
      "loss": 2.0772,
      "step": 147372
    },
    {
      "epoch": 1.92,
      "grad_norm": 0.9905099272727966,
      "learning_rate": 0.00017258060835040337,
      "loss": 2.1401,
      "step": 147373
    },
    {
      "epoch": 1.92,
      "grad_norm": 0.9573904275894165,
      "learning_rate": 0.0001725769050967265,
      "loss": 1.9998,
      "step": 147374
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.059090256690979,
      "learning_rate": 0.00017257320186674007,
      "loss": 2.1605,
      "step": 147375
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.1976581811904907,
      "learning_rate": 0.00017256949866044472,
      "loss": 2.065,
      "step": 147376
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.1247307062149048,
      "learning_rate": 0.0001725657954778411,
      "loss": 2.1816,
      "step": 147377
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.3698623180389404,
      "learning_rate": 0.00017256209231892976,
      "loss": 1.9979,
      "step": 147378
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.061324954032898,
      "learning_rate": 0.0001725583891837117,
      "loss": 2.0778,
      "step": 147379
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.0553733110427856,
      "learning_rate": 0.00017255468607218726,
      "loss": 2.3056,
      "step": 147380
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.4837262630462646,
      "learning_rate": 0.00017255098298435748,
      "loss": 2.1808,
      "step": 147381
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.0499156713485718,
      "learning_rate": 0.00017254727992022285,
      "loss": 2.0802,
      "step": 147382
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.228870153427124,
      "learning_rate": 0.00017254357687978409,
      "loss": 1.9076,
      "step": 147383
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.0832852125167847,
      "learning_rate": 0.00017253987386304175,
      "loss": 2.2056,
      "step": 147384
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.0174566507339478,
      "learning_rate": 0.00017253617086999682,
      "loss": 2.152,
      "step": 147385
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.103262186050415,
      "learning_rate": 0.00017253246790064968,
      "loss": 2.0072,
      "step": 147386
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.0112481117248535,
      "learning_rate": 0.00017252876495500125,
      "loss": 2.0937,
      "step": 147387
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.0767818689346313,
      "learning_rate": 0.00017252506203305218,
      "loss": 2.0643,
      "step": 147388
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.0158342123031616,
      "learning_rate": 0.00017252135913480308,
      "loss": 1.9352,
      "step": 147389
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.2454112768173218,
      "learning_rate": 0.00017251765626025454,
      "loss": 2.2065,
      "step": 147390
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.1510673761367798,
      "learning_rate": 0.0001725139534094075,
      "loss": 2.0252,
      "step": 147391
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.1188263893127441,
      "learning_rate": 0.00017251025058226242,
      "loss": 2.2463,
      "step": 147392
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.1071960926055908,
      "learning_rate": 0.0001725065477788202,
      "loss": 1.9292,
      "step": 147393
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.0209951400756836,
      "learning_rate": 0.00017250284499908142,
      "loss": 2.1783,
      "step": 147394
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.222567081451416,
      "learning_rate": 0.00017249914224304664,
      "loss": 2.0306,
      "step": 147395
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.0552922487258911,
      "learning_rate": 0.00017249543951071677,
      "loss": 1.9194,
      "step": 147396
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.0662949085235596,
      "learning_rate": 0.00017249173680209244,
      "loss": 2.0336,
      "step": 147397
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.062511682510376,
      "learning_rate": 0.00017248803411717417,
      "loss": 2.1611,
      "step": 147398
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.0409777164459229,
      "learning_rate": 0.00017248433145596295,
      "loss": 1.7901,
      "step": 147399
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.0155456066131592,
      "learning_rate": 0.0001724806288184591,
      "loss": 1.8509,
      "step": 147400
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.1517655849456787,
      "learning_rate": 0.00017247692620466369,
      "loss": 2.1807,
      "step": 147401
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.1906660795211792,
      "learning_rate": 0.00017247322361457719,
      "loss": 2.1569,
      "step": 147402
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.4194607734680176,
      "learning_rate": 0.0001724695210482002,
      "loss": 2.19,
      "step": 147403
    },
    {
      "epoch": 1.92,
      "grad_norm": 0.9970970749855042,
      "learning_rate": 0.0001724658185055337,
      "loss": 2.0877,
      "step": 147404
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.125678539276123,
      "learning_rate": 0.00017246211598657816,
      "loss": 1.9887,
      "step": 147405
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.1524419784545898,
      "learning_rate": 0.00017245841349133423,
      "loss": 2.1776,
      "step": 147406
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.0670992136001587,
      "learning_rate": 0.0001724547110198028,
      "loss": 1.886,
      "step": 147407
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.040339469909668,
      "learning_rate": 0.00017245100857198444,
      "loss": 2.0244,
      "step": 147408
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.1304017305374146,
      "learning_rate": 0.00017244730614787975,
      "loss": 1.8265,
      "step": 147409
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.1788008213043213,
      "learning_rate": 0.0001724436037474896,
      "loss": 1.7286,
      "step": 147410
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.2065969705581665,
      "learning_rate": 0.00017243990137081464,
      "loss": 2.0876,
      "step": 147411
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.14834463596344,
      "learning_rate": 0.00017243619901785535,
      "loss": 1.8684,
      "step": 147412
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.0363132953643799,
      "learning_rate": 0.0001724324966886127,
      "loss": 2.0954,
      "step": 147413
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.1036531925201416,
      "learning_rate": 0.00017242879438308727,
      "loss": 1.9782,
      "step": 147414
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.08348548412323,
      "learning_rate": 0.00017242509210127962,
      "loss": 2.0998,
      "step": 147415
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.1328400373458862,
      "learning_rate": 0.00017242138984319066,
      "loss": 2.2224,
      "step": 147416
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.1448307037353516,
      "learning_rate": 0.00017241768760882083,
      "loss": 2.2226,
      "step": 147417
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.0711675882339478,
      "learning_rate": 0.00017241398539817113,
      "loss": 2.0874,
      "step": 147418
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.0340206623077393,
      "learning_rate": 0.00017241028321124204,
      "loss": 1.9935,
      "step": 147419
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.1323968172073364,
      "learning_rate": 0.00017240658104803432,
      "loss": 1.9625,
      "step": 147420
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.1447235345840454,
      "learning_rate": 0.00017240287890854847,
      "loss": 1.9981,
      "step": 147421
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.085952639579773,
      "learning_rate": 0.00017239917679278546,
      "loss": 2.0116,
      "step": 147422
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.1382180452346802,
      "learning_rate": 0.00017239547470074572,
      "loss": 2.3193,
      "step": 147423
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.0988459587097168,
      "learning_rate": 0.0001723917726324302,
      "loss": 2.1779,
      "step": 147424
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.0848702192306519,
      "learning_rate": 0.00017238807058783946,
      "loss": 1.858,
      "step": 147425
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.0447770357131958,
      "learning_rate": 0.00017238436856697423,
      "loss": 2.0212,
      "step": 147426
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.1122798919677734,
      "learning_rate": 0.000172380666569835,
      "loss": 2.0854,
      "step": 147427
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.135898470878601,
      "learning_rate": 0.00017237696459642277,
      "loss": 2.115,
      "step": 147428
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.1563068628311157,
      "learning_rate": 0.0001723732626467379,
      "loss": 2.1228,
      "step": 147429
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.1667611598968506,
      "learning_rate": 0.00017236956072078137,
      "loss": 2.0641,
      "step": 147430
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.106176733970642,
      "learning_rate": 0.00017236585881855377,
      "loss": 2.1496,
      "step": 147431
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.0720853805541992,
      "learning_rate": 0.00017236215694005575,
      "loss": 2.1347,
      "step": 147432
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.0824551582336426,
      "learning_rate": 0.0001723584550852879,
      "loss": 2.0213,
      "step": 147433
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.1558430194854736,
      "learning_rate": 0.00017235475325425113,
      "loss": 2.0216,
      "step": 147434
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.13255774974823,
      "learning_rate": 0.00017235105144694594,
      "loss": 2.1313,
      "step": 147435
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.1364281177520752,
      "learning_rate": 0.0001723473496633732,
      "loss": 2.0377,
      "step": 147436
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.2112053632736206,
      "learning_rate": 0.0001723436479035335,
      "loss": 2.1797,
      "step": 147437
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.2270773649215698,
      "learning_rate": 0.00017233994616742748,
      "loss": 2.1379,
      "step": 147438
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.1058473587036133,
      "learning_rate": 0.00017233624445505582,
      "loss": 2.1137,
      "step": 147439
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.2181007862091064,
      "learning_rate": 0.00017233254276641935,
      "loss": 1.9625,
      "step": 147440
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.2504562139511108,
      "learning_rate": 0.00017232884110151853,
      "loss": 2.0666,
      "step": 147441
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.3251594305038452,
      "learning_rate": 0.00017232513946035434,
      "loss": 2.103,
      "step": 147442
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.0860048532485962,
      "learning_rate": 0.00017232143784292733,
      "loss": 2.3326,
      "step": 147443
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.1216498613357544,
      "learning_rate": 0.00017231773624923815,
      "loss": 2.0634,
      "step": 147444
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.063171625137329,
      "learning_rate": 0.0001723140346792874,
      "loss": 2.0909,
      "step": 147445
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.4890336990356445,
      "learning_rate": 0.00017231033313307598,
      "loss": 1.9449,
      "step": 147446
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.0243041515350342,
      "learning_rate": 0.0001723066316106044,
      "loss": 1.9279,
      "step": 147447
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.149780035018921,
      "learning_rate": 0.00017230293011187353,
      "loss": 2.0156,
      "step": 147448
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.1570485830307007,
      "learning_rate": 0.00017229922863688395,
      "loss": 2.0,
      "step": 147449
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.081352710723877,
      "learning_rate": 0.00017229552718563633,
      "loss": 2.1125,
      "step": 147450
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.0328490734100342,
      "learning_rate": 0.0001722918257581313,
      "loss": 2.0174,
      "step": 147451
    },
    {
      "epoch": 1.92,
      "grad_norm": 0.9787558913230896,
      "learning_rate": 0.00017228812435436973,
      "loss": 2.1714,
      "step": 147452
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.136074185371399,
      "learning_rate": 0.0001722844229743521,
      "loss": 2.1148,
      "step": 147453
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.122347354888916,
      "learning_rate": 0.0001722807216180793,
      "loss": 2.0045,
      "step": 147454
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.0814878940582275,
      "learning_rate": 0.00017227702028555194,
      "loss": 1.9841,
      "step": 147455
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.112558126449585,
      "learning_rate": 0.00017227331897677075,
      "loss": 2.1786,
      "step": 147456
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.1606577634811401,
      "learning_rate": 0.00017226961769173615,
      "loss": 1.999,
      "step": 147457
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.117684006690979,
      "learning_rate": 0.00017226591643044922,
      "loss": 2.2486,
      "step": 147458
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.1065802574157715,
      "learning_rate": 0.0001722622151929103,
      "loss": 2.245,
      "step": 147459
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.0427751541137695,
      "learning_rate": 0.00017225851397912042,
      "loss": 1.9528,
      "step": 147460
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.1207077503204346,
      "learning_rate": 0.00017225481278908007,
      "loss": 2.2169,
      "step": 147461
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.2162730693817139,
      "learning_rate": 0.00017225111162278995,
      "loss": 2.2001,
      "step": 147462
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.1537654399871826,
      "learning_rate": 0.00017224741048025064,
      "loss": 2.1734,
      "step": 147463
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.0545635223388672,
      "learning_rate": 0.00017224370936146305,
      "loss": 2.134,
      "step": 147464
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.0169380903244019,
      "learning_rate": 0.0001722400082664277,
      "loss": 2.1581,
      "step": 147465
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.0490695238113403,
      "learning_rate": 0.00017223630719514542,
      "loss": 2.1413,
      "step": 147466
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.1833323240280151,
      "learning_rate": 0.00017223260614761672,
      "loss": 2.1387,
      "step": 147467
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.0778318643569946,
      "learning_rate": 0.00017222890512384268,
      "loss": 2.244,
      "step": 147468
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.135406494140625,
      "learning_rate": 0.00017222520412382334,
      "loss": 2.2323,
      "step": 147469
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.044289469718933,
      "learning_rate": 0.00017222150314755996,
      "loss": 2.4006,
      "step": 147470
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.234334111213684,
      "learning_rate": 0.00017221780219505288,
      "loss": 2.102,
      "step": 147471
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.0918989181518555,
      "learning_rate": 0.00017221410126630303,
      "loss": 2.0329,
      "step": 147472
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.083030343055725,
      "learning_rate": 0.00017221040036131084,
      "loss": 2.0293,
      "step": 147473
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.0458396673202515,
      "learning_rate": 0.00017220669948007743,
      "loss": 2.0722,
      "step": 147474
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.2431164979934692,
      "learning_rate": 0.00017220299862260297,
      "loss": 1.9589,
      "step": 147475
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.0407826900482178,
      "learning_rate": 0.0001721992977888885,
      "loss": 1.886,
      "step": 147476
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.1074422597885132,
      "learning_rate": 0.00017219559697893445,
      "loss": 2.0146,
      "step": 147477
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.1958770751953125,
      "learning_rate": 0.0001721918961927418,
      "loss": 2.0023,
      "step": 147478
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.2422430515289307,
      "learning_rate": 0.00017218819543031092,
      "loss": 2.2122,
      "step": 147479
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.0964443683624268,
      "learning_rate": 0.00017218449469164285,
      "loss": 2.1054,
      "step": 147480
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.4760493040084839,
      "learning_rate": 0.00017218079397673806,
      "loss": 1.9903,
      "step": 147481
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.1678048372268677,
      "learning_rate": 0.0001721770932855973,
      "loss": 2.1045,
      "step": 147482
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.0582882165908813,
      "learning_rate": 0.0001721733926182211,
      "loss": 2.0644,
      "step": 147483
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.0997810363769531,
      "learning_rate": 0.00017216969197461035,
      "loss": 1.9256,
      "step": 147484
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.1329163312911987,
      "learning_rate": 0.0001721659913547656,
      "loss": 2.1926,
      "step": 147485
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.3954691886901855,
      "learning_rate": 0.00017216229075868773,
      "loss": 1.816,
      "step": 147486
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.1766084432601929,
      "learning_rate": 0.0001721585901863773,
      "loss": 2.2471,
      "step": 147487
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.0620328187942505,
      "learning_rate": 0.00017215488963783483,
      "loss": 2.2525,
      "step": 147488
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.195010781288147,
      "learning_rate": 0.00017215118911306136,
      "loss": 1.9296,
      "step": 147489
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.17131507396698,
      "learning_rate": 0.00017214748861205738,
      "loss": 1.9151,
      "step": 147490
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.0682157278060913,
      "learning_rate": 0.00017214378813482344,
      "loss": 1.9646,
      "step": 147491
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.2243839502334595,
      "learning_rate": 0.00017214008768136056,
      "loss": 2.1462,
      "step": 147492
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.0918426513671875,
      "learning_rate": 0.00017213638725166921,
      "loss": 2.0771,
      "step": 147493
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.2087308168411255,
      "learning_rate": 0.00017213268684575003,
      "loss": 2.0981,
      "step": 147494
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.2302954196929932,
      "learning_rate": 0.00017212898646360388,
      "loss": 2.3332,
      "step": 147495
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.15915048122406,
      "learning_rate": 0.00017212528610523138,
      "loss": 2.1521,
      "step": 147496
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.0968449115753174,
      "learning_rate": 0.00017212158577063308,
      "loss": 1.878,
      "step": 147497
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.2065953016281128,
      "learning_rate": 0.00017211788545980994,
      "loss": 2.2218,
      "step": 147498
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.0818051099777222,
      "learning_rate": 0.0001721141851727625,
      "loss": 1.9561,
      "step": 147499
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.1891759634017944,
      "learning_rate": 0.00017211048490949128,
      "loss": 2.2206,
      "step": 147500
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.0716761350631714,
      "learning_rate": 0.0001721067846699973,
      "loss": 2.1956,
      "step": 147501
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.0807433128356934,
      "learning_rate": 0.00017210308445428095,
      "loss": 2.1345,
      "step": 147502
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.1594244241714478,
      "learning_rate": 0.00017209938426234316,
      "loss": 1.7966,
      "step": 147503
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.046844720840454,
      "learning_rate": 0.0001720956840941845,
      "loss": 2.1086,
      "step": 147504
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.1730698347091675,
      "learning_rate": 0.00017209198394980567,
      "loss": 2.0605,
      "step": 147505
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.1004966497421265,
      "learning_rate": 0.00017208828382920724,
      "loss": 2.0427,
      "step": 147506
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.1160818338394165,
      "learning_rate": 0.00017208458373239014,
      "loss": 2.0834,
      "step": 147507
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.0907094478607178,
      "learning_rate": 0.0001720808836593548,
      "loss": 2.0358,
      "step": 147508
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.091140866279602,
      "learning_rate": 0.0001720771836101022,
      "loss": 1.9712,
      "step": 147509
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.13126802444458,
      "learning_rate": 0.0001720734835846328,
      "loss": 2.1702,
      "step": 147510
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.1016594171524048,
      "learning_rate": 0.0001720697835829474,
      "loss": 1.9997,
      "step": 147511
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.0663206577301025,
      "learning_rate": 0.0001720660836050465,
      "loss": 2.1857,
      "step": 147512
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.0542181730270386,
      "learning_rate": 0.00017206238365093105,
      "loss": 2.3177,
      "step": 147513
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.0748366117477417,
      "learning_rate": 0.00017205868372060154,
      "loss": 1.9513,
      "step": 147514
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.072461724281311,
      "learning_rate": 0.0001720549838140588,
      "loss": 1.9276,
      "step": 147515
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.1920865774154663,
      "learning_rate": 0.0001720512839313035,
      "loss": 1.9361,
      "step": 147516
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.1543452739715576,
      "learning_rate": 0.00017204758407233622,
      "loss": 2.1022,
      "step": 147517
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.1253914833068848,
      "learning_rate": 0.00017204388423715768,
      "loss": 2.1034,
      "step": 147518
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.182155728340149,
      "learning_rate": 0.00017204018442576862,
      "loss": 2.223,
      "step": 147519
    },
    {
      "epoch": 1.92,
      "grad_norm": 0.9603149890899658,
      "learning_rate": 0.00017203648463816966,
      "loss": 1.8952,
      "step": 147520
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.2531176805496216,
      "learning_rate": 0.0001720327848743616,
      "loss": 2.0515,
      "step": 147521
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.1259022951126099,
      "learning_rate": 0.00017202908513434508,
      "loss": 2.1239,
      "step": 147522
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.4062094688415527,
      "learning_rate": 0.00017202538541812078,
      "loss": 2.0899,
      "step": 147523
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.1357721090316772,
      "learning_rate": 0.00017202168572568927,
      "loss": 2.1528,
      "step": 147524
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.1674009561538696,
      "learning_rate": 0.00017201798605705142,
      "loss": 1.9608,
      "step": 147525
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.172736644744873,
      "learning_rate": 0.00017201428641220774,
      "loss": 2.2465,
      "step": 147526
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.2858707904815674,
      "learning_rate": 0.00017201058679115915,
      "loss": 2.1999,
      "step": 147527
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.0613951683044434,
      "learning_rate": 0.0001720068871939062,
      "loss": 2.0561,
      "step": 147528
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.232228398323059,
      "learning_rate": 0.00017200318762044959,
      "loss": 2.0677,
      "step": 147529
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.1294927597045898,
      "learning_rate": 0.00017199948807078988,
      "loss": 2.257,
      "step": 147530
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.3286893367767334,
      "learning_rate": 0.000171995788544928,
      "loss": 2.0432,
      "step": 147531
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.1500211954116821,
      "learning_rate": 0.00017199208904286438,
      "loss": 2.0785,
      "step": 147532
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.1789017915725708,
      "learning_rate": 0.00017198838956460001,
      "loss": 2.1246,
      "step": 147533
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.042785406112671,
      "learning_rate": 0.0001719846901101353,
      "loss": 1.8363,
      "step": 147534
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.0811057090759277,
      "learning_rate": 0.00017198099067947125,
      "loss": 2.1991,
      "step": 147535
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.2426187992095947,
      "learning_rate": 0.0001719772912726081,
      "loss": 1.844,
      "step": 147536
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.0637778043746948,
      "learning_rate": 0.00017197359188954697,
      "loss": 1.8728,
      "step": 147537
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.1177133321762085,
      "learning_rate": 0.0001719698925302882,
      "loss": 2.1155,
      "step": 147538
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.184910774230957,
      "learning_rate": 0.0001719661931948328,
      "loss": 2.2438,
      "step": 147539
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.1436556577682495,
      "learning_rate": 0.00017196249388318113,
      "loss": 1.9448,
      "step": 147540
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.0453418493270874,
      "learning_rate": 0.00017195879459533435,
      "loss": 2.0072,
      "step": 147541
    },
    {
      "epoch": 1.92,
      "grad_norm": 0.9958035349845886,
      "learning_rate": 0.00017195509533129253,
      "loss": 2.0242,
      "step": 147542
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.210808277130127,
      "learning_rate": 0.00017195139609105687,
      "loss": 2.1863,
      "step": 147543
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.1033803224563599,
      "learning_rate": 0.00017194769687462768,
      "loss": 1.9547,
      "step": 147544
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.0639612674713135,
      "learning_rate": 0.000171943997682006,
      "loss": 2.0795,
      "step": 147545
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.1340328454971313,
      "learning_rate": 0.0001719402985131922,
      "loss": 2.0075,
      "step": 147546
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.2134277820587158,
      "learning_rate": 0.00017193659936818735,
      "loss": 2.1577,
      "step": 147547
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.0412936210632324,
      "learning_rate": 0.0001719329002469917,
      "loss": 2.017,
      "step": 147548
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.3752859830856323,
      "learning_rate": 0.00017192920114960628,
      "loss": 2.0264,
      "step": 147549
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.0353721380233765,
      "learning_rate": 0.00017192550207603143,
      "loss": 2.0882,
      "step": 147550
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.1193745136260986,
      "learning_rate": 0.0001719218030262682,
      "loss": 1.9393,
      "step": 147551
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.0326019525527954,
      "learning_rate": 0.00017191810400031706,
      "loss": 2.0922,
      "step": 147552
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.3651580810546875,
      "learning_rate": 0.00017191440499817896,
      "loss": 1.9439,
      "step": 147553
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.132404088973999,
      "learning_rate": 0.00017191070601985415,
      "loss": 2.0653,
      "step": 147554
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.136754035949707,
      "learning_rate": 0.00017190700706534365,
      "loss": 2.0395,
      "step": 147555
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.3397668600082397,
      "learning_rate": 0.00017190330813464798,
      "loss": 2.0299,
      "step": 147556
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.0282037258148193,
      "learning_rate": 0.000171899609227768,
      "loss": 2.0868,
      "step": 147557
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.1202377080917358,
      "learning_rate": 0.00017189591034470422,
      "loss": 2.0892,
      "step": 147558
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.018298864364624,
      "learning_rate": 0.0001718922114854576,
      "loss": 2.1364,
      "step": 147559
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.1914719343185425,
      "learning_rate": 0.00017188851265002842,
      "loss": 2.0213,
      "step": 147560
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.2421525716781616,
      "learning_rate": 0.0001718848138384177,
      "loss": 1.9133,
      "step": 147561
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.3635958433151245,
      "learning_rate": 0.00017188111505062588,
      "loss": 2.1231,
      "step": 147562
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.1269910335540771,
      "learning_rate": 0.0001718774162866539,
      "loss": 2.4197,
      "step": 147563
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.1569442749023438,
      "learning_rate": 0.0001718737175465022,
      "loss": 2.1676,
      "step": 147564
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.5408542156219482,
      "learning_rate": 0.00017187001883017174,
      "loss": 2.1156,
      "step": 147565
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.063840389251709,
      "learning_rate": 0.00017186632013766305,
      "loss": 2.0381,
      "step": 147566
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.1217230558395386,
      "learning_rate": 0.00017186262146897684,
      "loss": 2.2731,
      "step": 147567
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.1295396089553833,
      "learning_rate": 0.00017185892282411364,
      "loss": 2.0305,
      "step": 147568
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.0652027130126953,
      "learning_rate": 0.00017185522420307445,
      "loss": 2.0933,
      "step": 147569
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.0847468376159668,
      "learning_rate": 0.0001718515256058596,
      "loss": 2.2535,
      "step": 147570
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.1069241762161255,
      "learning_rate": 0.0001718478270324702,
      "loss": 1.9374,
      "step": 147571
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.0774810314178467,
      "learning_rate": 0.00017184412848290662,
      "loss": 1.9305,
      "step": 147572
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.128453016281128,
      "learning_rate": 0.00017184042995716965,
      "loss": 2.3023,
      "step": 147573
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.1482709646224976,
      "learning_rate": 0.00017183673145525986,
      "loss": 2.0887,
      "step": 147574
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.0636327266693115,
      "learning_rate": 0.00017183303297717815,
      "loss": 2.0506,
      "step": 147575
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.0655436515808105,
      "learning_rate": 0.000171829334522925,
      "loss": 2.0838,
      "step": 147576
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.0870503187179565,
      "learning_rate": 0.00017182563609250128,
      "loss": 1.9831,
      "step": 147577
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.0517690181732178,
      "learning_rate": 0.00017182193768590763,
      "loss": 2.0319,
      "step": 147578
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.0542529821395874,
      "learning_rate": 0.00017181823930314456,
      "loss": 1.905,
      "step": 147579
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.1550660133361816,
      "learning_rate": 0.00017181454094421304,
      "loss": 2.2494,
      "step": 147580
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.0482943058013916,
      "learning_rate": 0.0001718108426091136,
      "loss": 2.0795,
      "step": 147581
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.1372110843658447,
      "learning_rate": 0.0001718071442978468,
      "loss": 1.9878,
      "step": 147582
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.1476719379425049,
      "learning_rate": 0.0001718034460104136,
      "loss": 2.258,
      "step": 147583
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.212161898612976,
      "learning_rate": 0.00017179974774681457,
      "loss": 2.0547,
      "step": 147584
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.3096837997436523,
      "learning_rate": 0.00017179604950705028,
      "loss": 2.1749,
      "step": 147585
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.2141488790512085,
      "learning_rate": 0.00017179235129112166,
      "loss": 2.1732,
      "step": 147586
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.1423505544662476,
      "learning_rate": 0.0001717886530990291,
      "loss": 2.0657,
      "step": 147587
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.2033143043518066,
      "learning_rate": 0.0001717849549307736,
      "loss": 1.9072,
      "step": 147588
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.131601095199585,
      "learning_rate": 0.00017178125678635572,
      "loss": 2.0526,
      "step": 147589
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.0873994827270508,
      "learning_rate": 0.00017177755866577607,
      "loss": 2.0342,
      "step": 147590
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.1417261362075806,
      "learning_rate": 0.00017177386056903526,
      "loss": 2.0807,
      "step": 147591
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.1275629997253418,
      "learning_rate": 0.00017177016249613427,
      "loss": 2.0733,
      "step": 147592
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.2144813537597656,
      "learning_rate": 0.0001717664644470735,
      "loss": 2.1553,
      "step": 147593
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.1221596002578735,
      "learning_rate": 0.0001717627664218539,
      "loss": 2.0698,
      "step": 147594
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.0498037338256836,
      "learning_rate": 0.000171759068420476,
      "loss": 2.1255,
      "step": 147595
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.0512815713882446,
      "learning_rate": 0.0001717553704429405,
      "loss": 2.1431,
      "step": 147596
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.1134284734725952,
      "learning_rate": 0.000171751672489248,
      "loss": 2.1706,
      "step": 147597
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.2410098314285278,
      "learning_rate": 0.00017174797455939942,
      "loss": 1.9768,
      "step": 147598
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.1281148195266724,
      "learning_rate": 0.00017174427665339514,
      "loss": 2.0715,
      "step": 147599
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.2177013158798218,
      "learning_rate": 0.00017174057877123618,
      "loss": 2.1311,
      "step": 147600
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.145238995552063,
      "learning_rate": 0.00017173688091292293,
      "loss": 2.1119,
      "step": 147601
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.2001523971557617,
      "learning_rate": 0.00017173318307845648,
      "loss": 1.9354,
      "step": 147602
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.1193488836288452,
      "learning_rate": 0.000171729485267837,
      "loss": 2.3811,
      "step": 147603
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.0675840377807617,
      "learning_rate": 0.00017172578748106556,
      "loss": 1.9519,
      "step": 147604
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.1809804439544678,
      "learning_rate": 0.00017172208971814258,
      "loss": 2.0969,
      "step": 147605
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.2314430475234985,
      "learning_rate": 0.00017171839197906898,
      "loss": 2.0468,
      "step": 147606
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.2173194885253906,
      "learning_rate": 0.00017171469426384526,
      "loss": 2.0549,
      "step": 147607
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.1246757507324219,
      "learning_rate": 0.00017171099657247246,
      "loss": 2.1629,
      "step": 147608
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.0040624141693115,
      "learning_rate": 0.00017170729890495076,
      "loss": 2.1409,
      "step": 147609
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.174220085144043,
      "learning_rate": 0.00017170360126128118,
      "loss": 2.0627,
      "step": 147610
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.1811028718948364,
      "learning_rate": 0.00017169990364146424,
      "loss": 2.1916,
      "step": 147611
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.057603120803833,
      "learning_rate": 0.0001716962060455008,
      "loss": 2.1494,
      "step": 147612
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.140661597251892,
      "learning_rate": 0.00017169250847339133,
      "loss": 2.0351,
      "step": 147613
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.1138375997543335,
      "learning_rate": 0.00017168881092513696,
      "loss": 2.2224,
      "step": 147614
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.1062266826629639,
      "learning_rate": 0.00017168511340073775,
      "loss": 1.8992,
      "step": 147615
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.0828101634979248,
      "learning_rate": 0.00017168141590019482,
      "loss": 2.0276,
      "step": 147616
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.1720645427703857,
      "learning_rate": 0.00017167771842350866,
      "loss": 2.0745,
      "step": 147617
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.1960933208465576,
      "learning_rate": 0.0001716740209706801,
      "loss": 2.1895,
      "step": 147618
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.0731898546218872,
      "learning_rate": 0.0001716703235417097,
      "loss": 2.0643,
      "step": 147619
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.1156108379364014,
      "learning_rate": 0.0001716666261365984,
      "loss": 2.0218,
      "step": 147620
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.2193775177001953,
      "learning_rate": 0.0001716629287553465,
      "loss": 2.3119,
      "step": 147621
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.0736407041549683,
      "learning_rate": 0.000171659231397955,
      "loss": 2.0523,
      "step": 147622
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.1520047187805176,
      "learning_rate": 0.00017165553406442433,
      "loss": 1.9169,
      "step": 147623
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.0459506511688232,
      "learning_rate": 0.00017165183675475542,
      "loss": 2.1288,
      "step": 147624
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.174715280532837,
      "learning_rate": 0.00017164813946894875,
      "loss": 2.1495,
      "step": 147625
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.087203860282898,
      "learning_rate": 0.00017164444220700537,
      "loss": 2.0697,
      "step": 147626
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.2164639234542847,
      "learning_rate": 0.00017164074496892545,
      "loss": 2.1456,
      "step": 147627
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.1217657327651978,
      "learning_rate": 0.0001716370477547101,
      "loss": 2.2576,
      "step": 147628
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.2077492475509644,
      "learning_rate": 0.0001716333505643597,
      "loss": 2.0899,
      "step": 147629
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.359159231185913,
      "learning_rate": 0.00017162965339787516,
      "loss": 1.9563,
      "step": 147630
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.343271255493164,
      "learning_rate": 0.000171625956255257,
      "loss": 1.961,
      "step": 147631
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.2981150150299072,
      "learning_rate": 0.00017162225913650627,
      "loss": 2.2245,
      "step": 147632
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.0528262853622437,
      "learning_rate": 0.00017161856204162313,
      "loss": 1.9331,
      "step": 147633
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.1415629386901855,
      "learning_rate": 0.00017161486497060862,
      "loss": 2.1994,
      "step": 147634
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.1473240852355957,
      "learning_rate": 0.00017161116792346326,
      "loss": 1.8694,
      "step": 147635
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.2115968465805054,
      "learning_rate": 0.0001716074709001879,
      "loss": 2.2104,
      "step": 147636
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.1113765239715576,
      "learning_rate": 0.000171603773900783,
      "loss": 2.0437,
      "step": 147637
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.1036968231201172,
      "learning_rate": 0.00017160007692524964,
      "loss": 1.9919,
      "step": 147638
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.106642246246338,
      "learning_rate": 0.00017159637997358798,
      "loss": 2.1385,
      "step": 147639
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.1992173194885254,
      "learning_rate": 0.00017159268304579913,
      "loss": 1.9705,
      "step": 147640
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.200549840927124,
      "learning_rate": 0.00017158898614188348,
      "loss": 1.728,
      "step": 147641
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.2364377975463867,
      "learning_rate": 0.000171585289261842,
      "loss": 2.2719,
      "step": 147642
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.1655521392822266,
      "learning_rate": 0.00017158159240567513,
      "loss": 1.7814,
      "step": 147643
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.1933585405349731,
      "learning_rate": 0.00017157789557338389,
      "loss": 2.1238,
      "step": 147644
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.1911938190460205,
      "learning_rate": 0.00017157419876496845,
      "loss": 2.1439,
      "step": 147645
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.0319855213165283,
      "learning_rate": 0.00017157050198042997,
      "loss": 2.1891,
      "step": 147646
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.1878787279129028,
      "learning_rate": 0.00017156680521976885,
      "loss": 2.1742,
      "step": 147647
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.2518786191940308,
      "learning_rate": 0.00017156310848298598,
      "loss": 2.1481,
      "step": 147648
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.2087743282318115,
      "learning_rate": 0.00017155941177008185,
      "loss": 2.4047,
      "step": 147649
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.121333122253418,
      "learning_rate": 0.00017155571508105735,
      "loss": 2.2736,
      "step": 147650
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.0546011924743652,
      "learning_rate": 0.00017155201841591308,
      "loss": 2.171,
      "step": 147651
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.1387519836425781,
      "learning_rate": 0.00017154832177464972,
      "loss": 2.1934,
      "step": 147652
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.11971914768219,
      "learning_rate": 0.0001715446251572678,
      "loss": 1.8431,
      "step": 147653
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.0813206434249878,
      "learning_rate": 0.00017154092856376828,
      "loss": 1.9233,
      "step": 147654
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.0632566213607788,
      "learning_rate": 0.00017153723199415165,
      "loss": 2.3035,
      "step": 147655
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.1180084943771362,
      "learning_rate": 0.00017153353544841876,
      "loss": 2.0776,
      "step": 147656
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.1149183511734009,
      "learning_rate": 0.0001715298389265702,
      "loss": 1.9946,
      "step": 147657
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.0515236854553223,
      "learning_rate": 0.00017152614242860672,
      "loss": 2.0747,
      "step": 147658
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.053786277770996,
      "learning_rate": 0.0001715224459545288,
      "loss": 2.1195,
      "step": 147659
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.008752465248108,
      "learning_rate": 0.0001715187495043374,
      "loss": 2.1094,
      "step": 147660
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.168669581413269,
      "learning_rate": 0.00017151505307803298,
      "loss": 2.1973,
      "step": 147661
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.0833748579025269,
      "learning_rate": 0.00017151135667561646,
      "loss": 2.0815,
      "step": 147662
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.1701685190200806,
      "learning_rate": 0.0001715076602970884,
      "loss": 1.953,
      "step": 147663
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.0697641372680664,
      "learning_rate": 0.00017150396394244937,
      "loss": 1.9001,
      "step": 147664
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.0865281820297241,
      "learning_rate": 0.0001715002676117003,
      "loss": 1.9893,
      "step": 147665
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.0562493801116943,
      "learning_rate": 0.00017149657130484174,
      "loss": 1.8683,
      "step": 147666
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.2493425607681274,
      "learning_rate": 0.0001714928750218743,
      "loss": 2.1196,
      "step": 147667
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.2226169109344482,
      "learning_rate": 0.00017148917876279885,
      "loss": 2.1244,
      "step": 147668
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.1176133155822754,
      "learning_rate": 0.000171485482527616,
      "loss": 2.1254,
      "step": 147669
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.1776798963546753,
      "learning_rate": 0.00017148178631632633,
      "loss": 2.1103,
      "step": 147670
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.0862785577774048,
      "learning_rate": 0.0001714780901289307,
      "loss": 2.0589,
      "step": 147671
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.1126686334609985,
      "learning_rate": 0.00017147439396542977,
      "loss": 2.0361,
      "step": 147672
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.1051578521728516,
      "learning_rate": 0.00017147069782582402,
      "loss": 1.9608,
      "step": 147673
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.1163129806518555,
      "learning_rate": 0.0001714670017101144,
      "loss": 2.1989,
      "step": 147674
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.095664620399475,
      "learning_rate": 0.0001714633056183015,
      "loss": 1.898,
      "step": 147675
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.1025574207305908,
      "learning_rate": 0.00017145960955038591,
      "loss": 2.3038,
      "step": 147676
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.026576042175293,
      "learning_rate": 0.0001714559135063685,
      "loss": 1.9786,
      "step": 147677
    },
    {
      "epoch": 1.92,
      "grad_norm": 0.980914294719696,
      "learning_rate": 0.00017145221748624974,
      "loss": 1.8668,
      "step": 147678
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.1543294191360474,
      "learning_rate": 0.00017144852149003056,
      "loss": 2.0422,
      "step": 147679
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.0894107818603516,
      "learning_rate": 0.00017144482551771154,
      "loss": 1.9942,
      "step": 147680
    },
    {
      "epoch": 1.92,
      "grad_norm": 2.0539190769195557,
      "learning_rate": 0.00017144112956929332,
      "loss": 2.0509,
      "step": 147681
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.0965532064437866,
      "learning_rate": 0.00017143743364477656,
      "loss": 2.0785,
      "step": 147682
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.1819108724594116,
      "learning_rate": 0.00017143373774416206,
      "loss": 2.0189,
      "step": 147683
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.2668609619140625,
      "learning_rate": 0.0001714300418674504,
      "loss": 2.0427,
      "step": 147684
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.0547189712524414,
      "learning_rate": 0.0001714263460146424,
      "loss": 1.951,
      "step": 147685
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.2140101194381714,
      "learning_rate": 0.00017142265018573854,
      "loss": 1.9269,
      "step": 147686
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.230987548828125,
      "learning_rate": 0.00017141895438073991,
      "loss": 2.0671,
      "step": 147687
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.1457102298736572,
      "learning_rate": 0.0001714152585996467,
      "loss": 2.2744,
      "step": 147688
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.1157156229019165,
      "learning_rate": 0.00017141156284245987,
      "loss": 2.1409,
      "step": 147689
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.2503767013549805,
      "learning_rate": 0.00017140786710918,
      "loss": 1.9499,
      "step": 147690
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.0750982761383057,
      "learning_rate": 0.00017140417139980792,
      "loss": 2.1072,
      "step": 147691
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.1112045049667358,
      "learning_rate": 0.00017140047571434413,
      "loss": 2.156,
      "step": 147692
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.0080047845840454,
      "learning_rate": 0.00017139678005278965,
      "loss": 2.1025,
      "step": 147693
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.2098201513290405,
      "learning_rate": 0.00017139308441514473,
      "loss": 2.1164,
      "step": 147694
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.053652048110962,
      "learning_rate": 0.00017138938880141032,
      "loss": 2.08,
      "step": 147695
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.097730040550232,
      "learning_rate": 0.00017138569321158698,
      "loss": 1.8966,
      "step": 147696
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.0555319786071777,
      "learning_rate": 0.00017138199764567556,
      "loss": 2.1841,
      "step": 147697
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.126882553100586,
      "learning_rate": 0.00017137830210367655,
      "loss": 2.0422,
      "step": 147698
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.1432150602340698,
      "learning_rate": 0.000171374606585591,
      "loss": 1.9468,
      "step": 147699
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.1578134298324585,
      "learning_rate": 0.00017137091109141906,
      "loss": 2.2306,
      "step": 147700
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.0978378057479858,
      "learning_rate": 0.0001713672156211618,
      "loss": 2.0035,
      "step": 147701
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.067999243736267,
      "learning_rate": 0.00017136352017481973,
      "loss": 1.9665,
      "step": 147702
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.032795786857605,
      "learning_rate": 0.00017135982475239374,
      "loss": 1.9395,
      "step": 147703
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.0648900270462036,
      "learning_rate": 0.00017135612935388422,
      "loss": 1.9095,
      "step": 147704
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.204110026359558,
      "learning_rate": 0.00017135243397929228,
      "loss": 1.8794,
      "step": 147705
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.0986926555633545,
      "learning_rate": 0.0001713487386286181,
      "loss": 2.0434,
      "step": 147706
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.07687509059906,
      "learning_rate": 0.00017134504330186277,
      "loss": 2.0883,
      "step": 147707
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.1033512353897095,
      "learning_rate": 0.0001713413479990267,
      "loss": 1.9458,
      "step": 147708
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.1947450637817383,
      "learning_rate": 0.00017133765272011082,
      "loss": 2.1542,
      "step": 147709
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.0782783031463623,
      "learning_rate": 0.00017133395746511554,
      "loss": 1.8882,
      "step": 147710
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.1633678674697876,
      "learning_rate": 0.000171330262234042,
      "loss": 1.8712,
      "step": 147711
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.1677125692367554,
      "learning_rate": 0.0001713265670268903,
      "loss": 2.2237,
      "step": 147712
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.0511705875396729,
      "learning_rate": 0.0001713228718436616,
      "loss": 2.1607,
      "step": 147713
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.4461981058120728,
      "learning_rate": 0.00017131917668435624,
      "loss": 1.9656,
      "step": 147714
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.1865286827087402,
      "learning_rate": 0.00017131548154897524,
      "loss": 2.1184,
      "step": 147715
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.367400884628296,
      "learning_rate": 0.00017131178643751895,
      "loss": 2.1096,
      "step": 147716
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.0490331649780273,
      "learning_rate": 0.00017130809134998845,
      "loss": 2.1817,
      "step": 147717
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.167014241218567,
      "learning_rate": 0.00017130439628638405,
      "loss": 2.2242,
      "step": 147718
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.056938648223877,
      "learning_rate": 0.00017130070124670666,
      "loss": 1.7321,
      "step": 147719
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.0999339818954468,
      "learning_rate": 0.00017129700623095677,
      "loss": 2.1998,
      "step": 147720
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.0501043796539307,
      "learning_rate": 0.0001712933112391353,
      "loss": 2.1496,
      "step": 147721
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.2757359743118286,
      "learning_rate": 0.0001712896162712428,
      "loss": 2.0193,
      "step": 147722
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.4780319929122925,
      "learning_rate": 0.00017128592132728014,
      "loss": 2.0618,
      "step": 147723
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.023358702659607,
      "learning_rate": 0.00017128222640724766,
      "loss": 2.3082,
      "step": 147724
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.08646559715271,
      "learning_rate": 0.0001712785315111464,
      "loss": 2.0484,
      "step": 147725
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.1299805641174316,
      "learning_rate": 0.0001712748366389767,
      "loss": 2.186,
      "step": 147726
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.2704086303710938,
      "learning_rate": 0.0001712711417907396,
      "loss": 1.9979,
      "step": 147727
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.1476147174835205,
      "learning_rate": 0.00017126744696643547,
      "loss": 2.0898,
      "step": 147728
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.036665678024292,
      "learning_rate": 0.00017126375216606544,
      "loss": 1.9926,
      "step": 147729
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.1163959503173828,
      "learning_rate": 0.00017126005738962963,
      "loss": 1.9682,
      "step": 147730
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.221991777420044,
      "learning_rate": 0.0001712563626371291,
      "loss": 2.4317,
      "step": 147731
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.065115213394165,
      "learning_rate": 0.00017125266790856439,
      "loss": 2.1034,
      "step": 147732
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.0421068668365479,
      "learning_rate": 0.00017124897320393631,
      "loss": 2.1624,
      "step": 147733
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.0855814218521118,
      "learning_rate": 0.00017124527852324534,
      "loss": 2.086,
      "step": 147734
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.056504249572754,
      "learning_rate": 0.00017124158386649246,
      "loss": 1.9467,
      "step": 147735
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.081846833229065,
      "learning_rate": 0.00017123788923367823,
      "loss": 2.0154,
      "step": 147736
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.170900583267212,
      "learning_rate": 0.00017123419462480323,
      "loss": 1.9411,
      "step": 147737
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.1773054599761963,
      "learning_rate": 0.00017123050003986815,
      "loss": 2.2905,
      "step": 147738
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.111819863319397,
      "learning_rate": 0.00017122680547887388,
      "loss": 1.9562,
      "step": 147739
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.117426872253418,
      "learning_rate": 0.00017122311094182086,
      "loss": 2.2062,
      "step": 147740
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.1456505060195923,
      "learning_rate": 0.00017121941642871,
      "loss": 2.0419,
      "step": 147741
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.1383968591690063,
      "learning_rate": 0.00017121572193954187,
      "loss": 2.1246,
      "step": 147742
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.1136854887008667,
      "learning_rate": 0.0001712120274743172,
      "loss": 2.2539,
      "step": 147743
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.232983946800232,
      "learning_rate": 0.00017120833303303648,
      "loss": 1.9787,
      "step": 147744
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.3043394088745117,
      "learning_rate": 0.00017120463861570074,
      "loss": 1.9952,
      "step": 147745
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.1552393436431885,
      "learning_rate": 0.00017120094422231033,
      "loss": 1.959,
      "step": 147746
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.120686650276184,
      "learning_rate": 0.0001711972498528662,
      "loss": 1.9495,
      "step": 147747
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.3689061403274536,
      "learning_rate": 0.00017119355550736893,
      "loss": 2.1491,
      "step": 147748
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.3428767919540405,
      "learning_rate": 0.0001711898611858191,
      "loss": 2.0334,
      "step": 147749
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.6551899909973145,
      "learning_rate": 0.00017118616688821767,
      "loss": 2.1558,
      "step": 147750
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.1027737855911255,
      "learning_rate": 0.00017118247261456513,
      "loss": 2.1922,
      "step": 147751
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.1309970617294312,
      "learning_rate": 0.00017117877836486206,
      "loss": 2.1305,
      "step": 147752
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.1802328824996948,
      "learning_rate": 0.00017117508413910944,
      "loss": 2.1962,
      "step": 147753
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.134260654449463,
      "learning_rate": 0.0001711713899373078,
      "loss": 2.3182,
      "step": 147754
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.0649257898330688,
      "learning_rate": 0.00017116769575945768,
      "loss": 1.9332,
      "step": 147755
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.0765269994735718,
      "learning_rate": 0.0001711640016055601,
      "loss": 2.1671,
      "step": 147756
    },
    {
      "epoch": 1.92,
      "grad_norm": 0.9848669171333313,
      "learning_rate": 0.00017116030747561548,
      "loss": 2.0224,
      "step": 147757
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.3894890546798706,
      "learning_rate": 0.0001711566133696245,
      "loss": 2.0954,
      "step": 147758
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.1157019138336182,
      "learning_rate": 0.00017115291928758813,
      "loss": 2.082,
      "step": 147759
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.0355116128921509,
      "learning_rate": 0.00017114922522950678,
      "loss": 2.0572,
      "step": 147760
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.1217236518859863,
      "learning_rate": 0.00017114553119538113,
      "loss": 1.8237,
      "step": 147761
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.248854637145996,
      "learning_rate": 0.00017114183718521209,
      "loss": 1.935,
      "step": 147762
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.3463581800460815,
      "learning_rate": 0.00017113814319900006,
      "loss": 2.0699,
      "step": 147763
    },
    {
      "epoch": 1.92,
      "grad_norm": 0.9967616200447083,
      "learning_rate": 0.000171134449236746,
      "loss": 2.0848,
      "step": 147764
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.2082258462905884,
      "learning_rate": 0.00017113075529845054,
      "loss": 2.0689,
      "step": 147765
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.195779800415039,
      "learning_rate": 0.00017112706138411427,
      "loss": 2.03,
      "step": 147766
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.1858701705932617,
      "learning_rate": 0.00017112336749373777,
      "loss": 2.0891,
      "step": 147767
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.0907337665557861,
      "learning_rate": 0.00017111967362732204,
      "loss": 2.1313,
      "step": 147768
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.0990279912948608,
      "learning_rate": 0.0001711159797848674,
      "loss": 1.9726,
      "step": 147769
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.0964587926864624,
      "learning_rate": 0.0001711122859663749,
      "loss": 1.9556,
      "step": 147770
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.1473044157028198,
      "learning_rate": 0.00017110859217184493,
      "loss": 1.9635,
      "step": 147771
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.2194228172302246,
      "learning_rate": 0.0001711048984012786,
      "loss": 2.1154,
      "step": 147772
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.1459217071533203,
      "learning_rate": 0.000171101204654676,
      "loss": 2.3332,
      "step": 147773
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.126468300819397,
      "learning_rate": 0.00017109751093203826,
      "loss": 1.8178,
      "step": 147774
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.1251530647277832,
      "learning_rate": 0.0001710938172333658,
      "loss": 1.9371,
      "step": 147775
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.268632411956787,
      "learning_rate": 0.0001710901235586595,
      "loss": 2.0096,
      "step": 147776
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.206549882888794,
      "learning_rate": 0.00017108642990791993,
      "loss": 2.1183,
      "step": 147777
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.0621438026428223,
      "learning_rate": 0.00017108273628114806,
      "loss": 2.1604,
      "step": 147778
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.1655702590942383,
      "learning_rate": 0.00017107904267834405,
      "loss": 1.9813,
      "step": 147779
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.2016243934631348,
      "learning_rate": 0.00017107534909950906,
      "loss": 2.1595,
      "step": 147780
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.1590970754623413,
      "learning_rate": 0.00017107165554464343,
      "loss": 2.1657,
      "step": 147781
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.1662323474884033,
      "learning_rate": 0.00017106796201374818,
      "loss": 2.2431,
      "step": 147782
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.0845803022384644,
      "learning_rate": 0.00017106426850682367,
      "loss": 2.0849,
      "step": 147783
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.0904520750045776,
      "learning_rate": 0.00017106057502387103,
      "loss": 1.9346,
      "step": 147784
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.1177568435668945,
      "learning_rate": 0.0001710568815648904,
      "loss": 2.0559,
      "step": 147785
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.1034131050109863,
      "learning_rate": 0.00017105318812988282,
      "loss": 1.8943,
      "step": 147786
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.1166023015975952,
      "learning_rate": 0.0001710494947188488,
      "loss": 1.8802,
      "step": 147787
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.0799944400787354,
      "learning_rate": 0.0001710458013317892,
      "loss": 2.2553,
      "step": 147788
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.0780526399612427,
      "learning_rate": 0.00017104210796870455,
      "loss": 2.1366,
      "step": 147789
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.038008689880371,
      "learning_rate": 0.00017103841462959583,
      "loss": 2.0868,
      "step": 147790
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.1048798561096191,
      "learning_rate": 0.00017103472131446324,
      "loss": 2.1415,
      "step": 147791
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.0457149744033813,
      "learning_rate": 0.00017103102802330789,
      "loss": 2.0422,
      "step": 147792
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.1504535675048828,
      "learning_rate": 0.00017102733475613012,
      "loss": 1.8962,
      "step": 147793
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.066605806350708,
      "learning_rate": 0.00017102364151293098,
      "loss": 2.0654,
      "step": 147794
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.205618143081665,
      "learning_rate": 0.00017101994829371085,
      "loss": 2.1264,
      "step": 147795
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.0247259140014648,
      "learning_rate": 0.0001710162550984708,
      "loss": 2.1288,
      "step": 147796
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.157348871231079,
      "learning_rate": 0.000171012561927211,
      "loss": 2.0409,
      "step": 147797
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.0967432260513306,
      "learning_rate": 0.00017100886877993252,
      "loss": 2.12,
      "step": 147798
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.0880037546157837,
      "learning_rate": 0.0001710051756566358,
      "loss": 2.1006,
      "step": 147799
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.280595064163208,
      "learning_rate": 0.0001710014825573218,
      "loss": 2.086,
      "step": 147800
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.1942732334136963,
      "learning_rate": 0.0001709977894819909,
      "loss": 2.1575,
      "step": 147801
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.0635673999786377,
      "learning_rate": 0.00017099409643064407,
      "loss": 2.0558,
      "step": 147802
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.2216672897338867,
      "learning_rate": 0.0001709904034032819,
      "loss": 1.9581,
      "step": 147803
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.22055983543396,
      "learning_rate": 0.000170986710399905,
      "loss": 2.1938,
      "step": 147804
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.084734559059143,
      "learning_rate": 0.00017098301742051405,
      "loss": 2.1816,
      "step": 147805
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.2500274181365967,
      "learning_rate": 0.00017097932446510992,
      "loss": 2.0602,
      "step": 147806
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.259432077407837,
      "learning_rate": 0.00017097563153369298,
      "loss": 2.1405,
      "step": 147807
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.1464130878448486,
      "learning_rate": 0.00017097193862626426,
      "loss": 2.206,
      "step": 147808
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.1685971021652222,
      "learning_rate": 0.00017096824574282428,
      "loss": 2.0444,
      "step": 147809
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.2026413679122925,
      "learning_rate": 0.00017096455288337378,
      "loss": 2.0158,
      "step": 147810
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.275611400604248,
      "learning_rate": 0.00017096086004791323,
      "loss": 1.8082,
      "step": 147811
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.1512057781219482,
      "learning_rate": 0.00017095716723644367,
      "loss": 2.0175,
      "step": 147812
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.2528623342514038,
      "learning_rate": 0.00017095347444896544,
      "loss": 2.2458,
      "step": 147813
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.189986228942871,
      "learning_rate": 0.00017094978168547955,
      "loss": 1.9792,
      "step": 147814
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.1177762746810913,
      "learning_rate": 0.00017094608894598648,
      "loss": 1.8074,
      "step": 147815
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.0218385457992554,
      "learning_rate": 0.00017094239623048705,
      "loss": 2.2827,
      "step": 147816
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.0481014251708984,
      "learning_rate": 0.00017093870353898167,
      "loss": 2.0753,
      "step": 147817
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.1084864139556885,
      "learning_rate": 0.00017093501087147136,
      "loss": 2.023,
      "step": 147818
    },
    {
      "epoch": 1.92,
      "grad_norm": 2.763049602508545,
      "learning_rate": 0.00017093131822795658,
      "loss": 2.096,
      "step": 147819
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.235295057296753,
      "learning_rate": 0.00017092762560843818,
      "loss": 2.2889,
      "step": 147820
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.0633573532104492,
      "learning_rate": 0.00017092393301291683,
      "loss": 2.03,
      "step": 147821
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.210181713104248,
      "learning_rate": 0.00017092024044139308,
      "loss": 1.9334,
      "step": 147822
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.2234548330307007,
      "learning_rate": 0.00017091654789386763,
      "loss": 2.2491,
      "step": 147823
    },
    {
      "epoch": 1.92,
      "grad_norm": 0.9948977828025818,
      "learning_rate": 0.00017091285537034132,
      "loss": 1.9256,
      "step": 147824
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.0764435529708862,
      "learning_rate": 0.00017090916287081464,
      "loss": 2.0507,
      "step": 147825
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.1095001697540283,
      "learning_rate": 0.0001709054703952885,
      "loss": 2.1346,
      "step": 147826
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.1136958599090576,
      "learning_rate": 0.0001709017779437635,
      "loss": 2.1109,
      "step": 147827
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.165098786354065,
      "learning_rate": 0.00017089808551624025,
      "loss": 2.0827,
      "step": 147828
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.0959962606430054,
      "learning_rate": 0.00017089439311271937,
      "loss": 2.1821,
      "step": 147829
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.0158213376998901,
      "learning_rate": 0.0001708907007332018,
      "loss": 2.0912,
      "step": 147830
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.2267440557479858,
      "learning_rate": 0.00017088700837768794,
      "loss": 2.0165,
      "step": 147831
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.1562645435333252,
      "learning_rate": 0.0001708833160461788,
      "loss": 2.3494,
      "step": 147832
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.1408051252365112,
      "learning_rate": 0.00017087962373867486,
      "loss": 2.0424,
      "step": 147833
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.439834713935852,
      "learning_rate": 0.00017087593145517676,
      "loss": 2.1078,
      "step": 147834
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.01907479763031,
      "learning_rate": 0.00017087223919568523,
      "loss": 1.7405,
      "step": 147835
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.2784180641174316,
      "learning_rate": 0.0001708685469602011,
      "loss": 2.0425,
      "step": 147836
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.114503264427185,
      "learning_rate": 0.0001708648547487248,
      "loss": 2.096,
      "step": 147837
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.0995275974273682,
      "learning_rate": 0.00017086116256125728,
      "loss": 2.073,
      "step": 147838
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.119755506515503,
      "learning_rate": 0.00017085747039779912,
      "loss": 2.1683,
      "step": 147839
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.0462067127227783,
      "learning_rate": 0.00017085377825835088,
      "loss": 2.1151,
      "step": 147840
    },
    {
      "epoch": 1.92,
      "grad_norm": 1.0941373109817505,
      "learning_rate": 0.0001708500861429135,
      "loss": 2.2717,
      "step": 147841
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.168912410736084,
      "learning_rate": 0.00017084639405148745,
      "loss": 2.1955,
      "step": 147842
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.1236084699630737,
      "learning_rate": 0.00017084270198407342,
      "loss": 2.121,
      "step": 147843
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.2999918460845947,
      "learning_rate": 0.0001708390099406723,
      "loss": 2.0273,
      "step": 147844
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.1075685024261475,
      "learning_rate": 0.00017083531792128466,
      "loss": 2.3897,
      "step": 147845
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.2010483741760254,
      "learning_rate": 0.00017083162592591104,
      "loss": 2.0873,
      "step": 147846
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.09039306640625,
      "learning_rate": 0.00017082793395455236,
      "loss": 2.103,
      "step": 147847
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.0640300512313843,
      "learning_rate": 0.0001708242420072091,
      "loss": 1.8657,
      "step": 147848
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.0908865928649902,
      "learning_rate": 0.00017082055008388214,
      "loss": 2.0404,
      "step": 147849
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.1609817743301392,
      "learning_rate": 0.00017081685818457214,
      "loss": 1.9909,
      "step": 147850
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.2836843729019165,
      "learning_rate": 0.00017081316630927965,
      "loss": 2.0458,
      "step": 147851
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.1644775867462158,
      "learning_rate": 0.00017080947445800537,
      "loss": 2.0969,
      "step": 147852
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.1681703329086304,
      "learning_rate": 0.00017080578263075014,
      "loss": 2.0246,
      "step": 147853
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.171836256980896,
      "learning_rate": 0.00017080209082751444,
      "loss": 2.1559,
      "step": 147854
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.2618944644927979,
      "learning_rate": 0.0001707983990482992,
      "loss": 2.4077,
      "step": 147855
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.1892834901809692,
      "learning_rate": 0.00017079470729310496,
      "loss": 2.0597,
      "step": 147856
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.0769062042236328,
      "learning_rate": 0.00017079101556193244,
      "loss": 2.307,
      "step": 147857
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.0606954097747803,
      "learning_rate": 0.0001707873238547822,
      "loss": 2.0564,
      "step": 147858
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.160900592803955,
      "learning_rate": 0.00017078363217165515,
      "loss": 2.2868,
      "step": 147859
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.2501919269561768,
      "learning_rate": 0.00017077994051255168,
      "loss": 1.9625,
      "step": 147860
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.0956066846847534,
      "learning_rate": 0.00017077624887747284,
      "loss": 1.8672,
      "step": 147861
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.0600066184997559,
      "learning_rate": 0.000170772557266419,
      "loss": 1.9473,
      "step": 147862
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.3500771522521973,
      "learning_rate": 0.00017076886567939122,
      "loss": 1.9715,
      "step": 147863
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.3417072296142578,
      "learning_rate": 0.0001707651741163897,
      "loss": 1.77,
      "step": 147864
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.1671339273452759,
      "learning_rate": 0.00017076148257741554,
      "loss": 2.247,
      "step": 147865
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.1315150260925293,
      "learning_rate": 0.00017075779106246906,
      "loss": 2.1415,
      "step": 147866
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.4387692213058472,
      "learning_rate": 0.00017075409957155132,
      "loss": 2.2319,
      "step": 147867
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.810982584953308,
      "learning_rate": 0.0001707504081046627,
      "loss": 2.1215,
      "step": 147868
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.2898757457733154,
      "learning_rate": 0.00017074671666180414,
      "loss": 2.0481,
      "step": 147869
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.3729853630065918,
      "learning_rate": 0.00017074302524297622,
      "loss": 1.9418,
      "step": 147870
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.2364239692687988,
      "learning_rate": 0.0001707393338481796,
      "loss": 2.2284,
      "step": 147871
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.3045947551727295,
      "learning_rate": 0.00017073564247741484,
      "loss": 2.0657,
      "step": 147872
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.1295908689498901,
      "learning_rate": 0.00017073195113068292,
      "loss": 2.0913,
      "step": 147873
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.1962323188781738,
      "learning_rate": 0.00017072825980798422,
      "loss": 1.9881,
      "step": 147874
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.1383758783340454,
      "learning_rate": 0.00017072456850931967,
      "loss": 2.0296,
      "step": 147875
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.1726971864700317,
      "learning_rate": 0.0001707208772346899,
      "loss": 2.1464,
      "step": 147876
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.3541423082351685,
      "learning_rate": 0.0001707171859840956,
      "loss": 2.0745,
      "step": 147877
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.1760152578353882,
      "learning_rate": 0.00017071349475753726,
      "loss": 2.2289,
      "step": 147878
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.13443922996521,
      "learning_rate": 0.00017070980355501583,
      "loss": 2.1005,
      "step": 147879
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.1837540864944458,
      "learning_rate": 0.00017070611237653176,
      "loss": 1.9581,
      "step": 147880
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.1170146465301514,
      "learning_rate": 0.000170702421222086,
      "loss": 2.0928,
      "step": 147881
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.0049582719802856,
      "learning_rate": 0.00017069873009167906,
      "loss": 2.2228,
      "step": 147882
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.098914384841919,
      "learning_rate": 0.00017069503898531173,
      "loss": 2.0617,
      "step": 147883
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.0868722200393677,
      "learning_rate": 0.0001706913479029845,
      "loss": 2.0459,
      "step": 147884
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.2262431383132935,
      "learning_rate": 0.00017068765684469833,
      "loss": 2.2665,
      "step": 147885
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.0413517951965332,
      "learning_rate": 0.00017068396581045356,
      "loss": 2.0128,
      "step": 147886
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.2260410785675049,
      "learning_rate": 0.00017068027480025125,
      "loss": 2.2822,
      "step": 147887
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.2314051389694214,
      "learning_rate": 0.00017067658381409194,
      "loss": 1.8929,
      "step": 147888
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.1035666465759277,
      "learning_rate": 0.00017067289285197627,
      "loss": 1.992,
      "step": 147889
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.1360307931900024,
      "learning_rate": 0.00017066920191390484,
      "loss": 2.2812,
      "step": 147890
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.0611989498138428,
      "learning_rate": 0.00017066551099987855,
      "loss": 2.008,
      "step": 147891
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.1665079593658447,
      "learning_rate": 0.00017066182010989792,
      "loss": 2.1222,
      "step": 147892
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.1104400157928467,
      "learning_rate": 0.00017065812924396375,
      "loss": 2.0751,
      "step": 147893
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.128389835357666,
      "learning_rate": 0.00017065443840207672,
      "loss": 2.039,
      "step": 147894
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.1120299100875854,
      "learning_rate": 0.00017065074758423744,
      "loss": 2.1621,
      "step": 147895
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.0293536186218262,
      "learning_rate": 0.0001706470567904465,
      "loss": 2.1172,
      "step": 147896
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.1135063171386719,
      "learning_rate": 0.00017064336602070487,
      "loss": 1.9581,
      "step": 147897
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.1012204885482788,
      "learning_rate": 0.00017063967527501295,
      "loss": 2.1398,
      "step": 147898
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.0517491102218628,
      "learning_rate": 0.00017063598455337167,
      "loss": 2.0128,
      "step": 147899
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.1885486841201782,
      "learning_rate": 0.00017063229385578163,
      "loss": 2.0456,
      "step": 147900
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.0678033828735352,
      "learning_rate": 0.00017062860318224344,
      "loss": 1.976,
      "step": 147901
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.1033909320831299,
      "learning_rate": 0.00017062491253275777,
      "loss": 2.0881,
      "step": 147902
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.0990629196166992,
      "learning_rate": 0.00017062122190732545,
      "loss": 2.124,
      "step": 147903
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.0935404300689697,
      "learning_rate": 0.00017061753130594693,
      "loss": 1.9653,
      "step": 147904
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.0155247449874878,
      "learning_rate": 0.00017061384072862326,
      "loss": 1.8475,
      "step": 147905
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.1763629913330078,
      "learning_rate": 0.00017061015017535487,
      "loss": 1.9104,
      "step": 147906
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.366660475730896,
      "learning_rate": 0.0001706064596461425,
      "loss": 2.2474,
      "step": 147907
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.0840057134628296,
      "learning_rate": 0.0001706027691409867,
      "loss": 2.2077,
      "step": 147908
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.1212512254714966,
      "learning_rate": 0.00017059907865988846,
      "loss": 2.0302,
      "step": 147909
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.0739350318908691,
      "learning_rate": 0.00017059538820284807,
      "loss": 2.1397,
      "step": 147910
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.1560568809509277,
      "learning_rate": 0.00017059169776986667,
      "loss": 2.0565,
      "step": 147911
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.1618032455444336,
      "learning_rate": 0.00017058800736094465,
      "loss": 2.1764,
      "step": 147912
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.0524122714996338,
      "learning_rate": 0.0001705843169760828,
      "loss": 1.9305,
      "step": 147913
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.0579500198364258,
      "learning_rate": 0.00017058062661528165,
      "loss": 2.2796,
      "step": 147914
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.1536840200424194,
      "learning_rate": 0.00017057693627854208,
      "loss": 2.193,
      "step": 147915
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.1099869012832642,
      "learning_rate": 0.0001705732459658646,
      "loss": 2.3733,
      "step": 147916
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.154872179031372,
      "learning_rate": 0.00017056955567725012,
      "loss": 1.888,
      "step": 147917
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.1490910053253174,
      "learning_rate": 0.00017056586541269923,
      "loss": 2.1804,
      "step": 147918
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.0846027135849,
      "learning_rate": 0.00017056217517221256,
      "loss": 2.114,
      "step": 147919
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.1450074911117554,
      "learning_rate": 0.0001705584849557907,
      "loss": 2.0307,
      "step": 147920
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.033586025238037,
      "learning_rate": 0.00017055479476343462,
      "loss": 2.0714,
      "step": 147921
    },
    {
      "epoch": 1.93,
      "grad_norm": 0.9918425679206848,
      "learning_rate": 0.00017055110459514468,
      "loss": 1.8628,
      "step": 147922
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.0104691982269287,
      "learning_rate": 0.00017054741445092188,
      "loss": 1.9616,
      "step": 147923
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.0653470754623413,
      "learning_rate": 0.00017054372433076678,
      "loss": 2.0977,
      "step": 147924
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.093509316444397,
      "learning_rate": 0.00017054003423467987,
      "loss": 2.2312,
      "step": 147925
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.221739411354065,
      "learning_rate": 0.00017053634416266212,
      "loss": 2.0307,
      "step": 147926
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.1452102661132812,
      "learning_rate": 0.0001705326541147142,
      "loss": 2.1621,
      "step": 147927
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.2842423915863037,
      "learning_rate": 0.00017052896409083652,
      "loss": 2.2545,
      "step": 147928
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.1501047611236572,
      "learning_rate": 0.00017052527409103005,
      "loss": 2.1311,
      "step": 147929
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.0462205410003662,
      "learning_rate": 0.00017052158411529544,
      "loss": 2.2118,
      "step": 147930
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.0992531776428223,
      "learning_rate": 0.00017051789416363313,
      "loss": 1.8688,
      "step": 147931
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.1132638454437256,
      "learning_rate": 0.00017051420423604415,
      "loss": 1.9923,
      "step": 147932
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.0636781454086304,
      "learning_rate": 0.0001705105143325289,
      "loss": 2.1035,
      "step": 147933
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.0036261081695557,
      "learning_rate": 0.0001705068244530883,
      "loss": 2.0557,
      "step": 147934
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.038608431816101,
      "learning_rate": 0.0001705031345977229,
      "loss": 2.2256,
      "step": 147935
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.0396910905838013,
      "learning_rate": 0.00017049944476643333,
      "loss": 2.3734,
      "step": 147936
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.0862540006637573,
      "learning_rate": 0.00017049575495922044,
      "loss": 2.0778,
      "step": 147937
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.0408788919448853,
      "learning_rate": 0.00017049206517608486,
      "loss": 2.0362,
      "step": 147938
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.1368820667266846,
      "learning_rate": 0.0001704883754170271,
      "loss": 1.9952,
      "step": 147939
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.1492465734481812,
      "learning_rate": 0.00017048468568204816,
      "loss": 2.0723,
      "step": 147940
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.058332920074463,
      "learning_rate": 0.00017048099597114855,
      "loss": 2.2521,
      "step": 147941
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.1356149911880493,
      "learning_rate": 0.00017047730628432883,
      "loss": 1.9067,
      "step": 147942
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.232174038887024,
      "learning_rate": 0.00017047361662158998,
      "loss": 2.1928,
      "step": 147943
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.0343043804168701,
      "learning_rate": 0.0001704699269829325,
      "loss": 1.9843,
      "step": 147944
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.1868923902511597,
      "learning_rate": 0.00017046623736835702,
      "loss": 2.0536,
      "step": 147945
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.0916533470153809,
      "learning_rate": 0.00017046254777786435,
      "loss": 2.1218,
      "step": 147946
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.1701074838638306,
      "learning_rate": 0.0001704588582114551,
      "loss": 2.2695,
      "step": 147947
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.1440316438674927,
      "learning_rate": 0.0001704551686691301,
      "loss": 2.2157,
      "step": 147948
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.1623181104660034,
      "learning_rate": 0.0001704514791508899,
      "loss": 2.1168,
      "step": 147949
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.1291255950927734,
      "learning_rate": 0.0001704477896567352,
      "loss": 2.0783,
      "step": 147950
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.2602640390396118,
      "learning_rate": 0.00017044410018666662,
      "loss": 2.1275,
      "step": 147951
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.0903149843215942,
      "learning_rate": 0.00017044041074068504,
      "loss": 2.0769,
      "step": 147952
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.0965348482131958,
      "learning_rate": 0.0001704367213187909,
      "loss": 2.0599,
      "step": 147953
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.4478721618652344,
      "learning_rate": 0.00017043303192098514,
      "loss": 2.1331,
      "step": 147954
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.1987570524215698,
      "learning_rate": 0.00017042934254726835,
      "loss": 1.9514,
      "step": 147955
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.1524113416671753,
      "learning_rate": 0.00017042565319764117,
      "loss": 1.7987,
      "step": 147956
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.1712161302566528,
      "learning_rate": 0.0001704219638721042,
      "loss": 1.9822,
      "step": 147957
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.1855682134628296,
      "learning_rate": 0.00017041827457065832,
      "loss": 2.1237,
      "step": 147958
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.1871395111083984,
      "learning_rate": 0.00017041458529330406,
      "loss": 2.0744,
      "step": 147959
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.2247180938720703,
      "learning_rate": 0.00017041089604004226,
      "loss": 2.1366,
      "step": 147960
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.0915569067001343,
      "learning_rate": 0.00017040720681087354,
      "loss": 1.9569,
      "step": 147961
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.2285858392715454,
      "learning_rate": 0.00017040351760579854,
      "loss": 2.137,
      "step": 147962
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.138507604598999,
      "learning_rate": 0.00017039982842481787,
      "loss": 2.0726,
      "step": 147963
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.1355808973312378,
      "learning_rate": 0.00017039613926793243,
      "loss": 2.1224,
      "step": 147964
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.3547414541244507,
      "learning_rate": 0.0001703924501351427,
      "loss": 2.1628,
      "step": 147965
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.119206190109253,
      "learning_rate": 0.00017038876102644954,
      "loss": 2.0336,
      "step": 147966
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.2360233068466187,
      "learning_rate": 0.00017038507194185357,
      "loss": 2.4215,
      "step": 147967
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.1299057006835938,
      "learning_rate": 0.0001703813828813555,
      "loss": 1.9644,
      "step": 147968
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.2595590353012085,
      "learning_rate": 0.0001703776938449558,
      "loss": 2.1227,
      "step": 147969
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.2336328029632568,
      "learning_rate": 0.0001703740048326555,
      "loss": 2.0891,
      "step": 147970
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.0833829641342163,
      "learning_rate": 0.00017037031584445498,
      "loss": 1.9303,
      "step": 147971
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.2699968814849854,
      "learning_rate": 0.00017036662688035516,
      "loss": 2.0548,
      "step": 147972
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.486961841583252,
      "learning_rate": 0.00017036293794035667,
      "loss": 2.0781,
      "step": 147973
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.2379776239395142,
      "learning_rate": 0.00017035924902446012,
      "loss": 2.1532,
      "step": 147974
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.1514712572097778,
      "learning_rate": 0.00017035556013266615,
      "loss": 1.9843,
      "step": 147975
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.1910464763641357,
      "learning_rate": 0.00017035187126497565,
      "loss": 1.9895,
      "step": 147976
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.2236956357955933,
      "learning_rate": 0.00017034818242138905,
      "loss": 2.1808,
      "step": 147977
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.0979788303375244,
      "learning_rate": 0.0001703444936019073,
      "loss": 1.9888,
      "step": 147978
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.165946364402771,
      "learning_rate": 0.00017034080480653093,
      "loss": 2.235,
      "step": 147979
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.140818476676941,
      "learning_rate": 0.00017033711603526067,
      "loss": 2.1892,
      "step": 147980
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.0718446969985962,
      "learning_rate": 0.00017033342728809707,
      "loss": 2.042,
      "step": 147981
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.278875708580017,
      "learning_rate": 0.00017032973856504104,
      "loss": 2.1036,
      "step": 147982
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.1368333101272583,
      "learning_rate": 0.00017032604986609305,
      "loss": 2.0371,
      "step": 147983
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.2457590103149414,
      "learning_rate": 0.000170322361191254,
      "loss": 2.0442,
      "step": 147984
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.198004961013794,
      "learning_rate": 0.00017031867254052447,
      "loss": 1.8621,
      "step": 147985
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.4271488189697266,
      "learning_rate": 0.00017031498391390518,
      "loss": 2.0494,
      "step": 147986
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.0771546363830566,
      "learning_rate": 0.00017031129531139662,
      "loss": 2.0871,
      "step": 147987
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.0320658683776855,
      "learning_rate": 0.00017030760673299977,
      "loss": 2.0366,
      "step": 147988
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.3588054180145264,
      "learning_rate": 0.0001703039181787151,
      "loss": 1.9901,
      "step": 147989
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.1557347774505615,
      "learning_rate": 0.00017030022964854347,
      "loss": 2.1787,
      "step": 147990
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.2613502740859985,
      "learning_rate": 0.0001702965411424855,
      "loss": 2.1093,
      "step": 147991
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.0313831567764282,
      "learning_rate": 0.0001702928526605418,
      "loss": 2.0712,
      "step": 147992
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.2031197547912598,
      "learning_rate": 0.00017028916420271304,
      "loss": 2.0159,
      "step": 147993
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.040298342704773,
      "learning_rate": 0.00017028547576900004,
      "loss": 2.1151,
      "step": 147994
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.1283124685287476,
      "learning_rate": 0.00017028178735940336,
      "loss": 2.1853,
      "step": 147995
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.0963712930679321,
      "learning_rate": 0.00017027809897392384,
      "loss": 2.4365,
      "step": 147996
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.1406418085098267,
      "learning_rate": 0.00017027441061256206,
      "loss": 2.075,
      "step": 147997
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.029355764389038,
      "learning_rate": 0.00017027072227531872,
      "loss": 2.1809,
      "step": 147998
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.1214662790298462,
      "learning_rate": 0.0001702670339621944,
      "loss": 2.0271,
      "step": 147999
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.0964007377624512,
      "learning_rate": 0.00017026334567318998,
      "loss": 2.1376,
      "step": 148000
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.0724914073944092,
      "learning_rate": 0.00017025965740830593,
      "loss": 1.9573,
      "step": 148001
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.1214755773544312,
      "learning_rate": 0.00017025596916754316,
      "loss": 2.0306,
      "step": 148002
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.2663986682891846,
      "learning_rate": 0.0001702522809509022,
      "loss": 2.2105,
      "step": 148003
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.086608648300171,
      "learning_rate": 0.00017024859275838398,
      "loss": 1.9711,
      "step": 148004
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.1134721040725708,
      "learning_rate": 0.00017024490458998875,
      "loss": 2.2864,
      "step": 148005
    },
    {
      "epoch": 1.93,
      "grad_norm": 0.9540960788726807,
      "learning_rate": 0.0001702412164457176,
      "loss": 1.9791,
      "step": 148006
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.0882683992385864,
      "learning_rate": 0.00017023752832557093,
      "loss": 1.938,
      "step": 148007
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.1819242238998413,
      "learning_rate": 0.00017023384022954964,
      "loss": 2.1802,
      "step": 148008
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.1572153568267822,
      "learning_rate": 0.00017023015215765423,
      "loss": 2.2209,
      "step": 148009
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.1489202976226807,
      "learning_rate": 0.0001702264641098856,
      "loss": 1.9714,
      "step": 148010
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.0549116134643555,
      "learning_rate": 0.00017022277608624433,
      "loss": 1.9717,
      "step": 148011
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.0607353448867798,
      "learning_rate": 0.00017021908808673112,
      "loss": 2.0189,
      "step": 148012
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.3355040550231934,
      "learning_rate": 0.00017021540011134643,
      "loss": 2.2044,
      "step": 148013
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.08633291721344,
      "learning_rate": 0.00017021171216009137,
      "loss": 1.8063,
      "step": 148014
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.0808557271957397,
      "learning_rate": 0.0001702080242329662,
      "loss": 1.808,
      "step": 148015
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.1904094219207764,
      "learning_rate": 0.00017020433632997198,
      "loss": 2.0368,
      "step": 148016
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.2117635011672974,
      "learning_rate": 0.0001702006484511092,
      "loss": 2.062,
      "step": 148017
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.1573965549468994,
      "learning_rate": 0.00017019696059637857,
      "loss": 2.1947,
      "step": 148018
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.1265708208084106,
      "learning_rate": 0.00017019327276578068,
      "loss": 1.8263,
      "step": 148019
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.0659089088439941,
      "learning_rate": 0.0001701895849593164,
      "loss": 2.2113,
      "step": 148020
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.0441584587097168,
      "learning_rate": 0.00017018589717698624,
      "loss": 2.0271,
      "step": 148021
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.1440950632095337,
      "learning_rate": 0.00017018220941879104,
      "loss": 1.9746,
      "step": 148022
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.0927562713623047,
      "learning_rate": 0.00017017852168473143,
      "loss": 2.0684,
      "step": 148023
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.3004931211471558,
      "learning_rate": 0.000170174833974808,
      "loss": 2.142,
      "step": 148024
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.042625904083252,
      "learning_rate": 0.0001701711462890216,
      "loss": 2.1886,
      "step": 148025
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.180755376815796,
      "learning_rate": 0.00017016745862737288,
      "loss": 1.9339,
      "step": 148026
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.0720634460449219,
      "learning_rate": 0.00017016377098986234,
      "loss": 1.9029,
      "step": 148027
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.066060185432434,
      "learning_rate": 0.00017016008337649092,
      "loss": 2.3073,
      "step": 148028
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.0162372589111328,
      "learning_rate": 0.00017015639578725917,
      "loss": 1.7844,
      "step": 148029
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.064106822013855,
      "learning_rate": 0.0001701527082221677,
      "loss": 2.0257,
      "step": 148030
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.0810941457748413,
      "learning_rate": 0.00017014902068121743,
      "loss": 1.9888,
      "step": 148031
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.1371418237686157,
      "learning_rate": 0.00017014533316440876,
      "loss": 1.901,
      "step": 148032
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.0764920711517334,
      "learning_rate": 0.00017014164567174264,
      "loss": 2.0352,
      "step": 148033
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.0544041395187378,
      "learning_rate": 0.00017013795820321965,
      "loss": 1.9621,
      "step": 148034
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.193399429321289,
      "learning_rate": 0.00017013427075884049,
      "loss": 1.929,
      "step": 148035
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.0934041738510132,
      "learning_rate": 0.00017013058333860566,
      "loss": 1.8826,
      "step": 148036
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.0685442686080933,
      "learning_rate": 0.00017012689594251616,
      "loss": 2.0665,
      "step": 148037
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.3412292003631592,
      "learning_rate": 0.00017012320857057238,
      "loss": 2.0361,
      "step": 148038
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.289230465888977,
      "learning_rate": 0.00017011952122277524,
      "loss": 2.229,
      "step": 148039
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.1823550462722778,
      "learning_rate": 0.00017011583389912538,
      "loss": 2.1792,
      "step": 148040
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.442469596862793,
      "learning_rate": 0.00017011214659962343,
      "loss": 2.009,
      "step": 148041
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.1792610883712769,
      "learning_rate": 0.00017010845932426998,
      "loss": 1.8591,
      "step": 148042
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.0872952938079834,
      "learning_rate": 0.00017010477207306588,
      "loss": 2.1448,
      "step": 148043
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.1314500570297241,
      "learning_rate": 0.00017010108484601165,
      "loss": 1.9157,
      "step": 148044
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.156456708908081,
      "learning_rate": 0.00017009739764310827,
      "loss": 1.9625,
      "step": 148045
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.1804907321929932,
      "learning_rate": 0.00017009371046435617,
      "loss": 1.8293,
      "step": 148046
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.1177688837051392,
      "learning_rate": 0.0001700900233097561,
      "loss": 2.2223,
      "step": 148047
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.050861120223999,
      "learning_rate": 0.00017008633617930866,
      "loss": 2.1931,
      "step": 148048
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.178255319595337,
      "learning_rate": 0.0001700826490730147,
      "loss": 2.1365,
      "step": 148049
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.0800085067749023,
      "learning_rate": 0.00017007896199087478,
      "loss": 1.923,
      "step": 148050
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.0866482257843018,
      "learning_rate": 0.0001700752749328897,
      "loss": 1.9699,
      "step": 148051
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.218116283416748,
      "learning_rate": 0.00017007158789906007,
      "loss": 2.1978,
      "step": 148052
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.166929006576538,
      "learning_rate": 0.00017006790088938662,
      "loss": 2.0069,
      "step": 148053
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.1128531694412231,
      "learning_rate": 0.00017006421390386984,
      "loss": 1.9668,
      "step": 148054
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.0966870784759521,
      "learning_rate": 0.00017006052694251074,
      "loss": 2.034,
      "step": 148055
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.1468017101287842,
      "learning_rate": 0.0001700568400053097,
      "loss": 1.9834,
      "step": 148056
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.2674754858016968,
      "learning_rate": 0.00017005315309226766,
      "loss": 2.1971,
      "step": 148057
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.102007508277893,
      "learning_rate": 0.0001700494662033852,
      "loss": 2.0859,
      "step": 148058
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.109352707862854,
      "learning_rate": 0.000170045779338663,
      "loss": 2.1494,
      "step": 148059
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.0930322408676147,
      "learning_rate": 0.0001700420924981016,
      "loss": 2.138,
      "step": 148060
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.201471209526062,
      "learning_rate": 0.00017003840568170195,
      "loss": 2.2162,
      "step": 148061
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.1422185897827148,
      "learning_rate": 0.00017003471888946454,
      "loss": 2.08,
      "step": 148062
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.1719214916229248,
      "learning_rate": 0.0001700310321213902,
      "loss": 1.9403,
      "step": 148063
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.1531758308410645,
      "learning_rate": 0.00017002734537747944,
      "loss": 2.0697,
      "step": 148064
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.2669177055358887,
      "learning_rate": 0.00017002365865773328,
      "loss": 2.1901,
      "step": 148065
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.304465413093567,
      "learning_rate": 0.00017001997196215197,
      "loss": 2.0998,
      "step": 148066
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.2407422065734863,
      "learning_rate": 0.00017001628529073648,
      "loss": 2.031,
      "step": 148067
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.1374821662902832,
      "learning_rate": 0.0001700125986434873,
      "loss": 2.1357,
      "step": 148068
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.0784881114959717,
      "learning_rate": 0.0001700089120204054,
      "loss": 1.9752,
      "step": 148069
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.3454835414886475,
      "learning_rate": 0.00017000522542149116,
      "loss": 1.857,
      "step": 148070
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.2748043537139893,
      "learning_rate": 0.0001700015388467456,
      "loss": 2.2304,
      "step": 148071
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.1265839338302612,
      "learning_rate": 0.000169997852296169,
      "loss": 2.0818,
      "step": 148072
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.1877435445785522,
      "learning_rate": 0.00016999416576976236,
      "loss": 2.2831,
      "step": 148073
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.1415431499481201,
      "learning_rate": 0.00016999047926752615,
      "loss": 2.1863,
      "step": 148074
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.2337554693222046,
      "learning_rate": 0.00016998679278946126,
      "loss": 1.9761,
      "step": 148075
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.1202408075332642,
      "learning_rate": 0.0001699831063355682,
      "loss": 1.8214,
      "step": 148076
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.160237431526184,
      "learning_rate": 0.00016997941990584795,
      "loss": 2.2394,
      "step": 148077
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.2635921239852905,
      "learning_rate": 0.00016997573350030073,
      "loss": 2.0139,
      "step": 148078
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.3723807334899902,
      "learning_rate": 0.00016997204711892763,
      "loss": 2.0512,
      "step": 148079
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.0748627185821533,
      "learning_rate": 0.00016996836076172903,
      "loss": 2.0794,
      "step": 148080
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.0472142696380615,
      "learning_rate": 0.0001699646744287059,
      "loss": 1.9969,
      "step": 148081
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.2351040840148926,
      "learning_rate": 0.00016996098811985865,
      "loss": 1.8469,
      "step": 148082
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.1018368005752563,
      "learning_rate": 0.00016995730183518837,
      "loss": 2.0399,
      "step": 148083
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.1268587112426758,
      "learning_rate": 0.00016995361557469523,
      "loss": 1.9285,
      "step": 148084
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.0391685962677002,
      "learning_rate": 0.0001699499293383803,
      "loss": 2.0798,
      "step": 148085
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.1334542036056519,
      "learning_rate": 0.00016994624312624402,
      "loss": 1.943,
      "step": 148086
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.1276532411575317,
      "learning_rate": 0.00016994255693828728,
      "loss": 2.0992,
      "step": 148087
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.0969542264938354,
      "learning_rate": 0.00016993887077451056,
      "loss": 2.3791,
      "step": 148088
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.0924324989318848,
      "learning_rate": 0.0001699351846349149,
      "loss": 2.1654,
      "step": 148089
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.0977959632873535,
      "learning_rate": 0.00016993149851950054,
      "loss": 2.1216,
      "step": 148090
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.1981772184371948,
      "learning_rate": 0.00016992781242826846,
      "loss": 2.1551,
      "step": 148091
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.104956030845642,
      "learning_rate": 0.00016992412636121913,
      "loss": 2.16,
      "step": 148092
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.1636736392974854,
      "learning_rate": 0.0001699204403183535,
      "loss": 2.22,
      "step": 148093
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.1796287298202515,
      "learning_rate": 0.00016991675429967198,
      "loss": 2.3815,
      "step": 148094
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.0823614597320557,
      "learning_rate": 0.00016991306830517553,
      "loss": 2.1739,
      "step": 148095
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.1633747816085815,
      "learning_rate": 0.00016990938233486468,
      "loss": 2.1469,
      "step": 148096
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.0540298223495483,
      "learning_rate": 0.0001699056963887401,
      "loss": 2.1035,
      "step": 148097
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.067699909210205,
      "learning_rate": 0.00016990201046680243,
      "loss": 1.782,
      "step": 148098
    },
    {
      "epoch": 1.93,
      "grad_norm": 0.998023271560669,
      "learning_rate": 0.00016989832456905255,
      "loss": 1.9493,
      "step": 148099
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.1066161394119263,
      "learning_rate": 0.00016989463869549091,
      "loss": 2.1059,
      "step": 148100
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.1239126920700073,
      "learning_rate": 0.0001698909528461184,
      "loss": 2.0444,
      "step": 148101
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.0656318664550781,
      "learning_rate": 0.00016988726702093566,
      "loss": 2.1418,
      "step": 148102
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.2645299434661865,
      "learning_rate": 0.0001698835812199433,
      "loss": 2.1401,
      "step": 148103
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.0937141180038452,
      "learning_rate": 0.00016987989544314192,
      "loss": 2.0449,
      "step": 148104
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.0905085802078247,
      "learning_rate": 0.00016987620969053245,
      "loss": 2.0971,
      "step": 148105
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.2491645812988281,
      "learning_rate": 0.00016987252396211534,
      "loss": 1.8875,
      "step": 148106
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.0215837955474854,
      "learning_rate": 0.00016986883825789148,
      "loss": 2.135,
      "step": 148107
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.1293326616287231,
      "learning_rate": 0.0001698651525778615,
      "loss": 2.1002,
      "step": 148108
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.0534451007843018,
      "learning_rate": 0.0001698614669220259,
      "loss": 1.7907,
      "step": 148109
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.107189416885376,
      "learning_rate": 0.00016985778129038564,
      "loss": 2.1869,
      "step": 148110
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.0054622888565063,
      "learning_rate": 0.00016985409568294127,
      "loss": 1.8849,
      "step": 148111
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.156646728515625,
      "learning_rate": 0.00016985041009969334,
      "loss": 2.1454,
      "step": 148112
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.0353000164031982,
      "learning_rate": 0.0001698467245406428,
      "loss": 2.1891,
      "step": 148113
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.0976970195770264,
      "learning_rate": 0.00016984303900579022,
      "loss": 1.9582,
      "step": 148114
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.0548704862594604,
      "learning_rate": 0.00016983935349513616,
      "loss": 2.0216,
      "step": 148115
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.2807821035385132,
      "learning_rate": 0.00016983566800868154,
      "loss": 2.2621,
      "step": 148116
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.1122498512268066,
      "learning_rate": 0.00016983198254642696,
      "loss": 2.2947,
      "step": 148117
    },
    {
      "epoch": 1.93,
      "grad_norm": 0.9525506496429443,
      "learning_rate": 0.00016982829710837289,
      "loss": 1.9766,
      "step": 148118
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.12222421169281,
      "learning_rate": 0.00016982461169452032,
      "loss": 2.0993,
      "step": 148119
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.001753568649292,
      "learning_rate": 0.00016982092630486988,
      "loss": 2.2177,
      "step": 148120
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.1340450048446655,
      "learning_rate": 0.00016981724093942198,
      "loss": 2.0962,
      "step": 148121
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.1841888427734375,
      "learning_rate": 0.00016981355559817768,
      "loss": 2.1224,
      "step": 148122
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.2725051641464233,
      "learning_rate": 0.00016980987028113742,
      "loss": 2.172,
      "step": 148123
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.1922053098678589,
      "learning_rate": 0.00016980618498830201,
      "loss": 2.1479,
      "step": 148124
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.0180777311325073,
      "learning_rate": 0.00016980249971967212,
      "loss": 2.1045,
      "step": 148125
    },
    {
      "epoch": 1.93,
      "grad_norm": 0.990947961807251,
      "learning_rate": 0.00016979881447524837,
      "loss": 1.9975,
      "step": 148126
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.2316986322402954,
      "learning_rate": 0.00016979512925503138,
      "loss": 2.0735,
      "step": 148127
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.1592601537704468,
      "learning_rate": 0.00016979144405902208,
      "loss": 2.1918,
      "step": 148128
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.1528089046478271,
      "learning_rate": 0.00016978775888722085,
      "loss": 2.2427,
      "step": 148129
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.1485345363616943,
      "learning_rate": 0.0001697840737396287,
      "loss": 2.2128,
      "step": 148130
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.5355408191680908,
      "learning_rate": 0.000169780388616246,
      "loss": 1.986,
      "step": 148131
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.05659818649292,
      "learning_rate": 0.00016977670351707382,
      "loss": 1.9903,
      "step": 148132
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.097685694694519,
      "learning_rate": 0.00016977301844211236,
      "loss": 2.154,
      "step": 148133
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.345525860786438,
      "learning_rate": 0.00016976933339136267,
      "loss": 2.1278,
      "step": 148134
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.2244681119918823,
      "learning_rate": 0.00016976564836482523,
      "loss": 2.0208,
      "step": 148135
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.0563627481460571,
      "learning_rate": 0.00016976196336250093,
      "loss": 2.0507,
      "step": 148136
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.0914874076843262,
      "learning_rate": 0.0001697582783843902,
      "loss": 2.0752,
      "step": 148137
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.1745206117630005,
      "learning_rate": 0.0001697545934304941,
      "loss": 2.046,
      "step": 148138
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.3009240627288818,
      "learning_rate": 0.00016975090850081286,
      "loss": 1.7922,
      "step": 148139
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.0224864482879639,
      "learning_rate": 0.0001697472235953475,
      "loss": 1.9354,
      "step": 148140
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.07981276512146,
      "learning_rate": 0.00016974353871409845,
      "loss": 1.936,
      "step": 148141
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.1587961912155151,
      "learning_rate": 0.0001697398538570667,
      "loss": 2.0684,
      "step": 148142
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.0270432233810425,
      "learning_rate": 0.0001697361690242526,
      "loss": 2.0536,
      "step": 148143
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.204103946685791,
      "learning_rate": 0.00016973248421565726,
      "loss": 2.0964,
      "step": 148144
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.095579743385315,
      "learning_rate": 0.0001697287994312809,
      "loss": 2.0352,
      "step": 148145
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.2488629817962646,
      "learning_rate": 0.0001697251146711245,
      "loss": 2.3168,
      "step": 148146
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.11295485496521,
      "learning_rate": 0.00016972142993518856,
      "loss": 1.9986,
      "step": 148147
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.0931421518325806,
      "learning_rate": 0.000169717745223474,
      "loss": 2.0866,
      "step": 148148
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.1128196716308594,
      "learning_rate": 0.0001697140605359812,
      "loss": 2.0144,
      "step": 148149
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.3243076801300049,
      "learning_rate": 0.0001697103758727113,
      "loss": 1.9093,
      "step": 148150
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.0949937105178833,
      "learning_rate": 0.00016970669123366443,
      "loss": 2.0822,
      "step": 148151
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.065545678138733,
      "learning_rate": 0.0001697030066188417,
      "loss": 2.3394,
      "step": 148152
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.0734543800354004,
      "learning_rate": 0.00016969932202824347,
      "loss": 1.8946,
      "step": 148153
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.0882240533828735,
      "learning_rate": 0.00016969563746187076,
      "loss": 1.9059,
      "step": 148154
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.080475091934204,
      "learning_rate": 0.00016969195291972395,
      "loss": 2.2005,
      "step": 148155
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.3256511688232422,
      "learning_rate": 0.00016968826840180412,
      "loss": 1.9459,
      "step": 148156
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.0392842292785645,
      "learning_rate": 0.00016968458390811145,
      "loss": 2.2146,
      "step": 148157
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.2020519971847534,
      "learning_rate": 0.000169680899438647,
      "loss": 2.0956,
      "step": 148158
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.1017800569534302,
      "learning_rate": 0.0001696772149934112,
      "loss": 2.1585,
      "step": 148159
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.1268773078918457,
      "learning_rate": 0.000169673530572405,
      "loss": 2.1179,
      "step": 148160
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.0902477502822876,
      "learning_rate": 0.00016966984617562884,
      "loss": 1.8982,
      "step": 148161
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.0230170488357544,
      "learning_rate": 0.00016966616180308372,
      "loss": 1.9191,
      "step": 148162
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.0522773265838623,
      "learning_rate": 0.00016966247745476992,
      "loss": 1.8559,
      "step": 148163
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.079626441001892,
      "learning_rate": 0.00016965879313068843,
      "loss": 1.9203,
      "step": 148164
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.1561505794525146,
      "learning_rate": 0.0001696551088308397,
      "loss": 2.207,
      "step": 148165
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.2235431671142578,
      "learning_rate": 0.00016965142455522468,
      "loss": 2.161,
      "step": 148166
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.052064299583435,
      "learning_rate": 0.00016964774030384382,
      "loss": 1.8998,
      "step": 148167
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.1063978672027588,
      "learning_rate": 0.00016964405607669813,
      "loss": 2.1748,
      "step": 148168
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.095811367034912,
      "learning_rate": 0.0001696403718737878,
      "loss": 2.2908,
      "step": 148169
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.0659799575805664,
      "learning_rate": 0.0001696366876951139,
      "loss": 2.0466,
      "step": 148170
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.0500301122665405,
      "learning_rate": 0.00016963300354067693,
      "loss": 1.9397,
      "step": 148171
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.144150972366333,
      "learning_rate": 0.00016962931941047775,
      "loss": 1.9786,
      "step": 148172
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.2094320058822632,
      "learning_rate": 0.00016962563530451683,
      "loss": 1.921,
      "step": 148173
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.0495171546936035,
      "learning_rate": 0.00016962195122279517,
      "loss": 1.9021,
      "step": 148174
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.0413954257965088,
      "learning_rate": 0.00016961826716531303,
      "loss": 2.0508,
      "step": 148175
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.2254321575164795,
      "learning_rate": 0.00016961458313207144,
      "loss": 2.1704,
      "step": 148176
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.0939384698867798,
      "learning_rate": 0.0001696108991230708,
      "loss": 1.8991,
      "step": 148177
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.0167322158813477,
      "learning_rate": 0.00016960721513831217,
      "loss": 2.1117,
      "step": 148178
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.0574575662612915,
      "learning_rate": 0.00016960353117779584,
      "loss": 2.1747,
      "step": 148179
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.095048427581787,
      "learning_rate": 0.00016959984724152288,
      "loss": 1.9126,
      "step": 148180
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.1533571481704712,
      "learning_rate": 0.0001695961633294936,
      "loss": 2.0212,
      "step": 148181
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.1029924154281616,
      "learning_rate": 0.0001695924794417089,
      "loss": 2.0829,
      "step": 148182
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.140714406967163,
      "learning_rate": 0.00016958879557816935,
      "loss": 2.1823,
      "step": 148183
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.224172592163086,
      "learning_rate": 0.0001695851117388758,
      "loss": 2.2367,
      "step": 148184
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.1107008457183838,
      "learning_rate": 0.00016958142792382875,
      "loss": 2.0888,
      "step": 148185
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.0952885150909424,
      "learning_rate": 0.00016957774413302906,
      "loss": 2.0893,
      "step": 148186
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.1555441617965698,
      "learning_rate": 0.00016957406036647736,
      "loss": 1.9262,
      "step": 148187
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.1509672403335571,
      "learning_rate": 0.0001695703766241743,
      "loss": 2.1697,
      "step": 148188
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.0669304132461548,
      "learning_rate": 0.00016956669290612046,
      "loss": 2.0153,
      "step": 148189
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.2728849649429321,
      "learning_rate": 0.00016956300921231672,
      "loss": 1.954,
      "step": 148190
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.0678727626800537,
      "learning_rate": 0.00016955932554276357,
      "loss": 2.271,
      "step": 148191
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.1450189352035522,
      "learning_rate": 0.00016955564189746194,
      "loss": 2.196,
      "step": 148192
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.0518100261688232,
      "learning_rate": 0.00016955195827641237,
      "loss": 2.1224,
      "step": 148193
    },
    {
      "epoch": 1.93,
      "grad_norm": 0.9937590956687927,
      "learning_rate": 0.0001695482746796154,
      "loss": 2.1232,
      "step": 148194
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.1971584558486938,
      "learning_rate": 0.00016954459110707205,
      "loss": 2.2006,
      "step": 148195
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.0552024841308594,
      "learning_rate": 0.0001695409075587828,
      "loss": 1.9639,
      "step": 148196
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.0502817630767822,
      "learning_rate": 0.00016953722403474823,
      "loss": 2.0305,
      "step": 148197
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.1110037565231323,
      "learning_rate": 0.00016953354053496925,
      "loss": 1.8583,
      "step": 148198
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.0762826204299927,
      "learning_rate": 0.00016952985705944647,
      "loss": 2.0705,
      "step": 148199
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.061356544494629,
      "learning_rate": 0.00016952617360818043,
      "loss": 2.0773,
      "step": 148200
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.0467132329940796,
      "learning_rate": 0.00016952249018117207,
      "loss": 2.3117,
      "step": 148201
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.1492072343826294,
      "learning_rate": 0.0001695188067784219,
      "loss": 1.9416,
      "step": 148202
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.0464673042297363,
      "learning_rate": 0.00016951512339993057,
      "loss": 2.2801,
      "step": 148203
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.0936411619186401,
      "learning_rate": 0.00016951144004569895,
      "loss": 2.1319,
      "step": 148204
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.086523175239563,
      "learning_rate": 0.00016950775671572762,
      "loss": 2.2594,
      "step": 148205
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.0875651836395264,
      "learning_rate": 0.00016950407341001711,
      "loss": 2.2127,
      "step": 148206
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.0229941606521606,
      "learning_rate": 0.00016950039012856842,
      "loss": 1.979,
      "step": 148207
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.107985019683838,
      "learning_rate": 0.00016949670687138198,
      "loss": 2.197,
      "step": 148208
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.1413013935089111,
      "learning_rate": 0.00016949302363845862,
      "loss": 2.0907,
      "step": 148209
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.0647621154785156,
      "learning_rate": 0.00016948934042979898,
      "loss": 2.2096,
      "step": 148210
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.0468415021896362,
      "learning_rate": 0.00016948565724540373,
      "loss": 2.1918,
      "step": 148211
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.1408747434616089,
      "learning_rate": 0.0001694819740852735,
      "loss": 2.0544,
      "step": 148212
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.0970587730407715,
      "learning_rate": 0.00016947829094940905,
      "loss": 1.8616,
      "step": 148213
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.1494773626327515,
      "learning_rate": 0.000169474607837811,
      "loss": 2.0656,
      "step": 148214
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.1158002614974976,
      "learning_rate": 0.00016947092475048023,
      "loss": 2.1202,
      "step": 148215
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.1076213121414185,
      "learning_rate": 0.0001694672416874171,
      "loss": 2.065,
      "step": 148216
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.0881820917129517,
      "learning_rate": 0.00016946355864862276,
      "loss": 2.0232,
      "step": 148217
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.1347123384475708,
      "learning_rate": 0.00016945987563409733,
      "loss": 1.9592,
      "step": 148218
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.1009972095489502,
      "learning_rate": 0.0001694561926438419,
      "loss": 2.2966,
      "step": 148219
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.2070581912994385,
      "learning_rate": 0.00016945250967785692,
      "loss": 2.1687,
      "step": 148220
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.0392742156982422,
      "learning_rate": 0.0001694488267361433,
      "loss": 1.98,
      "step": 148221
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.0767419338226318,
      "learning_rate": 0.0001694451438187015,
      "loss": 2.0043,
      "step": 148222
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.119938611984253,
      "learning_rate": 0.00016944146092553252,
      "loss": 2.2764,
      "step": 148223
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.1318992376327515,
      "learning_rate": 0.0001694377780566366,
      "loss": 2.3946,
      "step": 148224
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.1090328693389893,
      "learning_rate": 0.0001694340952120148,
      "loss": 1.9375,
      "step": 148225
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.1254205703735352,
      "learning_rate": 0.00016943041239166756,
      "loss": 1.991,
      "step": 148226
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.0886791944503784,
      "learning_rate": 0.00016942672959559574,
      "loss": 2.3095,
      "step": 148227
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.0873730182647705,
      "learning_rate": 0.0001694230468237999,
      "loss": 2.2203,
      "step": 148228
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.0816435813903809,
      "learning_rate": 0.000169419364076281,
      "loss": 2.3508,
      "step": 148229
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.1971994638442993,
      "learning_rate": 0.00016941568135303927,
      "loss": 2.0705,
      "step": 148230
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.1453555822372437,
      "learning_rate": 0.00016941199865407568,
      "loss": 2.125,
      "step": 148231
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.0963224172592163,
      "learning_rate": 0.0001694083159793908,
      "loss": 1.9905,
      "step": 148232
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.0510811805725098,
      "learning_rate": 0.00016940463332898553,
      "loss": 2.1311,
      "step": 148233
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.1823843717575073,
      "learning_rate": 0.00016940095070286024,
      "loss": 1.9703,
      "step": 148234
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.184369683265686,
      "learning_rate": 0.00016939726810101606,
      "loss": 2.2069,
      "step": 148235
    },
    {
      "epoch": 1.93,
      "grad_norm": 0.9851866960525513,
      "learning_rate": 0.00016939358552345312,
      "loss": 1.8582,
      "step": 148236
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.0451476573944092,
      "learning_rate": 0.00016938990297017247,
      "loss": 1.964,
      "step": 148237
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.4684041738510132,
      "learning_rate": 0.00016938622044117464,
      "loss": 2.1701,
      "step": 148238
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.135756254196167,
      "learning_rate": 0.00016938253793646047,
      "loss": 2.0,
      "step": 148239
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.2043757438659668,
      "learning_rate": 0.00016937885545603045,
      "loss": 2.1973,
      "step": 148240
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.0553228855133057,
      "learning_rate": 0.00016937517299988556,
      "loss": 2.0104,
      "step": 148241
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.2688281536102295,
      "learning_rate": 0.0001693714905680261,
      "loss": 2.1269,
      "step": 148242
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.1094144582748413,
      "learning_rate": 0.00016936780816045305,
      "loss": 2.1689,
      "step": 148243
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.0342955589294434,
      "learning_rate": 0.00016936412577716687,
      "loss": 2.0396,
      "step": 148244
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.2080469131469727,
      "learning_rate": 0.00016936044341816847,
      "loss": 2.3457,
      "step": 148245
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.149563193321228,
      "learning_rate": 0.0001693567610834583,
      "loss": 2.033,
      "step": 148246
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.1491559743881226,
      "learning_rate": 0.00016935307877303743,
      "loss": 2.2699,
      "step": 148247
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.168501615524292,
      "learning_rate": 0.00016934939648690609,
      "loss": 2.0395,
      "step": 148248
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.1961919069290161,
      "learning_rate": 0.00016934571422506524,
      "loss": 2.3432,
      "step": 148249
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.116320252418518,
      "learning_rate": 0.00016934203198751537,
      "loss": 2.165,
      "step": 148250
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.127325177192688,
      "learning_rate": 0.0001693383497742574,
      "loss": 2.069,
      "step": 148251
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.0835647583007812,
      "learning_rate": 0.0001693346675852918,
      "loss": 2.0042,
      "step": 148252
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.1147702932357788,
      "learning_rate": 0.00016933098542061957,
      "loss": 2.2226,
      "step": 148253
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.2091782093048096,
      "learning_rate": 0.00016932730328024097,
      "loss": 2.4017,
      "step": 148254
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.2404613494873047,
      "learning_rate": 0.00016932362116415698,
      "loss": 1.9211,
      "step": 148255
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.1682056188583374,
      "learning_rate": 0.00016931993907236804,
      "loss": 2.1129,
      "step": 148256
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.0933722257614136,
      "learning_rate": 0.00016931625700487518,
      "loss": 1.8361,
      "step": 148257
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.0786536931991577,
      "learning_rate": 0.00016931257496167874,
      "loss": 1.9789,
      "step": 148258
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.236550211906433,
      "learning_rate": 0.00016930889294277982,
      "loss": 1.944,
      "step": 148259
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.101744532585144,
      "learning_rate": 0.0001693052109481786,
      "loss": 1.8738,
      "step": 148260
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.1089001893997192,
      "learning_rate": 0.0001693015289778761,
      "loss": 2.0609,
      "step": 148261
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.1981391906738281,
      "learning_rate": 0.0001692978470318728,
      "loss": 2.18,
      "step": 148262
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.0728095769882202,
      "learning_rate": 0.00016929416511016963,
      "loss": 2.1574,
      "step": 148263
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.327125072479248,
      "learning_rate": 0.00016929048321276703,
      "loss": 1.9955,
      "step": 148264
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.38507080078125,
      "learning_rate": 0.0001692868013396659,
      "loss": 1.9362,
      "step": 148265
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.197636365890503,
      "learning_rate": 0.00016928311949086683,
      "loss": 2.0326,
      "step": 148266
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.0333243608474731,
      "learning_rate": 0.00016927943766637047,
      "loss": 1.9278,
      "step": 148267
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.3097102642059326,
      "learning_rate": 0.00016927575586617744,
      "loss": 2.1955,
      "step": 148268
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.1679165363311768,
      "learning_rate": 0.00016927207409028863,
      "loss": 2.0639,
      "step": 148269
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.0879000425338745,
      "learning_rate": 0.0001692683923387045,
      "loss": 2.04,
      "step": 148270
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.1041934490203857,
      "learning_rate": 0.00016926471061142593,
      "loss": 2.25,
      "step": 148271
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.150991439819336,
      "learning_rate": 0.00016926102890845357,
      "loss": 2.0134,
      "step": 148272
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.363764762878418,
      "learning_rate": 0.000169257347229788,
      "loss": 2.2261,
      "step": 148273
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.196974515914917,
      "learning_rate": 0.00016925366557542985,
      "loss": 2.012,
      "step": 148274
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.0635894536972046,
      "learning_rate": 0.00016924998394538004,
      "loss": 1.9289,
      "step": 148275
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.1451255083084106,
      "learning_rate": 0.00016924630233963904,
      "loss": 2.0553,
      "step": 148276
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.1537694931030273,
      "learning_rate": 0.00016924262075820767,
      "loss": 2.1061,
      "step": 148277
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.2268928289413452,
      "learning_rate": 0.0001692389392010866,
      "loss": 1.987,
      "step": 148278
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.255719542503357,
      "learning_rate": 0.00016923525766827647,
      "loss": 2.0497,
      "step": 148279
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.1674917936325073,
      "learning_rate": 0.0001692315761597779,
      "loss": 2.0925,
      "step": 148280
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.064774751663208,
      "learning_rate": 0.00016922789467559173,
      "loss": 1.8684,
      "step": 148281
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.3496661186218262,
      "learning_rate": 0.00016922421321571843,
      "loss": 2.0241,
      "step": 148282
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.2098512649536133,
      "learning_rate": 0.00016922053178015898,
      "loss": 2.0281,
      "step": 148283
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.2076380252838135,
      "learning_rate": 0.00016921685036891385,
      "loss": 2.0388,
      "step": 148284
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.2120295763015747,
      "learning_rate": 0.0001692131689819837,
      "loss": 2.1129,
      "step": 148285
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.3930999040603638,
      "learning_rate": 0.00016920948761936933,
      "loss": 2.1562,
      "step": 148286
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.0669194459915161,
      "learning_rate": 0.00016920580628107147,
      "loss": 2.0672,
      "step": 148287
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.1172572374343872,
      "learning_rate": 0.0001692021249670906,
      "loss": 2.2565,
      "step": 148288
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.0751100778579712,
      "learning_rate": 0.00016919844367742758,
      "loss": 2.0207,
      "step": 148289
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.17548406124115,
      "learning_rate": 0.0001691947624120831,
      "loss": 2.2942,
      "step": 148290
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.2232208251953125,
      "learning_rate": 0.00016919108117105765,
      "loss": 2.1006,
      "step": 148291
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.1674641370773315,
      "learning_rate": 0.00016918739995435217,
      "loss": 1.875,
      "step": 148292
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.1590251922607422,
      "learning_rate": 0.0001691837187619671,
      "loss": 2.0199,
      "step": 148293
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.2802194356918335,
      "learning_rate": 0.00016918003759390334,
      "loss": 2.0935,
      "step": 148294
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.3741565942764282,
      "learning_rate": 0.0001691763564501615,
      "loss": 2.0912,
      "step": 148295
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.1227951049804688,
      "learning_rate": 0.00016917267533074221,
      "loss": 1.9359,
      "step": 148296
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.1436384916305542,
      "learning_rate": 0.00016916899423564613,
      "loss": 2.1217,
      "step": 148297
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.2994446754455566,
      "learning_rate": 0.00016916531316487406,
      "loss": 2.1851,
      "step": 148298
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.3109627962112427,
      "learning_rate": 0.00016916163211842656,
      "loss": 2.0152,
      "step": 148299
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.3418512344360352,
      "learning_rate": 0.0001691579510963045,
      "loss": 2.0234,
      "step": 148300
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.0938116312026978,
      "learning_rate": 0.00016915427009850842,
      "loss": 2.0789,
      "step": 148301
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.2204476594924927,
      "learning_rate": 0.00016915058912503903,
      "loss": 2.2692,
      "step": 148302
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.281124472618103,
      "learning_rate": 0.00016914690817589694,
      "loss": 2.1578,
      "step": 148303
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.0706263780593872,
      "learning_rate": 0.00016914322725108299,
      "loss": 2.0093,
      "step": 148304
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.0162320137023926,
      "learning_rate": 0.00016913954635059767,
      "loss": 2.1449,
      "step": 148305
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.196730613708496,
      "learning_rate": 0.0001691358654744419,
      "loss": 2.1511,
      "step": 148306
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.183271050453186,
      "learning_rate": 0.0001691321846226161,
      "loss": 2.254,
      "step": 148307
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.079194188117981,
      "learning_rate": 0.0001691285037951214,
      "loss": 1.997,
      "step": 148308
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.125959038734436,
      "learning_rate": 0.00016912482299195788,
      "loss": 2.1559,
      "step": 148309
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.0309771299362183,
      "learning_rate": 0.0001691211422131267,
      "loss": 2.097,
      "step": 148310
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.1145411729812622,
      "learning_rate": 0.00016911746145862816,
      "loss": 2.3014,
      "step": 148311
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.1230782270431519,
      "learning_rate": 0.00016911378072846338,
      "loss": 1.8804,
      "step": 148312
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.0027915239334106,
      "learning_rate": 0.00016911010002263266,
      "loss": 1.9598,
      "step": 148313
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.0476876497268677,
      "learning_rate": 0.00016910641934113707,
      "loss": 2.1507,
      "step": 148314
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.1025508642196655,
      "learning_rate": 0.0001691027386839768,
      "loss": 2.0535,
      "step": 148315
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.3027268648147583,
      "learning_rate": 0.00016909905805115297,
      "loss": 1.9565,
      "step": 148316
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.0685725212097168,
      "learning_rate": 0.00016909537744266598,
      "loss": 1.8666,
      "step": 148317
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.026160717010498,
      "learning_rate": 0.0001690916968585167,
      "loss": 2.0329,
      "step": 148318
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.1961668729782104,
      "learning_rate": 0.00016908801629870565,
      "loss": 1.9296,
      "step": 148319
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.3303598165512085,
      "learning_rate": 0.00016908433576323385,
      "loss": 2.0491,
      "step": 148320
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.217625379562378,
      "learning_rate": 0.00016908065525210152,
      "loss": 1.9607,
      "step": 148321
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.1415939331054688,
      "learning_rate": 0.00016907697476530967,
      "loss": 1.8543,
      "step": 148322
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.0896306037902832,
      "learning_rate": 0.00016907329430285877,
      "loss": 2.1638,
      "step": 148323
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.180925965309143,
      "learning_rate": 0.0001690696138647497,
      "loss": 2.2147,
      "step": 148324
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.0844366550445557,
      "learning_rate": 0.00016906593345098302,
      "loss": 2.1941,
      "step": 148325
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.1715960502624512,
      "learning_rate": 0.00016906225306155965,
      "loss": 1.9991,
      "step": 148326
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.1583123207092285,
      "learning_rate": 0.00016905857269647982,
      "loss": 1.9164,
      "step": 148327
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.091442346572876,
      "learning_rate": 0.0001690548923557446,
      "loss": 2.0357,
      "step": 148328
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.155197024345398,
      "learning_rate": 0.00016905121203935442,
      "loss": 2.0095,
      "step": 148329
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.1010451316833496,
      "learning_rate": 0.00016904753174731024,
      "loss": 2.1688,
      "step": 148330
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.0693576335906982,
      "learning_rate": 0.00016904385147961245,
      "loss": 2.3892,
      "step": 148331
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.0756454467773438,
      "learning_rate": 0.00016904017123626206,
      "loss": 2.0342,
      "step": 148332
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.1092731952667236,
      "learning_rate": 0.00016903649101725953,
      "loss": 1.9201,
      "step": 148333
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.0430928468704224,
      "learning_rate": 0.00016903281082260557,
      "loss": 2.045,
      "step": 148334
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.094104528427124,
      "learning_rate": 0.00016902913065230078,
      "loss": 2.1655,
      "step": 148335
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.2017534971237183,
      "learning_rate": 0.00016902545050634604,
      "loss": 2.1018,
      "step": 148336
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.2534000873565674,
      "learning_rate": 0.00016902177038474186,
      "loss": 2.211,
      "step": 148337
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.1627510786056519,
      "learning_rate": 0.0001690180902874891,
      "loss": 1.9687,
      "step": 148338
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.251133680343628,
      "learning_rate": 0.0001690144102145884,
      "loss": 2.0733,
      "step": 148339
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.1167101860046387,
      "learning_rate": 0.00016901073016604032,
      "loss": 2.0214,
      "step": 148340
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.0380120277404785,
      "learning_rate": 0.00016900705014184555,
      "loss": 2.1638,
      "step": 148341
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.1983705759048462,
      "learning_rate": 0.00016900337014200494,
      "loss": 1.8929,
      "step": 148342
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.0577565431594849,
      "learning_rate": 0.00016899969016651896,
      "loss": 2.0148,
      "step": 148343
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.3280210494995117,
      "learning_rate": 0.00016899601021538855,
      "loss": 2.0698,
      "step": 148344
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.3620719909667969,
      "learning_rate": 0.00016899233028861422,
      "loss": 1.9989,
      "step": 148345
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.0621439218521118,
      "learning_rate": 0.0001689886503861967,
      "loss": 1.9601,
      "step": 148346
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.0748156309127808,
      "learning_rate": 0.00016898497050813656,
      "loss": 2.0144,
      "step": 148347
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.3829395771026611,
      "learning_rate": 0.0001689812906544347,
      "loss": 2.0263,
      "step": 148348
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.2862651348114014,
      "learning_rate": 0.00016897761082509158,
      "loss": 2.0475,
      "step": 148349
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.4090282917022705,
      "learning_rate": 0.00016897393102010806,
      "loss": 2.1718,
      "step": 148350
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.1503530740737915,
      "learning_rate": 0.0001689702512394848,
      "loss": 2.2957,
      "step": 148351
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.3832879066467285,
      "learning_rate": 0.00016896657148322247,
      "loss": 2.3119,
      "step": 148352
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.0321840047836304,
      "learning_rate": 0.00016896289175132156,
      "loss": 1.9481,
      "step": 148353
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.1295342445373535,
      "learning_rate": 0.00016895921204378308,
      "loss": 2.0077,
      "step": 148354
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.1119462251663208,
      "learning_rate": 0.00016895553236060743,
      "loss": 2.2026,
      "step": 148355
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.0664231777191162,
      "learning_rate": 0.00016895185270179554,
      "loss": 1.8855,
      "step": 148356
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.0757066011428833,
      "learning_rate": 0.00016894817306734792,
      "loss": 1.9157,
      "step": 148357
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.0446759462356567,
      "learning_rate": 0.00016894449345726537,
      "loss": 2.1,
      "step": 148358
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.0433335304260254,
      "learning_rate": 0.00016894081387154838,
      "loss": 2.0601,
      "step": 148359
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.519233226776123,
      "learning_rate": 0.00016893713431019792,
      "loss": 2.2683,
      "step": 148360
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.1143983602523804,
      "learning_rate": 0.00016893345477321433,
      "loss": 2.3328,
      "step": 148361
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.173683762550354,
      "learning_rate": 0.00016892977526059865,
      "loss": 2.0207,
      "step": 148362
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.1858646869659424,
      "learning_rate": 0.00016892609577235138,
      "loss": 2.0743,
      "step": 148363
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.09697425365448,
      "learning_rate": 0.00016892241630847322,
      "loss": 2.1591,
      "step": 148364
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.2244477272033691,
      "learning_rate": 0.00016891873686896477,
      "loss": 2.1972,
      "step": 148365
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.0798320770263672,
      "learning_rate": 0.00016891505745382688,
      "loss": 2.2918,
      "step": 148366
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.2434707880020142,
      "learning_rate": 0.00016891137806306007,
      "loss": 2.1488,
      "step": 148367
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.182410478591919,
      "learning_rate": 0.0001689076986966652,
      "loss": 2.1161,
      "step": 148368
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.121995449066162,
      "learning_rate": 0.00016890401935464288,
      "loss": 1.872,
      "step": 148369
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.0080151557922363,
      "learning_rate": 0.00016890034003699367,
      "loss": 2.3659,
      "step": 148370
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.2450584173202515,
      "learning_rate": 0.00016889666074371845,
      "loss": 2.3362,
      "step": 148371
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.0912243127822876,
      "learning_rate": 0.00016889298147481784,
      "loss": 2.1335,
      "step": 148372
    },
    {
      "epoch": 1.93,
      "grad_norm": 0.9961310625076294,
      "learning_rate": 0.00016888930223029236,
      "loss": 1.826,
      "step": 148373
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.0194827318191528,
      "learning_rate": 0.00016888562301014295,
      "loss": 1.8382,
      "step": 148374
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.0875675678253174,
      "learning_rate": 0.0001688819438143702,
      "loss": 2.2825,
      "step": 148375
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.2921775579452515,
      "learning_rate": 0.00016887826464297465,
      "loss": 2.1084,
      "step": 148376
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.1087841987609863,
      "learning_rate": 0.00016887458549595723,
      "loss": 1.9967,
      "step": 148377
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.109948992729187,
      "learning_rate": 0.00016887090637331836,
      "loss": 2.0297,
      "step": 148378
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.1849536895751953,
      "learning_rate": 0.000168867227275059,
      "loss": 2.2499,
      "step": 148379
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.1244637966156006,
      "learning_rate": 0.00016886354820117968,
      "loss": 1.9451,
      "step": 148380
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.1978687047958374,
      "learning_rate": 0.00016885986915168112,
      "loss": 2.1007,
      "step": 148381
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.1219043731689453,
      "learning_rate": 0.00016885619012656385,
      "loss": 2.121,
      "step": 148382
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.0729345083236694,
      "learning_rate": 0.00016885251112582882,
      "loss": 2.4021,
      "step": 148383
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.1812175512313843,
      "learning_rate": 0.00016884883214947648,
      "loss": 2.2393,
      "step": 148384
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.072546124458313,
      "learning_rate": 0.0001688451531975077,
      "loss": 2.0226,
      "step": 148385
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.1129977703094482,
      "learning_rate": 0.00016884147426992312,
      "loss": 1.9415,
      "step": 148386
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.1364076137542725,
      "learning_rate": 0.00016883779536672338,
      "loss": 2.2796,
      "step": 148387
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.1283544301986694,
      "learning_rate": 0.00016883411648790898,
      "loss": 2.0549,
      "step": 148388
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.1326783895492554,
      "learning_rate": 0.000168830437633481,
      "loss": 1.9685,
      "step": 148389
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.197460412979126,
      "learning_rate": 0.00016882675880343977,
      "loss": 2.0913,
      "step": 148390
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.099316954612732,
      "learning_rate": 0.00016882307999778625,
      "loss": 2.1252,
      "step": 148391
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.2472645044326782,
      "learning_rate": 0.00016881940121652085,
      "loss": 2.0878,
      "step": 148392
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.0421561002731323,
      "learning_rate": 0.00016881572245964466,
      "loss": 2.1794,
      "step": 148393
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.1906300783157349,
      "learning_rate": 0.00016881204372715786,
      "loss": 2.1833,
      "step": 148394
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.2447755336761475,
      "learning_rate": 0.00016880836501906147,
      "loss": 2.1338,
      "step": 148395
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.1831477880477905,
      "learning_rate": 0.00016880468633535603,
      "loss": 2.1969,
      "step": 148396
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.2328779697418213,
      "learning_rate": 0.0001688010076760423,
      "loss": 2.2144,
      "step": 148397
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.1747474670410156,
      "learning_rate": 0.00016879732904112092,
      "loss": 2.1829,
      "step": 148398
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.1734044551849365,
      "learning_rate": 0.00016879365043059264,
      "loss": 2.1994,
      "step": 148399
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.1811280250549316,
      "learning_rate": 0.00016878997184445815,
      "loss": 2.0418,
      "step": 148400
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.141349196434021,
      "learning_rate": 0.00016878629328271802,
      "loss": 1.9915,
      "step": 148401
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.2873563766479492,
      "learning_rate": 0.00016878261474537293,
      "loss": 2.0264,
      "step": 148402
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.1123828887939453,
      "learning_rate": 0.00016877893623242372,
      "loss": 1.8776,
      "step": 148403
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.3978044986724854,
      "learning_rate": 0.00016877525774387087,
      "loss": 2.1925,
      "step": 148404
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.1766186952590942,
      "learning_rate": 0.00016877157927971527,
      "loss": 1.9579,
      "step": 148405
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.1706949472427368,
      "learning_rate": 0.00016876790083995757,
      "loss": 2.1651,
      "step": 148406
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.1330091953277588,
      "learning_rate": 0.00016876422242459834,
      "loss": 2.0973,
      "step": 148407
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.1870112419128418,
      "learning_rate": 0.0001687605440336382,
      "loss": 1.9611,
      "step": 148408
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.1878621578216553,
      "learning_rate": 0.0001687568656670781,
      "loss": 2.0515,
      "step": 148409
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.0769991874694824,
      "learning_rate": 0.00016875318732491844,
      "loss": 2.0883,
      "step": 148410
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.1161319017410278,
      "learning_rate": 0.0001687495090071602,
      "loss": 2.0714,
      "step": 148411
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.2204567193984985,
      "learning_rate": 0.00016874583071380386,
      "loss": 2.2269,
      "step": 148412
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.068568229675293,
      "learning_rate": 0.00016874215244485016,
      "loss": 2.1068,
      "step": 148413
    },
    {
      "epoch": 1.93,
      "grad_norm": 0.9992848038673401,
      "learning_rate": 0.00016873847420029967,
      "loss": 1.777,
      "step": 148414
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.0792533159255981,
      "learning_rate": 0.00016873479598015326,
      "loss": 2.2234,
      "step": 148415
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.5006705522537231,
      "learning_rate": 0.0001687311177844114,
      "loss": 2.0582,
      "step": 148416
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.0317137241363525,
      "learning_rate": 0.00016872743961307508,
      "loss": 2.2071,
      "step": 148417
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.1329854726791382,
      "learning_rate": 0.00016872376146614477,
      "loss": 2.1246,
      "step": 148418
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.2121299505233765,
      "learning_rate": 0.0001687200833436212,
      "loss": 2.364,
      "step": 148419
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.121244192123413,
      "learning_rate": 0.00016871640524550492,
      "loss": 1.8801,
      "step": 148420
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.096989393234253,
      "learning_rate": 0.00016871272717179685,
      "loss": 2.1685,
      "step": 148421
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.1326130628585815,
      "learning_rate": 0.00016870904912249745,
      "loss": 2.0583,
      "step": 148422
    },
    {
      "epoch": 1.93,
      "grad_norm": 2.363556385040283,
      "learning_rate": 0.00016870537109760767,
      "loss": 1.9197,
      "step": 148423
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.1010165214538574,
      "learning_rate": 0.000168701693097128,
      "loss": 2.054,
      "step": 148424
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.0364396572113037,
      "learning_rate": 0.0001686980151210592,
      "loss": 2.1623,
      "step": 148425
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.172601342201233,
      "learning_rate": 0.00016869433716940174,
      "loss": 2.0127,
      "step": 148426
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.0389636754989624,
      "learning_rate": 0.00016869065924215662,
      "loss": 2.1021,
      "step": 148427
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.1458289623260498,
      "learning_rate": 0.00016868698133932426,
      "loss": 2.0518,
      "step": 148428
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.105421543121338,
      "learning_rate": 0.0001686833034609056,
      "loss": 2.0023,
      "step": 148429
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.3255528211593628,
      "learning_rate": 0.0001686796256069012,
      "loss": 1.9856,
      "step": 148430
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.1582648754119873,
      "learning_rate": 0.00016867594777731173,
      "loss": 2.0276,
      "step": 148431
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.1816493272781372,
      "learning_rate": 0.0001686722699721378,
      "loss": 2.228,
      "step": 148432
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.145655632019043,
      "learning_rate": 0.00016866859219138023,
      "loss": 2.2273,
      "step": 148433
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.1333168745040894,
      "learning_rate": 0.00016866491443503958,
      "loss": 1.9812,
      "step": 148434
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.2305134534835815,
      "learning_rate": 0.00016866123670311668,
      "loss": 2.2467,
      "step": 148435
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.2679721117019653,
      "learning_rate": 0.00016865755899561218,
      "loss": 2.2093,
      "step": 148436
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.1200599670410156,
      "learning_rate": 0.00016865388131252665,
      "loss": 2.2354,
      "step": 148437
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.417310118675232,
      "learning_rate": 0.00016865020365386076,
      "loss": 2.1808,
      "step": 148438
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.3980787992477417,
      "learning_rate": 0.00016864652601961542,
      "loss": 1.8523,
      "step": 148439
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.1095423698425293,
      "learning_rate": 0.00016864284840979103,
      "loss": 1.8764,
      "step": 148440
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.1968294382095337,
      "learning_rate": 0.0001686391708243885,
      "loss": 2.2674,
      "step": 148441
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.165407419204712,
      "learning_rate": 0.00016863549326340849,
      "loss": 2.2654,
      "step": 148442
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.384803056716919,
      "learning_rate": 0.0001686318157268516,
      "loss": 1.8672,
      "step": 148443
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.240482211112976,
      "learning_rate": 0.00016862813821471838,
      "loss": 2.2663,
      "step": 148444
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.0951114892959595,
      "learning_rate": 0.00016862446072700982,
      "loss": 1.9599,
      "step": 148445
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.7239539623260498,
      "learning_rate": 0.00016862078326372633,
      "loss": 1.9624,
      "step": 148446
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.2471694946289062,
      "learning_rate": 0.00016861710582486884,
      "loss": 2.0694,
      "step": 148447
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.2143712043762207,
      "learning_rate": 0.00016861342841043795,
      "loss": 2.0176,
      "step": 148448
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.0740433931350708,
      "learning_rate": 0.00016860975102043425,
      "loss": 2.1286,
      "step": 148449
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.5945165157318115,
      "learning_rate": 0.00016860607365485835,
      "loss": 2.1116,
      "step": 148450
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.2680283784866333,
      "learning_rate": 0.0001686023963137112,
      "loss": 2.0535,
      "step": 148451
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.177627682685852,
      "learning_rate": 0.00016859871899699324,
      "loss": 2.0685,
      "step": 148452
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.0566399097442627,
      "learning_rate": 0.00016859504170470537,
      "loss": 2.1986,
      "step": 148453
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.0222864151000977,
      "learning_rate": 0.00016859136443684813,
      "loss": 2.1795,
      "step": 148454
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.2616856098175049,
      "learning_rate": 0.00016858768719342214,
      "loss": 1.9185,
      "step": 148455
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.7061668634414673,
      "learning_rate": 0.00016858400997442832,
      "loss": 2.0095,
      "step": 148456
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.1229983568191528,
      "learning_rate": 0.0001685803327798672,
      "loss": 2.0876,
      "step": 148457
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.0661565065383911,
      "learning_rate": 0.00016857665560973933,
      "loss": 2.1066,
      "step": 148458
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.1602693796157837,
      "learning_rate": 0.00016857297846404568,
      "loss": 1.94,
      "step": 148459
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.2156063318252563,
      "learning_rate": 0.0001685693013427868,
      "loss": 2.2522,
      "step": 148460
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.0889272689819336,
      "learning_rate": 0.00016856562424596324,
      "loss": 1.8809,
      "step": 148461
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.0463947057724,
      "learning_rate": 0.00016856194717357594,
      "loss": 1.8974,
      "step": 148462
    },
    {
      "epoch": 1.93,
      "grad_norm": 0.9786050915718079,
      "learning_rate": 0.00016855827012562545,
      "loss": 2.1011,
      "step": 148463
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.124064326286316,
      "learning_rate": 0.00016855459310211233,
      "loss": 2.2108,
      "step": 148464
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.2421958446502686,
      "learning_rate": 0.00016855091610303754,
      "loss": 1.8977,
      "step": 148465
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.2175345420837402,
      "learning_rate": 0.00016854723912840147,
      "loss": 2.0993,
      "step": 148466
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.1585161685943604,
      "learning_rate": 0.00016854356217820509,
      "loss": 2.1978,
      "step": 148467
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.1602243185043335,
      "learning_rate": 0.00016853988525244892,
      "loss": 2.0889,
      "step": 148468
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.368393063545227,
      "learning_rate": 0.00016853620835113357,
      "loss": 2.2225,
      "step": 148469
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.268760085105896,
      "learning_rate": 0.00016853253147425988,
      "loss": 1.981,
      "step": 148470
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.1647982597351074,
      "learning_rate": 0.00016852885462182856,
      "loss": 2.0381,
      "step": 148471
    },
    {
      "epoch": 1.93,
      "grad_norm": 0.9748519659042358,
      "learning_rate": 0.00016852517779384002,
      "loss": 1.958,
      "step": 148472
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.1594823598861694,
      "learning_rate": 0.0001685215009902953,
      "loss": 2.063,
      "step": 148473
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.0777982473373413,
      "learning_rate": 0.00016851782421119487,
      "loss": 2.2444,
      "step": 148474
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.0358293056488037,
      "learning_rate": 0.0001685141474565394,
      "loss": 2.0265,
      "step": 148475
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.156132698059082,
      "learning_rate": 0.00016851047072632974,
      "loss": 2.0645,
      "step": 148476
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.0246562957763672,
      "learning_rate": 0.00016850679402056627,
      "loss": 1.9055,
      "step": 148477
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.1494104862213135,
      "learning_rate": 0.00016850311733925005,
      "loss": 2.1183,
      "step": 148478
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.2439146041870117,
      "learning_rate": 0.0001684994406823816,
      "loss": 2.059,
      "step": 148479
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.012485384941101,
      "learning_rate": 0.00016849576404996154,
      "loss": 2.2785,
      "step": 148480
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.0319634675979614,
      "learning_rate": 0.00016849208744199048,
      "loss": 1.855,
      "step": 148481
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.0348923206329346,
      "learning_rate": 0.00016848841085846937,
      "loss": 1.9149,
      "step": 148482
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.0470043420791626,
      "learning_rate": 0.00016848473429939862,
      "loss": 2.1361,
      "step": 148483
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.083430528640747,
      "learning_rate": 0.00016848105776477918,
      "loss": 1.934,
      "step": 148484
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.0612486600875854,
      "learning_rate": 0.00016847738125461153,
      "loss": 1.9726,
      "step": 148485
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.1221283674240112,
      "learning_rate": 0.00016847370476889647,
      "loss": 1.8475,
      "step": 148486
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.1422594785690308,
      "learning_rate": 0.00016847002830763452,
      "loss": 2.055,
      "step": 148487
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.1134661436080933,
      "learning_rate": 0.0001684663518708266,
      "loss": 2.2584,
      "step": 148488
    },
    {
      "epoch": 1.93,
      "grad_norm": 0.9755541086196899,
      "learning_rate": 0.00016846267545847312,
      "loss": 1.8923,
      "step": 148489
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.2349934577941895,
      "learning_rate": 0.00016845899907057501,
      "loss": 2.2056,
      "step": 148490
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.103792667388916,
      "learning_rate": 0.00016845532270713288,
      "loss": 2.1372,
      "step": 148491
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.071734070777893,
      "learning_rate": 0.0001684516463681474,
      "loss": 1.9633,
      "step": 148492
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.143188714981079,
      "learning_rate": 0.0001684479700536191,
      "loss": 2.2595,
      "step": 148493
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.2090470790863037,
      "learning_rate": 0.00016844429376354893,
      "loss": 1.9597,
      "step": 148494
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.128243327140808,
      "learning_rate": 0.00016844061749793734,
      "loss": 2.0832,
      "step": 148495
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.086230993270874,
      "learning_rate": 0.00016843694125678524,
      "loss": 1.9581,
      "step": 148496
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.3427784442901611,
      "learning_rate": 0.0001684332650400932,
      "loss": 2.1834,
      "step": 148497
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.1532444953918457,
      "learning_rate": 0.00016842958884786188,
      "loss": 2.3029,
      "step": 148498
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.212347388267517,
      "learning_rate": 0.0001684259126800919,
      "loss": 1.8981,
      "step": 148499
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.0691946744918823,
      "learning_rate": 0.00016842223653678413,
      "loss": 2.1782,
      "step": 148500
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.130761981010437,
      "learning_rate": 0.000168418560417939,
      "loss": 1.999,
      "step": 148501
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.1595470905303955,
      "learning_rate": 0.0001684148843235575,
      "loss": 1.9277,
      "step": 148502
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.2593426704406738,
      "learning_rate": 0.00016841120825364017,
      "loss": 2.096,
      "step": 148503
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.0926182270050049,
      "learning_rate": 0.00016840753220818765,
      "loss": 2.2196,
      "step": 148504
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.0608022212982178,
      "learning_rate": 0.0001684038561872005,
      "loss": 2.1113,
      "step": 148505
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.2067904472351074,
      "learning_rate": 0.00016840018019067973,
      "loss": 2.0247,
      "step": 148506
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.3687649965286255,
      "learning_rate": 0.00016839650421862572,
      "loss": 2.1003,
      "step": 148507
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.1820448637008667,
      "learning_rate": 0.00016839282827103938,
      "loss": 2.2306,
      "step": 148508
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.1444042921066284,
      "learning_rate": 0.00016838915234792135,
      "loss": 2.1618,
      "step": 148509
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.4218041896820068,
      "learning_rate": 0.00016838547644927221,
      "loss": 2.1134,
      "step": 148510
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.1987431049346924,
      "learning_rate": 0.00016838180057509257,
      "loss": 2.2458,
      "step": 148511
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.140077829360962,
      "learning_rate": 0.00016837812472538343,
      "loss": 2.0337,
      "step": 148512
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.1394050121307373,
      "learning_rate": 0.0001683744489001451,
      "loss": 2.0839,
      "step": 148513
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.0732738971710205,
      "learning_rate": 0.00016837077309937858,
      "loss": 1.9305,
      "step": 148514
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.2802248001098633,
      "learning_rate": 0.0001683670973230844,
      "loss": 2.3681,
      "step": 148515
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.3070361614227295,
      "learning_rate": 0.0001683634215712633,
      "loss": 1.9419,
      "step": 148516
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.2343116998672485,
      "learning_rate": 0.0001683597458439158,
      "loss": 2.0978,
      "step": 148517
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.1663119792938232,
      "learning_rate": 0.00016835607014104278,
      "loss": 2.2602,
      "step": 148518
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.1710456609725952,
      "learning_rate": 0.00016835239446264476,
      "loss": 1.9928,
      "step": 148519
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.1354219913482666,
      "learning_rate": 0.00016834871880872264,
      "loss": 2.0526,
      "step": 148520
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.2440067529678345,
      "learning_rate": 0.00016834504317927694,
      "loss": 2.1293,
      "step": 148521
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.1096466779708862,
      "learning_rate": 0.00016834136757430844,
      "loss": 2.0619,
      "step": 148522
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.112489104270935,
      "learning_rate": 0.00016833769199381762,
      "loss": 1.8891,
      "step": 148523
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.14485502243042,
      "learning_rate": 0.00016833401643780545,
      "loss": 2.1188,
      "step": 148524
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.222705602645874,
      "learning_rate": 0.0001683303409062723,
      "loss": 2.0861,
      "step": 148525
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.1219338178634644,
      "learning_rate": 0.0001683266653992192,
      "loss": 2.1516,
      "step": 148526
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.0413933992385864,
      "learning_rate": 0.00016832298991664664,
      "loss": 2.2182,
      "step": 148527
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.0840989351272583,
      "learning_rate": 0.00016831931445855528,
      "loss": 1.9904,
      "step": 148528
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.1648352146148682,
      "learning_rate": 0.00016831563902494577,
      "loss": 2.355,
      "step": 148529
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.2071146965026855,
      "learning_rate": 0.00016831196361581898,
      "loss": 1.8721,
      "step": 148530
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.1282421350479126,
      "learning_rate": 0.00016830828823117535,
      "loss": 2.1754,
      "step": 148531
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.125349998474121,
      "learning_rate": 0.00016830461287101585,
      "loss": 2.0761,
      "step": 148532
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.4813634157180786,
      "learning_rate": 0.00016830093753534086,
      "loss": 2.1644,
      "step": 148533
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.2523599863052368,
      "learning_rate": 0.00016829726222415147,
      "loss": 1.9435,
      "step": 148534
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.1317683458328247,
      "learning_rate": 0.00016829358693744783,
      "loss": 2.3324,
      "step": 148535
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.0026286840438843,
      "learning_rate": 0.00016828991167523104,
      "loss": 2.0606,
      "step": 148536
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.2230887413024902,
      "learning_rate": 0.00016828623643750152,
      "loss": 2.0297,
      "step": 148537
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.029844045639038,
      "learning_rate": 0.0001682825612242602,
      "loss": 2.0807,
      "step": 148538
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.2742974758148193,
      "learning_rate": 0.00016827888603550753,
      "loss": 2.1091,
      "step": 148539
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.080604076385498,
      "learning_rate": 0.00016827521087124441,
      "loss": 2.0947,
      "step": 148540
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.1112703084945679,
      "learning_rate": 0.0001682715357314714,
      "loss": 2.3151,
      "step": 148541
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.1227089166641235,
      "learning_rate": 0.0001682678606161892,
      "loss": 2.0258,
      "step": 148542
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.1393415927886963,
      "learning_rate": 0.00016826418552539838,
      "loss": 2.0405,
      "step": 148543
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.005488395690918,
      "learning_rate": 0.00016826051045909987,
      "loss": 2.1982,
      "step": 148544
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.0546613931655884,
      "learning_rate": 0.00016825683541729408,
      "loss": 2.0502,
      "step": 148545
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.1229102611541748,
      "learning_rate": 0.00016825316039998197,
      "loss": 2.1859,
      "step": 148546
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.110573649406433,
      "learning_rate": 0.00016824948540716404,
      "loss": 1.9143,
      "step": 148547
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.1534233093261719,
      "learning_rate": 0.00016824581043884104,
      "loss": 2.014,
      "step": 148548
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.2014473676681519,
      "learning_rate": 0.00016824213549501352,
      "loss": 1.8679,
      "step": 148549
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.1242133378982544,
      "learning_rate": 0.00016823846057568238,
      "loss": 2.0953,
      "step": 148550
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.1477142572402954,
      "learning_rate": 0.00016823478568084812,
      "loss": 2.1575,
      "step": 148551
    },
    {
      "epoch": 1.93,
      "grad_norm": 0.9964993596076965,
      "learning_rate": 0.00016823111081051156,
      "loss": 1.8519,
      "step": 148552
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.2746081352233887,
      "learning_rate": 0.0001682274359646734,
      "loss": 2.1001,
      "step": 148553
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.0552881956100464,
      "learning_rate": 0.00016822376114333402,
      "loss": 1.8798,
      "step": 148554
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.2046058177947998,
      "learning_rate": 0.00016822008634649453,
      "loss": 2.236,
      "step": 148555
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.2534193992614746,
      "learning_rate": 0.00016821641157415538,
      "loss": 1.9876,
      "step": 148556
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.0904678106307983,
      "learning_rate": 0.0001682127368263172,
      "loss": 2.1106,
      "step": 148557
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.062530517578125,
      "learning_rate": 0.00016820906210298085,
      "loss": 2.097,
      "step": 148558
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.118363618850708,
      "learning_rate": 0.00016820538740414695,
      "loss": 2.217,
      "step": 148559
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.2249610424041748,
      "learning_rate": 0.00016820171272981601,
      "loss": 2.1888,
      "step": 148560
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.2774758338928223,
      "learning_rate": 0.000168198038079989,
      "loss": 2.0136,
      "step": 148561
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.1137008666992188,
      "learning_rate": 0.00016819436345466644,
      "loss": 2.0678,
      "step": 148562
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.0515707731246948,
      "learning_rate": 0.00016819068885384894,
      "loss": 2.0087,
      "step": 148563
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.5586040019989014,
      "learning_rate": 0.00016818701427753738,
      "loss": 1.9942,
      "step": 148564
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.2824788093566895,
      "learning_rate": 0.00016818333972573235,
      "loss": 1.9218,
      "step": 148565
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.1292228698730469,
      "learning_rate": 0.00016817966519843438,
      "loss": 1.9454,
      "step": 148566
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.0935425758361816,
      "learning_rate": 0.00016817599069564448,
      "loss": 2.1061,
      "step": 148567
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.063279628753662,
      "learning_rate": 0.00016817231621736297,
      "loss": 2.0966,
      "step": 148568
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.2976737022399902,
      "learning_rate": 0.00016816864176359087,
      "loss": 2.1598,
      "step": 148569
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.2787216901779175,
      "learning_rate": 0.0001681649673343287,
      "loss": 2.1485,
      "step": 148570
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.1285072565078735,
      "learning_rate": 0.0001681612929295771,
      "loss": 2.1795,
      "step": 148571
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.430190920829773,
      "learning_rate": 0.00016815761854933677,
      "loss": 2.1299,
      "step": 148572
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.2020998001098633,
      "learning_rate": 0.0001681539441936085,
      "loss": 2.2525,
      "step": 148573
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.1170647144317627,
      "learning_rate": 0.00016815026986239282,
      "loss": 2.1044,
      "step": 148574
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.221888780593872,
      "learning_rate": 0.00016814659555569058,
      "loss": 1.9808,
      "step": 148575
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.1411802768707275,
      "learning_rate": 0.00016814292127350234,
      "loss": 2.2192,
      "step": 148576
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.0483537912368774,
      "learning_rate": 0.0001681392470158289,
      "loss": 2.0859,
      "step": 148577
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.162826418876648,
      "learning_rate": 0.00016813557278267072,
      "loss": 2.1004,
      "step": 148578
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.0802595615386963,
      "learning_rate": 0.0001681318985740287,
      "loss": 2.013,
      "step": 148579
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.2409240007400513,
      "learning_rate": 0.00016812822438990336,
      "loss": 1.8627,
      "step": 148580
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.0865532159805298,
      "learning_rate": 0.00016812455023029562,
      "loss": 1.8714,
      "step": 148581
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.2368417978286743,
      "learning_rate": 0.00016812087609520597,
      "loss": 2.0604,
      "step": 148582
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.334479570388794,
      "learning_rate": 0.00016811720198463514,
      "loss": 2.1794,
      "step": 148583
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.1846309900283813,
      "learning_rate": 0.00016811352789858371,
      "loss": 2.2311,
      "step": 148584
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.8296679258346558,
      "learning_rate": 0.00016810985383705257,
      "loss": 2.374,
      "step": 148585
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.3654764890670776,
      "learning_rate": 0.0001681061798000422,
      "loss": 1.9138,
      "step": 148586
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.2227094173431396,
      "learning_rate": 0.00016810250578755352,
      "loss": 2.0928,
      "step": 148587
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.4124243259429932,
      "learning_rate": 0.00016809883179958702,
      "loss": 2.0736,
      "step": 148588
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.2551480531692505,
      "learning_rate": 0.00016809515783614346,
      "loss": 2.0553,
      "step": 148589
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.1321927309036255,
      "learning_rate": 0.00016809148389722338,
      "loss": 2.2322,
      "step": 148590
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.6806795597076416,
      "learning_rate": 0.00016808780998282772,
      "loss": 1.8452,
      "step": 148591
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.6060700416564941,
      "learning_rate": 0.00016808413609295688,
      "loss": 2.0205,
      "step": 148592
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.1255323886871338,
      "learning_rate": 0.00016808046222761185,
      "loss": 2.2419,
      "step": 148593
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.1669166088104248,
      "learning_rate": 0.00016807678838679307,
      "loss": 2.0335,
      "step": 148594
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.2004318237304688,
      "learning_rate": 0.00016807311457050136,
      "loss": 2.236,
      "step": 148595
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.0894775390625,
      "learning_rate": 0.00016806944077873723,
      "loss": 2.1501,
      "step": 148596
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.1204105615615845,
      "learning_rate": 0.0001680657670115016,
      "loss": 2.0324,
      "step": 148597
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.0664640665054321,
      "learning_rate": 0.00016806209326879494,
      "loss": 2.1388,
      "step": 148598
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.0959773063659668,
      "learning_rate": 0.0001680584195506181,
      "loss": 2.2672,
      "step": 148599
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.1120887994766235,
      "learning_rate": 0.00016805474585697162,
      "loss": 2.2645,
      "step": 148600
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.2324336767196655,
      "learning_rate": 0.00016805107218785642,
      "loss": 2.0456,
      "step": 148601
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.0656828880310059,
      "learning_rate": 0.00016804739854327282,
      "loss": 2.2645,
      "step": 148602
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.1469241380691528,
      "learning_rate": 0.00016804372492322186,
      "loss": 2.091,
      "step": 148603
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.1073037385940552,
      "learning_rate": 0.00016804005132770388,
      "loss": 1.9158,
      "step": 148604
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.0750796794891357,
      "learning_rate": 0.00016803637775671988,
      "loss": 2.1054,
      "step": 148605
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.218628168106079,
      "learning_rate": 0.00016803270421027032,
      "loss": 1.9614,
      "step": 148606
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.140167236328125,
      "learning_rate": 0.0001680290306883562,
      "loss": 2.1425,
      "step": 148607
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.098486065864563,
      "learning_rate": 0.0001680253571909777,
      "loss": 2.2382,
      "step": 148608
    },
    {
      "epoch": 1.93,
      "grad_norm": 1.069545030593872,
      "learning_rate": 0.00016802168371813592,
      "loss": 1.9605,
      "step": 148609
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.1434543132781982,
      "learning_rate": 0.00016801801026983127,
      "loss": 1.9801,
      "step": 148610
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.2144423723220825,
      "learning_rate": 0.00016801433684606468,
      "loss": 2.0833,
      "step": 148611
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.1175357103347778,
      "learning_rate": 0.0001680106634468366,
      "loss": 1.8502,
      "step": 148612
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.1019623279571533,
      "learning_rate": 0.0001680069900721481,
      "loss": 2.1113,
      "step": 148613
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.1389250755310059,
      "learning_rate": 0.0001680033167219993,
      "loss": 2.1376,
      "step": 148614
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.2941817045211792,
      "learning_rate": 0.00016799964339639137,
      "loss": 2.2862,
      "step": 148615
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.1525332927703857,
      "learning_rate": 0.0001679959700953246,
      "loss": 2.1801,
      "step": 148616
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.1899837255477905,
      "learning_rate": 0.0001679922968188,
      "loss": 2.0614,
      "step": 148617
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.0823578834533691,
      "learning_rate": 0.00016798862356681805,
      "loss": 2.1598,
      "step": 148618
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.2399394512176514,
      "learning_rate": 0.00016798495033937976,
      "loss": 2.3284,
      "step": 148619
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.0797083377838135,
      "learning_rate": 0.00016798127713648525,
      "loss": 1.8461,
      "step": 148620
    },
    {
      "epoch": 1.94,
      "grad_norm": 0.9535485506057739,
      "learning_rate": 0.00016797760395813568,
      "loss": 2.036,
      "step": 148621
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.1249101161956787,
      "learning_rate": 0.00016797393080433142,
      "loss": 2.2239,
      "step": 148622
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.0848844051361084,
      "learning_rate": 0.00016797025767507342,
      "loss": 2.1036,
      "step": 148623
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.1728050708770752,
      "learning_rate": 0.00016796658457036217,
      "loss": 2.0573,
      "step": 148624
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.273886799812317,
      "learning_rate": 0.0001679629114901986,
      "loss": 1.925,
      "step": 148625
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.0677438974380493,
      "learning_rate": 0.000167959238434583,
      "loss": 2.1657,
      "step": 148626
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.1394826173782349,
      "learning_rate": 0.0001679555654035164,
      "loss": 2.0199,
      "step": 148627
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.1947699785232544,
      "learning_rate": 0.00016795189239699924,
      "loss": 2.2207,
      "step": 148628
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.0635149478912354,
      "learning_rate": 0.0001679482194150324,
      "loss": 2.197,
      "step": 148629
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.0825587511062622,
      "learning_rate": 0.00016794454645761638,
      "loss": 2.0086,
      "step": 148630
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.1738964319229126,
      "learning_rate": 0.00016794087352475207,
      "loss": 2.1005,
      "step": 148631
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.0871919393539429,
      "learning_rate": 0.00016793720061644007,
      "loss": 2.3056,
      "step": 148632
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.1905994415283203,
      "learning_rate": 0.00016793352773268106,
      "loss": 2.2209,
      "step": 148633
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.0519953966140747,
      "learning_rate": 0.00016792985487347551,
      "loss": 2.1128,
      "step": 148634
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.5255858898162842,
      "learning_rate": 0.00016792618203882446,
      "loss": 2.0613,
      "step": 148635
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.0759327411651611,
      "learning_rate": 0.00016792250922872829,
      "loss": 2.0624,
      "step": 148636
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.1359517574310303,
      "learning_rate": 0.0001679188364431879,
      "loss": 2.2844,
      "step": 148637
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.1984572410583496,
      "learning_rate": 0.00016791516368220397,
      "loss": 2.0393,
      "step": 148638
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.4249547719955444,
      "learning_rate": 0.00016791149094577695,
      "loss": 2.0342,
      "step": 148639
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.1585410833358765,
      "learning_rate": 0.00016790781823390778,
      "loss": 1.9739,
      "step": 148640
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.2568676471710205,
      "learning_rate": 0.00016790414554659707,
      "loss": 2.2421,
      "step": 148641
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.1516311168670654,
      "learning_rate": 0.00016790047288384532,
      "loss": 1.9456,
      "step": 148642
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.4172694683074951,
      "learning_rate": 0.00016789680024565352,
      "loss": 2.3025,
      "step": 148643
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.0806020498275757,
      "learning_rate": 0.00016789312763202218,
      "loss": 2.2692,
      "step": 148644
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.1711676120758057,
      "learning_rate": 0.00016788945504295186,
      "loss": 1.7755,
      "step": 148645
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.3598048686981201,
      "learning_rate": 0.00016788578247844355,
      "loss": 2.0542,
      "step": 148646
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.1033427715301514,
      "learning_rate": 0.00016788210993849773,
      "loss": 2.0409,
      "step": 148647
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.230425238609314,
      "learning_rate": 0.000167878437423115,
      "loss": 2.0936,
      "step": 148648
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.0826581716537476,
      "learning_rate": 0.0001678747649322963,
      "loss": 2.3852,
      "step": 148649
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.1941595077514648,
      "learning_rate": 0.00016787109246604217,
      "loss": 1.9996,
      "step": 148650
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.059390664100647,
      "learning_rate": 0.0001678674200243532,
      "loss": 1.9967,
      "step": 148651
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.0920957326889038,
      "learning_rate": 0.00016786374760723025,
      "loss": 1.8461,
      "step": 148652
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.156118392944336,
      "learning_rate": 0.00016786007521467383,
      "loss": 2.1342,
      "step": 148653
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.0884662866592407,
      "learning_rate": 0.00016785640284668483,
      "loss": 1.9883,
      "step": 148654
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.1267119646072388,
      "learning_rate": 0.00016785273050326385,
      "loss": 2.1636,
      "step": 148655
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.1864027976989746,
      "learning_rate": 0.0001678490581844115,
      "loss": 2.0856,
      "step": 148656
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.1637170314788818,
      "learning_rate": 0.00016784538589012838,
      "loss": 2.1351,
      "step": 148657
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.1175856590270996,
      "learning_rate": 0.00016784171362041545,
      "loss": 2.0375,
      "step": 148658
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.1255922317504883,
      "learning_rate": 0.00016783804137527307,
      "loss": 2.2036,
      "step": 148659
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.132729172706604,
      "learning_rate": 0.00016783436915470226,
      "loss": 2.1166,
      "step": 148660
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.0985338687896729,
      "learning_rate": 0.0001678306969587035,
      "loss": 2.2487,
      "step": 148661
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.180431842803955,
      "learning_rate": 0.00016782702478727757,
      "loss": 2.1013,
      "step": 148662
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.043053150177002,
      "learning_rate": 0.0001678233526404249,
      "loss": 2.1485,
      "step": 148663
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.0609369277954102,
      "learning_rate": 0.00016781968051814651,
      "loss": 2.2194,
      "step": 148664
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.1315462589263916,
      "learning_rate": 0.00016781600842044283,
      "loss": 2.0048,
      "step": 148665
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.1365293264389038,
      "learning_rate": 0.00016781233634731476,
      "loss": 2.0769,
      "step": 148666
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.088131070137024,
      "learning_rate": 0.00016780866429876273,
      "loss": 2.2269,
      "step": 148667
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.1357791423797607,
      "learning_rate": 0.00016780499227478786,
      "loss": 2.0799,
      "step": 148668
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.0845293998718262,
      "learning_rate": 0.00016780132027539025,
      "loss": 1.8947,
      "step": 148669
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.2016451358795166,
      "learning_rate": 0.00016779764830057104,
      "loss": 2.1803,
      "step": 148670
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.1371628046035767,
      "learning_rate": 0.00016779397635033056,
      "loss": 2.0193,
      "step": 148671
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.0564775466918945,
      "learning_rate": 0.00016779030442466982,
      "loss": 2.0386,
      "step": 148672
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.1674680709838867,
      "learning_rate": 0.00016778663252358923,
      "loss": 2.2954,
      "step": 148673
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.042125940322876,
      "learning_rate": 0.00016778296064708984,
      "loss": 2.1044,
      "step": 148674
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.1178345680236816,
      "learning_rate": 0.0001677792887951719,
      "loss": 2.2507,
      "step": 148675
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.2174419164657593,
      "learning_rate": 0.00016777561696783636,
      "loss": 2.0982,
      "step": 148676
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.2090606689453125,
      "learning_rate": 0.0001677719451650837,
      "loss": 2.1517,
      "step": 148677
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.2321351766586304,
      "learning_rate": 0.00016776827338691488,
      "loss": 2.0595,
      "step": 148678
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.1268631219863892,
      "learning_rate": 0.0001677646016333303,
      "loss": 2.0379,
      "step": 148679
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.0664881467819214,
      "learning_rate": 0.00016776092990433103,
      "loss": 2.1464,
      "step": 148680
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.0795201063156128,
      "learning_rate": 0.00016775725819991724,
      "loss": 2.0814,
      "step": 148681
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.1534454822540283,
      "learning_rate": 0.00016775358652009001,
      "loss": 2.0659,
      "step": 148682
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.1319066286087036,
      "learning_rate": 0.00016774991486484975,
      "loss": 1.7854,
      "step": 148683
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.1066417694091797,
      "learning_rate": 0.0001677462432341974,
      "loss": 2.1912,
      "step": 148684
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.1470530033111572,
      "learning_rate": 0.0001677425716281334,
      "loss": 2.3098,
      "step": 148685
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.1153658628463745,
      "learning_rate": 0.00016773890004665882,
      "loss": 2.1358,
      "step": 148686
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.2741175889968872,
      "learning_rate": 0.00016773522848977383,
      "loss": 2.1573,
      "step": 148687
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.1578994989395142,
      "learning_rate": 0.00016773155695747944,
      "loss": 2.184,
      "step": 148688
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.1158523559570312,
      "learning_rate": 0.00016772788544977614,
      "loss": 2.0874,
      "step": 148689
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.08823561668396,
      "learning_rate": 0.00016772421396666484,
      "loss": 2.0264,
      "step": 148690
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.148633360862732,
      "learning_rate": 0.00016772054250814605,
      "loss": 2.0296,
      "step": 148691
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.0523854494094849,
      "learning_rate": 0.00016771687107422067,
      "loss": 2.2474,
      "step": 148692
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.1542738676071167,
      "learning_rate": 0.000167713199664889,
      "loss": 1.9895,
      "step": 148693
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.0424836874008179,
      "learning_rate": 0.0001677095282801521,
      "loss": 1.7979,
      "step": 148694
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.1681240797042847,
      "learning_rate": 0.00016770585692001033,
      "loss": 2.0893,
      "step": 148695
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.1060441732406616,
      "learning_rate": 0.00016770218558446464,
      "loss": 2.2062,
      "step": 148696
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.044915795326233,
      "learning_rate": 0.00016769851427351552,
      "loss": 2.1117,
      "step": 148697
    },
    {
      "epoch": 1.94,
      "grad_norm": 0.9956285953521729,
      "learning_rate": 0.000167694842987164,
      "loss": 2.1817,
      "step": 148698
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.1826430559158325,
      "learning_rate": 0.00016769117172541026,
      "loss": 2.0945,
      "step": 148699
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.2726742029190063,
      "learning_rate": 0.00016768750048825535,
      "loss": 1.8854,
      "step": 148700
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.0199414491653442,
      "learning_rate": 0.0001676838292756997,
      "loss": 2.0776,
      "step": 148701
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.0721532106399536,
      "learning_rate": 0.00016768015808774428,
      "loss": 2.0783,
      "step": 148702
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.0608562231063843,
      "learning_rate": 0.00016767648692438947,
      "loss": 2.0604,
      "step": 148703
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.2030433416366577,
      "learning_rate": 0.00016767281578563629,
      "loss": 2.1978,
      "step": 148704
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.0925862789154053,
      "learning_rate": 0.00016766914467148506,
      "loss": 2.0324,
      "step": 148705
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.0328481197357178,
      "learning_rate": 0.00016766547358193674,
      "loss": 2.1264,
      "step": 148706
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.1336513757705688,
      "learning_rate": 0.00016766180251699176,
      "loss": 2.0851,
      "step": 148707
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.089361310005188,
      "learning_rate": 0.0001676581314766511,
      "loss": 2.1823,
      "step": 148708
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.3085261583328247,
      "learning_rate": 0.00016765446046091513,
      "loss": 2.0571,
      "step": 148709
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.1767631769180298,
      "learning_rate": 0.000167650789469785,
      "loss": 2.0556,
      "step": 148710
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.1622575521469116,
      "learning_rate": 0.0001676471185032608,
      "loss": 2.0711,
      "step": 148711
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.180651307106018,
      "learning_rate": 0.00016764344756134363,
      "loss": 2.25,
      "step": 148712
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.3742326498031616,
      "learning_rate": 0.0001676397766440339,
      "loss": 2.1348,
      "step": 148713
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.1150892972946167,
      "learning_rate": 0.00016763610575133258,
      "loss": 2.1706,
      "step": 148714
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.1989781856536865,
      "learning_rate": 0.00016763243488324007,
      "loss": 2.0019,
      "step": 148715
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.4298139810562134,
      "learning_rate": 0.0001676287640397573,
      "loss": 1.9393,
      "step": 148716
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.3813910484313965,
      "learning_rate": 0.0001676250932208849,
      "loss": 1.8921,
      "step": 148717
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.2511167526245117,
      "learning_rate": 0.00016762142242662343,
      "loss": 2.1985,
      "step": 148718
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.070841670036316,
      "learning_rate": 0.00016761775165697357,
      "loss": 2.038,
      "step": 148719
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.1433900594711304,
      "learning_rate": 0.00016761408091193613,
      "loss": 1.9435,
      "step": 148720
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.0596905946731567,
      "learning_rate": 0.00016761041019151165,
      "loss": 1.9703,
      "step": 148721
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.0693531036376953,
      "learning_rate": 0.000167606739495701,
      "loss": 2.0188,
      "step": 148722
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.154374122619629,
      "learning_rate": 0.00016760306882450477,
      "loss": 2.2826,
      "step": 148723
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.1352461576461792,
      "learning_rate": 0.0001675993981779236,
      "loss": 2.0789,
      "step": 148724
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.1978524923324585,
      "learning_rate": 0.0001675957275559581,
      "loss": 2.2171,
      "step": 148725
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.0916638374328613,
      "learning_rate": 0.00016759205695860915,
      "loss": 2.0884,
      "step": 148726
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.063313603401184,
      "learning_rate": 0.00016758838638587727,
      "loss": 2.0934,
      "step": 148727
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.0551109313964844,
      "learning_rate": 0.00016758471583776328,
      "loss": 1.8617,
      "step": 148728
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.1208181381225586,
      "learning_rate": 0.00016758104531426778,
      "loss": 2.0508,
      "step": 148729
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.098143458366394,
      "learning_rate": 0.00016757737481539142,
      "loss": 1.9309,
      "step": 148730
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.102862000465393,
      "learning_rate": 0.000167573704341135,
      "loss": 1.9896,
      "step": 148731
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.0762171745300293,
      "learning_rate": 0.00016757003389149911,
      "loss": 1.9369,
      "step": 148732
    },
    {
      "epoch": 1.94,
      "grad_norm": 0.9953158497810364,
      "learning_rate": 0.00016756636346648433,
      "loss": 2.1238,
      "step": 148733
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.141715407371521,
      "learning_rate": 0.0001675626930660916,
      "loss": 2.064,
      "step": 148734
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.1152034997940063,
      "learning_rate": 0.00016755902269032148,
      "loss": 2.188,
      "step": 148735
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.3322579860687256,
      "learning_rate": 0.0001675553523391745,
      "loss": 2.0525,
      "step": 148736
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.1884725093841553,
      "learning_rate": 0.00016755168201265162,
      "loss": 1.9218,
      "step": 148737
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.06514573097229,
      "learning_rate": 0.0001675480117107532,
      "loss": 1.9628,
      "step": 148738
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.0356003046035767,
      "learning_rate": 0.0001675443414334803,
      "loss": 1.9603,
      "step": 148739
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.197920799255371,
      "learning_rate": 0.00016754067118083336,
      "loss": 2.118,
      "step": 148740
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.1498416662216187,
      "learning_rate": 0.00016753700095281312,
      "loss": 2.1483,
      "step": 148741
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.1786282062530518,
      "learning_rate": 0.00016753333074942013,
      "loss": 1.9421,
      "step": 148742
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.2607591152191162,
      "learning_rate": 0.00016752966057065536,
      "loss": 1.9054,
      "step": 148743
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.1515835523605347,
      "learning_rate": 0.00016752599041651916,
      "loss": 2.0474,
      "step": 148744
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.5804425477981567,
      "learning_rate": 0.00016752232028701251,
      "loss": 1.9749,
      "step": 148745
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.115746021270752,
      "learning_rate": 0.00016751865018213593,
      "loss": 1.8256,
      "step": 148746
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.1204323768615723,
      "learning_rate": 0.0001675149801018902,
      "loss": 1.9996,
      "step": 148747
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.1296573877334595,
      "learning_rate": 0.00016751131004627578,
      "loss": 1.9372,
      "step": 148748
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.3401398658752441,
      "learning_rate": 0.0001675076400152936,
      "loss": 1.9932,
      "step": 148749
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.0544228553771973,
      "learning_rate": 0.00016750397000894418,
      "loss": 1.9481,
      "step": 148750
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.1174914836883545,
      "learning_rate": 0.00016750030002722835,
      "loss": 2.0229,
      "step": 148751
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.155915379524231,
      "learning_rate": 0.0001674966300701466,
      "loss": 2.3682,
      "step": 148752
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.2305840253829956,
      "learning_rate": 0.00016749296013770002,
      "loss": 2.1286,
      "step": 148753
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.1985981464385986,
      "learning_rate": 0.0001674892902298887,
      "loss": 2.2015,
      "step": 148754
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.1358188390731812,
      "learning_rate": 0.00016748562034671376,
      "loss": 2.0518,
      "step": 148755
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.2892029285430908,
      "learning_rate": 0.00016748195048817564,
      "loss": 2.0136,
      "step": 148756
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.3230584859848022,
      "learning_rate": 0.00016747828065427528,
      "loss": 2.2298,
      "step": 148757
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.117077112197876,
      "learning_rate": 0.000167474610845013,
      "loss": 2.1619,
      "step": 148758
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.032208800315857,
      "learning_rate": 0.00016747094106039,
      "loss": 2.0008,
      "step": 148759
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.0733836889266968,
      "learning_rate": 0.00016746727130040638,
      "loss": 2.0374,
      "step": 148760
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.192588448524475,
      "learning_rate": 0.00016746360156506324,
      "loss": 2.1247,
      "step": 148761
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.2337429523468018,
      "learning_rate": 0.000167459931854361,
      "loss": 2.0436,
      "step": 148762
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.2915153503417969,
      "learning_rate": 0.00016745626216830054,
      "loss": 1.9966,
      "step": 148763
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.0508836507797241,
      "learning_rate": 0.00016745259250688238,
      "loss": 2.0656,
      "step": 148764
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.1445339918136597,
      "learning_rate": 0.0001674489228701075,
      "loss": 2.0942,
      "step": 148765
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.1504632234573364,
      "learning_rate": 0.00016744525325797615,
      "loss": 1.9521,
      "step": 148766
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.147332787513733,
      "learning_rate": 0.00016744158367048935,
      "loss": 2.074,
      "step": 148767
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.1395245790481567,
      "learning_rate": 0.00016743791410764753,
      "loss": 2.0419,
      "step": 148768
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.1427737474441528,
      "learning_rate": 0.0001674342445694516,
      "loss": 1.9781,
      "step": 148769
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.0560752153396606,
      "learning_rate": 0.00016743057505590208,
      "loss": 2.161,
      "step": 148770
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.138999581336975,
      "learning_rate": 0.00016742690556699992,
      "loss": 1.9997,
      "step": 148771
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.0688318014144897,
      "learning_rate": 0.00016742323610274539,
      "loss": 1.9403,
      "step": 148772
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.0244653224945068,
      "learning_rate": 0.00016741956666313946,
      "loss": 2.0414,
      "step": 148773
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.0472491979599,
      "learning_rate": 0.00016741589724818265,
      "loss": 2.178,
      "step": 148774
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.2108924388885498,
      "learning_rate": 0.00016741222785787585,
      "loss": 2.223,
      "step": 148775
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.2391231060028076,
      "learning_rate": 0.00016740855849221948,
      "loss": 2.2591,
      "step": 148776
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.7530852556228638,
      "learning_rate": 0.0001674048891512146,
      "loss": 2.2015,
      "step": 148777
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.1421821117401123,
      "learning_rate": 0.00016740121983486146,
      "loss": 2.0332,
      "step": 148778
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.1512162685394287,
      "learning_rate": 0.000167397550543161,
      "loss": 2.2236,
      "step": 148779
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.5082544088363647,
      "learning_rate": 0.00016739388127611374,
      "loss": 1.8695,
      "step": 148780
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.0988733768463135,
      "learning_rate": 0.0001673902120337206,
      "loss": 1.967,
      "step": 148781
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.1355690956115723,
      "learning_rate": 0.00016738654281598194,
      "loss": 2.0731,
      "step": 148782
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.1943374872207642,
      "learning_rate": 0.0001673828736228989,
      "loss": 2.2028,
      "step": 148783
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.0461819171905518,
      "learning_rate": 0.0001673792044544717,
      "loss": 2.033,
      "step": 148784
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.2359501123428345,
      "learning_rate": 0.00016737553531070127,
      "loss": 2.0275,
      "step": 148785
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.1283074617385864,
      "learning_rate": 0.00016737186619158814,
      "loss": 2.0987,
      "step": 148786
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.3489327430725098,
      "learning_rate": 0.00016736819709713317,
      "loss": 2.1304,
      "step": 148787
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.0888010263442993,
      "learning_rate": 0.00016736452802733684,
      "loss": 2.01,
      "step": 148788
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.2982869148254395,
      "learning_rate": 0.00016736085898220025,
      "loss": 2.1526,
      "step": 148789
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.0420597791671753,
      "learning_rate": 0.00016735718996172347,
      "loss": 1.82,
      "step": 148790
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.0366121530532837,
      "learning_rate": 0.00016735352096590764,
      "loss": 2.0648,
      "step": 148791
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.3273717164993286,
      "learning_rate": 0.00016734985199475314,
      "loss": 2.2311,
      "step": 148792
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.040590763092041,
      "learning_rate": 0.00016734618304826095,
      "loss": 2.0897,
      "step": 148793
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.1226314306259155,
      "learning_rate": 0.00016734251412643152,
      "loss": 2.0701,
      "step": 148794
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.1259008646011353,
      "learning_rate": 0.00016733884522926584,
      "loss": 2.2089,
      "step": 148795
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.1188414096832275,
      "learning_rate": 0.0001673351763567641,
      "loss": 2.1642,
      "step": 148796
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.1223316192626953,
      "learning_rate": 0.00016733150750892744,
      "loss": 2.1133,
      "step": 148797
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.1285775899887085,
      "learning_rate": 0.00016732783868575617,
      "loss": 2.1008,
      "step": 148798
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.0549813508987427,
      "learning_rate": 0.0001673241698872513,
      "loss": 1.8591,
      "step": 148799
    },
    {
      "epoch": 1.94,
      "grad_norm": 0.997893214225769,
      "learning_rate": 0.0001673205011134133,
      "loss": 2.1164,
      "step": 148800
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.0808031558990479,
      "learning_rate": 0.00016731683236424297,
      "loss": 1.9409,
      "step": 148801
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.11011803150177,
      "learning_rate": 0.000167313163639741,
      "loss": 2.0053,
      "step": 148802
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.109771728515625,
      "learning_rate": 0.000167309494939908,
      "loss": 1.7927,
      "step": 148803
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.105075478553772,
      "learning_rate": 0.00016730582626474453,
      "loss": 1.9563,
      "step": 148804
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.1618304252624512,
      "learning_rate": 0.00016730215761425154,
      "loss": 2.0008,
      "step": 148805
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.2852919101715088,
      "learning_rate": 0.00016729848898842947,
      "loss": 2.1747,
      "step": 148806
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.1730785369873047,
      "learning_rate": 0.00016729482038727926,
      "loss": 2.1453,
      "step": 148807
    },
    {
      "epoch": 1.94,
      "grad_norm": 0.9937484860420227,
      "learning_rate": 0.00016729115181080142,
      "loss": 1.9808,
      "step": 148808
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.1707926988601685,
      "learning_rate": 0.0001672874832589967,
      "loss": 2.0469,
      "step": 148809
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.067674994468689,
      "learning_rate": 0.00016728381473186553,
      "loss": 1.7883,
      "step": 148810
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.060378074645996,
      "learning_rate": 0.000167280146229409,
      "loss": 2.225,
      "step": 148811
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.2063043117523193,
      "learning_rate": 0.00016727647775162746,
      "loss": 2.097,
      "step": 148812
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.0878093242645264,
      "learning_rate": 0.00016727280929852183,
      "loss": 2.3572,
      "step": 148813
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.1446824073791504,
      "learning_rate": 0.00016726914087009275,
      "loss": 2.114,
      "step": 148814
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.0840387344360352,
      "learning_rate": 0.00016726547246634065,
      "loss": 2.1592,
      "step": 148815
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.076850175857544,
      "learning_rate": 0.00016726180408726656,
      "loss": 2.1316,
      "step": 148816
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.197663426399231,
      "learning_rate": 0.000167258135732871,
      "loss": 2.1625,
      "step": 148817
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.2932301759719849,
      "learning_rate": 0.00016725446740315452,
      "loss": 2.18,
      "step": 148818
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.1911826133728027,
      "learning_rate": 0.0001672507990981181,
      "loss": 1.8383,
      "step": 148819
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.1044893264770508,
      "learning_rate": 0.00016724713081776223,
      "loss": 2.0107,
      "step": 148820
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.116624116897583,
      "learning_rate": 0.0001672434625620875,
      "loss": 1.9462,
      "step": 148821
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.1490803956985474,
      "learning_rate": 0.00016723979433109484,
      "loss": 2.0612,
      "step": 148822
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.096563458442688,
      "learning_rate": 0.0001672361261247847,
      "loss": 1.9815,
      "step": 148823
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.125158667564392,
      "learning_rate": 0.00016723245794315802,
      "loss": 2.0665,
      "step": 148824
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.0906840562820435,
      "learning_rate": 0.0001672287897862153,
      "loss": 2.0574,
      "step": 148825
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.281498908996582,
      "learning_rate": 0.00016722512165395724,
      "loss": 2.3644,
      "step": 148826
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.2192983627319336,
      "learning_rate": 0.0001672214535463844,
      "loss": 2.2023,
      "step": 148827
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.1043702363967896,
      "learning_rate": 0.0001672177854634978,
      "loss": 1.9876,
      "step": 148828
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.0829081535339355,
      "learning_rate": 0.00016721411740529773,
      "loss": 2.153,
      "step": 148829
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.0645360946655273,
      "learning_rate": 0.0001672104493717852,
      "loss": 2.0568,
      "step": 148830
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.0878788232803345,
      "learning_rate": 0.00016720678136296075,
      "loss": 2.3039,
      "step": 148831
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.16314697265625,
      "learning_rate": 0.00016720311337882506,
      "loss": 2.1537,
      "step": 148832
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.0661689043045044,
      "learning_rate": 0.00016719944541937872,
      "loss": 2.1266,
      "step": 148833
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.21342134475708,
      "learning_rate": 0.0001671957774846226,
      "loss": 2.1129,
      "step": 148834
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.1577794551849365,
      "learning_rate": 0.0001671921095745572,
      "loss": 1.9686,
      "step": 148835
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.0752105712890625,
      "learning_rate": 0.00016718844168918344,
      "loss": 1.954,
      "step": 148836
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.1872837543487549,
      "learning_rate": 0.00016718477382850167,
      "loss": 2.1546,
      "step": 148837
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.2044405937194824,
      "learning_rate": 0.00016718110599251308,
      "loss": 2.1125,
      "step": 148838
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.0930997133255005,
      "learning_rate": 0.00016717743818121773,
      "loss": 2.1376,
      "step": 148839
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.0712718963623047,
      "learning_rate": 0.00016717377039461668,
      "loss": 2.145,
      "step": 148840
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.1461750268936157,
      "learning_rate": 0.00016717010263271048,
      "loss": 2.1064,
      "step": 148841
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.0785309076309204,
      "learning_rate": 0.0001671664348955,
      "loss": 2.1595,
      "step": 148842
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.233899712562561,
      "learning_rate": 0.00016716276718298562,
      "loss": 2.2057,
      "step": 148843
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.1310421228408813,
      "learning_rate": 0.00016715909949516843,
      "loss": 2.2359,
      "step": 148844
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.1580798625946045,
      "learning_rate": 0.00016715543183204866,
      "loss": 2.0467,
      "step": 148845
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.2249382734298706,
      "learning_rate": 0.00016715176419362728,
      "loss": 1.8898,
      "step": 148846
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.1380270719528198,
      "learning_rate": 0.00016714809657990482,
      "loss": 2.1976,
      "step": 148847
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.0850672721862793,
      "learning_rate": 0.00016714442899088214,
      "loss": 1.8131,
      "step": 148848
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.163280963897705,
      "learning_rate": 0.0001671407614265597,
      "loss": 2.1211,
      "step": 148849
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.1244949102401733,
      "learning_rate": 0.00016713709388693854,
      "loss": 2.0307,
      "step": 148850
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.1731319427490234,
      "learning_rate": 0.00016713342637201888,
      "loss": 2.0733,
      "step": 148851
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.2605788707733154,
      "learning_rate": 0.00016712975888180168,
      "loss": 1.9595,
      "step": 148852
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.2729750871658325,
      "learning_rate": 0.00016712609141628752,
      "loss": 2.4243,
      "step": 148853
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.2847797870635986,
      "learning_rate": 0.0001671224239754772,
      "loss": 2.0716,
      "step": 148854
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.2057324647903442,
      "learning_rate": 0.00016711875655937123,
      "loss": 1.9932,
      "step": 148855
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.1425700187683105,
      "learning_rate": 0.00016711508916797063,
      "loss": 2.1177,
      "step": 148856
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.1469154357910156,
      "learning_rate": 0.0001671114218012756,
      "loss": 2.2692,
      "step": 148857
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.4736205339431763,
      "learning_rate": 0.0001671077544592872,
      "loss": 1.9172,
      "step": 148858
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.1033467054367065,
      "learning_rate": 0.00016710408714200583,
      "loss": 2.0718,
      "step": 148859
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.0851608514785767,
      "learning_rate": 0.00016710041984943245,
      "loss": 2.0679,
      "step": 148860
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.3161849975585938,
      "learning_rate": 0.0001670967525815675,
      "loss": 2.2479,
      "step": 148861
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.1777994632720947,
      "learning_rate": 0.00016709308533841203,
      "loss": 2.2525,
      "step": 148862
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.0686380863189697,
      "learning_rate": 0.0001670894181199662,
      "loss": 2.0486,
      "step": 148863
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.0223897695541382,
      "learning_rate": 0.00016708575092623106,
      "loss": 2.0753,
      "step": 148864
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.169661045074463,
      "learning_rate": 0.00016708208375720707,
      "loss": 2.1811,
      "step": 148865
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.254726529121399,
      "learning_rate": 0.0001670784166128952,
      "loss": 2.1882,
      "step": 148866
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.2627683877944946,
      "learning_rate": 0.0001670747494932958,
      "loss": 2.0321,
      "step": 148867
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.1121832132339478,
      "learning_rate": 0.00016707108239840985,
      "loss": 2.1791,
      "step": 148868
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.161260724067688,
      "learning_rate": 0.00016706741532823788,
      "loss": 1.9451,
      "step": 148869
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.2766010761260986,
      "learning_rate": 0.00016706374828278058,
      "loss": 2.3555,
      "step": 148870
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.2637797594070435,
      "learning_rate": 0.00016706008126203854,
      "loss": 2.209,
      "step": 148871
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.0971297025680542,
      "learning_rate": 0.00016705641426601264,
      "loss": 1.9803,
      "step": 148872
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.1755009889602661,
      "learning_rate": 0.00016705274729470336,
      "loss": 2.1793,
      "step": 148873
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.151497483253479,
      "learning_rate": 0.0001670490803481116,
      "loss": 2.0822,
      "step": 148874
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.1583575010299683,
      "learning_rate": 0.00016704541342623793,
      "loss": 2.1944,
      "step": 148875
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.216863751411438,
      "learning_rate": 0.000167041746529083,
      "loss": 1.9796,
      "step": 148876
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.4111307859420776,
      "learning_rate": 0.0001670380796566474,
      "loss": 1.9154,
      "step": 148877
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.05731999874115,
      "learning_rate": 0.0001670344128089321,
      "loss": 2.3277,
      "step": 148878
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.1089049577713013,
      "learning_rate": 0.00016703074598593747,
      "loss": 2.1674,
      "step": 148879
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.1408647298812866,
      "learning_rate": 0.0001670270791876644,
      "loss": 2.0474,
      "step": 148880
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.304481029510498,
      "learning_rate": 0.00016702341241411358,
      "loss": 2.0195,
      "step": 148881
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.321075201034546,
      "learning_rate": 0.0001670197456652856,
      "loss": 1.9916,
      "step": 148882
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.0947372913360596,
      "learning_rate": 0.000167016078941181,
      "loss": 2.1224,
      "step": 148883
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.1052271127700806,
      "learning_rate": 0.00016701241224180077,
      "loss": 2.114,
      "step": 148884
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.214467167854309,
      "learning_rate": 0.00016700874556714533,
      "loss": 2.0114,
      "step": 148885
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.8475984334945679,
      "learning_rate": 0.00016700507891721558,
      "loss": 2.1176,
      "step": 148886
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.2363611459732056,
      "learning_rate": 0.00016700141229201212,
      "loss": 2.2625,
      "step": 148887
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.1480841636657715,
      "learning_rate": 0.00016699774569153562,
      "loss": 2.0411,
      "step": 148888
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.2812960147857666,
      "learning_rate": 0.00016699407911578655,
      "loss": 2.0918,
      "step": 148889
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.0577418804168701,
      "learning_rate": 0.00016699041256476599,
      "loss": 1.8506,
      "step": 148890
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.133999228477478,
      "learning_rate": 0.0001669867460384743,
      "loss": 1.9689,
      "step": 148891
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.1681684255599976,
      "learning_rate": 0.00016698307953691235,
      "loss": 1.9157,
      "step": 148892
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.114521861076355,
      "learning_rate": 0.00016697941306008078,
      "loss": 1.8725,
      "step": 148893
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.1133240461349487,
      "learning_rate": 0.00016697574660798025,
      "loss": 1.9704,
      "step": 148894
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.1163164377212524,
      "learning_rate": 0.00016697208018061135,
      "loss": 2.2856,
      "step": 148895
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.1536377668380737,
      "learning_rate": 0.00016696841377797491,
      "loss": 1.8844,
      "step": 148896
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.1002345085144043,
      "learning_rate": 0.00016696474740007152,
      "loss": 2.1067,
      "step": 148897
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.1634689569473267,
      "learning_rate": 0.00016696108104690194,
      "loss": 2.1678,
      "step": 148898
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.244162678718567,
      "learning_rate": 0.00016695741471846684,
      "loss": 2.0947,
      "step": 148899
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.118442416191101,
      "learning_rate": 0.00016695374841476672,
      "loss": 2.1381,
      "step": 148900
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.2142853736877441,
      "learning_rate": 0.00016695008213580255,
      "loss": 1.9857,
      "step": 148901
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.3149741888046265,
      "learning_rate": 0.00016694641588157488,
      "loss": 2.0105,
      "step": 148902
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.3091069459915161,
      "learning_rate": 0.00016694274965208425,
      "loss": 2.0996,
      "step": 148903
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.3149642944335938,
      "learning_rate": 0.0001669390834473316,
      "loss": 1.988,
      "step": 148904
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.101758360862732,
      "learning_rate": 0.0001669354172673175,
      "loss": 2.0082,
      "step": 148905
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.054703950881958,
      "learning_rate": 0.00016693175111204245,
      "loss": 1.9894,
      "step": 148906
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.2046656608581543,
      "learning_rate": 0.00016692808498150748,
      "loss": 2.213,
      "step": 148907
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.1341173648834229,
      "learning_rate": 0.00016692441887571305,
      "loss": 2.1167,
      "step": 148908
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.1492178440093994,
      "learning_rate": 0.00016692075279465976,
      "loss": 2.1448,
      "step": 148909
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.0929855108261108,
      "learning_rate": 0.00016691708673834853,
      "loss": 2.0771,
      "step": 148910
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.0984351634979248,
      "learning_rate": 0.00016691342070677997,
      "loss": 2.087,
      "step": 148911
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.121598720550537,
      "learning_rate": 0.00016690975469995457,
      "loss": 2.1196,
      "step": 148912
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.0908187627792358,
      "learning_rate": 0.00016690608871787328,
      "loss": 1.9279,
      "step": 148913
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.1478346586227417,
      "learning_rate": 0.0001669024227605365,
      "loss": 1.9888,
      "step": 148914
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.2626094818115234,
      "learning_rate": 0.00016689875682794524,
      "loss": 2.1053,
      "step": 148915
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.1693834066390991,
      "learning_rate": 0.0001668950909201,
      "loss": 2.0217,
      "step": 148916
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.2110116481781006,
      "learning_rate": 0.00016689142503700154,
      "loss": 2.0847,
      "step": 148917
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.2179659605026245,
      "learning_rate": 0.00016688775917865026,
      "loss": 1.9794,
      "step": 148918
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.238630771636963,
      "learning_rate": 0.00016688409334504725,
      "loss": 2.0768,
      "step": 148919
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.2119438648223877,
      "learning_rate": 0.00016688042753619284,
      "loss": 1.9222,
      "step": 148920
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.1055271625518799,
      "learning_rate": 0.00016687676175208797,
      "loss": 1.8876,
      "step": 148921
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.1090071201324463,
      "learning_rate": 0.00016687309599273313,
      "loss": 2.154,
      "step": 148922
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.2962509393692017,
      "learning_rate": 0.00016686943025812935,
      "loss": 1.8161,
      "step": 148923
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.1574842929840088,
      "learning_rate": 0.00016686576454827678,
      "loss": 2.1201,
      "step": 148924
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.1463185548782349,
      "learning_rate": 0.00016686209886317653,
      "loss": 2.1277,
      "step": 148925
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.2325763702392578,
      "learning_rate": 0.00016685843320282898,
      "loss": 1.925,
      "step": 148926
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.102029800415039,
      "learning_rate": 0.00016685476756723514,
      "loss": 1.8775,
      "step": 148927
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.2585283517837524,
      "learning_rate": 0.00016685110195639532,
      "loss": 1.9609,
      "step": 148928
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.0397858619689941,
      "learning_rate": 0.00016684743637031065,
      "loss": 2.1697,
      "step": 148929
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.0968422889709473,
      "learning_rate": 0.00016684377080898133,
      "loss": 2.0612,
      "step": 148930
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.0967247486114502,
      "learning_rate": 0.00016684010527240838,
      "loss": 2.1019,
      "step": 148931
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.0894445180892944,
      "learning_rate": 0.00016683643976059224,
      "loss": 1.9344,
      "step": 148932
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.1041815280914307,
      "learning_rate": 0.0001668327742735339,
      "loss": 2.2705,
      "step": 148933
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.1462048292160034,
      "learning_rate": 0.0001668291088112337,
      "loss": 2.0337,
      "step": 148934
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.1676732301712036,
      "learning_rate": 0.00016682544337369255,
      "loss": 2.179,
      "step": 148935
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.2041418552398682,
      "learning_rate": 0.0001668217779609111,
      "loss": 2.1942,
      "step": 148936
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.1461559534072876,
      "learning_rate": 0.00016681811257289003,
      "loss": 2.1344,
      "step": 148937
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.0553817749023438,
      "learning_rate": 0.0001668144472096298,
      "loss": 1.849,
      "step": 148938
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.1823246479034424,
      "learning_rate": 0.00016681078187113146,
      "loss": 2.0987,
      "step": 148939
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.242201566696167,
      "learning_rate": 0.00016680711655739537,
      "loss": 1.9288,
      "step": 148940
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.121944785118103,
      "learning_rate": 0.0001668034512684225,
      "loss": 1.9636,
      "step": 148941
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.3178517818450928,
      "learning_rate": 0.00016679978600421335,
      "loss": 2.2854,
      "step": 148942
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.200156569480896,
      "learning_rate": 0.00016679612076476863,
      "loss": 2.069,
      "step": 148943
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.0864884853363037,
      "learning_rate": 0.00016679245555008894,
      "loss": 2.0738,
      "step": 148944
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.2632218599319458,
      "learning_rate": 0.00016678879036017508,
      "loss": 2.1851,
      "step": 148945
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.248413324356079,
      "learning_rate": 0.00016678512519502765,
      "loss": 2.1149,
      "step": 148946
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.222322702407837,
      "learning_rate": 0.00016678146005464752,
      "loss": 2.161,
      "step": 148947
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.3414418697357178,
      "learning_rate": 0.00016677779493903518,
      "loss": 2.1431,
      "step": 148948
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.1040747165679932,
      "learning_rate": 0.0001667741298481914,
      "loss": 2.0613,
      "step": 148949
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.2161545753479004,
      "learning_rate": 0.00016677046478211666,
      "loss": 1.9786,
      "step": 148950
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.2144780158996582,
      "learning_rate": 0.00016676679974081195,
      "loss": 2.0834,
      "step": 148951
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.0968183279037476,
      "learning_rate": 0.0001667631347242777,
      "loss": 2.1419,
      "step": 148952
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.1466635465621948,
      "learning_rate": 0.00016675946973251481,
      "loss": 2.2359,
      "step": 148953
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.1606457233428955,
      "learning_rate": 0.00016675580476552387,
      "loss": 2.1149,
      "step": 148954
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.1306846141815186,
      "learning_rate": 0.0001667521398233055,
      "loss": 2.0868,
      "step": 148955
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.3869562149047852,
      "learning_rate": 0.0001667484749058603,
      "loss": 1.9077,
      "step": 148956
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.3189067840576172,
      "learning_rate": 0.00016674481001318923,
      "loss": 2.0371,
      "step": 148957
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.0554187297821045,
      "learning_rate": 0.00016674114514529266,
      "loss": 2.0125,
      "step": 148958
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.4969184398651123,
      "learning_rate": 0.0001667374803021716,
      "loss": 2.2114,
      "step": 148959
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.3018947839736938,
      "learning_rate": 0.0001667338154838265,
      "loss": 2.1413,
      "step": 148960
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.5616471767425537,
      "learning_rate": 0.00016673015069025815,
      "loss": 2.051,
      "step": 148961
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.0723862648010254,
      "learning_rate": 0.000166726485921467,
      "loss": 1.6882,
      "step": 148962
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.1286364793777466,
      "learning_rate": 0.00016672282117745406,
      "loss": 2.1256,
      "step": 148963
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.0873674154281616,
      "learning_rate": 0.0001667191564582198,
      "loss": 2.1314,
      "step": 148964
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.243659257888794,
      "learning_rate": 0.00016671549176376502,
      "loss": 2.0415,
      "step": 148965
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.1462124586105347,
      "learning_rate": 0.00016671182709409037,
      "loss": 2.1337,
      "step": 148966
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.2231922149658203,
      "learning_rate": 0.0001667081624491965,
      "loss": 1.9128,
      "step": 148967
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.084006428718567,
      "learning_rate": 0.00016670449782908391,
      "loss": 1.9566,
      "step": 148968
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.059462070465088,
      "learning_rate": 0.00016670083323375368,
      "loss": 2.1688,
      "step": 148969
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.0823410749435425,
      "learning_rate": 0.00016669716866320612,
      "loss": 2.0466,
      "step": 148970
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.0471047163009644,
      "learning_rate": 0.00016669350411744222,
      "loss": 1.98,
      "step": 148971
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.079490303993225,
      "learning_rate": 0.00016668983959646253,
      "loss": 2.2182,
      "step": 148972
    },
    {
      "epoch": 1.94,
      "grad_norm": 0.9795377254486084,
      "learning_rate": 0.00016668617510026768,
      "loss": 2.0504,
      "step": 148973
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.047147274017334,
      "learning_rate": 0.0001666825106288583,
      "loss": 1.9589,
      "step": 148974
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.1948250532150269,
      "learning_rate": 0.00016667884618223522,
      "loss": 2.3174,
      "step": 148975
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.232820987701416,
      "learning_rate": 0.00016667518176039897,
      "loss": 2.114,
      "step": 148976
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.133127212524414,
      "learning_rate": 0.00016667151736335045,
      "loss": 2.1291,
      "step": 148977
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.0831704139709473,
      "learning_rate": 0.00016666785299109023,
      "loss": 1.901,
      "step": 148978
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.1329528093338013,
      "learning_rate": 0.00016666418864361894,
      "loss": 2.1875,
      "step": 148979
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.0828425884246826,
      "learning_rate": 0.00016666052432093715,
      "loss": 2.174,
      "step": 148980
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.2232266664505005,
      "learning_rate": 0.00016665686002304583,
      "loss": 2.0701,
      "step": 148981
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.0236374139785767,
      "learning_rate": 0.00016665319574994542,
      "loss": 1.9772,
      "step": 148982
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.043917179107666,
      "learning_rate": 0.0001666495315016368,
      "loss": 2.1962,
      "step": 148983
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.0328465700149536,
      "learning_rate": 0.00016664586727812056,
      "loss": 1.8818,
      "step": 148984
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.116127848625183,
      "learning_rate": 0.00016664220307939726,
      "loss": 2.0325,
      "step": 148985
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.155157446861267,
      "learning_rate": 0.0001666385389054678,
      "loss": 2.0045,
      "step": 148986
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.1042773723602295,
      "learning_rate": 0.00016663487475633275,
      "loss": 1.9287,
      "step": 148987
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.019740343093872,
      "learning_rate": 0.00016663121063199265,
      "loss": 2.0577,
      "step": 148988
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.2376363277435303,
      "learning_rate": 0.00016662754653244845,
      "loss": 2.0812,
      "step": 148989
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.094177484512329,
      "learning_rate": 0.00016662388245770076,
      "loss": 2.1166,
      "step": 148990
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.148755431175232,
      "learning_rate": 0.00016662021840775003,
      "loss": 1.8905,
      "step": 148991
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.3322222232818604,
      "learning_rate": 0.00016661655438259725,
      "loss": 2.1482,
      "step": 148992
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.0672798156738281,
      "learning_rate": 0.00016661289038224296,
      "loss": 2.0726,
      "step": 148993
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.0496433973312378,
      "learning_rate": 0.00016660922640668777,
      "loss": 2.0742,
      "step": 148994
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.4399676322937012,
      "learning_rate": 0.00016660556245593256,
      "loss": 2.0427,
      "step": 148995
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.3818458318710327,
      "learning_rate": 0.00016660189852997785,
      "loss": 2.0041,
      "step": 148996
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.388129472732544,
      "learning_rate": 0.00016659823462882432,
      "loss": 2.0425,
      "step": 148997
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.0919218063354492,
      "learning_rate": 0.00016659457075247273,
      "loss": 2.0509,
      "step": 148998
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.1926908493041992,
      "learning_rate": 0.00016659090690092365,
      "loss": 2.2236,
      "step": 148999
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.5884466171264648,
      "learning_rate": 0.000166587243074178,
      "loss": 2.1253,
      "step": 149000
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.2560392618179321,
      "learning_rate": 0.00016658357927223623,
      "loss": 2.1471,
      "step": 149001
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.0969489812850952,
      "learning_rate": 0.000166579915495099,
      "loss": 2.2247,
      "step": 149002
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.1402901411056519,
      "learning_rate": 0.0001665762517427672,
      "loss": 1.8276,
      "step": 149003
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.1528851985931396,
      "learning_rate": 0.0001665725880152414,
      "loss": 2.0273,
      "step": 149004
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.2511004209518433,
      "learning_rate": 0.00016656892431252213,
      "loss": 1.9761,
      "step": 149005
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.1713743209838867,
      "learning_rate": 0.00016656526063461036,
      "loss": 2.1323,
      "step": 149006
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.0929136276245117,
      "learning_rate": 0.0001665615969815066,
      "loss": 2.0686,
      "step": 149007
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.1931405067443848,
      "learning_rate": 0.00016655793335321146,
      "loss": 2.0518,
      "step": 149008
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.142547607421875,
      "learning_rate": 0.00016655426974972588,
      "loss": 2.1218,
      "step": 149009
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.0262420177459717,
      "learning_rate": 0.00016655060617105032,
      "loss": 1.9262,
      "step": 149010
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.0556926727294922,
      "learning_rate": 0.0001665469426171854,
      "loss": 2.0365,
      "step": 149011
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.185568928718567,
      "learning_rate": 0.00016654327908813206,
      "loss": 2.3033,
      "step": 149012
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.073487639427185,
      "learning_rate": 0.00016653961558389075,
      "loss": 2.0913,
      "step": 149013
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.1028566360473633,
      "learning_rate": 0.00016653595210446235,
      "loss": 2.0199,
      "step": 149014
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.2059483528137207,
      "learning_rate": 0.0001665322886498474,
      "loss": 2.2199,
      "step": 149015
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.2587791681289673,
      "learning_rate": 0.00016652862522004662,
      "loss": 2.037,
      "step": 149016
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.147101640701294,
      "learning_rate": 0.00016652496181506063,
      "loss": 1.9634,
      "step": 149017
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.1166006326675415,
      "learning_rate": 0.00016652129843489023,
      "loss": 2.1266,
      "step": 149018
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.182766079902649,
      "learning_rate": 0.00016651763507953595,
      "loss": 2.2221,
      "step": 149019
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.0777226686477661,
      "learning_rate": 0.00016651397174899865,
      "loss": 1.9832,
      "step": 149020
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.2215549945831299,
      "learning_rate": 0.00016651030844327895,
      "loss": 2.0856,
      "step": 149021
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.1134834289550781,
      "learning_rate": 0.00016650664516237748,
      "loss": 2.0343,
      "step": 149022
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.2179107666015625,
      "learning_rate": 0.00016650298190629486,
      "loss": 2.1976,
      "step": 149023
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.120589017868042,
      "learning_rate": 0.00016649931867503197,
      "loss": 2.0427,
      "step": 149024
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.2617802619934082,
      "learning_rate": 0.00016649565546858925,
      "loss": 1.8924,
      "step": 149025
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.1527155637741089,
      "learning_rate": 0.0001664919922869676,
      "loss": 1.9828,
      "step": 149026
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.1686735153198242,
      "learning_rate": 0.00016648832913016763,
      "loss": 1.9167,
      "step": 149027
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.1501057147979736,
      "learning_rate": 0.00016648466599819,
      "loss": 2.0119,
      "step": 149028
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.1689666509628296,
      "learning_rate": 0.00016648100289103526,
      "loss": 2.0692,
      "step": 149029
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.3544679880142212,
      "learning_rate": 0.0001664773398087044,
      "loss": 2.0644,
      "step": 149030
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.103590965270996,
      "learning_rate": 0.00016647367675119774,
      "loss": 2.0265,
      "step": 149031
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.2686097621917725,
      "learning_rate": 0.00016647001371851626,
      "loss": 2.3247,
      "step": 149032
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.0890672206878662,
      "learning_rate": 0.00016646635071066056,
      "loss": 2.0182,
      "step": 149033
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.1686564683914185,
      "learning_rate": 0.00016646268772763125,
      "loss": 2.1958,
      "step": 149034
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.0668267011642456,
      "learning_rate": 0.00016645902476942893,
      "loss": 2.0566,
      "step": 149035
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.2449783086776733,
      "learning_rate": 0.00016645536183605456,
      "loss": 2.0093,
      "step": 149036
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.1649986505508423,
      "learning_rate": 0.0001664516989275085,
      "loss": 1.9413,
      "step": 149037
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.1028220653533936,
      "learning_rate": 0.00016644803604379173,
      "loss": 2.3547,
      "step": 149038
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.201191782951355,
      "learning_rate": 0.00016644437318490477,
      "loss": 1.9561,
      "step": 149039
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.2566052675247192,
      "learning_rate": 0.0001664407103508484,
      "loss": 2.2629,
      "step": 149040
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.0793840885162354,
      "learning_rate": 0.000166437047541623,
      "loss": 1.9265,
      "step": 149041
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.0602368116378784,
      "learning_rate": 0.00016643338475722964,
      "loss": 2.2216,
      "step": 149042
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.0701078176498413,
      "learning_rate": 0.0001664297219976687,
      "loss": 2.0548,
      "step": 149043
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.1470781564712524,
      "learning_rate": 0.00016642605926294113,
      "loss": 2.0677,
      "step": 149044
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.1084885597229004,
      "learning_rate": 0.00016642239655304748,
      "loss": 2.1259,
      "step": 149045
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.1347218751907349,
      "learning_rate": 0.0001664187338679884,
      "loss": 2.0534,
      "step": 149046
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.1612308025360107,
      "learning_rate": 0.00016641507120776454,
      "loss": 2.1331,
      "step": 149047
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.0936553478240967,
      "learning_rate": 0.0001664114085723767,
      "loss": 2.1216,
      "step": 149048
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.1956071853637695,
      "learning_rate": 0.0001664077459618254,
      "loss": 1.9604,
      "step": 149049
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.1298943758010864,
      "learning_rate": 0.00016640408337611156,
      "loss": 2.1074,
      "step": 149050
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.2478679418563843,
      "learning_rate": 0.0001664004208152357,
      "loss": 2.0681,
      "step": 149051
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.0455564260482788,
      "learning_rate": 0.00016639675827919852,
      "loss": 2.0209,
      "step": 149052
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.1730149984359741,
      "learning_rate": 0.0001663930957680006,
      "loss": 2.0338,
      "step": 149053
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.1556487083435059,
      "learning_rate": 0.00016638943328164285,
      "loss": 2.269,
      "step": 149054
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.2429523468017578,
      "learning_rate": 0.00016638577082012568,
      "loss": 1.9607,
      "step": 149055
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.1462115049362183,
      "learning_rate": 0.00016638210838345008,
      "loss": 2.0599,
      "step": 149056
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.0405081510543823,
      "learning_rate": 0.00016637844597161653,
      "loss": 2.1772,
      "step": 149057
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.1489428281784058,
      "learning_rate": 0.00016637478358462574,
      "loss": 2.1257,
      "step": 149058
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.1881518363952637,
      "learning_rate": 0.0001663711212224783,
      "loss": 2.0436,
      "step": 149059
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.0459997653961182,
      "learning_rate": 0.0001663674588851751,
      "loss": 2.1871,
      "step": 149060
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.0894222259521484,
      "learning_rate": 0.00016636379657271663,
      "loss": 1.9978,
      "step": 149061
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.2133655548095703,
      "learning_rate": 0.00016636013428510376,
      "loss": 2.196,
      "step": 149062
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.1291097402572632,
      "learning_rate": 0.00016635647202233703,
      "loss": 2.0861,
      "step": 149063
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.1208866834640503,
      "learning_rate": 0.00016635280978441718,
      "loss": 2.0517,
      "step": 149064
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.173791527748108,
      "learning_rate": 0.00016634914757134473,
      "loss": 1.9645,
      "step": 149065
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.1962329149246216,
      "learning_rate": 0.00016634548538312064,
      "loss": 1.9007,
      "step": 149066
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.254090666770935,
      "learning_rate": 0.0001663418232197453,
      "loss": 2.0019,
      "step": 149067
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.1328524351119995,
      "learning_rate": 0.00016633816108121966,
      "loss": 1.9054,
      "step": 149068
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.1406246423721313,
      "learning_rate": 0.0001663344989675442,
      "loss": 1.9477,
      "step": 149069
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.1963858604431152,
      "learning_rate": 0.00016633083687871988,
      "loss": 2.1873,
      "step": 149070
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.1443142890930176,
      "learning_rate": 0.00016632717481474696,
      "loss": 2.0689,
      "step": 149071
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.1576688289642334,
      "learning_rate": 0.00016632351277562644,
      "loss": 1.9198,
      "step": 149072
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.318636178970337,
      "learning_rate": 0.0001663198507613588,
      "loss": 2.0027,
      "step": 149073
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.19364595413208,
      "learning_rate": 0.00016631618877194495,
      "loss": 1.9815,
      "step": 149074
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.086249589920044,
      "learning_rate": 0.00016631252680738532,
      "loss": 2.0747,
      "step": 149075
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.3459504842758179,
      "learning_rate": 0.0001663088648676808,
      "loss": 2.2008,
      "step": 149076
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.052298903465271,
      "learning_rate": 0.00016630520295283203,
      "loss": 1.9533,
      "step": 149077
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.1364762783050537,
      "learning_rate": 0.0001663015410628396,
      "loss": 2.1016,
      "step": 149078
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.1026240587234497,
      "learning_rate": 0.00016629787919770415,
      "loss": 1.9682,
      "step": 149079
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.0797394514083862,
      "learning_rate": 0.00016629421735742656,
      "loss": 2.07,
      "step": 149080
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.1200624704360962,
      "learning_rate": 0.00016629055554200732,
      "loss": 2.2738,
      "step": 149081
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.1257297992706299,
      "learning_rate": 0.00016628689375144725,
      "loss": 2.0237,
      "step": 149082
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.0910860300064087,
      "learning_rate": 0.000166283231985747,
      "loss": 2.1485,
      "step": 149083
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.3152854442596436,
      "learning_rate": 0.0001662795702449071,
      "loss": 2.1932,
      "step": 149084
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.012032151222229,
      "learning_rate": 0.00016627590852892846,
      "loss": 2.1497,
      "step": 149085
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.2035473585128784,
      "learning_rate": 0.00016627224683781166,
      "loss": 2.1049,
      "step": 149086
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.070120096206665,
      "learning_rate": 0.00016626858517155725,
      "loss": 2.1609,
      "step": 149087
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.2035393714904785,
      "learning_rate": 0.00016626492353016617,
      "loss": 2.1221,
      "step": 149088
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.1157519817352295,
      "learning_rate": 0.00016626126191363896,
      "loss": 2.1706,
      "step": 149089
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.0109459161758423,
      "learning_rate": 0.00016625760032197616,
      "loss": 2.1346,
      "step": 149090
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.2356587648391724,
      "learning_rate": 0.00016625393875517873,
      "loss": 1.9605,
      "step": 149091
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.0931799411773682,
      "learning_rate": 0.0001662502772132472,
      "loss": 2.1905,
      "step": 149092
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.1343128681182861,
      "learning_rate": 0.0001662466156961822,
      "loss": 1.9352,
      "step": 149093
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.312137484550476,
      "learning_rate": 0.00016624295420398456,
      "loss": 2.1264,
      "step": 149094
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.1447142362594604,
      "learning_rate": 0.0001662392927366549,
      "loss": 2.2432,
      "step": 149095
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.1800812482833862,
      "learning_rate": 0.00016623563129419374,
      "loss": 2.0281,
      "step": 149096
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.156885027885437,
      "learning_rate": 0.00016623196987660205,
      "loss": 2.08,
      "step": 149097
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.0059027671813965,
      "learning_rate": 0.0001662283084838802,
      "loss": 2.4287,
      "step": 149098
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.277162790298462,
      "learning_rate": 0.0001662246471160292,
      "loss": 2.2108,
      "step": 149099
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.1072067022323608,
      "learning_rate": 0.00016622098577304952,
      "loss": 2.1363,
      "step": 149100
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.1564520597457886,
      "learning_rate": 0.0001662173244549419,
      "loss": 2.0129,
      "step": 149101
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.1465004682540894,
      "learning_rate": 0.0001662136631617069,
      "loss": 2.0706,
      "step": 149102
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.0761475563049316,
      "learning_rate": 0.00016621000189334542,
      "loss": 1.9462,
      "step": 149103
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.1772478818893433,
      "learning_rate": 0.0001662063406498579,
      "loss": 2.1934,
      "step": 149104
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.132214903831482,
      "learning_rate": 0.00016620267943124524,
      "loss": 2.1687,
      "step": 149105
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.2425376176834106,
      "learning_rate": 0.00016619901823750806,
      "loss": 2.2843,
      "step": 149106
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.1921463012695312,
      "learning_rate": 0.000166195357068647,
      "loss": 2.2552,
      "step": 149107
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.3327933549880981,
      "learning_rate": 0.00016619169592466265,
      "loss": 2.1339,
      "step": 149108
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.1190040111541748,
      "learning_rate": 0.0001661880348055559,
      "loss": 2.112,
      "step": 149109
    },
    {
      "epoch": 1.94,
      "grad_norm": 0.9925463199615479,
      "learning_rate": 0.00016618437371132718,
      "loss": 2.1567,
      "step": 149110
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.0905689001083374,
      "learning_rate": 0.00016618071264197742,
      "loss": 2.1713,
      "step": 149111
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.08355712890625,
      "learning_rate": 0.0001661770515975072,
      "loss": 2.0549,
      "step": 149112
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.042109489440918,
      "learning_rate": 0.00016617339057791724,
      "loss": 1.8335,
      "step": 149113
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.059340476989746,
      "learning_rate": 0.000166169729583208,
      "loss": 2.2114,
      "step": 149114
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.0698636770248413,
      "learning_rate": 0.00016616606861338047,
      "loss": 1.9621,
      "step": 149115
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.1261036396026611,
      "learning_rate": 0.0001661624076684351,
      "loss": 2.0875,
      "step": 149116
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.0897023677825928,
      "learning_rate": 0.00016615874674837274,
      "loss": 1.9576,
      "step": 149117
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.0932071208953857,
      "learning_rate": 0.000166155085853194,
      "loss": 2.0722,
      "step": 149118
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.1258739233016968,
      "learning_rate": 0.00016615142498289958,
      "loss": 2.1399,
      "step": 149119
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.214683175086975,
      "learning_rate": 0.00016614776413749,
      "loss": 2.2562,
      "step": 149120
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.1988531351089478,
      "learning_rate": 0.00016614410331696622,
      "loss": 2.0044,
      "step": 149121
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.1428850889205933,
      "learning_rate": 0.00016614044252132864,
      "loss": 1.8949,
      "step": 149122
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.1545312404632568,
      "learning_rate": 0.0001661367817505782,
      "loss": 2.0925,
      "step": 149123
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.0790780782699585,
      "learning_rate": 0.00016613312100471547,
      "loss": 1.9025,
      "step": 149124
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.1400688886642456,
      "learning_rate": 0.00016612946028374113,
      "loss": 2.2058,
      "step": 149125
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.2720801830291748,
      "learning_rate": 0.0001661257995876557,
      "loss": 2.1258,
      "step": 149126
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.1192177534103394,
      "learning_rate": 0.00016612213891646017,
      "loss": 1.9625,
      "step": 149127
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.0551509857177734,
      "learning_rate": 0.00016611847827015488,
      "loss": 1.8794,
      "step": 149128
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.1699943542480469,
      "learning_rate": 0.00016611481764874084,
      "loss": 1.9131,
      "step": 149129
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.071945071220398,
      "learning_rate": 0.00016611115705221857,
      "loss": 2.0186,
      "step": 149130
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.0059608221054077,
      "learning_rate": 0.0001661074964805888,
      "loss": 2.1498,
      "step": 149131
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.178276538848877,
      "learning_rate": 0.00016610383593385202,
      "loss": 2.1399,
      "step": 149132
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.1437278985977173,
      "learning_rate": 0.0001661001754120092,
      "loss": 1.8997,
      "step": 149133
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.122969150543213,
      "learning_rate": 0.00016609651491506077,
      "loss": 2.0129,
      "step": 149134
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.0998139381408691,
      "learning_rate": 0.00016609285444300765,
      "loss": 1.9144,
      "step": 149135
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.0776678323745728,
      "learning_rate": 0.00016608919399585027,
      "loss": 1.9035,
      "step": 149136
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.5275615453720093,
      "learning_rate": 0.0001660855335735897,
      "loss": 1.9537,
      "step": 149137
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.1153326034545898,
      "learning_rate": 0.00016608187317622606,
      "loss": 2.129,
      "step": 149138
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.087941288948059,
      "learning_rate": 0.00016607821280376048,
      "loss": 2.1961,
      "step": 149139
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.0647367238998413,
      "learning_rate": 0.00016607455245619336,
      "loss": 1.9849,
      "step": 149140
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.1604870557785034,
      "learning_rate": 0.00016607089213352562,
      "loss": 2.1378,
      "step": 149141
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.1432265043258667,
      "learning_rate": 0.00016606723183575775,
      "loss": 2.0081,
      "step": 149142
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.3499763011932373,
      "learning_rate": 0.00016606357156289072,
      "loss": 2.2509,
      "step": 149143
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.0841751098632812,
      "learning_rate": 0.00016605991131492476,
      "loss": 2.1979,
      "step": 149144
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.1534292697906494,
      "learning_rate": 0.00016605625109186094,
      "loss": 2.3011,
      "step": 149145
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.2188538312911987,
      "learning_rate": 0.00016605259089369964,
      "loss": 2.1728,
      "step": 149146
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.1403045654296875,
      "learning_rate": 0.0001660489307204418,
      "loss": 2.0663,
      "step": 149147
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.1949164867401123,
      "learning_rate": 0.00016604527057208793,
      "loss": 2.4295,
      "step": 149148
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.1706323623657227,
      "learning_rate": 0.00016604161044863892,
      "loss": 2.0259,
      "step": 149149
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.0951385498046875,
      "learning_rate": 0.00016603795035009515,
      "loss": 2.0996,
      "step": 149150
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.1285400390625,
      "learning_rate": 0.0001660342902764575,
      "loss": 2.0642,
      "step": 149151
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.132424235343933,
      "learning_rate": 0.00016603063022772657,
      "loss": 1.9864,
      "step": 149152
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.1673113107681274,
      "learning_rate": 0.00016602697020390314,
      "loss": 2.0859,
      "step": 149153
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.0110048055648804,
      "learning_rate": 0.0001660233102049877,
      "loss": 1.9522,
      "step": 149154
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.1446878910064697,
      "learning_rate": 0.0001660196502309813,
      "loss": 2.1324,
      "step": 149155
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.0894707441329956,
      "learning_rate": 0.00016601599028188415,
      "loss": 2.0679,
      "step": 149156
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.246594786643982,
      "learning_rate": 0.0001660123303576973,
      "loss": 2.0085,
      "step": 149157
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.0783523321151733,
      "learning_rate": 0.00016600867045842114,
      "loss": 1.9651,
      "step": 149158
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.183952808380127,
      "learning_rate": 0.00016600501058405663,
      "loss": 1.95,
      "step": 149159
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.0865538120269775,
      "learning_rate": 0.00016600135073460416,
      "loss": 2.2576,
      "step": 149160
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.061466097831726,
      "learning_rate": 0.0001659976909100647,
      "loss": 2.1858,
      "step": 149161
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.0091009140014648,
      "learning_rate": 0.0001659940311104388,
      "loss": 1.9116,
      "step": 149162
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.1734950542449951,
      "learning_rate": 0.0001659903713357272,
      "loss": 2.086,
      "step": 149163
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.197653889656067,
      "learning_rate": 0.0001659867115859303,
      "loss": 2.0986,
      "step": 149164
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.2741936445236206,
      "learning_rate": 0.0001659830518610492,
      "loss": 2.065,
      "step": 149165
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.1345851421356201,
      "learning_rate": 0.00016597939216108422,
      "loss": 1.9424,
      "step": 149166
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.0661967992782593,
      "learning_rate": 0.00016597573248603635,
      "loss": 2.1069,
      "step": 149167
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.095435380935669,
      "learning_rate": 0.00016597207283590608,
      "loss": 2.0541,
      "step": 149168
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.1283130645751953,
      "learning_rate": 0.00016596841321069413,
      "loss": 2.045,
      "step": 149169
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.1274521350860596,
      "learning_rate": 0.00016596475361040113,
      "loss": 2.247,
      "step": 149170
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.2335292100906372,
      "learning_rate": 0.00016596109403502787,
      "loss": 2.0361,
      "step": 149171
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.1502410173416138,
      "learning_rate": 0.00016595743448457486,
      "loss": 2.1464,
      "step": 149172
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.3149691820144653,
      "learning_rate": 0.00016595377495904305,
      "loss": 2.0909,
      "step": 149173
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.2731595039367676,
      "learning_rate": 0.00016595011545843293,
      "loss": 2.0043,
      "step": 149174
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.1769745349884033,
      "learning_rate": 0.0001659464559827451,
      "loss": 2.0552,
      "step": 149175
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.1378376483917236,
      "learning_rate": 0.0001659427965319805,
      "loss": 2.1442,
      "step": 149176
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.1971244812011719,
      "learning_rate": 0.00016593913710613963,
      "loss": 2.168,
      "step": 149177
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.238351821899414,
      "learning_rate": 0.00016593547770522312,
      "loss": 2.1025,
      "step": 149178
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.2781370878219604,
      "learning_rate": 0.00016593181832923182,
      "loss": 1.967,
      "step": 149179
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.1623859405517578,
      "learning_rate": 0.00016592815897816633,
      "loss": 2.2142,
      "step": 149180
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.1800754070281982,
      "learning_rate": 0.0001659244996520272,
      "loss": 2.2033,
      "step": 149181
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.2006639242172241,
      "learning_rate": 0.00016592084035081537,
      "loss": 1.8642,
      "step": 149182
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.1887096166610718,
      "learning_rate": 0.00016591718107453128,
      "loss": 2.1174,
      "step": 149183
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.1977320909500122,
      "learning_rate": 0.00016591352182317582,
      "loss": 2.1269,
      "step": 149184
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.0646530389785767,
      "learning_rate": 0.0001659098625967496,
      "loss": 1.8589,
      "step": 149185
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.144789218902588,
      "learning_rate": 0.0001659062033952532,
      "loss": 1.8159,
      "step": 149186
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.1124508380889893,
      "learning_rate": 0.00016590254421868728,
      "loss": 2.1883,
      "step": 149187
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.1322695016860962,
      "learning_rate": 0.00016589888506705273,
      "loss": 2.0622,
      "step": 149188
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.2461072206497192,
      "learning_rate": 0.00016589522594035003,
      "loss": 1.8258,
      "step": 149189
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.1497547626495361,
      "learning_rate": 0.00016589156683858002,
      "loss": 1.9871,
      "step": 149190
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.3843891620635986,
      "learning_rate": 0.00016588790776174328,
      "loss": 1.8362,
      "step": 149191
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.565258502960205,
      "learning_rate": 0.00016588424870984055,
      "loss": 2.1026,
      "step": 149192
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.0607489347457886,
      "learning_rate": 0.00016588058968287233,
      "loss": 2.0054,
      "step": 149193
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.09200918674469,
      "learning_rate": 0.00016587693068083955,
      "loss": 1.9337,
      "step": 149194
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.2196458578109741,
      "learning_rate": 0.0001658732717037427,
      "loss": 2.0352,
      "step": 149195
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.343989372253418,
      "learning_rate": 0.0001658696127515826,
      "loss": 2.3855,
      "step": 149196
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.2276325225830078,
      "learning_rate": 0.0001658659538243599,
      "loss": 2.0511,
      "step": 149197
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.0778234004974365,
      "learning_rate": 0.00016586229492207528,
      "loss": 1.7644,
      "step": 149198
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.22886061668396,
      "learning_rate": 0.00016585863604472924,
      "loss": 2.154,
      "step": 149199
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.209040641784668,
      "learning_rate": 0.00016585497719232272,
      "loss": 2.2239,
      "step": 149200
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.1244361400604248,
      "learning_rate": 0.0001658513183648562,
      "loss": 2.1106,
      "step": 149201
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.044249415397644,
      "learning_rate": 0.00016584765956233056,
      "loss": 2.1059,
      "step": 149202
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.1692887544631958,
      "learning_rate": 0.00016584400078474627,
      "loss": 2.2124,
      "step": 149203
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.2075223922729492,
      "learning_rate": 0.00016584034203210435,
      "loss": 2.2051,
      "step": 149204
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.2366570234298706,
      "learning_rate": 0.00016583668330440496,
      "loss": 2.1823,
      "step": 149205
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.0612068176269531,
      "learning_rate": 0.00016583302460164922,
      "loss": 2.0392,
      "step": 149206
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.062125325202942,
      "learning_rate": 0.00016582936592383752,
      "loss": 2.116,
      "step": 149207
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.6010146141052246,
      "learning_rate": 0.0001658257072709708,
      "loss": 2.2351,
      "step": 149208
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.343011498451233,
      "learning_rate": 0.0001658220486430495,
      "loss": 1.993,
      "step": 149209
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.050162672996521,
      "learning_rate": 0.00016581839004007468,
      "loss": 1.8861,
      "step": 149210
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.1853103637695312,
      "learning_rate": 0.00016581473146204647,
      "loss": 2.1529,
      "step": 149211
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.1480921506881714,
      "learning_rate": 0.00016581107290896595,
      "loss": 1.9919,
      "step": 149212
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.278460144996643,
      "learning_rate": 0.00016580741438083357,
      "loss": 2.1025,
      "step": 149213
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.2045270204544067,
      "learning_rate": 0.00016580375587765028,
      "loss": 2.0879,
      "step": 149214
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.1666470766067505,
      "learning_rate": 0.00016580009739941643,
      "loss": 1.9287,
      "step": 149215
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.1744201183319092,
      "learning_rate": 0.00016579643894613317,
      "loss": 1.9717,
      "step": 149216
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.2591798305511475,
      "learning_rate": 0.0001657927805178006,
      "loss": 2.1572,
      "step": 149217
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.286976933479309,
      "learning_rate": 0.00016578912211441982,
      "loss": 2.1736,
      "step": 149218
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.0907310247421265,
      "learning_rate": 0.00016578546373599128,
      "loss": 2.175,
      "step": 149219
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.0433000326156616,
      "learning_rate": 0.00016578180538251588,
      "loss": 1.7878,
      "step": 149220
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.1754885911941528,
      "learning_rate": 0.000165778147053994,
      "loss": 1.9906,
      "step": 149221
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.2000041007995605,
      "learning_rate": 0.00016577448875042678,
      "loss": 2.1732,
      "step": 149222
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.1847598552703857,
      "learning_rate": 0.00016577083047181437,
      "loss": 2.1813,
      "step": 149223
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.085386872291565,
      "learning_rate": 0.0001657671722181578,
      "loss": 2.0225,
      "step": 149224
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.079276204109192,
      "learning_rate": 0.00016576351398945753,
      "loss": 2.1282,
      "step": 149225
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.3548290729522705,
      "learning_rate": 0.0001657598557857145,
      "loss": 2.1322,
      "step": 149226
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.101948618888855,
      "learning_rate": 0.00016575619760692911,
      "loss": 2.2441,
      "step": 149227
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.083339810371399,
      "learning_rate": 0.0001657525394531024,
      "loss": 1.9997,
      "step": 149228
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.0930774211883545,
      "learning_rate": 0.00016574888132423463,
      "loss": 2.1097,
      "step": 149229
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.0653480291366577,
      "learning_rate": 0.00016574522322032676,
      "loss": 2.2171,
      "step": 149230
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.067168116569519,
      "learning_rate": 0.0001657415651413793,
      "loss": 2.2707,
      "step": 149231
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.1346936225891113,
      "learning_rate": 0.0001657379070873931,
      "loss": 2.0659,
      "step": 149232
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.0118736028671265,
      "learning_rate": 0.00016573424905836866,
      "loss": 2.09,
      "step": 149233
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.176580786705017,
      "learning_rate": 0.00016573059105430698,
      "loss": 2.0419,
      "step": 149234
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.1436296701431274,
      "learning_rate": 0.00016572693307520832,
      "loss": 2.0704,
      "step": 149235
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.153772234916687,
      "learning_rate": 0.00016572327512107363,
      "loss": 2.1141,
      "step": 149236
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.158947467803955,
      "learning_rate": 0.00016571961719190342,
      "loss": 1.8778,
      "step": 149237
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.2328444719314575,
      "learning_rate": 0.00016571595928769857,
      "loss": 2.0401,
      "step": 149238
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.142118215560913,
      "learning_rate": 0.00016571230140845957,
      "loss": 1.9949,
      "step": 149239
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.2019445896148682,
      "learning_rate": 0.00016570864355418738,
      "loss": 1.933,
      "step": 149240
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.1512020826339722,
      "learning_rate": 0.0001657049857248823,
      "loss": 2.1548,
      "step": 149241
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.1943646669387817,
      "learning_rate": 0.00016570132792054532,
      "loss": 2.4339,
      "step": 149242
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.1984262466430664,
      "learning_rate": 0.00016569767014117684,
      "loss": 2.1685,
      "step": 149243
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.2327073812484741,
      "learning_rate": 0.00016569401238677783,
      "loss": 2.0333,
      "step": 149244
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.1544331312179565,
      "learning_rate": 0.0001656903546573487,
      "loss": 2.2545,
      "step": 149245
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.046106219291687,
      "learning_rate": 0.00016568669695289043,
      "loss": 2.168,
      "step": 149246
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.1169847249984741,
      "learning_rate": 0.0001656830392734035,
      "loss": 2.4016,
      "step": 149247
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.0324522256851196,
      "learning_rate": 0.00016567938161888868,
      "loss": 2.1489,
      "step": 149248
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.0698546171188354,
      "learning_rate": 0.00016567572398934642,
      "loss": 1.9584,
      "step": 149249
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.0920606851577759,
      "learning_rate": 0.00016567206638477776,
      "loss": 1.8206,
      "step": 149250
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.1219981908798218,
      "learning_rate": 0.000165668408805183,
      "loss": 2.1102,
      "step": 149251
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.1833995580673218,
      "learning_rate": 0.0001656647512505632,
      "loss": 1.9347,
      "step": 149252
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.0726399421691895,
      "learning_rate": 0.00016566109372091883,
      "loss": 2.0113,
      "step": 149253
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.2024893760681152,
      "learning_rate": 0.00016565743621625058,
      "loss": 2.0632,
      "step": 149254
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.1210371255874634,
      "learning_rate": 0.00016565377873655908,
      "loss": 1.9225,
      "step": 149255
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.0694583654403687,
      "learning_rate": 0.00016565012128184515,
      "loss": 2.071,
      "step": 149256
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.0800529718399048,
      "learning_rate": 0.00016564646385210932,
      "loss": 2.2095,
      "step": 149257
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.0949735641479492,
      "learning_rate": 0.00016564280644735244,
      "loss": 2.1083,
      "step": 149258
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.0715539455413818,
      "learning_rate": 0.00016563914906757513,
      "loss": 2.1139,
      "step": 149259
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.1889020204544067,
      "learning_rate": 0.00016563549171277789,
      "loss": 2.1369,
      "step": 149260
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.098327875137329,
      "learning_rate": 0.00016563183438296166,
      "loss": 2.0111,
      "step": 149261
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.1112757921218872,
      "learning_rate": 0.00016562817707812706,
      "loss": 2.2437,
      "step": 149262
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.1021462678909302,
      "learning_rate": 0.00016562451979827456,
      "loss": 2.1579,
      "step": 149263
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.3296946287155151,
      "learning_rate": 0.00016562086254340508,
      "loss": 2.1339,
      "step": 149264
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.119368076324463,
      "learning_rate": 0.0001656172053135193,
      "loss": 2.1374,
      "step": 149265
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.1844854354858398,
      "learning_rate": 0.00016561354810861763,
      "loss": 2.064,
      "step": 149266
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.044089913368225,
      "learning_rate": 0.00016560989092870111,
      "loss": 2.2175,
      "step": 149267
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.125488519668579,
      "learning_rate": 0.00016560623377377012,
      "loss": 2.051,
      "step": 149268
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.2276557683944702,
      "learning_rate": 0.00016560257664382556,
      "loss": 2.0881,
      "step": 149269
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.107940435409546,
      "learning_rate": 0.00016559891953886805,
      "loss": 2.0346,
      "step": 149270
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.2101377248764038,
      "learning_rate": 0.0001655952624588982,
      "loss": 2.2473,
      "step": 149271
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.0743950605392456,
      "learning_rate": 0.0001655916054039166,
      "loss": 2.094,
      "step": 149272
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.1497459411621094,
      "learning_rate": 0.00016558794837392423,
      "loss": 1.9924,
      "step": 149273
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.0779861211776733,
      "learning_rate": 0.00016558429136892143,
      "loss": 1.907,
      "step": 149274
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.1265357732772827,
      "learning_rate": 0.0001655806343889092,
      "loss": 2.1235,
      "step": 149275
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.0930571556091309,
      "learning_rate": 0.00016557697743388807,
      "loss": 2.2981,
      "step": 149276
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.1054542064666748,
      "learning_rate": 0.00016557332050385866,
      "loss": 2.1305,
      "step": 149277
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.0899146795272827,
      "learning_rate": 0.00016556966359882166,
      "loss": 2.1204,
      "step": 149278
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.1020060777664185,
      "learning_rate": 0.0001655660067187779,
      "loss": 2.3765,
      "step": 149279
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.2185828685760498,
      "learning_rate": 0.00016556234986372782,
      "loss": 2.1276,
      "step": 149280
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.3901668787002563,
      "learning_rate": 0.00016555869303367236,
      "loss": 1.843,
      "step": 149281
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.184731364250183,
      "learning_rate": 0.00016555503622861194,
      "loss": 2.1423,
      "step": 149282
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.2684394121170044,
      "learning_rate": 0.00016555137944854767,
      "loss": 2.2146,
      "step": 149283
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.0926029682159424,
      "learning_rate": 0.00016554772269347965,
      "loss": 1.9612,
      "step": 149284
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.1134085655212402,
      "learning_rate": 0.00016554406596340897,
      "loss": 1.9353,
      "step": 149285
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.0257741212844849,
      "learning_rate": 0.0001655404092583361,
      "loss": 2.098,
      "step": 149286
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.1812782287597656,
      "learning_rate": 0.0001655367525782619,
      "loss": 2.1112,
      "step": 149287
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.190139651298523,
      "learning_rate": 0.0001655330959231868,
      "loss": 2.0732,
      "step": 149288
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.099364161491394,
      "learning_rate": 0.00016552943929311193,
      "loss": 1.9208,
      "step": 149289
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.1597124338150024,
      "learning_rate": 0.00016552578268803743,
      "loss": 2.2731,
      "step": 149290
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.0667108297348022,
      "learning_rate": 0.0001655221261079643,
      "loss": 1.8994,
      "step": 149291
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.1778355836868286,
      "learning_rate": 0.00016551846955289306,
      "loss": 1.9778,
      "step": 149292
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.1240085363388062,
      "learning_rate": 0.0001655148130228246,
      "loss": 1.9559,
      "step": 149293
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.0020906925201416,
      "learning_rate": 0.00016551115651775936,
      "loss": 1.9443,
      "step": 149294
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.0733866691589355,
      "learning_rate": 0.00016550750003769838,
      "loss": 2.1398,
      "step": 149295
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.1896671056747437,
      "learning_rate": 0.00016550384358264181,
      "loss": 2.0717,
      "step": 149296
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.239820957183838,
      "learning_rate": 0.00016550018715259077,
      "loss": 2.0529,
      "step": 149297
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.0584555864334106,
      "learning_rate": 0.00016549653074754563,
      "loss": 2.1103,
      "step": 149298
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.0286800861358643,
      "learning_rate": 0.0001654928743675074,
      "loss": 2.0851,
      "step": 149299
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.0947861671447754,
      "learning_rate": 0.0001654892180124764,
      "loss": 2.1123,
      "step": 149300
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.227593183517456,
      "learning_rate": 0.0001654855616824538,
      "loss": 2.1928,
      "step": 149301
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.433427095413208,
      "learning_rate": 0.0001654819053774397,
      "loss": 2.0664,
      "step": 149302
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.214788556098938,
      "learning_rate": 0.00016547824909743513,
      "loss": 2.3114,
      "step": 149303
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.220332145690918,
      "learning_rate": 0.00016547459284244065,
      "loss": 1.8607,
      "step": 149304
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.239726185798645,
      "learning_rate": 0.00016547093661245708,
      "loss": 1.8861,
      "step": 149305
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.5134090185165405,
      "learning_rate": 0.00016546728040748484,
      "loss": 2.0343,
      "step": 149306
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.0682432651519775,
      "learning_rate": 0.00016546362422752503,
      "loss": 1.9895,
      "step": 149307
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.0712772607803345,
      "learning_rate": 0.00016545996807257785,
      "loss": 1.9317,
      "step": 149308
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.2304028272628784,
      "learning_rate": 0.00016545631194264435,
      "loss": 2.0857,
      "step": 149309
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.1464390754699707,
      "learning_rate": 0.00016545265583772488,
      "loss": 2.2137,
      "step": 149310
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.1160314083099365,
      "learning_rate": 0.00016544899975782045,
      "loss": 2.247,
      "step": 149311
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.1603577136993408,
      "learning_rate": 0.0001654453437029315,
      "loss": 1.7866,
      "step": 149312
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.1071813106536865,
      "learning_rate": 0.000165441687673059,
      "loss": 1.8391,
      "step": 149313
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.0912660360336304,
      "learning_rate": 0.0001654380316682032,
      "loss": 2.2119,
      "step": 149314
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.0992554426193237,
      "learning_rate": 0.00016543437568836513,
      "loss": 1.7876,
      "step": 149315
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.1264595985412598,
      "learning_rate": 0.00016543071973354522,
      "loss": 2.004,
      "step": 149316
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.1551820039749146,
      "learning_rate": 0.0001654270638037444,
      "loss": 1.9984,
      "step": 149317
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.107665777206421,
      "learning_rate": 0.0001654234078989631,
      "loss": 2.0606,
      "step": 149318
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.2080425024032593,
      "learning_rate": 0.0001654197520192024,
      "loss": 2.1606,
      "step": 149319
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.0213299989700317,
      "learning_rate": 0.00016541609616446245,
      "loss": 2.0342,
      "step": 149320
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.0149474143981934,
      "learning_rate": 0.0001654124403347443,
      "loss": 2.003,
      "step": 149321
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.0417957305908203,
      "learning_rate": 0.00016540878453004843,
      "loss": 1.9566,
      "step": 149322
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.1853874921798706,
      "learning_rate": 0.0001654051287503757,
      "loss": 2.1033,
      "step": 149323
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.0342800617218018,
      "learning_rate": 0.00016540147299572662,
      "loss": 1.994,
      "step": 149324
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.2080646753311157,
      "learning_rate": 0.00016539781726610214,
      "loss": 2.003,
      "step": 149325
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.067224144935608,
      "learning_rate": 0.00016539416156150248,
      "loss": 2.1202,
      "step": 149326
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.1361000537872314,
      "learning_rate": 0.00016539050588192878,
      "loss": 2.0031,
      "step": 149327
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.1319010257720947,
      "learning_rate": 0.00016538685022738137,
      "loss": 2.2499,
      "step": 149328
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.126193642616272,
      "learning_rate": 0.0001653831945978612,
      "loss": 2.0989,
      "step": 149329
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.1634641885757446,
      "learning_rate": 0.00016537953899336877,
      "loss": 1.8699,
      "step": 149330
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.294478416442871,
      "learning_rate": 0.00016537588341390496,
      "loss": 1.956,
      "step": 149331
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.136828064918518,
      "learning_rate": 0.00016537222785947012,
      "loss": 1.9631,
      "step": 149332
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.1489514112472534,
      "learning_rate": 0.00016536857233006528,
      "loss": 2.034,
      "step": 149333
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.0772374868392944,
      "learning_rate": 0.00016536491682569082,
      "loss": 2.0847,
      "step": 149334
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.2225141525268555,
      "learning_rate": 0.00016536126134634768,
      "loss": 2.0359,
      "step": 149335
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.031603455543518,
      "learning_rate": 0.0001653576058920363,
      "loss": 1.9042,
      "step": 149336
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.22607421875,
      "learning_rate": 0.0001653539504627576,
      "loss": 2.3649,
      "step": 149337
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.0614699125289917,
      "learning_rate": 0.0001653502950585122,
      "loss": 2.4099,
      "step": 149338
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.1248458623886108,
      "learning_rate": 0.00016534663967930066,
      "loss": 2.0296,
      "step": 149339
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.1578024625778198,
      "learning_rate": 0.0001653429843251236,
      "loss": 1.9747,
      "step": 149340
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.1549464464187622,
      "learning_rate": 0.00016533932899598196,
      "loss": 2.167,
      "step": 149341
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.325298547744751,
      "learning_rate": 0.00016533567369187614,
      "loss": 2.0138,
      "step": 149342
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.1564887762069702,
      "learning_rate": 0.0001653320184128071,
      "loss": 1.9362,
      "step": 149343
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.1485605239868164,
      "learning_rate": 0.0001653283631587754,
      "loss": 1.8849,
      "step": 149344
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.1870614290237427,
      "learning_rate": 0.00016532470792978153,
      "loss": 2.119,
      "step": 149345
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.0903339385986328,
      "learning_rate": 0.00016532105272582648,
      "loss": 2.2212,
      "step": 149346
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.0789251327514648,
      "learning_rate": 0.0001653173975469108,
      "loss": 2.1449,
      "step": 149347
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.050054669380188,
      "learning_rate": 0.00016531374239303503,
      "loss": 2.1142,
      "step": 149348
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.142541527748108,
      "learning_rate": 0.0001653100872642001,
      "loss": 2.0823,
      "step": 149349
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.0828889608383179,
      "learning_rate": 0.0001653064321604066,
      "loss": 2.212,
      "step": 149350
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.0858521461486816,
      "learning_rate": 0.000165302777081655,
      "loss": 2.1015,
      "step": 149351
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.1607353687286377,
      "learning_rate": 0.0001652991220279463,
      "loss": 2.0796,
      "step": 149352
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.1986647844314575,
      "learning_rate": 0.00016529546699928105,
      "loss": 2.1278,
      "step": 149353
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.0924863815307617,
      "learning_rate": 0.00016529181199565976,
      "loss": 2.0385,
      "step": 149354
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.1760698556900024,
      "learning_rate": 0.00016528815701708342,
      "loss": 2.1116,
      "step": 149355
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.1207200288772583,
      "learning_rate": 0.00016528450206355251,
      "loss": 2.0901,
      "step": 149356
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.1361241340637207,
      "learning_rate": 0.00016528084713506765,
      "loss": 2.3669,
      "step": 149357
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.3846758604049683,
      "learning_rate": 0.0001652771922316298,
      "loss": 2.0524,
      "step": 149358
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.119957447052002,
      "learning_rate": 0.00016527353735323929,
      "loss": 2.0267,
      "step": 149359
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.1625486612319946,
      "learning_rate": 0.0001652698824998971,
      "loss": 1.9539,
      "step": 149360
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.1968276500701904,
      "learning_rate": 0.00016526622767160378,
      "loss": 2.0218,
      "step": 149361
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.1431324481964111,
      "learning_rate": 0.00016526257286836004,
      "loss": 2.1687,
      "step": 149362
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.1901856660842896,
      "learning_rate": 0.00016525891809016639,
      "loss": 2.157,
      "step": 149363
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.0364497900009155,
      "learning_rate": 0.00016525526333702377,
      "loss": 2.0694,
      "step": 149364
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.1852967739105225,
      "learning_rate": 0.00016525160860893259,
      "loss": 2.1392,
      "step": 149365
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.0431723594665527,
      "learning_rate": 0.00016524795390589382,
      "loss": 1.8683,
      "step": 149366
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.253420352935791,
      "learning_rate": 0.0001652442992279079,
      "loss": 2.212,
      "step": 149367
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.1245671510696411,
      "learning_rate": 0.00016524064457497584,
      "loss": 2.0038,
      "step": 149368
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.0376720428466797,
      "learning_rate": 0.00016523698994709785,
      "loss": 2.1336,
      "step": 149369
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.2053205966949463,
      "learning_rate": 0.00016523333534427495,
      "loss": 2.1163,
      "step": 149370
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.0621932744979858,
      "learning_rate": 0.0001652296807665076,
      "loss": 2.0573,
      "step": 149371
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.2655553817749023,
      "learning_rate": 0.00016522602621379674,
      "loss": 2.2117,
      "step": 149372
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.188645601272583,
      "learning_rate": 0.00016522237168614277,
      "loss": 2.2541,
      "step": 149373
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.0289262533187866,
      "learning_rate": 0.00016521871718354675,
      "loss": 2.1998,
      "step": 149374
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.3561561107635498,
      "learning_rate": 0.00016521506270600887,
      "loss": 2.3417,
      "step": 149375
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.0565391778945923,
      "learning_rate": 0.00016521140825353017,
      "loss": 2.1609,
      "step": 149376
    },
    {
      "epoch": 1.94,
      "grad_norm": 1.2202955484390259,
      "learning_rate": 0.00016520775382611108,
      "loss": 2.0491,
      "step": 149377
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.2092175483703613,
      "learning_rate": 0.00016520409942375256,
      "loss": 2.1045,
      "step": 149378
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.1394355297088623,
      "learning_rate": 0.000165200445046455,
      "loss": 2.2096,
      "step": 149379
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.2069354057312012,
      "learning_rate": 0.0001651967906942195,
      "loss": 2.1811,
      "step": 149380
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.1036944389343262,
      "learning_rate": 0.00016519313636704618,
      "loss": 2.1898,
      "step": 149381
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.110486388206482,
      "learning_rate": 0.00016518948206493614,
      "loss": 2.2137,
      "step": 149382
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.089043378829956,
      "learning_rate": 0.0001651858277878898,
      "loss": 2.0761,
      "step": 149383
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.2027537822723389,
      "learning_rate": 0.0001651821735359081,
      "loss": 2.0767,
      "step": 149384
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.1450684070587158,
      "learning_rate": 0.0001651785193089914,
      "loss": 2.0524,
      "step": 149385
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.0423557758331299,
      "learning_rate": 0.00016517486510714087,
      "loss": 1.9699,
      "step": 149386
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.1555742025375366,
      "learning_rate": 0.00016517121093035658,
      "loss": 2.1638,
      "step": 149387
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.0751593112945557,
      "learning_rate": 0.00016516755677863963,
      "loss": 1.718,
      "step": 149388
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.175096869468689,
      "learning_rate": 0.0001651639026519905,
      "loss": 1.9954,
      "step": 149389
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.05254065990448,
      "learning_rate": 0.00016516024855041004,
      "loss": 2.003,
      "step": 149390
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.0594754219055176,
      "learning_rate": 0.0001651565944738987,
      "loss": 2.1787,
      "step": 149391
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.3174407482147217,
      "learning_rate": 0.00016515294042245758,
      "loss": 2.0245,
      "step": 149392
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.1106960773468018,
      "learning_rate": 0.00016514928639608676,
      "loss": 1.9587,
      "step": 149393
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.2294660806655884,
      "learning_rate": 0.00016514563239478742,
      "loss": 1.9955,
      "step": 149394
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.1355799436569214,
      "learning_rate": 0.0001651419784185599,
      "loss": 2.1692,
      "step": 149395
    },
    {
      "epoch": 1.95,
      "grad_norm": 0.9899928569793701,
      "learning_rate": 0.00016513832446740518,
      "loss": 2.0614,
      "step": 149396
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.1846758127212524,
      "learning_rate": 0.0001651346705413236,
      "loss": 2.1066,
      "step": 149397
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.0259515047073364,
      "learning_rate": 0.00016513101664031635,
      "loss": 1.9041,
      "step": 149398
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.140073299407959,
      "learning_rate": 0.00016512736276438352,
      "loss": 2.1677,
      "step": 149399
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.1409300565719604,
      "learning_rate": 0.0001651237089135262,
      "loss": 2.1139,
      "step": 149400
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.2428282499313354,
      "learning_rate": 0.0001651200550877448,
      "loss": 2.1665,
      "step": 149401
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.1231592893600464,
      "learning_rate": 0.00016511640128704023,
      "loss": 2.0391,
      "step": 149402
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.4386045932769775,
      "learning_rate": 0.00016511274751141293,
      "loss": 2.1265,
      "step": 149403
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.1555564403533936,
      "learning_rate": 0.00016510909376086385,
      "loss": 1.9626,
      "step": 149404
    },
    {
      "epoch": 1.95,
      "grad_norm": 0.9735507965087891,
      "learning_rate": 0.00016510544003539354,
      "loss": 2.1956,
      "step": 149405
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.07571280002594,
      "learning_rate": 0.00016510178633500272,
      "loss": 1.8815,
      "step": 149406
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.3204981088638306,
      "learning_rate": 0.00016509813265969188,
      "loss": 2.2365,
      "step": 149407
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.1280328035354614,
      "learning_rate": 0.0001650944790094619,
      "loss": 2.0548,
      "step": 149408
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.1069931983947754,
      "learning_rate": 0.00016509082538431334,
      "loss": 2.1106,
      "step": 149409
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.0958356857299805,
      "learning_rate": 0.00016508717178424705,
      "loss": 2.1629,
      "step": 149410
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.0775871276855469,
      "learning_rate": 0.00016508351820926365,
      "loss": 1.831,
      "step": 149411
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.096020221710205,
      "learning_rate": 0.0001650798646593637,
      "loss": 2.3264,
      "step": 149412
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.1322715282440186,
      "learning_rate": 0.00016507621113454786,
      "loss": 1.9223,
      "step": 149413
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.116212010383606,
      "learning_rate": 0.00016507255763481703,
      "loss": 1.9384,
      "step": 149414
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.1795414686203003,
      "learning_rate": 0.0001650689041601716,
      "loss": 2.0672,
      "step": 149415
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.1422892808914185,
      "learning_rate": 0.00016506525071061255,
      "loss": 1.9238,
      "step": 149416
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.1143956184387207,
      "learning_rate": 0.00016506159728614043,
      "loss": 2.3601,
      "step": 149417
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.1766291856765747,
      "learning_rate": 0.0001650579438867559,
      "loss": 2.3428,
      "step": 149418
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.3894155025482178,
      "learning_rate": 0.0001650542905124595,
      "loss": 2.1069,
      "step": 149419
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.0347535610198975,
      "learning_rate": 0.00016505063716325223,
      "loss": 2.1546,
      "step": 149420
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.0359604358673096,
      "learning_rate": 0.00016504698383913442,
      "loss": 2.2551,
      "step": 149421
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.179713487625122,
      "learning_rate": 0.00016504333054010706,
      "loss": 2.2112,
      "step": 149422
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.1525297164916992,
      "learning_rate": 0.0001650396772661707,
      "loss": 2.1371,
      "step": 149423
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.2305042743682861,
      "learning_rate": 0.00016503602401732602,
      "loss": 1.9703,
      "step": 149424
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.2631953954696655,
      "learning_rate": 0.00016503237079357354,
      "loss": 2.0525,
      "step": 149425
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.3376528024673462,
      "learning_rate": 0.00016502871759491424,
      "loss": 2.2278,
      "step": 149426
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.0970526933670044,
      "learning_rate": 0.00016502506442134853,
      "loss": 1.9062,
      "step": 149427
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.0863319635391235,
      "learning_rate": 0.0001650214112728773,
      "loss": 1.8766,
      "step": 149428
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.0991151332855225,
      "learning_rate": 0.00016501775814950116,
      "loss": 2.107,
      "step": 149429
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.2153971195220947,
      "learning_rate": 0.0001650141050512206,
      "loss": 2.2876,
      "step": 149430
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.1315521001815796,
      "learning_rate": 0.00016501045197803667,
      "loss": 1.8222,
      "step": 149431
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.150092363357544,
      "learning_rate": 0.00016500679892994977,
      "loss": 1.9737,
      "step": 149432
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.1666260957717896,
      "learning_rate": 0.0001650031459069606,
      "loss": 1.9549,
      "step": 149433
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.0728628635406494,
      "learning_rate": 0.00016499949290906996,
      "loss": 2.1468,
      "step": 149434
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.330346941947937,
      "learning_rate": 0.0001649958399362785,
      "loss": 1.9486,
      "step": 149435
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.1683515310287476,
      "learning_rate": 0.0001649921869885867,
      "loss": 2.08,
      "step": 149436
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.1495193243026733,
      "learning_rate": 0.00016498853406599556,
      "loss": 1.9911,
      "step": 149437
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.1388434171676636,
      "learning_rate": 0.00016498488116850558,
      "loss": 1.9852,
      "step": 149438
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.2027512788772583,
      "learning_rate": 0.00016498122829611736,
      "loss": 2.1283,
      "step": 149439
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.0357517004013062,
      "learning_rate": 0.00016497757544883178,
      "loss": 1.9844,
      "step": 149440
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.041684627532959,
      "learning_rate": 0.0001649739226266494,
      "loss": 2.0617,
      "step": 149441
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.1435390710830688,
      "learning_rate": 0.0001649702698295708,
      "loss": 1.8645,
      "step": 149442
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.1730560064315796,
      "learning_rate": 0.00016496661705759692,
      "loss": 2.0346,
      "step": 149443
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.1136829853057861,
      "learning_rate": 0.00016496296431072816,
      "loss": 2.0361,
      "step": 149444
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.106555700302124,
      "learning_rate": 0.00016495931158896546,
      "loss": 2.071,
      "step": 149445
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.1424723863601685,
      "learning_rate": 0.0001649556588923094,
      "loss": 2.1329,
      "step": 149446
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.1437530517578125,
      "learning_rate": 0.0001649520062207606,
      "loss": 2.133,
      "step": 149447
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.1349468231201172,
      "learning_rate": 0.00016494835357431964,
      "loss": 2.2819,
      "step": 149448
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.2968264818191528,
      "learning_rate": 0.00016494470095298748,
      "loss": 1.958,
      "step": 149449
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.1353120803833008,
      "learning_rate": 0.0001649410483567645,
      "loss": 2.0054,
      "step": 149450
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.0585392713546753,
      "learning_rate": 0.00016493739578565166,
      "loss": 1.8811,
      "step": 149451
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.0620578527450562,
      "learning_rate": 0.00016493374323964951,
      "loss": 2.0605,
      "step": 149452
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.064436435699463,
      "learning_rate": 0.0001649300907187587,
      "loss": 2.1494,
      "step": 149453
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.1586146354675293,
      "learning_rate": 0.00016492643822297988,
      "loss": 2.2938,
      "step": 149454
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.3103500604629517,
      "learning_rate": 0.00016492278575231387,
      "loss": 2.0404,
      "step": 149455
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.114757776260376,
      "learning_rate": 0.00016491913330676116,
      "loss": 2.3868,
      "step": 149456
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.1809545755386353,
      "learning_rate": 0.0001649154808863226,
      "loss": 2.0939,
      "step": 149457
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.2231082916259766,
      "learning_rate": 0.00016491182849099877,
      "loss": 2.0198,
      "step": 149458
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.0709244012832642,
      "learning_rate": 0.00016490817612079055,
      "loss": 1.9161,
      "step": 149459
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.0266304016113281,
      "learning_rate": 0.00016490452377569822,
      "loss": 2.196,
      "step": 149460
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.1042897701263428,
      "learning_rate": 0.0001649008714557228,
      "loss": 2.0521,
      "step": 149461
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.3467158079147339,
      "learning_rate": 0.00016489721916086478,
      "loss": 2.2405,
      "step": 149462
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.2014691829681396,
      "learning_rate": 0.000164893566891125,
      "loss": 2.1733,
      "step": 149463
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.1346590518951416,
      "learning_rate": 0.00016488991464650395,
      "loss": 2.0845,
      "step": 149464
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.220859408378601,
      "learning_rate": 0.00016488626242700265,
      "loss": 2.1313,
      "step": 149465
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.0733000040054321,
      "learning_rate": 0.00016488261023262134,
      "loss": 2.0778,
      "step": 149466
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.1434135437011719,
      "learning_rate": 0.000164878958063361,
      "loss": 2.1253,
      "step": 149467
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.017417311668396,
      "learning_rate": 0.00016487530591922207,
      "loss": 2.1865,
      "step": 149468
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.1719619035720825,
      "learning_rate": 0.0001648716538002055,
      "loss": 1.976,
      "step": 149469
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.0679426193237305,
      "learning_rate": 0.00016486800170631172,
      "loss": 2.0697,
      "step": 149470
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.226468563079834,
      "learning_rate": 0.00016486434963754166,
      "loss": 2.2348,
      "step": 149471
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.1140410900115967,
      "learning_rate": 0.00016486069759389586,
      "loss": 2.0058,
      "step": 149472
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.0972797870635986,
      "learning_rate": 0.00016485704557537501,
      "loss": 2.1637,
      "step": 149473
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.1414339542388916,
      "learning_rate": 0.00016485339358197969,
      "loss": 2.1368,
      "step": 149474
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.1870484352111816,
      "learning_rate": 0.0001648497416137108,
      "loss": 2.0242,
      "step": 149475
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.1452767848968506,
      "learning_rate": 0.0001648460896705687,
      "loss": 2.1267,
      "step": 149476
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.179583191871643,
      "learning_rate": 0.00016484243775255443,
      "loss": 2.1892,
      "step": 149477
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.0471282005310059,
      "learning_rate": 0.00016483878585966851,
      "loss": 1.9562,
      "step": 149478
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.0944864749908447,
      "learning_rate": 0.00016483513399191162,
      "loss": 1.9795,
      "step": 149479
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.036915898323059,
      "learning_rate": 0.00016483148214928428,
      "loss": 1.9593,
      "step": 149480
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.1380811929702759,
      "learning_rate": 0.00016482783033178747,
      "loss": 2.0786,
      "step": 149481
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.1789642572402954,
      "learning_rate": 0.00016482417853942155,
      "loss": 1.9642,
      "step": 149482
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.15142023563385,
      "learning_rate": 0.00016482052677218757,
      "loss": 2.0824,
      "step": 149483
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.1836326122283936,
      "learning_rate": 0.0001648168750300859,
      "loss": 1.9358,
      "step": 149484
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.003740668296814,
      "learning_rate": 0.00016481322331311738,
      "loss": 2.0525,
      "step": 149485
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.229062557220459,
      "learning_rate": 0.0001648095716212825,
      "loss": 1.9643,
      "step": 149486
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.3583669662475586,
      "learning_rate": 0.00016480591995458223,
      "loss": 1.9243,
      "step": 149487
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.0541763305664062,
      "learning_rate": 0.00016480226831301693,
      "loss": 1.9988,
      "step": 149488
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.215903401374817,
      "learning_rate": 0.0001647986166965876,
      "loss": 2.3416,
      "step": 149489
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.2311705350875854,
      "learning_rate": 0.00016479496510529473,
      "loss": 2.0748,
      "step": 149490
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.1283369064331055,
      "learning_rate": 0.000164791313539139,
      "loss": 2.2119,
      "step": 149491
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.1709961891174316,
      "learning_rate": 0.00016478766199812107,
      "loss": 1.9547,
      "step": 149492
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.1395734548568726,
      "learning_rate": 0.00016478401048224172,
      "loss": 2.2091,
      "step": 149493
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.2468805313110352,
      "learning_rate": 0.0001647803589915015,
      "loss": 2.2704,
      "step": 149494
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.1704427003860474,
      "learning_rate": 0.00016477670752590125,
      "loss": 2.033,
      "step": 149495
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.2141363620758057,
      "learning_rate": 0.0001647730560854416,
      "loss": 2.1823,
      "step": 149496
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.0430253744125366,
      "learning_rate": 0.00016476940467012317,
      "loss": 2.0058,
      "step": 149497
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.1251928806304932,
      "learning_rate": 0.00016476575327994654,
      "loss": 1.9415,
      "step": 149498
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.109063982963562,
      "learning_rate": 0.00016476210191491262,
      "loss": 1.8801,
      "step": 149499
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.2292815446853638,
      "learning_rate": 0.00016475845057502188,
      "loss": 2.2513,
      "step": 149500
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.085995078086853,
      "learning_rate": 0.0001647547992602752,
      "loss": 2.0601,
      "step": 149501
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.1332756280899048,
      "learning_rate": 0.00016475114797067316,
      "loss": 2.1926,
      "step": 149502
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.2460426092147827,
      "learning_rate": 0.00016474749670621644,
      "loss": 2.0816,
      "step": 149503
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.0633105039596558,
      "learning_rate": 0.0001647438454669056,
      "loss": 2.1507,
      "step": 149504
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.1398029327392578,
      "learning_rate": 0.00016474019425274152,
      "loss": 1.8811,
      "step": 149505
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.3586719036102295,
      "learning_rate": 0.0001647365430637247,
      "loss": 1.9323,
      "step": 149506
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.1974000930786133,
      "learning_rate": 0.000164732891899856,
      "loss": 2.0832,
      "step": 149507
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.223588466644287,
      "learning_rate": 0.00016472924076113606,
      "loss": 2.1926,
      "step": 149508
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.1625584363937378,
      "learning_rate": 0.00016472558964756543,
      "loss": 2.1079,
      "step": 149509
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.1382839679718018,
      "learning_rate": 0.00016472193855914484,
      "loss": 2.1431,
      "step": 149510
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.0545896291732788,
      "learning_rate": 0.00016471828749587507,
      "loss": 2.2035,
      "step": 149511
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.0887852907180786,
      "learning_rate": 0.00016471463645775658,
      "loss": 2.0067,
      "step": 149512
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.1688847541809082,
      "learning_rate": 0.00016471098544479034,
      "loss": 2.052,
      "step": 149513
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.0931427478790283,
      "learning_rate": 0.00016470733445697686,
      "loss": 1.9663,
      "step": 149514
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.0885549783706665,
      "learning_rate": 0.00016470368349431687,
      "loss": 2.1009,
      "step": 149515
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.0782432556152344,
      "learning_rate": 0.0001647000325568109,
      "loss": 1.7475,
      "step": 149516
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.1041582822799683,
      "learning_rate": 0.00016469638164445983,
      "loss": 2.1185,
      "step": 149517
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.1332367658615112,
      "learning_rate": 0.0001646927307572642,
      "loss": 1.9081,
      "step": 149518
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.1231496334075928,
      "learning_rate": 0.0001646890798952248,
      "loss": 2.0928,
      "step": 149519
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.1056612730026245,
      "learning_rate": 0.0001646854290583423,
      "loss": 2.0219,
      "step": 149520
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.1606029272079468,
      "learning_rate": 0.00016468177824661718,
      "loss": 1.9567,
      "step": 149521
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.4156485795974731,
      "learning_rate": 0.0001646781274600504,
      "loss": 2.0623,
      "step": 149522
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.2494635581970215,
      "learning_rate": 0.00016467447669864252,
      "loss": 2.2515,
      "step": 149523
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.1484575271606445,
      "learning_rate": 0.00016467082596239408,
      "loss": 1.9764,
      "step": 149524
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.06597101688385,
      "learning_rate": 0.00016466717525130602,
      "loss": 2.1195,
      "step": 149525
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.2397218942642212,
      "learning_rate": 0.00016466352456537885,
      "loss": 2.0818,
      "step": 149526
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.0927469730377197,
      "learning_rate": 0.00016465987390461325,
      "loss": 2.1534,
      "step": 149527
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.057449460029602,
      "learning_rate": 0.00016465622326900995,
      "loss": 2.0827,
      "step": 149528
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.2063207626342773,
      "learning_rate": 0.00016465257265856954,
      "loss": 1.7973,
      "step": 149529
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.1984838247299194,
      "learning_rate": 0.00016464892207329295,
      "loss": 1.8906,
      "step": 149530
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.1112117767333984,
      "learning_rate": 0.0001646452715131806,
      "loss": 2.1832,
      "step": 149531
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.2292038202285767,
      "learning_rate": 0.00016464162097823315,
      "loss": 2.3077,
      "step": 149532
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.2385802268981934,
      "learning_rate": 0.00016463797046845148,
      "loss": 2.1482,
      "step": 149533
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.2274950742721558,
      "learning_rate": 0.00016463431998383623,
      "loss": 2.0301,
      "step": 149534
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.2707397937774658,
      "learning_rate": 0.00016463066952438784,
      "loss": 2.1299,
      "step": 149535
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.3464113473892212,
      "learning_rate": 0.00016462701909010732,
      "loss": 2.1997,
      "step": 149536
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.326907992362976,
      "learning_rate": 0.00016462336868099513,
      "loss": 2.0801,
      "step": 149537
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.2353893518447876,
      "learning_rate": 0.00016461971829705193,
      "loss": 2.1019,
      "step": 149538
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.000888466835022,
      "learning_rate": 0.00016461606793827862,
      "loss": 1.9451,
      "step": 149539
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.3651795387268066,
      "learning_rate": 0.00016461241760467568,
      "loss": 2.159,
      "step": 149540
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.2079726457595825,
      "learning_rate": 0.0001646087672962438,
      "loss": 2.0808,
      "step": 149541
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.4999427795410156,
      "learning_rate": 0.0001646051170129838,
      "loss": 1.9023,
      "step": 149542
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.0491095781326294,
      "learning_rate": 0.0001646014667548961,
      "loss": 2.028,
      "step": 149543
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.2733817100524902,
      "learning_rate": 0.00016459781652198172,
      "loss": 1.911,
      "step": 149544
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.4192317724227905,
      "learning_rate": 0.00016459416631424114,
      "loss": 1.836,
      "step": 149545
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.170884132385254,
      "learning_rate": 0.00016459051613167507,
      "loss": 2.2575,
      "step": 149546
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.1784999370574951,
      "learning_rate": 0.00016458686597428402,
      "loss": 2.021,
      "step": 149547
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.162555456161499,
      "learning_rate": 0.000164583215842069,
      "loss": 1.8963,
      "step": 149548
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.1760658025741577,
      "learning_rate": 0.0001645795657350304,
      "loss": 1.9846,
      "step": 149549
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.3223401308059692,
      "learning_rate": 0.0001645759156531691,
      "loss": 2.1869,
      "step": 149550
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.1750268936157227,
      "learning_rate": 0.00016457226559648573,
      "loss": 2.0942,
      "step": 149551
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.1136574745178223,
      "learning_rate": 0.00016456861556498088,
      "loss": 2.1387,
      "step": 149552
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.372347354888916,
      "learning_rate": 0.00016456496555865516,
      "loss": 1.9677,
      "step": 149553
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.2314765453338623,
      "learning_rate": 0.00016456131557750954,
      "loss": 2.1142,
      "step": 149554
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.0516785383224487,
      "learning_rate": 0.0001645576656215444,
      "loss": 2.1135,
      "step": 149555
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.2016681432724,
      "learning_rate": 0.00016455401569076065,
      "loss": 2.1325,
      "step": 149556
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.108919382095337,
      "learning_rate": 0.00016455036578515888,
      "loss": 2.1089,
      "step": 149557
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.0871212482452393,
      "learning_rate": 0.0001645467159047397,
      "loss": 1.8381,
      "step": 149558
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.0386792421340942,
      "learning_rate": 0.0001645430660495038,
      "loss": 1.9486,
      "step": 149559
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.0604373216629028,
      "learning_rate": 0.00016453941621945195,
      "loss": 2.2976,
      "step": 149560
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.1016186475753784,
      "learning_rate": 0.0001645357664145847,
      "loss": 1.9965,
      "step": 149561
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.1614267826080322,
      "learning_rate": 0.00016453211663490294,
      "loss": 2.2114,
      "step": 149562
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.1552151441574097,
      "learning_rate": 0.00016452846688040719,
      "loss": 2.201,
      "step": 149563
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.163322925567627,
      "learning_rate": 0.00016452481715109816,
      "loss": 1.9198,
      "step": 149564
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.0869104862213135,
      "learning_rate": 0.00016452116744697642,
      "loss": 2.1833,
      "step": 149565
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.0612421035766602,
      "learning_rate": 0.00016451751776804286,
      "loss": 2.0294,
      "step": 149566
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.0970975160598755,
      "learning_rate": 0.00016451386811429792,
      "loss": 2.1059,
      "step": 149567
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.195894479751587,
      "learning_rate": 0.00016451021848574255,
      "loss": 2.1609,
      "step": 149568
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.0620882511138916,
      "learning_rate": 0.00016450656888237727,
      "loss": 2.1614,
      "step": 149569
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.1453431844711304,
      "learning_rate": 0.00016450291930420277,
      "loss": 2.3451,
      "step": 149570
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.1684949398040771,
      "learning_rate": 0.00016449926975121965,
      "loss": 1.9232,
      "step": 149571
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.3533684015274048,
      "learning_rate": 0.00016449562022342876,
      "loss": 2.0895,
      "step": 149572
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.0038974285125732,
      "learning_rate": 0.00016449197072083058,
      "loss": 2.0557,
      "step": 149573
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.167236089706421,
      "learning_rate": 0.00016448832124342603,
      "loss": 2.1095,
      "step": 149574
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.2382748126983643,
      "learning_rate": 0.00016448467179121568,
      "loss": 1.7062,
      "step": 149575
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.2044886350631714,
      "learning_rate": 0.00016448102236420016,
      "loss": 2.1193,
      "step": 149576
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.1874895095825195,
      "learning_rate": 0.00016447737296238,
      "loss": 2.0399,
      "step": 149577
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.0539292097091675,
      "learning_rate": 0.00016447372358575625,
      "loss": 2.2056,
      "step": 149578
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.2655888795852661,
      "learning_rate": 0.00016447007423432927,
      "loss": 2.4062,
      "step": 149579
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.0460163354873657,
      "learning_rate": 0.00016446642490809996,
      "loss": 2.1056,
      "step": 149580
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.1240084171295166,
      "learning_rate": 0.0001644627756070689,
      "loss": 2.1943,
      "step": 149581
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.111289143562317,
      "learning_rate": 0.00016445912633123678,
      "loss": 2.432,
      "step": 149582
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.1394121646881104,
      "learning_rate": 0.00016445547708060417,
      "loss": 1.8988,
      "step": 149583
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.1673227548599243,
      "learning_rate": 0.00016445182785517195,
      "loss": 2.0147,
      "step": 149584
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.1357786655426025,
      "learning_rate": 0.00016444817865494057,
      "loss": 2.0434,
      "step": 149585
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.059305191040039,
      "learning_rate": 0.0001644445294799109,
      "loss": 2.0427,
      "step": 149586
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.3274940252304077,
      "learning_rate": 0.00016444088033008361,
      "loss": 2.1056,
      "step": 149587
    },
    {
      "epoch": 1.95,
      "grad_norm": 0.979448676109314,
      "learning_rate": 0.0001644372312054593,
      "loss": 2.1069,
      "step": 149588
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.0941269397735596,
      "learning_rate": 0.00016443358210603854,
      "loss": 2.0775,
      "step": 149589
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.0790950059890747,
      "learning_rate": 0.00016442993303182225,
      "loss": 2.2592,
      "step": 149590
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.3321882486343384,
      "learning_rate": 0.00016442628398281091,
      "loss": 2.1841,
      "step": 149591
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.2186511754989624,
      "learning_rate": 0.00016442263495900538,
      "loss": 2.1155,
      "step": 149592
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.1539539098739624,
      "learning_rate": 0.00016441898596040628,
      "loss": 1.8332,
      "step": 149593
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.1098089218139648,
      "learning_rate": 0.0001644153369870142,
      "loss": 1.9763,
      "step": 149594
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.0634746551513672,
      "learning_rate": 0.0001644116880388298,
      "loss": 2.0668,
      "step": 149595
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.1748511791229248,
      "learning_rate": 0.00016440803911585387,
      "loss": 2.0239,
      "step": 149596
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.0637824535369873,
      "learning_rate": 0.00016440439021808698,
      "loss": 2.0694,
      "step": 149597
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.1035298109054565,
      "learning_rate": 0.00016440074134552995,
      "loss": 2.0987,
      "step": 149598
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.097537636756897,
      "learning_rate": 0.00016439709249818329,
      "loss": 2.1854,
      "step": 149599
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.4773876667022705,
      "learning_rate": 0.00016439344367604806,
      "loss": 1.9923,
      "step": 149600
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.13140070438385,
      "learning_rate": 0.0001643897948791243,
      "loss": 2.158,
      "step": 149601
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.2022427320480347,
      "learning_rate": 0.00016438614610741329,
      "loss": 2.2309,
      "step": 149602
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.1165437698364258,
      "learning_rate": 0.00016438249736091523,
      "loss": 2.0498,
      "step": 149603
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.1592224836349487,
      "learning_rate": 0.00016437884863963116,
      "loss": 2.0398,
      "step": 149604
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.0784978866577148,
      "learning_rate": 0.00016437519994356154,
      "loss": 2.11,
      "step": 149605
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.140660047531128,
      "learning_rate": 0.00016437155127270721,
      "loss": 2.0004,
      "step": 149606
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.157999873161316,
      "learning_rate": 0.0001643679026270688,
      "loss": 1.9428,
      "step": 149607
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.1502693891525269,
      "learning_rate": 0.00016436425400664695,
      "loss": 2.0623,
      "step": 149608
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.2107975482940674,
      "learning_rate": 0.0001643606054114422,
      "loss": 2.1551,
      "step": 149609
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.1801010370254517,
      "learning_rate": 0.0001643569568414555,
      "loss": 2.204,
      "step": 149610
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.1674985885620117,
      "learning_rate": 0.00016435330829668734,
      "loss": 2.3264,
      "step": 149611
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.3160988092422485,
      "learning_rate": 0.00016434965977713852,
      "loss": 1.9266,
      "step": 149612
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.123401403427124,
      "learning_rate": 0.00016434601128280968,
      "loss": 2.0633,
      "step": 149613
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.2361871004104614,
      "learning_rate": 0.00016434236281370147,
      "loss": 2.21,
      "step": 149614
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.106897234916687,
      "learning_rate": 0.00016433871436981445,
      "loss": 2.1329,
      "step": 149615
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.1135081052780151,
      "learning_rate": 0.0001643350659511496,
      "loss": 1.8996,
      "step": 149616
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.0739802122116089,
      "learning_rate": 0.00016433141755770722,
      "loss": 2.0104,
      "step": 149617
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.2360165119171143,
      "learning_rate": 0.00016432776918948836,
      "loss": 1.9495,
      "step": 149618
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.1614140272140503,
      "learning_rate": 0.00016432412084649352,
      "loss": 2.0491,
      "step": 149619
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.1702808141708374,
      "learning_rate": 0.00016432047252872323,
      "loss": 1.9598,
      "step": 149620
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.060738444328308,
      "learning_rate": 0.00016431682423617849,
      "loss": 2.0938,
      "step": 149621
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.042394995689392,
      "learning_rate": 0.0001643131759688598,
      "loss": 1.9197,
      "step": 149622
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.1229232549667358,
      "learning_rate": 0.0001643095277267677,
      "loss": 2.1032,
      "step": 149623
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.1404346227645874,
      "learning_rate": 0.00016430587950990323,
      "loss": 2.0559,
      "step": 149624
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.1152927875518799,
      "learning_rate": 0.00016430223131826677,
      "loss": 1.9576,
      "step": 149625
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.1840354204177856,
      "learning_rate": 0.000164298583151859,
      "loss": 2.2927,
      "step": 149626
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.1116323471069336,
      "learning_rate": 0.00016429493501068077,
      "loss": 2.0618,
      "step": 149627
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.0557208061218262,
      "learning_rate": 0.00016429128689473258,
      "loss": 1.8801,
      "step": 149628
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.099185824394226,
      "learning_rate": 0.00016428763880401535,
      "loss": 2.1673,
      "step": 149629
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.1341885328292847,
      "learning_rate": 0.00016428399073852959,
      "loss": 2.0935,
      "step": 149630
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.119837760925293,
      "learning_rate": 0.00016428034269827597,
      "loss": 2.1608,
      "step": 149631
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.1262074708938599,
      "learning_rate": 0.0001642766946832551,
      "loss": 2.0047,
      "step": 149632
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.037553071975708,
      "learning_rate": 0.0001642730466934679,
      "loss": 2.1904,
      "step": 149633
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.167836308479309,
      "learning_rate": 0.00016426939872891475,
      "loss": 2.0567,
      "step": 149634
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.1734288930892944,
      "learning_rate": 0.00016426575078959662,
      "loss": 1.9397,
      "step": 149635
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.0987077951431274,
      "learning_rate": 0.00016426210287551402,
      "loss": 1.9672,
      "step": 149636
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.2593520879745483,
      "learning_rate": 0.00016425845498666767,
      "loss": 2.0947,
      "step": 149637
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.8385143280029297,
      "learning_rate": 0.00016425480712305813,
      "loss": 2.0721,
      "step": 149638
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.093326449394226,
      "learning_rate": 0.0001642511592846863,
      "loss": 1.8313,
      "step": 149639
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.219653844833374,
      "learning_rate": 0.00016424751147155265,
      "loss": 2.046,
      "step": 149640
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.2775986194610596,
      "learning_rate": 0.000164243863683658,
      "loss": 2.1168,
      "step": 149641
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.1176859140396118,
      "learning_rate": 0.00016424021592100301,
      "loss": 2.2398,
      "step": 149642
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.0259209871292114,
      "learning_rate": 0.00016423656818358834,
      "loss": 1.837,
      "step": 149643
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.2256423234939575,
      "learning_rate": 0.00016423292047141453,
      "loss": 2.1078,
      "step": 149644
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.5184463262557983,
      "learning_rate": 0.00016422927278448248,
      "loss": 2.0683,
      "step": 149645
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.1507676839828491,
      "learning_rate": 0.0001642256251227927,
      "loss": 2.097,
      "step": 149646
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.2349071502685547,
      "learning_rate": 0.00016422197748634603,
      "loss": 2.119,
      "step": 149647
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.1220566034317017,
      "learning_rate": 0.00016421832987514304,
      "loss": 2.1096,
      "step": 149648
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.1061381101608276,
      "learning_rate": 0.00016421468228918442,
      "loss": 2.0582,
      "step": 149649
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.114216923713684,
      "learning_rate": 0.0001642110347284708,
      "loss": 2.1328,
      "step": 149650
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.07036292552948,
      "learning_rate": 0.00016420738719300296,
      "loss": 2.1017,
      "step": 149651
    },
    {
      "epoch": 1.95,
      "grad_norm": 0.9894770979881287,
      "learning_rate": 0.00016420373968278144,
      "loss": 2.0545,
      "step": 149652
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.1389803886413574,
      "learning_rate": 0.00016420009219780715,
      "loss": 2.0992,
      "step": 149653
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.1823972463607788,
      "learning_rate": 0.00016419644473808056,
      "loss": 1.9814,
      "step": 149654
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.1976438760757446,
      "learning_rate": 0.00016419279730360247,
      "loss": 2.1321,
      "step": 149655
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.610122561454773,
      "learning_rate": 0.00016418914989437336,
      "loss": 2.1087,
      "step": 149656
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.7613550424575806,
      "learning_rate": 0.00016418550251039418,
      "loss": 2.1473,
      "step": 149657
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.0740870237350464,
      "learning_rate": 0.0001641818551516654,
      "loss": 2.1057,
      "step": 149658
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.4134292602539062,
      "learning_rate": 0.0001641782078181878,
      "loss": 1.964,
      "step": 149659
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.1095741987228394,
      "learning_rate": 0.0001641745605099621,
      "loss": 2.0464,
      "step": 149660
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.299201488494873,
      "learning_rate": 0.00016417091322698893,
      "loss": 1.7793,
      "step": 149661
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.5736197233200073,
      "learning_rate": 0.00016416726596926877,
      "loss": 1.9186,
      "step": 149662
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.2479487657546997,
      "learning_rate": 0.00016416361873680265,
      "loss": 2.007,
      "step": 149663
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.2162548303604126,
      "learning_rate": 0.00016415997152959092,
      "loss": 2.0921,
      "step": 149664
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.1635624170303345,
      "learning_rate": 0.00016415632434763457,
      "loss": 2.3166,
      "step": 149665
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.1367110013961792,
      "learning_rate": 0.000164152677190934,
      "loss": 2.102,
      "step": 149666
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.1269422769546509,
      "learning_rate": 0.00016414903005949026,
      "loss": 2.0545,
      "step": 149667
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.0178451538085938,
      "learning_rate": 0.0001641453829533035,
      "loss": 2.1656,
      "step": 149668
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.073439598083496,
      "learning_rate": 0.0001641417358723748,
      "loss": 2.1598,
      "step": 149669
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.120779037475586,
      "learning_rate": 0.00016413808881670465,
      "loss": 2.2588,
      "step": 149670
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.1124485731124878,
      "learning_rate": 0.00016413444178629388,
      "loss": 2.121,
      "step": 149671
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.0587351322174072,
      "learning_rate": 0.00016413079478114293,
      "loss": 2.1566,
      "step": 149672
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.2314643859863281,
      "learning_rate": 0.00016412714780125292,
      "loss": 1.9585,
      "step": 149673
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.135120153427124,
      "learning_rate": 0.00016412350084662395,
      "loss": 1.9645,
      "step": 149674
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.0897951126098633,
      "learning_rate": 0.00016411985391725717,
      "loss": 2.0541,
      "step": 149675
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.1163326501846313,
      "learning_rate": 0.0001641162070131529,
      "loss": 2.2281,
      "step": 149676
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.1202574968338013,
      "learning_rate": 0.00016411256013431215,
      "loss": 2.1237,
      "step": 149677
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.150734305381775,
      "learning_rate": 0.0001641089132807353,
      "loss": 1.918,
      "step": 149678
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.4850172996520996,
      "learning_rate": 0.0001641052664524234,
      "loss": 2.0339,
      "step": 149679
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.0955694913864136,
      "learning_rate": 0.00016410161964937664,
      "loss": 1.9613,
      "step": 149680
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.0637288093566895,
      "learning_rate": 0.00016409797287159612,
      "loss": 2.2137,
      "step": 149681
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.0578891038894653,
      "learning_rate": 0.0001640943261190822,
      "loss": 1.7838,
      "step": 149682
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.244491696357727,
      "learning_rate": 0.00016409067939183585,
      "loss": 2.461,
      "step": 149683
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.0788698196411133,
      "learning_rate": 0.0001640870326898575,
      "loss": 1.8332,
      "step": 149684
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.176827311515808,
      "learning_rate": 0.00016408338601314816,
      "loss": 2.3647,
      "step": 149685
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.1539568901062012,
      "learning_rate": 0.000164079739361708,
      "loss": 2.1447,
      "step": 149686
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.1366593837738037,
      "learning_rate": 0.0001640760927355382,
      "loss": 1.9784,
      "step": 149687
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.051239252090454,
      "learning_rate": 0.00016407244613463902,
      "loss": 2.1861,
      "step": 149688
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.1507854461669922,
      "learning_rate": 0.0001640687995590115,
      "loss": 2.1875,
      "step": 149689
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.276354432106018,
      "learning_rate": 0.00016406515300865598,
      "loss": 2.0281,
      "step": 149690
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.1331439018249512,
      "learning_rate": 0.00016406150648357347,
      "loss": 2.1017,
      "step": 149691
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.2210808992385864,
      "learning_rate": 0.0001640578599837645,
      "loss": 1.994,
      "step": 149692
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.0518453121185303,
      "learning_rate": 0.00016405421350922973,
      "loss": 2.1815,
      "step": 149693
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.3610433340072632,
      "learning_rate": 0.00016405056705996976,
      "loss": 2.1839,
      "step": 149694
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.1320339441299438,
      "learning_rate": 0.00016404692063598544,
      "loss": 2.3085,
      "step": 149695
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.2316431999206543,
      "learning_rate": 0.0001640432742372772,
      "loss": 2.0943,
      "step": 149696
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.062909722328186,
      "learning_rate": 0.00016403962786384608,
      "loss": 2.0229,
      "step": 149697
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.0872195959091187,
      "learning_rate": 0.00016403598151569252,
      "loss": 2.1226,
      "step": 149698
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.2028220891952515,
      "learning_rate": 0.0001640323351928173,
      "loss": 2.0816,
      "step": 149699
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.0712827444076538,
      "learning_rate": 0.00016402868889522083,
      "loss": 2.0358,
      "step": 149700
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.0692894458770752,
      "learning_rate": 0.00016402504262290414,
      "loss": 2.1719,
      "step": 149701
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.1330095529556274,
      "learning_rate": 0.00016402139637586766,
      "loss": 2.3106,
      "step": 149702
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.0479016304016113,
      "learning_rate": 0.00016401775015411227,
      "loss": 1.9275,
      "step": 149703
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.1821846961975098,
      "learning_rate": 0.00016401410395763856,
      "loss": 2.0262,
      "step": 149704
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.423753023147583,
      "learning_rate": 0.00016401045778644707,
      "loss": 2.0908,
      "step": 149705
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.0624593496322632,
      "learning_rate": 0.00016400681164053873,
      "loss": 2.0255,
      "step": 149706
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.107123851776123,
      "learning_rate": 0.0001640031655199141,
      "loss": 1.9919,
      "step": 149707
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.2656465768814087,
      "learning_rate": 0.00016399951942457368,
      "loss": 2.3019,
      "step": 149708
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.1270229816436768,
      "learning_rate": 0.00016399587335451848,
      "loss": 2.1592,
      "step": 149709
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.1444201469421387,
      "learning_rate": 0.00016399222730974902,
      "loss": 2.3403,
      "step": 149710
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.2800360918045044,
      "learning_rate": 0.00016398858129026582,
      "loss": 1.9549,
      "step": 149711
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.0149784088134766,
      "learning_rate": 0.00016398493529606986,
      "loss": 1.986,
      "step": 149712
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.1117887496948242,
      "learning_rate": 0.00016398128932716154,
      "loss": 2.3829,
      "step": 149713
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.0802416801452637,
      "learning_rate": 0.00016397764338354177,
      "loss": 2.3024,
      "step": 149714
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.0161490440368652,
      "learning_rate": 0.00016397399746521113,
      "loss": 2.0713,
      "step": 149715
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.0351264476776123,
      "learning_rate": 0.0001639703515721703,
      "loss": 1.9988,
      "step": 149716
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.1482337713241577,
      "learning_rate": 0.00016396670570441982,
      "loss": 2.2996,
      "step": 149717
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.053321123123169,
      "learning_rate": 0.00016396305986196062,
      "loss": 2.1214,
      "step": 149718
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.172934651374817,
      "learning_rate": 0.00016395941404479317,
      "loss": 2.0624,
      "step": 149719
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.1479170322418213,
      "learning_rate": 0.00016395576825291828,
      "loss": 2.0326,
      "step": 149720
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.4458705186843872,
      "learning_rate": 0.00016395212248633664,
      "loss": 1.9773,
      "step": 149721
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.0982919931411743,
      "learning_rate": 0.00016394847674504884,
      "loss": 1.8816,
      "step": 149722
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.2634013891220093,
      "learning_rate": 0.00016394483102905548,
      "loss": 2.354,
      "step": 149723
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.0428478717803955,
      "learning_rate": 0.00016394118533835745,
      "loss": 1.7822,
      "step": 149724
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.1651978492736816,
      "learning_rate": 0.0001639375396729552,
      "loss": 2.0784,
      "step": 149725
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.0696181058883667,
      "learning_rate": 0.00016393389403284968,
      "loss": 1.9232,
      "step": 149726
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.139816164970398,
      "learning_rate": 0.00016393024841804138,
      "loss": 2.1226,
      "step": 149727
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.1395071744918823,
      "learning_rate": 0.00016392660282853105,
      "loss": 1.9732,
      "step": 149728
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.0969452857971191,
      "learning_rate": 0.0001639229572643192,
      "loss": 2.2171,
      "step": 149729
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.0854344367980957,
      "learning_rate": 0.00016391931172540677,
      "loss": 2.1198,
      "step": 149730
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.2296183109283447,
      "learning_rate": 0.00016391566621179414,
      "loss": 2.1706,
      "step": 149731
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.170149564743042,
      "learning_rate": 0.00016391202072348234,
      "loss": 2.2031,
      "step": 149732
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.1065210103988647,
      "learning_rate": 0.00016390837526047175,
      "loss": 2.0883,
      "step": 149733
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.1327896118164062,
      "learning_rate": 0.00016390472982276335,
      "loss": 2.0551,
      "step": 149734
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.0778911113739014,
      "learning_rate": 0.0001639010844103574,
      "loss": 2.1133,
      "step": 149735
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.174658179283142,
      "learning_rate": 0.0001638974390232549,
      "loss": 2.2266,
      "step": 149736
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.133371114730835,
      "learning_rate": 0.0001638937936614564,
      "loss": 2.0879,
      "step": 149737
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.1632745265960693,
      "learning_rate": 0.00016389014832496267,
      "loss": 2.1047,
      "step": 149738
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.1614874601364136,
      "learning_rate": 0.00016388650301377424,
      "loss": 2.2808,
      "step": 149739
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.0944581031799316,
      "learning_rate": 0.00016388285772789212,
      "loss": 2.1557,
      "step": 149740
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.1822558641433716,
      "learning_rate": 0.00016387921246731647,
      "loss": 2.0783,
      "step": 149741
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.1668046712875366,
      "learning_rate": 0.00016387556723204835,
      "loss": 2.1327,
      "step": 149742
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.1887027025222778,
      "learning_rate": 0.00016387192202208827,
      "loss": 2.0714,
      "step": 149743
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.2457783222198486,
      "learning_rate": 0.00016386827683743707,
      "loss": 2.3131,
      "step": 149744
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.2576916217803955,
      "learning_rate": 0.0001638646316780952,
      "loss": 1.988,
      "step": 149745
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.22412109375,
      "learning_rate": 0.0001638609865440637,
      "loss": 2.1081,
      "step": 149746
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.6406829357147217,
      "learning_rate": 0.00016385734143534278,
      "loss": 2.0109,
      "step": 149747
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.3456408977508545,
      "learning_rate": 0.00016385369635193346,
      "loss": 2.0753,
      "step": 149748
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.0401015281677246,
      "learning_rate": 0.00016385005129383623,
      "loss": 2.0071,
      "step": 149749
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.3744316101074219,
      "learning_rate": 0.0001638464062610519,
      "loss": 2.1132,
      "step": 149750
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.2965110540390015,
      "learning_rate": 0.00016384276125358104,
      "loss": 2.101,
      "step": 149751
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.2064763307571411,
      "learning_rate": 0.00016383911627142458,
      "loss": 2.0686,
      "step": 149752
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.0944163799285889,
      "learning_rate": 0.00016383547131458275,
      "loss": 2.0211,
      "step": 149753
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.161592960357666,
      "learning_rate": 0.00016383182638305663,
      "loss": 2.115,
      "step": 149754
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.2484334707260132,
      "learning_rate": 0.0001638281814768466,
      "loss": 2.2159,
      "step": 149755
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.0997366905212402,
      "learning_rate": 0.00016382453659595362,
      "loss": 1.956,
      "step": 149756
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.145882248878479,
      "learning_rate": 0.00016382089174037806,
      "loss": 2.2055,
      "step": 149757
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.1763437986373901,
      "learning_rate": 0.00016381724691012106,
      "loss": 2.1304,
      "step": 149758
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.1537997722625732,
      "learning_rate": 0.00016381360210518273,
      "loss": 2.0874,
      "step": 149759
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.0639020204544067,
      "learning_rate": 0.00016380995732556418,
      "loss": 2.0237,
      "step": 149760
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.1985743045806885,
      "learning_rate": 0.00016380631257126578,
      "loss": 2.0448,
      "step": 149761
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.0902669429779053,
      "learning_rate": 0.00016380266784228847,
      "loss": 2.125,
      "step": 149762
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.1611380577087402,
      "learning_rate": 0.00016379902313863273,
      "loss": 1.9499,
      "step": 149763
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.1496132612228394,
      "learning_rate": 0.00016379537846029953,
      "loss": 2.1874,
      "step": 149764
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.2255438566207886,
      "learning_rate": 0.00016379173380728915,
      "loss": 1.9014,
      "step": 149765
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.0849741697311401,
      "learning_rate": 0.00016378808917960253,
      "loss": 2.0945,
      "step": 149766
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.6715624332427979,
      "learning_rate": 0.0001637844445772402,
      "loss": 2.145,
      "step": 149767
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.1781766414642334,
      "learning_rate": 0.00016378080000020294,
      "loss": 2.0162,
      "step": 149768
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.2321950197219849,
      "learning_rate": 0.00016377715544849136,
      "loss": 2.0257,
      "step": 149769
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.1137239933013916,
      "learning_rate": 0.0001637735109221064,
      "loss": 2.0255,
      "step": 149770
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.1126655340194702,
      "learning_rate": 0.0001637698664210483,
      "loss": 2.1351,
      "step": 149771
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.1599878072738647,
      "learning_rate": 0.00016376622194531803,
      "loss": 2.0803,
      "step": 149772
    },
    {
      "epoch": 1.95,
      "grad_norm": 3.585338592529297,
      "learning_rate": 0.0001637625774949161,
      "loss": 2.0375,
      "step": 149773
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.0475950241088867,
      "learning_rate": 0.00016375893306984338,
      "loss": 1.9714,
      "step": 149774
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.30775785446167,
      "learning_rate": 0.00016375528867010032,
      "loss": 2.214,
      "step": 149775
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.3103500604629517,
      "learning_rate": 0.00016375164429568802,
      "loss": 2.0414,
      "step": 149776
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.1791584491729736,
      "learning_rate": 0.00016374799994660656,
      "loss": 1.9376,
      "step": 149777
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.1405471563339233,
      "learning_rate": 0.00016374435562285707,
      "loss": 2.1917,
      "step": 149778
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.2449586391448975,
      "learning_rate": 0.00016374071132443992,
      "loss": 2.1536,
      "step": 149779
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.2567007541656494,
      "learning_rate": 0.0001637370670513561,
      "loss": 1.9459,
      "step": 149780
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.196006178855896,
      "learning_rate": 0.00016373342280360598,
      "loss": 2.1684,
      "step": 149781
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.0934499502182007,
      "learning_rate": 0.00016372977858119053,
      "loss": 2.0218,
      "step": 149782
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.1549525260925293,
      "learning_rate": 0.00016372613438411028,
      "loss": 2.083,
      "step": 149783
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.2554086446762085,
      "learning_rate": 0.00016372249021236591,
      "loss": 2.0622,
      "step": 149784
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.4105311632156372,
      "learning_rate": 0.00016371884606595802,
      "loss": 2.0201,
      "step": 149785
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.221506953239441,
      "learning_rate": 0.00016371520194488745,
      "loss": 2.0856,
      "step": 149786
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.1274152994155884,
      "learning_rate": 0.00016371155784915468,
      "loss": 1.8617,
      "step": 149787
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.2670903205871582,
      "learning_rate": 0.00016370791377876064,
      "loss": 1.9807,
      "step": 149788
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.1017392873764038,
      "learning_rate": 0.00016370426973370586,
      "loss": 2.1408,
      "step": 149789
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.2757065296173096,
      "learning_rate": 0.0001637006257139909,
      "loss": 2.201,
      "step": 149790
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.0620170831680298,
      "learning_rate": 0.00016369698171961672,
      "loss": 2.1067,
      "step": 149791
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.0943474769592285,
      "learning_rate": 0.0001636933377505838,
      "loss": 2.303,
      "step": 149792
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.2049757242202759,
      "learning_rate": 0.00016368969380689278,
      "loss": 1.8587,
      "step": 149793
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.353472352027893,
      "learning_rate": 0.0001636860498885445,
      "loss": 1.9734,
      "step": 149794
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.0835031270980835,
      "learning_rate": 0.00016368240599553958,
      "loss": 2.2263,
      "step": 149795
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.4035838842391968,
      "learning_rate": 0.00016367876212787855,
      "loss": 1.9985,
      "step": 149796
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.4265216588974,
      "learning_rate": 0.00016367511828556231,
      "loss": 1.8793,
      "step": 149797
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.1379128694534302,
      "learning_rate": 0.0001636714744685915,
      "loss": 2.2206,
      "step": 149798
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.1502121686935425,
      "learning_rate": 0.00016366783067696657,
      "loss": 2.3587,
      "step": 149799
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.0819730758666992,
      "learning_rate": 0.0001636641869106885,
      "loss": 2.0116,
      "step": 149800
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.0456753969192505,
      "learning_rate": 0.00016366054316975783,
      "loss": 1.9935,
      "step": 149801
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.094560980796814,
      "learning_rate": 0.0001636568994541751,
      "loss": 2.1319,
      "step": 149802
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.2051664590835571,
      "learning_rate": 0.00016365325576394126,
      "loss": 2.0968,
      "step": 149803
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.1309126615524292,
      "learning_rate": 0.00016364961209905672,
      "loss": 1.7864,
      "step": 149804
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.112277626991272,
      "learning_rate": 0.00016364596845952243,
      "loss": 1.9084,
      "step": 149805
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.1246352195739746,
      "learning_rate": 0.00016364232484533892,
      "loss": 2.0457,
      "step": 149806
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.1769826412200928,
      "learning_rate": 0.00016363868125650688,
      "loss": 2.0355,
      "step": 149807
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.2296017408370972,
      "learning_rate": 0.00016363503769302687,
      "loss": 1.9502,
      "step": 149808
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.3678040504455566,
      "learning_rate": 0.00016363139415489976,
      "loss": 2.2514,
      "step": 149809
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.3278447389602661,
      "learning_rate": 0.00016362775064212605,
      "loss": 2.3369,
      "step": 149810
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.0308641195297241,
      "learning_rate": 0.00016362410715470667,
      "loss": 2.0831,
      "step": 149811
    },
    {
      "epoch": 1.95,
      "grad_norm": 0.956989049911499,
      "learning_rate": 0.000163620463692642,
      "loss": 2.1463,
      "step": 149812
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.143225908279419,
      "learning_rate": 0.0001636168202559331,
      "loss": 2.2654,
      "step": 149813
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.3105199337005615,
      "learning_rate": 0.00016361317684458014,
      "loss": 2.0786,
      "step": 149814
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.075985312461853,
      "learning_rate": 0.00016360953345858422,
      "loss": 1.9375,
      "step": 149815
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.0521554946899414,
      "learning_rate": 0.00016360589009794572,
      "loss": 2.1831,
      "step": 149816
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.251991868019104,
      "learning_rate": 0.0001636022467626656,
      "loss": 2.0462,
      "step": 149817
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.1563926935195923,
      "learning_rate": 0.00016359860345274428,
      "loss": 2.3337,
      "step": 149818
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.1855195760726929,
      "learning_rate": 0.0001635949601681828,
      "loss": 2.1445,
      "step": 149819
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.1678224802017212,
      "learning_rate": 0.00016359131690898133,
      "loss": 2.1181,
      "step": 149820
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.3941395282745361,
      "learning_rate": 0.00016358767367514093,
      "loss": 2.0925,
      "step": 149821
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.0600250959396362,
      "learning_rate": 0.00016358403046666205,
      "loss": 2.0921,
      "step": 149822
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.0796421766281128,
      "learning_rate": 0.00016358038728354557,
      "loss": 2.0008,
      "step": 149823
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.1306877136230469,
      "learning_rate": 0.00016357674412579196,
      "loss": 2.0786,
      "step": 149824
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.3662879467010498,
      "learning_rate": 0.00016357310099340225,
      "loss": 2.0851,
      "step": 149825
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.1657884120941162,
      "learning_rate": 0.00016356945788637662,
      "loss": 2.1048,
      "step": 149826
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.0587706565856934,
      "learning_rate": 0.00016356581480471617,
      "loss": 1.952,
      "step": 149827
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.430964469909668,
      "learning_rate": 0.00016356217174842125,
      "loss": 2.0447,
      "step": 149828
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.1852391958236694,
      "learning_rate": 0.00016355852871749284,
      "loss": 2.1415,
      "step": 149829
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.0149238109588623,
      "learning_rate": 0.00016355488571193131,
      "loss": 2.0213,
      "step": 149830
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.2157741785049438,
      "learning_rate": 0.00016355124273173782,
      "loss": 2.0609,
      "step": 149831
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.2449253797531128,
      "learning_rate": 0.0001635475997769124,
      "loss": 1.9575,
      "step": 149832
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.2693357467651367,
      "learning_rate": 0.00016354395684745624,
      "loss": 1.9715,
      "step": 149833
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.0634537935256958,
      "learning_rate": 0.00016354031394336972,
      "loss": 2.0648,
      "step": 149834
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.1534324884414673,
      "learning_rate": 0.0001635366710646537,
      "loss": 1.9806,
      "step": 149835
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.0422393083572388,
      "learning_rate": 0.00016353302821130874,
      "loss": 2.0065,
      "step": 149836
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.1872889995574951,
      "learning_rate": 0.00016352938538333574,
      "loss": 2.0603,
      "step": 149837
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.1902185678482056,
      "learning_rate": 0.000163525742580735,
      "loss": 2.15,
      "step": 149838
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.1651166677474976,
      "learning_rate": 0.0001635220998035075,
      "loss": 1.9161,
      "step": 149839
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.0849806070327759,
      "learning_rate": 0.00016351845705165371,
      "loss": 2.1451,
      "step": 149840
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.197279453277588,
      "learning_rate": 0.00016351481432517456,
      "loss": 1.948,
      "step": 149841
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.12437105178833,
      "learning_rate": 0.0001635111716240704,
      "loss": 2.0074,
      "step": 149842
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.0620958805084229,
      "learning_rate": 0.00016350752894834237,
      "loss": 2.1978,
      "step": 149843
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.201517105102539,
      "learning_rate": 0.00016350388629799063,
      "loss": 2.1265,
      "step": 149844
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.1967390775680542,
      "learning_rate": 0.0001635002436730162,
      "loss": 2.2879,
      "step": 149845
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.0514870882034302,
      "learning_rate": 0.0001634966010734195,
      "loss": 1.9901,
      "step": 149846
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.1597570180892944,
      "learning_rate": 0.00016349295849920153,
      "loss": 1.9681,
      "step": 149847
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.1194756031036377,
      "learning_rate": 0.0001634893159503627,
      "loss": 1.9072,
      "step": 149848
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.1893572807312012,
      "learning_rate": 0.00016348567342690397,
      "loss": 2.1476,
      "step": 149849
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.0555293560028076,
      "learning_rate": 0.0001634820309288256,
      "loss": 2.0749,
      "step": 149850
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.3980722427368164,
      "learning_rate": 0.00016347838845612861,
      "loss": 2.0804,
      "step": 149851
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.104699969291687,
      "learning_rate": 0.00016347474600881344,
      "loss": 2.017,
      "step": 149852
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.0548653602600098,
      "learning_rate": 0.00016347110358688104,
      "loss": 2.0996,
      "step": 149853
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.1500353813171387,
      "learning_rate": 0.0001634674611903318,
      "loss": 1.9965,
      "step": 149854
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.0239231586456299,
      "learning_rate": 0.0001634638188191668,
      "loss": 2.2653,
      "step": 149855
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.2519011497497559,
      "learning_rate": 0.0001634601764733862,
      "loss": 1.9703,
      "step": 149856
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.1837718486785889,
      "learning_rate": 0.00016345653415299105,
      "loss": 2.2305,
      "step": 149857
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.1480306386947632,
      "learning_rate": 0.00016345289185798177,
      "loss": 2.0517,
      "step": 149858
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.249659538269043,
      "learning_rate": 0.0001634492495883593,
      "loss": 2.173,
      "step": 149859
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.2128679752349854,
      "learning_rate": 0.0001634456073441241,
      "loss": 2.1277,
      "step": 149860
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.1587872505187988,
      "learning_rate": 0.00016344196512527715,
      "loss": 2.2068,
      "step": 149861
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.1358249187469482,
      "learning_rate": 0.0001634383229318187,
      "loss": 1.9013,
      "step": 149862
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.3669277429580688,
      "learning_rate": 0.00016343468076374974,
      "loss": 2.1691,
      "step": 149863
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.4589513540267944,
      "learning_rate": 0.0001634310386210707,
      "loss": 1.9877,
      "step": 149864
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.2981770038604736,
      "learning_rate": 0.00016342739650378264,
      "loss": 2.1679,
      "step": 149865
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.3268829584121704,
      "learning_rate": 0.0001634237544118858,
      "loss": 1.9644,
      "step": 149866
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.0153237581253052,
      "learning_rate": 0.00016342011234538118,
      "loss": 1.9988,
      "step": 149867
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.20950186252594,
      "learning_rate": 0.00016341647030426935,
      "loss": 1.8138,
      "step": 149868
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.0756596326828003,
      "learning_rate": 0.000163412828288551,
      "loss": 2.212,
      "step": 149869
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.2982196807861328,
      "learning_rate": 0.00016340918629822662,
      "loss": 1.9316,
      "step": 149870
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.1205533742904663,
      "learning_rate": 0.0001634055443332972,
      "loss": 2.1176,
      "step": 149871
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.0979993343353271,
      "learning_rate": 0.00016340190239376315,
      "loss": 2.1156,
      "step": 149872
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.0945048332214355,
      "learning_rate": 0.00016339826047962537,
      "loss": 2.1351,
      "step": 149873
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.0811147689819336,
      "learning_rate": 0.00016339461859088446,
      "loss": 2.2518,
      "step": 149874
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.1343848705291748,
      "learning_rate": 0.00016339097672754093,
      "loss": 2.023,
      "step": 149875
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.0897018909454346,
      "learning_rate": 0.0001633873348895957,
      "loss": 2.2057,
      "step": 149876
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.21999192237854,
      "learning_rate": 0.00016338369307704936,
      "loss": 2.2855,
      "step": 149877
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.241936206817627,
      "learning_rate": 0.00016338005128990244,
      "loss": 2.093,
      "step": 149878
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.1670104265213013,
      "learning_rate": 0.00016337640952815587,
      "loss": 1.8632,
      "step": 149879
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.1760073900222778,
      "learning_rate": 0.00016337276779181025,
      "loss": 1.9021,
      "step": 149880
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.210465669631958,
      "learning_rate": 0.00016336912608086606,
      "loss": 2.0422,
      "step": 149881
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.5989654064178467,
      "learning_rate": 0.0001633654843953242,
      "loss": 2.1844,
      "step": 149882
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.1698684692382812,
      "learning_rate": 0.00016336184273518534,
      "loss": 2.1572,
      "step": 149883
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.2056375741958618,
      "learning_rate": 0.00016335820110044994,
      "loss": 2.3386,
      "step": 149884
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.095374584197998,
      "learning_rate": 0.00016335455949111898,
      "loss": 1.8939,
      "step": 149885
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.1628998517990112,
      "learning_rate": 0.00016335091790719298,
      "loss": 2.0284,
      "step": 149886
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.2502838373184204,
      "learning_rate": 0.00016334727634867243,
      "loss": 2.1116,
      "step": 149887
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.3764740228652954,
      "learning_rate": 0.0001633436348155584,
      "loss": 2.1147,
      "step": 149888
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.0787560939788818,
      "learning_rate": 0.00016333999330785126,
      "loss": 2.2251,
      "step": 149889
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.1864848136901855,
      "learning_rate": 0.00016333635182555186,
      "loss": 2.0281,
      "step": 149890
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.2136237621307373,
      "learning_rate": 0.00016333271036866081,
      "loss": 2.0308,
      "step": 149891
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.9006174802780151,
      "learning_rate": 0.00016332906893717884,
      "loss": 2.0829,
      "step": 149892
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.100311040878296,
      "learning_rate": 0.00016332542753110644,
      "loss": 2.0382,
      "step": 149893
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.1978094577789307,
      "learning_rate": 0.00016332178615044454,
      "loss": 2.0084,
      "step": 149894
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.156977653503418,
      "learning_rate": 0.00016331814479519355,
      "loss": 1.9842,
      "step": 149895
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.4291120767593384,
      "learning_rate": 0.00016331450346535446,
      "loss": 2.1523,
      "step": 149896
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.1361924409866333,
      "learning_rate": 0.00016331086216092778,
      "loss": 2.076,
      "step": 149897
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.1331199407577515,
      "learning_rate": 0.00016330722088191412,
      "loss": 2.0617,
      "step": 149898
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.0960496664047241,
      "learning_rate": 0.00016330357962831418,
      "loss": 2.1216,
      "step": 149899
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.081137776374817,
      "learning_rate": 0.00016329993840012878,
      "loss": 2.1008,
      "step": 149900
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.1584025621414185,
      "learning_rate": 0.00016329629719735841,
      "loss": 2.035,
      "step": 149901
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.16221022605896,
      "learning_rate": 0.0001632926560200039,
      "loss": 2.1984,
      "step": 149902
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.0151735544204712,
      "learning_rate": 0.00016328901486806582,
      "loss": 1.9928,
      "step": 149903
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.071053385734558,
      "learning_rate": 0.0001632853737415451,
      "loss": 1.8639,
      "step": 149904
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.1608870029449463,
      "learning_rate": 0.00016328173264044198,
      "loss": 2.2519,
      "step": 149905
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.2916468381881714,
      "learning_rate": 0.00016327809156475745,
      "loss": 1.995,
      "step": 149906
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.132588505744934,
      "learning_rate": 0.00016327445051449199,
      "loss": 2.0054,
      "step": 149907
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.1630289554595947,
      "learning_rate": 0.00016327080948964657,
      "loss": 2.2053,
      "step": 149908
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.1399362087249756,
      "learning_rate": 0.0001632671684902215,
      "loss": 2.1991,
      "step": 149909
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.4630967378616333,
      "learning_rate": 0.0001632635275162179,
      "loss": 2.0859,
      "step": 149910
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.2044506072998047,
      "learning_rate": 0.00016325988656763595,
      "loss": 2.0563,
      "step": 149911
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.0985299348831177,
      "learning_rate": 0.00016325624564447673,
      "loss": 2.3458,
      "step": 149912
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.1586072444915771,
      "learning_rate": 0.00016325260474674063,
      "loss": 2.0775,
      "step": 149913
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.294258952140808,
      "learning_rate": 0.00016324896387442852,
      "loss": 2.0487,
      "step": 149914
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.4811357259750366,
      "learning_rate": 0.0001632453230275409,
      "loss": 2.1427,
      "step": 149915
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.2000277042388916,
      "learning_rate": 0.00016324168220607882,
      "loss": 1.8604,
      "step": 149916
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.0889428853988647,
      "learning_rate": 0.00016323804141004245,
      "loss": 2.1351,
      "step": 149917
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.152854323387146,
      "learning_rate": 0.00016323440063943282,
      "loss": 2.15,
      "step": 149918
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.4462257623672485,
      "learning_rate": 0.00016323075989425038,
      "loss": 2.0422,
      "step": 149919
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.374854326248169,
      "learning_rate": 0.00016322711917449604,
      "loss": 1.9296,
      "step": 149920
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.069008708000183,
      "learning_rate": 0.00016322347848017025,
      "loss": 1.9302,
      "step": 149921
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.22118079662323,
      "learning_rate": 0.000163219837811274,
      "loss": 1.8983,
      "step": 149922
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.2280722856521606,
      "learning_rate": 0.00016321619716780757,
      "loss": 2.1911,
      "step": 149923
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.648483395576477,
      "learning_rate": 0.00016321255654977193,
      "loss": 2.1694,
      "step": 149924
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.5299993753433228,
      "learning_rate": 0.00016320891595716756,
      "loss": 2.1479,
      "step": 149925
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.062459945678711,
      "learning_rate": 0.00016320527538999535,
      "loss": 2.147,
      "step": 149926
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.199242115020752,
      "learning_rate": 0.0001632016348482557,
      "loss": 1.9255,
      "step": 149927
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.4206759929656982,
      "learning_rate": 0.00016319799433194975,
      "loss": 1.8725,
      "step": 149928
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.2019593715667725,
      "learning_rate": 0.0001631943538410776,
      "loss": 1.939,
      "step": 149929
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.1763123273849487,
      "learning_rate": 0.0001631907133756403,
      "loss": 2.1557,
      "step": 149930
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.0420674085617065,
      "learning_rate": 0.00016318707293563837,
      "loss": 2.0898,
      "step": 149931
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.0698496103286743,
      "learning_rate": 0.0001631834325210726,
      "loss": 2.2649,
      "step": 149932
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.1948095560073853,
      "learning_rate": 0.00016317979213194354,
      "loss": 2.0118,
      "step": 149933
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.1824183464050293,
      "learning_rate": 0.00016317615176825203,
      "loss": 2.331,
      "step": 149934
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.0823419094085693,
      "learning_rate": 0.0001631725114299987,
      "loss": 2.0036,
      "step": 149935
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.2422950267791748,
      "learning_rate": 0.0001631688711171841,
      "loss": 2.0908,
      "step": 149936
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.0487117767333984,
      "learning_rate": 0.00016316523082980896,
      "loss": 1.9813,
      "step": 149937
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.436591625213623,
      "learning_rate": 0.00016316159056787408,
      "loss": 2.1721,
      "step": 149938
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.111868143081665,
      "learning_rate": 0.0001631579503313799,
      "loss": 2.2156,
      "step": 149939
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.0823378562927246,
      "learning_rate": 0.00016315431012032738,
      "loss": 2.0323,
      "step": 149940
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.0128321647644043,
      "learning_rate": 0.00016315066993471703,
      "loss": 1.9446,
      "step": 149941
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.1670427322387695,
      "learning_rate": 0.00016314702977454957,
      "loss": 2.0757,
      "step": 149942
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.0115375518798828,
      "learning_rate": 0.00016314338963982556,
      "loss": 2.1483,
      "step": 149943
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.1384433507919312,
      "learning_rate": 0.00016313974953054586,
      "loss": 2.1404,
      "step": 149944
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.0224685668945312,
      "learning_rate": 0.00016313610944671098,
      "loss": 2.1297,
      "step": 149945
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.2983016967773438,
      "learning_rate": 0.00016313246938832177,
      "loss": 2.1115,
      "step": 149946
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.129684329032898,
      "learning_rate": 0.00016312882935537882,
      "loss": 2.0214,
      "step": 149947
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.120313048362732,
      "learning_rate": 0.0001631251893478828,
      "loss": 2.1916,
      "step": 149948
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.0835938453674316,
      "learning_rate": 0.0001631215493658343,
      "loss": 2.0658,
      "step": 149949
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.1299866437911987,
      "learning_rate": 0.0001631179094092342,
      "loss": 2.1189,
      "step": 149950
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.1328636407852173,
      "learning_rate": 0.00016311426947808295,
      "loss": 2.0117,
      "step": 149951
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.1325843334197998,
      "learning_rate": 0.00016311062957238143,
      "loss": 2.0641,
      "step": 149952
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.136311411857605,
      "learning_rate": 0.00016310698969213027,
      "loss": 2.2516,
      "step": 149953
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.1542407274246216,
      "learning_rate": 0.00016310334983733006,
      "loss": 2.1238,
      "step": 149954
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.123460292816162,
      "learning_rate": 0.00016309971000798144,
      "loss": 2.1585,
      "step": 149955
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.2115997076034546,
      "learning_rate": 0.00016309607020408527,
      "loss": 2.1636,
      "step": 149956
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.0643260478973389,
      "learning_rate": 0.00016309243042564203,
      "loss": 2.1028,
      "step": 149957
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.1972531080245972,
      "learning_rate": 0.00016308879067265255,
      "loss": 2.1513,
      "step": 149958
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.0245332717895508,
      "learning_rate": 0.0001630851509451175,
      "loss": 2.0239,
      "step": 149959
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.104026198387146,
      "learning_rate": 0.00016308151124303752,
      "loss": 2.0386,
      "step": 149960
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.1613564491271973,
      "learning_rate": 0.0001630778715664131,
      "loss": 1.9412,
      "step": 149961
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.111950159072876,
      "learning_rate": 0.00016307423191524526,
      "loss": 2.1298,
      "step": 149962
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.1359091997146606,
      "learning_rate": 0.00016307059228953434,
      "loss": 2.0661,
      "step": 149963
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.1180881261825562,
      "learning_rate": 0.00016306695268928132,
      "loss": 2.0466,
      "step": 149964
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.160805106163025,
      "learning_rate": 0.0001630633131144867,
      "loss": 2.2285,
      "step": 149965
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.2061879634857178,
      "learning_rate": 0.00016305967356515112,
      "loss": 1.8401,
      "step": 149966
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.1638706922531128,
      "learning_rate": 0.00016305603404127546,
      "loss": 2.1297,
      "step": 149967
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.185083270072937,
      "learning_rate": 0.00016305239454286025,
      "loss": 1.8813,
      "step": 149968
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.1480202674865723,
      "learning_rate": 0.00016304875506990607,
      "loss": 2.0507,
      "step": 149969
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.1574934720993042,
      "learning_rate": 0.00016304511562241383,
      "loss": 2.0446,
      "step": 149970
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.1394062042236328,
      "learning_rate": 0.00016304147620038408,
      "loss": 2.0281,
      "step": 149971
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.2337900400161743,
      "learning_rate": 0.0001630378368038174,
      "loss": 2.0845,
      "step": 149972
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.194487452507019,
      "learning_rate": 0.00016303419743271468,
      "loss": 2.0781,
      "step": 149973
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.1053849458694458,
      "learning_rate": 0.00016303055808707638,
      "loss": 2.1145,
      "step": 149974
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.142925500869751,
      "learning_rate": 0.00016302691876690339,
      "loss": 1.9659,
      "step": 149975
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.1328884363174438,
      "learning_rate": 0.0001630232794721963,
      "loss": 2.001,
      "step": 149976
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.0833510160446167,
      "learning_rate": 0.00016301964020295573,
      "loss": 2.1967,
      "step": 149977
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.0771363973617554,
      "learning_rate": 0.00016301600095918232,
      "loss": 2.0289,
      "step": 149978
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.1389986276626587,
      "learning_rate": 0.0001630123617408769,
      "loss": 1.9549,
      "step": 149979
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.188496470451355,
      "learning_rate": 0.00016300872254803998,
      "loss": 1.9606,
      "step": 149980
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.1495189666748047,
      "learning_rate": 0.00016300508338067243,
      "loss": 2.0128,
      "step": 149981
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.2537213563919067,
      "learning_rate": 0.00016300144423877483,
      "loss": 2.1198,
      "step": 149982
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.4765713214874268,
      "learning_rate": 0.00016299780512234782,
      "loss": 2.1387,
      "step": 149983
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.085288405418396,
      "learning_rate": 0.00016299416603139202,
      "loss": 2.0092,
      "step": 149984
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.0415146350860596,
      "learning_rate": 0.00016299052696590828,
      "loss": 2.1739,
      "step": 149985
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.1260639429092407,
      "learning_rate": 0.00016298688792589712,
      "loss": 2.1032,
      "step": 149986
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.1492483615875244,
      "learning_rate": 0.00016298324891135936,
      "loss": 1.9439,
      "step": 149987
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.2495038509368896,
      "learning_rate": 0.00016297960992229548,
      "loss": 2.1374,
      "step": 149988
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.463281273841858,
      "learning_rate": 0.0001629759709587065,
      "loss": 2.3223,
      "step": 149989
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.1485580205917358,
      "learning_rate": 0.00016297233202059266,
      "loss": 2.3013,
      "step": 149990
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.1373076438903809,
      "learning_rate": 0.00016296869310795497,
      "loss": 2.0808,
      "step": 149991
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.069047451019287,
      "learning_rate": 0.00016296505422079387,
      "loss": 2.1198,
      "step": 149992
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.317007064819336,
      "learning_rate": 0.00016296141535911027,
      "loss": 2.0771,
      "step": 149993
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.0864297151565552,
      "learning_rate": 0.0001629577765229046,
      "loss": 1.9234,
      "step": 149994
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.1761363744735718,
      "learning_rate": 0.0001629541377121779,
      "loss": 1.9573,
      "step": 149995
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.5882714986801147,
      "learning_rate": 0.0001629504989269304,
      "loss": 1.9586,
      "step": 149996
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.128854751586914,
      "learning_rate": 0.0001629468601671631,
      "loss": 1.9444,
      "step": 149997
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.1485146284103394,
      "learning_rate": 0.00016294322143287646,
      "loss": 2.1802,
      "step": 149998
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.2628862857818604,
      "learning_rate": 0.00016293958272407135,
      "loss": 2.0773,
      "step": 149999
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.3171396255493164,
      "learning_rate": 0.00016293594404074823,
      "loss": 1.9228,
      "step": 150000
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.0910030603408813,
      "learning_rate": 0.00016293230538290808,
      "loss": 2.0972,
      "step": 150001
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.1320958137512207,
      "learning_rate": 0.00016292866675055137,
      "loss": 2.2687,
      "step": 150002
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.1009562015533447,
      "learning_rate": 0.00016292502814367878,
      "loss": 1.8788,
      "step": 150003
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.1292345523834229,
      "learning_rate": 0.00016292138956229093,
      "loss": 1.9295,
      "step": 150004
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.6703054904937744,
      "learning_rate": 0.0001629177510063887,
      "loss": 1.7749,
      "step": 150005
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.407307744026184,
      "learning_rate": 0.00016291411247597252,
      "loss": 2.0159,
      "step": 150006
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.1214537620544434,
      "learning_rate": 0.00016291047397104332,
      "loss": 2.1185,
      "step": 150007
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.1747639179229736,
      "learning_rate": 0.00016290683549160165,
      "loss": 2.2109,
      "step": 150008
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.3876841068267822,
      "learning_rate": 0.00016290319703764818,
      "loss": 2.3014,
      "step": 150009
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.1122711896896362,
      "learning_rate": 0.00016289955860918347,
      "loss": 2.0908,
      "step": 150010
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.1744917631149292,
      "learning_rate": 0.00016289592020620846,
      "loss": 1.905,
      "step": 150011
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.0984095335006714,
      "learning_rate": 0.00016289228182872354,
      "loss": 2.2492,
      "step": 150012
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.0170537233352661,
      "learning_rate": 0.0001628886434767297,
      "loss": 2.0036,
      "step": 150013
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.1231240034103394,
      "learning_rate": 0.00016288500515022737,
      "loss": 1.9411,
      "step": 150014
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.136631965637207,
      "learning_rate": 0.00016288136684921737,
      "loss": 2.1498,
      "step": 150015
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.353087306022644,
      "learning_rate": 0.0001628777285737002,
      "loss": 1.9189,
      "step": 150016
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.2077810764312744,
      "learning_rate": 0.00016287409032367673,
      "loss": 1.9476,
      "step": 150017
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.3758797645568848,
      "learning_rate": 0.00016287045209914744,
      "loss": 2.0394,
      "step": 150018
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.2291653156280518,
      "learning_rate": 0.00016286681390011325,
      "loss": 1.9906,
      "step": 150019
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.1024820804595947,
      "learning_rate": 0.00016286317572657472,
      "loss": 2.0502,
      "step": 150020
    },
    {
      "epoch": 1.95,
      "grad_norm": 0.9928257465362549,
      "learning_rate": 0.0001628595375785325,
      "loss": 2.0484,
      "step": 150021
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.1965090036392212,
      "learning_rate": 0.00016285589945598715,
      "loss": 2.0223,
      "step": 150022
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.1012698411941528,
      "learning_rate": 0.0001628522613589396,
      "loss": 2.0607,
      "step": 150023
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.0390644073486328,
      "learning_rate": 0.0001628486232873903,
      "loss": 2.1206,
      "step": 150024
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.0965955257415771,
      "learning_rate": 0.0001628449852413402,
      "loss": 2.1184,
      "step": 150025
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.087391972541809,
      "learning_rate": 0.00016284134722078969,
      "loss": 2.1806,
      "step": 150026
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.103535532951355,
      "learning_rate": 0.00016283770922573962,
      "loss": 2.2491,
      "step": 150027
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.1615136861801147,
      "learning_rate": 0.0001628340712561905,
      "loss": 2.053,
      "step": 150028
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.0441211462020874,
      "learning_rate": 0.00016283043331214323,
      "loss": 1.8932,
      "step": 150029
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.0448650121688843,
      "learning_rate": 0.00016282679539359826,
      "loss": 1.9542,
      "step": 150030
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.1026434898376465,
      "learning_rate": 0.00016282315750055644,
      "loss": 1.931,
      "step": 150031
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.0452806949615479,
      "learning_rate": 0.00016281951963301845,
      "loss": 2.0892,
      "step": 150032
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.0847653150558472,
      "learning_rate": 0.0001628158817909849,
      "loss": 1.9899,
      "step": 150033
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.201827883720398,
      "learning_rate": 0.00016281224397445633,
      "loss": 2.0217,
      "step": 150034
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.3936717510223389,
      "learning_rate": 0.00016280860618343368,
      "loss": 2.0002,
      "step": 150035
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.2186275720596313,
      "learning_rate": 0.00016280496841791735,
      "loss": 2.2595,
      "step": 150036
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.1279577016830444,
      "learning_rate": 0.0001628013306779083,
      "loss": 2.0613,
      "step": 150037
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.1214932203292847,
      "learning_rate": 0.00016279769296340707,
      "loss": 2.1045,
      "step": 150038
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.1431739330291748,
      "learning_rate": 0.00016279405527441437,
      "loss": 2.2137,
      "step": 150039
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.1132886409759521,
      "learning_rate": 0.0001627904176109307,
      "loss": 2.1156,
      "step": 150040
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.0292423963546753,
      "learning_rate": 0.000162786779972957,
      "loss": 2.2181,
      "step": 150041
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.1336060762405396,
      "learning_rate": 0.00016278314236049373,
      "loss": 2.1718,
      "step": 150042
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.2242027521133423,
      "learning_rate": 0.00016277950477354173,
      "loss": 2.0889,
      "step": 150043
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.2114992141723633,
      "learning_rate": 0.00016277586721210168,
      "loss": 2.2342,
      "step": 150044
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.5402899980545044,
      "learning_rate": 0.00016277222967617415,
      "loss": 2.1426,
      "step": 150045
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.1559056043624878,
      "learning_rate": 0.00016276859216575972,
      "loss": 1.8159,
      "step": 150046
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.0811229944229126,
      "learning_rate": 0.00016276495468085938,
      "loss": 1.9668,
      "step": 150047
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.169664978981018,
      "learning_rate": 0.00016276131722147344,
      "loss": 2.0464,
      "step": 150048
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.1997807025909424,
      "learning_rate": 0.0001627576797876029,
      "loss": 1.9355,
      "step": 150049
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.0969231128692627,
      "learning_rate": 0.0001627540423792483,
      "loss": 2.0973,
      "step": 150050
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.1377402544021606,
      "learning_rate": 0.00016275040499641019,
      "loss": 2.0358,
      "step": 150051
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.4543501138687134,
      "learning_rate": 0.00016274676763908952,
      "loss": 2.2826,
      "step": 150052
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.1921494007110596,
      "learning_rate": 0.00016274313030728677,
      "loss": 2.0564,
      "step": 150053
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.2453060150146484,
      "learning_rate": 0.00016273949300100256,
      "loss": 2.3272,
      "step": 150054
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.1631464958190918,
      "learning_rate": 0.0001627358557202378,
      "loss": 2.1175,
      "step": 150055
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.3722811937332153,
      "learning_rate": 0.00016273221846499302,
      "loss": 1.9577,
      "step": 150056
    },
    {
      "epoch": 1.95,
      "grad_norm": 3.011467218399048,
      "learning_rate": 0.00016272858123526882,
      "loss": 1.8198,
      "step": 150057
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.2048693895339966,
      "learning_rate": 0.00016272494403106607,
      "loss": 2.0354,
      "step": 150058
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.04088294506073,
      "learning_rate": 0.00016272130685238536,
      "loss": 2.0462,
      "step": 150059
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.1071298122406006,
      "learning_rate": 0.00016271766969922717,
      "loss": 2.0839,
      "step": 150060
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.0773371458053589,
      "learning_rate": 0.00016271403257159256,
      "loss": 2.1011,
      "step": 150061
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.216657042503357,
      "learning_rate": 0.00016271039546948194,
      "loss": 2.0771,
      "step": 150062
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.0644656419754028,
      "learning_rate": 0.00016270675839289597,
      "loss": 2.1843,
      "step": 150063
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.1752045154571533,
      "learning_rate": 0.00016270312134183552,
      "loss": 2.1524,
      "step": 150064
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.130178689956665,
      "learning_rate": 0.00016269948431630097,
      "loss": 2.1149,
      "step": 150065
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.0844238996505737,
      "learning_rate": 0.00016269584731629338,
      "loss": 2.0555,
      "step": 150066
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.1138670444488525,
      "learning_rate": 0.00016269221034181317,
      "loss": 1.9615,
      "step": 150067
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.0456448793411255,
      "learning_rate": 0.00016268857339286097,
      "loss": 2.0549,
      "step": 150068
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.0802617073059082,
      "learning_rate": 0.00016268493646943768,
      "loss": 1.97,
      "step": 150069
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.242822289466858,
      "learning_rate": 0.00016268129957154386,
      "loss": 1.9429,
      "step": 150070
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.1649034023284912,
      "learning_rate": 0.00016267766269918002,
      "loss": 2.2276,
      "step": 150071
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.0833102464675903,
      "learning_rate": 0.00016267402585234717,
      "loss": 1.9735,
      "step": 150072
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.1713602542877197,
      "learning_rate": 0.00016267038903104567,
      "loss": 2.305,
      "step": 150073
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.1329734325408936,
      "learning_rate": 0.00016266675223527643,
      "loss": 2.1481,
      "step": 150074
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.1509947776794434,
      "learning_rate": 0.00016266311546504007,
      "loss": 1.9389,
      "step": 150075
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.1595813035964966,
      "learning_rate": 0.00016265947872033722,
      "loss": 1.8541,
      "step": 150076
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.1724671125411987,
      "learning_rate": 0.0001626558420011684,
      "loss": 2.0291,
      "step": 150077
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.1409519910812378,
      "learning_rate": 0.00016265220530753465,
      "loss": 2.2453,
      "step": 150078
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.0375909805297852,
      "learning_rate": 0.00016264856863943628,
      "loss": 2.0192,
      "step": 150079
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.225608229637146,
      "learning_rate": 0.00016264493199687427,
      "loss": 2.0969,
      "step": 150080
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.1955595016479492,
      "learning_rate": 0.00016264129537984915,
      "loss": 2.3057,
      "step": 150081
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.149785041809082,
      "learning_rate": 0.0001626376587883616,
      "loss": 2.1345,
      "step": 150082
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.091022253036499,
      "learning_rate": 0.00016263402222241219,
      "loss": 2.176,
      "step": 150083
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.0964419841766357,
      "learning_rate": 0.00016263038568200183,
      "loss": 1.9865,
      "step": 150084
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.251570224761963,
      "learning_rate": 0.00016262674916713097,
      "loss": 2.1482,
      "step": 150085
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.0868324041366577,
      "learning_rate": 0.00016262311267780047,
      "loss": 2.0181,
      "step": 150086
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.3371365070343018,
      "learning_rate": 0.00016261947621401096,
      "loss": 1.9321,
      "step": 150087
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.2258267402648926,
      "learning_rate": 0.00016261583977576305,
      "loss": 2.1806,
      "step": 150088
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.0472675561904907,
      "learning_rate": 0.00016261220336305735,
      "loss": 1.9631,
      "step": 150089
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.1401653289794922,
      "learning_rate": 0.00016260856697589474,
      "loss": 1.9877,
      "step": 150090
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.1486573219299316,
      "learning_rate": 0.00016260493061427572,
      "loss": 2.1218,
      "step": 150091
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.2125805616378784,
      "learning_rate": 0.0001626012942782011,
      "loss": 1.8076,
      "step": 150092
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.1040010452270508,
      "learning_rate": 0.00016259765796767153,
      "loss": 1.91,
      "step": 150093
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.012121319770813,
      "learning_rate": 0.00016259402168268764,
      "loss": 2.0801,
      "step": 150094
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.350728154182434,
      "learning_rate": 0.00016259038542324998,
      "loss": 2.2597,
      "step": 150095
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.2627499103546143,
      "learning_rate": 0.0001625867491893595,
      "loss": 2.0287,
      "step": 150096
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.106492042541504,
      "learning_rate": 0.0001625831129810166,
      "loss": 2.0674,
      "step": 150097
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.1119024753570557,
      "learning_rate": 0.00016257947679822223,
      "loss": 1.8509,
      "step": 150098
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.1372689008712769,
      "learning_rate": 0.00016257584064097692,
      "loss": 1.9278,
      "step": 150099
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.1458245515823364,
      "learning_rate": 0.00016257220450928132,
      "loss": 2.0174,
      "step": 150100
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.2224512100219727,
      "learning_rate": 0.0001625685684031361,
      "loss": 1.9994,
      "step": 150101
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.1187961101531982,
      "learning_rate": 0.00016256493232254206,
      "loss": 2.1801,
      "step": 150102
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.111966848373413,
      "learning_rate": 0.0001625612962674997,
      "loss": 1.8305,
      "step": 150103
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.0734434127807617,
      "learning_rate": 0.00016255766023800985,
      "loss": 2.227,
      "step": 150104
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.267371416091919,
      "learning_rate": 0.0001625540242340731,
      "loss": 2.1842,
      "step": 150105
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.1002936363220215,
      "learning_rate": 0.00016255038825569023,
      "loss": 2.2016,
      "step": 150106
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.022973656654358,
      "learning_rate": 0.00016254675230286173,
      "loss": 1.8558,
      "step": 150107
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.117958903312683,
      "learning_rate": 0.0001625431163755884,
      "loss": 2.2482,
      "step": 150108
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.1370151042938232,
      "learning_rate": 0.00016253948047387088,
      "loss": 2.1856,
      "step": 150109
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.1763359308242798,
      "learning_rate": 0.00016253584459770994,
      "loss": 2.1617,
      "step": 150110
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.1370222568511963,
      "learning_rate": 0.00016253220874710622,
      "loss": 2.1006,
      "step": 150111
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.2560385465621948,
      "learning_rate": 0.00016252857292206032,
      "loss": 2.0219,
      "step": 150112
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.2041398286819458,
      "learning_rate": 0.00016252493712257284,
      "loss": 2.1359,
      "step": 150113
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.193184494972229,
      "learning_rate": 0.00016252130134864468,
      "loss": 2.3433,
      "step": 150114
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.2118629217147827,
      "learning_rate": 0.00016251766560027627,
      "loss": 2.0441,
      "step": 150115
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.0692423582077026,
      "learning_rate": 0.0001625140298774686,
      "loss": 1.7233,
      "step": 150116
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.104651927947998,
      "learning_rate": 0.00016251039418022214,
      "loss": 2.1871,
      "step": 150117
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.0897927284240723,
      "learning_rate": 0.00016250675850853756,
      "loss": 2.0464,
      "step": 150118
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.0827988386154175,
      "learning_rate": 0.00016250312286241546,
      "loss": 2.2039,
      "step": 150119
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.2637139558792114,
      "learning_rate": 0.00016249948724185678,
      "loss": 1.9578,
      "step": 150120
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.1826610565185547,
      "learning_rate": 0.00016249585164686191,
      "loss": 1.7673,
      "step": 150121
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.195185899734497,
      "learning_rate": 0.00016249221607743177,
      "loss": 2.2967,
      "step": 150122
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.1041624546051025,
      "learning_rate": 0.0001624885805335669,
      "loss": 2.0712,
      "step": 150123
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.113186001777649,
      "learning_rate": 0.00016248494501526804,
      "loss": 2.0751,
      "step": 150124
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.3864400386810303,
      "learning_rate": 0.00016248130952253569,
      "loss": 2.2903,
      "step": 150125
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.2367994785308838,
      "learning_rate": 0.00016247767405537077,
      "loss": 2.1914,
      "step": 150126
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.0915203094482422,
      "learning_rate": 0.00016247403861377376,
      "loss": 1.948,
      "step": 150127
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.2693485021591187,
      "learning_rate": 0.00016247040319774546,
      "loss": 2.2388,
      "step": 150128
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.214159607887268,
      "learning_rate": 0.0001624667678072866,
      "loss": 2.2957,
      "step": 150129
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.1358479261398315,
      "learning_rate": 0.0001624631324423977,
      "loss": 1.9468,
      "step": 150130
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.2250168323516846,
      "learning_rate": 0.00016245949710307943,
      "loss": 2.2106,
      "step": 150131
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.1928441524505615,
      "learning_rate": 0.00016245586178933262,
      "loss": 1.8424,
      "step": 150132
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.6573399305343628,
      "learning_rate": 0.00016245222650115775,
      "loss": 2.1407,
      "step": 150133
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.1497926712036133,
      "learning_rate": 0.00016244859123855571,
      "loss": 2.1442,
      "step": 150134
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.137300968170166,
      "learning_rate": 0.00016244495600152697,
      "loss": 2.2564,
      "step": 150135
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.1433053016662598,
      "learning_rate": 0.00016244132079007247,
      "loss": 2.1607,
      "step": 150136
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.0655813217163086,
      "learning_rate": 0.00016243768560419267,
      "loss": 1.8294,
      "step": 150137
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.0976238250732422,
      "learning_rate": 0.0001624340504438883,
      "loss": 2.1695,
      "step": 150138
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.1882972717285156,
      "learning_rate": 0.00016243041530915995,
      "loss": 1.7849,
      "step": 150139
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.2781100273132324,
      "learning_rate": 0.00016242678020000851,
      "loss": 1.9907,
      "step": 150140
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.1778771877288818,
      "learning_rate": 0.00016242314511643439,
      "loss": 2.1491,
      "step": 150141
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.131441593170166,
      "learning_rate": 0.00016241951005843848,
      "loss": 2.1332,
      "step": 150142
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.1254589557647705,
      "learning_rate": 0.00016241587502602146,
      "loss": 2.0832,
      "step": 150143
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.1832149028778076,
      "learning_rate": 0.00016241224001918388,
      "loss": 2.2529,
      "step": 150144
    },
    {
      "epoch": 1.95,
      "grad_norm": 1.5060133934020996,
      "learning_rate": 0.00016240860503792637,
      "loss": 2.2506,
      "step": 150145
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.1062169075012207,
      "learning_rate": 0.00016240497008224977,
      "loss": 2.1505,
      "step": 150146
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.121985912322998,
      "learning_rate": 0.0001624013351521546,
      "loss": 2.0469,
      "step": 150147
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.1911855936050415,
      "learning_rate": 0.00016239770024764174,
      "loss": 2.0316,
      "step": 150148
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.3098808526992798,
      "learning_rate": 0.00016239406536871172,
      "loss": 2.3741,
      "step": 150149
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.106825351715088,
      "learning_rate": 0.00016239043051536513,
      "loss": 2.3908,
      "step": 150150
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.2399502992630005,
      "learning_rate": 0.00016238679568760287,
      "loss": 2.0155,
      "step": 150151
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.344212293624878,
      "learning_rate": 0.00016238316088542553,
      "loss": 2.3501,
      "step": 150152
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.2542619705200195,
      "learning_rate": 0.00016237952610883365,
      "loss": 2.1128,
      "step": 150153
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.1207574605941772,
      "learning_rate": 0.00016237589135782807,
      "loss": 2.2142,
      "step": 150154
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.1410340070724487,
      "learning_rate": 0.00016237225663240945,
      "loss": 2.1701,
      "step": 150155
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.1482551097869873,
      "learning_rate": 0.0001623686219325783,
      "loss": 2.025,
      "step": 150156
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.2889816761016846,
      "learning_rate": 0.00016236498725833554,
      "loss": 1.8471,
      "step": 150157
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.211593508720398,
      "learning_rate": 0.0001623613526096816,
      "loss": 1.8679,
      "step": 150158
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.1908586025238037,
      "learning_rate": 0.00016235771798661744,
      "loss": 1.8851,
      "step": 150159
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.1310116052627563,
      "learning_rate": 0.00016235408338914354,
      "loss": 2.2376,
      "step": 150160
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.4084761142730713,
      "learning_rate": 0.0001623504488172606,
      "loss": 2.0072,
      "step": 150161
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.4534153938293457,
      "learning_rate": 0.00016234681427096926,
      "loss": 2.0901,
      "step": 150162
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.3317639827728271,
      "learning_rate": 0.00016234317975027032,
      "loss": 1.9031,
      "step": 150163
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.0674428939819336,
      "learning_rate": 0.00016233954525516424,
      "loss": 2.1316,
      "step": 150164
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.0923185348510742,
      "learning_rate": 0.00016233591078565195,
      "loss": 2.2409,
      "step": 150165
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.5967578887939453,
      "learning_rate": 0.00016233227634173406,
      "loss": 1.8767,
      "step": 150166
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.7197723388671875,
      "learning_rate": 0.00016232864192341118,
      "loss": 2.0711,
      "step": 150167
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.470054268836975,
      "learning_rate": 0.00016232500753068387,
      "loss": 2.0472,
      "step": 150168
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.1239362955093384,
      "learning_rate": 0.00016232137316355308,
      "loss": 2.1162,
      "step": 150169
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.2939354181289673,
      "learning_rate": 0.00016231773882201923,
      "loss": 2.1105,
      "step": 150170
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.6869056224822998,
      "learning_rate": 0.00016231410450608325,
      "loss": 2.1978,
      "step": 150171
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.9656445980072021,
      "learning_rate": 0.00016231047021574564,
      "loss": 2.0353,
      "step": 150172
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.1942309141159058,
      "learning_rate": 0.0001623068359510071,
      "loss": 1.9415,
      "step": 150173
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.125032901763916,
      "learning_rate": 0.00016230320171186822,
      "loss": 2.0222,
      "step": 150174
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.374318242073059,
      "learning_rate": 0.00016229956749832988,
      "loss": 2.0287,
      "step": 150175
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.72836172580719,
      "learning_rate": 0.00016229593331039256,
      "loss": 2.0705,
      "step": 150176
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.4763239622116089,
      "learning_rate": 0.0001622922991480571,
      "loss": 2.2433,
      "step": 150177
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.1058534383773804,
      "learning_rate": 0.00016228866501132415,
      "loss": 1.7467,
      "step": 150178
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.1805428266525269,
      "learning_rate": 0.00016228503090019435,
      "loss": 2.0535,
      "step": 150179
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.279473900794983,
      "learning_rate": 0.00016228139681466816,
      "loss": 2.0485,
      "step": 150180
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.4751094579696655,
      "learning_rate": 0.00016227776275474663,
      "loss": 2.0094,
      "step": 150181
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.140173316001892,
      "learning_rate": 0.0001622741287204302,
      "loss": 2.2229,
      "step": 150182
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.1449528932571411,
      "learning_rate": 0.00016227049471171962,
      "loss": 2.0605,
      "step": 150183
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.1447948217391968,
      "learning_rate": 0.0001622668607286156,
      "loss": 2.0383,
      "step": 150184
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.3120392560958862,
      "learning_rate": 0.00016226322677111876,
      "loss": 2.1315,
      "step": 150185
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.3858063220977783,
      "learning_rate": 0.0001622595928392297,
      "loss": 2.0958,
      "step": 150186
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.1338156461715698,
      "learning_rate": 0.00016225595893294928,
      "loss": 1.9466,
      "step": 150187
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.05892813205719,
      "learning_rate": 0.00016225232505227795,
      "loss": 2.0905,
      "step": 150188
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.0670987367630005,
      "learning_rate": 0.00016224869119721666,
      "loss": 1.9895,
      "step": 150189
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.1275004148483276,
      "learning_rate": 0.0001622450573677659,
      "loss": 2.1011,
      "step": 150190
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.1999822854995728,
      "learning_rate": 0.0001622414235639264,
      "loss": 2.1511,
      "step": 150191
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.196694016456604,
      "learning_rate": 0.00016223778978569873,
      "loss": 2.0997,
      "step": 150192
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.0449302196502686,
      "learning_rate": 0.00016223415603308373,
      "loss": 2.0182,
      "step": 150193
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.1764949560165405,
      "learning_rate": 0.0001622305223060819,
      "loss": 2.1055,
      "step": 150194
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.375590443611145,
      "learning_rate": 0.00016222688860469418,
      "loss": 2.3002,
      "step": 150195
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.1416176557540894,
      "learning_rate": 0.00016222325492892104,
      "loss": 2.1409,
      "step": 150196
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.0736587047576904,
      "learning_rate": 0.00016221962127876318,
      "loss": 2.2306,
      "step": 150197
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.0780987739562988,
      "learning_rate": 0.0001622159876542212,
      "loss": 1.8823,
      "step": 150198
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.054183840751648,
      "learning_rate": 0.00016221235405529594,
      "loss": 2.1016,
      "step": 150199
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.107346773147583,
      "learning_rate": 0.00016220872048198793,
      "loss": 2.1808,
      "step": 150200
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.0696600675582886,
      "learning_rate": 0.00016220508693429804,
      "loss": 2.0339,
      "step": 150201
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.1739401817321777,
      "learning_rate": 0.00016220145341222668,
      "loss": 2.1903,
      "step": 150202
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.026775598526001,
      "learning_rate": 0.00016219781991577497,
      "loss": 2.0759,
      "step": 150203
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.0247005224227905,
      "learning_rate": 0.00016219418644494298,
      "loss": 1.9792,
      "step": 150204
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.0665007829666138,
      "learning_rate": 0.00016219055299973185,
      "loss": 2.1477,
      "step": 150205
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.2961641550064087,
      "learning_rate": 0.000162186919580142,
      "loss": 2.0103,
      "step": 150206
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.1880722045898438,
      "learning_rate": 0.00016218328618617425,
      "loss": 2.0811,
      "step": 150207
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.1471956968307495,
      "learning_rate": 0.00016217965281782915,
      "loss": 2.0354,
      "step": 150208
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.0539544820785522,
      "learning_rate": 0.00016217601947510768,
      "loss": 2.1973,
      "step": 150209
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.955274224281311,
      "learning_rate": 0.0001621723861580101,
      "loss": 2.1003,
      "step": 150210
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.1850618124008179,
      "learning_rate": 0.00016216875286653734,
      "loss": 2.124,
      "step": 150211
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.019877314567566,
      "learning_rate": 0.0001621651196006899,
      "loss": 1.7673,
      "step": 150212
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.112830638885498,
      "learning_rate": 0.00016216148636046868,
      "loss": 1.8897,
      "step": 150213
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.15507173538208,
      "learning_rate": 0.00016215785314587414,
      "loss": 2.1219,
      "step": 150214
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.5793869495391846,
      "learning_rate": 0.00016215421995690728,
      "loss": 1.9294,
      "step": 150215
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.1188879013061523,
      "learning_rate": 0.0001621505867935683,
      "loss": 1.9807,
      "step": 150216
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.1221671104431152,
      "learning_rate": 0.0001621469536558583,
      "loss": 2.2999,
      "step": 150217
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.2131175994873047,
      "learning_rate": 0.00016214332054377765,
      "loss": 2.2124,
      "step": 150218
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.4286837577819824,
      "learning_rate": 0.00016213968745732723,
      "loss": 2.0468,
      "step": 150219
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.0483297109603882,
      "learning_rate": 0.0001621360543965076,
      "loss": 2.0533,
      "step": 150220
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.1839439868927002,
      "learning_rate": 0.00016213242136131967,
      "loss": 1.978,
      "step": 150221
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.2013375759124756,
      "learning_rate": 0.00016212878835176372,
      "loss": 2.1158,
      "step": 150222
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.4303158521652222,
      "learning_rate": 0.0001621251553678407,
      "loss": 2.0602,
      "step": 150223
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.1964887380599976,
      "learning_rate": 0.0001621215224095511,
      "loss": 1.9092,
      "step": 150224
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.1718082427978516,
      "learning_rate": 0.00016211788947689587,
      "loss": 1.9834,
      "step": 150225
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.2582610845565796,
      "learning_rate": 0.00016211425656987538,
      "loss": 1.9734,
      "step": 150226
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.2041041851043701,
      "learning_rate": 0.00016211062368849056,
      "loss": 2.1568,
      "step": 150227
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.4195911884307861,
      "learning_rate": 0.00016210699083274202,
      "loss": 2.1163,
      "step": 150228
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.1112064123153687,
      "learning_rate": 0.0001621033580026304,
      "loss": 1.8845,
      "step": 150229
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.093180775642395,
      "learning_rate": 0.0001620997251981562,
      "loss": 2.2227,
      "step": 150230
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.174034833908081,
      "learning_rate": 0.00016209609241932043,
      "loss": 2.1839,
      "step": 150231
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.181471586227417,
      "learning_rate": 0.00016209245966612343,
      "loss": 2.1166,
      "step": 150232
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.1450828313827515,
      "learning_rate": 0.0001620888269385662,
      "loss": 2.1882,
      "step": 150233
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.0142062902450562,
      "learning_rate": 0.00016208519423664924,
      "loss": 1.9535,
      "step": 150234
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.2369612455368042,
      "learning_rate": 0.00016208156156037314,
      "loss": 2.2113,
      "step": 150235
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.0773096084594727,
      "learning_rate": 0.00016207792890973884,
      "loss": 2.0304,
      "step": 150236
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.1552239656448364,
      "learning_rate": 0.0001620742962847468,
      "loss": 2.1139,
      "step": 150237
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.083242654800415,
      "learning_rate": 0.00016207066368539763,
      "loss": 1.9809,
      "step": 150238
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.1872910261154175,
      "learning_rate": 0.00016206703111169225,
      "loss": 2.0884,
      "step": 150239
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.1131162643432617,
      "learning_rate": 0.00016206339856363123,
      "loss": 2.1456,
      "step": 150240
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.1289961338043213,
      "learning_rate": 0.00016205976604121512,
      "loss": 1.8617,
      "step": 150241
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.1309282779693604,
      "learning_rate": 0.0001620561335444448,
      "loss": 2.2108,
      "step": 150242
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.1538091897964478,
      "learning_rate": 0.00016205250107332084,
      "loss": 2.1831,
      "step": 150243
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.2457505464553833,
      "learning_rate": 0.00016204886862784384,
      "loss": 2.3709,
      "step": 150244
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.1105910539627075,
      "learning_rate": 0.00016204523620801464,
      "loss": 2.3516,
      "step": 150245
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.2303972244262695,
      "learning_rate": 0.00016204160381383387,
      "loss": 2.0226,
      "step": 150246
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.071328043937683,
      "learning_rate": 0.00016203797144530204,
      "loss": 2.2763,
      "step": 150247
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.20670485496521,
      "learning_rate": 0.00016203433910242005,
      "loss": 1.9404,
      "step": 150248
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.3443514108657837,
      "learning_rate": 0.0001620307067851884,
      "loss": 2.0521,
      "step": 150249
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.3968040943145752,
      "learning_rate": 0.00016202707449360792,
      "loss": 2.0595,
      "step": 150250
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.217656135559082,
      "learning_rate": 0.00016202344222767924,
      "loss": 1.9621,
      "step": 150251
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.1402008533477783,
      "learning_rate": 0.000162019809987403,
      "loss": 2.2339,
      "step": 150252
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.197550654411316,
      "learning_rate": 0.0001620161777727798,
      "loss": 2.1935,
      "step": 150253
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.1492409706115723,
      "learning_rate": 0.00016201254558381048,
      "loss": 2.2033,
      "step": 150254
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.0967212915420532,
      "learning_rate": 0.0001620089134204955,
      "loss": 2.1196,
      "step": 150255
    },
    {
      "epoch": 1.96,
      "grad_norm": 0.9736312627792358,
      "learning_rate": 0.00016200528128283581,
      "loss": 2.2555,
      "step": 150256
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.0867812633514404,
      "learning_rate": 0.00016200164917083197,
      "loss": 1.8281,
      "step": 150257
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.1651899814605713,
      "learning_rate": 0.00016199801708448455,
      "loss": 2.1432,
      "step": 150258
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.1398138999938965,
      "learning_rate": 0.00016199438502379428,
      "loss": 2.3138,
      "step": 150259
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.096686601638794,
      "learning_rate": 0.00016199075298876188,
      "loss": 2.0765,
      "step": 150260
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.1375638246536255,
      "learning_rate": 0.00016198712097938796,
      "loss": 2.1392,
      "step": 150261
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.0972388982772827,
      "learning_rate": 0.0001619834889956733,
      "loss": 2.1544,
      "step": 150262
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.16118586063385,
      "learning_rate": 0.00016197985703761856,
      "loss": 2.0235,
      "step": 150263
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.0747777223587036,
      "learning_rate": 0.00016197622510522437,
      "loss": 1.9312,
      "step": 150264
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.1156882047653198,
      "learning_rate": 0.00016197259319849128,
      "loss": 1.9573,
      "step": 150265
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.2283753156661987,
      "learning_rate": 0.0001619689613174202,
      "loss": 2.0319,
      "step": 150266
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.177736759185791,
      "learning_rate": 0.00016196532946201157,
      "loss": 2.0343,
      "step": 150267
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.1312878131866455,
      "learning_rate": 0.0001619616976322663,
      "loss": 2.0681,
      "step": 150268
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.2056373357772827,
      "learning_rate": 0.00016195806582818486,
      "loss": 1.905,
      "step": 150269
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.143165111541748,
      "learning_rate": 0.00016195443404976826,
      "loss": 2.1077,
      "step": 150270
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.1186041831970215,
      "learning_rate": 0.0001619508022970167,
      "loss": 2.0388,
      "step": 150271
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.1251994371414185,
      "learning_rate": 0.00016194717056993119,
      "loss": 2.1514,
      "step": 150272
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.1203919649124146,
      "learning_rate": 0.0001619435388685122,
      "loss": 2.0365,
      "step": 150273
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.107419490814209,
      "learning_rate": 0.00016193990719276063,
      "loss": 2.0825,
      "step": 150274
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.1970999240875244,
      "learning_rate": 0.00016193627554267692,
      "loss": 2.1873,
      "step": 150275
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.0865542888641357,
      "learning_rate": 0.00016193264391826213,
      "loss": 1.8105,
      "step": 150276
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.2543057203292847,
      "learning_rate": 0.00016192901231951637,
      "loss": 2.1533,
      "step": 150277
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.2718859910964966,
      "learning_rate": 0.00016192538074644074,
      "loss": 2.0328,
      "step": 150278
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.0944328308105469,
      "learning_rate": 0.00016192174919903572,
      "loss": 2.0955,
      "step": 150279
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.1675115823745728,
      "learning_rate": 0.00016191811767730216,
      "loss": 2.0597,
      "step": 150280
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.1985596418380737,
      "learning_rate": 0.0001619144861812405,
      "loss": 1.9852,
      "step": 150281
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.2470775842666626,
      "learning_rate": 0.00016191085471085176,
      "loss": 1.9349,
      "step": 150282
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.2172191143035889,
      "learning_rate": 0.00016190722326613616,
      "loss": 1.9398,
      "step": 150283
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.2024849653244019,
      "learning_rate": 0.00016190359184709477,
      "loss": 2.0527,
      "step": 150284
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.1897761821746826,
      "learning_rate": 0.00016189996045372796,
      "loss": 1.9708,
      "step": 150285
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.0802077054977417,
      "learning_rate": 0.00016189632908603666,
      "loss": 2.188,
      "step": 150286
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.2974315881729126,
      "learning_rate": 0.00016189269774402138,
      "loss": 2.117,
      "step": 150287
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.0624363422393799,
      "learning_rate": 0.000161889066427683,
      "loss": 1.9256,
      "step": 150288
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.2013314962387085,
      "learning_rate": 0.00016188543513702191,
      "loss": 2.0189,
      "step": 150289
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.1814402341842651,
      "learning_rate": 0.00016188180387203898,
      "loss": 2.09,
      "step": 150290
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.091814637184143,
      "learning_rate": 0.00016187817263273472,
      "loss": 2.0428,
      "step": 150291
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.2767497301101685,
      "learning_rate": 0.00016187454141911004,
      "loss": 2.2426,
      "step": 150292
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.1296122074127197,
      "learning_rate": 0.0001618709102311654,
      "loss": 2.1324,
      "step": 150293
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.186985731124878,
      "learning_rate": 0.00016186727906890178,
      "loss": 2.0901,
      "step": 150294
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.0670199394226074,
      "learning_rate": 0.0001618636479323194,
      "loss": 2.0333,
      "step": 150295
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.1438766717910767,
      "learning_rate": 0.0001618600168214193,
      "loss": 2.1462,
      "step": 150296
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.053201675415039,
      "learning_rate": 0.0001618563857362019,
      "loss": 2.2089,
      "step": 150297
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.0154169797897339,
      "learning_rate": 0.00016185275467666812,
      "loss": 2.179,
      "step": 150298
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.137639045715332,
      "learning_rate": 0.00016184912364281847,
      "loss": 1.8433,
      "step": 150299
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.1179742813110352,
      "learning_rate": 0.0001618454926346538,
      "loss": 2.1554,
      "step": 150300
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.164189100265503,
      "learning_rate": 0.0001618418616521745,
      "loss": 2.2242,
      "step": 150301
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.1531673669815063,
      "learning_rate": 0.0001618382306953815,
      "loss": 2.0075,
      "step": 150302
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.2092409133911133,
      "learning_rate": 0.0001618345997642753,
      "loss": 2.1041,
      "step": 150303
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.0903791189193726,
      "learning_rate": 0.00016183096885885668,
      "loss": 1.9076,
      "step": 150304
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.031723976135254,
      "learning_rate": 0.0001618273379791263,
      "loss": 2.1495,
      "step": 150305
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.1491167545318604,
      "learning_rate": 0.00016182370712508495,
      "loss": 2.1713,
      "step": 150306
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.139438509941101,
      "learning_rate": 0.000161820076296733,
      "loss": 2.2447,
      "step": 150307
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.0842338800430298,
      "learning_rate": 0.0001618164454940714,
      "loss": 1.9553,
      "step": 150308
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.1272586584091187,
      "learning_rate": 0.00016181281471710064,
      "loss": 2.1435,
      "step": 150309
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.1988239288330078,
      "learning_rate": 0.0001618091839658216,
      "loss": 2.02,
      "step": 150310
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.0295698642730713,
      "learning_rate": 0.00016180555324023474,
      "loss": 1.8184,
      "step": 150311
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.1594375371932983,
      "learning_rate": 0.0001618019225403409,
      "loss": 2.2807,
      "step": 150312
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.1237361431121826,
      "learning_rate": 0.00016179829186614074,
      "loss": 2.001,
      "step": 150313
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.147081732749939,
      "learning_rate": 0.00016179466121763486,
      "loss": 2.0016,
      "step": 150314
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.0716031789779663,
      "learning_rate": 0.00016179103059482384,
      "loss": 2.4376,
      "step": 150315
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.0834795236587524,
      "learning_rate": 0.00016178739999770864,
      "loss": 2.0808,
      "step": 150316
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.3578308820724487,
      "learning_rate": 0.0001617837694262896,
      "loss": 2.2181,
      "step": 150317
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.3032289743423462,
      "learning_rate": 0.0001617801388805677,
      "loss": 2.0017,
      "step": 150318
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.194738745689392,
      "learning_rate": 0.0001617765083605435,
      "loss": 2.1493,
      "step": 150319
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.1737587451934814,
      "learning_rate": 0.00016177287786621749,
      "loss": 2.0276,
      "step": 150320
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.1911565065383911,
      "learning_rate": 0.0001617692473975907,
      "loss": 2.2754,
      "step": 150321
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.1370302438735962,
      "learning_rate": 0.00016176561695466354,
      "loss": 2.0751,
      "step": 150322
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.11050283908844,
      "learning_rate": 0.00016176198653743672,
      "loss": 1.988,
      "step": 150323
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.322898507118225,
      "learning_rate": 0.00016175835614591104,
      "loss": 2.3814,
      "step": 150324
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.1543647050857544,
      "learning_rate": 0.00016175472578008708,
      "loss": 2.2586,
      "step": 150325
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.5842114686965942,
      "learning_rate": 0.00016175109543996538,
      "loss": 2.092,
      "step": 150326
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.491623878479004,
      "learning_rate": 0.00016174746512554695,
      "loss": 2.168,
      "step": 150327
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.1223030090332031,
      "learning_rate": 0.00016174383483683223,
      "loss": 2.0659,
      "step": 150328
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.1417145729064941,
      "learning_rate": 0.00016174020457382184,
      "loss": 2.0921,
      "step": 150329
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.139953374862671,
      "learning_rate": 0.00016173657433651666,
      "loss": 2.0972,
      "step": 150330
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.0820646286010742,
      "learning_rate": 0.00016173294412491726,
      "loss": 2.3186,
      "step": 150331
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.1659682989120483,
      "learning_rate": 0.0001617293139390242,
      "loss": 2.0433,
      "step": 150332
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.0730082988739014,
      "learning_rate": 0.00016172568377883842,
      "loss": 1.8627,
      "step": 150333
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.1150569915771484,
      "learning_rate": 0.00016172205364436027,
      "loss": 1.9385,
      "step": 150334
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.1418275833129883,
      "learning_rate": 0.00016171842353559075,
      "loss": 1.9869,
      "step": 150335
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.114068865776062,
      "learning_rate": 0.0001617147934525304,
      "loss": 2.0728,
      "step": 150336
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.2106742858886719,
      "learning_rate": 0.00016171116339517986,
      "loss": 2.0721,
      "step": 150337
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.4172437191009521,
      "learning_rate": 0.0001617075333635397,
      "loss": 2.0511,
      "step": 150338
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.190262794494629,
      "learning_rate": 0.00016170390335761088,
      "loss": 2.0222,
      "step": 150339
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.0803263187408447,
      "learning_rate": 0.00016170027337739375,
      "loss": 1.85,
      "step": 150340
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.2016469240188599,
      "learning_rate": 0.00016169664342288931,
      "loss": 2.0872,
      "step": 150341
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.0891432762145996,
      "learning_rate": 0.00016169301349409804,
      "loss": 2.0409,
      "step": 150342
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.0996263027191162,
      "learning_rate": 0.00016168938359102064,
      "loss": 2.1982,
      "step": 150343
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.0160720348358154,
      "learning_rate": 0.00016168575371365768,
      "loss": 2.0353,
      "step": 150344
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.085709810256958,
      "learning_rate": 0.0001616821238620101,
      "loss": 2.0601,
      "step": 150345
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.2468137741088867,
      "learning_rate": 0.00016167849403607827,
      "loss": 1.9264,
      "step": 150346
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.1141456365585327,
      "learning_rate": 0.00016167486423586316,
      "loss": 2.0056,
      "step": 150347
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.1694574356079102,
      "learning_rate": 0.0001616712344613652,
      "loss": 2.0742,
      "step": 150348
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.2220205068588257,
      "learning_rate": 0.0001616676047125854,
      "loss": 2.0954,
      "step": 150349
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.065995454788208,
      "learning_rate": 0.00016166397498952394,
      "loss": 2.1567,
      "step": 150350
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.1142973899841309,
      "learning_rate": 0.00016166034529218189,
      "loss": 2.075,
      "step": 150351
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.0447207689285278,
      "learning_rate": 0.00016165671562055968,
      "loss": 2.1583,
      "step": 150352
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.270779013633728,
      "learning_rate": 0.00016165308597465818,
      "loss": 2.209,
      "step": 150353
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.0751605033874512,
      "learning_rate": 0.0001616494563544779,
      "loss": 2.0782,
      "step": 150354
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.0629897117614746,
      "learning_rate": 0.00016164582676001985,
      "loss": 2.0962,
      "step": 150355
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.0677247047424316,
      "learning_rate": 0.00016164219719128422,
      "loss": 2.015,
      "step": 150356
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.125746250152588,
      "learning_rate": 0.000161638567648272,
      "loss": 2.2184,
      "step": 150357
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.097176194190979,
      "learning_rate": 0.00016163493813098367,
      "loss": 1.8525,
      "step": 150358
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.1151245832443237,
      "learning_rate": 0.00016163130863942014,
      "loss": 2.191,
      "step": 150359
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.1070953607559204,
      "learning_rate": 0.00016162767917358187,
      "loss": 2.0965,
      "step": 150360
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.2972087860107422,
      "learning_rate": 0.0001616240497334698,
      "loss": 2.0515,
      "step": 150361
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.2304973602294922,
      "learning_rate": 0.00016162042031908424,
      "loss": 1.8649,
      "step": 150362
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.119093418121338,
      "learning_rate": 0.00016161679093042617,
      "loss": 2.1166,
      "step": 150363
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.1921814680099487,
      "learning_rate": 0.00016161316156749605,
      "loss": 2.1795,
      "step": 150364
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.113857626914978,
      "learning_rate": 0.00016160953223029473,
      "loss": 2.0703,
      "step": 150365
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.1750155687332153,
      "learning_rate": 0.0001616059029188227,
      "loss": 1.9063,
      "step": 150366
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.0717315673828125,
      "learning_rate": 0.00016160227363308099,
      "loss": 1.9285,
      "step": 150367
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.1171358823776245,
      "learning_rate": 0.0001615986443730698,
      "loss": 1.8555,
      "step": 150368
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.1135342121124268,
      "learning_rate": 0.00016159501513879015,
      "loss": 1.9902,
      "step": 150369
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.06912100315094,
      "learning_rate": 0.00016159138593024248,
      "loss": 2.1829,
      "step": 150370
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.1716984510421753,
      "learning_rate": 0.00016158775674742775,
      "loss": 2.1045,
      "step": 150371
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.0849823951721191,
      "learning_rate": 0.0001615841275903463,
      "loss": 1.906,
      "step": 150372
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.275797724723816,
      "learning_rate": 0.00016158049845899918,
      "loss": 2.097,
      "step": 150373
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.1694488525390625,
      "learning_rate": 0.00016157686935338667,
      "loss": 2.1295,
      "step": 150374
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.038383960723877,
      "learning_rate": 0.00016157324027350975,
      "loss": 1.7686,
      "step": 150375
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.0718008279800415,
      "learning_rate": 0.00016156961121936882,
      "loss": 1.9354,
      "step": 150376
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.1048624515533447,
      "learning_rate": 0.00016156598219096486,
      "loss": 2.0062,
      "step": 150377
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.1416195631027222,
      "learning_rate": 0.0001615623531882983,
      "loss": 1.9458,
      "step": 150378
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.0611274242401123,
      "learning_rate": 0.00016155872421137011,
      "loss": 2.1357,
      "step": 150379
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.109991192817688,
      "learning_rate": 0.00016155509526018054,
      "loss": 1.9675,
      "step": 150380
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.1996660232543945,
      "learning_rate": 0.0001615514663347306,
      "loss": 2.0496,
      "step": 150381
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.0432045459747314,
      "learning_rate": 0.0001615478374350208,
      "loss": 2.0051,
      "step": 150382
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.097838282585144,
      "learning_rate": 0.0001615442085610519,
      "loss": 2.1888,
      "step": 150383
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.1331764459609985,
      "learning_rate": 0.00016154057971282449,
      "loss": 2.2465,
      "step": 150384
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.0958927869796753,
      "learning_rate": 0.0001615369508903395,
      "loss": 2.2686,
      "step": 150385
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.184154987335205,
      "learning_rate": 0.00016153332209359718,
      "loss": 2.2083,
      "step": 150386
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.0774664878845215,
      "learning_rate": 0.00016152969332259857,
      "loss": 1.9403,
      "step": 150387
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.2715237140655518,
      "learning_rate": 0.00016152606457734407,
      "loss": 1.8978,
      "step": 150388
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.1839274168014526,
      "learning_rate": 0.0001615224358578346,
      "loss": 2.0243,
      "step": 150389
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.145256519317627,
      "learning_rate": 0.00016151880716407062,
      "loss": 1.8834,
      "step": 150390
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.1044262647628784,
      "learning_rate": 0.00016151517849605308,
      "loss": 2.0803,
      "step": 150391
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.4849143028259277,
      "learning_rate": 0.00016151154985378233,
      "loss": 2.1744,
      "step": 150392
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.1888372898101807,
      "learning_rate": 0.00016150792123725927,
      "loss": 2.2427,
      "step": 150393
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.1230565309524536,
      "learning_rate": 0.0001615042926464844,
      "loss": 2.0161,
      "step": 150394
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.2510758638381958,
      "learning_rate": 0.0001615006640814586,
      "loss": 2.1009,
      "step": 150395
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.3410956859588623,
      "learning_rate": 0.00016149703554218237,
      "loss": 2.0466,
      "step": 150396
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.0648198127746582,
      "learning_rate": 0.00016149340702865651,
      "loss": 2.0692,
      "step": 150397
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.0537420511245728,
      "learning_rate": 0.00016148977854088168,
      "loss": 1.954,
      "step": 150398
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.280335783958435,
      "learning_rate": 0.0001614861500788585,
      "loss": 2.2234,
      "step": 150399
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.0353046655654907,
      "learning_rate": 0.00016148252164258753,
      "loss": 2.1408,
      "step": 150400
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.1582012176513672,
      "learning_rate": 0.00016147889323206972,
      "loss": 2.0992,
      "step": 150401
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.2253705263137817,
      "learning_rate": 0.00016147526484730544,
      "loss": 2.0503,
      "step": 150402
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.2358521223068237,
      "learning_rate": 0.0001614716364882957,
      "loss": 2.154,
      "step": 150403
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.3024928569793701,
      "learning_rate": 0.00016146800815504094,
      "loss": 2.1013,
      "step": 150404
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.2534674406051636,
      "learning_rate": 0.00016146437984754193,
      "loss": 2.0161,
      "step": 150405
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.0555943250656128,
      "learning_rate": 0.0001614607515657992,
      "loss": 2.141,
      "step": 150406
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.1075758934020996,
      "learning_rate": 0.0001614571233098136,
      "loss": 2.1422,
      "step": 150407
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.1414822340011597,
      "learning_rate": 0.00016145349507958563,
      "loss": 2.0378,
      "step": 150408
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.2477096319198608,
      "learning_rate": 0.0001614498668751162,
      "loss": 2.0783,
      "step": 150409
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.2153102159500122,
      "learning_rate": 0.00016144623869640586,
      "loss": 2.0457,
      "step": 150410
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.1132152080535889,
      "learning_rate": 0.00016144261054345515,
      "loss": 2.1245,
      "step": 150411
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.211507797241211,
      "learning_rate": 0.000161438982416265,
      "loss": 2.2166,
      "step": 150412
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.3752306699752808,
      "learning_rate": 0.00016143535431483596,
      "loss": 2.1178,
      "step": 150413
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.0653196573257446,
      "learning_rate": 0.0001614317262391686,
      "loss": 2.2108,
      "step": 150414
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.1975970268249512,
      "learning_rate": 0.0001614280981892638,
      "loss": 2.2332,
      "step": 150415
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.243513584136963,
      "learning_rate": 0.00016142447016512213,
      "loss": 2.0418,
      "step": 150416
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.161806344985962,
      "learning_rate": 0.00016142084216674417,
      "loss": 2.1068,
      "step": 150417
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.178039789199829,
      "learning_rate": 0.00016141721419413076,
      "loss": 2.0376,
      "step": 150418
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.043810486793518,
      "learning_rate": 0.00016141358624728244,
      "loss": 1.9594,
      "step": 150419
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.2003848552703857,
      "learning_rate": 0.00016140995832620005,
      "loss": 2.1422,
      "step": 150420
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.1596026420593262,
      "learning_rate": 0.00016140633043088416,
      "loss": 2.0422,
      "step": 150421
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.3931865692138672,
      "learning_rate": 0.00016140270256133544,
      "loss": 2.1935,
      "step": 150422
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.4602978229522705,
      "learning_rate": 0.00016139907471755447,
      "loss": 2.1993,
      "step": 150423
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.0930415391921997,
      "learning_rate": 0.00016139544689954213,
      "loss": 2.2803,
      "step": 150424
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.4510831832885742,
      "learning_rate": 0.00016139181910729893,
      "loss": 1.8683,
      "step": 150425
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.1294599771499634,
      "learning_rate": 0.00016138819134082566,
      "loss": 2.0403,
      "step": 150426
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.517559289932251,
      "learning_rate": 0.00016138456360012296,
      "loss": 1.843,
      "step": 150427
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.2059646844863892,
      "learning_rate": 0.00016138093588519148,
      "loss": 1.9748,
      "step": 150428
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.06974196434021,
      "learning_rate": 0.0001613773081960318,
      "loss": 2.2567,
      "step": 150429
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.092029094696045,
      "learning_rate": 0.00016137368053264483,
      "loss": 2.0829,
      "step": 150430
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.1251481771469116,
      "learning_rate": 0.00016137005289503095,
      "loss": 2.055,
      "step": 150431
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.4732067584991455,
      "learning_rate": 0.00016136642528319112,
      "loss": 2.0024,
      "step": 150432
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.2084954977035522,
      "learning_rate": 0.0001613627976971258,
      "loss": 2.1763,
      "step": 150433
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.1691884994506836,
      "learning_rate": 0.00016135917013683596,
      "loss": 2.1001,
      "step": 150434
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.1961013078689575,
      "learning_rate": 0.00016135554260232184,
      "loss": 1.9288,
      "step": 150435
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.1617074012756348,
      "learning_rate": 0.00016135191509358444,
      "loss": 2.2297,
      "step": 150436
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.1316428184509277,
      "learning_rate": 0.00016134828761062427,
      "loss": 2.2585,
      "step": 150437
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.1439982652664185,
      "learning_rate": 0.00016134466015344214,
      "loss": 1.905,
      "step": 150438
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.102773666381836,
      "learning_rate": 0.00016134103272203855,
      "loss": 2.0707,
      "step": 150439
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.1427254676818848,
      "learning_rate": 0.00016133740531641455,
      "loss": 2.2347,
      "step": 150440
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.42665433883667,
      "learning_rate": 0.00016133377793657025,
      "loss": 2.1508,
      "step": 150441
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.2423063516616821,
      "learning_rate": 0.0001613301505825068,
      "loss": 2.0513,
      "step": 150442
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.1158270835876465,
      "learning_rate": 0.00016132652325422453,
      "loss": 2.171,
      "step": 150443
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.2171776294708252,
      "learning_rate": 0.0001613228959517244,
      "loss": 2.2329,
      "step": 150444
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.2136354446411133,
      "learning_rate": 0.00016131926867500685,
      "loss": 2.2332,
      "step": 150445
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.4975610971450806,
      "learning_rate": 0.00016131564142407288,
      "loss": 1.8923,
      "step": 150446
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.281079888343811,
      "learning_rate": 0.00016131201419892273,
      "loss": 2.2453,
      "step": 150447
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.0307815074920654,
      "learning_rate": 0.00016130838699955744,
      "loss": 2.0754,
      "step": 150448
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.2500369548797607,
      "learning_rate": 0.00016130475982597736,
      "loss": 1.9571,
      "step": 150449
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.4663586616516113,
      "learning_rate": 0.00016130113267818352,
      "loss": 1.9229,
      "step": 150450
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.1282023191452026,
      "learning_rate": 0.00016129750555617627,
      "loss": 1.9922,
      "step": 150451
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.097443699836731,
      "learning_rate": 0.00016129387845995668,
      "loss": 2.0667,
      "step": 150452
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.2038276195526123,
      "learning_rate": 0.00016129025138952494,
      "loss": 2.2668,
      "step": 150453
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.2147367000579834,
      "learning_rate": 0.00016128662434488207,
      "loss": 1.932,
      "step": 150454
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.1172173023223877,
      "learning_rate": 0.00016128299732602852,
      "loss": 2.0449,
      "step": 150455
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.0807925462722778,
      "learning_rate": 0.0001612793703329652,
      "loss": 2.2892,
      "step": 150456
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.1547770500183105,
      "learning_rate": 0.00016127574336569255,
      "loss": 2.1135,
      "step": 150457
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.261461615562439,
      "learning_rate": 0.00016127211642421156,
      "loss": 2.1743,
      "step": 150458
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.3086423873901367,
      "learning_rate": 0.0001612684895085225,
      "loss": 1.9918,
      "step": 150459
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.0977519750595093,
      "learning_rate": 0.00016126486261862637,
      "loss": 1.934,
      "step": 150460
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.1829109191894531,
      "learning_rate": 0.0001612612357545236,
      "loss": 2.0736,
      "step": 150461
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.2667571306228638,
      "learning_rate": 0.0001612576089162151,
      "loss": 2.1505,
      "step": 150462
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.1006942987442017,
      "learning_rate": 0.0001612539821037013,
      "loss": 2.1362,
      "step": 150463
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.0583947896957397,
      "learning_rate": 0.00016125035531698324,
      "loss": 2.0398,
      "step": 150464
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.155505657196045,
      "learning_rate": 0.00016124672855606114,
      "loss": 2.1181,
      "step": 150465
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.0739779472351074,
      "learning_rate": 0.00016124310182093597,
      "loss": 2.0546,
      "step": 150466
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.1610183715820312,
      "learning_rate": 0.00016123947511160826,
      "loss": 2.1753,
      "step": 150467
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.2243205308914185,
      "learning_rate": 0.00016123584842807885,
      "loss": 2.0897,
      "step": 150468
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.0197135210037231,
      "learning_rate": 0.00016123222177034818,
      "loss": 1.9622,
      "step": 150469
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.0899397134780884,
      "learning_rate": 0.00016122859513841722,
      "loss": 2.1053,
      "step": 150470
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.049023151397705,
      "learning_rate": 0.00016122496853228642,
      "loss": 1.9493,
      "step": 150471
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.064252257347107,
      "learning_rate": 0.00016122134195195657,
      "loss": 2.0423,
      "step": 150472
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.0971159934997559,
      "learning_rate": 0.00016121771539742816,
      "loss": 2.0665,
      "step": 150473
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.1054517030715942,
      "learning_rate": 0.0001612140888687021,
      "loss": 2.165,
      "step": 150474
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.030612826347351,
      "learning_rate": 0.00016121046236577884,
      "loss": 2.0986,
      "step": 150475
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.1305124759674072,
      "learning_rate": 0.00016120683588865932,
      "loss": 2.1289,
      "step": 150476
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.0924644470214844,
      "learning_rate": 0.00016120320943734405,
      "loss": 2.0656,
      "step": 150477
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.0802119970321655,
      "learning_rate": 0.00016119958301183372,
      "loss": 1.891,
      "step": 150478
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.0877526998519897,
      "learning_rate": 0.00016119595661212888,
      "loss": 2.0766,
      "step": 150479
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.3048787117004395,
      "learning_rate": 0.0001611923302382305,
      "loss": 1.9321,
      "step": 150480
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.150748372077942,
      "learning_rate": 0.00016118870389013893,
      "loss": 2.1925,
      "step": 150481
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.2908955812454224,
      "learning_rate": 0.00016118507756785513,
      "loss": 2.1825,
      "step": 150482
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.1420713663101196,
      "learning_rate": 0.00016118145127137963,
      "loss": 2.0445,
      "step": 150483
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.184635877609253,
      "learning_rate": 0.00016117782500071315,
      "loss": 1.9653,
      "step": 150484
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.1358448266983032,
      "learning_rate": 0.00016117419875585616,
      "loss": 2.1469,
      "step": 150485
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.1405420303344727,
      "learning_rate": 0.00016117057253680966,
      "loss": 1.9893,
      "step": 150486
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.1450821161270142,
      "learning_rate": 0.00016116694634357406,
      "loss": 2.236,
      "step": 150487
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.113051176071167,
      "learning_rate": 0.00016116332017615024,
      "loss": 2.1651,
      "step": 150488
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.0727089643478394,
      "learning_rate": 0.0001611596940345388,
      "loss": 1.855,
      "step": 150489
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.123772144317627,
      "learning_rate": 0.0001611560679187404,
      "loss": 2.2662,
      "step": 150490
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.118409276008606,
      "learning_rate": 0.00016115244182875554,
      "loss": 2.1658,
      "step": 150491
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.2438278198242188,
      "learning_rate": 0.00016114881576458523,
      "loss": 2.0424,
      "step": 150492
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.1196463108062744,
      "learning_rate": 0.00016114518972622986,
      "loss": 2.0904,
      "step": 150493
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.196271538734436,
      "learning_rate": 0.00016114156371369032,
      "loss": 2.1106,
      "step": 150494
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.2806748151779175,
      "learning_rate": 0.00016113793772696717,
      "loss": 1.9213,
      "step": 150495
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.0409085750579834,
      "learning_rate": 0.00016113431176606103,
      "loss": 2.0072,
      "step": 150496
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.0502790212631226,
      "learning_rate": 0.0001611306858309727,
      "loss": 1.9793,
      "step": 150497
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.1195265054702759,
      "learning_rate": 0.00016112705992170283,
      "loss": 2.336,
      "step": 150498
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.1051121950149536,
      "learning_rate": 0.00016112343403825195,
      "loss": 1.9762,
      "step": 150499
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.2608991861343384,
      "learning_rate": 0.00016111980818062094,
      "loss": 2.0449,
      "step": 150500
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.0772933959960938,
      "learning_rate": 0.0001611161823488104,
      "loss": 2.1426,
      "step": 150501
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.1424410343170166,
      "learning_rate": 0.00016111255654282087,
      "loss": 2.1507,
      "step": 150502
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.175165057182312,
      "learning_rate": 0.00016110893076265323,
      "loss": 2.2286,
      "step": 150503
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.1150676012039185,
      "learning_rate": 0.00016110530500830806,
      "loss": 1.8128,
      "step": 150504
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.112127423286438,
      "learning_rate": 0.00016110167927978593,
      "loss": 2.096,
      "step": 150505
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.1155385971069336,
      "learning_rate": 0.00016109805357708776,
      "loss": 1.9465,
      "step": 150506
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.15065336227417,
      "learning_rate": 0.00016109442790021404,
      "loss": 2.2112,
      "step": 150507
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.0976216793060303,
      "learning_rate": 0.0001610908022491654,
      "loss": 1.81,
      "step": 150508
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.1870958805084229,
      "learning_rate": 0.0001610871766239427,
      "loss": 2.1817,
      "step": 150509
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.1190965175628662,
      "learning_rate": 0.00016108355102454646,
      "loss": 2.0165,
      "step": 150510
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.1128392219543457,
      "learning_rate": 0.00016107992545097744,
      "loss": 2.1599,
      "step": 150511
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.1642488241195679,
      "learning_rate": 0.00016107629990323632,
      "loss": 2.254,
      "step": 150512
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.0580294132232666,
      "learning_rate": 0.00016107267438132377,
      "loss": 2.2961,
      "step": 150513
    },
    {
      "epoch": 1.96,
      "grad_norm": 2.053903818130493,
      "learning_rate": 0.00016106904888524027,
      "loss": 2.0511,
      "step": 150514
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.1013661623001099,
      "learning_rate": 0.0001610654234149868,
      "loss": 2.0344,
      "step": 150515
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.161868929862976,
      "learning_rate": 0.00016106179797056373,
      "loss": 2.0392,
      "step": 150516
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.1613013744354248,
      "learning_rate": 0.00016105817255197206,
      "loss": 2.0611,
      "step": 150517
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.1627269983291626,
      "learning_rate": 0.00016105454715921216,
      "loss": 2.2451,
      "step": 150518
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.2413233518600464,
      "learning_rate": 0.00016105092179228506,
      "loss": 1.973,
      "step": 150519
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.112707257270813,
      "learning_rate": 0.000161047296451191,
      "loss": 2.0942,
      "step": 150520
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.2151927947998047,
      "learning_rate": 0.00016104367113593102,
      "loss": 2.1701,
      "step": 150521
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.0373321771621704,
      "learning_rate": 0.00016104004584650548,
      "loss": 1.8274,
      "step": 150522
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.0262666940689087,
      "learning_rate": 0.00016103642058291537,
      "loss": 2.1423,
      "step": 150523
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.1968578100204468,
      "learning_rate": 0.00016103279534516112,
      "loss": 2.1913,
      "step": 150524
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.0720134973526,
      "learning_rate": 0.00016102917013324365,
      "loss": 2.1898,
      "step": 150525
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.2059146165847778,
      "learning_rate": 0.0001610255449471633,
      "loss": 2.2238,
      "step": 150526
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.1599905490875244,
      "learning_rate": 0.00016102191978692104,
      "loss": 2.0885,
      "step": 150527
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.1613109111785889,
      "learning_rate": 0.00016101829465251728,
      "loss": 2.1078,
      "step": 150528
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.0550158023834229,
      "learning_rate": 0.000161014669543953,
      "loss": 1.7754,
      "step": 150529
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.2362886667251587,
      "learning_rate": 0.00016101104446122858,
      "loss": 2.0691,
      "step": 150530
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.125177264213562,
      "learning_rate": 0.00016100741940434506,
      "loss": 1.9945,
      "step": 150531
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.1493971347808838,
      "learning_rate": 0.00016100379437330265,
      "loss": 2.3032,
      "step": 150532
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.276718258857727,
      "learning_rate": 0.00016100016936810237,
      "loss": 2.2251,
      "step": 150533
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.2731856107711792,
      "learning_rate": 0.0001609965443887447,
      "loss": 1.9669,
      "step": 150534
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.1923809051513672,
      "learning_rate": 0.00016099291943523047,
      "loss": 1.8991,
      "step": 150535
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.1611931324005127,
      "learning_rate": 0.00016098929450756018,
      "loss": 2.0179,
      "step": 150536
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.0914735794067383,
      "learning_rate": 0.00016098566960573473,
      "loss": 2.3514,
      "step": 150537
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.2432327270507812,
      "learning_rate": 0.00016098204472975465,
      "loss": 2.0422,
      "step": 150538
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.4236400127410889,
      "learning_rate": 0.00016097841987962064,
      "loss": 2.2104,
      "step": 150539
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.3125824928283691,
      "learning_rate": 0.00016097479505533326,
      "loss": 1.895,
      "step": 150540
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.169184923171997,
      "learning_rate": 0.0001609711702568934,
      "loss": 2.1681,
      "step": 150541
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.147646188735962,
      "learning_rate": 0.00016096754548430148,
      "loss": 2.1387,
      "step": 150542
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.147078037261963,
      "learning_rate": 0.00016096392073755849,
      "loss": 1.9963,
      "step": 150543
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.1753926277160645,
      "learning_rate": 0.00016096029601666488,
      "loss": 2.1037,
      "step": 150544
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.093397855758667,
      "learning_rate": 0.00016095667132162144,
      "loss": 2.0364,
      "step": 150545
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.1776299476623535,
      "learning_rate": 0.00016095304665242859,
      "loss": 2.044,
      "step": 150546
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.1249688863754272,
      "learning_rate": 0.00016094942200908734,
      "loss": 2.0951,
      "step": 150547
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.0467610359191895,
      "learning_rate": 0.0001609457973915981,
      "loss": 1.9133,
      "step": 150548
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.1618505716323853,
      "learning_rate": 0.0001609421727999618,
      "loss": 2.198,
      "step": 150549
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.1404731273651123,
      "learning_rate": 0.00016093854823417896,
      "loss": 2.1013,
      "step": 150550
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.272652506828308,
      "learning_rate": 0.00016093492369425024,
      "loss": 1.9638,
      "step": 150551
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.114588975906372,
      "learning_rate": 0.00016093129918017625,
      "loss": 2.2118,
      "step": 150552
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.053586721420288,
      "learning_rate": 0.00016092767469195787,
      "loss": 2.1204,
      "step": 150553
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.3978278636932373,
      "learning_rate": 0.00016092405022959555,
      "loss": 2.1937,
      "step": 150554
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.1975135803222656,
      "learning_rate": 0.00016092042579309016,
      "loss": 1.9544,
      "step": 150555
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.1394864320755005,
      "learning_rate": 0.00016091680138244236,
      "loss": 2.0713,
      "step": 150556
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.1360678672790527,
      "learning_rate": 0.00016091317699765267,
      "loss": 2.092,
      "step": 150557
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.2228480577468872,
      "learning_rate": 0.0001609095526387218,
      "loss": 1.8829,
      "step": 150558
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.2505970001220703,
      "learning_rate": 0.00016090592830565056,
      "loss": 2.0918,
      "step": 150559
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.1633888483047485,
      "learning_rate": 0.0001609023039984394,
      "loss": 2.193,
      "step": 150560
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.086639642715454,
      "learning_rate": 0.00016089867971708926,
      "loss": 2.1754,
      "step": 150561
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.1325759887695312,
      "learning_rate": 0.00016089505546160065,
      "loss": 1.8446,
      "step": 150562
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.2221795320510864,
      "learning_rate": 0.0001608914312319743,
      "loss": 1.8065,
      "step": 150563
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.1705727577209473,
      "learning_rate": 0.00016088780702821074,
      "loss": 2.0624,
      "step": 150564
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.1959812641143799,
      "learning_rate": 0.0001608841828503109,
      "loss": 1.8182,
      "step": 150565
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.2792572975158691,
      "learning_rate": 0.00016088055869827516,
      "loss": 2.0365,
      "step": 150566
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.321555495262146,
      "learning_rate": 0.0001608769345721045,
      "loss": 2.0406,
      "step": 150567
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.0820974111557007,
      "learning_rate": 0.0001608733104717994,
      "loss": 2.0981,
      "step": 150568
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.289375901222229,
      "learning_rate": 0.00016086968639736067,
      "loss": 1.908,
      "step": 150569
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.1572896242141724,
      "learning_rate": 0.0001608660623487887,
      "loss": 2.1868,
      "step": 150570
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.221032738685608,
      "learning_rate": 0.0001608624383260845,
      "loss": 1.8722,
      "step": 150571
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.1196045875549316,
      "learning_rate": 0.00016085881432924845,
      "loss": 2.0492,
      "step": 150572
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.1061630249023438,
      "learning_rate": 0.0001608551903582815,
      "loss": 2.1783,
      "step": 150573
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.2892323732376099,
      "learning_rate": 0.00016085156641318423,
      "loss": 2.2608,
      "step": 150574
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.2150224447250366,
      "learning_rate": 0.00016084794249395724,
      "loss": 2.0927,
      "step": 150575
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.0993692874908447,
      "learning_rate": 0.0001608443186006011,
      "loss": 2.0013,
      "step": 150576
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.0391790866851807,
      "learning_rate": 0.00016084069473311683,
      "loss": 1.9659,
      "step": 150577
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.1361653804779053,
      "learning_rate": 0.00016083707089150474,
      "loss": 2.0897,
      "step": 150578
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.0485146045684814,
      "learning_rate": 0.0001608334470757658,
      "loss": 2.1805,
      "step": 150579
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.2411056756973267,
      "learning_rate": 0.0001608298232859005,
      "loss": 2.1118,
      "step": 150580
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.1730175018310547,
      "learning_rate": 0.0001608261995219095,
      "loss": 2.1137,
      "step": 150581
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.1842097043991089,
      "learning_rate": 0.00016082257578379366,
      "loss": 1.9775,
      "step": 150582
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.221996784210205,
      "learning_rate": 0.00016081895207155348,
      "loss": 2.3728,
      "step": 150583
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.1560471057891846,
      "learning_rate": 0.0001608153283851896,
      "loss": 1.9891,
      "step": 150584
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.1848938465118408,
      "learning_rate": 0.00016081170472470287,
      "loss": 1.9118,
      "step": 150585
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.0836366415023804,
      "learning_rate": 0.0001608080810900939,
      "loss": 2.0677,
      "step": 150586
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.46324622631073,
      "learning_rate": 0.00016080445748136322,
      "loss": 2.0639,
      "step": 150587
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.2188401222229004,
      "learning_rate": 0.00016080083389851172,
      "loss": 2.2837,
      "step": 150588
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.1997886896133423,
      "learning_rate": 0.00016079721034153996,
      "loss": 1.9375,
      "step": 150589
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.1877849102020264,
      "learning_rate": 0.00016079358681044852,
      "loss": 1.8544,
      "step": 150590
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.1457440853118896,
      "learning_rate": 0.0001607899633052383,
      "loss": 2.0435,
      "step": 150591
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.1475590467453003,
      "learning_rate": 0.00016078633982590985,
      "loss": 1.865,
      "step": 150592
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.2083210945129395,
      "learning_rate": 0.00016078271637246373,
      "loss": 2.1442,
      "step": 150593
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.0846686363220215,
      "learning_rate": 0.00016077909294490083,
      "loss": 1.9333,
      "step": 150594
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.1429373025894165,
      "learning_rate": 0.00016077546954322168,
      "loss": 2.1033,
      "step": 150595
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.2056046724319458,
      "learning_rate": 0.00016077184616742704,
      "loss": 2.1865,
      "step": 150596
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.1064785718917847,
      "learning_rate": 0.00016076822281751755,
      "loss": 2.0726,
      "step": 150597
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.018621802330017,
      "learning_rate": 0.00016076459949349392,
      "loss": 2.206,
      "step": 150598
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.1786352396011353,
      "learning_rate": 0.00016076097619535662,
      "loss": 1.9198,
      "step": 150599
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.1124087572097778,
      "learning_rate": 0.00016075735292310663,
      "loss": 2.0698,
      "step": 150600
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.4991523027420044,
      "learning_rate": 0.00016075372967674432,
      "loss": 1.8941,
      "step": 150601
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.1173301935195923,
      "learning_rate": 0.00016075010645627067,
      "loss": 1.8989,
      "step": 150602
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.1978005170822144,
      "learning_rate": 0.0001607464832616861,
      "loss": 2.1017,
      "step": 150603
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.1968162059783936,
      "learning_rate": 0.00016074286009299144,
      "loss": 2.0918,
      "step": 150604
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.1996920108795166,
      "learning_rate": 0.00016073923695018736,
      "loss": 1.8511,
      "step": 150605
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.4790339469909668,
      "learning_rate": 0.0001607356138332745,
      "loss": 2.2593,
      "step": 150606
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.4866416454315186,
      "learning_rate": 0.00016073199074225338,
      "loss": 2.0566,
      "step": 150607
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.1758363246917725,
      "learning_rate": 0.00016072836767712495,
      "loss": 2.1024,
      "step": 150608
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.049596905708313,
      "learning_rate": 0.00016072474463788957,
      "loss": 2.0377,
      "step": 150609
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.3644392490386963,
      "learning_rate": 0.00016072112162454825,
      "loss": 2.0677,
      "step": 150610
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.4761959314346313,
      "learning_rate": 0.0001607174986371015,
      "loss": 2.1017,
      "step": 150611
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.1100844144821167,
      "learning_rate": 0.00016071387567554997,
      "loss": 1.9328,
      "step": 150612
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.2307548522949219,
      "learning_rate": 0.00016071025273989428,
      "loss": 2.2723,
      "step": 150613
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.1582258939743042,
      "learning_rate": 0.0001607066298301353,
      "loss": 2.1528,
      "step": 150614
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.2772598266601562,
      "learning_rate": 0.00016070300694627344,
      "loss": 2.1265,
      "step": 150615
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.6500836610794067,
      "learning_rate": 0.00016069938408830962,
      "loss": 2.2066,
      "step": 150616
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.2559740543365479,
      "learning_rate": 0.0001606957612562445,
      "loss": 1.9944,
      "step": 150617
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.1445043087005615,
      "learning_rate": 0.0001606921384500786,
      "loss": 2.0582,
      "step": 150618
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.2258682250976562,
      "learning_rate": 0.00016068851566981255,
      "loss": 2.1537,
      "step": 150619
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.1397701501846313,
      "learning_rate": 0.00016068489291544725,
      "loss": 1.7796,
      "step": 150620
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.6259183883666992,
      "learning_rate": 0.00016068127018698314,
      "loss": 2.0675,
      "step": 150621
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.163225531578064,
      "learning_rate": 0.00016067764748442113,
      "loss": 2.032,
      "step": 150622
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.1084171533584595,
      "learning_rate": 0.0001606740248077618,
      "loss": 1.8136,
      "step": 150623
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.1281260251998901,
      "learning_rate": 0.00016067040215700577,
      "loss": 1.6676,
      "step": 150624
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.4012640714645386,
      "learning_rate": 0.00016066677953215368,
      "loss": 2.0579,
      "step": 150625
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.370978832244873,
      "learning_rate": 0.00016066315693320633,
      "loss": 2.0889,
      "step": 150626
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.2030388116836548,
      "learning_rate": 0.0001606595343601642,
      "loss": 1.9678,
      "step": 150627
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.3618923425674438,
      "learning_rate": 0.00016065591181302827,
      "loss": 2.1696,
      "step": 150628
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.379966378211975,
      "learning_rate": 0.00016065228929179896,
      "loss": 1.7985,
      "step": 150629
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.1577152013778687,
      "learning_rate": 0.00016064866679647705,
      "loss": 2.1867,
      "step": 150630
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.1028170585632324,
      "learning_rate": 0.0001606450443270631,
      "loss": 2.2476,
      "step": 150631
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.2544969320297241,
      "learning_rate": 0.00016064142188355798,
      "loss": 2.0925,
      "step": 150632
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.1904603242874146,
      "learning_rate": 0.0001606377994659621,
      "loss": 2.1676,
      "step": 150633
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.1208707094192505,
      "learning_rate": 0.0001606341770742764,
      "loss": 2.2571,
      "step": 150634
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.098637580871582,
      "learning_rate": 0.00016063055470850145,
      "loss": 2.1478,
      "step": 150635
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.0743311643600464,
      "learning_rate": 0.00016062693236863792,
      "loss": 2.0537,
      "step": 150636
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.2770131826400757,
      "learning_rate": 0.00016062331005468636,
      "loss": 2.1916,
      "step": 150637
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.1152188777923584,
      "learning_rate": 0.00016061968776664764,
      "loss": 2.2362,
      "step": 150638
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.1892021894454956,
      "learning_rate": 0.00016061606550452225,
      "loss": 2.2512,
      "step": 150639
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.1382055282592773,
      "learning_rate": 0.00016061244326831108,
      "loss": 2.1295,
      "step": 150640
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.0533522367477417,
      "learning_rate": 0.00016060882105801473,
      "loss": 2.025,
      "step": 150641
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.169762372970581,
      "learning_rate": 0.00016060519887363379,
      "loss": 2.1593,
      "step": 150642
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.0129410028457642,
      "learning_rate": 0.00016060157671516886,
      "loss": 1.9736,
      "step": 150643
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.2348345518112183,
      "learning_rate": 0.00016059795458262086,
      "loss": 2.0733,
      "step": 150644
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.2125978469848633,
      "learning_rate": 0.0001605943324759902,
      "loss": 2.3315,
      "step": 150645
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.2154620885849,
      "learning_rate": 0.00016059071039527782,
      "loss": 2.2625,
      "step": 150646
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.3349157571792603,
      "learning_rate": 0.00016058708834048422,
      "loss": 2.2337,
      "step": 150647
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.5088210105895996,
      "learning_rate": 0.00016058346631161015,
      "loss": 1.9,
      "step": 150648
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.0739713907241821,
      "learning_rate": 0.0001605798443086561,
      "loss": 2.0204,
      "step": 150649
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.2616976499557495,
      "learning_rate": 0.00016057622233162303,
      "loss": 2.2111,
      "step": 150650
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.5137903690338135,
      "learning_rate": 0.00016057260038051133,
      "loss": 2.051,
      "step": 150651
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.3258060216903687,
      "learning_rate": 0.00016056897845532198,
      "loss": 1.7908,
      "step": 150652
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.2143386602401733,
      "learning_rate": 0.00016056535655605542,
      "loss": 2.0659,
      "step": 150653
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.1252115964889526,
      "learning_rate": 0.00016056173468271246,
      "loss": 2.012,
      "step": 150654
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.2081176042556763,
      "learning_rate": 0.00016055811283529356,
      "loss": 2.1678,
      "step": 150655
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.1855459213256836,
      "learning_rate": 0.00016055449101379967,
      "loss": 2.0496,
      "step": 150656
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.1918364763259888,
      "learning_rate": 0.0001605508692182312,
      "loss": 2.1198,
      "step": 150657
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.073759913444519,
      "learning_rate": 0.00016054724744858905,
      "loss": 2.1789,
      "step": 150658
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.141625165939331,
      "learning_rate": 0.00016054362570487382,
      "loss": 2.3303,
      "step": 150659
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.099273920059204,
      "learning_rate": 0.00016054000398708616,
      "loss": 1.9331,
      "step": 150660
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.2433805465698242,
      "learning_rate": 0.00016053638229522664,
      "loss": 2.169,
      "step": 150661
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.0801681280136108,
      "learning_rate": 0.00016053276062929616,
      "loss": 2.1211,
      "step": 150662
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.179178237915039,
      "learning_rate": 0.00016052913898929516,
      "loss": 2.0262,
      "step": 150663
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.1028724908828735,
      "learning_rate": 0.00016052551737522452,
      "loss": 1.9943,
      "step": 150664
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.1607195138931274,
      "learning_rate": 0.0001605218957870848,
      "loss": 2.0924,
      "step": 150665
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.1585458517074585,
      "learning_rate": 0.00016051827422487676,
      "loss": 2.0722,
      "step": 150666
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.0728769302368164,
      "learning_rate": 0.0001605146526886008,
      "loss": 2.3197,
      "step": 150667
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.0580856800079346,
      "learning_rate": 0.00016051103117825798,
      "loss": 2.1044,
      "step": 150668
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.032326579093933,
      "learning_rate": 0.00016050740969384864,
      "loss": 1.897,
      "step": 150669
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.125727891921997,
      "learning_rate": 0.00016050378823537375,
      "loss": 2.0816,
      "step": 150670
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.2374894618988037,
      "learning_rate": 0.0001605001668028337,
      "loss": 1.94,
      "step": 150671
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.1239701509475708,
      "learning_rate": 0.00016049654539622944,
      "loss": 2.0678,
      "step": 150672
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.1835572719573975,
      "learning_rate": 0.0001604929240155615,
      "loss": 1.8914,
      "step": 150673
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.1696772575378418,
      "learning_rate": 0.00016048930266083055,
      "loss": 1.9824,
      "step": 150674
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.099992036819458,
      "learning_rate": 0.00016048568133203718,
      "loss": 2.0374,
      "step": 150675
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.2946616411209106,
      "learning_rate": 0.00016048206002918227,
      "loss": 2.4042,
      "step": 150676
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.063734531402588,
      "learning_rate": 0.00016047843875226624,
      "loss": 1.9358,
      "step": 150677
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.106411099433899,
      "learning_rate": 0.00016047481750129002,
      "loss": 2.0653,
      "step": 150678
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.0092428922653198,
      "learning_rate": 0.00016047119627625415,
      "loss": 2.0397,
      "step": 150679
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.1731855869293213,
      "learning_rate": 0.00016046757507715923,
      "loss": 2.1426,
      "step": 150680
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.1519465446472168,
      "learning_rate": 0.00016046395390400617,
      "loss": 2.2302,
      "step": 150681
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.0493863821029663,
      "learning_rate": 0.00016046033275679542,
      "loss": 2.1574,
      "step": 150682
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.3520374298095703,
      "learning_rate": 0.00016045671163552764,
      "loss": 2.0569,
      "step": 150683
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.1822516918182373,
      "learning_rate": 0.00016045309054020372,
      "loss": 2.1098,
      "step": 150684
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.066677451133728,
      "learning_rate": 0.00016044946947082423,
      "loss": 2.0079,
      "step": 150685
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.2013883590698242,
      "learning_rate": 0.00016044584842738968,
      "loss": 2.0751,
      "step": 150686
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.15834379196167,
      "learning_rate": 0.00016044222740990095,
      "loss": 2.1158,
      "step": 150687
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.1999350786209106,
      "learning_rate": 0.0001604386064183587,
      "loss": 2.147,
      "step": 150688
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.1690446138381958,
      "learning_rate": 0.00016043498545276343,
      "loss": 2.0984,
      "step": 150689
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.2302196025848389,
      "learning_rate": 0.000160431364513116,
      "loss": 2.1634,
      "step": 150690
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.1268184185028076,
      "learning_rate": 0.00016042774359941705,
      "loss": 2.2444,
      "step": 150691
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.0234347581863403,
      "learning_rate": 0.0001604241227116671,
      "loss": 1.9437,
      "step": 150692
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.1425540447235107,
      "learning_rate": 0.00016042050184986707,
      "loss": 1.9037,
      "step": 150693
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.075233817100525,
      "learning_rate": 0.00016041688101401736,
      "loss": 2.0502,
      "step": 150694
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.092490315437317,
      "learning_rate": 0.00016041326020411892,
      "loss": 1.9459,
      "step": 150695
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.1699334383010864,
      "learning_rate": 0.00016040963942017228,
      "loss": 2.1522,
      "step": 150696
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.134425163269043,
      "learning_rate": 0.00016040601866217815,
      "loss": 1.8463,
      "step": 150697
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.1846058368682861,
      "learning_rate": 0.00016040239793013703,
      "loss": 2.0884,
      "step": 150698
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.0769906044006348,
      "learning_rate": 0.00016039877722404984,
      "loss": 1.9783,
      "step": 150699
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.2276676893234253,
      "learning_rate": 0.00016039515654391704,
      "loss": 2.0769,
      "step": 150700
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.224131464958191,
      "learning_rate": 0.00016039153588973956,
      "loss": 2.0967,
      "step": 150701
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.0498332977294922,
      "learning_rate": 0.00016038791526151792,
      "loss": 2.2108,
      "step": 150702
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.1043047904968262,
      "learning_rate": 0.0001603842946592528,
      "loss": 1.9079,
      "step": 150703
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.338093876838684,
      "learning_rate": 0.00016038067408294477,
      "loss": 2.2619,
      "step": 150704
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.1499531269073486,
      "learning_rate": 0.0001603770535325947,
      "loss": 2.1162,
      "step": 150705
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.1582863330841064,
      "learning_rate": 0.00016037343300820308,
      "loss": 1.9771,
      "step": 150706
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.1802666187286377,
      "learning_rate": 0.00016036981250977078,
      "loss": 2.192,
      "step": 150707
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.0492711067199707,
      "learning_rate": 0.00016036619203729834,
      "loss": 2.0239,
      "step": 150708
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.1925557851791382,
      "learning_rate": 0.0001603625715907865,
      "loss": 2.0155,
      "step": 150709
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.0519804954528809,
      "learning_rate": 0.00016035895117023577,
      "loss": 2.1744,
      "step": 150710
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.0930840969085693,
      "learning_rate": 0.00016035533077564704,
      "loss": 2.0115,
      "step": 150711
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.0714919567108154,
      "learning_rate": 0.0001603517104070208,
      "loss": 2.3504,
      "step": 150712
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.088170051574707,
      "learning_rate": 0.00016034809006435793,
      "loss": 2.0167,
      "step": 150713
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.244544267654419,
      "learning_rate": 0.00016034446974765897,
      "loss": 1.8957,
      "step": 150714
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.191541314125061,
      "learning_rate": 0.00016034084945692464,
      "loss": 2.0012,
      "step": 150715
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.0352482795715332,
      "learning_rate": 0.00016033722919215541,
      "loss": 1.9822,
      "step": 150716
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.1651413440704346,
      "learning_rate": 0.0001603336089533523,
      "loss": 2.0455,
      "step": 150717
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.0696780681610107,
      "learning_rate": 0.00016032998874051564,
      "loss": 1.8989,
      "step": 150718
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.079295039176941,
      "learning_rate": 0.00016032636855364642,
      "loss": 2.1714,
      "step": 150719
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.0813242197036743,
      "learning_rate": 0.00016032274839274517,
      "loss": 2.0371,
      "step": 150720
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.0432318449020386,
      "learning_rate": 0.00016031912825781254,
      "loss": 1.8817,
      "step": 150721
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.2461308240890503,
      "learning_rate": 0.0001603155081488491,
      "loss": 2.2184,
      "step": 150722
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.0945988893508911,
      "learning_rate": 0.00016031188806585577,
      "loss": 1.7514,
      "step": 150723
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.1618775129318237,
      "learning_rate": 0.000160308268008833,
      "loss": 2.1645,
      "step": 150724
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.0831018686294556,
      "learning_rate": 0.00016030464797778165,
      "loss": 2.0305,
      "step": 150725
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.1356401443481445,
      "learning_rate": 0.00016030102797270234,
      "loss": 1.9263,
      "step": 150726
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.1623884439468384,
      "learning_rate": 0.00016029740799359567,
      "loss": 2.053,
      "step": 150727
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.0846997499465942,
      "learning_rate": 0.00016029378804046222,
      "loss": 1.8429,
      "step": 150728
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.1911324262619019,
      "learning_rate": 0.00016029016811330293,
      "loss": 2.0758,
      "step": 150729
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.2059277296066284,
      "learning_rate": 0.00016028654821211821,
      "loss": 2.051,
      "step": 150730
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.151458501815796,
      "learning_rate": 0.000160282928336909,
      "loss": 1.8925,
      "step": 150731
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.1768091917037964,
      "learning_rate": 0.0001602793084876758,
      "loss": 2.1854,
      "step": 150732
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.0523650646209717,
      "learning_rate": 0.0001602756886644193,
      "loss": 1.9972,
      "step": 150733
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.2295645475387573,
      "learning_rate": 0.0001602720688671401,
      "loss": 2.1858,
      "step": 150734
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.0471992492675781,
      "learning_rate": 0.00016026844909583908,
      "loss": 2.0317,
      "step": 150735
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.2752187252044678,
      "learning_rate": 0.00016026482935051667,
      "loss": 2.1208,
      "step": 150736
    },
    {
      "epoch": 1.96,
      "grad_norm": 0.9726659655570984,
      "learning_rate": 0.0001602612096311738,
      "loss": 1.9936,
      "step": 150737
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.2667138576507568,
      "learning_rate": 0.00016025758993781086,
      "loss": 1.9684,
      "step": 150738
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.1143759489059448,
      "learning_rate": 0.00016025397027042896,
      "loss": 2.1321,
      "step": 150739
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.3190393447875977,
      "learning_rate": 0.0001602503506290282,
      "loss": 1.909,
      "step": 150740
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.2785322666168213,
      "learning_rate": 0.00016024673101360965,
      "loss": 2.3119,
      "step": 150741
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.1601250171661377,
      "learning_rate": 0.00016024311142417375,
      "loss": 2.1564,
      "step": 150742
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.2073469161987305,
      "learning_rate": 0.00016023949186072143,
      "loss": 1.9585,
      "step": 150743
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.161736249923706,
      "learning_rate": 0.00016023587232325313,
      "loss": 2.1597,
      "step": 150744
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.0369125604629517,
      "learning_rate": 0.00016023225281176982,
      "loss": 2.0167,
      "step": 150745
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.2223060131072998,
      "learning_rate": 0.00016022863332627174,
      "loss": 2.172,
      "step": 150746
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.1200584173202515,
      "learning_rate": 0.00016022501386675993,
      "loss": 2.0123,
      "step": 150747
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.1654863357543945,
      "learning_rate": 0.0001602213944332348,
      "loss": 2.087,
      "step": 150748
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.1417930126190186,
      "learning_rate": 0.00016021777502569728,
      "loss": 1.9096,
      "step": 150749
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.178953766822815,
      "learning_rate": 0.0001602141556441478,
      "loss": 2.1564,
      "step": 150750
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.3422834873199463,
      "learning_rate": 0.00016021053628858736,
      "loss": 2.1211,
      "step": 150751
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.1762276887893677,
      "learning_rate": 0.0001602069169590162,
      "loss": 1.829,
      "step": 150752
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.1810188293457031,
      "learning_rate": 0.0001602032976554353,
      "loss": 2.1462,
      "step": 150753
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.4453248977661133,
      "learning_rate": 0.0001601996783778452,
      "loss": 2.3288,
      "step": 150754
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.4311435222625732,
      "learning_rate": 0.00016019605912624666,
      "loss": 1.9184,
      "step": 150755
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.1610510349273682,
      "learning_rate": 0.00016019243990064025,
      "loss": 2.0419,
      "step": 150756
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.0392704010009766,
      "learning_rate": 0.0001601888207010268,
      "loss": 2.269,
      "step": 150757
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.0696543455123901,
      "learning_rate": 0.00016018520152740685,
      "loss": 2.0022,
      "step": 150758
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.4847424030303955,
      "learning_rate": 0.00016018158237978122,
      "loss": 1.9229,
      "step": 150759
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.1384316682815552,
      "learning_rate": 0.00016017796325815023,
      "loss": 2.0576,
      "step": 150760
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.2480602264404297,
      "learning_rate": 0.000160174344162515,
      "loss": 2.2765,
      "step": 150761
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.13019597530365,
      "learning_rate": 0.0001601707250928758,
      "loss": 1.9212,
      "step": 150762
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.1416151523590088,
      "learning_rate": 0.0001601671060492337,
      "loss": 2.0504,
      "step": 150763
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.1876038312911987,
      "learning_rate": 0.00016016348703158913,
      "loss": 1.9703,
      "step": 150764
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.1369943618774414,
      "learning_rate": 0.0001601598680399427,
      "loss": 2.0286,
      "step": 150765
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.13124418258667,
      "learning_rate": 0.0001601562490742953,
      "loss": 2.0643,
      "step": 150766
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.1479461193084717,
      "learning_rate": 0.0001601526301346475,
      "loss": 2.0439,
      "step": 150767
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.2779356241226196,
      "learning_rate": 0.00016014901122099985,
      "loss": 2.1604,
      "step": 150768
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.5393873453140259,
      "learning_rate": 0.00016014539233335324,
      "loss": 2.2081,
      "step": 150769
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.172420620918274,
      "learning_rate": 0.0001601417734717083,
      "loss": 2.2376,
      "step": 150770
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.0222123861312866,
      "learning_rate": 0.00016013815463606547,
      "loss": 1.986,
      "step": 150771
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.131173849105835,
      "learning_rate": 0.00016013453582642572,
      "loss": 1.9391,
      "step": 150772
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.159269094467163,
      "learning_rate": 0.0001601309170427896,
      "loss": 2.221,
      "step": 150773
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.247375249862671,
      "learning_rate": 0.0001601272982851577,
      "loss": 2.0429,
      "step": 150774
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.0944608449935913,
      "learning_rate": 0.00016012367955353088,
      "loss": 2.249,
      "step": 150775
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.2005398273468018,
      "learning_rate": 0.00016012006084790964,
      "loss": 1.88,
      "step": 150776
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.1381359100341797,
      "learning_rate": 0.00016011644216829468,
      "loss": 2.0829,
      "step": 150777
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.125111699104309,
      "learning_rate": 0.0001601128235146868,
      "loss": 2.1302,
      "step": 150778
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.2158403396606445,
      "learning_rate": 0.00016010920488708649,
      "loss": 1.9481,
      "step": 150779
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.0696758031845093,
      "learning_rate": 0.0001601055862854946,
      "loss": 1.9443,
      "step": 150780
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.1157636642456055,
      "learning_rate": 0.00016010196770991177,
      "loss": 2.008,
      "step": 150781
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.0406546592712402,
      "learning_rate": 0.00016009834916033865,
      "loss": 1.9475,
      "step": 150782
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.1350927352905273,
      "learning_rate": 0.0001600947306367757,
      "loss": 2.0545,
      "step": 150783
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.2102525234222412,
      "learning_rate": 0.00016009111213922391,
      "loss": 1.9943,
      "step": 150784
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.0534334182739258,
      "learning_rate": 0.00016008749366768372,
      "loss": 1.9292,
      "step": 150785
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.1654170751571655,
      "learning_rate": 0.000160083875222156,
      "loss": 2.2235,
      "step": 150786
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.0430612564086914,
      "learning_rate": 0.00016008025680264135,
      "loss": 2.1707,
      "step": 150787
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.117987871170044,
      "learning_rate": 0.00016007663840914043,
      "loss": 2.0673,
      "step": 150788
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.0937312841415405,
      "learning_rate": 0.00016007302004165377,
      "loss": 2.2571,
      "step": 150789
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.1108806133270264,
      "learning_rate": 0.00016006940170018233,
      "loss": 2.1362,
      "step": 150790
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.340082049369812,
      "learning_rate": 0.00016006578338472645,
      "loss": 2.091,
      "step": 150791
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.083786964416504,
      "learning_rate": 0.00016006216509528712,
      "loss": 2.0039,
      "step": 150792
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.0868325233459473,
      "learning_rate": 0.00016005854683186492,
      "loss": 1.8164,
      "step": 150793
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.0368893146514893,
      "learning_rate": 0.00016005492859446044,
      "loss": 1.9261,
      "step": 150794
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.2776340246200562,
      "learning_rate": 0.0001600513103830743,
      "loss": 2.1497,
      "step": 150795
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.1268912553787231,
      "learning_rate": 0.00016004769219770735,
      "loss": 2.0426,
      "step": 150796
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.2294689416885376,
      "learning_rate": 0.00016004407403836008,
      "loss": 2.2978,
      "step": 150797
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.2134740352630615,
      "learning_rate": 0.00016004045590503334,
      "loss": 2.155,
      "step": 150798
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.1162116527557373,
      "learning_rate": 0.00016003683779772777,
      "loss": 2.1825,
      "step": 150799
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.0871156454086304,
      "learning_rate": 0.00016003321971644395,
      "loss": 2.0535,
      "step": 150800
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.1740810871124268,
      "learning_rate": 0.00016002960166118253,
      "loss": 2.3179,
      "step": 150801
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.1765079498291016,
      "learning_rate": 0.00016002598363194431,
      "loss": 2.0138,
      "step": 150802
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.264224648475647,
      "learning_rate": 0.00016002236562872983,
      "loss": 2.0207,
      "step": 150803
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.233276128768921,
      "learning_rate": 0.0001600187476515399,
      "loss": 2.2211,
      "step": 150804
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.103058934211731,
      "learning_rate": 0.00016001512970037507,
      "loss": 1.9354,
      "step": 150805
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.1555572748184204,
      "learning_rate": 0.00016001151177523628,
      "loss": 1.8491,
      "step": 150806
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.1194766759872437,
      "learning_rate": 0.00016000789387612372,
      "loss": 2.0319,
      "step": 150807
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.1563061475753784,
      "learning_rate": 0.0001600042760030385,
      "loss": 1.9359,
      "step": 150808
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.1273962259292603,
      "learning_rate": 0.000160000658155981,
      "loss": 2.0585,
      "step": 150809
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.0366348028182983,
      "learning_rate": 0.00015999704033495214,
      "loss": 2.124,
      "step": 150810
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.3309420347213745,
      "learning_rate": 0.00015999342253995234,
      "loss": 1.8085,
      "step": 150811
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.2465956211090088,
      "learning_rate": 0.00015998980477098266,
      "loss": 1.987,
      "step": 150812
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.5320457220077515,
      "learning_rate": 0.00015998618702804324,
      "loss": 2.0117,
      "step": 150813
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.19082510471344,
      "learning_rate": 0.0001599825693111352,
      "loss": 1.9907,
      "step": 150814
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.1303431987762451,
      "learning_rate": 0.00015997895162025893,
      "loss": 2.0011,
      "step": 150815
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.2074743509292603,
      "learning_rate": 0.0001599753339554153,
      "loss": 1.9758,
      "step": 150816
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.4288573265075684,
      "learning_rate": 0.0001599717163166048,
      "loss": 2.24,
      "step": 150817
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.2527830600738525,
      "learning_rate": 0.00015996809870382842,
      "loss": 2.0953,
      "step": 150818
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.2948195934295654,
      "learning_rate": 0.00015996448111708637,
      "loss": 2.036,
      "step": 150819
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.0296733379364014,
      "learning_rate": 0.00015996086355637973,
      "loss": 2.2821,
      "step": 150820
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.0923075675964355,
      "learning_rate": 0.00015995724602170888,
      "loss": 2.0821,
      "step": 150821
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.134232997894287,
      "learning_rate": 0.0001599536285130747,
      "loss": 2.1514,
      "step": 150822
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.229288101196289,
      "learning_rate": 0.00015995001103047768,
      "loss": 2.3861,
      "step": 150823
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.410603642463684,
      "learning_rate": 0.00015994639357391885,
      "loss": 2.3236,
      "step": 150824
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.1747545003890991,
      "learning_rate": 0.00015994277614339838,
      "loss": 1.9227,
      "step": 150825
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.096648097038269,
      "learning_rate": 0.0001599391587389173,
      "loss": 2.1316,
      "step": 150826
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.3625167608261108,
      "learning_rate": 0.00015993554136047608,
      "loss": 2.092,
      "step": 150827
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.2635809183120728,
      "learning_rate": 0.00015993192400807558,
      "loss": 2.2352,
      "step": 150828
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.1175330877304077,
      "learning_rate": 0.0001599283066817163,
      "loss": 2.2975,
      "step": 150829
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.1209955215454102,
      "learning_rate": 0.0001599246893813992,
      "loss": 2.0076,
      "step": 150830
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.1539517641067505,
      "learning_rate": 0.00015992107210712451,
      "loss": 1.9791,
      "step": 150831
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.119450569152832,
      "learning_rate": 0.00015991745485889326,
      "loss": 2.0097,
      "step": 150832
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.1908128261566162,
      "learning_rate": 0.00015991383763670587,
      "loss": 2.2015,
      "step": 150833
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.1210167407989502,
      "learning_rate": 0.00015991022044056327,
      "loss": 2.1885,
      "step": 150834
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.191148042678833,
      "learning_rate": 0.0001599066032704659,
      "loss": 2.0805,
      "step": 150835
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.295399785041809,
      "learning_rate": 0.00015990298612641473,
      "loss": 2.0615,
      "step": 150836
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.3507298231124878,
      "learning_rate": 0.00015989936900841006,
      "loss": 1.9768,
      "step": 150837
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.2429195642471313,
      "learning_rate": 0.00015989575191645283,
      "loss": 2.0169,
      "step": 150838
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.1876893043518066,
      "learning_rate": 0.0001598921348505435,
      "loss": 1.849,
      "step": 150839
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.2099354267120361,
      "learning_rate": 0.000159888517810683,
      "loss": 2.0072,
      "step": 150840
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.254528522491455,
      "learning_rate": 0.00015988490079687172,
      "loss": 2.19,
      "step": 150841
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.1840386390686035,
      "learning_rate": 0.00015988128380911066,
      "loss": 2.239,
      "step": 150842
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.1165653467178345,
      "learning_rate": 0.00015987766684740024,
      "loss": 2.1417,
      "step": 150843
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.1619890928268433,
      "learning_rate": 0.00015987404991174125,
      "loss": 2.1657,
      "step": 150844
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.339021921157837,
      "learning_rate": 0.00015987043300213417,
      "loss": 2.0866,
      "step": 150845
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.3798147439956665,
      "learning_rate": 0.00015986681611858,
      "loss": 2.137,
      "step": 150846
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.2819966077804565,
      "learning_rate": 0.0001598631992610791,
      "loss": 2.1053,
      "step": 150847
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.1072838306427002,
      "learning_rate": 0.00015985958242963234,
      "loss": 1.9706,
      "step": 150848
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.3182430267333984,
      "learning_rate": 0.00015985596562424038,
      "loss": 2.0404,
      "step": 150849
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.118377923965454,
      "learning_rate": 0.00015985234884490382,
      "loss": 1.9303,
      "step": 150850
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.101434588432312,
      "learning_rate": 0.00015984873209162325,
      "loss": 2.1218,
      "step": 150851
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.148197054862976,
      "learning_rate": 0.00015984511536439958,
      "loss": 1.9338,
      "step": 150852
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.1917635202407837,
      "learning_rate": 0.00015984149866323316,
      "loss": 2.1254,
      "step": 150853
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.0128495693206787,
      "learning_rate": 0.00015983788198812503,
      "loss": 2.1449,
      "step": 150854
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.2177921533584595,
      "learning_rate": 0.00015983426533907566,
      "loss": 2.1055,
      "step": 150855
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.0844510793685913,
      "learning_rate": 0.00015983064871608562,
      "loss": 2.1682,
      "step": 150856
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.1263580322265625,
      "learning_rate": 0.00015982703211915589,
      "loss": 2.0337,
      "step": 150857
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.0784714221954346,
      "learning_rate": 0.0001598234155482869,
      "loss": 2.2217,
      "step": 150858
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.094670057296753,
      "learning_rate": 0.00015981979900347927,
      "loss": 2.0853,
      "step": 150859
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.10273015499115,
      "learning_rate": 0.00015981618248473394,
      "loss": 2.0933,
      "step": 150860
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.119555115699768,
      "learning_rate": 0.00015981256599205138,
      "loss": 2.092,
      "step": 150861
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.2428901195526123,
      "learning_rate": 0.0001598089495254322,
      "loss": 2.1633,
      "step": 150862
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.137553334236145,
      "learning_rate": 0.0001598053330848773,
      "loss": 2.2473,
      "step": 150863
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.20370352268219,
      "learning_rate": 0.00015980171667038715,
      "loss": 2.1448,
      "step": 150864
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.144136667251587,
      "learning_rate": 0.00015979810028196262,
      "loss": 2.0712,
      "step": 150865
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.0247299671173096,
      "learning_rate": 0.00015979448391960425,
      "loss": 1.9888,
      "step": 150866
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.1160646677017212,
      "learning_rate": 0.00015979086758331274,
      "loss": 2.1076,
      "step": 150867
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.2194993495941162,
      "learning_rate": 0.00015978725127308862,
      "loss": 2.1468,
      "step": 150868
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.0577044486999512,
      "learning_rate": 0.00015978363498893284,
      "loss": 1.9612,
      "step": 150869
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.2125811576843262,
      "learning_rate": 0.00015978001873084578,
      "loss": 2.1212,
      "step": 150870
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.169191598892212,
      "learning_rate": 0.0001597764024988284,
      "loss": 2.1281,
      "step": 150871
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.1343636512756348,
      "learning_rate": 0.00015977278629288126,
      "loss": 1.9896,
      "step": 150872
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.0284581184387207,
      "learning_rate": 0.000159769170113005,
      "loss": 2.4023,
      "step": 150873
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.300600290298462,
      "learning_rate": 0.00015976555395920013,
      "loss": 2.108,
      "step": 150874
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.3324763774871826,
      "learning_rate": 0.00015976193783146764,
      "loss": 1.8832,
      "step": 150875
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.1181856393814087,
      "learning_rate": 0.0001597583217298079,
      "loss": 2.2791,
      "step": 150876
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.0742065906524658,
      "learning_rate": 0.00015975470565422192,
      "loss": 2.1617,
      "step": 150877
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.1554484367370605,
      "learning_rate": 0.00015975108960471004,
      "loss": 2.178,
      "step": 150878
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.1448585987091064,
      "learning_rate": 0.00015974747358127328,
      "loss": 2.0119,
      "step": 150879
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.05828058719635,
      "learning_rate": 0.00015974385758391193,
      "loss": 1.8616,
      "step": 150880
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.092590093612671,
      "learning_rate": 0.00015974024161262694,
      "loss": 2.1038,
      "step": 150881
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.0747466087341309,
      "learning_rate": 0.00015973662566741876,
      "loss": 2.1792,
      "step": 150882
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.326662302017212,
      "learning_rate": 0.00015973300974828831,
      "loss": 2.0398,
      "step": 150883
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.1772408485412598,
      "learning_rate": 0.00015972939385523602,
      "loss": 2.0053,
      "step": 150884
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.141884684562683,
      "learning_rate": 0.00015972577798826297,
      "loss": 2.1045,
      "step": 150885
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.199733018875122,
      "learning_rate": 0.00015972216214736926,
      "loss": 2.0978,
      "step": 150886
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.0886473655700684,
      "learning_rate": 0.00015971854633255595,
      "loss": 1.9197,
      "step": 150887
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.0412824153900146,
      "learning_rate": 0.00015971493054382354,
      "loss": 1.9447,
      "step": 150888
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.1093964576721191,
      "learning_rate": 0.00015971131478117285,
      "loss": 1.9172,
      "step": 150889
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.1151008605957031,
      "learning_rate": 0.00015970769904460435,
      "loss": 2.2012,
      "step": 150890
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.1140445470809937,
      "learning_rate": 0.0001597040833341191,
      "loss": 1.9014,
      "step": 150891
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.0918281078338623,
      "learning_rate": 0.00015970046764971727,
      "loss": 2.068,
      "step": 150892
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.1583051681518555,
      "learning_rate": 0.00015969685199139993,
      "loss": 2.306,
      "step": 150893
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.0826183557510376,
      "learning_rate": 0.0001596932363591674,
      "loss": 1.9274,
      "step": 150894
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.0732648372650146,
      "learning_rate": 0.0001596896207530207,
      "loss": 2.2703,
      "step": 150895
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.0284621715545654,
      "learning_rate": 0.0001596860051729602,
      "loss": 1.9636,
      "step": 150896
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.0976523160934448,
      "learning_rate": 0.000159682389618987,
      "loss": 1.9479,
      "step": 150897
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.1850495338439941,
      "learning_rate": 0.00015967877409110126,
      "loss": 1.9969,
      "step": 150898
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.1876355409622192,
      "learning_rate": 0.00015967515858930398,
      "loss": 2.1707,
      "step": 150899
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.125340461730957,
      "learning_rate": 0.0001596715431135956,
      "loss": 2.036,
      "step": 150900
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.083269715309143,
      "learning_rate": 0.0001596679276639771,
      "loss": 1.9034,
      "step": 150901
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.1358708143234253,
      "learning_rate": 0.00015966431224044884,
      "loss": 2.0409,
      "step": 150902
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.2064125537872314,
      "learning_rate": 0.00015966069684301187,
      "loss": 2.2305,
      "step": 150903
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.0821449756622314,
      "learning_rate": 0.0001596570814716664,
      "loss": 1.9933,
      "step": 150904
    },
    {
      "epoch": 1.96,
      "grad_norm": 0.9791959524154663,
      "learning_rate": 0.0001596534661264134,
      "loss": 2.0607,
      "step": 150905
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.1310412883758545,
      "learning_rate": 0.00015964985080725336,
      "loss": 2.2673,
      "step": 150906
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.127593994140625,
      "learning_rate": 0.00015964623551418722,
      "loss": 2.0249,
      "step": 150907
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.147171974182129,
      "learning_rate": 0.00015964262024721536,
      "loss": 2.0853,
      "step": 150908
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.2471985816955566,
      "learning_rate": 0.0001596390050063388,
      "loss": 2.0989,
      "step": 150909
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.165576457977295,
      "learning_rate": 0.0001596353897915578,
      "loss": 1.9515,
      "step": 150910
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.2134389877319336,
      "learning_rate": 0.0001596317746028733,
      "loss": 2.2291,
      "step": 150911
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.0973671674728394,
      "learning_rate": 0.00015962815944028583,
      "loss": 1.88,
      "step": 150912
    },
    {
      "epoch": 1.96,
      "grad_norm": 1.0745158195495605,
      "learning_rate": 0.00015962454430379626,
      "loss": 2.1756,
      "step": 150913
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.1440174579620361,
      "learning_rate": 0.00015962092919340497,
      "loss": 1.9385,
      "step": 150914
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.05852210521698,
      "learning_rate": 0.0001596173141091131,
      "loss": 2.2458,
      "step": 150915
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.0824460983276367,
      "learning_rate": 0.00015961369905092068,
      "loss": 2.0623,
      "step": 150916
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.1153851747512817,
      "learning_rate": 0.00015961008401882895,
      "loss": 2.0134,
      "step": 150917
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.2586359977722168,
      "learning_rate": 0.0001596064690128382,
      "loss": 2.0166,
      "step": 150918
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.1886591911315918,
      "learning_rate": 0.00015960285403294938,
      "loss": 2.1289,
      "step": 150919
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.1887620687484741,
      "learning_rate": 0.00015959923907916292,
      "loss": 2.2189,
      "step": 150920
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.1276577711105347,
      "learning_rate": 0.0001595956241514798,
      "loss": 2.0457,
      "step": 150921
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.1719446182250977,
      "learning_rate": 0.0001595920092499003,
      "loss": 2.081,
      "step": 150922
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.267047643661499,
      "learning_rate": 0.00015958839437442543,
      "loss": 2.1553,
      "step": 150923
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.5290987491607666,
      "learning_rate": 0.00015958477952505558,
      "loss": 2.2337,
      "step": 150924
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.1529732942581177,
      "learning_rate": 0.00015958116470179173,
      "loss": 2.2119,
      "step": 150925
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.274090051651001,
      "learning_rate": 0.00015957754990463427,
      "loss": 1.9943,
      "step": 150926
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.2675200700759888,
      "learning_rate": 0.0001595739351335841,
      "loss": 2.0985,
      "step": 150927
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.155224323272705,
      "learning_rate": 0.00015957032038864177,
      "loss": 2.3718,
      "step": 150928
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.1277730464935303,
      "learning_rate": 0.000159566705669808,
      "loss": 1.8417,
      "step": 150929
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.2957158088684082,
      "learning_rate": 0.00015956309097708328,
      "loss": 2.1251,
      "step": 150930
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.2059658765792847,
      "learning_rate": 0.0001595594763104686,
      "loss": 2.1029,
      "step": 150931
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.0844755172729492,
      "learning_rate": 0.00015955586166996428,
      "loss": 1.9729,
      "step": 150932
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.0638995170593262,
      "learning_rate": 0.00015955224705557132,
      "loss": 1.7519,
      "step": 150933
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.2581675052642822,
      "learning_rate": 0.00015954863246729026,
      "loss": 2.1376,
      "step": 150934
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.147744059562683,
      "learning_rate": 0.00015954501790512175,
      "loss": 2.111,
      "step": 150935
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.163806438446045,
      "learning_rate": 0.00015954140336906636,
      "loss": 2.1496,
      "step": 150936
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.2000041007995605,
      "learning_rate": 0.000159537788859125,
      "loss": 2.319,
      "step": 150937
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.1224950551986694,
      "learning_rate": 0.0001595341743752981,
      "loss": 1.9485,
      "step": 150938
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.1818245649337769,
      "learning_rate": 0.00015953055991758654,
      "loss": 2.0996,
      "step": 150939
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.0035349130630493,
      "learning_rate": 0.0001595269454859909,
      "loss": 2.0249,
      "step": 150940
    },
    {
      "epoch": 1.97,
      "grad_norm": 0.9839731454849243,
      "learning_rate": 0.00015952333108051176,
      "loss": 2.317,
      "step": 150941
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.090927004814148,
      "learning_rate": 0.00015951971670114998,
      "loss": 1.8479,
      "step": 150942
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.1294304132461548,
      "learning_rate": 0.00015951610234790617,
      "loss": 2.1868,
      "step": 150943
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.1838618516921997,
      "learning_rate": 0.0001595124880207808,
      "loss": 1.8673,
      "step": 150944
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.0853883028030396,
      "learning_rate": 0.00015950887371977485,
      "loss": 1.7771,
      "step": 150945
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.1450759172439575,
      "learning_rate": 0.00015950525944488888,
      "loss": 2.1755,
      "step": 150946
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.1621272563934326,
      "learning_rate": 0.0001595016451961234,
      "loss": 2.0476,
      "step": 150947
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.1438266038894653,
      "learning_rate": 0.0001594980309734793,
      "loss": 2.068,
      "step": 150948
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.1677436828613281,
      "learning_rate": 0.00015949441677695725,
      "loss": 2.0224,
      "step": 150949
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.083006501197815,
      "learning_rate": 0.0001594908026065576,
      "loss": 1.9792,
      "step": 150950
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.1389992237091064,
      "learning_rate": 0.0001594871884622815,
      "loss": 2.0678,
      "step": 150951
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.1196562051773071,
      "learning_rate": 0.00015948357434412936,
      "loss": 1.944,
      "step": 150952
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.1794421672821045,
      "learning_rate": 0.0001594799602521017,
      "loss": 2.0871,
      "step": 150953
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.116117000579834,
      "learning_rate": 0.00015947634618619953,
      "loss": 2.0758,
      "step": 150954
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.1109808683395386,
      "learning_rate": 0.00015947273214642323,
      "loss": 1.9721,
      "step": 150955
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.324597716331482,
      "learning_rate": 0.0001594691181327737,
      "loss": 2.1338,
      "step": 150956
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.292611837387085,
      "learning_rate": 0.00015946550414525155,
      "loss": 2.0625,
      "step": 150957
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.2782819271087646,
      "learning_rate": 0.00015946189018385738,
      "loss": 1.8136,
      "step": 150958
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.1523865461349487,
      "learning_rate": 0.0001594582762485918,
      "loss": 2.0235,
      "step": 150959
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.3416324853897095,
      "learning_rate": 0.00015945466233945572,
      "loss": 2.1189,
      "step": 150960
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.1911802291870117,
      "learning_rate": 0.00015945104845644953,
      "loss": 2.2128,
      "step": 150961
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.1102936267852783,
      "learning_rate": 0.00015944743459957416,
      "loss": 2.2043,
      "step": 150962
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.0561579465866089,
      "learning_rate": 0.00015944382076883007,
      "loss": 2.0508,
      "step": 150963
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.1487247943878174,
      "learning_rate": 0.0001594402069642182,
      "loss": 2.2203,
      "step": 150964
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.18888258934021,
      "learning_rate": 0.00015943659318573886,
      "loss": 2.0577,
      "step": 150965
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.1591349840164185,
      "learning_rate": 0.00015943297943339302,
      "loss": 2.1654,
      "step": 150966
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.1706918478012085,
      "learning_rate": 0.00015942936570718114,
      "loss": 1.9377,
      "step": 150967
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.113774299621582,
      "learning_rate": 0.0001594257520071041,
      "loss": 1.8631,
      "step": 150968
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.5660439729690552,
      "learning_rate": 0.00015942213833316235,
      "loss": 1.9547,
      "step": 150969
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.5413470268249512,
      "learning_rate": 0.00015941852468535689,
      "loss": 2.2927,
      "step": 150970
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.3629869222640991,
      "learning_rate": 0.000159414911063688,
      "loss": 1.9348,
      "step": 150971
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.101546287536621,
      "learning_rate": 0.0001594112974681566,
      "loss": 2.1296,
      "step": 150972
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.153287410736084,
      "learning_rate": 0.0001594076838987632,
      "loss": 1.8428,
      "step": 150973
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.279462218284607,
      "learning_rate": 0.00015940407035550867,
      "loss": 2.0291,
      "step": 150974
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.2635613679885864,
      "learning_rate": 0.00015940045683839347,
      "loss": 1.9416,
      "step": 150975
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.2986055612564087,
      "learning_rate": 0.00015939684334741858,
      "loss": 2.0137,
      "step": 150976
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.0880708694458008,
      "learning_rate": 0.00015939322988258426,
      "loss": 2.1071,
      "step": 150977
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.2505244016647339,
      "learning_rate": 0.00015938961644389155,
      "loss": 2.173,
      "step": 150978
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.4138619899749756,
      "learning_rate": 0.00015938600303134077,
      "loss": 2.1327,
      "step": 150979
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.2023415565490723,
      "learning_rate": 0.00015938238964493295,
      "loss": 2.0616,
      "step": 150980
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.1518406867980957,
      "learning_rate": 0.0001593787762846685,
      "loss": 1.9472,
      "step": 150981
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.1324955224990845,
      "learning_rate": 0.00015937516295054838,
      "loss": 2.1635,
      "step": 150982
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.232806921005249,
      "learning_rate": 0.00015937154964257289,
      "loss": 1.9309,
      "step": 150983
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.2482868432998657,
      "learning_rate": 0.00015936793636074295,
      "loss": 2.0248,
      "step": 150984
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.19440758228302,
      "learning_rate": 0.0001593643231050591,
      "loss": 1.9959,
      "step": 150985
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.075994849205017,
      "learning_rate": 0.00015936070987552213,
      "loss": 1.9048,
      "step": 150986
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.1195640563964844,
      "learning_rate": 0.00015935709667213256,
      "loss": 1.9961,
      "step": 150987
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.208225131034851,
      "learning_rate": 0.00015935348349489144,
      "loss": 2.1398,
      "step": 150988
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.636054277420044,
      "learning_rate": 0.00015934987034379886,
      "loss": 1.9938,
      "step": 150989
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.2849613428115845,
      "learning_rate": 0.00015934625721885593,
      "loss": 2.0139,
      "step": 150990
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.1939760446548462,
      "learning_rate": 0.00015934264412006312,
      "loss": 1.9827,
      "step": 150991
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.120640516281128,
      "learning_rate": 0.00015933903104742124,
      "loss": 1.8361,
      "step": 150992
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.4163436889648438,
      "learning_rate": 0.0001593354180009308,
      "loss": 1.9755,
      "step": 150993
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.1540157794952393,
      "learning_rate": 0.00015933180498059276,
      "loss": 1.9322,
      "step": 150994
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.1115235090255737,
      "learning_rate": 0.00015932819198640738,
      "loss": 1.9336,
      "step": 150995
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.3692928552627563,
      "learning_rate": 0.00015932457901837564,
      "loss": 2.0555,
      "step": 150996
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.2420746088027954,
      "learning_rate": 0.00015932096607649803,
      "loss": 2.1834,
      "step": 150997
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.2086669206619263,
      "learning_rate": 0.00015931735316077544,
      "loss": 2.0223,
      "step": 150998
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.1906574964523315,
      "learning_rate": 0.00015931374027120825,
      "loss": 2.1835,
      "step": 150999
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.0779967308044434,
      "learning_rate": 0.0001593101274077974,
      "loss": 2.0288,
      "step": 151000
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.2006779909133911,
      "learning_rate": 0.00015930651457054347,
      "loss": 2.3301,
      "step": 151001
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.0238724946975708,
      "learning_rate": 0.00015930290175944714,
      "loss": 2.0822,
      "step": 151002
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.177718997001648,
      "learning_rate": 0.0001592992889745089,
      "loss": 2.0711,
      "step": 151003
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.300500512123108,
      "learning_rate": 0.00015929567621572973,
      "loss": 2.0706,
      "step": 151004
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.0557130575180054,
      "learning_rate": 0.00015929206348311002,
      "loss": 2.0953,
      "step": 151005
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.1058826446533203,
      "learning_rate": 0.00015928845077665068,
      "loss": 1.9742,
      "step": 151006
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.0989094972610474,
      "learning_rate": 0.00015928483809635226,
      "loss": 2.22,
      "step": 151007
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.17232084274292,
      "learning_rate": 0.00015928122544221546,
      "loss": 2.0946,
      "step": 151008
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.1281471252441406,
      "learning_rate": 0.00015927761281424083,
      "loss": 1.9866,
      "step": 151009
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.1991232633590698,
      "learning_rate": 0.00015927400021242927,
      "loss": 2.1029,
      "step": 151010
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.087743878364563,
      "learning_rate": 0.0001592703876367812,
      "loss": 2.3649,
      "step": 151011
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.3061702251434326,
      "learning_rate": 0.00015926677508729754,
      "loss": 2.2124,
      "step": 151012
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.1829572916030884,
      "learning_rate": 0.00015926316256397886,
      "loss": 2.081,
      "step": 151013
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.0612132549285889,
      "learning_rate": 0.00015925955006682578,
      "loss": 2.0157,
      "step": 151014
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.1346179246902466,
      "learning_rate": 0.0001592559375958389,
      "loss": 2.1905,
      "step": 151015
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.2327194213867188,
      "learning_rate": 0.00015925232515101915,
      "loss": 1.9269,
      "step": 151016
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.1508429050445557,
      "learning_rate": 0.00015924871273236688,
      "loss": 1.923,
      "step": 151017
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.0827476978302002,
      "learning_rate": 0.00015924510033988307,
      "loss": 1.968,
      "step": 151018
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.196077823638916,
      "learning_rate": 0.00015924148797356825,
      "loss": 1.8225,
      "step": 151019
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.0931299924850464,
      "learning_rate": 0.00015923787563342312,
      "loss": 1.9243,
      "step": 151020
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.188328742980957,
      "learning_rate": 0.00015923426331944818,
      "loss": 2.111,
      "step": 151021
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.139018177986145,
      "learning_rate": 0.00015923065103164437,
      "loss": 2.1129,
      "step": 151022
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.0694869756698608,
      "learning_rate": 0.00015922703877001214,
      "loss": 1.8423,
      "step": 151023
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.2700773477554321,
      "learning_rate": 0.00015922342653455234,
      "loss": 2.0817,
      "step": 151024
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.0368868112564087,
      "learning_rate": 0.00015921981432526558,
      "loss": 2.0163,
      "step": 151025
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.1030548810958862,
      "learning_rate": 0.00015921620214215244,
      "loss": 2.203,
      "step": 151026
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.2404119968414307,
      "learning_rate": 0.00015921258998521373,
      "loss": 1.994,
      "step": 151027
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.206098198890686,
      "learning_rate": 0.00015920897785445007,
      "loss": 2.1366,
      "step": 151028
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.0964863300323486,
      "learning_rate": 0.000159205365749862,
      "loss": 1.9979,
      "step": 151029
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.15367591381073,
      "learning_rate": 0.00015920175367145043,
      "loss": 1.8596,
      "step": 151030
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.1229326725006104,
      "learning_rate": 0.0001591981416192159,
      "loss": 2.0142,
      "step": 151031
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.1539549827575684,
      "learning_rate": 0.000159194529593159,
      "loss": 2.0782,
      "step": 151032
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.133880615234375,
      "learning_rate": 0.0001591909175932806,
      "loss": 2.112,
      "step": 151033
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.080224871635437,
      "learning_rate": 0.00015918730561958127,
      "loss": 2.0378,
      "step": 151034
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.1345760822296143,
      "learning_rate": 0.00015918369367206157,
      "loss": 1.9918,
      "step": 151035
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.0498669147491455,
      "learning_rate": 0.00015918008175072235,
      "loss": 1.9878,
      "step": 151036
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.0651476383209229,
      "learning_rate": 0.00015917646985556428,
      "loss": 1.9976,
      "step": 151037
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.152788519859314,
      "learning_rate": 0.0001591728579865878,
      "loss": 2.0852,
      "step": 151038
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.2085380554199219,
      "learning_rate": 0.00015916924614379387,
      "loss": 2.223,
      "step": 151039
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.0572218894958496,
      "learning_rate": 0.00015916563432718291,
      "loss": 1.7903,
      "step": 151040
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.287692666053772,
      "learning_rate": 0.00015916202253675585,
      "loss": 2.1618,
      "step": 151041
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.0469449758529663,
      "learning_rate": 0.00015915841077251324,
      "loss": 2.0778,
      "step": 151042
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.1059681177139282,
      "learning_rate": 0.00015915479903445574,
      "loss": 1.9321,
      "step": 151043
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.126476526260376,
      "learning_rate": 0.00015915118732258386,
      "loss": 2.1275,
      "step": 151044
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.2451831102371216,
      "learning_rate": 0.00015914757563689858,
      "loss": 2.0341,
      "step": 151045
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.1354258060455322,
      "learning_rate": 0.0001591439639774003,
      "loss": 1.9283,
      "step": 151046
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.1713026762008667,
      "learning_rate": 0.00015914035234408994,
      "loss": 2.083,
      "step": 151047
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.1036683320999146,
      "learning_rate": 0.00015913674073696793,
      "loss": 2.0189,
      "step": 151048
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.7957310676574707,
      "learning_rate": 0.0001591331291560353,
      "loss": 2.2548,
      "step": 151049
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.6779218912124634,
      "learning_rate": 0.00015912951760129224,
      "loss": 2.2356,
      "step": 151050
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.1528246402740479,
      "learning_rate": 0.0001591259060727398,
      "loss": 2.0647,
      "step": 151051
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.3506369590759277,
      "learning_rate": 0.00015912229457037836,
      "loss": 2.094,
      "step": 151052
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.3709070682525635,
      "learning_rate": 0.00015911868309420886,
      "loss": 2.0587,
      "step": 151053
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.1994510889053345,
      "learning_rate": 0.00015911507164423176,
      "loss": 1.9637,
      "step": 151054
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.209568738937378,
      "learning_rate": 0.00015911146022044807,
      "loss": 1.8802,
      "step": 151055
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.2138546705245972,
      "learning_rate": 0.00015910784882285797,
      "loss": 2.0757,
      "step": 151056
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.1912199258804321,
      "learning_rate": 0.0001591042374514625,
      "loss": 2.0579,
      "step": 151057
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.180574655532837,
      "learning_rate": 0.0001591006261062621,
      "loss": 2.1513,
      "step": 151058
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.1434777975082397,
      "learning_rate": 0.00015909701478725765,
      "loss": 2.0577,
      "step": 151059
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.140826940536499,
      "learning_rate": 0.0001590934034944496,
      "loss": 2.2765,
      "step": 151060
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.1529631614685059,
      "learning_rate": 0.000159089792227839,
      "loss": 2.0574,
      "step": 151061
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.2522233724594116,
      "learning_rate": 0.00015908618098742604,
      "loss": 2.1503,
      "step": 151062
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.3001337051391602,
      "learning_rate": 0.00015908256977321174,
      "loss": 2.159,
      "step": 151063
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.3140064477920532,
      "learning_rate": 0.00015907895858519651,
      "loss": 2.0447,
      "step": 151064
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.1827808618545532,
      "learning_rate": 0.00015907534742338127,
      "loss": 2.0363,
      "step": 151065
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.4266456365585327,
      "learning_rate": 0.0001590717362877665,
      "loss": 1.9967,
      "step": 151066
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.16629958152771,
      "learning_rate": 0.00015906812517835306,
      "loss": 2.0346,
      "step": 151067
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.123711347579956,
      "learning_rate": 0.00015906451409514153,
      "loss": 2.1354,
      "step": 151068
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.0921751260757446,
      "learning_rate": 0.00015906090303813256,
      "loss": 2.1387,
      "step": 151069
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.1734728813171387,
      "learning_rate": 0.00015905729200732668,
      "loss": 2.0513,
      "step": 151070
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.086786150932312,
      "learning_rate": 0.0001590536810027249,
      "loss": 2.0714,
      "step": 151071
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.0410231351852417,
      "learning_rate": 0.00015905007002432753,
      "loss": 1.9168,
      "step": 151072
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.07997727394104,
      "learning_rate": 0.00015904645907213554,
      "loss": 2.1236,
      "step": 151073
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.0616991519927979,
      "learning_rate": 0.00015904284814614945,
      "loss": 2.1105,
      "step": 151074
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.1752961874008179,
      "learning_rate": 0.00015903923724636996,
      "loss": 2.0646,
      "step": 151075
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.099312663078308,
      "learning_rate": 0.00015903562637279765,
      "loss": 2.0909,
      "step": 151076
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.1515806913375854,
      "learning_rate": 0.00015903201552543337,
      "loss": 1.961,
      "step": 151077
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.2319221496582031,
      "learning_rate": 0.0001590284047042776,
      "loss": 1.8638,
      "step": 151078
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.434983730316162,
      "learning_rate": 0.00015902479390933122,
      "loss": 2.035,
      "step": 151079
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.0385092496871948,
      "learning_rate": 0.00015902118314059478,
      "loss": 1.9042,
      "step": 151080
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.2044109106063843,
      "learning_rate": 0.00015901757239806898,
      "loss": 2.0256,
      "step": 151081
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.3601493835449219,
      "learning_rate": 0.00015901396168175433,
      "loss": 2.0999,
      "step": 151082
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.1322712898254395,
      "learning_rate": 0.0001590103509916518,
      "loss": 2.0993,
      "step": 151083
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.1485657691955566,
      "learning_rate": 0.0001590067403277618,
      "loss": 2.2182,
      "step": 151084
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.2574303150177002,
      "learning_rate": 0.00015900312969008518,
      "loss": 2.3162,
      "step": 151085
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.2530847787857056,
      "learning_rate": 0.00015899951907862253,
      "loss": 1.9469,
      "step": 151086
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.0250266790390015,
      "learning_rate": 0.0001589959084933746,
      "loss": 2.0686,
      "step": 151087
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.1032161712646484,
      "learning_rate": 0.0001589922979343418,
      "loss": 2.21,
      "step": 151088
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.2705090045928955,
      "learning_rate": 0.00015898868740152515,
      "loss": 2.1293,
      "step": 151089
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.1585320234298706,
      "learning_rate": 0.00015898507689492505,
      "loss": 2.1213,
      "step": 151090
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.3346260786056519,
      "learning_rate": 0.00015898146641454242,
      "loss": 2.2466,
      "step": 151091
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.122862696647644,
      "learning_rate": 0.00015897785596037775,
      "loss": 1.6766,
      "step": 151092
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.1161038875579834,
      "learning_rate": 0.0001589742455324318,
      "loss": 2.1461,
      "step": 151093
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.192842721939087,
      "learning_rate": 0.00015897063513070509,
      "loss": 2.1233,
      "step": 151094
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.2171566486358643,
      "learning_rate": 0.00015896702475519847,
      "loss": 1.901,
      "step": 151095
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.0863536596298218,
      "learning_rate": 0.00015896341440591247,
      "loss": 2.112,
      "step": 151096
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.0888087749481201,
      "learning_rate": 0.0001589598040828479,
      "loss": 2.0779,
      "step": 151097
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.1472702026367188,
      "learning_rate": 0.00015895619378600542,
      "loss": 2.1038,
      "step": 151098
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.278226613998413,
      "learning_rate": 0.00015895258351538563,
      "loss": 2.0438,
      "step": 151099
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.6642022132873535,
      "learning_rate": 0.00015894897327098907,
      "loss": 2.0108,
      "step": 151100
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.039746880531311,
      "learning_rate": 0.00015894536305281672,
      "loss": 1.9736,
      "step": 151101
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.1378698348999023,
      "learning_rate": 0.00015894175286086892,
      "loss": 1.953,
      "step": 151102
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.1425909996032715,
      "learning_rate": 0.00015893814269514668,
      "loss": 2.0844,
      "step": 151103
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.2058196067810059,
      "learning_rate": 0.00015893453255565046,
      "loss": 2.1162,
      "step": 151104
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.2830191850662231,
      "learning_rate": 0.00015893092244238102,
      "loss": 1.9639,
      "step": 151105
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.066425085067749,
      "learning_rate": 0.0001589273123553388,
      "loss": 1.9372,
      "step": 151106
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.1435198783874512,
      "learning_rate": 0.00015892370229452483,
      "loss": 1.9916,
      "step": 151107
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.429024338722229,
      "learning_rate": 0.0001589200922599395,
      "loss": 2.1188,
      "step": 151108
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.3043869733810425,
      "learning_rate": 0.00015891648225158363,
      "loss": 1.9917,
      "step": 151109
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.201501727104187,
      "learning_rate": 0.0001589128722694579,
      "loss": 2.1965,
      "step": 151110
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.0937542915344238,
      "learning_rate": 0.00015890926231356293,
      "loss": 2.1896,
      "step": 151111
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.262182593345642,
      "learning_rate": 0.00015890565238389927,
      "loss": 1.9643,
      "step": 151112
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.4999672174453735,
      "learning_rate": 0.00015890204248046784,
      "loss": 1.9132,
      "step": 151113
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.217629313468933,
      "learning_rate": 0.00015889843260326905,
      "loss": 1.9283,
      "step": 151114
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.1962229013442993,
      "learning_rate": 0.00015889482275230384,
      "loss": 1.9272,
      "step": 151115
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.1433812379837036,
      "learning_rate": 0.00015889121292757272,
      "loss": 1.9978,
      "step": 151116
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.2310094833374023,
      "learning_rate": 0.0001588876031290763,
      "loss": 2.0589,
      "step": 151117
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.2565363645553589,
      "learning_rate": 0.00015888399335681544,
      "loss": 2.0209,
      "step": 151118
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.2834450006484985,
      "learning_rate": 0.0001588803836107907,
      "loss": 2.1146,
      "step": 151119
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.3774547576904297,
      "learning_rate": 0.00015887677389100267,
      "loss": 2.1032,
      "step": 151120
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.0607407093048096,
      "learning_rate": 0.0001588731641974522,
      "loss": 2.1344,
      "step": 151121
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.4975157976150513,
      "learning_rate": 0.00015886955453013986,
      "loss": 1.9976,
      "step": 151122
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.2522423267364502,
      "learning_rate": 0.00015886594488906627,
      "loss": 2.1354,
      "step": 151123
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.160568118095398,
      "learning_rate": 0.00015886233527423227,
      "loss": 1.9636,
      "step": 151124
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.1438626050949097,
      "learning_rate": 0.00015885872568563827,
      "loss": 2.1355,
      "step": 151125
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.2543065547943115,
      "learning_rate": 0.00015885511612328523,
      "loss": 2.0379,
      "step": 151126
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.1219794750213623,
      "learning_rate": 0.00015885150658717369,
      "loss": 2.2652,
      "step": 151127
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.159557819366455,
      "learning_rate": 0.00015884789707730436,
      "loss": 2.2463,
      "step": 151128
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.1210297346115112,
      "learning_rate": 0.0001588442875936777,
      "loss": 1.9778,
      "step": 151129
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.0822192430496216,
      "learning_rate": 0.00015884067813629467,
      "loss": 2.0964,
      "step": 151130
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.1961617469787598,
      "learning_rate": 0.00015883706870515577,
      "loss": 2.0642,
      "step": 151131
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.2811965942382812,
      "learning_rate": 0.00015883345930026177,
      "loss": 2.0247,
      "step": 151132
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.1222271919250488,
      "learning_rate": 0.0001588298499216133,
      "loss": 2.1214,
      "step": 151133
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.2288347482681274,
      "learning_rate": 0.00015882624056921094,
      "loss": 2.1768,
      "step": 151134
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.1490697860717773,
      "learning_rate": 0.00015882263124305553,
      "loss": 1.9297,
      "step": 151135
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.0633114576339722,
      "learning_rate": 0.0001588190219431477,
      "loss": 2.2041,
      "step": 151136
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.0787805318832397,
      "learning_rate": 0.00015881541266948794,
      "loss": 2.2934,
      "step": 151137
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.1685694456100464,
      "learning_rate": 0.00015881180342207715,
      "loss": 2.1997,
      "step": 151138
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.140744924545288,
      "learning_rate": 0.00015880819420091582,
      "loss": 1.9788,
      "step": 151139
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.108701229095459,
      "learning_rate": 0.00015880458500600484,
      "loss": 2.0707,
      "step": 151140
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.1558940410614014,
      "learning_rate": 0.00015880097583734475,
      "loss": 2.0735,
      "step": 151141
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.249768853187561,
      "learning_rate": 0.0001587973666949362,
      "loss": 2.1379,
      "step": 151142
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.1087223291397095,
      "learning_rate": 0.0001587937575787798,
      "loss": 1.9581,
      "step": 151143
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.0279330015182495,
      "learning_rate": 0.0001587901484888764,
      "loss": 2.0053,
      "step": 151144
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.3683174848556519,
      "learning_rate": 0.00015878653942522647,
      "loss": 2.0744,
      "step": 151145
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.156245470046997,
      "learning_rate": 0.00015878293038783088,
      "loss": 2.1775,
      "step": 151146
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.2252955436706543,
      "learning_rate": 0.00015877932137669024,
      "loss": 1.9983,
      "step": 151147
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.1436262130737305,
      "learning_rate": 0.0001587757123918052,
      "loss": 2.0032,
      "step": 151148
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.0920273065567017,
      "learning_rate": 0.00015877210343317625,
      "loss": 1.9286,
      "step": 151149
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.1719956398010254,
      "learning_rate": 0.00015876849450080437,
      "loss": 1.9907,
      "step": 151150
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.1545560359954834,
      "learning_rate": 0.00015876488559469002,
      "loss": 1.9184,
      "step": 151151
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.2176899909973145,
      "learning_rate": 0.000158761276714834,
      "loss": 1.9256,
      "step": 151152
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.1172751188278198,
      "learning_rate": 0.00015875766786123697,
      "loss": 2.2511,
      "step": 151153
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.0784517526626587,
      "learning_rate": 0.00015875405903389956,
      "loss": 1.8492,
      "step": 151154
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.057206153869629,
      "learning_rate": 0.0001587504502328223,
      "loss": 1.952,
      "step": 151155
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.087700366973877,
      "learning_rate": 0.0001587468414580061,
      "loss": 1.9685,
      "step": 151156
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.275779128074646,
      "learning_rate": 0.0001587432327094514,
      "loss": 2.4149,
      "step": 151157
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.1391969919204712,
      "learning_rate": 0.00015873962398715917,
      "loss": 2.2053,
      "step": 151158
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.0802762508392334,
      "learning_rate": 0.0001587360152911299,
      "loss": 2.0279,
      "step": 151159
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.164361596107483,
      "learning_rate": 0.0001587324066213642,
      "loss": 2.1029,
      "step": 151160
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.1101038455963135,
      "learning_rate": 0.00015872879797786279,
      "loss": 2.0157,
      "step": 151161
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.1377520561218262,
      "learning_rate": 0.00015872518936062643,
      "loss": 2.28,
      "step": 151162
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.2481715679168701,
      "learning_rate": 0.0001587215807696556,
      "loss": 1.809,
      "step": 151163
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.1054308414459229,
      "learning_rate": 0.00015871797220495126,
      "loss": 1.9649,
      "step": 151164
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.0637016296386719,
      "learning_rate": 0.00015871436366651388,
      "loss": 2.1964,
      "step": 151165
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.1166326999664307,
      "learning_rate": 0.00015871075515434414,
      "loss": 2.0922,
      "step": 151166
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.0870764255523682,
      "learning_rate": 0.00015870714666844267,
      "loss": 2.0634,
      "step": 151167
    },
    {
      "epoch": 1.97,
      "grad_norm": 0.9890732169151306,
      "learning_rate": 0.0001587035382088103,
      "loss": 2.2212,
      "step": 151168
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.0747774839401245,
      "learning_rate": 0.00015869992977544752,
      "loss": 1.7817,
      "step": 151169
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.0758819580078125,
      "learning_rate": 0.00015869632136835519,
      "loss": 2.0315,
      "step": 151170
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.1114774942398071,
      "learning_rate": 0.00015869271298753386,
      "loss": 2.0522,
      "step": 151171
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.0791385173797607,
      "learning_rate": 0.00015868910463298424,
      "loss": 2.1502,
      "step": 151172
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.035285472869873,
      "learning_rate": 0.00015868549630470685,
      "loss": 2.1522,
      "step": 151173
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.1492438316345215,
      "learning_rate": 0.00015868188800270268,
      "loss": 2.0626,
      "step": 151174
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.160332441329956,
      "learning_rate": 0.00015867827972697201,
      "loss": 2.2364,
      "step": 151175
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.1232266426086426,
      "learning_rate": 0.00015867467147751585,
      "loss": 2.1386,
      "step": 151176
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.103589415550232,
      "learning_rate": 0.00015867106325433478,
      "loss": 2.0011,
      "step": 151177
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.1788766384124756,
      "learning_rate": 0.0001586674550574294,
      "loss": 1.9058,
      "step": 151178
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.003559947013855,
      "learning_rate": 0.00015866384688680027,
      "loss": 2.1287,
      "step": 151179
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.1409759521484375,
      "learning_rate": 0.00015866023874244834,
      "loss": 2.1034,
      "step": 151180
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.087768793106079,
      "learning_rate": 0.00015865663062437403,
      "loss": 2.2552,
      "step": 151181
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.097407579421997,
      "learning_rate": 0.00015865302253257823,
      "loss": 2.0085,
      "step": 151182
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.120740294456482,
      "learning_rate": 0.0001586494144670615,
      "loss": 2.0613,
      "step": 151183
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.2073746919631958,
      "learning_rate": 0.0001586458064278245,
      "loss": 2.2138,
      "step": 151184
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.1296557188034058,
      "learning_rate": 0.0001586421984148678,
      "loss": 1.9183,
      "step": 151185
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.1451489925384521,
      "learning_rate": 0.00015863859042819232,
      "loss": 2.2199,
      "step": 151186
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.052960991859436,
      "learning_rate": 0.00015863498246779844,
      "loss": 1.8405,
      "step": 151187
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.1276253461837769,
      "learning_rate": 0.00015863137453368717,
      "loss": 2.2362,
      "step": 151188
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.059358835220337,
      "learning_rate": 0.00015862776662585896,
      "loss": 2.0466,
      "step": 151189
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.1290864944458008,
      "learning_rate": 0.0001586241587443145,
      "loss": 2.1944,
      "step": 151190
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.2988263368606567,
      "learning_rate": 0.00015862055088905434,
      "loss": 2.1249,
      "step": 151191
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.0862807035446167,
      "learning_rate": 0.00015861694306007945,
      "loss": 2.0517,
      "step": 151192
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.072821021080017,
      "learning_rate": 0.0001586133352573902,
      "loss": 1.9218,
      "step": 151193
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.1498968601226807,
      "learning_rate": 0.00015860972748098754,
      "loss": 2.1831,
      "step": 151194
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.0517417192459106,
      "learning_rate": 0.00015860611973087196,
      "loss": 2.0294,
      "step": 151195
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.1101173162460327,
      "learning_rate": 0.00015860251200704422,
      "loss": 1.984,
      "step": 151196
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.23936128616333,
      "learning_rate": 0.00015859890430950478,
      "loss": 2.2203,
      "step": 151197
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.1066032648086548,
      "learning_rate": 0.00015859529663825463,
      "loss": 2.1146,
      "step": 151198
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.149777889251709,
      "learning_rate": 0.00015859168899329414,
      "loss": 1.9846,
      "step": 151199
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.096282720565796,
      "learning_rate": 0.00015858808137462428,
      "loss": 2.2958,
      "step": 151200
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.1274912357330322,
      "learning_rate": 0.00015858447378224537,
      "loss": 1.9379,
      "step": 151201
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.1074752807617188,
      "learning_rate": 0.00015858086621615844,
      "loss": 2.1242,
      "step": 151202
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.0657637119293213,
      "learning_rate": 0.00015857725867636402,
      "loss": 1.9444,
      "step": 151203
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.0968801975250244,
      "learning_rate": 0.0001585736511628627,
      "loss": 2.1332,
      "step": 151204
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.1165598630905151,
      "learning_rate": 0.00015857004367565513,
      "loss": 1.9711,
      "step": 151205
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.1901839971542358,
      "learning_rate": 0.00015856643621474218,
      "loss": 2.1612,
      "step": 151206
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.1287567615509033,
      "learning_rate": 0.00015856282878012426,
      "loss": 2.2017,
      "step": 151207
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.1079400777816772,
      "learning_rate": 0.00015855922137180233,
      "loss": 2.0609,
      "step": 151208
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.2917300462722778,
      "learning_rate": 0.0001585556139897769,
      "loss": 2.0057,
      "step": 151209
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.1538101434707642,
      "learning_rate": 0.0001585520066340485,
      "loss": 2.1022,
      "step": 151210
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.1753565073013306,
      "learning_rate": 0.0001585483993046181,
      "loss": 2.1319,
      "step": 151211
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.1548267602920532,
      "learning_rate": 0.0001585447920014862,
      "loss": 2.1659,
      "step": 151212
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.1465116739273071,
      "learning_rate": 0.00015854118472465343,
      "loss": 2.1971,
      "step": 151213
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.1739394664764404,
      "learning_rate": 0.00015853757747412057,
      "loss": 2.0742,
      "step": 151214
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.100726842880249,
      "learning_rate": 0.0001585339702498883,
      "loss": 2.2172,
      "step": 151215
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.1758123636245728,
      "learning_rate": 0.00015853036305195708,
      "loss": 1.9294,
      "step": 151216
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.455573558807373,
      "learning_rate": 0.00015852675588032787,
      "loss": 2.1687,
      "step": 151217
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.1282308101654053,
      "learning_rate": 0.0001585231487350012,
      "loss": 2.0752,
      "step": 151218
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.159279465675354,
      "learning_rate": 0.00015851954161597763,
      "loss": 2.0923,
      "step": 151219
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.0941191911697388,
      "learning_rate": 0.00015851593452325806,
      "loss": 2.1834,
      "step": 151220
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.2248398065567017,
      "learning_rate": 0.0001585123274568431,
      "loss": 2.043,
      "step": 151221
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.1222575902938843,
      "learning_rate": 0.00015850872041673318,
      "loss": 2.0312,
      "step": 151222
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.190027117729187,
      "learning_rate": 0.00015850511340292933,
      "loss": 2.0864,
      "step": 151223
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.194549560546875,
      "learning_rate": 0.0001585015064154319,
      "loss": 2.1147,
      "step": 151224
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.0894814729690552,
      "learning_rate": 0.00015849789945424184,
      "loss": 2.0692,
      "step": 151225
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.1588724851608276,
      "learning_rate": 0.0001584942925193597,
      "loss": 2.1417,
      "step": 151226
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.2331229448318481,
      "learning_rate": 0.00015849068561078612,
      "loss": 2.2926,
      "step": 151227
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.0881460905075073,
      "learning_rate": 0.0001584870787285217,
      "loss": 2.1152,
      "step": 151228
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.2222511768341064,
      "learning_rate": 0.00015848347187256729,
      "loss": 2.1622,
      "step": 151229
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.1296769380569458,
      "learning_rate": 0.0001584798650429234,
      "loss": 2.1629,
      "step": 151230
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.2307462692260742,
      "learning_rate": 0.00015847625823959088,
      "loss": 1.9741,
      "step": 151231
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.2567713260650635,
      "learning_rate": 0.00015847265146257025,
      "loss": 2.0241,
      "step": 151232
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.2150036096572876,
      "learning_rate": 0.00015846904471186225,
      "loss": 2.1858,
      "step": 151233
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.0782934427261353,
      "learning_rate": 0.00015846543798746744,
      "loss": 2.1103,
      "step": 151234
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.236106038093567,
      "learning_rate": 0.00015846183128938665,
      "loss": 2.1271,
      "step": 151235
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.4161230325698853,
      "learning_rate": 0.00015845822461762042,
      "loss": 2.1825,
      "step": 151236
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.0962419509887695,
      "learning_rate": 0.00015845461797216951,
      "loss": 2.0112,
      "step": 151237
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.2495007514953613,
      "learning_rate": 0.00015845101135303463,
      "loss": 2.0565,
      "step": 151238
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.3897005319595337,
      "learning_rate": 0.00015844740476021635,
      "loss": 2.0836,
      "step": 151239
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.1578009128570557,
      "learning_rate": 0.00015844379819371525,
      "loss": 2.2236,
      "step": 151240
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.141187310218811,
      "learning_rate": 0.00015844019165353224,
      "loss": 2.0273,
      "step": 151241
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.2477138042449951,
      "learning_rate": 0.00015843658513966776,
      "loss": 1.9995,
      "step": 151242
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.4429821968078613,
      "learning_rate": 0.0001584329786521227,
      "loss": 2.0178,
      "step": 151243
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.1023406982421875,
      "learning_rate": 0.00015842937219089762,
      "loss": 2.2069,
      "step": 151244
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.1505918502807617,
      "learning_rate": 0.00015842576575599322,
      "loss": 2.2011,
      "step": 151245
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.1425656080245972,
      "learning_rate": 0.00015842215934741002,
      "loss": 2.0317,
      "step": 151246
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.071742057800293,
      "learning_rate": 0.0001584185529651489,
      "loss": 2.2455,
      "step": 151247
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.1331136226654053,
      "learning_rate": 0.0001584149466092103,
      "loss": 2.1629,
      "step": 151248
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.1706345081329346,
      "learning_rate": 0.00015841134027959523,
      "loss": 1.8917,
      "step": 151249
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.1061607599258423,
      "learning_rate": 0.0001584077339763041,
      "loss": 2.1111,
      "step": 151250
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.0613781213760376,
      "learning_rate": 0.0001584041276993377,
      "loss": 2.2678,
      "step": 151251
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.1973698139190674,
      "learning_rate": 0.00015840052144869648,
      "loss": 2.2809,
      "step": 151252
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.261788010597229,
      "learning_rate": 0.0001583969152243814,
      "loss": 2.0075,
      "step": 151253
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.121820330619812,
      "learning_rate": 0.00015839330902639293,
      "loss": 1.9705,
      "step": 151254
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.1333913803100586,
      "learning_rate": 0.00015838970285473187,
      "loss": 2.2042,
      "step": 151255
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.1216422319412231,
      "learning_rate": 0.0001583860967093989,
      "loss": 2.0858,
      "step": 151256
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.0897252559661865,
      "learning_rate": 0.00015838249059039458,
      "loss": 2.0233,
      "step": 151257
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.1216031312942505,
      "learning_rate": 0.00015837888449771954,
      "loss": 2.0633,
      "step": 151258
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.1637629270553589,
      "learning_rate": 0.00015837527843137465,
      "loss": 2.1766,
      "step": 151259
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.1241720914840698,
      "learning_rate": 0.00015837167239136037,
      "loss": 2.3888,
      "step": 151260
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.0867552757263184,
      "learning_rate": 0.00015836806637767754,
      "loss": 2.0088,
      "step": 151261
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.156692385673523,
      "learning_rate": 0.00015836446039032683,
      "loss": 2.0485,
      "step": 151262
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.1131917238235474,
      "learning_rate": 0.00015836085442930877,
      "loss": 2.2567,
      "step": 151263
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.1609421968460083,
      "learning_rate": 0.000158357248494624,
      "loss": 1.9882,
      "step": 151264
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.0759004354476929,
      "learning_rate": 0.0001583536425862734,
      "loss": 2.122,
      "step": 151265
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.0526936054229736,
      "learning_rate": 0.00015835003670425744,
      "loss": 1.7179,
      "step": 151266
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.0574535131454468,
      "learning_rate": 0.000158346430848577,
      "loss": 1.9653,
      "step": 151267
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.1919351816177368,
      "learning_rate": 0.00015834282501923252,
      "loss": 2.1619,
      "step": 151268
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.2518028020858765,
      "learning_rate": 0.000158339219216225,
      "loss": 2.0702,
      "step": 151269
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.1075152158737183,
      "learning_rate": 0.00015833561343955464,
      "loss": 1.9332,
      "step": 151270
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.2607194185256958,
      "learning_rate": 0.00015833200768922254,
      "loss": 2.2021,
      "step": 151271
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.1795610189437866,
      "learning_rate": 0.000158328401965229,
      "loss": 2.0537,
      "step": 151272
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.4192421436309814,
      "learning_rate": 0.00015832479626757507,
      "loss": 1.8526,
      "step": 151273
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.0043349266052246,
      "learning_rate": 0.0001583211905962611,
      "loss": 1.8254,
      "step": 151274
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.08059561252594,
      "learning_rate": 0.0001583175849512881,
      "loss": 2.2635,
      "step": 151275
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.261941909790039,
      "learning_rate": 0.00015831397933265633,
      "loss": 1.8361,
      "step": 151276
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.1751612424850464,
      "learning_rate": 0.0001583103737403668,
      "loss": 2.0099,
      "step": 151277
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.2892305850982666,
      "learning_rate": 0.0001583067681744199,
      "loss": 2.0518,
      "step": 151278
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.263075351715088,
      "learning_rate": 0.00015830316263481655,
      "loss": 2.2025,
      "step": 151279
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.0750864744186401,
      "learning_rate": 0.00015829955712155723,
      "loss": 2.0369,
      "step": 151280
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.1426563262939453,
      "learning_rate": 0.0001582959516346429,
      "loss": 2.1111,
      "step": 151281
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.0805554389953613,
      "learning_rate": 0.00015829234617407382,
      "loss": 2.0628,
      "step": 151282
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.226261019706726,
      "learning_rate": 0.000158288740739851,
      "loss": 2.2721,
      "step": 151283
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.009067177772522,
      "learning_rate": 0.0001582851353319748,
      "loss": 1.8915,
      "step": 151284
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.1152558326721191,
      "learning_rate": 0.0001582815299504462,
      "loss": 2.1446,
      "step": 151285
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.1892582178115845,
      "learning_rate": 0.00015827792459526568,
      "loss": 2.1703,
      "step": 151286
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.1368886232376099,
      "learning_rate": 0.00015827431926643406,
      "loss": 2.0756,
      "step": 151287
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.1959733963012695,
      "learning_rate": 0.0001582707139639519,
      "loss": 1.9634,
      "step": 151288
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.0474580526351929,
      "learning_rate": 0.00015826710868781986,
      "loss": 2.024,
      "step": 151289
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.1327439546585083,
      "learning_rate": 0.00015826350343803858,
      "loss": 2.1561,
      "step": 151290
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.0994784832000732,
      "learning_rate": 0.0001582598982146089,
      "loss": 2.2024,
      "step": 151291
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.3628056049346924,
      "learning_rate": 0.00015825629301753123,
      "loss": 2.1653,
      "step": 151292
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.113439679145813,
      "learning_rate": 0.00015825268784680653,
      "loss": 2.0986,
      "step": 151293
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.1479229927062988,
      "learning_rate": 0.00015824908270243538,
      "loss": 1.9622,
      "step": 151294
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.174558401107788,
      "learning_rate": 0.00015824547758441833,
      "loss": 1.8901,
      "step": 151295
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.2154145240783691,
      "learning_rate": 0.00015824187249275599,
      "loss": 2.0342,
      "step": 151296
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.126278042793274,
      "learning_rate": 0.00015823826742744932,
      "loss": 2.0358,
      "step": 151297
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.0440832376480103,
      "learning_rate": 0.00015823466238849872,
      "loss": 1.9535,
      "step": 151298
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.1571329832077026,
      "learning_rate": 0.0001582310573759051,
      "loss": 2.0912,
      "step": 151299
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.205148458480835,
      "learning_rate": 0.00015822745238966898,
      "loss": 2.0862,
      "step": 151300
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.2162797451019287,
      "learning_rate": 0.0001582238474297909,
      "loss": 2.0572,
      "step": 151301
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.1723793745040894,
      "learning_rate": 0.00015822024249627188,
      "loss": 1.9586,
      "step": 151302
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.2353650331497192,
      "learning_rate": 0.00015821663758911237,
      "loss": 2.2434,
      "step": 151303
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.0807218551635742,
      "learning_rate": 0.0001582130327083129,
      "loss": 1.904,
      "step": 151304
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.1536306142807007,
      "learning_rate": 0.00015820942785387443,
      "loss": 2.0587,
      "step": 151305
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.1615890264511108,
      "learning_rate": 0.00015820582302579757,
      "loss": 1.842,
      "step": 151306
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.0593136548995972,
      "learning_rate": 0.00015820221822408274,
      "loss": 2.0165,
      "step": 151307
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.1072306632995605,
      "learning_rate": 0.0001581986134487309,
      "loss": 1.9838,
      "step": 151308
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.1181273460388184,
      "learning_rate": 0.00015819500869974253,
      "loss": 2.0979,
      "step": 151309
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.0949827432632446,
      "learning_rate": 0.0001581914039771185,
      "loss": 2.0826,
      "step": 151310
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.3382679224014282,
      "learning_rate": 0.00015818779928085936,
      "loss": 2.148,
      "step": 151311
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.2490988969802856,
      "learning_rate": 0.00015818419461096579,
      "loss": 2.176,
      "step": 151312
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.140391230583191,
      "learning_rate": 0.00015818058996743835,
      "loss": 2.1954,
      "step": 151313
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.7157151699066162,
      "learning_rate": 0.00015817698535027793,
      "loss": 2.2401,
      "step": 151314
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.3292793035507202,
      "learning_rate": 0.00015817338075948498,
      "loss": 2.0972,
      "step": 151315
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.2086178064346313,
      "learning_rate": 0.00015816977619506037,
      "loss": 2.0853,
      "step": 151316
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.0593438148498535,
      "learning_rate": 0.00015816617165700467,
      "loss": 2.016,
      "step": 151317
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.3179391622543335,
      "learning_rate": 0.00015816256714531855,
      "loss": 2.1037,
      "step": 151318
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.7024165391921997,
      "learning_rate": 0.0001581589626600026,
      "loss": 2.2385,
      "step": 151319
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.1753028631210327,
      "learning_rate": 0.00015815535820105768,
      "loss": 2.1916,
      "step": 151320
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.1113163232803345,
      "learning_rate": 0.0001581517537684842,
      "loss": 2.2178,
      "step": 151321
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.3144959211349487,
      "learning_rate": 0.00015814814936228314,
      "loss": 2.1792,
      "step": 151322
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.5757273435592651,
      "learning_rate": 0.000158144544982455,
      "loss": 2.1739,
      "step": 151323
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.9456815719604492,
      "learning_rate": 0.0001581409406290005,
      "loss": 2.2206,
      "step": 151324
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.2468206882476807,
      "learning_rate": 0.00015813733630192016,
      "loss": 2.0027,
      "step": 151325
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.1742388010025024,
      "learning_rate": 0.00015813373200121485,
      "loss": 2.0991,
      "step": 151326
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.5389119386672974,
      "learning_rate": 0.00015813012772688506,
      "loss": 1.9122,
      "step": 151327
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.5206325054168701,
      "learning_rate": 0.00015812652347893167,
      "loss": 2.1731,
      "step": 151328
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.206791639328003,
      "learning_rate": 0.00015812291925735523,
      "loss": 2.1044,
      "step": 151329
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.0630215406417847,
      "learning_rate": 0.00015811931506215645,
      "loss": 2.315,
      "step": 151330
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.2985763549804688,
      "learning_rate": 0.0001581157108933358,
      "loss": 2.0099,
      "step": 151331
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.1935997009277344,
      "learning_rate": 0.0001581121067508943,
      "loss": 2.3052,
      "step": 151332
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.1577690839767456,
      "learning_rate": 0.00015810850263483223,
      "loss": 2.1193,
      "step": 151333
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.1339116096496582,
      "learning_rate": 0.00015810489854515067,
      "loss": 1.9026,
      "step": 151334
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.1707652807235718,
      "learning_rate": 0.00015810129448184994,
      "loss": 2.249,
      "step": 151335
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.0868216753005981,
      "learning_rate": 0.0001580976904449311,
      "loss": 2.0671,
      "step": 151336
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.2429295778274536,
      "learning_rate": 0.0001580940864343943,
      "loss": 2.3466,
      "step": 151337
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.0831308364868164,
      "learning_rate": 0.00015809048245024065,
      "loss": 2.1846,
      "step": 151338
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.1605349779129028,
      "learning_rate": 0.0001580868784924705,
      "loss": 2.0663,
      "step": 151339
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.0792722702026367,
      "learning_rate": 0.00015808327456108486,
      "loss": 2.1689,
      "step": 151340
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.261080026626587,
      "learning_rate": 0.00015807967065608407,
      "loss": 2.1507,
      "step": 151341
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.1137107610702515,
      "learning_rate": 0.00015807606677746916,
      "loss": 2.0701,
      "step": 151342
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.0822315216064453,
      "learning_rate": 0.0001580724629252404,
      "loss": 1.98,
      "step": 151343
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.1217056512832642,
      "learning_rate": 0.0001580688590993987,
      "loss": 2.0354,
      "step": 151344
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.2075484991073608,
      "learning_rate": 0.0001580652552999446,
      "loss": 2.0738,
      "step": 151345
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.2711671590805054,
      "learning_rate": 0.00015806165152687895,
      "loss": 2.1821,
      "step": 151346
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.1399765014648438,
      "learning_rate": 0.00015805804778020222,
      "loss": 1.9333,
      "step": 151347
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.153370976448059,
      "learning_rate": 0.00015805444405991544,
      "loss": 2.1749,
      "step": 151348
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.1944423913955688,
      "learning_rate": 0.00015805084036601874,
      "loss": 2.1126,
      "step": 151349
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.1156740188598633,
      "learning_rate": 0.0001580472366985132,
      "loss": 2.1674,
      "step": 151350
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.2384363412857056,
      "learning_rate": 0.00015804363305739922,
      "loss": 2.2578,
      "step": 151351
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.2685739994049072,
      "learning_rate": 0.00015804002944267773,
      "loss": 2.0752,
      "step": 151352
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.2778582572937012,
      "learning_rate": 0.0001580364258543492,
      "loss": 2.144,
      "step": 151353
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.0953580141067505,
      "learning_rate": 0.00015803282229241461,
      "loss": 1.931,
      "step": 151354
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.1808042526245117,
      "learning_rate": 0.00015802921875687412,
      "loss": 2.084,
      "step": 151355
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.1847894191741943,
      "learning_rate": 0.00015802561524772883,
      "loss": 2.1675,
      "step": 151356
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.1827890872955322,
      "learning_rate": 0.0001580220117649791,
      "loss": 2.1303,
      "step": 151357
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.105372428894043,
      "learning_rate": 0.0001580184083086259,
      "loss": 2.1502,
      "step": 151358
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.1764938831329346,
      "learning_rate": 0.00015801480487866965,
      "loss": 1.9182,
      "step": 151359
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.15074622631073,
      "learning_rate": 0.00015801120147511136,
      "loss": 2.0647,
      "step": 151360
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.1362122297286987,
      "learning_rate": 0.0001580075980979512,
      "loss": 1.9539,
      "step": 151361
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.2205688953399658,
      "learning_rate": 0.00015800399474719028,
      "loss": 1.9254,
      "step": 151362
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.311436653137207,
      "learning_rate": 0.00015800039142282896,
      "loss": 2.0061,
      "step": 151363
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.1780046224594116,
      "learning_rate": 0.00015799678812486814,
      "loss": 2.0437,
      "step": 151364
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.2759543657302856,
      "learning_rate": 0.00015799318485330832,
      "loss": 2.0985,
      "step": 151365
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.1095553636550903,
      "learning_rate": 0.00015798958160815048,
      "loss": 1.9688,
      "step": 151366
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.1257288455963135,
      "learning_rate": 0.00015798597838939478,
      "loss": 1.926,
      "step": 151367
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.0705506801605225,
      "learning_rate": 0.00015798237519704228,
      "loss": 2.1346,
      "step": 151368
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.2164846658706665,
      "learning_rate": 0.00015797877203109348,
      "loss": 1.9249,
      "step": 151369
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.2705919742584229,
      "learning_rate": 0.0001579751688915492,
      "loss": 1.9909,
      "step": 151370
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.2097588777542114,
      "learning_rate": 0.00015797156577840988,
      "loss": 2.291,
      "step": 151371
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.2584036588668823,
      "learning_rate": 0.00015796796269167647,
      "loss": 2.1619,
      "step": 151372
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.1079931259155273,
      "learning_rate": 0.00015796435963134946,
      "loss": 1.7909,
      "step": 151373
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.2266285419464111,
      "learning_rate": 0.00015796075659742958,
      "loss": 2.2547,
      "step": 151374
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.1117581129074097,
      "learning_rate": 0.00015795715358991736,
      "loss": 2.1221,
      "step": 151375
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.100961685180664,
      "learning_rate": 0.0001579535506088137,
      "loss": 2.1775,
      "step": 151376
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.2134003639221191,
      "learning_rate": 0.00015794994765411908,
      "loss": 2.1347,
      "step": 151377
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.153549313545227,
      "learning_rate": 0.00015794634472583429,
      "loss": 2.0035,
      "step": 151378
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.1904910802841187,
      "learning_rate": 0.00015794274182395998,
      "loss": 2.1244,
      "step": 151379
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.1104217767715454,
      "learning_rate": 0.0001579391389484968,
      "loss": 1.9812,
      "step": 151380
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.234483003616333,
      "learning_rate": 0.00015793553609944532,
      "loss": 2.1628,
      "step": 151381
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.1053804159164429,
      "learning_rate": 0.0001579319332768064,
      "loss": 2.0831,
      "step": 151382
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.1043424606323242,
      "learning_rate": 0.00015792833048058053,
      "loss": 2.0088,
      "step": 151383
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.1128108501434326,
      "learning_rate": 0.00015792472771076859,
      "loss": 2.1242,
      "step": 151384
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.278960943222046,
      "learning_rate": 0.00015792112496737108,
      "loss": 2.11,
      "step": 151385
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.1513495445251465,
      "learning_rate": 0.0001579175222503886,
      "loss": 2.0965,
      "step": 151386
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.18489408493042,
      "learning_rate": 0.0001579139195598221,
      "loss": 2.0742,
      "step": 151387
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.1137827634811401,
      "learning_rate": 0.00015791031689567201,
      "loss": 2.0723,
      "step": 151388
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.1807900667190552,
      "learning_rate": 0.000157906714257939,
      "loss": 1.9436,
      "step": 151389
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.0548017024993896,
      "learning_rate": 0.00015790311164662395,
      "loss": 2.041,
      "step": 151390
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.2285082340240479,
      "learning_rate": 0.00015789950906172734,
      "loss": 2.1014,
      "step": 151391
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.0899122953414917,
      "learning_rate": 0.00015789590650324982,
      "loss": 2.0659,
      "step": 151392
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.2206083536148071,
      "learning_rate": 0.0001578923039711922,
      "loss": 1.8722,
      "step": 151393
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.1744972467422485,
      "learning_rate": 0.00015788870146555515,
      "loss": 2.3656,
      "step": 151394
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.180859923362732,
      "learning_rate": 0.00015788509898633914,
      "loss": 2.1843,
      "step": 151395
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.1351463794708252,
      "learning_rate": 0.00015788149653354506,
      "loss": 2.1837,
      "step": 151396
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.0809553861618042,
      "learning_rate": 0.00015787789410717353,
      "loss": 2.1873,
      "step": 151397
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.0973429679870605,
      "learning_rate": 0.000157874291707225,
      "loss": 2.0551,
      "step": 151398
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.2248111963272095,
      "learning_rate": 0.0001578706893337005,
      "loss": 2.0063,
      "step": 151399
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.1356372833251953,
      "learning_rate": 0.0001578670869866004,
      "loss": 2.1152,
      "step": 151400
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.1987311840057373,
      "learning_rate": 0.0001578634846659256,
      "loss": 1.9202,
      "step": 151401
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.1591321229934692,
      "learning_rate": 0.00015785988237167664,
      "loss": 2.0496,
      "step": 151402
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.377213478088379,
      "learning_rate": 0.00015785628010385423,
      "loss": 1.9511,
      "step": 151403
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.0889291763305664,
      "learning_rate": 0.0001578526778624589,
      "loss": 2.2037,
      "step": 151404
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.2025190591812134,
      "learning_rate": 0.00015784907564749158,
      "loss": 2.2902,
      "step": 151405
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.2371320724487305,
      "learning_rate": 0.00015784547345895267,
      "loss": 2.2083,
      "step": 151406
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.2192236185073853,
      "learning_rate": 0.00015784187129684306,
      "loss": 2.1286,
      "step": 151407
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.136940360069275,
      "learning_rate": 0.00015783826916116324,
      "loss": 2.2675,
      "step": 151408
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.077111005783081,
      "learning_rate": 0.0001578346670519142,
      "loss": 2.1092,
      "step": 151409
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.4468770027160645,
      "learning_rate": 0.00015783106496909614,
      "loss": 2.0089,
      "step": 151410
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.0851597785949707,
      "learning_rate": 0.00015782746291271007,
      "loss": 1.9491,
      "step": 151411
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.122512936592102,
      "learning_rate": 0.00015782386088275646,
      "loss": 1.7629,
      "step": 151412
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.2514723539352417,
      "learning_rate": 0.0001578202588792362,
      "loss": 2.0893,
      "step": 151413
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.1025110483169556,
      "learning_rate": 0.0001578166569021497,
      "loss": 1.9464,
      "step": 151414
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.1237741708755493,
      "learning_rate": 0.00015781305495149802,
      "loss": 2.0351,
      "step": 151415
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.1723352670669556,
      "learning_rate": 0.00015780945302728134,
      "loss": 2.1264,
      "step": 151416
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.1732077598571777,
      "learning_rate": 0.0001578058511295007,
      "loss": 2.1297,
      "step": 151417
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.1977636814117432,
      "learning_rate": 0.00015780224925815648,
      "loss": 1.8864,
      "step": 151418
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.1837241649627686,
      "learning_rate": 0.00015779864741324965,
      "loss": 1.9366,
      "step": 151419
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.0916917324066162,
      "learning_rate": 0.0001577950455947806,
      "loss": 2.126,
      "step": 151420
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.1820149421691895,
      "learning_rate": 0.00015779144380275042,
      "loss": 2.0182,
      "step": 151421
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.1939237117767334,
      "learning_rate": 0.00015778784203715918,
      "loss": 2.1909,
      "step": 151422
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.0818445682525635,
      "learning_rate": 0.00015778424029800806,
      "loss": 1.9715,
      "step": 151423
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.0675549507141113,
      "learning_rate": 0.00015778063858529738,
      "loss": 2.0013,
      "step": 151424
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.0883703231811523,
      "learning_rate": 0.00015777703689902808,
      "loss": 1.917,
      "step": 151425
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.150202989578247,
      "learning_rate": 0.00015777343523920055,
      "loss": 2.0896,
      "step": 151426
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.2219361066818237,
      "learning_rate": 0.00015776983360581595,
      "loss": 2.1769,
      "step": 151427
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.373491644859314,
      "learning_rate": 0.00015776623199887434,
      "loss": 2.0278,
      "step": 151428
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.0677043199539185,
      "learning_rate": 0.00015776263041837677,
      "loss": 1.9522,
      "step": 151429
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.2878141403198242,
      "learning_rate": 0.0001577590288643237,
      "loss": 2.0165,
      "step": 151430
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.2991671562194824,
      "learning_rate": 0.0001577554273367161,
      "loss": 2.2475,
      "step": 151431
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.216005802154541,
      "learning_rate": 0.00015775182583555427,
      "loss": 2.1311,
      "step": 151432
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.1334946155548096,
      "learning_rate": 0.0001577482243608393,
      "loss": 2.0568,
      "step": 151433
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.291043996810913,
      "learning_rate": 0.0001577446229125714,
      "loss": 2.2357,
      "step": 151434
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.7056833505630493,
      "learning_rate": 0.0001577410214907516,
      "loss": 2.2361,
      "step": 151435
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.190588355064392,
      "learning_rate": 0.00015773742009538027,
      "loss": 2.124,
      "step": 151436
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.1398029327392578,
      "learning_rate": 0.00015773381872645836,
      "loss": 2.0555,
      "step": 151437
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.1135036945343018,
      "learning_rate": 0.00015773021738398627,
      "loss": 2.1263,
      "step": 151438
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.3729039430618286,
      "learning_rate": 0.00015772661606796513,
      "loss": 2.0916,
      "step": 151439
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.2130681276321411,
      "learning_rate": 0.000157723014778395,
      "loss": 2.0377,
      "step": 151440
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.1870170831680298,
      "learning_rate": 0.00015771941351527702,
      "loss": 2.238,
      "step": 151441
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.1673762798309326,
      "learning_rate": 0.00015771581227861154,
      "loss": 2.0888,
      "step": 151442
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.1450544595718384,
      "learning_rate": 0.00015771221106839948,
      "loss": 2.2259,
      "step": 151443
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.2355456352233887,
      "learning_rate": 0.0001577086098846413,
      "loss": 2.0308,
      "step": 151444
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.1341208219528198,
      "learning_rate": 0.00015770500872733803,
      "loss": 2.4215,
      "step": 151445
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.0376088619232178,
      "learning_rate": 0.00015770140759648984,
      "loss": 2.0646,
      "step": 151446
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.1952297687530518,
      "learning_rate": 0.00015769780649209776,
      "loss": 1.9127,
      "step": 151447
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.0782564878463745,
      "learning_rate": 0.00015769420541416219,
      "loss": 2.0543,
      "step": 151448
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.1276367902755737,
      "learning_rate": 0.0001576906043626841,
      "loss": 1.9058,
      "step": 151449
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.1670994758605957,
      "learning_rate": 0.0001576870033376639,
      "loss": 1.956,
      "step": 151450
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.099980354309082,
      "learning_rate": 0.0001576834023391026,
      "loss": 1.9992,
      "step": 151451
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.1602087020874023,
      "learning_rate": 0.00015767980136700043,
      "loss": 1.9379,
      "step": 151452
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.1113406419754028,
      "learning_rate": 0.00015767620042135837,
      "loss": 2.2835,
      "step": 151453
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.1055411100387573,
      "learning_rate": 0.00015767259950217686,
      "loss": 2.2353,
      "step": 151454
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.1037123203277588,
      "learning_rate": 0.00015766899860945684,
      "loss": 2.3402,
      "step": 151455
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.0962704420089722,
      "learning_rate": 0.00015766539774319868,
      "loss": 2.005,
      "step": 151456
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.1333308219909668,
      "learning_rate": 0.00015766179690340348,
      "loss": 2.079,
      "step": 151457
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.160794734954834,
      "learning_rate": 0.00015765819609007145,
      "loss": 1.9745,
      "step": 151458
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.0842182636260986,
      "learning_rate": 0.0001576545953032035,
      "loss": 2.201,
      "step": 151459
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.0835000276565552,
      "learning_rate": 0.00015765099454280012,
      "loss": 2.0667,
      "step": 151460
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.0848031044006348,
      "learning_rate": 0.00015764739380886227,
      "loss": 2.0721,
      "step": 151461
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.1047121286392212,
      "learning_rate": 0.00015764379310139032,
      "loss": 1.9589,
      "step": 151462
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.1066112518310547,
      "learning_rate": 0.0001576401924203852,
      "loss": 2.178,
      "step": 151463
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.092107892036438,
      "learning_rate": 0.00015763659176584743,
      "loss": 2.1443,
      "step": 151464
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.1805087327957153,
      "learning_rate": 0.00015763299113777775,
      "loss": 2.1254,
      "step": 151465
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.1402915716171265,
      "learning_rate": 0.00015762939053617664,
      "loss": 2.0572,
      "step": 151466
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.0560563802719116,
      "learning_rate": 0.00015762578996104503,
      "loss": 2.1863,
      "step": 151467
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.3595386743545532,
      "learning_rate": 0.00015762218941238335,
      "loss": 2.1496,
      "step": 151468
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.0023326873779297,
      "learning_rate": 0.00015761858889019254,
      "loss": 1.773,
      "step": 151469
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.12550687789917,
      "learning_rate": 0.0001576149883944731,
      "loss": 2.4405,
      "step": 151470
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.3698550462722778,
      "learning_rate": 0.0001576113879252256,
      "loss": 2.0224,
      "step": 151471
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.08928382396698,
      "learning_rate": 0.00015760778748245095,
      "loss": 1.9906,
      "step": 151472
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.104901909828186,
      "learning_rate": 0.00015760418706614967,
      "loss": 2.0379,
      "step": 151473
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.1358184814453125,
      "learning_rate": 0.0001576005866763224,
      "loss": 2.0179,
      "step": 151474
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.4918339252471924,
      "learning_rate": 0.00015759698631296996,
      "loss": 1.9894,
      "step": 151475
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.209524154663086,
      "learning_rate": 0.00015759338597609292,
      "loss": 1.9931,
      "step": 151476
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.157953143119812,
      "learning_rate": 0.00015758978566569185,
      "loss": 2.098,
      "step": 151477
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.1206638813018799,
      "learning_rate": 0.00015758618538176767,
      "loss": 2.0664,
      "step": 151478
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.0957978963851929,
      "learning_rate": 0.0001575825851243209,
      "loss": 2.0409,
      "step": 151479
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.2701497077941895,
      "learning_rate": 0.00015757898489335208,
      "loss": 2.1393,
      "step": 151480
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.1491340398788452,
      "learning_rate": 0.0001575753846888621,
      "loss": 2.0591,
      "step": 151481
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.0956509113311768,
      "learning_rate": 0.00015757178451085159,
      "loss": 1.8986,
      "step": 151482
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.14766263961792,
      "learning_rate": 0.00015756818435932102,
      "loss": 2.0222,
      "step": 151483
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.1925231218338013,
      "learning_rate": 0.00015756458423427138,
      "loss": 2.2591,
      "step": 151484
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.055787444114685,
      "learning_rate": 0.00015756098413570303,
      "loss": 2.0491,
      "step": 151485
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.1016262769699097,
      "learning_rate": 0.0001575573840636169,
      "loss": 2.1764,
      "step": 151486
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.1641253232955933,
      "learning_rate": 0.00015755378401801354,
      "loss": 2.1319,
      "step": 151487
    },
    {
      "epoch": 1.97,
      "grad_norm": 0.9717761278152466,
      "learning_rate": 0.0001575501839988936,
      "loss": 1.87,
      "step": 151488
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.0622515678405762,
      "learning_rate": 0.00015754658400625769,
      "loss": 2.0454,
      "step": 151489
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.168851613998413,
      "learning_rate": 0.00015754298404010663,
      "loss": 2.1519,
      "step": 151490
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.145921230316162,
      "learning_rate": 0.00015753938410044093,
      "loss": 1.9613,
      "step": 151491
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.1447292566299438,
      "learning_rate": 0.00015753578418726147,
      "loss": 2.153,
      "step": 151492
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.1797842979431152,
      "learning_rate": 0.0001575321843005687,
      "loss": 2.1015,
      "step": 151493
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.2449309825897217,
      "learning_rate": 0.00015752858444036358,
      "loss": 2.1909,
      "step": 151494
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.1531091928482056,
      "learning_rate": 0.00015752498460664636,
      "loss": 2.0802,
      "step": 151495
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.0852681398391724,
      "learning_rate": 0.00015752138479941806,
      "loss": 1.9941,
      "step": 151496
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.1225208044052124,
      "learning_rate": 0.00015751778501867913,
      "loss": 1.855,
      "step": 151497
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.1590758562088013,
      "learning_rate": 0.00015751418526443038,
      "loss": 1.977,
      "step": 151498
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.2534412145614624,
      "learning_rate": 0.00015751058553667237,
      "loss": 2.3155,
      "step": 151499
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.0733164548873901,
      "learning_rate": 0.00015750698583540609,
      "loss": 2.1396,
      "step": 151500
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.2952343225479126,
      "learning_rate": 0.00015750338616063166,
      "loss": 1.9734,
      "step": 151501
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.0344135761260986,
      "learning_rate": 0.0001574997865123502,
      "loss": 2.0343,
      "step": 151502
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.0321494340896606,
      "learning_rate": 0.0001574961868905621,
      "loss": 2.1413,
      "step": 151503
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.1278504133224487,
      "learning_rate": 0.00015749258729526823,
      "loss": 2.0885,
      "step": 151504
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.1466190814971924,
      "learning_rate": 0.00015748898772646906,
      "loss": 2.0439,
      "step": 151505
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.0438976287841797,
      "learning_rate": 0.00015748538818416565,
      "loss": 1.9145,
      "step": 151506
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.0968645811080933,
      "learning_rate": 0.0001574817886683581,
      "loss": 2.1062,
      "step": 151507
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.0612901449203491,
      "learning_rate": 0.00015747818917904758,
      "loss": 2.0785,
      "step": 151508
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.1224864721298218,
      "learning_rate": 0.00015747458971623437,
      "loss": 1.77,
      "step": 151509
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.1590713262557983,
      "learning_rate": 0.0001574709902799195,
      "loss": 2.057,
      "step": 151510
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.157280445098877,
      "learning_rate": 0.0001574673908701033,
      "loss": 1.9769,
      "step": 151511
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.1446772813796997,
      "learning_rate": 0.00015746379148678683,
      "loss": 2.2761,
      "step": 151512
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.1195695400238037,
      "learning_rate": 0.00015746019212997033,
      "loss": 2.2388,
      "step": 151513
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.1696677207946777,
      "learning_rate": 0.00015745659279965476,
      "loss": 2.1799,
      "step": 151514
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.1278589963912964,
      "learning_rate": 0.00015745299349584062,
      "loss": 2.1932,
      "step": 151515
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.0558637380599976,
      "learning_rate": 0.00015744939421852874,
      "loss": 2.0401,
      "step": 151516
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.0997132062911987,
      "learning_rate": 0.00015744579496771962,
      "loss": 2.2885,
      "step": 151517
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.0158021450042725,
      "learning_rate": 0.00015744219574341422,
      "loss": 2.2721,
      "step": 151518
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.195427417755127,
      "learning_rate": 0.0001574385965456128,
      "loss": 2.181,
      "step": 151519
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.223463773727417,
      "learning_rate": 0.00015743499737431633,
      "loss": 1.8155,
      "step": 151520
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.07933509349823,
      "learning_rate": 0.0001574313982295253,
      "loss": 2.218,
      "step": 151521
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.1856564283370972,
      "learning_rate": 0.00015742779911124055,
      "loss": 2.2212,
      "step": 151522
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.2279984951019287,
      "learning_rate": 0.00015742420001946258,
      "loss": 2.1179,
      "step": 151523
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.086046576499939,
      "learning_rate": 0.00015742060095419235,
      "loss": 2.1357,
      "step": 151524
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.0881242752075195,
      "learning_rate": 0.0001574170019154301,
      "loss": 1.9454,
      "step": 151525
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.029836893081665,
      "learning_rate": 0.00015741340290317682,
      "loss": 1.9711,
      "step": 151526
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.189785122871399,
      "learning_rate": 0.00015740980391743299,
      "loss": 2.3182,
      "step": 151527
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.2245482206344604,
      "learning_rate": 0.0001574062049581994,
      "loss": 1.9199,
      "step": 151528
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.0464776754379272,
      "learning_rate": 0.0001574026060254767,
      "loss": 2.1771,
      "step": 151529
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.0680679082870483,
      "learning_rate": 0.0001573990071192657,
      "loss": 2.3188,
      "step": 151530
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.1977988481521606,
      "learning_rate": 0.00015739540823956672,
      "loss": 1.8514,
      "step": 151531
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.158195972442627,
      "learning_rate": 0.00015739180938638072,
      "loss": 2.2516,
      "step": 151532
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.1835237741470337,
      "learning_rate": 0.00015738821055970816,
      "loss": 2.1265,
      "step": 151533
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.192979335784912,
      "learning_rate": 0.00015738461175954994,
      "loss": 2.0792,
      "step": 151534
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.2034739255905151,
      "learning_rate": 0.00015738101298590654,
      "loss": 2.2013,
      "step": 151535
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.084801197052002,
      "learning_rate": 0.00015737741423877877,
      "loss": 2.2552,
      "step": 151536
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.2056167125701904,
      "learning_rate": 0.00015737381551816722,
      "loss": 2.0145,
      "step": 151537
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.1143015623092651,
      "learning_rate": 0.0001573702168240726,
      "loss": 2.0299,
      "step": 151538
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.1277084350585938,
      "learning_rate": 0.0001573666181564954,
      "loss": 1.9049,
      "step": 151539
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.2624545097351074,
      "learning_rate": 0.0001573630195154366,
      "loss": 2.0618,
      "step": 151540
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.5039840936660767,
      "learning_rate": 0.00015735942090089658,
      "loss": 1.9653,
      "step": 151541
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.0980019569396973,
      "learning_rate": 0.0001573558223128762,
      "loss": 1.9725,
      "step": 151542
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.2037794589996338,
      "learning_rate": 0.00015735222375137616,
      "loss": 2.0385,
      "step": 151543
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.6497215032577515,
      "learning_rate": 0.00015734862521639696,
      "loss": 1.9603,
      "step": 151544
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.331923007965088,
      "learning_rate": 0.00015734502670793922,
      "loss": 2.0575,
      "step": 151545
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.2151283025741577,
      "learning_rate": 0.00015734142822600392,
      "loss": 2.1351,
      "step": 151546
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.372138261795044,
      "learning_rate": 0.00015733782977059136,
      "loss": 2.0871,
      "step": 151547
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.0074348449707031,
      "learning_rate": 0.00015733423134170254,
      "loss": 1.9062,
      "step": 151548
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.1524543762207031,
      "learning_rate": 0.00015733063293933794,
      "loss": 2.0915,
      "step": 151549
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.3700528144836426,
      "learning_rate": 0.0001573270345634983,
      "loss": 2.0383,
      "step": 151550
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.0761946439743042,
      "learning_rate": 0.00015732343621418414,
      "loss": 1.8437,
      "step": 151551
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.1900748014450073,
      "learning_rate": 0.00015731983789139638,
      "loss": 2.1413,
      "step": 151552
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.5610986948013306,
      "learning_rate": 0.0001573162395951354,
      "loss": 1.9164,
      "step": 151553
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.2923457622528076,
      "learning_rate": 0.00015731264132540218,
      "loss": 1.955,
      "step": 151554
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.1309813261032104,
      "learning_rate": 0.0001573090430821972,
      "loss": 1.9002,
      "step": 151555
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.1397758722305298,
      "learning_rate": 0.00015730544486552116,
      "loss": 1.8905,
      "step": 151556
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.3784348964691162,
      "learning_rate": 0.00015730184667537463,
      "loss": 1.9541,
      "step": 151557
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.1294889450073242,
      "learning_rate": 0.0001572982485117585,
      "loss": 1.854,
      "step": 151558
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.1940370798110962,
      "learning_rate": 0.00015729465037467319,
      "loss": 1.9568,
      "step": 151559
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.0622622966766357,
      "learning_rate": 0.00015729105226411962,
      "loss": 1.9199,
      "step": 151560
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.2345837354660034,
      "learning_rate": 0.00015728745418009836,
      "loss": 2.1306,
      "step": 151561
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.1412190198898315,
      "learning_rate": 0.00015728385612260993,
      "loss": 2.1671,
      "step": 151562
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.2211235761642456,
      "learning_rate": 0.00015728025809165524,
      "loss": 2.0657,
      "step": 151563
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.1325981616973877,
      "learning_rate": 0.00015727666008723482,
      "loss": 2.3049,
      "step": 151564
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.0413228273391724,
      "learning_rate": 0.00015727306210934928,
      "loss": 1.7539,
      "step": 151565
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.1586720943450928,
      "learning_rate": 0.00015726946415799947,
      "loss": 2.0148,
      "step": 151566
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.2051043510437012,
      "learning_rate": 0.00015726586623318592,
      "loss": 2.2139,
      "step": 151567
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.1899007558822632,
      "learning_rate": 0.00015726226833490928,
      "loss": 1.9477,
      "step": 151568
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.1615475416183472,
      "learning_rate": 0.00015725867046317043,
      "loss": 1.9399,
      "step": 151569
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.1752899885177612,
      "learning_rate": 0.0001572550726179697,
      "loss": 2.0275,
      "step": 151570
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.2771786451339722,
      "learning_rate": 0.00015725147479930807,
      "loss": 1.9771,
      "step": 151571
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.2940287590026855,
      "learning_rate": 0.00015724787700718612,
      "loss": 2.035,
      "step": 151572
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.1239023208618164,
      "learning_rate": 0.00015724427924160446,
      "loss": 2.0567,
      "step": 151573
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.369031548500061,
      "learning_rate": 0.00015724068150256367,
      "loss": 2.1446,
      "step": 151574
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.2016890048980713,
      "learning_rate": 0.00015723708379006466,
      "loss": 2.3059,
      "step": 151575
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.143832802772522,
      "learning_rate": 0.00015723348610410784,
      "loss": 2.1539,
      "step": 151576
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.1801743507385254,
      "learning_rate": 0.00015722988844469414,
      "loss": 1.9284,
      "step": 151577
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.2540844678878784,
      "learning_rate": 0.00015722629081182406,
      "loss": 2.1089,
      "step": 151578
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.0547786951065063,
      "learning_rate": 0.00015722269320549834,
      "loss": 1.8318,
      "step": 151579
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.3055312633514404,
      "learning_rate": 0.00015721909562571752,
      "loss": 2.2169,
      "step": 151580
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.0998092889785767,
      "learning_rate": 0.00015721549807248245,
      "loss": 1.9862,
      "step": 151581
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.0790050029754639,
      "learning_rate": 0.0001572119005457936,
      "loss": 2.2544,
      "step": 151582
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.1837793588638306,
      "learning_rate": 0.00015720830304565188,
      "loss": 1.9983,
      "step": 151583
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.1772500276565552,
      "learning_rate": 0.00015720470557205774,
      "loss": 2.0973,
      "step": 151584
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.0781781673431396,
      "learning_rate": 0.00015720110812501213,
      "loss": 2.0345,
      "step": 151585
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.1405493021011353,
      "learning_rate": 0.0001571975107045153,
      "loss": 2.0471,
      "step": 151586
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.049085259437561,
      "learning_rate": 0.00015719391331056828,
      "loss": 2.1257,
      "step": 151587
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.1421526670455933,
      "learning_rate": 0.00015719031594317148,
      "loss": 2.0504,
      "step": 151588
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.1227418184280396,
      "learning_rate": 0.0001571867186023258,
      "loss": 2.0066,
      "step": 151589
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.0665942430496216,
      "learning_rate": 0.00015718312128803173,
      "loss": 1.968,
      "step": 151590
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.1187573671340942,
      "learning_rate": 0.0001571795240002902,
      "loss": 2.153,
      "step": 151591
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.1447398662567139,
      "learning_rate": 0.00015717592673910151,
      "loss": 1.9309,
      "step": 151592
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.17874014377594,
      "learning_rate": 0.0001571723295044666,
      "loss": 2.275,
      "step": 151593
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.2806720733642578,
      "learning_rate": 0.00015716873229638595,
      "loss": 2.1962,
      "step": 151594
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.09084951877594,
      "learning_rate": 0.00015716513511486043,
      "loss": 2.171,
      "step": 151595
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.1473124027252197,
      "learning_rate": 0.0001571615379598905,
      "loss": 2.084,
      "step": 151596
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.1559271812438965,
      "learning_rate": 0.00015715794083147718,
      "loss": 2.0411,
      "step": 151597
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.2481284141540527,
      "learning_rate": 0.00015715434372962062,
      "loss": 2.0196,
      "step": 151598
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.3747751712799072,
      "learning_rate": 0.0001571507466543219,
      "loss": 2.118,
      "step": 151599
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.6596622467041016,
      "learning_rate": 0.00015714714960558143,
      "loss": 2.1402,
      "step": 151600
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.0149439573287964,
      "learning_rate": 0.00015714355258340012,
      "loss": 1.9132,
      "step": 151601
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.1481517553329468,
      "learning_rate": 0.00015713995558777844,
      "loss": 1.9935,
      "step": 151602
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.1805051565170288,
      "learning_rate": 0.00015713635861871724,
      "loss": 2.315,
      "step": 151603
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.1748378276824951,
      "learning_rate": 0.00015713276167621704,
      "loss": 2.082,
      "step": 151604
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.0717216730117798,
      "learning_rate": 0.0001571291647602786,
      "loss": 2.1113,
      "step": 151605
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.089165449142456,
      "learning_rate": 0.00015712556787090244,
      "loss": 1.969,
      "step": 151606
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.031772494316101,
      "learning_rate": 0.00015712197100808942,
      "loss": 2.0287,
      "step": 151607
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.1735470294952393,
      "learning_rate": 0.00015711837417184003,
      "loss": 2.1917,
      "step": 151608
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.132904052734375,
      "learning_rate": 0.00015711477736215514,
      "loss": 2.1321,
      "step": 151609
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.1724053621292114,
      "learning_rate": 0.00015711118057903527,
      "loss": 2.0382,
      "step": 151610
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.164940595626831,
      "learning_rate": 0.0001571075838224812,
      "loss": 2.0061,
      "step": 151611
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.156291127204895,
      "learning_rate": 0.00015710398709249337,
      "loss": 2.1389,
      "step": 151612
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.1203176975250244,
      "learning_rate": 0.0001571003903890727,
      "loss": 2.0368,
      "step": 151613
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.2741926908493042,
      "learning_rate": 0.0001570967937122197,
      "loss": 2.2364,
      "step": 151614
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.0985851287841797,
      "learning_rate": 0.0001570931970619352,
      "loss": 2.0687,
      "step": 151615
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.0216751098632812,
      "learning_rate": 0.00015708960043821975,
      "loss": 1.9072,
      "step": 151616
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.2373952865600586,
      "learning_rate": 0.00015708600384107404,
      "loss": 2.2489,
      "step": 151617
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.11111319065094,
      "learning_rate": 0.00015708240727049867,
      "loss": 1.8711,
      "step": 151618
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.1377885341644287,
      "learning_rate": 0.00015707881072649446,
      "loss": 2.096,
      "step": 151619
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.0561823844909668,
      "learning_rate": 0.00015707521420906185,
      "loss": 2.1791,
      "step": 151620
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.1714162826538086,
      "learning_rate": 0.00015707161771820187,
      "loss": 2.0123,
      "step": 151621
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.5888880491256714,
      "learning_rate": 0.0001570680212539149,
      "loss": 1.7443,
      "step": 151622
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.1662155389785767,
      "learning_rate": 0.0001570644248162017,
      "loss": 2.114,
      "step": 151623
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.1692286729812622,
      "learning_rate": 0.0001570608284050628,
      "loss": 2.168,
      "step": 151624
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.1929272413253784,
      "learning_rate": 0.0001570572320204991,
      "loss": 2.0718,
      "step": 151625
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.3350964784622192,
      "learning_rate": 0.00015705363566251106,
      "loss": 1.9508,
      "step": 151626
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.1049349308013916,
      "learning_rate": 0.00015705003933109952,
      "loss": 1.9129,
      "step": 151627
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.167004108428955,
      "learning_rate": 0.00015704644302626513,
      "loss": 2.1528,
      "step": 151628
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.2940574884414673,
      "learning_rate": 0.00015704284674800847,
      "loss": 2.1274,
      "step": 151629
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.309664249420166,
      "learning_rate": 0.00015703925049633015,
      "loss": 2.0854,
      "step": 151630
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.2953667640686035,
      "learning_rate": 0.00015703565427123103,
      "loss": 2.1724,
      "step": 151631
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.1258615255355835,
      "learning_rate": 0.0001570320580727116,
      "loss": 2.1756,
      "step": 151632
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.1193983554840088,
      "learning_rate": 0.0001570284619007727,
      "loss": 2.2181,
      "step": 151633
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.2218101024627686,
      "learning_rate": 0.0001570248657554149,
      "loss": 1.9541,
      "step": 151634
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.126184105873108,
      "learning_rate": 0.0001570212696366389,
      "loss": 1.9879,
      "step": 151635
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.2163331508636475,
      "learning_rate": 0.0001570176735444452,
      "loss": 1.824,
      "step": 151636
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.1714330911636353,
      "learning_rate": 0.00015701407747883473,
      "loss": 2.1958,
      "step": 151637
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.1391013860702515,
      "learning_rate": 0.00015701048143980796,
      "loss": 2.2159,
      "step": 151638
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.134262204170227,
      "learning_rate": 0.00015700688542736574,
      "loss": 2.1803,
      "step": 151639
    },
    {
      "epoch": 1.97,
      "grad_norm": 0.9981777667999268,
      "learning_rate": 0.0001570032894415086,
      "loss": 2.0095,
      "step": 151640
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.1277228593826294,
      "learning_rate": 0.0001569996934822373,
      "loss": 2.1752,
      "step": 151641
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.1818180084228516,
      "learning_rate": 0.00015699609754955232,
      "loss": 2.4515,
      "step": 151642
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.0812370777130127,
      "learning_rate": 0.00015699250164345462,
      "loss": 2.082,
      "step": 151643
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.123444676399231,
      "learning_rate": 0.00015698890576394453,
      "loss": 2.1651,
      "step": 151644
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.0922974348068237,
      "learning_rate": 0.00015698530991102303,
      "loss": 2.16,
      "step": 151645
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.1523816585540771,
      "learning_rate": 0.0001569817140846907,
      "loss": 2.134,
      "step": 151646
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.1544313430786133,
      "learning_rate": 0.00015697811828494805,
      "loss": 2.2655,
      "step": 151647
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.0312883853912354,
      "learning_rate": 0.00015697452251179595,
      "loss": 2.2156,
      "step": 151648
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.0689013004302979,
      "learning_rate": 0.00015697092676523499,
      "loss": 2.1087,
      "step": 151649
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.183072566986084,
      "learning_rate": 0.00015696733104526573,
      "loss": 2.1334,
      "step": 151650
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.0982669591903687,
      "learning_rate": 0.0001569637353518891,
      "loss": 2.1979,
      "step": 151651
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.2080942392349243,
      "learning_rate": 0.00015696013968510554,
      "loss": 2.1541,
      "step": 151652
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.2083697319030762,
      "learning_rate": 0.0001569565440449157,
      "loss": 2.2766,
      "step": 151653
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.212041974067688,
      "learning_rate": 0.00015695294843132046,
      "loss": 2.0421,
      "step": 151654
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.1275603771209717,
      "learning_rate": 0.00015694935284432025,
      "loss": 1.9532,
      "step": 151655
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.6581611633300781,
      "learning_rate": 0.00015694575728391596,
      "loss": 2.0644,
      "step": 151656
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.3321572542190552,
      "learning_rate": 0.00015694216175010818,
      "loss": 2.2748,
      "step": 151657
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.122936725616455,
      "learning_rate": 0.00015693856624289754,
      "loss": 2.0188,
      "step": 151658
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.5290734767913818,
      "learning_rate": 0.00015693497076228462,
      "loss": 2.0584,
      "step": 151659
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.2737997770309448,
      "learning_rate": 0.0001569313753082703,
      "loss": 1.9915,
      "step": 151660
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.1654683351516724,
      "learning_rate": 0.00015692777988085497,
      "loss": 2.034,
      "step": 151661
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.1083078384399414,
      "learning_rate": 0.00015692418448003963,
      "loss": 1.8997,
      "step": 151662
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.0706671476364136,
      "learning_rate": 0.0001569205891058248,
      "loss": 2.2149,
      "step": 151663
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.1890910863876343,
      "learning_rate": 0.0001569169937582111,
      "loss": 2.3597,
      "step": 151664
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.1203347444534302,
      "learning_rate": 0.00015691339843719915,
      "loss": 2.2473,
      "step": 151665
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.1953487396240234,
      "learning_rate": 0.00015690980314278977,
      "loss": 1.7984,
      "step": 151666
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.0853650569915771,
      "learning_rate": 0.00015690620787498348,
      "loss": 2.225,
      "step": 151667
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.0375367403030396,
      "learning_rate": 0.00015690261263378113,
      "loss": 2.0343,
      "step": 151668
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.0804390907287598,
      "learning_rate": 0.00015689901741918318,
      "loss": 2.0622,
      "step": 151669
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.0801441669464111,
      "learning_rate": 0.00015689542223119046,
      "loss": 2.114,
      "step": 151670
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.5294792652130127,
      "learning_rate": 0.0001568918270698037,
      "loss": 2.2455,
      "step": 151671
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.0707144737243652,
      "learning_rate": 0.00015688823193502334,
      "loss": 2.1562,
      "step": 151672
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.1189665794372559,
      "learning_rate": 0.0001568846368268501,
      "loss": 2.0642,
      "step": 151673
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.0869591236114502,
      "learning_rate": 0.0001568810417452848,
      "loss": 2.0654,
      "step": 151674
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.1765302419662476,
      "learning_rate": 0.0001568774466903279,
      "loss": 1.9849,
      "step": 151675
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.2860124111175537,
      "learning_rate": 0.00015687385166198033,
      "loss": 2.2718,
      "step": 151676
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.0637381076812744,
      "learning_rate": 0.00015687025666024255,
      "loss": 1.946,
      "step": 151677
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.3372722864151,
      "learning_rate": 0.00015686666168511536,
      "loss": 2.2174,
      "step": 151678
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.1990679502487183,
      "learning_rate": 0.00015686306673659922,
      "loss": 2.1629,
      "step": 151679
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.204924464225769,
      "learning_rate": 0.00015685947181469506,
      "loss": 1.9978,
      "step": 151680
    },
    {
      "epoch": 1.97,
      "grad_norm": 1.0550875663757324,
      "learning_rate": 0.0001568558769194033,
      "loss": 1.9848,
      "step": 151681
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.2101014852523804,
      "learning_rate": 0.0001568522820507248,
      "loss": 2.0113,
      "step": 151682
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.1458231210708618,
      "learning_rate": 0.00015684868720866027,
      "loss": 2.1292,
      "step": 151683
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.2025552988052368,
      "learning_rate": 0.0001568450923932102,
      "loss": 2.2533,
      "step": 151684
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.1809653043746948,
      "learning_rate": 0.0001568414976043752,
      "loss": 2.0993,
      "step": 151685
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.3107928037643433,
      "learning_rate": 0.00015683790284215624,
      "loss": 1.9064,
      "step": 151686
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.1549206972122192,
      "learning_rate": 0.00015683430810655365,
      "loss": 1.8821,
      "step": 151687
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.0687320232391357,
      "learning_rate": 0.00015683071339756842,
      "loss": 1.8117,
      "step": 151688
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.1219196319580078,
      "learning_rate": 0.00015682711871520106,
      "loss": 2.149,
      "step": 151689
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.1790491342544556,
      "learning_rate": 0.00015682352405945222,
      "loss": 2.2094,
      "step": 151690
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.0999666452407837,
      "learning_rate": 0.00015681992943032246,
      "loss": 2.2509,
      "step": 151691
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.1589932441711426,
      "learning_rate": 0.00015681633482781274,
      "loss": 2.0654,
      "step": 151692
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.1837092638015747,
      "learning_rate": 0.00015681274025192341,
      "loss": 2.0705,
      "step": 151693
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.28005051612854,
      "learning_rate": 0.00015680914570265545,
      "loss": 1.9885,
      "step": 151694
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.1826878786087036,
      "learning_rate": 0.00015680555118000934,
      "loss": 2.1823,
      "step": 151695
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.210816740989685,
      "learning_rate": 0.0001568019566839858,
      "loss": 2.1032,
      "step": 151696
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.526877760887146,
      "learning_rate": 0.00015679836221458535,
      "loss": 2.0897,
      "step": 151697
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.3074661493301392,
      "learning_rate": 0.0001567947677718089,
      "loss": 2.0576,
      "step": 151698
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.1975153684616089,
      "learning_rate": 0.00015679117335565688,
      "loss": 2.3343,
      "step": 151699
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.234183430671692,
      "learning_rate": 0.00015678757896613022,
      "loss": 2.1406,
      "step": 151700
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.0441091060638428,
      "learning_rate": 0.00015678398460322948,
      "loss": 1.9542,
      "step": 151701
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.1203666925430298,
      "learning_rate": 0.00015678039026695525,
      "loss": 1.8802,
      "step": 151702
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.164661169052124,
      "learning_rate": 0.00015677679595730814,
      "loss": 2.14,
      "step": 151703
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.26746666431427,
      "learning_rate": 0.00015677320167428906,
      "loss": 1.9597,
      "step": 151704
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.1561379432678223,
      "learning_rate": 0.0001567696074178984,
      "loss": 1.887,
      "step": 151705
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.083677053451538,
      "learning_rate": 0.00015676601318813713,
      "loss": 2.0195,
      "step": 151706
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.4139039516448975,
      "learning_rate": 0.00015676241898500575,
      "loss": 2.1645,
      "step": 151707
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.2695623636245728,
      "learning_rate": 0.00015675882480850492,
      "loss": 1.9504,
      "step": 151708
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.10012948513031,
      "learning_rate": 0.00015675523065863527,
      "loss": 2.1336,
      "step": 151709
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.3010748624801636,
      "learning_rate": 0.00015675163653539758,
      "loss": 2.1154,
      "step": 151710
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.3101410865783691,
      "learning_rate": 0.00015674804243879237,
      "loss": 2.0655,
      "step": 151711
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.2804540395736694,
      "learning_rate": 0.00015674444836882054,
      "loss": 2.4336,
      "step": 151712
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.0848311185836792,
      "learning_rate": 0.0001567408543254826,
      "loss": 2.0474,
      "step": 151713
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.1766585111618042,
      "learning_rate": 0.00015673726030877923,
      "loss": 1.8107,
      "step": 151714
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.192870020866394,
      "learning_rate": 0.000156733666318711,
      "loss": 2.0256,
      "step": 151715
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.096157431602478,
      "learning_rate": 0.0001567300723552788,
      "loss": 1.8579,
      "step": 151716
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.1053571701049805,
      "learning_rate": 0.00015672647841848304,
      "loss": 2.2856,
      "step": 151717
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.0713262557983398,
      "learning_rate": 0.00015672288450832472,
      "loss": 2.2644,
      "step": 151718
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.1434248685836792,
      "learning_rate": 0.00015671929062480427,
      "loss": 2.1347,
      "step": 151719
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.1960442066192627,
      "learning_rate": 0.00015671569676792245,
      "loss": 2.1027,
      "step": 151720
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.2165257930755615,
      "learning_rate": 0.00015671210293767972,
      "loss": 2.1733,
      "step": 151721
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.2323120832443237,
      "learning_rate": 0.00015670850913407705,
      "loss": 2.1762,
      "step": 151722
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.175129771232605,
      "learning_rate": 0.00015670491535711484,
      "loss": 2.2129,
      "step": 151723
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.1647824048995972,
      "learning_rate": 0.00015670132160679404,
      "loss": 2.1258,
      "step": 151724
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.2488198280334473,
      "learning_rate": 0.00015669772788311513,
      "loss": 2.2502,
      "step": 151725
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.370152473449707,
      "learning_rate": 0.00015669413418607886,
      "loss": 2.1714,
      "step": 151726
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.144640326499939,
      "learning_rate": 0.00015669054051568572,
      "loss": 1.984,
      "step": 151727
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.2176518440246582,
      "learning_rate": 0.0001566869468719366,
      "loss": 2.1472,
      "step": 151728
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.2043368816375732,
      "learning_rate": 0.00015668335325483203,
      "loss": 2.084,
      "step": 151729
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.198565125465393,
      "learning_rate": 0.00015667975966437278,
      "loss": 2.1068,
      "step": 151730
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.1533652544021606,
      "learning_rate": 0.00015667616610055954,
      "loss": 2.1478,
      "step": 151731
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.1790486574172974,
      "learning_rate": 0.00015667257256339273,
      "loss": 2.0759,
      "step": 151732
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.7298475503921509,
      "learning_rate": 0.00015666897905287336,
      "loss": 1.8665,
      "step": 151733
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.2222530841827393,
      "learning_rate": 0.0001566653855690019,
      "loss": 2.2323,
      "step": 151734
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.19244384765625,
      "learning_rate": 0.0001566617921117789,
      "loss": 2.0904,
      "step": 151735
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.1828364133834839,
      "learning_rate": 0.00015665819868120533,
      "loss": 2.2456,
      "step": 151736
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.4510890245437622,
      "learning_rate": 0.0001566546052772816,
      "loss": 1.9527,
      "step": 151737
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.090525507926941,
      "learning_rate": 0.0001566510119000086,
      "loss": 2.127,
      "step": 151738
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.113497257232666,
      "learning_rate": 0.00015664741854938687,
      "loss": 2.1913,
      "step": 151739
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.1360353231430054,
      "learning_rate": 0.00015664382522541712,
      "loss": 2.1501,
      "step": 151740
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.233875036239624,
      "learning_rate": 0.00015664023192809984,
      "loss": 2.136,
      "step": 151741
    },
    {
      "epoch": 1.98,
      "grad_norm": 0.9982384443283081,
      "learning_rate": 0.00015663663865743597,
      "loss": 1.9506,
      "step": 151742
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.0715678930282593,
      "learning_rate": 0.00015663304541342594,
      "loss": 2.0632,
      "step": 151743
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.1006728410720825,
      "learning_rate": 0.00015662945219607063,
      "loss": 2.1575,
      "step": 151744
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.1742513179779053,
      "learning_rate": 0.00015662585900537066,
      "loss": 1.9375,
      "step": 151745
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.2044473886489868,
      "learning_rate": 0.0001566222658413265,
      "loss": 2.2149,
      "step": 151746
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.0101001262664795,
      "learning_rate": 0.00015661867270393907,
      "loss": 1.9596,
      "step": 151747
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.1294963359832764,
      "learning_rate": 0.00015661507959320898,
      "loss": 2.0301,
      "step": 151748
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.0628591775894165,
      "learning_rate": 0.0001566114865091367,
      "loss": 1.9658,
      "step": 151749
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.15535569190979,
      "learning_rate": 0.00015660789345172318,
      "loss": 2.0248,
      "step": 151750
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.242451548576355,
      "learning_rate": 0.00015660430042096897,
      "loss": 2.0092,
      "step": 151751
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.1482211351394653,
      "learning_rate": 0.0001566007074168746,
      "loss": 2.147,
      "step": 151752
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.0781993865966797,
      "learning_rate": 0.00015659711443944096,
      "loss": 2.1532,
      "step": 151753
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.0807353258132935,
      "learning_rate": 0.00015659352148866853,
      "loss": 2.208,
      "step": 151754
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.1322747468948364,
      "learning_rate": 0.00015658992856455817,
      "loss": 2.0475,
      "step": 151755
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.2612895965576172,
      "learning_rate": 0.00015658633566711046,
      "loss": 2.106,
      "step": 151756
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.2352008819580078,
      "learning_rate": 0.0001565827427963261,
      "loss": 2.0818,
      "step": 151757
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.2238335609436035,
      "learning_rate": 0.00015657914995220555,
      "loss": 2.1669,
      "step": 151758
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.147276759147644,
      "learning_rate": 0.00015657555713474976,
      "loss": 2.1092,
      "step": 151759
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.2645257711410522,
      "learning_rate": 0.00015657196434395917,
      "loss": 2.0095,
      "step": 151760
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.1232883930206299,
      "learning_rate": 0.0001565683715798347,
      "loss": 2.022,
      "step": 151761
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.1338505744934082,
      "learning_rate": 0.00015656477884237684,
      "loss": 1.9926,
      "step": 151762
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.1477878093719482,
      "learning_rate": 0.00015656118613158635,
      "loss": 1.9046,
      "step": 151763
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.1706454753875732,
      "learning_rate": 0.00015655759344746368,
      "loss": 2.1032,
      "step": 151764
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.0718612670898438,
      "learning_rate": 0.00015655400079000976,
      "loss": 2.0336,
      "step": 151765
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.0748178958892822,
      "learning_rate": 0.00015655040815922507,
      "loss": 2.0476,
      "step": 151766
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.1248456239700317,
      "learning_rate": 0.00015654681555511048,
      "loss": 2.0532,
      "step": 151767
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.1007180213928223,
      "learning_rate": 0.00015654322297766656,
      "loss": 2.1688,
      "step": 151768
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.1392180919647217,
      "learning_rate": 0.00015653963042689396,
      "loss": 2.273,
      "step": 151769
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.0837033987045288,
      "learning_rate": 0.00015653603790279322,
      "loss": 2.1841,
      "step": 151770
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.1594231128692627,
      "learning_rate": 0.00015653244540536526,
      "loss": 2.4189,
      "step": 151771
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.2233537435531616,
      "learning_rate": 0.00015652885293461052,
      "loss": 2.1421,
      "step": 151772
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.023817777633667,
      "learning_rate": 0.00015652526049052989,
      "loss": 2.1092,
      "step": 151773
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.1165688037872314,
      "learning_rate": 0.0001565216680731239,
      "loss": 1.8814,
      "step": 151774
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.067230224609375,
      "learning_rate": 0.00015651807568239326,
      "loss": 1.8034,
      "step": 151775
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.1701160669326782,
      "learning_rate": 0.00015651448331833852,
      "loss": 2.1998,
      "step": 151776
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.1435911655426025,
      "learning_rate": 0.0001565108909809605,
      "loss": 2.0634,
      "step": 151777
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.0926225185394287,
      "learning_rate": 0.00015650729867025973,
      "loss": 1.7762,
      "step": 151778
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.1406127214431763,
      "learning_rate": 0.0001565037063862371,
      "loss": 2.0273,
      "step": 151779
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.1856765747070312,
      "learning_rate": 0.00015650011412889314,
      "loss": 1.8996,
      "step": 151780
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.1089688539505005,
      "learning_rate": 0.00015649652189822846,
      "loss": 2.2078,
      "step": 151781
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.0931001901626587,
      "learning_rate": 0.00015649292969424373,
      "loss": 2.1732,
      "step": 151782
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.0153703689575195,
      "learning_rate": 0.0001564893375169398,
      "loss": 1.9766,
      "step": 151783
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.0405733585357666,
      "learning_rate": 0.00015648574536631705,
      "loss": 1.9129,
      "step": 151784
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.108312964439392,
      "learning_rate": 0.0001564821532423764,
      "loss": 2.0431,
      "step": 151785
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.277013897895813,
      "learning_rate": 0.0001564785611451185,
      "loss": 2.2539,
      "step": 151786
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.0442320108413696,
      "learning_rate": 0.0001564749690745439,
      "loss": 2.0393,
      "step": 151787
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.209347128868103,
      "learning_rate": 0.0001564713770306532,
      "loss": 2.3225,
      "step": 151788
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.0474404096603394,
      "learning_rate": 0.0001564677850134473,
      "loss": 1.9663,
      "step": 151789
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.111387848854065,
      "learning_rate": 0.00015646419302292667,
      "loss": 1.9389,
      "step": 151790
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.0895004272460938,
      "learning_rate": 0.0001564606010590921,
      "loss": 1.8807,
      "step": 151791
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.126534104347229,
      "learning_rate": 0.00015645700912194426,
      "loss": 2.3234,
      "step": 151792
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.1766853332519531,
      "learning_rate": 0.00015645341721148379,
      "loss": 2.099,
      "step": 151793
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.0641157627105713,
      "learning_rate": 0.00015644982532771117,
      "loss": 2.1437,
      "step": 151794
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.1768847703933716,
      "learning_rate": 0.0001564462334706274,
      "loss": 2.0958,
      "step": 151795
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.1298741102218628,
      "learning_rate": 0.00015644264164023283,
      "loss": 1.9767,
      "step": 151796
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.1026852130889893,
      "learning_rate": 0.00015643904983652843,
      "loss": 2.1136,
      "step": 151797
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.117889165878296,
      "learning_rate": 0.0001564354580595147,
      "loss": 2.0126,
      "step": 151798
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.279720664024353,
      "learning_rate": 0.00015643186630919233,
      "loss": 2.2543,
      "step": 151799
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.2797276973724365,
      "learning_rate": 0.00015642827458556188,
      "loss": 2.2379,
      "step": 151800
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.1087650060653687,
      "learning_rate": 0.00015642468288862426,
      "loss": 2.0962,
      "step": 151801
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.1386362314224243,
      "learning_rate": 0.00015642109121837987,
      "loss": 2.1299,
      "step": 151802
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.2935986518859863,
      "learning_rate": 0.00015641749957482964,
      "loss": 2.1333,
      "step": 151803
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.3679097890853882,
      "learning_rate": 0.00015641390795797394,
      "loss": 1.9957,
      "step": 151804
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.0740729570388794,
      "learning_rate": 0.00015641031636781387,
      "loss": 2.0814,
      "step": 151805
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.1037571430206299,
      "learning_rate": 0.0001564067248043496,
      "loss": 2.0057,
      "step": 151806
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.1870166063308716,
      "learning_rate": 0.00015640313326758214,
      "loss": 2.1273,
      "step": 151807
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.1706115007400513,
      "learning_rate": 0.00015639954175751192,
      "loss": 1.8975,
      "step": 151808
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.3266956806182861,
      "learning_rate": 0.0001563959502741399,
      "loss": 2.0837,
      "step": 151809
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.053093433380127,
      "learning_rate": 0.00015639235881746642,
      "loss": 2.2275,
      "step": 151810
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.1731889247894287,
      "learning_rate": 0.00015638876738749256,
      "loss": 2.2517,
      "step": 151811
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.104544758796692,
      "learning_rate": 0.0001563851759842185,
      "loss": 2.002,
      "step": 151812
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.2144287824630737,
      "learning_rate": 0.00015638158460764526,
      "loss": 1.9829,
      "step": 151813
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.0711662769317627,
      "learning_rate": 0.00015637799325777328,
      "loss": 2.2335,
      "step": 151814
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.1036261320114136,
      "learning_rate": 0.00015637440193460347,
      "loss": 2.1989,
      "step": 151815
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.0433539152145386,
      "learning_rate": 0.00015637081063813623,
      "loss": 2.1195,
      "step": 151816
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.0921423435211182,
      "learning_rate": 0.00015636721936837254,
      "loss": 2.1808,
      "step": 151817
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.1095131635665894,
      "learning_rate": 0.00015636362812531288,
      "loss": 2.0569,
      "step": 151818
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.1367923021316528,
      "learning_rate": 0.00015636003690895788,
      "loss": 2.083,
      "step": 151819
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.0892324447631836,
      "learning_rate": 0.0001563564457193082,
      "loss": 2.049,
      "step": 151820
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.1554428339004517,
      "learning_rate": 0.00015635285455636465,
      "loss": 2.1515,
      "step": 151821
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.1034986972808838,
      "learning_rate": 0.0001563492634201277,
      "loss": 1.9616,
      "step": 151822
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.108359456062317,
      "learning_rate": 0.00015634567231059823,
      "loss": 2.1297,
      "step": 151823
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.2641446590423584,
      "learning_rate": 0.00015634208122777688,
      "loss": 1.896,
      "step": 151824
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.2342277765274048,
      "learning_rate": 0.0001563384901716642,
      "loss": 2.2919,
      "step": 151825
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.2210205793380737,
      "learning_rate": 0.00015633489914226075,
      "loss": 2.0876,
      "step": 151826
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.2297496795654297,
      "learning_rate": 0.00015633130813956755,
      "loss": 1.9825,
      "step": 151827
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.0779346227645874,
      "learning_rate": 0.00015632771716358492,
      "loss": 2.0105,
      "step": 151828
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.1948273181915283,
      "learning_rate": 0.00015632412621431377,
      "loss": 1.9866,
      "step": 151829
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.1040515899658203,
      "learning_rate": 0.00015632053529175474,
      "loss": 2.2162,
      "step": 151830
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.0843851566314697,
      "learning_rate": 0.00015631694439590828,
      "loss": 2.2104,
      "step": 151831
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.0769540071487427,
      "learning_rate": 0.00015631335352677534,
      "loss": 1.886,
      "step": 151832
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.2661243677139282,
      "learning_rate": 0.00015630976268435646,
      "loss": 2.3668,
      "step": 151833
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.097057819366455,
      "learning_rate": 0.00015630617186865217,
      "loss": 2.1108,
      "step": 151834
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.1712363958358765,
      "learning_rate": 0.00015630258107966343,
      "loss": 2.1558,
      "step": 151835
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.170339584350586,
      "learning_rate": 0.0001562989903173907,
      "loss": 2.0142,
      "step": 151836
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.123020887374878,
      "learning_rate": 0.00015629539958183461,
      "loss": 2.2061,
      "step": 151837
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.1237536668777466,
      "learning_rate": 0.00015629180887299606,
      "loss": 2.0407,
      "step": 151838
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.1193147897720337,
      "learning_rate": 0.00015628821819087556,
      "loss": 2.0941,
      "step": 151839
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.1014457941055298,
      "learning_rate": 0.00015628462753547364,
      "loss": 2.1257,
      "step": 151840
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.1106574535369873,
      "learning_rate": 0.0001562810369067913,
      "loss": 1.9694,
      "step": 151841
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.0724656581878662,
      "learning_rate": 0.00015627744630482897,
      "loss": 2.1974,
      "step": 151842
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.0210446119308472,
      "learning_rate": 0.0001562738557295873,
      "loss": 1.8599,
      "step": 151843
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.122835636138916,
      "learning_rate": 0.00015627026518106712,
      "loss": 2.1647,
      "step": 151844
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.100814938545227,
      "learning_rate": 0.00015626667465926892,
      "loss": 2.1505,
      "step": 151845
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.3667678833007812,
      "learning_rate": 0.00015626308416419354,
      "loss": 2.0951,
      "step": 151846
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.1677592992782593,
      "learning_rate": 0.0001562594936958416,
      "loss": 2.2247,
      "step": 151847
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.2028719186782837,
      "learning_rate": 0.00015625590325421372,
      "loss": 2.0779,
      "step": 151848
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.1352397203445435,
      "learning_rate": 0.00015625231283931045,
      "loss": 2.1923,
      "step": 151849
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.1744487285614014,
      "learning_rate": 0.00015624872245113274,
      "loss": 2.0459,
      "step": 151850
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.0937414169311523,
      "learning_rate": 0.000156245132089681,
      "loss": 2.021,
      "step": 151851
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.0786014795303345,
      "learning_rate": 0.00015624154175495608,
      "loss": 2.0276,
      "step": 151852
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.029867172241211,
      "learning_rate": 0.00015623795144695856,
      "loss": 2.062,
      "step": 151853
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.227541446685791,
      "learning_rate": 0.00015623436116568915,
      "loss": 2.0469,
      "step": 151854
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.1395899057388306,
      "learning_rate": 0.00015623077091114835,
      "loss": 2.0783,
      "step": 151855
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.2076143026351929,
      "learning_rate": 0.0001562271806833371,
      "loss": 1.9906,
      "step": 151856
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.0999869108200073,
      "learning_rate": 0.00015622359048225582,
      "loss": 2.0973,
      "step": 151857
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.2077445983886719,
      "learning_rate": 0.00015622000030790536,
      "loss": 2.4068,
      "step": 151858
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.1008588075637817,
      "learning_rate": 0.00015621641016028633,
      "loss": 2.0299,
      "step": 151859
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.0442808866500854,
      "learning_rate": 0.00015621282003939942,
      "loss": 2.2157,
      "step": 151860
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.1097102165222168,
      "learning_rate": 0.00015620922994524514,
      "loss": 2.217,
      "step": 151861
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.1308027505874634,
      "learning_rate": 0.00015620563987782436,
      "loss": 1.8799,
      "step": 151862
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.1854664087295532,
      "learning_rate": 0.0001562020498371376,
      "loss": 2.1625,
      "step": 151863
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.156453013420105,
      "learning_rate": 0.00015619845982318566,
      "loss": 2.1898,
      "step": 151864
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.0708938837051392,
      "learning_rate": 0.00015619486983596914,
      "loss": 2.0012,
      "step": 151865
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.0216267108917236,
      "learning_rate": 0.00015619127987548877,
      "loss": 2.2214,
      "step": 151866
    },
    {
      "epoch": 1.98,
      "grad_norm": 0.9946944117546082,
      "learning_rate": 0.00015618768994174494,
      "loss": 2.0363,
      "step": 151867
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.2839385271072388,
      "learning_rate": 0.00015618410003473876,
      "loss": 2.1923,
      "step": 151868
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.248026967048645,
      "learning_rate": 0.00015618051015447047,
      "loss": 2.1004,
      "step": 151869
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.1492574214935303,
      "learning_rate": 0.00015617692030094106,
      "loss": 2.1425,
      "step": 151870
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.0135427713394165,
      "learning_rate": 0.000156173330474151,
      "loss": 2.1416,
      "step": 151871
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.1811007261276245,
      "learning_rate": 0.00015616974067410126,
      "loss": 2.1647,
      "step": 151872
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.187508225440979,
      "learning_rate": 0.00015616615090079203,
      "loss": 2.1547,
      "step": 151873
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.1693994998931885,
      "learning_rate": 0.00015616256115422435,
      "loss": 2.1912,
      "step": 151874
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.1253647804260254,
      "learning_rate": 0.00015615897143439865,
      "loss": 2.2574,
      "step": 151875
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.1653684377670288,
      "learning_rate": 0.00015615538174131583,
      "loss": 2.0816,
      "step": 151876
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.1246031522750854,
      "learning_rate": 0.00015615179207497632,
      "loss": 1.9816,
      "step": 151877
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.1387934684753418,
      "learning_rate": 0.00015614820243538114,
      "loss": 2.0114,
      "step": 151878
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.1230947971343994,
      "learning_rate": 0.0001561446128225305,
      "loss": 1.9546,
      "step": 151879
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.3197062015533447,
      "learning_rate": 0.0001561410232364254,
      "loss": 1.9397,
      "step": 151880
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.2216500043869019,
      "learning_rate": 0.00015613743367706628,
      "loss": 2.0062,
      "step": 151881
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.1593807935714722,
      "learning_rate": 0.00015613384414445408,
      "loss": 2.0008,
      "step": 151882
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.1779426336288452,
      "learning_rate": 0.00015613025463858912,
      "loss": 2.1863,
      "step": 151883
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.1582955121994019,
      "learning_rate": 0.00015612666515947256,
      "loss": 2.0215,
      "step": 151884
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.209001898765564,
      "learning_rate": 0.00015612307570710452,
      "loss": 1.9722,
      "step": 151885
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.1364524364471436,
      "learning_rate": 0.00015611948628148603,
      "loss": 1.9171,
      "step": 151886
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.1505813598632812,
      "learning_rate": 0.0001561158968826175,
      "loss": 2.1294,
      "step": 151887
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.1214600801467896,
      "learning_rate": 0.0001561123075104999,
      "loss": 2.2357,
      "step": 151888
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.1475245952606201,
      "learning_rate": 0.0001561087181651336,
      "loss": 2.0502,
      "step": 151889
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.1555231809616089,
      "learning_rate": 0.00015610512884651963,
      "loss": 1.9645,
      "step": 151890
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.1332588195800781,
      "learning_rate": 0.0001561015395546582,
      "loss": 2.0921,
      "step": 151891
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.3354709148406982,
      "learning_rate": 0.00015609795028955034,
      "loss": 1.9448,
      "step": 151892
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.173724889755249,
      "learning_rate": 0.0001560943610511965,
      "loss": 2.2922,
      "step": 151893
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.0155956745147705,
      "learning_rate": 0.00015609077183959748,
      "loss": 2.1097,
      "step": 151894
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.1654200553894043,
      "learning_rate": 0.00015608718265475386,
      "loss": 2.0301,
      "step": 151895
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.1578700542449951,
      "learning_rate": 0.00015608359349666656,
      "loss": 1.8196,
      "step": 151896
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.2064740657806396,
      "learning_rate": 0.00015608000436533577,
      "loss": 1.999,
      "step": 151897
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.1956933736801147,
      "learning_rate": 0.00015607641526076257,
      "loss": 2.0269,
      "step": 151898
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.2150806188583374,
      "learning_rate": 0.00015607282618294735,
      "loss": 1.8456,
      "step": 151899
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.2847156524658203,
      "learning_rate": 0.00015606923713189103,
      "loss": 2.3284,
      "step": 151900
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.1723023653030396,
      "learning_rate": 0.00015606564810759403,
      "loss": 2.0996,
      "step": 151901
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.0753065347671509,
      "learning_rate": 0.00015606205911005737,
      "loss": 2.1997,
      "step": 151902
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.2282761335372925,
      "learning_rate": 0.00015605847013928128,
      "loss": 1.8353,
      "step": 151903
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.2973190546035767,
      "learning_rate": 0.00015605488119526677,
      "loss": 1.9596,
      "step": 151904
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.174206256866455,
      "learning_rate": 0.00015605129227801425,
      "loss": 2.0433,
      "step": 151905
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.2886682748794556,
      "learning_rate": 0.00015604770338752458,
      "loss": 2.1646,
      "step": 151906
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.4814229011535645,
      "learning_rate": 0.00015604411452379828,
      "loss": 2.1506,
      "step": 151907
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.2692506313323975,
      "learning_rate": 0.00015604052568683624,
      "loss": 2.0307,
      "step": 151908
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.4201862812042236,
      "learning_rate": 0.00015603693687663895,
      "loss": 2.1917,
      "step": 151909
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.1764882802963257,
      "learning_rate": 0.0001560333480932071,
      "loss": 2.1838,
      "step": 151910
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.1383246183395386,
      "learning_rate": 0.00015602975933654128,
      "loss": 2.3203,
      "step": 151911
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.0604290962219238,
      "learning_rate": 0.00015602617060664234,
      "loss": 2.0907,
      "step": 151912
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.2627969980239868,
      "learning_rate": 0.00015602258190351075,
      "loss": 1.9367,
      "step": 151913
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.12042236328125,
      "learning_rate": 0.00015601899322714737,
      "loss": 2.0833,
      "step": 151914
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.1024471521377563,
      "learning_rate": 0.00015601540457755282,
      "loss": 2.0716,
      "step": 151915
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.084573745727539,
      "learning_rate": 0.0001560118159547276,
      "loss": 2.0327,
      "step": 151916
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.0444031953811646,
      "learning_rate": 0.0001560082273586726,
      "loss": 1.8525,
      "step": 151917
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.1681865453720093,
      "learning_rate": 0.00015600463878938842,
      "loss": 1.9997,
      "step": 151918
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.1720664501190186,
      "learning_rate": 0.00015600105024687552,
      "loss": 2.0606,
      "step": 151919
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.1911307573318481,
      "learning_rate": 0.0001559974617311349,
      "loss": 1.9931,
      "step": 151920
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.1297569274902344,
      "learning_rate": 0.00015599387324216705,
      "loss": 2.2795,
      "step": 151921
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.511666178703308,
      "learning_rate": 0.00015599028477997253,
      "loss": 1.9369,
      "step": 151922
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.3300447463989258,
      "learning_rate": 0.0001559866963445523,
      "loss": 2.413,
      "step": 151923
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.1842764616012573,
      "learning_rate": 0.00015598310793590682,
      "loss": 2.3206,
      "step": 151924
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.1529237031936646,
      "learning_rate": 0.00015597951955403672,
      "loss": 2.0956,
      "step": 151925
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.1552777290344238,
      "learning_rate": 0.0001559759311989428,
      "loss": 2.3512,
      "step": 151926
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.062320590019226,
      "learning_rate": 0.00015597234287062573,
      "loss": 2.0715,
      "step": 151927
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.1342449188232422,
      "learning_rate": 0.000155968754569086,
      "loss": 2.011,
      "step": 151928
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.238506555557251,
      "learning_rate": 0.00015596516629432446,
      "loss": 2.0967,
      "step": 151929
    },
    {
      "epoch": 1.98,
      "grad_norm": 0.9983130693435669,
      "learning_rate": 0.00015596157804634167,
      "loss": 2.294,
      "step": 151930
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.0815937519073486,
      "learning_rate": 0.00015595798982513844,
      "loss": 2.3647,
      "step": 151931
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.4161378145217896,
      "learning_rate": 0.00015595440163071532,
      "loss": 2.0483,
      "step": 151932
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.1223021745681763,
      "learning_rate": 0.00015595081346307297,
      "loss": 1.9725,
      "step": 151933
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.1813621520996094,
      "learning_rate": 0.000155947225322212,
      "loss": 1.9938,
      "step": 151934
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.110856533050537,
      "learning_rate": 0.00015594363720813328,
      "loss": 1.917,
      "step": 151935
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.1136080026626587,
      "learning_rate": 0.00015594004912083725,
      "loss": 1.7932,
      "step": 151936
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.248236894607544,
      "learning_rate": 0.00015593646106032476,
      "loss": 2.2171,
      "step": 151937
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.1885055303573608,
      "learning_rate": 0.00015593287302659634,
      "loss": 2.0824,
      "step": 151938
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.196168303489685,
      "learning_rate": 0.00015592928501965292,
      "loss": 2.061,
      "step": 151939
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.0874722003936768,
      "learning_rate": 0.00015592569703949474,
      "loss": 2.1665,
      "step": 151940
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.1715140342712402,
      "learning_rate": 0.00015592210908612277,
      "loss": 2.2409,
      "step": 151941
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.142581582069397,
      "learning_rate": 0.00015591852115953752,
      "loss": 2.2406,
      "step": 151942
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.0292681455612183,
      "learning_rate": 0.00015591493325973986,
      "loss": 2.128,
      "step": 151943
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.0933773517608643,
      "learning_rate": 0.00015591134538673023,
      "loss": 2.2264,
      "step": 151944
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.0628572702407837,
      "learning_rate": 0.00015590775754050957,
      "loss": 2.0277,
      "step": 151945
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.2765276432037354,
      "learning_rate": 0.00015590416972107823,
      "loss": 2.2284,
      "step": 151946
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.033349871635437,
      "learning_rate": 0.00015590058192843705,
      "loss": 2.1601,
      "step": 151947
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.014015555381775,
      "learning_rate": 0.00015589699416258663,
      "loss": 2.1007,
      "step": 151948
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.1081756353378296,
      "learning_rate": 0.00015589340642352775,
      "loss": 2.154,
      "step": 151949
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.0613617897033691,
      "learning_rate": 0.0001558898187112609,
      "loss": 2.0342,
      "step": 151950
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.1050382852554321,
      "learning_rate": 0.0001558862310257871,
      "loss": 2.039,
      "step": 151951
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.1586610078811646,
      "learning_rate": 0.00015588264336710653,
      "loss": 2.0467,
      "step": 151952
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.1517513990402222,
      "learning_rate": 0.0001558790557352202,
      "loss": 1.8596,
      "step": 151953
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.1050490140914917,
      "learning_rate": 0.00015587546813012855,
      "loss": 1.9722,
      "step": 151954
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.2719898223876953,
      "learning_rate": 0.00015587188055183254,
      "loss": 2.2825,
      "step": 151955
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.2015637159347534,
      "learning_rate": 0.00015586829300033253,
      "loss": 2.0183,
      "step": 151956
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.0849478244781494,
      "learning_rate": 0.00015586470547562954,
      "loss": 2.3425,
      "step": 151957
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.1776071786880493,
      "learning_rate": 0.0001558611179777238,
      "loss": 1.7803,
      "step": 151958
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.2391506433486938,
      "learning_rate": 0.00015585753050661628,
      "loss": 2.2463,
      "step": 151959
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.8045822381973267,
      "learning_rate": 0.0001558539430623075,
      "loss": 2.1438,
      "step": 151960
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.3049829006195068,
      "learning_rate": 0.00015585035564479827,
      "loss": 2.0478,
      "step": 151961
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.067946195602417,
      "learning_rate": 0.00015584676825408913,
      "loss": 1.9621,
      "step": 151962
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.1406123638153076,
      "learning_rate": 0.00015584318089018097,
      "loss": 2.1586,
      "step": 151963
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.4109044075012207,
      "learning_rate": 0.00015583959355307405,
      "loss": 1.9029,
      "step": 151964
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.2218990325927734,
      "learning_rate": 0.00015583600624276942,
      "loss": 1.9989,
      "step": 151965
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.3979334831237793,
      "learning_rate": 0.00015583241895926743,
      "loss": 2.1794,
      "step": 151966
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.2770721912384033,
      "learning_rate": 0.0001558288317025691,
      "loss": 1.926,
      "step": 151967
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.4622552394866943,
      "learning_rate": 0.00015582524447267475,
      "loss": 2.0627,
      "step": 151968
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.1274998188018799,
      "learning_rate": 0.00015582165726958544,
      "loss": 1.9215,
      "step": 151969
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.158445119857788,
      "learning_rate": 0.0001558180700933014,
      "loss": 1.9431,
      "step": 151970
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.2530213594436646,
      "learning_rate": 0.0001558144829438236,
      "loss": 2.0155,
      "step": 151971
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.068806767463684,
      "learning_rate": 0.00015581089582115246,
      "loss": 1.9543,
      "step": 151972
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.042126178741455,
      "learning_rate": 0.00015580730872528898,
      "loss": 2.0034,
      "step": 151973
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.2714264392852783,
      "learning_rate": 0.0001558037216562335,
      "loss": 2.2877,
      "step": 151974
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.1156240701675415,
      "learning_rate": 0.00015580013461398705,
      "loss": 1.9755,
      "step": 151975
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.139351487159729,
      "learning_rate": 0.00015579654759854985,
      "loss": 2.3312,
      "step": 151976
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.163468837738037,
      "learning_rate": 0.0001557929606099229,
      "loss": 2.1436,
      "step": 151977
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.242289423942566,
      "learning_rate": 0.00015578937364810666,
      "loss": 1.8879,
      "step": 151978
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.1468063592910767,
      "learning_rate": 0.00015578578671310205,
      "loss": 2.105,
      "step": 151979
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.1692126989364624,
      "learning_rate": 0.0001557821998049094,
      "loss": 2.0448,
      "step": 151980
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.1200509071350098,
      "learning_rate": 0.00015577861292352983,
      "loss": 2.4394,
      "step": 151981
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.038692831993103,
      "learning_rate": 0.0001557750260689635,
      "loss": 2.2201,
      "step": 151982
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.3592135906219482,
      "learning_rate": 0.00015577143924121143,
      "loss": 1.8506,
      "step": 151983
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.0598489046096802,
      "learning_rate": 0.00015576785244027404,
      "loss": 2.0624,
      "step": 151984
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.040425181388855,
      "learning_rate": 0.00015576426566615227,
      "loss": 2.1603,
      "step": 151985
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.1285794973373413,
      "learning_rate": 0.00015576067891884653,
      "loss": 1.9709,
      "step": 151986
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.1380001306533813,
      "learning_rate": 0.0001557570921983578,
      "loss": 1.934,
      "step": 151987
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.5095471143722534,
      "learning_rate": 0.00015575350550468635,
      "loss": 2.0804,
      "step": 151988
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.1304168701171875,
      "learning_rate": 0.00015574991883783316,
      "loss": 2.2265,
      "step": 151989
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.1798734664916992,
      "learning_rate": 0.00015574633219779866,
      "loss": 2.1642,
      "step": 151990
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.1329056024551392,
      "learning_rate": 0.00015574274558458376,
      "loss": 2.2349,
      "step": 151991
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.238937497138977,
      "learning_rate": 0.0001557391589981889,
      "loss": 2.1108,
      "step": 151992
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.2107504606246948,
      "learning_rate": 0.00015573557243861495,
      "loss": 2.1526,
      "step": 151993
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.111693263053894,
      "learning_rate": 0.00015573198590586247,
      "loss": 1.9873,
      "step": 151994
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.036407232284546,
      "learning_rate": 0.00015572839939993218,
      "loss": 2.0414,
      "step": 151995
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.182386875152588,
      "learning_rate": 0.00015572481292082458,
      "loss": 2.1009,
      "step": 151996
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.1244349479675293,
      "learning_rate": 0.00015572122646854056,
      "loss": 2.0212,
      "step": 151997
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.136394739151001,
      "learning_rate": 0.00015571764004308053,
      "loss": 2.0608,
      "step": 151998
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.1568001508712769,
      "learning_rate": 0.00015571405364444546,
      "loss": 2.2045,
      "step": 151999
    },
    {
      "epoch": 1.98,
      "grad_norm": 0.9895687699317932,
      "learning_rate": 0.0001557104672726359,
      "loss": 1.8816,
      "step": 152000
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.1019307374954224,
      "learning_rate": 0.0001557068809276525,
      "loss": 2.1626,
      "step": 152001
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.0906472206115723,
      "learning_rate": 0.00015570329460949574,
      "loss": 2.0417,
      "step": 152002
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.1738090515136719,
      "learning_rate": 0.0001556997083181666,
      "loss": 2.1888,
      "step": 152003
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.131919503211975,
      "learning_rate": 0.0001556961220536655,
      "loss": 1.9575,
      "step": 152004
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.1102696657180786,
      "learning_rate": 0.00015569253581599335,
      "loss": 2.0162,
      "step": 152005
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.1628295183181763,
      "learning_rate": 0.00015568894960515066,
      "loss": 1.89,
      "step": 152006
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.1880594491958618,
      "learning_rate": 0.000155685363421138,
      "loss": 2.1027,
      "step": 152007
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.1431416273117065,
      "learning_rate": 0.00015568177726395626,
      "loss": 2.0363,
      "step": 152008
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.266757845878601,
      "learning_rate": 0.00015567819113360604,
      "loss": 2.1364,
      "step": 152009
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.1122130155563354,
      "learning_rate": 0.0001556746050300878,
      "loss": 2.2895,
      "step": 152010
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.2371231317520142,
      "learning_rate": 0.00015567101895340256,
      "loss": 1.9552,
      "step": 152011
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.180238962173462,
      "learning_rate": 0.00015566743290355075,
      "loss": 2.245,
      "step": 152012
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.307481288909912,
      "learning_rate": 0.00015566384688053296,
      "loss": 2.4652,
      "step": 152013
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.1553610563278198,
      "learning_rate": 0.00015566026088435013,
      "loss": 2.1468,
      "step": 152014
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.1807587146759033,
      "learning_rate": 0.00015565667491500266,
      "loss": 2.2012,
      "step": 152015
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.1339884996414185,
      "learning_rate": 0.00015565308897249147,
      "loss": 1.9329,
      "step": 152016
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.1410837173461914,
      "learning_rate": 0.00015564950305681707,
      "loss": 2.2355,
      "step": 152017
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.2793523073196411,
      "learning_rate": 0.00015564591716798017,
      "loss": 2.0117,
      "step": 152018
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.3420698642730713,
      "learning_rate": 0.0001556423313059813,
      "loss": 2.1299,
      "step": 152019
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.1955806016921997,
      "learning_rate": 0.00015563874547082135,
      "loss": 2.0698,
      "step": 152020
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.1341131925582886,
      "learning_rate": 0.00015563515966250075,
      "loss": 1.956,
      "step": 152021
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.204378366470337,
      "learning_rate": 0.00015563157388102046,
      "loss": 2.16,
      "step": 152022
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.1279078722000122,
      "learning_rate": 0.00015562798812638096,
      "loss": 1.9529,
      "step": 152023
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.3065932989120483,
      "learning_rate": 0.00015562440239858295,
      "loss": 2.0682,
      "step": 152024
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.0824899673461914,
      "learning_rate": 0.00015562081669762696,
      "loss": 1.823,
      "step": 152025
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.263891339302063,
      "learning_rate": 0.00015561723102351394,
      "loss": 2.3203,
      "step": 152026
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.2491189241409302,
      "learning_rate": 0.00015561364537624423,
      "loss": 2.0601,
      "step": 152027
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.3015329837799072,
      "learning_rate": 0.00015561005975581882,
      "loss": 1.9634,
      "step": 152028
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.161474585533142,
      "learning_rate": 0.00015560647416223812,
      "loss": 1.9712,
      "step": 152029
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.1726278066635132,
      "learning_rate": 0.0001556028885955031,
      "loss": 2.1562,
      "step": 152030
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.1992441415786743,
      "learning_rate": 0.00015559930305561402,
      "loss": 2.1775,
      "step": 152031
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.6140888929367065,
      "learning_rate": 0.00015559571754257187,
      "loss": 1.9813,
      "step": 152032
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.186706304550171,
      "learning_rate": 0.0001555921320563771,
      "loss": 2.0827,
      "step": 152033
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.1595592498779297,
      "learning_rate": 0.00015558854659703054,
      "loss": 2.3189,
      "step": 152034
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.2170976400375366,
      "learning_rate": 0.00015558496116453274,
      "loss": 2.1211,
      "step": 152035
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.078643560409546,
      "learning_rate": 0.0001555813757588846,
      "loss": 2.1847,
      "step": 152036
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.297663688659668,
      "learning_rate": 0.0001555777903800864,
      "loss": 2.1481,
      "step": 152037
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.0241761207580566,
      "learning_rate": 0.00015557420502813915,
      "loss": 2.1603,
      "step": 152038
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.1164190769195557,
      "learning_rate": 0.00015557061970304325,
      "loss": 2.1033,
      "step": 152039
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.2090330123901367,
      "learning_rate": 0.00015556703440479963,
      "loss": 1.9549,
      "step": 152040
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.1306127309799194,
      "learning_rate": 0.0001555634491334087,
      "loss": 2.0286,
      "step": 152041
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.1675922870635986,
      "learning_rate": 0.00015555986388887148,
      "loss": 2.0214,
      "step": 152042
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.0950251817703247,
      "learning_rate": 0.00015555627867118815,
      "loss": 1.9646,
      "step": 152043
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.1887272596359253,
      "learning_rate": 0.0001555526934803598,
      "loss": 1.9848,
      "step": 152044
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.1035830974578857,
      "learning_rate": 0.00015554910831638675,
      "loss": 2.081,
      "step": 152045
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.106410026550293,
      "learning_rate": 0.00015554552317927002,
      "loss": 2.0216,
      "step": 152046
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.1924902200698853,
      "learning_rate": 0.00015554193806900998,
      "loss": 1.8401,
      "step": 152047
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.1987724304199219,
      "learning_rate": 0.00015553835298560764,
      "loss": 2.1834,
      "step": 152048
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.194094181060791,
      "learning_rate": 0.0001555347679290632,
      "loss": 1.8584,
      "step": 152049
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.0857787132263184,
      "learning_rate": 0.0001555311828993777,
      "loss": 1.9657,
      "step": 152050
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.2192329168319702,
      "learning_rate": 0.00015552759789655158,
      "loss": 2.028,
      "step": 152051
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.2672191858291626,
      "learning_rate": 0.00015552401292058573,
      "loss": 2.3004,
      "step": 152052
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.0864328145980835,
      "learning_rate": 0.00015552042797148058,
      "loss": 1.9687,
      "step": 152053
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.175663948059082,
      "learning_rate": 0.0001555168430492371,
      "loss": 2.091,
      "step": 152054
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.0633786916732788,
      "learning_rate": 0.00015551325815385555,
      "loss": 2.2272,
      "step": 152055
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.101615309715271,
      "learning_rate": 0.00015550967328533693,
      "loss": 2.1077,
      "step": 152056
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.2181203365325928,
      "learning_rate": 0.00015550608844368166,
      "loss": 1.9727,
      "step": 152057
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.1205863952636719,
      "learning_rate": 0.00015550250362889072,
      "loss": 1.8183,
      "step": 152058
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.3796387910842896,
      "learning_rate": 0.00015549891884096442,
      "loss": 2.1788,
      "step": 152059
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.2782050371170044,
      "learning_rate": 0.0001554953340799038,
      "loss": 1.9312,
      "step": 152060
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.1641045808792114,
      "learning_rate": 0.00015549174934570913,
      "loss": 2.0467,
      "step": 152061
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.0437806844711304,
      "learning_rate": 0.00015548816463838142,
      "loss": 1.9313,
      "step": 152062
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.2594791650772095,
      "learning_rate": 0.00015548457995792102,
      "loss": 2.2435,
      "step": 152063
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.1246371269226074,
      "learning_rate": 0.0001554809953043289,
      "loss": 2.0953,
      "step": 152064
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.1029912233352661,
      "learning_rate": 0.00015547741067760548,
      "loss": 2.1211,
      "step": 152065
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.1058107614517212,
      "learning_rate": 0.00015547382607775177,
      "loss": 2.1223,
      "step": 152066
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.1501952409744263,
      "learning_rate": 0.00015547024150476798,
      "loss": 2.0333,
      "step": 152067
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.1714496612548828,
      "learning_rate": 0.0001554666569586551,
      "loss": 1.8834,
      "step": 152068
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.1500532627105713,
      "learning_rate": 0.00015546307243941357,
      "loss": 2.2421,
      "step": 152069
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.1898771524429321,
      "learning_rate": 0.0001554594879470443,
      "loss": 1.8815,
      "step": 152070
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.2746070623397827,
      "learning_rate": 0.00015545590348154775,
      "loss": 1.9669,
      "step": 152071
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.1134670972824097,
      "learning_rate": 0.00015545231904292477,
      "loss": 2.208,
      "step": 152072
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.1292009353637695,
      "learning_rate": 0.00015544873463117594,
      "loss": 2.0915,
      "step": 152073
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.1230052709579468,
      "learning_rate": 0.00015544515024630196,
      "loss": 2.0068,
      "step": 152074
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.4926615953445435,
      "learning_rate": 0.0001554415658883033,
      "loss": 2.0141,
      "step": 152075
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.1868451833724976,
      "learning_rate": 0.00015543798155718088,
      "loss": 2.1014,
      "step": 152076
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.268487811088562,
      "learning_rate": 0.00015543439725293517,
      "loss": 2.0124,
      "step": 152077
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.3440823554992676,
      "learning_rate": 0.00015543081297556704,
      "loss": 2.1718,
      "step": 152078
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.1694819927215576,
      "learning_rate": 0.00015542722872507705,
      "loss": 1.9743,
      "step": 152079
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.0289738178253174,
      "learning_rate": 0.00015542364450146586,
      "loss": 1.9902,
      "step": 152080
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.1149662733078003,
      "learning_rate": 0.00015542006030473405,
      "loss": 1.996,
      "step": 152081
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.3187105655670166,
      "learning_rate": 0.00015541647613488252,
      "loss": 2.2355,
      "step": 152082
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.1153277158737183,
      "learning_rate": 0.00015541289199191163,
      "loss": 1.9921,
      "step": 152083
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.194665789604187,
      "learning_rate": 0.00015540930787582235,
      "loss": 2.1111,
      "step": 152084
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.2666432857513428,
      "learning_rate": 0.0001554057237866152,
      "loss": 2.1093,
      "step": 152085
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.1604585647583008,
      "learning_rate": 0.00015540213972429088,
      "loss": 1.9255,
      "step": 152086
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.2218753099441528,
      "learning_rate": 0.00015539855568884992,
      "loss": 2.1696,
      "step": 152087
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.1693434715270996,
      "learning_rate": 0.0001553949716802932,
      "loss": 2.0717,
      "step": 152088
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.1523892879486084,
      "learning_rate": 0.00015539138769862115,
      "loss": 2.1286,
      "step": 152089
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.1089756488800049,
      "learning_rate": 0.0001553878037438347,
      "loss": 2.1529,
      "step": 152090
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.2511667013168335,
      "learning_rate": 0.00015538421981593436,
      "loss": 2.0406,
      "step": 152091
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.1677521467208862,
      "learning_rate": 0.00015538063591492076,
      "loss": 2.0017,
      "step": 152092
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.1059871912002563,
      "learning_rate": 0.00015537705204079471,
      "loss": 2.1744,
      "step": 152093
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.1460450887680054,
      "learning_rate": 0.0001553734681935568,
      "loss": 2.0152,
      "step": 152094
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.2646522521972656,
      "learning_rate": 0.0001553698843732076,
      "loss": 2.0101,
      "step": 152095
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.2951065301895142,
      "learning_rate": 0.00015536630057974796,
      "loss": 1.9056,
      "step": 152096
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.1593929529190063,
      "learning_rate": 0.00015536271681317846,
      "loss": 2.0288,
      "step": 152097
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.123750925064087,
      "learning_rate": 0.00015535913307349964,
      "loss": 1.9997,
      "step": 152098
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.1599383354187012,
      "learning_rate": 0.00015535554936071242,
      "loss": 2.1559,
      "step": 152099
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.1816986799240112,
      "learning_rate": 0.00015535196567481718,
      "loss": 2.0406,
      "step": 152100
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.0845320224761963,
      "learning_rate": 0.0001553483820158149,
      "loss": 2.2543,
      "step": 152101
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.1439882516860962,
      "learning_rate": 0.00015534479838370607,
      "loss": 2.2525,
      "step": 152102
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.161837100982666,
      "learning_rate": 0.0001553412147784914,
      "loss": 2.3068,
      "step": 152103
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.3647925853729248,
      "learning_rate": 0.0001553376312001714,
      "loss": 2.1412,
      "step": 152104
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.2837258577346802,
      "learning_rate": 0.0001553340476487469,
      "loss": 2.265,
      "step": 152105
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.1356837749481201,
      "learning_rate": 0.00015533046412421851,
      "loss": 1.8902,
      "step": 152106
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.1381222009658813,
      "learning_rate": 0.00015532688062658702,
      "loss": 2.0267,
      "step": 152107
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.2657413482666016,
      "learning_rate": 0.00015532329715585295,
      "loss": 2.3117,
      "step": 152108
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.5726507902145386,
      "learning_rate": 0.00015531971371201707,
      "loss": 2.0972,
      "step": 152109
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.1966478824615479,
      "learning_rate": 0.00015531613029507984,
      "loss": 2.1721,
      "step": 152110
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.1688976287841797,
      "learning_rate": 0.00015531254690504217,
      "loss": 1.7595,
      "step": 152111
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.0770012140274048,
      "learning_rate": 0.00015530896354190452,
      "loss": 2.2859,
      "step": 152112
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.3267639875411987,
      "learning_rate": 0.0001553053802056678,
      "loss": 2.1494,
      "step": 152113
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.1686028242111206,
      "learning_rate": 0.0001553017968963324,
      "loss": 2.0365,
      "step": 152114
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.1218795776367188,
      "learning_rate": 0.00015529821361389939,
      "loss": 2.0209,
      "step": 152115
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.1585346460342407,
      "learning_rate": 0.00015529463035836893,
      "loss": 2.1016,
      "step": 152116
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.3631449937820435,
      "learning_rate": 0.00015529104712974206,
      "loss": 2.0766,
      "step": 152117
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.158881425857544,
      "learning_rate": 0.00015528746392801918,
      "loss": 2.1251,
      "step": 152118
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.168379545211792,
      "learning_rate": 0.0001552838807532012,
      "loss": 2.1616,
      "step": 152119
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.078126311302185,
      "learning_rate": 0.00015528029760528857,
      "loss": 2.1597,
      "step": 152120
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.241992712020874,
      "learning_rate": 0.00015527671448428232,
      "loss": 1.9351,
      "step": 152121
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.373425841331482,
      "learning_rate": 0.0001552731313901826,
      "loss": 2.1559,
      "step": 152122
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.1976048946380615,
      "learning_rate": 0.00015526954832299048,
      "loss": 2.1117,
      "step": 152123
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.1473479270935059,
      "learning_rate": 0.00015526596528270633,
      "loss": 1.9685,
      "step": 152124
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.1268831491470337,
      "learning_rate": 0.0001552623822693311,
      "loss": 1.9165,
      "step": 152125
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.23920476436615,
      "learning_rate": 0.00015525879928286523,
      "loss": 2.0325,
      "step": 152126
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.0962564945220947,
      "learning_rate": 0.00015525521632330972,
      "loss": 1.9528,
      "step": 152127
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.053018569946289,
      "learning_rate": 0.0001552516333906647,
      "loss": 2.0168,
      "step": 152128
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.2108957767486572,
      "learning_rate": 0.0001552480504849313,
      "loss": 2.3509,
      "step": 152129
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.2376794815063477,
      "learning_rate": 0.00015524446760610986,
      "loss": 1.9301,
      "step": 152130
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.084495186805725,
      "learning_rate": 0.00015524088475420136,
      "loss": 1.9495,
      "step": 152131
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.1239968538284302,
      "learning_rate": 0.00015523730192920624,
      "loss": 2.2093,
      "step": 152132
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.0631240606307983,
      "learning_rate": 0.0001552337191311254,
      "loss": 2.1428,
      "step": 152133
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.1175645589828491,
      "learning_rate": 0.00015523013635995912,
      "loss": 2.0035,
      "step": 152134
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.1670396327972412,
      "learning_rate": 0.00015522655361570846,
      "loss": 2.0259,
      "step": 152135
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.270248532295227,
      "learning_rate": 0.00015522297089837371,
      "loss": 1.9112,
      "step": 152136
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.0440691709518433,
      "learning_rate": 0.00015521938820795594,
      "loss": 2.0072,
      "step": 152137
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.2425150871276855,
      "learning_rate": 0.0001552158055444555,
      "loss": 2.1302,
      "step": 152138
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.1125876903533936,
      "learning_rate": 0.00015521222290787326,
      "loss": 1.9499,
      "step": 152139
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.2582042217254639,
      "learning_rate": 0.0001552086402982098,
      "loss": 2.0584,
      "step": 152140
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.1384485960006714,
      "learning_rate": 0.0001552050577154658,
      "loss": 2.1556,
      "step": 152141
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.0834085941314697,
      "learning_rate": 0.0001552014751596418,
      "loss": 2.0277,
      "step": 152142
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.2314085960388184,
      "learning_rate": 0.00015519789263073865,
      "loss": 2.1054,
      "step": 152143
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.390415906906128,
      "learning_rate": 0.00015519431012875684,
      "loss": 1.9871,
      "step": 152144
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.2485170364379883,
      "learning_rate": 0.00015519072765369732,
      "loss": 2.0537,
      "step": 152145
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.2603790760040283,
      "learning_rate": 0.00015518714520556054,
      "loss": 2.0668,
      "step": 152146
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.0926474332809448,
      "learning_rate": 0.00015518356278434718,
      "loss": 1.9764,
      "step": 152147
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.0970642566680908,
      "learning_rate": 0.00015517998039005785,
      "loss": 1.9953,
      "step": 152148
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.2310686111450195,
      "learning_rate": 0.0001551763980226934,
      "loss": 2.0907,
      "step": 152149
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.2517673969268799,
      "learning_rate": 0.00015517281568225427,
      "loss": 2.1715,
      "step": 152150
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.3088293075561523,
      "learning_rate": 0.00015516923336874133,
      "loss": 2.0498,
      "step": 152151
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.102001428604126,
      "learning_rate": 0.00015516565108215525,
      "loss": 2.0964,
      "step": 152152
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.2503737211227417,
      "learning_rate": 0.00015516206882249653,
      "loss": 1.9637,
      "step": 152153
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.2440086603164673,
      "learning_rate": 0.00015515848658976582,
      "loss": 2.0285,
      "step": 152154
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.141498327255249,
      "learning_rate": 0.00015515490438396402,
      "loss": 1.9165,
      "step": 152155
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.0760825872421265,
      "learning_rate": 0.0001551513222050915,
      "loss": 2.1771,
      "step": 152156
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.1509790420532227,
      "learning_rate": 0.00015514774005314922,
      "loss": 2.1655,
      "step": 152157
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.137554407119751,
      "learning_rate": 0.00015514415792813773,
      "loss": 1.8931,
      "step": 152158
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.1801728010177612,
      "learning_rate": 0.00015514057583005766,
      "loss": 2.2116,
      "step": 152159
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.0584654808044434,
      "learning_rate": 0.00015513699375890958,
      "loss": 2.0746,
      "step": 152160
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.0597201585769653,
      "learning_rate": 0.00015513341171469438,
      "loss": 2.0807,
      "step": 152161
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.11223566532135,
      "learning_rate": 0.00015512982969741252,
      "loss": 2.2049,
      "step": 152162
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.1694566011428833,
      "learning_rate": 0.00015512624770706485,
      "loss": 1.805,
      "step": 152163
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.291115641593933,
      "learning_rate": 0.00015512266574365193,
      "loss": 2.1413,
      "step": 152164
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.1057815551757812,
      "learning_rate": 0.00015511908380717447,
      "loss": 2.254,
      "step": 152165
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.2403075695037842,
      "learning_rate": 0.00015511550189763299,
      "loss": 1.8005,
      "step": 152166
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.104540228843689,
      "learning_rate": 0.00015511192001502837,
      "loss": 2.1678,
      "step": 152167
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.0977237224578857,
      "learning_rate": 0.00015510833815936105,
      "loss": 1.9766,
      "step": 152168
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.086052417755127,
      "learning_rate": 0.00015510475633063198,
      "loss": 2.0237,
      "step": 152169
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.2446165084838867,
      "learning_rate": 0.00015510117452884165,
      "loss": 2.2461,
      "step": 152170
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.0359827280044556,
      "learning_rate": 0.00015509759275399076,
      "loss": 2.0825,
      "step": 152171
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.213327407836914,
      "learning_rate": 0.00015509401100607983,
      "loss": 2.3075,
      "step": 152172
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.234149694442749,
      "learning_rate": 0.00015509042928510975,
      "loss": 2.1664,
      "step": 152173
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.2736618518829346,
      "learning_rate": 0.00015508684759108106,
      "loss": 2.2085,
      "step": 152174
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.1892449855804443,
      "learning_rate": 0.0001550832659239945,
      "loss": 2.0191,
      "step": 152175
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.1569322347640991,
      "learning_rate": 0.00015507968428385074,
      "loss": 2.0909,
      "step": 152176
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.1644080877304077,
      "learning_rate": 0.00015507610267065025,
      "loss": 2.017,
      "step": 152177
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.2944327592849731,
      "learning_rate": 0.000155072521084394,
      "loss": 2.2079,
      "step": 152178
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.3218921422958374,
      "learning_rate": 0.00015506893952508248,
      "loss": 2.0771,
      "step": 152179
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.1841678619384766,
      "learning_rate": 0.00015506535799271628,
      "loss": 2.0004,
      "step": 152180
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.193022608757019,
      "learning_rate": 0.00015506177648729627,
      "loss": 2.3541,
      "step": 152181
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.2780746221542358,
      "learning_rate": 0.000155058195008823,
      "loss": 1.9273,
      "step": 152182
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.1619839668273926,
      "learning_rate": 0.00015505461355729702,
      "loss": 2.2615,
      "step": 152183
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.2414082288742065,
      "learning_rate": 0.0001550510321327193,
      "loss": 2.2258,
      "step": 152184
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.1401777267456055,
      "learning_rate": 0.0001550474507350903,
      "loss": 2.1725,
      "step": 152185
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.1318920850753784,
      "learning_rate": 0.00015504386936441057,
      "loss": 2.0117,
      "step": 152186
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.50778329372406,
      "learning_rate": 0.00015504028802068107,
      "loss": 1.9957,
      "step": 152187
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.171721339225769,
      "learning_rate": 0.00015503670670390235,
      "loss": 2.0325,
      "step": 152188
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.0932282209396362,
      "learning_rate": 0.00015503312541407485,
      "loss": 2.1906,
      "step": 152189
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.086107850074768,
      "learning_rate": 0.0001550295441511996,
      "loss": 2.2787,
      "step": 152190
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.2886555194854736,
      "learning_rate": 0.0001550259629152769,
      "loss": 2.0187,
      "step": 152191
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.1954922676086426,
      "learning_rate": 0.0001550223817063078,
      "loss": 2.0731,
      "step": 152192
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.1779680252075195,
      "learning_rate": 0.00015501880052429278,
      "loss": 1.9808,
      "step": 152193
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.1247199773788452,
      "learning_rate": 0.0001550152193692325,
      "loss": 2.1287,
      "step": 152194
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.144336462020874,
      "learning_rate": 0.00015501163824112747,
      "loss": 2.2709,
      "step": 152195
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.1023094654083252,
      "learning_rate": 0.00015500805713997864,
      "loss": 1.9694,
      "step": 152196
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.1179039478302002,
      "learning_rate": 0.00015500447606578646,
      "loss": 1.9289,
      "step": 152197
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.116918683052063,
      "learning_rate": 0.00015500089501855178,
      "loss": 1.9998,
      "step": 152198
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.1686395406723022,
      "learning_rate": 0.00015499731399827506,
      "loss": 2.1468,
      "step": 152199
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.1340126991271973,
      "learning_rate": 0.00015499373300495732,
      "loss": 1.9603,
      "step": 152200
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.1763017177581787,
      "learning_rate": 0.00015499015203859876,
      "loss": 2.1685,
      "step": 152201
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.1351929903030396,
      "learning_rate": 0.00015498657109920034,
      "loss": 1.9025,
      "step": 152202
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.332618236541748,
      "learning_rate": 0.00015498299018676257,
      "loss": 2.1682,
      "step": 152203
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.046831727027893,
      "learning_rate": 0.0001549794093012863,
      "loss": 2.0307,
      "step": 152204
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.2101027965545654,
      "learning_rate": 0.000154975828442772,
      "loss": 2.2866,
      "step": 152205
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.2182483673095703,
      "learning_rate": 0.0001549722476112205,
      "loss": 2.0185,
      "step": 152206
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.1407712697982788,
      "learning_rate": 0.00015496866680663244,
      "loss": 2.1784,
      "step": 152207
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.0729492902755737,
      "learning_rate": 0.00015496508602900843,
      "loss": 2.0967,
      "step": 152208
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.1167560815811157,
      "learning_rate": 0.00015496150527834904,
      "loss": 2.133,
      "step": 152209
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.2153948545455933,
      "learning_rate": 0.00015495792455465517,
      "loss": 2.178,
      "step": 152210
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.068167805671692,
      "learning_rate": 0.0001549543438579272,
      "loss": 2.0343,
      "step": 152211
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.2503197193145752,
      "learning_rate": 0.0001549507631881661,
      "loss": 1.9527,
      "step": 152212
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.0856513977050781,
      "learning_rate": 0.00015494718254537238,
      "loss": 2.0814,
      "step": 152213
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.0871988534927368,
      "learning_rate": 0.00015494360192954673,
      "loss": 2.0126,
      "step": 152214
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.216992735862732,
      "learning_rate": 0.0001549400213406897,
      "loss": 1.9234,
      "step": 152215
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.3200432062149048,
      "learning_rate": 0.00015493644077880215,
      "loss": 1.9979,
      "step": 152216
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.186450481414795,
      "learning_rate": 0.00015493286024388455,
      "loss": 2.1908,
      "step": 152217
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.170630693435669,
      "learning_rate": 0.00015492927973593778,
      "loss": 2.069,
      "step": 152218
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.594951868057251,
      "learning_rate": 0.0001549256992549624,
      "loss": 2.1913,
      "step": 152219
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.2378419637680054,
      "learning_rate": 0.00015492211880095905,
      "loss": 2.1723,
      "step": 152220
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.164491057395935,
      "learning_rate": 0.00015491853837392835,
      "loss": 1.9631,
      "step": 152221
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.1127723455429077,
      "learning_rate": 0.0001549149579738711,
      "loss": 2.1103,
      "step": 152222
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.3383175134658813,
      "learning_rate": 0.00015491137760078783,
      "loss": 1.9991,
      "step": 152223
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.3273786306381226,
      "learning_rate": 0.00015490779725467932,
      "loss": 2.0821,
      "step": 152224
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.297302484512329,
      "learning_rate": 0.00015490421693554622,
      "loss": 1.9491,
      "step": 152225
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.08705735206604,
      "learning_rate": 0.00015490063664338918,
      "loss": 2.181,
      "step": 152226
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.4510606527328491,
      "learning_rate": 0.00015489705637820872,
      "loss": 2.3918,
      "step": 152227
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.3858712911605835,
      "learning_rate": 0.00015489347614000576,
      "loss": 2.0871,
      "step": 152228
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.1704775094985962,
      "learning_rate": 0.00015488989592878075,
      "loss": 2.1554,
      "step": 152229
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.1630589962005615,
      "learning_rate": 0.00015488631574453454,
      "loss": 2.02,
      "step": 152230
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.0653489828109741,
      "learning_rate": 0.0001548827355872677,
      "loss": 2.0694,
      "step": 152231
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.2886121273040771,
      "learning_rate": 0.00015487915545698088,
      "loss": 2.0967,
      "step": 152232
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.116451382637024,
      "learning_rate": 0.00015487557535367468,
      "loss": 2.3211,
      "step": 152233
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.102475643157959,
      "learning_rate": 0.00015487199527734996,
      "loss": 2.0109,
      "step": 152234
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.1688556671142578,
      "learning_rate": 0.00015486841522800715,
      "loss": 2.2103,
      "step": 152235
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.3339272737503052,
      "learning_rate": 0.00015486483520564718,
      "loss": 2.0191,
      "step": 152236
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.1378400325775146,
      "learning_rate": 0.00015486125521027057,
      "loss": 1.9803,
      "step": 152237
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.1859878301620483,
      "learning_rate": 0.00015485767524187798,
      "loss": 2.2184,
      "step": 152238
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.027843713760376,
      "learning_rate": 0.00015485409530046997,
      "loss": 2.1452,
      "step": 152239
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.1729018688201904,
      "learning_rate": 0.00015485051538604742,
      "loss": 2.1072,
      "step": 152240
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.2504831552505493,
      "learning_rate": 0.0001548469354986108,
      "loss": 2.1426,
      "step": 152241
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.0814622640609741,
      "learning_rate": 0.000154843355638161,
      "loss": 2.2223,
      "step": 152242
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.2120637893676758,
      "learning_rate": 0.00015483977580469857,
      "loss": 2.1575,
      "step": 152243
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.1257553100585938,
      "learning_rate": 0.00015483619599822415,
      "loss": 2.0982,
      "step": 152244
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.2380940914154053,
      "learning_rate": 0.0001548326162187383,
      "loss": 1.9202,
      "step": 152245
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.1229445934295654,
      "learning_rate": 0.00015482903646624194,
      "loss": 2.0482,
      "step": 152246
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.1955972909927368,
      "learning_rate": 0.00015482545674073547,
      "loss": 1.9865,
      "step": 152247
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.1184173822402954,
      "learning_rate": 0.00015482187704221986,
      "loss": 2.0846,
      "step": 152248
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.2472951412200928,
      "learning_rate": 0.0001548182973706955,
      "loss": 2.2158,
      "step": 152249
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.3058109283447266,
      "learning_rate": 0.00015481471772616325,
      "loss": 2.117,
      "step": 152250
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.2722069025039673,
      "learning_rate": 0.00015481113810862353,
      "loss": 1.9229,
      "step": 152251
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.1421116590499878,
      "learning_rate": 0.00015480755851807727,
      "loss": 2.0702,
      "step": 152252
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.1078897714614868,
      "learning_rate": 0.0001548039789545249,
      "loss": 2.086,
      "step": 152253
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.1893190145492554,
      "learning_rate": 0.00015480039941796736,
      "loss": 2.0505,
      "step": 152254
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.607973575592041,
      "learning_rate": 0.00015479681990840513,
      "loss": 1.9521,
      "step": 152255
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.1137405633926392,
      "learning_rate": 0.00015479324042583895,
      "loss": 2.0504,
      "step": 152256
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.15921151638031,
      "learning_rate": 0.0001547896609702693,
      "loss": 2.243,
      "step": 152257
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.0535030364990234,
      "learning_rate": 0.00015478608154169716,
      "loss": 1.896,
      "step": 152258
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.2965896129608154,
      "learning_rate": 0.00015478250214012287,
      "loss": 2.2464,
      "step": 152259
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.29099702835083,
      "learning_rate": 0.00015477892276554735,
      "loss": 1.7962,
      "step": 152260
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.0444543361663818,
      "learning_rate": 0.00015477534341797123,
      "loss": 1.7512,
      "step": 152261
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.0412300825119019,
      "learning_rate": 0.00015477176409739495,
      "loss": 2.0444,
      "step": 152262
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.2306123971939087,
      "learning_rate": 0.00015476818480381947,
      "loss": 1.8422,
      "step": 152263
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.4505853652954102,
      "learning_rate": 0.0001547646055372453,
      "loss": 2.3218,
      "step": 152264
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.1597884893417358,
      "learning_rate": 0.00015476102629767308,
      "loss": 2.0744,
      "step": 152265
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.2073643207550049,
      "learning_rate": 0.00015475744708510358,
      "loss": 2.2101,
      "step": 152266
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.0747474431991577,
      "learning_rate": 0.00015475386789953746,
      "loss": 1.9576,
      "step": 152267
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.2780888080596924,
      "learning_rate": 0.0001547502887409752,
      "loss": 2.1702,
      "step": 152268
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.3104573488235474,
      "learning_rate": 0.00015474670960941773,
      "loss": 2.1313,
      "step": 152269
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.1688200235366821,
      "learning_rate": 0.00015474313050486556,
      "loss": 2.0926,
      "step": 152270
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.2117419242858887,
      "learning_rate": 0.00015473955142731926,
      "loss": 2.1725,
      "step": 152271
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.050007700920105,
      "learning_rate": 0.0001547359723767798,
      "loss": 1.9526,
      "step": 152272
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.1696040630340576,
      "learning_rate": 0.00015473239335324753,
      "loss": 2.0489,
      "step": 152273
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.1442023515701294,
      "learning_rate": 0.0001547288143567233,
      "loss": 2.149,
      "step": 152274
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.125199794769287,
      "learning_rate": 0.00015472523538720778,
      "loss": 2.1909,
      "step": 152275
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.1911016702651978,
      "learning_rate": 0.00015472165644470146,
      "loss": 2.2244,
      "step": 152276
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.160614013671875,
      "learning_rate": 0.00015471807752920523,
      "loss": 2.1276,
      "step": 152277
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.2576353549957275,
      "learning_rate": 0.00015471449864071967,
      "loss": 1.9905,
      "step": 152278
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.2508970499038696,
      "learning_rate": 0.0001547109197792453,
      "loss": 2.13,
      "step": 152279
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.1935813426971436,
      "learning_rate": 0.00015470734094478306,
      "loss": 2.1587,
      "step": 152280
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.2232617139816284,
      "learning_rate": 0.0001547037621373334,
      "loss": 2.294,
      "step": 152281
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.087199091911316,
      "learning_rate": 0.00015470018335689701,
      "loss": 2.0708,
      "step": 152282
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.1511808633804321,
      "learning_rate": 0.00015469660460347466,
      "loss": 2.2065,
      "step": 152283
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.121272087097168,
      "learning_rate": 0.00015469302587706697,
      "loss": 2.1683,
      "step": 152284
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.0844229459762573,
      "learning_rate": 0.00015468944717767448,
      "loss": 2.0109,
      "step": 152285
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.2121036052703857,
      "learning_rate": 0.0001546858685052981,
      "loss": 2.3531,
      "step": 152286
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.05918288230896,
      "learning_rate": 0.00015468228985993833,
      "loss": 1.8602,
      "step": 152287
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.380375862121582,
      "learning_rate": 0.00015467871124159576,
      "loss": 2.0196,
      "step": 152288
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.1727672815322876,
      "learning_rate": 0.00015467513265027128,
      "loss": 1.8711,
      "step": 152289
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.174964189529419,
      "learning_rate": 0.0001546715540859653,
      "loss": 2.0062,
      "step": 152290
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.2018181085586548,
      "learning_rate": 0.00015466797554867875,
      "loss": 2.3083,
      "step": 152291
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.094620943069458,
      "learning_rate": 0.00015466439703841218,
      "loss": 2.2155,
      "step": 152292
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.127443552017212,
      "learning_rate": 0.00015466081855516623,
      "loss": 2.1093,
      "step": 152293
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.2788487672805786,
      "learning_rate": 0.00015465724009894145,
      "loss": 2.2499,
      "step": 152294
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.249723196029663,
      "learning_rate": 0.00015465366166973876,
      "loss": 2.2391,
      "step": 152295
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.2440418004989624,
      "learning_rate": 0.00015465008326755858,
      "loss": 2.0706,
      "step": 152296
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.116765022277832,
      "learning_rate": 0.0001546465048924018,
      "loss": 2.0371,
      "step": 152297
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.202702283859253,
      "learning_rate": 0.000154642926544269,
      "loss": 1.9719,
      "step": 152298
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.2708535194396973,
      "learning_rate": 0.00015463934822316085,
      "loss": 2.0612,
      "step": 152299
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.1111299991607666,
      "learning_rate": 0.0001546357699290778,
      "loss": 1.8858,
      "step": 152300
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.2168887853622437,
      "learning_rate": 0.00015463219166202087,
      "loss": 2.3397,
      "step": 152301
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.135034441947937,
      "learning_rate": 0.00015462861342199042,
      "loss": 2.1038,
      "step": 152302
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.439619779586792,
      "learning_rate": 0.0001546250352089874,
      "loss": 1.9207,
      "step": 152303
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.1758133172988892,
      "learning_rate": 0.0001546214570230123,
      "loss": 1.8231,
      "step": 152304
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.2044013738632202,
      "learning_rate": 0.00015461787886406585,
      "loss": 2.2448,
      "step": 152305
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.2207649946212769,
      "learning_rate": 0.00015461430073214851,
      "loss": 2.1554,
      "step": 152306
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.1673657894134521,
      "learning_rate": 0.00015461072262726132,
      "loss": 2.174,
      "step": 152307
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.1111401319503784,
      "learning_rate": 0.00015460714454940453,
      "loss": 2.0671,
      "step": 152308
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.0363430976867676,
      "learning_rate": 0.00015460356649857918,
      "loss": 2.0132,
      "step": 152309
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.0864592790603638,
      "learning_rate": 0.00015459998847478578,
      "loss": 2.0519,
      "step": 152310
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.2668511867523193,
      "learning_rate": 0.00015459641047802495,
      "loss": 2.0876,
      "step": 152311
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.1945221424102783,
      "learning_rate": 0.0001545928325082973,
      "loss": 2.0899,
      "step": 152312
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.1675422191619873,
      "learning_rate": 0.00015458925456560368,
      "loss": 2.3101,
      "step": 152313
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.1429442167282104,
      "learning_rate": 0.0001545856766499446,
      "loss": 2.1169,
      "step": 152314
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.158895492553711,
      "learning_rate": 0.00015458209876132085,
      "loss": 2.2389,
      "step": 152315
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.1428166627883911,
      "learning_rate": 0.00015457852089973304,
      "loss": 2.2466,
      "step": 152316
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.1943531036376953,
      "learning_rate": 0.00015457494306518184,
      "loss": 2.1346,
      "step": 152317
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.3782386779785156,
      "learning_rate": 0.0001545713652576678,
      "loss": 2.1894,
      "step": 152318
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.1990071535110474,
      "learning_rate": 0.00015456778747719178,
      "loss": 2.0166,
      "step": 152319
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.1864184141159058,
      "learning_rate": 0.00015456420972375428,
      "loss": 2.2417,
      "step": 152320
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.0332773923873901,
      "learning_rate": 0.0001545606319973561,
      "loss": 1.8996,
      "step": 152321
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.5503591299057007,
      "learning_rate": 0.00015455705429799787,
      "loss": 2.0255,
      "step": 152322
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.2294070720672607,
      "learning_rate": 0.00015455347662568023,
      "loss": 2.1559,
      "step": 152323
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.2271441221237183,
      "learning_rate": 0.00015454989898040374,
      "loss": 2.1444,
      "step": 152324
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.171378493309021,
      "learning_rate": 0.00015454632136216928,
      "loss": 2.0099,
      "step": 152325
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.2180365324020386,
      "learning_rate": 0.00015454274377097723,
      "loss": 2.0905,
      "step": 152326
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.1874279975891113,
      "learning_rate": 0.0001545391662068286,
      "loss": 2.0328,
      "step": 152327
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.1075025796890259,
      "learning_rate": 0.00015453558866972388,
      "loss": 2.0293,
      "step": 152328
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.1428465843200684,
      "learning_rate": 0.00015453201115966377,
      "loss": 1.9531,
      "step": 152329
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.1339540481567383,
      "learning_rate": 0.00015452843367664873,
      "loss": 2.0921,
      "step": 152330
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.1704703569412231,
      "learning_rate": 0.00015452485622067974,
      "loss": 1.9778,
      "step": 152331
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.3060978651046753,
      "learning_rate": 0.00015452127879175725,
      "loss": 2.0026,
      "step": 152332
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.1829723119735718,
      "learning_rate": 0.00015451770138988203,
      "loss": 1.9444,
      "step": 152333
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.2501071691513062,
      "learning_rate": 0.0001545141240150548,
      "loss": 2.0706,
      "step": 152334
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.1271673440933228,
      "learning_rate": 0.0001545105466672761,
      "loss": 1.8789,
      "step": 152335
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.152640700340271,
      "learning_rate": 0.00015450696934654653,
      "loss": 2.0736,
      "step": 152336
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.1606481075286865,
      "learning_rate": 0.00015450339205286699,
      "loss": 2.1595,
      "step": 152337
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.1123853921890259,
      "learning_rate": 0.00015449981478623785,
      "loss": 1.8019,
      "step": 152338
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.3332990407943726,
      "learning_rate": 0.0001544962375466601,
      "loss": 1.9739,
      "step": 152339
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.2109626531600952,
      "learning_rate": 0.00015449266033413413,
      "loss": 2.2269,
      "step": 152340
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.182235836982727,
      "learning_rate": 0.00015448908314866095,
      "loss": 2.0093,
      "step": 152341
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.1075866222381592,
      "learning_rate": 0.00015448550599024078,
      "loss": 2.2961,
      "step": 152342
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.1085469722747803,
      "learning_rate": 0.00015448192885887457,
      "loss": 2.0714,
      "step": 152343
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.1821094751358032,
      "learning_rate": 0.00015447835175456284,
      "loss": 2.3117,
      "step": 152344
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.1110647916793823,
      "learning_rate": 0.00015447477467730642,
      "loss": 2.095,
      "step": 152345
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.2046300172805786,
      "learning_rate": 0.00015447119762710578,
      "loss": 2.0988,
      "step": 152346
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.103846788406372,
      "learning_rate": 0.0001544676206039619,
      "loss": 2.1599,
      "step": 152347
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.156056523323059,
      "learning_rate": 0.00015446404360787503,
      "loss": 1.9784,
      "step": 152348
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.1443166732788086,
      "learning_rate": 0.0001544604666388462,
      "loss": 2.0367,
      "step": 152349
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.166542887687683,
      "learning_rate": 0.00015445688969687572,
      "loss": 2.158,
      "step": 152350
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.1639679670333862,
      "learning_rate": 0.00015445331278196461,
      "loss": 2.3877,
      "step": 152351
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.13325035572052,
      "learning_rate": 0.00015444973589411328,
      "loss": 2.1529,
      "step": 152352
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.2140570878982544,
      "learning_rate": 0.00015444615903332257,
      "loss": 2.0285,
      "step": 152353
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.1367249488830566,
      "learning_rate": 0.00015444258219959308,
      "loss": 2.1724,
      "step": 152354
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.2198675870895386,
      "learning_rate": 0.00015443900539292548,
      "loss": 2.0002,
      "step": 152355
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.1190541982650757,
      "learning_rate": 0.00015443542861332026,
      "loss": 2.0574,
      "step": 152356
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.1043050289154053,
      "learning_rate": 0.00015443185186077838,
      "loss": 2.0356,
      "step": 152357
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.128194808959961,
      "learning_rate": 0.00015442827513530022,
      "loss": 2.2191,
      "step": 152358
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.0917664766311646,
      "learning_rate": 0.0001544246984368868,
      "loss": 2.1344,
      "step": 152359
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.2010750770568848,
      "learning_rate": 0.00015442112176553848,
      "loss": 2.226,
      "step": 152360
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.308369517326355,
      "learning_rate": 0.00015441754512125595,
      "loss": 2.0509,
      "step": 152361
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.185155987739563,
      "learning_rate": 0.00015441396850404004,
      "loss": 1.9669,
      "step": 152362
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.2197362184524536,
      "learning_rate": 0.00015441039191389135,
      "loss": 1.9784,
      "step": 152363
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.2808884382247925,
      "learning_rate": 0.00015440681535081039,
      "loss": 2.2495,
      "step": 152364
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.1826871633529663,
      "learning_rate": 0.00015440323881479805,
      "loss": 1.7663,
      "step": 152365
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.0716077089309692,
      "learning_rate": 0.00015439966230585493,
      "loss": 2.0793,
      "step": 152366
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.1516398191452026,
      "learning_rate": 0.0001543960858239815,
      "loss": 2.1336,
      "step": 152367
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.2170296907424927,
      "learning_rate": 0.00015439250936917872,
      "loss": 2.2605,
      "step": 152368
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.2625865936279297,
      "learning_rate": 0.00015438893294144712,
      "loss": 2.0616,
      "step": 152369
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.1741094589233398,
      "learning_rate": 0.00015438535654078726,
      "loss": 2.1094,
      "step": 152370
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.1700481176376343,
      "learning_rate": 0.00015438178016720003,
      "loss": 2.0521,
      "step": 152371
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.2162535190582275,
      "learning_rate": 0.00015437820382068596,
      "loss": 2.0953,
      "step": 152372
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.146140217781067,
      "learning_rate": 0.00015437462750124564,
      "loss": 2.1041,
      "step": 152373
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.2778170108795166,
      "learning_rate": 0.00015437105120887993,
      "loss": 2.0717,
      "step": 152374
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.0495061874389648,
      "learning_rate": 0.00015436747494358927,
      "loss": 2.0221,
      "step": 152375
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.1204180717468262,
      "learning_rate": 0.00015436389870537455,
      "loss": 1.9344,
      "step": 152376
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.21564781665802,
      "learning_rate": 0.00015436032249423635,
      "loss": 2.031,
      "step": 152377
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.290149211883545,
      "learning_rate": 0.0001543567463101753,
      "loss": 1.9833,
      "step": 152378
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.4181088209152222,
      "learning_rate": 0.00015435317015319197,
      "loss": 2.0803,
      "step": 152379
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.297560453414917,
      "learning_rate": 0.00015434959402328726,
      "loss": 1.9767,
      "step": 152380
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.2458679676055908,
      "learning_rate": 0.00015434601792046159,
      "loss": 2.148,
      "step": 152381
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.1134886741638184,
      "learning_rate": 0.00015434244184471587,
      "loss": 1.9154,
      "step": 152382
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.1342766284942627,
      "learning_rate": 0.00015433886579605061,
      "loss": 1.9565,
      "step": 152383
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.1428673267364502,
      "learning_rate": 0.00015433528977446649,
      "loss": 2.152,
      "step": 152384
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.0928394794464111,
      "learning_rate": 0.00015433171377996413,
      "loss": 2.0494,
      "step": 152385
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.073696494102478,
      "learning_rate": 0.00015432813781254435,
      "loss": 2.1792,
      "step": 152386
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.1119064092636108,
      "learning_rate": 0.00015432456187220757,
      "loss": 2.0828,
      "step": 152387
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.2899320125579834,
      "learning_rate": 0.00015432098595895475,
      "loss": 2.1503,
      "step": 152388
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.1140727996826172,
      "learning_rate": 0.00015431741007278642,
      "loss": 2.0944,
      "step": 152389
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.1723321676254272,
      "learning_rate": 0.0001543138342137032,
      "loss": 2.1157,
      "step": 152390
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.12532639503479,
      "learning_rate": 0.0001543102583817057,
      "loss": 2.1998,
      "step": 152391
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.1967543363571167,
      "learning_rate": 0.00015430668257679478,
      "loss": 2.1302,
      "step": 152392
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.074665904045105,
      "learning_rate": 0.00015430310679897087,
      "loss": 2.2136,
      "step": 152393
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.063912034034729,
      "learning_rate": 0.0001542995310482349,
      "loss": 1.9479,
      "step": 152394
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.1745136976242065,
      "learning_rate": 0.00015429595532458738,
      "loss": 2.1174,
      "step": 152395
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.0837818384170532,
      "learning_rate": 0.000154292379628029,
      "loss": 2.1687,
      "step": 152396
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.2012388706207275,
      "learning_rate": 0.00015428880395856033,
      "loss": 2.0397,
      "step": 152397
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.0994938611984253,
      "learning_rate": 0.0001542852283161822,
      "loss": 1.8862,
      "step": 152398
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.4050602912902832,
      "learning_rate": 0.00015428165270089508,
      "loss": 2.0598,
      "step": 152399
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.1818703413009644,
      "learning_rate": 0.0001542780771126999,
      "loss": 2.088,
      "step": 152400
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.0986219644546509,
      "learning_rate": 0.00015427450155159703,
      "loss": 2.0376,
      "step": 152401
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.2611658573150635,
      "learning_rate": 0.00015427092601758753,
      "loss": 2.236,
      "step": 152402
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.3968626260757446,
      "learning_rate": 0.00015426735051067157,
      "loss": 2.0195,
      "step": 152403
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.1395694017410278,
      "learning_rate": 0.00015426377503085018,
      "loss": 1.9693,
      "step": 152404
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.0525139570236206,
      "learning_rate": 0.00015426019957812378,
      "loss": 1.9098,
      "step": 152405
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.1831741333007812,
      "learning_rate": 0.00015425662415249327,
      "loss": 2.0855,
      "step": 152406
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.4177358150482178,
      "learning_rate": 0.00015425304875395915,
      "loss": 2.1533,
      "step": 152407
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.52381432056427,
      "learning_rate": 0.0001542494733825223,
      "loss": 1.9592,
      "step": 152408
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.195473551750183,
      "learning_rate": 0.00015424589803818303,
      "loss": 1.8746,
      "step": 152409
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.1580249071121216,
      "learning_rate": 0.0001542423227209423,
      "loss": 2.1906,
      "step": 152410
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.2468334436416626,
      "learning_rate": 0.00015423874743080056,
      "loss": 2.266,
      "step": 152411
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.5612865686416626,
      "learning_rate": 0.00015423517216775866,
      "loss": 2.0975,
      "step": 152412
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.1917883157730103,
      "learning_rate": 0.00015423159693181716,
      "loss": 2.2478,
      "step": 152413
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.08646559715271,
      "learning_rate": 0.00015422802172297692,
      "loss": 2.0215,
      "step": 152414
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.0677956342697144,
      "learning_rate": 0.00015422444654123826,
      "loss": 1.8741,
      "step": 152415
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.1276698112487793,
      "learning_rate": 0.00015422087138660214,
      "loss": 1.8985,
      "step": 152416
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.2250099182128906,
      "learning_rate": 0.00015421729625906898,
      "loss": 2.0877,
      "step": 152417
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.1159706115722656,
      "learning_rate": 0.0001542137211586397,
      "loss": 1.9772,
      "step": 152418
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.1362347602844238,
      "learning_rate": 0.0001542101460853147,
      "loss": 2.2441,
      "step": 152419
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.0522366762161255,
      "learning_rate": 0.00015420657103909505,
      "loss": 2.0367,
      "step": 152420
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.161965250968933,
      "learning_rate": 0.0001542029960199809,
      "loss": 1.893,
      "step": 152421
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.0991337299346924,
      "learning_rate": 0.00015419942102797327,
      "loss": 2.0066,
      "step": 152422
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.0951817035675049,
      "learning_rate": 0.00015419584606307264,
      "loss": 2.0103,
      "step": 152423
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.039534330368042,
      "learning_rate": 0.00015419227112527982,
      "loss": 2.21,
      "step": 152424
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.1244778633117676,
      "learning_rate": 0.00015418869621459534,
      "loss": 2.1683,
      "step": 152425
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.2210005521774292,
      "learning_rate": 0.00015418512133102014,
      "loss": 2.3686,
      "step": 152426
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.0700620412826538,
      "learning_rate": 0.00015418154647455447,
      "loss": 2.191,
      "step": 152427
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.162508487701416,
      "learning_rate": 0.0001541779716451993,
      "loss": 2.0775,
      "step": 152428
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.2438799142837524,
      "learning_rate": 0.00015417439684295512,
      "loss": 2.101,
      "step": 152429
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.2064052820205688,
      "learning_rate": 0.00015417082206782275,
      "loss": 1.9634,
      "step": 152430
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.1162327527999878,
      "learning_rate": 0.00015416724731980263,
      "loss": 2.0089,
      "step": 152431
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.182379126548767,
      "learning_rate": 0.00015416367259889587,
      "loss": 2.0039,
      "step": 152432
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.1082364320755005,
      "learning_rate": 0.00015416009790510256,
      "loss": 2.0403,
      "step": 152433
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.167129635810852,
      "learning_rate": 0.00015415652323842375,
      "loss": 2.1484,
      "step": 152434
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.1872235536575317,
      "learning_rate": 0.00015415294859885993,
      "loss": 2.2232,
      "step": 152435
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.1497975587844849,
      "learning_rate": 0.0001541493739864119,
      "loss": 2.1288,
      "step": 152436
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.2423779964447021,
      "learning_rate": 0.00015414579940108017,
      "loss": 1.9338,
      "step": 152437
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.1426078081130981,
      "learning_rate": 0.00015414222484286556,
      "loss": 2.0722,
      "step": 152438
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.2265301942825317,
      "learning_rate": 0.0001541386503117687,
      "loss": 2.0725,
      "step": 152439
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.2408186197280884,
      "learning_rate": 0.0001541350758077902,
      "loss": 1.9091,
      "step": 152440
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.1414763927459717,
      "learning_rate": 0.00015413150133093063,
      "loss": 1.9397,
      "step": 152441
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.0825433731079102,
      "learning_rate": 0.0001541279268811909,
      "loss": 2.1754,
      "step": 152442
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.1196576356887817,
      "learning_rate": 0.00015412435245857142,
      "loss": 2.089,
      "step": 152443
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.4141664505004883,
      "learning_rate": 0.0001541207780630731,
      "loss": 2.0289,
      "step": 152444
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.107646107673645,
      "learning_rate": 0.0001541172036946965,
      "loss": 2.0872,
      "step": 152445
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.2317873239517212,
      "learning_rate": 0.0001541136293534422,
      "loss": 2.2199,
      "step": 152446
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.1554185152053833,
      "learning_rate": 0.00015411005503931084,
      "loss": 2.2901,
      "step": 152447
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.0665056705474854,
      "learning_rate": 0.00015410648075230333,
      "loss": 1.78,
      "step": 152448
    },
    {
      "epoch": 1.98,
      "grad_norm": 1.2024165391921997,
      "learning_rate": 0.00015410290649242,
      "loss": 2.3188,
      "step": 152449
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.263239860534668,
      "learning_rate": 0.00015409933225966188,
      "loss": 2.0632,
      "step": 152450
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.277282476425171,
      "learning_rate": 0.0001540957580540294,
      "loss": 2.1265,
      "step": 152451
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.3085663318634033,
      "learning_rate": 0.00015409218387552317,
      "loss": 2.1123,
      "step": 152452
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.0988314151763916,
      "learning_rate": 0.0001540886097241441,
      "loss": 2.4378,
      "step": 152453
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.0434489250183105,
      "learning_rate": 0.00015408503559989268,
      "loss": 2.1125,
      "step": 152454
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.4072272777557373,
      "learning_rate": 0.0001540814615027695,
      "loss": 2.0619,
      "step": 152455
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.057300329208374,
      "learning_rate": 0.00015407788743277543,
      "loss": 2.1587,
      "step": 152456
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.1827809810638428,
      "learning_rate": 0.00015407431338991103,
      "loss": 1.9258,
      "step": 152457
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.0652446746826172,
      "learning_rate": 0.0001540707393741769,
      "loss": 1.9966,
      "step": 152458
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.1294279098510742,
      "learning_rate": 0.0001540671653855738,
      "loss": 2.0056,
      "step": 152459
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.0674824714660645,
      "learning_rate": 0.0001540635914241023,
      "loss": 2.0213,
      "step": 152460
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.0353000164031982,
      "learning_rate": 0.0001540600174897633,
      "loss": 2.2325,
      "step": 152461
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.2176439762115479,
      "learning_rate": 0.00015405644358255726,
      "loss": 2.1544,
      "step": 152462
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.1392205953598022,
      "learning_rate": 0.00015405286970248487,
      "loss": 1.9119,
      "step": 152463
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.025108814239502,
      "learning_rate": 0.00015404929584954673,
      "loss": 2.0614,
      "step": 152464
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.2794691324234009,
      "learning_rate": 0.00015404572202374364,
      "loss": 2.0804,
      "step": 152465
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.4007889032363892,
      "learning_rate": 0.00015404214822507613,
      "loss": 2.2356,
      "step": 152466
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.1395916938781738,
      "learning_rate": 0.00015403857445354507,
      "loss": 2.2316,
      "step": 152467
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.165667176246643,
      "learning_rate": 0.00015403500070915096,
      "loss": 2.0351,
      "step": 152468
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.2671074867248535,
      "learning_rate": 0.00015403142699189448,
      "loss": 1.9636,
      "step": 152469
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.320034146308899,
      "learning_rate": 0.00015402785330177624,
      "loss": 2.0625,
      "step": 152470
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.0873503684997559,
      "learning_rate": 0.0001540242796387971,
      "loss": 2.2127,
      "step": 152471
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.208809494972229,
      "learning_rate": 0.00015402070600295745,
      "loss": 2.1441,
      "step": 152472
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.2100425958633423,
      "learning_rate": 0.0001540171323942582,
      "loss": 2.223,
      "step": 152473
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.1609776020050049,
      "learning_rate": 0.00015401355881269988,
      "loss": 2.0317,
      "step": 152474
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.1662224531173706,
      "learning_rate": 0.00015400998525828336,
      "loss": 2.2489,
      "step": 152475
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.1617575883865356,
      "learning_rate": 0.00015400641173100895,
      "loss": 2.1221,
      "step": 152476
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.172188639640808,
      "learning_rate": 0.0001540028382308776,
      "loss": 2.1881,
      "step": 152477
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.1453468799591064,
      "learning_rate": 0.00015399926475788975,
      "loss": 2.1056,
      "step": 152478
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.2928296327590942,
      "learning_rate": 0.00015399569131204632,
      "loss": 1.8745,
      "step": 152479
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.1248447895050049,
      "learning_rate": 0.00015399211789334777,
      "loss": 1.9159,
      "step": 152480
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.240902304649353,
      "learning_rate": 0.00015398854450179503,
      "loss": 1.8254,
      "step": 152481
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.1667464971542358,
      "learning_rate": 0.00015398497113738836,
      "loss": 2.1108,
      "step": 152482
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.173566222190857,
      "learning_rate": 0.00015398139780012872,
      "loss": 2.3015,
      "step": 152483
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.0777751207351685,
      "learning_rate": 0.00015397782449001662,
      "loss": 1.9104,
      "step": 152484
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.2157257795333862,
      "learning_rate": 0.00015397425120705293,
      "loss": 2.2293,
      "step": 152485
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.2629306316375732,
      "learning_rate": 0.00015397067795123803,
      "loss": 2.0844,
      "step": 152486
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.1548686027526855,
      "learning_rate": 0.00015396710472257298,
      "loss": 1.944,
      "step": 152487
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.1237868070602417,
      "learning_rate": 0.00015396353152105798,
      "loss": 1.9699,
      "step": 152488
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.090442180633545,
      "learning_rate": 0.000153959958346694,
      "loss": 2.0275,
      "step": 152489
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.1709531545639038,
      "learning_rate": 0.00015395638519948154,
      "loss": 1.9831,
      "step": 152490
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.04116952419281,
      "learning_rate": 0.00015395281207942143,
      "loss": 1.9733,
      "step": 152491
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.1208451986312866,
      "learning_rate": 0.00015394923898651416,
      "loss": 2.1315,
      "step": 152492
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.0453873872756958,
      "learning_rate": 0.00015394566592076073,
      "loss": 1.9903,
      "step": 152493
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.1181230545043945,
      "learning_rate": 0.00015394209288216126,
      "loss": 2.0427,
      "step": 152494
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.2845747470855713,
      "learning_rate": 0.00015393851987071687,
      "loss": 2.1457,
      "step": 152495
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.2733043432235718,
      "learning_rate": 0.00015393494688642798,
      "loss": 1.9369,
      "step": 152496
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.140230655670166,
      "learning_rate": 0.00015393137392929543,
      "loss": 1.9929,
      "step": 152497
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.1776686906814575,
      "learning_rate": 0.00015392780099931968,
      "loss": 1.91,
      "step": 152498
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.1763795614242554,
      "learning_rate": 0.00015392422809650173,
      "loss": 1.9179,
      "step": 152499
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.2364763021469116,
      "learning_rate": 0.00015392065522084177,
      "loss": 2.226,
      "step": 152500
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.116943120956421,
      "learning_rate": 0.00015391708237234088,
      "loss": 1.9197,
      "step": 152501
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.1804884672164917,
      "learning_rate": 0.00015391350955099945,
      "loss": 1.9976,
      "step": 152502
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.0290125608444214,
      "learning_rate": 0.00015390993675681833,
      "loss": 2.145,
      "step": 152503
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.116132140159607,
      "learning_rate": 0.000153906363989798,
      "loss": 2.0105,
      "step": 152504
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.1248492002487183,
      "learning_rate": 0.00015390279124993946,
      "loss": 2.1905,
      "step": 152505
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.272510290145874,
      "learning_rate": 0.00015389921853724295,
      "loss": 2.0277,
      "step": 152506
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.1971346139907837,
      "learning_rate": 0.00015389564585170944,
      "loss": 1.9575,
      "step": 152507
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.1932605504989624,
      "learning_rate": 0.00015389207319333937,
      "loss": 1.8157,
      "step": 152508
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.1613140106201172,
      "learning_rate": 0.00015388850056213363,
      "loss": 2.0342,
      "step": 152509
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.1737961769104004,
      "learning_rate": 0.00015388492795809268,
      "loss": 2.0925,
      "step": 152510
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.1158117055892944,
      "learning_rate": 0.00015388135538121748,
      "loss": 1.8918,
      "step": 152511
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.0473214387893677,
      "learning_rate": 0.0001538777828315083,
      "loss": 2.0308,
      "step": 152512
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.173379898071289,
      "learning_rate": 0.0001538742103089661,
      "loss": 2.0033,
      "step": 152513
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.1637227535247803,
      "learning_rate": 0.0001538706378135913,
      "loss": 2.1263,
      "step": 152514
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.1135183572769165,
      "learning_rate": 0.00015386706534538483,
      "loss": 2.2471,
      "step": 152515
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.2621986865997314,
      "learning_rate": 0.00015386349290434715,
      "loss": 2.0544,
      "step": 152516
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.1555593013763428,
      "learning_rate": 0.0001538599204904792,
      "loss": 2.0143,
      "step": 152517
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.2855206727981567,
      "learning_rate": 0.00015385634810378127,
      "loss": 2.1481,
      "step": 152518
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.1681631803512573,
      "learning_rate": 0.00015385277574425426,
      "loss": 2.0156,
      "step": 152519
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.1097099781036377,
      "learning_rate": 0.0001538492034118987,
      "loss": 2.1591,
      "step": 152520
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.1625938415527344,
      "learning_rate": 0.00015384563110671542,
      "loss": 1.9664,
      "step": 152521
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.1796119213104248,
      "learning_rate": 0.00015384205882870488,
      "loss": 2.3135,
      "step": 152522
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.1333857774734497,
      "learning_rate": 0.000153838486577868,
      "loss": 2.0202,
      "step": 152523
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.066308856010437,
      "learning_rate": 0.00015383491435420527,
      "loss": 2.088,
      "step": 152524
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.20624840259552,
      "learning_rate": 0.00015383134215771744,
      "loss": 1.9741,
      "step": 152525
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.2969034910202026,
      "learning_rate": 0.00015382776998840498,
      "loss": 1.9842,
      "step": 152526
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.206586241722107,
      "learning_rate": 0.0001538241978462688,
      "loss": 2.0232,
      "step": 152527
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.1568831205368042,
      "learning_rate": 0.00015382062573130934,
      "loss": 2.2048,
      "step": 152528
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.12462317943573,
      "learning_rate": 0.00015381705364352752,
      "loss": 1.925,
      "step": 152529
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.3512599468231201,
      "learning_rate": 0.00015381348158292386,
      "loss": 2.215,
      "step": 152530
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.1059890985488892,
      "learning_rate": 0.0001538099095494991,
      "loss": 1.7799,
      "step": 152531
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.1300935745239258,
      "learning_rate": 0.0001538063375432536,
      "loss": 2.0303,
      "step": 152532
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.27501380443573,
      "learning_rate": 0.00015380276556418846,
      "loss": 2.1173,
      "step": 152533
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.110042929649353,
      "learning_rate": 0.00015379919361230401,
      "loss": 2.0438,
      "step": 152534
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.2128539085388184,
      "learning_rate": 0.00015379562168760114,
      "loss": 1.9983,
      "step": 152535
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.059317946434021,
      "learning_rate": 0.00015379204979008043,
      "loss": 1.9545,
      "step": 152536
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.1772338151931763,
      "learning_rate": 0.00015378847791974246,
      "loss": 2.1592,
      "step": 152537
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.14720618724823,
      "learning_rate": 0.0001537849060765881,
      "loss": 2.0894,
      "step": 152538
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.0965162515640259,
      "learning_rate": 0.00015378133426061783,
      "loss": 2.1504,
      "step": 152539
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.1352434158325195,
      "learning_rate": 0.0001537777624718323,
      "loss": 2.0141,
      "step": 152540
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.0399492979049683,
      "learning_rate": 0.0001537741907102323,
      "loss": 2.2025,
      "step": 152541
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.0802966356277466,
      "learning_rate": 0.0001537706189758185,
      "loss": 2.1608,
      "step": 152542
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.136279582977295,
      "learning_rate": 0.00015376704726859132,
      "loss": 1.9578,
      "step": 152543
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.1021803617477417,
      "learning_rate": 0.0001537634755885518,
      "loss": 1.9974,
      "step": 152544
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.1367961168289185,
      "learning_rate": 0.00015375990393570025,
      "loss": 2.0108,
      "step": 152545
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.4627197980880737,
      "learning_rate": 0.00015375633231003762,
      "loss": 1.9576,
      "step": 152546
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.3224139213562012,
      "learning_rate": 0.00015375276071156446,
      "loss": 2.1893,
      "step": 152547
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.289491891860962,
      "learning_rate": 0.0001537491891402814,
      "loss": 1.906,
      "step": 152548
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.130111575126648,
      "learning_rate": 0.000153745617596189,
      "loss": 2.0724,
      "step": 152549
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.1780478954315186,
      "learning_rate": 0.0001537420460792882,
      "loss": 2.1848,
      "step": 152550
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.3209993839263916,
      "learning_rate": 0.00015373847458957938,
      "loss": 2.246,
      "step": 152551
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.127700686454773,
      "learning_rate": 0.00015373490312706347,
      "loss": 1.9521,
      "step": 152552
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.3742002248764038,
      "learning_rate": 0.000153731331691741,
      "loss": 2.063,
      "step": 152553
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.1784504652023315,
      "learning_rate": 0.0001537277602836126,
      "loss": 2.2189,
      "step": 152554
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.1647523641586304,
      "learning_rate": 0.0001537241889026789,
      "loss": 1.961,
      "step": 152555
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.1410117149353027,
      "learning_rate": 0.00015372061754894073,
      "loss": 1.9362,
      "step": 152556
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.1316510438919067,
      "learning_rate": 0.00015371704622239854,
      "loss": 1.926,
      "step": 152557
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.2243415117263794,
      "learning_rate": 0.0001537134749230532,
      "loss": 1.9653,
      "step": 152558
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.132077693939209,
      "learning_rate": 0.00015370990365090525,
      "loss": 1.9082,
      "step": 152559
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.2711613178253174,
      "learning_rate": 0.00015370633240595558,
      "loss": 2.1377,
      "step": 152560
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.2456377744674683,
      "learning_rate": 0.0001537027611882044,
      "loss": 2.2977,
      "step": 152561
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.2031326293945312,
      "learning_rate": 0.0001536991899976528,
      "loss": 2.1446,
      "step": 152562
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.1122714281082153,
      "learning_rate": 0.00015369561883430116,
      "loss": 2.1623,
      "step": 152563
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.1159392595291138,
      "learning_rate": 0.00015369204769815037,
      "loss": 1.8469,
      "step": 152564
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.1901655197143555,
      "learning_rate": 0.00015368847658920088,
      "loss": 1.971,
      "step": 152565
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.2801578044891357,
      "learning_rate": 0.0001536849055074537,
      "loss": 2.002,
      "step": 152566
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.128102421760559,
      "learning_rate": 0.000153681334452909,
      "loss": 1.9763,
      "step": 152567
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.1064434051513672,
      "learning_rate": 0.0001536777634255678,
      "loss": 2.2243,
      "step": 152568
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.1330299377441406,
      "learning_rate": 0.0001536741924254306,
      "loss": 2.123,
      "step": 152569
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.0163216590881348,
      "learning_rate": 0.00015367062145249825,
      "loss": 1.9643,
      "step": 152570
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.0622150897979736,
      "learning_rate": 0.00015366705050677112,
      "loss": 1.8617,
      "step": 152571
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.3366527557373047,
      "learning_rate": 0.00015366347958825036,
      "loss": 2.1073,
      "step": 152572
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.128373384475708,
      "learning_rate": 0.00015365990869693603,
      "loss": 2.0142,
      "step": 152573
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.1832116842269897,
      "learning_rate": 0.0001536563378328292,
      "loss": 2.0057,
      "step": 152574
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.2437793016433716,
      "learning_rate": 0.00015365276699593036,
      "loss": 2.026,
      "step": 152575
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.442135214805603,
      "learning_rate": 0.0001536491961862403,
      "loss": 2.2887,
      "step": 152576
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.305129885673523,
      "learning_rate": 0.00015364562540375954,
      "loss": 1.9611,
      "step": 152577
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.1014403104782104,
      "learning_rate": 0.000153642054648489,
      "loss": 1.8883,
      "step": 152578
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.2171355485916138,
      "learning_rate": 0.00015363848392042898,
      "loss": 2.1884,
      "step": 152579
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.277803659439087,
      "learning_rate": 0.00015363491321958037,
      "loss": 2.106,
      "step": 152580
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.2246161699295044,
      "learning_rate": 0.00015363134254594372,
      "loss": 2.2164,
      "step": 152581
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.1950958967208862,
      "learning_rate": 0.00015362777189951993,
      "loss": 2.2912,
      "step": 152582
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.1439303159713745,
      "learning_rate": 0.00015362420128030933,
      "loss": 2.1226,
      "step": 152583
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.126530647277832,
      "learning_rate": 0.00015362063068831303,
      "loss": 2.106,
      "step": 152584
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.1587471961975098,
      "learning_rate": 0.00015361706012353112,
      "loss": 2.1929,
      "step": 152585
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.276526927947998,
      "learning_rate": 0.0001536134895859647,
      "loss": 1.9402,
      "step": 152586
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.0942966938018799,
      "learning_rate": 0.00015360991907561422,
      "loss": 2.2112,
      "step": 152587
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.3161040544509888,
      "learning_rate": 0.00015360634859248052,
      "loss": 2.0857,
      "step": 152588
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.2611899375915527,
      "learning_rate": 0.00015360277813656405,
      "loss": 2.1807,
      "step": 152589
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.2343225479125977,
      "learning_rate": 0.00015359920770786578,
      "loss": 2.0258,
      "step": 152590
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.051275372505188,
      "learning_rate": 0.000153595637306386,
      "loss": 2.0841,
      "step": 152591
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.2357336282730103,
      "learning_rate": 0.00015359206693212564,
      "loss": 1.9957,
      "step": 152592
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.440398931503296,
      "learning_rate": 0.00015358849658508517,
      "loss": 2.1646,
      "step": 152593
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.1793676614761353,
      "learning_rate": 0.00015358492626526548,
      "loss": 2.3603,
      "step": 152594
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.1854223012924194,
      "learning_rate": 0.00015358135597266697,
      "loss": 2.0428,
      "step": 152595
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.198421835899353,
      "learning_rate": 0.00015357778570729074,
      "loss": 2.0402,
      "step": 152596
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.2749502658843994,
      "learning_rate": 0.00015357421546913688,
      "loss": 2.2421,
      "step": 152597
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.2684470415115356,
      "learning_rate": 0.00015357064525820647,
      "loss": 2.0582,
      "step": 152598
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.1559667587280273,
      "learning_rate": 0.00015356707507449994,
      "loss": 2.053,
      "step": 152599
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.245622992515564,
      "learning_rate": 0.00015356350491801816,
      "loss": 1.9926,
      "step": 152600
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.0826138257980347,
      "learning_rate": 0.00015355993478876159,
      "loss": 1.9422,
      "step": 152601
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.240805745124817,
      "learning_rate": 0.00015355636468673104,
      "loss": 2.1534,
      "step": 152602
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.202451229095459,
      "learning_rate": 0.0001535527946119272,
      "loss": 1.9532,
      "step": 152603
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.0124754905700684,
      "learning_rate": 0.00015354922456435063,
      "loss": 2.143,
      "step": 152604
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.2539632320404053,
      "learning_rate": 0.0001535456545440019,
      "loss": 2.2316,
      "step": 152605
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.1573011875152588,
      "learning_rate": 0.00015354208455088195,
      "loss": 1.7798,
      "step": 152606
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.3609609603881836,
      "learning_rate": 0.00015353851458499115,
      "loss": 2.1254,
      "step": 152607
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.1263738870620728,
      "learning_rate": 0.0001535349446463304,
      "loss": 2.0098,
      "step": 152608
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.1615803241729736,
      "learning_rate": 0.00015353137473490033,
      "loss": 2.2431,
      "step": 152609
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.1654961109161377,
      "learning_rate": 0.00015352780485070148,
      "loss": 1.9808,
      "step": 152610
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.1375608444213867,
      "learning_rate": 0.0001535242349937345,
      "loss": 1.9742,
      "step": 152611
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.3387665748596191,
      "learning_rate": 0.00015352066516400022,
      "loss": 2.0934,
      "step": 152612
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.1454073190689087,
      "learning_rate": 0.00015351709536149912,
      "loss": 1.9936,
      "step": 152613
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.1459693908691406,
      "learning_rate": 0.000153513525586232,
      "loss": 2.1219,
      "step": 152614
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.1964421272277832,
      "learning_rate": 0.00015350995583819958,
      "loss": 2.3396,
      "step": 152615
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.1236120462417603,
      "learning_rate": 0.00015350638611740238,
      "loss": 2.1022,
      "step": 152616
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.2443326711654663,
      "learning_rate": 0.00015350281642384102,
      "loss": 2.1496,
      "step": 152617
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.1084725856781006,
      "learning_rate": 0.00015349924675751636,
      "loss": 2.17,
      "step": 152618
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.315528392791748,
      "learning_rate": 0.00015349567711842878,
      "loss": 2.2858,
      "step": 152619
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.122389316558838,
      "learning_rate": 0.0001534921075065793,
      "loss": 1.9768,
      "step": 152620
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.2218530178070068,
      "learning_rate": 0.00015348853792196835,
      "loss": 2.0542,
      "step": 152621
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.0477930307388306,
      "learning_rate": 0.0001534849683645966,
      "loss": 1.9873,
      "step": 152622
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.1255156993865967,
      "learning_rate": 0.00015348139883446482,
      "loss": 2.0502,
      "step": 152623
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.170974850654602,
      "learning_rate": 0.00015347782933157367,
      "loss": 2.1491,
      "step": 152624
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.270087480545044,
      "learning_rate": 0.00015347425985592358,
      "loss": 1.9868,
      "step": 152625
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.1361042261123657,
      "learning_rate": 0.00015347069040751553,
      "loss": 2.1047,
      "step": 152626
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.2822054624557495,
      "learning_rate": 0.00015346712098635005,
      "loss": 2.0676,
      "step": 152627
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.4711029529571533,
      "learning_rate": 0.0001534635515924277,
      "loss": 1.9614,
      "step": 152628
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.2227040529251099,
      "learning_rate": 0.0001534599822257493,
      "loss": 2.1107,
      "step": 152629
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.1705304384231567,
      "learning_rate": 0.00015345641288631553,
      "loss": 2.3716,
      "step": 152630
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.1858747005462646,
      "learning_rate": 0.00015345284357412678,
      "loss": 2.0576,
      "step": 152631
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.1039303541183472,
      "learning_rate": 0.00015344927428918409,
      "loss": 2.3111,
      "step": 152632
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.1232938766479492,
      "learning_rate": 0.0001534457050314879,
      "loss": 1.9226,
      "step": 152633
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.1916123628616333,
      "learning_rate": 0.00015344213580103884,
      "loss": 2.196,
      "step": 152634
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.128639817237854,
      "learning_rate": 0.00015343856659783778,
      "loss": 2.0998,
      "step": 152635
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.1767925024032593,
      "learning_rate": 0.0001534349974218851,
      "loss": 2.0044,
      "step": 152636
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.1017366647720337,
      "learning_rate": 0.0001534314282731818,
      "loss": 1.9469,
      "step": 152637
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.136518120765686,
      "learning_rate": 0.00015342785915172828,
      "loss": 1.8422,
      "step": 152638
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.050731897354126,
      "learning_rate": 0.00015342429005752538,
      "loss": 2.1026,
      "step": 152639
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.1652404069900513,
      "learning_rate": 0.00015342072099057346,
      "loss": 2.1046,
      "step": 152640
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.1191412210464478,
      "learning_rate": 0.00015341715195087358,
      "loss": 2.2497,
      "step": 152641
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.068392038345337,
      "learning_rate": 0.0001534135829384261,
      "loss": 1.9464,
      "step": 152642
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.1209558248519897,
      "learning_rate": 0.0001534100139532319,
      "loss": 1.9145,
      "step": 152643
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.1134119033813477,
      "learning_rate": 0.0001534064449952914,
      "loss": 1.92,
      "step": 152644
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.1980639696121216,
      "learning_rate": 0.00015340287606460567,
      "loss": 2.0993,
      "step": 152645
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.0826501846313477,
      "learning_rate": 0.0001533993071611749,
      "loss": 2.0273,
      "step": 152646
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.1461161375045776,
      "learning_rate": 0.00015339573828500005,
      "loss": 2.0531,
      "step": 152647
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.1948119401931763,
      "learning_rate": 0.0001533921694360816,
      "loss": 2.1608,
      "step": 152648
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.211916446685791,
      "learning_rate": 0.0001533886006144204,
      "loss": 2.1839,
      "step": 152649
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.2316715717315674,
      "learning_rate": 0.00015338503182001694,
      "loss": 1.9276,
      "step": 152650
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.1399434804916382,
      "learning_rate": 0.0001533814630528722,
      "loss": 2.0645,
      "step": 152651
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.1453670263290405,
      "learning_rate": 0.00015337789431298638,
      "loss": 2.0524,
      "step": 152652
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.156503677368164,
      "learning_rate": 0.00015337432560036052,
      "loss": 1.9445,
      "step": 152653
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.0429022312164307,
      "learning_rate": 0.00015337075691499496,
      "loss": 1.977,
      "step": 152654
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.168908715248108,
      "learning_rate": 0.00015336718825689072,
      "loss": 2.1097,
      "step": 152655
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.099668025970459,
      "learning_rate": 0.00015336361962604815,
      "loss": 1.8553,
      "step": 152656
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.0850205421447754,
      "learning_rate": 0.00015336005102246827,
      "loss": 2.0025,
      "step": 152657
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.1135573387145996,
      "learning_rate": 0.00015335648244615128,
      "loss": 2.073,
      "step": 152658
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.2277885675430298,
      "learning_rate": 0.00015335291389709823,
      "loss": 2.0686,
      "step": 152659
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.2367255687713623,
      "learning_rate": 0.00015334934537530957,
      "loss": 1.8874,
      "step": 152660
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.0527222156524658,
      "learning_rate": 0.0001533457768807861,
      "loss": 1.9976,
      "step": 152661
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.1347109079360962,
      "learning_rate": 0.00015334220841352827,
      "loss": 1.7871,
      "step": 152662
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.1328210830688477,
      "learning_rate": 0.00015333863997353718,
      "loss": 2.2654,
      "step": 152663
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.1407749652862549,
      "learning_rate": 0.00015333507156081293,
      "loss": 2.1054,
      "step": 152664
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.0455058813095093,
      "learning_rate": 0.00015333150317535662,
      "loss": 2.1432,
      "step": 152665
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.2181987762451172,
      "learning_rate": 0.0001533279348171686,
      "loss": 2.0903,
      "step": 152666
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.0253667831420898,
      "learning_rate": 0.00015332436648624983,
      "loss": 2.123,
      "step": 152667
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.129977822303772,
      "learning_rate": 0.0001533207981826007,
      "loss": 1.8429,
      "step": 152668
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.1577364206314087,
      "learning_rate": 0.00015331722990622213,
      "loss": 1.936,
      "step": 152669
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.1637073755264282,
      "learning_rate": 0.0001533136616571146,
      "loss": 2.0688,
      "step": 152670
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.3696423768997192,
      "learning_rate": 0.0001533100934352789,
      "loss": 2.1702,
      "step": 152671
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.1765856742858887,
      "learning_rate": 0.00015330652524071548,
      "loss": 2.0822,
      "step": 152672
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.2732354402542114,
      "learning_rate": 0.00015330295707342528,
      "loss": 2.1442,
      "step": 152673
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.225453495979309,
      "learning_rate": 0.0001532993889334087,
      "loss": 1.9398,
      "step": 152674
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.2115845680236816,
      "learning_rate": 0.00015329582082066663,
      "loss": 2.289,
      "step": 152675
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.2663887739181519,
      "learning_rate": 0.00015329225273519963,
      "loss": 2.0952,
      "step": 152676
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.1535910367965698,
      "learning_rate": 0.0001532886846770084,
      "loss": 1.8883,
      "step": 152677
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.1564185619354248,
      "learning_rate": 0.00015328511664609344,
      "loss": 2.1338,
      "step": 152678
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.1720823049545288,
      "learning_rate": 0.00015328154864245567,
      "loss": 2.0461,
      "step": 152679
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.1308174133300781,
      "learning_rate": 0.00015327798066609553,
      "loss": 2.0224,
      "step": 152680
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.3843973875045776,
      "learning_rate": 0.0001532744127170139,
      "loss": 2.0974,
      "step": 152681
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.1535104513168335,
      "learning_rate": 0.00015327084479521127,
      "loss": 1.9943,
      "step": 152682
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.329533576965332,
      "learning_rate": 0.0001532672769006884,
      "loss": 1.8311,
      "step": 152683
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.1357927322387695,
      "learning_rate": 0.0001532637090334458,
      "loss": 2.2178,
      "step": 152684
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.111118197441101,
      "learning_rate": 0.00015326014119348432,
      "loss": 2.0321,
      "step": 152685
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.1904888153076172,
      "learning_rate": 0.0001532565733808045,
      "loss": 2.2921,
      "step": 152686
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.2828612327575684,
      "learning_rate": 0.00015325300559540712,
      "loss": 2.0288,
      "step": 152687
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.245782732963562,
      "learning_rate": 0.0001532494378372928,
      "loss": 2.188,
      "step": 152688
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.0867623090744019,
      "learning_rate": 0.0001532458701064622,
      "loss": 2.1423,
      "step": 152689
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.1351900100708008,
      "learning_rate": 0.00015324230240291583,
      "loss": 2.0144,
      "step": 152690
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.090192198753357,
      "learning_rate": 0.00015323873472665456,
      "loss": 2.1089,
      "step": 152691
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.3666545152664185,
      "learning_rate": 0.0001532351670776789,
      "loss": 2.036,
      "step": 152692
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.1955593824386597,
      "learning_rate": 0.0001532315994559897,
      "loss": 2.0593,
      "step": 152693
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.1849147081375122,
      "learning_rate": 0.00015322803186158752,
      "loss": 1.9323,
      "step": 152694
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.308846116065979,
      "learning_rate": 0.00015322446429447305,
      "loss": 2.2206,
      "step": 152695
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.2610441446304321,
      "learning_rate": 0.00015322089675464678,
      "loss": 2.0571,
      "step": 152696
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.1843287944793701,
      "learning_rate": 0.00015321732924210963,
      "loss": 2.2983,
      "step": 152697
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.190861701965332,
      "learning_rate": 0.00015321376175686204,
      "loss": 2.1583,
      "step": 152698
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.214263677597046,
      "learning_rate": 0.00015321019429890485,
      "loss": 2.014,
      "step": 152699
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.1677780151367188,
      "learning_rate": 0.00015320662686823874,
      "loss": 2.2273,
      "step": 152700
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.283740758895874,
      "learning_rate": 0.00015320305946486422,
      "loss": 2.0675,
      "step": 152701
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.1073026657104492,
      "learning_rate": 0.00015319949208878196,
      "loss": 1.9656,
      "step": 152702
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.0718493461608887,
      "learning_rate": 0.0001531959247399928,
      "loss": 1.9484,
      "step": 152703
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.101985216140747,
      "learning_rate": 0.0001531923574184971,
      "loss": 2.1431,
      "step": 152704
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.1652861833572388,
      "learning_rate": 0.0001531887901242959,
      "loss": 2.1367,
      "step": 152705
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.1297030448913574,
      "learning_rate": 0.00015318522285738965,
      "loss": 1.869,
      "step": 152706
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.2387619018554688,
      "learning_rate": 0.0001531816556177789,
      "loss": 2.2787,
      "step": 152707
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.1152199506759644,
      "learning_rate": 0.0001531780884054646,
      "loss": 1.7684,
      "step": 152708
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.1309717893600464,
      "learning_rate": 0.00015317452122044726,
      "loss": 2.0661,
      "step": 152709
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.1941554546356201,
      "learning_rate": 0.00015317095406272742,
      "loss": 2.0282,
      "step": 152710
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.1688653230667114,
      "learning_rate": 0.00015316738693230603,
      "loss": 2.1038,
      "step": 152711
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.1728990077972412,
      "learning_rate": 0.00015316381982918354,
      "loss": 2.084,
      "step": 152712
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.2118031978607178,
      "learning_rate": 0.00015316025275336054,
      "loss": 1.9815,
      "step": 152713
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.2245405912399292,
      "learning_rate": 0.00015315668570483798,
      "loss": 2.1856,
      "step": 152714
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.0991718769073486,
      "learning_rate": 0.00015315311868361635,
      "loss": 2.0551,
      "step": 152715
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.210830807685852,
      "learning_rate": 0.00015314955168969623,
      "loss": 1.9841,
      "step": 152716
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.2984812259674072,
      "learning_rate": 0.00015314598472307848,
      "loss": 2.0557,
      "step": 152717
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.3603841066360474,
      "learning_rate": 0.00015314241778376363,
      "loss": 2.148,
      "step": 152718
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.111060619354248,
      "learning_rate": 0.00015313885087175232,
      "loss": 1.9879,
      "step": 152719
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.1146161556243896,
      "learning_rate": 0.00015313528398704535,
      "loss": 2.1959,
      "step": 152720
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.1177047491073608,
      "learning_rate": 0.0001531317171296432,
      "loss": 2.1036,
      "step": 152721
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.2947715520858765,
      "learning_rate": 0.00015312815029954678,
      "loss": 2.1939,
      "step": 152722
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.1559443473815918,
      "learning_rate": 0.00015312458349675656,
      "loss": 2.0723,
      "step": 152723
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.123018503189087,
      "learning_rate": 0.00015312101672127328,
      "loss": 1.8743,
      "step": 152724
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.2681324481964111,
      "learning_rate": 0.00015311744997309744,
      "loss": 2.0839,
      "step": 152725
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.0575000047683716,
      "learning_rate": 0.00015311388325222995,
      "loss": 2.1098,
      "step": 152726
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.3361443281173706,
      "learning_rate": 0.00015311031655867126,
      "loss": 2.2923,
      "step": 152727
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.0673524141311646,
      "learning_rate": 0.00015310674989242224,
      "loss": 2.0341,
      "step": 152728
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.169183611869812,
      "learning_rate": 0.0001531031832534835,
      "loss": 2.0476,
      "step": 152729
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.299839735031128,
      "learning_rate": 0.00015309961664185557,
      "loss": 2.2047,
      "step": 152730
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.1203073263168335,
      "learning_rate": 0.00015309605005753914,
      "loss": 2.3117,
      "step": 152731
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.1243804693222046,
      "learning_rate": 0.000153092483500535,
      "loss": 1.9573,
      "step": 152732
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.069698452949524,
      "learning_rate": 0.0001530889169708437,
      "loss": 1.953,
      "step": 152733
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.1617968082427979,
      "learning_rate": 0.000153085350468466,
      "loss": 1.9144,
      "step": 152734
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.1341341733932495,
      "learning_rate": 0.00015308178399340238,
      "loss": 2.0594,
      "step": 152735
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.1936362981796265,
      "learning_rate": 0.00015307821754565376,
      "loss": 1.9499,
      "step": 152736
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.0426485538482666,
      "learning_rate": 0.0001530746511252207,
      "loss": 1.8203,
      "step": 152737
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.0778284072875977,
      "learning_rate": 0.00015307108473210378,
      "loss": 1.9729,
      "step": 152738
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.2022618055343628,
      "learning_rate": 0.00015306751836630363,
      "loss": 2.0331,
      "step": 152739
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.0300792455673218,
      "learning_rate": 0.0001530639520278211,
      "loss": 2.022,
      "step": 152740
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.1523605585098267,
      "learning_rate": 0.00015306038571665667,
      "loss": 2.1334,
      "step": 152741
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.2626076936721802,
      "learning_rate": 0.00015305681943281123,
      "loss": 2.1345,
      "step": 152742
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.2836642265319824,
      "learning_rate": 0.00015305325317628525,
      "loss": 2.1751,
      "step": 152743
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.165666937828064,
      "learning_rate": 0.00015304968694707945,
      "loss": 2.1139,
      "step": 152744
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.215273380279541,
      "learning_rate": 0.0001530461207451944,
      "loss": 2.1193,
      "step": 152745
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.235323190689087,
      "learning_rate": 0.00015304255457063094,
      "loss": 2.0142,
      "step": 152746
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.3136898279190063,
      "learning_rate": 0.0001530389884233895,
      "loss": 1.9177,
      "step": 152747
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.207432508468628,
      "learning_rate": 0.00015303542230347107,
      "loss": 1.9823,
      "step": 152748
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.236633539199829,
      "learning_rate": 0.0001530318562108761,
      "loss": 2.2417,
      "step": 152749
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.5192238092422485,
      "learning_rate": 0.00015302829014560525,
      "loss": 2.0317,
      "step": 152750
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.2534210681915283,
      "learning_rate": 0.00015302472410765912,
      "loss": 2.1565,
      "step": 152751
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.0817917585372925,
      "learning_rate": 0.0001530211580970386,
      "loss": 2.1196,
      "step": 152752
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.2865440845489502,
      "learning_rate": 0.0001530175921137441,
      "loss": 2.032,
      "step": 152753
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.210945963859558,
      "learning_rate": 0.00015301402615777653,
      "loss": 1.8524,
      "step": 152754
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.1837362051010132,
      "learning_rate": 0.00015301046022913638,
      "loss": 2.0735,
      "step": 152755
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.0901565551757812,
      "learning_rate": 0.00015300689432782443,
      "loss": 2.0633,
      "step": 152756
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.2889089584350586,
      "learning_rate": 0.0001530033284538411,
      "loss": 2.2408,
      "step": 152757
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.24961519241333,
      "learning_rate": 0.00015299976260718738,
      "loss": 2.0784,
      "step": 152758
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.0649791955947876,
      "learning_rate": 0.00015299619678786364,
      "loss": 1.7591,
      "step": 152759
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.182529330253601,
      "learning_rate": 0.0001529926309958708,
      "loss": 2.1729,
      "step": 152760
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.1140295267105103,
      "learning_rate": 0.00015298906523120943,
      "loss": 1.9573,
      "step": 152761
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.1037112474441528,
      "learning_rate": 0.0001529854994938801,
      "loss": 2.0584,
      "step": 152762
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.1255626678466797,
      "learning_rate": 0.00015298193378388347,
      "loss": 2.0965,
      "step": 152763
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.1714061498641968,
      "learning_rate": 0.00015297836810122041,
      "loss": 2.1997,
      "step": 152764
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.00812828540802,
      "learning_rate": 0.00015297480244589129,
      "loss": 1.8858,
      "step": 152765
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.0971319675445557,
      "learning_rate": 0.0001529712368178971,
      "loss": 2.289,
      "step": 152766
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.2882473468780518,
      "learning_rate": 0.00015296767121723829,
      "loss": 2.0771,
      "step": 152767
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.1398247480392456,
      "learning_rate": 0.00015296410564391558,
      "loss": 2.1408,
      "step": 152768
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.139726996421814,
      "learning_rate": 0.00015296054009792951,
      "loss": 2.0864,
      "step": 152769
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.2126632928848267,
      "learning_rate": 0.0001529569745792809,
      "loss": 2.0444,
      "step": 152770
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.118802785873413,
      "learning_rate": 0.0001529534090879703,
      "loss": 1.9348,
      "step": 152771
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.124826192855835,
      "learning_rate": 0.00015294984362399858,
      "loss": 1.9976,
      "step": 152772
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.231642484664917,
      "learning_rate": 0.00015294627818736623,
      "loss": 2.0731,
      "step": 152773
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.1254525184631348,
      "learning_rate": 0.00015294271277807396,
      "loss": 1.9634,
      "step": 152774
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.1756246089935303,
      "learning_rate": 0.0001529391473961223,
      "loss": 2.0048,
      "step": 152775
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.1921981573104858,
      "learning_rate": 0.0001529355820415121,
      "loss": 2.1214,
      "step": 152776
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.150163173675537,
      "learning_rate": 0.00015293201671424386,
      "loss": 1.9507,
      "step": 152777
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.179566740989685,
      "learning_rate": 0.00015292845141431848,
      "loss": 2.1731,
      "step": 152778
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.101016640663147,
      "learning_rate": 0.0001529248861417365,
      "loss": 1.9845,
      "step": 152779
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.1652735471725464,
      "learning_rate": 0.00015292132089649848,
      "loss": 2.0103,
      "step": 152780
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.1538187265396118,
      "learning_rate": 0.00015291775567860506,
      "loss": 2.1493,
      "step": 152781
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.082359790802002,
      "learning_rate": 0.00015291419048805717,
      "loss": 2.1006,
      "step": 152782
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.051679253578186,
      "learning_rate": 0.00015291062532485515,
      "loss": 1.9052,
      "step": 152783
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.1065579652786255,
      "learning_rate": 0.00015290706018899997,
      "loss": 1.9066,
      "step": 152784
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.1757352352142334,
      "learning_rate": 0.0001529034950804921,
      "loss": 2.2123,
      "step": 152785
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.3926905393600464,
      "learning_rate": 0.00015289992999933228,
      "loss": 1.9526,
      "step": 152786
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.1038613319396973,
      "learning_rate": 0.000152896364945521,
      "loss": 2.1922,
      "step": 152787
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.0405676364898682,
      "learning_rate": 0.0001528927999190592,
      "loss": 2.1172,
      "step": 152788
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.1833586692810059,
      "learning_rate": 0.00015288923491994728,
      "loss": 2.0994,
      "step": 152789
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.2191309928894043,
      "learning_rate": 0.00015288566994818616,
      "loss": 1.9736,
      "step": 152790
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.2552770376205444,
      "learning_rate": 0.00015288210500377636,
      "loss": 2.0481,
      "step": 152791
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.158355951309204,
      "learning_rate": 0.00015287854008671855,
      "loss": 2.3023,
      "step": 152792
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.2768746614456177,
      "learning_rate": 0.0001528749751970133,
      "loss": 2.1023,
      "step": 152793
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.1944656372070312,
      "learning_rate": 0.00015287141033466147,
      "loss": 2.1906,
      "step": 152794
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.1573054790496826,
      "learning_rate": 0.0001528678454996635,
      "loss": 1.9795,
      "step": 152795
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.3223495483398438,
      "learning_rate": 0.00015286428069202028,
      "loss": 2.1261,
      "step": 152796
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.074081540107727,
      "learning_rate": 0.0001528607159117324,
      "loss": 2.0409,
      "step": 152797
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.0541664361953735,
      "learning_rate": 0.00015285715115880035,
      "loss": 2.2794,
      "step": 152798
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.2053805589675903,
      "learning_rate": 0.00015285358643322508,
      "loss": 2.1053,
      "step": 152799
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.2517650127410889,
      "learning_rate": 0.00015285002173500708,
      "loss": 2.0459,
      "step": 152800
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.2040369510650635,
      "learning_rate": 0.0001528464570641469,
      "loss": 2.1159,
      "step": 152801
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.22930908203125,
      "learning_rate": 0.0001528428924206455,
      "loss": 2.138,
      "step": 152802
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.3462125062942505,
      "learning_rate": 0.00015283932780450323,
      "loss": 2.1986,
      "step": 152803
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.1706628799438477,
      "learning_rate": 0.00015283576321572107,
      "loss": 2.1882,
      "step": 152804
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.1895073652267456,
      "learning_rate": 0.0001528321986542995,
      "loss": 2.0528,
      "step": 152805
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.1262634992599487,
      "learning_rate": 0.00015282863412023905,
      "loss": 2.1573,
      "step": 152806
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.2647453546524048,
      "learning_rate": 0.0001528250696135407,
      "loss": 1.9546,
      "step": 152807
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.8480918407440186,
      "learning_rate": 0.00015282150513420495,
      "loss": 2.1336,
      "step": 152808
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.406776785850525,
      "learning_rate": 0.0001528179406822323,
      "loss": 2.3754,
      "step": 152809
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.2564886808395386,
      "learning_rate": 0.0001528143762576237,
      "loss": 2.184,
      "step": 152810
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.1802690029144287,
      "learning_rate": 0.00015281081186037973,
      "loss": 2.2005,
      "step": 152811
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.1858540773391724,
      "learning_rate": 0.00015280724749050086,
      "loss": 2.0452,
      "step": 152812
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.2210619449615479,
      "learning_rate": 0.00015280368314798802,
      "loss": 2.2002,
      "step": 152813
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.1591670513153076,
      "learning_rate": 0.0001528001188328417,
      "loss": 2.1198,
      "step": 152814
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.0706030130386353,
      "learning_rate": 0.00015279655454506259,
      "loss": 1.8872,
      "step": 152815
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.1542128324508667,
      "learning_rate": 0.00015279299028465144,
      "loss": 2.08,
      "step": 152816
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.2845978736877441,
      "learning_rate": 0.00015278942605160886,
      "loss": 2.0924,
      "step": 152817
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.170750617980957,
      "learning_rate": 0.0001527858618459354,
      "loss": 2.0692,
      "step": 152818
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.1762621402740479,
      "learning_rate": 0.00015278229766763196,
      "loss": 2.2267,
      "step": 152819
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.275362253189087,
      "learning_rate": 0.0001527787335166989,
      "loss": 1.9555,
      "step": 152820
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.2607028484344482,
      "learning_rate": 0.00015277516939313723,
      "loss": 2.182,
      "step": 152821
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.0965064764022827,
      "learning_rate": 0.00015277160529694742,
      "loss": 1.9194,
      "step": 152822
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.0735352039337158,
      "learning_rate": 0.0001527680412281301,
      "loss": 2.0971,
      "step": 152823
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.1257200241088867,
      "learning_rate": 0.00015276447718668595,
      "loss": 2.0797,
      "step": 152824
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.115543007850647,
      "learning_rate": 0.0001527609131726157,
      "loss": 1.921,
      "step": 152825
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.0681198835372925,
      "learning_rate": 0.0001527573491859199,
      "loss": 2.0623,
      "step": 152826
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.0633609294891357,
      "learning_rate": 0.00015275378522659937,
      "loss": 2.0774,
      "step": 152827
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.1005561351776123,
      "learning_rate": 0.00015275022129465475,
      "loss": 2.0698,
      "step": 152828
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.1564416885375977,
      "learning_rate": 0.00015274665739008659,
      "loss": 1.9807,
      "step": 152829
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.1600459814071655,
      "learning_rate": 0.00015274309351289553,
      "loss": 1.9014,
      "step": 152830
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.1732665300369263,
      "learning_rate": 0.0001527395296630824,
      "loss": 2.0085,
      "step": 152831
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.0528974533081055,
      "learning_rate": 0.00015273596584064765,
      "loss": 1.9413,
      "step": 152832
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.156144380569458,
      "learning_rate": 0.00015273240204559217,
      "loss": 2.0395,
      "step": 152833
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.148911476135254,
      "learning_rate": 0.00015272883827791654,
      "loss": 1.8586,
      "step": 152834
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.1580390930175781,
      "learning_rate": 0.00015272527453762138,
      "loss": 2.0735,
      "step": 152835
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.201479434967041,
      "learning_rate": 0.00015272171082470729,
      "loss": 2.145,
      "step": 152836
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.1944196224212646,
      "learning_rate": 0.00015271814713917512,
      "loss": 2.092,
      "step": 152837
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.1369572877883911,
      "learning_rate": 0.00015271458348102527,
      "loss": 2.0712,
      "step": 152838
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.3693369626998901,
      "learning_rate": 0.00015271101985025873,
      "loss": 2.0846,
      "step": 152839
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.1511015892028809,
      "learning_rate": 0.00015270745624687595,
      "loss": 2.0346,
      "step": 152840
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.0809345245361328,
      "learning_rate": 0.00015270389267087766,
      "loss": 2.1646,
      "step": 152841
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.1166021823883057,
      "learning_rate": 0.00015270032912226435,
      "loss": 1.9776,
      "step": 152842
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.123280644416809,
      "learning_rate": 0.00015269676560103696,
      "loss": 2.0197,
      "step": 152843
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.2282437086105347,
      "learning_rate": 0.0001526932021071959,
      "loss": 1.9457,
      "step": 152844
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.181764841079712,
      "learning_rate": 0.0001526896386407421,
      "loss": 1.7398,
      "step": 152845
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.0466831922531128,
      "learning_rate": 0.00015268607520167607,
      "loss": 2.2209,
      "step": 152846
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.1933029890060425,
      "learning_rate": 0.0001526825117899985,
      "loss": 1.9866,
      "step": 152847
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.2975071668624878,
      "learning_rate": 0.00015267894840570985,
      "loss": 2.2072,
      "step": 152848
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.1753170490264893,
      "learning_rate": 0.0001526753850488111,
      "loss": 2.2324,
      "step": 152849
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.192993402481079,
      "learning_rate": 0.0001526718217193027,
      "loss": 2.1235,
      "step": 152850
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.169417142868042,
      "learning_rate": 0.00015266825841718547,
      "loss": 1.9547,
      "step": 152851
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.1348885297775269,
      "learning_rate": 0.00015266469514245998,
      "loss": 1.9939,
      "step": 152852
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.1989151239395142,
      "learning_rate": 0.00015266113189512696,
      "loss": 2.003,
      "step": 152853
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.3772603273391724,
      "learning_rate": 0.00015265756867518685,
      "loss": 2.1187,
      "step": 152854
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.2692526578903198,
      "learning_rate": 0.0001526540054826406,
      "loss": 2.0419,
      "step": 152855
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.100732445716858,
      "learning_rate": 0.00015265044231748868,
      "loss": 1.8959,
      "step": 152856
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.2170114517211914,
      "learning_rate": 0.00015264687917973188,
      "loss": 2.2305,
      "step": 152857
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.1613154411315918,
      "learning_rate": 0.00015264331606937087,
      "loss": 1.9303,
      "step": 152858
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.1531871557235718,
      "learning_rate": 0.00015263975298640618,
      "loss": 2.0233,
      "step": 152859
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.0849326848983765,
      "learning_rate": 0.00015263618993083847,
      "loss": 1.7948,
      "step": 152860
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.0942635536193848,
      "learning_rate": 0.0001526326269026686,
      "loss": 2.0584,
      "step": 152861
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.122521996498108,
      "learning_rate": 0.00015262906390189698,
      "loss": 2.0386,
      "step": 152862
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.117922306060791,
      "learning_rate": 0.00015262550092852445,
      "loss": 2.184,
      "step": 152863
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.0760637521743774,
      "learning_rate": 0.00015262193798255174,
      "loss": 1.9793,
      "step": 152864
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.126765251159668,
      "learning_rate": 0.00015261837506397929,
      "loss": 2.0565,
      "step": 152865
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.102426528930664,
      "learning_rate": 0.0001526148121728078,
      "loss": 2.0675,
      "step": 152866
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.1022412776947021,
      "learning_rate": 0.00015261124930903813,
      "loss": 1.9639,
      "step": 152867
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.1728215217590332,
      "learning_rate": 0.00015260768647267064,
      "loss": 1.8787,
      "step": 152868
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.280942678451538,
      "learning_rate": 0.00015260412366370634,
      "loss": 2.3152,
      "step": 152869
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.280023455619812,
      "learning_rate": 0.00015260056088214557,
      "loss": 1.8968,
      "step": 152870
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.1891920566558838,
      "learning_rate": 0.00015259699812798935,
      "loss": 2.058,
      "step": 152871
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.3291873931884766,
      "learning_rate": 0.00015259343540123793,
      "loss": 1.9353,
      "step": 152872
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.4217334985733032,
      "learning_rate": 0.00015258987270189224,
      "loss": 2.0886,
      "step": 152873
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.3363323211669922,
      "learning_rate": 0.00015258631002995282,
      "loss": 1.9303,
      "step": 152874
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.0801297426223755,
      "learning_rate": 0.00015258274738542046,
      "loss": 2.0999,
      "step": 152875
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.171592354774475,
      "learning_rate": 0.00015257918476829565,
      "loss": 2.1297,
      "step": 152876
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.3346318006515503,
      "learning_rate": 0.0001525756221785794,
      "loss": 2.023,
      "step": 152877
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.4640352725982666,
      "learning_rate": 0.0001525720596162719,
      "loss": 2.1895,
      "step": 152878
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.2759997844696045,
      "learning_rate": 0.00015256849708137412,
      "loss": 2.0497,
      "step": 152879
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.1732221841812134,
      "learning_rate": 0.00015256493457388654,
      "loss": 1.8776,
      "step": 152880
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.2743878364562988,
      "learning_rate": 0.00015256137209381,
      "loss": 1.8241,
      "step": 152881
    },
    {
      "epoch": 1.99,
      "grad_norm": 2.1139864921569824,
      "learning_rate": 0.000152557809641145,
      "loss": 2.0299,
      "step": 152882
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.3249880075454712,
      "learning_rate": 0.00015255424721589243,
      "loss": 2.0676,
      "step": 152883
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.1549638509750366,
      "learning_rate": 0.00015255068481805276,
      "loss": 2.0725,
      "step": 152884
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.2389633655548096,
      "learning_rate": 0.00015254712244762675,
      "loss": 1.98,
      "step": 152885
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.2398675680160522,
      "learning_rate": 0.0001525435601046149,
      "loss": 1.9981,
      "step": 152886
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.311300277709961,
      "learning_rate": 0.00015253999778901803,
      "loss": 2.1063,
      "step": 152887
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.2365024089813232,
      "learning_rate": 0.00015253643550083672,
      "loss": 1.9676,
      "step": 152888
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.0912927389144897,
      "learning_rate": 0.00015253287324007173,
      "loss": 1.9148,
      "step": 152889
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.551074743270874,
      "learning_rate": 0.00015252931100672367,
      "loss": 2.2269,
      "step": 152890
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.3852287530899048,
      "learning_rate": 0.00015252574880079324,
      "loss": 2.2294,
      "step": 152891
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.109504222869873,
      "learning_rate": 0.00015252218662228088,
      "loss": 1.8958,
      "step": 152892
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.0806922912597656,
      "learning_rate": 0.0001525186244711876,
      "loss": 2.0703,
      "step": 152893
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.0714502334594727,
      "learning_rate": 0.00015251506234751375,
      "loss": 2.0572,
      "step": 152894
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.2978026866912842,
      "learning_rate": 0.00015251150025126026,
      "loss": 1.8665,
      "step": 152895
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.2369550466537476,
      "learning_rate": 0.00015250793818242763,
      "loss": 2.2002,
      "step": 152896
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.173224925994873,
      "learning_rate": 0.00015250437614101648,
      "loss": 1.9618,
      "step": 152897
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.1571732759475708,
      "learning_rate": 0.00015250081412702767,
      "loss": 1.9395,
      "step": 152898
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.235982060432434,
      "learning_rate": 0.00015249725214046172,
      "loss": 1.9946,
      "step": 152899
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.047480583190918,
      "learning_rate": 0.00015249369018131923,
      "loss": 2.0965,
      "step": 152900
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.0710664987564087,
      "learning_rate": 0.00015249012824960101,
      "loss": 2.147,
      "step": 152901
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.1006823778152466,
      "learning_rate": 0.0001524865663453077,
      "loss": 2.1494,
      "step": 152902
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.1255701780319214,
      "learning_rate": 0.00015248300446843983,
      "loss": 2.1151,
      "step": 152903
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.2339625358581543,
      "learning_rate": 0.00015247944261899822,
      "loss": 1.9529,
      "step": 152904
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.2699620723724365,
      "learning_rate": 0.00015247588079698337,
      "loss": 2.181,
      "step": 152905
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.1895782947540283,
      "learning_rate": 0.00015247231900239615,
      "loss": 1.978,
      "step": 152906
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.2285964488983154,
      "learning_rate": 0.00015246875723523708,
      "loss": 2.2674,
      "step": 152907
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.1039350032806396,
      "learning_rate": 0.00015246519549550692,
      "loss": 1.912,
      "step": 152908
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.2175976037979126,
      "learning_rate": 0.0001524616337832061,
      "loss": 2.2017,
      "step": 152909
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.114985704421997,
      "learning_rate": 0.0001524580720983356,
      "loss": 2.2761,
      "step": 152910
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.144052505493164,
      "learning_rate": 0.00015245451044089577,
      "loss": 2.0824,
      "step": 152911
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.08024263381958,
      "learning_rate": 0.00015245094881088757,
      "loss": 2.032,
      "step": 152912
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.1225931644439697,
      "learning_rate": 0.0001524473872083115,
      "loss": 2.1572,
      "step": 152913
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.1025582551956177,
      "learning_rate": 0.0001524438256331683,
      "loss": 2.316,
      "step": 152914
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.0471128225326538,
      "learning_rate": 0.00015244026408545841,
      "loss": 1.9768,
      "step": 152915
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.2083393335342407,
      "learning_rate": 0.00015243670256518283,
      "loss": 2.0659,
      "step": 152916
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.160129427909851,
      "learning_rate": 0.00015243314107234187,
      "loss": 1.8861,
      "step": 152917
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.4344115257263184,
      "learning_rate": 0.00015242957960693657,
      "loss": 2.1538,
      "step": 152918
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.2266359329223633,
      "learning_rate": 0.00015242601816896733,
      "loss": 2.2306,
      "step": 152919
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.140479326248169,
      "learning_rate": 0.00015242245675843488,
      "loss": 2.2782,
      "step": 152920
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.0646716356277466,
      "learning_rate": 0.00015241889537533977,
      "loss": 2.2639,
      "step": 152921
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.186720609664917,
      "learning_rate": 0.00015241533401968294,
      "loss": 2.0424,
      "step": 152922
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.1417800188064575,
      "learning_rate": 0.00015241177269146474,
      "loss": 2.1097,
      "step": 152923
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.1150480508804321,
      "learning_rate": 0.00015240821139068605,
      "loss": 2.0588,
      "step": 152924
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.4496934413909912,
      "learning_rate": 0.00015240465011734752,
      "loss": 2.4138,
      "step": 152925
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.1995065212249756,
      "learning_rate": 0.00015240108887144973,
      "loss": 2.1883,
      "step": 152926
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.2222903966903687,
      "learning_rate": 0.00015239752765299326,
      "loss": 2.1375,
      "step": 152927
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.185302972793579,
      "learning_rate": 0.00015239396646197898,
      "loss": 2.0695,
      "step": 152928
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.1660734415054321,
      "learning_rate": 0.0001523904052984073,
      "loss": 1.8573,
      "step": 152929
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.0733612775802612,
      "learning_rate": 0.00015238684416227918,
      "loss": 2.2739,
      "step": 152930
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.1456007957458496,
      "learning_rate": 0.00015238328305359515,
      "loss": 1.9999,
      "step": 152931
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.1620880365371704,
      "learning_rate": 0.00015237972197235583,
      "loss": 1.9118,
      "step": 152932
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.2391135692596436,
      "learning_rate": 0.00015237616091856178,
      "loss": 2.1415,
      "step": 152933
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.0764751434326172,
      "learning_rate": 0.0001523725998922139,
      "loss": 1.9546,
      "step": 152934
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.023220181465149,
      "learning_rate": 0.00015236903889331266,
      "loss": 1.8203,
      "step": 152935
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.0127842426300049,
      "learning_rate": 0.00015236547792185892,
      "loss": 2.1274,
      "step": 152936
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.0814948081970215,
      "learning_rate": 0.00015236191697785305,
      "loss": 2.1023,
      "step": 152937
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.121886968612671,
      "learning_rate": 0.00015235835606129615,
      "loss": 2.0982,
      "step": 152938
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.2071411609649658,
      "learning_rate": 0.00015235479517218843,
      "loss": 2.1023,
      "step": 152939
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.0856151580810547,
      "learning_rate": 0.0001523512343105308,
      "loss": 1.8421,
      "step": 152940
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.0982871055603027,
      "learning_rate": 0.00015234767347632375,
      "loss": 2.0066,
      "step": 152941
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.176430106163025,
      "learning_rate": 0.00015234411266956823,
      "loss": 2.1686,
      "step": 152942
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.2496237754821777,
      "learning_rate": 0.00015234055189026455,
      "loss": 2.0023,
      "step": 152943
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.1683865785598755,
      "learning_rate": 0.0001523369911384138,
      "loss": 2.1191,
      "step": 152944
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.234831690788269,
      "learning_rate": 0.00015233343041401613,
      "loss": 2.0273,
      "step": 152945
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.185185194015503,
      "learning_rate": 0.00015232986971707262,
      "loss": 2.0377,
      "step": 152946
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.0884861946105957,
      "learning_rate": 0.00015232630904758363,
      "loss": 1.9314,
      "step": 152947
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.2426754236221313,
      "learning_rate": 0.00015232274840555013,
      "loss": 1.9927,
      "step": 152948
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.1357003450393677,
      "learning_rate": 0.00015231918779097246,
      "loss": 2.2143,
      "step": 152949
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.2819286584854126,
      "learning_rate": 0.0001523156272038517,
      "loss": 2.0345,
      "step": 152950
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.1692836284637451,
      "learning_rate": 0.00015231206664418799,
      "loss": 2.232,
      "step": 152951
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.145365834236145,
      "learning_rate": 0.00015230850611198238,
      "loss": 2.0291,
      "step": 152952
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.1631652116775513,
      "learning_rate": 0.00015230494560723528,
      "loss": 2.217,
      "step": 152953
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.2258753776550293,
      "learning_rate": 0.0001523013851299476,
      "loss": 1.9672,
      "step": 152954
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.2836610078811646,
      "learning_rate": 0.00015229782468011976,
      "loss": 2.2743,
      "step": 152955
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.1405701637268066,
      "learning_rate": 0.00015229426425775284,
      "loss": 1.9019,
      "step": 152956
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.2652448415756226,
      "learning_rate": 0.0001522907038628469,
      "loss": 1.9388,
      "step": 152957
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.1150215864181519,
      "learning_rate": 0.00015228714349540303,
      "loss": 2.0192,
      "step": 152958
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.2014176845550537,
      "learning_rate": 0.00015228358315542168,
      "loss": 2.2762,
      "step": 152959
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.2437978982925415,
      "learning_rate": 0.0001522800228429037,
      "loss": 1.9917,
      "step": 152960
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.207801103591919,
      "learning_rate": 0.00015227646255784954,
      "loss": 2.1904,
      "step": 152961
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.1040050983428955,
      "learning_rate": 0.00015227290230026018,
      "loss": 1.9112,
      "step": 152962
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.503686547279358,
      "learning_rate": 0.00015226934207013588,
      "loss": 2.2015,
      "step": 152963
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.1729490756988525,
      "learning_rate": 0.0001522657818674776,
      "loss": 2.1053,
      "step": 152964
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.458837866783142,
      "learning_rate": 0.00015226222169228583,
      "loss": 2.2174,
      "step": 152965
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.0870239734649658,
      "learning_rate": 0.00015225866154456132,
      "loss": 2.0856,
      "step": 152966
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.3299609422683716,
      "learning_rate": 0.00015225510142430465,
      "loss": 2.0343,
      "step": 152967
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.5191115140914917,
      "learning_rate": 0.0001522515413315167,
      "loss": 2.0321,
      "step": 152968
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.1367751359939575,
      "learning_rate": 0.0001522479812661979,
      "loss": 2.1698,
      "step": 152969
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.133925199508667,
      "learning_rate": 0.00015224442122834902,
      "loss": 2.2703,
      "step": 152970
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.3190124034881592,
      "learning_rate": 0.0001522408612179706,
      "loss": 2.0874,
      "step": 152971
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.0844178199768066,
      "learning_rate": 0.0001522373012350635,
      "loss": 2.1607,
      "step": 152972
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.251942753791809,
      "learning_rate": 0.00015223374127962817,
      "loss": 2.3569,
      "step": 152973
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.1551882028579712,
      "learning_rate": 0.00015223018135166545,
      "loss": 2.1499,
      "step": 152974
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.0960028171539307,
      "learning_rate": 0.00015222662145117596,
      "loss": 2.151,
      "step": 152975
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.1060545444488525,
      "learning_rate": 0.00015222306157816035,
      "loss": 2.0803,
      "step": 152976
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.6276448965072632,
      "learning_rate": 0.00015221950173261912,
      "loss": 2.1222,
      "step": 152977
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.1304967403411865,
      "learning_rate": 0.0001522159419145532,
      "loss": 2.0354,
      "step": 152978
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.3404353857040405,
      "learning_rate": 0.00015221238212396296,
      "loss": 2.0372,
      "step": 152979
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.0788487195968628,
      "learning_rate": 0.00015220882236084942,
      "loss": 2.0743,
      "step": 152980
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.1648248434066772,
      "learning_rate": 0.00015220526262521298,
      "loss": 2.1572,
      "step": 152981
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.2975772619247437,
      "learning_rate": 0.00015220170291705432,
      "loss": 2.2924,
      "step": 152982
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.1871201992034912,
      "learning_rate": 0.0001521981432363742,
      "loss": 1.8464,
      "step": 152983
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.2377427816390991,
      "learning_rate": 0.0001521945835831733,
      "loss": 2.1239,
      "step": 152984
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.1279674768447876,
      "learning_rate": 0.00015219102395745208,
      "loss": 1.9221,
      "step": 152985
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.2669141292572021,
      "learning_rate": 0.00015218746435921145,
      "loss": 1.9904,
      "step": 152986
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.4881831407546997,
      "learning_rate": 0.00015218390478845197,
      "loss": 2.0265,
      "step": 152987
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.4124171733856201,
      "learning_rate": 0.00015218034524517413,
      "loss": 2.1943,
      "step": 152988
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.2713408470153809,
      "learning_rate": 0.00015217678572937895,
      "loss": 2.0918,
      "step": 152989
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.1327890157699585,
      "learning_rate": 0.0001521732262410667,
      "loss": 2.1535,
      "step": 152990
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.3306056261062622,
      "learning_rate": 0.00015216966678023843,
      "loss": 2.0477,
      "step": 152991
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.5574263334274292,
      "learning_rate": 0.00015216610734689455,
      "loss": 1.9745,
      "step": 152992
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.2321337461471558,
      "learning_rate": 0.00015216254794103576,
      "loss": 2.2075,
      "step": 152993
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.3484185934066772,
      "learning_rate": 0.0001521589885626627,
      "loss": 2.1584,
      "step": 152994
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.1854640245437622,
      "learning_rate": 0.00015215542921177618,
      "loss": 2.0619,
      "step": 152995
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.2394806146621704,
      "learning_rate": 0.0001521518698883766,
      "loss": 2.0789,
      "step": 152996
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.277858853340149,
      "learning_rate": 0.00015214831059246492,
      "loss": 2.0646,
      "step": 152997
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.4891571998596191,
      "learning_rate": 0.00015214475132404166,
      "loss": 2.0586,
      "step": 152998
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.0280694961547852,
      "learning_rate": 0.00015214119208310746,
      "loss": 1.8593,
      "step": 152999
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.240348219871521,
      "learning_rate": 0.0001521376328696629,
      "loss": 2.0651,
      "step": 153000
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.5301231145858765,
      "learning_rate": 0.00015213407368370888,
      "loss": 2.0537,
      "step": 153001
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.3030346632003784,
      "learning_rate": 0.00015213051452524578,
      "loss": 2.1406,
      "step": 153002
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.2078005075454712,
      "learning_rate": 0.00015212695539427456,
      "loss": 1.9873,
      "step": 153003
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.1737043857574463,
      "learning_rate": 0.00015212339629079558,
      "loss": 1.8664,
      "step": 153004
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.0744861364364624,
      "learning_rate": 0.0001521198372148099,
      "loss": 2.1714,
      "step": 153005
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.1171215772628784,
      "learning_rate": 0.00015211627816631772,
      "loss": 2.2143,
      "step": 153006
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.305349349975586,
      "learning_rate": 0.00015211271914532,
      "loss": 1.9091,
      "step": 153007
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.1426823139190674,
      "learning_rate": 0.00015210916015181723,
      "loss": 1.9573,
      "step": 153008
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.210210919380188,
      "learning_rate": 0.00015210560118581026,
      "loss": 2.1944,
      "step": 153009
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.078234076499939,
      "learning_rate": 0.0001521020422472995,
      "loss": 2.0357,
      "step": 153010
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.065609335899353,
      "learning_rate": 0.00015209848333628604,
      "loss": 2.1734,
      "step": 153011
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.0868017673492432,
      "learning_rate": 0.00015209492445276997,
      "loss": 1.9347,
      "step": 153012
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.2401841878890991,
      "learning_rate": 0.00015209136559675245,
      "loss": 2.0534,
      "step": 153013
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.129628300666809,
      "learning_rate": 0.00015208780676823376,
      "loss": 2.0698,
      "step": 153014
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.035097599029541,
      "learning_rate": 0.00015208424796721488,
      "loss": 2.1683,
      "step": 153015
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.1823078393936157,
      "learning_rate": 0.0001520806891936962,
      "loss": 1.8721,
      "step": 153016
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.176720380783081,
      "learning_rate": 0.0001520771304476788,
      "loss": 2.2523,
      "step": 153017
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.079406499862671,
      "learning_rate": 0.00015207357172916273,
      "loss": 2.1178,
      "step": 153018
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.1474050283432007,
      "learning_rate": 0.00015207001303814915,
      "loss": 1.9051,
      "step": 153019
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.2284009456634521,
      "learning_rate": 0.0001520664543746384,
      "loss": 2.2518,
      "step": 153020
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.147771954536438,
      "learning_rate": 0.00015206289573863144,
      "loss": 2.1226,
      "step": 153021
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.2061448097229004,
      "learning_rate": 0.00015205933713012863,
      "loss": 2.0388,
      "step": 153022
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.0995982885360718,
      "learning_rate": 0.00015205577854913101,
      "loss": 1.8044,
      "step": 153023
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.3720369338989258,
      "learning_rate": 0.00015205221999563885,
      "loss": 2.1103,
      "step": 153024
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.0650417804718018,
      "learning_rate": 0.00015204866146965304,
      "loss": 1.8944,
      "step": 153025
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.1631135940551758,
      "learning_rate": 0.00015204510297117405,
      "loss": 2.1344,
      "step": 153026
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.2138036489486694,
      "learning_rate": 0.0001520415445002028,
      "loss": 2.2006,
      "step": 153027
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.1884660720825195,
      "learning_rate": 0.0001520379860567397,
      "loss": 2.3418,
      "step": 153028
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.1099879741668701,
      "learning_rate": 0.00015203442764078577,
      "loss": 1.9888,
      "step": 153029
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.1156220436096191,
      "learning_rate": 0.00015203086925234122,
      "loss": 2.0978,
      "step": 153030
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.2385306358337402,
      "learning_rate": 0.000152027310891407,
      "loss": 1.9933,
      "step": 153031
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.393180251121521,
      "learning_rate": 0.00015202375255798359,
      "loss": 2.1609,
      "step": 153032
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.1258455514907837,
      "learning_rate": 0.00015202019425207186,
      "loss": 2.0686,
      "step": 153033
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.40916907787323,
      "learning_rate": 0.0001520166359736723,
      "loss": 2.2059,
      "step": 153034
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.33965265750885,
      "learning_rate": 0.00015201307772278583,
      "loss": 2.1691,
      "step": 153035
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.277604103088379,
      "learning_rate": 0.0001520095194994127,
      "loss": 2.077,
      "step": 153036
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.3074876070022583,
      "learning_rate": 0.00015200596130355394,
      "loss": 2.149,
      "step": 153037
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.1457844972610474,
      "learning_rate": 0.00015200240313520993,
      "loss": 2.1649,
      "step": 153038
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.434391736984253,
      "learning_rate": 0.00015199884499438154,
      "loss": 2.1191,
      "step": 153039
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.133318305015564,
      "learning_rate": 0.0001519952868810693,
      "loss": 2.0415,
      "step": 153040
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.4815576076507568,
      "learning_rate": 0.00015199172879527413,
      "loss": 2.1762,
      "step": 153041
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.3066215515136719,
      "learning_rate": 0.0001519881707369963,
      "loss": 2.228,
      "step": 153042
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.1857222318649292,
      "learning_rate": 0.00015198461270623674,
      "loss": 2.1655,
      "step": 153043
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.3910837173461914,
      "learning_rate": 0.0001519810547029959,
      "loss": 1.883,
      "step": 153044
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.0969600677490234,
      "learning_rate": 0.00015197749672727474,
      "loss": 2.1835,
      "step": 153045
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.2912532091140747,
      "learning_rate": 0.00015197393877907365,
      "loss": 1.8613,
      "step": 153046
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.1660072803497314,
      "learning_rate": 0.0001519703808583936,
      "loss": 1.922,
      "step": 153047
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.213229775428772,
      "learning_rate": 0.0001519668229652348,
      "loss": 2.2025,
      "step": 153048
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.2362207174301147,
      "learning_rate": 0.00015196326509959833,
      "loss": 2.1305,
      "step": 153049
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.491252064704895,
      "learning_rate": 0.0001519597072614845,
      "loss": 2.1452,
      "step": 153050
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.299581527709961,
      "learning_rate": 0.00015195614945089433,
      "loss": 2.1235,
      "step": 153051
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.1230367422103882,
      "learning_rate": 0.0001519525916678282,
      "loss": 2.0731,
      "step": 153052
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.4664311408996582,
      "learning_rate": 0.00015194903391228695,
      "loss": 2.2684,
      "step": 153053
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.5209906101226807,
      "learning_rate": 0.00015194547618427114,
      "loss": 2.1069,
      "step": 153054
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.2066318988800049,
      "learning_rate": 0.00015194191848378155,
      "loss": 2.1267,
      "step": 153055
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.0272600650787354,
      "learning_rate": 0.0001519383608108186,
      "loss": 1.8532,
      "step": 153056
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.2482854127883911,
      "learning_rate": 0.00015193480316538316,
      "loss": 2.0381,
      "step": 153057
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.180925726890564,
      "learning_rate": 0.0001519312455474758,
      "loss": 2.065,
      "step": 153058
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.1563373804092407,
      "learning_rate": 0.00015192768795709727,
      "loss": 2.3112,
      "step": 153059
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.1124205589294434,
      "learning_rate": 0.0001519241303942482,
      "loss": 2.0022,
      "step": 153060
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.3045698404312134,
      "learning_rate": 0.00015192057285892926,
      "loss": 2.1649,
      "step": 153061
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.22683584690094,
      "learning_rate": 0.0001519170153511409,
      "loss": 2.0392,
      "step": 153062
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.202666997909546,
      "learning_rate": 0.00015191345787088413,
      "loss": 1.8524,
      "step": 153063
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.1930330991744995,
      "learning_rate": 0.00015190990041815934,
      "loss": 1.8531,
      "step": 153064
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.1040382385253906,
      "learning_rate": 0.00015190634299296738,
      "loss": 2.2113,
      "step": 153065
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.2908004522323608,
      "learning_rate": 0.00015190278559530887,
      "loss": 1.936,
      "step": 153066
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.3493977785110474,
      "learning_rate": 0.00015189922822518427,
      "loss": 1.8521,
      "step": 153067
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.1719263792037964,
      "learning_rate": 0.0001518956708825945,
      "loss": 2.1495,
      "step": 153068
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.2054115533828735,
      "learning_rate": 0.00015189211356754015,
      "loss": 2.0306,
      "step": 153069
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.2389131784439087,
      "learning_rate": 0.00015188855628002175,
      "loss": 2.1896,
      "step": 153070
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.2824259996414185,
      "learning_rate": 0.00015188499902004014,
      "loss": 2.0658,
      "step": 153071
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.2998075485229492,
      "learning_rate": 0.00015188144178759593,
      "loss": 2.0672,
      "step": 153072
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.1426506042480469,
      "learning_rate": 0.00015187788458268962,
      "loss": 2.1409,
      "step": 153073
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.1184133291244507,
      "learning_rate": 0.00015187432740532212,
      "loss": 2.2104,
      "step": 153074
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.078246831893921,
      "learning_rate": 0.000151870770255494,
      "loss": 2.0481,
      "step": 153075
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.720152735710144,
      "learning_rate": 0.00015186721313320577,
      "loss": 2.0812,
      "step": 153076
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.2407079935073853,
      "learning_rate": 0.00015186365603845836,
      "loss": 2.0008,
      "step": 153077
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.5357896089553833,
      "learning_rate": 0.00015186009897125228,
      "loss": 2.0964,
      "step": 153078
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.2867037057876587,
      "learning_rate": 0.00015185654193158808,
      "loss": 2.1553,
      "step": 153079
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.204371452331543,
      "learning_rate": 0.00015185298491946666,
      "loss": 2.1452,
      "step": 153080
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.3516265153884888,
      "learning_rate": 0.00015184942793488847,
      "loss": 1.9418,
      "step": 153081
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.4899882078170776,
      "learning_rate": 0.00015184587097785438,
      "loss": 1.9739,
      "step": 153082
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.5546659231185913,
      "learning_rate": 0.0001518423140483649,
      "loss": 2.2306,
      "step": 153083
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.246062994003296,
      "learning_rate": 0.0001518387571464208,
      "loss": 2.1884,
      "step": 153084
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.1722971200942993,
      "learning_rate": 0.00015183520027202253,
      "loss": 2.1646,
      "step": 153085
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.4916744232177734,
      "learning_rate": 0.00015183164342517098,
      "loss": 2.0725,
      "step": 153086
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.455981969833374,
      "learning_rate": 0.0001518280866058666,
      "loss": 2.0383,
      "step": 153087
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.3834271430969238,
      "learning_rate": 0.00015182452981411034,
      "loss": 2.0805,
      "step": 153088
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.0005903244018555,
      "learning_rate": 0.00015182097304990253,
      "loss": 2.112,
      "step": 153089
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.2664752006530762,
      "learning_rate": 0.00015181741631324427,
      "loss": 2.023,
      "step": 153090
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.3473724126815796,
      "learning_rate": 0.00015181385960413572,
      "loss": 2.286,
      "step": 153091
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.2178630828857422,
      "learning_rate": 0.00015181030292257786,
      "loss": 2.0228,
      "step": 153092
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.1493430137634277,
      "learning_rate": 0.00015180674626857118,
      "loss": 1.8189,
      "step": 153093
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.1019943952560425,
      "learning_rate": 0.0001518031896421165,
      "loss": 2.0843,
      "step": 153094
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.2139979600906372,
      "learning_rate": 0.0001517996330432143,
      "loss": 2.1826,
      "step": 153095
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.050833821296692,
      "learning_rate": 0.0001517960764718656,
      "loss": 2.1345,
      "step": 153096
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.1333779096603394,
      "learning_rate": 0.00015179251992807053,
      "loss": 2.0792,
      "step": 153097
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.2384872436523438,
      "learning_rate": 0.00015178896341183016,
      "loss": 2.148,
      "step": 153098
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.1351065635681152,
      "learning_rate": 0.00015178540692314493,
      "loss": 2.108,
      "step": 153099
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.0490971803665161,
      "learning_rate": 0.00015178185046201567,
      "loss": 2.0185,
      "step": 153100
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.2265775203704834,
      "learning_rate": 0.0001517782940284429,
      "loss": 2.0859,
      "step": 153101
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.342597246170044,
      "learning_rate": 0.00015177473762242755,
      "loss": 2.1207,
      "step": 153102
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.1030036211013794,
      "learning_rate": 0.00015177118124396982,
      "loss": 2.0788,
      "step": 153103
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.2087618112564087,
      "learning_rate": 0.00015176762489307077,
      "loss": 2.0734,
      "step": 153104
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.2145671844482422,
      "learning_rate": 0.0001517640685697308,
      "loss": 2.3216,
      "step": 153105
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.250279426574707,
      "learning_rate": 0.0001517605122739508,
      "loss": 1.9496,
      "step": 153106
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.0979877710342407,
      "learning_rate": 0.00015175695600573114,
      "loss": 1.8852,
      "step": 153107
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.1327236890792847,
      "learning_rate": 0.00015175339976507298,
      "loss": 1.8655,
      "step": 153108
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.1313729286193848,
      "learning_rate": 0.00015174984355197639,
      "loss": 2.0453,
      "step": 153109
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.2035936117172241,
      "learning_rate": 0.0001517462873664424,
      "loss": 1.9519,
      "step": 153110
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.2696750164031982,
      "learning_rate": 0.00015174273120847152,
      "loss": 1.9511,
      "step": 153111
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.1686859130859375,
      "learning_rate": 0.0001517391750780645,
      "loss": 2.0625,
      "step": 153112
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.1053122282028198,
      "learning_rate": 0.00015173561897522192,
      "loss": 2.361,
      "step": 153113
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.1661217212677002,
      "learning_rate": 0.00015173206289994468,
      "loss": 2.0665,
      "step": 153114
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.291731834411621,
      "learning_rate": 0.00015172850685223304,
      "loss": 2.1045,
      "step": 153115
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.1971744298934937,
      "learning_rate": 0.00015172495083208795,
      "loss": 2.0271,
      "step": 153116
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.2857861518859863,
      "learning_rate": 0.00015172139483950992,
      "loss": 2.1546,
      "step": 153117
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.1019572019577026,
      "learning_rate": 0.00015171783887449978,
      "loss": 2.1826,
      "step": 153118
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.2363016605377197,
      "learning_rate": 0.00015171428293705793,
      "loss": 2.1258,
      "step": 153119
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.2836285829544067,
      "learning_rate": 0.00015171072702718552,
      "loss": 1.8878,
      "step": 153120
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.1775990724563599,
      "learning_rate": 0.00015170717114488258,
      "loss": 2.059,
      "step": 153121
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.089916467666626,
      "learning_rate": 0.0001517036152901502,
      "loss": 2.1271,
      "step": 153122
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.0324186086654663,
      "learning_rate": 0.00015170005946298884,
      "loss": 2.0962,
      "step": 153123
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.2853524684906006,
      "learning_rate": 0.00015169650366339934,
      "loss": 2.0224,
      "step": 153124
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.1489794254302979,
      "learning_rate": 0.00015169294789138212,
      "loss": 2.3417,
      "step": 153125
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.1421644687652588,
      "learning_rate": 0.0001516893921469382,
      "loss": 2.0714,
      "step": 153126
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.2154200077056885,
      "learning_rate": 0.00015168583643006785,
      "loss": 2.0612,
      "step": 153127
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.1611204147338867,
      "learning_rate": 0.00015168228074077198,
      "loss": 2.1256,
      "step": 153128
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.16482412815094,
      "learning_rate": 0.00015167872507905106,
      "loss": 2.1714,
      "step": 153129
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.1670759916305542,
      "learning_rate": 0.00015167516944490598,
      "loss": 1.8341,
      "step": 153130
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.2421990633010864,
      "learning_rate": 0.00015167161383833718,
      "loss": 2.1253,
      "step": 153131
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.1805108785629272,
      "learning_rate": 0.00015166805825934563,
      "loss": 2.1538,
      "step": 153132
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.183416724205017,
      "learning_rate": 0.00015166450270793157,
      "loss": 2.0855,
      "step": 153133
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.1413755416870117,
      "learning_rate": 0.00015166094718409602,
      "loss": 2.1378,
      "step": 153134
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.2270742654800415,
      "learning_rate": 0.00015165739168783934,
      "loss": 2.0307,
      "step": 153135
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.172787070274353,
      "learning_rate": 0.0001516538362191625,
      "loss": 2.0106,
      "step": 153136
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.236843466758728,
      "learning_rate": 0.0001516502807780659,
      "loss": 2.1186,
      "step": 153137
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.0791038274765015,
      "learning_rate": 0.0001516467253645504,
      "loss": 2.1099,
      "step": 153138
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.1790343523025513,
      "learning_rate": 0.0001516431699786166,
      "loss": 1.9385,
      "step": 153139
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.1192212104797363,
      "learning_rate": 0.00015163961462026515,
      "loss": 1.974,
      "step": 153140
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.221881628036499,
      "learning_rate": 0.00015163605928949657,
      "loss": 2.321,
      "step": 153141
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.1426249742507935,
      "learning_rate": 0.0001516325039863117,
      "loss": 2.0447,
      "step": 153142
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.3100121021270752,
      "learning_rate": 0.00015162894871071112,
      "loss": 2.1303,
      "step": 153143
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.0282047986984253,
      "learning_rate": 0.0001516253934626956,
      "loss": 1.9638,
      "step": 153144
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.1606711149215698,
      "learning_rate": 0.00015162183824226567,
      "loss": 2.0099,
      "step": 153145
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.1775696277618408,
      "learning_rate": 0.0001516182830494221,
      "loss": 2.0596,
      "step": 153146
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.1410069465637207,
      "learning_rate": 0.00015161472788416538,
      "loss": 2.2738,
      "step": 153147
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.2573587894439697,
      "learning_rate": 0.0001516111727464964,
      "loss": 2.113,
      "step": 153148
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.1101469993591309,
      "learning_rate": 0.00015160761763641556,
      "loss": 2.015,
      "step": 153149
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.1029353141784668,
      "learning_rate": 0.0001516040625539238,
      "loss": 1.9743,
      "step": 153150
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.1215161085128784,
      "learning_rate": 0.00015160050749902162,
      "loss": 2.0013,
      "step": 153151
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.1652292013168335,
      "learning_rate": 0.00015159695247170962,
      "loss": 2.0832,
      "step": 153152
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.3300156593322754,
      "learning_rate": 0.00015159339747198864,
      "loss": 2.0657,
      "step": 153153
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.2304236888885498,
      "learning_rate": 0.00015158984249985928,
      "loss": 2.3711,
      "step": 153154
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.1292743682861328,
      "learning_rate": 0.00015158628755532205,
      "loss": 2.0679,
      "step": 153155
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.3888453245162964,
      "learning_rate": 0.00015158273263837782,
      "loss": 1.9201,
      "step": 153156
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.1888351440429688,
      "learning_rate": 0.00015157917774902716,
      "loss": 2.086,
      "step": 153157
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.1313284635543823,
      "learning_rate": 0.00015157562288727064,
      "loss": 1.772,
      "step": 153158
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.0838277339935303,
      "learning_rate": 0.00015157206805310912,
      "loss": 2.1683,
      "step": 153159
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.2228014469146729,
      "learning_rate": 0.00015156851324654318,
      "loss": 1.9584,
      "step": 153160
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.1039259433746338,
      "learning_rate": 0.0001515649584675733,
      "loss": 2.1564,
      "step": 153161
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.1681357622146606,
      "learning_rate": 0.00015156140371620042,
      "loss": 2.3153,
      "step": 153162
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.1203821897506714,
      "learning_rate": 0.00015155784899242508,
      "loss": 2.0015,
      "step": 153163
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.1288279294967651,
      "learning_rate": 0.0001515542942962478,
      "loss": 2.1094,
      "step": 153164
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.194887399673462,
      "learning_rate": 0.00015155073962766955,
      "loss": 2.2013,
      "step": 153165
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.1215349435806274,
      "learning_rate": 0.00015154718498669068,
      "loss": 2.1099,
      "step": 153166
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.193284511566162,
      "learning_rate": 0.0001515436303733121,
      "loss": 2.018,
      "step": 153167
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.1253490447998047,
      "learning_rate": 0.00015154007578753437,
      "loss": 2.0059,
      "step": 153168
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.0651240348815918,
      "learning_rate": 0.00015153652122935814,
      "loss": 2.0732,
      "step": 153169
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.1002241373062134,
      "learning_rate": 0.00015153296669878391,
      "loss": 2.1735,
      "step": 153170
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.0976810455322266,
      "learning_rate": 0.00015152941219581268,
      "loss": 2.073,
      "step": 153171
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.0924392938613892,
      "learning_rate": 0.00015152585772044478,
      "loss": 2.073,
      "step": 153172
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.2116069793701172,
      "learning_rate": 0.0001515223032726812,
      "loss": 2.2791,
      "step": 153173
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.1585839986801147,
      "learning_rate": 0.0001515187488525224,
      "loss": 2.1888,
      "step": 153174
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.2092007398605347,
      "learning_rate": 0.00015151519445996906,
      "loss": 2.1329,
      "step": 153175
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.2572792768478394,
      "learning_rate": 0.00015151164009502167,
      "loss": 2.3277,
      "step": 153176
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.135130524635315,
      "learning_rate": 0.0001515080857576813,
      "loss": 1.9568,
      "step": 153177
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.2618489265441895,
      "learning_rate": 0.00015150453144794815,
      "loss": 2.2227,
      "step": 153178
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.0820343494415283,
      "learning_rate": 0.00015150097716582332,
      "loss": 2.0341,
      "step": 153179
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.1143648624420166,
      "learning_rate": 0.0001514974229113071,
      "loss": 2.1778,
      "step": 153180
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.1613150835037231,
      "learning_rate": 0.00015149386868440056,
      "loss": 1.8609,
      "step": 153181
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.138926386833191,
      "learning_rate": 0.0001514903144851039,
      "loss": 2.1761,
      "step": 153182
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.2289988994598389,
      "learning_rate": 0.00015148676031341807,
      "loss": 2.1463,
      "step": 153183
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.066800832748413,
      "learning_rate": 0.00015148320616934353,
      "loss": 1.7301,
      "step": 153184
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.178489089012146,
      "learning_rate": 0.00015147965205288124,
      "loss": 2.1535,
      "step": 153185
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.0976548194885254,
      "learning_rate": 0.00015147609796403152,
      "loss": 2.0594,
      "step": 153186
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.1755061149597168,
      "learning_rate": 0.00015147254390279543,
      "loss": 2.2159,
      "step": 153187
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.2555639743804932,
      "learning_rate": 0.00015146898986917318,
      "loss": 2.0281,
      "step": 153188
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.1837282180786133,
      "learning_rate": 0.00015146543586316575,
      "loss": 2.0512,
      "step": 153189
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.14889395236969,
      "learning_rate": 0.00015146188188477363,
      "loss": 2.0947,
      "step": 153190
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.049599051475525,
      "learning_rate": 0.00015145832793399762,
      "loss": 2.0534,
      "step": 153191
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.1804792881011963,
      "learning_rate": 0.0001514547740108382,
      "loss": 2.1343,
      "step": 153192
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.2760411500930786,
      "learning_rate": 0.0001514512201152964,
      "loss": 2.1477,
      "step": 153193
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.2191200256347656,
      "learning_rate": 0.00015144766624737235,
      "loss": 2.2771,
      "step": 153194
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.1234124898910522,
      "learning_rate": 0.00015144411240706712,
      "loss": 2.0212,
      "step": 153195
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.2706496715545654,
      "learning_rate": 0.00015144055859438112,
      "loss": 2.2079,
      "step": 153196
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.1242824792861938,
      "learning_rate": 0.00015143700480931525,
      "loss": 2.2413,
      "step": 153197
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.2299972772598267,
      "learning_rate": 0.0001514334510518699,
      "loss": 2.0841,
      "step": 153198
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.2116022109985352,
      "learning_rate": 0.00015142989732204617,
      "loss": 2.2256,
      "step": 153199
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.1247113943099976,
      "learning_rate": 0.0001514263436198441,
      "loss": 2.19,
      "step": 153200
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.194893717765808,
      "learning_rate": 0.00015142278994526484,
      "loss": 2.2393,
      "step": 153201
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.1200604438781738,
      "learning_rate": 0.00015141923629830882,
      "loss": 2.0115,
      "step": 153202
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.1876296997070312,
      "learning_rate": 0.0001514156826789768,
      "loss": 2.0397,
      "step": 153203
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.0846747159957886,
      "learning_rate": 0.00015141212908726935,
      "loss": 2.1058,
      "step": 153204
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.1783443689346313,
      "learning_rate": 0.0001514085755231873,
      "loss": 2.0715,
      "step": 153205
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.2063008546829224,
      "learning_rate": 0.00015140502198673117,
      "loss": 2.2818,
      "step": 153206
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.0347105264663696,
      "learning_rate": 0.00015140146847790172,
      "loss": 1.8758,
      "step": 153207
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.220404028892517,
      "learning_rate": 0.00015139791499669937,
      "loss": 2.2218,
      "step": 153208
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.16738760471344,
      "learning_rate": 0.00015139436154312504,
      "loss": 2.0084,
      "step": 153209
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.163744330406189,
      "learning_rate": 0.00015139080811717924,
      "loss": 2.0067,
      "step": 153210
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.0935004949569702,
      "learning_rate": 0.00015138725471886282,
      "loss": 2.066,
      "step": 153211
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.005417823791504,
      "learning_rate": 0.00015138370134817626,
      "loss": 2.24,
      "step": 153212
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.1840709447860718,
      "learning_rate": 0.00015138014800512032,
      "loss": 1.9709,
      "step": 153213
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.1188857555389404,
      "learning_rate": 0.00015137659468969546,
      "loss": 2.0908,
      "step": 153214
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.1857084035873413,
      "learning_rate": 0.00015137304140190263,
      "loss": 2.0363,
      "step": 153215
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.1177507638931274,
      "learning_rate": 0.00015136948814174224,
      "loss": 1.9571,
      "step": 153216
    },
    {
      "epoch": 1.99,
      "grad_norm": 1.0810083150863647,
      "learning_rate": 0.00015136593490921521,
      "loss": 1.9909,
      "step": 153217
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.2094550132751465,
      "learning_rate": 0.00015136238170432206,
      "loss": 1.9406,
      "step": 153218
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.1687999963760376,
      "learning_rate": 0.0001513588285270634,
      "loss": 2.0873,
      "step": 153219
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.2050455808639526,
      "learning_rate": 0.00015135527537743986,
      "loss": 1.923,
      "step": 153220
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.1556265354156494,
      "learning_rate": 0.00015135172225545229,
      "loss": 2.1476,
      "step": 153221
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.327239990234375,
      "learning_rate": 0.00015134816916110116,
      "loss": 2.1186,
      "step": 153222
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.0815588235855103,
      "learning_rate": 0.00015134461609438724,
      "loss": 1.8934,
      "step": 153223
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.1588047742843628,
      "learning_rate": 0.00015134106305531123,
      "loss": 2.1364,
      "step": 153224
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.2538577318191528,
      "learning_rate": 0.00015133751004387368,
      "loss": 2.2153,
      "step": 153225
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.134206771850586,
      "learning_rate": 0.00015133395706007526,
      "loss": 1.7141,
      "step": 153226
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.2199020385742188,
      "learning_rate": 0.0001513304041039167,
      "loss": 2.0045,
      "step": 153227
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.1179014444351196,
      "learning_rate": 0.00015132685117539852,
      "loss": 1.9345,
      "step": 153228
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.1362583637237549,
      "learning_rate": 0.00015132329827452162,
      "loss": 2.1317,
      "step": 153229
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.1367661952972412,
      "learning_rate": 0.00015131974540128653,
      "loss": 2.0146,
      "step": 153230
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.0971088409423828,
      "learning_rate": 0.00015131619255569388,
      "loss": 1.9126,
      "step": 153231
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.1737606525421143,
      "learning_rate": 0.00015131263973774424,
      "loss": 1.9368,
      "step": 153232
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.0463056564331055,
      "learning_rate": 0.00015130908694743853,
      "loss": 2.0598,
      "step": 153233
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.1210187673568726,
      "learning_rate": 0.00015130553418477713,
      "loss": 2.2776,
      "step": 153234
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.372896432876587,
      "learning_rate": 0.000151301981449761,
      "loss": 2.1889,
      "step": 153235
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.1913604736328125,
      "learning_rate": 0.00015129842874239056,
      "loss": 2.0288,
      "step": 153236
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.1458930969238281,
      "learning_rate": 0.0001512948760626666,
      "loss": 1.8934,
      "step": 153237
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.1137953996658325,
      "learning_rate": 0.00015129132341058958,
      "loss": 2.2747,
      "step": 153238
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.127564549446106,
      "learning_rate": 0.00015128777078616046,
      "loss": 2.22,
      "step": 153239
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.1528114080429077,
      "learning_rate": 0.00015128421818937963,
      "loss": 1.9318,
      "step": 153240
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.059850811958313,
      "learning_rate": 0.00015128066562024797,
      "loss": 2.0678,
      "step": 153241
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.166580080986023,
      "learning_rate": 0.00015127711307876603,
      "loss": 2.0153,
      "step": 153242
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.0793793201446533,
      "learning_rate": 0.0001512735605649344,
      "loss": 2.0206,
      "step": 153243
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.4951437711715698,
      "learning_rate": 0.0001512700080787539,
      "loss": 1.9856,
      "step": 153244
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.0749125480651855,
      "learning_rate": 0.00015126645562022517,
      "loss": 2.1593,
      "step": 153245
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.090501070022583,
      "learning_rate": 0.00015126290318934866,
      "loss": 2.1853,
      "step": 153246
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.1768758296966553,
      "learning_rate": 0.00015125935078612532,
      "loss": 2.0922,
      "step": 153247
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.1828216314315796,
      "learning_rate": 0.00015125579841055565,
      "loss": 1.8067,
      "step": 153248
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.2190035581588745,
      "learning_rate": 0.00015125224606264023,
      "loss": 2.1198,
      "step": 153249
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.088591456413269,
      "learning_rate": 0.00015124869374237993,
      "loss": 2.0531,
      "step": 153250
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.143782138824463,
      "learning_rate": 0.0001512451414497752,
      "loss": 2.0198,
      "step": 153251
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.0817502737045288,
      "learning_rate": 0.00015124158918482691,
      "loss": 2.2147,
      "step": 153252
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.2221705913543701,
      "learning_rate": 0.00015123803694753568,
      "loss": 2.0972,
      "step": 153253
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.105603575706482,
      "learning_rate": 0.00015123448473790204,
      "loss": 1.9619,
      "step": 153254
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.1291896104812622,
      "learning_rate": 0.0001512309325559266,
      "loss": 2.3254,
      "step": 153255
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.1459347009658813,
      "learning_rate": 0.00015122738040161028,
      "loss": 2.2154,
      "step": 153256
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.0888128280639648,
      "learning_rate": 0.0001512238282749535,
      "loss": 2.1327,
      "step": 153257
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.158191442489624,
      "learning_rate": 0.00015122027617595708,
      "loss": 2.0988,
      "step": 153258
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.1680647134780884,
      "learning_rate": 0.00015121672410462166,
      "loss": 2.1492,
      "step": 153259
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.0873968601226807,
      "learning_rate": 0.00015121317206094782,
      "loss": 2.2132,
      "step": 153260
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.0492020845413208,
      "learning_rate": 0.00015120962004493618,
      "loss": 2.0403,
      "step": 153261
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.1520116329193115,
      "learning_rate": 0.0001512060680565876,
      "loss": 1.8712,
      "step": 153262
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.0680592060089111,
      "learning_rate": 0.00015120251609590247,
      "loss": 1.9056,
      "step": 153263
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.3091368675231934,
      "learning_rate": 0.0001511989641628818,
      "loss": 1.9959,
      "step": 153264
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.163770079612732,
      "learning_rate": 0.00015119541225752585,
      "loss": 2.246,
      "step": 153265
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.2522916793823242,
      "learning_rate": 0.00015119186037983572,
      "loss": 2.1671,
      "step": 153266
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.6419252157211304,
      "learning_rate": 0.0001511883085298116,
      "loss": 2.1057,
      "step": 153267
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.2087351083755493,
      "learning_rate": 0.00015118475670745454,
      "loss": 1.9341,
      "step": 153268
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.222009301185608,
      "learning_rate": 0.00015118120491276489,
      "loss": 2.0644,
      "step": 153269
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.307517647743225,
      "learning_rate": 0.0001511776531457436,
      "loss": 2.0484,
      "step": 153270
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.1976906061172485,
      "learning_rate": 0.00015117410140639106,
      "loss": 2.0664,
      "step": 153271
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.265504240989685,
      "learning_rate": 0.0001511705496947082,
      "loss": 1.9055,
      "step": 153272
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.0530462265014648,
      "learning_rate": 0.0001511669980106955,
      "loss": 2.0406,
      "step": 153273
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.148978352546692,
      "learning_rate": 0.00015116344635435376,
      "loss": 1.8873,
      "step": 153274
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.2458114624023438,
      "learning_rate": 0.00015115989472568332,
      "loss": 2.3221,
      "step": 153275
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.323725700378418,
      "learning_rate": 0.00015115634312468527,
      "loss": 2.0804,
      "step": 153276
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.2054002285003662,
      "learning_rate": 0.0001511527915513599,
      "loss": 2.2668,
      "step": 153277
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.2324775457382202,
      "learning_rate": 0.00015114924000570815,
      "loss": 1.8713,
      "step": 153278
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.172315239906311,
      "learning_rate": 0.00015114568848773058,
      "loss": 2.0042,
      "step": 153279
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.3525524139404297,
      "learning_rate": 0.0001511421369974278,
      "loss": 1.9972,
      "step": 153280
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.1467093229293823,
      "learning_rate": 0.00015113858553480046,
      "loss": 2.1176,
      "step": 153281
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.116567850112915,
      "learning_rate": 0.00015113503409984935,
      "loss": 2.13,
      "step": 153282
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.1306120157241821,
      "learning_rate": 0.00015113148269257492,
      "loss": 1.9479,
      "step": 153283
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.221447229385376,
      "learning_rate": 0.00015112793131297808,
      "loss": 2.2659,
      "step": 153284
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.3381019830703735,
      "learning_rate": 0.00015112437996105937,
      "loss": 1.9721,
      "step": 153285
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.2863261699676514,
      "learning_rate": 0.00015112082863681943,
      "loss": 2.2604,
      "step": 153286
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.1510335206985474,
      "learning_rate": 0.00015111727734025887,
      "loss": 2.0197,
      "step": 153287
    },
    {
      "epoch": 2.0,
      "grad_norm": 0.9962593913078308,
      "learning_rate": 0.0001511137260713785,
      "loss": 2.0802,
      "step": 153288
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.2604808807373047,
      "learning_rate": 0.0001511101748301788,
      "loss": 2.0003,
      "step": 153289
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.2222076654434204,
      "learning_rate": 0.00015110662361666062,
      "loss": 2.1025,
      "step": 153290
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.1643939018249512,
      "learning_rate": 0.00015110307243082453,
      "loss": 2.1825,
      "step": 153291
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.1761077642440796,
      "learning_rate": 0.0001510995212726712,
      "loss": 2.1898,
      "step": 153292
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.1999834775924683,
      "learning_rate": 0.00015109597014220113,
      "loss": 1.9823,
      "step": 153293
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.0608552694320679,
      "learning_rate": 0.0001510924190394153,
      "loss": 2.0018,
      "step": 153294
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.1184331178665161,
      "learning_rate": 0.00015108886796431407,
      "loss": 2.2904,
      "step": 153295
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.2279833555221558,
      "learning_rate": 0.00015108531691689834,
      "loss": 2.0201,
      "step": 153296
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.2329288721084595,
      "learning_rate": 0.00015108176589716867,
      "loss": 1.9866,
      "step": 153297
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.2041475772857666,
      "learning_rate": 0.00015107821490512567,
      "loss": 2.1211,
      "step": 153298
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.2864294052124023,
      "learning_rate": 0.00015107466394077,
      "loss": 2.0863,
      "step": 153299
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.0941442251205444,
      "learning_rate": 0.0001510711130041024,
      "loss": 2.2515,
      "step": 153300
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.0792102813720703,
      "learning_rate": 0.00015106756209512344,
      "loss": 2.0506,
      "step": 153301
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.1749213933944702,
      "learning_rate": 0.0001510640112138339,
      "loss": 1.995,
      "step": 153302
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.0167092084884644,
      "learning_rate": 0.0001510604603602344,
      "loss": 1.9234,
      "step": 153303
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.142078161239624,
      "learning_rate": 0.00015105690953432557,
      "loss": 2.1514,
      "step": 153304
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.0881953239440918,
      "learning_rate": 0.0001510533587361079,
      "loss": 1.9485,
      "step": 153305
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.273404598236084,
      "learning_rate": 0.0001510498079655824,
      "loss": 2.0922,
      "step": 153306
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.1611064672470093,
      "learning_rate": 0.00015104625722274942,
      "loss": 2.0956,
      "step": 153307
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.2660783529281616,
      "learning_rate": 0.0001510427065076099,
      "loss": 2.1355,
      "step": 153308
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.1402543783187866,
      "learning_rate": 0.0001510391558201643,
      "loss": 2.1203,
      "step": 153309
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.100461483001709,
      "learning_rate": 0.00015103560516041336,
      "loss": 1.9053,
      "step": 153310
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.1345864534378052,
      "learning_rate": 0.00015103205452835758,
      "loss": 1.9524,
      "step": 153311
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.058478593826294,
      "learning_rate": 0.0001510285039239979,
      "loss": 2.0864,
      "step": 153312
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.0720502138137817,
      "learning_rate": 0.0001510249533473347,
      "loss": 2.0409,
      "step": 153313
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.1058416366577148,
      "learning_rate": 0.00015102140279836887,
      "loss": 2.1765,
      "step": 153314
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.1944968700408936,
      "learning_rate": 0.000151017852277101,
      "loss": 2.1132,
      "step": 153315
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.0846394300460815,
      "learning_rate": 0.0001510143017835317,
      "loss": 2.1084,
      "step": 153316
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.0826879739761353,
      "learning_rate": 0.00015101075131766155,
      "loss": 1.9291,
      "step": 153317
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.08280348777771,
      "learning_rate": 0.00015100720087949144,
      "loss": 2.1244,
      "step": 153318
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.2322887182235718,
      "learning_rate": 0.00015100365046902178,
      "loss": 2.1041,
      "step": 153319
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.0782177448272705,
      "learning_rate": 0.00015100010008625345,
      "loss": 1.8823,
      "step": 153320
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.2262202501296997,
      "learning_rate": 0.00015099654973118706,
      "loss": 2.0555,
      "step": 153321
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.0634117126464844,
      "learning_rate": 0.0001509929994038232,
      "loss": 2.2614,
      "step": 153322
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.2517189979553223,
      "learning_rate": 0.0001509894491041624,
      "loss": 1.9241,
      "step": 153323
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.2218586206436157,
      "learning_rate": 0.00015098589883220565,
      "loss": 2.1324,
      "step": 153324
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.0861881971359253,
      "learning_rate": 0.0001509823485879533,
      "loss": 2.0903,
      "step": 153325
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.2761716842651367,
      "learning_rate": 0.00015097879837140625,
      "loss": 2.1098,
      "step": 153326
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.4130029678344727,
      "learning_rate": 0.00015097524818256505,
      "loss": 1.8926,
      "step": 153327
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.0640430450439453,
      "learning_rate": 0.00015097169802143028,
      "loss": 2.096,
      "step": 153328
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.3152241706848145,
      "learning_rate": 0.00015096814788800278,
      "loss": 2.2235,
      "step": 153329
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.1426618099212646,
      "learning_rate": 0.00015096459778228308,
      "loss": 2.1158,
      "step": 153330
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.4584946632385254,
      "learning_rate": 0.00015096104770427181,
      "loss": 2.2302,
      "step": 153331
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.1649090051651,
      "learning_rate": 0.00015095749765396975,
      "loss": 1.9708,
      "step": 153332
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.2647926807403564,
      "learning_rate": 0.00015095394763137756,
      "loss": 2.1185,
      "step": 153333
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.144264578819275,
      "learning_rate": 0.00015095039763649573,
      "loss": 2.276,
      "step": 153334
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.2136298418045044,
      "learning_rate": 0.00015094684766932511,
      "loss": 2.1352,
      "step": 153335
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.124903917312622,
      "learning_rate": 0.00015094329772986632,
      "loss": 2.055,
      "step": 153336
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.1330586671829224,
      "learning_rate": 0.00015093974781811987,
      "loss": 1.8798,
      "step": 153337
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.296617865562439,
      "learning_rate": 0.00015093619793408663,
      "loss": 1.9876,
      "step": 153338
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.319940447807312,
      "learning_rate": 0.0001509326480777671,
      "loss": 2.1492,
      "step": 153339
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.1446285247802734,
      "learning_rate": 0.000150929098249162,
      "loss": 2.1115,
      "step": 153340
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.26852285861969,
      "learning_rate": 0.00015092554844827212,
      "loss": 2.082,
      "step": 153341
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.543286681175232,
      "learning_rate": 0.00015092199867509785,
      "loss": 2.2412,
      "step": 153342
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.3227670192718506,
      "learning_rate": 0.00015091844892964007,
      "loss": 2.2153,
      "step": 153343
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.0340749025344849,
      "learning_rate": 0.0001509148992118994,
      "loss": 1.9114,
      "step": 153344
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.2063249349594116,
      "learning_rate": 0.00015091134952187633,
      "loss": 2.0698,
      "step": 153345
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.498114824295044,
      "learning_rate": 0.0001509077998595718,
      "loss": 2.0945,
      "step": 153346
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.0535553693771362,
      "learning_rate": 0.00015090425022498626,
      "loss": 2.0414,
      "step": 153347
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.1924608945846558,
      "learning_rate": 0.00015090070061812033,
      "loss": 2.1629,
      "step": 153348
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.2297552824020386,
      "learning_rate": 0.00015089715103897494,
      "loss": 1.9965,
      "step": 153349
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.1555464267730713,
      "learning_rate": 0.00015089360148755045,
      "loss": 1.9243,
      "step": 153350
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.1726243495941162,
      "learning_rate": 0.0001508900519638478,
      "loss": 1.9577,
      "step": 153351
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.1612151861190796,
      "learning_rate": 0.00015088650246786745,
      "loss": 1.9551,
      "step": 153352
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.0697646141052246,
      "learning_rate": 0.00015088295299961013,
      "loss": 2.0538,
      "step": 153353
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.157094120979309,
      "learning_rate": 0.00015087940355907638,
      "loss": 2.0252,
      "step": 153354
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.2590028047561646,
      "learning_rate": 0.00015087585414626703,
      "loss": 1.9553,
      "step": 153355
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.2479453086853027,
      "learning_rate": 0.0001508723047611826,
      "loss": 2.231,
      "step": 153356
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.1445279121398926,
      "learning_rate": 0.00015086875540382396,
      "loss": 1.9679,
      "step": 153357
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.1342724561691284,
      "learning_rate": 0.0001508652060741916,
      "loss": 2.0473,
      "step": 153358
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.2761222124099731,
      "learning_rate": 0.00015086165677228626,
      "loss": 2.1615,
      "step": 153359
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.4248310327529907,
      "learning_rate": 0.00015085810749810837,
      "loss": 1.9148,
      "step": 153360
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.0500160455703735,
      "learning_rate": 0.00015085455825165892,
      "loss": 2.0337,
      "step": 153361
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.1126855611801147,
      "learning_rate": 0.0001508510090329383,
      "loss": 2.0831,
      "step": 153362
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.1598204374313354,
      "learning_rate": 0.00015084745984194744,
      "loss": 1.9624,
      "step": 153363
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.2598204612731934,
      "learning_rate": 0.00015084391067868684,
      "loss": 2.0613,
      "step": 153364
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.1020748615264893,
      "learning_rate": 0.00015084036154315713,
      "loss": 2.0011,
      "step": 153365
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.17671799659729,
      "learning_rate": 0.00015083681243535893,
      "loss": 1.902,
      "step": 153366
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.1526286602020264,
      "learning_rate": 0.0001508332633552931,
      "loss": 2.1723,
      "step": 153367
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.127646803855896,
      "learning_rate": 0.00015082971430296007,
      "loss": 2.0428,
      "step": 153368
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.3284777402877808,
      "learning_rate": 0.00015082616527836072,
      "loss": 2.0947,
      "step": 153369
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.2337816953659058,
      "learning_rate": 0.00015082261628149562,
      "loss": 2.3647,
      "step": 153370
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.1494109630584717,
      "learning_rate": 0.00015081906731236538,
      "loss": 1.8573,
      "step": 153371
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.1859182119369507,
      "learning_rate": 0.0001508155183709706,
      "loss": 2.0744,
      "step": 153372
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.305988073348999,
      "learning_rate": 0.00015081196945731213,
      "loss": 2.0994,
      "step": 153373
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.1931133270263672,
      "learning_rate": 0.0001508084205713904,
      "loss": 1.9669,
      "step": 153374
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.1227573156356812,
      "learning_rate": 0.00015080487171320635,
      "loss": 2.089,
      "step": 153375
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.4643828868865967,
      "learning_rate": 0.0001508013228827605,
      "loss": 2.1321,
      "step": 153376
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.157533884048462,
      "learning_rate": 0.00015079777408005346,
      "loss": 1.9945,
      "step": 153377
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.1381756067276,
      "learning_rate": 0.00015079422530508586,
      "loss": 2.2284,
      "step": 153378
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.2109383344650269,
      "learning_rate": 0.0001507906765578585,
      "loss": 2.1889,
      "step": 153379
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.1677889823913574,
      "learning_rate": 0.00015078712783837188,
      "loss": 2.0303,
      "step": 153380
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.3354718685150146,
      "learning_rate": 0.00015078357914662685,
      "loss": 2.0615,
      "step": 153381
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.2021286487579346,
      "learning_rate": 0.00015078003048262398,
      "loss": 1.9211,
      "step": 153382
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.1519559621810913,
      "learning_rate": 0.00015077648184636392,
      "loss": 2.1554,
      "step": 153383
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.1534819602966309,
      "learning_rate": 0.00015077293323784722,
      "loss": 2.0994,
      "step": 153384
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.0768710374832153,
      "learning_rate": 0.00015076938465707475,
      "loss": 1.9609,
      "step": 153385
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.0839364528656006,
      "learning_rate": 0.00015076583610404694,
      "loss": 1.9777,
      "step": 153386
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.1391680240631104,
      "learning_rate": 0.0001507622875787647,
      "loss": 2.1202,
      "step": 153387
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.1713358163833618,
      "learning_rate": 0.00015075873908122861,
      "loss": 2.0191,
      "step": 153388
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.1554462909698486,
      "learning_rate": 0.00015075519061143923,
      "loss": 2.0347,
      "step": 153389
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.1193231344223022,
      "learning_rate": 0.00015075164216939721,
      "loss": 2.0856,
      "step": 153390
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.2669084072113037,
      "learning_rate": 0.00015074809375510337,
      "loss": 2.1344,
      "step": 153391
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.178812861442566,
      "learning_rate": 0.00015074454536855817,
      "loss": 2.0934,
      "step": 153392
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.355338215827942,
      "learning_rate": 0.00015074099700976247,
      "loss": 2.108,
      "step": 153393
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.2091082334518433,
      "learning_rate": 0.00015073744867871686,
      "loss": 1.9443,
      "step": 153394
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.090244174003601,
      "learning_rate": 0.00015073390037542193,
      "loss": 2.0256,
      "step": 153395
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.170720100402832,
      "learning_rate": 0.00015073035209987835,
      "loss": 2.0418,
      "step": 153396
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.784633755683899,
      "learning_rate": 0.00015072680385208687,
      "loss": 2.2015,
      "step": 153397
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.39674973487854,
      "learning_rate": 0.000150723255632048,
      "loss": 2.2714,
      "step": 153398
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.1237478256225586,
      "learning_rate": 0.0001507197074397626,
      "loss": 2.1506,
      "step": 153399
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.0203930139541626,
      "learning_rate": 0.00015071615927523123,
      "loss": 1.8401,
      "step": 153400
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.1122527122497559,
      "learning_rate": 0.00015071261113845457,
      "loss": 2.0396,
      "step": 153401
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.1140105724334717,
      "learning_rate": 0.0001507090630294331,
      "loss": 2.0773,
      "step": 153402
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.4327988624572754,
      "learning_rate": 0.0001507055149481678,
      "loss": 1.9914,
      "step": 153403
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.1540248394012451,
      "learning_rate": 0.00015070196689465898,
      "loss": 2.1316,
      "step": 153404
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.1067461967468262,
      "learning_rate": 0.00015069841886890763,
      "loss": 2.1809,
      "step": 153405
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.2533705234527588,
      "learning_rate": 0.0001506948708709141,
      "loss": 2.2233,
      "step": 153406
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.1046779155731201,
      "learning_rate": 0.0001506913229006795,
      "loss": 2.0198,
      "step": 153407
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.2123082876205444,
      "learning_rate": 0.00015068777495820396,
      "loss": 1.9944,
      "step": 153408
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.2237812280654907,
      "learning_rate": 0.00015068422704348847,
      "loss": 2.3553,
      "step": 153409
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.2045156955718994,
      "learning_rate": 0.00015068067915653352,
      "loss": 2.3226,
      "step": 153410
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.2856642007827759,
      "learning_rate": 0.00015067713129733996,
      "loss": 2.1363,
      "step": 153411
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.1750140190124512,
      "learning_rate": 0.00015067358346590825,
      "loss": 2.0382,
      "step": 153412
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.3422235250473022,
      "learning_rate": 0.0001506700356622392,
      "loss": 2.0594,
      "step": 153413
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.2310855388641357,
      "learning_rate": 0.00015066648788633345,
      "loss": 2.0769,
      "step": 153414
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.1612290143966675,
      "learning_rate": 0.00015066294013819164,
      "loss": 2.0008,
      "step": 153415
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.2874541282653809,
      "learning_rate": 0.0001506593924178142,
      "loss": 2.1154,
      "step": 153416
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.5048595666885376,
      "learning_rate": 0.0001506558447252022,
      "loss": 2.3204,
      "step": 153417
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.4050524234771729,
      "learning_rate": 0.00015065229706035592,
      "loss": 2.1439,
      "step": 153418
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.0625532865524292,
      "learning_rate": 0.00015064874942327634,
      "loss": 2.0347,
      "step": 153419
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.0954468250274658,
      "learning_rate": 0.00015064520181396398,
      "loss": 2.0273,
      "step": 153420
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.4209963083267212,
      "learning_rate": 0.00015064165423241948,
      "loss": 2.008,
      "step": 153421
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.1385616064071655,
      "learning_rate": 0.00015063810667864346,
      "loss": 2.2372,
      "step": 153422
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.1352695226669312,
      "learning_rate": 0.0001506345591526367,
      "loss": 2.0867,
      "step": 153423
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.2299525737762451,
      "learning_rate": 0.0001506310116543997,
      "loss": 2.1102,
      "step": 153424
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.1876978874206543,
      "learning_rate": 0.0001506274641839333,
      "loss": 2.068,
      "step": 153425
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.214104175567627,
      "learning_rate": 0.00015062391674123808,
      "loss": 2.1816,
      "step": 153426
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.2469180822372437,
      "learning_rate": 0.00015062036932631458,
      "loss": 2.2967,
      "step": 153427
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.070879578590393,
      "learning_rate": 0.0001506168219391637,
      "loss": 2.0265,
      "step": 153428
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.0972737073898315,
      "learning_rate": 0.00015061327457978598,
      "loss": 1.9783,
      "step": 153429
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.1278241872787476,
      "learning_rate": 0.00015060972724818196,
      "loss": 2.0196,
      "step": 153430
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.2864995002746582,
      "learning_rate": 0.0001506061799443525,
      "loss": 2.0599,
      "step": 153431
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.063138484954834,
      "learning_rate": 0.0001506026326682982,
      "loss": 1.8516,
      "step": 153432
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.1795921325683594,
      "learning_rate": 0.00015059908542001957,
      "loss": 2.1601,
      "step": 153433
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.1550885438919067,
      "learning_rate": 0.0001505955381995175,
      "loss": 2.1731,
      "step": 153434
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.1350711584091187,
      "learning_rate": 0.0001505919910067924,
      "loss": 1.9872,
      "step": 153435
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.1042389869689941,
      "learning_rate": 0.0001505884438418452,
      "loss": 2.029,
      "step": 153436
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.2140921354293823,
      "learning_rate": 0.00015058489670467646,
      "loss": 1.9327,
      "step": 153437
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.1427373886108398,
      "learning_rate": 0.00015058134959528677,
      "loss": 1.8607,
      "step": 153438
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.1555671691894531,
      "learning_rate": 0.0001505778025136767,
      "loss": 1.9207,
      "step": 153439
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.1330657005310059,
      "learning_rate": 0.0001505742554598472,
      "loss": 1.9812,
      "step": 153440
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.1509606838226318,
      "learning_rate": 0.00015057070843379862,
      "loss": 1.9875,
      "step": 153441
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.1860307455062866,
      "learning_rate": 0.0001505671614355319,
      "loss": 2.0852,
      "step": 153442
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.1293611526489258,
      "learning_rate": 0.00015056361446504752,
      "loss": 1.8544,
      "step": 153443
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.1885403394699097,
      "learning_rate": 0.00015056006752234624,
      "loss": 2.0576,
      "step": 153444
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.195417046546936,
      "learning_rate": 0.00015055652060742855,
      "loss": 1.9889,
      "step": 153445
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.254604458808899,
      "learning_rate": 0.0001505529737202953,
      "loss": 1.9984,
      "step": 153446
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.210898756980896,
      "learning_rate": 0.00015054942686094696,
      "loss": 2.0085,
      "step": 153447
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.2243528366088867,
      "learning_rate": 0.00015054588002938446,
      "loss": 1.9831,
      "step": 153448
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.4081356525421143,
      "learning_rate": 0.00015054233322560825,
      "loss": 2.1935,
      "step": 153449
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.3480714559555054,
      "learning_rate": 0.00015053878644961908,
      "loss": 2.0812,
      "step": 153450
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.1593310832977295,
      "learning_rate": 0.00015053523970141744,
      "loss": 1.9977,
      "step": 153451
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.148897409439087,
      "learning_rate": 0.00015053169298100426,
      "loss": 1.8651,
      "step": 153452
    },
    {
      "epoch": 2.0,
      "grad_norm": 0.9647863507270813,
      "learning_rate": 0.00015052814628837993,
      "loss": 1.9205,
      "step": 153453
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.121233582496643,
      "learning_rate": 0.0001505245996235453,
      "loss": 2.0405,
      "step": 153454
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.123430848121643,
      "learning_rate": 0.00015052105298650107,
      "loss": 1.9036,
      "step": 153455
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.0856693983078003,
      "learning_rate": 0.00015051750637724772,
      "loss": 1.8841,
      "step": 153456
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.0655510425567627,
      "learning_rate": 0.00015051395979578586,
      "loss": 2.1251,
      "step": 153457
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.1918045282363892,
      "learning_rate": 0.00015051041324211646,
      "loss": 2.1243,
      "step": 153458
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.4753410816192627,
      "learning_rate": 0.00015050686671623986,
      "loss": 1.8567,
      "step": 153459
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.117319107055664,
      "learning_rate": 0.00015050332021815694,
      "loss": 1.9661,
      "step": 153460
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.1313939094543457,
      "learning_rate": 0.0001504997737478683,
      "loss": 2.2002,
      "step": 153461
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.2789157629013062,
      "learning_rate": 0.00015049622730537457,
      "loss": 2.0159,
      "step": 153462
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.1234809160232544,
      "learning_rate": 0.0001504926808906763,
      "loss": 2.3265,
      "step": 153463
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.0953236818313599,
      "learning_rate": 0.00015048913450377437,
      "loss": 2.0683,
      "step": 153464
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.1013598442077637,
      "learning_rate": 0.0001504855881446692,
      "loss": 1.8662,
      "step": 153465
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.0798128843307495,
      "learning_rate": 0.00015048204181336173,
      "loss": 2.0579,
      "step": 153466
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.319382905960083,
      "learning_rate": 0.00015047849550985247,
      "loss": 2.2348,
      "step": 153467
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.198728084564209,
      "learning_rate": 0.00015047494923414205,
      "loss": 2.0883,
      "step": 153468
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.0594615936279297,
      "learning_rate": 0.00015047140298623106,
      "loss": 1.8219,
      "step": 153469
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.1663190126419067,
      "learning_rate": 0.00015046785676612036,
      "loss": 1.9378,
      "step": 153470
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.261694312095642,
      "learning_rate": 0.00015046431057381043,
      "loss": 2.0198,
      "step": 153471
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.2513669729232788,
      "learning_rate": 0.00015046076440930213,
      "loss": 2.0736,
      "step": 153472
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.2611956596374512,
      "learning_rate": 0.00015045721827259585,
      "loss": 2.2085,
      "step": 153473
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.2177541255950928,
      "learning_rate": 0.0001504536721636926,
      "loss": 2.3654,
      "step": 153474
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.173005223274231,
      "learning_rate": 0.00015045012608259263,
      "loss": 2.1441,
      "step": 153475
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.236937403678894,
      "learning_rate": 0.00015044658002929693,
      "loss": 2.1077,
      "step": 153476
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.2143585681915283,
      "learning_rate": 0.00015044303400380592,
      "loss": 1.932,
      "step": 153477
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.2967489957809448,
      "learning_rate": 0.00015043948800612047,
      "loss": 1.9515,
      "step": 153478
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.100358009338379,
      "learning_rate": 0.00015043594203624107,
      "loss": 1.8566,
      "step": 153479
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.4770280122756958,
      "learning_rate": 0.00015043239609416863,
      "loss": 2.0642,
      "step": 153480
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.3788973093032837,
      "learning_rate": 0.0001504288501799034,
      "loss": 2.0656,
      "step": 153481
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.4236364364624023,
      "learning_rate": 0.0001504253042934464,
      "loss": 2.1527,
      "step": 153482
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.1678634881973267,
      "learning_rate": 0.00015042175843479807,
      "loss": 1.8418,
      "step": 153483
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.1932501792907715,
      "learning_rate": 0.00015041821260395927,
      "loss": 1.9276,
      "step": 153484
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.2497538328170776,
      "learning_rate": 0.00015041466680093042,
      "loss": 2.2954,
      "step": 153485
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.3424643278121948,
      "learning_rate": 0.00015041112102571254,
      "loss": 2.194,
      "step": 153486
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.2389715909957886,
      "learning_rate": 0.00015040757527830578,
      "loss": 2.0126,
      "step": 153487
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.6737483739852905,
      "learning_rate": 0.00015040402955871125,
      "loss": 1.9963,
      "step": 153488
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.2751195430755615,
      "learning_rate": 0.0001504004838669293,
      "loss": 2.0934,
      "step": 153489
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.2905360460281372,
      "learning_rate": 0.00015039693820296084,
      "loss": 2.1872,
      "step": 153490
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.426072120666504,
      "learning_rate": 0.0001503933925668063,
      "loss": 2.1311,
      "step": 153491
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.2213287353515625,
      "learning_rate": 0.0001503898469584667,
      "loss": 2.2668,
      "step": 153492
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.219771385192871,
      "learning_rate": 0.00015038630137794218,
      "loss": 2.1859,
      "step": 153493
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.316494345664978,
      "learning_rate": 0.00015038275582523382,
      "loss": 2.1618,
      "step": 153494
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.1442447900772095,
      "learning_rate": 0.000150379210300342,
      "loss": 1.9746,
      "step": 153495
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.2562999725341797,
      "learning_rate": 0.00015037566480326762,
      "loss": 1.9492,
      "step": 153496
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.2032220363616943,
      "learning_rate": 0.0001503721193340111,
      "loss": 1.9944,
      "step": 153497
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.1317994594573975,
      "learning_rate": 0.0001503685738925734,
      "loss": 2.0993,
      "step": 153498
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.0973711013793945,
      "learning_rate": 0.00015036502847895494,
      "loss": 2.0946,
      "step": 153499
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.1736259460449219,
      "learning_rate": 0.0001503614830931565,
      "loss": 2.063,
      "step": 153500
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.2420358657836914,
      "learning_rate": 0.0001503579377351785,
      "loss": 2.2648,
      "step": 153501
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.192039966583252,
      "learning_rate": 0.00015035439240502197,
      "loss": 2.2897,
      "step": 153502
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.1201118230819702,
      "learning_rate": 0.00015035084710268725,
      "loss": 2.096,
      "step": 153503
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.28070068359375,
      "learning_rate": 0.00015034730182817525,
      "loss": 2.0562,
      "step": 153504
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.195454478263855,
      "learning_rate": 0.0001503437565814865,
      "loss": 1.9819,
      "step": 153505
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.2249469757080078,
      "learning_rate": 0.00015034021136262166,
      "loss": 1.9566,
      "step": 153506
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.1937521696090698,
      "learning_rate": 0.0001503366661715813,
      "loss": 2.2399,
      "step": 153507
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.2406530380249023,
      "learning_rate": 0.0001503331210083663,
      "loss": 1.9582,
      "step": 153508
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.1153671741485596,
      "learning_rate": 0.00015032957587297706,
      "loss": 1.908,
      "step": 153509
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.1258032321929932,
      "learning_rate": 0.00015032603076541446,
      "loss": 2.2812,
      "step": 153510
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.032801628112793,
      "learning_rate": 0.00015032248568567916,
      "loss": 2.0782,
      "step": 153511
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.2647864818572998,
      "learning_rate": 0.0001503189406337716,
      "loss": 2.0266,
      "step": 153512
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.1066144704818726,
      "learning_rate": 0.00015031539560969267,
      "loss": 2.2328,
      "step": 153513
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.0811578035354614,
      "learning_rate": 0.00015031185061344291,
      "loss": 2.1262,
      "step": 153514
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.2484794855117798,
      "learning_rate": 0.0001503083056450229,
      "loss": 2.3235,
      "step": 153515
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.089115858078003,
      "learning_rate": 0.00015030476070443356,
      "loss": 2.0404,
      "step": 153516
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.1875391006469727,
      "learning_rate": 0.00015030121579167536,
      "loss": 2.0364,
      "step": 153517
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.1042256355285645,
      "learning_rate": 0.00015029767090674886,
      "loss": 2.2007,
      "step": 153518
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.1463497877120972,
      "learning_rate": 0.00015029412604965496,
      "loss": 1.9279,
      "step": 153519
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.1509792804718018,
      "learning_rate": 0.00015029058122039424,
      "loss": 2.2314,
      "step": 153520
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.2417024374008179,
      "learning_rate": 0.00015028703641896719,
      "loss": 2.1607,
      "step": 153521
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.3108160495758057,
      "learning_rate": 0.00015028349164537472,
      "loss": 2.1844,
      "step": 153522
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.2227791547775269,
      "learning_rate": 0.0001502799468996174,
      "loss": 1.9418,
      "step": 153523
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.1362392902374268,
      "learning_rate": 0.00015027640218169574,
      "loss": 1.9807,
      "step": 153524
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.1890507936477661,
      "learning_rate": 0.0001502728574916106,
      "loss": 2.1473,
      "step": 153525
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.2163845300674438,
      "learning_rate": 0.0001502693128293625,
      "loss": 2.0957,
      "step": 153526
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.1579651832580566,
      "learning_rate": 0.00015026576819495224,
      "loss": 1.9362,
      "step": 153527
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.1950358152389526,
      "learning_rate": 0.00015026222358838044,
      "loss": 2.142,
      "step": 153528
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.194067358970642,
      "learning_rate": 0.0001502586790096477,
      "loss": 2.3453,
      "step": 153529
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.1264833211898804,
      "learning_rate": 0.00015025513445875456,
      "loss": 1.9567,
      "step": 153530
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.1860630512237549,
      "learning_rate": 0.00015025158993570191,
      "loss": 2.1053,
      "step": 153531
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.1835263967514038,
      "learning_rate": 0.00015024804544049025,
      "loss": 2.2073,
      "step": 153532
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.1611056327819824,
      "learning_rate": 0.00015024450097312044,
      "loss": 2.1677,
      "step": 153533
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.145067572593689,
      "learning_rate": 0.00015024095653359297,
      "loss": 2.1217,
      "step": 153534
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.1268665790557861,
      "learning_rate": 0.00015023741212190854,
      "loss": 2.3537,
      "step": 153535
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.0916101932525635,
      "learning_rate": 0.00015023386773806768,
      "loss": 1.9262,
      "step": 153536
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.1502923965454102,
      "learning_rate": 0.0001502303233820713,
      "loss": 2.0111,
      "step": 153537
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.1729439496994019,
      "learning_rate": 0.00015022677905391979,
      "loss": 2.0723,
      "step": 153538
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.1310594081878662,
      "learning_rate": 0.00015022323475361406,
      "loss": 1.9515,
      "step": 153539
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.1596564054489136,
      "learning_rate": 0.0001502196904811546,
      "loss": 2.268,
      "step": 153540
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.1088842153549194,
      "learning_rate": 0.00015021614623654233,
      "loss": 2.2103,
      "step": 153541
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.150589108467102,
      "learning_rate": 0.00015021260201977744,
      "loss": 1.9986,
      "step": 153542
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.216599941253662,
      "learning_rate": 0.000150209057830861,
      "loss": 2.101,
      "step": 153543
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.3370896577835083,
      "learning_rate": 0.0001502055136697934,
      "loss": 1.9659,
      "step": 153544
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.1484678983688354,
      "learning_rate": 0.00015020196953657552,
      "loss": 2.0049,
      "step": 153545
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.0391483306884766,
      "learning_rate": 0.00015019842543120785,
      "loss": 2.1411,
      "step": 153546
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.1032812595367432,
      "learning_rate": 0.00015019488135369131,
      "loss": 1.9927,
      "step": 153547
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.2140817642211914,
      "learning_rate": 0.00015019133730402614,
      "loss": 1.9541,
      "step": 153548
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.249061107635498,
      "learning_rate": 0.0001501877932822134,
      "loss": 2.1807,
      "step": 153549
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.1165952682495117,
      "learning_rate": 0.00015018424928825344,
      "loss": 1.9457,
      "step": 153550
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.1682308912277222,
      "learning_rate": 0.00015018070532214713,
      "loss": 2.0728,
      "step": 153551
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.0978177785873413,
      "learning_rate": 0.00015017716138389497,
      "loss": 1.8393,
      "step": 153552
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.1161539554595947,
      "learning_rate": 0.00015017361747349792,
      "loss": 2.0733,
      "step": 153553
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.115235686302185,
      "learning_rate": 0.0001501700735909562,
      "loss": 1.9413,
      "step": 153554
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.0994246006011963,
      "learning_rate": 0.0001501665297362708,
      "loss": 2.0991,
      "step": 153555
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.1592568159103394,
      "learning_rate": 0.00015016298590944216,
      "loss": 1.95,
      "step": 153556
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.2456879615783691,
      "learning_rate": 0.0001501594421104712,
      "loss": 2.0821,
      "step": 153557
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.106142282485962,
      "learning_rate": 0.00015015589833935824,
      "loss": 1.9538,
      "step": 153558
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.1113468408584595,
      "learning_rate": 0.0001501523545961044,
      "loss": 1.9561,
      "step": 153559
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.0371549129486084,
      "learning_rate": 0.00015014881088070984,
      "loss": 2.1674,
      "step": 153560
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.094647765159607,
      "learning_rate": 0.00015014526719317556,
      "loss": 2.1471,
      "step": 153561
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.1161930561065674,
      "learning_rate": 0.00015014172353350195,
      "loss": 2.0281,
      "step": 153562
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.077893853187561,
      "learning_rate": 0.00015013817990168997,
      "loss": 1.9069,
      "step": 153563
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.127791166305542,
      "learning_rate": 0.00015013463629774004,
      "loss": 2.0619,
      "step": 153564
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.0544682741165161,
      "learning_rate": 0.00015013109272165313,
      "loss": 2.1608,
      "step": 153565
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.0315546989440918,
      "learning_rate": 0.00015012754917342942,
      "loss": 2.1005,
      "step": 153566
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.1760790348052979,
      "learning_rate": 0.00015012400565306993,
      "loss": 2.0183,
      "step": 153567
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.230446219444275,
      "learning_rate": 0.00015012046216057514,
      "loss": 2.2704,
      "step": 153568
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.1359636783599854,
      "learning_rate": 0.0001501169186959459,
      "loss": 2.154,
      "step": 153569
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.2503331899642944,
      "learning_rate": 0.00015011337525918262,
      "loss": 1.9616,
      "step": 153570
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.158118724822998,
      "learning_rate": 0.00015010983185028635,
      "loss": 2.2925,
      "step": 153571
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.1215994358062744,
      "learning_rate": 0.00015010628846925722,
      "loss": 1.8274,
      "step": 153572
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.0774691104888916,
      "learning_rate": 0.00015010274511609629,
      "loss": 1.8705,
      "step": 153573
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.082869052886963,
      "learning_rate": 0.00015009920179080396,
      "loss": 1.8708,
      "step": 153574
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.1897988319396973,
      "learning_rate": 0.00015009565849338115,
      "loss": 1.9041,
      "step": 153575
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.1173962354660034,
      "learning_rate": 0.0001500921152238283,
      "loss": 1.8512,
      "step": 153576
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.1297698020935059,
      "learning_rate": 0.00015008857198214637,
      "loss": 1.9956,
      "step": 153577
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.2599910497665405,
      "learning_rate": 0.00015008502876833556,
      "loss": 1.8333,
      "step": 153578
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.1087327003479004,
      "learning_rate": 0.00015008148558239687,
      "loss": 2.1439,
      "step": 153579
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.1336034536361694,
      "learning_rate": 0.0001500779424243308,
      "loss": 2.1334,
      "step": 153580
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.342043161392212,
      "learning_rate": 0.00015007439929413816,
      "loss": 2.1286,
      "step": 153581
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.1522122621536255,
      "learning_rate": 0.00015007085619181938,
      "loss": 1.7881,
      "step": 153582
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.1193562746047974,
      "learning_rate": 0.00015006731311737552,
      "loss": 2.0531,
      "step": 153583
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.1762694120407104,
      "learning_rate": 0.0001500637700708067,
      "loss": 1.9722,
      "step": 153584
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.6900120973587036,
      "learning_rate": 0.00015006022705211402,
      "loss": 2.4445,
      "step": 153585
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.0841147899627686,
      "learning_rate": 0.00015005668406129787,
      "loss": 2.1816,
      "step": 153586
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.0255590677261353,
      "learning_rate": 0.0001500531410983591,
      "loss": 2.0421,
      "step": 153587
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.167224407196045,
      "learning_rate": 0.00015004959816329816,
      "loss": 1.9228,
      "step": 153588
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.128424048423767,
      "learning_rate": 0.00015004605525611596,
      "loss": 2.1017,
      "step": 153589
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.1260387897491455,
      "learning_rate": 0.00015004251237681304,
      "loss": 1.9591,
      "step": 153590
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.2396821975708008,
      "learning_rate": 0.00015003896952539,
      "loss": 1.8057,
      "step": 153591
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.2409921884536743,
      "learning_rate": 0.00015003542670184746,
      "loss": 2.0651,
      "step": 153592
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.1078107357025146,
      "learning_rate": 0.0001500318839061863,
      "loss": 1.8916,
      "step": 153593
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.1614898443222046,
      "learning_rate": 0.0001500283411384069,
      "loss": 2.1114,
      "step": 153594
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.100901484489441,
      "learning_rate": 0.00015002479839851019,
      "loss": 2.1502,
      "step": 153595
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.2919291257858276,
      "learning_rate": 0.0001500212556864967,
      "loss": 2.2167,
      "step": 153596
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.2381852865219116,
      "learning_rate": 0.00015001771300236698,
      "loss": 1.9733,
      "step": 153597
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.124624252319336,
      "learning_rate": 0.00015001417034612188,
      "loss": 1.9585,
      "step": 153598
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.2088650465011597,
      "learning_rate": 0.00015001062771776203,
      "loss": 1.9493,
      "step": 153599
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.243162751197815,
      "learning_rate": 0.00015000708511728788,
      "loss": 1.8108,
      "step": 153600
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.1135320663452148,
      "learning_rate": 0.00015000354254470036,
      "loss": 2.1567,
      "step": 153601
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.4610968828201294,
      "learning_rate": 0.00015000000000000004,
      "loss": 1.6754,
      "step": 153602
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.1269882917404175,
      "learning_rate": 0.00014999645748318743,
      "loss": 2.0329,
      "step": 153603
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.1262472867965698,
      "learning_rate": 0.00014999291499426343,
      "loss": 2.0276,
      "step": 153604
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.1961473226547241,
      "learning_rate": 0.00014998937253322855,
      "loss": 1.9954,
      "step": 153605
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.1213626861572266,
      "learning_rate": 0.0001499858301000834,
      "loss": 2.1441,
      "step": 153606
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.2239792346954346,
      "learning_rate": 0.0001499822876948288,
      "loss": 1.9842,
      "step": 153607
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.139235258102417,
      "learning_rate": 0.00014997874531746532,
      "loss": 2.0916,
      "step": 153608
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.1804943084716797,
      "learning_rate": 0.00014997520296799354,
      "loss": 1.9359,
      "step": 153609
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.5049965381622314,
      "learning_rate": 0.00014997166064641427,
      "loss": 1.9428,
      "step": 153610
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.079288363456726,
      "learning_rate": 0.00014996811835272803,
      "loss": 1.9915,
      "step": 153611
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.1723291873931885,
      "learning_rate": 0.00014996457608693563,
      "loss": 1.9817,
      "step": 153612
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.07205331325531,
      "learning_rate": 0.00014996103384903766,
      "loss": 2.0173,
      "step": 153613
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.1784098148345947,
      "learning_rate": 0.00014995749163903478,
      "loss": 2.1636,
      "step": 153614
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.1491144895553589,
      "learning_rate": 0.00014995394945692753,
      "loss": 2.3347,
      "step": 153615
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.1643372774124146,
      "learning_rate": 0.00014995040730271677,
      "loss": 1.9948,
      "step": 153616
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.361291766166687,
      "learning_rate": 0.00014994686517640296,
      "loss": 1.9694,
      "step": 153617
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.3326895236968994,
      "learning_rate": 0.00014994332307798695,
      "loss": 2.1746,
      "step": 153618
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.1227580308914185,
      "learning_rate": 0.0001499397810074693,
      "loss": 1.9296,
      "step": 153619
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.0757547616958618,
      "learning_rate": 0.0001499362389648507,
      "loss": 1.9867,
      "step": 153620
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.2531461715698242,
      "learning_rate": 0.0001499326969501317,
      "loss": 2.2248,
      "step": 153621
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.4321908950805664,
      "learning_rate": 0.0001499291549633131,
      "loss": 2.0362,
      "step": 153622
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.1466456651687622,
      "learning_rate": 0.0001499256130043954,
      "loss": 2.2212,
      "step": 153623
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.1634430885314941,
      "learning_rate": 0.00014992207107337953,
      "loss": 1.9964,
      "step": 153624
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.0885908603668213,
      "learning_rate": 0.00014991852917026582,
      "loss": 1.9594,
      "step": 153625
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.184303641319275,
      "learning_rate": 0.0001499149872950553,
      "loss": 2.0196,
      "step": 153626
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.1736575365066528,
      "learning_rate": 0.00014991144544774824,
      "loss": 2.328,
      "step": 153627
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.0882354974746704,
      "learning_rate": 0.00014990790362834556,
      "loss": 1.8947,
      "step": 153628
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.1441961526870728,
      "learning_rate": 0.0001499043618368477,
      "loss": 1.9511,
      "step": 153629
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.2133375406265259,
      "learning_rate": 0.0001499008200732556,
      "loss": 1.9455,
      "step": 153630
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.3013807535171509,
      "learning_rate": 0.00014989727833756965,
      "loss": 2.0307,
      "step": 153631
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.298071265220642,
      "learning_rate": 0.00014989373662979084,
      "loss": 2.0155,
      "step": 153632
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.2988530397415161,
      "learning_rate": 0.0001498901949499194,
      "loss": 2.0947,
      "step": 153633
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.1968928575515747,
      "learning_rate": 0.00014988665329795635,
      "loss": 2.1372,
      "step": 153634
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.11270010471344,
      "learning_rate": 0.00014988311167390205,
      "loss": 2.2651,
      "step": 153635
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.7577805519104004,
      "learning_rate": 0.00014987957007775744,
      "loss": 1.7267,
      "step": 153636
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.2097141742706299,
      "learning_rate": 0.0001498760285095229,
      "loss": 1.9226,
      "step": 153637
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.131408929824829,
      "learning_rate": 0.00014987248696919953,
      "loss": 2.1413,
      "step": 153638
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.1926498413085938,
      "learning_rate": 0.00014986894545678745,
      "loss": 1.9964,
      "step": 153639
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.2589192390441895,
      "learning_rate": 0.0001498654039722877,
      "loss": 2.0351,
      "step": 153640
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.5447763204574585,
      "learning_rate": 0.00014986186251570065,
      "loss": 2.0147,
      "step": 153641
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.2282605171203613,
      "learning_rate": 0.00014985832108702725,
      "loss": 1.8455,
      "step": 153642
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.275292992591858,
      "learning_rate": 0.0001498547796862679,
      "loss": 2.0734,
      "step": 153643
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.3231934309005737,
      "learning_rate": 0.0001498512383134236,
      "loss": 2.0175,
      "step": 153644
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.293742299079895,
      "learning_rate": 0.0001498476969684946,
      "loss": 2.2637,
      "step": 153645
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.164232850074768,
      "learning_rate": 0.00014984415565148183,
      "loss": 2.0625,
      "step": 153646
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.1951812505722046,
      "learning_rate": 0.00014984061436238577,
      "loss": 1.9881,
      "step": 153647
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.1603525876998901,
      "learning_rate": 0.00014983707310120727,
      "loss": 2.0728,
      "step": 153648
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.390236258506775,
      "learning_rate": 0.00014983353186794678,
      "loss": 2.1131,
      "step": 153649
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.3870617151260376,
      "learning_rate": 0.0001498299906626053,
      "loss": 1.9356,
      "step": 153650
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.1436362266540527,
      "learning_rate": 0.00014982644948518305,
      "loss": 2.2737,
      "step": 153651
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.1678262948989868,
      "learning_rate": 0.00014982290833568098,
      "loss": 2.0763,
      "step": 153652
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.2894490957260132,
      "learning_rate": 0.0001498193672140996,
      "loss": 1.9494,
      "step": 153653
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.4850727319717407,
      "learning_rate": 0.0001498158261204397,
      "loss": 2.2508,
      "step": 153654
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.4427298307418823,
      "learning_rate": 0.00014981228505470178,
      "loss": 1.9932,
      "step": 153655
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.1382246017456055,
      "learning_rate": 0.00014980874401688684,
      "loss": 2.1272,
      "step": 153656
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.1828055381774902,
      "learning_rate": 0.000149805203006995,
      "loss": 2.1886,
      "step": 153657
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.1710889339447021,
      "learning_rate": 0.00014980166202502738,
      "loss": 1.9025,
      "step": 153658
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.0923199653625488,
      "learning_rate": 0.00014979812107098432,
      "loss": 2.118,
      "step": 153659
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.3418611288070679,
      "learning_rate": 0.00014979458014486674,
      "loss": 2.1714,
      "step": 153660
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.1104322671890259,
      "learning_rate": 0.00014979103924667509,
      "loss": 1.951,
      "step": 153661
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.2585573196411133,
      "learning_rate": 0.00014978749837641034,
      "loss": 2.1755,
      "step": 153662
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.1143109798431396,
      "learning_rate": 0.0001497839575340727,
      "loss": 1.9341,
      "step": 153663
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.16260826587677,
      "learning_rate": 0.0001497804167196632,
      "loss": 1.7999,
      "step": 153664
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.2120529413223267,
      "learning_rate": 0.0001497768759331822,
      "loss": 1.9534,
      "step": 153665
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.0971993207931519,
      "learning_rate": 0.00014977333517463066,
      "loss": 2.151,
      "step": 153666
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.1210863590240479,
      "learning_rate": 0.0001497697944440089,
      "loss": 1.9592,
      "step": 153667
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.1098898649215698,
      "learning_rate": 0.00014976625374131812,
      "loss": 2.0657,
      "step": 153668
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.2105544805526733,
      "learning_rate": 0.00014976271306655834,
      "loss": 2.0623,
      "step": 153669
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.1666827201843262,
      "learning_rate": 0.0001497591724197306,
      "loss": 2.2786,
      "step": 153670
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.085768461227417,
      "learning_rate": 0.00014975563180083537,
      "loss": 2.027,
      "step": 153671
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.1086970567703247,
      "learning_rate": 0.00014975209120987351,
      "loss": 1.9533,
      "step": 153672
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.0845004320144653,
      "learning_rate": 0.00014974855064684548,
      "loss": 2.0189,
      "step": 153673
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.1781388521194458,
      "learning_rate": 0.00014974501011175216,
      "loss": 1.9783,
      "step": 153674
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.1271103620529175,
      "learning_rate": 0.00014974146960459404,
      "loss": 2.0043,
      "step": 153675
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.1573022603988647,
      "learning_rate": 0.00014973792912537185,
      "loss": 1.9728,
      "step": 153676
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.257852554321289,
      "learning_rate": 0.00014973438867408605,
      "loss": 2.2795,
      "step": 153677
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.1461243629455566,
      "learning_rate": 0.00014973084825073763,
      "loss": 2.0473,
      "step": 153678
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.209452509880066,
      "learning_rate": 0.00014972730785532695,
      "loss": 1.9407,
      "step": 153679
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.12416410446167,
      "learning_rate": 0.00014972376748785492,
      "loss": 1.8549,
      "step": 153680
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.1274185180664062,
      "learning_rate": 0.00014972022714832207,
      "loss": 1.895,
      "step": 153681
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.1318683624267578,
      "learning_rate": 0.00014971668683672905,
      "loss": 1.8147,
      "step": 153682
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.2421759366989136,
      "learning_rate": 0.00014971314655307645,
      "loss": 2.0302,
      "step": 153683
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.0704736709594727,
      "learning_rate": 0.00014970960629736514,
      "loss": 1.761,
      "step": 153684
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.1609231233596802,
      "learning_rate": 0.00014970606606959548,
      "loss": 2.0611,
      "step": 153685
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.2836726903915405,
      "learning_rate": 0.00014970252586976845,
      "loss": 2.0445,
      "step": 153686
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.4288036823272705,
      "learning_rate": 0.00014969898569788456,
      "loss": 1.9616,
      "step": 153687
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.1912857294082642,
      "learning_rate": 0.00014969544555394436,
      "loss": 1.7775,
      "step": 153688
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.1959508657455444,
      "learning_rate": 0.0001496919054379487,
      "loss": 2.1546,
      "step": 153689
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.0518529415130615,
      "learning_rate": 0.00014968836534989818,
      "loss": 2.0999,
      "step": 153690
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.166511058807373,
      "learning_rate": 0.0001496848252897933,
      "loss": 2.0997,
      "step": 153691
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.2332288026809692,
      "learning_rate": 0.00014968128525763493,
      "loss": 1.9565,
      "step": 153692
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.1541892290115356,
      "learning_rate": 0.00014967774525342372,
      "loss": 2.3067,
      "step": 153693
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.0531290769577026,
      "learning_rate": 0.00014967420527716007,
      "loss": 2.0337,
      "step": 153694
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.129615306854248,
      "learning_rate": 0.00014967066532884496,
      "loss": 1.9633,
      "step": 153695
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.1975451707839966,
      "learning_rate": 0.00014966712540847885,
      "loss": 2.0882,
      "step": 153696
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.1809382438659668,
      "learning_rate": 0.00014966358551606252,
      "loss": 2.1936,
      "step": 153697
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.146999478340149,
      "learning_rate": 0.00014966004565159657,
      "loss": 2.0537,
      "step": 153698
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.0076888799667358,
      "learning_rate": 0.00014965650581508165,
      "loss": 1.9301,
      "step": 153699
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.1070702075958252,
      "learning_rate": 0.0001496529660065183,
      "loss": 1.9494,
      "step": 153700
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.3153362274169922,
      "learning_rate": 0.00014964942622590745,
      "loss": 1.8748,
      "step": 153701
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.2068812847137451,
      "learning_rate": 0.0001496458864732495,
      "loss": 2.1861,
      "step": 153702
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.1418966054916382,
      "learning_rate": 0.0001496423467485453,
      "loss": 2.0598,
      "step": 153703
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.1649245023727417,
      "learning_rate": 0.00014963880705179543,
      "loss": 2.0674,
      "step": 153704
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.2234803438186646,
      "learning_rate": 0.00014963526738300057,
      "loss": 2.0738,
      "step": 153705
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.0686817169189453,
      "learning_rate": 0.00014963172774216124,
      "loss": 1.9045,
      "step": 153706
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.0879290103912354,
      "learning_rate": 0.0001496281881292783,
      "loss": 2.1357,
      "step": 153707
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.14008629322052,
      "learning_rate": 0.0001496246485443522,
      "loss": 1.7609,
      "step": 153708
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.1937477588653564,
      "learning_rate": 0.00014962110898738385,
      "loss": 2.2277,
      "step": 153709
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.069307804107666,
      "learning_rate": 0.00014961756945837366,
      "loss": 2.0191,
      "step": 153710
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.1790715456008911,
      "learning_rate": 0.0001496140299573226,
      "loss": 1.8244,
      "step": 153711
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.122212529182434,
      "learning_rate": 0.00014961049048423092,
      "loss": 2.1492,
      "step": 153712
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.1636998653411865,
      "learning_rate": 0.00014960695103909962,
      "loss": 2.3041,
      "step": 153713
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.071863055229187,
      "learning_rate": 0.00014960341162192908,
      "loss": 2.1801,
      "step": 153714
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.1130621433258057,
      "learning_rate": 0.00014959987223272028,
      "loss": 2.1451,
      "step": 153715
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.130379319190979,
      "learning_rate": 0.00014959633287147352,
      "loss": 1.9341,
      "step": 153716
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.1234432458877563,
      "learning_rate": 0.0001495927935381899,
      "loss": 2.1546,
      "step": 153717
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.1679941415786743,
      "learning_rate": 0.00014958925423286959,
      "loss": 2.1322,
      "step": 153718
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.2063872814178467,
      "learning_rate": 0.0001495857149555136,
      "loss": 2.1732,
      "step": 153719
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.1781607866287231,
      "learning_rate": 0.00014958217570612233,
      "loss": 2.0218,
      "step": 153720
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.1724746227264404,
      "learning_rate": 0.0001495786364846967,
      "loss": 1.8733,
      "step": 153721
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.1979557275772095,
      "learning_rate": 0.00014957509729123712,
      "loss": 1.9669,
      "step": 153722
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.3222651481628418,
      "learning_rate": 0.0001495715581257446,
      "loss": 2.2742,
      "step": 153723
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.1291589736938477,
      "learning_rate": 0.00014956801898821935,
      "loss": 1.9227,
      "step": 153724
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.0571038722991943,
      "learning_rate": 0.00014956447987866234,
      "loss": 2.0415,
      "step": 153725
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.1501473188400269,
      "learning_rate": 0.000149560940797074,
      "loss": 1.9996,
      "step": 153726
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.190860629081726,
      "learning_rate": 0.0001495574017434553,
      "loss": 1.8713,
      "step": 153727
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.1178956031799316,
      "learning_rate": 0.00014955386271780656,
      "loss": 2.0426,
      "step": 153728
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.2642569541931152,
      "learning_rate": 0.0001495503237201288,
      "loss": 1.9515,
      "step": 153729
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.1737470626831055,
      "learning_rate": 0.0001495467847504223,
      "loss": 2.1134,
      "step": 153730
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.0992094278335571,
      "learning_rate": 0.000149543245808688,
      "loss": 2.0553,
      "step": 153731
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.1152501106262207,
      "learning_rate": 0.0001495397068949263,
      "loss": 2.0187,
      "step": 153732
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.1223759651184082,
      "learning_rate": 0.00014953616800913815,
      "loss": 2.1746,
      "step": 153733
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.261608600616455,
      "learning_rate": 0.00014953262915132397,
      "loss": 2.144,
      "step": 153734
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.1216908693313599,
      "learning_rate": 0.0001495290903214847,
      "loss": 2.1749,
      "step": 153735
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.2695791721343994,
      "learning_rate": 0.00014952555151962062,
      "loss": 2.043,
      "step": 153736
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.1954714059829712,
      "learning_rate": 0.00014952201274573266,
      "loss": 2.2426,
      "step": 153737
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.2279216051101685,
      "learning_rate": 0.0001495184739998213,
      "loss": 2.1122,
      "step": 153738
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.1378518342971802,
      "learning_rate": 0.0001495149352818874,
      "loss": 2.0352,
      "step": 153739
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.2452458143234253,
      "learning_rate": 0.0001495113965919314,
      "loss": 2.1398,
      "step": 153740
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.3898733854293823,
      "learning_rate": 0.00014950785792995424,
      "loss": 1.9982,
      "step": 153741
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.8078426122665405,
      "learning_rate": 0.00014950431929595633,
      "loss": 2.0679,
      "step": 153742
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.1700531244277954,
      "learning_rate": 0.00014950078068993848,
      "loss": 2.0306,
      "step": 153743
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.2063336372375488,
      "learning_rate": 0.00014949724211190112,
      "loss": 1.9091,
      "step": 153744
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.5529834032058716,
      "learning_rate": 0.00014949370356184518,
      "loss": 1.7766,
      "step": 153745
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.4114183187484741,
      "learning_rate": 0.00014949016503977108,
      "loss": 2.044,
      "step": 153746
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.1084563732147217,
      "learning_rate": 0.00014948662654567973,
      "loss": 2.1664,
      "step": 153747
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.1956866979599,
      "learning_rate": 0.00014948308807957166,
      "loss": 2.1975,
      "step": 153748
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.111560583114624,
      "learning_rate": 0.00014947954964144753,
      "loss": 1.8448,
      "step": 153749
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.5943468809127808,
      "learning_rate": 0.00014947601123130788,
      "loss": 2.006,
      "step": 153750
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.4100279808044434,
      "learning_rate": 0.00014947247284915354,
      "loss": 2.1352,
      "step": 153751
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.189207911491394,
      "learning_rate": 0.00014946893449498507,
      "loss": 2.0012,
      "step": 153752
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.194888710975647,
      "learning_rate": 0.00014946539616880324,
      "loss": 1.9861,
      "step": 153753
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.5841803550720215,
      "learning_rate": 0.00014946185787060863,
      "loss": 2.1462,
      "step": 153754
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.3184047937393188,
      "learning_rate": 0.00014945831960040193,
      "loss": 1.9897,
      "step": 153755
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.202091932296753,
      "learning_rate": 0.00014945478135818362,
      "loss": 1.9357,
      "step": 153756
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.142764925956726,
      "learning_rate": 0.00014945124314395467,
      "loss": 2.0096,
      "step": 153757
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.208809494972229,
      "learning_rate": 0.00014944770495771542,
      "loss": 2.3636,
      "step": 153758
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.1921257972717285,
      "learning_rate": 0.0001494441667994668,
      "loss": 2.022,
      "step": 153759
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.228547215461731,
      "learning_rate": 0.00014944062866920936,
      "loss": 1.8943,
      "step": 153760
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.3802040815353394,
      "learning_rate": 0.00014943709056694377,
      "loss": 2.0917,
      "step": 153761
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.1361658573150635,
      "learning_rate": 0.00014943355249267054,
      "loss": 2.1762,
      "step": 153762
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.1846529245376587,
      "learning_rate": 0.00014943001444639057,
      "loss": 2.2871,
      "step": 153763
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.3116272687911987,
      "learning_rate": 0.0001494264764281043,
      "loss": 2.0872,
      "step": 153764
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.175603985786438,
      "learning_rate": 0.00014942293843781256,
      "loss": 1.9515,
      "step": 153765
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.1757456064224243,
      "learning_rate": 0.00014941940047551595,
      "loss": 1.8838,
      "step": 153766
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.1747688055038452,
      "learning_rate": 0.00014941586254121515,
      "loss": 2.1305,
      "step": 153767
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.0481934547424316,
      "learning_rate": 0.00014941232463491064,
      "loss": 1.8425,
      "step": 153768
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.195794939994812,
      "learning_rate": 0.00014940878675660338,
      "loss": 1.9745,
      "step": 153769
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.2136552333831787,
      "learning_rate": 0.00014940524890629368,
      "loss": 1.999,
      "step": 153770
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.127260684967041,
      "learning_rate": 0.00014940171108398256,
      "loss": 2.1855,
      "step": 153771
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.1370223760604858,
      "learning_rate": 0.00014939817328967048,
      "loss": 2.0224,
      "step": 153772
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.0549647808074951,
      "learning_rate": 0.00014939463552335802,
      "loss": 1.9372,
      "step": 153773
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.1447631120681763,
      "learning_rate": 0.00014939109778504606,
      "loss": 1.9122,
      "step": 153774
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.0923352241516113,
      "learning_rate": 0.00014938756007473514,
      "loss": 2.1122,
      "step": 153775
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.1654456853866577,
      "learning_rate": 0.00014938402239242577,
      "loss": 2.0751,
      "step": 153776
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.0636394023895264,
      "learning_rate": 0.0001493804847381189,
      "loss": 1.9429,
      "step": 153777
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.6707977056503296,
      "learning_rate": 0.000149376947111815,
      "loss": 2.181,
      "step": 153778
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.2462056875228882,
      "learning_rate": 0.00014937340951351471,
      "loss": 2.2278,
      "step": 153779
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.1158093214035034,
      "learning_rate": 0.0001493698719432188,
      "loss": 2.0199,
      "step": 153780
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.2814443111419678,
      "learning_rate": 0.00014936633440092793,
      "loss": 2.1266,
      "step": 153781
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.2886450290679932,
      "learning_rate": 0.00014936279688664258,
      "loss": 2.2328,
      "step": 153782
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.2858388423919678,
      "learning_rate": 0.0001493592594003636,
      "loss": 2.2111,
      "step": 153783
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.2839466333389282,
      "learning_rate": 0.00014935572194209163,
      "loss": 2.0875,
      "step": 153784
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.1835626363754272,
      "learning_rate": 0.00014935218451182712,
      "loss": 2.0707,
      "step": 153785
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.192584753036499,
      "learning_rate": 0.000149348647109571,
      "loss": 2.0472,
      "step": 153786
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.1803219318389893,
      "learning_rate": 0.00014934510973532372,
      "loss": 2.2057,
      "step": 153787
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.2090201377868652,
      "learning_rate": 0.00014934157238908613,
      "loss": 2.091,
      "step": 153788
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.0181527137756348,
      "learning_rate": 0.0001493380350708588,
      "loss": 2.1497,
      "step": 153789
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.2226530313491821,
      "learning_rate": 0.00014933449778064235,
      "loss": 2.0313,
      "step": 153790
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.0367358922958374,
      "learning_rate": 0.00014933096051843734,
      "loss": 2.1991,
      "step": 153791
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.181419014930725,
      "learning_rate": 0.0001493274232842447,
      "loss": 2.2552,
      "step": 153792
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.1024894714355469,
      "learning_rate": 0.0001493238860780648,
      "loss": 2.0657,
      "step": 153793
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.2495429515838623,
      "learning_rate": 0.00014932034889989857,
      "loss": 2.0913,
      "step": 153794
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.1416531801223755,
      "learning_rate": 0.0001493168117497464,
      "loss": 2.1503,
      "step": 153795
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.3414801359176636,
      "learning_rate": 0.00014931327462760927,
      "loss": 1.6719,
      "step": 153796
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.1775943040847778,
      "learning_rate": 0.00014930973753348747,
      "loss": 2.1288,
      "step": 153797
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.1025954484939575,
      "learning_rate": 0.00014930620046738194,
      "loss": 2.0068,
      "step": 153798
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.2411093711853027,
      "learning_rate": 0.0001493026634292931,
      "loss": 2.1079,
      "step": 153799
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.1536407470703125,
      "learning_rate": 0.00014929912641922186,
      "loss": 2.0298,
      "step": 153800
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.2584112882614136,
      "learning_rate": 0.00014929558943716864,
      "loss": 2.1238,
      "step": 153801
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.1320470571517944,
      "learning_rate": 0.00014929205248313446,
      "loss": 1.9774,
      "step": 153802
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.192002296447754,
      "learning_rate": 0.00014928851555711948,
      "loss": 2.0629,
      "step": 153803
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.267635703086853,
      "learning_rate": 0.00014928497865912473,
      "loss": 2.1312,
      "step": 153804
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.1845118999481201,
      "learning_rate": 0.00014928144178915067,
      "loss": 2.1109,
      "step": 153805
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.1378775835037231,
      "learning_rate": 0.00014927790494719808,
      "loss": 2.0866,
      "step": 153806
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.1815623044967651,
      "learning_rate": 0.00014927436813326754,
      "loss": 2.0229,
      "step": 153807
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.1883578300476074,
      "learning_rate": 0.00014927083134735978,
      "loss": 2.019,
      "step": 153808
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.574903964996338,
      "learning_rate": 0.00014926729458947547,
      "loss": 1.9675,
      "step": 153809
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.2050315141677856,
      "learning_rate": 0.0001492637578596152,
      "loss": 1.9552,
      "step": 153810
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.0694427490234375,
      "learning_rate": 0.0001492602211577795,
      "loss": 1.978,
      "step": 153811
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.1284329891204834,
      "learning_rate": 0.00014925668448396932,
      "loss": 2.0261,
      "step": 153812
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.2597482204437256,
      "learning_rate": 0.00014925314783818504,
      "loss": 2.0914,
      "step": 153813
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.3062926530838013,
      "learning_rate": 0.00014924961122042758,
      "loss": 2.1024,
      "step": 153814
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.086484432220459,
      "learning_rate": 0.00014924607463069744,
      "loss": 2.1756,
      "step": 153815
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.1666386127471924,
      "learning_rate": 0.0001492425380689953,
      "loss": 1.9695,
      "step": 153816
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.3249874114990234,
      "learning_rate": 0.00014923900153532171,
      "loss": 2.1547,
      "step": 153817
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.2511357069015503,
      "learning_rate": 0.00014923546502967753,
      "loss": 2.1372,
      "step": 153818
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.351122498512268,
      "learning_rate": 0.00014923192855206322,
      "loss": 2.2109,
      "step": 153819
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.1681809425354004,
      "learning_rate": 0.00014922839210247966,
      "loss": 1.9652,
      "step": 153820
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.2208141088485718,
      "learning_rate": 0.00014922485568092742,
      "loss": 2.3995,
      "step": 153821
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.1827521324157715,
      "learning_rate": 0.00014922131928740706,
      "loss": 1.9173,
      "step": 153822
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.148188829421997,
      "learning_rate": 0.00014921778292191924,
      "loss": 2.1013,
      "step": 153823
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.2847291231155396,
      "learning_rate": 0.00014921424658446479,
      "loss": 1.9078,
      "step": 153824
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.226685881614685,
      "learning_rate": 0.00014921071027504412,
      "loss": 2.0395,
      "step": 153825
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.0750843286514282,
      "learning_rate": 0.00014920717399365813,
      "loss": 1.8424,
      "step": 153826
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.1719568967819214,
      "learning_rate": 0.00014920363774030744,
      "loss": 1.8695,
      "step": 153827
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.1282988786697388,
      "learning_rate": 0.00014920010151499256,
      "loss": 1.7928,
      "step": 153828
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.428159475326538,
      "learning_rate": 0.00014919656531771415,
      "loss": 2.1491,
      "step": 153829
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.1777489185333252,
      "learning_rate": 0.00014919302914847307,
      "loss": 2.1915,
      "step": 153830
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.1108819246292114,
      "learning_rate": 0.0001491894930072697,
      "loss": 1.8181,
      "step": 153831
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.202501893043518,
      "learning_rate": 0.000149185956894105,
      "loss": 2.0523,
      "step": 153832
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.657659888267517,
      "learning_rate": 0.00014918242080897944,
      "loss": 1.8726,
      "step": 153833
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.16102135181427,
      "learning_rate": 0.0001491788847518938,
      "loss": 1.8979,
      "step": 153834
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.1701453924179077,
      "learning_rate": 0.00014917534872284843,
      "loss": 2.0467,
      "step": 153835
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.0701990127563477,
      "learning_rate": 0.0001491718127218444,
      "loss": 1.8664,
      "step": 153836
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.148485779762268,
      "learning_rate": 0.00014916827674888204,
      "loss": 2.0335,
      "step": 153837
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.2443342208862305,
      "learning_rate": 0.00014916474080396225,
      "loss": 1.9793,
      "step": 153838
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.1887825727462769,
      "learning_rate": 0.00014916120488708557,
      "loss": 2.0534,
      "step": 153839
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.1658508777618408,
      "learning_rate": 0.00014915766899825268,
      "loss": 2.1013,
      "step": 153840
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.1254932880401611,
      "learning_rate": 0.00014915413313746414,
      "loss": 2.2217,
      "step": 153841
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.1448240280151367,
      "learning_rate": 0.0001491505973047208,
      "loss": 2.2977,
      "step": 153842
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.2163676023483276,
      "learning_rate": 0.00014914706150002306,
      "loss": 1.9973,
      "step": 153843
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.2083123922348022,
      "learning_rate": 0.00014914352572337186,
      "loss": 2.046,
      "step": 153844
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.326336145401001,
      "learning_rate": 0.00014913998997476773,
      "loss": 1.9793,
      "step": 153845
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.0815130472183228,
      "learning_rate": 0.00014913645425421133,
      "loss": 1.9212,
      "step": 153846
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.209125280380249,
      "learning_rate": 0.00014913291856170322,
      "loss": 2.1661,
      "step": 153847
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.1501426696777344,
      "learning_rate": 0.00014912938289724423,
      "loss": 2.072,
      "step": 153848
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.108651041984558,
      "learning_rate": 0.00014912584726083485,
      "loss": 1.8316,
      "step": 153849
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.1130106449127197,
      "learning_rate": 0.0001491223116524759,
      "loss": 2.1376,
      "step": 153850
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.0259530544281006,
      "learning_rate": 0.00014911877607216796,
      "loss": 2.1732,
      "step": 153851
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.1055936813354492,
      "learning_rate": 0.0001491152405199117,
      "loss": 1.9368,
      "step": 153852
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.1633076667785645,
      "learning_rate": 0.00014911170499570768,
      "loss": 2.1338,
      "step": 153853
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.2168142795562744,
      "learning_rate": 0.00014910816949955673,
      "loss": 2.1928,
      "step": 153854
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.232155442237854,
      "learning_rate": 0.00014910463403145928,
      "loss": 1.9548,
      "step": 153855
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.2456178665161133,
      "learning_rate": 0.0001491010985914163,
      "loss": 2.287,
      "step": 153856
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.1528745889663696,
      "learning_rate": 0.00014909756317942822,
      "loss": 1.9686,
      "step": 153857
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.147696614265442,
      "learning_rate": 0.00014909402779549565,
      "loss": 1.9264,
      "step": 153858
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.1583677530288696,
      "learning_rate": 0.00014909049243961946,
      "loss": 1.8305,
      "step": 153859
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.0426945686340332,
      "learning_rate": 0.00014908695711180023,
      "loss": 2.1002,
      "step": 153860
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.1929601430892944,
      "learning_rate": 0.00014908342181203845,
      "loss": 2.1217,
      "step": 153861
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.1761677265167236,
      "learning_rate": 0.00014907988654033503,
      "loss": 2.1341,
      "step": 153862
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.1890692710876465,
      "learning_rate": 0.0001490763512966905,
      "loss": 2.2332,
      "step": 153863
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.1862704753875732,
      "learning_rate": 0.0001490728160811054,
      "loss": 2.042,
      "step": 153864
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.158734679222107,
      "learning_rate": 0.00014906928089358066,
      "loss": 2.1586,
      "step": 153865
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.1252325773239136,
      "learning_rate": 0.00014906574573411675,
      "loss": 1.8728,
      "step": 153866
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.0958718061447144,
      "learning_rate": 0.00014906221060271427,
      "loss": 2.0699,
      "step": 153867
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.0879273414611816,
      "learning_rate": 0.00014905867549937408,
      "loss": 2.1786,
      "step": 153868
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.2358070611953735,
      "learning_rate": 0.0001490551404240966,
      "loss": 1.8874,
      "step": 153869
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.116192102432251,
      "learning_rate": 0.0001490516053768828,
      "loss": 2.1332,
      "step": 153870
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.2354305982589722,
      "learning_rate": 0.00014904807035773308,
      "loss": 2.1972,
      "step": 153871
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.1888477802276611,
      "learning_rate": 0.0001490445353666481,
      "loss": 2.1259,
      "step": 153872
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.0198609828948975,
      "learning_rate": 0.0001490410004036287,
      "loss": 1.9483,
      "step": 153873
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.0432319641113281,
      "learning_rate": 0.00014903746546867543,
      "loss": 1.7228,
      "step": 153874
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.115991234779358,
      "learning_rate": 0.00014903393056178883,
      "loss": 2.3052,
      "step": 153875
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.1807066202163696,
      "learning_rate": 0.0001490303956829698,
      "loss": 2.1048,
      "step": 153876
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.1208422183990479,
      "learning_rate": 0.00014902686083221884,
      "loss": 2.0631,
      "step": 153877
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.2720848321914673,
      "learning_rate": 0.0001490233260095365,
      "loss": 2.1884,
      "step": 153878
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.2217079401016235,
      "learning_rate": 0.00014901979121492373,
      "loss": 2.2047,
      "step": 153879
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.1697790622711182,
      "learning_rate": 0.00014901625644838092,
      "loss": 1.8888,
      "step": 153880
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.0899938344955444,
      "learning_rate": 0.00014901272170990895,
      "loss": 2.0186,
      "step": 153881
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.1014095544815063,
      "learning_rate": 0.00014900918699950832,
      "loss": 1.961,
      "step": 153882
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.2317218780517578,
      "learning_rate": 0.0001490056523171798,
      "loss": 2.1172,
      "step": 153883
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.054395079612732,
      "learning_rate": 0.0001490021176629238,
      "loss": 1.8568,
      "step": 153884
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.1047347784042358,
      "learning_rate": 0.00014899858303674135,
      "loss": 1.8917,
      "step": 153885
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.0932246446609497,
      "learning_rate": 0.00014899504843863274,
      "loss": 1.881,
      "step": 153886
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.2034357786178589,
      "learning_rate": 0.00014899151386859891,
      "loss": 2.0589,
      "step": 153887
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.2275809049606323,
      "learning_rate": 0.00014898797932664043,
      "loss": 1.9673,
      "step": 153888
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.2416746616363525,
      "learning_rate": 0.00014898444481275795,
      "loss": 2.0744,
      "step": 153889
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.0651792287826538,
      "learning_rate": 0.00014898091032695198,
      "loss": 1.9607,
      "step": 153890
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.1624579429626465,
      "learning_rate": 0.00014897737586922342,
      "loss": 2.3105,
      "step": 153891
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.2357051372528076,
      "learning_rate": 0.0001489738414395727,
      "loss": 2.0787,
      "step": 153892
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.286518931388855,
      "learning_rate": 0.00014897030703800072,
      "loss": 2.0699,
      "step": 153893
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.1888060569763184,
      "learning_rate": 0.00014896677266450797,
      "loss": 2.0075,
      "step": 153894
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.0775659084320068,
      "learning_rate": 0.0001489632383190952,
      "loss": 1.975,
      "step": 153895
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.1304081678390503,
      "learning_rate": 0.00014895970400176289,
      "loss": 2.0667,
      "step": 153896
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.0893588066101074,
      "learning_rate": 0.00014895616971251192,
      "loss": 2.0675,
      "step": 153897
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.188205361366272,
      "learning_rate": 0.00014895263545134277,
      "loss": 2.2878,
      "step": 153898
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.1969325542449951,
      "learning_rate": 0.00014894910121825624,
      "loss": 2.1775,
      "step": 153899
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.120813012123108,
      "learning_rate": 0.00014894556701325296,
      "loss": 2.128,
      "step": 153900
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.170041561126709,
      "learning_rate": 0.00014894203283633353,
      "loss": 1.8851,
      "step": 153901
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.397386074066162,
      "learning_rate": 0.0001489384986874985,
      "loss": 1.851,
      "step": 153902
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.2237766981124878,
      "learning_rate": 0.00014893496456674883,
      "loss": 2.0944,
      "step": 153903
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.1832090616226196,
      "learning_rate": 0.00014893143047408486,
      "loss": 2.1504,
      "step": 153904
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.2296839952468872,
      "learning_rate": 0.00014892789640950747,
      "loss": 2.1075,
      "step": 153905
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.2895069122314453,
      "learning_rate": 0.00014892436237301727,
      "loss": 2.2251,
      "step": 153906
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.2066353559494019,
      "learning_rate": 0.00014892082836461486,
      "loss": 2.1979,
      "step": 153907
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.1570775508880615,
      "learning_rate": 0.00014891729438430083,
      "loss": 2.2229,
      "step": 153908
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.127250075340271,
      "learning_rate": 0.000148913760432076,
      "loss": 2.0742,
      "step": 153909
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.1053657531738281,
      "learning_rate": 0.00014891022650794089,
      "loss": 1.8806,
      "step": 153910
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.0364121198654175,
      "learning_rate": 0.00014890669261189633,
      "loss": 1.807,
      "step": 153911
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.1939383745193481,
      "learning_rate": 0.00014890315874394283,
      "loss": 2.0304,
      "step": 153912
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.1335813999176025,
      "learning_rate": 0.0001488996249040811,
      "loss": 2.1889,
      "step": 153913
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.2111284732818604,
      "learning_rate": 0.00014889609109231168,
      "loss": 2.373,
      "step": 153914
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.1658037900924683,
      "learning_rate": 0.00014889255730863544,
      "loss": 2.0607,
      "step": 153915
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.1149561405181885,
      "learning_rate": 0.00014888902355305277,
      "loss": 2.1448,
      "step": 153916
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.7452392578125,
      "learning_rate": 0.00014888548982556465,
      "loss": 2.0991,
      "step": 153917
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.1048862934112549,
      "learning_rate": 0.00014888195612617155,
      "loss": 2.0864,
      "step": 153918
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.099463939666748,
      "learning_rate": 0.0001488784224548742,
      "loss": 2.0018,
      "step": 153919
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.1045912504196167,
      "learning_rate": 0.000148874888811673,
      "loss": 2.0309,
      "step": 153920
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.1440646648406982,
      "learning_rate": 0.00014887135519656898,
      "loss": 2.0764,
      "step": 153921
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.1739052534103394,
      "learning_rate": 0.00014886782160956249,
      "loss": 2.1492,
      "step": 153922
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.1573249101638794,
      "learning_rate": 0.00014886428805065444,
      "loss": 2.0928,
      "step": 153923
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.174523115158081,
      "learning_rate": 0.00014886075451984537,
      "loss": 2.072,
      "step": 153924
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.2365435361862183,
      "learning_rate": 0.00014885722101713598,
      "loss": 2.0892,
      "step": 153925
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.0863739252090454,
      "learning_rate": 0.00014885368754252674,
      "loss": 2.1805,
      "step": 153926
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.152138352394104,
      "learning_rate": 0.00014885015409601856,
      "loss": 2.2011,
      "step": 153927
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.1886959075927734,
      "learning_rate": 0.0001488466206776119,
      "loss": 2.0144,
      "step": 153928
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.2109737396240234,
      "learning_rate": 0.00014884308728730757,
      "loss": 2.1683,
      "step": 153929
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.1885666847229004,
      "learning_rate": 0.00014883955392510623,
      "loss": 2.1129,
      "step": 153930
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.0779812335968018,
      "learning_rate": 0.00014883602059100843,
      "loss": 1.9071,
      "step": 153931
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.1132915019989014,
      "learning_rate": 0.00014883248728501472,
      "loss": 1.993,
      "step": 153932
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.1076096296310425,
      "learning_rate": 0.0001488289540071261,
      "loss": 1.927,
      "step": 153933
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.1323426961898804,
      "learning_rate": 0.00014882542075734284,
      "loss": 1.8868,
      "step": 153934
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.214836835861206,
      "learning_rate": 0.00014882188753566596,
      "loss": 2.1164,
      "step": 153935
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.0234872102737427,
      "learning_rate": 0.0001488183543420958,
      "loss": 2.0705,
      "step": 153936
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.1668106317520142,
      "learning_rate": 0.0001488148211766334,
      "loss": 1.8901,
      "step": 153937
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.1329784393310547,
      "learning_rate": 0.00014881128803927895,
      "loss": 2.021,
      "step": 153938
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.1172287464141846,
      "learning_rate": 0.00014880775493003346,
      "loss": 2.0838,
      "step": 153939
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.1030640602111816,
      "learning_rate": 0.00014880422184889732,
      "loss": 2.0284,
      "step": 153940
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.3059024810791016,
      "learning_rate": 0.00014880068879587145,
      "loss": 2.0046,
      "step": 153941
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.1617575883865356,
      "learning_rate": 0.0001487971557709563,
      "loss": 1.9912,
      "step": 153942
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.196853756904602,
      "learning_rate": 0.00014879362277415273,
      "loss": 2.0412,
      "step": 153943
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.2619714736938477,
      "learning_rate": 0.00014879008980546123,
      "loss": 2.0893,
      "step": 153944
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.0246731042861938,
      "learning_rate": 0.00014878655686488253,
      "loss": 1.9063,
      "step": 153945
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.086756706237793,
      "learning_rate": 0.00014878302395241716,
      "loss": 1.9053,
      "step": 153946
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.492013692855835,
      "learning_rate": 0.000148779491068066,
      "loss": 1.9289,
      "step": 153947
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.0561460256576538,
      "learning_rate": 0.00014877595821182946,
      "loss": 2.0489,
      "step": 153948
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.0603711605072021,
      "learning_rate": 0.00014877242538370843,
      "loss": 2.2009,
      "step": 153949
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.1231343746185303,
      "learning_rate": 0.0001487688925837035,
      "loss": 2.0425,
      "step": 153950
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.1519994735717773,
      "learning_rate": 0.00014876535981181522,
      "loss": 2.105,
      "step": 153951
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.099722146987915,
      "learning_rate": 0.00014876182706804425,
      "loss": 2.028,
      "step": 153952
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.2704814672470093,
      "learning_rate": 0.0001487582943523914,
      "loss": 2.0506,
      "step": 153953
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.1127458810806274,
      "learning_rate": 0.00014875476166485713,
      "loss": 1.9427,
      "step": 153954
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.1128047704696655,
      "learning_rate": 0.00014875122900544235,
      "loss": 2.0234,
      "step": 153955
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.2053978443145752,
      "learning_rate": 0.00014874769637414755,
      "loss": 1.7935,
      "step": 153956
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.2063549757003784,
      "learning_rate": 0.0001487441637709733,
      "loss": 1.9826,
      "step": 153957
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.2435206174850464,
      "learning_rate": 0.00014874063119592044,
      "loss": 2.0127,
      "step": 153958
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.1057531833648682,
      "learning_rate": 0.0001487370986489896,
      "loss": 2.0253,
      "step": 153959
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.0802005529403687,
      "learning_rate": 0.00014873356613018124,
      "loss": 1.9494,
      "step": 153960
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.0645380020141602,
      "learning_rate": 0.00014873003363949632,
      "loss": 2.0753,
      "step": 153961
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.1864205598831177,
      "learning_rate": 0.00014872650117693533,
      "loss": 2.0343,
      "step": 153962
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.131169319152832,
      "learning_rate": 0.0001487229687424988,
      "loss": 1.9263,
      "step": 153963
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.1229077577590942,
      "learning_rate": 0.00014871943633618768,
      "loss": 1.958,
      "step": 153964
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.0908379554748535,
      "learning_rate": 0.00014871590395800242,
      "loss": 2.091,
      "step": 153965
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.118436336517334,
      "learning_rate": 0.00014871237160794367,
      "loss": 2.1305,
      "step": 153966
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.1681517362594604,
      "learning_rate": 0.00014870883928601222,
      "loss": 2.1975,
      "step": 153967
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.4589252471923828,
      "learning_rate": 0.00014870530699220865,
      "loss": 1.856,
      "step": 153968
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.1978758573532104,
      "learning_rate": 0.00014870177472653352,
      "loss": 2.068,
      "step": 153969
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.1840007305145264,
      "learning_rate": 0.00014869824248898772,
      "loss": 1.938,
      "step": 153970
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.1664915084838867,
      "learning_rate": 0.00014869471027957162,
      "loss": 2.1495,
      "step": 153971
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.2887996435165405,
      "learning_rate": 0.00014869117809828615,
      "loss": 2.0354,
      "step": 153972
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.1170494556427002,
      "learning_rate": 0.00014868764594513185,
      "loss": 1.9671,
      "step": 153973
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.1801660060882568,
      "learning_rate": 0.0001486841138201094,
      "loss": 2.1043,
      "step": 153974
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.1844100952148438,
      "learning_rate": 0.00014868058172321928,
      "loss": 1.9897,
      "step": 153975
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.4410114288330078,
      "learning_rate": 0.00014867704965446245,
      "loss": 1.9203,
      "step": 153976
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.1348263025283813,
      "learning_rate": 0.0001486735176138393,
      "loss": 2.174,
      "step": 153977
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.4431366920471191,
      "learning_rate": 0.00014866998560135065,
      "loss": 1.9945,
      "step": 153978
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.1257169246673584,
      "learning_rate": 0.00014866645361699714,
      "loss": 1.7335,
      "step": 153979
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.3118655681610107,
      "learning_rate": 0.00014866292166077941,
      "loss": 1.8974,
      "step": 153980
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.0912219285964966,
      "learning_rate": 0.00014865938973269798,
      "loss": 2.1791,
      "step": 153981
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.142412543296814,
      "learning_rate": 0.00014865585783275373,
      "loss": 1.9906,
      "step": 153982
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.2170076370239258,
      "learning_rate": 0.0001486523259609471,
      "loss": 2.1977,
      "step": 153983
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.3863475322723389,
      "learning_rate": 0.000148648794117279,
      "loss": 1.9638,
      "step": 153984
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.0623711347579956,
      "learning_rate": 0.00014864526230174988,
      "loss": 2.031,
      "step": 153985
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.215099573135376,
      "learning_rate": 0.00014864173051436054,
      "loss": 2.0084,
      "step": 153986
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.2159298658370972,
      "learning_rate": 0.0001486381987551114,
      "loss": 2.1384,
      "step": 153987
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.2977463006973267,
      "learning_rate": 0.0001486346670240034,
      "loss": 1.9311,
      "step": 153988
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.4175547361373901,
      "learning_rate": 0.00014863113532103696,
      "loss": 2.0622,
      "step": 153989
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.1107462644577026,
      "learning_rate": 0.00014862760364621296,
      "loss": 2.1379,
      "step": 153990
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.3093407154083252,
      "learning_rate": 0.00014862407199953197,
      "loss": 2.2103,
      "step": 153991
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.2041106224060059,
      "learning_rate": 0.0001486205403809946,
      "loss": 1.8903,
      "step": 153992
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.4377663135528564,
      "learning_rate": 0.0001486170087906014,
      "loss": 2.0546,
      "step": 153993
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.2667627334594727,
      "learning_rate": 0.0001486134772283533,
      "loss": 2.1634,
      "step": 153994
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.098128080368042,
      "learning_rate": 0.00014860994569425068,
      "loss": 2.1528,
      "step": 153995
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.0691713094711304,
      "learning_rate": 0.00014860641418829444,
      "loss": 2.1666,
      "step": 153996
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.1457406282424927,
      "learning_rate": 0.00014860288271048512,
      "loss": 2.0623,
      "step": 153997
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.1387354135513306,
      "learning_rate": 0.0001485993512608234,
      "loss": 2.1253,
      "step": 153998
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.3026483058929443,
      "learning_rate": 0.0001485958198393098,
      "loss": 1.9934,
      "step": 153999
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.0898520946502686,
      "learning_rate": 0.0001485922884459452,
      "loss": 2.0326,
      "step": 154000
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.2521713972091675,
      "learning_rate": 0.00014858875708073006,
      "loss": 2.1722,
      "step": 154001
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.1927968263626099,
      "learning_rate": 0.00014858522574366523,
      "loss": 2.2138,
      "step": 154002
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.0961685180664062,
      "learning_rate": 0.00014858169443475116,
      "loss": 2.0244,
      "step": 154003
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.1917896270751953,
      "learning_rate": 0.0001485781631539888,
      "loss": 1.8472,
      "step": 154004
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.205854892730713,
      "learning_rate": 0.00014857463190137842,
      "loss": 2.1227,
      "step": 154005
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.1417416334152222,
      "learning_rate": 0.000148571100676921,
      "loss": 1.947,
      "step": 154006
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.0910595655441284,
      "learning_rate": 0.0001485675694806169,
      "loss": 2.371,
      "step": 154007
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.0946599245071411,
      "learning_rate": 0.00014856403831246714,
      "loss": 1.919,
      "step": 154008
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.0792932510375977,
      "learning_rate": 0.00014856050717247204,
      "loss": 1.9409,
      "step": 154009
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.1454598903656006,
      "learning_rate": 0.0001485569760606326,
      "loss": 1.8586,
      "step": 154010
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.2086880207061768,
      "learning_rate": 0.00014855344497694909,
      "loss": 2.1505,
      "step": 154011
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.1444885730743408,
      "learning_rate": 0.00014854991392142244,
      "loss": 2.1175,
      "step": 154012
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.2305666208267212,
      "learning_rate": 0.00014854638289405313,
      "loss": 2.0069,
      "step": 154013
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.1314265727996826,
      "learning_rate": 0.00014854285189484198,
      "loss": 2.2184,
      "step": 154014
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.3654202222824097,
      "learning_rate": 0.0001485393209237895,
      "loss": 2.3553,
      "step": 154015
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.164649248123169,
      "learning_rate": 0.0001485357899808966,
      "loss": 2.0188,
      "step": 154016
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.1708463430404663,
      "learning_rate": 0.0001485322590661636,
      "loss": 2.0245,
      "step": 154017
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.1270021200180054,
      "learning_rate": 0.00014852872817959137,
      "loss": 2.0368,
      "step": 154018
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.3756251335144043,
      "learning_rate": 0.0001485251973211804,
      "loss": 2.0957,
      "step": 154019
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.1250801086425781,
      "learning_rate": 0.00014852166649093157,
      "loss": 2.1255,
      "step": 154020
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.098752737045288,
      "learning_rate": 0.00014851813568884532,
      "loss": 2.193,
      "step": 154021
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.119411826133728,
      "learning_rate": 0.00014851460491492263,
      "loss": 2.0245,
      "step": 154022
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.1011608839035034,
      "learning_rate": 0.00014851107416916375,
      "loss": 1.9115,
      "step": 154023
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.1544792652130127,
      "learning_rate": 0.0001485075434515696,
      "loss": 2.078,
      "step": 154024
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.1530629396438599,
      "learning_rate": 0.0001485040127621406,
      "loss": 2.0024,
      "step": 154025
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.0919246673583984,
      "learning_rate": 0.0001485004821008777,
      "loss": 2.1864,
      "step": 154026
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.0741692781448364,
      "learning_rate": 0.00014849695146778132,
      "loss": 2.0566,
      "step": 154027
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.0759705305099487,
      "learning_rate": 0.00014849342086285246,
      "loss": 1.9572,
      "step": 154028
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.1110206842422485,
      "learning_rate": 0.0001484898902860913,
      "loss": 1.9573,
      "step": 154029
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.1590219736099243,
      "learning_rate": 0.00014848635973749882,
      "loss": 2.0948,
      "step": 154030
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.2404381036758423,
      "learning_rate": 0.00014848282921707548,
      "loss": 1.9046,
      "step": 154031
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.1965539455413818,
      "learning_rate": 0.00014847929872482217,
      "loss": 2.0633,
      "step": 154032
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.1104929447174072,
      "learning_rate": 0.00014847576826073928,
      "loss": 1.8026,
      "step": 154033
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.2404136657714844,
      "learning_rate": 0.00014847223782482773,
      "loss": 1.9918,
      "step": 154034
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.2143394947052002,
      "learning_rate": 0.00014846870741708809,
      "loss": 2.2517,
      "step": 154035
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.3935660123825073,
      "learning_rate": 0.00014846517703752092,
      "loss": 1.9871,
      "step": 154036
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.085068702697754,
      "learning_rate": 0.00014846164668612684,
      "loss": 2.0273,
      "step": 154037
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.2192682027816772,
      "learning_rate": 0.00014845811636290672,
      "loss": 2.041,
      "step": 154038
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.5444538593292236,
      "learning_rate": 0.000148454586067861,
      "loss": 1.9848,
      "step": 154039
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.1682579517364502,
      "learning_rate": 0.00014845105580099053,
      "loss": 2.242,
      "step": 154040
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.1193902492523193,
      "learning_rate": 0.00014844752556229584,
      "loss": 1.6225,
      "step": 154041
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.1718933582305908,
      "learning_rate": 0.00014844399535177754,
      "loss": 2.0358,
      "step": 154042
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.1002219915390015,
      "learning_rate": 0.00014844046516943645,
      "loss": 2.0317,
      "step": 154043
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.2382258176803589,
      "learning_rate": 0.00014843693501527316,
      "loss": 2.1768,
      "step": 154044
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.152407169342041,
      "learning_rate": 0.0001484334048892882,
      "loss": 2.0505,
      "step": 154045
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.2039539813995361,
      "learning_rate": 0.00014842987479148244,
      "loss": 2.0245,
      "step": 154046
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.4505362510681152,
      "learning_rate": 0.0001484263447218564,
      "loss": 2.2834,
      "step": 154047
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.3492615222930908,
      "learning_rate": 0.00014842281468041067,
      "loss": 2.0545,
      "step": 154048
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.24264657497406,
      "learning_rate": 0.00014841928466714611,
      "loss": 1.8482,
      "step": 154049
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.3098204135894775,
      "learning_rate": 0.00014841575468206328,
      "loss": 2.0727,
      "step": 154050
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.2440624237060547,
      "learning_rate": 0.0001484122247251627,
      "loss": 2.1283,
      "step": 154051
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.1128833293914795,
      "learning_rate": 0.00014840869479644522,
      "loss": 2.366,
      "step": 154052
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.2320212125778198,
      "learning_rate": 0.0001484051648959115,
      "loss": 1.9814,
      "step": 154053
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.198930025100708,
      "learning_rate": 0.00014840163502356194,
      "loss": 2.1966,
      "step": 154054
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.1540591716766357,
      "learning_rate": 0.00014839810517939755,
      "loss": 1.9889,
      "step": 154055
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.0294727087020874,
      "learning_rate": 0.00014839457536341865,
      "loss": 2.1094,
      "step": 154056
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.184531569480896,
      "learning_rate": 0.00014839104557562622,
      "loss": 2.0348,
      "step": 154057
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.1361149549484253,
      "learning_rate": 0.00014838751581602072,
      "loss": 2.2416,
      "step": 154058
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.1677594184875488,
      "learning_rate": 0.00014838398608460287,
      "loss": 2.0979,
      "step": 154059
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.334319829940796,
      "learning_rate": 0.00014838045638137316,
      "loss": 1.9636,
      "step": 154060
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.0887168645858765,
      "learning_rate": 0.0001483769267063325,
      "loss": 2.0678,
      "step": 154061
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.2262829542160034,
      "learning_rate": 0.00014837339705948136,
      "loss": 2.0389,
      "step": 154062
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.1042320728302002,
      "learning_rate": 0.00014836986744082054,
      "loss": 2.1392,
      "step": 154063
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.2379348278045654,
      "learning_rate": 0.0001483663378503506,
      "loss": 2.0266,
      "step": 154064
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.1435277462005615,
      "learning_rate": 0.00014836280828807224,
      "loss": 2.0485,
      "step": 154065
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.2337993383407593,
      "learning_rate": 0.00014835927875398602,
      "loss": 2.0445,
      "step": 154066
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.3153557777404785,
      "learning_rate": 0.00014835574924809277,
      "loss": 1.9975,
      "step": 154067
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.1676249504089355,
      "learning_rate": 0.00014835221977039294,
      "loss": 2.0805,
      "step": 154068
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.1138941049575806,
      "learning_rate": 0.00014834869032088736,
      "loss": 2.1264,
      "step": 154069
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.1235764026641846,
      "learning_rate": 0.00014834516089957657,
      "loss": 2.07,
      "step": 154070
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.1965889930725098,
      "learning_rate": 0.00014834163150646146,
      "loss": 1.9487,
      "step": 154071
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.090038537979126,
      "learning_rate": 0.00014833810214154232,
      "loss": 2.1047,
      "step": 154072
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.1498022079467773,
      "learning_rate": 0.00014833457280482006,
      "loss": 2.384,
      "step": 154073
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.0803672075271606,
      "learning_rate": 0.0001483310434962952,
      "loss": 2.0646,
      "step": 154074
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.1999871730804443,
      "learning_rate": 0.00014832751421596852,
      "loss": 2.0982,
      "step": 154075
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.08849036693573,
      "learning_rate": 0.0001483239849638405,
      "loss": 2.0894,
      "step": 154076
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.2270315885543823,
      "learning_rate": 0.0001483204557399122,
      "loss": 2.2304,
      "step": 154077
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.0852922201156616,
      "learning_rate": 0.00014831692654418372,
      "loss": 2.1283,
      "step": 154078
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.1948503255844116,
      "learning_rate": 0.00014831339737665612,
      "loss": 2.0685,
      "step": 154079
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.202204704284668,
      "learning_rate": 0.0001483098682373298,
      "loss": 1.808,
      "step": 154080
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.2496724128723145,
      "learning_rate": 0.00014830633912620567,
      "loss": 2.0833,
      "step": 154081
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.059203028678894,
      "learning_rate": 0.0001483028100432841,
      "loss": 2.1711,
      "step": 154082
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.1933352947235107,
      "learning_rate": 0.0001482992809885662,
      "loss": 2.2426,
      "step": 154083
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.0903668403625488,
      "learning_rate": 0.00014829575196205204,
      "loss": 2.0763,
      "step": 154084
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.218058705329895,
      "learning_rate": 0.0001482922229637427,
      "loss": 2.025,
      "step": 154085
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.2281917333602905,
      "learning_rate": 0.00014828869399363854,
      "loss": 1.8561,
      "step": 154086
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.1822850704193115,
      "learning_rate": 0.00014828516505174054,
      "loss": 1.9346,
      "step": 154087
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.134442925453186,
      "learning_rate": 0.00014828163613804908,
      "loss": 1.883,
      "step": 154088
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.1695632934570312,
      "learning_rate": 0.00014827810725256512,
      "loss": 2.1363,
      "step": 154089
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.1950643062591553,
      "learning_rate": 0.00014827457839528894,
      "loss": 2.1409,
      "step": 154090
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.333469271659851,
      "learning_rate": 0.00014827104956622144,
      "loss": 2.0781,
      "step": 154091
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.1295194625854492,
      "learning_rate": 0.00014826752076536315,
      "loss": 2.024,
      "step": 154092
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.2159863710403442,
      "learning_rate": 0.0001482639919927149,
      "loss": 2.0514,
      "step": 154093
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.1455944776535034,
      "learning_rate": 0.0001482604632482771,
      "loss": 2.2931,
      "step": 154094
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.2064208984375,
      "learning_rate": 0.0001482569345320508,
      "loss": 2.1829,
      "step": 154095
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.1895482540130615,
      "learning_rate": 0.00014825340584403613,
      "loss": 2.2906,
      "step": 154096
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.3795928955078125,
      "learning_rate": 0.0001482498771842342,
      "loss": 1.9792,
      "step": 154097
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.2838068008422852,
      "learning_rate": 0.0001482463485526453,
      "loss": 2.1746,
      "step": 154098
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.1098628044128418,
      "learning_rate": 0.0001482428199492704,
      "loss": 2.0837,
      "step": 154099
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.1792877912521362,
      "learning_rate": 0.0001482392913741099,
      "loss": 2.15,
      "step": 154100
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.1224693059921265,
      "learning_rate": 0.00014823576282716486,
      "loss": 1.9055,
      "step": 154101
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.163970947265625,
      "learning_rate": 0.00014823223430843538,
      "loss": 1.9718,
      "step": 154102
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.0812734365463257,
      "learning_rate": 0.00014822870581792253,
      "loss": 1.9123,
      "step": 154103
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.198188066482544,
      "learning_rate": 0.0001482251773556267,
      "loss": 2.1493,
      "step": 154104
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.278175711631775,
      "learning_rate": 0.00014822164892154882,
      "loss": 2.1953,
      "step": 154105
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.0987162590026855,
      "learning_rate": 0.00014821812051568923,
      "loss": 2.159,
      "step": 154106
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.323540210723877,
      "learning_rate": 0.00014821459213804904,
      "loss": 1.9077,
      "step": 154107
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.111733078956604,
      "learning_rate": 0.00014821106378862836,
      "loss": 2.2273,
      "step": 154108
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.0496714115142822,
      "learning_rate": 0.00014820753546742824,
      "loss": 2.1337,
      "step": 154109
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.2468870878219604,
      "learning_rate": 0.00014820400717444908,
      "loss": 1.9143,
      "step": 154110
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.144347071647644,
      "learning_rate": 0.0001482004789096918,
      "loss": 1.9499,
      "step": 154111
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.1655972003936768,
      "learning_rate": 0.00014819695067315675,
      "loss": 2.1402,
      "step": 154112
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.1894131898880005,
      "learning_rate": 0.000148193422464845,
      "loss": 2.1413,
      "step": 154113
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.1058639287948608,
      "learning_rate": 0.00014818989428475675,
      "loss": 2.0423,
      "step": 154114
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.2172168493270874,
      "learning_rate": 0.00014818636613289299,
      "loss": 2.0084,
      "step": 154115
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.0662013292312622,
      "learning_rate": 0.0001481828380092541,
      "loss": 1.9301,
      "step": 154116
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.119631290435791,
      "learning_rate": 0.00014817930991384103,
      "loss": 2.0439,
      "step": 154117
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.2760534286499023,
      "learning_rate": 0.00014817578184665413,
      "loss": 2.0313,
      "step": 154118
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.1118667125701904,
      "learning_rate": 0.0001481722538076944,
      "loss": 2.0974,
      "step": 154119
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.2490206956863403,
      "learning_rate": 0.00014816872579696228,
      "loss": 2.0544,
      "step": 154120
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.1416970491409302,
      "learning_rate": 0.00014816519781445846,
      "loss": 2.0539,
      "step": 154121
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.1628717184066772,
      "learning_rate": 0.0001481616698601835,
      "loss": 2.1963,
      "step": 154122
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.0444350242614746,
      "learning_rate": 0.00014815814193413823,
      "loss": 1.9141,
      "step": 154123
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.054197072982788,
      "learning_rate": 0.00014815461403632315,
      "loss": 2.0407,
      "step": 154124
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.1420397758483887,
      "learning_rate": 0.00014815108616673907,
      "loss": 1.9772,
      "step": 154125
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.134135127067566,
      "learning_rate": 0.0001481475583253866,
      "loss": 2.2936,
      "step": 154126
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.0767208337783813,
      "learning_rate": 0.00014814403051226633,
      "loss": 1.9432,
      "step": 154127
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.244277834892273,
      "learning_rate": 0.00014814050272737885,
      "loss": 1.9735,
      "step": 154128
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.2342256307601929,
      "learning_rate": 0.00014813697497072505,
      "loss": 2.2605,
      "step": 154129
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.1629687547683716,
      "learning_rate": 0.00014813344724230534,
      "loss": 2.0391,
      "step": 154130
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.067971110343933,
      "learning_rate": 0.0001481299195421206,
      "loss": 1.9398,
      "step": 154131
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.1292704343795776,
      "learning_rate": 0.00014812639187017135,
      "loss": 2.2005,
      "step": 154132
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.2598166465759277,
      "learning_rate": 0.00014812286422645818,
      "loss": 2.1034,
      "step": 154133
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.233437180519104,
      "learning_rate": 0.00014811933661098195,
      "loss": 2.1099,
      "step": 154134
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.1862510442733765,
      "learning_rate": 0.0001481158090237432,
      "loss": 2.1063,
      "step": 154135
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.1572209596633911,
      "learning_rate": 0.00014811228146474246,
      "loss": 1.8888,
      "step": 154136
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.3016871213912964,
      "learning_rate": 0.00014810875393398066,
      "loss": 1.9902,
      "step": 154137
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.313707947731018,
      "learning_rate": 0.0001481052264314583,
      "loss": 2.1523,
      "step": 154138
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.168763518333435,
      "learning_rate": 0.0001481016989571759,
      "loss": 2.3409,
      "step": 154139
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.1527669429779053,
      "learning_rate": 0.00014809817151113441,
      "loss": 1.7508,
      "step": 154140
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.1576459407806396,
      "learning_rate": 0.0001480946440933342,
      "loss": 2.0878,
      "step": 154141
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.139971137046814,
      "learning_rate": 0.0001480911167037762,
      "loss": 1.9321,
      "step": 154142
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.1522661447525024,
      "learning_rate": 0.0001480875893424609,
      "loss": 1.7409,
      "step": 154143
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.2022106647491455,
      "learning_rate": 0.000148084062009389,
      "loss": 2.048,
      "step": 154144
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.243213415145874,
      "learning_rate": 0.00014808053470456102,
      "loss": 2.0112,
      "step": 154145
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.176005482673645,
      "learning_rate": 0.00014807700742797787,
      "loss": 2.0369,
      "step": 154146
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.2177221775054932,
      "learning_rate": 0.00014807348017963992,
      "loss": 2.1923,
      "step": 154147
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.1959117650985718,
      "learning_rate": 0.00014806995295954813,
      "loss": 2.0413,
      "step": 154148
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.2344601154327393,
      "learning_rate": 0.000148066425767703,
      "loss": 1.9578,
      "step": 154149
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.0813168287277222,
      "learning_rate": 0.00014806289860410518,
      "loss": 2.0362,
      "step": 154150
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.1969022750854492,
      "learning_rate": 0.00014805937146875522,
      "loss": 2.0351,
      "step": 154151
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.4482545852661133,
      "learning_rate": 0.000148055844361654,
      "loss": 2.3405,
      "step": 154152
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.2555712461471558,
      "learning_rate": 0.00014805231728280197,
      "loss": 1.9745,
      "step": 154153
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.2116622924804688,
      "learning_rate": 0.0001480487902322,
      "loss": 2.0575,
      "step": 154154
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.1339352130889893,
      "learning_rate": 0.0001480452632098485,
      "loss": 2.0629,
      "step": 154155
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.261839509010315,
      "learning_rate": 0.00014804173621574848,
      "loss": 2.0641,
      "step": 154156
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.2554399967193604,
      "learning_rate": 0.00014803820924990014,
      "loss": 2.078,
      "step": 154157
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.198263168334961,
      "learning_rate": 0.0001480346823123045,
      "loss": 2.0269,
      "step": 154158
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.2113703489303589,
      "learning_rate": 0.00014803115540296195,
      "loss": 1.9735,
      "step": 154159
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.0387686491012573,
      "learning_rate": 0.00014802762852187337,
      "loss": 2.1894,
      "step": 154160
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.1163933277130127,
      "learning_rate": 0.00014802410166903923,
      "loss": 1.988,
      "step": 154161
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.0725947618484497,
      "learning_rate": 0.0001480205748444605,
      "loss": 1.9088,
      "step": 154162
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.1420061588287354,
      "learning_rate": 0.0001480170480481374,
      "loss": 2.2024,
      "step": 154163
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.1548316478729248,
      "learning_rate": 0.0001480135212800709,
      "loss": 1.8899,
      "step": 154164
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.2019051313400269,
      "learning_rate": 0.00014800999454026145,
      "loss": 2.2126,
      "step": 154165
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.1386632919311523,
      "learning_rate": 0.0001480064678287099,
      "loss": 2.0268,
      "step": 154166
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.2870460748672485,
      "learning_rate": 0.00014800294114541675,
      "loss": 2.1358,
      "step": 154167
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.2508550882339478,
      "learning_rate": 0.0001479994144903829,
      "loss": 2.0338,
      "step": 154168
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.3113293647766113,
      "learning_rate": 0.00014799588786360864,
      "loss": 1.8723,
      "step": 154169
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.2064834833145142,
      "learning_rate": 0.00014799236126509494,
      "loss": 1.8544,
      "step": 154170
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.079856038093567,
      "learning_rate": 0.00014798883469484218,
      "loss": 1.8849,
      "step": 154171
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.1565358638763428,
      "learning_rate": 0.00014798530815285127,
      "loss": 1.9712,
      "step": 154172
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.1454042196273804,
      "learning_rate": 0.00014798178163912267,
      "loss": 2.2434,
      "step": 154173
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.2045427560806274,
      "learning_rate": 0.00014797825515365734,
      "loss": 2.0036,
      "step": 154174
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.1849043369293213,
      "learning_rate": 0.00014797472869645552,
      "loss": 2.2071,
      "step": 154175
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.1593353748321533,
      "learning_rate": 0.00014797120226751816,
      "loss": 1.8997,
      "step": 154176
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.1967928409576416,
      "learning_rate": 0.0001479676758668457,
      "loss": 2.1076,
      "step": 154177
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.2151583433151245,
      "learning_rate": 0.00014796414949443907,
      "loss": 2.1495,
      "step": 154178
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.0958805084228516,
      "learning_rate": 0.0001479606231502986,
      "loss": 2.2597,
      "step": 154179
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.3273251056671143,
      "learning_rate": 0.0001479570968344254,
      "loss": 2.1189,
      "step": 154180
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.2825357913970947,
      "learning_rate": 0.00014795357054681963,
      "loss": 2.0293,
      "step": 154181
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.2866790294647217,
      "learning_rate": 0.00014795004428748223,
      "loss": 2.2005,
      "step": 154182
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.1950204372406006,
      "learning_rate": 0.0001479465180564137,
      "loss": 2.0343,
      "step": 154183
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.1227242946624756,
      "learning_rate": 0.00014794299185361487,
      "loss": 1.9686,
      "step": 154184
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.1917492151260376,
      "learning_rate": 0.0001479394656790862,
      "loss": 2.0738,
      "step": 154185
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.1763508319854736,
      "learning_rate": 0.0001479359395328287,
      "loss": 2.1453,
      "step": 154186
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.155299186706543,
      "learning_rate": 0.00014793241341484257,
      "loss": 1.9643,
      "step": 154187
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.1709917783737183,
      "learning_rate": 0.00014792888732512874,
      "loss": 2.4376,
      "step": 154188
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.2834603786468506,
      "learning_rate": 0.00014792536126368772,
      "loss": 1.954,
      "step": 154189
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.2533910274505615,
      "learning_rate": 0.00014792183523052035,
      "loss": 2.1953,
      "step": 154190
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.2476446628570557,
      "learning_rate": 0.00014791830922562707,
      "loss": 2.2047,
      "step": 154191
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.1545114517211914,
      "learning_rate": 0.00014791478324900886,
      "loss": 1.8374,
      "step": 154192
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.1086969375610352,
      "learning_rate": 0.00014791125730066596,
      "loss": 1.9845,
      "step": 154193
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.087935447692871,
      "learning_rate": 0.0001479077313805993,
      "loss": 1.9306,
      "step": 154194
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.1938074827194214,
      "learning_rate": 0.00014790420548880943,
      "loss": 2.3527,
      "step": 154195
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.248257040977478,
      "learning_rate": 0.00014790067962529706,
      "loss": 2.2644,
      "step": 154196
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.0882408618927002,
      "learning_rate": 0.00014789715379006274,
      "loss": 2.0045,
      "step": 154197
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.1391888856887817,
      "learning_rate": 0.00014789362798310748,
      "loss": 2.0177,
      "step": 154198
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.1428234577178955,
      "learning_rate": 0.0001478901022044314,
      "loss": 1.9107,
      "step": 154199
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.1779553890228271,
      "learning_rate": 0.00014788657645403556,
      "loss": 1.8024,
      "step": 154200
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.1883773803710938,
      "learning_rate": 0.00014788305073192034,
      "loss": 2.1278,
      "step": 154201
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.152410626411438,
      "learning_rate": 0.00014787952503808667,
      "loss": 2.0314,
      "step": 154202
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.129244089126587,
      "learning_rate": 0.00014787599937253492,
      "loss": 2.2254,
      "step": 154203
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.1015201807022095,
      "learning_rate": 0.000147872473735266,
      "loss": 2.1175,
      "step": 154204
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.1345030069351196,
      "learning_rate": 0.0001478689481262805,
      "loss": 2.0998,
      "step": 154205
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.0634573698043823,
      "learning_rate": 0.00014786542254557902,
      "loss": 1.9919,
      "step": 154206
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.5046014785766602,
      "learning_rate": 0.00014786189699316212,
      "loss": 1.9176,
      "step": 154207
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.1608099937438965,
      "learning_rate": 0.00014785837146903067,
      "loss": 2.202,
      "step": 154208
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.303330898284912,
      "learning_rate": 0.0001478548459731851,
      "loss": 1.9739,
      "step": 154209
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.0737906694412231,
      "learning_rate": 0.0001478513205056263,
      "loss": 2.1029,
      "step": 154210
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.114035964012146,
      "learning_rate": 0.00014784779506635482,
      "loss": 2.2672,
      "step": 154211
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.1580458879470825,
      "learning_rate": 0.0001478442696553713,
      "loss": 1.8828,
      "step": 154212
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.137010931968689,
      "learning_rate": 0.0001478407442726763,
      "loss": 1.9629,
      "step": 154213
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.218605637550354,
      "learning_rate": 0.00014783721891827072,
      "loss": 1.9895,
      "step": 154214
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.3326059579849243,
      "learning_rate": 0.00014783369359215492,
      "loss": 2.0557,
      "step": 154215
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.239556908607483,
      "learning_rate": 0.00014783016829432983,
      "loss": 2.1255,
      "step": 154216
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.098563313484192,
      "learning_rate": 0.00014782664302479597,
      "loss": 2.1939,
      "step": 154217
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.2645448446273804,
      "learning_rate": 0.00014782311778355393,
      "loss": 1.9757,
      "step": 154218
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.0706920623779297,
      "learning_rate": 0.00014781959257060453,
      "loss": 2.0659,
      "step": 154219
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.111576795578003,
      "learning_rate": 0.00014781606738594835,
      "loss": 2.2052,
      "step": 154220
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.1815462112426758,
      "learning_rate": 0.0001478125422295859,
      "loss": 1.8368,
      "step": 154221
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.1629952192306519,
      "learning_rate": 0.0001478090171015181,
      "loss": 1.7719,
      "step": 154222
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.342336893081665,
      "learning_rate": 0.0001478054920017455,
      "loss": 2.068,
      "step": 154223
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.2543996572494507,
      "learning_rate": 0.00014780196693026858,
      "loss": 2.1302,
      "step": 154224
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.2225819826126099,
      "learning_rate": 0.0001477984418870883,
      "loss": 2.1883,
      "step": 154225
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.4569324254989624,
      "learning_rate": 0.00014779491687220514,
      "loss": 2.0372,
      "step": 154226
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.1724092960357666,
      "learning_rate": 0.0001477913918856197,
      "loss": 2.0901,
      "step": 154227
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.1978468894958496,
      "learning_rate": 0.00014778786692733276,
      "loss": 2.0453,
      "step": 154228
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.1402214765548706,
      "learning_rate": 0.00014778434199734497,
      "loss": 2.1029,
      "step": 154229
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.118786096572876,
      "learning_rate": 0.00014778081709565684,
      "loss": 2.0087,
      "step": 154230
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.1839858293533325,
      "learning_rate": 0.00014777729222226923,
      "loss": 1.9995,
      "step": 154231
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.1874046325683594,
      "learning_rate": 0.00014777376737718257,
      "loss": 2.0421,
      "step": 154232
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.1653107404708862,
      "learning_rate": 0.00014777024256039777,
      "loss": 2.0698,
      "step": 154233
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.0051450729370117,
      "learning_rate": 0.00014776671777191533,
      "loss": 1.9962,
      "step": 154234
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.1454107761383057,
      "learning_rate": 0.00014776319301173594,
      "loss": 2.0397,
      "step": 154235
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.280228853225708,
      "learning_rate": 0.00014775966827986018,
      "loss": 1.9922,
      "step": 154236
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.0708461999893188,
      "learning_rate": 0.00014775614357628882,
      "loss": 1.8997,
      "step": 154237
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.2983912229537964,
      "learning_rate": 0.00014775261890102238,
      "loss": 2.1982,
      "step": 154238
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.1235851049423218,
      "learning_rate": 0.00014774909425406172,
      "loss": 2.1674,
      "step": 154239
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.1753082275390625,
      "learning_rate": 0.00014774556963540725,
      "loss": 2.1259,
      "step": 154240
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.182407259941101,
      "learning_rate": 0.00014774204504506,
      "loss": 2.1878,
      "step": 154241
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.1513404846191406,
      "learning_rate": 0.00014773852048302014,
      "loss": 2.1241,
      "step": 154242
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.1067572832107544,
      "learning_rate": 0.00014773499594928868,
      "loss": 2.0489,
      "step": 154243
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.069257378578186,
      "learning_rate": 0.00014773147144386603,
      "loss": 2.1163,
      "step": 154244
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.17666757106781,
      "learning_rate": 0.0001477279469667531,
      "loss": 2.1318,
      "step": 154245
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.1936367750167847,
      "learning_rate": 0.00014772442251795033,
      "loss": 1.9964,
      "step": 154246
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.1271464824676514,
      "learning_rate": 0.00014772089809745863,
      "loss": 2.0505,
      "step": 154247
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.1760931015014648,
      "learning_rate": 0.00014771737370527832,
      "loss": 2.0992,
      "step": 154248
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.0577154159545898,
      "learning_rate": 0.00014771384934141033,
      "loss": 2.0729,
      "step": 154249
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.265842080116272,
      "learning_rate": 0.00014771032500585508,
      "loss": 2.1312,
      "step": 154250
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.055232286453247,
      "learning_rate": 0.00014770680069861345,
      "loss": 1.9642,
      "step": 154251
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.0821553468704224,
      "learning_rate": 0.0001477032764196859,
      "loss": 2.0292,
      "step": 154252
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.1400870084762573,
      "learning_rate": 0.00014769975216907341,
      "loss": 1.9226,
      "step": 154253
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.0904662609100342,
      "learning_rate": 0.00014769622794677623,
      "loss": 1.9688,
      "step": 154254
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.1442257165908813,
      "learning_rate": 0.00014769270375279526,
      "loss": 1.9591,
      "step": 154255
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.1460981369018555,
      "learning_rate": 0.000147689179587131,
      "loss": 2.0604,
      "step": 154256
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.2766470909118652,
      "learning_rate": 0.00014768565544978428,
      "loss": 2.0596,
      "step": 154257
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.1614859104156494,
      "learning_rate": 0.0001476821313407556,
      "loss": 2.1338,
      "step": 154258
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.2962415218353271,
      "learning_rate": 0.00014767860726004592,
      "loss": 1.9736,
      "step": 154259
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.1554721593856812,
      "learning_rate": 0.00014767508320765536,
      "loss": 2.0455,
      "step": 154260
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.1816632747650146,
      "learning_rate": 0.00014767155918358509,
      "loss": 2.2415,
      "step": 154261
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.121343731880188,
      "learning_rate": 0.0001476680351878354,
      "loss": 2.0747,
      "step": 154262
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.2607765197753906,
      "learning_rate": 0.00014766451122040717,
      "loss": 2.2826,
      "step": 154263
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.280932903289795,
      "learning_rate": 0.00014766098728130088,
      "loss": 1.9557,
      "step": 154264
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.0721293687820435,
      "learning_rate": 0.00014765746337051756,
      "loss": 2.1029,
      "step": 154265
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.125473976135254,
      "learning_rate": 0.00014765393948805734,
      "loss": 2.0192,
      "step": 154266
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.2274895906448364,
      "learning_rate": 0.00014765041563392125,
      "loss": 2.0252,
      "step": 154267
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.4542192220687866,
      "learning_rate": 0.0001476468918081097,
      "loss": 2.0474,
      "step": 154268
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.2733551263809204,
      "learning_rate": 0.00014764336801062358,
      "loss": 2.1308,
      "step": 154269
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.1099720001220703,
      "learning_rate": 0.00014763984424146333,
      "loss": 1.9837,
      "step": 154270
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.1795358657836914,
      "learning_rate": 0.00014763632050062981,
      "loss": 1.9035,
      "step": 154271
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.5213409662246704,
      "learning_rate": 0.0001476327967881236,
      "loss": 2.2828,
      "step": 154272
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.1403456926345825,
      "learning_rate": 0.0001476292731039453,
      "loss": 2.0039,
      "step": 154273
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.1810598373413086,
      "learning_rate": 0.00014762574944809549,
      "loss": 1.9103,
      "step": 154274
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.148843765258789,
      "learning_rate": 0.00014762222582057508,
      "loss": 2.0565,
      "step": 154275
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.3508965969085693,
      "learning_rate": 0.00014761870222138442,
      "loss": 2.1562,
      "step": 154276
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.2834678888320923,
      "learning_rate": 0.00014761517865052442,
      "loss": 2.108,
      "step": 154277
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.0328978300094604,
      "learning_rate": 0.00014761165510799568,
      "loss": 1.7661,
      "step": 154278
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.2830748558044434,
      "learning_rate": 0.00014760813159379876,
      "loss": 2.1394,
      "step": 154279
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.1401915550231934,
      "learning_rate": 0.00014760460810793428,
      "loss": 2.079,
      "step": 154280
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.1557397842407227,
      "learning_rate": 0.0001476010846504031,
      "loss": 2.1277,
      "step": 154281
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.0493032932281494,
      "learning_rate": 0.00014759756122120565,
      "loss": 2.0259,
      "step": 154282
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.0195938348770142,
      "learning_rate": 0.00014759403782034277,
      "loss": 2.1335,
      "step": 154283
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.3047860860824585,
      "learning_rate": 0.00014759051444781504,
      "loss": 1.9655,
      "step": 154284
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.2272984981536865,
      "learning_rate": 0.00014758699110362314,
      "loss": 2.1827,
      "step": 154285
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.2759616374969482,
      "learning_rate": 0.00014758346778776757,
      "loss": 2.1516,
      "step": 154286
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.0761799812316895,
      "learning_rate": 0.0001475799445002492,
      "loss": 1.9858,
      "step": 154287
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.0434417724609375,
      "learning_rate": 0.0001475764212410685,
      "loss": 1.919,
      "step": 154288
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.0880458354949951,
      "learning_rate": 0.00014757289801022635,
      "loss": 2.0771,
      "step": 154289
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.0663177967071533,
      "learning_rate": 0.00014756937480772325,
      "loss": 2.0385,
      "step": 154290
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.2714992761611938,
      "learning_rate": 0.0001475658516335599,
      "loss": 2.1027,
      "step": 154291
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.196044921875,
      "learning_rate": 0.00014756232848773678,
      "loss": 1.9814,
      "step": 154292
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.114463210105896,
      "learning_rate": 0.00014755880537025486,
      "loss": 1.9974,
      "step": 154293
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.189480185508728,
      "learning_rate": 0.0001475552822811145,
      "loss": 2.0636,
      "step": 154294
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.209469199180603,
      "learning_rate": 0.0001475517592203166,
      "loss": 1.9183,
      "step": 154295
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.1651694774627686,
      "learning_rate": 0.00014754823618786175,
      "loss": 1.7658,
      "step": 154296
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.0837677717208862,
      "learning_rate": 0.0001475447131837505,
      "loss": 1.9995,
      "step": 154297
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.1787152290344238,
      "learning_rate": 0.0001475411902079835,
      "loss": 2.2513,
      "step": 154298
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.0553251504898071,
      "learning_rate": 0.00014753766726056155,
      "loss": 2.0449,
      "step": 154299
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.2288234233856201,
      "learning_rate": 0.0001475341443414851,
      "loss": 1.9961,
      "step": 154300
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.1353009939193726,
      "learning_rate": 0.0001475306214507551,
      "loss": 2.2124,
      "step": 154301
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.185247778892517,
      "learning_rate": 0.000147527098588372,
      "loss": 2.078,
      "step": 154302
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.1366279125213623,
      "learning_rate": 0.0001475235757543364,
      "loss": 2.1665,
      "step": 154303
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.1518501043319702,
      "learning_rate": 0.00014752005294864912,
      "loss": 2.094,
      "step": 154304
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.1702171564102173,
      "learning_rate": 0.00014751653017131079,
      "loss": 2.0706,
      "step": 154305
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.150684118270874,
      "learning_rate": 0.00014751300742232187,
      "loss": 1.9516,
      "step": 154306
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.0877571105957031,
      "learning_rate": 0.0001475094847016833,
      "loss": 2.1203,
      "step": 154307
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.0299055576324463,
      "learning_rate": 0.00014750596200939554,
      "loss": 2.0499,
      "step": 154308
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.1717997789382935,
      "learning_rate": 0.00014750243934545922,
      "loss": 2.0979,
      "step": 154309
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.230536699295044,
      "learning_rate": 0.0001474989167098752,
      "loss": 2.0264,
      "step": 154310
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.1471154689788818,
      "learning_rate": 0.00014749539410264398,
      "loss": 2.1909,
      "step": 154311
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.1417336463928223,
      "learning_rate": 0.00014749187152376616,
      "loss": 2.0428,
      "step": 154312
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.0816184282302856,
      "learning_rate": 0.0001474883489732426,
      "loss": 2.1419,
      "step": 154313
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.195864200592041,
      "learning_rate": 0.0001474848264510738,
      "loss": 2.1655,
      "step": 154314
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.1246905326843262,
      "learning_rate": 0.00014748130395726036,
      "loss": 2.1095,
      "step": 154315
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.1909146308898926,
      "learning_rate": 0.00014747778149180313,
      "loss": 2.1443,
      "step": 154316
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.1276353597640991,
      "learning_rate": 0.00014747425905470255,
      "loss": 2.0357,
      "step": 154317
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.1094396114349365,
      "learning_rate": 0.0001474707366459595,
      "loss": 2.0279,
      "step": 154318
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.1186963319778442,
      "learning_rate": 0.00014746721426557452,
      "loss": 2.2149,
      "step": 154319
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.1297461986541748,
      "learning_rate": 0.00014746369191354823,
      "loss": 2.1437,
      "step": 154320
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.1922353506088257,
      "learning_rate": 0.00014746016958988128,
      "loss": 2.3435,
      "step": 154321
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.160809874534607,
      "learning_rate": 0.0001474566472945744,
      "loss": 1.9269,
      "step": 154322
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.3200181722640991,
      "learning_rate": 0.00014745312502762814,
      "loss": 2.042,
      "step": 154323
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.1037750244140625,
      "learning_rate": 0.00014744960278904333,
      "loss": 2.2035,
      "step": 154324
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.19392728805542,
      "learning_rate": 0.0001474460805788204,
      "loss": 1.8963,
      "step": 154325
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.159740686416626,
      "learning_rate": 0.00014744255839696036,
      "loss": 1.9664,
      "step": 154326
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.1338375806808472,
      "learning_rate": 0.00014743903624346339,
      "loss": 2.0162,
      "step": 154327
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.1406954526901245,
      "learning_rate": 0.00014743551411833047,
      "loss": 2.0423,
      "step": 154328
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.2065495252609253,
      "learning_rate": 0.0001474319920215621,
      "loss": 1.8455,
      "step": 154329
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.1116576194763184,
      "learning_rate": 0.0001474284699531591,
      "loss": 1.9569,
      "step": 154330
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.1352118253707886,
      "learning_rate": 0.0001474249479131219,
      "loss": 2.0499,
      "step": 154331
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.1917344331741333,
      "learning_rate": 0.0001474214259014515,
      "loss": 2.0677,
      "step": 154332
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.1322996616363525,
      "learning_rate": 0.00014741790391814815,
      "loss": 1.9805,
      "step": 154333
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.1728154420852661,
      "learning_rate": 0.0001474143819632128,
      "loss": 2.1358,
      "step": 154334
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.1619619131088257,
      "learning_rate": 0.00014741086003664588,
      "loss": 2.0794,
      "step": 154335
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.1646440029144287,
      "learning_rate": 0.00014740733813844826,
      "loss": 2.0864,
      "step": 154336
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.1486610174179077,
      "learning_rate": 0.0001474038162686204,
      "loss": 1.9135,
      "step": 154337
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.1952450275421143,
      "learning_rate": 0.00014740029442716313,
      "loss": 2.0673,
      "step": 154338
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.1755754947662354,
      "learning_rate": 0.00014739677261407707,
      "loss": 2.1085,
      "step": 154339
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.1802517175674438,
      "learning_rate": 0.00014739325082936282,
      "loss": 2.022,
      "step": 154340
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.2793508768081665,
      "learning_rate": 0.0001473897290730209,
      "loss": 2.1229,
      "step": 154341
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.256838083267212,
      "learning_rate": 0.00014738620734505223,
      "loss": 2.086,
      "step": 154342
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.2012512683868408,
      "learning_rate": 0.00014738268564545726,
      "loss": 2.171,
      "step": 154343
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.0769699811935425,
      "learning_rate": 0.00014737916397423683,
      "loss": 1.8979,
      "step": 154344
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.0948288440704346,
      "learning_rate": 0.0001473756423313915,
      "loss": 1.9222,
      "step": 154345
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.181228756904602,
      "learning_rate": 0.00014737212071692192,
      "loss": 2.1637,
      "step": 154346
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.2176648378372192,
      "learning_rate": 0.00014736859913082862,
      "loss": 2.2641,
      "step": 154347
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.3309131860733032,
      "learning_rate": 0.00014736507757311248,
      "loss": 1.8645,
      "step": 154348
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.243589997291565,
      "learning_rate": 0.00014736155604377395,
      "loss": 1.9533,
      "step": 154349
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.1087918281555176,
      "learning_rate": 0.00014735803454281392,
      "loss": 1.9163,
      "step": 154350
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.248852252960205,
      "learning_rate": 0.0001473545130702329,
      "loss": 2.0784,
      "step": 154351
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.2407740354537964,
      "learning_rate": 0.00014735099162603154,
      "loss": 2.226,
      "step": 154352
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.7418516874313354,
      "learning_rate": 0.00014734747021021043,
      "loss": 2.1175,
      "step": 154353
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.2126976251602173,
      "learning_rate": 0.00014734394882277038,
      "loss": 2.1167,
      "step": 154354
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.0824183225631714,
      "learning_rate": 0.0001473404274637119,
      "loss": 2.0904,
      "step": 154355
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.3380043506622314,
      "learning_rate": 0.00014733690613303582,
      "loss": 2.1908,
      "step": 154356
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.5103609561920166,
      "learning_rate": 0.00014733338483074266,
      "loss": 1.9686,
      "step": 154357
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.3124258518218994,
      "learning_rate": 0.00014732986355683314,
      "loss": 2.2557,
      "step": 154358
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.1108721494674683,
      "learning_rate": 0.00014732634231130775,
      "loss": 2.2909,
      "step": 154359
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.2119477987289429,
      "learning_rate": 0.00014732282109416743,
      "loss": 1.9087,
      "step": 154360
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.0688300132751465,
      "learning_rate": 0.00014731929990541252,
      "loss": 2.1286,
      "step": 154361
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.084224820137024,
      "learning_rate": 0.00014731577874504392,
      "loss": 1.8103,
      "step": 154362
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.2809784412384033,
      "learning_rate": 0.00014731225761306229,
      "loss": 2.0254,
      "step": 154363
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.1734869480133057,
      "learning_rate": 0.00014730873650946813,
      "loss": 2.088,
      "step": 154364
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.1553285121917725,
      "learning_rate": 0.00014730521543426204,
      "loss": 2.0654,
      "step": 154365
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.1733019351959229,
      "learning_rate": 0.00014730169438744494,
      "loss": 2.1683,
      "step": 154366
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.0613231658935547,
      "learning_rate": 0.0001472981733690172,
      "loss": 2.1133,
      "step": 154367
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.0704745054244995,
      "learning_rate": 0.0001472946523789797,
      "loss": 1.7209,
      "step": 154368
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.1507188081741333,
      "learning_rate": 0.00014729113141733307,
      "loss": 2.1213,
      "step": 154369
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.306801676750183,
      "learning_rate": 0.00014728761048407786,
      "loss": 2.0731,
      "step": 154370
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.1673446893692017,
      "learning_rate": 0.00014728408957921465,
      "loss": 2.2092,
      "step": 154371
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.168541669845581,
      "learning_rate": 0.00014728056870274436,
      "loss": 2.1639,
      "step": 154372
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.169532060623169,
      "learning_rate": 0.00014727704785466732,
      "loss": 2.1206,
      "step": 154373
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.1113834381103516,
      "learning_rate": 0.00014727352703498454,
      "loss": 1.986,
      "step": 154374
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.1089328527450562,
      "learning_rate": 0.00014727000624369644,
      "loss": 1.9399,
      "step": 154375
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.2525182962417603,
      "learning_rate": 0.0001472664854808038,
      "loss": 2.1334,
      "step": 154376
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.1196335554122925,
      "learning_rate": 0.000147262964746307,
      "loss": 2.2493,
      "step": 154377
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.1581923961639404,
      "learning_rate": 0.00014725944404020706,
      "loss": 2.2268,
      "step": 154378
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.2422739267349243,
      "learning_rate": 0.00014725592336250435,
      "loss": 2.0894,
      "step": 154379
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.1557140350341797,
      "learning_rate": 0.0001472524027131998,
      "loss": 2.2108,
      "step": 154380
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.1126149892807007,
      "learning_rate": 0.00014724888209229384,
      "loss": 2.1513,
      "step": 154381
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.2888115644454956,
      "learning_rate": 0.00014724536149978728,
      "loss": 2.0371,
      "step": 154382
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.2494535446166992,
      "learning_rate": 0.0001472418409356805,
      "loss": 2.2421,
      "step": 154383
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.172221064567566,
      "learning_rate": 0.00014723832039997448,
      "loss": 2.1426,
      "step": 154384
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.1218665838241577,
      "learning_rate": 0.00014723479989266964,
      "loss": 2.0446,
      "step": 154385
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.133026123046875,
      "learning_rate": 0.00014723127941376686,
      "loss": 2.3315,
      "step": 154386
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.1744836568832397,
      "learning_rate": 0.00014722775896326663,
      "loss": 2.0164,
      "step": 154387
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.222786784172058,
      "learning_rate": 0.00014722423854116954,
      "loss": 1.8876,
      "step": 154388
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.0631519556045532,
      "learning_rate": 0.00014722071814747648,
      "loss": 2.0108,
      "step": 154389
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.1440354585647583,
      "learning_rate": 0.00014721719778218798,
      "loss": 1.9909,
      "step": 154390
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.0752744674682617,
      "learning_rate": 0.00014721367744530454,
      "loss": 1.9877,
      "step": 154391
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.295396327972412,
      "learning_rate": 0.00014721015713682712,
      "loss": 1.9201,
      "step": 154392
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.4212661981582642,
      "learning_rate": 0.00014720663685675617,
      "loss": 1.9861,
      "step": 154393
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.1942745447158813,
      "learning_rate": 0.0001472031166050923,
      "loss": 2.0066,
      "step": 154394
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.2259931564331055,
      "learning_rate": 0.00014719959638183637,
      "loss": 1.8855,
      "step": 154395
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.2820556163787842,
      "learning_rate": 0.00014719607618698892,
      "loss": 2.0046,
      "step": 154396
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.0871011018753052,
      "learning_rate": 0.00014719255602055047,
      "loss": 2.3358,
      "step": 154397
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.1333011388778687,
      "learning_rate": 0.00014718903588252193,
      "loss": 2.2147,
      "step": 154398
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.364381194114685,
      "learning_rate": 0.00014718551577290386,
      "loss": 2.1125,
      "step": 154399
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.2145735025405884,
      "learning_rate": 0.00014718199569169673,
      "loss": 2.0919,
      "step": 154400
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.3110365867614746,
      "learning_rate": 0.0001471784756389015,
      "loss": 2.1782,
      "step": 154401
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.3635510206222534,
      "learning_rate": 0.00014717495561451853,
      "loss": 2.13,
      "step": 154402
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.4472172260284424,
      "learning_rate": 0.00014717143561854878,
      "loss": 1.9514,
      "step": 154403
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.2706754207611084,
      "learning_rate": 0.0001471679156509927,
      "loss": 2.2421,
      "step": 154404
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.2827646732330322,
      "learning_rate": 0.0001471643957118509,
      "loss": 2.0492,
      "step": 154405
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.4827367067337036,
      "learning_rate": 0.00014716087580112424,
      "loss": 1.8329,
      "step": 154406
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.4486466646194458,
      "learning_rate": 0.00014715735591881328,
      "loss": 2.0551,
      "step": 154407
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.168776273727417,
      "learning_rate": 0.0001471538360649185,
      "loss": 1.9581,
      "step": 154408
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.1985654830932617,
      "learning_rate": 0.0001471503162394408,
      "loss": 1.9825,
      "step": 154409
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.1966112852096558,
      "learning_rate": 0.00014714679644238079,
      "loss": 2.0039,
      "step": 154410
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.4461700916290283,
      "learning_rate": 0.00014714327667373893,
      "loss": 1.9205,
      "step": 154411
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.3787739276885986,
      "learning_rate": 0.00014713975693351615,
      "loss": 2.0207,
      "step": 154412
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.4052436351776123,
      "learning_rate": 0.00014713623722171297,
      "loss": 1.8257,
      "step": 154413
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.1122689247131348,
      "learning_rate": 0.00014713271753832993,
      "loss": 1.9842,
      "step": 154414
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.196115493774414,
      "learning_rate": 0.00014712919788336794,
      "loss": 2.0936,
      "step": 154415
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.4208869934082031,
      "learning_rate": 0.0001471256782568274,
      "loss": 2.2081,
      "step": 154416
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.403783917427063,
      "learning_rate": 0.00014712215865870918,
      "loss": 2.2503,
      "step": 154417
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.2950810194015503,
      "learning_rate": 0.00014711863908901384,
      "loss": 2.0871,
      "step": 154418
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.1452058553695679,
      "learning_rate": 0.00014711511954774206,
      "loss": 1.9265,
      "step": 154419
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.3517359495162964,
      "learning_rate": 0.0001471116000348943,
      "loss": 2.212,
      "step": 154420
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.2777425050735474,
      "learning_rate": 0.00014710808055047153,
      "loss": 2.3107,
      "step": 154421
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.2392646074295044,
      "learning_rate": 0.0001471045610944741,
      "loss": 2.1416,
      "step": 154422
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.2935858964920044,
      "learning_rate": 0.000147101041666903,
      "loss": 2.1495,
      "step": 154423
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.089084267616272,
      "learning_rate": 0.00014709752226775864,
      "loss": 1.9265,
      "step": 154424
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.4042333364486694,
      "learning_rate": 0.00014709400289704178,
      "loss": 2.0119,
      "step": 154425
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.1943283081054688,
      "learning_rate": 0.00014709048355475288,
      "loss": 1.8738,
      "step": 154426
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.2508609294891357,
      "learning_rate": 0.00014708696424089288,
      "loss": 2.0159,
      "step": 154427
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.1408799886703491,
      "learning_rate": 0.00014708344495546218,
      "loss": 2.0746,
      "step": 154428
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.0953361988067627,
      "learning_rate": 0.00014707992569846164,
      "loss": 1.9178,
      "step": 154429
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.2015740871429443,
      "learning_rate": 0.0001470764064698919,
      "loss": 2.1247,
      "step": 154430
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.2172396183013916,
      "learning_rate": 0.00014707288726975348,
      "loss": 2.1062,
      "step": 154431
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.3494985103607178,
      "learning_rate": 0.00014706936809804702,
      "loss": 2.1315,
      "step": 154432
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.0804270505905151,
      "learning_rate": 0.00014706584895477333,
      "loss": 1.9808,
      "step": 154433
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.2786469459533691,
      "learning_rate": 0.0001470623298399329,
      "loss": 2.1216,
      "step": 154434
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.4297130107879639,
      "learning_rate": 0.00014705881075352658,
      "loss": 1.9067,
      "step": 154435
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.2666049003601074,
      "learning_rate": 0.00014705529169555488,
      "loss": 1.9142,
      "step": 154436
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.1072622537612915,
      "learning_rate": 0.00014705177266601853,
      "loss": 2.0436,
      "step": 154437
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.2109971046447754,
      "learning_rate": 0.00014704825366491803,
      "loss": 1.9281,
      "step": 154438
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.2081502676010132,
      "learning_rate": 0.00014704473469225423,
      "loss": 2.0702,
      "step": 154439
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.5801489353179932,
      "learning_rate": 0.0001470412157480276,
      "loss": 2.0261,
      "step": 154440
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.2257236242294312,
      "learning_rate": 0.00014703769683223903,
      "loss": 2.3125,
      "step": 154441
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.115606665611267,
      "learning_rate": 0.00014703417794488904,
      "loss": 2.1255,
      "step": 154442
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.1495299339294434,
      "learning_rate": 0.00014703065908597824,
      "loss": 1.9987,
      "step": 154443
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.279772162437439,
      "learning_rate": 0.00014702714025550723,
      "loss": 2.0543,
      "step": 154444
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.241523265838623,
      "learning_rate": 0.00014702362145347687,
      "loss": 2.0886,
      "step": 154445
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.1196582317352295,
      "learning_rate": 0.00014702010267988764,
      "loss": 2.0369,
      "step": 154446
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.3373186588287354,
      "learning_rate": 0.00014701658393474035,
      "loss": 2.207,
      "step": 154447
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.3018523454666138,
      "learning_rate": 0.00014701306521803554,
      "loss": 2.0741,
      "step": 154448
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.2447420358657837,
      "learning_rate": 0.0001470095465297739,
      "loss": 2.2241,
      "step": 154449
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.1120989322662354,
      "learning_rate": 0.00014700602786995597,
      "loss": 2.4173,
      "step": 154450
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.1699308156967163,
      "learning_rate": 0.00014700250923858264,
      "loss": 1.8947,
      "step": 154451
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.223887324333191,
      "learning_rate": 0.00014699899063565428,
      "loss": 2.3382,
      "step": 154452
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.1504428386688232,
      "learning_rate": 0.00014699547206117183,
      "loss": 2.1116,
      "step": 154453
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.1270451545715332,
      "learning_rate": 0.00014699195351513582,
      "loss": 2.2386,
      "step": 154454
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.121940016746521,
      "learning_rate": 0.00014698843499754685,
      "loss": 2.1621,
      "step": 154455
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.2530670166015625,
      "learning_rate": 0.00014698491650840556,
      "loss": 1.9103,
      "step": 154456
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.1483306884765625,
      "learning_rate": 0.00014698139804771276,
      "loss": 2.1093,
      "step": 154457
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.1014539003372192,
      "learning_rate": 0.00014697787961546885,
      "loss": 2.0111,
      "step": 154458
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.1979701519012451,
      "learning_rate": 0.0001469743612116748,
      "loss": 1.9813,
      "step": 154459
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.323509931564331,
      "learning_rate": 0.0001469708428363311,
      "loss": 2.1793,
      "step": 154460
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.3263731002807617,
      "learning_rate": 0.00014696732448943838,
      "loss": 2.0509,
      "step": 154461
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.1673518419265747,
      "learning_rate": 0.00014696380617099724,
      "loss": 2.1115,
      "step": 154462
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.082395076751709,
      "learning_rate": 0.00014696028788100852,
      "loss": 2.0871,
      "step": 154463
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.1645704507827759,
      "learning_rate": 0.00014695676961947265,
      "loss": 1.7478,
      "step": 154464
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.0461225509643555,
      "learning_rate": 0.00014695325138639056,
      "loss": 1.8926,
      "step": 154465
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.1844333410263062,
      "learning_rate": 0.00014694973318176268,
      "loss": 1.9121,
      "step": 154466
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.0508711338043213,
      "learning_rate": 0.0001469462150055898,
      "loss": 2.1857,
      "step": 154467
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.1659846305847168,
      "learning_rate": 0.00014694269685787235,
      "loss": 2.1121,
      "step": 154468
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.1441700458526611,
      "learning_rate": 0.00014693917873861127,
      "loss": 1.9644,
      "step": 154469
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.2316055297851562,
      "learning_rate": 0.000146935660647807,
      "loss": 2.1571,
      "step": 154470
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.331931471824646,
      "learning_rate": 0.00014693214258546035,
      "loss": 2.2299,
      "step": 154471
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.1154297590255737,
      "learning_rate": 0.0001469286245515718,
      "loss": 2.0381,
      "step": 154472
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.0844887495040894,
      "learning_rate": 0.00014692510654614233,
      "loss": 1.8806,
      "step": 154473
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.2965850830078125,
      "learning_rate": 0.00014692158856917216,
      "loss": 2.0652,
      "step": 154474
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.1671754121780396,
      "learning_rate": 0.00014691807062066226,
      "loss": 1.9478,
      "step": 154475
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.1523523330688477,
      "learning_rate": 0.0001469145527006131,
      "loss": 2.0538,
      "step": 154476
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.2244073152542114,
      "learning_rate": 0.00014691103480902548,
      "loss": 1.8953,
      "step": 154477
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.1242879629135132,
      "learning_rate": 0.0001469075169458999,
      "loss": 2.0484,
      "step": 154478
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.0663478374481201,
      "learning_rate": 0.00014690399911123722,
      "loss": 1.8368,
      "step": 154479
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.2154217958450317,
      "learning_rate": 0.00014690048130503795,
      "loss": 2.0402,
      "step": 154480
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.1475510597229004,
      "learning_rate": 0.00014689696352730282,
      "loss": 2.1375,
      "step": 154481
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.1716049909591675,
      "learning_rate": 0.00014689344577803226,
      "loss": 2.1568,
      "step": 154482
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.1370017528533936,
      "learning_rate": 0.00014688992805722727,
      "loss": 2.1754,
      "step": 154483
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.1262016296386719,
      "learning_rate": 0.00014688641036488816,
      "loss": 1.8885,
      "step": 154484
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.037131905555725,
      "learning_rate": 0.00014688289270101593,
      "loss": 1.9535,
      "step": 154485
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.0973631143569946,
      "learning_rate": 0.00014687937506561102,
      "loss": 2.0286,
      "step": 154486
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.3346487283706665,
      "learning_rate": 0.00014687585745867403,
      "loss": 2.0857,
      "step": 154487
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.202164888381958,
      "learning_rate": 0.00014687233988020583,
      "loss": 1.8797,
      "step": 154488
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.2775330543518066,
      "learning_rate": 0.00014686882233020692,
      "loss": 2.1137,
      "step": 154489
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.1855053901672363,
      "learning_rate": 0.0001468653048086779,
      "loss": 1.9969,
      "step": 154490
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.1908271312713623,
      "learning_rate": 0.0001468617873156196,
      "loss": 1.9498,
      "step": 154491
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.15067720413208,
      "learning_rate": 0.00014685826985103261,
      "loss": 2.2857,
      "step": 154492
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.3639566898345947,
      "learning_rate": 0.00014685475241491743,
      "loss": 2.0603,
      "step": 154493
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.308298945426941,
      "learning_rate": 0.00014685123500727496,
      "loss": 2.014,
      "step": 154494
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.1632612943649292,
      "learning_rate": 0.00014684771762810573,
      "loss": 2.3385,
      "step": 154495
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.1597062349319458,
      "learning_rate": 0.00014684420027741028,
      "loss": 2.0972,
      "step": 154496
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.0930960178375244,
      "learning_rate": 0.00014684068295518955,
      "loss": 2.145,
      "step": 154497
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.2580243349075317,
      "learning_rate": 0.00014683716566144398,
      "loss": 1.9702,
      "step": 154498
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.1675740480422974,
      "learning_rate": 0.00014683364839617415,
      "loss": 1.9871,
      "step": 154499
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.1776628494262695,
      "learning_rate": 0.00014683013115938097,
      "loss": 2.0064,
      "step": 154500
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.4515528678894043,
      "learning_rate": 0.00014682661395106484,
      "loss": 2.2243,
      "step": 154501
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.1151211261749268,
      "learning_rate": 0.00014682309677122667,
      "loss": 2.0116,
      "step": 154502
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.084012746810913,
      "learning_rate": 0.00014681957961986693,
      "loss": 2.0867,
      "step": 154503
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.1386539936065674,
      "learning_rate": 0.00014681606249698637,
      "loss": 1.9419,
      "step": 154504
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.0967955589294434,
      "learning_rate": 0.00014681254540258544,
      "loss": 2.2646,
      "step": 154505
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.1105557680130005,
      "learning_rate": 0.00014680902833666512,
      "loss": 1.9221,
      "step": 154506
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.3389841318130493,
      "learning_rate": 0.00014680551129922574,
      "loss": 2.0329,
      "step": 154507
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.2370721101760864,
      "learning_rate": 0.00014680199429026824,
      "loss": 1.9853,
      "step": 154508
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.154227375984192,
      "learning_rate": 0.00014679847730979313,
      "loss": 1.9701,
      "step": 154509
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.1760324239730835,
      "learning_rate": 0.00014679496035780105,
      "loss": 2.1783,
      "step": 154510
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.360021710395813,
      "learning_rate": 0.00014679144343429258,
      "loss": 2.0685,
      "step": 154511
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.1944528818130493,
      "learning_rate": 0.0001467879265392686,
      "loss": 2.231,
      "step": 154512
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.2858092784881592,
      "learning_rate": 0.0001467844096727295,
      "loss": 2.1672,
      "step": 154513
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.125520944595337,
      "learning_rate": 0.0001467808928346762,
      "loss": 1.9111,
      "step": 154514
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.1572896242141724,
      "learning_rate": 0.00014677737602510925,
      "loss": 1.8592,
      "step": 154515
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.1593430042266846,
      "learning_rate": 0.00014677385924402925,
      "loss": 2.0379,
      "step": 154516
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.2344905138015747,
      "learning_rate": 0.00014677034249143675,
      "loss": 2.0651,
      "step": 154517
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.108096718788147,
      "learning_rate": 0.00014676682576733268,
      "loss": 2.1722,
      "step": 154518
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.1590100526809692,
      "learning_rate": 0.00014676330907171741,
      "loss": 2.0079,
      "step": 154519
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.0871845483779907,
      "learning_rate": 0.0001467597924045919,
      "loss": 1.9986,
      "step": 154520
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.1496241092681885,
      "learning_rate": 0.0001467562757659566,
      "loss": 2.0531,
      "step": 154521
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.1146899461746216,
      "learning_rate": 0.00014675275915581218,
      "loss": 2.0453,
      "step": 154522
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.180599570274353,
      "learning_rate": 0.00014674924257415922,
      "loss": 1.8493,
      "step": 154523
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.2671276330947876,
      "learning_rate": 0.0001467457260209986,
      "loss": 2.1449,
      "step": 154524
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.101584553718567,
      "learning_rate": 0.0001467422094963307,
      "loss": 2.0547,
      "step": 154525
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.33372962474823,
      "learning_rate": 0.00014673869300015646,
      "loss": 2.2224,
      "step": 154526
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.305299997329712,
      "learning_rate": 0.00014673517653247635,
      "loss": 2.1296,
      "step": 154527
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.194827914237976,
      "learning_rate": 0.0001467316600932911,
      "loss": 2.0979,
      "step": 154528
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.2040785551071167,
      "learning_rate": 0.0001467281436826012,
      "loss": 1.9641,
      "step": 154529
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.146251916885376,
      "learning_rate": 0.00014672462730040757,
      "loss": 2.2992,
      "step": 154530
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.2010855674743652,
      "learning_rate": 0.00014672111094671058,
      "loss": 2.2255,
      "step": 154531
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.1576251983642578,
      "learning_rate": 0.0001467175946215111,
      "loss": 2.0852,
      "step": 154532
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.112549066543579,
      "learning_rate": 0.00014671407832480982,
      "loss": 1.8286,
      "step": 154533
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.1964967250823975,
      "learning_rate": 0.0001467105620566072,
      "loss": 2.163,
      "step": 154534
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.0858511924743652,
      "learning_rate": 0.0001467070458169039,
      "loss": 2.0342,
      "step": 154535
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.064438819885254,
      "learning_rate": 0.00014670352960570077,
      "loss": 1.9807,
      "step": 154536
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.0700594186782837,
      "learning_rate": 0.00014670001342299824,
      "loss": 2.0441,
      "step": 154537
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.1447337865829468,
      "learning_rate": 0.00014669649726879716,
      "loss": 2.1687,
      "step": 154538
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.1362340450286865,
      "learning_rate": 0.000146692981143098,
      "loss": 2.0825,
      "step": 154539
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.1692415475845337,
      "learning_rate": 0.00014668946504590172,
      "loss": 1.9321,
      "step": 154540
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.2251585721969604,
      "learning_rate": 0.00014668594897720852,
      "loss": 1.9055,
      "step": 154541
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.2380502223968506,
      "learning_rate": 0.00014668243293701946,
      "loss": 2.065,
      "step": 154542
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.316258192062378,
      "learning_rate": 0.0001466789169253349,
      "loss": 1.8676,
      "step": 154543
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.1218724250793457,
      "learning_rate": 0.00014667540094215572,
      "loss": 2.2243,
      "step": 154544
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.210893988609314,
      "learning_rate": 0.00014667188498748235,
      "loss": 1.9054,
      "step": 154545
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.1341321468353271,
      "learning_rate": 0.00014666836906131585,
      "loss": 2.1391,
      "step": 154546
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.1183079481124878,
      "learning_rate": 0.00014666485316365633,
      "loss": 2.083,
      "step": 154547
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.161331057548523,
      "learning_rate": 0.00014666133729450482,
      "loss": 1.8754,
      "step": 154548
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.1801340579986572,
      "learning_rate": 0.00014665782145386176,
      "loss": 2.0867,
      "step": 154549
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.31455397605896,
      "learning_rate": 0.00014665430564172801,
      "loss": 1.9003,
      "step": 154550
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.3838838338851929,
      "learning_rate": 0.000146650789858104,
      "loss": 1.9957,
      "step": 154551
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.272077202796936,
      "learning_rate": 0.0001466472741029908,
      "loss": 1.9596,
      "step": 154552
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.1823188066482544,
      "learning_rate": 0.00014664375837638842,
      "loss": 1.9146,
      "step": 154553
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.2206463813781738,
      "learning_rate": 0.00014664024267829805,
      "loss": 1.9356,
      "step": 154554
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.2250289916992188,
      "learning_rate": 0.00014663672700872003,
      "loss": 2.1836,
      "step": 154555
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.2277555465698242,
      "learning_rate": 0.00014663321136765525,
      "loss": 2.1693,
      "step": 154556
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.1520358324050903,
      "learning_rate": 0.00014662969575510412,
      "loss": 1.8437,
      "step": 154557
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.2272765636444092,
      "learning_rate": 0.00014662618017106767,
      "loss": 1.9316,
      "step": 154558
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.140092134475708,
      "learning_rate": 0.00014662266461554606,
      "loss": 2.0406,
      "step": 154559
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.2196472883224487,
      "learning_rate": 0.00014661914908854033,
      "loss": 1.7833,
      "step": 154560
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.209377646446228,
      "learning_rate": 0.00014661563359005087,
      "loss": 2.2416,
      "step": 154561
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.1020362377166748,
      "learning_rate": 0.0001466121181200786,
      "loss": 2.0009,
      "step": 154562
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.0692813396453857,
      "learning_rate": 0.0001466086026786239,
      "loss": 2.0504,
      "step": 154563
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.1164100170135498,
      "learning_rate": 0.00014660508726568763,
      "loss": 2.0076,
      "step": 154564
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.2119746208190918,
      "learning_rate": 0.00014660157188127045,
      "loss": 1.8742,
      "step": 154565
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.2390358448028564,
      "learning_rate": 0.00014659805652537288,
      "loss": 1.9991,
      "step": 154566
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.2143259048461914,
      "learning_rate": 0.00014659454119799552,
      "loss": 2.093,
      "step": 154567
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.0668659210205078,
      "learning_rate": 0.00014659102589913923,
      "loss": 2.1445,
      "step": 154568
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.1294125318527222,
      "learning_rate": 0.00014658751062880448,
      "loss": 2.1426,
      "step": 154569
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.2512562274932861,
      "learning_rate": 0.00014658399538699209,
      "loss": 2.1006,
      "step": 154570
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.2140265703201294,
      "learning_rate": 0.00014658048017370266,
      "loss": 2.0572,
      "step": 154571
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.1535024642944336,
      "learning_rate": 0.0001465769649889368,
      "loss": 2.0012,
      "step": 154572
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.0830579996109009,
      "learning_rate": 0.00014657344983269507,
      "loss": 2.1719,
      "step": 154573
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.16521418094635,
      "learning_rate": 0.00014656993470497833,
      "loss": 1.7349,
      "step": 154574
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.2631534337997437,
      "learning_rate": 0.00014656641960578707,
      "loss": 2.1226,
      "step": 154575
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.0883028507232666,
      "learning_rate": 0.00014656290453512205,
      "loss": 2.0622,
      "step": 154576
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.199958086013794,
      "learning_rate": 0.00014655938949298392,
      "loss": 2.1439,
      "step": 154577
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.228137493133545,
      "learning_rate": 0.00014655587447937317,
      "loss": 1.7804,
      "step": 154578
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.195380687713623,
      "learning_rate": 0.00014655235949429072,
      "loss": 2.0345,
      "step": 154579
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.3650739192962646,
      "learning_rate": 0.00014654884453773706,
      "loss": 2.0596,
      "step": 154580
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.239502191543579,
      "learning_rate": 0.0001465453296097128,
      "loss": 2.0129,
      "step": 154581
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.2391482591629028,
      "learning_rate": 0.00014654181471021873,
      "loss": 2.1563,
      "step": 154582
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.1615949869155884,
      "learning_rate": 0.0001465382998392554,
      "loss": 2.0267,
      "step": 154583
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.2213866710662842,
      "learning_rate": 0.00014653478499682343,
      "loss": 2.2122,
      "step": 154584
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.0709961652755737,
      "learning_rate": 0.00014653127018292368,
      "loss": 1.974,
      "step": 154585
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.3792541027069092,
      "learning_rate": 0.0001465277553975565,
      "loss": 2.2158,
      "step": 154586
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.2538750171661377,
      "learning_rate": 0.00014652424064072285,
      "loss": 2.3296,
      "step": 154587
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.1252862215042114,
      "learning_rate": 0.00014652072591242322,
      "loss": 1.9059,
      "step": 154588
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.157126545906067,
      "learning_rate": 0.0001465172112126583,
      "loss": 2.0982,
      "step": 154589
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.0722458362579346,
      "learning_rate": 0.0001465136965414286,
      "loss": 2.0989,
      "step": 154590
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.1516293287277222,
      "learning_rate": 0.000146510181898735,
      "loss": 1.7715,
      "step": 154591
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.1232243776321411,
      "learning_rate": 0.00014650666728457799,
      "loss": 2.2933,
      "step": 154592
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.145808219909668,
      "learning_rate": 0.00014650315269895833,
      "loss": 2.0752,
      "step": 154593
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.2470545768737793,
      "learning_rate": 0.0001464996381418767,
      "loss": 1.8616,
      "step": 154594
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.1106433868408203,
      "learning_rate": 0.00014649612361333365,
      "loss": 1.9009,
      "step": 154595
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.1450629234313965,
      "learning_rate": 0.00014649260911332975,
      "loss": 2.0196,
      "step": 154596
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.1384291648864746,
      "learning_rate": 0.0001464890946418659,
      "loss": 2.0267,
      "step": 154597
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.230002760887146,
      "learning_rate": 0.00014648558019894254,
      "loss": 2.0314,
      "step": 154598
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.1636857986450195,
      "learning_rate": 0.0001464820657845605,
      "loss": 2.2134,
      "step": 154599
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.167875051498413,
      "learning_rate": 0.00014647855139872032,
      "loss": 2.0483,
      "step": 154600
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.212225079536438,
      "learning_rate": 0.00014647503704142266,
      "loss": 2.0678,
      "step": 154601
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.374011516571045,
      "learning_rate": 0.00014647152271266814,
      "loss": 1.8671,
      "step": 154602
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.2334439754486084,
      "learning_rate": 0.00014646800841245752,
      "loss": 1.9624,
      "step": 154603
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.1632558107376099,
      "learning_rate": 0.0001464644941407913,
      "loss": 2.0947,
      "step": 154604
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.1029345989227295,
      "learning_rate": 0.00014646097989767036,
      "loss": 1.9905,
      "step": 154605
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.1316823959350586,
      "learning_rate": 0.00014645746568309507,
      "loss": 2.0649,
      "step": 154606
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.2114639282226562,
      "learning_rate": 0.0001464539514970665,
      "loss": 2.0198,
      "step": 154607
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.2709264755249023,
      "learning_rate": 0.0001464504373395848,
      "loss": 2.1501,
      "step": 154608
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.2186769247055054,
      "learning_rate": 0.00014644692321065097,
      "loss": 1.9814,
      "step": 154609
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.2047125101089478,
      "learning_rate": 0.00014644340911026542,
      "loss": 1.9299,
      "step": 154610
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.2803678512573242,
      "learning_rate": 0.0001464398950384291,
      "loss": 2.0532,
      "step": 154611
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.2033531665802002,
      "learning_rate": 0.00014643638099514234,
      "loss": 2.0229,
      "step": 154612
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.1778459548950195,
      "learning_rate": 0.00014643286698040622,
      "loss": 1.8578,
      "step": 154613
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.1300936937332153,
      "learning_rate": 0.0001464293529942209,
      "loss": 2.0901,
      "step": 154614
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.1139483451843262,
      "learning_rate": 0.00014642583903658735,
      "loss": 2.0897,
      "step": 154615
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.1540292501449585,
      "learning_rate": 0.00014642232510750605,
      "loss": 1.9632,
      "step": 154616
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.1719304323196411,
      "learning_rate": 0.00014641881120697786,
      "loss": 1.9432,
      "step": 154617
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.2398427724838257,
      "learning_rate": 0.00014641529733500316,
      "loss": 2.2874,
      "step": 154618
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.1938776969909668,
      "learning_rate": 0.000146411783491583,
      "loss": 1.7926,
      "step": 154619
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.2561527490615845,
      "learning_rate": 0.00014640826967671756,
      "loss": 1.9307,
      "step": 154620
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.2338383197784424,
      "learning_rate": 0.00014640475589040785,
      "loss": 1.9305,
      "step": 154621
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.119035243988037,
      "learning_rate": 0.00014640124213265428,
      "loss": 1.8705,
      "step": 154622
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.1982887983322144,
      "learning_rate": 0.0001463977284034577,
      "loss": 1.9307,
      "step": 154623
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.0323817729949951,
      "learning_rate": 0.00014639421470281862,
      "loss": 2.1217,
      "step": 154624
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.1854867935180664,
      "learning_rate": 0.00014639070103073795,
      "loss": 2.093,
      "step": 154625
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.13011634349823,
      "learning_rate": 0.0001463871873872159,
      "loss": 2.0586,
      "step": 154626
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.528589129447937,
      "learning_rate": 0.00014638367377225352,
      "loss": 1.9437,
      "step": 154627
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.140939712524414,
      "learning_rate": 0.00014638016018585117,
      "loss": 1.8871,
      "step": 154628
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.1366699934005737,
      "learning_rate": 0.0001463766466280098,
      "loss": 2.1797,
      "step": 154629
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.1409502029418945,
      "learning_rate": 0.00014637313309872978,
      "loss": 2.2121,
      "step": 154630
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.1117324829101562,
      "learning_rate": 0.00014636961959801212,
      "loss": 1.9468,
      "step": 154631
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.1295497417449951,
      "learning_rate": 0.00014636610612585703,
      "loss": 2.0133,
      "step": 154632
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.1252763271331787,
      "learning_rate": 0.00014636259268226547,
      "loss": 2.1983,
      "step": 154633
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.1721845865249634,
      "learning_rate": 0.0001463590792672379,
      "loss": 2.0887,
      "step": 154634
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.3585176467895508,
      "learning_rate": 0.00014635556588077519,
      "loss": 2.162,
      "step": 154635
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.1735421419143677,
      "learning_rate": 0.0001463520525228778,
      "loss": 1.8049,
      "step": 154636
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.1046010255813599,
      "learning_rate": 0.00014634853919354667,
      "loss": 1.9829,
      "step": 154637
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.1464263200759888,
      "learning_rate": 0.00014634502589278201,
      "loss": 1.9696,
      "step": 154638
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.125479817390442,
      "learning_rate": 0.00014634151262058482,
      "loss": 2.1748,
      "step": 154639
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.048091173171997,
      "learning_rate": 0.00014633799937695553,
      "loss": 2.169,
      "step": 154640
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.1942838430404663,
      "learning_rate": 0.00014633448616189503,
      "loss": 2.0949,
      "step": 154641
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.1399123668670654,
      "learning_rate": 0.00014633097297540374,
      "loss": 1.9593,
      "step": 154642
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.1789097785949707,
      "learning_rate": 0.00014632745981748262,
      "loss": 2.0796,
      "step": 154643
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.1832044124603271,
      "learning_rate": 0.00014632394668813193,
      "loss": 2.3052,
      "step": 154644
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.1844711303710938,
      "learning_rate": 0.00014632043358735265,
      "loss": 1.8116,
      "step": 154645
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.193610668182373,
      "learning_rate": 0.00014631692051514513,
      "loss": 1.9117,
      "step": 154646
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.3046494722366333,
      "learning_rate": 0.00014631340747151038,
      "loss": 1.9646,
      "step": 154647
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.1411844491958618,
      "learning_rate": 0.0001463098944564487,
      "loss": 2.005,
      "step": 154648
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.1367366313934326,
      "learning_rate": 0.00014630638146996105,
      "loss": 2.0917,
      "step": 154649
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.1675348281860352,
      "learning_rate": 0.00014630286851204796,
      "loss": 2.1545,
      "step": 154650
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.246726632118225,
      "learning_rate": 0.00014629935558271002,
      "loss": 2.1213,
      "step": 154651
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.202715277671814,
      "learning_rate": 0.00014629584268194786,
      "loss": 2.1172,
      "step": 154652
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.1508756875991821,
      "learning_rate": 0.00014629232980976228,
      "loss": 2.2252,
      "step": 154653
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.254118800163269,
      "learning_rate": 0.00014628881696615378,
      "loss": 2.1231,
      "step": 154654
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.065989375114441,
      "learning_rate": 0.0001462853041511232,
      "loss": 2.0168,
      "step": 154655
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.1837397813796997,
      "learning_rate": 0.00014628179136467106,
      "loss": 1.8978,
      "step": 154656
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.3692903518676758,
      "learning_rate": 0.00014627827860679804,
      "loss": 1.9874,
      "step": 154657
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.2531931400299072,
      "learning_rate": 0.0001462747658775047,
      "loss": 2.2558,
      "step": 154658
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.2848973274230957,
      "learning_rate": 0.00014627125317679188,
      "loss": 2.0043,
      "step": 154659
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.1258505582809448,
      "learning_rate": 0.00014626774050466003,
      "loss": 2.1666,
      "step": 154660
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.0955911874771118,
      "learning_rate": 0.00014626422786111,
      "loss": 2.1013,
      "step": 154661
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.2941243648529053,
      "learning_rate": 0.00014626071524614237,
      "loss": 2.1083,
      "step": 154662
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.1154829263687134,
      "learning_rate": 0.0001462572026597577,
      "loss": 1.9085,
      "step": 154663
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.2128554582595825,
      "learning_rate": 0.00014625369010195678,
      "loss": 2.0879,
      "step": 154664
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.1614388227462769,
      "learning_rate": 0.0001462501775727402,
      "loss": 1.9581,
      "step": 154665
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.290523648262024,
      "learning_rate": 0.00014624666507210852,
      "loss": 1.8636,
      "step": 154666
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.2128872871398926,
      "learning_rate": 0.0001462431526000626,
      "loss": 2.3133,
      "step": 154667
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.0462034940719604,
      "learning_rate": 0.00014623964015660296,
      "loss": 1.9686,
      "step": 154668
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.1448049545288086,
      "learning_rate": 0.00014623612774173017,
      "loss": 2.2212,
      "step": 154669
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.2594997882843018,
      "learning_rate": 0.0001462326153554451,
      "loss": 2.1508,
      "step": 154670
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.0909432172775269,
      "learning_rate": 0.00014622910299774832,
      "loss": 1.9797,
      "step": 154671
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.1133880615234375,
      "learning_rate": 0.0001462255906686403,
      "loss": 1.9245,
      "step": 154672
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.2742232084274292,
      "learning_rate": 0.00014622207836812197,
      "loss": 2.0122,
      "step": 154673
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.0943149328231812,
      "learning_rate": 0.0001462185660961939,
      "loss": 2.2393,
      "step": 154674
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.4404600858688354,
      "learning_rate": 0.00014621505385285652,
      "loss": 1.974,
      "step": 154675
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.1676844358444214,
      "learning_rate": 0.00014621154163811081,
      "loss": 2.0228,
      "step": 154676
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.1612322330474854,
      "learning_rate": 0.00014620802945195717,
      "loss": 2.1012,
      "step": 154677
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.2654328346252441,
      "learning_rate": 0.00014620451729439646,
      "loss": 1.8735,
      "step": 154678
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.2358092069625854,
      "learning_rate": 0.00014620100516542922,
      "loss": 1.9064,
      "step": 154679
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.1456482410430908,
      "learning_rate": 0.00014619749306505613,
      "loss": 1.9434,
      "step": 154680
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.117396354675293,
      "learning_rate": 0.00014619398099327774,
      "loss": 2.0858,
      "step": 154681
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.2430944442749023,
      "learning_rate": 0.0001461904689500949,
      "loss": 2.2326,
      "step": 154682
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.1443616151809692,
      "learning_rate": 0.00014618695693550802,
      "loss": 1.9164,
      "step": 154683
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.2466398477554321,
      "learning_rate": 0.000146183444949518,
      "loss": 2.0607,
      "step": 154684
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.3205335140228271,
      "learning_rate": 0.0001461799329921253,
      "loss": 1.9447,
      "step": 154685
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.280401349067688,
      "learning_rate": 0.00014617642106333085,
      "loss": 2.1879,
      "step": 154686
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.1552561521530151,
      "learning_rate": 0.00014617290916313485,
      "loss": 2.1356,
      "step": 154687
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.152575135231018,
      "learning_rate": 0.00014616939729153837,
      "loss": 2.188,
      "step": 154688
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.212077021598816,
      "learning_rate": 0.0001461658854485418,
      "loss": 2.1371,
      "step": 154689
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.2193728685379028,
      "learning_rate": 0.00014616237363414598,
      "loss": 2.2113,
      "step": 154690
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.2433958053588867,
      "learning_rate": 0.00014615886184835139,
      "loss": 1.9821,
      "step": 154691
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.2024918794631958,
      "learning_rate": 0.00014615535009115897,
      "loss": 2.1224,
      "step": 154692
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.249171495437622,
      "learning_rate": 0.00014615183836256895,
      "loss": 2.0136,
      "step": 154693
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.132827877998352,
      "learning_rate": 0.00014614832666258234,
      "loss": 2.115,
      "step": 154694
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.2534452676773071,
      "learning_rate": 0.00014614481499119954,
      "loss": 1.9524,
      "step": 154695
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.1764276027679443,
      "learning_rate": 0.00014614130334842144,
      "loss": 2.352,
      "step": 154696
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.0969234704971313,
      "learning_rate": 0.00014613779173424846,
      "loss": 1.6787,
      "step": 154697
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.005136489868164,
      "learning_rate": 0.0001461342801486816,
      "loss": 2.0118,
      "step": 154698
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.173762559890747,
      "learning_rate": 0.00014613076859172104,
      "loss": 2.2554,
      "step": 154699
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.0868667364120483,
      "learning_rate": 0.0001461272570633678,
      "loss": 2.138,
      "step": 154700
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.1258161067962646,
      "learning_rate": 0.0001461237455636223,
      "loss": 1.9197,
      "step": 154701
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.2283200025558472,
      "learning_rate": 0.0001461202340924854,
      "loss": 1.9921,
      "step": 154702
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.0841636657714844,
      "learning_rate": 0.00014611672264995754,
      "loss": 1.9005,
      "step": 154703
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.2170329093933105,
      "learning_rate": 0.00014611321123603972,
      "loss": 2.0384,
      "step": 154704
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.314584732055664,
      "learning_rate": 0.00014610969985073214,
      "loss": 2.0029,
      "step": 154705
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.189821481704712,
      "learning_rate": 0.00014610618849403577,
      "loss": 2.0492,
      "step": 154706
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.062424898147583,
      "learning_rate": 0.00014610267716595103,
      "loss": 1.9671,
      "step": 154707
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.239209532737732,
      "learning_rate": 0.00014609916586647888,
      "loss": 2.1528,
      "step": 154708
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.1697125434875488,
      "learning_rate": 0.00014609565459561963,
      "loss": 1.7773,
      "step": 154709
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.1236785650253296,
      "learning_rate": 0.00014609214335337437,
      "loss": 2.0909,
      "step": 154710
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.1431846618652344,
      "learning_rate": 0.00014608863213974322,
      "loss": 2.0778,
      "step": 154711
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.1412720680236816,
      "learning_rate": 0.00014608512095472725,
      "loss": 2.1748,
      "step": 154712
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.0652564764022827,
      "learning_rate": 0.00014608160979832682,
      "loss": 1.962,
      "step": 154713
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.0442700386047363,
      "learning_rate": 0.00014607809867054285,
      "loss": 1.9378,
      "step": 154714
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.1502388715744019,
      "learning_rate": 0.00014607458757137577,
      "loss": 2.0754,
      "step": 154715
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.1320953369140625,
      "learning_rate": 0.00014607107650082654,
      "loss": 2.0751,
      "step": 154716
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.1221815347671509,
      "learning_rate": 0.0001460675654588954,
      "loss": 1.9816,
      "step": 154717
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.4985649585723877,
      "learning_rate": 0.00014606405444558324,
      "loss": 2.0683,
      "step": 154718
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.1659060716629028,
      "learning_rate": 0.00014606054346089065,
      "loss": 1.9904,
      "step": 154719
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.2125095129013062,
      "learning_rate": 0.00014605703250481836,
      "loss": 1.8355,
      "step": 154720
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.178729772567749,
      "learning_rate": 0.0001460535215773669,
      "loss": 1.9904,
      "step": 154721
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.4290533065795898,
      "learning_rate": 0.0001460500106785372,
      "loss": 2.0716,
      "step": 154722
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.1352847814559937,
      "learning_rate": 0.0001460464998083295,
      "loss": 2.17,
      "step": 154723
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.2003169059753418,
      "learning_rate": 0.0001460429889667448,
      "loss": 2.058,
      "step": 154724
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.52876877784729,
      "learning_rate": 0.00014603947815378348,
      "loss": 2.075,
      "step": 154725
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.1663295030593872,
      "learning_rate": 0.00014603596736944644,
      "loss": 1.9133,
      "step": 154726
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.101836919784546,
      "learning_rate": 0.0001460324566137341,
      "loss": 1.9824,
      "step": 154727
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.1037051677703857,
      "learning_rate": 0.00014602894588664744,
      "loss": 1.8604,
      "step": 154728
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.1710684299468994,
      "learning_rate": 0.00014602543518818671,
      "loss": 2.0304,
      "step": 154729
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.094399094581604,
      "learning_rate": 0.00014602192451835288,
      "loss": 2.1564,
      "step": 154730
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.2494533061981201,
      "learning_rate": 0.00014601841387714632,
      "loss": 2.032,
      "step": 154731
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.2942509651184082,
      "learning_rate": 0.000146014903264568,
      "loss": 2.1051,
      "step": 154732
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.3109087944030762,
      "learning_rate": 0.00014601139268061828,
      "loss": 1.921,
      "step": 154733
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.4405204057693481,
      "learning_rate": 0.00014600788212529818,
      "loss": 2.012,
      "step": 154734
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.10886549949646,
      "learning_rate": 0.0001460043715986079,
      "loss": 1.9126,
      "step": 154735
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.227365493774414,
      "learning_rate": 0.00014600086110054842,
      "loss": 1.9524,
      "step": 154736
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.2952686548233032,
      "learning_rate": 0.00014599735063112017,
      "loss": 1.9923,
      "step": 154737
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.47959566116333,
      "learning_rate": 0.00014599384019032408,
      "loss": 2.0898,
      "step": 154738
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.1513158082962036,
      "learning_rate": 0.00014599032977816048,
      "loss": 1.9634,
      "step": 154739
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.2395521402359009,
      "learning_rate": 0.00014598681939463032,
      "loss": 2.0196,
      "step": 154740
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.1880186796188354,
      "learning_rate": 0.0001459833090397341,
      "loss": 2.2464,
      "step": 154741
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.235691785812378,
      "learning_rate": 0.0001459797987134725,
      "loss": 2.0014,
      "step": 154742
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.2829819917678833,
      "learning_rate": 0.00014597628841584606,
      "loss": 2.0335,
      "step": 154743
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.2243242263793945,
      "learning_rate": 0.0001459727781468556,
      "loss": 2.203,
      "step": 154744
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.118794322013855,
      "learning_rate": 0.00014596926790650164,
      "loss": 2.1299,
      "step": 154745
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.4043024778366089,
      "learning_rate": 0.00014596575769478502,
      "loss": 2.0825,
      "step": 154746
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.5039795637130737,
      "learning_rate": 0.00014596224751170625,
      "loss": 1.8108,
      "step": 154747
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.308898687362671,
      "learning_rate": 0.0001459587373572659,
      "loss": 2.1523,
      "step": 154748
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.1877754926681519,
      "learning_rate": 0.00014595522723146484,
      "loss": 1.8959,
      "step": 154749
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.2770068645477295,
      "learning_rate": 0.0001459517171343036,
      "loss": 2.1749,
      "step": 154750
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.3220689296722412,
      "learning_rate": 0.00014594820706578275,
      "loss": 1.8435,
      "step": 154751
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.131980538368225,
      "learning_rate": 0.00014594469702590315,
      "loss": 2.1245,
      "step": 154752
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.2159969806671143,
      "learning_rate": 0.00014594118701466533,
      "loss": 1.9519,
      "step": 154753
    },
    {
      "epoch": 2.01,
      "grad_norm": 1.1481329202651978,
      "learning_rate": 0.00014593767703206982,
      "loss": 2.0077,
      "step": 154754
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.1178624629974365,
      "learning_rate": 0.00014593416707811752,
      "loss": 1.9771,
      "step": 154755
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.0994740724563599,
      "learning_rate": 0.000145930657152809,
      "loss": 2.1213,
      "step": 154756
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.1156203746795654,
      "learning_rate": 0.00014592714725614477,
      "loss": 1.9002,
      "step": 154757
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.1738723516464233,
      "learning_rate": 0.00014592363738812566,
      "loss": 2.1269,
      "step": 154758
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.145646095275879,
      "learning_rate": 0.0001459201275487523,
      "loss": 2.0841,
      "step": 154759
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.1705635786056519,
      "learning_rate": 0.00014591661773802515,
      "loss": 1.7198,
      "step": 154760
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.1214640140533447,
      "learning_rate": 0.00014591310795594514,
      "loss": 2.0246,
      "step": 154761
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.1400004625320435,
      "learning_rate": 0.0001459095982025127,
      "loss": 1.9594,
      "step": 154762
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.148234486579895,
      "learning_rate": 0.00014590608847772862,
      "loss": 2.1063,
      "step": 154763
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.205725908279419,
      "learning_rate": 0.00014590257878159356,
      "loss": 2.2796,
      "step": 154764
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.044846534729004,
      "learning_rate": 0.0001458990691141081,
      "loss": 1.9901,
      "step": 154765
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.2305011749267578,
      "learning_rate": 0.0001458955594752728,
      "loss": 2.0707,
      "step": 154766
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.1235630512237549,
      "learning_rate": 0.00014589204986508854,
      "loss": 2.2021,
      "step": 154767
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.0831080675125122,
      "learning_rate": 0.00014588854028355577,
      "loss": 1.9078,
      "step": 154768
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.1831278800964355,
      "learning_rate": 0.0001458850307306753,
      "loss": 2.321,
      "step": 154769
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.1464924812316895,
      "learning_rate": 0.00014588152120644765,
      "loss": 2.1957,
      "step": 154770
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.3131301403045654,
      "learning_rate": 0.00014587801171087373,
      "loss": 1.7927,
      "step": 154771
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.1992391347885132,
      "learning_rate": 0.00014587450224395375,
      "loss": 2.1478,
      "step": 154772
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.151235580444336,
      "learning_rate": 0.00014587099280568875,
      "loss": 2.0902,
      "step": 154773
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.1421023607254028,
      "learning_rate": 0.00014586748339607912,
      "loss": 1.8882,
      "step": 154774
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.0793472528457642,
      "learning_rate": 0.00014586397401512575,
      "loss": 2.0107,
      "step": 154775
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.359221339225769,
      "learning_rate": 0.00014586046466282906,
      "loss": 2.1712,
      "step": 154776
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.4065804481506348,
      "learning_rate": 0.00014585695533919003,
      "loss": 2.0926,
      "step": 154777
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.5525182485580444,
      "learning_rate": 0.0001458534460442089,
      "loss": 2.0981,
      "step": 154778
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.166700005531311,
      "learning_rate": 0.00014584993677788663,
      "loss": 2.1518,
      "step": 154779
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.1228468418121338,
      "learning_rate": 0.00014584642754022367,
      "loss": 1.9351,
      "step": 154780
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.122249722480774,
      "learning_rate": 0.00014584291833122086,
      "loss": 1.9993,
      "step": 154781
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.0934427976608276,
      "learning_rate": 0.00014583940915087866,
      "loss": 2.0575,
      "step": 154782
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.4562413692474365,
      "learning_rate": 0.00014583589999919805,
      "loss": 2.0547,
      "step": 154783
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.2403889894485474,
      "learning_rate": 0.00014583239087617922,
      "loss": 2.0263,
      "step": 154784
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.1673840284347534,
      "learning_rate": 0.00014582888178182315,
      "loss": 2.0476,
      "step": 154785
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.5690478086471558,
      "learning_rate": 0.00014582537271613035,
      "loss": 2.0373,
      "step": 154786
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.235809326171875,
      "learning_rate": 0.0001458218636791016,
      "loss": 2.0959,
      "step": 154787
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.2905495166778564,
      "learning_rate": 0.00014581835467073734,
      "loss": 1.9461,
      "step": 154788
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.2061526775360107,
      "learning_rate": 0.0001458148456910386,
      "loss": 2.125,
      "step": 154789
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.1054117679595947,
      "learning_rate": 0.0001458113367400056,
      "loss": 1.8047,
      "step": 154790
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.1765812635421753,
      "learning_rate": 0.00014580782781763927,
      "loss": 1.9764,
      "step": 154791
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.1856436729431152,
      "learning_rate": 0.00014580431892394002,
      "loss": 1.7572,
      "step": 154792
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.2312507629394531,
      "learning_rate": 0.00014580081005890883,
      "loss": 1.8736,
      "step": 154793
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.267182469367981,
      "learning_rate": 0.00014579730122254607,
      "loss": 2.1638,
      "step": 154794
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.164929986000061,
      "learning_rate": 0.0001457937924148527,
      "loss": 2.0715,
      "step": 154795
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.192928671836853,
      "learning_rate": 0.00014579028363582893,
      "loss": 1.9748,
      "step": 154796
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.1679688692092896,
      "learning_rate": 0.00014578677488547582,
      "loss": 2.2311,
      "step": 154797
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.2130768299102783,
      "learning_rate": 0.00014578326616379372,
      "loss": 2.2062,
      "step": 154798
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.0091283321380615,
      "learning_rate": 0.00014577975747078354,
      "loss": 2.1048,
      "step": 154799
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.1931297779083252,
      "learning_rate": 0.0001457762488064457,
      "loss": 2.0402,
      "step": 154800
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.2286075353622437,
      "learning_rate": 0.0001457727401707812,
      "loss": 2.025,
      "step": 154801
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.2330127954483032,
      "learning_rate": 0.00014576923156379022,
      "loss": 2.2088,
      "step": 154802
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.1508203744888306,
      "learning_rate": 0.00014576572298547377,
      "loss": 1.9936,
      "step": 154803
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.1640512943267822,
      "learning_rate": 0.00014576221443583225,
      "loss": 2.1444,
      "step": 154804
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.0642391443252563,
      "learning_rate": 0.0001457587059148666,
      "loss": 2.0894,
      "step": 154805
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.1560136079788208,
      "learning_rate": 0.00014575519742257716,
      "loss": 2.2923,
      "step": 154806
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.2986901998519897,
      "learning_rate": 0.00014575168895896492,
      "loss": 2.1655,
      "step": 154807
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.304799199104309,
      "learning_rate": 0.00014574818052403026,
      "loss": 2.1111,
      "step": 154808
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.1903717517852783,
      "learning_rate": 0.00014574467211777399,
      "loss": 1.8827,
      "step": 154809
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.200933575630188,
      "learning_rate": 0.00014574116374019657,
      "loss": 2.0569,
      "step": 154810
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.0893007516860962,
      "learning_rate": 0.0001457376553912989,
      "loss": 1.9351,
      "step": 154811
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.172920823097229,
      "learning_rate": 0.0001457341470710814,
      "loss": 2.0764,
      "step": 154812
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.2020195722579956,
      "learning_rate": 0.00014573063877954491,
      "loss": 2.0632,
      "step": 154813
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.4181818962097168,
      "learning_rate": 0.00014572713051669005,
      "loss": 2.0265,
      "step": 154814
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.3449575901031494,
      "learning_rate": 0.0001457236222825174,
      "loss": 1.9103,
      "step": 154815
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.2638031244277954,
      "learning_rate": 0.00014572011407702756,
      "loss": 2.0699,
      "step": 154816
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.2040661573410034,
      "learning_rate": 0.00014571660590022132,
      "loss": 1.8748,
      "step": 154817
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.258987307548523,
      "learning_rate": 0.00014571309775209923,
      "loss": 2.0795,
      "step": 154818
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.1882224082946777,
      "learning_rate": 0.00014570958963266206,
      "loss": 1.9004,
      "step": 154819
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.2777076959609985,
      "learning_rate": 0.00014570608154191038,
      "loss": 2.114,
      "step": 154820
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.217731237411499,
      "learning_rate": 0.00014570257347984488,
      "loss": 1.9238,
      "step": 154821
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.3081669807434082,
      "learning_rate": 0.00014569906544646607,
      "loss": 2.0005,
      "step": 154822
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.1234530210494995,
      "learning_rate": 0.00014569555744177484,
      "loss": 2.0284,
      "step": 154823
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.2358829975128174,
      "learning_rate": 0.00014569204946577157,
      "loss": 1.9235,
      "step": 154824
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.1806347370147705,
      "learning_rate": 0.00014568854151845718,
      "loss": 2.1062,
      "step": 154825
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.1886613368988037,
      "learning_rate": 0.00014568503359983222,
      "loss": 2.0279,
      "step": 154826
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.2950750589370728,
      "learning_rate": 0.00014568152570989732,
      "loss": 1.976,
      "step": 154827
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.3063325881958008,
      "learning_rate": 0.000145678017848653,
      "loss": 2.148,
      "step": 154828
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.4118833541870117,
      "learning_rate": 0.00014567451001610022,
      "loss": 2.0346,
      "step": 154829
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.303539514541626,
      "learning_rate": 0.00014567100221223928,
      "loss": 1.8345,
      "step": 154830
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.1818374395370483,
      "learning_rate": 0.00014566749443707116,
      "loss": 1.8972,
      "step": 154831
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.4616395235061646,
      "learning_rate": 0.0001456639866905964,
      "loss": 1.9718,
      "step": 154832
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.1063071489334106,
      "learning_rate": 0.00014566047897281544,
      "loss": 2.0479,
      "step": 154833
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.1836884021759033,
      "learning_rate": 0.00014565697128372928,
      "loss": 2.113,
      "step": 154834
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.1728519201278687,
      "learning_rate": 0.00014565346362333838,
      "loss": 1.9348,
      "step": 154835
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.1996716260910034,
      "learning_rate": 0.00014564995599164327,
      "loss": 2.0058,
      "step": 154836
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.210715889930725,
      "learning_rate": 0.00014564644838864493,
      "loss": 2.1479,
      "step": 154837
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.0962399244308472,
      "learning_rate": 0.00014564294081434377,
      "loss": 2.1374,
      "step": 154838
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.1996616125106812,
      "learning_rate": 0.00014563943326874042,
      "loss": 2.0179,
      "step": 154839
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.1600273847579956,
      "learning_rate": 0.0001456359257518357,
      "loss": 1.9765,
      "step": 154840
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.1750624179840088,
      "learning_rate": 0.00014563241826363017,
      "loss": 1.9619,
      "step": 154841
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.2838774919509888,
      "learning_rate": 0.00014562891080412442,
      "loss": 2.0965,
      "step": 154842
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.173438310623169,
      "learning_rate": 0.00014562540337331926,
      "loss": 2.3593,
      "step": 154843
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.1249207258224487,
      "learning_rate": 0.00014562189597121524,
      "loss": 2.1453,
      "step": 154844
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.1477079391479492,
      "learning_rate": 0.0001456183885978129,
      "loss": 1.9523,
      "step": 154845
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.1106504201889038,
      "learning_rate": 0.00014561488125311315,
      "loss": 1.948,
      "step": 154846
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.1188108921051025,
      "learning_rate": 0.00014561137393711638,
      "loss": 1.9112,
      "step": 154847
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.1578819751739502,
      "learning_rate": 0.00014560786664982348,
      "loss": 1.9535,
      "step": 154848
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.1864078044891357,
      "learning_rate": 0.00014560435939123503,
      "loss": 2.0922,
      "step": 154849
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.2348906993865967,
      "learning_rate": 0.00014560085216135162,
      "loss": 2.2919,
      "step": 154850
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.1455309391021729,
      "learning_rate": 0.0001455973449601738,
      "loss": 2.2113,
      "step": 154851
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.071078896522522,
      "learning_rate": 0.00014559383778770248,
      "loss": 2.065,
      "step": 154852
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.2169089317321777,
      "learning_rate": 0.00014559033064393806,
      "loss": 2.1445,
      "step": 154853
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.140110731124878,
      "learning_rate": 0.00014558682352888144,
      "loss": 2.0375,
      "step": 154854
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.2597891092300415,
      "learning_rate": 0.0001455833164425331,
      "loss": 2.1446,
      "step": 154855
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.065916657447815,
      "learning_rate": 0.00014557980938489378,
      "loss": 2.016,
      "step": 154856
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.2440791130065918,
      "learning_rate": 0.00014557630235596396,
      "loss": 1.9192,
      "step": 154857
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.125271201133728,
      "learning_rate": 0.00014557279535574452,
      "loss": 2.131,
      "step": 154858
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.0882387161254883,
      "learning_rate": 0.00014556928838423594,
      "loss": 1.8048,
      "step": 154859
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.0798254013061523,
      "learning_rate": 0.00014556578144143904,
      "loss": 2.0013,
      "step": 154860
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.1127173900604248,
      "learning_rate": 0.00014556227452735423,
      "loss": 2.2173,
      "step": 154861
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.086134910583496,
      "learning_rate": 0.00014555876764198256,
      "loss": 1.9497,
      "step": 154862
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.1688352823257446,
      "learning_rate": 0.0001455552607853242,
      "loss": 2.1342,
      "step": 154863
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.0734561681747437,
      "learning_rate": 0.00014555175395738015,
      "loss": 1.8646,
      "step": 154864
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.270621657371521,
      "learning_rate": 0.00014554824715815083,
      "loss": 1.9853,
      "step": 154865
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.2537082433700562,
      "learning_rate": 0.00014554474038763713,
      "loss": 2.0466,
      "step": 154866
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.2821146249771118,
      "learning_rate": 0.00014554123364583947,
      "loss": 2.103,
      "step": 154867
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.127410888671875,
      "learning_rate": 0.0001455377269327588,
      "loss": 2.0526,
      "step": 154868
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.1800490617752075,
      "learning_rate": 0.0001455342202483954,
      "loss": 2.0308,
      "step": 154869
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.1200681924819946,
      "learning_rate": 0.00014553071359275018,
      "loss": 1.977,
      "step": 154870
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.0609796047210693,
      "learning_rate": 0.0001455272069658236,
      "loss": 2.1153,
      "step": 154871
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.1826452016830444,
      "learning_rate": 0.00014552370036761657,
      "loss": 2.0064,
      "step": 154872
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.1493844985961914,
      "learning_rate": 0.00014552019379812948,
      "loss": 1.9421,
      "step": 154873
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.268125057220459,
      "learning_rate": 0.0001455166872573632,
      "loss": 1.9968,
      "step": 154874
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.2566624879837036,
      "learning_rate": 0.00014551318074531832,
      "loss": 2.0798,
      "step": 154875
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.1170488595962524,
      "learning_rate": 0.0001455096742619954,
      "loss": 2.2389,
      "step": 154876
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.040922999382019,
      "learning_rate": 0.0001455061678073951,
      "loss": 2.1178,
      "step": 154877
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.223291277885437,
      "learning_rate": 0.00014550266138151817,
      "loss": 1.991,
      "step": 154878
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.1276921033859253,
      "learning_rate": 0.00014549915498436516,
      "loss": 1.9936,
      "step": 154879
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.4263057708740234,
      "learning_rate": 0.00014549564861593685,
      "loss": 2.1776,
      "step": 154880
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.0765942335128784,
      "learning_rate": 0.00014549214227623379,
      "loss": 2.1431,
      "step": 154881
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.3619569540023804,
      "learning_rate": 0.0001454886359652567,
      "loss": 2.1593,
      "step": 154882
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.1846733093261719,
      "learning_rate": 0.00014548512968300605,
      "loss": 2.1052,
      "step": 154883
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.2346216440200806,
      "learning_rate": 0.00014548162342948277,
      "loss": 2.1688,
      "step": 154884
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.1937236785888672,
      "learning_rate": 0.00014547811720468727,
      "loss": 2.0818,
      "step": 154885
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.0794309377670288,
      "learning_rate": 0.0001454746110086204,
      "loss": 2.0563,
      "step": 154886
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.425777792930603,
      "learning_rate": 0.00014547110484128267,
      "loss": 2.1273,
      "step": 154887
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.1474988460540771,
      "learning_rate": 0.00014546759870267485,
      "loss": 2.0828,
      "step": 154888
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.1576769351959229,
      "learning_rate": 0.00014546409259279738,
      "loss": 2.4186,
      "step": 154889
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.2036560773849487,
      "learning_rate": 0.00014546058651165113,
      "loss": 1.9502,
      "step": 154890
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.056888222694397,
      "learning_rate": 0.00014545708045923662,
      "loss": 2.2684,
      "step": 154891
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.1612492799758911,
      "learning_rate": 0.00014545357443555463,
      "loss": 1.9894,
      "step": 154892
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.0860360860824585,
      "learning_rate": 0.00014545006844060573,
      "loss": 2.0393,
      "step": 154893
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.2486953735351562,
      "learning_rate": 0.00014544656247439063,
      "loss": 2.1473,
      "step": 154894
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.1986384391784668,
      "learning_rate": 0.0001454430565369098,
      "loss": 2.1049,
      "step": 154895
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.0685268640518188,
      "learning_rate": 0.0001454395506281641,
      "loss": 2.1589,
      "step": 154896
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.267567753791809,
      "learning_rate": 0.000145436044748154,
      "loss": 2.0281,
      "step": 154897
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.1881681680679321,
      "learning_rate": 0.00014543253889688043,
      "loss": 1.958,
      "step": 154898
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.142305612564087,
      "learning_rate": 0.0001454290330743438,
      "loss": 1.7727,
      "step": 154899
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.0660877227783203,
      "learning_rate": 0.00014542552728054485,
      "loss": 2.1878,
      "step": 154900
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.198330044746399,
      "learning_rate": 0.00014542202151548414,
      "loss": 2.0677,
      "step": 154901
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.2231061458587646,
      "learning_rate": 0.00014541851577916246,
      "loss": 2.0247,
      "step": 154902
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.1628844738006592,
      "learning_rate": 0.0001454150100715803,
      "loss": 2.0588,
      "step": 154903
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.124487280845642,
      "learning_rate": 0.0001454115043927385,
      "loss": 1.9477,
      "step": 154904
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.1616061925888062,
      "learning_rate": 0.00014540799874263762,
      "loss": 2.1508,
      "step": 154905
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.186233639717102,
      "learning_rate": 0.00014540449312127834,
      "loss": 1.8341,
      "step": 154906
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.1583623886108398,
      "learning_rate": 0.00014540098752866118,
      "loss": 2.0716,
      "step": 154907
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.2777341604232788,
      "learning_rate": 0.00014539748196478695,
      "loss": 1.9694,
      "step": 154908
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.1962225437164307,
      "learning_rate": 0.0001453939764296562,
      "loss": 1.9137,
      "step": 154909
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.0999040603637695,
      "learning_rate": 0.0001453904709232697,
      "loss": 1.7922,
      "step": 154910
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.108151912689209,
      "learning_rate": 0.00014538696544562805,
      "loss": 2.0401,
      "step": 154911
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.1534337997436523,
      "learning_rate": 0.00014538345999673185,
      "loss": 2.0299,
      "step": 154912
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.512715220451355,
      "learning_rate": 0.0001453799545765817,
      "loss": 1.983,
      "step": 154913
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.1279360055923462,
      "learning_rate": 0.00014537644918517843,
      "loss": 1.9855,
      "step": 154914
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.1127023696899414,
      "learning_rate": 0.00014537294382252247,
      "loss": 2.0912,
      "step": 154915
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.1878381967544556,
      "learning_rate": 0.00014536943848861473,
      "loss": 1.8352,
      "step": 154916
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.2042595148086548,
      "learning_rate": 0.00014536593318345578,
      "loss": 2.0182,
      "step": 154917
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.325337290763855,
      "learning_rate": 0.00014536242790704618,
      "loss": 2.1969,
      "step": 154918
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.3201237916946411,
      "learning_rate": 0.00014535892265938647,
      "loss": 2.116,
      "step": 154919
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.5860952138900757,
      "learning_rate": 0.00014535541744047762,
      "loss": 2.0978,
      "step": 154920
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.1548864841461182,
      "learning_rate": 0.00014535191225032,
      "loss": 2.0943,
      "step": 154921
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.1680161952972412,
      "learning_rate": 0.00014534840708891448,
      "loss": 2.0917,
      "step": 154922
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.1683828830718994,
      "learning_rate": 0.0001453449019562616,
      "loss": 2.1288,
      "step": 154923
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.2556127309799194,
      "learning_rate": 0.0001453413968523619,
      "loss": 2.3547,
      "step": 154924
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.2768272161483765,
      "learning_rate": 0.00014533789177721633,
      "loss": 1.8743,
      "step": 154925
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.157503604888916,
      "learning_rate": 0.0001453343867308253,
      "loss": 1.9345,
      "step": 154926
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.1158558130264282,
      "learning_rate": 0.00014533088171318943,
      "loss": 2.2778,
      "step": 154927
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.058653712272644,
      "learning_rate": 0.0001453273767243096,
      "loss": 1.8494,
      "step": 154928
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.2836395502090454,
      "learning_rate": 0.00014532387176418633,
      "loss": 1.9633,
      "step": 154929
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.0472666025161743,
      "learning_rate": 0.0001453203668328201,
      "loss": 2.0436,
      "step": 154930
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.1224700212478638,
      "learning_rate": 0.00014531686193021195,
      "loss": 2.0221,
      "step": 154931
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.1687462329864502,
      "learning_rate": 0.00014531335705636215,
      "loss": 1.8079,
      "step": 154932
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.1489423513412476,
      "learning_rate": 0.00014530985221127162,
      "loss": 2.0321,
      "step": 154933
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.195988416671753,
      "learning_rate": 0.00014530634739494088,
      "loss": 2.0426,
      "step": 154934
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.107960820198059,
      "learning_rate": 0.00014530284260737066,
      "loss": 1.9175,
      "step": 154935
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.0834581851959229,
      "learning_rate": 0.00014529933784856143,
      "loss": 1.9413,
      "step": 154936
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.1130056381225586,
      "learning_rate": 0.00014529583311851413,
      "loss": 2.1365,
      "step": 154937
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.162933588027954,
      "learning_rate": 0.00014529232841722906,
      "loss": 2.204,
      "step": 154938
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.0853606462478638,
      "learning_rate": 0.00014528882374470725,
      "loss": 1.9026,
      "step": 154939
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.1784803867340088,
      "learning_rate": 0.00014528531910094913,
      "loss": 2.0783,
      "step": 154940
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.184124231338501,
      "learning_rate": 0.0001452818144859553,
      "loss": 2.0936,
      "step": 154941
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.1145626306533813,
      "learning_rate": 0.00014527830989972658,
      "loss": 2.1976,
      "step": 154942
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.3142824172973633,
      "learning_rate": 0.00014527480534226359,
      "loss": 2.1955,
      "step": 154943
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.249153733253479,
      "learning_rate": 0.00014527130081356678,
      "loss": 1.8667,
      "step": 154944
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.1803394556045532,
      "learning_rate": 0.00014526779631363708,
      "loss": 2.0629,
      "step": 154945
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.094132661819458,
      "learning_rate": 0.0001452642918424749,
      "loss": 1.9082,
      "step": 154946
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.1874858140945435,
      "learning_rate": 0.00014526078740008115,
      "loss": 2.0669,
      "step": 154947
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.1863677501678467,
      "learning_rate": 0.0001452572829864563,
      "loss": 2.064,
      "step": 154948
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.2398790121078491,
      "learning_rate": 0.00014525377860160105,
      "loss": 2.3011,
      "step": 154949
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.2208216190338135,
      "learning_rate": 0.00014525027424551598,
      "loss": 1.9757,
      "step": 154950
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.0975421667099,
      "learning_rate": 0.0001452467699182019,
      "loss": 2.1519,
      "step": 154951
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.1544315814971924,
      "learning_rate": 0.0001452432656196592,
      "loss": 2.0419,
      "step": 154952
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.089326024055481,
      "learning_rate": 0.00014523976134988884,
      "loss": 2.2684,
      "step": 154953
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.1626687049865723,
      "learning_rate": 0.00014523625710889134,
      "loss": 2.0813,
      "step": 154954
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.1366995573043823,
      "learning_rate": 0.00014523275289666734,
      "loss": 2.135,
      "step": 154955
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.253505825996399,
      "learning_rate": 0.00014522924871321737,
      "loss": 2.1144,
      "step": 154956
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.3021397590637207,
      "learning_rate": 0.00014522574455854234,
      "loss": 2.0137,
      "step": 154957
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.263671636581421,
      "learning_rate": 0.00014522224043264263,
      "loss": 2.1123,
      "step": 154958
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.2948205471038818,
      "learning_rate": 0.00014521873633551913,
      "loss": 1.9972,
      "step": 154959
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.1871933937072754,
      "learning_rate": 0.0001452152322671724,
      "loss": 2.0665,
      "step": 154960
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.1892050504684448,
      "learning_rate": 0.0001452117282276031,
      "loss": 2.2048,
      "step": 154961
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.1529840230941772,
      "learning_rate": 0.00014520822421681175,
      "loss": 2.1923,
      "step": 154962
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.2322566509246826,
      "learning_rate": 0.0001452047202347992,
      "loss": 2.033,
      "step": 154963
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.0700236558914185,
      "learning_rate": 0.00014520121628156592,
      "loss": 1.997,
      "step": 154964
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.0860300064086914,
      "learning_rate": 0.00014519771235711276,
      "loss": 2.1113,
      "step": 154965
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.2784448862075806,
      "learning_rate": 0.00014519420846144024,
      "loss": 2.0528,
      "step": 154966
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.315293550491333,
      "learning_rate": 0.00014519070459454908,
      "loss": 1.8942,
      "step": 154967
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.1403039693832397,
      "learning_rate": 0.00014518720075643978,
      "loss": 1.9069,
      "step": 154968
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.2704981565475464,
      "learning_rate": 0.0001451836969471132,
      "loss": 2.0843,
      "step": 154969
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.14597487449646,
      "learning_rate": 0.00014518019316656975,
      "loss": 2.0413,
      "step": 154970
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.0912867784500122,
      "learning_rate": 0.0001451766894148104,
      "loss": 1.8005,
      "step": 154971
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.1077214479446411,
      "learning_rate": 0.0001451731856918356,
      "loss": 1.9863,
      "step": 154972
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.0760688781738281,
      "learning_rate": 0.000145169681997646,
      "loss": 1.9833,
      "step": 154973
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.177148699760437,
      "learning_rate": 0.0001451661783322422,
      "loss": 2.1814,
      "step": 154974
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.3666490316390991,
      "learning_rate": 0.000145162674695625,
      "loss": 1.9998,
      "step": 154975
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.1217344999313354,
      "learning_rate": 0.0001451591710877949,
      "loss": 2.0891,
      "step": 154976
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.26343834400177,
      "learning_rate": 0.00014515566750875276,
      "loss": 2.165,
      "step": 154977
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.1250559091567993,
      "learning_rate": 0.00014515216395849908,
      "loss": 1.9159,
      "step": 154978
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.1313998699188232,
      "learning_rate": 0.00014514866043703454,
      "loss": 2.2401,
      "step": 154979
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.2525306940078735,
      "learning_rate": 0.0001451451569443597,
      "loss": 2.1124,
      "step": 154980
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.2744308710098267,
      "learning_rate": 0.00014514165348047535,
      "loss": 2.2586,
      "step": 154981
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.1752439737319946,
      "learning_rate": 0.00014513815004538202,
      "loss": 2.0328,
      "step": 154982
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.1380321979522705,
      "learning_rate": 0.00014513464663908058,
      "loss": 2.0972,
      "step": 154983
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.2851961851119995,
      "learning_rate": 0.00014513114326157147,
      "loss": 2.0491,
      "step": 154984
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.793387532234192,
      "learning_rate": 0.00014512763991285545,
      "loss": 1.9769,
      "step": 154985
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.1362195014953613,
      "learning_rate": 0.00014512413659293296,
      "loss": 2.2049,
      "step": 154986
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.1078108549118042,
      "learning_rate": 0.00014512063330180495,
      "loss": 1.9882,
      "step": 154987
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.1349319219589233,
      "learning_rate": 0.00014511713003947184,
      "loss": 2.0092,
      "step": 154988
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.2469515800476074,
      "learning_rate": 0.0001451136268059345,
      "loss": 2.1338,
      "step": 154989
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.1881059408187866,
      "learning_rate": 0.00014511012360119343,
      "loss": 2.1003,
      "step": 154990
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.1673840284347534,
      "learning_rate": 0.00014510662042524933,
      "loss": 2.1505,
      "step": 154991
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.0697671175003052,
      "learning_rate": 0.0001451031172781027,
      "loss": 1.962,
      "step": 154992
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.0917377471923828,
      "learning_rate": 0.00014509961415975448,
      "loss": 1.9714,
      "step": 154993
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.1255851984024048,
      "learning_rate": 0.00014509611107020503,
      "loss": 2.1164,
      "step": 154994
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.2213382720947266,
      "learning_rate": 0.00014509260800945521,
      "loss": 2.0629,
      "step": 154995
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.3393522500991821,
      "learning_rate": 0.00014508910497750568,
      "loss": 2.1355,
      "step": 154996
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.1229549646377563,
      "learning_rate": 0.00014508560197435696,
      "loss": 1.9473,
      "step": 154997
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.322973608970642,
      "learning_rate": 0.00014508209900000962,
      "loss": 2.1951,
      "step": 154998
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.1560871601104736,
      "learning_rate": 0.0001450785960544646,
      "loss": 2.0131,
      "step": 154999
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.0543619394302368,
      "learning_rate": 0.00014507509313772227,
      "loss": 1.7637,
      "step": 155000
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.181410312652588,
      "learning_rate": 0.0001450715902497835,
      "loss": 2.2492,
      "step": 155001
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.1657540798187256,
      "learning_rate": 0.00014506808739064882,
      "loss": 2.1137,
      "step": 155002
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.2663007974624634,
      "learning_rate": 0.00014506458456031894,
      "loss": 2.0343,
      "step": 155003
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.234769582748413,
      "learning_rate": 0.00014506108175879438,
      "loss": 2.1302,
      "step": 155004
    },
    {
      "epoch": 2.02,
      "grad_norm": 2.8086421489715576,
      "learning_rate": 0.000145057578986076,
      "loss": 1.9398,
      "step": 155005
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.0556949377059937,
      "learning_rate": 0.0001450540762421642,
      "loss": 2.1303,
      "step": 155006
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.2073363065719604,
      "learning_rate": 0.0001450505735270599,
      "loss": 2.1351,
      "step": 155007
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.120957374572754,
      "learning_rate": 0.0001450470708407635,
      "loss": 2.0736,
      "step": 155008
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.038164496421814,
      "learning_rate": 0.00014504356818327593,
      "loss": 2.0187,
      "step": 155009
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.218173861503601,
      "learning_rate": 0.00014504006555459765,
      "loss": 1.8536,
      "step": 155010
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.1981589794158936,
      "learning_rate": 0.00014503656295472935,
      "loss": 2.0098,
      "step": 155011
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.1038267612457275,
      "learning_rate": 0.00014503306038367154,
      "loss": 1.8737,
      "step": 155012
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.150221347808838,
      "learning_rate": 0.00014502955784142515,
      "loss": 2.0822,
      "step": 155013
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.061468243598938,
      "learning_rate": 0.0001450260553279906,
      "loss": 2.1568,
      "step": 155014
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.1849197149276733,
      "learning_rate": 0.0001450225528433687,
      "loss": 2.021,
      "step": 155015
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.1509164571762085,
      "learning_rate": 0.00014501905038756007,
      "loss": 1.9925,
      "step": 155016
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.211626648902893,
      "learning_rate": 0.0001450155479605653,
      "loss": 1.8896,
      "step": 155017
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.118848443031311,
      "learning_rate": 0.00014501204556238495,
      "loss": 2.0844,
      "step": 155018
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.1436845064163208,
      "learning_rate": 0.00014500854319301993,
      "loss": 1.9693,
      "step": 155019
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.239536166191101,
      "learning_rate": 0.00014500504085247057,
      "loss": 1.8951,
      "step": 155020
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.273800015449524,
      "learning_rate": 0.00014500153854073787,
      "loss": 2.0873,
      "step": 155021
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.1503102779388428,
      "learning_rate": 0.00014499803625782232,
      "loss": 1.8125,
      "step": 155022
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.0947539806365967,
      "learning_rate": 0.0001449945340037244,
      "loss": 2.0841,
      "step": 155023
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.1580466032028198,
      "learning_rate": 0.00014499103177844505,
      "loss": 2.1823,
      "step": 155024
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.1584166288375854,
      "learning_rate": 0.0001449875295819848,
      "loss": 1.8761,
      "step": 155025
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.2141542434692383,
      "learning_rate": 0.0001449840274143442,
      "loss": 2.0074,
      "step": 155026
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.228361964225769,
      "learning_rate": 0.0001449805252755241,
      "loss": 2.11,
      "step": 155027
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.1664226055145264,
      "learning_rate": 0.00014497702316552502,
      "loss": 1.9025,
      "step": 155028
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.2356253862380981,
      "learning_rate": 0.00014497352108434753,
      "loss": 2.08,
      "step": 155029
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.203211784362793,
      "learning_rate": 0.00014497001903199252,
      "loss": 2.1836,
      "step": 155030
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.284461259841919,
      "learning_rate": 0.00014496651700846038,
      "loss": 2.0188,
      "step": 155031
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.188683271408081,
      "learning_rate": 0.00014496301501375203,
      "loss": 2.1474,
      "step": 155032
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.2030009031295776,
      "learning_rate": 0.00014495951304786795,
      "loss": 2.088,
      "step": 155033
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.1986720561981201,
      "learning_rate": 0.00014495601111080885,
      "loss": 1.9287,
      "step": 155034
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.4736093282699585,
      "learning_rate": 0.00014495250920257524,
      "loss": 2.1426,
      "step": 155035
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.086405634880066,
      "learning_rate": 0.000144949007323168,
      "loss": 1.8796,
      "step": 155036
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.1206940412521362,
      "learning_rate": 0.00014494550547258754,
      "loss": 2.1777,
      "step": 155037
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.1248029470443726,
      "learning_rate": 0.00014494200365083472,
      "loss": 2.088,
      "step": 155038
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.0458145141601562,
      "learning_rate": 0.00014493850185791016,
      "loss": 1.9431,
      "step": 155039
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.1979728937149048,
      "learning_rate": 0.00014493500009381442,
      "loss": 2.1221,
      "step": 155040
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.1946467161178589,
      "learning_rate": 0.0001449314983585481,
      "loss": 2.1156,
      "step": 155041
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.2805359363555908,
      "learning_rate": 0.00014492799665211207,
      "loss": 2.0006,
      "step": 155042
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.141329050064087,
      "learning_rate": 0.00014492449497450672,
      "loss": 2.0033,
      "step": 155043
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.1255005598068237,
      "learning_rate": 0.00014492099332573293,
      "loss": 2.08,
      "step": 155044
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.3845752477645874,
      "learning_rate": 0.0001449174917057913,
      "loss": 2.0143,
      "step": 155045
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.214401125907898,
      "learning_rate": 0.00014491399011468242,
      "loss": 2.112,
      "step": 155046
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.1915867328643799,
      "learning_rate": 0.0001449104885524068,
      "loss": 1.9679,
      "step": 155047
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.221471905708313,
      "learning_rate": 0.0001449069870189654,
      "loss": 1.999,
      "step": 155048
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.1674424409866333,
      "learning_rate": 0.0001449034855143586,
      "loss": 1.8463,
      "step": 155049
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.1577141284942627,
      "learning_rate": 0.00014489998403858728,
      "loss": 2.1317,
      "step": 155050
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.1690689325332642,
      "learning_rate": 0.000144896482591652,
      "loss": 2.3318,
      "step": 155051
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.1737343072891235,
      "learning_rate": 0.00014489298117355336,
      "loss": 1.9244,
      "step": 155052
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.2083593606948853,
      "learning_rate": 0.00014488947978429198,
      "loss": 2.2307,
      "step": 155053
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.2248069047927856,
      "learning_rate": 0.00014488597842386865,
      "loss": 2.0222,
      "step": 155054
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.3155417442321777,
      "learning_rate": 0.00014488247709228382,
      "loss": 1.97,
      "step": 155055
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.24451744556427,
      "learning_rate": 0.00014487897578953837,
      "loss": 1.9534,
      "step": 155056
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.589095115661621,
      "learning_rate": 0.00014487547451563288,
      "loss": 2.0822,
      "step": 155057
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.2280914783477783,
      "learning_rate": 0.00014487197327056796,
      "loss": 2.1782,
      "step": 155058
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.2031062841415405,
      "learning_rate": 0.00014486847205434416,
      "loss": 2.1128,
      "step": 155059
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.2531366348266602,
      "learning_rate": 0.00014486497086696235,
      "loss": 2.0325,
      "step": 155060
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.3512773513793945,
      "learning_rate": 0.0001448614697084229,
      "loss": 1.9888,
      "step": 155061
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.2759348154067993,
      "learning_rate": 0.00014485796857872683,
      "loss": 1.8834,
      "step": 155062
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.2585077285766602,
      "learning_rate": 0.00014485446747787456,
      "loss": 2.2431,
      "step": 155063
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.1145964860916138,
      "learning_rate": 0.00014485096640586676,
      "loss": 2.1332,
      "step": 155064
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.0995122194290161,
      "learning_rate": 0.000144847465362704,
      "loss": 2.1715,
      "step": 155065
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.280064344406128,
      "learning_rate": 0.0001448439643483871,
      "loss": 2.3957,
      "step": 155066
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.3082553148269653,
      "learning_rate": 0.0001448404633629166,
      "loss": 2.0476,
      "step": 155067
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.1549183130264282,
      "learning_rate": 0.00014483696240629322,
      "loss": 1.897,
      "step": 155068
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.1558507680892944,
      "learning_rate": 0.00014483346147851762,
      "loss": 1.9698,
      "step": 155069
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.1772401332855225,
      "learning_rate": 0.0001448299605795904,
      "loss": 2.0109,
      "step": 155070
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.3106634616851807,
      "learning_rate": 0.0001448264597095121,
      "loss": 2.1159,
      "step": 155071
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.2547101974487305,
      "learning_rate": 0.00014482295886828362,
      "loss": 1.9579,
      "step": 155072
    },
    {
      "epoch": 2.02,
      "grad_norm": 0.9956779479980469,
      "learning_rate": 0.00014481945805590535,
      "loss": 2.0719,
      "step": 155073
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.182401180267334,
      "learning_rate": 0.0001448159572723782,
      "loss": 1.9779,
      "step": 155074
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.3267006874084473,
      "learning_rate": 0.00014481245651770254,
      "loss": 2.0498,
      "step": 155075
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.1019080877304077,
      "learning_rate": 0.00014480895579187947,
      "loss": 2.1625,
      "step": 155076
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.1289621591567993,
      "learning_rate": 0.00014480545509490904,
      "loss": 2.3236,
      "step": 155077
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.1766539812088013,
      "learning_rate": 0.00014480195442679236,
      "loss": 1.9956,
      "step": 155078
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.2794787883758545,
      "learning_rate": 0.00014479845378752984,
      "loss": 2.2403,
      "step": 155079
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.215041160583496,
      "learning_rate": 0.00014479495317712228,
      "loss": 1.9936,
      "step": 155080
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.1626313924789429,
      "learning_rate": 0.00014479145259557017,
      "loss": 2.179,
      "step": 155081
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.1507614850997925,
      "learning_rate": 0.0001447879520428745,
      "loss": 1.9249,
      "step": 155082
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.1785386800765991,
      "learning_rate": 0.00014478445151903545,
      "loss": 2.0645,
      "step": 155083
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.2108956575393677,
      "learning_rate": 0.000144780951024054,
      "loss": 2.0799,
      "step": 155084
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.114983320236206,
      "learning_rate": 0.00014477745055793058,
      "loss": 2.0959,
      "step": 155085
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.1369432210922241,
      "learning_rate": 0.0001447739501206661,
      "loss": 2.0136,
      "step": 155086
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.1518691778182983,
      "learning_rate": 0.00014477044971226096,
      "loss": 1.9138,
      "step": 155087
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.1901005506515503,
      "learning_rate": 0.00014476694933271615,
      "loss": 2.0134,
      "step": 155088
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.2051910161972046,
      "learning_rate": 0.00014476344898203185,
      "loss": 2.1843,
      "step": 155089
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.1867119073867798,
      "learning_rate": 0.0001447599486602091,
      "loss": 2.2116,
      "step": 155090
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.2447588443756104,
      "learning_rate": 0.00014475644836724823,
      "loss": 2.1131,
      "step": 155091
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.040411353111267,
      "learning_rate": 0.00014475294810315024,
      "loss": 1.9239,
      "step": 155092
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.139430046081543,
      "learning_rate": 0.00014474944786791545,
      "loss": 2.0239,
      "step": 155093
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.0873438119888306,
      "learning_rate": 0.0001447459476615448,
      "loss": 1.8641,
      "step": 155094
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.2150259017944336,
      "learning_rate": 0.00014474244748403884,
      "loss": 2.1896,
      "step": 155095
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.1983506679534912,
      "learning_rate": 0.00014473894733539814,
      "loss": 2.1447,
      "step": 155096
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.2515592575073242,
      "learning_rate": 0.00014473544721562331,
      "loss": 2.129,
      "step": 155097
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.3508269786834717,
      "learning_rate": 0.00014473194712471518,
      "loss": 1.9708,
      "step": 155098
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.1987258195877075,
      "learning_rate": 0.00014472844706267426,
      "loss": 2.0831,
      "step": 155099
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.200820803642273,
      "learning_rate": 0.0001447249470295013,
      "loss": 1.9178,
      "step": 155100
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.2770609855651855,
      "learning_rate": 0.00014472144702519697,
      "loss": 1.9374,
      "step": 155101
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.215820074081421,
      "learning_rate": 0.0001447179470497618,
      "loss": 2.0255,
      "step": 155102
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.2418437004089355,
      "learning_rate": 0.0001447144471031964,
      "loss": 2.126,
      "step": 155103
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.1723263263702393,
      "learning_rate": 0.00014471094718550162,
      "loss": 2.0161,
      "step": 155104
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.1636197566986084,
      "learning_rate": 0.0001447074472966779,
      "loss": 2.0947,
      "step": 155105
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.1125273704528809,
      "learning_rate": 0.00014470394743672612,
      "loss": 2.223,
      "step": 155106
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.2173445224761963,
      "learning_rate": 0.0001447004476056468,
      "loss": 2.0715,
      "step": 155107
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.0867477655410767,
      "learning_rate": 0.00014469694780344048,
      "loss": 2.0904,
      "step": 155108
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.1609405279159546,
      "learning_rate": 0.00014469344803010804,
      "loss": 2.0737,
      "step": 155109
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.4242630004882812,
      "learning_rate": 0.00014468994828565005,
      "loss": 2.0058,
      "step": 155110
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.128633737564087,
      "learning_rate": 0.00014468644857006696,
      "loss": 1.9345,
      "step": 155111
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.083918571472168,
      "learning_rate": 0.00014468294888335975,
      "loss": 1.8873,
      "step": 155112
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.2294673919677734,
      "learning_rate": 0.00014467944922552892,
      "loss": 2.1506,
      "step": 155113
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.1933645009994507,
      "learning_rate": 0.00014467594959657494,
      "loss": 2.1535,
      "step": 155114
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.5578144788742065,
      "learning_rate": 0.0001446724499964988,
      "loss": 1.9481,
      "step": 155115
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.1144952774047852,
      "learning_rate": 0.00014466895042530097,
      "loss": 2.0576,
      "step": 155116
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.1094768047332764,
      "learning_rate": 0.000144665450882982,
      "loss": 1.9384,
      "step": 155117
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.0820801258087158,
      "learning_rate": 0.00014466195136954277,
      "loss": 1.9599,
      "step": 155118
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.3422516584396362,
      "learning_rate": 0.0001446584518849838,
      "loss": 2.2868,
      "step": 155119
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.115815281867981,
      "learning_rate": 0.00014465495242930567,
      "loss": 1.8992,
      "step": 155120
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.096152901649475,
      "learning_rate": 0.00014465145300250918,
      "loss": 1.8021,
      "step": 155121
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.115899682044983,
      "learning_rate": 0.00014464795360459484,
      "loss": 2.0006,
      "step": 155122
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.773846983909607,
      "learning_rate": 0.0001446444542355635,
      "loss": 1.9073,
      "step": 155123
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.0611778497695923,
      "learning_rate": 0.0001446409548954157,
      "loss": 1.7067,
      "step": 155124
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.1783013343811035,
      "learning_rate": 0.00014463745558415203,
      "loss": 2.0493,
      "step": 155125
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.0200413465499878,
      "learning_rate": 0.0001446339563017731,
      "loss": 1.8205,
      "step": 155126
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.12032151222229,
      "learning_rate": 0.00014463045704827976,
      "loss": 2.0153,
      "step": 155127
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.3595253229141235,
      "learning_rate": 0.00014462695782367248,
      "loss": 2.0543,
      "step": 155128
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.1623539924621582,
      "learning_rate": 0.00014462345862795205,
      "loss": 2.1075,
      "step": 155129
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.2542093992233276,
      "learning_rate": 0.00014461995946111905,
      "loss": 1.9274,
      "step": 155130
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.2021933794021606,
      "learning_rate": 0.00014461646032317414,
      "loss": 2.0744,
      "step": 155131
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.180981159210205,
      "learning_rate": 0.00014461296121411787,
      "loss": 1.9383,
      "step": 155132
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.2405987977981567,
      "learning_rate": 0.00014460946213395109,
      "loss": 2.0509,
      "step": 155133
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.1628077030181885,
      "learning_rate": 0.0001446059630826742,
      "loss": 1.9855,
      "step": 155134
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.1259989738464355,
      "learning_rate": 0.00014460246406028813,
      "loss": 1.9397,
      "step": 155135
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.1748318672180176,
      "learning_rate": 0.0001445989650667934,
      "loss": 2.2147,
      "step": 155136
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.2347596883773804,
      "learning_rate": 0.00014459546610219066,
      "loss": 1.8912,
      "step": 155137
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.2370576858520508,
      "learning_rate": 0.00014459196716648043,
      "loss": 2.0838,
      "step": 155138
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.1487431526184082,
      "learning_rate": 0.0001445884682596636,
      "loss": 2.0494,
      "step": 155139
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.1538490056991577,
      "learning_rate": 0.00014458496938174058,
      "loss": 2.0515,
      "step": 155140
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.1932604312896729,
      "learning_rate": 0.00014458147053271228,
      "loss": 2.0238,
      "step": 155141
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.1804776191711426,
      "learning_rate": 0.0001445779717125791,
      "loss": 2.3129,
      "step": 155142
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.3057613372802734,
      "learning_rate": 0.00014457447292134208,
      "loss": 2.0479,
      "step": 155143
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.0443098545074463,
      "learning_rate": 0.00014457097415900133,
      "loss": 1.855,
      "step": 155144
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.1758216619491577,
      "learning_rate": 0.00014456747542555785,
      "loss": 1.9504,
      "step": 155145
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.1198842525482178,
      "learning_rate": 0.00014456397672101215,
      "loss": 2.2238,
      "step": 155146
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.1568959951400757,
      "learning_rate": 0.00014456047804536505,
      "loss": 2.0908,
      "step": 155147
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.3181343078613281,
      "learning_rate": 0.00014455697939861692,
      "loss": 2.1937,
      "step": 155148
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.1662904024124146,
      "learning_rate": 0.00014455348078076886,
      "loss": 2.0261,
      "step": 155149
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.109276294708252,
      "learning_rate": 0.000144549982191821,
      "loss": 1.8161,
      "step": 155150
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.2957934141159058,
      "learning_rate": 0.0001445464836317744,
      "loss": 2.2917,
      "step": 155151
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.2493597269058228,
      "learning_rate": 0.00014454298510062938,
      "loss": 2.1275,
      "step": 155152
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.0623235702514648,
      "learning_rate": 0.00014453948659838686,
      "loss": 1.9325,
      "step": 155153
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.2373683452606201,
      "learning_rate": 0.00014453598812504728,
      "loss": 2.2288,
      "step": 155154
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.1888734102249146,
      "learning_rate": 0.00014453248968061163,
      "loss": 2.086,
      "step": 155155
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.2237610816955566,
      "learning_rate": 0.0001445289912650801,
      "loss": 2.115,
      "step": 155156
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.1381194591522217,
      "learning_rate": 0.00014452549287845363,
      "loss": 1.9038,
      "step": 155157
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.122831106185913,
      "learning_rate": 0.00014452199452073277,
      "loss": 1.8626,
      "step": 155158
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.228837490081787,
      "learning_rate": 0.00014451849619191826,
      "loss": 2.0416,
      "step": 155159
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.1386667490005493,
      "learning_rate": 0.0001445149978920106,
      "loss": 2.1106,
      "step": 155160
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.4858847856521606,
      "learning_rate": 0.0001445114996210108,
      "loss": 2.0716,
      "step": 155161
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.1253161430358887,
      "learning_rate": 0.00014450800137891896,
      "loss": 2.2603,
      "step": 155162
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.1202950477600098,
      "learning_rate": 0.00014450450316573615,
      "loss": 2.0287,
      "step": 155163
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.151802897453308,
      "learning_rate": 0.0001445010049814628,
      "loss": 1.9598,
      "step": 155164
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.1289337873458862,
      "learning_rate": 0.00014449750682609977,
      "loss": 1.9595,
      "step": 155165
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.0671437978744507,
      "learning_rate": 0.00014449400869964742,
      "loss": 2.0378,
      "step": 155166
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.0729845762252808,
      "learning_rate": 0.00014449051060210687,
      "loss": 1.9055,
      "step": 155167
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.1331398487091064,
      "learning_rate": 0.00014448701253347816,
      "loss": 1.8509,
      "step": 155168
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.122878074645996,
      "learning_rate": 0.00014448351449376242,
      "loss": 1.9645,
      "step": 155169
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.1746301651000977,
      "learning_rate": 0.00014448001648296,
      "loss": 2.009,
      "step": 155170
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.2498857975006104,
      "learning_rate": 0.0001444765185010718,
      "loss": 1.8247,
      "step": 155171
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.1711376905441284,
      "learning_rate": 0.00014447302054809824,
      "loss": 1.949,
      "step": 155172
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.1349329948425293,
      "learning_rate": 0.00014446952262404028,
      "loss": 2.0374,
      "step": 155173
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.1543952226638794,
      "learning_rate": 0.00014446602472889818,
      "loss": 1.9419,
      "step": 155174
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.1876627206802368,
      "learning_rate": 0.00014446252686267286,
      "loss": 2.0691,
      "step": 155175
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.1287312507629395,
      "learning_rate": 0.00014445902902536481,
      "loss": 2.073,
      "step": 155176
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.1303014755249023,
      "learning_rate": 0.00014445553121697488,
      "loss": 1.9253,
      "step": 155177
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.1289764642715454,
      "learning_rate": 0.00014445203343750344,
      "loss": 1.985,
      "step": 155178
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.1684179306030273,
      "learning_rate": 0.0001444485356869516,
      "loss": 2.0563,
      "step": 155179
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.123934268951416,
      "learning_rate": 0.0001444450379653194,
      "loss": 2.0872,
      "step": 155180
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.1380114555358887,
      "learning_rate": 0.000144441540272608,
      "loss": 2.0211,
      "step": 155181
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.2401224374771118,
      "learning_rate": 0.00014443804260881772,
      "loss": 1.9775,
      "step": 155182
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.1839625835418701,
      "learning_rate": 0.00014443454497394944,
      "loss": 2.1597,
      "step": 155183
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.1717960834503174,
      "learning_rate": 0.0001444310473680036,
      "loss": 2.1099,
      "step": 155184
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.192399263381958,
      "learning_rate": 0.00014442754979098108,
      "loss": 1.8466,
      "step": 155185
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.1932684183120728,
      "learning_rate": 0.00014442405224288245,
      "loss": 2.2351,
      "step": 155186
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.2621759176254272,
      "learning_rate": 0.00014442055472370827,
      "loss": 1.9942,
      "step": 155187
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.1199437379837036,
      "learning_rate": 0.0001444170572334592,
      "loss": 1.9747,
      "step": 155188
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.1325031518936157,
      "learning_rate": 0.00014441355977213598,
      "loss": 2.0547,
      "step": 155189
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.4021881818771362,
      "learning_rate": 0.00014441006233973915,
      "loss": 1.9558,
      "step": 155190
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.2515215873718262,
      "learning_rate": 0.0001444065649362695,
      "loss": 2.1213,
      "step": 155191
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.1389578580856323,
      "learning_rate": 0.00014440306756172765,
      "loss": 1.9652,
      "step": 155192
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.0467826128005981,
      "learning_rate": 0.00014439957021611407,
      "loss": 1.9329,
      "step": 155193
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.422727346420288,
      "learning_rate": 0.00014439607289942963,
      "loss": 1.7996,
      "step": 155194
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.2830179929733276,
      "learning_rate": 0.00014439257561167494,
      "loss": 1.9336,
      "step": 155195
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.164516806602478,
      "learning_rate": 0.00014438907835285047,
      "loss": 2.1112,
      "step": 155196
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.2218652963638306,
      "learning_rate": 0.00014438558112295715,
      "loss": 1.9619,
      "step": 155197
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.213955283164978,
      "learning_rate": 0.00014438208392199547,
      "loss": 2.2848,
      "step": 155198
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.7178525924682617,
      "learning_rate": 0.00014437858674996596,
      "loss": 2.2444,
      "step": 155199
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.5980535745620728,
      "learning_rate": 0.00014437508960686956,
      "loss": 2.1349,
      "step": 155200
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.1739760637283325,
      "learning_rate": 0.00014437159249270677,
      "loss": 2.0689,
      "step": 155201
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.2705323696136475,
      "learning_rate": 0.0001443680954074781,
      "loss": 1.9367,
      "step": 155202
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.3000993728637695,
      "learning_rate": 0.00014436459835118448,
      "loss": 2.1917,
      "step": 155203
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.4607597589492798,
      "learning_rate": 0.0001443611013238264,
      "loss": 2.1816,
      "step": 155204
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.0570017099380493,
      "learning_rate": 0.00014435760432540443,
      "loss": 2.3654,
      "step": 155205
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.193442702293396,
      "learning_rate": 0.00014435410735591942,
      "loss": 1.9884,
      "step": 155206
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.2061147689819336,
      "learning_rate": 0.0001443506104153718,
      "loss": 2.0112,
      "step": 155207
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.3533000946044922,
      "learning_rate": 0.00014434711350376248,
      "loss": 1.9021,
      "step": 155208
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.4101803302764893,
      "learning_rate": 0.00014434361662109195,
      "loss": 2.0641,
      "step": 155209
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.0903064012527466,
      "learning_rate": 0.00014434011976736087,
      "loss": 1.9059,
      "step": 155210
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.4415333271026611,
      "learning_rate": 0.00014433662294256982,
      "loss": 2.1736,
      "step": 155211
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.4091376066207886,
      "learning_rate": 0.0001443331261467196,
      "loss": 2.211,
      "step": 155212
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.3638787269592285,
      "learning_rate": 0.0001443296293798107,
      "loss": 2.134,
      "step": 155213
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.1224133968353271,
      "learning_rate": 0.00014432613264184397,
      "loss": 1.9944,
      "step": 155214
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.6117346286773682,
      "learning_rate": 0.00014432263593281984,
      "loss": 2.2156,
      "step": 155215
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.2654651403427124,
      "learning_rate": 0.0001443191392527393,
      "loss": 1.8822,
      "step": 155216
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.6228166818618774,
      "learning_rate": 0.00014431564260160255,
      "loss": 2.0511,
      "step": 155217
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.427671194076538,
      "learning_rate": 0.00014431214597941056,
      "loss": 1.946,
      "step": 155218
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.2415220737457275,
      "learning_rate": 0.00014430864938616377,
      "loss": 2.2698,
      "step": 155219
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.2839536666870117,
      "learning_rate": 0.00014430515282186306,
      "loss": 1.8991,
      "step": 155220
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.3172413110733032,
      "learning_rate": 0.00014430165628650883,
      "loss": 2.3717,
      "step": 155221
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.2061532735824585,
      "learning_rate": 0.0001442981597801021,
      "loss": 2.0119,
      "step": 155222
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.2733346223831177,
      "learning_rate": 0.00014429466330264306,
      "loss": 2.1965,
      "step": 155223
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.2498270273208618,
      "learning_rate": 0.0001442911668541327,
      "loss": 1.885,
      "step": 155224
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.2676368951797485,
      "learning_rate": 0.00014428767043457143,
      "loss": 2.1216,
      "step": 155225
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.241623044013977,
      "learning_rate": 0.00014428417404396013,
      "loss": 1.9289,
      "step": 155226
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.3794913291931152,
      "learning_rate": 0.00014428067768229924,
      "loss": 1.8989,
      "step": 155227
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.3139029741287231,
      "learning_rate": 0.0001442771813495897,
      "loss": 2.1141,
      "step": 155228
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.1610480546951294,
      "learning_rate": 0.0001442736850458318,
      "loss": 2.0213,
      "step": 155229
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.3119263648986816,
      "learning_rate": 0.00014427018877102642,
      "loss": 1.8624,
      "step": 155230
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.1872777938842773,
      "learning_rate": 0.0001442666925251741,
      "loss": 1.8694,
      "step": 155231
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.184274673461914,
      "learning_rate": 0.0001442631963082756,
      "loss": 1.9594,
      "step": 155232
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.0374329090118408,
      "learning_rate": 0.00014425970012033144,
      "loss": 1.9953,
      "step": 155233
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.1458402872085571,
      "learning_rate": 0.00014425620396134255,
      "loss": 1.9614,
      "step": 155234
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.146113634109497,
      "learning_rate": 0.00014425270783130912,
      "loss": 1.9986,
      "step": 155235
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.107276201248169,
      "learning_rate": 0.0001442492117302322,
      "loss": 1.9188,
      "step": 155236
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.114186406135559,
      "learning_rate": 0.00014424571565811216,
      "loss": 2.1328,
      "step": 155237
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.21533203125,
      "learning_rate": 0.00014424221961494992,
      "loss": 2.0343,
      "step": 155238
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.1697288751602173,
      "learning_rate": 0.00014423872360074587,
      "loss": 2.148,
      "step": 155239
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.3682738542556763,
      "learning_rate": 0.00014423522761550098,
      "loss": 1.946,
      "step": 155240
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.2954827547073364,
      "learning_rate": 0.0001442317316592155,
      "loss": 2.1353,
      "step": 155241
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.0976208448410034,
      "learning_rate": 0.00014422823573189038,
      "loss": 1.9009,
      "step": 155242
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.2763352394104004,
      "learning_rate": 0.00014422473983352605,
      "loss": 2.01,
      "step": 155243
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.1336945295333862,
      "learning_rate": 0.00014422124396412344,
      "loss": 2.1364,
      "step": 155244
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.2077349424362183,
      "learning_rate": 0.00014421774812368288,
      "loss": 2.1197,
      "step": 155245
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.1372864246368408,
      "learning_rate": 0.00014421425231220546,
      "loss": 1.9745,
      "step": 155246
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.2186297178268433,
      "learning_rate": 0.00014421075652969125,
      "loss": 2.2209,
      "step": 155247
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.2214348316192627,
      "learning_rate": 0.0001442072607761414,
      "loss": 1.9313,
      "step": 155248
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.1775896549224854,
      "learning_rate": 0.00014420376505155618,
      "loss": 1.9719,
      "step": 155249
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.1368796825408936,
      "learning_rate": 0.00014420026935593654,
      "loss": 2.008,
      "step": 155250
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.1672368049621582,
      "learning_rate": 0.00014419677368928294,
      "loss": 2.1833,
      "step": 155251
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.1177456378936768,
      "learning_rate": 0.0001441932780515963,
      "loss": 2.0173,
      "step": 155252
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.2088706493377686,
      "learning_rate": 0.00014418978244287685,
      "loss": 1.9906,
      "step": 155253
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.0734411478042603,
      "learning_rate": 0.00014418628686312554,
      "loss": 2.0764,
      "step": 155254
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.1434626579284668,
      "learning_rate": 0.00014418279131234285,
      "loss": 1.7841,
      "step": 155255
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.3295390605926514,
      "learning_rate": 0.00014417929579052966,
      "loss": 2.0202,
      "step": 155256
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.2749783992767334,
      "learning_rate": 0.00014417580029768635,
      "loss": 2.1805,
      "step": 155257
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.0758622884750366,
      "learning_rate": 0.00014417230483381393,
      "loss": 2.0494,
      "step": 155258
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.2190325260162354,
      "learning_rate": 0.00014416880939891258,
      "loss": 1.9819,
      "step": 155259
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.0688785314559937,
      "learning_rate": 0.0001441653139929833,
      "loss": 2.2288,
      "step": 155260
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.1833046674728394,
      "learning_rate": 0.00014416181861602655,
      "loss": 2.2275,
      "step": 155261
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.1190053224563599,
      "learning_rate": 0.00014415832326804316,
      "loss": 2.0226,
      "step": 155262
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.1051703691482544,
      "learning_rate": 0.00014415482794903357,
      "loss": 2.1654,
      "step": 155263
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.1076154708862305,
      "learning_rate": 0.00014415133265899873,
      "loss": 1.9737,
      "step": 155264
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.282590389251709,
      "learning_rate": 0.0001441478373979389,
      "loss": 2.0549,
      "step": 155265
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.3640809059143066,
      "learning_rate": 0.00014414434216585505,
      "loss": 2.0464,
      "step": 155266
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.1202924251556396,
      "learning_rate": 0.00014414084696274756,
      "loss": 2.0659,
      "step": 155267
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.127519130706787,
      "learning_rate": 0.00014413735178861738,
      "loss": 2.1332,
      "step": 155268
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.1712538003921509,
      "learning_rate": 0.0001441338566434649,
      "loss": 2.0439,
      "step": 155269
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.1339937448501587,
      "learning_rate": 0.00014413036152729095,
      "loss": 2.045,
      "step": 155270
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.322988510131836,
      "learning_rate": 0.0001441268664400961,
      "loss": 2.0184,
      "step": 155271
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.1443171501159668,
      "learning_rate": 0.00014412337138188107,
      "loss": 2.1342,
      "step": 155272
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.1385663747787476,
      "learning_rate": 0.00014411987635264627,
      "loss": 1.946,
      "step": 155273
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.1223186254501343,
      "learning_rate": 0.0001441163813523927,
      "loss": 2.1271,
      "step": 155274
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.106210470199585,
      "learning_rate": 0.0001441128863811207,
      "loss": 2.0002,
      "step": 155275
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.2723517417907715,
      "learning_rate": 0.00014410939143883115,
      "loss": 2.0116,
      "step": 155276
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.1925660371780396,
      "learning_rate": 0.0001441058965255246,
      "loss": 2.035,
      "step": 155277
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.205068588256836,
      "learning_rate": 0.00014410240164120164,
      "loss": 2.1577,
      "step": 155278
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.355600118637085,
      "learning_rate": 0.0001440989067858631,
      "loss": 1.8084,
      "step": 155279
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.1442996263504028,
      "learning_rate": 0.00014409541195950948,
      "loss": 1.806,
      "step": 155280
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.2715173959732056,
      "learning_rate": 0.00014409191716214137,
      "loss": 1.968,
      "step": 155281
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.2074369192123413,
      "learning_rate": 0.00014408842239375962,
      "loss": 1.8999,
      "step": 155282
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.6201932430267334,
      "learning_rate": 0.00014408492765436478,
      "loss": 1.9121,
      "step": 155283
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.2383184432983398,
      "learning_rate": 0.0001440814329439574,
      "loss": 2.0656,
      "step": 155284
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.2588016986846924,
      "learning_rate": 0.00014407793826253833,
      "loss": 2.1691,
      "step": 155285
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.197502851486206,
      "learning_rate": 0.00014407444361010807,
      "loss": 2.133,
      "step": 155286
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.4524078369140625,
      "learning_rate": 0.00014407094898666726,
      "loss": 2.0436,
      "step": 155287
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.2190237045288086,
      "learning_rate": 0.00014406745439221667,
      "loss": 2.1066,
      "step": 155288
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.2406731843948364,
      "learning_rate": 0.0001440639598267569,
      "loss": 2.1768,
      "step": 155289
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.1451908349990845,
      "learning_rate": 0.00014406046529028848,
      "loss": 1.9551,
      "step": 155290
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.088191032409668,
      "learning_rate": 0.00014405697078281228,
      "loss": 1.9087,
      "step": 155291
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.2590054273605347,
      "learning_rate": 0.0001440534763043287,
      "loss": 2.1504,
      "step": 155292
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.1287028789520264,
      "learning_rate": 0.00014404998185483866,
      "loss": 1.7643,
      "step": 155293
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.3288737535476685,
      "learning_rate": 0.00014404648743434267,
      "loss": 1.9901,
      "step": 155294
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.2448614835739136,
      "learning_rate": 0.00014404299304284136,
      "loss": 2.1335,
      "step": 155295
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.1586575508117676,
      "learning_rate": 0.0001440394986803353,
      "loss": 1.9011,
      "step": 155296
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.1279810667037964,
      "learning_rate": 0.00014403600434682536,
      "loss": 2.2374,
      "step": 155297
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.19585382938385,
      "learning_rate": 0.00014403251004231195,
      "loss": 2.1936,
      "step": 155298
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.2662460803985596,
      "learning_rate": 0.00014402901576679594,
      "loss": 2.0681,
      "step": 155299
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.1412440538406372,
      "learning_rate": 0.0001440255215202779,
      "loss": 1.9815,
      "step": 155300
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.2248504161834717,
      "learning_rate": 0.00014402202730275844,
      "loss": 1.9761,
      "step": 155301
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.1165701150894165,
      "learning_rate": 0.00014401853311423815,
      "loss": 1.9872,
      "step": 155302
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.247202754020691,
      "learning_rate": 0.00014401503895471787,
      "loss": 2.1466,
      "step": 155303
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.311798095703125,
      "learning_rate": 0.000144011544824198,
      "loss": 2.1491,
      "step": 155304
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.1864348649978638,
      "learning_rate": 0.00014400805072267947,
      "loss": 2.2426,
      "step": 155305
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.1381826400756836,
      "learning_rate": 0.00014400455665016264,
      "loss": 1.9267,
      "step": 155306
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.1831116676330566,
      "learning_rate": 0.00014400106260664856,
      "loss": 2.0092,
      "step": 155307
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.1915689706802368,
      "learning_rate": 0.00014399756859213736,
      "loss": 2.1097,
      "step": 155308
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.2352051734924316,
      "learning_rate": 0.0001439940746066301,
      "loss": 1.951,
      "step": 155309
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.098846197128296,
      "learning_rate": 0.0001439905806501272,
      "loss": 2.1081,
      "step": 155310
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.0630998611450195,
      "learning_rate": 0.00014398708672262947,
      "loss": 2.1598,
      "step": 155311
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.3447954654693604,
      "learning_rate": 0.0001439835928241374,
      "loss": 1.8273,
      "step": 155312
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.3618319034576416,
      "learning_rate": 0.00014398009895465191,
      "loss": 2.0201,
      "step": 155313
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.0858720541000366,
      "learning_rate": 0.0001439766051141733,
      "loss": 2.105,
      "step": 155314
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.130751609802246,
      "learning_rate": 0.00014397311130270247,
      "loss": 2.0262,
      "step": 155315
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.1169334650039673,
      "learning_rate": 0.00014396961752023983,
      "loss": 2.06,
      "step": 155316
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.1927826404571533,
      "learning_rate": 0.00014396612376678635,
      "loss": 2.0719,
      "step": 155317
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.2131047248840332,
      "learning_rate": 0.00014396263004234236,
      "loss": 1.9453,
      "step": 155318
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.1232253313064575,
      "learning_rate": 0.0001439591363469089,
      "loss": 1.9628,
      "step": 155319
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.2414062023162842,
      "learning_rate": 0.0001439556426804862,
      "loss": 1.9791,
      "step": 155320
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.137700080871582,
      "learning_rate": 0.00014395214904307514,
      "loss": 2.193,
      "step": 155321
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.1431432962417603,
      "learning_rate": 0.00014394865543467623,
      "loss": 1.8264,
      "step": 155322
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.2425562143325806,
      "learning_rate": 0.00014394516185529035,
      "loss": 1.9325,
      "step": 155323
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.1723756790161133,
      "learning_rate": 0.00014394166830491786,
      "loss": 2.0557,
      "step": 155324
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.1425615549087524,
      "learning_rate": 0.00014393817478355983,
      "loss": 1.8072,
      "step": 155325
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.124226689338684,
      "learning_rate": 0.0001439346812912164,
      "loss": 2.1147,
      "step": 155326
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.334839105606079,
      "learning_rate": 0.00014393118782788855,
      "loss": 2.1128,
      "step": 155327
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.131143569946289,
      "learning_rate": 0.0001439276943935767,
      "loss": 1.9892,
      "step": 155328
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.0760079622268677,
      "learning_rate": 0.0001439242009882818,
      "loss": 2.3,
      "step": 155329
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.391158938407898,
      "learning_rate": 0.00014392070761200422,
      "loss": 2.1018,
      "step": 155330
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.260137677192688,
      "learning_rate": 0.00014391721426474491,
      "loss": 2.2909,
      "step": 155331
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.149957537651062,
      "learning_rate": 0.00014391372094650417,
      "loss": 1.917,
      "step": 155332
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.1738576889038086,
      "learning_rate": 0.0001439102276572829,
      "loss": 2.0462,
      "step": 155333
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.2813448905944824,
      "learning_rate": 0.00014390673439708155,
      "loss": 2.0856,
      "step": 155334
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.0764491558074951,
      "learning_rate": 0.000143903241165901,
      "loss": 2.0755,
      "step": 155335
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.1124504804611206,
      "learning_rate": 0.00014389974796374166,
      "loss": 2.1035,
      "step": 155336
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.4705320596694946,
      "learning_rate": 0.00014389625479060446,
      "loss": 1.9959,
      "step": 155337
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.218762755393982,
      "learning_rate": 0.0001438927616464898,
      "loss": 2.2408,
      "step": 155338
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.0902667045593262,
      "learning_rate": 0.0001438892685313985,
      "loss": 2.2066,
      "step": 155339
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.1276670694351196,
      "learning_rate": 0.000143885775445331,
      "loss": 1.9534,
      "step": 155340
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.230857491493225,
      "learning_rate": 0.00014388228238828822,
      "loss": 2.0705,
      "step": 155341
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.1744049787521362,
      "learning_rate": 0.00014387878936027054,
      "loss": 2.0779,
      "step": 155342
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.193400502204895,
      "learning_rate": 0.00014387529636127885,
      "loss": 2.0197,
      "step": 155343
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.1634835004806519,
      "learning_rate": 0.00014387180339131369,
      "loss": 1.9831,
      "step": 155344
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.2149903774261475,
      "learning_rate": 0.00014386831045037568,
      "loss": 2.0281,
      "step": 155345
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.294970154762268,
      "learning_rate": 0.00014386481753846542,
      "loss": 1.986,
      "step": 155346
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.1417312622070312,
      "learning_rate": 0.00014386132465558373,
      "loss": 2.128,
      "step": 155347
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.171616554260254,
      "learning_rate": 0.00014385783180173107,
      "loss": 2.1455,
      "step": 155348
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.0927854776382446,
      "learning_rate": 0.0001438543389769083,
      "loss": 1.9135,
      "step": 155349
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.2700653076171875,
      "learning_rate": 0.00014385084618111595,
      "loss": 2.0436,
      "step": 155350
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.159705400466919,
      "learning_rate": 0.00014384735341435467,
      "loss": 2.0481,
      "step": 155351
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.2151894569396973,
      "learning_rate": 0.000143843860676625,
      "loss": 2.0623,
      "step": 155352
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.3968662023544312,
      "learning_rate": 0.00014384036796792783,
      "loss": 2.2034,
      "step": 155353
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.1035387516021729,
      "learning_rate": 0.00014383687528826358,
      "loss": 2.1473,
      "step": 155354
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.238128423690796,
      "learning_rate": 0.0001438333826376331,
      "loss": 2.0994,
      "step": 155355
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.0924220085144043,
      "learning_rate": 0.00014382989001603694,
      "loss": 1.8255,
      "step": 155356
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.2821972370147705,
      "learning_rate": 0.00014382639742347575,
      "loss": 2.2048,
      "step": 155357
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.1408594846725464,
      "learning_rate": 0.00014382290485995006,
      "loss": 2.0961,
      "step": 155358
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.137993335723877,
      "learning_rate": 0.00014381941232546078,
      "loss": 2.0382,
      "step": 155359
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.1751668453216553,
      "learning_rate": 0.00014381591982000826,
      "loss": 2.0448,
      "step": 155360
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.20536470413208,
      "learning_rate": 0.00014381242734359345,
      "loss": 2.0019,
      "step": 155361
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.1180928945541382,
      "learning_rate": 0.00014380893489621685,
      "loss": 2.0938,
      "step": 155362
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.2545877695083618,
      "learning_rate": 0.00014380544247787915,
      "loss": 2.1328,
      "step": 155363
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.1235876083374023,
      "learning_rate": 0.00014380195008858085,
      "loss": 1.7709,
      "step": 155364
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.2186670303344727,
      "learning_rate": 0.0001437984577283228,
      "loss": 2.1332,
      "step": 155365
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.2221580743789673,
      "learning_rate": 0.00014379496539710545,
      "loss": 2.1841,
      "step": 155366
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.2099961042404175,
      "learning_rate": 0.0001437914730949297,
      "loss": 2.1415,
      "step": 155367
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.1151669025421143,
      "learning_rate": 0.00014378798082179608,
      "loss": 2.0745,
      "step": 155368
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.1374722719192505,
      "learning_rate": 0.00014378448857770506,
      "loss": 1.9548,
      "step": 155369
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.2471450567245483,
      "learning_rate": 0.0001437809963626576,
      "loss": 2.2819,
      "step": 155370
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.2007551193237305,
      "learning_rate": 0.0001437775041766542,
      "loss": 2.1209,
      "step": 155371
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.2112109661102295,
      "learning_rate": 0.00014377401201969537,
      "loss": 2.081,
      "step": 155372
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.1427364349365234,
      "learning_rate": 0.00014377051989178204,
      "loss": 2.1092,
      "step": 155373
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.1621147394180298,
      "learning_rate": 0.0001437670277929147,
      "loss": 2.0413,
      "step": 155374
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.1743404865264893,
      "learning_rate": 0.0001437635357230939,
      "loss": 2.0778,
      "step": 155375
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.2805551290512085,
      "learning_rate": 0.00014376004368232056,
      "loss": 1.9907,
      "step": 155376
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.1414239406585693,
      "learning_rate": 0.000143756551670595,
      "loss": 2.1544,
      "step": 155377
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.1971142292022705,
      "learning_rate": 0.00014375305968791823,
      "loss": 2.0082,
      "step": 155378
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.1505956649780273,
      "learning_rate": 0.0001437495677342907,
      "loss": 2.0679,
      "step": 155379
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.194790244102478,
      "learning_rate": 0.00014374607580971307,
      "loss": 2.0015,
      "step": 155380
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.222486972808838,
      "learning_rate": 0.00014374258391418586,
      "loss": 2.1443,
      "step": 155381
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.1727945804595947,
      "learning_rate": 0.00014373909204771,
      "loss": 2.283,
      "step": 155382
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.1461222171783447,
      "learning_rate": 0.00014373560021028586,
      "loss": 2.0701,
      "step": 155383
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.2492817640304565,
      "learning_rate": 0.0001437321084019143,
      "loss": 2.0001,
      "step": 155384
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.1167312860488892,
      "learning_rate": 0.00014372861662259595,
      "loss": 2.1118,
      "step": 155385
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.120098352432251,
      "learning_rate": 0.00014372512487233134,
      "loss": 1.9935,
      "step": 155386
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.1171718835830688,
      "learning_rate": 0.0001437216331511211,
      "loss": 2.0474,
      "step": 155387
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.2042055130004883,
      "learning_rate": 0.00014371814145896608,
      "loss": 2.2971,
      "step": 155388
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.1164426803588867,
      "learning_rate": 0.0001437146497958667,
      "loss": 1.6888,
      "step": 155389
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.0855369567871094,
      "learning_rate": 0.00014371115816182381,
      "loss": 2.0722,
      "step": 155390
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.1060757637023926,
      "learning_rate": 0.00014370766655683785,
      "loss": 1.8904,
      "step": 155391
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.097657322883606,
      "learning_rate": 0.00014370417498090984,
      "loss": 2.0535,
      "step": 155392
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.1214669942855835,
      "learning_rate": 0.0001437006834340399,
      "loss": 2.0934,
      "step": 155393
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.1308962106704712,
      "learning_rate": 0.00014369719191622911,
      "loss": 2.1078,
      "step": 155394
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.1557506322860718,
      "learning_rate": 0.0001436937004274778,
      "loss": 2.0286,
      "step": 155395
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.1193132400512695,
      "learning_rate": 0.00014369020896778695,
      "loss": 1.9073,
      "step": 155396
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.1455072164535522,
      "learning_rate": 0.0001436867175371569,
      "loss": 2.1084,
      "step": 155397
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.1659107208251953,
      "learning_rate": 0.0001436832261355887,
      "loss": 2.0845,
      "step": 155398
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.2333402633666992,
      "learning_rate": 0.00014367973476308248,
      "loss": 2.0094,
      "step": 155399
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.2363568544387817,
      "learning_rate": 0.00014367624341963926,
      "loss": 2.0696,
      "step": 155400
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.178494930267334,
      "learning_rate": 0.00014367275210525946,
      "loss": 2.0546,
      "step": 155401
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.2292215824127197,
      "learning_rate": 0.00014366926081994398,
      "loss": 2.046,
      "step": 155402
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.2733502388000488,
      "learning_rate": 0.00014366576956369322,
      "loss": 2.1585,
      "step": 155403
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.2349056005477905,
      "learning_rate": 0.00014366227833650805,
      "loss": 2.0678,
      "step": 155404
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.293325662612915,
      "learning_rate": 0.00014365878713838901,
      "loss": 2.1336,
      "step": 155405
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.1662297248840332,
      "learning_rate": 0.00014365529596933676,
      "loss": 2.0299,
      "step": 155406
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.1278834342956543,
      "learning_rate": 0.00014365180482935183,
      "loss": 2.1598,
      "step": 155407
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.088496208190918,
      "learning_rate": 0.00014364831371843506,
      "loss": 1.8536,
      "step": 155408
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.4025046825408936,
      "learning_rate": 0.00014364482263658695,
      "loss": 2.0209,
      "step": 155409
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.179439663887024,
      "learning_rate": 0.0001436413315838083,
      "loss": 2.127,
      "step": 155410
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.0655511617660522,
      "learning_rate": 0.00014363784056009972,
      "loss": 2.0141,
      "step": 155411
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.0909323692321777,
      "learning_rate": 0.0001436343495654618,
      "loss": 1.9621,
      "step": 155412
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.1063356399536133,
      "learning_rate": 0.00014363085859989507,
      "loss": 2.1068,
      "step": 155413
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.6041719913482666,
      "learning_rate": 0.00014362736766340042,
      "loss": 2.0786,
      "step": 155414
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.1506496667861938,
      "learning_rate": 0.00014362387675597834,
      "loss": 2.151,
      "step": 155415
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.1619268655776978,
      "learning_rate": 0.0001436203858776296,
      "loss": 1.8782,
      "step": 155416
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.0199270248413086,
      "learning_rate": 0.0001436168950283548,
      "loss": 1.8515,
      "step": 155417
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.2393603324890137,
      "learning_rate": 0.00014361340420815456,
      "loss": 1.9263,
      "step": 155418
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.5218738317489624,
      "learning_rate": 0.00014360991341702943,
      "loss": 2.0416,
      "step": 155419
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.2107938528060913,
      "learning_rate": 0.00014360642265498028,
      "loss": 2.1227,
      "step": 155420
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.8093535900115967,
      "learning_rate": 0.00014360293192200752,
      "loss": 2.1707,
      "step": 155421
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.2348333597183228,
      "learning_rate": 0.00014359944121811207,
      "loss": 2.2229,
      "step": 155422
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.3074012994766235,
      "learning_rate": 0.00014359595054329444,
      "loss": 2.0449,
      "step": 155423
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.2415437698364258,
      "learning_rate": 0.00014359245989755524,
      "loss": 2.0953,
      "step": 155424
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.170426368713379,
      "learning_rate": 0.00014358896928089507,
      "loss": 2.1179,
      "step": 155425
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.3613709211349487,
      "learning_rate": 0.00014358547869331475,
      "loss": 1.9239,
      "step": 155426
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.1329947710037231,
      "learning_rate": 0.00014358198813481477,
      "loss": 2.0144,
      "step": 155427
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.2395766973495483,
      "learning_rate": 0.00014357849760539593,
      "loss": 1.9944,
      "step": 155428
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.3178564310073853,
      "learning_rate": 0.00014357500710505882,
      "loss": 2.2474,
      "step": 155429
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.1380800008773804,
      "learning_rate": 0.00014357151663380404,
      "loss": 2.0779,
      "step": 155430
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.1570398807525635,
      "learning_rate": 0.00014356802619163217,
      "loss": 2.0365,
      "step": 155431
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.2873034477233887,
      "learning_rate": 0.00014356453577854409,
      "loss": 2.1702,
      "step": 155432
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.4020220041275024,
      "learning_rate": 0.00014356104539454015,
      "loss": 1.992,
      "step": 155433
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.1799662113189697,
      "learning_rate": 0.00014355755503962135,
      "loss": 1.8921,
      "step": 155434
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.128505825996399,
      "learning_rate": 0.00014355406471378812,
      "loss": 2.0941,
      "step": 155435
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.1889806985855103,
      "learning_rate": 0.00014355057441704113,
      "loss": 1.8618,
      "step": 155436
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.5451419353485107,
      "learning_rate": 0.00014354708414938093,
      "loss": 2.058,
      "step": 155437
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.1287585496902466,
      "learning_rate": 0.0001435435939108084,
      "loss": 2.1719,
      "step": 155438
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.203033685684204,
      "learning_rate": 0.00014354010370132395,
      "loss": 1.8617,
      "step": 155439
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.1658718585968018,
      "learning_rate": 0.0001435366135209285,
      "loss": 1.9475,
      "step": 155440
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.196638822555542,
      "learning_rate": 0.0001435331233696225,
      "loss": 1.925,
      "step": 155441
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.287469744682312,
      "learning_rate": 0.00014352963324740667,
      "loss": 1.9425,
      "step": 155442
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.1715917587280273,
      "learning_rate": 0.0001435261431542815,
      "loss": 1.9097,
      "step": 155443
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.1523354053497314,
      "learning_rate": 0.0001435226530902479,
      "loss": 2.154,
      "step": 155444
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.0864391326904297,
      "learning_rate": 0.00014351916305530628,
      "loss": 1.9464,
      "step": 155445
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.2301557064056396,
      "learning_rate": 0.00014351567304945754,
      "loss": 2.0449,
      "step": 155446
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.1259195804595947,
      "learning_rate": 0.00014351218307270214,
      "loss": 1.9338,
      "step": 155447
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.1659811735153198,
      "learning_rate": 0.00014350869312504082,
      "loss": 2.0562,
      "step": 155448
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.1931425333023071,
      "learning_rate": 0.00014350520320647404,
      "loss": 2.0476,
      "step": 155449
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.103777527809143,
      "learning_rate": 0.00014350171331700275,
      "loss": 2.0838,
      "step": 155450
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.1696375608444214,
      "learning_rate": 0.0001434982234566273,
      "loss": 1.9088,
      "step": 155451
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.2077504396438599,
      "learning_rate": 0.0001434947336253486,
      "loss": 1.8462,
      "step": 155452
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.1303385496139526,
      "learning_rate": 0.0001434912438231672,
      "loss": 2.2322,
      "step": 155453
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.097804307937622,
      "learning_rate": 0.00014348775405008363,
      "loss": 2.068,
      "step": 155454
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.066390872001648,
      "learning_rate": 0.00014348426430609872,
      "loss": 2.1523,
      "step": 155455
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.2590429782867432,
      "learning_rate": 0.00014348077459121303,
      "loss": 1.8784,
      "step": 155456
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.1728262901306152,
      "learning_rate": 0.00014347728490542714,
      "loss": 2.2008,
      "step": 155457
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.204329490661621,
      "learning_rate": 0.00014347379524874189,
      "loss": 2.1613,
      "step": 155458
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.175495982170105,
      "learning_rate": 0.0001434703056211578,
      "loss": 1.8844,
      "step": 155459
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.0907261371612549,
      "learning_rate": 0.00014346681602267546,
      "loss": 1.9515,
      "step": 155460
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.1757944822311401,
      "learning_rate": 0.00014346332645329566,
      "loss": 1.8178,
      "step": 155461
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.1520802974700928,
      "learning_rate": 0.000143459836913019,
      "loss": 1.9476,
      "step": 155462
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.1806148290634155,
      "learning_rate": 0.00014345634740184598,
      "loss": 2.0751,
      "step": 155463
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.0950050354003906,
      "learning_rate": 0.0001434528579197775,
      "loss": 1.9352,
      "step": 155464
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.0954711437225342,
      "learning_rate": 0.00014344936846681414,
      "loss": 2.1739,
      "step": 155465
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.1584343910217285,
      "learning_rate": 0.0001434458790429563,
      "loss": 1.9972,
      "step": 155466
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.2425782680511475,
      "learning_rate": 0.000143442389648205,
      "loss": 2.3455,
      "step": 155467
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.0771368741989136,
      "learning_rate": 0.00014343890028256057,
      "loss": 2.1711,
      "step": 155468
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.1088935136795044,
      "learning_rate": 0.00014343541094602395,
      "loss": 2.1756,
      "step": 155469
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.1148101091384888,
      "learning_rate": 0.00014343192163859562,
      "loss": 1.989,
      "step": 155470
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.0937286615371704,
      "learning_rate": 0.00014342843236027616,
      "loss": 2.1209,
      "step": 155471
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.146885633468628,
      "learning_rate": 0.0001434249431110664,
      "loss": 2.062,
      "step": 155472
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.222710132598877,
      "learning_rate": 0.0001434214538909669,
      "loss": 1.9736,
      "step": 155473
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.2827852964401245,
      "learning_rate": 0.0001434179646999782,
      "loss": 2.1256,
      "step": 155474
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.1204473972320557,
      "learning_rate": 0.00014341447553810113,
      "loss": 2.0966,
      "step": 155475
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.1719706058502197,
      "learning_rate": 0.0001434109864053362,
      "loss": 2.0082,
      "step": 155476
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.1322076320648193,
      "learning_rate": 0.00014340749730168423,
      "loss": 2.3057,
      "step": 155477
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.2636678218841553,
      "learning_rate": 0.00014340400822714574,
      "loss": 2.0564,
      "step": 155478
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.1658191680908203,
      "learning_rate": 0.00014340051918172142,
      "loss": 1.9977,
      "step": 155479
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.2203495502471924,
      "learning_rate": 0.0001433970301654118,
      "loss": 2.1301,
      "step": 155480
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.162664771080017,
      "learning_rate": 0.0001433935411782177,
      "loss": 1.9765,
      "step": 155481
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.043965220451355,
      "learning_rate": 0.00014339005222013964,
      "loss": 1.8324,
      "step": 155482
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.308941125869751,
      "learning_rate": 0.0001433865632911784,
      "loss": 2.3013,
      "step": 155483
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.1966207027435303,
      "learning_rate": 0.00014338307439133456,
      "loss": 1.8802,
      "step": 155484
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.3495866060256958,
      "learning_rate": 0.0001433795855206088,
      "loss": 2.2612,
      "step": 155485
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.2778072357177734,
      "learning_rate": 0.00014337609667900152,
      "loss": 2.0269,
      "step": 155486
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.1390368938446045,
      "learning_rate": 0.00014337260786651378,
      "loss": 1.9759,
      "step": 155487
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.3111506700515747,
      "learning_rate": 0.0001433691190831459,
      "loss": 1.8455,
      "step": 155488
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.1811261177062988,
      "learning_rate": 0.00014336563032889876,
      "loss": 2.0038,
      "step": 155489
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.3016211986541748,
      "learning_rate": 0.0001433621416037729,
      "loss": 1.9573,
      "step": 155490
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.2074023485183716,
      "learning_rate": 0.00014335865290776898,
      "loss": 2.0557,
      "step": 155491
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.336186408996582,
      "learning_rate": 0.00014335516424088753,
      "loss": 2.1983,
      "step": 155492
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.1391514539718628,
      "learning_rate": 0.00014335167560312943,
      "loss": 2.0091,
      "step": 155493
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.1302204132080078,
      "learning_rate": 0.0001433481869944951,
      "loss": 2.0556,
      "step": 155494
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.0709530115127563,
      "learning_rate": 0.00014334469841498543,
      "loss": 1.9248,
      "step": 155495
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.1488150358200073,
      "learning_rate": 0.00014334120986460089,
      "loss": 2.1224,
      "step": 155496
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.2834935188293457,
      "learning_rate": 0.00014333772134334218,
      "loss": 2.0274,
      "step": 155497
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.0608829259872437,
      "learning_rate": 0.0001433342328512099,
      "loss": 2.1992,
      "step": 155498
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.1018784046173096,
      "learning_rate": 0.0001433307443882048,
      "loss": 1.8408,
      "step": 155499
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.1794620752334595,
      "learning_rate": 0.00014332725595432737,
      "loss": 2.1791,
      "step": 155500
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.2629166841506958,
      "learning_rate": 0.00014332376754957847,
      "loss": 2.0562,
      "step": 155501
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.2712624073028564,
      "learning_rate": 0.00014332027917395864,
      "loss": 2.0729,
      "step": 155502
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.4965026378631592,
      "learning_rate": 0.0001433167908274685,
      "loss": 2.1675,
      "step": 155503
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.0953776836395264,
      "learning_rate": 0.00014331330251010868,
      "loss": 2.1716,
      "step": 155504
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.0519691705703735,
      "learning_rate": 0.00014330981422187993,
      "loss": 2.1676,
      "step": 155505
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.3222806453704834,
      "learning_rate": 0.00014330632596278275,
      "loss": 2.052,
      "step": 155506
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.1389763355255127,
      "learning_rate": 0.00014330283773281798,
      "loss": 1.8829,
      "step": 155507
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.1381134986877441,
      "learning_rate": 0.0001432993495319862,
      "loss": 2.0052,
      "step": 155508
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.1565049886703491,
      "learning_rate": 0.00014329586136028802,
      "loss": 1.8508,
      "step": 155509
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.2284060716629028,
      "learning_rate": 0.00014329237321772397,
      "loss": 2.1306,
      "step": 155510
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.4328316450119019,
      "learning_rate": 0.00014328888510429497,
      "loss": 2.0157,
      "step": 155511
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.1805462837219238,
      "learning_rate": 0.00014328539702000137,
      "loss": 2.1185,
      "step": 155512
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.2257853746414185,
      "learning_rate": 0.0001432819089648441,
      "loss": 2.0333,
      "step": 155513
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.236720323562622,
      "learning_rate": 0.00014327842093882367,
      "loss": 2.057,
      "step": 155514
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.1522908210754395,
      "learning_rate": 0.00014327493294194078,
      "loss": 2.0492,
      "step": 155515
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.3999847173690796,
      "learning_rate": 0.00014327144497419588,
      "loss": 2.2569,
      "step": 155516
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.051383137702942,
      "learning_rate": 0.00014326795703558992,
      "loss": 2.0941,
      "step": 155517
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.1739437580108643,
      "learning_rate": 0.0001432644691261233,
      "loss": 2.0634,
      "step": 155518
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.09956693649292,
      "learning_rate": 0.0001432609812457969,
      "loss": 2.3002,
      "step": 155519
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.0769169330596924,
      "learning_rate": 0.00014325749339461118,
      "loss": 2.0029,
      "step": 155520
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.1712820529937744,
      "learning_rate": 0.0001432540055725669,
      "loss": 1.9369,
      "step": 155521
    },
    {
      "epoch": 2.02,
      "grad_norm": 1.1606519222259521,
      "learning_rate": 0.0001432505177796645,
      "loss": 2.1677,
      "step": 155522
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.2504099607467651,
      "learning_rate": 0.00014324703001590493,
      "loss": 1.9926,
      "step": 155523
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.1978700160980225,
      "learning_rate": 0.0001432435422812886,
      "loss": 2.238,
      "step": 155524
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.2262414693832397,
      "learning_rate": 0.00014324005457581638,
      "loss": 2.051,
      "step": 155525
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.1351494789123535,
      "learning_rate": 0.00014323656689948873,
      "loss": 2.2944,
      "step": 155526
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.2703865766525269,
      "learning_rate": 0.0001432330792523064,
      "loss": 2.1559,
      "step": 155527
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.0881028175354004,
      "learning_rate": 0.00014322959163426988,
      "loss": 1.8391,
      "step": 155528
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.1976351737976074,
      "learning_rate": 0.00014322610404538005,
      "loss": 2.0217,
      "step": 155529
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.152834415435791,
      "learning_rate": 0.00014322261648563737,
      "loss": 1.9727,
      "step": 155530
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.178815484046936,
      "learning_rate": 0.0001432191289550426,
      "loss": 2.0433,
      "step": 155531
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.3149033784866333,
      "learning_rate": 0.0001432156414535964,
      "loss": 1.9484,
      "step": 155532
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.2524528503417969,
      "learning_rate": 0.0001432121539812994,
      "loss": 2.22,
      "step": 155533
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.1919891834259033,
      "learning_rate": 0.00014320866653815207,
      "loss": 2.0605,
      "step": 155534
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.446480631828308,
      "learning_rate": 0.00014320517912415532,
      "loss": 2.1174,
      "step": 155535
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.3009971380233765,
      "learning_rate": 0.00014320169173930958,
      "loss": 2.0234,
      "step": 155536
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.202722191810608,
      "learning_rate": 0.00014319820438361572,
      "loss": 2.091,
      "step": 155537
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.2206746339797974,
      "learning_rate": 0.00014319471705707414,
      "loss": 2.0937,
      "step": 155538
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.2523388862609863,
      "learning_rate": 0.00014319122975968578,
      "loss": 2.0895,
      "step": 155539
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.2571974992752075,
      "learning_rate": 0.0001431877424914511,
      "loss": 1.9982,
      "step": 155540
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.1731548309326172,
      "learning_rate": 0.0001431842552523708,
      "loss": 1.8626,
      "step": 155541
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.1470142602920532,
      "learning_rate": 0.0001431807680424454,
      "loss": 1.9607,
      "step": 155542
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.2413804531097412,
      "learning_rate": 0.00014317728086167574,
      "loss": 2.0494,
      "step": 155543
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.1633731126785278,
      "learning_rate": 0.00014317379371006227,
      "loss": 2.0387,
      "step": 155544
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.1161561012268066,
      "learning_rate": 0.0001431703065876059,
      "loss": 2.2492,
      "step": 155545
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.31184983253479,
      "learning_rate": 0.00014316681949430707,
      "loss": 1.8799,
      "step": 155546
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.1862013339996338,
      "learning_rate": 0.0001431633324301665,
      "loss": 2.1267,
      "step": 155547
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.239381194114685,
      "learning_rate": 0.00014315984539518475,
      "loss": 2.1722,
      "step": 155548
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.0621473789215088,
      "learning_rate": 0.00014315635838936267,
      "loss": 2.0092,
      "step": 155549
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.0655043125152588,
      "learning_rate": 0.0001431528714127006,
      "loss": 2.2993,
      "step": 155550
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.187073826789856,
      "learning_rate": 0.00014314938446519955,
      "loss": 2.1455,
      "step": 155551
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.2767232656478882,
      "learning_rate": 0.00014314589754685998,
      "loss": 2.1509,
      "step": 155552
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.3260319232940674,
      "learning_rate": 0.0001431424106576824,
      "loss": 2.1515,
      "step": 155553
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.127017617225647,
      "learning_rate": 0.0001431389237976678,
      "loss": 2.2068,
      "step": 155554
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.1860790252685547,
      "learning_rate": 0.00014313543696681656,
      "loss": 2.1147,
      "step": 155555
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.0771293640136719,
      "learning_rate": 0.00014313195016512928,
      "loss": 2.0996,
      "step": 155556
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.0451947450637817,
      "learning_rate": 0.0001431284633926069,
      "loss": 2.0142,
      "step": 155557
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.18098783493042,
      "learning_rate": 0.00014312497664924987,
      "loss": 1.9954,
      "step": 155558
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.1443415880203247,
      "learning_rate": 0.00014312148993505876,
      "loss": 2.0294,
      "step": 155559
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.408978819847107,
      "learning_rate": 0.00014311800325003443,
      "loss": 2.0344,
      "step": 155560
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.2041188478469849,
      "learning_rate": 0.00014311451659417747,
      "loss": 2.1174,
      "step": 155561
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.0447306632995605,
      "learning_rate": 0.0001431110299674883,
      "loss": 1.8327,
      "step": 155562
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.0415552854537964,
      "learning_rate": 0.00014310754336996793,
      "loss": 2.0063,
      "step": 155563
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.1808024644851685,
      "learning_rate": 0.00014310405680161679,
      "loss": 2.1013,
      "step": 155564
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.215535283088684,
      "learning_rate": 0.00014310057026243545,
      "loss": 2.0135,
      "step": 155565
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.3402903079986572,
      "learning_rate": 0.00014309708375242482,
      "loss": 2.1603,
      "step": 155566
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.1375536918640137,
      "learning_rate": 0.00014309359727158525,
      "loss": 2.0177,
      "step": 155567
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.1442030668258667,
      "learning_rate": 0.00014309011081991768,
      "loss": 2.3424,
      "step": 155568
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.1403617858886719,
      "learning_rate": 0.00014308662439742263,
      "loss": 2.0072,
      "step": 155569
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.0672966241836548,
      "learning_rate": 0.00014308313800410073,
      "loss": 2.0395,
      "step": 155570
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.1758270263671875,
      "learning_rate": 0.0001430796516399525,
      "loss": 2.1474,
      "step": 155571
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.1814028024673462,
      "learning_rate": 0.0001430761653049789,
      "loss": 1.9906,
      "step": 155572
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.1640857458114624,
      "learning_rate": 0.0001430726789991802,
      "loss": 1.9471,
      "step": 155573
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.208779215812683,
      "learning_rate": 0.00014306919272255743,
      "loss": 2.2148,
      "step": 155574
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.073675274848938,
      "learning_rate": 0.00014306570647511105,
      "loss": 2.1755,
      "step": 155575
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.1252251863479614,
      "learning_rate": 0.0001430622202568417,
      "loss": 2.1743,
      "step": 155576
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.0615999698638916,
      "learning_rate": 0.00014305873406774997,
      "loss": 1.9297,
      "step": 155577
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.1416057348251343,
      "learning_rate": 0.00014305524790783665,
      "loss": 1.991,
      "step": 155578
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.1894807815551758,
      "learning_rate": 0.00014305176177710222,
      "loss": 2.2512,
      "step": 155579
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.4985779523849487,
      "learning_rate": 0.00014304827567554755,
      "loss": 2.1964,
      "step": 155580
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.2923940420150757,
      "learning_rate": 0.0001430447896031732,
      "loss": 2.0156,
      "step": 155581
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.3338558673858643,
      "learning_rate": 0.00014304130355997975,
      "loss": 2.1122,
      "step": 155582
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.0808390378952026,
      "learning_rate": 0.00014303781754596777,
      "loss": 2.0095,
      "step": 155583
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.2331950664520264,
      "learning_rate": 0.00014303433156113814,
      "loss": 1.9001,
      "step": 155584
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.4640345573425293,
      "learning_rate": 0.00014303084560549127,
      "loss": 2.0902,
      "step": 155585
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.2380032539367676,
      "learning_rate": 0.00014302735967902807,
      "loss": 1.9943,
      "step": 155586
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.1768518686294556,
      "learning_rate": 0.000143023873781749,
      "loss": 2.1307,
      "step": 155587
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.2002263069152832,
      "learning_rate": 0.00014302038791365478,
      "loss": 2.3159,
      "step": 155588
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.1827095746994019,
      "learning_rate": 0.00014301690207474594,
      "loss": 2.1835,
      "step": 155589
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.2395174503326416,
      "learning_rate": 0.00014301341626502334,
      "loss": 1.9749,
      "step": 155590
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.066079020500183,
      "learning_rate": 0.00014300993048448735,
      "loss": 1.9404,
      "step": 155591
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.0697455406188965,
      "learning_rate": 0.00014300644473313893,
      "loss": 1.9844,
      "step": 155592
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.1820343732833862,
      "learning_rate": 0.00014300295901097852,
      "loss": 2.0606,
      "step": 155593
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.2368767261505127,
      "learning_rate": 0.0001429994733180069,
      "loss": 1.9598,
      "step": 155594
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.1981576681137085,
      "learning_rate": 0.0001429959876542245,
      "loss": 2.1615,
      "step": 155595
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.066297173500061,
      "learning_rate": 0.0001429925020196322,
      "loss": 1.9436,
      "step": 155596
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.3568224906921387,
      "learning_rate": 0.00014298901641423046,
      "loss": 1.9984,
      "step": 155597
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.050899863243103,
      "learning_rate": 0.00014298553083802015,
      "loss": 2.078,
      "step": 155598
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.299560308456421,
      "learning_rate": 0.00014298204529100176,
      "loss": 2.1639,
      "step": 155599
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.2030802965164185,
      "learning_rate": 0.00014297855977317599,
      "loss": 2.0727,
      "step": 155600
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.2057348489761353,
      "learning_rate": 0.00014297507428454338,
      "loss": 2.2168,
      "step": 155601
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.2041082382202148,
      "learning_rate": 0.00014297158882510473,
      "loss": 2.1266,
      "step": 155602
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.1089534759521484,
      "learning_rate": 0.00014296810339486056,
      "loss": 2.1384,
      "step": 155603
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.2334814071655273,
      "learning_rate": 0.00014296461799381168,
      "loss": 2.0873,
      "step": 155604
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.3634127378463745,
      "learning_rate": 0.00014296113262195854,
      "loss": 2.0469,
      "step": 155605
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.0939655303955078,
      "learning_rate": 0.0001429576472793021,
      "loss": 1.9216,
      "step": 155606
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.1453028917312622,
      "learning_rate": 0.00014295416196584255,
      "loss": 2.0401,
      "step": 155607
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.2688438892364502,
      "learning_rate": 0.00014295067668158097,
      "loss": 1.8535,
      "step": 155608
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.1622768640518188,
      "learning_rate": 0.00014294719142651766,
      "loss": 2.0081,
      "step": 155609
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.1294561624526978,
      "learning_rate": 0.00014294370620065354,
      "loss": 1.9395,
      "step": 155610
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.2043418884277344,
      "learning_rate": 0.00014294022100398905,
      "loss": 2.0183,
      "step": 155611
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.051612138748169,
      "learning_rate": 0.00014293673583652515,
      "loss": 1.9515,
      "step": 155612
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.1558302640914917,
      "learning_rate": 0.00014293325069826207,
      "loss": 2.1424,
      "step": 155613
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.141223669052124,
      "learning_rate": 0.00014292976558920076,
      "loss": 2.0663,
      "step": 155614
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.1687333583831787,
      "learning_rate": 0.00014292628050934167,
      "loss": 2.0738,
      "step": 155615
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.1205155849456787,
      "learning_rate": 0.00014292279545868568,
      "loss": 1.9642,
      "step": 155616
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.131182074546814,
      "learning_rate": 0.00014291931043723314,
      "loss": 1.9983,
      "step": 155617
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.306327223777771,
      "learning_rate": 0.00014291582544498512,
      "loss": 2.1595,
      "step": 155618
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.3588937520980835,
      "learning_rate": 0.00014291234048194182,
      "loss": 2.0908,
      "step": 155619
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.204918622970581,
      "learning_rate": 0.00014290885554810414,
      "loss": 2.0926,
      "step": 155620
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.272504448890686,
      "learning_rate": 0.0001429053706434726,
      "loss": 2.1179,
      "step": 155621
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.1979129314422607,
      "learning_rate": 0.000142901885768048,
      "loss": 2.0938,
      "step": 155622
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.072102427482605,
      "learning_rate": 0.00014289840092183084,
      "loss": 1.8548,
      "step": 155623
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.1674537658691406,
      "learning_rate": 0.000142894916104822,
      "loss": 2.1137,
      "step": 155624
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.2311087846755981,
      "learning_rate": 0.00014289143131702176,
      "loss": 2.1465,
      "step": 155625
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.2669727802276611,
      "learning_rate": 0.0001428879465584311,
      "loss": 2.089,
      "step": 155626
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.3734174966812134,
      "learning_rate": 0.00014288446182905037,
      "loss": 2.0871,
      "step": 155627
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.2438441514968872,
      "learning_rate": 0.00014288097712888054,
      "loss": 2.1998,
      "step": 155628
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.0699632167816162,
      "learning_rate": 0.00014287749245792196,
      "loss": 2.0828,
      "step": 155629
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.1286721229553223,
      "learning_rate": 0.00014287400781617558,
      "loss": 1.9828,
      "step": 155630
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.3051173686981201,
      "learning_rate": 0.00014287052320364186,
      "loss": 2.0488,
      "step": 155631
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.2459031343460083,
      "learning_rate": 0.00014286703862032146,
      "loss": 2.0701,
      "step": 155632
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.15506112575531,
      "learning_rate": 0.00014286355406621494,
      "loss": 2.047,
      "step": 155633
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.2302902936935425,
      "learning_rate": 0.00014286006954132316,
      "loss": 1.9614,
      "step": 155634
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.1297470331192017,
      "learning_rate": 0.00014285658504564657,
      "loss": 2.0093,
      "step": 155635
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.2205843925476074,
      "learning_rate": 0.000142853100579186,
      "loss": 1.9762,
      "step": 155636
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.1308990716934204,
      "learning_rate": 0.00014284961614194203,
      "loss": 2.0432,
      "step": 155637
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.1664340496063232,
      "learning_rate": 0.00014284613173391517,
      "loss": 1.9612,
      "step": 155638
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.242201566696167,
      "learning_rate": 0.00014284264735510625,
      "loss": 2.236,
      "step": 155639
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.1233004331588745,
      "learning_rate": 0.0001428391630055159,
      "loss": 2.0591,
      "step": 155640
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.1133369207382202,
      "learning_rate": 0.00014283567868514455,
      "loss": 1.7708,
      "step": 155641
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.1048260927200317,
      "learning_rate": 0.00014283219439399316,
      "loss": 2.1151,
      "step": 155642
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.1295137405395508,
      "learning_rate": 0.0001428287101320622,
      "loss": 2.2379,
      "step": 155643
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.2052624225616455,
      "learning_rate": 0.00014282522589935227,
      "loss": 2.1969,
      "step": 155644
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.1496373414993286,
      "learning_rate": 0.0001428217416958642,
      "loss": 2.0026,
      "step": 155645
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.2193113565444946,
      "learning_rate": 0.00014281825752159854,
      "loss": 1.9212,
      "step": 155646
    },
    {
      "epoch": 2.03,
      "grad_norm": 0.9998840093612671,
      "learning_rate": 0.00014281477337655584,
      "loss": 2.0003,
      "step": 155647
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.1675994396209717,
      "learning_rate": 0.0001428112892607369,
      "loss": 1.8998,
      "step": 155648
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.127153754234314,
      "learning_rate": 0.00014280780517414234,
      "loss": 1.9292,
      "step": 155649
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.240261435508728,
      "learning_rate": 0.00014280432111677267,
      "loss": 2.1034,
      "step": 155650
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.208113670349121,
      "learning_rate": 0.00014280083708862875,
      "loss": 2.1311,
      "step": 155651
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.217509388923645,
      "learning_rate": 0.00014279735308971104,
      "loss": 2.3264,
      "step": 155652
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.231157660484314,
      "learning_rate": 0.00014279386912002036,
      "loss": 2.1809,
      "step": 155653
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.2050403356552124,
      "learning_rate": 0.00014279038517955723,
      "loss": 2.1664,
      "step": 155654
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.1770514249801636,
      "learning_rate": 0.00014278690126832237,
      "loss": 2.0177,
      "step": 155655
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.2189871072769165,
      "learning_rate": 0.00014278341738631627,
      "loss": 2.0977,
      "step": 155656
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.2154195308685303,
      "learning_rate": 0.0001427799335335398,
      "loss": 2.2479,
      "step": 155657
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.511500358581543,
      "learning_rate": 0.00014277644970999342,
      "loss": 2.133,
      "step": 155658
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.2581398487091064,
      "learning_rate": 0.00014277296591567798,
      "loss": 1.9449,
      "step": 155659
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.593090534210205,
      "learning_rate": 0.00014276948215059403,
      "loss": 2.1627,
      "step": 155660
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.2206225395202637,
      "learning_rate": 0.00014276599841474218,
      "loss": 1.843,
      "step": 155661
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.181517243385315,
      "learning_rate": 0.00014276251470812296,
      "loss": 1.9336,
      "step": 155662
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.384557843208313,
      "learning_rate": 0.00014275903103073729,
      "loss": 2.3217,
      "step": 155663
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.2655810117721558,
      "learning_rate": 0.00014275554738258557,
      "loss": 2.0495,
      "step": 155664
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.0712815523147583,
      "learning_rate": 0.00014275206376366867,
      "loss": 1.9294,
      "step": 155665
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.1778696775436401,
      "learning_rate": 0.00014274858017398716,
      "loss": 1.9896,
      "step": 155666
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.3395832777023315,
      "learning_rate": 0.00014274509661354163,
      "loss": 1.918,
      "step": 155667
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.181357979774475,
      "learning_rate": 0.00014274161308233262,
      "loss": 2.1051,
      "step": 155668
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.1397669315338135,
      "learning_rate": 0.0001427381295803611,
      "loss": 2.1343,
      "step": 155669
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.1979061365127563,
      "learning_rate": 0.00014273464610762735,
      "loss": 2.0648,
      "step": 155670
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.218321442604065,
      "learning_rate": 0.00014273116266413233,
      "loss": 2.0374,
      "step": 155671
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.1879804134368896,
      "learning_rate": 0.00014272767924987644,
      "loss": 2.1996,
      "step": 155672
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.2619125843048096,
      "learning_rate": 0.00014272419586486068,
      "loss": 2.0779,
      "step": 155673
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.1377642154693604,
      "learning_rate": 0.00014272071250908522,
      "loss": 1.9624,
      "step": 155674
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.166473627090454,
      "learning_rate": 0.00014271722918255108,
      "loss": 1.9252,
      "step": 155675
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.4262062311172485,
      "learning_rate": 0.00014271374588525863,
      "loss": 2.2417,
      "step": 155676
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.3845911026000977,
      "learning_rate": 0.00014271026261720882,
      "loss": 2.0594,
      "step": 155677
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.142473578453064,
      "learning_rate": 0.00014270677937840197,
      "loss": 2.1789,
      "step": 155678
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.3275099992752075,
      "learning_rate": 0.00014270329616883917,
      "loss": 2.038,
      "step": 155679
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.5826282501220703,
      "learning_rate": 0.00014269981298852056,
      "loss": 2.0738,
      "step": 155680
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.119394302368164,
      "learning_rate": 0.00014269632983744717,
      "loss": 1.9588,
      "step": 155681
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.0423667430877686,
      "learning_rate": 0.0001426928467156193,
      "loss": 1.8771,
      "step": 155682
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.144787311553955,
      "learning_rate": 0.000142689363623038,
      "loss": 1.8831,
      "step": 155683
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.3631004095077515,
      "learning_rate": 0.00014268588055970358,
      "loss": 2.1623,
      "step": 155684
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.2536627054214478,
      "learning_rate": 0.00014268239752561704,
      "loss": 1.8958,
      "step": 155685
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.1702849864959717,
      "learning_rate": 0.00014267891452077858,
      "loss": 2.0229,
      "step": 155686
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.0954560041427612,
      "learning_rate": 0.00014267543154518922,
      "loss": 1.9724,
      "step": 155687
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.6229404211044312,
      "learning_rate": 0.00014267194859884935,
      "loss": 2.1618,
      "step": 155688
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.1994017362594604,
      "learning_rate": 0.0001426684656817598,
      "loss": 1.9274,
      "step": 155689
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.3533049821853638,
      "learning_rate": 0.00014266498279392106,
      "loss": 2.0167,
      "step": 155690
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.1690870523452759,
      "learning_rate": 0.00014266149993533413,
      "loss": 2.172,
      "step": 155691
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.2089855670928955,
      "learning_rate": 0.00014265801710599914,
      "loss": 1.9861,
      "step": 155692
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.2472032308578491,
      "learning_rate": 0.00014265453430591714,
      "loss": 2.2541,
      "step": 155693
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.0970696210861206,
      "learning_rate": 0.00014265105153508847,
      "loss": 1.8446,
      "step": 155694
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.284891963005066,
      "learning_rate": 0.0001426475687935141,
      "loss": 2.1795,
      "step": 155695
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.444352388381958,
      "learning_rate": 0.00014264408608119438,
      "loss": 2.1367,
      "step": 155696
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.6872036457061768,
      "learning_rate": 0.00014264060339813032,
      "loss": 2.2249,
      "step": 155697
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.3441685438156128,
      "learning_rate": 0.00014263712074432213,
      "loss": 1.9922,
      "step": 155698
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.1725280284881592,
      "learning_rate": 0.00014263363811977076,
      "loss": 2.1617,
      "step": 155699
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.17466139793396,
      "learning_rate": 0.00014263015552447666,
      "loss": 1.9394,
      "step": 155700
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.3460122346878052,
      "learning_rate": 0.0001426266729584407,
      "loss": 2.1066,
      "step": 155701
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.5241456031799316,
      "learning_rate": 0.00014262319042166333,
      "loss": 2.0933,
      "step": 155702
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.0980855226516724,
      "learning_rate": 0.00014261970791414546,
      "loss": 2.1334,
      "step": 155703
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.1788252592086792,
      "learning_rate": 0.00014261622543588736,
      "loss": 2.1682,
      "step": 155704
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.1476619243621826,
      "learning_rate": 0.00014261274298688997,
      "loss": 2.2066,
      "step": 155705
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.149782419204712,
      "learning_rate": 0.00014260926056715374,
      "loss": 2.048,
      "step": 155706
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.2455122470855713,
      "learning_rate": 0.00014260577817667953,
      "loss": 2.0383,
      "step": 155707
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.2510184049606323,
      "learning_rate": 0.00014260229581546777,
      "loss": 1.9298,
      "step": 155708
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.0806738138198853,
      "learning_rate": 0.00014259881348351944,
      "loss": 1.9399,
      "step": 155709
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.1389236450195312,
      "learning_rate": 0.00014259533118083474,
      "loss": 1.8419,
      "step": 155710
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.314407229423523,
      "learning_rate": 0.0001425918489074146,
      "loss": 1.8896,
      "step": 155711
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.0742970705032349,
      "learning_rate": 0.00014258836666325954,
      "loss": 2.109,
      "step": 155712
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.373651385307312,
      "learning_rate": 0.00014258488444837038,
      "loss": 2.2186,
      "step": 155713
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.211748480796814,
      "learning_rate": 0.00014258140226274755,
      "loss": 2.0539,
      "step": 155714
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.0764738321304321,
      "learning_rate": 0.00014257792010639194,
      "loss": 1.92,
      "step": 155715
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.1904157400131226,
      "learning_rate": 0.00014257443797930405,
      "loss": 2.0433,
      "step": 155716
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.242738127708435,
      "learning_rate": 0.00014257095588148455,
      "loss": 2.2248,
      "step": 155717
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.2106504440307617,
      "learning_rate": 0.00014256747381293395,
      "loss": 1.9722,
      "step": 155718
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.2251524925231934,
      "learning_rate": 0.00014256399177365312,
      "loss": 1.934,
      "step": 155719
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.2461673021316528,
      "learning_rate": 0.00014256050976364258,
      "loss": 1.9838,
      "step": 155720
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.2720199823379517,
      "learning_rate": 0.00014255702778290306,
      "loss": 2.1384,
      "step": 155721
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.113515853881836,
      "learning_rate": 0.0001425535458314352,
      "loss": 2.0177,
      "step": 155722
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.2095407247543335,
      "learning_rate": 0.00014255006390923944,
      "loss": 1.9977,
      "step": 155723
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.1326792240142822,
      "learning_rate": 0.00014254658201631675,
      "loss": 1.8863,
      "step": 155724
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.2231630086898804,
      "learning_rate": 0.00014254310015266762,
      "loss": 2.2323,
      "step": 155725
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.1981269121170044,
      "learning_rate": 0.0001425396183182926,
      "loss": 1.9779,
      "step": 155726
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.135547161102295,
      "learning_rate": 0.00014253613651319253,
      "loss": 2.132,
      "step": 155727
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.202022910118103,
      "learning_rate": 0.00014253265473736796,
      "loss": 1.9948,
      "step": 155728
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.2065198421478271,
      "learning_rate": 0.00014252917299081944,
      "loss": 2.1237,
      "step": 155729
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.2046931982040405,
      "learning_rate": 0.00014252569127354783,
      "loss": 2.0149,
      "step": 155730
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.219977617263794,
      "learning_rate": 0.00014252220958555368,
      "loss": 2.2735,
      "step": 155731
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.2639676332473755,
      "learning_rate": 0.00014251872792683747,
      "loss": 2.1541,
      "step": 155732
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.4105404615402222,
      "learning_rate": 0.00014251524629740016,
      "loss": 2.1314,
      "step": 155733
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.2999897003173828,
      "learning_rate": 0.00014251176469724222,
      "loss": 2.0536,
      "step": 155734
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.182814359664917,
      "learning_rate": 0.0001425082831263642,
      "loss": 2.0583,
      "step": 155735
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.2460192441940308,
      "learning_rate": 0.00014250480158476696,
      "loss": 2.1244,
      "step": 155736
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.2788805961608887,
      "learning_rate": 0.00014250132007245098,
      "loss": 1.9842,
      "step": 155737
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.240860104560852,
      "learning_rate": 0.00014249783858941707,
      "loss": 2.0046,
      "step": 155738
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.5235011577606201,
      "learning_rate": 0.00014249435713566575,
      "loss": 2.0534,
      "step": 155739
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.182458519935608,
      "learning_rate": 0.00014249087571119773,
      "loss": 1.9647,
      "step": 155740
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.272551417350769,
      "learning_rate": 0.00014248739431601355,
      "loss": 2.0129,
      "step": 155741
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.2563725709915161,
      "learning_rate": 0.000142483912950114,
      "loss": 2.1743,
      "step": 155742
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.55473792552948,
      "learning_rate": 0.00014248043161349958,
      "loss": 2.0925,
      "step": 155743
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.26302170753479,
      "learning_rate": 0.0001424769503061711,
      "loss": 2.1833,
      "step": 155744
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.1271700859069824,
      "learning_rate": 0.00014247346902812916,
      "loss": 2.1397,
      "step": 155745
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.2741395235061646,
      "learning_rate": 0.00014246998777937438,
      "loss": 1.8317,
      "step": 155746
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.3494170904159546,
      "learning_rate": 0.00014246650655990728,
      "loss": 2.0579,
      "step": 155747
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.2906527519226074,
      "learning_rate": 0.00014246302536972876,
      "loss": 1.9296,
      "step": 155748
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.0583655834197998,
      "learning_rate": 0.0001424595442088392,
      "loss": 1.8868,
      "step": 155749
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.1008894443511963,
      "learning_rate": 0.00014245606307723948,
      "loss": 2.0438,
      "step": 155750
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.4203510284423828,
      "learning_rate": 0.00014245258197493008,
      "loss": 1.9723,
      "step": 155751
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.1673976182937622,
      "learning_rate": 0.0001424491009019119,
      "loss": 2.2004,
      "step": 155752
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.223818063735962,
      "learning_rate": 0.00014244561985818523,
      "loss": 1.9988,
      "step": 155753
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.088728427886963,
      "learning_rate": 0.000142442138843751,
      "loss": 2.1912,
      "step": 155754
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.2692896127700806,
      "learning_rate": 0.0001424386578586096,
      "loss": 2.0315,
      "step": 155755
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.325492262840271,
      "learning_rate": 0.00014243517690276197,
      "loss": 2.0983,
      "step": 155756
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.401237964630127,
      "learning_rate": 0.0001424316959762085,
      "loss": 1.9797,
      "step": 155757
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.2640442848205566,
      "learning_rate": 0.00014242821507895015,
      "loss": 2.1916,
      "step": 155758
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.1357760429382324,
      "learning_rate": 0.0001424247342109872,
      "loss": 1.9267,
      "step": 155759
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.3129711151123047,
      "learning_rate": 0.00014242125337232052,
      "loss": 1.9844,
      "step": 155760
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.3758232593536377,
      "learning_rate": 0.0001424177725629506,
      "loss": 2.154,
      "step": 155761
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.3824491500854492,
      "learning_rate": 0.00014241429178287834,
      "loss": 2.1556,
      "step": 155762
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.1722397804260254,
      "learning_rate": 0.0001424108110321041,
      "loss": 2.1189,
      "step": 155763
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.1718902587890625,
      "learning_rate": 0.00014240733031062889,
      "loss": 2.3582,
      "step": 155764
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.3262531757354736,
      "learning_rate": 0.00014240384961845291,
      "loss": 2.0676,
      "step": 155765
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.280096173286438,
      "learning_rate": 0.00014240036895557716,
      "loss": 1.7919,
      "step": 155766
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.2167630195617676,
      "learning_rate": 0.000142396888322002,
      "loss": 2.1492,
      "step": 155767
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.1620240211486816,
      "learning_rate": 0.00014239340771772838,
      "loss": 1.9188,
      "step": 155768
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.1327416896820068,
      "learning_rate": 0.0001423899271427567,
      "loss": 1.9753,
      "step": 155769
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.2835323810577393,
      "learning_rate": 0.0001423864465970879,
      "loss": 2.1808,
      "step": 155770
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.1848207712173462,
      "learning_rate": 0.00014238296608072224,
      "loss": 2.1532,
      "step": 155771
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.193430781364441,
      "learning_rate": 0.00014237948559366065,
      "loss": 2.0364,
      "step": 155772
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.3669120073318481,
      "learning_rate": 0.00014237600513590358,
      "loss": 2.0074,
      "step": 155773
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.2010022401809692,
      "learning_rate": 0.0001423725247074519,
      "loss": 2.2489,
      "step": 155774
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.42262864112854,
      "learning_rate": 0.000142369044308306,
      "loss": 2.2052,
      "step": 155775
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.3740397691726685,
      "learning_rate": 0.00014236556393846695,
      "loss": 2.0971,
      "step": 155776
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.6118930578231812,
      "learning_rate": 0.00014236208359793488,
      "loss": 1.7869,
      "step": 155777
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.135482907295227,
      "learning_rate": 0.00014235860328671077,
      "loss": 2.2022,
      "step": 155778
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.1782971620559692,
      "learning_rate": 0.00014235512300479509,
      "loss": 1.8585,
      "step": 155779
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.1893911361694336,
      "learning_rate": 0.00014235164275218864,
      "loss": 2.0138,
      "step": 155780
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.1996649503707886,
      "learning_rate": 0.0001423481625288919,
      "loss": 1.9961,
      "step": 155781
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.3511042594909668,
      "learning_rate": 0.00014234468233490584,
      "loss": 1.8705,
      "step": 155782
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.2524763345718384,
      "learning_rate": 0.00014234120217023063,
      "loss": 1.982,
      "step": 155783
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.0951064825057983,
      "learning_rate": 0.0001423377220348673,
      "loss": 2.0409,
      "step": 155784
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.1968586444854736,
      "learning_rate": 0.00014233424192881626,
      "loss": 2.0474,
      "step": 155785
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.3541992902755737,
      "learning_rate": 0.00014233076185207836,
      "loss": 2.2104,
      "step": 155786
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.2190576791763306,
      "learning_rate": 0.00014232728180465406,
      "loss": 1.979,
      "step": 155787
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.1242059469223022,
      "learning_rate": 0.00014232380178654428,
      "loss": 1.9818,
      "step": 155788
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.2710307836532593,
      "learning_rate": 0.0001423203217977493,
      "loss": 2.0305,
      "step": 155789
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.2680072784423828,
      "learning_rate": 0.00014231684183827004,
      "loss": 2.2651,
      "step": 155790
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.232263445854187,
      "learning_rate": 0.0001423133619081069,
      "loss": 2.1627,
      "step": 155791
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.1853306293487549,
      "learning_rate": 0.00014230988200726084,
      "loss": 1.9467,
      "step": 155792
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.311753511428833,
      "learning_rate": 0.00014230640213573227,
      "loss": 2.1127,
      "step": 155793
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.1497855186462402,
      "learning_rate": 0.00014230292229352207,
      "loss": 2.1997,
      "step": 155794
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.3530629873275757,
      "learning_rate": 0.00014229944248063056,
      "loss": 1.8447,
      "step": 155795
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.2537370920181274,
      "learning_rate": 0.00014229596269705865,
      "loss": 2.1803,
      "step": 155796
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.2372268438339233,
      "learning_rate": 0.00014229248294280678,
      "loss": 1.9476,
      "step": 155797
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.4335815906524658,
      "learning_rate": 0.00014228900321787587,
      "loss": 2.0273,
      "step": 155798
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.1829197406768799,
      "learning_rate": 0.00014228552352226625,
      "loss": 2.2141,
      "step": 155799
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.146070122718811,
      "learning_rate": 0.00014228204385597883,
      "loss": 1.7423,
      "step": 155800
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.194019079208374,
      "learning_rate": 0.0001422785642190142,
      "loss": 2.1232,
      "step": 155801
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.2625527381896973,
      "learning_rate": 0.00014227508461137295,
      "loss": 1.9901,
      "step": 155802
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.1504350900650024,
      "learning_rate": 0.0001422716050330556,
      "loss": 2.0436,
      "step": 155803
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.1262803077697754,
      "learning_rate": 0.0001422681254840631,
      "loss": 2.1573,
      "step": 155804
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.1099427938461304,
      "learning_rate": 0.00014226464596439576,
      "loss": 1.9997,
      "step": 155805
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.1840529441833496,
      "learning_rate": 0.00014226116647405457,
      "loss": 1.8949,
      "step": 155806
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.0959539413452148,
      "learning_rate": 0.00014225768701303998,
      "loss": 1.9794,
      "step": 155807
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.1051846742630005,
      "learning_rate": 0.00014225420758135263,
      "loss": 2.0046,
      "step": 155808
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.1272046566009521,
      "learning_rate": 0.00014225072817899314,
      "loss": 2.161,
      "step": 155809
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.1007916927337646,
      "learning_rate": 0.00014224724880596232,
      "loss": 1.8804,
      "step": 155810
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.404526710510254,
      "learning_rate": 0.00014224376946226058,
      "loss": 1.8691,
      "step": 155811
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.261332631111145,
      "learning_rate": 0.00014224029014788884,
      "loss": 2.3367,
      "step": 155812
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.1752066612243652,
      "learning_rate": 0.00014223681086284758,
      "loss": 2.278,
      "step": 155813
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.2172249555587769,
      "learning_rate": 0.00014223333160713738,
      "loss": 2.0154,
      "step": 155814
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.2309726476669312,
      "learning_rate": 0.00014222985238075908,
      "loss": 2.0235,
      "step": 155815
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.3217023611068726,
      "learning_rate": 0.00014222637318371325,
      "loss": 1.9121,
      "step": 155816
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.335385799407959,
      "learning_rate": 0.0001422228940160004,
      "loss": 1.995,
      "step": 155817
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.1769628524780273,
      "learning_rate": 0.0001422194148776214,
      "loss": 1.9984,
      "step": 155818
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.1293524503707886,
      "learning_rate": 0.00014221593576857678,
      "loss": 2.0594,
      "step": 155819
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.3219534158706665,
      "learning_rate": 0.00014221245668886708,
      "loss": 2.0107,
      "step": 155820
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.1781507730484009,
      "learning_rate": 0.0001422089776384932,
      "loss": 2.3108,
      "step": 155821
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.2060627937316895,
      "learning_rate": 0.0001422054986174555,
      "loss": 1.8749,
      "step": 155822
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.1489700078964233,
      "learning_rate": 0.00014220201962575492,
      "loss": 2.0944,
      "step": 155823
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.1659648418426514,
      "learning_rate": 0.00014219854066339197,
      "loss": 2.0095,
      "step": 155824
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.1508125066757202,
      "learning_rate": 0.00014219506173036725,
      "loss": 2.1213,
      "step": 155825
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.1679953336715698,
      "learning_rate": 0.00014219158282668137,
      "loss": 2.0187,
      "step": 155826
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.1028577089309692,
      "learning_rate": 0.00014218810395233514,
      "loss": 1.829,
      "step": 155827
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.0282337665557861,
      "learning_rate": 0.000142184625107329,
      "loss": 2.0557,
      "step": 155828
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.2249819040298462,
      "learning_rate": 0.00014218114629166388,
      "loss": 1.8561,
      "step": 155829
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.2241771221160889,
      "learning_rate": 0.00014217766750534023,
      "loss": 2.1182,
      "step": 155830
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.1067547798156738,
      "learning_rate": 0.00014217418874835875,
      "loss": 2.1631,
      "step": 155831
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.2566940784454346,
      "learning_rate": 0.00014217071002071993,
      "loss": 1.9733,
      "step": 155832
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.1893013715744019,
      "learning_rate": 0.00014216723132242469,
      "loss": 1.9877,
      "step": 155833
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.291077733039856,
      "learning_rate": 0.00014216375265347343,
      "loss": 1.9077,
      "step": 155834
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.1104071140289307,
      "learning_rate": 0.00014216027401386697,
      "loss": 1.9413,
      "step": 155835
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.2815473079681396,
      "learning_rate": 0.00014215679540360583,
      "loss": 1.9012,
      "step": 155836
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.3813495635986328,
      "learning_rate": 0.00014215331682269092,
      "loss": 1.9425,
      "step": 155837
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.091362476348877,
      "learning_rate": 0.0001421498382711225,
      "loss": 1.9536,
      "step": 155838
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.108415961265564,
      "learning_rate": 0.0001421463597489015,
      "loss": 2.0608,
      "step": 155839
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.344765067100525,
      "learning_rate": 0.00014214288125602835,
      "loss": 1.9941,
      "step": 155840
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.1862945556640625,
      "learning_rate": 0.00014213940279250394,
      "loss": 2.0908,
      "step": 155841
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.2518048286437988,
      "learning_rate": 0.00014213592435832867,
      "loss": 2.0647,
      "step": 155842
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.1493103504180908,
      "learning_rate": 0.00014213244595350354,
      "loss": 2.0388,
      "step": 155843
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.2790470123291016,
      "learning_rate": 0.00014212896757802873,
      "loss": 2.0388,
      "step": 155844
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.1530588865280151,
      "learning_rate": 0.00014212548923190528,
      "loss": 2.1411,
      "step": 155845
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.1661100387573242,
      "learning_rate": 0.00014212201091513352,
      "loss": 2.2245,
      "step": 155846
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.093324065208435,
      "learning_rate": 0.00014211853262771442,
      "loss": 2.0305,
      "step": 155847
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.0696790218353271,
      "learning_rate": 0.0001421150543696483,
      "loss": 2.0308,
      "step": 155848
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.2014259099960327,
      "learning_rate": 0.00014211157614093625,
      "loss": 1.9808,
      "step": 155849
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.1361377239227295,
      "learning_rate": 0.00014210809794157838,
      "loss": 1.8718,
      "step": 155850
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.2917399406433105,
      "learning_rate": 0.00014210461977157572,
      "loss": 2.0808,
      "step": 155851
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.1279977560043335,
      "learning_rate": 0.0001421011416309287,
      "loss": 2.1507,
      "step": 155852
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.2424778938293457,
      "learning_rate": 0.0001420976635196381,
      "loss": 2.0434,
      "step": 155853
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.1010687351226807,
      "learning_rate": 0.00014209418543770447,
      "loss": 2.0633,
      "step": 155854
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.1571671962738037,
      "learning_rate": 0.00014209070738512873,
      "loss": 2.1683,
      "step": 155855
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.3393336534500122,
      "learning_rate": 0.00014208722936191108,
      "loss": 2.1474,
      "step": 155856
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.2443287372589111,
      "learning_rate": 0.00014208375136805248,
      "loss": 2.2725,
      "step": 155857
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.102349042892456,
      "learning_rate": 0.00014208027340355343,
      "loss": 2.1378,
      "step": 155858
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.3279231786727905,
      "learning_rate": 0.0001420767954684147,
      "loss": 1.8309,
      "step": 155859
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.730883240699768,
      "learning_rate": 0.0001420733175626368,
      "loss": 1.943,
      "step": 155860
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.136255145072937,
      "learning_rate": 0.00014206983968622069,
      "loss": 2.2274,
      "step": 155861
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.2110320329666138,
      "learning_rate": 0.0001420663618391665,
      "loss": 2.0078,
      "step": 155862
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.2468760013580322,
      "learning_rate": 0.0001420628840214753,
      "loss": 2.0955,
      "step": 155863
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.225590467453003,
      "learning_rate": 0.00014205940623314748,
      "loss": 2.1225,
      "step": 155864
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.169309139251709,
      "learning_rate": 0.0001420559284741839,
      "loss": 1.9964,
      "step": 155865
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.1192190647125244,
      "learning_rate": 0.00014205245074458495,
      "loss": 1.9979,
      "step": 155866
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.2377806901931763,
      "learning_rate": 0.00014204897304435175,
      "loss": 2.082,
      "step": 155867
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.1474429368972778,
      "learning_rate": 0.00014204549537348432,
      "loss": 2.0034,
      "step": 155868
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.0888190269470215,
      "learning_rate": 0.00014204201773198376,
      "loss": 1.9441,
      "step": 155869
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.1271438598632812,
      "learning_rate": 0.00014203854011985047,
      "loss": 1.9425,
      "step": 155870
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.0993198156356812,
      "learning_rate": 0.00014203506253708528,
      "loss": 2.2375,
      "step": 155871
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.221711277961731,
      "learning_rate": 0.00014203158498368867,
      "loss": 1.949,
      "step": 155872
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.1503113508224487,
      "learning_rate": 0.00014202810745966147,
      "loss": 2.0658,
      "step": 155873
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.2694388628005981,
      "learning_rate": 0.00014202462996500421,
      "loss": 2.0213,
      "step": 155874
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.1927804946899414,
      "learning_rate": 0.0001420211524997176,
      "loss": 1.9742,
      "step": 155875
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.1275019645690918,
      "learning_rate": 0.00014201767506380206,
      "loss": 2.0156,
      "step": 155876
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.1477693319320679,
      "learning_rate": 0.0001420141976572586,
      "loss": 2.3093,
      "step": 155877
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.1025906801223755,
      "learning_rate": 0.00014201072028008753,
      "loss": 2.0299,
      "step": 155878
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.0843267440795898,
      "learning_rate": 0.00014200724293228977,
      "loss": 1.8966,
      "step": 155879
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.2308831214904785,
      "learning_rate": 0.00014200376561386582,
      "loss": 2.0515,
      "step": 155880
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.2030062675476074,
      "learning_rate": 0.0001420002883248164,
      "loss": 2.0464,
      "step": 155881
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.1443885564804077,
      "learning_rate": 0.00014199681106514197,
      "loss": 1.8191,
      "step": 155882
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.036957859992981,
      "learning_rate": 0.00014199333383484343,
      "loss": 2.0046,
      "step": 155883
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.1919373273849487,
      "learning_rate": 0.00014198985663392119,
      "loss": 2.1585,
      "step": 155884
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.2874666452407837,
      "learning_rate": 0.00014198637946237616,
      "loss": 1.9649,
      "step": 155885
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.1595475673675537,
      "learning_rate": 0.00014198290232020884,
      "loss": 1.9753,
      "step": 155886
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.1048941612243652,
      "learning_rate": 0.00014197942520741984,
      "loss": 1.9357,
      "step": 155887
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.13511323928833,
      "learning_rate": 0.00014197594812400977,
      "loss": 1.8757,
      "step": 155888
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.1784604787826538,
      "learning_rate": 0.00014197247106997944,
      "loss": 2.1037,
      "step": 155889
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.0741775035858154,
      "learning_rate": 0.00014196899404532933,
      "loss": 2.1907,
      "step": 155890
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.1901601552963257,
      "learning_rate": 0.00014196551705006027,
      "loss": 1.9625,
      "step": 155891
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.1473561525344849,
      "learning_rate": 0.0001419620400841728,
      "loss": 2.0124,
      "step": 155892
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.1808139085769653,
      "learning_rate": 0.00014195856314766755,
      "loss": 2.1861,
      "step": 155893
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.3147274255752563,
      "learning_rate": 0.00014195508624054512,
      "loss": 1.9306,
      "step": 155894
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.2622489929199219,
      "learning_rate": 0.0001419516093628063,
      "loss": 2.2094,
      "step": 155895
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.1541388034820557,
      "learning_rate": 0.00014194813251445156,
      "loss": 2.1785,
      "step": 155896
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.1629873514175415,
      "learning_rate": 0.00014194465569548174,
      "loss": 1.8703,
      "step": 155897
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.1628473997116089,
      "learning_rate": 0.00014194117890589742,
      "loss": 2.0335,
      "step": 155898
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.1565372943878174,
      "learning_rate": 0.0001419377021456991,
      "loss": 1.9442,
      "step": 155899
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.2454442977905273,
      "learning_rate": 0.00014193422541488766,
      "loss": 1.9957,
      "step": 155900
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.19973623752594,
      "learning_rate": 0.00014193074871346361,
      "loss": 2.1509,
      "step": 155901
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.3178935050964355,
      "learning_rate": 0.0001419272720414275,
      "loss": 2.0175,
      "step": 155902
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.3437529802322388,
      "learning_rate": 0.00014192379539878025,
      "loss": 2.3171,
      "step": 155903
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.345750331878662,
      "learning_rate": 0.00014192031878552234,
      "loss": 1.8933,
      "step": 155904
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.2021948099136353,
      "learning_rate": 0.00014191684220165427,
      "loss": 2.2574,
      "step": 155905
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.2281553745269775,
      "learning_rate": 0.00014191336564717697,
      "loss": 1.8345,
      "step": 155906
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.189418077468872,
      "learning_rate": 0.000141909889122091,
      "loss": 1.9383,
      "step": 155907
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.1767754554748535,
      "learning_rate": 0.0001419064126263968,
      "loss": 1.8316,
      "step": 155908
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.1042345762252808,
      "learning_rate": 0.00014190293616009534,
      "loss": 2.2795,
      "step": 155909
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.1949042081832886,
      "learning_rate": 0.00014189945972318708,
      "loss": 2.1883,
      "step": 155910
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.2230521440505981,
      "learning_rate": 0.0001418959833156726,
      "loss": 1.955,
      "step": 155911
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.1719311475753784,
      "learning_rate": 0.00014189250693755274,
      "loss": 1.9753,
      "step": 155912
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.360542893409729,
      "learning_rate": 0.00014188903058882793,
      "loss": 2.0993,
      "step": 155913
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.1638038158416748,
      "learning_rate": 0.00014188555426949905,
      "loss": 1.866,
      "step": 155914
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.331143856048584,
      "learning_rate": 0.00014188207797956665,
      "loss": 2.1328,
      "step": 155915
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.1740169525146484,
      "learning_rate": 0.00014187860171903133,
      "loss": 1.9136,
      "step": 155916
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.1748408079147339,
      "learning_rate": 0.00014187512548789367,
      "loss": 2.2248,
      "step": 155917
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.1761480569839478,
      "learning_rate": 0.0001418716492861545,
      "loss": 2.0117,
      "step": 155918
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.3576807975769043,
      "learning_rate": 0.0001418681731138143,
      "loss": 2.1188,
      "step": 155919
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.3208236694335938,
      "learning_rate": 0.00014186469697087385,
      "loss": 1.7561,
      "step": 155920
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.1587638854980469,
      "learning_rate": 0.00014186122085733369,
      "loss": 2.0892,
      "step": 155921
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.1284223794937134,
      "learning_rate": 0.00014185774477319468,
      "loss": 1.9116,
      "step": 155922
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.2251726388931274,
      "learning_rate": 0.0001418542687184571,
      "loss": 2.2763,
      "step": 155923
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.2814397811889648,
      "learning_rate": 0.00014185079269312192,
      "loss": 2.0505,
      "step": 155924
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.1391719579696655,
      "learning_rate": 0.00014184731669718954,
      "loss": 1.9438,
      "step": 155925
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.2767583131790161,
      "learning_rate": 0.00014184384073066082,
      "loss": 1.987,
      "step": 155926
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.2878718376159668,
      "learning_rate": 0.00014184036479353626,
      "loss": 2.0713,
      "step": 155927
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.1351672410964966,
      "learning_rate": 0.00014183688888581673,
      "loss": 1.8957,
      "step": 155928
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.3435969352722168,
      "learning_rate": 0.0001418334130075025,
      "loss": 1.9735,
      "step": 155929
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.1913396120071411,
      "learning_rate": 0.00014182993715859454,
      "loss": 1.9222,
      "step": 155930
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.340600848197937,
      "learning_rate": 0.0001418264613390933,
      "loss": 1.9514,
      "step": 155931
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.1205207109451294,
      "learning_rate": 0.00014182298554899959,
      "loss": 2.137,
      "step": 155932
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.2200533151626587,
      "learning_rate": 0.00014181950978831384,
      "loss": 2.0689,
      "step": 155933
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.122502326965332,
      "learning_rate": 0.0001418160340570371,
      "loss": 2.048,
      "step": 155934
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.2183457612991333,
      "learning_rate": 0.00014181255835516948,
      "loss": 2.1341,
      "step": 155935
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.1541334390640259,
      "learning_rate": 0.000141809082682712,
      "loss": 1.8456,
      "step": 155936
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.201593279838562,
      "learning_rate": 0.0001418056070396651,
      "loss": 2.1323,
      "step": 155937
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.0700416564941406,
      "learning_rate": 0.00014180213142602963,
      "loss": 2.0043,
      "step": 155938
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.092239260673523,
      "learning_rate": 0.00014179865584180605,
      "loss": 1.9732,
      "step": 155939
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.0947211980819702,
      "learning_rate": 0.00014179518028699517,
      "loss": 1.8036,
      "step": 155940
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.062272548675537,
      "learning_rate": 0.00014179170476159755,
      "loss": 1.8779,
      "step": 155941
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.21086585521698,
      "learning_rate": 0.00014178822926561385,
      "loss": 1.9186,
      "step": 155942
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.323593258857727,
      "learning_rate": 0.00014178475379904458,
      "loss": 2.1548,
      "step": 155943
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.3283841609954834,
      "learning_rate": 0.00014178127836189064,
      "loss": 2.0212,
      "step": 155944
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.1497316360473633,
      "learning_rate": 0.0001417778029541524,
      "loss": 2.0527,
      "step": 155945
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.196974515914917,
      "learning_rate": 0.0001417743275758308,
      "loss": 2.0335,
      "step": 155946
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.1686838865280151,
      "learning_rate": 0.00014177085222692636,
      "loss": 1.9676,
      "step": 155947
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.280230164527893,
      "learning_rate": 0.00014176737690743967,
      "loss": 2.0981,
      "step": 155948
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.1870659589767456,
      "learning_rate": 0.00014176390161737128,
      "loss": 2.1756,
      "step": 155949
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.2416508197784424,
      "learning_rate": 0.00014176042635672212,
      "loss": 1.8682,
      "step": 155950
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.5819789171218872,
      "learning_rate": 0.00014175695112549254,
      "loss": 2.1629,
      "step": 155951
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.1328070163726807,
      "learning_rate": 0.0001417534759236835,
      "loss": 1.7399,
      "step": 155952
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.1031811237335205,
      "learning_rate": 0.00014175000075129538,
      "loss": 2.0039,
      "step": 155953
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.5998951196670532,
      "learning_rate": 0.000141746525608329,
      "loss": 2.0281,
      "step": 155954
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.1250921487808228,
      "learning_rate": 0.00014174305049478481,
      "loss": 2.1721,
      "step": 155955
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.1951589584350586,
      "learning_rate": 0.00014173957541066364,
      "loss": 2.1416,
      "step": 155956
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.2620364427566528,
      "learning_rate": 0.000141736100355966,
      "loss": 1.9707,
      "step": 155957
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.1405431032180786,
      "learning_rate": 0.0001417326253306927,
      "loss": 2.4578,
      "step": 155958
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.1762295961380005,
      "learning_rate": 0.0001417291503348443,
      "loss": 2.0986,
      "step": 155959
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.1613889932632446,
      "learning_rate": 0.00014172567536842141,
      "loss": 2.2395,
      "step": 155960
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.366094708442688,
      "learning_rate": 0.00014172220043142463,
      "loss": 2.1549,
      "step": 155961
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.125584363937378,
      "learning_rate": 0.0001417187255238548,
      "loss": 2.0382,
      "step": 155962
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.4587332010269165,
      "learning_rate": 0.00014171525064571232,
      "loss": 2.0906,
      "step": 155963
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.1926268339157104,
      "learning_rate": 0.00014171177579699805,
      "loss": 1.8887,
      "step": 155964
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.1381720304489136,
      "learning_rate": 0.00014170830097771258,
      "loss": 1.921,
      "step": 155965
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.2333892583847046,
      "learning_rate": 0.00014170482618785653,
      "loss": 2.0515,
      "step": 155966
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.165364146232605,
      "learning_rate": 0.0001417013514274304,
      "loss": 2.1876,
      "step": 155967
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.1257543563842773,
      "learning_rate": 0.0001416978766964351,
      "loss": 2.0135,
      "step": 155968
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.5642377138137817,
      "learning_rate": 0.00014169440199487105,
      "loss": 1.7696,
      "step": 155969
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.1369221210479736,
      "learning_rate": 0.0001416909273227391,
      "loss": 2.1657,
      "step": 155970
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.051979422569275,
      "learning_rate": 0.00014168745268003982,
      "loss": 1.869,
      "step": 155971
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.132441759109497,
      "learning_rate": 0.0001416839780667738,
      "loss": 1.8734,
      "step": 155972
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.5470396280288696,
      "learning_rate": 0.0001416805034829416,
      "loss": 2.0225,
      "step": 155973
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.1858891248703003,
      "learning_rate": 0.00014167702892854412,
      "loss": 1.9292,
      "step": 155974
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.121041178703308,
      "learning_rate": 0.00014167355440358173,
      "loss": 1.9341,
      "step": 155975
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.1393322944641113,
      "learning_rate": 0.00014167007990805534,
      "loss": 2.019,
      "step": 155976
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.1472779512405396,
      "learning_rate": 0.00014166660544196548,
      "loss": 2.1093,
      "step": 155977
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.301074743270874,
      "learning_rate": 0.0001416631310053128,
      "loss": 2.1537,
      "step": 155978
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.1200156211853027,
      "learning_rate": 0.00014165965659809779,
      "loss": 1.9028,
      "step": 155979
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.2261910438537598,
      "learning_rate": 0.0001416561822203214,
      "loss": 1.9828,
      "step": 155980
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.24386727809906,
      "learning_rate": 0.0001416527078719839,
      "loss": 2.1472,
      "step": 155981
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.3891311883926392,
      "learning_rate": 0.00014164923355308636,
      "loss": 2.08,
      "step": 155982
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.4615792036056519,
      "learning_rate": 0.00014164575926362919,
      "loss": 2.0082,
      "step": 155983
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.0928932428359985,
      "learning_rate": 0.00014164228500361292,
      "loss": 2.0013,
      "step": 155984
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.1480350494384766,
      "learning_rate": 0.0001416388107730385,
      "loss": 2.2497,
      "step": 155985
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.2005585432052612,
      "learning_rate": 0.0001416353365719064,
      "loss": 2.0837,
      "step": 155986
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.3785650730133057,
      "learning_rate": 0.00014163186240021716,
      "loss": 2.4011,
      "step": 155987
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.2298420667648315,
      "learning_rate": 0.00014162838825797162,
      "loss": 2.2077,
      "step": 155988
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.106806993484497,
      "learning_rate": 0.0001416249141451704,
      "loss": 2.1563,
      "step": 155989
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.1551731824874878,
      "learning_rate": 0.00014162144006181398,
      "loss": 2.0899,
      "step": 155990
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.2050448656082153,
      "learning_rate": 0.00014161796600790322,
      "loss": 2.1692,
      "step": 155991
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.2658092975616455,
      "learning_rate": 0.0001416144919834387,
      "loss": 2.1022,
      "step": 155992
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.148303747177124,
      "learning_rate": 0.0001416110179884209,
      "loss": 2.0717,
      "step": 155993
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.0785349607467651,
      "learning_rate": 0.00014160754402285072,
      "loss": 2.0481,
      "step": 155994
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.1260366439819336,
      "learning_rate": 0.0001416040700867287,
      "loss": 1.8263,
      "step": 155995
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.0175338983535767,
      "learning_rate": 0.00014160059618005533,
      "loss": 1.9664,
      "step": 155996
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.0209497213363647,
      "learning_rate": 0.00014159712230283152,
      "loss": 2.0347,
      "step": 155997
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.188046932220459,
      "learning_rate": 0.0001415936484550577,
      "loss": 2.1394,
      "step": 155998
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.1478989124298096,
      "learning_rate": 0.0001415901746367347,
      "loss": 2.1131,
      "step": 155999
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.309402585029602,
      "learning_rate": 0.0001415867008478631,
      "loss": 2.283,
      "step": 156000
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.267439365386963,
      "learning_rate": 0.0001415832270884435,
      "loss": 1.9688,
      "step": 156001
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.1473548412322998,
      "learning_rate": 0.00014157975335847648,
      "loss": 2.0225,
      "step": 156002
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.1984933614730835,
      "learning_rate": 0.00014157627965796289,
      "loss": 2.1155,
      "step": 156003
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.2910090684890747,
      "learning_rate": 0.0001415728059869031,
      "loss": 2.1011,
      "step": 156004
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.1810630559921265,
      "learning_rate": 0.0001415693323452981,
      "loss": 2.1732,
      "step": 156005
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.2432600259780884,
      "learning_rate": 0.0001415658587331483,
      "loss": 2.1804,
      "step": 156006
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.0686713457107544,
      "learning_rate": 0.0001415623851504543,
      "loss": 1.9657,
      "step": 156007
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.2373661994934082,
      "learning_rate": 0.00014155891159721695,
      "loss": 2.2195,
      "step": 156008
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.2671891450881958,
      "learning_rate": 0.0001415554380734368,
      "loss": 1.9852,
      "step": 156009
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.309977650642395,
      "learning_rate": 0.00014155196457911435,
      "loss": 2.1107,
      "step": 156010
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.235498309135437,
      "learning_rate": 0.0001415484911142505,
      "loss": 1.9494,
      "step": 156011
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.1730550527572632,
      "learning_rate": 0.00014154501767884567,
      "loss": 1.9635,
      "step": 156012
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.129149079322815,
      "learning_rate": 0.00014154154427290075,
      "loss": 1.8667,
      "step": 156013
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.1920907497406006,
      "learning_rate": 0.0001415380708964162,
      "loss": 2.0353,
      "step": 156014
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.2911144495010376,
      "learning_rate": 0.00014153459754939274,
      "loss": 2.0331,
      "step": 156015
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.2378840446472168,
      "learning_rate": 0.00014153112423183087,
      "loss": 2.1592,
      "step": 156016
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.1673541069030762,
      "learning_rate": 0.0001415276509437315,
      "loss": 2.0618,
      "step": 156017
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.130959391593933,
      "learning_rate": 0.00014152417768509499,
      "loss": 2.2542,
      "step": 156018
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.1073788404464722,
      "learning_rate": 0.00014152070445592224,
      "loss": 2.0872,
      "step": 156019
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.1277680397033691,
      "learning_rate": 0.00014151723125621377,
      "loss": 2.1046,
      "step": 156020
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.2562917470932007,
      "learning_rate": 0.00014151375808597029,
      "loss": 1.8258,
      "step": 156021
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.2838014364242554,
      "learning_rate": 0.00014151028494519224,
      "loss": 1.9286,
      "step": 156022
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.1222909688949585,
      "learning_rate": 0.0001415068118338805,
      "loss": 1.8167,
      "step": 156023
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.1694270372390747,
      "learning_rate": 0.0001415033387520356,
      "loss": 2.0528,
      "step": 156024
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.148193597793579,
      "learning_rate": 0.0001414998656996583,
      "loss": 2.0006,
      "step": 156025
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.193167805671692,
      "learning_rate": 0.00014149639267674915,
      "loss": 1.8374,
      "step": 156026
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.2211315631866455,
      "learning_rate": 0.00014149291968330882,
      "loss": 2.1479,
      "step": 156027
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.2094025611877441,
      "learning_rate": 0.00014148944671933783,
      "loss": 2.1216,
      "step": 156028
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.2240797281265259,
      "learning_rate": 0.00014148597378483706,
      "loss": 2.1255,
      "step": 156029
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.1476409435272217,
      "learning_rate": 0.00014148250087980693,
      "loss": 1.9712,
      "step": 156030
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.1584370136260986,
      "learning_rate": 0.00014147902800424835,
      "loss": 2.3739,
      "step": 156031
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.1248854398727417,
      "learning_rate": 0.00014147555515816175,
      "loss": 1.951,
      "step": 156032
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.241492748260498,
      "learning_rate": 0.00014147208234154785,
      "loss": 2.1049,
      "step": 156033
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.184123158454895,
      "learning_rate": 0.0001414686095544072,
      "loss": 2.2923,
      "step": 156034
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.2365219593048096,
      "learning_rate": 0.0001414651367967406,
      "loss": 2.1382,
      "step": 156035
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.12613844871521,
      "learning_rate": 0.00014146166406854853,
      "loss": 1.9281,
      "step": 156036
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.4899640083312988,
      "learning_rate": 0.00014145819136983183,
      "loss": 1.9754,
      "step": 156037
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.2515770196914673,
      "learning_rate": 0.00014145471870059107,
      "loss": 1.9287,
      "step": 156038
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.4862152338027954,
      "learning_rate": 0.00014145124606082686,
      "loss": 2.0374,
      "step": 156039
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.1816027164459229,
      "learning_rate": 0.00014144777345053973,
      "loss": 2.138,
      "step": 156040
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.2561708688735962,
      "learning_rate": 0.0001414443008697306,
      "loss": 2.1017,
      "step": 156041
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.2566308975219727,
      "learning_rate": 0.0001414408283183998,
      "loss": 1.9822,
      "step": 156042
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.198156476020813,
      "learning_rate": 0.0001414373557965483,
      "loss": 1.8067,
      "step": 156043
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.3154431581497192,
      "learning_rate": 0.0001414338833041766,
      "loss": 2.1841,
      "step": 156044
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.4446725845336914,
      "learning_rate": 0.0001414304108412853,
      "loss": 2.1174,
      "step": 156045
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.2332489490509033,
      "learning_rate": 0.00014142693840787495,
      "loss": 1.9816,
      "step": 156046
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.2677379846572876,
      "learning_rate": 0.00014142346600394648,
      "loss": 2.1179,
      "step": 156047
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.3059608936309814,
      "learning_rate": 0.00014141999362950023,
      "loss": 2.1026,
      "step": 156048
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.248673677444458,
      "learning_rate": 0.00014141652128453713,
      "loss": 2.0028,
      "step": 156049
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.225616216659546,
      "learning_rate": 0.0001414130489690577,
      "loss": 2.0282,
      "step": 156050
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.2029482126235962,
      "learning_rate": 0.0001414095766830626,
      "loss": 2.1403,
      "step": 156051
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.1958402395248413,
      "learning_rate": 0.0001414061044265523,
      "loss": 1.9952,
      "step": 156052
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.252134084701538,
      "learning_rate": 0.0001414026321995277,
      "loss": 2.1781,
      "step": 156053
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.221954107284546,
      "learning_rate": 0.00014139916000198928,
      "loss": 2.0877,
      "step": 156054
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.2730417251586914,
      "learning_rate": 0.00014139568783393783,
      "loss": 2.0772,
      "step": 156055
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.462729811668396,
      "learning_rate": 0.00014139221569537392,
      "loss": 2.0465,
      "step": 156056
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.2582143545150757,
      "learning_rate": 0.0001413887435862982,
      "loss": 2.037,
      "step": 156057
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.228635549545288,
      "learning_rate": 0.00014138527150671119,
      "loss": 2.3333,
      "step": 156058
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.163598895072937,
      "learning_rate": 0.00014138179945661375,
      "loss": 2.1635,
      "step": 156059
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.1507107019424438,
      "learning_rate": 0.00014137832743600634,
      "loss": 1.9516,
      "step": 156060
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.3264490365982056,
      "learning_rate": 0.0001413748554448898,
      "loss": 2.1915,
      "step": 156061
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.3061587810516357,
      "learning_rate": 0.00014137138348326465,
      "loss": 2.0359,
      "step": 156062
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.1773595809936523,
      "learning_rate": 0.00014136791155113155,
      "loss": 2.0023,
      "step": 156063
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.2179604768753052,
      "learning_rate": 0.00014136443964849109,
      "loss": 2.2535,
      "step": 156064
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.1279500722885132,
      "learning_rate": 0.00014136096777534404,
      "loss": 2.0212,
      "step": 156065
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.1712311506271362,
      "learning_rate": 0.00014135749593169086,
      "loss": 2.1732,
      "step": 156066
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.1135345697402954,
      "learning_rate": 0.00014135402411753246,
      "loss": 1.9789,
      "step": 156067
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.30255126953125,
      "learning_rate": 0.00014135055233286935,
      "loss": 2.1196,
      "step": 156068
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.1289310455322266,
      "learning_rate": 0.00014134708057770214,
      "loss": 2.2511,
      "step": 156069
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.0683900117874146,
      "learning_rate": 0.0001413436088520314,
      "loss": 1.9439,
      "step": 156070
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.2872822284698486,
      "learning_rate": 0.00014134013715585799,
      "loss": 2.0712,
      "step": 156071
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.193196177482605,
      "learning_rate": 0.0001413366654891823,
      "loss": 1.8947,
      "step": 156072
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.1147186756134033,
      "learning_rate": 0.00014133319385200527,
      "loss": 2.072,
      "step": 156073
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.2323052883148193,
      "learning_rate": 0.0001413297222443273,
      "loss": 2.0262,
      "step": 156074
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.0946365594863892,
      "learning_rate": 0.00014132625066614918,
      "loss": 2.0901,
      "step": 156075
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.2587586641311646,
      "learning_rate": 0.00014132277911747156,
      "loss": 2.12,
      "step": 156076
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.1787687540054321,
      "learning_rate": 0.000141319307598295,
      "loss": 1.8405,
      "step": 156077
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.0899009704589844,
      "learning_rate": 0.0001413158361086201,
      "loss": 1.9588,
      "step": 156078
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.2313951253890991,
      "learning_rate": 0.00014131236464844763,
      "loss": 2.0172,
      "step": 156079
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.1743595600128174,
      "learning_rate": 0.00014130889321777815,
      "loss": 1.881,
      "step": 156080
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.1620887517929077,
      "learning_rate": 0.0001413054218166124,
      "loss": 2.1668,
      "step": 156081
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.2499439716339111,
      "learning_rate": 0.00014130195044495097,
      "loss": 2.0393,
      "step": 156082
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.1439188718795776,
      "learning_rate": 0.0001412984791027944,
      "loss": 1.8908,
      "step": 156083
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.1795176267623901,
      "learning_rate": 0.00014129500779014358,
      "loss": 2.0745,
      "step": 156084
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.2140839099884033,
      "learning_rate": 0.00014129153650699897,
      "loss": 2.189,
      "step": 156085
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.3232735395431519,
      "learning_rate": 0.0001412880652533612,
      "loss": 1.9244,
      "step": 156086
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.367598533630371,
      "learning_rate": 0.000141284594029231,
      "loss": 2.1003,
      "step": 156087
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.1987687349319458,
      "learning_rate": 0.00014128112283460907,
      "loss": 2.0581,
      "step": 156088
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.2626841068267822,
      "learning_rate": 0.00014127765166949583,
      "loss": 2.2977,
      "step": 156089
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.3959405422210693,
      "learning_rate": 0.00014127418053389221,
      "loss": 2.0092,
      "step": 156090
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.2673038244247437,
      "learning_rate": 0.00014127070942779869,
      "loss": 2.0704,
      "step": 156091
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.1748967170715332,
      "learning_rate": 0.00014126723835121584,
      "loss": 2.0214,
      "step": 156092
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.1538805961608887,
      "learning_rate": 0.0001412637673041445,
      "loss": 2.1009,
      "step": 156093
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.2794586420059204,
      "learning_rate": 0.00014126029628658526,
      "loss": 1.7702,
      "step": 156094
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.4890737533569336,
      "learning_rate": 0.00014125682529853858,
      "loss": 2.0386,
      "step": 156095
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.2180883884429932,
      "learning_rate": 0.00014125335434000537,
      "loss": 2.1139,
      "step": 156096
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.2226663827896118,
      "learning_rate": 0.00014124988341098605,
      "loss": 1.9393,
      "step": 156097
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.194439172744751,
      "learning_rate": 0.00014124641251148147,
      "loss": 1.9957,
      "step": 156098
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.2306604385375977,
      "learning_rate": 0.0001412429416414922,
      "loss": 2.1117,
      "step": 156099
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.2883127927780151,
      "learning_rate": 0.00014123947080101884,
      "loss": 2.1594,
      "step": 156100
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.2263673543930054,
      "learning_rate": 0.00014123599999006197,
      "loss": 2.1339,
      "step": 156101
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.3494110107421875,
      "learning_rate": 0.00014123252920862246,
      "loss": 2.067,
      "step": 156102
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.1159718036651611,
      "learning_rate": 0.00014122905845670067,
      "loss": 2.18,
      "step": 156103
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.2351984977722168,
      "learning_rate": 0.0001412255877342975,
      "loss": 2.1465,
      "step": 156104
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.1122727394104004,
      "learning_rate": 0.00014122211704141352,
      "loss": 2.1727,
      "step": 156105
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.22121000289917,
      "learning_rate": 0.00014121864637804934,
      "loss": 2.1031,
      "step": 156106
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.171355128288269,
      "learning_rate": 0.00014121517574420552,
      "loss": 2.2596,
      "step": 156107
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.0740628242492676,
      "learning_rate": 0.00014121170513988284,
      "loss": 2.1214,
      "step": 156108
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.3511167764663696,
      "learning_rate": 0.00014120823456508183,
      "loss": 1.8375,
      "step": 156109
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.111289381980896,
      "learning_rate": 0.00014120476401980334,
      "loss": 2.226,
      "step": 156110
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.1822459697723389,
      "learning_rate": 0.00014120129350404787,
      "loss": 2.1779,
      "step": 156111
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.0959519147872925,
      "learning_rate": 0.00014119782301781606,
      "loss": 2.07,
      "step": 156112
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.3042534589767456,
      "learning_rate": 0.00014119435256110846,
      "loss": 2.126,
      "step": 156113
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.102500081062317,
      "learning_rate": 0.00014119088213392596,
      "loss": 2.0884,
      "step": 156114
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.2265901565551758,
      "learning_rate": 0.00014118741173626894,
      "loss": 1.9334,
      "step": 156115
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.2836319208145142,
      "learning_rate": 0.00014118394136813827,
      "loss": 1.9915,
      "step": 156116
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.2748922109603882,
      "learning_rate": 0.00014118047102953458,
      "loss": 2.013,
      "step": 156117
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.1080557107925415,
      "learning_rate": 0.00014117700072045837,
      "loss": 2.0854,
      "step": 156118
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.0987461805343628,
      "learning_rate": 0.00014117353044091024,
      "loss": 1.9026,
      "step": 156119
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.135509729385376,
      "learning_rate": 0.0001411700601908911,
      "loss": 1.9822,
      "step": 156120
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.111102819442749,
      "learning_rate": 0.0001411665899704013,
      "loss": 2.1471,
      "step": 156121
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.135417103767395,
      "learning_rate": 0.00014116311977944177,
      "loss": 2.1521,
      "step": 156122
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.069008469581604,
      "learning_rate": 0.000141159649618013,
      "loss": 1.8501,
      "step": 156123
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.2454540729522705,
      "learning_rate": 0.00014115617948611564,
      "loss": 2.0046,
      "step": 156124
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.186147689819336,
      "learning_rate": 0.00014115270938375025,
      "loss": 2.1134,
      "step": 156125
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.2449843883514404,
      "learning_rate": 0.00014114923931091768,
      "loss": 1.9134,
      "step": 156126
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.2127748727798462,
      "learning_rate": 0.00014114576926761835,
      "loss": 1.9516,
      "step": 156127
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.1554887294769287,
      "learning_rate": 0.00014114229925385311,
      "loss": 2.1032,
      "step": 156128
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.1887303590774536,
      "learning_rate": 0.00014113882926962255,
      "loss": 2.0063,
      "step": 156129
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.3598253726959229,
      "learning_rate": 0.00014113535931492725,
      "loss": 2.1885,
      "step": 156130
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.2434308528900146,
      "learning_rate": 0.0001411318893897678,
      "loss": 2.0068,
      "step": 156131
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.1378847360610962,
      "learning_rate": 0.00014112841949414504,
      "loss": 2.0923,
      "step": 156132
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.3609615564346313,
      "learning_rate": 0.0001411249496280594,
      "loss": 1.9952,
      "step": 156133
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.3455618619918823,
      "learning_rate": 0.0001411214797915117,
      "loss": 1.9625,
      "step": 156134
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.035414695739746,
      "learning_rate": 0.00014111800998450254,
      "loss": 2.1408,
      "step": 156135
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.0949161052703857,
      "learning_rate": 0.00014111454020703253,
      "loss": 1.9541,
      "step": 156136
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.200019359588623,
      "learning_rate": 0.00014111107045910223,
      "loss": 1.8213,
      "step": 156137
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.136318325996399,
      "learning_rate": 0.00014110760074071248,
      "loss": 1.9559,
      "step": 156138
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.1482654809951782,
      "learning_rate": 0.00014110413105186375,
      "loss": 2.2511,
      "step": 156139
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.3296337127685547,
      "learning_rate": 0.00014110066139255687,
      "loss": 2.0625,
      "step": 156140
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.350521445274353,
      "learning_rate": 0.00014109719176279222,
      "loss": 2.2145,
      "step": 156141
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.1460094451904297,
      "learning_rate": 0.00014109372216257086,
      "loss": 2.0496,
      "step": 156142
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.2287551164627075,
      "learning_rate": 0.00014109025259189295,
      "loss": 2.0593,
      "step": 156143
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.2168422937393188,
      "learning_rate": 0.00014108678305075944,
      "loss": 1.978,
      "step": 156144
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.0981800556182861,
      "learning_rate": 0.00014108331353917083,
      "loss": 2.1749,
      "step": 156145
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.2784796953201294,
      "learning_rate": 0.0001410798440571279,
      "loss": 1.8816,
      "step": 156146
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.2607685327529907,
      "learning_rate": 0.00014107637460463116,
      "loss": 2.0162,
      "step": 156147
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.241567850112915,
      "learning_rate": 0.00014107290518168152,
      "loss": 1.8259,
      "step": 156148
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.2022849321365356,
      "learning_rate": 0.0001410694357882792,
      "loss": 2.2706,
      "step": 156149
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.1417361497879028,
      "learning_rate": 0.0001410659664244252,
      "loss": 2.0739,
      "step": 156150
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.0790594816207886,
      "learning_rate": 0.00014106249709011992,
      "loss": 1.9773,
      "step": 156151
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.269118309020996,
      "learning_rate": 0.0001410590277853642,
      "loss": 2.0432,
      "step": 156152
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.2314032316207886,
      "learning_rate": 0.00014105555851015857,
      "loss": 1.8722,
      "step": 156153
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.2511526346206665,
      "learning_rate": 0.00014105208926450387,
      "loss": 2.068,
      "step": 156154
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.2784112691879272,
      "learning_rate": 0.00014104862004840038,
      "loss": 2.085,
      "step": 156155
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.3888863325119019,
      "learning_rate": 0.0001410451508618491,
      "loss": 2.0974,
      "step": 156156
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.180991291999817,
      "learning_rate": 0.00014104168170485035,
      "loss": 2.2013,
      "step": 156157
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.2230018377304077,
      "learning_rate": 0.0001410382125774051,
      "loss": 2.0219,
      "step": 156158
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.128273844718933,
      "learning_rate": 0.00014103474347951372,
      "loss": 1.9409,
      "step": 156159
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.266012191772461,
      "learning_rate": 0.00014103127441117714,
      "loss": 2.1143,
      "step": 156160
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.1782302856445312,
      "learning_rate": 0.00014102780537239578,
      "loss": 2.0857,
      "step": 156161
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.1660810708999634,
      "learning_rate": 0.0001410243363631704,
      "loss": 2.0712,
      "step": 156162
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.1463559865951538,
      "learning_rate": 0.00014102086738350148,
      "loss": 2.3125,
      "step": 156163
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.1116725206375122,
      "learning_rate": 0.00014101739843338986,
      "loss": 1.8636,
      "step": 156164
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.2158421277999878,
      "learning_rate": 0.00014101392951283603,
      "loss": 2.1491,
      "step": 156165
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.2440485954284668,
      "learning_rate": 0.0001410104606218408,
      "loss": 1.9869,
      "step": 156166
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.1073461771011353,
      "learning_rate": 0.00014100699176040476,
      "loss": 2.0477,
      "step": 156167
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.103105068206787,
      "learning_rate": 0.00014100352292852847,
      "loss": 1.8947,
      "step": 156168
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.2682098150253296,
      "learning_rate": 0.00014100005412621256,
      "loss": 1.9081,
      "step": 156169
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.249384880065918,
      "learning_rate": 0.00014099658535345785,
      "loss": 2.064,
      "step": 156170
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.19265615940094,
      "learning_rate": 0.00014099311661026475,
      "loss": 2.075,
      "step": 156171
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.1635520458221436,
      "learning_rate": 0.00014098964789663416,
      "loss": 2.1355,
      "step": 156172
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.1526154279708862,
      "learning_rate": 0.00014098617921256656,
      "loss": 2.0942,
      "step": 156173
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.3365362882614136,
      "learning_rate": 0.00014098271055806256,
      "loss": 1.9018,
      "step": 156174
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.1384146213531494,
      "learning_rate": 0.00014097924193312296,
      "loss": 1.9065,
      "step": 156175
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.1926891803741455,
      "learning_rate": 0.00014097577333774833,
      "loss": 2.2762,
      "step": 156176
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.1738530397415161,
      "learning_rate": 0.0001409723047719392,
      "loss": 2.0992,
      "step": 156177
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.2599915266036987,
      "learning_rate": 0.00014096883623569638,
      "loss": 2.1152,
      "step": 156178
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.130918025970459,
      "learning_rate": 0.00014096536772902052,
      "loss": 1.8986,
      "step": 156179
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.098777174949646,
      "learning_rate": 0.00014096189925191205,
      "loss": 2.0512,
      "step": 156180
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.6615711450576782,
      "learning_rate": 0.0001409584308043719,
      "loss": 2.0979,
      "step": 156181
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.2901076078414917,
      "learning_rate": 0.0001409549623864005,
      "loss": 1.8707,
      "step": 156182
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.1576184034347534,
      "learning_rate": 0.00014095149399799862,
      "loss": 1.9205,
      "step": 156183
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.0982288122177124,
      "learning_rate": 0.0001409480256391669,
      "loss": 2.1038,
      "step": 156184
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.242749571800232,
      "learning_rate": 0.00014094455730990593,
      "loss": 1.9011,
      "step": 156185
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.0567882061004639,
      "learning_rate": 0.00014094108901021625,
      "loss": 2.0705,
      "step": 156186
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.142348051071167,
      "learning_rate": 0.00014093762074009875,
      "loss": 1.9398,
      "step": 156187
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.0883671045303345,
      "learning_rate": 0.00014093415249955384,
      "loss": 2.036,
      "step": 156188
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.132136583328247,
      "learning_rate": 0.0001409306842885824,
      "loss": 1.8428,
      "step": 156189
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.1447173357009888,
      "learning_rate": 0.0001409272161071849,
      "loss": 1.9832,
      "step": 156190
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.2896349430084229,
      "learning_rate": 0.00014092374795536205,
      "loss": 2.0015,
      "step": 156191
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.0927374362945557,
      "learning_rate": 0.0001409202798331144,
      "loss": 2.1286,
      "step": 156192
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.0984963178634644,
      "learning_rate": 0.00014091681174044276,
      "loss": 2.0226,
      "step": 156193
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.2099037170410156,
      "learning_rate": 0.00014091334367734756,
      "loss": 2.1906,
      "step": 156194
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.1704524755477905,
      "learning_rate": 0.0001409098756438297,
      "loss": 2.2274,
      "step": 156195
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.1698024272918701,
      "learning_rate": 0.0001409064076398897,
      "loss": 2.0036,
      "step": 156196
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.1705315113067627,
      "learning_rate": 0.00014090293966552818,
      "loss": 2.1959,
      "step": 156197
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.1545876264572144,
      "learning_rate": 0.00014089947172074574,
      "loss": 2.0862,
      "step": 156198
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.2447415590286255,
      "learning_rate": 0.00014089600380554318,
      "loss": 2.0353,
      "step": 156199
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.1538975238800049,
      "learning_rate": 0.00014089253591992095,
      "loss": 2.2217,
      "step": 156200
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.240535020828247,
      "learning_rate": 0.0001408890680638799,
      "loss": 2.0933,
      "step": 156201
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.267139196395874,
      "learning_rate": 0.0001408856002374206,
      "loss": 1.9875,
      "step": 156202
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.1760869026184082,
      "learning_rate": 0.00014088213244054364,
      "loss": 2.0068,
      "step": 156203
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.2841211557388306,
      "learning_rate": 0.0001408786646732496,
      "loss": 2.0889,
      "step": 156204
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.3023468255996704,
      "learning_rate": 0.0001408751969355393,
      "loss": 2.2105,
      "step": 156205
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.2185173034667969,
      "learning_rate": 0.00014087172922741322,
      "loss": 2.3204,
      "step": 156206
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.142409086227417,
      "learning_rate": 0.0001408682615488722,
      "loss": 2.1041,
      "step": 156207
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.1106398105621338,
      "learning_rate": 0.00014086479389991666,
      "loss": 2.02,
      "step": 156208
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.1290055513381958,
      "learning_rate": 0.0001408613262805476,
      "loss": 2.0199,
      "step": 156209
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.2018898725509644,
      "learning_rate": 0.00014085785869076518,
      "loss": 2.2048,
      "step": 156210
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.3511022329330444,
      "learning_rate": 0.0001408543911305704,
      "loss": 1.9213,
      "step": 156211
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.2389174699783325,
      "learning_rate": 0.00014085092359996372,
      "loss": 1.8573,
      "step": 156212
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.2426178455352783,
      "learning_rate": 0.00014084745609894593,
      "loss": 2.2591,
      "step": 156213
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.2153903245925903,
      "learning_rate": 0.00014084398862751748,
      "loss": 1.9217,
      "step": 156214
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.1238694190979004,
      "learning_rate": 0.00014084052118567937,
      "loss": 1.9849,
      "step": 156215
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.1639386415481567,
      "learning_rate": 0.00014083705377343177,
      "loss": 2.1911,
      "step": 156216
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.1394354104995728,
      "learning_rate": 0.0001408335863907757,
      "loss": 1.8766,
      "step": 156217
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.6331250667572021,
      "learning_rate": 0.0001408301190377116,
      "loss": 2.0242,
      "step": 156218
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.2186206579208374,
      "learning_rate": 0.00014082665171424026,
      "loss": 2.1203,
      "step": 156219
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.2011362314224243,
      "learning_rate": 0.00014082318442036214,
      "loss": 1.9731,
      "step": 156220
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.1958739757537842,
      "learning_rate": 0.0001408197171560782,
      "loss": 2.2006,
      "step": 156221
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.0972557067871094,
      "learning_rate": 0.00014081624992138867,
      "loss": 2.1524,
      "step": 156222
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.3862305879592896,
      "learning_rate": 0.00014081278271629454,
      "loss": 2.1815,
      "step": 156223
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.2372275590896606,
      "learning_rate": 0.00014080931554079614,
      "loss": 2.1627,
      "step": 156224
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.175645351409912,
      "learning_rate": 0.00014080584839489446,
      "loss": 1.9583,
      "step": 156225
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.3676444292068481,
      "learning_rate": 0.00014080238127858985,
      "loss": 1.7718,
      "step": 156226
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.3091415166854858,
      "learning_rate": 0.0001407989141918833,
      "loss": 1.9145,
      "step": 156227
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.196192741394043,
      "learning_rate": 0.00014079544713477504,
      "loss": 2.084,
      "step": 156228
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.1747779846191406,
      "learning_rate": 0.00014079198010726602,
      "loss": 2.0977,
      "step": 156229
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.3402115106582642,
      "learning_rate": 0.00014078851310935663,
      "loss": 2.0564,
      "step": 156230
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.253816843032837,
      "learning_rate": 0.0001407850461410478,
      "loss": 1.9237,
      "step": 156231
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.2597260475158691,
      "learning_rate": 0.00014078157920233993,
      "loss": 2.0808,
      "step": 156232
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.0826350450515747,
      "learning_rate": 0.00014077811229323397,
      "loss": 2.0974,
      "step": 156233
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.1157917976379395,
      "learning_rate": 0.00014077464541373017,
      "loss": 2.0339,
      "step": 156234
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.3035207986831665,
      "learning_rate": 0.00014077117856382942,
      "loss": 1.9645,
      "step": 156235
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.1388487815856934,
      "learning_rate": 0.00014076771174353225,
      "loss": 2.0188,
      "step": 156236
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.1391884088516235,
      "learning_rate": 0.00014076424495283952,
      "loss": 2.05,
      "step": 156237
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.0742628574371338,
      "learning_rate": 0.00014076077819175154,
      "loss": 2.1529,
      "step": 156238
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.0900660753250122,
      "learning_rate": 0.0001407573114602694,
      "loss": 2.1287,
      "step": 156239
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.176112413406372,
      "learning_rate": 0.00014075384475839323,
      "loss": 1.8788,
      "step": 156240
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.1041944026947021,
      "learning_rate": 0.00014075037808612405,
      "loss": 1.8067,
      "step": 156241
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.1805784702301025,
      "learning_rate": 0.00014074691144346229,
      "loss": 2.0751,
      "step": 156242
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.1752017736434937,
      "learning_rate": 0.00014074344483040876,
      "loss": 2.2071,
      "step": 156243
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.2598967552185059,
      "learning_rate": 0.00014073997824696393,
      "loss": 1.9392,
      "step": 156244
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.1629890203475952,
      "learning_rate": 0.00014073651169312868,
      "loss": 2.0717,
      "step": 156245
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.0671658515930176,
      "learning_rate": 0.00014073304516890348,
      "loss": 1.8187,
      "step": 156246
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.1881521940231323,
      "learning_rate": 0.00014072957867428904,
      "loss": 2.0271,
      "step": 156247
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.085355520248413,
      "learning_rate": 0.00014072611220928587,
      "loss": 2.0432,
      "step": 156248
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.2929739952087402,
      "learning_rate": 0.0001407226457738948,
      "loss": 2.0099,
      "step": 156249
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.0883787870407104,
      "learning_rate": 0.00014071917936811634,
      "loss": 1.8527,
      "step": 156250
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.0915147066116333,
      "learning_rate": 0.00014071571299195126,
      "loss": 2.2662,
      "step": 156251
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.1361958980560303,
      "learning_rate": 0.00014071224664540015,
      "loss": 1.8752,
      "step": 156252
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.2013218402862549,
      "learning_rate": 0.00014070878032846365,
      "loss": 2.013,
      "step": 156253
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.1971640586853027,
      "learning_rate": 0.00014070531404114227,
      "loss": 1.9987,
      "step": 156254
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.1812012195587158,
      "learning_rate": 0.00014070184778343687,
      "loss": 1.8518,
      "step": 156255
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.243308663368225,
      "learning_rate": 0.00014069838155534792,
      "loss": 2.3014,
      "step": 156256
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.1811491250991821,
      "learning_rate": 0.00014069491535687628,
      "loss": 2.1419,
      "step": 156257
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.2121764421463013,
      "learning_rate": 0.0001406914491880224,
      "loss": 2.0328,
      "step": 156258
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.090014934539795,
      "learning_rate": 0.00014068798304878692,
      "loss": 1.7928,
      "step": 156259
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.0252324342727661,
      "learning_rate": 0.00014068451693917064,
      "loss": 2.0487,
      "step": 156260
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.1427751779556274,
      "learning_rate": 0.00014068105085917416,
      "loss": 2.3491,
      "step": 156261
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.2756024599075317,
      "learning_rate": 0.00014067758480879793,
      "loss": 1.9127,
      "step": 156262
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.2877461910247803,
      "learning_rate": 0.00014067411878804285,
      "loss": 2.085,
      "step": 156263
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.1421154737472534,
      "learning_rate": 0.00014067065279690948,
      "loss": 2.0239,
      "step": 156264
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.2249399423599243,
      "learning_rate": 0.0001406671868353983,
      "loss": 2.0286,
      "step": 156265
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.1430634260177612,
      "learning_rate": 0.00014066372090351025,
      "loss": 2.0184,
      "step": 156266
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.2733080387115479,
      "learning_rate": 0.0001406602550012457,
      "loss": 2.0735,
      "step": 156267
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.267730712890625,
      "learning_rate": 0.00014065678912860552,
      "loss": 2.2038,
      "step": 156268
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.1181659698486328,
      "learning_rate": 0.00014065332328559027,
      "loss": 1.9196,
      "step": 156269
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.211732029914856,
      "learning_rate": 0.00014064985747220055,
      "loss": 2.1578,
      "step": 156270
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.3389084339141846,
      "learning_rate": 0.00014064639168843692,
      "loss": 2.2905,
      "step": 156271
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.186684250831604,
      "learning_rate": 0.00014064292593430022,
      "loss": 2.0543,
      "step": 156272
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.3245658874511719,
      "learning_rate": 0.00014063946020979095,
      "loss": 2.0225,
      "step": 156273
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.2079721689224243,
      "learning_rate": 0.00014063599451490992,
      "loss": 2.0722,
      "step": 156274
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.1797940731048584,
      "learning_rate": 0.00014063252884965763,
      "loss": 2.1724,
      "step": 156275
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.123500943183899,
      "learning_rate": 0.00014062906321403478,
      "loss": 2.3065,
      "step": 156276
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.126691222190857,
      "learning_rate": 0.00014062559760804187,
      "loss": 1.9596,
      "step": 156277
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.2532681226730347,
      "learning_rate": 0.00014062213203167983,
      "loss": 2.0258,
      "step": 156278
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.0628290176391602,
      "learning_rate": 0.00014061866648494898,
      "loss": 1.9411,
      "step": 156279
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.2473156452178955,
      "learning_rate": 0.00014061520096785024,
      "loss": 2.0037,
      "step": 156280
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.2179224491119385,
      "learning_rate": 0.00014061173548038405,
      "loss": 1.9469,
      "step": 156281
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.1026180982589722,
      "learning_rate": 0.0001406082700225514,
      "loss": 1.9807,
      "step": 156282
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.2011967897415161,
      "learning_rate": 0.0001406048045943524,
      "loss": 2.1065,
      "step": 156283
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.2982097864151,
      "learning_rate": 0.00014060133919578812,
      "loss": 2.0458,
      "step": 156284
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.1506928205490112,
      "learning_rate": 0.00014059787382685894,
      "loss": 1.7413,
      "step": 156285
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.4155044555664062,
      "learning_rate": 0.00014059440848756576,
      "loss": 2.0531,
      "step": 156286
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.1103535890579224,
      "learning_rate": 0.00014059094317790894,
      "loss": 1.9995,
      "step": 156287
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.2679884433746338,
      "learning_rate": 0.0001405874778978895,
      "loss": 2.1281,
      "step": 156288
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.3227126598358154,
      "learning_rate": 0.00014058401264750766,
      "loss": 2.2927,
      "step": 156289
    },
    {
      "epoch": 2.03,
      "grad_norm": 1.1999664306640625,
      "learning_rate": 0.00014058054742676433,
      "loss": 1.8764,
      "step": 156290
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.325951337814331,
      "learning_rate": 0.00014057708223566003,
      "loss": 2.0949,
      "step": 156291
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.0463752746582031,
      "learning_rate": 0.00014057361707419552,
      "loss": 1.9313,
      "step": 156292
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.2577329874038696,
      "learning_rate": 0.0001405701519423713,
      "loss": 2.2169,
      "step": 156293
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.2392809391021729,
      "learning_rate": 0.00014056668684018833,
      "loss": 2.2248,
      "step": 156294
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.0556472539901733,
      "learning_rate": 0.0001405632217676468,
      "loss": 1.9918,
      "step": 156295
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.193801760673523,
      "learning_rate": 0.00014055975672474765,
      "loss": 1.8847,
      "step": 156296
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.2254825830459595,
      "learning_rate": 0.00014055629171149138,
      "loss": 2.0173,
      "step": 156297
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.2342556715011597,
      "learning_rate": 0.0001405528267278788,
      "loss": 1.983,
      "step": 156298
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.1924123764038086,
      "learning_rate": 0.00014054936177391035,
      "loss": 2.1957,
      "step": 156299
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.2074233293533325,
      "learning_rate": 0.000140545896849587,
      "loss": 1.8216,
      "step": 156300
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.2106246948242188,
      "learning_rate": 0.00014054243195490895,
      "loss": 2.2151,
      "step": 156301
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.256287932395935,
      "learning_rate": 0.0001405389670898772,
      "loss": 2.2583,
      "step": 156302
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.2064422369003296,
      "learning_rate": 0.00014053550225449216,
      "loss": 2.1256,
      "step": 156303
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.0901063680648804,
      "learning_rate": 0.00014053203744875467,
      "loss": 1.8218,
      "step": 156304
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.1470897197723389,
      "learning_rate": 0.00014052857267266518,
      "loss": 2.0883,
      "step": 156305
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.183631420135498,
      "learning_rate": 0.00014052510792622467,
      "loss": 2.0291,
      "step": 156306
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.1897766590118408,
      "learning_rate": 0.0001405216432094333,
      "loss": 1.9162,
      "step": 156307
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.1526365280151367,
      "learning_rate": 0.0001405181785222921,
      "loss": 2.1624,
      "step": 156308
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.319083571434021,
      "learning_rate": 0.0001405147138648015,
      "loss": 2.0869,
      "step": 156309
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.1969653367996216,
      "learning_rate": 0.0001405112492369623,
      "loss": 1.9861,
      "step": 156310
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.1400524377822876,
      "learning_rate": 0.00014050778463877496,
      "loss": 2.0802,
      "step": 156311
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.041771650314331,
      "learning_rate": 0.00014050432007024048,
      "loss": 2.0273,
      "step": 156312
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.2602744102478027,
      "learning_rate": 0.000140500855531359,
      "loss": 2.0596,
      "step": 156313
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.138471007347107,
      "learning_rate": 0.00014049739102213156,
      "loss": 1.9098,
      "step": 156314
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.2244080305099487,
      "learning_rate": 0.00014049392654255852,
      "loss": 1.8858,
      "step": 156315
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.202223300933838,
      "learning_rate": 0.00014049046209264082,
      "loss": 2.0172,
      "step": 156316
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.151068091392517,
      "learning_rate": 0.00014048699767237882,
      "loss": 1.9646,
      "step": 156317
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.213737964630127,
      "learning_rate": 0.00014048353328177353,
      "loss": 2.0709,
      "step": 156318
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.1042671203613281,
      "learning_rate": 0.00014048006892082514,
      "loss": 2.042,
      "step": 156319
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.396662950515747,
      "learning_rate": 0.00014047660458953463,
      "loss": 1.9588,
      "step": 156320
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.1735862493515015,
      "learning_rate": 0.00014047314028790242,
      "loss": 2.0619,
      "step": 156321
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.1118417978286743,
      "learning_rate": 0.00014046967601592938,
      "loss": 2.0228,
      "step": 156322
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.1821755170822144,
      "learning_rate": 0.00014046621177361593,
      "loss": 2.1485,
      "step": 156323
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.149759292602539,
      "learning_rate": 0.00014046274756096303,
      "loss": 2.0926,
      "step": 156324
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.1525249481201172,
      "learning_rate": 0.0001404592833779709,
      "loss": 2.1164,
      "step": 156325
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.1718276739120483,
      "learning_rate": 0.00014045581922464049,
      "loss": 2.31,
      "step": 156326
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.167849063873291,
      "learning_rate": 0.00014045235510097228,
      "loss": 2.0758,
      "step": 156327
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.0572662353515625,
      "learning_rate": 0.00014044889100696707,
      "loss": 2.0591,
      "step": 156328
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.2111467123031616,
      "learning_rate": 0.00014044542694262532,
      "loss": 2.2338,
      "step": 156329
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.264762043952942,
      "learning_rate": 0.0001404419629079479,
      "loss": 1.9635,
      "step": 156330
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.1499342918395996,
      "learning_rate": 0.00014043849890293532,
      "loss": 2.1322,
      "step": 156331
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.2310892343521118,
      "learning_rate": 0.00014043503492758825,
      "loss": 2.1327,
      "step": 156332
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.1637057065963745,
      "learning_rate": 0.0001404315709819072,
      "loss": 1.9572,
      "step": 156333
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.4457148313522339,
      "learning_rate": 0.00014042810706589305,
      "loss": 2.2463,
      "step": 156334
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.1875696182250977,
      "learning_rate": 0.00014042464317954619,
      "loss": 1.9494,
      "step": 156335
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.2791731357574463,
      "learning_rate": 0.00014042117932286754,
      "loss": 2.0033,
      "step": 156336
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.116620659828186,
      "learning_rate": 0.0001404177154958576,
      "loss": 1.9615,
      "step": 156337
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.124412178993225,
      "learning_rate": 0.000140414251698517,
      "loss": 2.2935,
      "step": 156338
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.170433759689331,
      "learning_rate": 0.00014041078793084632,
      "loss": 2.0455,
      "step": 156339
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.2007341384887695,
      "learning_rate": 0.0001404073241928464,
      "loss": 2.081,
      "step": 156340
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.2123711109161377,
      "learning_rate": 0.00014040386048451764,
      "loss": 2.0897,
      "step": 156341
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.1241590976715088,
      "learning_rate": 0.00014040039680586093,
      "loss": 1.9715,
      "step": 156342
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.1621237993240356,
      "learning_rate": 0.0001403969331568768,
      "loss": 1.9487,
      "step": 156343
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.2153321504592896,
      "learning_rate": 0.00014039346953756575,
      "loss": 2.1371,
      "step": 156344
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.1270604133605957,
      "learning_rate": 0.00014039000594792873,
      "loss": 1.9868,
      "step": 156345
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.3483966588974,
      "learning_rate": 0.00014038654238796619,
      "loss": 1.9508,
      "step": 156346
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.0746477842330933,
      "learning_rate": 0.0001403830788576787,
      "loss": 2.0284,
      "step": 156347
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.2034107446670532,
      "learning_rate": 0.00014037961535706713,
      "loss": 2.1008,
      "step": 156348
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.30422842502594,
      "learning_rate": 0.00014037615188613198,
      "loss": 2.0806,
      "step": 156349
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.1407960653305054,
      "learning_rate": 0.00014037268844487383,
      "loss": 1.8773,
      "step": 156350
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.108951449394226,
      "learning_rate": 0.0001403692250332935,
      "loss": 2.1265,
      "step": 156351
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.0895946025848389,
      "learning_rate": 0.00014036576165139158,
      "loss": 2.1678,
      "step": 156352
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.2788257598876953,
      "learning_rate": 0.00014036229829916854,
      "loss": 2.1901,
      "step": 156353
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.2765015363693237,
      "learning_rate": 0.0001403588349766253,
      "loss": 2.2463,
      "step": 156354
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.3826260566711426,
      "learning_rate": 0.00014035537168376232,
      "loss": 2.3168,
      "step": 156355
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.1873000860214233,
      "learning_rate": 0.00014035190842058017,
      "loss": 2.0413,
      "step": 156356
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.1688194274902344,
      "learning_rate": 0.00014034844518707977,
      "loss": 2.0375,
      "step": 156357
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.124485731124878,
      "learning_rate": 0.00014034498198326148,
      "loss": 2.0434,
      "step": 156358
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.0186052322387695,
      "learning_rate": 0.0001403415188091262,
      "loss": 1.8965,
      "step": 156359
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.2131891250610352,
      "learning_rate": 0.00014033805566467445,
      "loss": 2.1412,
      "step": 156360
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.1521520614624023,
      "learning_rate": 0.00014033459254990686,
      "loss": 1.7429,
      "step": 156361
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.1190229654312134,
      "learning_rate": 0.00014033112946482395,
      "loss": 2.0603,
      "step": 156362
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.1513922214508057,
      "learning_rate": 0.0001403276664094266,
      "loss": 2.218,
      "step": 156363
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.0817763805389404,
      "learning_rate": 0.00014032420338371527,
      "loss": 1.8743,
      "step": 156364
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.1557341814041138,
      "learning_rate": 0.00014032074038769078,
      "loss": 1.9978,
      "step": 156365
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.2977546453475952,
      "learning_rate": 0.0001403172774213536,
      "loss": 1.9571,
      "step": 156366
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.2355999946594238,
      "learning_rate": 0.00014031381448470463,
      "loss": 2.0461,
      "step": 156367
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.2220759391784668,
      "learning_rate": 0.00014031035157774415,
      "loss": 2.3777,
      "step": 156368
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.1533738374710083,
      "learning_rate": 0.00014030688870047308,
      "loss": 1.9328,
      "step": 156369
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.0878654718399048,
      "learning_rate": 0.00014030342585289186,
      "loss": 2.1153,
      "step": 156370
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.1367628574371338,
      "learning_rate": 0.00014029996303500136,
      "loss": 1.9263,
      "step": 156371
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.1486064195632935,
      "learning_rate": 0.000140296500246802,
      "loss": 2.0444,
      "step": 156372
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.4057648181915283,
      "learning_rate": 0.00014029303748829477,
      "loss": 2.1508,
      "step": 156373
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.2192648649215698,
      "learning_rate": 0.00014028957475947985,
      "loss": 2.0814,
      "step": 156374
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.2735379934310913,
      "learning_rate": 0.00014028611206035822,
      "loss": 2.0355,
      "step": 156375
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.140696406364441,
      "learning_rate": 0.0001402826493909303,
      "loss": 1.8306,
      "step": 156376
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.5498884916305542,
      "learning_rate": 0.000140279186751197,
      "loss": 1.9044,
      "step": 156377
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.2634531259536743,
      "learning_rate": 0.00014027572414115867,
      "loss": 1.9521,
      "step": 156378
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.1830062866210938,
      "learning_rate": 0.00014027226156081632,
      "loss": 2.1865,
      "step": 156379
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.079524040222168,
      "learning_rate": 0.00014026879901017015,
      "loss": 1.9025,
      "step": 156380
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.5560013055801392,
      "learning_rate": 0.00014026533648922116,
      "loss": 1.8777,
      "step": 156381
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.3579078912734985,
      "learning_rate": 0.0001402618739979697,
      "loss": 2.1208,
      "step": 156382
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.1791865825653076,
      "learning_rate": 0.00014025841153641672,
      "loss": 2.2404,
      "step": 156383
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.083534836769104,
      "learning_rate": 0.0001402549491045626,
      "loss": 2.1581,
      "step": 156384
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.1718121767044067,
      "learning_rate": 0.0001402514867024083,
      "loss": 2.1381,
      "step": 156385
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.294828176498413,
      "learning_rate": 0.00014024802432995406,
      "loss": 2.0507,
      "step": 156386
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.375257968902588,
      "learning_rate": 0.00014024456198720087,
      "loss": 1.9059,
      "step": 156387
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.1691503524780273,
      "learning_rate": 0.00014024109967414908,
      "loss": 1.8237,
      "step": 156388
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.0756851434707642,
      "learning_rate": 0.00014023763739079962,
      "loss": 1.8971,
      "step": 156389
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.2699183225631714,
      "learning_rate": 0.00014023417513715287,
      "loss": 2.1077,
      "step": 156390
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.2998391389846802,
      "learning_rate": 0.00014023071291320983,
      "loss": 2.0457,
      "step": 156391
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.1595124006271362,
      "learning_rate": 0.0001402272507189707,
      "loss": 2.1323,
      "step": 156392
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.0912338495254517,
      "learning_rate": 0.00014022378855443643,
      "loss": 1.8479,
      "step": 156393
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.3550559282302856,
      "learning_rate": 0.00014022032641960747,
      "loss": 2.0026,
      "step": 156394
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.2896621227264404,
      "learning_rate": 0.0001402168643144847,
      "loss": 1.871,
      "step": 156395
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.2550581693649292,
      "learning_rate": 0.0001402134022390685,
      "loss": 1.7785,
      "step": 156396
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.1265887022018433,
      "learning_rate": 0.0001402099401933599,
      "loss": 2.1245,
      "step": 156397
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.1670985221862793,
      "learning_rate": 0.00014020647817735902,
      "loss": 1.888,
      "step": 156398
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.5145899057388306,
      "learning_rate": 0.00014020301619106692,
      "loss": 2.1063,
      "step": 156399
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.1037479639053345,
      "learning_rate": 0.00014019955423448394,
      "loss": 2.0239,
      "step": 156400
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.0644930601119995,
      "learning_rate": 0.00014019609230761103,
      "loss": 1.9211,
      "step": 156401
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.1155283451080322,
      "learning_rate": 0.00014019263041044856,
      "loss": 2.1985,
      "step": 156402
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.1611433029174805,
      "learning_rate": 0.00014018916854299752,
      "loss": 2.1335,
      "step": 156403
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.2770408391952515,
      "learning_rate": 0.00014018570670525812,
      "loss": 1.9405,
      "step": 156404
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.2276092767715454,
      "learning_rate": 0.0001401822448972313,
      "loss": 2.177,
      "step": 156405
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.0596119165420532,
      "learning_rate": 0.0001401787831189175,
      "loss": 1.6651,
      "step": 156406
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.1010385751724243,
      "learning_rate": 0.00014017532137031762,
      "loss": 1.934,
      "step": 156407
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.1349213123321533,
      "learning_rate": 0.00014017185965143204,
      "loss": 2.1689,
      "step": 156408
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.1688933372497559,
      "learning_rate": 0.00014016839796226167,
      "loss": 1.8085,
      "step": 156409
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.144822359085083,
      "learning_rate": 0.00014016493630280698,
      "loss": 2.1395,
      "step": 156410
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.2751182317733765,
      "learning_rate": 0.00014016147467306862,
      "loss": 2.27,
      "step": 156411
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.2330117225646973,
      "learning_rate": 0.0001401580130730472,
      "loss": 2.2164,
      "step": 156412
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.3544138669967651,
      "learning_rate": 0.0001401545515027435,
      "loss": 2.4029,
      "step": 156413
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.3656773567199707,
      "learning_rate": 0.000140151089962158,
      "loss": 2.241,
      "step": 156414
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.0751136541366577,
      "learning_rate": 0.00014014762845129152,
      "loss": 1.8764,
      "step": 156415
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.164139986038208,
      "learning_rate": 0.0001401441669701446,
      "loss": 2.0794,
      "step": 156416
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.165922999382019,
      "learning_rate": 0.00014014070551871796,
      "loss": 1.9219,
      "step": 156417
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.1985759735107422,
      "learning_rate": 0.00014013724409701203,
      "loss": 2.1377,
      "step": 156418
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.0578371286392212,
      "learning_rate": 0.00014013378270502768,
      "loss": 2.0198,
      "step": 156419
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.2007962465286255,
      "learning_rate": 0.00014013032134276541,
      "loss": 1.9634,
      "step": 156420
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.1070291996002197,
      "learning_rate": 0.00014012686001022606,
      "loss": 1.8755,
      "step": 156421
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.1351220607757568,
      "learning_rate": 0.00014012339870741008,
      "loss": 1.9555,
      "step": 156422
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.1418477296829224,
      "learning_rate": 0.00014011993743431823,
      "loss": 2.0302,
      "step": 156423
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.3798421621322632,
      "learning_rate": 0.00014011647619095098,
      "loss": 2.1379,
      "step": 156424
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.1921842098236084,
      "learning_rate": 0.0001401130149773092,
      "loss": 1.8457,
      "step": 156425
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.2202929258346558,
      "learning_rate": 0.0001401095537933933,
      "loss": 2.1909,
      "step": 156426
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.2120213508605957,
      "learning_rate": 0.0001401060926392042,
      "loss": 1.9568,
      "step": 156427
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.1869720220565796,
      "learning_rate": 0.00014010263151474235,
      "loss": 2.2507,
      "step": 156428
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.1609174013137817,
      "learning_rate": 0.00014009917042000837,
      "loss": 2.0999,
      "step": 156429
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.4949123859405518,
      "learning_rate": 0.0001400957093550031,
      "loss": 1.9675,
      "step": 156430
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.1509044170379639,
      "learning_rate": 0.000140092248319727,
      "loss": 1.9353,
      "step": 156431
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.1567152738571167,
      "learning_rate": 0.0001400887873141807,
      "loss": 2.2792,
      "step": 156432
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.221396565437317,
      "learning_rate": 0.00014008532633836497,
      "loss": 2.0548,
      "step": 156433
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.3726389408111572,
      "learning_rate": 0.00014008186539228045,
      "loss": 1.8757,
      "step": 156434
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.119657039642334,
      "learning_rate": 0.00014007840447592763,
      "loss": 1.8945,
      "step": 156435
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.0723868608474731,
      "learning_rate": 0.0001400749435893073,
      "loss": 2.0035,
      "step": 156436
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.2593798637390137,
      "learning_rate": 0.00014007148273242009,
      "loss": 1.8708,
      "step": 156437
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.146824836730957,
      "learning_rate": 0.0001400680219052665,
      "loss": 2.1467,
      "step": 156438
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.1434563398361206,
      "learning_rate": 0.00014006456110784743,
      "loss": 1.9315,
      "step": 156439
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.1548725366592407,
      "learning_rate": 0.00014006110034016334,
      "loss": 1.9909,
      "step": 156440
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.2021170854568481,
      "learning_rate": 0.0001400576396022148,
      "loss": 2.0091,
      "step": 156441
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.3229955434799194,
      "learning_rate": 0.0001400541788940027,
      "loss": 2.0788,
      "step": 156442
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.282094955444336,
      "learning_rate": 0.0001400507182155274,
      "loss": 2.0882,
      "step": 156443
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.0565567016601562,
      "learning_rate": 0.0001400472575667898,
      "loss": 1.9987,
      "step": 156444
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.1991097927093506,
      "learning_rate": 0.00014004379694779046,
      "loss": 2.184,
      "step": 156445
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.1205366849899292,
      "learning_rate": 0.00014004033635852997,
      "loss": 2.0158,
      "step": 156446
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.2240278720855713,
      "learning_rate": 0.00014003687579900893,
      "loss": 2.0923,
      "step": 156447
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.2359462976455688,
      "learning_rate": 0.00014003341526922815,
      "loss": 2.0353,
      "step": 156448
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.328809380531311,
      "learning_rate": 0.00014002995476918806,
      "loss": 1.9959,
      "step": 156449
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.3230466842651367,
      "learning_rate": 0.00014002649429888952,
      "loss": 2.211,
      "step": 156450
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.0846134424209595,
      "learning_rate": 0.0001400230338583331,
      "loss": 1.9812,
      "step": 156451
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.6562864780426025,
      "learning_rate": 0.0001400195734475194,
      "loss": 2.0766,
      "step": 156452
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.1673821210861206,
      "learning_rate": 0.00014001611306644897,
      "loss": 2.073,
      "step": 156453
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.1748466491699219,
      "learning_rate": 0.0001400126527151227,
      "loss": 2.0421,
      "step": 156454
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.146917700767517,
      "learning_rate": 0.00014000919239354095,
      "loss": 2.2983,
      "step": 156455
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.1341419219970703,
      "learning_rate": 0.00014000573210170466,
      "loss": 2.0337,
      "step": 156456
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.1914228200912476,
      "learning_rate": 0.00014000227183961418,
      "loss": 2.0611,
      "step": 156457
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.268841028213501,
      "learning_rate": 0.00013999881160727052,
      "loss": 2.0913,
      "step": 156458
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.1804782152175903,
      "learning_rate": 0.0001399953514046739,
      "loss": 1.8261,
      "step": 156459
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.3389060497283936,
      "learning_rate": 0.00013999189123182527,
      "loss": 2.1422,
      "step": 156460
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.075276255607605,
      "learning_rate": 0.00013998843108872506,
      "loss": 2.0721,
      "step": 156461
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.1676039695739746,
      "learning_rate": 0.0001399849709753741,
      "loss": 2.032,
      "step": 156462
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.1033241748809814,
      "learning_rate": 0.00013998151089177288,
      "loss": 2.1694,
      "step": 156463
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.2486169338226318,
      "learning_rate": 0.00013997805083792235,
      "loss": 1.9146,
      "step": 156464
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.1499186754226685,
      "learning_rate": 0.00013997459081382268,
      "loss": 1.8844,
      "step": 156465
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.2143466472625732,
      "learning_rate": 0.00013997113081947485,
      "loss": 2.244,
      "step": 156466
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.3094127178192139,
      "learning_rate": 0.00013996767085487933,
      "loss": 2.1964,
      "step": 156467
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.1382304430007935,
      "learning_rate": 0.00013996421092003693,
      "loss": 2.1109,
      "step": 156468
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.3593565225601196,
      "learning_rate": 0.0001399607510149481,
      "loss": 2.3041,
      "step": 156469
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.3535293340682983,
      "learning_rate": 0.0001399572911396138,
      "loss": 2.0828,
      "step": 156470
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.1618729829788208,
      "learning_rate": 0.00013995383129403427,
      "loss": 2.0977,
      "step": 156471
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.1240051984786987,
      "learning_rate": 0.00013995037147821038,
      "loss": 1.7008,
      "step": 156472
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.160193681716919,
      "learning_rate": 0.00013994691169214268,
      "loss": 2.1804,
      "step": 156473
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.1475813388824463,
      "learning_rate": 0.000139943451935832,
      "loss": 2.0532,
      "step": 156474
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.0833022594451904,
      "learning_rate": 0.0001399399922092787,
      "loss": 2.0399,
      "step": 156475
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.1565560102462769,
      "learning_rate": 0.00013993653251248372,
      "loss": 1.9579,
      "step": 156476
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.1908293962478638,
      "learning_rate": 0.00013993307284544755,
      "loss": 2.0638,
      "step": 156477
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.166230320930481,
      "learning_rate": 0.00013992961320817083,
      "loss": 2.0443,
      "step": 156478
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.0925971269607544,
      "learning_rate": 0.00013992615360065413,
      "loss": 2.0156,
      "step": 156479
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.191961646080017,
      "learning_rate": 0.00013992269402289827,
      "loss": 1.9394,
      "step": 156480
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.1153011322021484,
      "learning_rate": 0.00013991923447490367,
      "loss": 1.9404,
      "step": 156481
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.0969808101654053,
      "learning_rate": 0.00013991577495667127,
      "loss": 1.9033,
      "step": 156482
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.0749540328979492,
      "learning_rate": 0.00013991231546820154,
      "loss": 1.9091,
      "step": 156483
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.097703218460083,
      "learning_rate": 0.00013990885600949512,
      "loss": 1.7574,
      "step": 156484
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.248616337776184,
      "learning_rate": 0.00013990539658055253,
      "loss": 1.9334,
      "step": 156485
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.1736059188842773,
      "learning_rate": 0.0001399019371813747,
      "loss": 2.1373,
      "step": 156486
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.2311551570892334,
      "learning_rate": 0.000139898477811962,
      "loss": 2.3219,
      "step": 156487
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.1459611654281616,
      "learning_rate": 0.00013989501847231527,
      "loss": 2.191,
      "step": 156488
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.2619576454162598,
      "learning_rate": 0.00013989155916243508,
      "loss": 2.1951,
      "step": 156489
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.4846456050872803,
      "learning_rate": 0.00013988809988232213,
      "loss": 2.1479,
      "step": 156490
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.1888623237609863,
      "learning_rate": 0.00013988464063197686,
      "loss": 2.0703,
      "step": 156491
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.3662641048431396,
      "learning_rate": 0.00013988118141140018,
      "loss": 2.2048,
      "step": 156492
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.2960045337677002,
      "learning_rate": 0.00013987772222059247,
      "loss": 1.9028,
      "step": 156493
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.468876600265503,
      "learning_rate": 0.00013987426305955467,
      "loss": 2.0932,
      "step": 156494
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.1649134159088135,
      "learning_rate": 0.00013987080392828725,
      "loss": 2.1448,
      "step": 156495
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.2384017705917358,
      "learning_rate": 0.00013986734482679083,
      "loss": 2.0394,
      "step": 156496
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.3455767631530762,
      "learning_rate": 0.00013986388575506603,
      "loss": 2.0307,
      "step": 156497
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.2995129823684692,
      "learning_rate": 0.00013986042671311366,
      "loss": 2.1931,
      "step": 156498
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.1992088556289673,
      "learning_rate": 0.00013985696770093414,
      "loss": 1.9644,
      "step": 156499
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.1718769073486328,
      "learning_rate": 0.00013985350871852833,
      "loss": 1.9509,
      "step": 156500
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.2174595594406128,
      "learning_rate": 0.00013985004976589682,
      "loss": 1.9954,
      "step": 156501
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.468308448791504,
      "learning_rate": 0.00013984659084304013,
      "loss": 2.0602,
      "step": 156502
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.275273323059082,
      "learning_rate": 0.00013984313194995894,
      "loss": 1.8624,
      "step": 156503
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.2955408096313477,
      "learning_rate": 0.00013983967308665404,
      "loss": 2.2383,
      "step": 156504
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.2113516330718994,
      "learning_rate": 0.00013983621425312583,
      "loss": 2.0595,
      "step": 156505
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.2452478408813477,
      "learning_rate": 0.0001398327554493752,
      "loss": 1.9765,
      "step": 156506
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.567997932434082,
      "learning_rate": 0.0001398292966754027,
      "loss": 1.9334,
      "step": 156507
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.3062610626220703,
      "learning_rate": 0.00013982583793120895,
      "loss": 1.7632,
      "step": 156508
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.209852933883667,
      "learning_rate": 0.00013982237921679453,
      "loss": 2.189,
      "step": 156509
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.157272458076477,
      "learning_rate": 0.00013981892053216023,
      "loss": 2.0578,
      "step": 156510
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.4714100360870361,
      "learning_rate": 0.00013981546187730648,
      "loss": 2.0264,
      "step": 156511
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.2585394382476807,
      "learning_rate": 0.0001398120032522342,
      "loss": 1.9859,
      "step": 156512
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.1329439878463745,
      "learning_rate": 0.00013980854465694392,
      "loss": 2.1422,
      "step": 156513
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.121399998664856,
      "learning_rate": 0.00013980508609143618,
      "loss": 2.1986,
      "step": 156514
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.3578073978424072,
      "learning_rate": 0.00013980162755571166,
      "loss": 1.938,
      "step": 156515
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.5970652103424072,
      "learning_rate": 0.00013979816904977112,
      "loss": 2.0457,
      "step": 156516
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.240751028060913,
      "learning_rate": 0.00013979471057361503,
      "loss": 1.9141,
      "step": 156517
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.2227593660354614,
      "learning_rate": 0.0001397912521272442,
      "loss": 2.108,
      "step": 156518
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.0962953567504883,
      "learning_rate": 0.00013978779371065923,
      "loss": 1.9689,
      "step": 156519
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.2775003910064697,
      "learning_rate": 0.0001397843353238606,
      "loss": 2.0423,
      "step": 156520
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.2342190742492676,
      "learning_rate": 0.00013978087696684923,
      "loss": 1.9264,
      "step": 156521
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.2778741121292114,
      "learning_rate": 0.00013977741863962562,
      "loss": 1.9557,
      "step": 156522
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.0405446290969849,
      "learning_rate": 0.00013977396034219028,
      "loss": 1.9684,
      "step": 156523
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.2202537059783936,
      "learning_rate": 0.0001397705020745441,
      "loss": 2.1243,
      "step": 156524
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.0897878408432007,
      "learning_rate": 0.00013976704383668762,
      "loss": 2.0233,
      "step": 156525
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.3216389417648315,
      "learning_rate": 0.00013976358562862133,
      "loss": 2.211,
      "step": 156526
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.258119821548462,
      "learning_rate": 0.00013976012745034615,
      "loss": 2.2223,
      "step": 156527
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.0979948043823242,
      "learning_rate": 0.0001397566693018625,
      "loss": 2.0725,
      "step": 156528
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.2608333826065063,
      "learning_rate": 0.00013975321118317116,
      "loss": 2.0331,
      "step": 156529
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.1272883415222168,
      "learning_rate": 0.00013974975309427275,
      "loss": 2.0554,
      "step": 156530
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.1334500312805176,
      "learning_rate": 0.00013974629503516793,
      "loss": 1.8802,
      "step": 156531
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.18997061252594,
      "learning_rate": 0.0001397428370058571,
      "loss": 1.9625,
      "step": 156532
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.1924939155578613,
      "learning_rate": 0.00013973937900634128,
      "loss": 1.8988,
      "step": 156533
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.1324090957641602,
      "learning_rate": 0.00013973592103662082,
      "loss": 1.9501,
      "step": 156534
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.2320191860198975,
      "learning_rate": 0.00013973246309669657,
      "loss": 2.2545,
      "step": 156535
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.1283546686172485,
      "learning_rate": 0.0001397290051865691,
      "loss": 2.1694,
      "step": 156536
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.1298290491104126,
      "learning_rate": 0.000139725547306239,
      "loss": 1.9716,
      "step": 156537
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.0587679147720337,
      "learning_rate": 0.00013972208945570684,
      "loss": 1.9668,
      "step": 156538
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.2365005016326904,
      "learning_rate": 0.0001397186316349735,
      "loss": 2.2492,
      "step": 156539
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.0864306688308716,
      "learning_rate": 0.0001397151738440394,
      "loss": 1.7964,
      "step": 156540
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.0819077491760254,
      "learning_rate": 0.00013971171608290534,
      "loss": 1.9766,
      "step": 156541
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.1168862581253052,
      "learning_rate": 0.00013970825835157182,
      "loss": 2.1613,
      "step": 156542
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.1704115867614746,
      "learning_rate": 0.00013970480065003964,
      "loss": 2.1684,
      "step": 156543
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.1280639171600342,
      "learning_rate": 0.00013970134297830938,
      "loss": 2.1155,
      "step": 156544
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.1468807458877563,
      "learning_rate": 0.00013969788533638167,
      "loss": 1.9854,
      "step": 156545
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.2248952388763428,
      "learning_rate": 0.00013969442772425702,
      "loss": 2.3681,
      "step": 156546
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.2005281448364258,
      "learning_rate": 0.00013969097014193633,
      "loss": 1.9007,
      "step": 156547
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.2251540422439575,
      "learning_rate": 0.00013968751258942,
      "loss": 2.2579,
      "step": 156548
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.0733839273452759,
      "learning_rate": 0.00013968405506670892,
      "loss": 1.9837,
      "step": 156549
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.213058590888977,
      "learning_rate": 0.00013968059757380355,
      "loss": 2.0618,
      "step": 156550
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.2866694927215576,
      "learning_rate": 0.00013967714011070464,
      "loss": 2.1094,
      "step": 156551
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.2454723119735718,
      "learning_rate": 0.00013967368267741264,
      "loss": 2.1517,
      "step": 156552
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.1875243186950684,
      "learning_rate": 0.00013967022527392844,
      "loss": 2.2517,
      "step": 156553
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.1708430051803589,
      "learning_rate": 0.00013966676790025246,
      "loss": 2.0267,
      "step": 156554
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.3189457654953003,
      "learning_rate": 0.00013966331055638556,
      "loss": 1.9674,
      "step": 156555
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.175864815711975,
      "learning_rate": 0.00013965985324232829,
      "loss": 2.0654,
      "step": 156556
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.1100953817367554,
      "learning_rate": 0.00013965639595808126,
      "loss": 2.1392,
      "step": 156557
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.1492220163345337,
      "learning_rate": 0.00013965293870364504,
      "loss": 1.8365,
      "step": 156558
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.0670039653778076,
      "learning_rate": 0.00013964948147902046,
      "loss": 2.2591,
      "step": 156559
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.249610424041748,
      "learning_rate": 0.00013964602428420797,
      "loss": 2.1922,
      "step": 156560
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.0778443813323975,
      "learning_rate": 0.00013964256711920844,
      "loss": 2.0012,
      "step": 156561
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.0700880289077759,
      "learning_rate": 0.00013963910998402235,
      "loss": 2.0835,
      "step": 156562
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.2116109132766724,
      "learning_rate": 0.0001396356528786504,
      "loss": 2.038,
      "step": 156563
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.1094847917556763,
      "learning_rate": 0.00013963219580309308,
      "loss": 2.0295,
      "step": 156564
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.1701312065124512,
      "learning_rate": 0.00013962873875735129,
      "loss": 1.9557,
      "step": 156565
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.1584115028381348,
      "learning_rate": 0.00013962528174142543,
      "loss": 1.8878,
      "step": 156566
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.28248131275177,
      "learning_rate": 0.00013962182475531635,
      "loss": 2.0966,
      "step": 156567
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.1414387226104736,
      "learning_rate": 0.00013961836779902463,
      "loss": 2.0042,
      "step": 156568
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.3053700923919678,
      "learning_rate": 0.00013961491087255084,
      "loss": 1.9967,
      "step": 156569
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.1663198471069336,
      "learning_rate": 0.00013961145397589557,
      "loss": 2.2965,
      "step": 156570
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.1369707584381104,
      "learning_rate": 0.0001396079971090597,
      "loss": 1.7864,
      "step": 156571
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.3099839687347412,
      "learning_rate": 0.0001396045402720436,
      "loss": 2.1792,
      "step": 156572
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.1929900646209717,
      "learning_rate": 0.00013960108346484815,
      "loss": 2.019,
      "step": 156573
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.2088156938552856,
      "learning_rate": 0.0001395976266874739,
      "loss": 2.2345,
      "step": 156574
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.1484293937683105,
      "learning_rate": 0.00013959416993992147,
      "loss": 1.9879,
      "step": 156575
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.077317714691162,
      "learning_rate": 0.0001395907132221914,
      "loss": 1.9142,
      "step": 156576
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.1472526788711548,
      "learning_rate": 0.00013958725653428456,
      "loss": 2.0883,
      "step": 156577
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.171021819114685,
      "learning_rate": 0.00013958379987620136,
      "loss": 2.0081,
      "step": 156578
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.2193700075149536,
      "learning_rate": 0.00013958034324794267,
      "loss": 2.0842,
      "step": 156579
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.1204802989959717,
      "learning_rate": 0.00013957688664950903,
      "loss": 2.1201,
      "step": 156580
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.1675609350204468,
      "learning_rate": 0.00013957343008090106,
      "loss": 2.1469,
      "step": 156581
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.1856962442398071,
      "learning_rate": 0.00013956997354211932,
      "loss": 1.9545,
      "step": 156582
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.269242525100708,
      "learning_rate": 0.00013956651703316467,
      "loss": 2.0723,
      "step": 156583
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.1785781383514404,
      "learning_rate": 0.0001395630605540375,
      "loss": 2.052,
      "step": 156584
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.0966805219650269,
      "learning_rate": 0.0001395596041047387,
      "loss": 2.0895,
      "step": 156585
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.1565335988998413,
      "learning_rate": 0.0001395561476852688,
      "loss": 2.2358,
      "step": 156586
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.1884509325027466,
      "learning_rate": 0.00013955269129562847,
      "loss": 1.9769,
      "step": 156587
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.132812261581421,
      "learning_rate": 0.00013954923493581816,
      "loss": 1.9862,
      "step": 156588
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.1805088520050049,
      "learning_rate": 0.0001395457786058388,
      "loss": 1.8155,
      "step": 156589
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.13912034034729,
      "learning_rate": 0.00013954232230569083,
      "loss": 2.0706,
      "step": 156590
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.1495648622512817,
      "learning_rate": 0.00013953886603537506,
      "loss": 2.0145,
      "step": 156591
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.1393380165100098,
      "learning_rate": 0.00013953540979489207,
      "loss": 1.8764,
      "step": 156592
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.2207890748977661,
      "learning_rate": 0.00013953195358424244,
      "loss": 1.8571,
      "step": 156593
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.3158180713653564,
      "learning_rate": 0.00013952849740342675,
      "loss": 2.2023,
      "step": 156594
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.2342453002929688,
      "learning_rate": 0.00013952504125244587,
      "loss": 2.1478,
      "step": 156595
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.1444180011749268,
      "learning_rate": 0.0001395215851313002,
      "loss": 2.2142,
      "step": 156596
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.234843373298645,
      "learning_rate": 0.0001395181290399906,
      "loss": 2.3765,
      "step": 156597
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.1801011562347412,
      "learning_rate": 0.0001395146729785176,
      "loss": 1.8995,
      "step": 156598
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.1138643026351929,
      "learning_rate": 0.00013951121694688183,
      "loss": 2.1843,
      "step": 156599
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.2099963426589966,
      "learning_rate": 0.00013950776094508386,
      "loss": 1.9821,
      "step": 156600
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.1770188808441162,
      "learning_rate": 0.00013950430497312457,
      "loss": 2.3849,
      "step": 156601
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.0884008407592773,
      "learning_rate": 0.0001395008490310043,
      "loss": 2.0123,
      "step": 156602
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.187384843826294,
      "learning_rate": 0.000139497393118724,
      "loss": 2.0335,
      "step": 156603
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.3802634477615356,
      "learning_rate": 0.00013949393723628416,
      "loss": 1.8746,
      "step": 156604
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.0915160179138184,
      "learning_rate": 0.00013949048138368528,
      "loss": 1.9898,
      "step": 156605
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.1120764017105103,
      "learning_rate": 0.0001394870255609283,
      "loss": 1.9368,
      "step": 156606
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.2603778839111328,
      "learning_rate": 0.0001394835697680137,
      "loss": 2.1323,
      "step": 156607
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.1732667684555054,
      "learning_rate": 0.00013948011400494203,
      "loss": 1.9169,
      "step": 156608
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.0748066902160645,
      "learning_rate": 0.00013947665827171416,
      "loss": 2.1153,
      "step": 156609
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.1153314113616943,
      "learning_rate": 0.00013947320256833047,
      "loss": 2.1072,
      "step": 156610
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.1713268756866455,
      "learning_rate": 0.00013946974689479187,
      "loss": 2.1351,
      "step": 156611
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.2252928018569946,
      "learning_rate": 0.00013946629125109885,
      "loss": 1.9847,
      "step": 156612
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.2950713634490967,
      "learning_rate": 0.00013946283563725213,
      "loss": 2.1194,
      "step": 156613
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.238121747970581,
      "learning_rate": 0.00013945938005325214,
      "loss": 2.055,
      "step": 156614
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.6424514055252075,
      "learning_rate": 0.00013945592449909978,
      "loss": 2.0719,
      "step": 156615
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.3470278978347778,
      "learning_rate": 0.00013945246897479553,
      "loss": 1.912,
      "step": 156616
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.2694463729858398,
      "learning_rate": 0.00013944901348034018,
      "loss": 2.2733,
      "step": 156617
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.4180344343185425,
      "learning_rate": 0.0001394455580157343,
      "loss": 2.0356,
      "step": 156618
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.1835511922836304,
      "learning_rate": 0.00013944210258097842,
      "loss": 1.8951,
      "step": 156619
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.1778780221939087,
      "learning_rate": 0.00013943864717607337,
      "loss": 2.102,
      "step": 156620
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.1606804132461548,
      "learning_rate": 0.0001394351918010197,
      "loss": 2.0955,
      "step": 156621
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.1089571714401245,
      "learning_rate": 0.00013943173645581796,
      "loss": 2.0058,
      "step": 156622
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.1508440971374512,
      "learning_rate": 0.000139428281140469,
      "loss": 2.1065,
      "step": 156623
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.193993091583252,
      "learning_rate": 0.00013942482585497338,
      "loss": 1.8139,
      "step": 156624
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.1810033321380615,
      "learning_rate": 0.00013942137059933158,
      "loss": 1.9849,
      "step": 156625
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.0695587396621704,
      "learning_rate": 0.00013941791537354448,
      "loss": 1.9959,
      "step": 156626
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.2096928358078003,
      "learning_rate": 0.00013941446017761252,
      "loss": 2.1424,
      "step": 156627
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.2280244827270508,
      "learning_rate": 0.00013941100501153658,
      "loss": 1.9887,
      "step": 156628
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.2124196290969849,
      "learning_rate": 0.00013940754987531716,
      "loss": 2.1776,
      "step": 156629
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.2154808044433594,
      "learning_rate": 0.00013940409476895487,
      "loss": 2.051,
      "step": 156630
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.1815946102142334,
      "learning_rate": 0.0001394006396924503,
      "loss": 2.0314,
      "step": 156631
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.7209545373916626,
      "learning_rate": 0.0001393971846458043,
      "loss": 2.1002,
      "step": 156632
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.284866213798523,
      "learning_rate": 0.00013939372962901728,
      "loss": 1.9589,
      "step": 156633
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.0937443971633911,
      "learning_rate": 0.00013939027464209008,
      "loss": 2.0749,
      "step": 156634
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.2728668451309204,
      "learning_rate": 0.00013938681968502333,
      "loss": 2.0442,
      "step": 156635
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.4250456094741821,
      "learning_rate": 0.00013938336475781754,
      "loss": 1.9261,
      "step": 156636
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.2842121124267578,
      "learning_rate": 0.0001393799098604733,
      "loss": 1.954,
      "step": 156637
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.0824024677276611,
      "learning_rate": 0.00013937645499299153,
      "loss": 1.9738,
      "step": 156638
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.233228325843811,
      "learning_rate": 0.00013937300015537255,
      "loss": 1.9002,
      "step": 156639
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.23141348361969,
      "learning_rate": 0.00013936954534761727,
      "loss": 1.9032,
      "step": 156640
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.349801778793335,
      "learning_rate": 0.00013936609056972629,
      "loss": 2.0083,
      "step": 156641
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.2228193283081055,
      "learning_rate": 0.00013936263582170014,
      "loss": 2.1583,
      "step": 156642
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.2068841457366943,
      "learning_rate": 0.0001393591811035394,
      "loss": 2.0196,
      "step": 156643
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.129223346710205,
      "learning_rate": 0.00013935572641524493,
      "loss": 1.8571,
      "step": 156644
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.32637619972229,
      "learning_rate": 0.0001393522717568171,
      "loss": 1.9237,
      "step": 156645
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.1698393821716309,
      "learning_rate": 0.0001393488171282569,
      "loss": 2.0256,
      "step": 156646
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.0511260032653809,
      "learning_rate": 0.00013934536252956477,
      "loss": 1.8685,
      "step": 156647
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.206870198249817,
      "learning_rate": 0.00013934190796074136,
      "loss": 2.2063,
      "step": 156648
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.2459683418273926,
      "learning_rate": 0.0001393384534217872,
      "loss": 1.9413,
      "step": 156649
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.1933043003082275,
      "learning_rate": 0.00013933499891270322,
      "loss": 2.2514,
      "step": 156650
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.161844253540039,
      "learning_rate": 0.00013933154443348972,
      "loss": 2.0582,
      "step": 156651
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.2474435567855835,
      "learning_rate": 0.00013932808998414766,
      "loss": 1.9426,
      "step": 156652
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.298567533493042,
      "learning_rate": 0.00013932463556467754,
      "loss": 2.2046,
      "step": 156653
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.46568763256073,
      "learning_rate": 0.00013932118117508,
      "loss": 1.8709,
      "step": 156654
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.1545579433441162,
      "learning_rate": 0.0001393177268153556,
      "loss": 2.1,
      "step": 156655
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.2508732080459595,
      "learning_rate": 0.0001393142724855051,
      "loss": 2.0481,
      "step": 156656
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.2031521797180176,
      "learning_rate": 0.00013931081818552906,
      "loss": 2.2131,
      "step": 156657
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.4401612281799316,
      "learning_rate": 0.00013930736391542828,
      "loss": 2.1441,
      "step": 156658
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.1850559711456299,
      "learning_rate": 0.0001393039096752033,
      "loss": 2.1293,
      "step": 156659
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.1656842231750488,
      "learning_rate": 0.00013930045546485474,
      "loss": 2.3483,
      "step": 156660
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.1444283723831177,
      "learning_rate": 0.00013929700128438314,
      "loss": 1.9623,
      "step": 156661
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.1784756183624268,
      "learning_rate": 0.0001392935471337894,
      "loss": 1.8332,
      "step": 156662
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.3988808393478394,
      "learning_rate": 0.0001392900930130739,
      "loss": 2.3248,
      "step": 156663
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.083243727684021,
      "learning_rate": 0.00013928663892223749,
      "loss": 1.9067,
      "step": 156664
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.132359504699707,
      "learning_rate": 0.00013928318486128077,
      "loss": 2.025,
      "step": 156665
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.1050477027893066,
      "learning_rate": 0.00013927973083020432,
      "loss": 1.82,
      "step": 156666
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.1589971780776978,
      "learning_rate": 0.0001392762768290087,
      "loss": 2.0612,
      "step": 156667
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.239527702331543,
      "learning_rate": 0.00013927282285769473,
      "loss": 2.2214,
      "step": 156668
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.1184643507003784,
      "learning_rate": 0.0001392693689162629,
      "loss": 2.1617,
      "step": 156669
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.1410714387893677,
      "learning_rate": 0.00013926591500471402,
      "loss": 2.2574,
      "step": 156670
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.1285207271575928,
      "learning_rate": 0.00013926246112304868,
      "loss": 1.9947,
      "step": 156671
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.1187115907669067,
      "learning_rate": 0.00013925900727126744,
      "loss": 2.05,
      "step": 156672
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.139037847518921,
      "learning_rate": 0.0001392555534493709,
      "loss": 2.0839,
      "step": 156673
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.2831681966781616,
      "learning_rate": 0.0001392520996573599,
      "loss": 2.07,
      "step": 156674
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.1735409498214722,
      "learning_rate": 0.00013924864589523483,
      "loss": 1.9776,
      "step": 156675
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.129179835319519,
      "learning_rate": 0.0001392451921629966,
      "loss": 2.23,
      "step": 156676
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.1294026374816895,
      "learning_rate": 0.00013924173846064557,
      "loss": 2.0781,
      "step": 156677
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.2243512868881226,
      "learning_rate": 0.00013923828478818283,
      "loss": 2.0521,
      "step": 156678
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.1262638568878174,
      "learning_rate": 0.00013923483114560846,
      "loss": 2.0407,
      "step": 156679
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.3308640718460083,
      "learning_rate": 0.0001392313775329235,
      "loss": 1.9794,
      "step": 156680
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.0928846597671509,
      "learning_rate": 0.00013922792395012833,
      "loss": 2.0171,
      "step": 156681
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.299318790435791,
      "learning_rate": 0.00013922447039722385,
      "loss": 2.1131,
      "step": 156682
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.1401578187942505,
      "learning_rate": 0.00013922101687421047,
      "loss": 2.2207,
      "step": 156683
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.1085933446884155,
      "learning_rate": 0.00013921756338108914,
      "loss": 2.1836,
      "step": 156684
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.133041262626648,
      "learning_rate": 0.00013921410991786008,
      "loss": 2.1355,
      "step": 156685
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.20174241065979,
      "learning_rate": 0.00013921065648452428,
      "loss": 1.8733,
      "step": 156686
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.1586899757385254,
      "learning_rate": 0.00013920720308108214,
      "loss": 2.1112,
      "step": 156687
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.176427960395813,
      "learning_rate": 0.0001392037497075345,
      "loss": 2.0593,
      "step": 156688
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.2158046960830688,
      "learning_rate": 0.00013920029636388184,
      "loss": 2.214,
      "step": 156689
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.295591115951538,
      "learning_rate": 0.000139196843050125,
      "loss": 2.2064,
      "step": 156690
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.3094067573547363,
      "learning_rate": 0.0001391933897662645,
      "loss": 2.0137,
      "step": 156691
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.1376193761825562,
      "learning_rate": 0.00013918993651230093,
      "loss": 2.1417,
      "step": 156692
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.3230959177017212,
      "learning_rate": 0.00013918648328823494,
      "loss": 1.9615,
      "step": 156693
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.2437063455581665,
      "learning_rate": 0.0001391830300940673,
      "loss": 2.2261,
      "step": 156694
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.2333694696426392,
      "learning_rate": 0.00013917957692979847,
      "loss": 2.0884,
      "step": 156695
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.4886775016784668,
      "learning_rate": 0.0001391761237954293,
      "loss": 1.9807,
      "step": 156696
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.15876042842865,
      "learning_rate": 0.0001391726706909603,
      "loss": 2.2896,
      "step": 156697
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.183458924293518,
      "learning_rate": 0.0001391692176163922,
      "loss": 2.0885,
      "step": 156698
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.157285213470459,
      "learning_rate": 0.0001391657645717254,
      "loss": 2.1141,
      "step": 156699
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.3236429691314697,
      "learning_rate": 0.00013916231155696086,
      "loss": 1.8952,
      "step": 156700
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.474424123764038,
      "learning_rate": 0.00013915885857209895,
      "loss": 2.2374,
      "step": 156701
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.2550115585327148,
      "learning_rate": 0.00013915540561714056,
      "loss": 2.1408,
      "step": 156702
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.323608160018921,
      "learning_rate": 0.00013915195269208625,
      "loss": 2.1779,
      "step": 156703
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.2510744333267212,
      "learning_rate": 0.00013914849979693647,
      "loss": 2.0786,
      "step": 156704
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.3226561546325684,
      "learning_rate": 0.00013914504693169217,
      "loss": 2.1647,
      "step": 156705
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.0841953754425049,
      "learning_rate": 0.00013914159409635383,
      "loss": 1.996,
      "step": 156706
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.2408638000488281,
      "learning_rate": 0.000139138141290922,
      "loss": 1.9913,
      "step": 156707
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.1505827903747559,
      "learning_rate": 0.00013913468851539751,
      "loss": 2.0514,
      "step": 156708
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.2268837690353394,
      "learning_rate": 0.00013913123576978097,
      "loss": 1.7618,
      "step": 156709
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.2591451406478882,
      "learning_rate": 0.00013912778305407282,
      "loss": 2.1772,
      "step": 156710
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.193216323852539,
      "learning_rate": 0.00013912433036827396,
      "loss": 1.9622,
      "step": 156711
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.2571628093719482,
      "learning_rate": 0.00013912087771238482,
      "loss": 2.2808,
      "step": 156712
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.1763650178909302,
      "learning_rate": 0.00013911742508640625,
      "loss": 1.9063,
      "step": 156713
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.2035154104232788,
      "learning_rate": 0.0001391139724903388,
      "loss": 2.2424,
      "step": 156714
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.2286816835403442,
      "learning_rate": 0.0001391105199241831,
      "loss": 1.6476,
      "step": 156715
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.3308465480804443,
      "learning_rate": 0.00013910706738793965,
      "loss": 2.2653,
      "step": 156716
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.1924645900726318,
      "learning_rate": 0.00013910361488160937,
      "loss": 1.8376,
      "step": 156717
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.2300186157226562,
      "learning_rate": 0.0001391001624051927,
      "loss": 1.9261,
      "step": 156718
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.1355476379394531,
      "learning_rate": 0.00013909670995869037,
      "loss": 2.0675,
      "step": 156719
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.1319979429244995,
      "learning_rate": 0.00013909325754210306,
      "loss": 2.0969,
      "step": 156720
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.1676387786865234,
      "learning_rate": 0.00013908980515543132,
      "loss": 1.9287,
      "step": 156721
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.1473510265350342,
      "learning_rate": 0.00013908635279867573,
      "loss": 2.1176,
      "step": 156722
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.291231393814087,
      "learning_rate": 0.00013908290047183713,
      "loss": 2.1191,
      "step": 156723
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.2492799758911133,
      "learning_rate": 0.00013907944817491592,
      "loss": 2.2392,
      "step": 156724
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.076176404953003,
      "learning_rate": 0.00013907599590791303,
      "loss": 2.1041,
      "step": 156725
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.147463083267212,
      "learning_rate": 0.00013907254367082894,
      "loss": 1.7302,
      "step": 156726
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.1762495040893555,
      "learning_rate": 0.0001390690914636643,
      "loss": 1.9686,
      "step": 156727
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.192279577255249,
      "learning_rate": 0.00013906563928641963,
      "loss": 2.1319,
      "step": 156728
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.3142343759536743,
      "learning_rate": 0.00013906218713909582,
      "loss": 2.2998,
      "step": 156729
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.2157866954803467,
      "learning_rate": 0.00013905873502169326,
      "loss": 2.2251,
      "step": 156730
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.174418330192566,
      "learning_rate": 0.00013905528293421286,
      "loss": 1.8847,
      "step": 156731
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.1896235942840576,
      "learning_rate": 0.00013905183087665508,
      "loss": 1.9997,
      "step": 156732
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.221389889717102,
      "learning_rate": 0.00013904837884902064,
      "loss": 2.1011,
      "step": 156733
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.104488492012024,
      "learning_rate": 0.00013904492685130998,
      "loss": 2.3004,
      "step": 156734
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.2746580839157104,
      "learning_rate": 0.00013904147488352406,
      "loss": 1.8452,
      "step": 156735
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.1989939212799072,
      "learning_rate": 0.0001390380229456633,
      "loss": 2.3643,
      "step": 156736
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.2225035429000854,
      "learning_rate": 0.00013903457103772845,
      "loss": 1.9619,
      "step": 156737
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.1406716108322144,
      "learning_rate": 0.00013903111915972014,
      "loss": 2.044,
      "step": 156738
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.2695190906524658,
      "learning_rate": 0.00013902766731163895,
      "loss": 2.2354,
      "step": 156739
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.1897246837615967,
      "learning_rate": 0.00013902421549348549,
      "loss": 2.0913,
      "step": 156740
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.2094767093658447,
      "learning_rate": 0.00013902076370526055,
      "loss": 1.9123,
      "step": 156741
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.2017611265182495,
      "learning_rate": 0.0001390173119469646,
      "loss": 2.0743,
      "step": 156742
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.161483645439148,
      "learning_rate": 0.00013901386021859848,
      "loss": 2.0545,
      "step": 156743
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.1333225965499878,
      "learning_rate": 0.0001390104085201626,
      "loss": 1.9847,
      "step": 156744
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.1603325605392456,
      "learning_rate": 0.00013900695685165794,
      "loss": 1.8722,
      "step": 156745
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.2056055068969727,
      "learning_rate": 0.00013900350521308473,
      "loss": 1.9003,
      "step": 156746
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.2726194858551025,
      "learning_rate": 0.00013900005360444387,
      "loss": 1.894,
      "step": 156747
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.1670435667037964,
      "learning_rate": 0.00013899660202573586,
      "loss": 1.9784,
      "step": 156748
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.2844668626785278,
      "learning_rate": 0.00013899315047696155,
      "loss": 2.012,
      "step": 156749
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.2433421611785889,
      "learning_rate": 0.00013898969895812131,
      "loss": 2.2005,
      "step": 156750
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.1418286561965942,
      "learning_rate": 0.00013898624746921617,
      "loss": 2.2135,
      "step": 156751
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.2197794914245605,
      "learning_rate": 0.00013898279601024628,
      "loss": 2.1067,
      "step": 156752
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.228223204612732,
      "learning_rate": 0.00013897934458121267,
      "loss": 1.9666,
      "step": 156753
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.1076233386993408,
      "learning_rate": 0.00013897589318211567,
      "loss": 2.0211,
      "step": 156754
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.4372682571411133,
      "learning_rate": 0.00013897244181295627,
      "loss": 1.8872,
      "step": 156755
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.1846387386322021,
      "learning_rate": 0.0001389689904737348,
      "loss": 2.1851,
      "step": 156756
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.1778873205184937,
      "learning_rate": 0.00013896553916445222,
      "loss": 2.0859,
      "step": 156757
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.196229338645935,
      "learning_rate": 0.00013896208788510882,
      "loss": 2.1949,
      "step": 156758
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.1218594312667847,
      "learning_rate": 0.00013895863663570548,
      "loss": 2.1832,
      "step": 156759
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.1804077625274658,
      "learning_rate": 0.00013895518541624268,
      "loss": 2.0864,
      "step": 156760
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.4215829372406006,
      "learning_rate": 0.00013895173422672123,
      "loss": 2.2697,
      "step": 156761
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.153030276298523,
      "learning_rate": 0.0001389482830671416,
      "loss": 2.0373,
      "step": 156762
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.2013604640960693,
      "learning_rate": 0.00013894483193750476,
      "loss": 2.0535,
      "step": 156763
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.177579402923584,
      "learning_rate": 0.00013894138083781087,
      "loss": 2.1316,
      "step": 156764
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.1211552619934082,
      "learning_rate": 0.00013893792976806096,
      "loss": 2.1533,
      "step": 156765
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.1350955963134766,
      "learning_rate": 0.00013893447872825537,
      "loss": 1.8927,
      "step": 156766
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.2432429790496826,
      "learning_rate": 0.00013893102771839506,
      "loss": 1.9535,
      "step": 156767
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.2057907581329346,
      "learning_rate": 0.00013892757673848039,
      "loss": 2.1958,
      "step": 156768
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.1615469455718994,
      "learning_rate": 0.00013892412578851235,
      "loss": 2.0204,
      "step": 156769
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.1632238626480103,
      "learning_rate": 0.0001389206748684911,
      "loss": 2.1617,
      "step": 156770
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.2046371698379517,
      "learning_rate": 0.00013891722397841767,
      "loss": 1.8651,
      "step": 156771
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.1643598079681396,
      "learning_rate": 0.00013891377311829245,
      "loss": 1.9264,
      "step": 156772
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.0968024730682373,
      "learning_rate": 0.00013891032228811634,
      "loss": 2.064,
      "step": 156773
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.0882550477981567,
      "learning_rate": 0.0001389068714878897,
      "loss": 1.9657,
      "step": 156774
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.3962962627410889,
      "learning_rate": 0.00013890342071761345,
      "loss": 2.134,
      "step": 156775
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.2201820611953735,
      "learning_rate": 0.00013889996997728805,
      "loss": 2.071,
      "step": 156776
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.1601948738098145,
      "learning_rate": 0.00013889651926691422,
      "loss": 2.0456,
      "step": 156777
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.2281990051269531,
      "learning_rate": 0.00013889306858649246,
      "loss": 1.9488,
      "step": 156778
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.2730090618133545,
      "learning_rate": 0.00013888961793602362,
      "loss": 1.9684,
      "step": 156779
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.1335982084274292,
      "learning_rate": 0.00013888616731550814,
      "loss": 2.1688,
      "step": 156780
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.134519100189209,
      "learning_rate": 0.0001388827167249469,
      "loss": 2.2768,
      "step": 156781
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.216564416885376,
      "learning_rate": 0.00013887926616434033,
      "loss": 2.1233,
      "step": 156782
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.2998124361038208,
      "learning_rate": 0.0001388758156336892,
      "loss": 2.0778,
      "step": 156783
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.1710975170135498,
      "learning_rate": 0.00013887236513299397,
      "loss": 1.9686,
      "step": 156784
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.135440707206726,
      "learning_rate": 0.00013886891466225555,
      "loss": 2.2043,
      "step": 156785
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.1053228378295898,
      "learning_rate": 0.00013886546422147428,
      "loss": 1.8947,
      "step": 156786
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.113607406616211,
      "learning_rate": 0.00013886201381065112,
      "loss": 2.0169,
      "step": 156787
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.102849006652832,
      "learning_rate": 0.00013885856342978653,
      "loss": 1.7639,
      "step": 156788
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.1843829154968262,
      "learning_rate": 0.00013885511307888104,
      "loss": 1.9319,
      "step": 156789
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.202621340751648,
      "learning_rate": 0.00013885166275793556,
      "loss": 1.8679,
      "step": 156790
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.172698736190796,
      "learning_rate": 0.0001388482124669506,
      "loss": 2.3798,
      "step": 156791
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.2979564666748047,
      "learning_rate": 0.00013884476220592668,
      "loss": 2.0825,
      "step": 156792
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.1088879108428955,
      "learning_rate": 0.0001388413119748647,
      "loss": 1.856,
      "step": 156793
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.3409188985824585,
      "learning_rate": 0.00013883786177376514,
      "loss": 2.1821,
      "step": 156794
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.2751317024230957,
      "learning_rate": 0.00013883441160262856,
      "loss": 2.043,
      "step": 156795
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.2397078275680542,
      "learning_rate": 0.0001388309614614558,
      "loss": 2.0092,
      "step": 156796
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.241087794303894,
      "learning_rate": 0.00013882751135024741,
      "loss": 1.9578,
      "step": 156797
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.285827875137329,
      "learning_rate": 0.0001388240612690039,
      "loss": 2.1503,
      "step": 156798
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.1731255054473877,
      "learning_rate": 0.00013882061121772623,
      "loss": 2.1714,
      "step": 156799
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.2309799194335938,
      "learning_rate": 0.0001388171611964148,
      "loss": 2.0627,
      "step": 156800
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.1077708005905151,
      "learning_rate": 0.00013881371120507015,
      "loss": 1.7759,
      "step": 156801
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.1592473983764648,
      "learning_rate": 0.00013881026124369324,
      "loss": 2.0183,
      "step": 156802
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.195185899734497,
      "learning_rate": 0.0001388068113122844,
      "loss": 2.1306,
      "step": 156803
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.2324981689453125,
      "learning_rate": 0.00013880336141084454,
      "loss": 1.9264,
      "step": 156804
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.0793569087982178,
      "learning_rate": 0.0001387999115393742,
      "loss": 2.0486,
      "step": 156805
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.1360342502593994,
      "learning_rate": 0.00013879646169787397,
      "loss": 2.1235,
      "step": 156806
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.1555238962173462,
      "learning_rate": 0.00013879301188634443,
      "loss": 2.0353,
      "step": 156807
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.2147507667541504,
      "learning_rate": 0.0001387895621047864,
      "loss": 2.218,
      "step": 156808
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.2689697742462158,
      "learning_rate": 0.00013878611235320035,
      "loss": 2.0915,
      "step": 156809
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.1922526359558105,
      "learning_rate": 0.00013878266263158708,
      "loss": 2.068,
      "step": 156810
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.0937291383743286,
      "learning_rate": 0.00013877921293994707,
      "loss": 1.9993,
      "step": 156811
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.105924367904663,
      "learning_rate": 0.00013877576327828126,
      "loss": 1.9438,
      "step": 156812
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.1865899562835693,
      "learning_rate": 0.00013877231364658988,
      "loss": 2.0018,
      "step": 156813
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.2950451374053955,
      "learning_rate": 0.00013876886404487387,
      "loss": 2.1504,
      "step": 156814
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.097734808921814,
      "learning_rate": 0.00013876541447313363,
      "loss": 2.0091,
      "step": 156815
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.1511715650558472,
      "learning_rate": 0.00013876196493137007,
      "loss": 2.1471,
      "step": 156816
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.2826616764068604,
      "learning_rate": 0.00013875851541958362,
      "loss": 2.0114,
      "step": 156817
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.238629698753357,
      "learning_rate": 0.0001387550659377752,
      "loss": 2.27,
      "step": 156818
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.1081089973449707,
      "learning_rate": 0.00013875161648594506,
      "loss": 1.9509,
      "step": 156819
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.1555339097976685,
      "learning_rate": 0.00013874816706409415,
      "loss": 2.0791,
      "step": 156820
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.1644164323806763,
      "learning_rate": 0.00013874471767222284,
      "loss": 1.933,
      "step": 156821
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.3662021160125732,
      "learning_rate": 0.0001387412683103321,
      "loss": 2.1751,
      "step": 156822
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.5135408639907837,
      "learning_rate": 0.00013873781897842227,
      "loss": 2.1018,
      "step": 156823
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.1994166374206543,
      "learning_rate": 0.0001387343696764943,
      "loss": 2.1909,
      "step": 156824
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.2290047407150269,
      "learning_rate": 0.0001387309204045485,
      "loss": 2.0178,
      "step": 156825
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.2517180442810059,
      "learning_rate": 0.00013872747116258576,
      "loss": 1.8188,
      "step": 156826
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.1433771848678589,
      "learning_rate": 0.0001387240219506065,
      "loss": 2.0891,
      "step": 156827
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.1553561687469482,
      "learning_rate": 0.0001387205727686116,
      "loss": 2.0587,
      "step": 156828
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.0856380462646484,
      "learning_rate": 0.0001387171236166015,
      "loss": 2.2666,
      "step": 156829
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.235587477684021,
      "learning_rate": 0.0001387136744945771,
      "loss": 2.0622,
      "step": 156830
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.1917897462844849,
      "learning_rate": 0.0001387102254025387,
      "loss": 2.2166,
      "step": 156831
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.2526066303253174,
      "learning_rate": 0.00013870677634048723,
      "loss": 1.8578,
      "step": 156832
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.0941466093063354,
      "learning_rate": 0.00013870332730842307,
      "loss": 1.9101,
      "step": 156833
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.1734473705291748,
      "learning_rate": 0.00013869987830634713,
      "loss": 2.1026,
      "step": 156834
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.1595101356506348,
      "learning_rate": 0.00013869642933425985,
      "loss": 2.0737,
      "step": 156835
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.1900914907455444,
      "learning_rate": 0.00013869298039216214,
      "loss": 1.822,
      "step": 156836
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.2398890256881714,
      "learning_rate": 0.00013868953148005422,
      "loss": 2.1046,
      "step": 156837
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.1926867961883545,
      "learning_rate": 0.0001386860825979371,
      "loss": 1.9754,
      "step": 156838
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.115513801574707,
      "learning_rate": 0.00013868263374581114,
      "loss": 2.0727,
      "step": 156839
    },
    {
      "epoch": 2.04,
      "grad_norm": 2.058074712753296,
      "learning_rate": 0.00013867918492367726,
      "loss": 2.0557,
      "step": 156840
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.225688099861145,
      "learning_rate": 0.00013867573613153582,
      "loss": 2.0617,
      "step": 156841
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.1662325859069824,
      "learning_rate": 0.00013867228736938786,
      "loss": 2.1897,
      "step": 156842
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.7034755945205688,
      "learning_rate": 0.00013866883863723355,
      "loss": 2.0203,
      "step": 156843
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.607214331626892,
      "learning_rate": 0.00013866538993507384,
      "loss": 1.9414,
      "step": 156844
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.2383029460906982,
      "learning_rate": 0.00013866194126290916,
      "loss": 2.1711,
      "step": 156845
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.2030400037765503,
      "learning_rate": 0.0001386584926207404,
      "loss": 2.0923,
      "step": 156846
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.2606685161590576,
      "learning_rate": 0.00013865504400856796,
      "loss": 2.1716,
      "step": 156847
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.2218126058578491,
      "learning_rate": 0.0001386515954263928,
      "loss": 2.1632,
      "step": 156848
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.15437650680542,
      "learning_rate": 0.00013864814687421516,
      "loss": 2.0245,
      "step": 156849
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.2099064588546753,
      "learning_rate": 0.00013864469835203597,
      "loss": 1.9807,
      "step": 156850
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.2584645748138428,
      "learning_rate": 0.00013864124985985565,
      "loss": 2.1628,
      "step": 156851
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.1357053518295288,
      "learning_rate": 0.00013863780139767508,
      "loss": 2.0368,
      "step": 156852
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.1049132347106934,
      "learning_rate": 0.0001386343529654947,
      "loss": 1.8558,
      "step": 156853
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.3466423749923706,
      "learning_rate": 0.00013863090456331547,
      "loss": 2.1744,
      "step": 156854
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.2493215799331665,
      "learning_rate": 0.00013862745619113755,
      "loss": 1.9541,
      "step": 156855
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.2799440622329712,
      "learning_rate": 0.00013862400784896194,
      "loss": 2.02,
      "step": 156856
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.6431297063827515,
      "learning_rate": 0.00013862055953678908,
      "loss": 1.9976,
      "step": 156857
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.2885079383850098,
      "learning_rate": 0.0001386171112546198,
      "loss": 2.1304,
      "step": 156858
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.171075701713562,
      "learning_rate": 0.00013861366300245456,
      "loss": 1.8856,
      "step": 156859
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.227421522140503,
      "learning_rate": 0.0001386102147802943,
      "loss": 2.1465,
      "step": 156860
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.3778125047683716,
      "learning_rate": 0.0001386067665881392,
      "loss": 2.0704,
      "step": 156861
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.3792917728424072,
      "learning_rate": 0.00013860331842599027,
      "loss": 2.4143,
      "step": 156862
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.1097005605697632,
      "learning_rate": 0.00013859987029384797,
      "loss": 2.0786,
      "step": 156863
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.2731552124023438,
      "learning_rate": 0.00013859642219171304,
      "loss": 1.8599,
      "step": 156864
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.1848301887512207,
      "learning_rate": 0.000138592974119586,
      "loss": 2.068,
      "step": 156865
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.323279619216919,
      "learning_rate": 0.0001385895260774677,
      "loss": 2.0463,
      "step": 156866
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.1156320571899414,
      "learning_rate": 0.00013858607806535863,
      "loss": 2.1988,
      "step": 156867
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.2012149095535278,
      "learning_rate": 0.00013858263008325946,
      "loss": 2.0438,
      "step": 156868
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.1286495923995972,
      "learning_rate": 0.00013857918213117072,
      "loss": 1.9149,
      "step": 156869
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.1344605684280396,
      "learning_rate": 0.00013857573420909328,
      "loss": 2.2177,
      "step": 156870
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.265479326248169,
      "learning_rate": 0.00013857228631702751,
      "loss": 2.1441,
      "step": 156871
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.1002073287963867,
      "learning_rate": 0.00013856883845497436,
      "loss": 1.9362,
      "step": 156872
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.2526334524154663,
      "learning_rate": 0.00013856539062293428,
      "loss": 2.0406,
      "step": 156873
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.1918050050735474,
      "learning_rate": 0.0001385619428209078,
      "loss": 2.0297,
      "step": 156874
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.0924935340881348,
      "learning_rate": 0.0001385584950488959,
      "loss": 1.9212,
      "step": 156875
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.203291654586792,
      "learning_rate": 0.00013855504730689894,
      "loss": 2.1546,
      "step": 156876
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.1616628170013428,
      "learning_rate": 0.00013855159959491758,
      "loss": 1.9078,
      "step": 156877
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.108967661857605,
      "learning_rate": 0.0001385481519129526,
      "loss": 2.1214,
      "step": 156878
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.2679041624069214,
      "learning_rate": 0.0001385447042610046,
      "loss": 2.041,
      "step": 156879
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.2139301300048828,
      "learning_rate": 0.00013854125663907406,
      "loss": 2.2506,
      "step": 156880
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.136597990989685,
      "learning_rate": 0.00013853780904716186,
      "loss": 1.8793,
      "step": 156881
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.3081746101379395,
      "learning_rate": 0.00013853436148526853,
      "loss": 1.9167,
      "step": 156882
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.147759199142456,
      "learning_rate": 0.0001385309139533946,
      "loss": 2.1202,
      "step": 156883
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.1653876304626465,
      "learning_rate": 0.00013852746645154094,
      "loss": 2.1453,
      "step": 156884
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.1805026531219482,
      "learning_rate": 0.00013852401897970805,
      "loss": 1.8748,
      "step": 156885
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.2245131731033325,
      "learning_rate": 0.00013852057153789649,
      "loss": 2.142,
      "step": 156886
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.7543694972991943,
      "learning_rate": 0.00013851712412610713,
      "loss": 2.0223,
      "step": 156887
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.3764156103134155,
      "learning_rate": 0.00013851367674434033,
      "loss": 2.0692,
      "step": 156888
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.282150387763977,
      "learning_rate": 0.00013851022939259704,
      "loss": 2.1213,
      "step": 156889
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.1361572742462158,
      "learning_rate": 0.00013850678207087775,
      "loss": 1.9637,
      "step": 156890
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.5102182626724243,
      "learning_rate": 0.00013850333477918308,
      "loss": 2.0532,
      "step": 156891
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.1408202648162842,
      "learning_rate": 0.0001384998875175136,
      "loss": 2.0926,
      "step": 156892
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.3369108438491821,
      "learning_rate": 0.00013849644028587012,
      "loss": 2.04,
      "step": 156893
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.1572990417480469,
      "learning_rate": 0.0001384929930842531,
      "loss": 2.1496,
      "step": 156894
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.2571609020233154,
      "learning_rate": 0.0001384895459126634,
      "loss": 2.0223,
      "step": 156895
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.2709565162658691,
      "learning_rate": 0.00013848609877110152,
      "loss": 2.174,
      "step": 156896
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.1026389598846436,
      "learning_rate": 0.00013848265165956815,
      "loss": 2.1911,
      "step": 156897
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.1977403163909912,
      "learning_rate": 0.0001384792045780638,
      "loss": 2.2737,
      "step": 156898
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.092568039894104,
      "learning_rate": 0.00013847575752658933,
      "loss": 1.8594,
      "step": 156899
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.2014527320861816,
      "learning_rate": 0.0001384723105051451,
      "loss": 2.1421,
      "step": 156900
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.1668938398361206,
      "learning_rate": 0.00013846886351373205,
      "loss": 1.9453,
      "step": 156901
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.1048542261123657,
      "learning_rate": 0.0001384654165523506,
      "loss": 1.9807,
      "step": 156902
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.192672848701477,
      "learning_rate": 0.0001384619696210017,
      "loss": 2.0049,
      "step": 156903
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.1362417936325073,
      "learning_rate": 0.0001384585227196855,
      "loss": 1.8796,
      "step": 156904
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.1434712409973145,
      "learning_rate": 0.00013845507584840308,
      "loss": 2.3364,
      "step": 156905
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.2929906845092773,
      "learning_rate": 0.00013845162900715476,
      "loss": 1.9914,
      "step": 156906
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.1660763025283813,
      "learning_rate": 0.0001384481821959415,
      "loss": 2.072,
      "step": 156907
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.1437265872955322,
      "learning_rate": 0.0001384447354147636,
      "loss": 2.0339,
      "step": 156908
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.2292609214782715,
      "learning_rate": 0.0001384412886636221,
      "loss": 1.8659,
      "step": 156909
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.3772830963134766,
      "learning_rate": 0.0001384378419425172,
      "loss": 2.1409,
      "step": 156910
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.1524688005447388,
      "learning_rate": 0.0001384343952514499,
      "loss": 2.0852,
      "step": 156911
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.099813461303711,
      "learning_rate": 0.00013843094859042056,
      "loss": 1.9575,
      "step": 156912
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.429339051246643,
      "learning_rate": 0.00013842750195943007,
      "loss": 2.0727,
      "step": 156913
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.3762255907058716,
      "learning_rate": 0.00013842405535847882,
      "loss": 2.0403,
      "step": 156914
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.1155022382736206,
      "learning_rate": 0.00013842060878756785,
      "loss": 2.1343,
      "step": 156915
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.0446888208389282,
      "learning_rate": 0.00013841716224669728,
      "loss": 1.9305,
      "step": 156916
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.1364656686782837,
      "learning_rate": 0.00013841371573586816,
      "loss": 1.9802,
      "step": 156917
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.1353763341903687,
      "learning_rate": 0.00013841026925508082,
      "loss": 2.0383,
      "step": 156918
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.136690616607666,
      "learning_rate": 0.0001384068228043362,
      "loss": 2.1306,
      "step": 156919
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.1004180908203125,
      "learning_rate": 0.0001384033763836347,
      "loss": 1.9457,
      "step": 156920
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.2379168272018433,
      "learning_rate": 0.00013839992999297728,
      "loss": 1.9268,
      "step": 156921
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.1534459590911865,
      "learning_rate": 0.00013839648363236413,
      "loss": 2.0476,
      "step": 156922
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.1938713788986206,
      "learning_rate": 0.00013839303730179628,
      "loss": 2.2074,
      "step": 156923
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.0775165557861328,
      "learning_rate": 0.00013838959100127404,
      "loss": 1.9449,
      "step": 156924
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.0888153314590454,
      "learning_rate": 0.00013838614473079836,
      "loss": 1.6687,
      "step": 156925
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.1346570253372192,
      "learning_rate": 0.00013838269849036965,
      "loss": 1.9063,
      "step": 156926
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.25700044631958,
      "learning_rate": 0.00013837925227998886,
      "loss": 1.9206,
      "step": 156927
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.241014003753662,
      "learning_rate": 0.0001383758060996562,
      "loss": 2.3113,
      "step": 156928
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.0548765659332275,
      "learning_rate": 0.00013837235994937263,
      "loss": 1.9304,
      "step": 156929
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.1344155073165894,
      "learning_rate": 0.0001383689138291386,
      "loss": 2.0129,
      "step": 156930
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.3406023979187012,
      "learning_rate": 0.0001383654677389549,
      "loss": 2.0412,
      "step": 156931
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.2763431072235107,
      "learning_rate": 0.00013836202167882204,
      "loss": 2.1304,
      "step": 156932
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.1532328128814697,
      "learning_rate": 0.00013835857564874093,
      "loss": 1.9719,
      "step": 156933
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.7180145978927612,
      "learning_rate": 0.00013835512964871185,
      "loss": 2.1743,
      "step": 156934
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.2725688219070435,
      "learning_rate": 0.00013835168367873563,
      "loss": 2.0933,
      "step": 156935
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.1767914295196533,
      "learning_rate": 0.00013834823773881282,
      "loss": 2.1241,
      "step": 156936
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.1841751337051392,
      "learning_rate": 0.00013834479182894425,
      "loss": 2.0587,
      "step": 156937
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.4952179193496704,
      "learning_rate": 0.0001383413459491303,
      "loss": 1.9208,
      "step": 156938
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.25347101688385,
      "learning_rate": 0.00013833790009937181,
      "loss": 2.234,
      "step": 156939
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.1490830183029175,
      "learning_rate": 0.0001383344542796694,
      "loss": 2.1622,
      "step": 156940
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.200989007949829,
      "learning_rate": 0.0001383310084900237,
      "loss": 1.878,
      "step": 156941
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.157727837562561,
      "learning_rate": 0.00013832756273043516,
      "loss": 1.8372,
      "step": 156942
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.3845571279525757,
      "learning_rate": 0.00013832411700090473,
      "loss": 2.2256,
      "step": 156943
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.194359302520752,
      "learning_rate": 0.00013832067130143275,
      "loss": 2.1024,
      "step": 156944
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.2287473678588867,
      "learning_rate": 0.00013831722563202012,
      "loss": 2.248,
      "step": 156945
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.0968048572540283,
      "learning_rate": 0.0001383137799926674,
      "loss": 2.3143,
      "step": 156946
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.2281227111816406,
      "learning_rate": 0.00013831033438337518,
      "loss": 1.9031,
      "step": 156947
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.1750974655151367,
      "learning_rate": 0.000138306888804144,
      "loss": 1.8697,
      "step": 156948
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.2079561948776245,
      "learning_rate": 0.00013830344325497476,
      "loss": 2.1343,
      "step": 156949
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.174593448638916,
      "learning_rate": 0.00013829999773586784,
      "loss": 2.1697,
      "step": 156950
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.1604743003845215,
      "learning_rate": 0.00013829655224682408,
      "loss": 2.0145,
      "step": 156951
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.274245023727417,
      "learning_rate": 0.0001382931067878441,
      "loss": 1.9389,
      "step": 156952
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.2216136455535889,
      "learning_rate": 0.00013828966135892847,
      "loss": 2.2071,
      "step": 156953
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.1124311685562134,
      "learning_rate": 0.0001382862159600777,
      "loss": 1.9126,
      "step": 156954
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.3246124982833862,
      "learning_rate": 0.0001382827705912927,
      "loss": 2.2572,
      "step": 156955
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.1418248414993286,
      "learning_rate": 0.0001382793252525739,
      "loss": 1.8898,
      "step": 156956
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.1465599536895752,
      "learning_rate": 0.0001382758799439221,
      "loss": 1.9572,
      "step": 156957
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.1903860569000244,
      "learning_rate": 0.0001382724346653379,
      "loss": 1.9037,
      "step": 156958
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.2050979137420654,
      "learning_rate": 0.00013826898941682187,
      "loss": 2.1076,
      "step": 156959
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.4238356351852417,
      "learning_rate": 0.00013826554419837461,
      "loss": 2.3286,
      "step": 156960
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.2624632120132446,
      "learning_rate": 0.00013826209900999692,
      "loss": 1.9079,
      "step": 156961
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.2236000299453735,
      "learning_rate": 0.0001382586538516893,
      "loss": 2.1117,
      "step": 156962
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.173543930053711,
      "learning_rate": 0.0001382552087234525,
      "loss": 2.1927,
      "step": 156963
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.3612169027328491,
      "learning_rate": 0.00013825176362528715,
      "loss": 1.8459,
      "step": 156964
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.1327083110809326,
      "learning_rate": 0.00013824831855719372,
      "loss": 2.0179,
      "step": 156965
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.2244937419891357,
      "learning_rate": 0.0001382448735191731,
      "loss": 1.9687,
      "step": 156966
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.144025206565857,
      "learning_rate": 0.00013824142851122582,
      "loss": 1.9677,
      "step": 156967
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.1068520545959473,
      "learning_rate": 0.0001382379835333524,
      "loss": 2.0511,
      "step": 156968
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.1993987560272217,
      "learning_rate": 0.00013823453858555367,
      "loss": 1.9303,
      "step": 156969
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.1830934286117554,
      "learning_rate": 0.00013823109366783022,
      "loss": 1.8456,
      "step": 156970
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.1639751195907593,
      "learning_rate": 0.00013822764878018253,
      "loss": 1.9225,
      "step": 156971
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.1927564144134521,
      "learning_rate": 0.00013822420392261154,
      "loss": 2.0991,
      "step": 156972
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.1829949617385864,
      "learning_rate": 0.00013822075909511756,
      "loss": 2.0307,
      "step": 156973
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.094956398010254,
      "learning_rate": 0.00013821731429770152,
      "loss": 2.0255,
      "step": 156974
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.1550368070602417,
      "learning_rate": 0.00013821386953036396,
      "loss": 2.1598,
      "step": 156975
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.1604537963867188,
      "learning_rate": 0.00013821042479310544,
      "loss": 2.093,
      "step": 156976
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.178788185119629,
      "learning_rate": 0.00013820698008592657,
      "loss": 2.0691,
      "step": 156977
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.3120157718658447,
      "learning_rate": 0.0001382035354088282,
      "loss": 2.1489,
      "step": 156978
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.2244188785552979,
      "learning_rate": 0.00013820009076181074,
      "loss": 2.319,
      "step": 156979
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.1018956899642944,
      "learning_rate": 0.00013819664614487503,
      "loss": 1.8732,
      "step": 156980
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.146338701248169,
      "learning_rate": 0.00013819320155802165,
      "loss": 2.0231,
      "step": 156981
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.2120931148529053,
      "learning_rate": 0.00013818975700125117,
      "loss": 1.8096,
      "step": 156982
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.1144230365753174,
      "learning_rate": 0.00013818631247456418,
      "loss": 1.8979,
      "step": 156983
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.1382354497909546,
      "learning_rate": 0.00013818286797796152,
      "loss": 1.815,
      "step": 156984
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.1314542293548584,
      "learning_rate": 0.0001381794235114436,
      "loss": 2.1842,
      "step": 156985
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.2810592651367188,
      "learning_rate": 0.0001381759790750113,
      "loss": 2.1226,
      "step": 156986
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.2451692819595337,
      "learning_rate": 0.00013817253466866502,
      "loss": 2.132,
      "step": 156987
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.2350621223449707,
      "learning_rate": 0.00013816909029240574,
      "loss": 2.1834,
      "step": 156988
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.1682413816452026,
      "learning_rate": 0.00013816564594623367,
      "loss": 1.7683,
      "step": 156989
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.145227313041687,
      "learning_rate": 0.00013816220163014975,
      "loss": 2.0529,
      "step": 156990
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.13601815700531,
      "learning_rate": 0.00013815875734415445,
      "loss": 2.2624,
      "step": 156991
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.0986133813858032,
      "learning_rate": 0.0001381553130882486,
      "loss": 2.0275,
      "step": 156992
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.3939546346664429,
      "learning_rate": 0.0001381518688624326,
      "loss": 2.2403,
      "step": 156993
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.3775182962417603,
      "learning_rate": 0.00013814842466670745,
      "loss": 2.0372,
      "step": 156994
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.1395107507705688,
      "learning_rate": 0.00013814498050107336,
      "loss": 1.9671,
      "step": 156995
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.133266568183899,
      "learning_rate": 0.00013814153636553124,
      "loss": 1.8131,
      "step": 156996
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.0367056131362915,
      "learning_rate": 0.0001381380922600816,
      "loss": 2.0884,
      "step": 156997
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.28495192527771,
      "learning_rate": 0.00013813464818472523,
      "loss": 1.9575,
      "step": 156998
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.1560689210891724,
      "learning_rate": 0.00013813120413946257,
      "loss": 1.8694,
      "step": 156999
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.13516104221344,
      "learning_rate": 0.00013812776012429463,
      "loss": 2.2721,
      "step": 157000
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.1498968601226807,
      "learning_rate": 0.00013812431613922157,
      "loss": 2.0376,
      "step": 157001
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.2813719511032104,
      "learning_rate": 0.00013812087218424433,
      "loss": 2.0252,
      "step": 157002
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.1058061122894287,
      "learning_rate": 0.00013811742825936338,
      "loss": 2.3392,
      "step": 157003
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.1993001699447632,
      "learning_rate": 0.00013811398436457958,
      "loss": 1.8358,
      "step": 157004
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.5547109842300415,
      "learning_rate": 0.00013811054049989331,
      "loss": 2.1482,
      "step": 157005
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.1381388902664185,
      "learning_rate": 0.00013810709666530548,
      "loss": 2.1182,
      "step": 157006
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.1769596338272095,
      "learning_rate": 0.00013810365286081656,
      "loss": 2.0893,
      "step": 157007
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.335371971130371,
      "learning_rate": 0.00013810020908642726,
      "loss": 2.0332,
      "step": 157008
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.3020915985107422,
      "learning_rate": 0.00013809676534213807,
      "loss": 2.1462,
      "step": 157009
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.2279987335205078,
      "learning_rate": 0.00013809332162794985,
      "loss": 2.215,
      "step": 157010
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.1991058588027954,
      "learning_rate": 0.00013808987794386307,
      "loss": 2.1586,
      "step": 157011
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.3036627769470215,
      "learning_rate": 0.0001380864342898785,
      "loss": 2.067,
      "step": 157012
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.205919623374939,
      "learning_rate": 0.00013808299066599676,
      "loss": 2.0528,
      "step": 157013
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.3114941120147705,
      "learning_rate": 0.00013807954707221838,
      "loss": 1.9086,
      "step": 157014
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.2685126066207886,
      "learning_rate": 0.00013807610350854402,
      "loss": 2.0906,
      "step": 157015
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.2053719758987427,
      "learning_rate": 0.00013807265997497445,
      "loss": 2.0866,
      "step": 157016
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.2133866548538208,
      "learning_rate": 0.00013806921647151015,
      "loss": 1.9929,
      "step": 157017
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.349597454071045,
      "learning_rate": 0.0001380657729981519,
      "loss": 2.0492,
      "step": 157018
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.115535020828247,
      "learning_rate": 0.00013806232955490032,
      "loss": 2.2471,
      "step": 157019
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.5189942121505737,
      "learning_rate": 0.000138058886141756,
      "loss": 2.1525,
      "step": 157020
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.276853084564209,
      "learning_rate": 0.00013805544275871946,
      "loss": 2.0143,
      "step": 157021
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.244089126586914,
      "learning_rate": 0.00013805199940579161,
      "loss": 1.9083,
      "step": 157022
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.4108062982559204,
      "learning_rate": 0.00013804855608297282,
      "loss": 2.0244,
      "step": 157023
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.4898617267608643,
      "learning_rate": 0.000138045112790264,
      "loss": 1.8086,
      "step": 157024
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.1772302389144897,
      "learning_rate": 0.0001380416695276656,
      "loss": 1.8652,
      "step": 157025
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.0998040437698364,
      "learning_rate": 0.00013803822629517837,
      "loss": 1.9808,
      "step": 157026
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.3368124961853027,
      "learning_rate": 0.00013803478309280275,
      "loss": 2.2243,
      "step": 157027
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.1177765130996704,
      "learning_rate": 0.00013803133992053963,
      "loss": 2.1944,
      "step": 157028
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.1587198972702026,
      "learning_rate": 0.0001380278967783894,
      "loss": 1.9754,
      "step": 157029
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.267379641532898,
      "learning_rate": 0.00013802445366635298,
      "loss": 2.0962,
      "step": 157030
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.286325216293335,
      "learning_rate": 0.0001380210105844309,
      "loss": 2.2516,
      "step": 157031
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.210384488105774,
      "learning_rate": 0.00013801756753262375,
      "loss": 2.3361,
      "step": 157032
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.179650068283081,
      "learning_rate": 0.00013801412451093208,
      "loss": 1.9721,
      "step": 157033
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.2492523193359375,
      "learning_rate": 0.00013801068151935675,
      "loss": 1.9681,
      "step": 157034
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.2565604448318481,
      "learning_rate": 0.0001380072385578982,
      "loss": 2.0203,
      "step": 157035
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.0786852836608887,
      "learning_rate": 0.00013800379562655725,
      "loss": 1.8508,
      "step": 157036
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.1744945049285889,
      "learning_rate": 0.00013800035272533445,
      "loss": 2.1555,
      "step": 157037
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.2071858644485474,
      "learning_rate": 0.00013799690985423043,
      "loss": 1.8995,
      "step": 157038
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.1246888637542725,
      "learning_rate": 0.00013799346701324578,
      "loss": 1.9298,
      "step": 157039
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.2259353399276733,
      "learning_rate": 0.00013799002420238128,
      "loss": 2.0723,
      "step": 157040
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.3278311491012573,
      "learning_rate": 0.0001379865814216374,
      "loss": 1.998,
      "step": 157041
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.219969391822815,
      "learning_rate": 0.00013798313867101493,
      "loss": 2.2003,
      "step": 157042
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.2654255628585815,
      "learning_rate": 0.00013797969595051452,
      "loss": 2.0746,
      "step": 157043
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.1810953617095947,
      "learning_rate": 0.0001379762532601367,
      "loss": 2.0464,
      "step": 157044
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.1335662603378296,
      "learning_rate": 0.00013797281059988207,
      "loss": 2.0771,
      "step": 157045
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.2470669746398926,
      "learning_rate": 0.00013796936796975145,
      "loss": 2.1223,
      "step": 157046
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.1753603219985962,
      "learning_rate": 0.00013796592536974528,
      "loss": 1.7394,
      "step": 157047
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.2543865442276,
      "learning_rate": 0.00013796248279986442,
      "loss": 2.0895,
      "step": 157048
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.1024742126464844,
      "learning_rate": 0.0001379590402601094,
      "loss": 1.8542,
      "step": 157049
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.1864290237426758,
      "learning_rate": 0.0001379555977504807,
      "loss": 1.9279,
      "step": 157050
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.1034928560256958,
      "learning_rate": 0.00013795215527097923,
      "loss": 1.8037,
      "step": 157051
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.2077745199203491,
      "learning_rate": 0.00013794871282160555,
      "loss": 2.1702,
      "step": 157052
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.0973864793777466,
      "learning_rate": 0.0001379452704023601,
      "loss": 1.9782,
      "step": 157053
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.2169243097305298,
      "learning_rate": 0.00013794182801324387,
      "loss": 2.1482,
      "step": 157054
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.1739517450332642,
      "learning_rate": 0.00013793838565425727,
      "loss": 1.9797,
      "step": 157055
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.2225477695465088,
      "learning_rate": 0.00013793494332540088,
      "loss": 2.0905,
      "step": 157056
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.0653795003890991,
      "learning_rate": 0.00013793150102667554,
      "loss": 1.9323,
      "step": 157057
    },
    {
      "epoch": 2.04,
      "grad_norm": 1.1341015100479126,
      "learning_rate": 0.0001379280587580818,
      "loss": 1.9911,
      "step": 157058
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.0579191446304321,
      "learning_rate": 0.00013792461651962018,
      "loss": 2.0547,
      "step": 157059
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.0629236698150635,
      "learning_rate": 0.00013792117431129155,
      "loss": 1.9252,
      "step": 157060
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.2456976175308228,
      "learning_rate": 0.0001379177321330964,
      "loss": 2.047,
      "step": 157061
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.404958963394165,
      "learning_rate": 0.00013791428998503535,
      "loss": 1.9316,
      "step": 157062
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.2694109678268433,
      "learning_rate": 0.00013791084786710916,
      "loss": 2.2675,
      "step": 157063
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.3285285234451294,
      "learning_rate": 0.00013790740577931832,
      "loss": 1.9225,
      "step": 157064
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.3000761270523071,
      "learning_rate": 0.00013790396372166365,
      "loss": 2.0118,
      "step": 157065
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.2129684686660767,
      "learning_rate": 0.0001379005216941457,
      "loss": 1.9239,
      "step": 157066
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.1941497325897217,
      "learning_rate": 0.00013789707969676508,
      "loss": 2.1473,
      "step": 157067
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.3309619426727295,
      "learning_rate": 0.0001378936377295224,
      "loss": 2.2544,
      "step": 157068
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.3467928171157837,
      "learning_rate": 0.0001378901957924184,
      "loss": 2.0972,
      "step": 157069
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.1562026739120483,
      "learning_rate": 0.0001378867538854536,
      "loss": 2.1476,
      "step": 157070
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.2195466756820679,
      "learning_rate": 0.0001378833120086288,
      "loss": 2.1308,
      "step": 157071
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.1947822570800781,
      "learning_rate": 0.00013787987016194444,
      "loss": 1.8639,
      "step": 157072
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.195505142211914,
      "learning_rate": 0.0001378764283454014,
      "loss": 1.9491,
      "step": 157073
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.3512531518936157,
      "learning_rate": 0.0001378729865590002,
      "loss": 2.2669,
      "step": 157074
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.3292735815048218,
      "learning_rate": 0.00013786954480274148,
      "loss": 2.2249,
      "step": 157075
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.1214954853057861,
      "learning_rate": 0.00013786610307662574,
      "loss": 1.9459,
      "step": 157076
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.1671594381332397,
      "learning_rate": 0.00013786266138065385,
      "loss": 2.1519,
      "step": 157077
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.2615727186203003,
      "learning_rate": 0.00013785921971482624,
      "loss": 2.0254,
      "step": 157078
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.2738845348358154,
      "learning_rate": 0.0001378557780791438,
      "loss": 1.8997,
      "step": 157079
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.1257978677749634,
      "learning_rate": 0.00013785233647360703,
      "loss": 1.9886,
      "step": 157080
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.1634560823440552,
      "learning_rate": 0.00013784889489821655,
      "loss": 2.1138,
      "step": 157081
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.1566760540008545,
      "learning_rate": 0.0001378454533529729,
      "loss": 2.1698,
      "step": 157082
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.2277532815933228,
      "learning_rate": 0.00013784201183787698,
      "loss": 1.8345,
      "step": 157083
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.331215739250183,
      "learning_rate": 0.00013783857035292916,
      "loss": 2.1197,
      "step": 157084
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.1164708137512207,
      "learning_rate": 0.00013783512889813037,
      "loss": 1.986,
      "step": 157085
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.0691683292388916,
      "learning_rate": 0.00013783168747348104,
      "loss": 1.9034,
      "step": 157086
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.385959267616272,
      "learning_rate": 0.00013782824607898187,
      "loss": 2.1659,
      "step": 157087
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.296913743019104,
      "learning_rate": 0.00013782480471463338,
      "loss": 2.1838,
      "step": 157088
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.1270947456359863,
      "learning_rate": 0.00013782136338043644,
      "loss": 1.8737,
      "step": 157089
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.1625441312789917,
      "learning_rate": 0.00013781792207639146,
      "loss": 2.022,
      "step": 157090
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.1090196371078491,
      "learning_rate": 0.00013781448080249926,
      "loss": 2.0691,
      "step": 157091
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.2164052724838257,
      "learning_rate": 0.00013781103955876048,
      "loss": 1.9977,
      "step": 157092
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.3853178024291992,
      "learning_rate": 0.00013780759834517564,
      "loss": 2.0834,
      "step": 157093
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.1182290315628052,
      "learning_rate": 0.00013780415716174534,
      "loss": 1.9339,
      "step": 157094
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.2404271364212036,
      "learning_rate": 0.0001378007160084704,
      "loss": 2.2159,
      "step": 157095
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.1263597011566162,
      "learning_rate": 0.0001377972748853513,
      "loss": 2.0149,
      "step": 157096
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.171995997428894,
      "learning_rate": 0.00013779383379238882,
      "loss": 1.9101,
      "step": 157097
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.2027512788772583,
      "learning_rate": 0.00013779039272958352,
      "loss": 2.2375,
      "step": 157098
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.2235349416732788,
      "learning_rate": 0.00013778695169693606,
      "loss": 2.1505,
      "step": 157099
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.3092238903045654,
      "learning_rate": 0.00013778351069444698,
      "loss": 2.1592,
      "step": 157100
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.1982216835021973,
      "learning_rate": 0.00013778006972211708,
      "loss": 2.0335,
      "step": 157101
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.212721586227417,
      "learning_rate": 0.00013777662877994686,
      "loss": 2.0527,
      "step": 157102
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.2941620349884033,
      "learning_rate": 0.0001377731878679371,
      "loss": 2.2991,
      "step": 157103
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.1287345886230469,
      "learning_rate": 0.00013776974698608837,
      "loss": 2.2859,
      "step": 157104
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.0819474458694458,
      "learning_rate": 0.00013776630613440133,
      "loss": 2.0076,
      "step": 157105
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.1237128973007202,
      "learning_rate": 0.00013776286531287646,
      "loss": 1.9815,
      "step": 157106
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.3114631175994873,
      "learning_rate": 0.00013775942452151467,
      "loss": 1.9754,
      "step": 157107
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.4083679914474487,
      "learning_rate": 0.00013775598376031636,
      "loss": 2.0654,
      "step": 157108
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.3444242477416992,
      "learning_rate": 0.00013775254302928237,
      "loss": 2.0906,
      "step": 157109
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.2232086658477783,
      "learning_rate": 0.00013774910232841324,
      "loss": 2.1481,
      "step": 157110
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.087265133857727,
      "learning_rate": 0.0001377456616577096,
      "loss": 2.2773,
      "step": 157111
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.2949153184890747,
      "learning_rate": 0.000137742221017172,
      "loss": 2.1051,
      "step": 157112
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.2718782424926758,
      "learning_rate": 0.00013773878040680134,
      "loss": 1.9691,
      "step": 157113
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.1347429752349854,
      "learning_rate": 0.00013773533982659793,
      "loss": 1.9026,
      "step": 157114
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.2764254808425903,
      "learning_rate": 0.00013773189927656273,
      "loss": 2.2575,
      "step": 157115
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.2176508903503418,
      "learning_rate": 0.00013772845875669623,
      "loss": 2.0725,
      "step": 157116
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.3124089241027832,
      "learning_rate": 0.00013772501826699908,
      "loss": 2.1273,
      "step": 157117
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.3282082080841064,
      "learning_rate": 0.00013772157780747178,
      "loss": 2.022,
      "step": 157118
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.0671133995056152,
      "learning_rate": 0.00013771813737811526,
      "loss": 1.8995,
      "step": 157119
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.189087986946106,
      "learning_rate": 0.00013771469697892983,
      "loss": 2.1411,
      "step": 157120
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.4489871263504028,
      "learning_rate": 0.00013771125660991645,
      "loss": 2.1211,
      "step": 157121
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.207780122756958,
      "learning_rate": 0.0001377078162710756,
      "loss": 2.1357,
      "step": 157122
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.2664072513580322,
      "learning_rate": 0.0001377043759624079,
      "loss": 2.0682,
      "step": 157123
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.2900807857513428,
      "learning_rate": 0.00013770093568391396,
      "loss": 2.0007,
      "step": 157124
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.1956095695495605,
      "learning_rate": 0.00013769749543559456,
      "loss": 1.9527,
      "step": 157125
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.215419054031372,
      "learning_rate": 0.00013769405521745016,
      "loss": 2.1521,
      "step": 157126
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.3143748044967651,
      "learning_rate": 0.0001376906150294816,
      "loss": 2.1754,
      "step": 157127
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.23943293094635,
      "learning_rate": 0.00013768717487168942,
      "loss": 2.09,
      "step": 157128
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.1696089506149292,
      "learning_rate": 0.00013768373474407427,
      "loss": 1.8592,
      "step": 157129
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.1156877279281616,
      "learning_rate": 0.00013768029464663668,
      "loss": 2.1242,
      "step": 157130
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.234885811805725,
      "learning_rate": 0.00013767685457937744,
      "loss": 2.0609,
      "step": 157131
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.329280972480774,
      "learning_rate": 0.00013767341454229708,
      "loss": 2.0612,
      "step": 157132
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.107374668121338,
      "learning_rate": 0.00013766997453539638,
      "loss": 1.9002,
      "step": 157133
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.1734418869018555,
      "learning_rate": 0.0001376665345586759,
      "loss": 2.1474,
      "step": 157134
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.1692681312561035,
      "learning_rate": 0.00013766309461213619,
      "loss": 2.1524,
      "step": 157135
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.182737946510315,
      "learning_rate": 0.00013765965469577808,
      "loss": 1.9062,
      "step": 157136
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.3230129480361938,
      "learning_rate": 0.0001376562148096021,
      "loss": 1.9334,
      "step": 157137
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.3662686347961426,
      "learning_rate": 0.00013765277495360873,
      "loss": 1.9448,
      "step": 157138
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.1040390729904175,
      "learning_rate": 0.00013764933512779896,
      "loss": 2.1061,
      "step": 157139
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.3383458852767944,
      "learning_rate": 0.00013764589533217308,
      "loss": 1.9632,
      "step": 157140
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.0691194534301758,
      "learning_rate": 0.00013764245556673204,
      "loss": 2.0007,
      "step": 157141
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.1684669256210327,
      "learning_rate": 0.0001376390158314763,
      "loss": 1.9683,
      "step": 157142
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.1343754529953003,
      "learning_rate": 0.00013763557612640656,
      "loss": 1.9793,
      "step": 157143
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.135980486869812,
      "learning_rate": 0.0001376321364515233,
      "loss": 2.1591,
      "step": 157144
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.1480541229248047,
      "learning_rate": 0.00013762869680682745,
      "loss": 2.0611,
      "step": 157145
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.265503168106079,
      "learning_rate": 0.00013762525719231933,
      "loss": 2.1237,
      "step": 157146
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.1589046716690063,
      "learning_rate": 0.00013762181760799987,
      "loss": 1.902,
      "step": 157147
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.1356348991394043,
      "learning_rate": 0.00013761837805386954,
      "loss": 2.0624,
      "step": 157148
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.1221927404403687,
      "learning_rate": 0.00013761493852992894,
      "loss": 2.0171,
      "step": 157149
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.25641667842865,
      "learning_rate": 0.0001376114990361789,
      "loss": 1.9356,
      "step": 157150
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.1585662364959717,
      "learning_rate": 0.00013760805957261998,
      "loss": 2.0792,
      "step": 157151
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.3030602931976318,
      "learning_rate": 0.00013760462013925263,
      "loss": 2.0724,
      "step": 157152
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.1752387285232544,
      "learning_rate": 0.00013760118073607777,
      "loss": 1.9634,
      "step": 157153
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.2592169046401978,
      "learning_rate": 0.0001375977413630959,
      "loss": 2.121,
      "step": 157154
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.1213057041168213,
      "learning_rate": 0.00013759430202030758,
      "loss": 2.2605,
      "step": 157155
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.2422293424606323,
      "learning_rate": 0.00013759086270771366,
      "loss": 2.0337,
      "step": 157156
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.2073737382888794,
      "learning_rate": 0.00013758742342531455,
      "loss": 1.9619,
      "step": 157157
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.0744191408157349,
      "learning_rate": 0.0001375839841731111,
      "loss": 2.0189,
      "step": 157158
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.2605777978897095,
      "learning_rate": 0.00013758054495110388,
      "loss": 1.9695,
      "step": 157159
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.1674387454986572,
      "learning_rate": 0.0001375771057592935,
      "loss": 1.8747,
      "step": 157160
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.1132677793502808,
      "learning_rate": 0.00013757366659768048,
      "loss": 1.9139,
      "step": 157161
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.3157732486724854,
      "learning_rate": 0.00013757022746626568,
      "loss": 2.0907,
      "step": 157162
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.1877312660217285,
      "learning_rate": 0.00013756678836504953,
      "loss": 2.094,
      "step": 157163
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.082093596458435,
      "learning_rate": 0.0001375633492940329,
      "loss": 2.0416,
      "step": 157164
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.2156895399093628,
      "learning_rate": 0.00013755991025321632,
      "loss": 2.1288,
      "step": 157165
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.2833750247955322,
      "learning_rate": 0.00013755647124260044,
      "loss": 2.134,
      "step": 157166
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.172599196434021,
      "learning_rate": 0.00013755303226218574,
      "loss": 2.0656,
      "step": 157167
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.0957707166671753,
      "learning_rate": 0.0001375495933119731,
      "loss": 2.1291,
      "step": 157168
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.2980217933654785,
      "learning_rate": 0.00013754615439196295,
      "loss": 2.0864,
      "step": 157169
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.3222010135650635,
      "learning_rate": 0.00013754271550215616,
      "loss": 2.0391,
      "step": 157170
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.1862404346466064,
      "learning_rate": 0.0001375392766425532,
      "loss": 1.9944,
      "step": 157171
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.2191507816314697,
      "learning_rate": 0.0001375358378131548,
      "loss": 1.9986,
      "step": 157172
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.2235186100006104,
      "learning_rate": 0.00013753239901396144,
      "loss": 2.1404,
      "step": 157173
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.1744993925094604,
      "learning_rate": 0.00013752896024497402,
      "loss": 2.1086,
      "step": 157174
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.2281280755996704,
      "learning_rate": 0.00013752552150619285,
      "loss": 2.2357,
      "step": 157175
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.1654655933380127,
      "learning_rate": 0.00013752208279761892,
      "loss": 2.161,
      "step": 157176
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.2135847806930542,
      "learning_rate": 0.0001375186441192527,
      "loss": 2.0525,
      "step": 157177
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.1258535385131836,
      "learning_rate": 0.00013751520547109476,
      "loss": 2.0787,
      "step": 157178
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.1352962255477905,
      "learning_rate": 0.00013751176685314578,
      "loss": 1.9759,
      "step": 157179
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.1419209241867065,
      "learning_rate": 0.0001375083282654065,
      "loss": 2.1184,
      "step": 157180
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.1749379634857178,
      "learning_rate": 0.0001375048897078774,
      "loss": 2.0661,
      "step": 157181
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.179822325706482,
      "learning_rate": 0.00013750145118055927,
      "loss": 2.103,
      "step": 157182
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.12269127368927,
      "learning_rate": 0.00013749801268345273,
      "loss": 1.98,
      "step": 157183
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.0656782388687134,
      "learning_rate": 0.00013749457421655838,
      "loss": 2.0062,
      "step": 157184
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.302931547164917,
      "learning_rate": 0.00013749113577987674,
      "loss": 2.2121,
      "step": 157185
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.187973976135254,
      "learning_rate": 0.0001374876973734087,
      "loss": 2.1295,
      "step": 157186
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.1639471054077148,
      "learning_rate": 0.00013748425899715462,
      "loss": 1.8805,
      "step": 157187
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.1423966884613037,
      "learning_rate": 0.0001374808206511154,
      "loss": 2.2174,
      "step": 157188
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.1270607709884644,
      "learning_rate": 0.00013747738233529157,
      "loss": 1.8192,
      "step": 157189
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.1432640552520752,
      "learning_rate": 0.00013747394404968374,
      "loss": 1.9434,
      "step": 157190
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.0446555614471436,
      "learning_rate": 0.00013747050579429249,
      "loss": 1.8861,
      "step": 157191
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.1527135372161865,
      "learning_rate": 0.00013746706756911865,
      "loss": 1.9247,
      "step": 157192
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.0600467920303345,
      "learning_rate": 0.00013746362937416262,
      "loss": 1.9565,
      "step": 157193
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.1455518007278442,
      "learning_rate": 0.0001374601912094253,
      "loss": 2.067,
      "step": 157194
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.3082828521728516,
      "learning_rate": 0.00013745675307490717,
      "loss": 2.0862,
      "step": 157195
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.3800315856933594,
      "learning_rate": 0.00013745331497060892,
      "loss": 2.2021,
      "step": 157196
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.2876931428909302,
      "learning_rate": 0.00013744987689653105,
      "loss": 2.0778,
      "step": 157197
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.4966565370559692,
      "learning_rate": 0.00013744643885267444,
      "loss": 2.2459,
      "step": 157198
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.3719462156295776,
      "learning_rate": 0.0001374430008390395,
      "loss": 2.0947,
      "step": 157199
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.2105978727340698,
      "learning_rate": 0.00013743956285562707,
      "loss": 1.9816,
      "step": 157200
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.0889238119125366,
      "learning_rate": 0.00013743612490243768,
      "loss": 1.981,
      "step": 157201
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.1478118896484375,
      "learning_rate": 0.000137432686979472,
      "loss": 2.1457,
      "step": 157202
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.161171317100525,
      "learning_rate": 0.00013742924908673055,
      "loss": 1.8339,
      "step": 157203
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.1583330631256104,
      "learning_rate": 0.00013742581122421415,
      "loss": 1.9875,
      "step": 157204
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.4183030128479004,
      "learning_rate": 0.00013742237339192326,
      "loss": 1.9413,
      "step": 157205
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.1399681568145752,
      "learning_rate": 0.00013741893558985875,
      "loss": 2.2603,
      "step": 157206
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.2117506265640259,
      "learning_rate": 0.00013741549781802102,
      "loss": 1.9136,
      "step": 157207
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.2127827405929565,
      "learning_rate": 0.00013741206007641103,
      "loss": 2.0558,
      "step": 157208
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.1661717891693115,
      "learning_rate": 0.00013740862236502898,
      "loss": 2.1273,
      "step": 157209
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.2916008234024048,
      "learning_rate": 0.00013740518468387586,
      "loss": 2.0968,
      "step": 157210
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.191402792930603,
      "learning_rate": 0.00013740174703295204,
      "loss": 2.0863,
      "step": 157211
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.326392650604248,
      "learning_rate": 0.00013739830941225846,
      "loss": 2.0107,
      "step": 157212
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.567186713218689,
      "learning_rate": 0.00013739487182179547,
      "loss": 1.9755,
      "step": 157213
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.4338033199310303,
      "learning_rate": 0.00013739143426156407,
      "loss": 2.2064,
      "step": 157214
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.7091362476348877,
      "learning_rate": 0.00013738799673156444,
      "loss": 2.1327,
      "step": 157215
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.2973661422729492,
      "learning_rate": 0.0001373845592317976,
      "loss": 2.0344,
      "step": 157216
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.4651074409484863,
      "learning_rate": 0.00013738112176226388,
      "loss": 2.013,
      "step": 157217
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.234676480293274,
      "learning_rate": 0.00013737768432296423,
      "loss": 2.2679,
      "step": 157218
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.1478561162948608,
      "learning_rate": 0.000137374246913899,
      "loss": 1.9413,
      "step": 157219
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.211003303527832,
      "learning_rate": 0.00013737080953506905,
      "loss": 1.954,
      "step": 157220
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.1593005657196045,
      "learning_rate": 0.000137367372186475,
      "loss": 2.0611,
      "step": 157221
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.2204951047897339,
      "learning_rate": 0.00013736393486811737,
      "loss": 1.9647,
      "step": 157222
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.2238250970840454,
      "learning_rate": 0.0001373604975799968,
      "loss": 1.938,
      "step": 157223
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.1371736526489258,
      "learning_rate": 0.00013735706032211406,
      "loss": 1.9823,
      "step": 157224
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.1702419519424438,
      "learning_rate": 0.0001373536230944696,
      "loss": 2.0722,
      "step": 157225
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.2173492908477783,
      "learning_rate": 0.0001373501858970643,
      "loss": 1.9253,
      "step": 157226
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.3170055150985718,
      "learning_rate": 0.00013734674872989863,
      "loss": 2.1154,
      "step": 157227
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.2553457021713257,
      "learning_rate": 0.00013734331159297332,
      "loss": 2.2132,
      "step": 157228
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.28862726688385,
      "learning_rate": 0.00013733987448628882,
      "loss": 2.1205,
      "step": 157229
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.5119400024414062,
      "learning_rate": 0.00013733643740984603,
      "loss": 2.1214,
      "step": 157230
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.1348814964294434,
      "learning_rate": 0.00013733300036364535,
      "loss": 1.8037,
      "step": 157231
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.2800698280334473,
      "learning_rate": 0.00013732956334768764,
      "loss": 1.931,
      "step": 157232
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.1646124124526978,
      "learning_rate": 0.00013732612636197344,
      "loss": 1.939,
      "step": 157233
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.3630707263946533,
      "learning_rate": 0.0001373226894065033,
      "loss": 2.061,
      "step": 157234
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.2489211559295654,
      "learning_rate": 0.00013731925248127802,
      "loss": 1.9852,
      "step": 157235
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.2606841325759888,
      "learning_rate": 0.00013731581558629817,
      "loss": 1.9655,
      "step": 157236
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.1081469058990479,
      "learning_rate": 0.00013731237872156428,
      "loss": 2.1312,
      "step": 157237
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.358599305152893,
      "learning_rate": 0.00013730894188707717,
      "loss": 1.9476,
      "step": 157238
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.4409210681915283,
      "learning_rate": 0.00013730550508283745,
      "loss": 2.0197,
      "step": 157239
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.255307674407959,
      "learning_rate": 0.00013730206830884554,
      "loss": 1.9572,
      "step": 157240
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.2545394897460938,
      "learning_rate": 0.00013729863156510238,
      "loss": 1.967,
      "step": 157241
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.2397984266281128,
      "learning_rate": 0.00013729519485160851,
      "loss": 2.0885,
      "step": 157242
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.1986629962921143,
      "learning_rate": 0.00013729175816836438,
      "loss": 2.1096,
      "step": 157243
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.3018887042999268,
      "learning_rate": 0.0001372883215153709,
      "loss": 1.9855,
      "step": 157244
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.5732157230377197,
      "learning_rate": 0.00013728488489262862,
      "loss": 2.2982,
      "step": 157245
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.2351505756378174,
      "learning_rate": 0.000137281448300138,
      "loss": 2.225,
      "step": 157246
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.2806059122085571,
      "learning_rate": 0.0001372780117379,
      "loss": 2.2748,
      "step": 157247
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.2570726871490479,
      "learning_rate": 0.0001372745752059149,
      "loss": 2.2342,
      "step": 157248
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.2046098709106445,
      "learning_rate": 0.0001372711387041837,
      "loss": 1.9034,
      "step": 157249
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.1315585374832153,
      "learning_rate": 0.00013726770223270685,
      "loss": 2.0911,
      "step": 157250
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.2920074462890625,
      "learning_rate": 0.00013726426579148502,
      "loss": 1.9671,
      "step": 157251
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.2395955324172974,
      "learning_rate": 0.0001372608293805187,
      "loss": 2.1616,
      "step": 157252
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.1919834613800049,
      "learning_rate": 0.00013725739299980878,
      "loss": 2.1583,
      "step": 157253
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.1001713275909424,
      "learning_rate": 0.00013725395664935567,
      "loss": 1.9767,
      "step": 157254
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.1876814365386963,
      "learning_rate": 0.00013725052032916024,
      "loss": 2.1048,
      "step": 157255
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.16527259349823,
      "learning_rate": 0.00013724708403922302,
      "loss": 2.1584,
      "step": 157256
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.2509671449661255,
      "learning_rate": 0.00013724364777954463,
      "loss": 2.1642,
      "step": 157257
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.049814224243164,
      "learning_rate": 0.00013724021155012557,
      "loss": 2.0627,
      "step": 157258
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.10073721408844,
      "learning_rate": 0.0001372367753509668,
      "loss": 2.1515,
      "step": 157259
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.251838207244873,
      "learning_rate": 0.00013723333918206864,
      "loss": 2.219,
      "step": 157260
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.2180334329605103,
      "learning_rate": 0.000137229903043432,
      "loss": 2.1596,
      "step": 157261
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.2134467363357544,
      "learning_rate": 0.00013722646693505735,
      "loss": 2.0988,
      "step": 157262
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.1937335729599,
      "learning_rate": 0.00013722303085694543,
      "loss": 2.0706,
      "step": 157263
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.2051422595977783,
      "learning_rate": 0.0001372195948090967,
      "loss": 1.9968,
      "step": 157264
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.2192431688308716,
      "learning_rate": 0.000137216158791512,
      "loss": 2.1928,
      "step": 157265
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.1379445791244507,
      "learning_rate": 0.0001372127228041918,
      "loss": 2.2005,
      "step": 157266
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.1901923418045044,
      "learning_rate": 0.00013720928684713693,
      "loss": 2.279,
      "step": 157267
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.2926520109176636,
      "learning_rate": 0.00013720585092034794,
      "loss": 2.1781,
      "step": 157268
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.208806037902832,
      "learning_rate": 0.00013720241502382545,
      "loss": 2.2197,
      "step": 157269
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.2618401050567627,
      "learning_rate": 0.00013719897915757,
      "loss": 2.0758,
      "step": 157270
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.1563899517059326,
      "learning_rate": 0.00013719554332158242,
      "loss": 1.9846,
      "step": 157271
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.1126476526260376,
      "learning_rate": 0.00013719210751586317,
      "loss": 2.0924,
      "step": 157272
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.1448098421096802,
      "learning_rate": 0.00013718867174041308,
      "loss": 1.9994,
      "step": 157273
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.2879698276519775,
      "learning_rate": 0.0001371852359952326,
      "loss": 1.93,
      "step": 157274
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.3501248359680176,
      "learning_rate": 0.00013718180028032264,
      "loss": 2.0768,
      "step": 157275
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.198951005935669,
      "learning_rate": 0.00013717836459568346,
      "loss": 2.1531,
      "step": 157276
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.2289286851882935,
      "learning_rate": 0.00013717492894131604,
      "loss": 1.9126,
      "step": 157277
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.178823709487915,
      "learning_rate": 0.0001371714933172207,
      "loss": 1.9266,
      "step": 157278
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.1751129627227783,
      "learning_rate": 0.00013716805772339842,
      "loss": 1.9893,
      "step": 157279
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.2812062501907349,
      "learning_rate": 0.0001371646221598495,
      "loss": 2.0484,
      "step": 157280
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.2378233671188354,
      "learning_rate": 0.00013716118662657504,
      "loss": 1.9899,
      "step": 157281
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.2181929349899292,
      "learning_rate": 0.00013715775112357512,
      "loss": 1.9389,
      "step": 157282
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.2471094131469727,
      "learning_rate": 0.00013715431565085076,
      "loss": 1.9143,
      "step": 157283
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.2601072788238525,
      "learning_rate": 0.0001371508802084024,
      "loss": 1.974,
      "step": 157284
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.3698862791061401,
      "learning_rate": 0.00013714744479623086,
      "loss": 2.0135,
      "step": 157285
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.2408510446548462,
      "learning_rate": 0.00013714400941433657,
      "loss": 2.1268,
      "step": 157286
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.149236798286438,
      "learning_rate": 0.00013714057406272054,
      "loss": 2.0315,
      "step": 157287
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.3272308111190796,
      "learning_rate": 0.0001371371387413829,
      "loss": 2.0868,
      "step": 157288
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.2599608898162842,
      "learning_rate": 0.00013713370345032466,
      "loss": 2.1645,
      "step": 157289
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.2614495754241943,
      "learning_rate": 0.0001371302681895462,
      "loss": 2.119,
      "step": 157290
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.185215711593628,
      "learning_rate": 0.00013712683295904842,
      "loss": 2.2508,
      "step": 157291
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.3316062688827515,
      "learning_rate": 0.00013712339775883174,
      "loss": 2.0882,
      "step": 157292
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.3838660717010498,
      "learning_rate": 0.00013711996258889713,
      "loss": 1.951,
      "step": 157293
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.1751762628555298,
      "learning_rate": 0.00013711652744924474,
      "loss": 2.0407,
      "step": 157294
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.0886086225509644,
      "learning_rate": 0.0001371130923398756,
      "loss": 2.1127,
      "step": 157295
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.321735143661499,
      "learning_rate": 0.00013710965726079007,
      "loss": 2.0452,
      "step": 157296
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.3725526332855225,
      "learning_rate": 0.00013710622221198906,
      "loss": 1.9092,
      "step": 157297
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.251542329788208,
      "learning_rate": 0.00013710278719347295,
      "loss": 1.9117,
      "step": 157298
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.2831225395202637,
      "learning_rate": 0.00013709935220524276,
      "loss": 2.0903,
      "step": 157299
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.2033624649047852,
      "learning_rate": 0.00013709591724729862,
      "loss": 2.2952,
      "step": 157300
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.3329745531082153,
      "learning_rate": 0.00013709248231964154,
      "loss": 1.9741,
      "step": 157301
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.175262212753296,
      "learning_rate": 0.000137089047422272,
      "loss": 2.1819,
      "step": 157302
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.1485326290130615,
      "learning_rate": 0.00013708561255519071,
      "loss": 1.961,
      "step": 157303
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.3397730588912964,
      "learning_rate": 0.0001370821777183982,
      "loss": 2.0177,
      "step": 157304
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.1518161296844482,
      "learning_rate": 0.0001370787429118954,
      "loss": 1.8823,
      "step": 157305
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.1267592906951904,
      "learning_rate": 0.0001370753081356825,
      "loss": 1.9108,
      "step": 157306
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.2287440299987793,
      "learning_rate": 0.00013707187338976053,
      "loss": 1.9906,
      "step": 157307
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.2355858087539673,
      "learning_rate": 0.00013706843867412983,
      "loss": 2.033,
      "step": 157308
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.1295924186706543,
      "learning_rate": 0.0001370650039887913,
      "loss": 2.0958,
      "step": 157309
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.2787084579467773,
      "learning_rate": 0.00013706156933374536,
      "loss": 2.01,
      "step": 157310
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.0486496686935425,
      "learning_rate": 0.00013705813470899286,
      "loss": 2.1392,
      "step": 157311
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.1974691152572632,
      "learning_rate": 0.00013705470011453433,
      "loss": 2.1825,
      "step": 157312
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.1334697008132935,
      "learning_rate": 0.0001370512655503704,
      "loss": 2.028,
      "step": 157313
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.2432490587234497,
      "learning_rate": 0.00013704783101650157,
      "loss": 1.9999,
      "step": 157314
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.2357574701309204,
      "learning_rate": 0.0001370443965129288,
      "loss": 2.0383,
      "step": 157315
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.1810816526412964,
      "learning_rate": 0.0001370409620396524,
      "loss": 2.3543,
      "step": 157316
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.1970224380493164,
      "learning_rate": 0.00013703752759667328,
      "loss": 2.0339,
      "step": 157317
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.103117823600769,
      "learning_rate": 0.00013703409318399198,
      "loss": 1.85,
      "step": 157318
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.075274109840393,
      "learning_rate": 0.00013703065880160898,
      "loss": 1.8445,
      "step": 157319
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.3146196603775024,
      "learning_rate": 0.00013702722444952517,
      "loss": 1.827,
      "step": 157320
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.0864051580429077,
      "learning_rate": 0.0001370237901277411,
      "loss": 2.0795,
      "step": 157321
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.1325818300247192,
      "learning_rate": 0.00013702035583625726,
      "loss": 2.0403,
      "step": 157322
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.2043083906173706,
      "learning_rate": 0.0001370169215750745,
      "loss": 2.1492,
      "step": 157323
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.1161243915557861,
      "learning_rate": 0.0001370134873441934,
      "loss": 2.1161,
      "step": 157324
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.430751085281372,
      "learning_rate": 0.00013701005314361447,
      "loss": 2.0705,
      "step": 157325
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.3675528764724731,
      "learning_rate": 0.00013700661897333852,
      "loss": 1.9874,
      "step": 157326
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.1442898511886597,
      "learning_rate": 0.00013700318483336613,
      "loss": 2.1778,
      "step": 157327
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.220157504081726,
      "learning_rate": 0.00013699975072369781,
      "loss": 1.9852,
      "step": 157328
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.3220293521881104,
      "learning_rate": 0.00013699631664433444,
      "loss": 2.1497,
      "step": 157329
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.2637096643447876,
      "learning_rate": 0.00013699288259527652,
      "loss": 2.2158,
      "step": 157330
    },
    {
      "epoch": 2.05,
      "grad_norm": 2.001157283782959,
      "learning_rate": 0.0001369894485765246,
      "loss": 2.1564,
      "step": 157331
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.1523219347000122,
      "learning_rate": 0.00013698601458807954,
      "loss": 2.0837,
      "step": 157332
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.2657432556152344,
      "learning_rate": 0.00013698258062994173,
      "loss": 2.0735,
      "step": 157333
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.1296058893203735,
      "learning_rate": 0.00013697914670211204,
      "loss": 2.2374,
      "step": 157334
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.1773498058319092,
      "learning_rate": 0.000136975712804591,
      "loss": 1.961,
      "step": 157335
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.3473888635635376,
      "learning_rate": 0.00013697227893737926,
      "loss": 2.3128,
      "step": 157336
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.2659300565719604,
      "learning_rate": 0.00013696884510047734,
      "loss": 1.8908,
      "step": 157337
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.1780180931091309,
      "learning_rate": 0.00013696541129388612,
      "loss": 1.9785,
      "step": 157338
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.2993518114089966,
      "learning_rate": 0.00013696197751760596,
      "loss": 1.8869,
      "step": 157339
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.1820701360702515,
      "learning_rate": 0.00013695854377163776,
      "loss": 2.123,
      "step": 157340
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.1447734832763672,
      "learning_rate": 0.0001369551100559821,
      "loss": 1.9086,
      "step": 157341
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.2159432172775269,
      "learning_rate": 0.0001369516763706395,
      "loss": 2.0031,
      "step": 157342
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.2166191339492798,
      "learning_rate": 0.0001369482427156106,
      "loss": 2.2135,
      "step": 157343
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.172863483428955,
      "learning_rate": 0.00013694480909089618,
      "loss": 2.1369,
      "step": 157344
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.1373927593231201,
      "learning_rate": 0.00013694137549649665,
      "loss": 2.1068,
      "step": 157345
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.3628332614898682,
      "learning_rate": 0.000136937941932413,
      "loss": 2.1237,
      "step": 157346
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.1683977842330933,
      "learning_rate": 0.00013693450839864546,
      "loss": 1.9257,
      "step": 157347
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.1624393463134766,
      "learning_rate": 0.00013693107489519516,
      "loss": 1.9143,
      "step": 157348
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.1493502855300903,
      "learning_rate": 0.00013692764142206218,
      "loss": 2.1083,
      "step": 157349
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.1898643970489502,
      "learning_rate": 0.00013692420797924759,
      "loss": 2.192,
      "step": 157350
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.2252813577651978,
      "learning_rate": 0.0001369207745667517,
      "loss": 1.9456,
      "step": 157351
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.1962019205093384,
      "learning_rate": 0.00013691734118457546,
      "loss": 2.3057,
      "step": 157352
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.0986194610595703,
      "learning_rate": 0.00013691390783271927,
      "loss": 2.0006,
      "step": 157353
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.2755632400512695,
      "learning_rate": 0.00013691047451118408,
      "loss": 1.9313,
      "step": 157354
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.1211520433425903,
      "learning_rate": 0.00013690704121997006,
      "loss": 1.8896,
      "step": 157355
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.0907719135284424,
      "learning_rate": 0.00013690360795907822,
      "loss": 2.1469,
      "step": 157356
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.2112855911254883,
      "learning_rate": 0.000136900174728509,
      "loss": 1.976,
      "step": 157357
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.2211717367172241,
      "learning_rate": 0.0001368967415282632,
      "loss": 2.0006,
      "step": 157358
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.1021941900253296,
      "learning_rate": 0.00013689330835834124,
      "loss": 2.1041,
      "step": 157359
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.235049843788147,
      "learning_rate": 0.00013688987521874417,
      "loss": 1.8517,
      "step": 157360
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.2196202278137207,
      "learning_rate": 0.0001368864421094721,
      "loss": 1.9914,
      "step": 157361
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.3477652072906494,
      "learning_rate": 0.00013688300903052603,
      "loss": 2.2326,
      "step": 157362
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.3039629459381104,
      "learning_rate": 0.00013687957598190636,
      "loss": 1.9305,
      "step": 157363
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.1754727363586426,
      "learning_rate": 0.00013687614296361398,
      "loss": 2.1191,
      "step": 157364
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.0946553945541382,
      "learning_rate": 0.00013687270997564933,
      "loss": 1.831,
      "step": 157365
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.228468418121338,
      "learning_rate": 0.00013686927701801329,
      "loss": 1.8367,
      "step": 157366
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.372209906578064,
      "learning_rate": 0.00013686584409070614,
      "loss": 2.1445,
      "step": 157367
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.3199000358581543,
      "learning_rate": 0.00013686241119372882,
      "loss": 2.1355,
      "step": 157368
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.157584309577942,
      "learning_rate": 0.00013685897832708174,
      "loss": 2.261,
      "step": 157369
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.2139830589294434,
      "learning_rate": 0.00013685554549076577,
      "loss": 2.0694,
      "step": 157370
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.1731802225112915,
      "learning_rate": 0.0001368521126847813,
      "loss": 1.9291,
      "step": 157371
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.1067824363708496,
      "learning_rate": 0.0001368486799091293,
      "loss": 2.0657,
      "step": 157372
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.0694698095321655,
      "learning_rate": 0.00013684524716381006,
      "loss": 2.2238,
      "step": 157373
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.2404078245162964,
      "learning_rate": 0.00013684181444882445,
      "loss": 1.9285,
      "step": 157374
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.1485093832015991,
      "learning_rate": 0.0001368383817641729,
      "loss": 1.8904,
      "step": 157375
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.2379512786865234,
      "learning_rate": 0.0001368349491098563,
      "loss": 2.1125,
      "step": 157376
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.1275911331176758,
      "learning_rate": 0.00013683151648587506,
      "loss": 1.9506,
      "step": 157377
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.082059621810913,
      "learning_rate": 0.00013682808389223013,
      "loss": 2.3538,
      "step": 157378
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.0606228113174438,
      "learning_rate": 0.00013682465132892172,
      "loss": 1.9198,
      "step": 157379
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.0945895910263062,
      "learning_rate": 0.0001368212187959508,
      "loss": 2.0283,
      "step": 157380
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.281073808670044,
      "learning_rate": 0.00013681778629331777,
      "loss": 2.015,
      "step": 157381
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.075637936592102,
      "learning_rate": 0.00013681435382102355,
      "loss": 2.1326,
      "step": 157382
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.189146876335144,
      "learning_rate": 0.0001368109213790685,
      "loss": 2.1257,
      "step": 157383
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.1406837701797485,
      "learning_rate": 0.00013680748896745354,
      "loss": 1.8943,
      "step": 157384
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.2912960052490234,
      "learning_rate": 0.00013680405658617897,
      "loss": 2.001,
      "step": 157385
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.215174913406372,
      "learning_rate": 0.00013680062423524573,
      "loss": 1.945,
      "step": 157386
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.0748405456542969,
      "learning_rate": 0.00013679719191465422,
      "loss": 1.9467,
      "step": 157387
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.3556543588638306,
      "learning_rate": 0.00013679375962440528,
      "loss": 2.0549,
      "step": 157388
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.1454235315322876,
      "learning_rate": 0.00013679032736449936,
      "loss": 2.1838,
      "step": 157389
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.4148002862930298,
      "learning_rate": 0.00013678689513493743,
      "loss": 2.1076,
      "step": 157390
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.2320080995559692,
      "learning_rate": 0.00013678346293571965,
      "loss": 2.0781,
      "step": 157391
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.1888378858566284,
      "learning_rate": 0.00013678003076684705,
      "loss": 1.9335,
      "step": 157392
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.2328298091888428,
      "learning_rate": 0.00013677659862832,
      "loss": 1.8669,
      "step": 157393
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.1767873764038086,
      "learning_rate": 0.00013677316652013937,
      "loss": 1.998,
      "step": 157394
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.1316417455673218,
      "learning_rate": 0.00013676973444230557,
      "loss": 2.1028,
      "step": 157395
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.2229523658752441,
      "learning_rate": 0.0001367663023948195,
      "loss": 2.1637,
      "step": 157396
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.051997184753418,
      "learning_rate": 0.00013676287037768162,
      "loss": 2.047,
      "step": 157397
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.061346411705017,
      "learning_rate": 0.00013675943839089262,
      "loss": 2.12,
      "step": 157398
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.1800518035888672,
      "learning_rate": 0.00013675600643445302,
      "loss": 2.0702,
      "step": 157399
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.1924545764923096,
      "learning_rate": 0.00013675257450836365,
      "loss": 1.9688,
      "step": 157400
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.1352229118347168,
      "learning_rate": 0.00013674914261262494,
      "loss": 2.0096,
      "step": 157401
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.1936092376708984,
      "learning_rate": 0.00013674571074723778,
      "loss": 2.1697,
      "step": 157402
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.3209623098373413,
      "learning_rate": 0.00013674227891220266,
      "loss": 2.1357,
      "step": 157403
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.2892799377441406,
      "learning_rate": 0.00013673884710752027,
      "loss": 2.1628,
      "step": 157404
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.266029953956604,
      "learning_rate": 0.00013673541533319107,
      "loss": 2.0274,
      "step": 157405
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.2105036973953247,
      "learning_rate": 0.00013673198358921592,
      "loss": 2.2051,
      "step": 157406
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.3499804735183716,
      "learning_rate": 0.0001367285518755953,
      "loss": 2.0158,
      "step": 157407
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.203398585319519,
      "learning_rate": 0.00013672512019233005,
      "loss": 2.1857,
      "step": 157408
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.0869094133377075,
      "learning_rate": 0.00013672168853942065,
      "loss": 2.0603,
      "step": 157409
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.1524639129638672,
      "learning_rate": 0.00013671825691686766,
      "loss": 2.2244,
      "step": 157410
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.2234907150268555,
      "learning_rate": 0.00013671482532467196,
      "loss": 1.9582,
      "step": 157411
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.195914626121521,
      "learning_rate": 0.00013671139376283402,
      "loss": 1.8983,
      "step": 157412
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.2490239143371582,
      "learning_rate": 0.00013670796223135443,
      "loss": 2.2231,
      "step": 157413
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.165019154548645,
      "learning_rate": 0.000136704530730234,
      "loss": 2.1493,
      "step": 157414
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.0943022966384888,
      "learning_rate": 0.00013670109925947334,
      "loss": 1.973,
      "step": 157415
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.1713035106658936,
      "learning_rate": 0.00013669766781907288,
      "loss": 2.158,
      "step": 157416
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.2862130403518677,
      "learning_rate": 0.0001366942364090335,
      "loss": 1.9376,
      "step": 157417
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.1693944931030273,
      "learning_rate": 0.00013669080502935564,
      "loss": 1.8758,
      "step": 157418
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.3273146152496338,
      "learning_rate": 0.00013668737368004017,
      "loss": 2.3186,
      "step": 157419
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.3793063163757324,
      "learning_rate": 0.00013668394236108757,
      "loss": 1.9598,
      "step": 157420
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.2189199924468994,
      "learning_rate": 0.00013668051107249856,
      "loss": 1.8977,
      "step": 157421
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.1430083513259888,
      "learning_rate": 0.00013667707981427355,
      "loss": 2.0842,
      "step": 157422
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.186689019203186,
      "learning_rate": 0.00013667364858641352,
      "loss": 2.0103,
      "step": 157423
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.3549848794937134,
      "learning_rate": 0.00013667021738891883,
      "loss": 2.278,
      "step": 157424
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.1441807746887207,
      "learning_rate": 0.0001366667862217903,
      "loss": 1.8935,
      "step": 157425
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.1738356351852417,
      "learning_rate": 0.00013666335508502852,
      "loss": 2.0697,
      "step": 157426
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.1049914360046387,
      "learning_rate": 0.00013665992397863408,
      "loss": 1.9436,
      "step": 157427
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.128534197807312,
      "learning_rate": 0.00013665649290260756,
      "loss": 2.0544,
      "step": 157428
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.2363680601119995,
      "learning_rate": 0.0001366530618569498,
      "loss": 2.2686,
      "step": 157429
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.2949069738388062,
      "learning_rate": 0.0001366496308416612,
      "loss": 1.9588,
      "step": 157430
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.2172529697418213,
      "learning_rate": 0.00013664619985674263,
      "loss": 1.9506,
      "step": 157431
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.3135653734207153,
      "learning_rate": 0.0001366427689021945,
      "loss": 2.1084,
      "step": 157432
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.2017160654067993,
      "learning_rate": 0.00013663933797801776,
      "loss": 2.2992,
      "step": 157433
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.3589794635772705,
      "learning_rate": 0.00013663590708421267,
      "loss": 2.0554,
      "step": 157434
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.1920874118804932,
      "learning_rate": 0.00013663247622078015,
      "loss": 2.0641,
      "step": 157435
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.1605746746063232,
      "learning_rate": 0.0001366290453877206,
      "loss": 2.1569,
      "step": 157436
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.1117515563964844,
      "learning_rate": 0.00013662561458503494,
      "loss": 2.0166,
      "step": 157437
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.315335750579834,
      "learning_rate": 0.0001366221838127235,
      "loss": 1.8857,
      "step": 157438
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.1449207067489624,
      "learning_rate": 0.00013661875307078734,
      "loss": 2.0015,
      "step": 157439
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.2634234428405762,
      "learning_rate": 0.00013661532235922664,
      "loss": 2.0613,
      "step": 157440
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.2082279920578003,
      "learning_rate": 0.0001366118916780423,
      "loss": 2.1066,
      "step": 157441
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.1886122226715088,
      "learning_rate": 0.0001366084610272348,
      "loss": 1.8452,
      "step": 157442
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.409496545791626,
      "learning_rate": 0.00013660503040680496,
      "loss": 2.1255,
      "step": 157443
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.4357409477233887,
      "learning_rate": 0.00013660159981675326,
      "loss": 2.1185,
      "step": 157444
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.151572346687317,
      "learning_rate": 0.00013659816925708064,
      "loss": 2.0283,
      "step": 157445
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.1241648197174072,
      "learning_rate": 0.00013659473872778722,
      "loss": 1.9125,
      "step": 157446
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.7734723091125488,
      "learning_rate": 0.00013659130822887407,
      "loss": 2.1486,
      "step": 157447
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.3903002738952637,
      "learning_rate": 0.0001365878777603416,
      "loss": 1.9997,
      "step": 157448
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.1820528507232666,
      "learning_rate": 0.00013658444732219058,
      "loss": 2.2224,
      "step": 157449
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.2613651752471924,
      "learning_rate": 0.0001365810169144215,
      "loss": 1.9615,
      "step": 157450
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.2701793909072876,
      "learning_rate": 0.00013657758653703533,
      "loss": 2.1557,
      "step": 157451
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.2333296537399292,
      "learning_rate": 0.00013657415619003225,
      "loss": 2.2825,
      "step": 157452
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.303471565246582,
      "learning_rate": 0.00013657072587341322,
      "loss": 2.1505,
      "step": 157453
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.381894826889038,
      "learning_rate": 0.00013656729558717868,
      "loss": 1.9087,
      "step": 157454
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.2288272380828857,
      "learning_rate": 0.0001365638653313295,
      "loss": 2.0534,
      "step": 157455
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.3033751249313354,
      "learning_rate": 0.000136560435105866,
      "loss": 1.8837,
      "step": 157456
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.2644044160842896,
      "learning_rate": 0.00013655700491078926,
      "loss": 1.842,
      "step": 157457
    },
    {
      "epoch": 2.05,
      "grad_norm": 2.1516599655151367,
      "learning_rate": 0.00013655357474609943,
      "loss": 2.0428,
      "step": 157458
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.3079043626785278,
      "learning_rate": 0.0001365501446117975,
      "loss": 2.2316,
      "step": 157459
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.259958028793335,
      "learning_rate": 0.00013654671450788388,
      "loss": 1.9327,
      "step": 157460
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.1316094398498535,
      "learning_rate": 0.00013654328443435938,
      "loss": 2.0238,
      "step": 157461
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.3204970359802246,
      "learning_rate": 0.0001365398543912245,
      "loss": 2.0411,
      "step": 157462
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.6811162233352661,
      "learning_rate": 0.00013653642437848017,
      "loss": 2.0044,
      "step": 157463
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.3312935829162598,
      "learning_rate": 0.00013653299439612656,
      "loss": 1.9574,
      "step": 157464
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.5020201206207275,
      "learning_rate": 0.00013652956444416467,
      "loss": 2.112,
      "step": 157465
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.109351634979248,
      "learning_rate": 0.0001365261345225949,
      "loss": 1.8733,
      "step": 157466
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.2856439352035522,
      "learning_rate": 0.00013652270463141813,
      "loss": 2.1959,
      "step": 157467
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.0969575643539429,
      "learning_rate": 0.00013651927477063478,
      "loss": 1.9218,
      "step": 157468
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.4384558200836182,
      "learning_rate": 0.00013651584494024578,
      "loss": 1.8331,
      "step": 157469
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.272098422050476,
      "learning_rate": 0.00013651241514025137,
      "loss": 1.9073,
      "step": 157470
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.5045503377914429,
      "learning_rate": 0.00013650898537065248,
      "loss": 1.7921,
      "step": 157471
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.4138721227645874,
      "learning_rate": 0.00013650555563144952,
      "loss": 2.0914,
      "step": 157472
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.3124616146087646,
      "learning_rate": 0.0001365021259226434,
      "loss": 2.0127,
      "step": 157473
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.1971540451049805,
      "learning_rate": 0.00013649869624423453,
      "loss": 2.1477,
      "step": 157474
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.3485465049743652,
      "learning_rate": 0.00013649526659622373,
      "loss": 2.0653,
      "step": 157475
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.1521509885787964,
      "learning_rate": 0.00013649183697861153,
      "loss": 2.0585,
      "step": 157476
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.1773797273635864,
      "learning_rate": 0.00013648840739139862,
      "loss": 2.1633,
      "step": 157477
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.2808876037597656,
      "learning_rate": 0.00013648497783458547,
      "loss": 1.9467,
      "step": 157478
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.3809821605682373,
      "learning_rate": 0.000136481548308173,
      "loss": 2.3729,
      "step": 157479
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.2846989631652832,
      "learning_rate": 0.00013647811881216152,
      "loss": 2.0103,
      "step": 157480
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.091424822807312,
      "learning_rate": 0.000136474689346552,
      "loss": 2.0617,
      "step": 157481
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.1666325330734253,
      "learning_rate": 0.0001364712599113449,
      "loss": 1.9039,
      "step": 157482
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.3453072309494019,
      "learning_rate": 0.00013646783050654088,
      "loss": 2.0552,
      "step": 157483
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.282531976699829,
      "learning_rate": 0.00013646440113214048,
      "loss": 2.1962,
      "step": 157484
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.2283133268356323,
      "learning_rate": 0.00013646097178814453,
      "loss": 2.1437,
      "step": 157485
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.1707974672317505,
      "learning_rate": 0.00013645754247455346,
      "loss": 1.9956,
      "step": 157486
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.150841474533081,
      "learning_rate": 0.00013645411319136816,
      "loss": 2.1114,
      "step": 157487
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.178274154663086,
      "learning_rate": 0.00013645068393858912,
      "loss": 2.0311,
      "step": 157488
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.2660589218139648,
      "learning_rate": 0.00013644725471621694,
      "loss": 2.1077,
      "step": 157489
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.1807869672775269,
      "learning_rate": 0.00013644382552425224,
      "loss": 2.0962,
      "step": 157490
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.1341636180877686,
      "learning_rate": 0.00013644039636269583,
      "loss": 1.9096,
      "step": 157491
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.220300555229187,
      "learning_rate": 0.00013643696723154812,
      "loss": 2.0784,
      "step": 157492
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.2773807048797607,
      "learning_rate": 0.00013643353813080995,
      "loss": 1.9913,
      "step": 157493
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.1902248859405518,
      "learning_rate": 0.00013643010906048188,
      "loss": 2.1178,
      "step": 157494
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.37015700340271,
      "learning_rate": 0.00013642668002056445,
      "loss": 2.1378,
      "step": 157495
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.1496005058288574,
      "learning_rate": 0.00013642325101105847,
      "loss": 2.1033,
      "step": 157496
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.1792428493499756,
      "learning_rate": 0.00013641982203196453,
      "loss": 1.953,
      "step": 157497
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.1460697650909424,
      "learning_rate": 0.00013641639308328312,
      "loss": 2.0847,
      "step": 157498
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.1593883037567139,
      "learning_rate": 0.0001364129641650151,
      "loss": 2.0483,
      "step": 157499
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.0920885801315308,
      "learning_rate": 0.00013640953527716096,
      "loss": 2.0107,
      "step": 157500
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.1363190412521362,
      "learning_rate": 0.00013640610641972128,
      "loss": 2.0872,
      "step": 157501
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.0872787237167358,
      "learning_rate": 0.00013640267759269694,
      "loss": 1.9575,
      "step": 157502
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.078924298286438,
      "learning_rate": 0.0001363992487960884,
      "loss": 1.9725,
      "step": 157503
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.1230896711349487,
      "learning_rate": 0.00013639582002989624,
      "loss": 2.1363,
      "step": 157504
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.1234619617462158,
      "learning_rate": 0.00013639239129412128,
      "loss": 1.936,
      "step": 157505
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.2340741157531738,
      "learning_rate": 0.00013638896258876405,
      "loss": 2.0814,
      "step": 157506
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.1907521486282349,
      "learning_rate": 0.00013638553391382511,
      "loss": 1.9843,
      "step": 157507
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.174243688583374,
      "learning_rate": 0.00013638210526930529,
      "loss": 1.9657,
      "step": 157508
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.305381178855896,
      "learning_rate": 0.000136378676655205,
      "loss": 2.0376,
      "step": 157509
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.2242599725723267,
      "learning_rate": 0.00013637524807152512,
      "loss": 2.036,
      "step": 157510
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.1842279434204102,
      "learning_rate": 0.00013637181951826617,
      "loss": 2.1993,
      "step": 157511
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.2570198774337769,
      "learning_rate": 0.00013636839099542882,
      "loss": 2.1081,
      "step": 157512
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.1661853790283203,
      "learning_rate": 0.00013636496250301355,
      "loss": 1.96,
      "step": 157513
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.3439507484436035,
      "learning_rate": 0.0001363615340410212,
      "loss": 1.9276,
      "step": 157514
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.0531138181686401,
      "learning_rate": 0.00013635810560945223,
      "loss": 2.0475,
      "step": 157515
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.1693373918533325,
      "learning_rate": 0.00013635467720830751,
      "loss": 2.1082,
      "step": 157516
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.4943032264709473,
      "learning_rate": 0.00013635124883758744,
      "loss": 2.1859,
      "step": 157517
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.1646541357040405,
      "learning_rate": 0.00013634782049729295,
      "loss": 2.115,
      "step": 157518
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.2540878057479858,
      "learning_rate": 0.0001363443921874243,
      "loss": 2.1803,
      "step": 157519
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.1686440706253052,
      "learning_rate": 0.0001363409639079824,
      "loss": 2.0402,
      "step": 157520
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.1614792346954346,
      "learning_rate": 0.0001363375356589677,
      "loss": 2.064,
      "step": 157521
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.2441433668136597,
      "learning_rate": 0.00013633410744038105,
      "loss": 1.879,
      "step": 157522
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.1583997011184692,
      "learning_rate": 0.0001363306792522229,
      "loss": 1.9469,
      "step": 157523
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.2681496143341064,
      "learning_rate": 0.00013632725109449413,
      "loss": 2.0545,
      "step": 157524
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.1685372591018677,
      "learning_rate": 0.00013632382296719503,
      "loss": 2.2556,
      "step": 157525
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.967779278755188,
      "learning_rate": 0.00013632039487032654,
      "loss": 1.9414,
      "step": 157526
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.1976749897003174,
      "learning_rate": 0.00013631696680388905,
      "loss": 2.0604,
      "step": 157527
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.1639124155044556,
      "learning_rate": 0.00013631353876788346,
      "loss": 2.0268,
      "step": 157528
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.0510002374649048,
      "learning_rate": 0.00013631011076231013,
      "loss": 2.2449,
      "step": 157529
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.1221632957458496,
      "learning_rate": 0.00013630668278717008,
      "loss": 2.1594,
      "step": 157530
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.111911654472351,
      "learning_rate": 0.00013630325484246348,
      "loss": 2.0924,
      "step": 157531
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.1031402349472046,
      "learning_rate": 0.0001362998269281913,
      "loss": 2.0491,
      "step": 157532
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.245531678199768,
      "learning_rate": 0.00013629639904435395,
      "loss": 2.0244,
      "step": 157533
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.1841639280319214,
      "learning_rate": 0.00013629297119095233,
      "loss": 2.2542,
      "step": 157534
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.2457661628723145,
      "learning_rate": 0.00013628954336798683,
      "loss": 1.9727,
      "step": 157535
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.1732150316238403,
      "learning_rate": 0.00013628611557545839,
      "loss": 2.2656,
      "step": 157536
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.1714893579483032,
      "learning_rate": 0.00013628268781336724,
      "loss": 2.1396,
      "step": 157537
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.1403177976608276,
      "learning_rate": 0.00013627926008171433,
      "loss": 2.0599,
      "step": 157538
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.19217848777771,
      "learning_rate": 0.0001362758323805001,
      "loss": 1.958,
      "step": 157539
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.258434772491455,
      "learning_rate": 0.00013627240470972537,
      "loss": 2.0671,
      "step": 157540
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.2310782670974731,
      "learning_rate": 0.0001362689770693906,
      "loss": 1.9236,
      "step": 157541
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.2178884744644165,
      "learning_rate": 0.00013626554945949664,
      "loss": 2.0421,
      "step": 157542
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.213536024093628,
      "learning_rate": 0.000136262121880044,
      "loss": 1.8526,
      "step": 157543
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.203863501548767,
      "learning_rate": 0.00013625869433103334,
      "loss": 2.1172,
      "step": 157544
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.0264629125595093,
      "learning_rate": 0.00013625526681246514,
      "loss": 1.7396,
      "step": 157545
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.1547373533248901,
      "learning_rate": 0.00013625183932434027,
      "loss": 2.1049,
      "step": 157546
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.075940728187561,
      "learning_rate": 0.00013624841186665916,
      "loss": 1.9539,
      "step": 157547
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.2016291618347168,
      "learning_rate": 0.0001362449844394227,
      "loss": 2.3089,
      "step": 157548
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.1835103034973145,
      "learning_rate": 0.0001362415570426314,
      "loss": 2.0869,
      "step": 157549
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.098198413848877,
      "learning_rate": 0.00013623812967628586,
      "loss": 2.0272,
      "step": 157550
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.2306580543518066,
      "learning_rate": 0.00013623470234038664,
      "loss": 1.984,
      "step": 157551
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.1651809215545654,
      "learning_rate": 0.0001362312750349346,
      "loss": 2.0742,
      "step": 157552
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.1419274806976318,
      "learning_rate": 0.00013622784775993015,
      "loss": 2.1518,
      "step": 157553
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.2722452878952026,
      "learning_rate": 0.00013622442051537415,
      "loss": 1.944,
      "step": 157554
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.130893349647522,
      "learning_rate": 0.0001362209933012671,
      "loss": 2.2694,
      "step": 157555
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.1531116962432861,
      "learning_rate": 0.0001362175661176097,
      "loss": 2.0163,
      "step": 157556
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.1246074438095093,
      "learning_rate": 0.00013621413896440236,
      "loss": 2.0388,
      "step": 157557
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.2116411924362183,
      "learning_rate": 0.00013621071184164607,
      "loss": 1.9223,
      "step": 157558
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.264933705329895,
      "learning_rate": 0.00013620728474934116,
      "loss": 2.0256,
      "step": 157559
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.1685811281204224,
      "learning_rate": 0.00013620385768748852,
      "loss": 2.2885,
      "step": 157560
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.1605761051177979,
      "learning_rate": 0.0001362004306560887,
      "loss": 1.9262,
      "step": 157561
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.1279020309448242,
      "learning_rate": 0.0001361970036551423,
      "loss": 1.9028,
      "step": 157562
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.2525755167007446,
      "learning_rate": 0.00013619357668464985,
      "loss": 2.1926,
      "step": 157563
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.1731014251708984,
      "learning_rate": 0.00013619014974461225,
      "loss": 2.1209,
      "step": 157564
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.4089847803115845,
      "learning_rate": 0.00013618672283502984,
      "loss": 2.1714,
      "step": 157565
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.3042628765106201,
      "learning_rate": 0.00013618329595590354,
      "loss": 2.115,
      "step": 157566
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.1277543306350708,
      "learning_rate": 0.00013617986910723385,
      "loss": 2.1173,
      "step": 157567
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.3015458583831787,
      "learning_rate": 0.00013617644228902144,
      "loss": 2.0827,
      "step": 157568
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.1541062593460083,
      "learning_rate": 0.0001361730155012668,
      "loss": 2.0695,
      "step": 157569
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.3086965084075928,
      "learning_rate": 0.0001361695887439708,
      "loss": 1.9285,
      "step": 157570
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.0778859853744507,
      "learning_rate": 0.00013616616201713382,
      "loss": 1.9156,
      "step": 157571
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.1753178834915161,
      "learning_rate": 0.0001361627353207568,
      "loss": 2.0636,
      "step": 157572
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.304821491241455,
      "learning_rate": 0.0001361593086548402,
      "loss": 1.832,
      "step": 157573
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.0424742698669434,
      "learning_rate": 0.00013615588201938467,
      "loss": 1.7574,
      "step": 157574
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.0970797538757324,
      "learning_rate": 0.00013615245541439075,
      "loss": 2.1401,
      "step": 157575
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.2323572635650635,
      "learning_rate": 0.0001361490288398593,
      "loss": 2.0851,
      "step": 157576
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.506845235824585,
      "learning_rate": 0.00013614560229579072,
      "loss": 2.1782,
      "step": 157577
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.31168794631958,
      "learning_rate": 0.00013614217578218586,
      "loss": 2.1049,
      "step": 157578
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.1051068305969238,
      "learning_rate": 0.0001361387492990453,
      "loss": 1.902,
      "step": 157579
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.1869678497314453,
      "learning_rate": 0.00013613532284636952,
      "loss": 2.1876,
      "step": 157580
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.1412665843963623,
      "learning_rate": 0.00013613189642415935,
      "loss": 1.944,
      "step": 157581
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.1666826009750366,
      "learning_rate": 0.0001361284700324154,
      "loss": 1.9635,
      "step": 157582
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.0765105485916138,
      "learning_rate": 0.00013612504367113812,
      "loss": 2.0767,
      "step": 157583
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.2948791980743408,
      "learning_rate": 0.0001361216173403284,
      "loss": 2.4771,
      "step": 157584
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.1613856554031372,
      "learning_rate": 0.0001361181910399868,
      "loss": 2.1035,
      "step": 157585
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.215928316116333,
      "learning_rate": 0.00013611476477011378,
      "loss": 1.9327,
      "step": 157586
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.2204477787017822,
      "learning_rate": 0.00013611133853071023,
      "loss": 1.9489,
      "step": 157587
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.132305383682251,
      "learning_rate": 0.0001361079123217767,
      "loss": 2.1184,
      "step": 157588
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.1916977167129517,
      "learning_rate": 0.00013610448614331366,
      "loss": 2.0319,
      "step": 157589
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.2296103239059448,
      "learning_rate": 0.00013610105999532201,
      "loss": 1.982,
      "step": 157590
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.2130526304244995,
      "learning_rate": 0.00013609763387780228,
      "loss": 1.9128,
      "step": 157591
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.1362993717193604,
      "learning_rate": 0.00013609420779075494,
      "loss": 1.9384,
      "step": 157592
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.2075484991073608,
      "learning_rate": 0.00013609078173418095,
      "loss": 1.8026,
      "step": 157593
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.1838405132293701,
      "learning_rate": 0.00013608735570808063,
      "loss": 2.198,
      "step": 157594
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.0997806787490845,
      "learning_rate": 0.0001360839297124549,
      "loss": 1.951,
      "step": 157595
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.1698260307312012,
      "learning_rate": 0.00013608050374730422,
      "loss": 2.0811,
      "step": 157596
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.1265040636062622,
      "learning_rate": 0.0001360770778126293,
      "loss": 1.8879,
      "step": 157597
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.1461939811706543,
      "learning_rate": 0.00013607365190843065,
      "loss": 2.1085,
      "step": 157598
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.1467399597167969,
      "learning_rate": 0.00013607022603470911,
      "loss": 1.8306,
      "step": 157599
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.1134172677993774,
      "learning_rate": 0.00013606680019146504,
      "loss": 2.2974,
      "step": 157600
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.201456069946289,
      "learning_rate": 0.00013606337437869942,
      "loss": 2.1471,
      "step": 157601
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.1402920484542847,
      "learning_rate": 0.0001360599485964126,
      "loss": 2.1233,
      "step": 157602
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.2564433813095093,
      "learning_rate": 0.00013605652284460554,
      "loss": 2.0496,
      "step": 157603
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.2194836139678955,
      "learning_rate": 0.00013605309712327843,
      "loss": 1.9338,
      "step": 157604
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.0733474493026733,
      "learning_rate": 0.00013604967143243226,
      "loss": 1.9829,
      "step": 157605
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.1179771423339844,
      "learning_rate": 0.00013604624577206746,
      "loss": 2.0409,
      "step": 157606
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.205466866493225,
      "learning_rate": 0.00013604282014218483,
      "loss": 2.09,
      "step": 157607
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.1867456436157227,
      "learning_rate": 0.00013603939454278485,
      "loss": 2.4179,
      "step": 157608
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.2368086576461792,
      "learning_rate": 0.00013603596897386838,
      "loss": 1.9697,
      "step": 157609
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.1806150674819946,
      "learning_rate": 0.00013603254343543585,
      "loss": 1.8992,
      "step": 157610
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.112499475479126,
      "learning_rate": 0.00013602911792748803,
      "loss": 2.1578,
      "step": 157611
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.1603304147720337,
      "learning_rate": 0.00013602569245002534,
      "loss": 1.8987,
      "step": 157612
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.3491568565368652,
      "learning_rate": 0.0001360222670030487,
      "loss": 2.0796,
      "step": 157613
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.1690077781677246,
      "learning_rate": 0.00013601884158655848,
      "loss": 2.275,
      "step": 157614
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.1336641311645508,
      "learning_rate": 0.00013601541620055562,
      "loss": 2.2131,
      "step": 157615
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.0814857482910156,
      "learning_rate": 0.00013601199084504055,
      "loss": 2.0665,
      "step": 157616
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.145187497138977,
      "learning_rate": 0.00013600856552001394,
      "loss": 2.0963,
      "step": 157617
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.1209449768066406,
      "learning_rate": 0.00013600514022547633,
      "loss": 2.0552,
      "step": 157618
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.0387742519378662,
      "learning_rate": 0.00013600171496142856,
      "loss": 2.0037,
      "step": 157619
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.1193058490753174,
      "learning_rate": 0.00013599828972787108,
      "loss": 2.1257,
      "step": 157620
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.3896304368972778,
      "learning_rate": 0.0001359948645248047,
      "loss": 2.0375,
      "step": 157621
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.2774949073791504,
      "learning_rate": 0.00013599143935223,
      "loss": 2.2226,
      "step": 157622
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.096411943435669,
      "learning_rate": 0.00013598801421014757,
      "loss": 2.0166,
      "step": 157623
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.5911099910736084,
      "learning_rate": 0.00013598458909855796,
      "loss": 2.18,
      "step": 157624
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.226631999015808,
      "learning_rate": 0.00013598116401746202,
      "loss": 2.0044,
      "step": 157625
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.2413885593414307,
      "learning_rate": 0.00013597773896686016,
      "loss": 2.2247,
      "step": 157626
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.3341351747512817,
      "learning_rate": 0.00013597431394675325,
      "loss": 2.1219,
      "step": 157627
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.3069730997085571,
      "learning_rate": 0.0001359708889571418,
      "loss": 2.1025,
      "step": 157628
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.3226300477981567,
      "learning_rate": 0.0001359674639980265,
      "loss": 2.1215,
      "step": 157629
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.2675248384475708,
      "learning_rate": 0.00013596403906940776,
      "loss": 2.064,
      "step": 157630
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.3361111879348755,
      "learning_rate": 0.00013596061417128657,
      "loss": 1.8505,
      "step": 157631
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.260111689567566,
      "learning_rate": 0.00013595718930366327,
      "loss": 2.0525,
      "step": 157632
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.3018356561660767,
      "learning_rate": 0.00013595376446653873,
      "loss": 2.0334,
      "step": 157633
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.1958993673324585,
      "learning_rate": 0.0001359503396599135,
      "loss": 1.8778,
      "step": 157634
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.2973443269729614,
      "learning_rate": 0.0001359469148837882,
      "loss": 1.9754,
      "step": 157635
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.2507085800170898,
      "learning_rate": 0.00013594349013816334,
      "loss": 1.9377,
      "step": 157636
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.1074410676956177,
      "learning_rate": 0.0001359400654230398,
      "loss": 2.015,
      "step": 157637
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.1161714792251587,
      "learning_rate": 0.00013593664073841796,
      "loss": 1.7876,
      "step": 157638
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.269374966621399,
      "learning_rate": 0.00013593321608429876,
      "loss": 1.9775,
      "step": 157639
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.1655882596969604,
      "learning_rate": 0.00013592979146068263,
      "loss": 1.8543,
      "step": 157640
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.2254923582077026,
      "learning_rate": 0.00013592636686757023,
      "loss": 1.939,
      "step": 157641
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.1461656093597412,
      "learning_rate": 0.00013592294230496216,
      "loss": 2.0815,
      "step": 157642
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.0779951810836792,
      "learning_rate": 0.0001359195177728592,
      "loss": 1.9573,
      "step": 157643
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.295996069908142,
      "learning_rate": 0.00013591609327126179,
      "loss": 1.9616,
      "step": 157644
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.3488212823867798,
      "learning_rate": 0.00013591266880017076,
      "loss": 2.025,
      "step": 157645
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.0899797677993774,
      "learning_rate": 0.0001359092443595867,
      "loss": 2.0371,
      "step": 157646
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.2552400827407837,
      "learning_rate": 0.0001359058199495102,
      "loss": 2.298,
      "step": 157647
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.2625802755355835,
      "learning_rate": 0.0001359023955699418,
      "loss": 2.0266,
      "step": 157648
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.0843315124511719,
      "learning_rate": 0.0001358989712208823,
      "loss": 2.2713,
      "step": 157649
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.2447937726974487,
      "learning_rate": 0.00013589554690233223,
      "loss": 2.1597,
      "step": 157650
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.1252442598342896,
      "learning_rate": 0.00013589212261429237,
      "loss": 2.2402,
      "step": 157651
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.1432064771652222,
      "learning_rate": 0.00013588869835676326,
      "loss": 1.8999,
      "step": 157652
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.2278460264205933,
      "learning_rate": 0.00013588527412974555,
      "loss": 2.2855,
      "step": 157653
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.2731664180755615,
      "learning_rate": 0.00013588184993323975,
      "loss": 2.1093,
      "step": 157654
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.1939276456832886,
      "learning_rate": 0.00013587842576724668,
      "loss": 2.244,
      "step": 157655
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.1331589221954346,
      "learning_rate": 0.00013587500163176683,
      "loss": 1.8388,
      "step": 157656
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.1018513441085815,
      "learning_rate": 0.00013587157752680104,
      "loss": 2.1106,
      "step": 157657
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.397322654724121,
      "learning_rate": 0.00013586815345234982,
      "loss": 1.985,
      "step": 157658
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.200408697128296,
      "learning_rate": 0.00013586472940841377,
      "loss": 1.8082,
      "step": 157659
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.165884017944336,
      "learning_rate": 0.0001358613053949935,
      "loss": 2.0926,
      "step": 157660
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.1106212139129639,
      "learning_rate": 0.0001358578814120898,
      "loss": 2.1851,
      "step": 157661
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.3443081378936768,
      "learning_rate": 0.00013585445745970312,
      "loss": 2.158,
      "step": 157662
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.115801453590393,
      "learning_rate": 0.0001358510335378343,
      "loss": 1.8226,
      "step": 157663
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.1665453910827637,
      "learning_rate": 0.00013584760964648385,
      "loss": 1.8747,
      "step": 157664
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.3475897312164307,
      "learning_rate": 0.00013584418578565236,
      "loss": 2.0173,
      "step": 157665
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.1675634384155273,
      "learning_rate": 0.0001358407619553406,
      "loss": 2.1686,
      "step": 157666
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.4946955442428589,
      "learning_rate": 0.0001358373381555492,
      "loss": 2.2119,
      "step": 157667
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.128043293952942,
      "learning_rate": 0.00013583391438627855,
      "loss": 2.135,
      "step": 157668
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.197287917137146,
      "learning_rate": 0.00013583049064752966,
      "loss": 2.0101,
      "step": 157669
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.114902138710022,
      "learning_rate": 0.00013582706693930294,
      "loss": 2.0265,
      "step": 157670
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.1447910070419312,
      "learning_rate": 0.00013582364326159897,
      "loss": 2.0246,
      "step": 157671
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.0814265012741089,
      "learning_rate": 0.00013582021961441857,
      "loss": 1.9418,
      "step": 157672
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.1456823348999023,
      "learning_rate": 0.00013581679599776233,
      "loss": 2.0614,
      "step": 157673
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.1809415817260742,
      "learning_rate": 0.0001358133724116307,
      "loss": 2.1921,
      "step": 157674
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.2213815450668335,
      "learning_rate": 0.0001358099488560246,
      "loss": 2.1327,
      "step": 157675
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.3200252056121826,
      "learning_rate": 0.00013580652533094442,
      "loss": 1.9899,
      "step": 157676
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.0882327556610107,
      "learning_rate": 0.000135803101836391,
      "loss": 1.9581,
      "step": 157677
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.1925878524780273,
      "learning_rate": 0.00013579967837236485,
      "loss": 2.0668,
      "step": 157678
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.0898702144622803,
      "learning_rate": 0.00013579625493886658,
      "loss": 2.1274,
      "step": 157679
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.2511910200119019,
      "learning_rate": 0.00013579283153589697,
      "loss": 1.8941,
      "step": 157680
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.2160497903823853,
      "learning_rate": 0.00013578940816345658,
      "loss": 2.0658,
      "step": 157681
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.1481742858886719,
      "learning_rate": 0.00013578598482154593,
      "loss": 2.0346,
      "step": 157682
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.2279846668243408,
      "learning_rate": 0.0001357825615101659,
      "loss": 1.847,
      "step": 157683
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.2317203283309937,
      "learning_rate": 0.00013577913822931698,
      "loss": 2.0745,
      "step": 157684
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.2044763565063477,
      "learning_rate": 0.0001357757149789997,
      "loss": 1.9995,
      "step": 157685
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.273694396018982,
      "learning_rate": 0.0001357722917592149,
      "loss": 1.8421,
      "step": 157686
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.1357274055480957,
      "learning_rate": 0.00013576886856996316,
      "loss": 2.1507,
      "step": 157687
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.3309745788574219,
      "learning_rate": 0.00013576544541124498,
      "loss": 2.1675,
      "step": 157688
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.3800889253616333,
      "learning_rate": 0.0001357620222830612,
      "loss": 2.1915,
      "step": 157689
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.2469598054885864,
      "learning_rate": 0.00013575859918541237,
      "loss": 2.1599,
      "step": 157690
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.4295662641525269,
      "learning_rate": 0.000135755176118299,
      "loss": 2.1952,
      "step": 157691
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.1844793558120728,
      "learning_rate": 0.00013575175308172194,
      "loss": 1.8169,
      "step": 157692
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.190180778503418,
      "learning_rate": 0.00013574833007568162,
      "loss": 2.0007,
      "step": 157693
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.1794743537902832,
      "learning_rate": 0.0001357449071001789,
      "loss": 1.9481,
      "step": 157694
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.185269832611084,
      "learning_rate": 0.00013574148415521436,
      "loss": 1.9304,
      "step": 157695
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.1653591394424438,
      "learning_rate": 0.00013573806124078853,
      "loss": 2.1499,
      "step": 157696
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.1331257820129395,
      "learning_rate": 0.00013573463835690196,
      "loss": 2.2256,
      "step": 157697
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.182194709777832,
      "learning_rate": 0.00013573121550355556,
      "loss": 2.0287,
      "step": 157698
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.172429084777832,
      "learning_rate": 0.00013572779268074972,
      "loss": 2.2162,
      "step": 157699
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.302418828010559,
      "learning_rate": 0.00013572436988848528,
      "loss": 2.0847,
      "step": 157700
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.1763263940811157,
      "learning_rate": 0.00013572094712676282,
      "loss": 1.9394,
      "step": 157701
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.1034170389175415,
      "learning_rate": 0.0001357175243955829,
      "loss": 2.1215,
      "step": 157702
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.4013184309005737,
      "learning_rate": 0.00013571410169494608,
      "loss": 2.2446,
      "step": 157703
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.1589897871017456,
      "learning_rate": 0.00013571067902485323,
      "loss": 2.133,
      "step": 157704
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.1577394008636475,
      "learning_rate": 0.00013570725638530477,
      "loss": 1.9573,
      "step": 157705
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.358819603919983,
      "learning_rate": 0.0001357038337763015,
      "loss": 2.3772,
      "step": 157706
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.53285813331604,
      "learning_rate": 0.00013570041119784405,
      "loss": 2.023,
      "step": 157707
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.7135655879974365,
      "learning_rate": 0.00013569698864993295,
      "loss": 2.0961,
      "step": 157708
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.1641753911972046,
      "learning_rate": 0.00013569356613256877,
      "loss": 2.1796,
      "step": 157709
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.3959637880325317,
      "learning_rate": 0.00013569014364575238,
      "loss": 2.0355,
      "step": 157710
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.2135957479476929,
      "learning_rate": 0.0001356867211894842,
      "loss": 2.0935,
      "step": 157711
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.2980549335479736,
      "learning_rate": 0.00013568329876376505,
      "loss": 2.1105,
      "step": 157712
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.179633617401123,
      "learning_rate": 0.0001356798763685955,
      "loss": 2.2343,
      "step": 157713
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.2387551069259644,
      "learning_rate": 0.00013567645400397612,
      "loss": 1.9686,
      "step": 157714
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.1634509563446045,
      "learning_rate": 0.00013567303166990753,
      "loss": 1.9769,
      "step": 157715
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.374922275543213,
      "learning_rate": 0.00013566960936639048,
      "loss": 1.8919,
      "step": 157716
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.1350271701812744,
      "learning_rate": 0.0001356661870934255,
      "loss": 2.0159,
      "step": 157717
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.1589045524597168,
      "learning_rate": 0.00013566276485101335,
      "loss": 2.1322,
      "step": 157718
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.2182302474975586,
      "learning_rate": 0.00013565934263915457,
      "loss": 1.8955,
      "step": 157719
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.247715950012207,
      "learning_rate": 0.00013565592045784987,
      "loss": 1.9665,
      "step": 157720
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.140119194984436,
      "learning_rate": 0.0001356524983070997,
      "loss": 2.16,
      "step": 157721
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.246517539024353,
      "learning_rate": 0.00013564907618690496,
      "loss": 2.021,
      "step": 157722
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.2272310256958008,
      "learning_rate": 0.00013564565409726606,
      "loss": 2.061,
      "step": 157723
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.1633832454681396,
      "learning_rate": 0.0001356422320381838,
      "loss": 2.1262,
      "step": 157724
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.1472275257110596,
      "learning_rate": 0.0001356388100096588,
      "loss": 2.0792,
      "step": 157725
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.2246479988098145,
      "learning_rate": 0.00013563538801169162,
      "loss": 1.9977,
      "step": 157726
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.2150369882583618,
      "learning_rate": 0.0001356319660442828,
      "loss": 2.0242,
      "step": 157727
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.3266627788543701,
      "learning_rate": 0.00013562854410743323,
      "loss": 2.0313,
      "step": 157728
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.1554367542266846,
      "learning_rate": 0.0001356251222011433,
      "loss": 2.25,
      "step": 157729
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.216081142425537,
      "learning_rate": 0.00013562170032541392,
      "loss": 2.1826,
      "step": 157730
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.3505475521087646,
      "learning_rate": 0.0001356182784802455,
      "loss": 2.037,
      "step": 157731
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.1006211042404175,
      "learning_rate": 0.00013561485666563878,
      "loss": 2.1772,
      "step": 157732
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.1345734596252441,
      "learning_rate": 0.00013561143488159425,
      "loss": 2.0372,
      "step": 157733
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.1116201877593994,
      "learning_rate": 0.00013560801312811275,
      "loss": 1.988,
      "step": 157734
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.0974950790405273,
      "learning_rate": 0.00013560459140519473,
      "loss": 2.1403,
      "step": 157735
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.2184511423110962,
      "learning_rate": 0.00013560116971284103,
      "loss": 2.1349,
      "step": 157736
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.103761911392212,
      "learning_rate": 0.00013559774805105217,
      "loss": 2.0156,
      "step": 157737
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.2907969951629639,
      "learning_rate": 0.0001355943264198288,
      "loss": 1.9802,
      "step": 157738
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.0824463367462158,
      "learning_rate": 0.00013559090481917143,
      "loss": 2.196,
      "step": 157739
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.16464364528656,
      "learning_rate": 0.00013558748324908092,
      "loss": 2.1208,
      "step": 157740
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.2117505073547363,
      "learning_rate": 0.0001355840617095577,
      "loss": 2.0815,
      "step": 157741
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.1763030290603638,
      "learning_rate": 0.0001355806402006026,
      "loss": 1.8556,
      "step": 157742
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.1475640535354614,
      "learning_rate": 0.00013557721872221607,
      "loss": 2.0617,
      "step": 157743
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.2232388257980347,
      "learning_rate": 0.00013557379727439903,
      "loss": 2.1193,
      "step": 157744
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.1205233335494995,
      "learning_rate": 0.00013557037585715175,
      "loss": 2.2052,
      "step": 157745
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.1611965894699097,
      "learning_rate": 0.00013556695447047512,
      "loss": 2.0436,
      "step": 157746
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.270889163017273,
      "learning_rate": 0.00013556353311436957,
      "loss": 2.1226,
      "step": 157747
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.07374906539917,
      "learning_rate": 0.00013556011178883604,
      "loss": 2.2884,
      "step": 157748
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.0902847051620483,
      "learning_rate": 0.00013555669049387483,
      "loss": 2.0729,
      "step": 157749
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.2700283527374268,
      "learning_rate": 0.00013555326922948697,
      "loss": 2.0449,
      "step": 157750
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.1939632892608643,
      "learning_rate": 0.00013554984799567262,
      "loss": 2.1988,
      "step": 157751
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.1204249858856201,
      "learning_rate": 0.00013554642679243278,
      "loss": 1.9151,
      "step": 157752
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.1674168109893799,
      "learning_rate": 0.00013554300561976785,
      "loss": 1.9896,
      "step": 157753
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.0644471645355225,
      "learning_rate": 0.0001355395844776787,
      "loss": 1.784,
      "step": 157754
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.1798430681228638,
      "learning_rate": 0.00013553616336616574,
      "loss": 2.0753,
      "step": 157755
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.1106239557266235,
      "learning_rate": 0.0001355327422852298,
      "loss": 1.8282,
      "step": 157756
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.390807032585144,
      "learning_rate": 0.00013552932123487145,
      "loss": 2.2074,
      "step": 157757
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.131651759147644,
      "learning_rate": 0.0001355259002150913,
      "loss": 1.9536,
      "step": 157758
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.2643377780914307,
      "learning_rate": 0.00013552247922588988,
      "loss": 2.0229,
      "step": 157759
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.1690924167633057,
      "learning_rate": 0.00013551905826726803,
      "loss": 2.1585,
      "step": 157760
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.1478972434997559,
      "learning_rate": 0.0001355156373392262,
      "loss": 2.04,
      "step": 157761
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.291926622390747,
      "learning_rate": 0.0001355122164417652,
      "loss": 2.1041,
      "step": 157762
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.1697602272033691,
      "learning_rate": 0.0001355087955748856,
      "loss": 2.0807,
      "step": 157763
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.2148298025131226,
      "learning_rate": 0.00013550537473858794,
      "loss": 1.9444,
      "step": 157764
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.1119271516799927,
      "learning_rate": 0.00013550195393287302,
      "loss": 1.9697,
      "step": 157765
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.1733453273773193,
      "learning_rate": 0.0001354985331577414,
      "loss": 1.9679,
      "step": 157766
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.2315654754638672,
      "learning_rate": 0.00013549511241319357,
      "loss": 2.0302,
      "step": 157767
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.1844125986099243,
      "learning_rate": 0.00013549169169923045,
      "loss": 2.2246,
      "step": 157768
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.22698974609375,
      "learning_rate": 0.0001354882710158525,
      "loss": 2.0922,
      "step": 157769
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.204043984413147,
      "learning_rate": 0.00013548485036306027,
      "loss": 1.931,
      "step": 157770
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.3172441720962524,
      "learning_rate": 0.00013548142974085468,
      "loss": 2.2167,
      "step": 157771
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.1250567436218262,
      "learning_rate": 0.00013547800914923616,
      "loss": 2.0023,
      "step": 157772
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.1532503366470337,
      "learning_rate": 0.00013547458858820526,
      "loss": 2.1071,
      "step": 157773
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.2724483013153076,
      "learning_rate": 0.00013547116805776287,
      "loss": 2.0599,
      "step": 157774
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.1935839653015137,
      "learning_rate": 0.0001354677475579095,
      "loss": 1.9271,
      "step": 157775
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.1995949745178223,
      "learning_rate": 0.00013546432708864563,
      "loss": 2.1993,
      "step": 157776
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.0898016691207886,
      "learning_rate": 0.0001354609066499722,
      "loss": 2.0434,
      "step": 157777
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.0256637334823608,
      "learning_rate": 0.00013545748624188957,
      "loss": 2.0079,
      "step": 157778
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.1910698413848877,
      "learning_rate": 0.0001354540658643986,
      "loss": 1.8663,
      "step": 157779
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.189192295074463,
      "learning_rate": 0.00013545064551749984,
      "loss": 2.054,
      "step": 157780
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.2147135734558105,
      "learning_rate": 0.0001354472252011939,
      "loss": 2.0716,
      "step": 157781
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.1971408128738403,
      "learning_rate": 0.0001354438049154813,
      "loss": 1.9782,
      "step": 157782
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.2321949005126953,
      "learning_rate": 0.00013544038466036296,
      "loss": 2.3003,
      "step": 157783
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.1553452014923096,
      "learning_rate": 0.00013543696443583922,
      "loss": 2.0083,
      "step": 157784
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.329958200454712,
      "learning_rate": 0.000135433544241911,
      "loss": 2.0368,
      "step": 157785
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.180688738822937,
      "learning_rate": 0.00013543012407857873,
      "loss": 2.2068,
      "step": 157786
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.649653673171997,
      "learning_rate": 0.00013542670394584315,
      "loss": 1.8388,
      "step": 157787
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.1630884408950806,
      "learning_rate": 0.00013542328384370472,
      "loss": 2.1448,
      "step": 157788
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.2051411867141724,
      "learning_rate": 0.00013541986377216432,
      "loss": 2.2206,
      "step": 157789
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.2381383180618286,
      "learning_rate": 0.00013541644373122236,
      "loss": 2.1226,
      "step": 157790
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.2263610363006592,
      "learning_rate": 0.0001354130237208797,
      "loss": 1.9598,
      "step": 157791
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.1710199117660522,
      "learning_rate": 0.0001354096037411369,
      "loss": 2.0486,
      "step": 157792
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.2474181652069092,
      "learning_rate": 0.00013540618379199453,
      "loss": 2.183,
      "step": 157793
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.2294217348098755,
      "learning_rate": 0.00013540276387345317,
      "loss": 2.0556,
      "step": 157794
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.2093243598937988,
      "learning_rate": 0.00013539934398551363,
      "loss": 2.0268,
      "step": 157795
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.2592530250549316,
      "learning_rate": 0.00013539592412817634,
      "loss": 1.9591,
      "step": 157796
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.1499427556991577,
      "learning_rate": 0.00013539250430144218,
      "loss": 2.3592,
      "step": 157797
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.1866027116775513,
      "learning_rate": 0.00013538908450531168,
      "loss": 2.1822,
      "step": 157798
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.1872498989105225,
      "learning_rate": 0.00013538566473978545,
      "loss": 2.0286,
      "step": 157799
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.329341173171997,
      "learning_rate": 0.000135382245004864,
      "loss": 2.1726,
      "step": 157800
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.124753475189209,
      "learning_rate": 0.00013537882530054822,
      "loss": 2.1269,
      "step": 157801
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.1543303728103638,
      "learning_rate": 0.0001353754056268385,
      "loss": 2.0703,
      "step": 157802
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.286530613899231,
      "learning_rate": 0.00013537198598373573,
      "loss": 1.9084,
      "step": 157803
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.3831692934036255,
      "learning_rate": 0.00013536856637124043,
      "loss": 2.1486,
      "step": 157804
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.1526281833648682,
      "learning_rate": 0.00013536514678935316,
      "loss": 1.9308,
      "step": 157805
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.1803981065750122,
      "learning_rate": 0.00013536172723807458,
      "loss": 1.8539,
      "step": 157806
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.1440082788467407,
      "learning_rate": 0.00013535830771740546,
      "loss": 2.2465,
      "step": 157807
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.1195729970932007,
      "learning_rate": 0.00013535488822734616,
      "loss": 2.0276,
      "step": 157808
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.4069018363952637,
      "learning_rate": 0.00013535146876789768,
      "loss": 1.9297,
      "step": 157809
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.207342267036438,
      "learning_rate": 0.00013534804933906032,
      "loss": 1.9672,
      "step": 157810
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.182096242904663,
      "learning_rate": 0.0001353446299408351,
      "loss": 2.2776,
      "step": 157811
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.3246443271636963,
      "learning_rate": 0.0001353412105732222,
      "loss": 2.0259,
      "step": 157812
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.1876667737960815,
      "learning_rate": 0.0001353377912362226,
      "loss": 1.8555,
      "step": 157813
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.1971933841705322,
      "learning_rate": 0.00013533437192983665,
      "loss": 2.0242,
      "step": 157814
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.210149884223938,
      "learning_rate": 0.0001353309526540653,
      "loss": 2.092,
      "step": 157815
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.225455641746521,
      "learning_rate": 0.00013532753340890892,
      "loss": 1.9849,
      "step": 157816
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.1894299983978271,
      "learning_rate": 0.0001353241141943685,
      "loss": 2.0399,
      "step": 157817
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.1623609066009521,
      "learning_rate": 0.00013532069501044421,
      "loss": 2.262,
      "step": 157818
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.1433398723602295,
      "learning_rate": 0.000135317275857137,
      "loss": 1.9207,
      "step": 157819
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.1543519496917725,
      "learning_rate": 0.00013531385673444733,
      "loss": 1.8682,
      "step": 157820
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.1110364198684692,
      "learning_rate": 0.000135310437642376,
      "loss": 2.0666,
      "step": 157821
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.235561728477478,
      "learning_rate": 0.00013530701858092349,
      "loss": 1.9532,
      "step": 157822
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.3855836391448975,
      "learning_rate": 0.0001353035995500907,
      "loss": 2.0247,
      "step": 157823
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.1488733291625977,
      "learning_rate": 0.0001353001805498779,
      "loss": 2.0058,
      "step": 157824
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.14961576461792,
      "learning_rate": 0.00013529676158028598,
      "loss": 2.068,
      "step": 157825
    },
    {
      "epoch": 2.05,
      "grad_norm": 1.3462963104248047,
      "learning_rate": 0.00013529334264131542,
      "loss": 2.0875,
      "step": 157826
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.2366533279418945,
      "learning_rate": 0.00013528992373296701,
      "loss": 1.9874,
      "step": 157827
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.3611656427383423,
      "learning_rate": 0.00013528650485524122,
      "loss": 2.2046,
      "step": 157828
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.2142939567565918,
      "learning_rate": 0.00013528308600813905,
      "loss": 2.2707,
      "step": 157829
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.1123031377792358,
      "learning_rate": 0.0001352796671916606,
      "loss": 2.1092,
      "step": 157830
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.1783267259597778,
      "learning_rate": 0.00013527624840580684,
      "loss": 2.2459,
      "step": 157831
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.3240472078323364,
      "learning_rate": 0.00013527282965057827,
      "loss": 2.2963,
      "step": 157832
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.1938916444778442,
      "learning_rate": 0.00013526941092597568,
      "loss": 1.9397,
      "step": 157833
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.1928174495697021,
      "learning_rate": 0.00013526599223199955,
      "loss": 1.9454,
      "step": 157834
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.171263337135315,
      "learning_rate": 0.00013526257356865078,
      "loss": 2.0387,
      "step": 157835
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.317020297050476,
      "learning_rate": 0.00013525915493592955,
      "loss": 1.8906,
      "step": 157836
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.0860991477966309,
      "learning_rate": 0.00013525573633383687,
      "loss": 1.9298,
      "step": 157837
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.181779146194458,
      "learning_rate": 0.00013525231776237318,
      "loss": 1.9461,
      "step": 157838
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.250972032546997,
      "learning_rate": 0.0001352488992215393,
      "loss": 1.8945,
      "step": 157839
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.1820287704467773,
      "learning_rate": 0.00013524548071133566,
      "loss": 2.0698,
      "step": 157840
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.1680564880371094,
      "learning_rate": 0.00013524206223176308,
      "loss": 1.8563,
      "step": 157841
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.2188081741333008,
      "learning_rate": 0.00013523864378282212,
      "loss": 2.0488,
      "step": 157842
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.508673906326294,
      "learning_rate": 0.0001352352253645134,
      "loss": 2.1603,
      "step": 157843
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.1637436151504517,
      "learning_rate": 0.00013523180697683744,
      "loss": 1.9085,
      "step": 157844
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.26406991481781,
      "learning_rate": 0.00013522838861979516,
      "loss": 2.0743,
      "step": 157845
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.2660812139511108,
      "learning_rate": 0.00013522497029338685,
      "loss": 2.0121,
      "step": 157846
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.1798666715621948,
      "learning_rate": 0.00013522155199761348,
      "loss": 2.1628,
      "step": 157847
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.1241075992584229,
      "learning_rate": 0.00013521813373247552,
      "loss": 1.9037,
      "step": 157848
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.2157031297683716,
      "learning_rate": 0.00013521471549797365,
      "loss": 2.1907,
      "step": 157849
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.1848386526107788,
      "learning_rate": 0.00013521129729410832,
      "loss": 2.2284,
      "step": 157850
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.270724892616272,
      "learning_rate": 0.00013520787912088046,
      "loss": 2.0655,
      "step": 157851
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.2737956047058105,
      "learning_rate": 0.00013520446097829044,
      "loss": 2.1776,
      "step": 157852
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.0596636533737183,
      "learning_rate": 0.00013520104286633914,
      "loss": 1.9714,
      "step": 157853
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.2312312126159668,
      "learning_rate": 0.0001351976247850271,
      "loss": 2.0823,
      "step": 157854
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.1850557327270508,
      "learning_rate": 0.00013519420673435478,
      "loss": 1.901,
      "step": 157855
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.1283869743347168,
      "learning_rate": 0.0001351907887143231,
      "loss": 2.0108,
      "step": 157856
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.1055258512496948,
      "learning_rate": 0.00013518737072493254,
      "loss": 1.8912,
      "step": 157857
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.102646827697754,
      "learning_rate": 0.0001351839527661837,
      "loss": 2.2192,
      "step": 157858
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.1502416133880615,
      "learning_rate": 0.00013518053483807735,
      "loss": 2.1765,
      "step": 157859
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.1983798742294312,
      "learning_rate": 0.00013517711694061406,
      "loss": 1.8201,
      "step": 157860
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.271917700767517,
      "learning_rate": 0.00013517369907379434,
      "loss": 1.9636,
      "step": 157861
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.166746973991394,
      "learning_rate": 0.00013517028123761904,
      "loss": 2.0374,
      "step": 157862
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.2998534440994263,
      "learning_rate": 0.0001351668634320886,
      "loss": 1.983,
      "step": 157863
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.2321288585662842,
      "learning_rate": 0.00013516344565720384,
      "loss": 1.9332,
      "step": 157864
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.2010784149169922,
      "learning_rate": 0.00013516002791296534,
      "loss": 2.1391,
      "step": 157865
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.3473960161209106,
      "learning_rate": 0.0001351566101993737,
      "loss": 2.0163,
      "step": 157866
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.294031023979187,
      "learning_rate": 0.00013515319251642942,
      "loss": 2.0168,
      "step": 157867
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.1597267389297485,
      "learning_rate": 0.0001351497748641334,
      "loss": 2.0448,
      "step": 157868
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.117926836013794,
      "learning_rate": 0.00013514635724248602,
      "loss": 1.7843,
      "step": 157869
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.5206336975097656,
      "learning_rate": 0.0001351429396514882,
      "loss": 2.1142,
      "step": 157870
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.207126259803772,
      "learning_rate": 0.00013513952209114036,
      "loss": 1.9423,
      "step": 157871
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.8062695264816284,
      "learning_rate": 0.00013513610456144325,
      "loss": 2.0696,
      "step": 157872
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.1921377182006836,
      "learning_rate": 0.0001351326870623973,
      "loss": 2.0034,
      "step": 157873
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.237237572669983,
      "learning_rate": 0.00013512926959400345,
      "loss": 2.0445,
      "step": 157874
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.35868239402771,
      "learning_rate": 0.00013512585215626202,
      "loss": 2.0721,
      "step": 157875
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.1811250448226929,
      "learning_rate": 0.00013512243474917394,
      "loss": 2.1289,
      "step": 157876
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.2202242612838745,
      "learning_rate": 0.0001351190173727396,
      "loss": 2.1993,
      "step": 157877
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.130260944366455,
      "learning_rate": 0.00013511560002695997,
      "loss": 1.9991,
      "step": 157878
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.2790634632110596,
      "learning_rate": 0.00013511218271183525,
      "loss": 2.2397,
      "step": 157879
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.156959056854248,
      "learning_rate": 0.00013510876542736637,
      "loss": 2.1159,
      "step": 157880
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.2235760688781738,
      "learning_rate": 0.0001351053481735538,
      "loss": 1.9748,
      "step": 157881
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.2740247249603271,
      "learning_rate": 0.00013510193095039833,
      "loss": 2.1143,
      "step": 157882
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.1370813846588135,
      "learning_rate": 0.00013509851375790048,
      "loss": 1.9644,
      "step": 157883
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.1204495429992676,
      "learning_rate": 0.00013509509659606107,
      "loss": 1.6525,
      "step": 157884
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.1543749570846558,
      "learning_rate": 0.00013509167946488044,
      "loss": 2.0634,
      "step": 157885
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.1982711553573608,
      "learning_rate": 0.00013508826236435946,
      "loss": 1.9684,
      "step": 157886
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.3571430444717407,
      "learning_rate": 0.0001350848452944986,
      "loss": 1.9311,
      "step": 157887
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.1877049207687378,
      "learning_rate": 0.00013508142825529866,
      "loss": 1.9915,
      "step": 157888
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.1304383277893066,
      "learning_rate": 0.00013507801124676007,
      "loss": 2.021,
      "step": 157889
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.3796753883361816,
      "learning_rate": 0.00013507459426888386,
      "loss": 2.0129,
      "step": 157890
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.1521053314208984,
      "learning_rate": 0.00013507117732167016,
      "loss": 1.986,
      "step": 157891
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.162298560142517,
      "learning_rate": 0.00013506776040511992,
      "loss": 2.1434,
      "step": 157892
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.0993657112121582,
      "learning_rate": 0.00013506434351923363,
      "loss": 1.835,
      "step": 157893
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.1505030393600464,
      "learning_rate": 0.0001350609266640121,
      "loss": 1.9323,
      "step": 157894
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.187178373336792,
      "learning_rate": 0.0001350575098394557,
      "loss": 1.9838,
      "step": 157895
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.3681474924087524,
      "learning_rate": 0.00013505409304556549,
      "loss": 2.0825,
      "step": 157896
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.0943727493286133,
      "learning_rate": 0.0001350506762823416,
      "loss": 2.2119,
      "step": 157897
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.0979325771331787,
      "learning_rate": 0.00013504725954978502,
      "loss": 1.974,
      "step": 157898
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.1975033283233643,
      "learning_rate": 0.00013504384284789613,
      "loss": 1.9842,
      "step": 157899
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.1505422592163086,
      "learning_rate": 0.00013504042617667585,
      "loss": 1.9371,
      "step": 157900
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.2771031856536865,
      "learning_rate": 0.00013503700953612454,
      "loss": 1.9767,
      "step": 157901
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.2599458694458008,
      "learning_rate": 0.00013503359292624318,
      "loss": 1.9632,
      "step": 157902
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.460039734840393,
      "learning_rate": 0.000135030176347032,
      "loss": 2.1913,
      "step": 157903
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.4754787683486938,
      "learning_rate": 0.0001350267597984919,
      "loss": 1.9248,
      "step": 157904
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.248738169670105,
      "learning_rate": 0.0001350233432806233,
      "loss": 2.0467,
      "step": 157905
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.2925372123718262,
      "learning_rate": 0.00013501992679342713,
      "loss": 1.9219,
      "step": 157906
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.4169585704803467,
      "learning_rate": 0.0001350165103369038,
      "loss": 2.1121,
      "step": 157907
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.3120976686477661,
      "learning_rate": 0.00013501309391105416,
      "loss": 2.0923,
      "step": 157908
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.1891142129898071,
      "learning_rate": 0.0001350096775158785,
      "loss": 1.9038,
      "step": 157909
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.1236740350723267,
      "learning_rate": 0.0001350062611513778,
      "loss": 2.1115,
      "step": 157910
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.2657281160354614,
      "learning_rate": 0.0001350028448175524,
      "loss": 2.093,
      "step": 157911
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.8665845394134521,
      "learning_rate": 0.00013499942851440323,
      "loss": 2.1565,
      "step": 157912
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.1856838464736938,
      "learning_rate": 0.00013499601224193066,
      "loss": 1.9753,
      "step": 157913
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.389969825744629,
      "learning_rate": 0.00013499259600013562,
      "loss": 2.0574,
      "step": 157914
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.166795015335083,
      "learning_rate": 0.0001349891797890184,
      "loss": 1.9981,
      "step": 157915
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.1897411346435547,
      "learning_rate": 0.00013498576360857991,
      "loss": 2.0255,
      "step": 157916
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.2141788005828857,
      "learning_rate": 0.00013498234745882058,
      "loss": 2.05,
      "step": 157917
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.2161517143249512,
      "learning_rate": 0.00013497893133974123,
      "loss": 1.9725,
      "step": 157918
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.125217080116272,
      "learning_rate": 0.00013497551525134234,
      "loss": 2.1952,
      "step": 157919
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.4000673294067383,
      "learning_rate": 0.0001349720991936248,
      "loss": 1.7716,
      "step": 157920
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.3083922863006592,
      "learning_rate": 0.00013496868316658882,
      "loss": 2.1779,
      "step": 157921
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.1766629219055176,
      "learning_rate": 0.00013496526717023544,
      "loss": 2.1028,
      "step": 157922
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.177018165588379,
      "learning_rate": 0.00013496185120456501,
      "loss": 1.9848,
      "step": 157923
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.1729049682617188,
      "learning_rate": 0.00013495843526957837,
      "loss": 1.9406,
      "step": 157924
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.6489323377609253,
      "learning_rate": 0.000134955019365276,
      "loss": 2.0142,
      "step": 157925
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.1887428760528564,
      "learning_rate": 0.0001349516034916587,
      "loss": 2.1844,
      "step": 157926
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.24497652053833,
      "learning_rate": 0.000134948187648727,
      "loss": 2.2544,
      "step": 157927
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.5348387956619263,
      "learning_rate": 0.00013494477183648157,
      "loss": 2.1118,
      "step": 157928
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.3537814617156982,
      "learning_rate": 0.0001349413560549229,
      "loss": 2.1194,
      "step": 157929
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.171160340309143,
      "learning_rate": 0.0001349379403040519,
      "loss": 2.0681,
      "step": 157930
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.1601859331130981,
      "learning_rate": 0.00013493452458386888,
      "loss": 1.9815,
      "step": 157931
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.1922030448913574,
      "learning_rate": 0.00013493110889437477,
      "loss": 2.1441,
      "step": 157932
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.2246922254562378,
      "learning_rate": 0.00013492769323557012,
      "loss": 2.0278,
      "step": 157933
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.2303038835525513,
      "learning_rate": 0.00013492427760745547,
      "loss": 2.1463,
      "step": 157934
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.2093051671981812,
      "learning_rate": 0.00013492086201003143,
      "loss": 2.0466,
      "step": 157935
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.1861495971679688,
      "learning_rate": 0.00013491744644329886,
      "loss": 2.1699,
      "step": 157936
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.1327961683273315,
      "learning_rate": 0.0001349140309072581,
      "loss": 1.9213,
      "step": 157937
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.4707151651382446,
      "learning_rate": 0.00013491061540191006,
      "loss": 2.0112,
      "step": 157938
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.2844833135604858,
      "learning_rate": 0.00013490719992725525,
      "loss": 1.9227,
      "step": 157939
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.2039211988449097,
      "learning_rate": 0.0001349037844832942,
      "loss": 2.0518,
      "step": 157940
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.1963495016098022,
      "learning_rate": 0.0001349003690700278,
      "loss": 2.0624,
      "step": 157941
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.9181393384933472,
      "learning_rate": 0.0001348969536874565,
      "loss": 1.9029,
      "step": 157942
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.3824933767318726,
      "learning_rate": 0.00013489353833558086,
      "loss": 2.1134,
      "step": 157943
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.1803205013275146,
      "learning_rate": 0.00013489012301440173,
      "loss": 1.9315,
      "step": 157944
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.1657620668411255,
      "learning_rate": 0.00013488670772391967,
      "loss": 1.9223,
      "step": 157945
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.2994974851608276,
      "learning_rate": 0.00013488329246413516,
      "loss": 1.9386,
      "step": 157946
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.5783991813659668,
      "learning_rate": 0.00013487987723504908,
      "loss": 1.918,
      "step": 157947
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.1375941038131714,
      "learning_rate": 0.00013487646203666197,
      "loss": 1.8565,
      "step": 157948
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.2331738471984863,
      "learning_rate": 0.00013487304686897429,
      "loss": 1.9737,
      "step": 157949
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.2930092811584473,
      "learning_rate": 0.00013486963173198693,
      "loss": 2.1691,
      "step": 157950
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.2816544771194458,
      "learning_rate": 0.00013486621662570048,
      "loss": 2.0352,
      "step": 157951
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.227189064025879,
      "learning_rate": 0.00013486280155011534,
      "loss": 2.1256,
      "step": 157952
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.232671856880188,
      "learning_rate": 0.0001348593865052325,
      "loss": 1.9587,
      "step": 157953
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.2966535091400146,
      "learning_rate": 0.00013485597149105226,
      "loss": 1.9887,
      "step": 157954
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.1566678285598755,
      "learning_rate": 0.00013485255650757557,
      "loss": 2.3313,
      "step": 157955
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.2115466594696045,
      "learning_rate": 0.00013484914155480286,
      "loss": 1.8253,
      "step": 157956
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.0697625875473022,
      "learning_rate": 0.0001348457266327348,
      "loss": 1.8911,
      "step": 157957
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.2373616695404053,
      "learning_rate": 0.00013484231174137198,
      "loss": 1.9209,
      "step": 157958
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.1490602493286133,
      "learning_rate": 0.00013483889688071518,
      "loss": 1.7309,
      "step": 157959
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.1603578329086304,
      "learning_rate": 0.00013483548205076484,
      "loss": 2.0712,
      "step": 157960
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.182876467704773,
      "learning_rate": 0.0001348320672515218,
      "loss": 1.8229,
      "step": 157961
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.1716663837432861,
      "learning_rate": 0.00013482865248298648,
      "loss": 2.081,
      "step": 157962
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.060303807258606,
      "learning_rate": 0.00013482523774515986,
      "loss": 2.0009,
      "step": 157963
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.2221384048461914,
      "learning_rate": 0.00013482182303804213,
      "loss": 1.8691,
      "step": 157964
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.2386884689331055,
      "learning_rate": 0.00013481840836163426,
      "loss": 2.0595,
      "step": 157965
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.3062752485275269,
      "learning_rate": 0.00013481499371593663,
      "loss": 1.9547,
      "step": 157966
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.1467351913452148,
      "learning_rate": 0.00013481157910095012,
      "loss": 1.9958,
      "step": 157967
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.095990777015686,
      "learning_rate": 0.0001348081645166752,
      "loss": 2.1612,
      "step": 157968
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.2663705348968506,
      "learning_rate": 0.00013480474996311272,
      "loss": 1.9189,
      "step": 157969
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.2380797863006592,
      "learning_rate": 0.00013480133544026298,
      "loss": 1.9749,
      "step": 157970
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.225754976272583,
      "learning_rate": 0.00013479792094812687,
      "loss": 2.0608,
      "step": 157971
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.420652985572815,
      "learning_rate": 0.00013479450648670483,
      "loss": 2.0978,
      "step": 157972
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.3685839176177979,
      "learning_rate": 0.00013479109205599772,
      "loss": 2.2437,
      "step": 157973
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.1197103261947632,
      "learning_rate": 0.000134787677656006,
      "loss": 1.9233,
      "step": 157974
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.498243808746338,
      "learning_rate": 0.00013478426328673054,
      "loss": 1.9944,
      "step": 157975
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.1577720642089844,
      "learning_rate": 0.0001347808489481716,
      "loss": 2.0932,
      "step": 157976
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.2309244871139526,
      "learning_rate": 0.00013477743464033014,
      "loss": 2.2699,
      "step": 157977
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.1578624248504639,
      "learning_rate": 0.00013477402036320656,
      "loss": 2.0661,
      "step": 157978
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.3214669227600098,
      "learning_rate": 0.00013477060611680167,
      "loss": 1.8895,
      "step": 157979
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.2117226123809814,
      "learning_rate": 0.000134767191901116,
      "loss": 2.0824,
      "step": 157980
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.45634126663208,
      "learning_rate": 0.00013476377771615044,
      "loss": 2.0151,
      "step": 157981
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.2269607782363892,
      "learning_rate": 0.00013476036356190515,
      "loss": 2.0084,
      "step": 157982
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.1232025623321533,
      "learning_rate": 0.0001347569494383812,
      "loss": 1.8958,
      "step": 157983
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.3942270278930664,
      "learning_rate": 0.0001347535353455789,
      "loss": 2.1116,
      "step": 157984
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.3505793809890747,
      "learning_rate": 0.00013475012128349914,
      "loss": 2.0326,
      "step": 157985
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.3120479583740234,
      "learning_rate": 0.00013474670725214236,
      "loss": 2.2008,
      "step": 157986
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.1888504028320312,
      "learning_rate": 0.0001347432932515095,
      "loss": 1.8968,
      "step": 157987
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.136184573173523,
      "learning_rate": 0.00013473987928160075,
      "loss": 1.7271,
      "step": 157988
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.1134337186813354,
      "learning_rate": 0.0001347364653424171,
      "loss": 2.0649,
      "step": 157989
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.2085444927215576,
      "learning_rate": 0.00013473305143395895,
      "loss": 1.9663,
      "step": 157990
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.1414411067962646,
      "learning_rate": 0.00013472963755622715,
      "loss": 1.9528,
      "step": 157991
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.2733104228973389,
      "learning_rate": 0.00013472622370922212,
      "loss": 2.0607,
      "step": 157992
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.1760492324829102,
      "learning_rate": 0.00013472280989294482,
      "loss": 2.1427,
      "step": 157993
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.2047646045684814,
      "learning_rate": 0.0001347193961073955,
      "loss": 1.9678,
      "step": 157994
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.0788626670837402,
      "learning_rate": 0.00013471598235257503,
      "loss": 2.059,
      "step": 157995
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.3117129802703857,
      "learning_rate": 0.00013471256862848387,
      "loss": 2.3057,
      "step": 157996
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.366686224937439,
      "learning_rate": 0.00013470915493512286,
      "loss": 1.9173,
      "step": 157997
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.1557296514511108,
      "learning_rate": 0.00013470574127249244,
      "loss": 2.0691,
      "step": 157998
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.3377265930175781,
      "learning_rate": 0.00013470232764059358,
      "loss": 1.998,
      "step": 157999
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.2378793954849243,
      "learning_rate": 0.00013469891403942644,
      "loss": 2.1429,
      "step": 158000
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.1983898878097534,
      "learning_rate": 0.00013469550046899198,
      "loss": 2.0009,
      "step": 158001
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.2392455339431763,
      "learning_rate": 0.0001346920869292907,
      "loss": 1.863,
      "step": 158002
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.2885257005691528,
      "learning_rate": 0.00013468867342032333,
      "loss": 2.0236,
      "step": 158003
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.105871319770813,
      "learning_rate": 0.00013468525994209037,
      "loss": 1.8692,
      "step": 158004
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.1890861988067627,
      "learning_rate": 0.00013468184649459276,
      "loss": 1.9125,
      "step": 158005
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.1851849555969238,
      "learning_rate": 0.0001346784330778307,
      "loss": 1.9949,
      "step": 158006
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.3486441373825073,
      "learning_rate": 0.00013467501969180516,
      "loss": 1.9028,
      "step": 158007
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.1785759925842285,
      "learning_rate": 0.0001346716063365165,
      "loss": 2.2111,
      "step": 158008
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.160903811454773,
      "learning_rate": 0.00013466819301196568,
      "loss": 1.9952,
      "step": 158009
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.4615757465362549,
      "learning_rate": 0.00013466477971815302,
      "loss": 2.1621,
      "step": 158010
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.2646584510803223,
      "learning_rate": 0.0001346613664550794,
      "loss": 2.1863,
      "step": 158011
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.076697826385498,
      "learning_rate": 0.00013465795322274536,
      "loss": 1.896,
      "step": 158012
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.1770001649856567,
      "learning_rate": 0.00013465454002115152,
      "loss": 1.939,
      "step": 158013
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.2346049547195435,
      "learning_rate": 0.00013465112685029845,
      "loss": 1.958,
      "step": 158014
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.274141788482666,
      "learning_rate": 0.00013464771371018695,
      "loss": 1.9925,
      "step": 158015
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.2567408084869385,
      "learning_rate": 0.00013464430060081744,
      "loss": 1.9698,
      "step": 158016
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.354040503501892,
      "learning_rate": 0.00013464088752219078,
      "loss": 2.2036,
      "step": 158017
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.18661630153656,
      "learning_rate": 0.0001346374744743075,
      "loss": 1.9068,
      "step": 158018
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.2020456790924072,
      "learning_rate": 0.00013463406145716825,
      "loss": 1.9456,
      "step": 158019
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.4037750959396362,
      "learning_rate": 0.00013463064847077353,
      "loss": 2.0641,
      "step": 158020
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.1749602556228638,
      "learning_rate": 0.00013462723551512423,
      "loss": 1.9116,
      "step": 158021
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.1977568864822388,
      "learning_rate": 0.0001346238225902207,
      "loss": 2.0272,
      "step": 158022
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.8125851154327393,
      "learning_rate": 0.00013462040969606387,
      "loss": 1.9254,
      "step": 158023
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.3408658504486084,
      "learning_rate": 0.00013461699683265418,
      "loss": 2.0268,
      "step": 158024
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.1432592868804932,
      "learning_rate": 0.00013461358399999223,
      "loss": 2.1744,
      "step": 158025
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.0970571041107178,
      "learning_rate": 0.00013461017119807887,
      "loss": 2.0979,
      "step": 158026
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.30012047290802,
      "learning_rate": 0.00013460675842691457,
      "loss": 2.2586,
      "step": 158027
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.269545555114746,
      "learning_rate": 0.0001346033456864999,
      "loss": 1.8662,
      "step": 158028
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.1607050895690918,
      "learning_rate": 0.00013459993297683566,
      "loss": 1.9633,
      "step": 158029
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.1117510795593262,
      "learning_rate": 0.00013459652029792246,
      "loss": 2.0033,
      "step": 158030
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.1969420909881592,
      "learning_rate": 0.00013459310764976079,
      "loss": 2.1211,
      "step": 158031
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.0996084213256836,
      "learning_rate": 0.00013458969503235145,
      "loss": 1.6955,
      "step": 158032
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.1517760753631592,
      "learning_rate": 0.00013458628244569502,
      "loss": 1.9842,
      "step": 158033
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.340273380279541,
      "learning_rate": 0.00013458286988979204,
      "loss": 2.151,
      "step": 158034
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.1020135879516602,
      "learning_rate": 0.0001345794573646433,
      "loss": 2.2393,
      "step": 158035
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.216365933418274,
      "learning_rate": 0.00013457604487024938,
      "loss": 2.2967,
      "step": 158036
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.1252305507659912,
      "learning_rate": 0.00013457263240661081,
      "loss": 1.9103,
      "step": 158037
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.1580537557601929,
      "learning_rate": 0.00013456921997372838,
      "loss": 2.0991,
      "step": 158038
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.1937400102615356,
      "learning_rate": 0.0001345658075716026,
      "loss": 2.1651,
      "step": 158039
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.2749321460723877,
      "learning_rate": 0.0001345623952002342,
      "loss": 2.2272,
      "step": 158040
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.3277873992919922,
      "learning_rate": 0.0001345589828596238,
      "loss": 2.0093,
      "step": 158041
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.5407168865203857,
      "learning_rate": 0.00013455557054977202,
      "loss": 1.8289,
      "step": 158042
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.39358389377594,
      "learning_rate": 0.00013455215827067935,
      "loss": 2.0214,
      "step": 158043
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.1032545566558838,
      "learning_rate": 0.00013454874602234667,
      "loss": 2.2262,
      "step": 158044
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.368779182434082,
      "learning_rate": 0.00013454533380477442,
      "loss": 2.4106,
      "step": 158045
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.3255329132080078,
      "learning_rate": 0.0001345419216179634,
      "loss": 2.1713,
      "step": 158046
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.2185505628585815,
      "learning_rate": 0.00013453850946191408,
      "loss": 2.0734,
      "step": 158047
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.220502257347107,
      "learning_rate": 0.00013453509733662736,
      "loss": 1.8696,
      "step": 158048
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.098070502281189,
      "learning_rate": 0.00013453168524210345,
      "loss": 1.8656,
      "step": 158049
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.0976006984710693,
      "learning_rate": 0.0001345282731783434,
      "loss": 2.1579,
      "step": 158050
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.2811880111694336,
      "learning_rate": 0.0001345248611453475,
      "loss": 2.0738,
      "step": 158051
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.2525545358657837,
      "learning_rate": 0.0001345214491431167,
      "loss": 2.0066,
      "step": 158052
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.3452942371368408,
      "learning_rate": 0.00013451803717165134,
      "loss": 2.0131,
      "step": 158053
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.1539270877838135,
      "learning_rate": 0.00013451462523095243,
      "loss": 2.0662,
      "step": 158054
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.186599850654602,
      "learning_rate": 0.0001345112133210202,
      "loss": 1.8971,
      "step": 158055
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.4906558990478516,
      "learning_rate": 0.00013450780144185553,
      "loss": 2.1447,
      "step": 158056
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.0830307006835938,
      "learning_rate": 0.00013450438959345884,
      "loss": 1.7812,
      "step": 158057
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.2316226959228516,
      "learning_rate": 0.00013450097777583108,
      "loss": 2.1457,
      "step": 158058
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.1772668361663818,
      "learning_rate": 0.00013449756598897256,
      "loss": 2.0743,
      "step": 158059
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.25916588306427,
      "learning_rate": 0.00013449415423288433,
      "loss": 2.1242,
      "step": 158060
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.1954388618469238,
      "learning_rate": 0.0001344907425075665,
      "loss": 1.994,
      "step": 158061
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.1834460496902466,
      "learning_rate": 0.0001344873308130201,
      "loss": 2.0765,
      "step": 158062
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.1534481048583984,
      "learning_rate": 0.0001344839191492455,
      "loss": 2.0509,
      "step": 158063
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.1034795045852661,
      "learning_rate": 0.0001344805075162436,
      "loss": 1.9562,
      "step": 158064
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.322421908378601,
      "learning_rate": 0.00013447709591401477,
      "loss": 2.1908,
      "step": 158065
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.306764841079712,
      "learning_rate": 0.00013447368434256,
      "loss": 2.0665,
      "step": 158066
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.4418526887893677,
      "learning_rate": 0.00013447027280187945,
      "loss": 2.1904,
      "step": 158067
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.2470202445983887,
      "learning_rate": 0.00013446686129197415,
      "loss": 1.9623,
      "step": 158068
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.1073148250579834,
      "learning_rate": 0.00013446344981284445,
      "loss": 2.1518,
      "step": 158069
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.2419122457504272,
      "learning_rate": 0.00013446003836449125,
      "loss": 1.9739,
      "step": 158070
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.1244635581970215,
      "learning_rate": 0.00013445662694691493,
      "loss": 1.957,
      "step": 158071
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.1494423151016235,
      "learning_rate": 0.00013445321556011645,
      "loss": 1.8049,
      "step": 158072
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.2116039991378784,
      "learning_rate": 0.00013444980420409606,
      "loss": 2.1634,
      "step": 158073
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.097358226776123,
      "learning_rate": 0.00013444639287885465,
      "loss": 2.064,
      "step": 158074
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.2127957344055176,
      "learning_rate": 0.00013444298158439268,
      "loss": 2.1326,
      "step": 158075
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.3023557662963867,
      "learning_rate": 0.00013443957032071102,
      "loss": 2.1867,
      "step": 158076
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.132737636566162,
      "learning_rate": 0.00013443615908781005,
      "loss": 2.2362,
      "step": 158077
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.201481580734253,
      "learning_rate": 0.00013443274788569058,
      "loss": 1.7041,
      "step": 158078
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.1104062795639038,
      "learning_rate": 0.00013442933671435327,
      "loss": 1.9651,
      "step": 158079
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.1010479927062988,
      "learning_rate": 0.00013442592557379862,
      "loss": 2.2121,
      "step": 158080
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.101211428642273,
      "learning_rate": 0.00013442251446402724,
      "loss": 2.1657,
      "step": 158081
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.296833872795105,
      "learning_rate": 0.00013441910338503994,
      "loss": 1.8876,
      "step": 158082
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.193341612815857,
      "learning_rate": 0.0001344156923368371,
      "loss": 2.0762,
      "step": 158083
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.3358649015426636,
      "learning_rate": 0.00013441228131941965,
      "loss": 2.2207,
      "step": 158084
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.1807283163070679,
      "learning_rate": 0.0001344088703327881,
      "loss": 2.0136,
      "step": 158085
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.1821370124816895,
      "learning_rate": 0.00013440545937694307,
      "loss": 2.0669,
      "step": 158086
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.1346580982208252,
      "learning_rate": 0.00013440204845188502,
      "loss": 2.0353,
      "step": 158087
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.0998743772506714,
      "learning_rate": 0.00013439863755761492,
      "loss": 2.1337,
      "step": 158088
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.33843195438385,
      "learning_rate": 0.0001343952266941331,
      "loss": 2.1595,
      "step": 158089
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.1097049713134766,
      "learning_rate": 0.00013439181586144048,
      "loss": 1.9458,
      "step": 158090
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.2644866704940796,
      "learning_rate": 0.00013438840505953755,
      "loss": 2.0541,
      "step": 158091
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.182438850402832,
      "learning_rate": 0.0001343849942884249,
      "loss": 2.0449,
      "step": 158092
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.1942074298858643,
      "learning_rate": 0.00013438158354810314,
      "loss": 2.1171,
      "step": 158093
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.144123911857605,
      "learning_rate": 0.000134378172838573,
      "loss": 2.0108,
      "step": 158094
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.177453875541687,
      "learning_rate": 0.00013437476215983504,
      "loss": 2.0558,
      "step": 158095
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.0624821186065674,
      "learning_rate": 0.00013437135151189003,
      "loss": 1.8926,
      "step": 158096
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.3907856941223145,
      "learning_rate": 0.00013436794089473854,
      "loss": 2.0351,
      "step": 158097
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.1175097227096558,
      "learning_rate": 0.00013436453030838118,
      "loss": 1.986,
      "step": 158098
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.185665488243103,
      "learning_rate": 0.00013436111975281844,
      "loss": 1.996,
      "step": 158099
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.1772116422653198,
      "learning_rate": 0.00013435770922805122,
      "loss": 2.0666,
      "step": 158100
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.119863510131836,
      "learning_rate": 0.00013435429873407987,
      "loss": 2.1234,
      "step": 158101
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.2013800144195557,
      "learning_rate": 0.00013435088827090536,
      "loss": 2.1331,
      "step": 158102
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.2060996294021606,
      "learning_rate": 0.0001343474778385281,
      "loss": 1.9239,
      "step": 158103
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.4386588335037231,
      "learning_rate": 0.0001343440674369488,
      "loss": 1.7725,
      "step": 158104
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.29684579372406,
      "learning_rate": 0.00013434065706616796,
      "loss": 1.987,
      "step": 158105
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.2779163122177124,
      "learning_rate": 0.0001343372467261864,
      "loss": 2.2535,
      "step": 158106
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.4366562366485596,
      "learning_rate": 0.0001343338364170046,
      "loss": 1.9606,
      "step": 158107
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.2668046951293945,
      "learning_rate": 0.00013433042613862333,
      "loss": 1.8744,
      "step": 158108
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.1365305185317993,
      "learning_rate": 0.00013432701589104318,
      "loss": 2.0045,
      "step": 158109
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.165400505065918,
      "learning_rate": 0.00013432360567426467,
      "loss": 2.2386,
      "step": 158110
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.3621509075164795,
      "learning_rate": 0.00013432019548828864,
      "loss": 2.1553,
      "step": 158111
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.2529122829437256,
      "learning_rate": 0.0001343167853331156,
      "loss": 1.929,
      "step": 158112
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.227465271949768,
      "learning_rate": 0.00013431337520874607,
      "loss": 2.0416,
      "step": 158113
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.153562307357788,
      "learning_rate": 0.00013430996511518096,
      "loss": 2.0868,
      "step": 158114
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.229862093925476,
      "learning_rate": 0.00013430655505242072,
      "loss": 2.2333,
      "step": 158115
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.314829707145691,
      "learning_rate": 0.00013430314502046594,
      "loss": 2.042,
      "step": 158116
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.2651437520980835,
      "learning_rate": 0.00013429973501931743,
      "loss": 2.2213,
      "step": 158117
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.1962417364120483,
      "learning_rate": 0.0001342963250489757,
      "loss": 2.0355,
      "step": 158118
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.281536340713501,
      "learning_rate": 0.00013429291510944132,
      "loss": 1.949,
      "step": 158119
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.47113835811615,
      "learning_rate": 0.00013428950520071514,
      "loss": 2.2879,
      "step": 158120
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.2442090511322021,
      "learning_rate": 0.00013428609532279765,
      "loss": 2.0842,
      "step": 158121
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.0738987922668457,
      "learning_rate": 0.00013428268547568942,
      "loss": 2.2309,
      "step": 158122
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.3264226913452148,
      "learning_rate": 0.00013427927565939126,
      "loss": 2.1427,
      "step": 158123
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.2352203130722046,
      "learning_rate": 0.0001342758658739036,
      "loss": 2.0128,
      "step": 158124
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.1651287078857422,
      "learning_rate": 0.0001342724561192273,
      "loss": 2.324,
      "step": 158125
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.348059892654419,
      "learning_rate": 0.00013426904639536288,
      "loss": 2.044,
      "step": 158126
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.3607419729232788,
      "learning_rate": 0.00013426563670231097,
      "loss": 2.0198,
      "step": 158127
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.2684910297393799,
      "learning_rate": 0.00013426222704007208,
      "loss": 1.9763,
      "step": 158128
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.161714792251587,
      "learning_rate": 0.0001342588174086471,
      "loss": 2.1565,
      "step": 158129
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.1327531337738037,
      "learning_rate": 0.0001342554078080364,
      "loss": 2.2691,
      "step": 158130
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.1918810606002808,
      "learning_rate": 0.0001342519982382409,
      "loss": 2.16,
      "step": 158131
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.1750190258026123,
      "learning_rate": 0.00013424858869926106,
      "loss": 2.0068,
      "step": 158132
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.190858244895935,
      "learning_rate": 0.00013424517919109752,
      "loss": 2.0627,
      "step": 158133
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.1618086099624634,
      "learning_rate": 0.00013424176971375083,
      "loss": 2.1087,
      "step": 158134
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.2767897844314575,
      "learning_rate": 0.00013423836026722183,
      "loss": 2.1543,
      "step": 158135
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.264174461364746,
      "learning_rate": 0.00013423495085151095,
      "loss": 1.9273,
      "step": 158136
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.343205451965332,
      "learning_rate": 0.00013423154146661903,
      "loss": 1.992,
      "step": 158137
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.1742066144943237,
      "learning_rate": 0.00013422813211254648,
      "loss": 2.1192,
      "step": 158138
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.1353965997695923,
      "learning_rate": 0.00013422472278929427,
      "loss": 2.0137,
      "step": 158139
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.1701722145080566,
      "learning_rate": 0.0001342213134968626,
      "loss": 2.105,
      "step": 158140
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.1812015771865845,
      "learning_rate": 0.00013421790423525244,
      "loss": 2.0294,
      "step": 158141
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.334215521812439,
      "learning_rate": 0.00013421449500446416,
      "loss": 2.1486,
      "step": 158142
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.1731665134429932,
      "learning_rate": 0.00013421108580449867,
      "loss": 2.0255,
      "step": 158143
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.1567721366882324,
      "learning_rate": 0.00013420767663535634,
      "loss": 2.1076,
      "step": 158144
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.315800666809082,
      "learning_rate": 0.00013420426749703807,
      "loss": 2.1418,
      "step": 158145
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.3023149967193604,
      "learning_rate": 0.00013420085838954435,
      "loss": 2.1533,
      "step": 158146
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.170814871788025,
      "learning_rate": 0.0001341974493128758,
      "loss": 2.1235,
      "step": 158147
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.2263460159301758,
      "learning_rate": 0.00013419404026703298,
      "loss": 2.0688,
      "step": 158148
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.120566964149475,
      "learning_rate": 0.00013419063125201668,
      "loss": 1.9825,
      "step": 158149
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.1241263151168823,
      "learning_rate": 0.00013418722226782743,
      "loss": 2.0068,
      "step": 158150
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.2115821838378906,
      "learning_rate": 0.00013418381331446597,
      "loss": 1.9512,
      "step": 158151
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.1551522016525269,
      "learning_rate": 0.00013418040439193288,
      "loss": 2.1259,
      "step": 158152
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.3063125610351562,
      "learning_rate": 0.00013417699550022878,
      "loss": 2.0183,
      "step": 158153
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.063908576965332,
      "learning_rate": 0.00013417358663935422,
      "loss": 1.9791,
      "step": 158154
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.2099978923797607,
      "learning_rate": 0.00013417017780931,
      "loss": 1.8899,
      "step": 158155
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.2558186054229736,
      "learning_rate": 0.0001341667690100966,
      "loss": 1.9848,
      "step": 158156
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.2090656757354736,
      "learning_rate": 0.00013416336024171482,
      "loss": 2.1302,
      "step": 158157
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.0560450553894043,
      "learning_rate": 0.0001341599515041652,
      "loss": 2.1001,
      "step": 158158
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.1286920309066772,
      "learning_rate": 0.00013415654279744837,
      "loss": 2.0329,
      "step": 158159
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.0773555040359497,
      "learning_rate": 0.00013415313412156488,
      "loss": 2.1318,
      "step": 158160
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.1739051342010498,
      "learning_rate": 0.00013414972547651552,
      "loss": 1.8217,
      "step": 158161
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.1781556606292725,
      "learning_rate": 0.00013414631686230077,
      "loss": 2.1011,
      "step": 158162
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.1902939081192017,
      "learning_rate": 0.00013414290827892147,
      "loss": 2.119,
      "step": 158163
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.1568331718444824,
      "learning_rate": 0.00013413949972637814,
      "loss": 2.147,
      "step": 158164
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.1572741270065308,
      "learning_rate": 0.0001341360912046714,
      "loss": 1.9863,
      "step": 158165
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.3362292051315308,
      "learning_rate": 0.00013413268271380177,
      "loss": 2.3299,
      "step": 158166
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.178476333618164,
      "learning_rate": 0.00013412927425377011,
      "loss": 1.8759,
      "step": 158167
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.1998218297958374,
      "learning_rate": 0.00013412586582457688,
      "loss": 2.0851,
      "step": 158168
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.124739408493042,
      "learning_rate": 0.00013412245742622285,
      "loss": 2.1182,
      "step": 158169
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.337365746498108,
      "learning_rate": 0.0001341190490587086,
      "loss": 2.2561,
      "step": 158170
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.2862021923065186,
      "learning_rate": 0.00013411564072203473,
      "loss": 2.3347,
      "step": 158171
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.1942853927612305,
      "learning_rate": 0.0001341122324162018,
      "loss": 2.2089,
      "step": 158172
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.1484456062316895,
      "learning_rate": 0.00013410882414121068,
      "loss": 2.0033,
      "step": 158173
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.2693480253219604,
      "learning_rate": 0.00013410541589706172,
      "loss": 2.0009,
      "step": 158174
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.1472963094711304,
      "learning_rate": 0.00013410200768375582,
      "loss": 2.0101,
      "step": 158175
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.2040801048278809,
      "learning_rate": 0.00013409859950129347,
      "loss": 2.0022,
      "step": 158176
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.2637925148010254,
      "learning_rate": 0.00013409519134967532,
      "loss": 2.0472,
      "step": 158177
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.4544188976287842,
      "learning_rate": 0.00013409178322890187,
      "loss": 1.9605,
      "step": 158178
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.1539506912231445,
      "learning_rate": 0.00013408837513897404,
      "loss": 2.0239,
      "step": 158179
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.2696644067764282,
      "learning_rate": 0.0001340849670798922,
      "loss": 2.0232,
      "step": 158180
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.2860913276672363,
      "learning_rate": 0.00013408155905165721,
      "loss": 2.1367,
      "step": 158181
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.1229037046432495,
      "learning_rate": 0.00013407815105426958,
      "loss": 1.7515,
      "step": 158182
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.1069629192352295,
      "learning_rate": 0.00013407474308773,
      "loss": 2.3383,
      "step": 158183
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.1835660934448242,
      "learning_rate": 0.00013407133515203886,
      "loss": 1.8199,
      "step": 158184
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.1259033679962158,
      "learning_rate": 0.00013406792724719716,
      "loss": 2.1022,
      "step": 158185
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.1977216005325317,
      "learning_rate": 0.00013406451937320526,
      "loss": 2.0688,
      "step": 158186
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.1815465688705444,
      "learning_rate": 0.00013406111153006398,
      "loss": 2.2053,
      "step": 158187
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.363010287284851,
      "learning_rate": 0.00013405770371777384,
      "loss": 1.9647,
      "step": 158188
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.1560720205307007,
      "learning_rate": 0.00013405429593633556,
      "loss": 2.0111,
      "step": 158189
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.2042607069015503,
      "learning_rate": 0.0001340508881857496,
      "loss": 1.8915,
      "step": 158190
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.2418493032455444,
      "learning_rate": 0.0001340474804660168,
      "loss": 1.9621,
      "step": 158191
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.420832633972168,
      "learning_rate": 0.00013404407277713763,
      "loss": 2.0851,
      "step": 158192
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.1851543188095093,
      "learning_rate": 0.0001340406651191129,
      "loss": 2.1253,
      "step": 158193
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.2152254581451416,
      "learning_rate": 0.00013403725749194312,
      "loss": 1.937,
      "step": 158194
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.2977616786956787,
      "learning_rate": 0.00013403384989562898,
      "loss": 2.0968,
      "step": 158195
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.2360299825668335,
      "learning_rate": 0.00013403044233017093,
      "loss": 2.2492,
      "step": 158196
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.237379550933838,
      "learning_rate": 0.00013402703479556988,
      "loss": 2.0543,
      "step": 158197
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.3112794160842896,
      "learning_rate": 0.00013402362729182622,
      "loss": 2.2162,
      "step": 158198
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.4136837720870972,
      "learning_rate": 0.00013402021981894082,
      "loss": 2.1729,
      "step": 158199
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.1398369073867798,
      "learning_rate": 0.00013401681237691423,
      "loss": 2.235,
      "step": 158200
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.1512069702148438,
      "learning_rate": 0.00013401340496574688,
      "loss": 1.908,
      "step": 158201
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.21701979637146,
      "learning_rate": 0.0001340099975854397,
      "loss": 2.4016,
      "step": 158202
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.460242509841919,
      "learning_rate": 0.0001340065902359932,
      "loss": 1.8766,
      "step": 158203
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.3463809490203857,
      "learning_rate": 0.00013400318291740792,
      "loss": 1.993,
      "step": 158204
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.273255467414856,
      "learning_rate": 0.00013399977562968465,
      "loss": 1.8736,
      "step": 158205
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.063949704170227,
      "learning_rate": 0.000133996368372824,
      "loss": 2.064,
      "step": 158206
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.2638338804244995,
      "learning_rate": 0.00013399296114682642,
      "loss": 2.397,
      "step": 158207
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.2832951545715332,
      "learning_rate": 0.0001339895539516928,
      "loss": 1.9145,
      "step": 158208
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.1057307720184326,
      "learning_rate": 0.00013398614678742355,
      "loss": 1.8411,
      "step": 158209
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.0528923273086548,
      "learning_rate": 0.0001339827396540195,
      "loss": 2.0415,
      "step": 158210
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.1614954471588135,
      "learning_rate": 0.0001339793325514812,
      "loss": 2.0888,
      "step": 158211
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.1589921712875366,
      "learning_rate": 0.00013397592547980918,
      "loss": 2.1003,
      "step": 158212
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.1492185592651367,
      "learning_rate": 0.00013397251843900426,
      "loss": 2.0269,
      "step": 158213
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.315619707107544,
      "learning_rate": 0.000133969111429067,
      "loss": 2.0509,
      "step": 158214
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.1986333131790161,
      "learning_rate": 0.0001339657044499979,
      "loss": 2.0285,
      "step": 158215
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.1591336727142334,
      "learning_rate": 0.0001339622975017978,
      "loss": 1.9105,
      "step": 158216
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.2078853845596313,
      "learning_rate": 0.00013395889058446725,
      "loss": 2.0339,
      "step": 158217
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.2470107078552246,
      "learning_rate": 0.00013395548369800678,
      "loss": 2.0092,
      "step": 158218
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.252947211265564,
      "learning_rate": 0.00013395207684241723,
      "loss": 2.101,
      "step": 158219
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.3043122291564941,
      "learning_rate": 0.00013394867001769914,
      "loss": 2.0567,
      "step": 158220
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.2455923557281494,
      "learning_rate": 0.00013394526322385298,
      "loss": 1.9682,
      "step": 158221
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.2467868328094482,
      "learning_rate": 0.00013394185646087964,
      "loss": 1.9784,
      "step": 158222
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.5027825832366943,
      "learning_rate": 0.00013393844972877956,
      "loss": 2.2841,
      "step": 158223
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.2813739776611328,
      "learning_rate": 0.0001339350430275536,
      "loss": 1.9651,
      "step": 158224
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.2346750497817993,
      "learning_rate": 0.00013393163635720218,
      "loss": 2.0863,
      "step": 158225
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.1389436721801758,
      "learning_rate": 0.00013392822971772602,
      "loss": 2.0085,
      "step": 158226
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.2810156345367432,
      "learning_rate": 0.00013392482310912565,
      "loss": 2.0597,
      "step": 158227
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.1693713665008545,
      "learning_rate": 0.00013392141653140188,
      "loss": 2.1831,
      "step": 158228
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.1882994174957275,
      "learning_rate": 0.00013391800998455516,
      "loss": 1.9176,
      "step": 158229
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.2628216743469238,
      "learning_rate": 0.00013391460346858633,
      "loss": 2.2183,
      "step": 158230
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.175492286682129,
      "learning_rate": 0.00013391119698349588,
      "loss": 2.1272,
      "step": 158231
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.1673164367675781,
      "learning_rate": 0.0001339077905292845,
      "loss": 2.1649,
      "step": 158232
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.134286880493164,
      "learning_rate": 0.00013390438410595268,
      "loss": 2.0443,
      "step": 158233
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.1776646375656128,
      "learning_rate": 0.00013390097771350128,
      "loss": 2.0855,
      "step": 158234
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.197577714920044,
      "learning_rate": 0.0001338975713519307,
      "loss": 2.127,
      "step": 158235
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.1181645393371582,
      "learning_rate": 0.00013389416502124183,
      "loss": 1.8118,
      "step": 158236
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.100663423538208,
      "learning_rate": 0.0001338907587214352,
      "loss": 2.0529,
      "step": 158237
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.200081706047058,
      "learning_rate": 0.00013388735245251136,
      "loss": 1.9741,
      "step": 158238
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.175613522529602,
      "learning_rate": 0.0001338839462144709,
      "loss": 2.0924,
      "step": 158239
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.1878306865692139,
      "learning_rate": 0.00013388054000731462,
      "loss": 1.9052,
      "step": 158240
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.3304784297943115,
      "learning_rate": 0.00013387713383104303,
      "loss": 1.9692,
      "step": 158241
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.1322765350341797,
      "learning_rate": 0.00013387372768565691,
      "loss": 2.192,
      "step": 158242
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.2171872854232788,
      "learning_rate": 0.0001338703215711568,
      "loss": 1.953,
      "step": 158243
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.191216230392456,
      "learning_rate": 0.00013386691548754334,
      "loss": 1.9499,
      "step": 158244
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.1916331052780151,
      "learning_rate": 0.00013386350943481705,
      "loss": 1.9877,
      "step": 158245
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.0816768407821655,
      "learning_rate": 0.00013386010341297875,
      "loss": 1.9297,
      "step": 158246
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.072420597076416,
      "learning_rate": 0.00013385669742202891,
      "loss": 1.9417,
      "step": 158247
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.1338642835617065,
      "learning_rate": 0.00013385329146196835,
      "loss": 1.8008,
      "step": 158248
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.1776790618896484,
      "learning_rate": 0.00013384988553279763,
      "loss": 1.9584,
      "step": 158249
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.1362749338150024,
      "learning_rate": 0.00013384647963451732,
      "loss": 1.9955,
      "step": 158250
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.2367750406265259,
      "learning_rate": 0.00013384307376712796,
      "loss": 2.0189,
      "step": 158251
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.3838058710098267,
      "learning_rate": 0.00013383966793063045,
      "loss": 2.1834,
      "step": 158252
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.1494786739349365,
      "learning_rate": 0.00013383626212502513,
      "loss": 2.0733,
      "step": 158253
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.2024825811386108,
      "learning_rate": 0.00013383285635031294,
      "loss": 1.9849,
      "step": 158254
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.2380902767181396,
      "learning_rate": 0.00013382945060649436,
      "loss": 2.0146,
      "step": 158255
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.1173170804977417,
      "learning_rate": 0.00013382604489356998,
      "loss": 2.0396,
      "step": 158256
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.2396304607391357,
      "learning_rate": 0.00013382263921154038,
      "loss": 1.9765,
      "step": 158257
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.2580012083053589,
      "learning_rate": 0.00013381923356040642,
      "loss": 1.8694,
      "step": 158258
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.113911509513855,
      "learning_rate": 0.00013381582794016846,
      "loss": 2.1419,
      "step": 158259
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.243171215057373,
      "learning_rate": 0.00013381242235082743,
      "loss": 2.1036,
      "step": 158260
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.2050440311431885,
      "learning_rate": 0.0001338090167923838,
      "loss": 1.8712,
      "step": 158261
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.1277117729187012,
      "learning_rate": 0.00013380561126483814,
      "loss": 1.5629,
      "step": 158262
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.1758263111114502,
      "learning_rate": 0.00013380220576819115,
      "loss": 1.7214,
      "step": 158263
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.1825889348983765,
      "learning_rate": 0.0001337988003024435,
      "loss": 2.2163,
      "step": 158264
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.1467781066894531,
      "learning_rate": 0.0001337953948675957,
      "loss": 2.0328,
      "step": 158265
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.127986192703247,
      "learning_rate": 0.0001337919894636486,
      "loss": 1.9818,
      "step": 158266
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.193324089050293,
      "learning_rate": 0.00013378858409060268,
      "loss": 2.099,
      "step": 158267
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.2838833332061768,
      "learning_rate": 0.00013378517874845861,
      "loss": 2.1365,
      "step": 158268
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.3212616443634033,
      "learning_rate": 0.0001337817734372169,
      "loss": 2.087,
      "step": 158269
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.1227830648422241,
      "learning_rate": 0.00013377836815687842,
      "loss": 1.9864,
      "step": 158270
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.4457085132598877,
      "learning_rate": 0.00013377496290744358,
      "loss": 2.2848,
      "step": 158271
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.103339672088623,
      "learning_rate": 0.0001337715576889132,
      "loss": 2.0559,
      "step": 158272
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.2798653841018677,
      "learning_rate": 0.0001337681525012877,
      "loss": 2.1031,
      "step": 158273
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.295150637626648,
      "learning_rate": 0.0001337647473445681,
      "loss": 2.228,
      "step": 158274
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.2794129848480225,
      "learning_rate": 0.0001337613422187545,
      "loss": 2.2546,
      "step": 158275
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.2058607339859009,
      "learning_rate": 0.00013375793712384795,
      "loss": 1.8544,
      "step": 158276
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.5701225996017456,
      "learning_rate": 0.00013375453205984882,
      "loss": 1.986,
      "step": 158277
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.2599929571151733,
      "learning_rate": 0.000133751127026758,
      "loss": 2.0184,
      "step": 158278
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.2679293155670166,
      "learning_rate": 0.00013374772202457582,
      "loss": 1.9824,
      "step": 158279
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.1059671640396118,
      "learning_rate": 0.00013374431705330333,
      "loss": 1.9216,
      "step": 158280
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.342424750328064,
      "learning_rate": 0.00013374091211294068,
      "loss": 2.0338,
      "step": 158281
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.2120959758758545,
      "learning_rate": 0.00013373750720348883,
      "loss": 2.215,
      "step": 158282
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.4212833642959595,
      "learning_rate": 0.00013373410232494823,
      "loss": 2.1611,
      "step": 158283
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.4841392040252686,
      "learning_rate": 0.00013373069747731972,
      "loss": 2.3018,
      "step": 158284
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.195974588394165,
      "learning_rate": 0.00013372729266060367,
      "loss": 2.0541,
      "step": 158285
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.2420995235443115,
      "learning_rate": 0.00013372388787480098,
      "loss": 2.0123,
      "step": 158286
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.4186127185821533,
      "learning_rate": 0.00013372048311991215,
      "loss": 2.0596,
      "step": 158287
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.4762252569198608,
      "learning_rate": 0.0001337170783959378,
      "loss": 2.1969,
      "step": 158288
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.344983458518982,
      "learning_rate": 0.00013371367370287852,
      "loss": 2.1791,
      "step": 158289
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.2274999618530273,
      "learning_rate": 0.0001337102690407351,
      "loss": 2.1125,
      "step": 158290
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.4748612642288208,
      "learning_rate": 0.00013370686440950796,
      "loss": 2.0501,
      "step": 158291
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.3756771087646484,
      "learning_rate": 0.00013370345980919795,
      "loss": 2.0066,
      "step": 158292
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.2323294878005981,
      "learning_rate": 0.00013370005523980565,
      "loss": 1.9864,
      "step": 158293
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.1161030530929565,
      "learning_rate": 0.00013369665070133162,
      "loss": 2.0582,
      "step": 158294
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.3157795667648315,
      "learning_rate": 0.0001336932461937764,
      "loss": 2.1207,
      "step": 158295
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.44249427318573,
      "learning_rate": 0.00013368984171714088,
      "loss": 2.2792,
      "step": 158296
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.080722451210022,
      "learning_rate": 0.00013368643727142542,
      "loss": 1.925,
      "step": 158297
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.15050208568573,
      "learning_rate": 0.0001336830328566309,
      "loss": 1.9815,
      "step": 158298
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.0941272974014282,
      "learning_rate": 0.0001336796284727579,
      "loss": 2.0699,
      "step": 158299
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.3995076417922974,
      "learning_rate": 0.00013367622411980683,
      "loss": 2.1739,
      "step": 158300
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.4916949272155762,
      "learning_rate": 0.00013367281979777864,
      "loss": 2.0202,
      "step": 158301
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.2536219358444214,
      "learning_rate": 0.00013366941550667377,
      "loss": 2.0225,
      "step": 158302
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.0965489149093628,
      "learning_rate": 0.00013366601124649284,
      "loss": 2.186,
      "step": 158303
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.1124643087387085,
      "learning_rate": 0.0001336626070172366,
      "loss": 1.8717,
      "step": 158304
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.2224966287612915,
      "learning_rate": 0.0001336592028189056,
      "loss": 1.9258,
      "step": 158305
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.3477336168289185,
      "learning_rate": 0.00013365579865150045,
      "loss": 1.921,
      "step": 158306
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.183162808418274,
      "learning_rate": 0.00013365239451502192,
      "loss": 2.0035,
      "step": 158307
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.0854039192199707,
      "learning_rate": 0.0001336489904094704,
      "loss": 2.0777,
      "step": 158308
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.0770279169082642,
      "learning_rate": 0.00013364558633484683,
      "loss": 2.234,
      "step": 158309
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.2672383785247803,
      "learning_rate": 0.0001336421822911517,
      "loss": 1.9532,
      "step": 158310
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.1654926538467407,
      "learning_rate": 0.00013363877827838556,
      "loss": 2.0785,
      "step": 158311
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.1767749786376953,
      "learning_rate": 0.00013363537429654905,
      "loss": 2.0062,
      "step": 158312
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.1867016553878784,
      "learning_rate": 0.00013363197034564294,
      "loss": 1.9329,
      "step": 158313
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.2143473625183105,
      "learning_rate": 0.00013362856642566768,
      "loss": 1.9084,
      "step": 158314
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.1027711629867554,
      "learning_rate": 0.00013362516253662414,
      "loss": 1.8976,
      "step": 158315
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.1959738731384277,
      "learning_rate": 0.00013362175867851284,
      "loss": 2.0029,
      "step": 158316
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.1638578176498413,
      "learning_rate": 0.00013361835485133436,
      "loss": 1.8469,
      "step": 158317
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.2796698808670044,
      "learning_rate": 0.00013361495105508928,
      "loss": 2.1455,
      "step": 158318
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.4271742105484009,
      "learning_rate": 0.00013361154728977844,
      "loss": 1.8373,
      "step": 158319
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.1018472909927368,
      "learning_rate": 0.00013360814355540222,
      "loss": 2.2593,
      "step": 158320
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.1501445770263672,
      "learning_rate": 0.00013360473985196152,
      "loss": 2.0282,
      "step": 158321
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.27059805393219,
      "learning_rate": 0.00013360133617945681,
      "loss": 2.079,
      "step": 158322
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.2672061920166016,
      "learning_rate": 0.0001335979325378888,
      "loss": 2.0331,
      "step": 158323
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.0750819444656372,
      "learning_rate": 0.0001335945289272579,
      "loss": 2.215,
      "step": 158324
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.202765941619873,
      "learning_rate": 0.00013359112534756505,
      "loss": 2.2142,
      "step": 158325
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.3230668306350708,
      "learning_rate": 0.00013358772179881067,
      "loss": 2.0944,
      "step": 158326
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.1739277839660645,
      "learning_rate": 0.00013358431828099554,
      "loss": 2.1565,
      "step": 158327
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.1614775657653809,
      "learning_rate": 0.0001335809147941203,
      "loss": 2.0873,
      "step": 158328
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.1695116758346558,
      "learning_rate": 0.00013357751133818543,
      "loss": 1.8429,
      "step": 158329
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.129421353340149,
      "learning_rate": 0.00013357410791319159,
      "loss": 2.135,
      "step": 158330
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.1815509796142578,
      "learning_rate": 0.00013357070451913952,
      "loss": 1.9778,
      "step": 158331
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.107343316078186,
      "learning_rate": 0.00013356730115602974,
      "loss": 1.9193,
      "step": 158332
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.2545150518417358,
      "learning_rate": 0.000133563897823863,
      "loss": 2.1703,
      "step": 158333
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.1686489582061768,
      "learning_rate": 0.00013356049452263995,
      "loss": 1.997,
      "step": 158334
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.1663215160369873,
      "learning_rate": 0.00013355709125236108,
      "loss": 1.9655,
      "step": 158335
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.121368408203125,
      "learning_rate": 0.00013355368801302696,
      "loss": 2.0638,
      "step": 158336
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.1408743858337402,
      "learning_rate": 0.00013355028480463855,
      "loss": 1.7939,
      "step": 158337
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.205865740776062,
      "learning_rate": 0.0001335468816271961,
      "loss": 2.3137,
      "step": 158338
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.166877269744873,
      "learning_rate": 0.00013354347848070056,
      "loss": 2.0731,
      "step": 158339
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.1783311367034912,
      "learning_rate": 0.00013354007536515232,
      "loss": 1.8978,
      "step": 158340
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.1595704555511475,
      "learning_rate": 0.00013353667228055234,
      "loss": 1.9318,
      "step": 158341
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.1541696786880493,
      "learning_rate": 0.0001335332692269008,
      "loss": 2.0197,
      "step": 158342
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.3159457445144653,
      "learning_rate": 0.0001335298662041987,
      "loss": 1.8311,
      "step": 158343
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.1492561101913452,
      "learning_rate": 0.00013352646321244642,
      "loss": 1.9625,
      "step": 158344
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.2130601406097412,
      "learning_rate": 0.00013352306025164482,
      "loss": 2.0738,
      "step": 158345
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.1440184116363525,
      "learning_rate": 0.00013351965732179432,
      "loss": 2.0981,
      "step": 158346
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.1303086280822754,
      "learning_rate": 0.0001335162544228959,
      "loss": 1.9941,
      "step": 158347
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.197213888168335,
      "learning_rate": 0.0001335128515549497,
      "loss": 2.0749,
      "step": 158348
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.2914323806762695,
      "learning_rate": 0.00013350944871795672,
      "loss": 2.1293,
      "step": 158349
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.1861021518707275,
      "learning_rate": 0.0001335060459119174,
      "loss": 2.0396,
      "step": 158350
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.292281150817871,
      "learning_rate": 0.0001335026431368325,
      "loss": 2.1052,
      "step": 158351
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.1992136240005493,
      "learning_rate": 0.00013349924039270252,
      "loss": 2.077,
      "step": 158352
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.163770318031311,
      "learning_rate": 0.00013349583767952839,
      "loss": 1.9361,
      "step": 158353
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.400818109512329,
      "learning_rate": 0.00013349243499731028,
      "loss": 2.1336,
      "step": 158354
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.1388156414031982,
      "learning_rate": 0.00013348903234604924,
      "loss": 2.0355,
      "step": 158355
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.1404997110366821,
      "learning_rate": 0.0001334856297257456,
      "loss": 1.9021,
      "step": 158356
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.2446430921554565,
      "learning_rate": 0.00013348222713640022,
      "loss": 2.1304,
      "step": 158357
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.2843166589736938,
      "learning_rate": 0.00013347882457801353,
      "loss": 2.1619,
      "step": 158358
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.096591591835022,
      "learning_rate": 0.0001334754220505865,
      "loss": 1.9971,
      "step": 158359
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.1399251222610474,
      "learning_rate": 0.00013347201955411932,
      "loss": 2.2121,
      "step": 158360
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.1079461574554443,
      "learning_rate": 0.0001334686170886129,
      "loss": 2.0325,
      "step": 158361
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.1454118490219116,
      "learning_rate": 0.00013346521465406772,
      "loss": 1.9428,
      "step": 158362
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.2940155267715454,
      "learning_rate": 0.00013346181225048465,
      "loss": 1.9779,
      "step": 158363
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.1599693298339844,
      "learning_rate": 0.00013345840987786403,
      "loss": 2.206,
      "step": 158364
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.220658779144287,
      "learning_rate": 0.00013345500753620684,
      "loss": 2.0077,
      "step": 158365
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.243301272392273,
      "learning_rate": 0.00013345160522551332,
      "loss": 2.0423,
      "step": 158366
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.2128201723098755,
      "learning_rate": 0.00013344820294578438,
      "loss": 2.0292,
      "step": 158367
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.1430045366287231,
      "learning_rate": 0.00013344480069702046,
      "loss": 2.1913,
      "step": 158368
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.2686830759048462,
      "learning_rate": 0.0001334413984792224,
      "loss": 2.0325,
      "step": 158369
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.1056370735168457,
      "learning_rate": 0.0001334379962923906,
      "loss": 1.985,
      "step": 158370
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.2152677774429321,
      "learning_rate": 0.00013343459413652599,
      "loss": 2.0919,
      "step": 158371
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.1082768440246582,
      "learning_rate": 0.00013343119201162897,
      "loss": 2.1036,
      "step": 158372
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.126582145690918,
      "learning_rate": 0.00013342778991770027,
      "loss": 1.9603,
      "step": 158373
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.3593660593032837,
      "learning_rate": 0.00013342438785474036,
      "loss": 1.9987,
      "step": 158374
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.085946798324585,
      "learning_rate": 0.00013342098582275015,
      "loss": 1.917,
      "step": 158375
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.2141220569610596,
      "learning_rate": 0.00013341758382173,
      "loss": 2.3117,
      "step": 158376
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.3485814332962036,
      "learning_rate": 0.00013341418185168073,
      "loss": 2.0946,
      "step": 158377
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.5300824642181396,
      "learning_rate": 0.00013341077991260296,
      "loss": 2.0889,
      "step": 158378
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.4023158550262451,
      "learning_rate": 0.00013340737800449724,
      "loss": 1.8909,
      "step": 158379
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.4046543836593628,
      "learning_rate": 0.00013340397612736412,
      "loss": 2.1269,
      "step": 158380
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.2059162855148315,
      "learning_rate": 0.00013340057428120448,
      "loss": 1.8422,
      "step": 158381
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.120316743850708,
      "learning_rate": 0.00013339717246601868,
      "loss": 1.9082,
      "step": 158382
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.1712514162063599,
      "learning_rate": 0.0001333937706818076,
      "loss": 1.9721,
      "step": 158383
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.118467926979065,
      "learning_rate": 0.00013339036892857177,
      "loss": 1.9294,
      "step": 158384
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.1795063018798828,
      "learning_rate": 0.00013338696720631175,
      "loss": 2.069,
      "step": 158385
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.2934821844100952,
      "learning_rate": 0.0001333835655150283,
      "loss": 2.1188,
      "step": 158386
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.2438101768493652,
      "learning_rate": 0.000133380163854722,
      "loss": 2.1654,
      "step": 158387
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.1567680835723877,
      "learning_rate": 0.00013337676222539336,
      "loss": 2.1179,
      "step": 158388
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.2450292110443115,
      "learning_rate": 0.00013337336062704323,
      "loss": 1.9323,
      "step": 158389
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.2556047439575195,
      "learning_rate": 0.00013336995905967216,
      "loss": 2.1337,
      "step": 158390
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.2218009233474731,
      "learning_rate": 0.00013336655752328063,
      "loss": 2.1233,
      "step": 158391
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.1616944074630737,
      "learning_rate": 0.00013336315601786953,
      "loss": 2.1314,
      "step": 158392
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.1759421825408936,
      "learning_rate": 0.00013335975454343933,
      "loss": 1.9592,
      "step": 158393
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.1893037557601929,
      "learning_rate": 0.00013335635309999065,
      "loss": 2.2241,
      "step": 158394
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.2425318956375122,
      "learning_rate": 0.00013335295168752422,
      "loss": 1.9008,
      "step": 158395
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.1801775693893433,
      "learning_rate": 0.00013334955030604065,
      "loss": 2.0416,
      "step": 158396
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.2644414901733398,
      "learning_rate": 0.00013334614895554043,
      "loss": 2.0791,
      "step": 158397
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.2068172693252563,
      "learning_rate": 0.00013334274763602443,
      "loss": 1.9729,
      "step": 158398
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.3174911737442017,
      "learning_rate": 0.000133339346347493,
      "loss": 1.9588,
      "step": 158399
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.4565256834030151,
      "learning_rate": 0.00013333594508994708,
      "loss": 1.9124,
      "step": 158400
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.281476616859436,
      "learning_rate": 0.00013333254386338716,
      "loss": 2.2384,
      "step": 158401
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.4044846296310425,
      "learning_rate": 0.00013332914266781385,
      "loss": 2.1213,
      "step": 158402
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.1706593036651611,
      "learning_rate": 0.00013332574150322772,
      "loss": 1.9969,
      "step": 158403
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.1464303731918335,
      "learning_rate": 0.00013332234036962954,
      "loss": 1.9778,
      "step": 158404
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.185646891593933,
      "learning_rate": 0.00013331893926701982,
      "loss": 1.9674,
      "step": 158405
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.1518303155899048,
      "learning_rate": 0.0001333155381953993,
      "loss": 2.2164,
      "step": 158406
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.0941351652145386,
      "learning_rate": 0.0001333121371547685,
      "loss": 2.094,
      "step": 158407
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.1751360893249512,
      "learning_rate": 0.00013330873614512837,
      "loss": 1.931,
      "step": 158408
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.150978446006775,
      "learning_rate": 0.00013330533516647903,
      "loss": 1.9118,
      "step": 158409
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.170244574546814,
      "learning_rate": 0.0001333019342188215,
      "loss": 1.9258,
      "step": 158410
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.251703143119812,
      "learning_rate": 0.0001332985333021562,
      "loss": 2.0152,
      "step": 158411
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.0964157581329346,
      "learning_rate": 0.0001332951324164839,
      "loss": 1.96,
      "step": 158412
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.2218917608261108,
      "learning_rate": 0.00013329173156180515,
      "loss": 1.8711,
      "step": 158413
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.1984996795654297,
      "learning_rate": 0.00013328833073812077,
      "loss": 2.0097,
      "step": 158414
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.1414597034454346,
      "learning_rate": 0.00013328492994543102,
      "loss": 2.0765,
      "step": 158415
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.236628532409668,
      "learning_rate": 0.0001332815291837369,
      "loss": 2.087,
      "step": 158416
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.2938064336776733,
      "learning_rate": 0.00013327812845303875,
      "loss": 1.8317,
      "step": 158417
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.1913561820983887,
      "learning_rate": 0.00013327472775333745,
      "loss": 2.0906,
      "step": 158418
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.1329745054244995,
      "learning_rate": 0.0001332713270846334,
      "loss": 1.9481,
      "step": 158419
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.2417181730270386,
      "learning_rate": 0.00013326792644692762,
      "loss": 2.091,
      "step": 158420
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.134546160697937,
      "learning_rate": 0.00013326452584022026,
      "loss": 2.0241,
      "step": 158421
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.2406359910964966,
      "learning_rate": 0.00013326112526451227,
      "loss": 1.9793,
      "step": 158422
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.1482183933258057,
      "learning_rate": 0.00013325772471980404,
      "loss": 1.9309,
      "step": 158423
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.2239469289779663,
      "learning_rate": 0.0001332543242060965,
      "loss": 2.1609,
      "step": 158424
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.2145190238952637,
      "learning_rate": 0.00013325092372339,
      "loss": 1.8698,
      "step": 158425
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.2487717866897583,
      "learning_rate": 0.00013324752327168552,
      "loss": 1.9444,
      "step": 158426
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.2043718099594116,
      "learning_rate": 0.00013324412285098324,
      "loss": 2.0077,
      "step": 158427
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.3997255563735962,
      "learning_rate": 0.00013324072246128418,
      "loss": 1.8421,
      "step": 158428
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.1305687427520752,
      "learning_rate": 0.00013323732210258869,
      "loss": 1.9564,
      "step": 158429
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.2199541330337524,
      "learning_rate": 0.0001332339217748976,
      "loss": 2.1289,
      "step": 158430
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.2506203651428223,
      "learning_rate": 0.0001332305214782114,
      "loss": 1.7392,
      "step": 158431
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.256121039390564,
      "learning_rate": 0.00013322712121253104,
      "loss": 2.1432,
      "step": 158432
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.2503597736358643,
      "learning_rate": 0.00013322372097785665,
      "loss": 1.9493,
      "step": 158433
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.1538139581680298,
      "learning_rate": 0.00013322032077418922,
      "loss": 2.15,
      "step": 158434
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.1698874235153198,
      "learning_rate": 0.00013321692060152916,
      "loss": 2.1637,
      "step": 158435
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.2464284896850586,
      "learning_rate": 0.00013321352045987734,
      "loss": 1.8616,
      "step": 158436
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.2356544733047485,
      "learning_rate": 0.0001332101203492342,
      "loss": 2.1895,
      "step": 158437
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.210866928100586,
      "learning_rate": 0.00013320672026960063,
      "loss": 2.101,
      "step": 158438
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.3365322351455688,
      "learning_rate": 0.00013320332022097688,
      "loss": 2.0234,
      "step": 158439
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.0966023206710815,
      "learning_rate": 0.00013319992020336392,
      "loss": 2.0224,
      "step": 158440
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.2485636472702026,
      "learning_rate": 0.0001331965202167621,
      "loss": 2.0263,
      "step": 158441
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.228965163230896,
      "learning_rate": 0.0001331931202611723,
      "loss": 2.0836,
      "step": 158442
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.2228654623031616,
      "learning_rate": 0.0001331897203365949,
      "loss": 2.0602,
      "step": 158443
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.3807430267333984,
      "learning_rate": 0.00013318632044303095,
      "loss": 2.0041,
      "step": 158444
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.1605138778686523,
      "learning_rate": 0.0001331829205804806,
      "loss": 1.9457,
      "step": 158445
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.2555180788040161,
      "learning_rate": 0.0001331795207489448,
      "loss": 2.3211,
      "step": 158446
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.2334896326065063,
      "learning_rate": 0.00013317612094842395,
      "loss": 2.122,
      "step": 158447
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.5667914152145386,
      "learning_rate": 0.00013317272117891892,
      "loss": 2.3599,
      "step": 158448
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.1987321376800537,
      "learning_rate": 0.0001331693214404301,
      "loss": 2.0371,
      "step": 158449
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.1138514280319214,
      "learning_rate": 0.0001331659217329585,
      "loss": 1.9682,
      "step": 158450
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.2491306066513062,
      "learning_rate": 0.0001331625220565043,
      "loss": 1.9463,
      "step": 158451
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.1449592113494873,
      "learning_rate": 0.00013315912241106841,
      "loss": 2.0191,
      "step": 158452
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.2747374773025513,
      "learning_rate": 0.00013315572279665132,
      "loss": 2.0663,
      "step": 158453
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.1998416185379028,
      "learning_rate": 0.00013315232321325378,
      "loss": 2.0053,
      "step": 158454
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.1521048545837402,
      "learning_rate": 0.0001331489236608763,
      "loss": 2.1263,
      "step": 158455
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.1430063247680664,
      "learning_rate": 0.00013314552413951983,
      "loss": 2.0602,
      "step": 158456
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.2454333305358887,
      "learning_rate": 0.0001331421246491845,
      "loss": 2.119,
      "step": 158457
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.0941585302352905,
      "learning_rate": 0.00013313872518987133,
      "loss": 2.159,
      "step": 158458
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.3633794784545898,
      "learning_rate": 0.00013313532576158068,
      "loss": 2.0218,
      "step": 158459
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.0881268978118896,
      "learning_rate": 0.00013313192636431345,
      "loss": 1.8996,
      "step": 158460
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.2052667140960693,
      "learning_rate": 0.00013312852699807006,
      "loss": 1.9395,
      "step": 158461
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.221685528755188,
      "learning_rate": 0.00013312512766285132,
      "loss": 1.8197,
      "step": 158462
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.1956051588058472,
      "learning_rate": 0.00013312172835865774,
      "loss": 2.1795,
      "step": 158463
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.2390838861465454,
      "learning_rate": 0.00013311832908549,
      "loss": 2.1986,
      "step": 158464
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.2806226015090942,
      "learning_rate": 0.0001331149298433486,
      "loss": 2.0727,
      "step": 158465
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.1918354034423828,
      "learning_rate": 0.0001331115306322344,
      "loss": 2.08,
      "step": 158466
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.2658640146255493,
      "learning_rate": 0.0001331081314521478,
      "loss": 2.1828,
      "step": 158467
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.243695616722107,
      "learning_rate": 0.0001331047323030897,
      "loss": 2.0704,
      "step": 158468
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.0921404361724854,
      "learning_rate": 0.00013310133318506054,
      "loss": 2.2169,
      "step": 158469
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.233073353767395,
      "learning_rate": 0.00013309793409806085,
      "loss": 2.2355,
      "step": 158470
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.1734023094177246,
      "learning_rate": 0.00013309453504209157,
      "loss": 1.9585,
      "step": 158471
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.3117798566818237,
      "learning_rate": 0.00013309113601715316,
      "loss": 2.136,
      "step": 158472
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.2357653379440308,
      "learning_rate": 0.00013308773702324614,
      "loss": 2.1802,
      "step": 158473
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.4018501043319702,
      "learning_rate": 0.00013308433806037136,
      "loss": 2.1489,
      "step": 158474
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.1537293195724487,
      "learning_rate": 0.00013308093912852936,
      "loss": 2.1853,
      "step": 158475
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.1938732862472534,
      "learning_rate": 0.00013307754022772066,
      "loss": 2.168,
      "step": 158476
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.116098165512085,
      "learning_rate": 0.0001330741413579461,
      "loss": 1.9286,
      "step": 158477
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.1949385404586792,
      "learning_rate": 0.00013307074251920622,
      "loss": 1.9347,
      "step": 158478
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.2529557943344116,
      "learning_rate": 0.00013306734371150149,
      "loss": 2.1892,
      "step": 158479
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.13974130153656,
      "learning_rate": 0.00013306394493483285,
      "loss": 1.9199,
      "step": 158480
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.0968948602676392,
      "learning_rate": 0.00013306054618920075,
      "loss": 1.7943,
      "step": 158481
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.2557193040847778,
      "learning_rate": 0.00013305714747460573,
      "loss": 2.0775,
      "step": 158482
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.2370595932006836,
      "learning_rate": 0.00013305374879104867,
      "loss": 2.0376,
      "step": 158483
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.1915847063064575,
      "learning_rate": 0.00013305035013852997,
      "loss": 1.8578,
      "step": 158484
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.2133227586746216,
      "learning_rate": 0.00013304695151705043,
      "loss": 2.0015,
      "step": 158485
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.210411787033081,
      "learning_rate": 0.00013304355292661065,
      "loss": 2.1733,
      "step": 158486
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.2940011024475098,
      "learning_rate": 0.0001330401543672112,
      "loss": 2.0401,
      "step": 158487
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.2297958135604858,
      "learning_rate": 0.00013303675583885264,
      "loss": 2.0582,
      "step": 158488
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.2133398056030273,
      "learning_rate": 0.00013303335734153577,
      "loss": 2.1504,
      "step": 158489
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.2649579048156738,
      "learning_rate": 0.00013302995887526107,
      "loss": 1.8725,
      "step": 158490
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.2008144855499268,
      "learning_rate": 0.00013302656044002938,
      "loss": 2.1978,
      "step": 158491
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.145785927772522,
      "learning_rate": 0.0001330231620358411,
      "loss": 1.9399,
      "step": 158492
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.2098519802093506,
      "learning_rate": 0.00013301976366269716,
      "loss": 2.1222,
      "step": 158493
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.1217598915100098,
      "learning_rate": 0.00013301636532059776,
      "loss": 2.0357,
      "step": 158494
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.14174485206604,
      "learning_rate": 0.00013301296700954393,
      "loss": 2.0076,
      "step": 158495
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.1612120866775513,
      "learning_rate": 0.000133009568729536,
      "loss": 2.3157,
      "step": 158496
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.2371916770935059,
      "learning_rate": 0.00013300617048057485,
      "loss": 2.0516,
      "step": 158497
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.1231287717819214,
      "learning_rate": 0.0001330027722626609,
      "loss": 2.0525,
      "step": 158498
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.2192604541778564,
      "learning_rate": 0.00013299937407579513,
      "loss": 2.1526,
      "step": 158499
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.1914373636245728,
      "learning_rate": 0.00013299597591997768,
      "loss": 2.1493,
      "step": 158500
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.1406184434890747,
      "learning_rate": 0.00013299257779520956,
      "loss": 2.005,
      "step": 158501
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.1899776458740234,
      "learning_rate": 0.00013298917970149115,
      "loss": 2.2309,
      "step": 158502
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.157997727394104,
      "learning_rate": 0.0001329857816388233,
      "loss": 2.0693,
      "step": 158503
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.1225382089614868,
      "learning_rate": 0.0001329823836072065,
      "loss": 2.0469,
      "step": 158504
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.1677643060684204,
      "learning_rate": 0.00013297898560664156,
      "loss": 1.9983,
      "step": 158505
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.1108168363571167,
      "learning_rate": 0.00013297558763712878,
      "loss": 1.967,
      "step": 158506
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.1614551544189453,
      "learning_rate": 0.00013297218969866915,
      "loss": 2.0787,
      "step": 158507
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.1454429626464844,
      "learning_rate": 0.000132968791791263,
      "loss": 1.9796,
      "step": 158508
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.2147691249847412,
      "learning_rate": 0.0001329653939149112,
      "loss": 2.1275,
      "step": 158509
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.0951517820358276,
      "learning_rate": 0.0001329619960696142,
      "loss": 2.0163,
      "step": 158510
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.2378041744232178,
      "learning_rate": 0.00013295859825537292,
      "loss": 1.9436,
      "step": 158511
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.1395007371902466,
      "learning_rate": 0.00013295520047218763,
      "loss": 2.2529,
      "step": 158512
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.209111213684082,
      "learning_rate": 0.00013295180272005917,
      "loss": 2.1077,
      "step": 158513
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.114003300666809,
      "learning_rate": 0.000132948404998988,
      "loss": 1.8054,
      "step": 158514
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.2271431684494019,
      "learning_rate": 0.00013294500730897502,
      "loss": 2.1691,
      "step": 158515
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.2213025093078613,
      "learning_rate": 0.0001329416096500206,
      "loss": 2.2205,
      "step": 158516
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.0685561895370483,
      "learning_rate": 0.00013293821202212572,
      "loss": 1.9007,
      "step": 158517
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.148185133934021,
      "learning_rate": 0.00013293481442529052,
      "loss": 2.2526,
      "step": 158518
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.2456456422805786,
      "learning_rate": 0.00013293141685951604,
      "loss": 2.1824,
      "step": 158519
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.3257381916046143,
      "learning_rate": 0.00013292801932480265,
      "loss": 2.0895,
      "step": 158520
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.3402011394500732,
      "learning_rate": 0.00013292462182115121,
      "loss": 1.9472,
      "step": 158521
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.2288016080856323,
      "learning_rate": 0.00013292122434856212,
      "loss": 2.2406,
      "step": 158522
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.2510490417480469,
      "learning_rate": 0.00013291782690703634,
      "loss": 2.135,
      "step": 158523
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.23127281665802,
      "learning_rate": 0.00013291442949657403,
      "loss": 2.0539,
      "step": 158524
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.2494032382965088,
      "learning_rate": 0.00013291103211717626,
      "loss": 2.1099,
      "step": 158525
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.185032844543457,
      "learning_rate": 0.00013290763476884334,
      "loss": 1.9889,
      "step": 158526
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.204982042312622,
      "learning_rate": 0.00013290423745157616,
      "loss": 1.9937,
      "step": 158527
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.1917563676834106,
      "learning_rate": 0.0001329008401653751,
      "loss": 2.1906,
      "step": 158528
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.1574413776397705,
      "learning_rate": 0.0001328974429102412,
      "loss": 2.0149,
      "step": 158529
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.1817420721054077,
      "learning_rate": 0.0001328940456861746,
      "loss": 1.9646,
      "step": 158530
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.1256537437438965,
      "learning_rate": 0.0001328906484931762,
      "loss": 2.0922,
      "step": 158531
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.193930983543396,
      "learning_rate": 0.0001328872513312465,
      "loss": 1.8486,
      "step": 158532
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.2721301317214966,
      "learning_rate": 0.00013288385420038635,
      "loss": 1.8358,
      "step": 158533
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.2291513681411743,
      "learning_rate": 0.0001328804571005961,
      "loss": 1.9582,
      "step": 158534
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.3403823375701904,
      "learning_rate": 0.0001328770600318768,
      "loss": 2.1124,
      "step": 158535
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.1440938711166382,
      "learning_rate": 0.00013287366299422854,
      "loss": 2.039,
      "step": 158536
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.0452383756637573,
      "learning_rate": 0.00013287026598765233,
      "loss": 1.8165,
      "step": 158537
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.3304804563522339,
      "learning_rate": 0.00013286686901214863,
      "loss": 2.0253,
      "step": 158538
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.2870371341705322,
      "learning_rate": 0.0001328634720677182,
      "loss": 1.8658,
      "step": 158539
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.192679524421692,
      "learning_rate": 0.00013286007515436152,
      "loss": 2.1844,
      "step": 158540
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.3702322244644165,
      "learning_rate": 0.0001328566782720794,
      "loss": 2.1961,
      "step": 158541
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.4062505960464478,
      "learning_rate": 0.00013285328142087236,
      "loss": 1.9854,
      "step": 158542
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.2345715761184692,
      "learning_rate": 0.0001328498846007411,
      "loss": 1.9494,
      "step": 158543
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.2725411653518677,
      "learning_rate": 0.00013284648781168606,
      "loss": 1.9768,
      "step": 158544
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.1941187381744385,
      "learning_rate": 0.00013284309105370814,
      "loss": 2.1025,
      "step": 158545
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.2607569694519043,
      "learning_rate": 0.0001328396943268077,
      "loss": 2.0574,
      "step": 158546
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.1004070043563843,
      "learning_rate": 0.00013283629763098567,
      "loss": 1.8687,
      "step": 158547
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.3457143306732178,
      "learning_rate": 0.0001328329009662425,
      "loss": 2.0495,
      "step": 158548
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.1236906051635742,
      "learning_rate": 0.00013282950433257887,
      "loss": 2.1822,
      "step": 158549
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.1733590364456177,
      "learning_rate": 0.00013282610772999526,
      "loss": 2.0058,
      "step": 158550
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.1967490911483765,
      "learning_rate": 0.00013282271115849255,
      "loss": 2.0411,
      "step": 158551
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.5595818758010864,
      "learning_rate": 0.00013281931461807113,
      "loss": 2.1327,
      "step": 158552
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.1323270797729492,
      "learning_rate": 0.00013281591810873188,
      "loss": 1.9244,
      "step": 158553
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.2123231887817383,
      "learning_rate": 0.0001328125216304753,
      "loss": 1.9477,
      "step": 158554
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.2092945575714111,
      "learning_rate": 0.00013280912518330194,
      "loss": 2.1077,
      "step": 158555
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.1181098222732544,
      "learning_rate": 0.00013280572876721262,
      "loss": 2.1337,
      "step": 158556
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.3722078800201416,
      "learning_rate": 0.00013280233238220786,
      "loss": 2.178,
      "step": 158557
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.25238835811615,
      "learning_rate": 0.0001327989360282882,
      "loss": 2.0438,
      "step": 158558
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.1391644477844238,
      "learning_rate": 0.00013279553970545449,
      "loss": 1.9865,
      "step": 158559
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.5068436861038208,
      "learning_rate": 0.00013279214341370722,
      "loss": 1.9581,
      "step": 158560
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.1587249040603638,
      "learning_rate": 0.00013278874715304698,
      "loss": 2.0666,
      "step": 158561
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.2289406061172485,
      "learning_rate": 0.00013278535092347453,
      "loss": 2.1409,
      "step": 158562
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.3098629713058472,
      "learning_rate": 0.00013278195472499045,
      "loss": 2.1149,
      "step": 158563
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.2055997848510742,
      "learning_rate": 0.00013277855855759529,
      "loss": 2.0598,
      "step": 158564
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.2264621257781982,
      "learning_rate": 0.0001327751624212898,
      "loss": 2.0576,
      "step": 158565
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.2630306482315063,
      "learning_rate": 0.00013277176631607462,
      "loss": 2.1432,
      "step": 158566
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.3209799528121948,
      "learning_rate": 0.00013276837024195018,
      "loss": 1.9703,
      "step": 158567
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.176196813583374,
      "learning_rate": 0.0001327649741989174,
      "loss": 2.1124,
      "step": 158568
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.1799132823944092,
      "learning_rate": 0.00013276157818697663,
      "loss": 2.0509,
      "step": 158569
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.2991305589675903,
      "learning_rate": 0.00013275818220612876,
      "loss": 2.134,
      "step": 158570
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.093374252319336,
      "learning_rate": 0.00013275478625637428,
      "loss": 1.954,
      "step": 158571
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.136715054512024,
      "learning_rate": 0.00013275139033771388,
      "loss": 2.0652,
      "step": 158572
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.4739806652069092,
      "learning_rate": 0.00013274799445014802,
      "loss": 2.2226,
      "step": 158573
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.1322840452194214,
      "learning_rate": 0.00013274459859367757,
      "loss": 2.2307,
      "step": 158574
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.3020182847976685,
      "learning_rate": 0.00013274120276830298,
      "loss": 1.9615,
      "step": 158575
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.1752467155456543,
      "learning_rate": 0.00013273780697402505,
      "loss": 2.2257,
      "step": 158576
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.1664767265319824,
      "learning_rate": 0.00013273441121084429,
      "loss": 2.1391,
      "step": 158577
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.3430190086364746,
      "learning_rate": 0.00013273101547876137,
      "loss": 2.2606,
      "step": 158578
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.2856322526931763,
      "learning_rate": 0.00013272761977777683,
      "loss": 2.0808,
      "step": 158579
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.2245079278945923,
      "learning_rate": 0.00013272422410789148,
      "loss": 1.9649,
      "step": 158580
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.113317608833313,
      "learning_rate": 0.00013272082846910575,
      "loss": 2.024,
      "step": 158581
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.1656568050384521,
      "learning_rate": 0.00013271743286142048,
      "loss": 2.0282,
      "step": 158582
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.509967565536499,
      "learning_rate": 0.00013271403728483608,
      "loss": 2.2306,
      "step": 158583
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.1008158922195435,
      "learning_rate": 0.00013271064173935352,
      "loss": 1.7939,
      "step": 158584
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.2089170217514038,
      "learning_rate": 0.00013270724622497298,
      "loss": 1.9664,
      "step": 158585
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.1849300861358643,
      "learning_rate": 0.00013270385074169543,
      "loss": 1.9972,
      "step": 158586
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.1694632768630981,
      "learning_rate": 0.0001327004552895213,
      "loss": 1.9741,
      "step": 158587
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.3864655494689941,
      "learning_rate": 0.00013269705986845143,
      "loss": 1.9778,
      "step": 158588
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.1916980743408203,
      "learning_rate": 0.00013269366447848623,
      "loss": 1.7613,
      "step": 158589
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.418013334274292,
      "learning_rate": 0.00013269026911962663,
      "loss": 2.1778,
      "step": 158590
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.2894971370697021,
      "learning_rate": 0.00013268687379187287,
      "loss": 2.0028,
      "step": 158591
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.2216408252716064,
      "learning_rate": 0.00013268347849522587,
      "loss": 2.2122,
      "step": 158592
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.3986366987228394,
      "learning_rate": 0.00013268008322968604,
      "loss": 2.126,
      "step": 158593
    },
    {
      "epoch": 2.06,
      "grad_norm": 1.1462396383285522,
      "learning_rate": 0.0001326766879952543,
      "loss": 2.0172,
      "step": 158594
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.233061671257019,
      "learning_rate": 0.00013267329279193098,
      "loss": 2.0924,
      "step": 158595
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.176859974861145,
      "learning_rate": 0.00013266989761971708,
      "loss": 1.8504,
      "step": 158596
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.197169303894043,
      "learning_rate": 0.0001326665024786128,
      "loss": 2.2076,
      "step": 158597
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.1738488674163818,
      "learning_rate": 0.00013266310736861905,
      "loss": 2.2808,
      "step": 158598
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.1694176197052002,
      "learning_rate": 0.0001326597122897363,
      "loss": 1.9682,
      "step": 158599
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.1685338020324707,
      "learning_rate": 0.00013265631724196538,
      "loss": 1.9443,
      "step": 158600
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.1856539249420166,
      "learning_rate": 0.0001326529222253067,
      "loss": 2.0213,
      "step": 158601
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.5137499570846558,
      "learning_rate": 0.00013264952723976122,
      "loss": 2.1091,
      "step": 158602
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.2490110397338867,
      "learning_rate": 0.00013264613228532914,
      "loss": 2.0544,
      "step": 158603
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.214280128479004,
      "learning_rate": 0.0001326427373620114,
      "loss": 1.994,
      "step": 158604
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.4727824926376343,
      "learning_rate": 0.00013263934246980841,
      "loss": 2.0492,
      "step": 158605
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.2809077501296997,
      "learning_rate": 0.00013263594760872107,
      "loss": 2.1821,
      "step": 158606
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.1504706144332886,
      "learning_rate": 0.00013263255277874974,
      "loss": 1.9174,
      "step": 158607
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.1668314933776855,
      "learning_rate": 0.00013262915797989527,
      "loss": 1.9375,
      "step": 158608
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.3535789251327515,
      "learning_rate": 0.0001326257632121582,
      "loss": 1.8108,
      "step": 158609
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.1877609491348267,
      "learning_rate": 0.0001326223684755392,
      "loss": 1.9992,
      "step": 158610
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.152534008026123,
      "learning_rate": 0.00013261897377003873,
      "loss": 2.0491,
      "step": 158611
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.3326681852340698,
      "learning_rate": 0.00013261557909565762,
      "loss": 2.0574,
      "step": 158612
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.2245409488677979,
      "learning_rate": 0.0001326121844523964,
      "loss": 1.9663,
      "step": 158613
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.163145661354065,
      "learning_rate": 0.0001326087898402558,
      "loss": 2.0014,
      "step": 158614
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.2857402563095093,
      "learning_rate": 0.0001326053952592364,
      "loss": 2.0775,
      "step": 158615
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.091845154762268,
      "learning_rate": 0.0001326020007093388,
      "loss": 2.0665,
      "step": 158616
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.2228976488113403,
      "learning_rate": 0.00013259860619056353,
      "loss": 2.0779,
      "step": 158617
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.3260232210159302,
      "learning_rate": 0.00013259521170291147,
      "loss": 1.9329,
      "step": 158618
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.1722922325134277,
      "learning_rate": 0.00013259181724638303,
      "loss": 2.0592,
      "step": 158619
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.0184009075164795,
      "learning_rate": 0.000132588422820979,
      "loss": 1.8857,
      "step": 158620
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.499698281288147,
      "learning_rate": 0.00013258502842669996,
      "loss": 1.9678,
      "step": 158621
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.202644944190979,
      "learning_rate": 0.0001325816340635465,
      "loss": 2.034,
      "step": 158622
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.1498757600784302,
      "learning_rate": 0.00013257823973151923,
      "loss": 2.0337,
      "step": 158623
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.2626503705978394,
      "learning_rate": 0.00013257484543061887,
      "loss": 1.9828,
      "step": 158624
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.4051138162612915,
      "learning_rate": 0.00013257145116084593,
      "loss": 2.1087,
      "step": 158625
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.3173785209655762,
      "learning_rate": 0.00013256805692220123,
      "loss": 2.1005,
      "step": 158626
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.2198500633239746,
      "learning_rate": 0.00013256466271468527,
      "loss": 2.1362,
      "step": 158627
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.1608020067214966,
      "learning_rate": 0.00013256126853829873,
      "loss": 2.0258,
      "step": 158628
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.1759135723114014,
      "learning_rate": 0.00013255787439304207,
      "loss": 2.1042,
      "step": 158629
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.137668490409851,
      "learning_rate": 0.00013255448027891614,
      "loss": 2.0458,
      "step": 158630
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.23060142993927,
      "learning_rate": 0.00013255108619592145,
      "loss": 2.0782,
      "step": 158631
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.1865874528884888,
      "learning_rate": 0.00013254769214405876,
      "loss": 2.0481,
      "step": 158632
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.6062918901443481,
      "learning_rate": 0.0001325442981233286,
      "loss": 1.8995,
      "step": 158633
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.1422754526138306,
      "learning_rate": 0.00013254090413373162,
      "loss": 1.9307,
      "step": 158634
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.2048472166061401,
      "learning_rate": 0.00013253751017526836,
      "loss": 2.0337,
      "step": 158635
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.2908759117126465,
      "learning_rate": 0.00013253411624793962,
      "loss": 1.9284,
      "step": 158636
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.3956100940704346,
      "learning_rate": 0.00013253072235174584,
      "loss": 2.0502,
      "step": 158637
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.2541687488555908,
      "learning_rate": 0.00013252732848668787,
      "loss": 2.0203,
      "step": 158638
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.3419266939163208,
      "learning_rate": 0.00013252393465276624,
      "loss": 2.1602,
      "step": 158639
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.187955617904663,
      "learning_rate": 0.0001325205408499816,
      "loss": 2.0373,
      "step": 158640
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.3191170692443848,
      "learning_rate": 0.0001325171470783344,
      "loss": 1.8841,
      "step": 158641
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.1114197969436646,
      "learning_rate": 0.00013251375333782556,
      "loss": 2.1193,
      "step": 158642
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.1856653690338135,
      "learning_rate": 0.00013251035962845548,
      "loss": 2.166,
      "step": 158643
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.2225821018218994,
      "learning_rate": 0.00013250696595022493,
      "loss": 2.0266,
      "step": 158644
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.2537519931793213,
      "learning_rate": 0.00013250357230313456,
      "loss": 1.6908,
      "step": 158645
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.1643168926239014,
      "learning_rate": 0.0001325001786871848,
      "loss": 2.1231,
      "step": 158646
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.1791585683822632,
      "learning_rate": 0.00013249678510237654,
      "loss": 2.0437,
      "step": 158647
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.1497896909713745,
      "learning_rate": 0.00013249339154871025,
      "loss": 1.7786,
      "step": 158648
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.205336093902588,
      "learning_rate": 0.00013248999802618653,
      "loss": 2.0144,
      "step": 158649
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.233270287513733,
      "learning_rate": 0.0001324866045348062,
      "loss": 2.2296,
      "step": 158650
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.2365670204162598,
      "learning_rate": 0.00013248321107456977,
      "loss": 2.134,
      "step": 158651
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.1836473941802979,
      "learning_rate": 0.00013247981764547774,
      "loss": 2.228,
      "step": 158652
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.1572306156158447,
      "learning_rate": 0.000132476424247531,
      "loss": 1.9299,
      "step": 158653
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.1743559837341309,
      "learning_rate": 0.00013247303088072994,
      "loss": 1.9157,
      "step": 158654
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.1624679565429688,
      "learning_rate": 0.0001324696375450754,
      "loss": 2.1321,
      "step": 158655
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.1251040697097778,
      "learning_rate": 0.00013246624424056793,
      "loss": 2.0902,
      "step": 158656
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.1343731880187988,
      "learning_rate": 0.00013246285096720818,
      "loss": 2.3522,
      "step": 158657
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.443182110786438,
      "learning_rate": 0.00013245945772499658,
      "loss": 2.0678,
      "step": 158658
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.1147469282150269,
      "learning_rate": 0.00013245606451393408,
      "loss": 1.9147,
      "step": 158659
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.172195553779602,
      "learning_rate": 0.00013245267133402103,
      "loss": 2.132,
      "step": 158660
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.1653087139129639,
      "learning_rate": 0.0001324492781852583,
      "loss": 2.0768,
      "step": 158661
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.3190397024154663,
      "learning_rate": 0.00013244588506764642,
      "loss": 2.3512,
      "step": 158662
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.1648088693618774,
      "learning_rate": 0.00013244249198118602,
      "loss": 2.018,
      "step": 158663
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.166239857673645,
      "learning_rate": 0.00013243909892587762,
      "loss": 2.1131,
      "step": 158664
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.131867527961731,
      "learning_rate": 0.00013243570590172203,
      "loss": 2.0522,
      "step": 158665
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.1221108436584473,
      "learning_rate": 0.00013243231290871975,
      "loss": 2.1439,
      "step": 158666
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.265036702156067,
      "learning_rate": 0.00013242891994687153,
      "loss": 2.138,
      "step": 158667
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.2580276727676392,
      "learning_rate": 0.00013242552701617783,
      "loss": 2.0224,
      "step": 158668
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.1311742067337036,
      "learning_rate": 0.00013242213411663963,
      "loss": 2.0088,
      "step": 158669
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.2582582235336304,
      "learning_rate": 0.0001324187412482571,
      "loss": 1.9822,
      "step": 158670
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.2093615531921387,
      "learning_rate": 0.00013241534841103116,
      "loss": 2.0599,
      "step": 158671
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.1571900844573975,
      "learning_rate": 0.0001324119556049623,
      "loss": 2.1392,
      "step": 158672
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.1570496559143066,
      "learning_rate": 0.00013240856283005132,
      "loss": 2.1687,
      "step": 158673
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.3082681894302368,
      "learning_rate": 0.00013240517008629863,
      "loss": 2.3015,
      "step": 158674
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.2946239709854126,
      "learning_rate": 0.0001324017773737051,
      "loss": 1.7883,
      "step": 158675
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.2177050113677979,
      "learning_rate": 0.00013239838469227124,
      "loss": 2.1281,
      "step": 158676
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.2144107818603516,
      "learning_rate": 0.0001323949920419977,
      "loss": 2.0661,
      "step": 158677
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.1621637344360352,
      "learning_rate": 0.00013239159942288496,
      "loss": 2.1366,
      "step": 158678
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.2678362131118774,
      "learning_rate": 0.00013238820683493394,
      "loss": 2.1199,
      "step": 158679
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.3169876337051392,
      "learning_rate": 0.00013238481427814492,
      "loss": 2.1857,
      "step": 158680
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.097593903541565,
      "learning_rate": 0.00013238142175251891,
      "loss": 1.9467,
      "step": 158681
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.2461323738098145,
      "learning_rate": 0.00013237802925805637,
      "loss": 2.0253,
      "step": 158682
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.1802425384521484,
      "learning_rate": 0.0001323746367947579,
      "loss": 2.1807,
      "step": 158683
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.3058360815048218,
      "learning_rate": 0.00013237124436262402,
      "loss": 2.0492,
      "step": 158684
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.2210456132888794,
      "learning_rate": 0.00013236785196165556,
      "loss": 1.8351,
      "step": 158685
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.3595013618469238,
      "learning_rate": 0.00013236445959185304,
      "loss": 1.9289,
      "step": 158686
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.202566385269165,
      "learning_rate": 0.00013236106725321724,
      "loss": 2.0628,
      "step": 158687
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.1616977453231812,
      "learning_rate": 0.00013235767494574864,
      "loss": 1.9838,
      "step": 158688
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.0114079713821411,
      "learning_rate": 0.00013235428266944793,
      "loss": 2.1329,
      "step": 158689
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.2039767503738403,
      "learning_rate": 0.00013235089042431563,
      "loss": 2.0242,
      "step": 158690
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.2083557844161987,
      "learning_rate": 0.00013234749821035256,
      "loss": 1.9011,
      "step": 158691
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.1437422037124634,
      "learning_rate": 0.00013234410602755913,
      "loss": 2.1744,
      "step": 158692
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.3544299602508545,
      "learning_rate": 0.00013234071387593623,
      "loss": 2.209,
      "step": 158693
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.231849193572998,
      "learning_rate": 0.00013233732175548435,
      "loss": 2.0517,
      "step": 158694
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.1237144470214844,
      "learning_rate": 0.00013233392966620412,
      "loss": 2.168,
      "step": 158695
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.1217378377914429,
      "learning_rate": 0.00013233053760809606,
      "loss": 1.8355,
      "step": 158696
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.1575325727462769,
      "learning_rate": 0.00013232714558116104,
      "loss": 2.1598,
      "step": 158697
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.204807996749878,
      "learning_rate": 0.00013232375358539946,
      "loss": 2.1272,
      "step": 158698
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.2046633958816528,
      "learning_rate": 0.00013232036162081216,
      "loss": 1.855,
      "step": 158699
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.1044378280639648,
      "learning_rate": 0.00013231696968739968,
      "loss": 2.1785,
      "step": 158700
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.1738816499710083,
      "learning_rate": 0.00013231357778516268,
      "loss": 1.9857,
      "step": 158701
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.182748556137085,
      "learning_rate": 0.0001323101859141016,
      "loss": 2.0105,
      "step": 158702
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.0632517337799072,
      "learning_rate": 0.0001323067940742173,
      "loss": 1.9908,
      "step": 158703
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.1131870746612549,
      "learning_rate": 0.00013230340226551027,
      "loss": 2.024,
      "step": 158704
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.154493808746338,
      "learning_rate": 0.0001323000104879813,
      "loss": 2.1124,
      "step": 158705
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.2151851654052734,
      "learning_rate": 0.0001322966187416309,
      "loss": 2.0597,
      "step": 158706
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.1754629611968994,
      "learning_rate": 0.00013229322702645977,
      "loss": 1.9307,
      "step": 158707
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.3498932123184204,
      "learning_rate": 0.00013228983534246833,
      "loss": 1.8958,
      "step": 158708
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.1225183010101318,
      "learning_rate": 0.00013228644368965754,
      "loss": 2.1222,
      "step": 158709
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.3407609462738037,
      "learning_rate": 0.00013228305206802773,
      "loss": 2.1606,
      "step": 158710
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.2757829427719116,
      "learning_rate": 0.00013227966047757976,
      "loss": 1.9927,
      "step": 158711
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.2213066816329956,
      "learning_rate": 0.0001322762689183142,
      "loss": 2.1636,
      "step": 158712
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.334420919418335,
      "learning_rate": 0.00013227287739023163,
      "loss": 2.2191,
      "step": 158713
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.3568710088729858,
      "learning_rate": 0.00013226948589333258,
      "loss": 2.1066,
      "step": 158714
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.0764338970184326,
      "learning_rate": 0.0001322660944276179,
      "loss": 1.8102,
      "step": 158715
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.496621012687683,
      "learning_rate": 0.00013226270299308807,
      "loss": 2.1305,
      "step": 158716
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.2114717960357666,
      "learning_rate": 0.0001322593115897438,
      "loss": 2.0346,
      "step": 158717
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.1304247379302979,
      "learning_rate": 0.00013225592021758574,
      "loss": 2.0619,
      "step": 158718
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.2179830074310303,
      "learning_rate": 0.00013225252887661445,
      "loss": 2.093,
      "step": 158719
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.2877141237258911,
      "learning_rate": 0.00013224913756683049,
      "loss": 2.2262,
      "step": 158720
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.2247778177261353,
      "learning_rate": 0.00013224574628823468,
      "loss": 2.0585,
      "step": 158721
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.1997134685516357,
      "learning_rate": 0.00013224235504082747,
      "loss": 2.0864,
      "step": 158722
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.2022476196289062,
      "learning_rate": 0.00013223896382460966,
      "loss": 1.8792,
      "step": 158723
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.501280665397644,
      "learning_rate": 0.00013223557263958177,
      "loss": 2.3008,
      "step": 158724
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.4195553064346313,
      "learning_rate": 0.0001322321814857445,
      "loss": 2.1172,
      "step": 158725
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.2161228656768799,
      "learning_rate": 0.00013222879036309826,
      "loss": 1.9767,
      "step": 158726
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.2028605937957764,
      "learning_rate": 0.00013222539927164402,
      "loss": 2.1048,
      "step": 158727
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.1546101570129395,
      "learning_rate": 0.00013222200821138213,
      "loss": 1.9931,
      "step": 158728
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.2785319089889526,
      "learning_rate": 0.00013221861718231342,
      "loss": 2.267,
      "step": 158729
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.2105395793914795,
      "learning_rate": 0.00013221522618443845,
      "loss": 1.9659,
      "step": 158730
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.3284612894058228,
      "learning_rate": 0.00013221183521775775,
      "loss": 2.1541,
      "step": 158731
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.2921260595321655,
      "learning_rate": 0.0001322084442822721,
      "loss": 1.8492,
      "step": 158732
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.1968082189559937,
      "learning_rate": 0.00013220505337798211,
      "loss": 2.0504,
      "step": 158733
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.7117677927017212,
      "learning_rate": 0.00013220166250488825,
      "loss": 2.0311,
      "step": 158734
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.416415810585022,
      "learning_rate": 0.00013219827166299138,
      "loss": 2.1139,
      "step": 158735
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.1582083702087402,
      "learning_rate": 0.000132194880852292,
      "loss": 2.096,
      "step": 158736
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.1909081935882568,
      "learning_rate": 0.00013219149007279065,
      "loss": 1.8005,
      "step": 158737
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.2102094888687134,
      "learning_rate": 0.00013218809932448818,
      "loss": 1.9998,
      "step": 158738
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.1398035287857056,
      "learning_rate": 0.00013218470860738512,
      "loss": 2.0828,
      "step": 158739
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.0793383121490479,
      "learning_rate": 0.00013218131792148196,
      "loss": 2.0278,
      "step": 158740
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.2101786136627197,
      "learning_rate": 0.0001321779272667796,
      "loss": 2.0639,
      "step": 158741
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.102568507194519,
      "learning_rate": 0.00013217453664327838,
      "loss": 2.0393,
      "step": 158742
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.1640985012054443,
      "learning_rate": 0.00013217114605097925,
      "loss": 2.1544,
      "step": 158743
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.1294525861740112,
      "learning_rate": 0.00013216775548988266,
      "loss": 1.8073,
      "step": 158744
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.1278412342071533,
      "learning_rate": 0.0001321643649599891,
      "loss": 1.9722,
      "step": 158745
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.1464587450027466,
      "learning_rate": 0.0001321609744612995,
      "loss": 2.0928,
      "step": 158746
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.2330026626586914,
      "learning_rate": 0.00013215758399381431,
      "loss": 1.9719,
      "step": 158747
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.1942836046218872,
      "learning_rate": 0.00013215419355753412,
      "loss": 1.6138,
      "step": 158748
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.2404900789260864,
      "learning_rate": 0.00013215080315245973,
      "loss": 2.1516,
      "step": 158749
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.2864036560058594,
      "learning_rate": 0.00013214741277859168,
      "loss": 2.0803,
      "step": 158750
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.131794810295105,
      "learning_rate": 0.00013214402243593047,
      "loss": 2.2412,
      "step": 158751
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.1319538354873657,
      "learning_rate": 0.00013214063212447694,
      "loss": 1.9467,
      "step": 158752
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.2517035007476807,
      "learning_rate": 0.00013213724184423157,
      "loss": 1.8606,
      "step": 158753
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.1074568033218384,
      "learning_rate": 0.00013213385159519518,
      "loss": 2.2816,
      "step": 158754
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.0825457572937012,
      "learning_rate": 0.00013213046137736825,
      "loss": 1.8433,
      "step": 158755
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.252012848854065,
      "learning_rate": 0.0001321270711907514,
      "loss": 2.0201,
      "step": 158756
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.140404224395752,
      "learning_rate": 0.00013212368103534522,
      "loss": 1.9992,
      "step": 158757
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.242641806602478,
      "learning_rate": 0.00013212029091115053,
      "loss": 2.1286,
      "step": 158758
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.267412781715393,
      "learning_rate": 0.00013211690081816775,
      "loss": 1.6335,
      "step": 158759
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.118792176246643,
      "learning_rate": 0.00013211351075639768,
      "loss": 2.0636,
      "step": 158760
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.0578769445419312,
      "learning_rate": 0.00013211012072584093,
      "loss": 1.9311,
      "step": 158761
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.117245078086853,
      "learning_rate": 0.00013210673072649805,
      "loss": 2.0847,
      "step": 158762
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.2997359037399292,
      "learning_rate": 0.00013210334075836958,
      "loss": 2.0376,
      "step": 158763
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.1513218879699707,
      "learning_rate": 0.00013209995082145637,
      "loss": 2.0827,
      "step": 158764
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.0814698934555054,
      "learning_rate": 0.00013209656091575883,
      "loss": 2.1977,
      "step": 158765
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.2849178314208984,
      "learning_rate": 0.00013209317104127786,
      "loss": 2.0538,
      "step": 158766
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.0850443840026855,
      "learning_rate": 0.0001320897811980139,
      "loss": 2.147,
      "step": 158767
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.0779762268066406,
      "learning_rate": 0.00013208639138596764,
      "loss": 1.802,
      "step": 158768
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.0718774795532227,
      "learning_rate": 0.00013208300160513957,
      "loss": 2.1333,
      "step": 158769
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.0957571268081665,
      "learning_rate": 0.00013207961185553058,
      "loss": 1.9369,
      "step": 158770
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.1831364631652832,
      "learning_rate": 0.00013207622213714103,
      "loss": 2.0188,
      "step": 158771
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.1809037923812866,
      "learning_rate": 0.00013207283244997175,
      "loss": 2.105,
      "step": 158772
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.1812347173690796,
      "learning_rate": 0.00013206944279402338,
      "loss": 1.9895,
      "step": 158773
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.3948071002960205,
      "learning_rate": 0.0001320660531692964,
      "loss": 2.1687,
      "step": 158774
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.3293813467025757,
      "learning_rate": 0.00013206266357579144,
      "loss": 1.9534,
      "step": 158775
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.1955738067626953,
      "learning_rate": 0.0001320592740135093,
      "loss": 2.2043,
      "step": 158776
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.0850392580032349,
      "learning_rate": 0.00013205588448245037,
      "loss": 1.9043,
      "step": 158777
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.1570513248443604,
      "learning_rate": 0.00013205249498261557,
      "loss": 1.8746,
      "step": 158778
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.1739150285720825,
      "learning_rate": 0.0001320491055140054,
      "loss": 1.9629,
      "step": 158779
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.1691913604736328,
      "learning_rate": 0.00013204571607662043,
      "loss": 1.995,
      "step": 158780
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.157710075378418,
      "learning_rate": 0.00013204232667046124,
      "loss": 2.21,
      "step": 158781
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.2944904565811157,
      "learning_rate": 0.00013203893729552864,
      "loss": 1.9335,
      "step": 158782
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.2862629890441895,
      "learning_rate": 0.0001320355479518231,
      "loss": 2.228,
      "step": 158783
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.0851709842681885,
      "learning_rate": 0.00013203215863934542,
      "loss": 1.9279,
      "step": 158784
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.0994263887405396,
      "learning_rate": 0.00013202876935809613,
      "loss": 1.9259,
      "step": 158785
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.1529446840286255,
      "learning_rate": 0.00013202538010807584,
      "loss": 1.9617,
      "step": 158786
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.1169849634170532,
      "learning_rate": 0.0001320219908892851,
      "loss": 2.1338,
      "step": 158787
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.1729682683944702,
      "learning_rate": 0.00013201860170172476,
      "loss": 1.9194,
      "step": 158788
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.2352044582366943,
      "learning_rate": 0.0001320152125453952,
      "loss": 2.0285,
      "step": 158789
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.2205020189285278,
      "learning_rate": 0.00013201182342029735,
      "loss": 2.3222,
      "step": 158790
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.1696062088012695,
      "learning_rate": 0.00013200843432643165,
      "loss": 2.0252,
      "step": 158791
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.5230422019958496,
      "learning_rate": 0.00013200504526379874,
      "loss": 2.0428,
      "step": 158792
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.0864641666412354,
      "learning_rate": 0.00013200165623239918,
      "loss": 2.0202,
      "step": 158793
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.276969313621521,
      "learning_rate": 0.00013199826723223374,
      "loss": 2.01,
      "step": 158794
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.1420358419418335,
      "learning_rate": 0.00013199487826330293,
      "loss": 1.9441,
      "step": 158795
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.1635661125183105,
      "learning_rate": 0.00013199148932560754,
      "loss": 2.0324,
      "step": 158796
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.2274158000946045,
      "learning_rate": 0.00013198810041914808,
      "loss": 2.1701,
      "step": 158797
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.1878799200057983,
      "learning_rate": 0.00013198471154392524,
      "loss": 1.7611,
      "step": 158798
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.173872709274292,
      "learning_rate": 0.00013198132269993948,
      "loss": 1.8822,
      "step": 158799
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.2295781373977661,
      "learning_rate": 0.0001319779338871917,
      "loss": 2.3292,
      "step": 158800
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.214312195777893,
      "learning_rate": 0.00013197454510568225,
      "loss": 1.9263,
      "step": 158801
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.23612380027771,
      "learning_rate": 0.00013197115635541203,
      "loss": 2.1608,
      "step": 158802
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.1750473976135254,
      "learning_rate": 0.00013196776763638155,
      "loss": 2.0461,
      "step": 158803
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.156233310699463,
      "learning_rate": 0.0001319643789485914,
      "loss": 1.9292,
      "step": 158804
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.8048205375671387,
      "learning_rate": 0.00013196099029204217,
      "loss": 1.953,
      "step": 158805
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.2976871728897095,
      "learning_rate": 0.0001319576016667347,
      "loss": 2.1751,
      "step": 158806
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.1946125030517578,
      "learning_rate": 0.0001319542130726693,
      "loss": 1.9928,
      "step": 158807
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.1966891288757324,
      "learning_rate": 0.00013195082450984694,
      "loss": 1.9453,
      "step": 158808
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.2337852716445923,
      "learning_rate": 0.000131947435978268,
      "loss": 2.2193,
      "step": 158809
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.3325233459472656,
      "learning_rate": 0.0001319440474779334,
      "loss": 2.0366,
      "step": 158810
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.1873310804367065,
      "learning_rate": 0.00013194065900884333,
      "loss": 1.9637,
      "step": 158811
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.2549067735671997,
      "learning_rate": 0.0001319372705709988,
      "loss": 2.1147,
      "step": 158812
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.225154995918274,
      "learning_rate": 0.0001319338821644002,
      "loss": 2.2114,
      "step": 158813
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.1657991409301758,
      "learning_rate": 0.00013193049378904835,
      "loss": 2.3294,
      "step": 158814
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.2434840202331543,
      "learning_rate": 0.00013192710544494374,
      "loss": 2.1556,
      "step": 158815
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.362196922302246,
      "learning_rate": 0.0001319237171320871,
      "loss": 2.0599,
      "step": 158816
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.221717357635498,
      "learning_rate": 0.00013192032885047902,
      "loss": 2.1454,
      "step": 158817
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.0847593545913696,
      "learning_rate": 0.0001319169406001202,
      "loss": 2.1,
      "step": 158818
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.2387633323669434,
      "learning_rate": 0.000131913552381011,
      "loss": 1.7953,
      "step": 158819
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.2945195436477661,
      "learning_rate": 0.00013191016419315235,
      "loss": 2.2878,
      "step": 158820
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.289628505706787,
      "learning_rate": 0.00013190677603654472,
      "loss": 2.1485,
      "step": 158821
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.1062935590744019,
      "learning_rate": 0.00013190338791118888,
      "loss": 2.0995,
      "step": 158822
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.2854567766189575,
      "learning_rate": 0.00013189999981708537,
      "loss": 1.9646,
      "step": 158823
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.1548370122909546,
      "learning_rate": 0.0001318966117542348,
      "loss": 1.9863,
      "step": 158824
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.2506626844406128,
      "learning_rate": 0.00013189322372263773,
      "loss": 2.1371,
      "step": 158825
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.202385663986206,
      "learning_rate": 0.00013188983572229502,
      "loss": 2.0909,
      "step": 158826
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.185189962387085,
      "learning_rate": 0.000131886447753207,
      "loss": 1.9735,
      "step": 158827
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.138010025024414,
      "learning_rate": 0.00013188305981537462,
      "loss": 2.1452,
      "step": 158828
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.2188845872879028,
      "learning_rate": 0.00013187967190879835,
      "loss": 2.0267,
      "step": 158829
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.1561763286590576,
      "learning_rate": 0.00013187628403347866,
      "loss": 2.0311,
      "step": 158830
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.1579318046569824,
      "learning_rate": 0.00013187289618941651,
      "loss": 2.0021,
      "step": 158831
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.2812364101409912,
      "learning_rate": 0.00013186950837661235,
      "loss": 1.9129,
      "step": 158832
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.25013267993927,
      "learning_rate": 0.0001318661205950667,
      "loss": 2.1389,
      "step": 158833
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.2311747074127197,
      "learning_rate": 0.00013186273284478042,
      "loss": 1.9873,
      "step": 158834
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.5936859846115112,
      "learning_rate": 0.00013185934512575405,
      "loss": 1.8335,
      "step": 158835
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.3487523794174194,
      "learning_rate": 0.00013185595743798808,
      "loss": 2.1464,
      "step": 158836
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.1888394355773926,
      "learning_rate": 0.00013185256978148337,
      "loss": 2.0242,
      "step": 158837
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.1969627141952515,
      "learning_rate": 0.00013184918215624044,
      "loss": 1.9738,
      "step": 158838
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.1003005504608154,
      "learning_rate": 0.0001318457945622598,
      "loss": 1.8887,
      "step": 158839
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.2942806482315063,
      "learning_rate": 0.00013184240699954235,
      "loss": 1.9992,
      "step": 158840
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.298396110534668,
      "learning_rate": 0.00013183901946808851,
      "loss": 2.0209,
      "step": 158841
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.124408483505249,
      "learning_rate": 0.0001318356319678989,
      "loss": 1.7837,
      "step": 158842
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.128885269165039,
      "learning_rate": 0.00013183224449897434,
      "loss": 1.9576,
      "step": 158843
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.1381551027297974,
      "learning_rate": 0.00013182885706131524,
      "loss": 2.258,
      "step": 158844
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.2394945621490479,
      "learning_rate": 0.00013182546965492242,
      "loss": 2.2427,
      "step": 158845
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.2708927392959595,
      "learning_rate": 0.0001318220822797964,
      "loss": 2.12,
      "step": 158846
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.2154616117477417,
      "learning_rate": 0.00013181869493593783,
      "loss": 2.1979,
      "step": 158847
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.1890900135040283,
      "learning_rate": 0.00013181530762334726,
      "loss": 2.0714,
      "step": 158848
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.2145955562591553,
      "learning_rate": 0.00013181192034202548,
      "loss": 2.0366,
      "step": 158849
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.3141931295394897,
      "learning_rate": 0.00013180853309197296,
      "loss": 2.0482,
      "step": 158850
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.1089541912078857,
      "learning_rate": 0.00013180514587319048,
      "loss": 1.8856,
      "step": 158851
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.1682745218276978,
      "learning_rate": 0.00013180175868567865,
      "loss": 2.2064,
      "step": 158852
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.3494718074798584,
      "learning_rate": 0.000131798371529438,
      "loss": 2.0207,
      "step": 158853
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.1297061443328857,
      "learning_rate": 0.00013179498440446912,
      "loss": 1.7736,
      "step": 158854
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.1404163837432861,
      "learning_rate": 0.00013179159731077284,
      "loss": 2.0734,
      "step": 158855
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.2111501693725586,
      "learning_rate": 0.00013178821024834955,
      "loss": 1.981,
      "step": 158856
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.1400493383407593,
      "learning_rate": 0.00013178482321720013,
      "loss": 2.0618,
      "step": 158857
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.145569086074829,
      "learning_rate": 0.0001317814362173251,
      "loss": 1.9349,
      "step": 158858
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.2040998935699463,
      "learning_rate": 0.00013177804924872504,
      "loss": 1.9281,
      "step": 158859
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.1245590448379517,
      "learning_rate": 0.00013177466231140052,
      "loss": 1.9933,
      "step": 158860
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.1207541227340698,
      "learning_rate": 0.0001317712754053524,
      "loss": 2.0449,
      "step": 158861
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.5908539295196533,
      "learning_rate": 0.00013176788853058106,
      "loss": 2.1174,
      "step": 158862
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.2821061611175537,
      "learning_rate": 0.00013176450168708733,
      "loss": 2.1657,
      "step": 158863
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.2214041948318481,
      "learning_rate": 0.00013176111487487176,
      "loss": 2.0184,
      "step": 158864
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.295301079750061,
      "learning_rate": 0.000131757728093935,
      "loss": 2.1093,
      "step": 158865
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.5966789722442627,
      "learning_rate": 0.0001317543413442775,
      "loss": 2.0357,
      "step": 158866
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.1485297679901123,
      "learning_rate": 0.0001317509546259002,
      "loss": 1.9412,
      "step": 158867
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.1102261543273926,
      "learning_rate": 0.00013174756793880345,
      "loss": 2.0117,
      "step": 158868
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.073770523071289,
      "learning_rate": 0.00013174418128298807,
      "loss": 1.9866,
      "step": 158869
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.2728476524353027,
      "learning_rate": 0.00013174079465845466,
      "loss": 2.1638,
      "step": 158870
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.2131741046905518,
      "learning_rate": 0.00013173740806520385,
      "loss": 2.112,
      "step": 158871
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.1402034759521484,
      "learning_rate": 0.00013173402150323606,
      "loss": 1.8572,
      "step": 158872
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.221798062324524,
      "learning_rate": 0.00013173063497255223,
      "loss": 1.9787,
      "step": 158873
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.2475333213806152,
      "learning_rate": 0.00013172724847315275,
      "loss": 1.962,
      "step": 158874
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.106471300125122,
      "learning_rate": 0.00013172386200503843,
      "loss": 2.2219,
      "step": 158875
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.1485294103622437,
      "learning_rate": 0.00013172047556820975,
      "loss": 2.1096,
      "step": 158876
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.1623131036758423,
      "learning_rate": 0.0001317170891626676,
      "loss": 1.9277,
      "step": 158877
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.1581151485443115,
      "learning_rate": 0.0001317137027884122,
      "loss": 1.9602,
      "step": 158878
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.245703101158142,
      "learning_rate": 0.00013171031644544453,
      "loss": 2.2668,
      "step": 158879
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.2164199352264404,
      "learning_rate": 0.00013170693013376495,
      "loss": 2.2866,
      "step": 158880
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.2530006170272827,
      "learning_rate": 0.00013170354385337436,
      "loss": 2.2623,
      "step": 158881
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.1955360174179077,
      "learning_rate": 0.00013170015760427314,
      "loss": 2.1239,
      "step": 158882
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.112205147743225,
      "learning_rate": 0.00013169677138646224,
      "loss": 2.0655,
      "step": 158883
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.1961281299591064,
      "learning_rate": 0.0001316933851999419,
      "loss": 1.9038,
      "step": 158884
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.1375370025634766,
      "learning_rate": 0.00013168999904471305,
      "loss": 2.1083,
      "step": 158885
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.1916862726211548,
      "learning_rate": 0.0001316866129207761,
      "loss": 2.0331,
      "step": 158886
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.1335068941116333,
      "learning_rate": 0.00013168322682813186,
      "loss": 1.8052,
      "step": 158887
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.1768957376480103,
      "learning_rate": 0.0001316798407667808,
      "loss": 1.9414,
      "step": 158888
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.2003793716430664,
      "learning_rate": 0.0001316764547367238,
      "loss": 2.1009,
      "step": 158889
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.1478297710418701,
      "learning_rate": 0.0001316730687379612,
      "loss": 2.0104,
      "step": 158890
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.1706697940826416,
      "learning_rate": 0.00013166968277049382,
      "loss": 2.1657,
      "step": 158891
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.1571764945983887,
      "learning_rate": 0.0001316662968343221,
      "loss": 2.0459,
      "step": 158892
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.2405866384506226,
      "learning_rate": 0.0001316629109294469,
      "loss": 1.8687,
      "step": 158893
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.2495898008346558,
      "learning_rate": 0.00013165952505586866,
      "loss": 2.0349,
      "step": 158894
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.240020513534546,
      "learning_rate": 0.00013165613921358826,
      "loss": 1.8578,
      "step": 158895
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.2491182088851929,
      "learning_rate": 0.000131652753402606,
      "loss": 2.1378,
      "step": 158896
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.1286234855651855,
      "learning_rate": 0.00013164936762292278,
      "loss": 1.9882,
      "step": 158897
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.1592868566513062,
      "learning_rate": 0.00013164598187453895,
      "loss": 1.9192,
      "step": 158898
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.0670326948165894,
      "learning_rate": 0.00013164259615745547,
      "loss": 1.7766,
      "step": 158899
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.172562837600708,
      "learning_rate": 0.0001316392104716727,
      "loss": 1.9059,
      "step": 158900
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.1494688987731934,
      "learning_rate": 0.00013163582481719158,
      "loss": 2.0261,
      "step": 158901
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.1211916208267212,
      "learning_rate": 0.0001316324391940123,
      "loss": 1.9519,
      "step": 158902
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.197849988937378,
      "learning_rate": 0.0001316290536021359,
      "loss": 2.0348,
      "step": 158903
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.2616119384765625,
      "learning_rate": 0.00013162566804156268,
      "loss": 2.0913,
      "step": 158904
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.2907664775848389,
      "learning_rate": 0.00013162228251229356,
      "loss": 2.152,
      "step": 158905
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.139399766921997,
      "learning_rate": 0.00013161889701432888,
      "loss": 1.9572,
      "step": 158906
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.1304717063903809,
      "learning_rate": 0.00013161551154766956,
      "loss": 2.0616,
      "step": 158907
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.1565651893615723,
      "learning_rate": 0.00013161212611231612,
      "loss": 2.0333,
      "step": 158908
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.2536284923553467,
      "learning_rate": 0.00013160874070826915,
      "loss": 2.1185,
      "step": 158909
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.1454558372497559,
      "learning_rate": 0.00013160535533552917,
      "loss": 2.0611,
      "step": 158910
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.1791253089904785,
      "learning_rate": 0.00013160196999409708,
      "loss": 2.3117,
      "step": 158911
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.210989236831665,
      "learning_rate": 0.00013159858468397322,
      "loss": 2.0307,
      "step": 158912
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.2070810794830322,
      "learning_rate": 0.00013159519940515846,
      "loss": 2.2311,
      "step": 158913
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.1799548864364624,
      "learning_rate": 0.0001315918141576534,
      "loss": 2.0376,
      "step": 158914
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.322217345237732,
      "learning_rate": 0.00013158842894145842,
      "loss": 1.8992,
      "step": 158915
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.239663004875183,
      "learning_rate": 0.00013158504375657444,
      "loss": 2.0158,
      "step": 158916
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.2038376331329346,
      "learning_rate": 0.00013158165860300202,
      "loss": 2.1921,
      "step": 158917
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.0982027053833008,
      "learning_rate": 0.0001315782734807416,
      "loss": 1.9768,
      "step": 158918
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.2524495124816895,
      "learning_rate": 0.0001315748883897941,
      "loss": 2.0671,
      "step": 158919
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.2259478569030762,
      "learning_rate": 0.00013157150333016,
      "loss": 2.0572,
      "step": 158920
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.1763982772827148,
      "learning_rate": 0.00013156811830183983,
      "loss": 2.0859,
      "step": 158921
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.159549593925476,
      "learning_rate": 0.0001315647333048344,
      "loss": 1.9788,
      "step": 158922
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.2774405479431152,
      "learning_rate": 0.00013156134833914432,
      "loss": 2.2583,
      "step": 158923
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.197609782218933,
      "learning_rate": 0.00013155796340477002,
      "loss": 1.9817,
      "step": 158924
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.1813461780548096,
      "learning_rate": 0.0001315545785017124,
      "loss": 2.0431,
      "step": 158925
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.2205852270126343,
      "learning_rate": 0.00013155119362997196,
      "loss": 2.2146,
      "step": 158926
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.2215497493743896,
      "learning_rate": 0.0001315478087895492,
      "loss": 2.0283,
      "step": 158927
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.2175605297088623,
      "learning_rate": 0.000131544423980445,
      "loss": 2.1443,
      "step": 158928
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.2059282064437866,
      "learning_rate": 0.00013154103920265977,
      "loss": 2.1278,
      "step": 158929
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.2536414861679077,
      "learning_rate": 0.00013153765445619438,
      "loss": 1.8963,
      "step": 158930
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.2120206356048584,
      "learning_rate": 0.00013153426974104929,
      "loss": 2.0508,
      "step": 158931
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.3346763849258423,
      "learning_rate": 0.00013153088505722516,
      "loss": 2.29,
      "step": 158932
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.1442310810089111,
      "learning_rate": 0.0001315275004047225,
      "loss": 2.1078,
      "step": 158933
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.167891502380371,
      "learning_rate": 0.00013152411578354217,
      "loss": 2.0013,
      "step": 158934
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.1960278749465942,
      "learning_rate": 0.00013152073119368458,
      "loss": 2.1059,
      "step": 158935
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.2212774753570557,
      "learning_rate": 0.0001315173466351506,
      "loss": 2.0911,
      "step": 158936
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.240684986114502,
      "learning_rate": 0.00013151396210794072,
      "loss": 1.7893,
      "step": 158937
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.2349283695220947,
      "learning_rate": 0.00013151057761205554,
      "loss": 2.1089,
      "step": 158938
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.248291254043579,
      "learning_rate": 0.00013150719314749564,
      "loss": 2.1212,
      "step": 158939
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.1765707731246948,
      "learning_rate": 0.00013150380871426184,
      "loss": 2.2964,
      "step": 158940
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.2006982564926147,
      "learning_rate": 0.0001315004243123546,
      "loss": 1.9155,
      "step": 158941
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.458398699760437,
      "learning_rate": 0.00013149703994177464,
      "loss": 2.0073,
      "step": 158942
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.2015103101730347,
      "learning_rate": 0.0001314936556025225,
      "loss": 1.8927,
      "step": 158943
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.1149837970733643,
      "learning_rate": 0.00013149027129459904,
      "loss": 2.0312,
      "step": 158944
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.4493945837020874,
      "learning_rate": 0.00013148688701800456,
      "loss": 2.3342,
      "step": 158945
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.245242953300476,
      "learning_rate": 0.00013148350277273992,
      "loss": 1.9326,
      "step": 158946
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.2388182878494263,
      "learning_rate": 0.0001314801185588056,
      "loss": 1.9588,
      "step": 158947
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.2145341634750366,
      "learning_rate": 0.00013147673437620237,
      "loss": 2.0077,
      "step": 158948
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.1710400581359863,
      "learning_rate": 0.00013147335022493074,
      "loss": 2.2048,
      "step": 158949
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.1358767747879028,
      "learning_rate": 0.0001314699661049916,
      "loss": 2.0448,
      "step": 158950
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.1449228525161743,
      "learning_rate": 0.00013146658201638512,
      "loss": 2.1676,
      "step": 158951
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.3136852979660034,
      "learning_rate": 0.00013146319795911233,
      "loss": 1.9867,
      "step": 158952
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.213556170463562,
      "learning_rate": 0.00013145981393317358,
      "loss": 2.0973,
      "step": 158953
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.263257384300232,
      "learning_rate": 0.00013145642993856975,
      "loss": 2.0248,
      "step": 158954
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.3291447162628174,
      "learning_rate": 0.00013145304597530125,
      "loss": 2.1426,
      "step": 158955
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.1737927198410034,
      "learning_rate": 0.00013144966204336905,
      "loss": 1.9575,
      "step": 158956
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.470578670501709,
      "learning_rate": 0.0001314462781427733,
      "loss": 2.2587,
      "step": 158957
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.3124052286148071,
      "learning_rate": 0.00013144289427351495,
      "loss": 1.9368,
      "step": 158958
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.6037702560424805,
      "learning_rate": 0.00013143951043559444,
      "loss": 2.0524,
      "step": 158959
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.2605736255645752,
      "learning_rate": 0.00013143612662901266,
      "loss": 2.2544,
      "step": 158960
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.257636308670044,
      "learning_rate": 0.00013143274285376994,
      "loss": 1.9436,
      "step": 158961
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.5922770500183105,
      "learning_rate": 0.0001314293591098673,
      "loss": 1.7293,
      "step": 158962
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.3312469720840454,
      "learning_rate": 0.0001314259753973049,
      "loss": 1.9252,
      "step": 158963
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.170683741569519,
      "learning_rate": 0.00013142259171608367,
      "loss": 1.8052,
      "step": 158964
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.1077710390090942,
      "learning_rate": 0.0001314192080662041,
      "loss": 2.048,
      "step": 158965
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.2479195594787598,
      "learning_rate": 0.00013141582444766697,
      "loss": 2.0361,
      "step": 158966
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.1051898002624512,
      "learning_rate": 0.00013141244086047272,
      "loss": 1.9124,
      "step": 158967
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.1522231101989746,
      "learning_rate": 0.00013140905730462228,
      "loss": 1.9809,
      "step": 158968
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.1992961168289185,
      "learning_rate": 0.00013140567378011587,
      "loss": 2.0162,
      "step": 158969
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.3571504354476929,
      "learning_rate": 0.0001314022902869544,
      "loss": 2.0376,
      "step": 158970
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.125969409942627,
      "learning_rate": 0.0001313989068251384,
      "loss": 1.971,
      "step": 158971
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.3453599214553833,
      "learning_rate": 0.0001313955233946686,
      "loss": 1.9142,
      "step": 158972
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.0964531898498535,
      "learning_rate": 0.0001313921399955454,
      "loss": 1.8972,
      "step": 158973
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.1068023443222046,
      "learning_rate": 0.00013138875662776987,
      "loss": 1.9955,
      "step": 158974
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.3332359790802002,
      "learning_rate": 0.00013138537329134207,
      "loss": 2.1779,
      "step": 158975
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.2224427461624146,
      "learning_rate": 0.00013138198998626307,
      "loss": 2.1592,
      "step": 158976
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.2080434560775757,
      "learning_rate": 0.00013137860671253321,
      "loss": 2.1638,
      "step": 158977
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.1467458009719849,
      "learning_rate": 0.00013137522347015336,
      "loss": 2.0104,
      "step": 158978
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.1502138376235962,
      "learning_rate": 0.00013137184025912392,
      "loss": 1.8893,
      "step": 158979
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.1761585474014282,
      "learning_rate": 0.0001313684570794459,
      "loss": 1.9018,
      "step": 158980
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.1905657052993774,
      "learning_rate": 0.0001313650739311194,
      "loss": 2.1355,
      "step": 158981
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.2090871334075928,
      "learning_rate": 0.00013136169081414545,
      "loss": 1.9711,
      "step": 158982
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.2766976356506348,
      "learning_rate": 0.00013135830772852442,
      "loss": 2.1953,
      "step": 158983
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.2688308954238892,
      "learning_rate": 0.0001313549246742572,
      "loss": 2.0156,
      "step": 158984
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.2217003107070923,
      "learning_rate": 0.00013135154165134412,
      "loss": 2.0011,
      "step": 158985
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.1350113153457642,
      "learning_rate": 0.00013134815865978623,
      "loss": 2.1136,
      "step": 158986
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.140234112739563,
      "learning_rate": 0.0001313447756995837,
      "loss": 1.9521,
      "step": 158987
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.0736680030822754,
      "learning_rate": 0.00013134139277073744,
      "loss": 1.7803,
      "step": 158988
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.1596041917800903,
      "learning_rate": 0.0001313380098732479,
      "loss": 1.8437,
      "step": 158989
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.195334792137146,
      "learning_rate": 0.00013133462700711595,
      "loss": 1.9654,
      "step": 158990
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.2433414459228516,
      "learning_rate": 0.0001313312441723419,
      "loss": 2.2511,
      "step": 158991
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.1829720735549927,
      "learning_rate": 0.0001313278613689267,
      "loss": 2.0148,
      "step": 158992
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.1120071411132812,
      "learning_rate": 0.00013132447859687084,
      "loss": 1.7782,
      "step": 158993
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.1087689399719238,
      "learning_rate": 0.0001313210958561749,
      "loss": 1.8715,
      "step": 158994
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.2552599906921387,
      "learning_rate": 0.0001313177131468395,
      "loss": 1.9062,
      "step": 158995
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.2779130935668945,
      "learning_rate": 0.0001313143304688654,
      "loss": 1.8555,
      "step": 158996
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.2176202535629272,
      "learning_rate": 0.00013131094782225307,
      "loss": 2.0437,
      "step": 158997
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.220507264137268,
      "learning_rate": 0.0001313075652070033,
      "loss": 2.266,
      "step": 158998
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.1505242586135864,
      "learning_rate": 0.00013130418262311664,
      "loss": 2.0264,
      "step": 158999
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.330235481262207,
      "learning_rate": 0.0001313008000705936,
      "loss": 2.2319,
      "step": 159000
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.1625643968582153,
      "learning_rate": 0.00013129741754943505,
      "loss": 2.0713,
      "step": 159001
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.3485888242721558,
      "learning_rate": 0.0001312940350596415,
      "loss": 2.0089,
      "step": 159002
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.2054282426834106,
      "learning_rate": 0.00013129065260121346,
      "loss": 1.9628,
      "step": 159003
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.2446634769439697,
      "learning_rate": 0.00013128727017415177,
      "loss": 2.0648,
      "step": 159004
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.1566797494888306,
      "learning_rate": 0.00013128388777845697,
      "loss": 2.1503,
      "step": 159005
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.2911419868469238,
      "learning_rate": 0.00013128050541412958,
      "loss": 2.201,
      "step": 159006
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.301757574081421,
      "learning_rate": 0.00013127712308117043,
      "loss": 2.207,
      "step": 159007
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.3268723487854004,
      "learning_rate": 0.00013127374077958004,
      "loss": 1.981,
      "step": 159008
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.177397608757019,
      "learning_rate": 0.00013127035850935895,
      "loss": 2.1349,
      "step": 159009
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.208677887916565,
      "learning_rate": 0.00013126697627050796,
      "loss": 2.0745,
      "step": 159010
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.1976933479309082,
      "learning_rate": 0.00013126359406302767,
      "loss": 2.0462,
      "step": 159011
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.1934080123901367,
      "learning_rate": 0.00013126021188691853,
      "loss": 2.0029,
      "step": 159012
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.1745482683181763,
      "learning_rate": 0.00013125682974218144,
      "loss": 1.904,
      "step": 159013
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.3200733661651611,
      "learning_rate": 0.00013125344762881673,
      "loss": 1.9282,
      "step": 159014
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.1580474376678467,
      "learning_rate": 0.0001312500655468253,
      "loss": 1.9925,
      "step": 159015
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.1614351272583008,
      "learning_rate": 0.00013124668349620772,
      "loss": 2.0086,
      "step": 159016
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.1607849597930908,
      "learning_rate": 0.00013124330147696456,
      "loss": 1.8329,
      "step": 159017
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.1949689388275146,
      "learning_rate": 0.0001312399194890963,
      "loss": 2.1262,
      "step": 159018
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.1778674125671387,
      "learning_rate": 0.0001312365375326039,
      "loss": 2.1371,
      "step": 159019
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.1076141595840454,
      "learning_rate": 0.00013123315560748764,
      "loss": 2.1251,
      "step": 159020
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.2115955352783203,
      "learning_rate": 0.00013122977371374847,
      "loss": 2.2836,
      "step": 159021
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.252925157546997,
      "learning_rate": 0.00013122639185138686,
      "loss": 2.3386,
      "step": 159022
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.1294910907745361,
      "learning_rate": 0.00013122301002040345,
      "loss": 2.1214,
      "step": 159023
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.1659064292907715,
      "learning_rate": 0.00013121962822079878,
      "loss": 2.0784,
      "step": 159024
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.2592614889144897,
      "learning_rate": 0.00013121624645257363,
      "loss": 2.0372,
      "step": 159025
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.2930593490600586,
      "learning_rate": 0.00013121286471572845,
      "loss": 2.2195,
      "step": 159026
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.383791446685791,
      "learning_rate": 0.00013120948301026413,
      "loss": 2.1733,
      "step": 159027
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.1787139177322388,
      "learning_rate": 0.00013120610133618104,
      "loss": 2.0513,
      "step": 159028
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.2816667556762695,
      "learning_rate": 0.00013120271969348015,
      "loss": 2.1282,
      "step": 159029
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.2767655849456787,
      "learning_rate": 0.00013119933808216163,
      "loss": 1.9372,
      "step": 159030
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.3068238496780396,
      "learning_rate": 0.00013119595650222646,
      "loss": 2.1845,
      "step": 159031
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.2645556926727295,
      "learning_rate": 0.000131192574953675,
      "loss": 1.94,
      "step": 159032
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.3187164068222046,
      "learning_rate": 0.0001311891934365082,
      "loss": 1.8796,
      "step": 159033
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.1223872900009155,
      "learning_rate": 0.00013118581195072635,
      "loss": 2.0453,
      "step": 159034
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.1577980518341064,
      "learning_rate": 0.00013118243049633048,
      "loss": 1.8455,
      "step": 159035
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.2051700353622437,
      "learning_rate": 0.00013117904907332075,
      "loss": 2.2487,
      "step": 159036
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.5271623134613037,
      "learning_rate": 0.00013117566768169816,
      "loss": 1.9226,
      "step": 159037
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.2168294191360474,
      "learning_rate": 0.00013117228632146308,
      "loss": 2.1273,
      "step": 159038
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.2195090055465698,
      "learning_rate": 0.00013116890499261638,
      "loss": 2.0768,
      "step": 159039
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.235708475112915,
      "learning_rate": 0.00013116552369515843,
      "loss": 2.0474,
      "step": 159040
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.4274101257324219,
      "learning_rate": 0.00013116214242909024,
      "loss": 1.9929,
      "step": 159041
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.3193310499191284,
      "learning_rate": 0.00013115876119441198,
      "loss": 2.0591,
      "step": 159042
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.2263110876083374,
      "learning_rate": 0.0001311553799911246,
      "loss": 2.1909,
      "step": 159043
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.0663261413574219,
      "learning_rate": 0.0001311519988192285,
      "loss": 2.1094,
      "step": 159044
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.2979767322540283,
      "learning_rate": 0.00013114861767872453,
      "loss": 2.2774,
      "step": 159045
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.199098825454712,
      "learning_rate": 0.00013114523656961312,
      "loss": 2.0267,
      "step": 159046
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.1752336025238037,
      "learning_rate": 0.00013114185549189524,
      "loss": 1.9822,
      "step": 159047
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.3143529891967773,
      "learning_rate": 0.000131138474445571,
      "loss": 1.9869,
      "step": 159048
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.1379308700561523,
      "learning_rate": 0.00013113509343064145,
      "loss": 1.8661,
      "step": 159049
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.1889896392822266,
      "learning_rate": 0.000131131712447107,
      "loss": 2.1469,
      "step": 159050
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.250788688659668,
      "learning_rate": 0.0001311283314949684,
      "loss": 1.9902,
      "step": 159051
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.3255822658538818,
      "learning_rate": 0.00013112495057422617,
      "loss": 1.9218,
      "step": 159052
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.1481133699417114,
      "learning_rate": 0.00013112156968488119,
      "loss": 1.9987,
      "step": 159053
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.253799319267273,
      "learning_rate": 0.0001311181888269337,
      "loss": 1.9618,
      "step": 159054
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.2618063688278198,
      "learning_rate": 0.0001311148080003846,
      "loss": 2.0235,
      "step": 159055
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.071450114250183,
      "learning_rate": 0.00013111142720523438,
      "loss": 1.958,
      "step": 159056
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.1358047723770142,
      "learning_rate": 0.00013110804644148378,
      "loss": 2.0023,
      "step": 159057
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.1669024229049683,
      "learning_rate": 0.00013110466570913332,
      "loss": 1.8998,
      "step": 159058
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.3324989080429077,
      "learning_rate": 0.0001311012850081839,
      "loss": 1.9711,
      "step": 159059
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.1777266263961792,
      "learning_rate": 0.00013109790433863574,
      "loss": 1.8462,
      "step": 159060
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.121828317642212,
      "learning_rate": 0.00013109452370048977,
      "loss": 1.9951,
      "step": 159061
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.454520344734192,
      "learning_rate": 0.0001310911430937464,
      "loss": 2.0091,
      "step": 159062
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.2065845727920532,
      "learning_rate": 0.00013108776251840646,
      "loss": 2.0339,
      "step": 159063
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.1298044919967651,
      "learning_rate": 0.00013108438197447043,
      "loss": 1.8725,
      "step": 159064
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.3940898180007935,
      "learning_rate": 0.00013108100146193917,
      "loss": 2.1779,
      "step": 159065
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.3729857206344604,
      "learning_rate": 0.00013107762098081295,
      "loss": 2.3,
      "step": 159066
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.135115146636963,
      "learning_rate": 0.00013107424053109272,
      "loss": 2.021,
      "step": 159067
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.3564656972885132,
      "learning_rate": 0.00013107086011277883,
      "loss": 2.0837,
      "step": 159068
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.0796585083007812,
      "learning_rate": 0.00013106747972587218,
      "loss": 1.7495,
      "step": 159069
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.1467986106872559,
      "learning_rate": 0.00013106409937037317,
      "loss": 2.3417,
      "step": 159070
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.1935882568359375,
      "learning_rate": 0.00013106071904628273,
      "loss": 2.1659,
      "step": 159071
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.2807601690292358,
      "learning_rate": 0.0001310573387536011,
      "loss": 2.1659,
      "step": 159072
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.2068464756011963,
      "learning_rate": 0.00013105395849232921,
      "loss": 2.0465,
      "step": 159073
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.1170274019241333,
      "learning_rate": 0.00013105057826246747,
      "loss": 2.1229,
      "step": 159074
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.1627955436706543,
      "learning_rate": 0.00013104719806401668,
      "loss": 2.034,
      "step": 159075
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.2669329643249512,
      "learning_rate": 0.00013104381789697734,
      "loss": 2.1666,
      "step": 159076
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.2295684814453125,
      "learning_rate": 0.00013104043776135025,
      "loss": 1.9227,
      "step": 159077
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.2688604593276978,
      "learning_rate": 0.00013103705765713593,
      "loss": 2.0329,
      "step": 159078
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.1384779214859009,
      "learning_rate": 0.00013103367758433499,
      "loss": 2.1596,
      "step": 159079
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.4265940189361572,
      "learning_rate": 0.000131030297542948,
      "loss": 1.9161,
      "step": 159080
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.295608639717102,
      "learning_rate": 0.00013102691753297575,
      "loss": 1.8473,
      "step": 159081
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.0918883085250854,
      "learning_rate": 0.00013102353755441867,
      "loss": 1.9311,
      "step": 159082
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.1721397638320923,
      "learning_rate": 0.00013102015760727765,
      "loss": 2.0267,
      "step": 159083
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.3778769969940186,
      "learning_rate": 0.00013101677769155313,
      "loss": 1.9946,
      "step": 159084
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.4280036687850952,
      "learning_rate": 0.0001310133978072458,
      "loss": 1.9453,
      "step": 159085
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.276219367980957,
      "learning_rate": 0.0001310100179543562,
      "loss": 1.8963,
      "step": 159086
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.330956220626831,
      "learning_rate": 0.0001310066381328851,
      "loss": 2.0527,
      "step": 159087
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.1873059272766113,
      "learning_rate": 0.00013100325834283292,
      "loss": 1.9115,
      "step": 159088
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.2134881019592285,
      "learning_rate": 0.00013099987858420056,
      "loss": 2.0285,
      "step": 159089
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.184920310974121,
      "learning_rate": 0.00013099649885698852,
      "loss": 1.8856,
      "step": 159090
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.1462349891662598,
      "learning_rate": 0.0001309931191611973,
      "loss": 1.7586,
      "step": 159091
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.1673575639724731,
      "learning_rate": 0.00013098973949682776,
      "loss": 1.9212,
      "step": 159092
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.3364512920379639,
      "learning_rate": 0.00013098635986388043,
      "loss": 2.2311,
      "step": 159093
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.2864652872085571,
      "learning_rate": 0.0001309829802623558,
      "loss": 2.0511,
      "step": 159094
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.2689716815948486,
      "learning_rate": 0.00013097960069225476,
      "loss": 2.0892,
      "step": 159095
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.2414578199386597,
      "learning_rate": 0.00013097622115357778,
      "loss": 2.1935,
      "step": 159096
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.158872365951538,
      "learning_rate": 0.0001309728416463254,
      "loss": 2.1809,
      "step": 159097
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.1007733345031738,
      "learning_rate": 0.00013096946217049847,
      "loss": 2.0915,
      "step": 159098
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.1843847036361694,
      "learning_rate": 0.0001309660827260974,
      "loss": 1.9811,
      "step": 159099
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.07993745803833,
      "learning_rate": 0.00013096270331312303,
      "loss": 2.0589,
      "step": 159100
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.1611641645431519,
      "learning_rate": 0.00013095932393157592,
      "loss": 2.0105,
      "step": 159101
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.2370203733444214,
      "learning_rate": 0.0001309559445814566,
      "loss": 2.0913,
      "step": 159102
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.224583625793457,
      "learning_rate": 0.0001309525652627657,
      "loss": 1.8005,
      "step": 159103
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.286805272102356,
      "learning_rate": 0.000130949185975504,
      "loss": 2.0188,
      "step": 159104
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.2336621284484863,
      "learning_rate": 0.00013094580671967195,
      "loss": 2.0273,
      "step": 159105
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.095521092414856,
      "learning_rate": 0.00013094242749527035,
      "loss": 2.1839,
      "step": 159106
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.1446003913879395,
      "learning_rate": 0.00013093904830229974,
      "loss": 1.8079,
      "step": 159107
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.1562196016311646,
      "learning_rate": 0.00013093566914076078,
      "loss": 2.0226,
      "step": 159108
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.0718239545822144,
      "learning_rate": 0.00013093229001065397,
      "loss": 2.1612,
      "step": 159109
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.2719085216522217,
      "learning_rate": 0.00013092891091198009,
      "loss": 1.9353,
      "step": 159110
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.1817213296890259,
      "learning_rate": 0.00013092553184473963,
      "loss": 2.1505,
      "step": 159111
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.1064507961273193,
      "learning_rate": 0.00013092215280893345,
      "loss": 2.0007,
      "step": 159112
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.1945968866348267,
      "learning_rate": 0.0001309187738045619,
      "loss": 1.9568,
      "step": 159113
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.312108039855957,
      "learning_rate": 0.00013091539483162595,
      "loss": 2.1444,
      "step": 159114
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.22392737865448,
      "learning_rate": 0.0001309120158901258,
      "loss": 1.8967,
      "step": 159115
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.2828172445297241,
      "learning_rate": 0.00013090863698006245,
      "loss": 1.9018,
      "step": 159116
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.2860107421875,
      "learning_rate": 0.00013090525810143623,
      "loss": 2.1068,
      "step": 159117
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.1423594951629639,
      "learning_rate": 0.00013090187925424802,
      "loss": 2.199,
      "step": 159118
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.2580516338348389,
      "learning_rate": 0.00013089850043849826,
      "loss": 1.9321,
      "step": 159119
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.1481605768203735,
      "learning_rate": 0.00013089512165418785,
      "loss": 1.8648,
      "step": 159120
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.1907391548156738,
      "learning_rate": 0.00013089174290131702,
      "loss": 2.1537,
      "step": 159121
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.211562156677246,
      "learning_rate": 0.00013088836417988672,
      "loss": 2.1532,
      "step": 159122
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.4664850234985352,
      "learning_rate": 0.00013088498548989734,
      "loss": 2.0315,
      "step": 159123
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.291440725326538,
      "learning_rate": 0.00013088160683134977,
      "loss": 1.9982,
      "step": 159124
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.107584834098816,
      "learning_rate": 0.0001308782282042444,
      "loss": 1.8876,
      "step": 159125
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.2703760862350464,
      "learning_rate": 0.00013087484960858214,
      "loss": 2.2091,
      "step": 159126
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.1655781269073486,
      "learning_rate": 0.00013087147104436322,
      "loss": 1.7716,
      "step": 159127
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.18604576587677,
      "learning_rate": 0.00013086809251158863,
      "loss": 1.9857,
      "step": 159128
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.0801877975463867,
      "learning_rate": 0.00013086471401025875,
      "loss": 1.7161,
      "step": 159129
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.2831236124038696,
      "learning_rate": 0.0001308613355403744,
      "loss": 2.1257,
      "step": 159130
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.3189448118209839,
      "learning_rate": 0.000130857957101936,
      "loss": 2.1513,
      "step": 159131
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.2938300371170044,
      "learning_rate": 0.00013085457869494452,
      "loss": 2.1524,
      "step": 159132
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.1793746948242188,
      "learning_rate": 0.00013085120031940014,
      "loss": 2.123,
      "step": 159133
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.318257212638855,
      "learning_rate": 0.00013084782197530384,
      "loss": 2.1191,
      "step": 159134
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.1921155452728271,
      "learning_rate": 0.00013084444366265603,
      "loss": 2.1329,
      "step": 159135
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.3255627155303955,
      "learning_rate": 0.00013084106538145752,
      "loss": 2.1596,
      "step": 159136
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.1254041194915771,
      "learning_rate": 0.00013083768713170876,
      "loss": 2.0497,
      "step": 159137
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.2802287340164185,
      "learning_rate": 0.00013083430891341064,
      "loss": 1.7993,
      "step": 159138
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.248512864112854,
      "learning_rate": 0.00013083093072656347,
      "loss": 2.2499,
      "step": 159139
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.3569517135620117,
      "learning_rate": 0.00013082755257116808,
      "loss": 2.1609,
      "step": 159140
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.1112505197525024,
      "learning_rate": 0.00013082417444722493,
      "loss": 2.2198,
      "step": 159141
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.20451021194458,
      "learning_rate": 0.00013082079635473487,
      "loss": 2.2427,
      "step": 159142
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.1714816093444824,
      "learning_rate": 0.00013081741829369832,
      "loss": 2.1334,
      "step": 159143
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.5669149160385132,
      "learning_rate": 0.0001308140402641161,
      "loss": 2.1985,
      "step": 159144
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.3076075315475464,
      "learning_rate": 0.00013081066226598874,
      "loss": 2.0163,
      "step": 159145
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.1357320547103882,
      "learning_rate": 0.0001308072842993169,
      "loss": 2.1774,
      "step": 159146
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.3028987646102905,
      "learning_rate": 0.00013080390636410107,
      "loss": 2.1438,
      "step": 159147
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.5245345830917358,
      "learning_rate": 0.0001308005284603421,
      "loss": 2.0698,
      "step": 159148
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.1388946771621704,
      "learning_rate": 0.00013079715058804034,
      "loss": 2.0305,
      "step": 159149
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.1589593887329102,
      "learning_rate": 0.00013079377274719675,
      "loss": 2.2334,
      "step": 159150
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.1873657703399658,
      "learning_rate": 0.0001307903949378118,
      "loss": 2.0857,
      "step": 159151
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.1535245180130005,
      "learning_rate": 0.00013078701715988605,
      "loss": 2.1435,
      "step": 159152
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.2844350337982178,
      "learning_rate": 0.00013078363941342015,
      "loss": 2.0619,
      "step": 159153
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.526787519454956,
      "learning_rate": 0.0001307802616984148,
      "loss": 2.1028,
      "step": 159154
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.3239161968231201,
      "learning_rate": 0.00013077688401487057,
      "loss": 2.096,
      "step": 159155
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.09053373336792,
      "learning_rate": 0.00013077350636278817,
      "loss": 2.5363,
      "step": 159156
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.3834089040756226,
      "learning_rate": 0.00013077012874216815,
      "loss": 2.2149,
      "step": 159157
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.3258907794952393,
      "learning_rate": 0.00013076675115301117,
      "loss": 1.7826,
      "step": 159158
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.4993444681167603,
      "learning_rate": 0.0001307633735953178,
      "loss": 2.1737,
      "step": 159159
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.197608232498169,
      "learning_rate": 0.00013075999606908874,
      "loss": 2.091,
      "step": 159160
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.2164385318756104,
      "learning_rate": 0.0001307566185743245,
      "loss": 2.2208,
      "step": 159161
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.3567010164260864,
      "learning_rate": 0.00013075324111102595,
      "loss": 2.0212,
      "step": 159162
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.3061853647232056,
      "learning_rate": 0.0001307498636791935,
      "loss": 2.264,
      "step": 159163
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.1954048871994019,
      "learning_rate": 0.00013074648627882789,
      "loss": 2.0121,
      "step": 159164
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.1595344543457031,
      "learning_rate": 0.0001307431089099296,
      "loss": 2.0369,
      "step": 159165
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.2929284572601318,
      "learning_rate": 0.00013073973157249946,
      "loss": 2.0468,
      "step": 159166
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.7443366050720215,
      "learning_rate": 0.0001307363542665379,
      "loss": 2.0478,
      "step": 159167
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.216579556465149,
      "learning_rate": 0.00013073297699204572,
      "loss": 1.947,
      "step": 159168
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.2980005741119385,
      "learning_rate": 0.00013072959974902348,
      "loss": 2.1564,
      "step": 159169
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.1522361040115356,
      "learning_rate": 0.00013072622253747183,
      "loss": 2.0457,
      "step": 159170
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.1921406984329224,
      "learning_rate": 0.00013072284535739124,
      "loss": 2.0054,
      "step": 159171
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.1947746276855469,
      "learning_rate": 0.00013071946820878262,
      "loss": 1.8821,
      "step": 159172
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.308388113975525,
      "learning_rate": 0.00013071609109164628,
      "loss": 2.1357,
      "step": 159173
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.1966651678085327,
      "learning_rate": 0.00013071271400598317,
      "loss": 1.7655,
      "step": 159174
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.0989238023757935,
      "learning_rate": 0.00013070933695179374,
      "loss": 2.056,
      "step": 159175
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.3596649169921875,
      "learning_rate": 0.00013070595992907853,
      "loss": 2.197,
      "step": 159176
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.3234280347824097,
      "learning_rate": 0.0001307025829378384,
      "loss": 2.156,
      "step": 159177
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.298812985420227,
      "learning_rate": 0.00013069920597807383,
      "loss": 2.1651,
      "step": 159178
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.2471719980239868,
      "learning_rate": 0.00013069582904978537,
      "loss": 2.0575,
      "step": 159179
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.4917670488357544,
      "learning_rate": 0.0001306924521529739,
      "loss": 2.2075,
      "step": 159180
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.2702124118804932,
      "learning_rate": 0.00013068907528763986,
      "loss": 2.0177,
      "step": 159181
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.205496072769165,
      "learning_rate": 0.00013068569845378378,
      "loss": 2.2028,
      "step": 159182
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.2300769090652466,
      "learning_rate": 0.00013068232165140662,
      "loss": 2.1091,
      "step": 159183
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.1768760681152344,
      "learning_rate": 0.00013067894488050875,
      "loss": 1.9863,
      "step": 159184
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.2207454442977905,
      "learning_rate": 0.00013067556814109075,
      "loss": 1.9379,
      "step": 159185
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.1018457412719727,
      "learning_rate": 0.0001306721914331535,
      "loss": 1.8702,
      "step": 159186
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.1703970432281494,
      "learning_rate": 0.00013066881475669746,
      "loss": 2.0192,
      "step": 159187
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.1541815996170044,
      "learning_rate": 0.0001306654381117232,
      "loss": 2.3572,
      "step": 159188
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.188435435295105,
      "learning_rate": 0.0001306620614982315,
      "loss": 1.9756,
      "step": 159189
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.182400107383728,
      "learning_rate": 0.00013065868491622285,
      "loss": 1.9339,
      "step": 159190
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.226466417312622,
      "learning_rate": 0.00013065530836569802,
      "loss": 1.993,
      "step": 159191
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.1662697792053223,
      "learning_rate": 0.0001306519318466576,
      "loss": 1.9891,
      "step": 159192
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.2877205610275269,
      "learning_rate": 0.00013064855535910216,
      "loss": 2.0631,
      "step": 159193
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.0537259578704834,
      "learning_rate": 0.00013064517890303223,
      "loss": 2.1479,
      "step": 159194
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.1448715925216675,
      "learning_rate": 0.00013064180247844868,
      "loss": 2.0232,
      "step": 159195
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.1348549127578735,
      "learning_rate": 0.0001306384260853519,
      "loss": 2.1363,
      "step": 159196
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.1195169687271118,
      "learning_rate": 0.00013063504972374277,
      "loss": 2.0845,
      "step": 159197
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.1064035892486572,
      "learning_rate": 0.00013063167339362166,
      "loss": 2.0827,
      "step": 159198
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.2977393865585327,
      "learning_rate": 0.00013062829709498952,
      "loss": 1.699,
      "step": 159199
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.147621750831604,
      "learning_rate": 0.0001306249208278466,
      "loss": 2.1122,
      "step": 159200
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.1847128868103027,
      "learning_rate": 0.00013062154459219378,
      "loss": 2.1709,
      "step": 159201
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.1451084613800049,
      "learning_rate": 0.0001306181683880315,
      "loss": 2.1532,
      "step": 159202
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.2333261966705322,
      "learning_rate": 0.00013061479221536062,
      "loss": 2.3852,
      "step": 159203
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.2418522834777832,
      "learning_rate": 0.00013061141607418155,
      "loss": 2.032,
      "step": 159204
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.1923346519470215,
      "learning_rate": 0.0001306080399644952,
      "loss": 2.218,
      "step": 159205
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.3135424852371216,
      "learning_rate": 0.00013060466388630182,
      "loss": 2.1726,
      "step": 159206
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.1729010343551636,
      "learning_rate": 0.00013060128783960232,
      "loss": 2.1638,
      "step": 159207
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.1350338459014893,
      "learning_rate": 0.00013059791182439717,
      "loss": 2.2282,
      "step": 159208
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.275233268737793,
      "learning_rate": 0.00013059453584068713,
      "loss": 1.989,
      "step": 159209
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.1704769134521484,
      "learning_rate": 0.00013059115988847266,
      "loss": 2.0816,
      "step": 159210
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.2255908250808716,
      "learning_rate": 0.00013058778396775463,
      "loss": 2.0517,
      "step": 159211
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.214180827140808,
      "learning_rate": 0.0001305844080785335,
      "loss": 1.9391,
      "step": 159212
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.171651840209961,
      "learning_rate": 0.00013058103222080993,
      "loss": 1.9011,
      "step": 159213
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.0846787691116333,
      "learning_rate": 0.00013057765639458446,
      "loss": 2.0068,
      "step": 159214
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.3141661882400513,
      "learning_rate": 0.00013057428059985788,
      "loss": 2.1357,
      "step": 159215
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.1840189695358276,
      "learning_rate": 0.00013057090483663066,
      "loss": 2.2947,
      "step": 159216
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.2737681865692139,
      "learning_rate": 0.0001305675291049036,
      "loss": 2.1063,
      "step": 159217
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.1344183683395386,
      "learning_rate": 0.00013056415340467723,
      "loss": 1.9794,
      "step": 159218
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.1861460208892822,
      "learning_rate": 0.00013056077773595216,
      "loss": 2.2926,
      "step": 159219
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.3635724782943726,
      "learning_rate": 0.000130557402098729,
      "loss": 1.9765,
      "step": 159220
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.188766598701477,
      "learning_rate": 0.00013055402649300845,
      "loss": 2.074,
      "step": 159221
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.1591283082962036,
      "learning_rate": 0.00013055065091879104,
      "loss": 2.2697,
      "step": 159222
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.2417352199554443,
      "learning_rate": 0.00013054727537607755,
      "loss": 1.9662,
      "step": 159223
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.358657717704773,
      "learning_rate": 0.00013054389986486855,
      "loss": 2.0456,
      "step": 159224
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.163714051246643,
      "learning_rate": 0.0001305405243851646,
      "loss": 2.0204,
      "step": 159225
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.177709937095642,
      "learning_rate": 0.00013053714893696626,
      "loss": 2.1122,
      "step": 159226
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.2525944709777832,
      "learning_rate": 0.00013053377352027442,
      "loss": 2.2301,
      "step": 159227
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.1330125331878662,
      "learning_rate": 0.0001305303981350894,
      "loss": 2.0258,
      "step": 159228
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.30535888671875,
      "learning_rate": 0.0001305270227814121,
      "loss": 2.0644,
      "step": 159229
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.2698121070861816,
      "learning_rate": 0.000130523647459243,
      "loss": 2.0243,
      "step": 159230
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.24288010597229,
      "learning_rate": 0.0001305202721685828,
      "loss": 2.1156,
      "step": 159231
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.2586504220962524,
      "learning_rate": 0.00013051689690943193,
      "loss": 2.1968,
      "step": 159232
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.1813985109329224,
      "learning_rate": 0.00013051352168179131,
      "loss": 2.1481,
      "step": 159233
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.143319010734558,
      "learning_rate": 0.0001305101464856613,
      "loss": 2.1559,
      "step": 159234
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.3028563261032104,
      "learning_rate": 0.00013050677132104275,
      "loss": 1.9074,
      "step": 159235
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.2719948291778564,
      "learning_rate": 0.0001305033961879362,
      "loss": 2.0882,
      "step": 159236
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.261247158050537,
      "learning_rate": 0.00013050002108634228,
      "loss": 1.7826,
      "step": 159237
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.1094681024551392,
      "learning_rate": 0.0001304966460162615,
      "loss": 2.0422,
      "step": 159238
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.214309573173523,
      "learning_rate": 0.00013049327097769468,
      "loss": 2.2748,
      "step": 159239
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.1868999004364014,
      "learning_rate": 0.00013048989597064223,
      "loss": 2.1436,
      "step": 159240
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.2029705047607422,
      "learning_rate": 0.00013048652099510507,
      "loss": 1.8872,
      "step": 159241
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.2108428478240967,
      "learning_rate": 0.00013048314605108367,
      "loss": 2.1305,
      "step": 159242
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.2516398429870605,
      "learning_rate": 0.00013047977113857863,
      "loss": 2.2749,
      "step": 159243
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.1661449670791626,
      "learning_rate": 0.00013047639625759046,
      "loss": 1.9966,
      "step": 159244
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.2287731170654297,
      "learning_rate": 0.00013047302140812008,
      "loss": 2.0678,
      "step": 159245
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.2694246768951416,
      "learning_rate": 0.00013046964659016786,
      "loss": 2.0781,
      "step": 159246
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.299930214881897,
      "learning_rate": 0.0001304662718037346,
      "loss": 2.0358,
      "step": 159247
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.2150012254714966,
      "learning_rate": 0.00013046289704882088,
      "loss": 2.1325,
      "step": 159248
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.1554131507873535,
      "learning_rate": 0.0001304595223254273,
      "loss": 2.1023,
      "step": 159249
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.2428679466247559,
      "learning_rate": 0.0001304561476335544,
      "loss": 2.226,
      "step": 159250
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.1529508829116821,
      "learning_rate": 0.00013045277297320303,
      "loss": 1.9428,
      "step": 159251
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.2548540830612183,
      "learning_rate": 0.00013044939834437356,
      "loss": 1.9646,
      "step": 159252
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.2886526584625244,
      "learning_rate": 0.00013044602374706686,
      "loss": 1.8841,
      "step": 159253
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.1716084480285645,
      "learning_rate": 0.00013044264918128342,
      "loss": 1.9323,
      "step": 159254
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.2474955320358276,
      "learning_rate": 0.00013043927464702392,
      "loss": 2.2654,
      "step": 159255
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.2447435855865479,
      "learning_rate": 0.00013043590014428884,
      "loss": 2.014,
      "step": 159256
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.0636813640594482,
      "learning_rate": 0.00013043252567307905,
      "loss": 1.9731,
      "step": 159257
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.247976303100586,
      "learning_rate": 0.00013042915123339494,
      "loss": 1.9129,
      "step": 159258
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.184707522392273,
      "learning_rate": 0.00013042577682523734,
      "loss": 2.0224,
      "step": 159259
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.320314884185791,
      "learning_rate": 0.0001304224024486068,
      "loss": 2.4375,
      "step": 159260
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.2323492765426636,
      "learning_rate": 0.00013041902810350384,
      "loss": 1.9986,
      "step": 159261
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.1351264715194702,
      "learning_rate": 0.00013041565378992928,
      "loss": 1.992,
      "step": 159262
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.1471248865127563,
      "learning_rate": 0.00013041227950788366,
      "loss": 1.8031,
      "step": 159263
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.2432732582092285,
      "learning_rate": 0.0001304089052573675,
      "loss": 2.0029,
      "step": 159264
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.1660817861557007,
      "learning_rate": 0.0001304055310383816,
      "loss": 2.1364,
      "step": 159265
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.1981755495071411,
      "learning_rate": 0.00013040215685092654,
      "loss": 2.0571,
      "step": 159266
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.1405786275863647,
      "learning_rate": 0.00013039878269500283,
      "loss": 1.949,
      "step": 159267
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.1474653482437134,
      "learning_rate": 0.00013039540857061126,
      "loss": 1.9565,
      "step": 159268
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.331767201423645,
      "learning_rate": 0.0001303920344777524,
      "loss": 1.8683,
      "step": 159269
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.2601279020309448,
      "learning_rate": 0.00013038866041642676,
      "loss": 2.1114,
      "step": 159270
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.1637580394744873,
      "learning_rate": 0.00013038528638663517,
      "loss": 2.0278,
      "step": 159271
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.1610450744628906,
      "learning_rate": 0.00013038191238837814,
      "loss": 2.109,
      "step": 159272
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.1846100091934204,
      "learning_rate": 0.00013037853842165622,
      "loss": 2.1381,
      "step": 159273
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.1240204572677612,
      "learning_rate": 0.00013037516448647022,
      "loss": 2.0695,
      "step": 159274
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.2177155017852783,
      "learning_rate": 0.0001303717905828206,
      "loss": 2.1272,
      "step": 159275
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.1814533472061157,
      "learning_rate": 0.00013036841671070819,
      "loss": 2.0069,
      "step": 159276
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.2005473375320435,
      "learning_rate": 0.00013036504287013346,
      "loss": 2.015,
      "step": 159277
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.29123055934906,
      "learning_rate": 0.00013036166906109695,
      "loss": 2.0275,
      "step": 159278
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.226485252380371,
      "learning_rate": 0.00013035829528359953,
      "loss": 2.0971,
      "step": 159279
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.4644179344177246,
      "learning_rate": 0.00013035492153764175,
      "loss": 1.974,
      "step": 159280
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.2071174383163452,
      "learning_rate": 0.00013035154782322405,
      "loss": 2.1531,
      "step": 159281
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.0669102668762207,
      "learning_rate": 0.0001303481741403473,
      "loss": 1.7768,
      "step": 159282
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.1807332038879395,
      "learning_rate": 0.00013034480048901203,
      "loss": 2.1769,
      "step": 159283
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.1760196685791016,
      "learning_rate": 0.00013034142686921876,
      "loss": 1.9192,
      "step": 159284
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.2606201171875,
      "learning_rate": 0.0001303380532809683,
      "loss": 2.2009,
      "step": 159285
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.0987505912780762,
      "learning_rate": 0.00013033467972426127,
      "loss": 1.9476,
      "step": 159286
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.3749228715896606,
      "learning_rate": 0.00013033130619909803,
      "loss": 2.1125,
      "step": 159287
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.2966336011886597,
      "learning_rate": 0.0001303279327054796,
      "loss": 2.2388,
      "step": 159288
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.1954203844070435,
      "learning_rate": 0.00013032455924340624,
      "loss": 2.1985,
      "step": 159289
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.2016563415527344,
      "learning_rate": 0.0001303211858128789,
      "loss": 2.2151,
      "step": 159290
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.3137547969818115,
      "learning_rate": 0.000130317812413898,
      "loss": 2.0746,
      "step": 159291
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.2537345886230469,
      "learning_rate": 0.00013031443904646422,
      "loss": 1.9543,
      "step": 159292
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.2137845754623413,
      "learning_rate": 0.00013031106571057815,
      "loss": 2.1766,
      "step": 159293
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.406123399734497,
      "learning_rate": 0.00013030769240624052,
      "loss": 1.8954,
      "step": 159294
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.379842758178711,
      "learning_rate": 0.0001303043191334518,
      "loss": 2.0236,
      "step": 159295
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.231864333152771,
      "learning_rate": 0.00013030094589221278,
      "loss": 2.0105,
      "step": 159296
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.3520619869232178,
      "learning_rate": 0.00013029757268252405,
      "loss": 1.8138,
      "step": 159297
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.2723214626312256,
      "learning_rate": 0.0001302941995043862,
      "loss": 2.2093,
      "step": 159298
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.2443792819976807,
      "learning_rate": 0.00013029082635779976,
      "loss": 2.0794,
      "step": 159299
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.1042885780334473,
      "learning_rate": 0.00013028745324276557,
      "loss": 1.8636,
      "step": 159300
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.1952686309814453,
      "learning_rate": 0.000130284080159284,
      "loss": 2.0643,
      "step": 159301
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.3591843843460083,
      "learning_rate": 0.00013028070710735596,
      "loss": 1.8886,
      "step": 159302
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.1868395805358887,
      "learning_rate": 0.00013027733408698194,
      "loss": 2.1789,
      "step": 159303
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.1508036851882935,
      "learning_rate": 0.00013027396109816257,
      "loss": 2.1923,
      "step": 159304
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.1771565675735474,
      "learning_rate": 0.00013027058814089835,
      "loss": 2.1367,
      "step": 159305
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.23529052734375,
      "learning_rate": 0.00013026721521519014,
      "loss": 1.9318,
      "step": 159306
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.189759612083435,
      "learning_rate": 0.00013026384232103837,
      "loss": 2.3219,
      "step": 159307
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.1621602773666382,
      "learning_rate": 0.0001302604694584439,
      "loss": 1.956,
      "step": 159308
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.1851729154586792,
      "learning_rate": 0.00013025709662740716,
      "loss": 1.9415,
      "step": 159309
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.1326558589935303,
      "learning_rate": 0.00013025372382792882,
      "loss": 2.065,
      "step": 159310
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.3793344497680664,
      "learning_rate": 0.00013025035106000943,
      "loss": 2.1792,
      "step": 159311
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.1736363172531128,
      "learning_rate": 0.0001302469783236498,
      "loss": 2.1008,
      "step": 159312
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.1795012950897217,
      "learning_rate": 0.00013024360561885036,
      "loss": 2.2132,
      "step": 159313
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.1807615756988525,
      "learning_rate": 0.00013024023294561196,
      "loss": 2.132,
      "step": 159314
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.1032322645187378,
      "learning_rate": 0.0001302368603039351,
      "loss": 1.781,
      "step": 159315
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.2132556438446045,
      "learning_rate": 0.0001302334876938204,
      "loss": 2.0883,
      "step": 159316
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.0790718793869019,
      "learning_rate": 0.0001302301151152684,
      "loss": 1.7246,
      "step": 159317
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.1433064937591553,
      "learning_rate": 0.00013022674256827995,
      "loss": 1.8077,
      "step": 159318
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.0901128053665161,
      "learning_rate": 0.00013022337005285541,
      "loss": 2.1477,
      "step": 159319
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.1651268005371094,
      "learning_rate": 0.00013021999756899566,
      "loss": 1.9379,
      "step": 159320
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.1964092254638672,
      "learning_rate": 0.00013021662511670124,
      "loss": 2.1361,
      "step": 159321
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.220553994178772,
      "learning_rate": 0.00013021325269597277,
      "loss": 2.0226,
      "step": 159322
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.3069175481796265,
      "learning_rate": 0.0001302098803068107,
      "loss": 2.2269,
      "step": 159323
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.4620935916900635,
      "learning_rate": 0.00013020650794921595,
      "loss": 2.1529,
      "step": 159324
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.1288312673568726,
      "learning_rate": 0.00013020313562318894,
      "loss": 2.0591,
      "step": 159325
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.1872634887695312,
      "learning_rate": 0.00013019976332873044,
      "loss": 1.7776,
      "step": 159326
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.1467299461364746,
      "learning_rate": 0.00013019639106584104,
      "loss": 1.8578,
      "step": 159327
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.2641711235046387,
      "learning_rate": 0.00013019301883452134,
      "loss": 2.0546,
      "step": 159328
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.0915940999984741,
      "learning_rate": 0.00013018964663477183,
      "loss": 2.1492,
      "step": 159329
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.3732340335845947,
      "learning_rate": 0.00013018627446659338,
      "loss": 2.3956,
      "step": 159330
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.5525448322296143,
      "learning_rate": 0.00013018290232998637,
      "loss": 2.0238,
      "step": 159331
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.2193742990493774,
      "learning_rate": 0.00013017953022495171,
      "loss": 1.9253,
      "step": 159332
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.2991688251495361,
      "learning_rate": 0.0001301761581514899,
      "loss": 2.0171,
      "step": 159333
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.5626922845840454,
      "learning_rate": 0.0001301727861096015,
      "loss": 2.1051,
      "step": 159334
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.4727085828781128,
      "learning_rate": 0.00013016941409928713,
      "loss": 2.1715,
      "step": 159335
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.869162678718567,
      "learning_rate": 0.00013016604212054753,
      "loss": 1.8572,
      "step": 159336
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.1965934038162231,
      "learning_rate": 0.00013016267017338317,
      "loss": 2.0292,
      "step": 159337
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.105933427810669,
      "learning_rate": 0.0001301592982577949,
      "loss": 1.8803,
      "step": 159338
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.4733102321624756,
      "learning_rate": 0.00013015592637378323,
      "loss": 2.1706,
      "step": 159339
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.281825065612793,
      "learning_rate": 0.00013015255452134876,
      "loss": 2.1502,
      "step": 159340
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.1933553218841553,
      "learning_rate": 0.00013014918270049203,
      "loss": 1.9604,
      "step": 159341
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.1897985935211182,
      "learning_rate": 0.00013014581091121388,
      "loss": 2.0208,
      "step": 159342
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.1849589347839355,
      "learning_rate": 0.00013014243915351474,
      "loss": 1.9397,
      "step": 159343
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.2513809204101562,
      "learning_rate": 0.0001301390674273954,
      "loss": 2.0904,
      "step": 159344
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.6986839771270752,
      "learning_rate": 0.00013013569573285634,
      "loss": 2.0411,
      "step": 159345
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.2629960775375366,
      "learning_rate": 0.00013013232406989847,
      "loss": 2.0062,
      "step": 159346
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.2185837030410767,
      "learning_rate": 0.00013012895243852198,
      "loss": 2.0872,
      "step": 159347
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.2959601879119873,
      "learning_rate": 0.00013012558083872781,
      "loss": 1.6822,
      "step": 159348
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.1929761171340942,
      "learning_rate": 0.00013012220927051646,
      "loss": 2.1666,
      "step": 159349
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.3096468448638916,
      "learning_rate": 0.00013011883773388866,
      "loss": 2.034,
      "step": 159350
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.3032937049865723,
      "learning_rate": 0.00013011546622884488,
      "loss": 1.9301,
      "step": 159351
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.1354619264602661,
      "learning_rate": 0.00013011209475538593,
      "loss": 2.0266,
      "step": 159352
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.1726040840148926,
      "learning_rate": 0.00013010872331351237,
      "loss": 2.2926,
      "step": 159353
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.308197021484375,
      "learning_rate": 0.0001301053519032248,
      "loss": 2.2572,
      "step": 159354
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.1347907781600952,
      "learning_rate": 0.00013010198052452372,
      "loss": 2.2456,
      "step": 159355
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.4358611106872559,
      "learning_rate": 0.00013009860917741005,
      "loss": 2.1154,
      "step": 159356
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.2288105487823486,
      "learning_rate": 0.0001300952378618841,
      "loss": 2.1324,
      "step": 159357
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.146781086921692,
      "learning_rate": 0.00013009186657794678,
      "loss": 2.1028,
      "step": 159358
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.1781336069107056,
      "learning_rate": 0.00013008849532559856,
      "loss": 2.0405,
      "step": 159359
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.4211573600769043,
      "learning_rate": 0.00013008512410484,
      "loss": 2.2006,
      "step": 159360
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.3970366716384888,
      "learning_rate": 0.00013008175291567196,
      "loss": 1.8886,
      "step": 159361
    },
    {
      "epoch": 2.07,
      "grad_norm": 1.1675925254821777,
      "learning_rate": 0.00013007838175809494,
      "loss": 2.2126,
      "step": 159362
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.1269104480743408,
      "learning_rate": 0.0001300750106321094,
      "loss": 2.1511,
      "step": 159363
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.2353236675262451,
      "learning_rate": 0.00013007163953771626,
      "loss": 1.8985,
      "step": 159364
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.223008632659912,
      "learning_rate": 0.000130068268474916,
      "loss": 2.0205,
      "step": 159365
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.1502704620361328,
      "learning_rate": 0.00013006489744370916,
      "loss": 1.9026,
      "step": 159366
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.0894399881362915,
      "learning_rate": 0.00013006152644409657,
      "loss": 2.2099,
      "step": 159367
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.1594367027282715,
      "learning_rate": 0.00013005815547607874,
      "loss": 2.0187,
      "step": 159368
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.1699373722076416,
      "learning_rate": 0.00013005478453965622,
      "loss": 2.0407,
      "step": 159369
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.4414541721343994,
      "learning_rate": 0.00013005141363482982,
      "loss": 2.1174,
      "step": 159370
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.1851072311401367,
      "learning_rate": 0.00013004804276160005,
      "loss": 1.7594,
      "step": 159371
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.3487956523895264,
      "learning_rate": 0.00013004467191996745,
      "loss": 2.2058,
      "step": 159372
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.1264539957046509,
      "learning_rate": 0.00013004130110993287,
      "loss": 2.0311,
      "step": 159373
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.11106538772583,
      "learning_rate": 0.00013003793033149674,
      "loss": 1.6823,
      "step": 159374
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.4757740497589111,
      "learning_rate": 0.00013003455958465981,
      "loss": 2.1248,
      "step": 159375
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.1999198198318481,
      "learning_rate": 0.00013003118886942272,
      "loss": 2.1162,
      "step": 159376
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.230963945388794,
      "learning_rate": 0.00013002781818578606,
      "loss": 1.9695,
      "step": 159377
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.2344601154327393,
      "learning_rate": 0.00013002444753375028,
      "loss": 2.0069,
      "step": 159378
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.1416356563568115,
      "learning_rate": 0.00013002107691331629,
      "loss": 1.7694,
      "step": 159379
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.4050103425979614,
      "learning_rate": 0.00013001770632448448,
      "loss": 2.0435,
      "step": 159380
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.405335545539856,
      "learning_rate": 0.00013001433576725568,
      "loss": 2.1201,
      "step": 159381
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.1323221921920776,
      "learning_rate": 0.00013001096524163045,
      "loss": 1.949,
      "step": 159382
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.2621424198150635,
      "learning_rate": 0.00013000759474760936,
      "loss": 2.1187,
      "step": 159383
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.1965320110321045,
      "learning_rate": 0.000130004224285193,
      "loss": 2.0005,
      "step": 159384
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.270504117012024,
      "learning_rate": 0.00013000085385438215,
      "loss": 2.169,
      "step": 159385
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.266819715499878,
      "learning_rate": 0.00012999748345517723,
      "loss": 1.9707,
      "step": 159386
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.2040051221847534,
      "learning_rate": 0.00012999411308757913,
      "loss": 2.1111,
      "step": 159387
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.2141271829605103,
      "learning_rate": 0.00012999074275158828,
      "loss": 2.0166,
      "step": 159388
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.2994487285614014,
      "learning_rate": 0.00012998737244720537,
      "loss": 2.2048,
      "step": 159389
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.1819860935211182,
      "learning_rate": 0.00012998400217443094,
      "loss": 1.8718,
      "step": 159390
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.158512830734253,
      "learning_rate": 0.0001299806319332658,
      "loss": 2.097,
      "step": 159391
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.1614166498184204,
      "learning_rate": 0.00012997726172371034,
      "loss": 1.9619,
      "step": 159392
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.2150739431381226,
      "learning_rate": 0.00012997389154576546,
      "loss": 1.9684,
      "step": 159393
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.1986244916915894,
      "learning_rate": 0.00012997052139943164,
      "loss": 2.1927,
      "step": 159394
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.2338645458221436,
      "learning_rate": 0.00012996715128470946,
      "loss": 2.1798,
      "step": 159395
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.1271792650222778,
      "learning_rate": 0.00012996378120159953,
      "loss": 2.0602,
      "step": 159396
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.098016381263733,
      "learning_rate": 0.00012996041115010263,
      "loss": 2.0266,
      "step": 159397
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.1454707384109497,
      "learning_rate": 0.0001299570411302192,
      "loss": 2.0078,
      "step": 159398
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.2680598497390747,
      "learning_rate": 0.00012995367114195007,
      "loss": 2.0886,
      "step": 159399
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.1263900995254517,
      "learning_rate": 0.00012995030118529577,
      "loss": 1.8982,
      "step": 159400
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.132628321647644,
      "learning_rate": 0.0001299469312602569,
      "loss": 1.8561,
      "step": 159401
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.0536885261535645,
      "learning_rate": 0.00012994356136683403,
      "loss": 1.972,
      "step": 159402
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.1101843118667603,
      "learning_rate": 0.00012994019150502795,
      "loss": 1.9344,
      "step": 159403
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.1095036268234253,
      "learning_rate": 0.00012993682167483907,
      "loss": 2.107,
      "step": 159404
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.1728112697601318,
      "learning_rate": 0.0001299334518762683,
      "loss": 1.8231,
      "step": 159405
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.2080082893371582,
      "learning_rate": 0.0001299300821093161,
      "loss": 2.1924,
      "step": 159406
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.269684910774231,
      "learning_rate": 0.0001299267123739831,
      "loss": 2.1586,
      "step": 159407
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.3254014253616333,
      "learning_rate": 0.0001299233426702698,
      "loss": 1.9983,
      "step": 159408
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.2198907136917114,
      "learning_rate": 0.00012991997299817708,
      "loss": 2.2299,
      "step": 159409
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.2420461177825928,
      "learning_rate": 0.00012991660335770534,
      "loss": 2.1488,
      "step": 159410
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.2769821882247925,
      "learning_rate": 0.00012991323374885545,
      "loss": 2.2953,
      "step": 159411
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.1275707483291626,
      "learning_rate": 0.00012990986417162778,
      "loss": 2.0496,
      "step": 159412
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.1920884847640991,
      "learning_rate": 0.00012990649462602327,
      "loss": 1.9757,
      "step": 159413
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.1287983655929565,
      "learning_rate": 0.00012990312511204215,
      "loss": 2.1884,
      "step": 159414
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.1911004781723022,
      "learning_rate": 0.0001298997556296854,
      "loss": 1.9606,
      "step": 159415
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.182804822921753,
      "learning_rate": 0.00012989638617895332,
      "loss": 2.1905,
      "step": 159416
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.2865166664123535,
      "learning_rate": 0.00012989301675984686,
      "loss": 1.9788,
      "step": 159417
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.2394646406173706,
      "learning_rate": 0.00012988964737236638,
      "loss": 2.0249,
      "step": 159418
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.137513518333435,
      "learning_rate": 0.00012988627801651287,
      "loss": 1.9115,
      "step": 159419
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.2278027534484863,
      "learning_rate": 0.00012988290869228643,
      "loss": 2.0124,
      "step": 159420
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.1579952239990234,
      "learning_rate": 0.00012987953939968814,
      "loss": 2.0837,
      "step": 159421
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.1806142330169678,
      "learning_rate": 0.00012987617013871831,
      "loss": 1.8577,
      "step": 159422
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.2644851207733154,
      "learning_rate": 0.00012987280090937783,
      "loss": 2.2475,
      "step": 159423
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.1440809965133667,
      "learning_rate": 0.0001298694317116671,
      "loss": 2.0531,
      "step": 159424
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.1320650577545166,
      "learning_rate": 0.00012986606254558708,
      "loss": 1.9094,
      "step": 159425
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.2905399799346924,
      "learning_rate": 0.00012986269341113795,
      "loss": 1.8071,
      "step": 159426
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.1969213485717773,
      "learning_rate": 0.00012985932430832067,
      "loss": 2.0498,
      "step": 159427
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.2437694072723389,
      "learning_rate": 0.00012985595523713563,
      "loss": 1.9929,
      "step": 159428
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.1813552379608154,
      "learning_rate": 0.0001298525861975837,
      "loss": 1.8746,
      "step": 159429
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.3113296031951904,
      "learning_rate": 0.00012984921718966528,
      "loss": 1.9728,
      "step": 159430
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.2585197687149048,
      "learning_rate": 0.0001298458482133813,
      "loss": 2.0571,
      "step": 159431
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.2244744300842285,
      "learning_rate": 0.00012984247926873198,
      "loss": 1.9131,
      "step": 159432
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.1570855379104614,
      "learning_rate": 0.00012983911035571827,
      "loss": 2.0321,
      "step": 159433
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.1725053787231445,
      "learning_rate": 0.00012983574147434058,
      "loss": 2.0063,
      "step": 159434
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.1960402727127075,
      "learning_rate": 0.00012983237262459974,
      "loss": 2.1902,
      "step": 159435
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.2431542873382568,
      "learning_rate": 0.00012982900380649617,
      "loss": 2.0601,
      "step": 159436
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.1060919761657715,
      "learning_rate": 0.0001298256350200307,
      "loss": 2.0139,
      "step": 159437
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.2498679161071777,
      "learning_rate": 0.00012982226626520385,
      "loss": 2.1951,
      "step": 159438
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.1856063604354858,
      "learning_rate": 0.00012981889754201626,
      "loss": 2.0165,
      "step": 159439
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.2054682970046997,
      "learning_rate": 0.00012981552885046844,
      "loss": 1.7336,
      "step": 159440
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.2728662490844727,
      "learning_rate": 0.00012981216019056124,
      "loss": 2.1577,
      "step": 159441
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.3502200841903687,
      "learning_rate": 0.00012980879156229507,
      "loss": 2.3065,
      "step": 159442
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.1334210634231567,
      "learning_rate": 0.00012980542296567074,
      "loss": 2.1169,
      "step": 159443
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.2268939018249512,
      "learning_rate": 0.0001298020544006888,
      "loss": 2.0968,
      "step": 159444
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.388001799583435,
      "learning_rate": 0.0001297986858673498,
      "loss": 2.0366,
      "step": 159445
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.2148730754852295,
      "learning_rate": 0.0001297953173656545,
      "loss": 1.8345,
      "step": 159446
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.2245209217071533,
      "learning_rate": 0.00012979194889560348,
      "loss": 1.8253,
      "step": 159447
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.3365627527236938,
      "learning_rate": 0.00012978858045719724,
      "loss": 2.211,
      "step": 159448
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.1663727760314941,
      "learning_rate": 0.00012978521205043666,
      "loss": 2.0996,
      "step": 159449
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.1805294752120972,
      "learning_rate": 0.0001297818436753222,
      "loss": 1.9085,
      "step": 159450
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.2860205173492432,
      "learning_rate": 0.00012977847533185438,
      "loss": 1.9248,
      "step": 159451
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.2083570957183838,
      "learning_rate": 0.00012977510702003405,
      "loss": 2.1013,
      "step": 159452
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.3293626308441162,
      "learning_rate": 0.00012977173873986178,
      "loss": 2.2029,
      "step": 159453
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.164628505706787,
      "learning_rate": 0.00012976837049133802,
      "loss": 2.1202,
      "step": 159454
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.34396493434906,
      "learning_rate": 0.00012976500227446366,
      "loss": 1.984,
      "step": 159455
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.0544859170913696,
      "learning_rate": 0.0001297616340892392,
      "loss": 1.9679,
      "step": 159456
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.145007610321045,
      "learning_rate": 0.00012975826593566517,
      "loss": 2.0035,
      "step": 159457
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.2308228015899658,
      "learning_rate": 0.00012975489781374237,
      "loss": 2.184,
      "step": 159458
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.1581122875213623,
      "learning_rate": 0.00012975152972347125,
      "loss": 2.0219,
      "step": 159459
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.1435827016830444,
      "learning_rate": 0.00012974816166485265,
      "loss": 1.9039,
      "step": 159460
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.2254791259765625,
      "learning_rate": 0.00012974479363788707,
      "loss": 2.1437,
      "step": 159461
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.3247944116592407,
      "learning_rate": 0.00012974142564257514,
      "loss": 2.1573,
      "step": 159462
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.179053783416748,
      "learning_rate": 0.00012973805767891737,
      "loss": 1.9478,
      "step": 159463
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.179012656211853,
      "learning_rate": 0.00012973468974691467,
      "loss": 2.0705,
      "step": 159464
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.27918541431427,
      "learning_rate": 0.00012973132184656736,
      "loss": 2.0429,
      "step": 159465
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.1266684532165527,
      "learning_rate": 0.00012972795397787632,
      "loss": 2.008,
      "step": 159466
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.1474162340164185,
      "learning_rate": 0.0001297245861408421,
      "loss": 1.9496,
      "step": 159467
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.1015410423278809,
      "learning_rate": 0.00012972121833546528,
      "loss": 2.2612,
      "step": 159468
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.1582677364349365,
      "learning_rate": 0.00012971785056174637,
      "loss": 1.8444,
      "step": 159469
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.260970950126648,
      "learning_rate": 0.00012971448281968626,
      "loss": 2.0556,
      "step": 159470
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.162519097328186,
      "learning_rate": 0.0001297111151092853,
      "loss": 1.9381,
      "step": 159471
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.1008354425430298,
      "learning_rate": 0.0001297077474305444,
      "loss": 1.9352,
      "step": 159472
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.4667928218841553,
      "learning_rate": 0.00012970437978346403,
      "loss": 2.061,
      "step": 159473
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.241538166999817,
      "learning_rate": 0.00012970101216804483,
      "loss": 1.9576,
      "step": 159474
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.267451286315918,
      "learning_rate": 0.00012969764458428733,
      "loss": 2.0416,
      "step": 159475
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.2256114482879639,
      "learning_rate": 0.00012969427703219235,
      "loss": 1.9321,
      "step": 159476
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.1957578659057617,
      "learning_rate": 0.0001296909095117603,
      "loss": 2.1296,
      "step": 159477
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.1667447090148926,
      "learning_rate": 0.00012968754202299202,
      "loss": 1.8906,
      "step": 159478
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.0892890691757202,
      "learning_rate": 0.00012968417456588798,
      "loss": 2.309,
      "step": 159479
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.1200624704360962,
      "learning_rate": 0.00012968080714044903,
      "loss": 2.0827,
      "step": 159480
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.1386964321136475,
      "learning_rate": 0.0001296774397466755,
      "loss": 1.8266,
      "step": 159481
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.237805962562561,
      "learning_rate": 0.00012967407238456815,
      "loss": 2.0344,
      "step": 159482
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.251680850982666,
      "learning_rate": 0.00012967070505412757,
      "loss": 2.0443,
      "step": 159483
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.1102838516235352,
      "learning_rate": 0.0001296673377553545,
      "loss": 1.9207,
      "step": 159484
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.1696066856384277,
      "learning_rate": 0.00012966397048824938,
      "loss": 1.9981,
      "step": 159485
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.1245331764221191,
      "learning_rate": 0.0001296606032528132,
      "loss": 2.26,
      "step": 159486
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.198647141456604,
      "learning_rate": 0.00012965723604904604,
      "loss": 1.9469,
      "step": 159487
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.1700694561004639,
      "learning_rate": 0.00012965386887694901,
      "loss": 1.8701,
      "step": 159488
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.069867491722107,
      "learning_rate": 0.00012965050173652237,
      "loss": 1.9107,
      "step": 159489
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.2100945711135864,
      "learning_rate": 0.00012964713462776704,
      "loss": 1.988,
      "step": 159490
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.163157343864441,
      "learning_rate": 0.00012964376755068342,
      "loss": 2.0545,
      "step": 159491
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.1041761636734009,
      "learning_rate": 0.00012964040050527246,
      "loss": 2.1584,
      "step": 159492
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.186867117881775,
      "learning_rate": 0.00012963703349153433,
      "loss": 2.1246,
      "step": 159493
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.1082208156585693,
      "learning_rate": 0.00012963366650947005,
      "loss": 2.0419,
      "step": 159494
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.2088862657546997,
      "learning_rate": 0.0001296302995590799,
      "loss": 2.203,
      "step": 159495
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.088736653327942,
      "learning_rate": 0.00012962693264036485,
      "loss": 2.0549,
      "step": 159496
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.3738934993743896,
      "learning_rate": 0.00012962356575332525,
      "loss": 2.0012,
      "step": 159497
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.6031287908554077,
      "learning_rate": 0.00012962019889796206,
      "loss": 1.9197,
      "step": 159498
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.2551709413528442,
      "learning_rate": 0.0001296168320742755,
      "loss": 2.002,
      "step": 159499
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.2290065288543701,
      "learning_rate": 0.00012961346528226646,
      "loss": 2.2331,
      "step": 159500
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.2812654972076416,
      "learning_rate": 0.0001296100985219354,
      "loss": 1.9732,
      "step": 159501
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.1534371376037598,
      "learning_rate": 0.00012960673179328314,
      "loss": 2.038,
      "step": 159502
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.2545496225357056,
      "learning_rate": 0.00012960336509631012,
      "loss": 2.2582,
      "step": 159503
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.0848236083984375,
      "learning_rate": 0.00012959999843101727,
      "loss": 2.0708,
      "step": 159504
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.4497361183166504,
      "learning_rate": 0.00012959663179740473,
      "loss": 2.2105,
      "step": 159505
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.7167407274246216,
      "learning_rate": 0.00012959326519547355,
      "loss": 2.0596,
      "step": 159506
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.237950086593628,
      "learning_rate": 0.00012958989862522406,
      "loss": 2.046,
      "step": 159507
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.2741469144821167,
      "learning_rate": 0.00012958653208665715,
      "loss": 2.2368,
      "step": 159508
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.1904958486557007,
      "learning_rate": 0.0001295831655797732,
      "loss": 2.0507,
      "step": 159509
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.6049312353134155,
      "learning_rate": 0.0001295797991045732,
      "loss": 1.9881,
      "step": 159510
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.192470669746399,
      "learning_rate": 0.0001295764326610573,
      "loss": 2.1673,
      "step": 159511
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.5679707527160645,
      "learning_rate": 0.00012957306624922647,
      "loss": 1.9265,
      "step": 159512
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.3110560178756714,
      "learning_rate": 0.0001295696998690811,
      "loss": 2.018,
      "step": 159513
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.5758363008499146,
      "learning_rate": 0.00012956633352062208,
      "loss": 2.1338,
      "step": 159514
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.203128457069397,
      "learning_rate": 0.00012956296720384973,
      "loss": 2.0525,
      "step": 159515
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.2856398820877075,
      "learning_rate": 0.00012955960091876512,
      "loss": 2.1785,
      "step": 159516
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.113877296447754,
      "learning_rate": 0.00012955623466536835,
      "loss": 2.0842,
      "step": 159517
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.1138501167297363,
      "learning_rate": 0.00012955286844366042,
      "loss": 1.9317,
      "step": 159518
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.1767574548721313,
      "learning_rate": 0.0001295495022536417,
      "loss": 2.0199,
      "step": 159519
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.3608577251434326,
      "learning_rate": 0.00012954613609531305,
      "loss": 2.0944,
      "step": 159520
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.166520118713379,
      "learning_rate": 0.0001295427699686749,
      "loss": 1.9854,
      "step": 159521
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.2458263635635376,
      "learning_rate": 0.00012953940387372807,
      "loss": 1.8535,
      "step": 159522
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.1359844207763672,
      "learning_rate": 0.00012953603781047307,
      "loss": 2.1123,
      "step": 159523
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.243345022201538,
      "learning_rate": 0.00012953267177891057,
      "loss": 2.0021,
      "step": 159524
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.055691123008728,
      "learning_rate": 0.00012952930577904103,
      "loss": 2.0676,
      "step": 159525
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.0591895580291748,
      "learning_rate": 0.0001295259398108653,
      "loss": 2.2842,
      "step": 159526
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.0819008350372314,
      "learning_rate": 0.00012952257387438383,
      "loss": 1.8255,
      "step": 159527
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.3064557313919067,
      "learning_rate": 0.00012951920796959744,
      "loss": 2.0497,
      "step": 159528
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.0969146490097046,
      "learning_rate": 0.00012951584209650667,
      "loss": 2.088,
      "step": 159529
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.082687497138977,
      "learning_rate": 0.00012951247625511206,
      "loss": 1.8952,
      "step": 159530
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.1620099544525146,
      "learning_rate": 0.00012950911044541424,
      "loss": 1.9014,
      "step": 159531
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.1064034700393677,
      "learning_rate": 0.00012950574466741402,
      "loss": 1.9622,
      "step": 159532
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.156267523765564,
      "learning_rate": 0.00012950237892111175,
      "loss": 1.9881,
      "step": 159533
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.205776572227478,
      "learning_rate": 0.00012949901320650833,
      "loss": 1.9058,
      "step": 159534
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.1787099838256836,
      "learning_rate": 0.0001294956475236043,
      "loss": 2.0156,
      "step": 159535
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.2417153120040894,
      "learning_rate": 0.00012949228187240008,
      "loss": 2.2042,
      "step": 159536
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.231407880783081,
      "learning_rate": 0.00012948891625289658,
      "loss": 1.9743,
      "step": 159537
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.115893840789795,
      "learning_rate": 0.00012948555066509435,
      "loss": 1.9876,
      "step": 159538
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.2312568426132202,
      "learning_rate": 0.00012948218510899383,
      "loss": 1.9035,
      "step": 159539
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.20171058177948,
      "learning_rate": 0.0001294788195845959,
      "loss": 1.9524,
      "step": 159540
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.1276484727859497,
      "learning_rate": 0.0001294754540919011,
      "loss": 1.9257,
      "step": 159541
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.327562928199768,
      "learning_rate": 0.00012947208863090987,
      "loss": 2.0411,
      "step": 159542
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.1468534469604492,
      "learning_rate": 0.00012946872320162317,
      "loss": 2.1178,
      "step": 159543
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.1802769899368286,
      "learning_rate": 0.00012946535780404133,
      "loss": 1.9686,
      "step": 159544
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.1214996576309204,
      "learning_rate": 0.00012946199243816516,
      "loss": 2.0987,
      "step": 159545
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.2550818920135498,
      "learning_rate": 0.00012945862710399527,
      "loss": 1.9711,
      "step": 159546
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.1464937925338745,
      "learning_rate": 0.00012945526180153225,
      "loss": 2.0642,
      "step": 159547
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.2319374084472656,
      "learning_rate": 0.00012945189653077658,
      "loss": 2.0097,
      "step": 159548
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.2735135555267334,
      "learning_rate": 0.00012944853129172914,
      "loss": 1.9423,
      "step": 159549
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.1961805820465088,
      "learning_rate": 0.00012944516608439033,
      "loss": 1.9242,
      "step": 159550
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.2932441234588623,
      "learning_rate": 0.000129441800908761,
      "loss": 2.0287,
      "step": 159551
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.140519380569458,
      "learning_rate": 0.00012943843576484164,
      "loss": 1.8887,
      "step": 159552
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.3138989210128784,
      "learning_rate": 0.0001294350706526329,
      "loss": 2.1657,
      "step": 159553
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.1992297172546387,
      "learning_rate": 0.0001294317055721353,
      "loss": 2.0496,
      "step": 159554
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.1977505683898926,
      "learning_rate": 0.00012942834052334965,
      "loss": 2.093,
      "step": 159555
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.4240456819534302,
      "learning_rate": 0.0001294249755062764,
      "loss": 2.0712,
      "step": 159556
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.1981756687164307,
      "learning_rate": 0.00012942161052091638,
      "loss": 2.0416,
      "step": 159557
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.1892976760864258,
      "learning_rate": 0.00012941824556726998,
      "loss": 1.9844,
      "step": 159558
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.3109058141708374,
      "learning_rate": 0.0001294148806453382,
      "loss": 2.2216,
      "step": 159559
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.253839135169983,
      "learning_rate": 0.00012941151575512119,
      "loss": 2.0832,
      "step": 159560
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.3666601181030273,
      "learning_rate": 0.00012940815089661987,
      "loss": 2.1887,
      "step": 159561
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.1048070192337036,
      "learning_rate": 0.00012940478606983472,
      "loss": 2.0682,
      "step": 159562
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.154591679573059,
      "learning_rate": 0.00012940142127476653,
      "loss": 2.2455,
      "step": 159563
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.2329325675964355,
      "learning_rate": 0.00012939805651141572,
      "loss": 1.9528,
      "step": 159564
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.3074511289596558,
      "learning_rate": 0.00012939469177978327,
      "loss": 1.9905,
      "step": 159565
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.206009864807129,
      "learning_rate": 0.00012939132707986937,
      "loss": 2.0357,
      "step": 159566
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.1849493980407715,
      "learning_rate": 0.0001293879624116749,
      "loss": 2.0736,
      "step": 159567
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.108133316040039,
      "learning_rate": 0.00012938459777520037,
      "loss": 2.1893,
      "step": 159568
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.3057899475097656,
      "learning_rate": 0.00012938123317044652,
      "loss": 2.0023,
      "step": 159569
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.1860930919647217,
      "learning_rate": 0.00012937786859741385,
      "loss": 2.2441,
      "step": 159570
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.1077501773834229,
      "learning_rate": 0.00012937450405610325,
      "loss": 1.9832,
      "step": 159571
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.152228593826294,
      "learning_rate": 0.00012937113954651495,
      "loss": 1.7212,
      "step": 159572
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.1417049169540405,
      "learning_rate": 0.00012936777506864988,
      "loss": 2.0615,
      "step": 159573
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.2647089958190918,
      "learning_rate": 0.00012936441062250847,
      "loss": 1.9566,
      "step": 159574
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.1201667785644531,
      "learning_rate": 0.00012936104620809148,
      "loss": 1.9208,
      "step": 159575
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.3159008026123047,
      "learning_rate": 0.00012935768182539945,
      "loss": 2.1561,
      "step": 159576
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.1744037866592407,
      "learning_rate": 0.00012935431747443323,
      "loss": 1.9549,
      "step": 159577
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.1373605728149414,
      "learning_rate": 0.00012935095315519307,
      "loss": 1.6951,
      "step": 159578
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.1801203489303589,
      "learning_rate": 0.00012934758886767989,
      "loss": 2.0702,
      "step": 159579
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.3427239656448364,
      "learning_rate": 0.00012934422461189412,
      "loss": 2.1563,
      "step": 159580
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.0891145467758179,
      "learning_rate": 0.00012934086038783658,
      "loss": 2.2857,
      "step": 159581
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.2355350255966187,
      "learning_rate": 0.00012933749619550765,
      "loss": 2.1411,
      "step": 159582
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.0722287893295288,
      "learning_rate": 0.00012933413203490836,
      "loss": 2.0078,
      "step": 159583
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.2530659437179565,
      "learning_rate": 0.00012933076790603887,
      "loss": 2.2614,
      "step": 159584
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.1783865690231323,
      "learning_rate": 0.0001293274038089001,
      "loss": 1.9489,
      "step": 159585
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.2490216493606567,
      "learning_rate": 0.00012932403974349246,
      "loss": 2.0973,
      "step": 159586
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.1179921627044678,
      "learning_rate": 0.00012932067570981684,
      "loss": 2.2252,
      "step": 159587
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.2056437730789185,
      "learning_rate": 0.0001293173117078736,
      "loss": 1.9403,
      "step": 159588
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.1881815195083618,
      "learning_rate": 0.0001293139477376637,
      "loss": 2.1019,
      "step": 159589
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.1492549180984497,
      "learning_rate": 0.00012931058379918737,
      "loss": 2.0224,
      "step": 159590
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.2280387878417969,
      "learning_rate": 0.00012930721989244553,
      "loss": 2.0827,
      "step": 159591
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.2224878072738647,
      "learning_rate": 0.0001293038560174386,
      "loss": 2.0348,
      "step": 159592
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.3639485836029053,
      "learning_rate": 0.0001293004921741674,
      "loss": 2.0872,
      "step": 159593
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.2336516380310059,
      "learning_rate": 0.00012929712836263235,
      "loss": 1.8497,
      "step": 159594
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.3074826002120972,
      "learning_rate": 0.0001292937645828344,
      "loss": 2.0265,
      "step": 159595
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.1803970336914062,
      "learning_rate": 0.00012929040083477374,
      "loss": 2.0325,
      "step": 159596
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.3309903144836426,
      "learning_rate": 0.00012928703711845134,
      "loss": 2.1385,
      "step": 159597
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.491295576095581,
      "learning_rate": 0.0001292836734338676,
      "loss": 1.9796,
      "step": 159598
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.137749195098877,
      "learning_rate": 0.00012928030978102335,
      "loss": 1.8138,
      "step": 159599
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.2124853134155273,
      "learning_rate": 0.00012927694615991897,
      "loss": 1.9251,
      "step": 159600
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.344604253768921,
      "learning_rate": 0.0001292735825705555,
      "loss": 2.0412,
      "step": 159601
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.186181664466858,
      "learning_rate": 0.00012927021901293302,
      "loss": 2.2118,
      "step": 159602
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.1534510850906372,
      "learning_rate": 0.00012926685548705257,
      "loss": 2.1303,
      "step": 159603
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.1357574462890625,
      "learning_rate": 0.00012926349199291455,
      "loss": 2.0401,
      "step": 159604
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.1548470258712769,
      "learning_rate": 0.00012926012853051974,
      "loss": 1.9608,
      "step": 159605
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.1684819459915161,
      "learning_rate": 0.00012925676509986862,
      "loss": 2.1311,
      "step": 159606
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.174639105796814,
      "learning_rate": 0.00012925340170096195,
      "loss": 1.8984,
      "step": 159607
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.1887520551681519,
      "learning_rate": 0.00012925003833380035,
      "loss": 2.0266,
      "step": 159608
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.2655655145645142,
      "learning_rate": 0.00012924667499838432,
      "loss": 2.308,
      "step": 159609
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.333808422088623,
      "learning_rate": 0.0001292433116947145,
      "loss": 2.2495,
      "step": 159610
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.2969818115234375,
      "learning_rate": 0.00012923994842279168,
      "loss": 1.9051,
      "step": 159611
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.131779670715332,
      "learning_rate": 0.00012923658518261628,
      "loss": 2.0535,
      "step": 159612
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.2416365146636963,
      "learning_rate": 0.00012923322197418908,
      "loss": 2.0383,
      "step": 159613
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.3315091133117676,
      "learning_rate": 0.00012922985879751072,
      "loss": 2.1274,
      "step": 159614
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.151585578918457,
      "learning_rate": 0.0001292264956525817,
      "loss": 2.0597,
      "step": 159615
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.2306350469589233,
      "learning_rate": 0.00012922313253940258,
      "loss": 2.0356,
      "step": 159616
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.5105968713760376,
      "learning_rate": 0.00012921976945797425,
      "loss": 1.9946,
      "step": 159617
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.3469064235687256,
      "learning_rate": 0.00012921640640829708,
      "loss": 1.9527,
      "step": 159618
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.3743222951889038,
      "learning_rate": 0.00012921304339037187,
      "loss": 1.9661,
      "step": 159619
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.2066597938537598,
      "learning_rate": 0.0001292096804041992,
      "loss": 2.0265,
      "step": 159620
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.1563137769699097,
      "learning_rate": 0.00012920631744977958,
      "loss": 1.9704,
      "step": 159621
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.2726908922195435,
      "learning_rate": 0.00012920295452711384,
      "loss": 1.9785,
      "step": 159622
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.2969454526901245,
      "learning_rate": 0.00012919959163620247,
      "loss": 2.1988,
      "step": 159623
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.2417196035385132,
      "learning_rate": 0.00012919622877704604,
      "loss": 1.9771,
      "step": 159624
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.2348906993865967,
      "learning_rate": 0.00012919286594964536,
      "loss": 2.2228,
      "step": 159625
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.345192551612854,
      "learning_rate": 0.0001291895031540009,
      "loss": 2.1496,
      "step": 159626
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.2178311347961426,
      "learning_rate": 0.0001291861403901133,
      "loss": 1.9019,
      "step": 159627
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.3286123275756836,
      "learning_rate": 0.00012918277765798327,
      "loss": 2.1344,
      "step": 159628
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.2878199815750122,
      "learning_rate": 0.00012917941495761142,
      "loss": 2.0706,
      "step": 159629
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.203452229499817,
      "learning_rate": 0.00012917605228899822,
      "loss": 2.0341,
      "step": 159630
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.0857570171356201,
      "learning_rate": 0.00012917268965214452,
      "loss": 1.932,
      "step": 159631
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.2644211053848267,
      "learning_rate": 0.0001291693270470509,
      "loss": 2.1265,
      "step": 159632
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.4212511777877808,
      "learning_rate": 0.00012916596447371773,
      "loss": 2.1079,
      "step": 159633
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.1440346240997314,
      "learning_rate": 0.000129162601932146,
      "loss": 1.9113,
      "step": 159634
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.206109881401062,
      "learning_rate": 0.00012915923942233604,
      "loss": 2.013,
      "step": 159635
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.1658190488815308,
      "learning_rate": 0.00012915587694428872,
      "loss": 2.1591,
      "step": 159636
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.0899603366851807,
      "learning_rate": 0.0001291525144980045,
      "loss": 2.1547,
      "step": 159637
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.2591307163238525,
      "learning_rate": 0.00012914915208348413,
      "loss": 1.9987,
      "step": 159638
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.2319835424423218,
      "learning_rate": 0.00012914578970072798,
      "loss": 2.3066,
      "step": 159639
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.151068925857544,
      "learning_rate": 0.000129142427349737,
      "loss": 1.8383,
      "step": 159640
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.2468175888061523,
      "learning_rate": 0.00012913906503051156,
      "loss": 1.8096,
      "step": 159641
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.1357899904251099,
      "learning_rate": 0.00012913570274305247,
      "loss": 2.0759,
      "step": 159642
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.15821373462677,
      "learning_rate": 0.00012913234048736019,
      "loss": 2.1925,
      "step": 159643
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.366197943687439,
      "learning_rate": 0.00012912897826343567,
      "loss": 2.0001,
      "step": 159644
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.1409376859664917,
      "learning_rate": 0.00012912561607127904,
      "loss": 1.9236,
      "step": 159645
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.1209628582000732,
      "learning_rate": 0.00012912225391089132,
      "loss": 2.0558,
      "step": 159646
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.2182505130767822,
      "learning_rate": 0.00012911889178227287,
      "loss": 2.0352,
      "step": 159647
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.555269479751587,
      "learning_rate": 0.00012911552968542458,
      "loss": 1.9936,
      "step": 159648
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.2151334285736084,
      "learning_rate": 0.0001291121676203468,
      "loss": 1.8793,
      "step": 159649
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.1811163425445557,
      "learning_rate": 0.00012910880558704055,
      "loss": 1.8963,
      "step": 159650
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.310498833656311,
      "learning_rate": 0.00012910544358550596,
      "loss": 2.1815,
      "step": 159651
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.3323872089385986,
      "learning_rate": 0.000129102081615744,
      "loss": 2.0486,
      "step": 159652
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.7335587739944458,
      "learning_rate": 0.00012909871967775507,
      "loss": 2.1119,
      "step": 159653
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.1501935720443726,
      "learning_rate": 0.00012909535777154006,
      "loss": 2.0803,
      "step": 159654
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.1949970722198486,
      "learning_rate": 0.00012909199589709929,
      "loss": 2.0366,
      "step": 159655
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.4086735248565674,
      "learning_rate": 0.00012908863405443382,
      "loss": 2.1303,
      "step": 159656
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.20902419090271,
      "learning_rate": 0.00012908527224354374,
      "loss": 1.8258,
      "step": 159657
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.58522367477417,
      "learning_rate": 0.00012908191046443005,
      "loss": 1.9623,
      "step": 159658
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.1956323385238647,
      "learning_rate": 0.00012907854871709314,
      "loss": 2.0658,
      "step": 159659
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.1830778121948242,
      "learning_rate": 0.0001290751870015339,
      "loss": 1.821,
      "step": 159660
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.2333790063858032,
      "learning_rate": 0.00012907182531775264,
      "loss": 2.2502,
      "step": 159661
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.1458988189697266,
      "learning_rate": 0.0001290684636657504,
      "loss": 2.1583,
      "step": 159662
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.3153079748153687,
      "learning_rate": 0.00012906510204552737,
      "loss": 1.7426,
      "step": 159663
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.4759293794631958,
      "learning_rate": 0.00012906174045708446,
      "loss": 2.1157,
      "step": 159664
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.300091028213501,
      "learning_rate": 0.00012905837890042206,
      "loss": 2.0297,
      "step": 159665
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.409045696258545,
      "learning_rate": 0.00012905501737554107,
      "loss": 2.0052,
      "step": 159666
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.5676034688949585,
      "learning_rate": 0.00012905165588244187,
      "loss": 1.9124,
      "step": 159667
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.3396483659744263,
      "learning_rate": 0.0001290482944211254,
      "loss": 1.9,
      "step": 159668
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.252843976020813,
      "learning_rate": 0.00012904493299159187,
      "loss": 1.9732,
      "step": 159669
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.2340716123580933,
      "learning_rate": 0.00012904157159384223,
      "loss": 2.1494,
      "step": 159670
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.4491089582443237,
      "learning_rate": 0.00012903821022787685,
      "loss": 2.109,
      "step": 159671
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.0528442859649658,
      "learning_rate": 0.00012903484889369664,
      "loss": 1.8703,
      "step": 159672
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.2102638483047485,
      "learning_rate": 0.00012903148759130193,
      "loss": 1.9738,
      "step": 159673
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.1996468305587769,
      "learning_rate": 0.00012902812632069376,
      "loss": 2.1969,
      "step": 159674
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.3069758415222168,
      "learning_rate": 0.00012902476508187225,
      "loss": 1.9308,
      "step": 159675
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.2852104902267456,
      "learning_rate": 0.00012902140387483835,
      "loss": 2.1729,
      "step": 159676
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.1370185613632202,
      "learning_rate": 0.0001290180426995925,
      "loss": 2.1305,
      "step": 159677
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.1377967596054077,
      "learning_rate": 0.00012901468155613555,
      "loss": 2.2606,
      "step": 159678
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.1549925804138184,
      "learning_rate": 0.00012901132044446786,
      "loss": 2.0268,
      "step": 159679
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.1139510869979858,
      "learning_rate": 0.00012900795936459034,
      "loss": 2.2513,
      "step": 159680
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.1696197986602783,
      "learning_rate": 0.00012900459831650347,
      "loss": 1.9293,
      "step": 159681
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.120888590812683,
      "learning_rate": 0.00012900123730020784,
      "loss": 2.0034,
      "step": 159682
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.1193697452545166,
      "learning_rate": 0.00012899787631570403,
      "loss": 2.0812,
      "step": 159683
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.1961148977279663,
      "learning_rate": 0.00012899451536299282,
      "loss": 1.8458,
      "step": 159684
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.4309874773025513,
      "learning_rate": 0.00012899115444207465,
      "loss": 1.8259,
      "step": 159685
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.2037845849990845,
      "learning_rate": 0.00012898779355295038,
      "loss": 2.0432,
      "step": 159686
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.2681808471679688,
      "learning_rate": 0.0001289844326956205,
      "loss": 1.976,
      "step": 159687
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.092079520225525,
      "learning_rate": 0.00012898107187008564,
      "loss": 2.0623,
      "step": 159688
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.2184141874313354,
      "learning_rate": 0.00012897771107634633,
      "loss": 2.1276,
      "step": 159689
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.1425222158432007,
      "learning_rate": 0.0001289743503144034,
      "loss": 2.1803,
      "step": 159690
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.1102607250213623,
      "learning_rate": 0.00012897098958425724,
      "loss": 2.0786,
      "step": 159691
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.1690826416015625,
      "learning_rate": 0.0001289676288859087,
      "loss": 2.1863,
      "step": 159692
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.1980212926864624,
      "learning_rate": 0.00012896426821935837,
      "loss": 2.0077,
      "step": 159693
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.1852484941482544,
      "learning_rate": 0.00012896090758460674,
      "loss": 2.2123,
      "step": 159694
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.2115174531936646,
      "learning_rate": 0.00012895754698165443,
      "loss": 2.0436,
      "step": 159695
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.1678012609481812,
      "learning_rate": 0.00012895418641050224,
      "loss": 1.9947,
      "step": 159696
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.1392050981521606,
      "learning_rate": 0.00012895082587115056,
      "loss": 1.9019,
      "step": 159697
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.2331138849258423,
      "learning_rate": 0.0001289474653636003,
      "loss": 1.8231,
      "step": 159698
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.1447765827178955,
      "learning_rate": 0.00012894410488785193,
      "loss": 2.0813,
      "step": 159699
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.1461960077285767,
      "learning_rate": 0.0001289407444439061,
      "loss": 2.0271,
      "step": 159700
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.2489837408065796,
      "learning_rate": 0.0001289373840317633,
      "loss": 1.8892,
      "step": 159701
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.1358399391174316,
      "learning_rate": 0.00012893402365142433,
      "loss": 2.2353,
      "step": 159702
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.1466633081436157,
      "learning_rate": 0.00012893066330288967,
      "loss": 2.0307,
      "step": 159703
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.1781017780303955,
      "learning_rate": 0.00012892730298616015,
      "loss": 2.0349,
      "step": 159704
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.106951355934143,
      "learning_rate": 0.00012892394270123624,
      "loss": 1.8568,
      "step": 159705
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.290797233581543,
      "learning_rate": 0.00012892058244811852,
      "loss": 2.0801,
      "step": 159706
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.2556811571121216,
      "learning_rate": 0.00012891722222680782,
      "loss": 1.7961,
      "step": 159707
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.5989282131195068,
      "learning_rate": 0.0001289138620373046,
      "loss": 1.9777,
      "step": 159708
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.3766429424285889,
      "learning_rate": 0.00012891050187960942,
      "loss": 1.9795,
      "step": 159709
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.187079668045044,
      "learning_rate": 0.00012890714175372312,
      "loss": 2.025,
      "step": 159710
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.3037853240966797,
      "learning_rate": 0.0001289037816596462,
      "loss": 2.3103,
      "step": 159711
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.1541097164154053,
      "learning_rate": 0.00012890042159737923,
      "loss": 2.0981,
      "step": 159712
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.2449432611465454,
      "learning_rate": 0.000128897061566923,
      "loss": 2.061,
      "step": 159713
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.2647297382354736,
      "learning_rate": 0.00012889370156827801,
      "loss": 1.8812,
      "step": 159714
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.2400434017181396,
      "learning_rate": 0.00012889034160144482,
      "loss": 2.055,
      "step": 159715
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.0998010635375977,
      "learning_rate": 0.00012888698166642424,
      "loss": 2.0217,
      "step": 159716
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.2244318723678589,
      "learning_rate": 0.00012888362176321678,
      "loss": 2.0104,
      "step": 159717
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.1609218120574951,
      "learning_rate": 0.000128880261891823,
      "loss": 1.9911,
      "step": 159718
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.3235855102539062,
      "learning_rate": 0.00012887690205224376,
      "loss": 2.0264,
      "step": 159719
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.0777562856674194,
      "learning_rate": 0.00012887354224447935,
      "loss": 1.9323,
      "step": 159720
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.1316196918487549,
      "learning_rate": 0.00012887018246853074,
      "loss": 1.9726,
      "step": 159721
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.2030608654022217,
      "learning_rate": 0.00012886682272439837,
      "loss": 2.0287,
      "step": 159722
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.2090249061584473,
      "learning_rate": 0.00012886346301208291,
      "loss": 2.018,
      "step": 159723
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.3080157041549683,
      "learning_rate": 0.00012886010333158483,
      "loss": 2.037,
      "step": 159724
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.2796801328659058,
      "learning_rate": 0.00012885674368290503,
      "loss": 2.2672,
      "step": 159725
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.2392545938491821,
      "learning_rate": 0.00012885338406604384,
      "loss": 2.198,
      "step": 159726
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.168276071548462,
      "learning_rate": 0.00012885002448100215,
      "loss": 2.0862,
      "step": 159727
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.394223690032959,
      "learning_rate": 0.00012884666492778049,
      "loss": 2.0833,
      "step": 159728
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.152343988418579,
      "learning_rate": 0.00012884330540637945,
      "loss": 2.0943,
      "step": 159729
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.1120706796646118,
      "learning_rate": 0.00012883994591679956,
      "loss": 1.8383,
      "step": 159730
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.1385990381240845,
      "learning_rate": 0.0001288365864590417,
      "loss": 2.0299,
      "step": 159731
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.1364353895187378,
      "learning_rate": 0.00012883322703310623,
      "loss": 1.912,
      "step": 159732
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.3502781391143799,
      "learning_rate": 0.00012882986763899397,
      "loss": 2.0294,
      "step": 159733
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.2569514513015747,
      "learning_rate": 0.00012882650827670538,
      "loss": 2.0751,
      "step": 159734
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.1563403606414795,
      "learning_rate": 0.00012882314894624138,
      "loss": 1.9983,
      "step": 159735
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.0694140195846558,
      "learning_rate": 0.00012881978964760218,
      "loss": 1.9474,
      "step": 159736
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.2360541820526123,
      "learning_rate": 0.00012881643038078873,
      "loss": 2.1,
      "step": 159737
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.1462279558181763,
      "learning_rate": 0.00012881307114580144,
      "loss": 2.0948,
      "step": 159738
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.1789849996566772,
      "learning_rate": 0.0001288097119426411,
      "loss": 2.1083,
      "step": 159739
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.3094534873962402,
      "learning_rate": 0.0001288063527713082,
      "loss": 2.1114,
      "step": 159740
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.3132543563842773,
      "learning_rate": 0.0001288029936318035,
      "loss": 2.1552,
      "step": 159741
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.240604281425476,
      "learning_rate": 0.00012879963452412755,
      "loss": 2.1098,
      "step": 159742
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.2214289903640747,
      "learning_rate": 0.00012879627544828102,
      "loss": 2.2476,
      "step": 159743
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.174999475479126,
      "learning_rate": 0.00012879291640426437,
      "loss": 2.1021,
      "step": 159744
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.3045048713684082,
      "learning_rate": 0.00012878955739207846,
      "loss": 2.1051,
      "step": 159745
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.1784305572509766,
      "learning_rate": 0.0001287861984117237,
      "loss": 1.9433,
      "step": 159746
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.3228436708450317,
      "learning_rate": 0.00012878283946320089,
      "loss": 2.1551,
      "step": 159747
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.2937371730804443,
      "learning_rate": 0.0001287794805465106,
      "loss": 2.1346,
      "step": 159748
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.1591870784759521,
      "learning_rate": 0.00012877612166165345,
      "loss": 2.0194,
      "step": 159749
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.156943917274475,
      "learning_rate": 0.00012877276280862993,
      "loss": 1.9737,
      "step": 159750
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.2669004201889038,
      "learning_rate": 0.00012876940398744093,
      "loss": 1.8368,
      "step": 159751
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.1485655307769775,
      "learning_rate": 0.00012876604519808678,
      "loss": 2.2278,
      "step": 159752
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.3606822490692139,
      "learning_rate": 0.00012876268644056835,
      "loss": 1.812,
      "step": 159753
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.2398252487182617,
      "learning_rate": 0.00012875932771488623,
      "loss": 2.1038,
      "step": 159754
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.208027720451355,
      "learning_rate": 0.00012875596902104094,
      "loss": 1.9653,
      "step": 159755
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.1895304918289185,
      "learning_rate": 0.00012875261035903303,
      "loss": 2.0995,
      "step": 159756
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.2137792110443115,
      "learning_rate": 0.0001287492517288634,
      "loss": 1.8726,
      "step": 159757
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.289939284324646,
      "learning_rate": 0.00012874589313053233,
      "loss": 2.0819,
      "step": 159758
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.3355261087417603,
      "learning_rate": 0.00012874253456404081,
      "loss": 2.1008,
      "step": 159759
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.3407386541366577,
      "learning_rate": 0.00012873917602938925,
      "loss": 1.8253,
      "step": 159760
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.2243218421936035,
      "learning_rate": 0.00012873581752657833,
      "loss": 2.0173,
      "step": 159761
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.286199927330017,
      "learning_rate": 0.00012873245905560855,
      "loss": 1.8872,
      "step": 159762
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.155676007270813,
      "learning_rate": 0.0001287291006164807,
      "loss": 1.9615,
      "step": 159763
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.2059216499328613,
      "learning_rate": 0.00012872574220919528,
      "loss": 1.8065,
      "step": 159764
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.309259295463562,
      "learning_rate": 0.00012872238383375306,
      "loss": 2.0802,
      "step": 159765
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.1902258396148682,
      "learning_rate": 0.00012871902549015462,
      "loss": 2.3375,
      "step": 159766
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.2485533952713013,
      "learning_rate": 0.0001287156671784005,
      "loss": 1.9273,
      "step": 159767
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.1928929090499878,
      "learning_rate": 0.0001287123088984913,
      "loss": 2.3098,
      "step": 159768
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.0975909233093262,
      "learning_rate": 0.0001287089506504278,
      "loss": 1.9686,
      "step": 159769
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.2724182605743408,
      "learning_rate": 0.00012870559243421042,
      "loss": 2.0761,
      "step": 159770
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.3704922199249268,
      "learning_rate": 0.00012870223424984005,
      "loss": 2.0078,
      "step": 159771
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.2024271488189697,
      "learning_rate": 0.00012869887609731715,
      "loss": 1.9991,
      "step": 159772
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.2749818563461304,
      "learning_rate": 0.00012869551797664236,
      "loss": 1.8561,
      "step": 159773
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.3463507890701294,
      "learning_rate": 0.0001286921598878162,
      "loss": 1.9738,
      "step": 159774
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.1277707815170288,
      "learning_rate": 0.0001286888018308395,
      "loss": 2.0213,
      "step": 159775
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.3568247556686401,
      "learning_rate": 0.0001286854438057127,
      "loss": 1.9649,
      "step": 159776
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.1698179244995117,
      "learning_rate": 0.0001286820858124366,
      "loss": 2.0234,
      "step": 159777
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.1265199184417725,
      "learning_rate": 0.00012867872785101176,
      "loss": 2.1192,
      "step": 159778
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.378464937210083,
      "learning_rate": 0.00012867536992143875,
      "loss": 1.8736,
      "step": 159779
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.1939560174942017,
      "learning_rate": 0.0001286720120237181,
      "loss": 1.9591,
      "step": 159780
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.2180149555206299,
      "learning_rate": 0.0001286686541578507,
      "loss": 1.9824,
      "step": 159781
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.3034950494766235,
      "learning_rate": 0.00012866529632383692,
      "loss": 2.0911,
      "step": 159782
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.2354094982147217,
      "learning_rate": 0.00012866193852167758,
      "loss": 2.2058,
      "step": 159783
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.1595430374145508,
      "learning_rate": 0.00012865858075137326,
      "loss": 2.1265,
      "step": 159784
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.108752965927124,
      "learning_rate": 0.00012865522301292451,
      "loss": 1.8279,
      "step": 159785
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.3341093063354492,
      "learning_rate": 0.00012865186530633188,
      "loss": 2.2874,
      "step": 159786
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.1903012990951538,
      "learning_rate": 0.00012864850763159621,
      "loss": 2.2049,
      "step": 159787
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.3110591173171997,
      "learning_rate": 0.0001286451499887179,
      "loss": 2.2881,
      "step": 159788
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.2700316905975342,
      "learning_rate": 0.00012864179237769783,
      "loss": 2.0553,
      "step": 159789
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.428443193435669,
      "learning_rate": 0.00012863843479853651,
      "loss": 2.1341,
      "step": 159790
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.1494240760803223,
      "learning_rate": 0.00012863507725123448,
      "loss": 2.2527,
      "step": 159791
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.1708776950836182,
      "learning_rate": 0.00012863171973579235,
      "loss": 2.0606,
      "step": 159792
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.2142878770828247,
      "learning_rate": 0.00012862836225221088,
      "loss": 2.3355,
      "step": 159793
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.167634129524231,
      "learning_rate": 0.0001286250048004906,
      "loss": 2.1499,
      "step": 159794
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.0793546438217163,
      "learning_rate": 0.00012862164738063222,
      "loss": 2.097,
      "step": 159795
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.323643684387207,
      "learning_rate": 0.00012861828999263634,
      "loss": 2.126,
      "step": 159796
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.259604811668396,
      "learning_rate": 0.00012861493263650344,
      "loss": 2.0567,
      "step": 159797
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.3180382251739502,
      "learning_rate": 0.00012861157531223434,
      "loss": 2.1014,
      "step": 159798
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.2506053447723389,
      "learning_rate": 0.0001286082180198296,
      "loss": 2.1246,
      "step": 159799
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.188991665840149,
      "learning_rate": 0.0001286048607592897,
      "loss": 2.3207,
      "step": 159800
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.1708821058273315,
      "learning_rate": 0.00012860150353061553,
      "loss": 2.1195,
      "step": 159801
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.1302595138549805,
      "learning_rate": 0.00012859814633380757,
      "loss": 1.9541,
      "step": 159802
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.1882655620574951,
      "learning_rate": 0.00012859478916886632,
      "loss": 2.0575,
      "step": 159803
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.294020175933838,
      "learning_rate": 0.0001285914320357927,
      "loss": 2.0808,
      "step": 159804
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.4041823148727417,
      "learning_rate": 0.000128588074934587,
      "loss": 1.974,
      "step": 159805
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.2478244304656982,
      "learning_rate": 0.00012858471786525015,
      "loss": 1.8369,
      "step": 159806
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.3314685821533203,
      "learning_rate": 0.00012858136082778265,
      "loss": 1.9127,
      "step": 159807
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.2820674180984497,
      "learning_rate": 0.00012857800382218496,
      "loss": 1.9692,
      "step": 159808
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.2233343124389648,
      "learning_rate": 0.00012857464684845797,
      "loss": 1.9955,
      "step": 159809
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.6397336721420288,
      "learning_rate": 0.0001285712899066022,
      "loss": 2.2097,
      "step": 159810
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.1059969663619995,
      "learning_rate": 0.0001285679329966182,
      "loss": 1.93,
      "step": 159811
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.735722303390503,
      "learning_rate": 0.0001285645761185067,
      "loss": 2.0873,
      "step": 159812
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.346580982208252,
      "learning_rate": 0.0001285612192722683,
      "loss": 2.1815,
      "step": 159813
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.3416194915771484,
      "learning_rate": 0.00012855786245790352,
      "loss": 2.2022,
      "step": 159814
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.2976419925689697,
      "learning_rate": 0.00012855450567541315,
      "loss": 2.0254,
      "step": 159815
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.2315737009048462,
      "learning_rate": 0.00012855114892479776,
      "loss": 1.9698,
      "step": 159816
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.6150879859924316,
      "learning_rate": 0.00012854779220605782,
      "loss": 1.9339,
      "step": 159817
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.759075403213501,
      "learning_rate": 0.0001285444355191942,
      "loss": 2.2551,
      "step": 159818
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.2384412288665771,
      "learning_rate": 0.0001285410788642073,
      "loss": 2.0683,
      "step": 159819
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.0856163501739502,
      "learning_rate": 0.00012853772224109792,
      "loss": 1.9393,
      "step": 159820
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.2599287033081055,
      "learning_rate": 0.00012853436564986667,
      "loss": 1.9605,
      "step": 159821
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.459790825843811,
      "learning_rate": 0.00012853100909051407,
      "loss": 1.9916,
      "step": 159822
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.414473295211792,
      "learning_rate": 0.0001285276525630407,
      "loss": 2.0226,
      "step": 159823
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.1545734405517578,
      "learning_rate": 0.00012852429606744742,
      "loss": 1.7804,
      "step": 159824
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.1805715560913086,
      "learning_rate": 0.00012852093960373455,
      "loss": 2.0114,
      "step": 159825
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.2942053079605103,
      "learning_rate": 0.000128517583171903,
      "loss": 2.2129,
      "step": 159826
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.2180237770080566,
      "learning_rate": 0.00012851422677195328,
      "loss": 2.1365,
      "step": 159827
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.401620626449585,
      "learning_rate": 0.00012851087040388598,
      "loss": 2.1346,
      "step": 159828
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.2820781469345093,
      "learning_rate": 0.0001285075140677017,
      "loss": 1.9469,
      "step": 159829
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.1090826988220215,
      "learning_rate": 0.00012850415776340113,
      "loss": 1.9101,
      "step": 159830
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.554487943649292,
      "learning_rate": 0.0001285008014909848,
      "loss": 1.9083,
      "step": 159831
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.1191431283950806,
      "learning_rate": 0.00012849744525045354,
      "loss": 1.9377,
      "step": 159832
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.1795142889022827,
      "learning_rate": 0.00012849408904180786,
      "loss": 2.2155,
      "step": 159833
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.2671241760253906,
      "learning_rate": 0.0001284907328650483,
      "loss": 1.9982,
      "step": 159834
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.171830415725708,
      "learning_rate": 0.0001284873767201755,
      "loss": 1.9634,
      "step": 159835
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.2676385641098022,
      "learning_rate": 0.00012848402060719021,
      "loss": 2.1246,
      "step": 159836
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.3135976791381836,
      "learning_rate": 0.0001284806645260929,
      "loss": 2.0565,
      "step": 159837
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.3444690704345703,
      "learning_rate": 0.0001284773084768843,
      "loss": 2.008,
      "step": 159838
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.1311333179473877,
      "learning_rate": 0.0001284739524595651,
      "loss": 2.1074,
      "step": 159839
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.2392553091049194,
      "learning_rate": 0.00012847059647413578,
      "loss": 1.7937,
      "step": 159840
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.2390626668930054,
      "learning_rate": 0.0001284672405205969,
      "loss": 2.1916,
      "step": 159841
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.2364662885665894,
      "learning_rate": 0.00012846388459894934,
      "loss": 2.0311,
      "step": 159842
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.3189220428466797,
      "learning_rate": 0.00012846052870919345,
      "loss": 1.932,
      "step": 159843
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.1036312580108643,
      "learning_rate": 0.00012845717285133013,
      "loss": 2.3361,
      "step": 159844
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.2065407037734985,
      "learning_rate": 0.00012845381702535983,
      "loss": 2.2707,
      "step": 159845
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.3025121688842773,
      "learning_rate": 0.0001284504612312832,
      "loss": 2.2096,
      "step": 159846
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.6841161251068115,
      "learning_rate": 0.00012844710546910078,
      "loss": 1.9104,
      "step": 159847
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.2781449556350708,
      "learning_rate": 0.00012844374973881335,
      "loss": 2.1322,
      "step": 159848
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.180704951286316,
      "learning_rate": 0.0001284403940404214,
      "loss": 1.8989,
      "step": 159849
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.371579647064209,
      "learning_rate": 0.00012843703837392572,
      "loss": 2.203,
      "step": 159850
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.5201871395111084,
      "learning_rate": 0.00012843368273932683,
      "loss": 2.1991,
      "step": 159851
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.161690354347229,
      "learning_rate": 0.0001284303271366253,
      "loss": 2.0575,
      "step": 159852
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.2001463174819946,
      "learning_rate": 0.00012842697156582178,
      "loss": 2.0462,
      "step": 159853
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.165518045425415,
      "learning_rate": 0.000128423616026917,
      "loss": 2.2187,
      "step": 159854
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.235591173171997,
      "learning_rate": 0.0001284202605199114,
      "loss": 1.8948,
      "step": 159855
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.211187481880188,
      "learning_rate": 0.00012841690504480585,
      "loss": 1.8917,
      "step": 159856
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.1038997173309326,
      "learning_rate": 0.00012841354960160082,
      "loss": 2.0614,
      "step": 159857
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.128204584121704,
      "learning_rate": 0.00012841019419029695,
      "loss": 2.0161,
      "step": 159858
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.3018492460250854,
      "learning_rate": 0.00012840683881089473,
      "loss": 2.2567,
      "step": 159859
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.1152828931808472,
      "learning_rate": 0.00012840348346339505,
      "loss": 2.1235,
      "step": 159860
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.1890772581100464,
      "learning_rate": 0.00012840012814779828,
      "loss": 2.1258,
      "step": 159861
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.2292579412460327,
      "learning_rate": 0.00012839677286410528,
      "loss": 1.9749,
      "step": 159862
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.198487639427185,
      "learning_rate": 0.00012839341761231656,
      "loss": 2.0123,
      "step": 159863
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.2702351808547974,
      "learning_rate": 0.00012839006239243276,
      "loss": 2.1865,
      "step": 159864
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.1597408056259155,
      "learning_rate": 0.00012838670720445434,
      "loss": 2.1226,
      "step": 159865
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.1747794151306152,
      "learning_rate": 0.0001283833520483822,
      "loss": 1.9656,
      "step": 159866
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.2686842679977417,
      "learning_rate": 0.00012837999692421673,
      "loss": 1.8622,
      "step": 159867
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.482521891593933,
      "learning_rate": 0.00012837664183195876,
      "loss": 2.135,
      "step": 159868
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.3520309925079346,
      "learning_rate": 0.00012837328677160883,
      "loss": 2.1184,
      "step": 159869
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.1555278301239014,
      "learning_rate": 0.00012836993174316751,
      "loss": 2.238,
      "step": 159870
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.1421018838882446,
      "learning_rate": 0.0001283665767466354,
      "loss": 1.9279,
      "step": 159871
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.190279245376587,
      "learning_rate": 0.00012836322178201324,
      "loss": 1.9954,
      "step": 159872
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.2183260917663574,
      "learning_rate": 0.00012835986684930155,
      "loss": 2.0393,
      "step": 159873
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.2799224853515625,
      "learning_rate": 0.00012835651194850105,
      "loss": 1.9671,
      "step": 159874
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.1999751329421997,
      "learning_rate": 0.00012835315707961223,
      "loss": 2.1712,
      "step": 159875
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.1924852132797241,
      "learning_rate": 0.00012834980224263602,
      "loss": 2.0154,
      "step": 159876
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.126888394355774,
      "learning_rate": 0.00012834644743757258,
      "loss": 1.9519,
      "step": 159877
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.1484743356704712,
      "learning_rate": 0.0001283430926644229,
      "loss": 1.9313,
      "step": 159878
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.1984378099441528,
      "learning_rate": 0.0001283397379231874,
      "loss": 1.9331,
      "step": 159879
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.0915250778198242,
      "learning_rate": 0.00012833638321386686,
      "loss": 1.9234,
      "step": 159880
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.1471138000488281,
      "learning_rate": 0.00012833302853646172,
      "loss": 2.0209,
      "step": 159881
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.1690714359283447,
      "learning_rate": 0.0001283296738909728,
      "loss": 2.0694,
      "step": 159882
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.0914103984832764,
      "learning_rate": 0.00012832631927740066,
      "loss": 1.9363,
      "step": 159883
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.2033756971359253,
      "learning_rate": 0.00012832296469574592,
      "loss": 2.2081,
      "step": 159884
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.2454160451889038,
      "learning_rate": 0.00012831961014600902,
      "loss": 1.9433,
      "step": 159885
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.2419922351837158,
      "learning_rate": 0.0001283162556281909,
      "loss": 2.2605,
      "step": 159886
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.1499688625335693,
      "learning_rate": 0.0001283129011422919,
      "loss": 1.9323,
      "step": 159887
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.2581309080123901,
      "learning_rate": 0.00012830954668831287,
      "loss": 2.0299,
      "step": 159888
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.2770605087280273,
      "learning_rate": 0.00012830619226625434,
      "loss": 2.07,
      "step": 159889
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.2670719623565674,
      "learning_rate": 0.00012830283787611686,
      "loss": 2.1116,
      "step": 159890
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.0978561639785767,
      "learning_rate": 0.00012829948351790119,
      "loss": 1.9589,
      "step": 159891
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.1486730575561523,
      "learning_rate": 0.0001282961291916079,
      "loss": 1.9786,
      "step": 159892
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.140220284461975,
      "learning_rate": 0.00012829277489723748,
      "loss": 1.7818,
      "step": 159893
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.3121615648269653,
      "learning_rate": 0.00012828942063479077,
      "loss": 2.1872,
      "step": 159894
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.4519625902175903,
      "learning_rate": 0.00012828606640426834,
      "loss": 2.1001,
      "step": 159895
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.3252956867218018,
      "learning_rate": 0.00012828271220567062,
      "loss": 2.0561,
      "step": 159896
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.1424890756607056,
      "learning_rate": 0.00012827935803899856,
      "loss": 2.0492,
      "step": 159897
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.1686129570007324,
      "learning_rate": 0.00012827600390425255,
      "loss": 2.1672,
      "step": 159898
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.3947350978851318,
      "learning_rate": 0.0001282726498014332,
      "loss": 2.0216,
      "step": 159899
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.2914316654205322,
      "learning_rate": 0.00012826929573054127,
      "loss": 2.1079,
      "step": 159900
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.25800359249115,
      "learning_rate": 0.00012826594169157736,
      "loss": 2.0204,
      "step": 159901
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.2253518104553223,
      "learning_rate": 0.00012826258768454193,
      "loss": 1.9059,
      "step": 159902
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.2218196392059326,
      "learning_rate": 0.00012825923370943582,
      "loss": 2.114,
      "step": 159903
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.2142322063446045,
      "learning_rate": 0.00012825587976625946,
      "loss": 2.1353,
      "step": 159904
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.2088688611984253,
      "learning_rate": 0.0001282525258550137,
      "loss": 2.3313,
      "step": 159905
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.2138786315917969,
      "learning_rate": 0.00012824917197569902,
      "loss": 2.0659,
      "step": 159906
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.2056548595428467,
      "learning_rate": 0.00012824581812831607,
      "loss": 2.0136,
      "step": 159907
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.2476121187210083,
      "learning_rate": 0.00012824246431286535,
      "loss": 2.1233,
      "step": 159908
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.2142863273620605,
      "learning_rate": 0.0001282391105293477,
      "loss": 1.9096,
      "step": 159909
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.2346271276474,
      "learning_rate": 0.00012823575677776356,
      "loss": 2.1217,
      "step": 159910
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.1633696556091309,
      "learning_rate": 0.00012823240305811373,
      "loss": 2.2262,
      "step": 159911
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.3330328464508057,
      "learning_rate": 0.00012822904937039874,
      "loss": 2.1435,
      "step": 159912
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.3614133596420288,
      "learning_rate": 0.0001282256957146192,
      "loss": 2.0561,
      "step": 159913
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.2270560264587402,
      "learning_rate": 0.00012822234209077565,
      "loss": 1.8939,
      "step": 159914
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.1671117544174194,
      "learning_rate": 0.0001282189884988689,
      "loss": 1.9801,
      "step": 159915
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.3411338329315186,
      "learning_rate": 0.0001282156349388994,
      "loss": 1.9401,
      "step": 159916
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.2814806699752808,
      "learning_rate": 0.00012821228141086794,
      "loss": 2.0629,
      "step": 159917
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.1397982835769653,
      "learning_rate": 0.0001282089279147751,
      "loss": 2.0108,
      "step": 159918
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.1535906791687012,
      "learning_rate": 0.00012820557445062143,
      "loss": 1.9869,
      "step": 159919
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.0818778276443481,
      "learning_rate": 0.0001282022210184075,
      "loss": 2.1715,
      "step": 159920
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.2961214780807495,
      "learning_rate": 0.00012819886761813407,
      "loss": 2.1877,
      "step": 159921
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.1237579584121704,
      "learning_rate": 0.00012819551424980168,
      "loss": 2.0526,
      "step": 159922
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.172516107559204,
      "learning_rate": 0.00012819216091341102,
      "loss": 1.8067,
      "step": 159923
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.1164798736572266,
      "learning_rate": 0.00012818880760896275,
      "loss": 2.0397,
      "step": 159924
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.1545345783233643,
      "learning_rate": 0.00012818545433645738,
      "loss": 2.046,
      "step": 159925
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.2180047035217285,
      "learning_rate": 0.0001281821010958955,
      "loss": 2.0001,
      "step": 159926
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.1611618995666504,
      "learning_rate": 0.00012817874788727788,
      "loss": 2.1374,
      "step": 159927
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.2325519323349,
      "learning_rate": 0.000128175394710605,
      "loss": 2.0282,
      "step": 159928
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.3532650470733643,
      "learning_rate": 0.00012817204156587766,
      "loss": 1.9609,
      "step": 159929
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.198374629020691,
      "learning_rate": 0.00012816868845309638,
      "loss": 1.9159,
      "step": 159930
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.2330448627471924,
      "learning_rate": 0.00012816533537226174,
      "loss": 1.8714,
      "step": 159931
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.2937694787979126,
      "learning_rate": 0.00012816198232337437,
      "loss": 2.2768,
      "step": 159932
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.1664336919784546,
      "learning_rate": 0.00012815862930643502,
      "loss": 1.8704,
      "step": 159933
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.0994082689285278,
      "learning_rate": 0.0001281552763214441,
      "loss": 2.1136,
      "step": 159934
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.1749730110168457,
      "learning_rate": 0.00012815192336840248,
      "loss": 2.1083,
      "step": 159935
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.2899900674819946,
      "learning_rate": 0.00012814857044731064,
      "loss": 2.0645,
      "step": 159936
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.0982677936553955,
      "learning_rate": 0.00012814521755816923,
      "loss": 1.9202,
      "step": 159937
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.262208342552185,
      "learning_rate": 0.00012814186470097875,
      "loss": 2.0491,
      "step": 159938
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.2906392812728882,
      "learning_rate": 0.0001281385118757401,
      "loss": 1.9999,
      "step": 159939
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.2217737436294556,
      "learning_rate": 0.0001281351590824536,
      "loss": 2.2804,
      "step": 159940
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.2925388813018799,
      "learning_rate": 0.0001281318063211201,
      "loss": 2.0248,
      "step": 159941
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.0971626043319702,
      "learning_rate": 0.0001281284535917401,
      "loss": 2.1974,
      "step": 159942
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.326420545578003,
      "learning_rate": 0.00012812510089431443,
      "loss": 2.0981,
      "step": 159943
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.1900051832199097,
      "learning_rate": 0.00012812174822884336,
      "loss": 1.9307,
      "step": 159944
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.3122327327728271,
      "learning_rate": 0.00012811839559532777,
      "loss": 2.0777,
      "step": 159945
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.2174488306045532,
      "learning_rate": 0.00012811504299376815,
      "loss": 1.8946,
      "step": 159946
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.2176055908203125,
      "learning_rate": 0.00012811169042416528,
      "loss": 2.0365,
      "step": 159947
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.2258622646331787,
      "learning_rate": 0.00012810833788651959,
      "loss": 2.0793,
      "step": 159948
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.1649338006973267,
      "learning_rate": 0.000128104985380832,
      "loss": 1.9655,
      "step": 159949
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.2424545288085938,
      "learning_rate": 0.00012810163290710274,
      "loss": 2.0808,
      "step": 159950
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.2058546543121338,
      "learning_rate": 0.00012809828046533275,
      "loss": 1.8536,
      "step": 159951
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.2779253721237183,
      "learning_rate": 0.0001280949280555224,
      "loss": 2.1116,
      "step": 159952
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.2471699714660645,
      "learning_rate": 0.00012809157567767254,
      "loss": 1.922,
      "step": 159953
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.123736023902893,
      "learning_rate": 0.00012808822333178362,
      "loss": 1.87,
      "step": 159954
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.654094934463501,
      "learning_rate": 0.00012808487101785657,
      "loss": 2.1182,
      "step": 159955
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.2622919082641602,
      "learning_rate": 0.00012808151873589159,
      "loss": 2.1747,
      "step": 159956
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.2146519422531128,
      "learning_rate": 0.00012807816648588956,
      "loss": 2.1506,
      "step": 159957
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.2757959365844727,
      "learning_rate": 0.00012807481426785095,
      "loss": 1.9378,
      "step": 159958
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.270119309425354,
      "learning_rate": 0.0001280714620817766,
      "loss": 1.9703,
      "step": 159959
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.1979702711105347,
      "learning_rate": 0.00012806810992766688,
      "loss": 2.239,
      "step": 159960
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.1732829809188843,
      "learning_rate": 0.00012806475780552279,
      "loss": 2.0663,
      "step": 159961
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.2596855163574219,
      "learning_rate": 0.00012806140571534446,
      "loss": 2.1185,
      "step": 159962
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.230486512184143,
      "learning_rate": 0.0001280580536571329,
      "loss": 2.0167,
      "step": 159963
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.2422165870666504,
      "learning_rate": 0.00012805470163088846,
      "loss": 1.9601,
      "step": 159964
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.0533764362335205,
      "learning_rate": 0.00012805134963661202,
      "loss": 1.8914,
      "step": 159965
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.1379767656326294,
      "learning_rate": 0.00012804799767430396,
      "loss": 2.0969,
      "step": 159966
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.2821464538574219,
      "learning_rate": 0.00012804464574396515,
      "loss": 2.131,
      "step": 159967
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.1898037195205688,
      "learning_rate": 0.00012804129384559607,
      "loss": 2.003,
      "step": 159968
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.1113455295562744,
      "learning_rate": 0.00012803794197919737,
      "loss": 2.0618,
      "step": 159969
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.1221321821212769,
      "learning_rate": 0.00012803459014476952,
      "loss": 2.24,
      "step": 159970
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.1284027099609375,
      "learning_rate": 0.00012803123834231344,
      "loss": 2.0936,
      "step": 159971
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.1983580589294434,
      "learning_rate": 0.00012802788657182947,
      "loss": 2.1655,
      "step": 159972
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.2559205293655396,
      "learning_rate": 0.00012802453483331847,
      "loss": 2.2554,
      "step": 159973
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.2448492050170898,
      "learning_rate": 0.00012802118312678096,
      "loss": 2.1124,
      "step": 159974
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.3242299556732178,
      "learning_rate": 0.00012801783145221756,
      "loss": 2.185,
      "step": 159975
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.1971269845962524,
      "learning_rate": 0.00012801447980962877,
      "loss": 2.1526,
      "step": 159976
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.1584689617156982,
      "learning_rate": 0.00012801112819901545,
      "loss": 2.1307,
      "step": 159977
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.3087776899337769,
      "learning_rate": 0.000128007776620378,
      "loss": 2.0174,
      "step": 159978
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.2034016847610474,
      "learning_rate": 0.00012800442507371728,
      "loss": 2.1506,
      "step": 159979
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.0874712467193604,
      "learning_rate": 0.00012800107355903374,
      "loss": 2.262,
      "step": 159980
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.2510440349578857,
      "learning_rate": 0.00012799772207632798,
      "loss": 2.1586,
      "step": 159981
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.1528449058532715,
      "learning_rate": 0.00012799437062560077,
      "loss": 2.0185,
      "step": 159982
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.1497828960418701,
      "learning_rate": 0.00012799101920685268,
      "loss": 2.1436,
      "step": 159983
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.396647572517395,
      "learning_rate": 0.0001279876678200842,
      "loss": 2.0637,
      "step": 159984
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.1760871410369873,
      "learning_rate": 0.00012798431646529615,
      "loss": 1.8841,
      "step": 159985
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.2511752843856812,
      "learning_rate": 0.00012798096514248905,
      "loss": 1.9665,
      "step": 159986
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.051693081855774,
      "learning_rate": 0.00012797761385166344,
      "loss": 1.9727,
      "step": 159987
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.1704376935958862,
      "learning_rate": 0.00012797426259282012,
      "loss": 2.0648,
      "step": 159988
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.11698317527771,
      "learning_rate": 0.00012797091136595957,
      "loss": 2.0862,
      "step": 159989
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.1749738454818726,
      "learning_rate": 0.00012796756017108258,
      "loss": 2.3191,
      "step": 159990
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.2238097190856934,
      "learning_rate": 0.00012796420900818963,
      "loss": 2.2476,
      "step": 159991
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.275434970855713,
      "learning_rate": 0.00012796085787728143,
      "loss": 1.9468,
      "step": 159992
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.229485034942627,
      "learning_rate": 0.00012795750677835842,
      "loss": 2.1611,
      "step": 159993
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.1662368774414062,
      "learning_rate": 0.00012795415571142146,
      "loss": 1.9487,
      "step": 159994
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.339975118637085,
      "learning_rate": 0.00012795080467647098,
      "loss": 1.978,
      "step": 159995
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.2013412714004517,
      "learning_rate": 0.00012794745367350775,
      "loss": 2.131,
      "step": 159996
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.2478071451187134,
      "learning_rate": 0.00012794410270253239,
      "loss": 2.2625,
      "step": 159997
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.2461528778076172,
      "learning_rate": 0.00012794075176354546,
      "loss": 2.0939,
      "step": 159998
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.2335976362228394,
      "learning_rate": 0.00012793740085654748,
      "loss": 2.1341,
      "step": 159999
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.1975266933441162,
      "learning_rate": 0.0001279340499815393,
      "loss": 1.9966,
      "step": 160000
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.1516212224960327,
      "learning_rate": 0.0001279306991385213,
      "loss": 2.1351,
      "step": 160001
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.1229050159454346,
      "learning_rate": 0.00012792734832749435,
      "loss": 2.055,
      "step": 160002
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.144397258758545,
      "learning_rate": 0.00012792399754845896,
      "loss": 2.2207,
      "step": 160003
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.236606240272522,
      "learning_rate": 0.00012792064680141574,
      "loss": 2.0518,
      "step": 160004
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.3024780750274658,
      "learning_rate": 0.00012791729608636519,
      "loss": 2.0065,
      "step": 160005
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.1790084838867188,
      "learning_rate": 0.00012791394540330822,
      "loss": 1.8111,
      "step": 160006
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.2154722213745117,
      "learning_rate": 0.00012791059475224513,
      "loss": 2.1673,
      "step": 160007
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.1698191165924072,
      "learning_rate": 0.00012790724413317688,
      "loss": 1.8713,
      "step": 160008
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.3941587209701538,
      "learning_rate": 0.0001279038935461038,
      "loss": 2.0298,
      "step": 160009
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.0842269659042358,
      "learning_rate": 0.0001279005429910268,
      "loss": 2.1982,
      "step": 160010
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.3151448965072632,
      "learning_rate": 0.00012789719246794615,
      "loss": 2.0582,
      "step": 160011
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.2413039207458496,
      "learning_rate": 0.0001278938419768628,
      "loss": 2.1142,
      "step": 160012
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.316375732421875,
      "learning_rate": 0.0001278904915177771,
      "loss": 2.3221,
      "step": 160013
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.2005596160888672,
      "learning_rate": 0.00012788714109068992,
      "loss": 1.9585,
      "step": 160014
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.23622465133667,
      "learning_rate": 0.00012788379069560165,
      "loss": 2.0847,
      "step": 160015
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.2418500185012817,
      "learning_rate": 0.00012788044033251324,
      "loss": 2.027,
      "step": 160016
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.1393955945968628,
      "learning_rate": 0.00012787709000142491,
      "loss": 1.8248,
      "step": 160017
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.2103081941604614,
      "learning_rate": 0.00012787373970233755,
      "loss": 2.1072,
      "step": 160018
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.0964457988739014,
      "learning_rate": 0.00012787038943525163,
      "loss": 2.1129,
      "step": 160019
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.2445549964904785,
      "learning_rate": 0.00012786703920016797,
      "loss": 2.2069,
      "step": 160020
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.4194047451019287,
      "learning_rate": 0.00012786368899708696,
      "loss": 1.9472,
      "step": 160021
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.278136968612671,
      "learning_rate": 0.00012786033882600958,
      "loss": 1.9514,
      "step": 160022
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.2566426992416382,
      "learning_rate": 0.00012785698868693595,
      "loss": 1.8615,
      "step": 160023
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.112320065498352,
      "learning_rate": 0.0001278536385798671,
      "loss": 1.7935,
      "step": 160024
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.1811858415603638,
      "learning_rate": 0.0001278502885048034,
      "loss": 1.9743,
      "step": 160025
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.2653310298919678,
      "learning_rate": 0.0001278469384617457,
      "loss": 2.1143,
      "step": 160026
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.1221989393234253,
      "learning_rate": 0.00012784358845069436,
      "loss": 2.0781,
      "step": 160027
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.096147894859314,
      "learning_rate": 0.00012784023847165035,
      "loss": 2.029,
      "step": 160028
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.172958254814148,
      "learning_rate": 0.00012783688852461392,
      "loss": 2.142,
      "step": 160029
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.2376439571380615,
      "learning_rate": 0.00012783353860958596,
      "loss": 1.773,
      "step": 160030
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.204606533050537,
      "learning_rate": 0.00012783018872656685,
      "loss": 2.0325,
      "step": 160031
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.1722806692123413,
      "learning_rate": 0.0001278268388755575,
      "loss": 2.2256,
      "step": 160032
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.1950117349624634,
      "learning_rate": 0.00012782348905655827,
      "loss": 1.8324,
      "step": 160033
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.2104347944259644,
      "learning_rate": 0.0001278201392695701,
      "loss": 2.3063,
      "step": 160034
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.1567409038543701,
      "learning_rate": 0.00012781678951459324,
      "loss": 2.1016,
      "step": 160035
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.260873794555664,
      "learning_rate": 0.00012781343979162857,
      "loss": 1.8933,
      "step": 160036
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.329551339149475,
      "learning_rate": 0.00012781009010067655,
      "loss": 2.107,
      "step": 160037
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.1265738010406494,
      "learning_rate": 0.00012780674044173795,
      "loss": 1.9222,
      "step": 160038
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.1242226362228394,
      "learning_rate": 0.00012780339081481327,
      "loss": 2.0763,
      "step": 160039
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.2227435111999512,
      "learning_rate": 0.0001278000412199034,
      "loss": 2.1027,
      "step": 160040
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.148239016532898,
      "learning_rate": 0.0001277966916570085,
      "loss": 1.8351,
      "step": 160041
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.2242316007614136,
      "learning_rate": 0.00012779334212612958,
      "loss": 2.1443,
      "step": 160042
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.2971954345703125,
      "learning_rate": 0.000127789992627267,
      "loss": 2.2244,
      "step": 160043
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.2961041927337646,
      "learning_rate": 0.00012778664316042163,
      "loss": 1.9976,
      "step": 160044
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.21011221408844,
      "learning_rate": 0.00012778329372559386,
      "loss": 2.0812,
      "step": 160045
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.2567143440246582,
      "learning_rate": 0.00012777994432278467,
      "loss": 1.7795,
      "step": 160046
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.2148151397705078,
      "learning_rate": 0.0001277765949519942,
      "loss": 2.1432,
      "step": 160047
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.278112769126892,
      "learning_rate": 0.00012777324561322342,
      "loss": 2.0592,
      "step": 160048
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.2041069269180298,
      "learning_rate": 0.00012776989630647274,
      "loss": 1.9926,
      "step": 160049
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.3039519786834717,
      "learning_rate": 0.000127766547031743,
      "loss": 2.3167,
      "step": 160050
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.2220855951309204,
      "learning_rate": 0.0001277631977890346,
      "loss": 1.9421,
      "step": 160051
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.494673490524292,
      "learning_rate": 0.0001277598485783484,
      "loss": 2.2696,
      "step": 160052
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.2064166069030762,
      "learning_rate": 0.00012775649939968486,
      "loss": 1.9443,
      "step": 160053
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.3018299341201782,
      "learning_rate": 0.00012775315025304466,
      "loss": 2.1997,
      "step": 160054
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.1106386184692383,
      "learning_rate": 0.00012774980113842827,
      "loss": 1.9277,
      "step": 160055
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.233824610710144,
      "learning_rate": 0.0001277464520558366,
      "loss": 1.9725,
      "step": 160056
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.3548305034637451,
      "learning_rate": 0.00012774310300526994,
      "loss": 1.9414,
      "step": 160057
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.4053720235824585,
      "learning_rate": 0.00012773975398672926,
      "loss": 2.1975,
      "step": 160058
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.406009554862976,
      "learning_rate": 0.00012773640500021494,
      "loss": 2.0142,
      "step": 160059
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.2689447402954102,
      "learning_rate": 0.00012773305604572776,
      "loss": 1.8522,
      "step": 160060
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.1550959348678589,
      "learning_rate": 0.00012772970712326807,
      "loss": 2.0342,
      "step": 160061
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.3484399318695068,
      "learning_rate": 0.00012772635823283686,
      "loss": 1.9062,
      "step": 160062
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.1556282043457031,
      "learning_rate": 0.00012772300937443439,
      "loss": 2.1119,
      "step": 160063
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.1311991214752197,
      "learning_rate": 0.00012771966054806163,
      "loss": 2.0927,
      "step": 160064
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.0979973077774048,
      "learning_rate": 0.00012771631175371903,
      "loss": 2.1159,
      "step": 160065
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.304807186126709,
      "learning_rate": 0.0001277129629914071,
      "loss": 2.0695,
      "step": 160066
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.614655613899231,
      "learning_rate": 0.00012770961426112666,
      "loss": 2.3598,
      "step": 160067
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.3053405284881592,
      "learning_rate": 0.0001277062655628783,
      "loss": 2.0918,
      "step": 160068
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.196531891822815,
      "learning_rate": 0.00012770291689666245,
      "loss": 2.0041,
      "step": 160069
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.2461626529693604,
      "learning_rate": 0.00012769956826248,
      "loss": 2.1398,
      "step": 160070
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.1515954732894897,
      "learning_rate": 0.0001276962196603315,
      "loss": 2.1187,
      "step": 160071
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.1837259531021118,
      "learning_rate": 0.0001276928710902174,
      "loss": 2.1751,
      "step": 160072
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.1847540140151978,
      "learning_rate": 0.0001276895225521385,
      "loss": 1.9749,
      "step": 160073
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.2622301578521729,
      "learning_rate": 0.00012768617404609542,
      "loss": 1.9821,
      "step": 160074
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.2518491744995117,
      "learning_rate": 0.0001276828255720886,
      "loss": 2.3164,
      "step": 160075
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.1783162355422974,
      "learning_rate": 0.00012767947713011892,
      "loss": 2.2594,
      "step": 160076
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.359398365020752,
      "learning_rate": 0.00012767612872018687,
      "loss": 2.1838,
      "step": 160077
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.233520746231079,
      "learning_rate": 0.000127672780342293,
      "loss": 2.1697,
      "step": 160078
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.2649880647659302,
      "learning_rate": 0.00012766943199643807,
      "loss": 1.9627,
      "step": 160079
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.1391395330429077,
      "learning_rate": 0.0001276660836826226,
      "loss": 1.9849,
      "step": 160080
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.3023066520690918,
      "learning_rate": 0.00012766273540084733,
      "loss": 2.1988,
      "step": 160081
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.1931170225143433,
      "learning_rate": 0.00012765938715111278,
      "loss": 1.9476,
      "step": 160082
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.2374491691589355,
      "learning_rate": 0.00012765603893341964,
      "loss": 2.0655,
      "step": 160083
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.2926796674728394,
      "learning_rate": 0.00012765269074776838,
      "loss": 2.1444,
      "step": 160084
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.2773874998092651,
      "learning_rate": 0.00012764934259415985,
      "loss": 2.0617,
      "step": 160085
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.1724841594696045,
      "learning_rate": 0.00012764599447259446,
      "loss": 2.1812,
      "step": 160086
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.31493079662323,
      "learning_rate": 0.00012764264638307304,
      "loss": 2.0604,
      "step": 160087
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.2169685363769531,
      "learning_rate": 0.00012763929832559597,
      "loss": 2.093,
      "step": 160088
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.202674388885498,
      "learning_rate": 0.00012763595030016424,
      "loss": 1.9846,
      "step": 160089
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.3164719343185425,
      "learning_rate": 0.00012763260230677805,
      "loss": 1.9189,
      "step": 160090
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.196885347366333,
      "learning_rate": 0.0001276292543454383,
      "loss": 2.0174,
      "step": 160091
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.157482385635376,
      "learning_rate": 0.0001276259064161454,
      "loss": 1.8969,
      "step": 160092
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.2923728227615356,
      "learning_rate": 0.00012762255851890024,
      "loss": 2.1818,
      "step": 160093
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.1379733085632324,
      "learning_rate": 0.00012761921065370315,
      "loss": 2.1453,
      "step": 160094
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.183763027191162,
      "learning_rate": 0.00012761586282055516,
      "loss": 2.0487,
      "step": 160095
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.1593055725097656,
      "learning_rate": 0.00012761251501945637,
      "loss": 2.1225,
      "step": 160096
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.3202910423278809,
      "learning_rate": 0.00012760916725040785,
      "loss": 2.1318,
      "step": 160097
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.459277629852295,
      "learning_rate": 0.00012760581951340986,
      "loss": 2.1423,
      "step": 160098
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.2876478433609009,
      "learning_rate": 0.00012760247180846338,
      "loss": 2.0916,
      "step": 160099
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.288086175918579,
      "learning_rate": 0.00012759912413556868,
      "loss": 1.9805,
      "step": 160100
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.3434098958969116,
      "learning_rate": 0.00012759577649472682,
      "loss": 2.2755,
      "step": 160101
    },
    {
      "epoch": 2.08,
      "grad_norm": 2.0945546627044678,
      "learning_rate": 0.0001275924288859379,
      "loss": 1.8893,
      "step": 160102
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.6781775951385498,
      "learning_rate": 0.00012758908130920292,
      "loss": 2.1862,
      "step": 160103
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.212032437324524,
      "learning_rate": 0.0001275857337645223,
      "loss": 2.1217,
      "step": 160104
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.1551755666732788,
      "learning_rate": 0.00012758238625189684,
      "loss": 2.0,
      "step": 160105
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.1719536781311035,
      "learning_rate": 0.00012757903877132698,
      "loss": 2.041,
      "step": 160106
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.7741965055465698,
      "learning_rate": 0.00012757569132281364,
      "loss": 2.0544,
      "step": 160107
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.5973528623580933,
      "learning_rate": 0.000127572343906357,
      "loss": 1.9793,
      "step": 160108
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.2082908153533936,
      "learning_rate": 0.00012756899652195805,
      "loss": 1.9331,
      "step": 160109
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.2323458194732666,
      "learning_rate": 0.00012756564916961713,
      "loss": 2.056,
      "step": 160110
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.2362581491470337,
      "learning_rate": 0.00012756230184933514,
      "loss": 2.0898,
      "step": 160111
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.7859660387039185,
      "learning_rate": 0.00012755895456111247,
      "loss": 2.0573,
      "step": 160112
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.3077011108398438,
      "learning_rate": 0.00012755560730495007,
      "loss": 2.2424,
      "step": 160113
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.19304621219635,
      "learning_rate": 0.0001275522600808481,
      "loss": 1.9498,
      "step": 160114
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.1468989849090576,
      "learning_rate": 0.00012754891288880758,
      "loss": 2.0681,
      "step": 160115
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.3837573528289795,
      "learning_rate": 0.0001275455657288288,
      "loss": 1.9873,
      "step": 160116
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.7584412097930908,
      "learning_rate": 0.00012754221860091267,
      "loss": 2.0518,
      "step": 160117
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.793164610862732,
      "learning_rate": 0.00012753887150505964,
      "loss": 1.9348,
      "step": 160118
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.3502098321914673,
      "learning_rate": 0.00012753552444127054,
      "loss": 2.1568,
      "step": 160119
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.1330022811889648,
      "learning_rate": 0.00012753217740954566,
      "loss": 2.0941,
      "step": 160120
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.2729239463806152,
      "learning_rate": 0.00012752883040988592,
      "loss": 1.9789,
      "step": 160121
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.5113015174865723,
      "learning_rate": 0.0001275254834422917,
      "loss": 2.031,
      "step": 160122
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.6259568929672241,
      "learning_rate": 0.00012752213650676388,
      "loss": 2.0323,
      "step": 160123
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.3148504495620728,
      "learning_rate": 0.0001275187896033028,
      "loss": 2.0482,
      "step": 160124
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.247807502746582,
      "learning_rate": 0.0001275154427319095,
      "loss": 2.0211,
      "step": 160125
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.1864278316497803,
      "learning_rate": 0.00012751209589258408,
      "loss": 1.7828,
      "step": 160126
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.816392183303833,
      "learning_rate": 0.00012750874908532757,
      "loss": 2.1732,
      "step": 160127
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.4756196737289429,
      "learning_rate": 0.00012750540231014026,
      "loss": 2.1217,
      "step": 160128
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.2717010974884033,
      "learning_rate": 0.0001275020555670231,
      "loss": 2.2236,
      "step": 160129
    },
    {
      "epoch": 2.08,
      "grad_norm": 1.1278700828552246,
      "learning_rate": 0.00012749870885597647,
      "loss": 1.9375,
      "step": 160130
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.2525073289871216,
      "learning_rate": 0.0001274953621770013,
      "loss": 2.0126,
      "step": 160131
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.3422085046768188,
      "learning_rate": 0.00012749201553009776,
      "loss": 1.9046,
      "step": 160132
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.2494486570358276,
      "learning_rate": 0.00012748866891526683,
      "loss": 2.0029,
      "step": 160133
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.199251651763916,
      "learning_rate": 0.0001274853223325089,
      "loss": 1.986,
      "step": 160134
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.2886559963226318,
      "learning_rate": 0.00012748197578182482,
      "loss": 2.0003,
      "step": 160135
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.1345893144607544,
      "learning_rate": 0.00012747862926321498,
      "loss": 2.0436,
      "step": 160136
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.1317949295043945,
      "learning_rate": 0.00012747528277668035,
      "loss": 2.0083,
      "step": 160137
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.2159180641174316,
      "learning_rate": 0.0001274719363222211,
      "loss": 1.9858,
      "step": 160138
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.1763386726379395,
      "learning_rate": 0.00012746858989983817,
      "loss": 2.2074,
      "step": 160139
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.1973950862884521,
      "learning_rate": 0.00012746524350953198,
      "loss": 1.9333,
      "step": 160140
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.3821314573287964,
      "learning_rate": 0.0001274618971513034,
      "loss": 1.8446,
      "step": 160141
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.2052005529403687,
      "learning_rate": 0.00012745855082515274,
      "loss": 2.3683,
      "step": 160142
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.1493629217147827,
      "learning_rate": 0.00012745520453108093,
      "loss": 1.9528,
      "step": 160143
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.260176181793213,
      "learning_rate": 0.00012745185826908847,
      "loss": 1.9955,
      "step": 160144
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.2904424667358398,
      "learning_rate": 0.00012744851203917598,
      "loss": 1.9831,
      "step": 160145
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.200225830078125,
      "learning_rate": 0.00012744516584134392,
      "loss": 2.0195,
      "step": 160146
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.9353572130203247,
      "learning_rate": 0.00012744181967559322,
      "loss": 2.0886,
      "step": 160147
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.2424236536026,
      "learning_rate": 0.00012743847354192416,
      "loss": 2.0211,
      "step": 160148
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.2740299701690674,
      "learning_rate": 0.00012743512744033773,
      "loss": 1.9647,
      "step": 160149
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.242794156074524,
      "learning_rate": 0.0001274317813708343,
      "loss": 2.1305,
      "step": 160150
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.2831883430480957,
      "learning_rate": 0.0001274284353334145,
      "loss": 2.233,
      "step": 160151
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.2940688133239746,
      "learning_rate": 0.0001274250893280791,
      "loss": 1.8612,
      "step": 160152
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.164306402206421,
      "learning_rate": 0.00012742174335482863,
      "loss": 1.9467,
      "step": 160153
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.1308270692825317,
      "learning_rate": 0.00012741839741366363,
      "loss": 2.1587,
      "step": 160154
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.2304795980453491,
      "learning_rate": 0.0001274150515045849,
      "loss": 2.1271,
      "step": 160155
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.1013360023498535,
      "learning_rate": 0.00012741170562759297,
      "loss": 1.7936,
      "step": 160156
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.2354201078414917,
      "learning_rate": 0.00012740835978268834,
      "loss": 1.8904,
      "step": 160157
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.1504746675491333,
      "learning_rate": 0.00012740501396987187,
      "loss": 1.9502,
      "step": 160158
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.2827601432800293,
      "learning_rate": 0.00012740166818914407,
      "loss": 2.3249,
      "step": 160159
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.0800386667251587,
      "learning_rate": 0.00012739832244050545,
      "loss": 2.1494,
      "step": 160160
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.1990801095962524,
      "learning_rate": 0.00012739497672395686,
      "loss": 2.0038,
      "step": 160161
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.2867512702941895,
      "learning_rate": 0.0001273916310394988,
      "loss": 2.0695,
      "step": 160162
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.3111944198608398,
      "learning_rate": 0.00012738828538713177,
      "loss": 2.1038,
      "step": 160163
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.074720025062561,
      "learning_rate": 0.00012738493976685664,
      "loss": 1.992,
      "step": 160164
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.2318302392959595,
      "learning_rate": 0.0001273815941786738,
      "loss": 2.128,
      "step": 160165
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.1954829692840576,
      "learning_rate": 0.00012737824862258407,
      "loss": 2.081,
      "step": 160166
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.1791189908981323,
      "learning_rate": 0.000127374903098588,
      "loss": 2.064,
      "step": 160167
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.1849137544631958,
      "learning_rate": 0.00012737155760668618,
      "loss": 1.8616,
      "step": 160168
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.3653229475021362,
      "learning_rate": 0.00012736821214687913,
      "loss": 2.0817,
      "step": 160169
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.1980319023132324,
      "learning_rate": 0.00012736486671916772,
      "loss": 2.1071,
      "step": 160170
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.1860538721084595,
      "learning_rate": 0.00012736152132355232,
      "loss": 2.079,
      "step": 160171
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.2672113180160522,
      "learning_rate": 0.00012735817596003378,
      "loss": 2.0544,
      "step": 160172
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.328469157218933,
      "learning_rate": 0.0001273548306286126,
      "loss": 2.1567,
      "step": 160173
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.1810959577560425,
      "learning_rate": 0.00012735148532928945,
      "loss": 1.9272,
      "step": 160174
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.340712547302246,
      "learning_rate": 0.00012734814006206478,
      "loss": 1.9957,
      "step": 160175
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.305227518081665,
      "learning_rate": 0.00012734479482693947,
      "loss": 2.4376,
      "step": 160176
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.2873077392578125,
      "learning_rate": 0.00012734144962391392,
      "loss": 1.9472,
      "step": 160177
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.3108668327331543,
      "learning_rate": 0.00012733810445298893,
      "loss": 2.2152,
      "step": 160178
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.1029754877090454,
      "learning_rate": 0.00012733475931416498,
      "loss": 2.0513,
      "step": 160179
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.2934495210647583,
      "learning_rate": 0.00012733141420744293,
      "loss": 2.1518,
      "step": 160180
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.2715249061584473,
      "learning_rate": 0.00012732806913282307,
      "loss": 2.2232,
      "step": 160181
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.1637892723083496,
      "learning_rate": 0.00012732472409030627,
      "loss": 1.779,
      "step": 160182
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.226804494857788,
      "learning_rate": 0.00012732137907989295,
      "loss": 1.9245,
      "step": 160183
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.2101068496704102,
      "learning_rate": 0.00012731803410158393,
      "loss": 2.0245,
      "step": 160184
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.2058680057525635,
      "learning_rate": 0.00012731468915537963,
      "loss": 2.1394,
      "step": 160185
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.1857819557189941,
      "learning_rate": 0.00012731134424128103,
      "loss": 2.0167,
      "step": 160186
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.1142683029174805,
      "learning_rate": 0.00012730799935928834,
      "loss": 1.7927,
      "step": 160187
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.2554495334625244,
      "learning_rate": 0.0001273046545094024,
      "loss": 2.2369,
      "step": 160188
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.1662338972091675,
      "learning_rate": 0.00012730130969162368,
      "loss": 2.0032,
      "step": 160189
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.2383960485458374,
      "learning_rate": 0.00012729796490595306,
      "loss": 2.0592,
      "step": 160190
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.1445597410202026,
      "learning_rate": 0.00012729462015239083,
      "loss": 1.9653,
      "step": 160191
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.2515044212341309,
      "learning_rate": 0.00012729127543093806,
      "loss": 2.1712,
      "step": 160192
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.148820400238037,
      "learning_rate": 0.00012728793074159487,
      "loss": 2.1427,
      "step": 160193
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.1549452543258667,
      "learning_rate": 0.00012728458608436222,
      "loss": 2.2896,
      "step": 160194
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.2054253816604614,
      "learning_rate": 0.0001272812414592405,
      "loss": 2.04,
      "step": 160195
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.1711959838867188,
      "learning_rate": 0.00012727789686623058,
      "loss": 2.0054,
      "step": 160196
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.2135299444198608,
      "learning_rate": 0.0001272745523053329,
      "loss": 1.9785,
      "step": 160197
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.1795557737350464,
      "learning_rate": 0.0001272712077765483,
      "loss": 1.9687,
      "step": 160198
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.1415737867355347,
      "learning_rate": 0.00012726786327987703,
      "loss": 1.9982,
      "step": 160199
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.1649751663208008,
      "learning_rate": 0.00012726451881532006,
      "loss": 1.8984,
      "step": 160200
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.2745442390441895,
      "learning_rate": 0.00012726117438287773,
      "loss": 1.9968,
      "step": 160201
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.1535313129425049,
      "learning_rate": 0.00012725782998255093,
      "loss": 1.991,
      "step": 160202
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.326518177986145,
      "learning_rate": 0.00012725448561434007,
      "loss": 2.1178,
      "step": 160203
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.3919943571090698,
      "learning_rate": 0.00012725114127824607,
      "loss": 2.047,
      "step": 160204
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.2704535722732544,
      "learning_rate": 0.00012724779697426911,
      "loss": 2.0366,
      "step": 160205
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.28584623336792,
      "learning_rate": 0.0001272444527024102,
      "loss": 1.9407,
      "step": 160206
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.157382607460022,
      "learning_rate": 0.00012724110846266964,
      "loss": 1.7904,
      "step": 160207
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.2883588075637817,
      "learning_rate": 0.0001272377642550484,
      "loss": 2.0584,
      "step": 160208
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.1771584749221802,
      "learning_rate": 0.00012723442007954675,
      "loss": 1.7626,
      "step": 160209
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.2028082609176636,
      "learning_rate": 0.0001272310759361656,
      "loss": 1.9989,
      "step": 160210
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.1742887496948242,
      "learning_rate": 0.00012722773182490546,
      "loss": 2.0336,
      "step": 160211
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.14651620388031,
      "learning_rate": 0.00012722438774576695,
      "loss": 2.1535,
      "step": 160212
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.118177890777588,
      "learning_rate": 0.0001272210436987506,
      "loss": 2.0502,
      "step": 160213
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.1662348508834839,
      "learning_rate": 0.0001272176996838572,
      "loss": 2.0433,
      "step": 160214
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.264225721359253,
      "learning_rate": 0.0001272143557010872,
      "loss": 2.0295,
      "step": 160215
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.2680301666259766,
      "learning_rate": 0.0001272110117504414,
      "loss": 1.9602,
      "step": 160216
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.2929970026016235,
      "learning_rate": 0.00012720766783192036,
      "loss": 2.1581,
      "step": 160217
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.14114248752594,
      "learning_rate": 0.00012720432394552463,
      "loss": 2.2474,
      "step": 160218
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.1515107154846191,
      "learning_rate": 0.00012720098009125482,
      "loss": 1.9382,
      "step": 160219
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.2223411798477173,
      "learning_rate": 0.00012719763626911168,
      "loss": 2.066,
      "step": 160220
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.2936809062957764,
      "learning_rate": 0.00012719429247909567,
      "loss": 2.1187,
      "step": 160221
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.3714730739593506,
      "learning_rate": 0.00012719094872120764,
      "loss": 1.9229,
      "step": 160222
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.126953363418579,
      "learning_rate": 0.000127187604995448,
      "loss": 1.9216,
      "step": 160223
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.1171646118164062,
      "learning_rate": 0.00012718426130181753,
      "loss": 1.9195,
      "step": 160224
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.2648062705993652,
      "learning_rate": 0.00012718091764031664,
      "loss": 1.9812,
      "step": 160225
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.171121597290039,
      "learning_rate": 0.00012717757401094616,
      "loss": 2.0476,
      "step": 160226
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.3538411855697632,
      "learning_rate": 0.0001271742304137065,
      "loss": 2.0573,
      "step": 160227
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.3854331970214844,
      "learning_rate": 0.0001271708868485986,
      "loss": 2.0554,
      "step": 160228
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.1050939559936523,
      "learning_rate": 0.00012716754331562285,
      "loss": 1.9906,
      "step": 160229
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.3267234563827515,
      "learning_rate": 0.00012716419981477988,
      "loss": 1.9504,
      "step": 160230
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.2366902828216553,
      "learning_rate": 0.0001271608563460703,
      "loss": 2.1049,
      "step": 160231
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.5814857482910156,
      "learning_rate": 0.00012715751290949485,
      "loss": 2.0811,
      "step": 160232
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.2475361824035645,
      "learning_rate": 0.000127154169505054,
      "loss": 1.9573,
      "step": 160233
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.2167987823486328,
      "learning_rate": 0.00012715082613274856,
      "loss": 2.1468,
      "step": 160234
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.5422166585922241,
      "learning_rate": 0.00012714748279257903,
      "loss": 2.1503,
      "step": 160235
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.3877075910568237,
      "learning_rate": 0.00012714413948454605,
      "loss": 2.0329,
      "step": 160236
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.1530396938323975,
      "learning_rate": 0.00012714079620865012,
      "loss": 1.9691,
      "step": 160237
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.4307594299316406,
      "learning_rate": 0.0001271374529648921,
      "loss": 2.1587,
      "step": 160238
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.212027668952942,
      "learning_rate": 0.00012713410975327238,
      "loss": 1.9399,
      "step": 160239
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.216124415397644,
      "learning_rate": 0.0001271307665737918,
      "loss": 2.0426,
      "step": 160240
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.2795288562774658,
      "learning_rate": 0.00012712742342645085,
      "loss": 1.8872,
      "step": 160241
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.5434356927871704,
      "learning_rate": 0.00012712408031125008,
      "loss": 2.0247,
      "step": 160242
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.1599912643432617,
      "learning_rate": 0.00012712073722819035,
      "loss": 2.1749,
      "step": 160243
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.2912516593933105,
      "learning_rate": 0.0001271173941772721,
      "loss": 2.1006,
      "step": 160244
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.558675765991211,
      "learning_rate": 0.00012711405115849587,
      "loss": 2.1827,
      "step": 160245
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.3993169069290161,
      "learning_rate": 0.00012711070817186254,
      "loss": 2.0413,
      "step": 160246
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.1976336240768433,
      "learning_rate": 0.0001271073652173726,
      "loss": 2.1195,
      "step": 160247
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.3797121047973633,
      "learning_rate": 0.00012710402229502654,
      "loss": 2.0773,
      "step": 160248
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.2945730686187744,
      "learning_rate": 0.0001271006794048252,
      "loss": 2.1537,
      "step": 160249
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.468432068824768,
      "learning_rate": 0.000127097336546769,
      "loss": 2.1437,
      "step": 160250
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.2040890455245972,
      "learning_rate": 0.0001270939937208588,
      "loss": 2.0814,
      "step": 160251
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.2494516372680664,
      "learning_rate": 0.0001270906509270951,
      "loss": 2.1559,
      "step": 160252
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.4020240306854248,
      "learning_rate": 0.0001270873081654785,
      "loss": 1.9857,
      "step": 160253
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.3288850784301758,
      "learning_rate": 0.00012708396543600951,
      "loss": 1.9055,
      "step": 160254
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.2760670185089111,
      "learning_rate": 0.000127080622738689,
      "loss": 2.1401,
      "step": 160255
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.1632083654403687,
      "learning_rate": 0.00012707728007351735,
      "loss": 2.1647,
      "step": 160256
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.2797605991363525,
      "learning_rate": 0.0001270739374404954,
      "loss": 1.9682,
      "step": 160257
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.1323432922363281,
      "learning_rate": 0.00012707059483962367,
      "loss": 1.9648,
      "step": 160258
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.1811362504959106,
      "learning_rate": 0.00012706725227090278,
      "loss": 2.0182,
      "step": 160259
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.1774396896362305,
      "learning_rate": 0.00012706390973433326,
      "loss": 1.9957,
      "step": 160260
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.2787730693817139,
      "learning_rate": 0.00012706056722991592,
      "loss": 2.167,
      "step": 160261
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.3543620109558105,
      "learning_rate": 0.00012705722475765118,
      "loss": 2.0487,
      "step": 160262
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.1259710788726807,
      "learning_rate": 0.00012705388231753988,
      "loss": 1.9141,
      "step": 160263
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.2490317821502686,
      "learning_rate": 0.0001270505399095824,
      "loss": 2.094,
      "step": 160264
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.2060720920562744,
      "learning_rate": 0.00012704719753377973,
      "loss": 2.2989,
      "step": 160265
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.3261523246765137,
      "learning_rate": 0.000127043855190132,
      "loss": 1.9567,
      "step": 160266
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.167388677597046,
      "learning_rate": 0.00012704051287864022,
      "loss": 1.8912,
      "step": 160267
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.2676585912704468,
      "learning_rate": 0.00012703717059930474,
      "loss": 2.0697,
      "step": 160268
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.2665767669677734,
      "learning_rate": 0.00012703382835212648,
      "loss": 2.0189,
      "step": 160269
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.2812551259994507,
      "learning_rate": 0.00012703048613710573,
      "loss": 2.2085,
      "step": 160270
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.2246021032333374,
      "learning_rate": 0.00012702714395424353,
      "loss": 1.9822,
      "step": 160271
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.2090922594070435,
      "learning_rate": 0.00012702380180354,
      "loss": 1.8727,
      "step": 160272
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.2317368984222412,
      "learning_rate": 0.00012702045968499616,
      "loss": 2.0701,
      "step": 160273
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.3870197534561157,
      "learning_rate": 0.00012701711759861232,
      "loss": 2.0474,
      "step": 160274
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.1817799806594849,
      "learning_rate": 0.0001270137755443894,
      "loss": 1.95,
      "step": 160275
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.2218070030212402,
      "learning_rate": 0.00012701043352232778,
      "loss": 1.9227,
      "step": 160276
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.2798317670822144,
      "learning_rate": 0.0001270070915324283,
      "loss": 2.0732,
      "step": 160277
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.2815591096878052,
      "learning_rate": 0.00012700374957469145,
      "loss": 2.0547,
      "step": 160278
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.184171438217163,
      "learning_rate": 0.00012700040764911788,
      "loss": 1.8281,
      "step": 160279
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.639359474182129,
      "learning_rate": 0.0001269970657557081,
      "loss": 2.2808,
      "step": 160280
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.446518063545227,
      "learning_rate": 0.00012699372389446292,
      "loss": 1.8895,
      "step": 160281
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.1980971097946167,
      "learning_rate": 0.00012699038206538275,
      "loss": 1.9366,
      "step": 160282
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.2292057275772095,
      "learning_rate": 0.00012698704026846847,
      "loss": 2.0729,
      "step": 160283
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.2305364608764648,
      "learning_rate": 0.00012698369850372058,
      "loss": 1.9197,
      "step": 160284
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.261321783065796,
      "learning_rate": 0.00012698035677113966,
      "loss": 2.0725,
      "step": 160285
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.1002758741378784,
      "learning_rate": 0.00012697701507072623,
      "loss": 1.9047,
      "step": 160286
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.1157718896865845,
      "learning_rate": 0.0001269736734024812,
      "loss": 2.1244,
      "step": 160287
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.1599879264831543,
      "learning_rate": 0.0001269703317664049,
      "loss": 1.9168,
      "step": 160288
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.1666295528411865,
      "learning_rate": 0.00012696699016249817,
      "loss": 2.0228,
      "step": 160289
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.311533808708191,
      "learning_rate": 0.00012696364859076154,
      "loss": 1.9144,
      "step": 160290
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.2574987411499023,
      "learning_rate": 0.00012696030705119566,
      "loss": 2.112,
      "step": 160291
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.1971362829208374,
      "learning_rate": 0.000126956965543801,
      "loss": 1.8502,
      "step": 160292
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.1177719831466675,
      "learning_rate": 0.00012695362406857843,
      "loss": 1.8291,
      "step": 160293
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.132798433303833,
      "learning_rate": 0.00012695028262552834,
      "loss": 1.9954,
      "step": 160294
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.424961805343628,
      "learning_rate": 0.00012694694121465156,
      "loss": 2.1107,
      "step": 160295
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.3326716423034668,
      "learning_rate": 0.0001269435998359486,
      "loss": 2.0519,
      "step": 160296
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.2330766916275024,
      "learning_rate": 0.0001269402584894201,
      "loss": 1.949,
      "step": 160297
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.185973048210144,
      "learning_rate": 0.00012693691717506657,
      "loss": 1.9555,
      "step": 160298
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.1633453369140625,
      "learning_rate": 0.00012693357589288882,
      "loss": 2.082,
      "step": 160299
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.2494597434997559,
      "learning_rate": 0.00012693023464288732,
      "loss": 1.7797,
      "step": 160300
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.210567593574524,
      "learning_rate": 0.00012692689342506282,
      "loss": 2.0607,
      "step": 160301
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.3838456869125366,
      "learning_rate": 0.0001269235522394159,
      "loss": 1.9248,
      "step": 160302
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.0956425666809082,
      "learning_rate": 0.00012692021108594715,
      "loss": 1.9985,
      "step": 160303
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.3830708265304565,
      "learning_rate": 0.00012691686996465712,
      "loss": 2.2222,
      "step": 160304
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.4100656509399414,
      "learning_rate": 0.00012691352887554658,
      "loss": 2.0729,
      "step": 160305
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.2003878355026245,
      "learning_rate": 0.00012691018781861598,
      "loss": 1.7486,
      "step": 160306
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.197898030281067,
      "learning_rate": 0.00012690684679386615,
      "loss": 2.2958,
      "step": 160307
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.231284737586975,
      "learning_rate": 0.00012690350580129764,
      "loss": 2.1427,
      "step": 160308
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.1853742599487305,
      "learning_rate": 0.00012690016484091102,
      "loss": 2.0679,
      "step": 160309
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.3395192623138428,
      "learning_rate": 0.0001268968239127068,
      "loss": 1.9683,
      "step": 160310
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.4095629453659058,
      "learning_rate": 0.00012689348301668583,
      "loss": 1.9886,
      "step": 160311
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.2304553985595703,
      "learning_rate": 0.00012689014215284857,
      "loss": 2.04,
      "step": 160312
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.1460984945297241,
      "learning_rate": 0.00012688680132119575,
      "loss": 2.1991,
      "step": 160313
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.19529390335083,
      "learning_rate": 0.00012688346052172798,
      "loss": 1.9096,
      "step": 160314
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.2396891117095947,
      "learning_rate": 0.00012688011975444584,
      "loss": 2.0842,
      "step": 160315
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.2641658782958984,
      "learning_rate": 0.0001268767790193498,
      "loss": 2.0871,
      "step": 160316
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.2599269151687622,
      "learning_rate": 0.0001268734383164408,
      "loss": 2.1962,
      "step": 160317
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.2377278804779053,
      "learning_rate": 0.00012687009764571916,
      "loss": 2.151,
      "step": 160318
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.1022911071777344,
      "learning_rate": 0.00012686675700718575,
      "loss": 2.1379,
      "step": 160319
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.1020712852478027,
      "learning_rate": 0.0001268634164008411,
      "loss": 2.2708,
      "step": 160320
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.2746368646621704,
      "learning_rate": 0.00012686007582668577,
      "loss": 2.1743,
      "step": 160321
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.3208333253860474,
      "learning_rate": 0.00012685673528472032,
      "loss": 1.8236,
      "step": 160322
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.2791433334350586,
      "learning_rate": 0.0001268533947749456,
      "loss": 2.1731,
      "step": 160323
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.2580751180648804,
      "learning_rate": 0.00012685005429736198,
      "loss": 2.0059,
      "step": 160324
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.1441603899002075,
      "learning_rate": 0.0001268467138519703,
      "loss": 2.0215,
      "step": 160325
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.1461181640625,
      "learning_rate": 0.00012684337343877114,
      "loss": 2.1496,
      "step": 160326
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.2445334196090698,
      "learning_rate": 0.0001268400330577649,
      "loss": 2.1616,
      "step": 160327
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.073998212814331,
      "learning_rate": 0.00012683669270895247,
      "loss": 2.1007,
      "step": 160328
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.2785942554473877,
      "learning_rate": 0.0001268333523923344,
      "loss": 2.0644,
      "step": 160329
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.4338431358337402,
      "learning_rate": 0.0001268300121079112,
      "loss": 1.9026,
      "step": 160330
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.156830906867981,
      "learning_rate": 0.0001268266718556836,
      "loss": 1.9571,
      "step": 160331
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.2440800666809082,
      "learning_rate": 0.00012682333163565224,
      "loss": 2.0663,
      "step": 160332
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.3020992279052734,
      "learning_rate": 0.00012681999144781755,
      "loss": 1.7993,
      "step": 160333
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.3061143159866333,
      "learning_rate": 0.00012681665129218042,
      "loss": 1.8513,
      "step": 160334
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.1450014114379883,
      "learning_rate": 0.00012681331116874123,
      "loss": 1.992,
      "step": 160335
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.1478885412216187,
      "learning_rate": 0.00012680997107750085,
      "loss": 1.8475,
      "step": 160336
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.2770695686340332,
      "learning_rate": 0.00012680663101845974,
      "loss": 2.0962,
      "step": 160337
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.4560984373092651,
      "learning_rate": 0.00012680329099161856,
      "loss": 1.9252,
      "step": 160338
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.3048447370529175,
      "learning_rate": 0.0001267999509969778,
      "loss": 1.8067,
      "step": 160339
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.174670696258545,
      "learning_rate": 0.00012679661103453833,
      "loss": 1.9178,
      "step": 160340
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.2204499244689941,
      "learning_rate": 0.00012679327110430048,
      "loss": 2.0456,
      "step": 160341
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.382436990737915,
      "learning_rate": 0.00012678993120626518,
      "loss": 1.9696,
      "step": 160342
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.181946873664856,
      "learning_rate": 0.00012678659134043288,
      "loss": 2.0463,
      "step": 160343
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.3382023572921753,
      "learning_rate": 0.00012678325150680412,
      "loss": 1.9539,
      "step": 160344
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.2366529703140259,
      "learning_rate": 0.00012677991170537974,
      "loss": 2.1881,
      "step": 160345
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.232391357421875,
      "learning_rate": 0.00012677657193616026,
      "loss": 2.1504,
      "step": 160346
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.2167212963104248,
      "learning_rate": 0.00012677323219914614,
      "loss": 2.0544,
      "step": 160347
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.2002437114715576,
      "learning_rate": 0.00012676989249433827,
      "loss": 2.1264,
      "step": 160348
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.360663652420044,
      "learning_rate": 0.00012676655282173704,
      "loss": 2.0484,
      "step": 160349
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.281902551651001,
      "learning_rate": 0.00012676321318134327,
      "loss": 2.2455,
      "step": 160350
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.1613857746124268,
      "learning_rate": 0.00012675987357315752,
      "loss": 2.0964,
      "step": 160351
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.371661901473999,
      "learning_rate": 0.00012675653399718035,
      "loss": 2.0435,
      "step": 160352
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.2675665616989136,
      "learning_rate": 0.00012675319445341234,
      "loss": 2.1847,
      "step": 160353
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.2305490970611572,
      "learning_rate": 0.00012674985494185427,
      "loss": 2.0506,
      "step": 160354
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.3147975206375122,
      "learning_rate": 0.00012674651546250654,
      "loss": 1.9887,
      "step": 160355
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.191348671913147,
      "learning_rate": 0.00012674317601537005,
      "loss": 2.1607,
      "step": 160356
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.2165215015411377,
      "learning_rate": 0.00012673983660044526,
      "loss": 2.0066,
      "step": 160357
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.2711951732635498,
      "learning_rate": 0.00012673649721773277,
      "loss": 2.0407,
      "step": 160358
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.2492960691452026,
      "learning_rate": 0.0001267331578672332,
      "loss": 1.8972,
      "step": 160359
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.121902346611023,
      "learning_rate": 0.00012672981854894725,
      "loss": 1.6893,
      "step": 160360
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.1773083209991455,
      "learning_rate": 0.00012672647926287542,
      "loss": 1.9775,
      "step": 160361
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.157332420349121,
      "learning_rate": 0.00012672314000901848,
      "loss": 2.0633,
      "step": 160362
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.2637114524841309,
      "learning_rate": 0.00012671980078737702,
      "loss": 1.9648,
      "step": 160363
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.2411836385726929,
      "learning_rate": 0.00012671646159795162,
      "loss": 1.835,
      "step": 160364
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.199127435684204,
      "learning_rate": 0.0001267131224407428,
      "loss": 2.0422,
      "step": 160365
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.2565511465072632,
      "learning_rate": 0.00012670978331575134,
      "loss": 1.8421,
      "step": 160366
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.1942107677459717,
      "learning_rate": 0.00012670644422297774,
      "loss": 2.0379,
      "step": 160367
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.3693904876708984,
      "learning_rate": 0.00012670310516242278,
      "loss": 2.1084,
      "step": 160368
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.098665475845337,
      "learning_rate": 0.00012669976613408695,
      "loss": 1.9745,
      "step": 160369
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.1628994941711426,
      "learning_rate": 0.00012669642713797097,
      "loss": 2.0495,
      "step": 160370
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.139337182044983,
      "learning_rate": 0.0001266930881740752,
      "loss": 1.8921,
      "step": 160371
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.242754578590393,
      "learning_rate": 0.00012668974924240063,
      "loss": 2.1209,
      "step": 160372
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.2408716678619385,
      "learning_rate": 0.00012668641034294762,
      "loss": 2.147,
      "step": 160373
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.2880455255508423,
      "learning_rate": 0.0001266830714757169,
      "loss": 2.2592,
      "step": 160374
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.1777220964431763,
      "learning_rate": 0.00012667973264070915,
      "loss": 2.1517,
      "step": 160375
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.5249500274658203,
      "learning_rate": 0.00012667639383792488,
      "loss": 2.0293,
      "step": 160376
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.1501713991165161,
      "learning_rate": 0.00012667305506736465,
      "loss": 2.057,
      "step": 160377
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.2807366847991943,
      "learning_rate": 0.00012666971632902926,
      "loss": 2.1371,
      "step": 160378
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.3525733947753906,
      "learning_rate": 0.00012666637762291912,
      "loss": 2.0277,
      "step": 160379
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.2036486864089966,
      "learning_rate": 0.00012666303894903508,
      "loss": 2.1312,
      "step": 160380
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.3543221950531006,
      "learning_rate": 0.00012665970030737768,
      "loss": 1.9687,
      "step": 160381
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.1705946922302246,
      "learning_rate": 0.0001266563616979475,
      "loss": 2.2821,
      "step": 160382
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.1530989408493042,
      "learning_rate": 0.00012665302312074507,
      "loss": 1.9759,
      "step": 160383
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.2408983707427979,
      "learning_rate": 0.00012664968457577123,
      "loss": 2.1447,
      "step": 160384
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.2390202283859253,
      "learning_rate": 0.00012664634606302636,
      "loss": 2.1245,
      "step": 160385
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.2489659786224365,
      "learning_rate": 0.00012664300758251135,
      "loss": 2.0343,
      "step": 160386
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.3015477657318115,
      "learning_rate": 0.00012663966913422663,
      "loss": 2.0876,
      "step": 160387
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.2816485166549683,
      "learning_rate": 0.00012663633071817288,
      "loss": 2.1002,
      "step": 160388
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.3994306325912476,
      "learning_rate": 0.0001266329923343506,
      "loss": 2.0796,
      "step": 160389
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.3424259424209595,
      "learning_rate": 0.00012662965398276064,
      "loss": 1.9843,
      "step": 160390
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.217720627784729,
      "learning_rate": 0.0001266263156634034,
      "loss": 1.9402,
      "step": 160391
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.3946563005447388,
      "learning_rate": 0.0001266229773762797,
      "loss": 1.8124,
      "step": 160392
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.4914984703063965,
      "learning_rate": 0.00012661963912139005,
      "loss": 2.2044,
      "step": 160393
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.3015269041061401,
      "learning_rate": 0.0001266163008987351,
      "loss": 2.1201,
      "step": 160394
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.1982595920562744,
      "learning_rate": 0.00012661296270831532,
      "loss": 2.0837,
      "step": 160395
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.2178945541381836,
      "learning_rate": 0.00012660962455013158,
      "loss": 1.9903,
      "step": 160396
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.2512660026550293,
      "learning_rate": 0.00012660628642418428,
      "loss": 2.0503,
      "step": 160397
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.258446455001831,
      "learning_rate": 0.00012660294833047425,
      "loss": 2.0612,
      "step": 160398
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.1567515134811401,
      "learning_rate": 0.00012659961026900203,
      "loss": 2.0693,
      "step": 160399
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.2771846055984497,
      "learning_rate": 0.00012659627223976817,
      "loss": 1.9029,
      "step": 160400
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.1430450677871704,
      "learning_rate": 0.00012659293424277325,
      "loss": 1.9904,
      "step": 160401
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.1769542694091797,
      "learning_rate": 0.0001265895962780181,
      "loss": 2.0296,
      "step": 160402
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.1217126846313477,
      "learning_rate": 0.0001265862583455031,
      "loss": 2.0496,
      "step": 160403
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.129676103591919,
      "learning_rate": 0.0001265829204452291,
      "loss": 1.9866,
      "step": 160404
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.4181592464447021,
      "learning_rate": 0.0001265795825771966,
      "loss": 2.0084,
      "step": 160405
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.2673802375793457,
      "learning_rate": 0.00012657624474140622,
      "loss": 2.0143,
      "step": 160406
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.4265947341918945,
      "learning_rate": 0.0001265729069378585,
      "loss": 1.7935,
      "step": 160407
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.2873121500015259,
      "learning_rate": 0.00012656956916655426,
      "loss": 2.0429,
      "step": 160408
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.2199299335479736,
      "learning_rate": 0.0001265662314274939,
      "loss": 2.0162,
      "step": 160409
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.131343960762024,
      "learning_rate": 0.00012656289372067826,
      "loss": 1.7213,
      "step": 160410
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.1405054330825806,
      "learning_rate": 0.00012655955604610772,
      "loss": 2.1266,
      "step": 160411
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.1756806373596191,
      "learning_rate": 0.00012655621840378315,
      "loss": 2.2489,
      "step": 160412
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.2367442846298218,
      "learning_rate": 0.00012655288079370505,
      "loss": 2.0562,
      "step": 160413
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.193180799484253,
      "learning_rate": 0.00012654954321587405,
      "loss": 1.8985,
      "step": 160414
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.19965660572052,
      "learning_rate": 0.00012654620567029067,
      "loss": 2.0215,
      "step": 160415
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.2833632230758667,
      "learning_rate": 0.00012654286815695572,
      "loss": 2.0176,
      "step": 160416
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.2280163764953613,
      "learning_rate": 0.00012653953067586962,
      "loss": 2.1299,
      "step": 160417
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.3876004219055176,
      "learning_rate": 0.00012653619322703323,
      "loss": 2.3699,
      "step": 160418
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.311776876449585,
      "learning_rate": 0.000126532855810447,
      "loss": 1.9904,
      "step": 160419
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.1170645952224731,
      "learning_rate": 0.00012652951842611156,
      "loss": 1.9544,
      "step": 160420
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.184043526649475,
      "learning_rate": 0.00012652618107402753,
      "loss": 2.0741,
      "step": 160421
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.2432583570480347,
      "learning_rate": 0.00012652284375419561,
      "loss": 1.8908,
      "step": 160422
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.3006285429000854,
      "learning_rate": 0.00012651950646661624,
      "loss": 2.0598,
      "step": 160423
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.125030517578125,
      "learning_rate": 0.00012651616921129032,
      "loss": 1.8871,
      "step": 160424
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.7002776861190796,
      "learning_rate": 0.00012651283198821832,
      "loss": 2.0058,
      "step": 160425
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.287958025932312,
      "learning_rate": 0.00012650949479740072,
      "loss": 2.1259,
      "step": 160426
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.2179696559906006,
      "learning_rate": 0.0001265061576388384,
      "loss": 2.0528,
      "step": 160427
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.1369224786758423,
      "learning_rate": 0.00012650282051253188,
      "loss": 2.2179,
      "step": 160428
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.3303487300872803,
      "learning_rate": 0.00012649948341848162,
      "loss": 1.9641,
      "step": 160429
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.2345834970474243,
      "learning_rate": 0.00012649614635668852,
      "loss": 1.9357,
      "step": 160430
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.25783371925354,
      "learning_rate": 0.00012649280932715306,
      "loss": 2.2462,
      "step": 160431
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.2352924346923828,
      "learning_rate": 0.0001264894723298757,
      "loss": 2.0811,
      "step": 160432
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.261083960533142,
      "learning_rate": 0.0001264861353648574,
      "loss": 1.8778,
      "step": 160433
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.1971336603164673,
      "learning_rate": 0.00012648279843209846,
      "loss": 2.0714,
      "step": 160434
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.1991949081420898,
      "learning_rate": 0.00012647946153159978,
      "loss": 2.0695,
      "step": 160435
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.1715036630630493,
      "learning_rate": 0.00012647612466336181,
      "loss": 1.9659,
      "step": 160436
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.167587399482727,
      "learning_rate": 0.00012647278782738523,
      "loss": 2.08,
      "step": 160437
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.163009524345398,
      "learning_rate": 0.0001264694510236705,
      "loss": 1.8887,
      "step": 160438
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.1729011535644531,
      "learning_rate": 0.00012646611425221846,
      "loss": 2.0082,
      "step": 160439
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.2060267925262451,
      "learning_rate": 0.00012646277751302958,
      "loss": 2.0573,
      "step": 160440
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.2023251056671143,
      "learning_rate": 0.00012645944080610466,
      "loss": 1.9056,
      "step": 160441
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.1918848752975464,
      "learning_rate": 0.0001264561041314442,
      "loss": 2.0887,
      "step": 160442
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.0875396728515625,
      "learning_rate": 0.0001264527674890488,
      "loss": 1.8405,
      "step": 160443
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.185556411743164,
      "learning_rate": 0.00012644943087891902,
      "loss": 2.1629,
      "step": 160444
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.331735610961914,
      "learning_rate": 0.00012644609430105568,
      "loss": 2.0704,
      "step": 160445
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.165996789932251,
      "learning_rate": 0.00012644275775545918,
      "loss": 2.0874,
      "step": 160446
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.3337496519088745,
      "learning_rate": 0.00012643942124213032,
      "loss": 2.0473,
      "step": 160447
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.2508114576339722,
      "learning_rate": 0.0001264360847610697,
      "loss": 1.999,
      "step": 160448
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.22100031375885,
      "learning_rate": 0.00012643274831227784,
      "loss": 1.9762,
      "step": 160449
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.3508877754211426,
      "learning_rate": 0.00012642941189575536,
      "loss": 1.9699,
      "step": 160450
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.1758060455322266,
      "learning_rate": 0.000126426075511503,
      "loss": 2.012,
      "step": 160451
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.3044015169143677,
      "learning_rate": 0.00012642273915952116,
      "loss": 2.0639,
      "step": 160452
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.1113874912261963,
      "learning_rate": 0.00012641940283981075,
      "loss": 1.9135,
      "step": 160453
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.1532715559005737,
      "learning_rate": 0.00012641606655237227,
      "loss": 1.9843,
      "step": 160454
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.0728713274002075,
      "learning_rate": 0.00012641273029720632,
      "loss": 1.9938,
      "step": 160455
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.1672072410583496,
      "learning_rate": 0.0001264093940743134,
      "loss": 2.2574,
      "step": 160456
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.1937264204025269,
      "learning_rate": 0.00012640605788369433,
      "loss": 2.0003,
      "step": 160457
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.1853364706039429,
      "learning_rate": 0.00012640272172534958,
      "loss": 2.0806,
      "step": 160458
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.119884967803955,
      "learning_rate": 0.00012639938559927996,
      "loss": 1.7362,
      "step": 160459
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.293819785118103,
      "learning_rate": 0.00012639604950548595,
      "loss": 2.0796,
      "step": 160460
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.2390679121017456,
      "learning_rate": 0.00012639271344396817,
      "loss": 2.1159,
      "step": 160461
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.2366011142730713,
      "learning_rate": 0.00012638937741472718,
      "loss": 2.1965,
      "step": 160462
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.1715381145477295,
      "learning_rate": 0.00012638604141776378,
      "loss": 1.7497,
      "step": 160463
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.2073566913604736,
      "learning_rate": 0.00012638270545307838,
      "loss": 2.1154,
      "step": 160464
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.1219745874404907,
      "learning_rate": 0.00012637936952067185,
      "loss": 1.7385,
      "step": 160465
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.2744140625,
      "learning_rate": 0.00012637603362054465,
      "loss": 1.956,
      "step": 160466
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.1291896104812622,
      "learning_rate": 0.00012637269775269742,
      "loss": 2.0062,
      "step": 160467
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.1466739177703857,
      "learning_rate": 0.0001263693619171307,
      "loss": 2.0104,
      "step": 160468
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.168272614479065,
      "learning_rate": 0.00012636602611384526,
      "loss": 2.1512,
      "step": 160469
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.342342734336853,
      "learning_rate": 0.00012636269034284157,
      "loss": 1.9695,
      "step": 160470
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.236630916595459,
      "learning_rate": 0.00012635935460412041,
      "loss": 2.3065,
      "step": 160471
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.3493119478225708,
      "learning_rate": 0.00012635601889768238,
      "loss": 1.9835,
      "step": 160472
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.2343451976776123,
      "learning_rate": 0.000126352683223528,
      "loss": 1.9658,
      "step": 160473
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.288488507270813,
      "learning_rate": 0.00012634934758165784,
      "loss": 2.1428,
      "step": 160474
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.2326452732086182,
      "learning_rate": 0.00012634601197207275,
      "loss": 2.1691,
      "step": 160475
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.3581113815307617,
      "learning_rate": 0.00012634267639477308,
      "loss": 2.0268,
      "step": 160476
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.243874192237854,
      "learning_rate": 0.00012633934084975968,
      "loss": 2.0853,
      "step": 160477
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.214250922203064,
      "learning_rate": 0.00012633600533703298,
      "loss": 2.1842,
      "step": 160478
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.2425274848937988,
      "learning_rate": 0.0001263326698565939,
      "loss": 1.9576,
      "step": 160479
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.4175317287445068,
      "learning_rate": 0.00012632933440844267,
      "loss": 2.1257,
      "step": 160480
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.2306053638458252,
      "learning_rate": 0.00012632599899258017,
      "loss": 2.2486,
      "step": 160481
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.1772220134735107,
      "learning_rate": 0.00012632266360900683,
      "loss": 2.1239,
      "step": 160482
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.4050506353378296,
      "learning_rate": 0.00012631932825772353,
      "loss": 2.0635,
      "step": 160483
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.3133654594421387,
      "learning_rate": 0.0001263159929387306,
      "loss": 1.8907,
      "step": 160484
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.3613295555114746,
      "learning_rate": 0.00012631265765202906,
      "loss": 2.0994,
      "step": 160485
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.155480146408081,
      "learning_rate": 0.00012630932239761903,
      "loss": 1.9791,
      "step": 160486
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.328283429145813,
      "learning_rate": 0.00012630598717550148,
      "loss": 2.1097,
      "step": 160487
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.1120994091033936,
      "learning_rate": 0.00012630265198567683,
      "loss": 2.1538,
      "step": 160488
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.1615047454833984,
      "learning_rate": 0.0001262993168281459,
      "loss": 2.0875,
      "step": 160489
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.1629129648208618,
      "learning_rate": 0.00012629598170290913,
      "loss": 2.0865,
      "step": 160490
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.140268087387085,
      "learning_rate": 0.00012629264660996738,
      "loss": 2.0768,
      "step": 160491
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.135750651359558,
      "learning_rate": 0.00012628931154932092,
      "loss": 2.1318,
      "step": 160492
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.1542203426361084,
      "learning_rate": 0.00012628597652097065,
      "loss": 2.1016,
      "step": 160493
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.1961228847503662,
      "learning_rate": 0.000126282641524917,
      "loss": 2.0082,
      "step": 160494
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.1628930568695068,
      "learning_rate": 0.00012627930656116082,
      "loss": 1.8735,
      "step": 160495
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.1651792526245117,
      "learning_rate": 0.00012627597162970244,
      "loss": 2.1851,
      "step": 160496
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.1775950193405151,
      "learning_rate": 0.00012627263673054278,
      "loss": 2.162,
      "step": 160497
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.20232355594635,
      "learning_rate": 0.00012626930186368228,
      "loss": 2.1245,
      "step": 160498
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.1990081071853638,
      "learning_rate": 0.00012626596702912163,
      "loss": 1.7791,
      "step": 160499
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.2355519533157349,
      "learning_rate": 0.00012626263222686128,
      "loss": 1.8248,
      "step": 160500
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.1839045286178589,
      "learning_rate": 0.0001262592974569021,
      "loss": 1.9035,
      "step": 160501
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.439834713935852,
      "learning_rate": 0.00012625596271924452,
      "loss": 1.9715,
      "step": 160502
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.3060978651046753,
      "learning_rate": 0.0001262526280138893,
      "loss": 2.0863,
      "step": 160503
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.186990737915039,
      "learning_rate": 0.00012624929334083702,
      "loss": 2.2092,
      "step": 160504
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.260542869567871,
      "learning_rate": 0.0001262459587000883,
      "loss": 2.1818,
      "step": 160505
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.2277084589004517,
      "learning_rate": 0.0001262426240916436,
      "loss": 1.9149,
      "step": 160506
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.2013181447982788,
      "learning_rate": 0.00012623928951550377,
      "loss": 2.0562,
      "step": 160507
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.1873314380645752,
      "learning_rate": 0.00012623595497166925,
      "loss": 2.0338,
      "step": 160508
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.2563371658325195,
      "learning_rate": 0.00012623262046014086,
      "loss": 1.9659,
      "step": 160509
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.2501293420791626,
      "learning_rate": 0.0001262292859809191,
      "loss": 1.9538,
      "step": 160510
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.3069593906402588,
      "learning_rate": 0.00012622595153400447,
      "loss": 1.7266,
      "step": 160511
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.207196831703186,
      "learning_rate": 0.00012622261711939782,
      "loss": 1.8527,
      "step": 160512
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.1372027397155762,
      "learning_rate": 0.00012621928273709966,
      "loss": 2.0873,
      "step": 160513
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.1894983053207397,
      "learning_rate": 0.00012621594838711057,
      "loss": 1.936,
      "step": 160514
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.2762373685836792,
      "learning_rate": 0.00012621261406943127,
      "loss": 2.0352,
      "step": 160515
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.2399277687072754,
      "learning_rate": 0.00012620927978406233,
      "loss": 2.0324,
      "step": 160516
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.2795944213867188,
      "learning_rate": 0.00012620594553100427,
      "loss": 2.1482,
      "step": 160517
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.1601086854934692,
      "learning_rate": 0.0001262026113102579,
      "loss": 2.0006,
      "step": 160518
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.3866045475006104,
      "learning_rate": 0.00012619927712182377,
      "loss": 2.0973,
      "step": 160519
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.189205288887024,
      "learning_rate": 0.00012619594296570233,
      "loss": 2.0466,
      "step": 160520
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.2568979263305664,
      "learning_rate": 0.0001261926088418945,
      "loss": 1.9761,
      "step": 160521
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.2656110525131226,
      "learning_rate": 0.0001261892747504007,
      "loss": 2.2571,
      "step": 160522
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.202768087387085,
      "learning_rate": 0.0001261859406912215,
      "loss": 1.9382,
      "step": 160523
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.317748785018921,
      "learning_rate": 0.00012618260666435774,
      "loss": 1.9853,
      "step": 160524
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.2468003034591675,
      "learning_rate": 0.00012617927266980977,
      "loss": 1.9589,
      "step": 160525
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.195391297340393,
      "learning_rate": 0.0001261759387075785,
      "loss": 1.9829,
      "step": 160526
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.1794722080230713,
      "learning_rate": 0.00012617260477766441,
      "loss": 1.9418,
      "step": 160527
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.1335980892181396,
      "learning_rate": 0.00012616927088006808,
      "loss": 1.9795,
      "step": 160528
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.227332592010498,
      "learning_rate": 0.00012616593701479008,
      "loss": 1.9217,
      "step": 160529
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.2124649286270142,
      "learning_rate": 0.0001261626031818312,
      "loss": 1.9986,
      "step": 160530
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.1484392881393433,
      "learning_rate": 0.0001261592693811919,
      "loss": 2.0184,
      "step": 160531
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.3685051202774048,
      "learning_rate": 0.00012615593561287293,
      "loss": 2.1588,
      "step": 160532
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.2337030172348022,
      "learning_rate": 0.0001261526018768749,
      "loss": 2.0063,
      "step": 160533
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.2693909406661987,
      "learning_rate": 0.00012614926817319837,
      "loss": 2.166,
      "step": 160534
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.3208001852035522,
      "learning_rate": 0.00012614593450184386,
      "loss": 1.8637,
      "step": 160535
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.1841487884521484,
      "learning_rate": 0.0001261426008628122,
      "loss": 2.1004,
      "step": 160536
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.1442245244979858,
      "learning_rate": 0.00012613926725610382,
      "loss": 1.8466,
      "step": 160537
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.2143936157226562,
      "learning_rate": 0.00012613593368171952,
      "loss": 1.9653,
      "step": 160538
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.177727460861206,
      "learning_rate": 0.00012613260013965986,
      "loss": 2.0715,
      "step": 160539
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.1041823625564575,
      "learning_rate": 0.0001261292666299254,
      "loss": 1.9051,
      "step": 160540
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.356963038444519,
      "learning_rate": 0.0001261259331525167,
      "loss": 2.1817,
      "step": 160541
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.1489005088806152,
      "learning_rate": 0.00012612259970743461,
      "loss": 2.0126,
      "step": 160542
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.259541630744934,
      "learning_rate": 0.00012611926629467945,
      "loss": 2.0653,
      "step": 160543
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.1768828630447388,
      "learning_rate": 0.00012611593291425214,
      "loss": 2.1461,
      "step": 160544
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.1585274934768677,
      "learning_rate": 0.00012611259956615305,
      "loss": 1.9811,
      "step": 160545
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.1680701971054077,
      "learning_rate": 0.00012610926625038313,
      "loss": 2.1608,
      "step": 160546
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.21812105178833,
      "learning_rate": 0.00012610593296694255,
      "loss": 2.1887,
      "step": 160547
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.0638599395751953,
      "learning_rate": 0.00012610259971583226,
      "loss": 2.1925,
      "step": 160548
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.301376223564148,
      "learning_rate": 0.0001260992664970527,
      "loss": 1.9864,
      "step": 160549
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.26784086227417,
      "learning_rate": 0.00012609593331060467,
      "loss": 2.2581,
      "step": 160550
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.2211713790893555,
      "learning_rate": 0.00012609260015648854,
      "loss": 1.9346,
      "step": 160551
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.2015259265899658,
      "learning_rate": 0.00012608926703470532,
      "loss": 2.2376,
      "step": 160552
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.339826226234436,
      "learning_rate": 0.00012608593394525518,
      "loss": 2.049,
      "step": 160553
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.1897099018096924,
      "learning_rate": 0.0001260826008881391,
      "loss": 1.9587,
      "step": 160554
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.2108831405639648,
      "learning_rate": 0.0001260792678633574,
      "loss": 2.1002,
      "step": 160555
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.2976524829864502,
      "learning_rate": 0.0001260759348709109,
      "loss": 2.1754,
      "step": 160556
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.1444308757781982,
      "learning_rate": 0.00012607260191080011,
      "loss": 2.2873,
      "step": 160557
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.3077577352523804,
      "learning_rate": 0.00012606926898302593,
      "loss": 2.1426,
      "step": 160558
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.1117215156555176,
      "learning_rate": 0.00012606593608758854,
      "loss": 2.0285,
      "step": 160559
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.1974714994430542,
      "learning_rate": 0.00012606260322448886,
      "loss": 2.0122,
      "step": 160560
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.3426271677017212,
      "learning_rate": 0.00012605927039372733,
      "loss": 2.0814,
      "step": 160561
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.2161664962768555,
      "learning_rate": 0.00012605593759530478,
      "loss": 1.9352,
      "step": 160562
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.281083583831787,
      "learning_rate": 0.00012605260482922162,
      "loss": 2.2602,
      "step": 160563
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.1677517890930176,
      "learning_rate": 0.00012604927209547873,
      "loss": 1.823,
      "step": 160564
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.232061743736267,
      "learning_rate": 0.0001260459393940764,
      "loss": 2.179,
      "step": 160565
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.4520177841186523,
      "learning_rate": 0.0001260426067250155,
      "loss": 1.9257,
      "step": 160566
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.1401698589324951,
      "learning_rate": 0.00012603927408829646,
      "loss": 2.0015,
      "step": 160567
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.201036810874939,
      "learning_rate": 0.00012603594148392008,
      "loss": 2.1294,
      "step": 160568
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.1487290859222412,
      "learning_rate": 0.00012603260891188683,
      "loss": 2.2491,
      "step": 160569
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.2925679683685303,
      "learning_rate": 0.0001260292763721976,
      "loss": 2.0825,
      "step": 160570
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.1795731782913208,
      "learning_rate": 0.0001260259438648526,
      "loss": 1.945,
      "step": 160571
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.1768878698349,
      "learning_rate": 0.0001260226113898528,
      "loss": 2.1083,
      "step": 160572
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.2442349195480347,
      "learning_rate": 0.00012601927894719854,
      "loss": 1.9531,
      "step": 160573
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.2359318733215332,
      "learning_rate": 0.0001260159465368907,
      "loss": 2.1476,
      "step": 160574
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.1181517839431763,
      "learning_rate": 0.00012601261415892966,
      "loss": 1.9392,
      "step": 160575
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.2738031148910522,
      "learning_rate": 0.00012600928181331637,
      "loss": 1.9889,
      "step": 160576
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.3575235605239868,
      "learning_rate": 0.00012600594950005106,
      "loss": 1.8266,
      "step": 160577
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.1542153358459473,
      "learning_rate": 0.00012600261721913458,
      "loss": 1.9336,
      "step": 160578
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.312278151512146,
      "learning_rate": 0.00012599928497056745,
      "loss": 2.0326,
      "step": 160579
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.212186336517334,
      "learning_rate": 0.00012599595275435038,
      "loss": 2.1006,
      "step": 160580
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.2000212669372559,
      "learning_rate": 0.00012599262057048387,
      "loss": 1.9989,
      "step": 160581
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.3541501760482788,
      "learning_rate": 0.00012598928841896885,
      "loss": 2.025,
      "step": 160582
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.2661820650100708,
      "learning_rate": 0.0001259859562998055,
      "loss": 1.9201,
      "step": 160583
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.1984788179397583,
      "learning_rate": 0.0001259826242129947,
      "loss": 2.1776,
      "step": 160584
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.1375809907913208,
      "learning_rate": 0.00012597929215853697,
      "loss": 1.924,
      "step": 160585
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.3182257413864136,
      "learning_rate": 0.00012597596013643302,
      "loss": 2.45,
      "step": 160586
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.3221440315246582,
      "learning_rate": 0.00012597262814668334,
      "loss": 2.2303,
      "step": 160587
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.243130087852478,
      "learning_rate": 0.00012596929618928875,
      "loss": 1.9785,
      "step": 160588
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.2035969495773315,
      "learning_rate": 0.00012596596426424976,
      "loss": 2.0591,
      "step": 160589
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.055349588394165,
      "learning_rate": 0.00012596263237156697,
      "loss": 2.1121,
      "step": 160590
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.2403590679168701,
      "learning_rate": 0.0001259593005112409,
      "loss": 1.9721,
      "step": 160591
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.3933430910110474,
      "learning_rate": 0.0001259559686832724,
      "loss": 2.044,
      "step": 160592
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.128872036933899,
      "learning_rate": 0.00012595263688766186,
      "loss": 2.139,
      "step": 160593
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.2768170833587646,
      "learning_rate": 0.0001259493051244101,
      "loss": 1.9546,
      "step": 160594
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.234511137008667,
      "learning_rate": 0.0001259459733935177,
      "loss": 2.0592,
      "step": 160595
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.1861690282821655,
      "learning_rate": 0.00012594264169498508,
      "loss": 1.9533,
      "step": 160596
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.1784849166870117,
      "learning_rate": 0.00012593931002881313,
      "loss": 2.0366,
      "step": 160597
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.2017936706542969,
      "learning_rate": 0.0001259359783950023,
      "loss": 2.049,
      "step": 160598
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.3845747709274292,
      "learning_rate": 0.0001259326467935532,
      "loss": 1.9996,
      "step": 160599
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.4758964776992798,
      "learning_rate": 0.00012592931522446664,
      "loss": 1.6353,
      "step": 160600
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.1763516664505005,
      "learning_rate": 0.00012592598368774304,
      "loss": 1.9654,
      "step": 160601
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.3410003185272217,
      "learning_rate": 0.00012592265218338303,
      "loss": 2.0105,
      "step": 160602
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.2675020694732666,
      "learning_rate": 0.00012591932071138735,
      "loss": 2.1068,
      "step": 160603
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.2722547054290771,
      "learning_rate": 0.0001259159892717566,
      "loss": 2.1047,
      "step": 160604
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.2059566974639893,
      "learning_rate": 0.0001259126578644912,
      "loss": 2.2024,
      "step": 160605
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.2833013534545898,
      "learning_rate": 0.00012590932648959207,
      "loss": 2.1804,
      "step": 160606
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.256158709526062,
      "learning_rate": 0.00012590599514705967,
      "loss": 1.8336,
      "step": 160607
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.1453348398208618,
      "learning_rate": 0.0001259026638368945,
      "loss": 2.0185,
      "step": 160608
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.1638685464859009,
      "learning_rate": 0.00012589933255909743,
      "loss": 1.8624,
      "step": 160609
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.12894606590271,
      "learning_rate": 0.0001258960013136689,
      "loss": 2.1261,
      "step": 160610
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.0790667533874512,
      "learning_rate": 0.00012589267010060964,
      "loss": 1.9088,
      "step": 160611
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.3609259128570557,
      "learning_rate": 0.00012588933891992025,
      "loss": 1.9118,
      "step": 160612
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.400302529335022,
      "learning_rate": 0.00012588600777160128,
      "loss": 2.2203,
      "step": 160613
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.1581209897994995,
      "learning_rate": 0.00012588267665565331,
      "loss": 2.0596,
      "step": 160614
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.17173433303833,
      "learning_rate": 0.00012587934557207716,
      "loss": 1.8947,
      "step": 160615
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.2216954231262207,
      "learning_rate": 0.0001258760145208732,
      "loss": 2.3606,
      "step": 160616
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.3905112743377686,
      "learning_rate": 0.00012587268350204225,
      "loss": 1.9822,
      "step": 160617
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.2312936782836914,
      "learning_rate": 0.0001258693525155849,
      "loss": 1.9292,
      "step": 160618
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.166441559791565,
      "learning_rate": 0.00012586602156150175,
      "loss": 1.894,
      "step": 160619
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.2454544305801392,
      "learning_rate": 0.0001258626906397932,
      "loss": 2.0299,
      "step": 160620
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.1500438451766968,
      "learning_rate": 0.00012585935975046022,
      "loss": 1.8131,
      "step": 160621
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.209338665008545,
      "learning_rate": 0.00012585602889350319,
      "loss": 2.0695,
      "step": 160622
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.4159647226333618,
      "learning_rate": 0.0001258526980689229,
      "loss": 2.1478,
      "step": 160623
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.1656798124313354,
      "learning_rate": 0.00012584936727671976,
      "loss": 2.05,
      "step": 160624
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.1543551683425903,
      "learning_rate": 0.00012584603651689472,
      "loss": 2.0004,
      "step": 160625
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.294061303138733,
      "learning_rate": 0.000125842705789448,
      "loss": 1.9882,
      "step": 160626
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.1982676982879639,
      "learning_rate": 0.0001258393750943805,
      "loss": 1.9346,
      "step": 160627
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.2600017786026,
      "learning_rate": 0.00012583604443169264,
      "loss": 2.2087,
      "step": 160628
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.184126377105713,
      "learning_rate": 0.00012583271380138525,
      "loss": 2.026,
      "step": 160629
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.3099918365478516,
      "learning_rate": 0.00012582938320345876,
      "loss": 2.0162,
      "step": 160630
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.1338547468185425,
      "learning_rate": 0.00012582605263791404,
      "loss": 2.1941,
      "step": 160631
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.2294543981552124,
      "learning_rate": 0.0001258227221047514,
      "loss": 2.1248,
      "step": 160632
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.1117393970489502,
      "learning_rate": 0.00012581939160397166,
      "loss": 2.1568,
      "step": 160633
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.1339248418807983,
      "learning_rate": 0.00012581606113557528,
      "loss": 2.0387,
      "step": 160634
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.2466723918914795,
      "learning_rate": 0.00012581273069956313,
      "loss": 2.0113,
      "step": 160635
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.1263049840927124,
      "learning_rate": 0.0001258094002959355,
      "loss": 2.041,
      "step": 160636
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.2694000005722046,
      "learning_rate": 0.00012580606992469346,
      "loss": 2.085,
      "step": 160637
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.1568591594696045,
      "learning_rate": 0.00012580273958583715,
      "loss": 2.1609,
      "step": 160638
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.3991296291351318,
      "learning_rate": 0.0001257994092793675,
      "loss": 2.0197,
      "step": 160639
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.1035207509994507,
      "learning_rate": 0.0001257960790052849,
      "loss": 2.0923,
      "step": 160640
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.162644386291504,
      "learning_rate": 0.00012579274876359022,
      "loss": 2.0992,
      "step": 160641
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.248318076133728,
      "learning_rate": 0.00012578941855428385,
      "loss": 2.0502,
      "step": 160642
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.372697114944458,
      "learning_rate": 0.00012578608837736672,
      "loss": 2.0315,
      "step": 160643
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.2762876749038696,
      "learning_rate": 0.00012578275823283905,
      "loss": 2.1606,
      "step": 160644
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.2207245826721191,
      "learning_rate": 0.00012577942812070175,
      "loss": 2.2029,
      "step": 160645
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.131981611251831,
      "learning_rate": 0.0001257760980409552,
      "loss": 2.207,
      "step": 160646
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.1967031955718994,
      "learning_rate": 0.00012577276799360032,
      "loss": 2.1689,
      "step": 160647
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.195477843284607,
      "learning_rate": 0.00012576943797863746,
      "loss": 2.1041,
      "step": 160648
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.2120102643966675,
      "learning_rate": 0.00012576610799606755,
      "loss": 2.06,
      "step": 160649
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.3613488674163818,
      "learning_rate": 0.00012576277804589078,
      "loss": 1.7739,
      "step": 160650
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.3092057704925537,
      "learning_rate": 0.00012575944812810813,
      "loss": 2.1951,
      "step": 160651
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.2203662395477295,
      "learning_rate": 0.00012575611824271998,
      "loss": 1.9745,
      "step": 160652
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.2302290201187134,
      "learning_rate": 0.00012575278838972714,
      "loss": 2.1287,
      "step": 160653
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.189490556716919,
      "learning_rate": 0.00012574945856913005,
      "loss": 2.0303,
      "step": 160654
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.3038681745529175,
      "learning_rate": 0.00012574612878092964,
      "loss": 2.0994,
      "step": 160655
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.1690775156021118,
      "learning_rate": 0.00012574279902512615,
      "loss": 2.0636,
      "step": 160656
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.2125039100646973,
      "learning_rate": 0.0001257394693017204,
      "loss": 2.0172,
      "step": 160657
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.3900398015975952,
      "learning_rate": 0.0001257361396107129,
      "loss": 2.3342,
      "step": 160658
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.498611330986023,
      "learning_rate": 0.00012573280995210442,
      "loss": 1.9018,
      "step": 160659
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.0693891048431396,
      "learning_rate": 0.00012572948032589542,
      "loss": 1.884,
      "step": 160660
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.2526119947433472,
      "learning_rate": 0.00012572615073208683,
      "loss": 2.1701,
      "step": 160661
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.1792008876800537,
      "learning_rate": 0.00012572282117067878,
      "loss": 1.9946,
      "step": 160662
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.1452962160110474,
      "learning_rate": 0.00012571949164167227,
      "loss": 2.2141,
      "step": 160663
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.2958143949508667,
      "learning_rate": 0.0001257161621450677,
      "loss": 1.892,
      "step": 160664
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.2330948114395142,
      "learning_rate": 0.00012571283268086587,
      "loss": 2.155,
      "step": 160665
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.1785178184509277,
      "learning_rate": 0.0001257095032490672,
      "loss": 1.897,
      "step": 160666
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.2964388132095337,
      "learning_rate": 0.0001257061738496727,
      "loss": 2.1057,
      "step": 160667
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.208535075187683,
      "learning_rate": 0.00012570284448268243,
      "loss": 1.9452,
      "step": 160668
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.2909314632415771,
      "learning_rate": 0.00012569951514809744,
      "loss": 2.2089,
      "step": 160669
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.208154559135437,
      "learning_rate": 0.00012569618584591807,
      "loss": 1.9611,
      "step": 160670
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.1612935066223145,
      "learning_rate": 0.0001256928565761452,
      "loss": 2.0542,
      "step": 160671
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.1665213108062744,
      "learning_rate": 0.00012568952733877918,
      "loss": 2.0276,
      "step": 160672
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.251569151878357,
      "learning_rate": 0.0001256861981338209,
      "loss": 2.1104,
      "step": 160673
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.258832573890686,
      "learning_rate": 0.0001256828689612708,
      "loss": 1.9714,
      "step": 160674
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.1937129497528076,
      "learning_rate": 0.00012567953982112962,
      "loss": 1.9129,
      "step": 160675
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.1666620969772339,
      "learning_rate": 0.00012567621071339773,
      "loss": 2.1177,
      "step": 160676
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.0904072523117065,
      "learning_rate": 0.0001256728816380761,
      "loss": 2.0722,
      "step": 160677
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.2610279321670532,
      "learning_rate": 0.00012566955259516499,
      "loss": 1.8774,
      "step": 160678
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.2430952787399292,
      "learning_rate": 0.00012566622358466535,
      "loss": 2.0073,
      "step": 160679
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.3820621967315674,
      "learning_rate": 0.00012566289460657764,
      "loss": 2.2137,
      "step": 160680
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.362621545791626,
      "learning_rate": 0.00012565956566090253,
      "loss": 1.8456,
      "step": 160681
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.3210184574127197,
      "learning_rate": 0.0001256562367476404,
      "loss": 2.0963,
      "step": 160682
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.1181161403656006,
      "learning_rate": 0.00012565290786679224,
      "loss": 1.9975,
      "step": 160683
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.1453372240066528,
      "learning_rate": 0.00012564957901835838,
      "loss": 1.9738,
      "step": 160684
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.1850059032440186,
      "learning_rate": 0.00012564625020233965,
      "loss": 2.0681,
      "step": 160685
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.2028694152832031,
      "learning_rate": 0.00012564292141873659,
      "loss": 2.0153,
      "step": 160686
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.327669382095337,
      "learning_rate": 0.00012563959266754967,
      "loss": 2.0254,
      "step": 160687
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.3227434158325195,
      "learning_rate": 0.00012563626394877979,
      "loss": 1.8679,
      "step": 160688
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.2916109561920166,
      "learning_rate": 0.00012563293526242735,
      "loss": 2.0771,
      "step": 160689
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.2018239498138428,
      "learning_rate": 0.00012562960660849297,
      "loss": 2.0933,
      "step": 160690
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.1439558267593384,
      "learning_rate": 0.00012562627798697745,
      "loss": 2.0355,
      "step": 160691
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.1739929914474487,
      "learning_rate": 0.0001256229493978813,
      "loss": 2.0911,
      "step": 160692
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.3992539644241333,
      "learning_rate": 0.00012561962084120502,
      "loss": 2.1063,
      "step": 160693
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.1226333379745483,
      "learning_rate": 0.00012561629231694945,
      "loss": 2.1745,
      "step": 160694
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.309095025062561,
      "learning_rate": 0.000125612963825115,
      "loss": 1.7712,
      "step": 160695
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.2477502822875977,
      "learning_rate": 0.0001256096353657025,
      "loss": 2.024,
      "step": 160696
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.2690403461456299,
      "learning_rate": 0.00012560630693871246,
      "loss": 2.0611,
      "step": 160697
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.1555920839309692,
      "learning_rate": 0.0001256029785441455,
      "loss": 2.2561,
      "step": 160698
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.214867353439331,
      "learning_rate": 0.00012559965018200215,
      "loss": 1.9272,
      "step": 160699
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.2836897373199463,
      "learning_rate": 0.00012559632185228317,
      "loss": 1.9937,
      "step": 160700
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.2518773078918457,
      "learning_rate": 0.00012559299355498907,
      "loss": 2.1213,
      "step": 160701
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.3210643529891968,
      "learning_rate": 0.00012558966529012064,
      "loss": 1.9457,
      "step": 160702
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.1994256973266602,
      "learning_rate": 0.00012558633705767833,
      "loss": 2.0264,
      "step": 160703
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.2513593435287476,
      "learning_rate": 0.00012558300885766283,
      "loss": 2.244,
      "step": 160704
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.1857414245605469,
      "learning_rate": 0.00012557968069007466,
      "loss": 2.1007,
      "step": 160705
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.4011000394821167,
      "learning_rate": 0.00012557635255491462,
      "loss": 1.7803,
      "step": 160706
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.2295244932174683,
      "learning_rate": 0.00012557302445218312,
      "loss": 2.145,
      "step": 160707
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.2425799369812012,
      "learning_rate": 0.00012556969638188097,
      "loss": 1.9052,
      "step": 160708
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.1749852895736694,
      "learning_rate": 0.0001255663683440086,
      "loss": 1.9621,
      "step": 160709
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.1449928283691406,
      "learning_rate": 0.000125563040338567,
      "loss": 2.0435,
      "step": 160710
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.2542743682861328,
      "learning_rate": 0.00012555971236555624,
      "loss": 2.2284,
      "step": 160711
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.110076665878296,
      "learning_rate": 0.00012555638442497737,
      "loss": 2.2951,
      "step": 160712
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.235764741897583,
      "learning_rate": 0.00012555305651683075,
      "loss": 1.8667,
      "step": 160713
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.2845947742462158,
      "learning_rate": 0.00012554972864111722,
      "loss": 2.103,
      "step": 160714
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.380218744277954,
      "learning_rate": 0.00012554640079783717,
      "loss": 2.0071,
      "step": 160715
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.2419555187225342,
      "learning_rate": 0.00012554307298699155,
      "loss": 2.1695,
      "step": 160716
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.1609079837799072,
      "learning_rate": 0.00012553974520858056,
      "loss": 1.959,
      "step": 160717
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.103594422340393,
      "learning_rate": 0.00012553641746260513,
      "loss": 1.9123,
      "step": 160718
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.2530757188796997,
      "learning_rate": 0.00012553308974906567,
      "loss": 2.0702,
      "step": 160719
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.1540615558624268,
      "learning_rate": 0.00012552976206796302,
      "loss": 2.0199,
      "step": 160720
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.2241805791854858,
      "learning_rate": 0.00012552643441929756,
      "loss": 1.9142,
      "step": 160721
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.1558499336242676,
      "learning_rate": 0.00012552310680307024,
      "loss": 2.1561,
      "step": 160722
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.2092255353927612,
      "learning_rate": 0.00012551977921928124,
      "loss": 1.9284,
      "step": 160723
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.09292733669281,
      "learning_rate": 0.00012551645166793154,
      "loss": 1.9098,
      "step": 160724
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.2902395725250244,
      "learning_rate": 0.0001255131241490215,
      "loss": 1.9344,
      "step": 160725
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.1934555768966675,
      "learning_rate": 0.00012550979666255197,
      "loss": 2.0692,
      "step": 160726
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.210572600364685,
      "learning_rate": 0.0001255064692085234,
      "loss": 2.0874,
      "step": 160727
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.2444700002670288,
      "learning_rate": 0.00012550314178693664,
      "loss": 2.0241,
      "step": 160728
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.241635799407959,
      "learning_rate": 0.00012549981439779193,
      "loss": 2.3527,
      "step": 160729
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.2568989992141724,
      "learning_rate": 0.00012549648704109022,
      "loss": 1.9279,
      "step": 160730
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.2457319498062134,
      "learning_rate": 0.00012549315971683187,
      "loss": 2.3051,
      "step": 160731
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.2132712602615356,
      "learning_rate": 0.00012548983242501775,
      "loss": 2.0529,
      "step": 160732
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.3014018535614014,
      "learning_rate": 0.00012548650516564828,
      "loss": 1.9247,
      "step": 160733
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.250356912612915,
      "learning_rate": 0.00012548317793872437,
      "loss": 1.9347,
      "step": 160734
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.1371155977249146,
      "learning_rate": 0.00012547985074424625,
      "loss": 2.1792,
      "step": 160735
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.1932387351989746,
      "learning_rate": 0.0001254765235822148,
      "loss": 2.1401,
      "step": 160736
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.1478310823440552,
      "learning_rate": 0.00012547319645263042,
      "loss": 1.9593,
      "step": 160737
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.2053303718566895,
      "learning_rate": 0.00012546986935549397,
      "loss": 2.0621,
      "step": 160738
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.2123256921768188,
      "learning_rate": 0.0001254665422908059,
      "loss": 1.994,
      "step": 160739
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.1736961603164673,
      "learning_rate": 0.00012546321525856713,
      "loss": 1.8437,
      "step": 160740
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.3372564315795898,
      "learning_rate": 0.00012545988825877778,
      "loss": 2.0736,
      "step": 160741
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.2536817789077759,
      "learning_rate": 0.00012545656129143884,
      "loss": 2.2373,
      "step": 160742
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.2590893507003784,
      "learning_rate": 0.00012545323435655072,
      "loss": 2.1835,
      "step": 160743
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.2088780403137207,
      "learning_rate": 0.00012544990745411426,
      "loss": 1.9876,
      "step": 160744
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.1551274061203003,
      "learning_rate": 0.00012544658058412985,
      "loss": 2.1149,
      "step": 160745
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.2499005794525146,
      "learning_rate": 0.00012544325374659832,
      "loss": 2.0842,
      "step": 160746
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.3367043733596802,
      "learning_rate": 0.00012543992694152015,
      "loss": 1.9593,
      "step": 160747
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.2819360494613647,
      "learning_rate": 0.00012543660016889606,
      "loss": 2.1254,
      "step": 160748
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.1021862030029297,
      "learning_rate": 0.00012543327342872645,
      "loss": 2.0567,
      "step": 160749
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.205686092376709,
      "learning_rate": 0.00012542994672101217,
      "loss": 2.0178,
      "step": 160750
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.181841492652893,
      "learning_rate": 0.0001254266200457537,
      "loss": 1.8331,
      "step": 160751
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.2080672979354858,
      "learning_rate": 0.00012542329340295178,
      "loss": 1.9839,
      "step": 160752
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.1378172636032104,
      "learning_rate": 0.00012541996679260703,
      "loss": 1.9494,
      "step": 160753
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.2617360353469849,
      "learning_rate": 0.00012541664021471994,
      "loss": 2.1207,
      "step": 160754
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.2357549667358398,
      "learning_rate": 0.00012541331366929112,
      "loss": 2.157,
      "step": 160755
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.2183504104614258,
      "learning_rate": 0.00012540998715632136,
      "loss": 1.816,
      "step": 160756
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.2957377433776855,
      "learning_rate": 0.00012540666067581107,
      "loss": 1.9928,
      "step": 160757
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.1761252880096436,
      "learning_rate": 0.0001254033342277611,
      "loss": 2.0107,
      "step": 160758
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.1238853931427002,
      "learning_rate": 0.0001254000078121719,
      "loss": 1.9265,
      "step": 160759
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.2461744546890259,
      "learning_rate": 0.00012539668142904416,
      "loss": 2.0937,
      "step": 160760
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.2774252891540527,
      "learning_rate": 0.00012539335507837839,
      "loss": 1.9889,
      "step": 160761
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.1351439952850342,
      "learning_rate": 0.00012539002876017538,
      "loss": 1.9676,
      "step": 160762
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.21894109249115,
      "learning_rate": 0.00012538670247443556,
      "loss": 2.3801,
      "step": 160763
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.1719269752502441,
      "learning_rate": 0.00012538337622115974,
      "loss": 2.0495,
      "step": 160764
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.2996736764907837,
      "learning_rate": 0.00012538005000034848,
      "loss": 1.9815,
      "step": 160765
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.1482282876968384,
      "learning_rate": 0.00012537672381200232,
      "loss": 2.0982,
      "step": 160766
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.223525881767273,
      "learning_rate": 0.0001253733976561218,
      "loss": 2.1288,
      "step": 160767
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.1691501140594482,
      "learning_rate": 0.00012537007153270782,
      "loss": 1.9989,
      "step": 160768
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.2129677534103394,
      "learning_rate": 0.00012536674544176072,
      "loss": 2.2723,
      "step": 160769
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.2739924192428589,
      "learning_rate": 0.00012536341938328133,
      "loss": 2.1287,
      "step": 160770
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.2813788652420044,
      "learning_rate": 0.0001253600933572702,
      "loss": 2.0188,
      "step": 160771
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.2913830280303955,
      "learning_rate": 0.00012535676736372781,
      "loss": 2.04,
      "step": 160772
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.3563016653060913,
      "learning_rate": 0.000125353441402655,
      "loss": 1.8974,
      "step": 160773
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.2076668739318848,
      "learning_rate": 0.00012535011547405226,
      "loss": 2.0424,
      "step": 160774
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.2469313144683838,
      "learning_rate": 0.00012534678957792017,
      "loss": 2.3118,
      "step": 160775
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.2997502088546753,
      "learning_rate": 0.0001253434637142595,
      "loss": 2.0443,
      "step": 160776
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.2022714614868164,
      "learning_rate": 0.00012534013788307073,
      "loss": 2.0529,
      "step": 160777
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.1606782674789429,
      "learning_rate": 0.00012533681208435447,
      "loss": 1.9829,
      "step": 160778
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.4780555963516235,
      "learning_rate": 0.00012533348631811147,
      "loss": 1.982,
      "step": 160779
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.1169883012771606,
      "learning_rate": 0.0001253301605843422,
      "loss": 2.1294,
      "step": 160780
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.1477147340774536,
      "learning_rate": 0.00012532683488304742,
      "loss": 2.0321,
      "step": 160781
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.2508645057678223,
      "learning_rate": 0.00012532350921422772,
      "loss": 2.1052,
      "step": 160782
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.2280888557434082,
      "learning_rate": 0.00012532018357788363,
      "loss": 1.8995,
      "step": 160783
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.1771894693374634,
      "learning_rate": 0.00012531685797401572,
      "loss": 1.9676,
      "step": 160784
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.1308249235153198,
      "learning_rate": 0.00012531353240262486,
      "loss": 2.0024,
      "step": 160785
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.340185284614563,
      "learning_rate": 0.00012531020686371133,
      "loss": 2.0311,
      "step": 160786
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.2933634519577026,
      "learning_rate": 0.00012530688135727608,
      "loss": 2.0019,
      "step": 160787
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.3335061073303223,
      "learning_rate": 0.00012530355588331958,
      "loss": 2.1548,
      "step": 160788
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.2897603511810303,
      "learning_rate": 0.00012530023044184243,
      "loss": 1.9755,
      "step": 160789
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.1566332578659058,
      "learning_rate": 0.00012529690503284516,
      "loss": 2.023,
      "step": 160790
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.1807926893234253,
      "learning_rate": 0.00012529357965632861,
      "loss": 2.146,
      "step": 160791
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.1117300987243652,
      "learning_rate": 0.00012529025431229316,
      "loss": 2.4342,
      "step": 160792
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.1434444189071655,
      "learning_rate": 0.00012528692900073968,
      "loss": 2.1111,
      "step": 160793
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.2563629150390625,
      "learning_rate": 0.00012528360372166854,
      "loss": 2.0324,
      "step": 160794
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.1433106660842896,
      "learning_rate": 0.0001252802784750807,
      "loss": 2.188,
      "step": 160795
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.172531247138977,
      "learning_rate": 0.0001252769532609763,
      "loss": 1.9403,
      "step": 160796
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.1788150072097778,
      "learning_rate": 0.00012527362807935636,
      "loss": 1.8489,
      "step": 160797
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.2163760662078857,
      "learning_rate": 0.00012527030293022122,
      "loss": 1.9107,
      "step": 160798
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.2748292684555054,
      "learning_rate": 0.00012526697781357177,
      "loss": 1.821,
      "step": 160799
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.193843960762024,
      "learning_rate": 0.00012526365272940836,
      "loss": 2.1299,
      "step": 160800
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.340633511543274,
      "learning_rate": 0.00012526032767773193,
      "loss": 2.1497,
      "step": 160801
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.2306634187698364,
      "learning_rate": 0.00012525700265854272,
      "loss": 1.933,
      "step": 160802
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.2207090854644775,
      "learning_rate": 0.0001252536776718416,
      "loss": 2.0743,
      "step": 160803
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.1364799737930298,
      "learning_rate": 0.00012525035271762907,
      "loss": 1.8948,
      "step": 160804
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.1265201568603516,
      "learning_rate": 0.00012524702779590584,
      "loss": 1.9297,
      "step": 160805
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.4402343034744263,
      "learning_rate": 0.00012524370290667244,
      "loss": 2.0127,
      "step": 160806
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.2196935415267944,
      "learning_rate": 0.0001252403780499297,
      "loss": 2.1267,
      "step": 160807
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.334855318069458,
      "learning_rate": 0.0001252370532256779,
      "loss": 2.0025,
      "step": 160808
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.1364432573318481,
      "learning_rate": 0.00012523372843391788,
      "loss": 2.0309,
      "step": 160809
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.1465754508972168,
      "learning_rate": 0.00012523040367465015,
      "loss": 2.1254,
      "step": 160810
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.2382452487945557,
      "learning_rate": 0.0001252270789478755,
      "loss": 2.0662,
      "step": 160811
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.2809104919433594,
      "learning_rate": 0.00012522375425359428,
      "loss": 1.9188,
      "step": 160812
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.2904146909713745,
      "learning_rate": 0.0001252204295918074,
      "loss": 2.0699,
      "step": 160813
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.2783418893814087,
      "learning_rate": 0.00012521710496251538,
      "loss": 2.137,
      "step": 160814
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.2095084190368652,
      "learning_rate": 0.00012521378036571876,
      "loss": 2.2464,
      "step": 160815
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.4332003593444824,
      "learning_rate": 0.0001252104558014181,
      "loss": 2.0958,
      "step": 160816
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.623682975769043,
      "learning_rate": 0.00012520713126961422,
      "loss": 2.0719,
      "step": 160817
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.2112542390823364,
      "learning_rate": 0.0001252038067703075,
      "loss": 2.009,
      "step": 160818
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.200314998626709,
      "learning_rate": 0.00012520048230349882,
      "loss": 2.125,
      "step": 160819
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.2561562061309814,
      "learning_rate": 0.00012519715786918866,
      "loss": 1.9947,
      "step": 160820
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.2682101726531982,
      "learning_rate": 0.00012519383346737767,
      "loss": 2.0489,
      "step": 160821
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.5813649892807007,
      "learning_rate": 0.00012519050909806631,
      "loss": 2.0727,
      "step": 160822
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.2071096897125244,
      "learning_rate": 0.00012518718476125547,
      "loss": 1.845,
      "step": 160823
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.1221894025802612,
      "learning_rate": 0.0001251838604569455,
      "loss": 2.0167,
      "step": 160824
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.2373250722885132,
      "learning_rate": 0.00012518053618513727,
      "loss": 1.9652,
      "step": 160825
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.1990349292755127,
      "learning_rate": 0.0001251772119458313,
      "loss": 1.9549,
      "step": 160826
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.5101674795150757,
      "learning_rate": 0.00012517388773902816,
      "loss": 2.0262,
      "step": 160827
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.163482427597046,
      "learning_rate": 0.0001251705635647284,
      "loss": 2.0609,
      "step": 160828
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.3336268663406372,
      "learning_rate": 0.00012516723942293282,
      "loss": 2.2705,
      "step": 160829
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.0914688110351562,
      "learning_rate": 0.00012516391531364185,
      "loss": 1.9606,
      "step": 160830
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.189221739768982,
      "learning_rate": 0.00012516059123685632,
      "loss": 2.0127,
      "step": 160831
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.2433627843856812,
      "learning_rate": 0.00012515726719257677,
      "loss": 2.0863,
      "step": 160832
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.2285386323928833,
      "learning_rate": 0.00012515394318080375,
      "loss": 2.3321,
      "step": 160833
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.2003085613250732,
      "learning_rate": 0.00012515061920153781,
      "loss": 2.0321,
      "step": 160834
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.234312891960144,
      "learning_rate": 0.00012514729525477976,
      "loss": 2.0498,
      "step": 160835
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.1292555332183838,
      "learning_rate": 0.00012514397134053007,
      "loss": 1.9642,
      "step": 160836
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.2115888595581055,
      "learning_rate": 0.00012514064745878947,
      "loss": 2.0327,
      "step": 160837
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.10592782497406,
      "learning_rate": 0.00012513732360955852,
      "loss": 2.0816,
      "step": 160838
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.1152743101119995,
      "learning_rate": 0.0001251339997928379,
      "loss": 1.8191,
      "step": 160839
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.2183914184570312,
      "learning_rate": 0.000125130676008628,
      "loss": 2.1452,
      "step": 160840
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.2036038637161255,
      "learning_rate": 0.00012512735225692976,
      "loss": 1.9366,
      "step": 160841
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.469508171081543,
      "learning_rate": 0.00012512402853774352,
      "loss": 2.066,
      "step": 160842
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.2945584058761597,
      "learning_rate": 0.00012512070485107014,
      "loss": 2.1206,
      "step": 160843
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.292188286781311,
      "learning_rate": 0.0001251173811969101,
      "loss": 2.0335,
      "step": 160844
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.235277771949768,
      "learning_rate": 0.0001251140575752641,
      "loss": 1.9243,
      "step": 160845
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.2453320026397705,
      "learning_rate": 0.00012511073398613253,
      "loss": 2.2495,
      "step": 160846
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.7906484603881836,
      "learning_rate": 0.00012510741042951628,
      "loss": 2.0882,
      "step": 160847
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.20167076587677,
      "learning_rate": 0.0001251040869054158,
      "loss": 2.1223,
      "step": 160848
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.6078505516052246,
      "learning_rate": 0.00012510076341383186,
      "loss": 2.1278,
      "step": 160849
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.282530426979065,
      "learning_rate": 0.00012509743995476496,
      "loss": 1.9716,
      "step": 160850
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.1236023902893066,
      "learning_rate": 0.0001250941165282158,
      "loss": 1.8332,
      "step": 160851
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.3666502237319946,
      "learning_rate": 0.0001250907931341848,
      "loss": 2.0085,
      "step": 160852
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.1403049230575562,
      "learning_rate": 0.00012508746977267285,
      "loss": 1.8743,
      "step": 160853
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.268200397491455,
      "learning_rate": 0.0001250841464436803,
      "loss": 1.8632,
      "step": 160854
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.1668105125427246,
      "learning_rate": 0.00012508082314720807,
      "loss": 2.0307,
      "step": 160855
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.2145766019821167,
      "learning_rate": 0.0001250774998832566,
      "loss": 1.8982,
      "step": 160856
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.110891580581665,
      "learning_rate": 0.0001250741766518264,
      "loss": 2.1354,
      "step": 160857
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.206388235092163,
      "learning_rate": 0.0001250708534529183,
      "loss": 2.1739,
      "step": 160858
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.2336505651474,
      "learning_rate": 0.00012506753028653287,
      "loss": 2.1606,
      "step": 160859
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.2778462171554565,
      "learning_rate": 0.0001250642071526706,
      "loss": 2.081,
      "step": 160860
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.1193617582321167,
      "learning_rate": 0.00012506088405133225,
      "loss": 2.2095,
      "step": 160861
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.1933245658874512,
      "learning_rate": 0.0001250575609825184,
      "loss": 1.9516,
      "step": 160862
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.192047119140625,
      "learning_rate": 0.00012505423794622955,
      "loss": 2.1182,
      "step": 160863
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.1970549821853638,
      "learning_rate": 0.00012505091494246652,
      "loss": 1.9817,
      "step": 160864
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.1761749982833862,
      "learning_rate": 0.00012504759197122986,
      "loss": 2.0682,
      "step": 160865
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.1871731281280518,
      "learning_rate": 0.00012504426903252,
      "loss": 2.0272,
      "step": 160866
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.2422566413879395,
      "learning_rate": 0.00012504094612633785,
      "loss": 1.9978,
      "step": 160867
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.2176425457000732,
      "learning_rate": 0.00012503762325268388,
      "loss": 1.8844,
      "step": 160868
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.1686193943023682,
      "learning_rate": 0.00012503430041155863,
      "loss": 2.1979,
      "step": 160869
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.195101261138916,
      "learning_rate": 0.0001250309776029629,
      "loss": 1.7435,
      "step": 160870
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.3647593259811401,
      "learning_rate": 0.0001250276548268971,
      "loss": 2.1492,
      "step": 160871
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.1100867986679077,
      "learning_rate": 0.00012502433208336203,
      "loss": 1.7984,
      "step": 160872
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.1247799396514893,
      "learning_rate": 0.0001250210093723583,
      "loss": 2.0303,
      "step": 160873
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.3116481304168701,
      "learning_rate": 0.00012501768669388649,
      "loss": 1.9502,
      "step": 160874
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.2899757623672485,
      "learning_rate": 0.000125014364047947,
      "loss": 2.2728,
      "step": 160875
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.2116076946258545,
      "learning_rate": 0.00012501104143454082,
      "loss": 2.2104,
      "step": 160876
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.2374626398086548,
      "learning_rate": 0.00012500771885366825,
      "loss": 2.0829,
      "step": 160877
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.2400850057601929,
      "learning_rate": 0.0001250043963053302,
      "loss": 1.8908,
      "step": 160878
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.5172655582427979,
      "learning_rate": 0.00012500107378952697,
      "loss": 2.0572,
      "step": 160879
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.2274717092514038,
      "learning_rate": 0.00012499775130625948,
      "loss": 2.0729,
      "step": 160880
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.29153573513031,
      "learning_rate": 0.0001249944288555282,
      "loss": 2.1117,
      "step": 160881
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.0906785726547241,
      "learning_rate": 0.00012499110643733377,
      "loss": 1.9907,
      "step": 160882
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.3288140296936035,
      "learning_rate": 0.0001249877840516767,
      "loss": 2.0726,
      "step": 160883
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.488141417503357,
      "learning_rate": 0.0001249844616985578,
      "loss": 1.8082,
      "step": 160884
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.2711784839630127,
      "learning_rate": 0.00012498113937797748,
      "loss": 1.9358,
      "step": 160885
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.3080426454544067,
      "learning_rate": 0.00012497781708993658,
      "loss": 2.1147,
      "step": 160886
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.2329528331756592,
      "learning_rate": 0.0001249744948344356,
      "loss": 2.1249,
      "step": 160887
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.1463571786880493,
      "learning_rate": 0.00012497117261147517,
      "loss": 2.1301,
      "step": 160888
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.2404940128326416,
      "learning_rate": 0.0001249678504210558,
      "loss": 1.9406,
      "step": 160889
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.2127412557601929,
      "learning_rate": 0.00012496452826317835,
      "loss": 1.8903,
      "step": 160890
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.218214750289917,
      "learning_rate": 0.00012496120613784316,
      "loss": 1.7558,
      "step": 160891
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.1270078420639038,
      "learning_rate": 0.00012495788404505106,
      "loss": 2.0197,
      "step": 160892
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.390573263168335,
      "learning_rate": 0.00012495456198480264,
      "loss": 1.9982,
      "step": 160893
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.279426097869873,
      "learning_rate": 0.00012495123995709845,
      "loss": 2.0868,
      "step": 160894
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.224665880203247,
      "learning_rate": 0.00012494791796193903,
      "loss": 1.8496,
      "step": 160895
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.153228759765625,
      "learning_rate": 0.0001249445959993252,
      "loss": 2.0699,
      "step": 160896
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.1662116050720215,
      "learning_rate": 0.00012494127406925738,
      "loss": 2.0979,
      "step": 160897
    },
    {
      "epoch": 2.09,
      "grad_norm": 1.1359256505966187,
      "learning_rate": 0.00012493795217173635,
      "loss": 1.9748,
      "step": 160898
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.2363954782485962,
      "learning_rate": 0.0001249346303067627,
      "loss": 2.0881,
      "step": 160899
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.3022195100784302,
      "learning_rate": 0.000124931308474337,
      "loss": 1.9833,
      "step": 160900
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.238338828086853,
      "learning_rate": 0.00012492798667445976,
      "loss": 2.0284,
      "step": 160901
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.1776199340820312,
      "learning_rate": 0.0001249246649071318,
      "loss": 1.9275,
      "step": 160902
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.4572031497955322,
      "learning_rate": 0.00012492134317235355,
      "loss": 2.1548,
      "step": 160903
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.1868712902069092,
      "learning_rate": 0.00012491802147012587,
      "loss": 2.0087,
      "step": 160904
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.2208402156829834,
      "learning_rate": 0.0001249146998004492,
      "loss": 1.929,
      "step": 160905
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.227473258972168,
      "learning_rate": 0.0001249113781633242,
      "loss": 2.077,
      "step": 160906
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.2562443017959595,
      "learning_rate": 0.00012490805655875138,
      "loss": 2.015,
      "step": 160907
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.330186128616333,
      "learning_rate": 0.0001249047349867316,
      "loss": 1.9148,
      "step": 160908
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.3853240013122559,
      "learning_rate": 0.00012490141344726516,
      "loss": 1.9471,
      "step": 160909
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.2117443084716797,
      "learning_rate": 0.000124898091940353,
      "loss": 2.2287,
      "step": 160910
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.4701757431030273,
      "learning_rate": 0.0001248947704659956,
      "loss": 2.2042,
      "step": 160911
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.3625497817993164,
      "learning_rate": 0.00012489144902419353,
      "loss": 2.1511,
      "step": 160912
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.0935297012329102,
      "learning_rate": 0.00012488812761494736,
      "loss": 2.2597,
      "step": 160913
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.0948539972305298,
      "learning_rate": 0.00012488480623825787,
      "loss": 1.8898,
      "step": 160914
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.2470232248306274,
      "learning_rate": 0.00012488148489412553,
      "loss": 2.0723,
      "step": 160915
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.2891590595245361,
      "learning_rate": 0.00012487816358255113,
      "loss": 2.1549,
      "step": 160916
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.096938133239746,
      "learning_rate": 0.00012487484230353516,
      "loss": 1.9291,
      "step": 160917
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.2187855243682861,
      "learning_rate": 0.0001248715210570783,
      "loss": 1.8403,
      "step": 160918
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.2512013912200928,
      "learning_rate": 0.000124868199843181,
      "loss": 1.8563,
      "step": 160919
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.1823046207427979,
      "learning_rate": 0.0001248648786618441,
      "loss": 1.9252,
      "step": 160920
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.1927021741867065,
      "learning_rate": 0.000124861557513068,
      "loss": 1.9933,
      "step": 160921
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.2336323261260986,
      "learning_rate": 0.0001248582363968536,
      "loss": 2.0303,
      "step": 160922
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.2164736986160278,
      "learning_rate": 0.00012485491531320135,
      "loss": 2.1493,
      "step": 160923
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.091709017753601,
      "learning_rate": 0.0001248515942621119,
      "loss": 1.951,
      "step": 160924
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.1716729402542114,
      "learning_rate": 0.00012484827324358568,
      "loss": 1.9534,
      "step": 160925
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.2012076377868652,
      "learning_rate": 0.0001248449522576236,
      "loss": 2.058,
      "step": 160926
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.1913959980010986,
      "learning_rate": 0.00012484163130422606,
      "loss": 1.9257,
      "step": 160927
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.1568142175674438,
      "learning_rate": 0.00012483831038339387,
      "loss": 2.1946,
      "step": 160928
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.168205738067627,
      "learning_rate": 0.00012483498949512754,
      "loss": 2.1927,
      "step": 160929
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.24117910861969,
      "learning_rate": 0.0001248316686394277,
      "loss": 1.9665,
      "step": 160930
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.3526369333267212,
      "learning_rate": 0.0001248283478162948,
      "loss": 1.8552,
      "step": 160931
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.1566864252090454,
      "learning_rate": 0.00012482502702572978,
      "loss": 2.0275,
      "step": 160932
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.1915181875228882,
      "learning_rate": 0.000124821706267733,
      "loss": 2.251,
      "step": 160933
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.383568525314331,
      "learning_rate": 0.00012481838554230523,
      "loss": 1.9539,
      "step": 160934
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.2700728178024292,
      "learning_rate": 0.00012481506484944706,
      "loss": 2.0092,
      "step": 160935
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.195448875427246,
      "learning_rate": 0.00012481174418915906,
      "loss": 1.9828,
      "step": 160936
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.3782994747161865,
      "learning_rate": 0.00012480842356144177,
      "loss": 2.2567,
      "step": 160937
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.2638779878616333,
      "learning_rate": 0.00012480510296629602,
      "loss": 2.0242,
      "step": 160938
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.2016853094100952,
      "learning_rate": 0.00012480178240372217,
      "loss": 2.0981,
      "step": 160939
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.2238143682479858,
      "learning_rate": 0.00012479846187372112,
      "loss": 1.8466,
      "step": 160940
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.2604451179504395,
      "learning_rate": 0.0001247951413762933,
      "loss": 1.6695,
      "step": 160941
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.1452373266220093,
      "learning_rate": 0.0001247918209114393,
      "loss": 2.0829,
      "step": 160942
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.1949317455291748,
      "learning_rate": 0.0001247885004791599,
      "loss": 2.139,
      "step": 160943
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.3483617305755615,
      "learning_rate": 0.0001247851800794556,
      "loss": 1.9651,
      "step": 160944
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.4319039583206177,
      "learning_rate": 0.00012478185971232696,
      "loss": 2.0905,
      "step": 160945
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.22065269947052,
      "learning_rate": 0.0001247785393777748,
      "loss": 2.0086,
      "step": 160946
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.150267481803894,
      "learning_rate": 0.0001247752190757995,
      "loss": 2.2076,
      "step": 160947
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.2493356466293335,
      "learning_rate": 0.0001247718988064019,
      "loss": 2.0621,
      "step": 160948
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.2345911264419556,
      "learning_rate": 0.0001247685785695825,
      "loss": 1.9391,
      "step": 160949
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.10334050655365,
      "learning_rate": 0.0001247652583653419,
      "loss": 1.9965,
      "step": 160950
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.1856383085250854,
      "learning_rate": 0.0001247619381936807,
      "loss": 2.0948,
      "step": 160951
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.218695044517517,
      "learning_rate": 0.00012475861805459965,
      "loss": 1.9617,
      "step": 160952
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.1999304294586182,
      "learning_rate": 0.0001247552979480992,
      "loss": 2.0144,
      "step": 160953
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.148751974105835,
      "learning_rate": 0.0001247519778741801,
      "loss": 1.8881,
      "step": 160954
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.0696678161621094,
      "learning_rate": 0.00012474865783284296,
      "loss": 1.8488,
      "step": 160955
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.2628673315048218,
      "learning_rate": 0.0001247453378240882,
      "loss": 1.8708,
      "step": 160956
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.4136778116226196,
      "learning_rate": 0.00012474201784791675,
      "loss": 2.1586,
      "step": 160957
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.1783479452133179,
      "learning_rate": 0.00012473869790432905,
      "loss": 1.9949,
      "step": 160958
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.3789359331130981,
      "learning_rate": 0.0001247353779933256,
      "loss": 2.1708,
      "step": 160959
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.3530333042144775,
      "learning_rate": 0.00012473205811490727,
      "loss": 2.0356,
      "step": 160960
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.361306071281433,
      "learning_rate": 0.00012472873826907457,
      "loss": 2.1579,
      "step": 160961
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.4940749406814575,
      "learning_rate": 0.000124725418455828,
      "loss": 2.1038,
      "step": 160962
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.2979469299316406,
      "learning_rate": 0.0001247220986751684,
      "loss": 1.9965,
      "step": 160963
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.1653947830200195,
      "learning_rate": 0.00012471877892709625,
      "loss": 1.9635,
      "step": 160964
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.263514757156372,
      "learning_rate": 0.00012471545921161208,
      "loss": 2.024,
      "step": 160965
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.2419029474258423,
      "learning_rate": 0.00012471213952871673,
      "loss": 2.2018,
      "step": 160966
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.1835110187530518,
      "learning_rate": 0.00012470881987841073,
      "loss": 2.0987,
      "step": 160967
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.1620550155639648,
      "learning_rate": 0.00012470550026069452,
      "loss": 2.1697,
      "step": 160968
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.186844825744629,
      "learning_rate": 0.000124702180675569,
      "loss": 2.0889,
      "step": 160969
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.2425789833068848,
      "learning_rate": 0.00012469886112303452,
      "loss": 1.9439,
      "step": 160970
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.1758569478988647,
      "learning_rate": 0.00012469554160309196,
      "loss": 1.985,
      "step": 160971
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.2726101875305176,
      "learning_rate": 0.0001246922221157418,
      "loss": 2.1685,
      "step": 160972
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.247862696647644,
      "learning_rate": 0.00012468890266098466,
      "loss": 2.1981,
      "step": 160973
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.1845506429672241,
      "learning_rate": 0.00012468558323882104,
      "loss": 1.9658,
      "step": 160974
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.2364710569381714,
      "learning_rate": 0.0001246822638492518,
      "loss": 2.326,
      "step": 160975
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.2495083808898926,
      "learning_rate": 0.00012467894449227736,
      "loss": 1.9242,
      "step": 160976
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.1396108865737915,
      "learning_rate": 0.0001246756251678985,
      "loss": 2.0106,
      "step": 160977
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.294460415840149,
      "learning_rate": 0.00012467230587611574,
      "loss": 2.3705,
      "step": 160978
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.15725839138031,
      "learning_rate": 0.00012466898661692974,
      "loss": 1.972,
      "step": 160979
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.2094955444335938,
      "learning_rate": 0.0001246656673903409,
      "loss": 1.7063,
      "step": 160980
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.1768865585327148,
      "learning_rate": 0.00012466234819635022,
      "loss": 2.0308,
      "step": 160981
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.3072844743728638,
      "learning_rate": 0.00012465902903495795,
      "loss": 1.9176,
      "step": 160982
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.2213475704193115,
      "learning_rate": 0.000124655709906165,
      "loss": 2.0722,
      "step": 160983
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.3237863779067993,
      "learning_rate": 0.0001246523908099719,
      "loss": 2.154,
      "step": 160984
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.2644950151443481,
      "learning_rate": 0.0001246490717463792,
      "loss": 2.0534,
      "step": 160985
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.216286301612854,
      "learning_rate": 0.00012464575271538744,
      "loss": 2.1182,
      "step": 160986
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.2676684856414795,
      "learning_rate": 0.00012464243371699746,
      "loss": 2.0873,
      "step": 160987
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.1985331773757935,
      "learning_rate": 0.00012463911475120966,
      "loss": 2.0265,
      "step": 160988
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.2019457817077637,
      "learning_rate": 0.00012463579581802484,
      "loss": 2.121,
      "step": 160989
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.156102180480957,
      "learning_rate": 0.00012463247691744353,
      "loss": 1.9318,
      "step": 160990
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.1700596809387207,
      "learning_rate": 0.0001246291580494664,
      "loss": 2.2173,
      "step": 160991
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.2351619005203247,
      "learning_rate": 0.00012462583921409387,
      "loss": 1.9498,
      "step": 160992
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.1536519527435303,
      "learning_rate": 0.00012462252041132684,
      "loss": 1.9569,
      "step": 160993
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.246532917022705,
      "learning_rate": 0.00012461920164116567,
      "loss": 2.2158,
      "step": 160994
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.2575657367706299,
      "learning_rate": 0.0001246158829036112,
      "loss": 2.2842,
      "step": 160995
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.165831446647644,
      "learning_rate": 0.00012461256419866396,
      "loss": 1.8475,
      "step": 160996
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.1869080066680908,
      "learning_rate": 0.00012460924552632456,
      "loss": 1.8989,
      "step": 160997
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.2632595300674438,
      "learning_rate": 0.0001246059268865935,
      "loss": 2.0129,
      "step": 160998
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.3178582191467285,
      "learning_rate": 0.00012460260827947163,
      "loss": 2.1803,
      "step": 160999
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.2502357959747314,
      "learning_rate": 0.00012459928970495933,
      "loss": 1.8698,
      "step": 161000
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.1903460025787354,
      "learning_rate": 0.00012459597116305744,
      "loss": 1.9743,
      "step": 161001
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.4289028644561768,
      "learning_rate": 0.00012459265265376646,
      "loss": 2.0956,
      "step": 161002
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.090778112411499,
      "learning_rate": 0.00012458933417708703,
      "loss": 1.9611,
      "step": 161003
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.3409801721572876,
      "learning_rate": 0.00012458601573301964,
      "loss": 2.038,
      "step": 161004
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.3376270532608032,
      "learning_rate": 0.00012458269732156514,
      "loss": 2.1202,
      "step": 161005
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.349446177482605,
      "learning_rate": 0.0001245793789427239,
      "loss": 2.156,
      "step": 161006
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.2582417726516724,
      "learning_rate": 0.00012457606059649678,
      "loss": 2.1829,
      "step": 161007
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.1960071325302124,
      "learning_rate": 0.0001245727422828843,
      "loss": 1.9296,
      "step": 161008
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.2497055530548096,
      "learning_rate": 0.00012456942400188703,
      "loss": 2.2124,
      "step": 161009
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.2225701808929443,
      "learning_rate": 0.0001245661057535055,
      "loss": 2.175,
      "step": 161010
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.195375919342041,
      "learning_rate": 0.0001245627875377406,
      "loss": 1.9634,
      "step": 161011
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.1613736152648926,
      "learning_rate": 0.00012455946935459265,
      "loss": 2.3853,
      "step": 161012
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.3329815864562988,
      "learning_rate": 0.0001245561512040625,
      "loss": 1.9658,
      "step": 161013
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.197265386581421,
      "learning_rate": 0.00012455283308615057,
      "loss": 1.8605,
      "step": 161014
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.169717788696289,
      "learning_rate": 0.00012454951500085783,
      "loss": 1.8841,
      "step": 161015
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.2311697006225586,
      "learning_rate": 0.0001245461969481844,
      "loss": 2.0704,
      "step": 161016
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.3429009914398193,
      "learning_rate": 0.00012454287892813126,
      "loss": 2.15,
      "step": 161017
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.0913352966308594,
      "learning_rate": 0.00012453956094069882,
      "loss": 2.1751,
      "step": 161018
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.2321598529815674,
      "learning_rate": 0.00012453624298588788,
      "loss": 2.0305,
      "step": 161019
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.240041732788086,
      "learning_rate": 0.00012453292506369885,
      "loss": 2.1006,
      "step": 161020
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.202069640159607,
      "learning_rate": 0.00012452960717413265,
      "loss": 1.978,
      "step": 161021
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.0776420831680298,
      "learning_rate": 0.00012452628931718953,
      "loss": 1.9955,
      "step": 161022
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.0768705606460571,
      "learning_rate": 0.00012452297149287043,
      "loss": 1.7574,
      "step": 161023
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.1675676107406616,
      "learning_rate": 0.00012451965370117565,
      "loss": 1.9061,
      "step": 161024
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.2077105045318604,
      "learning_rate": 0.0001245163359421061,
      "loss": 2.1089,
      "step": 161025
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.1073737144470215,
      "learning_rate": 0.0001245130182156622,
      "loss": 1.9382,
      "step": 161026
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.2056716680526733,
      "learning_rate": 0.00012450970052184482,
      "loss": 1.9663,
      "step": 161027
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.1203372478485107,
      "learning_rate": 0.0001245063828606542,
      "loss": 2.2428,
      "step": 161028
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.2133136987686157,
      "learning_rate": 0.00012450306523209128,
      "loss": 1.9613,
      "step": 161029
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.162529706954956,
      "learning_rate": 0.0001244997476361564,
      "loss": 2.1952,
      "step": 161030
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.1119604110717773,
      "learning_rate": 0.00012449643007285047,
      "loss": 1.9162,
      "step": 161031
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.1815447807312012,
      "learning_rate": 0.00012449311254217387,
      "loss": 2.1189,
      "step": 161032
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.383223533630371,
      "learning_rate": 0.00012448979504412738,
      "loss": 1.8769,
      "step": 161033
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.21353018283844,
      "learning_rate": 0.0001244864775787116,
      "loss": 2.0396,
      "step": 161034
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.2026396989822388,
      "learning_rate": 0.00012448316014592709,
      "loss": 1.994,
      "step": 161035
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.3086858987808228,
      "learning_rate": 0.00012447984274577433,
      "loss": 2.1056,
      "step": 161036
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.234777808189392,
      "learning_rate": 0.0001244765253782542,
      "loss": 2.1901,
      "step": 161037
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.3629631996154785,
      "learning_rate": 0.0001244732080433671,
      "loss": 2.1274,
      "step": 161038
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.408830165863037,
      "learning_rate": 0.00012446989074111386,
      "loss": 2.165,
      "step": 161039
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.2759323120117188,
      "learning_rate": 0.00012446657347149498,
      "loss": 2.1055,
      "step": 161040
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.122857689857483,
      "learning_rate": 0.00012446325623451094,
      "loss": 1.9161,
      "step": 161041
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.1468687057495117,
      "learning_rate": 0.00012445993903016262,
      "loss": 1.7064,
      "step": 161042
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.206926941871643,
      "learning_rate": 0.0001244566218584505,
      "loss": 2.0029,
      "step": 161043
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.114902377128601,
      "learning_rate": 0.00012445330471937512,
      "loss": 2.1821,
      "step": 161044
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.4552302360534668,
      "learning_rate": 0.0001244499876129373,
      "loss": 2.1589,
      "step": 161045
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.2049424648284912,
      "learning_rate": 0.0001244466705391375,
      "loss": 1.9384,
      "step": 161046
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.4266326427459717,
      "learning_rate": 0.00012444335349797632,
      "loss": 2.1518,
      "step": 161047
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.1540013551712036,
      "learning_rate": 0.0001244400364894545,
      "loss": 2.1211,
      "step": 161048
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.2763031721115112,
      "learning_rate": 0.0001244367195135726,
      "loss": 1.955,
      "step": 161049
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.1709935665130615,
      "learning_rate": 0.00012443340257033112,
      "loss": 2.168,
      "step": 161050
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.3280912637710571,
      "learning_rate": 0.0001244300856597309,
      "loss": 2.0986,
      "step": 161051
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.2088887691497803,
      "learning_rate": 0.00012442676878177245,
      "loss": 2.0783,
      "step": 161052
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.358777642250061,
      "learning_rate": 0.00012442345193645625,
      "loss": 2.1033,
      "step": 161053
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.221519947052002,
      "learning_rate": 0.00012442013512378318,
      "loss": 1.9405,
      "step": 161054
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.3636367321014404,
      "learning_rate": 0.0001244168183437536,
      "loss": 2.1143,
      "step": 161055
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.1580256223678589,
      "learning_rate": 0.00012441350159636834,
      "loss": 1.816,
      "step": 161056
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.274889588356018,
      "learning_rate": 0.00012441018488162795,
      "loss": 2.1248,
      "step": 161057
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.3902186155319214,
      "learning_rate": 0.00012440686819953302,
      "loss": 2.0578,
      "step": 161058
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.3126946687698364,
      "learning_rate": 0.00012440355155008404,
      "loss": 2.0952,
      "step": 161059
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.2085357904434204,
      "learning_rate": 0.00012440023493328188,
      "loss": 2.0928,
      "step": 161060
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.1866425275802612,
      "learning_rate": 0.0001243969183491269,
      "loss": 2.2207,
      "step": 161061
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.7740976810455322,
      "learning_rate": 0.00012439360179761997,
      "loss": 2.0031,
      "step": 161062
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.2054684162139893,
      "learning_rate": 0.0001243902852787616,
      "loss": 1.9669,
      "step": 161063
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.1818517446517944,
      "learning_rate": 0.00012438696879255238,
      "loss": 1.9386,
      "step": 161064
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.2476024627685547,
      "learning_rate": 0.0001243836523389928,
      "loss": 2.0056,
      "step": 161065
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.294346809387207,
      "learning_rate": 0.00012438033591808374,
      "loss": 1.8777,
      "step": 161066
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.3264027833938599,
      "learning_rate": 0.00012437701952982557,
      "loss": 1.7725,
      "step": 161067
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.2704354524612427,
      "learning_rate": 0.00012437370317421918,
      "loss": 1.9904,
      "step": 161068
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.2345565557479858,
      "learning_rate": 0.000124370386851265,
      "loss": 1.9197,
      "step": 161069
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.1503028869628906,
      "learning_rate": 0.0001243670705609637,
      "loss": 2.0523,
      "step": 161070
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.1839923858642578,
      "learning_rate": 0.00012436375430331575,
      "loss": 2.2362,
      "step": 161071
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.2007414102554321,
      "learning_rate": 0.000124360438078322,
      "loss": 2.1545,
      "step": 161072
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.355404019355774,
      "learning_rate": 0.00012435712188598288,
      "loss": 1.9367,
      "step": 161073
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.318947672843933,
      "learning_rate": 0.00012435380572629914,
      "loss": 2.2146,
      "step": 161074
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.3128833770751953,
      "learning_rate": 0.0001243504895992714,
      "loss": 1.9152,
      "step": 161075
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.3391715288162231,
      "learning_rate": 0.00012434717350490022,
      "loss": 2.1549,
      "step": 161076
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.156876802444458,
      "learning_rate": 0.00012434385744318605,
      "loss": 2.127,
      "step": 161077
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.0939500331878662,
      "learning_rate": 0.00012434054141412986,
      "loss": 2.0734,
      "step": 161078
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.200933814048767,
      "learning_rate": 0.00012433722541773192,
      "loss": 1.9952,
      "step": 161079
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.2003567218780518,
      "learning_rate": 0.0001243339094539931,
      "loss": 1.8911,
      "step": 161080
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.0914573669433594,
      "learning_rate": 0.00012433059352291384,
      "loss": 2.2656,
      "step": 161081
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.2411646842956543,
      "learning_rate": 0.00012432727762449507,
      "loss": 2.2656,
      "step": 161082
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.4943779706954956,
      "learning_rate": 0.00012432396175873694,
      "loss": 2.0015,
      "step": 161083
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.204775333404541,
      "learning_rate": 0.0001243206459256404,
      "loss": 2.0563,
      "step": 161084
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.063836693763733,
      "learning_rate": 0.00012431733012520583,
      "loss": 1.9722,
      "step": 161085
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.4383741617202759,
      "learning_rate": 0.00012431401435743415,
      "loss": 2.0933,
      "step": 161086
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.213606595993042,
      "learning_rate": 0.0001243106986223257,
      "loss": 2.0551,
      "step": 161087
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.0931340456008911,
      "learning_rate": 0.0001243073829198814,
      "loss": 2.0675,
      "step": 161088
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.139527440071106,
      "learning_rate": 0.00012430406725010148,
      "loss": 2.161,
      "step": 161089
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.2513830661773682,
      "learning_rate": 0.00012430075161298684,
      "loss": 2.0971,
      "step": 161090
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.255110263824463,
      "learning_rate": 0.0001242974360085379,
      "loss": 2.1739,
      "step": 161091
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.1923671960830688,
      "learning_rate": 0.00012429412043675548,
      "loss": 1.8765,
      "step": 161092
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.2309672832489014,
      "learning_rate": 0.00012429080489763998,
      "loss": 2.1075,
      "step": 161093
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.1502670049667358,
      "learning_rate": 0.00012428748939119242,
      "loss": 2.2078,
      "step": 161094
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.4875515699386597,
      "learning_rate": 0.00012428417391741287,
      "loss": 1.9745,
      "step": 161095
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.3519399166107178,
      "learning_rate": 0.0001242808584763023,
      "loss": 1.9826,
      "step": 161096
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.4330824613571167,
      "learning_rate": 0.00012427754306786116,
      "loss": 2.062,
      "step": 161097
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.4143978357315063,
      "learning_rate": 0.00012427422769209024,
      "loss": 2.0555,
      "step": 161098
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.524595856666565,
      "learning_rate": 0.00012427091234898995,
      "loss": 2.127,
      "step": 161099
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.2767549753189087,
      "learning_rate": 0.00012426759703856123,
      "loss": 2.2596,
      "step": 161100
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.3374075889587402,
      "learning_rate": 0.00012426428176080426,
      "loss": 2.0094,
      "step": 161101
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.2506985664367676,
      "learning_rate": 0.00012426096651571998,
      "loss": 2.2102,
      "step": 161102
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.2527105808258057,
      "learning_rate": 0.00012425765130330882,
      "loss": 2.2164,
      "step": 161103
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.1988849639892578,
      "learning_rate": 0.0001242543361235716,
      "loss": 1.8171,
      "step": 161104
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.256217122077942,
      "learning_rate": 0.00012425102097650867,
      "loss": 1.8519,
      "step": 161105
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.283414602279663,
      "learning_rate": 0.00012424770586212102,
      "loss": 1.8317,
      "step": 161106
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.1500225067138672,
      "learning_rate": 0.00012424439078040882,
      "loss": 2.1196,
      "step": 161107
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.202886700630188,
      "learning_rate": 0.000124241075731373,
      "loss": 1.9974,
      "step": 161108
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.30105721950531,
      "learning_rate": 0.00012423776071501398,
      "loss": 2.176,
      "step": 161109
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.253944754600525,
      "learning_rate": 0.00012423444573133262,
      "loss": 1.8109,
      "step": 161110
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.1653680801391602,
      "learning_rate": 0.00012423113078032926,
      "loss": 2.0188,
      "step": 161111
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.2076367139816284,
      "learning_rate": 0.00012422781586200486,
      "loss": 2.1106,
      "step": 161112
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.4287663698196411,
      "learning_rate": 0.00012422450097635966,
      "loss": 1.9152,
      "step": 161113
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.3181557655334473,
      "learning_rate": 0.0001242211861233945,
      "loss": 1.9812,
      "step": 161114
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.160586953163147,
      "learning_rate": 0.00012421787130310983,
      "loss": 2.0511,
      "step": 161115
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.0768941640853882,
      "learning_rate": 0.0001242145565155065,
      "loss": 2.2429,
      "step": 161116
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.4901959896087646,
      "learning_rate": 0.00012421124176058488,
      "loss": 1.9503,
      "step": 161117
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.1947247982025146,
      "learning_rate": 0.00012420792703834584,
      "loss": 1.9978,
      "step": 161118
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.1229411363601685,
      "learning_rate": 0.00012420461234878986,
      "loss": 2.0283,
      "step": 161119
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.1786787509918213,
      "learning_rate": 0.00012420129769191756,
      "loss": 2.0889,
      "step": 161120
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.2510738372802734,
      "learning_rate": 0.00012419798306772942,
      "loss": 2.0365,
      "step": 161121
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.3581020832061768,
      "learning_rate": 0.00012419466847622633,
      "loss": 2.1211,
      "step": 161122
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.189955711364746,
      "learning_rate": 0.00012419135391740865,
      "loss": 2.3498,
      "step": 161123
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.2961581945419312,
      "learning_rate": 0.00012418803939127725,
      "loss": 1.9303,
      "step": 161124
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.2406257390975952,
      "learning_rate": 0.0001241847248978326,
      "loss": 2.1727,
      "step": 161125
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.1965285539627075,
      "learning_rate": 0.0001241814104370753,
      "loss": 2.0308,
      "step": 161126
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.272765040397644,
      "learning_rate": 0.0001241780960090059,
      "loss": 1.8815,
      "step": 161127
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.2779459953308105,
      "learning_rate": 0.00012417478161362526,
      "loss": 2.2989,
      "step": 161128
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.1232738494873047,
      "learning_rate": 0.00012417146725093368,
      "loss": 2.0404,
      "step": 161129
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.2069820165634155,
      "learning_rate": 0.00012416815292093207,
      "loss": 2.0943,
      "step": 161130
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.3211193084716797,
      "learning_rate": 0.00012416483862362094,
      "loss": 2.0445,
      "step": 161131
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.1825916767120361,
      "learning_rate": 0.00012416152435900077,
      "loss": 2.1132,
      "step": 161132
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.2537682056427002,
      "learning_rate": 0.0001241582101270724,
      "loss": 1.8557,
      "step": 161133
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.1770349740982056,
      "learning_rate": 0.0001241548959278363,
      "loss": 2.0627,
      "step": 161134
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.1157844066619873,
      "learning_rate": 0.00012415158176129308,
      "loss": 2.1327,
      "step": 161135
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.1778920888900757,
      "learning_rate": 0.00012414826762744346,
      "loss": 2.3491,
      "step": 161136
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.2180240154266357,
      "learning_rate": 0.00012414495352628803,
      "loss": 1.9919,
      "step": 161137
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.2179425954818726,
      "learning_rate": 0.00012414163945782724,
      "loss": 2.0748,
      "step": 161138
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.1138356924057007,
      "learning_rate": 0.00012413832542206195,
      "loss": 1.9366,
      "step": 161139
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.3216997385025024,
      "learning_rate": 0.00012413501141899256,
      "loss": 2.2545,
      "step": 161140
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.2852683067321777,
      "learning_rate": 0.00012413169744861992,
      "loss": 2.0661,
      "step": 161141
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.2544118165969849,
      "learning_rate": 0.00012412838351094445,
      "loss": 2.2058,
      "step": 161142
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.5968303680419922,
      "learning_rate": 0.0001241250696059669,
      "loss": 2.1728,
      "step": 161143
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.2862107753753662,
      "learning_rate": 0.00012412175573368772,
      "loss": 2.1159,
      "step": 161144
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.148768424987793,
      "learning_rate": 0.0001241184418941077,
      "loss": 1.9331,
      "step": 161145
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.510616660118103,
      "learning_rate": 0.00012411512808722727,
      "loss": 2.0296,
      "step": 161146
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.2654533386230469,
      "learning_rate": 0.00012411181431304727,
      "loss": 1.964,
      "step": 161147
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.2815446853637695,
      "learning_rate": 0.00012410850057156822,
      "loss": 2.1709,
      "step": 161148
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.1627823114395142,
      "learning_rate": 0.0001241051868627907,
      "loss": 1.8394,
      "step": 161149
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.3079735040664673,
      "learning_rate": 0.00012410187318671526,
      "loss": 2.1562,
      "step": 161150
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.2259794473648071,
      "learning_rate": 0.00012409855954334268,
      "loss": 2.1345,
      "step": 161151
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.2183507680892944,
      "learning_rate": 0.0001240952459326734,
      "loss": 1.9896,
      "step": 161152
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.0970780849456787,
      "learning_rate": 0.00012409193235470825,
      "loss": 2.2138,
      "step": 161153
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.2584924697875977,
      "learning_rate": 0.00012408861880944765,
      "loss": 1.9676,
      "step": 161154
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.2898597717285156,
      "learning_rate": 0.0001240853052968925,
      "loss": 1.9414,
      "step": 161155
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.203453540802002,
      "learning_rate": 0.00012408199181704294,
      "loss": 2.2646,
      "step": 161156
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.235564947128296,
      "learning_rate": 0.0001240786783699,
      "loss": 2.1858,
      "step": 161157
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.185075283050537,
      "learning_rate": 0.00012407536495546406,
      "loss": 1.8862,
      "step": 161158
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.2592480182647705,
      "learning_rate": 0.00012407205157373595,
      "loss": 2.2921,
      "step": 161159
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.2717323303222656,
      "learning_rate": 0.00012406873822471602,
      "loss": 1.8074,
      "step": 161160
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.1871662139892578,
      "learning_rate": 0.00012406542490840528,
      "loss": 1.9154,
      "step": 161161
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.20516037940979,
      "learning_rate": 0.00012406211162480386,
      "loss": 2.1217,
      "step": 161162
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.3569252490997314,
      "learning_rate": 0.00012405879837391274,
      "loss": 2.2956,
      "step": 161163
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.138345718383789,
      "learning_rate": 0.0001240554851557323,
      "loss": 1.9139,
      "step": 161164
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.2454485893249512,
      "learning_rate": 0.0001240521719702634,
      "loss": 2.1012,
      "step": 161165
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.2128074169158936,
      "learning_rate": 0.00012404885881750634,
      "loss": 2.1204,
      "step": 161166
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.2258754968643188,
      "learning_rate": 0.0001240455456974622,
      "loss": 2.2816,
      "step": 161167
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.2272133827209473,
      "learning_rate": 0.0001240422326101311,
      "loss": 1.7997,
      "step": 161168
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.2202969789505005,
      "learning_rate": 0.00012403891955551394,
      "loss": 2.1409,
      "step": 161169
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.1823269128799438,
      "learning_rate": 0.0001240356065336112,
      "loss": 1.9636,
      "step": 161170
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.3509403467178345,
      "learning_rate": 0.00012403229354442363,
      "loss": 1.9246,
      "step": 161171
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.1191973686218262,
      "learning_rate": 0.00012402898058795167,
      "loss": 1.9504,
      "step": 161172
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.2297303676605225,
      "learning_rate": 0.00012402566766419627,
      "loss": 2.2322,
      "step": 161173
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.2010537385940552,
      "learning_rate": 0.00012402235477315763,
      "loss": 2.1453,
      "step": 161174
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.1309932470321655,
      "learning_rate": 0.00012401904191483665,
      "loss": 2.1247,
      "step": 161175
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.156728744506836,
      "learning_rate": 0.00012401572908923373,
      "loss": 2.0859,
      "step": 161176
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.3133145570755005,
      "learning_rate": 0.00012401241629634972,
      "loss": 2.15,
      "step": 161177
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.1334842443466187,
      "learning_rate": 0.000124009103536185,
      "loss": 2.0001,
      "step": 161178
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.583559274673462,
      "learning_rate": 0.00012400579080874057,
      "loss": 2.0764,
      "step": 161179
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.2262485027313232,
      "learning_rate": 0.00012400247811401654,
      "loss": 2.3358,
      "step": 161180
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.1067109107971191,
      "learning_rate": 0.0001239991654520139,
      "loss": 1.9088,
      "step": 161181
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.1753132343292236,
      "learning_rate": 0.000123995852822733,
      "loss": 1.9473,
      "step": 161182
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.2725204229354858,
      "learning_rate": 0.00012399254022617474,
      "loss": 2.0249,
      "step": 161183
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.2488899230957031,
      "learning_rate": 0.00012398922766233944,
      "loss": 1.9973,
      "step": 161184
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.2179529666900635,
      "learning_rate": 0.00012398591513122808,
      "loss": 2.3185,
      "step": 161185
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.2709202766418457,
      "learning_rate": 0.00012398260263284087,
      "loss": 1.9944,
      "step": 161186
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.2412586212158203,
      "learning_rate": 0.00012397929016717872,
      "loss": 1.9893,
      "step": 161187
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.2569880485534668,
      "learning_rate": 0.00012397597773424203,
      "loss": 2.2103,
      "step": 161188
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.2841026782989502,
      "learning_rate": 0.00012397266533403163,
      "loss": 2.2279,
      "step": 161189
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.6010514497756958,
      "learning_rate": 0.00012396935296654795,
      "loss": 2.0488,
      "step": 161190
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.198398470878601,
      "learning_rate": 0.0001239660406317919,
      "loss": 2.1115,
      "step": 161191
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.0426889657974243,
      "learning_rate": 0.00012396272832976364,
      "loss": 1.9732,
      "step": 161192
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.3289449214935303,
      "learning_rate": 0.00012395941606046417,
      "loss": 1.9703,
      "step": 161193
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.228318452835083,
      "learning_rate": 0.00012395610382389384,
      "loss": 2.3689,
      "step": 161194
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.2242028713226318,
      "learning_rate": 0.0001239527916200535,
      "loss": 1.8156,
      "step": 161195
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.2855302095413208,
      "learning_rate": 0.0001239494794489436,
      "loss": 2.04,
      "step": 161196
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.2423056364059448,
      "learning_rate": 0.00012394616731056497,
      "loss": 1.9821,
      "step": 161197
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.4748520851135254,
      "learning_rate": 0.00012394285520491788,
      "loss": 2.0064,
      "step": 161198
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.1709141731262207,
      "learning_rate": 0.00012393954313200326,
      "loss": 2.0939,
      "step": 161199
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.2430661916732788,
      "learning_rate": 0.0001239362310918215,
      "loss": 2.105,
      "step": 161200
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.1382904052734375,
      "learning_rate": 0.00012393291908437338,
      "loss": 2.0135,
      "step": 161201
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.1645915508270264,
      "learning_rate": 0.00012392960710965936,
      "loss": 2.1815,
      "step": 161202
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.1305428743362427,
      "learning_rate": 0.00012392629516768026,
      "loss": 2.0688,
      "step": 161203
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.4005367755889893,
      "learning_rate": 0.0001239229832584366,
      "loss": 1.9341,
      "step": 161204
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.2168303728103638,
      "learning_rate": 0.00012391967138192898,
      "loss": 2.0031,
      "step": 161205
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.1823928356170654,
      "learning_rate": 0.0001239163595381579,
      "loss": 1.9709,
      "step": 161206
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.2752145528793335,
      "learning_rate": 0.0001239130477271242,
      "loss": 2.0662,
      "step": 161207
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.212563395500183,
      "learning_rate": 0.0001239097359488283,
      "loss": 1.8183,
      "step": 161208
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.3067858219146729,
      "learning_rate": 0.00012390642420327097,
      "loss": 1.9605,
      "step": 161209
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.2349567413330078,
      "learning_rate": 0.0001239031124904528,
      "loss": 2.1567,
      "step": 161210
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.209265947341919,
      "learning_rate": 0.00012389980081037433,
      "loss": 1.7884,
      "step": 161211
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.2572712898254395,
      "learning_rate": 0.00012389648916303616,
      "loss": 2.1957,
      "step": 161212
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.3568358421325684,
      "learning_rate": 0.000123893177548439,
      "loss": 2.0367,
      "step": 161213
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.3135414123535156,
      "learning_rate": 0.00012388986596658337,
      "loss": 2.2057,
      "step": 161214
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.2952715158462524,
      "learning_rate": 0.00012388655441747,
      "loss": 1.9547,
      "step": 161215
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.195703148841858,
      "learning_rate": 0.0001238832429010995,
      "loss": 1.8796,
      "step": 161216
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.3552827835083008,
      "learning_rate": 0.00012387993141747228,
      "loss": 2.2219,
      "step": 161217
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.4824671745300293,
      "learning_rate": 0.00012387661996658922,
      "loss": 2.0853,
      "step": 161218
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.3606696128845215,
      "learning_rate": 0.0001238733085484508,
      "loss": 2.2913,
      "step": 161219
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.1668728590011597,
      "learning_rate": 0.00012386999716305756,
      "loss": 2.0908,
      "step": 161220
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.3457350730895996,
      "learning_rate": 0.00012386668581041034,
      "loss": 1.8086,
      "step": 161221
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.2727901935577393,
      "learning_rate": 0.00012386337449050964,
      "loss": 2.2065,
      "step": 161222
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.284845232963562,
      "learning_rate": 0.00012386006320335593,
      "loss": 2.1548,
      "step": 161223
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.277774691581726,
      "learning_rate": 0.00012385675194895006,
      "loss": 2.0138,
      "step": 161224
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.101370096206665,
      "learning_rate": 0.0001238534407272924,
      "loss": 1.826,
      "step": 161225
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.2830621004104614,
      "learning_rate": 0.0001238501295383839,
      "loss": 2.2632,
      "step": 161226
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.2036107778549194,
      "learning_rate": 0.00012384681838222494,
      "loss": 2.0975,
      "step": 161227
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.176379919052124,
      "learning_rate": 0.00012384350725881622,
      "loss": 2.0144,
      "step": 161228
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.2300398349761963,
      "learning_rate": 0.00012384019616815817,
      "loss": 2.1257,
      "step": 161229
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.332962155342102,
      "learning_rate": 0.00012383688511025166,
      "loss": 1.9164,
      "step": 161230
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.1598427295684814,
      "learning_rate": 0.0001238335740850971,
      "loss": 2.0554,
      "step": 161231
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.192403793334961,
      "learning_rate": 0.0001238302630926953,
      "loss": 1.778,
      "step": 161232
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.1539448499679565,
      "learning_rate": 0.0001238269521330468,
      "loss": 2.2586,
      "step": 161233
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.1802074909210205,
      "learning_rate": 0.0001238236412061522,
      "loss": 1.9475,
      "step": 161234
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.2307342290878296,
      "learning_rate": 0.00012382033031201197,
      "loss": 1.886,
      "step": 161235
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.4540525674819946,
      "learning_rate": 0.000123817019450627,
      "loss": 2.1225,
      "step": 161236
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.1565172672271729,
      "learning_rate": 0.00012381370862199768,
      "loss": 2.073,
      "step": 161237
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.2338265180587769,
      "learning_rate": 0.0001238103978261248,
      "loss": 1.9071,
      "step": 161238
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.2443760633468628,
      "learning_rate": 0.0001238070870630088,
      "loss": 2.0107,
      "step": 161239
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.247105360031128,
      "learning_rate": 0.00012380377633265058,
      "loss": 1.9846,
      "step": 161240
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.2154799699783325,
      "learning_rate": 0.00012380046563505035,
      "loss": 2.0071,
      "step": 161241
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.6928837299346924,
      "learning_rate": 0.0001237971549702091,
      "loss": 2.089,
      "step": 161242
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.1186332702636719,
      "learning_rate": 0.0001237938443381271,
      "loss": 2.1663,
      "step": 161243
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.295355200767517,
      "learning_rate": 0.0001237905337388053,
      "loss": 2.2021,
      "step": 161244
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.13450026512146,
      "learning_rate": 0.00012378722317224404,
      "loss": 2.1198,
      "step": 161245
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.7917560338974,
      "learning_rate": 0.0001237839126384443,
      "loss": 2.0035,
      "step": 161246
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.2985905408859253,
      "learning_rate": 0.00012378060213740624,
      "loss": 2.0441,
      "step": 161247
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.253159999847412,
      "learning_rate": 0.00012377729166913081,
      "loss": 1.8442,
      "step": 161248
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.120760202407837,
      "learning_rate": 0.00012377398123361835,
      "loss": 2.0063,
      "step": 161249
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.1931252479553223,
      "learning_rate": 0.0001237706708308698,
      "loss": 1.9573,
      "step": 161250
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.2849364280700684,
      "learning_rate": 0.00012376736046088547,
      "loss": 2.0515,
      "step": 161251
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.2532545328140259,
      "learning_rate": 0.00012376405012366632,
      "loss": 2.0001,
      "step": 161252
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.140717625617981,
      "learning_rate": 0.00012376073981921259,
      "loss": 2.0532,
      "step": 161253
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.180099606513977,
      "learning_rate": 0.0001237574295475251,
      "loss": 1.916,
      "step": 161254
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.1942644119262695,
      "learning_rate": 0.00012375411930860437,
      "loss": 2.0656,
      "step": 161255
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.5221301317214966,
      "learning_rate": 0.0001237508091024512,
      "loss": 2.0217,
      "step": 161256
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.2115553617477417,
      "learning_rate": 0.00012374749892906593,
      "loss": 2.0754,
      "step": 161257
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.0950406789779663,
      "learning_rate": 0.0001237441887884496,
      "loss": 2.0864,
      "step": 161258
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.5654137134552002,
      "learning_rate": 0.00012374087868060227,
      "loss": 2.1649,
      "step": 161259
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.1692217588424683,
      "learning_rate": 0.000123737568605525,
      "loss": 2.0008,
      "step": 161260
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.2189545631408691,
      "learning_rate": 0.00012373425856321808,
      "loss": 1.922,
      "step": 161261
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.2514363527297974,
      "learning_rate": 0.00012373094855368244,
      "loss": 2.2468,
      "step": 161262
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.2214586734771729,
      "learning_rate": 0.00012372763857691844,
      "loss": 2.1124,
      "step": 161263
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.2918123006820679,
      "learning_rate": 0.000123724328632927,
      "loss": 2.0302,
      "step": 161264
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.3969354629516602,
      "learning_rate": 0.0001237210187217083,
      "loss": 1.7914,
      "step": 161265
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.2180107831954956,
      "learning_rate": 0.00012371770884326332,
      "loss": 1.925,
      "step": 161266
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.1968443393707275,
      "learning_rate": 0.00012371439899759245,
      "loss": 2.0093,
      "step": 161267
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.2807914018630981,
      "learning_rate": 0.00012371108918469647,
      "loss": 2.0899,
      "step": 161268
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.3273823261260986,
      "learning_rate": 0.00012370777940457585,
      "loss": 2.0976,
      "step": 161269
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.2371315956115723,
      "learning_rate": 0.0001237044696572315,
      "loss": 1.9645,
      "step": 161270
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.1327507495880127,
      "learning_rate": 0.00012370115994266357,
      "loss": 2.0247,
      "step": 161271
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.2434645891189575,
      "learning_rate": 0.00012369785026087308,
      "loss": 2.072,
      "step": 161272
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.4083670377731323,
      "learning_rate": 0.00012369454061186033,
      "loss": 1.9633,
      "step": 161273
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.2352203130722046,
      "learning_rate": 0.0001236912309956262,
      "loss": 2.0814,
      "step": 161274
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.236267328262329,
      "learning_rate": 0.00012368792141217113,
      "loss": 2.1696,
      "step": 161275
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.180164098739624,
      "learning_rate": 0.0001236846118614959,
      "loss": 2.2041,
      "step": 161276
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.1943697929382324,
      "learning_rate": 0.00012368130234360103,
      "loss": 1.9955,
      "step": 161277
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.3213998079299927,
      "learning_rate": 0.00012367799285848712,
      "loss": 1.9371,
      "step": 161278
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.312768816947937,
      "learning_rate": 0.0001236746834061547,
      "loss": 2.0006,
      "step": 161279
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.211673617362976,
      "learning_rate": 0.00012367137398660457,
      "loss": 2.0483,
      "step": 161280
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.1918210983276367,
      "learning_rate": 0.00012366806459983718,
      "loss": 1.9659,
      "step": 161281
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.1499353647232056,
      "learning_rate": 0.00012366475524585335,
      "loss": 2.1489,
      "step": 161282
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.1437491178512573,
      "learning_rate": 0.00012366144592465355,
      "loss": 2.0797,
      "step": 161283
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.2782738208770752,
      "learning_rate": 0.0001236581366362384,
      "loss": 2.1121,
      "step": 161284
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.2088408470153809,
      "learning_rate": 0.00012365482738060842,
      "loss": 2.2363,
      "step": 161285
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.16580069065094,
      "learning_rate": 0.00012365151815776446,
      "loss": 2.2783,
      "step": 161286
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.1106382608413696,
      "learning_rate": 0.00012364820896770693,
      "loss": 2.0201,
      "step": 161287
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.2093307971954346,
      "learning_rate": 0.0001236448998104366,
      "loss": 2.2163,
      "step": 161288
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.3612136840820312,
      "learning_rate": 0.00012364159068595402,
      "loss": 1.7951,
      "step": 161289
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.249575138092041,
      "learning_rate": 0.0001236382815942598,
      "loss": 2.2799,
      "step": 161290
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.2874492406845093,
      "learning_rate": 0.00012363497253535443,
      "loss": 2.1555,
      "step": 161291
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.1516649723052979,
      "learning_rate": 0.00012363166350923876,
      "loss": 2.136,
      "step": 161292
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.3692975044250488,
      "learning_rate": 0.0001236283545159132,
      "loss": 2.0568,
      "step": 161293
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.2028390169143677,
      "learning_rate": 0.00012362504555537857,
      "loss": 2.212,
      "step": 161294
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.1837338209152222,
      "learning_rate": 0.00012362173662763536,
      "loss": 2.1785,
      "step": 161295
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.3522469997406006,
      "learning_rate": 0.0001236184277326842,
      "loss": 1.9286,
      "step": 161296
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.1422946453094482,
      "learning_rate": 0.00012361511887052558,
      "loss": 1.9819,
      "step": 161297
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.19627845287323,
      "learning_rate": 0.00012361181004116035,
      "loss": 1.9343,
      "step": 161298
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.1020762920379639,
      "learning_rate": 0.00012360850124458893,
      "loss": 1.8532,
      "step": 161299
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.1513053178787231,
      "learning_rate": 0.00012360519248081213,
      "loss": 1.9546,
      "step": 161300
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.335597038269043,
      "learning_rate": 0.00012360188374983043,
      "loss": 2.2758,
      "step": 161301
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.14317786693573,
      "learning_rate": 0.0001235985750516444,
      "loss": 1.9754,
      "step": 161302
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.2773774862289429,
      "learning_rate": 0.0001235952663862548,
      "loss": 2.156,
      "step": 161303
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.333045244216919,
      "learning_rate": 0.00012359195775366215,
      "loss": 1.8941,
      "step": 161304
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.2080713510513306,
      "learning_rate": 0.00012358864915386702,
      "loss": 2.055,
      "step": 161305
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.170446515083313,
      "learning_rate": 0.00012358534058687018,
      "loss": 2.2836,
      "step": 161306
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.2836538553237915,
      "learning_rate": 0.00012358203205267213,
      "loss": 1.9559,
      "step": 161307
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.3365020751953125,
      "learning_rate": 0.00012357872355127345,
      "loss": 2.0004,
      "step": 161308
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.5310423374176025,
      "learning_rate": 0.0001235754150826749,
      "loss": 2.0176,
      "step": 161309
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.1850203275680542,
      "learning_rate": 0.000123572106646877,
      "loss": 2.001,
      "step": 161310
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.4207453727722168,
      "learning_rate": 0.0001235687982438803,
      "loss": 2.0274,
      "step": 161311
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.2172272205352783,
      "learning_rate": 0.00012356548987368555,
      "loss": 2.1681,
      "step": 161312
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.2629555463790894,
      "learning_rate": 0.00012356218153629334,
      "loss": 2.2218,
      "step": 161313
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.2445313930511475,
      "learning_rate": 0.00012355887323170413,
      "loss": 2.0952,
      "step": 161314
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.69542396068573,
      "learning_rate": 0.0001235555649599188,
      "loss": 1.913,
      "step": 161315
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.3380119800567627,
      "learning_rate": 0.00012355225672093767,
      "loss": 2.1701,
      "step": 161316
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.1092453002929688,
      "learning_rate": 0.00012354894851476162,
      "loss": 2.2022,
      "step": 161317
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.3107826709747314,
      "learning_rate": 0.00012354564034139118,
      "loss": 1.9758,
      "step": 161318
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.3802568912506104,
      "learning_rate": 0.0001235423322008269,
      "loss": 2.2016,
      "step": 161319
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.358923077583313,
      "learning_rate": 0.00012353902409306935,
      "loss": 2.1196,
      "step": 161320
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.2574143409729004,
      "learning_rate": 0.00012353571601811934,
      "loss": 2.0262,
      "step": 161321
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.1722794771194458,
      "learning_rate": 0.00012353240797597723,
      "loss": 1.8189,
      "step": 161322
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.1729366779327393,
      "learning_rate": 0.0001235290999666439,
      "loss": 1.9995,
      "step": 161323
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.1949318647384644,
      "learning_rate": 0.00012352579199011973,
      "loss": 2.0745,
      "step": 161324
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.2242324352264404,
      "learning_rate": 0.00012352248404640567,
      "loss": 2.3067,
      "step": 161325
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.2363042831420898,
      "learning_rate": 0.00012351917613550188,
      "loss": 2.0581,
      "step": 161326
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.4110182523727417,
      "learning_rate": 0.00012351586825740934,
      "loss": 1.942,
      "step": 161327
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.334913730621338,
      "learning_rate": 0.00012351256041212843,
      "loss": 2.1099,
      "step": 161328
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.176788091659546,
      "learning_rate": 0.00012350925259965997,
      "loss": 2.1688,
      "step": 161329
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.3399145603179932,
      "learning_rate": 0.00012350594482000432,
      "loss": 2.077,
      "step": 161330
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.3204569816589355,
      "learning_rate": 0.00012350263707316248,
      "loss": 1.8996,
      "step": 161331
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.3198128938674927,
      "learning_rate": 0.00012349932935913464,
      "loss": 2.1621,
      "step": 161332
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.1563725471496582,
      "learning_rate": 0.0001234960216779217,
      "loss": 1.9764,
      "step": 161333
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.1524943113327026,
      "learning_rate": 0.00012349271402952407,
      "loss": 1.9378,
      "step": 161334
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.2866216897964478,
      "learning_rate": 0.0001234894064139426,
      "loss": 2.0378,
      "step": 161335
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.1530479192733765,
      "learning_rate": 0.00012348609883117766,
      "loss": 2.0299,
      "step": 161336
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.1706651449203491,
      "learning_rate": 0.0001234827912812302,
      "loss": 2.0018,
      "step": 161337
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.1737127304077148,
      "learning_rate": 0.0001234794837641004,
      "loss": 1.8119,
      "step": 161338
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.1691406965255737,
      "learning_rate": 0.0001234761762797892,
      "loss": 1.8502,
      "step": 161339
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.2720218896865845,
      "learning_rate": 0.00012347286882829703,
      "loss": 1.925,
      "step": 161340
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.1627917289733887,
      "learning_rate": 0.0001234695614096247,
      "loss": 2.131,
      "step": 161341
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.2668328285217285,
      "learning_rate": 0.00012346625402377256,
      "loss": 2.1282,
      "step": 161342
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.196055293083191,
      "learning_rate": 0.00012346294667074152,
      "loss": 2.0337,
      "step": 161343
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.176822304725647,
      "learning_rate": 0.00012345963935053207,
      "loss": 1.9526,
      "step": 161344
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.1138007640838623,
      "learning_rate": 0.00012345633206314473,
      "loss": 2.2613,
      "step": 161345
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.4349595308303833,
      "learning_rate": 0.00012345302480858017,
      "loss": 2.1388,
      "step": 161346
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.286394476890564,
      "learning_rate": 0.00012344971758683905,
      "loss": 2.1019,
      "step": 161347
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.295393466949463,
      "learning_rate": 0.0001234464103979219,
      "loss": 2.1868,
      "step": 161348
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.3029013872146606,
      "learning_rate": 0.0001234431032418295,
      "loss": 2.0258,
      "step": 161349
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.158297061920166,
      "learning_rate": 0.00012343979611856236,
      "loss": 1.8552,
      "step": 161350
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.303139090538025,
      "learning_rate": 0.0001234364890281211,
      "loss": 2.2051,
      "step": 161351
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.2643163204193115,
      "learning_rate": 0.0001234331819705062,
      "loss": 2.0138,
      "step": 161352
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.2047553062438965,
      "learning_rate": 0.0001234298749457185,
      "loss": 2.3655,
      "step": 161353
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.3388181924819946,
      "learning_rate": 0.00012342656795375845,
      "loss": 2.1424,
      "step": 161354
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.287652611732483,
      "learning_rate": 0.00012342326099462682,
      "loss": 2.1345,
      "step": 161355
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.224291205406189,
      "learning_rate": 0.00012341995406832415,
      "loss": 2.1503,
      "step": 161356
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.1837992668151855,
      "learning_rate": 0.00012341664717485103,
      "loss": 1.9873,
      "step": 161357
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.3690271377563477,
      "learning_rate": 0.000123413340314208,
      "loss": 2.0614,
      "step": 161358
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.199374794960022,
      "learning_rate": 0.00012341003348639585,
      "loss": 2.1444,
      "step": 161359
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.1991548538208008,
      "learning_rate": 0.000123406726691415,
      "loss": 2.05,
      "step": 161360
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.1899396181106567,
      "learning_rate": 0.0001234034199292663,
      "loss": 2.083,
      "step": 161361
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.3562138080596924,
      "learning_rate": 0.00012340011319995025,
      "loss": 2.1836,
      "step": 161362
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.2617830038070679,
      "learning_rate": 0.0001233968065034674,
      "loss": 2.0878,
      "step": 161363
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.1878786087036133,
      "learning_rate": 0.00012339349983981837,
      "loss": 2.1456,
      "step": 161364
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.1700330972671509,
      "learning_rate": 0.00012339019320900389,
      "loss": 2.0118,
      "step": 161365
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.1444942951202393,
      "learning_rate": 0.00012338688661102444,
      "loss": 1.8307,
      "step": 161366
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.3570016622543335,
      "learning_rate": 0.00012338358004588076,
      "loss": 2.3081,
      "step": 161367
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.2675458192825317,
      "learning_rate": 0.00012338027351357342,
      "loss": 2.0142,
      "step": 161368
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.2735564708709717,
      "learning_rate": 0.00012337696701410304,
      "loss": 1.9143,
      "step": 161369
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.370862603187561,
      "learning_rate": 0.0001233736605474701,
      "loss": 1.9002,
      "step": 161370
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.4025264978408813,
      "learning_rate": 0.0001233703541136754,
      "loss": 2.0749,
      "step": 161371
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.1812481880187988,
      "learning_rate": 0.00012336704771271942,
      "loss": 2.1112,
      "step": 161372
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.2964576482772827,
      "learning_rate": 0.00012336374134460295,
      "loss": 1.9695,
      "step": 161373
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.2412761449813843,
      "learning_rate": 0.00012336043500932648,
      "loss": 2.1169,
      "step": 161374
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.2854955196380615,
      "learning_rate": 0.00012335712870689065,
      "loss": 1.8383,
      "step": 161375
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.1624811887741089,
      "learning_rate": 0.00012335382243729596,
      "loss": 2.1335,
      "step": 161376
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.1925938129425049,
      "learning_rate": 0.0001233505162005432,
      "loss": 2.1111,
      "step": 161377
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.1743649244308472,
      "learning_rate": 0.00012334720999663285,
      "loss": 2.0886,
      "step": 161378
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.2071784734725952,
      "learning_rate": 0.00012334390382556567,
      "loss": 1.9151,
      "step": 161379
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.2149791717529297,
      "learning_rate": 0.0001233405976873422,
      "loss": 2.2617,
      "step": 161380
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.1950862407684326,
      "learning_rate": 0.00012333729158196306,
      "loss": 1.9537,
      "step": 161381
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.2011016607284546,
      "learning_rate": 0.00012333398550942872,
      "loss": 1.9764,
      "step": 161382
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.3088409900665283,
      "learning_rate": 0.00012333067946974008,
      "loss": 2.1163,
      "step": 161383
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.2378079891204834,
      "learning_rate": 0.00012332737346289744,
      "loss": 2.1831,
      "step": 161384
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.2998301982879639,
      "learning_rate": 0.0001233240674889017,
      "loss": 2.0517,
      "step": 161385
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.3673611879348755,
      "learning_rate": 0.00012332076154775336,
      "loss": 1.9819,
      "step": 161386
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.1589853763580322,
      "learning_rate": 0.0001233174556394529,
      "loss": 1.9956,
      "step": 161387
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.2767798900604248,
      "learning_rate": 0.0001233141497640012,
      "loss": 1.9894,
      "step": 161388
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.2067917585372925,
      "learning_rate": 0.0001233108439213987,
      "loss": 2.1063,
      "step": 161389
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.1511062383651733,
      "learning_rate": 0.00012330753811164594,
      "loss": 1.9761,
      "step": 161390
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.4115458726882935,
      "learning_rate": 0.0001233042323347438,
      "loss": 1.8935,
      "step": 161391
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.2113698720932007,
      "learning_rate": 0.0001233009265906927,
      "loss": 2.1203,
      "step": 161392
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.2147692441940308,
      "learning_rate": 0.00012329762087949319,
      "loss": 2.0641,
      "step": 161393
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.3056960105895996,
      "learning_rate": 0.0001232943152011461,
      "loss": 1.93,
      "step": 161394
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.2176376581192017,
      "learning_rate": 0.00012329100955565193,
      "loss": 2.034,
      "step": 161395
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.2855981588363647,
      "learning_rate": 0.0001232877039430112,
      "loss": 2.146,
      "step": 161396
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.2614489793777466,
      "learning_rate": 0.00012328439836322474,
      "loss": 2.0141,
      "step": 161397
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.1644889116287231,
      "learning_rate": 0.00012328109281629302,
      "loss": 2.1027,
      "step": 161398
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.1877696514129639,
      "learning_rate": 0.00012327778730221656,
      "loss": 2.1036,
      "step": 161399
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.5372825860977173,
      "learning_rate": 0.00012327448182099623,
      "loss": 1.8697,
      "step": 161400
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.4631633758544922,
      "learning_rate": 0.00012327117637263242,
      "loss": 2.0518,
      "step": 161401
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.161595344543457,
      "learning_rate": 0.0001232678709571259,
      "loss": 2.1032,
      "step": 161402
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.2313753366470337,
      "learning_rate": 0.00012326456557447724,
      "loss": 2.0662,
      "step": 161403
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.4931998252868652,
      "learning_rate": 0.000123261260224687,
      "loss": 2.0534,
      "step": 161404
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.3870950937271118,
      "learning_rate": 0.00012325795490775578,
      "loss": 1.8038,
      "step": 161405
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.1910487413406372,
      "learning_rate": 0.00012325464962368432,
      "loss": 1.9986,
      "step": 161406
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.318398356437683,
      "learning_rate": 0.00012325134437247306,
      "loss": 1.9649,
      "step": 161407
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.4003815650939941,
      "learning_rate": 0.0001232480391541228,
      "loss": 2.0591,
      "step": 161408
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.3090828657150269,
      "learning_rate": 0.00012324473396863408,
      "loss": 2.1605,
      "step": 161409
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.1995962858200073,
      "learning_rate": 0.00012324142881600737,
      "loss": 2.3833,
      "step": 161410
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.1145730018615723,
      "learning_rate": 0.00012323812369624353,
      "loss": 1.9529,
      "step": 161411
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.139673113822937,
      "learning_rate": 0.00012323481860934308,
      "loss": 2.153,
      "step": 161412
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.2767024040222168,
      "learning_rate": 0.0001232315135553065,
      "loss": 2.1315,
      "step": 161413
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.3045217990875244,
      "learning_rate": 0.00012322820853413461,
      "loss": 2.2009,
      "step": 161414
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.2369933128356934,
      "learning_rate": 0.00012322490354582782,
      "loss": 1.9288,
      "step": 161415
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.2979806661605835,
      "learning_rate": 0.000123221598590387,
      "loss": 2.295,
      "step": 161416
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.1825631856918335,
      "learning_rate": 0.0001232182936678126,
      "loss": 2.0213,
      "step": 161417
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.19191575050354,
      "learning_rate": 0.00012321498877810525,
      "loss": 1.7911,
      "step": 161418
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.5445613861083984,
      "learning_rate": 0.00012321168392126546,
      "loss": 2.1889,
      "step": 161419
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.287155032157898,
      "learning_rate": 0.0001232083790972941,
      "loss": 1.7643,
      "step": 161420
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.4339519739151,
      "learning_rate": 0.0001232050743061915,
      "loss": 2.1259,
      "step": 161421
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.1306971311569214,
      "learning_rate": 0.00012320176954795853,
      "loss": 1.8388,
      "step": 161422
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.2087246179580688,
      "learning_rate": 0.00012319846482259568,
      "loss": 1.9023,
      "step": 161423
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.233167052268982,
      "learning_rate": 0.00012319516013010358,
      "loss": 2.1618,
      "step": 161424
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.5428533554077148,
      "learning_rate": 0.00012319185547048274,
      "loss": 2.1085,
      "step": 161425
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.1384135484695435,
      "learning_rate": 0.00012318855084373395,
      "loss": 2.0671,
      "step": 161426
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.2557377815246582,
      "learning_rate": 0.00012318524624985768,
      "loss": 2.042,
      "step": 161427
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.1538463830947876,
      "learning_rate": 0.00012318194168885467,
      "loss": 1.9968,
      "step": 161428
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.3144922256469727,
      "learning_rate": 0.0001231786371607255,
      "loss": 2.169,
      "step": 161429
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.1794219017028809,
      "learning_rate": 0.00012317533266547077,
      "loss": 2.1412,
      "step": 161430
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.181914210319519,
      "learning_rate": 0.000123172028203091,
      "loss": 1.9145,
      "step": 161431
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.1957882642745972,
      "learning_rate": 0.00012316872377358693,
      "loss": 1.9826,
      "step": 161432
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.2805871963500977,
      "learning_rate": 0.00012316541937695908,
      "loss": 1.9141,
      "step": 161433
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.258868932723999,
      "learning_rate": 0.0001231621150132082,
      "loss": 2.0646,
      "step": 161434
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.1810730695724487,
      "learning_rate": 0.0001231588106823348,
      "loss": 1.8019,
      "step": 161435
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.1882153749465942,
      "learning_rate": 0.00012315550638433955,
      "loss": 2.1867,
      "step": 161436
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.2697325944900513,
      "learning_rate": 0.00012315220211922288,
      "loss": 2.1397,
      "step": 161437
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.144612193107605,
      "learning_rate": 0.0001231488978869857,
      "loss": 1.9181,
      "step": 161438
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.1038954257965088,
      "learning_rate": 0.00012314559368762834,
      "loss": 2.0169,
      "step": 161439
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.1836044788360596,
      "learning_rate": 0.00012314228952115168,
      "loss": 2.1789,
      "step": 161440
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.3254038095474243,
      "learning_rate": 0.00012313898538755623,
      "loss": 2.0458,
      "step": 161441
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.2117527723312378,
      "learning_rate": 0.00012313568128684255,
      "loss": 1.8541,
      "step": 161442
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.2264549732208252,
      "learning_rate": 0.00012313237721901118,
      "loss": 2.0496,
      "step": 161443
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.15958571434021,
      "learning_rate": 0.00012312907318406296,
      "loss": 2.0543,
      "step": 161444
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.298825979232788,
      "learning_rate": 0.00012312576918199826,
      "loss": 2.2537,
      "step": 161445
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.2963526248931885,
      "learning_rate": 0.00012312246521281793,
      "loss": 1.9826,
      "step": 161446
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.2541570663452148,
      "learning_rate": 0.00012311916127652248,
      "loss": 1.8423,
      "step": 161447
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.2296133041381836,
      "learning_rate": 0.00012311585737311253,
      "loss": 2.2514,
      "step": 161448
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.1389931440353394,
      "learning_rate": 0.00012311255350258853,
      "loss": 2.1001,
      "step": 161449
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.1740080118179321,
      "learning_rate": 0.00012310924966495138,
      "loss": 1.9219,
      "step": 161450
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.1743398904800415,
      "learning_rate": 0.00012310594586020145,
      "loss": 2.146,
      "step": 161451
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.3040069341659546,
      "learning_rate": 0.0001231026420883396,
      "loss": 2.1058,
      "step": 161452
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.240913987159729,
      "learning_rate": 0.00012309933834936626,
      "loss": 2.0041,
      "step": 161453
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.1480473279953003,
      "learning_rate": 0.00012309603464328214,
      "loss": 1.9826,
      "step": 161454
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.206555724143982,
      "learning_rate": 0.00012309273097008765,
      "loss": 2.1283,
      "step": 161455
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.2339690923690796,
      "learning_rate": 0.0001230894273297837,
      "loss": 2.0323,
      "step": 161456
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.122782826423645,
      "learning_rate": 0.00012308612372237065,
      "loss": 2.0891,
      "step": 161457
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.203747272491455,
      "learning_rate": 0.00012308282014784934,
      "loss": 2.0929,
      "step": 161458
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.2607396841049194,
      "learning_rate": 0.00012307951660622026,
      "loss": 2.0129,
      "step": 161459
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.1856281757354736,
      "learning_rate": 0.00012307621309748406,
      "loss": 1.7869,
      "step": 161460
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.1385884284973145,
      "learning_rate": 0.00012307290962164117,
      "loss": 2.1553,
      "step": 161461
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.2326655387878418,
      "learning_rate": 0.00012306960617869254,
      "loss": 2.017,
      "step": 161462
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.2487692832946777,
      "learning_rate": 0.00012306630276863847,
      "loss": 2.0702,
      "step": 161463
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.3432986736297607,
      "learning_rate": 0.00012306299939147985,
      "loss": 2.2977,
      "step": 161464
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.1905711889266968,
      "learning_rate": 0.0001230596960472171,
      "loss": 2.0943,
      "step": 161465
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.169653296470642,
      "learning_rate": 0.00012305639273585094,
      "loss": 2.0865,
      "step": 161466
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.2444719076156616,
      "learning_rate": 0.00012305308945738183,
      "loss": 1.9356,
      "step": 161467
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.2031688690185547,
      "learning_rate": 0.00012304978621181057,
      "loss": 1.8706,
      "step": 161468
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.1983377933502197,
      "learning_rate": 0.00012304648299913762,
      "loss": 1.8227,
      "step": 161469
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.3027950525283813,
      "learning_rate": 0.00012304317981936375,
      "loss": 2.0723,
      "step": 161470
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.2137784957885742,
      "learning_rate": 0.00012303987667248948,
      "loss": 1.9899,
      "step": 161471
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.1966603994369507,
      "learning_rate": 0.0001230365735585155,
      "loss": 2.0745,
      "step": 161472
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.1737111806869507,
      "learning_rate": 0.0001230332704774422,
      "loss": 1.9239,
      "step": 161473
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.195645809173584,
      "learning_rate": 0.00012302996742927049,
      "loss": 2.1364,
      "step": 161474
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.245741367340088,
      "learning_rate": 0.00012302666441400072,
      "loss": 2.2518,
      "step": 161475
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.2381683588027954,
      "learning_rate": 0.00012302336143163374,
      "loss": 2.3014,
      "step": 161476
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.2161798477172852,
      "learning_rate": 0.00012302005848216998,
      "loss": 1.8472,
      "step": 161477
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.5731332302093506,
      "learning_rate": 0.0001230167555656102,
      "loss": 2.2279,
      "step": 161478
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.3362022638320923,
      "learning_rate": 0.00012301345268195496,
      "loss": 1.9907,
      "step": 161479
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.2394444942474365,
      "learning_rate": 0.00012301014983120487,
      "loss": 2.034,
      "step": 161480
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.1693323850631714,
      "learning_rate": 0.00012300684701336043,
      "loss": 1.9111,
      "step": 161481
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.1296049356460571,
      "learning_rate": 0.00012300354422842245,
      "loss": 1.8169,
      "step": 161482
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.1987049579620361,
      "learning_rate": 0.00012300024147639135,
      "loss": 2.1054,
      "step": 161483
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.2566372156143188,
      "learning_rate": 0.00012299693875726797,
      "loss": 2.291,
      "step": 161484
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.2795953750610352,
      "learning_rate": 0.00012299363607105277,
      "loss": 2.2102,
      "step": 161485
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.3088221549987793,
      "learning_rate": 0.0001229903334177463,
      "loss": 2.1035,
      "step": 161486
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.175507664680481,
      "learning_rate": 0.0001229870307973494,
      "loss": 2.0553,
      "step": 161487
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.1314330101013184,
      "learning_rate": 0.00012298372820986254,
      "loss": 1.961,
      "step": 161488
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.1518588066101074,
      "learning_rate": 0.0001229804256552862,
      "loss": 2.015,
      "step": 161489
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.1841633319854736,
      "learning_rate": 0.00012297712313362133,
      "loss": 1.8911,
      "step": 161490
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.3294912576675415,
      "learning_rate": 0.0001229738206448683,
      "loss": 2.1715,
      "step": 161491
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.1870976686477661,
      "learning_rate": 0.00012297051818902765,
      "loss": 1.9543,
      "step": 161492
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.2483022212982178,
      "learning_rate": 0.00012296721576610028,
      "loss": 2.1882,
      "step": 161493
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.247006893157959,
      "learning_rate": 0.0001229639133760866,
      "loss": 2.0438,
      "step": 161494
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.4685899019241333,
      "learning_rate": 0.00012296061101898717,
      "loss": 2.0544,
      "step": 161495
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.2068207263946533,
      "learning_rate": 0.00012295730869480283,
      "loss": 1.7407,
      "step": 161496
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.7435786724090576,
      "learning_rate": 0.00012295400640353408,
      "loss": 2.1585,
      "step": 161497
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.1999140977859497,
      "learning_rate": 0.00012295070414518138,
      "loss": 2.0891,
      "step": 161498
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.390730857849121,
      "learning_rate": 0.0001229474019197456,
      "loss": 1.9676,
      "step": 161499
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.3722220659255981,
      "learning_rate": 0.00012294409972722715,
      "loss": 1.9736,
      "step": 161500
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.192014217376709,
      "learning_rate": 0.00012294079756762684,
      "loss": 1.9435,
      "step": 161501
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.1363447904586792,
      "learning_rate": 0.00012293749544094518,
      "loss": 2.0193,
      "step": 161502
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.417621374130249,
      "learning_rate": 0.00012293419334718277,
      "loss": 1.9953,
      "step": 161503
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.2493743896484375,
      "learning_rate": 0.00012293089128634013,
      "loss": 2.0186,
      "step": 161504
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.230392336845398,
      "learning_rate": 0.00012292758925841813,
      "loss": 1.869,
      "step": 161505
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.2253224849700928,
      "learning_rate": 0.00012292428726341705,
      "loss": 2.0829,
      "step": 161506
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.39948570728302,
      "learning_rate": 0.00012292098530133783,
      "loss": 2.1686,
      "step": 161507
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.282211184501648,
      "learning_rate": 0.00012291768337218098,
      "loss": 2.2876,
      "step": 161508
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.172228455543518,
      "learning_rate": 0.000122914381475947,
      "loss": 2.0355,
      "step": 161509
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.1831227540969849,
      "learning_rate": 0.00012291107961263654,
      "loss": 1.9132,
      "step": 161510
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.36112642288208,
      "learning_rate": 0.00012290777778225032,
      "loss": 1.9424,
      "step": 161511
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.1301392316818237,
      "learning_rate": 0.0001229044759847888,
      "loss": 1.9969,
      "step": 161512
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.2463867664337158,
      "learning_rate": 0.00012290117422025277,
      "loss": 1.9725,
      "step": 161513
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.451485514640808,
      "learning_rate": 0.00012289787248864276,
      "loss": 2.0553,
      "step": 161514
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.2421088218688965,
      "learning_rate": 0.00012289457078995937,
      "loss": 2.1788,
      "step": 161515
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.2242631912231445,
      "learning_rate": 0.00012289126912420315,
      "loss": 2.0235,
      "step": 161516
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.189211130142212,
      "learning_rate": 0.00012288796749137487,
      "loss": 1.9617,
      "step": 161517
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.2236828804016113,
      "learning_rate": 0.00012288466589147493,
      "loss": 1.9323,
      "step": 161518
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.2067121267318726,
      "learning_rate": 0.00012288136432450422,
      "loss": 2.091,
      "step": 161519
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.275443434715271,
      "learning_rate": 0.0001228780627904632,
      "loss": 2.0902,
      "step": 161520
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.2941843271255493,
      "learning_rate": 0.00012287476128935247,
      "loss": 2.1191,
      "step": 161521
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.1626836061477661,
      "learning_rate": 0.0001228714598211726,
      "loss": 2.0189,
      "step": 161522
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.1091240644454956,
      "learning_rate": 0.00012286815838592435,
      "loss": 1.9471,
      "step": 161523
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.1747556924819946,
      "learning_rate": 0.0001228648569836081,
      "loss": 1.9946,
      "step": 161524
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.0671813488006592,
      "learning_rate": 0.00012286155561422478,
      "loss": 1.9311,
      "step": 161525
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.1460864543914795,
      "learning_rate": 0.00012285825427777483,
      "loss": 2.0212,
      "step": 161526
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.3921316862106323,
      "learning_rate": 0.00012285495297425887,
      "loss": 2.0483,
      "step": 161527
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.256531834602356,
      "learning_rate": 0.0001228516517036774,
      "loss": 2.0609,
      "step": 161528
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.3005356788635254,
      "learning_rate": 0.0001228483504660313,
      "loss": 1.8758,
      "step": 161529
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.2986366748809814,
      "learning_rate": 0.00012284504926132091,
      "loss": 2.1679,
      "step": 161530
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.152044653892517,
      "learning_rate": 0.00012284174808954707,
      "loss": 2.024,
      "step": 161531
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.1959538459777832,
      "learning_rate": 0.00012283844695071028,
      "loss": 1.8922,
      "step": 161532
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.1430779695510864,
      "learning_rate": 0.00012283514584481118,
      "loss": 2.1849,
      "step": 161533
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.2487072944641113,
      "learning_rate": 0.00012283184477185027,
      "loss": 2.1231,
      "step": 161534
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.1813157796859741,
      "learning_rate": 0.00012282854373182837,
      "loss": 2.0382,
      "step": 161535
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.2908003330230713,
      "learning_rate": 0.00012282524272474586,
      "loss": 2.221,
      "step": 161536
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.218320369720459,
      "learning_rate": 0.0001228219417506036,
      "loss": 2.1679,
      "step": 161537
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.2120200395584106,
      "learning_rate": 0.00012281864080940209,
      "loss": 2.1286,
      "step": 161538
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.2139015197753906,
      "learning_rate": 0.00012281533990114195,
      "loss": 2.0965,
      "step": 161539
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.1698521375656128,
      "learning_rate": 0.00012281203902582367,
      "loss": 1.9283,
      "step": 161540
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.308406114578247,
      "learning_rate": 0.00012280873818344806,
      "loss": 2.1329,
      "step": 161541
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.4263759851455688,
      "learning_rate": 0.00012280543737401556,
      "loss": 2.0664,
      "step": 161542
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.3124459981918335,
      "learning_rate": 0.000122802136597527,
      "loss": 2.1411,
      "step": 161543
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.1871143579483032,
      "learning_rate": 0.00012279883585398273,
      "loss": 1.9705,
      "step": 161544
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.3339229822158813,
      "learning_rate": 0.00012279553514338374,
      "loss": 1.9456,
      "step": 161545
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.2771425247192383,
      "learning_rate": 0.00012279223446573016,
      "loss": 1.9189,
      "step": 161546
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.1782584190368652,
      "learning_rate": 0.00012278893382102298,
      "loss": 2.3932,
      "step": 161547
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.2479008436203003,
      "learning_rate": 0.00012278563320926257,
      "loss": 2.0681,
      "step": 161548
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.299769401550293,
      "learning_rate": 0.00012278233263044976,
      "loss": 2.1838,
      "step": 161549
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.227781891822815,
      "learning_rate": 0.00012277903208458496,
      "loss": 1.981,
      "step": 161550
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.2021219730377197,
      "learning_rate": 0.0001227757315716691,
      "loss": 2.1508,
      "step": 161551
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.4984803199768066,
      "learning_rate": 0.00012277243109170234,
      "loss": 2.2026,
      "step": 161552
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.2309612035751343,
      "learning_rate": 0.00012276913064468567,
      "loss": 2.0916,
      "step": 161553
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.171849250793457,
      "learning_rate": 0.00012276583023061944,
      "loss": 2.0213,
      "step": 161554
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.3013187646865845,
      "learning_rate": 0.0001227625298495045,
      "loss": 2.118,
      "step": 161555
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.1709797382354736,
      "learning_rate": 0.00012275922950134123,
      "loss": 1.9635,
      "step": 161556
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.1738134622573853,
      "learning_rate": 0.0001227559291861306,
      "loss": 1.7938,
      "step": 161557
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.1779370307922363,
      "learning_rate": 0.00012275262890387275,
      "loss": 2.2016,
      "step": 161558
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.2108819484710693,
      "learning_rate": 0.00012274932865456863,
      "loss": 1.8639,
      "step": 161559
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.0860211849212646,
      "learning_rate": 0.0001227460284382187,
      "loss": 2.109,
      "step": 161560
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.2552542686462402,
      "learning_rate": 0.00012274272825482373,
      "loss": 1.9931,
      "step": 161561
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.3224178552627563,
      "learning_rate": 0.00012273942810438407,
      "loss": 2.1643,
      "step": 161562
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.520612120628357,
      "learning_rate": 0.00012273612798690064,
      "loss": 1.985,
      "step": 161563
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.2476447820663452,
      "learning_rate": 0.0001227328279023739,
      "loss": 2.1478,
      "step": 161564
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.2389557361602783,
      "learning_rate": 0.00012272952785080449,
      "loss": 2.1664,
      "step": 161565
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.2947285175323486,
      "learning_rate": 0.00012272622783219287,
      "loss": 2.1989,
      "step": 161566
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.302821397781372,
      "learning_rate": 0.00012272292784653994,
      "loss": 1.8775,
      "step": 161567
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.2167402505874634,
      "learning_rate": 0.00012271962789384605,
      "loss": 2.2807,
      "step": 161568
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.15566885471344,
      "learning_rate": 0.00012271632797411202,
      "loss": 1.9394,
      "step": 161569
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.3777509927749634,
      "learning_rate": 0.00012271302808733835,
      "loss": 1.9662,
      "step": 161570
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.1383137702941895,
      "learning_rate": 0.0001227097282335257,
      "loss": 2.1088,
      "step": 161571
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.2854838371276855,
      "learning_rate": 0.00012270642841267454,
      "loss": 2.2423,
      "step": 161572
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.1572656631469727,
      "learning_rate": 0.0001227031286247857,
      "loss": 2.1431,
      "step": 161573
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.187288522720337,
      "learning_rate": 0.00012269982886985961,
      "loss": 1.9901,
      "step": 161574
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.2530277967453003,
      "learning_rate": 0.00012269652914789707,
      "loss": 1.879,
      "step": 161575
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.2424347400665283,
      "learning_rate": 0.0001226932294588986,
      "loss": 2.0522,
      "step": 161576
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.3923426866531372,
      "learning_rate": 0.00012268992980286468,
      "loss": 2.0825,
      "step": 161577
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.1049150228500366,
      "learning_rate": 0.00012268663017979615,
      "loss": 2.075,
      "step": 161578
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.2402204275131226,
      "learning_rate": 0.00012268333058969354,
      "loss": 1.8574,
      "step": 161579
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.1843509674072266,
      "learning_rate": 0.00012268003103255737,
      "loss": 2.1277,
      "step": 161580
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.2016043663024902,
      "learning_rate": 0.00012267673150838837,
      "loss": 1.8247,
      "step": 161581
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.1620464324951172,
      "learning_rate": 0.00012267343201718714,
      "loss": 2.0544,
      "step": 161582
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.2481580972671509,
      "learning_rate": 0.00012267013255895417,
      "loss": 2.0283,
      "step": 161583
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.1601736545562744,
      "learning_rate": 0.00012266683313369027,
      "loss": 2.134,
      "step": 161584
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.2297481298446655,
      "learning_rate": 0.00012266353374139588,
      "loss": 1.9907,
      "step": 161585
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.4269332885742188,
      "learning_rate": 0.00012266023438207174,
      "loss": 2.1828,
      "step": 161586
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.2994649410247803,
      "learning_rate": 0.00012265693505571844,
      "loss": 1.9607,
      "step": 161587
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.8519269227981567,
      "learning_rate": 0.00012265363576233654,
      "loss": 1.9855,
      "step": 161588
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.2853947877883911,
      "learning_rate": 0.00012265033650192658,
      "loss": 2.0647,
      "step": 161589
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.209420919418335,
      "learning_rate": 0.00012264703727448938,
      "loss": 2.1338,
      "step": 161590
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.116689682006836,
      "learning_rate": 0.00012264373808002535,
      "loss": 1.8793,
      "step": 161591
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.211927056312561,
      "learning_rate": 0.00012264043891853528,
      "loss": 2.0026,
      "step": 161592
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.0982544422149658,
      "learning_rate": 0.00012263713979001971,
      "loss": 2.0772,
      "step": 161593
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.3069876432418823,
      "learning_rate": 0.00012263384069447926,
      "loss": 2.1074,
      "step": 161594
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.1606242656707764,
      "learning_rate": 0.00012263054163191443,
      "loss": 2.065,
      "step": 161595
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.2021979093551636,
      "learning_rate": 0.00012262724260232597,
      "loss": 2.2485,
      "step": 161596
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.2810431718826294,
      "learning_rate": 0.00012262394360571443,
      "loss": 2.0765,
      "step": 161597
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.413687825202942,
      "learning_rate": 0.00012262064464208048,
      "loss": 1.9399,
      "step": 161598
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.171779751777649,
      "learning_rate": 0.00012261734571142473,
      "loss": 1.9252,
      "step": 161599
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.303662657737732,
      "learning_rate": 0.00012261404681374779,
      "loss": 2.1053,
      "step": 161600
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.1578484773635864,
      "learning_rate": 0.0001226107479490501,
      "loss": 1.9616,
      "step": 161601
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.9664994478225708,
      "learning_rate": 0.00012260744911733257,
      "loss": 2.033,
      "step": 161602
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.108152151107788,
      "learning_rate": 0.00012260415031859555,
      "loss": 1.9304,
      "step": 161603
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.1530210971832275,
      "learning_rate": 0.00012260085155283985,
      "loss": 2.0208,
      "step": 161604
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.2516711950302124,
      "learning_rate": 0.000122597552820066,
      "loss": 2.0463,
      "step": 161605
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.4405492544174194,
      "learning_rate": 0.0001225942541202746,
      "loss": 2.1945,
      "step": 161606
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.1138890981674194,
      "learning_rate": 0.00012259095545346619,
      "loss": 1.8974,
      "step": 161607
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.2000353336334229,
      "learning_rate": 0.0001225876568196416,
      "loss": 1.9712,
      "step": 161608
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.1527971029281616,
      "learning_rate": 0.00012258435821880118,
      "loss": 1.8858,
      "step": 161609
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.3009326457977295,
      "learning_rate": 0.0001225810596509458,
      "loss": 2.1545,
      "step": 161610
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.0811889171600342,
      "learning_rate": 0.00012257776111607583,
      "loss": 1.7314,
      "step": 161611
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.0956685543060303,
      "learning_rate": 0.0001225744626141922,
      "loss": 2.0942,
      "step": 161612
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.2313557863235474,
      "learning_rate": 0.00012257116414529515,
      "loss": 1.8632,
      "step": 161613
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.2277098894119263,
      "learning_rate": 0.00012256786570938555,
      "loss": 2.3417,
      "step": 161614
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.153676152229309,
      "learning_rate": 0.00012256456730646385,
      "loss": 1.9783,
      "step": 161615
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.1567827463150024,
      "learning_rate": 0.00012256126893653082,
      "loss": 2.0203,
      "step": 161616
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.1103183031082153,
      "learning_rate": 0.0001225579705995869,
      "loss": 2.166,
      "step": 161617
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.1115936040878296,
      "learning_rate": 0.00012255467229563304,
      "loss": 1.9335,
      "step": 161618
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.1936143636703491,
      "learning_rate": 0.0001225513740246694,
      "loss": 1.8776,
      "step": 161619
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.276158094406128,
      "learning_rate": 0.0001225480757866969,
      "loss": 1.9503,
      "step": 161620
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.1666492223739624,
      "learning_rate": 0.00012254477758171596,
      "loss": 2.026,
      "step": 161621
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.153610110282898,
      "learning_rate": 0.0001225414794097274,
      "loss": 2.3424,
      "step": 161622
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.231710433959961,
      "learning_rate": 0.00012253818127073163,
      "loss": 2.3842,
      "step": 161623
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.1463537216186523,
      "learning_rate": 0.00012253488316472957,
      "loss": 2.1645,
      "step": 161624
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.1140596866607666,
      "learning_rate": 0.0001225315850917214,
      "loss": 1.8933,
      "step": 161625
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.2506358623504639,
      "learning_rate": 0.00012252828705170806,
      "loss": 2.1786,
      "step": 161626
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.2892056703567505,
      "learning_rate": 0.00012252498904469,
      "loss": 2.0745,
      "step": 161627
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.2175768613815308,
      "learning_rate": 0.00012252169107066795,
      "loss": 2.0524,
      "step": 161628
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.2566877603530884,
      "learning_rate": 0.0001225183931296424,
      "loss": 1.8362,
      "step": 161629
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.141677737236023,
      "learning_rate": 0.00012251509522161422,
      "loss": 2.1281,
      "step": 161630
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.299226999282837,
      "learning_rate": 0.00012251179734658362,
      "loss": 2.3925,
      "step": 161631
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.2800533771514893,
      "learning_rate": 0.00012250849950455156,
      "loss": 2.0637,
      "step": 161632
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.2957338094711304,
      "learning_rate": 0.0001225052016955184,
      "loss": 2.0732,
      "step": 161633
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.2046570777893066,
      "learning_rate": 0.00012250190391948496,
      "loss": 2.0574,
      "step": 161634
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.2771707773208618,
      "learning_rate": 0.00012249860617645166,
      "loss": 1.8526,
      "step": 161635
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.3033185005187988,
      "learning_rate": 0.00012249530846641945,
      "loss": 2.0938,
      "step": 161636
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.1848794221878052,
      "learning_rate": 0.0001224920107893885,
      "loss": 2.1576,
      "step": 161637
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.146981954574585,
      "learning_rate": 0.00012248871314535977,
      "loss": 2.1217,
      "step": 161638
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.3541725873947144,
      "learning_rate": 0.0001224854155343336,
      "loss": 2.0935,
      "step": 161639
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.2692118883132935,
      "learning_rate": 0.00012248211795631085,
      "loss": 1.8471,
      "step": 161640
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.4123977422714233,
      "learning_rate": 0.0001224788204112919,
      "loss": 2.0746,
      "step": 161641
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.1832607984542847,
      "learning_rate": 0.00012247552289927776,
      "loss": 1.9938,
      "step": 161642
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.2740752696990967,
      "learning_rate": 0.0001224722254202685,
      "loss": 2.1482,
      "step": 161643
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.222819209098816,
      "learning_rate": 0.00012246892797426515,
      "loss": 1.8851,
      "step": 161644
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.2689276933670044,
      "learning_rate": 0.00012246563056126806,
      "loss": 2.0784,
      "step": 161645
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.32014799118042,
      "learning_rate": 0.00012246233318127803,
      "loss": 2.0271,
      "step": 161646
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.179270625114441,
      "learning_rate": 0.00012245903583429555,
      "loss": 1.9725,
      "step": 161647
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.2619694471359253,
      "learning_rate": 0.00012245573852032136,
      "loss": 2.1095,
      "step": 161648
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.5740430355072021,
      "learning_rate": 0.00012245244123935603,
      "loss": 1.9688,
      "step": 161649
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.2655190229415894,
      "learning_rate": 0.00012244914399140014,
      "loss": 2.0688,
      "step": 161650
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.3231985569000244,
      "learning_rate": 0.00012244584677645416,
      "loss": 2.1015,
      "step": 161651
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.1536061763763428,
      "learning_rate": 0.00012244254959451896,
      "loss": 1.9226,
      "step": 161652
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.4474762678146362,
      "learning_rate": 0.00012243925244559496,
      "loss": 2.1417,
      "step": 161653
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.496713638305664,
      "learning_rate": 0.00012243595532968296,
      "loss": 2.0272,
      "step": 161654
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.3194993734359741,
      "learning_rate": 0.00012243265824678345,
      "loss": 1.9575,
      "step": 161655
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.1604583263397217,
      "learning_rate": 0.0001224293611968971,
      "loss": 2.0257,
      "step": 161656
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.4007817506790161,
      "learning_rate": 0.00012242606418002433,
      "loss": 2.0029,
      "step": 161657
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.2636393308639526,
      "learning_rate": 0.00012242276719616604,
      "loss": 2.1666,
      "step": 161658
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.3229875564575195,
      "learning_rate": 0.00012241947024532258,
      "loss": 2.1985,
      "step": 161659
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.3735215663909912,
      "learning_rate": 0.00012241617332749483,
      "loss": 2.0586,
      "step": 161660
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.1626465320587158,
      "learning_rate": 0.00012241287644268323,
      "loss": 2.0789,
      "step": 161661
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.2297701835632324,
      "learning_rate": 0.00012240957959088833,
      "loss": 1.8934,
      "step": 161662
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.308945655822754,
      "learning_rate": 0.00012240628277211096,
      "loss": 2.1488,
      "step": 161663
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.31512451171875,
      "learning_rate": 0.00012240298598635164,
      "loss": 2.0066,
      "step": 161664
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.1889764070510864,
      "learning_rate": 0.0001223996892336108,
      "loss": 2.1983,
      "step": 161665
    },
    {
      "epoch": 2.1,
      "grad_norm": 1.313606858253479,
      "learning_rate": 0.00012239639251388935,
      "loss": 2.0188,
      "step": 161666
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.1920464038848877,
      "learning_rate": 0.00012239309582718778,
      "loss": 1.7604,
      "step": 161667
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.262913703918457,
      "learning_rate": 0.00012238979917350654,
      "loss": 1.9222,
      "step": 161668
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.4505752325057983,
      "learning_rate": 0.00012238650255284652,
      "loss": 2.072,
      "step": 161669
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.2642699480056763,
      "learning_rate": 0.00012238320596520807,
      "loss": 1.9815,
      "step": 161670
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.1921395063400269,
      "learning_rate": 0.00012237990941059207,
      "loss": 1.8945,
      "step": 161671
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.0952223539352417,
      "learning_rate": 0.00012237661288899896,
      "loss": 2.3081,
      "step": 161672
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.2003759145736694,
      "learning_rate": 0.0001223733164004294,
      "loss": 2.1061,
      "step": 161673
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.5133270025253296,
      "learning_rate": 0.00012237001994488392,
      "loss": 1.9817,
      "step": 161674
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.308260440826416,
      "learning_rate": 0.0001223667235223633,
      "loss": 2.2366,
      "step": 161675
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.2530150413513184,
      "learning_rate": 0.00012236342713286793,
      "loss": 2.0402,
      "step": 161676
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.1622424125671387,
      "learning_rate": 0.00012236013077639868,
      "loss": 1.9241,
      "step": 161677
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.153471827507019,
      "learning_rate": 0.00012235683445295604,
      "loss": 2.025,
      "step": 161678
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.2161197662353516,
      "learning_rate": 0.0001223535381625406,
      "loss": 2.1464,
      "step": 161679
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.2134344577789307,
      "learning_rate": 0.0001223502419051529,
      "loss": 2.1648,
      "step": 161680
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.1566951274871826,
      "learning_rate": 0.00012234694568079375,
      "loss": 2.0205,
      "step": 161681
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.2028480768203735,
      "learning_rate": 0.00012234364948946353,
      "loss": 1.7827,
      "step": 161682
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.2298845052719116,
      "learning_rate": 0.0001223403533311631,
      "loss": 2.1485,
      "step": 161683
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.141709566116333,
      "learning_rate": 0.00012233705720589283,
      "loss": 1.8406,
      "step": 161684
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.1773720979690552,
      "learning_rate": 0.0001223337611136537,
      "loss": 1.9205,
      "step": 161685
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.1425350904464722,
      "learning_rate": 0.00012233046505444586,
      "loss": 1.8476,
      "step": 161686
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.1438204050064087,
      "learning_rate": 0.0001223271690282702,
      "loss": 1.9038,
      "step": 161687
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.1869553327560425,
      "learning_rate": 0.0001223238730351272,
      "loss": 2.0617,
      "step": 161688
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.151194453239441,
      "learning_rate": 0.0001223205770750177,
      "loss": 2.0179,
      "step": 161689
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.2305673360824585,
      "learning_rate": 0.000122317281147942,
      "loss": 2.0935,
      "step": 161690
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.1127780675888062,
      "learning_rate": 0.00012231398525390107,
      "loss": 1.8446,
      "step": 161691
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.2658562660217285,
      "learning_rate": 0.0001223106893928951,
      "loss": 2.0063,
      "step": 161692
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.2280468940734863,
      "learning_rate": 0.00012230739356492508,
      "loss": 2.0287,
      "step": 161693
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.2290502786636353,
      "learning_rate": 0.00012230409776999135,
      "loss": 2.0718,
      "step": 161694
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.1888335943222046,
      "learning_rate": 0.00012230080200809475,
      "loss": 1.889,
      "step": 161695
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.2172287702560425,
      "learning_rate": 0.00012229750627923566,
      "loss": 1.9449,
      "step": 161696
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.2248361110687256,
      "learning_rate": 0.00012229421058341505,
      "loss": 2.2565,
      "step": 161697
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.2350983619689941,
      "learning_rate": 0.00012229091492063307,
      "loss": 2.1183,
      "step": 161698
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.22723388671875,
      "learning_rate": 0.00012228761929089068,
      "loss": 1.8571,
      "step": 161699
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.213516354560852,
      "learning_rate": 0.0001222843236941883,
      "loss": 1.9579,
      "step": 161700
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.3069279193878174,
      "learning_rate": 0.00012228102813052668,
      "loss": 2.0442,
      "step": 161701
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.3039426803588867,
      "learning_rate": 0.00012227773259990626,
      "loss": 1.8692,
      "step": 161702
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.27108633518219,
      "learning_rate": 0.000122274437102328,
      "loss": 1.983,
      "step": 161703
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.3314012289047241,
      "learning_rate": 0.0001222711416377921,
      "loss": 2.2139,
      "step": 161704
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.2302461862564087,
      "learning_rate": 0.00012226784620629938,
      "loss": 2.035,
      "step": 161705
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.1586343050003052,
      "learning_rate": 0.00012226455080785036,
      "loss": 1.8952,
      "step": 161706
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.2329249382019043,
      "learning_rate": 0.00012226125544244584,
      "loss": 2.0446,
      "step": 161707
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.2288378477096558,
      "learning_rate": 0.00012225796011008615,
      "loss": 2.3309,
      "step": 161708
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.3228073120117188,
      "learning_rate": 0.00012225466481077235,
      "loss": 2.131,
      "step": 161709
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.2790318727493286,
      "learning_rate": 0.00012225136954450449,
      "loss": 2.0878,
      "step": 161710
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.2268315553665161,
      "learning_rate": 0.00012224807431128357,
      "loss": 2.08,
      "step": 161711
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.2121267318725586,
      "learning_rate": 0.00012224477911110997,
      "loss": 2.0052,
      "step": 161712
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.2573376893997192,
      "learning_rate": 0.00012224148394398453,
      "loss": 2.0705,
      "step": 161713
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.1976478099822998,
      "learning_rate": 0.00012223818880990767,
      "loss": 1.8806,
      "step": 161714
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.2697794437408447,
      "learning_rate": 0.00012223489370888032,
      "loss": 2.0925,
      "step": 161715
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.286996603012085,
      "learning_rate": 0.0001222315986409026,
      "loss": 2.0862,
      "step": 161716
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.3276021480560303,
      "learning_rate": 0.0001222283036059755,
      "loss": 2.2468,
      "step": 161717
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.3667384386062622,
      "learning_rate": 0.00012222500860409942,
      "loss": 2.0354,
      "step": 161718
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.7149442434310913,
      "learning_rate": 0.00012222171363527516,
      "loss": 2.1137,
      "step": 161719
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.1609810590744019,
      "learning_rate": 0.00012221841869950313,
      "loss": 1.7896,
      "step": 161720
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.2149665355682373,
      "learning_rate": 0.00012221512379678427,
      "loss": 2.0375,
      "step": 161721
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.204925537109375,
      "learning_rate": 0.00012221182892711876,
      "loss": 2.0503,
      "step": 161722
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.2705248594284058,
      "learning_rate": 0.00012220853409050753,
      "loss": 2.0325,
      "step": 161723
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.2835642099380493,
      "learning_rate": 0.000122205239286951,
      "loss": 1.9783,
      "step": 161724
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.2757515907287598,
      "learning_rate": 0.00012220194451644997,
      "loss": 2.099,
      "step": 161725
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.2195430994033813,
      "learning_rate": 0.00012219864977900484,
      "loss": 2.1485,
      "step": 161726
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.2118093967437744,
      "learning_rate": 0.00012219535507461656,
      "loss": 1.8296,
      "step": 161727
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.1978732347488403,
      "learning_rate": 0.00012219206040328532,
      "loss": 2.1023,
      "step": 161728
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.1956844329833984,
      "learning_rate": 0.00012218876576501202,
      "loss": 2.0398,
      "step": 161729
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.2186431884765625,
      "learning_rate": 0.00012218547115979707,
      "loss": 1.841,
      "step": 161730
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.1559975147247314,
      "learning_rate": 0.00012218217658764132,
      "loss": 1.9189,
      "step": 161731
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.3469940423965454,
      "learning_rate": 0.00012217888204854515,
      "loss": 2.0505,
      "step": 161732
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.2742336988449097,
      "learning_rate": 0.00012217558754250947,
      "loss": 1.7732,
      "step": 161733
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.2283949851989746,
      "learning_rate": 0.00012217229306953452,
      "loss": 2.152,
      "step": 161734
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.208727240562439,
      "learning_rate": 0.00012216899862962122,
      "loss": 2.1477,
      "step": 161735
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.2962647676467896,
      "learning_rate": 0.00012216570422276992,
      "loss": 1.9393,
      "step": 161736
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.3132790327072144,
      "learning_rate": 0.00012216240984898148,
      "loss": 2.0566,
      "step": 161737
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.446223497390747,
      "learning_rate": 0.0001221591155082563,
      "loss": 2.0554,
      "step": 161738
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.240370512008667,
      "learning_rate": 0.00012215582120059522,
      "loss": 2.1753,
      "step": 161739
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.4032654762268066,
      "learning_rate": 0.00012215252692599873,
      "loss": 1.9129,
      "step": 161740
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.246682047843933,
      "learning_rate": 0.00012214923268446745,
      "loss": 2.1143,
      "step": 161741
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.4266144037246704,
      "learning_rate": 0.00012214593847600187,
      "loss": 2.0529,
      "step": 161742
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.5083250999450684,
      "learning_rate": 0.0001221426443006028,
      "loss": 2.0702,
      "step": 161743
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.1522070169448853,
      "learning_rate": 0.0001221393501582707,
      "loss": 1.9869,
      "step": 161744
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.275871992111206,
      "learning_rate": 0.00012213605604900635,
      "loss": 2.2796,
      "step": 161745
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.2958629131317139,
      "learning_rate": 0.00012213276197281024,
      "loss": 2.0398,
      "step": 161746
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.218910574913025,
      "learning_rate": 0.0001221294679296829,
      "loss": 1.806,
      "step": 161747
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.1311715841293335,
      "learning_rate": 0.00012212617391962516,
      "loss": 2.093,
      "step": 161748
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.1967442035675049,
      "learning_rate": 0.00012212287994263756,
      "loss": 1.8224,
      "step": 161749
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.219713807106018,
      "learning_rate": 0.00012211958599872054,
      "loss": 1.9694,
      "step": 161750
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.2668678760528564,
      "learning_rate": 0.00012211629208787496,
      "loss": 2.0537,
      "step": 161751
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.1953555345535278,
      "learning_rate": 0.0001221129982101013,
      "loss": 1.7828,
      "step": 161752
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.2334710359573364,
      "learning_rate": 0.0001221097043654001,
      "loss": 1.8512,
      "step": 161753
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.3613545894622803,
      "learning_rate": 0.00012210641055377218,
      "loss": 2.1524,
      "step": 161754
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.4030195474624634,
      "learning_rate": 0.000122103116775218,
      "loss": 1.9044,
      "step": 161755
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.2243977785110474,
      "learning_rate": 0.00012209982302973816,
      "loss": 1.9938,
      "step": 161756
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.1571173667907715,
      "learning_rate": 0.00012209652931733338,
      "loss": 2.1606,
      "step": 161757
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.2693465948104858,
      "learning_rate": 0.00012209323563800424,
      "loss": 1.7942,
      "step": 161758
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.1631675958633423,
      "learning_rate": 0.0001220899419917512,
      "loss": 1.9988,
      "step": 161759
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.6657451391220093,
      "learning_rate": 0.0001220866483785751,
      "loss": 2.1996,
      "step": 161760
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.3256819248199463,
      "learning_rate": 0.00012208335479847634,
      "loss": 2.1207,
      "step": 161761
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.290088415145874,
      "learning_rate": 0.00012208006125145578,
      "loss": 1.9727,
      "step": 161762
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.2532397508621216,
      "learning_rate": 0.0001220767677375139,
      "loss": 1.9793,
      "step": 161763
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.4925614595413208,
      "learning_rate": 0.00012207347425665128,
      "loss": 1.9985,
      "step": 161764
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.2578037977218628,
      "learning_rate": 0.00012207018080886845,
      "loss": 1.8657,
      "step": 161765
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.3447027206420898,
      "learning_rate": 0.00012206688739416625,
      "loss": 1.9074,
      "step": 161766
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.1999812126159668,
      "learning_rate": 0.00012206359401254507,
      "loss": 2.1639,
      "step": 161767
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.340370535850525,
      "learning_rate": 0.00012206030066400573,
      "loss": 2.161,
      "step": 161768
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.1312777996063232,
      "learning_rate": 0.00012205700734854865,
      "loss": 1.8795,
      "step": 161769
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.2120734453201294,
      "learning_rate": 0.00012205371406617471,
      "loss": 2.0485,
      "step": 161770
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.1125785112380981,
      "learning_rate": 0.00012205042081688415,
      "loss": 1.9001,
      "step": 161771
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.3452569246292114,
      "learning_rate": 0.00012204712760067787,
      "loss": 1.9741,
      "step": 161772
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.1541931629180908,
      "learning_rate": 0.00012204383441755631,
      "loss": 1.9227,
      "step": 161773
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.2306417226791382,
      "learning_rate": 0.00012204054126752024,
      "loss": 1.8733,
      "step": 161774
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.2059476375579834,
      "learning_rate": 0.0001220372481505701,
      "loss": 2.0153,
      "step": 161775
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.227973461151123,
      "learning_rate": 0.00012203395506670681,
      "loss": 2.3732,
      "step": 161776
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.1379798650741577,
      "learning_rate": 0.00012203066201593054,
      "loss": 1.8821,
      "step": 161777
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.2430909872055054,
      "learning_rate": 0.00012202736899824224,
      "loss": 1.8477,
      "step": 161778
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.237058401107788,
      "learning_rate": 0.00012202407601364232,
      "loss": 2.0054,
      "step": 161779
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.2956119775772095,
      "learning_rate": 0.00012202078306213158,
      "loss": 2.0771,
      "step": 161780
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.2604944705963135,
      "learning_rate": 0.0001220174901437104,
      "loss": 2.0178,
      "step": 161781
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.269915223121643,
      "learning_rate": 0.0001220141972583798,
      "loss": 1.8554,
      "step": 161782
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.158417820930481,
      "learning_rate": 0.00012201090440613986,
      "loss": 1.8317,
      "step": 161783
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.2518154382705688,
      "learning_rate": 0.00012200761158699157,
      "loss": 2.1366,
      "step": 161784
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.296260952949524,
      "learning_rate": 0.00012200431880093532,
      "loss": 1.9961,
      "step": 161785
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.2632375955581665,
      "learning_rate": 0.00012200102604797197,
      "loss": 2.0102,
      "step": 161786
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.1513922214508057,
      "learning_rate": 0.00012199773332810183,
      "loss": 1.7839,
      "step": 161787
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.175887107849121,
      "learning_rate": 0.0001219944406413259,
      "loss": 2.0037,
      "step": 161788
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.3621066808700562,
      "learning_rate": 0.00012199114798764436,
      "loss": 2.2022,
      "step": 161789
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.256218433380127,
      "learning_rate": 0.00012198785536705814,
      "loss": 1.9793,
      "step": 161790
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.309709072113037,
      "learning_rate": 0.00012198456277956762,
      "loss": 2.1475,
      "step": 161791
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.2703996896743774,
      "learning_rate": 0.00012198127022517363,
      "loss": 1.9461,
      "step": 161792
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.2882190942764282,
      "learning_rate": 0.00012197797770387659,
      "loss": 1.9962,
      "step": 161793
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.2032326459884644,
      "learning_rate": 0.0001219746852156774,
      "loss": 1.9028,
      "step": 161794
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.3465566635131836,
      "learning_rate": 0.00012197139276057627,
      "loss": 2.1972,
      "step": 161795
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.2925945520401,
      "learning_rate": 0.00012196810033857411,
      "loss": 2.1871,
      "step": 161796
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.207312822341919,
      "learning_rate": 0.00012196480794967137,
      "loss": 2.1338,
      "step": 161797
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.3678182363510132,
      "learning_rate": 0.00012196151559386882,
      "loss": 2.1119,
      "step": 161798
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.244459867477417,
      "learning_rate": 0.00012195822327116688,
      "loss": 2.1005,
      "step": 161799
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.1909594535827637,
      "learning_rate": 0.00012195493098156647,
      "loss": 2.0393,
      "step": 161800
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.274152159690857,
      "learning_rate": 0.00012195163872506779,
      "loss": 1.9137,
      "step": 161801
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.234344244003296,
      "learning_rate": 0.00012194834650167179,
      "loss": 2.0664,
      "step": 161802
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.1215059757232666,
      "learning_rate": 0.00012194505431137884,
      "loss": 2.006,
      "step": 161803
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.1703541278839111,
      "learning_rate": 0.00012194176215418975,
      "loss": 2.2345,
      "step": 161804
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.2028976678848267,
      "learning_rate": 0.00012193847003010494,
      "loss": 2.0104,
      "step": 161805
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.1734144687652588,
      "learning_rate": 0.00012193517793912538,
      "loss": 1.8326,
      "step": 161806
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.3627530336380005,
      "learning_rate": 0.00012193188588125118,
      "loss": 2.2405,
      "step": 161807
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.2372384071350098,
      "learning_rate": 0.00012192859385648333,
      "loss": 1.9571,
      "step": 161808
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.2407076358795166,
      "learning_rate": 0.00012192530186482222,
      "loss": 2.1638,
      "step": 161809
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.2188689708709717,
      "learning_rate": 0.00012192200990626864,
      "loss": 1.9882,
      "step": 161810
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.5598454475402832,
      "learning_rate": 0.000121918717980823,
      "loss": 1.9909,
      "step": 161811
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.2495917081832886,
      "learning_rate": 0.00012191542608848618,
      "loss": 1.9582,
      "step": 161812
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.6243281364440918,
      "learning_rate": 0.0001219121342292586,
      "loss": 2.064,
      "step": 161813
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.284233808517456,
      "learning_rate": 0.00012190884240314095,
      "loss": 2.2324,
      "step": 161814
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.2605056762695312,
      "learning_rate": 0.0001219055506101337,
      "loss": 1.8278,
      "step": 161815
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.3016966581344604,
      "learning_rate": 0.00012190225885023767,
      "loss": 2.1792,
      "step": 161816
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.1011704206466675,
      "learning_rate": 0.00012189896712345327,
      "loss": 1.8849,
      "step": 161817
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.3645151853561401,
      "learning_rate": 0.00012189567542978129,
      "loss": 1.8464,
      "step": 161818
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.3458964824676514,
      "learning_rate": 0.0001218923837692223,
      "loss": 2.0077,
      "step": 161819
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.25779390335083,
      "learning_rate": 0.00012188909214177689,
      "loss": 1.9998,
      "step": 161820
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.2983344793319702,
      "learning_rate": 0.00012188580054744552,
      "loss": 1.9095,
      "step": 161821
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.1233998537063599,
      "learning_rate": 0.00012188250898622907,
      "loss": 1.8247,
      "step": 161822
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.2465102672576904,
      "learning_rate": 0.00012187921745812788,
      "loss": 2.1139,
      "step": 161823
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.4999154806137085,
      "learning_rate": 0.00012187592596314285,
      "loss": 1.8969,
      "step": 161824
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.1962865591049194,
      "learning_rate": 0.00012187263450127443,
      "loss": 2.0216,
      "step": 161825
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.1613754034042358,
      "learning_rate": 0.00012186934307252328,
      "loss": 2.2073,
      "step": 161826
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.3100850582122803,
      "learning_rate": 0.00012186605167688983,
      "loss": 2.1633,
      "step": 161827
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.3341578245162964,
      "learning_rate": 0.00012186276031437497,
      "loss": 2.1593,
      "step": 161828
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.1774680614471436,
      "learning_rate": 0.00012185946898497911,
      "loss": 2.1886,
      "step": 161829
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.1515135765075684,
      "learning_rate": 0.00012185617768870298,
      "loss": 2.1189,
      "step": 161830
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.2525500059127808,
      "learning_rate": 0.0001218528864255472,
      "loss": 2.0871,
      "step": 161831
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.1989585161209106,
      "learning_rate": 0.0001218495951955122,
      "loss": 1.8418,
      "step": 161832
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.202161431312561,
      "learning_rate": 0.00012184630399859882,
      "loss": 2.0542,
      "step": 161833
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.3000375032424927,
      "learning_rate": 0.00012184301283480763,
      "loss": 1.9725,
      "step": 161834
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.1979502439498901,
      "learning_rate": 0.00012183972170413903,
      "loss": 2.0742,
      "step": 161835
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.318670392036438,
      "learning_rate": 0.00012183643060659391,
      "loss": 2.2121,
      "step": 161836
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.3762190341949463,
      "learning_rate": 0.00012183313954217276,
      "loss": 2.1842,
      "step": 161837
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.2901315689086914,
      "learning_rate": 0.00012182984851087608,
      "loss": 2.0849,
      "step": 161838
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.4458086490631104,
      "learning_rate": 0.0001218265575127047,
      "loss": 2.0843,
      "step": 161839
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.1907936334609985,
      "learning_rate": 0.00012182326654765912,
      "loss": 2.0863,
      "step": 161840
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.1984187364578247,
      "learning_rate": 0.00012181997561573988,
      "loss": 1.8952,
      "step": 161841
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.2479398250579834,
      "learning_rate": 0.00012181668471694777,
      "loss": 1.9594,
      "step": 161842
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.3435531854629517,
      "learning_rate": 0.0001218133938512833,
      "loss": 2.0177,
      "step": 161843
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.6200213432312012,
      "learning_rate": 0.00012181010301874696,
      "loss": 2.0212,
      "step": 161844
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.3485592603683472,
      "learning_rate": 0.00012180681221933959,
      "loss": 2.124,
      "step": 161845
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.1535371541976929,
      "learning_rate": 0.00012180352145306161,
      "loss": 2.1444,
      "step": 161846
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.1223366260528564,
      "learning_rate": 0.0001218002307199138,
      "loss": 2.0935,
      "step": 161847
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.619866132736206,
      "learning_rate": 0.00012179694001989673,
      "loss": 1.9518,
      "step": 161848
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.345848798751831,
      "learning_rate": 0.00012179364935301094,
      "loss": 2.1888,
      "step": 161849
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.2095128297805786,
      "learning_rate": 0.000121790358719257,
      "loss": 2.0767,
      "step": 161850
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.1850117444992065,
      "learning_rate": 0.00012178706811863568,
      "loss": 1.9804,
      "step": 161851
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.2195647954940796,
      "learning_rate": 0.00012178377755114739,
      "loss": 1.9259,
      "step": 161852
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.2639333009719849,
      "learning_rate": 0.000121780487016793,
      "loss": 2.0416,
      "step": 161853
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.617436170578003,
      "learning_rate": 0.00012177719651557292,
      "loss": 2.0008,
      "step": 161854
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.1069341897964478,
      "learning_rate": 0.00012177390604748789,
      "loss": 1.835,
      "step": 161855
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.1855312585830688,
      "learning_rate": 0.00012177061561253832,
      "loss": 1.9802,
      "step": 161856
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.5636807680130005,
      "learning_rate": 0.00012176732521072506,
      "loss": 2.0197,
      "step": 161857
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.7351127862930298,
      "learning_rate": 0.00012176403484204854,
      "loss": 2.0304,
      "step": 161858
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.313781976699829,
      "learning_rate": 0.00012176074450650952,
      "loss": 1.9419,
      "step": 161859
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.318304181098938,
      "learning_rate": 0.00012175745420410844,
      "loss": 2.1148,
      "step": 161860
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.7180969715118408,
      "learning_rate": 0.00012175416393484626,
      "loss": 2.0715,
      "step": 161861
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.3456941843032837,
      "learning_rate": 0.0001217508736987231,
      "loss": 2.0885,
      "step": 161862
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.2873250246047974,
      "learning_rate": 0.00012174758349573994,
      "loss": 2.1155,
      "step": 161863
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.1700886487960815,
      "learning_rate": 0.00012174429332589717,
      "loss": 1.9557,
      "step": 161864
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.4282780885696411,
      "learning_rate": 0.00012174100318919558,
      "loss": 1.9863,
      "step": 161865
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.1661498546600342,
      "learning_rate": 0.00012173771308563563,
      "loss": 1.901,
      "step": 161866
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.467224359512329,
      "learning_rate": 0.00012173442301521819,
      "loss": 2.037,
      "step": 161867
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.5614949464797974,
      "learning_rate": 0.00012173113297794347,
      "loss": 2.0332,
      "step": 161868
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.401253342628479,
      "learning_rate": 0.00012172784297381241,
      "loss": 1.9367,
      "step": 161869
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.3552968502044678,
      "learning_rate": 0.00012172455300282539,
      "loss": 2.1485,
      "step": 161870
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.155727505683899,
      "learning_rate": 0.00012172126306498327,
      "loss": 1.8918,
      "step": 161871
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.2791368961334229,
      "learning_rate": 0.00012171797316028645,
      "loss": 2.0237,
      "step": 161872
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.2392429113388062,
      "learning_rate": 0.0001217146832887358,
      "loss": 2.2662,
      "step": 161873
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.2220356464385986,
      "learning_rate": 0.00012171139345033153,
      "loss": 2.0673,
      "step": 161874
    },
    {
      "epoch": 2.11,
      "grad_norm": 2.3012475967407227,
      "learning_rate": 0.00012170810364507461,
      "loss": 1.9877,
      "step": 161875
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.2682263851165771,
      "learning_rate": 0.00012170481387296543,
      "loss": 1.9492,
      "step": 161876
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.2822299003601074,
      "learning_rate": 0.00012170152413400474,
      "loss": 1.887,
      "step": 161877
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.3598488569259644,
      "learning_rate": 0.00012169823442819306,
      "loss": 1.9193,
      "step": 161878
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.2999162673950195,
      "learning_rate": 0.00012169494475553111,
      "loss": 1.9707,
      "step": 161879
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.1634764671325684,
      "learning_rate": 0.00012169165511601945,
      "loss": 1.9179,
      "step": 161880
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.2503799200057983,
      "learning_rate": 0.00012168836550965866,
      "loss": 2.1201,
      "step": 161881
    },
    {
      "epoch": 2.11,
      "grad_norm": 2.294428825378418,
      "learning_rate": 0.00012168507593644928,
      "loss": 2.0026,
      "step": 161882
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.3878430128097534,
      "learning_rate": 0.00012168178639639213,
      "loss": 2.1632,
      "step": 161883
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.2423714399337769,
      "learning_rate": 0.00012167849688948756,
      "loss": 2.2467,
      "step": 161884
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.1996554136276245,
      "learning_rate": 0.00012167520741573643,
      "loss": 2.0699,
      "step": 161885
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.2739160060882568,
      "learning_rate": 0.00012167191797513928,
      "loss": 2.1893,
      "step": 161886
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.3504748344421387,
      "learning_rate": 0.00012166862856769666,
      "loss": 1.9459,
      "step": 161887
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.0791642665863037,
      "learning_rate": 0.0001216653391934091,
      "loss": 2.2097,
      "step": 161888
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.2380666732788086,
      "learning_rate": 0.00012166204985227746,
      "loss": 1.9351,
      "step": 161889
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.242866039276123,
      "learning_rate": 0.00012165876054430208,
      "loss": 2.0783,
      "step": 161890
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.2496752738952637,
      "learning_rate": 0.00012165547126948382,
      "loss": 2.1539,
      "step": 161891
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.236061930656433,
      "learning_rate": 0.00012165218202782315,
      "loss": 1.9926,
      "step": 161892
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.2224984169006348,
      "learning_rate": 0.00012164889281932071,
      "loss": 2.0174,
      "step": 161893
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.2031099796295166,
      "learning_rate": 0.00012164560364397701,
      "loss": 1.9996,
      "step": 161894
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.266802430152893,
      "learning_rate": 0.00012164231450179287,
      "loss": 2.0186,
      "step": 161895
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.2258296012878418,
      "learning_rate": 0.00012163902539276869,
      "loss": 2.0304,
      "step": 161896
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.2228165864944458,
      "learning_rate": 0.0001216357363169053,
      "loss": 2.1559,
      "step": 161897
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.2167267799377441,
      "learning_rate": 0.00012163244727420317,
      "loss": 1.8812,
      "step": 161898
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.2002931833267212,
      "learning_rate": 0.00012162915826466291,
      "loss": 2.09,
      "step": 161899
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.304746150970459,
      "learning_rate": 0.0001216258692882851,
      "loss": 2.0638,
      "step": 161900
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.1581259965896606,
      "learning_rate": 0.0001216225803450705,
      "loss": 1.9244,
      "step": 161901
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.1957379579544067,
      "learning_rate": 0.0001216192914350195,
      "loss": 2.2642,
      "step": 161902
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.1338571310043335,
      "learning_rate": 0.00012161600255813297,
      "loss": 1.8753,
      "step": 161903
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.1092501878738403,
      "learning_rate": 0.00012161271371441137,
      "loss": 2.2595,
      "step": 161904
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.2161579132080078,
      "learning_rate": 0.00012160942490385538,
      "loss": 1.9297,
      "step": 161905
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.2606587409973145,
      "learning_rate": 0.00012160613612646542,
      "loss": 2.308,
      "step": 161906
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.2074943780899048,
      "learning_rate": 0.00012160284738224234,
      "loss": 1.9489,
      "step": 161907
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.1450155973434448,
      "learning_rate": 0.0001215995586711866,
      "loss": 2.027,
      "step": 161908
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.3082677125930786,
      "learning_rate": 0.00012159626999329895,
      "loss": 2.0782,
      "step": 161909
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.306038498878479,
      "learning_rate": 0.0001215929813485799,
      "loss": 2.0951,
      "step": 161910
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.2750344276428223,
      "learning_rate": 0.00012158969273703012,
      "loss": 1.9337,
      "step": 161911
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.2590711116790771,
      "learning_rate": 0.00012158640415865004,
      "loss": 2.1545,
      "step": 161912
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.2316159009933472,
      "learning_rate": 0.00012158311561344057,
      "loss": 2.0822,
      "step": 161913
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.263494849205017,
      "learning_rate": 0.00012157982710140201,
      "loss": 2.1832,
      "step": 161914
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.1807653903961182,
      "learning_rate": 0.00012157653862253528,
      "loss": 2.1776,
      "step": 161915
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.4591702222824097,
      "learning_rate": 0.00012157325017684079,
      "loss": 1.693,
      "step": 161916
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.2416603565216064,
      "learning_rate": 0.00012156996176431924,
      "loss": 2.1043,
      "step": 161917
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.2357600927352905,
      "learning_rate": 0.00012156667338497108,
      "loss": 2.0558,
      "step": 161918
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.1904548406600952,
      "learning_rate": 0.00012156338503879715,
      "loss": 2.1525,
      "step": 161919
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.3076367378234863,
      "learning_rate": 0.00012156009672579786,
      "loss": 1.9306,
      "step": 161920
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.2296191453933716,
      "learning_rate": 0.000121556808445974,
      "loss": 2.1709,
      "step": 161921
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.287527322769165,
      "learning_rate": 0.00012155352019932613,
      "loss": 1.7778,
      "step": 161922
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.3455880880355835,
      "learning_rate": 0.00012155023198585468,
      "loss": 2.0837,
      "step": 161923
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.165561318397522,
      "learning_rate": 0.00012154694380556057,
      "loss": 2.0966,
      "step": 161924
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.347764492034912,
      "learning_rate": 0.00012154365565844422,
      "loss": 2.2007,
      "step": 161925
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.3590574264526367,
      "learning_rate": 0.00012154036754450616,
      "loss": 1.951,
      "step": 161926
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.289872407913208,
      "learning_rate": 0.00012153707946374723,
      "loss": 2.0863,
      "step": 161927
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.0809972286224365,
      "learning_rate": 0.00012153379141616794,
      "loss": 1.9357,
      "step": 161928
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.3061938285827637,
      "learning_rate": 0.00012153050340176877,
      "loss": 1.9665,
      "step": 161929
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.3149994611740112,
      "learning_rate": 0.00012152721542055054,
      "loss": 1.9598,
      "step": 161930
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.1896631717681885,
      "learning_rate": 0.00012152392747251367,
      "loss": 2.0758,
      "step": 161931
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.4146736860275269,
      "learning_rate": 0.00012152063955765899,
      "loss": 1.9324,
      "step": 161932
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.1237597465515137,
      "learning_rate": 0.00012151735167598698,
      "loss": 2.189,
      "step": 161933
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.2718333005905151,
      "learning_rate": 0.00012151406382749828,
      "loss": 2.1005,
      "step": 161934
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.3673779964447021,
      "learning_rate": 0.00012151077601219337,
      "loss": 2.1536,
      "step": 161935
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.2545028924942017,
      "learning_rate": 0.00012150748823007304,
      "loss": 2.1742,
      "step": 161936
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.0989618301391602,
      "learning_rate": 0.00012150420048113777,
      "loss": 2.0809,
      "step": 161937
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.252875566482544,
      "learning_rate": 0.00012150091276538836,
      "loss": 2.3305,
      "step": 161938
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.1958626508712769,
      "learning_rate": 0.00012149762508282528,
      "loss": 1.9626,
      "step": 161939
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.2122875452041626,
      "learning_rate": 0.00012149433743344917,
      "loss": 2.0948,
      "step": 161940
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.4864981174468994,
      "learning_rate": 0.00012149104981726053,
      "loss": 1.8782,
      "step": 161941
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.2123045921325684,
      "learning_rate": 0.00012148776223426015,
      "loss": 2.2766,
      "step": 161942
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.1315094232559204,
      "learning_rate": 0.0001214844746844485,
      "loss": 2.1063,
      "step": 161943
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.2804794311523438,
      "learning_rate": 0.00012148118716782635,
      "loss": 2.0177,
      "step": 161944
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.3594211339950562,
      "learning_rate": 0.0001214778996843941,
      "loss": 2.1351,
      "step": 161945
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.1475067138671875,
      "learning_rate": 0.00012147461223415257,
      "loss": 1.8421,
      "step": 161946
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.2089667320251465,
      "learning_rate": 0.0001214713248171023,
      "loss": 2.1775,
      "step": 161947
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.158781886100769,
      "learning_rate": 0.00012146803743324389,
      "loss": 1.8913,
      "step": 161948
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.2323594093322754,
      "learning_rate": 0.00012146475008257781,
      "loss": 1.8292,
      "step": 161949
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.1878777742385864,
      "learning_rate": 0.00012146146276510492,
      "loss": 1.9513,
      "step": 161950
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.28135347366333,
      "learning_rate": 0.00012145817548082563,
      "loss": 1.9631,
      "step": 161951
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.2186399698257446,
      "learning_rate": 0.00012145488822974069,
      "loss": 2.1914,
      "step": 161952
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.1904958486557007,
      "learning_rate": 0.00012145160101185069,
      "loss": 1.9913,
      "step": 161953
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.205387830734253,
      "learning_rate": 0.00012144831382715622,
      "loss": 2.3055,
      "step": 161954
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.3763834238052368,
      "learning_rate": 0.00012144502667565775,
      "loss": 1.9455,
      "step": 161955
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.259607195854187,
      "learning_rate": 0.00012144173955735613,
      "loss": 2.01,
      "step": 161956
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.2504514455795288,
      "learning_rate": 0.00012143845247225176,
      "loss": 1.8904,
      "step": 161957
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.2758517265319824,
      "learning_rate": 0.00012143516542034546,
      "loss": 1.924,
      "step": 161958
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.3035497665405273,
      "learning_rate": 0.00012143187840163773,
      "loss": 2.0832,
      "step": 161959
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.2927778959274292,
      "learning_rate": 0.0001214285914161292,
      "loss": 2.1982,
      "step": 161960
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.2313032150268555,
      "learning_rate": 0.00012142530446382032,
      "loss": 2.2009,
      "step": 161961
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.251108169555664,
      "learning_rate": 0.000121422017544712,
      "loss": 2.0876,
      "step": 161962
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.169373631477356,
      "learning_rate": 0.00012141873065880459,
      "loss": 2.0744,
      "step": 161963
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.253196358680725,
      "learning_rate": 0.00012141544380609887,
      "loss": 2.0926,
      "step": 161964
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.1536277532577515,
      "learning_rate": 0.0001214121569865954,
      "loss": 2.1091,
      "step": 161965
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.2216732501983643,
      "learning_rate": 0.00012140887020029481,
      "loss": 2.1215,
      "step": 161966
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.1769530773162842,
      "learning_rate": 0.00012140558344719756,
      "loss": 1.8975,
      "step": 161967
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.382184624671936,
      "learning_rate": 0.00012140229672730451,
      "loss": 1.9138,
      "step": 161968
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.3900727033615112,
      "learning_rate": 0.00012139901004061601,
      "loss": 2.1909,
      "step": 161969
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.1951987743377686,
      "learning_rate": 0.00012139572338713294,
      "loss": 1.9663,
      "step": 161970
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.1746010780334473,
      "learning_rate": 0.00012139243676685573,
      "loss": 2.0461,
      "step": 161971
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.1457386016845703,
      "learning_rate": 0.00012138915017978508,
      "loss": 2.1054,
      "step": 161972
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.4182169437408447,
      "learning_rate": 0.00012138586362592142,
      "loss": 2.1859,
      "step": 161973
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.2115497589111328,
      "learning_rate": 0.00012138257710526563,
      "loss": 2.1837,
      "step": 161974
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.2696439027786255,
      "learning_rate": 0.00012137929061781808,
      "loss": 1.9747,
      "step": 161975
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.255157470703125,
      "learning_rate": 0.0001213760041635796,
      "loss": 2.0169,
      "step": 161976
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.1937870979309082,
      "learning_rate": 0.00012137271774255069,
      "loss": 2.1081,
      "step": 161977
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.268523931503296,
      "learning_rate": 0.00012136943135473197,
      "loss": 1.9103,
      "step": 161978
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.2438267469406128,
      "learning_rate": 0.00012136614500012393,
      "loss": 1.9908,
      "step": 161979
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.3109817504882812,
      "learning_rate": 0.00012136285867872739,
      "loss": 2.0699,
      "step": 161980
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.1627544164657593,
      "learning_rate": 0.00012135957239054277,
      "loss": 2.0263,
      "step": 161981
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.4754455089569092,
      "learning_rate": 0.0001213562861355709,
      "loss": 2.0504,
      "step": 161982
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.1218852996826172,
      "learning_rate": 0.00012135299991381224,
      "loss": 2.2685,
      "step": 161983
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.2564833164215088,
      "learning_rate": 0.00012134971372526747,
      "loss": 2.012,
      "step": 161984
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.3282287120819092,
      "learning_rate": 0.00012134642756993702,
      "loss": 1.945,
      "step": 161985
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.184231162071228,
      "learning_rate": 0.00012134314144782175,
      "loss": 2.0331,
      "step": 161986
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.1147326231002808,
      "learning_rate": 0.00012133985535892206,
      "loss": 1.9735,
      "step": 161987
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.222377896308899,
      "learning_rate": 0.00012133656930323878,
      "loss": 2.0347,
      "step": 161988
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.2007299661636353,
      "learning_rate": 0.00012133328328077237,
      "loss": 2.0977,
      "step": 161989
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.1568800210952759,
      "learning_rate": 0.00012132999729152354,
      "loss": 2.1225,
      "step": 161990
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.1344237327575684,
      "learning_rate": 0.00012132671133549266,
      "loss": 1.8921,
      "step": 161991
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.2936182022094727,
      "learning_rate": 0.00012132342541268069,
      "loss": 2.1679,
      "step": 161992
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.1404088735580444,
      "learning_rate": 0.0001213201395230879,
      "loss": 2.2847,
      "step": 161993
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.1147013902664185,
      "learning_rate": 0.00012131685366671521,
      "loss": 1.8363,
      "step": 161994
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.3032448291778564,
      "learning_rate": 0.00012131356784356307,
      "loss": 2.1663,
      "step": 161995
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.2201447486877441,
      "learning_rate": 0.00012131028205363213,
      "loss": 1.9654,
      "step": 161996
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.1888543367385864,
      "learning_rate": 0.00012130699629692288,
      "loss": 2.0789,
      "step": 161997
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.1442514657974243,
      "learning_rate": 0.00012130371057343612,
      "loss": 1.9051,
      "step": 161998
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.3316315412521362,
      "learning_rate": 0.00012130042488317226,
      "loss": 2.1346,
      "step": 161999
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.3311129808425903,
      "learning_rate": 0.00012129713922613216,
      "loss": 2.2389,
      "step": 162000
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.3172918558120728,
      "learning_rate": 0.00012129385360231626,
      "loss": 2.2074,
      "step": 162001
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.263685703277588,
      "learning_rate": 0.00012129056801172525,
      "loss": 2.0166,
      "step": 162002
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.20194411277771,
      "learning_rate": 0.00012128728245435956,
      "loss": 2.1559,
      "step": 162003
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.1566636562347412,
      "learning_rate": 0.00012128399693022005,
      "loss": 2.1228,
      "step": 162004
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.210753083229065,
      "learning_rate": 0.00012128071143930712,
      "loss": 1.8675,
      "step": 162005
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.1402215957641602,
      "learning_rate": 0.00012127742598162158,
      "loss": 1.924,
      "step": 162006
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.2018221616744995,
      "learning_rate": 0.00012127414055716392,
      "loss": 1.9325,
      "step": 162007
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.2468035221099854,
      "learning_rate": 0.00012127085516593468,
      "loss": 1.9353,
      "step": 162008
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.2848297357559204,
      "learning_rate": 0.00012126756980793467,
      "loss": 2.1307,
      "step": 162009
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.3278990983963013,
      "learning_rate": 0.00012126428448316441,
      "loss": 2.2711,
      "step": 162010
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.276963233947754,
      "learning_rate": 0.00012126099919162437,
      "loss": 1.9623,
      "step": 162011
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.2149953842163086,
      "learning_rate": 0.00012125771393331537,
      "loss": 1.8406,
      "step": 162012
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.1255483627319336,
      "learning_rate": 0.00012125442870823786,
      "loss": 1.9844,
      "step": 162013
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.2042748928070068,
      "learning_rate": 0.00012125114351639264,
      "loss": 2.345,
      "step": 162014
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.0876215696334839,
      "learning_rate": 0.00012124785835778016,
      "loss": 2.275,
      "step": 162015
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.20828378200531,
      "learning_rate": 0.0001212445732324011,
      "loss": 1.9819,
      "step": 162016
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.2874963283538818,
      "learning_rate": 0.00012124128814025596,
      "loss": 2.033,
      "step": 162017
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.1645957231521606,
      "learning_rate": 0.00012123800308134553,
      "loss": 2.1197,
      "step": 162018
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.0971726179122925,
      "learning_rate": 0.00012123471805567025,
      "loss": 2.1738,
      "step": 162019
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.2289077043533325,
      "learning_rate": 0.00012123143306323087,
      "loss": 2.1451,
      "step": 162020
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.1384752988815308,
      "learning_rate": 0.00012122814810402796,
      "loss": 1.9545,
      "step": 162021
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.1298800706863403,
      "learning_rate": 0.00012122486317806202,
      "loss": 1.9181,
      "step": 162022
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.2446421384811401,
      "learning_rate": 0.00012122157828533385,
      "loss": 2.0464,
      "step": 162023
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.2110852003097534,
      "learning_rate": 0.00012121829342584397,
      "loss": 2.0395,
      "step": 162024
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.0905797481536865,
      "learning_rate": 0.00012121500859959286,
      "loss": 1.97,
      "step": 162025
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.3870755434036255,
      "learning_rate": 0.00012121172380658138,
      "loss": 1.9251,
      "step": 162026
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.2335937023162842,
      "learning_rate": 0.00012120843904680998,
      "loss": 2.2351,
      "step": 162027
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.2514880895614624,
      "learning_rate": 0.00012120515432027921,
      "loss": 2.2148,
      "step": 162028
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.2031179666519165,
      "learning_rate": 0.00012120186962698992,
      "loss": 1.966,
      "step": 162029
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.2983429431915283,
      "learning_rate": 0.00012119858496694244,
      "loss": 2.0062,
      "step": 162030
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.3515868186950684,
      "learning_rate": 0.00012119530034013766,
      "loss": 2.0425,
      "step": 162031
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.3338485956192017,
      "learning_rate": 0.00012119201574657603,
      "loss": 2.0907,
      "step": 162032
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.329093337059021,
      "learning_rate": 0.00012118873118625815,
      "loss": 1.9142,
      "step": 162033
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.2085283994674683,
      "learning_rate": 0.00012118544665918457,
      "loss": 1.754,
      "step": 162034
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.4140825271606445,
      "learning_rate": 0.00012118216216535613,
      "loss": 2.0749,
      "step": 162035
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.263856053352356,
      "learning_rate": 0.00012117887770477314,
      "loss": 2.1094,
      "step": 162036
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.1909680366516113,
      "learning_rate": 0.00012117559327743651,
      "loss": 1.9612,
      "step": 162037
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.208892822265625,
      "learning_rate": 0.00012117230888334671,
      "loss": 1.8662,
      "step": 162038
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.3124819993972778,
      "learning_rate": 0.00012116902452250435,
      "loss": 2.1649,
      "step": 162039
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.1623082160949707,
      "learning_rate": 0.00012116574019490997,
      "loss": 1.9585,
      "step": 162040
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.227263331413269,
      "learning_rate": 0.00012116245590056432,
      "loss": 2.1427,
      "step": 162041
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.207511067390442,
      "learning_rate": 0.00012115917163946784,
      "loss": 2.1396,
      "step": 162042
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.276288390159607,
      "learning_rate": 0.00012115588741162136,
      "loss": 2.2791,
      "step": 162043
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.1256558895111084,
      "learning_rate": 0.00012115260321702538,
      "loss": 1.8994,
      "step": 162044
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.2277448177337646,
      "learning_rate": 0.0001211493190556805,
      "loss": 2.1806,
      "step": 162045
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.2346876859664917,
      "learning_rate": 0.00012114603492758727,
      "loss": 2.2294,
      "step": 162046
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.2256819009780884,
      "learning_rate": 0.00012114275083274643,
      "loss": 2.1485,
      "step": 162047
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.2155879735946655,
      "learning_rate": 0.00012113946677115842,
      "loss": 2.2704,
      "step": 162048
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.1356513500213623,
      "learning_rate": 0.00012113618274282409,
      "loss": 2.159,
      "step": 162049
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.3437342643737793,
      "learning_rate": 0.00012113289874774392,
      "loss": 1.9672,
      "step": 162050
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.2214884757995605,
      "learning_rate": 0.0001211296147859185,
      "loss": 2.0587,
      "step": 162051
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.2616262435913086,
      "learning_rate": 0.00012112633085734836,
      "loss": 2.1669,
      "step": 162052
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.2204536199569702,
      "learning_rate": 0.00012112304696203432,
      "loss": 2.0895,
      "step": 162053
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.2062082290649414,
      "learning_rate": 0.00012111976309997681,
      "loss": 2.0018,
      "step": 162054
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.1266363859176636,
      "learning_rate": 0.00012111647927117656,
      "loss": 1.9289,
      "step": 162055
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.2881871461868286,
      "learning_rate": 0.00012111319547563418,
      "loss": 2.0477,
      "step": 162056
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.1025056838989258,
      "learning_rate": 0.00012110991171335021,
      "loss": 1.998,
      "step": 162057
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.2260037660598755,
      "learning_rate": 0.00012110662798432519,
      "loss": 2.0123,
      "step": 162058
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.136339545249939,
      "learning_rate": 0.00012110334428855992,
      "loss": 1.9844,
      "step": 162059
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.3863306045532227,
      "learning_rate": 0.00012110006062605481,
      "loss": 1.9774,
      "step": 162060
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.187211513519287,
      "learning_rate": 0.00012109677699681068,
      "loss": 2.0251,
      "step": 162061
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.248236060142517,
      "learning_rate": 0.00012109349340082806,
      "loss": 2.1468,
      "step": 162062
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.196139931678772,
      "learning_rate": 0.0001210902098381075,
      "loss": 1.9034,
      "step": 162063
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.3036084175109863,
      "learning_rate": 0.00012108692630864956,
      "loss": 2.105,
      "step": 162064
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.3001285791397095,
      "learning_rate": 0.00012108364281245506,
      "loss": 2.0971,
      "step": 162065
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.2566107511520386,
      "learning_rate": 0.00012108035934952435,
      "loss": 1.8994,
      "step": 162066
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.1821868419647217,
      "learning_rate": 0.0001210770759198583,
      "loss": 2.1437,
      "step": 162067
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.2626450061798096,
      "learning_rate": 0.0001210737925234574,
      "loss": 2.0439,
      "step": 162068
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.2411843538284302,
      "learning_rate": 0.00012107050916032225,
      "loss": 2.1725,
      "step": 162069
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.0922069549560547,
      "learning_rate": 0.0001210672258304534,
      "loss": 1.9591,
      "step": 162070
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.2134169340133667,
      "learning_rate": 0.0001210639425338516,
      "loss": 2.1266,
      "step": 162071
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.2671018838882446,
      "learning_rate": 0.00012106065927051728,
      "loss": 2.1286,
      "step": 162072
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.167576551437378,
      "learning_rate": 0.00012105737604045128,
      "loss": 2.2489,
      "step": 162073
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.3963818550109863,
      "learning_rate": 0.00012105409284365411,
      "loss": 1.9754,
      "step": 162074
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.2953095436096191,
      "learning_rate": 0.00012105080968012633,
      "loss": 2.1518,
      "step": 162075
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.1599268913269043,
      "learning_rate": 0.00012104752654986851,
      "loss": 2.0879,
      "step": 162076
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.1157866716384888,
      "learning_rate": 0.00012104424345288139,
      "loss": 1.9808,
      "step": 162077
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.2879446744918823,
      "learning_rate": 0.00012104096038916544,
      "loss": 1.987,
      "step": 162078
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.234675407409668,
      "learning_rate": 0.00012103767735872148,
      "loss": 2.1301,
      "step": 162079
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.3609665632247925,
      "learning_rate": 0.00012103439436154989,
      "loss": 2.022,
      "step": 162080
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.0801595449447632,
      "learning_rate": 0.00012103111139765156,
      "loss": 2.1206,
      "step": 162081
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.2529643774032593,
      "learning_rate": 0.00012102782846702674,
      "loss": 2.1926,
      "step": 162082
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.2169663906097412,
      "learning_rate": 0.00012102454556967631,
      "loss": 2.0491,
      "step": 162083
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.1353402137756348,
      "learning_rate": 0.00012102126270560073,
      "loss": 2.2495,
      "step": 162084
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.3021948337554932,
      "learning_rate": 0.00012101797987480074,
      "loss": 2.0558,
      "step": 162085
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.3992276191711426,
      "learning_rate": 0.00012101469707727682,
      "loss": 1.9255,
      "step": 162086
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.4356247186660767,
      "learning_rate": 0.0001210114143130298,
      "loss": 2.2674,
      "step": 162087
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.2277740240097046,
      "learning_rate": 0.00012100813158205999,
      "loss": 2.0294,
      "step": 162088
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.4171719551086426,
      "learning_rate": 0.00012100484888436822,
      "loss": 2.2113,
      "step": 162089
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.3400802612304688,
      "learning_rate": 0.00012100156621995492,
      "loss": 2.0694,
      "step": 162090
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.2820006608963013,
      "learning_rate": 0.00012099828358882089,
      "loss": 1.9856,
      "step": 162091
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.319138526916504,
      "learning_rate": 0.00012099500099096659,
      "loss": 2.0724,
      "step": 162092
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.1598981618881226,
      "learning_rate": 0.0001209917184263928,
      "loss": 1.9936,
      "step": 162093
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.2706663608551025,
      "learning_rate": 0.00012098843589510002,
      "loss": 2.0172,
      "step": 162094
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.7184029817581177,
      "learning_rate": 0.00012098515339708885,
      "loss": 2.0644,
      "step": 162095
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.2261528968811035,
      "learning_rate": 0.00012098187093235985,
      "loss": 1.8897,
      "step": 162096
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.2064121961593628,
      "learning_rate": 0.00012097858850091378,
      "loss": 1.9146,
      "step": 162097
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.199978232383728,
      "learning_rate": 0.00012097530610275107,
      "loss": 1.9529,
      "step": 162098
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.2479907274246216,
      "learning_rate": 0.00012097202373787254,
      "loss": 1.9334,
      "step": 162099
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.3857160806655884,
      "learning_rate": 0.00012096874140627871,
      "loss": 2.1695,
      "step": 162100
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.3973151445388794,
      "learning_rate": 0.00012096545910797012,
      "loss": 1.8613,
      "step": 162101
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.385504126548767,
      "learning_rate": 0.00012096217684294736,
      "loss": 2.3026,
      "step": 162102
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.238471269607544,
      "learning_rate": 0.00012095889461121122,
      "loss": 1.9227,
      "step": 162103
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.279177188873291,
      "learning_rate": 0.00012095561241276209,
      "loss": 2.1054,
      "step": 162104
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.227360725402832,
      "learning_rate": 0.00012095233024760081,
      "loss": 2.1234,
      "step": 162105
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.3616294860839844,
      "learning_rate": 0.00012094904811572786,
      "loss": 1.9917,
      "step": 162106
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.3092703819274902,
      "learning_rate": 0.00012094576601714375,
      "loss": 2.0509,
      "step": 162107
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.200255036354065,
      "learning_rate": 0.0001209424839518493,
      "loss": 2.2078,
      "step": 162108
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.323152780532837,
      "learning_rate": 0.00012093920191984504,
      "loss": 1.9737,
      "step": 162109
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.197926640510559,
      "learning_rate": 0.00012093591992113142,
      "loss": 2.0451,
      "step": 162110
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.3752855062484741,
      "learning_rate": 0.00012093263795570934,
      "loss": 1.9381,
      "step": 162111
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.2914185523986816,
      "learning_rate": 0.00012092935602357924,
      "loss": 2.2651,
      "step": 162112
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.2244532108306885,
      "learning_rate": 0.00012092607412474165,
      "loss": 2.1719,
      "step": 162113
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.175898551940918,
      "learning_rate": 0.00012092279225919739,
      "loss": 1.9193,
      "step": 162114
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.3184705972671509,
      "learning_rate": 0.00012091951042694687,
      "loss": 2.2188,
      "step": 162115
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.2084542512893677,
      "learning_rate": 0.00012091622862799091,
      "loss": 1.8749,
      "step": 162116
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.1884727478027344,
      "learning_rate": 0.00012091294686232998,
      "loss": 2.2068,
      "step": 162117
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.3312230110168457,
      "learning_rate": 0.0001209096651299647,
      "loss": 2.02,
      "step": 162118
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.576978325843811,
      "learning_rate": 0.0001209063834308956,
      "loss": 2.0801,
      "step": 162119
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.157301902770996,
      "learning_rate": 0.00012090310176512351,
      "loss": 1.8693,
      "step": 162120
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.1271240711212158,
      "learning_rate": 0.00012089982013264879,
      "loss": 2.0902,
      "step": 162121
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.5807174444198608,
      "learning_rate": 0.00012089653853347228,
      "loss": 2.0725,
      "step": 162122
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.2569810152053833,
      "learning_rate": 0.00012089325696759448,
      "loss": 1.9766,
      "step": 162123
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.1164220571517944,
      "learning_rate": 0.00012088997543501604,
      "loss": 2.0328,
      "step": 162124
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.2378301620483398,
      "learning_rate": 0.00012088669393573741,
      "loss": 2.0409,
      "step": 162125
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.25564706325531,
      "learning_rate": 0.0001208834124697594,
      "loss": 2.1889,
      "step": 162126
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.154946208000183,
      "learning_rate": 0.00012088013103708248,
      "loss": 1.9676,
      "step": 162127
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.3272494077682495,
      "learning_rate": 0.0001208768496377074,
      "loss": 1.8691,
      "step": 162128
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.287008285522461,
      "learning_rate": 0.0001208735682716347,
      "loss": 2.024,
      "step": 162129
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.2270004749298096,
      "learning_rate": 0.000120870286938865,
      "loss": 2.1411,
      "step": 162130
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.2767449617385864,
      "learning_rate": 0.0001208670056393988,
      "loss": 2.2307,
      "step": 162131
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.1223622560501099,
      "learning_rate": 0.00012086372437323687,
      "loss": 2.0649,
      "step": 162132
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.1102595329284668,
      "learning_rate": 0.00012086044314037969,
      "loss": 1.8817,
      "step": 162133
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.2516615390777588,
      "learning_rate": 0.000120857161940828,
      "loss": 2.0946,
      "step": 162134
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.3406739234924316,
      "learning_rate": 0.0001208538807745824,
      "loss": 1.8996,
      "step": 162135
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.1159164905548096,
      "learning_rate": 0.00012085059964164337,
      "loss": 1.7604,
      "step": 162136
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.2060139179229736,
      "learning_rate": 0.00012084731854201155,
      "loss": 2.186,
      "step": 162137
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.443195104598999,
      "learning_rate": 0.00012084403747568768,
      "loss": 2.0093,
      "step": 162138
    },
    {
      "epoch": 2.11,
      "grad_norm": 2.058333158493042,
      "learning_rate": 0.00012084075644267217,
      "loss": 2.1346,
      "step": 162139
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.2728540897369385,
      "learning_rate": 0.00012083747544296588,
      "loss": 2.0807,
      "step": 162140
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.2382738590240479,
      "learning_rate": 0.00012083419447656928,
      "loss": 2.1387,
      "step": 162141
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.1251987218856812,
      "learning_rate": 0.00012083091354348296,
      "loss": 1.9364,
      "step": 162142
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.2061582803726196,
      "learning_rate": 0.00012082763264370747,
      "loss": 2.1374,
      "step": 162143
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.2503609657287598,
      "learning_rate": 0.0001208243517772436,
      "loss": 2.2964,
      "step": 162144
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.1801382303237915,
      "learning_rate": 0.00012082107094409177,
      "loss": 1.8229,
      "step": 162145
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.2008041143417358,
      "learning_rate": 0.00012081779014425278,
      "loss": 2.1498,
      "step": 162146
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.2313393354415894,
      "learning_rate": 0.00012081450937772704,
      "loss": 1.9433,
      "step": 162147
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.1489193439483643,
      "learning_rate": 0.0001208112286445155,
      "loss": 2.0318,
      "step": 162148
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.2958824634552002,
      "learning_rate": 0.00012080794794461829,
      "loss": 1.9471,
      "step": 162149
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.14060378074646,
      "learning_rate": 0.00012080466727803639,
      "loss": 2.0062,
      "step": 162150
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.0839157104492188,
      "learning_rate": 0.00012080138664477018,
      "loss": 2.0095,
      "step": 162151
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.2229743003845215,
      "learning_rate": 0.00012079810604482046,
      "loss": 1.9933,
      "step": 162152
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.2999815940856934,
      "learning_rate": 0.00012079482547818767,
      "loss": 2.0441,
      "step": 162153
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.0677928924560547,
      "learning_rate": 0.0001207915449448727,
      "loss": 2.0749,
      "step": 162154
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.2358545064926147,
      "learning_rate": 0.00012078826444487576,
      "loss": 2.1943,
      "step": 162155
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.2019346952438354,
      "learning_rate": 0.00012078498397819779,
      "loss": 2.2158,
      "step": 162156
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.375077247619629,
      "learning_rate": 0.00012078170354483913,
      "loss": 2.1794,
      "step": 162157
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.2482671737670898,
      "learning_rate": 0.0001207784231448007,
      "loss": 2.1396,
      "step": 162158
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.4463551044464111,
      "learning_rate": 0.00012077514277808277,
      "loss": 1.9922,
      "step": 162159
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.2222959995269775,
      "learning_rate": 0.00012077186244468639,
      "loss": 1.8506,
      "step": 162160
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.4235252141952515,
      "learning_rate": 0.00012076858214461164,
      "loss": 1.9021,
      "step": 162161
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.266378402709961,
      "learning_rate": 0.00012076530187785956,
      "loss": 2.0621,
      "step": 162162
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.1361749172210693,
      "learning_rate": 0.00012076202164443044,
      "loss": 1.842,
      "step": 162163
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.2699326276779175,
      "learning_rate": 0.00012075874144432516,
      "loss": 2.0126,
      "step": 162164
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.1697968244552612,
      "learning_rate": 0.00012075546127754411,
      "loss": 2.1226,
      "step": 162165
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.24073326587677,
      "learning_rate": 0.00012075218114408823,
      "loss": 1.9835,
      "step": 162166
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.1722584962844849,
      "learning_rate": 0.00012074890104395768,
      "loss": 1.9474,
      "step": 162167
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.1040928363800049,
      "learning_rate": 0.00012074562097715342,
      "loss": 1.9689,
      "step": 162168
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.1718955039978027,
      "learning_rate": 0.00012074234094367579,
      "loss": 2.2076,
      "step": 162169
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.1507223844528198,
      "learning_rate": 0.00012073906094352567,
      "loss": 2.0532,
      "step": 162170
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.287697196006775,
      "learning_rate": 0.00012073578097670342,
      "loss": 2.0005,
      "step": 162171
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.0899854898452759,
      "learning_rate": 0.00012073250104320998,
      "loss": 2.0718,
      "step": 162172
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.1602624654769897,
      "learning_rate": 0.00012072922114304558,
      "loss": 1.9369,
      "step": 162173
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.2275278568267822,
      "learning_rate": 0.00012072594127621106,
      "loss": 2.1264,
      "step": 162174
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.264319896697998,
      "learning_rate": 0.00012072266144270691,
      "loss": 2.1698,
      "step": 162175
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.0875030755996704,
      "learning_rate": 0.00012071938164253388,
      "loss": 1.778,
      "step": 162176
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.2493876218795776,
      "learning_rate": 0.00012071610187569241,
      "loss": 2.1632,
      "step": 162177
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.2298259735107422,
      "learning_rate": 0.0001207128221421834,
      "loss": 2.4116,
      "step": 162178
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.2138862609863281,
      "learning_rate": 0.00012070954244200706,
      "loss": 1.9581,
      "step": 162179
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.3234463930130005,
      "learning_rate": 0.00012070626277516429,
      "loss": 2.1034,
      "step": 162180
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.1565498113632202,
      "learning_rate": 0.00012070298314165549,
      "loss": 2.2105,
      "step": 162181
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.166155457496643,
      "learning_rate": 0.00012069970354148152,
      "loss": 1.9645,
      "step": 162182
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.184144139289856,
      "learning_rate": 0.00012069642397464276,
      "loss": 2.0385,
      "step": 162183
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.1284774541854858,
      "learning_rate": 0.00012069314444114002,
      "loss": 1.9002,
      "step": 162184
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.1874430179595947,
      "learning_rate": 0.00012068986494097379,
      "loss": 2.0755,
      "step": 162185
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.188518762588501,
      "learning_rate": 0.00012068658547414473,
      "loss": 2.2818,
      "step": 162186
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.141107439994812,
      "learning_rate": 0.00012068330604065329,
      "loss": 1.973,
      "step": 162187
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.2779110670089722,
      "learning_rate": 0.00012068002664050033,
      "loss": 2.0241,
      "step": 162188
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.3701894283294678,
      "learning_rate": 0.00012067674727368623,
      "loss": 2.0067,
      "step": 162189
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.2202335596084595,
      "learning_rate": 0.00012067346794021182,
      "loss": 1.9678,
      "step": 162190
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.2482210397720337,
      "learning_rate": 0.00012067018864007758,
      "loss": 2.0472,
      "step": 162191
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.2916209697723389,
      "learning_rate": 0.00012066690937328403,
      "loss": 2.1516,
      "step": 162192
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.5610344409942627,
      "learning_rate": 0.00012066363013983199,
      "loss": 2.0988,
      "step": 162193
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.2008826732635498,
      "learning_rate": 0.00012066035093972197,
      "loss": 2.2033,
      "step": 162194
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.2110588550567627,
      "learning_rate": 0.00012065707177295449,
      "loss": 2.239,
      "step": 162195
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.2333823442459106,
      "learning_rate": 0.00012065379263953032,
      "loss": 2.1838,
      "step": 162196
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.33053719997406,
      "learning_rate": 0.00012065051353945002,
      "loss": 2.1208,
      "step": 162197
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.6161181926727295,
      "learning_rate": 0.00012064723447271408,
      "loss": 2.1249,
      "step": 162198
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.3146438598632812,
      "learning_rate": 0.00012064395543932327,
      "loss": 2.1411,
      "step": 162199
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.1580820083618164,
      "learning_rate": 0.00012064067643927818,
      "loss": 1.9898,
      "step": 162200
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.2485461235046387,
      "learning_rate": 0.00012063739747257923,
      "loss": 2.1075,
      "step": 162201
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.281667947769165,
      "learning_rate": 0.00012063411853922729,
      "loss": 1.9861,
      "step": 162202
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.2099549770355225,
      "learning_rate": 0.00012063083963922287,
      "loss": 2.2772,
      "step": 162203
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.419761300086975,
      "learning_rate": 0.00012062756077256643,
      "loss": 1.828,
      "step": 162204
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.4353777170181274,
      "learning_rate": 0.00012062428193925882,
      "loss": 1.9509,
      "step": 162205
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.210101842880249,
      "learning_rate": 0.00012062100313930045,
      "loss": 1.9783,
      "step": 162206
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.6374542713165283,
      "learning_rate": 0.00012061772437269213,
      "loss": 1.9148,
      "step": 162207
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.2457730770111084,
      "learning_rate": 0.00012061444563943438,
      "loss": 1.9984,
      "step": 162208
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.2782710790634155,
      "learning_rate": 0.00012061116693952777,
      "loss": 2.2835,
      "step": 162209
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.178091049194336,
      "learning_rate": 0.00012060788827297281,
      "loss": 1.9674,
      "step": 162210
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.286177158355713,
      "learning_rate": 0.00012060460963977036,
      "loss": 2.2157,
      "step": 162211
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.3064169883728027,
      "learning_rate": 0.00012060133103992078,
      "loss": 2.3182,
      "step": 162212
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.3534184694290161,
      "learning_rate": 0.00012059805247342488,
      "loss": 2.0178,
      "step": 162213
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.296134114265442,
      "learning_rate": 0.00012059477394028312,
      "loss": 2.167,
      "step": 162214
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.1801499128341675,
      "learning_rate": 0.00012059149544049639,
      "loss": 1.9749,
      "step": 162215
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.7816665172576904,
      "learning_rate": 0.00012058821697406487,
      "loss": 1.977,
      "step": 162216
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.2561357021331787,
      "learning_rate": 0.00012058493854098948,
      "loss": 2.081,
      "step": 162217
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.203636646270752,
      "learning_rate": 0.00012058166014127062,
      "loss": 2.0136,
      "step": 162218
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.1578454971313477,
      "learning_rate": 0.00012057838177490914,
      "loss": 2.0959,
      "step": 162219
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.25641930103302,
      "learning_rate": 0.00012057510344190544,
      "loss": 2.1393,
      "step": 162220
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.1105241775512695,
      "learning_rate": 0.0001205718251422604,
      "loss": 1.9908,
      "step": 162221
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.321834683418274,
      "learning_rate": 0.00012056854687597422,
      "loss": 1.9827,
      "step": 162222
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.2060688734054565,
      "learning_rate": 0.00012056526864304788,
      "loss": 2.034,
      "step": 162223
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.166336178779602,
      "learning_rate": 0.0001205619904434817,
      "loss": 2.0597,
      "step": 162224
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.284890055656433,
      "learning_rate": 0.00012055871227727658,
      "loss": 2.0529,
      "step": 162225
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.3261677026748657,
      "learning_rate": 0.00012055543414443283,
      "loss": 1.9527,
      "step": 162226
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.1457419395446777,
      "learning_rate": 0.00012055215604495148,
      "loss": 1.9686,
      "step": 162227
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.1170815229415894,
      "learning_rate": 0.00012054887797883262,
      "loss": 1.8588,
      "step": 162228
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.1973602771759033,
      "learning_rate": 0.00012054559994607723,
      "loss": 1.9954,
      "step": 162229
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.1953076124191284,
      "learning_rate": 0.00012054232194668566,
      "loss": 2.1899,
      "step": 162230
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.1971585750579834,
      "learning_rate": 0.00012053904398065882,
      "loss": 2.0801,
      "step": 162231
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.3515772819519043,
      "learning_rate": 0.000120535766047997,
      "loss": 2.0015,
      "step": 162232
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.1481170654296875,
      "learning_rate": 0.0001205324881487012,
      "loss": 1.9784,
      "step": 162233
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.1892387866973877,
      "learning_rate": 0.00012052921028277162,
      "loss": 2.1338,
      "step": 162234
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.2894704341888428,
      "learning_rate": 0.00012052593245020912,
      "loss": 2.2674,
      "step": 162235
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.3030844926834106,
      "learning_rate": 0.00012052265465101415,
      "loss": 2.1962,
      "step": 162236
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.1665641069412231,
      "learning_rate": 0.00012051937688518749,
      "loss": 1.9907,
      "step": 162237
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.325128197669983,
      "learning_rate": 0.00012051609915272957,
      "loss": 2.0137,
      "step": 162238
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.2299553155899048,
      "learning_rate": 0.00012051282145364133,
      "loss": 2.2304,
      "step": 162239
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.273118257522583,
      "learning_rate": 0.00012050954378792289,
      "loss": 1.9344,
      "step": 162240
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.1429492235183716,
      "learning_rate": 0.00012050626615557526,
      "loss": 1.9352,
      "step": 162241
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.2582838535308838,
      "learning_rate": 0.00012050298855659877,
      "loss": 1.9036,
      "step": 162242
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.2250531911849976,
      "learning_rate": 0.00012049971099099428,
      "loss": 1.9538,
      "step": 162243
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.0594063997268677,
      "learning_rate": 0.00012049643345876216,
      "loss": 1.7829,
      "step": 162244
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.153228521347046,
      "learning_rate": 0.00012049315595990338,
      "loss": 1.9021,
      "step": 162245
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.2082760334014893,
      "learning_rate": 0.00012048987849441808,
      "loss": 2.1304,
      "step": 162246
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.2700884342193604,
      "learning_rate": 0.00012048660106230721,
      "loss": 2.1104,
      "step": 162247
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.2059575319290161,
      "learning_rate": 0.00012048332366357117,
      "loss": 2.3609,
      "step": 162248
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.2765825986862183,
      "learning_rate": 0.00012048004629821077,
      "loss": 1.9589,
      "step": 162249
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.2141075134277344,
      "learning_rate": 0.00012047676896622642,
      "loss": 2.083,
      "step": 162250
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.214163899421692,
      "learning_rate": 0.00012047349166761901,
      "loss": 2.1741,
      "step": 162251
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.197314977645874,
      "learning_rate": 0.00012047021440238879,
      "loss": 2.0476,
      "step": 162252
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.1902445554733276,
      "learning_rate": 0.00012046693717053663,
      "loss": 1.9164,
      "step": 162253
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.2199947834014893,
      "learning_rate": 0.00012046365997206296,
      "loss": 1.8481,
      "step": 162254
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.1681551933288574,
      "learning_rate": 0.0001204603828069686,
      "loss": 1.7063,
      "step": 162255
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.3895008563995361,
      "learning_rate": 0.00012045710567525391,
      "loss": 1.9724,
      "step": 162256
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.4820116758346558,
      "learning_rate": 0.00012045382857691988,
      "loss": 2.0979,
      "step": 162257
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.3328537940979004,
      "learning_rate": 0.00012045055151196664,
      "loss": 1.9518,
      "step": 162258
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.290877103805542,
      "learning_rate": 0.00012044727448039514,
      "loss": 1.9308,
      "step": 162259
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.3257921934127808,
      "learning_rate": 0.00012044399748220577,
      "loss": 2.2019,
      "step": 162260
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.2829675674438477,
      "learning_rate": 0.00012044072051739936,
      "loss": 2.0031,
      "step": 162261
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.2857462167739868,
      "learning_rate": 0.00012043744358597627,
      "loss": 1.954,
      "step": 162262
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.430902361869812,
      "learning_rate": 0.00012043416668793753,
      "loss": 2.1754,
      "step": 162263
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.349767804145813,
      "learning_rate": 0.00012043088982328321,
      "loss": 1.8045,
      "step": 162264
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.2737047672271729,
      "learning_rate": 0.00012042761299201432,
      "loss": 1.9618,
      "step": 162265
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.2984930276870728,
      "learning_rate": 0.00012042433619413118,
      "loss": 2.1432,
      "step": 162266
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.2209296226501465,
      "learning_rate": 0.00012042105942963467,
      "loss": 1.808,
      "step": 162267
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.1277474164962769,
      "learning_rate": 0.00012041778269852516,
      "loss": 2.1731,
      "step": 162268
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.263280987739563,
      "learning_rate": 0.00012041450600080346,
      "loss": 2.1715,
      "step": 162269
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.214980959892273,
      "learning_rate": 0.00012041122933647014,
      "loss": 2.2423,
      "step": 162270
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.1769859790802002,
      "learning_rate": 0.00012040795270552573,
      "loss": 1.9485,
      "step": 162271
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.1800886392593384,
      "learning_rate": 0.00012040467610797079,
      "loss": 1.8731,
      "step": 162272
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.1955442428588867,
      "learning_rate": 0.0001204013995438061,
      "loss": 1.8562,
      "step": 162273
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.2644206285476685,
      "learning_rate": 0.00012039812301303206,
      "loss": 1.9087,
      "step": 162274
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.2739567756652832,
      "learning_rate": 0.00012039484651564953,
      "loss": 2.0978,
      "step": 162275
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.2175133228302002,
      "learning_rate": 0.00012039157005165897,
      "loss": 1.9463,
      "step": 162276
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.227587342262268,
      "learning_rate": 0.00012038829362106094,
      "loss": 2.1283,
      "step": 162277
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.4346041679382324,
      "learning_rate": 0.00012038501722385619,
      "loss": 2.0048,
      "step": 162278
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.076621413230896,
      "learning_rate": 0.00012038174086004524,
      "loss": 1.9584,
      "step": 162279
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.2056794166564941,
      "learning_rate": 0.00012037846452962864,
      "loss": 2.172,
      "step": 162280
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.2470743656158447,
      "learning_rate": 0.00012037518823260715,
      "loss": 1.924,
      "step": 162281
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.1302396059036255,
      "learning_rate": 0.00012037191196898133,
      "loss": 1.8953,
      "step": 162282
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.3167920112609863,
      "learning_rate": 0.00012036863573875165,
      "loss": 1.687,
      "step": 162283
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.1921486854553223,
      "learning_rate": 0.00012036535954191896,
      "loss": 1.8129,
      "step": 162284
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.3351109027862549,
      "learning_rate": 0.00012036208337848372,
      "loss": 2.1025,
      "step": 162285
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.2283179759979248,
      "learning_rate": 0.00012035880724844644,
      "loss": 2.1218,
      "step": 162286
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.1901863813400269,
      "learning_rate": 0.00012035553115180796,
      "loss": 2.2025,
      "step": 162287
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.2771692276000977,
      "learning_rate": 0.00012035225508856877,
      "loss": 2.0261,
      "step": 162288
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.3387506008148193,
      "learning_rate": 0.0001203489790587294,
      "loss": 2.0112,
      "step": 162289
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.229074478149414,
      "learning_rate": 0.00012034570306229064,
      "loss": 2.1846,
      "step": 162290
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.222747802734375,
      "learning_rate": 0.0001203424270992529,
      "loss": 2.1931,
      "step": 162291
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.2076919078826904,
      "learning_rate": 0.00012033915116961701,
      "loss": 1.7615,
      "step": 162292
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.236246109008789,
      "learning_rate": 0.00012033587527338347,
      "loss": 2.1378,
      "step": 162293
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.1832791566848755,
      "learning_rate": 0.00012033259941055286,
      "loss": 1.9826,
      "step": 162294
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.203272819519043,
      "learning_rate": 0.00012032932358112571,
      "loss": 2.1974,
      "step": 162295
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.3141140937805176,
      "learning_rate": 0.00012032604778510287,
      "loss": 2.2751,
      "step": 162296
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.1824455261230469,
      "learning_rate": 0.00012032277202248464,
      "loss": 1.9843,
      "step": 162297
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.193520188331604,
      "learning_rate": 0.00012031949629327197,
      "loss": 2.0287,
      "step": 162298
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.1627817153930664,
      "learning_rate": 0.00012031622059746527,
      "loss": 2.1525,
      "step": 162299
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.2687373161315918,
      "learning_rate": 0.00012031294493506517,
      "loss": 2.1382,
      "step": 162300
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.1671531200408936,
      "learning_rate": 0.00012030966930607218,
      "loss": 2.1922,
      "step": 162301
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.2714093923568726,
      "learning_rate": 0.00012030639371048711,
      "loss": 2.0305,
      "step": 162302
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.1297781467437744,
      "learning_rate": 0.00012030311814831038,
      "loss": 2.0625,
      "step": 162303
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.217023253440857,
      "learning_rate": 0.00012029984261954279,
      "loss": 1.9969,
      "step": 162304
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.0938823223114014,
      "learning_rate": 0.00012029656712418479,
      "loss": 2.1365,
      "step": 162305
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.295852780342102,
      "learning_rate": 0.00012029329166223722,
      "loss": 2.1386,
      "step": 162306
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.1001969575881958,
      "learning_rate": 0.00012029001623370031,
      "loss": 1.9553,
      "step": 162307
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.2815114259719849,
      "learning_rate": 0.000120286740838575,
      "loss": 2.1307,
      "step": 162308
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.2526861429214478,
      "learning_rate": 0.00012028346547686165,
      "loss": 2.013,
      "step": 162309
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.3353006839752197,
      "learning_rate": 0.0001202801901485611,
      "loss": 1.8329,
      "step": 162310
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.2165963649749756,
      "learning_rate": 0.00012027691485367376,
      "loss": 2.0071,
      "step": 162311
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.1881928443908691,
      "learning_rate": 0.00012027363959220055,
      "loss": 1.9298,
      "step": 162312
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.2438760995864868,
      "learning_rate": 0.00012027036436414165,
      "loss": 2.0596,
      "step": 162313
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.2695202827453613,
      "learning_rate": 0.000120267089169498,
      "loss": 2.332,
      "step": 162314
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.1733559370040894,
      "learning_rate": 0.00012026381400826994,
      "loss": 2.0518,
      "step": 162315
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.2362486124038696,
      "learning_rate": 0.0001202605388804584,
      "loss": 2.1191,
      "step": 162316
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.3263025283813477,
      "learning_rate": 0.00012025726378606367,
      "loss": 1.9761,
      "step": 162317
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.2759759426116943,
      "learning_rate": 0.00012025398872508672,
      "loss": 1.9975,
      "step": 162318
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.147919774055481,
      "learning_rate": 0.00012025071369752775,
      "loss": 1.8273,
      "step": 162319
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.1565163135528564,
      "learning_rate": 0.00012024743870338767,
      "loss": 1.9291,
      "step": 162320
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.210719108581543,
      "learning_rate": 0.00012024416374266686,
      "loss": 2.1427,
      "step": 162321
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.2212895154953003,
      "learning_rate": 0.00012024088881536617,
      "loss": 2.1833,
      "step": 162322
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.2875001430511475,
      "learning_rate": 0.000120237613921486,
      "loss": 2.2875,
      "step": 162323
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.0939033031463623,
      "learning_rate": 0.00012023433906102725,
      "loss": 1.879,
      "step": 162324
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.1691436767578125,
      "learning_rate": 0.00012023106423399011,
      "loss": 2.1472,
      "step": 162325
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.268230676651001,
      "learning_rate": 0.00012022778944037552,
      "loss": 2.0399,
      "step": 162326
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.3077272176742554,
      "learning_rate": 0.00012022451468018387,
      "loss": 2.0469,
      "step": 162327
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.1523158550262451,
      "learning_rate": 0.00012022123995341598,
      "loss": 1.8739,
      "step": 162328
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.205236554145813,
      "learning_rate": 0.00012021796526007226,
      "loss": 2.1865,
      "step": 162329
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.6044769287109375,
      "learning_rate": 0.00012021469060015363,
      "loss": 2.0545,
      "step": 162330
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.2785271406173706,
      "learning_rate": 0.00012021141597366027,
      "loss": 1.8707,
      "step": 162331
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.3555431365966797,
      "learning_rate": 0.00012020814138059309,
      "loss": 2.0348,
      "step": 162332
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.2746652364730835,
      "learning_rate": 0.00012020486682095252,
      "loss": 2.0089,
      "step": 162333
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.636351227760315,
      "learning_rate": 0.00012020159229473937,
      "loss": 1.9891,
      "step": 162334
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.3051942586898804,
      "learning_rate": 0.00012019831780195401,
      "loss": 1.8393,
      "step": 162335
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.201653003692627,
      "learning_rate": 0.00012019504334259738,
      "loss": 1.8984,
      "step": 162336
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.2218286991119385,
      "learning_rate": 0.00012019176891666972,
      "loss": 2.1213,
      "step": 162337
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.2616850137710571,
      "learning_rate": 0.00012018849452417188,
      "loss": 1.9988,
      "step": 162338
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.2720471620559692,
      "learning_rate": 0.00012018522016510429,
      "loss": 2.181,
      "step": 162339
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.1829036474227905,
      "learning_rate": 0.00012018194583946776,
      "loss": 1.8889,
      "step": 162340
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.0912718772888184,
      "learning_rate": 0.00012017867154726269,
      "loss": 2.2138,
      "step": 162341
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.3216454982757568,
      "learning_rate": 0.00012017539728849005,
      "loss": 2.0979,
      "step": 162342
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.2066521644592285,
      "learning_rate": 0.00012017212306314996,
      "loss": 2.1142,
      "step": 162343
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.190807580947876,
      "learning_rate": 0.00012016884887124336,
      "loss": 1.9106,
      "step": 162344
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.1772598028182983,
      "learning_rate": 0.00012016557471277069,
      "loss": 2.0723,
      "step": 162345
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.1896106004714966,
      "learning_rate": 0.00012016230058773273,
      "loss": 2.2437,
      "step": 162346
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.1995357275009155,
      "learning_rate": 0.00012015902649612986,
      "loss": 2.1164,
      "step": 162347
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.3572548627853394,
      "learning_rate": 0.00012015575243796295,
      "loss": 2.0058,
      "step": 162348
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.259055256843567,
      "learning_rate": 0.00012015247841323249,
      "loss": 2.2154,
      "step": 162349
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.2448251247406006,
      "learning_rate": 0.00012014920442193904,
      "loss": 2.0379,
      "step": 162350
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.1858066320419312,
      "learning_rate": 0.00012014593046408319,
      "loss": 2.0229,
      "step": 162351
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.2454519271850586,
      "learning_rate": 0.0001201426565396657,
      "loss": 2.001,
      "step": 162352
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.1992777585983276,
      "learning_rate": 0.00012013938264868698,
      "loss": 2.1079,
      "step": 162353
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.1241499185562134,
      "learning_rate": 0.0001201361087911478,
      "loss": 1.7892,
      "step": 162354
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.1329340934753418,
      "learning_rate": 0.0001201328349670488,
      "loss": 1.9799,
      "step": 162355
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.179403305053711,
      "learning_rate": 0.00012012956117639043,
      "loss": 1.8995,
      "step": 162356
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.162527084350586,
      "learning_rate": 0.00012012628741917328,
      "loss": 1.8447,
      "step": 162357
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.4199166297912598,
      "learning_rate": 0.00012012301369539818,
      "loss": 2.1281,
      "step": 162358
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.2733609676361084,
      "learning_rate": 0.00012011974000506546,
      "loss": 2.017,
      "step": 162359
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.1875511407852173,
      "learning_rate": 0.00012011646634817603,
      "loss": 2.0219,
      "step": 162360
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.1316713094711304,
      "learning_rate": 0.0001201131927247303,
      "loss": 2.0277,
      "step": 162361
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.2019715309143066,
      "learning_rate": 0.00012010991913472898,
      "loss": 2.4689,
      "step": 162362
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.3179057836532593,
      "learning_rate": 0.00012010664557817248,
      "loss": 2.0909,
      "step": 162363
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.186659574508667,
      "learning_rate": 0.00012010337205506163,
      "loss": 2.1033,
      "step": 162364
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.1768990755081177,
      "learning_rate": 0.0001201000985653969,
      "loss": 2.0284,
      "step": 162365
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.1761215925216675,
      "learning_rate": 0.00012009682510917904,
      "loss": 2.1636,
      "step": 162366
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.1581887006759644,
      "learning_rate": 0.00012009355168640856,
      "loss": 2.1086,
      "step": 162367
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.1569370031356812,
      "learning_rate": 0.00012009027829708599,
      "loss": 1.9038,
      "step": 162368
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.089162826538086,
      "learning_rate": 0.00012008700494121217,
      "loss": 1.762,
      "step": 162369
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.1481982469558716,
      "learning_rate": 0.00012008373161878756,
      "loss": 1.9656,
      "step": 162370
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.1493186950683594,
      "learning_rate": 0.00012008045832981266,
      "loss": 2.1031,
      "step": 162371
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.2875081300735474,
      "learning_rate": 0.0001200771850742883,
      "loss": 2.0234,
      "step": 162372
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.4621397256851196,
      "learning_rate": 0.00012007391185221498,
      "loss": 2.0752,
      "step": 162373
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.2589287757873535,
      "learning_rate": 0.00012007063866359324,
      "loss": 2.0637,
      "step": 162374
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.143946886062622,
      "learning_rate": 0.00012006736550842386,
      "loss": 2.0943,
      "step": 162375
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.2379255294799805,
      "learning_rate": 0.00012006409238670723,
      "loss": 1.9271,
      "step": 162376
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.4386075735092163,
      "learning_rate": 0.0001200608192984442,
      "loss": 1.9804,
      "step": 162377
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.5460010766983032,
      "learning_rate": 0.00012005754624363526,
      "loss": 2.1067,
      "step": 162378
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.2140285968780518,
      "learning_rate": 0.00012005427322228105,
      "loss": 1.9966,
      "step": 162379
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.246273159980774,
      "learning_rate": 0.000120051000234382,
      "loss": 2.3774,
      "step": 162380
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.5428684949874878,
      "learning_rate": 0.00012004772727993897,
      "loss": 2.0739,
      "step": 162381
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.1930855512619019,
      "learning_rate": 0.00012004445435895238,
      "loss": 2.1017,
      "step": 162382
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.1456342935562134,
      "learning_rate": 0.00012004118147142302,
      "loss": 2.2314,
      "step": 162383
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.1564494371414185,
      "learning_rate": 0.00012003790861735139,
      "loss": 2.0175,
      "step": 162384
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.2052302360534668,
      "learning_rate": 0.00012003463579673814,
      "loss": 2.394,
      "step": 162385
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.2408698797225952,
      "learning_rate": 0.00012003136300958373,
      "loss": 2.0184,
      "step": 162386
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.2575143575668335,
      "learning_rate": 0.00012002809025588898,
      "loss": 2.0177,
      "step": 162387
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.229358434677124,
      "learning_rate": 0.00012002481753565429,
      "loss": 2.3438,
      "step": 162388
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.1501781940460205,
      "learning_rate": 0.00012002154484888051,
      "loss": 2.0651,
      "step": 162389
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.2109923362731934,
      "learning_rate": 0.00012001827219556803,
      "loss": 1.7929,
      "step": 162390
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.1683499813079834,
      "learning_rate": 0.00012001499957571775,
      "loss": 1.9952,
      "step": 162391
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.318008542060852,
      "learning_rate": 0.00012001172698932987,
      "loss": 2.0901,
      "step": 162392
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.3786996603012085,
      "learning_rate": 0.00012000845443640529,
      "loss": 2.1421,
      "step": 162393
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.2908713817596436,
      "learning_rate": 0.00012000518191694444,
      "loss": 2.1252,
      "step": 162394
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.2819271087646484,
      "learning_rate": 0.00012000190943094816,
      "loss": 2.1082,
      "step": 162395
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.250559687614441,
      "learning_rate": 0.00011999863697841676,
      "loss": 1.9419,
      "step": 162396
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.2933940887451172,
      "learning_rate": 0.00011999536455935127,
      "loss": 1.9736,
      "step": 162397
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.2849836349487305,
      "learning_rate": 0.00011999209217375181,
      "loss": 2.1198,
      "step": 162398
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.2211902141571045,
      "learning_rate": 0.00011998881982161934,
      "loss": 2.3631,
      "step": 162399
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.125756859779358,
      "learning_rate": 0.00011998554750295424,
      "loss": 1.9277,
      "step": 162400
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.202530026435852,
      "learning_rate": 0.00011998227521775729,
      "loss": 2.004,
      "step": 162401
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.2509407997131348,
      "learning_rate": 0.00011997900296602898,
      "loss": 2.0478,
      "step": 162402
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.1553136110305786,
      "learning_rate": 0.00011997573074777016,
      "loss": 2.0548,
      "step": 162403
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.2291285991668701,
      "learning_rate": 0.00011997245856298103,
      "loss": 1.9953,
      "step": 162404
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.2067129611968994,
      "learning_rate": 0.00011996918641166253,
      "loss": 2.2219,
      "step": 162405
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.2080602645874023,
      "learning_rate": 0.00011996591429381506,
      "loss": 2.1326,
      "step": 162406
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.2876795530319214,
      "learning_rate": 0.00011996264220943941,
      "loss": 2.0859,
      "step": 162407
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.1866930723190308,
      "learning_rate": 0.00011995937015853602,
      "loss": 2.1686,
      "step": 162408
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.3557722568511963,
      "learning_rate": 0.00011995609814110578,
      "loss": 1.8679,
      "step": 162409
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.319968342781067,
      "learning_rate": 0.00011995282615714891,
      "loss": 2.0105,
      "step": 162410
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.2272487878799438,
      "learning_rate": 0.00011994955420666629,
      "loss": 2.0863,
      "step": 162411
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.1508804559707642,
      "learning_rate": 0.00011994628228965833,
      "loss": 1.981,
      "step": 162412
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.2640334367752075,
      "learning_rate": 0.00011994301040612589,
      "loss": 1.7905,
      "step": 162413
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.2384568452835083,
      "learning_rate": 0.00011993973855606933,
      "loss": 2.3213,
      "step": 162414
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.1211365461349487,
      "learning_rate": 0.00011993646673948946,
      "loss": 1.9463,
      "step": 162415
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.1805768013000488,
      "learning_rate": 0.00011993319495638683,
      "loss": 2.0513,
      "step": 162416
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.2371731996536255,
      "learning_rate": 0.00011992992320676197,
      "loss": 2.0429,
      "step": 162417
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.1781704425811768,
      "learning_rate": 0.00011992665149061546,
      "loss": 2.1561,
      "step": 162418
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.2211437225341797,
      "learning_rate": 0.00011992337980794812,
      "loss": 2.0634,
      "step": 162419
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.1336263418197632,
      "learning_rate": 0.00011992010815876029,
      "loss": 1.9396,
      "step": 162420
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.4308528900146484,
      "learning_rate": 0.00011991683654305281,
      "loss": 2.3176,
      "step": 162421
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.1843030452728271,
      "learning_rate": 0.00011991356496082618,
      "loss": 2.0194,
      "step": 162422
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.3253363370895386,
      "learning_rate": 0.00011991029341208103,
      "loss": 2.0332,
      "step": 162423
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.2030149698257446,
      "learning_rate": 0.00011990702189681784,
      "loss": 1.7905,
      "step": 162424
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.1285996437072754,
      "learning_rate": 0.00011990375041503748,
      "loss": 2.187,
      "step": 162425
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.212049961090088,
      "learning_rate": 0.00011990047896674028,
      "loss": 2.0592,
      "step": 162426
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.355513334274292,
      "learning_rate": 0.00011989720755192707,
      "loss": 2.1717,
      "step": 162427
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.198695182800293,
      "learning_rate": 0.0001198939361705984,
      "loss": 2.0949,
      "step": 162428
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.507850170135498,
      "learning_rate": 0.00011989066482275482,
      "loss": 1.9044,
      "step": 162429
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.3116610050201416,
      "learning_rate": 0.00011988739350839686,
      "loss": 1.8196,
      "step": 162430
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.3390229940414429,
      "learning_rate": 0.00011988412222752538,
      "loss": 2.0609,
      "step": 162431
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.257741928100586,
      "learning_rate": 0.0001198808509801407,
      "loss": 2.1729,
      "step": 162432
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.274822473526001,
      "learning_rate": 0.00011987757976624367,
      "loss": 2.0575,
      "step": 162433
    },
    {
      "epoch": 2.11,
      "grad_norm": 1.458094596862793,
      "learning_rate": 0.00011987430858583482,
      "loss": 2.0553,
      "step": 162434
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.1433252096176147,
      "learning_rate": 0.00011987103743891473,
      "loss": 1.9423,
      "step": 162435
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.2091916799545288,
      "learning_rate": 0.00011986776632548389,
      "loss": 2.0938,
      "step": 162436
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.6130677461624146,
      "learning_rate": 0.00011986449524554315,
      "loss": 2.2665,
      "step": 162437
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.4005632400512695,
      "learning_rate": 0.00011986122419909291,
      "loss": 2.0742,
      "step": 162438
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.403923749923706,
      "learning_rate": 0.00011985795318613395,
      "loss": 2.1213,
      "step": 162439
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.256643295288086,
      "learning_rate": 0.0001198546822066668,
      "loss": 1.9801,
      "step": 162440
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.127568244934082,
      "learning_rate": 0.00011985141126069209,
      "loss": 1.9887,
      "step": 162441
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.4675929546356201,
      "learning_rate": 0.00011984814034821029,
      "loss": 1.8861,
      "step": 162442
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.5279210805892944,
      "learning_rate": 0.00011984486946922219,
      "loss": 2.2632,
      "step": 162443
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.333606243133545,
      "learning_rate": 0.00011984159862372827,
      "loss": 2.0138,
      "step": 162444
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.235802412033081,
      "learning_rate": 0.0001198383278117293,
      "loss": 2.1017,
      "step": 162445
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.1232246160507202,
      "learning_rate": 0.00011983505703322575,
      "loss": 2.0907,
      "step": 162446
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.3899812698364258,
      "learning_rate": 0.00011983178628821829,
      "loss": 1.8217,
      "step": 162447
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.3721686601638794,
      "learning_rate": 0.00011982851557670739,
      "loss": 2.0049,
      "step": 162448
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.4157336950302124,
      "learning_rate": 0.00011982524489869388,
      "loss": 2.0225,
      "step": 162449
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.1680604219436646,
      "learning_rate": 0.00011982197425417815,
      "loss": 2.0038,
      "step": 162450
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.1742733716964722,
      "learning_rate": 0.000119818703643161,
      "loss": 1.9258,
      "step": 162451
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.4296571016311646,
      "learning_rate": 0.000119815433065643,
      "loss": 2.1624,
      "step": 162452
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.3544725179672241,
      "learning_rate": 0.00011981216252162456,
      "loss": 2.0759,
      "step": 162453
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.223893404006958,
      "learning_rate": 0.00011980889201110658,
      "loss": 2.1751,
      "step": 162454
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.3523041009902954,
      "learning_rate": 0.00011980562153408952,
      "loss": 1.9945,
      "step": 162455
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.2471927404403687,
      "learning_rate": 0.00011980235109057388,
      "loss": 2.0114,
      "step": 162456
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.2789791822433472,
      "learning_rate": 0.00011979908068056051,
      "loss": 1.8979,
      "step": 162457
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.3836146593093872,
      "learning_rate": 0.00011979581030404988,
      "loss": 2.1825,
      "step": 162458
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.3157265186309814,
      "learning_rate": 0.00011979253996104248,
      "loss": 2.0384,
      "step": 162459
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.2341995239257812,
      "learning_rate": 0.00011978926965153918,
      "loss": 2.0232,
      "step": 162460
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.2710895538330078,
      "learning_rate": 0.00011978599937554047,
      "loss": 2.052,
      "step": 162461
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.156674861907959,
      "learning_rate": 0.00011978272913304681,
      "loss": 2.2768,
      "step": 162462
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.1976828575134277,
      "learning_rate": 0.00011977945892405907,
      "loss": 2.0171,
      "step": 162463
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.1969326734542847,
      "learning_rate": 0.00011977618874857773,
      "loss": 2.0018,
      "step": 162464
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.2142813205718994,
      "learning_rate": 0.00011977291860660329,
      "loss": 2.2357,
      "step": 162465
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.1264837980270386,
      "learning_rate": 0.00011976964849813654,
      "loss": 1.8866,
      "step": 162466
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.1240112781524658,
      "learning_rate": 0.00011976637842317796,
      "loss": 1.8941,
      "step": 162467
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.2279303073883057,
      "learning_rate": 0.00011976310838172827,
      "loss": 2.1785,
      "step": 162468
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.2314002513885498,
      "learning_rate": 0.00011975983837378804,
      "loss": 2.087,
      "step": 162469
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.230292797088623,
      "learning_rate": 0.00011975656839935788,
      "loss": 2.1011,
      "step": 162470
    },
    {
      "epoch": 2.12,
      "grad_norm": 2.457087755203247,
      "learning_rate": 0.00011975329845843824,
      "loss": 2.3092,
      "step": 162471
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.3916337490081787,
      "learning_rate": 0.00011975002855103,
      "loss": 1.886,
      "step": 162472
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.31077241897583,
      "learning_rate": 0.00011974675867713352,
      "loss": 2.1185,
      "step": 162473
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.315006136894226,
      "learning_rate": 0.00011974348883674958,
      "loss": 2.1919,
      "step": 162474
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.3018196821212769,
      "learning_rate": 0.0001197402190298787,
      "loss": 1.9863,
      "step": 162475
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.2842813730239868,
      "learning_rate": 0.00011973694925652165,
      "loss": 1.9534,
      "step": 162476
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.119321346282959,
      "learning_rate": 0.00011973367951667874,
      "loss": 1.924,
      "step": 162477
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.181926965713501,
      "learning_rate": 0.00011973040981035085,
      "loss": 1.8387,
      "step": 162478
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.3255674839019775,
      "learning_rate": 0.00011972714013753835,
      "loss": 2.1312,
      "step": 162479
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.2152576446533203,
      "learning_rate": 0.0001197238704982421,
      "loss": 2.2093,
      "step": 162480
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.2894214391708374,
      "learning_rate": 0.00011972060089246244,
      "loss": 1.9672,
      "step": 162481
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.204566240310669,
      "learning_rate": 0.00011971733132020029,
      "loss": 2.0502,
      "step": 162482
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.1701273918151855,
      "learning_rate": 0.00011971406178145608,
      "loss": 1.9408,
      "step": 162483
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.2514533996582031,
      "learning_rate": 0.00011971079227623038,
      "loss": 2.0223,
      "step": 162484
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.1655492782592773,
      "learning_rate": 0.00011970752280452378,
      "loss": 2.0072,
      "step": 162485
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.1524004936218262,
      "learning_rate": 0.00011970425336633707,
      "loss": 2.117,
      "step": 162486
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.1693315505981445,
      "learning_rate": 0.0001197009839616706,
      "loss": 2.0344,
      "step": 162487
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.1984913349151611,
      "learning_rate": 0.00011969771459052526,
      "loss": 2.0059,
      "step": 162488
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.2407195568084717,
      "learning_rate": 0.00011969444525290151,
      "loss": 2.109,
      "step": 162489
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.2846723794937134,
      "learning_rate": 0.00011969117594879994,
      "loss": 2.0384,
      "step": 162490
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.266751766204834,
      "learning_rate": 0.00011968790667822112,
      "loss": 2.0209,
      "step": 162491
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.1870542764663696,
      "learning_rate": 0.00011968463744116581,
      "loss": 1.824,
      "step": 162492
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.1527950763702393,
      "learning_rate": 0.00011968136823763439,
      "loss": 2.1353,
      "step": 162493
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.3075499534606934,
      "learning_rate": 0.00011967809906762775,
      "loss": 1.9442,
      "step": 162494
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.2220854759216309,
      "learning_rate": 0.00011967482993114637,
      "loss": 2.3781,
      "step": 162495
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.1514142751693726,
      "learning_rate": 0.0001196715608281908,
      "loss": 2.091,
      "step": 162496
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.3887220621109009,
      "learning_rate": 0.00011966829175876161,
      "loss": 1.9409,
      "step": 162497
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.2432771921157837,
      "learning_rate": 0.00011966502272285959,
      "loss": 2.1034,
      "step": 162498
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.2546603679656982,
      "learning_rate": 0.00011966175372048513,
      "loss": 2.0017,
      "step": 162499
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.2346597909927368,
      "learning_rate": 0.00011965848475163908,
      "loss": 1.9578,
      "step": 162500
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.1104390621185303,
      "learning_rate": 0.00011965521581632192,
      "loss": 2.1521,
      "step": 162501
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.3353886604309082,
      "learning_rate": 0.00011965194691453426,
      "loss": 2.2196,
      "step": 162502
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.251059889793396,
      "learning_rate": 0.00011964867804627659,
      "loss": 1.9808,
      "step": 162503
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.3243513107299805,
      "learning_rate": 0.00011964540921154975,
      "loss": 2.0794,
      "step": 162504
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.1316924095153809,
      "learning_rate": 0.00011964214041035412,
      "loss": 2.0413,
      "step": 162505
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.146370530128479,
      "learning_rate": 0.00011963887164269052,
      "loss": 2.0084,
      "step": 162506
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.1404166221618652,
      "learning_rate": 0.00011963560290855948,
      "loss": 2.0335,
      "step": 162507
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.306221842765808,
      "learning_rate": 0.0001196323342079616,
      "loss": 1.9851,
      "step": 162508
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.1662541627883911,
      "learning_rate": 0.00011962906554089731,
      "loss": 2.0932,
      "step": 162509
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.1407684087753296,
      "learning_rate": 0.00011962579690736754,
      "loss": 2.0572,
      "step": 162510
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.1735711097717285,
      "learning_rate": 0.0001196225283073726,
      "loss": 2.1426,
      "step": 162511
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.1141465902328491,
      "learning_rate": 0.00011961925974091334,
      "loss": 1.8459,
      "step": 162512
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.28107488155365,
      "learning_rate": 0.00011961599120799029,
      "loss": 2.0926,
      "step": 162513
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.3044450283050537,
      "learning_rate": 0.00011961272270860401,
      "loss": 2.1173,
      "step": 162514
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.179931879043579,
      "learning_rate": 0.00011960945424275504,
      "loss": 2.1401,
      "step": 162515
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.2984987497329712,
      "learning_rate": 0.00011960618581044417,
      "loss": 2.1391,
      "step": 162516
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.1941672563552856,
      "learning_rate": 0.00011960291741167186,
      "loss": 2.2366,
      "step": 162517
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.2067195177078247,
      "learning_rate": 0.00011959964904643883,
      "loss": 1.9349,
      "step": 162518
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.2343626022338867,
      "learning_rate": 0.00011959638071474562,
      "loss": 1.9629,
      "step": 162519
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.2751058340072632,
      "learning_rate": 0.00011959311241659286,
      "loss": 2.1027,
      "step": 162520
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.1250749826431274,
      "learning_rate": 0.00011958984415198105,
      "loss": 2.1391,
      "step": 162521
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.1806930303573608,
      "learning_rate": 0.00011958657592091101,
      "loss": 2.0985,
      "step": 162522
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.3593779802322388,
      "learning_rate": 0.00011958330772338311,
      "loss": 1.9564,
      "step": 162523
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.1072114706039429,
      "learning_rate": 0.00011958003955939816,
      "loss": 1.9933,
      "step": 162524
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.1833575963974,
      "learning_rate": 0.00011957677142895673,
      "loss": 1.8709,
      "step": 162525
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.20431387424469,
      "learning_rate": 0.00011957350333205936,
      "loss": 2.2624,
      "step": 162526
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.2837713956832886,
      "learning_rate": 0.0001195702352687066,
      "loss": 1.8828,
      "step": 162527
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.1615521907806396,
      "learning_rate": 0.00011956696723889923,
      "loss": 1.9453,
      "step": 162528
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.2843883037567139,
      "learning_rate": 0.00011956369924263766,
      "loss": 1.959,
      "step": 162529
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.2009788751602173,
      "learning_rate": 0.00011956043127992269,
      "loss": 2.1984,
      "step": 162530
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.194487452507019,
      "learning_rate": 0.00011955716335075485,
      "loss": 2.0057,
      "step": 162531
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.2702643871307373,
      "learning_rate": 0.00011955389545513478,
      "loss": 1.7737,
      "step": 162532
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.187474250793457,
      "learning_rate": 0.00011955062759306289,
      "loss": 2.0982,
      "step": 162533
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.3277342319488525,
      "learning_rate": 0.00011954735976454008,
      "loss": 2.2097,
      "step": 162534
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.4757009744644165,
      "learning_rate": 0.00011954409196956671,
      "loss": 1.9764,
      "step": 162535
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.159978985786438,
      "learning_rate": 0.00011954082420814359,
      "loss": 2.0101,
      "step": 162536
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.12193763256073,
      "learning_rate": 0.00011953755648027125,
      "loss": 1.9893,
      "step": 162537
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.1921075582504272,
      "learning_rate": 0.00011953428878595016,
      "loss": 1.9889,
      "step": 162538
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.2905219793319702,
      "learning_rate": 0.0001195310211251812,
      "loss": 1.9024,
      "step": 162539
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.3138617277145386,
      "learning_rate": 0.00011952775349796478,
      "loss": 2.044,
      "step": 162540
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.1101195812225342,
      "learning_rate": 0.0001195244859043015,
      "loss": 1.8704,
      "step": 162541
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.1143018007278442,
      "learning_rate": 0.00011952121834419212,
      "loss": 2.1813,
      "step": 162542
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.227685809135437,
      "learning_rate": 0.00011951795081763712,
      "loss": 2.033,
      "step": 162543
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.2958728075027466,
      "learning_rate": 0.00011951468332463705,
      "loss": 2.1419,
      "step": 162544
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.2391486167907715,
      "learning_rate": 0.00011951141586519273,
      "loss": 2.236,
      "step": 162545
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.3021653890609741,
      "learning_rate": 0.00011950814843930461,
      "loss": 1.9077,
      "step": 162546
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.297084093093872,
      "learning_rate": 0.00011950488104697325,
      "loss": 2.1233,
      "step": 162547
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.2930350303649902,
      "learning_rate": 0.00011950161368819948,
      "loss": 2.0329,
      "step": 162548
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.185874104499817,
      "learning_rate": 0.0001194983463629836,
      "loss": 1.9214,
      "step": 162549
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.2225334644317627,
      "learning_rate": 0.00011949507907132653,
      "loss": 1.9703,
      "step": 162550
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.230383038520813,
      "learning_rate": 0.00011949181181322873,
      "loss": 1.9476,
      "step": 162551
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.227847695350647,
      "learning_rate": 0.00011948854458869068,
      "loss": 2.0266,
      "step": 162552
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.1848435401916504,
      "learning_rate": 0.00011948527739771322,
      "loss": 1.9873,
      "step": 162553
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.222363829612732,
      "learning_rate": 0.00011948201024029688,
      "loss": 2.1511,
      "step": 162554
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.1318989992141724,
      "learning_rate": 0.00011947874311644212,
      "loss": 1.7181,
      "step": 162555
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.3065272569656372,
      "learning_rate": 0.00011947547602614979,
      "loss": 1.9836,
      "step": 162556
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.1820896863937378,
      "learning_rate": 0.00011947220896942036,
      "loss": 2.1918,
      "step": 162557
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.3695186376571655,
      "learning_rate": 0.00011946894194625437,
      "loss": 2.1204,
      "step": 162558
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.168823480606079,
      "learning_rate": 0.00011946567495665259,
      "loss": 1.9884,
      "step": 162559
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.6499437093734741,
      "learning_rate": 0.00011946240800061546,
      "loss": 2.0306,
      "step": 162560
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.1516941785812378,
      "learning_rate": 0.00011945914107814378,
      "loss": 1.9897,
      "step": 162561
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.1954843997955322,
      "learning_rate": 0.00011945587418923805,
      "loss": 2.2034,
      "step": 162562
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.301231861114502,
      "learning_rate": 0.00011945260733389888,
      "loss": 1.9314,
      "step": 162563
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.2296713590621948,
      "learning_rate": 0.00011944934051212678,
      "loss": 2.3531,
      "step": 162564
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.159380316734314,
      "learning_rate": 0.00011944607372392256,
      "loss": 2.118,
      "step": 162565
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.201134204864502,
      "learning_rate": 0.0001194428069692866,
      "loss": 2.1915,
      "step": 162566
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.1588562726974487,
      "learning_rate": 0.00011943954024821974,
      "loss": 2.1107,
      "step": 162567
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.3287863731384277,
      "learning_rate": 0.00011943627356072251,
      "loss": 2.1566,
      "step": 162568
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.2591948509216309,
      "learning_rate": 0.00011943300690679544,
      "loss": 2.034,
      "step": 162569
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.1454542875289917,
      "learning_rate": 0.00011942974028643913,
      "loss": 1.8941,
      "step": 162570
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.2375872135162354,
      "learning_rate": 0.0001194264736996543,
      "loss": 1.9656,
      "step": 162571
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.2876263856887817,
      "learning_rate": 0.00011942320714644141,
      "loss": 2.1023,
      "step": 162572
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.2563809156417847,
      "learning_rate": 0.00011941994062680126,
      "loss": 1.9545,
      "step": 162573
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.1712201833724976,
      "learning_rate": 0.00011941667414073433,
      "loss": 2.1257,
      "step": 162574
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.1962097883224487,
      "learning_rate": 0.00011941340768824128,
      "loss": 2.0441,
      "step": 162575
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.2961058616638184,
      "learning_rate": 0.00011941014126932257,
      "loss": 1.9166,
      "step": 162576
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.293835163116455,
      "learning_rate": 0.00011940687488397903,
      "loss": 2.0966,
      "step": 162577
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.1001520156860352,
      "learning_rate": 0.00011940360853221107,
      "loss": 2.1608,
      "step": 162578
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.1925801038742065,
      "learning_rate": 0.00011940034221401945,
      "loss": 2.206,
      "step": 162579
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.2261712551116943,
      "learning_rate": 0.00011939707592940477,
      "loss": 2.0204,
      "step": 162580
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.1932097673416138,
      "learning_rate": 0.00011939380967836754,
      "loss": 2.0278,
      "step": 162581
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.2822352647781372,
      "learning_rate": 0.00011939054346090831,
      "loss": 2.3956,
      "step": 162582
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.192415714263916,
      "learning_rate": 0.00011938727727702792,
      "loss": 2.1245,
      "step": 162583
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.2830727100372314,
      "learning_rate": 0.00011938401112672673,
      "loss": 2.2642,
      "step": 162584
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.1957850456237793,
      "learning_rate": 0.00011938074501000555,
      "loss": 1.9676,
      "step": 162585
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.2306339740753174,
      "learning_rate": 0.00011937747892686491,
      "loss": 1.8993,
      "step": 162586
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.1662688255310059,
      "learning_rate": 0.00011937421287730541,
      "loss": 2.0322,
      "step": 162587
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.286683201789856,
      "learning_rate": 0.00011937094686132751,
      "loss": 1.7712,
      "step": 162588
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.2155696153640747,
      "learning_rate": 0.00011936768087893212,
      "loss": 1.9825,
      "step": 162589
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.1875009536743164,
      "learning_rate": 0.00011936441493011955,
      "loss": 2.0226,
      "step": 162590
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.3001354932785034,
      "learning_rate": 0.00011936114901489066,
      "loss": 2.1461,
      "step": 162591
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.1567130088806152,
      "learning_rate": 0.00011935788313324595,
      "loss": 2.2473,
      "step": 162592
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.156282901763916,
      "learning_rate": 0.00011935461728518599,
      "loss": 2.0689,
      "step": 162593
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.217657208442688,
      "learning_rate": 0.00011935135147071134,
      "loss": 2.262,
      "step": 162594
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.3009748458862305,
      "learning_rate": 0.00011934808568982279,
      "loss": 2.0447,
      "step": 162595
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.2482093572616577,
      "learning_rate": 0.00011934481994252074,
      "loss": 2.0134,
      "step": 162596
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.2294223308563232,
      "learning_rate": 0.00011934155422880599,
      "loss": 2.1691,
      "step": 162597
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.1753215789794922,
      "learning_rate": 0.00011933828854867904,
      "loss": 2.2112,
      "step": 162598
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.4193873405456543,
      "learning_rate": 0.00011933502290214055,
      "loss": 1.9759,
      "step": 162599
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.2284280061721802,
      "learning_rate": 0.00011933175728919096,
      "loss": 2.1725,
      "step": 162600
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.2586294412612915,
      "learning_rate": 0.0001193284917098311,
      "loss": 1.979,
      "step": 162601
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.1917169094085693,
      "learning_rate": 0.00011932522616406141,
      "loss": 2.1375,
      "step": 162602
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.1482897996902466,
      "learning_rate": 0.00011932196065188266,
      "loss": 1.9401,
      "step": 162603
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.356123924255371,
      "learning_rate": 0.00011931869517329535,
      "loss": 2.2133,
      "step": 162604
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.3439019918441772,
      "learning_rate": 0.00011931542972830015,
      "loss": 2.1061,
      "step": 162605
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.2547515630722046,
      "learning_rate": 0.00011931216431689748,
      "loss": 2.0076,
      "step": 162606
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.3510850667953491,
      "learning_rate": 0.0001193088989390882,
      "loss": 2.1753,
      "step": 162607
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.246882677078247,
      "learning_rate": 0.0001193056335948727,
      "loss": 1.9848,
      "step": 162608
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.262860655784607,
      "learning_rate": 0.00011930236828425181,
      "loss": 2.1699,
      "step": 162609
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.153061032295227,
      "learning_rate": 0.00011929910300722592,
      "loss": 1.7568,
      "step": 162610
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.2260994911193848,
      "learning_rate": 0.00011929583776379595,
      "loss": 2.211,
      "step": 162611
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.277705192565918,
      "learning_rate": 0.00011929257255396207,
      "loss": 2.1121,
      "step": 162612
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.141351580619812,
      "learning_rate": 0.0001192893073777252,
      "loss": 2.0083,
      "step": 162613
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.3456289768218994,
      "learning_rate": 0.00011928604223508577,
      "loss": 2.0866,
      "step": 162614
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.0678521394729614,
      "learning_rate": 0.0001192827771260446,
      "loss": 1.8147,
      "step": 162615
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.1196231842041016,
      "learning_rate": 0.00011927951205060205,
      "loss": 1.9419,
      "step": 162616
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.2162073850631714,
      "learning_rate": 0.00011927624700875906,
      "loss": 2.1066,
      "step": 162617
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.2711244821548462,
      "learning_rate": 0.00011927298200051582,
      "loss": 2.0061,
      "step": 162618
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.2855974435806274,
      "learning_rate": 0.00011926971702587325,
      "loss": 1.9146,
      "step": 162619
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.4734702110290527,
      "learning_rate": 0.00011926645208483175,
      "loss": 1.9376,
      "step": 162620
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.2697956562042236,
      "learning_rate": 0.00011926318717739212,
      "loss": 1.9883,
      "step": 162621
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.5056294202804565,
      "learning_rate": 0.0001192599223035548,
      "loss": 2.1149,
      "step": 162622
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.2586219310760498,
      "learning_rate": 0.00011925665746332064,
      "loss": 2.2224,
      "step": 162623
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.1298911571502686,
      "learning_rate": 0.0001192533926566899,
      "loss": 2.2206,
      "step": 162624
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.4006778001785278,
      "learning_rate": 0.00011925012788366347,
      "loss": 2.3329,
      "step": 162625
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.3377039432525635,
      "learning_rate": 0.00011924686314424174,
      "loss": 2.2153,
      "step": 162626
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.3344767093658447,
      "learning_rate": 0.00011924359843842552,
      "loss": 2.2214,
      "step": 162627
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.1825518608093262,
      "learning_rate": 0.00011924033376621524,
      "loss": 2.1999,
      "step": 162628
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.2193986177444458,
      "learning_rate": 0.0001192370691276117,
      "loss": 2.051,
      "step": 162629
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.2330559492111206,
      "learning_rate": 0.0001192338045226154,
      "loss": 1.9591,
      "step": 162630
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.2938060760498047,
      "learning_rate": 0.00011923053995122694,
      "loss": 1.8201,
      "step": 162631
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.2593696117401123,
      "learning_rate": 0.00011922727541344684,
      "loss": 1.9789,
      "step": 162632
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.323598027229309,
      "learning_rate": 0.00011922401090927589,
      "loss": 1.9622,
      "step": 162633
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.208813190460205,
      "learning_rate": 0.00011922074643871452,
      "loss": 2.0456,
      "step": 162634
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.17935311794281,
      "learning_rate": 0.00011921748200176353,
      "loss": 1.9033,
      "step": 162635
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.3834532499313354,
      "learning_rate": 0.00011921421759842342,
      "loss": 2.162,
      "step": 162636
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.2778631448745728,
      "learning_rate": 0.0001192109532286947,
      "loss": 2.0462,
      "step": 162637
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.2103458642959595,
      "learning_rate": 0.00011920768889257818,
      "loss": 1.7729,
      "step": 162638
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.3248522281646729,
      "learning_rate": 0.00011920442459007434,
      "loss": 2.2291,
      "step": 162639
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.3057918548583984,
      "learning_rate": 0.00011920116032118372,
      "loss": 2.1653,
      "step": 162640
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.4796411991119385,
      "learning_rate": 0.00011919789608590714,
      "loss": 1.9257,
      "step": 162641
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.7179399728775024,
      "learning_rate": 0.00011919463188424507,
      "loss": 2.0292,
      "step": 162642
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.1747214794158936,
      "learning_rate": 0.00011919136771619802,
      "loss": 2.0713,
      "step": 162643
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.231839895248413,
      "learning_rate": 0.0001191881035817668,
      "loss": 2.0744,
      "step": 162644
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.1530349254608154,
      "learning_rate": 0.00011918483948095196,
      "loss": 2.0709,
      "step": 162645
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.2677220106124878,
      "learning_rate": 0.00011918157541375394,
      "loss": 2.0963,
      "step": 162646
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.2508141994476318,
      "learning_rate": 0.00011917831138017359,
      "loss": 2.0515,
      "step": 162647
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.1792515516281128,
      "learning_rate": 0.0001191750473802114,
      "loss": 1.8958,
      "step": 162648
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.2964818477630615,
      "learning_rate": 0.0001191717834138679,
      "loss": 2.2201,
      "step": 162649
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.1994600296020508,
      "learning_rate": 0.00011916851948114387,
      "loss": 2.0585,
      "step": 162650
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.1974297761917114,
      "learning_rate": 0.00011916525558203972,
      "loss": 2.1271,
      "step": 162651
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.2133779525756836,
      "learning_rate": 0.00011916199171655627,
      "loss": 1.9572,
      "step": 162652
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.170280933380127,
      "learning_rate": 0.00011915872788469403,
      "loss": 2.0132,
      "step": 162653
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.3356231451034546,
      "learning_rate": 0.00011915546408645356,
      "loss": 2.07,
      "step": 162654
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.3691565990447998,
      "learning_rate": 0.00011915220032183543,
      "loss": 2.1793,
      "step": 162655
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.1928231716156006,
      "learning_rate": 0.00011914893659084042,
      "loss": 2.2657,
      "step": 162656
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.4829676151275635,
      "learning_rate": 0.0001191456728934689,
      "loss": 2.1729,
      "step": 162657
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.1954416036605835,
      "learning_rate": 0.00011914240922972174,
      "loss": 2.0932,
      "step": 162658
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.3181612491607666,
      "learning_rate": 0.00011913914559959946,
      "loss": 2.2338,
      "step": 162659
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.2054579257965088,
      "learning_rate": 0.00011913588200310258,
      "loss": 2.1159,
      "step": 162660
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.4349340200424194,
      "learning_rate": 0.00011913261844023167,
      "loss": 2.0851,
      "step": 162661
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.0470103025436401,
      "learning_rate": 0.00011912935491098751,
      "loss": 2.0201,
      "step": 162662
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.1291611194610596,
      "learning_rate": 0.00011912609141537051,
      "loss": 1.7865,
      "step": 162663
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.2277745008468628,
      "learning_rate": 0.00011912282795338149,
      "loss": 2.087,
      "step": 162664
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.2345296144485474,
      "learning_rate": 0.00011911956452502098,
      "loss": 1.9239,
      "step": 162665
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.1684530973434448,
      "learning_rate": 0.00011911630113028954,
      "loss": 2.1941,
      "step": 162666
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.2044243812561035,
      "learning_rate": 0.00011911303776918766,
      "loss": 1.9641,
      "step": 162667
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.104026198387146,
      "learning_rate": 0.00011910977444171621,
      "loss": 1.7759,
      "step": 162668
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.1315476894378662,
      "learning_rate": 0.00011910651114787557,
      "loss": 2.0088,
      "step": 162669
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.2883837223052979,
      "learning_rate": 0.00011910324788766654,
      "loss": 1.7837,
      "step": 162670
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.14174222946167,
      "learning_rate": 0.00011909998466108962,
      "loss": 2.0092,
      "step": 162671
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.2189708948135376,
      "learning_rate": 0.00011909672146814547,
      "loss": 1.9762,
      "step": 162672
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.2866894006729126,
      "learning_rate": 0.0001190934583088345,
      "loss": 1.9266,
      "step": 162673
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.2045786380767822,
      "learning_rate": 0.0001190901951831576,
      "loss": 2.0465,
      "step": 162674
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.4813625812530518,
      "learning_rate": 0.00011908693209111512,
      "loss": 2.0777,
      "step": 162675
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.1991474628448486,
      "learning_rate": 0.00011908366903270792,
      "loss": 1.9933,
      "step": 162676
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.1724342107772827,
      "learning_rate": 0.00011908040600793637,
      "loss": 2.1739,
      "step": 162677
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.2399790287017822,
      "learning_rate": 0.00011907714301680142,
      "loss": 2.0058,
      "step": 162678
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.3344039916992188,
      "learning_rate": 0.00011907388005930321,
      "loss": 1.995,
      "step": 162679
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.39952552318573,
      "learning_rate": 0.00011907061713544268,
      "loss": 2.0992,
      "step": 162680
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.1914972066879272,
      "learning_rate": 0.00011906735424522024,
      "loss": 2.0802,
      "step": 162681
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.1785787343978882,
      "learning_rate": 0.00011906409138863671,
      "loss": 2.1126,
      "step": 162682
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.1832919120788574,
      "learning_rate": 0.00011906082856569249,
      "loss": 2.022,
      "step": 162683
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.2414146661758423,
      "learning_rate": 0.00011905756577638845,
      "loss": 1.6996,
      "step": 162684
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.1734271049499512,
      "learning_rate": 0.00011905430302072483,
      "loss": 1.8417,
      "step": 162685
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.192345142364502,
      "learning_rate": 0.00011905104029870255,
      "loss": 2.1095,
      "step": 162686
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.260527491569519,
      "learning_rate": 0.00011904777761032195,
      "loss": 2.055,
      "step": 162687
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.314613938331604,
      "learning_rate": 0.00011904451495558393,
      "loss": 2.0083,
      "step": 162688
    },
    {
      "epoch": 2.12,
      "grad_norm": 2.0391907691955566,
      "learning_rate": 0.00011904125233448885,
      "loss": 2.0759,
      "step": 162689
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.1887953281402588,
      "learning_rate": 0.00011903798974703762,
      "loss": 1.9335,
      "step": 162690
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.8112835884094238,
      "learning_rate": 0.0001190347271932304,
      "loss": 1.9706,
      "step": 162691
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.437064528465271,
      "learning_rate": 0.00011903146467306816,
      "loss": 2.0748,
      "step": 162692
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.212219476699829,
      "learning_rate": 0.00011902820218655129,
      "loss": 1.9707,
      "step": 162693
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.2803053855895996,
      "learning_rate": 0.0001190249397336806,
      "loss": 1.9828,
      "step": 162694
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.2161903381347656,
      "learning_rate": 0.0001190216773144565,
      "loss": 1.9639,
      "step": 162695
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.2573422193527222,
      "learning_rate": 0.00011901841492887987,
      "loss": 1.9516,
      "step": 162696
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.4350019693374634,
      "learning_rate": 0.00011901515257695091,
      "loss": 2.1309,
      "step": 162697
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.307445764541626,
      "learning_rate": 0.00011901189025867059,
      "loss": 2.0984,
      "step": 162698
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.2066510915756226,
      "learning_rate": 0.00011900862797403926,
      "loss": 2.0293,
      "step": 162699
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.2188471555709839,
      "learning_rate": 0.00011900536572305774,
      "loss": 2.2523,
      "step": 162700
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.348563551902771,
      "learning_rate": 0.00011900210350572645,
      "loss": 1.8593,
      "step": 162701
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.1306923627853394,
      "learning_rate": 0.00011899884132204629,
      "loss": 2.1526,
      "step": 162702
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.2268431186676025,
      "learning_rate": 0.00011899557917201746,
      "loss": 2.1272,
      "step": 162703
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.1931289434432983,
      "learning_rate": 0.00011899231705564083,
      "loss": 2.0088,
      "step": 162704
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.2745493650436401,
      "learning_rate": 0.00011898905497291688,
      "loss": 1.991,
      "step": 162705
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.2621428966522217,
      "learning_rate": 0.00011898579292384639,
      "loss": 2.1496,
      "step": 162706
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.308536410331726,
      "learning_rate": 0.00011898253090842973,
      "loss": 1.9294,
      "step": 162707
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.2805637121200562,
      "learning_rate": 0.00011897926892666787,
      "loss": 2.1224,
      "step": 162708
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.2228138446807861,
      "learning_rate": 0.00011897600697856094,
      "loss": 2.1106,
      "step": 162709
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.2251120805740356,
      "learning_rate": 0.0001189727450641099,
      "loss": 2.0506,
      "step": 162710
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.1105650663375854,
      "learning_rate": 0.00011896948318331516,
      "loss": 1.9883,
      "step": 162711
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.3411239385604858,
      "learning_rate": 0.00011896622133617751,
      "loss": 2.0183,
      "step": 162712
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.2191312313079834,
      "learning_rate": 0.00011896295952269732,
      "loss": 2.1158,
      "step": 162713
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.3208107948303223,
      "learning_rate": 0.00011895969774287548,
      "loss": 2.0654,
      "step": 162714
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.2026041746139526,
      "learning_rate": 0.00011895643599671242,
      "loss": 2.2618,
      "step": 162715
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.182509422302246,
      "learning_rate": 0.0001189531742842088,
      "loss": 2.0622,
      "step": 162716
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.33885657787323,
      "learning_rate": 0.00011894991260536509,
      "loss": 1.9048,
      "step": 162717
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.2456170320510864,
      "learning_rate": 0.00011894665096018212,
      "loss": 1.8348,
      "step": 162718
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.1689012050628662,
      "learning_rate": 0.00011894338934866025,
      "loss": 2.0547,
      "step": 162719
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.3291624784469604,
      "learning_rate": 0.00011894012777080035,
      "loss": 2.1777,
      "step": 162720
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.1815319061279297,
      "learning_rate": 0.0001189368662266029,
      "loss": 2.0947,
      "step": 162721
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.1726572513580322,
      "learning_rate": 0.00011893360471606839,
      "loss": 1.9321,
      "step": 162722
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.1590663194656372,
      "learning_rate": 0.00011893034323919765,
      "loss": 2.019,
      "step": 162723
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.2969069480895996,
      "learning_rate": 0.00011892708179599117,
      "loss": 2.1361,
      "step": 162724
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.228405237197876,
      "learning_rate": 0.00011892382038644947,
      "loss": 2.1986,
      "step": 162725
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.1286242008209229,
      "learning_rate": 0.00011892055901057335,
      "loss": 1.866,
      "step": 162726
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.334358811378479,
      "learning_rate": 0.00011891729766836332,
      "loss": 2.2135,
      "step": 162727
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.207794189453125,
      "learning_rate": 0.00011891403635981987,
      "loss": 1.964,
      "step": 162728
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.2803454399108887,
      "learning_rate": 0.00011891077508494382,
      "loss": 1.9216,
      "step": 162729
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.1871808767318726,
      "learning_rate": 0.00011890751384373569,
      "loss": 1.9625,
      "step": 162730
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.2132614850997925,
      "learning_rate": 0.00011890425263619594,
      "loss": 2.2544,
      "step": 162731
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.301195502281189,
      "learning_rate": 0.00011890099146232545,
      "loss": 1.944,
      "step": 162732
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.3065320253372192,
      "learning_rate": 0.00011889773032212467,
      "loss": 2.1212,
      "step": 162733
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.3192752599716187,
      "learning_rate": 0.00011889446921559409,
      "loss": 2.0591,
      "step": 162734
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.1153035163879395,
      "learning_rate": 0.00011889120814273457,
      "loss": 1.9835,
      "step": 162735
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.1670126914978027,
      "learning_rate": 0.00011888794710354652,
      "loss": 2.1933,
      "step": 162736
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.2526025772094727,
      "learning_rate": 0.00011888468609803067,
      "loss": 2.1167,
      "step": 162737
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.2743322849273682,
      "learning_rate": 0.00011888142512618762,
      "loss": 2.1469,
      "step": 162738
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.3789409399032593,
      "learning_rate": 0.00011887816418801793,
      "loss": 2.0002,
      "step": 162739
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.2716389894485474,
      "learning_rate": 0.00011887490328352206,
      "loss": 2.2866,
      "step": 162740
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.6208688020706177,
      "learning_rate": 0.0001188716424127009,
      "loss": 1.9347,
      "step": 162741
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.1511951684951782,
      "learning_rate": 0.00011886838157555483,
      "loss": 2.0258,
      "step": 162742
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.2253013849258423,
      "learning_rate": 0.00011886512077208462,
      "loss": 2.1104,
      "step": 162743
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.119120717048645,
      "learning_rate": 0.00011886186000229082,
      "loss": 1.9522,
      "step": 162744
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.3542282581329346,
      "learning_rate": 0.00011885859926617403,
      "loss": 2.2727,
      "step": 162745
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.1370913982391357,
      "learning_rate": 0.0001188553385637347,
      "loss": 1.9454,
      "step": 162746
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.2183058261871338,
      "learning_rate": 0.00011885207789497371,
      "loss": 2.0773,
      "step": 162747
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.313712239265442,
      "learning_rate": 0.00011884881725989143,
      "loss": 2.1905,
      "step": 162748
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.2318568229675293,
      "learning_rate": 0.00011884555665848869,
      "loss": 2.1788,
      "step": 162749
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.212969422340393,
      "learning_rate": 0.00011884229609076588,
      "loss": 2.0409,
      "step": 162750
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.1606365442276,
      "learning_rate": 0.00011883903555672391,
      "loss": 2.2182,
      "step": 162751
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.2519757747650146,
      "learning_rate": 0.00011883577505636298,
      "loss": 2.0721,
      "step": 162752
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.357399582862854,
      "learning_rate": 0.000118832514589684,
      "loss": 1.9953,
      "step": 162753
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.3146159648895264,
      "learning_rate": 0.00011882925415668738,
      "loss": 2.0732,
      "step": 162754
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.3043392896652222,
      "learning_rate": 0.00011882599375737388,
      "loss": 1.6951,
      "step": 162755
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.1517292261123657,
      "learning_rate": 0.000118822733391744,
      "loss": 2.1182,
      "step": 162756
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.1217678785324097,
      "learning_rate": 0.00011881947305979858,
      "loss": 1.976,
      "step": 162757
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.2496086359024048,
      "learning_rate": 0.00011881621276153782,
      "loss": 1.942,
      "step": 162758
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.14009690284729,
      "learning_rate": 0.00011881295249696268,
      "loss": 2.287,
      "step": 162759
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.303604245185852,
      "learning_rate": 0.00011880969226607349,
      "loss": 2.2944,
      "step": 162760
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.1606473922729492,
      "learning_rate": 0.00011880643206887112,
      "loss": 2.037,
      "step": 162761
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.2374337911605835,
      "learning_rate": 0.00011880317190535597,
      "loss": 2.0643,
      "step": 162762
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.2583900690078735,
      "learning_rate": 0.0001187999117755289,
      "loss": 1.9745,
      "step": 162763
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.1499650478363037,
      "learning_rate": 0.00011879665167939016,
      "loss": 1.8315,
      "step": 162764
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.2037882804870605,
      "learning_rate": 0.00011879339161694066,
      "loss": 2.1983,
      "step": 162765
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.1591123342514038,
      "learning_rate": 0.00011879013158818075,
      "loss": 1.887,
      "step": 162766
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.3298299312591553,
      "learning_rate": 0.00011878687159311132,
      "loss": 2.1312,
      "step": 162767
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.340002417564392,
      "learning_rate": 0.0001187836116317327,
      "loss": 2.0756,
      "step": 162768
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.1741937398910522,
      "learning_rate": 0.00011878035170404585,
      "loss": 1.9401,
      "step": 162769
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.1348540782928467,
      "learning_rate": 0.00011877709181005094,
      "loss": 2.0689,
      "step": 162770
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.404858946800232,
      "learning_rate": 0.00011877383194974887,
      "loss": 2.0141,
      "step": 162771
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.3281760215759277,
      "learning_rate": 0.0001187705721231401,
      "loss": 2.0426,
      "step": 162772
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.1897552013397217,
      "learning_rate": 0.0001187673123302254,
      "loss": 1.8905,
      "step": 162773
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.2384976148605347,
      "learning_rate": 0.00011876405257100514,
      "loss": 1.8902,
      "step": 162774
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.1399750709533691,
      "learning_rate": 0.0001187607928454803,
      "loss": 1.8378,
      "step": 162775
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.2485569715499878,
      "learning_rate": 0.00011875753315365104,
      "loss": 2.0002,
      "step": 162776
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.1372534036636353,
      "learning_rate": 0.00011875427349551829,
      "loss": 2.049,
      "step": 162777
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.2734464406967163,
      "learning_rate": 0.00011875101387108243,
      "loss": 2.0482,
      "step": 162778
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.2060338258743286,
      "learning_rate": 0.00011874775428034426,
      "loss": 1.7371,
      "step": 162779
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.2821706533432007,
      "learning_rate": 0.00011874449472330422,
      "loss": 1.9485,
      "step": 162780
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.168294072151184,
      "learning_rate": 0.00011874123519996322,
      "loss": 2.014,
      "step": 162781
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.2568328380584717,
      "learning_rate": 0.00011873797571032139,
      "loss": 2.1097,
      "step": 162782
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.159205436706543,
      "learning_rate": 0.00011873471625437973,
      "loss": 2.1767,
      "step": 162783
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.2165842056274414,
      "learning_rate": 0.00011873145683213859,
      "loss": 2.2583,
      "step": 162784
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.3801321983337402,
      "learning_rate": 0.00011872819744359883,
      "loss": 1.9574,
      "step": 162785
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.2231289148330688,
      "learning_rate": 0.0001187249380887608,
      "loss": 1.9022,
      "step": 162786
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.1435818672180176,
      "learning_rate": 0.00011872167876762542,
      "loss": 1.8635,
      "step": 162787
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.4191077947616577,
      "learning_rate": 0.0001187184194801929,
      "loss": 2.116,
      "step": 162788
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.1677618026733398,
      "learning_rate": 0.00011871516022646417,
      "loss": 2.1224,
      "step": 162789
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.3547272682189941,
      "learning_rate": 0.00011871190100643956,
      "loss": 2.2003,
      "step": 162790
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.2301439046859741,
      "learning_rate": 0.00011870864182011998,
      "loss": 1.926,
      "step": 162791
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.2276911735534668,
      "learning_rate": 0.00011870538266750579,
      "loss": 2.0396,
      "step": 162792
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.3357839584350586,
      "learning_rate": 0.00011870212354859787,
      "loss": 1.8901,
      "step": 162793
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.3491668701171875,
      "learning_rate": 0.00011869886446339646,
      "loss": 2.072,
      "step": 162794
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.2439848184585571,
      "learning_rate": 0.00011869560541190246,
      "loss": 2.1241,
      "step": 162795
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.160951018333435,
      "learning_rate": 0.00011869234639411621,
      "loss": 2.3183,
      "step": 162796
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.1192526817321777,
      "learning_rate": 0.00011868908741003864,
      "loss": 2.0232,
      "step": 162797
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.2584251165390015,
      "learning_rate": 0.00011868582845967007,
      "loss": 1.8696,
      "step": 162798
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.3636317253112793,
      "learning_rate": 0.0001186825695430113,
      "loss": 2.0693,
      "step": 162799
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.1345725059509277,
      "learning_rate": 0.00011867931066006291,
      "loss": 1.95,
      "step": 162800
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.338240146636963,
      "learning_rate": 0.00011867605181082545,
      "loss": 2.1257,
      "step": 162801
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.159058690071106,
      "learning_rate": 0.00011867279299529941,
      "loss": 2.0418,
      "step": 162802
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.2469100952148438,
      "learning_rate": 0.00011866953421348565,
      "loss": 2.0707,
      "step": 162803
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.1332299709320068,
      "learning_rate": 0.00011866627546538452,
      "loss": 2.0478,
      "step": 162804
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.2957388162612915,
      "learning_rate": 0.00011866301675099686,
      "loss": 2.0153,
      "step": 162805
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.2263429164886475,
      "learning_rate": 0.00011865975807032318,
      "loss": 2.1696,
      "step": 162806
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.0801239013671875,
      "learning_rate": 0.00011865649942336408,
      "loss": 2.1485,
      "step": 162807
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.3683658838272095,
      "learning_rate": 0.00011865324081012004,
      "loss": 2.2422,
      "step": 162808
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.161632776260376,
      "learning_rate": 0.00011864998223059187,
      "loss": 1.9227,
      "step": 162809
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.1393224000930786,
      "learning_rate": 0.00011864672368478002,
      "loss": 1.9288,
      "step": 162810
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.6642955541610718,
      "learning_rate": 0.00011864346517268527,
      "loss": 1.9779,
      "step": 162811
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.228818416595459,
      "learning_rate": 0.00011864020669430812,
      "loss": 2.3048,
      "step": 162812
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.1644326448440552,
      "learning_rate": 0.00011863694824964906,
      "loss": 2.0394,
      "step": 162813
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.852452278137207,
      "learning_rate": 0.00011863368983870893,
      "loss": 1.8433,
      "step": 162814
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.3430238962173462,
      "learning_rate": 0.00011863043146148822,
      "loss": 2.02,
      "step": 162815
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.3296549320220947,
      "learning_rate": 0.00011862717311798744,
      "loss": 2.0273,
      "step": 162816
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.1569440364837646,
      "learning_rate": 0.00011862391480820738,
      "loss": 1.9981,
      "step": 162817
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.1920229196548462,
      "learning_rate": 0.00011862065653214856,
      "loss": 2.1477,
      "step": 162818
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.2638801336288452,
      "learning_rate": 0.0001186173982898115,
      "loss": 1.9757,
      "step": 162819
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.1987849473953247,
      "learning_rate": 0.00011861414008119695,
      "loss": 1.8931,
      "step": 162820
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.3946430683135986,
      "learning_rate": 0.0001186108819063054,
      "loss": 2.1925,
      "step": 162821
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.3357864618301392,
      "learning_rate": 0.00011860762376513758,
      "loss": 2.0087,
      "step": 162822
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.3046777248382568,
      "learning_rate": 0.00011860436565769403,
      "loss": 2.1262,
      "step": 162823
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.2178289890289307,
      "learning_rate": 0.00011860110758397539,
      "loss": 2.0396,
      "step": 162824
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.322034239768982,
      "learning_rate": 0.0001185978495439821,
      "loss": 2.2123,
      "step": 162825
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.1834713220596313,
      "learning_rate": 0.000118594591537715,
      "loss": 2.1277,
      "step": 162826
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.1832672357559204,
      "learning_rate": 0.00011859133356517447,
      "loss": 2.2579,
      "step": 162827
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.181194543838501,
      "learning_rate": 0.00011858807562636134,
      "loss": 2.0515,
      "step": 162828
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.1609011888504028,
      "learning_rate": 0.00011858481772127614,
      "loss": 1.9235,
      "step": 162829
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.146477460861206,
      "learning_rate": 0.00011858155984991942,
      "loss": 2.086,
      "step": 162830
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.2455179691314697,
      "learning_rate": 0.00011857830201229171,
      "loss": 2.0312,
      "step": 162831
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.2327499389648438,
      "learning_rate": 0.00011857504420839383,
      "loss": 2.3888,
      "step": 162832
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.2197574377059937,
      "learning_rate": 0.0001185717864382262,
      "loss": 2.0444,
      "step": 162833
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.2403194904327393,
      "learning_rate": 0.00011856852870178955,
      "loss": 2.1175,
      "step": 162834
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.1049991846084595,
      "learning_rate": 0.00011856527099908437,
      "loss": 1.7418,
      "step": 162835
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.2441012859344482,
      "learning_rate": 0.00011856201333011152,
      "loss": 2.0024,
      "step": 162836
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.3382067680358887,
      "learning_rate": 0.00011855875569487122,
      "loss": 2.2776,
      "step": 162837
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.2439554929733276,
      "learning_rate": 0.00011855549809336437,
      "loss": 2.2109,
      "step": 162838
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.2403554916381836,
      "learning_rate": 0.00011855224052559139,
      "loss": 2.1658,
      "step": 162839
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.3538379669189453,
      "learning_rate": 0.00011854898299155303,
      "loss": 2.0354,
      "step": 162840
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.1759141683578491,
      "learning_rate": 0.00011854572549124977,
      "loss": 2.0772,
      "step": 162841
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.2273186445236206,
      "learning_rate": 0.00011854246802468248,
      "loss": 2.0476,
      "step": 162842
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.2225637435913086,
      "learning_rate": 0.00011853921059185137,
      "loss": 1.929,
      "step": 162843
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.4420255422592163,
      "learning_rate": 0.00011853595319275738,
      "loss": 1.8055,
      "step": 162844
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.1877182722091675,
      "learning_rate": 0.00011853269582740084,
      "loss": 2.139,
      "step": 162845
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.1570391654968262,
      "learning_rate": 0.0001185294384957826,
      "loss": 2.0078,
      "step": 162846
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.2424638271331787,
      "learning_rate": 0.00011852618119790309,
      "loss": 1.9595,
      "step": 162847
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.450253963470459,
      "learning_rate": 0.00011852292393376317,
      "loss": 2.1498,
      "step": 162848
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.3101272583007812,
      "learning_rate": 0.00011851966670336307,
      "loss": 2.3251,
      "step": 162849
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.2702020406723022,
      "learning_rate": 0.00011851640950670368,
      "loss": 2.116,
      "step": 162850
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.0856034755706787,
      "learning_rate": 0.0001185131523437854,
      "loss": 1.9385,
      "step": 162851
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.3932693004608154,
      "learning_rate": 0.0001185098952146091,
      "loss": 2.1986,
      "step": 162852
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.5140714645385742,
      "learning_rate": 0.00011850663811917507,
      "loss": 2.0583,
      "step": 162853
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.1461780071258545,
      "learning_rate": 0.00011850338105748435,
      "loss": 2.0735,
      "step": 162854
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.1221672296524048,
      "learning_rate": 0.00011850012402953704,
      "loss": 2.012,
      "step": 162855
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.174505352973938,
      "learning_rate": 0.0001184968670353341,
      "loss": 2.0056,
      "step": 162856
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.3257215023040771,
      "learning_rate": 0.0001184936100748759,
      "loss": 1.9858,
      "step": 162857
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.2893122434616089,
      "learning_rate": 0.00011849035314816327,
      "loss": 1.9893,
      "step": 162858
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.2389963865280151,
      "learning_rate": 0.0001184870962551966,
      "loss": 2.1958,
      "step": 162859
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.3228110074996948,
      "learning_rate": 0.00011848383939597681,
      "loss": 2.1863,
      "step": 162860
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.1535383462905884,
      "learning_rate": 0.00011848058257050413,
      "loss": 2.1676,
      "step": 162861
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.4756817817687988,
      "learning_rate": 0.0001184773257787794,
      "loss": 2.0027,
      "step": 162862
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.388622522354126,
      "learning_rate": 0.00011847406902080306,
      "loss": 2.1356,
      "step": 162863
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.2253493070602417,
      "learning_rate": 0.00011847081229657593,
      "loss": 1.9033,
      "step": 162864
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.2792222499847412,
      "learning_rate": 0.0001184675556060984,
      "loss": 2.1265,
      "step": 162865
    },
    {
      "epoch": 2.12,
      "grad_norm": 2.1181530952453613,
      "learning_rate": 0.00011846429894937141,
      "loss": 2.0078,
      "step": 162866
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.3536220788955688,
      "learning_rate": 0.00011846104232639509,
      "loss": 2.2358,
      "step": 162867
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.2839733362197876,
      "learning_rate": 0.0001184577857371704,
      "loss": 2.0376,
      "step": 162868
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.2622522115707397,
      "learning_rate": 0.00011845452918169776,
      "loss": 2.0939,
      "step": 162869
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.214003086090088,
      "learning_rate": 0.00011845127265997794,
      "loss": 2.2283,
      "step": 162870
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.5223796367645264,
      "learning_rate": 0.00011844801617201132,
      "loss": 2.0215,
      "step": 162871
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.4881926774978638,
      "learning_rate": 0.0001184447597177989,
      "loss": 2.0754,
      "step": 162872
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.2197680473327637,
      "learning_rate": 0.00011844150329734077,
      "loss": 2.1516,
      "step": 162873
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.1939529180526733,
      "learning_rate": 0.00011843824691063792,
      "loss": 1.9406,
      "step": 162874
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.2404085397720337,
      "learning_rate": 0.00011843499055769073,
      "loss": 2.1137,
      "step": 162875
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.1039247512817383,
      "learning_rate": 0.00011843173423849997,
      "loss": 1.8475,
      "step": 162876
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.2160073518753052,
      "learning_rate": 0.00011842847795306612,
      "loss": 1.8413,
      "step": 162877
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.2812918424606323,
      "learning_rate": 0.00011842522170138992,
      "loss": 1.9849,
      "step": 162878
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.2555516958236694,
      "learning_rate": 0.00011842196548347191,
      "loss": 2.0468,
      "step": 162879
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.1637235879898071,
      "learning_rate": 0.00011841870929931267,
      "loss": 2.1306,
      "step": 162880
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.1781089305877686,
      "learning_rate": 0.00011841545314891273,
      "loss": 1.9898,
      "step": 162881
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.420400857925415,
      "learning_rate": 0.00011841219703227288,
      "loss": 2.2048,
      "step": 162882
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.2349140644073486,
      "learning_rate": 0.00011840894094939352,
      "loss": 2.1785,
      "step": 162883
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.1089757680892944,
      "learning_rate": 0.0001184056849002755,
      "loss": 2.0549,
      "step": 162884
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.4882320165634155,
      "learning_rate": 0.00011840242888491926,
      "loss": 2.1116,
      "step": 162885
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.2098172903060913,
      "learning_rate": 0.00011839917290332546,
      "loss": 1.8703,
      "step": 162886
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.3186150789260864,
      "learning_rate": 0.00011839591695549457,
      "loss": 2.076,
      "step": 162887
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.1513254642486572,
      "learning_rate": 0.00011839266104142738,
      "loss": 2.0007,
      "step": 162888
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.2394819259643555,
      "learning_rate": 0.00011838940516112436,
      "loss": 2.2263,
      "step": 162889
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.193669319152832,
      "learning_rate": 0.00011838614931458628,
      "loss": 1.9171,
      "step": 162890
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.517969012260437,
      "learning_rate": 0.00011838289350181363,
      "loss": 2.0958,
      "step": 162891
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.1789493560791016,
      "learning_rate": 0.00011837963772280703,
      "loss": 1.8504,
      "step": 162892
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.2293123006820679,
      "learning_rate": 0.00011837638197756696,
      "loss": 1.9758,
      "step": 162893
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.2771354913711548,
      "learning_rate": 0.00011837312626609427,
      "loss": 2.0951,
      "step": 162894
    },
    {
      "epoch": 2.12,
      "grad_norm": 2.2934632301330566,
      "learning_rate": 0.00011836987058838935,
      "loss": 1.9463,
      "step": 162895
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.148592472076416,
      "learning_rate": 0.00011836661494445297,
      "loss": 1.8677,
      "step": 162896
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.2089779376983643,
      "learning_rate": 0.00011836335933428572,
      "loss": 1.8702,
      "step": 162897
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.1591519117355347,
      "learning_rate": 0.000118360103757888,
      "loss": 2.0052,
      "step": 162898
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.2111351490020752,
      "learning_rate": 0.00011835684821526067,
      "loss": 2.1644,
      "step": 162899
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.2692463397979736,
      "learning_rate": 0.00011835359270640426,
      "loss": 2.2111,
      "step": 162900
    },
    {
      "epoch": 2.12,
      "grad_norm": 2.0202584266662598,
      "learning_rate": 0.00011835033723131925,
      "loss": 2.0233,
      "step": 162901
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.1958595514297485,
      "learning_rate": 0.00011834708179000639,
      "loss": 2.1112,
      "step": 162902
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.2626497745513916,
      "learning_rate": 0.00011834382638246628,
      "loss": 2.0932,
      "step": 162903
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.17851722240448,
      "learning_rate": 0.00011834057100869937,
      "loss": 2.0977,
      "step": 162904
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.2160321474075317,
      "learning_rate": 0.00011833731566870647,
      "loss": 2.0517,
      "step": 162905
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.1178195476531982,
      "learning_rate": 0.00011833406036248812,
      "loss": 2.0851,
      "step": 162906
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.1862181425094604,
      "learning_rate": 0.00011833080509004478,
      "loss": 1.9709,
      "step": 162907
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.2652148008346558,
      "learning_rate": 0.00011832754985137726,
      "loss": 2.1286,
      "step": 162908
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.2346253395080566,
      "learning_rate": 0.00011832429464648608,
      "loss": 2.1158,
      "step": 162909
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.2513729333877563,
      "learning_rate": 0.00011832103947537176,
      "loss": 1.9402,
      "step": 162910
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.186043381690979,
      "learning_rate": 0.00011831778433803508,
      "loss": 2.1447,
      "step": 162911
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.2912076711654663,
      "learning_rate": 0.00011831452923447645,
      "loss": 2.0898,
      "step": 162912
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.232061743736267,
      "learning_rate": 0.00011831127416469667,
      "loss": 2.1503,
      "step": 162913
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.8718633651733398,
      "learning_rate": 0.00011830801912869626,
      "loss": 1.8451,
      "step": 162914
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.156887412071228,
      "learning_rate": 0.00011830476412647584,
      "loss": 1.7858,
      "step": 162915
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.2260866165161133,
      "learning_rate": 0.00011830150915803586,
      "loss": 2.1328,
      "step": 162916
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.396469235420227,
      "learning_rate": 0.00011829825422337718,
      "loss": 2.0354,
      "step": 162917
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.1796059608459473,
      "learning_rate": 0.00011829499932250016,
      "loss": 2.0429,
      "step": 162918
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.2577581405639648,
      "learning_rate": 0.00011829174445540568,
      "loss": 2.0083,
      "step": 162919
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.1881264448165894,
      "learning_rate": 0.00011828848962209404,
      "loss": 1.8267,
      "step": 162920
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.3308870792388916,
      "learning_rate": 0.00011828523482256624,
      "loss": 2.0377,
      "step": 162921
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.2715288400650024,
      "learning_rate": 0.00011828198005682243,
      "loss": 2.0655,
      "step": 162922
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.3774381875991821,
      "learning_rate": 0.00011827872532486355,
      "loss": 1.988,
      "step": 162923
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.2308293581008911,
      "learning_rate": 0.00011827547062668994,
      "loss": 2.0202,
      "step": 162924
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.3550441265106201,
      "learning_rate": 0.00011827221596230248,
      "loss": 2.0958,
      "step": 162925
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.4554011821746826,
      "learning_rate": 0.00011826896133170151,
      "loss": 2.2156,
      "step": 162926
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.274478793144226,
      "learning_rate": 0.00011826570673488803,
      "loss": 1.9731,
      "step": 162927
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.249493956565857,
      "learning_rate": 0.00011826245217186216,
      "loss": 2.0869,
      "step": 162928
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.2531002759933472,
      "learning_rate": 0.00011825919764262481,
      "loss": 2.0694,
      "step": 162929
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.4804257154464722,
      "learning_rate": 0.00011825594314717645,
      "loss": 2.1475,
      "step": 162930
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.2752574682235718,
      "learning_rate": 0.00011825268868551781,
      "loss": 2.1131,
      "step": 162931
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.2328100204467773,
      "learning_rate": 0.00011824943425764932,
      "loss": 1.9803,
      "step": 162932
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.3355473279953003,
      "learning_rate": 0.0001182461798635719,
      "loss": 2.0488,
      "step": 162933
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.212963581085205,
      "learning_rate": 0.00011824292550328574,
      "loss": 1.9852,
      "step": 162934
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.5737988948822021,
      "learning_rate": 0.00011823967117679176,
      "loss": 2.2164,
      "step": 162935
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.2237755060195923,
      "learning_rate": 0.00011823641688409034,
      "loss": 2.0539,
      "step": 162936
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.2773181200027466,
      "learning_rate": 0.00011823316262518233,
      "loss": 1.8308,
      "step": 162937
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.2124390602111816,
      "learning_rate": 0.00011822990840006808,
      "loss": 2.1596,
      "step": 162938
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.349526286125183,
      "learning_rate": 0.00011822665420874854,
      "loss": 1.8836,
      "step": 162939
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.2494293451309204,
      "learning_rate": 0.00011822340005122389,
      "loss": 2.0398,
      "step": 162940
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.2561888694763184,
      "learning_rate": 0.00011822014592749505,
      "loss": 1.895,
      "step": 162941
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.3225678205490112,
      "learning_rate": 0.0001182168918375624,
      "loss": 2.0964,
      "step": 162942
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.1270278692245483,
      "learning_rate": 0.00011821363778142678,
      "loss": 2.0081,
      "step": 162943
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.2396247386932373,
      "learning_rate": 0.00011821038375908856,
      "loss": 2.1512,
      "step": 162944
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.3675625324249268,
      "learning_rate": 0.00011820712977054854,
      "loss": 2.0414,
      "step": 162945
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.1468946933746338,
      "learning_rate": 0.0001182038758158073,
      "loss": 2.2566,
      "step": 162946
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.2658392190933228,
      "learning_rate": 0.00011820062189486537,
      "loss": 2.0038,
      "step": 162947
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.1975098848342896,
      "learning_rate": 0.00011819736800772323,
      "loss": 2.082,
      "step": 162948
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.2582701444625854,
      "learning_rate": 0.0001181941141543818,
      "loss": 1.968,
      "step": 162949
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.2055513858795166,
      "learning_rate": 0.00011819086033484136,
      "loss": 1.8095,
      "step": 162950
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.154242992401123,
      "learning_rate": 0.00011818760654910283,
      "loss": 2.0566,
      "step": 162951
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.401584267616272,
      "learning_rate": 0.00011818435279716661,
      "loss": 1.9326,
      "step": 162952
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.230750322341919,
      "learning_rate": 0.00011818109907903337,
      "loss": 2.1026,
      "step": 162953
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.4356091022491455,
      "learning_rate": 0.00011817784539470359,
      "loss": 2.0691,
      "step": 162954
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.1285851001739502,
      "learning_rate": 0.00011817459174417808,
      "loss": 2.0935,
      "step": 162955
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.1983824968338013,
      "learning_rate": 0.00011817133812745724,
      "loss": 2.0481,
      "step": 162956
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.3858169317245483,
      "learning_rate": 0.00011816808454454188,
      "loss": 1.9029,
      "step": 162957
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.2343556880950928,
      "learning_rate": 0.00011816483099543253,
      "loss": 2.1361,
      "step": 162958
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.3053792715072632,
      "learning_rate": 0.00011816157748012975,
      "loss": 2.0112,
      "step": 162959
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.1687012910842896,
      "learning_rate": 0.00011815832399863404,
      "loss": 2.0646,
      "step": 162960
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.19275963306427,
      "learning_rate": 0.00011815507055094628,
      "loss": 1.9611,
      "step": 162961
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.2581113576889038,
      "learning_rate": 0.00011815181713706678,
      "loss": 1.872,
      "step": 162962
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.2730010747909546,
      "learning_rate": 0.00011814856375699644,
      "loss": 1.938,
      "step": 162963
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.2951977252960205,
      "learning_rate": 0.00011814531041073568,
      "loss": 2.1464,
      "step": 162964
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.121374487876892,
      "learning_rate": 0.00011814205709828515,
      "loss": 2.0244,
      "step": 162965
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.2197065353393555,
      "learning_rate": 0.00011813880381964534,
      "loss": 2.0106,
      "step": 162966
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.1816829442977905,
      "learning_rate": 0.00011813555057481704,
      "loss": 2.026,
      "step": 162967
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.3347798585891724,
      "learning_rate": 0.0001181322973638007,
      "loss": 2.21,
      "step": 162968
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.3292278051376343,
      "learning_rate": 0.00011812904418659708,
      "loss": 2.1236,
      "step": 162969
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.3660321235656738,
      "learning_rate": 0.00011812579104320674,
      "loss": 2.1613,
      "step": 162970
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.3859872817993164,
      "learning_rate": 0.00011812253793363022,
      "loss": 2.066,
      "step": 162971
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.1843390464782715,
      "learning_rate": 0.00011811928485786806,
      "loss": 2.149,
      "step": 162972
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.2750064134597778,
      "learning_rate": 0.00011811603181592104,
      "loss": 1.8449,
      "step": 162973
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.253193974494934,
      "learning_rate": 0.0001181127788077896,
      "loss": 2.0387,
      "step": 162974
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.2407808303833008,
      "learning_rate": 0.00011810952583347453,
      "loss": 1.9448,
      "step": 162975
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.2277867794036865,
      "learning_rate": 0.00011810627289297633,
      "loss": 1.9664,
      "step": 162976
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.1963642835617065,
      "learning_rate": 0.00011810301998629561,
      "loss": 2.0384,
      "step": 162977
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.269246220588684,
      "learning_rate": 0.00011809976711343287,
      "loss": 1.8999,
      "step": 162978
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.178013563156128,
      "learning_rate": 0.00011809651427438892,
      "loss": 1.8562,
      "step": 162979
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.2277286052703857,
      "learning_rate": 0.00011809326146916415,
      "loss": 2.2223,
      "step": 162980
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.1782344579696655,
      "learning_rate": 0.00011809000869775938,
      "loss": 2.0286,
      "step": 162981
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.3900730609893799,
      "learning_rate": 0.00011808675596017514,
      "loss": 2.1452,
      "step": 162982
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.2335766553878784,
      "learning_rate": 0.00011808350325641187,
      "loss": 2.0522,
      "step": 162983
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.319056749343872,
      "learning_rate": 0.00011808025058647042,
      "loss": 2.1425,
      "step": 162984
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.2406543493270874,
      "learning_rate": 0.0001180769979503513,
      "loss": 1.9505,
      "step": 162985
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.4727144241333008,
      "learning_rate": 0.00011807374534805497,
      "loss": 2.1677,
      "step": 162986
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.2799972295761108,
      "learning_rate": 0.00011807049277958228,
      "loss": 2.5377,
      "step": 162987
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.2478910684585571,
      "learning_rate": 0.00011806724024493374,
      "loss": 2.1836,
      "step": 162988
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.1593959331512451,
      "learning_rate": 0.0001180639877441098,
      "loss": 1.9113,
      "step": 162989
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.210205316543579,
      "learning_rate": 0.00011806073527711134,
      "loss": 2.1941,
      "step": 162990
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.2717022895812988,
      "learning_rate": 0.0001180574828439388,
      "loss": 1.9523,
      "step": 162991
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.2094013690948486,
      "learning_rate": 0.00011805423044459271,
      "loss": 1.8684,
      "step": 162992
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.381160855293274,
      "learning_rate": 0.00011805097807907388,
      "loss": 2.2115,
      "step": 162993
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.232149600982666,
      "learning_rate": 0.00011804772574738278,
      "loss": 1.9175,
      "step": 162994
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.2231485843658447,
      "learning_rate": 0.00011804447344951996,
      "loss": 1.9808,
      "step": 162995
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.2881886959075928,
      "learning_rate": 0.00011804122118548622,
      "loss": 2.061,
      "step": 162996
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.2545931339263916,
      "learning_rate": 0.00011803796895528193,
      "loss": 2.2893,
      "step": 162997
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.2795642614364624,
      "learning_rate": 0.00011803471675890794,
      "loss": 1.8754,
      "step": 162998
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.1697689294815063,
      "learning_rate": 0.0001180314645963647,
      "loss": 1.859,
      "step": 162999
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.2905999422073364,
      "learning_rate": 0.00011802821246765288,
      "loss": 2.2367,
      "step": 163000
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.1686030626296997,
      "learning_rate": 0.00011802496037277292,
      "loss": 2.0634,
      "step": 163001
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.3076903820037842,
      "learning_rate": 0.00011802170831172567,
      "loss": 1.9868,
      "step": 163002
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.2953022718429565,
      "learning_rate": 0.00011801845628451151,
      "loss": 2.1116,
      "step": 163003
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.290204405784607,
      "learning_rate": 0.00011801520429113125,
      "loss": 2.009,
      "step": 163004
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.348690390586853,
      "learning_rate": 0.00011801195233158531,
      "loss": 2.0377,
      "step": 163005
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.2350717782974243,
      "learning_rate": 0.0001180087004058746,
      "loss": 2.1014,
      "step": 163006
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.1858956813812256,
      "learning_rate": 0.00011800544851399929,
      "loss": 2.0636,
      "step": 163007
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.3191256523132324,
      "learning_rate": 0.00011800219665596031,
      "loss": 2.0891,
      "step": 163008
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.2667371034622192,
      "learning_rate": 0.00011799894483175803,
      "loss": 2.2884,
      "step": 163009
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.3191108703613281,
      "learning_rate": 0.00011799569304139333,
      "loss": 2.1332,
      "step": 163010
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.233725905418396,
      "learning_rate": 0.00011799244128486651,
      "loss": 2.1264,
      "step": 163011
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.239607572555542,
      "learning_rate": 0.00011798918956217847,
      "loss": 1.9253,
      "step": 163012
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.1605428457260132,
      "learning_rate": 0.00011798593787332969,
      "loss": 2.1952,
      "step": 163013
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.2384272813796997,
      "learning_rate": 0.00011798268621832073,
      "loss": 2.1941,
      "step": 163014
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.186612606048584,
      "learning_rate": 0.0001179794345971521,
      "loss": 2.1421,
      "step": 163015
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.2127634286880493,
      "learning_rate": 0.00011797618300982468,
      "loss": 2.1325,
      "step": 163016
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.4193843603134155,
      "learning_rate": 0.00011797293145633879,
      "loss": 1.9949,
      "step": 163017
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.1172583103179932,
      "learning_rate": 0.0001179696799366953,
      "loss": 1.9772,
      "step": 163018
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.117217779159546,
      "learning_rate": 0.00011796642845089465,
      "loss": 1.8964,
      "step": 163019
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.3699733018875122,
      "learning_rate": 0.00011796317699893752,
      "loss": 1.9708,
      "step": 163020
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.3074127435684204,
      "learning_rate": 0.00011795992558082433,
      "loss": 2.0658,
      "step": 163021
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.2705844640731812,
      "learning_rate": 0.00011795667419655594,
      "loss": 2.0584,
      "step": 163022
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.3979660272598267,
      "learning_rate": 0.00011795342284613272,
      "loss": 2.0141,
      "step": 163023
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.1336781978607178,
      "learning_rate": 0.00011795017152955551,
      "loss": 2.1026,
      "step": 163024
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.2383027076721191,
      "learning_rate": 0.0001179469202468248,
      "loss": 2.1516,
      "step": 163025
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.1594489812850952,
      "learning_rate": 0.00011794366899794122,
      "loss": 2.1916,
      "step": 163026
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.3059589862823486,
      "learning_rate": 0.00011794041778290522,
      "loss": 1.9199,
      "step": 163027
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.193699836730957,
      "learning_rate": 0.00011793716660171763,
      "loss": 2.0751,
      "step": 163028
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.3815789222717285,
      "learning_rate": 0.00011793391545437889,
      "loss": 2.1199,
      "step": 163029
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.205137014389038,
      "learning_rate": 0.00011793066434088972,
      "loss": 2.0128,
      "step": 163030
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.3855340480804443,
      "learning_rate": 0.00011792741326125073,
      "loss": 1.9858,
      "step": 163031
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.1957224607467651,
      "learning_rate": 0.00011792416221546245,
      "loss": 1.9806,
      "step": 163032
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.151390790939331,
      "learning_rate": 0.00011792091120352539,
      "loss": 1.9838,
      "step": 163033
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.2448915243148804,
      "learning_rate": 0.00011791766022544036,
      "loss": 1.9235,
      "step": 163034
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.3385835886001587,
      "learning_rate": 0.00011791440928120781,
      "loss": 2.1446,
      "step": 163035
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.1511176824569702,
      "learning_rate": 0.00011791115837082849,
      "loss": 1.9722,
      "step": 163036
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.1998159885406494,
      "learning_rate": 0.00011790790749430291,
      "loss": 2.1611,
      "step": 163037
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.2135814428329468,
      "learning_rate": 0.00011790465665163173,
      "loss": 2.0422,
      "step": 163038
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.3287558555603027,
      "learning_rate": 0.00011790140584281539,
      "loss": 1.907,
      "step": 163039
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.2349576950073242,
      "learning_rate": 0.0001178981550678547,
      "loss": 2.0306,
      "step": 163040
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.235352873802185,
      "learning_rate": 0.00011789490432675007,
      "loss": 2.1456,
      "step": 163041
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.1115391254425049,
      "learning_rate": 0.00011789165361950235,
      "loss": 2.1886,
      "step": 163042
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.2002863883972168,
      "learning_rate": 0.00011788840294611199,
      "loss": 2.1193,
      "step": 163043
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.1457010507583618,
      "learning_rate": 0.00011788515230657961,
      "loss": 1.9505,
      "step": 163044
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.1541244983673096,
      "learning_rate": 0.00011788190170090575,
      "loss": 1.8936,
      "step": 163045
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.2667542695999146,
      "learning_rate": 0.00011787865112909117,
      "loss": 2.1446,
      "step": 163046
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.2061351537704468,
      "learning_rate": 0.00011787540059113626,
      "loss": 1.9619,
      "step": 163047
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.2753421068191528,
      "learning_rate": 0.00011787215008704187,
      "loss": 2.0291,
      "step": 163048
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.1987396478652954,
      "learning_rate": 0.00011786889961680846,
      "loss": 1.8983,
      "step": 163049
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.2152525186538696,
      "learning_rate": 0.0001178656491804367,
      "loss": 2.1012,
      "step": 163050
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.1688312292099,
      "learning_rate": 0.00011786239877792703,
      "loss": 2.0666,
      "step": 163051
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.2022548913955688,
      "learning_rate": 0.00011785914840928029,
      "loss": 2.3251,
      "step": 163052
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.138031244277954,
      "learning_rate": 0.00011785589807449684,
      "loss": 2.0662,
      "step": 163053
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.1163707971572876,
      "learning_rate": 0.00011785264777357756,
      "loss": 2.0594,
      "step": 163054
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.2043912410736084,
      "learning_rate": 0.00011784939750652289,
      "loss": 2.3078,
      "step": 163055
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.233702540397644,
      "learning_rate": 0.00011784614727333347,
      "loss": 2.1498,
      "step": 163056
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.3639901876449585,
      "learning_rate": 0.00011784289707400976,
      "loss": 2.0293,
      "step": 163057
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.2001203298568726,
      "learning_rate": 0.00011783964690855263,
      "loss": 1.9619,
      "step": 163058
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.1749929189682007,
      "learning_rate": 0.00011783639677696243,
      "loss": 1.9234,
      "step": 163059
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.1910507678985596,
      "learning_rate": 0.00011783314667923997,
      "loss": 1.8967,
      "step": 163060
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.2060920000076294,
      "learning_rate": 0.00011782989661538576,
      "loss": 2.151,
      "step": 163061
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.173726201057434,
      "learning_rate": 0.00011782664658540046,
      "loss": 1.9636,
      "step": 163062
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.2178388833999634,
      "learning_rate": 0.00011782339658928447,
      "loss": 2.207,
      "step": 163063
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.1714603900909424,
      "learning_rate": 0.00011782014662703868,
      "loss": 2.1725,
      "step": 163064
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.3969683647155762,
      "learning_rate": 0.00011781689669866347,
      "loss": 1.9255,
      "step": 163065
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.2661141157150269,
      "learning_rate": 0.00011781364680415962,
      "loss": 2.033,
      "step": 163066
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.2200685739517212,
      "learning_rate": 0.00011781039694352765,
      "loss": 2.0445,
      "step": 163067
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.3089797496795654,
      "learning_rate": 0.00011780714711676818,
      "loss": 2.0769,
      "step": 163068
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.177325963973999,
      "learning_rate": 0.00011780389732388169,
      "loss": 1.9293,
      "step": 163069
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.2810856103897095,
      "learning_rate": 0.000117800647564869,
      "loss": 2.2435,
      "step": 163070
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.265441656112671,
      "learning_rate": 0.00011779739783973052,
      "loss": 2.2047,
      "step": 163071
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.2212285995483398,
      "learning_rate": 0.00011779414814846702,
      "loss": 2.1945,
      "step": 163072
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.2111681699752808,
      "learning_rate": 0.00011779089849107905,
      "loss": 2.0976,
      "step": 163073
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.2252249717712402,
      "learning_rate": 0.00011778764886756706,
      "loss": 2.0433,
      "step": 163074
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.2002307176589966,
      "learning_rate": 0.00011778439927793192,
      "loss": 1.8871,
      "step": 163075
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.265616774559021,
      "learning_rate": 0.0001177811497221741,
      "loss": 1.9779,
      "step": 163076
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.2482229471206665,
      "learning_rate": 0.00011777790020029409,
      "loss": 2.0134,
      "step": 163077
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.3720035552978516,
      "learning_rate": 0.00011777465071229271,
      "loss": 2.1592,
      "step": 163078
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.202693223953247,
      "learning_rate": 0.00011777140125817033,
      "loss": 2.0875,
      "step": 163079
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.279783010482788,
      "learning_rate": 0.00011776815183792783,
      "loss": 2.0137,
      "step": 163080
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.090251088142395,
      "learning_rate": 0.00011776490245156569,
      "loss": 1.9822,
      "step": 163081
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.1886721849441528,
      "learning_rate": 0.00011776165309908433,
      "loss": 2.0204,
      "step": 163082
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.295192003250122,
      "learning_rate": 0.00011775840378048464,
      "loss": 2.0195,
      "step": 163083
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.131201148033142,
      "learning_rate": 0.00011775515449576712,
      "loss": 2.0822,
      "step": 163084
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.3289663791656494,
      "learning_rate": 0.00011775190524493227,
      "loss": 2.1462,
      "step": 163085
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.2220802307128906,
      "learning_rate": 0.00011774865602798084,
      "loss": 1.991,
      "step": 163086
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.2394179105758667,
      "learning_rate": 0.00011774540684491341,
      "loss": 1.9592,
      "step": 163087
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.3539575338363647,
      "learning_rate": 0.00011774215769573043,
      "loss": 1.9996,
      "step": 163088
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.2102328538894653,
      "learning_rate": 0.00011773890858043272,
      "loss": 2.0141,
      "step": 163089
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.2155580520629883,
      "learning_rate": 0.00011773565949902081,
      "loss": 1.9205,
      "step": 163090
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.1132595539093018,
      "learning_rate": 0.00011773241045149516,
      "loss": 1.9546,
      "step": 163091
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.1553877592086792,
      "learning_rate": 0.00011772916143785658,
      "loss": 2.1813,
      "step": 163092
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.4408271312713623,
      "learning_rate": 0.00011772591245810566,
      "loss": 2.0437,
      "step": 163093
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.311929702758789,
      "learning_rate": 0.00011772266351224277,
      "loss": 1.9459,
      "step": 163094
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.2631150484085083,
      "learning_rate": 0.00011771941460026879,
      "loss": 2.1303,
      "step": 163095
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.2164735794067383,
      "learning_rate": 0.0001177161657221841,
      "loss": 1.9418,
      "step": 163096
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.3029876947402954,
      "learning_rate": 0.00011771291687798952,
      "loss": 2.127,
      "step": 163097
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.2708889245986938,
      "learning_rate": 0.00011770966806768558,
      "loss": 2.2733,
      "step": 163098
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.2617305517196655,
      "learning_rate": 0.00011770641929127282,
      "loss": 2.1287,
      "step": 163099
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.1974481344223022,
      "learning_rate": 0.00011770317054875182,
      "loss": 1.8868,
      "step": 163100
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.210722804069519,
      "learning_rate": 0.00011769992184012332,
      "loss": 2.1145,
      "step": 163101
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.1540356874465942,
      "learning_rate": 0.00011769667316538772,
      "loss": 2.1322,
      "step": 163102
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.1982640027999878,
      "learning_rate": 0.00011769342452454586,
      "loss": 1.8849,
      "step": 163103
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.2551565170288086,
      "learning_rate": 0.00011769017591759824,
      "loss": 2.03,
      "step": 163104
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.2207571268081665,
      "learning_rate": 0.00011768692734454548,
      "loss": 2.0978,
      "step": 163105
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.1276607513427734,
      "learning_rate": 0.00011768367880538804,
      "loss": 2.0838,
      "step": 163106
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.3181402683258057,
      "learning_rate": 0.00011768043030012675,
      "loss": 1.9852,
      "step": 163107
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.4191826581954956,
      "learning_rate": 0.00011767718182876201,
      "loss": 2.143,
      "step": 163108
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.1162004470825195,
      "learning_rate": 0.00011767393339129459,
      "loss": 1.8081,
      "step": 163109
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.2389098405838013,
      "learning_rate": 0.0001176706849877251,
      "loss": 2.2245,
      "step": 163110
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.253185749053955,
      "learning_rate": 0.00011766743661805402,
      "loss": 2.015,
      "step": 163111
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.3502845764160156,
      "learning_rate": 0.00011766418828228191,
      "loss": 2.0189,
      "step": 163112
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.2464553117752075,
      "learning_rate": 0.00011766093998040956,
      "loss": 2.1557,
      "step": 163113
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.445258617401123,
      "learning_rate": 0.0001176576917124374,
      "loss": 2.1034,
      "step": 163114
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.2642879486083984,
      "learning_rate": 0.00011765444347836621,
      "loss": 2.014,
      "step": 163115
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.2350670099258423,
      "learning_rate": 0.0001176511952781965,
      "loss": 2.2186,
      "step": 163116
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.1650768518447876,
      "learning_rate": 0.00011764794711192889,
      "loss": 2.0298,
      "step": 163117
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.1481413841247559,
      "learning_rate": 0.00011764469897956387,
      "loss": 2.0532,
      "step": 163118
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.3397467136383057,
      "learning_rate": 0.00011764145088110222,
      "loss": 1.9142,
      "step": 163119
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.2935279607772827,
      "learning_rate": 0.00011763820281654436,
      "loss": 2.049,
      "step": 163120
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.2547733783721924,
      "learning_rate": 0.00011763495478589113,
      "loss": 2.0822,
      "step": 163121
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.2606326341629028,
      "learning_rate": 0.00011763170678914301,
      "loss": 1.9476,
      "step": 163122
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.1643362045288086,
      "learning_rate": 0.00011762845882630057,
      "loss": 2.1357,
      "step": 163123
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.2265914678573608,
      "learning_rate": 0.00011762521089736436,
      "loss": 2.0171,
      "step": 163124
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.2147146463394165,
      "learning_rate": 0.00011762196300233517,
      "loss": 1.8699,
      "step": 163125
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.3021718263626099,
      "learning_rate": 0.00011761871514121336,
      "loss": 1.9954,
      "step": 163126
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.2446051836013794,
      "learning_rate": 0.00011761546731399983,
      "loss": 2.0813,
      "step": 163127
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.2143222093582153,
      "learning_rate": 0.000117612219520695,
      "loss": 2.052,
      "step": 163128
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.0824074745178223,
      "learning_rate": 0.00011760897176129951,
      "loss": 2.0337,
      "step": 163129
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.1882898807525635,
      "learning_rate": 0.00011760572403581383,
      "loss": 1.95,
      "step": 163130
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.1550588607788086,
      "learning_rate": 0.0001176024763442388,
      "loss": 2.1574,
      "step": 163131
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.2020944356918335,
      "learning_rate": 0.00011759922868657483,
      "loss": 2.104,
      "step": 163132
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.234549641609192,
      "learning_rate": 0.00011759598106282267,
      "loss": 2.007,
      "step": 163133
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.313208818435669,
      "learning_rate": 0.00011759273347298289,
      "loss": 2.1545,
      "step": 163134
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.2746471166610718,
      "learning_rate": 0.00011758948591705608,
      "loss": 2.1176,
      "step": 163135
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.2926580905914307,
      "learning_rate": 0.00011758623839504267,
      "loss": 2.0447,
      "step": 163136
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.2358478307724,
      "learning_rate": 0.00011758299090694358,
      "loss": 2.2001,
      "step": 163137
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.3843368291854858,
      "learning_rate": 0.00011757974345275915,
      "loss": 2.2014,
      "step": 163138
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.244232416152954,
      "learning_rate": 0.00011757649603249015,
      "loss": 2.0606,
      "step": 163139
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.353674054145813,
      "learning_rate": 0.00011757324864613716,
      "loss": 2.0613,
      "step": 163140
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.2396022081375122,
      "learning_rate": 0.00011757000129370072,
      "loss": 1.9802,
      "step": 163141
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.3348355293273926,
      "learning_rate": 0.00011756675397518137,
      "loss": 1.9889,
      "step": 163142
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.2841284275054932,
      "learning_rate": 0.00011756350669057991,
      "loss": 2.1443,
      "step": 163143
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.3419959545135498,
      "learning_rate": 0.00011756025943989674,
      "loss": 2.0009,
      "step": 163144
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.2271583080291748,
      "learning_rate": 0.00011755701222313265,
      "loss": 2.0488,
      "step": 163145
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.2254374027252197,
      "learning_rate": 0.00011755376504028807,
      "loss": 1.806,
      "step": 163146
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.2682257890701294,
      "learning_rate": 0.00011755051789136387,
      "loss": 2.2238,
      "step": 163147
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.2660889625549316,
      "learning_rate": 0.00011754727077636028,
      "loss": 2.0722,
      "step": 163148
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.2675586938858032,
      "learning_rate": 0.0001175440236952782,
      "loss": 1.9515,
      "step": 163149
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.1716423034667969,
      "learning_rate": 0.00011754077664811801,
      "loss": 1.9177,
      "step": 163150
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.734958291053772,
      "learning_rate": 0.00011753752963488057,
      "loss": 2.0308,
      "step": 163151
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.1668835878372192,
      "learning_rate": 0.00011753428265556625,
      "loss": 2.3695,
      "step": 163152
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.285285234451294,
      "learning_rate": 0.00011753103571017593,
      "loss": 2.0235,
      "step": 163153
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.3239606618881226,
      "learning_rate": 0.00011752778879870984,
      "loss": 2.0608,
      "step": 163154
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.229231357574463,
      "learning_rate": 0.00011752454192116886,
      "loss": 2.3778,
      "step": 163155
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.1658774614334106,
      "learning_rate": 0.00011752129507755343,
      "loss": 2.0465,
      "step": 163156
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.1098018884658813,
      "learning_rate": 0.00011751804826786432,
      "loss": 2.1451,
      "step": 163157
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.2212556600570679,
      "learning_rate": 0.00011751480149210199,
      "loss": 2.1437,
      "step": 163158
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.1781971454620361,
      "learning_rate": 0.00011751155475026717,
      "loss": 2.1871,
      "step": 163159
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.2058017253875732,
      "learning_rate": 0.00011750830804236038,
      "loss": 2.1578,
      "step": 163160
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.1422524452209473,
      "learning_rate": 0.00011750506136838228,
      "loss": 1.8161,
      "step": 163161
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.1614296436309814,
      "learning_rate": 0.00011750181472833331,
      "loss": 1.9316,
      "step": 163162
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.2748432159423828,
      "learning_rate": 0.0001174985681222143,
      "loss": 2.0261,
      "step": 163163
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.4440405368804932,
      "learning_rate": 0.00011749532155002565,
      "loss": 2.1399,
      "step": 163164
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.1902174949645996,
      "learning_rate": 0.00011749207501176818,
      "loss": 2.2269,
      "step": 163165
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.275221586227417,
      "learning_rate": 0.00011748882850744238,
      "loss": 2.0997,
      "step": 163166
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.0989376306533813,
      "learning_rate": 0.00011748558203704873,
      "loss": 1.903,
      "step": 163167
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.29860520362854,
      "learning_rate": 0.00011748233560058808,
      "loss": 1.9903,
      "step": 163168
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.2090539932250977,
      "learning_rate": 0.00011747908919806088,
      "loss": 1.977,
      "step": 163169
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.193284273147583,
      "learning_rate": 0.0001174758428294677,
      "loss": 1.945,
      "step": 163170
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.312934160232544,
      "learning_rate": 0.00011747259649480926,
      "loss": 1.96,
      "step": 163171
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.2597399950027466,
      "learning_rate": 0.00011746935019408613,
      "loss": 2.2342,
      "step": 163172
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.296841025352478,
      "learning_rate": 0.00011746610392729882,
      "loss": 2.0476,
      "step": 163173
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.6550686359405518,
      "learning_rate": 0.00011746285769444805,
      "loss": 1.8838,
      "step": 163174
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.4918497800827026,
      "learning_rate": 0.00011745961149553444,
      "loss": 1.9197,
      "step": 163175
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.2479383945465088,
      "learning_rate": 0.00011745636533055839,
      "loss": 2.2478,
      "step": 163176
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.2497363090515137,
      "learning_rate": 0.00011745311919952078,
      "loss": 1.886,
      "step": 163177
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.273037075996399,
      "learning_rate": 0.00011744987310242206,
      "loss": 1.9441,
      "step": 163178
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.1868220567703247,
      "learning_rate": 0.00011744662703926276,
      "loss": 1.9595,
      "step": 163179
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.1743850708007812,
      "learning_rate": 0.00011744338101004369,
      "loss": 2.0257,
      "step": 163180
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.2175966501235962,
      "learning_rate": 0.00011744013501476521,
      "loss": 2.1221,
      "step": 163181
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.350752830505371,
      "learning_rate": 0.0001174368890534282,
      "loss": 1.9007,
      "step": 163182
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.3485751152038574,
      "learning_rate": 0.0001174336431260331,
      "loss": 2.047,
      "step": 163183
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.2143139839172363,
      "learning_rate": 0.00011743039723258051,
      "loss": 1.96,
      "step": 163184
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.204833984375,
      "learning_rate": 0.00011742715137307098,
      "loss": 1.9957,
      "step": 163185
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.323154091835022,
      "learning_rate": 0.00011742390554750528,
      "loss": 2.2717,
      "step": 163186
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.3228638172149658,
      "learning_rate": 0.0001174206597558838,
      "loss": 2.2678,
      "step": 163187
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.6111105680465698,
      "learning_rate": 0.00011741741399820741,
      "loss": 2.2564,
      "step": 163188
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.26348876953125,
      "learning_rate": 0.00011741416827447653,
      "loss": 2.1289,
      "step": 163189
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.1910682916641235,
      "learning_rate": 0.00011741092258469182,
      "loss": 1.8518,
      "step": 163190
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.227537751197815,
      "learning_rate": 0.00011740767692885376,
      "loss": 1.9314,
      "step": 163191
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.1342836618423462,
      "learning_rate": 0.00011740443130696318,
      "loss": 1.9603,
      "step": 163192
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.1751958131790161,
      "learning_rate": 0.00011740118571902045,
      "loss": 1.9974,
      "step": 163193
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.3520475625991821,
      "learning_rate": 0.0001173979401650264,
      "loss": 2.0597,
      "step": 163194
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.0918782949447632,
      "learning_rate": 0.00011739469464498148,
      "loss": 1.9519,
      "step": 163195
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.2019262313842773,
      "learning_rate": 0.00011739144915888637,
      "loss": 2.1213,
      "step": 163196
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.183922290802002,
      "learning_rate": 0.0001173882037067415,
      "loss": 2.0991,
      "step": 163197
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.1690003871917725,
      "learning_rate": 0.00011738495828854776,
      "loss": 1.9094,
      "step": 163198
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.3188802003860474,
      "learning_rate": 0.00011738171290430547,
      "loss": 1.8728,
      "step": 163199
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.139872670173645,
      "learning_rate": 0.00011737846755401547,
      "loss": 1.9444,
      "step": 163200
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.3073549270629883,
      "learning_rate": 0.00011737522223767827,
      "loss": 2.3139,
      "step": 163201
    },
    {
      "epoch": 2.12,
      "grad_norm": 1.5293978452682495,
      "learning_rate": 0.00011737197695529449,
      "loss": 1.8851,
      "step": 163202
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.2562930583953857,
      "learning_rate": 0.00011736873170686458,
      "loss": 2.2587,
      "step": 163203
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.3974342346191406,
      "learning_rate": 0.00011736548649238935,
      "loss": 2.0049,
      "step": 163204
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.0979032516479492,
      "learning_rate": 0.00011736224131186921,
      "loss": 1.9594,
      "step": 163205
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.184909701347351,
      "learning_rate": 0.00011735899616530501,
      "loss": 1.9749,
      "step": 163206
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.3670183420181274,
      "learning_rate": 0.00011735575105269722,
      "loss": 2.1697,
      "step": 163207
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.1819127798080444,
      "learning_rate": 0.00011735250597404646,
      "loss": 1.9043,
      "step": 163208
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.1385226249694824,
      "learning_rate": 0.0001173492609293532,
      "loss": 1.8082,
      "step": 163209
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.3864836692810059,
      "learning_rate": 0.00011734601591861825,
      "loss": 2.1265,
      "step": 163210
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.1976162195205688,
      "learning_rate": 0.00011734277094184204,
      "loss": 1.9184,
      "step": 163211
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.2832295894622803,
      "learning_rate": 0.00011733952599902535,
      "loss": 1.8149,
      "step": 163212
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.2698485851287842,
      "learning_rate": 0.00011733628109016857,
      "loss": 2.1464,
      "step": 163213
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.1611868143081665,
      "learning_rate": 0.00011733303621527265,
      "loss": 2.0888,
      "step": 163214
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.101396083831787,
      "learning_rate": 0.00011732979137433774,
      "loss": 1.9273,
      "step": 163215
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.1406686305999756,
      "learning_rate": 0.0001173265465673648,
      "loss": 2.0372,
      "step": 163216
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.2241603136062622,
      "learning_rate": 0.00011732330179435418,
      "loss": 2.1819,
      "step": 163217
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.135719895362854,
      "learning_rate": 0.00011732005705530674,
      "loss": 1.9654,
      "step": 163218
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.1547383069992065,
      "learning_rate": 0.00011731681235022282,
      "loss": 2.263,
      "step": 163219
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.2582181692123413,
      "learning_rate": 0.00011731356767910337,
      "loss": 2.1874,
      "step": 163220
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.1829657554626465,
      "learning_rate": 0.00011731032304194855,
      "loss": 2.2479,
      "step": 163221
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.2562092542648315,
      "learning_rate": 0.00011730707843875933,
      "loss": 2.0674,
      "step": 163222
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.1049336194992065,
      "learning_rate": 0.00011730383386953604,
      "loss": 2.0616,
      "step": 163223
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.1987643241882324,
      "learning_rate": 0.0001173005893342795,
      "loss": 1.8616,
      "step": 163224
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.2944648265838623,
      "learning_rate": 0.00011729734483299016,
      "loss": 2.0138,
      "step": 163225
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.2952855825424194,
      "learning_rate": 0.00011729410036566889,
      "loss": 2.2721,
      "step": 163226
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.2538992166519165,
      "learning_rate": 0.00011729085593231588,
      "loss": 1.9564,
      "step": 163227
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.2797250747680664,
      "learning_rate": 0.00011728761153293207,
      "loss": 1.8847,
      "step": 163228
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.2598390579223633,
      "learning_rate": 0.00011728436716751785,
      "loss": 2.2224,
      "step": 163229
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.1550095081329346,
      "learning_rate": 0.00011728112283607399,
      "loss": 1.9616,
      "step": 163230
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.1988099813461304,
      "learning_rate": 0.00011727787853860091,
      "loss": 2.0458,
      "step": 163231
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.28267240524292,
      "learning_rate": 0.00011727463427509957,
      "loss": 2.0149,
      "step": 163232
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.2902103662490845,
      "learning_rate": 0.00011727139004557008,
      "loss": 1.9528,
      "step": 163233
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.2745789289474487,
      "learning_rate": 0.00011726814585001338,
      "loss": 2.0678,
      "step": 163234
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.2506303787231445,
      "learning_rate": 0.00011726490168842992,
      "loss": 2.0481,
      "step": 163235
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.1692708730697632,
      "learning_rate": 0.00011726165756082044,
      "loss": 1.868,
      "step": 163236
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.202866554260254,
      "learning_rate": 0.00011725841346718537,
      "loss": 1.9481,
      "step": 163237
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.3872913122177124,
      "learning_rate": 0.0001172551694075256,
      "loss": 2.1581,
      "step": 163238
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.3307944536209106,
      "learning_rate": 0.00011725192538184136,
      "loss": 1.8764,
      "step": 163239
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.2304844856262207,
      "learning_rate": 0.00011724868139013354,
      "loss": 2.0591,
      "step": 163240
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.2339038848876953,
      "learning_rate": 0.00011724543743240254,
      "loss": 1.9762,
      "step": 163241
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.1386152505874634,
      "learning_rate": 0.00011724219350864915,
      "loss": 2.023,
      "step": 163242
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.813196063041687,
      "learning_rate": 0.00011723894961887379,
      "loss": 2.2458,
      "step": 163243
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.3581650257110596,
      "learning_rate": 0.00011723570576307728,
      "loss": 2.0672,
      "step": 163244
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.383061408996582,
      "learning_rate": 0.00011723246194126006,
      "loss": 1.945,
      "step": 163245
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.157390832901001,
      "learning_rate": 0.00011722921815342282,
      "loss": 2.2079,
      "step": 163246
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.2470660209655762,
      "learning_rate": 0.00011722597439956597,
      "loss": 2.0606,
      "step": 163247
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.2302676439285278,
      "learning_rate": 0.0001172227306796904,
      "loss": 2.0895,
      "step": 163248
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.1052019596099854,
      "learning_rate": 0.00011721948699379645,
      "loss": 2.1209,
      "step": 163249
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.2852678298950195,
      "learning_rate": 0.00011721624334188498,
      "loss": 2.2245,
      "step": 163250
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.197633147239685,
      "learning_rate": 0.00011721299972395644,
      "loss": 2.0843,
      "step": 163251
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.1814736127853394,
      "learning_rate": 0.00011720975614001147,
      "loss": 2.1468,
      "step": 163252
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.1847164630889893,
      "learning_rate": 0.00011720651259005051,
      "loss": 2.133,
      "step": 163253
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.2877006530761719,
      "learning_rate": 0.00011720326907407445,
      "loss": 1.9883,
      "step": 163254
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.162853717803955,
      "learning_rate": 0.00011720002559208366,
      "loss": 2.082,
      "step": 163255
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.3453166484832764,
      "learning_rate": 0.00011719678214407892,
      "loss": 2.1042,
      "step": 163256
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.2164193391799927,
      "learning_rate": 0.00011719353873006075,
      "loss": 1.8689,
      "step": 163257
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.165602207183838,
      "learning_rate": 0.00011719029535002964,
      "loss": 1.9403,
      "step": 163258
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.202692985534668,
      "learning_rate": 0.00011718705200398643,
      "loss": 1.9462,
      "step": 163259
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.1687833070755005,
      "learning_rate": 0.00011718380869193158,
      "loss": 1.8366,
      "step": 163260
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.2931674718856812,
      "learning_rate": 0.0001171805654138656,
      "loss": 1.9618,
      "step": 163261
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.2086083889007568,
      "learning_rate": 0.00011717732216978933,
      "loss": 1.9907,
      "step": 163262
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.272968053817749,
      "learning_rate": 0.00011717407895970324,
      "loss": 2.1587,
      "step": 163263
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.2072254419326782,
      "learning_rate": 0.00011717083578360783,
      "loss": 2.3347,
      "step": 163264
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.313968300819397,
      "learning_rate": 0.00011716759264150393,
      "loss": 1.8602,
      "step": 163265
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.245625376701355,
      "learning_rate": 0.0001171643495333919,
      "loss": 1.976,
      "step": 163266
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.300015926361084,
      "learning_rate": 0.00011716110645927258,
      "loss": 2.1111,
      "step": 163267
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.1655237674713135,
      "learning_rate": 0.00011715786341914647,
      "loss": 2.0797,
      "step": 163268
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.2481422424316406,
      "learning_rate": 0.00011715462041301417,
      "loss": 2.0449,
      "step": 163269
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.280962347984314,
      "learning_rate": 0.00011715137744087614,
      "loss": 1.9644,
      "step": 163270
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.3574960231781006,
      "learning_rate": 0.00011714813450273325,
      "loss": 2.2627,
      "step": 163271
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.1357632875442505,
      "learning_rate": 0.00011714489159858587,
      "loss": 1.8398,
      "step": 163272
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.1792157888412476,
      "learning_rate": 0.00011714164872843482,
      "loss": 2.233,
      "step": 163273
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.1700979471206665,
      "learning_rate": 0.00011713840589228059,
      "loss": 2.1413,
      "step": 163274
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.296413779258728,
      "learning_rate": 0.00011713516309012374,
      "loss": 2.1955,
      "step": 163275
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.4454092979431152,
      "learning_rate": 0.00011713192032196485,
      "loss": 2.1248,
      "step": 163276
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.4715286493301392,
      "learning_rate": 0.00011712867758780469,
      "loss": 1.9981,
      "step": 163277
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.255277395248413,
      "learning_rate": 0.00011712543488764364,
      "loss": 2.1538,
      "step": 163278
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.1829283237457275,
      "learning_rate": 0.00011712219222148255,
      "loss": 1.9446,
      "step": 163279
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.3343604803085327,
      "learning_rate": 0.00011711894958932178,
      "loss": 2.5186,
      "step": 163280
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.5179376602172852,
      "learning_rate": 0.00011711570699116226,
      "loss": 1.9489,
      "step": 163281
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.3026052713394165,
      "learning_rate": 0.00011711246442700415,
      "loss": 2.208,
      "step": 163282
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.2866618633270264,
      "learning_rate": 0.00011710922189684844,
      "loss": 2.2538,
      "step": 163283
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.322220802307129,
      "learning_rate": 0.00011710597940069543,
      "loss": 1.9992,
      "step": 163284
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.1360605955123901,
      "learning_rate": 0.00011710273693854598,
      "loss": 1.9998,
      "step": 163285
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.1812376976013184,
      "learning_rate": 0.00011709949451040051,
      "loss": 2.0078,
      "step": 163286
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.2717339992523193,
      "learning_rate": 0.00011709625211625985,
      "loss": 2.2601,
      "step": 163287
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.5801972150802612,
      "learning_rate": 0.00011709300975612428,
      "loss": 2.0244,
      "step": 163288
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.1262084245681763,
      "learning_rate": 0.00011708976742999469,
      "loss": 2.0607,
      "step": 163289
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.0789929628372192,
      "learning_rate": 0.00011708652513787144,
      "loss": 1.8825,
      "step": 163290
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.301927924156189,
      "learning_rate": 0.00011708328287975536,
      "loss": 2.0227,
      "step": 163291
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.1739500761032104,
      "learning_rate": 0.00011708004065564683,
      "loss": 1.9875,
      "step": 163292
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.2203984260559082,
      "learning_rate": 0.0001170767984655468,
      "loss": 1.9693,
      "step": 163293
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.1062825918197632,
      "learning_rate": 0.00011707355630945545,
      "loss": 1.9966,
      "step": 163294
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.1806447505950928,
      "learning_rate": 0.00011707031418737367,
      "loss": 2.1009,
      "step": 163295
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.233583927154541,
      "learning_rate": 0.00011706707209930187,
      "loss": 1.995,
      "step": 163296
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.2192280292510986,
      "learning_rate": 0.00011706383004524087,
      "loss": 2.0343,
      "step": 163297
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.3058714866638184,
      "learning_rate": 0.00011706058802519102,
      "loss": 2.2124,
      "step": 163298
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.2607516050338745,
      "learning_rate": 0.0001170573460391533,
      "loss": 2.1197,
      "step": 163299
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.2582917213439941,
      "learning_rate": 0.00011705410408712787,
      "loss": 2.0219,
      "step": 163300
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.2108166217803955,
      "learning_rate": 0.00011705086216911563,
      "loss": 2.3051,
      "step": 163301
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.2330278158187866,
      "learning_rate": 0.00011704762028511698,
      "loss": 2.0652,
      "step": 163302
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.2715182304382324,
      "learning_rate": 0.00011704437843513272,
      "loss": 2.0838,
      "step": 163303
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.192309021949768,
      "learning_rate": 0.00011704113661916328,
      "loss": 2.0981,
      "step": 163304
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.1562529802322388,
      "learning_rate": 0.00011703789483720954,
      "loss": 2.132,
      "step": 163305
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.1454615592956543,
      "learning_rate": 0.00011703465308927171,
      "loss": 2.1305,
      "step": 163306
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.236935019493103,
      "learning_rate": 0.00011703141137535068,
      "loss": 2.0242,
      "step": 163307
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.2471529245376587,
      "learning_rate": 0.00011702816969544687,
      "loss": 2.1146,
      "step": 163308
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.3298019170761108,
      "learning_rate": 0.00011702492804956108,
      "loss": 1.9854,
      "step": 163309
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.1986589431762695,
      "learning_rate": 0.00011702168643769372,
      "loss": 2.0774,
      "step": 163310
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.1667089462280273,
      "learning_rate": 0.00011701844485984567,
      "loss": 1.9945,
      "step": 163311
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.1516265869140625,
      "learning_rate": 0.00011701520331601714,
      "loss": 1.8563,
      "step": 163312
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.2411245107650757,
      "learning_rate": 0.00011701196180620908,
      "loss": 1.9719,
      "step": 163313
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.1833891868591309,
      "learning_rate": 0.00011700872033042182,
      "loss": 1.8707,
      "step": 163314
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.3118325471878052,
      "learning_rate": 0.0001170054788886562,
      "loss": 2.0374,
      "step": 163315
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.2803114652633667,
      "learning_rate": 0.00011700223748091257,
      "loss": 2.1337,
      "step": 163316
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.2022560834884644,
      "learning_rate": 0.00011699899610719195,
      "loss": 2.0794,
      "step": 163317
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.160287618637085,
      "learning_rate": 0.00011699575476749444,
      "loss": 1.9531,
      "step": 163318
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.3626717329025269,
      "learning_rate": 0.00011699251346182097,
      "loss": 1.9015,
      "step": 163319
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.1961411237716675,
      "learning_rate": 0.00011698927219017192,
      "loss": 1.9491,
      "step": 163320
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.2259637117385864,
      "learning_rate": 0.00011698603095254816,
      "loss": 1.983,
      "step": 163321
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.2464791536331177,
      "learning_rate": 0.00011698278974895005,
      "loss": 1.7435,
      "step": 163322
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.359703779220581,
      "learning_rate": 0.00011697954857937848,
      "loss": 2.0567,
      "step": 163323
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.2596275806427002,
      "learning_rate": 0.00011697630744383366,
      "loss": 2.0543,
      "step": 163324
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.3096797466278076,
      "learning_rate": 0.0001169730663423165,
      "loss": 2.1338,
      "step": 163325
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.2657597064971924,
      "learning_rate": 0.0001169698252748274,
      "loss": 2.1553,
      "step": 163326
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.1921948194503784,
      "learning_rate": 0.00011696658424136719,
      "loss": 2.1193,
      "step": 163327
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.1558133363723755,
      "learning_rate": 0.00011696334324193623,
      "loss": 1.8568,
      "step": 163328
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.303431749343872,
      "learning_rate": 0.00011696010227653534,
      "loss": 2.1398,
      "step": 163329
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.2634592056274414,
      "learning_rate": 0.00011695686134516503,
      "loss": 1.8996,
      "step": 163330
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.2057881355285645,
      "learning_rate": 0.00011695362044782586,
      "loss": 2.1893,
      "step": 163331
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.2182873487472534,
      "learning_rate": 0.00011695037958451842,
      "loss": 2.0541,
      "step": 163332
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.1362159252166748,
      "learning_rate": 0.00011694713875524344,
      "loss": 2.0694,
      "step": 163333
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.3071860074996948,
      "learning_rate": 0.00011694389796000131,
      "loss": 1.9925,
      "step": 163334
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.4218984842300415,
      "learning_rate": 0.00011694065719879292,
      "loss": 1.8396,
      "step": 163335
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.2413398027420044,
      "learning_rate": 0.00011693741647161868,
      "loss": 1.811,
      "step": 163336
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.342071533203125,
      "learning_rate": 0.00011693417577847923,
      "loss": 2.0723,
      "step": 163337
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.4307202100753784,
      "learning_rate": 0.00011693093511937508,
      "loss": 2.0434,
      "step": 163338
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.2594090700149536,
      "learning_rate": 0.00011692769449430705,
      "loss": 2.2882,
      "step": 163339
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.2426655292510986,
      "learning_rate": 0.00011692445390327546,
      "loss": 1.9095,
      "step": 163340
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.092074990272522,
      "learning_rate": 0.00011692121334628122,
      "loss": 1.88,
      "step": 163341
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.2527755498886108,
      "learning_rate": 0.00011691797282332478,
      "loss": 2.1173,
      "step": 163342
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.1762316226959229,
      "learning_rate": 0.00011691473233440662,
      "loss": 1.8872,
      "step": 163343
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.165982723236084,
      "learning_rate": 0.00011691149187952758,
      "loss": 1.7765,
      "step": 163344
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.234324336051941,
      "learning_rate": 0.00011690825145868813,
      "loss": 2.0322,
      "step": 163345
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.3541337251663208,
      "learning_rate": 0.00011690501107188881,
      "loss": 2.1371,
      "step": 163346
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.1783030033111572,
      "learning_rate": 0.00011690177071913043,
      "loss": 1.9091,
      "step": 163347
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.3434547185897827,
      "learning_rate": 0.00011689853040041341,
      "loss": 1.9686,
      "step": 163348
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.2157762050628662,
      "learning_rate": 0.00011689529011573834,
      "loss": 1.9138,
      "step": 163349
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.1524916887283325,
      "learning_rate": 0.000116892049865106,
      "loss": 1.9895,
      "step": 163350
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.3171643018722534,
      "learning_rate": 0.00011688880964851689,
      "loss": 2.0315,
      "step": 163351
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.1876256465911865,
      "learning_rate": 0.00011688556946597148,
      "loss": 1.9722,
      "step": 163352
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.354355812072754,
      "learning_rate": 0.00011688232931747062,
      "loss": 2.1485,
      "step": 163353
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.274905800819397,
      "learning_rate": 0.00011687908920301477,
      "loss": 1.9813,
      "step": 163354
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.1377241611480713,
      "learning_rate": 0.00011687584912260447,
      "loss": 2.0123,
      "step": 163355
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.2496196031570435,
      "learning_rate": 0.00011687260907624054,
      "loss": 2.0525,
      "step": 163356
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.2241863012313843,
      "learning_rate": 0.00011686936906392328,
      "loss": 1.905,
      "step": 163357
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.2258398532867432,
      "learning_rate": 0.00011686612908565361,
      "loss": 2.0857,
      "step": 163358
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.275884747505188,
      "learning_rate": 0.00011686288914143197,
      "loss": 2.0343,
      "step": 163359
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.2087892293930054,
      "learning_rate": 0.00011685964923125895,
      "loss": 2.0396,
      "step": 163360
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.4190571308135986,
      "learning_rate": 0.00011685640935513511,
      "loss": 2.1568,
      "step": 163361
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.3778314590454102,
      "learning_rate": 0.00011685316951306122,
      "loss": 2.3557,
      "step": 163362
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.2155494689941406,
      "learning_rate": 0.00011684992970503769,
      "loss": 2.2992,
      "step": 163363
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.275774359703064,
      "learning_rate": 0.0001168466899310653,
      "loss": 2.096,
      "step": 163364
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.1429343223571777,
      "learning_rate": 0.00011684345019114446,
      "loss": 1.9728,
      "step": 163365
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.2535310983657837,
      "learning_rate": 0.00011684021048527609,
      "loss": 2.0956,
      "step": 163366
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.147424578666687,
      "learning_rate": 0.00011683697081346037,
      "loss": 2.1607,
      "step": 163367
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.1891613006591797,
      "learning_rate": 0.00011683373117569826,
      "loss": 2.002,
      "step": 163368
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.2341701984405518,
      "learning_rate": 0.00011683049157199005,
      "loss": 1.9812,
      "step": 163369
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.3008474111557007,
      "learning_rate": 0.00011682725200233665,
      "loss": 2.1573,
      "step": 163370
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.2401330471038818,
      "learning_rate": 0.00011682401246673842,
      "loss": 2.0635,
      "step": 163371
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.2546592950820923,
      "learning_rate": 0.00011682077296519627,
      "loss": 2.0071,
      "step": 163372
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.2253540754318237,
      "learning_rate": 0.00011681753349771037,
      "loss": 2.3059,
      "step": 163373
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.14430832862854,
      "learning_rate": 0.00011681429406428167,
      "loss": 2.0831,
      "step": 163374
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.1497917175292969,
      "learning_rate": 0.00011681105466491051,
      "loss": 1.9164,
      "step": 163375
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.1610121726989746,
      "learning_rate": 0.0001168078152995978,
      "loss": 2.0101,
      "step": 163376
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.2494010925292969,
      "learning_rate": 0.00011680457596834382,
      "loss": 2.083,
      "step": 163377
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.2864408493041992,
      "learning_rate": 0.00011680133667114955,
      "loss": 1.9783,
      "step": 163378
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.1584118604660034,
      "learning_rate": 0.00011679809740801515,
      "loss": 1.9943,
      "step": 163379
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.1045116186141968,
      "learning_rate": 0.00011679485817894159,
      "loss": 2.0088,
      "step": 163380
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.1937395334243774,
      "learning_rate": 0.00011679161898392916,
      "loss": 1.9821,
      "step": 163381
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.3140146732330322,
      "learning_rate": 0.00011678837982297877,
      "loss": 2.1464,
      "step": 163382
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.268481969833374,
      "learning_rate": 0.00011678514069609077,
      "loss": 2.0967,
      "step": 163383
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.2480436563491821,
      "learning_rate": 0.00011678190160326606,
      "loss": 2.0787,
      "step": 163384
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.2307264804840088,
      "learning_rate": 0.0001167786625445049,
      "loss": 1.9568,
      "step": 163385
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.165363073348999,
      "learning_rate": 0.00011677542351980807,
      "loss": 2.0261,
      "step": 163386
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.2393230199813843,
      "learning_rate": 0.0001167721845291761,
      "loss": 2.0315,
      "step": 163387
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.2064380645751953,
      "learning_rate": 0.00011676894557260972,
      "loss": 2.0017,
      "step": 163388
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.1926491260528564,
      "learning_rate": 0.00011676570665010934,
      "loss": 1.9661,
      "step": 163389
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.2200666666030884,
      "learning_rate": 0.00011676246776167593,
      "loss": 1.9762,
      "step": 163390
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.3934974670410156,
      "learning_rate": 0.00011675922890730958,
      "loss": 2.3104,
      "step": 163391
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.4250999689102173,
      "learning_rate": 0.00011675599008701127,
      "loss": 1.8727,
      "step": 163392
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.1822476387023926,
      "learning_rate": 0.00011675275130078135,
      "loss": 1.9142,
      "step": 163393
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.3168059587478638,
      "learning_rate": 0.0001167495125486207,
      "loss": 2.1093,
      "step": 163394
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.2406697273254395,
      "learning_rate": 0.00011674627383052965,
      "loss": 2.0547,
      "step": 163395
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.2185550928115845,
      "learning_rate": 0.00011674303514650915,
      "loss": 2.1261,
      "step": 163396
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.2753844261169434,
      "learning_rate": 0.00011673979649655932,
      "loss": 2.2035,
      "step": 163397
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.1829850673675537,
      "learning_rate": 0.00011673655788068119,
      "loss": 1.9046,
      "step": 163398
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.2461572885513306,
      "learning_rate": 0.00011673331929887505,
      "loss": 2.07,
      "step": 163399
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.1762394905090332,
      "learning_rate": 0.00011673008075114176,
      "loss": 2.0219,
      "step": 163400
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.0691044330596924,
      "learning_rate": 0.00011672684223748168,
      "loss": 2.012,
      "step": 163401
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.1870962381362915,
      "learning_rate": 0.00011672360375789578,
      "loss": 2.1017,
      "step": 163402
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.2468541860580444,
      "learning_rate": 0.0001167203653123842,
      "loss": 1.8489,
      "step": 163403
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.2139564752578735,
      "learning_rate": 0.0001167171269009479,
      "loss": 2.217,
      "step": 163404
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.2691766023635864,
      "learning_rate": 0.00011671388852358719,
      "loss": 1.9793,
      "step": 163405
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.227699637413025,
      "learning_rate": 0.00011671065018030296,
      "loss": 1.9758,
      "step": 163406
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.2267025709152222,
      "learning_rate": 0.00011670741187109557,
      "loss": 1.9206,
      "step": 163407
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.1652040481567383,
      "learning_rate": 0.00011670417359596596,
      "loss": 1.9132,
      "step": 163408
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.2089818716049194,
      "learning_rate": 0.00011670093535491425,
      "loss": 2.0029,
      "step": 163409
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.3144066333770752,
      "learning_rate": 0.00011669769714794145,
      "loss": 1.8649,
      "step": 163410
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.2816842794418335,
      "learning_rate": 0.00011669445897504789,
      "loss": 2.0293,
      "step": 163411
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.2902082204818726,
      "learning_rate": 0.00011669122083623436,
      "loss": 1.9223,
      "step": 163412
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.2679171562194824,
      "learning_rate": 0.00011668798273150132,
      "loss": 2.0664,
      "step": 163413
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.3245395421981812,
      "learning_rate": 0.00011668474466084954,
      "loss": 2.026,
      "step": 163414
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.2033425569534302,
      "learning_rate": 0.00011668150662427953,
      "loss": 1.9979,
      "step": 163415
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.4175868034362793,
      "learning_rate": 0.0001166782686217919,
      "loss": 2.1357,
      "step": 163416
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.2345129251480103,
      "learning_rate": 0.00011667503065338713,
      "loss": 2.0563,
      "step": 163417
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.1441997289657593,
      "learning_rate": 0.00011667179271906603,
      "loss": 2.0627,
      "step": 163418
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.294348120689392,
      "learning_rate": 0.000116668554818829,
      "loss": 2.1096,
      "step": 163419
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.2722080945968628,
      "learning_rate": 0.00011666531695267684,
      "loss": 2.1068,
      "step": 163420
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.128589153289795,
      "learning_rate": 0.00011666207912061008,
      "loss": 2.1082,
      "step": 163421
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.2112795114517212,
      "learning_rate": 0.00011665884132262929,
      "loss": 2.1177,
      "step": 163422
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.2695530652999878,
      "learning_rate": 0.00011665560355873496,
      "loss": 2.0957,
      "step": 163423
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.1414462327957153,
      "learning_rate": 0.00011665236582892794,
      "loss": 1.9435,
      "step": 163424
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.1610239744186401,
      "learning_rate": 0.00011664912813320861,
      "loss": 1.6933,
      "step": 163425
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.3286283016204834,
      "learning_rate": 0.00011664589047157775,
      "loss": 2.0572,
      "step": 163426
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.2052370309829712,
      "learning_rate": 0.00011664265284403588,
      "loss": 1.8796,
      "step": 163427
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.2641361951828003,
      "learning_rate": 0.00011663941525058352,
      "loss": 2.2407,
      "step": 163428
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.2810044288635254,
      "learning_rate": 0.00011663617769122145,
      "loss": 2.0936,
      "step": 163429
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.1437269449234009,
      "learning_rate": 0.00011663294016595017,
      "loss": 1.9557,
      "step": 163430
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.2533230781555176,
      "learning_rate": 0.00011662970267477016,
      "loss": 1.985,
      "step": 163431
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.3159953355789185,
      "learning_rate": 0.00011662646521768228,
      "loss": 1.9842,
      "step": 163432
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.1101590394973755,
      "learning_rate": 0.000116623227794687,
      "loss": 2.0122,
      "step": 163433
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.2490999698638916,
      "learning_rate": 0.00011661999040578485,
      "loss": 1.99,
      "step": 163434
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.2747966051101685,
      "learning_rate": 0.00011661675305097655,
      "loss": 2.2803,
      "step": 163435
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.2564321756362915,
      "learning_rate": 0.00011661351573026273,
      "loss": 1.9683,
      "step": 163436
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.350342035293579,
      "learning_rate": 0.00011661027844364375,
      "loss": 2.1613,
      "step": 163437
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.2124611139297485,
      "learning_rate": 0.0001166070411911205,
      "loss": 1.9342,
      "step": 163438
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.3364590406417847,
      "learning_rate": 0.00011660380397269348,
      "loss": 2.0241,
      "step": 163439
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.197628378868103,
      "learning_rate": 0.00011660056678836315,
      "loss": 1.9752,
      "step": 163440
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.1525958776474,
      "learning_rate": 0.00011659732963813034,
      "loss": 1.9952,
      "step": 163441
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.2793335914611816,
      "learning_rate": 0.00011659409252199547,
      "loss": 2.278,
      "step": 163442
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.2537778615951538,
      "learning_rate": 0.00011659085543995931,
      "loss": 1.9506,
      "step": 163443
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.3582884073257446,
      "learning_rate": 0.00011658761839202238,
      "loss": 2.0806,
      "step": 163444
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.3554868698120117,
      "learning_rate": 0.00011658438137818527,
      "loss": 1.9479,
      "step": 163445
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.3202402591705322,
      "learning_rate": 0.00011658114439844848,
      "loss": 2.1106,
      "step": 163446
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.2450685501098633,
      "learning_rate": 0.00011657790745281281,
      "loss": 2.0369,
      "step": 163447
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.279032588005066,
      "learning_rate": 0.00011657467054127867,
      "loss": 2.0189,
      "step": 163448
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.1470918655395508,
      "learning_rate": 0.00011657143366384688,
      "loss": 1.8645,
      "step": 163449
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.2895793914794922,
      "learning_rate": 0.00011656819682051778,
      "loss": 2.2275,
      "step": 163450
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.2398724555969238,
      "learning_rate": 0.00011656496001129236,
      "loss": 2.076,
      "step": 163451
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.2163474559783936,
      "learning_rate": 0.00011656172323617077,
      "loss": 1.9918,
      "step": 163452
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.200682282447815,
      "learning_rate": 0.00011655848649515389,
      "loss": 2.1725,
      "step": 163453
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.161582112312317,
      "learning_rate": 0.00011655524978824217,
      "loss": 2.012,
      "step": 163454
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.2508447170257568,
      "learning_rate": 0.0001165520131154364,
      "loss": 1.9452,
      "step": 163455
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.1632784605026245,
      "learning_rate": 0.00011654877647673698,
      "loss": 2.1643,
      "step": 163456
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.1700611114501953,
      "learning_rate": 0.0001165455398721448,
      "loss": 2.149,
      "step": 163457
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.1957000494003296,
      "learning_rate": 0.00011654230330166004,
      "loss": 2.0818,
      "step": 163458
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.2192391157150269,
      "learning_rate": 0.00011653906676528364,
      "loss": 1.974,
      "step": 163459
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.1555604934692383,
      "learning_rate": 0.00011653583026301602,
      "loss": 1.9372,
      "step": 163460
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.1821521520614624,
      "learning_rate": 0.0001165325937948579,
      "loss": 2.0903,
      "step": 163461
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.1856153011322021,
      "learning_rate": 0.00011652935736080979,
      "loss": 2.0179,
      "step": 163462
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.181551218032837,
      "learning_rate": 0.0001165261209608725,
      "loss": 2.2584,
      "step": 163463
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.1181211471557617,
      "learning_rate": 0.00011652288459504627,
      "loss": 2.0916,
      "step": 163464
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.2241071462631226,
      "learning_rate": 0.00011651964826333203,
      "loss": 2.2914,
      "step": 163465
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.220605731010437,
      "learning_rate": 0.00011651641196573012,
      "loss": 2.1072,
      "step": 163466
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.143459677696228,
      "learning_rate": 0.00011651317570224139,
      "loss": 2.0442,
      "step": 163467
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.1434522867202759,
      "learning_rate": 0.0001165099394728662,
      "loss": 2.0221,
      "step": 163468
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.2070096731185913,
      "learning_rate": 0.0001165067032776055,
      "loss": 2.1812,
      "step": 163469
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.15496826171875,
      "learning_rate": 0.00011650346711645945,
      "loss": 2.0912,
      "step": 163470
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.2760738134384155,
      "learning_rate": 0.00011650023098942897,
      "loss": 1.8253,
      "step": 163471
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.1609899997711182,
      "learning_rate": 0.00011649699489651446,
      "loss": 1.9712,
      "step": 163472
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.329363226890564,
      "learning_rate": 0.00011649375883771673,
      "loss": 2.0839,
      "step": 163473
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.5233429670333862,
      "learning_rate": 0.00011649052281303618,
      "loss": 1.9536,
      "step": 163474
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.103095531463623,
      "learning_rate": 0.0001164872868224737,
      "loss": 1.9474,
      "step": 163475
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.2335797548294067,
      "learning_rate": 0.0001164840508660295,
      "loss": 1.9144,
      "step": 163476
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.3552063703536987,
      "learning_rate": 0.00011648081494370447,
      "loss": 2.0408,
      "step": 163477
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.3677150011062622,
      "learning_rate": 0.00011647757905549904,
      "loss": 2.5116,
      "step": 163478
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.218013882637024,
      "learning_rate": 0.00011647434320141397,
      "loss": 2.1453,
      "step": 163479
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.1984577178955078,
      "learning_rate": 0.0001164711073814497,
      "loss": 2.1343,
      "step": 163480
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.239133358001709,
      "learning_rate": 0.00011646787159560702,
      "loss": 2.0974,
      "step": 163481
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.213524580001831,
      "learning_rate": 0.00011646463584388643,
      "loss": 1.9882,
      "step": 163482
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.2866684198379517,
      "learning_rate": 0.0001164614001262885,
      "loss": 2.1499,
      "step": 163483
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.233379602432251,
      "learning_rate": 0.00011645816444281378,
      "loss": 2.1944,
      "step": 163484
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.2892417907714844,
      "learning_rate": 0.00011645492879346306,
      "loss": 1.9178,
      "step": 163485
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.2473522424697876,
      "learning_rate": 0.00011645169317823671,
      "loss": 2.3259,
      "step": 163486
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.1332213878631592,
      "learning_rate": 0.00011644845759713555,
      "loss": 1.8492,
      "step": 163487
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.2204910516738892,
      "learning_rate": 0.00011644522205016012,
      "loss": 2.0514,
      "step": 163488
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.2994601726531982,
      "learning_rate": 0.00011644198653731097,
      "loss": 1.8292,
      "step": 163489
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.1932464838027954,
      "learning_rate": 0.0001164387510585886,
      "loss": 2.0617,
      "step": 163490
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.1655617952346802,
      "learning_rate": 0.00011643551561399386,
      "loss": 2.1801,
      "step": 163491
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.2332831621170044,
      "learning_rate": 0.00011643228020352711,
      "loss": 2.1134,
      "step": 163492
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.1658227443695068,
      "learning_rate": 0.00011642904482718918,
      "loss": 1.8234,
      "step": 163493
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.241544485092163,
      "learning_rate": 0.00011642580948498054,
      "loss": 2.0396,
      "step": 163494
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.2904638051986694,
      "learning_rate": 0.00011642257417690179,
      "loss": 2.0066,
      "step": 163495
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.2529398202896118,
      "learning_rate": 0.00011641933890295348,
      "loss": 1.913,
      "step": 163496
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.2512481212615967,
      "learning_rate": 0.00011641610366313638,
      "loss": 2.1212,
      "step": 163497
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.1474932432174683,
      "learning_rate": 0.00011641286845745086,
      "loss": 2.0922,
      "step": 163498
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.3363196849822998,
      "learning_rate": 0.00011640963328589778,
      "loss": 2.1998,
      "step": 163499
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.527456283569336,
      "learning_rate": 0.00011640639814847762,
      "loss": 1.827,
      "step": 163500
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.3551421165466309,
      "learning_rate": 0.00011640316304519096,
      "loss": 2.1141,
      "step": 163501
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.3001444339752197,
      "learning_rate": 0.00011639992797603833,
      "loss": 2.0235,
      "step": 163502
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.2715433835983276,
      "learning_rate": 0.0001163966929410205,
      "loss": 1.9644,
      "step": 163503
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.5174208879470825,
      "learning_rate": 0.00011639345794013793,
      "loss": 2.0448,
      "step": 163504
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.4720529317855835,
      "learning_rate": 0.00011639022297339134,
      "loss": 1.9039,
      "step": 163505
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.2547574043273926,
      "learning_rate": 0.00011638698804078129,
      "loss": 2.2405,
      "step": 163506
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.2338162660598755,
      "learning_rate": 0.00011638375314230837,
      "loss": 1.9597,
      "step": 163507
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.2908879518508911,
      "learning_rate": 0.00011638051827797306,
      "loss": 2.0355,
      "step": 163508
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.527782678604126,
      "learning_rate": 0.0001163772834477762,
      "loss": 2.0582,
      "step": 163509
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.2097421884536743,
      "learning_rate": 0.00011637404865171815,
      "loss": 1.9627,
      "step": 163510
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.3861356973648071,
      "learning_rate": 0.00011637081388979972,
      "loss": 1.9729,
      "step": 163511
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.3617470264434814,
      "learning_rate": 0.00011636757916202145,
      "loss": 1.9628,
      "step": 163512
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.3613611459732056,
      "learning_rate": 0.00011636434446838391,
      "loss": 2.0313,
      "step": 163513
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.332259178161621,
      "learning_rate": 0.00011636110980888759,
      "loss": 2.2841,
      "step": 163514
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.2675434350967407,
      "learning_rate": 0.00011635787518353332,
      "loss": 2.1662,
      "step": 163515
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.278943419456482,
      "learning_rate": 0.00011635464059232147,
      "loss": 1.9187,
      "step": 163516
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.4380728006362915,
      "learning_rate": 0.00011635140603525289,
      "loss": 1.844,
      "step": 163517
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.3723313808441162,
      "learning_rate": 0.00011634817151232803,
      "loss": 2.0258,
      "step": 163518
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.2654637098312378,
      "learning_rate": 0.00011634493702354741,
      "loss": 2.0832,
      "step": 163519
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.1859855651855469,
      "learning_rate": 0.00011634170256891182,
      "loss": 2.1196,
      "step": 163520
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.1798416376113892,
      "learning_rate": 0.00011633846814842179,
      "loss": 2.0382,
      "step": 163521
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.310340404510498,
      "learning_rate": 0.0001163352337620778,
      "loss": 2.2078,
      "step": 163522
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.310422658920288,
      "learning_rate": 0.00011633199940988067,
      "loss": 2.2846,
      "step": 163523
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.260224461555481,
      "learning_rate": 0.00011632876509183087,
      "loss": 2.0535,
      "step": 163524
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.179739236831665,
      "learning_rate": 0.00011632553080792895,
      "loss": 1.8905,
      "step": 163525
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.2167259454727173,
      "learning_rate": 0.00011632229655817567,
      "loss": 2.1581,
      "step": 163526
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.156917929649353,
      "learning_rate": 0.0001163190623425714,
      "loss": 2.1518,
      "step": 163527
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.1296908855438232,
      "learning_rate": 0.00011631582816111702,
      "loss": 2.0308,
      "step": 163528
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.2126352787017822,
      "learning_rate": 0.00011631259401381298,
      "loss": 1.945,
      "step": 163529
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.3454653024673462,
      "learning_rate": 0.00011630935990065993,
      "loss": 2.0664,
      "step": 163530
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.2523503303527832,
      "learning_rate": 0.0001163061258216583,
      "loss": 2.1465,
      "step": 163531
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.2078917026519775,
      "learning_rate": 0.00011630289177680895,
      "loss": 1.9811,
      "step": 163532
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.3566088676452637,
      "learning_rate": 0.00011629965776611224,
      "loss": 1.9358,
      "step": 163533
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.2197308540344238,
      "learning_rate": 0.00011629642378956902,
      "loss": 1.7752,
      "step": 163534
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.3107002973556519,
      "learning_rate": 0.00011629318984717977,
      "loss": 1.9405,
      "step": 163535
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.149580478668213,
      "learning_rate": 0.00011628995593894504,
      "loss": 2.1001,
      "step": 163536
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.1475580930709839,
      "learning_rate": 0.00011628672206486541,
      "loss": 2.0058,
      "step": 163537
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.3871477842330933,
      "learning_rate": 0.00011628348822494163,
      "loss": 1.9901,
      "step": 163538
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.5002501010894775,
      "learning_rate": 0.0001162802544191741,
      "loss": 2.4423,
      "step": 163539
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.343225121498108,
      "learning_rate": 0.00011627702064756368,
      "loss": 2.1292,
      "step": 163540
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.245652437210083,
      "learning_rate": 0.0001162737869101107,
      "loss": 2.0232,
      "step": 163541
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.2400283813476562,
      "learning_rate": 0.0001162705532068161,
      "loss": 2.0155,
      "step": 163542
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.320921778678894,
      "learning_rate": 0.00011626731953768006,
      "loss": 1.957,
      "step": 163543
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.2989981174468994,
      "learning_rate": 0.00011626408590270351,
      "loss": 2.108,
      "step": 163544
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.3060224056243896,
      "learning_rate": 0.00011626085230188684,
      "loss": 1.9396,
      "step": 163545
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.2472134828567505,
      "learning_rate": 0.00011625761873523084,
      "loss": 1.8336,
      "step": 163546
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.1795507669448853,
      "learning_rate": 0.00011625438520273591,
      "loss": 2.0944,
      "step": 163547
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.2144492864608765,
      "learning_rate": 0.00011625115170440286,
      "loss": 1.8609,
      "step": 163548
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.3196449279785156,
      "learning_rate": 0.00011624791824023217,
      "loss": 2.1394,
      "step": 163549
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.2931592464447021,
      "learning_rate": 0.0001162446848102245,
      "loss": 1.9462,
      "step": 163550
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.2678101062774658,
      "learning_rate": 0.0001162414514143803,
      "loss": 2.0752,
      "step": 163551
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.1724742650985718,
      "learning_rate": 0.00011623821805270036,
      "loss": 2.1174,
      "step": 163552
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.2358267307281494,
      "learning_rate": 0.00011623498472518513,
      "loss": 2.2204,
      "step": 163553
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.205236792564392,
      "learning_rate": 0.00011623175143183537,
      "loss": 1.8735,
      "step": 163554
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.237964153289795,
      "learning_rate": 0.00011622851817265158,
      "loss": 2.0361,
      "step": 163555
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.2551175355911255,
      "learning_rate": 0.00011622528494763444,
      "loss": 2.2124,
      "step": 163556
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.2491645812988281,
      "learning_rate": 0.00011622205175678432,
      "loss": 2.1255,
      "step": 163557
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.1410959959030151,
      "learning_rate": 0.00011621881860010214,
      "loss": 2.0886,
      "step": 163558
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.1355340480804443,
      "learning_rate": 0.00011621558547758819,
      "loss": 2.1172,
      "step": 163559
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.2381222248077393,
      "learning_rate": 0.00011621235238924337,
      "loss": 2.0785,
      "step": 163560
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.2696608304977417,
      "learning_rate": 0.00011620911933506817,
      "loss": 1.9531,
      "step": 163561
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.3910574913024902,
      "learning_rate": 0.00011620588631506312,
      "loss": 2.223,
      "step": 163562
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.153147578239441,
      "learning_rate": 0.00011620265332922878,
      "loss": 2.0451,
      "step": 163563
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.2310229539871216,
      "learning_rate": 0.00011619942037756593,
      "loss": 2.2114,
      "step": 163564
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.190887689590454,
      "learning_rate": 0.00011619618746007498,
      "loss": 2.1161,
      "step": 163565
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.205117106437683,
      "learning_rate": 0.00011619295457675677,
      "loss": 2.1582,
      "step": 163566
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.1150883436203003,
      "learning_rate": 0.00011618972172761172,
      "loss": 1.8863,
      "step": 163567
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.397005558013916,
      "learning_rate": 0.00011618648891264046,
      "loss": 2.2406,
      "step": 163568
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.073305606842041,
      "learning_rate": 0.0001161832561318435,
      "loss": 2.0241,
      "step": 163569
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.2280504703521729,
      "learning_rate": 0.00011618002338522167,
      "loss": 1.8537,
      "step": 163570
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.2814337015151978,
      "learning_rate": 0.00011617679067277535,
      "loss": 2.1228,
      "step": 163571
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.1941697597503662,
      "learning_rate": 0.00011617355799450531,
      "loss": 2.0813,
      "step": 163572
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.2408593893051147,
      "learning_rate": 0.00011617032535041206,
      "loss": 2.0621,
      "step": 163573
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.3725991249084473,
      "learning_rate": 0.00011616709274049625,
      "loss": 2.1636,
      "step": 163574
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.0866819620132446,
      "learning_rate": 0.00011616386016475833,
      "loss": 2.0307,
      "step": 163575
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.1739590167999268,
      "learning_rate": 0.00011616062762319913,
      "loss": 1.9228,
      "step": 163576
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.1913678646087646,
      "learning_rate": 0.00011615739511581903,
      "loss": 2.191,
      "step": 163577
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.2470641136169434,
      "learning_rate": 0.00011615416264261885,
      "loss": 2.0356,
      "step": 163578
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.208092212677002,
      "learning_rate": 0.00011615093020359907,
      "loss": 2.1409,
      "step": 163579
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.2535098791122437,
      "learning_rate": 0.00011614769779876031,
      "loss": 2.0352,
      "step": 163580
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.3760416507720947,
      "learning_rate": 0.00011614446542810305,
      "loss": 2.3578,
      "step": 163581
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.2178142070770264,
      "learning_rate": 0.00011614123309162812,
      "loss": 2.2641,
      "step": 163582
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.457133173942566,
      "learning_rate": 0.0001161380007893359,
      "loss": 2.0393,
      "step": 163583
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.3058518171310425,
      "learning_rate": 0.0001161347685212272,
      "loss": 2.1272,
      "step": 163584
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.281815767288208,
      "learning_rate": 0.00011613153628730253,
      "loss": 2.1134,
      "step": 163585
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.184440016746521,
      "learning_rate": 0.00011612830408756245,
      "loss": 1.9501,
      "step": 163586
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.5216740369796753,
      "learning_rate": 0.00011612507192200751,
      "loss": 2.1254,
      "step": 163587
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.20845365524292,
      "learning_rate": 0.00011612183979063846,
      "loss": 1.9449,
      "step": 163588
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.177051305770874,
      "learning_rate": 0.00011611860769345577,
      "loss": 2.1032,
      "step": 163589
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.1265063285827637,
      "learning_rate": 0.00011611537563046016,
      "loss": 1.916,
      "step": 163590
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.1604607105255127,
      "learning_rate": 0.0001161121436016522,
      "loss": 2.1164,
      "step": 163591
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.2019433975219727,
      "learning_rate": 0.00011610891160703248,
      "loss": 2.0667,
      "step": 163592
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.2024130821228027,
      "learning_rate": 0.00011610567964660146,
      "loss": 1.9167,
      "step": 163593
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.2878531217575073,
      "learning_rate": 0.00011610244772035995,
      "loss": 2.063,
      "step": 163594
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.1894564628601074,
      "learning_rate": 0.0001160992158283084,
      "loss": 1.8921,
      "step": 163595
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.3337359428405762,
      "learning_rate": 0.00011609598397044758,
      "loss": 2.1429,
      "step": 163596
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.2057572603225708,
      "learning_rate": 0.00011609275214677796,
      "loss": 2.0987,
      "step": 163597
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.1677074432373047,
      "learning_rate": 0.0001160895203573002,
      "loss": 1.9114,
      "step": 163598
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.409965991973877,
      "learning_rate": 0.00011608628860201475,
      "loss": 1.963,
      "step": 163599
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.2265125513076782,
      "learning_rate": 0.00011608305688092242,
      "loss": 1.953,
      "step": 163600
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.2392737865447998,
      "learning_rate": 0.00011607982519402364,
      "loss": 1.9558,
      "step": 163601
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.2274798154830933,
      "learning_rate": 0.0001160765935413192,
      "loss": 1.8922,
      "step": 163602
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.1486786603927612,
      "learning_rate": 0.00011607336192280958,
      "loss": 2.1243,
      "step": 163603
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.1947169303894043,
      "learning_rate": 0.00011607013033849526,
      "loss": 2.1251,
      "step": 163604
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.286152958869934,
      "learning_rate": 0.00011606689878837712,
      "loss": 1.9507,
      "step": 163605
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.2552680969238281,
      "learning_rate": 0.00011606366727245561,
      "loss": 2.1741,
      "step": 163606
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.2634220123291016,
      "learning_rate": 0.00011606043579073123,
      "loss": 2.1749,
      "step": 163607
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.1981770992279053,
      "learning_rate": 0.00011605720434320477,
      "loss": 1.9922,
      "step": 163608
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.2302355766296387,
      "learning_rate": 0.00011605397292987678,
      "loss": 1.9696,
      "step": 163609
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.1528379917144775,
      "learning_rate": 0.0001160507415507477,
      "loss": 1.9283,
      "step": 163610
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.2955600023269653,
      "learning_rate": 0.00011604751020581837,
      "loss": 2.0367,
      "step": 163611
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.3612972497940063,
      "learning_rate": 0.00011604427889508918,
      "loss": 2.1261,
      "step": 163612
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.1878759860992432,
      "learning_rate": 0.00011604104761856093,
      "loss": 1.9935,
      "step": 163613
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.2120097875595093,
      "learning_rate": 0.00011603781637623412,
      "loss": 2.0809,
      "step": 163614
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.329667091369629,
      "learning_rate": 0.00011603458516810926,
      "loss": 2.0378,
      "step": 163615
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.2101130485534668,
      "learning_rate": 0.00011603135399418714,
      "loss": 2.0115,
      "step": 163616
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.2000541687011719,
      "learning_rate": 0.00011602812285446829,
      "loss": 2.1078,
      "step": 163617
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.2910940647125244,
      "learning_rate": 0.00011602489174895314,
      "loss": 1.982,
      "step": 163618
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.1898043155670166,
      "learning_rate": 0.00011602166067764256,
      "loss": 2.1443,
      "step": 163619
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.2100061178207397,
      "learning_rate": 0.00011601842964053704,
      "loss": 1.9142,
      "step": 163620
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.1763588190078735,
      "learning_rate": 0.00011601519863763704,
      "loss": 2.1746,
      "step": 163621
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.253908395767212,
      "learning_rate": 0.00011601196766894339,
      "loss": 2.0649,
      "step": 163622
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.133123517036438,
      "learning_rate": 0.0001160087367344566,
      "loss": 2.1134,
      "step": 163623
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.29555082321167,
      "learning_rate": 0.00011600550583417714,
      "loss": 1.8263,
      "step": 163624
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.136921763420105,
      "learning_rate": 0.00011600227496810584,
      "loss": 1.9527,
      "step": 163625
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.2158316373825073,
      "learning_rate": 0.0001159990441362431,
      "loss": 1.9891,
      "step": 163626
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.3258740901947021,
      "learning_rate": 0.00011599581333858968,
      "loss": 2.052,
      "step": 163627
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.4130799770355225,
      "learning_rate": 0.00011599258257514615,
      "loss": 1.9116,
      "step": 163628
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.1621986627578735,
      "learning_rate": 0.00011598935184591307,
      "loss": 1.9797,
      "step": 163629
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.325954556465149,
      "learning_rate": 0.00011598612115089093,
      "loss": 2.127,
      "step": 163630
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.3009166717529297,
      "learning_rate": 0.00011598289049008052,
      "loss": 2.2225,
      "step": 163631
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.2042044401168823,
      "learning_rate": 0.00011597965986348229,
      "loss": 2.0621,
      "step": 163632
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.288197636604309,
      "learning_rate": 0.00011597642927109701,
      "loss": 2.0058,
      "step": 163633
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.2873579263687134,
      "learning_rate": 0.0001159731987129252,
      "loss": 2.229,
      "step": 163634
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.1936867237091064,
      "learning_rate": 0.00011596996818896743,
      "loss": 1.7187,
      "step": 163635
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.2730417251586914,
      "learning_rate": 0.00011596673769922422,
      "loss": 2.1395,
      "step": 163636
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.1267075538635254,
      "learning_rate": 0.00011596350724369637,
      "loss": 1.9873,
      "step": 163637
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.4227055311203003,
      "learning_rate": 0.00011596027682238429,
      "loss": 2.0379,
      "step": 163638
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.1971828937530518,
      "learning_rate": 0.00011595704643528878,
      "loss": 1.9611,
      "step": 163639
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.2487940788269043,
      "learning_rate": 0.00011595381608241027,
      "loss": 1.9696,
      "step": 163640
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.1538156270980835,
      "learning_rate": 0.00011595058576374948,
      "loss": 2.1552,
      "step": 163641
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.4168328046798706,
      "learning_rate": 0.00011594735547930682,
      "loss": 1.9919,
      "step": 163642
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.2380582094192505,
      "learning_rate": 0.00011594412522908311,
      "loss": 1.9885,
      "step": 163643
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.3339592218399048,
      "learning_rate": 0.00011594089501307879,
      "loss": 2.1466,
      "step": 163644
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.1993557214736938,
      "learning_rate": 0.0001159376648312946,
      "loss": 2.0538,
      "step": 163645
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.2580360174179077,
      "learning_rate": 0.00011593443468373111,
      "loss": 1.8745,
      "step": 163646
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.4475924968719482,
      "learning_rate": 0.00011593120457038885,
      "loss": 1.9784,
      "step": 163647
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.1425931453704834,
      "learning_rate": 0.00011592797449126838,
      "loss": 1.8922,
      "step": 163648
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.2053494453430176,
      "learning_rate": 0.00011592474444637046,
      "loss": 2.1582,
      "step": 163649
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.3038898706436157,
      "learning_rate": 0.00011592151443569549,
      "loss": 2.2551,
      "step": 163650
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.193529486656189,
      "learning_rate": 0.00011591828445924432,
      "loss": 1.9858,
      "step": 163651
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.3253016471862793,
      "learning_rate": 0.00011591505451701738,
      "loss": 2.1938,
      "step": 163652
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.1445585489273071,
      "learning_rate": 0.00011591182460901533,
      "loss": 1.8499,
      "step": 163653
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.287235975265503,
      "learning_rate": 0.00011590859473523865,
      "loss": 2.0385,
      "step": 163654
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.3064565658569336,
      "learning_rate": 0.00011590536489568811,
      "loss": 1.9552,
      "step": 163655
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.2149157524108887,
      "learning_rate": 0.00011590213509036415,
      "loss": 2.067,
      "step": 163656
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.1417421102523804,
      "learning_rate": 0.00011589890531926758,
      "loss": 2.2805,
      "step": 163657
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.320277214050293,
      "learning_rate": 0.00011589567558239888,
      "loss": 2.0556,
      "step": 163658
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.1994695663452148,
      "learning_rate": 0.0001158924458797586,
      "loss": 1.9461,
      "step": 163659
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.3116120100021362,
      "learning_rate": 0.00011588921621134733,
      "loss": 2.0898,
      "step": 163660
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.1870362758636475,
      "learning_rate": 0.00011588598657716585,
      "loss": 1.907,
      "step": 163661
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.2294365167617798,
      "learning_rate": 0.00011588275697721451,
      "loss": 2.0027,
      "step": 163662
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.2272511720657349,
      "learning_rate": 0.00011587952741149412,
      "loss": 1.8699,
      "step": 163663
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.2119497060775757,
      "learning_rate": 0.00011587629788000525,
      "loss": 2.2017,
      "step": 163664
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.2696459293365479,
      "learning_rate": 0.0001158730683827484,
      "loss": 2.1275,
      "step": 163665
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.2309520244598389,
      "learning_rate": 0.00011586983891972417,
      "loss": 2.1444,
      "step": 163666
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.3272722959518433,
      "learning_rate": 0.00011586660949093329,
      "loss": 1.8964,
      "step": 163667
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.5047096014022827,
      "learning_rate": 0.00011586338009637618,
      "loss": 2.2736,
      "step": 163668
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.341866374015808,
      "learning_rate": 0.00011586015073605366,
      "loss": 1.9502,
      "step": 163669
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.3606802225112915,
      "learning_rate": 0.00011585692140996619,
      "loss": 2.0185,
      "step": 163670
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.1896908283233643,
      "learning_rate": 0.0001158536921181144,
      "loss": 2.1117,
      "step": 163671
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.4858973026275635,
      "learning_rate": 0.00011585046286049879,
      "loss": 2.0096,
      "step": 163672
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.3393504619598389,
      "learning_rate": 0.00011584723363712015,
      "loss": 1.7638,
      "step": 163673
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.2508493661880493,
      "learning_rate": 0.0001158440044479789,
      "loss": 1.9608,
      "step": 163674
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.1383253335952759,
      "learning_rate": 0.00011584077529307582,
      "loss": 2.1406,
      "step": 163675
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.2925457954406738,
      "learning_rate": 0.0001158375461724114,
      "loss": 2.1673,
      "step": 163676
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.3374298810958862,
      "learning_rate": 0.00011583431708598628,
      "loss": 1.8633,
      "step": 163677
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.2453324794769287,
      "learning_rate": 0.00011583108803380091,
      "loss": 1.9146,
      "step": 163678
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.2710144519805908,
      "learning_rate": 0.00011582785901585616,
      "loss": 2.022,
      "step": 163679
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.162659764289856,
      "learning_rate": 0.00011582463003215237,
      "loss": 1.9003,
      "step": 163680
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.158024549484253,
      "learning_rate": 0.00011582140108269032,
      "loss": 2.0186,
      "step": 163681
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.3780033588409424,
      "learning_rate": 0.00011581817216747049,
      "loss": 2.0489,
      "step": 163682
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.345738410949707,
      "learning_rate": 0.00011581494328649373,
      "loss": 1.9936,
      "step": 163683
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.2004444599151611,
      "learning_rate": 0.00011581171443976025,
      "loss": 2.1499,
      "step": 163684
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.3120700120925903,
      "learning_rate": 0.00011580848562727094,
      "loss": 2.2506,
      "step": 163685
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.1818896532058716,
      "learning_rate": 0.00011580525684902621,
      "loss": 2.0119,
      "step": 163686
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.33573317527771,
      "learning_rate": 0.00011580202810502685,
      "loss": 2.022,
      "step": 163687
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.309762954711914,
      "learning_rate": 0.00011579879939527329,
      "loss": 2.3133,
      "step": 163688
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.2613517045974731,
      "learning_rate": 0.00011579557071976631,
      "loss": 2.0712,
      "step": 163689
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.384125828742981,
      "learning_rate": 0.0001157923420785064,
      "loss": 2.0484,
      "step": 163690
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.184157133102417,
      "learning_rate": 0.00011578911347149419,
      "loss": 1.8531,
      "step": 163691
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.1784462928771973,
      "learning_rate": 0.00011578588489873011,
      "loss": 1.9808,
      "step": 163692
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.2598505020141602,
      "learning_rate": 0.00011578265636021505,
      "loss": 2.0725,
      "step": 163693
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.2816319465637207,
      "learning_rate": 0.00011577942785594939,
      "loss": 2.2132,
      "step": 163694
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.398343563079834,
      "learning_rate": 0.00011577619938593387,
      "loss": 2.2708,
      "step": 163695
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.136872410774231,
      "learning_rate": 0.00011577297095016908,
      "loss": 2.1426,
      "step": 163696
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.594691514968872,
      "learning_rate": 0.00011576974254865552,
      "loss": 1.9947,
      "step": 163697
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.2059776782989502,
      "learning_rate": 0.00011576651418139377,
      "loss": 1.9504,
      "step": 163698
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.1880888938903809,
      "learning_rate": 0.00011576328584838462,
      "loss": 2.0126,
      "step": 163699
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.1336380243301392,
      "learning_rate": 0.00011576005754962842,
      "loss": 2.0551,
      "step": 163700
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.160566806793213,
      "learning_rate": 0.00011575682928512601,
      "loss": 1.8142,
      "step": 163701
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.2201751470565796,
      "learning_rate": 0.00011575360105487788,
      "loss": 2.04,
      "step": 163702
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.3102983236312866,
      "learning_rate": 0.00011575037285888456,
      "loss": 1.9987,
      "step": 163703
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.2608609199523926,
      "learning_rate": 0.0001157471446971468,
      "loss": 2.1736,
      "step": 163704
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.370210886001587,
      "learning_rate": 0.00011574391656966512,
      "loss": 2.039,
      "step": 163705
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.3512836694717407,
      "learning_rate": 0.00011574068847644003,
      "loss": 1.8987,
      "step": 163706
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.1843076944351196,
      "learning_rate": 0.00011573746041747233,
      "loss": 2.081,
      "step": 163707
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.377507209777832,
      "learning_rate": 0.00011573423239276253,
      "loss": 1.939,
      "step": 163708
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.296399712562561,
      "learning_rate": 0.00011573100440231108,
      "loss": 2.119,
      "step": 163709
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.2227281332015991,
      "learning_rate": 0.00011572777644611885,
      "loss": 2.0158,
      "step": 163710
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.3058918714523315,
      "learning_rate": 0.00011572454852418616,
      "loss": 1.934,
      "step": 163711
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.3038312196731567,
      "learning_rate": 0.00011572132063651388,
      "loss": 1.8956,
      "step": 163712
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.1575160026550293,
      "learning_rate": 0.0001157180927831025,
      "loss": 1.9212,
      "step": 163713
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.2395148277282715,
      "learning_rate": 0.00011571486496395259,
      "loss": 2.1259,
      "step": 163714
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.2540955543518066,
      "learning_rate": 0.00011571163717906466,
      "loss": 2.212,
      "step": 163715
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.4561971426010132,
      "learning_rate": 0.00011570840942843954,
      "loss": 1.9427,
      "step": 163716
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.5026390552520752,
      "learning_rate": 0.00011570518171207759,
      "loss": 1.823,
      "step": 163717
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.3503471612930298,
      "learning_rate": 0.00011570195402997963,
      "loss": 1.9197,
      "step": 163718
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.3489329814910889,
      "learning_rate": 0.00011569872638214613,
      "loss": 1.8677,
      "step": 163719
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.1278640031814575,
      "learning_rate": 0.00011569549876857775,
      "loss": 2.131,
      "step": 163720
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.1302205324172974,
      "learning_rate": 0.00011569227118927492,
      "loss": 2.1021,
      "step": 163721
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.1433945894241333,
      "learning_rate": 0.00011568904364423849,
      "loss": 2.099,
      "step": 163722
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.2559266090393066,
      "learning_rate": 0.00011568581613346884,
      "loss": 2.106,
      "step": 163723
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.138576865196228,
      "learning_rate": 0.00011568258865696682,
      "loss": 1.8719,
      "step": 163724
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.1107306480407715,
      "learning_rate": 0.00011567936121473286,
      "loss": 1.7702,
      "step": 163725
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.2252742052078247,
      "learning_rate": 0.00011567613380676758,
      "loss": 1.9953,
      "step": 163726
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.4555636644363403,
      "learning_rate": 0.00011567290643307145,
      "loss": 2.342,
      "step": 163727
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.1897794008255005,
      "learning_rate": 0.00011566967909364535,
      "loss": 2.1859,
      "step": 163728
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.2211724519729614,
      "learning_rate": 0.00011566645178848964,
      "loss": 1.9719,
      "step": 163729
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.5162700414657593,
      "learning_rate": 0.0001156632245176051,
      "loss": 2.1009,
      "step": 163730
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.281967282295227,
      "learning_rate": 0.00011565999728099226,
      "loss": 1.9704,
      "step": 163731
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.1011043787002563,
      "learning_rate": 0.00011565677007865169,
      "loss": 2.0137,
      "step": 163732
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.2722995281219482,
      "learning_rate": 0.00011565354291058392,
      "loss": 2.3255,
      "step": 163733
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.1616417169570923,
      "learning_rate": 0.00011565031577678973,
      "loss": 2.0537,
      "step": 163734
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.177328109741211,
      "learning_rate": 0.00011564708867726952,
      "loss": 2.1091,
      "step": 163735
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.294311285018921,
      "learning_rate": 0.0001156438616120241,
      "loss": 2.1644,
      "step": 163736
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.1901010274887085,
      "learning_rate": 0.00011564063458105397,
      "loss": 2.2268,
      "step": 163737
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.1959192752838135,
      "learning_rate": 0.00011563740758435973,
      "loss": 2.1511,
      "step": 163738
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.2864185571670532,
      "learning_rate": 0.00011563418062194188,
      "loss": 2.174,
      "step": 163739
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.130807638168335,
      "learning_rate": 0.00011563095369380118,
      "loss": 2.1638,
      "step": 163740
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.1799719333648682,
      "learning_rate": 0.00011562772679993809,
      "loss": 2.2095,
      "step": 163741
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.160133719444275,
      "learning_rate": 0.00011562449994035339,
      "loss": 2.0568,
      "step": 163742
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.1688807010650635,
      "learning_rate": 0.0001156212731150476,
      "loss": 1.9895,
      "step": 163743
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.249626636505127,
      "learning_rate": 0.00011561804632402126,
      "loss": 2.158,
      "step": 163744
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.105953574180603,
      "learning_rate": 0.00011561481956727492,
      "loss": 1.836,
      "step": 163745
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.169501781463623,
      "learning_rate": 0.00011561159284480934,
      "loss": 2.0618,
      "step": 163746
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.2477083206176758,
      "learning_rate": 0.00011560836615662495,
      "loss": 1.9029,
      "step": 163747
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.2644462585449219,
      "learning_rate": 0.0001156051395027226,
      "loss": 1.9467,
      "step": 163748
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.1647244691848755,
      "learning_rate": 0.00011560191288310256,
      "loss": 1.8245,
      "step": 163749
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.233820915222168,
      "learning_rate": 0.00011559868629776585,
      "loss": 1.9065,
      "step": 163750
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.199233055114746,
      "learning_rate": 0.00011559545974671262,
      "loss": 1.9978,
      "step": 163751
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.2723056077957153,
      "learning_rate": 0.00011559223322994378,
      "loss": 2.1105,
      "step": 163752
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.2876695394515991,
      "learning_rate": 0.0001155890067474597,
      "loss": 2.2068,
      "step": 163753
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.3022624254226685,
      "learning_rate": 0.00011558578029926123,
      "loss": 2.0056,
      "step": 163754
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.2845169305801392,
      "learning_rate": 0.00011558255388534875,
      "loss": 2.4068,
      "step": 163755
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.2039424180984497,
      "learning_rate": 0.00011557932750572316,
      "loss": 2.119,
      "step": 163756
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.6691133975982666,
      "learning_rate": 0.0001155761011603846,
      "loss": 2.0338,
      "step": 163757
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.2226223945617676,
      "learning_rate": 0.00011557287484933407,
      "loss": 2.01,
      "step": 163758
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.2901231050491333,
      "learning_rate": 0.00011556964857257189,
      "loss": 1.9992,
      "step": 163759
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.1319260597229004,
      "learning_rate": 0.00011556642233009892,
      "loss": 2.0212,
      "step": 163760
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.883424997329712,
      "learning_rate": 0.00011556319612191555,
      "loss": 1.9673,
      "step": 163761
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.5487102270126343,
      "learning_rate": 0.00011555996994802265,
      "loss": 1.9267,
      "step": 163762
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.2373253107070923,
      "learning_rate": 0.00011555674380842042,
      "loss": 2.1122,
      "step": 163763
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.183869481086731,
      "learning_rate": 0.00011555351770310979,
      "loss": 1.9238,
      "step": 163764
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.2622891664505005,
      "learning_rate": 0.00011555029163209109,
      "loss": 2.0772,
      "step": 163765
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.544255018234253,
      "learning_rate": 0.00011554706559536524,
      "loss": 2.1561,
      "step": 163766
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.610308289527893,
      "learning_rate": 0.00011554383959293255,
      "loss": 2.224,
      "step": 163767
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.925586462020874,
      "learning_rate": 0.00011554061362479393,
      "loss": 1.9818,
      "step": 163768
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.1270633935928345,
      "learning_rate": 0.00011553738769094962,
      "loss": 1.8385,
      "step": 163769
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.3686000108718872,
      "learning_rate": 0.00011553416179140046,
      "loss": 2.1235,
      "step": 163770
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.6055912971496582,
      "learning_rate": 0.00011553093592614687,
      "loss": 1.9246,
      "step": 163771
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.6455129384994507,
      "learning_rate": 0.00011552771009518971,
      "loss": 2.2253,
      "step": 163772
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.2772499322891235,
      "learning_rate": 0.0001155244842985293,
      "loss": 1.9773,
      "step": 163773
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.2842086553573608,
      "learning_rate": 0.00011552125853616647,
      "loss": 2.0028,
      "step": 163774
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.3558008670806885,
      "learning_rate": 0.00011551803280810175,
      "loss": 2.2134,
      "step": 163775
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.340976357460022,
      "learning_rate": 0.00011551480711433569,
      "loss": 1.9794,
      "step": 163776
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.1663439273834229,
      "learning_rate": 0.00011551158145486879,
      "loss": 2.0122,
      "step": 163777
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.2458415031433105,
      "learning_rate": 0.00011550835582970188,
      "loss": 2.0778,
      "step": 163778
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.159658432006836,
      "learning_rate": 0.00011550513023883534,
      "loss": 2.3204,
      "step": 163779
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.471137523651123,
      "learning_rate": 0.00011550190468226999,
      "loss": 1.9823,
      "step": 163780
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.2214750051498413,
      "learning_rate": 0.00011549867916000635,
      "loss": 1.9525,
      "step": 163781
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.215883731842041,
      "learning_rate": 0.00011549545367204496,
      "loss": 2.1764,
      "step": 163782
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.3505009412765503,
      "learning_rate": 0.00011549222821838633,
      "loss": 2.1955,
      "step": 163783
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.1312499046325684,
      "learning_rate": 0.00011548900279903129,
      "loss": 2.0617,
      "step": 163784
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.2431352138519287,
      "learning_rate": 0.00011548577741398025,
      "loss": 2.0695,
      "step": 163785
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.1638836860656738,
      "learning_rate": 0.00011548255206323397,
      "loss": 2.0351,
      "step": 163786
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.1935217380523682,
      "learning_rate": 0.00011547932674679298,
      "loss": 2.0446,
      "step": 163787
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.4900885820388794,
      "learning_rate": 0.00011547610146465774,
      "loss": 2.0909,
      "step": 163788
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.1754649877548218,
      "learning_rate": 0.0001154728762168291,
      "loss": 1.9152,
      "step": 163789
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.1246228218078613,
      "learning_rate": 0.00011546965100330752,
      "loss": 1.9784,
      "step": 163790
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.2283539772033691,
      "learning_rate": 0.00011546642582409352,
      "loss": 2.2282,
      "step": 163791
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.174182653427124,
      "learning_rate": 0.00011546320067918788,
      "loss": 1.7877,
      "step": 163792
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.0924780368804932,
      "learning_rate": 0.00011545997556859113,
      "loss": 2.0769,
      "step": 163793
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.1500509977340698,
      "learning_rate": 0.00011545675049230377,
      "loss": 2.0956,
      "step": 163794
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.191514015197754,
      "learning_rate": 0.00011545352545032657,
      "loss": 2.0886,
      "step": 163795
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.1344764232635498,
      "learning_rate": 0.00011545030044266005,
      "loss": 2.1235,
      "step": 163796
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.264803171157837,
      "learning_rate": 0.00011544707546930468,
      "loss": 1.8559,
      "step": 163797
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.1563693284988403,
      "learning_rate": 0.0001154438505302613,
      "loss": 2.1595,
      "step": 163798
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.1106593608856201,
      "learning_rate": 0.00011544062562553038,
      "loss": 2.1037,
      "step": 163799
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.2521480321884155,
      "learning_rate": 0.00011543740075511244,
      "loss": 1.9678,
      "step": 163800
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.1660616397857666,
      "learning_rate": 0.00011543417591900825,
      "loss": 2.0489,
      "step": 163801
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.3164324760437012,
      "learning_rate": 0.00011543095111721825,
      "loss": 2.2308,
      "step": 163802
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.2484124898910522,
      "learning_rate": 0.00011542772634974323,
      "loss": 2.5045,
      "step": 163803
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.1697075366973877,
      "learning_rate": 0.00011542450161658366,
      "loss": 2.0924,
      "step": 163804
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.252134919166565,
      "learning_rate": 0.00011542127691774018,
      "loss": 2.0948,
      "step": 163805
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.5550968647003174,
      "learning_rate": 0.00011541805225321328,
      "loss": 2.0995,
      "step": 163806
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.2265894412994385,
      "learning_rate": 0.00011541482762300373,
      "loss": 2.1028,
      "step": 163807
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.2037290334701538,
      "learning_rate": 0.00011541160302711192,
      "loss": 2.2098,
      "step": 163808
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.1721080541610718,
      "learning_rate": 0.0001154083784655387,
      "loss": 1.9708,
      "step": 163809
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.125026822090149,
      "learning_rate": 0.00011540515393828456,
      "loss": 1.876,
      "step": 163810
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.4485821723937988,
      "learning_rate": 0.0001154019294453501,
      "loss": 2.0475,
      "step": 163811
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.1892939805984497,
      "learning_rate": 0.00011539870498673575,
      "loss": 1.9686,
      "step": 163812
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.2876620292663574,
      "learning_rate": 0.00011539548056244241,
      "loss": 1.8686,
      "step": 163813
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.1745790243148804,
      "learning_rate": 0.00011539225617247043,
      "loss": 2.0399,
      "step": 163814
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.374024510383606,
      "learning_rate": 0.00011538903181682057,
      "loss": 2.145,
      "step": 163815
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.33562433719635,
      "learning_rate": 0.00011538580749549334,
      "loss": 2.0427,
      "step": 163816
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.1469873189926147,
      "learning_rate": 0.00011538258320848956,
      "loss": 1.9926,
      "step": 163817
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.588127851486206,
      "learning_rate": 0.0001153793589558094,
      "loss": 2.0698,
      "step": 163818
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.1520167589187622,
      "learning_rate": 0.00011537613473745383,
      "loss": 2.1024,
      "step": 163819
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.2600295543670654,
      "learning_rate": 0.00011537291055342321,
      "loss": 1.8632,
      "step": 163820
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.1909610033035278,
      "learning_rate": 0.00011536968640371837,
      "loss": 1.9347,
      "step": 163821
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.1726062297821045,
      "learning_rate": 0.0001153664622883397,
      "loss": 2.0755,
      "step": 163822
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.3281816244125366,
      "learning_rate": 0.00011536323820728806,
      "loss": 2.063,
      "step": 163823
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.164157748222351,
      "learning_rate": 0.00011536001416056367,
      "loss": 2.0685,
      "step": 163824
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.2236307859420776,
      "learning_rate": 0.00011535679014816747,
      "loss": 1.8501,
      "step": 163825
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.248459815979004,
      "learning_rate": 0.0001153535661700998,
      "loss": 1.9914,
      "step": 163826
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.2256443500518799,
      "learning_rate": 0.00011535034222636152,
      "loss": 1.9839,
      "step": 163827
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.3487509489059448,
      "learning_rate": 0.00011534711831695299,
      "loss": 2.0719,
      "step": 163828
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.2616249322891235,
      "learning_rate": 0.0001153438944418751,
      "loss": 1.9676,
      "step": 163829
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.124448537826538,
      "learning_rate": 0.00011534067060112807,
      "loss": 2.046,
      "step": 163830
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.4041738510131836,
      "learning_rate": 0.00011533744679471282,
      "loss": 1.9617,
      "step": 163831
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.3172231912612915,
      "learning_rate": 0.00011533422302262966,
      "loss": 2.0765,
      "step": 163832
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.210878610610962,
      "learning_rate": 0.00011533099928487952,
      "loss": 2.0769,
      "step": 163833
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.4654853343963623,
      "learning_rate": 0.00011532777558146272,
      "loss": 2.2505,
      "step": 163834
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.3060225248336792,
      "learning_rate": 0.00011532455191238017,
      "loss": 2.0659,
      "step": 163835
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.3197754621505737,
      "learning_rate": 0.00011532132827763204,
      "loss": 2.0321,
      "step": 163836
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.1891154050827026,
      "learning_rate": 0.0001153181046772193,
      "loss": 2.0534,
      "step": 163837
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.276868224143982,
      "learning_rate": 0.00011531488111114229,
      "loss": 2.1427,
      "step": 163838
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.260233759880066,
      "learning_rate": 0.00011531165757940181,
      "loss": 2.1712,
      "step": 163839
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.3137284517288208,
      "learning_rate": 0.00011530843408199831,
      "loss": 2.135,
      "step": 163840
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.5678865909576416,
      "learning_rate": 0.00011530521061893268,
      "loss": 2.0327,
      "step": 163841
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.4123644828796387,
      "learning_rate": 0.00011530198719020507,
      "loss": 2.1871,
      "step": 163842
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.163212537765503,
      "learning_rate": 0.0001152987637958164,
      "loss": 2.0601,
      "step": 163843
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.2245655059814453,
      "learning_rate": 0.00011529554043576709,
      "loss": 1.9599,
      "step": 163844
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.1925854682922363,
      "learning_rate": 0.0001152923171100579,
      "loss": 1.9166,
      "step": 163845
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.1391700506210327,
      "learning_rate": 0.0001152890938186893,
      "loss": 2.1442,
      "step": 163846
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.2714430093765259,
      "learning_rate": 0.00011528587056166211,
      "loss": 1.819,
      "step": 163847
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.2812215089797974,
      "learning_rate": 0.00011528264733897656,
      "loss": 1.9174,
      "step": 163848
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.2278894186019897,
      "learning_rate": 0.00011527942415063355,
      "loss": 2.1227,
      "step": 163849
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.187319040298462,
      "learning_rate": 0.00011527620099663353,
      "loss": 1.973,
      "step": 163850
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.16709566116333,
      "learning_rate": 0.00011527297787697719,
      "loss": 2.1774,
      "step": 163851
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.1983294486999512,
      "learning_rate": 0.00011526975479166501,
      "loss": 2.1676,
      "step": 163852
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.3171992301940918,
      "learning_rate": 0.0001152665317406979,
      "loss": 2.0873,
      "step": 163853
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.3664147853851318,
      "learning_rate": 0.000115263308724076,
      "loss": 1.9829,
      "step": 163854
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.1956143379211426,
      "learning_rate": 0.00011526008574180025,
      "loss": 2.0148,
      "step": 163855
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.1786909103393555,
      "learning_rate": 0.00011525686279387104,
      "loss": 2.1141,
      "step": 163856
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.7306697368621826,
      "learning_rate": 0.00011525363988028917,
      "loss": 2.067,
      "step": 163857
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.311644434928894,
      "learning_rate": 0.000115250417001055,
      "loss": 2.1971,
      "step": 163858
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.26376211643219,
      "learning_rate": 0.00011524719415616951,
      "loss": 1.9908,
      "step": 163859
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.1924693584442139,
      "learning_rate": 0.00011524397134563284,
      "loss": 1.9584,
      "step": 163860
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.7204078435897827,
      "learning_rate": 0.00011524074856944588,
      "loss": 1.7865,
      "step": 163861
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.3102463483810425,
      "learning_rate": 0.00011523752582760906,
      "loss": 2.0483,
      "step": 163862
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.1863555908203125,
      "learning_rate": 0.00011523430312012319,
      "loss": 2.1538,
      "step": 163863
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.3821086883544922,
      "learning_rate": 0.00011523108044698864,
      "loss": 1.9523,
      "step": 163864
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.190542459487915,
      "learning_rate": 0.00011522785780820621,
      "loss": 1.8511,
      "step": 163865
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.1562507152557373,
      "learning_rate": 0.00011522463520377643,
      "loss": 2.0137,
      "step": 163866
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.2163894176483154,
      "learning_rate": 0.00011522141263369987,
      "loss": 2.0562,
      "step": 163867
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.2579386234283447,
      "learning_rate": 0.00011521819009797704,
      "loss": 1.9764,
      "step": 163868
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.2064164876937866,
      "learning_rate": 0.00011521496759660872,
      "loss": 2.1272,
      "step": 163869
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.2708849906921387,
      "learning_rate": 0.00011521174512959536,
      "loss": 1.9883,
      "step": 163870
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.3018710613250732,
      "learning_rate": 0.00011520852269693768,
      "loss": 2.1172,
      "step": 163871
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.2152471542358398,
      "learning_rate": 0.00011520530029863625,
      "loss": 2.1138,
      "step": 163872
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.279585838317871,
      "learning_rate": 0.00011520207793469154,
      "loss": 1.8188,
      "step": 163873
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.4115853309631348,
      "learning_rate": 0.00011519885560510435,
      "loss": 2.1044,
      "step": 163874
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.1726648807525635,
      "learning_rate": 0.0001151956333098752,
      "loss": 2.2923,
      "step": 163875
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.2370041608810425,
      "learning_rate": 0.00011519241104900454,
      "loss": 2.0776,
      "step": 163876
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.303823471069336,
      "learning_rate": 0.00011518918882249323,
      "loss": 2.0848,
      "step": 163877
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.2882592678070068,
      "learning_rate": 0.00011518596663034169,
      "loss": 1.9088,
      "step": 163878
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.2468231916427612,
      "learning_rate": 0.00011518274447255051,
      "loss": 2.0862,
      "step": 163879
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.2294745445251465,
      "learning_rate": 0.00011517952234912045,
      "loss": 2.0518,
      "step": 163880
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.19770085811615,
      "learning_rate": 0.000115176300260052,
      "loss": 2.0645,
      "step": 163881
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.240563154220581,
      "learning_rate": 0.00011517307820534562,
      "loss": 2.1184,
      "step": 163882
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.2704620361328125,
      "learning_rate": 0.00011516985618500219,
      "loss": 1.9913,
      "step": 163883
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.1572977304458618,
      "learning_rate": 0.00011516663419902217,
      "loss": 1.9905,
      "step": 163884
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.138718843460083,
      "learning_rate": 0.00011516341224740606,
      "loss": 2.0376,
      "step": 163885
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.2032783031463623,
      "learning_rate": 0.00011516019033015466,
      "loss": 2.2193,
      "step": 163886
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.120550274848938,
      "learning_rate": 0.00011515696844726838,
      "loss": 2.0519,
      "step": 163887
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.444061279296875,
      "learning_rate": 0.00011515374659874801,
      "loss": 2.1563,
      "step": 163888
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.2808040380477905,
      "learning_rate": 0.00011515052478459404,
      "loss": 1.8881,
      "step": 163889
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.4022986888885498,
      "learning_rate": 0.00011514730300480709,
      "loss": 2.0131,
      "step": 163890
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.286078691482544,
      "learning_rate": 0.00011514408125938761,
      "loss": 2.0043,
      "step": 163891
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.311193823814392,
      "learning_rate": 0.00011514085954833648,
      "loss": 1.9706,
      "step": 163892
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.3904529809951782,
      "learning_rate": 0.00011513763787165401,
      "loss": 1.9897,
      "step": 163893
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.2255603075027466,
      "learning_rate": 0.00011513441622934106,
      "loss": 1.8998,
      "step": 163894
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.2459815740585327,
      "learning_rate": 0.00011513119462139801,
      "loss": 2.0729,
      "step": 163895
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.327148199081421,
      "learning_rate": 0.0001151279730478258,
      "loss": 2.2068,
      "step": 163896
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.4052402973175049,
      "learning_rate": 0.00011512475150862453,
      "loss": 1.9474,
      "step": 163897
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.2057521343231201,
      "learning_rate": 0.00011512153000379519,
      "loss": 1.8319,
      "step": 163898
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.134946584701538,
      "learning_rate": 0.00011511830853333814,
      "loss": 1.9463,
      "step": 163899
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.3226393461227417,
      "learning_rate": 0.00011511508709725417,
      "loss": 1.9049,
      "step": 163900
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.146723747253418,
      "learning_rate": 0.00011511186569554372,
      "loss": 1.8265,
      "step": 163901
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.457932710647583,
      "learning_rate": 0.00011510864432820767,
      "loss": 1.9914,
      "step": 163902
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.288024663925171,
      "learning_rate": 0.0001151054229952462,
      "loss": 2.0119,
      "step": 163903
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.3033747673034668,
      "learning_rate": 0.00011510220169666022,
      "loss": 1.949,
      "step": 163904
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.263545036315918,
      "learning_rate": 0.00011509898043245011,
      "loss": 1.7832,
      "step": 163905
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.1071149110794067,
      "learning_rate": 0.00011509575920261674,
      "loss": 1.9027,
      "step": 163906
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.272850751876831,
      "learning_rate": 0.0001150925380071604,
      "loss": 2.0251,
      "step": 163907
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.4027669429779053,
      "learning_rate": 0.00011508931684608211,
      "loss": 1.9453,
      "step": 163908
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.2126085758209229,
      "learning_rate": 0.00011508609571938197,
      "loss": 2.1159,
      "step": 163909
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.2805049419403076,
      "learning_rate": 0.00011508287462706093,
      "loss": 1.9377,
      "step": 163910
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.373887300491333,
      "learning_rate": 0.00011507965356911936,
      "loss": 1.9638,
      "step": 163911
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.1400030851364136,
      "learning_rate": 0.00011507643254555808,
      "loss": 2.1417,
      "step": 163912
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.39197838306427,
      "learning_rate": 0.00011507321155637748,
      "loss": 2.1885,
      "step": 163913
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.2308580875396729,
      "learning_rate": 0.00011506999060157846,
      "loss": 2.2454,
      "step": 163914
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.1650198698043823,
      "learning_rate": 0.00011506676968116123,
      "loss": 2.0847,
      "step": 163915
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.1944931745529175,
      "learning_rate": 0.00011506354879512663,
      "loss": 2.1886,
      "step": 163916
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.266006588935852,
      "learning_rate": 0.00011506032794347515,
      "loss": 2.1134,
      "step": 163917
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.2547482252120972,
      "learning_rate": 0.00011505710712620756,
      "loss": 2.2049,
      "step": 163918
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.4739404916763306,
      "learning_rate": 0.00011505388634332423,
      "loss": 2.1247,
      "step": 163919
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.1578929424285889,
      "learning_rate": 0.00011505066559482606,
      "loss": 1.881,
      "step": 163920
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.2123544216156006,
      "learning_rate": 0.0001150474448807133,
      "loss": 1.9476,
      "step": 163921
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.1368221044540405,
      "learning_rate": 0.00011504422420098677,
      "loss": 2.2201,
      "step": 163922
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.3984863758087158,
      "learning_rate": 0.00011504100355564693,
      "loss": 1.9366,
      "step": 163923
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.2492179870605469,
      "learning_rate": 0.00011503778294469454,
      "loss": 2.1375,
      "step": 163924
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.2072505950927734,
      "learning_rate": 0.00011503456236813003,
      "loss": 1.8845,
      "step": 163925
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.1827279329299927,
      "learning_rate": 0.00011503134182595429,
      "loss": 1.9899,
      "step": 163926
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.1523464918136597,
      "learning_rate": 0.0001150281213181675,
      "loss": 2.1096,
      "step": 163927
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.2098686695098877,
      "learning_rate": 0.00011502490084477058,
      "loss": 1.8863,
      "step": 163928
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.339250087738037,
      "learning_rate": 0.00011502168040576395,
      "loss": 2.1775,
      "step": 163929
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.2395602464675903,
      "learning_rate": 0.00011501846000114834,
      "loss": 2.0156,
      "step": 163930
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.1081308126449585,
      "learning_rate": 0.00011501523963092422,
      "loss": 1.6244,
      "step": 163931
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.3094552755355835,
      "learning_rate": 0.00011501201929509245,
      "loss": 1.9893,
      "step": 163932
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.2365275621414185,
      "learning_rate": 0.00011500879899365324,
      "loss": 1.8322,
      "step": 163933
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.9567673206329346,
      "learning_rate": 0.00011500557872660749,
      "loss": 1.8927,
      "step": 163934
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.4029819965362549,
      "learning_rate": 0.00011500235849395558,
      "loss": 2.0723,
      "step": 163935
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.3712252378463745,
      "learning_rate": 0.00011499913829569834,
      "loss": 1.8635,
      "step": 163936
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.271619439125061,
      "learning_rate": 0.00011499591813183614,
      "loss": 1.9459,
      "step": 163937
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.3030121326446533,
      "learning_rate": 0.00011499269800236989,
      "loss": 2.078,
      "step": 163938
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.4111891984939575,
      "learning_rate": 0.00011498947790729977,
      "loss": 2.0709,
      "step": 163939
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.3306902647018433,
      "learning_rate": 0.00011498625784662674,
      "loss": 2.0965,
      "step": 163940
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.2331076860427856,
      "learning_rate": 0.00011498303782035116,
      "loss": 1.8997,
      "step": 163941
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.150984764099121,
      "learning_rate": 0.00011497981782847379,
      "loss": 2.0196,
      "step": 163942
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.3254820108413696,
      "learning_rate": 0.00011497659787099507,
      "loss": 2.1941,
      "step": 163943
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.5326632261276245,
      "learning_rate": 0.0001149733779479159,
      "loss": 2.0845,
      "step": 163944
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.27225923538208,
      "learning_rate": 0.00011497015805923648,
      "loss": 1.825,
      "step": 163945
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.2948366403579712,
      "learning_rate": 0.00011496693820495772,
      "loss": 1.9461,
      "step": 163946
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.170307993888855,
      "learning_rate": 0.00011496371838507996,
      "loss": 1.913,
      "step": 163947
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.887025237083435,
      "learning_rate": 0.00011496049859960405,
      "loss": 2.3716,
      "step": 163948
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.4855916500091553,
      "learning_rate": 0.00011495727884853037,
      "loss": 2.0641,
      "step": 163949
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.3910350799560547,
      "learning_rate": 0.00011495405913185973,
      "loss": 1.9565,
      "step": 163950
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.2165336608886719,
      "learning_rate": 0.00011495083944959263,
      "loss": 2.0505,
      "step": 163951
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.3142186403274536,
      "learning_rate": 0.00011494761980172967,
      "loss": 2.0246,
      "step": 163952
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.2793235778808594,
      "learning_rate": 0.0001149444001882713,
      "loss": 2.1039,
      "step": 163953
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.3538696765899658,
      "learning_rate": 0.00011494118060921841,
      "loss": 2.1092,
      "step": 163954
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.2525053024291992,
      "learning_rate": 0.00011493796106457132,
      "loss": 2.1087,
      "step": 163955
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.3868460655212402,
      "learning_rate": 0.00011493474155433086,
      "loss": 1.997,
      "step": 163956
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.3131266832351685,
      "learning_rate": 0.00011493152207849753,
      "loss": 1.9487,
      "step": 163957
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.2985914945602417,
      "learning_rate": 0.0001149283026370719,
      "loss": 1.9977,
      "step": 163958
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.290135383605957,
      "learning_rate": 0.00011492508323005454,
      "loss": 2.2421,
      "step": 163959
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.3065608739852905,
      "learning_rate": 0.00011492186385744617,
      "loss": 1.9689,
      "step": 163960
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.2591701745986938,
      "learning_rate": 0.00011491864451924722,
      "loss": 2.0172,
      "step": 163961
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.188451886177063,
      "learning_rate": 0.00011491542521545847,
      "loss": 1.9999,
      "step": 163962
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.222213864326477,
      "learning_rate": 0.00011491220594608046,
      "loss": 2.1653,
      "step": 163963
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.18159019947052,
      "learning_rate": 0.00011490898671111366,
      "loss": 2.2012,
      "step": 163964
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.1541446447372437,
      "learning_rate": 0.00011490576751055886,
      "loss": 1.9954,
      "step": 163965
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.2473862171173096,
      "learning_rate": 0.00011490254834441662,
      "loss": 2.1447,
      "step": 163966
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.3254728317260742,
      "learning_rate": 0.00011489932921268733,
      "loss": 2.0809,
      "step": 163967
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.0870258808135986,
      "learning_rate": 0.00011489611011537189,
      "loss": 1.9855,
      "step": 163968
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.1700681447982788,
      "learning_rate": 0.00011489289105247075,
      "loss": 2.1467,
      "step": 163969
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.3781925439834595,
      "learning_rate": 0.00011488967202398438,
      "loss": 2.1801,
      "step": 163970
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.1118054389953613,
      "learning_rate": 0.00011488645302991366,
      "loss": 2.1324,
      "step": 163971
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.1930568218231201,
      "learning_rate": 0.00011488323407025892,
      "loss": 2.0839,
      "step": 163972
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.1711227893829346,
      "learning_rate": 0.000114880015145021,
      "loss": 1.9592,
      "step": 163973
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.2884660959243774,
      "learning_rate": 0.00011487679625420034,
      "loss": 1.7552,
      "step": 163974
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.1688034534454346,
      "learning_rate": 0.00011487357739779762,
      "loss": 1.9154,
      "step": 163975
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.1378016471862793,
      "learning_rate": 0.00011487035857581326,
      "loss": 1.7077,
      "step": 163976
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.3044246435165405,
      "learning_rate": 0.00011486713978824813,
      "loss": 2.005,
      "step": 163977
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.1856452226638794,
      "learning_rate": 0.00011486392103510257,
      "loss": 1.8397,
      "step": 163978
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.2378333806991577,
      "learning_rate": 0.00011486070231637741,
      "loss": 1.9301,
      "step": 163979
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.1977531909942627,
      "learning_rate": 0.00011485748363207314,
      "loss": 2.0703,
      "step": 163980
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.1438114643096924,
      "learning_rate": 0.00011485426498219039,
      "loss": 1.7109,
      "step": 163981
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.3285226821899414,
      "learning_rate": 0.00011485104636672956,
      "loss": 2.0738,
      "step": 163982
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.453567385673523,
      "learning_rate": 0.00011484782778569158,
      "loss": 1.9513,
      "step": 163983
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.2739931344985962,
      "learning_rate": 0.00011484460923907675,
      "loss": 2.2819,
      "step": 163984
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.2682243585586548,
      "learning_rate": 0.00011484139072688588,
      "loss": 2.1738,
      "step": 163985
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.3578928709030151,
      "learning_rate": 0.00011483817224911941,
      "loss": 2.1454,
      "step": 163986
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.3617990016937256,
      "learning_rate": 0.00011483495380577825,
      "loss": 2.0266,
      "step": 163987
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.2655233144760132,
      "learning_rate": 0.00011483173539686255,
      "loss": 2.1529,
      "step": 163988
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.1354879140853882,
      "learning_rate": 0.0001148285170223732,
      "loss": 1.9002,
      "step": 163989
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.2078309059143066,
      "learning_rate": 0.00011482529868231065,
      "loss": 2.1379,
      "step": 163990
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.394586443901062,
      "learning_rate": 0.00011482208037667566,
      "loss": 2.088,
      "step": 163991
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.4853869676589966,
      "learning_rate": 0.00011481886210546863,
      "loss": 1.9393,
      "step": 163992
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.1905220746994019,
      "learning_rate": 0.0001148156438686905,
      "loss": 1.9745,
      "step": 163993
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.341603398323059,
      "learning_rate": 0.00011481242566634142,
      "loss": 2.0642,
      "step": 163994
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.3426040410995483,
      "learning_rate": 0.00011480920749842231,
      "loss": 1.9417,
      "step": 163995
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.4525601863861084,
      "learning_rate": 0.00011480598936493358,
      "loss": 2.0337,
      "step": 163996
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.315795660018921,
      "learning_rate": 0.00011480277126587598,
      "loss": 2.0735,
      "step": 163997
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.335125207901001,
      "learning_rate": 0.00011479955320124997,
      "loss": 1.955,
      "step": 163998
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.1238223314285278,
      "learning_rate": 0.00011479633517105641,
      "loss": 2.0519,
      "step": 163999
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.3058967590332031,
      "learning_rate": 0.00011479311717529549,
      "loss": 1.9242,
      "step": 164000
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.3240364789962769,
      "learning_rate": 0.00011478989921396815,
      "loss": 2.0731,
      "step": 164001
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.1880253553390503,
      "learning_rate": 0.00011478668128707474,
      "loss": 2.0794,
      "step": 164002
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.2532954216003418,
      "learning_rate": 0.0001147834633946161,
      "loss": 2.127,
      "step": 164003
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.1305363178253174,
      "learning_rate": 0.0001147802455365926,
      "loss": 1.869,
      "step": 164004
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.3114901781082153,
      "learning_rate": 0.00011477702771300514,
      "loss": 1.8862,
      "step": 164005
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.2275148630142212,
      "learning_rate": 0.00011477380992385394,
      "loss": 2.2571,
      "step": 164006
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.184953212738037,
      "learning_rate": 0.0001147705921691399,
      "loss": 1.8423,
      "step": 164007
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.1586483716964722,
      "learning_rate": 0.0001147673744488634,
      "loss": 2.0329,
      "step": 164008
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.182813286781311,
      "learning_rate": 0.00011476415676302522,
      "loss": 2.0557,
      "step": 164009
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.1464855670928955,
      "learning_rate": 0.00011476093911162578,
      "loss": 1.9425,
      "step": 164010
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.2229655981063843,
      "learning_rate": 0.000114757721494666,
      "loss": 2.0024,
      "step": 164011
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.1318554878234863,
      "learning_rate": 0.00011475450391214604,
      "loss": 1.8983,
      "step": 164012
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.1644525527954102,
      "learning_rate": 0.0001147512863640668,
      "loss": 1.9746,
      "step": 164013
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.2656121253967285,
      "learning_rate": 0.00011474806885042872,
      "loss": 2.2434,
      "step": 164014
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.1968954801559448,
      "learning_rate": 0.00011474485137123258,
      "loss": 1.7745,
      "step": 164015
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.3071001768112183,
      "learning_rate": 0.00011474163392647877,
      "loss": 1.9232,
      "step": 164016
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.2941328287124634,
      "learning_rate": 0.00011473841651616806,
      "loss": 2.0769,
      "step": 164017
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.1694056987762451,
      "learning_rate": 0.00011473519914030096,
      "loss": 1.8413,
      "step": 164018
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.1981414556503296,
      "learning_rate": 0.00011473198179887811,
      "loss": 2.0905,
      "step": 164019
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.2714786529541016,
      "learning_rate": 0.00011472876449189998,
      "loss": 2.077,
      "step": 164020
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.199232578277588,
      "learning_rate": 0.00011472554721936736,
      "loss": 1.9871,
      "step": 164021
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.226493239402771,
      "learning_rate": 0.0001147223299812807,
      "loss": 1.9918,
      "step": 164022
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.2335208654403687,
      "learning_rate": 0.0001147191127776407,
      "loss": 2.2011,
      "step": 164023
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.1361942291259766,
      "learning_rate": 0.00011471589560844795,
      "loss": 1.9806,
      "step": 164024
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.24188232421875,
      "learning_rate": 0.00011471267847370301,
      "loss": 2.1017,
      "step": 164025
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.3068265914916992,
      "learning_rate": 0.00011470946137340635,
      "loss": 2.0179,
      "step": 164026
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.3128106594085693,
      "learning_rate": 0.00011470624430755884,
      "loss": 2.1343,
      "step": 164027
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.1061744689941406,
      "learning_rate": 0.0001147030272761608,
      "loss": 1.978,
      "step": 164028
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.2175171375274658,
      "learning_rate": 0.0001146998102792131,
      "loss": 1.9013,
      "step": 164029
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.2324378490447998,
      "learning_rate": 0.00011469659331671618,
      "loss": 1.8597,
      "step": 164030
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.2870876789093018,
      "learning_rate": 0.00011469337638867066,
      "loss": 2.0097,
      "step": 164031
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.328062891960144,
      "learning_rate": 0.00011469015949507706,
      "loss": 2.1257,
      "step": 164032
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.2122046947479248,
      "learning_rate": 0.00011468694263593612,
      "loss": 1.9943,
      "step": 164033
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.2017643451690674,
      "learning_rate": 0.0001146837258112483,
      "loss": 2.0829,
      "step": 164034
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.1601002216339111,
      "learning_rate": 0.00011468050902101439,
      "loss": 1.7813,
      "step": 164035
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.1590638160705566,
      "learning_rate": 0.00011467729226523484,
      "loss": 2.0018,
      "step": 164036
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.4596209526062012,
      "learning_rate": 0.0001146740755439103,
      "loss": 2.0116,
      "step": 164037
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.4262140989303589,
      "learning_rate": 0.00011467085885704124,
      "loss": 2.029,
      "step": 164038
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.434141755104065,
      "learning_rate": 0.00011466764220462847,
      "loss": 2.2826,
      "step": 164039
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.3290126323699951,
      "learning_rate": 0.00011466442558667239,
      "loss": 2.1351,
      "step": 164040
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.2699413299560547,
      "learning_rate": 0.00011466120900317376,
      "loss": 2.0191,
      "step": 164041
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.2606297731399536,
      "learning_rate": 0.00011465799245413314,
      "loss": 2.1045,
      "step": 164042
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.2699981927871704,
      "learning_rate": 0.00011465477593955109,
      "loss": 2.1332,
      "step": 164043
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.2367048263549805,
      "learning_rate": 0.00011465155945942811,
      "loss": 1.9551,
      "step": 164044
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.1630656719207764,
      "learning_rate": 0.00011464834301376499,
      "loss": 2.1773,
      "step": 164045
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.2418056726455688,
      "learning_rate": 0.00011464512660256219,
      "loss": 1.8221,
      "step": 164046
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.1623060703277588,
      "learning_rate": 0.0001146419102258204,
      "loss": 1.9926,
      "step": 164047
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.1711046695709229,
      "learning_rate": 0.00011463869388354023,
      "loss": 2.1173,
      "step": 164048
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.233229160308838,
      "learning_rate": 0.0001146354775757221,
      "loss": 2.2298,
      "step": 164049
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.2729952335357666,
      "learning_rate": 0.00011463226130236685,
      "loss": 2.0407,
      "step": 164050
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.1995030641555786,
      "learning_rate": 0.00011462904506347493,
      "loss": 2.1414,
      "step": 164051
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.3178945779800415,
      "learning_rate": 0.00011462582885904689,
      "loss": 2.1263,
      "step": 164052
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.221255898475647,
      "learning_rate": 0.00011462261268908351,
      "loss": 2.1552,
      "step": 164053
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.3650628328323364,
      "learning_rate": 0.00011461939655358528,
      "loss": 1.9909,
      "step": 164054
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.2299591302871704,
      "learning_rate": 0.00011461618045255274,
      "loss": 2.2359,
      "step": 164055
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.3835760354995728,
      "learning_rate": 0.0001146129643859866,
      "loss": 2.115,
      "step": 164056
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.290877342224121,
      "learning_rate": 0.00011460974835388731,
      "loss": 1.8535,
      "step": 164057
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.144715428352356,
      "learning_rate": 0.00011460653235625571,
      "loss": 2.0486,
      "step": 164058
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.1476185321807861,
      "learning_rate": 0.0001146033163930922,
      "loss": 2.0928,
      "step": 164059
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.2682808637619019,
      "learning_rate": 0.0001146001004643975,
      "loss": 1.8652,
      "step": 164060
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.2326526641845703,
      "learning_rate": 0.00011459688457017198,
      "loss": 2.2356,
      "step": 164061
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.305690884590149,
      "learning_rate": 0.00011459366871041651,
      "loss": 2.1866,
      "step": 164062
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.3281289339065552,
      "learning_rate": 0.0001145904528851315,
      "loss": 2.1538,
      "step": 164063
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.319280743598938,
      "learning_rate": 0.00011458723709431772,
      "loss": 1.9422,
      "step": 164064
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.0934808254241943,
      "learning_rate": 0.00011458402133797567,
      "loss": 1.9907,
      "step": 164065
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.269270896911621,
      "learning_rate": 0.00011458080561610595,
      "loss": 2.1324,
      "step": 164066
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.333174705505371,
      "learning_rate": 0.00011457758992870906,
      "loss": 2.0571,
      "step": 164067
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.224406123161316,
      "learning_rate": 0.00011457437427578578,
      "loss": 2.0755,
      "step": 164068
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.2071921825408936,
      "learning_rate": 0.00011457115865733653,
      "loss": 2.0308,
      "step": 164069
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.173783540725708,
      "learning_rate": 0.00011456794307336211,
      "loss": 1.9961,
      "step": 164070
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.2190864086151123,
      "learning_rate": 0.00011456472752386291,
      "loss": 1.8864,
      "step": 164071
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.1546909809112549,
      "learning_rate": 0.00011456151200883983,
      "loss": 1.9511,
      "step": 164072
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.3904463052749634,
      "learning_rate": 0.00011455829652829303,
      "loss": 2.1684,
      "step": 164073
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.2501424551010132,
      "learning_rate": 0.0001145550810822235,
      "loss": 2.0579,
      "step": 164074
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.1762316226959229,
      "learning_rate": 0.00011455186567063155,
      "loss": 1.9824,
      "step": 164075
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.2306082248687744,
      "learning_rate": 0.00011454865029351802,
      "loss": 2.0361,
      "step": 164076
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.1588327884674072,
      "learning_rate": 0.00011454543495088328,
      "loss": 1.9637,
      "step": 164077
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.3028792142868042,
      "learning_rate": 0.00011454221964272815,
      "loss": 2.0288,
      "step": 164078
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.2321076393127441,
      "learning_rate": 0.00011453900436905314,
      "loss": 2.2466,
      "step": 164079
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.2080864906311035,
      "learning_rate": 0.00011453578912985883,
      "loss": 2.0608,
      "step": 164080
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.2416789531707764,
      "learning_rate": 0.00011453257392514568,
      "loss": 2.1254,
      "step": 164081
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.4164958000183105,
      "learning_rate": 0.00011452935875491453,
      "loss": 2.3131,
      "step": 164082
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.1550036668777466,
      "learning_rate": 0.0001145261436191658,
      "loss": 2.2434,
      "step": 164083
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.284844160079956,
      "learning_rate": 0.00011452292851790025,
      "loss": 2.2255,
      "step": 164084
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.3785783052444458,
      "learning_rate": 0.00011451971345111841,
      "loss": 2.2641,
      "step": 164085
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.1827861070632935,
      "learning_rate": 0.00011451649841882082,
      "loss": 2.0154,
      "step": 164086
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.2433031797409058,
      "learning_rate": 0.00011451328342100805,
      "loss": 2.1134,
      "step": 164087
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.2317023277282715,
      "learning_rate": 0.00011451006845768086,
      "loss": 2.1384,
      "step": 164088
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.1394492387771606,
      "learning_rate": 0.00011450685352883961,
      "loss": 2.0157,
      "step": 164089
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.1892977952957153,
      "learning_rate": 0.00011450363863448515,
      "loss": 2.0703,
      "step": 164090
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.3690763711929321,
      "learning_rate": 0.000114500423774618,
      "loss": 2.0138,
      "step": 164091
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.3042782545089722,
      "learning_rate": 0.0001144972089492387,
      "loss": 2.1543,
      "step": 164092
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.2999500036239624,
      "learning_rate": 0.00011449399415834777,
      "loss": 2.0685,
      "step": 164093
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.1941553354263306,
      "learning_rate": 0.00011449077940194601,
      "loss": 2.0657,
      "step": 164094
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.33687424659729,
      "learning_rate": 0.00011448756468003381,
      "loss": 2.1175,
      "step": 164095
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.1911827325820923,
      "learning_rate": 0.000114484349992612,
      "loss": 2.0578,
      "step": 164096
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.2794610261917114,
      "learning_rate": 0.00011448113533968101,
      "loss": 2.0152,
      "step": 164097
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.2892335653305054,
      "learning_rate": 0.0001144779207212415,
      "loss": 2.0611,
      "step": 164098
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.2808033227920532,
      "learning_rate": 0.00011447470613729393,
      "loss": 1.9773,
      "step": 164099
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.305942416191101,
      "learning_rate": 0.00011447149158783914,
      "loss": 2.1305,
      "step": 164100
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.2697786092758179,
      "learning_rate": 0.00011446827707287746,
      "loss": 2.1258,
      "step": 164101
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.2772523164749146,
      "learning_rate": 0.0001144650625924098,
      "loss": 1.8488,
      "step": 164102
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.337534785270691,
      "learning_rate": 0.00011446184814643653,
      "loss": 2.303,
      "step": 164103
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.155883550643921,
      "learning_rate": 0.00011445863373495834,
      "loss": 1.9412,
      "step": 164104
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.163921594619751,
      "learning_rate": 0.00011445541935797565,
      "loss": 2.0482,
      "step": 164105
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.1961387395858765,
      "learning_rate": 0.00011445220501548933,
      "loss": 2.1064,
      "step": 164106
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.26370370388031,
      "learning_rate": 0.00011444899070749973,
      "loss": 2.0392,
      "step": 164107
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.3667162656784058,
      "learning_rate": 0.0001144457764340077,
      "loss": 1.8124,
      "step": 164108
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.4004430770874023,
      "learning_rate": 0.00011444256219501366,
      "loss": 1.9395,
      "step": 164109
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.2711964845657349,
      "learning_rate": 0.00011443934799051827,
      "loss": 2.0156,
      "step": 164110
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.1736664772033691,
      "learning_rate": 0.00011443613382052202,
      "loss": 1.8102,
      "step": 164111
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.3332616090774536,
      "learning_rate": 0.00011443291968502569,
      "loss": 1.9456,
      "step": 164112
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.4188497066497803,
      "learning_rate": 0.00011442970558402968,
      "loss": 2.241,
      "step": 164113
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.2283599376678467,
      "learning_rate": 0.00011442649151753475,
      "loss": 2.1534,
      "step": 164114
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.2615656852722168,
      "learning_rate": 0.00011442327748554151,
      "loss": 2.2021,
      "step": 164115
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.334675669670105,
      "learning_rate": 0.00011442006348805043,
      "loss": 2.0948,
      "step": 164116
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.314321756362915,
      "learning_rate": 0.00011441684952506208,
      "loss": 1.994,
      "step": 164117
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.244042158126831,
      "learning_rate": 0.00011441363559657727,
      "loss": 2.1235,
      "step": 164118
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.2050927877426147,
      "learning_rate": 0.00011441042170259632,
      "loss": 1.9117,
      "step": 164119
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.2427760362625122,
      "learning_rate": 0.0001144072078431201,
      "loss": 2.1232,
      "step": 164120
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.3023937940597534,
      "learning_rate": 0.00011440399401814909,
      "loss": 2.0808,
      "step": 164121
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.1911591291427612,
      "learning_rate": 0.00011440078022768387,
      "loss": 2.092,
      "step": 164122
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.250677466392517,
      "learning_rate": 0.00011439756647172495,
      "loss": 1.8876,
      "step": 164123
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.3887622356414795,
      "learning_rate": 0.00011439435275027313,
      "loss": 2.285,
      "step": 164124
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.2363721132278442,
      "learning_rate": 0.0001143911390633288,
      "loss": 2.0672,
      "step": 164125
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.249353051185608,
      "learning_rate": 0.00011438792541089277,
      "loss": 1.9811,
      "step": 164126
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.2810925245285034,
      "learning_rate": 0.0001143847117929655,
      "loss": 1.9694,
      "step": 164127
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.2908241748809814,
      "learning_rate": 0.00011438149820954767,
      "loss": 2.0258,
      "step": 164128
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.417784333229065,
      "learning_rate": 0.00011437828466063966,
      "loss": 2.0068,
      "step": 164129
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.4367321729660034,
      "learning_rate": 0.00011437507114624238,
      "loss": 2.0061,
      "step": 164130
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.39176344871521,
      "learning_rate": 0.00011437185766635615,
      "loss": 2.067,
      "step": 164131
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.4523659944534302,
      "learning_rate": 0.0001143686442209818,
      "loss": 2.0261,
      "step": 164132
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.2850170135498047,
      "learning_rate": 0.00011436543081011983,
      "loss": 2.1244,
      "step": 164133
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.2449724674224854,
      "learning_rate": 0.00011436221743377072,
      "loss": 2.2214,
      "step": 164134
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.148235559463501,
      "learning_rate": 0.00011435900409193531,
      "loss": 1.9033,
      "step": 164135
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.3662426471710205,
      "learning_rate": 0.00011435579078461405,
      "loss": 1.9789,
      "step": 164136
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.331294059753418,
      "learning_rate": 0.00011435257751180741,
      "loss": 2.1805,
      "step": 164137
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.1408582925796509,
      "learning_rate": 0.00011434936427351626,
      "loss": 2.1819,
      "step": 164138
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.21476411819458,
      "learning_rate": 0.00011434615106974107,
      "loss": 1.8162,
      "step": 164139
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.3764580488204956,
      "learning_rate": 0.00011434293790048235,
      "loss": 1.9536,
      "step": 164140
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.437548041343689,
      "learning_rate": 0.00011433972476574088,
      "loss": 2.1507,
      "step": 164141
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.2790107727050781,
      "learning_rate": 0.00011433651166551715,
      "loss": 2.1003,
      "step": 164142
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.190569281578064,
      "learning_rate": 0.00011433329859981166,
      "loss": 2.2103,
      "step": 164143
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.3658114671707153,
      "learning_rate": 0.00011433008556862521,
      "loss": 1.977,
      "step": 164144
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.2679049968719482,
      "learning_rate": 0.00011432687257195821,
      "loss": 1.7925,
      "step": 164145
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.2718758583068848,
      "learning_rate": 0.00011432365960981148,
      "loss": 1.945,
      "step": 164146
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.1423579454421997,
      "learning_rate": 0.00011432044668218542,
      "loss": 2.0735,
      "step": 164147
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.1399508714675903,
      "learning_rate": 0.00011431723378908066,
      "loss": 2.1394,
      "step": 164148
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.4162349700927734,
      "learning_rate": 0.00011431402093049788,
      "loss": 1.9135,
      "step": 164149
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.3944638967514038,
      "learning_rate": 0.00011431080810643769,
      "loss": 1.9515,
      "step": 164150
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.259652018547058,
      "learning_rate": 0.00011430759531690047,
      "loss": 2.0075,
      "step": 164151
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.2247892618179321,
      "learning_rate": 0.00011430438256188708,
      "loss": 1.8639,
      "step": 164152
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.303488850593567,
      "learning_rate": 0.00011430116984139804,
      "loss": 2.0841,
      "step": 164153
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.2215561866760254,
      "learning_rate": 0.00011429795715543379,
      "loss": 2.1314,
      "step": 164154
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.2590845823287964,
      "learning_rate": 0.00011429474450399517,
      "loss": 1.9579,
      "step": 164155
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.1937941312789917,
      "learning_rate": 0.00011429153188708255,
      "loss": 1.8615,
      "step": 164156
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.3191449642181396,
      "learning_rate": 0.00011428831930469677,
      "loss": 2.01,
      "step": 164157
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.197967767715454,
      "learning_rate": 0.00011428510675683827,
      "loss": 1.91,
      "step": 164158
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.2698907852172852,
      "learning_rate": 0.00011428189424350768,
      "loss": 2.2933,
      "step": 164159
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.2300301790237427,
      "learning_rate": 0.00011427868176470547,
      "loss": 1.8212,
      "step": 164160
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.2966620922088623,
      "learning_rate": 0.00011427546932043248,
      "loss": 1.9819,
      "step": 164161
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.1326624155044556,
      "learning_rate": 0.0001142722569106891,
      "loss": 2.0858,
      "step": 164162
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.1608630418777466,
      "learning_rate": 0.0001142690445354761,
      "loss": 2.0251,
      "step": 164163
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.3646576404571533,
      "learning_rate": 0.00011426583219479397,
      "loss": 2.1155,
      "step": 164164
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.1804479360580444,
      "learning_rate": 0.00011426261988864336,
      "loss": 1.8623,
      "step": 164165
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.1820919513702393,
      "learning_rate": 0.00011425940761702472,
      "loss": 2.0612,
      "step": 164166
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.1636685132980347,
      "learning_rate": 0.00011425619537993887,
      "loss": 2.2064,
      "step": 164167
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.2761551141738892,
      "learning_rate": 0.00011425298317738619,
      "loss": 2.0792,
      "step": 164168
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.2359589338302612,
      "learning_rate": 0.00011424977100936747,
      "loss": 2.1899,
      "step": 164169
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.1739150285720825,
      "learning_rate": 0.00011424655887588324,
      "loss": 2.0998,
      "step": 164170
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.4924404621124268,
      "learning_rate": 0.00011424334677693412,
      "loss": 2.2024,
      "step": 164171
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.3580727577209473,
      "learning_rate": 0.00011424013471252051,
      "loss": 2.3625,
      "step": 164172
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.299149513244629,
      "learning_rate": 0.0001142369226826433,
      "loss": 2.2514,
      "step": 164173
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.291245698928833,
      "learning_rate": 0.00011423371068730283,
      "loss": 2.2448,
      "step": 164174
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.5275932550430298,
      "learning_rate": 0.00011423049872649993,
      "loss": 1.9737,
      "step": 164175
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.2298073768615723,
      "learning_rate": 0.00011422728680023511,
      "loss": 1.9939,
      "step": 164176
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.151038408279419,
      "learning_rate": 0.00011422407490850892,
      "loss": 2.1464,
      "step": 164177
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.2153269052505493,
      "learning_rate": 0.00011422086305132187,
      "loss": 1.8802,
      "step": 164178
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.2198508977890015,
      "learning_rate": 0.00011421765122867477,
      "loss": 2.1681,
      "step": 164179
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.3993594646453857,
      "learning_rate": 0.00011421443944056803,
      "loss": 2.1809,
      "step": 164180
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.297387957572937,
      "learning_rate": 0.00011421122768700243,
      "loss": 2.1007,
      "step": 164181
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.2103586196899414,
      "learning_rate": 0.0001142080159679785,
      "loss": 2.2401,
      "step": 164182
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.4075802564620972,
      "learning_rate": 0.00011420480428349677,
      "loss": 2.1122,
      "step": 164183
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.352946162223816,
      "learning_rate": 0.00011420159263355777,
      "loss": 2.0573,
      "step": 164184
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.3443031311035156,
      "learning_rate": 0.00011419838101816231,
      "loss": 2.1025,
      "step": 164185
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.3900673389434814,
      "learning_rate": 0.00011419516943731078,
      "loss": 2.189,
      "step": 164186
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.2654073238372803,
      "learning_rate": 0.00011419195789100396,
      "loss": 1.9933,
      "step": 164187
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.2556145191192627,
      "learning_rate": 0.0001141887463792424,
      "loss": 1.9053,
      "step": 164188
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.4147937297821045,
      "learning_rate": 0.00011418553490202665,
      "loss": 2.3205,
      "step": 164189
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.2520701885223389,
      "learning_rate": 0.00011418232345935718,
      "loss": 2.2377,
      "step": 164190
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.2244516611099243,
      "learning_rate": 0.00011417911205123486,
      "loss": 1.8565,
      "step": 164191
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.195016860961914,
      "learning_rate": 0.00011417590067766003,
      "loss": 2.1622,
      "step": 164192
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.1303282976150513,
      "learning_rate": 0.00011417268933863353,
      "loss": 1.8669,
      "step": 164193
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.1934484243392944,
      "learning_rate": 0.00011416947803415582,
      "loss": 2.2607,
      "step": 164194
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.2670637369155884,
      "learning_rate": 0.00011416626676422749,
      "loss": 2.0637,
      "step": 164195
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.1792926788330078,
      "learning_rate": 0.00011416305552884908,
      "loss": 2.2136,
      "step": 164196
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.3428525924682617,
      "learning_rate": 0.00011415984432802136,
      "loss": 2.1308,
      "step": 164197
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.330000638961792,
      "learning_rate": 0.00011415663316174475,
      "loss": 1.9942,
      "step": 164198
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.3556171655654907,
      "learning_rate": 0.00011415342203002002,
      "loss": 2.0837,
      "step": 164199
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.3055061101913452,
      "learning_rate": 0.0001141502109328477,
      "loss": 2.1865,
      "step": 164200
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.1307446956634521,
      "learning_rate": 0.00011414699987022832,
      "loss": 2.1474,
      "step": 164201
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.1665971279144287,
      "learning_rate": 0.00011414378884216246,
      "loss": 1.9845,
      "step": 164202
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.187209963798523,
      "learning_rate": 0.00011414057784865087,
      "loss": 2.2907,
      "step": 164203
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.2758709192276,
      "learning_rate": 0.00011413736688969391,
      "loss": 2.1446,
      "step": 164204
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.2640810012817383,
      "learning_rate": 0.00011413415596529248,
      "loss": 2.1521,
      "step": 164205
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.2232246398925781,
      "learning_rate": 0.00011413094507544701,
      "loss": 1.89,
      "step": 164206
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.1242986917495728,
      "learning_rate": 0.00011412773422015806,
      "loss": 1.8138,
      "step": 164207
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.3208789825439453,
      "learning_rate": 0.00011412452339942621,
      "loss": 2.2016,
      "step": 164208
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.2313727140426636,
      "learning_rate": 0.00011412131261325223,
      "loss": 1.9588,
      "step": 164209
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.1428300142288208,
      "learning_rate": 0.0001141181018616365,
      "loss": 1.9365,
      "step": 164210
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.2549062967300415,
      "learning_rate": 0.0001141148911445798,
      "loss": 2.1967,
      "step": 164211
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.2903234958648682,
      "learning_rate": 0.00011411168046208256,
      "loss": 2.0535,
      "step": 164212
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.444273591041565,
      "learning_rate": 0.00011410846981414569,
      "loss": 2.017,
      "step": 164213
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.2473270893096924,
      "learning_rate": 0.00011410525920076933,
      "loss": 2.1661,
      "step": 164214
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.2107861042022705,
      "learning_rate": 0.00011410204862195442,
      "loss": 1.9066,
      "step": 164215
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.200632929801941,
      "learning_rate": 0.00011409883807770137,
      "loss": 2.0366,
      "step": 164216
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.3258023262023926,
      "learning_rate": 0.00011409562756801094,
      "loss": 2.0138,
      "step": 164217
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.231679916381836,
      "learning_rate": 0.00011409241709288356,
      "loss": 2.0596,
      "step": 164218
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.269660472869873,
      "learning_rate": 0.00011408920665232,
      "loss": 2.2098,
      "step": 164219
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.1802141666412354,
      "learning_rate": 0.00011408599624632075,
      "loss": 2.0016,
      "step": 164220
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.4420628547668457,
      "learning_rate": 0.00011408278587488643,
      "loss": 2.1829,
      "step": 164221
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.2651124000549316,
      "learning_rate": 0.00011407957553801752,
      "loss": 2.039,
      "step": 164222
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.5904788970947266,
      "learning_rate": 0.00011407636523571486,
      "loss": 2.0052,
      "step": 164223
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.1970653533935547,
      "learning_rate": 0.00011407315496797879,
      "loss": 1.9777,
      "step": 164224
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.5176845788955688,
      "learning_rate": 0.00011406994473481013,
      "loss": 1.8,
      "step": 164225
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.4894850254058838,
      "learning_rate": 0.00011406673453620938,
      "loss": 2.0127,
      "step": 164226
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.2476367950439453,
      "learning_rate": 0.00011406352437217712,
      "loss": 1.9855,
      "step": 164227
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.1968209743499756,
      "learning_rate": 0.00011406031424271387,
      "loss": 2.1326,
      "step": 164228
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.2141616344451904,
      "learning_rate": 0.00011405710414782044,
      "loss": 1.9666,
      "step": 164229
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.5116024017333984,
      "learning_rate": 0.00011405389408749718,
      "loss": 2.075,
      "step": 164230
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.408639669418335,
      "learning_rate": 0.0001140506840617449,
      "loss": 1.838,
      "step": 164231
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.14224374294281,
      "learning_rate": 0.00011404747407056414,
      "loss": 2.0131,
      "step": 164232
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.218239188194275,
      "learning_rate": 0.00011404426411395532,
      "loss": 2.0522,
      "step": 164233
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.2444995641708374,
      "learning_rate": 0.00011404105419191932,
      "loss": 2.0884,
      "step": 164234
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.2039859294891357,
      "learning_rate": 0.00011403784430445658,
      "loss": 2.0103,
      "step": 164235
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.3908841609954834,
      "learning_rate": 0.00011403463445156763,
      "loss": 1.9534,
      "step": 164236
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.3760626316070557,
      "learning_rate": 0.00011403142463325325,
      "loss": 2.244,
      "step": 164237
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.202232837677002,
      "learning_rate": 0.00011402821484951393,
      "loss": 1.9027,
      "step": 164238
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.231429934501648,
      "learning_rate": 0.00011402500510035016,
      "loss": 2.1472,
      "step": 164239
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.3749020099639893,
      "learning_rate": 0.00011402179538576278,
      "loss": 2.2816,
      "step": 164240
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.2327100038528442,
      "learning_rate": 0.00011401858570575227,
      "loss": 1.7886,
      "step": 164241
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.2162463665008545,
      "learning_rate": 0.00011401537606031907,
      "loss": 1.8832,
      "step": 164242
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.1818009614944458,
      "learning_rate": 0.00011401216644946405,
      "loss": 2.0503,
      "step": 164243
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.3137311935424805,
      "learning_rate": 0.0001140089568731877,
      "loss": 1.9483,
      "step": 164244
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.445197343826294,
      "learning_rate": 0.00011400574733149045,
      "loss": 1.9266,
      "step": 164245
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.221331238746643,
      "learning_rate": 0.00011400253782437317,
      "loss": 1.942,
      "step": 164246
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.3588274717330933,
      "learning_rate": 0.00011399932835183624,
      "loss": 2.1163,
      "step": 164247
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.188848853111267,
      "learning_rate": 0.00011399611891388041,
      "loss": 1.9339,
      "step": 164248
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.2382838726043701,
      "learning_rate": 0.00011399290951050624,
      "loss": 2.075,
      "step": 164249
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.2256479263305664,
      "learning_rate": 0.0001139897001417143,
      "loss": 1.9843,
      "step": 164250
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.2558237314224243,
      "learning_rate": 0.00011398649080750507,
      "loss": 2.1037,
      "step": 164251
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.2390923500061035,
      "learning_rate": 0.00011398328150787938,
      "loss": 1.8889,
      "step": 164252
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.230324387550354,
      "learning_rate": 0.00011398007224283761,
      "loss": 2.139,
      "step": 164253
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.3397986888885498,
      "learning_rate": 0.00011397686301238052,
      "loss": 2.1622,
      "step": 164254
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.2243975400924683,
      "learning_rate": 0.00011397365381650868,
      "loss": 1.9792,
      "step": 164255
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.2193681001663208,
      "learning_rate": 0.00011397044465522263,
      "loss": 1.9913,
      "step": 164256
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.2003856897354126,
      "learning_rate": 0.00011396723552852289,
      "loss": 1.9706,
      "step": 164257
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.2496345043182373,
      "learning_rate": 0.00011396402643641025,
      "loss": 2.0431,
      "step": 164258
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.2470558881759644,
      "learning_rate": 0.00011396081737888514,
      "loss": 2.0491,
      "step": 164259
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.1443078517913818,
      "learning_rate": 0.00011395760835594829,
      "loss": 2.2014,
      "step": 164260
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.2864437103271484,
      "learning_rate": 0.00011395439936760024,
      "loss": 1.9419,
      "step": 164261
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.1549841165542603,
      "learning_rate": 0.0001139511904138416,
      "loss": 1.9375,
      "step": 164262
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.2334471940994263,
      "learning_rate": 0.0001139479814946728,
      "loss": 2.1241,
      "step": 164263
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.2106949090957642,
      "learning_rate": 0.00011394477261009472,
      "loss": 1.8134,
      "step": 164264
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.3459911346435547,
      "learning_rate": 0.0001139415637601077,
      "loss": 2.2263,
      "step": 164265
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.2709367275238037,
      "learning_rate": 0.00011393835494471257,
      "loss": 2.1988,
      "step": 164266
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.3677339553833008,
      "learning_rate": 0.0001139351461639098,
      "loss": 2.1531,
      "step": 164267
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.3015027046203613,
      "learning_rate": 0.00011393193741770004,
      "loss": 2.0028,
      "step": 164268
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.3942699432373047,
      "learning_rate": 0.0001139287287060837,
      "loss": 1.8283,
      "step": 164269
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.2059884071350098,
      "learning_rate": 0.0001139255200290616,
      "loss": 1.9367,
      "step": 164270
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.1565382480621338,
      "learning_rate": 0.00011392231138663422,
      "loss": 2.0234,
      "step": 164271
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.1484637260437012,
      "learning_rate": 0.00011391910277880227,
      "loss": 1.9238,
      "step": 164272
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.1771358251571655,
      "learning_rate": 0.00011391589420556627,
      "loss": 1.8481,
      "step": 164273
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.2372506856918335,
      "learning_rate": 0.00011391268566692682,
      "loss": 2.0607,
      "step": 164274
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.4559537172317505,
      "learning_rate": 0.0001139094771628844,
      "loss": 2.2493,
      "step": 164275
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.3114696741104126,
      "learning_rate": 0.0001139062686934398,
      "loss": 2.0403,
      "step": 164276
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.4860455989837646,
      "learning_rate": 0.00011390306025859347,
      "loss": 2.0727,
      "step": 164277
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.1547027826309204,
      "learning_rate": 0.0001138998518583462,
      "loss": 2.0516,
      "step": 164278
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.2137043476104736,
      "learning_rate": 0.0001138966434926983,
      "loss": 2.0153,
      "step": 164279
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.2829915285110474,
      "learning_rate": 0.00011389343516165077,
      "loss": 2.2036,
      "step": 164280
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.2616941928863525,
      "learning_rate": 0.00011389022686520376,
      "loss": 2.0366,
      "step": 164281
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.2476439476013184,
      "learning_rate": 0.00011388701860335817,
      "loss": 2.0563,
      "step": 164282
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.234004020690918,
      "learning_rate": 0.00011388381037611436,
      "loss": 2.0309,
      "step": 164283
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.2177021503448486,
      "learning_rate": 0.00011388060218347323,
      "loss": 2.043,
      "step": 164284
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.255426049232483,
      "learning_rate": 0.00011387739402543506,
      "loss": 1.9659,
      "step": 164285
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.1953822374343872,
      "learning_rate": 0.0001138741859020008,
      "loss": 2.063,
      "step": 164286
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.1287897825241089,
      "learning_rate": 0.00011387097781317064,
      "loss": 1.9012,
      "step": 164287
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.3338263034820557,
      "learning_rate": 0.00011386776975894551,
      "loss": 2.1167,
      "step": 164288
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.2313661575317383,
      "learning_rate": 0.00011386456173932572,
      "loss": 2.0772,
      "step": 164289
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.1570992469787598,
      "learning_rate": 0.00011386135375431215,
      "loss": 2.1063,
      "step": 164290
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.2264630794525146,
      "learning_rate": 0.00011385814580390517,
      "loss": 1.8714,
      "step": 164291
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.2721647024154663,
      "learning_rate": 0.00011385493788810572,
      "loss": 2.4493,
      "step": 164292
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.1107748746871948,
      "learning_rate": 0.0001138517300069139,
      "loss": 2.1671,
      "step": 164293
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.198175072669983,
      "learning_rate": 0.00011384852216033064,
      "loss": 1.9631,
      "step": 164294
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.3482062816619873,
      "learning_rate": 0.00011384531434835639,
      "loss": 2.0183,
      "step": 164295
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.3159551620483398,
      "learning_rate": 0.00011384210657099188,
      "loss": 2.2216,
      "step": 164296
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.2659317255020142,
      "learning_rate": 0.00011383889882823756,
      "loss": 2.0567,
      "step": 164297
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.1956194639205933,
      "learning_rate": 0.0001138356911200943,
      "loss": 1.9672,
      "step": 164298
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.0852406024932861,
      "learning_rate": 0.0001138324834465623,
      "loss": 1.9323,
      "step": 164299
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.5207451581954956,
      "learning_rate": 0.00011382927580764246,
      "loss": 1.8702,
      "step": 164300
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.3927572965621948,
      "learning_rate": 0.00011382606820333517,
      "loss": 2.1133,
      "step": 164301
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.327260136604309,
      "learning_rate": 0.00011382286063364121,
      "loss": 2.172,
      "step": 164302
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.2065794467926025,
      "learning_rate": 0.000113819653098561,
      "loss": 2.0677,
      "step": 164303
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.3317127227783203,
      "learning_rate": 0.00011381644559809545,
      "loss": 2.2489,
      "step": 164304
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.1562267541885376,
      "learning_rate": 0.00011381323813224472,
      "loss": 1.9762,
      "step": 164305
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.142441749572754,
      "learning_rate": 0.00011381003070100971,
      "loss": 1.9763,
      "step": 164306
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.353194236755371,
      "learning_rate": 0.00011380682330439085,
      "loss": 1.9119,
      "step": 164307
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.5113155841827393,
      "learning_rate": 0.00011380361594238892,
      "loss": 2.0946,
      "step": 164308
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.2963683605194092,
      "learning_rate": 0.00011380040861500431,
      "loss": 1.937,
      "step": 164309
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.2228933572769165,
      "learning_rate": 0.0001137972013222378,
      "loss": 2.163,
      "step": 164310
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.2985539436340332,
      "learning_rate": 0.00011379399406408994,
      "loss": 2.2339,
      "step": 164311
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.4056587219238281,
      "learning_rate": 0.00011379078684056129,
      "loss": 2.0371,
      "step": 164312
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.3041623830795288,
      "learning_rate": 0.0001137875796516523,
      "loss": 2.0639,
      "step": 164313
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.3257050514221191,
      "learning_rate": 0.00011378437249736387,
      "loss": 1.7908,
      "step": 164314
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.21409010887146,
      "learning_rate": 0.00011378116537769629,
      "loss": 2.0136,
      "step": 164315
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.2916247844696045,
      "learning_rate": 0.00011377795829265042,
      "loss": 2.0958,
      "step": 164316
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.1327848434448242,
      "learning_rate": 0.00011377475124222676,
      "loss": 2.0327,
      "step": 164317
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.2922544479370117,
      "learning_rate": 0.00011377154422642578,
      "loss": 2.0276,
      "step": 164318
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.4108301401138306,
      "learning_rate": 0.0001137683372452483,
      "loss": 2.0947,
      "step": 164319
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.3874117136001587,
      "learning_rate": 0.00011376513029869479,
      "loss": 2.1187,
      "step": 164320
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.3302420377731323,
      "learning_rate": 0.00011376192338676574,
      "loss": 1.9921,
      "step": 164321
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.2354987859725952,
      "learning_rate": 0.000113758716509462,
      "loss": 1.8499,
      "step": 164322
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.4357365369796753,
      "learning_rate": 0.000113755509666784,
      "loss": 1.7904,
      "step": 164323
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.3271305561065674,
      "learning_rate": 0.00011375230285873229,
      "loss": 1.9801,
      "step": 164324
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.3241064548492432,
      "learning_rate": 0.00011374909608530762,
      "loss": 1.7958,
      "step": 164325
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.2191065549850464,
      "learning_rate": 0.00011374588934651053,
      "loss": 2.1133,
      "step": 164326
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.1960252523422241,
      "learning_rate": 0.00011374268264234148,
      "loss": 1.9374,
      "step": 164327
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.2975391149520874,
      "learning_rate": 0.00011373947597280128,
      "loss": 1.7697,
      "step": 164328
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.3118153810501099,
      "learning_rate": 0.00011373626933789045,
      "loss": 1.6921,
      "step": 164329
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.1936343908309937,
      "learning_rate": 0.00011373306273760941,
      "loss": 2.0785,
      "step": 164330
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.2934991121292114,
      "learning_rate": 0.00011372985617195908,
      "loss": 1.9198,
      "step": 164331
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.1512384414672852,
      "learning_rate": 0.00011372664964093973,
      "loss": 2.1878,
      "step": 164332
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.1423962116241455,
      "learning_rate": 0.00011372344314455223,
      "loss": 1.8398,
      "step": 164333
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.3310331106185913,
      "learning_rate": 0.00011372023668279709,
      "loss": 1.8177,
      "step": 164334
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.265727162361145,
      "learning_rate": 0.00011371703025567483,
      "loss": 2.0578,
      "step": 164335
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.2033538818359375,
      "learning_rate": 0.00011371382386318602,
      "loss": 2.1006,
      "step": 164336
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.2427762746810913,
      "learning_rate": 0.0001137106175053314,
      "loss": 1.9347,
      "step": 164337
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.3630917072296143,
      "learning_rate": 0.00011370741118211139,
      "loss": 2.0837,
      "step": 164338
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.447205901145935,
      "learning_rate": 0.00011370420489352684,
      "loss": 1.8103,
      "step": 164339
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.1969151496887207,
      "learning_rate": 0.00011370099863957812,
      "loss": 1.8731,
      "step": 164340
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.1469639539718628,
      "learning_rate": 0.00011369779242026598,
      "loss": 2.0358,
      "step": 164341
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.2478832006454468,
      "learning_rate": 0.00011369458623559079,
      "loss": 2.0303,
      "step": 164342
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.4445819854736328,
      "learning_rate": 0.0001136913800855534,
      "loss": 2.0397,
      "step": 164343
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.2126221656799316,
      "learning_rate": 0.0001136881739701542,
      "loss": 2.2016,
      "step": 164344
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.1158137321472168,
      "learning_rate": 0.00011368496788939399,
      "loss": 1.9876,
      "step": 164345
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.466062307357788,
      "learning_rate": 0.00011368176184327313,
      "loss": 2.3219,
      "step": 164346
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.3863800764083862,
      "learning_rate": 0.00011367855583179257,
      "loss": 2.1096,
      "step": 164347
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.3880847692489624,
      "learning_rate": 0.0001136753498549525,
      "loss": 1.8627,
      "step": 164348
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.2847833633422852,
      "learning_rate": 0.00011367214391275379,
      "loss": 2.3073,
      "step": 164349
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.151703953742981,
      "learning_rate": 0.00011366893800519683,
      "loss": 1.9684,
      "step": 164350
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.4342424869537354,
      "learning_rate": 0.00011366573213228246,
      "loss": 2.0629,
      "step": 164351
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.1738356351852417,
      "learning_rate": 0.00011366252629401101,
      "loss": 2.1941,
      "step": 164352
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.3067244291305542,
      "learning_rate": 0.00011365932049038345,
      "loss": 2.1563,
      "step": 164353
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.2228143215179443,
      "learning_rate": 0.00011365611472139991,
      "loss": 2.1451,
      "step": 164354
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.142961859703064,
      "learning_rate": 0.00011365290898706136,
      "loss": 2.0524,
      "step": 164355
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.2650023698806763,
      "learning_rate": 0.0001136497032873681,
      "loss": 2.2257,
      "step": 164356
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.2821893692016602,
      "learning_rate": 0.00011364649762232102,
      "loss": 1.9387,
      "step": 164357
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.1914408206939697,
      "learning_rate": 0.00011364329199192047,
      "loss": 2.0827,
      "step": 164358
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.2593159675598145,
      "learning_rate": 0.00011364008639616732,
      "loss": 2.0308,
      "step": 164359
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.1862998008728027,
      "learning_rate": 0.0001136368808350618,
      "loss": 1.8013,
      "step": 164360
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.2839521169662476,
      "learning_rate": 0.00011363367530860483,
      "loss": 2.3063,
      "step": 164361
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.1056318283081055,
      "learning_rate": 0.00011363046981679673,
      "loss": 1.8418,
      "step": 164362
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.1426987648010254,
      "learning_rate": 0.00011362726435963839,
      "loss": 2.116,
      "step": 164363
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.3551082611083984,
      "learning_rate": 0.00011362405893713014,
      "loss": 1.8194,
      "step": 164364
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.2438105344772339,
      "learning_rate": 0.0001136208535492729,
      "loss": 2.0033,
      "step": 164365
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.3148138523101807,
      "learning_rate": 0.00011361764819606687,
      "loss": 1.9969,
      "step": 164366
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.202699065208435,
      "learning_rate": 0.00011361444287751293,
      "loss": 1.9746,
      "step": 164367
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.2439442873001099,
      "learning_rate": 0.00011361123759361146,
      "loss": 2.0712,
      "step": 164368
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.2781602144241333,
      "learning_rate": 0.00011360803234436331,
      "loss": 2.1814,
      "step": 164369
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.269179105758667,
      "learning_rate": 0.00011360482712976884,
      "loss": 2.1617,
      "step": 164370
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.2467341423034668,
      "learning_rate": 0.00011360162194982894,
      "loss": 2.3065,
      "step": 164371
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.2443227767944336,
      "learning_rate": 0.00011359841680454383,
      "loss": 2.1289,
      "step": 164372
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.2422534227371216,
      "learning_rate": 0.0001135952116939144,
      "loss": 2.069,
      "step": 164373
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.2297674417495728,
      "learning_rate": 0.00011359200661794101,
      "loss": 2.0116,
      "step": 164374
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.2897945642471313,
      "learning_rate": 0.0001135888015766245,
      "loss": 1.8614,
      "step": 164375
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.3262425661087036,
      "learning_rate": 0.00011358559656996527,
      "loss": 2.2147,
      "step": 164376
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.1642706394195557,
      "learning_rate": 0.00011358239159796418,
      "loss": 2.0139,
      "step": 164377
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.3755775690078735,
      "learning_rate": 0.00011357918666062143,
      "loss": 2.0921,
      "step": 164378
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.2175136804580688,
      "learning_rate": 0.00011357598175793792,
      "loss": 1.8906,
      "step": 164379
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.1283464431762695,
      "learning_rate": 0.00011357277688991408,
      "loss": 2.0967,
      "step": 164380
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.139338493347168,
      "learning_rate": 0.00011356957205655066,
      "loss": 1.8524,
      "step": 164381
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.2724732160568237,
      "learning_rate": 0.00011356636725784808,
      "loss": 1.917,
      "step": 164382
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.2847329378128052,
      "learning_rate": 0.00011356316249380727,
      "loss": 1.9738,
      "step": 164383
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.2431689500808716,
      "learning_rate": 0.00011355995776442833,
      "loss": 1.9749,
      "step": 164384
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.2287613153457642,
      "learning_rate": 0.00011355675306971227,
      "loss": 2.0356,
      "step": 164385
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.3175221681594849,
      "learning_rate": 0.00011355354840965937,
      "loss": 1.9004,
      "step": 164386
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.223132848739624,
      "learning_rate": 0.00011355034378427052,
      "loss": 1.9205,
      "step": 164387
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.3333202600479126,
      "learning_rate": 0.00011354713919354608,
      "loss": 2.0112,
      "step": 164388
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.2003098726272583,
      "learning_rate": 0.00011354393463748693,
      "loss": 2.0499,
      "step": 164389
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.3153133392333984,
      "learning_rate": 0.00011354073011609329,
      "loss": 2.1666,
      "step": 164390
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.1547778844833374,
      "learning_rate": 0.00011353752562936605,
      "loss": 2.1768,
      "step": 164391
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.300389051437378,
      "learning_rate": 0.00011353432117730562,
      "loss": 2.0611,
      "step": 164392
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.3238056898117065,
      "learning_rate": 0.00011353111675991275,
      "loss": 2.1797,
      "step": 164393
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.1214306354522705,
      "learning_rate": 0.00011352791237718789,
      "loss": 1.858,
      "step": 164394
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.1133476495742798,
      "learning_rate": 0.00011352470802913179,
      "loss": 2.177,
      "step": 164395
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.2207868099212646,
      "learning_rate": 0.000113521503715745,
      "loss": 2.0537,
      "step": 164396
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.2181569337844849,
      "learning_rate": 0.00011351829943702804,
      "loss": 2.1326,
      "step": 164397
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.2841321229934692,
      "learning_rate": 0.00011351509519298148,
      "loss": 2.0795,
      "step": 164398
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.154463529586792,
      "learning_rate": 0.00011351189098360605,
      "loss": 2.0019,
      "step": 164399
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.2278271913528442,
      "learning_rate": 0.00011350868680890222,
      "loss": 2.1931,
      "step": 164400
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.2211709022521973,
      "learning_rate": 0.00011350548266887072,
      "loss": 2.1999,
      "step": 164401
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.2147806882858276,
      "learning_rate": 0.00011350227856351209,
      "loss": 2.0677,
      "step": 164402
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.2920399904251099,
      "learning_rate": 0.00011349907449282688,
      "loss": 2.0544,
      "step": 164403
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.2629756927490234,
      "learning_rate": 0.00011349587045681565,
      "loss": 2.1735,
      "step": 164404
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.5219454765319824,
      "learning_rate": 0.00011349266645547909,
      "loss": 2.2412,
      "step": 164405
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.7791966199874878,
      "learning_rate": 0.00011348946248881772,
      "loss": 2.1302,
      "step": 164406
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.1154886484146118,
      "learning_rate": 0.00011348625855683227,
      "loss": 1.876,
      "step": 164407
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.2663774490356445,
      "learning_rate": 0.00011348305465952324,
      "loss": 2.2115,
      "step": 164408
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.228949785232544,
      "learning_rate": 0.00011347985079689115,
      "loss": 2.036,
      "step": 164409
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.330079436302185,
      "learning_rate": 0.00011347664696893675,
      "loss": 2.0361,
      "step": 164410
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.3353427648544312,
      "learning_rate": 0.00011347344317566057,
      "loss": 1.8249,
      "step": 164411
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.3202271461486816,
      "learning_rate": 0.00011347023941706313,
      "loss": 2.0213,
      "step": 164412
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.4092496633529663,
      "learning_rate": 0.00011346703569314514,
      "loss": 2.1942,
      "step": 164413
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.3517329692840576,
      "learning_rate": 0.00011346383200390717,
      "loss": 2.1271,
      "step": 164414
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.2923412322998047,
      "learning_rate": 0.0001134606283493497,
      "loss": 1.877,
      "step": 164415
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.3885670900344849,
      "learning_rate": 0.00011345742472947352,
      "loss": 2.0715,
      "step": 164416
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.4259003400802612,
      "learning_rate": 0.00011345422114427904,
      "loss": 2.0849,
      "step": 164417
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.1579128503799438,
      "learning_rate": 0.00011345101759376704,
      "loss": 2.0544,
      "step": 164418
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.354287028312683,
      "learning_rate": 0.000113447814077938,
      "loss": 2.1132,
      "step": 164419
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.4010093212127686,
      "learning_rate": 0.00011344461059679254,
      "loss": 2.1013,
      "step": 164420
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.1932623386383057,
      "learning_rate": 0.00011344140715033114,
      "loss": 2.1112,
      "step": 164421
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.2371368408203125,
      "learning_rate": 0.0001134382037385546,
      "loss": 2.037,
      "step": 164422
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.3015769720077515,
      "learning_rate": 0.00011343500036146331,
      "loss": 1.9211,
      "step": 164423
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.330052375793457,
      "learning_rate": 0.00011343179701905811,
      "loss": 2.1484,
      "step": 164424
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.3377037048339844,
      "learning_rate": 0.00011342859371133944,
      "loss": 1.8512,
      "step": 164425
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.287721037864685,
      "learning_rate": 0.00011342539043830794,
      "loss": 1.8825,
      "step": 164426
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.25851571559906,
      "learning_rate": 0.00011342218719996407,
      "loss": 1.9113,
      "step": 164427
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.2146806716918945,
      "learning_rate": 0.0001134189839963086,
      "loss": 1.9199,
      "step": 164428
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.3023695945739746,
      "learning_rate": 0.000113415780827342,
      "loss": 2.212,
      "step": 164429
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.3146523237228394,
      "learning_rate": 0.00011341257769306502,
      "loss": 2.1558,
      "step": 164430
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.377874732017517,
      "learning_rate": 0.00011340937459347805,
      "loss": 2.1392,
      "step": 164431
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.2866426706314087,
      "learning_rate": 0.00011340617152858203,
      "loss": 1.9584,
      "step": 164432
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.2592885494232178,
      "learning_rate": 0.0001134029684983771,
      "loss": 2.1266,
      "step": 164433
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.3911105394363403,
      "learning_rate": 0.00011339976550286421,
      "loss": 1.9846,
      "step": 164434
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.288533091545105,
      "learning_rate": 0.0001133965625420437,
      "loss": 2.072,
      "step": 164435
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.2004163265228271,
      "learning_rate": 0.0001133933596159164,
      "loss": 2.08,
      "step": 164436
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.1650687456130981,
      "learning_rate": 0.00011339015672448269,
      "loss": 2.0935,
      "step": 164437
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.4812259674072266,
      "learning_rate": 0.00011338695386774349,
      "loss": 2.2142,
      "step": 164438
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.2526887655258179,
      "learning_rate": 0.00011338375104569895,
      "loss": 2.0172,
      "step": 164439
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.480087161064148,
      "learning_rate": 0.00011338054825835002,
      "loss": 2.1476,
      "step": 164440
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.3126291036605835,
      "learning_rate": 0.00011337734550569706,
      "loss": 1.9124,
      "step": 164441
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.1314321756362915,
      "learning_rate": 0.00011337414278774085,
      "loss": 1.9156,
      "step": 164442
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.1477910280227661,
      "learning_rate": 0.00011337094010448182,
      "loss": 1.8821,
      "step": 164443
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.5153510570526123,
      "learning_rate": 0.0001133677374559209,
      "loss": 2.0596,
      "step": 164444
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.3875178098678589,
      "learning_rate": 0.0001133645348420582,
      "loss": 1.9859,
      "step": 164445
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.5437740087509155,
      "learning_rate": 0.00011336133226289466,
      "loss": 2.1472,
      "step": 164446
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.2239056825637817,
      "learning_rate": 0.00011335812971843068,
      "loss": 2.0404,
      "step": 164447
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.2583551406860352,
      "learning_rate": 0.00011335492720866704,
      "loss": 1.9728,
      "step": 164448
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.2833775281906128,
      "learning_rate": 0.00011335172473360414,
      "loss": 2.3124,
      "step": 164449
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.32084059715271,
      "learning_rate": 0.00011334852229324288,
      "loss": 2.0487,
      "step": 164450
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.10597562789917,
      "learning_rate": 0.00011334531988758346,
      "loss": 1.9775,
      "step": 164451
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.1970428228378296,
      "learning_rate": 0.00011334211751662675,
      "loss": 1.9987,
      "step": 164452
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.227691411972046,
      "learning_rate": 0.00011333891518037317,
      "loss": 1.9009,
      "step": 164453
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.3150379657745361,
      "learning_rate": 0.0001133357128788235,
      "loss": 1.9648,
      "step": 164454
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.2433584928512573,
      "learning_rate": 0.00011333251061197817,
      "loss": 2.0486,
      "step": 164455
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.2888821363449097,
      "learning_rate": 0.00011332930837983806,
      "loss": 2.0828,
      "step": 164456
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.202636957168579,
      "learning_rate": 0.0001133261061824033,
      "loss": 1.9697,
      "step": 164457
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.2562991380691528,
      "learning_rate": 0.0001133229040196749,
      "loss": 2.0867,
      "step": 164458
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.2261276245117188,
      "learning_rate": 0.00011331970189165316,
      "loss": 1.9495,
      "step": 164459
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.2568773031234741,
      "learning_rate": 0.00011331649979833892,
      "loss": 2.0027,
      "step": 164460
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.2837669849395752,
      "learning_rate": 0.00011331329773973259,
      "loss": 2.0373,
      "step": 164461
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.3635488748550415,
      "learning_rate": 0.00011331009571583502,
      "loss": 2.0706,
      "step": 164462
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.187445878982544,
      "learning_rate": 0.00011330689372664643,
      "loss": 2.0387,
      "step": 164463
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.343216896057129,
      "learning_rate": 0.0001133036917721677,
      "loss": 1.9668,
      "step": 164464
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.3251086473464966,
      "learning_rate": 0.00011330048985239925,
      "loss": 2.0213,
      "step": 164465
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.165446162223816,
      "learning_rate": 0.00011329728796734185,
      "loss": 2.0687,
      "step": 164466
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.2394462823867798,
      "learning_rate": 0.00011329408611699593,
      "loss": 1.8722,
      "step": 164467
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.2312802076339722,
      "learning_rate": 0.00011329088430136237,
      "loss": 1.7075,
      "step": 164468
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.2162822484970093,
      "learning_rate": 0.00011328768252044132,
      "loss": 2.1019,
      "step": 164469
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.3346294164657593,
      "learning_rate": 0.00011328448077423376,
      "loss": 2.1136,
      "step": 164470
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.2079505920410156,
      "learning_rate": 0.00011328127906274,
      "loss": 2.0236,
      "step": 164471
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.2412339448928833,
      "learning_rate": 0.0001132780773859609,
      "loss": 1.9951,
      "step": 164472
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.1464786529541016,
      "learning_rate": 0.00011327487574389679,
      "loss": 2.1057,
      "step": 164473
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.20722496509552,
      "learning_rate": 0.00011327167413654866,
      "loss": 1.8681,
      "step": 164474
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.3024789094924927,
      "learning_rate": 0.00011326847256391663,
      "loss": 1.9405,
      "step": 164475
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.2651009559631348,
      "learning_rate": 0.00011326527102600162,
      "loss": 2.1529,
      "step": 164476
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.1771622896194458,
      "learning_rate": 0.000113262069522804,
      "loss": 2.0294,
      "step": 164477
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.1912274360656738,
      "learning_rate": 0.00011325886805432462,
      "loss": 2.0078,
      "step": 164478
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.2421660423278809,
      "learning_rate": 0.00011325566662056379,
      "loss": 2.0357,
      "step": 164479
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.217223048210144,
      "learning_rate": 0.00011325246522152239,
      "loss": 2.0414,
      "step": 164480
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.386896014213562,
      "learning_rate": 0.00011324926385720089,
      "loss": 1.9686,
      "step": 164481
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.1860095262527466,
      "learning_rate": 0.00011324606252759988,
      "loss": 1.9727,
      "step": 164482
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.2214902639389038,
      "learning_rate": 0.00011324286123271982,
      "loss": 1.8883,
      "step": 164483
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.226534366607666,
      "learning_rate": 0.00011323965997256156,
      "loss": 1.9165,
      "step": 164484
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.246307373046875,
      "learning_rate": 0.00011323645874712546,
      "loss": 1.9646,
      "step": 164485
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.2014145851135254,
      "learning_rate": 0.00011323325755641236,
      "loss": 2.0828,
      "step": 164486
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.2051331996917725,
      "learning_rate": 0.00011323005640042271,
      "loss": 2.033,
      "step": 164487
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.4220432043075562,
      "learning_rate": 0.00011322685527915711,
      "loss": 1.9588,
      "step": 164488
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.192487120628357,
      "learning_rate": 0.00011322365419261604,
      "loss": 1.8703,
      "step": 164489
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.1395087242126465,
      "learning_rate": 0.00011322045314080036,
      "loss": 2.1273,
      "step": 164490
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.2237414121627808,
      "learning_rate": 0.0001132172521237104,
      "loss": 2.2426,
      "step": 164491
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.1912295818328857,
      "learning_rate": 0.00011321405114134697,
      "loss": 1.9091,
      "step": 164492
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.2205301523208618,
      "learning_rate": 0.00011321085019371058,
      "loss": 1.9609,
      "step": 164493
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.218309998512268,
      "learning_rate": 0.00011320764928080175,
      "loss": 2.1698,
      "step": 164494
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.2266819477081299,
      "learning_rate": 0.00011320444840262121,
      "loss": 2.043,
      "step": 164495
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.2470073699951172,
      "learning_rate": 0.00011320124755916948,
      "loss": 2.0152,
      "step": 164496
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.2998546361923218,
      "learning_rate": 0.00011319804675044708,
      "loss": 2.3914,
      "step": 164497
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.2599915266036987,
      "learning_rate": 0.0001131948459764548,
      "loss": 2.0443,
      "step": 164498
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.2887552976608276,
      "learning_rate": 0.00011319164523719312,
      "loss": 2.1937,
      "step": 164499
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.2405668497085571,
      "learning_rate": 0.00011318844453266255,
      "loss": 2.009,
      "step": 164500
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.2988427877426147,
      "learning_rate": 0.00011318524386286386,
      "loss": 1.922,
      "step": 164501
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.255628228187561,
      "learning_rate": 0.00011318204322779744,
      "loss": 1.9569,
      "step": 164502
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.2837575674057007,
      "learning_rate": 0.00011317884262746413,
      "loss": 2.0747,
      "step": 164503
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.237032175064087,
      "learning_rate": 0.0001131756420618644,
      "loss": 2.06,
      "step": 164504
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.21477210521698,
      "learning_rate": 0.00011317244153099882,
      "loss": 2.12,
      "step": 164505
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.2490438222885132,
      "learning_rate": 0.00011316924103486794,
      "loss": 2.2277,
      "step": 164506
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.307318925857544,
      "learning_rate": 0.0001131660405734725,
      "loss": 2.3353,
      "step": 164507
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.2830294370651245,
      "learning_rate": 0.00011316284014681294,
      "loss": 2.0314,
      "step": 164508
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.3454679250717163,
      "learning_rate": 0.00011315963975489005,
      "loss": 2.1592,
      "step": 164509
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.2213457822799683,
      "learning_rate": 0.00011315643939770428,
      "loss": 2.2522,
      "step": 164510
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.1312499046325684,
      "learning_rate": 0.00011315323907525626,
      "loss": 1.9653,
      "step": 164511
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.2566708326339722,
      "learning_rate": 0.00011315003878754646,
      "loss": 1.9184,
      "step": 164512
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.1692216396331787,
      "learning_rate": 0.00011314683853457574,
      "loss": 2.2626,
      "step": 164513
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.2867350578308105,
      "learning_rate": 0.00011314363831634443,
      "loss": 1.921,
      "step": 164514
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.2664029598236084,
      "learning_rate": 0.00011314043813285331,
      "loss": 1.9251,
      "step": 164515
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.186943531036377,
      "learning_rate": 0.00011313723798410285,
      "loss": 1.9624,
      "step": 164516
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.3800280094146729,
      "learning_rate": 0.0001131340378700939,
      "loss": 2.1205,
      "step": 164517
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.2129634618759155,
      "learning_rate": 0.00011313083779082667,
      "loss": 2.1418,
      "step": 164518
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.1759330034255981,
      "learning_rate": 0.00011312763774630203,
      "loss": 2.0962,
      "step": 164519
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.2365838289260864,
      "learning_rate": 0.00011312443773652038,
      "loss": 1.7886,
      "step": 164520
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.1876366138458252,
      "learning_rate": 0.00011312123776148253,
      "loss": 1.9164,
      "step": 164521
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.1901824474334717,
      "learning_rate": 0.00011311803782118889,
      "loss": 2.0612,
      "step": 164522
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.1850852966308594,
      "learning_rate": 0.00011311483791564032,
      "loss": 2.1251,
      "step": 164523
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.3418505191802979,
      "learning_rate": 0.00011311163804483703,
      "loss": 2.0657,
      "step": 164524
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.151049017906189,
      "learning_rate": 0.0001131084382087799,
      "loss": 2.1893,
      "step": 164525
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.4326515197753906,
      "learning_rate": 0.00011310523840746935,
      "loss": 1.9744,
      "step": 164526
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.162462830543518,
      "learning_rate": 0.00011310203864090618,
      "loss": 2.146,
      "step": 164527
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.3584409952163696,
      "learning_rate": 0.00011309883890909073,
      "loss": 2.1505,
      "step": 164528
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.254070520401001,
      "learning_rate": 0.00011309563921202397,
      "loss": 1.9995,
      "step": 164529
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.2811429500579834,
      "learning_rate": 0.00011309243954970603,
      "loss": 1.9896,
      "step": 164530
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.3137162923812866,
      "learning_rate": 0.00011308923992213788,
      "loss": 2.1723,
      "step": 164531
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.4625818729400635,
      "learning_rate": 0.00011308604032931982,
      "loss": 2.1683,
      "step": 164532
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.1838780641555786,
      "learning_rate": 0.00011308284077125272,
      "loss": 2.0151,
      "step": 164533
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.1997523307800293,
      "learning_rate": 0.00011307964124793693,
      "loss": 1.8264,
      "step": 164534
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.173775553703308,
      "learning_rate": 0.00011307644175937338,
      "loss": 1.9619,
      "step": 164535
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.128287672996521,
      "learning_rate": 0.00011307324230556226,
      "loss": 2.0196,
      "step": 164536
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.284833550453186,
      "learning_rate": 0.0001130700428865044,
      "loss": 2.1149,
      "step": 164537
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.2447642087936401,
      "learning_rate": 0.00011306684350220029,
      "loss": 2.0912,
      "step": 164538
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.170799732208252,
      "learning_rate": 0.0001130636441526507,
      "loss": 2.0598,
      "step": 164539
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.1860255002975464,
      "learning_rate": 0.00011306044483785598,
      "loss": 1.9402,
      "step": 164540
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.2510374784469604,
      "learning_rate": 0.00011305724555781705,
      "loss": 1.9406,
      "step": 164541
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.2366081476211548,
      "learning_rate": 0.0001130540463125341,
      "loss": 2.1979,
      "step": 164542
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.2528260946273804,
      "learning_rate": 0.00011305084710200803,
      "loss": 2.2082,
      "step": 164543
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.2426260709762573,
      "learning_rate": 0.00011304764792623928,
      "loss": 2.1456,
      "step": 164544
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.123165488243103,
      "learning_rate": 0.00011304444878522857,
      "loss": 2.12,
      "step": 164545
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.2479639053344727,
      "learning_rate": 0.0001130412496789763,
      "loss": 1.8866,
      "step": 164546
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.1850736141204834,
      "learning_rate": 0.00011303805060748332,
      "loss": 2.0873,
      "step": 164547
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.3915563821792603,
      "learning_rate": 0.0001130348515707501,
      "loss": 2.0263,
      "step": 164548
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.1879026889801025,
      "learning_rate": 0.00011303165256877722,
      "loss": 2.0072,
      "step": 164549
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.2018799781799316,
      "learning_rate": 0.00011302845360156519,
      "loss": 2.002,
      "step": 164550
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.2552237510681152,
      "learning_rate": 0.0001130252546691148,
      "loss": 1.9806,
      "step": 164551
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.4064679145812988,
      "learning_rate": 0.00011302205577142646,
      "loss": 2.284,
      "step": 164552
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.2654645442962646,
      "learning_rate": 0.00011301885690850091,
      "loss": 2.1354,
      "step": 164553
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.2986118793487549,
      "learning_rate": 0.00011301565808033872,
      "loss": 2.0302,
      "step": 164554
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.1740306615829468,
      "learning_rate": 0.00011301245928694042,
      "loss": 2.0591,
      "step": 164555
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.332950472831726,
      "learning_rate": 0.00011300926052830656,
      "loss": 1.9852,
      "step": 164556
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.2407387495040894,
      "learning_rate": 0.0001130060618044379,
      "loss": 2.2803,
      "step": 164557
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.094563364982605,
      "learning_rate": 0.00011300286311533482,
      "loss": 1.9875,
      "step": 164558
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.088448405265808,
      "learning_rate": 0.00011299966446099815,
      "loss": 1.8481,
      "step": 164559
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.340093731880188,
      "learning_rate": 0.00011299646584142841,
      "loss": 2.2876,
      "step": 164560
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.3954144716262817,
      "learning_rate": 0.00011299326725662613,
      "loss": 2.046,
      "step": 164561
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.1944464445114136,
      "learning_rate": 0.00011299006870659182,
      "loss": 2.0156,
      "step": 164562
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.416060447692871,
      "learning_rate": 0.00011298687019132627,
      "loss": 1.9504,
      "step": 164563
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.2372792959213257,
      "learning_rate": 0.00011298367171082991,
      "loss": 1.9985,
      "step": 164564
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.2149122953414917,
      "learning_rate": 0.00011298047326510354,
      "loss": 1.8934,
      "step": 164565
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.1711366176605225,
      "learning_rate": 0.00011297727485414762,
      "loss": 1.971,
      "step": 164566
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.2889201641082764,
      "learning_rate": 0.00011297407647796272,
      "loss": 1.9836,
      "step": 164567
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.2463245391845703,
      "learning_rate": 0.00011297087813654941,
      "loss": 2.3446,
      "step": 164568
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.297491431236267,
      "learning_rate": 0.00011296767982990842,
      "loss": 2.0545,
      "step": 164569
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.359238862991333,
      "learning_rate": 0.00011296448155804017,
      "loss": 2.0951,
      "step": 164570
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.3185304403305054,
      "learning_rate": 0.00011296128332094549,
      "loss": 2.1327,
      "step": 164571
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.2394139766693115,
      "learning_rate": 0.00011295808511862481,
      "loss": 2.2659,
      "step": 164572
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.200280785560608,
      "learning_rate": 0.00011295488695107875,
      "loss": 1.827,
      "step": 164573
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.2715013027191162,
      "learning_rate": 0.00011295168881830781,
      "loss": 1.9171,
      "step": 164574
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.1888843774795532,
      "learning_rate": 0.00011294849072031278,
      "loss": 1.8041,
      "step": 164575
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.0826047658920288,
      "learning_rate": 0.00011294529265709407,
      "loss": 1.946,
      "step": 164576
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.2323713302612305,
      "learning_rate": 0.00011294209462865244,
      "loss": 1.9297,
      "step": 164577
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.3196189403533936,
      "learning_rate": 0.00011293889663498842,
      "loss": 2.0578,
      "step": 164578
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.287312388420105,
      "learning_rate": 0.00011293569867610248,
      "loss": 2.0505,
      "step": 164579
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.2205480337142944,
      "learning_rate": 0.00011293250075199542,
      "loss": 2.3521,
      "step": 164580
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.2748106718063354,
      "learning_rate": 0.00011292930286266775,
      "loss": 2.1417,
      "step": 164581
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.1321799755096436,
      "learning_rate": 0.00011292610500811995,
      "loss": 2.1009,
      "step": 164582
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.3759325742721558,
      "learning_rate": 0.00011292290718835283,
      "loss": 2.0312,
      "step": 164583
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.1932560205459595,
      "learning_rate": 0.00011291970940336685,
      "loss": 1.988,
      "step": 164584
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.1817432641983032,
      "learning_rate": 0.00011291651165316256,
      "loss": 2.0931,
      "step": 164585
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.313454508781433,
      "learning_rate": 0.00011291331393774071,
      "loss": 1.9356,
      "step": 164586
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.3375380039215088,
      "learning_rate": 0.00011291011625710179,
      "loss": 1.961,
      "step": 164587
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.236613392829895,
      "learning_rate": 0.00011290691861124632,
      "loss": 1.9285,
      "step": 164588
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.2747993469238281,
      "learning_rate": 0.00011290372100017509,
      "loss": 2.0743,
      "step": 164589
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.2621594667434692,
      "learning_rate": 0.00011290052342388858,
      "loss": 2.2241,
      "step": 164590
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.4691438674926758,
      "learning_rate": 0.0001128973258823873,
      "loss": 1.9346,
      "step": 164591
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.2778990268707275,
      "learning_rate": 0.00011289412837567204,
      "loss": 2.0639,
      "step": 164592
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.237245798110962,
      "learning_rate": 0.00011289093090374321,
      "loss": 2.0724,
      "step": 164593
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.2711961269378662,
      "learning_rate": 0.00011288773346660155,
      "loss": 2.0951,
      "step": 164594
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.2781567573547363,
      "learning_rate": 0.00011288453606424764,
      "loss": 2.1197,
      "step": 164595
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.2718231678009033,
      "learning_rate": 0.000112881338696682,
      "loss": 2.1196,
      "step": 164596
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.2078208923339844,
      "learning_rate": 0.00011287814136390512,
      "loss": 1.8473,
      "step": 164597
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.2113981246948242,
      "learning_rate": 0.00011287494406591788,
      "loss": 1.9012,
      "step": 164598
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.2515720129013062,
      "learning_rate": 0.00011287174680272059,
      "loss": 2.0746,
      "step": 164599
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.3108139038085938,
      "learning_rate": 0.0001128685495743141,
      "loss": 1.9005,
      "step": 164600
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.4179883003234863,
      "learning_rate": 0.00011286535238069874,
      "loss": 2.1346,
      "step": 164601
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.2347973585128784,
      "learning_rate": 0.00011286215522187545,
      "loss": 2.0424,
      "step": 164602
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.2433466911315918,
      "learning_rate": 0.00011285895809784444,
      "loss": 2.085,
      "step": 164603
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.2231249809265137,
      "learning_rate": 0.00011285576100860658,
      "loss": 2.1867,
      "step": 164604
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.4158002138137817,
      "learning_rate": 0.00011285256395416227,
      "loss": 2.2491,
      "step": 164605
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.1778976917266846,
      "learning_rate": 0.00011284936693451229,
      "loss": 2.1914,
      "step": 164606
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.1096736192703247,
      "learning_rate": 0.00011284616994965705,
      "loss": 1.9273,
      "step": 164607
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.4493900537490845,
      "learning_rate": 0.00011284297299959743,
      "loss": 2.1968,
      "step": 164608
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.129852294921875,
      "learning_rate": 0.00011283977608433365,
      "loss": 1.8276,
      "step": 164609
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.3321064710617065,
      "learning_rate": 0.00011283657920386657,
      "loss": 2.0502,
      "step": 164610
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.450932502746582,
      "learning_rate": 0.00011283338235819661,
      "loss": 2.0496,
      "step": 164611
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.3033398389816284,
      "learning_rate": 0.00011283018554732456,
      "loss": 2.0394,
      "step": 164612
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.1759283542633057,
      "learning_rate": 0.00011282698877125081,
      "loss": 2.0228,
      "step": 164613
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.3857429027557373,
      "learning_rate": 0.00011282379202997618,
      "loss": 2.1065,
      "step": 164614
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.4036177396774292,
      "learning_rate": 0.00011282059532350112,
      "loss": 1.6945,
      "step": 164615
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.2188292741775513,
      "learning_rate": 0.00011281739865182622,
      "loss": 2.2233,
      "step": 164616
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.230325698852539,
      "learning_rate": 0.00011281420201495202,
      "loss": 2.0802,
      "step": 164617
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.1904808282852173,
      "learning_rate": 0.0001128110054128793,
      "loss": 1.942,
      "step": 164618
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.3592681884765625,
      "learning_rate": 0.00011280780884560846,
      "loss": 2.0034,
      "step": 164619
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.303830623626709,
      "learning_rate": 0.00011280461231314024,
      "loss": 2.0696,
      "step": 164620
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.2368768453598022,
      "learning_rate": 0.00011280141581547521,
      "loss": 2.0577,
      "step": 164621
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.1400270462036133,
      "learning_rate": 0.00011279821935261391,
      "loss": 2.0108,
      "step": 164622
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.2139296531677246,
      "learning_rate": 0.00011279502292455687,
      "loss": 2.0433,
      "step": 164623
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.2993531227111816,
      "learning_rate": 0.00011279182653130486,
      "loss": 2.1374,
      "step": 164624
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.5516517162322998,
      "learning_rate": 0.0001127886301728583,
      "loss": 2.2316,
      "step": 164625
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.204332709312439,
      "learning_rate": 0.00011278543384921794,
      "loss": 1.755,
      "step": 164626
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.3468859195709229,
      "learning_rate": 0.00011278223756038433,
      "loss": 2.1861,
      "step": 164627
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.3586875200271606,
      "learning_rate": 0.00011277904130635802,
      "loss": 2.1119,
      "step": 164628
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.6800971031188965,
      "learning_rate": 0.00011277584508713953,
      "loss": 2.194,
      "step": 164629
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.18281090259552,
      "learning_rate": 0.00011277264890272966,
      "loss": 2.2466,
      "step": 164630
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.2591915130615234,
      "learning_rate": 0.00011276945275312875,
      "loss": 2.0512,
      "step": 164631
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.197514295578003,
      "learning_rate": 0.00011276625663833765,
      "loss": 2.0982,
      "step": 164632
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.2561622858047485,
      "learning_rate": 0.00011276306055835685,
      "loss": 2.1085,
      "step": 164633
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.23611581325531,
      "learning_rate": 0.00011275986451318689,
      "loss": 2.2237,
      "step": 164634
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.1684685945510864,
      "learning_rate": 0.00011275666850282836,
      "loss": 2.1404,
      "step": 164635
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.1846157312393188,
      "learning_rate": 0.00011275347252728195,
      "loss": 2.0463,
      "step": 164636
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.22538161277771,
      "learning_rate": 0.00011275027658654813,
      "loss": 1.951,
      "step": 164637
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.350329875946045,
      "learning_rate": 0.00011274708068062769,
      "loss": 2.1572,
      "step": 164638
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.1525927782058716,
      "learning_rate": 0.00011274388480952106,
      "loss": 2.2091,
      "step": 164639
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.1465641260147095,
      "learning_rate": 0.0001127406889732289,
      "loss": 1.7818,
      "step": 164640
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.220314383506775,
      "learning_rate": 0.00011273749317175167,
      "loss": 2.185,
      "step": 164641
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.224708914756775,
      "learning_rate": 0.00011273429740509017,
      "loss": 2.1933,
      "step": 164642
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.2110716104507446,
      "learning_rate": 0.00011273110167324483,
      "loss": 2.1154,
      "step": 164643
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.1814252138137817,
      "learning_rate": 0.00011272790597621639,
      "loss": 1.9271,
      "step": 164644
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.2144697904586792,
      "learning_rate": 0.00011272471031400537,
      "loss": 1.8429,
      "step": 164645
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.1009893417358398,
      "learning_rate": 0.00011272151468661237,
      "loss": 2.0697,
      "step": 164646
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.2609832286834717,
      "learning_rate": 0.00011271831909403784,
      "loss": 2.0505,
      "step": 164647
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.2022147178649902,
      "learning_rate": 0.00011271512353628267,
      "loss": 2.136,
      "step": 164648
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.247279167175293,
      "learning_rate": 0.00011271192801334714,
      "loss": 2.0165,
      "step": 164649
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.2920305728912354,
      "learning_rate": 0.00011270873252523213,
      "loss": 2.0237,
      "step": 164650
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.4159095287322998,
      "learning_rate": 0.0001127055370719381,
      "loss": 1.9479,
      "step": 164651
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.236362099647522,
      "learning_rate": 0.00011270234165346563,
      "loss": 1.9999,
      "step": 164652
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.1483274698257446,
      "learning_rate": 0.00011269914626981522,
      "loss": 1.9955,
      "step": 164653
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.1510149240493774,
      "learning_rate": 0.00011269595092098772,
      "loss": 2.0436,
      "step": 164654
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.3811520338058472,
      "learning_rate": 0.00011269275560698343,
      "loss": 2.196,
      "step": 164655
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.114163875579834,
      "learning_rate": 0.00011268956032780322,
      "loss": 1.9522,
      "step": 164656
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.286196231842041,
      "learning_rate": 0.00011268636508344755,
      "loss": 1.9976,
      "step": 164657
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.1648489236831665,
      "learning_rate": 0.00011268316987391702,
      "loss": 2.059,
      "step": 164658
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.583983302116394,
      "learning_rate": 0.0001126799746992121,
      "loss": 2.0349,
      "step": 164659
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.2944256067276,
      "learning_rate": 0.00011267677955933366,
      "loss": 2.1979,
      "step": 164660
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.7606446743011475,
      "learning_rate": 0.00011267358445428199,
      "loss": 2.1551,
      "step": 164661
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.2070448398590088,
      "learning_rate": 0.00011267038938405797,
      "loss": 1.8724,
      "step": 164662
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.1590689420700073,
      "learning_rate": 0.00011266719434866205,
      "loss": 1.9196,
      "step": 164663
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.1015293598175049,
      "learning_rate": 0.00011266399934809473,
      "loss": 2.0177,
      "step": 164664
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.3901393413543701,
      "learning_rate": 0.00011266080438235683,
      "loss": 2.2727,
      "step": 164665
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.2701998949050903,
      "learning_rate": 0.00011265760945144882,
      "loss": 2.1011,
      "step": 164666
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.2418462038040161,
      "learning_rate": 0.00011265441455537116,
      "loss": 1.9962,
      "step": 164667
    },
    {
      "epoch": 2.14,
      "grad_norm": 2.895280361175537,
      "learning_rate": 0.00011265121969412474,
      "loss": 2.1739,
      "step": 164668
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.1534241437911987,
      "learning_rate": 0.00011264802486770995,
      "loss": 2.0383,
      "step": 164669
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.3084791898727417,
      "learning_rate": 0.00011264483007612733,
      "loss": 2.1342,
      "step": 164670
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.2997277975082397,
      "learning_rate": 0.0001126416353193777,
      "loss": 2.2183,
      "step": 164671
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.307287335395813,
      "learning_rate": 0.00011263844059746151,
      "loss": 2.1836,
      "step": 164672
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.1920576095581055,
      "learning_rate": 0.00011263524591037927,
      "loss": 1.9857,
      "step": 164673
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.3215357065200806,
      "learning_rate": 0.00011263205125813176,
      "loss": 2.0742,
      "step": 164674
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.2585018873214722,
      "learning_rate": 0.00011262885664071952,
      "loss": 1.9204,
      "step": 164675
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.141369342803955,
      "learning_rate": 0.00011262566205814298,
      "loss": 2.0611,
      "step": 164676
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.4614300727844238,
      "learning_rate": 0.00011262246751040298,
      "loss": 2.0022,
      "step": 164677
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.1947218179702759,
      "learning_rate": 0.0001126192729974999,
      "loss": 2.1247,
      "step": 164678
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.2154604196548462,
      "learning_rate": 0.00011261607851943453,
      "loss": 1.7416,
      "step": 164679
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.3775770664215088,
      "learning_rate": 0.0001126128840762074,
      "loss": 1.9316,
      "step": 164680
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.4213271141052246,
      "learning_rate": 0.00011260968966781893,
      "loss": 1.9533,
      "step": 164681
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.1824733018875122,
      "learning_rate": 0.00011260649529426998,
      "loss": 1.9016,
      "step": 164682
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.2283896207809448,
      "learning_rate": 0.000112603300955561,
      "loss": 2.1687,
      "step": 164683
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.1617074012756348,
      "learning_rate": 0.00011260010665169252,
      "loss": 2.0334,
      "step": 164684
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.4045002460479736,
      "learning_rate": 0.0001125969123826653,
      "loss": 2.1569,
      "step": 164685
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.361636757850647,
      "learning_rate": 0.00011259371814847989,
      "loss": 2.1678,
      "step": 164686
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.2712444067001343,
      "learning_rate": 0.00011259052394913673,
      "loss": 2.0298,
      "step": 164687
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.3855905532836914,
      "learning_rate": 0.0001125873297846366,
      "loss": 1.8614,
      "step": 164688
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.325745940208435,
      "learning_rate": 0.00011258413565498005,
      "loss": 2.1401,
      "step": 164689
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.3389848470687866,
      "learning_rate": 0.00011258094156016754,
      "loss": 2.0514,
      "step": 164690
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.2707843780517578,
      "learning_rate": 0.00011257774750019987,
      "loss": 2.3581,
      "step": 164691
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.1405763626098633,
      "learning_rate": 0.00011257455347507741,
      "loss": 2.053,
      "step": 164692
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.060273289680481,
      "learning_rate": 0.00011257135948480102,
      "loss": 1.8551,
      "step": 164693
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.1496691703796387,
      "learning_rate": 0.00011256816552937114,
      "loss": 2.0113,
      "step": 164694
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.0738403797149658,
      "learning_rate": 0.00011256497160878837,
      "loss": 1.9309,
      "step": 164695
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.503548502922058,
      "learning_rate": 0.00011256177772305322,
      "loss": 2.2931,
      "step": 164696
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.2460254430770874,
      "learning_rate": 0.00011255858387216645,
      "loss": 1.931,
      "step": 164697
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.1407530307769775,
      "learning_rate": 0.00011255539005612852,
      "loss": 2.0743,
      "step": 164698
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.2423113584518433,
      "learning_rate": 0.00011255219627494014,
      "loss": 2.1331,
      "step": 164699
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.303175926208496,
      "learning_rate": 0.00011254900252860188,
      "loss": 1.9872,
      "step": 164700
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.3369287252426147,
      "learning_rate": 0.00011254580881711426,
      "loss": 2.054,
      "step": 164701
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.2420732975006104,
      "learning_rate": 0.00011254261514047782,
      "loss": 2.1348,
      "step": 164702
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.240798830986023,
      "learning_rate": 0.00011253942149869336,
      "loss": 2.0803,
      "step": 164703
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.2236424684524536,
      "learning_rate": 0.00011253622789176126,
      "loss": 2.0204,
      "step": 164704
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.2237416505813599,
      "learning_rate": 0.0001125330343196823,
      "loss": 2.15,
      "step": 164705
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.4482526779174805,
      "learning_rate": 0.00011252984078245701,
      "loss": 2.0059,
      "step": 164706
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.5486401319503784,
      "learning_rate": 0.00011252664728008596,
      "loss": 1.9023,
      "step": 164707
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.2879763841629028,
      "learning_rate": 0.00011252345381256963,
      "loss": 2.1003,
      "step": 164708
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.2957837581634521,
      "learning_rate": 0.00011252026037990882,
      "loss": 1.8863,
      "step": 164709
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.2064546346664429,
      "learning_rate": 0.00011251706698210397,
      "loss": 1.9287,
      "step": 164710
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.2787188291549683,
      "learning_rate": 0.00011251387361915579,
      "loss": 2.2202,
      "step": 164711
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.280213475227356,
      "learning_rate": 0.00011251068029106483,
      "loss": 1.9055,
      "step": 164712
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.2075707912445068,
      "learning_rate": 0.0001125074869978317,
      "loss": 2.0905,
      "step": 164713
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.0703145265579224,
      "learning_rate": 0.00011250429373945688,
      "loss": 2.1638,
      "step": 164714
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.2965977191925049,
      "learning_rate": 0.0001125011005159411,
      "loss": 2.0433,
      "step": 164715
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.183860421180725,
      "learning_rate": 0.00011249790732728484,
      "loss": 2.1058,
      "step": 164716
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.2605273723602295,
      "learning_rate": 0.00011249471417348883,
      "loss": 2.298,
      "step": 164717
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.130549430847168,
      "learning_rate": 0.00011249152105455363,
      "loss": 2.0043,
      "step": 164718
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.1729987859725952,
      "learning_rate": 0.00011248832797047979,
      "loss": 2.1237,
      "step": 164719
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.2861801385879517,
      "learning_rate": 0.00011248513492126778,
      "loss": 1.7984,
      "step": 164720
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.1959328651428223,
      "learning_rate": 0.00011248194190691842,
      "loss": 1.9439,
      "step": 164721
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.3004727363586426,
      "learning_rate": 0.00011247874892743214,
      "loss": 2.0265,
      "step": 164722
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.2499464750289917,
      "learning_rate": 0.00011247555598280969,
      "loss": 2.011,
      "step": 164723
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.144457221031189,
      "learning_rate": 0.00011247236307305157,
      "loss": 2.17,
      "step": 164724
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.191028356552124,
      "learning_rate": 0.00011246917019815839,
      "loss": 2.0038,
      "step": 164725
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.2929044961929321,
      "learning_rate": 0.00011246597735813059,
      "loss": 2.0416,
      "step": 164726
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.2964677810668945,
      "learning_rate": 0.00011246278455296903,
      "loss": 2.0049,
      "step": 164727
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.2407454252243042,
      "learning_rate": 0.00011245959178267408,
      "loss": 2.337,
      "step": 164728
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.1869417428970337,
      "learning_rate": 0.00011245639904724655,
      "loss": 1.9939,
      "step": 164729
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.2367863655090332,
      "learning_rate": 0.00011245320634668689,
      "loss": 2.0289,
      "step": 164730
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.363131046295166,
      "learning_rate": 0.00011245001368099573,
      "loss": 2.1889,
      "step": 164731
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.4694362878799438,
      "learning_rate": 0.00011244682105017356,
      "loss": 1.7655,
      "step": 164732
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.1851798295974731,
      "learning_rate": 0.00011244362845422115,
      "loss": 2.011,
      "step": 164733
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.243288516998291,
      "learning_rate": 0.00011244043589313891,
      "loss": 1.9269,
      "step": 164734
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.177234172821045,
      "learning_rate": 0.00011243724336692765,
      "loss": 1.9901,
      "step": 164735
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.4422950744628906,
      "learning_rate": 0.00011243405087558783,
      "loss": 1.9925,
      "step": 164736
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.4044911861419678,
      "learning_rate": 0.00011243085841912008,
      "loss": 2.0085,
      "step": 164737
    },
    {
      "epoch": 2.14,
      "grad_norm": 1.3340482711791992,
      "learning_rate": 0.00011242766599752484,
      "loss": 2.0358,
      "step": 164738
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.4679051637649536,
      "learning_rate": 0.00011242447361080297,
      "loss": 1.9328,
      "step": 164739
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.2470109462738037,
      "learning_rate": 0.00011242128125895483,
      "loss": 1.9076,
      "step": 164740
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.1678900718688965,
      "learning_rate": 0.0001124180889419812,
      "loss": 1.8444,
      "step": 164741
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.205256462097168,
      "learning_rate": 0.0001124148966598826,
      "loss": 2.1676,
      "step": 164742
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.24590265750885,
      "learning_rate": 0.00011241170441265961,
      "loss": 1.8911,
      "step": 164743
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.2288225889205933,
      "learning_rate": 0.00011240851220031272,
      "loss": 1.9747,
      "step": 164744
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.2812511920928955,
      "learning_rate": 0.00011240532002284273,
      "loss": 2.0146,
      "step": 164745
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.2926573753356934,
      "learning_rate": 0.00011240212788025001,
      "loss": 1.9346,
      "step": 164746
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.2256317138671875,
      "learning_rate": 0.00011239893577253542,
      "loss": 1.8473,
      "step": 164747
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.2875958681106567,
      "learning_rate": 0.00011239574369969926,
      "loss": 2.0476,
      "step": 164748
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.303795576095581,
      "learning_rate": 0.00011239255166174251,
      "loss": 2.0699,
      "step": 164749
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.204628586769104,
      "learning_rate": 0.00011238935965866531,
      "loss": 1.9384,
      "step": 164750
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.157418131828308,
      "learning_rate": 0.00011238616769046858,
      "loss": 2.0696,
      "step": 164751
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.2575393915176392,
      "learning_rate": 0.0001123829757571527,
      "loss": 1.9861,
      "step": 164752
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.1503081321716309,
      "learning_rate": 0.00011237978385871847,
      "loss": 2.1001,
      "step": 164753
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.1754686832427979,
      "learning_rate": 0.00011237659199516626,
      "loss": 1.946,
      "step": 164754
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.167123556137085,
      "learning_rate": 0.00011237340016649689,
      "loss": 2.2226,
      "step": 164755
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.1894526481628418,
      "learning_rate": 0.00011237020837271086,
      "loss": 1.9168,
      "step": 164756
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.1802058219909668,
      "learning_rate": 0.00011236701661380875,
      "loss": 1.967,
      "step": 164757
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.2518689632415771,
      "learning_rate": 0.00011236382488979103,
      "loss": 1.9819,
      "step": 164758
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.2803024053573608,
      "learning_rate": 0.00011236063320065855,
      "loss": 2.1257,
      "step": 164759
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.2965675592422485,
      "learning_rate": 0.00011235744154641165,
      "loss": 2.0844,
      "step": 164760
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.2322993278503418,
      "learning_rate": 0.00011235424992705116,
      "loss": 2.2192,
      "step": 164761
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.3635876178741455,
      "learning_rate": 0.00011235105834257751,
      "loss": 2.0169,
      "step": 164762
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.1564959287643433,
      "learning_rate": 0.00011234786679299127,
      "loss": 2.0656,
      "step": 164763
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.22786545753479,
      "learning_rate": 0.00011234467527829321,
      "loss": 1.9491,
      "step": 164764
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.2192143201828003,
      "learning_rate": 0.00011234148379848382,
      "loss": 2.0794,
      "step": 164765
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.3427871465682983,
      "learning_rate": 0.00011233829235356358,
      "loss": 2.3008,
      "step": 164766
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.204496145248413,
      "learning_rate": 0.0001123351009435333,
      "loss": 2.0302,
      "step": 164767
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.292574405670166,
      "learning_rate": 0.00011233190956839348,
      "loss": 2.1536,
      "step": 164768
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.217433214187622,
      "learning_rate": 0.00011232871822814459,
      "loss": 2.0561,
      "step": 164769
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.0880430936813354,
      "learning_rate": 0.00011232552692278745,
      "loss": 2.1944,
      "step": 164770
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.2319828271865845,
      "learning_rate": 0.00011232233565232249,
      "loss": 2.2762,
      "step": 164771
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.2067975997924805,
      "learning_rate": 0.0001123191444167503,
      "loss": 2.2174,
      "step": 164772
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.5932421684265137,
      "learning_rate": 0.00011231595321607158,
      "loss": 2.0314,
      "step": 164773
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.228916883468628,
      "learning_rate": 0.00011231276205028689,
      "loss": 1.9099,
      "step": 164774
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.1460981369018555,
      "learning_rate": 0.00011230957091939669,
      "loss": 1.9575,
      "step": 164775
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.3726822137832642,
      "learning_rate": 0.00011230637982340182,
      "loss": 2.2128,
      "step": 164776
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.5335460901260376,
      "learning_rate": 0.00011230318876230258,
      "loss": 1.9777,
      "step": 164777
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.492352843284607,
      "learning_rate": 0.00011229999773609988,
      "loss": 2.2357,
      "step": 164778
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.2537014484405518,
      "learning_rate": 0.00011229680674479412,
      "loss": 1.8912,
      "step": 164779
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.2083112001419067,
      "learning_rate": 0.00011229361578838594,
      "loss": 2.1526,
      "step": 164780
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.3524911403656006,
      "learning_rate": 0.0001122904248668758,
      "loss": 2.1849,
      "step": 164781
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.3365954160690308,
      "learning_rate": 0.00011228723398026453,
      "loss": 1.8442,
      "step": 164782
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.249132752418518,
      "learning_rate": 0.0001122840431285525,
      "loss": 1.9918,
      "step": 164783
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.1936620473861694,
      "learning_rate": 0.0001122808523117405,
      "loss": 1.9683,
      "step": 164784
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.134747862815857,
      "learning_rate": 0.00011227766152982904,
      "loss": 2.0862,
      "step": 164785
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.231008768081665,
      "learning_rate": 0.00011227447078281874,
      "loss": 2.2223,
      "step": 164786
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.4175846576690674,
      "learning_rate": 0.00011227128007071001,
      "loss": 2.0777,
      "step": 164787
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.1854251623153687,
      "learning_rate": 0.0001122680893935037,
      "loss": 1.978,
      "step": 164788
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.1634795665740967,
      "learning_rate": 0.00011226489875120023,
      "loss": 2.1271,
      "step": 164789
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.3442840576171875,
      "learning_rate": 0.00011226170814380033,
      "loss": 2.1956,
      "step": 164790
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.374567985534668,
      "learning_rate": 0.00011225851757130452,
      "loss": 2.069,
      "step": 164791
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.2316374778747559,
      "learning_rate": 0.00011225532703371343,
      "loss": 2.0356,
      "step": 164792
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.4724677801132202,
      "learning_rate": 0.00011225213653102749,
      "loss": 2.0876,
      "step": 164793
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.1493979692459106,
      "learning_rate": 0.00011224894606324752,
      "loss": 1.8865,
      "step": 164794
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.2166054248809814,
      "learning_rate": 0.00011224575563037393,
      "loss": 2.0035,
      "step": 164795
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.2581000328063965,
      "learning_rate": 0.00011224256523240748,
      "loss": 1.9971,
      "step": 164796
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.2413431406021118,
      "learning_rate": 0.00011223937486934868,
      "loss": 2.1836,
      "step": 164797
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.3787397146224976,
      "learning_rate": 0.00011223618454119815,
      "loss": 2.1147,
      "step": 164798
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.2102302312850952,
      "learning_rate": 0.00011223299424795634,
      "loss": 2.1195,
      "step": 164799
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.578706979751587,
      "learning_rate": 0.00011222980398962406,
      "loss": 1.9445,
      "step": 164800
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.2056193351745605,
      "learning_rate": 0.00011222661376620169,
      "loss": 2.0388,
      "step": 164801
    },
    {
      "epoch": 2.15,
      "grad_norm": 2.481139659881592,
      "learning_rate": 0.00011222342357769007,
      "loss": 1.8567,
      "step": 164802
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.2062715291976929,
      "learning_rate": 0.00011222023342408965,
      "loss": 1.9291,
      "step": 164803
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.2592490911483765,
      "learning_rate": 0.00011221704330540103,
      "loss": 2.0547,
      "step": 164804
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.2134394645690918,
      "learning_rate": 0.00011221385322162469,
      "loss": 2.1403,
      "step": 164805
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.3702778816223145,
      "learning_rate": 0.00011221066317276145,
      "loss": 1.9255,
      "step": 164806
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.290557861328125,
      "learning_rate": 0.0001122074731588117,
      "loss": 2.1548,
      "step": 164807
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.1780790090560913,
      "learning_rate": 0.00011220428317977622,
      "loss": 2.0752,
      "step": 164808
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.184767484664917,
      "learning_rate": 0.00011220109323565553,
      "loss": 1.8847,
      "step": 164809
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.3513896465301514,
      "learning_rate": 0.00011219790332645018,
      "loss": 2.1066,
      "step": 164810
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.2894266843795776,
      "learning_rate": 0.00011219471345216066,
      "loss": 2.0283,
      "step": 164811
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.2362815141677856,
      "learning_rate": 0.00011219152361278782,
      "loss": 1.9466,
      "step": 164812
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.3544337749481201,
      "learning_rate": 0.00011218833380833203,
      "loss": 2.2591,
      "step": 164813
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.180332899093628,
      "learning_rate": 0.00011218514403879405,
      "loss": 2.191,
      "step": 164814
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.1751549243927002,
      "learning_rate": 0.00011218195430417431,
      "loss": 2.0871,
      "step": 164815
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.2249951362609863,
      "learning_rate": 0.00011217876460447368,
      "loss": 1.993,
      "step": 164816
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.1549947261810303,
      "learning_rate": 0.00011217557493969237,
      "loss": 2.0006,
      "step": 164817
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.2319496870040894,
      "learning_rate": 0.00011217238530983128,
      "loss": 2.2277,
      "step": 164818
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.1446417570114136,
      "learning_rate": 0.00011216919571489075,
      "loss": 2.2619,
      "step": 164819
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.238173246383667,
      "learning_rate": 0.00011216600615487165,
      "loss": 1.9962,
      "step": 164820
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.1409798860549927,
      "learning_rate": 0.00011216281662977432,
      "loss": 2.125,
      "step": 164821
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.3798539638519287,
      "learning_rate": 0.00011215962713959967,
      "loss": 2.1069,
      "step": 164822
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.2872129678726196,
      "learning_rate": 0.0001121564376843479,
      "loss": 1.9998,
      "step": 164823
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.2317954301834106,
      "learning_rate": 0.00011215324826401985,
      "loss": 2.1417,
      "step": 164824
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.2357017993927002,
      "learning_rate": 0.00011215005887861601,
      "loss": 1.8563,
      "step": 164825
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.202575922012329,
      "learning_rate": 0.0001121468695281371,
      "loss": 2.0894,
      "step": 164826
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.1788069009780884,
      "learning_rate": 0.00011214368021258354,
      "loss": 2.0968,
      "step": 164827
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.144616961479187,
      "learning_rate": 0.00011214049093195621,
      "loss": 1.8637,
      "step": 164828
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.2595210075378418,
      "learning_rate": 0.00011213730168625532,
      "loss": 2.0762,
      "step": 164829
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.2930083274841309,
      "learning_rate": 0.00011213411247548172,
      "loss": 2.1241,
      "step": 164830
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.2783558368682861,
      "learning_rate": 0.00011213092329963588,
      "loss": 1.9395,
      "step": 164831
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.329300880432129,
      "learning_rate": 0.0001121277341587185,
      "loss": 2.1178,
      "step": 164832
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.1703704595565796,
      "learning_rate": 0.00011212454505273004,
      "loss": 2.1434,
      "step": 164833
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.3289949893951416,
      "learning_rate": 0.00011212135598167138,
      "loss": 1.9185,
      "step": 164834
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.2745083570480347,
      "learning_rate": 0.0001121181669455427,
      "loss": 1.9995,
      "step": 164835
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.2623682022094727,
      "learning_rate": 0.00011211497794434493,
      "loss": 2.0558,
      "step": 164836
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.1517919301986694,
      "learning_rate": 0.00011211178897807841,
      "loss": 2.2447,
      "step": 164837
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.3001266717910767,
      "learning_rate": 0.00011210860004674395,
      "loss": 1.8191,
      "step": 164838
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.2280346155166626,
      "learning_rate": 0.00011210541115034193,
      "loss": 2.0231,
      "step": 164839
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.3141995668411255,
      "learning_rate": 0.00011210222228887319,
      "loss": 2.2771,
      "step": 164840
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.2082270383834839,
      "learning_rate": 0.00011209903346233819,
      "loss": 2.3071,
      "step": 164841
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.1276898384094238,
      "learning_rate": 0.00011209584467073752,
      "loss": 2.2816,
      "step": 164842
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.1579455137252808,
      "learning_rate": 0.00011209265591407167,
      "loss": 1.9601,
      "step": 164843
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.1659971475601196,
      "learning_rate": 0.00011208946719234144,
      "loss": 2.0837,
      "step": 164844
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.2362558841705322,
      "learning_rate": 0.00011208627850554723,
      "loss": 1.813,
      "step": 164845
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.2272869348526,
      "learning_rate": 0.00011208308985368984,
      "loss": 2.2152,
      "step": 164846
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.3380587100982666,
      "learning_rate": 0.00011207990123676974,
      "loss": 2.1555,
      "step": 164847
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.2516447305679321,
      "learning_rate": 0.00011207671265478756,
      "loss": 2.0777,
      "step": 164848
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.2515745162963867,
      "learning_rate": 0.00011207352410774373,
      "loss": 2.3223,
      "step": 164849
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.2504587173461914,
      "learning_rate": 0.00011207033559563912,
      "loss": 1.8639,
      "step": 164850
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.1777056455612183,
      "learning_rate": 0.00011206714711847406,
      "loss": 1.9323,
      "step": 164851
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.2142359018325806,
      "learning_rate": 0.00011206395867624935,
      "loss": 1.9586,
      "step": 164852
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.2403714656829834,
      "learning_rate": 0.00011206077026896552,
      "loss": 1.9222,
      "step": 164853
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.167378544807434,
      "learning_rate": 0.00011205758189662301,
      "loss": 2.0795,
      "step": 164854
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.2077455520629883,
      "learning_rate": 0.00011205439355922268,
      "loss": 2.011,
      "step": 164855
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.5913434028625488,
      "learning_rate": 0.00011205120525676497,
      "loss": 2.2041,
      "step": 164856
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.1750552654266357,
      "learning_rate": 0.00011204801698925036,
      "loss": 2.1211,
      "step": 164857
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.2166835069656372,
      "learning_rate": 0.00011204482875667971,
      "loss": 2.0318,
      "step": 164858
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.2062532901763916,
      "learning_rate": 0.00011204164055905348,
      "loss": 2.2507,
      "step": 164859
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.487374186515808,
      "learning_rate": 0.00011203845239637213,
      "loss": 2.1151,
      "step": 164860
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.2932851314544678,
      "learning_rate": 0.00011203526426863648,
      "loss": 2.018,
      "step": 164861
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.3263565301895142,
      "learning_rate": 0.00011203207617584695,
      "loss": 1.939,
      "step": 164862
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.148859977722168,
      "learning_rate": 0.00011202888811800426,
      "loss": 2.0157,
      "step": 164863
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.4222540855407715,
      "learning_rate": 0.000112025700095109,
      "loss": 2.1666,
      "step": 164864
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.2961361408233643,
      "learning_rate": 0.0001120225121071617,
      "loss": 2.0219,
      "step": 164865
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.367600679397583,
      "learning_rate": 0.00011201932415416282,
      "loss": 2.1898,
      "step": 164866
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.2095961570739746,
      "learning_rate": 0.00011201613623611324,
      "loss": 2.0098,
      "step": 164867
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.2771756649017334,
      "learning_rate": 0.00011201294835301328,
      "loss": 2.1301,
      "step": 164868
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.3036396503448486,
      "learning_rate": 0.00011200976050486379,
      "loss": 1.9853,
      "step": 164869
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.2719398736953735,
      "learning_rate": 0.0001120065726916652,
      "loss": 1.9337,
      "step": 164870
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.1183651685714722,
      "learning_rate": 0.00011200338491341817,
      "loss": 1.8962,
      "step": 164871
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.3038541078567505,
      "learning_rate": 0.00011200019717012314,
      "loss": 2.0587,
      "step": 164872
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.2760298252105713,
      "learning_rate": 0.00011199700946178095,
      "loss": 2.0816,
      "step": 164873
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.2844536304473877,
      "learning_rate": 0.00011199382178839193,
      "loss": 1.8773,
      "step": 164874
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.339812159538269,
      "learning_rate": 0.00011199063414995693,
      "loss": 1.9002,
      "step": 164875
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.18430757522583,
      "learning_rate": 0.00011198744654647643,
      "loss": 1.9647,
      "step": 164876
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.2234041690826416,
      "learning_rate": 0.00011198425897795096,
      "loss": 1.9613,
      "step": 164877
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.1965080499649048,
      "learning_rate": 0.0001119810714443811,
      "loss": 2.1044,
      "step": 164878
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.1856533288955688,
      "learning_rate": 0.0001119778839457676,
      "loss": 2.1324,
      "step": 164879
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.205840826034546,
      "learning_rate": 0.0001119746964821109,
      "loss": 1.9498,
      "step": 164880
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.181302785873413,
      "learning_rate": 0.00011197150905341172,
      "loss": 2.0662,
      "step": 164881
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.2285374402999878,
      "learning_rate": 0.00011196832165967047,
      "loss": 2.1266,
      "step": 164882
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.4446384906768799,
      "learning_rate": 0.0001119651343008881,
      "loss": 2.1496,
      "step": 164883
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.2920788526535034,
      "learning_rate": 0.00011196194697706472,
      "loss": 2.0063,
      "step": 164884
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.1621476411819458,
      "learning_rate": 0.00011195875968820128,
      "loss": 2.0309,
      "step": 164885
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.1719073057174683,
      "learning_rate": 0.00011195557243429815,
      "loss": 1.9131,
      "step": 164886
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.3560960292816162,
      "learning_rate": 0.00011195238521535614,
      "loss": 2.2072,
      "step": 164887
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.1742558479309082,
      "learning_rate": 0.00011194919803137561,
      "loss": 1.8782,
      "step": 164888
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.0522247552871704,
      "learning_rate": 0.0001119460108823575,
      "loss": 1.7394,
      "step": 164889
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.2523629665374756,
      "learning_rate": 0.00011194282376830196,
      "loss": 1.9984,
      "step": 164890
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.202029824256897,
      "learning_rate": 0.0001119396366892099,
      "loss": 2.0605,
      "step": 164891
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.2436790466308594,
      "learning_rate": 0.00011193644964508173,
      "loss": 2.2436,
      "step": 164892
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.381239652633667,
      "learning_rate": 0.00011193326263591818,
      "loss": 1.9811,
      "step": 164893
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.3075014352798462,
      "learning_rate": 0.00011193007566171975,
      "loss": 2.0919,
      "step": 164894
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.435110330581665,
      "learning_rate": 0.00011192688872248723,
      "loss": 2.0486,
      "step": 164895
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.0927315950393677,
      "learning_rate": 0.00011192370181822084,
      "loss": 1.9238,
      "step": 164896
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.3772361278533936,
      "learning_rate": 0.00011192051494892153,
      "loss": 2.2164,
      "step": 164897
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.234062671661377,
      "learning_rate": 0.0001119173281145896,
      "loss": 2.318,
      "step": 164898
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.2440859079360962,
      "learning_rate": 0.0001119141413152259,
      "loss": 1.8215,
      "step": 164899
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.154745101928711,
      "learning_rate": 0.00011191095455083084,
      "loss": 1.9975,
      "step": 164900
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.2438421249389648,
      "learning_rate": 0.00011190776782140526,
      "loss": 2.0072,
      "step": 164901
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.3460570573806763,
      "learning_rate": 0.00011190458112694938,
      "loss": 2.0821,
      "step": 164902
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.1455551385879517,
      "learning_rate": 0.00011190139446746406,
      "loss": 2.0245,
      "step": 164903
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.4043359756469727,
      "learning_rate": 0.00011189820784294977,
      "loss": 2.1048,
      "step": 164904
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.4262596368789673,
      "learning_rate": 0.00011189502125340722,
      "loss": 2.0078,
      "step": 164905
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.127736210823059,
      "learning_rate": 0.00011189183469883684,
      "loss": 2.0189,
      "step": 164906
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.290776252746582,
      "learning_rate": 0.00011188864817923955,
      "loss": 1.9249,
      "step": 164907
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.535966157913208,
      "learning_rate": 0.00011188546169461546,
      "loss": 2.1908,
      "step": 164908
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.3324705362319946,
      "learning_rate": 0.00011188227524496556,
      "loss": 2.0501,
      "step": 164909
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.3275914192199707,
      "learning_rate": 0.00011187908883029017,
      "loss": 2.1508,
      "step": 164910
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.3542298078536987,
      "learning_rate": 0.00011187590245059011,
      "loss": 1.9214,
      "step": 164911
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.2305632829666138,
      "learning_rate": 0.00011187271610586578,
      "loss": 1.989,
      "step": 164912
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.1056797504425049,
      "learning_rate": 0.00011186952979611809,
      "loss": 1.9072,
      "step": 164913
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.2566657066345215,
      "learning_rate": 0.00011186634352134717,
      "loss": 2.1828,
      "step": 164914
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.1717535257339478,
      "learning_rate": 0.00011186315728155396,
      "loss": 1.9547,
      "step": 164915
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.1689412593841553,
      "learning_rate": 0.00011185997107673881,
      "loss": 2.0085,
      "step": 164916
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.371599793434143,
      "learning_rate": 0.00011185678490690257,
      "loss": 2.0857,
      "step": 164917
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.325851321220398,
      "learning_rate": 0.00011185359877204563,
      "loss": 2.2126,
      "step": 164918
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.365077018737793,
      "learning_rate": 0.00011185041267216884,
      "loss": 2.0999,
      "step": 164919
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.325548768043518,
      "learning_rate": 0.00011184722660727239,
      "loss": 2.1791,
      "step": 164920
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.3124239444732666,
      "learning_rate": 0.0001118440405773572,
      "loss": 1.9848,
      "step": 164921
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.1819827556610107,
      "learning_rate": 0.00011184085458242365,
      "loss": 1.9165,
      "step": 164922
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.1087554693222046,
      "learning_rate": 0.00011183766862247253,
      "loss": 1.9916,
      "step": 164923
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.2719963788986206,
      "learning_rate": 0.00011183448269750425,
      "loss": 1.993,
      "step": 164924
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.1381590366363525,
      "learning_rate": 0.0001118312968075196,
      "loss": 2.0387,
      "step": 164925
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.287074327468872,
      "learning_rate": 0.00011182811095251905,
      "loss": 2.0742,
      "step": 164926
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.283611536026001,
      "learning_rate": 0.00011182492513250323,
      "loss": 1.9529,
      "step": 164927
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.2548470497131348,
      "learning_rate": 0.00011182173934747258,
      "loss": 2.1095,
      "step": 164928
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.3701348304748535,
      "learning_rate": 0.00011181855359742796,
      "loss": 2.2307,
      "step": 164929
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.1985291242599487,
      "learning_rate": 0.0001118153678823697,
      "loss": 2.0109,
      "step": 164930
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.1472574472427368,
      "learning_rate": 0.0001118121822022986,
      "loss": 1.9253,
      "step": 164931
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.3067702054977417,
      "learning_rate": 0.0001118089965572152,
      "loss": 2.168,
      "step": 164932
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.2427587509155273,
      "learning_rate": 0.00011180581094712002,
      "loss": 1.8452,
      "step": 164933
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.2967371940612793,
      "learning_rate": 0.00011180262537201363,
      "loss": 2.0128,
      "step": 164934
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.0591485500335693,
      "learning_rate": 0.00011179943983189677,
      "loss": 1.9776,
      "step": 164935
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.3865855932235718,
      "learning_rate": 0.00011179625432676982,
      "loss": 1.8995,
      "step": 164936
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.2321261167526245,
      "learning_rate": 0.00011179306885663364,
      "loss": 2.1691,
      "step": 164937
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.2863048315048218,
      "learning_rate": 0.00011178988342148866,
      "loss": 1.9714,
      "step": 164938
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.1885547637939453,
      "learning_rate": 0.00011178669802133541,
      "loss": 2.0924,
      "step": 164939
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.4126853942871094,
      "learning_rate": 0.00011178351265617463,
      "loss": 2.0457,
      "step": 164940
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.2928531169891357,
      "learning_rate": 0.00011178032732600687,
      "loss": 2.127,
      "step": 164941
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.241120457649231,
      "learning_rate": 0.00011177714203083261,
      "loss": 1.9516,
      "step": 164942
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.3100091218948364,
      "learning_rate": 0.00011177395677065263,
      "loss": 2.092,
      "step": 164943
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.257203221321106,
      "learning_rate": 0.00011177077154546742,
      "loss": 1.9731,
      "step": 164944
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.2394057512283325,
      "learning_rate": 0.00011176758635527744,
      "loss": 2.0662,
      "step": 164945
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.2455400228500366,
      "learning_rate": 0.00011176440120008358,
      "loss": 2.0104,
      "step": 164946
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.3254384994506836,
      "learning_rate": 0.00011176121607988612,
      "loss": 2.1022,
      "step": 164947
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.3625560998916626,
      "learning_rate": 0.00011175803099468594,
      "loss": 1.9607,
      "step": 164948
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.1927452087402344,
      "learning_rate": 0.00011175484594448349,
      "loss": 1.9142,
      "step": 164949
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.6522711515426636,
      "learning_rate": 0.00011175166092927936,
      "loss": 2.0987,
      "step": 164950
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.6885648965835571,
      "learning_rate": 0.00011174847594907404,
      "loss": 2.1208,
      "step": 164951
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.2287635803222656,
      "learning_rate": 0.00011174529100386834,
      "loss": 2.2701,
      "step": 164952
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.4868816137313843,
      "learning_rate": 0.00011174210609366268,
      "loss": 1.99,
      "step": 164953
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.5283393859863281,
      "learning_rate": 0.00011173892121845778,
      "loss": 2.0041,
      "step": 164954
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.3013635873794556,
      "learning_rate": 0.00011173573637825415,
      "loss": 2.068,
      "step": 164955
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.2692614793777466,
      "learning_rate": 0.00011173255157305242,
      "loss": 2.185,
      "step": 164956
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.3386356830596924,
      "learning_rate": 0.00011172936680285307,
      "loss": 1.9952,
      "step": 164957
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.44248366355896,
      "learning_rate": 0.00011172618206765687,
      "loss": 2.1074,
      "step": 164958
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.3340500593185425,
      "learning_rate": 0.00011172299736746424,
      "loss": 1.9344,
      "step": 164959
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.3820171356201172,
      "learning_rate": 0.00011171981270227595,
      "loss": 1.9133,
      "step": 164960
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.0452461242675781,
      "learning_rate": 0.00011171662807209242,
      "loss": 1.8897,
      "step": 164961
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.2691010236740112,
      "learning_rate": 0.00011171344347691449,
      "loss": 2.0077,
      "step": 164962
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.3001631498336792,
      "learning_rate": 0.00011171025891674243,
      "loss": 2.2124,
      "step": 164963
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.2332547903060913,
      "learning_rate": 0.00011170707439157705,
      "loss": 1.9373,
      "step": 164964
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.2400623559951782,
      "learning_rate": 0.00011170388990141879,
      "loss": 2.0429,
      "step": 164965
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.1671433448791504,
      "learning_rate": 0.00011170070544626841,
      "loss": 1.9625,
      "step": 164966
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.2075790166854858,
      "learning_rate": 0.00011169752102612633,
      "loss": 2.0653,
      "step": 164967
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.3487716913223267,
      "learning_rate": 0.00011169433664099349,
      "loss": 1.9932,
      "step": 164968
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.267049789428711,
      "learning_rate": 0.00011169115229086996,
      "loss": 2.169,
      "step": 164969
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.211788535118103,
      "learning_rate": 0.00011168796797575675,
      "loss": 2.0805,
      "step": 164970
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.3060277700424194,
      "learning_rate": 0.00011168478369565414,
      "loss": 1.8759,
      "step": 164971
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.2534016370773315,
      "learning_rate": 0.00011168159945056304,
      "loss": 2.0919,
      "step": 164972
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.2344642877578735,
      "learning_rate": 0.00011167841524048376,
      "loss": 2.1484,
      "step": 164973
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.4137003421783447,
      "learning_rate": 0.00011167523106541721,
      "loss": 2.4217,
      "step": 164974
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.2704064846038818,
      "learning_rate": 0.00011167204692536362,
      "loss": 1.8895,
      "step": 164975
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.1062023639678955,
      "learning_rate": 0.00011166886282032382,
      "loss": 2.1329,
      "step": 164976
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.1930114030838013,
      "learning_rate": 0.00011166567875029824,
      "loss": 2.0294,
      "step": 164977
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.4314517974853516,
      "learning_rate": 0.0001116624947152877,
      "loss": 1.9493,
      "step": 164978
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.309455394744873,
      "learning_rate": 0.00011165931071529251,
      "loss": 2.0836,
      "step": 164979
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.1875232458114624,
      "learning_rate": 0.00011165612675031363,
      "loss": 1.9763,
      "step": 164980
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.2399004697799683,
      "learning_rate": 0.00011165294282035126,
      "loss": 2.1791,
      "step": 164981
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.107105016708374,
      "learning_rate": 0.00011164975892540623,
      "loss": 2.0205,
      "step": 164982
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.2911618947982788,
      "learning_rate": 0.00011164657506547895,
      "loss": 2.0285,
      "step": 164983
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.34837007522583,
      "learning_rate": 0.00011164339124057024,
      "loss": 1.7174,
      "step": 164984
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.3199560642242432,
      "learning_rate": 0.00011164020745068049,
      "loss": 1.9689,
      "step": 164985
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.202399730682373,
      "learning_rate": 0.00011163702369581057,
      "loss": 2.1442,
      "step": 164986
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.2025763988494873,
      "learning_rate": 0.00011163383997596069,
      "loss": 2.07,
      "step": 164987
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.3606250286102295,
      "learning_rate": 0.00011163065629113176,
      "loss": 1.9447,
      "step": 164988
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.2620408535003662,
      "learning_rate": 0.00011162747264132412,
      "loss": 2.0809,
      "step": 164989
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.2597465515136719,
      "learning_rate": 0.00011162428902653858,
      "loss": 1.82,
      "step": 164990
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.1769920587539673,
      "learning_rate": 0.00011162110544677553,
      "loss": 2.089,
      "step": 164991
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.2776038646697998,
      "learning_rate": 0.00011161792190203591,
      "loss": 1.9535,
      "step": 164992
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.4741418361663818,
      "learning_rate": 0.00011161473839231987,
      "loss": 2.0088,
      "step": 164993
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.249177098274231,
      "learning_rate": 0.00011161155491762827,
      "loss": 2.0742,
      "step": 164994
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.3051971197128296,
      "learning_rate": 0.00011160837147796157,
      "loss": 2.097,
      "step": 164995
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.451686143875122,
      "learning_rate": 0.00011160518807332051,
      "loss": 2.1735,
      "step": 164996
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.4745041131973267,
      "learning_rate": 0.00011160200470370553,
      "loss": 1.9955,
      "step": 164997
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.117340087890625,
      "learning_rate": 0.00011159882136911747,
      "loss": 1.986,
      "step": 164998
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.2195460796356201,
      "learning_rate": 0.00011159563806955655,
      "loss": 2.1115,
      "step": 164999
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.3000506162643433,
      "learning_rate": 0.00011159245480502365,
      "loss": 2.0673,
      "step": 165000
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.2425884008407593,
      "learning_rate": 0.00011158927157551919,
      "loss": 2.0006,
      "step": 165001
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.1879163980484009,
      "learning_rate": 0.00011158608838104391,
      "loss": 2.0194,
      "step": 165002
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.2428785562515259,
      "learning_rate": 0.00011158290522159824,
      "loss": 1.8179,
      "step": 165003
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.385493516921997,
      "learning_rate": 0.0001115797220971831,
      "loss": 1.9455,
      "step": 165004
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.292950987815857,
      "learning_rate": 0.00011157653900779863,
      "loss": 2.253,
      "step": 165005
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.1958982944488525,
      "learning_rate": 0.00011157335595344573,
      "loss": 1.8151,
      "step": 165006
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.272298812866211,
      "learning_rate": 0.00011157017293412482,
      "loss": 2.2422,
      "step": 165007
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.1302993297576904,
      "learning_rate": 0.00011156698994983663,
      "loss": 1.8448,
      "step": 165008
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.327278733253479,
      "learning_rate": 0.00011156380700058161,
      "loss": 2.1597,
      "step": 165009
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.1936393976211548,
      "learning_rate": 0.00011156062408636066,
      "loss": 2.0338,
      "step": 165010
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.223414659500122,
      "learning_rate": 0.00011155744120717391,
      "loss": 1.8779,
      "step": 165011
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.2184334993362427,
      "learning_rate": 0.00011155425836302232,
      "loss": 2.0128,
      "step": 165012
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.252929925918579,
      "learning_rate": 0.00011155107555390623,
      "loss": 1.9554,
      "step": 165013
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.1489734649658203,
      "learning_rate": 0.00011154789277982647,
      "loss": 2.0744,
      "step": 165014
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.0453852415084839,
      "learning_rate": 0.00011154471004078339,
      "loss": 1.8381,
      "step": 165015
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.3306334018707275,
      "learning_rate": 0.00011154152733677781,
      "loss": 2.0763,
      "step": 165016
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.3486480712890625,
      "learning_rate": 0.00011153834466781024,
      "loss": 2.0167,
      "step": 165017
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.2582846879959106,
      "learning_rate": 0.00011153516203388125,
      "loss": 1.9401,
      "step": 165018
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.277405023574829,
      "learning_rate": 0.00011153197943499131,
      "loss": 2.1579,
      "step": 165019
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.2136491537094116,
      "learning_rate": 0.00011152879687114122,
      "loss": 2.1713,
      "step": 165020
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.1848934888839722,
      "learning_rate": 0.00011152561434233144,
      "loss": 1.9767,
      "step": 165021
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.2268900871276855,
      "learning_rate": 0.00011152243184856265,
      "loss": 2.1206,
      "step": 165022
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.394936442375183,
      "learning_rate": 0.0001115192493898354,
      "loss": 1.8066,
      "step": 165023
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.1218764781951904,
      "learning_rate": 0.00011151606696615023,
      "loss": 1.7923,
      "step": 165024
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.2480303049087524,
      "learning_rate": 0.00011151288457750785,
      "loss": 2.1364,
      "step": 165025
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.2302151918411255,
      "learning_rate": 0.00011150970222390877,
      "loss": 1.9106,
      "step": 165026
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.1574792861938477,
      "learning_rate": 0.00011150651990535351,
      "loss": 1.6948,
      "step": 165027
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.2430541515350342,
      "learning_rate": 0.00011150333762184284,
      "loss": 2.103,
      "step": 165028
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.2697213888168335,
      "learning_rate": 0.00011150015537337726,
      "loss": 1.9845,
      "step": 165029
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.1127734184265137,
      "learning_rate": 0.00011149697315995726,
      "loss": 1.8673,
      "step": 165030
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.4726433753967285,
      "learning_rate": 0.00011149379098158361,
      "loss": 2.0714,
      "step": 165031
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.334329605102539,
      "learning_rate": 0.00011149060883825685,
      "loss": 2.4015,
      "step": 165032
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.324507236480713,
      "learning_rate": 0.00011148742672997743,
      "loss": 2.04,
      "step": 165033
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.1392390727996826,
      "learning_rate": 0.00011148424465674619,
      "loss": 2.0818,
      "step": 165034
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.383800983428955,
      "learning_rate": 0.00011148106261856354,
      "loss": 2.222,
      "step": 165035
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.1545535326004028,
      "learning_rate": 0.00011147788061543005,
      "loss": 2.1131,
      "step": 165036
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.225809097290039,
      "learning_rate": 0.00011147469864734647,
      "loss": 2.0196,
      "step": 165037
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.22307550907135,
      "learning_rate": 0.00011147151671431319,
      "loss": 1.9477,
      "step": 165038
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.1372634172439575,
      "learning_rate": 0.00011146833481633102,
      "loss": 1.9129,
      "step": 165039
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.2496700286865234,
      "learning_rate": 0.00011146515295340045,
      "loss": 2.0624,
      "step": 165040
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.4459455013275146,
      "learning_rate": 0.00011146197112552208,
      "loss": 2.2555,
      "step": 165041
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.283339500427246,
      "learning_rate": 0.00011145878933269636,
      "loss": 2.0403,
      "step": 165042
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.252940058708191,
      "learning_rate": 0.00011145560757492414,
      "loss": 1.642,
      "step": 165043
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.3365308046340942,
      "learning_rate": 0.00011145242585220578,
      "loss": 1.9276,
      "step": 165044
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.1929436922073364,
      "learning_rate": 0.00011144924416454206,
      "loss": 1.8991,
      "step": 165045
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.1523492336273193,
      "learning_rate": 0.00011144606251193338,
      "loss": 2.0892,
      "step": 165046
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.4741262197494507,
      "learning_rate": 0.00011144288089438065,
      "loss": 2.1126,
      "step": 165047
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.2142703533172607,
      "learning_rate": 0.00011143969931188404,
      "loss": 1.8955,
      "step": 165048
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.2574162483215332,
      "learning_rate": 0.00011143651776444443,
      "loss": 2.2128,
      "step": 165049
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.3510090112686157,
      "learning_rate": 0.00011143333625206226,
      "loss": 1.9778,
      "step": 165050
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.186218023300171,
      "learning_rate": 0.00011143015477473827,
      "loss": 1.9859,
      "step": 165051
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.7596709728240967,
      "learning_rate": 0.0001114269733324729,
      "loss": 2.0948,
      "step": 165052
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.3890056610107422,
      "learning_rate": 0.00011142379192526699,
      "loss": 2.0775,
      "step": 165053
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.4134376049041748,
      "learning_rate": 0.00011142061055312077,
      "loss": 2.0969,
      "step": 165054
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.2445592880249023,
      "learning_rate": 0.00011141742921603513,
      "loss": 2.1484,
      "step": 165055
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.280571460723877,
      "learning_rate": 0.00011141424791401042,
      "loss": 2.0905,
      "step": 165056
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.2362322807312012,
      "learning_rate": 0.0001114110666470475,
      "loss": 1.8764,
      "step": 165057
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.524288296699524,
      "learning_rate": 0.00011140788541514669,
      "loss": 1.8542,
      "step": 165058
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.1915156841278076,
      "learning_rate": 0.0001114047042183089,
      "loss": 1.985,
      "step": 165059
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.3292236328125,
      "learning_rate": 0.00011140152305653436,
      "loss": 2.2113,
      "step": 165060
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.2340097427368164,
      "learning_rate": 0.00011139834192982395,
      "loss": 2.0671,
      "step": 165061
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.2652353048324585,
      "learning_rate": 0.00011139516083817802,
      "loss": 1.9214,
      "step": 165062
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.3507070541381836,
      "learning_rate": 0.00011139197978159741,
      "loss": 2.318,
      "step": 165063
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.2431329488754272,
      "learning_rate": 0.00011138879876008252,
      "loss": 1.9949,
      "step": 165064
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.2307298183441162,
      "learning_rate": 0.00011138561777363416,
      "loss": 2.0232,
      "step": 165065
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.2548904418945312,
      "learning_rate": 0.0001113824368222526,
      "loss": 2.1137,
      "step": 165066
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.3511403799057007,
      "learning_rate": 0.00011137925590593868,
      "loss": 2.0072,
      "step": 165067
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.3342549800872803,
      "learning_rate": 0.00011137607502469285,
      "loss": 2.055,
      "step": 165068
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.369338035583496,
      "learning_rate": 0.00011137289417851582,
      "loss": 1.964,
      "step": 165069
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.1590250730514526,
      "learning_rate": 0.00011136971336740808,
      "loss": 2.0066,
      "step": 165070
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.1784509420394897,
      "learning_rate": 0.00011136653259137046,
      "loss": 2.1035,
      "step": 165071
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.3221955299377441,
      "learning_rate": 0.00011136335185040315,
      "loss": 1.9282,
      "step": 165072
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.1257293224334717,
      "learning_rate": 0.00011136017114450705,
      "loss": 1.7452,
      "step": 165073
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.2518672943115234,
      "learning_rate": 0.00011135699047368256,
      "loss": 1.9328,
      "step": 165074
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.2727198600769043,
      "learning_rate": 0.00011135380983793048,
      "loss": 2.1139,
      "step": 165075
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.1821892261505127,
      "learning_rate": 0.0001113506292372512,
      "loss": 2.1282,
      "step": 165076
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.1041738986968994,
      "learning_rate": 0.00011134744867164556,
      "loss": 2.13,
      "step": 165077
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.2466522455215454,
      "learning_rate": 0.00011134426814111383,
      "loss": 2.2317,
      "step": 165078
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.2349687814712524,
      "learning_rate": 0.00011134108764565682,
      "loss": 2.2418,
      "step": 165079
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.2661751508712769,
      "learning_rate": 0.00011133790718527496,
      "loss": 2.0576,
      "step": 165080
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.1953182220458984,
      "learning_rate": 0.00011133472675996909,
      "loss": 2.2969,
      "step": 165081
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.1873124837875366,
      "learning_rate": 0.00011133154636973951,
      "loss": 1.9856,
      "step": 165082
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.242357850074768,
      "learning_rate": 0.00011132836601458712,
      "loss": 2.001,
      "step": 165083
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.2722976207733154,
      "learning_rate": 0.0001113251856945123,
      "loss": 1.9741,
      "step": 165084
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.1444780826568604,
      "learning_rate": 0.00011132200540951567,
      "loss": 1.996,
      "step": 165085
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.1827702522277832,
      "learning_rate": 0.00011131882515959779,
      "loss": 1.9066,
      "step": 165086
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.2011324167251587,
      "learning_rate": 0.00011131564494475936,
      "loss": 1.9779,
      "step": 165087
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.2029534578323364,
      "learning_rate": 0.00011131246476500085,
      "loss": 2.3209,
      "step": 165088
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.1871222257614136,
      "learning_rate": 0.000111309284620323,
      "loss": 2.0184,
      "step": 165089
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.2208681106567383,
      "learning_rate": 0.00011130610451072631,
      "loss": 1.9095,
      "step": 165090
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.1430095434188843,
      "learning_rate": 0.00011130292443621138,
      "loss": 2.0497,
      "step": 165091
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.362022876739502,
      "learning_rate": 0.0001112997443967787,
      "loss": 1.9925,
      "step": 165092
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.2575901746749878,
      "learning_rate": 0.00011129656439242906,
      "loss": 1.9447,
      "step": 165093
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.26068115234375,
      "learning_rate": 0.00011129338442316285,
      "loss": 1.9842,
      "step": 165094
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.3540226221084595,
      "learning_rate": 0.00011129020448898084,
      "loss": 2.0562,
      "step": 165095
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.3755663633346558,
      "learning_rate": 0.00011128702458988356,
      "loss": 1.9412,
      "step": 165096
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.338465690612793,
      "learning_rate": 0.00011128384472587159,
      "loss": 1.8938,
      "step": 165097
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.1903730630874634,
      "learning_rate": 0.00011128066489694542,
      "loss": 1.8713,
      "step": 165098
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.3252415657043457,
      "learning_rate": 0.00011127748510310582,
      "loss": 2.0915,
      "step": 165099
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.294696569442749,
      "learning_rate": 0.00011127430534435321,
      "loss": 2.0678,
      "step": 165100
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.3526716232299805,
      "learning_rate": 0.00011127112562068838,
      "loss": 2.0114,
      "step": 165101
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.240097165107727,
      "learning_rate": 0.00011126794593211182,
      "loss": 2.1669,
      "step": 165102
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.3523200750350952,
      "learning_rate": 0.00011126476627862407,
      "loss": 1.8615,
      "step": 165103
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.2254395484924316,
      "learning_rate": 0.00011126158666022567,
      "loss": 2.0661,
      "step": 165104
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.2219913005828857,
      "learning_rate": 0.00011125840707691742,
      "loss": 1.9434,
      "step": 165105
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.224061131477356,
      "learning_rate": 0.0001112552275286997,
      "loss": 1.9935,
      "step": 165106
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.4473692178726196,
      "learning_rate": 0.00011125204801557329,
      "loss": 2.2422,
      "step": 165107
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.365218997001648,
      "learning_rate": 0.00011124886853753867,
      "loss": 1.9008,
      "step": 165108
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.2263356447219849,
      "learning_rate": 0.00011124568909459638,
      "loss": 2.0885,
      "step": 165109
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.1190063953399658,
      "learning_rate": 0.00011124250968674716,
      "loss": 2.0088,
      "step": 165110
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.3753070831298828,
      "learning_rate": 0.00011123933031399151,
      "loss": 2.1496,
      "step": 165111
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.1840077638626099,
      "learning_rate": 0.00011123615097632994,
      "loss": 2.1135,
      "step": 165112
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.538699746131897,
      "learning_rate": 0.00011123297167376323,
      "loss": 2.3535,
      "step": 165113
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.2612143754959106,
      "learning_rate": 0.00011122979240629191,
      "loss": 1.9221,
      "step": 165114
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.1350111961364746,
      "learning_rate": 0.00011122661317391638,
      "loss": 1.9439,
      "step": 165115
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.3170970678329468,
      "learning_rate": 0.0001112234339766375,
      "loss": 2.0625,
      "step": 165116
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.1930750608444214,
      "learning_rate": 0.00011122025481445577,
      "loss": 2.1962,
      "step": 165117
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.2731343507766724,
      "learning_rate": 0.00011121707568737164,
      "loss": 1.9822,
      "step": 165118
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.2214150428771973,
      "learning_rate": 0.00011121389659538596,
      "loss": 1.8789,
      "step": 165119
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.2422205209732056,
      "learning_rate": 0.00011121071753849912,
      "loss": 1.9143,
      "step": 165120
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.3168978691101074,
      "learning_rate": 0.0001112075385167117,
      "loss": 2.072,
      "step": 165121
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.4701769351959229,
      "learning_rate": 0.00011120435953002445,
      "loss": 2.2394,
      "step": 165122
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.2264437675476074,
      "learning_rate": 0.00011120118057843778,
      "loss": 1.7183,
      "step": 165123
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.201217770576477,
      "learning_rate": 0.00011119800166195249,
      "loss": 2.2059,
      "step": 165124
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.1685400009155273,
      "learning_rate": 0.00011119482278056904,
      "loss": 1.7654,
      "step": 165125
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.3300957679748535,
      "learning_rate": 0.00011119164393428804,
      "loss": 2.1744,
      "step": 165126
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.4141972064971924,
      "learning_rate": 0.00011118846512310998,
      "loss": 2.2082,
      "step": 165127
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.3862828016281128,
      "learning_rate": 0.00011118528634703566,
      "loss": 2.0894,
      "step": 165128
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.085670828819275,
      "learning_rate": 0.00011118210760606544,
      "loss": 1.8689,
      "step": 165129
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.1126816272735596,
      "learning_rate": 0.00011117892890020011,
      "loss": 1.8989,
      "step": 165130
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.3719204664230347,
      "learning_rate": 0.00011117575022944021,
      "loss": 2.243,
      "step": 165131
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.169136643409729,
      "learning_rate": 0.00011117257159378633,
      "loss": 2.1525,
      "step": 165132
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.3770290613174438,
      "learning_rate": 0.00011116939299323888,
      "loss": 2.1944,
      "step": 165133
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.16963791847229,
      "learning_rate": 0.00011116621442779873,
      "loss": 1.928,
      "step": 165134
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.1389286518096924,
      "learning_rate": 0.00011116303589746623,
      "loss": 2.0388,
      "step": 165135
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.2305974960327148,
      "learning_rate": 0.0001111598574022422,
      "loss": 1.9698,
      "step": 165136
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.2593191862106323,
      "learning_rate": 0.000111156678942127,
      "loss": 1.9796,
      "step": 165137
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.2270171642303467,
      "learning_rate": 0.00011115350051712158,
      "loss": 2.1252,
      "step": 165138
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.175209641456604,
      "learning_rate": 0.00011115032212722606,
      "loss": 1.9548,
      "step": 165139
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.239857792854309,
      "learning_rate": 0.00011114714377244134,
      "loss": 2.0733,
      "step": 165140
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.2392024993896484,
      "learning_rate": 0.00011114396545276786,
      "loss": 2.0206,
      "step": 165141
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.4286773204803467,
      "learning_rate": 0.00011114078716820639,
      "loss": 2.0723,
      "step": 165142
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.1879005432128906,
      "learning_rate": 0.00011113760891875727,
      "loss": 1.7711,
      "step": 165143
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.1550120115280151,
      "learning_rate": 0.00011113443070442147,
      "loss": 2.333,
      "step": 165144
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.4852924346923828,
      "learning_rate": 0.00011113125252519913,
      "loss": 2.01,
      "step": 165145
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.3699805736541748,
      "learning_rate": 0.00011112807438109118,
      "loss": 2.1435,
      "step": 165146
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.8946138620376587,
      "learning_rate": 0.00011112489627209794,
      "loss": 1.8779,
      "step": 165147
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.1814931631088257,
      "learning_rate": 0.00011112171819822028,
      "loss": 2.1754,
      "step": 165148
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.1679607629776,
      "learning_rate": 0.00011111854015945856,
      "loss": 2.2009,
      "step": 165149
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.2618108987808228,
      "learning_rate": 0.00011111536215581355,
      "loss": 2.2369,
      "step": 165150
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.2233257293701172,
      "learning_rate": 0.00011111218418728577,
      "loss": 2.0103,
      "step": 165151
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.141615390777588,
      "learning_rate": 0.00011110900625387582,
      "loss": 1.819,
      "step": 165152
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.1910061836242676,
      "learning_rate": 0.00011110582835558413,
      "loss": 2.148,
      "step": 165153
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.2810767889022827,
      "learning_rate": 0.00011110265049241156,
      "loss": 2.1857,
      "step": 165154
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.2684414386749268,
      "learning_rate": 0.00011109947266435846,
      "loss": 2.1679,
      "step": 165155
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.3171281814575195,
      "learning_rate": 0.00011109629487142567,
      "loss": 2.0401,
      "step": 165156
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.2005609273910522,
      "learning_rate": 0.00011109311711361361,
      "loss": 1.8344,
      "step": 165157
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.3465852737426758,
      "learning_rate": 0.00011108993939092288,
      "loss": 2.14,
      "step": 165158
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.3129574060440063,
      "learning_rate": 0.00011108676170335405,
      "loss": 1.969,
      "step": 165159
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.2347605228424072,
      "learning_rate": 0.00011108358405090784,
      "loss": 1.9444,
      "step": 165160
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.2180891036987305,
      "learning_rate": 0.00011108040643358464,
      "loss": 1.9754,
      "step": 165161
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.343244194984436,
      "learning_rate": 0.00011107722885138526,
      "loss": 2.0986,
      "step": 165162
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.2203916311264038,
      "learning_rate": 0.00011107405130431021,
      "loss": 1.9997,
      "step": 165163
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.3817782402038574,
      "learning_rate": 0.00011107087379236005,
      "loss": 2.0735,
      "step": 165164
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.256462574005127,
      "learning_rate": 0.00011106769631553529,
      "loss": 2.3136,
      "step": 165165
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.3964594602584839,
      "learning_rate": 0.00011106451887383671,
      "loss": 2.0604,
      "step": 165166
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.2018603086471558,
      "learning_rate": 0.00011106134146726469,
      "loss": 1.8968,
      "step": 165167
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.1812255382537842,
      "learning_rate": 0.00011105816409582004,
      "loss": 2.1484,
      "step": 165168
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.3519091606140137,
      "learning_rate": 0.00011105498675950322,
      "loss": 2.0785,
      "step": 165169
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.3334479331970215,
      "learning_rate": 0.00011105180945831486,
      "loss": 2.0096,
      "step": 165170
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.2110013961791992,
      "learning_rate": 0.00011104863219225545,
      "loss": 2.0882,
      "step": 165171
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.1439707279205322,
      "learning_rate": 0.00011104545496132577,
      "loss": 2.0555,
      "step": 165172
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.313315987586975,
      "learning_rate": 0.00011104227776552618,
      "loss": 1.8742,
      "step": 165173
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.2171803712844849,
      "learning_rate": 0.00011103910060485753,
      "loss": 1.9794,
      "step": 165174
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.363176703453064,
      "learning_rate": 0.00011103592347932026,
      "loss": 2.0591,
      "step": 165175
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.1692979335784912,
      "learning_rate": 0.00011103274638891498,
      "loss": 1.7627,
      "step": 165176
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.7315902709960938,
      "learning_rate": 0.00011102956933364217,
      "loss": 2.1226,
      "step": 165177
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.2191100120544434,
      "learning_rate": 0.00011102639231350264,
      "loss": 2.2605,
      "step": 165178
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.0861058235168457,
      "learning_rate": 0.00011102321532849677,
      "loss": 2.0356,
      "step": 165179
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.1982461214065552,
      "learning_rate": 0.00011102003837862533,
      "loss": 1.9274,
      "step": 165180
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.1754924058914185,
      "learning_rate": 0.00011101686146388885,
      "loss": 1.8353,
      "step": 165181
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.1778078079223633,
      "learning_rate": 0.00011101368458428793,
      "loss": 2.0166,
      "step": 165182
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.1875874996185303,
      "learning_rate": 0.000111010507739823,
      "loss": 1.9677,
      "step": 165183
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.114606499671936,
      "learning_rate": 0.00011100733093049491,
      "loss": 2.175,
      "step": 165184
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.1108421087265015,
      "learning_rate": 0.000111004154156304,
      "loss": 1.8781,
      "step": 165185
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.2422949075698853,
      "learning_rate": 0.00011100097741725109,
      "loss": 1.9644,
      "step": 165186
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.2359809875488281,
      "learning_rate": 0.00011099780071333667,
      "loss": 2.0937,
      "step": 165187
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.2830637693405151,
      "learning_rate": 0.00011099462404456131,
      "loss": 2.0214,
      "step": 165188
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.3185259103775024,
      "learning_rate": 0.00011099144741092554,
      "loss": 1.9397,
      "step": 165189
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.2639167308807373,
      "learning_rate": 0.00011098827081243013,
      "loss": 2.2108,
      "step": 165190
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.1957545280456543,
      "learning_rate": 0.00011098509424907543,
      "loss": 1.8412,
      "step": 165191
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.4017795324325562,
      "learning_rate": 0.0001109819177208623,
      "loss": 1.8619,
      "step": 165192
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.1672521829605103,
      "learning_rate": 0.00011097874122779123,
      "loss": 2.0322,
      "step": 165193
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.2850464582443237,
      "learning_rate": 0.00011097556476986274,
      "loss": 2.179,
      "step": 165194
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.1825518608093262,
      "learning_rate": 0.00011097238834707736,
      "loss": 2.0507,
      "step": 165195
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.3009214401245117,
      "learning_rate": 0.00011096921195943588,
      "loss": 2.3017,
      "step": 165196
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.175902247428894,
      "learning_rate": 0.0001109660356069387,
      "loss": 1.9115,
      "step": 165197
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.2078951597213745,
      "learning_rate": 0.00011096285928958663,
      "loss": 1.922,
      "step": 165198
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.301749348640442,
      "learning_rate": 0.00011095968300738007,
      "loss": 2.1896,
      "step": 165199
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.0682930946350098,
      "learning_rate": 0.0001109565067603196,
      "loss": 2.0791,
      "step": 165200
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.3854742050170898,
      "learning_rate": 0.00011095333054840602,
      "loss": 2.1749,
      "step": 165201
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.1790581941604614,
      "learning_rate": 0.00011095015437163977,
      "loss": 2.1174,
      "step": 165202
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.2366763353347778,
      "learning_rate": 0.00011094697823002132,
      "loss": 2.0428,
      "step": 165203
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.3865489959716797,
      "learning_rate": 0.00011094380212355152,
      "loss": 1.8943,
      "step": 165204
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.2933523654937744,
      "learning_rate": 0.00011094062605223085,
      "loss": 2.2292,
      "step": 165205
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.2820419073104858,
      "learning_rate": 0.00011093745001605976,
      "loss": 2.0106,
      "step": 165206
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.4339083433151245,
      "learning_rate": 0.00011093427401503909,
      "loss": 2.1858,
      "step": 165207
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.51681387424469,
      "learning_rate": 0.00011093109804916919,
      "loss": 2.0463,
      "step": 165208
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.3228474855422974,
      "learning_rate": 0.00011092792211845087,
      "loss": 2.2304,
      "step": 165209
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.4796134233474731,
      "learning_rate": 0.0001109247462228846,
      "loss": 1.8971,
      "step": 165210
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.261028528213501,
      "learning_rate": 0.00011092157036247103,
      "loss": 1.9626,
      "step": 165211
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.257586121559143,
      "learning_rate": 0.00011091839453721058,
      "loss": 2.0904,
      "step": 165212
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.1522502899169922,
      "learning_rate": 0.00011091521874710407,
      "loss": 1.9877,
      "step": 165213
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.2379990816116333,
      "learning_rate": 0.00011091204299215188,
      "loss": 2.0026,
      "step": 165214
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.2705459594726562,
      "learning_rate": 0.00011090886727235485,
      "loss": 2.1409,
      "step": 165215
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.209046483039856,
      "learning_rate": 0.00011090569158771339,
      "loss": 2.0989,
      "step": 165216
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.2418794631958008,
      "learning_rate": 0.00011090251593822806,
      "loss": 1.8634,
      "step": 165217
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.1659808158874512,
      "learning_rate": 0.00011089934032389962,
      "loss": 1.8383,
      "step": 165218
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.3282438516616821,
      "learning_rate": 0.00011089616474472856,
      "loss": 2.0398,
      "step": 165219
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.116574764251709,
      "learning_rate": 0.00011089298920071535,
      "loss": 1.9798,
      "step": 165220
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.2510584592819214,
      "learning_rate": 0.00011088981369186081,
      "loss": 2.0511,
      "step": 165221
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.3079808950424194,
      "learning_rate": 0.00011088663821816536,
      "loss": 2.1758,
      "step": 165222
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.287519097328186,
      "learning_rate": 0.00011088346277962974,
      "loss": 2.0183,
      "step": 165223
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.3568079471588135,
      "learning_rate": 0.00011088028737625443,
      "loss": 1.9874,
      "step": 165224
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.4953652620315552,
      "learning_rate": 0.00011087711200804005,
      "loss": 2.1174,
      "step": 165225
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.4420040845870972,
      "learning_rate": 0.00011087393667498709,
      "loss": 2.1464,
      "step": 165226
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.2431142330169678,
      "learning_rate": 0.00011087076137709634,
      "loss": 2.2097,
      "step": 165227
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.2477797269821167,
      "learning_rate": 0.00011086758611436819,
      "loss": 1.8536,
      "step": 165228
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.3936694860458374,
      "learning_rate": 0.00011086441088680341,
      "loss": 2.035,
      "step": 165229
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.4400172233581543,
      "learning_rate": 0.0001108612356944025,
      "loss": 1.9371,
      "step": 165230
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.4657717943191528,
      "learning_rate": 0.00011085806053716607,
      "loss": 1.9033,
      "step": 165231
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.2050877809524536,
      "learning_rate": 0.0001108548854150946,
      "loss": 2.2673,
      "step": 165232
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.2422854900360107,
      "learning_rate": 0.00011085171032818889,
      "loss": 2.0217,
      "step": 165233
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.6716161966323853,
      "learning_rate": 0.0001108485352764493,
      "loss": 2.3013,
      "step": 165234
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.420913815498352,
      "learning_rate": 0.00011084536025987663,
      "loss": 2.0077,
      "step": 165235
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.170846700668335,
      "learning_rate": 0.00011084218527847138,
      "loss": 2.0765,
      "step": 165236
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.2068113088607788,
      "learning_rate": 0.00011083901033223417,
      "loss": 2.0184,
      "step": 165237
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.2168140411376953,
      "learning_rate": 0.00011083583542116542,
      "loss": 1.9637,
      "step": 165238
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.413162350654602,
      "learning_rate": 0.00011083266054526597,
      "loss": 2.0397,
      "step": 165239
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.281254768371582,
      "learning_rate": 0.0001108294857045362,
      "loss": 2.2031,
      "step": 165240
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.187675952911377,
      "learning_rate": 0.0001108263108989769,
      "loss": 1.9995,
      "step": 165241
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.1266661882400513,
      "learning_rate": 0.00011082313612858856,
      "loss": 2.0475,
      "step": 165242
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.1937010288238525,
      "learning_rate": 0.00011081996139337178,
      "loss": 2.1069,
      "step": 165243
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.3347749710083008,
      "learning_rate": 0.00011081678669332704,
      "loss": 2.124,
      "step": 165244
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.2959175109863281,
      "learning_rate": 0.00011081361202845511,
      "loss": 1.963,
      "step": 165245
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.255433201789856,
      "learning_rate": 0.00011081043739875642,
      "loss": 2.0346,
      "step": 165246
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.2074531316757202,
      "learning_rate": 0.00011080726280423173,
      "loss": 1.7806,
      "step": 165247
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.2133386135101318,
      "learning_rate": 0.00011080408824488154,
      "loss": 2.0903,
      "step": 165248
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.1920140981674194,
      "learning_rate": 0.00011080091372070647,
      "loss": 2.048,
      "step": 165249
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.2293980121612549,
      "learning_rate": 0.00011079773923170694,
      "loss": 2.1169,
      "step": 165250
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.195298671722412,
      "learning_rate": 0.0001107945647778838,
      "loss": 1.9786,
      "step": 165251
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.1675503253936768,
      "learning_rate": 0.00011079139035923742,
      "loss": 2.1267,
      "step": 165252
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.2041231393814087,
      "learning_rate": 0.0001107882159757686,
      "loss": 2.0676,
      "step": 165253
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.2616913318634033,
      "learning_rate": 0.00011078504162747781,
      "loss": 1.9639,
      "step": 165254
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.3394380807876587,
      "learning_rate": 0.00011078186731436568,
      "loss": 1.891,
      "step": 165255
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.2973390817642212,
      "learning_rate": 0.00011077869303643265,
      "loss": 2.0855,
      "step": 165256
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.2834086418151855,
      "learning_rate": 0.00011077551879367953,
      "loss": 2.3481,
      "step": 165257
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.1796374320983887,
      "learning_rate": 0.00011077234458610671,
      "loss": 1.9289,
      "step": 165258
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.4418323040008545,
      "learning_rate": 0.00011076917041371498,
      "loss": 2.0452,
      "step": 165259
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.1780906915664673,
      "learning_rate": 0.00011076599627650484,
      "loss": 2.046,
      "step": 165260
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.2220125198364258,
      "learning_rate": 0.00011076282217447689,
      "loss": 2.0592,
      "step": 165261
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.281110405921936,
      "learning_rate": 0.00011075964810763161,
      "loss": 2.1132,
      "step": 165262
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.1671534776687622,
      "learning_rate": 0.00011075647407596973,
      "loss": 1.8214,
      "step": 165263
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.2212172746658325,
      "learning_rate": 0.00011075330007949176,
      "loss": 2.0834,
      "step": 165264
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.325648307800293,
      "learning_rate": 0.00011075012611819837,
      "loss": 2.2804,
      "step": 165265
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.2398375272750854,
      "learning_rate": 0.00011074695219209014,
      "loss": 1.8526,
      "step": 165266
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.1414246559143066,
      "learning_rate": 0.00011074377830116762,
      "loss": 2.0444,
      "step": 165267
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.2612719535827637,
      "learning_rate": 0.00011074060444543126,
      "loss": 2.0682,
      "step": 165268
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.346699833869934,
      "learning_rate": 0.00011073743062488195,
      "loss": 1.9955,
      "step": 165269
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.248264193534851,
      "learning_rate": 0.00011073425683952,
      "loss": 2.0016,
      "step": 165270
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.246001958847046,
      "learning_rate": 0.00011073108308934623,
      "loss": 2.1212,
      "step": 165271
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.1838877201080322,
      "learning_rate": 0.00011072790937436112,
      "loss": 2.2549,
      "step": 165272
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.2297636270523071,
      "learning_rate": 0.00011072473569456525,
      "loss": 2.0225,
      "step": 165273
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.1767849922180176,
      "learning_rate": 0.00011072156204995913,
      "loss": 2.0413,
      "step": 165274
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.3481872081756592,
      "learning_rate": 0.00011071838844054357,
      "loss": 2.1248,
      "step": 165275
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.2942620515823364,
      "learning_rate": 0.00011071521486631892,
      "loss": 2.0889,
      "step": 165276
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.1154048442840576,
      "learning_rate": 0.00011071204132728599,
      "loss": 2.0788,
      "step": 165277
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.2552582025527954,
      "learning_rate": 0.00011070886782344525,
      "loss": 2.1951,
      "step": 165278
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.4993958473205566,
      "learning_rate": 0.00011070569435479731,
      "loss": 2.0206,
      "step": 165279
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.4691675901412964,
      "learning_rate": 0.00011070252092134267,
      "loss": 2.2827,
      "step": 165280
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.2362749576568604,
      "learning_rate": 0.0001106993475230821,
      "loss": 2.057,
      "step": 165281
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.211223840713501,
      "learning_rate": 0.00011069617416001597,
      "loss": 1.9661,
      "step": 165282
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.1976094245910645,
      "learning_rate": 0.0001106930008321451,
      "loss": 1.997,
      "step": 165283
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.2736461162567139,
      "learning_rate": 0.0001106898275394699,
      "loss": 2.1084,
      "step": 165284
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.2928436994552612,
      "learning_rate": 0.00011068665428199111,
      "loss": 2.2619,
      "step": 165285
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.1879799365997314,
      "learning_rate": 0.00011068348105970925,
      "loss": 2.0227,
      "step": 165286
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.1440397500991821,
      "learning_rate": 0.00011068030787262491,
      "loss": 2.1494,
      "step": 165287
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.3258146047592163,
      "learning_rate": 0.00011067713472073856,
      "loss": 2.0845,
      "step": 165288
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.232254147529602,
      "learning_rate": 0.00011067396160405103,
      "loss": 1.9933,
      "step": 165289
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.3430036306381226,
      "learning_rate": 0.00011067078852256267,
      "loss": 2.2518,
      "step": 165290
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.1846024990081787,
      "learning_rate": 0.00011066761547627429,
      "loss": 1.9431,
      "step": 165291
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.2245113849639893,
      "learning_rate": 0.00011066444246518638,
      "loss": 1.9643,
      "step": 165292
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.194503903388977,
      "learning_rate": 0.0001106612694892995,
      "loss": 2.2613,
      "step": 165293
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.2142034769058228,
      "learning_rate": 0.00011065809654861418,
      "loss": 2.0691,
      "step": 165294
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.3392678499221802,
      "learning_rate": 0.0001106549236431312,
      "loss": 1.9232,
      "step": 165295
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.4299685955047607,
      "learning_rate": 0.00011065175077285096,
      "loss": 2.1985,
      "step": 165296
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.1908897161483765,
      "learning_rate": 0.00011064857793777419,
      "loss": 1.9046,
      "step": 165297
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.1644231081008911,
      "learning_rate": 0.00011064540513790145,
      "loss": 2.2606,
      "step": 165298
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.3028795719146729,
      "learning_rate": 0.00011064223237323319,
      "loss": 2.337,
      "step": 165299
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.191904902458191,
      "learning_rate": 0.00011063905964377025,
      "loss": 2.0157,
      "step": 165300
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.190665602684021,
      "learning_rate": 0.00011063588694951303,
      "loss": 2.005,
      "step": 165301
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.2704236507415771,
      "learning_rate": 0.00011063271429046209,
      "loss": 2.0489,
      "step": 165302
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.2119327783584595,
      "learning_rate": 0.00011062954166661822,
      "loss": 1.8815,
      "step": 165303
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.289076805114746,
      "learning_rate": 0.00011062636907798188,
      "loss": 2.1725,
      "step": 165304
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.3269191980361938,
      "learning_rate": 0.00011062319652455358,
      "loss": 2.1432,
      "step": 165305
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.385324239730835,
      "learning_rate": 0.00011062002400633408,
      "loss": 2.0323,
      "step": 165306
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.2806450128555298,
      "learning_rate": 0.00011061685152332383,
      "loss": 2.1221,
      "step": 165307
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.3568897247314453,
      "learning_rate": 0.00011061367907552356,
      "loss": 2.0785,
      "step": 165308
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.2308276891708374,
      "learning_rate": 0.00011061050666293378,
      "loss": 1.9539,
      "step": 165309
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.2344961166381836,
      "learning_rate": 0.00011060733428555508,
      "loss": 2.0758,
      "step": 165310
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.2379264831542969,
      "learning_rate": 0.00011060416194338795,
      "loss": 2.1115,
      "step": 165311
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.2886685132980347,
      "learning_rate": 0.0001106009896364332,
      "loss": 2.0214,
      "step": 165312
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.2663326263427734,
      "learning_rate": 0.00011059781736469121,
      "loss": 2.095,
      "step": 165313
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.164502501487732,
      "learning_rate": 0.00011059464512816276,
      "loss": 1.9912,
      "step": 165314
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.239280343055725,
      "learning_rate": 0.00011059147292684829,
      "loss": 1.8503,
      "step": 165315
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.209322452545166,
      "learning_rate": 0.00011058830076074849,
      "loss": 1.9439,
      "step": 165316
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.1589391231536865,
      "learning_rate": 0.0001105851286298638,
      "loss": 1.9738,
      "step": 165317
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.1939924955368042,
      "learning_rate": 0.00011058195653419495,
      "loss": 1.8045,
      "step": 165318
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.2209432125091553,
      "learning_rate": 0.00011057878447374246,
      "loss": 1.9575,
      "step": 165319
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.4277393817901611,
      "learning_rate": 0.00011057561244850701,
      "loss": 2.0025,
      "step": 165320
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.3288401365280151,
      "learning_rate": 0.00011057244045848916,
      "loss": 2.0591,
      "step": 165321
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.2557179927825928,
      "learning_rate": 0.00011056926850368945,
      "loss": 1.8091,
      "step": 165322
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.3559882640838623,
      "learning_rate": 0.00011056609658410839,
      "loss": 1.9384,
      "step": 165323
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.1748571395874023,
      "learning_rate": 0.00011056292469974676,
      "loss": 1.9246,
      "step": 165324
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.381654977798462,
      "learning_rate": 0.00011055975285060496,
      "loss": 2.1758,
      "step": 165325
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.3333373069763184,
      "learning_rate": 0.00011055658103668382,
      "loss": 1.924,
      "step": 165326
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.278944969177246,
      "learning_rate": 0.00011055340925798373,
      "loss": 2.0556,
      "step": 165327
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.2652257680892944,
      "learning_rate": 0.00011055023751450537,
      "loss": 1.9132,
      "step": 165328
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.2396725416183472,
      "learning_rate": 0.00011054706580624917,
      "loss": 1.9789,
      "step": 165329
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.462856411933899,
      "learning_rate": 0.00011054389413321596,
      "loss": 2.0587,
      "step": 165330
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.242291808128357,
      "learning_rate": 0.00011054072249540614,
      "loss": 2.0671,
      "step": 165331
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.1326465606689453,
      "learning_rate": 0.00011053755089282045,
      "loss": 1.8443,
      "step": 165332
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.1498653888702393,
      "learning_rate": 0.00011053437932545941,
      "loss": 2.1087,
      "step": 165333
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.3458185195922852,
      "learning_rate": 0.0001105312077933236,
      "loss": 2.1826,
      "step": 165334
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.2924624681472778,
      "learning_rate": 0.00011052803629641353,
      "loss": 1.9686,
      "step": 165335
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.2728440761566162,
      "learning_rate": 0.00011052486483472997,
      "loss": 2.2298,
      "step": 165336
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.1990797519683838,
      "learning_rate": 0.00011052169340827331,
      "loss": 1.9381,
      "step": 165337
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.1751946210861206,
      "learning_rate": 0.00011051852201704435,
      "loss": 2.2196,
      "step": 165338
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.238914966583252,
      "learning_rate": 0.00011051535066104355,
      "loss": 2.2127,
      "step": 165339
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.464498519897461,
      "learning_rate": 0.00011051217934027156,
      "loss": 1.9659,
      "step": 165340
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.1365259885787964,
      "learning_rate": 0.00011050900805472878,
      "loss": 1.9902,
      "step": 165341
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.1655324697494507,
      "learning_rate": 0.0001105058368044161,
      "loss": 1.9041,
      "step": 165342
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.289034366607666,
      "learning_rate": 0.00011050266558933387,
      "loss": 2.0198,
      "step": 165343
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.1812591552734375,
      "learning_rate": 0.00011049949440948282,
      "loss": 2.2474,
      "step": 165344
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.2454012632369995,
      "learning_rate": 0.00011049632326486353,
      "loss": 2.0834,
      "step": 165345
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.2633599042892456,
      "learning_rate": 0.00011049315215547655,
      "loss": 2.1076,
      "step": 165346
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.1653425693511963,
      "learning_rate": 0.00011048998108132235,
      "loss": 1.8836,
      "step": 165347
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.1714417934417725,
      "learning_rate": 0.00011048681004240178,
      "loss": 1.8998,
      "step": 165348
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.3317348957061768,
      "learning_rate": 0.00011048363903871514,
      "loss": 2.0261,
      "step": 165349
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.3121832609176636,
      "learning_rate": 0.0001104804680702633,
      "loss": 1.983,
      "step": 165350
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.3311233520507812,
      "learning_rate": 0.0001104772971370466,
      "loss": 2.04,
      "step": 165351
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.2063847780227661,
      "learning_rate": 0.00011047412623906598,
      "loss": 2.1194,
      "step": 165352
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.2138665914535522,
      "learning_rate": 0.00011047095537632157,
      "loss": 2.1388,
      "step": 165353
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.4222465753555298,
      "learning_rate": 0.00011046778454881427,
      "loss": 2.2999,
      "step": 165354
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.2772750854492188,
      "learning_rate": 0.0001104646137565445,
      "loss": 2.2129,
      "step": 165355
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.2095813751220703,
      "learning_rate": 0.00011046144299951306,
      "loss": 2.2464,
      "step": 165356
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.2509711980819702,
      "learning_rate": 0.0001104582722777203,
      "loss": 2.1213,
      "step": 165357
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.277217149734497,
      "learning_rate": 0.00011045510159116713,
      "loss": 1.9571,
      "step": 165358
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.3101465702056885,
      "learning_rate": 0.00011045193093985373,
      "loss": 2.1449,
      "step": 165359
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.2760471105575562,
      "learning_rate": 0.00011044876032378097,
      "loss": 2.088,
      "step": 165360
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.0937963724136353,
      "learning_rate": 0.00011044558974294929,
      "loss": 2.1009,
      "step": 165361
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.288395881652832,
      "learning_rate": 0.00011044241919735945,
      "loss": 1.9912,
      "step": 165362
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.2078028917312622,
      "learning_rate": 0.00011043924868701182,
      "loss": 2.0114,
      "step": 165363
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.3667171001434326,
      "learning_rate": 0.00011043607821190738,
      "loss": 2.1599,
      "step": 165364
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.099259376525879,
      "learning_rate": 0.0001104329077720462,
      "loss": 2.0096,
      "step": 165365
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.2847480773925781,
      "learning_rate": 0.0001104297373674292,
      "loss": 1.9888,
      "step": 165366
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.661790132522583,
      "learning_rate": 0.00011042656699805683,
      "loss": 1.9695,
      "step": 165367
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.1940536499023438,
      "learning_rate": 0.00011042339666392981,
      "loss": 1.9364,
      "step": 165368
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.225728988647461,
      "learning_rate": 0.00011042022636504859,
      "loss": 2.0884,
      "step": 165369
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.324150562286377,
      "learning_rate": 0.00011041705610141392,
      "loss": 1.9013,
      "step": 165370
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.2063524723052979,
      "learning_rate": 0.00011041388587302628,
      "loss": 1.8302,
      "step": 165371
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.3933464288711548,
      "learning_rate": 0.00011041071567988628,
      "loss": 2.0734,
      "step": 165372
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.3117130994796753,
      "learning_rate": 0.00011040754552199441,
      "loss": 2.0941,
      "step": 165373
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.2973054647445679,
      "learning_rate": 0.00011040437539935149,
      "loss": 1.9176,
      "step": 165374
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.3908318281173706,
      "learning_rate": 0.00011040120531195783,
      "loss": 2.0487,
      "step": 165375
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.218269944190979,
      "learning_rate": 0.00011039803525981428,
      "loss": 2.0077,
      "step": 165376
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.2040379047393799,
      "learning_rate": 0.00011039486524292132,
      "loss": 2.1941,
      "step": 165377
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.2257874011993408,
      "learning_rate": 0.00011039169526127955,
      "loss": 2.0753,
      "step": 165378
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.31593656539917,
      "learning_rate": 0.0001103885253148894,
      "loss": 2.1533,
      "step": 165379
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.2586733102798462,
      "learning_rate": 0.00011038535540375172,
      "loss": 2.0657,
      "step": 165380
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.3066412210464478,
      "learning_rate": 0.00011038218552786687,
      "loss": 2.1914,
      "step": 165381
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.2001395225524902,
      "learning_rate": 0.0001103790156872357,
      "loss": 2.3074,
      "step": 165382
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.2374718189239502,
      "learning_rate": 0.00011037584588185862,
      "loss": 1.8699,
      "step": 165383
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.3551712036132812,
      "learning_rate": 0.00011037267611173613,
      "loss": 2.3307,
      "step": 165384
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.168400526046753,
      "learning_rate": 0.00011036950637686905,
      "loss": 2.179,
      "step": 165385
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.1743134260177612,
      "learning_rate": 0.00011036633667725787,
      "loss": 2.0898,
      "step": 165386
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.1873406171798706,
      "learning_rate": 0.00011036316701290305,
      "loss": 1.9463,
      "step": 165387
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.3268595933914185,
      "learning_rate": 0.00011035999738380543,
      "loss": 2.0069,
      "step": 165388
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.1495916843414307,
      "learning_rate": 0.00011035682778996544,
      "loss": 2.053,
      "step": 165389
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.1524152755737305,
      "learning_rate": 0.00011035365823138362,
      "loss": 2.1484,
      "step": 165390
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.1788841485977173,
      "learning_rate": 0.00011035048870806071,
      "loss": 2.1015,
      "step": 165391
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.2443718910217285,
      "learning_rate": 0.00011034731921999711,
      "loss": 1.9949,
      "step": 165392
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.2508103847503662,
      "learning_rate": 0.00011034414976719367,
      "loss": 1.9533,
      "step": 165393
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.227195382118225,
      "learning_rate": 0.00011034098034965079,
      "loss": 2.0692,
      "step": 165394
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.3604915142059326,
      "learning_rate": 0.00011033781096736913,
      "loss": 1.9954,
      "step": 165395
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.523634433746338,
      "learning_rate": 0.00011033464162034914,
      "loss": 2.0029,
      "step": 165396
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.2201329469680786,
      "learning_rate": 0.00011033147230859161,
      "loss": 2.0372,
      "step": 165397
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.1871956586837769,
      "learning_rate": 0.00011032830303209694,
      "loss": 1.9829,
      "step": 165398
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.401584267616272,
      "learning_rate": 0.00011032513379086592,
      "loss": 2.0679,
      "step": 165399
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.355103850364685,
      "learning_rate": 0.00011032196458489906,
      "loss": 1.9352,
      "step": 165400
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.302660346031189,
      "learning_rate": 0.00011031879541419692,
      "loss": 2.0988,
      "step": 165401
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.2205612659454346,
      "learning_rate": 0.00011031562627875999,
      "loss": 2.054,
      "step": 165402
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.2268080711364746,
      "learning_rate": 0.00011031245717858906,
      "loss": 1.6751,
      "step": 165403
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.3209362030029297,
      "learning_rate": 0.00011030928811368454,
      "loss": 2.2306,
      "step": 165404
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.2455726861953735,
      "learning_rate": 0.00011030611908404716,
      "loss": 1.9713,
      "step": 165405
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.3180984258651733,
      "learning_rate": 0.0001103029500896775,
      "loss": 2.2652,
      "step": 165406
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.159780740737915,
      "learning_rate": 0.00011029978113057608,
      "loss": 1.9308,
      "step": 165407
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.162319302558899,
      "learning_rate": 0.00011029661220674345,
      "loss": 2.0091,
      "step": 165408
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.321791648864746,
      "learning_rate": 0.00011029344331818032,
      "loss": 1.9477,
      "step": 165409
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.0833121538162231,
      "learning_rate": 0.00011029027446488711,
      "loss": 2.0102,
      "step": 165410
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.2877615690231323,
      "learning_rate": 0.00011028710564686467,
      "loss": 2.008,
      "step": 165411
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.3733209371566772,
      "learning_rate": 0.00011028393686411342,
      "loss": 2.0367,
      "step": 165412
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.3547865152359009,
      "learning_rate": 0.00011028076811663393,
      "loss": 2.0587,
      "step": 165413
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.1257684230804443,
      "learning_rate": 0.00011027759940442674,
      "loss": 2.0915,
      "step": 165414
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.426007866859436,
      "learning_rate": 0.00011027443072749266,
      "loss": 2.171,
      "step": 165415
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.2812612056732178,
      "learning_rate": 0.00011027126208583202,
      "loss": 2.1693,
      "step": 165416
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.524894118309021,
      "learning_rate": 0.00011026809347944564,
      "loss": 2.5942,
      "step": 165417
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.1831084489822388,
      "learning_rate": 0.00011026492490833388,
      "loss": 2.0321,
      "step": 165418
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.3904908895492554,
      "learning_rate": 0.00011026175637249768,
      "loss": 2.2127,
      "step": 165419
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.2756474018096924,
      "learning_rate": 0.00011025858787193719,
      "loss": 1.9128,
      "step": 165420
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.3324124813079834,
      "learning_rate": 0.00011025541940665331,
      "loss": 2.1281,
      "step": 165421
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.2704520225524902,
      "learning_rate": 0.00011025225097664645,
      "loss": 2.1768,
      "step": 165422
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.3164364099502563,
      "learning_rate": 0.00011024908258191733,
      "loss": 2.0548,
      "step": 165423
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.1984729766845703,
      "learning_rate": 0.00011024591422246643,
      "loss": 1.9951,
      "step": 165424
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.2245762348175049,
      "learning_rate": 0.00011024274589829458,
      "loss": 2.1205,
      "step": 165425
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.1851333379745483,
      "learning_rate": 0.00011023957760940201,
      "loss": 2.0563,
      "step": 165426
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.2194082736968994,
      "learning_rate": 0.00011023640935578955,
      "loss": 2.1863,
      "step": 165427
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.3035982847213745,
      "learning_rate": 0.00011023324113745763,
      "loss": 2.1861,
      "step": 165428
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.15513014793396,
      "learning_rate": 0.00011023007295440706,
      "loss": 2.041,
      "step": 165429
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.3321199417114258,
      "learning_rate": 0.00011022690480663817,
      "loss": 2.3816,
      "step": 165430
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.2177534103393555,
      "learning_rate": 0.00011022373669415191,
      "loss": 2.0251,
      "step": 165431
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.202360987663269,
      "learning_rate": 0.0001102205686169484,
      "loss": 2.105,
      "step": 165432
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.170820951461792,
      "learning_rate": 0.0001102174005750286,
      "loss": 2.1516,
      "step": 165433
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.213918924331665,
      "learning_rate": 0.00011021423256839285,
      "loss": 2.1451,
      "step": 165434
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.2073975801467896,
      "learning_rate": 0.00011021106459704197,
      "loss": 2.1327,
      "step": 165435
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.0864393711090088,
      "learning_rate": 0.00011020789666097633,
      "loss": 1.7658,
      "step": 165436
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.2143516540527344,
      "learning_rate": 0.00011020472876019682,
      "loss": 2.1621,
      "step": 165437
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.3030242919921875,
      "learning_rate": 0.00011020156089470363,
      "loss": 2.0515,
      "step": 165438
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.0739943981170654,
      "learning_rate": 0.00011019839306449768,
      "loss": 2.0046,
      "step": 165439
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.3695756196975708,
      "learning_rate": 0.0001101952252695793,
      "loss": 2.0116,
      "step": 165440
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.6472370624542236,
      "learning_rate": 0.00011019205750994932,
      "loss": 2.0555,
      "step": 165441
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.155221700668335,
      "learning_rate": 0.00011018888978560813,
      "loss": 2.0878,
      "step": 165442
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.2411655187606812,
      "learning_rate": 0.00011018572209655662,
      "loss": 1.8219,
      "step": 165443
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.16805899143219,
      "learning_rate": 0.00011018255444279494,
      "loss": 2.1291,
      "step": 165444
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.20851469039917,
      "learning_rate": 0.00011017938682432401,
      "loss": 1.9818,
      "step": 165445
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.2633525133132935,
      "learning_rate": 0.00011017621924114422,
      "loss": 2.0739,
      "step": 165446
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.3171684741973877,
      "learning_rate": 0.00011017305169325634,
      "loss": 2.0048,
      "step": 165447
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.197064995765686,
      "learning_rate": 0.0001101698841806608,
      "loss": 1.9257,
      "step": 165448
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.1991828680038452,
      "learning_rate": 0.00011016671670335845,
      "loss": 2.1975,
      "step": 165449
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.1561038494110107,
      "learning_rate": 0.0001101635492613495,
      "loss": 1.8881,
      "step": 165450
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.2743940353393555,
      "learning_rate": 0.00011016038185463483,
      "loss": 2.0763,
      "step": 165451
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.3079562187194824,
      "learning_rate": 0.00011015721448321478,
      "loss": 2.1574,
      "step": 165452
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.3122730255126953,
      "learning_rate": 0.00011015404714709024,
      "loss": 2.0445,
      "step": 165453
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.3985342979431152,
      "learning_rate": 0.00011015087984626153,
      "loss": 2.0235,
      "step": 165454
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.1322734355926514,
      "learning_rate": 0.00011014771258072956,
      "loss": 2.202,
      "step": 165455
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.1796082258224487,
      "learning_rate": 0.00011014454535049451,
      "loss": 1.9391,
      "step": 165456
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.2390536069869995,
      "learning_rate": 0.00011014137815555726,
      "loss": 1.9362,
      "step": 165457
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.2328659296035767,
      "learning_rate": 0.00011013821099591824,
      "loss": 1.9734,
      "step": 165458
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.3274105787277222,
      "learning_rate": 0.00011013504387157819,
      "loss": 2.1369,
      "step": 165459
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.2081016302108765,
      "learning_rate": 0.00011013187678253753,
      "loss": 2.2913,
      "step": 165460
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.2358604669570923,
      "learning_rate": 0.00011012870972879704,
      "loss": 2.2024,
      "step": 165461
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.61985182762146,
      "learning_rate": 0.00011012554271035721,
      "loss": 1.9867,
      "step": 165462
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.302582025527954,
      "learning_rate": 0.0001101223757272186,
      "loss": 1.9291,
      "step": 165463
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.2438663244247437,
      "learning_rate": 0.00011011920877938173,
      "loss": 2.0908,
      "step": 165464
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.2970185279846191,
      "learning_rate": 0.00011011604186684738,
      "loss": 2.122,
      "step": 165465
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.2605180740356445,
      "learning_rate": 0.00011011287498961596,
      "loss": 2.1308,
      "step": 165466
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.360210657119751,
      "learning_rate": 0.00011010970814768823,
      "loss": 2.0121,
      "step": 165467
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.254920244216919,
      "learning_rate": 0.0001101065413410647,
      "loss": 1.8878,
      "step": 165468
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.277504324913025,
      "learning_rate": 0.00011010337456974587,
      "loss": 1.8534,
      "step": 165469
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.2039543390274048,
      "learning_rate": 0.00011010020783373248,
      "loss": 1.8412,
      "step": 165470
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.334261178970337,
      "learning_rate": 0.00011009704113302506,
      "loss": 1.878,
      "step": 165471
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.4585355520248413,
      "learning_rate": 0.00011009387446762405,
      "loss": 2.0666,
      "step": 165472
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.138423204421997,
      "learning_rate": 0.00011009070783753031,
      "loss": 1.9523,
      "step": 165473
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.2106813192367554,
      "learning_rate": 0.00011008754124274426,
      "loss": 1.9411,
      "step": 165474
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.2767949104309082,
      "learning_rate": 0.00011008437468326643,
      "loss": 2.2146,
      "step": 165475
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.2928649187088013,
      "learning_rate": 0.00011008120815909762,
      "loss": 2.0057,
      "step": 165476
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.3180173635482788,
      "learning_rate": 0.0001100780416702383,
      "loss": 1.8438,
      "step": 165477
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.3574130535125732,
      "learning_rate": 0.00011007487521668893,
      "loss": 2.1253,
      "step": 165478
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.2579723596572876,
      "learning_rate": 0.00011007170879845034,
      "loss": 2.179,
      "step": 165479
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.243059754371643,
      "learning_rate": 0.00011006854241552301,
      "loss": 1.8715,
      "step": 165480
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.1877976655960083,
      "learning_rate": 0.0001100653760679074,
      "loss": 1.9602,
      "step": 165481
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.4464229345321655,
      "learning_rate": 0.00011006220975560436,
      "loss": 1.9826,
      "step": 165482
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.2522438764572144,
      "learning_rate": 0.00011005904347861423,
      "loss": 2.0314,
      "step": 165483
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.409157156944275,
      "learning_rate": 0.00011005587723693778,
      "loss": 1.9444,
      "step": 165484
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.1416692733764648,
      "learning_rate": 0.00011005271103057556,
      "loss": 1.9784,
      "step": 165485
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.1700550317764282,
      "learning_rate": 0.00011004954485952812,
      "loss": 1.9432,
      "step": 165486
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.1997849941253662,
      "learning_rate": 0.00011004637872379592,
      "loss": 1.7767,
      "step": 165487
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.2072330713272095,
      "learning_rate": 0.00011004321262337986,
      "loss": 1.8517,
      "step": 165488
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.3018567562103271,
      "learning_rate": 0.00011004004655828019,
      "loss": 1.9167,
      "step": 165489
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.3521394729614258,
      "learning_rate": 0.00011003688052849775,
      "loss": 2.299,
      "step": 165490
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.2475265264511108,
      "learning_rate": 0.000110033714534033,
      "loss": 1.8361,
      "step": 165491
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.313702940940857,
      "learning_rate": 0.00011003054857488672,
      "loss": 2.1551,
      "step": 165492
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.1969367265701294,
      "learning_rate": 0.00011002738265105915,
      "loss": 1.9618,
      "step": 165493
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.573532223701477,
      "learning_rate": 0.0001100242167625512,
      "loss": 2.1242,
      "step": 165494
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.2307417392730713,
      "learning_rate": 0.00011002105090936323,
      "loss": 1.9065,
      "step": 165495
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.202378749847412,
      "learning_rate": 0.000110017885091496,
      "loss": 2.4082,
      "step": 165496
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.2924655675888062,
      "learning_rate": 0.00011001471930894996,
      "loss": 2.1514,
      "step": 165497
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.1968027353286743,
      "learning_rate": 0.00011001155356172596,
      "loss": 1.9669,
      "step": 165498
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.5426064729690552,
      "learning_rate": 0.0001100083878498242,
      "loss": 1.9828,
      "step": 165499
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.3311710357666016,
      "learning_rate": 0.0001100052221732456,
      "loss": 2.2934,
      "step": 165500
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.2236946821212769,
      "learning_rate": 0.00011000205653199047,
      "loss": 2.0031,
      "step": 165501
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.238837480545044,
      "learning_rate": 0.00010999889092605968,
      "loss": 2.0366,
      "step": 165502
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.0860493183135986,
      "learning_rate": 0.00010999572535545354,
      "loss": 1.9806,
      "step": 165503
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.2984024286270142,
      "learning_rate": 0.00010999255982017303,
      "loss": 2.0296,
      "step": 165504
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.2777725458145142,
      "learning_rate": 0.00010998939432021824,
      "loss": 1.9215,
      "step": 165505
    },
    {
      "epoch": 2.15,
      "grad_norm": 1.1874780654907227,
      "learning_rate": 0.00010998622885559016,
      "loss": 2.0498,
      "step": 165506
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.2272199392318726,
      "learning_rate": 0.00010998306342628908,
      "loss": 2.0161,
      "step": 165507
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.213084101676941,
      "learning_rate": 0.00010997989803231587,
      "loss": 2.0152,
      "step": 165508
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.1337796449661255,
      "learning_rate": 0.00010997673267367084,
      "loss": 2.0717,
      "step": 165509
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.5041577816009521,
      "learning_rate": 0.00010997356735035498,
      "loss": 2.1317,
      "step": 165510
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.3599276542663574,
      "learning_rate": 0.00010997040206236836,
      "loss": 2.0239,
      "step": 165511
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.2201462984085083,
      "learning_rate": 0.00010996723680971199,
      "loss": 2.01,
      "step": 165512
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.5191031694412231,
      "learning_rate": 0.00010996407159238615,
      "loss": 2.1077,
      "step": 165513
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.1544981002807617,
      "learning_rate": 0.00010996090641039171,
      "loss": 1.9558,
      "step": 165514
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.0721434354782104,
      "learning_rate": 0.00010995774126372902,
      "loss": 2.0356,
      "step": 165515
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.2493395805358887,
      "learning_rate": 0.00010995457615239895,
      "loss": 2.0951,
      "step": 165516
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.2335189580917358,
      "learning_rate": 0.00010995141107640172,
      "loss": 1.9871,
      "step": 165517
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.2269123792648315,
      "learning_rate": 0.00010994824603573824,
      "loss": 2.0806,
      "step": 165518
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.1156173944473267,
      "learning_rate": 0.00010994508103040881,
      "loss": 1.9959,
      "step": 165519
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.4038217067718506,
      "learning_rate": 0.00010994191606041434,
      "loss": 2.0809,
      "step": 165520
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.1659412384033203,
      "learning_rate": 0.00010993875112575512,
      "loss": 1.9195,
      "step": 165521
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.2448660135269165,
      "learning_rate": 0.00010993558622643207,
      "loss": 1.9712,
      "step": 165522
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.1508944034576416,
      "learning_rate": 0.00010993242136244539,
      "loss": 2.2104,
      "step": 165523
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.2596718072891235,
      "learning_rate": 0.00010992925653379597,
      "loss": 1.8385,
      "step": 165524
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.2780170440673828,
      "learning_rate": 0.00010992609174048419,
      "loss": 1.9656,
      "step": 165525
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.235183596611023,
      "learning_rate": 0.00010992292698251083,
      "loss": 2.2024,
      "step": 165526
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.2507272958755493,
      "learning_rate": 0.00010991976225987629,
      "loss": 2.0892,
      "step": 165527
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.307665228843689,
      "learning_rate": 0.00010991659757258143,
      "loss": 2.1948,
      "step": 165528
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.3273944854736328,
      "learning_rate": 0.00010991343292062652,
      "loss": 1.971,
      "step": 165529
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.3301165103912354,
      "learning_rate": 0.00010991026830401232,
      "loss": 1.9921,
      "step": 165530
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.2211929559707642,
      "learning_rate": 0.00010990710372273932,
      "loss": 1.9087,
      "step": 165531
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.249861478805542,
      "learning_rate": 0.00010990393917680829,
      "loss": 2.1871,
      "step": 165532
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.2390764951705933,
      "learning_rate": 0.00010990077466621959,
      "loss": 2.2108,
      "step": 165533
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.1391551494598389,
      "learning_rate": 0.00010989761019097415,
      "loss": 2.0603,
      "step": 165534
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.3151514530181885,
      "learning_rate": 0.0001098944457510721,
      "loss": 2.0514,
      "step": 165535
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.1805763244628906,
      "learning_rate": 0.00010989128134651437,
      "loss": 1.9904,
      "step": 165536
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.3043993711471558,
      "learning_rate": 0.00010988811697730134,
      "loss": 2.1301,
      "step": 165537
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.156256914138794,
      "learning_rate": 0.0001098849526434338,
      "loss": 2.1042,
      "step": 165538
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.2989228963851929,
      "learning_rate": 0.00010988178834491214,
      "loss": 1.9826,
      "step": 165539
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.1924872398376465,
      "learning_rate": 0.00010987862408173725,
      "loss": 1.8658,
      "step": 165540
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.2741724252700806,
      "learning_rate": 0.00010987545985390929,
      "loss": 2.1601,
      "step": 165541
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.2596856355667114,
      "learning_rate": 0.0001098722956614292,
      "loss": 2.1273,
      "step": 165542
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.3236114978790283,
      "learning_rate": 0.00010986913150429732,
      "loss": 2.1271,
      "step": 165543
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.3271245956420898,
      "learning_rate": 0.00010986596738251447,
      "loss": 2.2357,
      "step": 165544
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.2554141283035278,
      "learning_rate": 0.00010986280329608102,
      "loss": 1.9463,
      "step": 165545
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.2629135847091675,
      "learning_rate": 0.00010985963924499778,
      "loss": 2.1713,
      "step": 165546
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.4242297410964966,
      "learning_rate": 0.0001098564752292652,
      "loss": 1.9687,
      "step": 165547
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.1962809562683105,
      "learning_rate": 0.00010985331124888386,
      "loss": 2.1163,
      "step": 165548
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.3333102464675903,
      "learning_rate": 0.00010985014730385434,
      "loss": 2.0336,
      "step": 165549
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.2892104387283325,
      "learning_rate": 0.00010984698339417735,
      "loss": 1.9736,
      "step": 165550
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.2008365392684937,
      "learning_rate": 0.00010984381951985328,
      "loss": 2.2234,
      "step": 165551
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.1443114280700684,
      "learning_rate": 0.00010984065568088294,
      "loss": 1.9342,
      "step": 165552
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.093488335609436,
      "learning_rate": 0.00010983749187726681,
      "loss": 2.0342,
      "step": 165553
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.2369728088378906,
      "learning_rate": 0.00010983432810900537,
      "loss": 2.1939,
      "step": 165554
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.1936726570129395,
      "learning_rate": 0.00010983116437609943,
      "loss": 2.0268,
      "step": 165555
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.1803690195083618,
      "learning_rate": 0.00010982800067854945,
      "loss": 1.8222,
      "step": 165556
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.3079884052276611,
      "learning_rate": 0.00010982483701635598,
      "loss": 1.8522,
      "step": 165557
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.1562634706497192,
      "learning_rate": 0.0001098216733895197,
      "loss": 1.9703,
      "step": 165558
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.2696300745010376,
      "learning_rate": 0.00010981850979804122,
      "loss": 2.0609,
      "step": 165559
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.177844524383545,
      "learning_rate": 0.00010981534624192092,
      "loss": 1.9859,
      "step": 165560
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.125422477722168,
      "learning_rate": 0.00010981218272115967,
      "loss": 1.7858,
      "step": 165561
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.296156406402588,
      "learning_rate": 0.0001098090192357579,
      "loss": 2.1306,
      "step": 165562
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.306302785873413,
      "learning_rate": 0.00010980585578571615,
      "loss": 1.9773,
      "step": 165563
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.2365517616271973,
      "learning_rate": 0.00010980269237103517,
      "loss": 1.9687,
      "step": 165564
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.3416069746017456,
      "learning_rate": 0.00010979952899171545,
      "loss": 1.9433,
      "step": 165565
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.2702399492263794,
      "learning_rate": 0.0001097963656477575,
      "loss": 2.015,
      "step": 165566
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.1881529092788696,
      "learning_rate": 0.0001097932023391621,
      "loss": 1.9299,
      "step": 165567
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.1593858003616333,
      "learning_rate": 0.00010979003906592961,
      "loss": 2.2126,
      "step": 165568
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.1350950002670288,
      "learning_rate": 0.00010978687582806087,
      "loss": 2.23,
      "step": 165569
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.1697510480880737,
      "learning_rate": 0.00010978371262555635,
      "loss": 2.1053,
      "step": 165570
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.3566910028457642,
      "learning_rate": 0.00010978054945841659,
      "loss": 2.2258,
      "step": 165571
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.2611567974090576,
      "learning_rate": 0.00010977738632664215,
      "loss": 2.0629,
      "step": 165572
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.2952998876571655,
      "learning_rate": 0.00010977422323023375,
      "loss": 2.1323,
      "step": 165573
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.2622523307800293,
      "learning_rate": 0.00010977106016919183,
      "loss": 2.0755,
      "step": 165574
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.186279535293579,
      "learning_rate": 0.00010976789714351718,
      "loss": 1.9911,
      "step": 165575
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.296927809715271,
      "learning_rate": 0.00010976473415321024,
      "loss": 2.0101,
      "step": 165576
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.1334880590438843,
      "learning_rate": 0.00010976157119827165,
      "loss": 2.0696,
      "step": 165577
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.3182159662246704,
      "learning_rate": 0.00010975840827870186,
      "loss": 1.9271,
      "step": 165578
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.1379942893981934,
      "learning_rate": 0.00010975524539450168,
      "loss": 2.1748,
      "step": 165579
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.3693166971206665,
      "learning_rate": 0.00010975208254567147,
      "loss": 2.099,
      "step": 165580
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.178301215171814,
      "learning_rate": 0.00010974891973221207,
      "loss": 2.2083,
      "step": 165581
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.1761945486068726,
      "learning_rate": 0.00010974575695412381,
      "loss": 1.9199,
      "step": 165582
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.2211158275604248,
      "learning_rate": 0.00010974259421140762,
      "loss": 2.2367,
      "step": 165583
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.5988619327545166,
      "learning_rate": 0.00010973943150406367,
      "loss": 1.8846,
      "step": 165584
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.4378509521484375,
      "learning_rate": 0.00010973626883209285,
      "loss": 2.2071,
      "step": 165585
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.2220630645751953,
      "learning_rate": 0.00010973310619549558,
      "loss": 1.8331,
      "step": 165586
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.1998140811920166,
      "learning_rate": 0.00010972994359427258,
      "loss": 2.003,
      "step": 165587
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.1790039539337158,
      "learning_rate": 0.00010972678102842427,
      "loss": 2.3032,
      "step": 165588
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.6450244188308716,
      "learning_rate": 0.00010972361849795156,
      "loss": 2.0027,
      "step": 165589
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.2977190017700195,
      "learning_rate": 0.00010972045600285464,
      "loss": 2.1744,
      "step": 165590
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.2184337377548218,
      "learning_rate": 0.00010971729354313435,
      "loss": 2.0474,
      "step": 165591
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.3942092657089233,
      "learning_rate": 0.00010971413111879112,
      "loss": 1.9268,
      "step": 165592
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.330419659614563,
      "learning_rate": 0.00010971096872982572,
      "loss": 1.9423,
      "step": 165593
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.418708086013794,
      "learning_rate": 0.00010970780637623853,
      "loss": 2.065,
      "step": 165594
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.4391509294509888,
      "learning_rate": 0.00010970464405803046,
      "loss": 1.8954,
      "step": 165595
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.2103941440582275,
      "learning_rate": 0.00010970148177520171,
      "loss": 2.2618,
      "step": 165596
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.3087310791015625,
      "learning_rate": 0.00010969831952775311,
      "loss": 2.2882,
      "step": 165597
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.323468565940857,
      "learning_rate": 0.00010969515731568511,
      "loss": 2.0069,
      "step": 165598
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.304192304611206,
      "learning_rate": 0.00010969199513899844,
      "loss": 2.0203,
      "step": 165599
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.2648814916610718,
      "learning_rate": 0.00010968883299769358,
      "loss": 2.1666,
      "step": 165600
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.420822262763977,
      "learning_rate": 0.0001096856708917713,
      "loss": 2.0293,
      "step": 165601
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.236667275428772,
      "learning_rate": 0.00010968250882123189,
      "loss": 1.9197,
      "step": 165602
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.2343586683273315,
      "learning_rate": 0.00010967934678607618,
      "loss": 2.0619,
      "step": 165603
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.3497787714004517,
      "learning_rate": 0.00010967618478630457,
      "loss": 2.0886,
      "step": 165604
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.3131301403045654,
      "learning_rate": 0.00010967302282191785,
      "loss": 1.9667,
      "step": 165605
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.1881271600723267,
      "learning_rate": 0.00010966986089291639,
      "loss": 1.8984,
      "step": 165606
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.1636370420455933,
      "learning_rate": 0.00010966669899930111,
      "loss": 1.9722,
      "step": 165607
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.313628077507019,
      "learning_rate": 0.00010966353714107218,
      "loss": 1.9879,
      "step": 165608
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.2031899690628052,
      "learning_rate": 0.00010966037531823049,
      "loss": 2.0784,
      "step": 165609
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.2909424304962158,
      "learning_rate": 0.00010965721353077645,
      "loss": 2.2531,
      "step": 165610
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.324474811553955,
      "learning_rate": 0.00010965405177871079,
      "loss": 2.1507,
      "step": 165611
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.356183648109436,
      "learning_rate": 0.00010965089006203395,
      "loss": 2.2014,
      "step": 165612
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.2032487392425537,
      "learning_rate": 0.00010964772838074674,
      "loss": 1.8305,
      "step": 165613
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.18035089969635,
      "learning_rate": 0.00010964456673484954,
      "loss": 2.0952,
      "step": 165614
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.347934365272522,
      "learning_rate": 0.00010964140512434308,
      "loss": 2.4685,
      "step": 165615
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.2308504581451416,
      "learning_rate": 0.00010963824354922774,
      "loss": 2.0309,
      "step": 165616
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.1639090776443481,
      "learning_rate": 0.00010963508200950432,
      "loss": 2.3378,
      "step": 165617
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.2040269374847412,
      "learning_rate": 0.00010963192050517326,
      "loss": 1.8295,
      "step": 165618
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.2563199996948242,
      "learning_rate": 0.00010962875903623533,
      "loss": 2.0013,
      "step": 165619
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.2380598783493042,
      "learning_rate": 0.000109625597602691,
      "loss": 2.016,
      "step": 165620
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.390472173690796,
      "learning_rate": 0.00010962243620454087,
      "loss": 2.1443,
      "step": 165621
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.2892253398895264,
      "learning_rate": 0.0001096192748417854,
      "loss": 1.9764,
      "step": 165622
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.2734904289245605,
      "learning_rate": 0.00010961611351442541,
      "loss": 2.1043,
      "step": 165623
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.376217246055603,
      "learning_rate": 0.0001096129522224613,
      "loss": 2.0568,
      "step": 165624
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.3313286304473877,
      "learning_rate": 0.00010960979096589382,
      "loss": 1.9669,
      "step": 165625
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.2755109071731567,
      "learning_rate": 0.00010960662974472346,
      "loss": 1.8634,
      "step": 165626
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.2199339866638184,
      "learning_rate": 0.00010960346855895084,
      "loss": 2.028,
      "step": 165627
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.3046081066131592,
      "learning_rate": 0.0001096003074085764,
      "loss": 1.9482,
      "step": 165628
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.1972051858901978,
      "learning_rate": 0.00010959714629360099,
      "loss": 2.0498,
      "step": 165629
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.1744941473007202,
      "learning_rate": 0.00010959398521402497,
      "loss": 1.9509,
      "step": 165630
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.28343665599823,
      "learning_rate": 0.0001095908241698491,
      "loss": 1.9747,
      "step": 165631
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.3051739931106567,
      "learning_rate": 0.00010958766316107391,
      "loss": 1.9264,
      "step": 165632
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.3407394886016846,
      "learning_rate": 0.00010958450218769993,
      "loss": 2.0127,
      "step": 165633
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.3082376718521118,
      "learning_rate": 0.00010958134124972773,
      "loss": 2.3542,
      "step": 165634
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.2663726806640625,
      "learning_rate": 0.00010957818034715804,
      "loss": 2.2446,
      "step": 165635
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.1996490955352783,
      "learning_rate": 0.00010957501947999124,
      "loss": 2.1408,
      "step": 165636
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.579945683479309,
      "learning_rate": 0.00010957185864822813,
      "loss": 1.9659,
      "step": 165637
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.2297455072402954,
      "learning_rate": 0.00010956869785186926,
      "loss": 2.0058,
      "step": 165638
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.1996124982833862,
      "learning_rate": 0.00010956553709091512,
      "loss": 2.0252,
      "step": 165639
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.3234645128250122,
      "learning_rate": 0.00010956237636536626,
      "loss": 2.0114,
      "step": 165640
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.3739814758300781,
      "learning_rate": 0.00010955921567522342,
      "loss": 2.0494,
      "step": 165641
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.2393372058868408,
      "learning_rate": 0.00010955605502048706,
      "loss": 1.9128,
      "step": 165642
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.4071478843688965,
      "learning_rate": 0.0001095528944011579,
      "loss": 2.1427,
      "step": 165643
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.3745830059051514,
      "learning_rate": 0.00010954973381723645,
      "loss": 2.1155,
      "step": 165644
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.1854802370071411,
      "learning_rate": 0.00010954657326872322,
      "loss": 2.0054,
      "step": 165645
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.1647696495056152,
      "learning_rate": 0.00010954341275561895,
      "loss": 2.055,
      "step": 165646
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.114882469177246,
      "learning_rate": 0.00010954025227792416,
      "loss": 2.0988,
      "step": 165647
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.2739542722702026,
      "learning_rate": 0.00010953709183563936,
      "loss": 1.8695,
      "step": 165648
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.6218690872192383,
      "learning_rate": 0.00010953393142876529,
      "loss": 2.1689,
      "step": 165649
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.2577980756759644,
      "learning_rate": 0.00010953077105730246,
      "loss": 2.1485,
      "step": 165650
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.3970674276351929,
      "learning_rate": 0.00010952761072125133,
      "loss": 1.9859,
      "step": 165651
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.2343817949295044,
      "learning_rate": 0.00010952445042061272,
      "loss": 2.1094,
      "step": 165652
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.3198833465576172,
      "learning_rate": 0.00010952129015538705,
      "loss": 2.0981,
      "step": 165653
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.463475227355957,
      "learning_rate": 0.00010951812992557504,
      "loss": 1.8377,
      "step": 165654
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.3419148921966553,
      "learning_rate": 0.0001095149697311772,
      "loss": 2.12,
      "step": 165655
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.2992780208587646,
      "learning_rate": 0.00010951180957219414,
      "loss": 2.0658,
      "step": 165656
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.268477201461792,
      "learning_rate": 0.00010950864944862632,
      "loss": 2.0727,
      "step": 165657
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.3648953437805176,
      "learning_rate": 0.00010950548936047454,
      "loss": 1.8398,
      "step": 165658
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.1985307931900024,
      "learning_rate": 0.00010950232930773918,
      "loss": 2.0935,
      "step": 165659
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.2042285203933716,
      "learning_rate": 0.00010949916929042103,
      "loss": 1.9323,
      "step": 165660
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.3575841188430786,
      "learning_rate": 0.00010949600930852058,
      "loss": 2.0764,
      "step": 165661
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.2815101146697998,
      "learning_rate": 0.00010949284936203844,
      "loss": 2.1767,
      "step": 165662
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.2754961252212524,
      "learning_rate": 0.00010948968945097507,
      "loss": 1.8697,
      "step": 165663
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.1880313158035278,
      "learning_rate": 0.00010948652957533122,
      "loss": 2.1297,
      "step": 165664
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.2526251077651978,
      "learning_rate": 0.00010948336973510737,
      "loss": 2.0872,
      "step": 165665
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.253569483757019,
      "learning_rate": 0.00010948020993030424,
      "loss": 1.9939,
      "step": 165666
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.1774849891662598,
      "learning_rate": 0.00010947705016092222,
      "loss": 1.8093,
      "step": 165667
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.181014060974121,
      "learning_rate": 0.00010947389042696224,
      "loss": 2.1746,
      "step": 165668
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.2389674186706543,
      "learning_rate": 0.00010947073072842446,
      "loss": 1.9374,
      "step": 165669
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.1788591146469116,
      "learning_rate": 0.00010946757106530975,
      "loss": 2.1897,
      "step": 165670
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.1334986686706543,
      "learning_rate": 0.00010946441143761852,
      "loss": 1.9997,
      "step": 165671
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.159887671470642,
      "learning_rate": 0.00010946125184535154,
      "loss": 1.9433,
      "step": 165672
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.3688359260559082,
      "learning_rate": 0.00010945809228850925,
      "loss": 2.1626,
      "step": 165673
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.2465077638626099,
      "learning_rate": 0.00010945493276709245,
      "loss": 1.9494,
      "step": 165674
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.1964106559753418,
      "learning_rate": 0.00010945177328110142,
      "loss": 1.9761,
      "step": 165675
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.4090025424957275,
      "learning_rate": 0.00010944861383053698,
      "loss": 1.9635,
      "step": 165676
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.1256804466247559,
      "learning_rate": 0.00010944545441539953,
      "loss": 1.9632,
      "step": 165677
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.1312272548675537,
      "learning_rate": 0.00010944229503568989,
      "loss": 1.9392,
      "step": 165678
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.3512725830078125,
      "learning_rate": 0.00010943913569140839,
      "loss": 2.306,
      "step": 165679
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.215684175491333,
      "learning_rate": 0.00010943597638255587,
      "loss": 2.0767,
      "step": 165680
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.1885663270950317,
      "learning_rate": 0.0001094328171091328,
      "loss": 2.1826,
      "step": 165681
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.2697429656982422,
      "learning_rate": 0.00010942965787113979,
      "loss": 2.0977,
      "step": 165682
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.2793548107147217,
      "learning_rate": 0.00010942649866857726,
      "loss": 2.0698,
      "step": 165683
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.1889771223068237,
      "learning_rate": 0.00010942333950144607,
      "loss": 1.7746,
      "step": 165684
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.3428969383239746,
      "learning_rate": 0.00010942018036974657,
      "loss": 2.1407,
      "step": 165685
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.2581554651260376,
      "learning_rate": 0.00010941702127347956,
      "loss": 2.0516,
      "step": 165686
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.1784991025924683,
      "learning_rate": 0.00010941386221264554,
      "loss": 1.979,
      "step": 165687
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.1937007904052734,
      "learning_rate": 0.00010941070318724503,
      "loss": 1.9819,
      "step": 165688
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.2469240427017212,
      "learning_rate": 0.00010940754419727862,
      "loss": 2.0707,
      "step": 165689
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.2964658737182617,
      "learning_rate": 0.00010940438524274703,
      "loss": 2.2273,
      "step": 165690
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.210120439529419,
      "learning_rate": 0.00010940122632365062,
      "loss": 2.0307,
      "step": 165691
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.1986074447631836,
      "learning_rate": 0.00010939806743999028,
      "loss": 2.1836,
      "step": 165692
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.2296786308288574,
      "learning_rate": 0.00010939490859176642,
      "loss": 1.9456,
      "step": 165693
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.215217113494873,
      "learning_rate": 0.00010939174977897965,
      "loss": 1.8874,
      "step": 165694
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.1490590572357178,
      "learning_rate": 0.00010938859100163042,
      "loss": 2.1729,
      "step": 165695
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.2799789905548096,
      "learning_rate": 0.00010938543225971955,
      "loss": 2.0988,
      "step": 165696
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.3858375549316406,
      "learning_rate": 0.00010938227355324747,
      "loss": 2.1245,
      "step": 165697
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.2067002058029175,
      "learning_rate": 0.0001093791148822149,
      "loss": 1.9181,
      "step": 165698
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.1710429191589355,
      "learning_rate": 0.00010937595624662235,
      "loss": 1.9733,
      "step": 165699
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.2778493165969849,
      "learning_rate": 0.0001093727976464704,
      "loss": 2.0962,
      "step": 165700
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.302754521369934,
      "learning_rate": 0.00010936963908175955,
      "loss": 2.1395,
      "step": 165701
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.2952090501785278,
      "learning_rate": 0.00010936648055249058,
      "loss": 1.8458,
      "step": 165702
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.0789755582809448,
      "learning_rate": 0.00010936332205866389,
      "loss": 2.0268,
      "step": 165703
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.35623037815094,
      "learning_rate": 0.00010936016360028026,
      "loss": 1.9562,
      "step": 165704
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.2263357639312744,
      "learning_rate": 0.00010935700517734017,
      "loss": 1.8967,
      "step": 165705
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.3580490350723267,
      "learning_rate": 0.0001093538467898442,
      "loss": 1.9954,
      "step": 165706
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.352290391921997,
      "learning_rate": 0.00010935068843779284,
      "loss": 2.1284,
      "step": 165707
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.2828906774520874,
      "learning_rate": 0.00010934753012118693,
      "loss": 2.1074,
      "step": 165708
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.2902069091796875,
      "learning_rate": 0.00010934437184002676,
      "loss": 2.0659,
      "step": 165709
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.190665602684021,
      "learning_rate": 0.00010934121359431321,
      "loss": 2.0081,
      "step": 165710
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.404584288597107,
      "learning_rate": 0.0001093380553840467,
      "loss": 2.2658,
      "step": 165711
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.3431984186172485,
      "learning_rate": 0.00010933489720922787,
      "loss": 2.2979,
      "step": 165712
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.5563507080078125,
      "learning_rate": 0.00010933173906985715,
      "loss": 1.9253,
      "step": 165713
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.5329177379608154,
      "learning_rate": 0.00010932858096593539,
      "loss": 1.8832,
      "step": 165714
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.2530065774917603,
      "learning_rate": 0.00010932542289746295,
      "loss": 1.9446,
      "step": 165715
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.3328776359558105,
      "learning_rate": 0.0001093222648644406,
      "loss": 2.0923,
      "step": 165716
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.2697033882141113,
      "learning_rate": 0.00010931910686686882,
      "loss": 2.0616,
      "step": 165717
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.180349349975586,
      "learning_rate": 0.00010931594890474825,
      "loss": 2.0765,
      "step": 165718
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.2279536724090576,
      "learning_rate": 0.00010931279097807933,
      "loss": 2.0267,
      "step": 165719
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.21426522731781,
      "learning_rate": 0.00010930963308686283,
      "loss": 2.0202,
      "step": 165720
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.3869144916534424,
      "learning_rate": 0.00010930647523109922,
      "loss": 2.1108,
      "step": 165721
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.253922700881958,
      "learning_rate": 0.00010930331741078921,
      "loss": 2.1277,
      "step": 165722
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.2873706817626953,
      "learning_rate": 0.00010930015962593331,
      "loss": 2.2291,
      "step": 165723
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.2593742609024048,
      "learning_rate": 0.00010929700187653211,
      "loss": 1.9557,
      "step": 165724
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.2826472520828247,
      "learning_rate": 0.00010929384416258613,
      "loss": 2.1647,
      "step": 165725
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.2842203378677368,
      "learning_rate": 0.00010929068648409611,
      "loss": 2.0434,
      "step": 165726
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.3533174991607666,
      "learning_rate": 0.00010928752884106244,
      "loss": 2.1877,
      "step": 165727
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.2166213989257812,
      "learning_rate": 0.00010928437123348592,
      "loss": 1.8283,
      "step": 165728
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.279937505722046,
      "learning_rate": 0.00010928121366136701,
      "loss": 2.0076,
      "step": 165729
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.2386380434036255,
      "learning_rate": 0.00010927805612470625,
      "loss": 2.2986,
      "step": 165730
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.2420929670333862,
      "learning_rate": 0.00010927489862350441,
      "loss": 2.2274,
      "step": 165731
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.1832753419876099,
      "learning_rate": 0.00010927174115776194,
      "loss": 2.0516,
      "step": 165732
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.2836735248565674,
      "learning_rate": 0.00010926858372747938,
      "loss": 2.1099,
      "step": 165733
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.176451325416565,
      "learning_rate": 0.0001092654263326575,
      "loss": 1.9925,
      "step": 165734
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.2840707302093506,
      "learning_rate": 0.00010926226897329673,
      "loss": 1.9345,
      "step": 165735
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.404617190361023,
      "learning_rate": 0.00010925911164939763,
      "loss": 1.9309,
      "step": 165736
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.3284224271774292,
      "learning_rate": 0.00010925595436096097,
      "loss": 1.9523,
      "step": 165737
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.1627780199050903,
      "learning_rate": 0.00010925279710798722,
      "loss": 1.951,
      "step": 165738
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.24189031124115,
      "learning_rate": 0.00010924963989047688,
      "loss": 1.8282,
      "step": 165739
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.336936593055725,
      "learning_rate": 0.00010924648270843071,
      "loss": 2.1545,
      "step": 165740
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.2387804985046387,
      "learning_rate": 0.00010924332556184925,
      "loss": 1.8648,
      "step": 165741
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.1760896444320679,
      "learning_rate": 0.00010924016845073294,
      "loss": 2.0331,
      "step": 165742
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.3445371389389038,
      "learning_rate": 0.00010923701137508261,
      "loss": 1.987,
      "step": 165743
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.3041027784347534,
      "learning_rate": 0.00010923385433489859,
      "loss": 1.946,
      "step": 165744
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.2210673093795776,
      "learning_rate": 0.00010923069733018171,
      "loss": 2.1532,
      "step": 165745
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.1869494915008545,
      "learning_rate": 0.00010922754036093243,
      "loss": 1.8671,
      "step": 165746
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.28811514377594,
      "learning_rate": 0.00010922438342715127,
      "loss": 1.8958,
      "step": 165747
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.3331763744354248,
      "learning_rate": 0.00010922122652883898,
      "loss": 2.2881,
      "step": 165748
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.2090327739715576,
      "learning_rate": 0.00010921806966599607,
      "loss": 2.1872,
      "step": 165749
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.2275727987289429,
      "learning_rate": 0.00010921491283862304,
      "loss": 2.1814,
      "step": 165750
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.378595232963562,
      "learning_rate": 0.00010921175604672065,
      "loss": 2.1039,
      "step": 165751
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.3496166467666626,
      "learning_rate": 0.0001092085992902893,
      "loss": 2.0164,
      "step": 165752
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.2313522100448608,
      "learning_rate": 0.00010920544256932973,
      "loss": 1.9864,
      "step": 165753
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.213972806930542,
      "learning_rate": 0.00010920228588384253,
      "loss": 2.0339,
      "step": 165754
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.2750343084335327,
      "learning_rate": 0.00010919912923382818,
      "loss": 2.0236,
      "step": 165755
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.4075640439987183,
      "learning_rate": 0.00010919597261928726,
      "loss": 2.1627,
      "step": 165756
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.1635509729385376,
      "learning_rate": 0.0001091928160402205,
      "loss": 2.0521,
      "step": 165757
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.1110084056854248,
      "learning_rate": 0.00010918965949662826,
      "loss": 1.7848,
      "step": 165758
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.188368320465088,
      "learning_rate": 0.0001091865029885114,
      "loss": 2.1992,
      "step": 165759
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.2850018739700317,
      "learning_rate": 0.00010918334651587033,
      "loss": 2.0034,
      "step": 165760
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.1893056631088257,
      "learning_rate": 0.00010918019007870573,
      "loss": 2.0246,
      "step": 165761
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.2150145769119263,
      "learning_rate": 0.00010917703367701802,
      "loss": 2.0758,
      "step": 165762
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.2759073972702026,
      "learning_rate": 0.00010917387731080798,
      "loss": 2.1278,
      "step": 165763
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.3766732215881348,
      "learning_rate": 0.00010917072098007603,
      "loss": 1.9862,
      "step": 165764
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.2510031461715698,
      "learning_rate": 0.00010916756468482294,
      "loss": 2.079,
      "step": 165765
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.358068585395813,
      "learning_rate": 0.00010916440842504923,
      "loss": 2.1223,
      "step": 165766
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.271298885345459,
      "learning_rate": 0.00010916125220075542,
      "loss": 2.148,
      "step": 165767
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.212301254272461,
      "learning_rate": 0.00010915809601194204,
      "loss": 2.0206,
      "step": 165768
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.2780479192733765,
      "learning_rate": 0.00010915493985860989,
      "loss": 2.0033,
      "step": 165769
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.1654316186904907,
      "learning_rate": 0.00010915178374075936,
      "loss": 2.0052,
      "step": 165770
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.2820683717727661,
      "learning_rate": 0.00010914862765839117,
      "loss": 2.0711,
      "step": 165771
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.3026282787322998,
      "learning_rate": 0.00010914547161150587,
      "loss": 2.0136,
      "step": 165772
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.2094407081604004,
      "learning_rate": 0.00010914231560010403,
      "loss": 2.1687,
      "step": 165773
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.1576993465423584,
      "learning_rate": 0.00010913915962418616,
      "loss": 1.9299,
      "step": 165774
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.1870967149734497,
      "learning_rate": 0.000109136003683753,
      "loss": 1.9457,
      "step": 165775
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.2213798761367798,
      "learning_rate": 0.00010913284777880493,
      "loss": 2.1591,
      "step": 165776
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.1880921125411987,
      "learning_rate": 0.00010912969190934281,
      "loss": 1.8938,
      "step": 165777
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.2607632875442505,
      "learning_rate": 0.00010912653607536707,
      "loss": 1.919,
      "step": 165778
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.1934901475906372,
      "learning_rate": 0.00010912338027687834,
      "loss": 2.0401,
      "step": 165779
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.361781358718872,
      "learning_rate": 0.00010912022451387703,
      "loss": 1.8801,
      "step": 165780
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.5250064134597778,
      "learning_rate": 0.00010911706878636398,
      "loss": 2.1046,
      "step": 165781
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.216602087020874,
      "learning_rate": 0.00010911391309433958,
      "loss": 2.1007,
      "step": 165782
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.2904454469680786,
      "learning_rate": 0.0001091107574378046,
      "loss": 2.159,
      "step": 165783
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.3080812692642212,
      "learning_rate": 0.00010910760181675958,
      "loss": 2.0647,
      "step": 165784
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.354724645614624,
      "learning_rate": 0.00010910444623120501,
      "loss": 2.1488,
      "step": 165785
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.3382691144943237,
      "learning_rate": 0.00010910129068114143,
      "loss": 1.9625,
      "step": 165786
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.2669129371643066,
      "learning_rate": 0.00010909813516656962,
      "loss": 2.1569,
      "step": 165787
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.2475453615188599,
      "learning_rate": 0.00010909497968749,
      "loss": 2.0199,
      "step": 165788
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.3960201740264893,
      "learning_rate": 0.0001090918242439033,
      "loss": 1.9316,
      "step": 165789
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.4706881046295166,
      "learning_rate": 0.00010908866883581006,
      "loss": 2.0554,
      "step": 165790
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.3873614072799683,
      "learning_rate": 0.00010908551346321082,
      "loss": 1.8085,
      "step": 165791
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.1494909524917603,
      "learning_rate": 0.00010908235812610608,
      "loss": 2.0988,
      "step": 165792
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.3763476610183716,
      "learning_rate": 0.00010907920282449662,
      "loss": 1.9475,
      "step": 165793
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.430835485458374,
      "learning_rate": 0.00010907604755838287,
      "loss": 2.1847,
      "step": 165794
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.2852585315704346,
      "learning_rate": 0.0001090728923277656,
      "loss": 2.0025,
      "step": 165795
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.1818095445632935,
      "learning_rate": 0.00010906973713264526,
      "loss": 2.2413,
      "step": 165796
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.3143235445022583,
      "learning_rate": 0.00010906658197302244,
      "loss": 1.9938,
      "step": 165797
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.2359493970870972,
      "learning_rate": 0.00010906342684889768,
      "loss": 2.1171,
      "step": 165798
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.3030093908309937,
      "learning_rate": 0.00010906027176027172,
      "loss": 1.8235,
      "step": 165799
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.3771342039108276,
      "learning_rate": 0.00010905711670714496,
      "loss": 2.034,
      "step": 165800
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.2340729236602783,
      "learning_rate": 0.00010905396168951821,
      "loss": 1.8011,
      "step": 165801
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.3732398748397827,
      "learning_rate": 0.0001090508067073919,
      "loss": 2.3082,
      "step": 165802
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.2900512218475342,
      "learning_rate": 0.00010904765176076667,
      "loss": 2.1611,
      "step": 165803
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.2942770719528198,
      "learning_rate": 0.00010904449684964297,
      "loss": 1.8973,
      "step": 165804
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.1949537992477417,
      "learning_rate": 0.00010904134197402161,
      "loss": 2.0686,
      "step": 165805
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.22261381149292,
      "learning_rate": 0.00010903818713390296,
      "loss": 2.1013,
      "step": 165806
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.209315299987793,
      "learning_rate": 0.00010903503232928782,
      "loss": 1.9512,
      "step": 165807
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.1843669414520264,
      "learning_rate": 0.0001090318775601767,
      "loss": 2.0581,
      "step": 165808
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.2594596147537231,
      "learning_rate": 0.00010902872282657012,
      "loss": 2.2616,
      "step": 165809
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.649478554725647,
      "learning_rate": 0.00010902556812846861,
      "loss": 2.1544,
      "step": 165810
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.282220721244812,
      "learning_rate": 0.00010902241346587299,
      "loss": 2.0449,
      "step": 165811
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.4723412990570068,
      "learning_rate": 0.00010901925883878357,
      "loss": 1.9491,
      "step": 165812
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.303246021270752,
      "learning_rate": 0.00010901610424720119,
      "loss": 1.8754,
      "step": 165813
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.1617567539215088,
      "learning_rate": 0.0001090129496911262,
      "loss": 1.9041,
      "step": 165814
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.2293576002120972,
      "learning_rate": 0.00010900979517055943,
      "loss": 1.8982,
      "step": 165815
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.2890130281448364,
      "learning_rate": 0.00010900664068550134,
      "loss": 2.1351,
      "step": 165816
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.3245757818222046,
      "learning_rate": 0.00010900348623595253,
      "loss": 2.1598,
      "step": 165817
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.3325482606887817,
      "learning_rate": 0.00010900033182191346,
      "loss": 2.1517,
      "step": 165818
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.1827162504196167,
      "learning_rate": 0.00010899717744338494,
      "loss": 1.8177,
      "step": 165819
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.1739557981491089,
      "learning_rate": 0.00010899402310036736,
      "loss": 2.0781,
      "step": 165820
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.3403300046920776,
      "learning_rate": 0.00010899086879286149,
      "loss": 2.0009,
      "step": 165821
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.1771094799041748,
      "learning_rate": 0.00010898771452086783,
      "loss": 2.0805,
      "step": 165822
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.2455195188522339,
      "learning_rate": 0.00010898456028438692,
      "loss": 1.9886,
      "step": 165823
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.2092344760894775,
      "learning_rate": 0.00010898140608341934,
      "loss": 1.8507,
      "step": 165824
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.138877034187317,
      "learning_rate": 0.00010897825191796579,
      "loss": 2.2052,
      "step": 165825
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.3579391241073608,
      "learning_rate": 0.00010897509778802669,
      "loss": 2.1267,
      "step": 165826
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.336789608001709,
      "learning_rate": 0.00010897194369360285,
      "loss": 2.2767,
      "step": 165827
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.1840591430664062,
      "learning_rate": 0.0001089687896346947,
      "loss": 2.2061,
      "step": 165828
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.241519570350647,
      "learning_rate": 0.00010896563561130276,
      "loss": 1.9624,
      "step": 165829
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.3894116878509521,
      "learning_rate": 0.00010896248162342783,
      "loss": 2.0456,
      "step": 165830
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.3522350788116455,
      "learning_rate": 0.00010895932767107038,
      "loss": 2.1885,
      "step": 165831
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.541907787322998,
      "learning_rate": 0.0001089561737542309,
      "loss": 1.9549,
      "step": 165832
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.2034671306610107,
      "learning_rate": 0.00010895301987291019,
      "loss": 1.9724,
      "step": 165833
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.1863453388214111,
      "learning_rate": 0.0001089498660271087,
      "loss": 1.9001,
      "step": 165834
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.218964695930481,
      "learning_rate": 0.00010894671221682691,
      "loss": 1.8158,
      "step": 165835
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.239907145500183,
      "learning_rate": 0.00010894355844206567,
      "loss": 2.1191,
      "step": 165836
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.421850323677063,
      "learning_rate": 0.0001089404047028253,
      "loss": 1.8113,
      "step": 165837
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.3607298135757446,
      "learning_rate": 0.00010893725099910664,
      "loss": 1.9546,
      "step": 165838
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.1715906858444214,
      "learning_rate": 0.00010893409733091015,
      "loss": 2.1534,
      "step": 165839
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.5593891143798828,
      "learning_rate": 0.0001089309436982364,
      "loss": 1.8697,
      "step": 165840
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.5153992176055908,
      "learning_rate": 0.00010892779010108593,
      "loss": 1.8203,
      "step": 165841
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.2035118341445923,
      "learning_rate": 0.00010892463653945945,
      "loss": 1.9927,
      "step": 165842
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.212689995765686,
      "learning_rate": 0.0001089214830133574,
      "loss": 2.0205,
      "step": 165843
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.086538314819336,
      "learning_rate": 0.00010891832952278057,
      "loss": 2.0036,
      "step": 165844
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.3080211877822876,
      "learning_rate": 0.00010891517606772942,
      "loss": 1.837,
      "step": 165845
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.3010410070419312,
      "learning_rate": 0.00010891202264820451,
      "loss": 2.1682,
      "step": 165846
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.1480716466903687,
      "learning_rate": 0.00010890886926420639,
      "loss": 2.0554,
      "step": 165847
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.2292336225509644,
      "learning_rate": 0.00010890571591573585,
      "loss": 2.0686,
      "step": 165848
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.3068749904632568,
      "learning_rate": 0.00010890256260279319,
      "loss": 2.0667,
      "step": 165849
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.1764676570892334,
      "learning_rate": 0.00010889940932537927,
      "loss": 1.9837,
      "step": 165850
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.3029242753982544,
      "learning_rate": 0.00010889625608349457,
      "loss": 1.9619,
      "step": 165851
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.2369067668914795,
      "learning_rate": 0.00010889310287713968,
      "loss": 2.0383,
      "step": 165852
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.4168498516082764,
      "learning_rate": 0.00010888994970631503,
      "loss": 2.2317,
      "step": 165853
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.1680814027786255,
      "learning_rate": 0.00010888679657102148,
      "loss": 1.8314,
      "step": 165854
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.2398689985275269,
      "learning_rate": 0.00010888364347125936,
      "loss": 1.9355,
      "step": 165855
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.2008377313613892,
      "learning_rate": 0.00010888049040702948,
      "loss": 1.925,
      "step": 165856
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.3574750423431396,
      "learning_rate": 0.00010887733737833234,
      "loss": 2.1125,
      "step": 165857
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.2669404745101929,
      "learning_rate": 0.00010887418438516848,
      "loss": 2.0722,
      "step": 165858
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.1208783388137817,
      "learning_rate": 0.00010887103142753844,
      "loss": 2.0862,
      "step": 165859
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.1950379610061646,
      "learning_rate": 0.00010886787850544297,
      "loss": 2.1136,
      "step": 165860
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.125224232673645,
      "learning_rate": 0.00010886472561888249,
      "loss": 2.0044,
      "step": 165861
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.2724515199661255,
      "learning_rate": 0.00010886157276785775,
      "loss": 1.977,
      "step": 165862
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.2020111083984375,
      "learning_rate": 0.00010885841995236926,
      "loss": 1.9099,
      "step": 165863
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.1825848817825317,
      "learning_rate": 0.00010885526717241759,
      "loss": 2.0583,
      "step": 165864
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.2230983972549438,
      "learning_rate": 0.00010885211442800324,
      "loss": 2.0862,
      "step": 165865
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.1612457036972046,
      "learning_rate": 0.00010884896171912698,
      "loss": 2.1417,
      "step": 165866
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.2124128341674805,
      "learning_rate": 0.00010884580904578922,
      "loss": 2.092,
      "step": 165867
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.3537325859069824,
      "learning_rate": 0.0001088426564079907,
      "loss": 2.049,
      "step": 165868
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.2866156101226807,
      "learning_rate": 0.00010883950380573199,
      "loss": 1.9998,
      "step": 165869
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.28312087059021,
      "learning_rate": 0.00010883635123901358,
      "loss": 2.1635,
      "step": 165870
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.3181766271591187,
      "learning_rate": 0.00010883319870783604,
      "loss": 2.1494,
      "step": 165871
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.2510995864868164,
      "learning_rate": 0.0001088300462122001,
      "loss": 1.8474,
      "step": 165872
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.2824478149414062,
      "learning_rate": 0.00010882689375210613,
      "loss": 2.1011,
      "step": 165873
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.2930216789245605,
      "learning_rate": 0.00010882374132755499,
      "loss": 1.9786,
      "step": 165874
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.384375810623169,
      "learning_rate": 0.00010882058893854713,
      "loss": 2.2702,
      "step": 165875
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.2845849990844727,
      "learning_rate": 0.0001088174365850831,
      "loss": 2.2211,
      "step": 165876
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.180809497833252,
      "learning_rate": 0.00010881428426716343,
      "loss": 2.1733,
      "step": 165877
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.2374619245529175,
      "learning_rate": 0.0001088111319847889,
      "loss": 2.0544,
      "step": 165878
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.2531039714813232,
      "learning_rate": 0.00010880797973795987,
      "loss": 1.9533,
      "step": 165879
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.3241347074508667,
      "learning_rate": 0.00010880482752667717,
      "loss": 1.9221,
      "step": 165880
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.1836881637573242,
      "learning_rate": 0.00010880167535094114,
      "loss": 2.0689,
      "step": 165881
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.272039771080017,
      "learning_rate": 0.0001087985232107527,
      "loss": 2.0347,
      "step": 165882
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.1548240184783936,
      "learning_rate": 0.00010879537110611204,
      "loss": 2.1269,
      "step": 165883
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.3012301921844482,
      "learning_rate": 0.00010879221903702,
      "loss": 2.0006,
      "step": 165884
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.2303247451782227,
      "learning_rate": 0.00010878906700347703,
      "loss": 2.2381,
      "step": 165885
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.178916573524475,
      "learning_rate": 0.00010878591500548385,
      "loss": 2.2893,
      "step": 165886
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.1976587772369385,
      "learning_rate": 0.0001087827630430409,
      "loss": 1.9165,
      "step": 165887
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.1726323366165161,
      "learning_rate": 0.00010877961111614901,
      "loss": 2.0653,
      "step": 165888
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.5071642398834229,
      "learning_rate": 0.00010877645922480845,
      "loss": 2.107,
      "step": 165889
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.382444143295288,
      "learning_rate": 0.00010877330736902003,
      "loss": 2.145,
      "step": 165890
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.1383260488510132,
      "learning_rate": 0.00010877015554878416,
      "loss": 2.0447,
      "step": 165891
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.1663018465042114,
      "learning_rate": 0.00010876700376410162,
      "loss": 2.2001,
      "step": 165892
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.2216174602508545,
      "learning_rate": 0.00010876385201497279,
      "loss": 1.9796,
      "step": 165893
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.3092750310897827,
      "learning_rate": 0.00010876070030139861,
      "loss": 1.8673,
      "step": 165894
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.283742070198059,
      "learning_rate": 0.0001087575486233792,
      "loss": 2.0596,
      "step": 165895
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.1362698078155518,
      "learning_rate": 0.0001087543969809155,
      "loss": 2.2244,
      "step": 165896
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.152262806892395,
      "learning_rate": 0.00010875124537400782,
      "loss": 1.9961,
      "step": 165897
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.2191038131713867,
      "learning_rate": 0.00010874809380265703,
      "loss": 2.0948,
      "step": 165898
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.3542989492416382,
      "learning_rate": 0.00010874494226686347,
      "loss": 1.9194,
      "step": 165899
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.3078721761703491,
      "learning_rate": 0.00010874179076662806,
      "loss": 1.9832,
      "step": 165900
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.2542264461517334,
      "learning_rate": 0.0001087386393019509,
      "loss": 2.3663,
      "step": 165901
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.1780056953430176,
      "learning_rate": 0.000108735487872833,
      "loss": 1.7419,
      "step": 165902
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.419335961341858,
      "learning_rate": 0.00010873233647927466,
      "loss": 2.2929,
      "step": 165903
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.3376137018203735,
      "learning_rate": 0.00010872918512127667,
      "loss": 2.1318,
      "step": 165904
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.2482346296310425,
      "learning_rate": 0.00010872603379883948,
      "loss": 1.9749,
      "step": 165905
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.131230115890503,
      "learning_rate": 0.0001087228825119638,
      "loss": 2.0489,
      "step": 165906
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.231217622756958,
      "learning_rate": 0.00010871973126065014,
      "loss": 1.9738,
      "step": 165907
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.247527837753296,
      "learning_rate": 0.0001087165800448991,
      "loss": 2.2173,
      "step": 165908
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.1662659645080566,
      "learning_rate": 0.00010871342886471113,
      "loss": 1.9146,
      "step": 165909
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.2558807134628296,
      "learning_rate": 0.0001087102777200871,
      "loss": 2.1317,
      "step": 165910
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.3333250284194946,
      "learning_rate": 0.00010870712661102731,
      "loss": 1.9894,
      "step": 165911
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.2103294134140015,
      "learning_rate": 0.00010870397553753258,
      "loss": 1.8906,
      "step": 165912
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.243752121925354,
      "learning_rate": 0.0001087008244996034,
      "loss": 1.9245,
      "step": 165913
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.3579189777374268,
      "learning_rate": 0.00010869767349724023,
      "loss": 1.8892,
      "step": 165914
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.3161934614181519,
      "learning_rate": 0.00010869452253044388,
      "loss": 2.092,
      "step": 165915
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.3265436887741089,
      "learning_rate": 0.00010869137159921485,
      "loss": 2.2468,
      "step": 165916
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.2615618705749512,
      "learning_rate": 0.00010868822070355359,
      "loss": 2.19,
      "step": 165917
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.274428129196167,
      "learning_rate": 0.0001086850698434609,
      "loss": 2.0069,
      "step": 165918
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.592183232307434,
      "learning_rate": 0.00010868191901893727,
      "loss": 1.9479,
      "step": 165919
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.1725260019302368,
      "learning_rate": 0.00010867876822998316,
      "loss": 1.9991,
      "step": 165920
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.2770445346832275,
      "learning_rate": 0.00010867561747659939,
      "loss": 2.1467,
      "step": 165921
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.2764739990234375,
      "learning_rate": 0.00010867246675878648,
      "loss": 1.9112,
      "step": 165922
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.2844487428665161,
      "learning_rate": 0.0001086693160765448,
      "loss": 1.9885,
      "step": 165923
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.3014637231826782,
      "learning_rate": 0.00010866616542987525,
      "loss": 2.008,
      "step": 165924
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.272271990776062,
      "learning_rate": 0.00010866301481877827,
      "loss": 1.9756,
      "step": 165925
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.1740853786468506,
      "learning_rate": 0.00010865986424325433,
      "loss": 2.0815,
      "step": 165926
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.3080236911773682,
      "learning_rate": 0.00010865671370330424,
      "loss": 2.0368,
      "step": 165927
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.213747262954712,
      "learning_rate": 0.00010865356319892839,
      "loss": 2.1818,
      "step": 165928
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.3900375366210938,
      "learning_rate": 0.00010865041273012753,
      "loss": 1.9775,
      "step": 165929
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.3426722288131714,
      "learning_rate": 0.0001086472622969022,
      "loss": 2.2386,
      "step": 165930
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.2847671508789062,
      "learning_rate": 0.00010864411189925294,
      "loss": 2.0135,
      "step": 165931
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.3034237623214722,
      "learning_rate": 0.00010864096153718023,
      "loss": 1.9095,
      "step": 165932
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.1632194519042969,
      "learning_rate": 0.00010863781121068493,
      "loss": 2.1048,
      "step": 165933
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.3000824451446533,
      "learning_rate": 0.00010863466091976732,
      "loss": 1.9085,
      "step": 165934
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.2090389728546143,
      "learning_rate": 0.00010863151066442826,
      "loss": 2.2096,
      "step": 165935
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.1890414953231812,
      "learning_rate": 0.00010862836044466822,
      "loss": 2.0283,
      "step": 165936
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.2136210203170776,
      "learning_rate": 0.00010862521026048777,
      "loss": 2.0749,
      "step": 165937
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.2279088497161865,
      "learning_rate": 0.0001086220601118874,
      "loss": 2.0788,
      "step": 165938
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.2000675201416016,
      "learning_rate": 0.00010861890999886794,
      "loss": 1.9741,
      "step": 165939
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.2260524034500122,
      "learning_rate": 0.0001086157599214297,
      "loss": 1.9955,
      "step": 165940
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.285685420036316,
      "learning_rate": 0.0001086126098795735,
      "loss": 2.0227,
      "step": 165941
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.2254379987716675,
      "learning_rate": 0.00010860945987329986,
      "loss": 2.077,
      "step": 165942
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.2868176698684692,
      "learning_rate": 0.0001086063099026093,
      "loss": 2.1549,
      "step": 165943
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.1406630277633667,
      "learning_rate": 0.00010860315996750236,
      "loss": 1.8611,
      "step": 165944
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.2101086378097534,
      "learning_rate": 0.0001086000100679798,
      "loss": 1.8584,
      "step": 165945
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.2191113233566284,
      "learning_rate": 0.00010859686020404202,
      "loss": 1.9484,
      "step": 165946
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.3578732013702393,
      "learning_rate": 0.00010859371037568979,
      "loss": 2.128,
      "step": 165947
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.2053135633468628,
      "learning_rate": 0.00010859056058292351,
      "loss": 1.9812,
      "step": 165948
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.2481837272644043,
      "learning_rate": 0.00010858741082574402,
      "loss": 1.8022,
      "step": 165949
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.358330488204956,
      "learning_rate": 0.00010858426110415158,
      "loss": 2.1004,
      "step": 165950
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.2523778676986694,
      "learning_rate": 0.00010858111141814704,
      "loss": 1.9958,
      "step": 165951
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.2535085678100586,
      "learning_rate": 0.00010857796176773079,
      "loss": 1.9888,
      "step": 165952
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.2277859449386597,
      "learning_rate": 0.00010857481215290358,
      "loss": 1.9929,
      "step": 165953
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.2961634397506714,
      "learning_rate": 0.00010857166257366583,
      "loss": 1.9621,
      "step": 165954
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.165695309638977,
      "learning_rate": 0.00010856851303001845,
      "loss": 1.9189,
      "step": 165955
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.144431471824646,
      "learning_rate": 0.00010856536352196158,
      "loss": 2.0101,
      "step": 165956
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.2307994365692139,
      "learning_rate": 0.00010856221404949613,
      "loss": 1.9024,
      "step": 165957
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.1824194192886353,
      "learning_rate": 0.00010855906461262246,
      "loss": 2.0465,
      "step": 165958
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.3017083406448364,
      "learning_rate": 0.0001085559152113414,
      "loss": 2.1971,
      "step": 165959
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.2854130268096924,
      "learning_rate": 0.0001085527658456533,
      "loss": 1.8285,
      "step": 165960
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.2790827751159668,
      "learning_rate": 0.00010854961651555903,
      "loss": 2.1605,
      "step": 165961
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.3384977579116821,
      "learning_rate": 0.00010854646722105884,
      "loss": 2.2442,
      "step": 165962
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.1749674081802368,
      "learning_rate": 0.00010854331796215355,
      "loss": 1.8216,
      "step": 165963
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.263847827911377,
      "learning_rate": 0.00010854016873884356,
      "loss": 2.0972,
      "step": 165964
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.1649298667907715,
      "learning_rate": 0.00010853701955112968,
      "loss": 1.9517,
      "step": 165965
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.3699077367782593,
      "learning_rate": 0.00010853387039901228,
      "loss": 1.9323,
      "step": 165966
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.2440102100372314,
      "learning_rate": 0.00010853072128249226,
      "loss": 2.0802,
      "step": 165967
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.2153558731079102,
      "learning_rate": 0.0001085275722015698,
      "loss": 2.1465,
      "step": 165968
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.2782459259033203,
      "learning_rate": 0.00010852442315624577,
      "loss": 2.2337,
      "step": 165969
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.633833408355713,
      "learning_rate": 0.00010852127414652055,
      "loss": 1.9613,
      "step": 165970
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.388782024383545,
      "learning_rate": 0.00010851812517239493,
      "loss": 1.9965,
      "step": 165971
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.324729323387146,
      "learning_rate": 0.00010851497623386934,
      "loss": 2.277,
      "step": 165972
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.220557451248169,
      "learning_rate": 0.00010851182733094462,
      "loss": 2.1173,
      "step": 165973
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.4871610403060913,
      "learning_rate": 0.00010850867846362096,
      "loss": 2.0473,
      "step": 165974
    },
    {
      "epoch": 2.16,
      "grad_norm": 2.254509925842285,
      "learning_rate": 0.00010850552963189923,
      "loss": 2.202,
      "step": 165975
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.1853135824203491,
      "learning_rate": 0.00010850238083577986,
      "loss": 2.0576,
      "step": 165976
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.3296658992767334,
      "learning_rate": 0.00010849923207526363,
      "loss": 2.0916,
      "step": 165977
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.5136933326721191,
      "learning_rate": 0.0001084960833503509,
      "loss": 2.1208,
      "step": 165978
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.2776837348937988,
      "learning_rate": 0.00010849293466104254,
      "loss": 1.9582,
      "step": 165979
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.2444411516189575,
      "learning_rate": 0.00010848978600733878,
      "loss": 2.0176,
      "step": 165980
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.2137290239334106,
      "learning_rate": 0.00010848663738924046,
      "loss": 2.161,
      "step": 165981
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.3131978511810303,
      "learning_rate": 0.00010848348880674799,
      "loss": 2.2897,
      "step": 165982
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.429840087890625,
      "learning_rate": 0.00010848034025986219,
      "loss": 2.07,
      "step": 165983
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.2718918323516846,
      "learning_rate": 0.00010847719174858336,
      "loss": 2.097,
      "step": 165984
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.2201077938079834,
      "learning_rate": 0.00010847404327291248,
      "loss": 2.0803,
      "step": 165985
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.1558665037155151,
      "learning_rate": 0.00010847089483284965,
      "loss": 2.1223,
      "step": 165986
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.2366501092910767,
      "learning_rate": 0.00010846774642839583,
      "loss": 2.1326,
      "step": 165987
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.2517906427383423,
      "learning_rate": 0.00010846459805955138,
      "loss": 1.8163,
      "step": 165988
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.3009116649627686,
      "learning_rate": 0.00010846144972631702,
      "loss": 2.247,
      "step": 165989
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.2644503116607666,
      "learning_rate": 0.00010845830142869326,
      "loss": 2.0585,
      "step": 165990
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.1122885942459106,
      "learning_rate": 0.00010845515316668077,
      "loss": 2.1879,
      "step": 165991
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.2017923593521118,
      "learning_rate": 0.0001084520049402801,
      "loss": 2.1396,
      "step": 165992
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.174200177192688,
      "learning_rate": 0.0001084488567494918,
      "loss": 1.9076,
      "step": 165993
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.396292805671692,
      "learning_rate": 0.00010844570859431642,
      "loss": 1.9756,
      "step": 165994
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.3046281337738037,
      "learning_rate": 0.00010844256047475466,
      "loss": 2.2519,
      "step": 165995
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.2314491271972656,
      "learning_rate": 0.00010843941239080696,
      "loss": 1.9017,
      "step": 165996
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.3611263036727905,
      "learning_rate": 0.00010843626434247409,
      "loss": 2.2094,
      "step": 165997
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.160340428352356,
      "learning_rate": 0.00010843311632975652,
      "loss": 2.1282,
      "step": 165998
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.1974574327468872,
      "learning_rate": 0.00010842996835265476,
      "loss": 1.8846,
      "step": 165999
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.3122146129608154,
      "learning_rate": 0.00010842682041116958,
      "loss": 1.7238,
      "step": 166000
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.4437596797943115,
      "learning_rate": 0.00010842367250530148,
      "loss": 2.0661,
      "step": 166001
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.134102463722229,
      "learning_rate": 0.0001084205246350509,
      "loss": 2.0452,
      "step": 166002
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.2362935543060303,
      "learning_rate": 0.00010841737680041868,
      "loss": 2.0372,
      "step": 166003
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.2840508222579956,
      "learning_rate": 0.0001084142290014053,
      "loss": 2.0742,
      "step": 166004
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.1776844263076782,
      "learning_rate": 0.00010841108123801123,
      "loss": 2.2406,
      "step": 166005
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.289180874824524,
      "learning_rate": 0.00010840793351023727,
      "loss": 2.1459,
      "step": 166006
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.210585117340088,
      "learning_rate": 0.00010840478581808388,
      "loss": 2.124,
      "step": 166007
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.315471887588501,
      "learning_rate": 0.00010840163816155155,
      "loss": 2.1772,
      "step": 166008
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.265416145324707,
      "learning_rate": 0.00010839849054064107,
      "loss": 2.2106,
      "step": 166009
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.1361021995544434,
      "learning_rate": 0.00010839534295535293,
      "loss": 1.9417,
      "step": 166010
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.2358269691467285,
      "learning_rate": 0.00010839219540568759,
      "loss": 1.9036,
      "step": 166011
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.3426591157913208,
      "learning_rate": 0.00010838904789164589,
      "loss": 2.1445,
      "step": 166012
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.1945620775222778,
      "learning_rate": 0.00010838590041322819,
      "loss": 1.9474,
      "step": 166013
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.1643823385238647,
      "learning_rate": 0.00010838275297043522,
      "loss": 1.9298,
      "step": 166014
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.152718186378479,
      "learning_rate": 0.00010837960556326757,
      "loss": 1.9063,
      "step": 166015
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.1900900602340698,
      "learning_rate": 0.00010837645819172575,
      "loss": 1.9594,
      "step": 166016
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.129260778427124,
      "learning_rate": 0.00010837331085581026,
      "loss": 1.9239,
      "step": 166017
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.2817468643188477,
      "learning_rate": 0.00010837016355552188,
      "loss": 2.1362,
      "step": 166018
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.1042041778564453,
      "learning_rate": 0.00010836701629086098,
      "loss": 2.1586,
      "step": 166019
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.2049330472946167,
      "learning_rate": 0.00010836386906182843,
      "loss": 2.1038,
      "step": 166020
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.1116443872451782,
      "learning_rate": 0.00010836072186842461,
      "loss": 1.8979,
      "step": 166021
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.2109335660934448,
      "learning_rate": 0.00010835757471065017,
      "loss": 2.1434,
      "step": 166022
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.3348742723464966,
      "learning_rate": 0.00010835442758850558,
      "loss": 1.9243,
      "step": 166023
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.2924376726150513,
      "learning_rate": 0.0001083512805019916,
      "loss": 2.1344,
      "step": 166024
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.1829791069030762,
      "learning_rate": 0.00010834813345110863,
      "loss": 1.912,
      "step": 166025
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.2253433465957642,
      "learning_rate": 0.00010834498643585747,
      "loss": 1.8152,
      "step": 166026
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.2128934860229492,
      "learning_rate": 0.00010834183945623847,
      "loss": 2.2752,
      "step": 166027
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.2854470014572144,
      "learning_rate": 0.00010833869251225257,
      "loss": 1.9732,
      "step": 166028
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.2878140211105347,
      "learning_rate": 0.00010833554560389995,
      "loss": 2.1669,
      "step": 166029
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.3098254203796387,
      "learning_rate": 0.00010833239873118143,
      "loss": 1.9897,
      "step": 166030
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.106806993484497,
      "learning_rate": 0.00010832925189409746,
      "loss": 1.8566,
      "step": 166031
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.4582608938217163,
      "learning_rate": 0.00010832610509264877,
      "loss": 1.9544,
      "step": 166032
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.1940420866012573,
      "learning_rate": 0.00010832295832683579,
      "loss": 1.8874,
      "step": 166033
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.1979866027832031,
      "learning_rate": 0.00010831981159665939,
      "loss": 2.0603,
      "step": 166034
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.2224615812301636,
      "learning_rate": 0.00010831666490211976,
      "loss": 2.0402,
      "step": 166035
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.2806379795074463,
      "learning_rate": 0.00010831351824321778,
      "loss": 1.9067,
      "step": 166036
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.2612234354019165,
      "learning_rate": 0.00010831037161995383,
      "loss": 2.0659,
      "step": 166037
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.208268642425537,
      "learning_rate": 0.0001083072250323287,
      "loss": 2.1292,
      "step": 166038
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.3199905157089233,
      "learning_rate": 0.00010830407848034277,
      "loss": 2.0196,
      "step": 166039
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.1670889854431152,
      "learning_rate": 0.00010830093196399693,
      "loss": 2.1828,
      "step": 166040
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.2603548765182495,
      "learning_rate": 0.00010829778548329138,
      "loss": 2.0503,
      "step": 166041
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.5348085165023804,
      "learning_rate": 0.00010829463903822699,
      "loss": 2.0841,
      "step": 166042
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.2435100078582764,
      "learning_rate": 0.00010829149262880415,
      "loss": 2.1965,
      "step": 166043
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.2153853178024292,
      "learning_rate": 0.00010828834625502363,
      "loss": 2.2219,
      "step": 166044
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.227393627166748,
      "learning_rate": 0.0001082851999168858,
      "loss": 2.0576,
      "step": 166045
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.4098172187805176,
      "learning_rate": 0.00010828205361439159,
      "loss": 2.1299,
      "step": 166046
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.2354823350906372,
      "learning_rate": 0.00010827890734754117,
      "loss": 1.8645,
      "step": 166047
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.562834620475769,
      "learning_rate": 0.00010827576111633545,
      "loss": 1.8706,
      "step": 166048
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.1499954462051392,
      "learning_rate": 0.00010827261492077474,
      "loss": 1.9553,
      "step": 166049
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.1805448532104492,
      "learning_rate": 0.00010826946876085988,
      "loss": 2.0679,
      "step": 166050
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.407051920890808,
      "learning_rate": 0.00010826632263659124,
      "loss": 2.2668,
      "step": 166051
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.19187331199646,
      "learning_rate": 0.00010826317654796971,
      "loss": 1.9018,
      "step": 166052
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.2492876052856445,
      "learning_rate": 0.00010826003049499552,
      "loss": 2.0797,
      "step": 166053
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.169878602027893,
      "learning_rate": 0.00010825688447766946,
      "loss": 1.7687,
      "step": 166054
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.283589482307434,
      "learning_rate": 0.00010825373849599197,
      "loss": 2.2191,
      "step": 166055
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.5190656185150146,
      "learning_rate": 0.00010825059254996385,
      "loss": 1.9884,
      "step": 166056
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.3568532466888428,
      "learning_rate": 0.00010824744663958546,
      "loss": 1.7673,
      "step": 166057
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.238454818725586,
      "learning_rate": 0.00010824430076485767,
      "loss": 2.0147,
      "step": 166058
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.3356236219406128,
      "learning_rate": 0.00010824115492578071,
      "loss": 2.1544,
      "step": 166059
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.290765643119812,
      "learning_rate": 0.00010823800912235542,
      "loss": 2.0305,
      "step": 166060
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.4830176830291748,
      "learning_rate": 0.00010823486335458221,
      "loss": 2.2396,
      "step": 166061
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.6576687097549438,
      "learning_rate": 0.00010823171762246185,
      "loss": 2.1286,
      "step": 166062
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.200339436531067,
      "learning_rate": 0.00010822857192599474,
      "loss": 1.9972,
      "step": 166063
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.241773247718811,
      "learning_rate": 0.00010822542626518175,
      "loss": 1.9753,
      "step": 166064
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.2252018451690674,
      "learning_rate": 0.0001082222806400231,
      "loss": 2.0318,
      "step": 166065
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.2287418842315674,
      "learning_rate": 0.00010821913505051962,
      "loss": 1.8835,
      "step": 166066
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.180242657661438,
      "learning_rate": 0.00010821598949667173,
      "loss": 1.9781,
      "step": 166067
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.4005643129348755,
      "learning_rate": 0.00010821284397848022,
      "loss": 1.9852,
      "step": 166068
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.389850378036499,
      "learning_rate": 0.00010820969849594544,
      "loss": 2.0273,
      "step": 166069
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.205684781074524,
      "learning_rate": 0.0001082065530490683,
      "loss": 2.0205,
      "step": 166070
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.2547928094863892,
      "learning_rate": 0.00010820340763784902,
      "loss": 1.9875,
      "step": 166071
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.1950504779815674,
      "learning_rate": 0.00010820026226228841,
      "loss": 1.9219,
      "step": 166072
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.2686724662780762,
      "learning_rate": 0.00010819711692238691,
      "loss": 2.2518,
      "step": 166073
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.2083133459091187,
      "learning_rate": 0.00010819397161814529,
      "loss": 2.0896,
      "step": 166074
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.163678765296936,
      "learning_rate": 0.00010819082634956391,
      "loss": 1.9612,
      "step": 166075
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.2371301651000977,
      "learning_rate": 0.0001081876811166436,
      "loss": 2.0572,
      "step": 166076
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.280368685722351,
      "learning_rate": 0.00010818453591938479,
      "loss": 2.0829,
      "step": 166077
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.3075464963912964,
      "learning_rate": 0.00010818139075778813,
      "loss": 2.1672,
      "step": 166078
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.2647820711135864,
      "learning_rate": 0.00010817824563185404,
      "loss": 2.0826,
      "step": 166079
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.310616374015808,
      "learning_rate": 0.00010817510054158336,
      "loss": 2.2168,
      "step": 166080
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.281380295753479,
      "learning_rate": 0.00010817195548697641,
      "loss": 1.9047,
      "step": 166081
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.2858632802963257,
      "learning_rate": 0.00010816881046803407,
      "loss": 1.9439,
      "step": 166082
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.2031517028808594,
      "learning_rate": 0.00010816566548475673,
      "loss": 2.0807,
      "step": 166083
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.2639232873916626,
      "learning_rate": 0.00010816252053714503,
      "loss": 2.0338,
      "step": 166084
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.2720938920974731,
      "learning_rate": 0.00010815937562519943,
      "loss": 1.9933,
      "step": 166085
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.2460891008377075,
      "learning_rate": 0.00010815623074892074,
      "loss": 1.9911,
      "step": 166086
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.5394256114959717,
      "learning_rate": 0.0001081530859083093,
      "loss": 1.9848,
      "step": 166087
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.2454988956451416,
      "learning_rate": 0.00010814994110336593,
      "loss": 2.0968,
      "step": 166088
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.3129572868347168,
      "learning_rate": 0.00010814679633409115,
      "loss": 2.0761,
      "step": 166089
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.262751579284668,
      "learning_rate": 0.00010814365160048536,
      "loss": 1.976,
      "step": 166090
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.1859004497528076,
      "learning_rate": 0.00010814050690254937,
      "loss": 2.039,
      "step": 166091
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.541990041732788,
      "learning_rate": 0.00010813736224028372,
      "loss": 2.1468,
      "step": 166092
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.362030267715454,
      "learning_rate": 0.00010813421761368883,
      "loss": 1.963,
      "step": 166093
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.2016726732254028,
      "learning_rate": 0.00010813107302276554,
      "loss": 2.0811,
      "step": 166094
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.0800230503082275,
      "learning_rate": 0.00010812792846751429,
      "loss": 1.8893,
      "step": 166095
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.1653324365615845,
      "learning_rate": 0.00010812478394793556,
      "loss": 1.8899,
      "step": 166096
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.1863054037094116,
      "learning_rate": 0.00010812163946403018,
      "loss": 1.9482,
      "step": 166097
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.263875961303711,
      "learning_rate": 0.00010811849501579852,
      "loss": 2.2962,
      "step": 166098
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.1945292949676514,
      "learning_rate": 0.00010811535060324131,
      "loss": 2.0337,
      "step": 166099
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.5540797710418701,
      "learning_rate": 0.00010811220622635911,
      "loss": 1.9683,
      "step": 166100
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.2672326564788818,
      "learning_rate": 0.00010810906188515247,
      "loss": 1.9689,
      "step": 166101
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.3051509857177734,
      "learning_rate": 0.0001081059175796219,
      "loss": 2.1445,
      "step": 166102
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.9182947874069214,
      "learning_rate": 0.00010810277330976811,
      "loss": 2.0464,
      "step": 166103
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.2349705696105957,
      "learning_rate": 0.00010809962907559155,
      "loss": 2.0448,
      "step": 166104
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.2748427391052246,
      "learning_rate": 0.00010809648487709303,
      "loss": 2.0785,
      "step": 166105
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.2170634269714355,
      "learning_rate": 0.00010809334071427295,
      "loss": 2.0165,
      "step": 166106
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.316232442855835,
      "learning_rate": 0.00010809019658713198,
      "loss": 1.8536,
      "step": 166107
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.262129545211792,
      "learning_rate": 0.00010808705249567054,
      "loss": 2.0133,
      "step": 166108
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.4764549732208252,
      "learning_rate": 0.00010808390843988945,
      "loss": 2.0512,
      "step": 166109
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.1554855108261108,
      "learning_rate": 0.00010808076441978907,
      "loss": 1.9427,
      "step": 166110
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.2619768381118774,
      "learning_rate": 0.00010807762043537021,
      "loss": 2.0848,
      "step": 166111
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.4096301794052124,
      "learning_rate": 0.00010807447648663325,
      "loss": 2.0666,
      "step": 166112
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.2291141748428345,
      "learning_rate": 0.00010807133257357906,
      "loss": 2.1351,
      "step": 166113
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.263658881187439,
      "learning_rate": 0.00010806818869620782,
      "loss": 1.8682,
      "step": 166114
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.2509742975234985,
      "learning_rate": 0.00010806504485452044,
      "loss": 2.0873,
      "step": 166115
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.2360191345214844,
      "learning_rate": 0.00010806190104851729,
      "loss": 1.8549,
      "step": 166116
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.2989929914474487,
      "learning_rate": 0.00010805875727819919,
      "loss": 1.9792,
      "step": 166117
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.3004772663116455,
      "learning_rate": 0.00010805561354356644,
      "loss": 2.077,
      "step": 166118
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.2613873481750488,
      "learning_rate": 0.00010805246984461997,
      "loss": 2.087,
      "step": 166119
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.1557917594909668,
      "learning_rate": 0.00010804932618135999,
      "loss": 1.9636,
      "step": 166120
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.3607004880905151,
      "learning_rate": 0.00010804618255378737,
      "loss": 2.0521,
      "step": 166121
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.132642388343811,
      "learning_rate": 0.00010804303896190248,
      "loss": 2.1777,
      "step": 166122
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.211452841758728,
      "learning_rate": 0.00010803989540570612,
      "loss": 2.0738,
      "step": 166123
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.4093979597091675,
      "learning_rate": 0.00010803675188519865,
      "loss": 2.087,
      "step": 166124
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.306789517402649,
      "learning_rate": 0.000108033608400381,
      "loss": 2.1875,
      "step": 166125
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.3675493001937866,
      "learning_rate": 0.00010803046495125331,
      "loss": 2.3137,
      "step": 166126
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.189807415008545,
      "learning_rate": 0.00010802732153781647,
      "loss": 1.8839,
      "step": 166127
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.2291717529296875,
      "learning_rate": 0.00010802417816007088,
      "loss": 1.8851,
      "step": 166128
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.226845622062683,
      "learning_rate": 0.00010802103481801733,
      "loss": 2.2822,
      "step": 166129
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.2662042379379272,
      "learning_rate": 0.0001080178915116562,
      "loss": 1.9078,
      "step": 166130
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.2967095375061035,
      "learning_rate": 0.00010801474824098835,
      "loss": 2.0093,
      "step": 166131
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.2193928956985474,
      "learning_rate": 0.00010801160500601398,
      "loss": 2.0493,
      "step": 166132
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.2760859727859497,
      "learning_rate": 0.00010800846180673395,
      "loss": 2.1665,
      "step": 166133
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.2763034105300903,
      "learning_rate": 0.00010800531864314873,
      "loss": 2.0924,
      "step": 166134
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.1869601011276245,
      "learning_rate": 0.00010800217551525902,
      "loss": 2.0455,
      "step": 166135
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.290278673171997,
      "learning_rate": 0.00010799903242306522,
      "loss": 1.8622,
      "step": 166136
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.2176973819732666,
      "learning_rate": 0.00010799588936656825,
      "loss": 2.1222,
      "step": 166137
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.2320290803909302,
      "learning_rate": 0.00010799274634576826,
      "loss": 2.3022,
      "step": 166138
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.1659003496170044,
      "learning_rate": 0.00010798960336066612,
      "loss": 2.0185,
      "step": 166139
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.1834534406661987,
      "learning_rate": 0.00010798646041126225,
      "loss": 2.1737,
      "step": 166140
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.1223852634429932,
      "learning_rate": 0.00010798331749755742,
      "loss": 1.844,
      "step": 166141
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.6688278913497925,
      "learning_rate": 0.000107980174619552,
      "loss": 2.0071,
      "step": 166142
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.5298075675964355,
      "learning_rate": 0.00010797703177724693,
      "loss": 2.0496,
      "step": 166143
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.1498706340789795,
      "learning_rate": 0.00010797388897064234,
      "loss": 1.9214,
      "step": 166144
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.570957899093628,
      "learning_rate": 0.00010797074619973912,
      "loss": 1.9632,
      "step": 166145
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.154938817024231,
      "learning_rate": 0.00010796760346453765,
      "loss": 1.9334,
      "step": 166146
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.2385320663452148,
      "learning_rate": 0.00010796446076503874,
      "loss": 2.1311,
      "step": 166147
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.3256288766860962,
      "learning_rate": 0.00010796131810124278,
      "loss": 2.2935,
      "step": 166148
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.400917649269104,
      "learning_rate": 0.00010795817547315052,
      "loss": 2.2811,
      "step": 166149
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.0699504613876343,
      "learning_rate": 0.00010795503288076248,
      "loss": 1.9917,
      "step": 166150
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.279802680015564,
      "learning_rate": 0.00010795189032407919,
      "loss": 2.007,
      "step": 166151
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.2179529666900635,
      "learning_rate": 0.00010794874780310118,
      "loss": 1.9993,
      "step": 166152
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.2641444206237793,
      "learning_rate": 0.00010794560531782922,
      "loss": 2.1358,
      "step": 166153
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.1497944593429565,
      "learning_rate": 0.00010794246286826373,
      "loss": 2.0642,
      "step": 166154
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.3371676206588745,
      "learning_rate": 0.00010793932045440543,
      "loss": 1.951,
      "step": 166155
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.268101453781128,
      "learning_rate": 0.00010793617807625484,
      "loss": 1.8757,
      "step": 166156
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.2108925580978394,
      "learning_rate": 0.00010793303573381253,
      "loss": 2.2145,
      "step": 166157
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.3200430870056152,
      "learning_rate": 0.00010792989342707901,
      "loss": 2.0248,
      "step": 166158
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.2243560552597046,
      "learning_rate": 0.00010792675115605506,
      "loss": 1.9932,
      "step": 166159
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.176790475845337,
      "learning_rate": 0.00010792360892074101,
      "loss": 2.0194,
      "step": 166160
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.175911545753479,
      "learning_rate": 0.00010792046672113773,
      "loss": 1.9408,
      "step": 166161
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.4905154705047607,
      "learning_rate": 0.00010791732455724565,
      "loss": 2.131,
      "step": 166162
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.2335717678070068,
      "learning_rate": 0.00010791418242906534,
      "loss": 2.0466,
      "step": 166163
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.2855896949768066,
      "learning_rate": 0.00010791104033659733,
      "loss": 2.1479,
      "step": 166164
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.2555173635482788,
      "learning_rate": 0.00010790789827984239,
      "loss": 1.9603,
      "step": 166165
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.2931677103042603,
      "learning_rate": 0.0001079047562588009,
      "loss": 1.9636,
      "step": 166166
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.1848300695419312,
      "learning_rate": 0.00010790161427347362,
      "loss": 2.023,
      "step": 166167
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.3081934452056885,
      "learning_rate": 0.00010789847232386107,
      "loss": 1.8884,
      "step": 166168
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.2311033010482788,
      "learning_rate": 0.00010789533040996382,
      "loss": 1.9276,
      "step": 166169
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.234163522720337,
      "learning_rate": 0.00010789218853178237,
      "loss": 2.1016,
      "step": 166170
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.2904839515686035,
      "learning_rate": 0.00010788904668931747,
      "loss": 2.3229,
      "step": 166171
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.2436467409133911,
      "learning_rate": 0.00010788590488256951,
      "loss": 2.0814,
      "step": 166172
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.2727346420288086,
      "learning_rate": 0.0001078827631115393,
      "loss": 2.1273,
      "step": 166173
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.2215124368667603,
      "learning_rate": 0.00010787962137622733,
      "loss": 1.9313,
      "step": 166174
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.3523463010787964,
      "learning_rate": 0.00010787647967663402,
      "loss": 1.9353,
      "step": 166175
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.2019466161727905,
      "learning_rate": 0.00010787333801276024,
      "loss": 2.1156,
      "step": 166176
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.265912652015686,
      "learning_rate": 0.00010787019638460641,
      "loss": 1.8854,
      "step": 166177
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.2516441345214844,
      "learning_rate": 0.00010786705479217305,
      "loss": 2.1658,
      "step": 166178
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.1657763719558716,
      "learning_rate": 0.00010786391323546095,
      "loss": 2.1,
      "step": 166179
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.2191118001937866,
      "learning_rate": 0.00010786077171447053,
      "loss": 1.9379,
      "step": 166180
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.3102176189422607,
      "learning_rate": 0.00010785763022920234,
      "loss": 2.0577,
      "step": 166181
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.2137969732284546,
      "learning_rate": 0.00010785448877965714,
      "loss": 2.0965,
      "step": 166182
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.2245184183120728,
      "learning_rate": 0.00010785134736583544,
      "loss": 2.0742,
      "step": 166183
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.2856495380401611,
      "learning_rate": 0.00010784820598773769,
      "loss": 2.0395,
      "step": 166184
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.2853699922561646,
      "learning_rate": 0.00010784506464536466,
      "loss": 2.0019,
      "step": 166185
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.3548418283462524,
      "learning_rate": 0.0001078419233387169,
      "loss": 2.0681,
      "step": 166186
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.2444756031036377,
      "learning_rate": 0.00010783878206779482,
      "loss": 2.2494,
      "step": 166187
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.1025530099868774,
      "learning_rate": 0.00010783564083259925,
      "loss": 2.2411,
      "step": 166188
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.2386500835418701,
      "learning_rate": 0.00010783249963313058,
      "loss": 2.0556,
      "step": 166189
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.2703317403793335,
      "learning_rate": 0.00010782935846938953,
      "loss": 2.0081,
      "step": 166190
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.2219048738479614,
      "learning_rate": 0.00010782621734137667,
      "loss": 2.3049,
      "step": 166191
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.134036660194397,
      "learning_rate": 0.00010782307624909253,
      "loss": 2.0857,
      "step": 166192
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.259293794631958,
      "learning_rate": 0.0001078199351925376,
      "loss": 2.1966,
      "step": 166193
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.1312634944915771,
      "learning_rate": 0.00010781679417171271,
      "loss": 2.0338,
      "step": 166194
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.2500896453857422,
      "learning_rate": 0.00010781365318661819,
      "loss": 2.1325,
      "step": 166195
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.1789672374725342,
      "learning_rate": 0.00010781051223725482,
      "loss": 1.9822,
      "step": 166196
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.2056715488433838,
      "learning_rate": 0.000107807371323623,
      "loss": 2.068,
      "step": 166197
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.2962069511413574,
      "learning_rate": 0.00010780423044572363,
      "loss": 1.9976,
      "step": 166198
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.2032419443130493,
      "learning_rate": 0.0001078010896035569,
      "loss": 2.1226,
      "step": 166199
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.180329442024231,
      "learning_rate": 0.00010779794879712363,
      "loss": 1.7729,
      "step": 166200
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.1563340425491333,
      "learning_rate": 0.00010779480802642427,
      "loss": 2.1988,
      "step": 166201
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.2354880571365356,
      "learning_rate": 0.00010779166729145961,
      "loss": 1.8273,
      "step": 166202
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.1305416822433472,
      "learning_rate": 0.00010778852659222999,
      "loss": 1.9187,
      "step": 166203
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.2414742708206177,
      "learning_rate": 0.00010778538592873631,
      "loss": 2.2053,
      "step": 166204
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.2998322248458862,
      "learning_rate": 0.00010778224530097875,
      "loss": 2.0531,
      "step": 166205
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.1892739534378052,
      "learning_rate": 0.00010777910470895818,
      "loss": 2.2217,
      "step": 166206
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.3988698720932007,
      "learning_rate": 0.00010777596415267505,
      "loss": 2.0898,
      "step": 166207
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.2012871503829956,
      "learning_rate": 0.00010777282363213005,
      "loss": 1.9103,
      "step": 166208
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.3462127447128296,
      "learning_rate": 0.00010776968314732365,
      "loss": 2.0003,
      "step": 166209
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.1844180822372437,
      "learning_rate": 0.00010776654269825665,
      "loss": 2.1405,
      "step": 166210
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.3531087636947632,
      "learning_rate": 0.0001077634022849293,
      "loss": 1.9643,
      "step": 166211
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.2608085870742798,
      "learning_rate": 0.00010776026190734249,
      "loss": 2.033,
      "step": 166212
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.2544432878494263,
      "learning_rate": 0.00010775712156549653,
      "loss": 2.013,
      "step": 166213
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.296581745147705,
      "learning_rate": 0.00010775398125939227,
      "loss": 2.1438,
      "step": 166214
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.3472464084625244,
      "learning_rate": 0.0001077508409890301,
      "loss": 1.9385,
      "step": 166215
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.3240386247634888,
      "learning_rate": 0.00010774770075441075,
      "loss": 1.9587,
      "step": 166216
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.2447595596313477,
      "learning_rate": 0.00010774456055553473,
      "loss": 2.0332,
      "step": 166217
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.2940465211868286,
      "learning_rate": 0.00010774142039240265,
      "loss": 2.0704,
      "step": 166218
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.2962071895599365,
      "learning_rate": 0.00010773828026501494,
      "loss": 1.8312,
      "step": 166219
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.35772705078125,
      "learning_rate": 0.00010773514017337242,
      "loss": 1.9882,
      "step": 166220
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.2000279426574707,
      "learning_rate": 0.00010773200011747546,
      "loss": 1.8908,
      "step": 166221
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.1751724481582642,
      "learning_rate": 0.00010772886009732486,
      "loss": 2.0906,
      "step": 166222
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.2447024583816528,
      "learning_rate": 0.0001077257201129211,
      "loss": 1.9962,
      "step": 166223
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.255866289138794,
      "learning_rate": 0.00010772258016426476,
      "loss": 2.0443,
      "step": 166224
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.3074792623519897,
      "learning_rate": 0.00010771944025135632,
      "loss": 2.0802,
      "step": 166225
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.2888895273208618,
      "learning_rate": 0.00010771630037419654,
      "loss": 2.1151,
      "step": 166226
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.302165150642395,
      "learning_rate": 0.00010771316053278587,
      "loss": 2.0463,
      "step": 166227
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.1633415222167969,
      "learning_rate": 0.00010771002072712504,
      "loss": 1.9992,
      "step": 166228
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.1872082948684692,
      "learning_rate": 0.00010770688095721454,
      "loss": 2.3206,
      "step": 166229
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.3271374702453613,
      "learning_rate": 0.00010770374122305493,
      "loss": 2.0044,
      "step": 166230
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.3108850717544556,
      "learning_rate": 0.00010770060152464677,
      "loss": 1.9765,
      "step": 166231
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.2882862091064453,
      "learning_rate": 0.00010769746186199079,
      "loss": 2.1049,
      "step": 166232
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.1722562313079834,
      "learning_rate": 0.00010769432223508736,
      "loss": 1.943,
      "step": 166233
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.2279839515686035,
      "learning_rate": 0.0001076911826439373,
      "loss": 1.7727,
      "step": 166234
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.2152820825576782,
      "learning_rate": 0.00010768804308854108,
      "loss": 2.1546,
      "step": 166235
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.274696946144104,
      "learning_rate": 0.00010768490356889929,
      "loss": 1.9032,
      "step": 166236
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.1960501670837402,
      "learning_rate": 0.00010768176408501238,
      "loss": 2.0967,
      "step": 166237
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.3994572162628174,
      "learning_rate": 0.0001076786246368812,
      "loss": 1.9621,
      "step": 166238
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.2764402627944946,
      "learning_rate": 0.00010767548522450606,
      "loss": 1.9163,
      "step": 166239
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.2994256019592285,
      "learning_rate": 0.0001076723458478878,
      "loss": 2.0978,
      "step": 166240
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.1470695734024048,
      "learning_rate": 0.00010766920650702687,
      "loss": 2.0881,
      "step": 166241
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.3050609827041626,
      "learning_rate": 0.00010766606720192388,
      "loss": 2.1167,
      "step": 166242
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.171714425086975,
      "learning_rate": 0.00010766292793257927,
      "loss": 2.0456,
      "step": 166243
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.378639578819275,
      "learning_rate": 0.00010765978869899387,
      "loss": 2.2383,
      "step": 166244
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.2698513269424438,
      "learning_rate": 0.00010765664950116805,
      "loss": 2.0295,
      "step": 166245
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.0857700109481812,
      "learning_rate": 0.00010765351033910258,
      "loss": 1.8353,
      "step": 166246
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.248706579208374,
      "learning_rate": 0.00010765037121279796,
      "loss": 1.9458,
      "step": 166247
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.372597336769104,
      "learning_rate": 0.00010764723212225476,
      "loss": 2.0243,
      "step": 166248
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.2164753675460815,
      "learning_rate": 0.00010764409306747347,
      "loss": 1.8538,
      "step": 166249
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.5101654529571533,
      "learning_rate": 0.0001076409540484549,
      "loss": 1.8854,
      "step": 166250
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.6881853342056274,
      "learning_rate": 0.00010763781506519937,
      "loss": 1.8853,
      "step": 166251
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.2006393671035767,
      "learning_rate": 0.00010763467611770774,
      "loss": 2.023,
      "step": 166252
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.1396663188934326,
      "learning_rate": 0.00010763153720598044,
      "loss": 1.8967,
      "step": 166253
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.2125680446624756,
      "learning_rate": 0.00010762839833001805,
      "loss": 1.8922,
      "step": 166254
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.5256482362747192,
      "learning_rate": 0.00010762525948982106,
      "loss": 1.9472,
      "step": 166255
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.2160226106643677,
      "learning_rate": 0.00010762212068539029,
      "loss": 1.8275,
      "step": 166256
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.5174669027328491,
      "learning_rate": 0.00010761898191672611,
      "loss": 2.1879,
      "step": 166257
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.3969472646713257,
      "learning_rate": 0.00010761584318382928,
      "loss": 2.2953,
      "step": 166258
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.2244503498077393,
      "learning_rate": 0.00010761270448670028,
      "loss": 2.2615,
      "step": 166259
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.325308918952942,
      "learning_rate": 0.00010760956582533963,
      "loss": 1.9199,
      "step": 166260
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.4016298055648804,
      "learning_rate": 0.00010760642719974811,
      "loss": 1.9097,
      "step": 166261
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.3290278911590576,
      "learning_rate": 0.00010760328860992618,
      "loss": 2.1466,
      "step": 166262
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.2603752613067627,
      "learning_rate": 0.00010760015005587429,
      "loss": 2.0693,
      "step": 166263
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.2665114402770996,
      "learning_rate": 0.00010759701153759329,
      "loss": 1.8711,
      "step": 166264
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.4789477586746216,
      "learning_rate": 0.00010759387305508364,
      "loss": 1.9171,
      "step": 166265
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.298275113105774,
      "learning_rate": 0.00010759073460834582,
      "loss": 2.0627,
      "step": 166266
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.352878451347351,
      "learning_rate": 0.00010758759619738061,
      "loss": 2.2276,
      "step": 166267
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.2461682558059692,
      "learning_rate": 0.00010758445782218852,
      "loss": 2.1317,
      "step": 166268
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.1244767904281616,
      "learning_rate": 0.00010758131948276998,
      "loss": 2.1433,
      "step": 166269
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.1489994525909424,
      "learning_rate": 0.00010757818117912583,
      "loss": 1.9136,
      "step": 166270
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.4360960721969604,
      "learning_rate": 0.00010757504291125651,
      "loss": 1.9818,
      "step": 166271
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.284184455871582,
      "learning_rate": 0.00010757190467916252,
      "loss": 1.8778,
      "step": 166272
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.172919511795044,
      "learning_rate": 0.00010756876648284468,
      "loss": 1.9438,
      "step": 166273
    },
    {
      "epoch": 2.16,
      "grad_norm": 1.1436872482299805,
      "learning_rate": 0.00010756562832230329,
      "loss": 1.8418,
      "step": 166274
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.2380244731903076,
      "learning_rate": 0.00010756249019753921,
      "loss": 2.064,
      "step": 166275
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.2079839706420898,
      "learning_rate": 0.0001075593521085529,
      "loss": 2.0049,
      "step": 166276
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.517320156097412,
      "learning_rate": 0.00010755621405534493,
      "loss": 2.0788,
      "step": 166277
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.262072205543518,
      "learning_rate": 0.00010755307603791578,
      "loss": 1.9882,
      "step": 166278
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.4877052307128906,
      "learning_rate": 0.00010754993805626626,
      "loss": 2.036,
      "step": 166279
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.2519872188568115,
      "learning_rate": 0.00010754680011039675,
      "loss": 1.9633,
      "step": 166280
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.339975357055664,
      "learning_rate": 0.00010754366220030801,
      "loss": 1.9161,
      "step": 166281
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.2428983449935913,
      "learning_rate": 0.00010754052432600043,
      "loss": 2.0571,
      "step": 166282
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.271963357925415,
      "learning_rate": 0.0001075373864874748,
      "loss": 2.1962,
      "step": 166283
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.1903337240219116,
      "learning_rate": 0.00010753424868473161,
      "loss": 2.0645,
      "step": 166284
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.3334773778915405,
      "learning_rate": 0.00010753111091777145,
      "loss": 1.841,
      "step": 166285
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.2726551294326782,
      "learning_rate": 0.00010752797318659476,
      "loss": 2.131,
      "step": 166286
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.1913822889328003,
      "learning_rate": 0.00010752483549120235,
      "loss": 2.1164,
      "step": 166287
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.226962685585022,
      "learning_rate": 0.00010752169783159465,
      "loss": 1.8535,
      "step": 166288
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.2412152290344238,
      "learning_rate": 0.00010751856020777237,
      "loss": 2.3193,
      "step": 166289
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.2703691720962524,
      "learning_rate": 0.00010751542261973602,
      "loss": 2.1857,
      "step": 166290
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.2646312713623047,
      "learning_rate": 0.0001075122850674862,
      "loss": 2.0782,
      "step": 166291
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.386204719543457,
      "learning_rate": 0.00010750914755102336,
      "loss": 2.187,
      "step": 166292
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.2478514909744263,
      "learning_rate": 0.00010750601007034833,
      "loss": 1.9387,
      "step": 166293
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.201648473739624,
      "learning_rate": 0.00010750287262546145,
      "loss": 1.9572,
      "step": 166294
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.2224904298782349,
      "learning_rate": 0.00010749973521636352,
      "loss": 2.1014,
      "step": 166295
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.2899260520935059,
      "learning_rate": 0.00010749659784305501,
      "loss": 1.8524,
      "step": 166296
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.143367886543274,
      "learning_rate": 0.00010749346050553653,
      "loss": 2.0563,
      "step": 166297
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.1773247718811035,
      "learning_rate": 0.00010749032320380855,
      "loss": 2.1237,
      "step": 166298
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.15234375,
      "learning_rate": 0.00010748718593787185,
      "loss": 2.1246,
      "step": 166299
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.3272281885147095,
      "learning_rate": 0.00010748404870772682,
      "loss": 2.0086,
      "step": 166300
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.1129629611968994,
      "learning_rate": 0.00010748091151337425,
      "loss": 1.8624,
      "step": 166301
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.2308775186538696,
      "learning_rate": 0.00010747777435481459,
      "loss": 2.1968,
      "step": 166302
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.295140027999878,
      "learning_rate": 0.00010747463723204847,
      "loss": 1.9206,
      "step": 166303
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.2935272455215454,
      "learning_rate": 0.00010747150014507634,
      "loss": 1.9191,
      "step": 166304
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.3530676364898682,
      "learning_rate": 0.00010746836309389898,
      "loss": 2.129,
      "step": 166305
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.3352599143981934,
      "learning_rate": 0.00010746522607851679,
      "loss": 2.4183,
      "step": 166306
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.124234676361084,
      "learning_rate": 0.00010746208909893055,
      "loss": 1.8421,
      "step": 166307
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.2089524269104004,
      "learning_rate": 0.00010745895215514077,
      "loss": 2.231,
      "step": 166308
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.4147653579711914,
      "learning_rate": 0.00010745581524714797,
      "loss": 2.1564,
      "step": 166309
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.227447509765625,
      "learning_rate": 0.00010745267837495267,
      "loss": 2.181,
      "step": 166310
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.2963109016418457,
      "learning_rate": 0.00010744954153855568,
      "loss": 1.9067,
      "step": 166311
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.3051129579544067,
      "learning_rate": 0.0001074464047379573,
      "loss": 2.0149,
      "step": 166312
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.2045793533325195,
      "learning_rate": 0.00010744326797315843,
      "loss": 2.1003,
      "step": 166313
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.2635538578033447,
      "learning_rate": 0.00010744013124415945,
      "loss": 1.9531,
      "step": 166314
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.2720345258712769,
      "learning_rate": 0.00010743699455096102,
      "loss": 2.0421,
      "step": 166315
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.346522569656372,
      "learning_rate": 0.00010743385789356356,
      "loss": 1.9803,
      "step": 166316
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.18808114528656,
      "learning_rate": 0.00010743072127196787,
      "loss": 2.0582,
      "step": 166317
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.2517694234848022,
      "learning_rate": 0.00010742758468617437,
      "loss": 1.7957,
      "step": 166318
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.205051302909851,
      "learning_rate": 0.00010742444813618381,
      "loss": 1.8686,
      "step": 166319
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.2148996591567993,
      "learning_rate": 0.00010742131162199667,
      "loss": 2.1228,
      "step": 166320
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.2186435461044312,
      "learning_rate": 0.00010741817514361355,
      "loss": 1.8322,
      "step": 166321
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.1979875564575195,
      "learning_rate": 0.00010741503870103493,
      "loss": 1.9308,
      "step": 166322
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.2789523601531982,
      "learning_rate": 0.0001074119022942616,
      "loss": 1.8291,
      "step": 166323
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.1865136623382568,
      "learning_rate": 0.00010740876592329393,
      "loss": 2.0918,
      "step": 166324
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.234877109527588,
      "learning_rate": 0.00010740562958813268,
      "loss": 2.001,
      "step": 166325
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.2796454429626465,
      "learning_rate": 0.0001074024932887784,
      "loss": 1.9283,
      "step": 166326
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.2455235719680786,
      "learning_rate": 0.0001073993570252316,
      "loss": 2.2031,
      "step": 166327
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.328136920928955,
      "learning_rate": 0.00010739622079749278,
      "loss": 1.9331,
      "step": 166328
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.1953628063201904,
      "learning_rate": 0.00010739308460556276,
      "loss": 2.0693,
      "step": 166329
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.1181484460830688,
      "learning_rate": 0.00010738994844944193,
      "loss": 2.0211,
      "step": 166330
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.0666857957839966,
      "learning_rate": 0.000107386812329131,
      "loss": 2.1039,
      "step": 166331
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.1670352220535278,
      "learning_rate": 0.00010738367624463054,
      "loss": 1.9468,
      "step": 166332
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.2515290975570679,
      "learning_rate": 0.00010738054019594103,
      "loss": 2.1369,
      "step": 166333
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.2339131832122803,
      "learning_rate": 0.00010737740418306305,
      "loss": 2.0887,
      "step": 166334
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.406250238418579,
      "learning_rate": 0.00010737426820599736,
      "loss": 1.8591,
      "step": 166335
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.1561695337295532,
      "learning_rate": 0.00010737113226474432,
      "loss": 2.05,
      "step": 166336
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.2210437059402466,
      "learning_rate": 0.0001073679963593047,
      "loss": 2.1256,
      "step": 166337
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.2230883836746216,
      "learning_rate": 0.00010736486048967903,
      "loss": 2.0926,
      "step": 166338
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.3276550769805908,
      "learning_rate": 0.00010736172465586788,
      "loss": 1.9632,
      "step": 166339
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.3326236009597778,
      "learning_rate": 0.00010735858885787167,
      "loss": 2.1088,
      "step": 166340
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.1900908946990967,
      "learning_rate": 0.00010735545309569129,
      "loss": 2.0785,
      "step": 166341
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.3274176120758057,
      "learning_rate": 0.00010735231736932702,
      "loss": 2.0729,
      "step": 166342
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.2071622610092163,
      "learning_rate": 0.00010734918167877973,
      "loss": 1.9919,
      "step": 166343
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.4455560445785522,
      "learning_rate": 0.00010734604602404987,
      "loss": 1.97,
      "step": 166344
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.2042113542556763,
      "learning_rate": 0.00010734291040513799,
      "loss": 1.8756,
      "step": 166345
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.2416173219680786,
      "learning_rate": 0.00010733977482204462,
      "loss": 2.0235,
      "step": 166346
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.3527377843856812,
      "learning_rate": 0.00010733663927477048,
      "loss": 2.012,
      "step": 166347
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.2574169635772705,
      "learning_rate": 0.00010733350376331604,
      "loss": 2.0227,
      "step": 166348
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.4228272438049316,
      "learning_rate": 0.00010733036828768201,
      "loss": 1.9963,
      "step": 166349
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.095393419265747,
      "learning_rate": 0.00010732723284786883,
      "loss": 2.0675,
      "step": 166350
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.1409085988998413,
      "learning_rate": 0.00010732409744387722,
      "loss": 2.1354,
      "step": 166351
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.2671362161636353,
      "learning_rate": 0.00010732096207570772,
      "loss": 2.1145,
      "step": 166352
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.2008986473083496,
      "learning_rate": 0.00010731782674336088,
      "loss": 2.1173,
      "step": 166353
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.1349921226501465,
      "learning_rate": 0.00010731469144683721,
      "loss": 1.8483,
      "step": 166354
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.1267800331115723,
      "learning_rate": 0.00010731155618613749,
      "loss": 1.8553,
      "step": 166355
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.4126917123794556,
      "learning_rate": 0.00010730842096126203,
      "loss": 2.0845,
      "step": 166356
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.2467576265335083,
      "learning_rate": 0.00010730528577221172,
      "loss": 2.1915,
      "step": 166357
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.3294566869735718,
      "learning_rate": 0.00010730215061898699,
      "loss": 1.9375,
      "step": 166358
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.2818704843521118,
      "learning_rate": 0.0001072990155015883,
      "loss": 2.0783,
      "step": 166359
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.1474432945251465,
      "learning_rate": 0.0001072958804200165,
      "loss": 2.2001,
      "step": 166360
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.3128385543823242,
      "learning_rate": 0.00010729274537427202,
      "loss": 1.9453,
      "step": 166361
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.2040101289749146,
      "learning_rate": 0.00010728961036435535,
      "loss": 2.2178,
      "step": 166362
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.2490274906158447,
      "learning_rate": 0.00010728647539026726,
      "loss": 2.0516,
      "step": 166363
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.172918677330017,
      "learning_rate": 0.00010728334045200827,
      "loss": 2.0228,
      "step": 166364
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.194481611251831,
      "learning_rate": 0.00010728020554957885,
      "loss": 2.0149,
      "step": 166365
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.2188432216644287,
      "learning_rate": 0.00010727707068297978,
      "loss": 1.914,
      "step": 166366
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.1644158363342285,
      "learning_rate": 0.00010727393585221151,
      "loss": 1.9651,
      "step": 166367
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.3037550449371338,
      "learning_rate": 0.00010727080105727458,
      "loss": 1.8636,
      "step": 166368
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.5359209775924683,
      "learning_rate": 0.00010726766629816976,
      "loss": 2.1412,
      "step": 166369
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.313383936882019,
      "learning_rate": 0.0001072645315748975,
      "loss": 2.0933,
      "step": 166370
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.2168850898742676,
      "learning_rate": 0.00010726139688745831,
      "loss": 2.1127,
      "step": 166371
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.3524311780929565,
      "learning_rate": 0.00010725826223585296,
      "loss": 2.0946,
      "step": 166372
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.2341275215148926,
      "learning_rate": 0.00010725512762008184,
      "loss": 2.0174,
      "step": 166373
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.2852317094802856,
      "learning_rate": 0.00010725199304014573,
      "loss": 2.0359,
      "step": 166374
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.2981210947036743,
      "learning_rate": 0.00010724885849604507,
      "loss": 2.0998,
      "step": 166375
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.3402612209320068,
      "learning_rate": 0.00010724572398778053,
      "loss": 1.977,
      "step": 166376
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.5296810865402222,
      "learning_rate": 0.00010724258951535255,
      "loss": 1.9748,
      "step": 166377
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.2803362607955933,
      "learning_rate": 0.0001072394550787619,
      "loss": 2.0994,
      "step": 166378
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.1723190546035767,
      "learning_rate": 0.000107236320678009,
      "loss": 2.0008,
      "step": 166379
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.2154513597488403,
      "learning_rate": 0.00010723318631309457,
      "loss": 2.1811,
      "step": 166380
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.248897910118103,
      "learning_rate": 0.00010723005198401913,
      "loss": 1.949,
      "step": 166381
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.3485075235366821,
      "learning_rate": 0.00010722691769078327,
      "loss": 1.7981,
      "step": 166382
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.2518904209136963,
      "learning_rate": 0.00010722378343338749,
      "loss": 1.9089,
      "step": 166383
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.2040235996246338,
      "learning_rate": 0.00010722064921183253,
      "loss": 1.9249,
      "step": 166384
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.1441971063613892,
      "learning_rate": 0.00010721751502611877,
      "loss": 2.0355,
      "step": 166385
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.5492969751358032,
      "learning_rate": 0.00010721438087624704,
      "loss": 2.0431,
      "step": 166386
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.2511438131332397,
      "learning_rate": 0.00010721124676221781,
      "loss": 1.9661,
      "step": 166387
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.258737564086914,
      "learning_rate": 0.0001072081126840316,
      "loss": 2.0632,
      "step": 166388
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.1646173000335693,
      "learning_rate": 0.000107204978641689,
      "loss": 2.1657,
      "step": 166389
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.3755038976669312,
      "learning_rate": 0.00010720184463519067,
      "loss": 2.2386,
      "step": 166390
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.3394532203674316,
      "learning_rate": 0.00010719871066453708,
      "loss": 1.8464,
      "step": 166391
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.7944632768630981,
      "learning_rate": 0.00010719557672972902,
      "loss": 1.9647,
      "step": 166392
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.332227349281311,
      "learning_rate": 0.00010719244283076691,
      "loss": 1.8875,
      "step": 166393
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.105930209159851,
      "learning_rate": 0.00010718930896765137,
      "loss": 1.9692,
      "step": 166394
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.4352048635482788,
      "learning_rate": 0.00010718617514038288,
      "loss": 1.7363,
      "step": 166395
    },
    {
      "epoch": 2.17,
      "grad_norm": 2.0344533920288086,
      "learning_rate": 0.00010718304134896221,
      "loss": 2.0916,
      "step": 166396
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.487160086631775,
      "learning_rate": 0.00010717990759338978,
      "loss": 1.8745,
      "step": 166397
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.189368724822998,
      "learning_rate": 0.00010717677387366634,
      "loss": 2.2107,
      "step": 166398
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.2783929109573364,
      "learning_rate": 0.00010717364018979234,
      "loss": 2.0489,
      "step": 166399
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.2462842464447021,
      "learning_rate": 0.00010717050654176844,
      "loss": 1.9601,
      "step": 166400
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.2168185710906982,
      "learning_rate": 0.00010716737292959507,
      "loss": 1.9652,
      "step": 166401
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.260548710823059,
      "learning_rate": 0.000107164239353273,
      "loss": 1.8577,
      "step": 166402
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.3269906044006348,
      "learning_rate": 0.00010716110581280268,
      "loss": 2.0362,
      "step": 166403
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.1494437456130981,
      "learning_rate": 0.00010715797230818483,
      "loss": 2.1949,
      "step": 166404
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.155286192893982,
      "learning_rate": 0.00010715483883941993,
      "loss": 2.0191,
      "step": 166405
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.2729201316833496,
      "learning_rate": 0.00010715170540650863,
      "loss": 2.1518,
      "step": 166406
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.5190515518188477,
      "learning_rate": 0.00010714857200945135,
      "loss": 1.8006,
      "step": 166407
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.2346755266189575,
      "learning_rate": 0.00010714543864824887,
      "loss": 2.1914,
      "step": 166408
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.2138234376907349,
      "learning_rate": 0.00010714230532290162,
      "loss": 2.0483,
      "step": 166409
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.15488600730896,
      "learning_rate": 0.00010713917203341036,
      "loss": 2.0773,
      "step": 166410
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.13345205783844,
      "learning_rate": 0.00010713603877977555,
      "loss": 2.0722,
      "step": 166411
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.2122807502746582,
      "learning_rate": 0.00010713290556199779,
      "loss": 1.9781,
      "step": 166412
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.1615288257598877,
      "learning_rate": 0.00010712977238007758,
      "loss": 1.9083,
      "step": 166413
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.375041127204895,
      "learning_rate": 0.00010712663923401567,
      "loss": 2.061,
      "step": 166414
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.1973868608474731,
      "learning_rate": 0.00010712350612381247,
      "loss": 1.8271,
      "step": 166415
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.2331123352050781,
      "learning_rate": 0.00010712037304946873,
      "loss": 2.1969,
      "step": 166416
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.2136913537979126,
      "learning_rate": 0.00010711724001098488,
      "loss": 2.1799,
      "step": 166417
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.18898344039917,
      "learning_rate": 0.00010711410700836174,
      "loss": 1.9237,
      "step": 166418
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.1616674661636353,
      "learning_rate": 0.00010711097404159956,
      "loss": 1.9296,
      "step": 166419
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.2646139860153198,
      "learning_rate": 0.00010710784111069919,
      "loss": 1.9837,
      "step": 166420
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.1806532144546509,
      "learning_rate": 0.00010710470821566103,
      "loss": 1.917,
      "step": 166421
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.257667899131775,
      "learning_rate": 0.00010710157535648582,
      "loss": 2.3382,
      "step": 166422
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.2912333011627197,
      "learning_rate": 0.00010709844253317397,
      "loss": 2.0851,
      "step": 166423
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.2998881340026855,
      "learning_rate": 0.00010709530974572636,
      "loss": 1.946,
      "step": 166424
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.4455984830856323,
      "learning_rate": 0.00010709217699414318,
      "loss": 2.1255,
      "step": 166425
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.1957868337631226,
      "learning_rate": 0.0001070890442784253,
      "loss": 2.0309,
      "step": 166426
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.3355079889297485,
      "learning_rate": 0.00010708591159857311,
      "loss": 2.195,
      "step": 166427
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.2577632665634155,
      "learning_rate": 0.0001070827789545874,
      "loss": 1.8227,
      "step": 166428
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.1242140531539917,
      "learning_rate": 0.00010707964634646853,
      "loss": 2.0474,
      "step": 166429
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.235187292098999,
      "learning_rate": 0.0001070765137742174,
      "loss": 2.0035,
      "step": 166430
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.3105707168579102,
      "learning_rate": 0.00010707338123783417,
      "loss": 2.0527,
      "step": 166431
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.2115387916564941,
      "learning_rate": 0.00010707024873731977,
      "loss": 1.9394,
      "step": 166432
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.1826462745666504,
      "learning_rate": 0.00010706711627267456,
      "loss": 2.0872,
      "step": 166433
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.3428187370300293,
      "learning_rate": 0.0001070639838438993,
      "loss": 2.0457,
      "step": 166434
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.334054708480835,
      "learning_rate": 0.00010706085145099437,
      "loss": 1.9279,
      "step": 166435
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.1870601177215576,
      "learning_rate": 0.00010705771909396058,
      "loss": 1.9689,
      "step": 166436
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.2256278991699219,
      "learning_rate": 0.0001070545867727984,
      "loss": 2.1209,
      "step": 166437
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.2785413265228271,
      "learning_rate": 0.00010705145448750842,
      "loss": 1.8972,
      "step": 166438
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.216633915901184,
      "learning_rate": 0.00010704832223809112,
      "loss": 1.9447,
      "step": 166439
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.1839934587478638,
      "learning_rate": 0.00010704519002454729,
      "loss": 1.9065,
      "step": 166440
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.4468061923980713,
      "learning_rate": 0.00010704205784687725,
      "loss": 1.9827,
      "step": 166441
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.2220836877822876,
      "learning_rate": 0.0001070389257050819,
      "loss": 1.9937,
      "step": 166442
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.2415363788604736,
      "learning_rate": 0.00010703579359916161,
      "loss": 2.0868,
      "step": 166443
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.1633740663528442,
      "learning_rate": 0.00010703266152911693,
      "loss": 2.0944,
      "step": 166444
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.1965315341949463,
      "learning_rate": 0.00010702952949494861,
      "loss": 1.7944,
      "step": 166445
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.1919790506362915,
      "learning_rate": 0.00010702639749665716,
      "loss": 2.0164,
      "step": 166446
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.1907020807266235,
      "learning_rate": 0.00010702326553424304,
      "loss": 1.8083,
      "step": 166447
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.268930196762085,
      "learning_rate": 0.00010702013360770703,
      "loss": 2.1688,
      "step": 166448
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.2137386798858643,
      "learning_rate": 0.00010701700171704961,
      "loss": 2.0114,
      "step": 166449
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.219558596611023,
      "learning_rate": 0.0001070138698622713,
      "loss": 2.1275,
      "step": 166450
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.1598869562149048,
      "learning_rate": 0.00010701073804337284,
      "loss": 2.0451,
      "step": 166451
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.2343404293060303,
      "learning_rate": 0.00010700760626035473,
      "loss": 2.0761,
      "step": 166452
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.2490779161453247,
      "learning_rate": 0.00010700447451321744,
      "loss": 2.0345,
      "step": 166453
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.236288070678711,
      "learning_rate": 0.00010700134280196176,
      "loss": 2.0856,
      "step": 166454
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.3887351751327515,
      "learning_rate": 0.00010699821112658816,
      "loss": 1.9725,
      "step": 166455
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.2722055912017822,
      "learning_rate": 0.00010699507948709716,
      "loss": 2.1913,
      "step": 166456
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.1605411767959595,
      "learning_rate": 0.00010699194788348953,
      "loss": 2.1112,
      "step": 166457
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.219006896018982,
      "learning_rate": 0.00010698881631576561,
      "loss": 1.8849,
      "step": 166458
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.167187213897705,
      "learning_rate": 0.00010698568478392623,
      "loss": 2.0032,
      "step": 166459
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.1922354698181152,
      "learning_rate": 0.00010698255328797185,
      "loss": 1.9601,
      "step": 166460
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.2129294872283936,
      "learning_rate": 0.00010697942182790307,
      "loss": 2.0903,
      "step": 166461
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.3460205793380737,
      "learning_rate": 0.00010697629040372034,
      "loss": 2.1564,
      "step": 166462
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.1293071508407593,
      "learning_rate": 0.00010697315901542445,
      "loss": 1.795,
      "step": 166463
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.2109382152557373,
      "learning_rate": 0.00010697002766301581,
      "loss": 2.0062,
      "step": 166464
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.2379441261291504,
      "learning_rate": 0.0001069668963464952,
      "loss": 2.1181,
      "step": 166465
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.2363792657852173,
      "learning_rate": 0.00010696376506586308,
      "loss": 1.9728,
      "step": 166466
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.263450264930725,
      "learning_rate": 0.00010696063382112002,
      "loss": 2.0328,
      "step": 166467
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.2278852462768555,
      "learning_rate": 0.00010695750261226656,
      "loss": 2.0861,
      "step": 166468
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.2225573062896729,
      "learning_rate": 0.00010695437143930341,
      "loss": 2.1616,
      "step": 166469
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.2429102659225464,
      "learning_rate": 0.000106951240302231,
      "loss": 2.2001,
      "step": 166470
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.3330373764038086,
      "learning_rate": 0.00010694810920105009,
      "loss": 2.0206,
      "step": 166471
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.1283459663391113,
      "learning_rate": 0.00010694497813576119,
      "loss": 1.9396,
      "step": 166472
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.1884591579437256,
      "learning_rate": 0.00010694184710636484,
      "loss": 1.7694,
      "step": 166473
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.368433952331543,
      "learning_rate": 0.00010693871611286154,
      "loss": 2.0599,
      "step": 166474
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.2883152961730957,
      "learning_rate": 0.00010693558515525208,
      "loss": 2.215,
      "step": 166475
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.3384054899215698,
      "learning_rate": 0.00010693245423353686,
      "loss": 2.0443,
      "step": 166476
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.2547515630722046,
      "learning_rate": 0.00010692932334771664,
      "loss": 2.0169,
      "step": 166477
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.294993281364441,
      "learning_rate": 0.00010692619249779189,
      "loss": 2.0085,
      "step": 166478
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.2122008800506592,
      "learning_rate": 0.00010692306168376319,
      "loss": 2.1882,
      "step": 166479
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.2992316484451294,
      "learning_rate": 0.00010691993090563107,
      "loss": 1.9921,
      "step": 166480
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.331078290939331,
      "learning_rate": 0.00010691680016339625,
      "loss": 2.1103,
      "step": 166481
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.3832775354385376,
      "learning_rate": 0.00010691366945705918,
      "loss": 2.1266,
      "step": 166482
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.235445261001587,
      "learning_rate": 0.00010691053878662056,
      "loss": 1.9695,
      "step": 166483
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.2022309303283691,
      "learning_rate": 0.00010690740815208086,
      "loss": 2.0212,
      "step": 166484
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.186995267868042,
      "learning_rate": 0.0001069042775534409,
      "loss": 2.062,
      "step": 166485
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.2817668914794922,
      "learning_rate": 0.00010690114699070086,
      "loss": 1.9645,
      "step": 166486
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.217376470565796,
      "learning_rate": 0.00010689801646386167,
      "loss": 2.146,
      "step": 166487
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.2755300998687744,
      "learning_rate": 0.00010689488597292366,
      "loss": 1.9475,
      "step": 166488
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.2720749378204346,
      "learning_rate": 0.00010689175551788764,
      "loss": 2.2045,
      "step": 166489
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.2094790935516357,
      "learning_rate": 0.00010688862509875399,
      "loss": 1.8612,
      "step": 166490
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.937140941619873,
      "learning_rate": 0.00010688549471552358,
      "loss": 2.03,
      "step": 166491
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.289513111114502,
      "learning_rate": 0.00010688236436819664,
      "loss": 2.1897,
      "step": 166492
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.2738103866577148,
      "learning_rate": 0.000106879234056774,
      "loss": 1.9833,
      "step": 166493
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.498224139213562,
      "learning_rate": 0.00010687610378125603,
      "loss": 2.1162,
      "step": 166494
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.1103465557098389,
      "learning_rate": 0.00010687297354164356,
      "loss": 2.0252,
      "step": 166495
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.2943116426467896,
      "learning_rate": 0.00010686984333793693,
      "loss": 1.9711,
      "step": 166496
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.1873631477355957,
      "learning_rate": 0.00010686671317013706,
      "loss": 2.1613,
      "step": 166497
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.1658198833465576,
      "learning_rate": 0.00010686358303824412,
      "loss": 1.8344,
      "step": 166498
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.2149831056594849,
      "learning_rate": 0.00010686045294225894,
      "loss": 1.9078,
      "step": 166499
    },
    {
      "epoch": 2.17,
      "grad_norm": 2.0332939624786377,
      "learning_rate": 0.00010685732288218202,
      "loss": 2.0278,
      "step": 166500
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.298004150390625,
      "learning_rate": 0.000106854192858014,
      "loss": 2.0827,
      "step": 166501
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.2452819347381592,
      "learning_rate": 0.00010685106286975539,
      "loss": 1.8799,
      "step": 166502
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.2721915245056152,
      "learning_rate": 0.00010684793291740698,
      "loss": 2.067,
      "step": 166503
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.3042271137237549,
      "learning_rate": 0.00010684480300096899,
      "loss": 2.1226,
      "step": 166504
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.2621926069259644,
      "learning_rate": 0.00010684167312044231,
      "loss": 1.9442,
      "step": 166505
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.2606523036956787,
      "learning_rate": 0.00010683854327582729,
      "loss": 2.1065,
      "step": 166506
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.2041065692901611,
      "learning_rate": 0.00010683541346712472,
      "loss": 2.0004,
      "step": 166507
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.3001984357833862,
      "learning_rate": 0.00010683228369433501,
      "loss": 2.4095,
      "step": 166508
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.1861587762832642,
      "learning_rate": 0.00010682915395745902,
      "loss": 2.0478,
      "step": 166509
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.3393503427505493,
      "learning_rate": 0.00010682602425649697,
      "loss": 2.3208,
      "step": 166510
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.304183006286621,
      "learning_rate": 0.00010682289459144968,
      "loss": 2.0012,
      "step": 166511
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.1665593385696411,
      "learning_rate": 0.00010681976496231754,
      "loss": 1.9328,
      "step": 166512
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.1369484663009644,
      "learning_rate": 0.0001068166353691014,
      "loss": 1.9437,
      "step": 166513
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.3303966522216797,
      "learning_rate": 0.00010681350581180155,
      "loss": 2.2606,
      "step": 166514
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.1968181133270264,
      "learning_rate": 0.00010681037629041894,
      "loss": 2.1096,
      "step": 166515
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.134175181388855,
      "learning_rate": 0.00010680724680495371,
      "loss": 2.0013,
      "step": 166516
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.187392234802246,
      "learning_rate": 0.00010680411735540677,
      "loss": 2.1082,
      "step": 166517
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.14407479763031,
      "learning_rate": 0.00010680098794177851,
      "loss": 1.9387,
      "step": 166518
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.1382324695587158,
      "learning_rate": 0.00010679785856406967,
      "loss": 2.1648,
      "step": 166519
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.1867088079452515,
      "learning_rate": 0.00010679472922228067,
      "loss": 1.9967,
      "step": 166520
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.2216079235076904,
      "learning_rate": 0.00010679159991641224,
      "loss": 2.0949,
      "step": 166521
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.1453523635864258,
      "learning_rate": 0.00010678847064646494,
      "loss": 2.2788,
      "step": 166522
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.2276614904403687,
      "learning_rate": 0.00010678534141243931,
      "loss": 2.0703,
      "step": 166523
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.3415261507034302,
      "learning_rate": 0.00010678221221433581,
      "loss": 1.9975,
      "step": 166524
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.163769006729126,
      "learning_rate": 0.00010677908305215527,
      "loss": 1.9367,
      "step": 166525
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.3630496263504028,
      "learning_rate": 0.000106775953925898,
      "loss": 1.941,
      "step": 166526
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.2411096096038818,
      "learning_rate": 0.00010677282483556485,
      "loss": 2.0566,
      "step": 166527
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.2402042150497437,
      "learning_rate": 0.0001067696957811563,
      "loss": 2.0622,
      "step": 166528
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.4590126276016235,
      "learning_rate": 0.00010676656676267289,
      "loss": 2.0506,
      "step": 166529
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.261094570159912,
      "learning_rate": 0.00010676343778011513,
      "loss": 2.011,
      "step": 166530
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.2300788164138794,
      "learning_rate": 0.00010676030883348381,
      "loss": 2.1842,
      "step": 166531
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.536744236946106,
      "learning_rate": 0.00010675717992277928,
      "loss": 2.0747,
      "step": 166532
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.2050421237945557,
      "learning_rate": 0.00010675405104800235,
      "loss": 2.0828,
      "step": 166533
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.3106142282485962,
      "learning_rate": 0.00010675092220915346,
      "loss": 2.0273,
      "step": 166534
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.5749417543411255,
      "learning_rate": 0.00010674779340623315,
      "loss": 2.1187,
      "step": 166535
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.4292386770248413,
      "learning_rate": 0.00010674466463924215,
      "loss": 2.1131,
      "step": 166536
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.1890162229537964,
      "learning_rate": 0.00010674153590818099,
      "loss": 2.125,
      "step": 166537
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.2917022705078125,
      "learning_rate": 0.0001067384072130501,
      "loss": 1.9602,
      "step": 166538
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.5095843076705933,
      "learning_rate": 0.00010673527855385032,
      "loss": 2.0491,
      "step": 166539
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.2644884586334229,
      "learning_rate": 0.00010673214993058209,
      "loss": 1.8966,
      "step": 166540
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.4254331588745117,
      "learning_rate": 0.0001067290213432459,
      "loss": 2.1554,
      "step": 166541
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.1987437009811401,
      "learning_rate": 0.00010672589279184252,
      "loss": 1.907,
      "step": 166542
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.352278470993042,
      "learning_rate": 0.00010672276427637233,
      "loss": 2.1117,
      "step": 166543
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.2181545495986938,
      "learning_rate": 0.00010671963579683617,
      "loss": 2.2022,
      "step": 166544
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.2532864809036255,
      "learning_rate": 0.00010671650735323443,
      "loss": 1.9428,
      "step": 166545
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.306099534034729,
      "learning_rate": 0.00010671337894556778,
      "loss": 2.1576,
      "step": 166546
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.29542875289917,
      "learning_rate": 0.00010671025057383664,
      "loss": 1.9178,
      "step": 166547
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.4659512042999268,
      "learning_rate": 0.00010670712223804183,
      "loss": 1.8618,
      "step": 166548
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.2480816841125488,
      "learning_rate": 0.00010670399393818371,
      "loss": 2.2389,
      "step": 166549
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.2684385776519775,
      "learning_rate": 0.00010670086567426303,
      "loss": 2.1938,
      "step": 166550
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.2496732473373413,
      "learning_rate": 0.00010669773744628035,
      "loss": 2.3043,
      "step": 166551
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.1961349248886108,
      "learning_rate": 0.00010669460925423622,
      "loss": 1.8811,
      "step": 166552
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.3433326482772827,
      "learning_rate": 0.00010669148109813107,
      "loss": 2.0906,
      "step": 166553
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.249577522277832,
      "learning_rate": 0.00010668835297796574,
      "loss": 2.0358,
      "step": 166554
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.4228390455245972,
      "learning_rate": 0.00010668522489374059,
      "loss": 2.1604,
      "step": 166555
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.2465871572494507,
      "learning_rate": 0.00010668209684545644,
      "loss": 1.9447,
      "step": 166556
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.2048163414001465,
      "learning_rate": 0.00010667896883311359,
      "loss": 2.1244,
      "step": 166557
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.327225923538208,
      "learning_rate": 0.00010667584085671298,
      "loss": 2.1917,
      "step": 166558
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.2232478857040405,
      "learning_rate": 0.00010667271291625481,
      "loss": 2.1224,
      "step": 166559
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.266112208366394,
      "learning_rate": 0.00010666958501173992,
      "loss": 2.0771,
      "step": 166560
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.2864329814910889,
      "learning_rate": 0.0001066664571431687,
      "loss": 2.0485,
      "step": 166561
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.2948729991912842,
      "learning_rate": 0.00010666332931054192,
      "loss": 2.082,
      "step": 166562
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.2420541048049927,
      "learning_rate": 0.00010666020151386,
      "loss": 2.1719,
      "step": 166563
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.1561137437820435,
      "learning_rate": 0.00010665707375312376,
      "loss": 2.2176,
      "step": 166564
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.2100176811218262,
      "learning_rate": 0.00010665394602833345,
      "loss": 1.9374,
      "step": 166565
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.2190901041030884,
      "learning_rate": 0.00010665081833948991,
      "loss": 1.987,
      "step": 166566
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.148711919784546,
      "learning_rate": 0.00010664769068659353,
      "loss": 1.9536,
      "step": 166567
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.1754909753799438,
      "learning_rate": 0.0001066445630696451,
      "loss": 1.9846,
      "step": 166568
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.245072364807129,
      "learning_rate": 0.00010664143548864502,
      "loss": 2.1427,
      "step": 166569
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.19086492061615,
      "learning_rate": 0.00010663830794359412,
      "loss": 2.0792,
      "step": 166570
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.2342931032180786,
      "learning_rate": 0.00010663518043449262,
      "loss": 2.0986,
      "step": 166571
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.2534712553024292,
      "learning_rate": 0.0001066320529613414,
      "loss": 2.0461,
      "step": 166572
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.2380454540252686,
      "learning_rate": 0.00010662892552414079,
      "loss": 1.97,
      "step": 166573
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.2514806985855103,
      "learning_rate": 0.00010662579812289165,
      "loss": 2.0651,
      "step": 166574
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.2924957275390625,
      "learning_rate": 0.00010662267075759432,
      "loss": 2.2268,
      "step": 166575
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.4376153945922852,
      "learning_rate": 0.00010661954342824967,
      "loss": 1.9792,
      "step": 166576
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.5411642789840698,
      "learning_rate": 0.00010661641613485788,
      "loss": 1.9861,
      "step": 166577
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.1867566108703613,
      "learning_rate": 0.00010661328887741991,
      "loss": 1.9541,
      "step": 166578
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.373972773551941,
      "learning_rate": 0.00010661016165593604,
      "loss": 2.0816,
      "step": 166579
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.2704555988311768,
      "learning_rate": 0.00010660703447040708,
      "loss": 2.016,
      "step": 166580
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.186523675918579,
      "learning_rate": 0.00010660390732083342,
      "loss": 2.1545,
      "step": 166581
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.2364771366119385,
      "learning_rate": 0.00010660078020721597,
      "loss": 1.9256,
      "step": 166582
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.1813973188400269,
      "learning_rate": 0.0001065976531295549,
      "loss": 1.958,
      "step": 166583
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.2242218255996704,
      "learning_rate": 0.00010659452608785105,
      "loss": 1.897,
      "step": 166584
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.194360375404358,
      "learning_rate": 0.00010659139908210484,
      "loss": 2.1913,
      "step": 166585
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.2641935348510742,
      "learning_rate": 0.00010658827211231704,
      "loss": 2.2034,
      "step": 166586
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.2005001306533813,
      "learning_rate": 0.000106585145178488,
      "loss": 1.9055,
      "step": 166587
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.194929599761963,
      "learning_rate": 0.00010658201828061868,
      "loss": 2.1283,
      "step": 166588
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.2072588205337524,
      "learning_rate": 0.00010657889141870917,
      "loss": 1.8881,
      "step": 166589
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.1689256429672241,
      "learning_rate": 0.00010657576459276042,
      "loss": 2.0155,
      "step": 166590
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.2177834510803223,
      "learning_rate": 0.00010657263780277279,
      "loss": 1.7487,
      "step": 166591
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.275284767150879,
      "learning_rate": 0.00010656951104874706,
      "loss": 2.0241,
      "step": 166592
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.2189496755599976,
      "learning_rate": 0.0001065663843306836,
      "loss": 1.8849,
      "step": 166593
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.2078526020050049,
      "learning_rate": 0.0001065632576485833,
      "loss": 1.8056,
      "step": 166594
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.1939129829406738,
      "learning_rate": 0.00010656013100244634,
      "loss": 1.906,
      "step": 166595
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.366563081741333,
      "learning_rate": 0.0001065570043922736,
      "loss": 1.9956,
      "step": 166596
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.4635210037231445,
      "learning_rate": 0.00010655387781806546,
      "loss": 2.004,
      "step": 166597
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.4668182134628296,
      "learning_rate": 0.00010655075127982274,
      "loss": 1.9564,
      "step": 166598
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.2933528423309326,
      "learning_rate": 0.00010654762477754572,
      "loss": 2.0799,
      "step": 166599
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.2413949966430664,
      "learning_rate": 0.00010654449831123542,
      "loss": 1.944,
      "step": 166600
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.2006325721740723,
      "learning_rate": 0.00010654137188089191,
      "loss": 2.0176,
      "step": 166601
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.4413516521453857,
      "learning_rate": 0.00010653824548651611,
      "loss": 2.1384,
      "step": 166602
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.416642189025879,
      "learning_rate": 0.00010653511912810843,
      "loss": 2.2225,
      "step": 166603
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.3678086996078491,
      "learning_rate": 0.0001065319928056696,
      "loss": 1.9218,
      "step": 166604
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.2498180866241455,
      "learning_rate": 0.00010652886651920003,
      "loss": 1.9743,
      "step": 166605
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.32511305809021,
      "learning_rate": 0.00010652574026870049,
      "loss": 1.9598,
      "step": 166606
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.2311793565750122,
      "learning_rate": 0.00010652261405417148,
      "loss": 1.9523,
      "step": 166607
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.3289496898651123,
      "learning_rate": 0.00010651948787561355,
      "loss": 2.2136,
      "step": 166608
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.3544361591339111,
      "learning_rate": 0.00010651636173302722,
      "loss": 2.0789,
      "step": 166609
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.1651514768600464,
      "learning_rate": 0.00010651323562641325,
      "loss": 2.0559,
      "step": 166610
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.2313635349273682,
      "learning_rate": 0.000106510109555772,
      "loss": 2.1766,
      "step": 166611
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.2590959072113037,
      "learning_rate": 0.00010650698352110428,
      "loss": 2.2136,
      "step": 166612
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.3182790279388428,
      "learning_rate": 0.00010650385752241059,
      "loss": 2.3348,
      "step": 166613
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.1659505367279053,
      "learning_rate": 0.00010650073155969147,
      "loss": 2.0349,
      "step": 166614
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.2077738046646118,
      "learning_rate": 0.00010649760563294739,
      "loss": 2.3409,
      "step": 166615
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.2434966564178467,
      "learning_rate": 0.0001064944797421792,
      "loss": 1.8874,
      "step": 166616
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.2431665658950806,
      "learning_rate": 0.00010649135388738722,
      "loss": 2.2529,
      "step": 166617
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.2798372507095337,
      "learning_rate": 0.00010648822806857223,
      "loss": 2.1773,
      "step": 166618
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.2059420347213745,
      "learning_rate": 0.00010648510228573477,
      "loss": 2.1597,
      "step": 166619
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.1904884576797485,
      "learning_rate": 0.00010648197653887526,
      "loss": 2.0419,
      "step": 166620
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.489724040031433,
      "learning_rate": 0.0001064788508279945,
      "loss": 1.8724,
      "step": 166621
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.7392619848251343,
      "learning_rate": 0.000106475725153093,
      "loss": 1.9925,
      "step": 166622
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.2094019651412964,
      "learning_rate": 0.00010647259951417117,
      "loss": 2.0923,
      "step": 166623
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.1798429489135742,
      "learning_rate": 0.00010646947391122988,
      "loss": 1.9365,
      "step": 166624
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.2092314958572388,
      "learning_rate": 0.00010646634834426956,
      "loss": 2.0363,
      "step": 166625
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.208932638168335,
      "learning_rate": 0.00010646322281329066,
      "loss": 1.7872,
      "step": 166626
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.3181474208831787,
      "learning_rate": 0.00010646009731829404,
      "loss": 2.0245,
      "step": 166627
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.3141899108886719,
      "learning_rate": 0.00010645697185928014,
      "loss": 1.984,
      "step": 166628
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.297648549079895,
      "learning_rate": 0.00010645384643624944,
      "loss": 2.1824,
      "step": 166629
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.1919641494750977,
      "learning_rate": 0.0001064507210492027,
      "loss": 2.0888,
      "step": 166630
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.470934510231018,
      "learning_rate": 0.00010644759569814045,
      "loss": 1.923,
      "step": 166631
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.3662619590759277,
      "learning_rate": 0.00010644447038306315,
      "loss": 1.8908,
      "step": 166632
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.1265344619750977,
      "learning_rate": 0.00010644134510397157,
      "loss": 2.0782,
      "step": 166633
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.252021312713623,
      "learning_rate": 0.0001064382198608661,
      "loss": 2.1863,
      "step": 166634
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.1567332744598389,
      "learning_rate": 0.00010643509465374749,
      "loss": 2.073,
      "step": 166635
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.2764315605163574,
      "learning_rate": 0.00010643196948261628,
      "loss": 2.3309,
      "step": 166636
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.1685415506362915,
      "learning_rate": 0.00010642884434747302,
      "loss": 1.8912,
      "step": 166637
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.2767506837844849,
      "learning_rate": 0.00010642571924831817,
      "loss": 2.1508,
      "step": 166638
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.2464042901992798,
      "learning_rate": 0.00010642259418515256,
      "loss": 1.9295,
      "step": 166639
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.1861414909362793,
      "learning_rate": 0.00010641946915797652,
      "loss": 1.9245,
      "step": 166640
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.107231855392456,
      "learning_rate": 0.00010641634416679087,
      "loss": 2.0262,
      "step": 166641
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.1274858713150024,
      "learning_rate": 0.00010641321921159595,
      "loss": 1.9054,
      "step": 166642
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.544765830039978,
      "learning_rate": 0.0001064100942923927,
      "loss": 2.0748,
      "step": 166643
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.2077492475509644,
      "learning_rate": 0.00010640696940918126,
      "loss": 2.0593,
      "step": 166644
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.1219655275344849,
      "learning_rate": 0.00010640384456196251,
      "loss": 1.9008,
      "step": 166645
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.3276718854904175,
      "learning_rate": 0.00010640071975073685,
      "loss": 1.7779,
      "step": 166646
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.1138670444488525,
      "learning_rate": 0.00010639759497550504,
      "loss": 2.0583,
      "step": 166647
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.2188783884048462,
      "learning_rate": 0.00010639447023626744,
      "loss": 2.0626,
      "step": 166648
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.149925708770752,
      "learning_rate": 0.00010639134553302504,
      "loss": 1.9299,
      "step": 166649
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.245069980621338,
      "learning_rate": 0.00010638822086577788,
      "loss": 2.2961,
      "step": 166650
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.2486974000930786,
      "learning_rate": 0.00010638509623452694,
      "loss": 2.0403,
      "step": 166651
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.2501037120819092,
      "learning_rate": 0.00010638197163927255,
      "loss": 1.8842,
      "step": 166652
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.2288708686828613,
      "learning_rate": 0.00010637884708001548,
      "loss": 2.0871,
      "step": 166653
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.3115586042404175,
      "learning_rate": 0.00010637572255675619,
      "loss": 2.1283,
      "step": 166654
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.1522578001022339,
      "learning_rate": 0.00010637259806949548,
      "loss": 2.0285,
      "step": 166655
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.3535112142562866,
      "learning_rate": 0.00010636947361823356,
      "loss": 2.1733,
      "step": 166656
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.2449537515640259,
      "learning_rate": 0.00010636634920297131,
      "loss": 1.9929,
      "step": 166657
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.2781264781951904,
      "learning_rate": 0.0001063632248237091,
      "loss": 2.0114,
      "step": 166658
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.273987054824829,
      "learning_rate": 0.00010636010048044777,
      "loss": 1.7546,
      "step": 166659
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.3632192611694336,
      "learning_rate": 0.0001063569761731876,
      "loss": 2.0407,
      "step": 166660
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.3720678091049194,
      "learning_rate": 0.00010635385190192955,
      "loss": 1.9641,
      "step": 166661
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.4401646852493286,
      "learning_rate": 0.00010635072766667375,
      "loss": 2.3396,
      "step": 166662
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.2632266283035278,
      "learning_rate": 0.00010634760346742111,
      "loss": 1.9714,
      "step": 166663
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.3396756649017334,
      "learning_rate": 0.00010634447930417202,
      "loss": 2.2002,
      "step": 166664
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.4435049295425415,
      "learning_rate": 0.00010634135517692725,
      "loss": 1.9429,
      "step": 166665
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.2470202445983887,
      "learning_rate": 0.00010633823108568718,
      "loss": 2.0247,
      "step": 166666
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.3249832391738892,
      "learning_rate": 0.00010633510703045267,
      "loss": 1.9514,
      "step": 166667
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.3259851932525635,
      "learning_rate": 0.00010633198301122393,
      "loss": 2.2078,
      "step": 166668
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.3319989442825317,
      "learning_rate": 0.00010632885902800182,
      "loss": 1.8322,
      "step": 166669
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.201853632926941,
      "learning_rate": 0.00010632573508078675,
      "loss": 2.0744,
      "step": 166670
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.3686238527297974,
      "learning_rate": 0.00010632261116957947,
      "loss": 2.3017,
      "step": 166671
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.2408238649368286,
      "learning_rate": 0.00010631948729438038,
      "loss": 2.0667,
      "step": 166672
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.190260887145996,
      "learning_rate": 0.00010631636345519032,
      "loss": 2.1865,
      "step": 166673
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.4830632209777832,
      "learning_rate": 0.00010631323965200956,
      "loss": 1.9322,
      "step": 166674
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.5131351947784424,
      "learning_rate": 0.00010631011588483889,
      "loss": 2.0809,
      "step": 166675
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.6250170469284058,
      "learning_rate": 0.00010630699215367872,
      "loss": 2.197,
      "step": 166676
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.186886191368103,
      "learning_rate": 0.00010630386845852985,
      "loss": 1.7926,
      "step": 166677
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.303533673286438,
      "learning_rate": 0.00010630074479939263,
      "loss": 1.8907,
      "step": 166678
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.670432448387146,
      "learning_rate": 0.00010629762117626798,
      "loss": 2.0488,
      "step": 166679
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.6128731966018677,
      "learning_rate": 0.00010629449758915609,
      "loss": 2.0336,
      "step": 166680
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.4393677711486816,
      "learning_rate": 0.00010629137403805774,
      "loss": 1.8643,
      "step": 166681
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.249639630317688,
      "learning_rate": 0.00010628825052297343,
      "loss": 2.0964,
      "step": 166682
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.574544906616211,
      "learning_rate": 0.00010628512704390387,
      "loss": 2.0113,
      "step": 166683
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.3152498006820679,
      "learning_rate": 0.00010628200360084945,
      "loss": 2.1061,
      "step": 166684
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.6156232357025146,
      "learning_rate": 0.00010627888019381097,
      "loss": 2.0712,
      "step": 166685
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.3167470693588257,
      "learning_rate": 0.00010627575682278891,
      "loss": 2.115,
      "step": 166686
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.308005928993225,
      "learning_rate": 0.00010627263348778385,
      "loss": 1.91,
      "step": 166687
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.3138447999954224,
      "learning_rate": 0.00010626951018879625,
      "loss": 1.9761,
      "step": 166688
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.4794766902923584,
      "learning_rate": 0.00010626638692582692,
      "loss": 1.8017,
      "step": 166689
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.2763789892196655,
      "learning_rate": 0.0001062632636988762,
      "loss": 2.0517,
      "step": 166690
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.387089729309082,
      "learning_rate": 0.00010626014050794492,
      "loss": 2.28,
      "step": 166691
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.3633404970169067,
      "learning_rate": 0.00010625701735303353,
      "loss": 2.0776,
      "step": 166692
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.213261365890503,
      "learning_rate": 0.00010625389423414263,
      "loss": 2.1549,
      "step": 166693
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.200297236442566,
      "learning_rate": 0.00010625077115127269,
      "loss": 1.914,
      "step": 166694
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.3088866472244263,
      "learning_rate": 0.00010624764810442446,
      "loss": 1.957,
      "step": 166695
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.3263556957244873,
      "learning_rate": 0.00010624452509359837,
      "loss": 2.0728,
      "step": 166696
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.3573797941207886,
      "learning_rate": 0.00010624140211879519,
      "loss": 2.096,
      "step": 166697
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.3552154302597046,
      "learning_rate": 0.00010623827918001534,
      "loss": 2.0558,
      "step": 166698
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.3073375225067139,
      "learning_rate": 0.00010623515627725951,
      "loss": 1.996,
      "step": 166699
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.2914563417434692,
      "learning_rate": 0.00010623203341052811,
      "loss": 1.9802,
      "step": 166700
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.35857355594635,
      "learning_rate": 0.00010622891057982191,
      "loss": 2.0105,
      "step": 166701
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.2369401454925537,
      "learning_rate": 0.0001062257877851413,
      "loss": 2.1729,
      "step": 166702
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.3058172464370728,
      "learning_rate": 0.00010622266502648711,
      "loss": 2.2611,
      "step": 166703
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.1832727193832397,
      "learning_rate": 0.0001062195423038598,
      "loss": 1.9249,
      "step": 166704
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.3667165040969849,
      "learning_rate": 0.00010621641961725979,
      "loss": 1.9552,
      "step": 166705
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.2116881608963013,
      "learning_rate": 0.00010621329696668792,
      "loss": 2.0836,
      "step": 166706
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.1778799295425415,
      "learning_rate": 0.00010621017435214468,
      "loss": 1.8763,
      "step": 166707
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.2826905250549316,
      "learning_rate": 0.00010620705177363048,
      "loss": 2.1408,
      "step": 166708
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.2069149017333984,
      "learning_rate": 0.00010620392923114617,
      "loss": 1.9122,
      "step": 166709
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.2319772243499756,
      "learning_rate": 0.00010620080672469224,
      "loss": 1.8537,
      "step": 166710
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.3142900466918945,
      "learning_rate": 0.00010619768425426912,
      "loss": 2.1056,
      "step": 166711
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.2952020168304443,
      "learning_rate": 0.00010619456181987756,
      "loss": 2.211,
      "step": 166712
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.1769522428512573,
      "learning_rate": 0.00010619143942151813,
      "loss": 1.9936,
      "step": 166713
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.2375143766403198,
      "learning_rate": 0.00010618831705919127,
      "loss": 1.9326,
      "step": 166714
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.371113896369934,
      "learning_rate": 0.00010618519473289776,
      "loss": 2.2088,
      "step": 166715
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.259922981262207,
      "learning_rate": 0.00010618207244263809,
      "loss": 1.949,
      "step": 166716
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.259675145149231,
      "learning_rate": 0.00010617895018841273,
      "loss": 2.2411,
      "step": 166717
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.287957787513733,
      "learning_rate": 0.00010617582797022243,
      "loss": 1.9445,
      "step": 166718
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.2126123905181885,
      "learning_rate": 0.00010617270578806762,
      "loss": 2.1384,
      "step": 166719
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.3367247581481934,
      "learning_rate": 0.00010616958364194908,
      "loss": 2.1886,
      "step": 166720
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.3171507120132446,
      "learning_rate": 0.00010616646153186725,
      "loss": 2.1053,
      "step": 166721
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.335219383239746,
      "learning_rate": 0.00010616333945782277,
      "loss": 1.984,
      "step": 166722
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.3694453239440918,
      "learning_rate": 0.00010616021741981604,
      "loss": 2.1549,
      "step": 166723
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.1313353776931763,
      "learning_rate": 0.00010615709541784793,
      "loss": 2.1884,
      "step": 166724
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.1606866121292114,
      "learning_rate": 0.00010615397345191872,
      "loss": 2.0191,
      "step": 166725
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.307842493057251,
      "learning_rate": 0.00010615085152202928,
      "loss": 1.9525,
      "step": 166726
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.251167893409729,
      "learning_rate": 0.00010614772962817995,
      "loss": 2.0786,
      "step": 166727
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.2050279378890991,
      "learning_rate": 0.00010614460777037162,
      "loss": 1.9614,
      "step": 166728
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.5418447256088257,
      "learning_rate": 0.00010614148594860445,
      "loss": 2.0204,
      "step": 166729
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.2411383390426636,
      "learning_rate": 0.00010613836416287937,
      "loss": 1.9699,
      "step": 166730
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.1662683486938477,
      "learning_rate": 0.0001061352424131967,
      "loss": 1.7348,
      "step": 166731
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.2030233144760132,
      "learning_rate": 0.00010613212069955725,
      "loss": 1.8078,
      "step": 166732
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.1922746896743774,
      "learning_rate": 0.0001061289990219614,
      "loss": 2.1853,
      "step": 166733
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.2788324356079102,
      "learning_rate": 0.00010612587738041006,
      "loss": 1.9708,
      "step": 166734
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.293142557144165,
      "learning_rate": 0.00010612275577490334,
      "loss": 2.1968,
      "step": 166735
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.2551318407058716,
      "learning_rate": 0.00010611963420544216,
      "loss": 2.0036,
      "step": 166736
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.2415121793746948,
      "learning_rate": 0.0001061165126720269,
      "loss": 1.9547,
      "step": 166737
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.2008370161056519,
      "learning_rate": 0.00010611339117465834,
      "loss": 2.1358,
      "step": 166738
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.4056217670440674,
      "learning_rate": 0.0001061102697133369,
      "loss": 2.0575,
      "step": 166739
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.268198013305664,
      "learning_rate": 0.00010610714828806337,
      "loss": 2.0259,
      "step": 166740
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.3465549945831299,
      "learning_rate": 0.00010610402689883801,
      "loss": 2.0406,
      "step": 166741
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.3642873764038086,
      "learning_rate": 0.00010610090554566168,
      "loss": 2.0963,
      "step": 166742
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.5654906034469604,
      "learning_rate": 0.00010609778422853472,
      "loss": 1.7712,
      "step": 166743
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.2135567665100098,
      "learning_rate": 0.00010609466294745795,
      "loss": 2.0163,
      "step": 166744
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.2700445652008057,
      "learning_rate": 0.00010609154170243174,
      "loss": 2.1039,
      "step": 166745
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.2760448455810547,
      "learning_rate": 0.00010608842049345701,
      "loss": 1.9745,
      "step": 166746
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.226383090019226,
      "learning_rate": 0.00010608529932053389,
      "loss": 1.9354,
      "step": 166747
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.2304863929748535,
      "learning_rate": 0.00010608217818366324,
      "loss": 2.0898,
      "step": 166748
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.2277567386627197,
      "learning_rate": 0.00010607905708284552,
      "loss": 2.0612,
      "step": 166749
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.2465636730194092,
      "learning_rate": 0.0001060759360180814,
      "loss": 2.0098,
      "step": 166750
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.2356207370758057,
      "learning_rate": 0.00010607281498937137,
      "loss": 2.1418,
      "step": 166751
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.2486954927444458,
      "learning_rate": 0.00010606969399671615,
      "loss": 2.0665,
      "step": 166752
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.0780991315841675,
      "learning_rate": 0.00010606657304011626,
      "loss": 1.8861,
      "step": 166753
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.2298328876495361,
      "learning_rate": 0.00010606345211957224,
      "loss": 1.8842,
      "step": 166754
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.233685851097107,
      "learning_rate": 0.00010606033123508457,
      "loss": 2.1798,
      "step": 166755
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.4882086515426636,
      "learning_rate": 0.00010605721038665404,
      "loss": 2.1998,
      "step": 166756
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.3298419713974,
      "learning_rate": 0.00010605408957428106,
      "loss": 2.2645,
      "step": 166757
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.242964744567871,
      "learning_rate": 0.00010605096879796639,
      "loss": 2.0559,
      "step": 166758
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.2149871587753296,
      "learning_rate": 0.00010604784805771048,
      "loss": 2.1285,
      "step": 166759
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.2157231569290161,
      "learning_rate": 0.00010604472735351394,
      "loss": 1.8842,
      "step": 166760
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.1920055150985718,
      "learning_rate": 0.00010604160668537728,
      "loss": 1.9071,
      "step": 166761
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.246071219444275,
      "learning_rate": 0.0001060384860533012,
      "loss": 1.9885,
      "step": 166762
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.3799729347229004,
      "learning_rate": 0.00010603536545728613,
      "loss": 2.0857,
      "step": 166763
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.2519334554672241,
      "learning_rate": 0.0001060322448973329,
      "loss": 2.0168,
      "step": 166764
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.188373327255249,
      "learning_rate": 0.00010602912437344188,
      "loss": 2.1204,
      "step": 166765
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.0663881301879883,
      "learning_rate": 0.00010602600388561375,
      "loss": 1.9663,
      "step": 166766
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.4728327989578247,
      "learning_rate": 0.00010602288343384892,
      "loss": 1.9652,
      "step": 166767
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.3228752613067627,
      "learning_rate": 0.00010601976301814819,
      "loss": 2.1434,
      "step": 166768
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.3207005262374878,
      "learning_rate": 0.00010601664263851198,
      "loss": 2.042,
      "step": 166769
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.1806416511535645,
      "learning_rate": 0.00010601352229494102,
      "loss": 1.9343,
      "step": 166770
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.2054225206375122,
      "learning_rate": 0.00010601040198743581,
      "loss": 1.9946,
      "step": 166771
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.4103126525878906,
      "learning_rate": 0.00010600728171599692,
      "loss": 1.9815,
      "step": 166772
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.4644793272018433,
      "learning_rate": 0.00010600416148062483,
      "loss": 2.1905,
      "step": 166773
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.1792113780975342,
      "learning_rate": 0.00010600104128132034,
      "loss": 1.9176,
      "step": 166774
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.2839335203170776,
      "learning_rate": 0.00010599792111808383,
      "loss": 2.0418,
      "step": 166775
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.249739170074463,
      "learning_rate": 0.00010599480099091607,
      "loss": 2.0222,
      "step": 166776
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.3440454006195068,
      "learning_rate": 0.00010599168089981755,
      "loss": 1.9791,
      "step": 166777
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.2649955749511719,
      "learning_rate": 0.00010598856084478881,
      "loss": 1.8472,
      "step": 166778
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.2606346607208252,
      "learning_rate": 0.00010598544082583034,
      "loss": 2.4028,
      "step": 166779
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.1933423280715942,
      "learning_rate": 0.00010598232084294297,
      "loss": 1.9189,
      "step": 166780
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.2688045501708984,
      "learning_rate": 0.00010597920089612704,
      "loss": 2.1336,
      "step": 166781
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.2272148132324219,
      "learning_rate": 0.00010597608098538335,
      "loss": 2.0091,
      "step": 166782
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.110366702079773,
      "learning_rate": 0.00010597296111071237,
      "loss": 1.9718,
      "step": 166783
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.1139556169509888,
      "learning_rate": 0.00010596984127211468,
      "loss": 1.9251,
      "step": 166784
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.2283523082733154,
      "learning_rate": 0.00010596672146959072,
      "loss": 2.0614,
      "step": 166785
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.2804033756256104,
      "learning_rate": 0.00010596360170314135,
      "loss": 2.1224,
      "step": 166786
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.5690070390701294,
      "learning_rate": 0.00010596048197276691,
      "loss": 2.1035,
      "step": 166787
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.2945911884307861,
      "learning_rate": 0.00010595736227846815,
      "loss": 2.0347,
      "step": 166788
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.2357642650604248,
      "learning_rate": 0.0001059542426202456,
      "loss": 2.0335,
      "step": 166789
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.259954571723938,
      "learning_rate": 0.00010595112299809983,
      "loss": 2.1131,
      "step": 166790
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.2749234437942505,
      "learning_rate": 0.00010594800341203129,
      "loss": 2.0035,
      "step": 166791
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.2967758178710938,
      "learning_rate": 0.0001059448838620408,
      "loss": 2.1983,
      "step": 166792
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.2095389366149902,
      "learning_rate": 0.00010594176434812868,
      "loss": 2.1192,
      "step": 166793
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.2097203731536865,
      "learning_rate": 0.00010593864487029577,
      "loss": 1.6732,
      "step": 166794
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.3715951442718506,
      "learning_rate": 0.00010593552542854252,
      "loss": 2.1469,
      "step": 166795
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.2432851791381836,
      "learning_rate": 0.00010593240602286945,
      "loss": 2.0942,
      "step": 166796
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.604296088218689,
      "learning_rate": 0.0001059292866532773,
      "loss": 1.856,
      "step": 166797
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.5603280067443848,
      "learning_rate": 0.00010592616731976654,
      "loss": 2.0067,
      "step": 166798
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.32188081741333,
      "learning_rate": 0.00010592304802233768,
      "loss": 2.0267,
      "step": 166799
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.3781808614730835,
      "learning_rate": 0.0001059199287609915,
      "loss": 1.9687,
      "step": 166800
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.4123345613479614,
      "learning_rate": 0.00010591680953572844,
      "loss": 1.8564,
      "step": 166801
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.3833357095718384,
      "learning_rate": 0.00010591369034654902,
      "loss": 1.9892,
      "step": 166802
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.2909770011901855,
      "learning_rate": 0.00010591057119345402,
      "loss": 2.1914,
      "step": 166803
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.1744554042816162,
      "learning_rate": 0.00010590745207644377,
      "loss": 2.1205,
      "step": 166804
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.274294376373291,
      "learning_rate": 0.00010590433299551913,
      "loss": 2.1924,
      "step": 166805
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.2309634685516357,
      "learning_rate": 0.00010590121395068052,
      "loss": 1.9281,
      "step": 166806
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.3411973714828491,
      "learning_rate": 0.00010589809494192855,
      "loss": 2.029,
      "step": 166807
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.3598233461380005,
      "learning_rate": 0.00010589497596926371,
      "loss": 2.2312,
      "step": 166808
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.1814992427825928,
      "learning_rate": 0.00010589185703268671,
      "loss": 1.8652,
      "step": 166809
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.3387912511825562,
      "learning_rate": 0.00010588873813219801,
      "loss": 2.1796,
      "step": 166810
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.2063626050949097,
      "learning_rate": 0.00010588561926779836,
      "loss": 2.2186,
      "step": 166811
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.4794609546661377,
      "learning_rate": 0.00010588250043948822,
      "loss": 2.0058,
      "step": 166812
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.2126293182373047,
      "learning_rate": 0.0001058793816472682,
      "loss": 2.1436,
      "step": 166813
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.3676468133926392,
      "learning_rate": 0.00010587626289113875,
      "loss": 2.1205,
      "step": 166814
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.2199559211730957,
      "learning_rate": 0.00010587314417110067,
      "loss": 2.0722,
      "step": 166815
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.1344355344772339,
      "learning_rate": 0.00010587002548715434,
      "loss": 1.9678,
      "step": 166816
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.2398310899734497,
      "learning_rate": 0.00010586690683930054,
      "loss": 1.9777,
      "step": 166817
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.1963552236557007,
      "learning_rate": 0.00010586378822753965,
      "loss": 2.0357,
      "step": 166818
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.3100318908691406,
      "learning_rate": 0.00010586066965187244,
      "loss": 2.1297,
      "step": 166819
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.245361566543579,
      "learning_rate": 0.0001058575511122994,
      "loss": 2.0467,
      "step": 166820
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.4212965965270996,
      "learning_rate": 0.00010585443260882111,
      "loss": 2.0462,
      "step": 166821
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.3270658254623413,
      "learning_rate": 0.00010585131414143806,
      "loss": 2.1549,
      "step": 166822
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.2685054540634155,
      "learning_rate": 0.000105848195710151,
      "loss": 1.63,
      "step": 166823
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.1897315979003906,
      "learning_rate": 0.00010584507731496033,
      "loss": 1.9172,
      "step": 166824
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.295343041419983,
      "learning_rate": 0.00010584195895586681,
      "loss": 2.0118,
      "step": 166825
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.2799946069717407,
      "learning_rate": 0.00010583884063287093,
      "loss": 2.0656,
      "step": 166826
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.501322865486145,
      "learning_rate": 0.0001058357223459733,
      "loss": 2.1914,
      "step": 166827
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.431107521057129,
      "learning_rate": 0.00010583260409517435,
      "loss": 1.7718,
      "step": 166828
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.3933510780334473,
      "learning_rate": 0.00010582948588047491,
      "loss": 2.1721,
      "step": 166829
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.2679091691970825,
      "learning_rate": 0.00010582636770187531,
      "loss": 2.3299,
      "step": 166830
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.2205742597579956,
      "learning_rate": 0.00010582324955937634,
      "loss": 1.9764,
      "step": 166831
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.2237739562988281,
      "learning_rate": 0.00010582013145297853,
      "loss": 2.0474,
      "step": 166832
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.2608067989349365,
      "learning_rate": 0.0001058170133826824,
      "loss": 1.9892,
      "step": 166833
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.250312328338623,
      "learning_rate": 0.00010581389534848846,
      "loss": 2.0243,
      "step": 166834
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.2842386960983276,
      "learning_rate": 0.00010581077735039752,
      "loss": 1.9656,
      "step": 166835
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.3519418239593506,
      "learning_rate": 0.00010580765938840985,
      "loss": 2.1246,
      "step": 166836
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.23273503780365,
      "learning_rate": 0.00010580454146252634,
      "loss": 1.9657,
      "step": 166837
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.2882709503173828,
      "learning_rate": 0.00010580142357274745,
      "loss": 1.8233,
      "step": 166838
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.0311497449874878,
      "learning_rate": 0.0001057983057190737,
      "loss": 1.864,
      "step": 166839
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.2737867832183838,
      "learning_rate": 0.00010579518790150565,
      "loss": 2.0703,
      "step": 166840
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.2164921760559082,
      "learning_rate": 0.00010579207012004401,
      "loss": 2.0838,
      "step": 166841
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.2688136100769043,
      "learning_rate": 0.0001057889523746892,
      "loss": 1.9074,
      "step": 166842
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.2465041875839233,
      "learning_rate": 0.00010578583466544199,
      "loss": 1.8042,
      "step": 166843
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.1501431465148926,
      "learning_rate": 0.00010578271699230288,
      "loss": 2.0853,
      "step": 166844
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.3391860723495483,
      "learning_rate": 0.00010577959935527241,
      "loss": 2.2506,
      "step": 166845
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.2398571968078613,
      "learning_rate": 0.00010577648175435108,
      "loss": 2.0159,
      "step": 166846
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.1741416454315186,
      "learning_rate": 0.00010577336418953964,
      "loss": 2.1864,
      "step": 166847
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.3998976945877075,
      "learning_rate": 0.00010577024666083855,
      "loss": 1.9533,
      "step": 166848
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.195191740989685,
      "learning_rate": 0.0001057671291682485,
      "loss": 1.9963,
      "step": 166849
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.2734407186508179,
      "learning_rate": 0.00010576401171177001,
      "loss": 2.008,
      "step": 166850
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.3326306343078613,
      "learning_rate": 0.0001057608942914037,
      "loss": 2.1914,
      "step": 166851
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.215414047241211,
      "learning_rate": 0.00010575777690714997,
      "loss": 2.0507,
      "step": 166852
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.226327657699585,
      "learning_rate": 0.00010575465955900964,
      "loss": 1.9499,
      "step": 166853
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.3459914922714233,
      "learning_rate": 0.00010575154224698306,
      "loss": 2.1007,
      "step": 166854
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.2825175523757935,
      "learning_rate": 0.00010574842497107106,
      "loss": 1.9011,
      "step": 166855
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.1721594333648682,
      "learning_rate": 0.0001057453077312741,
      "loss": 1.9163,
      "step": 166856
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.2271183729171753,
      "learning_rate": 0.00010574219052759276,
      "loss": 2.0217,
      "step": 166857
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.264500379562378,
      "learning_rate": 0.00010573907336002748,
      "loss": 1.9523,
      "step": 166858
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.2786725759506226,
      "learning_rate": 0.00010573595622857912,
      "loss": 2.0454,
      "step": 166859
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.3172404766082764,
      "learning_rate": 0.00010573283913324796,
      "loss": 2.061,
      "step": 166860
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.2478969097137451,
      "learning_rate": 0.00010572972207403486,
      "loss": 2.1236,
      "step": 166861
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.260366678237915,
      "learning_rate": 0.0001057266050509403,
      "loss": 2.2825,
      "step": 166862
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.1571080684661865,
      "learning_rate": 0.0001057234880639648,
      "loss": 1.9993,
      "step": 166863
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.1634366512298584,
      "learning_rate": 0.00010572037111310884,
      "loss": 1.966,
      "step": 166864
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.2344629764556885,
      "learning_rate": 0.0001057172541983733,
      "loss": 2.1693,
      "step": 166865
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.2826076745986938,
      "learning_rate": 0.00010571413731975843,
      "loss": 2.0461,
      "step": 166866
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.1686896085739136,
      "learning_rate": 0.0001057110204772651,
      "loss": 2.0967,
      "step": 166867
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.2493408918380737,
      "learning_rate": 0.00010570790367089377,
      "loss": 1.8736,
      "step": 166868
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.282690405845642,
      "learning_rate": 0.00010570478690064498,
      "loss": 1.8392,
      "step": 166869
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.2825068235397339,
      "learning_rate": 0.00010570167016651925,
      "loss": 2.2078,
      "step": 166870
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.228622317314148,
      "learning_rate": 0.00010569855346851734,
      "loss": 2.0525,
      "step": 166871
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.2221108675003052,
      "learning_rate": 0.00010569543680663963,
      "loss": 1.9221,
      "step": 166872
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.1437747478485107,
      "learning_rate": 0.00010569232018088692,
      "loss": 1.9502,
      "step": 166873
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.128296136856079,
      "learning_rate": 0.00010568920359125968,
      "loss": 2.0958,
      "step": 166874
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.2331191301345825,
      "learning_rate": 0.00010568608703775849,
      "loss": 1.9133,
      "step": 166875
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.2907311916351318,
      "learning_rate": 0.00010568297052038381,
      "loss": 1.69,
      "step": 166876
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.2127043008804321,
      "learning_rate": 0.00010567985403913645,
      "loss": 1.8839,
      "step": 166877
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.2562241554260254,
      "learning_rate": 0.00010567673759401674,
      "loss": 2.1152,
      "step": 166878
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.2870919704437256,
      "learning_rate": 0.00010567362118502551,
      "loss": 2.056,
      "step": 166879
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.5318751335144043,
      "learning_rate": 0.00010567050481216322,
      "loss": 2.0972,
      "step": 166880
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.1635698080062866,
      "learning_rate": 0.00010566738847543035,
      "loss": 1.8273,
      "step": 166881
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.2314631938934326,
      "learning_rate": 0.00010566427217482772,
      "loss": 2.0971,
      "step": 166882
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.2620279788970947,
      "learning_rate": 0.00010566115591035572,
      "loss": 1.9922,
      "step": 166883
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.2691880464553833,
      "learning_rate": 0.0001056580396820149,
      "loss": 1.9028,
      "step": 166884
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.1329458951950073,
      "learning_rate": 0.000105654923489806,
      "loss": 1.8658,
      "step": 166885
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.2962347269058228,
      "learning_rate": 0.00010565180733372946,
      "loss": 2.0958,
      "step": 166886
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.2059868574142456,
      "learning_rate": 0.00010564869121378597,
      "loss": 2.0888,
      "step": 166887
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.2877423763275146,
      "learning_rate": 0.0001056455751299761,
      "loss": 2.0376,
      "step": 166888
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.3514564037322998,
      "learning_rate": 0.00010564245908230024,
      "loss": 2.0328,
      "step": 166889
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.3596165180206299,
      "learning_rate": 0.00010563934307075924,
      "loss": 1.9691,
      "step": 166890
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.2841492891311646,
      "learning_rate": 0.00010563622709535358,
      "loss": 2.0344,
      "step": 166891
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.214646339416504,
      "learning_rate": 0.00010563311115608367,
      "loss": 2.1858,
      "step": 166892
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.3376491069793701,
      "learning_rate": 0.00010562999525295037,
      "loss": 1.9321,
      "step": 166893
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.3088041543960571,
      "learning_rate": 0.00010562687938595408,
      "loss": 2.0676,
      "step": 166894
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.2162021398544312,
      "learning_rate": 0.00010562376355509536,
      "loss": 2.0963,
      "step": 166895
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.360969066619873,
      "learning_rate": 0.00010562064776037492,
      "loss": 1.9514,
      "step": 166896
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.2149721384048462,
      "learning_rate": 0.00010561753200179332,
      "loss": 1.9359,
      "step": 166897
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.2628127336502075,
      "learning_rate": 0.00010561441627935094,
      "loss": 2.1837,
      "step": 166898
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.1660873889923096,
      "learning_rate": 0.00010561130059304864,
      "loss": 1.9923,
      "step": 166899
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.5136789083480835,
      "learning_rate": 0.00010560818494288685,
      "loss": 2.1821,
      "step": 166900
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.3228203058242798,
      "learning_rate": 0.00010560506932886607,
      "loss": 1.9928,
      "step": 166901
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.1536105871200562,
      "learning_rate": 0.00010560195375098708,
      "loss": 2.1599,
      "step": 166902
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.231080412864685,
      "learning_rate": 0.00010559883820925027,
      "loss": 2.2318,
      "step": 166903
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.4155102968215942,
      "learning_rate": 0.00010559572270365641,
      "loss": 1.8005,
      "step": 166904
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.2191354036331177,
      "learning_rate": 0.00010559260723420595,
      "loss": 1.9726,
      "step": 166905
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.277992844581604,
      "learning_rate": 0.00010558949180089952,
      "loss": 2.1976,
      "step": 166906
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.3783117532730103,
      "learning_rate": 0.00010558637640373754,
      "loss": 2.0026,
      "step": 166907
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.358518362045288,
      "learning_rate": 0.00010558326104272084,
      "loss": 2.23,
      "step": 166908
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.1530516147613525,
      "learning_rate": 0.00010558014571784978,
      "loss": 2.0616,
      "step": 166909
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.257689118385315,
      "learning_rate": 0.00010557703042912516,
      "loss": 1.9289,
      "step": 166910
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.3968636989593506,
      "learning_rate": 0.00010557391517654742,
      "loss": 2.2333,
      "step": 166911
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.23379385471344,
      "learning_rate": 0.00010557079996011713,
      "loss": 2.0999,
      "step": 166912
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.1842392683029175,
      "learning_rate": 0.00010556768477983483,
      "loss": 2.0419,
      "step": 166913
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.3594738245010376,
      "learning_rate": 0.00010556456963570129,
      "loss": 1.9188,
      "step": 166914
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.2718255519866943,
      "learning_rate": 0.00010556145452771683,
      "loss": 2.1001,
      "step": 166915
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.434496521949768,
      "learning_rate": 0.00010555833945588228,
      "loss": 1.9948,
      "step": 166916
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.0964573621749878,
      "learning_rate": 0.00010555522442019811,
      "loss": 1.7858,
      "step": 166917
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.209319829940796,
      "learning_rate": 0.00010555210942066489,
      "loss": 2.0259,
      "step": 166918
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.252572774887085,
      "learning_rate": 0.0001055489944572831,
      "loss": 2.0866,
      "step": 166919
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.3619849681854248,
      "learning_rate": 0.00010554587953005354,
      "loss": 2.1783,
      "step": 166920
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.2642385959625244,
      "learning_rate": 0.00010554276463897654,
      "loss": 2.3897,
      "step": 166921
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.2265222072601318,
      "learning_rate": 0.00010553964978405295,
      "loss": 1.9986,
      "step": 166922
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.2928043603897095,
      "learning_rate": 0.00010553653496528317,
      "loss": 2.0061,
      "step": 166923
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.2966078519821167,
      "learning_rate": 0.00010553342018266784,
      "loss": 2.1877,
      "step": 166924
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.183750867843628,
      "learning_rate": 0.00010553030543620741,
      "loss": 2.2137,
      "step": 166925
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.3131427764892578,
      "learning_rate": 0.00010552719072590268,
      "loss": 1.8397,
      "step": 166926
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.2735804319381714,
      "learning_rate": 0.00010552407605175402,
      "loss": 2.1361,
      "step": 166927
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.2791892290115356,
      "learning_rate": 0.00010552096141376217,
      "loss": 2.043,
      "step": 166928
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.0876576900482178,
      "learning_rate": 0.00010551784681192768,
      "loss": 2.006,
      "step": 166929
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.2581154108047485,
      "learning_rate": 0.0001055147322462511,
      "loss": 2.2446,
      "step": 166930
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.3060704469680786,
      "learning_rate": 0.00010551161771673287,
      "loss": 2.0797,
      "step": 166931
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.3244972229003906,
      "learning_rate": 0.00010550850322337379,
      "loss": 2.1586,
      "step": 166932
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.2179443836212158,
      "learning_rate": 0.00010550538876617428,
      "loss": 2.0982,
      "step": 166933
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.309633731842041,
      "learning_rate": 0.00010550227434513509,
      "loss": 1.966,
      "step": 166934
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.5452380180358887,
      "learning_rate": 0.0001054991599602567,
      "loss": 1.863,
      "step": 166935
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.2328193187713623,
      "learning_rate": 0.00010549604561153969,
      "loss": 2.0049,
      "step": 166936
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.3850806951522827,
      "learning_rate": 0.0001054929312989845,
      "loss": 2.2054,
      "step": 166937
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.2625683546066284,
      "learning_rate": 0.000105489817022592,
      "loss": 2.1628,
      "step": 166938
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.3381434679031372,
      "learning_rate": 0.00010548670278236247,
      "loss": 2.0428,
      "step": 166939
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.255752444267273,
      "learning_rate": 0.00010548358857829676,
      "loss": 1.9145,
      "step": 166940
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.2008634805679321,
      "learning_rate": 0.0001054804744103953,
      "loss": 2.2642,
      "step": 166941
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.26755952835083,
      "learning_rate": 0.00010547736027865872,
      "loss": 2.0767,
      "step": 166942
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.1423156261444092,
      "learning_rate": 0.00010547424618308744,
      "loss": 2.001,
      "step": 166943
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.2011421918869019,
      "learning_rate": 0.0001054711321236823,
      "loss": 2.2077,
      "step": 166944
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.294820785522461,
      "learning_rate": 0.00010546801810044363,
      "loss": 2.1322,
      "step": 166945
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.2823582887649536,
      "learning_rate": 0.00010546490411337223,
      "loss": 2.1439,
      "step": 166946
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.1694883108139038,
      "learning_rate": 0.00010546179016246859,
      "loss": 2.1314,
      "step": 166947
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.1173547506332397,
      "learning_rate": 0.00010545867624773326,
      "loss": 1.9793,
      "step": 166948
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.0859005451202393,
      "learning_rate": 0.00010545556236916674,
      "loss": 1.9042,
      "step": 166949
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.2957916259765625,
      "learning_rate": 0.00010545244852676981,
      "loss": 2.1104,
      "step": 166950
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.3196018934249878,
      "learning_rate": 0.00010544933472054281,
      "loss": 1.8504,
      "step": 166951
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.2336245775222778,
      "learning_rate": 0.00010544622095048658,
      "loss": 2.1006,
      "step": 166952
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.220159888267517,
      "learning_rate": 0.00010544310721660148,
      "loss": 2.1206,
      "step": 166953
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.2803058624267578,
      "learning_rate": 0.00010543999351888835,
      "loss": 1.9621,
      "step": 166954
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.279707431793213,
      "learning_rate": 0.00010543687985734742,
      "loss": 1.8744,
      "step": 166955
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.410938024520874,
      "learning_rate": 0.00010543376623197955,
      "loss": 2.091,
      "step": 166956
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.3879773616790771,
      "learning_rate": 0.00010543065264278508,
      "loss": 1.8765,
      "step": 166957
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.275801420211792,
      "learning_rate": 0.00010542753908976487,
      "loss": 1.9191,
      "step": 166958
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.3469088077545166,
      "learning_rate": 0.00010542442557291923,
      "loss": 1.7979,
      "step": 166959
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.5179955959320068,
      "learning_rate": 0.00010542131209224905,
      "loss": 1.9933,
      "step": 166960
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.3560770750045776,
      "learning_rate": 0.00010541819864775454,
      "loss": 2.1025,
      "step": 166961
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.1965781450271606,
      "learning_rate": 0.00010541508523943654,
      "loss": 1.72,
      "step": 166962
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.2155786752700806,
      "learning_rate": 0.00010541197186729549,
      "loss": 1.8649,
      "step": 166963
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.242930293083191,
      "learning_rate": 0.00010540885853133209,
      "loss": 1.9074,
      "step": 166964
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.266808271408081,
      "learning_rate": 0.00010540574523154677,
      "loss": 2.1377,
      "step": 166965
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.4586410522460938,
      "learning_rate": 0.0001054026319679403,
      "loss": 2.1281,
      "step": 166966
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.376950740814209,
      "learning_rate": 0.00010539951874051315,
      "loss": 1.8607,
      "step": 166967
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.1780343055725098,
      "learning_rate": 0.00010539640554926589,
      "loss": 1.8426,
      "step": 166968
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.3939616680145264,
      "learning_rate": 0.00010539329239419901,
      "loss": 2.1672,
      "step": 166969
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.2867496013641357,
      "learning_rate": 0.00010539017927531334,
      "loss": 1.8567,
      "step": 166970
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.1589241027832031,
      "learning_rate": 0.00010538706619260915,
      "loss": 1.907,
      "step": 166971
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.103023886680603,
      "learning_rate": 0.00010538395314608729,
      "loss": 1.8568,
      "step": 166972
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.1722584962844849,
      "learning_rate": 0.00010538084013574824,
      "loss": 2.0014,
      "step": 166973
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.3604958057403564,
      "learning_rate": 0.00010537772716159257,
      "loss": 2.1056,
      "step": 166974
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.2703192234039307,
      "learning_rate": 0.00010537461422362073,
      "loss": 1.9775,
      "step": 166975
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.2328373193740845,
      "learning_rate": 0.00010537150132183357,
      "loss": 2.0318,
      "step": 166976
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.1979186534881592,
      "learning_rate": 0.00010536838845623137,
      "loss": 2.0112,
      "step": 166977
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.1193411350250244,
      "learning_rate": 0.00010536527562681498,
      "loss": 1.9625,
      "step": 166978
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.2681570053100586,
      "learning_rate": 0.00010536216283358487,
      "loss": 1.9462,
      "step": 166979
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.2974084615707397,
      "learning_rate": 0.00010535905007654148,
      "loss": 2.0957,
      "step": 166980
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.2685492038726807,
      "learning_rate": 0.00010535593735568565,
      "loss": 2.1106,
      "step": 166981
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.1970775127410889,
      "learning_rate": 0.0001053528246710178,
      "loss": 2.0073,
      "step": 166982
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.1816964149475098,
      "learning_rate": 0.00010534971202253847,
      "loss": 1.8664,
      "step": 166983
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.2275044918060303,
      "learning_rate": 0.00010534659941024836,
      "loss": 2.2197,
      "step": 166984
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.4432100057601929,
      "learning_rate": 0.000105343486834148,
      "loss": 2.2083,
      "step": 166985
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.2789857387542725,
      "learning_rate": 0.00010534037429423789,
      "loss": 2.0748,
      "step": 166986
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.4204742908477783,
      "learning_rate": 0.00010533726179051876,
      "loss": 1.8488,
      "step": 166987
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.2736252546310425,
      "learning_rate": 0.00010533414932299098,
      "loss": 1.9866,
      "step": 166988
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.2219178676605225,
      "learning_rate": 0.00010533103689165538,
      "loss": 2.2159,
      "step": 166989
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.3096519708633423,
      "learning_rate": 0.00010532792449651244,
      "loss": 1.9823,
      "step": 166990
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.219376564025879,
      "learning_rate": 0.00010532481213756268,
      "loss": 1.8332,
      "step": 166991
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.2176742553710938,
      "learning_rate": 0.00010532169981480663,
      "loss": 2.1749,
      "step": 166992
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.3052685260772705,
      "learning_rate": 0.00010531858752824502,
      "loss": 2.1209,
      "step": 166993
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.3524322509765625,
      "learning_rate": 0.00010531547527787831,
      "loss": 2.0062,
      "step": 166994
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.242340326309204,
      "learning_rate": 0.00010531236306370719,
      "loss": 2.0949,
      "step": 166995
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.3281790018081665,
      "learning_rate": 0.0001053092508857322,
      "loss": 2.0151,
      "step": 166996
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.1984307765960693,
      "learning_rate": 0.00010530613874395388,
      "loss": 1.8848,
      "step": 166997
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.2305748462677002,
      "learning_rate": 0.0001053030266383727,
      "loss": 2.1114,
      "step": 166998
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.412730097770691,
      "learning_rate": 0.00010529991456898952,
      "loss": 2.0614,
      "step": 166999
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.280324101448059,
      "learning_rate": 0.00010529680253580463,
      "loss": 2.1405,
      "step": 167000
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.1838923692703247,
      "learning_rate": 0.00010529369053881883,
      "loss": 2.0114,
      "step": 167001
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.1569594144821167,
      "learning_rate": 0.00010529057857803264,
      "loss": 1.897,
      "step": 167002
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.312554121017456,
      "learning_rate": 0.00010528746665344658,
      "loss": 2.0963,
      "step": 167003
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.1088188886642456,
      "learning_rate": 0.00010528435476506114,
      "loss": 1.7161,
      "step": 167004
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.4039928913116455,
      "learning_rate": 0.00010528124291287713,
      "loss": 2.0627,
      "step": 167005
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.2551419734954834,
      "learning_rate": 0.00010527813109689494,
      "loss": 2.0248,
      "step": 167006
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.222047209739685,
      "learning_rate": 0.0001052750193171153,
      "loss": 1.941,
      "step": 167007
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.373894214630127,
      "learning_rate": 0.00010527190757353868,
      "loss": 1.964,
      "step": 167008
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.2539697885513306,
      "learning_rate": 0.00010526879586616574,
      "loss": 1.8933,
      "step": 167009
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.241939663887024,
      "learning_rate": 0.00010526568419499688,
      "loss": 2.081,
      "step": 167010
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.2235264778137207,
      "learning_rate": 0.00010526257256003292,
      "loss": 2.2038,
      "step": 167011
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.095887541770935,
      "learning_rate": 0.0001052594609612742,
      "loss": 1.8677,
      "step": 167012
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.3469197750091553,
      "learning_rate": 0.00010525634939872151,
      "loss": 1.9281,
      "step": 167013
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.1811141967773438,
      "learning_rate": 0.00010525323787237527,
      "loss": 2.0058,
      "step": 167014
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.2612767219543457,
      "learning_rate": 0.00010525012638223634,
      "loss": 2.1611,
      "step": 167015
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.48202383518219,
      "learning_rate": 0.00010524701492830487,
      "loss": 2.1776,
      "step": 167016
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.23655366897583,
      "learning_rate": 0.00010524390351058176,
      "loss": 1.9867,
      "step": 167017
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.2162927389144897,
      "learning_rate": 0.00010524079212906739,
      "loss": 1.8741,
      "step": 167018
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.3196098804473877,
      "learning_rate": 0.00010523768078376253,
      "loss": 1.9985,
      "step": 167019
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.217261552810669,
      "learning_rate": 0.00010523456947466754,
      "loss": 1.8346,
      "step": 167020
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.1899199485778809,
      "learning_rate": 0.00010523145820178334,
      "loss": 2.2072,
      "step": 167021
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.166166067123413,
      "learning_rate": 0.00010522834696511009,
      "loss": 1.8854,
      "step": 167022
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.3886845111846924,
      "learning_rate": 0.00010522523576464866,
      "loss": 2.1653,
      "step": 167023
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.2573707103729248,
      "learning_rate": 0.00010522212460039943,
      "loss": 1.9813,
      "step": 167024
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.3117318153381348,
      "learning_rate": 0.00010521901347236321,
      "loss": 2.209,
      "step": 167025
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.3381565809249878,
      "learning_rate": 0.00010521590238054031,
      "loss": 1.9421,
      "step": 167026
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.1606996059417725,
      "learning_rate": 0.00010521279132493168,
      "loss": 2.0426,
      "step": 167027
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.224592685699463,
      "learning_rate": 0.00010520968030553749,
      "loss": 1.9374,
      "step": 167028
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.3074852228164673,
      "learning_rate": 0.00010520656932235855,
      "loss": 2.0213,
      "step": 167029
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.2506498098373413,
      "learning_rate": 0.00010520345837539533,
      "loss": 2.1439,
      "step": 167030
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.2855290174484253,
      "learning_rate": 0.00010520034746464852,
      "loss": 2.1601,
      "step": 167031
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.2627394199371338,
      "learning_rate": 0.00010519723659011857,
      "loss": 2.0544,
      "step": 167032
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.3323829174041748,
      "learning_rate": 0.00010519412575180633,
      "loss": 1.8973,
      "step": 167033
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.2740370035171509,
      "learning_rate": 0.00010519101494971198,
      "loss": 2.0834,
      "step": 167034
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.3309342861175537,
      "learning_rate": 0.00010518790418383639,
      "loss": 2.2328,
      "step": 167035
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.2546032667160034,
      "learning_rate": 0.00010518479345417995,
      "loss": 1.9296,
      "step": 167036
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.3562911748886108,
      "learning_rate": 0.00010518168276074342,
      "loss": 2.1458,
      "step": 167037
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.497357964515686,
      "learning_rate": 0.0001051785721035272,
      "loss": 2.0569,
      "step": 167038
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.2445333003997803,
      "learning_rate": 0.00010517546148253216,
      "loss": 2.0429,
      "step": 167039
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.3328275680541992,
      "learning_rate": 0.0001051723508977585,
      "loss": 1.8985,
      "step": 167040
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.3255854845046997,
      "learning_rate": 0.00010516924034920706,
      "loss": 1.9197,
      "step": 167041
    },
    {
      "epoch": 2.17,
      "grad_norm": 1.2777819633483887,
      "learning_rate": 0.00010516612983687822,
      "loss": 2.2523,
      "step": 167042
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.3250491619110107,
      "learning_rate": 0.0001051630193607728,
      "loss": 2.0012,
      "step": 167043
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.3428093194961548,
      "learning_rate": 0.00010515990892089115,
      "loss": 1.841,
      "step": 167044
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.2524659633636475,
      "learning_rate": 0.00010515679851723414,
      "loss": 2.185,
      "step": 167045
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.3991860151290894,
      "learning_rate": 0.00010515368814980197,
      "loss": 2.0527,
      "step": 167046
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.2157546281814575,
      "learning_rate": 0.00010515057781859551,
      "loss": 2.0203,
      "step": 167047
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.4605363607406616,
      "learning_rate": 0.0001051474675236151,
      "loss": 2.0696,
      "step": 167048
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.1632275581359863,
      "learning_rate": 0.00010514435726486158,
      "loss": 2.0107,
      "step": 167049
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.3081281185150146,
      "learning_rate": 0.00010514124704233533,
      "loss": 2.1155,
      "step": 167050
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.231722354888916,
      "learning_rate": 0.00010513813685603706,
      "loss": 2.0509,
      "step": 167051
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.4256266355514526,
      "learning_rate": 0.0001051350267059673,
      "loss": 1.9659,
      "step": 167052
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.2853591442108154,
      "learning_rate": 0.00010513191659212664,
      "loss": 2.1087,
      "step": 167053
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.2138822078704834,
      "learning_rate": 0.00010512880651451549,
      "loss": 2.0069,
      "step": 167054
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.3192657232284546,
      "learning_rate": 0.0001051256964731347,
      "loss": 2.0052,
      "step": 167055
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.2863019704818726,
      "learning_rate": 0.00010512258646798462,
      "loss": 2.1318,
      "step": 167056
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.3304803371429443,
      "learning_rate": 0.00010511947649906603,
      "loss": 2.0452,
      "step": 167057
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.4672949314117432,
      "learning_rate": 0.00010511636656637942,
      "loss": 2.2629,
      "step": 167058
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.4013676643371582,
      "learning_rate": 0.00010511325666992534,
      "loss": 1.9724,
      "step": 167059
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.3129756450653076,
      "learning_rate": 0.0001051101468097043,
      "loss": 2.0872,
      "step": 167060
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.3544238805770874,
      "learning_rate": 0.00010510703698571704,
      "loss": 1.848,
      "step": 167061
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.7878726720809937,
      "learning_rate": 0.00010510392719796397,
      "loss": 1.9928,
      "step": 167062
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.313874363899231,
      "learning_rate": 0.00010510081744644587,
      "loss": 1.8937,
      "step": 167063
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.3484934568405151,
      "learning_rate": 0.0001050977077311632,
      "loss": 2.0433,
      "step": 167064
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.4040331840515137,
      "learning_rate": 0.00010509459805211644,
      "loss": 2.2319,
      "step": 167065
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.345786213874817,
      "learning_rate": 0.00010509148840930638,
      "loss": 2.0046,
      "step": 167066
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.5658786296844482,
      "learning_rate": 0.00010508837880273348,
      "loss": 2.084,
      "step": 167067
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.4303672313690186,
      "learning_rate": 0.00010508526923239824,
      "loss": 2.0698,
      "step": 167068
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.2261251211166382,
      "learning_rate": 0.00010508215969830144,
      "loss": 1.8607,
      "step": 167069
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.3210433721542358,
      "learning_rate": 0.00010507905020044351,
      "loss": 2.1614,
      "step": 167070
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.4200732707977295,
      "learning_rate": 0.00010507594073882498,
      "loss": 2.0927,
      "step": 167071
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.3170958757400513,
      "learning_rate": 0.00010507283131344664,
      "loss": 1.9757,
      "step": 167072
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.5023574829101562,
      "learning_rate": 0.0001050697219243089,
      "loss": 2.0146,
      "step": 167073
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.4609992504119873,
      "learning_rate": 0.00010506661257141231,
      "loss": 2.0016,
      "step": 167074
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.2675906419754028,
      "learning_rate": 0.0001050635032547576,
      "loss": 2.0109,
      "step": 167075
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.2762067317962646,
      "learning_rate": 0.00010506039397434525,
      "loss": 2.0782,
      "step": 167076
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.2370299100875854,
      "learning_rate": 0.00010505728473017576,
      "loss": 2.142,
      "step": 167077
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.275960922241211,
      "learning_rate": 0.0001050541755222499,
      "loss": 2.1184,
      "step": 167078
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.1580684185028076,
      "learning_rate": 0.00010505106635056809,
      "loss": 1.8558,
      "step": 167079
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.3150748014450073,
      "learning_rate": 0.00010504795721513102,
      "loss": 2.0957,
      "step": 167080
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.1202192306518555,
      "learning_rate": 0.00010504484811593925,
      "loss": 1.9312,
      "step": 167081
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.284946084022522,
      "learning_rate": 0.0001050417390529933,
      "loss": 2.0432,
      "step": 167082
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.2425248622894287,
      "learning_rate": 0.00010503863002629367,
      "loss": 1.9344,
      "step": 167083
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.3344007730484009,
      "learning_rate": 0.00010503552103584111,
      "loss": 2.2062,
      "step": 167084
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.282286524772644,
      "learning_rate": 0.00010503241208163607,
      "loss": 2.1713,
      "step": 167085
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.1345689296722412,
      "learning_rate": 0.00010502930316367927,
      "loss": 2.0969,
      "step": 167086
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.7001392841339111,
      "learning_rate": 0.00010502619428197113,
      "loss": 2.0118,
      "step": 167087
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.37958562374115,
      "learning_rate": 0.00010502308543651247,
      "loss": 2.0805,
      "step": 167088
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.3217436075210571,
      "learning_rate": 0.0001050199766273035,
      "loss": 1.6911,
      "step": 167089
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.3358585834503174,
      "learning_rate": 0.00010501686785434511,
      "loss": 2.2411,
      "step": 167090
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.4833828210830688,
      "learning_rate": 0.00010501375911763765,
      "loss": 2.2241,
      "step": 167091
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.1628764867782593,
      "learning_rate": 0.00010501065041718193,
      "loss": 1.8881,
      "step": 167092
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.2835687398910522,
      "learning_rate": 0.00010500754175297831,
      "loss": 2.069,
      "step": 167093
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.3216865062713623,
      "learning_rate": 0.00010500443312502765,
      "loss": 2.035,
      "step": 167094
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.3456921577453613,
      "learning_rate": 0.00010500132453333016,
      "loss": 2.1147,
      "step": 167095
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.2536282539367676,
      "learning_rate": 0.00010499821597788671,
      "loss": 2.0416,
      "step": 167096
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.3211270570755005,
      "learning_rate": 0.00010499510745869765,
      "loss": 1.9764,
      "step": 167097
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.2728370428085327,
      "learning_rate": 0.0001049919989757638,
      "loss": 1.8623,
      "step": 167098
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.2825123071670532,
      "learning_rate": 0.00010498889052908551,
      "loss": 2.047,
      "step": 167099
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.5828275680541992,
      "learning_rate": 0.00010498578211866366,
      "loss": 2.0358,
      "step": 167100
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.206762433052063,
      "learning_rate": 0.00010498267374449846,
      "loss": 2.0466,
      "step": 167101
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.196315050125122,
      "learning_rate": 0.00010497956540659073,
      "loss": 1.8618,
      "step": 167102
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.3492721319198608,
      "learning_rate": 0.00010497645710494089,
      "loss": 1.9077,
      "step": 167103
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.143349528312683,
      "learning_rate": 0.00010497334883954972,
      "loss": 1.8655,
      "step": 167104
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.348564624786377,
      "learning_rate": 0.00010497024061041758,
      "loss": 2.097,
      "step": 167105
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.2989453077316284,
      "learning_rate": 0.00010496713241754537,
      "loss": 2.2536,
      "step": 167106
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.3505330085754395,
      "learning_rate": 0.00010496402426093325,
      "loss": 2.1419,
      "step": 167107
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.2379560470581055,
      "learning_rate": 0.00010496091614058206,
      "loss": 2.1135,
      "step": 167108
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.3414740562438965,
      "learning_rate": 0.00010495780805649225,
      "loss": 1.9823,
      "step": 167109
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.2867214679718018,
      "learning_rate": 0.00010495470000866454,
      "loss": 2.0848,
      "step": 167110
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.1942064762115479,
      "learning_rate": 0.00010495159199709938,
      "loss": 2.1588,
      "step": 167111
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.5312999486923218,
      "learning_rate": 0.00010494848402179755,
      "loss": 1.8494,
      "step": 167112
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.3114997148513794,
      "learning_rate": 0.00010494537608275929,
      "loss": 2.116,
      "step": 167113
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.250321388244629,
      "learning_rate": 0.00010494226817998549,
      "loss": 2.0271,
      "step": 167114
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.1231180429458618,
      "learning_rate": 0.00010493916031347648,
      "loss": 1.991,
      "step": 167115
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.1566513776779175,
      "learning_rate": 0.00010493605248323309,
      "loss": 1.857,
      "step": 167116
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.3167539834976196,
      "learning_rate": 0.00010493294468925566,
      "loss": 1.9042,
      "step": 167117
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.265298843383789,
      "learning_rate": 0.00010492983693154505,
      "loss": 1.9723,
      "step": 167118
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.3107467889785767,
      "learning_rate": 0.0001049267292101015,
      "loss": 2.0545,
      "step": 167119
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.285941243171692,
      "learning_rate": 0.00010492362152492584,
      "loss": 2.0639,
      "step": 167120
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.2655187845230103,
      "learning_rate": 0.00010492051387601846,
      "loss": 1.9547,
      "step": 167121
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.2049179077148438,
      "learning_rate": 0.00010491740626338014,
      "loss": 1.8131,
      "step": 167122
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.244145393371582,
      "learning_rate": 0.00010491429868701125,
      "loss": 2.0336,
      "step": 167123
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.1605439186096191,
      "learning_rate": 0.0001049111911469127,
      "loss": 2.0686,
      "step": 167124
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.1526157855987549,
      "learning_rate": 0.00010490808364308461,
      "loss": 1.9099,
      "step": 167125
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.287863850593567,
      "learning_rate": 0.0001049049761755279,
      "loss": 2.0305,
      "step": 167126
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.1956480741500854,
      "learning_rate": 0.00010490186874424294,
      "loss": 2.1772,
      "step": 167127
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.2279118299484253,
      "learning_rate": 0.00010489876134923051,
      "loss": 2.2439,
      "step": 167128
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.1810880899429321,
      "learning_rate": 0.00010489565399049096,
      "loss": 1.8795,
      "step": 167129
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.3540027141571045,
      "learning_rate": 0.00010489254666802522,
      "loss": 1.9675,
      "step": 167130
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.187890887260437,
      "learning_rate": 0.00010488943938183344,
      "loss": 2.1024,
      "step": 167131
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.20383620262146,
      "learning_rate": 0.00010488633213191648,
      "loss": 1.9058,
      "step": 167132
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.3021275997161865,
      "learning_rate": 0.00010488322491827472,
      "loss": 2.1353,
      "step": 167133
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.1532596349716187,
      "learning_rate": 0.00010488011774090898,
      "loss": 1.9838,
      "step": 167134
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.2944387197494507,
      "learning_rate": 0.0001048770105998196,
      "loss": 1.9564,
      "step": 167135
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.1862411499023438,
      "learning_rate": 0.00010487390349500746,
      "loss": 1.8521,
      "step": 167136
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.1571720838546753,
      "learning_rate": 0.00010487079642647278,
      "loss": 2.0318,
      "step": 167137
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.2129566669464111,
      "learning_rate": 0.00010486768939421635,
      "loss": 1.8814,
      "step": 167138
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.1871193647384644,
      "learning_rate": 0.00010486458239823862,
      "loss": 1.9902,
      "step": 167139
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.223529577255249,
      "learning_rate": 0.00010486147543854034,
      "loss": 2.1132,
      "step": 167140
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.335025668144226,
      "learning_rate": 0.0001048583685151219,
      "loss": 1.8822,
      "step": 167141
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.236844778060913,
      "learning_rate": 0.00010485526162798411,
      "loss": 1.9352,
      "step": 167142
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.2613368034362793,
      "learning_rate": 0.00010485215477712736,
      "loss": 2.3239,
      "step": 167143
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.278692364692688,
      "learning_rate": 0.00010484904796255229,
      "loss": 2.0059,
      "step": 167144
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.3747756481170654,
      "learning_rate": 0.0001048459411842594,
      "loss": 1.981,
      "step": 167145
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.3338078260421753,
      "learning_rate": 0.00010484283444224938,
      "loss": 1.9185,
      "step": 167146
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.2149094343185425,
      "learning_rate": 0.0001048397277365227,
      "loss": 2.0318,
      "step": 167147
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.1625354290008545,
      "learning_rate": 0.00010483662106708007,
      "loss": 2.1225,
      "step": 167148
    },
    {
      "epoch": 2.18,
      "grad_norm": 2.033911943435669,
      "learning_rate": 0.00010483351443392202,
      "loss": 2.3769,
      "step": 167149
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.2076467275619507,
      "learning_rate": 0.000104830407837049,
      "loss": 1.6695,
      "step": 167150
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.167836308479309,
      "learning_rate": 0.00010482730127646179,
      "loss": 2.0553,
      "step": 167151
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.23300039768219,
      "learning_rate": 0.00010482419475216086,
      "loss": 1.8782,
      "step": 167152
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.2659045457839966,
      "learning_rate": 0.0001048210882641467,
      "loss": 2.0047,
      "step": 167153
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.2211613655090332,
      "learning_rate": 0.00010481798181242008,
      "loss": 2.0134,
      "step": 167154
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.160780906677246,
      "learning_rate": 0.0001048148753969815,
      "loss": 2.1199,
      "step": 167155
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.2651352882385254,
      "learning_rate": 0.00010481176901783143,
      "loss": 1.931,
      "step": 167156
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.2036949396133423,
      "learning_rate": 0.00010480866267497059,
      "loss": 2.0388,
      "step": 167157
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.3137513399124146,
      "learning_rate": 0.00010480555636839952,
      "loss": 2.0993,
      "step": 167158
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.3393659591674805,
      "learning_rate": 0.00010480245009811869,
      "loss": 1.7887,
      "step": 167159
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.4954674243927002,
      "learning_rate": 0.00010479934386412887,
      "loss": 1.9966,
      "step": 167160
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.2500779628753662,
      "learning_rate": 0.00010479623766643055,
      "loss": 1.9314,
      "step": 167161
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.203484058380127,
      "learning_rate": 0.00010479313150502417,
      "loss": 1.8205,
      "step": 167162
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.3506367206573486,
      "learning_rate": 0.00010479002537991055,
      "loss": 1.9896,
      "step": 167163
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.4428596496582031,
      "learning_rate": 0.00010478691929109004,
      "loss": 2.0238,
      "step": 167164
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.292662262916565,
      "learning_rate": 0.00010478381323856342,
      "loss": 1.9193,
      "step": 167165
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.2380305528640747,
      "learning_rate": 0.00010478070722233118,
      "loss": 2.1274,
      "step": 167166
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.2006253004074097,
      "learning_rate": 0.00010477760124239389,
      "loss": 1.7844,
      "step": 167167
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.3404370546340942,
      "learning_rate": 0.00010477449529875204,
      "loss": 1.9053,
      "step": 167168
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.2690354585647583,
      "learning_rate": 0.0001047713893914064,
      "loss": 2.1398,
      "step": 167169
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.2999372482299805,
      "learning_rate": 0.00010476828352035732,
      "loss": 2.0137,
      "step": 167170
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.237182378768921,
      "learning_rate": 0.0001047651776856056,
      "loss": 2.0027,
      "step": 167171
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.1808983087539673,
      "learning_rate": 0.00010476207188715165,
      "loss": 1.9662,
      "step": 167172
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.240836501121521,
      "learning_rate": 0.0001047589661249963,
      "loss": 2.1313,
      "step": 167173
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.4159189462661743,
      "learning_rate": 0.00010475586039913975,
      "loss": 2.1777,
      "step": 167174
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.2259209156036377,
      "learning_rate": 0.00010475275470958285,
      "loss": 1.8847,
      "step": 167175
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.3551499843597412,
      "learning_rate": 0.00010474964905632603,
      "loss": 2.1685,
      "step": 167176
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.3890386819839478,
      "learning_rate": 0.00010474654343936999,
      "loss": 2.1812,
      "step": 167177
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.304517149925232,
      "learning_rate": 0.00010474343785871518,
      "loss": 2.038,
      "step": 167178
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.3417283296585083,
      "learning_rate": 0.00010474033231436247,
      "loss": 1.9496,
      "step": 167179
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.1949864625930786,
      "learning_rate": 0.000104737226806312,
      "loss": 2.1305,
      "step": 167180
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.4338107109069824,
      "learning_rate": 0.00010473412133456467,
      "loss": 2.1309,
      "step": 167181
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.364601492881775,
      "learning_rate": 0.00010473101589912084,
      "loss": 1.8335,
      "step": 167182
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.1231107711791992,
      "learning_rate": 0.00010472791049998131,
      "loss": 2.0869,
      "step": 167183
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.1892261505126953,
      "learning_rate": 0.00010472480513714643,
      "loss": 2.2147,
      "step": 167184
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.186395525932312,
      "learning_rate": 0.0001047216998106171,
      "loss": 2.2071,
      "step": 167185
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.8397886753082275,
      "learning_rate": 0.00010471859452039352,
      "loss": 2.1544,
      "step": 167186
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.3100109100341797,
      "learning_rate": 0.00010471548926647651,
      "loss": 2.0725,
      "step": 167187
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.3010505437850952,
      "learning_rate": 0.0001047123840488665,
      "loss": 2.1011,
      "step": 167188
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.3479336500167847,
      "learning_rate": 0.00010470927886756422,
      "loss": 1.9257,
      "step": 167189
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.124514102935791,
      "learning_rate": 0.00010470617372257009,
      "loss": 2.0489,
      "step": 167190
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.2025541067123413,
      "learning_rate": 0.00010470306861388498,
      "loss": 2.0959,
      "step": 167191
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.3003840446472168,
      "learning_rate": 0.00010469996354150903,
      "loss": 1.8746,
      "step": 167192
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.358174204826355,
      "learning_rate": 0.00010469685850544316,
      "loss": 2.1269,
      "step": 167193
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.2659920454025269,
      "learning_rate": 0.00010469375350568772,
      "loss": 2.0238,
      "step": 167194
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.2367750406265259,
      "learning_rate": 0.0001046906485422435,
      "loss": 2.0199,
      "step": 167195
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.5303908586502075,
      "learning_rate": 0.00010468754361511086,
      "loss": 2.0074,
      "step": 167196
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.160074234008789,
      "learning_rate": 0.0001046844387242907,
      "loss": 2.1621,
      "step": 167197
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.3057551383972168,
      "learning_rate": 0.00010468133386978322,
      "loss": 1.9677,
      "step": 167198
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.254837155342102,
      "learning_rate": 0.00010467822905158924,
      "loss": 2.0108,
      "step": 167199
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.3468018770217896,
      "learning_rate": 0.00010467512426970917,
      "loss": 1.6924,
      "step": 167200
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.3315881490707397,
      "learning_rate": 0.00010467201952414377,
      "loss": 2.1709,
      "step": 167201
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.181807041168213,
      "learning_rate": 0.00010466891481489344,
      "loss": 2.2698,
      "step": 167202
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.258419394493103,
      "learning_rate": 0.00010466581014195904,
      "loss": 1.9827,
      "step": 167203
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.2430402040481567,
      "learning_rate": 0.00010466270550534079,
      "loss": 1.7828,
      "step": 167204
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.2786916494369507,
      "learning_rate": 0.00010465960090503947,
      "loss": 2.1264,
      "step": 167205
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.307063102722168,
      "learning_rate": 0.00010465649634105555,
      "loss": 2.0927,
      "step": 167206
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.1890411376953125,
      "learning_rate": 0.00010465339181338977,
      "loss": 2.0109,
      "step": 167207
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.2859363555908203,
      "learning_rate": 0.00010465028732204251,
      "loss": 2.0906,
      "step": 167208
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.322585105895996,
      "learning_rate": 0.00010464718286701467,
      "loss": 2.2364,
      "step": 167209
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.2237961292266846,
      "learning_rate": 0.00010464407844830637,
      "loss": 1.9609,
      "step": 167210
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.2116705179214478,
      "learning_rate": 0.00010464097406591854,
      "loss": 1.936,
      "step": 167211
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.7427704334259033,
      "learning_rate": 0.00010463786971985154,
      "loss": 2.0222,
      "step": 167212
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.2711471319198608,
      "learning_rate": 0.00010463476541010617,
      "loss": 1.9745,
      "step": 167213
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.4036977291107178,
      "learning_rate": 0.00010463166113668276,
      "loss": 1.9621,
      "step": 167214
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.396634817123413,
      "learning_rate": 0.00010462855689958213,
      "loss": 1.8738,
      "step": 167215
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.3391469717025757,
      "learning_rate": 0.00010462545269880471,
      "loss": 2.0482,
      "step": 167216
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.201756238937378,
      "learning_rate": 0.00010462234853435112,
      "loss": 2.014,
      "step": 167217
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.3235509395599365,
      "learning_rate": 0.00010461924440622185,
      "loss": 2.0533,
      "step": 167218
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.2375620603561401,
      "learning_rate": 0.00010461614031441761,
      "loss": 1.8901,
      "step": 167219
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.5248808860778809,
      "learning_rate": 0.00010461303625893883,
      "loss": 2.2397,
      "step": 167220
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.2825379371643066,
      "learning_rate": 0.00010460993223978629,
      "loss": 1.9493,
      "step": 167221
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.4415943622589111,
      "learning_rate": 0.00010460682825696042,
      "loss": 2.097,
      "step": 167222
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.2905555963516235,
      "learning_rate": 0.00010460372431046186,
      "loss": 1.7035,
      "step": 167223
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.3368020057678223,
      "learning_rate": 0.00010460062040029103,
      "loss": 1.8586,
      "step": 167224
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.196226954460144,
      "learning_rate": 0.00010459751652644876,
      "loss": 1.9182,
      "step": 167225
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.3617647886276245,
      "learning_rate": 0.00010459441268893537,
      "loss": 1.9272,
      "step": 167226
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.2895166873931885,
      "learning_rate": 0.00010459130888775171,
      "loss": 2.176,
      "step": 167227
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.31106698513031,
      "learning_rate": 0.00010458820512289818,
      "loss": 2.0232,
      "step": 167228
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.3937023878097534,
      "learning_rate": 0.00010458510139437543,
      "loss": 1.9843,
      "step": 167229
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.2506436109542847,
      "learning_rate": 0.00010458199770218387,
      "loss": 1.9978,
      "step": 167230
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.6519873142242432,
      "learning_rate": 0.00010457889404632429,
      "loss": 2.0342,
      "step": 167231
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.2161065340042114,
      "learning_rate": 0.00010457579042679712,
      "loss": 1.9618,
      "step": 167232
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.2521755695343018,
      "learning_rate": 0.00010457268684360307,
      "loss": 1.9888,
      "step": 167233
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.1010366678237915,
      "learning_rate": 0.00010456958329674263,
      "loss": 1.8834,
      "step": 167234
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.164913535118103,
      "learning_rate": 0.00010456647978621643,
      "loss": 2.0399,
      "step": 167235
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.3474868535995483,
      "learning_rate": 0.0001045633763120249,
      "loss": 1.963,
      "step": 167236
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.2482469081878662,
      "learning_rate": 0.00010456027287416883,
      "loss": 2.1164,
      "step": 167237
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.1945973634719849,
      "learning_rate": 0.0001045571694726486,
      "loss": 2.2462,
      "step": 167238
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.140431523323059,
      "learning_rate": 0.00010455406610746496,
      "loss": 2.1826,
      "step": 167239
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.5725616216659546,
      "learning_rate": 0.00010455096277861843,
      "loss": 2.1652,
      "step": 167240
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.2719670534133911,
      "learning_rate": 0.00010454785948610943,
      "loss": 1.9731,
      "step": 167241
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.3546634912490845,
      "learning_rate": 0.0001045447562299388,
      "loss": 2.0266,
      "step": 167242
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.1847819089889526,
      "learning_rate": 0.000104541653010107,
      "loss": 2.093,
      "step": 167243
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.3480987548828125,
      "learning_rate": 0.00010453854982661449,
      "loss": 2.0466,
      "step": 167244
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.1145380735397339,
      "learning_rate": 0.00010453544667946206,
      "loss": 2.0069,
      "step": 167245
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.39374577999115,
      "learning_rate": 0.00010453234356865014,
      "loss": 1.8377,
      "step": 167246
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.1946702003479004,
      "learning_rate": 0.00010452924049417927,
      "loss": 2.1728,
      "step": 167247
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.1574606895446777,
      "learning_rate": 0.00010452613745605021,
      "loss": 1.8327,
      "step": 167248
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.3213133811950684,
      "learning_rate": 0.00010452303445426334,
      "loss": 1.9754,
      "step": 167249
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.2403782606124878,
      "learning_rate": 0.00010451993148881941,
      "loss": 2.2428,
      "step": 167250
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.1568996906280518,
      "learning_rate": 0.00010451682855971891,
      "loss": 1.9662,
      "step": 167251
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.2823325395584106,
      "learning_rate": 0.00010451372566696244,
      "loss": 1.8655,
      "step": 167252
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.3122496604919434,
      "learning_rate": 0.00010451062281055047,
      "loss": 2.2102,
      "step": 167253
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.3311941623687744,
      "learning_rate": 0.00010450751999048372,
      "loss": 2.0264,
      "step": 167254
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.2071411609649658,
      "learning_rate": 0.00010450441720676267,
      "loss": 1.8241,
      "step": 167255
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.2005864381790161,
      "learning_rate": 0.00010450131445938801,
      "loss": 2.2727,
      "step": 167256
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.330844759941101,
      "learning_rate": 0.00010449821174836023,
      "loss": 2.2096,
      "step": 167257
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.1572827100753784,
      "learning_rate": 0.00010449510907367993,
      "loss": 1.9665,
      "step": 167258
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.2122339010238647,
      "learning_rate": 0.0001044920064353476,
      "loss": 1.9621,
      "step": 167259
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.5257095098495483,
      "learning_rate": 0.00010448890383336398,
      "loss": 1.8796,
      "step": 167260
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.1293553113937378,
      "learning_rate": 0.00010448580126772948,
      "loss": 2.1678,
      "step": 167261
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.1281994581222534,
      "learning_rate": 0.00010448269873844482,
      "loss": 2.1198,
      "step": 167262
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.1936098337173462,
      "learning_rate": 0.00010447959624551047,
      "loss": 2.0928,
      "step": 167263
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.3305296897888184,
      "learning_rate": 0.0001044764937889272,
      "loss": 2.0335,
      "step": 167264
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.3230431079864502,
      "learning_rate": 0.00010447339136869528,
      "loss": 1.9592,
      "step": 167265
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.1477769613265991,
      "learning_rate": 0.00010447028898481552,
      "loss": 1.7551,
      "step": 167266
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.264404535293579,
      "learning_rate": 0.00010446718663728835,
      "loss": 2.1453,
      "step": 167267
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.282923936843872,
      "learning_rate": 0.00010446408432611453,
      "loss": 1.7557,
      "step": 167268
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.1961034536361694,
      "learning_rate": 0.0001044609820512944,
      "loss": 2.0385,
      "step": 167269
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.2213528156280518,
      "learning_rate": 0.00010445787981282888,
      "loss": 2.2116,
      "step": 167270
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.2228986024856567,
      "learning_rate": 0.00010445477761071813,
      "loss": 1.9187,
      "step": 167271
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.21907377243042,
      "learning_rate": 0.00010445167544496302,
      "loss": 2.1359,
      "step": 167272
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.4303385019302368,
      "learning_rate": 0.00010444857331556393,
      "loss": 1.7825,
      "step": 167273
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.3576042652130127,
      "learning_rate": 0.00010444547122252165,
      "loss": 2.0211,
      "step": 167274
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.2202717065811157,
      "learning_rate": 0.00010444236916583653,
      "loss": 2.1092,
      "step": 167275
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.4479135274887085,
      "learning_rate": 0.00010443926714550948,
      "loss": 1.9205,
      "step": 167276
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.4957304000854492,
      "learning_rate": 0.00010443616516154067,
      "loss": 2.024,
      "step": 167277
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.3855828046798706,
      "learning_rate": 0.00010443306321393095,
      "loss": 2.1964,
      "step": 167278
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.1911197900772095,
      "learning_rate": 0.00010442996130268073,
      "loss": 2.0851,
      "step": 167279
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.2238287925720215,
      "learning_rate": 0.0001044268594277908,
      "loss": 2.1339,
      "step": 167280
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.3129445314407349,
      "learning_rate": 0.00010442375758926147,
      "loss": 1.9106,
      "step": 167281
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.3806672096252441,
      "learning_rate": 0.00010442065578709354,
      "loss": 2.0088,
      "step": 167282
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.230561375617981,
      "learning_rate": 0.00010441755402128755,
      "loss": 1.8831,
      "step": 167283
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.2376124858856201,
      "learning_rate": 0.00010441445229184399,
      "loss": 2.2909,
      "step": 167284
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.3600866794586182,
      "learning_rate": 0.00010441135059876337,
      "loss": 1.919,
      "step": 167285
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.3533515930175781,
      "learning_rate": 0.00010440824894204651,
      "loss": 2.1685,
      "step": 167286
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.3472750186920166,
      "learning_rate": 0.00010440514732169373,
      "loss": 2.0775,
      "step": 167287
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.089118480682373,
      "learning_rate": 0.00010440204573770583,
      "loss": 2.0064,
      "step": 167288
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.1872317790985107,
      "learning_rate": 0.00010439894419008325,
      "loss": 2.0619,
      "step": 167289
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.3284556865692139,
      "learning_rate": 0.00010439584267882664,
      "loss": 1.9547,
      "step": 167290
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.3241522312164307,
      "learning_rate": 0.00010439274120393643,
      "loss": 2.1368,
      "step": 167291
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.322723150253296,
      "learning_rate": 0.0001043896397654134,
      "loss": 2.1375,
      "step": 167292
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.1474498510360718,
      "learning_rate": 0.0001043865383632579,
      "loss": 1.9634,
      "step": 167293
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.3211907148361206,
      "learning_rate": 0.00010438343699747075,
      "loss": 1.9813,
      "step": 167294
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.1879945993423462,
      "learning_rate": 0.00010438033566805244,
      "loss": 2.1365,
      "step": 167295
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.2281246185302734,
      "learning_rate": 0.00010437723437500348,
      "loss": 2.0261,
      "step": 167296
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.3491404056549072,
      "learning_rate": 0.00010437413311832442,
      "loss": 2.1449,
      "step": 167297
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.2326436042785645,
      "learning_rate": 0.00010437103189801596,
      "loss": 2.3517,
      "step": 167298
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.2210614681243896,
      "learning_rate": 0.00010436793071407855,
      "loss": 2.0729,
      "step": 167299
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.2073252201080322,
      "learning_rate": 0.00010436482956651293,
      "loss": 2.4229,
      "step": 167300
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.1731833219528198,
      "learning_rate": 0.00010436172845531956,
      "loss": 1.9345,
      "step": 167301
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.2026431560516357,
      "learning_rate": 0.00010435862738049907,
      "loss": 2.0831,
      "step": 167302
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.1873533725738525,
      "learning_rate": 0.00010435552634205187,
      "loss": 2.107,
      "step": 167303
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.5570793151855469,
      "learning_rate": 0.00010435242533997878,
      "loss": 1.7779,
      "step": 167304
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.296302318572998,
      "learning_rate": 0.00010434932437428019,
      "loss": 1.958,
      "step": 167305
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.207594394683838,
      "learning_rate": 0.00010434622344495684,
      "loss": 1.8946,
      "step": 167306
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.356095552444458,
      "learning_rate": 0.00010434312255200922,
      "loss": 2.0766,
      "step": 167307
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.3113164901733398,
      "learning_rate": 0.00010434002169543791,
      "loss": 2.2413,
      "step": 167308
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.3580195903778076,
      "learning_rate": 0.00010433692087524335,
      "loss": 2.0768,
      "step": 167309
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.3455692529678345,
      "learning_rate": 0.00010433382009142637,
      "loss": 2.2385,
      "step": 167310
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.2611147165298462,
      "learning_rate": 0.00010433071934398737,
      "loss": 1.9688,
      "step": 167311
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.152001142501831,
      "learning_rate": 0.00010432761863292702,
      "loss": 2.3343,
      "step": 167312
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.1734870672225952,
      "learning_rate": 0.00010432451795824591,
      "loss": 1.9112,
      "step": 167313
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.3043488264083862,
      "learning_rate": 0.00010432141731994453,
      "loss": 2.0096,
      "step": 167314
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.2809041738510132,
      "learning_rate": 0.00010431831671802342,
      "loss": 2.1659,
      "step": 167315
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.2294071912765503,
      "learning_rate": 0.0001043152161524833,
      "loss": 2.1181,
      "step": 167316
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.2306121587753296,
      "learning_rate": 0.0001043121156233246,
      "loss": 2.014,
      "step": 167317
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.2927682399749756,
      "learning_rate": 0.00010430901513054807,
      "loss": 1.9711,
      "step": 167318
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.3302170038223267,
      "learning_rate": 0.00010430591467415415,
      "loss": 2.1059,
      "step": 167319
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.300759196281433,
      "learning_rate": 0.00010430281425414349,
      "loss": 1.9013,
      "step": 167320
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.275197982788086,
      "learning_rate": 0.00010429971387051652,
      "loss": 1.9623,
      "step": 167321
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.3701424598693848,
      "learning_rate": 0.00010429661352327405,
      "loss": 2.1338,
      "step": 167322
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.3490256071090698,
      "learning_rate": 0.00010429351321241641,
      "loss": 2.0433,
      "step": 167323
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.2422523498535156,
      "learning_rate": 0.0001042904129379444,
      "loss": 2.0653,
      "step": 167324
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.1694767475128174,
      "learning_rate": 0.00010428731269985852,
      "loss": 2.1189,
      "step": 167325
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.2108798027038574,
      "learning_rate": 0.0001042842124981592,
      "loss": 2.0958,
      "step": 167326
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.3383790254592896,
      "learning_rate": 0.00010428111233284723,
      "loss": 1.9356,
      "step": 167327
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.5343372821807861,
      "learning_rate": 0.00010427801220392313,
      "loss": 2.0062,
      "step": 167328
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.2636518478393555,
      "learning_rate": 0.00010427491211138731,
      "loss": 2.0383,
      "step": 167329
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.146201729774475,
      "learning_rate": 0.00010427181205524059,
      "loss": 1.8812,
      "step": 167330
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.2366735935211182,
      "learning_rate": 0.00010426871203548343,
      "loss": 1.9512,
      "step": 167331
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.3701951503753662,
      "learning_rate": 0.00010426561205211632,
      "loss": 2.11,
      "step": 167332
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.2994974851608276,
      "learning_rate": 0.00010426251210514,
      "loss": 1.9612,
      "step": 167333
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.3575352430343628,
      "learning_rate": 0.00010425941219455492,
      "loss": 2.1173,
      "step": 167334
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.2579426765441895,
      "learning_rate": 0.00010425631232036178,
      "loss": 2.0284,
      "step": 167335
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.2666863203048706,
      "learning_rate": 0.0001042532124825611,
      "loss": 1.9442,
      "step": 167336
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.2532405853271484,
      "learning_rate": 0.00010425011268115344,
      "loss": 1.9508,
      "step": 167337
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.6155571937561035,
      "learning_rate": 0.00010424701291613927,
      "loss": 2.0224,
      "step": 167338
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.274421215057373,
      "learning_rate": 0.00010424391318751939,
      "loss": 2.0686,
      "step": 167339
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.2781853675842285,
      "learning_rate": 0.00010424081349529412,
      "loss": 2.0797,
      "step": 167340
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.229277491569519,
      "learning_rate": 0.00010423771383946432,
      "loss": 2.0776,
      "step": 167341
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.2628623247146606,
      "learning_rate": 0.00010423461422003043,
      "loss": 1.986,
      "step": 167342
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.342524528503418,
      "learning_rate": 0.000104231514636993,
      "loss": 2.2102,
      "step": 167343
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.3599880933761597,
      "learning_rate": 0.00010422841509035254,
      "loss": 1.9602,
      "step": 167344
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.3487063646316528,
      "learning_rate": 0.00010422531558010981,
      "loss": 2.1,
      "step": 167345
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.2803387641906738,
      "learning_rate": 0.00010422221610626517,
      "loss": 2.2014,
      "step": 167346
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.253261923789978,
      "learning_rate": 0.00010421911666881943,
      "loss": 1.9741,
      "step": 167347
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.3926187753677368,
      "learning_rate": 0.00010421601726777295,
      "loss": 2.0914,
      "step": 167348
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.388689637184143,
      "learning_rate": 0.00010421291790312652,
      "loss": 2.1143,
      "step": 167349
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.4856737852096558,
      "learning_rate": 0.00010420981857488061,
      "loss": 2.1497,
      "step": 167350
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.3683799505233765,
      "learning_rate": 0.00010420671928303577,
      "loss": 1.9614,
      "step": 167351
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.3225558996200562,
      "learning_rate": 0.00010420362002759249,
      "loss": 1.9646,
      "step": 167352
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.3749197721481323,
      "learning_rate": 0.00010420052080855158,
      "loss": 1.764,
      "step": 167353
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.3814024925231934,
      "learning_rate": 0.00010419742162591337,
      "loss": 1.9488,
      "step": 167354
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.1817694902420044,
      "learning_rate": 0.00010419432247967866,
      "loss": 2.0818,
      "step": 167355
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.3788057565689087,
      "learning_rate": 0.00010419122336984793,
      "loss": 2.2609,
      "step": 167356
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.1834298372268677,
      "learning_rate": 0.00010418812429642172,
      "loss": 2.115,
      "step": 167357
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.5768847465515137,
      "learning_rate": 0.00010418502525940058,
      "loss": 2.3623,
      "step": 167358
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.3673498630523682,
      "learning_rate": 0.00010418192625878519,
      "loss": 2.0522,
      "step": 167359
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.3345321416854858,
      "learning_rate": 0.000104178827294576,
      "loss": 2.126,
      "step": 167360
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.228326439857483,
      "learning_rate": 0.00010417572836677377,
      "loss": 1.9441,
      "step": 167361
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.3782167434692383,
      "learning_rate": 0.00010417262947537897,
      "loss": 2.1195,
      "step": 167362
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.1751865148544312,
      "learning_rate": 0.00010416953062039216,
      "loss": 1.9736,
      "step": 167363
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.265571117401123,
      "learning_rate": 0.00010416643180181382,
      "loss": 2.1315,
      "step": 167364
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.2727246284484863,
      "learning_rate": 0.00010416333301964475,
      "loss": 2.0641,
      "step": 167365
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.1931642293930054,
      "learning_rate": 0.00010416023427388533,
      "loss": 2.277,
      "step": 167366
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.2775472402572632,
      "learning_rate": 0.00010415713556453632,
      "loss": 1.9598,
      "step": 167367
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.2827357053756714,
      "learning_rate": 0.00010415403689159818,
      "loss": 2.074,
      "step": 167368
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.6838579177856445,
      "learning_rate": 0.00010415093825507148,
      "loss": 2.2905,
      "step": 167369
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.2119466066360474,
      "learning_rate": 0.00010414783965495676,
      "loss": 1.8195,
      "step": 167370
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.255471110343933,
      "learning_rate": 0.00010414474109125474,
      "loss": 1.9851,
      "step": 167371
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.189778208732605,
      "learning_rate": 0.00010414164256396579,
      "loss": 1.9807,
      "step": 167372
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.1842223405838013,
      "learning_rate": 0.00010413854407309071,
      "loss": 2.127,
      "step": 167373
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.1895983219146729,
      "learning_rate": 0.00010413544561862997,
      "loss": 1.9362,
      "step": 167374
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.1261727809906006,
      "learning_rate": 0.00010413234720058417,
      "loss": 2.0436,
      "step": 167375
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.3060779571533203,
      "learning_rate": 0.00010412924881895374,
      "loss": 2.2159,
      "step": 167376
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.3291274309158325,
      "learning_rate": 0.00010412615047373952,
      "loss": 2.2318,
      "step": 167377
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.328586220741272,
      "learning_rate": 0.0001041230521649418,
      "loss": 2.0351,
      "step": 167378
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.241816520690918,
      "learning_rate": 0.00010411995389256141,
      "loss": 1.8387,
      "step": 167379
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.3751124143600464,
      "learning_rate": 0.00010411685565659887,
      "loss": 2.0096,
      "step": 167380
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.085025429725647,
      "learning_rate": 0.00010411375745705465,
      "loss": 1.9358,
      "step": 167381
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.2117335796356201,
      "learning_rate": 0.0001041106592939293,
      "loss": 2.031,
      "step": 167382
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.226792812347412,
      "learning_rate": 0.00010410756116722355,
      "loss": 2.2272,
      "step": 167383
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.2167837619781494,
      "learning_rate": 0.00010410446307693784,
      "loss": 2.2179,
      "step": 167384
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.3302092552185059,
      "learning_rate": 0.00010410136502307286,
      "loss": 2.083,
      "step": 167385
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.2246917486190796,
      "learning_rate": 0.00010409826700562917,
      "loss": 1.7751,
      "step": 167386
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.2124232053756714,
      "learning_rate": 0.00010409516902460732,
      "loss": 1.9486,
      "step": 167387
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.2620753049850464,
      "learning_rate": 0.00010409207108000774,
      "loss": 2.2946,
      "step": 167388
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.3746927976608276,
      "learning_rate": 0.00010408897317183126,
      "loss": 2.0271,
      "step": 167389
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.792405605316162,
      "learning_rate": 0.00010408587530007823,
      "loss": 2.0644,
      "step": 167390
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.2145118713378906,
      "learning_rate": 0.00010408277746474948,
      "loss": 1.8994,
      "step": 167391
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.2538236379623413,
      "learning_rate": 0.00010407967966584538,
      "loss": 1.8612,
      "step": 167392
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.3055297136306763,
      "learning_rate": 0.00010407658190336658,
      "loss": 2.0097,
      "step": 167393
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.2070422172546387,
      "learning_rate": 0.00010407348417731358,
      "loss": 1.962,
      "step": 167394
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.487158179283142,
      "learning_rate": 0.00010407038648768708,
      "loss": 1.9821,
      "step": 167395
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.3375028371810913,
      "learning_rate": 0.0001040672888344875,
      "loss": 2.0964,
      "step": 167396
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.2331386804580688,
      "learning_rate": 0.00010406419121771561,
      "loss": 2.0922,
      "step": 167397
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.1896252632141113,
      "learning_rate": 0.00010406109363737188,
      "loss": 2.0952,
      "step": 167398
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.277394413948059,
      "learning_rate": 0.00010405799609345691,
      "loss": 1.8613,
      "step": 167399
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.2824991941452026,
      "learning_rate": 0.00010405489858597115,
      "loss": 2.1985,
      "step": 167400
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.3613907098770142,
      "learning_rate": 0.00010405180111491537,
      "loss": 2.3204,
      "step": 167401
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.2400611639022827,
      "learning_rate": 0.00010404870368028999,
      "loss": 2.0368,
      "step": 167402
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.2572864294052124,
      "learning_rate": 0.00010404560628209572,
      "loss": 1.8263,
      "step": 167403
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.2220426797866821,
      "learning_rate": 0.00010404250892033309,
      "loss": 1.9984,
      "step": 167404
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.6882836818695068,
      "learning_rate": 0.00010403941159500263,
      "loss": 2.0721,
      "step": 167405
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.3849958181381226,
      "learning_rate": 0.00010403631430610487,
      "loss": 2.0589,
      "step": 167406
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.3517073392868042,
      "learning_rate": 0.00010403321705364054,
      "loss": 2.0669,
      "step": 167407
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.4793955087661743,
      "learning_rate": 0.00010403011983761006,
      "loss": 2.1445,
      "step": 167408
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.254392147064209,
      "learning_rate": 0.00010402702265801418,
      "loss": 2.0471,
      "step": 167409
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.3185964822769165,
      "learning_rate": 0.00010402392551485333,
      "loss": 1.8994,
      "step": 167410
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.2532002925872803,
      "learning_rate": 0.00010402082840812809,
      "loss": 2.0412,
      "step": 167411
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.413695216178894,
      "learning_rate": 0.00010401773133783913,
      "loss": 2.1156,
      "step": 167412
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.2484606504440308,
      "learning_rate": 0.00010401463430398699,
      "loss": 2.3631,
      "step": 167413
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.2822778224945068,
      "learning_rate": 0.00010401153730657214,
      "loss": 1.9678,
      "step": 167414
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.44454026222229,
      "learning_rate": 0.00010400844034559534,
      "loss": 2.1327,
      "step": 167415
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.5224508047103882,
      "learning_rate": 0.00010400534342105695,
      "loss": 2.1157,
      "step": 167416
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.119551658630371,
      "learning_rate": 0.00010400224653295778,
      "loss": 2.0606,
      "step": 167417
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.2876297235488892,
      "learning_rate": 0.00010399914968129831,
      "loss": 2.1805,
      "step": 167418
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.2466959953308105,
      "learning_rate": 0.00010399605286607908,
      "loss": 2.1117,
      "step": 167419
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.2001312971115112,
      "learning_rate": 0.0001039929560873006,
      "loss": 2.1878,
      "step": 167420
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.1913822889328003,
      "learning_rate": 0.00010398985934496361,
      "loss": 2.0427,
      "step": 167421
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.228426456451416,
      "learning_rate": 0.0001039867626390685,
      "loss": 2.1988,
      "step": 167422
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.2580043077468872,
      "learning_rate": 0.00010398366596961607,
      "loss": 2.2242,
      "step": 167423
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.3157957792282104,
      "learning_rate": 0.00010398056933660675,
      "loss": 1.9631,
      "step": 167424
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.2970343828201294,
      "learning_rate": 0.0001039774727400411,
      "loss": 2.2142,
      "step": 167425
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.4786911010742188,
      "learning_rate": 0.00010397437617991977,
      "loss": 2.1186,
      "step": 167426
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.3108689785003662,
      "learning_rate": 0.00010397127965624334,
      "loss": 1.9749,
      "step": 167427
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.2205626964569092,
      "learning_rate": 0.00010396818316901224,
      "loss": 2.1705,
      "step": 167428
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.1540961265563965,
      "learning_rate": 0.00010396508671822727,
      "loss": 1.9798,
      "step": 167429
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.2776129245758057,
      "learning_rate": 0.00010396199030388886,
      "loss": 2.1101,
      "step": 167430
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.3207414150238037,
      "learning_rate": 0.00010395889392599754,
      "loss": 2.2113,
      "step": 167431
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.1824144124984741,
      "learning_rate": 0.00010395579758455407,
      "loss": 1.8881,
      "step": 167432
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.2110350131988525,
      "learning_rate": 0.00010395270127955881,
      "loss": 1.8425,
      "step": 167433
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.233124852180481,
      "learning_rate": 0.00010394960501101251,
      "loss": 1.9224,
      "step": 167434
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.2824580669403076,
      "learning_rate": 0.00010394650877891574,
      "loss": 1.9678,
      "step": 167435
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.2593175172805786,
      "learning_rate": 0.00010394341258326897,
      "loss": 2.1694,
      "step": 167436
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.470921277999878,
      "learning_rate": 0.00010394031642407275,
      "loss": 2.0733,
      "step": 167437
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.3080064058303833,
      "learning_rate": 0.00010393722030132779,
      "loss": 2.0012,
      "step": 167438
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.3970304727554321,
      "learning_rate": 0.00010393412421503452,
      "loss": 1.9185,
      "step": 167439
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.2745888233184814,
      "learning_rate": 0.00010393102816519368,
      "loss": 2.1117,
      "step": 167440
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.276613473892212,
      "learning_rate": 0.0001039279321518058,
      "loss": 2.2439,
      "step": 167441
    },
    {
      "epoch": 2.18,
      "grad_norm": 2.156160831451416,
      "learning_rate": 0.00010392483617487141,
      "loss": 2.009,
      "step": 167442
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.2246822118759155,
      "learning_rate": 0.00010392174023439096,
      "loss": 2.1193,
      "step": 167443
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.2410974502563477,
      "learning_rate": 0.00010391864433036528,
      "loss": 2.1207,
      "step": 167444
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.2364810705184937,
      "learning_rate": 0.00010391554846279475,
      "loss": 2.0472,
      "step": 167445
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.269963026046753,
      "learning_rate": 0.00010391245263168009,
      "loss": 1.9333,
      "step": 167446
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.5196599960327148,
      "learning_rate": 0.00010390935683702183,
      "loss": 1.9818,
      "step": 167447
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.2480075359344482,
      "learning_rate": 0.00010390626107882051,
      "loss": 1.8963,
      "step": 167448
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.2355586290359497,
      "learning_rate": 0.00010390316535707662,
      "loss": 1.6888,
      "step": 167449
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.230120301246643,
      "learning_rate": 0.00010390006967179091,
      "loss": 2.1333,
      "step": 167450
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.61599600315094,
      "learning_rate": 0.0001038969740229638,
      "loss": 2.1842,
      "step": 167451
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.3942553997039795,
      "learning_rate": 0.00010389387841059605,
      "loss": 2.2995,
      "step": 167452
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.1870369911193848,
      "learning_rate": 0.00010389078283468812,
      "loss": 2.0521,
      "step": 167453
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.4979848861694336,
      "learning_rate": 0.00010388768729524061,
      "loss": 2.0464,
      "step": 167454
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.3610928058624268,
      "learning_rate": 0.00010388459179225396,
      "loss": 1.8652,
      "step": 167455
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.1756601333618164,
      "learning_rate": 0.00010388149632572896,
      "loss": 1.9249,
      "step": 167456
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.2519036531448364,
      "learning_rate": 0.00010387840089566603,
      "loss": 1.961,
      "step": 167457
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.2583800554275513,
      "learning_rate": 0.00010387530550206588,
      "loss": 1.9136,
      "step": 167458
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.2758135795593262,
      "learning_rate": 0.00010387221014492902,
      "loss": 2.0046,
      "step": 167459
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.1905313730239868,
      "learning_rate": 0.00010386911482425601,
      "loss": 1.8254,
      "step": 167460
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.4217585325241089,
      "learning_rate": 0.00010386601954004733,
      "loss": 2.1819,
      "step": 167461
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.2602252960205078,
      "learning_rate": 0.00010386292429230379,
      "loss": 2.0392,
      "step": 167462
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.3676197528839111,
      "learning_rate": 0.0001038598290810257,
      "loss": 2.1814,
      "step": 167463
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.1237874031066895,
      "learning_rate": 0.00010385673390621388,
      "loss": 1.9721,
      "step": 167464
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.2565289735794067,
      "learning_rate": 0.0001038536387678688,
      "loss": 2.0922,
      "step": 167465
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.4106690883636475,
      "learning_rate": 0.00010385054366599104,
      "loss": 2.0582,
      "step": 167466
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.2684246301651,
      "learning_rate": 0.00010384744860058104,
      "loss": 1.9972,
      "step": 167467
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.2509865760803223,
      "learning_rate": 0.00010384435357163961,
      "loss": 2.1099,
      "step": 167468
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.1821050643920898,
      "learning_rate": 0.00010384125857916714,
      "loss": 2.0585,
      "step": 167469
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.2191145420074463,
      "learning_rate": 0.00010383816362316435,
      "loss": 2.3105,
      "step": 167470
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.2612628936767578,
      "learning_rate": 0.00010383506870363176,
      "loss": 2.0019,
      "step": 167471
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.0729024410247803,
      "learning_rate": 0.00010383197382056998,
      "loss": 1.8768,
      "step": 167472
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.2462232112884521,
      "learning_rate": 0.00010382887897397937,
      "loss": 2.0365,
      "step": 167473
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.212431788444519,
      "learning_rate": 0.00010382578416386083,
      "loss": 2.1156,
      "step": 167474
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.131056308746338,
      "learning_rate": 0.00010382268939021464,
      "loss": 1.9053,
      "step": 167475
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.266594409942627,
      "learning_rate": 0.00010381959465304161,
      "loss": 2.3385,
      "step": 167476
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.1788485050201416,
      "learning_rate": 0.00010381649995234224,
      "loss": 1.8939,
      "step": 167477
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.2717552185058594,
      "learning_rate": 0.0001038134052881171,
      "loss": 2.1447,
      "step": 167478
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.3859179019927979,
      "learning_rate": 0.00010381031066036665,
      "loss": 2.0894,
      "step": 167479
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.3129023313522339,
      "learning_rate": 0.00010380721606909165,
      "loss": 1.9571,
      "step": 167480
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.2892886400222778,
      "learning_rate": 0.00010380412151429247,
      "loss": 2.0245,
      "step": 167481
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.3229843378067017,
      "learning_rate": 0.00010380102699596994,
      "loss": 2.0523,
      "step": 167482
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.423231840133667,
      "learning_rate": 0.00010379793251412439,
      "loss": 2.2641,
      "step": 167483
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.384442687034607,
      "learning_rate": 0.00010379483806875674,
      "loss": 1.9935,
      "step": 167484
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.238099217414856,
      "learning_rate": 0.00010379174365986708,
      "loss": 2.0067,
      "step": 167485
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.3132613897323608,
      "learning_rate": 0.00010378864928745638,
      "loss": 2.2066,
      "step": 167486
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.2421963214874268,
      "learning_rate": 0.00010378555495152496,
      "loss": 1.9463,
      "step": 167487
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.3868998289108276,
      "learning_rate": 0.00010378246065207359,
      "loss": 2.0008,
      "step": 167488
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.233304738998413,
      "learning_rate": 0.0001037793663891027,
      "loss": 2.0242,
      "step": 167489
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.1009392738342285,
      "learning_rate": 0.0001037762721626131,
      "loss": 1.9059,
      "step": 167490
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.4019025564193726,
      "learning_rate": 0.00010377317797260502,
      "loss": 2.0734,
      "step": 167491
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.2866482734680176,
      "learning_rate": 0.0001037700838190793,
      "loss": 2.043,
      "step": 167492
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.2770642042160034,
      "learning_rate": 0.00010376698970203632,
      "loss": 2.2066,
      "step": 167493
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.2094329595565796,
      "learning_rate": 0.00010376389562147686,
      "loss": 2.1306,
      "step": 167494
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.4433648586273193,
      "learning_rate": 0.0001037608015774013,
      "loss": 2.0955,
      "step": 167495
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.317348837852478,
      "learning_rate": 0.00010375770756981042,
      "loss": 2.1023,
      "step": 167496
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.2732999324798584,
      "learning_rate": 0.00010375461359870467,
      "loss": 1.984,
      "step": 167497
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.1917345523834229,
      "learning_rate": 0.00010375151966408462,
      "loss": 2.1317,
      "step": 167498
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.6109521389007568,
      "learning_rate": 0.0001037484257659508,
      "loss": 2.1873,
      "step": 167499
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.3017072677612305,
      "learning_rate": 0.00010374533190430396,
      "loss": 2.0447,
      "step": 167500
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.263087511062622,
      "learning_rate": 0.00010374223807914446,
      "loss": 2.283,
      "step": 167501
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.2618680000305176,
      "learning_rate": 0.00010373914429047309,
      "loss": 2.1111,
      "step": 167502
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.3406075239181519,
      "learning_rate": 0.00010373605053829028,
      "loss": 1.9883,
      "step": 167503
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.353255033493042,
      "learning_rate": 0.00010373295682259669,
      "loss": 2.1597,
      "step": 167504
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.3458882570266724,
      "learning_rate": 0.0001037298631433927,
      "loss": 1.9176,
      "step": 167505
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.2152818441390991,
      "learning_rate": 0.00010372676950067918,
      "loss": 2.2926,
      "step": 167506
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.2260128259658813,
      "learning_rate": 0.00010372367589445645,
      "loss": 1.9772,
      "step": 167507
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.2686458826065063,
      "learning_rate": 0.00010372058232472527,
      "loss": 2.1785,
      "step": 167508
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.4303268194198608,
      "learning_rate": 0.00010371748879148616,
      "loss": 1.8072,
      "step": 167509
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.2129729986190796,
      "learning_rate": 0.00010371439529473956,
      "loss": 2.1585,
      "step": 167510
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.3324816226959229,
      "learning_rate": 0.00010371130183448628,
      "loss": 2.0946,
      "step": 167511
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.1200042963027954,
      "learning_rate": 0.00010370820841072675,
      "loss": 2.1821,
      "step": 167512
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.3858848810195923,
      "learning_rate": 0.00010370511502346148,
      "loss": 2.0296,
      "step": 167513
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.9440537691116333,
      "learning_rate": 0.00010370202167269122,
      "loss": 2.0481,
      "step": 167514
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.3525125980377197,
      "learning_rate": 0.00010369892835841649,
      "loss": 2.0518,
      "step": 167515
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.202992558479309,
      "learning_rate": 0.0001036958350806377,
      "loss": 2.0088,
      "step": 167516
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.432294249534607,
      "learning_rate": 0.00010369274183935569,
      "loss": 2.1848,
      "step": 167517
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.235365629196167,
      "learning_rate": 0.00010368964863457088,
      "loss": 2.0922,
      "step": 167518
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.2708061933517456,
      "learning_rate": 0.00010368655546628378,
      "loss": 2.0455,
      "step": 167519
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.334869623184204,
      "learning_rate": 0.00010368346233449517,
      "loss": 2.3332,
      "step": 167520
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.2378695011138916,
      "learning_rate": 0.00010368036923920548,
      "loss": 1.8283,
      "step": 167521
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.2485294342041016,
      "learning_rate": 0.00010367727618041525,
      "loss": 1.9659,
      "step": 167522
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.3278990983963013,
      "learning_rate": 0.00010367418315812519,
      "loss": 2.2168,
      "step": 167523
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.2697240114212036,
      "learning_rate": 0.00010367109017233571,
      "loss": 2.2004,
      "step": 167524
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.1713740825653076,
      "learning_rate": 0.00010366799722304758,
      "loss": 2.0567,
      "step": 167525
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.5015015602111816,
      "learning_rate": 0.00010366490431026129,
      "loss": 2.1779,
      "step": 167526
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.230705738067627,
      "learning_rate": 0.00010366181143397742,
      "loss": 2.0611,
      "step": 167527
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.2364823818206787,
      "learning_rate": 0.00010365871859419637,
      "loss": 1.9374,
      "step": 167528
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.172186017036438,
      "learning_rate": 0.00010365562579091901,
      "loss": 2.122,
      "step": 167529
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.300839900970459,
      "learning_rate": 0.00010365253302414567,
      "loss": 2.125,
      "step": 167530
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.2334061861038208,
      "learning_rate": 0.00010364944029387712,
      "loss": 1.9909,
      "step": 167531
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.3715931177139282,
      "learning_rate": 0.00010364634760011387,
      "loss": 2.0735,
      "step": 167532
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.1434623003005981,
      "learning_rate": 0.00010364325494285644,
      "loss": 2.126,
      "step": 167533
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.3763816356658936,
      "learning_rate": 0.00010364016232210532,
      "loss": 2.1739,
      "step": 167534
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.3144725561141968,
      "learning_rate": 0.00010363706973786133,
      "loss": 2.1334,
      "step": 167535
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.2118326425552368,
      "learning_rate": 0.00010363397719012484,
      "loss": 2.0459,
      "step": 167536
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.266062617301941,
      "learning_rate": 0.00010363088467889657,
      "loss": 2.1619,
      "step": 167537
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.1804853677749634,
      "learning_rate": 0.000103627792204177,
      "loss": 2.2467,
      "step": 167538
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.1877458095550537,
      "learning_rate": 0.00010362469976596677,
      "loss": 2.0228,
      "step": 167539
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.2312865257263184,
      "learning_rate": 0.00010362160736426632,
      "loss": 1.9967,
      "step": 167540
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.147417426109314,
      "learning_rate": 0.00010361851499907641,
      "loss": 1.9898,
      "step": 167541
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.197622299194336,
      "learning_rate": 0.00010361542267039738,
      "loss": 1.924,
      "step": 167542
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.3481006622314453,
      "learning_rate": 0.00010361233037823013,
      "loss": 1.9293,
      "step": 167543
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.45406973361969,
      "learning_rate": 0.00010360923812257499,
      "loss": 2.0682,
      "step": 167544
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.3824810981750488,
      "learning_rate": 0.00010360614590343262,
      "loss": 2.2742,
      "step": 167545
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.1860462427139282,
      "learning_rate": 0.0001036030537208035,
      "loss": 1.9279,
      "step": 167546
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.2581028938293457,
      "learning_rate": 0.00010359996157468835,
      "loss": 1.9957,
      "step": 167547
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.3188780546188354,
      "learning_rate": 0.00010359686946508758,
      "loss": 2.2965,
      "step": 167548
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.3518555164337158,
      "learning_rate": 0.00010359377739200196,
      "loss": 1.9547,
      "step": 167549
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.38079035282135,
      "learning_rate": 0.00010359068535543185,
      "loss": 1.8807,
      "step": 167550
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.3794927597045898,
      "learning_rate": 0.00010358759335537815,
      "loss": 2.0554,
      "step": 167551
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.1395812034606934,
      "learning_rate": 0.00010358450139184104,
      "loss": 1.9392,
      "step": 167552
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.4958176612854004,
      "learning_rate": 0.00010358140946482137,
      "loss": 2.1869,
      "step": 167553
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.761798620223999,
      "learning_rate": 0.00010357831757431953,
      "loss": 2.1995,
      "step": 167554
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.6094809770584106,
      "learning_rate": 0.00010357522572033626,
      "loss": 1.9707,
      "step": 167555
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.200628638267517,
      "learning_rate": 0.00010357213390287198,
      "loss": 2.148,
      "step": 167556
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.258697748184204,
      "learning_rate": 0.00010356904212192758,
      "loss": 2.0046,
      "step": 167557
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.2125194072723389,
      "learning_rate": 0.00010356595037750316,
      "loss": 1.9958,
      "step": 167558
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.3053337335586548,
      "learning_rate": 0.00010356285866959965,
      "loss": 2.149,
      "step": 167559
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.3228830099105835,
      "learning_rate": 0.00010355976699821741,
      "loss": 1.6802,
      "step": 167560
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.2162529230117798,
      "learning_rate": 0.00010355667536335722,
      "loss": 1.9151,
      "step": 167561
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.1030070781707764,
      "learning_rate": 0.00010355358376501945,
      "loss": 2.0831,
      "step": 167562
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.2351046800613403,
      "learning_rate": 0.000103550492203205,
      "loss": 2.0435,
      "step": 167563
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.2412902116775513,
      "learning_rate": 0.000103547400677914,
      "loss": 2.1039,
      "step": 167564
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.5508087873458862,
      "learning_rate": 0.00010354430918914736,
      "loss": 1.944,
      "step": 167565
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.4353611469268799,
      "learning_rate": 0.00010354121773690543,
      "loss": 2.0988,
      "step": 167566
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.2469708919525146,
      "learning_rate": 0.00010353812632118898,
      "loss": 1.859,
      "step": 167567
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.3011980056762695,
      "learning_rate": 0.00010353503494199841,
      "loss": 2.0388,
      "step": 167568
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.2963413000106812,
      "learning_rate": 0.00010353194359933458,
      "loss": 2.0054,
      "step": 167569
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.2895725965499878,
      "learning_rate": 0.00010352885229319772,
      "loss": 1.9422,
      "step": 167570
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.1922065019607544,
      "learning_rate": 0.0001035257610235886,
      "loss": 2.0166,
      "step": 167571
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.3060911893844604,
      "learning_rate": 0.00010352266979050769,
      "loss": 2.0667,
      "step": 167572
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.259270429611206,
      "learning_rate": 0.00010351957859395569,
      "loss": 1.8352,
      "step": 167573
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.3069729804992676,
      "learning_rate": 0.00010351648743393301,
      "loss": 1.8445,
      "step": 167574
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.2332254648208618,
      "learning_rate": 0.00010351339631044053,
      "loss": 1.9686,
      "step": 167575
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.3266329765319824,
      "learning_rate": 0.00010351030522347844,
      "loss": 2.2373,
      "step": 167576
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.268106460571289,
      "learning_rate": 0.0001035072141730476,
      "loss": 1.9899,
      "step": 167577
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.2496252059936523,
      "learning_rate": 0.00010350412315914832,
      "loss": 1.9417,
      "step": 167578
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.1938226222991943,
      "learning_rate": 0.0001035010321817815,
      "loss": 1.9541,
      "step": 167579
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.268876552581787,
      "learning_rate": 0.00010349794124094741,
      "loss": 1.8845,
      "step": 167580
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.2475897073745728,
      "learning_rate": 0.00010349485033664694,
      "loss": 2.0345,
      "step": 167581
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.2110662460327148,
      "learning_rate": 0.00010349175946888034,
      "loss": 2.0468,
      "step": 167582
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.4117891788482666,
      "learning_rate": 0.00010348866863764839,
      "loss": 2.0969,
      "step": 167583
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.2863399982452393,
      "learning_rate": 0.00010348557784295154,
      "loss": 1.9862,
      "step": 167584
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.2858021259307861,
      "learning_rate": 0.0001034824870847905,
      "loss": 1.9897,
      "step": 167585
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.2030680179595947,
      "learning_rate": 0.0001034793963631657,
      "loss": 2.054,
      "step": 167586
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.302065372467041,
      "learning_rate": 0.00010347630567807787,
      "loss": 1.8712,
      "step": 167587
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.306540846824646,
      "learning_rate": 0.00010347321502952749,
      "loss": 2.137,
      "step": 167588
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.3207850456237793,
      "learning_rate": 0.00010347012441751517,
      "loss": 1.9102,
      "step": 167589
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.2471747398376465,
      "learning_rate": 0.00010346703384204137,
      "loss": 1.9995,
      "step": 167590
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.2781927585601807,
      "learning_rate": 0.00010346394330310687,
      "loss": 2.2258,
      "step": 167591
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.207139015197754,
      "learning_rate": 0.00010346085280071201,
      "loss": 2.0783,
      "step": 167592
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.1636261940002441,
      "learning_rate": 0.0001034577623348576,
      "loss": 2.1335,
      "step": 167593
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.4067071676254272,
      "learning_rate": 0.00010345467190554409,
      "loss": 2.0102,
      "step": 167594
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.275091528892517,
      "learning_rate": 0.00010345158151277196,
      "loss": 2.1024,
      "step": 167595
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.1768440008163452,
      "learning_rate": 0.00010344849115654201,
      "loss": 2.093,
      "step": 167596
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.3163938522338867,
      "learning_rate": 0.0001034454008368547,
      "loss": 2.0196,
      "step": 167597
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.2730752229690552,
      "learning_rate": 0.00010344231055371048,
      "loss": 2.0566,
      "step": 167598
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.2774909734725952,
      "learning_rate": 0.00010343922030711014,
      "loss": 1.9765,
      "step": 167599
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.1915252208709717,
      "learning_rate": 0.00010343613009705419,
      "loss": 1.9096,
      "step": 167600
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.4023675918579102,
      "learning_rate": 0.00010343303992354305,
      "loss": 2.16,
      "step": 167601
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.6142370700836182,
      "learning_rate": 0.00010342994978657754,
      "loss": 2.2207,
      "step": 167602
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.5006684064865112,
      "learning_rate": 0.00010342685968615809,
      "loss": 2.05,
      "step": 167603
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.3434220552444458,
      "learning_rate": 0.00010342376962228522,
      "loss": 2.1038,
      "step": 167604
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.2980831861495972,
      "learning_rate": 0.00010342067959495964,
      "loss": 1.9673,
      "step": 167605
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.149372935295105,
      "learning_rate": 0.00010341758960418189,
      "loss": 2.1071,
      "step": 167606
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.234894037246704,
      "learning_rate": 0.00010341449964995243,
      "loss": 2.0086,
      "step": 167607
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.565529465675354,
      "learning_rate": 0.00010341140973227201,
      "loss": 1.9864,
      "step": 167608
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.3443007469177246,
      "learning_rate": 0.00010340831985114105,
      "loss": 1.9969,
      "step": 167609
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.1654448509216309,
      "learning_rate": 0.00010340523000656026,
      "loss": 2.0286,
      "step": 167610
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.339596152305603,
      "learning_rate": 0.00010340214019853016,
      "loss": 1.9058,
      "step": 167611
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.343468427658081,
      "learning_rate": 0.00010339905042705132,
      "loss": 2.0375,
      "step": 167612
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.4394267797470093,
      "learning_rate": 0.00010339596069212421,
      "loss": 2.2265,
      "step": 167613
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.3949613571166992,
      "learning_rate": 0.00010339287099374959,
      "loss": 2.0013,
      "step": 167614
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.2846163511276245,
      "learning_rate": 0.00010338978133192785,
      "loss": 2.1879,
      "step": 167615
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.2699439525604248,
      "learning_rate": 0.00010338669170665975,
      "loss": 2.1607,
      "step": 167616
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.2402857542037964,
      "learning_rate": 0.00010338360211794572,
      "loss": 1.9445,
      "step": 167617
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.4153990745544434,
      "learning_rate": 0.00010338051256578654,
      "loss": 2.0538,
      "step": 167618
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.3450531959533691,
      "learning_rate": 0.00010337742305018246,
      "loss": 1.8578,
      "step": 167619
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.324751853942871,
      "learning_rate": 0.0001033743335711343,
      "loss": 2.1301,
      "step": 167620
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.233984112739563,
      "learning_rate": 0.00010337124412864251,
      "loss": 2.2471,
      "step": 167621
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.1584360599517822,
      "learning_rate": 0.0001033681547227078,
      "loss": 2.0104,
      "step": 167622
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.253612995147705,
      "learning_rate": 0.00010336506535333055,
      "loss": 2.3645,
      "step": 167623
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.2827991247177124,
      "learning_rate": 0.00010336197602051165,
      "loss": 2.0678,
      "step": 167624
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.6605730056762695,
      "learning_rate": 0.00010335888672425124,
      "loss": 2.0855,
      "step": 167625
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.2453769445419312,
      "learning_rate": 0.00010335579746455026,
      "loss": 1.8819,
      "step": 167626
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.2825111150741577,
      "learning_rate": 0.00010335270824140906,
      "loss": 2.1323,
      "step": 167627
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.5041754245758057,
      "learning_rate": 0.0001033496190548284,
      "loss": 2.0724,
      "step": 167628
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.2912570238113403,
      "learning_rate": 0.00010334652990480864,
      "loss": 2.0746,
      "step": 167629
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.1684058904647827,
      "learning_rate": 0.00010334344079135068,
      "loss": 1.8059,
      "step": 167630
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.382759928703308,
      "learning_rate": 0.00010334035171445466,
      "loss": 2.1139,
      "step": 167631
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.1746139526367188,
      "learning_rate": 0.0001033372626741215,
      "loss": 2.0847,
      "step": 167632
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.2462100982666016,
      "learning_rate": 0.00010333417367035156,
      "loss": 2.1315,
      "step": 167633
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.2724403142929077,
      "learning_rate": 0.00010333108470314561,
      "loss": 1.9407,
      "step": 167634
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.2966996431350708,
      "learning_rate": 0.00010332799577250404,
      "loss": 1.9695,
      "step": 167635
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.2059587240219116,
      "learning_rate": 0.00010332490687842769,
      "loss": 1.9931,
      "step": 167636
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.3965338468551636,
      "learning_rate": 0.00010332181802091675,
      "loss": 1.931,
      "step": 167637
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.1979544162750244,
      "learning_rate": 0.0001033187291999721,
      "loss": 2.1814,
      "step": 167638
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.220744013786316,
      "learning_rate": 0.00010331564041559409,
      "loss": 1.9846,
      "step": 167639
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.224318027496338,
      "learning_rate": 0.00010331255166778353,
      "loss": 2.0191,
      "step": 167640
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.3754513263702393,
      "learning_rate": 0.0001033094629565408,
      "loss": 1.8788,
      "step": 167641
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.3289709091186523,
      "learning_rate": 0.00010330637428186675,
      "loss": 1.8091,
      "step": 167642
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.202918291091919,
      "learning_rate": 0.00010330328564376153,
      "loss": 1.7833,
      "step": 167643
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.2224478721618652,
      "learning_rate": 0.00010330019704222607,
      "loss": 1.8351,
      "step": 167644
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.3225653171539307,
      "learning_rate": 0.00010329710847726073,
      "loss": 2.0624,
      "step": 167645
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.358145833015442,
      "learning_rate": 0.00010329401994886623,
      "loss": 1.8986,
      "step": 167646
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.3130764961242676,
      "learning_rate": 0.000103290931457043,
      "loss": 2.0182,
      "step": 167647
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.2625868320465088,
      "learning_rate": 0.0001032878430017919,
      "loss": 2.1945,
      "step": 167648
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.2528494596481323,
      "learning_rate": 0.0001032847545831131,
      "loss": 1.9103,
      "step": 167649
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.3508890867233276,
      "learning_rate": 0.00010328166620100748,
      "loss": 1.8728,
      "step": 167650
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.279849648475647,
      "learning_rate": 0.0001032785778554754,
      "loss": 1.9731,
      "step": 167651
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.21781325340271,
      "learning_rate": 0.00010327548954651767,
      "loss": 2.0993,
      "step": 167652
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.2132835388183594,
      "learning_rate": 0.00010327240127413464,
      "loss": 1.985,
      "step": 167653
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.31646728515625,
      "learning_rate": 0.00010326931303832716,
      "loss": 1.9505,
      "step": 167654
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.3482623100280762,
      "learning_rate": 0.00010326622483909546,
      "loss": 2.1445,
      "step": 167655
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.287940263748169,
      "learning_rate": 0.00010326313667644037,
      "loss": 2.102,
      "step": 167656
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.1939597129821777,
      "learning_rate": 0.00010326004855036224,
      "loss": 1.8366,
      "step": 167657
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.3148880004882812,
      "learning_rate": 0.00010325696046086193,
      "loss": 2.1695,
      "step": 167658
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.294821858406067,
      "learning_rate": 0.00010325387240793972,
      "loss": 2.09,
      "step": 167659
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.3057094812393188,
      "learning_rate": 0.00010325078439159656,
      "loss": 1.9924,
      "step": 167660
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.2831499576568604,
      "learning_rate": 0.00010324769641183259,
      "loss": 2.0774,
      "step": 167661
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.2348673343658447,
      "learning_rate": 0.00010324460846864867,
      "loss": 2.126,
      "step": 167662
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.2112122774124146,
      "learning_rate": 0.00010324152056204522,
      "loss": 2.1037,
      "step": 167663
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.2913942337036133,
      "learning_rate": 0.00010323843269202298,
      "loss": 2.0834,
      "step": 167664
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.2804754972457886,
      "learning_rate": 0.00010323534485858231,
      "loss": 1.8804,
      "step": 167665
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.2774477005004883,
      "learning_rate": 0.00010323225706172414,
      "loss": 1.9953,
      "step": 167666
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.1884245872497559,
      "learning_rate": 0.00010322916930144856,
      "loss": 1.9362,
      "step": 167667
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.2378212213516235,
      "learning_rate": 0.0001032260815777565,
      "loss": 2.2071,
      "step": 167668
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.1966302394866943,
      "learning_rate": 0.00010322299389064834,
      "loss": 1.8199,
      "step": 167669
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.3797012567520142,
      "learning_rate": 0.00010321990624012486,
      "loss": 2.0468,
      "step": 167670
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.2235902547836304,
      "learning_rate": 0.00010321681862618639,
      "loss": 2.1292,
      "step": 167671
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.1474627256393433,
      "learning_rate": 0.00010321373104883374,
      "loss": 2.0687,
      "step": 167672
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.195459246635437,
      "learning_rate": 0.00010321064350806734,
      "loss": 2.1089,
      "step": 167673
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.2555092573165894,
      "learning_rate": 0.00010320755600388783,
      "loss": 1.8696,
      "step": 167674
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.3640652894973755,
      "learning_rate": 0.00010320446853629564,
      "loss": 2.1075,
      "step": 167675
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.1827261447906494,
      "learning_rate": 0.00010320138110529155,
      "loss": 1.7516,
      "step": 167676
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.340697169303894,
      "learning_rate": 0.00010319829371087596,
      "loss": 2.027,
      "step": 167677
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.3167269229888916,
      "learning_rate": 0.0001031952063530496,
      "loss": 2.1084,
      "step": 167678
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.1865403652191162,
      "learning_rate": 0.00010319211903181297,
      "loss": 2.0161,
      "step": 167679
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.3805384635925293,
      "learning_rate": 0.00010318903174716664,
      "loss": 2.271,
      "step": 167680
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.2780640125274658,
      "learning_rate": 0.00010318594449911108,
      "loss": 2.1033,
      "step": 167681
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.3457132577896118,
      "learning_rate": 0.00010318285728764707,
      "loss": 2.174,
      "step": 167682
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.2031078338623047,
      "learning_rate": 0.00010317977011277497,
      "loss": 1.8835,
      "step": 167683
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.3226983547210693,
      "learning_rate": 0.00010317668297449558,
      "loss": 2.1937,
      "step": 167684
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.1515722274780273,
      "learning_rate": 0.00010317359587280937,
      "loss": 2.1654,
      "step": 167685
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.3565560579299927,
      "learning_rate": 0.00010317050880771677,
      "loss": 2.0837,
      "step": 167686
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.1856634616851807,
      "learning_rate": 0.00010316742177921862,
      "loss": 2.1647,
      "step": 167687
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.317880630493164,
      "learning_rate": 0.00010316433478731539,
      "loss": 2.0168,
      "step": 167688
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.1719285249710083,
      "learning_rate": 0.00010316124783200748,
      "loss": 1.8386,
      "step": 167689
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.6114344596862793,
      "learning_rate": 0.0001031581609132957,
      "loss": 2.1994,
      "step": 167690
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.2764191627502441,
      "learning_rate": 0.00010315507403118057,
      "loss": 1.9807,
      "step": 167691
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.1610718965530396,
      "learning_rate": 0.00010315198718566251,
      "loss": 1.9592,
      "step": 167692
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.367319941520691,
      "learning_rate": 0.0001031489003767423,
      "loss": 2.1671,
      "step": 167693
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.331384301185608,
      "learning_rate": 0.00010314581360442033,
      "loss": 1.9504,
      "step": 167694
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.3069789409637451,
      "learning_rate": 0.00010314272686869736,
      "loss": 2.0535,
      "step": 167695
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.2445714473724365,
      "learning_rate": 0.0001031396401695739,
      "loss": 2.1176,
      "step": 167696
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.2259950637817383,
      "learning_rate": 0.00010313655350705049,
      "loss": 2.0318,
      "step": 167697
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.2867298126220703,
      "learning_rate": 0.00010313346688112761,
      "loss": 2.1204,
      "step": 167698
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.1925336122512817,
      "learning_rate": 0.00010313038029180603,
      "loss": 1.995,
      "step": 167699
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.2571449279785156,
      "learning_rate": 0.00010312729373908613,
      "loss": 2.1105,
      "step": 167700
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.2113784551620483,
      "learning_rate": 0.00010312420722296871,
      "loss": 2.0212,
      "step": 167701
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.3278580904006958,
      "learning_rate": 0.00010312112074345417,
      "loss": 2.1268,
      "step": 167702
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.1984541416168213,
      "learning_rate": 0.00010311803430054316,
      "loss": 1.9834,
      "step": 167703
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.1409595012664795,
      "learning_rate": 0.00010311494789423609,
      "loss": 1.8655,
      "step": 167704
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.2455558776855469,
      "learning_rate": 0.00010311186152453381,
      "loss": 1.8552,
      "step": 167705
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.1969244480133057,
      "learning_rate": 0.00010310877519143662,
      "loss": 2.0258,
      "step": 167706
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.1984840631484985,
      "learning_rate": 0.00010310568889494534,
      "loss": 2.0057,
      "step": 167707
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.1891666650772095,
      "learning_rate": 0.00010310260263506033,
      "loss": 1.9647,
      "step": 167708
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.6328763961791992,
      "learning_rate": 0.00010309951641178246,
      "loss": 2.15,
      "step": 167709
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.3185653686523438,
      "learning_rate": 0.0001030964302251119,
      "loss": 1.8291,
      "step": 167710
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.2027599811553955,
      "learning_rate": 0.00010309334407504954,
      "loss": 2.0115,
      "step": 167711
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.2711392641067505,
      "learning_rate": 0.00010309025796159575,
      "loss": 2.0217,
      "step": 167712
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.3390289545059204,
      "learning_rate": 0.00010308717188475128,
      "loss": 2.1333,
      "step": 167713
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.2175167798995972,
      "learning_rate": 0.00010308408584451652,
      "loss": 2.0132,
      "step": 167714
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.200494408607483,
      "learning_rate": 0.00010308099984089234,
      "loss": 2.043,
      "step": 167715
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.2608232498168945,
      "learning_rate": 0.00010307791387387895,
      "loss": 2.0107,
      "step": 167716
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.2886388301849365,
      "learning_rate": 0.00010307482794347717,
      "loss": 2.1264,
      "step": 167717
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.2930614948272705,
      "learning_rate": 0.0001030717420496874,
      "loss": 2.1455,
      "step": 167718
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.210675597190857,
      "learning_rate": 0.00010306865619251041,
      "loss": 2.1752,
      "step": 167719
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.2871277332305908,
      "learning_rate": 0.0001030655703719466,
      "loss": 2.1042,
      "step": 167720
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.25806725025177,
      "learning_rate": 0.0001030624845879968,
      "loss": 2.0129,
      "step": 167721
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.2141718864440918,
      "learning_rate": 0.00010305939884066119,
      "loss": 2.0268,
      "step": 167722
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.1152273416519165,
      "learning_rate": 0.00010305631312994069,
      "loss": 1.9504,
      "step": 167723
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.336155891418457,
      "learning_rate": 0.00010305322745583558,
      "loss": 2.0366,
      "step": 167724
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.4470964670181274,
      "learning_rate": 0.00010305014181834674,
      "loss": 2.1352,
      "step": 167725
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.2127968072891235,
      "learning_rate": 0.00010304705621747449,
      "loss": 1.9951,
      "step": 167726
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.2798434495925903,
      "learning_rate": 0.00010304397065321969,
      "loss": 1.8557,
      "step": 167727
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.3286277055740356,
      "learning_rate": 0.00010304088512558256,
      "loss": 1.9549,
      "step": 167728
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.2138407230377197,
      "learning_rate": 0.00010303779963456391,
      "loss": 1.9848,
      "step": 167729
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.214290976524353,
      "learning_rate": 0.00010303471418016418,
      "loss": 1.8833,
      "step": 167730
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.2248868942260742,
      "learning_rate": 0.0001030316287623841,
      "loss": 2.0869,
      "step": 167731
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.3092434406280518,
      "learning_rate": 0.00010302854338122407,
      "loss": 1.9978,
      "step": 167732
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.4182490110397339,
      "learning_rate": 0.00010302545803668495,
      "loss": 2.0413,
      "step": 167733
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.4218707084655762,
      "learning_rate": 0.00010302237272876694,
      "loss": 2.2338,
      "step": 167734
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.2499414682388306,
      "learning_rate": 0.00010301928745747086,
      "loss": 2.0765,
      "step": 167735
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.2056199312210083,
      "learning_rate": 0.00010301620222279712,
      "loss": 1.9611,
      "step": 167736
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.1819688081741333,
      "learning_rate": 0.00010301311702474644,
      "loss": 1.9084,
      "step": 167737
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.4486466646194458,
      "learning_rate": 0.00010301003186331932,
      "loss": 2.102,
      "step": 167738
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.4838457107543945,
      "learning_rate": 0.00010300694673851649,
      "loss": 1.9241,
      "step": 167739
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.275818109512329,
      "learning_rate": 0.00010300386165033824,
      "loss": 2.1314,
      "step": 167740
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.8026622533798218,
      "learning_rate": 0.00010300077659878536,
      "loss": 1.9271,
      "step": 167741
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.3073501586914062,
      "learning_rate": 0.00010299769158385823,
      "loss": 1.9579,
      "step": 167742
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.4682103395462036,
      "learning_rate": 0.00010299460660555768,
      "loss": 2.0047,
      "step": 167743
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.2635456323623657,
      "learning_rate": 0.00010299152166388406,
      "loss": 2.2225,
      "step": 167744
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.1759244203567505,
      "learning_rate": 0.00010298843675883823,
      "loss": 2.0217,
      "step": 167745
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.2651729583740234,
      "learning_rate": 0.00010298535189042034,
      "loss": 1.9353,
      "step": 167746
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.357538104057312,
      "learning_rate": 0.00010298226705863132,
      "loss": 2.1175,
      "step": 167747
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.2436052560806274,
      "learning_rate": 0.00010297918226347152,
      "loss": 2.0624,
      "step": 167748
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.229529857635498,
      "learning_rate": 0.0001029760975049417,
      "loss": 1.9979,
      "step": 167749
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.2534583806991577,
      "learning_rate": 0.00010297301278304222,
      "loss": 1.8759,
      "step": 167750
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.350386142730713,
      "learning_rate": 0.0001029699280977739,
      "loss": 2.08,
      "step": 167751
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.3246163129806519,
      "learning_rate": 0.00010296684344913718,
      "loss": 1.9916,
      "step": 167752
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.2823952436447144,
      "learning_rate": 0.00010296375883713265,
      "loss": 1.8925,
      "step": 167753
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.19684898853302,
      "learning_rate": 0.0001029606742617608,
      "loss": 2.1341,
      "step": 167754
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.2763965129852295,
      "learning_rate": 0.00010295758972302234,
      "loss": 2.0456,
      "step": 167755
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.267569899559021,
      "learning_rate": 0.0001029545052209177,
      "loss": 2.0188,
      "step": 167756
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.1487443447113037,
      "learning_rate": 0.00010295142075544764,
      "loss": 1.9283,
      "step": 167757
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.5256233215332031,
      "learning_rate": 0.00010294833632661263,
      "loss": 2.0853,
      "step": 167758
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.234513521194458,
      "learning_rate": 0.00010294525193441324,
      "loss": 2.0664,
      "step": 167759
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.1773759126663208,
      "learning_rate": 0.00010294216757884996,
      "loss": 2.0321,
      "step": 167760
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.2791908979415894,
      "learning_rate": 0.00010293908325992353,
      "loss": 1.8042,
      "step": 167761
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.2270008325576782,
      "learning_rate": 0.00010293599897763435,
      "loss": 2.0446,
      "step": 167762
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.4779620170593262,
      "learning_rate": 0.00010293291473198319,
      "loss": 1.9147,
      "step": 167763
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.3587334156036377,
      "learning_rate": 0.00010292983052297054,
      "loss": 2.1096,
      "step": 167764
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.3445303440093994,
      "learning_rate": 0.00010292674635059691,
      "loss": 1.9263,
      "step": 167765
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.1940895318984985,
      "learning_rate": 0.00010292366221486285,
      "loss": 2.1013,
      "step": 167766
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.3103359937667847,
      "learning_rate": 0.00010292057811576911,
      "loss": 2.093,
      "step": 167767
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.4412039518356323,
      "learning_rate": 0.00010291749405331602,
      "loss": 2.0602,
      "step": 167768
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.297241449356079,
      "learning_rate": 0.00010291441002750442,
      "loss": 1.8053,
      "step": 167769
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.1552889347076416,
      "learning_rate": 0.00010291132603833472,
      "loss": 1.9333,
      "step": 167770
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.5482832193374634,
      "learning_rate": 0.00010290824208580747,
      "loss": 2.1222,
      "step": 167771
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.5115973949432373,
      "learning_rate": 0.00010290515816992337,
      "loss": 2.0404,
      "step": 167772
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.8838756084442139,
      "learning_rate": 0.0001029020742906829,
      "loss": 2.051,
      "step": 167773
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.2344677448272705,
      "learning_rate": 0.00010289899044808659,
      "loss": 2.0492,
      "step": 167774
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.358410358428955,
      "learning_rate": 0.00010289590664213518,
      "loss": 2.1102,
      "step": 167775
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.2784045934677124,
      "learning_rate": 0.00010289282287282911,
      "loss": 2.0009,
      "step": 167776
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.3447316884994507,
      "learning_rate": 0.00010288973914016891,
      "loss": 1.9151,
      "step": 167777
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.2353475093841553,
      "learning_rate": 0.00010288665544415531,
      "loss": 2.0853,
      "step": 167778
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.2580877542495728,
      "learning_rate": 0.00010288357178478872,
      "loss": 1.984,
      "step": 167779
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.3875962495803833,
      "learning_rate": 0.00010288048816206988,
      "loss": 2.1768,
      "step": 167780
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.3408420085906982,
      "learning_rate": 0.0001028774045759993,
      "loss": 1.9999,
      "step": 167781
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.2000634670257568,
      "learning_rate": 0.00010287432102657753,
      "loss": 2.0681,
      "step": 167782
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.237200140953064,
      "learning_rate": 0.00010287123751380502,
      "loss": 1.9912,
      "step": 167783
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.257803201675415,
      "learning_rate": 0.00010286815403768258,
      "loss": 1.9691,
      "step": 167784
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.3567132949829102,
      "learning_rate": 0.00010286507059821056,
      "loss": 2.0693,
      "step": 167785
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.2440149784088135,
      "learning_rate": 0.00010286198719538976,
      "loss": 1.9557,
      "step": 167786
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.3169516324996948,
      "learning_rate": 0.00010285890382922064,
      "loss": 2.2126,
      "step": 167787
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.2727785110473633,
      "learning_rate": 0.00010285582049970377,
      "loss": 1.86,
      "step": 167788
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.201387643814087,
      "learning_rate": 0.00010285273720683961,
      "loss": 1.9341,
      "step": 167789
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.1006922721862793,
      "learning_rate": 0.00010284965395062896,
      "loss": 1.8823,
      "step": 167790
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.1211704015731812,
      "learning_rate": 0.0001028465707310722,
      "loss": 1.9534,
      "step": 167791
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.3266968727111816,
      "learning_rate": 0.00010284348754817004,
      "loss": 2.1799,
      "step": 167792
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.3653029203414917,
      "learning_rate": 0.00010284040440192294,
      "loss": 2.005,
      "step": 167793
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.2110074758529663,
      "learning_rate": 0.00010283732129233171,
      "loss": 1.9821,
      "step": 167794
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.2775191068649292,
      "learning_rate": 0.00010283423821939653,
      "loss": 2.0119,
      "step": 167795
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.2637906074523926,
      "learning_rate": 0.00010283115518311831,
      "loss": 1.8898,
      "step": 167796
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.3388220071792603,
      "learning_rate": 0.00010282807218349739,
      "loss": 2.0251,
      "step": 167797
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.1864327192306519,
      "learning_rate": 0.00010282498922053454,
      "loss": 2.1637,
      "step": 167798
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.184829831123352,
      "learning_rate": 0.00010282190629423016,
      "loss": 2.2102,
      "step": 167799
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.2687345743179321,
      "learning_rate": 0.00010281882340458513,
      "loss": 2.1239,
      "step": 167800
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.3859556913375854,
      "learning_rate": 0.00010281574055159957,
      "loss": 2.016,
      "step": 167801
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.2358719110488892,
      "learning_rate": 0.00010281265773527441,
      "loss": 2.104,
      "step": 167802
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.2382957935333252,
      "learning_rate": 0.00010280957495560998,
      "loss": 1.8836,
      "step": 167803
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.2972010374069214,
      "learning_rate": 0.00010280649221260708,
      "loss": 2.0736,
      "step": 167804
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.3054381608963013,
      "learning_rate": 0.00010280340950626608,
      "loss": 1.8715,
      "step": 167805
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.1844483613967896,
      "learning_rate": 0.00010280032683658786,
      "loss": 2.0869,
      "step": 167806
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.2013813257217407,
      "learning_rate": 0.00010279724420357254,
      "loss": 2.0057,
      "step": 167807
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.263885259628296,
      "learning_rate": 0.00010279416160722106,
      "loss": 2.0166,
      "step": 167808
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.1840581893920898,
      "learning_rate": 0.00010279107904753379,
      "loss": 1.9696,
      "step": 167809
    },
    {
      "epoch": 2.18,
      "grad_norm": 1.1661711931228638,
      "learning_rate": 0.00010278799652451145,
      "loss": 1.8659,
      "step": 167810
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.264947772026062,
      "learning_rate": 0.00010278491403815445,
      "loss": 1.8555,
      "step": 167811
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.2263777256011963,
      "learning_rate": 0.00010278183158846366,
      "loss": 2.192,
      "step": 167812
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.1922564506530762,
      "learning_rate": 0.00010277874917543927,
      "loss": 2.1451,
      "step": 167813
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.275023341178894,
      "learning_rate": 0.00010277566679908211,
      "loss": 2.1797,
      "step": 167814
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.320407748222351,
      "learning_rate": 0.00010277258445939256,
      "loss": 1.9451,
      "step": 167815
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.2459588050842285,
      "learning_rate": 0.00010276950215637145,
      "loss": 2.0424,
      "step": 167816
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.285272479057312,
      "learning_rate": 0.00010276641989001909,
      "loss": 1.8534,
      "step": 167817
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.3550317287445068,
      "learning_rate": 0.00010276333766033627,
      "loss": 2.1309,
      "step": 167818
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.2753466367721558,
      "learning_rate": 0.0001027602554673235,
      "loss": 1.8896,
      "step": 167819
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.3815131187438965,
      "learning_rate": 0.00010275717331098128,
      "loss": 2.0267,
      "step": 167820
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.3206628561019897,
      "learning_rate": 0.00010275409119131016,
      "loss": 1.8677,
      "step": 167821
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.2157657146453857,
      "learning_rate": 0.00010275100910831085,
      "loss": 2.0066,
      "step": 167822
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.4309383630752563,
      "learning_rate": 0.00010274792706198376,
      "loss": 1.9127,
      "step": 167823
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.4083997011184692,
      "learning_rate": 0.00010274484505232967,
      "loss": 2.1367,
      "step": 167824
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.2182178497314453,
      "learning_rate": 0.00010274176307934901,
      "loss": 2.0525,
      "step": 167825
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.257328987121582,
      "learning_rate": 0.00010273868114304239,
      "loss": 2.1085,
      "step": 167826
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.491361379623413,
      "learning_rate": 0.00010273559924341029,
      "loss": 2.1059,
      "step": 167827
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.718074917793274,
      "learning_rate": 0.00010273251738045343,
      "loss": 1.9252,
      "step": 167828
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.1899418830871582,
      "learning_rate": 0.00010272943555417226,
      "loss": 1.9189,
      "step": 167829
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.2262628078460693,
      "learning_rate": 0.00010272635376456748,
      "loss": 2.0283,
      "step": 167830
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.1794577836990356,
      "learning_rate": 0.0001027232720116396,
      "loss": 2.0704,
      "step": 167831
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.386623740196228,
      "learning_rate": 0.00010272019029538922,
      "loss": 2.1984,
      "step": 167832
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.3208427429199219,
      "learning_rate": 0.00010271710861581675,
      "loss": 2.0882,
      "step": 167833
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.326432228088379,
      "learning_rate": 0.00010271402697292299,
      "loss": 2.2657,
      "step": 167834
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.2190659046173096,
      "learning_rate": 0.00010271094536670832,
      "loss": 2.176,
      "step": 167835
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.188187599182129,
      "learning_rate": 0.00010270786379717351,
      "loss": 1.9899,
      "step": 167836
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.171578049659729,
      "learning_rate": 0.00010270478226431905,
      "loss": 1.9119,
      "step": 167837
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.2564655542373657,
      "learning_rate": 0.00010270170076814549,
      "loss": 2.0697,
      "step": 167838
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.1552602052688599,
      "learning_rate": 0.00010269861930865328,
      "loss": 2.0827,
      "step": 167839
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.2766940593719482,
      "learning_rate": 0.00010269553788584324,
      "loss": 1.9342,
      "step": 167840
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.2573782205581665,
      "learning_rate": 0.00010269245649971575,
      "loss": 1.981,
      "step": 167841
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.2559057474136353,
      "learning_rate": 0.00010268937515027152,
      "loss": 1.9903,
      "step": 167842
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.3351874351501465,
      "learning_rate": 0.00010268629383751107,
      "loss": 2.0319,
      "step": 167843
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.2774373292922974,
      "learning_rate": 0.00010268321256143494,
      "loss": 1.904,
      "step": 167844
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.3784130811691284,
      "learning_rate": 0.00010268013132204368,
      "loss": 2.0948,
      "step": 167845
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.4294220209121704,
      "learning_rate": 0.00010267705011933797,
      "loss": 2.119,
      "step": 167846
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.1513301134109497,
      "learning_rate": 0.00010267396895331821,
      "loss": 2.0199,
      "step": 167847
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.4044665098190308,
      "learning_rate": 0.0001026708878239852,
      "loss": 2.0023,
      "step": 167848
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.2160251140594482,
      "learning_rate": 0.00010266780673133942,
      "loss": 2.0221,
      "step": 167849
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.4992070198059082,
      "learning_rate": 0.00010266472567538139,
      "loss": 1.8851,
      "step": 167850
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.275691270828247,
      "learning_rate": 0.00010266164465611164,
      "loss": 1.8097,
      "step": 167851
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.2873525619506836,
      "learning_rate": 0.00010265856367353088,
      "loss": 2.1004,
      "step": 167852
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.2379124164581299,
      "learning_rate": 0.00010265548272763954,
      "loss": 2.0682,
      "step": 167853
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.1940056085586548,
      "learning_rate": 0.00010265240181843838,
      "loss": 2.0883,
      "step": 167854
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.201668620109558,
      "learning_rate": 0.00010264932094592783,
      "loss": 2.2489,
      "step": 167855
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.4073185920715332,
      "learning_rate": 0.00010264624011010841,
      "loss": 1.9807,
      "step": 167856
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.3629504442214966,
      "learning_rate": 0.00010264315931098091,
      "loss": 2.0285,
      "step": 167857
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.2088477611541748,
      "learning_rate": 0.00010264007854854574,
      "loss": 2.1639,
      "step": 167858
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.3040465116500854,
      "learning_rate": 0.0001026369978228034,
      "loss": 1.9555,
      "step": 167859
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.3691763877868652,
      "learning_rate": 0.00010263391713375467,
      "loss": 2.1808,
      "step": 167860
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.186281442642212,
      "learning_rate": 0.00010263083648140005,
      "loss": 1.9369,
      "step": 167861
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.3139464855194092,
      "learning_rate": 0.00010262775586573996,
      "loss": 2.2497,
      "step": 167862
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.2497684955596924,
      "learning_rate": 0.0001026246752867752,
      "loss": 2.0597,
      "step": 167863
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.2584574222564697,
      "learning_rate": 0.00010262159474450623,
      "loss": 2.0999,
      "step": 167864
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.3825569152832031,
      "learning_rate": 0.00010261851423893354,
      "loss": 1.8901,
      "step": 167865
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.451672077178955,
      "learning_rate": 0.00010261543377005788,
      "loss": 2.107,
      "step": 167866
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.1946982145309448,
      "learning_rate": 0.00010261235333787979,
      "loss": 1.9886,
      "step": 167867
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.2294812202453613,
      "learning_rate": 0.00010260927294239962,
      "loss": 1.9517,
      "step": 167868
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.2841925621032715,
      "learning_rate": 0.00010260619258361824,
      "loss": 2.1,
      "step": 167869
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.332443118095398,
      "learning_rate": 0.000102603112261536,
      "loss": 2.1055,
      "step": 167870
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.2607601881027222,
      "learning_rate": 0.00010260003197615367,
      "loss": 2.0428,
      "step": 167871
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.1563318967819214,
      "learning_rate": 0.00010259695172747173,
      "loss": 1.8618,
      "step": 167872
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.2184187173843384,
      "learning_rate": 0.00010259387151549071,
      "loss": 1.9431,
      "step": 167873
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.2571965456008911,
      "learning_rate": 0.00010259079134021114,
      "loss": 2.0353,
      "step": 167874
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.2140836715698242,
      "learning_rate": 0.00010258771120163374,
      "loss": 1.838,
      "step": 167875
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.2831326723098755,
      "learning_rate": 0.00010258463109975896,
      "loss": 1.9188,
      "step": 167876
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.2382456064224243,
      "learning_rate": 0.00010258155103458748,
      "loss": 2.3284,
      "step": 167877
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.3798669576644897,
      "learning_rate": 0.00010257847100611975,
      "loss": 2.225,
      "step": 167878
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.3672348260879517,
      "learning_rate": 0.00010257539101435658,
      "loss": 2.0747,
      "step": 167879
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.1563115119934082,
      "learning_rate": 0.00010257231105929819,
      "loss": 2.229,
      "step": 167880
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.2459559440612793,
      "learning_rate": 0.00010256923114094543,
      "loss": 2.1174,
      "step": 167881
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.2714345455169678,
      "learning_rate": 0.00010256615125929867,
      "loss": 1.937,
      "step": 167882
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.2030330896377563,
      "learning_rate": 0.0001025630714143587,
      "loss": 2.0382,
      "step": 167883
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.290929913520813,
      "learning_rate": 0.00010255999160612586,
      "loss": 2.1983,
      "step": 167884
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.2253587245941162,
      "learning_rate": 0.00010255691183460095,
      "loss": 1.9013,
      "step": 167885
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.1760815382003784,
      "learning_rate": 0.00010255383209978445,
      "loss": 1.9038,
      "step": 167886
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.1832275390625,
      "learning_rate": 0.00010255075240167692,
      "loss": 2.0683,
      "step": 167887
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.4492493867874146,
      "learning_rate": 0.00010254767274027881,
      "loss": 1.9957,
      "step": 167888
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.3270436525344849,
      "learning_rate": 0.00010254459311559096,
      "loss": 2.2381,
      "step": 167889
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.3383128643035889,
      "learning_rate": 0.00010254151352761365,
      "loss": 2.2225,
      "step": 167890
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.4168798923492432,
      "learning_rate": 0.00010253843397634774,
      "loss": 2.0915,
      "step": 167891
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.273034930229187,
      "learning_rate": 0.00010253535446179364,
      "loss": 2.0373,
      "step": 167892
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.209718942642212,
      "learning_rate": 0.00010253227498395195,
      "loss": 1.9762,
      "step": 167893
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.2320232391357422,
      "learning_rate": 0.00010252919554282311,
      "loss": 1.971,
      "step": 167894
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.391793131828308,
      "learning_rate": 0.00010252611613840796,
      "loss": 1.9007,
      "step": 167895
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.1604633331298828,
      "learning_rate": 0.00010252303677070682,
      "loss": 1.9243,
      "step": 167896
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.5232409238815308,
      "learning_rate": 0.0001025199574397205,
      "loss": 2.0719,
      "step": 167897
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.1398475170135498,
      "learning_rate": 0.00010251687814544939,
      "loss": 1.909,
      "step": 167898
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.3527982234954834,
      "learning_rate": 0.00010251379888789415,
      "loss": 2.1039,
      "step": 167899
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.1641952991485596,
      "learning_rate": 0.00010251071966705524,
      "loss": 1.8205,
      "step": 167900
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.2933555841445923,
      "learning_rate": 0.0001025076404829334,
      "loss": 1.9857,
      "step": 167901
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.2941844463348389,
      "learning_rate": 0.00010250456133552901,
      "loss": 2.1282,
      "step": 167902
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.210243821144104,
      "learning_rate": 0.00010250148222484286,
      "loss": 2.1048,
      "step": 167903
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.1350175142288208,
      "learning_rate": 0.0001024984031508754,
      "loss": 1.874,
      "step": 167904
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.3295716047286987,
      "learning_rate": 0.00010249532411362723,
      "loss": 2.2997,
      "step": 167905
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.3341351747512817,
      "learning_rate": 0.0001024922451130988,
      "loss": 2.1607,
      "step": 167906
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.2262117862701416,
      "learning_rate": 0.00010248916614929089,
      "loss": 1.9615,
      "step": 167907
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.2776732444763184,
      "learning_rate": 0.00010248608722220386,
      "loss": 2.1548,
      "step": 167908
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.2477483749389648,
      "learning_rate": 0.00010248300833183853,
      "loss": 2.0829,
      "step": 167909
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.2694045305252075,
      "learning_rate": 0.0001024799294781953,
      "loss": 2.1103,
      "step": 167910
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.8664462566375732,
      "learning_rate": 0.00010247685066127479,
      "loss": 2.1283,
      "step": 167911
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.218117356300354,
      "learning_rate": 0.00010247377188107747,
      "loss": 2.1182,
      "step": 167912
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.1496632099151611,
      "learning_rate": 0.00010247069313760412,
      "loss": 2.1694,
      "step": 167913
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.3084194660186768,
      "learning_rate": 0.00010246761443085506,
      "loss": 1.8012,
      "step": 167914
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.3998600244522095,
      "learning_rate": 0.00010246453576083113,
      "loss": 1.9026,
      "step": 167915
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.3047915697097778,
      "learning_rate": 0.00010246145712753276,
      "loss": 1.9243,
      "step": 167916
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.2883251905441284,
      "learning_rate": 0.00010245837853096052,
      "loss": 2.1961,
      "step": 167917
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.2341176271438599,
      "learning_rate": 0.00010245529997111492,
      "loss": 1.9528,
      "step": 167918
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.1209721565246582,
      "learning_rate": 0.00010245222144799669,
      "loss": 1.9303,
      "step": 167919
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.2480173110961914,
      "learning_rate": 0.00010244914296160625,
      "loss": 1.9288,
      "step": 167920
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.3370132446289062,
      "learning_rate": 0.00010244606451194433,
      "loss": 2.1749,
      "step": 167921
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.2820600271224976,
      "learning_rate": 0.0001024429860990114,
      "loss": 1.8567,
      "step": 167922
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.3424423933029175,
      "learning_rate": 0.00010243990772280805,
      "loss": 2.2993,
      "step": 167923
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.3208906650543213,
      "learning_rate": 0.00010243682938333475,
      "loss": 2.2353,
      "step": 167924
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.2638472318649292,
      "learning_rate": 0.00010243375108059228,
      "loss": 1.9503,
      "step": 167925
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.2685704231262207,
      "learning_rate": 0.00010243067281458102,
      "loss": 2.1835,
      "step": 167926
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.2735395431518555,
      "learning_rate": 0.0001024275945853017,
      "loss": 2.0742,
      "step": 167927
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.255962610244751,
      "learning_rate": 0.00010242451639275482,
      "loss": 2.185,
      "step": 167928
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.3069301843643188,
      "learning_rate": 0.00010242143823694097,
      "loss": 1.9861,
      "step": 167929
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.3953220844268799,
      "learning_rate": 0.00010241836011786059,
      "loss": 1.9485,
      "step": 167930
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.3058035373687744,
      "learning_rate": 0.00010241528203551448,
      "loss": 2.0816,
      "step": 167931
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.6247222423553467,
      "learning_rate": 0.00010241220398990298,
      "loss": 2.0199,
      "step": 167932
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.1999177932739258,
      "learning_rate": 0.00010240912598102691,
      "loss": 1.8636,
      "step": 167933
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.2619619369506836,
      "learning_rate": 0.00010240604800888671,
      "loss": 1.9977,
      "step": 167934
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.2374799251556396,
      "learning_rate": 0.00010240297007348293,
      "loss": 2.0638,
      "step": 167935
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.3478296995162964,
      "learning_rate": 0.00010239989217481609,
      "loss": 2.1727,
      "step": 167936
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.2836437225341797,
      "learning_rate": 0.00010239681431288694,
      "loss": 2.007,
      "step": 167937
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.2369574308395386,
      "learning_rate": 0.00010239373648769581,
      "loss": 1.8815,
      "step": 167938
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.478940725326538,
      "learning_rate": 0.00010239065869924356,
      "loss": 1.9672,
      "step": 167939
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.2425765991210938,
      "learning_rate": 0.0001023875809475306,
      "loss": 1.9536,
      "step": 167940
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.2978273630142212,
      "learning_rate": 0.00010238450323255742,
      "loss": 1.9668,
      "step": 167941
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.2842109203338623,
      "learning_rate": 0.00010238142555432479,
      "loss": 1.9481,
      "step": 167942
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.363802433013916,
      "learning_rate": 0.0001023783479128332,
      "loss": 2.1105,
      "step": 167943
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.3834812641143799,
      "learning_rate": 0.0001023752703080831,
      "loss": 2.0567,
      "step": 167944
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.3618206977844238,
      "learning_rate": 0.00010237219274007526,
      "loss": 1.9096,
      "step": 167945
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.3343451023101807,
      "learning_rate": 0.00010236911520881017,
      "loss": 1.9136,
      "step": 167946
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.1962577104568481,
      "learning_rate": 0.00010236603771428828,
      "loss": 1.8263,
      "step": 167947
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.411455512046814,
      "learning_rate": 0.00010236296025651037,
      "loss": 1.8368,
      "step": 167948
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.2782323360443115,
      "learning_rate": 0.00010235988283547692,
      "loss": 2.1718,
      "step": 167949
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.5026037693023682,
      "learning_rate": 0.00010235680545118839,
      "loss": 1.9626,
      "step": 167950
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.2102123498916626,
      "learning_rate": 0.00010235372810364557,
      "loss": 2.0199,
      "step": 167951
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.212066411972046,
      "learning_rate": 0.00010235065079284886,
      "loss": 2.2628,
      "step": 167952
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.3198833465576172,
      "learning_rate": 0.00010234757351879894,
      "loss": 2.1464,
      "step": 167953
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.3362535238265991,
      "learning_rate": 0.00010234449628149638,
      "loss": 1.9044,
      "step": 167954
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.2638951539993286,
      "learning_rate": 0.00010234141908094157,
      "loss": 2.2002,
      "step": 167955
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.1106137037277222,
      "learning_rate": 0.00010233834191713533,
      "loss": 2.072,
      "step": 167956
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.239206314086914,
      "learning_rate": 0.00010233526479007814,
      "loss": 2.0007,
      "step": 167957
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.201142430305481,
      "learning_rate": 0.00010233218769977045,
      "loss": 2.1421,
      "step": 167958
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.2670483589172363,
      "learning_rate": 0.00010232911064621303,
      "loss": 2.0194,
      "step": 167959
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.2470329999923706,
      "learning_rate": 0.00010232603362940634,
      "loss": 1.9684,
      "step": 167960
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.4408053159713745,
      "learning_rate": 0.0001023229566493509,
      "loss": 2.1585,
      "step": 167961
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.1837562322616577,
      "learning_rate": 0.00010231987970604745,
      "loss": 2.0643,
      "step": 167962
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.2381746768951416,
      "learning_rate": 0.00010231680279949646,
      "loss": 2.1214,
      "step": 167963
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.2515918016433716,
      "learning_rate": 0.00010231372592969841,
      "loss": 2.2179,
      "step": 167964
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.279667615890503,
      "learning_rate": 0.00010231064909665411,
      "loss": 2.0016,
      "step": 167965
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.2784149646759033,
      "learning_rate": 0.00010230757230036393,
      "loss": 2.0772,
      "step": 167966
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.2968840599060059,
      "learning_rate": 0.00010230449554082845,
      "loss": 1.875,
      "step": 167967
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.2606291770935059,
      "learning_rate": 0.00010230141881804837,
      "loss": 1.9131,
      "step": 167968
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.2982885837554932,
      "learning_rate": 0.00010229834213202408,
      "loss": 2.0035,
      "step": 167969
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.1712255477905273,
      "learning_rate": 0.00010229526548275639,
      "loss": 1.8389,
      "step": 167970
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.3694192171096802,
      "learning_rate": 0.00010229218887024571,
      "loss": 1.9765,
      "step": 167971
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.1856728792190552,
      "learning_rate": 0.00010228911229449264,
      "loss": 2.0089,
      "step": 167972
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.2007014751434326,
      "learning_rate": 0.00010228603575549769,
      "loss": 2.0152,
      "step": 167973
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.3399966955184937,
      "learning_rate": 0.00010228295925326159,
      "loss": 2.0794,
      "step": 167974
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.1871212720870972,
      "learning_rate": 0.00010227988278778471,
      "loss": 2.2543,
      "step": 167975
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.1858989000320435,
      "learning_rate": 0.00010227680635906784,
      "loss": 2.0613,
      "step": 167976
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.214133858680725,
      "learning_rate": 0.00010227372996711143,
      "loss": 2.2327,
      "step": 167977
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.2281235456466675,
      "learning_rate": 0.00010227065361191608,
      "loss": 1.9567,
      "step": 167978
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.2711660861968994,
      "learning_rate": 0.00010226757729348225,
      "loss": 1.9705,
      "step": 167979
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.2367030382156372,
      "learning_rate": 0.00010226450101181071,
      "loss": 2.0972,
      "step": 167980
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.2497895956039429,
      "learning_rate": 0.00010226142476690182,
      "loss": 2.1255,
      "step": 167981
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.2577552795410156,
      "learning_rate": 0.00010225834855875637,
      "loss": 2.1293,
      "step": 167982
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.2631258964538574,
      "learning_rate": 0.00010225527238737481,
      "loss": 1.8596,
      "step": 167983
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.1729106903076172,
      "learning_rate": 0.00010225219625275775,
      "loss": 2.0186,
      "step": 167984
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.3423633575439453,
      "learning_rate": 0.00010224912015490564,
      "loss": 2.1707,
      "step": 167985
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.2574516534805298,
      "learning_rate": 0.00010224604409381926,
      "loss": 1.9216,
      "step": 167986
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.2975207567214966,
      "learning_rate": 0.00010224296806949893,
      "loss": 1.9985,
      "step": 167987
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.2674741744995117,
      "learning_rate": 0.0001022398920819455,
      "loss": 1.956,
      "step": 167988
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.2011362314224243,
      "learning_rate": 0.00010223681613115941,
      "loss": 1.9541,
      "step": 167989
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.3570144176483154,
      "learning_rate": 0.00010223374021714123,
      "loss": 2.2113,
      "step": 167990
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.3965076208114624,
      "learning_rate": 0.00010223066433989138,
      "loss": 2.0617,
      "step": 167991
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.265867829322815,
      "learning_rate": 0.00010222758849941074,
      "loss": 2.1349,
      "step": 167992
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.3104604482650757,
      "learning_rate": 0.00010222451269569962,
      "loss": 2.1404,
      "step": 167993
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.283953309059143,
      "learning_rate": 0.00010222143692875877,
      "loss": 2.0296,
      "step": 167994
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.3033809661865234,
      "learning_rate": 0.00010221836119858869,
      "loss": 1.8115,
      "step": 167995
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.2301758527755737,
      "learning_rate": 0.00010221528550518997,
      "loss": 2.0338,
      "step": 167996
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.3148783445358276,
      "learning_rate": 0.00010221220984856306,
      "loss": 2.0491,
      "step": 167997
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.3482682704925537,
      "learning_rate": 0.00010220913422870874,
      "loss": 2.1933,
      "step": 167998
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.2410889863967896,
      "learning_rate": 0.00010220605864562735,
      "loss": 2.2624,
      "step": 167999
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.315548300743103,
      "learning_rate": 0.00010220298309931969,
      "loss": 2.0643,
      "step": 168000
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.2224438190460205,
      "learning_rate": 0.00010219990758978625,
      "loss": 1.9144,
      "step": 168001
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.3297544717788696,
      "learning_rate": 0.00010219683211702755,
      "loss": 2.0855,
      "step": 168002
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.2697018384933472,
      "learning_rate": 0.0001021937566810441,
      "loss": 2.032,
      "step": 168003
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.3191570043563843,
      "learning_rate": 0.0001021906812818367,
      "loss": 1.8464,
      "step": 168004
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.4836640357971191,
      "learning_rate": 0.00010218760591940566,
      "loss": 2.0121,
      "step": 168005
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.1942262649536133,
      "learning_rate": 0.00010218453059375179,
      "loss": 2.1386,
      "step": 168006
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.1005136966705322,
      "learning_rate": 0.00010218145530487554,
      "loss": 1.9475,
      "step": 168007
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.259685754776001,
      "learning_rate": 0.00010217838005277747,
      "loss": 1.9223,
      "step": 168008
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.1469143629074097,
      "learning_rate": 0.00010217530483745811,
      "loss": 1.7363,
      "step": 168009
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.2570867538452148,
      "learning_rate": 0.0001021722296589182,
      "loss": 1.8756,
      "step": 168010
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.2565397024154663,
      "learning_rate": 0.00010216915451715806,
      "loss": 1.9466,
      "step": 168011
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.2847964763641357,
      "learning_rate": 0.00010216607941217853,
      "loss": 1.8795,
      "step": 168012
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.2437916994094849,
      "learning_rate": 0.00010216300434398005,
      "loss": 1.9813,
      "step": 168013
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.25717031955719,
      "learning_rate": 0.00010215992931256326,
      "loss": 2.0083,
      "step": 168014
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.387294054031372,
      "learning_rate": 0.00010215685431792851,
      "loss": 1.881,
      "step": 168015
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.2877590656280518,
      "learning_rate": 0.00010215377936007666,
      "loss": 1.9168,
      "step": 168016
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.2746940851211548,
      "learning_rate": 0.00010215070443900805,
      "loss": 2.185,
      "step": 168017
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.3260176181793213,
      "learning_rate": 0.00010214762955472347,
      "loss": 1.9756,
      "step": 168018
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.2434664964675903,
      "learning_rate": 0.00010214455470722329,
      "loss": 2.0612,
      "step": 168019
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.174668312072754,
      "learning_rate": 0.00010214147989650834,
      "loss": 1.8206,
      "step": 168020
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.2728840112686157,
      "learning_rate": 0.00010213840512257884,
      "loss": 1.8795,
      "step": 168021
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.252169132232666,
      "learning_rate": 0.00010213533038543564,
      "loss": 2.2416,
      "step": 168022
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.2190954685211182,
      "learning_rate": 0.00010213225568507912,
      "loss": 1.9743,
      "step": 168023
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.3317514657974243,
      "learning_rate": 0.00010212918102151005,
      "loss": 2.192,
      "step": 168024
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.1969329118728638,
      "learning_rate": 0.0001021261063947288,
      "loss": 1.9894,
      "step": 168025
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.2597748041152954,
      "learning_rate": 0.00010212303180473624,
      "loss": 1.9081,
      "step": 168026
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.3826091289520264,
      "learning_rate": 0.00010211995725153257,
      "loss": 1.8559,
      "step": 168027
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.2788846492767334,
      "learning_rate": 0.00010211688273511861,
      "loss": 2.0607,
      "step": 168028
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.5950655937194824,
      "learning_rate": 0.00010211380825549475,
      "loss": 2.2272,
      "step": 168029
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.2963448762893677,
      "learning_rate": 0.00010211073381266177,
      "loss": 1.8416,
      "step": 168030
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.207611083984375,
      "learning_rate": 0.00010210765940662006,
      "loss": 1.9353,
      "step": 168031
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.2312778234481812,
      "learning_rate": 0.00010210458503737035,
      "loss": 2.0715,
      "step": 168032
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.2790907621383667,
      "learning_rate": 0.00010210151070491314,
      "loss": 2.1174,
      "step": 168033
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.3646305799484253,
      "learning_rate": 0.000102098436409249,
      "loss": 2.1115,
      "step": 168034
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.2527310848236084,
      "learning_rate": 0.00010209536215037839,
      "loss": 2.121,
      "step": 168035
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.392409086227417,
      "learning_rate": 0.00010209228792830209,
      "loss": 2.135,
      "step": 168036
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.2839075326919556,
      "learning_rate": 0.00010208921374302048,
      "loss": 2.1885,
      "step": 168037
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.1923120021820068,
      "learning_rate": 0.00010208613959453431,
      "loss": 1.8582,
      "step": 168038
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.2644504308700562,
      "learning_rate": 0.00010208306548284405,
      "loss": 2.0599,
      "step": 168039
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.3716115951538086,
      "learning_rate": 0.0001020799914079502,
      "loss": 2.1156,
      "step": 168040
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.269951581954956,
      "learning_rate": 0.00010207691736985353,
      "loss": 1.952,
      "step": 168041
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.331652045249939,
      "learning_rate": 0.00010207384336855446,
      "loss": 2.1253,
      "step": 168042
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.4504343271255493,
      "learning_rate": 0.00010207076940405351,
      "loss": 1.9591,
      "step": 168043
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.2821619510650635,
      "learning_rate": 0.00010206769547635146,
      "loss": 1.942,
      "step": 168044
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.1744163036346436,
      "learning_rate": 0.00010206462158544875,
      "loss": 2.0094,
      "step": 168045
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.275048851966858,
      "learning_rate": 0.00010206154773134586,
      "loss": 2.0475,
      "step": 168046
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.4066438674926758,
      "learning_rate": 0.00010205847391404357,
      "loss": 1.9907,
      "step": 168047
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.259373664855957,
      "learning_rate": 0.00010205540013354235,
      "loss": 1.9619,
      "step": 168048
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.2440932989120483,
      "learning_rate": 0.00010205232638984268,
      "loss": 2.17,
      "step": 168049
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.2901240587234497,
      "learning_rate": 0.0001020492526829453,
      "loss": 1.904,
      "step": 168050
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.438718557357788,
      "learning_rate": 0.00010204617901285069,
      "loss": 1.9895,
      "step": 168051
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.2271361351013184,
      "learning_rate": 0.00010204310537955934,
      "loss": 1.8329,
      "step": 168052
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.176339030265808,
      "learning_rate": 0.00010204003178307202,
      "loss": 2.1403,
      "step": 168053
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.24777090549469,
      "learning_rate": 0.00010203695822338909,
      "loss": 2.2984,
      "step": 168054
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.300451397895813,
      "learning_rate": 0.00010203388470051135,
      "loss": 2.0761,
      "step": 168055
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.3751899003982544,
      "learning_rate": 0.00010203081121443921,
      "loss": 1.9871,
      "step": 168056
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.1100763082504272,
      "learning_rate": 0.00010202773776517331,
      "loss": 1.9296,
      "step": 168057
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.3304717540740967,
      "learning_rate": 0.00010202466435271405,
      "loss": 2.1033,
      "step": 168058
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.2440837621688843,
      "learning_rate": 0.00010202159097706227,
      "loss": 2.0334,
      "step": 168059
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.2386996746063232,
      "learning_rate": 0.00010201851763821833,
      "loss": 1.7426,
      "step": 168060
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.2999110221862793,
      "learning_rate": 0.00010201544433618296,
      "loss": 2.1693,
      "step": 168061
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.2343155145645142,
      "learning_rate": 0.00010201237107095666,
      "loss": 2.187,
      "step": 168062
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.1927953958511353,
      "learning_rate": 0.00010200929784254001,
      "loss": 2.1232,
      "step": 168063
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.310329794883728,
      "learning_rate": 0.00010200622465093345,
      "loss": 2.205,
      "step": 168064
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.2404277324676514,
      "learning_rate": 0.00010200315149613777,
      "loss": 1.921,
      "step": 168065
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.257171869277954,
      "learning_rate": 0.00010200007837815335,
      "loss": 2.1038,
      "step": 168066
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.2945213317871094,
      "learning_rate": 0.00010199700529698095,
      "loss": 2.225,
      "step": 168067
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.4689525365829468,
      "learning_rate": 0.00010199393225262107,
      "loss": 2.0392,
      "step": 168068
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.2026128768920898,
      "learning_rate": 0.00010199085924507424,
      "loss": 1.7829,
      "step": 168069
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.2709848880767822,
      "learning_rate": 0.00010198778627434094,
      "loss": 1.9892,
      "step": 168070
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.2374558448791504,
      "learning_rate": 0.00010198471334042195,
      "loss": 2.1921,
      "step": 168071
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.2131822109222412,
      "learning_rate": 0.00010198164044331764,
      "loss": 2.089,
      "step": 168072
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.2661068439483643,
      "learning_rate": 0.00010197856758302879,
      "loss": 2.0614,
      "step": 168073
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.220136284828186,
      "learning_rate": 0.00010197549475955585,
      "loss": 2.1083,
      "step": 168074
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.2996594905853271,
      "learning_rate": 0.00010197242197289939,
      "loss": 2.218,
      "step": 168075
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.1636683940887451,
      "learning_rate": 0.00010196934922305992,
      "loss": 2.1863,
      "step": 168076
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.2785941362380981,
      "learning_rate": 0.00010196627651003815,
      "loss": 2.1036,
      "step": 168077
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.3121299743652344,
      "learning_rate": 0.00010196320383383452,
      "loss": 2.0237,
      "step": 168078
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.2668007612228394,
      "learning_rate": 0.00010196013119444976,
      "loss": 1.9125,
      "step": 168079
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.2947100400924683,
      "learning_rate": 0.00010195705859188434,
      "loss": 2.048,
      "step": 168080
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.2293375730514526,
      "learning_rate": 0.00010195398602613888,
      "loss": 1.9008,
      "step": 168081
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.221624493598938,
      "learning_rate": 0.0001019509134972138,
      "loss": 2.1623,
      "step": 168082
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.3934674263000488,
      "learning_rate": 0.00010194784100510985,
      "loss": 2.225,
      "step": 168083
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.2127803564071655,
      "learning_rate": 0.00010194476854982746,
      "loss": 2.0135,
      "step": 168084
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.2696678638458252,
      "learning_rate": 0.00010194169613136735,
      "loss": 1.7876,
      "step": 168085
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.205304741859436,
      "learning_rate": 0.00010193862374972994,
      "loss": 2.0697,
      "step": 168086
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.2588798999786377,
      "learning_rate": 0.00010193555140491608,
      "loss": 2.0019,
      "step": 168087
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.2984097003936768,
      "learning_rate": 0.00010193247909692594,
      "loss": 2.2456,
      "step": 168088
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.3018357753753662,
      "learning_rate": 0.0001019294068257604,
      "loss": 2.2181,
      "step": 168089
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.2576335668563843,
      "learning_rate": 0.0001019263345914198,
      "loss": 2.0026,
      "step": 168090
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.3193097114562988,
      "learning_rate": 0.00010192326239390495,
      "loss": 2.1164,
      "step": 168091
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.2902944087982178,
      "learning_rate": 0.00010192019023321617,
      "loss": 2.0259,
      "step": 168092
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.308154821395874,
      "learning_rate": 0.0001019171181093544,
      "loss": 1.9043,
      "step": 168093
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.3646589517593384,
      "learning_rate": 0.00010191404602231977,
      "loss": 2.0028,
      "step": 168094
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.3228312730789185,
      "learning_rate": 0.00010191097397211316,
      "loss": 1.7397,
      "step": 168095
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.496169090270996,
      "learning_rate": 0.00010190790195873494,
      "loss": 2.0588,
      "step": 168096
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.6203938722610474,
      "learning_rate": 0.00010190482998218586,
      "loss": 2.062,
      "step": 168097
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.370174527168274,
      "learning_rate": 0.00010190175804246636,
      "loss": 1.943,
      "step": 168098
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.2014620304107666,
      "learning_rate": 0.00010189868613957724,
      "loss": 1.8847,
      "step": 168099
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.5496362447738647,
      "learning_rate": 0.00010189561427351867,
      "loss": 2.0804,
      "step": 168100
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.3211923837661743,
      "learning_rate": 0.00010189254244429156,
      "loss": 2.1405,
      "step": 168101
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.446791648864746,
      "learning_rate": 0.00010188947065189624,
      "loss": 2.0768,
      "step": 168102
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.2194181680679321,
      "learning_rate": 0.00010188639889633356,
      "loss": 1.9955,
      "step": 168103
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.310716152191162,
      "learning_rate": 0.00010188332717760379,
      "loss": 2.026,
      "step": 168104
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.2630021572113037,
      "learning_rate": 0.00010188025549570789,
      "loss": 2.2341,
      "step": 168105
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.247260332107544,
      "learning_rate": 0.00010187718385064594,
      "loss": 2.0184,
      "step": 168106
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.5923510789871216,
      "learning_rate": 0.00010187411224241891,
      "loss": 2.0376,
      "step": 168107
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.4859058856964111,
      "learning_rate": 0.00010187104067102712,
      "loss": 2.1828,
      "step": 168108
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.1525006294250488,
      "learning_rate": 0.00010186796913647132,
      "loss": 2.308,
      "step": 168109
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.1772220134735107,
      "learning_rate": 0.00010186489763875192,
      "loss": 1.6716,
      "step": 168110
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.2253202199935913,
      "learning_rate": 0.00010186182617786975,
      "loss": 1.9461,
      "step": 168111
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.228214979171753,
      "learning_rate": 0.00010185875475382502,
      "loss": 2.2749,
      "step": 168112
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.50703763961792,
      "learning_rate": 0.00010185568336661859,
      "loss": 1.9914,
      "step": 168113
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.2833495140075684,
      "learning_rate": 0.00010185261201625086,
      "loss": 2.0618,
      "step": 168114
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.282655119895935,
      "learning_rate": 0.00010184954070272253,
      "loss": 1.9925,
      "step": 168115
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.3841124773025513,
      "learning_rate": 0.00010184646942603401,
      "loss": 2.2228,
      "step": 168116
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.889034628868103,
      "learning_rate": 0.00010184339818618607,
      "loss": 2.1736,
      "step": 168117
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.5528301000595093,
      "learning_rate": 0.0001018403269831792,
      "loss": 2.1599,
      "step": 168118
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.2902578115463257,
      "learning_rate": 0.00010183725581701396,
      "loss": 1.9014,
      "step": 168119
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.2653465270996094,
      "learning_rate": 0.0001018341846876908,
      "loss": 2.105,
      "step": 168120
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.3641084432601929,
      "learning_rate": 0.00010183111359521052,
      "loss": 2.2617,
      "step": 168121
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.613121747970581,
      "learning_rate": 0.00010182804253957345,
      "loss": 1.7902,
      "step": 168122
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.4432666301727295,
      "learning_rate": 0.00010182497152078042,
      "loss": 2.0929,
      "step": 168123
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.3452266454696655,
      "learning_rate": 0.00010182190053883185,
      "loss": 2.1121,
      "step": 168124
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.2466822862625122,
      "learning_rate": 0.00010181882959372834,
      "loss": 1.9393,
      "step": 168125
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.5451140403747559,
      "learning_rate": 0.00010181575868547035,
      "loss": 2.1127,
      "step": 168126
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.394511103630066,
      "learning_rate": 0.00010181268781405862,
      "loss": 2.0022,
      "step": 168127
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.588139533996582,
      "learning_rate": 0.00010180961697949359,
      "loss": 1.8699,
      "step": 168128
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.1929534673690796,
      "learning_rate": 0.00010180654618177599,
      "loss": 2.2198,
      "step": 168129
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.2373062372207642,
      "learning_rate": 0.00010180347542090627,
      "loss": 1.8231,
      "step": 168130
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.2609018087387085,
      "learning_rate": 0.00010180040469688494,
      "loss": 1.9297,
      "step": 168131
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.187620997428894,
      "learning_rate": 0.00010179733400971277,
      "loss": 1.8387,
      "step": 168132
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.6009910106658936,
      "learning_rate": 0.0001017942633593902,
      "loss": 2.2021,
      "step": 168133
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.2915630340576172,
      "learning_rate": 0.0001017911927459177,
      "loss": 2.0557,
      "step": 168134
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.1319388151168823,
      "learning_rate": 0.0001017881221692961,
      "loss": 2.0862,
      "step": 168135
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.184619426727295,
      "learning_rate": 0.0001017850516295258,
      "loss": 1.968,
      "step": 168136
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.2650842666625977,
      "learning_rate": 0.0001017819811266073,
      "loss": 1.9289,
      "step": 168137
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.2581607103347778,
      "learning_rate": 0.0001017789106605414,
      "loss": 1.9354,
      "step": 168138
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.2455503940582275,
      "learning_rate": 0.00010177584023132842,
      "loss": 1.993,
      "step": 168139
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.2827847003936768,
      "learning_rate": 0.00010177276983896916,
      "loss": 1.7889,
      "step": 168140
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.2321171760559082,
      "learning_rate": 0.00010176969948346408,
      "loss": 1.827,
      "step": 168141
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.1753443479537964,
      "learning_rate": 0.00010176662916481377,
      "loss": 1.9227,
      "step": 168142
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.3259894847869873,
      "learning_rate": 0.00010176355888301866,
      "loss": 2.0298,
      "step": 168143
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.1123872995376587,
      "learning_rate": 0.00010176048863807954,
      "loss": 2.0659,
      "step": 168144
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.2031928300857544,
      "learning_rate": 0.00010175741842999681,
      "loss": 1.8727,
      "step": 168145
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.2058641910552979,
      "learning_rate": 0.00010175434825877123,
      "loss": 2.0764,
      "step": 168146
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.2558802366256714,
      "learning_rate": 0.00010175127812440322,
      "loss": 2.1133,
      "step": 168147
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.177405595779419,
      "learning_rate": 0.0001017482080268934,
      "loss": 2.1709,
      "step": 168148
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.3736214637756348,
      "learning_rate": 0.00010174513796624224,
      "loss": 2.2825,
      "step": 168149
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.37581205368042,
      "learning_rate": 0.00010174206794245052,
      "loss": 2.0994,
      "step": 168150
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.2277315855026245,
      "learning_rate": 0.00010173899795551859,
      "loss": 1.9885,
      "step": 168151
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.2443455457687378,
      "learning_rate": 0.0001017359280054472,
      "loss": 1.9713,
      "step": 168152
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.2924187183380127,
      "learning_rate": 0.00010173285809223672,
      "loss": 1.8589,
      "step": 168153
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.308986783027649,
      "learning_rate": 0.0001017297882158881,
      "loss": 2.1186,
      "step": 168154
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.3078862428665161,
      "learning_rate": 0.00010172671837640142,
      "loss": 1.9634,
      "step": 168155
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.193859338760376,
      "learning_rate": 0.00010172364857377759,
      "loss": 1.9154,
      "step": 168156
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.5158461332321167,
      "learning_rate": 0.000101720578808017,
      "loss": 1.87,
      "step": 168157
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.3451248407363892,
      "learning_rate": 0.00010171750907912039,
      "loss": 2.016,
      "step": 168158
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.2216752767562866,
      "learning_rate": 0.00010171443938708812,
      "loss": 2.0719,
      "step": 168159
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.2057639360427856,
      "learning_rate": 0.00010171136973192107,
      "loss": 2.0529,
      "step": 168160
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.1843748092651367,
      "learning_rate": 0.00010170830011361942,
      "loss": 2.0173,
      "step": 168161
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.326967716217041,
      "learning_rate": 0.00010170523053218406,
      "loss": 2.078,
      "step": 168162
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.3727033138275146,
      "learning_rate": 0.00010170216098761534,
      "loss": 2.008,
      "step": 168163
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.2388958930969238,
      "learning_rate": 0.00010169909147991405,
      "loss": 1.9561,
      "step": 168164
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.2562572956085205,
      "learning_rate": 0.00010169602200908051,
      "loss": 2.0057,
      "step": 168165
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.2861130237579346,
      "learning_rate": 0.00010169295257511562,
      "loss": 2.1345,
      "step": 168166
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.2320873737335205,
      "learning_rate": 0.00010168988317801956,
      "loss": 1.9988,
      "step": 168167
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.4382587671279907,
      "learning_rate": 0.00010168681381779321,
      "loss": 2.0118,
      "step": 168168
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.2664765119552612,
      "learning_rate": 0.0001016837444944369,
      "loss": 2.0217,
      "step": 168169
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.3486801385879517,
      "learning_rate": 0.00010168067520795147,
      "loss": 2.162,
      "step": 168170
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.2942290306091309,
      "learning_rate": 0.0001016776059583372,
      "loss": 1.9347,
      "step": 168171
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.2705730199813843,
      "learning_rate": 0.00010167453674559501,
      "loss": 2.1168,
      "step": 168172
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.37567937374115,
      "learning_rate": 0.0001016714675697251,
      "loss": 2.1136,
      "step": 168173
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.2449842691421509,
      "learning_rate": 0.00010166839843072829,
      "loss": 2.0324,
      "step": 168174
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.234944462776184,
      "learning_rate": 0.00010166532932860498,
      "loss": 2.279,
      "step": 168175
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.1187782287597656,
      "learning_rate": 0.0001016622602633559,
      "loss": 2.1165,
      "step": 168176
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.5357310771942139,
      "learning_rate": 0.00010165919123498146,
      "loss": 1.8539,
      "step": 168177
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.3113186359405518,
      "learning_rate": 0.00010165612224348253,
      "loss": 1.8849,
      "step": 168178
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.3256547451019287,
      "learning_rate": 0.00010165305328885927,
      "loss": 2.0501,
      "step": 168179
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.6620373725891113,
      "learning_rate": 0.00010164998437111254,
      "loss": 2.1041,
      "step": 168180
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.2685966491699219,
      "learning_rate": 0.00010164691549024275,
      "loss": 1.949,
      "step": 168181
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.2938473224639893,
      "learning_rate": 0.00010164384664625061,
      "loss": 2.0997,
      "step": 168182
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.2513489723205566,
      "learning_rate": 0.00010164077783913652,
      "loss": 1.7584,
      "step": 168183
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.3662906885147095,
      "learning_rate": 0.0001016377090689014,
      "loss": 1.8658,
      "step": 168184
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.1843913793563843,
      "learning_rate": 0.00010163464033554534,
      "loss": 2.0003,
      "step": 168185
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.2508448362350464,
      "learning_rate": 0.00010163157163906924,
      "loss": 2.1974,
      "step": 168186
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.3974767923355103,
      "learning_rate": 0.00010162850297947348,
      "loss": 2.2193,
      "step": 168187
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.1513768434524536,
      "learning_rate": 0.00010162543435675882,
      "loss": 2.0011,
      "step": 168188
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.1860451698303223,
      "learning_rate": 0.00010162236577092566,
      "loss": 1.9667,
      "step": 168189
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.3149669170379639,
      "learning_rate": 0.00010161929722197482,
      "loss": 2.1415,
      "step": 168190
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.1968300342559814,
      "learning_rate": 0.00010161622870990651,
      "loss": 2.2371,
      "step": 168191
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.2566823959350586,
      "learning_rate": 0.0001016131602347216,
      "loss": 2.1302,
      "step": 168192
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.2855710983276367,
      "learning_rate": 0.00010161009179642042,
      "loss": 1.7326,
      "step": 168193
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.2824020385742188,
      "learning_rate": 0.00010160702339500381,
      "loss": 2.0486,
      "step": 168194
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.2417460680007935,
      "learning_rate": 0.00010160395503047205,
      "loss": 1.9987,
      "step": 168195
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.379490852355957,
      "learning_rate": 0.00010160088670282609,
      "loss": 2.124,
      "step": 168196
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.4122939109802246,
      "learning_rate": 0.00010159781841206606,
      "loss": 2.2555,
      "step": 168197
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.4063130617141724,
      "learning_rate": 0.00010159475015819285,
      "loss": 1.8716,
      "step": 168198
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.2454876899719238,
      "learning_rate": 0.00010159168194120684,
      "loss": 1.888,
      "step": 168199
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.17750883102417,
      "learning_rate": 0.00010158861376110875,
      "loss": 2.0175,
      "step": 168200
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.3101636171340942,
      "learning_rate": 0.00010158554561789899,
      "loss": 1.987,
      "step": 168201
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.2227216958999634,
      "learning_rate": 0.00010158247751157832,
      "loss": 2.0578,
      "step": 168202
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.2798347473144531,
      "learning_rate": 0.0001015794094421472,
      "loss": 2.0652,
      "step": 168203
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.2443965673446655,
      "learning_rate": 0.00010157634140960625,
      "loss": 2.0537,
      "step": 168204
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.2613226175308228,
      "learning_rate": 0.00010157327341395586,
      "loss": 2.0925,
      "step": 168205
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.2203060388565063,
      "learning_rate": 0.00010157020545519689,
      "loss": 2.1758,
      "step": 168206
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.2903227806091309,
      "learning_rate": 0.00010156713753332964,
      "loss": 2.2064,
      "step": 168207
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.211775302886963,
      "learning_rate": 0.0001015640696483549,
      "loss": 2.1559,
      "step": 168208
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.1483049392700195,
      "learning_rate": 0.00010156100180027313,
      "loss": 1.891,
      "step": 168209
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.4026519060134888,
      "learning_rate": 0.00010155793398908496,
      "loss": 1.9287,
      "step": 168210
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.38504159450531,
      "learning_rate": 0.00010155486621479078,
      "loss": 1.9719,
      "step": 168211
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.1838445663452148,
      "learning_rate": 0.00010155179847739141,
      "loss": 2.0539,
      "step": 168212
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.2741665840148926,
      "learning_rate": 0.00010154873077688722,
      "loss": 2.0438,
      "step": 168213
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.4434112310409546,
      "learning_rate": 0.00010154566311327896,
      "loss": 1.9238,
      "step": 168214
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.254034399986267,
      "learning_rate": 0.00010154259548656706,
      "loss": 1.9246,
      "step": 168215
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.2723644971847534,
      "learning_rate": 0.0001015395278967521,
      "loss": 2.027,
      "step": 168216
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.263439416885376,
      "learning_rate": 0.00010153646034383476,
      "loss": 2.1188,
      "step": 168217
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.2193559408187866,
      "learning_rate": 0.00010153339282781557,
      "loss": 1.9274,
      "step": 168218
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.2875460386276245,
      "learning_rate": 0.00010153032534869492,
      "loss": 2.0452,
      "step": 168219
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.2577744722366333,
      "learning_rate": 0.00010152725790647366,
      "loss": 1.9638,
      "step": 168220
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.2206920385360718,
      "learning_rate": 0.00010152419050115222,
      "loss": 1.9053,
      "step": 168221
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.3634521961212158,
      "learning_rate": 0.00010152112313273108,
      "loss": 2.0689,
      "step": 168222
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.3700320720672607,
      "learning_rate": 0.00010151805580121101,
      "loss": 2.1306,
      "step": 168223
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.6164880990982056,
      "learning_rate": 0.00010151498850659238,
      "loss": 2.0013,
      "step": 168224
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.3647754192352295,
      "learning_rate": 0.00010151192124887596,
      "loss": 2.0522,
      "step": 168225
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.2733391523361206,
      "learning_rate": 0.00010150885402806222,
      "loss": 2.0162,
      "step": 168226
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.2309045791625977,
      "learning_rate": 0.00010150578684415174,
      "loss": 2.0013,
      "step": 168227
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.269828200340271,
      "learning_rate": 0.00010150271969714496,
      "loss": 2.0855,
      "step": 168228
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.40334951877594,
      "learning_rate": 0.0001014996525870427,
      "loss": 2.042,
      "step": 168229
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.4504759311676025,
      "learning_rate": 0.00010149658551384529,
      "loss": 1.9891,
      "step": 168230
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.41117262840271,
      "learning_rate": 0.00010149351847755348,
      "loss": 2.0616,
      "step": 168231
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.2148782014846802,
      "learning_rate": 0.00010149045147816782,
      "loss": 2.069,
      "step": 168232
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.3214129209518433,
      "learning_rate": 0.0001014873845156888,
      "loss": 2.0419,
      "step": 168233
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.3427460193634033,
      "learning_rate": 0.00010148431759011693,
      "loss": 2.2129,
      "step": 168234
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.2496083974838257,
      "learning_rate": 0.00010148125070145298,
      "loss": 2.009,
      "step": 168235
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.3369064331054688,
      "learning_rate": 0.00010147818384969733,
      "loss": 1.9911,
      "step": 168236
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.2606275081634521,
      "learning_rate": 0.00010147511703485068,
      "loss": 2.0328,
      "step": 168237
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.612023949623108,
      "learning_rate": 0.00010147205025691348,
      "loss": 1.819,
      "step": 168238
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.5536640882492065,
      "learning_rate": 0.0001014689835158866,
      "loss": 2.2527,
      "step": 168239
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.4047480821609497,
      "learning_rate": 0.00010146591681177017,
      "loss": 2.0477,
      "step": 168240
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.2382633686065674,
      "learning_rate": 0.00010146285014456507,
      "loss": 2.0202,
      "step": 168241
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.3490601778030396,
      "learning_rate": 0.00010145978351427167,
      "loss": 2.0796,
      "step": 168242
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.2525787353515625,
      "learning_rate": 0.00010145671692089073,
      "loss": 2.0648,
      "step": 168243
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.2559685707092285,
      "learning_rate": 0.00010145365036442264,
      "loss": 2.1846,
      "step": 168244
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.490734338760376,
      "learning_rate": 0.0001014505838448683,
      "loss": 1.855,
      "step": 168245
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.3260716199874878,
      "learning_rate": 0.00010144751736222781,
      "loss": 2.0951,
      "step": 168246
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.322824239730835,
      "learning_rate": 0.00010144445091650207,
      "loss": 2.0736,
      "step": 168247
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.2074705362319946,
      "learning_rate": 0.00010144138450769147,
      "loss": 2.2996,
      "step": 168248
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.4399101734161377,
      "learning_rate": 0.00010143831813579677,
      "loss": 1.9966,
      "step": 168249
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.693134069442749,
      "learning_rate": 0.00010143525180081831,
      "loss": 2.1821,
      "step": 168250
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.3579224348068237,
      "learning_rate": 0.00010143218550275701,
      "loss": 1.918,
      "step": 168251
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.2218170166015625,
      "learning_rate": 0.00010142911924161298,
      "loss": 2.3936,
      "step": 168252
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.2538121938705444,
      "learning_rate": 0.00010142605301738717,
      "loss": 1.9442,
      "step": 168253
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.233670949935913,
      "learning_rate": 0.00010142298683007988,
      "loss": 2.1431,
      "step": 168254
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.3317654132843018,
      "learning_rate": 0.00010141992067969192,
      "loss": 2.2048,
      "step": 168255
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.1877596378326416,
      "learning_rate": 0.00010141685456622363,
      "loss": 2.0186,
      "step": 168256
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.3767489194869995,
      "learning_rate": 0.00010141378848967592,
      "loss": 2.1046,
      "step": 168257
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.1745411157608032,
      "learning_rate": 0.0001014107224500489,
      "loss": 2.2721,
      "step": 168258
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.33585786819458,
      "learning_rate": 0.0001014076564473435,
      "loss": 1.9802,
      "step": 168259
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.5518704652786255,
      "learning_rate": 0.00010140459048156006,
      "loss": 2.0607,
      "step": 168260
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.4258087873458862,
      "learning_rate": 0.00010140152455269933,
      "loss": 2.0012,
      "step": 168261
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.4025145769119263,
      "learning_rate": 0.00010139845866076174,
      "loss": 2.0828,
      "step": 168262
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.3689336776733398,
      "learning_rate": 0.00010139539280574811,
      "loss": 2.2811,
      "step": 168263
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.4095174074172974,
      "learning_rate": 0.00010139232698765865,
      "loss": 2.0747,
      "step": 168264
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.6379237174987793,
      "learning_rate": 0.00010138926120649418,
      "loss": 1.9754,
      "step": 168265
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.458441138267517,
      "learning_rate": 0.0001013861954622551,
      "loss": 1.9276,
      "step": 168266
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.216768503189087,
      "learning_rate": 0.00010138312975494216,
      "loss": 2.0657,
      "step": 168267
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.6207332611083984,
      "learning_rate": 0.00010138006408455577,
      "loss": 1.871,
      "step": 168268
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.518068552017212,
      "learning_rate": 0.00010137699845109676,
      "loss": 2.1809,
      "step": 168269
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.3321211338043213,
      "learning_rate": 0.00010137393285456533,
      "loss": 1.8788,
      "step": 168270
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.294183373451233,
      "learning_rate": 0.00010137086729496233,
      "loss": 2.1752,
      "step": 168271
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.2480976581573486,
      "learning_rate": 0.00010136780177228812,
      "loss": 2.1104,
      "step": 168272
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.297297477722168,
      "learning_rate": 0.00010136473628654351,
      "loss": 2.0396,
      "step": 168273
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.3991281986236572,
      "learning_rate": 0.00010136167083772883,
      "loss": 1.9526,
      "step": 168274
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.3070828914642334,
      "learning_rate": 0.00010135860542584497,
      "loss": 1.9366,
      "step": 168275
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.3000481128692627,
      "learning_rate": 0.00010135554005089209,
      "loss": 2.0406,
      "step": 168276
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.284485936164856,
      "learning_rate": 0.00010135247471287105,
      "loss": 2.0986,
      "step": 168277
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.2469615936279297,
      "learning_rate": 0.00010134940941178226,
      "loss": 1.8333,
      "step": 168278
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.1983139514923096,
      "learning_rate": 0.00010134634414762645,
      "loss": 1.954,
      "step": 168279
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.2062631845474243,
      "learning_rate": 0.00010134327892040401,
      "loss": 1.9636,
      "step": 168280
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.1725554466247559,
      "learning_rate": 0.00010134021373011579,
      "loss": 1.9692,
      "step": 168281
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.3009910583496094,
      "learning_rate": 0.000101337148576762,
      "loss": 1.9869,
      "step": 168282
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.2770172357559204,
      "learning_rate": 0.00010133408346034348,
      "loss": 2.1734,
      "step": 168283
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.148231863975525,
      "learning_rate": 0.00010133101838086059,
      "loss": 2.0833,
      "step": 168284
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.1605019569396973,
      "learning_rate": 0.00010132795333831412,
      "loss": 1.9717,
      "step": 168285
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.496066689491272,
      "learning_rate": 0.00010132488833270444,
      "loss": 2.1181,
      "step": 168286
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.2938554286956787,
      "learning_rate": 0.00010132182336403232,
      "loss": 1.9628,
      "step": 168287
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.2847275733947754,
      "learning_rate": 0.00010131875843229821,
      "loss": 1.8569,
      "step": 168288
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.4899749755859375,
      "learning_rate": 0.00010131569353750268,
      "loss": 2.0201,
      "step": 168289
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.2603532075881958,
      "learning_rate": 0.00010131262867964625,
      "loss": 2.0555,
      "step": 168290
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.288387656211853,
      "learning_rate": 0.00010130956385872962,
      "loss": 2.0362,
      "step": 168291
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.2014576196670532,
      "learning_rate": 0.00010130649907475319,
      "loss": 2.1452,
      "step": 168292
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.207175374031067,
      "learning_rate": 0.00010130343432771777,
      "loss": 1.9712,
      "step": 168293
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.2855640649795532,
      "learning_rate": 0.00010130036961762376,
      "loss": 2.1091,
      "step": 168294
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.1802703142166138,
      "learning_rate": 0.00010129730494447178,
      "loss": 2.065,
      "step": 168295
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.3262085914611816,
      "learning_rate": 0.00010129424030826224,
      "loss": 2.078,
      "step": 168296
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.2792729139328003,
      "learning_rate": 0.00010129117570899598,
      "loss": 2.0669,
      "step": 168297
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.1813079118728638,
      "learning_rate": 0.00010128811114667338,
      "loss": 2.1623,
      "step": 168298
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.1787493228912354,
      "learning_rate": 0.00010128504662129511,
      "loss": 1.8922,
      "step": 168299
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.413230538368225,
      "learning_rate": 0.00010128198213286174,
      "loss": 2.0402,
      "step": 168300
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.1576632261276245,
      "learning_rate": 0.00010127891768137367,
      "loss": 2.0461,
      "step": 168301
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.2140947580337524,
      "learning_rate": 0.00010127585326683172,
      "loss": 1.9476,
      "step": 168302
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.3363407850265503,
      "learning_rate": 0.00010127278888923634,
      "loss": 1.8786,
      "step": 168303
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.3107949495315552,
      "learning_rate": 0.00010126972454858802,
      "loss": 1.8553,
      "step": 168304
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.1922703981399536,
      "learning_rate": 0.00010126666024488747,
      "loss": 2.0999,
      "step": 168305
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.5437346696853638,
      "learning_rate": 0.00010126359597813522,
      "loss": 2.0944,
      "step": 168306
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.237181544303894,
      "learning_rate": 0.00010126053174833173,
      "loss": 2.1625,
      "step": 168307
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.381195068359375,
      "learning_rate": 0.00010125746755547774,
      "loss": 2.1148,
      "step": 168308
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.342706561088562,
      "learning_rate": 0.00010125440339957373,
      "loss": 2.0543,
      "step": 168309
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.2385573387145996,
      "learning_rate": 0.0001012513392806202,
      "loss": 1.9987,
      "step": 168310
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.2151869535446167,
      "learning_rate": 0.00010124827519861789,
      "loss": 2.1651,
      "step": 168311
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.38113534450531,
      "learning_rate": 0.00010124521115356726,
      "loss": 2.1752,
      "step": 168312
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.342819333076477,
      "learning_rate": 0.00010124214714546883,
      "loss": 1.9507,
      "step": 168313
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.2522313594818115,
      "learning_rate": 0.00010123908317432332,
      "loss": 2.1273,
      "step": 168314
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.191490650177002,
      "learning_rate": 0.0001012360192401311,
      "loss": 2.0308,
      "step": 168315
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.1847952604293823,
      "learning_rate": 0.00010123295534289298,
      "loss": 1.9668,
      "step": 168316
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.320504069328308,
      "learning_rate": 0.00010122989148260939,
      "loss": 1.8986,
      "step": 168317
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.2350770235061646,
      "learning_rate": 0.00010122682765928092,
      "loss": 2.1516,
      "step": 168318
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.5273970365524292,
      "learning_rate": 0.00010122376387290805,
      "loss": 2.0042,
      "step": 168319
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.2945290803909302,
      "learning_rate": 0.00010122070012349151,
      "loss": 2.0032,
      "step": 168320
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.179337501525879,
      "learning_rate": 0.00010121763641103172,
      "loss": 1.9616,
      "step": 168321
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.3910167217254639,
      "learning_rate": 0.0001012145727355294,
      "loss": 2.1532,
      "step": 168322
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.3717749118804932,
      "learning_rate": 0.00010121150909698495,
      "loss": 2.0236,
      "step": 168323
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.1239473819732666,
      "learning_rate": 0.00010120844549539922,
      "loss": 1.8829,
      "step": 168324
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.269738793373108,
      "learning_rate": 0.00010120538193077244,
      "loss": 2.043,
      "step": 168325
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.2297545671463013,
      "learning_rate": 0.00010120231840310539,
      "loss": 1.8602,
      "step": 168326
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.317679762840271,
      "learning_rate": 0.00010119925491239849,
      "loss": 2.0354,
      "step": 168327
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.307321548461914,
      "learning_rate": 0.0001011961914586525,
      "loss": 2.1441,
      "step": 168328
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.2079464197158813,
      "learning_rate": 0.0001011931280418678,
      "loss": 2.045,
      "step": 168329
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.152518630027771,
      "learning_rate": 0.00010119006466204526,
      "loss": 1.9853,
      "step": 168330
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.2635440826416016,
      "learning_rate": 0.00010118700131918501,
      "loss": 2.1022,
      "step": 168331
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.181104063987732,
      "learning_rate": 0.00010118393801328797,
      "loss": 1.9193,
      "step": 168332
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.331639289855957,
      "learning_rate": 0.00010118087474435446,
      "loss": 2.0776,
      "step": 168333
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.1763737201690674,
      "learning_rate": 0.00010117781151238531,
      "loss": 1.9296,
      "step": 168334
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.1584835052490234,
      "learning_rate": 0.00010117474831738086,
      "loss": 2.027,
      "step": 168335
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.26483154296875,
      "learning_rate": 0.00010117168515934194,
      "loss": 2.0184,
      "step": 168336
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.4189614057540894,
      "learning_rate": 0.00010116862203826879,
      "loss": 2.0866,
      "step": 168337
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.2520573139190674,
      "learning_rate": 0.00010116555895416226,
      "loss": 2.0471,
      "step": 168338
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.2960437536239624,
      "learning_rate": 0.00010116249590702267,
      "loss": 2.0646,
      "step": 168339
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.2070406675338745,
      "learning_rate": 0.00010115943289685084,
      "loss": 1.9142,
      "step": 168340
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.3369860649108887,
      "learning_rate": 0.00010115636992364713,
      "loss": 2.0652,
      "step": 168341
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.2343236207962036,
      "learning_rate": 0.00010115330698741239,
      "loss": 1.9094,
      "step": 168342
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.1428114175796509,
      "learning_rate": 0.00010115024408814684,
      "loss": 1.9543,
      "step": 168343
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.153238296508789,
      "learning_rate": 0.00010114718122585129,
      "loss": 1.9152,
      "step": 168344
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.2562158107757568,
      "learning_rate": 0.00010114411840052612,
      "loss": 1.802,
      "step": 168345
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.244907259941101,
      "learning_rate": 0.00010114105561217212,
      "loss": 1.9105,
      "step": 168346
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.2061783075332642,
      "learning_rate": 0.00010113799286078964,
      "loss": 2.0704,
      "step": 168347
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.4823154211044312,
      "learning_rate": 0.0001011349301463796,
      "loss": 1.9348,
      "step": 168348
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.2038421630859375,
      "learning_rate": 0.00010113186746894206,
      "loss": 2.135,
      "step": 168349
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.1385798454284668,
      "learning_rate": 0.00010112880482847803,
      "loss": 1.7856,
      "step": 168350
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.3198431730270386,
      "learning_rate": 0.00010112574222498779,
      "loss": 2.0311,
      "step": 168351
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.4580183029174805,
      "learning_rate": 0.00010112267965847209,
      "loss": 1.7367,
      "step": 168352
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.2577295303344727,
      "learning_rate": 0.00010111961712893139,
      "loss": 1.9833,
      "step": 168353
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.3245360851287842,
      "learning_rate": 0.00010111655463636639,
      "loss": 2.064,
      "step": 168354
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.3686368465423584,
      "learning_rate": 0.00010111349218077757,
      "loss": 2.2067,
      "step": 168355
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.373712420463562,
      "learning_rate": 0.0001011104297621655,
      "loss": 2.0783,
      "step": 168356
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.2876973152160645,
      "learning_rate": 0.00010110736738053065,
      "loss": 1.9942,
      "step": 168357
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.354809284210205,
      "learning_rate": 0.00010110430503587384,
      "loss": 2.1297,
      "step": 168358
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.4228384494781494,
      "learning_rate": 0.00010110124272819537,
      "loss": 2.1842,
      "step": 168359
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.3679838180541992,
      "learning_rate": 0.00010109818045749602,
      "loss": 2.227,
      "step": 168360
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.3009161949157715,
      "learning_rate": 0.00010109511822377629,
      "loss": 1.82,
      "step": 168361
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.2630168199539185,
      "learning_rate": 0.0001010920560270367,
      "loss": 2.0335,
      "step": 168362
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.2249739170074463,
      "learning_rate": 0.00010108899386727777,
      "loss": 2.1148,
      "step": 168363
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.3779326677322388,
      "learning_rate": 0.00010108593174450027,
      "loss": 2.1083,
      "step": 168364
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.351203203201294,
      "learning_rate": 0.00010108286965870451,
      "loss": 2.1935,
      "step": 168365
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.1846404075622559,
      "learning_rate": 0.00010107980760989134,
      "loss": 1.7523,
      "step": 168366
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.2450157403945923,
      "learning_rate": 0.00010107674559806118,
      "loss": 1.9999,
      "step": 168367
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.271854043006897,
      "learning_rate": 0.00010107368362321461,
      "loss": 2.064,
      "step": 168368
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.3707364797592163,
      "learning_rate": 0.00010107062168535208,
      "loss": 1.9927,
      "step": 168369
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.4357765913009644,
      "learning_rate": 0.00010106755978447438,
      "loss": 2.206,
      "step": 168370
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.3507280349731445,
      "learning_rate": 0.00010106449792058187,
      "loss": 2.1767,
      "step": 168371
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.211774230003357,
      "learning_rate": 0.00010106143609367533,
      "loss": 2.1812,
      "step": 168372
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.4023559093475342,
      "learning_rate": 0.00010105837430375523,
      "loss": 2.0379,
      "step": 168373
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.2728110551834106,
      "learning_rate": 0.00010105531255082214,
      "loss": 2.1868,
      "step": 168374
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.2712363004684448,
      "learning_rate": 0.0001010522508348765,
      "loss": 2.0814,
      "step": 168375
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.3753535747528076,
      "learning_rate": 0.00010104918915591915,
      "loss": 1.9395,
      "step": 168376
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.232099175453186,
      "learning_rate": 0.00010104612751395036,
      "loss": 2.0775,
      "step": 168377
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.2219231128692627,
      "learning_rate": 0.00010104306590897099,
      "loss": 1.9216,
      "step": 168378
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.170039415359497,
      "learning_rate": 0.00010104000434098143,
      "loss": 1.9084,
      "step": 168379
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.2278367280960083,
      "learning_rate": 0.0001010369428099823,
      "loss": 2.0087,
      "step": 168380
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.2311302423477173,
      "learning_rate": 0.00010103388131597408,
      "loss": 1.9986,
      "step": 168381
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.382502555847168,
      "learning_rate": 0.00010103081985895751,
      "loss": 2.2294,
      "step": 168382
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.3374888896942139,
      "learning_rate": 0.00010102775843893296,
      "loss": 2.1321,
      "step": 168383
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.2983806133270264,
      "learning_rate": 0.00010102469705590119,
      "loss": 2.096,
      "step": 168384
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.3110158443450928,
      "learning_rate": 0.00010102163570986271,
      "loss": 2.1766,
      "step": 168385
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.3819454908370972,
      "learning_rate": 0.00010101857440081795,
      "loss": 1.8333,
      "step": 168386
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.2911813259124756,
      "learning_rate": 0.0001010155131287677,
      "loss": 2.0094,
      "step": 168387
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.3043688535690308,
      "learning_rate": 0.00010101245189371242,
      "loss": 1.9983,
      "step": 168388
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.244477391242981,
      "learning_rate": 0.00010100939069565258,
      "loss": 2.0297,
      "step": 168389
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.3439637422561646,
      "learning_rate": 0.00010100632953458895,
      "loss": 2.1397,
      "step": 168390
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.2086323499679565,
      "learning_rate": 0.000101003268410522,
      "loss": 2.234,
      "step": 168391
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.3096990585327148,
      "learning_rate": 0.0001010002073234522,
      "loss": 2.2163,
      "step": 168392
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.2893085479736328,
      "learning_rate": 0.00010099714627338031,
      "loss": 2.1971,
      "step": 168393
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.3180720806121826,
      "learning_rate": 0.00010099408526030683,
      "loss": 1.8953,
      "step": 168394
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.503326177597046,
      "learning_rate": 0.00010099102428423219,
      "loss": 2.062,
      "step": 168395
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.214974284172058,
      "learning_rate": 0.00010098796334515716,
      "loss": 1.8782,
      "step": 168396
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.2150053977966309,
      "learning_rate": 0.00010098490244308225,
      "loss": 1.9343,
      "step": 168397
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.34797203540802,
      "learning_rate": 0.00010098184157800789,
      "loss": 1.9735,
      "step": 168398
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.2151423692703247,
      "learning_rate": 0.00010097878074993484,
      "loss": 2.0393,
      "step": 168399
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.2000497579574585,
      "learning_rate": 0.00010097571995886355,
      "loss": 1.9412,
      "step": 168400
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.2395399808883667,
      "learning_rate": 0.0001009726592047947,
      "loss": 2.0837,
      "step": 168401
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.3232616186141968,
      "learning_rate": 0.00010096959848772876,
      "loss": 1.9246,
      "step": 168402
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.264822244644165,
      "learning_rate": 0.00010096653780766637,
      "loss": 1.9153,
      "step": 168403
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.369503378868103,
      "learning_rate": 0.00010096347716460794,
      "loss": 2.0189,
      "step": 168404
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.2254124879837036,
      "learning_rate": 0.00010096041655855427,
      "loss": 2.1385,
      "step": 168405
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.2847012281417847,
      "learning_rate": 0.00010095735598950571,
      "loss": 2.0811,
      "step": 168406
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.2121224403381348,
      "learning_rate": 0.00010095429545746302,
      "loss": 2.2207,
      "step": 168407
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.3125616312026978,
      "learning_rate": 0.00010095123496242669,
      "loss": 1.8833,
      "step": 168408
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.3814702033996582,
      "learning_rate": 0.00010094817450439728,
      "loss": 1.8642,
      "step": 168409
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.23008131980896,
      "learning_rate": 0.00010094511408337528,
      "loss": 2.0835,
      "step": 168410
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.2242469787597656,
      "learning_rate": 0.00010094205369936144,
      "loss": 2.1472,
      "step": 168411
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.3820044994354248,
      "learning_rate": 0.00010093899335235613,
      "loss": 2.2036,
      "step": 168412
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.3962719440460205,
      "learning_rate": 0.00010093593304236011,
      "loss": 2.0086,
      "step": 168413
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.107881784439087,
      "learning_rate": 0.00010093287276937376,
      "loss": 1.7911,
      "step": 168414
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.2737613916397095,
      "learning_rate": 0.00010092981253339793,
      "loss": 2.175,
      "step": 168415
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.148024082183838,
      "learning_rate": 0.00010092675233443279,
      "loss": 1.8852,
      "step": 168416
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.2473421096801758,
      "learning_rate": 0.00010092369217247925,
      "loss": 1.8608,
      "step": 168417
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.285006046295166,
      "learning_rate": 0.00010092063204753767,
      "loss": 2.0194,
      "step": 168418
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.2053759098052979,
      "learning_rate": 0.0001009175719596088,
      "loss": 2.1108,
      "step": 168419
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.2929846048355103,
      "learning_rate": 0.000100914511908693,
      "loss": 2.0586,
      "step": 168420
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.2551356554031372,
      "learning_rate": 0.00010091145189479103,
      "loss": 2.2183,
      "step": 168421
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.2909826040267944,
      "learning_rate": 0.00010090839191790342,
      "loss": 1.9336,
      "step": 168422
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.4587795734405518,
      "learning_rate": 0.00010090533197803067,
      "loss": 1.8844,
      "step": 168423
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.273787498474121,
      "learning_rate": 0.00010090227207517327,
      "loss": 2.1805,
      "step": 168424
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.540624737739563,
      "learning_rate": 0.00010089921220933204,
      "loss": 1.9873,
      "step": 168425
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.2599716186523438,
      "learning_rate": 0.00010089615238050725,
      "loss": 1.9422,
      "step": 168426
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.2305504083633423,
      "learning_rate": 0.00010089309258869977,
      "loss": 2.1403,
      "step": 168427
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.2213122844696045,
      "learning_rate": 0.00010089003283391,
      "loss": 2.165,
      "step": 168428
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.2727439403533936,
      "learning_rate": 0.00010088697311613851,
      "loss": 2.0534,
      "step": 168429
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.2073423862457275,
      "learning_rate": 0.00010088391343538582,
      "loss": 2.1931,
      "step": 168430
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.1286214590072632,
      "learning_rate": 0.00010088085379165265,
      "loss": 1.9182,
      "step": 168431
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.3058782815933228,
      "learning_rate": 0.00010087779418493943,
      "loss": 2.1653,
      "step": 168432
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.164267659187317,
      "learning_rate": 0.00010087473461524683,
      "loss": 2.034,
      "step": 168433
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.2774571180343628,
      "learning_rate": 0.00010087167508257542,
      "loss": 1.9386,
      "step": 168434
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.2305324077606201,
      "learning_rate": 0.0001008686155869257,
      "loss": 2.0277,
      "step": 168435
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.3331626653671265,
      "learning_rate": 0.00010086555612829819,
      "loss": 1.7092,
      "step": 168436
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.3621879816055298,
      "learning_rate": 0.0001008624967066936,
      "loss": 2.3001,
      "step": 168437
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.379530429840088,
      "learning_rate": 0.00010085943732211233,
      "loss": 1.9475,
      "step": 168438
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.2298282384872437,
      "learning_rate": 0.00010085637797455516,
      "loss": 2.1527,
      "step": 168439
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.1710724830627441,
      "learning_rate": 0.00010085331866402257,
      "loss": 2.0645,
      "step": 168440
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.3408125638961792,
      "learning_rate": 0.00010085025939051507,
      "loss": 1.9949,
      "step": 168441
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.1891322135925293,
      "learning_rate": 0.0001008472001540332,
      "loss": 2.0988,
      "step": 168442
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.2858506441116333,
      "learning_rate": 0.00010084414095457768,
      "loss": 1.9899,
      "step": 168443
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.1691761016845703,
      "learning_rate": 0.00010084108179214887,
      "loss": 1.9248,
      "step": 168444
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.3241010904312134,
      "learning_rate": 0.00010083802266674761,
      "loss": 2.0207,
      "step": 168445
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.261368989944458,
      "learning_rate": 0.0001008349635783743,
      "loss": 1.9968,
      "step": 168446
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.186686635017395,
      "learning_rate": 0.0001008319045270295,
      "loss": 1.7747,
      "step": 168447
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.288021445274353,
      "learning_rate": 0.00010082884551271373,
      "loss": 2.0114,
      "step": 168448
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.3301738500595093,
      "learning_rate": 0.00010082578653542777,
      "loss": 2.0527,
      "step": 168449
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.2902272939682007,
      "learning_rate": 0.0001008227275951719,
      "loss": 2.0049,
      "step": 168450
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.3911689519882202,
      "learning_rate": 0.00010081966869194699,
      "loss": 2.1497,
      "step": 168451
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.2329390048980713,
      "learning_rate": 0.00010081660982575344,
      "loss": 2.151,
      "step": 168452
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.2956141233444214,
      "learning_rate": 0.00010081355099659186,
      "loss": 2.0926,
      "step": 168453
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.6846842765808105,
      "learning_rate": 0.00010081049220446267,
      "loss": 2.1279,
      "step": 168454
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.1964683532714844,
      "learning_rate": 0.00010080743344936669,
      "loss": 2.1651,
      "step": 168455
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.1519283056259155,
      "learning_rate": 0.00010080437473130426,
      "loss": 1.9193,
      "step": 168456
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.2732573747634888,
      "learning_rate": 0.00010080131605027617,
      "loss": 1.8514,
      "step": 168457
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.2631540298461914,
      "learning_rate": 0.00010079825740628289,
      "loss": 2.0668,
      "step": 168458
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.3948397636413574,
      "learning_rate": 0.00010079519879932496,
      "loss": 2.2212,
      "step": 168459
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.364882469177246,
      "learning_rate": 0.00010079214022940282,
      "loss": 2.1326,
      "step": 168460
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.4850945472717285,
      "learning_rate": 0.00010078908169651732,
      "loss": 2.0581,
      "step": 168461
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.2388883829116821,
      "learning_rate": 0.00010078602320066881,
      "loss": 2.1969,
      "step": 168462
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.1456595659255981,
      "learning_rate": 0.000100782964741858,
      "loss": 2.0318,
      "step": 168463
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.3419920206069946,
      "learning_rate": 0.00010077990632008542,
      "loss": 1.9882,
      "step": 168464
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.2753832340240479,
      "learning_rate": 0.00010077684793535158,
      "loss": 1.9967,
      "step": 168465
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.255643367767334,
      "learning_rate": 0.00010077378958765702,
      "loss": 2.1095,
      "step": 168466
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.156821370124817,
      "learning_rate": 0.00010077073127700247,
      "loss": 2.0109,
      "step": 168467
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.284590244293213,
      "learning_rate": 0.00010076767300338832,
      "loss": 1.8249,
      "step": 168468
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.3444468975067139,
      "learning_rate": 0.0001007646147668153,
      "loss": 1.9765,
      "step": 168469
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.2714269161224365,
      "learning_rate": 0.00010076155656728388,
      "loss": 1.9784,
      "step": 168470
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.2096658945083618,
      "learning_rate": 0.00010075849840479468,
      "loss": 2.076,
      "step": 168471
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.312579870223999,
      "learning_rate": 0.00010075544027934813,
      "loss": 2.1222,
      "step": 168472
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.212007999420166,
      "learning_rate": 0.00010075238219094499,
      "loss": 2.0781,
      "step": 168473
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.2547125816345215,
      "learning_rate": 0.00010074932413958565,
      "loss": 2.0776,
      "step": 168474
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.342167854309082,
      "learning_rate": 0.00010074626612527088,
      "loss": 2.0261,
      "step": 168475
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.358457088470459,
      "learning_rate": 0.00010074320814800111,
      "loss": 2.2177,
      "step": 168476
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.2570586204528809,
      "learning_rate": 0.00010074015020777687,
      "loss": 2.1468,
      "step": 168477
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.359755039215088,
      "learning_rate": 0.0001007370923045989,
      "loss": 1.9408,
      "step": 168478
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.1752817630767822,
      "learning_rate": 0.00010073403443846766,
      "loss": 2.1315,
      "step": 168479
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.2122869491577148,
      "learning_rate": 0.00010073097660938358,
      "loss": 2.1362,
      "step": 168480
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.2702791690826416,
      "learning_rate": 0.00010072791881734754,
      "loss": 2.0186,
      "step": 168481
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.3288692235946655,
      "learning_rate": 0.0001007248610623598,
      "loss": 2.0989,
      "step": 168482
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.1788469552993774,
      "learning_rate": 0.00010072180334442118,
      "loss": 2.0017,
      "step": 168483
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.2082699537277222,
      "learning_rate": 0.00010071874566353216,
      "loss": 1.8339,
      "step": 168484
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.24970543384552,
      "learning_rate": 0.00010071568801969318,
      "loss": 2.2074,
      "step": 168485
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.32517409324646,
      "learning_rate": 0.00010071263041290498,
      "loss": 1.9352,
      "step": 168486
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.1984310150146484,
      "learning_rate": 0.00010070957284316808,
      "loss": 2.2929,
      "step": 168487
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.2827938795089722,
      "learning_rate": 0.00010070651531048295,
      "loss": 2.0122,
      "step": 168488
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.323233723640442,
      "learning_rate": 0.00010070345781485034,
      "loss": 1.8591,
      "step": 168489
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.3574802875518799,
      "learning_rate": 0.0001007004003562707,
      "loss": 2.03,
      "step": 168490
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.3170111179351807,
      "learning_rate": 0.00010069734293474454,
      "loss": 2.2725,
      "step": 168491
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.1604046821594238,
      "learning_rate": 0.00010069428555027258,
      "loss": 1.8298,
      "step": 168492
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.3625004291534424,
      "learning_rate": 0.00010069122820285531,
      "loss": 1.9432,
      "step": 168493
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.4750970602035522,
      "learning_rate": 0.00010068817089249324,
      "loss": 1.9303,
      "step": 168494
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.2183282375335693,
      "learning_rate": 0.00010068511361918709,
      "loss": 2.0174,
      "step": 168495
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.2026952505111694,
      "learning_rate": 0.00010068205638293732,
      "loss": 2.0702,
      "step": 168496
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.1971302032470703,
      "learning_rate": 0.00010067899918374442,
      "loss": 2.0078,
      "step": 168497
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.386845350265503,
      "learning_rate": 0.00010067594202160918,
      "loss": 2.184,
      "step": 168498
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.355595588684082,
      "learning_rate": 0.00010067288489653192,
      "loss": 2.0982,
      "step": 168499
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.242311716079712,
      "learning_rate": 0.00010066982780851346,
      "loss": 2.0892,
      "step": 168500
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.2520960569381714,
      "learning_rate": 0.00010066677075755422,
      "loss": 2.1297,
      "step": 168501
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.1986820697784424,
      "learning_rate": 0.00010066371374365479,
      "loss": 1.9188,
      "step": 168502
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.2676677703857422,
      "learning_rate": 0.00010066065676681566,
      "loss": 2.1849,
      "step": 168503
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.2244492769241333,
      "learning_rate": 0.00010065759982703755,
      "loss": 1.9356,
      "step": 168504
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.40299654006958,
      "learning_rate": 0.00010065454292432086,
      "loss": 2.2916,
      "step": 168505
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.3033292293548584,
      "learning_rate": 0.00010065148605866639,
      "loss": 1.9336,
      "step": 168506
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.2515742778778076,
      "learning_rate": 0.00010064842923007453,
      "loss": 2.1079,
      "step": 168507
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.2335342168807983,
      "learning_rate": 0.0001006453724385459,
      "loss": 2.0022,
      "step": 168508
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.1760872602462769,
      "learning_rate": 0.00010064231568408096,
      "loss": 1.8499,
      "step": 168509
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.3084499835968018,
      "learning_rate": 0.00010063925896668045,
      "loss": 2.2453,
      "step": 168510
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.2785543203353882,
      "learning_rate": 0.00010063620228634481,
      "loss": 1.8299,
      "step": 168511
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.261942744255066,
      "learning_rate": 0.00010063314564307476,
      "loss": 2.0186,
      "step": 168512
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.2321605682373047,
      "learning_rate": 0.00010063008903687075,
      "loss": 1.9022,
      "step": 168513
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.305095911026001,
      "learning_rate": 0.00010062703246773339,
      "loss": 2.1483,
      "step": 168514
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.3253846168518066,
      "learning_rate": 0.00010062397593566311,
      "loss": 2.1254,
      "step": 168515
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.1974338293075562,
      "learning_rate": 0.00010062091944066067,
      "loss": 1.855,
      "step": 168516
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.1942574977874756,
      "learning_rate": 0.00010061786298272653,
      "loss": 2.0015,
      "step": 168517
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.1434180736541748,
      "learning_rate": 0.00010061480656186132,
      "loss": 1.8024,
      "step": 168518
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.3497518301010132,
      "learning_rate": 0.00010061175017806564,
      "loss": 1.8818,
      "step": 168519
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.2398425340652466,
      "learning_rate": 0.00010060869383133997,
      "loss": 2.2944,
      "step": 168520
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.309314250946045,
      "learning_rate": 0.0001006056375216848,
      "loss": 2.2269,
      "step": 168521
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.2393980026245117,
      "learning_rate": 0.00010060258124910097,
      "loss": 1.9785,
      "step": 168522
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.3811511993408203,
      "learning_rate": 0.00010059952501358873,
      "loss": 2.04,
      "step": 168523
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.279160499572754,
      "learning_rate": 0.00010059646881514892,
      "loss": 2.2965,
      "step": 168524
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.2637078762054443,
      "learning_rate": 0.00010059341265378197,
      "loss": 2.0958,
      "step": 168525
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.224035382270813,
      "learning_rate": 0.00010059035652948847,
      "loss": 1.8675,
      "step": 168526
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.3587896823883057,
      "learning_rate": 0.00010058730044226889,
      "loss": 2.1714,
      "step": 168527
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.292812705039978,
      "learning_rate": 0.00010058424439212401,
      "loss": 2.0898,
      "step": 168528
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.3772943019866943,
      "learning_rate": 0.00010058118837905417,
      "loss": 2.1034,
      "step": 168529
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.2039635181427002,
      "learning_rate": 0.00010057813240306014,
      "loss": 1.8676,
      "step": 168530
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.3441438674926758,
      "learning_rate": 0.00010057507646414242,
      "loss": 2.3194,
      "step": 168531
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.346813440322876,
      "learning_rate": 0.00010057202056230155,
      "loss": 2.0919,
      "step": 168532
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.3225030899047852,
      "learning_rate": 0.00010056896469753802,
      "loss": 2.0891,
      "step": 168533
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.4321556091308594,
      "learning_rate": 0.00010056590886985256,
      "loss": 2.0673,
      "step": 168534
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.3827699422836304,
      "learning_rate": 0.00010056285307924557,
      "loss": 2.0417,
      "step": 168535
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.3316208124160767,
      "learning_rate": 0.00010055979732571782,
      "loss": 1.9552,
      "step": 168536
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.3483000993728638,
      "learning_rate": 0.00010055674160926978,
      "loss": 1.9202,
      "step": 168537
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.3589589595794678,
      "learning_rate": 0.00010055368592990197,
      "loss": 2.061,
      "step": 168538
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.1541552543640137,
      "learning_rate": 0.00010055063028761491,
      "loss": 2.0971,
      "step": 168539
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.2889496088027954,
      "learning_rate": 0.00010054757468240935,
      "loss": 2.1267,
      "step": 168540
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.2575597763061523,
      "learning_rate": 0.00010054451911428569,
      "loss": 1.9351,
      "step": 168541
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.240244746208191,
      "learning_rate": 0.00010054146358324464,
      "loss": 1.9716,
      "step": 168542
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.219850778579712,
      "learning_rate": 0.00010053840808928669,
      "loss": 2.0502,
      "step": 168543
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.297590732574463,
      "learning_rate": 0.00010053535263241245,
      "loss": 2.2841,
      "step": 168544
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.2381764650344849,
      "learning_rate": 0.00010053229721262233,
      "loss": 2.0771,
      "step": 168545
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.3185206651687622,
      "learning_rate": 0.00010052924182991713,
      "loss": 2.1532,
      "step": 168546
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.3833733797073364,
      "learning_rate": 0.0001005261864842972,
      "loss": 2.0671,
      "step": 168547
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.1680941581726074,
      "learning_rate": 0.00010052313117576331,
      "loss": 2.2035,
      "step": 168548
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.4036976099014282,
      "learning_rate": 0.00010052007590431584,
      "loss": 2.0473,
      "step": 168549
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.3314427137374878,
      "learning_rate": 0.00010051702066995567,
      "loss": 1.9496,
      "step": 168550
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.1956403255462646,
      "learning_rate": 0.00010051396547268294,
      "loss": 2.0897,
      "step": 168551
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.196367859840393,
      "learning_rate": 0.00010051091031249853,
      "loss": 2.1571,
      "step": 168552
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.204533576965332,
      "learning_rate": 0.00010050785518940279,
      "loss": 1.9783,
      "step": 168553
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.2533385753631592,
      "learning_rate": 0.00010050480010339651,
      "loss": 1.9408,
      "step": 168554
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.2175370454788208,
      "learning_rate": 0.00010050174505448007,
      "loss": 1.9175,
      "step": 168555
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.2081884145736694,
      "learning_rate": 0.00010049869004265432,
      "loss": 1.9623,
      "step": 168556
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.2480733394622803,
      "learning_rate": 0.00010049563506791943,
      "loss": 1.7995,
      "step": 168557
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.2381755113601685,
      "learning_rate": 0.00010049258013027624,
      "loss": 2.1167,
      "step": 168558
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.2805017232894897,
      "learning_rate": 0.00010048952522972518,
      "loss": 2.0724,
      "step": 168559
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.261064052581787,
      "learning_rate": 0.00010048647036626699,
      "loss": 1.8321,
      "step": 168560
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.3739466667175293,
      "learning_rate": 0.00010048341553990199,
      "loss": 1.9181,
      "step": 168561
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.3338532447814941,
      "learning_rate": 0.00010048036075063101,
      "loss": 2.0003,
      "step": 168562
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.2781850099563599,
      "learning_rate": 0.0001004773059984545,
      "loss": 2.0307,
      "step": 168563
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.177484393119812,
      "learning_rate": 0.00010047425128337306,
      "loss": 2.1476,
      "step": 168564
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.2256985902786255,
      "learning_rate": 0.00010047119660538706,
      "loss": 2.0152,
      "step": 168565
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.35622239112854,
      "learning_rate": 0.00010046814196449736,
      "loss": 2.1327,
      "step": 168566
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.160021185874939,
      "learning_rate": 0.00010046508736070432,
      "loss": 1.8523,
      "step": 168567
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.181471824645996,
      "learning_rate": 0.00010046203279400864,
      "loss": 2.0762,
      "step": 168568
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.22563636302948,
      "learning_rate": 0.00010045897826441088,
      "loss": 2.2166,
      "step": 168569
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.48106849193573,
      "learning_rate": 0.00010045592377191152,
      "loss": 2.1252,
      "step": 168570
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.2699687480926514,
      "learning_rate": 0.0001004528693165111,
      "loss": 2.0965,
      "step": 168571
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.2114150524139404,
      "learning_rate": 0.00010044981489821035,
      "loss": 2.0576,
      "step": 168572
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.2309802770614624,
      "learning_rate": 0.00010044676051700966,
      "loss": 2.0672,
      "step": 168573
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.2220828533172607,
      "learning_rate": 0.00010044370617290975,
      "loss": 1.9934,
      "step": 168574
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.3410147428512573,
      "learning_rate": 0.00010044065186591114,
      "loss": 2.0006,
      "step": 168575
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.3385542631149292,
      "learning_rate": 0.00010043759759601431,
      "loss": 2.0091,
      "step": 168576
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.3425477743148804,
      "learning_rate": 0.00010043454336321997,
      "loss": 2.092,
      "step": 168577
    },
    {
      "epoch": 2.19,
      "grad_norm": 1.1420236825942993,
      "learning_rate": 0.00010043148916752861,
      "loss": 1.9086,
      "step": 168578
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.2699874639511108,
      "learning_rate": 0.00010042843500894072,
      "loss": 2.0506,
      "step": 168579
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.2614036798477173,
      "learning_rate": 0.00010042538088745703,
      "loss": 1.9606,
      "step": 168580
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.157070517539978,
      "learning_rate": 0.00010042232680307803,
      "loss": 2.2428,
      "step": 168581
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.2605842351913452,
      "learning_rate": 0.00010041927275580419,
      "loss": 2.0492,
      "step": 168582
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.3117222785949707,
      "learning_rate": 0.00010041621874563626,
      "loss": 2.1323,
      "step": 168583
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.3102818727493286,
      "learning_rate": 0.00010041316477257466,
      "loss": 2.0787,
      "step": 168584
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.2734644412994385,
      "learning_rate": 0.00010041011083662011,
      "loss": 1.9013,
      "step": 168585
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.2687723636627197,
      "learning_rate": 0.00010040705693777306,
      "loss": 2.2801,
      "step": 168586
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.1664822101593018,
      "learning_rate": 0.00010040400307603415,
      "loss": 2.0288,
      "step": 168587
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.1475533246994019,
      "learning_rate": 0.00010040094925140377,
      "loss": 1.9274,
      "step": 168588
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.2741705179214478,
      "learning_rate": 0.00010039789546388275,
      "loss": 1.8892,
      "step": 168589
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.2702560424804688,
      "learning_rate": 0.00010039484171347141,
      "loss": 2.1049,
      "step": 168590
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.3088797330856323,
      "learning_rate": 0.00010039178800017051,
      "loss": 2.096,
      "step": 168591
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.256406307220459,
      "learning_rate": 0.00010038873432398057,
      "loss": 2.0503,
      "step": 168592
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.2134467363357544,
      "learning_rate": 0.00010038568068490214,
      "loss": 2.0625,
      "step": 168593
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.2916533946990967,
      "learning_rate": 0.00010038262708293568,
      "loss": 2.1188,
      "step": 168594
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.1845825910568237,
      "learning_rate": 0.00010037957351808195,
      "loss": 2.0566,
      "step": 168595
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.278926134109497,
      "learning_rate": 0.00010037651999034132,
      "loss": 1.9157,
      "step": 168596
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.2242755889892578,
      "learning_rate": 0.00010037346649971457,
      "loss": 2.1747,
      "step": 168597
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.5503312349319458,
      "learning_rate": 0.00010037041304620217,
      "loss": 1.9575,
      "step": 168598
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.3175790309906006,
      "learning_rate": 0.00010036735962980468,
      "loss": 1.9739,
      "step": 168599
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.1761283874511719,
      "learning_rate": 0.00010036430625052253,
      "loss": 1.9876,
      "step": 168600
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.2174335718154907,
      "learning_rate": 0.00010036125290835655,
      "loss": 1.992,
      "step": 168601
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.2935081720352173,
      "learning_rate": 0.0001003581996033071,
      "loss": 2.02,
      "step": 168602
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.2839224338531494,
      "learning_rate": 0.00010035514633537491,
      "loss": 1.9188,
      "step": 168603
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.2966928482055664,
      "learning_rate": 0.00010035209310456047,
      "loss": 2.0564,
      "step": 168604
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.1211283206939697,
      "learning_rate": 0.00010034903991086434,
      "loss": 1.7074,
      "step": 168605
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.2229995727539062,
      "learning_rate": 0.00010034598675428698,
      "loss": 1.928,
      "step": 168606
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.3194878101348877,
      "learning_rate": 0.00010034293363482919,
      "loss": 1.9681,
      "step": 168607
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.4964885711669922,
      "learning_rate": 0.00010033988055249132,
      "loss": 1.8879,
      "step": 168608
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.4780644178390503,
      "learning_rate": 0.00010033682750727411,
      "loss": 2.2128,
      "step": 168609
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.1807470321655273,
      "learning_rate": 0.00010033377449917808,
      "loss": 1.9928,
      "step": 168610
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.339552879333496,
      "learning_rate": 0.00010033072152820374,
      "loss": 1.9293,
      "step": 168611
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.7210978269577026,
      "learning_rate": 0.00010032766859435163,
      "loss": 1.9638,
      "step": 168612
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.7173430919647217,
      "learning_rate": 0.00010032461569762244,
      "loss": 1.8843,
      "step": 168613
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.235914945602417,
      "learning_rate": 0.0001003215628380166,
      "loss": 2.0091,
      "step": 168614
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.1677621603012085,
      "learning_rate": 0.00010031851001553482,
      "loss": 1.8639,
      "step": 168615
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.3481544256210327,
      "learning_rate": 0.00010031545723017752,
      "loss": 1.9211,
      "step": 168616
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.2736037969589233,
      "learning_rate": 0.00010031240448194555,
      "loss": 2.1734,
      "step": 168617
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.2509957551956177,
      "learning_rate": 0.00010030935177083905,
      "loss": 1.9396,
      "step": 168618
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.1506460905075073,
      "learning_rate": 0.00010030629909685892,
      "loss": 2.0584,
      "step": 168619
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.221053957939148,
      "learning_rate": 0.00010030324646000553,
      "loss": 1.9068,
      "step": 168620
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.4687587022781372,
      "learning_rate": 0.00010030019386027963,
      "loss": 2.128,
      "step": 168621
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.2310843467712402,
      "learning_rate": 0.00010029714129768161,
      "loss": 1.9382,
      "step": 168622
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.2561039924621582,
      "learning_rate": 0.00010029408877221228,
      "loss": 2.1473,
      "step": 168623
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.1719896793365479,
      "learning_rate": 0.00010029103628387188,
      "loss": 1.8649,
      "step": 168624
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.324600100517273,
      "learning_rate": 0.00010028798383266122,
      "loss": 2.1994,
      "step": 168625
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.215148687362671,
      "learning_rate": 0.00010028493141858072,
      "loss": 2.2691,
      "step": 168626
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.2257585525512695,
      "learning_rate": 0.00010028187904163114,
      "loss": 1.8593,
      "step": 168627
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.3096169233322144,
      "learning_rate": 0.00010027882670181279,
      "loss": 2.0654,
      "step": 168628
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.1400465965270996,
      "learning_rate": 0.0001002757743991266,
      "loss": 1.9007,
      "step": 168629
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.1660230159759521,
      "learning_rate": 0.0001002727221335727,
      "loss": 1.8757,
      "step": 168630
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.3294693231582642,
      "learning_rate": 0.00010026966990515195,
      "loss": 2.0184,
      "step": 168631
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.3497329950332642,
      "learning_rate": 0.00010026661771386478,
      "loss": 2.0529,
      "step": 168632
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.2464607954025269,
      "learning_rate": 0.0001002635655597119,
      "loss": 1.9376,
      "step": 168633
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.401672124862671,
      "learning_rate": 0.00010026051344269367,
      "loss": 1.9592,
      "step": 168634
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.1920605897903442,
      "learning_rate": 0.000100257461362811,
      "loss": 1.9707,
      "step": 168635
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.3850274085998535,
      "learning_rate": 0.00010025440932006404,
      "loss": 2.0312,
      "step": 168636
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.3774060010910034,
      "learning_rate": 0.00010025135731445367,
      "loss": 1.9001,
      "step": 168637
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.2473113536834717,
      "learning_rate": 0.00010024830534598022,
      "loss": 1.9235,
      "step": 168638
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.2959452867507935,
      "learning_rate": 0.00010024525341464451,
      "loss": 2.0348,
      "step": 168639
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.180212378501892,
      "learning_rate": 0.00010024220152044685,
      "loss": 2.1891,
      "step": 168640
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.2650184631347656,
      "learning_rate": 0.00010023914966338816,
      "loss": 1.9325,
      "step": 168641
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.2209519147872925,
      "learning_rate": 0.00010023609784346854,
      "loss": 1.924,
      "step": 168642
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.2080726623535156,
      "learning_rate": 0.00010023304606068893,
      "loss": 1.9649,
      "step": 168643
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.2270933389663696,
      "learning_rate": 0.00010022999431504967,
      "loss": 2.1724,
      "step": 168644
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.195540189743042,
      "learning_rate": 0.00010022694260655153,
      "loss": 2.0251,
      "step": 168645
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.4327143430709839,
      "learning_rate": 0.00010022389093519485,
      "loss": 2.2192,
      "step": 168646
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.5043736696243286,
      "learning_rate": 0.00010022083930098043,
      "loss": 2.0829,
      "step": 168647
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.2418261766433716,
      "learning_rate": 0.00010021778770390873,
      "loss": 2.0718,
      "step": 168648
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.2738800048828125,
      "learning_rate": 0.00010021473614398031,
      "loss": 2.211,
      "step": 168649
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.30001962184906,
      "learning_rate": 0.00010021168462119565,
      "loss": 1.8413,
      "step": 168650
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.316856026649475,
      "learning_rate": 0.00010020863313555549,
      "loss": 1.9161,
      "step": 168651
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.4301469326019287,
      "learning_rate": 0.00010020558168706023,
      "loss": 2.1621,
      "step": 168652
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.274510145187378,
      "learning_rate": 0.00010020253027571063,
      "loss": 1.9526,
      "step": 168653
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.3755698204040527,
      "learning_rate": 0.00010019947890150716,
      "loss": 1.9801,
      "step": 168654
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.2259913682937622,
      "learning_rate": 0.00010019642756445037,
      "loss": 2.0692,
      "step": 168655
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.3590985536575317,
      "learning_rate": 0.00010019337626454073,
      "loss": 2.1298,
      "step": 168656
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.304552435874939,
      "learning_rate": 0.000100190325001779,
      "loss": 2.0278,
      "step": 168657
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.3266665935516357,
      "learning_rate": 0.00010018727377616556,
      "loss": 1.9304,
      "step": 168658
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.5190097093582153,
      "learning_rate": 0.00010018422258770118,
      "loss": 1.996,
      "step": 168659
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.352522611618042,
      "learning_rate": 0.00010018117143638635,
      "loss": 2.1123,
      "step": 168660
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.2354848384857178,
      "learning_rate": 0.0001001781203222215,
      "loss": 2.0876,
      "step": 168661
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.2892415523529053,
      "learning_rate": 0.00010017506924520742,
      "loss": 1.9215,
      "step": 168662
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.221137285232544,
      "learning_rate": 0.00010017201820534453,
      "loss": 2.1334,
      "step": 168663
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.440687894821167,
      "learning_rate": 0.00010016896720263335,
      "loss": 2.1673,
      "step": 168664
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.2405576705932617,
      "learning_rate": 0.00010016591623707466,
      "loss": 2.0486,
      "step": 168665
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.2322555780410767,
      "learning_rate": 0.00010016286530866888,
      "loss": 2.1216,
      "step": 168666
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.2096608877182007,
      "learning_rate": 0.00010015981441741648,
      "loss": 1.8903,
      "step": 168667
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.2341066598892212,
      "learning_rate": 0.00010015676356331827,
      "loss": 2.064,
      "step": 168668
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.2561383247375488,
      "learning_rate": 0.00010015371274637456,
      "loss": 2.096,
      "step": 168669
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.212354302406311,
      "learning_rate": 0.00010015066196658617,
      "loss": 2.0327,
      "step": 168670
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.2619990110397339,
      "learning_rate": 0.00010014761122395352,
      "loss": 2.1406,
      "step": 168671
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.2359951734542847,
      "learning_rate": 0.00010014456051847721,
      "loss": 2.1195,
      "step": 168672
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.2486895322799683,
      "learning_rate": 0.00010014150985015771,
      "loss": 2.1065,
      "step": 168673
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.3019925355911255,
      "learning_rate": 0.00010013845921899577,
      "loss": 1.8549,
      "step": 168674
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.2829424142837524,
      "learning_rate": 0.00010013540862499177,
      "loss": 2.0517,
      "step": 168675
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.1250767707824707,
      "learning_rate": 0.00010013235806814647,
      "loss": 1.9097,
      "step": 168676
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.3945231437683105,
      "learning_rate": 0.00010012930754846035,
      "loss": 1.9884,
      "step": 168677
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.1705161333084106,
      "learning_rate": 0.00010012625706593394,
      "loss": 2.0627,
      "step": 168678
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.290006399154663,
      "learning_rate": 0.00010012320662056773,
      "loss": 2.1643,
      "step": 168679
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.2265183925628662,
      "learning_rate": 0.0001001201562123625,
      "loss": 2.2707,
      "step": 168680
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.2627393007278442,
      "learning_rate": 0.00010011710584131861,
      "loss": 1.8107,
      "step": 168681
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.2815409898757935,
      "learning_rate": 0.00010011405550743683,
      "loss": 2.048,
      "step": 168682
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.3765685558319092,
      "learning_rate": 0.00010011100521071748,
      "loss": 1.9737,
      "step": 168683
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.3676695823669434,
      "learning_rate": 0.00010010795495116152,
      "loss": 1.8405,
      "step": 168684
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.1858783960342407,
      "learning_rate": 0.00010010490472876902,
      "loss": 1.9859,
      "step": 168685
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.2938055992126465,
      "learning_rate": 0.00010010185454354093,
      "loss": 2.3261,
      "step": 168686
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.2154676914215088,
      "learning_rate": 0.00010009880439547754,
      "loss": 2.0998,
      "step": 168687
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.4084078073501587,
      "learning_rate": 0.00010009575428457968,
      "loss": 2.2069,
      "step": 168688
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.2451510429382324,
      "learning_rate": 0.0001000927042108477,
      "loss": 2.1421,
      "step": 168689
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.5592281818389893,
      "learning_rate": 0.00010008965417428244,
      "loss": 1.9714,
      "step": 168690
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.254421353340149,
      "learning_rate": 0.00010008660417488409,
      "loss": 2.2131,
      "step": 168691
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.3126322031021118,
      "learning_rate": 0.00010008355421265352,
      "loss": 1.8626,
      "step": 168692
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.3865132331848145,
      "learning_rate": 0.00010008050428759111,
      "loss": 2.1654,
      "step": 168693
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.3326869010925293,
      "learning_rate": 0.00010007745439969758,
      "loss": 2.1882,
      "step": 168694
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.179550290107727,
      "learning_rate": 0.00010007440454897335,
      "loss": 1.8639,
      "step": 168695
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.281388521194458,
      "learning_rate": 0.0001000713547354192,
      "loss": 2.0902,
      "step": 168696
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.2105686664581299,
      "learning_rate": 0.00010006830495903543,
      "loss": 1.8718,
      "step": 168697
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.3000624179840088,
      "learning_rate": 0.0001000652552198228,
      "loss": 2.1566,
      "step": 168698
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.316766381263733,
      "learning_rate": 0.00010006220551778174,
      "loss": 2.0325,
      "step": 168699
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.234083890914917,
      "learning_rate": 0.00010005915585291296,
      "loss": 2.1379,
      "step": 168700
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.3563776016235352,
      "learning_rate": 0.00010005610622521686,
      "loss": 2.1268,
      "step": 168701
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.1670827865600586,
      "learning_rate": 0.00010005305663469432,
      "loss": 1.9821,
      "step": 168702
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.2848715782165527,
      "learning_rate": 0.00010005000708134551,
      "loss": 1.9714,
      "step": 168703
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.3232569694519043,
      "learning_rate": 0.00010004695756517122,
      "loss": 2.2039,
      "step": 168704
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.2096680402755737,
      "learning_rate": 0.00010004390808617191,
      "loss": 1.8754,
      "step": 168705
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.5084160566329956,
      "learning_rate": 0.00010004085864434832,
      "loss": 2.2271,
      "step": 168706
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.2086142301559448,
      "learning_rate": 0.0001000378092397008,
      "loss": 1.9642,
      "step": 168707
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.321122407913208,
      "learning_rate": 0.00010003475987223023,
      "loss": 1.912,
      "step": 168708
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.2512661218643188,
      "learning_rate": 0.00010003171054193676,
      "loss": 1.913,
      "step": 168709
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.2887465953826904,
      "learning_rate": 0.00010002866124882128,
      "loss": 2.1506,
      "step": 168710
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.1902717351913452,
      "learning_rate": 0.00010002561199288414,
      "loss": 1.9688,
      "step": 168711
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.2655435800552368,
      "learning_rate": 0.00010002256277412611,
      "loss": 2.2484,
      "step": 168712
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.263802170753479,
      "learning_rate": 0.00010001951359254758,
      "loss": 2.0853,
      "step": 168713
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.2457146644592285,
      "learning_rate": 0.00010001646444814939,
      "loss": 2.2084,
      "step": 168714
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.222707748413086,
      "learning_rate": 0.0001000134153409317,
      "loss": 1.9596,
      "step": 168715
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.2628127336502075,
      "learning_rate": 0.00010001036627089541,
      "loss": 1.9536,
      "step": 168716
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.127102017402649,
      "learning_rate": 0.00010000731723804086,
      "loss": 2.3075,
      "step": 168717
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.3298019170761108,
      "learning_rate": 0.00010000426824236881,
      "loss": 2.0219,
      "step": 168718
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.147656798362732,
      "learning_rate": 0.00010000121928387966,
      "loss": 2.0863,
      "step": 168719
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.2415201663970947,
      "learning_rate": 9.999817036257427e-05,
      "loss": 1.937,
      "step": 168720
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.266534686088562,
      "learning_rate": 9.999512147845277e-05,
      "loss": 2.1537,
      "step": 168721
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.22476065158844,
      "learning_rate": 9.999207263151609e-05,
      "loss": 2.3307,
      "step": 168722
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.2934778928756714,
      "learning_rate": 9.998902382176453e-05,
      "loss": 2.0229,
      "step": 168723
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.2267253398895264,
      "learning_rate": 9.998597504919885e-05,
      "loss": 1.9264,
      "step": 168724
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.2838733196258545,
      "learning_rate": 9.99829263138195e-05,
      "loss": 2.0903,
      "step": 168725
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.192097544670105,
      "learning_rate": 9.99798776156273e-05,
      "loss": 1.993,
      "step": 168726
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.2481359243392944,
      "learning_rate": 9.997682895462239e-05,
      "loss": 2.0769,
      "step": 168727
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.3281114101409912,
      "learning_rate": 9.99737803308057e-05,
      "loss": 2.0334,
      "step": 168728
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.2655961513519287,
      "learning_rate": 9.997073174417752e-05,
      "loss": 1.9214,
      "step": 168729
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.208952784538269,
      "learning_rate": 9.99676831947387e-05,
      "loss": 1.9386,
      "step": 168730
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.2302476167678833,
      "learning_rate": 9.996463468248953e-05,
      "loss": 2.1401,
      "step": 168731
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.3749642372131348,
      "learning_rate": 9.99615862074309e-05,
      "loss": 2.0591,
      "step": 168732
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.4566611051559448,
      "learning_rate": 9.995853776956305e-05,
      "loss": 2.0861,
      "step": 168733
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.285783290863037,
      "learning_rate": 9.995548936888672e-05,
      "loss": 1.9656,
      "step": 168734
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.208028793334961,
      "learning_rate": 9.995244100540236e-05,
      "loss": 2.1468,
      "step": 168735
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.4454553127288818,
      "learning_rate": 9.994939267911073e-05,
      "loss": 1.9454,
      "step": 168736
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.3818237781524658,
      "learning_rate": 9.994634439001216e-05,
      "loss": 2.1393,
      "step": 168737
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.2001436948776245,
      "learning_rate": 9.994329613810746e-05,
      "loss": 2.009,
      "step": 168738
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.295086145401001,
      "learning_rate": 9.994024792339706e-05,
      "loss": 2.0832,
      "step": 168739
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.4081581830978394,
      "learning_rate": 9.993719974588155e-05,
      "loss": 1.9853,
      "step": 168740
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.264931321144104,
      "learning_rate": 9.99341516055614e-05,
      "loss": 1.8978,
      "step": 168741
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.2401095628738403,
      "learning_rate": 9.993110350243738e-05,
      "loss": 1.7948,
      "step": 168742
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.193615436553955,
      "learning_rate": 9.992805543650982e-05,
      "loss": 1.9586,
      "step": 168743
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.2857786417007446,
      "learning_rate": 9.992500740777954e-05,
      "loss": 1.9595,
      "step": 168744
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.1406420469284058,
      "learning_rate": 9.992195941624694e-05,
      "loss": 1.831,
      "step": 168745
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.2299952507019043,
      "learning_rate": 9.991891146191254e-05,
      "loss": 1.87,
      "step": 168746
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.2709486484527588,
      "learning_rate": 9.99158635447771e-05,
      "loss": 2.0692,
      "step": 168747
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.257859230041504,
      "learning_rate": 9.991281566484106e-05,
      "loss": 2.1811,
      "step": 168748
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.2649176120758057,
      "learning_rate": 9.990976782210494e-05,
      "loss": 1.9938,
      "step": 168749
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.2026431560516357,
      "learning_rate": 9.990672001656945e-05,
      "loss": 2.117,
      "step": 168750
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.1828923225402832,
      "learning_rate": 9.990367224823507e-05,
      "loss": 1.9174,
      "step": 168751
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.2332563400268555,
      "learning_rate": 9.990062451710227e-05,
      "loss": 2.0294,
      "step": 168752
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.0869159698486328,
      "learning_rate": 9.989757682317182e-05,
      "loss": 2.0397,
      "step": 168753
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.1624456644058228,
      "learning_rate": 9.98945291664442e-05,
      "loss": 2.0894,
      "step": 168754
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.324293613433838,
      "learning_rate": 9.989148154691986e-05,
      "loss": 2.1576,
      "step": 168755
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.3727036714553833,
      "learning_rate": 9.98884339645996e-05,
      "loss": 1.8523,
      "step": 168756
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.218498945236206,
      "learning_rate": 9.988538641948382e-05,
      "loss": 2.0807,
      "step": 168757
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.2904512882232666,
      "learning_rate": 9.988233891157303e-05,
      "loss": 2.1308,
      "step": 168758
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.3431874513626099,
      "learning_rate": 9.987929144086802e-05,
      "loss": 2.0387,
      "step": 168759
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.217008113861084,
      "learning_rate": 9.98762440073691e-05,
      "loss": 1.7655,
      "step": 168760
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.2209700345993042,
      "learning_rate": 9.987319661107705e-05,
      "loss": 1.9269,
      "step": 168761
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.252625584602356,
      "learning_rate": 9.987014925199239e-05,
      "loss": 1.9315,
      "step": 168762
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.148203730583191,
      "learning_rate": 9.986710193011562e-05,
      "loss": 1.9988,
      "step": 168763
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.349799633026123,
      "learning_rate": 9.986405464544722e-05,
      "loss": 2.1038,
      "step": 168764
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.2296022176742554,
      "learning_rate": 9.986100739798798e-05,
      "loss": 2.0105,
      "step": 168765
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.2049793004989624,
      "learning_rate": 9.985796018773828e-05,
      "loss": 1.9351,
      "step": 168766
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.2936941385269165,
      "learning_rate": 9.985491301469886e-05,
      "loss": 2.1352,
      "step": 168767
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.2249187231063843,
      "learning_rate": 9.985186587887006e-05,
      "loss": 1.8849,
      "step": 168768
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.2154985666275024,
      "learning_rate": 9.984881878025281e-05,
      "loss": 1.9991,
      "step": 168769
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.2672004699707031,
      "learning_rate": 9.984577171884721e-05,
      "loss": 2.1091,
      "step": 168770
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.1452440023422241,
      "learning_rate": 9.984272469465419e-05,
      "loss": 2.0701,
      "step": 168771
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.150267481803894,
      "learning_rate": 9.983967770767405e-05,
      "loss": 2.1348,
      "step": 168772
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.1866244077682495,
      "learning_rate": 9.983663075790763e-05,
      "loss": 2.1385,
      "step": 168773
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.2522205114364624,
      "learning_rate": 9.983358384535523e-05,
      "loss": 2.0248,
      "step": 168774
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.2193061113357544,
      "learning_rate": 9.983053697001776e-05,
      "loss": 1.9077,
      "step": 168775
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.8721851110458374,
      "learning_rate": 9.982749013189537e-05,
      "loss": 2.0064,
      "step": 168776
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.2491443157196045,
      "learning_rate": 9.982444333098893e-05,
      "loss": 2.0194,
      "step": 168777
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.30741548538208,
      "learning_rate": 9.98213965672988e-05,
      "loss": 2.0045,
      "step": 168778
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.2565314769744873,
      "learning_rate": 9.981834984082575e-05,
      "loss": 1.9766,
      "step": 168779
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.3378251791000366,
      "learning_rate": 9.981530315157016e-05,
      "loss": 2.2416,
      "step": 168780
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.159960150718689,
      "learning_rate": 9.981225649953289e-05,
      "loss": 1.7936,
      "step": 168781
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.2823837995529175,
      "learning_rate": 9.980920988471405e-05,
      "loss": 2.042,
      "step": 168782
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.3184890747070312,
      "learning_rate": 9.980616330711463e-05,
      "loss": 2.1255,
      "step": 168783
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.2969037294387817,
      "learning_rate": 9.980311676673486e-05,
      "loss": 2.0684,
      "step": 168784
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.263647198677063,
      "learning_rate": 9.980007026357561e-05,
      "loss": 1.9634,
      "step": 168785
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.2259273529052734,
      "learning_rate": 9.97970237976372e-05,
      "loss": 1.9649,
      "step": 168786
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.1717650890350342,
      "learning_rate": 9.979397736892048e-05,
      "loss": 1.9746,
      "step": 168787
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.3707592487335205,
      "learning_rate": 9.979093097742565e-05,
      "loss": 2.187,
      "step": 168788
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.2695456743240356,
      "learning_rate": 9.978788462315356e-05,
      "loss": 2.2056,
      "step": 168789
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.2788736820220947,
      "learning_rate": 9.978483830610459e-05,
      "loss": 2.0332,
      "step": 168790
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.1799161434173584,
      "learning_rate": 9.97817920262795e-05,
      "loss": 2.096,
      "step": 168791
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.3425532579421997,
      "learning_rate": 9.977874578367866e-05,
      "loss": 2.1791,
      "step": 168792
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.105081558227539,
      "learning_rate": 9.977569957830292e-05,
      "loss": 1.9205,
      "step": 168793
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.1926392316818237,
      "learning_rate": 9.977265341015246e-05,
      "loss": 1.8816,
      "step": 168794
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.329238772392273,
      "learning_rate": 9.976960727922812e-05,
      "loss": 1.9883,
      "step": 168795
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.3666322231292725,
      "learning_rate": 9.976656118553032e-05,
      "loss": 2.0844,
      "step": 168796
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.4309685230255127,
      "learning_rate": 9.97635151290598e-05,
      "loss": 2.0517,
      "step": 168797
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.2542674541473389,
      "learning_rate": 9.976046910981692e-05,
      "loss": 2.0811,
      "step": 168798
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.2699718475341797,
      "learning_rate": 9.975742312780254e-05,
      "loss": 1.9554,
      "step": 168799
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.221260666847229,
      "learning_rate": 9.975437718301682e-05,
      "loss": 1.963,
      "step": 168800
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.2462804317474365,
      "learning_rate": 9.975133127546066e-05,
      "loss": 1.7979,
      "step": 168801
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.2499520778656006,
      "learning_rate": 9.974828540513442e-05,
      "loss": 2.1379,
      "step": 168802
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.211517572402954,
      "learning_rate": 9.97452395720388e-05,
      "loss": 1.8579,
      "step": 168803
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.3018487691879272,
      "learning_rate": 9.974219377617425e-05,
      "loss": 1.9451,
      "step": 168804
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.4364923238754272,
      "learning_rate": 9.973914801754161e-05,
      "loss": 2.1406,
      "step": 168805
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.3474160432815552,
      "learning_rate": 9.973610229614106e-05,
      "loss": 2.0272,
      "step": 168806
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.2676752805709839,
      "learning_rate": 9.973305661197343e-05,
      "loss": 1.9263,
      "step": 168807
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.1983810663223267,
      "learning_rate": 9.973001096503911e-05,
      "loss": 1.8952,
      "step": 168808
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.308148741722107,
      "learning_rate": 9.972696535533887e-05,
      "loss": 2.0651,
      "step": 168809
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.5834903717041016,
      "learning_rate": 9.972391978287306e-05,
      "loss": 1.9951,
      "step": 168810
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.30647873878479,
      "learning_rate": 9.972087424764254e-05,
      "loss": 2.306,
      "step": 168811
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.2839285135269165,
      "learning_rate": 9.971782874964749e-05,
      "loss": 1.9403,
      "step": 168812
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.4055246114730835,
      "learning_rate": 9.971478328888882e-05,
      "loss": 2.1491,
      "step": 168813
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.2034624814987183,
      "learning_rate": 9.971173786536681e-05,
      "loss": 2.1417,
      "step": 168814
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.2231009006500244,
      "learning_rate": 9.970869247908227e-05,
      "loss": 1.9471,
      "step": 168815
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.2005256414413452,
      "learning_rate": 9.970564713003556e-05,
      "loss": 2.0809,
      "step": 168816
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.315524697303772,
      "learning_rate": 9.970260181822747e-05,
      "loss": 2.1158,
      "step": 168817
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.2543011903762817,
      "learning_rate": 9.969955654365846e-05,
      "loss": 2.1905,
      "step": 168818
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.3433582782745361,
      "learning_rate": 9.969651130632905e-05,
      "loss": 1.9413,
      "step": 168819
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.2263495922088623,
      "learning_rate": 9.969346610623976e-05,
      "loss": 2.0417,
      "step": 168820
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.3303552865982056,
      "learning_rate": 9.969042094339135e-05,
      "loss": 2.0189,
      "step": 168821
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.1903443336486816,
      "learning_rate": 9.968737581778414e-05,
      "loss": 1.8274,
      "step": 168822
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.1935333013534546,
      "learning_rate": 9.968433072941895e-05,
      "loss": 1.9096,
      "step": 168823
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.2823774814605713,
      "learning_rate": 9.96812856782962e-05,
      "loss": 2.1122,
      "step": 168824
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.1978440284729004,
      "learning_rate": 9.967824066441649e-05,
      "loss": 2.1352,
      "step": 168825
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.3178319931030273,
      "learning_rate": 9.967519568778028e-05,
      "loss": 1.9169,
      "step": 168826
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.2153453826904297,
      "learning_rate": 9.967215074838836e-05,
      "loss": 2.0384,
      "step": 168827
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.3179101943969727,
      "learning_rate": 9.966910584624104e-05,
      "loss": 1.8021,
      "step": 168828
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.3467828035354614,
      "learning_rate": 9.96660609813391e-05,
      "loss": 1.7533,
      "step": 168829
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.2406055927276611,
      "learning_rate": 9.966301615368306e-05,
      "loss": 2.1529,
      "step": 168830
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.2350233793258667,
      "learning_rate": 9.965997136327332e-05,
      "loss": 2.0691,
      "step": 168831
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.3017578125,
      "learning_rate": 9.965692661011066e-05,
      "loss": 2.0295,
      "step": 168832
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.2924572229385376,
      "learning_rate": 9.96538818941956e-05,
      "loss": 2.0144,
      "step": 168833
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.0918470621109009,
      "learning_rate": 9.965083721552851e-05,
      "loss": 1.9886,
      "step": 168834
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.3761882781982422,
      "learning_rate": 9.964779257411027e-05,
      "loss": 2.126,
      "step": 168835
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.2793793678283691,
      "learning_rate": 9.964474796994126e-05,
      "loss": 2.1632,
      "step": 168836
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.2153583765029907,
      "learning_rate": 9.964170340302198e-05,
      "loss": 2.0854,
      "step": 168837
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.336114525794983,
      "learning_rate": 9.963865887335317e-05,
      "loss": 2.2011,
      "step": 168838
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.2026894092559814,
      "learning_rate": 9.963561438093533e-05,
      "loss": 1.8951,
      "step": 168839
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.2556006908416748,
      "learning_rate": 9.96325699257689e-05,
      "loss": 2.1254,
      "step": 168840
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.2215059995651245,
      "learning_rate": 9.962952550785467e-05,
      "loss": 2.0942,
      "step": 168841
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.3982869386672974,
      "learning_rate": 9.96264811271931e-05,
      "loss": 2.0669,
      "step": 168842
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.2921186685562134,
      "learning_rate": 9.962343678378464e-05,
      "loss": 1.9294,
      "step": 168843
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.2653703689575195,
      "learning_rate": 9.962039247763008e-05,
      "loss": 2.0168,
      "step": 168844
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.2056471109390259,
      "learning_rate": 9.961734820872975e-05,
      "loss": 1.7963,
      "step": 168845
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.368133306503296,
      "learning_rate": 9.961430397708444e-05,
      "loss": 2.1332,
      "step": 168846
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.2214696407318115,
      "learning_rate": 9.96112597826946e-05,
      "loss": 2.0668,
      "step": 168847
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.1895411014556885,
      "learning_rate": 9.960821562556084e-05,
      "loss": 2.0024,
      "step": 168848
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.2227585315704346,
      "learning_rate": 9.96051715056836e-05,
      "loss": 2.2051,
      "step": 168849
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.2300761938095093,
      "learning_rate": 9.960212742306362e-05,
      "loss": 1.9462,
      "step": 168850
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.3193458318710327,
      "learning_rate": 9.959908337770129e-05,
      "loss": 1.9886,
      "step": 168851
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.3518693447113037,
      "learning_rate": 9.959603936959737e-05,
      "loss": 2.0429,
      "step": 168852
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.37654447555542,
      "learning_rate": 9.959299539875235e-05,
      "loss": 2.2493,
      "step": 168853
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.2374054193496704,
      "learning_rate": 9.958995146516676e-05,
      "loss": 2.2059,
      "step": 168854
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.3077924251556396,
      "learning_rate": 9.958690756884106e-05,
      "loss": 2.0031,
      "step": 168855
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.2716060876846313,
      "learning_rate": 9.958386370977606e-05,
      "loss": 2.016,
      "step": 168856
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.4095803499221802,
      "learning_rate": 9.958081988797214e-05,
      "loss": 1.886,
      "step": 168857
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.359207272529602,
      "learning_rate": 9.957777610342995e-05,
      "loss": 1.8497,
      "step": 168858
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.2287181615829468,
      "learning_rate": 9.957473235615e-05,
      "loss": 1.988,
      "step": 168859
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.183150291442871,
      "learning_rate": 9.957168864613304e-05,
      "loss": 2.0917,
      "step": 168860
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.3080335855484009,
      "learning_rate": 9.956864497337934e-05,
      "loss": 1.948,
      "step": 168861
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.3360021114349365,
      "learning_rate": 9.956560133788967e-05,
      "loss": 1.9583,
      "step": 168862
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.1255381107330322,
      "learning_rate": 9.956255773966445e-05,
      "loss": 2.0335,
      "step": 168863
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.2708567380905151,
      "learning_rate": 9.955951417870443e-05,
      "loss": 2.1937,
      "step": 168864
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.2206157445907593,
      "learning_rate": 9.955647065501e-05,
      "loss": 1.937,
      "step": 168865
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.2282661199569702,
      "learning_rate": 9.955342716858197e-05,
      "loss": 1.8875,
      "step": 168866
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.3538168668746948,
      "learning_rate": 9.955038371942058e-05,
      "loss": 2.0606,
      "step": 168867
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.1921449899673462,
      "learning_rate": 9.954734030752663e-05,
      "loss": 1.8736,
      "step": 168868
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.19607412815094,
      "learning_rate": 9.954429693290054e-05,
      "loss": 1.9873,
      "step": 168869
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.4722182750701904,
      "learning_rate": 9.954125359554303e-05,
      "loss": 2.2944,
      "step": 168870
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.2454261779785156,
      "learning_rate": 9.95382102954545e-05,
      "loss": 1.8924,
      "step": 168871
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.193631649017334,
      "learning_rate": 9.953516703263577e-05,
      "loss": 1.9628,
      "step": 168872
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.2027006149291992,
      "learning_rate": 9.953212380708705e-05,
      "loss": 1.9691,
      "step": 168873
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.15169358253479,
      "learning_rate": 9.952908061880921e-05,
      "loss": 2.0937,
      "step": 168874
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.2451106309890747,
      "learning_rate": 9.952603746780259e-05,
      "loss": 2.1125,
      "step": 168875
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.3993010520935059,
      "learning_rate": 9.952299435406796e-05,
      "loss": 2.2313,
      "step": 168876
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.1790313720703125,
      "learning_rate": 9.951995127760567e-05,
      "loss": 2.1042,
      "step": 168877
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.4636961221694946,
      "learning_rate": 9.951690823841663e-05,
      "loss": 2.0282,
      "step": 168878
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.2556004524230957,
      "learning_rate": 9.951386523650098e-05,
      "loss": 1.8917,
      "step": 168879
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.181814432144165,
      "learning_rate": 9.95108222718596e-05,
      "loss": 1.8202,
      "step": 168880
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.3271602392196655,
      "learning_rate": 9.950777934449281e-05,
      "loss": 2.1115,
      "step": 168881
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.2986249923706055,
      "learning_rate": 9.950473645440141e-05,
      "loss": 2.087,
      "step": 168882
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.1975557804107666,
      "learning_rate": 9.95016936015858e-05,
      "loss": 2.0,
      "step": 168883
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.334036946296692,
      "learning_rate": 9.949865078604669e-05,
      "loss": 2.1066,
      "step": 168884
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.1832141876220703,
      "learning_rate": 9.949560800778455e-05,
      "loss": 2.2627,
      "step": 168885
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.2651828527450562,
      "learning_rate": 9.949256526679998e-05,
      "loss": 1.9303,
      "step": 168886
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.3666411638259888,
      "learning_rate": 9.948952256309342e-05,
      "loss": 1.9962,
      "step": 168887
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.2483502626419067,
      "learning_rate": 9.948647989666564e-05,
      "loss": 1.9853,
      "step": 168888
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.262701392173767,
      "learning_rate": 9.9483437267517e-05,
      "loss": 2.173,
      "step": 168889
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.285294532775879,
      "learning_rate": 9.948039467564833e-05,
      "loss": 2.06,
      "step": 168890
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.2955421209335327,
      "learning_rate": 9.947735212105998e-05,
      "loss": 1.9323,
      "step": 168891
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.1730355024337769,
      "learning_rate": 9.947430960375259e-05,
      "loss": 1.8587,
      "step": 168892
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.1727691888809204,
      "learning_rate": 9.94712671237266e-05,
      "loss": 1.9954,
      "step": 168893
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.2206817865371704,
      "learning_rate": 9.94682246809828e-05,
      "loss": 1.8218,
      "step": 168894
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.288122534751892,
      "learning_rate": 9.946518227552156e-05,
      "loss": 2.0026,
      "step": 168895
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.1824673414230347,
      "learning_rate": 9.946213990734362e-05,
      "loss": 1.971,
      "step": 168896
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.1435750722885132,
      "learning_rate": 9.945909757644944e-05,
      "loss": 2.2074,
      "step": 168897
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.270937442779541,
      "learning_rate": 9.945605528283958e-05,
      "loss": 1.9999,
      "step": 168898
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.2578966617584229,
      "learning_rate": 9.945301302651454e-05,
      "loss": 2.0549,
      "step": 168899
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.473497748374939,
      "learning_rate": 9.94499708074751e-05,
      "loss": 1.9857,
      "step": 168900
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.3287338018417358,
      "learning_rate": 9.944692862572159e-05,
      "loss": 2.1068,
      "step": 168901
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.2989641427993774,
      "learning_rate": 9.944388648125475e-05,
      "loss": 2.1535,
      "step": 168902
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.2098116874694824,
      "learning_rate": 9.944084437407509e-05,
      "loss": 2.035,
      "step": 168903
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.2271764278411865,
      "learning_rate": 9.943780230418316e-05,
      "loss": 1.988,
      "step": 168904
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.2328764200210571,
      "learning_rate": 9.943476027157945e-05,
      "loss": 1.851,
      "step": 168905
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.2471966743469238,
      "learning_rate": 9.943171827626467e-05,
      "loss": 2.0431,
      "step": 168906
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.2082712650299072,
      "learning_rate": 9.942867631823927e-05,
      "loss": 2.1746,
      "step": 168907
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.3657219409942627,
      "learning_rate": 9.942563439750393e-05,
      "loss": 2.1178,
      "step": 168908
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.5572428703308105,
      "learning_rate": 9.942259251405914e-05,
      "loss": 2.0478,
      "step": 168909
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.3853861093521118,
      "learning_rate": 9.94195506679055e-05,
      "loss": 1.9892,
      "step": 168910
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.3706343173980713,
      "learning_rate": 9.941650885904344e-05,
      "loss": 2.0864,
      "step": 168911
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.307286262512207,
      "learning_rate": 9.941346708747373e-05,
      "loss": 2.0649,
      "step": 168912
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.267634391784668,
      "learning_rate": 9.941042535319677e-05,
      "loss": 2.3303,
      "step": 168913
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.5129871368408203,
      "learning_rate": 9.940738365621328e-05,
      "loss": 1.9993,
      "step": 168914
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.390584111213684,
      "learning_rate": 9.940434199652372e-05,
      "loss": 2.0048,
      "step": 168915
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.3468003273010254,
      "learning_rate": 9.94013003741287e-05,
      "loss": 1.8312,
      "step": 168916
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.373095989227295,
      "learning_rate": 9.93982587890287e-05,
      "loss": 1.9754,
      "step": 168917
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.1468313932418823,
      "learning_rate": 9.93952172412244e-05,
      "loss": 1.7799,
      "step": 168918
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.513927698135376,
      "learning_rate": 9.939217573071623e-05,
      "loss": 2.1946,
      "step": 168919
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.3614567518234253,
      "learning_rate": 9.938913425750494e-05,
      "loss": 2.013,
      "step": 168920
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.501265525817871,
      "learning_rate": 9.938609282159102e-05,
      "loss": 1.8938,
      "step": 168921
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.281266689300537,
      "learning_rate": 9.938305142297491e-05,
      "loss": 1.9803,
      "step": 168922
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.2659599781036377,
      "learning_rate": 9.938001006165735e-05,
      "loss": 2.0971,
      "step": 168923
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.4187678098678589,
      "learning_rate": 9.937696873763885e-05,
      "loss": 1.8729,
      "step": 168924
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.231677532196045,
      "learning_rate": 9.937392745091985e-05,
      "loss": 1.8457,
      "step": 168925
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.4060616493225098,
      "learning_rate": 9.937088620150112e-05,
      "loss": 1.9764,
      "step": 168926
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.274780511856079,
      "learning_rate": 9.936784498938315e-05,
      "loss": 2.1133,
      "step": 168927
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.216569423675537,
      "learning_rate": 9.936480381456637e-05,
      "loss": 1.9153,
      "step": 168928
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.3375937938690186,
      "learning_rate": 9.936176267705156e-05,
      "loss": 1.9773,
      "step": 168929
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.4007683992385864,
      "learning_rate": 9.935872157683911e-05,
      "loss": 2.0241,
      "step": 168930
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.5489195585250854,
      "learning_rate": 9.935568051392971e-05,
      "loss": 1.9403,
      "step": 168931
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.3422693014144897,
      "learning_rate": 9.935263948832391e-05,
      "loss": 2.4197,
      "step": 168932
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.298936367034912,
      "learning_rate": 9.934959850002225e-05,
      "loss": 2.103,
      "step": 168933
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.584492564201355,
      "learning_rate": 9.934655754902516e-05,
      "loss": 1.9158,
      "step": 168934
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.2705965042114258,
      "learning_rate": 9.934351663533347e-05,
      "loss": 1.899,
      "step": 168935
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.3195173740386963,
      "learning_rate": 9.934047575894748e-05,
      "loss": 1.778,
      "step": 168936
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.239922046661377,
      "learning_rate": 9.933743491986797e-05,
      "loss": 1.9258,
      "step": 168937
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.3296109437942505,
      "learning_rate": 9.933439411809547e-05,
      "loss": 1.9419,
      "step": 168938
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.2561213970184326,
      "learning_rate": 9.933135335363044e-05,
      "loss": 2.2626,
      "step": 168939
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.5294121503829956,
      "learning_rate": 9.932831262647344e-05,
      "loss": 1.9188,
      "step": 168940
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.3863359689712524,
      "learning_rate": 9.932527193662516e-05,
      "loss": 2.1671,
      "step": 168941
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.3667149543762207,
      "learning_rate": 9.932223128408601e-05,
      "loss": 2.2064,
      "step": 168942
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.290285587310791,
      "learning_rate": 9.931919066885678e-05,
      "loss": 1.9558,
      "step": 168943
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.2984907627105713,
      "learning_rate": 9.931615009093776e-05,
      "loss": 2.0914,
      "step": 168944
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.3068695068359375,
      "learning_rate": 9.931310955032987e-05,
      "loss": 2.1574,
      "step": 168945
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.4300637245178223,
      "learning_rate": 9.931006904703324e-05,
      "loss": 1.8563,
      "step": 168946
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.4828534126281738,
      "learning_rate": 9.930702858104879e-05,
      "loss": 1.974,
      "step": 168947
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.3433371782302856,
      "learning_rate": 9.930398815237683e-05,
      "loss": 2.0007,
      "step": 168948
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.3439241647720337,
      "learning_rate": 9.930094776101815e-05,
      "loss": 1.9688,
      "step": 168949
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.4406343698501587,
      "learning_rate": 9.929790740697312e-05,
      "loss": 2.0287,
      "step": 168950
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.3739216327667236,
      "learning_rate": 9.929486709024252e-05,
      "loss": 2.089,
      "step": 168951
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.4250046014785767,
      "learning_rate": 9.929182681082676e-05,
      "loss": 2.0271,
      "step": 168952
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.3084579706192017,
      "learning_rate": 9.928878656872641e-05,
      "loss": 2.1261,
      "step": 168953
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.2588081359863281,
      "learning_rate": 9.928574636394203e-05,
      "loss": 2.2101,
      "step": 168954
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.308896780014038,
      "learning_rate": 9.928270619647427e-05,
      "loss": 1.8093,
      "step": 168955
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.3824445009231567,
      "learning_rate": 9.927966606632357e-05,
      "loss": 1.8561,
      "step": 168956
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.31108820438385,
      "learning_rate": 9.927662597349067e-05,
      "loss": 1.8983,
      "step": 168957
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.31541907787323,
      "learning_rate": 9.927358591797601e-05,
      "loss": 1.9741,
      "step": 168958
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.3451029062271118,
      "learning_rate": 9.927054589978019e-05,
      "loss": 1.9418,
      "step": 168959
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.4300850629806519,
      "learning_rate": 9.926750591890366e-05,
      "loss": 2.0189,
      "step": 168960
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.2383447885513306,
      "learning_rate": 9.926446597534722e-05,
      "loss": 2.0023,
      "step": 168961
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.6710044145584106,
      "learning_rate": 9.926142606911118e-05,
      "loss": 2.2261,
      "step": 168962
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.289299726486206,
      "learning_rate": 9.925838620019633e-05,
      "loss": 2.1046,
      "step": 168963
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.2894091606140137,
      "learning_rate": 9.925534636860312e-05,
      "loss": 1.9989,
      "step": 168964
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.294852375984192,
      "learning_rate": 9.925230657433216e-05,
      "loss": 2.1316,
      "step": 168965
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.2582275867462158,
      "learning_rate": 9.924926681738386e-05,
      "loss": 1.789,
      "step": 168966
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.3233758211135864,
      "learning_rate": 9.924622709775902e-05,
      "loss": 1.9844,
      "step": 168967
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.4132989645004272,
      "learning_rate": 9.924318741545803e-05,
      "loss": 1.9326,
      "step": 168968
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.321648359298706,
      "learning_rate": 9.924014777048157e-05,
      "loss": 1.8692,
      "step": 168969
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.3148291110992432,
      "learning_rate": 9.923710816283019e-05,
      "loss": 1.9121,
      "step": 168970
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.3121017217636108,
      "learning_rate": 9.923406859250439e-05,
      "loss": 2.3562,
      "step": 168971
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.1825627088546753,
      "learning_rate": 9.923102905950469e-05,
      "loss": 1.9828,
      "step": 168972
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.3253308534622192,
      "learning_rate": 9.922798956383184e-05,
      "loss": 2.1061,
      "step": 168973
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.4126938581466675,
      "learning_rate": 9.922495010548617e-05,
      "loss": 2.0508,
      "step": 168974
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.2791465520858765,
      "learning_rate": 9.92219106844685e-05,
      "loss": 1.9372,
      "step": 168975
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.2101043462753296,
      "learning_rate": 9.921887130077923e-05,
      "loss": 2.3098,
      "step": 168976
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.293144941329956,
      "learning_rate": 9.921583195441902e-05,
      "loss": 2.0726,
      "step": 168977
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.2662255764007568,
      "learning_rate": 9.921279264538823e-05,
      "loss": 2.0672,
      "step": 168978
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.1918073892593384,
      "learning_rate": 9.920975337368769e-05,
      "loss": 2.0891,
      "step": 168979
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.257799744606018,
      "learning_rate": 9.920671413931771e-05,
      "loss": 2.1529,
      "step": 168980
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.1760106086730957,
      "learning_rate": 9.920367494227914e-05,
      "loss": 1.9723,
      "step": 168981
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.345467448234558,
      "learning_rate": 9.920063578257238e-05,
      "loss": 1.919,
      "step": 168982
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.1314082145690918,
      "learning_rate": 9.919759666019802e-05,
      "loss": 2.0252,
      "step": 168983
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.1837786436080933,
      "learning_rate": 9.91945575751565e-05,
      "loss": 2.0613,
      "step": 168984
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.2219009399414062,
      "learning_rate": 9.91915185274486e-05,
      "loss": 2.0499,
      "step": 168985
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.204296350479126,
      "learning_rate": 9.918847951707469e-05,
      "loss": 2.2321,
      "step": 168986
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.290530800819397,
      "learning_rate": 9.918544054403554e-05,
      "loss": 2.1278,
      "step": 168987
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.1715296506881714,
      "learning_rate": 9.918240160833163e-05,
      "loss": 1.916,
      "step": 168988
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.316856861114502,
      "learning_rate": 9.917936270996348e-05,
      "loss": 2.2201,
      "step": 168989
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.223860740661621,
      "learning_rate": 9.917632384893157e-05,
      "loss": 1.8902,
      "step": 168990
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.5562186241149902,
      "learning_rate": 9.91732850252367e-05,
      "loss": 2.0378,
      "step": 168991
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.5855673551559448,
      "learning_rate": 9.917024623887918e-05,
      "loss": 1.8559,
      "step": 168992
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.3572804927825928,
      "learning_rate": 9.91672074898598e-05,
      "loss": 2.0168,
      "step": 168993
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.6409035921096802,
      "learning_rate": 9.916416877817902e-05,
      "loss": 2.0224,
      "step": 168994
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.296718955039978,
      "learning_rate": 9.916113010383743e-05,
      "loss": 2.0525,
      "step": 168995
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.2898002862930298,
      "learning_rate": 9.915809146683549e-05,
      "loss": 1.9084,
      "step": 168996
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.3321480751037598,
      "learning_rate": 9.915505286717393e-05,
      "loss": 2.0197,
      "step": 168997
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.4152053594589233,
      "learning_rate": 9.915201430485315e-05,
      "loss": 1.9128,
      "step": 168998
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.4463765621185303,
      "learning_rate": 9.914897577987389e-05,
      "loss": 2.1494,
      "step": 168999
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.5808141231536865,
      "learning_rate": 9.914593729223662e-05,
      "loss": 1.9367,
      "step": 169000
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.320494294166565,
      "learning_rate": 9.914289884194192e-05,
      "loss": 2.1603,
      "step": 169001
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.24046790599823,
      "learning_rate": 9.913986042899028e-05,
      "loss": 1.833,
      "step": 169002
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.4541289806365967,
      "learning_rate": 9.91368220533824e-05,
      "loss": 2.1622,
      "step": 169003
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.3565970659255981,
      "learning_rate": 9.913378371511865e-05,
      "loss": 2.1526,
      "step": 169004
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.1138916015625,
      "learning_rate": 9.913074541419987e-05,
      "loss": 1.817,
      "step": 169005
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.4475693702697754,
      "learning_rate": 9.912770715062647e-05,
      "loss": 2.0447,
      "step": 169006
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.1981805562973022,
      "learning_rate": 9.91246689243989e-05,
      "loss": 2.0204,
      "step": 169007
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.2113533020019531,
      "learning_rate": 9.912163073551795e-05,
      "loss": 1.7082,
      "step": 169008
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.3027970790863037,
      "learning_rate": 9.911859258398408e-05,
      "loss": 2.3355,
      "step": 169009
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.5217324495315552,
      "learning_rate": 9.911555446979774e-05,
      "loss": 2.168,
      "step": 169010
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.2392159700393677,
      "learning_rate": 9.911251639295973e-05,
      "loss": 1.9655,
      "step": 169011
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.1995351314544678,
      "learning_rate": 9.91094783534705e-05,
      "loss": 1.8876,
      "step": 169012
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.1421400308609009,
      "learning_rate": 9.910644035133051e-05,
      "loss": 1.8028,
      "step": 169013
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.1924806833267212,
      "learning_rate": 9.910340238654051e-05,
      "loss": 2.044,
      "step": 169014
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.2320752143859863,
      "learning_rate": 9.9100364459101e-05,
      "loss": 2.1031,
      "step": 169015
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.2480143308639526,
      "learning_rate": 9.909732656901241e-05,
      "loss": 2.0813,
      "step": 169016
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.1746773719787598,
      "learning_rate": 9.909428871627552e-05,
      "loss": 2.0442,
      "step": 169017
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.2747927904129028,
      "learning_rate": 9.909125090089071e-05,
      "loss": 2.0416,
      "step": 169018
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.2173981666564941,
      "learning_rate": 9.908821312285871e-05,
      "loss": 2.0482,
      "step": 169019
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.4555989503860474,
      "learning_rate": 9.908517538218002e-05,
      "loss": 1.909,
      "step": 169020
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.2044860124588013,
      "learning_rate": 9.908213767885506e-05,
      "loss": 1.7926,
      "step": 169021
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.2926256656646729,
      "learning_rate": 9.907910001288467e-05,
      "loss": 2.0715,
      "step": 169022
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.1505203247070312,
      "learning_rate": 9.907606238426923e-05,
      "loss": 2.1052,
      "step": 169023
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.2188204526901245,
      "learning_rate": 9.907302479300923e-05,
      "loss": 2.2171,
      "step": 169024
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.119231104850769,
      "learning_rate": 9.90699872391055e-05,
      "loss": 2.1003,
      "step": 169025
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.2003896236419678,
      "learning_rate": 9.906694972255841e-05,
      "loss": 2.0234,
      "step": 169026
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.2893376350402832,
      "learning_rate": 9.906391224336847e-05,
      "loss": 2.0072,
      "step": 169027
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.255055546760559,
      "learning_rate": 9.906087480153644e-05,
      "loss": 2.0192,
      "step": 169028
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.2018754482269287,
      "learning_rate": 9.905783739706271e-05,
      "loss": 1.9848,
      "step": 169029
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.138411521911621,
      "learning_rate": 9.905480002994803e-05,
      "loss": 2.0192,
      "step": 169030
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.2924141883850098,
      "learning_rate": 9.905176270019282e-05,
      "loss": 1.9064,
      "step": 169031
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.2076897621154785,
      "learning_rate": 9.904872540779769e-05,
      "loss": 2.0916,
      "step": 169032
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.2045044898986816,
      "learning_rate": 9.904568815276311e-05,
      "loss": 2.053,
      "step": 169033
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.1926977634429932,
      "learning_rate": 9.904265093508982e-05,
      "loss": 1.8598,
      "step": 169034
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.3044649362564087,
      "learning_rate": 9.90396137547782e-05,
      "loss": 2.0299,
      "step": 169035
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.2286194562911987,
      "learning_rate": 9.9036576611829e-05,
      "loss": 2.0368,
      "step": 169036
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.2920368909835815,
      "learning_rate": 9.903353950624267e-05,
      "loss": 2.0123,
      "step": 169037
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.2287384271621704,
      "learning_rate": 9.903050243801984e-05,
      "loss": 2.1106,
      "step": 169038
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.3064067363739014,
      "learning_rate": 9.902746540716093e-05,
      "loss": 2.2012,
      "step": 169039
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.186415672302246,
      "learning_rate": 9.902442841366668e-05,
      "loss": 2.1353,
      "step": 169040
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.3039216995239258,
      "learning_rate": 9.90213914575375e-05,
      "loss": 2.0086,
      "step": 169041
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.2454299926757812,
      "learning_rate": 9.901835453877413e-05,
      "loss": 2.1029,
      "step": 169042
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.2225347757339478,
      "learning_rate": 9.901531765737705e-05,
      "loss": 1.9501,
      "step": 169043
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.259851098060608,
      "learning_rate": 9.901228081334684e-05,
      "loss": 2.0059,
      "step": 169044
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.2199169397354126,
      "learning_rate": 9.900924400668389e-05,
      "loss": 2.1811,
      "step": 169045
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.1999390125274658,
      "learning_rate": 9.900620723738907e-05,
      "loss": 2.015,
      "step": 169046
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.1831260919570923,
      "learning_rate": 9.900317050546264e-05,
      "loss": 1.994,
      "step": 169047
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.279819369316101,
      "learning_rate": 9.900013381090544e-05,
      "loss": 2.1229,
      "step": 169048
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.3588508367538452,
      "learning_rate": 9.899709715371789e-05,
      "loss": 2.2473,
      "step": 169049
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.4598021507263184,
      "learning_rate": 9.89940605339006e-05,
      "loss": 2.1237,
      "step": 169050
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.1761565208435059,
      "learning_rate": 9.899102395145401e-05,
      "loss": 2.0309,
      "step": 169051
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.3267408609390259,
      "learning_rate": 9.898798740637885e-05,
      "loss": 1.9529,
      "step": 169052
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.2302316427230835,
      "learning_rate": 9.898495089867554e-05,
      "loss": 1.9577,
      "step": 169053
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.2706760168075562,
      "learning_rate": 9.898191442834483e-05,
      "loss": 1.7668,
      "step": 169054
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.4322599172592163,
      "learning_rate": 9.897887799538717e-05,
      "loss": 2.2215,
      "step": 169055
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.230145812034607,
      "learning_rate": 9.897584159980313e-05,
      "loss": 1.9622,
      "step": 169056
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.1255286931991577,
      "learning_rate": 9.897280524159317e-05,
      "loss": 2.12,
      "step": 169057
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.194116473197937,
      "learning_rate": 9.896976892075807e-05,
      "loss": 2.0695,
      "step": 169058
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.130985975265503,
      "learning_rate": 9.896673263729815e-05,
      "loss": 2.0817,
      "step": 169059
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.2803746461868286,
      "learning_rate": 9.896369639121426e-05,
      "loss": 1.9932,
      "step": 169060
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.1711642742156982,
      "learning_rate": 9.896066018250678e-05,
      "loss": 2.0007,
      "step": 169061
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.4382600784301758,
      "learning_rate": 9.895762401117633e-05,
      "loss": 2.0707,
      "step": 169062
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.326385736465454,
      "learning_rate": 9.895458787722334e-05,
      "loss": 2.0022,
      "step": 169063
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.2754523754119873,
      "learning_rate": 9.895155178064859e-05,
      "loss": 2.1479,
      "step": 169064
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.1537703275680542,
      "learning_rate": 9.894851572145244e-05,
      "loss": 2.1676,
      "step": 169065
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.333943247795105,
      "learning_rate": 9.894547969963567e-05,
      "loss": 1.563,
      "step": 169066
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.3509254455566406,
      "learning_rate": 9.894244371519871e-05,
      "loss": 2.06,
      "step": 169067
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.3234974145889282,
      "learning_rate": 9.893940776814214e-05,
      "loss": 2.0018,
      "step": 169068
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.2058570384979248,
      "learning_rate": 9.893637185846644e-05,
      "loss": 1.8611,
      "step": 169069
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.3433879613876343,
      "learning_rate": 9.893333598617235e-05,
      "loss": 1.9149,
      "step": 169070
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.259344458580017,
      "learning_rate": 9.89303001512603e-05,
      "loss": 1.9977,
      "step": 169071
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.3738638162612915,
      "learning_rate": 9.892726435373095e-05,
      "loss": 2.0831,
      "step": 169072
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.1661041975021362,
      "learning_rate": 9.892422859358485e-05,
      "loss": 2.1221,
      "step": 169073
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.2278549671173096,
      "learning_rate": 9.892119287082251e-05,
      "loss": 2.1144,
      "step": 169074
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.2763530015945435,
      "learning_rate": 9.89181571854444e-05,
      "loss": 2.0938,
      "step": 169075
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.2945873737335205,
      "learning_rate": 9.891512153745132e-05,
      "loss": 2.1367,
      "step": 169076
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.303181529045105,
      "learning_rate": 9.891208592684365e-05,
      "loss": 2.0644,
      "step": 169077
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.325111746788025,
      "learning_rate": 9.890905035362205e-05,
      "loss": 2.0404,
      "step": 169078
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.1910701990127563,
      "learning_rate": 9.89060148177871e-05,
      "loss": 1.8841,
      "step": 169079
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.2608568668365479,
      "learning_rate": 9.89029793193393e-05,
      "loss": 2.1734,
      "step": 169080
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.152805209159851,
      "learning_rate": 9.889994385827915e-05,
      "loss": 1.9899,
      "step": 169081
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.3075573444366455,
      "learning_rate": 9.889690843460737e-05,
      "loss": 2.1225,
      "step": 169082
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.4870247840881348,
      "learning_rate": 9.889387304832438e-05,
      "loss": 1.9848,
      "step": 169083
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.3059953451156616,
      "learning_rate": 9.889083769943091e-05,
      "loss": 2.0336,
      "step": 169084
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.3025023937225342,
      "learning_rate": 9.88878023879273e-05,
      "loss": 1.9356,
      "step": 169085
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.2404910326004028,
      "learning_rate": 9.888476711381446e-05,
      "loss": 2.1126,
      "step": 169086
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.3063158988952637,
      "learning_rate": 9.888173187709256e-05,
      "loss": 1.9581,
      "step": 169087
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.3185709714889526,
      "learning_rate": 9.887869667776242e-05,
      "loss": 1.9624,
      "step": 169088
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.2753512859344482,
      "learning_rate": 9.887566151582445e-05,
      "loss": 2.1921,
      "step": 169089
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.318650245666504,
      "learning_rate": 9.887262639127935e-05,
      "loss": 2.1004,
      "step": 169090
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.2643331289291382,
      "learning_rate": 9.886959130412755e-05,
      "loss": 1.9835,
      "step": 169091
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.1974512338638306,
      "learning_rate": 9.88665562543698e-05,
      "loss": 2.1132,
      "step": 169092
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.1999614238739014,
      "learning_rate": 9.886352124200654e-05,
      "loss": 2.0695,
      "step": 169093
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.2139637470245361,
      "learning_rate": 9.886048626703833e-05,
      "loss": 2.001,
      "step": 169094
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.4016987085342407,
      "learning_rate": 9.885745132946568e-05,
      "loss": 2.1465,
      "step": 169095
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.2095719575881958,
      "learning_rate": 9.885441642928931e-05,
      "loss": 2.0526,
      "step": 169096
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.3424357175827026,
      "learning_rate": 9.885138156650962e-05,
      "loss": 2.0435,
      "step": 169097
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.320546269416809,
      "learning_rate": 9.884834674112736e-05,
      "loss": 2.1644,
      "step": 169098
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.3394036293029785,
      "learning_rate": 9.884531195314294e-05,
      "loss": 1.9938,
      "step": 169099
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.2686799764633179,
      "learning_rate": 9.884227720255702e-05,
      "loss": 1.9907,
      "step": 169100
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.3144540786743164,
      "learning_rate": 9.883924248937001e-05,
      "loss": 2.0907,
      "step": 169101
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.2672388553619385,
      "learning_rate": 9.883620781358268e-05,
      "loss": 2.1803,
      "step": 169102
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.1783759593963623,
      "learning_rate": 9.88331731751954e-05,
      "loss": 2.2565,
      "step": 169103
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.1910597085952759,
      "learning_rate": 9.883013857420891e-05,
      "loss": 2.1278,
      "step": 169104
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.418986201286316,
      "learning_rate": 9.882710401062373e-05,
      "loss": 1.9089,
      "step": 169105
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.2708460092544556,
      "learning_rate": 9.882406948444026e-05,
      "loss": 1.9705,
      "step": 169106
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.3898887634277344,
      "learning_rate": 9.882103499565928e-05,
      "loss": 2.1564,
      "step": 169107
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.2209283113479614,
      "learning_rate": 9.881800054428131e-05,
      "loss": 1.9514,
      "step": 169108
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.326943039894104,
      "learning_rate": 9.881496613030674e-05,
      "loss": 1.9117,
      "step": 169109
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.2529547214508057,
      "learning_rate": 9.881193175373637e-05,
      "loss": 2.1182,
      "step": 169110
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.3537226915359497,
      "learning_rate": 9.880889741457068e-05,
      "loss": 2.1278,
      "step": 169111
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.132611870765686,
      "learning_rate": 9.880586311281009e-05,
      "loss": 2.0179,
      "step": 169112
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.267456293106079,
      "learning_rate": 9.880282884845539e-05,
      "loss": 2.0061,
      "step": 169113
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.3516948223114014,
      "learning_rate": 9.879979462150695e-05,
      "loss": 2.329,
      "step": 169114
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.1456489562988281,
      "learning_rate": 9.879676043196553e-05,
      "loss": 2.0472,
      "step": 169115
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.4203969240188599,
      "learning_rate": 9.879372627983156e-05,
      "loss": 2.0679,
      "step": 169116
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.1571977138519287,
      "learning_rate": 9.879069216510566e-05,
      "loss": 1.9665,
      "step": 169117
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.3030123710632324,
      "learning_rate": 9.878765808778828e-05,
      "loss": 1.7976,
      "step": 169118
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.305688738822937,
      "learning_rate": 9.878462404788017e-05,
      "loss": 2.3837,
      "step": 169119
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.23833429813385,
      "learning_rate": 9.878159004538167e-05,
      "loss": 1.8797,
      "step": 169120
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.234220027923584,
      "learning_rate": 9.87785560802936e-05,
      "loss": 2.3079,
      "step": 169121
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.3676059246063232,
      "learning_rate": 9.877552215261638e-05,
      "loss": 2.0625,
      "step": 169122
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.1943633556365967,
      "learning_rate": 9.877248826235056e-05,
      "loss": 2.3529,
      "step": 169123
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.214343547821045,
      "learning_rate": 9.87694544094967e-05,
      "loss": 1.9289,
      "step": 169124
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.2449584007263184,
      "learning_rate": 9.876642059405546e-05,
      "loss": 1.8376,
      "step": 169125
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.272111415863037,
      "learning_rate": 9.876338681602724e-05,
      "loss": 1.926,
      "step": 169126
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.2256383895874023,
      "learning_rate": 9.876035307541278e-05,
      "loss": 1.9823,
      "step": 169127
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.1996219158172607,
      "learning_rate": 9.87573193722126e-05,
      "loss": 1.9357,
      "step": 169128
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.3675808906555176,
      "learning_rate": 9.875428570642725e-05,
      "loss": 2.0917,
      "step": 169129
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.1898925304412842,
      "learning_rate": 9.875125207805714e-05,
      "loss": 1.9927,
      "step": 169130
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.211897373199463,
      "learning_rate": 9.874821848710311e-05,
      "loss": 1.9741,
      "step": 169131
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.138098955154419,
      "learning_rate": 9.874518493356544e-05,
      "loss": 1.9236,
      "step": 169132
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.7139427661895752,
      "learning_rate": 9.874215141744497e-05,
      "loss": 1.9287,
      "step": 169133
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.1550188064575195,
      "learning_rate": 9.873911793874211e-05,
      "loss": 1.8132,
      "step": 169134
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.2609542608261108,
      "learning_rate": 9.873608449745748e-05,
      "loss": 1.9056,
      "step": 169135
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.166911005973816,
      "learning_rate": 9.873305109359148e-05,
      "loss": 1.945,
      "step": 169136
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.2892615795135498,
      "learning_rate": 9.873001772714493e-05,
      "loss": 1.9761,
      "step": 169137
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.129868507385254,
      "learning_rate": 9.872698439811813e-05,
      "loss": 2.0423,
      "step": 169138
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.2850568294525146,
      "learning_rate": 9.872395110651192e-05,
      "loss": 2.0333,
      "step": 169139
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.2920976877212524,
      "learning_rate": 9.872091785232673e-05,
      "loss": 1.7921,
      "step": 169140
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.1963460445404053,
      "learning_rate": 9.87178846355631e-05,
      "loss": 1.8779,
      "step": 169141
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.2405236959457397,
      "learning_rate": 9.87148514562215e-05,
      "loss": 2.2217,
      "step": 169142
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.247808575630188,
      "learning_rate": 9.871181831430274e-05,
      "loss": 2.1671,
      "step": 169143
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.2811179161071777,
      "learning_rate": 9.870878520980715e-05,
      "loss": 2.0873,
      "step": 169144
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.2006677389144897,
      "learning_rate": 9.870575214273548e-05,
      "loss": 2.083,
      "step": 169145
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.239635944366455,
      "learning_rate": 9.870271911308824e-05,
      "loss": 2.039,
      "step": 169146
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.2215094566345215,
      "learning_rate": 9.869968612086593e-05,
      "loss": 1.9745,
      "step": 169147
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.3714381456375122,
      "learning_rate": 9.869665316606905e-05,
      "loss": 1.8483,
      "step": 169148
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.263767123222351,
      "learning_rate": 9.869362024869837e-05,
      "loss": 2.0438,
      "step": 169149
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.3430585861206055,
      "learning_rate": 9.869058736875425e-05,
      "loss": 2.0438,
      "step": 169150
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.3970897197723389,
      "learning_rate": 9.868755452623745e-05,
      "loss": 2.0433,
      "step": 169151
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.2581697702407837,
      "learning_rate": 9.868452172114836e-05,
      "loss": 1.9603,
      "step": 169152
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.2356269359588623,
      "learning_rate": 9.868148895348779e-05,
      "loss": 1.9535,
      "step": 169153
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.163728952407837,
      "learning_rate": 9.86784562232559e-05,
      "loss": 1.9056,
      "step": 169154
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.2471566200256348,
      "learning_rate": 9.867542353045363e-05,
      "loss": 1.8846,
      "step": 169155
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.3381317853927612,
      "learning_rate": 9.867239087508129e-05,
      "loss": 2.1842,
      "step": 169156
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.2762798070907593,
      "learning_rate": 9.866935825713967e-05,
      "loss": 2.1076,
      "step": 169157
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.2382866144180298,
      "learning_rate": 9.866632567662909e-05,
      "loss": 1.9886,
      "step": 169158
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.2000353336334229,
      "learning_rate": 9.866329313355043e-05,
      "loss": 1.9038,
      "step": 169159
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.2421461343765259,
      "learning_rate": 9.866026062790387e-05,
      "loss": 1.8172,
      "step": 169160
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.464224934577942,
      "learning_rate": 9.86572281596903e-05,
      "loss": 2.2094,
      "step": 169161
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.4106944799423218,
      "learning_rate": 9.865419572891003e-05,
      "loss": 2.0705,
      "step": 169162
    },
    {
      "epoch": 2.2,
      "grad_norm": 2.104872465133667,
      "learning_rate": 9.865116333556383e-05,
      "loss": 2.0881,
      "step": 169163
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.2637566328048706,
      "learning_rate": 9.86481309796521e-05,
      "loss": 2.0299,
      "step": 169164
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.3497616052627563,
      "learning_rate": 9.864509866117568e-05,
      "loss": 2.1585,
      "step": 169165
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.4291497468948364,
      "learning_rate": 9.864206638013476e-05,
      "loss": 2.2759,
      "step": 169166
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.2312618494033813,
      "learning_rate": 9.863903413653015e-05,
      "loss": 1.9305,
      "step": 169167
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.261726975440979,
      "learning_rate": 9.863600193036222e-05,
      "loss": 2.0797,
      "step": 169168
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.2496784925460815,
      "learning_rate": 9.86329697616318e-05,
      "loss": 2.0934,
      "step": 169169
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.2838165760040283,
      "learning_rate": 9.862993763033922e-05,
      "loss": 1.8997,
      "step": 169170
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.2476799488067627,
      "learning_rate": 9.862690553648533e-05,
      "loss": 1.9844,
      "step": 169171
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.248594880104065,
      "learning_rate": 9.862387348007028e-05,
      "loss": 2.045,
      "step": 169172
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.1798382997512817,
      "learning_rate": 9.862084146109496e-05,
      "loss": 2.1107,
      "step": 169173
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.2928729057312012,
      "learning_rate": 9.861780947955971e-05,
      "loss": 2.0242,
      "step": 169174
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.3612152338027954,
      "learning_rate": 9.861477753546532e-05,
      "loss": 2.1196,
      "step": 169175
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.2923098802566528,
      "learning_rate": 9.861174562881216e-05,
      "loss": 2.0697,
      "step": 169176
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.3873202800750732,
      "learning_rate": 9.860871375960107e-05,
      "loss": 2.0104,
      "step": 169177
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.3308669328689575,
      "learning_rate": 9.860568192783223e-05,
      "loss": 1.878,
      "step": 169178
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.332202434539795,
      "learning_rate": 9.86026501335065e-05,
      "loss": 1.9613,
      "step": 169179
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.2649540901184082,
      "learning_rate": 9.859961837662422e-05,
      "loss": 1.8614,
      "step": 169180
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.2661523818969727,
      "learning_rate": 9.859658665718616e-05,
      "loss": 2.2718,
      "step": 169181
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.4079631567001343,
      "learning_rate": 9.859355497519271e-05,
      "loss": 1.7804,
      "step": 169182
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.3381543159484863,
      "learning_rate": 9.859052333064464e-05,
      "loss": 2.0317,
      "step": 169183
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.1983600854873657,
      "learning_rate": 9.858749172354239e-05,
      "loss": 2.0567,
      "step": 169184
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.3169530630111694,
      "learning_rate": 9.858446015388651e-05,
      "loss": 1.9353,
      "step": 169185
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.2879054546356201,
      "learning_rate": 9.858142862167748e-05,
      "loss": 2.0938,
      "step": 169186
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.2562402486801147,
      "learning_rate": 9.857839712691606e-05,
      "loss": 2.233,
      "step": 169187
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.3123778104782104,
      "learning_rate": 9.857536566960265e-05,
      "loss": 1.8709,
      "step": 169188
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.30484139919281,
      "learning_rate": 9.857233424973796e-05,
      "loss": 2.0115,
      "step": 169189
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.2610580921173096,
      "learning_rate": 9.856930286732247e-05,
      "loss": 1.9364,
      "step": 169190
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.2881457805633545,
      "learning_rate": 9.856627152235666e-05,
      "loss": 1.9762,
      "step": 169191
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.2912226915359497,
      "learning_rate": 9.856324021484126e-05,
      "loss": 2.4222,
      "step": 169192
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.288448691368103,
      "learning_rate": 9.856020894477677e-05,
      "loss": 2.1314,
      "step": 169193
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.2377839088439941,
      "learning_rate": 9.855717771216365e-05,
      "loss": 1.9687,
      "step": 169194
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.5962485074996948,
      "learning_rate": 9.855414651700265e-05,
      "loss": 2.1481,
      "step": 169195
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.3189884424209595,
      "learning_rate": 9.855111535929425e-05,
      "loss": 2.0273,
      "step": 169196
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.2636159658432007,
      "learning_rate": 9.85480842390389e-05,
      "loss": 2.1405,
      "step": 169197
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.2837159633636475,
      "learning_rate": 9.854505315623735e-05,
      "loss": 1.8625,
      "step": 169198
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.2305721044540405,
      "learning_rate": 9.854202211089006e-05,
      "loss": 2.0249,
      "step": 169199
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.297182559967041,
      "learning_rate": 9.853899110299752e-05,
      "loss": 2.1079,
      "step": 169200
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.3048902750015259,
      "learning_rate": 9.853596013256048e-05,
      "loss": 2.0218,
      "step": 169201
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.369798183441162,
      "learning_rate": 9.853292919957942e-05,
      "loss": 2.0582,
      "step": 169202
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.150739312171936,
      "learning_rate": 9.85298983040548e-05,
      "loss": 1.8861,
      "step": 169203
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.2967288494110107,
      "learning_rate": 9.852686744598734e-05,
      "loss": 2.2393,
      "step": 169204
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.4012470245361328,
      "learning_rate": 9.852383662537749e-05,
      "loss": 2.173,
      "step": 169205
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.2459361553192139,
      "learning_rate": 9.852080584222591e-05,
      "loss": 2.0017,
      "step": 169206
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.3071175813674927,
      "learning_rate": 9.851777509653317e-05,
      "loss": 1.9524,
      "step": 169207
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.201411247253418,
      "learning_rate": 9.851474438829972e-05,
      "loss": 1.9055,
      "step": 169208
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.4417303800582886,
      "learning_rate": 9.851171371752613e-05,
      "loss": 1.8541,
      "step": 169209
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.3122059106826782,
      "learning_rate": 9.850868308421311e-05,
      "loss": 1.9577,
      "step": 169210
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.2803895473480225,
      "learning_rate": 9.850565248836101e-05,
      "loss": 2.0506,
      "step": 169211
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.4802916049957275,
      "learning_rate": 9.85026219299706e-05,
      "loss": 2.0757,
      "step": 169212
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.187743902206421,
      "learning_rate": 9.84995914090424e-05,
      "loss": 2.0176,
      "step": 169213
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.2827943563461304,
      "learning_rate": 9.84965609255769e-05,
      "loss": 2.2565,
      "step": 169214
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.302795648574829,
      "learning_rate": 9.84935304795746e-05,
      "loss": 1.9769,
      "step": 169215
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.2370810508728027,
      "learning_rate": 9.849050007103624e-05,
      "loss": 2.0492,
      "step": 169216
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.307261347770691,
      "learning_rate": 9.848746969996224e-05,
      "loss": 2.0451,
      "step": 169217
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.3740408420562744,
      "learning_rate": 9.848443936635329e-05,
      "loss": 1.8795,
      "step": 169218
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.2695187330245972,
      "learning_rate": 9.848140907020979e-05,
      "loss": 2.0485,
      "step": 169219
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.1618763208389282,
      "learning_rate": 9.84783788115326e-05,
      "loss": 1.8788,
      "step": 169220
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.2268809080123901,
      "learning_rate": 9.847534859032187e-05,
      "loss": 2.0024,
      "step": 169221
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.2106951475143433,
      "learning_rate": 9.847231840657852e-05,
      "loss": 1.6668,
      "step": 169222
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.2562594413757324,
      "learning_rate": 9.846928826030283e-05,
      "loss": 1.9554,
      "step": 169223
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.329216480255127,
      "learning_rate": 9.846625815149563e-05,
      "loss": 2.2408,
      "step": 169224
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.378848910331726,
      "learning_rate": 9.846322808015725e-05,
      "loss": 2.2796,
      "step": 169225
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.2232530117034912,
      "learning_rate": 9.846019804628855e-05,
      "loss": 2.0263,
      "step": 169226
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.2354816198349,
      "learning_rate": 9.845716804988969e-05,
      "loss": 1.9875,
      "step": 169227
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.2182395458221436,
      "learning_rate": 9.845413809096156e-05,
      "loss": 1.7639,
      "step": 169228
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.57694673538208,
      "learning_rate": 9.845110816950453e-05,
      "loss": 1.9622,
      "step": 169229
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.4558837413787842,
      "learning_rate": 9.844807828551933e-05,
      "loss": 2.1064,
      "step": 169230
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.2388509511947632,
      "learning_rate": 9.844504843900634e-05,
      "loss": 2.003,
      "step": 169231
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.2853589057922363,
      "learning_rate": 9.844201862996643e-05,
      "loss": 1.7757,
      "step": 169232
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.2447704076766968,
      "learning_rate": 9.843898885839973e-05,
      "loss": 2.211,
      "step": 169233
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.4057847261428833,
      "learning_rate": 9.843595912430712e-05,
      "loss": 1.982,
      "step": 169234
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.2993812561035156,
      "learning_rate": 9.8432929427689e-05,
      "loss": 2.1294,
      "step": 169235
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.5936462879180908,
      "learning_rate": 9.84298997685461e-05,
      "loss": 1.885,
      "step": 169236
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.2316334247589111,
      "learning_rate": 9.842687014687877e-05,
      "loss": 2.18,
      "step": 169237
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.2533221244812012,
      "learning_rate": 9.842384056268788e-05,
      "loss": 1.8327,
      "step": 169238
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.5415246486663818,
      "learning_rate": 9.842081101597363e-05,
      "loss": 1.9594,
      "step": 169239
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.5151469707489014,
      "learning_rate": 9.841778150673685e-05,
      "loss": 2.2858,
      "step": 169240
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.475452184677124,
      "learning_rate": 9.84147520349779e-05,
      "loss": 2.0381,
      "step": 169241
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.1840286254882812,
      "learning_rate": 9.841172260069756e-05,
      "loss": 1.9658,
      "step": 169242
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.2697941064834595,
      "learning_rate": 9.840869320389618e-05,
      "loss": 2.2614,
      "step": 169243
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.3677918910980225,
      "learning_rate": 9.840566384457462e-05,
      "loss": 2.268,
      "step": 169244
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.2544870376586914,
      "learning_rate": 9.840263452273309e-05,
      "loss": 1.8416,
      "step": 169245
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.2485231161117554,
      "learning_rate": 9.839960523837239e-05,
      "loss": 2.1125,
      "step": 169246
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.4465380907058716,
      "learning_rate": 9.839657599149292e-05,
      "loss": 2.1074,
      "step": 169247
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.3364996910095215,
      "learning_rate": 9.83935467820954e-05,
      "loss": 1.9649,
      "step": 169248
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.355919361114502,
      "learning_rate": 9.839051761018026e-05,
      "loss": 2.0215,
      "step": 169249
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.7259883880615234,
      "learning_rate": 9.838748847574831e-05,
      "loss": 2.2157,
      "step": 169250
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.2711679935455322,
      "learning_rate": 9.838445937879974e-05,
      "loss": 2.222,
      "step": 169251
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.2327839136123657,
      "learning_rate": 9.838143031933539e-05,
      "loss": 2.3222,
      "step": 169252
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.2928560972213745,
      "learning_rate": 9.837840129735564e-05,
      "loss": 1.8277,
      "step": 169253
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.2777827978134155,
      "learning_rate": 9.837537231286129e-05,
      "loss": 2.014,
      "step": 169254
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.5258073806762695,
      "learning_rate": 9.837234336585265e-05,
      "loss": 1.9943,
      "step": 169255
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.1738499402999878,
      "learning_rate": 9.836931445633056e-05,
      "loss": 2.0054,
      "step": 169256
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.1272802352905273,
      "learning_rate": 9.836628558429524e-05,
      "loss": 2.1142,
      "step": 169257
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.1507165431976318,
      "learning_rate": 9.836325674974753e-05,
      "loss": 1.9854,
      "step": 169258
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.245642900466919,
      "learning_rate": 9.836022795268781e-05,
      "loss": 2.2751,
      "step": 169259
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.3008075952529907,
      "learning_rate": 9.835719919311683e-05,
      "loss": 1.961,
      "step": 169260
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.2451342344284058,
      "learning_rate": 9.835417047103497e-05,
      "loss": 1.919,
      "step": 169261
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.2082895040512085,
      "learning_rate": 9.835114178644304e-05,
      "loss": 1.9595,
      "step": 169262
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.250292420387268,
      "learning_rate": 9.834811313934124e-05,
      "loss": 1.9177,
      "step": 169263
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.2713159322738647,
      "learning_rate": 9.834508452973043e-05,
      "loss": 1.9325,
      "step": 169264
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.3917005062103271,
      "learning_rate": 9.834205595761101e-05,
      "loss": 1.9061,
      "step": 169265
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.1795426607131958,
      "learning_rate": 9.833902742298369e-05,
      "loss": 1.8866,
      "step": 169266
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.2003731727600098,
      "learning_rate": 9.833599892584886e-05,
      "loss": 1.9082,
      "step": 169267
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.2781178951263428,
      "learning_rate": 9.833297046620727e-05,
      "loss": 2.0121,
      "step": 169268
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.279667854309082,
      "learning_rate": 9.832994204405938e-05,
      "loss": 1.8914,
      "step": 169269
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.231223225593567,
      "learning_rate": 9.832691365940572e-05,
      "loss": 1.9209,
      "step": 169270
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.1446870565414429,
      "learning_rate": 9.832388531224685e-05,
      "loss": 2.0469,
      "step": 169271
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.317879319190979,
      "learning_rate": 9.832085700258349e-05,
      "loss": 2.1403,
      "step": 169272
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.313549518585205,
      "learning_rate": 9.831782873041596e-05,
      "loss": 2.1314,
      "step": 169273
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.2646983861923218,
      "learning_rate": 9.831480049574504e-05,
      "loss": 2.0715,
      "step": 169274
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.2917077541351318,
      "learning_rate": 9.831177229857122e-05,
      "loss": 2.3526,
      "step": 169275
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.281516671180725,
      "learning_rate": 9.830874413889495e-05,
      "loss": 1.9043,
      "step": 169276
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.2497645616531372,
      "learning_rate": 9.830571601671697e-05,
      "loss": 1.844,
      "step": 169277
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.221367359161377,
      "learning_rate": 9.83026879320378e-05,
      "loss": 2.0485,
      "step": 169278
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.312303900718689,
      "learning_rate": 9.829965988485787e-05,
      "loss": 2.0417,
      "step": 169279
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.2306002378463745,
      "learning_rate": 9.829663187517791e-05,
      "loss": 2.0034,
      "step": 169280
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.3434150218963623,
      "learning_rate": 9.829360390299843e-05,
      "loss": 2.0816,
      "step": 169281
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.3287404775619507,
      "learning_rate": 9.829057596831987e-05,
      "loss": 1.9633,
      "step": 169282
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.2897306680679321,
      "learning_rate": 9.828754807114301e-05,
      "loss": 1.9064,
      "step": 169283
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.2175066471099854,
      "learning_rate": 9.828452021146829e-05,
      "loss": 1.9491,
      "step": 169284
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.3207370042800903,
      "learning_rate": 9.82814923892962e-05,
      "loss": 2.1161,
      "step": 169285
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.3004010915756226,
      "learning_rate": 9.827846460462749e-05,
      "loss": 1.9217,
      "step": 169286
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.3566147089004517,
      "learning_rate": 9.82754368574626e-05,
      "loss": 2.035,
      "step": 169287
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.4205695390701294,
      "learning_rate": 9.827240914780201e-05,
      "loss": 1.9182,
      "step": 169288
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.2702240943908691,
      "learning_rate": 9.826938147564653e-05,
      "loss": 2.1416,
      "step": 169289
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.3519179821014404,
      "learning_rate": 9.826635384099643e-05,
      "loss": 1.9154,
      "step": 169290
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.2766344547271729,
      "learning_rate": 9.826332624385255e-05,
      "loss": 2.197,
      "step": 169291
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.2392573356628418,
      "learning_rate": 9.826029868421533e-05,
      "loss": 2.1162,
      "step": 169292
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.273827075958252,
      "learning_rate": 9.825727116208529e-05,
      "loss": 2.0916,
      "step": 169293
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.2267957925796509,
      "learning_rate": 9.825424367746296e-05,
      "loss": 1.9683,
      "step": 169294
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.441213846206665,
      "learning_rate": 9.825121623034909e-05,
      "loss": 2.1509,
      "step": 169295
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.3537036180496216,
      "learning_rate": 9.824818882074402e-05,
      "loss": 2.0924,
      "step": 169296
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.452087640762329,
      "learning_rate": 9.824516144864851e-05,
      "loss": 1.8982,
      "step": 169297
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.5604784488677979,
      "learning_rate": 9.824213411406303e-05,
      "loss": 2.1512,
      "step": 169298
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.3972276449203491,
      "learning_rate": 9.823910681698815e-05,
      "loss": 1.9805,
      "step": 169299
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.310486912727356,
      "learning_rate": 9.823607955742431e-05,
      "loss": 1.9979,
      "step": 169300
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.371522068977356,
      "learning_rate": 9.823305233537232e-05,
      "loss": 1.8704,
      "step": 169301
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.3916834592819214,
      "learning_rate": 9.823002515083252e-05,
      "loss": 2.1116,
      "step": 169302
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.2499186992645264,
      "learning_rate": 9.822699800380563e-05,
      "loss": 2.0365,
      "step": 169303
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.2372112274169922,
      "learning_rate": 9.822397089429208e-05,
      "loss": 2.1765,
      "step": 169304
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.349921703338623,
      "learning_rate": 9.822094382229272e-05,
      "loss": 2.116,
      "step": 169305
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.5079234838485718,
      "learning_rate": 9.821791678780766e-05,
      "loss": 2.0113,
      "step": 169306
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.246932029724121,
      "learning_rate": 9.82148897908378e-05,
      "loss": 2.1458,
      "step": 169307
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.2216299772262573,
      "learning_rate": 9.821186283138352e-05,
      "loss": 2.0025,
      "step": 169308
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.387977123260498,
      "learning_rate": 9.820883590944555e-05,
      "loss": 2.1108,
      "step": 169309
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.1747993230819702,
      "learning_rate": 9.820580902502427e-05,
      "loss": 1.9009,
      "step": 169310
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.1988364458084106,
      "learning_rate": 9.820278217812052e-05,
      "loss": 2.0711,
      "step": 169311
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.2370654344558716,
      "learning_rate": 9.819975536873452e-05,
      "loss": 1.7944,
      "step": 169312
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.342516303062439,
      "learning_rate": 9.819672859686708e-05,
      "loss": 1.9107,
      "step": 169313
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.1916221380233765,
      "learning_rate": 9.819370186251856e-05,
      "loss": 2.1451,
      "step": 169314
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.297698974609375,
      "learning_rate": 9.819067516568976e-05,
      "loss": 1.8577,
      "step": 169315
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.2539942264556885,
      "learning_rate": 9.818764850638102e-05,
      "loss": 1.9748,
      "step": 169316
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.279850959777832,
      "learning_rate": 9.818462188459319e-05,
      "loss": 2.1274,
      "step": 169317
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.4638370275497437,
      "learning_rate": 9.818159530032648e-05,
      "loss": 1.9183,
      "step": 169318
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.268197774887085,
      "learning_rate": 9.817856875358169e-05,
      "loss": 2.154,
      "step": 169319
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.191617488861084,
      "learning_rate": 9.817554224435922e-05,
      "loss": 1.9495,
      "step": 169320
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.266472578048706,
      "learning_rate": 9.817251577265983e-05,
      "loss": 2.1106,
      "step": 169321
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.3478339910507202,
      "learning_rate": 9.816948933848389e-05,
      "loss": 2.0344,
      "step": 169322
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.3638603687286377,
      "learning_rate": 9.816646294183224e-05,
      "loss": 2.1049,
      "step": 169323
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.2094515562057495,
      "learning_rate": 9.816343658270505e-05,
      "loss": 2.0285,
      "step": 169324
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.2948729991912842,
      "learning_rate": 9.816041026110319e-05,
      "loss": 1.9616,
      "step": 169325
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.4937092065811157,
      "learning_rate": 9.815738397702703e-05,
      "loss": 2.1569,
      "step": 169326
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.6096590757369995,
      "learning_rate": 9.81543577304773e-05,
      "loss": 2.1355,
      "step": 169327
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.2836525440216064,
      "learning_rate": 9.81513315214544e-05,
      "loss": 2.0005,
      "step": 169328
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.2356666326522827,
      "learning_rate": 9.814830534995918e-05,
      "loss": 1.9802,
      "step": 169329
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.234765648841858,
      "learning_rate": 9.81452792159918e-05,
      "loss": 2.068,
      "step": 169330
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.513190507888794,
      "learning_rate": 9.814225311955313e-05,
      "loss": 2.112,
      "step": 169331
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.2032114267349243,
      "learning_rate": 9.813922706064352e-05,
      "loss": 1.9857,
      "step": 169332
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.2132147550582886,
      "learning_rate": 9.813620103926371e-05,
      "loss": 1.9959,
      "step": 169333
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.2694941759109497,
      "learning_rate": 9.813317505541412e-05,
      "loss": 2.1189,
      "step": 169334
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.3193868398666382,
      "learning_rate": 9.813014910909562e-05,
      "loss": 1.9018,
      "step": 169335
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.2717981338500977,
      "learning_rate": 9.812712320030828e-05,
      "loss": 2.1676,
      "step": 169336
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.2702820301055908,
      "learning_rate": 9.812409732905302e-05,
      "loss": 2.1103,
      "step": 169337
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.1060702800750732,
      "learning_rate": 9.812107149533021e-05,
      "loss": 1.9744,
      "step": 169338
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.5204325914382935,
      "learning_rate": 9.811804569914064e-05,
      "loss": 1.8588,
      "step": 169339
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.2175740003585815,
      "learning_rate": 9.811501994048457e-05,
      "loss": 2.0181,
      "step": 169340
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.3456034660339355,
      "learning_rate": 9.811199421936299e-05,
      "loss": 1.8574,
      "step": 169341
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.2946947813034058,
      "learning_rate": 9.810896853577594e-05,
      "loss": 1.909,
      "step": 169342
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.1252493858337402,
      "learning_rate": 9.810594288972441e-05,
      "loss": 1.9564,
      "step": 169343
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.7411315441131592,
      "learning_rate": 9.810291728120865e-05,
      "loss": 2.0439,
      "step": 169344
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.2555019855499268,
      "learning_rate": 9.80998917102295e-05,
      "loss": 1.901,
      "step": 169345
    },
    {
      "epoch": 2.2,
      "grad_norm": 1.3408966064453125,
      "learning_rate": 9.809686617678724e-05,
      "loss": 2.158,
      "step": 169346
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.3468908071517944,
      "learning_rate": 9.809384068088284e-05,
      "loss": 2.2177,
      "step": 169347
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.2807698249816895,
      "learning_rate": 9.809081522251637e-05,
      "loss": 1.9778,
      "step": 169348
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.37259042263031,
      "learning_rate": 9.808778980168874e-05,
      "loss": 1.994,
      "step": 169349
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.5754246711730957,
      "learning_rate": 9.808476441840031e-05,
      "loss": 2.2117,
      "step": 169350
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.287693977355957,
      "learning_rate": 9.808173907265183e-05,
      "loss": 2.0964,
      "step": 169351
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.197519063949585,
      "learning_rate": 9.807871376444369e-05,
      "loss": 2.2762,
      "step": 169352
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.16605544090271,
      "learning_rate": 9.807568849377657e-05,
      "loss": 1.9191,
      "step": 169353
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.7578990459442139,
      "learning_rate": 9.807266326065102e-05,
      "loss": 1.7478,
      "step": 169354
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.5411958694458008,
      "learning_rate": 9.806963806506759e-05,
      "loss": 2.0352,
      "step": 169355
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.1652923822402954,
      "learning_rate": 9.806661290702672e-05,
      "loss": 1.976,
      "step": 169356
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.3215770721435547,
      "learning_rate": 9.806358778652918e-05,
      "loss": 2.2383,
      "step": 169357
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.2384508848190308,
      "learning_rate": 9.806056270357532e-05,
      "loss": 1.983,
      "step": 169358
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.354689598083496,
      "learning_rate": 9.805753765816591e-05,
      "loss": 2.215,
      "step": 169359
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.2083593606948853,
      "learning_rate": 9.805451265030142e-05,
      "loss": 2.1025,
      "step": 169360
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.304840326309204,
      "learning_rate": 9.805148767998242e-05,
      "loss": 1.9537,
      "step": 169361
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.2869952917099,
      "learning_rate": 9.804846274720936e-05,
      "loss": 2.0136,
      "step": 169362
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.157771110534668,
      "learning_rate": 9.804543785198298e-05,
      "loss": 2.0918,
      "step": 169363
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.3039571046829224,
      "learning_rate": 9.80424129943037e-05,
      "loss": 1.9446,
      "step": 169364
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.2797993421554565,
      "learning_rate": 9.803938817417222e-05,
      "loss": 1.9887,
      "step": 169365
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.2310181856155396,
      "learning_rate": 9.803636339158902e-05,
      "loss": 1.8044,
      "step": 169366
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.2008414268493652,
      "learning_rate": 9.803333864655461e-05,
      "loss": 1.9795,
      "step": 169367
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.389004111289978,
      "learning_rate": 9.803031393906965e-05,
      "loss": 1.9097,
      "step": 169368
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.4219924211502075,
      "learning_rate": 9.802728926913474e-05,
      "loss": 2.0553,
      "step": 169369
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.5973789691925049,
      "learning_rate": 9.802426463675023e-05,
      "loss": 2.1649,
      "step": 169370
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.5498942136764526,
      "learning_rate": 9.802124004191694e-05,
      "loss": 2.2076,
      "step": 169371
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.5833523273468018,
      "learning_rate": 9.801821548463531e-05,
      "loss": 1.9577,
      "step": 169372
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.2814064025878906,
      "learning_rate": 9.801519096490581e-05,
      "loss": 2.0416,
      "step": 169373
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.6995545625686646,
      "learning_rate": 9.801216648272917e-05,
      "loss": 2.1026,
      "step": 169374
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.2864432334899902,
      "learning_rate": 9.80091420381058e-05,
      "loss": 2.018,
      "step": 169375
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.2182456254959106,
      "learning_rate": 9.800611763103645e-05,
      "loss": 2.1733,
      "step": 169376
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.2897694110870361,
      "learning_rate": 9.800309326152158e-05,
      "loss": 1.9962,
      "step": 169377
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.2998464107513428,
      "learning_rate": 9.800006892956175e-05,
      "loss": 2.1367,
      "step": 169378
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.279422402381897,
      "learning_rate": 9.79970446351574e-05,
      "loss": 2.0508,
      "step": 169379
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.3081117868423462,
      "learning_rate": 9.799402037830932e-05,
      "loss": 2.0636,
      "step": 169380
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.3780752420425415,
      "learning_rate": 9.799099615901787e-05,
      "loss": 2.0961,
      "step": 169381
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.3395692110061646,
      "learning_rate": 9.79879719772838e-05,
      "loss": 1.9911,
      "step": 169382
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.2538208961486816,
      "learning_rate": 9.798494783310759e-05,
      "loss": 1.8495,
      "step": 169383
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.3228932619094849,
      "learning_rate": 9.798192372648976e-05,
      "loss": 1.9072,
      "step": 169384
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.2928807735443115,
      "learning_rate": 9.797889965743083e-05,
      "loss": 1.9996,
      "step": 169385
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.338079810142517,
      "learning_rate": 9.797587562593155e-05,
      "loss": 2.1479,
      "step": 169386
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.2701878547668457,
      "learning_rate": 9.797285163199227e-05,
      "loss": 2.1325,
      "step": 169387
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.2718080282211304,
      "learning_rate": 9.79698276756137e-05,
      "loss": 1.906,
      "step": 169388
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.3936151266098022,
      "learning_rate": 9.796680375679628e-05,
      "loss": 2.1178,
      "step": 169389
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.4511464834213257,
      "learning_rate": 9.796377987554087e-05,
      "loss": 2.1608,
      "step": 169390
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.3950986862182617,
      "learning_rate": 9.796075603184759e-05,
      "loss": 1.8415,
      "step": 169391
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.3068228960037231,
      "learning_rate": 9.79577322257173e-05,
      "loss": 1.9487,
      "step": 169392
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.4229055643081665,
      "learning_rate": 9.79547084571504e-05,
      "loss": 1.9518,
      "step": 169393
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.5744718313217163,
      "learning_rate": 9.795168472614765e-05,
      "loss": 2.1317,
      "step": 169394
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.440630555152893,
      "learning_rate": 9.794866103270938e-05,
      "loss": 1.9711,
      "step": 169395
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.2181051969528198,
      "learning_rate": 9.794563737683648e-05,
      "loss": 2.1801,
      "step": 169396
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.1569066047668457,
      "learning_rate": 9.794261375852907e-05,
      "loss": 1.909,
      "step": 169397
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.2996255159378052,
      "learning_rate": 9.793959017778807e-05,
      "loss": 2.1794,
      "step": 169398
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.3272910118103027,
      "learning_rate": 9.79365666346138e-05,
      "loss": 2.1214,
      "step": 169399
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.403742790222168,
      "learning_rate": 9.793354312900702e-05,
      "loss": 2.0703,
      "step": 169400
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.1976547241210938,
      "learning_rate": 9.793051966096813e-05,
      "loss": 1.7872,
      "step": 169401
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.2930675745010376,
      "learning_rate": 9.792749623049798e-05,
      "loss": 2.0221,
      "step": 169402
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.2345651388168335,
      "learning_rate": 9.79244728375967e-05,
      "loss": 1.8714,
      "step": 169403
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.4539766311645508,
      "learning_rate": 9.792144948226518e-05,
      "loss": 2.1671,
      "step": 169404
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.2153040170669556,
      "learning_rate": 9.79184261645038e-05,
      "loss": 1.8375,
      "step": 169405
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.2200345993041992,
      "learning_rate": 9.791540288431327e-05,
      "loss": 2.1277,
      "step": 169406
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.236536979675293,
      "learning_rate": 9.7912379641694e-05,
      "loss": 2.105,
      "step": 169407
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.1915043592453003,
      "learning_rate": 9.790935643664683e-05,
      "loss": 2.0783,
      "step": 169408
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.46245539188385,
      "learning_rate": 9.79063332691719e-05,
      "loss": 2.0435,
      "step": 169409
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.377558708190918,
      "learning_rate": 9.790331013927011e-05,
      "loss": 2.2077,
      "step": 169410
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.2950955629348755,
      "learning_rate": 9.790028704694185e-05,
      "loss": 2.0493,
      "step": 169411
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.2796096801757812,
      "learning_rate": 9.789726399218778e-05,
      "loss": 2.24,
      "step": 169412
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.2798454761505127,
      "learning_rate": 9.789424097500838e-05,
      "loss": 2.0283,
      "step": 169413
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.5033135414123535,
      "learning_rate": 9.789121799540443e-05,
      "loss": 2.0462,
      "step": 169414
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.6169142723083496,
      "learning_rate": 9.78881950533761e-05,
      "loss": 1.9412,
      "step": 169415
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.2121031284332275,
      "learning_rate": 9.788517214892432e-05,
      "loss": 2.0124,
      "step": 169416
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.2448759078979492,
      "learning_rate": 9.788214928204934e-05,
      "loss": 2.0838,
      "step": 169417
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.254618763923645,
      "learning_rate": 9.787912645275203e-05,
      "loss": 1.8998,
      "step": 169418
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.351167917251587,
      "learning_rate": 9.787610366103266e-05,
      "loss": 2.1506,
      "step": 169419
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.2669166326522827,
      "learning_rate": 9.787308090689208e-05,
      "loss": 2.0014,
      "step": 169420
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.2331864833831787,
      "learning_rate": 9.78700581903307e-05,
      "loss": 1.974,
      "step": 169421
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.2860627174377441,
      "learning_rate": 9.786703551134907e-05,
      "loss": 1.9946,
      "step": 169422
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.2180144786834717,
      "learning_rate": 9.786401286994771e-05,
      "loss": 2.0149,
      "step": 169423
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.2040647268295288,
      "learning_rate": 9.786099026612733e-05,
      "loss": 2.0816,
      "step": 169424
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.2939198017120361,
      "learning_rate": 9.78579676998883e-05,
      "loss": 1.9869,
      "step": 169425
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.304527997970581,
      "learning_rate": 9.785494517123142e-05,
      "loss": 2.0308,
      "step": 169426
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.3970905542373657,
      "learning_rate": 9.785192268015706e-05,
      "loss": 1.9283,
      "step": 169427
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.2735751867294312,
      "learning_rate": 9.78489002266659e-05,
      "loss": 2.0719,
      "step": 169428
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.2523905038833618,
      "learning_rate": 9.784587781075832e-05,
      "loss": 2.0303,
      "step": 169429
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.3308640718460083,
      "learning_rate": 9.784285543243511e-05,
      "loss": 1.8261,
      "step": 169430
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.4489600658416748,
      "learning_rate": 9.783983309169663e-05,
      "loss": 1.9614,
      "step": 169431
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.179038405418396,
      "learning_rate": 9.783681078854365e-05,
      "loss": 2.1663,
      "step": 169432
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.2905783653259277,
      "learning_rate": 9.78337885229766e-05,
      "loss": 1.9859,
      "step": 169433
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.2807948589324951,
      "learning_rate": 9.78307662949961e-05,
      "loss": 2.2273,
      "step": 169434
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.258984923362732,
      "learning_rate": 9.782774410460256e-05,
      "loss": 1.9592,
      "step": 169435
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.2214398384094238,
      "learning_rate": 9.782472195179674e-05,
      "loss": 2.0592,
      "step": 169436
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.3463616371154785,
      "learning_rate": 9.782169983657904e-05,
      "loss": 2.0136,
      "step": 169437
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.4173297882080078,
      "learning_rate": 9.781867775895019e-05,
      "loss": 1.8918,
      "step": 169438
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.3072245121002197,
      "learning_rate": 9.781565571891069e-05,
      "loss": 2.0383,
      "step": 169439
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.218727707862854,
      "learning_rate": 9.781263371646103e-05,
      "loss": 2.0565,
      "step": 169440
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.3536630868911743,
      "learning_rate": 9.780961175160175e-05,
      "loss": 2.1205,
      "step": 169441
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.1920572519302368,
      "learning_rate": 9.78065898243336e-05,
      "loss": 2.1598,
      "step": 169442
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.2589824199676514,
      "learning_rate": 9.780356793465686e-05,
      "loss": 2.1172,
      "step": 169443
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.2963857650756836,
      "learning_rate": 9.780054608257241e-05,
      "loss": 1.9887,
      "step": 169444
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.279413104057312,
      "learning_rate": 9.779752426808063e-05,
      "loss": 1.9942,
      "step": 169445
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.2131032943725586,
      "learning_rate": 9.779450249118213e-05,
      "loss": 2.0311,
      "step": 169446
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.1258820295333862,
      "learning_rate": 9.779148075187733e-05,
      "loss": 1.9354,
      "step": 169447
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.259324312210083,
      "learning_rate": 9.778845905016701e-05,
      "loss": 2.1934,
      "step": 169448
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.178911566734314,
      "learning_rate": 9.778543738605153e-05,
      "loss": 2.0407,
      "step": 169449
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.3263976573944092,
      "learning_rate": 9.778241575953167e-05,
      "loss": 1.9915,
      "step": 169450
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.198617696762085,
      "learning_rate": 9.777939417060786e-05,
      "loss": 1.9903,
      "step": 169451
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.1236510276794434,
      "learning_rate": 9.777637261928058e-05,
      "loss": 1.9387,
      "step": 169452
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.6708482503890991,
      "learning_rate": 9.77733511055506e-05,
      "loss": 2.0721,
      "step": 169453
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.1953412294387817,
      "learning_rate": 9.777032962941837e-05,
      "loss": 2.0317,
      "step": 169454
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.2605750560760498,
      "learning_rate": 9.776730819088434e-05,
      "loss": 2.0275,
      "step": 169455
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.2174855470657349,
      "learning_rate": 9.77642867899493e-05,
      "loss": 1.8866,
      "step": 169456
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.2770193815231323,
      "learning_rate": 9.776126542661367e-05,
      "loss": 1.8895,
      "step": 169457
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.337666392326355,
      "learning_rate": 9.775824410087797e-05,
      "loss": 2.086,
      "step": 169458
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.3952902555465698,
      "learning_rate": 9.775522281274293e-05,
      "loss": 1.9392,
      "step": 169459
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.2715855836868286,
      "learning_rate": 9.775220156220899e-05,
      "loss": 2.0621,
      "step": 169460
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.2386199235916138,
      "learning_rate": 9.774918034927665e-05,
      "loss": 1.9501,
      "step": 169461
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.2952638864517212,
      "learning_rate": 9.774615917394664e-05,
      "loss": 2.2496,
      "step": 169462
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.2151662111282349,
      "learning_rate": 9.774313803621946e-05,
      "loss": 1.9027,
      "step": 169463
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.2539029121398926,
      "learning_rate": 9.774011693609553e-05,
      "loss": 1.8956,
      "step": 169464
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.2337870597839355,
      "learning_rate": 9.773709587357562e-05,
      "loss": 1.9892,
      "step": 169465
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.25789213180542,
      "learning_rate": 9.773407484866012e-05,
      "loss": 2.2289,
      "step": 169466
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.2650113105773926,
      "learning_rate": 9.773105386134978e-05,
      "loss": 1.8571,
      "step": 169467
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.285662055015564,
      "learning_rate": 9.772803291164504e-05,
      "loss": 2.1456,
      "step": 169468
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.3325345516204834,
      "learning_rate": 9.772501199954647e-05,
      "loss": 2.0583,
      "step": 169469
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.3565007448196411,
      "learning_rate": 9.772199112505455e-05,
      "loss": 2.0417,
      "step": 169470
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.2536134719848633,
      "learning_rate": 9.771897028817006e-05,
      "loss": 1.9428,
      "step": 169471
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.2773785591125488,
      "learning_rate": 9.771594948889327e-05,
      "loss": 1.8706,
      "step": 169472
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.2940303087234497,
      "learning_rate": 9.771292872722506e-05,
      "loss": 2.0973,
      "step": 169473
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.3334938287734985,
      "learning_rate": 9.770990800316571e-05,
      "loss": 2.0825,
      "step": 169474
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.292372226715088,
      "learning_rate": 9.77068873167161e-05,
      "loss": 1.9953,
      "step": 169475
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.2544500827789307,
      "learning_rate": 9.77038666678764e-05,
      "loss": 2.0082,
      "step": 169476
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.3022352457046509,
      "learning_rate": 9.770084605664746e-05,
      "loss": 1.8565,
      "step": 169477
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.457971215248108,
      "learning_rate": 9.769782548302967e-05,
      "loss": 2.0793,
      "step": 169478
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.3262416124343872,
      "learning_rate": 9.769480494702378e-05,
      "loss": 2.0934,
      "step": 169479
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.2193398475646973,
      "learning_rate": 9.769178444863011e-05,
      "loss": 2.1722,
      "step": 169480
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.2375996112823486,
      "learning_rate": 9.768876398784959e-05,
      "loss": 1.9323,
      "step": 169481
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.2274367809295654,
      "learning_rate": 9.768574356468234e-05,
      "loss": 2.0821,
      "step": 169482
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.2780189514160156,
      "learning_rate": 9.768272317912922e-05,
      "loss": 1.7816,
      "step": 169483
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.2356586456298828,
      "learning_rate": 9.767970283119064e-05,
      "loss": 2.0529,
      "step": 169484
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.5185942649841309,
      "learning_rate": 9.767668252086729e-05,
      "loss": 2.1167,
      "step": 169485
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.2428252696990967,
      "learning_rate": 9.767366224815956e-05,
      "loss": 1.8582,
      "step": 169486
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.2773951292037964,
      "learning_rate": 9.76706420130682e-05,
      "loss": 2.038,
      "step": 169487
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.4085396528244019,
      "learning_rate": 9.766762181559372e-05,
      "loss": 1.9354,
      "step": 169488
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.3238376379013062,
      "learning_rate": 9.766460165573662e-05,
      "loss": 1.9998,
      "step": 169489
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.545812964439392,
      "learning_rate": 9.766158153349742e-05,
      "loss": 2.0277,
      "step": 169490
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.4308868646621704,
      "learning_rate": 9.765856144887684e-05,
      "loss": 2.2558,
      "step": 169491
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.2098608016967773,
      "learning_rate": 9.765554140187522e-05,
      "loss": 1.901,
      "step": 169492
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.3050814867019653,
      "learning_rate": 9.765252139249338e-05,
      "loss": 2.381,
      "step": 169493
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.627558946609497,
      "learning_rate": 9.764950142073179e-05,
      "loss": 1.9675,
      "step": 169494
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.766646385192871,
      "learning_rate": 9.764648148659093e-05,
      "loss": 1.903,
      "step": 169495
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.3260058164596558,
      "learning_rate": 9.764346159007134e-05,
      "loss": 2.1107,
      "step": 169496
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.1732640266418457,
      "learning_rate": 9.764044173117374e-05,
      "loss": 2.008,
      "step": 169497
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.3084691762924194,
      "learning_rate": 9.76374219098985e-05,
      "loss": 1.9001,
      "step": 169498
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.2806897163391113,
      "learning_rate": 9.763440212624636e-05,
      "loss": 2.0976,
      "step": 169499
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.5727055072784424,
      "learning_rate": 9.763138238021782e-05,
      "loss": 2.0162,
      "step": 169500
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.1498609781265259,
      "learning_rate": 9.762836267181343e-05,
      "loss": 1.8791,
      "step": 169501
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.2900865077972412,
      "learning_rate": 9.762534300103368e-05,
      "loss": 2.0377,
      "step": 169502
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.1189427375793457,
      "learning_rate": 9.762232336787925e-05,
      "loss": 2.0984,
      "step": 169503
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.2311933040618896,
      "learning_rate": 9.761930377235056e-05,
      "loss": 2.0135,
      "step": 169504
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.1816853284835815,
      "learning_rate": 9.761628421444839e-05,
      "loss": 2.0505,
      "step": 169505
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.4136672019958496,
      "learning_rate": 9.761326469417313e-05,
      "loss": 1.9029,
      "step": 169506
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.2980021238327026,
      "learning_rate": 9.761024521152542e-05,
      "loss": 2.016,
      "step": 169507
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.1453348398208618,
      "learning_rate": 9.760722576650567e-05,
      "loss": 2.1022,
      "step": 169508
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.1861950159072876,
      "learning_rate": 9.760420635911466e-05,
      "loss": 2.1977,
      "step": 169509
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.2721728086471558,
      "learning_rate": 9.760118698935274e-05,
      "loss": 2.2183,
      "step": 169510
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.1623281240463257,
      "learning_rate": 9.759816765722069e-05,
      "loss": 1.9441,
      "step": 169511
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.2348030805587769,
      "learning_rate": 9.759514836271895e-05,
      "loss": 2.0562,
      "step": 169512
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.5233066082000732,
      "learning_rate": 9.759212910584813e-05,
      "loss": 1.9072,
      "step": 169513
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.3922698497772217,
      "learning_rate": 9.75891098866086e-05,
      "loss": 2.135,
      "step": 169514
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.1669975519180298,
      "learning_rate": 9.758609070500122e-05,
      "loss": 1.9139,
      "step": 169515
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.2246180772781372,
      "learning_rate": 9.758307156102629e-05,
      "loss": 2.0185,
      "step": 169516
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.2402229309082031,
      "learning_rate": 9.758005245468457e-05,
      "loss": 2.2184,
      "step": 169517
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.39846670627594,
      "learning_rate": 9.757703338597656e-05,
      "loss": 2.2147,
      "step": 169518
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.309190034866333,
      "learning_rate": 9.75740143549028e-05,
      "loss": 2.0339,
      "step": 169519
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.3496304750442505,
      "learning_rate": 9.757099536146376e-05,
      "loss": 2.0812,
      "step": 169520
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.2689915895462036,
      "learning_rate": 9.756797640566018e-05,
      "loss": 1.9063,
      "step": 169521
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.3453624248504639,
      "learning_rate": 9.756495748749243e-05,
      "loss": 2.0915,
      "step": 169522
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.5633151531219482,
      "learning_rate": 9.756193860696126e-05,
      "loss": 2.1092,
      "step": 169523
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.2554019689559937,
      "learning_rate": 9.755891976406717e-05,
      "loss": 2.0958,
      "step": 169524
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.2160255908966064,
      "learning_rate": 9.75559009588107e-05,
      "loss": 1.8922,
      "step": 169525
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.2503538131713867,
      "learning_rate": 9.755288219119232e-05,
      "loss": 1.9922,
      "step": 169526
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.741940975189209,
      "learning_rate": 9.754986346121276e-05,
      "loss": 1.9063,
      "step": 169527
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.1892707347869873,
      "learning_rate": 9.75468447688724e-05,
      "loss": 1.9519,
      "step": 169528
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.2671576738357544,
      "learning_rate": 9.7543826114172e-05,
      "loss": 2.1743,
      "step": 169529
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.323997974395752,
      "learning_rate": 9.754080749711205e-05,
      "loss": 1.9838,
      "step": 169530
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.4578022956848145,
      "learning_rate": 9.753778891769307e-05,
      "loss": 2.1357,
      "step": 169531
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.536614179611206,
      "learning_rate": 9.753477037591553e-05,
      "loss": 2.056,
      "step": 169532
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.5323673486709595,
      "learning_rate": 9.753175187178022e-05,
      "loss": 1.9759,
      "step": 169533
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.329576849937439,
      "learning_rate": 9.752873340528746e-05,
      "loss": 2.2157,
      "step": 169534
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.204416036605835,
      "learning_rate": 9.752571497643802e-05,
      "loss": 2.1446,
      "step": 169535
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.4194916486740112,
      "learning_rate": 9.752269658523239e-05,
      "loss": 2.209,
      "step": 169536
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.3176621198654175,
      "learning_rate": 9.751967823167103e-05,
      "loss": 2.1484,
      "step": 169537
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.3509796857833862,
      "learning_rate": 9.751665991575466e-05,
      "loss": 2.2909,
      "step": 169538
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.238065481185913,
      "learning_rate": 9.751364163748377e-05,
      "loss": 1.9204,
      "step": 169539
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.2309362888336182,
      "learning_rate": 9.751062339685881e-05,
      "loss": 1.9877,
      "step": 169540
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.2774059772491455,
      "learning_rate": 9.750760519388054e-05,
      "loss": 1.9165,
      "step": 169541
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.2302433252334595,
      "learning_rate": 9.750458702854943e-05,
      "loss": 1.9633,
      "step": 169542
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.253556728363037,
      "learning_rate": 9.750156890086596e-05,
      "loss": 1.9676,
      "step": 169543
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.2141729593276978,
      "learning_rate": 9.749855081083087e-05,
      "loss": 2.1637,
      "step": 169544
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.2969785928726196,
      "learning_rate": 9.74955327584446e-05,
      "loss": 2.2038,
      "step": 169545
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.2199116945266724,
      "learning_rate": 9.749251474370765e-05,
      "loss": 1.7794,
      "step": 169546
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.132362961769104,
      "learning_rate": 9.748949676662077e-05,
      "loss": 2.0898,
      "step": 169547
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.1504565477371216,
      "learning_rate": 9.748647882718439e-05,
      "loss": 2.0427,
      "step": 169548
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.3237953186035156,
      "learning_rate": 9.748346092539902e-05,
      "loss": 1.9672,
      "step": 169549
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.3292053937911987,
      "learning_rate": 9.748044306126539e-05,
      "loss": 2.0146,
      "step": 169550
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.216059684753418,
      "learning_rate": 9.747742523478388e-05,
      "loss": 2.0395,
      "step": 169551
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.1818121671676636,
      "learning_rate": 9.747440744595521e-05,
      "loss": 1.9112,
      "step": 169552
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.1401383876800537,
      "learning_rate": 9.74713896947799e-05,
      "loss": 2.0782,
      "step": 169553
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.2206944227218628,
      "learning_rate": 9.746837198125836e-05,
      "loss": 2.2,
      "step": 169554
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.1998317241668701,
      "learning_rate": 9.746535430539141e-05,
      "loss": 2.0127,
      "step": 169555
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.2658137083053589,
      "learning_rate": 9.746233666717943e-05,
      "loss": 2.3306,
      "step": 169556
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.114611268043518,
      "learning_rate": 9.745931906662294e-05,
      "loss": 2.0689,
      "step": 169557
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.2308393716812134,
      "learning_rate": 9.745630150372267e-05,
      "loss": 2.1065,
      "step": 169558
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.3359816074371338,
      "learning_rate": 9.745328397847904e-05,
      "loss": 1.9686,
      "step": 169559
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.230032205581665,
      "learning_rate": 9.745026649089273e-05,
      "loss": 1.9789,
      "step": 169560
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.2492995262145996,
      "learning_rate": 9.744724904096425e-05,
      "loss": 1.8389,
      "step": 169561
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.1806648969650269,
      "learning_rate": 9.744423162869414e-05,
      "loss": 2.3543,
      "step": 169562
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.4369224309921265,
      "learning_rate": 9.74412142540829e-05,
      "loss": 2.0793,
      "step": 169563
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.3959003686904907,
      "learning_rate": 9.743819691713124e-05,
      "loss": 2.1584,
      "step": 169564
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.219951868057251,
      "learning_rate": 9.743517961783955e-05,
      "loss": 1.998,
      "step": 169565
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.2914936542510986,
      "learning_rate": 9.743216235620858e-05,
      "loss": 1.995,
      "step": 169566
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.2974073886871338,
      "learning_rate": 9.742914513223878e-05,
      "loss": 1.9567,
      "step": 169567
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.2903997898101807,
      "learning_rate": 9.742612794593074e-05,
      "loss": 2.0991,
      "step": 169568
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.2803722620010376,
      "learning_rate": 9.742311079728493e-05,
      "loss": 1.9878,
      "step": 169569
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.3516230583190918,
      "learning_rate": 9.742009368630204e-05,
      "loss": 1.9857,
      "step": 169570
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.2793724536895752,
      "learning_rate": 9.741707661298252e-05,
      "loss": 2.068,
      "step": 169571
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.2303563356399536,
      "learning_rate": 9.741405957732705e-05,
      "loss": 2.0068,
      "step": 169572
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.3301341533660889,
      "learning_rate": 9.741104257933615e-05,
      "loss": 1.9616,
      "step": 169573
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.3518420457839966,
      "learning_rate": 9.740802561901039e-05,
      "loss": 2.0436,
      "step": 169574
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.2881948947906494,
      "learning_rate": 9.740500869635017e-05,
      "loss": 2.0283,
      "step": 169575
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.3725385665893555,
      "learning_rate": 9.740199181135627e-05,
      "loss": 1.9725,
      "step": 169576
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.328666090965271,
      "learning_rate": 9.739897496402908e-05,
      "loss": 2.0284,
      "step": 169577
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.241067886352539,
      "learning_rate": 9.739595815436938e-05,
      "loss": 1.8762,
      "step": 169578
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.400660753250122,
      "learning_rate": 9.739294138237756e-05,
      "loss": 2.2414,
      "step": 169579
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.2938247919082642,
      "learning_rate": 9.738992464805421e-05,
      "loss": 1.8455,
      "step": 169580
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.4162986278533936,
      "learning_rate": 9.738690795139983e-05,
      "loss": 1.9692,
      "step": 169581
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.5416339635849,
      "learning_rate": 9.738389129241512e-05,
      "loss": 1.9944,
      "step": 169582
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.3095097541809082,
      "learning_rate": 9.738087467110047e-05,
      "loss": 2.0554,
      "step": 169583
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.2167305946350098,
      "learning_rate": 9.737785808745667e-05,
      "loss": 1.918,
      "step": 169584
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.3533324003219604,
      "learning_rate": 9.737484154148411e-05,
      "loss": 2.0722,
      "step": 169585
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.2755591869354248,
      "learning_rate": 9.73718250331834e-05,
      "loss": 1.958,
      "step": 169586
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.1843887567520142,
      "learning_rate": 9.736880856255498e-05,
      "loss": 1.8834,
      "step": 169587
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.2992274761199951,
      "learning_rate": 9.736579212959968e-05,
      "loss": 1.9834,
      "step": 169588
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.2391679286956787,
      "learning_rate": 9.736277573431776e-05,
      "loss": 2.0841,
      "step": 169589
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.3836711645126343,
      "learning_rate": 9.735975937671003e-05,
      "loss": 2.1043,
      "step": 169590
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.1640688180923462,
      "learning_rate": 9.735674305677694e-05,
      "loss": 2.0103,
      "step": 169591
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.2066144943237305,
      "learning_rate": 9.735372677451908e-05,
      "loss": 2.385,
      "step": 169592
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.2279423475265503,
      "learning_rate": 9.735071052993688e-05,
      "loss": 2.2457,
      "step": 169593
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.2469614744186401,
      "learning_rate": 9.73476943230311e-05,
      "loss": 1.8797,
      "step": 169594
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.2642332315444946,
      "learning_rate": 9.73446781538021e-05,
      "loss": 1.7772,
      "step": 169595
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.2808504104614258,
      "learning_rate": 9.734166202225069e-05,
      "loss": 2.08,
      "step": 169596
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.3240180015563965,
      "learning_rate": 9.733864592837724e-05,
      "loss": 1.9909,
      "step": 169597
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.275323748588562,
      "learning_rate": 9.733562987218241e-05,
      "loss": 1.9163,
      "step": 169598
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.3125925064086914,
      "learning_rate": 9.733261385366657e-05,
      "loss": 2.0697,
      "step": 169599
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.3636459112167358,
      "learning_rate": 9.732959787283054e-05,
      "loss": 2.1398,
      "step": 169600
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.2018086910247803,
      "learning_rate": 9.732658192967465e-05,
      "loss": 2.1435,
      "step": 169601
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.3108364343643188,
      "learning_rate": 9.73235660241997e-05,
      "loss": 2.0011,
      "step": 169602
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.2405561208724976,
      "learning_rate": 9.732055015640609e-05,
      "loss": 1.9195,
      "step": 169603
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.3226858377456665,
      "learning_rate": 9.731753432629445e-05,
      "loss": 2.0925,
      "step": 169604
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.3721710443496704,
      "learning_rate": 9.73145185338652e-05,
      "loss": 2.0143,
      "step": 169605
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.2917295694351196,
      "learning_rate": 9.731150277911908e-05,
      "loss": 2.0548,
      "step": 169606
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.2575898170471191,
      "learning_rate": 9.730848706205651e-05,
      "loss": 1.9849,
      "step": 169607
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.223042368888855,
      "learning_rate": 9.730547138267819e-05,
      "loss": 2.0459,
      "step": 169608
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.301079511642456,
      "learning_rate": 9.730245574098463e-05,
      "loss": 2.1138,
      "step": 169609
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.2715835571289062,
      "learning_rate": 9.729944013697636e-05,
      "loss": 1.7617,
      "step": 169610
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.251320481300354,
      "learning_rate": 9.729642457065387e-05,
      "loss": 1.8592,
      "step": 169611
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.2710163593292236,
      "learning_rate": 9.729340904201787e-05,
      "loss": 1.9905,
      "step": 169612
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.2332881689071655,
      "learning_rate": 9.729039355106881e-05,
      "loss": 2.1253,
      "step": 169613
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.2067803144454956,
      "learning_rate": 9.728737809780731e-05,
      "loss": 2.2155,
      "step": 169614
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.254764437675476,
      "learning_rate": 9.728436268223398e-05,
      "loss": 2.0453,
      "step": 169615
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.3260587453842163,
      "learning_rate": 9.728134730434925e-05,
      "loss": 2.0396,
      "step": 169616
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.4312976598739624,
      "learning_rate": 9.727833196415369e-05,
      "loss": 2.0046,
      "step": 169617
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.2332104444503784,
      "learning_rate": 9.7275316661648e-05,
      "loss": 1.9817,
      "step": 169618
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.2652318477630615,
      "learning_rate": 9.727230139683258e-05,
      "loss": 2.0805,
      "step": 169619
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.2298814058303833,
      "learning_rate": 9.726928616970816e-05,
      "loss": 2.0153,
      "step": 169620
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.3057708740234375,
      "learning_rate": 9.726627098027506e-05,
      "loss": 1.9485,
      "step": 169621
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.2158383131027222,
      "learning_rate": 9.726325582853425e-05,
      "loss": 2.0843,
      "step": 169622
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.3022682666778564,
      "learning_rate": 9.726024071448578e-05,
      "loss": 1.9164,
      "step": 169623
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.3976342678070068,
      "learning_rate": 9.725722563813057e-05,
      "loss": 2.1566,
      "step": 169624
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.1990329027175903,
      "learning_rate": 9.725421059946899e-05,
      "loss": 1.6739,
      "step": 169625
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.1972100734710693,
      "learning_rate": 9.725119559850177e-05,
      "loss": 2.0281,
      "step": 169626
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.2429239749908447,
      "learning_rate": 9.724818063522927e-05,
      "loss": 2.1916,
      "step": 169627
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.149183750152588,
      "learning_rate": 9.724516570965227e-05,
      "loss": 2.0902,
      "step": 169628
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.1527503728866577,
      "learning_rate": 9.72421508217712e-05,
      "loss": 2.0968,
      "step": 169629
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.2205145359039307,
      "learning_rate": 9.723913597158666e-05,
      "loss": 1.9211,
      "step": 169630
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.2728649377822876,
      "learning_rate": 9.72361211590991e-05,
      "loss": 2.1551,
      "step": 169631
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.4073714017868042,
      "learning_rate": 9.723310638430926e-05,
      "loss": 2.1369,
      "step": 169632
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.2703759670257568,
      "learning_rate": 9.723009164721753e-05,
      "loss": 2.1841,
      "step": 169633
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.2049819231033325,
      "learning_rate": 9.72270769478246e-05,
      "loss": 2.2316,
      "step": 169634
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.1570650339126587,
      "learning_rate": 9.722406228613102e-05,
      "loss": 1.9877,
      "step": 169635
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.1756079196929932,
      "learning_rate": 9.722104766213725e-05,
      "loss": 2.1325,
      "step": 169636
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.373021125793457,
      "learning_rate": 9.721803307584396e-05,
      "loss": 2.0748,
      "step": 169637
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.3284024000167847,
      "learning_rate": 9.721501852725171e-05,
      "loss": 1.9672,
      "step": 169638
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.311011791229248,
      "learning_rate": 9.721200401636087e-05,
      "loss": 2.0329,
      "step": 169639
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.2728705406188965,
      "learning_rate": 9.720898954317226e-05,
      "loss": 2.0592,
      "step": 169640
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.2305277585983276,
      "learning_rate": 9.720597510768634e-05,
      "loss": 2.1545,
      "step": 169641
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.2965342998504639,
      "learning_rate": 9.720296070990355e-05,
      "loss": 2.2041,
      "step": 169642
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.2654632329940796,
      "learning_rate": 9.719994634982468e-05,
      "loss": 1.9978,
      "step": 169643
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.6104756593704224,
      "learning_rate": 9.719693202745014e-05,
      "loss": 2.005,
      "step": 169644
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.151917815208435,
      "learning_rate": 9.719391774278044e-05,
      "loss": 1.9967,
      "step": 169645
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.2980469465255737,
      "learning_rate": 9.719090349581631e-05,
      "loss": 2.1427,
      "step": 169646
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.3502178192138672,
      "learning_rate": 9.718788928655821e-05,
      "loss": 1.9697,
      "step": 169647
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.249168872833252,
      "learning_rate": 9.718487511500659e-05,
      "loss": 1.8771,
      "step": 169648
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.1089963912963867,
      "learning_rate": 9.718186098116229e-05,
      "loss": 1.7865,
      "step": 169649
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.2698012590408325,
      "learning_rate": 9.717884688502558e-05,
      "loss": 2.0814,
      "step": 169650
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.2929046154022217,
      "learning_rate": 9.717583282659723e-05,
      "loss": 2.131,
      "step": 169651
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.2843453884124756,
      "learning_rate": 9.717281880587773e-05,
      "loss": 1.9144,
      "step": 169652
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.1617376804351807,
      "learning_rate": 9.716980482286766e-05,
      "loss": 2.0763,
      "step": 169653
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.545378565788269,
      "learning_rate": 9.716679087756743e-05,
      "loss": 2.0928,
      "step": 169654
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.2346818447113037,
      "learning_rate": 9.716377696997782e-05,
      "loss": 2.0651,
      "step": 169655
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.2121944427490234,
      "learning_rate": 9.71607631000992e-05,
      "loss": 2.042,
      "step": 169656
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.310172200202942,
      "learning_rate": 9.715774926793233e-05,
      "loss": 2.2486,
      "step": 169657
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.5537900924682617,
      "learning_rate": 9.715473547347762e-05,
      "loss": 2.0213,
      "step": 169658
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.2015900611877441,
      "learning_rate": 9.715172171673574e-05,
      "loss": 2.1482,
      "step": 169659
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.2921699285507202,
      "learning_rate": 9.714870799770704e-05,
      "loss": 2.0217,
      "step": 169660
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.344779133796692,
      "learning_rate": 9.714569431639234e-05,
      "loss": 2.1344,
      "step": 169661
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.2726410627365112,
      "learning_rate": 9.7142680672792e-05,
      "loss": 2.1124,
      "step": 169662
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.7908036708831787,
      "learning_rate": 9.713966706690672e-05,
      "loss": 2.0567,
      "step": 169663
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.2259180545806885,
      "learning_rate": 9.713665349873704e-05,
      "loss": 2.2117,
      "step": 169664
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.5332249402999878,
      "learning_rate": 9.713363996828348e-05,
      "loss": 2.0599,
      "step": 169665
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.3561664819717407,
      "learning_rate": 9.713062647554649e-05,
      "loss": 1.7786,
      "step": 169666
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.2462413311004639,
      "learning_rate": 9.712761302052687e-05,
      "loss": 1.9969,
      "step": 169667
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.3135409355163574,
      "learning_rate": 9.712459960322495e-05,
      "loss": 2.182,
      "step": 169668
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.2605245113372803,
      "learning_rate": 9.71215862236415e-05,
      "loss": 2.1119,
      "step": 169669
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.1702423095703125,
      "learning_rate": 9.711857288177696e-05,
      "loss": 1.9303,
      "step": 169670
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.5184839963912964,
      "learning_rate": 9.711555957763188e-05,
      "loss": 1.9512,
      "step": 169671
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.263648509979248,
      "learning_rate": 9.71125463112068e-05,
      "loss": 2.3072,
      "step": 169672
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.4808393716812134,
      "learning_rate": 9.710953308250242e-05,
      "loss": 2.1551,
      "step": 169673
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.2204171419143677,
      "learning_rate": 9.710651989151909e-05,
      "loss": 2.082,
      "step": 169674
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.2482593059539795,
      "learning_rate": 9.710350673825759e-05,
      "loss": 1.9588,
      "step": 169675
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.1423698663711548,
      "learning_rate": 9.710049362271839e-05,
      "loss": 2.052,
      "step": 169676
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.272392749786377,
      "learning_rate": 9.7097480544902e-05,
      "loss": 2.1248,
      "step": 169677
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.2384685277938843,
      "learning_rate": 9.709446750480898e-05,
      "loss": 2.0667,
      "step": 169678
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.3411914110183716,
      "learning_rate": 9.709145450243998e-05,
      "loss": 1.9571,
      "step": 169679
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.3003225326538086,
      "learning_rate": 9.708844153779544e-05,
      "loss": 2.2678,
      "step": 169680
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.3283025026321411,
      "learning_rate": 9.708542861087608e-05,
      "loss": 1.9623,
      "step": 169681
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.196808934211731,
      "learning_rate": 9.708241572168236e-05,
      "loss": 1.9519,
      "step": 169682
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.14666748046875,
      "learning_rate": 9.707940287021485e-05,
      "loss": 2.1281,
      "step": 169683
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.7050879001617432,
      "learning_rate": 9.707639005647402e-05,
      "loss": 2.0429,
      "step": 169684
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.2998493909835815,
      "learning_rate": 9.707337728046061e-05,
      "loss": 1.8895,
      "step": 169685
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.1646672487258911,
      "learning_rate": 9.7070364542175e-05,
      "loss": 2.0997,
      "step": 169686
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.2542203664779663,
      "learning_rate": 9.706735184161792e-05,
      "loss": 1.8774,
      "step": 169687
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.3248077630996704,
      "learning_rate": 9.706433917878977e-05,
      "loss": 2.1426,
      "step": 169688
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.266830563545227,
      "learning_rate": 9.70613265536914e-05,
      "loss": 1.9707,
      "step": 169689
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.1354435682296753,
      "learning_rate": 9.705831396632292e-05,
      "loss": 1.9457,
      "step": 169690
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.2956726551055908,
      "learning_rate": 9.705530141668524e-05,
      "loss": 2.0546,
      "step": 169691
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.2777043581008911,
      "learning_rate": 9.705228890477874e-05,
      "loss": 2.0508,
      "step": 169692
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.2804179191589355,
      "learning_rate": 9.704927643060413e-05,
      "loss": 2.0429,
      "step": 169693
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.3328615427017212,
      "learning_rate": 9.70462639941618e-05,
      "loss": 2.0533,
      "step": 169694
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.239631175994873,
      "learning_rate": 9.704325159545259e-05,
      "loss": 2.1137,
      "step": 169695
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.2381099462509155,
      "learning_rate": 9.704023923447666e-05,
      "loss": 1.8702,
      "step": 169696
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.3304613828659058,
      "learning_rate": 9.70372269112349e-05,
      "loss": 2.0653,
      "step": 169697
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.2555197477340698,
      "learning_rate": 9.703421462572765e-05,
      "loss": 1.8916,
      "step": 169698
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.4758793115615845,
      "learning_rate": 9.703120237795566e-05,
      "loss": 2.3135,
      "step": 169699
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.2769581079483032,
      "learning_rate": 9.702819016791933e-05,
      "loss": 1.8671,
      "step": 169700
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.3092360496520996,
      "learning_rate": 9.702517799561948e-05,
      "loss": 2.0221,
      "step": 169701
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.203151822090149,
      "learning_rate": 9.702216586105626e-05,
      "loss": 2.1727,
      "step": 169702
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.2431204319000244,
      "learning_rate": 9.701915376423054e-05,
      "loss": 2.0931,
      "step": 169703
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.311782717704773,
      "learning_rate": 9.701614170514273e-05,
      "loss": 2.1204,
      "step": 169704
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.297727346420288,
      "learning_rate": 9.701312968379352e-05,
      "loss": 2.1264,
      "step": 169705
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.2329027652740479,
      "learning_rate": 9.701011770018333e-05,
      "loss": 1.9728,
      "step": 169706
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.284680724143982,
      "learning_rate": 9.700710575431296e-05,
      "loss": 1.8122,
      "step": 169707
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.2175507545471191,
      "learning_rate": 9.700409384618262e-05,
      "loss": 2.0655,
      "step": 169708
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.2223924398422241,
      "learning_rate": 9.700108197579317e-05,
      "loss": 2.0327,
      "step": 169709
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.180959701538086,
      "learning_rate": 9.699807014314491e-05,
      "loss": 1.8979,
      "step": 169710
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.231024980545044,
      "learning_rate": 9.699505834823864e-05,
      "loss": 2.216,
      "step": 169711
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.4463601112365723,
      "learning_rate": 9.699204659107474e-05,
      "loss": 2.295,
      "step": 169712
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.366230845451355,
      "learning_rate": 9.698903487165396e-05,
      "loss": 1.8731,
      "step": 169713
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.4130220413208008,
      "learning_rate": 9.698602318997675e-05,
      "loss": 2.0522,
      "step": 169714
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.4016070365905762,
      "learning_rate": 9.698301154604364e-05,
      "loss": 2.1717,
      "step": 169715
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.4010072946548462,
      "learning_rate": 9.697999993985514e-05,
      "loss": 1.9974,
      "step": 169716
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.1663997173309326,
      "learning_rate": 9.697698837141199e-05,
      "loss": 1.8801,
      "step": 169717
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.3015462160110474,
      "learning_rate": 9.697397684071457e-05,
      "loss": 2.1017,
      "step": 169718
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.2508656978607178,
      "learning_rate": 9.69709653477636e-05,
      "loss": 2.1904,
      "step": 169719
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.2457053661346436,
      "learning_rate": 9.696795389255955e-05,
      "loss": 2.0258,
      "step": 169720
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.321387767791748,
      "learning_rate": 9.696494247510289e-05,
      "loss": 1.9462,
      "step": 169721
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.3056782484054565,
      "learning_rate": 9.69619310953944e-05,
      "loss": 2.0479,
      "step": 169722
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.3703253269195557,
      "learning_rate": 9.695891975343451e-05,
      "loss": 2.1883,
      "step": 169723
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.2857252359390259,
      "learning_rate": 9.695590844922368e-05,
      "loss": 2.1324,
      "step": 169724
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.2649943828582764,
      "learning_rate": 9.69528971827627e-05,
      "loss": 2.0543,
      "step": 169725
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.4134290218353271,
      "learning_rate": 9.694988595405199e-05,
      "loss": 1.9448,
      "step": 169726
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.409780740737915,
      "learning_rate": 9.694687476309204e-05,
      "loss": 2.1543,
      "step": 169727
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.277116298675537,
      "learning_rate": 9.69438636098836e-05,
      "loss": 1.8802,
      "step": 169728
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.262656331062317,
      "learning_rate": 9.694085249442712e-05,
      "loss": 2.053,
      "step": 169729
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.181839108467102,
      "learning_rate": 9.693784141672308e-05,
      "loss": 1.9573,
      "step": 169730
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.611749291419983,
      "learning_rate": 9.693483037677222e-05,
      "loss": 2.0841,
      "step": 169731
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.3134857416152954,
      "learning_rate": 9.693181937457501e-05,
      "loss": 2.0833,
      "step": 169732
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.2990162372589111,
      "learning_rate": 9.692880841013189e-05,
      "loss": 2.2214,
      "step": 169733
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.3055757284164429,
      "learning_rate": 9.692579748344369e-05,
      "loss": 2.1569,
      "step": 169734
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.2240649461746216,
      "learning_rate": 9.692278659451068e-05,
      "loss": 2.0188,
      "step": 169735
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.2380001544952393,
      "learning_rate": 9.691977574333364e-05,
      "loss": 2.1664,
      "step": 169736
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.27766752243042,
      "learning_rate": 9.691676492991309e-05,
      "loss": 2.0245,
      "step": 169737
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.2161798477172852,
      "learning_rate": 9.691375415424952e-05,
      "loss": 2.0005,
      "step": 169738
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.1846803426742554,
      "learning_rate": 9.691074341634342e-05,
      "loss": 1.9858,
      "step": 169739
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.3293390274047852,
      "learning_rate": 9.690773271619554e-05,
      "loss": 1.8822,
      "step": 169740
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.1580238342285156,
      "learning_rate": 9.690472205380625e-05,
      "loss": 1.9345,
      "step": 169741
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.3028124570846558,
      "learning_rate": 9.690171142917633e-05,
      "loss": 2.1016,
      "step": 169742
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.2648621797561646,
      "learning_rate": 9.689870084230619e-05,
      "loss": 2.1239,
      "step": 169743
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.3590424060821533,
      "learning_rate": 9.689569029319641e-05,
      "loss": 2.0831,
      "step": 169744
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.239490032196045,
      "learning_rate": 9.689267978184748e-05,
      "loss": 1.9435,
      "step": 169745
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.2367522716522217,
      "learning_rate": 9.688966930826011e-05,
      "loss": 2.0425,
      "step": 169746
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.192319393157959,
      "learning_rate": 9.68866588724347e-05,
      "loss": 1.7262,
      "step": 169747
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.201417326927185,
      "learning_rate": 9.6883648474372e-05,
      "loss": 1.9955,
      "step": 169748
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.4465336799621582,
      "learning_rate": 9.688063811407243e-05,
      "loss": 2.0514,
      "step": 169749
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.2376840114593506,
      "learning_rate": 9.687762779153661e-05,
      "loss": 2.0828,
      "step": 169750
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.277099609375,
      "learning_rate": 9.687461750676497e-05,
      "loss": 2.0281,
      "step": 169751
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.238427996635437,
      "learning_rate": 9.687160725975827e-05,
      "loss": 2.1225,
      "step": 169752
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.2081694602966309,
      "learning_rate": 9.686859705051689e-05,
      "loss": 2.1136,
      "step": 169753
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.2521085739135742,
      "learning_rate": 9.686558687904154e-05,
      "loss": 2.1103,
      "step": 169754
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.2539012432098389,
      "learning_rate": 9.686257674533265e-05,
      "loss": 2.0084,
      "step": 169755
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.2749966382980347,
      "learning_rate": 9.685956664939101e-05,
      "loss": 2.0012,
      "step": 169756
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.5065830945968628,
      "learning_rate": 9.685655659121684e-05,
      "loss": 1.9037,
      "step": 169757
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.2752629518508911,
      "learning_rate": 9.685354657081095e-05,
      "loss": 1.9314,
      "step": 169758
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.1929728984832764,
      "learning_rate": 9.685053658817371e-05,
      "loss": 2.0274,
      "step": 169759
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.3824306726455688,
      "learning_rate": 9.68475266433059e-05,
      "loss": 2.1084,
      "step": 169760
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.2336503267288208,
      "learning_rate": 9.684451673620787e-05,
      "loss": 2.0341,
      "step": 169761
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.1397684812545776,
      "learning_rate": 9.684150686688049e-05,
      "loss": 1.8651,
      "step": 169762
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.2327885627746582,
      "learning_rate": 9.68384970353239e-05,
      "loss": 2.0044,
      "step": 169763
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.3352339267730713,
      "learning_rate": 9.683548724153896e-05,
      "loss": 2.085,
      "step": 169764
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.4407285451889038,
      "learning_rate": 9.683247748552605e-05,
      "loss": 1.7882,
      "step": 169765
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.3574119806289673,
      "learning_rate": 9.682946776728591e-05,
      "loss": 2.0891,
      "step": 169766
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.2397276163101196,
      "learning_rate": 9.682645808681894e-05,
      "loss": 1.9581,
      "step": 169767
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.2590479850769043,
      "learning_rate": 9.68234484441259e-05,
      "loss": 1.9343,
      "step": 169768
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.20254385471344,
      "learning_rate": 9.682043883920709e-05,
      "loss": 1.8677,
      "step": 169769
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.2924429178237915,
      "learning_rate": 9.681742927206327e-05,
      "loss": 2.1433,
      "step": 169770
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.3151105642318726,
      "learning_rate": 9.681441974269478e-05,
      "loss": 2.2234,
      "step": 169771
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.2363221645355225,
      "learning_rate": 9.681141025110248e-05,
      "loss": 2.0812,
      "step": 169772
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.2219287157058716,
      "learning_rate": 9.680840079728663e-05,
      "loss": 2.2267,
      "step": 169773
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.3538012504577637,
      "learning_rate": 9.680539138124816e-05,
      "loss": 2.0355,
      "step": 169774
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.2055143117904663,
      "learning_rate": 9.680238200298721e-05,
      "loss": 1.9118,
      "step": 169775
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.219092607498169,
      "learning_rate": 9.679937266250463e-05,
      "loss": 2.0685,
      "step": 169776
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.3464971780776978,
      "learning_rate": 9.679636335980077e-05,
      "loss": 2.1368,
      "step": 169777
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.1718266010284424,
      "learning_rate": 9.679335409487643e-05,
      "loss": 2.2457,
      "step": 169778
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.30674409866333,
      "learning_rate": 9.679034486773189e-05,
      "loss": 2.0253,
      "step": 169779
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.3491575717926025,
      "learning_rate": 9.678733567836811e-05,
      "loss": 1.9128,
      "step": 169780
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.2695280313491821,
      "learning_rate": 9.67843265267852e-05,
      "loss": 2.2136,
      "step": 169781
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.4937012195587158,
      "learning_rate": 9.678131741298399e-05,
      "loss": 2.1875,
      "step": 169782
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.2653101682662964,
      "learning_rate": 9.677830833696488e-05,
      "loss": 2.001,
      "step": 169783
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.2005361318588257,
      "learning_rate": 9.677529929872863e-05,
      "loss": 2.0492,
      "step": 169784
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.3081345558166504,
      "learning_rate": 9.677229029827556e-05,
      "loss": 2.0428,
      "step": 169785
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.2978439331054688,
      "learning_rate": 9.676928133560657e-05,
      "loss": 1.9767,
      "step": 169786
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.2331624031066895,
      "learning_rate": 9.676627241072182e-05,
      "loss": 2.1585,
      "step": 169787
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.2574790716171265,
      "learning_rate": 9.676326352362214e-05,
      "loss": 1.9557,
      "step": 169788
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.2208410501480103,
      "learning_rate": 9.67602546743079e-05,
      "loss": 1.8424,
      "step": 169789
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.3776746988296509,
      "learning_rate": 9.675724586277988e-05,
      "loss": 1.8817,
      "step": 169790
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.3377435207366943,
      "learning_rate": 9.675423708903843e-05,
      "loss": 2.3889,
      "step": 169791
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.2806674242019653,
      "learning_rate": 9.67512283530844e-05,
      "loss": 1.93,
      "step": 169792
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.298679232597351,
      "learning_rate": 9.674821965491794e-05,
      "loss": 2.1364,
      "step": 169793
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.2139954566955566,
      "learning_rate": 9.674521099453994e-05,
      "loss": 2.1304,
      "step": 169794
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.3373805284500122,
      "learning_rate": 9.674220237195074e-05,
      "loss": 2.0867,
      "step": 169795
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.3016026020050049,
      "learning_rate": 9.673919378715108e-05,
      "loss": 2.0723,
      "step": 169796
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.4078636169433594,
      "learning_rate": 9.673618524014136e-05,
      "loss": 2.097,
      "step": 169797
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.1561182737350464,
      "learning_rate": 9.67331767309223e-05,
      "loss": 2.0045,
      "step": 169798
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.3231596946716309,
      "learning_rate": 9.673016825949444e-05,
      "loss": 1.8793,
      "step": 169799
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.2446006536483765,
      "learning_rate": 9.672715982585822e-05,
      "loss": 2.0558,
      "step": 169800
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.339539885520935,
      "learning_rate": 9.672415143001419e-05,
      "loss": 2.2952,
      "step": 169801
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.1608513593673706,
      "learning_rate": 9.672114307196305e-05,
      "loss": 2.1345,
      "step": 169802
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.239166021347046,
      "learning_rate": 9.671813475170517e-05,
      "loss": 2.0567,
      "step": 169803
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.3179222345352173,
      "learning_rate": 9.671512646924138e-05,
      "loss": 2.1229,
      "step": 169804
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.7809631824493408,
      "learning_rate": 9.671211822457204e-05,
      "loss": 2.0555,
      "step": 169805
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.3326698541641235,
      "learning_rate": 9.670911001769778e-05,
      "loss": 1.8131,
      "step": 169806
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.1830271482467651,
      "learning_rate": 9.670610184861901e-05,
      "loss": 1.8747,
      "step": 169807
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.289158821105957,
      "learning_rate": 9.670309371733654e-05,
      "loss": 1.8205,
      "step": 169808
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.26628577709198,
      "learning_rate": 9.670008562385071e-05,
      "loss": 1.9105,
      "step": 169809
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.2193409204483032,
      "learning_rate": 9.669707756816223e-05,
      "loss": 2.0846,
      "step": 169810
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.2649344205856323,
      "learning_rate": 9.669406955027164e-05,
      "loss": 1.8559,
      "step": 169811
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.3665162324905396,
      "learning_rate": 9.669106157017933e-05,
      "loss": 2.1224,
      "step": 169812
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.2470378875732422,
      "learning_rate": 9.668805362788611e-05,
      "loss": 1.8901,
      "step": 169813
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.2785604000091553,
      "learning_rate": 9.668504572339241e-05,
      "loss": 2.2282,
      "step": 169814
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.35199773311615,
      "learning_rate": 9.668203785669873e-05,
      "loss": 2.1931,
      "step": 169815
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.257735013961792,
      "learning_rate": 9.667903002780574e-05,
      "loss": 1.9869,
      "step": 169816
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.4020605087280273,
      "learning_rate": 9.667602223671398e-05,
      "loss": 1.985,
      "step": 169817
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.2531956434249878,
      "learning_rate": 9.667301448342391e-05,
      "loss": 2.0118,
      "step": 169818
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.4562708139419556,
      "learning_rate": 9.667000676793625e-05,
      "loss": 1.9766,
      "step": 169819
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.3440914154052734,
      "learning_rate": 9.666699909025134e-05,
      "loss": 2.0044,
      "step": 169820
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.4224936962127686,
      "learning_rate": 9.666399145037002e-05,
      "loss": 2.2245,
      "step": 169821
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.1834402084350586,
      "learning_rate": 9.666098384829267e-05,
      "loss": 1.9788,
      "step": 169822
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.3513894081115723,
      "learning_rate": 9.665797628401991e-05,
      "loss": 2.2517,
      "step": 169823
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.353535771369934,
      "learning_rate": 9.665496875755214e-05,
      "loss": 1.9707,
      "step": 169824
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.271444320678711,
      "learning_rate": 9.665196126889016e-05,
      "loss": 2.0767,
      "step": 169825
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.2882535457611084,
      "learning_rate": 9.664895381803435e-05,
      "loss": 2.0109,
      "step": 169826
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.331100583076477,
      "learning_rate": 9.664594640498538e-05,
      "loss": 1.9931,
      "step": 169827
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.599288821220398,
      "learning_rate": 9.664293902974382e-05,
      "loss": 2.0094,
      "step": 169828
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.2751719951629639,
      "learning_rate": 9.663993169231017e-05,
      "loss": 1.9466,
      "step": 169829
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.2235721349716187,
      "learning_rate": 9.663692439268488e-05,
      "loss": 2.1102,
      "step": 169830
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.2492483854293823,
      "learning_rate": 9.663391713086871e-05,
      "loss": 2.1093,
      "step": 169831
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.4558675289154053,
      "learning_rate": 9.663090990686207e-05,
      "loss": 2.0979,
      "step": 169832
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.4382529258728027,
      "learning_rate": 9.662790272066564e-05,
      "loss": 2.0597,
      "step": 169833
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.2302857637405396,
      "learning_rate": 9.662489557227983e-05,
      "loss": 1.8632,
      "step": 169834
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.3221269845962524,
      "learning_rate": 9.662188846170553e-05,
      "loss": 2.1324,
      "step": 169835
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.4153655767440796,
      "learning_rate": 9.661888138894283e-05,
      "loss": 1.915,
      "step": 169836
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.3554980754852295,
      "learning_rate": 9.661587435399262e-05,
      "loss": 1.9291,
      "step": 169837
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.3729264736175537,
      "learning_rate": 9.661286735685526e-05,
      "loss": 2.0718,
      "step": 169838
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.4963103532791138,
      "learning_rate": 9.660986039753152e-05,
      "loss": 1.7345,
      "step": 169839
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.3006845712661743,
      "learning_rate": 9.660685347602174e-05,
      "loss": 2.1343,
      "step": 169840
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.4627964496612549,
      "learning_rate": 9.660384659232677e-05,
      "loss": 1.8998,
      "step": 169841
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.1701529026031494,
      "learning_rate": 9.66008397464468e-05,
      "loss": 2.107,
      "step": 169842
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.5820790529251099,
      "learning_rate": 9.659783293838265e-05,
      "loss": 1.9464,
      "step": 169843
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.3005011081695557,
      "learning_rate": 9.659482616813475e-05,
      "loss": 2.1302,
      "step": 169844
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.5170094966888428,
      "learning_rate": 9.659181943570378e-05,
      "loss": 1.9474,
      "step": 169845
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.2531241178512573,
      "learning_rate": 9.658881274109013e-05,
      "loss": 1.8729,
      "step": 169846
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.4169450998306274,
      "learning_rate": 9.658580608429466e-05,
      "loss": 2.0812,
      "step": 169847
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.214794635772705,
      "learning_rate": 9.65827994653175e-05,
      "loss": 1.9302,
      "step": 169848
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.2356314659118652,
      "learning_rate": 9.65797928841596e-05,
      "loss": 1.815,
      "step": 169849
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.353895664215088,
      "learning_rate": 9.657678634082121e-05,
      "loss": 2.158,
      "step": 169850
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.3263343572616577,
      "learning_rate": 9.657377983530313e-05,
      "loss": 2.2134,
      "step": 169851
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.2154594659805298,
      "learning_rate": 9.657077336760577e-05,
      "loss": 1.8464,
      "step": 169852
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.2827430963516235,
      "learning_rate": 9.656776693772992e-05,
      "loss": 1.8869,
      "step": 169853
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.2656983137130737,
      "learning_rate": 9.656476054567575e-05,
      "loss": 1.8408,
      "step": 169854
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.2976133823394775,
      "learning_rate": 9.656175419144415e-05,
      "loss": 1.9837,
      "step": 169855
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.219449520111084,
      "learning_rate": 9.655874787503544e-05,
      "loss": 2.315,
      "step": 169856
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.3762588500976562,
      "learning_rate": 9.65557415964504e-05,
      "loss": 1.8955,
      "step": 169857
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.3748271465301514,
      "learning_rate": 9.65527353556894e-05,
      "loss": 2.1063,
      "step": 169858
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.27260422706604,
      "learning_rate": 9.654972915275329e-05,
      "loss": 2.1763,
      "step": 169859
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.241351842880249,
      "learning_rate": 9.654672298764221e-05,
      "loss": 2.0836,
      "step": 169860
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.2198108434677124,
      "learning_rate": 9.654371686035705e-05,
      "loss": 2.3275,
      "step": 169861
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.1609829664230347,
      "learning_rate": 9.654071077089816e-05,
      "loss": 1.9065,
      "step": 169862
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.3619780540466309,
      "learning_rate": 9.653770471926627e-05,
      "loss": 2.1071,
      "step": 169863
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.2246224880218506,
      "learning_rate": 9.653469870546177e-05,
      "loss": 2.1702,
      "step": 169864
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.1916803121566772,
      "learning_rate": 9.65316927294855e-05,
      "loss": 1.8841,
      "step": 169865
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.2786868810653687,
      "learning_rate": 9.652868679133768e-05,
      "loss": 2.1674,
      "step": 169866
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.221850037574768,
      "learning_rate": 9.652568089101905e-05,
      "loss": 1.9204,
      "step": 169867
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.234824776649475,
      "learning_rate": 9.652267502853007e-05,
      "loss": 1.9436,
      "step": 169868
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.3177576065063477,
      "learning_rate": 9.651966920387146e-05,
      "loss": 2.1535,
      "step": 169869
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.1934372186660767,
      "learning_rate": 9.65166634170436e-05,
      "loss": 1.8891,
      "step": 169870
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.2379300594329834,
      "learning_rate": 9.651365766804729e-05,
      "loss": 1.9855,
      "step": 169871
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.2715606689453125,
      "learning_rate": 9.651065195688276e-05,
      "loss": 2.0403,
      "step": 169872
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.2795110940933228,
      "learning_rate": 9.650764628355088e-05,
      "loss": 2.081,
      "step": 169873
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.1711591482162476,
      "learning_rate": 9.650464064805192e-05,
      "loss": 2.2456,
      "step": 169874
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.3100920915603638,
      "learning_rate": 9.650163505038671e-05,
      "loss": 2.0371,
      "step": 169875
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.187556266784668,
      "learning_rate": 9.649862949055556e-05,
      "loss": 2.1604,
      "step": 169876
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.249412178993225,
      "learning_rate": 9.649562396855937e-05,
      "loss": 2.0942,
      "step": 169877
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.1736050844192505,
      "learning_rate": 9.64926184843983e-05,
      "loss": 1.8764,
      "step": 169878
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.3162741661071777,
      "learning_rate": 9.648961303807319e-05,
      "loss": 2.304,
      "step": 169879
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.3073533773422241,
      "learning_rate": 9.648660762958438e-05,
      "loss": 1.9285,
      "step": 169880
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.4504114389419556,
      "learning_rate": 9.648360225893267e-05,
      "loss": 1.9953,
      "step": 169881
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.2078708410263062,
      "learning_rate": 9.64805969261184e-05,
      "loss": 1.8747,
      "step": 169882
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.2562158107757568,
      "learning_rate": 9.647759163114232e-05,
      "loss": 2.1156,
      "step": 169883
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.3568382263183594,
      "learning_rate": 9.64745863740049e-05,
      "loss": 2.0701,
      "step": 169884
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.439323902130127,
      "learning_rate": 9.647158115470672e-05,
      "loss": 1.858,
      "step": 169885
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.321075439453125,
      "learning_rate": 9.646857597324818e-05,
      "loss": 2.0999,
      "step": 169886
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.315268874168396,
      "learning_rate": 9.646557082963008e-05,
      "loss": 1.8675,
      "step": 169887
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.3101550340652466,
      "learning_rate": 9.64625657238528e-05,
      "loss": 1.9988,
      "step": 169888
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.2640305757522583,
      "learning_rate": 9.645956065591704e-05,
      "loss": 1.8657,
      "step": 169889
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.2843271493911743,
      "learning_rate": 9.645655562582332e-05,
      "loss": 1.9741,
      "step": 169890
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.2006251811981201,
      "learning_rate": 9.645355063357212e-05,
      "loss": 2.126,
      "step": 169891
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.2363120317459106,
      "learning_rate": 9.645054567916398e-05,
      "loss": 2.1154,
      "step": 169892
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.2690811157226562,
      "learning_rate": 9.644754076259964e-05,
      "loss": 2.1803,
      "step": 169893
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.2575637102127075,
      "learning_rate": 9.644453588387943e-05,
      "loss": 1.9601,
      "step": 169894
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.2441884279251099,
      "learning_rate": 9.64415310430041e-05,
      "loss": 1.8842,
      "step": 169895
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.2446662187576294,
      "learning_rate": 9.643852623997418e-05,
      "loss": 2.0867,
      "step": 169896
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.21505868434906,
      "learning_rate": 9.643552147479005e-05,
      "loss": 1.888,
      "step": 169897
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.4220411777496338,
      "learning_rate": 9.64325167474525e-05,
      "loss": 2.1988,
      "step": 169898
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.235723853111267,
      "learning_rate": 9.642951205796199e-05,
      "loss": 1.9481,
      "step": 169899
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.3639904260635376,
      "learning_rate": 9.642650740631896e-05,
      "loss": 2.1191,
      "step": 169900
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.4284366369247437,
      "learning_rate": 9.64235027925242e-05,
      "loss": 1.9828,
      "step": 169901
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.1280301809310913,
      "learning_rate": 9.642049821657816e-05,
      "loss": 2.1077,
      "step": 169902
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.1805343627929688,
      "learning_rate": 9.641749367848127e-05,
      "loss": 2.0161,
      "step": 169903
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.3552290201187134,
      "learning_rate": 9.641448917823434e-05,
      "loss": 2.0659,
      "step": 169904
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.3681628704071045,
      "learning_rate": 9.64114847158378e-05,
      "loss": 2.0356,
      "step": 169905
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.3069703578948975,
      "learning_rate": 9.640848029129209e-05,
      "loss": 1.991,
      "step": 169906
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.3484752178192139,
      "learning_rate": 9.640547590459799e-05,
      "loss": 1.8768,
      "step": 169907
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.1659033298492432,
      "learning_rate": 9.640247155575595e-05,
      "loss": 2.2457,
      "step": 169908
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.3041070699691772,
      "learning_rate": 9.639946724476643e-05,
      "loss": 2.0096,
      "step": 169909
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.1277399063110352,
      "learning_rate": 9.63964629716302e-05,
      "loss": 1.9878,
      "step": 169910
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.4493647813796997,
      "learning_rate": 9.639345873634762e-05,
      "loss": 2.174,
      "step": 169911
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.1642392873764038,
      "learning_rate": 9.639045453891943e-05,
      "loss": 1.9765,
      "step": 169912
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.1374591588974,
      "learning_rate": 9.638745037934613e-05,
      "loss": 1.9296,
      "step": 169913
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.361222743988037,
      "learning_rate": 9.638444625762821e-05,
      "loss": 2.0943,
      "step": 169914
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.2849470376968384,
      "learning_rate": 9.638144217376617e-05,
      "loss": 2.2052,
      "step": 169915
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.3341093063354492,
      "learning_rate": 9.637843812776076e-05,
      "loss": 2.0649,
      "step": 169916
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.4656704664230347,
      "learning_rate": 9.637543411961237e-05,
      "loss": 1.9028,
      "step": 169917
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.2465895414352417,
      "learning_rate": 9.637243014932168e-05,
      "loss": 2.1963,
      "step": 169918
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.1895751953125,
      "learning_rate": 9.636942621688915e-05,
      "loss": 1.8881,
      "step": 169919
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.3240429162979126,
      "learning_rate": 9.636642232231557e-05,
      "loss": 2.042,
      "step": 169920
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.2382031679153442,
      "learning_rate": 9.636341846560109e-05,
      "loss": 2.0287,
      "step": 169921
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.2177493572235107,
      "learning_rate": 9.63604146467466e-05,
      "loss": 1.8718,
      "step": 169922
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.2651863098144531,
      "learning_rate": 9.63574108657525e-05,
      "loss": 1.917,
      "step": 169923
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.246091604232788,
      "learning_rate": 9.635440712261945e-05,
      "loss": 1.8675,
      "step": 169924
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.2801613807678223,
      "learning_rate": 9.635140341734791e-05,
      "loss": 1.7818,
      "step": 169925
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.5313857793807983,
      "learning_rate": 9.634839974993864e-05,
      "loss": 2.1982,
      "step": 169926
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.3194133043289185,
      "learning_rate": 9.634539612039186e-05,
      "loss": 1.9854,
      "step": 169927
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.348137378692627,
      "learning_rate": 9.634239252870842e-05,
      "loss": 2.0836,
      "step": 169928
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.2444398403167725,
      "learning_rate": 9.633938897488866e-05,
      "loss": 2.127,
      "step": 169929
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.364960789680481,
      "learning_rate": 9.633638545893339e-05,
      "loss": 2.0539,
      "step": 169930
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.2792439460754395,
      "learning_rate": 9.633338198084289e-05,
      "loss": 1.8303,
      "step": 169931
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.4000564813613892,
      "learning_rate": 9.633037854061809e-05,
      "loss": 2.1475,
      "step": 169932
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.1665061712265015,
      "learning_rate": 9.632737513825911e-05,
      "loss": 1.8017,
      "step": 169933
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.2744042873382568,
      "learning_rate": 9.632437177376681e-05,
      "loss": 1.9388,
      "step": 169934
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.1858144998550415,
      "learning_rate": 9.632136844714155e-05,
      "loss": 2.3285,
      "step": 169935
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.9286632537841797,
      "learning_rate": 9.631836515838409e-05,
      "loss": 1.8392,
      "step": 169936
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.2211734056472778,
      "learning_rate": 9.631536190749478e-05,
      "loss": 1.919,
      "step": 169937
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.2847315073013306,
      "learning_rate": 9.631235869447451e-05,
      "loss": 2.0255,
      "step": 169938
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.3202787637710571,
      "learning_rate": 9.630935551932341e-05,
      "loss": 1.9734,
      "step": 169939
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.342737078666687,
      "learning_rate": 9.630635238204232e-05,
      "loss": 2.0191,
      "step": 169940
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.338062047958374,
      "learning_rate": 9.630334928263168e-05,
      "loss": 2.2401,
      "step": 169941
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.3999545574188232,
      "learning_rate": 9.630034622109214e-05,
      "loss": 1.8474,
      "step": 169942
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.2872076034545898,
      "learning_rate": 9.629734319742414e-05,
      "loss": 2.1353,
      "step": 169943
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.2243661880493164,
      "learning_rate": 9.62943402116285e-05,
      "loss": 2.0669,
      "step": 169944
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.281921148300171,
      "learning_rate": 9.629133726370541e-05,
      "loss": 1.9627,
      "step": 169945
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.262274146080017,
      "learning_rate": 9.62883343536557e-05,
      "loss": 1.8425,
      "step": 169946
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.53175950050354,
      "learning_rate": 9.62853314814797e-05,
      "loss": 1.9263,
      "step": 169947
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.3090236186981201,
      "learning_rate": 9.628232864717824e-05,
      "loss": 1.9911,
      "step": 169948
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.1561610698699951,
      "learning_rate": 9.627932585075162e-05,
      "loss": 2.0712,
      "step": 169949
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.268550992012024,
      "learning_rate": 9.627632309220062e-05,
      "loss": 2.1079,
      "step": 169950
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.2439132928848267,
      "learning_rate": 9.627332037152567e-05,
      "loss": 2.0326,
      "step": 169951
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.3930957317352295,
      "learning_rate": 9.62703176887274e-05,
      "loss": 1.9966,
      "step": 169952
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.226506233215332,
      "learning_rate": 9.62673150438062e-05,
      "loss": 1.958,
      "step": 169953
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.162902593612671,
      "learning_rate": 9.626431243676287e-05,
      "loss": 2.1406,
      "step": 169954
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.3413676023483276,
      "learning_rate": 9.626130986759773e-05,
      "loss": 2.2778,
      "step": 169955
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.4759072065353394,
      "learning_rate": 9.625830733631156e-05,
      "loss": 2.2203,
      "step": 169956
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.1066670417785645,
      "learning_rate": 9.625530484290483e-05,
      "loss": 2.1276,
      "step": 169957
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.2896010875701904,
      "learning_rate": 9.62523023873781e-05,
      "loss": 1.9572,
      "step": 169958
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.2250769138336182,
      "learning_rate": 9.62492999697318e-05,
      "loss": 2.3133,
      "step": 169959
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.2504761219024658,
      "learning_rate": 9.62462975899667e-05,
      "loss": 2.1784,
      "step": 169960
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.27815580368042,
      "learning_rate": 9.624329524808312e-05,
      "loss": 2.1222,
      "step": 169961
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.2354742288589478,
      "learning_rate": 9.624029294408192e-05,
      "loss": 2.058,
      "step": 169962
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.1884669065475464,
      "learning_rate": 9.623729067796346e-05,
      "loss": 2.1008,
      "step": 169963
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.407968282699585,
      "learning_rate": 9.623428844972831e-05,
      "loss": 2.0955,
      "step": 169964
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.2547940015792847,
      "learning_rate": 9.6231286259377e-05,
      "loss": 1.9131,
      "step": 169965
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.1326781511306763,
      "learning_rate": 9.622828410691024e-05,
      "loss": 1.9761,
      "step": 169966
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.2899943590164185,
      "learning_rate": 9.622528199232837e-05,
      "loss": 1.892,
      "step": 169967
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.2167646884918213,
      "learning_rate": 9.622227991563218e-05,
      "loss": 1.8942,
      "step": 169968
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.1848111152648926,
      "learning_rate": 9.621927787682207e-05,
      "loss": 2.1341,
      "step": 169969
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.7682116031646729,
      "learning_rate": 9.621627587589868e-05,
      "loss": 2.1019,
      "step": 169970
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.3186099529266357,
      "learning_rate": 9.621327391286243e-05,
      "loss": 2.1318,
      "step": 169971
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.2534453868865967,
      "learning_rate": 9.621027198771408e-05,
      "loss": 1.9397,
      "step": 169972
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.1951518058776855,
      "learning_rate": 9.620727010045397e-05,
      "loss": 2.0153,
      "step": 169973
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.2634038925170898,
      "learning_rate": 9.620426825108289e-05,
      "loss": 2.1464,
      "step": 169974
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.2455168962478638,
      "learning_rate": 9.620126643960128e-05,
      "loss": 1.8713,
      "step": 169975
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.3092845678329468,
      "learning_rate": 9.619826466600972e-05,
      "loss": 1.8365,
      "step": 169976
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.4513907432556152,
      "learning_rate": 9.619526293030861e-05,
      "loss": 2.1616,
      "step": 169977
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.4822099208831787,
      "learning_rate": 9.61922612324988e-05,
      "loss": 1.7876,
      "step": 169978
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.2098346948623657,
      "learning_rate": 9.618925957258055e-05,
      "loss": 1.8677,
      "step": 169979
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.501379132270813,
      "learning_rate": 9.618625795055468e-05,
      "loss": 1.9026,
      "step": 169980
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.470482349395752,
      "learning_rate": 9.618325636642163e-05,
      "loss": 1.9142,
      "step": 169981
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.2676117420196533,
      "learning_rate": 9.618025482018187e-05,
      "loss": 1.759,
      "step": 169982
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.3217977285385132,
      "learning_rate": 9.617725331183614e-05,
      "loss": 1.9748,
      "step": 169983
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.2318084239959717,
      "learning_rate": 9.61742518413849e-05,
      "loss": 1.8528,
      "step": 169984
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.2378093004226685,
      "learning_rate": 9.617125040882864e-05,
      "loss": 2.0073,
      "step": 169985
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.2783881425857544,
      "learning_rate": 9.616824901416807e-05,
      "loss": 1.8534,
      "step": 169986
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.3641022443771362,
      "learning_rate": 9.616524765740367e-05,
      "loss": 2.1085,
      "step": 169987
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.3598467111587524,
      "learning_rate": 9.61622463385359e-05,
      "loss": 2.1353,
      "step": 169988
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.2346611022949219,
      "learning_rate": 9.615924505756552e-05,
      "loss": 2.1332,
      "step": 169989
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.1848810911178589,
      "learning_rate": 9.6156243814493e-05,
      "loss": 1.9761,
      "step": 169990
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.3847169876098633,
      "learning_rate": 9.615324260931874e-05,
      "loss": 1.8454,
      "step": 169991
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.2289644479751587,
      "learning_rate": 9.615024144204358e-05,
      "loss": 2.0812,
      "step": 169992
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.26504385471344,
      "learning_rate": 9.61472403126679e-05,
      "loss": 2.1704,
      "step": 169993
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.1989463567733765,
      "learning_rate": 9.614423922119225e-05,
      "loss": 1.8912,
      "step": 169994
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.4326649904251099,
      "learning_rate": 9.61412381676173e-05,
      "loss": 1.9722,
      "step": 169995
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.2554959058761597,
      "learning_rate": 9.613823715194343e-05,
      "loss": 2.061,
      "step": 169996
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.307563066482544,
      "learning_rate": 9.613523617417142e-05,
      "loss": 2.2172,
      "step": 169997
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.3330457210540771,
      "learning_rate": 9.613223523430172e-05,
      "loss": 1.7523,
      "step": 169998
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.3261207342147827,
      "learning_rate": 9.61292343323349e-05,
      "loss": 2.1011,
      "step": 169999
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.198190450668335,
      "learning_rate": 9.612623346827139e-05,
      "loss": 2.0951,
      "step": 170000
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.1180930137634277,
      "learning_rate": 9.612323264211196e-05,
      "loss": 2.0738,
      "step": 170001
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.3443681001663208,
      "learning_rate": 9.612023185385698e-05,
      "loss": 2.1704,
      "step": 170002
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.4290320873260498,
      "learning_rate": 9.611723110350717e-05,
      "loss": 2.0475,
      "step": 170003
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.369531512260437,
      "learning_rate": 9.611423039106295e-05,
      "loss": 2.1195,
      "step": 170004
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.315220832824707,
      "learning_rate": 9.61112297165251e-05,
      "loss": 2.1529,
      "step": 170005
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.3201731443405151,
      "learning_rate": 9.610822907989387e-05,
      "loss": 2.0529,
      "step": 170006
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.6830337047576904,
      "learning_rate": 9.610522848117005e-05,
      "loss": 1.9441,
      "step": 170007
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.23439359664917,
      "learning_rate": 9.6102227920354e-05,
      "loss": 1.9962,
      "step": 170008
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.3028504848480225,
      "learning_rate": 9.609922739744654e-05,
      "loss": 2.305,
      "step": 170009
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.2010000944137573,
      "learning_rate": 9.609622691244793e-05,
      "loss": 2.0071,
      "step": 170010
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.2637484073638916,
      "learning_rate": 9.609322646535912e-05,
      "loss": 1.9015,
      "step": 170011
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.252359390258789,
      "learning_rate": 9.609022605618021e-05,
      "loss": 2.0269,
      "step": 170012
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.406622052192688,
      "learning_rate": 9.608722568491207e-05,
      "loss": 1.9654,
      "step": 170013
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.2978827953338623,
      "learning_rate": 9.608422535155508e-05,
      "loss": 2.4381,
      "step": 170014
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.5014880895614624,
      "learning_rate": 9.608122505610996e-05,
      "loss": 2.1451,
      "step": 170015
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.8650225400924683,
      "learning_rate": 9.60782247985771e-05,
      "loss": 1.9031,
      "step": 170016
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.2266759872436523,
      "learning_rate": 9.607522457895727e-05,
      "loss": 1.9945,
      "step": 170017
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.3504700660705566,
      "learning_rate": 9.607222439725086e-05,
      "loss": 2.1398,
      "step": 170018
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.2114390134811401,
      "learning_rate": 9.60692242534585e-05,
      "loss": 2.0884,
      "step": 170019
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.1695420742034912,
      "learning_rate": 9.606622414758061e-05,
      "loss": 1.9607,
      "step": 170020
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.4119746685028076,
      "learning_rate": 9.606322407961798e-05,
      "loss": 2.039,
      "step": 170021
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.2807085514068604,
      "learning_rate": 9.606022404957093e-05,
      "loss": 1.9662,
      "step": 170022
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.283659815788269,
      "learning_rate": 9.605722405744025e-05,
      "loss": 1.9998,
      "step": 170023
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.2247675657272339,
      "learning_rate": 9.605422410322635e-05,
      "loss": 2.1314,
      "step": 170024
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.2698309421539307,
      "learning_rate": 9.605122418692984e-05,
      "loss": 1.9611,
      "step": 170025
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.2264779806137085,
      "learning_rate": 9.604822430855116e-05,
      "loss": 1.9632,
      "step": 170026
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.3460851907730103,
      "learning_rate": 9.604522446809105e-05,
      "loss": 1.9269,
      "step": 170027
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.4994399547576904,
      "learning_rate": 9.604222466554992e-05,
      "loss": 2.0658,
      "step": 170028
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.1676069498062134,
      "learning_rate": 9.603922490092844e-05,
      "loss": 2.1133,
      "step": 170029
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.328398585319519,
      "learning_rate": 9.603622517422714e-05,
      "loss": 2.1106,
      "step": 170030
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.2858788967132568,
      "learning_rate": 9.603322548544658e-05,
      "loss": 1.899,
      "step": 170031
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.2644518613815308,
      "learning_rate": 9.603022583458717e-05,
      "loss": 1.8939,
      "step": 170032
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.1765108108520508,
      "learning_rate": 9.602722622164967e-05,
      "loss": 2.1655,
      "step": 170033
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.2778385877609253,
      "learning_rate": 9.60242266466345e-05,
      "loss": 2.1813,
      "step": 170034
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.3047717809677124,
      "learning_rate": 9.602122710954235e-05,
      "loss": 1.9055,
      "step": 170035
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.2458611726760864,
      "learning_rate": 9.60182276103737e-05,
      "loss": 2.04,
      "step": 170036
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.2936609983444214,
      "learning_rate": 9.601522814912913e-05,
      "loss": 2.1415,
      "step": 170037
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.3458960056304932,
      "learning_rate": 9.601222872580909e-05,
      "loss": 2.2521,
      "step": 170038
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.1764557361602783,
      "learning_rate": 9.600922934041429e-05,
      "loss": 2.1004,
      "step": 170039
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.2395011186599731,
      "learning_rate": 9.600622999294517e-05,
      "loss": 1.9801,
      "step": 170040
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.338010549545288,
      "learning_rate": 9.600323068340242e-05,
      "loss": 1.9624,
      "step": 170041
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.2078243494033813,
      "learning_rate": 9.60002314117865e-05,
      "loss": 1.9204,
      "step": 170042
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.49977445602417,
      "learning_rate": 9.5997232178098e-05,
      "loss": 1.9987,
      "step": 170043
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.2898050546646118,
      "learning_rate": 9.599423298233739e-05,
      "loss": 2.0151,
      "step": 170044
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.2074828147888184,
      "learning_rate": 9.599123382450538e-05,
      "loss": 2.0695,
      "step": 170045
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.19218111038208,
      "learning_rate": 9.598823470460235e-05,
      "loss": 2.0011,
      "step": 170046
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.5318706035614014,
      "learning_rate": 9.598523562262908e-05,
      "loss": 1.9406,
      "step": 170047
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.263134479522705,
      "learning_rate": 9.598223657858598e-05,
      "loss": 2.331,
      "step": 170048
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.2570972442626953,
      "learning_rate": 9.597923757247363e-05,
      "loss": 1.9842,
      "step": 170049
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.2968233823776245,
      "learning_rate": 9.597623860429251e-05,
      "loss": 1.8301,
      "step": 170050
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.2760199308395386,
      "learning_rate": 9.597323967404335e-05,
      "loss": 1.9271,
      "step": 170051
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.628914475440979,
      "learning_rate": 9.597024078172654e-05,
      "loss": 2.0633,
      "step": 170052
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.4836195707321167,
      "learning_rate": 9.596724192734278e-05,
      "loss": 2.0341,
      "step": 170053
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.2443389892578125,
      "learning_rate": 9.596424311089256e-05,
      "loss": 2.0487,
      "step": 170054
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.184301495552063,
      "learning_rate": 9.596124433237645e-05,
      "loss": 2.0361,
      "step": 170055
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.2487869262695312,
      "learning_rate": 9.59582455917949e-05,
      "loss": 2.0031,
      "step": 170056
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.6527220010757446,
      "learning_rate": 9.595524688914868e-05,
      "loss": 1.9453,
      "step": 170057
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.2223371267318726,
      "learning_rate": 9.59522482244381e-05,
      "loss": 2.0357,
      "step": 170058
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.2721501588821411,
      "learning_rate": 9.594924959766397e-05,
      "loss": 2.0435,
      "step": 170059
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.445001482963562,
      "learning_rate": 9.594625100882671e-05,
      "loss": 2.0495,
      "step": 170060
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.3308532238006592,
      "learning_rate": 9.59432524579269e-05,
      "loss": 1.8873,
      "step": 170061
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.4303090572357178,
      "learning_rate": 9.594025394496501e-05,
      "loss": 2.1216,
      "step": 170062
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.2458559274673462,
      "learning_rate": 9.593725546994176e-05,
      "loss": 1.9597,
      "step": 170063
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.240795373916626,
      "learning_rate": 9.59342570328575e-05,
      "loss": 1.8755,
      "step": 170064
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.2986021041870117,
      "learning_rate": 9.593125863371307e-05,
      "loss": 2.1371,
      "step": 170065
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.2991044521331787,
      "learning_rate": 9.592826027250882e-05,
      "loss": 1.9329,
      "step": 170066
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.3249281644821167,
      "learning_rate": 9.592526194924538e-05,
      "loss": 2.2047,
      "step": 170067
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.2490817308425903,
      "learning_rate": 9.59222636639232e-05,
      "loss": 2.0089,
      "step": 170068
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.2911417484283447,
      "learning_rate": 9.591926541654302e-05,
      "loss": 2.0988,
      "step": 170069
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.4950814247131348,
      "learning_rate": 9.591626720710516e-05,
      "loss": 1.9779,
      "step": 170070
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.362874984741211,
      "learning_rate": 9.591326903561047e-05,
      "loss": 2.0602,
      "step": 170071
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.217006802558899,
      "learning_rate": 9.591027090205934e-05,
      "loss": 2.0934,
      "step": 170072
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.245293378829956,
      "learning_rate": 9.590727280645223e-05,
      "loss": 1.7737,
      "step": 170073
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.2776931524276733,
      "learning_rate": 9.590427474878993e-05,
      "loss": 1.8723,
      "step": 170074
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.244959831237793,
      "learning_rate": 9.590127672907285e-05,
      "loss": 2.1275,
      "step": 170075
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.3488837480545044,
      "learning_rate": 9.589827874730152e-05,
      "loss": 2.1219,
      "step": 170076
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.2223261594772339,
      "learning_rate": 9.589528080347662e-05,
      "loss": 1.8149,
      "step": 170077
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.3677176237106323,
      "learning_rate": 9.589228289759865e-05,
      "loss": 1.9526,
      "step": 170078
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.1737651824951172,
      "learning_rate": 9.588928502966806e-05,
      "loss": 2.1254,
      "step": 170079
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.232680320739746,
      "learning_rate": 9.588628719968557e-05,
      "loss": 1.9722,
      "step": 170080
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.2369316816329956,
      "learning_rate": 9.588328940765161e-05,
      "loss": 1.9126,
      "step": 170081
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.271277666091919,
      "learning_rate": 9.588029165356688e-05,
      "loss": 2.0242,
      "step": 170082
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.2906901836395264,
      "learning_rate": 9.587729393743187e-05,
      "loss": 2.2907,
      "step": 170083
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.211848258972168,
      "learning_rate": 9.587429625924702e-05,
      "loss": 1.938,
      "step": 170084
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.1514050960540771,
      "learning_rate": 9.587129861901308e-05,
      "loss": 2.1348,
      "step": 170085
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.193103313446045,
      "learning_rate": 9.586830101673056e-05,
      "loss": 1.9264,
      "step": 170086
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.3428575992584229,
      "learning_rate": 9.586530345239984e-05,
      "loss": 2.0527,
      "step": 170087
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.2974941730499268,
      "learning_rate": 9.586230592602172e-05,
      "loss": 2.0835,
      "step": 170088
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.2129348516464233,
      "learning_rate": 9.585930843759665e-05,
      "loss": 1.8901,
      "step": 170089
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.4213435649871826,
      "learning_rate": 9.585631098712509e-05,
      "loss": 2.0688,
      "step": 170090
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.309194564819336,
      "learning_rate": 9.585331357460779e-05,
      "loss": 2.0133,
      "step": 170091
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.296830177307129,
      "learning_rate": 9.58503162000452e-05,
      "loss": 2.0543,
      "step": 170092
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.3608070611953735,
      "learning_rate": 9.584731886343783e-05,
      "loss": 2.1108,
      "step": 170093
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.1897315979003906,
      "learning_rate": 9.584432156478635e-05,
      "loss": 2.1993,
      "step": 170094
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.3628566265106201,
      "learning_rate": 9.584132430409121e-05,
      "loss": 1.9117,
      "step": 170095
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.3054802417755127,
      "learning_rate": 9.583832708135308e-05,
      "loss": 1.9714,
      "step": 170096
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.215790867805481,
      "learning_rate": 9.583532989657248e-05,
      "loss": 2.0276,
      "step": 170097
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.2536171674728394,
      "learning_rate": 9.583233274974997e-05,
      "loss": 2.1198,
      "step": 170098
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.1983773708343506,
      "learning_rate": 9.582933564088594e-05,
      "loss": 2.0229,
      "step": 170099
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.3033201694488525,
      "learning_rate": 9.582633856998121e-05,
      "loss": 2.1156,
      "step": 170100
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.3130208253860474,
      "learning_rate": 9.58233415370361e-05,
      "loss": 2.0395,
      "step": 170101
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.208912968635559,
      "learning_rate": 9.582034454205142e-05,
      "loss": 2.0458,
      "step": 170102
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.2852128744125366,
      "learning_rate": 9.581734758502757e-05,
      "loss": 2.1457,
      "step": 170103
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.1810810565948486,
      "learning_rate": 9.581435066596513e-05,
      "loss": 2.0713,
      "step": 170104
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.1506984233856201,
      "learning_rate": 9.581135378486454e-05,
      "loss": 2.0328,
      "step": 170105
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.2075273990631104,
      "learning_rate": 9.58083569417266e-05,
      "loss": 1.9695,
      "step": 170106
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.3137893676757812,
      "learning_rate": 9.580536013655165e-05,
      "loss": 2.2173,
      "step": 170107
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.2353492975234985,
      "learning_rate": 9.580236336934042e-05,
      "loss": 1.8143,
      "step": 170108
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.1982585191726685,
      "learning_rate": 9.579936664009338e-05,
      "loss": 1.9441,
      "step": 170109
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.3479127883911133,
      "learning_rate": 9.579636994881109e-05,
      "loss": 2.113,
      "step": 170110
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.3141449689865112,
      "learning_rate": 9.579337329549404e-05,
      "loss": 1.9083,
      "step": 170111
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.23691987991333,
      "learning_rate": 9.57903766801429e-05,
      "loss": 2.1985,
      "step": 170112
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.2663017511367798,
      "learning_rate": 9.578738010275814e-05,
      "loss": 1.9787,
      "step": 170113
    },
    {
      "epoch": 2.21,
      "grad_norm": 1.250063180923462,
      "learning_rate": 9.578438356334045e-05,
      "loss": 2.1728,
      "step": 170114
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.3285695314407349,
      "learning_rate": 9.578138706189028e-05,
      "loss": 1.9732,
      "step": 170115
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.2935031652450562,
      "learning_rate": 9.577839059840821e-05,
      "loss": 1.9112,
      "step": 170116
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.2001432180404663,
      "learning_rate": 9.577539417289469e-05,
      "loss": 2.0376,
      "step": 170117
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.1790876388549805,
      "learning_rate": 9.577239778535051e-05,
      "loss": 2.126,
      "step": 170118
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.1404329538345337,
      "learning_rate": 9.576940143577595e-05,
      "loss": 2.1088,
      "step": 170119
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.1965177059173584,
      "learning_rate": 9.576640512417186e-05,
      "loss": 1.8174,
      "step": 170120
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.2925844192504883,
      "learning_rate": 9.576340885053861e-05,
      "loss": 2.2359,
      "step": 170121
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.2624157667160034,
      "learning_rate": 9.57604126148768e-05,
      "loss": 1.9959,
      "step": 170122
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.1989771127700806,
      "learning_rate": 9.575741641718688e-05,
      "loss": 1.9885,
      "step": 170123
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.1743178367614746,
      "learning_rate": 9.575442025746966e-05,
      "loss": 2.1056,
      "step": 170124
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.3158361911773682,
      "learning_rate": 9.575142413572541e-05,
      "loss": 2.1636,
      "step": 170125
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.367117166519165,
      "learning_rate": 9.574842805195494e-05,
      "loss": 2.0182,
      "step": 170126
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.3997876644134521,
      "learning_rate": 9.574543200615868e-05,
      "loss": 2.2808,
      "step": 170127
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.3043620586395264,
      "learning_rate": 9.574243599833718e-05,
      "loss": 2.0966,
      "step": 170128
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.2905694246292114,
      "learning_rate": 9.573944002849094e-05,
      "loss": 1.8744,
      "step": 170129
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.3510433435440063,
      "learning_rate": 9.573644409662072e-05,
      "loss": 1.966,
      "step": 170130
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.4907101392745972,
      "learning_rate": 9.573344820272683e-05,
      "loss": 2.0687,
      "step": 170131
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.2824043035507202,
      "learning_rate": 9.573045234681005e-05,
      "loss": 1.981,
      "step": 170132
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.4586457014083862,
      "learning_rate": 9.572745652887083e-05,
      "loss": 1.9261,
      "step": 170133
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.2731839418411255,
      "learning_rate": 9.572446074890972e-05,
      "loss": 1.82,
      "step": 170134
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.624660849571228,
      "learning_rate": 9.572146500692721e-05,
      "loss": 2.2184,
      "step": 170135
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.2810322046279907,
      "learning_rate": 9.571846930292402e-05,
      "loss": 2.0703,
      "step": 170136
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.314087986946106,
      "learning_rate": 9.571547363690051e-05,
      "loss": 2.0396,
      "step": 170137
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.1866662502288818,
      "learning_rate": 9.571247800885748e-05,
      "loss": 2.2582,
      "step": 170138
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.2319223880767822,
      "learning_rate": 9.570948241879534e-05,
      "loss": 2.0497,
      "step": 170139
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.3784395456314087,
      "learning_rate": 9.570648686671466e-05,
      "loss": 2.3807,
      "step": 170140
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.3180601596832275,
      "learning_rate": 9.570349135261592e-05,
      "loss": 2.2071,
      "step": 170141
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.580407977104187,
      "learning_rate": 9.570049587649984e-05,
      "loss": 2.0882,
      "step": 170142
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.2363144159317017,
      "learning_rate": 9.56975004383668e-05,
      "loss": 2.1413,
      "step": 170143
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.2988444566726685,
      "learning_rate": 9.569450503821755e-05,
      "loss": 1.9349,
      "step": 170144
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.7890430688858032,
      "learning_rate": 9.569150967605255e-05,
      "loss": 2.0886,
      "step": 170145
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.2726284265518188,
      "learning_rate": 9.568851435187236e-05,
      "loss": 2.2257,
      "step": 170146
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.3168740272521973,
      "learning_rate": 9.568551906567742e-05,
      "loss": 1.8469,
      "step": 170147
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.359904408454895,
      "learning_rate": 9.56825238174685e-05,
      "loss": 1.7902,
      "step": 170148
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.2527084350585938,
      "learning_rate": 9.567952860724598e-05,
      "loss": 2.2396,
      "step": 170149
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.367418646812439,
      "learning_rate": 9.567653343501054e-05,
      "loss": 2.0515,
      "step": 170150
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.2953859567642212,
      "learning_rate": 9.567353830076262e-05,
      "loss": 2.0309,
      "step": 170151
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.3930939435958862,
      "learning_rate": 9.567054320450306e-05,
      "loss": 2.077,
      "step": 170152
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.1569613218307495,
      "learning_rate": 9.5667548146232e-05,
      "loss": 2.1767,
      "step": 170153
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.3171018362045288,
      "learning_rate": 9.566455312595029e-05,
      "loss": 2.008,
      "step": 170154
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.2838163375854492,
      "learning_rate": 9.566155814365832e-05,
      "loss": 2.0679,
      "step": 170155
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.294761061668396,
      "learning_rate": 9.565856319935679e-05,
      "loss": 1.9561,
      "step": 170156
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.4969990253448486,
      "learning_rate": 9.56555682930461e-05,
      "loss": 2.0769,
      "step": 170157
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.2254425287246704,
      "learning_rate": 9.5652573424727e-05,
      "loss": 2.0456,
      "step": 170158
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.3577438592910767,
      "learning_rate": 9.564957859439994e-05,
      "loss": 2.0575,
      "step": 170159
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.340722680091858,
      "learning_rate": 9.564658380206549e-05,
      "loss": 1.85,
      "step": 170160
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.3415992259979248,
      "learning_rate": 9.564358904772407e-05,
      "loss": 1.9802,
      "step": 170161
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.2583528757095337,
      "learning_rate": 9.56405943313765e-05,
      "loss": 2.1505,
      "step": 170162
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.2851728200912476,
      "learning_rate": 9.56375996530231e-05,
      "loss": 2.1337,
      "step": 170163
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.2698640823364258,
      "learning_rate": 9.563460501266461e-05,
      "loss": 1.7899,
      "step": 170164
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.1992640495300293,
      "learning_rate": 9.56316104103015e-05,
      "loss": 1.9291,
      "step": 170165
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.2095896005630493,
      "learning_rate": 9.562861584593425e-05,
      "loss": 2.1612,
      "step": 170166
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.278504490852356,
      "learning_rate": 9.562562131956358e-05,
      "loss": 1.9098,
      "step": 170167
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.2246983051300049,
      "learning_rate": 9.562262683118998e-05,
      "loss": 1.8678,
      "step": 170168
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.1920770406723022,
      "learning_rate": 9.56196323808139e-05,
      "loss": 2.27,
      "step": 170169
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.2438068389892578,
      "learning_rate": 9.561663796843604e-05,
      "loss": 1.9519,
      "step": 170170
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.401271104812622,
      "learning_rate": 9.561364359405695e-05,
      "loss": 2.1989,
      "step": 170171
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.326719045639038,
      "learning_rate": 9.561064925767704e-05,
      "loss": 2.2304,
      "step": 170172
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.243630290031433,
      "learning_rate": 9.560765495929704e-05,
      "loss": 1.9528,
      "step": 170173
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.180835485458374,
      "learning_rate": 9.560466069891745e-05,
      "loss": 2.2307,
      "step": 170174
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.286733865737915,
      "learning_rate": 9.560166647653872e-05,
      "loss": 1.9546,
      "step": 170175
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.2253379821777344,
      "learning_rate": 9.559867229216161e-05,
      "loss": 1.8879,
      "step": 170176
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.290690541267395,
      "learning_rate": 9.559567814578655e-05,
      "loss": 2.0674,
      "step": 170177
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.358669638633728,
      "learning_rate": 9.559268403741401e-05,
      "loss": 2.0406,
      "step": 170178
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.3689466714859009,
      "learning_rate": 9.558968996704474e-05,
      "loss": 2.2477,
      "step": 170179
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.3643497228622437,
      "learning_rate": 9.558669593467911e-05,
      "loss": 2.2341,
      "step": 170180
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.2205674648284912,
      "learning_rate": 9.558370194031788e-05,
      "loss": 1.9747,
      "step": 170181
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.2534242868423462,
      "learning_rate": 9.558070798396151e-05,
      "loss": 2.1863,
      "step": 170182
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.2264959812164307,
      "learning_rate": 9.557771406561051e-05,
      "loss": 2.1753,
      "step": 170183
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.6926779747009277,
      "learning_rate": 9.55747201852654e-05,
      "loss": 2.2989,
      "step": 170184
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.2974168062210083,
      "learning_rate": 9.55717263429269e-05,
      "loss": 2.1151,
      "step": 170185
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.2576488256454468,
      "learning_rate": 9.556873253859537e-05,
      "loss": 2.0251,
      "step": 170186
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.3446459770202637,
      "learning_rate": 9.556573877227159e-05,
      "loss": 1.9995,
      "step": 170187
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.2770565748214722,
      "learning_rate": 9.556274504395596e-05,
      "loss": 2.0322,
      "step": 170188
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.2430826425552368,
      "learning_rate": 9.555975135364912e-05,
      "loss": 2.11,
      "step": 170189
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.3132373094558716,
      "learning_rate": 9.555675770135145e-05,
      "loss": 1.8859,
      "step": 170190
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.4347971677780151,
      "learning_rate": 9.555376408706376e-05,
      "loss": 2.0436,
      "step": 170191
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.1830133199691772,
      "learning_rate": 9.555077051078636e-05,
      "loss": 2.1118,
      "step": 170192
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.2908263206481934,
      "learning_rate": 9.554777697252007e-05,
      "loss": 2.1281,
      "step": 170193
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.4285483360290527,
      "learning_rate": 9.554478347226525e-05,
      "loss": 1.8871,
      "step": 170194
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.351157546043396,
      "learning_rate": 9.554179001002257e-05,
      "loss": 2.2041,
      "step": 170195
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.3138257265090942,
      "learning_rate": 9.553879658579241e-05,
      "loss": 2.0596,
      "step": 170196
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.1233991384506226,
      "learning_rate": 9.553580319957555e-05,
      "loss": 2.0007,
      "step": 170197
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.3088706731796265,
      "learning_rate": 9.553280985137234e-05,
      "loss": 1.9857,
      "step": 170198
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.1880089044570923,
      "learning_rate": 9.552981654118352e-05,
      "loss": 2.1541,
      "step": 170199
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.2810286283493042,
      "learning_rate": 9.55268232690096e-05,
      "loss": 1.9721,
      "step": 170200
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.280955195426941,
      "learning_rate": 9.552383003485107e-05,
      "loss": 1.9437,
      "step": 170201
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.3975353240966797,
      "learning_rate": 9.552083683870843e-05,
      "loss": 2.044,
      "step": 170202
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.2073042392730713,
      "learning_rate": 9.551784368058244e-05,
      "loss": 1.9871,
      "step": 170203
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.2090039253234863,
      "learning_rate": 9.551485056047345e-05,
      "loss": 2.1521,
      "step": 170204
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.3195335865020752,
      "learning_rate": 9.55118574783822e-05,
      "loss": 2.2823,
      "step": 170205
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.1626380681991577,
      "learning_rate": 9.550886443430915e-05,
      "loss": 1.7712,
      "step": 170206
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.2288658618927002,
      "learning_rate": 9.550587142825485e-05,
      "loss": 1.8776,
      "step": 170207
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.3299057483673096,
      "learning_rate": 9.550287846021978e-05,
      "loss": 2.0452,
      "step": 170208
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.281551718711853,
      "learning_rate": 9.54998855302047e-05,
      "loss": 1.9785,
      "step": 170209
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.3357008695602417,
      "learning_rate": 9.549689263820993e-05,
      "loss": 1.9517,
      "step": 170210
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.259819746017456,
      "learning_rate": 9.549389978423627e-05,
      "loss": 1.6108,
      "step": 170211
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.2608991861343384,
      "learning_rate": 9.549090696828414e-05,
      "loss": 1.8283,
      "step": 170212
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.2764999866485596,
      "learning_rate": 9.54879141903541e-05,
      "loss": 1.9335,
      "step": 170213
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.1718239784240723,
      "learning_rate": 9.548492145044663e-05,
      "loss": 2.1348,
      "step": 170214
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.3113996982574463,
      "learning_rate": 9.548192874856246e-05,
      "loss": 2.2104,
      "step": 170215
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.3881975412368774,
      "learning_rate": 9.5478936084702e-05,
      "loss": 1.882,
      "step": 170216
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.364365816116333,
      "learning_rate": 9.547594345886594e-05,
      "loss": 1.8515,
      "step": 170217
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.2637215852737427,
      "learning_rate": 9.547295087105465e-05,
      "loss": 2.0997,
      "step": 170218
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.2419613599777222,
      "learning_rate": 9.5469958321269e-05,
      "loss": 2.0769,
      "step": 170219
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.3224982023239136,
      "learning_rate": 9.546696580950917e-05,
      "loss": 2.1055,
      "step": 170220
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.2424322366714478,
      "learning_rate": 9.546397333577597e-05,
      "loss": 2.0536,
      "step": 170221
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.280276894569397,
      "learning_rate": 9.546098090006981e-05,
      "loss": 2.0094,
      "step": 170222
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.2544057369232178,
      "learning_rate": 9.54579885023914e-05,
      "loss": 2.1133,
      "step": 170223
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.1817678213119507,
      "learning_rate": 9.545499614274109e-05,
      "loss": 2.0524,
      "step": 170224
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.206925630569458,
      "learning_rate": 9.545200382111976e-05,
      "loss": 2.0579,
      "step": 170225
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.3279343843460083,
      "learning_rate": 9.544901153752757e-05,
      "loss": 1.8655,
      "step": 170226
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.5340521335601807,
      "learning_rate": 9.544601929196537e-05,
      "loss": 2.0093,
      "step": 170227
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.3741496801376343,
      "learning_rate": 9.544302708443352e-05,
      "loss": 2.1375,
      "step": 170228
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.3138415813446045,
      "learning_rate": 9.544003491493278e-05,
      "loss": 2.2425,
      "step": 170229
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.3060106039047241,
      "learning_rate": 9.543704278346349e-05,
      "loss": 2.1949,
      "step": 170230
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.2133959531784058,
      "learning_rate": 9.543405069002651e-05,
      "loss": 2.076,
      "step": 170231
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.2074414491653442,
      "learning_rate": 9.543105863462203e-05,
      "loss": 1.9069,
      "step": 170232
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.238945722579956,
      "learning_rate": 9.542806661725084e-05,
      "loss": 2.0853,
      "step": 170233
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.280924677848816,
      "learning_rate": 9.542507463791335e-05,
      "loss": 1.9992,
      "step": 170234
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.2889230251312256,
      "learning_rate": 9.542208269661031e-05,
      "loss": 1.7882,
      "step": 170235
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.256358027458191,
      "learning_rate": 9.541909079334204e-05,
      "loss": 2.0876,
      "step": 170236
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.280639886856079,
      "learning_rate": 9.541609892810945e-05,
      "loss": 2.2038,
      "step": 170237
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.2487038373947144,
      "learning_rate": 9.541310710091267e-05,
      "loss": 2.1269,
      "step": 170238
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.236081838607788,
      "learning_rate": 9.541011531175256e-05,
      "loss": 1.9592,
      "step": 170239
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.3593947887420654,
      "learning_rate": 9.540712356062945e-05,
      "loss": 1.8703,
      "step": 170240
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.216044545173645,
      "learning_rate": 9.540413184754412e-05,
      "loss": 1.8555,
      "step": 170241
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.2405856847763062,
      "learning_rate": 9.540114017249693e-05,
      "loss": 2.0998,
      "step": 170242
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.349910855293274,
      "learning_rate": 9.539814853548861e-05,
      "loss": 1.9392,
      "step": 170243
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.277213215827942,
      "learning_rate": 9.539515693651965e-05,
      "loss": 2.0638,
      "step": 170244
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.3531583547592163,
      "learning_rate": 9.53921653755906e-05,
      "loss": 1.9155,
      "step": 170245
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.159929871559143,
      "learning_rate": 9.53891738527019e-05,
      "loss": 1.9616,
      "step": 170246
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.3469178676605225,
      "learning_rate": 9.538618236785434e-05,
      "loss": 2.0633,
      "step": 170247
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.3152027130126953,
      "learning_rate": 9.538319092104824e-05,
      "loss": 2.0477,
      "step": 170248
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.3010536432266235,
      "learning_rate": 9.538019951228434e-05,
      "loss": 1.9429,
      "step": 170249
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.340232491493225,
      "learning_rate": 9.537720814156314e-05,
      "loss": 2.1336,
      "step": 170250
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.3614686727523804,
      "learning_rate": 9.537421680888519e-05,
      "loss": 2.0872,
      "step": 170251
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.2600207328796387,
      "learning_rate": 9.537122551425093e-05,
      "loss": 2.0742,
      "step": 170252
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.2243850231170654,
      "learning_rate": 9.53682342576611e-05,
      "loss": 1.9679,
      "step": 170253
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.7940819263458252,
      "learning_rate": 9.536524303911611e-05,
      "loss": 2.0269,
      "step": 170254
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.4041755199432373,
      "learning_rate": 9.536225185861666e-05,
      "loss": 1.995,
      "step": 170255
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.2668766975402832,
      "learning_rate": 9.535926071616324e-05,
      "loss": 2.1835,
      "step": 170256
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.2283411026000977,
      "learning_rate": 9.535626961175631e-05,
      "loss": 2.0264,
      "step": 170257
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.2666923999786377,
      "learning_rate": 9.535327854539658e-05,
      "loss": 1.9358,
      "step": 170258
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.3567041158676147,
      "learning_rate": 9.535028751708455e-05,
      "loss": 2.3082,
      "step": 170259
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.3650034666061401,
      "learning_rate": 9.53472965268207e-05,
      "loss": 2.1109,
      "step": 170260
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.3847655057907104,
      "learning_rate": 9.534430557460573e-05,
      "loss": 2.2342,
      "step": 170261
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.3399626016616821,
      "learning_rate": 9.53413146604401e-05,
      "loss": 2.0488,
      "step": 170262
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.2275773286819458,
      "learning_rate": 9.533832378432429e-05,
      "loss": 2.1613,
      "step": 170263
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.2754768133163452,
      "learning_rate": 9.533533294625904e-05,
      "loss": 2.0216,
      "step": 170264
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.1526989936828613,
      "learning_rate": 9.533234214624473e-05,
      "loss": 1.9732,
      "step": 170265
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.3823424577713013,
      "learning_rate": 9.532935138428211e-05,
      "loss": 2.1974,
      "step": 170266
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.3271336555480957,
      "learning_rate": 9.532636066037163e-05,
      "loss": 2.0326,
      "step": 170267
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.2410391569137573,
      "learning_rate": 9.53233699745138e-05,
      "loss": 2.0676,
      "step": 170268
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.3489333391189575,
      "learning_rate": 9.532037932670917e-05,
      "loss": 2.1855,
      "step": 170269
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.2250577211380005,
      "learning_rate": 9.531738871695842e-05,
      "loss": 2.0318,
      "step": 170270
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.350055456161499,
      "learning_rate": 9.531439814526193e-05,
      "loss": 2.0467,
      "step": 170271
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.2963619232177734,
      "learning_rate": 9.531140761162045e-05,
      "loss": 1.9064,
      "step": 170272
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.2760345935821533,
      "learning_rate": 9.530841711603445e-05,
      "loss": 1.9595,
      "step": 170273
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.2696846723556519,
      "learning_rate": 9.530542665850448e-05,
      "loss": 2.1967,
      "step": 170274
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.1659653186798096,
      "learning_rate": 9.5302436239031e-05,
      "loss": 1.7482,
      "step": 170275
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.2884700298309326,
      "learning_rate": 9.529944585761478e-05,
      "loss": 1.9763,
      "step": 170276
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.1505281925201416,
      "learning_rate": 9.52964555142561e-05,
      "loss": 1.9394,
      "step": 170277
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.254974126815796,
      "learning_rate": 9.529346520895583e-05,
      "loss": 2.0601,
      "step": 170278
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.3528987169265747,
      "learning_rate": 9.529047494171435e-05,
      "loss": 1.9908,
      "step": 170279
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.286232590675354,
      "learning_rate": 9.528748471253221e-05,
      "loss": 2.2063,
      "step": 170280
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.4101922512054443,
      "learning_rate": 9.528449452140994e-05,
      "loss": 2.0401,
      "step": 170281
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.5678945779800415,
      "learning_rate": 9.528150436834822e-05,
      "loss": 2.005,
      "step": 170282
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.2068207263946533,
      "learning_rate": 9.527851425334742e-05,
      "loss": 2.1169,
      "step": 170283
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.2729637622833252,
      "learning_rate": 9.527552417640833e-05,
      "loss": 2.0643,
      "step": 170284
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.1128642559051514,
      "learning_rate": 9.527253413753128e-05,
      "loss": 1.9843,
      "step": 170285
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.4161790609359741,
      "learning_rate": 9.526954413671714e-05,
      "loss": 2.0786,
      "step": 170286
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.395922064781189,
      "learning_rate": 9.526655417396601e-05,
      "loss": 1.9419,
      "step": 170287
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.2994173765182495,
      "learning_rate": 9.526356424927885e-05,
      "loss": 2.227,
      "step": 170288
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.3452082872390747,
      "learning_rate": 9.526057436265593e-05,
      "loss": 2.1012,
      "step": 170289
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.5276514291763306,
      "learning_rate": 9.525758451409802e-05,
      "loss": 1.8052,
      "step": 170290
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.1998988389968872,
      "learning_rate": 9.525459470360551e-05,
      "loss": 2.1688,
      "step": 170291
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.29972243309021,
      "learning_rate": 9.525160493117925e-05,
      "loss": 2.0937,
      "step": 170292
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.3875330686569214,
      "learning_rate": 9.524861519681937e-05,
      "loss": 2.1559,
      "step": 170293
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.3857150077819824,
      "learning_rate": 9.52456255005267e-05,
      "loss": 1.965,
      "step": 170294
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.234256625175476,
      "learning_rate": 9.524263584230168e-05,
      "loss": 2.0935,
      "step": 170295
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.2233260869979858,
      "learning_rate": 9.523964622214501e-05,
      "loss": 2.0044,
      "step": 170296
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.3286455869674683,
      "learning_rate": 9.523665664005706e-05,
      "loss": 1.8193,
      "step": 170297
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.3613780736923218,
      "learning_rate": 9.523366709603866e-05,
      "loss": 1.9654,
      "step": 170298
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.2731642723083496,
      "learning_rate": 9.523067759009e-05,
      "loss": 2.1462,
      "step": 170299
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.198883056640625,
      "learning_rate": 9.522768812221192e-05,
      "loss": 1.8423,
      "step": 170300
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.2237532138824463,
      "learning_rate": 9.522469869240475e-05,
      "loss": 2.0392,
      "step": 170301
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.2180428504943848,
      "learning_rate": 9.522170930066931e-05,
      "loss": 2.0701,
      "step": 170302
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.3452197313308716,
      "learning_rate": 9.521871994700592e-05,
      "loss": 2.0491,
      "step": 170303
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.1742795705795288,
      "learning_rate": 9.521573063141544e-05,
      "loss": 2.0459,
      "step": 170304
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.279209017753601,
      "learning_rate": 9.521274135389798e-05,
      "loss": 2.0512,
      "step": 170305
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.3835052251815796,
      "learning_rate": 9.520975211445447e-05,
      "loss": 2.0747,
      "step": 170306
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.2406845092773438,
      "learning_rate": 9.520676291308524e-05,
      "loss": 2.0,
      "step": 170307
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.319427728652954,
      "learning_rate": 9.520377374979102e-05,
      "loss": 2.0094,
      "step": 170308
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.228082299232483,
      "learning_rate": 9.520078462457218e-05,
      "loss": 2.0503,
      "step": 170309
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.1923874616622925,
      "learning_rate": 9.519779553742961e-05,
      "loss": 1.9873,
      "step": 170310
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.2040002346038818,
      "learning_rate": 9.519480648836342e-05,
      "loss": 2.0671,
      "step": 170311
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.316030740737915,
      "learning_rate": 9.519181747737447e-05,
      "loss": 2.055,
      "step": 170312
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.3093223571777344,
      "learning_rate": 9.518882850446313e-05,
      "loss": 2.2372,
      "step": 170313
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.2671066522598267,
      "learning_rate": 9.518583956963016e-05,
      "loss": 1.9819,
      "step": 170314
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.225355863571167,
      "learning_rate": 9.518285067287589e-05,
      "loss": 1.9981,
      "step": 170315
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.3153512477874756,
      "learning_rate": 9.517986181420123e-05,
      "loss": 2.0273,
      "step": 170316
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.2447679042816162,
      "learning_rate": 9.517687299360628e-05,
      "loss": 1.984,
      "step": 170317
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.271852731704712,
      "learning_rate": 9.51738842110919e-05,
      "loss": 2.085,
      "step": 170318
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.2359399795532227,
      "learning_rate": 9.517089546665849e-05,
      "loss": 2.2021,
      "step": 170319
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.371078372001648,
      "learning_rate": 9.516790676030675e-05,
      "loss": 2.1175,
      "step": 170320
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.2621328830718994,
      "learning_rate": 9.516491809203709e-05,
      "loss": 2.0777,
      "step": 170321
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.2982463836669922,
      "learning_rate": 9.516192946185033e-05,
      "loss": 2.0331,
      "step": 170322
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.2601627111434937,
      "learning_rate": 9.51589408697466e-05,
      "loss": 2.1598,
      "step": 170323
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.2292665243148804,
      "learning_rate": 9.515595231572685e-05,
      "loss": 1.8357,
      "step": 170324
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.2682636976242065,
      "learning_rate": 9.515296379979133e-05,
      "loss": 1.8893,
      "step": 170325
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.248757243156433,
      "learning_rate": 9.514997532194084e-05,
      "loss": 2.2173,
      "step": 170326
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.4070223569869995,
      "learning_rate": 9.514698688217575e-05,
      "loss": 1.9886,
      "step": 170327
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.239111065864563,
      "learning_rate": 9.51439984804968e-05,
      "loss": 1.8878,
      "step": 170328
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.2148290872573853,
      "learning_rate": 9.514101011690446e-05,
      "loss": 1.9999,
      "step": 170329
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.237714409828186,
      "learning_rate": 9.513802179139927e-05,
      "loss": 2.1679,
      "step": 170330
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.2255908250808716,
      "learning_rate": 9.51350335039817e-05,
      "loss": 2.1726,
      "step": 170331
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.1560559272766113,
      "learning_rate": 9.513204525465248e-05,
      "loss": 1.8604,
      "step": 170332
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.2531609535217285,
      "learning_rate": 9.512905704341196e-05,
      "loss": 1.7069,
      "step": 170333
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.641975998878479,
      "learning_rate": 9.512606887026095e-05,
      "loss": 2.1544,
      "step": 170334
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.266162633895874,
      "learning_rate": 9.512308073519985e-05,
      "loss": 2.1538,
      "step": 170335
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.268783450126648,
      "learning_rate": 9.512009263822922e-05,
      "loss": 1.9877,
      "step": 170336
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.1649249792099,
      "learning_rate": 9.511710457934957e-05,
      "loss": 1.9498,
      "step": 170337
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.3172749280929565,
      "learning_rate": 9.51141165585616e-05,
      "loss": 1.9609,
      "step": 170338
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.3073041439056396,
      "learning_rate": 9.511112857586567e-05,
      "loss": 1.9759,
      "step": 170339
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.2350589036941528,
      "learning_rate": 9.510814063126256e-05,
      "loss": 1.906,
      "step": 170340
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.2215070724487305,
      "learning_rate": 9.510515272475272e-05,
      "loss": 2.2202,
      "step": 170341
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.2743397951126099,
      "learning_rate": 9.51021648563366e-05,
      "loss": 2.0307,
      "step": 170342
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.4923830032348633,
      "learning_rate": 9.509917702601495e-05,
      "loss": 2.172,
      "step": 170343
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.2071926593780518,
      "learning_rate": 9.509618923378822e-05,
      "loss": 1.8808,
      "step": 170344
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.1815464496612549,
      "learning_rate": 9.509320147965688e-05,
      "loss": 2.0168,
      "step": 170345
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.348459005355835,
      "learning_rate": 9.509021376362171e-05,
      "loss": 2.1942,
      "step": 170346
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.2729825973510742,
      "learning_rate": 9.508722608568313e-05,
      "loss": 1.8734,
      "step": 170347
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.3902543783187866,
      "learning_rate": 9.508423844584158e-05,
      "loss": 2.1548,
      "step": 170348
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.1674411296844482,
      "learning_rate": 9.508125084409785e-05,
      "loss": 1.9543,
      "step": 170349
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.170812964439392,
      "learning_rate": 9.507826328045237e-05,
      "loss": 1.9288,
      "step": 170350
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.2421200275421143,
      "learning_rate": 9.507527575490562e-05,
      "loss": 2.055,
      "step": 170351
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.282814621925354,
      "learning_rate": 9.507228826745837e-05,
      "loss": 2.0915,
      "step": 170352
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.4097272157669067,
      "learning_rate": 9.506930081811103e-05,
      "loss": 2.1268,
      "step": 170353
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.2015458345413208,
      "learning_rate": 9.506631340686405e-05,
      "loss": 1.9118,
      "step": 170354
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.2397000789642334,
      "learning_rate": 9.506332603371822e-05,
      "loss": 2.037,
      "step": 170355
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.2047134637832642,
      "learning_rate": 9.506033869867393e-05,
      "loss": 2.06,
      "step": 170356
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.2188613414764404,
      "learning_rate": 9.505735140173184e-05,
      "loss": 1.8369,
      "step": 170357
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.284722924232483,
      "learning_rate": 9.505436414289248e-05,
      "loss": 2.3271,
      "step": 170358
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.1168246269226074,
      "learning_rate": 9.505137692215638e-05,
      "loss": 1.8737,
      "step": 170359
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.1731113195419312,
      "learning_rate": 9.5048389739524e-05,
      "loss": 1.8266,
      "step": 170360
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.3381726741790771,
      "learning_rate": 9.504540259499609e-05,
      "loss": 2.0093,
      "step": 170361
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.3209202289581299,
      "learning_rate": 9.5042415488573e-05,
      "loss": 1.9155,
      "step": 170362
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.1851601600646973,
      "learning_rate": 9.503942842025551e-05,
      "loss": 1.8528,
      "step": 170363
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.1983418464660645,
      "learning_rate": 9.503644139004394e-05,
      "loss": 2.0449,
      "step": 170364
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.3125046491622925,
      "learning_rate": 9.503345439793918e-05,
      "loss": 2.1133,
      "step": 170365
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.2712873220443726,
      "learning_rate": 9.503046744394137e-05,
      "loss": 2.0157,
      "step": 170366
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.2061960697174072,
      "learning_rate": 9.502748052805135e-05,
      "loss": 2.0523,
      "step": 170367
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.370855450630188,
      "learning_rate": 9.502449365026951e-05,
      "loss": 2.0317,
      "step": 170368
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.2987823486328125,
      "learning_rate": 9.502150681059658e-05,
      "loss": 2.0383,
      "step": 170369
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.2553812265396118,
      "learning_rate": 9.501852000903291e-05,
      "loss": 2.1661,
      "step": 170370
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.1845321655273438,
      "learning_rate": 9.501553324557938e-05,
      "loss": 1.912,
      "step": 170371
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.3217676877975464,
      "learning_rate": 9.501254652023613e-05,
      "loss": 2.0215,
      "step": 170372
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.2639034986495972,
      "learning_rate": 9.5009559833004e-05,
      "loss": 1.9897,
      "step": 170373
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.2854814529418945,
      "learning_rate": 9.500657318388338e-05,
      "loss": 1.9137,
      "step": 170374
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.276861548423767,
      "learning_rate": 9.500358657287501e-05,
      "loss": 1.9714,
      "step": 170375
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.354454517364502,
      "learning_rate": 9.500059999997926e-05,
      "loss": 2.1397,
      "step": 170376
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.2296167612075806,
      "learning_rate": 9.499761346519691e-05,
      "loss": 2.0291,
      "step": 170377
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.22537362575531,
      "learning_rate": 9.499462696852821e-05,
      "loss": 2.0896,
      "step": 170378
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.2539902925491333,
      "learning_rate": 9.499164050997398e-05,
      "loss": 1.9715,
      "step": 170379
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.272330403327942,
      "learning_rate": 9.498865408953458e-05,
      "loss": 2.0627,
      "step": 170380
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.2444500923156738,
      "learning_rate": 9.498566770721074e-05,
      "loss": 1.9169,
      "step": 170381
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.2647689580917358,
      "learning_rate": 9.498268136300285e-05,
      "loss": 2.0732,
      "step": 170382
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.2544677257537842,
      "learning_rate": 9.497969505691174e-05,
      "loss": 2.1162,
      "step": 170383
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.1857550144195557,
      "learning_rate": 9.497670878893757e-05,
      "loss": 2.1196,
      "step": 170384
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.2959245443344116,
      "learning_rate": 9.497372255908121e-05,
      "loss": 2.2333,
      "step": 170385
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.1458897590637207,
      "learning_rate": 9.497073636734303e-05,
      "loss": 2.049,
      "step": 170386
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.2144291400909424,
      "learning_rate": 9.496775021372372e-05,
      "loss": 2.2234,
      "step": 170387
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.4168190956115723,
      "learning_rate": 9.496476409822368e-05,
      "loss": 1.7256,
      "step": 170388
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.1995906829833984,
      "learning_rate": 9.496177802084378e-05,
      "loss": 2.1621,
      "step": 170389
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.4201709032058716,
      "learning_rate": 9.495879198158412e-05,
      "loss": 2.0496,
      "step": 170390
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.4146422147750854,
      "learning_rate": 9.495580598044565e-05,
      "loss": 2.1478,
      "step": 170391
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.280534267425537,
      "learning_rate": 9.495282001742863e-05,
      "loss": 2.0064,
      "step": 170392
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.2976367473602295,
      "learning_rate": 9.494983409253387e-05,
      "loss": 2.1389,
      "step": 170393
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.3590216636657715,
      "learning_rate": 9.494684820576166e-05,
      "loss": 2.0576,
      "step": 170394
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.2584539651870728,
      "learning_rate": 9.494386235711293e-05,
      "loss": 2.0337,
      "step": 170395
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.2812286615371704,
      "learning_rate": 9.49408765465878e-05,
      "loss": 2.1076,
      "step": 170396
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.2159173488616943,
      "learning_rate": 9.493789077418715e-05,
      "loss": 1.9675,
      "step": 170397
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.1780370473861694,
      "learning_rate": 9.49349050399113e-05,
      "loss": 2.0649,
      "step": 170398
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.3079057931900024,
      "learning_rate": 9.493191934376105e-05,
      "loss": 1.7929,
      "step": 170399
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.1532665491104126,
      "learning_rate": 9.492893368573671e-05,
      "loss": 2.2116,
      "step": 170400
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.3573123216629028,
      "learning_rate": 9.492594806583914e-05,
      "loss": 2.1374,
      "step": 170401
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.3641060590744019,
      "learning_rate": 9.492296248406852e-05,
      "loss": 2.0137,
      "step": 170402
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.2795264720916748,
      "learning_rate": 9.491997694042568e-05,
      "loss": 2.0401,
      "step": 170403
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.3212063312530518,
      "learning_rate": 9.4916991434911e-05,
      "loss": 2.0248,
      "step": 170404
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.3889801502227783,
      "learning_rate": 9.491400596752523e-05,
      "loss": 1.9594,
      "step": 170405
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.247304916381836,
      "learning_rate": 9.491102053826869e-05,
      "loss": 2.1055,
      "step": 170406
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.232197642326355,
      "learning_rate": 9.490803514714228e-05,
      "loss": 2.0645,
      "step": 170407
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.2377160787582397,
      "learning_rate": 9.490504979414614e-05,
      "loss": 1.8203,
      "step": 170408
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.203910231590271,
      "learning_rate": 9.490206447928112e-05,
      "loss": 1.7807,
      "step": 170409
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.345482587814331,
      "learning_rate": 9.489907920254758e-05,
      "loss": 1.9,
      "step": 170410
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.2558300495147705,
      "learning_rate": 9.489609396394626e-05,
      "loss": 1.9705,
      "step": 170411
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.1828176975250244,
      "learning_rate": 9.489310876347754e-05,
      "loss": 1.8426,
      "step": 170412
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.220539927482605,
      "learning_rate": 9.489012360114226e-05,
      "loss": 2.0898,
      "step": 170413
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.3540573120117188,
      "learning_rate": 9.48871384769406e-05,
      "loss": 1.887,
      "step": 170414
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.3948289155960083,
      "learning_rate": 9.488415339087336e-05,
      "loss": 2.0503,
      "step": 170415
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.2287707328796387,
      "learning_rate": 9.488116834294092e-05,
      "loss": 2.1123,
      "step": 170416
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.2051297426223755,
      "learning_rate": 9.487818333314408e-05,
      "loss": 1.9552,
      "step": 170417
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.326153039932251,
      "learning_rate": 9.487519836148314e-05,
      "loss": 2.1456,
      "step": 170418
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.2093143463134766,
      "learning_rate": 9.487221342795885e-05,
      "loss": 1.7187,
      "step": 170419
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.3472055196762085,
      "learning_rate": 9.486922853257173e-05,
      "loss": 2.1459,
      "step": 170420
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.4645742177963257,
      "learning_rate": 9.486624367532227e-05,
      "loss": 2.065,
      "step": 170421
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.2563273906707764,
      "learning_rate": 9.486325885621095e-05,
      "loss": 2.0182,
      "step": 170422
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.174920678138733,
      "learning_rate": 9.486027407523851e-05,
      "loss": 1.9301,
      "step": 170423
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.192947506904602,
      "learning_rate": 9.485728933240535e-05,
      "loss": 2.0242,
      "step": 170424
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.3073070049285889,
      "learning_rate": 9.485430462771216e-05,
      "loss": 2.0037,
      "step": 170425
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.2979010343551636,
      "learning_rate": 9.485131996115945e-05,
      "loss": 2.0841,
      "step": 170426
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.272249460220337,
      "learning_rate": 9.484833533274762e-05,
      "loss": 1.7143,
      "step": 170427
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.2232612371444702,
      "learning_rate": 9.48453507424775e-05,
      "loss": 1.9727,
      "step": 170428
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.2397818565368652,
      "learning_rate": 9.484236619034946e-05,
      "loss": 2.0134,
      "step": 170429
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.253196120262146,
      "learning_rate": 9.483938167636401e-05,
      "loss": 2.0627,
      "step": 170430
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.267091155052185,
      "learning_rate": 9.483639720052192e-05,
      "loss": 1.9773,
      "step": 170431
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.2549430131912231,
      "learning_rate": 9.483341276282357e-05,
      "loss": 1.8169,
      "step": 170432
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.2259624004364014,
      "learning_rate": 9.48304283632695e-05,
      "loss": 2.1985,
      "step": 170433
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.2295674085617065,
      "learning_rate": 9.482744400186043e-05,
      "loss": 2.1158,
      "step": 170434
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.406249761581421,
      "learning_rate": 9.482445967859678e-05,
      "loss": 1.9661,
      "step": 170435
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.2509766817092896,
      "learning_rate": 9.482147539347906e-05,
      "loss": 2.0887,
      "step": 170436
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.229045033454895,
      "learning_rate": 9.481849114650798e-05,
      "loss": 2.1492,
      "step": 170437
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.2722933292388916,
      "learning_rate": 9.481550693768404e-05,
      "loss": 2.1594,
      "step": 170438
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.2295935153961182,
      "learning_rate": 9.481252276700763e-05,
      "loss": 2.1096,
      "step": 170439
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.2564138174057007,
      "learning_rate": 9.480953863447958e-05,
      "loss": 1.9828,
      "step": 170440
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.318777084350586,
      "learning_rate": 9.480655454010023e-05,
      "loss": 2.1338,
      "step": 170441
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.1559169292449951,
      "learning_rate": 9.480357048387027e-05,
      "loss": 2.2024,
      "step": 170442
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.2235913276672363,
      "learning_rate": 9.480058646579021e-05,
      "loss": 2.0384,
      "step": 170443
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.2954436540603638,
      "learning_rate": 9.479760248586059e-05,
      "loss": 2.0761,
      "step": 170444
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.2950022220611572,
      "learning_rate": 9.47946185440819e-05,
      "loss": 2.0193,
      "step": 170445
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.3855143785476685,
      "learning_rate": 9.479163464045483e-05,
      "loss": 2.1711,
      "step": 170446
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.3106319904327393,
      "learning_rate": 9.478865077497976e-05,
      "loss": 2.1206,
      "step": 170447
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.1290746927261353,
      "learning_rate": 9.478566694765751e-05,
      "loss": 1.8603,
      "step": 170448
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.249518871307373,
      "learning_rate": 9.478268315848833e-05,
      "loss": 1.9678,
      "step": 170449
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.3523802757263184,
      "learning_rate": 9.477969940747315e-05,
      "loss": 2.0734,
      "step": 170450
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.4183776378631592,
      "learning_rate": 9.477671569461206e-05,
      "loss": 2.1455,
      "step": 170451
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.389565110206604,
      "learning_rate": 9.477373201990599e-05,
      "loss": 2.0506,
      "step": 170452
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.2928135395050049,
      "learning_rate": 9.477074838335523e-05,
      "loss": 1.9899,
      "step": 170453
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.2863184213638306,
      "learning_rate": 9.476776478496059e-05,
      "loss": 2.0518,
      "step": 170454
    },
    {
      "epoch": 2.22,
      "grad_norm": 2.0231356620788574,
      "learning_rate": 9.476478122472238e-05,
      "loss": 2.0945,
      "step": 170455
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.2805792093276978,
      "learning_rate": 9.476179770264146e-05,
      "loss": 2.1955,
      "step": 170456
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.39009690284729,
      "learning_rate": 9.475881421871802e-05,
      "loss": 2.0841,
      "step": 170457
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.2753149271011353,
      "learning_rate": 9.475583077295288e-05,
      "loss": 1.97,
      "step": 170458
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.41262686252594,
      "learning_rate": 9.475284736534642e-05,
      "loss": 1.8625,
      "step": 170459
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.484565019607544,
      "learning_rate": 9.474986399589937e-05,
      "loss": 1.8187,
      "step": 170460
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.574163794517517,
      "learning_rate": 9.47468806646121e-05,
      "loss": 1.9792,
      "step": 170461
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.3479905128479004,
      "learning_rate": 9.474389737148543e-05,
      "loss": 1.9385,
      "step": 170462
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.2792071104049683,
      "learning_rate": 9.474091411651956e-05,
      "loss": 1.8727,
      "step": 170463
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.363914966583252,
      "learning_rate": 9.473793089971534e-05,
      "loss": 2.0405,
      "step": 170464
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.7878623008728027,
      "learning_rate": 9.473494772107309e-05,
      "loss": 2.0951,
      "step": 170465
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.6958712339401245,
      "learning_rate": 9.473196458059362e-05,
      "loss": 2.0495,
      "step": 170466
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.2970038652420044,
      "learning_rate": 9.472898147827722e-05,
      "loss": 1.9738,
      "step": 170467
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.245658040046692,
      "learning_rate": 9.47259984141248e-05,
      "loss": 2.133,
      "step": 170468
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.2923535108566284,
      "learning_rate": 9.47230153881365e-05,
      "loss": 2.0613,
      "step": 170469
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.2414177656173706,
      "learning_rate": 9.472003240031315e-05,
      "loss": 2.1934,
      "step": 170470
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.9846264123916626,
      "learning_rate": 9.47170494506551e-05,
      "loss": 1.9629,
      "step": 170471
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.4388806819915771,
      "learning_rate": 9.471406653916316e-05,
      "loss": 2.0707,
      "step": 170472
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.2379860877990723,
      "learning_rate": 9.471108366583765e-05,
      "loss": 2.0926,
      "step": 170473
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.2061007022857666,
      "learning_rate": 9.47081008306794e-05,
      "loss": 2.0531,
      "step": 170474
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.4353123903274536,
      "learning_rate": 9.470511803368861e-05,
      "loss": 1.9047,
      "step": 170475
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.4283956289291382,
      "learning_rate": 9.470213527486611e-05,
      "loss": 2.1365,
      "step": 170476
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.5129741430282593,
      "learning_rate": 9.469915255421225e-05,
      "loss": 2.1169,
      "step": 170477
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.152236819267273,
      "learning_rate": 9.469616987172778e-05,
      "loss": 1.9029,
      "step": 170478
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.2952277660369873,
      "learning_rate": 9.469318722741308e-05,
      "loss": 1.9473,
      "step": 170479
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.2891415357589722,
      "learning_rate": 9.469020462126902e-05,
      "loss": 2.1296,
      "step": 170480
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.317989706993103,
      "learning_rate": 9.468722205329568e-05,
      "loss": 2.1355,
      "step": 170481
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.2407437562942505,
      "learning_rate": 9.468423952349399e-05,
      "loss": 1.8757,
      "step": 170482
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.3499997854232788,
      "learning_rate": 9.468125703186428e-05,
      "loss": 2.0073,
      "step": 170483
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.2812265157699585,
      "learning_rate": 9.467827457840728e-05,
      "loss": 2.0596,
      "step": 170484
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.3042619228363037,
      "learning_rate": 9.467529216312339e-05,
      "loss": 1.8469,
      "step": 170485
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.4123905897140503,
      "learning_rate": 9.46723097860133e-05,
      "loss": 2.0827,
      "step": 170486
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.3899147510528564,
      "learning_rate": 9.466932744707753e-05,
      "loss": 2.0134,
      "step": 170487
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.2194526195526123,
      "learning_rate": 9.466634514631663e-05,
      "loss": 2.1248,
      "step": 170488
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.3500382900238037,
      "learning_rate": 9.466336288373097e-05,
      "loss": 1.9681,
      "step": 170489
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.2897186279296875,
      "learning_rate": 9.466038065932142e-05,
      "loss": 2.1495,
      "step": 170490
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.7241804599761963,
      "learning_rate": 9.465739847308826e-05,
      "loss": 2.1205,
      "step": 170491
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.2475401163101196,
      "learning_rate": 9.465441632503226e-05,
      "loss": 2.0469,
      "step": 170492
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.2705202102661133,
      "learning_rate": 9.46514342151539e-05,
      "loss": 2.0041,
      "step": 170493
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.4622265100479126,
      "learning_rate": 9.464845214345372e-05,
      "loss": 1.9662,
      "step": 170494
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.6169370412826538,
      "learning_rate": 9.464547010993214e-05,
      "loss": 1.8404,
      "step": 170495
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.2326842546463013,
      "learning_rate": 9.464248811458996e-05,
      "loss": 2.105,
      "step": 170496
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.216895580291748,
      "learning_rate": 9.463950615742752e-05,
      "loss": 2.167,
      "step": 170497
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.4259289503097534,
      "learning_rate": 9.463652423844559e-05,
      "loss": 2.1908,
      "step": 170498
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.2755253314971924,
      "learning_rate": 9.463354235764458e-05,
      "loss": 2.0883,
      "step": 170499
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.389909029006958,
      "learning_rate": 9.463056051502505e-05,
      "loss": 1.9492,
      "step": 170500
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.3738199472427368,
      "learning_rate": 9.462757871058752e-05,
      "loss": 2.1363,
      "step": 170501
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.2250643968582153,
      "learning_rate": 9.462459694433269e-05,
      "loss": 2.012,
      "step": 170502
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.237144112586975,
      "learning_rate": 9.46216152162609e-05,
      "loss": 1.9243,
      "step": 170503
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.2792519330978394,
      "learning_rate": 9.461863352637294e-05,
      "loss": 2.0124,
      "step": 170504
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.1987799406051636,
      "learning_rate": 9.461565187466928e-05,
      "loss": 2.0002,
      "step": 170505
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.290895938873291,
      "learning_rate": 9.46126702611504e-05,
      "loss": 2.0226,
      "step": 170506
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.1774706840515137,
      "learning_rate": 9.460968868581683e-05,
      "loss": 1.9143,
      "step": 170507
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.1498260498046875,
      "learning_rate": 9.460670714866929e-05,
      "loss": 1.8841,
      "step": 170508
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.2616643905639648,
      "learning_rate": 9.460372564970814e-05,
      "loss": 2.2007,
      "step": 170509
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.3364698886871338,
      "learning_rate": 9.460074418893415e-05,
      "loss": 2.1261,
      "step": 170510
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.3067383766174316,
      "learning_rate": 9.459776276634773e-05,
      "loss": 1.9039,
      "step": 170511
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.4640482664108276,
      "learning_rate": 9.459478138194947e-05,
      "loss": 2.0605,
      "step": 170512
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.2670003175735474,
      "learning_rate": 9.459180003573983e-05,
      "loss": 2.0046,
      "step": 170513
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.3675868511199951,
      "learning_rate": 9.458881872771957e-05,
      "loss": 2.0587,
      "step": 170514
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.2349207401275635,
      "learning_rate": 9.4585837457889e-05,
      "loss": 1.814,
      "step": 170515
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.2269657850265503,
      "learning_rate": 9.458285622624889e-05,
      "loss": 1.9215,
      "step": 170516
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.2704628705978394,
      "learning_rate": 9.457987503279971e-05,
      "loss": 2.1536,
      "step": 170517
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.2853199243545532,
      "learning_rate": 9.457689387754193e-05,
      "loss": 1.9076,
      "step": 170518
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.376635193824768,
      "learning_rate": 9.457391276047626e-05,
      "loss": 2.0877,
      "step": 170519
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.3321806192398071,
      "learning_rate": 9.45709316816032e-05,
      "loss": 2.154,
      "step": 170520
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.2530237436294556,
      "learning_rate": 9.456795064092316e-05,
      "loss": 1.9534,
      "step": 170521
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.2191773653030396,
      "learning_rate": 9.456496963843694e-05,
      "loss": 1.8204,
      "step": 170522
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.291974663734436,
      "learning_rate": 9.456198867414496e-05,
      "loss": 2.1272,
      "step": 170523
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.3144361972808838,
      "learning_rate": 9.455900774804766e-05,
      "loss": 1.9932,
      "step": 170524
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.3008702993392944,
      "learning_rate": 9.455602686014586e-05,
      "loss": 2.0224,
      "step": 170525
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.216878056526184,
      "learning_rate": 9.455304601043984e-05,
      "loss": 1.9588,
      "step": 170526
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.2084441184997559,
      "learning_rate": 9.45500651989304e-05,
      "loss": 1.9698,
      "step": 170527
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.3183103799819946,
      "learning_rate": 9.454708442561799e-05,
      "loss": 1.927,
      "step": 170528
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.2282384634017944,
      "learning_rate": 9.454410369050312e-05,
      "loss": 2.1147,
      "step": 170529
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.3747955560684204,
      "learning_rate": 9.454112299358633e-05,
      "loss": 1.8003,
      "step": 170530
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.2113779783248901,
      "learning_rate": 9.453814233486828e-05,
      "loss": 2.047,
      "step": 170531
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.2133029699325562,
      "learning_rate": 9.453516171434938e-05,
      "loss": 2.011,
      "step": 170532
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.3329927921295166,
      "learning_rate": 9.453218113203039e-05,
      "loss": 2.0085,
      "step": 170533
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.5337177515029907,
      "learning_rate": 9.452920058791171e-05,
      "loss": 2.0805,
      "step": 170534
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.3284729719161987,
      "learning_rate": 9.452622008199398e-05,
      "loss": 2.221,
      "step": 170535
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.2103021144866943,
      "learning_rate": 9.452323961427755e-05,
      "loss": 2.1841,
      "step": 170536
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.2385133504867554,
      "learning_rate": 9.452025918476326e-05,
      "loss": 1.8531,
      "step": 170537
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.3906399011611938,
      "learning_rate": 9.451727879345145e-05,
      "loss": 2.0899,
      "step": 170538
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.3035743236541748,
      "learning_rate": 9.45142984403428e-05,
      "loss": 2.1035,
      "step": 170539
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.3423397541046143,
      "learning_rate": 9.451131812543778e-05,
      "loss": 2.1236,
      "step": 170540
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.269511342048645,
      "learning_rate": 9.450833784873714e-05,
      "loss": 2.0948,
      "step": 170541
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.188447117805481,
      "learning_rate": 9.450535761024113e-05,
      "loss": 2.0606,
      "step": 170542
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.2512987852096558,
      "learning_rate": 9.450237740995051e-05,
      "loss": 2.0396,
      "step": 170543
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.286810278892517,
      "learning_rate": 9.449939724786567e-05,
      "loss": 1.9262,
      "step": 170544
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.2111005783081055,
      "learning_rate": 9.44964171239874e-05,
      "loss": 2.2125,
      "step": 170545
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.2000423669815063,
      "learning_rate": 9.449343703831604e-05,
      "loss": 2.0428,
      "step": 170546
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.3098479509353638,
      "learning_rate": 9.449045699085237e-05,
      "loss": 2.0518,
      "step": 170547
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.3802601099014282,
      "learning_rate": 9.448747698159664e-05,
      "loss": 2.246,
      "step": 170548
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.2869040966033936,
      "learning_rate": 9.448449701054969e-05,
      "loss": 2.2142,
      "step": 170549
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.2839733362197876,
      "learning_rate": 9.44815170777118e-05,
      "loss": 1.9716,
      "step": 170550
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.2665408849716187,
      "learning_rate": 9.447853718308381e-05,
      "loss": 1.8384,
      "step": 170551
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.4687271118164062,
      "learning_rate": 9.447555732666602e-05,
      "loss": 2.2309,
      "step": 170552
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.3199119567871094,
      "learning_rate": 9.447257750845925e-05,
      "loss": 2.1554,
      "step": 170553
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.2875475883483887,
      "learning_rate": 9.446959772846386e-05,
      "loss": 1.9572,
      "step": 170554
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.3074109554290771,
      "learning_rate": 9.446661798668045e-05,
      "loss": 2.018,
      "step": 170555
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.2818338871002197,
      "learning_rate": 9.44636382831095e-05,
      "loss": 1.9637,
      "step": 170556
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.1592481136322021,
      "learning_rate": 9.446065861775171e-05,
      "loss": 1.8055,
      "step": 170557
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.2882229089736938,
      "learning_rate": 9.44576789906075e-05,
      "loss": 2.039,
      "step": 170558
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.3382166624069214,
      "learning_rate": 9.445469940167758e-05,
      "loss": 2.0176,
      "step": 170559
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.3634239435195923,
      "learning_rate": 9.445171985096239e-05,
      "loss": 2.0157,
      "step": 170560
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.2504448890686035,
      "learning_rate": 9.44487403384625e-05,
      "loss": 1.956,
      "step": 170561
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.3730958700180054,
      "learning_rate": 9.44457608641784e-05,
      "loss": 2.1164,
      "step": 170562
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.2899799346923828,
      "learning_rate": 9.444278142811082e-05,
      "loss": 2.2841,
      "step": 170563
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.1702052354812622,
      "learning_rate": 9.443980203026008e-05,
      "loss": 1.9214,
      "step": 170564
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.1531658172607422,
      "learning_rate": 9.443682267062695e-05,
      "loss": 1.9585,
      "step": 170565
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.09519362449646,
      "learning_rate": 9.443384334921194e-05,
      "loss": 1.8157,
      "step": 170566
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.1170369386672974,
      "learning_rate": 9.443086406601552e-05,
      "loss": 1.9554,
      "step": 170567
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.4124761819839478,
      "learning_rate": 9.44278848210382e-05,
      "loss": 1.9223,
      "step": 170568
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.2139724493026733,
      "learning_rate": 9.442490561428071e-05,
      "loss": 2.2815,
      "step": 170569
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.2754638195037842,
      "learning_rate": 9.442192644574341e-05,
      "loss": 1.9321,
      "step": 170570
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.2628490924835205,
      "learning_rate": 9.441894731542705e-05,
      "loss": 1.9779,
      "step": 170571
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.283467173576355,
      "learning_rate": 9.44159682233321e-05,
      "loss": 1.9418,
      "step": 170572
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.4077723026275635,
      "learning_rate": 9.441298916945907e-05,
      "loss": 2.1016,
      "step": 170573
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.2249667644500732,
      "learning_rate": 9.441001015380848e-05,
      "loss": 2.1382,
      "step": 170574
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.2580918073654175,
      "learning_rate": 9.440703117638105e-05,
      "loss": 2.0259,
      "step": 170575
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.3199474811553955,
      "learning_rate": 9.440405223717713e-05,
      "loss": 1.7598,
      "step": 170576
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.3611565828323364,
      "learning_rate": 9.440107333619747e-05,
      "loss": 2.026,
      "step": 170577
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.2151695489883423,
      "learning_rate": 9.439809447344254e-05,
      "loss": 2.0929,
      "step": 170578
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.2714290618896484,
      "learning_rate": 9.439511564891287e-05,
      "loss": 2.0253,
      "step": 170579
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.258562445640564,
      "learning_rate": 9.439213686260895e-05,
      "loss": 2.1507,
      "step": 170580
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.2169016599655151,
      "learning_rate": 9.438915811453148e-05,
      "loss": 1.8763,
      "step": 170581
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.204799771308899,
      "learning_rate": 9.438617940468088e-05,
      "loss": 1.9463,
      "step": 170582
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.6589467525482178,
      "learning_rate": 9.438320073305786e-05,
      "loss": 2.0249,
      "step": 170583
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.2021598815917969,
      "learning_rate": 9.438022209966288e-05,
      "loss": 2.0502,
      "step": 170584
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.2990467548370361,
      "learning_rate": 9.437724350449651e-05,
      "loss": 1.8624,
      "step": 170585
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.405834674835205,
      "learning_rate": 9.437426494755917e-05,
      "loss": 2.0063,
      "step": 170586
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.291374683380127,
      "learning_rate": 9.437128642885166e-05,
      "loss": 2.1016,
      "step": 170587
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.2715719938278198,
      "learning_rate": 9.436830794837431e-05,
      "loss": 2.2219,
      "step": 170588
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.2629295587539673,
      "learning_rate": 9.436532950612784e-05,
      "loss": 2.1505,
      "step": 170589
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.1672240495681763,
      "learning_rate": 9.436235110211276e-05,
      "loss": 2.069,
      "step": 170590
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.2231478691101074,
      "learning_rate": 9.435937273632962e-05,
      "loss": 2.1025,
      "step": 170591
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.2994542121887207,
      "learning_rate": 9.435639440877883e-05,
      "loss": 2.0028,
      "step": 170592
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.2203749418258667,
      "learning_rate": 9.435341611946116e-05,
      "loss": 1.9629,
      "step": 170593
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.3062200546264648,
      "learning_rate": 9.435043786837698e-05,
      "loss": 1.8961,
      "step": 170594
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.2747730016708374,
      "learning_rate": 9.434745965552704e-05,
      "loss": 1.827,
      "step": 170595
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.234697699546814,
      "learning_rate": 9.43444814809118e-05,
      "loss": 1.955,
      "step": 170596
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.3624556064605713,
      "learning_rate": 9.434150334453178e-05,
      "loss": 2.0595,
      "step": 170597
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.1418577432632446,
      "learning_rate": 9.433852524638746e-05,
      "loss": 2.0496,
      "step": 170598
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.353240728378296,
      "learning_rate": 9.433554718647962e-05,
      "loss": 1.8288,
      "step": 170599
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.2922919988632202,
      "learning_rate": 9.433256916480856e-05,
      "loss": 1.8232,
      "step": 170600
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.204841136932373,
      "learning_rate": 9.432959118137504e-05,
      "loss": 1.975,
      "step": 170601
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.2191609144210815,
      "learning_rate": 9.432661323617957e-05,
      "loss": 1.935,
      "step": 170602
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.2625336647033691,
      "learning_rate": 9.432363532922255e-05,
      "loss": 1.8687,
      "step": 170603
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.1450964212417603,
      "learning_rate": 9.432065746050473e-05,
      "loss": 2.0333,
      "step": 170604
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.3086588382720947,
      "learning_rate": 9.43176796300266e-05,
      "loss": 2.0913,
      "step": 170605
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.449058175086975,
      "learning_rate": 9.431470183778862e-05,
      "loss": 2.0385,
      "step": 170606
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.2979055643081665,
      "learning_rate": 9.431172408379146e-05,
      "loss": 1.9624,
      "step": 170607
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.3190767765045166,
      "learning_rate": 9.430874636803569e-05,
      "loss": 1.99,
      "step": 170608
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.318716049194336,
      "learning_rate": 9.430576869052166e-05,
      "loss": 2.1272,
      "step": 170609
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.3517900705337524,
      "learning_rate": 9.430279105125021e-05,
      "loss": 1.8387,
      "step": 170610
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.2644296884536743,
      "learning_rate": 9.429981345022164e-05,
      "loss": 2.1149,
      "step": 170611
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.2440470457077026,
      "learning_rate": 9.429683588743671e-05,
      "loss": 2.0988,
      "step": 170612
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.2475123405456543,
      "learning_rate": 9.429385836289588e-05,
      "loss": 2.0864,
      "step": 170613
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.2442399263381958,
      "learning_rate": 9.429088087659972e-05,
      "loss": 1.9889,
      "step": 170614
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.2396165132522583,
      "learning_rate": 9.428790342854865e-05,
      "loss": 2.0163,
      "step": 170615
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.391141653060913,
      "learning_rate": 9.428492601874345e-05,
      "loss": 2.0128,
      "step": 170616
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.3199431896209717,
      "learning_rate": 9.428194864718448e-05,
      "loss": 2.1029,
      "step": 170617
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.1325335502624512,
      "learning_rate": 9.427897131387245e-05,
      "loss": 1.7786,
      "step": 170618
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.1653430461883545,
      "learning_rate": 9.427599401880787e-05,
      "loss": 1.9666,
      "step": 170619
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.4344754219055176,
      "learning_rate": 9.427301676199115e-05,
      "loss": 1.9592,
      "step": 170620
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.360898733139038,
      "learning_rate": 9.427003954342306e-05,
      "loss": 1.9084,
      "step": 170621
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.2435258626937866,
      "learning_rate": 9.426706236310406e-05,
      "loss": 2.0921,
      "step": 170622
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.253523826599121,
      "learning_rate": 9.426408522103459e-05,
      "loss": 2.1387,
      "step": 170623
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.286210298538208,
      "learning_rate": 9.426110811721543e-05,
      "loss": 2.1145,
      "step": 170624
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.741296410560608,
      "learning_rate": 9.425813105164689e-05,
      "loss": 1.9777,
      "step": 170625
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.2571651935577393,
      "learning_rate": 9.425515402432973e-05,
      "loss": 2.0481,
      "step": 170626
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.340151071548462,
      "learning_rate": 9.425217703526446e-05,
      "loss": 2.0809,
      "step": 170627
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.254350185394287,
      "learning_rate": 9.424920008445157e-05,
      "loss": 2.0172,
      "step": 170628
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.272185206413269,
      "learning_rate": 9.424622317189155e-05,
      "loss": 1.687,
      "step": 170629
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.245542049407959,
      "learning_rate": 9.424324629758515e-05,
      "loss": 2.0763,
      "step": 170630
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.188369631767273,
      "learning_rate": 9.42402694615327e-05,
      "loss": 1.773,
      "step": 170631
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.3085085153579712,
      "learning_rate": 9.423729266373496e-05,
      "loss": 2.114,
      "step": 170632
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.257077932357788,
      "learning_rate": 9.42343159041924e-05,
      "loss": 1.9575,
      "step": 170633
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.2696555852890015,
      "learning_rate": 9.423133918290557e-05,
      "loss": 2.1524,
      "step": 170634
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.3548479080200195,
      "learning_rate": 9.42283624998749e-05,
      "loss": 1.9188,
      "step": 170635
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.2469247579574585,
      "learning_rate": 9.422538585510118e-05,
      "loss": 1.9457,
      "step": 170636
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.3471386432647705,
      "learning_rate": 9.422240924858473e-05,
      "loss": 2.1059,
      "step": 170637
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.2826449871063232,
      "learning_rate": 9.421943268032633e-05,
      "loss": 2.0804,
      "step": 170638
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.2381868362426758,
      "learning_rate": 9.421645615032642e-05,
      "loss": 2.1246,
      "step": 170639
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.46675705909729,
      "learning_rate": 9.421347965858558e-05,
      "loss": 2.1217,
      "step": 170640
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.3421906232833862,
      "learning_rate": 9.42105032051042e-05,
      "loss": 1.9904,
      "step": 170641
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.437476396560669,
      "learning_rate": 9.42075267898831e-05,
      "loss": 2.126,
      "step": 170642
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.3546619415283203,
      "learning_rate": 9.420455041292258e-05,
      "loss": 2.021,
      "step": 170643
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.2177882194519043,
      "learning_rate": 9.420157407422344e-05,
      "loss": 1.9393,
      "step": 170644
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.2671300172805786,
      "learning_rate": 9.41985977737861e-05,
      "loss": 1.8021,
      "step": 170645
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.3142579793930054,
      "learning_rate": 9.419562151161111e-05,
      "loss": 2.2138,
      "step": 170646
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.2124217748641968,
      "learning_rate": 9.419264528769896e-05,
      "loss": 1.8786,
      "step": 170647
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.2513065338134766,
      "learning_rate": 9.41896691020504e-05,
      "loss": 1.8793,
      "step": 170648
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.1927762031555176,
      "learning_rate": 9.418669295466573e-05,
      "loss": 1.8318,
      "step": 170649
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.2417351007461548,
      "learning_rate": 9.418371684554575e-05,
      "loss": 2.0732,
      "step": 170650
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.4753718376159668,
      "learning_rate": 9.418074077469093e-05,
      "loss": 2.1286,
      "step": 170651
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.3068097829818726,
      "learning_rate": 9.417776474210177e-05,
      "loss": 2.0908,
      "step": 170652
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.3825150728225708,
      "learning_rate": 9.417478874777873e-05,
      "loss": 2.0148,
      "step": 170653
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.3639222383499146,
      "learning_rate": 9.417181279172262e-05,
      "loss": 1.9312,
      "step": 170654
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.2840805053710938,
      "learning_rate": 9.416883687393371e-05,
      "loss": 1.9406,
      "step": 170655
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.2581658363342285,
      "learning_rate": 9.416586099441284e-05,
      "loss": 1.8461,
      "step": 170656
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.200427770614624,
      "learning_rate": 9.416288515316043e-05,
      "loss": 2.1163,
      "step": 170657
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.4785165786743164,
      "learning_rate": 9.4159909350177e-05,
      "loss": 2.0386,
      "step": 170658
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.4316415786743164,
      "learning_rate": 9.415693358546302e-05,
      "loss": 2.137,
      "step": 170659
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.348918080329895,
      "learning_rate": 9.415395785901927e-05,
      "loss": 1.9635,
      "step": 170660
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.2651362419128418,
      "learning_rate": 9.415098217084606e-05,
      "loss": 2.0714,
      "step": 170661
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.340475082397461,
      "learning_rate": 9.414800652094419e-05,
      "loss": 1.9346,
      "step": 170662
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.224721074104309,
      "learning_rate": 9.414503090931409e-05,
      "loss": 1.8799,
      "step": 170663
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.2470340728759766,
      "learning_rate": 9.414205533595631e-05,
      "loss": 1.8797,
      "step": 170664
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.2441152334213257,
      "learning_rate": 9.413907980087132e-05,
      "loss": 2.0767,
      "step": 170665
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.126006841659546,
      "learning_rate": 9.413610430405985e-05,
      "loss": 1.8892,
      "step": 170666
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.2913230657577515,
      "learning_rate": 9.413312884552225e-05,
      "loss": 2.1827,
      "step": 170667
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.1922106742858887,
      "learning_rate": 9.413015342525931e-05,
      "loss": 1.9201,
      "step": 170668
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.2718372344970703,
      "learning_rate": 9.412717804327145e-05,
      "loss": 2.0115,
      "step": 170669
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.285934329032898,
      "learning_rate": 9.412420269955925e-05,
      "loss": 2.005,
      "step": 170670
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.2499370574951172,
      "learning_rate": 9.412122739412312e-05,
      "loss": 1.9954,
      "step": 170671
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.3271325826644897,
      "learning_rate": 9.411825212696385e-05,
      "loss": 2.0701,
      "step": 170672
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.240238070487976,
      "learning_rate": 9.411527689808178e-05,
      "loss": 1.9995,
      "step": 170673
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.3472191095352173,
      "learning_rate": 9.41123017074777e-05,
      "loss": 2.1355,
      "step": 170674
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.3547090291976929,
      "learning_rate": 9.4109326555152e-05,
      "loss": 2.1079,
      "step": 170675
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.2526524066925049,
      "learning_rate": 9.410635144110524e-05,
      "loss": 1.8261,
      "step": 170676
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.3413200378417969,
      "learning_rate": 9.410337636533793e-05,
      "loss": 1.9352,
      "step": 170677
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.4400527477264404,
      "learning_rate": 9.41004013278508e-05,
      "loss": 2.1413,
      "step": 170678
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.2932196855545044,
      "learning_rate": 9.409742632864416e-05,
      "loss": 2.0104,
      "step": 170679
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.24410879611969,
      "learning_rate": 9.409445136771882e-05,
      "loss": 2.0023,
      "step": 170680
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.3305351734161377,
      "learning_rate": 9.409147644507518e-05,
      "loss": 1.9027,
      "step": 170681
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.3239318132400513,
      "learning_rate": 9.408850156071383e-05,
      "loss": 1.9127,
      "step": 170682
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.3297808170318604,
      "learning_rate": 9.408552671463522e-05,
      "loss": 2.0751,
      "step": 170683
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.3109238147735596,
      "learning_rate": 9.408255190684009e-05,
      "loss": 2.3684,
      "step": 170684
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.3149722814559937,
      "learning_rate": 9.407957713732885e-05,
      "loss": 2.0141,
      "step": 170685
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.436521053314209,
      "learning_rate": 9.407660240610213e-05,
      "loss": 2.0038,
      "step": 170686
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.301321029663086,
      "learning_rate": 9.407362771316039e-05,
      "loss": 2.128,
      "step": 170687
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.2839025259017944,
      "learning_rate": 9.407065305850437e-05,
      "loss": 1.9861,
      "step": 170688
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.2731387615203857,
      "learning_rate": 9.406767844213448e-05,
      "loss": 2.09,
      "step": 170689
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.2363988161087036,
      "learning_rate": 9.40647038640513e-05,
      "loss": 2.0873,
      "step": 170690
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.2723743915557861,
      "learning_rate": 9.406172932425529e-05,
      "loss": 2.0114,
      "step": 170691
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.1697943210601807,
      "learning_rate": 9.405875482274715e-05,
      "loss": 1.8529,
      "step": 170692
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.2650059461593628,
      "learning_rate": 9.405578035952732e-05,
      "loss": 1.8568,
      "step": 170693
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.3650628328323364,
      "learning_rate": 9.40528059345965e-05,
      "loss": 2.0936,
      "step": 170694
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.3094204664230347,
      "learning_rate": 9.404983154795515e-05,
      "loss": 2.1517,
      "step": 170695
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.5459750890731812,
      "learning_rate": 9.404685719960379e-05,
      "loss": 1.9027,
      "step": 170696
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.289904236793518,
      "learning_rate": 9.404388288954295e-05,
      "loss": 2.1185,
      "step": 170697
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.280115008354187,
      "learning_rate": 9.404090861777332e-05,
      "loss": 2.1587,
      "step": 170698
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.27627694606781,
      "learning_rate": 9.403793438429527e-05,
      "loss": 2.1972,
      "step": 170699
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.156083106994629,
      "learning_rate": 9.403496018910958e-05,
      "loss": 2.0687,
      "step": 170700
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.1636300086975098,
      "learning_rate": 9.403198603221665e-05,
      "loss": 1.9432,
      "step": 170701
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.2549201250076294,
      "learning_rate": 9.402901191361697e-05,
      "loss": 2.0009,
      "step": 170702
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.3088009357452393,
      "learning_rate": 9.402603783331128e-05,
      "loss": 1.9915,
      "step": 170703
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.296991229057312,
      "learning_rate": 9.402306379130006e-05,
      "loss": 2.0957,
      "step": 170704
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.3630001544952393,
      "learning_rate": 9.402008978758372e-05,
      "loss": 2.3736,
      "step": 170705
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.2352185249328613,
      "learning_rate": 9.401711582216304e-05,
      "loss": 1.86,
      "step": 170706
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.246398687362671,
      "learning_rate": 9.401414189503847e-05,
      "loss": 1.878,
      "step": 170707
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.780754804611206,
      "learning_rate": 9.401116800621044e-05,
      "loss": 2.09,
      "step": 170708
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.4251006841659546,
      "learning_rate": 9.400819415567975e-05,
      "loss": 1.9862,
      "step": 170709
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.2097456455230713,
      "learning_rate": 9.400522034344671e-05,
      "loss": 2.0156,
      "step": 170710
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.3454570770263672,
      "learning_rate": 9.400224656951208e-05,
      "loss": 1.9708,
      "step": 170711
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.364988088607788,
      "learning_rate": 9.399927283387635e-05,
      "loss": 2.0401,
      "step": 170712
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.2269337177276611,
      "learning_rate": 9.399629913654001e-05,
      "loss": 2.0537,
      "step": 170713
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.319003701210022,
      "learning_rate": 9.399332547750357e-05,
      "loss": 2.2355,
      "step": 170714
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.224422812461853,
      "learning_rate": 9.399035185676775e-05,
      "loss": 2.0734,
      "step": 170715
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.3936797380447388,
      "learning_rate": 9.39873782743329e-05,
      "loss": 2.0291,
      "step": 170716
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.2006231546401978,
      "learning_rate": 9.398440473019981e-05,
      "loss": 1.8863,
      "step": 170717
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.2620750665664673,
      "learning_rate": 9.398143122436888e-05,
      "loss": 2.1899,
      "step": 170718
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.1877328157424927,
      "learning_rate": 9.397845775684072e-05,
      "loss": 1.9101,
      "step": 170719
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.3579357862472534,
      "learning_rate": 9.397548432761574e-05,
      "loss": 2.1214,
      "step": 170720
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.2985832691192627,
      "learning_rate": 9.39725109366947e-05,
      "loss": 2.3093,
      "step": 170721
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.2763839960098267,
      "learning_rate": 9.396953758407798e-05,
      "loss": 1.9717,
      "step": 170722
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.217939019203186,
      "learning_rate": 9.396656426976629e-05,
      "loss": 1.9688,
      "step": 170723
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.358009696006775,
      "learning_rate": 9.396359099376012e-05,
      "loss": 2.0073,
      "step": 170724
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.2333869934082031,
      "learning_rate": 9.396061775606001e-05,
      "loss": 1.8543,
      "step": 170725
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.2525742053985596,
      "learning_rate": 9.395764455666639e-05,
      "loss": 2.1549,
      "step": 170726
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.2079575061798096,
      "learning_rate": 9.395467139558006e-05,
      "loss": 2.0784,
      "step": 170727
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.2252755165100098,
      "learning_rate": 9.395169827280132e-05,
      "loss": 2.0908,
      "step": 170728
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.3288785219192505,
      "learning_rate": 9.394872518833098e-05,
      "loss": 1.8542,
      "step": 170729
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.3559139966964722,
      "learning_rate": 9.394575214216941e-05,
      "loss": 2.1089,
      "step": 170730
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.1217460632324219,
      "learning_rate": 9.394277913431725e-05,
      "loss": 1.7749,
      "step": 170731
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.2293436527252197,
      "learning_rate": 9.393980616477493e-05,
      "loss": 2.1045,
      "step": 170732
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.3494809865951538,
      "learning_rate": 9.393683323354317e-05,
      "loss": 1.9433,
      "step": 170733
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.2040528059005737,
      "learning_rate": 9.393386034062234e-05,
      "loss": 1.8761,
      "step": 170734
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.1826919317245483,
      "learning_rate": 9.393088748601319e-05,
      "loss": 1.7462,
      "step": 170735
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.3622263669967651,
      "learning_rate": 9.392791466971619e-05,
      "loss": 2.1403,
      "step": 170736
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.323758602142334,
      "learning_rate": 9.392494189173188e-05,
      "loss": 2.1291,
      "step": 170737
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.2385287284851074,
      "learning_rate": 9.392196915206071e-05,
      "loss": 2.0259,
      "step": 170738
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.2806476354599,
      "learning_rate": 9.391899645070345e-05,
      "loss": 2.1698,
      "step": 170739
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.3052754402160645,
      "learning_rate": 9.391602378766043e-05,
      "loss": 2.1321,
      "step": 170740
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.4159152507781982,
      "learning_rate": 9.391305116293239e-05,
      "loss": 2.3168,
      "step": 170741
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.1951079368591309,
      "learning_rate": 9.391007857651983e-05,
      "loss": 2.2149,
      "step": 170742
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.1545782089233398,
      "learning_rate": 9.390710602842326e-05,
      "loss": 2.0709,
      "step": 170743
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.2556730508804321,
      "learning_rate": 9.390413351864318e-05,
      "loss": 1.9483,
      "step": 170744
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.3733612298965454,
      "learning_rate": 9.390116104718028e-05,
      "loss": 2.0895,
      "step": 170745
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.4080400466918945,
      "learning_rate": 9.389818861403495e-05,
      "loss": 1.9564,
      "step": 170746
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.3016853332519531,
      "learning_rate": 9.389521621920795e-05,
      "loss": 2.1713,
      "step": 170747
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.3826395273208618,
      "learning_rate": 9.389224386269969e-05,
      "loss": 1.9995,
      "step": 170748
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.2774146795272827,
      "learning_rate": 9.388927154451078e-05,
      "loss": 1.9989,
      "step": 170749
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.2816370725631714,
      "learning_rate": 9.388629926464165e-05,
      "loss": 2.3136,
      "step": 170750
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.2508366107940674,
      "learning_rate": 9.388332702309304e-05,
      "loss": 2.0909,
      "step": 170751
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.2126107215881348,
      "learning_rate": 9.38803548198653e-05,
      "loss": 2.0748,
      "step": 170752
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.1410834789276123,
      "learning_rate": 9.387738265495921e-05,
      "loss": 2.1265,
      "step": 170753
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.5344104766845703,
      "learning_rate": 9.38744105283751e-05,
      "loss": 2.1966,
      "step": 170754
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.2979464530944824,
      "learning_rate": 9.38714384401138e-05,
      "loss": 2.2096,
      "step": 170755
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.5099754333496094,
      "learning_rate": 9.38684663901755e-05,
      "loss": 2.1116,
      "step": 170756
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.2703473567962646,
      "learning_rate": 9.386549437856106e-05,
      "loss": 2.0085,
      "step": 170757
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.5084702968597412,
      "learning_rate": 9.386252240527083e-05,
      "loss": 2.0872,
      "step": 170758
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.3759740591049194,
      "learning_rate": 9.385955047030551e-05,
      "loss": 2.0295,
      "step": 170759
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.488345742225647,
      "learning_rate": 9.385657857366553e-05,
      "loss": 1.9487,
      "step": 170760
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.2246050834655762,
      "learning_rate": 9.385360671535169e-05,
      "loss": 2.1526,
      "step": 170761
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.2158854007720947,
      "learning_rate": 9.385063489536412e-05,
      "loss": 2.1749,
      "step": 170762
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.1632450819015503,
      "learning_rate": 9.384766311370377e-05,
      "loss": 1.9315,
      "step": 170763
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.2017731666564941,
      "learning_rate": 9.38446913703709e-05,
      "loss": 1.8728,
      "step": 170764
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.238671898841858,
      "learning_rate": 9.384171966536629e-05,
      "loss": 2.0631,
      "step": 170765
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.225703477859497,
      "learning_rate": 9.38387479986903e-05,
      "loss": 2.0272,
      "step": 170766
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.2065060138702393,
      "learning_rate": 9.383577637034376e-05,
      "loss": 2.0191,
      "step": 170767
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.3461180925369263,
      "learning_rate": 9.383280478032687e-05,
      "loss": 2.1119,
      "step": 170768
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.3042491674423218,
      "learning_rate": 9.382983322864042e-05,
      "loss": 2.2784,
      "step": 170769
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.2217698097229004,
      "learning_rate": 9.382686171528485e-05,
      "loss": 2.0171,
      "step": 170770
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.1518911123275757,
      "learning_rate": 9.382389024026081e-05,
      "loss": 2.0951,
      "step": 170771
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.292814016342163,
      "learning_rate": 9.382091880356872e-05,
      "loss": 2.0344,
      "step": 170772
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.247400164604187,
      "learning_rate": 9.381794740520931e-05,
      "loss": 1.8873,
      "step": 170773
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.2549716234207153,
      "learning_rate": 9.381497604518302e-05,
      "loss": 1.8346,
      "step": 170774
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.2930907011032104,
      "learning_rate": 9.381200472349043e-05,
      "loss": 2.1178,
      "step": 170775
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.2460365295410156,
      "learning_rate": 9.380903344013201e-05,
      "loss": 1.9526,
      "step": 170776
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.1524968147277832,
      "learning_rate": 9.380606219510845e-05,
      "loss": 1.7597,
      "step": 170777
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.3632824420928955,
      "learning_rate": 9.380309098842014e-05,
      "loss": 2.0857,
      "step": 170778
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.223013997077942,
      "learning_rate": 9.380011982006782e-05,
      "loss": 2.1951,
      "step": 170779
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.166279673576355,
      "learning_rate": 9.379714869005197e-05,
      "loss": 1.825,
      "step": 170780
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.2769275903701782,
      "learning_rate": 9.379417759837309e-05,
      "loss": 1.9223,
      "step": 170781
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.3047455549240112,
      "learning_rate": 9.37912065450317e-05,
      "loss": 2.1489,
      "step": 170782
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.1635932922363281,
      "learning_rate": 9.378823553002848e-05,
      "loss": 2.0458,
      "step": 170783
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.2195770740509033,
      "learning_rate": 9.378526455336387e-05,
      "loss": 2.0293,
      "step": 170784
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.204264760017395,
      "learning_rate": 9.378229361503855e-05,
      "loss": 2.1195,
      "step": 170785
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.380171537399292,
      "learning_rate": 9.377932271505296e-05,
      "loss": 2.019,
      "step": 170786
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.2814042568206787,
      "learning_rate": 9.377635185340763e-05,
      "loss": 1.9977,
      "step": 170787
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.3414347171783447,
      "learning_rate": 9.377338103010325e-05,
      "loss": 2.0412,
      "step": 170788
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.549426555633545,
      "learning_rate": 9.377041024514031e-05,
      "loss": 2.058,
      "step": 170789
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.2937599420547485,
      "learning_rate": 9.376743949851921e-05,
      "loss": 2.0375,
      "step": 170790
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.24337899684906,
      "learning_rate": 9.376446879024078e-05,
      "loss": 2.0027,
      "step": 170791
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.2203369140625,
      "learning_rate": 9.37614981203054e-05,
      "loss": 1.9365,
      "step": 170792
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.240718960762024,
      "learning_rate": 9.375852748871355e-05,
      "loss": 2.1313,
      "step": 170793
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.2448968887329102,
      "learning_rate": 9.375555689546597e-05,
      "loss": 2.087,
      "step": 170794
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.5226024389266968,
      "learning_rate": 9.375258634056314e-05,
      "loss": 2.1061,
      "step": 170795
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.3035703897476196,
      "learning_rate": 9.37496158240055e-05,
      "loss": 2.1057,
      "step": 170796
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.3048036098480225,
      "learning_rate": 9.37466453457938e-05,
      "loss": 1.8933,
      "step": 170797
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.3121446371078491,
      "learning_rate": 9.37436749059285e-05,
      "loss": 1.9489,
      "step": 170798
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.26835036277771,
      "learning_rate": 9.374070450441003e-05,
      "loss": 2.0415,
      "step": 170799
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.1894830465316772,
      "learning_rate": 9.373773414123917e-05,
      "loss": 2.019,
      "step": 170800
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.4623403549194336,
      "learning_rate": 9.373476381641627e-05,
      "loss": 2.2378,
      "step": 170801
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.2382464408874512,
      "learning_rate": 9.373179352994202e-05,
      "loss": 1.8945,
      "step": 170802
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.2632007598876953,
      "learning_rate": 9.372882328181697e-05,
      "loss": 2.1537,
      "step": 170803
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.1903183460235596,
      "learning_rate": 9.37258530720416e-05,
      "loss": 1.8454,
      "step": 170804
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.3291771411895752,
      "learning_rate": 9.37228829006164e-05,
      "loss": 2.1625,
      "step": 170805
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.574641227722168,
      "learning_rate": 9.37199127675421e-05,
      "loss": 2.0273,
      "step": 170806
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.3940753936767578,
      "learning_rate": 9.371694267281907e-05,
      "loss": 2.0189,
      "step": 170807
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.2890112400054932,
      "learning_rate": 9.371397261644806e-05,
      "loss": 2.2085,
      "step": 170808
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.2095507383346558,
      "learning_rate": 9.371100259842951e-05,
      "loss": 1.7247,
      "step": 170809
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.2628077268600464,
      "learning_rate": 9.370803261876398e-05,
      "loss": 2.1193,
      "step": 170810
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.27585768699646,
      "learning_rate": 9.370506267745193e-05,
      "loss": 2.0514,
      "step": 170811
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.1901631355285645,
      "learning_rate": 9.370209277449409e-05,
      "loss": 2.0223,
      "step": 170812
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.15492582321167,
      "learning_rate": 9.369912290989083e-05,
      "loss": 1.953,
      "step": 170813
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.3916019201278687,
      "learning_rate": 9.36961530836429e-05,
      "loss": 2.0335,
      "step": 170814
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.2335113286972046,
      "learning_rate": 9.369318329575073e-05,
      "loss": 2.2012,
      "step": 170815
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.2318918704986572,
      "learning_rate": 9.369021354621491e-05,
      "loss": 2.0869,
      "step": 170816
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.2588989734649658,
      "learning_rate": 9.368724383503587e-05,
      "loss": 2.1008,
      "step": 170817
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.2348588705062866,
      "learning_rate": 9.368427416221435e-05,
      "loss": 2.0523,
      "step": 170818
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.3168494701385498,
      "learning_rate": 9.368130452775075e-05,
      "loss": 2.0609,
      "step": 170819
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.437656283378601,
      "learning_rate": 9.367833493164577e-05,
      "loss": 2.1985,
      "step": 170820
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.3860775232315063,
      "learning_rate": 9.36753653738998e-05,
      "loss": 2.0553,
      "step": 170821
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.4541058540344238,
      "learning_rate": 9.367239585451364e-05,
      "loss": 2.1347,
      "step": 170822
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.2490016222000122,
      "learning_rate": 9.366942637348752e-05,
      "loss": 1.9019,
      "step": 170823
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.1676702499389648,
      "learning_rate": 9.366645693082222e-05,
      "loss": 2.0037,
      "step": 170824
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.4727463722229004,
      "learning_rate": 9.366348752651811e-05,
      "loss": 1.9886,
      "step": 170825
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.2359249591827393,
      "learning_rate": 9.366051816057598e-05,
      "loss": 1.8023,
      "step": 170826
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.4503161907196045,
      "learning_rate": 9.365754883299613e-05,
      "loss": 2.273,
      "step": 170827
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.286625862121582,
      "learning_rate": 9.365457954377946e-05,
      "loss": 2.1578,
      "step": 170828
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.227079153060913,
      "learning_rate": 9.36516102929261e-05,
      "loss": 1.8508,
      "step": 170829
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.3978489637374878,
      "learning_rate": 9.364864108043689e-05,
      "loss": 1.9593,
      "step": 170830
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.3302245140075684,
      "learning_rate": 9.36456719063122e-05,
      "loss": 2.1234,
      "step": 170831
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.393218994140625,
      "learning_rate": 9.364270277055274e-05,
      "loss": 2.058,
      "step": 170832
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.4856007099151611,
      "learning_rate": 9.363973367315892e-05,
      "loss": 1.9408,
      "step": 170833
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.2160435914993286,
      "learning_rate": 9.363676461413158e-05,
      "loss": 2.0639,
      "step": 170834
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.13507878780365,
      "learning_rate": 9.363379559347087e-05,
      "loss": 1.996,
      "step": 170835
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.4617282152175903,
      "learning_rate": 9.363082661117762e-05,
      "loss": 2.0406,
      "step": 170836
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.5526303052902222,
      "learning_rate": 9.362785766725218e-05,
      "loss": 2.218,
      "step": 170837
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.2243586778640747,
      "learning_rate": 9.36248887616953e-05,
      "loss": 2.0747,
      "step": 170838
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.2792079448699951,
      "learning_rate": 9.362191989450741e-05,
      "loss": 2.0982,
      "step": 170839
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.2249356508255005,
      "learning_rate": 9.361895106568927e-05,
      "loss": 1.9891,
      "step": 170840
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.585574984550476,
      "learning_rate": 9.361598227524105e-05,
      "loss": 2.0365,
      "step": 170841
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.2583345174789429,
      "learning_rate": 9.361301352316361e-05,
      "loss": 2.1384,
      "step": 170842
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.2822099924087524,
      "learning_rate": 9.361004480945731e-05,
      "loss": 1.936,
      "step": 170843
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.2958362102508545,
      "learning_rate": 9.360707613412293e-05,
      "loss": 2.0751,
      "step": 170844
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.2939344644546509,
      "learning_rate": 9.360410749716077e-05,
      "loss": 2.0599,
      "step": 170845
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.2020577192306519,
      "learning_rate": 9.36011388985717e-05,
      "loss": 2.0947,
      "step": 170846
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.3180996179580688,
      "learning_rate": 9.359817033835588e-05,
      "loss": 2.0928,
      "step": 170847
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.280335545539856,
      "learning_rate": 9.359520181651414e-05,
      "loss": 1.9594,
      "step": 170848
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.3435354232788086,
      "learning_rate": 9.359223333304688e-05,
      "loss": 1.9512,
      "step": 170849
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.4583683013916016,
      "learning_rate": 9.35892648879548e-05,
      "loss": 2.112,
      "step": 170850
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.1529204845428467,
      "learning_rate": 9.358629648123828e-05,
      "loss": 2.0035,
      "step": 170851
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.3069926500320435,
      "learning_rate": 9.358332811289815e-05,
      "loss": 1.9663,
      "step": 170852
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.243806004524231,
      "learning_rate": 9.358035978293459e-05,
      "loss": 2.0698,
      "step": 170853
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.2270145416259766,
      "learning_rate": 9.357739149134843e-05,
      "loss": 2.0997,
      "step": 170854
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.289167046546936,
      "learning_rate": 9.357442323814003e-05,
      "loss": 1.8817,
      "step": 170855
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.1535801887512207,
      "learning_rate": 9.357145502331014e-05,
      "loss": 2.0429,
      "step": 170856
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.2884761095046997,
      "learning_rate": 9.356848684685913e-05,
      "loss": 1.9529,
      "step": 170857
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.6150329113006592,
      "learning_rate": 9.35655187087878e-05,
      "loss": 1.9698,
      "step": 170858
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.360073447227478,
      "learning_rate": 9.356255060909634e-05,
      "loss": 2.031,
      "step": 170859
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.3479984998703003,
      "learning_rate": 9.355958254778562e-05,
      "loss": 1.9907,
      "step": 170860
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.2551307678222656,
      "learning_rate": 9.355661452485599e-05,
      "loss": 1.9696,
      "step": 170861
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.2046916484832764,
      "learning_rate": 9.355364654030815e-05,
      "loss": 2.0166,
      "step": 170862
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.3081694841384888,
      "learning_rate": 9.355067859414249e-05,
      "loss": 2.0291,
      "step": 170863
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.2790671586990356,
      "learning_rate": 9.354771068635975e-05,
      "loss": 1.9928,
      "step": 170864
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.2873566150665283,
      "learning_rate": 9.354474281696042e-05,
      "loss": 2.0781,
      "step": 170865
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.2014026641845703,
      "learning_rate": 9.354177498594497e-05,
      "loss": 2.1966,
      "step": 170866
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.3494409322738647,
      "learning_rate": 9.353880719331392e-05,
      "loss": 1.9495,
      "step": 170867
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.34373140335083,
      "learning_rate": 9.353583943906801e-05,
      "loss": 2.1333,
      "step": 170868
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.309180498123169,
      "learning_rate": 9.353287172320759e-05,
      "loss": 2.1463,
      "step": 170869
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.2456954717636108,
      "learning_rate": 9.352990404573342e-05,
      "loss": 2.2587,
      "step": 170870
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.2744791507720947,
      "learning_rate": 9.352693640664591e-05,
      "loss": 2.2392,
      "step": 170871
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.2353715896606445,
      "learning_rate": 9.352396880594553e-05,
      "loss": 1.8578,
      "step": 170872
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.2275896072387695,
      "learning_rate": 9.352100124363309e-05,
      "loss": 2.0355,
      "step": 170873
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.312217116355896,
      "learning_rate": 9.351803371970896e-05,
      "loss": 2.0476,
      "step": 170874
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.341160535812378,
      "learning_rate": 9.351506623417362e-05,
      "loss": 1.9134,
      "step": 170875
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.2111700773239136,
      "learning_rate": 9.351209878702783e-05,
      "loss": 2.3641,
      "step": 170876
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.5082457065582275,
      "learning_rate": 9.350913137827203e-05,
      "loss": 1.8627,
      "step": 170877
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.1966007947921753,
      "learning_rate": 9.350616400790671e-05,
      "loss": 1.9607,
      "step": 170878
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.2283204793930054,
      "learning_rate": 9.350319667593258e-05,
      "loss": 1.9043,
      "step": 170879
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.357287883758545,
      "learning_rate": 9.35002293823501e-05,
      "loss": 2.255,
      "step": 170880
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.2283083200454712,
      "learning_rate": 9.349726212715974e-05,
      "loss": 1.8678,
      "step": 170881
    },
    {
      "epoch": 2.22,
      "grad_norm": 1.1782827377319336,
      "learning_rate": 9.349429491036221e-05,
      "loss": 1.9727,
      "step": 170882
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.1986587047576904,
      "learning_rate": 9.3491327731958e-05,
      "loss": 1.9072,
      "step": 170883
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.2850987911224365,
      "learning_rate": 9.348836059194755e-05,
      "loss": 1.9547,
      "step": 170884
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.4945141077041626,
      "learning_rate": 9.34853934903316e-05,
      "loss": 1.9727,
      "step": 170885
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.2960354089736938,
      "learning_rate": 9.348242642711055e-05,
      "loss": 2.2476,
      "step": 170886
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.2012293338775635,
      "learning_rate": 9.34794594022851e-05,
      "loss": 2.0765,
      "step": 170887
    },
    {
      "epoch": 2.23,
      "grad_norm": 2.0231564044952393,
      "learning_rate": 9.347649241585569e-05,
      "loss": 1.8977,
      "step": 170888
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.2042862176895142,
      "learning_rate": 9.347352546782293e-05,
      "loss": 1.9508,
      "step": 170889
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.1914520263671875,
      "learning_rate": 9.347055855818722e-05,
      "loss": 1.9535,
      "step": 170890
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.4029184579849243,
      "learning_rate": 9.346759168694933e-05,
      "loss": 2.081,
      "step": 170891
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.2685645818710327,
      "learning_rate": 9.346462485410964e-05,
      "loss": 2.0906,
      "step": 170892
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.5759912729263306,
      "learning_rate": 9.346165805966885e-05,
      "loss": 1.9752,
      "step": 170893
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.2295984029769897,
      "learning_rate": 9.345869130362735e-05,
      "loss": 1.9031,
      "step": 170894
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.2646441459655762,
      "learning_rate": 9.345572458598597e-05,
      "loss": 1.9538,
      "step": 170895
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.3682957887649536,
      "learning_rate": 9.345275790674486e-05,
      "loss": 2.1416,
      "step": 170896
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.4175209999084473,
      "learning_rate": 9.344979126590492e-05,
      "loss": 1.9812,
      "step": 170897
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.3463571071624756,
      "learning_rate": 9.34468246634664e-05,
      "loss": 2.0544,
      "step": 170898
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.3252809047698975,
      "learning_rate": 9.34438580994302e-05,
      "loss": 1.8585,
      "step": 170899
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.3231866359710693,
      "learning_rate": 9.344089157379652e-05,
      "loss": 2.0027,
      "step": 170900
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.2622050046920776,
      "learning_rate": 9.343792508656629e-05,
      "loss": 2.0067,
      "step": 170901
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.255935549736023,
      "learning_rate": 9.343495863773964e-05,
      "loss": 1.9104,
      "step": 170902
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.331382155418396,
      "learning_rate": 9.343199222731745e-05,
      "loss": 2.0217,
      "step": 170903
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.3384528160095215,
      "learning_rate": 9.342902585530005e-05,
      "loss": 2.1013,
      "step": 170904
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.1648445129394531,
      "learning_rate": 9.342605952168818e-05,
      "loss": 1.819,
      "step": 170905
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.382319450378418,
      "learning_rate": 9.342309322648221e-05,
      "loss": 1.9779,
      "step": 170906
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.2281924486160278,
      "learning_rate": 9.3420126969683e-05,
      "loss": 2.045,
      "step": 170907
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.3514348268508911,
      "learning_rate": 9.341716075129069e-05,
      "loss": 2.0906,
      "step": 170908
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.1649048328399658,
      "learning_rate": 9.341419457130609e-05,
      "loss": 2.0324,
      "step": 170909
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.3574377298355103,
      "learning_rate": 9.341122842972963e-05,
      "loss": 1.9215,
      "step": 170910
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.2922213077545166,
      "learning_rate": 9.340826232656203e-05,
      "loss": 1.8712,
      "step": 170911
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.2493451833724976,
      "learning_rate": 9.340529626180359e-05,
      "loss": 1.873,
      "step": 170912
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.1571422815322876,
      "learning_rate": 9.340233023545523e-05,
      "loss": 2.2341,
      "step": 170913
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.159549355506897,
      "learning_rate": 9.339936424751707e-05,
      "loss": 2.053,
      "step": 170914
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.3589504957199097,
      "learning_rate": 9.339639829798999e-05,
      "loss": 1.9364,
      "step": 170915
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.4285979270935059,
      "learning_rate": 9.339343238687429e-05,
      "loss": 2.0362,
      "step": 170916
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.316009283065796,
      "learning_rate": 9.339046651417076e-05,
      "loss": 2.1241,
      "step": 170917
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.2938392162322998,
      "learning_rate": 9.338750067987976e-05,
      "loss": 1.9912,
      "step": 170918
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.9352003335952759,
      "learning_rate": 9.338453488400212e-05,
      "loss": 2.2237,
      "step": 170919
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.3704171180725098,
      "learning_rate": 9.338156912653796e-05,
      "loss": 1.8861,
      "step": 170920
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.277825117111206,
      "learning_rate": 9.33786034074882e-05,
      "loss": 2.0753,
      "step": 170921
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.8494330644607544,
      "learning_rate": 9.337563772685316e-05,
      "loss": 1.9939,
      "step": 170922
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.26808500289917,
      "learning_rate": 9.337267208463357e-05,
      "loss": 2.2332,
      "step": 170923
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.3099353313446045,
      "learning_rate": 9.336970648082983e-05,
      "loss": 2.0123,
      "step": 170924
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.295624852180481,
      "learning_rate": 9.336674091544274e-05,
      "loss": 1.9865,
      "step": 170925
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.290047526359558,
      "learning_rate": 9.336377538847244e-05,
      "loss": 1.8439,
      "step": 170926
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.2919418811798096,
      "learning_rate": 9.336080989991986e-05,
      "loss": 2.0463,
      "step": 170927
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.264069676399231,
      "learning_rate": 9.335784444978532e-05,
      "loss": 1.8785,
      "step": 170928
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.1501495838165283,
      "learning_rate": 9.335487903806954e-05,
      "loss": 1.9115,
      "step": 170929
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.2822186946868896,
      "learning_rate": 9.335191366477288e-05,
      "loss": 2.1008,
      "step": 170930
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.141530156135559,
      "learning_rate": 9.334894832989617e-05,
      "loss": 1.8834,
      "step": 170931
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.311116337776184,
      "learning_rate": 9.334598303343965e-05,
      "loss": 2.0073,
      "step": 170932
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.2113959789276123,
      "learning_rate": 9.334301777540408e-05,
      "loss": 2.0445,
      "step": 170933
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.3212260007858276,
      "learning_rate": 9.334005255578986e-05,
      "loss": 2.0754,
      "step": 170934
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.2876462936401367,
      "learning_rate": 9.333708737459771e-05,
      "loss": 2.1287,
      "step": 170935
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.2246519327163696,
      "learning_rate": 9.3334122231828e-05,
      "loss": 2.06,
      "step": 170936
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.2751531600952148,
      "learning_rate": 9.333115712748158e-05,
      "loss": 1.9849,
      "step": 170937
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.245527744293213,
      "learning_rate": 9.332819206155858e-05,
      "loss": 1.913,
      "step": 170938
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.3111968040466309,
      "learning_rate": 9.332522703405989e-05,
      "loss": 2.07,
      "step": 170939
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.2404781579971313,
      "learning_rate": 9.332226204498586e-05,
      "loss": 2.0272,
      "step": 170940
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.229203701019287,
      "learning_rate": 9.33192970943372e-05,
      "loss": 1.7811,
      "step": 170941
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.4543836116790771,
      "learning_rate": 9.33163321821143e-05,
      "loss": 1.8454,
      "step": 170942
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.258563756942749,
      "learning_rate": 9.331336730831795e-05,
      "loss": 1.9679,
      "step": 170943
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.2672958374023438,
      "learning_rate": 9.331040247294837e-05,
      "loss": 2.1611,
      "step": 170944
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.2644424438476562,
      "learning_rate": 9.330743767600642e-05,
      "loss": 2.2519,
      "step": 170945
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.284814715385437,
      "learning_rate": 9.33044729174924e-05,
      "loss": 2.3029,
      "step": 170946
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.2455257177352905,
      "learning_rate": 9.330150819740707e-05,
      "loss": 1.9414,
      "step": 170947
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.2318475246429443,
      "learning_rate": 9.32985435157508e-05,
      "loss": 2.0413,
      "step": 170948
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.350231409072876,
      "learning_rate": 9.329557887252432e-05,
      "loss": 2.1722,
      "step": 170949
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.3059382438659668,
      "learning_rate": 9.32926142677281e-05,
      "loss": 2.0253,
      "step": 170950
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.3944673538208008,
      "learning_rate": 9.328964970136272e-05,
      "loss": 2.1387,
      "step": 170951
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.4714893102645874,
      "learning_rate": 9.328668517342855e-05,
      "loss": 1.8564,
      "step": 170952
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.2050000429153442,
      "learning_rate": 9.32837206839264e-05,
      "loss": 2.2711,
      "step": 170953
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.2572565078735352,
      "learning_rate": 9.328075623285661e-05,
      "loss": 1.9544,
      "step": 170954
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.3842689990997314,
      "learning_rate": 9.327779182021991e-05,
      "loss": 2.1624,
      "step": 170955
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.2009496688842773,
      "learning_rate": 9.32748274460168e-05,
      "loss": 2.1774,
      "step": 170956
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.3013503551483154,
      "learning_rate": 9.327186311024779e-05,
      "loss": 1.7022,
      "step": 170957
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.3747483491897583,
      "learning_rate": 9.326889881291337e-05,
      "loss": 2.1548,
      "step": 170958
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.3516515493392944,
      "learning_rate": 9.326593455401422e-05,
      "loss": 1.9421,
      "step": 170959
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.2955021858215332,
      "learning_rate": 9.326297033355076e-05,
      "loss": 2.0772,
      "step": 170960
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.2214624881744385,
      "learning_rate": 9.32600061515237e-05,
      "loss": 1.9565,
      "step": 170961
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.6385293006896973,
      "learning_rate": 9.325704200793353e-05,
      "loss": 1.9605,
      "step": 170962
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.2908686399459839,
      "learning_rate": 9.325407790278066e-05,
      "loss": 2.0404,
      "step": 170963
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.1570959091186523,
      "learning_rate": 9.325111383606586e-05,
      "loss": 2.1131,
      "step": 170964
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.2215605974197388,
      "learning_rate": 9.324814980778961e-05,
      "loss": 2.0378,
      "step": 170965
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.335052490234375,
      "learning_rate": 9.324518581795231e-05,
      "loss": 1.9414,
      "step": 170966
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.445865511894226,
      "learning_rate": 9.324222186655472e-05,
      "loss": 1.9398,
      "step": 170967
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.3231712579727173,
      "learning_rate": 9.323925795359732e-05,
      "loss": 2.2584,
      "step": 170968
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.2273576259613037,
      "learning_rate": 9.323629407908055e-05,
      "loss": 1.9216,
      "step": 170969
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.1982812881469727,
      "learning_rate": 9.323333024300515e-05,
      "loss": 2.1674,
      "step": 170970
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.2651455402374268,
      "learning_rate": 9.323036644537146e-05,
      "loss": 1.9384,
      "step": 170971
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.1453804969787598,
      "learning_rate": 9.322740268618027e-05,
      "loss": 2.1291,
      "step": 170972
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.4087867736816406,
      "learning_rate": 9.322443896543198e-05,
      "loss": 2.0517,
      "step": 170973
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.230491280555725,
      "learning_rate": 9.322147528312721e-05,
      "loss": 1.7148,
      "step": 170974
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.1614254713058472,
      "learning_rate": 9.321851163926633e-05,
      "loss": 1.9261,
      "step": 170975
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.4888712167739868,
      "learning_rate": 9.321554803385015e-05,
      "loss": 2.3108,
      "step": 170976
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.3430705070495605,
      "learning_rate": 9.321258446687903e-05,
      "loss": 2.1188,
      "step": 170977
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.4626225233078003,
      "learning_rate": 9.320962093835364e-05,
      "loss": 2.1199,
      "step": 170978
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.420243501663208,
      "learning_rate": 9.320665744827455e-05,
      "loss": 2.1087,
      "step": 170979
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.3255847692489624,
      "learning_rate": 9.32036939966422e-05,
      "loss": 2.1392,
      "step": 170980
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.3459160327911377,
      "learning_rate": 9.320073058345708e-05,
      "loss": 2.2238,
      "step": 170981
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.3104937076568604,
      "learning_rate": 9.319776720871998e-05,
      "loss": 1.8812,
      "step": 170982
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.243045449256897,
      "learning_rate": 9.31948038724312e-05,
      "loss": 2.0294,
      "step": 170983
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.1331974267959595,
      "learning_rate": 9.319184057459152e-05,
      "loss": 2.0299,
      "step": 170984
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.332951545715332,
      "learning_rate": 9.318887731520126e-05,
      "loss": 2.1147,
      "step": 170985
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.4912269115447998,
      "learning_rate": 9.31859140942613e-05,
      "loss": 1.8785,
      "step": 170986
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.3240901231765747,
      "learning_rate": 9.318295091177179e-05,
      "loss": 2.0115,
      "step": 170987
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.221449851989746,
      "learning_rate": 9.317998776773358e-05,
      "loss": 1.9388,
      "step": 170988
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.381923794746399,
      "learning_rate": 9.317702466214703e-05,
      "loss": 2.0078,
      "step": 170989
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.2757112979888916,
      "learning_rate": 9.317406159501284e-05,
      "loss": 1.8829,
      "step": 170990
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.37442147731781,
      "learning_rate": 9.317109856633139e-05,
      "loss": 2.1583,
      "step": 170991
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.1036148071289062,
      "learning_rate": 9.316813557610355e-05,
      "loss": 1.9507,
      "step": 170992
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.2318310737609863,
      "learning_rate": 9.316517262432948e-05,
      "loss": 2.107,
      "step": 170993
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.226494550704956,
      "learning_rate": 9.316220971100999e-05,
      "loss": 1.9856,
      "step": 170994
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.2816226482391357,
      "learning_rate": 9.315924683614544e-05,
      "loss": 2.1234,
      "step": 170995
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.2074822187423706,
      "learning_rate": 9.31562839997366e-05,
      "loss": 1.8981,
      "step": 170996
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.2347385883331299,
      "learning_rate": 9.315332120178382e-05,
      "loss": 2.0331,
      "step": 170997
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.2885031700134277,
      "learning_rate": 9.315035844228794e-05,
      "loss": 2.0699,
      "step": 170998
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.260356068611145,
      "learning_rate": 9.314739572124911e-05,
      "loss": 2.1797,
      "step": 170999
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.201302409172058,
      "learning_rate": 9.314443303866817e-05,
      "loss": 1.8507,
      "step": 171000
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.2484118938446045,
      "learning_rate": 9.314147039454551e-05,
      "loss": 2.1533,
      "step": 171001
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.200427770614624,
      "learning_rate": 9.313850778888183e-05,
      "loss": 2.2259,
      "step": 171002
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.2676045894622803,
      "learning_rate": 9.313554522167753e-05,
      "loss": 1.9456,
      "step": 171003
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.2389318943023682,
      "learning_rate": 9.313258269293342e-05,
      "loss": 1.8874,
      "step": 171004
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.33085298538208,
      "learning_rate": 9.312962020264966e-05,
      "loss": 1.9599,
      "step": 171005
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.2765403985977173,
      "learning_rate": 9.312665775082714e-05,
      "loss": 2.1976,
      "step": 171006
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.4312204122543335,
      "learning_rate": 9.312369533746615e-05,
      "loss": 1.9895,
      "step": 171007
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.2731471061706543,
      "learning_rate": 9.312073296256747e-05,
      "loss": 2.1718,
      "step": 171008
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.425146460533142,
      "learning_rate": 9.311777062613148e-05,
      "loss": 2.0605,
      "step": 171009
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.181046724319458,
      "learning_rate": 9.311480832815897e-05,
      "loss": 1.976,
      "step": 171010
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.2808067798614502,
      "learning_rate": 9.311184606865014e-05,
      "loss": 2.074,
      "step": 171011
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.1730899810791016,
      "learning_rate": 9.310888384760583e-05,
      "loss": 1.9752,
      "step": 171012
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.2514207363128662,
      "learning_rate": 9.310592166502636e-05,
      "loss": 2.0111,
      "step": 171013
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.3254574537277222,
      "learning_rate": 9.31029595209125e-05,
      "loss": 2.042,
      "step": 171014
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.2630884647369385,
      "learning_rate": 9.309999741526464e-05,
      "loss": 2.127,
      "step": 171015
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.3096072673797607,
      "learning_rate": 9.309703534808357e-05,
      "loss": 1.9275,
      "step": 171016
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.169352412223816,
      "learning_rate": 9.309407331936947e-05,
      "loss": 1.8847,
      "step": 171017
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.2891839742660522,
      "learning_rate": 9.309111132912319e-05,
      "loss": 2.0373,
      "step": 171018
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.1981137990951538,
      "learning_rate": 9.308814937734508e-05,
      "loss": 2.0979,
      "step": 171019
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.4762072563171387,
      "learning_rate": 9.308518746403589e-05,
      "loss": 2.1389,
      "step": 171020
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.1694674491882324,
      "learning_rate": 9.308222558919597e-05,
      "loss": 2.0332,
      "step": 171021
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.3708819150924683,
      "learning_rate": 9.307926375282607e-05,
      "loss": 2.0782,
      "step": 171022
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.3731259107589722,
      "learning_rate": 9.307630195492664e-05,
      "loss": 1.8979,
      "step": 171023
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.5513498783111572,
      "learning_rate": 9.307334019549823e-05,
      "loss": 2.1189,
      "step": 171024
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.4165912866592407,
      "learning_rate": 9.307037847454131e-05,
      "loss": 2.1435,
      "step": 171025
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.3287030458450317,
      "learning_rate": 9.306741679205658e-05,
      "loss": 2.0493,
      "step": 171026
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.254292368888855,
      "learning_rate": 9.306445514804445e-05,
      "loss": 1.844,
      "step": 171027
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.634398102760315,
      "learning_rate": 9.306149354250563e-05,
      "loss": 2.072,
      "step": 171028
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.3857042789459229,
      "learning_rate": 9.305853197544059e-05,
      "loss": 2.0798,
      "step": 171029
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.782947063446045,
      "learning_rate": 9.305557044684985e-05,
      "loss": 1.8852,
      "step": 171030
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.2649611234664917,
      "learning_rate": 9.305260895673392e-05,
      "loss": 1.9469,
      "step": 171031
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.3051106929779053,
      "learning_rate": 9.304964750509351e-05,
      "loss": 2.2157,
      "step": 171032
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.7556051015853882,
      "learning_rate": 9.304668609192897e-05,
      "loss": 1.9272,
      "step": 171033
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.3241044282913208,
      "learning_rate": 9.304372471724108e-05,
      "loss": 2.0947,
      "step": 171034
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.2832727432250977,
      "learning_rate": 9.304076338103023e-05,
      "loss": 2.099,
      "step": 171035
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.2970932722091675,
      "learning_rate": 9.303780208329703e-05,
      "loss": 1.8544,
      "step": 171036
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.2541676759719849,
      "learning_rate": 9.303484082404192e-05,
      "loss": 1.9968,
      "step": 171037
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.311233401298523,
      "learning_rate": 9.30318796032656e-05,
      "loss": 2.1082,
      "step": 171038
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.4664359092712402,
      "learning_rate": 9.30289184209685e-05,
      "loss": 1.8341,
      "step": 171039
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.1605738401412964,
      "learning_rate": 9.302595727715132e-05,
      "loss": 1.9638,
      "step": 171040
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.2930084466934204,
      "learning_rate": 9.302299617181452e-05,
      "loss": 1.8347,
      "step": 171041
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.5366649627685547,
      "learning_rate": 9.302003510495864e-05,
      "loss": 2.0747,
      "step": 171042
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.2160335779190063,
      "learning_rate": 9.301707407658415e-05,
      "loss": 1.8891,
      "step": 171043
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.3795952796936035,
      "learning_rate": 9.301411308669181e-05,
      "loss": 1.9231,
      "step": 171044
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.2769086360931396,
      "learning_rate": 9.301115213528195e-05,
      "loss": 2.256,
      "step": 171045
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.3161934614181519,
      "learning_rate": 9.300819122235531e-05,
      "loss": 2.0313,
      "step": 171046
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.2275387048721313,
      "learning_rate": 9.300523034791236e-05,
      "loss": 1.9453,
      "step": 171047
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.2991936206817627,
      "learning_rate": 9.300226951195354e-05,
      "loss": 2.297,
      "step": 171048
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.3030709028244019,
      "learning_rate": 9.299930871447962e-05,
      "loss": 2.0404,
      "step": 171049
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.2507174015045166,
      "learning_rate": 9.2996347955491e-05,
      "loss": 1.9548,
      "step": 171050
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.2771596908569336,
      "learning_rate": 9.29933872349882e-05,
      "loss": 1.8976,
      "step": 171051
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.3092396259307861,
      "learning_rate": 9.299042655297191e-05,
      "loss": 1.9493,
      "step": 171052
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.2328214645385742,
      "learning_rate": 9.298746590944265e-05,
      "loss": 2.1528,
      "step": 171053
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.190104603767395,
      "learning_rate": 9.29845053044008e-05,
      "loss": 2.0508,
      "step": 171054
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.2714158296585083,
      "learning_rate": 9.298154473784715e-05,
      "loss": 1.9499,
      "step": 171055
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.2970020771026611,
      "learning_rate": 9.297858420978201e-05,
      "loss": 1.9175,
      "step": 171056
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.2367444038391113,
      "learning_rate": 9.297562372020619e-05,
      "loss": 2.1917,
      "step": 171057
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.312681794166565,
      "learning_rate": 9.297266326912009e-05,
      "loss": 1.9033,
      "step": 171058
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.3566250801086426,
      "learning_rate": 9.296970285652428e-05,
      "loss": 2.1751,
      "step": 171059
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.277586817741394,
      "learning_rate": 9.296674248241923e-05,
      "loss": 1.9666,
      "step": 171060
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.2681174278259277,
      "learning_rate": 9.296378214680565e-05,
      "loss": 2.1641,
      "step": 171061
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.1913059949874878,
      "learning_rate": 9.296082184968392e-05,
      "loss": 1.9502,
      "step": 171062
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.3130733966827393,
      "learning_rate": 9.295786159105479e-05,
      "loss": 2.1781,
      "step": 171063
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.2807207107543945,
      "learning_rate": 9.29549013709187e-05,
      "loss": 1.9478,
      "step": 171064
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.3092387914657593,
      "learning_rate": 9.29519411892762e-05,
      "loss": 1.7601,
      "step": 171065
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.4093549251556396,
      "learning_rate": 9.294898104612774e-05,
      "loss": 2.2019,
      "step": 171066
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.289232611656189,
      "learning_rate": 9.294602094147407e-05,
      "loss": 1.9032,
      "step": 171067
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.2402617931365967,
      "learning_rate": 9.294306087531552e-05,
      "loss": 1.9417,
      "step": 171068
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.3872140645980835,
      "learning_rate": 9.294010084765292e-05,
      "loss": 2.0956,
      "step": 171069
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.2707834243774414,
      "learning_rate": 9.29371408584865e-05,
      "loss": 1.9947,
      "step": 171070
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.2275582551956177,
      "learning_rate": 9.29341809078172e-05,
      "loss": 2.0701,
      "step": 171071
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.3956012725830078,
      "learning_rate": 9.293122099564517e-05,
      "loss": 2.0242,
      "step": 171072
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.214205026626587,
      "learning_rate": 9.292826112197122e-05,
      "loss": 1.9291,
      "step": 171073
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.2942909002304077,
      "learning_rate": 9.292530128679569e-05,
      "loss": 2.1422,
      "step": 171074
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.4550426006317139,
      "learning_rate": 9.292234149011937e-05,
      "loss": 1.9892,
      "step": 171075
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.258508563041687,
      "learning_rate": 9.291938173194263e-05,
      "loss": 2.1448,
      "step": 171076
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.2596021890640259,
      "learning_rate": 9.291642201226625e-05,
      "loss": 2.0896,
      "step": 171077
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.3250211477279663,
      "learning_rate": 9.291346233109041e-05,
      "loss": 1.8019,
      "step": 171078
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.394705057144165,
      "learning_rate": 9.291050268841598e-05,
      "loss": 2.1724,
      "step": 171079
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.390360713005066,
      "learning_rate": 9.29075430842433e-05,
      "loss": 2.1878,
      "step": 171080
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.4341869354248047,
      "learning_rate": 9.29045835185731e-05,
      "loss": 1.9562,
      "step": 171081
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.2846170663833618,
      "learning_rate": 9.290162399140577e-05,
      "loss": 2.1253,
      "step": 171082
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.4310349225997925,
      "learning_rate": 9.289866450274214e-05,
      "loss": 2.1424,
      "step": 171083
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.2299658060073853,
      "learning_rate": 9.289570505258234e-05,
      "loss": 2.0198,
      "step": 171084
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.1181126832962036,
      "learning_rate": 9.289274564092723e-05,
      "loss": 1.9516,
      "step": 171085
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.4559566974639893,
      "learning_rate": 9.28897862677772e-05,
      "loss": 1.9954,
      "step": 171086
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.2122597694396973,
      "learning_rate": 9.288682693313294e-05,
      "loss": 2.0112,
      "step": 171087
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.3961318731307983,
      "learning_rate": 9.288386763699483e-05,
      "loss": 2.072,
      "step": 171088
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.3860034942626953,
      "learning_rate": 9.288090837936362e-05,
      "loss": 1.8534,
      "step": 171089
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.3810501098632812,
      "learning_rate": 9.287794916023977e-05,
      "loss": 2.2494,
      "step": 171090
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.2603013515472412,
      "learning_rate": 9.287498997962382e-05,
      "loss": 2.0817,
      "step": 171091
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.2756837606430054,
      "learning_rate": 9.28720308375162e-05,
      "loss": 1.9676,
      "step": 171092
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.1760457754135132,
      "learning_rate": 9.286907173391768e-05,
      "loss": 1.8992,
      "step": 171093
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.089289903640747,
      "learning_rate": 9.286611266882867e-05,
      "loss": 1.9804,
      "step": 171094
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.2789490222930908,
      "learning_rate": 9.28631536422498e-05,
      "loss": 2.041,
      "step": 171095
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.3081517219543457,
      "learning_rate": 9.286019465418157e-05,
      "loss": 2.5325,
      "step": 171096
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.1950393915176392,
      "learning_rate": 9.285723570462456e-05,
      "loss": 1.8084,
      "step": 171097
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.26652193069458,
      "learning_rate": 9.28542767935792e-05,
      "loss": 1.9345,
      "step": 171098
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.2971552610397339,
      "learning_rate": 9.285131792104626e-05,
      "loss": 1.9035,
      "step": 171099
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.2387220859527588,
      "learning_rate": 9.284835908702605e-05,
      "loss": 1.9689,
      "step": 171100
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.3690385818481445,
      "learning_rate": 9.284540029151933e-05,
      "loss": 2.0374,
      "step": 171101
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.4999980926513672,
      "learning_rate": 9.284244153452658e-05,
      "loss": 2.2378,
      "step": 171102
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.3172523975372314,
      "learning_rate": 9.283948281604833e-05,
      "loss": 2.1801,
      "step": 171103
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.2047847509384155,
      "learning_rate": 9.2836524136085e-05,
      "loss": 2.1917,
      "step": 171104
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.2629761695861816,
      "learning_rate": 9.28335654946374e-05,
      "loss": 1.8125,
      "step": 171105
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.2413697242736816,
      "learning_rate": 9.283060689170585e-05,
      "loss": 2.1251,
      "step": 171106
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.3632463216781616,
      "learning_rate": 9.282764832729109e-05,
      "loss": 2.1238,
      "step": 171107
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.1597011089324951,
      "learning_rate": 9.28246898013936e-05,
      "loss": 1.8264,
      "step": 171108
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.2447484731674194,
      "learning_rate": 9.282173131401387e-05,
      "loss": 1.9678,
      "step": 171109
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.2085707187652588,
      "learning_rate": 9.281877286515243e-05,
      "loss": 1.9163,
      "step": 171110
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.3269370794296265,
      "learning_rate": 9.281581445480999e-05,
      "loss": 2.1255,
      "step": 171111
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.272228717803955,
      "learning_rate": 9.281285608298688e-05,
      "loss": 2.1738,
      "step": 171112
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.4731948375701904,
      "learning_rate": 9.280989774968388e-05,
      "loss": 2.0436,
      "step": 171113
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.527018427848816,
      "learning_rate": 9.280693945490144e-05,
      "loss": 2.0685,
      "step": 171114
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.346287488937378,
      "learning_rate": 9.280398119864009e-05,
      "loss": 1.9808,
      "step": 171115
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.5392519235610962,
      "learning_rate": 9.280102298090029e-05,
      "loss": 1.9118,
      "step": 171116
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.3709734678268433,
      "learning_rate": 9.27980648016828e-05,
      "loss": 1.928,
      "step": 171117
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.4596445560455322,
      "learning_rate": 9.279510666098797e-05,
      "loss": 2.2826,
      "step": 171118
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.2700830698013306,
      "learning_rate": 9.279214855881652e-05,
      "loss": 2.0715,
      "step": 171119
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.1267142295837402,
      "learning_rate": 9.278919049516892e-05,
      "loss": 2.1729,
      "step": 171120
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.354956865310669,
      "learning_rate": 9.278623247004571e-05,
      "loss": 1.8554,
      "step": 171121
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.2688801288604736,
      "learning_rate": 9.278327448344739e-05,
      "loss": 2.1057,
      "step": 171122
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.2252694368362427,
      "learning_rate": 9.278031653537464e-05,
      "loss": 1.9171,
      "step": 171123
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.434061884880066,
      "learning_rate": 9.277735862582788e-05,
      "loss": 2.2225,
      "step": 171124
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.285428524017334,
      "learning_rate": 9.277440075480776e-05,
      "loss": 1.936,
      "step": 171125
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.2961008548736572,
      "learning_rate": 9.277144292231483e-05,
      "loss": 1.8869,
      "step": 171126
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.2871054410934448,
      "learning_rate": 9.276848512834962e-05,
      "loss": 2.0472,
      "step": 171127
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.4281251430511475,
      "learning_rate": 9.276552737291251e-05,
      "loss": 1.929,
      "step": 171128
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.244107961654663,
      "learning_rate": 9.276256965600431e-05,
      "loss": 1.8063,
      "step": 171129
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.2569857835769653,
      "learning_rate": 9.27596119776254e-05,
      "loss": 2.2151,
      "step": 171130
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.3390499353408813,
      "learning_rate": 9.275665433777647e-05,
      "loss": 2.1947,
      "step": 171131
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.2565021514892578,
      "learning_rate": 9.275369673645798e-05,
      "loss": 2.0101,
      "step": 171132
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.394570231437683,
      "learning_rate": 9.275073917367037e-05,
      "loss": 2.1355,
      "step": 171133
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.2427575588226318,
      "learning_rate": 9.274778164941445e-05,
      "loss": 2.0593,
      "step": 171134
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.199122667312622,
      "learning_rate": 9.274482416369063e-05,
      "loss": 2.0968,
      "step": 171135
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.3252477645874023,
      "learning_rate": 9.274186671649931e-05,
      "loss": 1.9492,
      "step": 171136
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.200394868850708,
      "learning_rate": 9.273890930784135e-05,
      "loss": 1.7292,
      "step": 171137
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.2483361959457397,
      "learning_rate": 9.273595193771709e-05,
      "loss": 2.2576,
      "step": 171138
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.355699062347412,
      "learning_rate": 9.273299460612704e-05,
      "loss": 2.0247,
      "step": 171139
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.2299786806106567,
      "learning_rate": 9.273003731307197e-05,
      "loss": 2.0834,
      "step": 171140
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.2840838432312012,
      "learning_rate": 9.272708005855227e-05,
      "loss": 1.7925,
      "step": 171141
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.4007455110549927,
      "learning_rate": 9.272412284256842e-05,
      "loss": 2.0144,
      "step": 171142
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.1993638277053833,
      "learning_rate": 9.272116566512117e-05,
      "loss": 2.1184,
      "step": 171143
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.1829992532730103,
      "learning_rate": 9.271820852621099e-05,
      "loss": 2.0615,
      "step": 171144
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.3626627922058105,
      "learning_rate": 9.271525142583829e-05,
      "loss": 2.0447,
      "step": 171145
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.2441331148147583,
      "learning_rate": 9.271229436400382e-05,
      "loss": 2.0882,
      "step": 171146
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.2221448421478271,
      "learning_rate": 9.270933734070796e-05,
      "loss": 2.0393,
      "step": 171147
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.2663737535476685,
      "learning_rate": 9.270638035595148e-05,
      "loss": 1.9815,
      "step": 171148
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.2430481910705566,
      "learning_rate": 9.270342340973477e-05,
      "loss": 2.1555,
      "step": 171149
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.54197359085083,
      "learning_rate": 9.27004665020584e-05,
      "loss": 2.1792,
      "step": 171150
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.3024382591247559,
      "learning_rate": 9.269750963292281e-05,
      "loss": 1.8249,
      "step": 171151
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.3765807151794434,
      "learning_rate": 9.269455280232878e-05,
      "loss": 1.8327,
      "step": 171152
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.4918581247329712,
      "learning_rate": 9.269159601027668e-05,
      "loss": 2.068,
      "step": 171153
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.223265290260315,
      "learning_rate": 9.268863925676718e-05,
      "loss": 1.9209,
      "step": 171154
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.311079978942871,
      "learning_rate": 9.268568254180072e-05,
      "loss": 2.3222,
      "step": 171155
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.2456961870193481,
      "learning_rate": 9.268272586537795e-05,
      "loss": 2.1348,
      "step": 171156
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.2835935354232788,
      "learning_rate": 9.267976922749943e-05,
      "loss": 1.9792,
      "step": 171157
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.2078218460083008,
      "learning_rate": 9.267681262816561e-05,
      "loss": 2.0715,
      "step": 171158
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.3554517030715942,
      "learning_rate": 9.267385606737702e-05,
      "loss": 2.1304,
      "step": 171159
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.2118138074874878,
      "learning_rate": 9.267089954513436e-05,
      "loss": 2.0211,
      "step": 171160
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.178096055984497,
      "learning_rate": 9.266794306143797e-05,
      "loss": 2.0156,
      "step": 171161
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.2635010480880737,
      "learning_rate": 9.266498661628863e-05,
      "loss": 2.0075,
      "step": 171162
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.256505012512207,
      "learning_rate": 9.26620302096868e-05,
      "loss": 2.1179,
      "step": 171163
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.2959973812103271,
      "learning_rate": 9.265907384163301e-05,
      "loss": 1.9851,
      "step": 171164
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.2320972681045532,
      "learning_rate": 9.265611751212772e-05,
      "loss": 2.1029,
      "step": 171165
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.2587902545928955,
      "learning_rate": 9.265316122117165e-05,
      "loss": 2.1158,
      "step": 171166
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.3468679189682007,
      "learning_rate": 9.26502049687652e-05,
      "loss": 1.8428,
      "step": 171167
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.3198778629302979,
      "learning_rate": 9.264724875490907e-05,
      "loss": 2.0144,
      "step": 171168
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.467374324798584,
      "learning_rate": 9.264429257960372e-05,
      "loss": 2.0179,
      "step": 171169
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.3490500450134277,
      "learning_rate": 9.264133644284973e-05,
      "loss": 2.1415,
      "step": 171170
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.1808918714523315,
      "learning_rate": 9.263838034464754e-05,
      "loss": 1.7573,
      "step": 171171
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.2747074365615845,
      "learning_rate": 9.263542428499785e-05,
      "loss": 2.1901,
      "step": 171172
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.2334856986999512,
      "learning_rate": 9.263246826390109e-05,
      "loss": 1.7915,
      "step": 171173
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.2318731546401978,
      "learning_rate": 9.262951228135794e-05,
      "loss": 2.0532,
      "step": 171174
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.2289533615112305,
      "learning_rate": 9.262655633736889e-05,
      "loss": 1.9634,
      "step": 171175
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.3707834482192993,
      "learning_rate": 9.262360043193448e-05,
      "loss": 2.0972,
      "step": 171176
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.334188461303711,
      "learning_rate": 9.262064456505514e-05,
      "loss": 1.9211,
      "step": 171177
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.3008654117584229,
      "learning_rate": 9.261768873673168e-05,
      "loss": 2.0442,
      "step": 171178
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.1449836492538452,
      "learning_rate": 9.261473294696437e-05,
      "loss": 1.96,
      "step": 171179
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.2854204177856445,
      "learning_rate": 9.261177719575401e-05,
      "loss": 1.915,
      "step": 171180
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.3208884000778198,
      "learning_rate": 9.260882148310102e-05,
      "loss": 2.1331,
      "step": 171181
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.337463617324829,
      "learning_rate": 9.260586580900598e-05,
      "loss": 2.0768,
      "step": 171182
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.3504536151885986,
      "learning_rate": 9.260291017346929e-05,
      "loss": 2.161,
      "step": 171183
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.3118386268615723,
      "learning_rate": 9.259995457649176e-05,
      "loss": 2.1154,
      "step": 171184
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.4532088041305542,
      "learning_rate": 9.259699901807373e-05,
      "loss": 2.0915,
      "step": 171185
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.3245090246200562,
      "learning_rate": 9.259404349821591e-05,
      "loss": 2.0671,
      "step": 171186
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.296355962753296,
      "learning_rate": 9.259108801691878e-05,
      "loss": 2.0448,
      "step": 171187
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.301528811454773,
      "learning_rate": 9.25881325741829e-05,
      "loss": 2.2511,
      "step": 171188
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.4129583835601807,
      "learning_rate": 9.258517717000868e-05,
      "loss": 2.2566,
      "step": 171189
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.1874321699142456,
      "learning_rate": 9.258222180439689e-05,
      "loss": 2.2277,
      "step": 171190
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.2121683359146118,
      "learning_rate": 9.257926647734787e-05,
      "loss": 2.0801,
      "step": 171191
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.1424965858459473,
      "learning_rate": 9.25763111888624e-05,
      "loss": 1.8407,
      "step": 171192
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.3328275680541992,
      "learning_rate": 9.25733559389409e-05,
      "loss": 1.917,
      "step": 171193
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.312960147857666,
      "learning_rate": 9.257040072758393e-05,
      "loss": 1.9645,
      "step": 171194
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.2246767282485962,
      "learning_rate": 9.256744555479196e-05,
      "loss": 2.0788,
      "step": 171195
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.2646043300628662,
      "learning_rate": 9.256449042056568e-05,
      "loss": 1.834,
      "step": 171196
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.2957789897918701,
      "learning_rate": 9.25615353249055e-05,
      "loss": 2.0323,
      "step": 171197
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.3846557140350342,
      "learning_rate": 9.255858026781214e-05,
      "loss": 2.1892,
      "step": 171198
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.2093642950057983,
      "learning_rate": 9.255562524928607e-05,
      "loss": 1.82,
      "step": 171199
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.1925863027572632,
      "learning_rate": 9.255267026932781e-05,
      "loss": 1.7865,
      "step": 171200
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.3393547534942627,
      "learning_rate": 9.254971532793783e-05,
      "loss": 2.2139,
      "step": 171201
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.376458764076233,
      "learning_rate": 9.254676042511687e-05,
      "loss": 2.133,
      "step": 171202
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.417280912399292,
      "learning_rate": 9.254380556086526e-05,
      "loss": 2.0898,
      "step": 171203
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.622346043586731,
      "learning_rate": 9.254085073518379e-05,
      "loss": 2.0091,
      "step": 171204
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.3067431449890137,
      "learning_rate": 9.25378959480729e-05,
      "loss": 2.0907,
      "step": 171205
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.3110454082489014,
      "learning_rate": 9.253494119953312e-05,
      "loss": 2.2424,
      "step": 171206
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.2541308403015137,
      "learning_rate": 9.253198648956493e-05,
      "loss": 1.9597,
      "step": 171207
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.1959748268127441,
      "learning_rate": 9.252903181816905e-05,
      "loss": 1.8999,
      "step": 171208
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.2435507774353027,
      "learning_rate": 9.252607718534584e-05,
      "loss": 2.2267,
      "step": 171209
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.3795394897460938,
      "learning_rate": 9.252312259109606e-05,
      "loss": 2.1766,
      "step": 171210
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.3369921445846558,
      "learning_rate": 9.252016803542013e-05,
      "loss": 2.0144,
      "step": 171211
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.1254396438598633,
      "learning_rate": 9.251721351831864e-05,
      "loss": 1.86,
      "step": 171212
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.2076889276504517,
      "learning_rate": 9.251425903979197e-05,
      "loss": 1.6842,
      "step": 171213
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.2321968078613281,
      "learning_rate": 9.251130459984095e-05,
      "loss": 2.1019,
      "step": 171214
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.1948667764663696,
      "learning_rate": 9.25083501984659e-05,
      "loss": 1.9396,
      "step": 171215
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.3012712001800537,
      "learning_rate": 9.250539583566754e-05,
      "loss": 2.2869,
      "step": 171216
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.407430648803711,
      "learning_rate": 9.250244151144638e-05,
      "loss": 2.1695,
      "step": 171217
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.295770287513733,
      "learning_rate": 9.249948722580279e-05,
      "loss": 2.1364,
      "step": 171218
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.2942038774490356,
      "learning_rate": 9.249653297873758e-05,
      "loss": 2.13,
      "step": 171219
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.1468946933746338,
      "learning_rate": 9.24935787702512e-05,
      "loss": 1.6868,
      "step": 171220
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.262618064880371,
      "learning_rate": 9.249062460034408e-05,
      "loss": 2.0218,
      "step": 171221
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.3181666135787964,
      "learning_rate": 9.248767046901695e-05,
      "loss": 1.9793,
      "step": 171222
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.2930476665496826,
      "learning_rate": 9.248471637627016e-05,
      "loss": 2.0069,
      "step": 171223
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.1817653179168701,
      "learning_rate": 9.248176232210451e-05,
      "loss": 2.1306,
      "step": 171224
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.2775276899337769,
      "learning_rate": 9.247880830652041e-05,
      "loss": 2.0845,
      "step": 171225
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.2578117847442627,
      "learning_rate": 9.247585432951842e-05,
      "loss": 1.8865,
      "step": 171226
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.15538489818573,
      "learning_rate": 9.247290039109899e-05,
      "loss": 2.137,
      "step": 171227
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.2071224451065063,
      "learning_rate": 9.246994649126286e-05,
      "loss": 1.9555,
      "step": 171228
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.2417389154434204,
      "learning_rate": 9.246699263001039e-05,
      "loss": 2.2364,
      "step": 171229
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.9160155057907104,
      "learning_rate": 9.24640388073423e-05,
      "loss": 2.0306,
      "step": 171230
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.513578176498413,
      "learning_rate": 9.246108502325907e-05,
      "loss": 2.0769,
      "step": 171231
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.2572922706604004,
      "learning_rate": 9.245813127776118e-05,
      "loss": 1.9652,
      "step": 171232
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.4148112535476685,
      "learning_rate": 9.24551775708493e-05,
      "loss": 1.8775,
      "step": 171233
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.2961288690567017,
      "learning_rate": 9.245222390252394e-05,
      "loss": 1.8085,
      "step": 171234
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.558533787727356,
      "learning_rate": 9.244927027278551e-05,
      "loss": 2.0187,
      "step": 171235
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.2674241065979004,
      "learning_rate": 9.244631668163479e-05,
      "loss": 2.0465,
      "step": 171236
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.2812738418579102,
      "learning_rate": 9.244336312907223e-05,
      "loss": 2.0863,
      "step": 171237
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.4682780504226685,
      "learning_rate": 9.244040961509824e-05,
      "loss": 2.0673,
      "step": 171238
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.7188018560409546,
      "learning_rate": 9.243745613971362e-05,
      "loss": 2.1641,
      "step": 171239
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.237302541732788,
      "learning_rate": 9.243450270291879e-05,
      "loss": 2.0098,
      "step": 171240
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.3274874687194824,
      "learning_rate": 9.243154930471419e-05,
      "loss": 1.9557,
      "step": 171241
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.3676947355270386,
      "learning_rate": 9.242859594510059e-05,
      "loss": 2.0865,
      "step": 171242
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.5904921293258667,
      "learning_rate": 9.242564262407842e-05,
      "loss": 2.0977,
      "step": 171243
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.3493735790252686,
      "learning_rate": 9.242268934164818e-05,
      "loss": 1.9844,
      "step": 171244
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.1834930181503296,
      "learning_rate": 9.241973609781054e-05,
      "loss": 2.0687,
      "step": 171245
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.4406390190124512,
      "learning_rate": 9.241678289256592e-05,
      "loss": 2.0322,
      "step": 171246
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.2042458057403564,
      "learning_rate": 9.241382972591503e-05,
      "loss": 2.0348,
      "step": 171247
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.4220603704452515,
      "learning_rate": 9.241087659785831e-05,
      "loss": 1.8339,
      "step": 171248
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.5410752296447754,
      "learning_rate": 9.240792350839634e-05,
      "loss": 1.9906,
      "step": 171249
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.2799952030181885,
      "learning_rate": 9.240497045752954e-05,
      "loss": 2.2488,
      "step": 171250
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.438931941986084,
      "learning_rate": 9.240201744525868e-05,
      "loss": 2.0286,
      "step": 171251
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.267730951309204,
      "learning_rate": 9.239906447158414e-05,
      "loss": 2.1021,
      "step": 171252
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.1933653354644775,
      "learning_rate": 9.239611153650661e-05,
      "loss": 1.9205,
      "step": 171253
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.5084408521652222,
      "learning_rate": 9.239315864002654e-05,
      "loss": 2.1158,
      "step": 171254
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.3222442865371704,
      "learning_rate": 9.239020578214453e-05,
      "loss": 1.9022,
      "step": 171255
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.2134435176849365,
      "learning_rate": 9.238725296286098e-05,
      "loss": 2.05,
      "step": 171256
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.2423409223556519,
      "learning_rate": 9.238430018217666e-05,
      "loss": 2.1374,
      "step": 171257
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.32660710811615,
      "learning_rate": 9.238134744009195e-05,
      "loss": 2.1024,
      "step": 171258
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.2602516412734985,
      "learning_rate": 9.237839473660753e-05,
      "loss": 2.038,
      "step": 171259
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.237384557723999,
      "learning_rate": 9.237544207172387e-05,
      "loss": 2.0275,
      "step": 171260
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.2305196523666382,
      "learning_rate": 9.237248944544156e-05,
      "loss": 2.2176,
      "step": 171261
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.255836009979248,
      "learning_rate": 9.236953685776106e-05,
      "loss": 2.0637,
      "step": 171262
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.3844428062438965,
      "learning_rate": 9.236658430868302e-05,
      "loss": 2.0513,
      "step": 171263
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.3805779218673706,
      "learning_rate": 9.236363179820788e-05,
      "loss": 2.461,
      "step": 171264
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.1557118892669678,
      "learning_rate": 9.236067932633636e-05,
      "loss": 2.0107,
      "step": 171265
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.241841435432434,
      "learning_rate": 9.23577268930689e-05,
      "loss": 1.8627,
      "step": 171266
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.2491189241409302,
      "learning_rate": 9.235477449840609e-05,
      "loss": 1.9053,
      "step": 171267
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.249447226524353,
      "learning_rate": 9.235182214234828e-05,
      "loss": 2.1434,
      "step": 171268
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.4019545316696167,
      "learning_rate": 9.234886982489635e-05,
      "loss": 1.8099,
      "step": 171269
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.45294189453125,
      "learning_rate": 9.234591754605056e-05,
      "loss": 1.8831,
      "step": 171270
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.2340017557144165,
      "learning_rate": 9.234296530581168e-05,
      "loss": 2.1676,
      "step": 171271
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.2567750215530396,
      "learning_rate": 9.234001310418019e-05,
      "loss": 1.9117,
      "step": 171272
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.394519567489624,
      "learning_rate": 9.233706094115659e-05,
      "loss": 1.9779,
      "step": 171273
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.2609156370162964,
      "learning_rate": 9.233410881674133e-05,
      "loss": 2.0709,
      "step": 171274
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.2264673709869385,
      "learning_rate": 9.233115673093522e-05,
      "loss": 2.2002,
      "step": 171275
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.2374509572982788,
      "learning_rate": 9.232820468373853e-05,
      "loss": 1.9762,
      "step": 171276
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.262271523475647,
      "learning_rate": 9.232525267515209e-05,
      "loss": 2.0035,
      "step": 171277
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.2292442321777344,
      "learning_rate": 9.232230070517626e-05,
      "loss": 1.9636,
      "step": 171278
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.2172203063964844,
      "learning_rate": 9.231934877381164e-05,
      "loss": 2.133,
      "step": 171279
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.2205461263656616,
      "learning_rate": 9.231639688105871e-05,
      "loss": 2.0715,
      "step": 171280
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.1681640148162842,
      "learning_rate": 9.231344502691816e-05,
      "loss": 2.0519,
      "step": 171281
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.3536313772201538,
      "learning_rate": 9.231049321139035e-05,
      "loss": 2.1126,
      "step": 171282
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.348156452178955,
      "learning_rate": 9.230754143447605e-05,
      "loss": 2.1046,
      "step": 171283
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.254071593284607,
      "learning_rate": 9.23045896961757e-05,
      "loss": 2.2443,
      "step": 171284
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.3097782135009766,
      "learning_rate": 9.230163799648986e-05,
      "loss": 1.9561,
      "step": 171285
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.2243138551712036,
      "learning_rate": 9.229868633541893e-05,
      "loss": 1.9707,
      "step": 171286
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.356095790863037,
      "learning_rate": 9.229573471296372e-05,
      "loss": 2.1155,
      "step": 171287
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.2390207052230835,
      "learning_rate": 9.229278312912458e-05,
      "loss": 2.0501,
      "step": 171288
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.2533962726593018,
      "learning_rate": 9.228983158390219e-05,
      "loss": 2.18,
      "step": 171289
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.302720546722412,
      "learning_rate": 9.228688007729697e-05,
      "loss": 2.0992,
      "step": 171290
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.3583632707595825,
      "learning_rate": 9.228392860930972e-05,
      "loss": 2.1181,
      "step": 171291
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.3222438097000122,
      "learning_rate": 9.228097717994062e-05,
      "loss": 1.946,
      "step": 171292
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.3708208799362183,
      "learning_rate": 9.227802578919049e-05,
      "loss": 1.9996,
      "step": 171293
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.2937654256820679,
      "learning_rate": 9.22750744370597e-05,
      "loss": 2.0087,
      "step": 171294
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.3258661031723022,
      "learning_rate": 9.2272123123549e-05,
      "loss": 2.0201,
      "step": 171295
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.1696453094482422,
      "learning_rate": 9.226917184865875e-05,
      "loss": 2.188,
      "step": 171296
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.3733092546463013,
      "learning_rate": 9.226622061238975e-05,
      "loss": 2.0091,
      "step": 171297
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.2789921760559082,
      "learning_rate": 9.226326941474222e-05,
      "loss": 2.0807,
      "step": 171298
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.2850961685180664,
      "learning_rate": 9.226031825571693e-05,
      "loss": 1.9799,
      "step": 171299
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.355028510093689,
      "learning_rate": 9.225736713531429e-05,
      "loss": 2.059,
      "step": 171300
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.2959855794906616,
      "learning_rate": 9.225441605353504e-05,
      "loss": 2.245,
      "step": 171301
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.3182085752487183,
      "learning_rate": 9.22514650103795e-05,
      "loss": 2.0163,
      "step": 171302
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.4341726303100586,
      "learning_rate": 9.224851400584853e-05,
      "loss": 2.1286,
      "step": 171303
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.30539870262146,
      "learning_rate": 9.224556303994226e-05,
      "loss": 1.9561,
      "step": 171304
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.2412028312683105,
      "learning_rate": 9.22426121126616e-05,
      "loss": 2.161,
      "step": 171305
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.282182216644287,
      "learning_rate": 9.223966122400683e-05,
      "loss": 1.9305,
      "step": 171306
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.3356895446777344,
      "learning_rate": 9.223671037397877e-05,
      "loss": 2.0468,
      "step": 171307
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.2355949878692627,
      "learning_rate": 9.223375956257772e-05,
      "loss": 1.8168,
      "step": 171308
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.3409461975097656,
      "learning_rate": 9.223080878980439e-05,
      "loss": 2.0556,
      "step": 171309
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.2661384344100952,
      "learning_rate": 9.22278580556593e-05,
      "loss": 2.1572,
      "step": 171310
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.1845660209655762,
      "learning_rate": 9.222490736014297e-05,
      "loss": 1.9335,
      "step": 171311
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.197235107421875,
      "learning_rate": 9.222195670325588e-05,
      "loss": 2.123,
      "step": 171312
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.3382759094238281,
      "learning_rate": 9.221900608499872e-05,
      "loss": 1.9183,
      "step": 171313
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.2701689004898071,
      "learning_rate": 9.22160555053719e-05,
      "loss": 1.9669,
      "step": 171314
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.387734293937683,
      "learning_rate": 9.22131049643761e-05,
      "loss": 1.907,
      "step": 171315
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.3205151557922363,
      "learning_rate": 9.221015446201181e-05,
      "loss": 1.9235,
      "step": 171316
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.1796815395355225,
      "learning_rate": 9.220720399827951e-05,
      "loss": 1.9595,
      "step": 171317
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.227303385734558,
      "learning_rate": 9.220425357317988e-05,
      "loss": 2.0831,
      "step": 171318
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.282137393951416,
      "learning_rate": 9.22013031867134e-05,
      "loss": 1.8892,
      "step": 171319
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.2231366634368896,
      "learning_rate": 9.219835283888052e-05,
      "loss": 1.9825,
      "step": 171320
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.2525402307510376,
      "learning_rate": 9.219540252968201e-05,
      "loss": 1.8164,
      "step": 171321
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.1837884187698364,
      "learning_rate": 9.21924522591183e-05,
      "loss": 1.9488,
      "step": 171322
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.2194048166275024,
      "learning_rate": 9.218950202718985e-05,
      "loss": 2.1828,
      "step": 171323
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.2717323303222656,
      "learning_rate": 9.218655183389735e-05,
      "loss": 1.9725,
      "step": 171324
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.2863686084747314,
      "learning_rate": 9.218360167924131e-05,
      "loss": 1.8759,
      "step": 171325
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.2738220691680908,
      "learning_rate": 9.218065156322215e-05,
      "loss": 1.9128,
      "step": 171326
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.272310495376587,
      "learning_rate": 9.217770148584066e-05,
      "loss": 1.9119,
      "step": 171327
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.2606059312820435,
      "learning_rate": 9.217475144709724e-05,
      "loss": 1.9145,
      "step": 171328
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.505797266960144,
      "learning_rate": 9.217180144699234e-05,
      "loss": 1.9684,
      "step": 171329
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.5328923463821411,
      "learning_rate": 9.216885148552674e-05,
      "loss": 1.9291,
      "step": 171330
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.3925145864486694,
      "learning_rate": 9.216590156270079e-05,
      "loss": 2.1769,
      "step": 171331
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.4039338827133179,
      "learning_rate": 9.216295167851521e-05,
      "loss": 1.9968,
      "step": 171332
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.2434519529342651,
      "learning_rate": 9.216000183297045e-05,
      "loss": 1.7771,
      "step": 171333
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.3021458387374878,
      "learning_rate": 9.215705202606707e-05,
      "loss": 1.8654,
      "step": 171334
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.3490279912948608,
      "learning_rate": 9.215410225780552e-05,
      "loss": 2.1497,
      "step": 171335
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.235486388206482,
      "learning_rate": 9.215115252818655e-05,
      "loss": 1.9533,
      "step": 171336
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.3052765130996704,
      "learning_rate": 9.214820283721051e-05,
      "loss": 1.7868,
      "step": 171337
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.2993265390396118,
      "learning_rate": 9.214525318487813e-05,
      "loss": 2.1888,
      "step": 171338
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.3347489833831787,
      "learning_rate": 9.21423035711899e-05,
      "loss": 2.0594,
      "step": 171339
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.272215723991394,
      "learning_rate": 9.213935399614629e-05,
      "loss": 1.9341,
      "step": 171340
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.4250426292419434,
      "learning_rate": 9.213640445974784e-05,
      "loss": 1.8943,
      "step": 171341
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.3743053674697876,
      "learning_rate": 9.213345496199524e-05,
      "loss": 2.0797,
      "step": 171342
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.2063113451004028,
      "learning_rate": 9.213050550288886e-05,
      "loss": 1.9926,
      "step": 171343
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.1908663511276245,
      "learning_rate": 9.212755608242942e-05,
      "loss": 2.0209,
      "step": 171344
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.4321202039718628,
      "learning_rate": 9.212460670061741e-05,
      "loss": 1.9527,
      "step": 171345
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.2735201120376587,
      "learning_rate": 9.212165735745338e-05,
      "loss": 2.1871,
      "step": 171346
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.2954753637313843,
      "learning_rate": 9.211870805293773e-05,
      "loss": 2.0813,
      "step": 171347
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.2140212059020996,
      "learning_rate": 9.211575878707124e-05,
      "loss": 1.806,
      "step": 171348
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.3514474630355835,
      "learning_rate": 9.211280955985427e-05,
      "loss": 1.9034,
      "step": 171349
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.2255253791809082,
      "learning_rate": 9.210986037128754e-05,
      "loss": 1.9809,
      "step": 171350
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.1137957572937012,
      "learning_rate": 9.21069112213714e-05,
      "loss": 1.8309,
      "step": 171351
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.281823754310608,
      "learning_rate": 9.210396211010671e-05,
      "loss": 2.1127,
      "step": 171352
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.3042148351669312,
      "learning_rate": 9.210101303749363e-05,
      "loss": 2.0476,
      "step": 171353
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.2823207378387451,
      "learning_rate": 9.2098064003533e-05,
      "loss": 2.0909,
      "step": 171354
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.262742519378662,
      "learning_rate": 9.209511500822518e-05,
      "loss": 2.1793,
      "step": 171355
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.2171249389648438,
      "learning_rate": 9.209216605157088e-05,
      "loss": 2.0883,
      "step": 171356
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.2792668342590332,
      "learning_rate": 9.208921713357046e-05,
      "loss": 2.3179,
      "step": 171357
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.314716100692749,
      "learning_rate": 9.208626825422482e-05,
      "loss": 2.2767,
      "step": 171358
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.4550502300262451,
      "learning_rate": 9.208331941353401e-05,
      "loss": 2.0291,
      "step": 171359
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.404228925704956,
      "learning_rate": 9.208037061149897e-05,
      "loss": 1.9672,
      "step": 171360
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.3254092931747437,
      "learning_rate": 9.207742184812003e-05,
      "loss": 2.0228,
      "step": 171361
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.313980221748352,
      "learning_rate": 9.20744731233979e-05,
      "loss": 1.8815,
      "step": 171362
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.8006038665771484,
      "learning_rate": 9.207152443733295e-05,
      "loss": 2.0056,
      "step": 171363
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.3362644910812378,
      "learning_rate": 9.2068575789926e-05,
      "loss": 1.8186,
      "step": 171364
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.3306803703308105,
      "learning_rate": 9.206562718117727e-05,
      "loss": 2.1708,
      "step": 171365
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.359788179397583,
      "learning_rate": 9.206267861108754e-05,
      "loss": 1.8346,
      "step": 171366
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.2194277048110962,
      "learning_rate": 9.20597300796572e-05,
      "loss": 2.024,
      "step": 171367
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.2333391904830933,
      "learning_rate": 9.2056781586887e-05,
      "loss": 2.008,
      "step": 171368
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.306584358215332,
      "learning_rate": 9.205383313277724e-05,
      "loss": 1.8791,
      "step": 171369
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.2135486602783203,
      "learning_rate": 9.20508847173288e-05,
      "loss": 2.1993,
      "step": 171370
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.2168978452682495,
      "learning_rate": 9.204793634054183e-05,
      "loss": 1.9616,
      "step": 171371
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.405215859413147,
      "learning_rate": 9.204498800241719e-05,
      "loss": 1.9817,
      "step": 171372
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.2521113157272339,
      "learning_rate": 9.204203970295518e-05,
      "loss": 2.1366,
      "step": 171373
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.3622887134552002,
      "learning_rate": 9.203909144215659e-05,
      "loss": 1.9748,
      "step": 171374
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.2665272951126099,
      "learning_rate": 9.203614322002176e-05,
      "loss": 1.8649,
      "step": 171375
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.339547872543335,
      "learning_rate": 9.203319503655154e-05,
      "loss": 2.0456,
      "step": 171376
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.194045066833496,
      "learning_rate": 9.20302468917461e-05,
      "loss": 1.917,
      "step": 171377
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.1562819480895996,
      "learning_rate": 9.202729878560622e-05,
      "loss": 2.3531,
      "step": 171378
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.248671054840088,
      "learning_rate": 9.202435071813231e-05,
      "loss": 2.0891,
      "step": 171379
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.2747397422790527,
      "learning_rate": 9.202140268932511e-05,
      "loss": 2.1268,
      "step": 171380
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.286611795425415,
      "learning_rate": 9.201845469918495e-05,
      "loss": 2.0812,
      "step": 171381
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.2843910455703735,
      "learning_rate": 9.201550674771267e-05,
      "loss": 2.0204,
      "step": 171382
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.221009612083435,
      "learning_rate": 9.201255883490844e-05,
      "loss": 1.914,
      "step": 171383
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.236707091331482,
      "learning_rate": 9.200961096077312e-05,
      "loss": 1.9054,
      "step": 171384
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.289199709892273,
      "learning_rate": 9.200666312530703e-05,
      "loss": 1.8596,
      "step": 171385
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.1411668062210083,
      "learning_rate": 9.200371532851092e-05,
      "loss": 1.8671,
      "step": 171386
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.2014259099960327,
      "learning_rate": 9.200076757038514e-05,
      "loss": 1.8404,
      "step": 171387
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.281305193901062,
      "learning_rate": 9.199781985093055e-05,
      "loss": 1.9941,
      "step": 171388
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.3330950736999512,
      "learning_rate": 9.199487217014732e-05,
      "loss": 1.9288,
      "step": 171389
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.2674957513809204,
      "learning_rate": 9.199192452803622e-05,
      "loss": 1.8932,
      "step": 171390
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.252370834350586,
      "learning_rate": 9.19889769245977e-05,
      "loss": 2.1063,
      "step": 171391
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.1764981746673584,
      "learning_rate": 9.198602935983239e-05,
      "loss": 1.8594,
      "step": 171392
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.172787070274353,
      "learning_rate": 9.198308183374076e-05,
      "loss": 2.0555,
      "step": 171393
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.3199183940887451,
      "learning_rate": 9.198013434632349e-05,
      "loss": 1.9894,
      "step": 171394
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.228324055671692,
      "learning_rate": 9.1977186897581e-05,
      "loss": 2.0728,
      "step": 171395
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.2323137521743774,
      "learning_rate": 9.197423948751394e-05,
      "loss": 2.0884,
      "step": 171396
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.2698990106582642,
      "learning_rate": 9.197129211612265e-05,
      "loss": 1.973,
      "step": 171397
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.250770092010498,
      "learning_rate": 9.196834478340793e-05,
      "loss": 1.9774,
      "step": 171398
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.2179423570632935,
      "learning_rate": 9.196539748937015e-05,
      "loss": 1.9398,
      "step": 171399
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.346703290939331,
      "learning_rate": 9.196245023401001e-05,
      "loss": 2.1203,
      "step": 171400
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.192901372909546,
      "learning_rate": 9.195950301732798e-05,
      "loss": 1.8458,
      "step": 171401
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.24705171585083,
      "learning_rate": 9.19565558393246e-05,
      "loss": 1.9797,
      "step": 171402
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.6840214729309082,
      "learning_rate": 9.195360870000031e-05,
      "loss": 1.9129,
      "step": 171403
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.2629541158676147,
      "learning_rate": 9.195066159935589e-05,
      "loss": 1.762,
      "step": 171404
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.2298475503921509,
      "learning_rate": 9.194771453739166e-05,
      "loss": 1.6904,
      "step": 171405
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.323128342628479,
      "learning_rate": 9.194476751410838e-05,
      "loss": 2.1702,
      "step": 171406
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.2760200500488281,
      "learning_rate": 9.19418205295065e-05,
      "loss": 2.1191,
      "step": 171407
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.1814334392547607,
      "learning_rate": 9.193887358358648e-05,
      "loss": 1.8066,
      "step": 171408
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.3403760194778442,
      "learning_rate": 9.193592667634905e-05,
      "loss": 2.0754,
      "step": 171409
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.220762848854065,
      "learning_rate": 9.193297980779463e-05,
      "loss": 2.1217,
      "step": 171410
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.3392807245254517,
      "learning_rate": 9.193003297792372e-05,
      "loss": 1.8687,
      "step": 171411
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.2766519784927368,
      "learning_rate": 9.192708618673702e-05,
      "loss": 2.1027,
      "step": 171412
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.3387115001678467,
      "learning_rate": 9.192413943423501e-05,
      "loss": 1.878,
      "step": 171413
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.3490101099014282,
      "learning_rate": 9.192119272041812e-05,
      "loss": 2.0027,
      "step": 171414
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.2303627729415894,
      "learning_rate": 9.191824604528713e-05,
      "loss": 1.9735,
      "step": 171415
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.2807565927505493,
      "learning_rate": 9.191529940884238e-05,
      "loss": 1.9557,
      "step": 171416
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.3468910455703735,
      "learning_rate": 9.191235281108458e-05,
      "loss": 1.9475,
      "step": 171417
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.2203688621520996,
      "learning_rate": 9.190940625201418e-05,
      "loss": 2.1585,
      "step": 171418
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.3746943473815918,
      "learning_rate": 9.19064597316318e-05,
      "loss": 2.124,
      "step": 171419
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.3884831666946411,
      "learning_rate": 9.19035132499378e-05,
      "loss": 1.9807,
      "step": 171420
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.2834023237228394,
      "learning_rate": 9.190056680693298e-05,
      "loss": 2.0437,
      "step": 171421
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.3804000616073608,
      "learning_rate": 9.189762040261766e-05,
      "loss": 2.1342,
      "step": 171422
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.2081540822982788,
      "learning_rate": 9.189467403699261e-05,
      "loss": 2.1622,
      "step": 171423
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.2785470485687256,
      "learning_rate": 9.189172771005829e-05,
      "loss": 2.2734,
      "step": 171424
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.3300940990447998,
      "learning_rate": 9.18887814218152e-05,
      "loss": 2.0489,
      "step": 171425
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.2943511009216309,
      "learning_rate": 9.188583517226382e-05,
      "loss": 1.7569,
      "step": 171426
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.4060683250427246,
      "learning_rate": 9.188288896140487e-05,
      "loss": 2.1169,
      "step": 171427
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.2859842777252197,
      "learning_rate": 9.187994278923873e-05,
      "loss": 2.1654,
      "step": 171428
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.2897621393203735,
      "learning_rate": 9.187699665576615e-05,
      "loss": 1.9985,
      "step": 171429
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.477025032043457,
      "learning_rate": 9.18740505609875e-05,
      "loss": 2.3045,
      "step": 171430
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.242337942123413,
      "learning_rate": 9.187110450490354e-05,
      "loss": 2.0452,
      "step": 171431
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.2401131391525269,
      "learning_rate": 9.186815848751446e-05,
      "loss": 2.0241,
      "step": 171432
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.348181128501892,
      "learning_rate": 9.186521250882117e-05,
      "loss": 1.9895,
      "step": 171433
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.3230093717575073,
      "learning_rate": 9.186226656882392e-05,
      "loss": 2.1359,
      "step": 171434
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.19496750831604,
      "learning_rate": 9.185932066752353e-05,
      "loss": 1.9366,
      "step": 171435
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.419616460800171,
      "learning_rate": 9.18563748049203e-05,
      "loss": 2.1854,
      "step": 171436
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.213092565536499,
      "learning_rate": 9.185342898101513e-05,
      "loss": 2.201,
      "step": 171437
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.3119028806686401,
      "learning_rate": 9.18504831958081e-05,
      "loss": 1.839,
      "step": 171438
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.3996199369430542,
      "learning_rate": 9.184753744930011e-05,
      "loss": 2.0469,
      "step": 171439
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.4289437532424927,
      "learning_rate": 9.18445917414915e-05,
      "loss": 1.9326,
      "step": 171440
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.4535502195358276,
      "learning_rate": 9.184164607238303e-05,
      "loss": 2.2729,
      "step": 171441
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.2913641929626465,
      "learning_rate": 9.183870044197501e-05,
      "loss": 1.9698,
      "step": 171442
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.3943969011306763,
      "learning_rate": 9.183575485026826e-05,
      "loss": 2.0545,
      "step": 171443
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.3754340410232544,
      "learning_rate": 9.183280929726302e-05,
      "loss": 1.9791,
      "step": 171444
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.2344180345535278,
      "learning_rate": 9.18298637829601e-05,
      "loss": 1.836,
      "step": 171445
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.6920663118362427,
      "learning_rate": 9.18269183073598e-05,
      "loss": 1.989,
      "step": 171446
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.2380980253219604,
      "learning_rate": 9.182397287046292e-05,
      "loss": 1.9123,
      "step": 171447
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.34416663646698,
      "learning_rate": 9.182102747226981e-05,
      "loss": 2.1397,
      "step": 171448
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.260313868522644,
      "learning_rate": 9.181808211278126e-05,
      "loss": 1.9802,
      "step": 171449
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.317687749862671,
      "learning_rate": 9.18151367919975e-05,
      "loss": 2.1123,
      "step": 171450
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.350301742553711,
      "learning_rate": 9.181219150991934e-05,
      "loss": 2.0259,
      "step": 171451
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.2529181241989136,
      "learning_rate": 9.180924626654709e-05,
      "loss": 2.2056,
      "step": 171452
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.3047946691513062,
      "learning_rate": 9.180630106188154e-05,
      "loss": 2.1346,
      "step": 171453
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.2641737461090088,
      "learning_rate": 9.180335589592306e-05,
      "loss": 2.1446,
      "step": 171454
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.2668097019195557,
      "learning_rate": 9.180041076867242e-05,
      "loss": 1.8919,
      "step": 171455
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.2931424379348755,
      "learning_rate": 9.179746568012985e-05,
      "loss": 2.088,
      "step": 171456
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.2277990579605103,
      "learning_rate": 9.179452063029614e-05,
      "loss": 2.0732,
      "step": 171457
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.2857246398925781,
      "learning_rate": 9.179157561917167e-05,
      "loss": 2.0763,
      "step": 171458
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.2093483209609985,
      "learning_rate": 9.178863064675714e-05,
      "loss": 2.0657,
      "step": 171459
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.373616099357605,
      "learning_rate": 9.178568571305298e-05,
      "loss": 2.1997,
      "step": 171460
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.534906029701233,
      "learning_rate": 9.178274081805998e-05,
      "loss": 2.0513,
      "step": 171461
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.313959002494812,
      "learning_rate": 9.177979596177829e-05,
      "loss": 2.0478,
      "step": 171462
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.3304698467254639,
      "learning_rate": 9.177685114420877e-05,
      "loss": 2.1181,
      "step": 171463
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.3066866397857666,
      "learning_rate": 9.177390636535177e-05,
      "loss": 1.9635,
      "step": 171464
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.4520982503890991,
      "learning_rate": 9.177096162520803e-05,
      "loss": 2.1355,
      "step": 171465
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.3386904001235962,
      "learning_rate": 9.176801692377791e-05,
      "loss": 1.8947,
      "step": 171466
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.2286555767059326,
      "learning_rate": 9.17650722610622e-05,
      "loss": 2.1444,
      "step": 171467
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.3087055683135986,
      "learning_rate": 9.176212763706115e-05,
      "loss": 2.0614,
      "step": 171468
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.3252699375152588,
      "learning_rate": 9.17591830517755e-05,
      "loss": 1.9931,
      "step": 171469
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.2319486141204834,
      "learning_rate": 9.17562385052057e-05,
      "loss": 2.0903,
      "step": 171470
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.3858966827392578,
      "learning_rate": 9.17532939973524e-05,
      "loss": 1.8973,
      "step": 171471
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.2558283805847168,
      "learning_rate": 9.175034952821602e-05,
      "loss": 2.1692,
      "step": 171472
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.4922866821289062,
      "learning_rate": 9.174740509779736e-05,
      "loss": 2.1904,
      "step": 171473
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.4874353408813477,
      "learning_rate": 9.174446070609661e-05,
      "loss": 2.132,
      "step": 171474
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.359809398651123,
      "learning_rate": 9.174151635311462e-05,
      "loss": 1.9126,
      "step": 171475
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.3753001689910889,
      "learning_rate": 9.173857203885165e-05,
      "loss": 1.9183,
      "step": 171476
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.275503396987915,
      "learning_rate": 9.173562776330856e-05,
      "loss": 2.0163,
      "step": 171477
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.2001502513885498,
      "learning_rate": 9.173268352648564e-05,
      "loss": 1.8262,
      "step": 171478
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.4845913648605347,
      "learning_rate": 9.172973932838362e-05,
      "loss": 2.0916,
      "step": 171479
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.1929161548614502,
      "learning_rate": 9.1726795169003e-05,
      "loss": 1.9608,
      "step": 171480
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.2669484615325928,
      "learning_rate": 9.17238510483443e-05,
      "loss": 2.055,
      "step": 171481
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.3436245918273926,
      "learning_rate": 9.172090696640796e-05,
      "loss": 1.8376,
      "step": 171482
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.2614184617996216,
      "learning_rate": 9.171796292319474e-05,
      "loss": 1.7956,
      "step": 171483
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.35814368724823,
      "learning_rate": 9.171501891870495e-05,
      "loss": 2.0883,
      "step": 171484
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.305163025856018,
      "learning_rate": 9.171207495293942e-05,
      "loss": 1.9973,
      "step": 171485
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.3735214471817017,
      "learning_rate": 9.17091310258985e-05,
      "loss": 1.8293,
      "step": 171486
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.1837072372436523,
      "learning_rate": 9.170618713758282e-05,
      "loss": 1.8537,
      "step": 171487
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.234401822090149,
      "learning_rate": 9.170324328799279e-05,
      "loss": 2.0526,
      "step": 171488
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.2992171049118042,
      "learning_rate": 9.170029947712915e-05,
      "loss": 2.1245,
      "step": 171489
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.377700924873352,
      "learning_rate": 9.169735570499222e-05,
      "loss": 2.1063,
      "step": 171490
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.3304938077926636,
      "learning_rate": 9.169441197158283e-05,
      "loss": 2.0152,
      "step": 171491
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.3084160089492798,
      "learning_rate": 9.169146827690139e-05,
      "loss": 1.9888,
      "step": 171492
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.246443510055542,
      "learning_rate": 9.16885246209483e-05,
      "loss": 1.9544,
      "step": 171493
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.148430585861206,
      "learning_rate": 9.168558100372435e-05,
      "loss": 1.7804,
      "step": 171494
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.3564468622207642,
      "learning_rate": 9.168263742523e-05,
      "loss": 1.989,
      "step": 171495
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.526452898979187,
      "learning_rate": 9.16796938854657e-05,
      "loss": 1.9519,
      "step": 171496
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.3039944171905518,
      "learning_rate": 9.167675038443216e-05,
      "loss": 2.0111,
      "step": 171497
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.3820873498916626,
      "learning_rate": 9.167380692212982e-05,
      "loss": 1.9318,
      "step": 171498
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.432115077972412,
      "learning_rate": 9.167086349855918e-05,
      "loss": 2.0173,
      "step": 171499
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.516491174697876,
      "learning_rate": 9.166792011372092e-05,
      "loss": 2.0022,
      "step": 171500
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.3298944234848022,
      "learning_rate": 9.166497676761548e-05,
      "loss": 2.0398,
      "step": 171501
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.4385590553283691,
      "learning_rate": 9.16620334602435e-05,
      "loss": 1.9191,
      "step": 171502
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.254011631011963,
      "learning_rate": 9.16590901916055e-05,
      "loss": 1.7443,
      "step": 171503
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.3598127365112305,
      "learning_rate": 9.165614696170204e-05,
      "loss": 2.2215,
      "step": 171504
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.2980685234069824,
      "learning_rate": 9.165320377053349e-05,
      "loss": 2.1831,
      "step": 171505
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.3232450485229492,
      "learning_rate": 9.165026061810069e-05,
      "loss": 2.1244,
      "step": 171506
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.204687476158142,
      "learning_rate": 9.164731750440391e-05,
      "loss": 2.0982,
      "step": 171507
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.697716236114502,
      "learning_rate": 9.16443744294439e-05,
      "loss": 2.0255,
      "step": 171508
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.342002034187317,
      "learning_rate": 9.164143139322118e-05,
      "loss": 1.9149,
      "step": 171509
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.2347886562347412,
      "learning_rate": 9.163848839573623e-05,
      "loss": 2.0348,
      "step": 171510
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.1467523574829102,
      "learning_rate": 9.163554543698953e-05,
      "loss": 1.9386,
      "step": 171511
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.2800973653793335,
      "learning_rate": 9.163260251698182e-05,
      "loss": 2.091,
      "step": 171512
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.3771553039550781,
      "learning_rate": 9.16296596357134e-05,
      "loss": 1.8306,
      "step": 171513
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.243255615234375,
      "learning_rate": 9.16267167931851e-05,
      "loss": 2.0042,
      "step": 171514
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.4277679920196533,
      "learning_rate": 9.162377398939722e-05,
      "loss": 2.0385,
      "step": 171515
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.5912033319473267,
      "learning_rate": 9.162083122435062e-05,
      "loss": 2.2793,
      "step": 171516
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.2461966276168823,
      "learning_rate": 9.161788849804541e-05,
      "loss": 1.9625,
      "step": 171517
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.2785227298736572,
      "learning_rate": 9.161494581048249e-05,
      "loss": 2.1197,
      "step": 171518
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.2908564805984497,
      "learning_rate": 9.16120031616622e-05,
      "loss": 2.0189,
      "step": 171519
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.3214049339294434,
      "learning_rate": 9.160906055158524e-05,
      "loss": 1.7338,
      "step": 171520
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.3231215476989746,
      "learning_rate": 9.160611798025201e-05,
      "loss": 2.2353,
      "step": 171521
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.2801867723464966,
      "learning_rate": 9.160317544766331e-05,
      "loss": 2.0818,
      "step": 171522
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.3636964559555054,
      "learning_rate": 9.160023295381937e-05,
      "loss": 1.9882,
      "step": 171523
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.8813270330429077,
      "learning_rate": 9.159729049872091e-05,
      "loss": 2.0827,
      "step": 171524
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.4455945491790771,
      "learning_rate": 9.159434808236842e-05,
      "loss": 1.9657,
      "step": 171525
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.1559226512908936,
      "learning_rate": 9.159140570476252e-05,
      "loss": 1.8541,
      "step": 171526
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.2129238843917847,
      "learning_rate": 9.158846336590363e-05,
      "loss": 2.1398,
      "step": 171527
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.3695560693740845,
      "learning_rate": 9.158552106579258e-05,
      "loss": 2.1308,
      "step": 171528
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.3116568326950073,
      "learning_rate": 9.158257880442953e-05,
      "loss": 2.004,
      "step": 171529
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.3099110126495361,
      "learning_rate": 9.157963658181528e-05,
      "loss": 1.9755,
      "step": 171530
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.263556718826294,
      "learning_rate": 9.157669439795025e-05,
      "loss": 1.9032,
      "step": 171531
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.350244402885437,
      "learning_rate": 9.157375225283511e-05,
      "loss": 2.146,
      "step": 171532
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.5556707382202148,
      "learning_rate": 9.157081014647025e-05,
      "loss": 2.2507,
      "step": 171533
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.3530162572860718,
      "learning_rate": 9.15678680788565e-05,
      "loss": 2.0033,
      "step": 171534
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.5102421045303345,
      "learning_rate": 9.156492604999405e-05,
      "loss": 1.9698,
      "step": 171535
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.5011963844299316,
      "learning_rate": 9.156198405988372e-05,
      "loss": 2.1753,
      "step": 171536
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.3685137033462524,
      "learning_rate": 9.155904210852582e-05,
      "loss": 2.2519,
      "step": 171537
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.2696444988250732,
      "learning_rate": 9.155610019592113e-05,
      "loss": 2.0928,
      "step": 171538
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.265550136566162,
      "learning_rate": 9.155315832207001e-05,
      "loss": 1.9075,
      "step": 171539
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.2327972650527954,
      "learning_rate": 9.155021648697329e-05,
      "loss": 1.7222,
      "step": 171540
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.1995999813079834,
      "learning_rate": 9.154727469063112e-05,
      "loss": 2.0788,
      "step": 171541
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.213869571685791,
      "learning_rate": 9.154433293304436e-05,
      "loss": 1.9219,
      "step": 171542
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.2117689847946167,
      "learning_rate": 9.154139121421334e-05,
      "loss": 2.1138,
      "step": 171543
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.2511405944824219,
      "learning_rate": 9.153844953413877e-05,
      "loss": 1.7165,
      "step": 171544
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.5262010097503662,
      "learning_rate": 9.153550789282107e-05,
      "loss": 1.8439,
      "step": 171545
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.3766801357269287,
      "learning_rate": 9.153256629026107e-05,
      "loss": 2.2138,
      "step": 171546
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.1972273588180542,
      "learning_rate": 9.152962472645886e-05,
      "loss": 1.7897,
      "step": 171547
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.4004414081573486,
      "learning_rate": 9.152668320141535e-05,
      "loss": 1.9421,
      "step": 171548
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.3535324335098267,
      "learning_rate": 9.152374171513087e-05,
      "loss": 1.871,
      "step": 171549
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.222142219543457,
      "learning_rate": 9.152080026760615e-05,
      "loss": 1.9261,
      "step": 171550
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.3823944330215454,
      "learning_rate": 9.151785885884156e-05,
      "loss": 2.1857,
      "step": 171551
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.291286587715149,
      "learning_rate": 9.151491748883784e-05,
      "loss": 2.0058,
      "step": 171552
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.1617449522018433,
      "learning_rate": 9.15119761575954e-05,
      "loss": 1.8765,
      "step": 171553
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.2799090147018433,
      "learning_rate": 9.150903486511486e-05,
      "loss": 1.9058,
      "step": 171554
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.2487218379974365,
      "learning_rate": 9.150609361139662e-05,
      "loss": 2.0658,
      "step": 171555
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.313209891319275,
      "learning_rate": 9.15031523964414e-05,
      "loss": 2.1245,
      "step": 171556
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.4475574493408203,
      "learning_rate": 9.15002112202496e-05,
      "loss": 2.1356,
      "step": 171557
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.2796400785446167,
      "learning_rate": 9.149727008282196e-05,
      "loss": 1.8554,
      "step": 171558
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.3654388189315796,
      "learning_rate": 9.149432898415889e-05,
      "loss": 1.9557,
      "step": 171559
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.31952702999115,
      "learning_rate": 9.149138792426096e-05,
      "loss": 1.8598,
      "step": 171560
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.192138671875,
      "learning_rate": 9.148844690312863e-05,
      "loss": 2.0656,
      "step": 171561
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.3241791725158691,
      "learning_rate": 9.148550592076262e-05,
      "loss": 1.9357,
      "step": 171562
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.2763258218765259,
      "learning_rate": 9.148256497716331e-05,
      "loss": 2.2435,
      "step": 171563
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.2797950506210327,
      "learning_rate": 9.147962407233142e-05,
      "loss": 2.0575,
      "step": 171564
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.2439831495285034,
      "learning_rate": 9.147668320626741e-05,
      "loss": 2.2344,
      "step": 171565
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.1741372346878052,
      "learning_rate": 9.147374237897182e-05,
      "loss": 2.0043,
      "step": 171566
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.3483206033706665,
      "learning_rate": 9.147080159044509e-05,
      "loss": 2.0027,
      "step": 171567
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.4902106523513794,
      "learning_rate": 9.146786084068796e-05,
      "loss": 2.03,
      "step": 171568
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.3307335376739502,
      "learning_rate": 9.14649201297008e-05,
      "loss": 2.0902,
      "step": 171569
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.2917585372924805,
      "learning_rate": 9.146197945748437e-05,
      "loss": 1.9035,
      "step": 171570
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.4464342594146729,
      "learning_rate": 9.145903882403908e-05,
      "loss": 2.2066,
      "step": 171571
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.1591548919677734,
      "learning_rate": 9.145609822936549e-05,
      "loss": 1.8623,
      "step": 171572
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.278969168663025,
      "learning_rate": 9.145315767346405e-05,
      "loss": 2.0728,
      "step": 171573
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.2849587202072144,
      "learning_rate": 9.145021715633549e-05,
      "loss": 1.8858,
      "step": 171574
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.4952291250228882,
      "learning_rate": 9.144727667798019e-05,
      "loss": 1.988,
      "step": 171575
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.1048686504364014,
      "learning_rate": 9.144433623839886e-05,
      "loss": 2.0709,
      "step": 171576
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.3579390048980713,
      "learning_rate": 9.144139583759199e-05,
      "loss": 2.0211,
      "step": 171577
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.2245267629623413,
      "learning_rate": 9.143845547555997e-05,
      "loss": 2.1498,
      "step": 171578
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.3824127912521362,
      "learning_rate": 9.14355151523036e-05,
      "loss": 2.0727,
      "step": 171579
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.369002103805542,
      "learning_rate": 9.143257486782331e-05,
      "loss": 2.2073,
      "step": 171580
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.227810025215149,
      "learning_rate": 9.142963462211953e-05,
      "loss": 2.3257,
      "step": 171581
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.4464316368103027,
      "learning_rate": 9.1426694415193e-05,
      "loss": 2.335,
      "step": 171582
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.1899253129959106,
      "learning_rate": 9.142375424704422e-05,
      "loss": 1.9344,
      "step": 171583
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.2898283004760742,
      "learning_rate": 9.142081411767359e-05,
      "loss": 1.8136,
      "step": 171584
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.2532742023468018,
      "learning_rate": 9.141787402708187e-05,
      "loss": 1.8891,
      "step": 171585
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.201250672340393,
      "learning_rate": 9.141493397526947e-05,
      "loss": 1.8439,
      "step": 171586
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.2801847457885742,
      "learning_rate": 9.141199396223691e-05,
      "loss": 1.8446,
      "step": 171587
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.2191581726074219,
      "learning_rate": 9.140905398798489e-05,
      "loss": 1.9496,
      "step": 171588
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.2648682594299316,
      "learning_rate": 9.140611405251385e-05,
      "loss": 1.9748,
      "step": 171589
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.1790908575057983,
      "learning_rate": 9.140317415582426e-05,
      "loss": 1.8405,
      "step": 171590
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.4631705284118652,
      "learning_rate": 9.140023429791688e-05,
      "loss": 2.0233,
      "step": 171591
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.2474538087844849,
      "learning_rate": 9.139729447879201e-05,
      "loss": 2.0445,
      "step": 171592
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.3572951555252075,
      "learning_rate": 9.139435469845042e-05,
      "loss": 1.7972,
      "step": 171593
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.3020256757736206,
      "learning_rate": 9.139141495689257e-05,
      "loss": 2.1096,
      "step": 171594
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.4634922742843628,
      "learning_rate": 9.138847525411896e-05,
      "loss": 2.1586,
      "step": 171595
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.3548033237457275,
      "learning_rate": 9.13855355901301e-05,
      "loss": 1.9945,
      "step": 171596
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.4178396463394165,
      "learning_rate": 9.138259596492671e-05,
      "loss": 1.8975,
      "step": 171597
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.4232540130615234,
      "learning_rate": 9.137965637850913e-05,
      "loss": 2.1285,
      "step": 171598
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.3596962690353394,
      "learning_rate": 9.137671683087812e-05,
      "loss": 2.0567,
      "step": 171599
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.278137445449829,
      "learning_rate": 9.137377732203398e-05,
      "loss": 1.9852,
      "step": 171600
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.2510489225387573,
      "learning_rate": 9.137083785197762e-05,
      "loss": 1.8826,
      "step": 171601
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.2828872203826904,
      "learning_rate": 9.136789842070916e-05,
      "loss": 1.9072,
      "step": 171602
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.2713512182235718,
      "learning_rate": 9.136495902822944e-05,
      "loss": 2.0728,
      "step": 171603
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.4018499851226807,
      "learning_rate": 9.13620196745388e-05,
      "loss": 2.0901,
      "step": 171604
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.2994205951690674,
      "learning_rate": 9.135908035963803e-05,
      "loss": 2.0927,
      "step": 171605
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.3717364072799683,
      "learning_rate": 9.135614108352743e-05,
      "loss": 2.0846,
      "step": 171606
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.267842411994934,
      "learning_rate": 9.135320184620783e-05,
      "loss": 2.098,
      "step": 171607
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.278775691986084,
      "learning_rate": 9.135026264767947e-05,
      "loss": 1.9331,
      "step": 171608
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.2064393758773804,
      "learning_rate": 9.134732348794308e-05,
      "loss": 2.0299,
      "step": 171609
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.4106380939483643,
      "learning_rate": 9.134438436699908e-05,
      "loss": 2.1707,
      "step": 171610
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.3580398559570312,
      "learning_rate": 9.13414452848482e-05,
      "loss": 2.1099,
      "step": 171611
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.2290769815444946,
      "learning_rate": 9.133850624149077e-05,
      "loss": 1.9852,
      "step": 171612
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.3902721405029297,
      "learning_rate": 9.133556723692768e-05,
      "loss": 2.027,
      "step": 171613
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.2829633951187134,
      "learning_rate": 9.133262827115904e-05,
      "loss": 2.0238,
      "step": 171614
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.2414190769195557,
      "learning_rate": 9.132968934418569e-05,
      "loss": 2.0659,
      "step": 171615
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.2787123918533325,
      "learning_rate": 9.1326750456008e-05,
      "loss": 1.9558,
      "step": 171616
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.2748222351074219,
      "learning_rate": 9.132381160662668e-05,
      "loss": 1.9046,
      "step": 171617
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.5551412105560303,
      "learning_rate": 9.132087279604213e-05,
      "loss": 1.9791,
      "step": 171618
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.2099039554595947,
      "learning_rate": 9.131793402425508e-05,
      "loss": 2.0694,
      "step": 171619
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.337801456451416,
      "learning_rate": 9.131499529126592e-05,
      "loss": 1.8597,
      "step": 171620
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.4313974380493164,
      "learning_rate": 9.131205659707526e-05,
      "loss": 2.2449,
      "step": 171621
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.2461392879486084,
      "learning_rate": 9.130911794168358e-05,
      "loss": 1.9426,
      "step": 171622
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.4091194868087769,
      "learning_rate": 9.130617932509152e-05,
      "loss": 1.973,
      "step": 171623
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.235715389251709,
      "learning_rate": 9.130324074729947e-05,
      "loss": 1.9285,
      "step": 171624
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.2328001260757446,
      "learning_rate": 9.13003022083082e-05,
      "loss": 2.1568,
      "step": 171625
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.232216238975525,
      "learning_rate": 9.129736370811816e-05,
      "loss": 2.057,
      "step": 171626
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.2435791492462158,
      "learning_rate": 9.129442524672989e-05,
      "loss": 1.9756,
      "step": 171627
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.414997935295105,
      "learning_rate": 9.12914868241438e-05,
      "loss": 2.1436,
      "step": 171628
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.5749311447143555,
      "learning_rate": 9.128854844036068e-05,
      "loss": 2.0639,
      "step": 171629
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.335202932357788,
      "learning_rate": 9.128561009538085e-05,
      "loss": 2.0389,
      "step": 171630
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.2844823598861694,
      "learning_rate": 9.128267178920505e-05,
      "loss": 1.7185,
      "step": 171631
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.3567949533462524,
      "learning_rate": 9.127973352183376e-05,
      "loss": 2.1581,
      "step": 171632
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.360833764076233,
      "learning_rate": 9.127679529326752e-05,
      "loss": 2.12,
      "step": 171633
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.2211456298828125,
      "learning_rate": 9.127385710350672e-05,
      "loss": 2.0348,
      "step": 171634
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.2359298467636108,
      "learning_rate": 9.127091895255218e-05,
      "loss": 1.8821,
      "step": 171635
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.3901753425598145,
      "learning_rate": 9.126798084040421e-05,
      "loss": 2.0407,
      "step": 171636
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.2332394123077393,
      "learning_rate": 9.126504276706356e-05,
      "loss": 1.8771,
      "step": 171637
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.3125516176223755,
      "learning_rate": 9.12621047325307e-05,
      "loss": 2.0176,
      "step": 171638
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.198312520980835,
      "learning_rate": 9.125916673680613e-05,
      "loss": 1.8768,
      "step": 171639
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.1798394918441772,
      "learning_rate": 9.125622877989035e-05,
      "loss": 1.8926,
      "step": 171640
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.3237239122390747,
      "learning_rate": 9.125329086178404e-05,
      "loss": 1.9747,
      "step": 171641
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.385951042175293,
      "learning_rate": 9.125035298248761e-05,
      "loss": 1.896,
      "step": 171642
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.3612346649169922,
      "learning_rate": 9.124741514200175e-05,
      "loss": 2.2006,
      "step": 171643
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.303057312965393,
      "learning_rate": 9.124447734032696e-05,
      "loss": 2.0624,
      "step": 171644
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.2470608949661255,
      "learning_rate": 9.124153957746375e-05,
      "loss": 1.93,
      "step": 171645
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.2896267175674438,
      "learning_rate": 9.123860185341259e-05,
      "loss": 1.9424,
      "step": 171646
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.1616796255111694,
      "learning_rate": 9.123566416817422e-05,
      "loss": 1.9784,
      "step": 171647
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.3973805904388428,
      "learning_rate": 9.123272652174898e-05,
      "loss": 1.928,
      "step": 171648
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.2870028018951416,
      "learning_rate": 9.122978891413762e-05,
      "loss": 2.1055,
      "step": 171649
    },
    {
      "epoch": 2.23,
      "grad_norm": 1.251486897468567,
      "learning_rate": 9.122685134534056e-05,
      "loss": 1.9064,
      "step": 171650
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.2011768817901611,
      "learning_rate": 9.122391381535838e-05,
      "loss": 1.8447,
      "step": 171651
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.1986770629882812,
      "learning_rate": 9.12209763241915e-05,
      "loss": 2.1544,
      "step": 171652
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.2216767072677612,
      "learning_rate": 9.12180388718407e-05,
      "loss": 2.1137,
      "step": 171653
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.3298784494400024,
      "learning_rate": 9.121510145830633e-05,
      "loss": 1.9123,
      "step": 171654
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.4558990001678467,
      "learning_rate": 9.121216408358906e-05,
      "loss": 1.9191,
      "step": 171655
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.2181072235107422,
      "learning_rate": 9.120922674768942e-05,
      "loss": 1.9303,
      "step": 171656
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.2762047052383423,
      "learning_rate": 9.120628945060793e-05,
      "loss": 2.1421,
      "step": 171657
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.1911544799804688,
      "learning_rate": 9.120335219234504e-05,
      "loss": 1.9573,
      "step": 171658
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.3259847164154053,
      "learning_rate": 9.120041497290148e-05,
      "loss": 2.0949,
      "step": 171659
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.3674442768096924,
      "learning_rate": 9.11974777922776e-05,
      "loss": 2.1095,
      "step": 171660
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.2649511098861694,
      "learning_rate": 9.119454065047416e-05,
      "loss": 1.9762,
      "step": 171661
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.2846647500991821,
      "learning_rate": 9.119160354749157e-05,
      "loss": 1.9756,
      "step": 171662
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.247775912284851,
      "learning_rate": 9.118866648333033e-05,
      "loss": 1.9706,
      "step": 171663
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.239874243736267,
      "learning_rate": 9.118572945799115e-05,
      "loss": 1.9102,
      "step": 171664
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.3546943664550781,
      "learning_rate": 9.118279247147446e-05,
      "loss": 2.1627,
      "step": 171665
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.356062412261963,
      "learning_rate": 9.117985552378075e-05,
      "loss": 1.9438,
      "step": 171666
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.1538089513778687,
      "learning_rate": 9.117691861491075e-05,
      "loss": 2.0323,
      "step": 171667
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.3398970365524292,
      "learning_rate": 9.117398174486493e-05,
      "loss": 2.0145,
      "step": 171668
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.2269697189331055,
      "learning_rate": 9.117104491364368e-05,
      "loss": 2.0405,
      "step": 171669
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.3444651365280151,
      "learning_rate": 9.116810812124775e-05,
      "loss": 2.1437,
      "step": 171670
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.2231208086013794,
      "learning_rate": 9.116517136767766e-05,
      "loss": 2.0789,
      "step": 171671
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.3193902969360352,
      "learning_rate": 9.116223465293378e-05,
      "loss": 2.283,
      "step": 171672
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.3707258701324463,
      "learning_rate": 9.115929797701687e-05,
      "loss": 1.8993,
      "step": 171673
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.270244836807251,
      "learning_rate": 9.11563613399274e-05,
      "loss": 2.1416,
      "step": 171674
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.2196710109710693,
      "learning_rate": 9.115342474166584e-05,
      "loss": 2.1102,
      "step": 171675
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.2779687643051147,
      "learning_rate": 9.115048818223284e-05,
      "loss": 2.0252,
      "step": 171676
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.2978096008300781,
      "learning_rate": 9.114755166162884e-05,
      "loss": 2.0906,
      "step": 171677
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.2760273218154907,
      "learning_rate": 9.114461517985456e-05,
      "loss": 2.1136,
      "step": 171678
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.2958065271377563,
      "learning_rate": 9.114167873691043e-05,
      "loss": 2.035,
      "step": 171679
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.3573658466339111,
      "learning_rate": 9.1138742332797e-05,
      "loss": 2.0483,
      "step": 171680
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.3409347534179688,
      "learning_rate": 9.11358059675147e-05,
      "loss": 2.1483,
      "step": 171681
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.195863962173462,
      "learning_rate": 9.113286964106435e-05,
      "loss": 2.2064,
      "step": 171682
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.2942839860916138,
      "learning_rate": 9.112993335344621e-05,
      "loss": 2.1336,
      "step": 171683
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.2189133167266846,
      "learning_rate": 9.112699710466105e-05,
      "loss": 1.7966,
      "step": 171684
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.499840259552002,
      "learning_rate": 9.112406089470932e-05,
      "loss": 2.0428,
      "step": 171685
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.294104814529419,
      "learning_rate": 9.11211247235915e-05,
      "loss": 1.8177,
      "step": 171686
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.2997102737426758,
      "learning_rate": 9.111818859130828e-05,
      "loss": 1.9625,
      "step": 171687
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.250511884689331,
      "learning_rate": 9.111525249786014e-05,
      "loss": 1.8403,
      "step": 171688
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.2010375261306763,
      "learning_rate": 9.111231644324753e-05,
      "loss": 2.0394,
      "step": 171689
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.183279037475586,
      "learning_rate": 9.110938042747117e-05,
      "loss": 1.9356,
      "step": 171690
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.3013324737548828,
      "learning_rate": 9.11064444505314e-05,
      "loss": 2.2101,
      "step": 171691
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.370002031326294,
      "learning_rate": 9.110350851242904e-05,
      "loss": 2.2582,
      "step": 171692
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.2519968748092651,
      "learning_rate": 9.110057261316444e-05,
      "loss": 2.0955,
      "step": 171693
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.1667451858520508,
      "learning_rate": 9.109763675273817e-05,
      "loss": 2.0173,
      "step": 171694
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.2533271312713623,
      "learning_rate": 9.109470093115073e-05,
      "loss": 2.0186,
      "step": 171695
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.2405530214309692,
      "learning_rate": 9.109176514840279e-05,
      "loss": 2.1114,
      "step": 171696
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.2895023822784424,
      "learning_rate": 9.108882940449478e-05,
      "loss": 2.0067,
      "step": 171697
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.332787275314331,
      "learning_rate": 9.108589369942738e-05,
      "loss": 2.1149,
      "step": 171698
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.1592888832092285,
      "learning_rate": 9.108295803320108e-05,
      "loss": 2.0126,
      "step": 171699
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.3256006240844727,
      "learning_rate": 9.108002240581635e-05,
      "loss": 2.1604,
      "step": 171700
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.2943564653396606,
      "learning_rate": 9.10770868172737e-05,
      "loss": 2.0063,
      "step": 171701
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.3004428148269653,
      "learning_rate": 9.107415126757388e-05,
      "loss": 2.2594,
      "step": 171702
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.231127381324768,
      "learning_rate": 9.107121575671722e-05,
      "loss": 2.1421,
      "step": 171703
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.2990866899490356,
      "learning_rate": 9.106828028470449e-05,
      "loss": 2.1275,
      "step": 171704
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.5071892738342285,
      "learning_rate": 9.106534485153605e-05,
      "loss": 2.1843,
      "step": 171705
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.253769040107727,
      "learning_rate": 9.106240945721254e-05,
      "loss": 1.7683,
      "step": 171706
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.3583239316940308,
      "learning_rate": 9.105947410173439e-05,
      "loss": 2.0896,
      "step": 171707
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.2219170331954956,
      "learning_rate": 9.10565387851023e-05,
      "loss": 1.9318,
      "step": 171708
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.326904296875,
      "learning_rate": 9.105360350731667e-05,
      "loss": 1.9884,
      "step": 171709
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.2706338167190552,
      "learning_rate": 9.105066826837819e-05,
      "loss": 2.0199,
      "step": 171710
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.306619644165039,
      "learning_rate": 9.104773306828734e-05,
      "loss": 2.1305,
      "step": 171711
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.363390326499939,
      "learning_rate": 9.104479790704465e-05,
      "loss": 2.0682,
      "step": 171712
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.2294272184371948,
      "learning_rate": 9.104186278465062e-05,
      "loss": 1.9081,
      "step": 171713
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.3500335216522217,
      "learning_rate": 9.103892770110591e-05,
      "loss": 1.9459,
      "step": 171714
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.420902967453003,
      "learning_rate": 9.103599265641093e-05,
      "loss": 1.9402,
      "step": 171715
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.2794318199157715,
      "learning_rate": 9.10330576505664e-05,
      "loss": 2.1536,
      "step": 171716
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.294034719467163,
      "learning_rate": 9.103012268357276e-05,
      "loss": 2.0537,
      "step": 171717
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.52594792842865,
      "learning_rate": 9.102718775543057e-05,
      "loss": 1.9762,
      "step": 171718
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.2286896705627441,
      "learning_rate": 9.102425286614027e-05,
      "loss": 1.9616,
      "step": 171719
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.317375898361206,
      "learning_rate": 9.10213180157026e-05,
      "loss": 1.9774,
      "step": 171720
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.381467342376709,
      "learning_rate": 9.101838320411792e-05,
      "loss": 2.1866,
      "step": 171721
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.2645962238311768,
      "learning_rate": 9.101544843138699e-05,
      "loss": 2.0098,
      "step": 171722
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.4243828058242798,
      "learning_rate": 9.10125136975102e-05,
      "loss": 1.899,
      "step": 171723
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.2660858631134033,
      "learning_rate": 9.100957900248815e-05,
      "loss": 2.1335,
      "step": 171724
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.272013545036316,
      "learning_rate": 9.100664434632123e-05,
      "loss": 2.0518,
      "step": 171725
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.4950367212295532,
      "learning_rate": 9.100370972901024e-05,
      "loss": 2.0059,
      "step": 171726
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.338112235069275,
      "learning_rate": 9.100077515055551e-05,
      "loss": 2.1159,
      "step": 171727
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.280381679534912,
      "learning_rate": 9.099784061095778e-05,
      "loss": 2.169,
      "step": 171728
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.331487774848938,
      "learning_rate": 9.099490611021747e-05,
      "loss": 1.9081,
      "step": 171729
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.2328987121582031,
      "learning_rate": 9.099197164833519e-05,
      "loss": 1.8433,
      "step": 171730
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.2485495805740356,
      "learning_rate": 9.098903722531135e-05,
      "loss": 1.7187,
      "step": 171731
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.1532200574874878,
      "learning_rate": 9.098610284114666e-05,
      "loss": 1.8586,
      "step": 171732
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.268520474433899,
      "learning_rate": 9.098316849584153e-05,
      "loss": 1.9717,
      "step": 171733
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.2561767101287842,
      "learning_rate": 9.098023418939668e-05,
      "loss": 1.8855,
      "step": 171734
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.6311646699905396,
      "learning_rate": 9.097729992181253e-05,
      "loss": 2.2554,
      "step": 171735
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.48164963722229,
      "learning_rate": 9.097436569308963e-05,
      "loss": 2.1653,
      "step": 171736
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.3141453266143799,
      "learning_rate": 9.097143150322846e-05,
      "loss": 1.9553,
      "step": 171737
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.3518368005752563,
      "learning_rate": 9.096849735222975e-05,
      "loss": 2.251,
      "step": 171738
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.3529181480407715,
      "learning_rate": 9.096556324009383e-05,
      "loss": 2.0616,
      "step": 171739
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.3450462818145752,
      "learning_rate": 9.09626291668215e-05,
      "loss": 2.1608,
      "step": 171740
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.4023942947387695,
      "learning_rate": 9.095969513241309e-05,
      "loss": 2.0326,
      "step": 171741
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.1989233493804932,
      "learning_rate": 9.095676113686927e-05,
      "loss": 2.115,
      "step": 171742
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.248700499534607,
      "learning_rate": 9.095382718019043e-05,
      "loss": 1.9321,
      "step": 171743
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.3587998151779175,
      "learning_rate": 9.09508932623773e-05,
      "loss": 2.0956,
      "step": 171744
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.4425841569900513,
      "learning_rate": 9.094795938343025e-05,
      "loss": 1.8565,
      "step": 171745
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.4629502296447754,
      "learning_rate": 9.094502554335003e-05,
      "loss": 2.0457,
      "step": 171746
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.3590397834777832,
      "learning_rate": 9.094209174213706e-05,
      "loss": 2.0261,
      "step": 171747
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.419087529182434,
      "learning_rate": 9.09391579797919e-05,
      "loss": 2.0151,
      "step": 171748
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.518864393234253,
      "learning_rate": 9.093622425631503e-05,
      "loss": 1.9552,
      "step": 171749
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.4823452234268188,
      "learning_rate": 9.093329057170715e-05,
      "loss": 1.7752,
      "step": 171750
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.3127144575119019,
      "learning_rate": 9.093035692596861e-05,
      "loss": 2.0954,
      "step": 171751
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.311691403388977,
      "learning_rate": 9.092742331910016e-05,
      "loss": 1.9505,
      "step": 171752
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.5800584554672241,
      "learning_rate": 9.092448975110217e-05,
      "loss": 2.0711,
      "step": 171753
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.6192911863327026,
      "learning_rate": 9.092155622197532e-05,
      "loss": 1.9259,
      "step": 171754
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.2839913368225098,
      "learning_rate": 9.091862273172015e-05,
      "loss": 2.107,
      "step": 171755
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.275760293006897,
      "learning_rate": 9.091568928033712e-05,
      "loss": 1.9189,
      "step": 171756
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.3894445896148682,
      "learning_rate": 9.091275586782672e-05,
      "loss": 1.9543,
      "step": 171757
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.5630500316619873,
      "learning_rate": 9.090982249418969e-05,
      "loss": 2.0427,
      "step": 171758
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.4919424057006836,
      "learning_rate": 9.090688915942637e-05,
      "loss": 2.0329,
      "step": 171759
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.4070556163787842,
      "learning_rate": 9.090395586353749e-05,
      "loss": 2.2123,
      "step": 171760
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.2113245725631714,
      "learning_rate": 9.090102260652354e-05,
      "loss": 1.895,
      "step": 171761
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.9973465204238892,
      "learning_rate": 9.089808938838491e-05,
      "loss": 1.8978,
      "step": 171762
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.6190077066421509,
      "learning_rate": 9.089515620912238e-05,
      "loss": 2.1277,
      "step": 171763
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.3538143634796143,
      "learning_rate": 9.08922230687364e-05,
      "loss": 2.12,
      "step": 171764
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.1228076219558716,
      "learning_rate": 9.08892899672274e-05,
      "loss": 1.9223,
      "step": 171765
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.2126587629318237,
      "learning_rate": 9.088635690459612e-05,
      "loss": 1.8965,
      "step": 171766
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.3535257577896118,
      "learning_rate": 9.088342388084299e-05,
      "loss": 1.9294,
      "step": 171767
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.2693769931793213,
      "learning_rate": 9.088049089596851e-05,
      "loss": 1.7483,
      "step": 171768
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.166994571685791,
      "learning_rate": 9.087755794997342e-05,
      "loss": 2.092,
      "step": 171769
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.2439589500427246,
      "learning_rate": 9.087462504285807e-05,
      "loss": 2.0435,
      "step": 171770
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.3820409774780273,
      "learning_rate": 9.087169217462303e-05,
      "loss": 2.0083,
      "step": 171771
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.357204556465149,
      "learning_rate": 9.086875934526895e-05,
      "loss": 1.9376,
      "step": 171772
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.8836603164672852,
      "learning_rate": 9.086582655479635e-05,
      "loss": 1.9239,
      "step": 171773
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.2425631284713745,
      "learning_rate": 9.086289380320564e-05,
      "loss": 2.1392,
      "step": 171774
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.3823186159133911,
      "learning_rate": 9.085996109049756e-05,
      "loss": 2.0389,
      "step": 171775
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.2304298877716064,
      "learning_rate": 9.085702841667244e-05,
      "loss": 2.0847,
      "step": 171776
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.450005054473877,
      "learning_rate": 9.085409578173108e-05,
      "loss": 2.0736,
      "step": 171777
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.1656831502914429,
      "learning_rate": 9.085116318567386e-05,
      "loss": 1.7673,
      "step": 171778
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.2136479616165161,
      "learning_rate": 9.084823062850136e-05,
      "loss": 2.0815,
      "step": 171779
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.2621291875839233,
      "learning_rate": 9.084529811021406e-05,
      "loss": 2.046,
      "step": 171780
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.2179971933364868,
      "learning_rate": 9.084236563081262e-05,
      "loss": 2.1654,
      "step": 171781
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.3385393619537354,
      "learning_rate": 9.083943319029748e-05,
      "loss": 2.2338,
      "step": 171782
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.2072807550430298,
      "learning_rate": 9.083650078866929e-05,
      "loss": 2.1406,
      "step": 171783
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.4545823335647583,
      "learning_rate": 9.083356842592856e-05,
      "loss": 1.8712,
      "step": 171784
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.3064806461334229,
      "learning_rate": 9.083063610207582e-05,
      "loss": 2.1723,
      "step": 171785
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.3672006130218506,
      "learning_rate": 9.082770381711151e-05,
      "loss": 1.7507,
      "step": 171786
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.2799068689346313,
      "learning_rate": 9.08247715710364e-05,
      "loss": 1.9961,
      "step": 171787
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.2378711700439453,
      "learning_rate": 9.082183936385078e-05,
      "loss": 2.0798,
      "step": 171788
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.2513562440872192,
      "learning_rate": 9.081890719555545e-05,
      "loss": 1.7431,
      "step": 171789
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.6152982711791992,
      "learning_rate": 9.081597506615083e-05,
      "loss": 1.7519,
      "step": 171790
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.197325587272644,
      "learning_rate": 9.081304297563745e-05,
      "loss": 2.0096,
      "step": 171791
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.4843852519989014,
      "learning_rate": 9.081011092401583e-05,
      "loss": 1.9592,
      "step": 171792
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.210950493812561,
      "learning_rate": 9.080717891128658e-05,
      "loss": 1.8227,
      "step": 171793
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.3192918300628662,
      "learning_rate": 9.08042469374502e-05,
      "loss": 2.1537,
      "step": 171794
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.4658119678497314,
      "learning_rate": 9.08013150025073e-05,
      "loss": 2.1575,
      "step": 171795
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.274894118309021,
      "learning_rate": 9.079838310645839e-05,
      "loss": 2.2694,
      "step": 171796
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.2731940746307373,
      "learning_rate": 9.079545124930404e-05,
      "loss": 2.1854,
      "step": 171797
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.2559654712677002,
      "learning_rate": 9.079251943104468e-05,
      "loss": 1.929,
      "step": 171798
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.1653121709823608,
      "learning_rate": 9.0789587651681e-05,
      "loss": 2.0581,
      "step": 171799
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.2952425479888916,
      "learning_rate": 9.07866559112134e-05,
      "loss": 1.9205,
      "step": 171800
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.23611319065094,
      "learning_rate": 9.078372420964259e-05,
      "loss": 1.9969,
      "step": 171801
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.3202098608016968,
      "learning_rate": 9.078079254696907e-05,
      "loss": 2.1385,
      "step": 171802
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.18634831905365,
      "learning_rate": 9.07778609231933e-05,
      "loss": 1.9317,
      "step": 171803
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.2167527675628662,
      "learning_rate": 9.077492933831581e-05,
      "loss": 2.0298,
      "step": 171804
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.2298030853271484,
      "learning_rate": 9.077199779233731e-05,
      "loss": 2.029,
      "step": 171805
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.252119779586792,
      "learning_rate": 9.076906628525816e-05,
      "loss": 2.0473,
      "step": 171806
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.2460225820541382,
      "learning_rate": 9.076613481707905e-05,
      "loss": 1.9568,
      "step": 171807
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.4815157651901245,
      "learning_rate": 9.076320338780051e-05,
      "loss": 1.8892,
      "step": 171808
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.5189074277877808,
      "learning_rate": 9.0760271997423e-05,
      "loss": 1.9033,
      "step": 171809
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.4539003372192383,
      "learning_rate": 9.075734064594702e-05,
      "loss": 2.061,
      "step": 171810
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.3438085317611694,
      "learning_rate": 9.075440933337331e-05,
      "loss": 1.8901,
      "step": 171811
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.2397453784942627,
      "learning_rate": 9.075147805970219e-05,
      "loss": 2.1188,
      "step": 171812
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.2426952123641968,
      "learning_rate": 9.074854682493445e-05,
      "loss": 1.8947,
      "step": 171813
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.307734727859497,
      "learning_rate": 9.074561562907048e-05,
      "loss": 1.9917,
      "step": 171814
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.2868443727493286,
      "learning_rate": 9.074268447211083e-05,
      "loss": 1.7675,
      "step": 171815
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.3482024669647217,
      "learning_rate": 9.073975335405599e-05,
      "loss": 2.0883,
      "step": 171816
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.3219051361083984,
      "learning_rate": 9.07368222749067e-05,
      "loss": 2.3081,
      "step": 171817
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.4339962005615234,
      "learning_rate": 9.073389123466325e-05,
      "loss": 1.9685,
      "step": 171818
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.669001817703247,
      "learning_rate": 9.073096023332644e-05,
      "loss": 2.1804,
      "step": 171819
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.4585319757461548,
      "learning_rate": 9.072802927089658e-05,
      "loss": 2.2116,
      "step": 171820
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.3640141487121582,
      "learning_rate": 9.072509834737454e-05,
      "loss": 1.9949,
      "step": 171821
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.2321804761886597,
      "learning_rate": 9.072216746276046e-05,
      "loss": 2.0178,
      "step": 171822
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.3457069396972656,
      "learning_rate": 9.071923661705517e-05,
      "loss": 2.0402,
      "step": 171823
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.4771878719329834,
      "learning_rate": 9.071630581025901e-05,
      "loss": 2.0478,
      "step": 171824
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.503260612487793,
      "learning_rate": 9.071337504237277e-05,
      "loss": 2.1264,
      "step": 171825
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.2612379789352417,
      "learning_rate": 9.071044431339673e-05,
      "loss": 1.8284,
      "step": 171826
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.2957983016967773,
      "learning_rate": 9.070751362333181e-05,
      "loss": 1.8405,
      "step": 171827
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.2265523672103882,
      "learning_rate": 9.070458297217809e-05,
      "loss": 2.3624,
      "step": 171828
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.4672316312789917,
      "learning_rate": 9.070165235993644e-05,
      "loss": 2.0413,
      "step": 171829
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.3465572595596313,
      "learning_rate": 9.06987217866072e-05,
      "loss": 1.9795,
      "step": 171830
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.3123221397399902,
      "learning_rate": 9.06957912521911e-05,
      "loss": 1.8793,
      "step": 171831
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.3021693229675293,
      "learning_rate": 9.069286075668853e-05,
      "loss": 2.1896,
      "step": 171832
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.2924295663833618,
      "learning_rate": 9.068993030010031e-05,
      "loss": 2.0811,
      "step": 171833
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.37467360496521,
      "learning_rate": 9.068699988242659e-05,
      "loss": 2.0491,
      "step": 171834
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.6762957572937012,
      "learning_rate": 9.068406950366818e-05,
      "loss": 2.3001,
      "step": 171835
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.2599200010299683,
      "learning_rate": 9.068113916382546e-05,
      "loss": 2.0818,
      "step": 171836
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.2456612586975098,
      "learning_rate": 9.06782088628992e-05,
      "loss": 2.1244,
      "step": 171837
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.3593144416809082,
      "learning_rate": 9.06752786008897e-05,
      "loss": 2.1963,
      "step": 171838
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.2168574333190918,
      "learning_rate": 9.067234837779768e-05,
      "loss": 1.98,
      "step": 171839
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.4521111249923706,
      "learning_rate": 9.066941819362366e-05,
      "loss": 2.0303,
      "step": 171840
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.2839632034301758,
      "learning_rate": 9.066648804836812e-05,
      "loss": 2.0754,
      "step": 171841
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.2755099534988403,
      "learning_rate": 9.066355794203153e-05,
      "loss": 2.0937,
      "step": 171842
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.2445281744003296,
      "learning_rate": 9.066062787461466e-05,
      "loss": 2.1002,
      "step": 171843
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.2857742309570312,
      "learning_rate": 9.065769784611784e-05,
      "loss": 1.8691,
      "step": 171844
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.4688849449157715,
      "learning_rate": 9.065476785654178e-05,
      "loss": 1.8969,
      "step": 171845
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.3936759233474731,
      "learning_rate": 9.065183790588695e-05,
      "loss": 2.3791,
      "step": 171846
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.2791773080825806,
      "learning_rate": 9.064890799415388e-05,
      "loss": 1.8008,
      "step": 171847
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.2451584339141846,
      "learning_rate": 9.064597812134306e-05,
      "loss": 1.8947,
      "step": 171848
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.4071464538574219,
      "learning_rate": 9.064304828745519e-05,
      "loss": 2.001,
      "step": 171849
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.6615450382232666,
      "learning_rate": 9.064011849249065e-05,
      "loss": 1.7451,
      "step": 171850
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.676937460899353,
      "learning_rate": 9.063718873645015e-05,
      "loss": 1.7375,
      "step": 171851
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.2694894075393677,
      "learning_rate": 9.063425901933414e-05,
      "loss": 2.1284,
      "step": 171852
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.356816053390503,
      "learning_rate": 9.06313293411431e-05,
      "loss": 2.1041,
      "step": 171853
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.4945186376571655,
      "learning_rate": 9.062839970187772e-05,
      "loss": 1.9566,
      "step": 171854
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.203718662261963,
      "learning_rate": 9.062547010153846e-05,
      "loss": 2.0735,
      "step": 171855
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.2113220691680908,
      "learning_rate": 9.06225405401258e-05,
      "loss": 2.0378,
      "step": 171856
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.4981721639633179,
      "learning_rate": 9.061961101764048e-05,
      "loss": 1.9933,
      "step": 171857
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.2947059869766235,
      "learning_rate": 9.061668153408291e-05,
      "loss": 1.9754,
      "step": 171858
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.272269368171692,
      "learning_rate": 9.061375208945354e-05,
      "loss": 2.0894,
      "step": 171859
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.3323708772659302,
      "learning_rate": 9.061082268375313e-05,
      "loss": 2.2182,
      "step": 171860
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.3227882385253906,
      "learning_rate": 9.060789331698204e-05,
      "loss": 2.1297,
      "step": 171861
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.2966227531433105,
      "learning_rate": 9.060496398914098e-05,
      "loss": 2.1813,
      "step": 171862
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.3629446029663086,
      "learning_rate": 9.060203470023041e-05,
      "loss": 1.8091,
      "step": 171863
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.1977871656417847,
      "learning_rate": 9.059910545025086e-05,
      "loss": 2.1019,
      "step": 171864
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.3421564102172852,
      "learning_rate": 9.059617623920281e-05,
      "loss": 1.9711,
      "step": 171865
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.2883471250534058,
      "learning_rate": 9.059324706708698e-05,
      "loss": 2.0379,
      "step": 171866
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.3981359004974365,
      "learning_rate": 9.059031793390372e-05,
      "loss": 1.8962,
      "step": 171867
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.4032270908355713,
      "learning_rate": 9.058738883965376e-05,
      "loss": 1.8899,
      "step": 171868
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.332879900932312,
      "learning_rate": 9.058445978433757e-05,
      "loss": 2.2924,
      "step": 171869
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.4391200542449951,
      "learning_rate": 9.05815307679557e-05,
      "loss": 2.177,
      "step": 171870
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.3085981607437134,
      "learning_rate": 9.057860179050855e-05,
      "loss": 1.853,
      "step": 171871
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.2805556058883667,
      "learning_rate": 9.057567285199689e-05,
      "loss": 1.898,
      "step": 171872
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.340413212776184,
      "learning_rate": 9.057274395242108e-05,
      "loss": 1.9039,
      "step": 171873
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.2592592239379883,
      "learning_rate": 9.056981509178184e-05,
      "loss": 1.993,
      "step": 171874
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.4753986597061157,
      "learning_rate": 9.056688627007961e-05,
      "loss": 1.9812,
      "step": 171875
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.346527099609375,
      "learning_rate": 9.056395748731496e-05,
      "loss": 2.0519,
      "step": 171876
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.4411473274230957,
      "learning_rate": 9.056102874348835e-05,
      "loss": 2.0057,
      "step": 171877
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.5299320220947266,
      "learning_rate": 9.055810003860047e-05,
      "loss": 1.9824,
      "step": 171878
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.2688207626342773,
      "learning_rate": 9.055517137265171e-05,
      "loss": 1.9865,
      "step": 171879
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.4057201147079468,
      "learning_rate": 9.055224274564278e-05,
      "loss": 2.1479,
      "step": 171880
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.4921261072158813,
      "learning_rate": 9.054931415757416e-05,
      "loss": 1.7964,
      "step": 171881
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.1875842809677124,
      "learning_rate": 9.054638560844636e-05,
      "loss": 1.8912,
      "step": 171882
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.2390416860580444,
      "learning_rate": 9.054345709825984e-05,
      "loss": 1.9672,
      "step": 171883
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.2156636714935303,
      "learning_rate": 9.054052862701534e-05,
      "loss": 2.161,
      "step": 171884
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.2305184602737427,
      "learning_rate": 9.053760019471323e-05,
      "loss": 1.9343,
      "step": 171885
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.3486961126327515,
      "learning_rate": 9.053467180135422e-05,
      "loss": 2.0262,
      "step": 171886
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.4470707178115845,
      "learning_rate": 9.05317434469387e-05,
      "loss": 2.0298,
      "step": 171887
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.234625220298767,
      "learning_rate": 9.052881513146742e-05,
      "loss": 2.1966,
      "step": 171888
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.294299840927124,
      "learning_rate": 9.052588685494063e-05,
      "loss": 2.0826,
      "step": 171889
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.2350502014160156,
      "learning_rate": 9.052295861735912e-05,
      "loss": 2.1553,
      "step": 171890
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.3157850503921509,
      "learning_rate": 9.052003041872325e-05,
      "loss": 1.9307,
      "step": 171891
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.2360916137695312,
      "learning_rate": 9.051710225903377e-05,
      "loss": 1.9573,
      "step": 171892
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.3118923902511597,
      "learning_rate": 9.051417413829103e-05,
      "loss": 2.1498,
      "step": 171893
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.2686752080917358,
      "learning_rate": 9.051124605649583e-05,
      "loss": 2.0635,
      "step": 171894
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.1868919134140015,
      "learning_rate": 9.050831801364836e-05,
      "loss": 1.8127,
      "step": 171895
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.2383028268814087,
      "learning_rate": 9.050539000974942e-05,
      "loss": 1.9607,
      "step": 171896
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.2890771627426147,
      "learning_rate": 9.050246204479942e-05,
      "loss": 2.023,
      "step": 171897
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.4044897556304932,
      "learning_rate": 9.049953411879904e-05,
      "loss": 1.9772,
      "step": 171898
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.3356318473815918,
      "learning_rate": 9.049660623174866e-05,
      "loss": 2.1366,
      "step": 171899
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.359190583229065,
      "learning_rate": 9.049367838364909e-05,
      "loss": 2.267,
      "step": 171900
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.3150608539581299,
      "learning_rate": 9.049075057450054e-05,
      "loss": 1.9632,
      "step": 171901
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.292535662651062,
      "learning_rate": 9.048782280430379e-05,
      "loss": 2.0715,
      "step": 171902
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.3267006874084473,
      "learning_rate": 9.048489507305922e-05,
      "loss": 1.9176,
      "step": 171903
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.2741074562072754,
      "learning_rate": 9.048196738076755e-05,
      "loss": 1.8983,
      "step": 171904
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.1878432035446167,
      "learning_rate": 9.047903972742916e-05,
      "loss": 1.9898,
      "step": 171905
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.3035155534744263,
      "learning_rate": 9.047611211304487e-05,
      "loss": 1.9437,
      "step": 171906
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.3679428100585938,
      "learning_rate": 9.047318453761479e-05,
      "loss": 2.0008,
      "step": 171907
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.4316575527191162,
      "learning_rate": 9.047025700113982e-05,
      "loss": 2.2384,
      "step": 171908
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.2623370885849,
      "learning_rate": 9.046732950362032e-05,
      "loss": 2.0699,
      "step": 171909
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.1487765312194824,
      "learning_rate": 9.046440204505696e-05,
      "loss": 1.9102,
      "step": 171910
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.2702103853225708,
      "learning_rate": 9.046147462545013e-05,
      "loss": 1.986,
      "step": 171911
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.3154230117797852,
      "learning_rate": 9.045854724480066e-05,
      "loss": 2.1775,
      "step": 171912
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.2541680335998535,
      "learning_rate": 9.045561990310873e-05,
      "loss": 2.0033,
      "step": 171913
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.3163597583770752,
      "learning_rate": 9.045269260037512e-05,
      "loss": 1.7572,
      "step": 171914
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.3241175413131714,
      "learning_rate": 9.044976533660025e-05,
      "loss": 2.0798,
      "step": 171915
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.3482463359832764,
      "learning_rate": 9.044683811178477e-05,
      "loss": 2.1154,
      "step": 171916
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.2746469974517822,
      "learning_rate": 9.044391092592913e-05,
      "loss": 2.1955,
      "step": 171917
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.343887209892273,
      "learning_rate": 9.044098377903409e-05,
      "loss": 2.0689,
      "step": 171918
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.2548881769180298,
      "learning_rate": 9.043805667109982e-05,
      "loss": 1.974,
      "step": 171919
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.2694989442825317,
      "learning_rate": 9.043512960212719e-05,
      "loss": 2.1333,
      "step": 171920
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.2174524068832397,
      "learning_rate": 9.043220257211656e-05,
      "loss": 1.9466,
      "step": 171921
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.3074482679367065,
      "learning_rate": 9.042927558106858e-05,
      "loss": 1.994,
      "step": 171922
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.369210124015808,
      "learning_rate": 9.042634862898368e-05,
      "loss": 1.8925,
      "step": 171923
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.3736048936843872,
      "learning_rate": 9.042342171586257e-05,
      "loss": 2.021,
      "step": 171924
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.4037508964538574,
      "learning_rate": 9.04204948417057e-05,
      "loss": 2.1069,
      "step": 171925
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.2784093618392944,
      "learning_rate": 9.041756800651363e-05,
      "loss": 2.0137,
      "step": 171926
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.2556339502334595,
      "learning_rate": 9.041464121028679e-05,
      "loss": 2.0411,
      "step": 171927
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.2493404150009155,
      "learning_rate": 9.04117144530259e-05,
      "loss": 1.9601,
      "step": 171928
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.3478765487670898,
      "learning_rate": 9.040878773473136e-05,
      "loss": 1.8519,
      "step": 171929
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.2633346319198608,
      "learning_rate": 9.040586105540387e-05,
      "loss": 2.0673,
      "step": 171930
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.3199235200881958,
      "learning_rate": 9.040293441504388e-05,
      "loss": 2.1383,
      "step": 171931
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.3846895694732666,
      "learning_rate": 9.040000781365195e-05,
      "loss": 2.0944,
      "step": 171932
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.3486993312835693,
      "learning_rate": 9.039708125122852e-05,
      "loss": 2.0762,
      "step": 171933
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.3138681650161743,
      "learning_rate": 9.039415472777431e-05,
      "loss": 2.1727,
      "step": 171934
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.3930987119674683,
      "learning_rate": 9.03912282432897e-05,
      "loss": 2.2069,
      "step": 171935
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.238094687461853,
      "learning_rate": 9.038830179777543e-05,
      "loss": 2.2872,
      "step": 171936
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.3257886171340942,
      "learning_rate": 9.038537539123189e-05,
      "loss": 1.9527,
      "step": 171937
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.2981839179992676,
      "learning_rate": 9.038244902365959e-05,
      "loss": 1.9209,
      "step": 171938
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.2958416938781738,
      "learning_rate": 9.037952269505926e-05,
      "loss": 1.9224,
      "step": 171939
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.3188589811325073,
      "learning_rate": 9.037659640543131e-05,
      "loss": 2.0635,
      "step": 171940
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.2838730812072754,
      "learning_rate": 9.037367015477622e-05,
      "loss": 2.0913,
      "step": 171941
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.2988814115524292,
      "learning_rate": 9.037074394309473e-05,
      "loss": 2.0479,
      "step": 171942
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.2134884595870972,
      "learning_rate": 9.036781777038725e-05,
      "loss": 2.2313,
      "step": 171943
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.3199338912963867,
      "learning_rate": 9.036489163665427e-05,
      "loss": 1.9625,
      "step": 171944
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.3505265712738037,
      "learning_rate": 9.036196554189652e-05,
      "loss": 2.0727,
      "step": 171945
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.3082715272903442,
      "learning_rate": 9.035903948611431e-05,
      "loss": 2.4521,
      "step": 171946
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.1943352222442627,
      "learning_rate": 9.035611346930843e-05,
      "loss": 1.9796,
      "step": 171947
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.491131067276001,
      "learning_rate": 9.03531874914793e-05,
      "loss": 2.0349,
      "step": 171948
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.3015679121017456,
      "learning_rate": 9.035026155262748e-05,
      "loss": 1.9974,
      "step": 171949
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.356113076210022,
      "learning_rate": 9.03473356527534e-05,
      "loss": 2.2292,
      "step": 171950
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.3219331502914429,
      "learning_rate": 9.034440979185781e-05,
      "loss": 2.0508,
      "step": 171951
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.2545818090438843,
      "learning_rate": 9.034148396994106e-05,
      "loss": 2.0418,
      "step": 171952
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.6056926250457764,
      "learning_rate": 9.033855818700388e-05,
      "loss": 2.3024,
      "step": 171953
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.3354662656784058,
      "learning_rate": 9.03356324430467e-05,
      "loss": 2.092,
      "step": 171954
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.2118306159973145,
      "learning_rate": 9.033270673807009e-05,
      "loss": 2.0253,
      "step": 171955
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.3416600227355957,
      "learning_rate": 9.032978107207451e-05,
      "loss": 2.1483,
      "step": 171956
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.2587783336639404,
      "learning_rate": 9.032685544506067e-05,
      "loss": 2.0806,
      "step": 171957
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.7183990478515625,
      "learning_rate": 9.032392985702894e-05,
      "loss": 1.838,
      "step": 171958
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.3522919416427612,
      "learning_rate": 9.032100430798002e-05,
      "loss": 2.132,
      "step": 171959
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.3497145175933838,
      "learning_rate": 9.031807879791428e-05,
      "loss": 2.2057,
      "step": 171960
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.1930278539657593,
      "learning_rate": 9.031515332683257e-05,
      "loss": 1.7998,
      "step": 171961
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.2416104078292847,
      "learning_rate": 9.031222789473506e-05,
      "loss": 1.9171,
      "step": 171962
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.2476149797439575,
      "learning_rate": 9.030930250162255e-05,
      "loss": 2.0544,
      "step": 171963
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.187916874885559,
      "learning_rate": 9.030637714749538e-05,
      "loss": 1.8045,
      "step": 171964
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.332348108291626,
      "learning_rate": 9.030345183235436e-05,
      "loss": 2.2042,
      "step": 171965
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.2520009279251099,
      "learning_rate": 9.030052655619975e-05,
      "loss": 2.0637,
      "step": 171966
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.3083080053329468,
      "learning_rate": 9.029760131903242e-05,
      "loss": 1.9041,
      "step": 171967
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.2570301294326782,
      "learning_rate": 9.029467612085254e-05,
      "loss": 2.0777,
      "step": 171968
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.2588344812393188,
      "learning_rate": 9.029175096166093e-05,
      "loss": 2.196,
      "step": 171969
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.2110902070999146,
      "learning_rate": 9.028882584145798e-05,
      "loss": 2.3012,
      "step": 171970
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.467472791671753,
      "learning_rate": 9.028590076024437e-05,
      "loss": 2.2096,
      "step": 171971
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.345657467842102,
      "learning_rate": 9.028297571802047e-05,
      "loss": 2.0139,
      "step": 171972
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.2216545343399048,
      "learning_rate": 9.028005071478712e-05,
      "loss": 1.9134,
      "step": 171973
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.2687227725982666,
      "learning_rate": 9.027712575054446e-05,
      "loss": 1.8811,
      "step": 171974
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.2189099788665771,
      "learning_rate": 9.027420082529334e-05,
      "loss": 1.8333,
      "step": 171975
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.603366494178772,
      "learning_rate": 9.02712759390341e-05,
      "loss": 2.1878,
      "step": 171976
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.150742530822754,
      "learning_rate": 9.026835109176753e-05,
      "loss": 2.0898,
      "step": 171977
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.263243556022644,
      "learning_rate": 9.026542628349391e-05,
      "loss": 2.1178,
      "step": 171978
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.2862372398376465,
      "learning_rate": 9.026250151421408e-05,
      "loss": 2.0655,
      "step": 171979
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.3260741233825684,
      "learning_rate": 9.025957678392825e-05,
      "loss": 1.8023,
      "step": 171980
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.2932405471801758,
      "learning_rate": 9.025665209263721e-05,
      "loss": 1.9662,
      "step": 171981
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.1807962656021118,
      "learning_rate": 9.02537274403413e-05,
      "loss": 1.9218,
      "step": 171982
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.3630454540252686,
      "learning_rate": 9.02508028270413e-05,
      "loss": 2.3183,
      "step": 171983
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.2309201955795288,
      "learning_rate": 9.024787825273752e-05,
      "loss": 1.9764,
      "step": 171984
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.4740946292877197,
      "learning_rate": 9.024495371743083e-05,
      "loss": 2.209,
      "step": 171985
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.311877727508545,
      "learning_rate": 9.024202922112135e-05,
      "loss": 2.0507,
      "step": 171986
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.2746014595031738,
      "learning_rate": 9.023910476380996e-05,
      "loss": 2.1956,
      "step": 171987
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.2512072324752808,
      "learning_rate": 9.023618034549695e-05,
      "loss": 2.2829,
      "step": 171988
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.2844101190567017,
      "learning_rate": 9.023325596618312e-05,
      "loss": 1.9743,
      "step": 171989
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.3332710266113281,
      "learning_rate": 9.023033162586878e-05,
      "loss": 1.8318,
      "step": 171990
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.2499009370803833,
      "learning_rate": 9.022740732455478e-05,
      "loss": 2.0883,
      "step": 171991
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.3067033290863037,
      "learning_rate": 9.022448306224129e-05,
      "loss": 2.0231,
      "step": 171992
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.2486732006072998,
      "learning_rate": 9.022155883892909e-05,
      "loss": 2.0895,
      "step": 171993
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.2066185474395752,
      "learning_rate": 9.021863465461855e-05,
      "loss": 2.2012,
      "step": 171994
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.3586318492889404,
      "learning_rate": 9.021571050931047e-05,
      "loss": 2.2677,
      "step": 171995
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.4524643421173096,
      "learning_rate": 9.021278640300514e-05,
      "loss": 2.1133,
      "step": 171996
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.2647058963775635,
      "learning_rate": 9.020986233570339e-05,
      "loss": 2.0875,
      "step": 171997
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.3330979347229004,
      "learning_rate": 9.02069383074054e-05,
      "loss": 1.9297,
      "step": 171998
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.1909950971603394,
      "learning_rate": 9.020401431811201e-05,
      "loss": 2.392,
      "step": 171999
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.40838623046875,
      "learning_rate": 9.020109036782355e-05,
      "loss": 2.0992,
      "step": 172000
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.3139876127243042,
      "learning_rate": 9.019816645654075e-05,
      "loss": 1.9977,
      "step": 172001
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.2406585216522217,
      "learning_rate": 9.0195242584264e-05,
      "loss": 2.0884,
      "step": 172002
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.4108902215957642,
      "learning_rate": 9.019231875099406e-05,
      "loss": 2.2439,
      "step": 172003
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.3455510139465332,
      "learning_rate": 9.018939495673119e-05,
      "loss": 1.9318,
      "step": 172004
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.3841078281402588,
      "learning_rate": 9.018647120147613e-05,
      "loss": 1.944,
      "step": 172005
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.2877922058105469,
      "learning_rate": 9.018354748522929e-05,
      "loss": 2.0319,
      "step": 172006
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.286394715309143,
      "learning_rate": 9.018062380799138e-05,
      "loss": 1.98,
      "step": 172007
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.3225921392440796,
      "learning_rate": 9.017770016976277e-05,
      "loss": 2.2427,
      "step": 172008
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.3879677057266235,
      "learning_rate": 9.017477657054425e-05,
      "loss": 2.1204,
      "step": 172009
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.4298802614212036,
      "learning_rate": 9.017185301033603e-05,
      "loss": 1.7529,
      "step": 172010
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.276819109916687,
      "learning_rate": 9.016892948913893e-05,
      "loss": 1.8763,
      "step": 172011
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.1861587762832642,
      "learning_rate": 9.016600600695327e-05,
      "loss": 1.9274,
      "step": 172012
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.2871373891830444,
      "learning_rate": 9.016308256377978e-05,
      "loss": 2.0827,
      "step": 172013
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.2507268190383911,
      "learning_rate": 9.01601591596189e-05,
      "loss": 2.2286,
      "step": 172014
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.2457116842269897,
      "learning_rate": 9.015723579447125e-05,
      "loss": 1.9661,
      "step": 172015
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.434726357460022,
      "learning_rate": 9.015431246833736e-05,
      "loss": 2.2301,
      "step": 172016
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.505271553993225,
      "learning_rate": 9.015138918121775e-05,
      "loss": 2.2082,
      "step": 172017
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.2758104801177979,
      "learning_rate": 9.014846593311283e-05,
      "loss": 2.2202,
      "step": 172018
    },
    {
      "epoch": 2.24,
      "grad_norm": 2.3144679069519043,
      "learning_rate": 9.014554272402341e-05,
      "loss": 2.0063,
      "step": 172019
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.3079848289489746,
      "learning_rate": 9.014261955394977e-05,
      "loss": 2.0343,
      "step": 172020
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.4018304347991943,
      "learning_rate": 9.013969642289267e-05,
      "loss": 2.2299,
      "step": 172021
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.4234929084777832,
      "learning_rate": 9.013677333085258e-05,
      "loss": 2.0115,
      "step": 172022
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.301382064819336,
      "learning_rate": 9.013385027782993e-05,
      "loss": 1.9034,
      "step": 172023
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.3227168321609497,
      "learning_rate": 9.013092726382545e-05,
      "loss": 1.8148,
      "step": 172024
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.2068989276885986,
      "learning_rate": 9.012800428883956e-05,
      "loss": 1.9595,
      "step": 172025
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.2510154247283936,
      "learning_rate": 9.012508135287276e-05,
      "loss": 1.9778,
      "step": 172026
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.3369146585464478,
      "learning_rate": 9.012215845592579e-05,
      "loss": 2.0203,
      "step": 172027
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.3913323879241943,
      "learning_rate": 9.011923559799906e-05,
      "loss": 2.079,
      "step": 172028
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.2608989477157593,
      "learning_rate": 9.011631277909302e-05,
      "loss": 1.9569,
      "step": 172029
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.335734486579895,
      "learning_rate": 9.01133899992084e-05,
      "loss": 2.1201,
      "step": 172030
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.4313911199569702,
      "learning_rate": 9.01104672583457e-05,
      "loss": 1.9912,
      "step": 172031
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.2570024728775024,
      "learning_rate": 9.010754455650533e-05,
      "loss": 1.8168,
      "step": 172032
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.2995623350143433,
      "learning_rate": 9.0104621893688e-05,
      "loss": 1.9445,
      "step": 172033
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.223323106765747,
      "learning_rate": 9.01016992698942e-05,
      "loss": 2.0183,
      "step": 172034
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.1448800563812256,
      "learning_rate": 9.009877668512433e-05,
      "loss": 1.9445,
      "step": 172035
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.2630610466003418,
      "learning_rate": 9.009585413937919e-05,
      "loss": 1.888,
      "step": 172036
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.2264528274536133,
      "learning_rate": 9.00929316326591e-05,
      "loss": 2.0056,
      "step": 172037
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.2967298030853271,
      "learning_rate": 9.009000916496475e-05,
      "loss": 1.9842,
      "step": 172038
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.168010950088501,
      "learning_rate": 9.008708673629666e-05,
      "loss": 1.7964,
      "step": 172039
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.3512388467788696,
      "learning_rate": 9.008416434665534e-05,
      "loss": 1.9064,
      "step": 172040
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.2754501104354858,
      "learning_rate": 9.008124199604124e-05,
      "loss": 1.8143,
      "step": 172041
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.3525052070617676,
      "learning_rate": 9.00783196844551e-05,
      "loss": 1.8842,
      "step": 172042
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.3269520998001099,
      "learning_rate": 9.007539741189725e-05,
      "loss": 1.9382,
      "step": 172043
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.2695246934890747,
      "learning_rate": 9.007247517836847e-05,
      "loss": 1.9309,
      "step": 172044
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.4143491983413696,
      "learning_rate": 9.006955298386906e-05,
      "loss": 2.3118,
      "step": 172045
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.2798970937728882,
      "learning_rate": 9.00666308283999e-05,
      "loss": 2.0748,
      "step": 172046
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.4403802156448364,
      "learning_rate": 9.00637087119611e-05,
      "loss": 2.3108,
      "step": 172047
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.5475974082946777,
      "learning_rate": 9.006078663455353e-05,
      "loss": 1.9844,
      "step": 172048
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.2788171768188477,
      "learning_rate": 9.00578645961775e-05,
      "loss": 1.8309,
      "step": 172049
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.346944808959961,
      "learning_rate": 9.005494259683381e-05,
      "loss": 1.9645,
      "step": 172050
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.345637321472168,
      "learning_rate": 9.005202063652273e-05,
      "loss": 1.9772,
      "step": 172051
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.2540602684020996,
      "learning_rate": 9.004909871524517e-05,
      "loss": 2.0117,
      "step": 172052
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.3299977779388428,
      "learning_rate": 9.004617683300124e-05,
      "loss": 2.1357,
      "step": 172053
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.220801591873169,
      "learning_rate": 9.00432549897918e-05,
      "loss": 1.8855,
      "step": 172054
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.2022234201431274,
      "learning_rate": 9.004033318561717e-05,
      "loss": 2.2367,
      "step": 172055
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.285423994064331,
      "learning_rate": 9.00374114204781e-05,
      "loss": 2.1372,
      "step": 172056
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.2726086378097534,
      "learning_rate": 9.003448969437495e-05,
      "loss": 1.943,
      "step": 172057
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.1998168230056763,
      "learning_rate": 9.003156800730853e-05,
      "loss": 2.0162,
      "step": 172058
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.2403569221496582,
      "learning_rate": 9.002864635927902e-05,
      "loss": 2.1577,
      "step": 172059
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.5000953674316406,
      "learning_rate": 9.002572475028726e-05,
      "loss": 1.911,
      "step": 172060
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.3628649711608887,
      "learning_rate": 9.002280318033359e-05,
      "loss": 1.9847,
      "step": 172061
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.3677749633789062,
      "learning_rate": 9.001988164941871e-05,
      "loss": 1.8263,
      "step": 172062
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.2508829832077026,
      "learning_rate": 9.001696015754302e-05,
      "loss": 1.969,
      "step": 172063
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.4028047323226929,
      "learning_rate": 9.001403870470734e-05,
      "loss": 1.9837,
      "step": 172064
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.4669866561889648,
      "learning_rate": 9.00111172909118e-05,
      "loss": 2.2035,
      "step": 172065
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.293653964996338,
      "learning_rate": 9.000819591615725e-05,
      "loss": 2.1266,
      "step": 172066
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.3805502653121948,
      "learning_rate": 9.000527458044405e-05,
      "loss": 1.9725,
      "step": 172067
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.2145960330963135,
      "learning_rate": 9.000235328377296e-05,
      "loss": 2.1199,
      "step": 172068
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.4214107990264893,
      "learning_rate": 8.999943202614426e-05,
      "loss": 2.0737,
      "step": 172069
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.272851824760437,
      "learning_rate": 8.999651080755886e-05,
      "loss": 1.9704,
      "step": 172070
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.3181004524230957,
      "learning_rate": 8.999358962801684e-05,
      "loss": 2.1134,
      "step": 172071
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.2342262268066406,
      "learning_rate": 8.999066848751911e-05,
      "loss": 2.0322,
      "step": 172072
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.2414802312850952,
      "learning_rate": 8.998774738606595e-05,
      "loss": 1.9689,
      "step": 172073
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.1890642642974854,
      "learning_rate": 8.998482632365816e-05,
      "loss": 1.9777,
      "step": 172074
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.3414198160171509,
      "learning_rate": 8.998190530029604e-05,
      "loss": 2.1646,
      "step": 172075
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.5701323747634888,
      "learning_rate": 8.997898431598045e-05,
      "loss": 2.0574,
      "step": 172076
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.2852624654769897,
      "learning_rate": 8.997606337071153e-05,
      "loss": 2.194,
      "step": 172077
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.3318312168121338,
      "learning_rate": 8.99731424644901e-05,
      "loss": 1.9743,
      "step": 172078
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.3316597938537598,
      "learning_rate": 8.997022159731657e-05,
      "loss": 1.9363,
      "step": 172079
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.2627615928649902,
      "learning_rate": 8.99673007691916e-05,
      "loss": 2.1305,
      "step": 172080
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.4201024770736694,
      "learning_rate": 8.99643799801156e-05,
      "loss": 2.2018,
      "step": 172081
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.4109013080596924,
      "learning_rate": 8.996145923008941e-05,
      "loss": 1.8793,
      "step": 172082
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.4869645833969116,
      "learning_rate": 8.99585385191131e-05,
      "loss": 2.0565,
      "step": 172083
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.5099729299545288,
      "learning_rate": 8.995561784718758e-05,
      "loss": 2.0209,
      "step": 172084
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.2628891468048096,
      "learning_rate": 8.995269721431317e-05,
      "loss": 2.2539,
      "step": 172085
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.4283525943756104,
      "learning_rate": 8.994977662049066e-05,
      "loss": 1.8887,
      "step": 172086
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.2029228210449219,
      "learning_rate": 8.99468560657203e-05,
      "loss": 2.2056,
      "step": 172087
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.474071741104126,
      "learning_rate": 8.99439355500029e-05,
      "loss": 1.8477,
      "step": 172088
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.4467695951461792,
      "learning_rate": 8.99410150733389e-05,
      "loss": 1.9948,
      "step": 172089
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.2737572193145752,
      "learning_rate": 8.993809463572882e-05,
      "loss": 2.092,
      "step": 172090
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.3337243795394897,
      "learning_rate": 8.99351742371731e-05,
      "loss": 1.7689,
      "step": 172091
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.3605005741119385,
      "learning_rate": 8.993225387767251e-05,
      "loss": 1.6876,
      "step": 172092
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.4859033823013306,
      "learning_rate": 8.992933355722741e-05,
      "loss": 1.9613,
      "step": 172093
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.251889944076538,
      "learning_rate": 8.992641327583847e-05,
      "loss": 1.983,
      "step": 172094
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.3536180257797241,
      "learning_rate": 8.992349303350619e-05,
      "loss": 2.1709,
      "step": 172095
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.2602378129959106,
      "learning_rate": 8.992057283023109e-05,
      "loss": 2.0315,
      "step": 172096
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.2979555130004883,
      "learning_rate": 8.991765266601363e-05,
      "loss": 1.8466,
      "step": 172097
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.5824114084243774,
      "learning_rate": 8.991473254085455e-05,
      "loss": 2.0773,
      "step": 172098
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.2645338773727417,
      "learning_rate": 8.991181245475416e-05,
      "loss": 2.0883,
      "step": 172099
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.3782764673233032,
      "learning_rate": 8.990889240771326e-05,
      "loss": 2.0838,
      "step": 172100
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.612713098526001,
      "learning_rate": 8.990597239973226e-05,
      "loss": 1.9693,
      "step": 172101
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.3269734382629395,
      "learning_rate": 8.99030524308117e-05,
      "loss": 1.9377,
      "step": 172102
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.341637134552002,
      "learning_rate": 8.990013250095203e-05,
      "loss": 1.863,
      "step": 172103
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.3068073987960815,
      "learning_rate": 8.989721261015398e-05,
      "loss": 2.2235,
      "step": 172104
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.3842568397521973,
      "learning_rate": 8.989429275841788e-05,
      "loss": 1.9821,
      "step": 172105
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.5562551021575928,
      "learning_rate": 8.989137294574454e-05,
      "loss": 2.0741,
      "step": 172106
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.295069694519043,
      "learning_rate": 8.988845317213436e-05,
      "loss": 2.056,
      "step": 172107
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.235887885093689,
      "learning_rate": 8.988553343758776e-05,
      "loss": 1.8526,
      "step": 172108
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.3611935377120972,
      "learning_rate": 8.988261374210551e-05,
      "loss": 1.9525,
      "step": 172109
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.2229256629943848,
      "learning_rate": 8.987969408568807e-05,
      "loss": 1.8615,
      "step": 172110
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.425424575805664,
      "learning_rate": 8.987677446833582e-05,
      "loss": 2.1209,
      "step": 172111
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.330569863319397,
      "learning_rate": 8.987385489004959e-05,
      "loss": 2.2005,
      "step": 172112
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.4852370023727417,
      "learning_rate": 8.987093535082974e-05,
      "loss": 2.1738,
      "step": 172113
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.3181488513946533,
      "learning_rate": 8.986801585067678e-05,
      "loss": 1.8337,
      "step": 172114
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.3357609510421753,
      "learning_rate": 8.98650963895914e-05,
      "loss": 2.0733,
      "step": 172115
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.791759967803955,
      "learning_rate": 8.986217696757406e-05,
      "loss": 1.9746,
      "step": 172116
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.4483059644699097,
      "learning_rate": 8.985925758462524e-05,
      "loss": 2.0185,
      "step": 172117
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.3824225664138794,
      "learning_rate": 8.985633824074564e-05,
      "loss": 2.1267,
      "step": 172118
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.1843866109848022,
      "learning_rate": 8.985341893593569e-05,
      "loss": 2.0013,
      "step": 172119
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.3058871030807495,
      "learning_rate": 8.985049967019587e-05,
      "loss": 2.1705,
      "step": 172120
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.3066275119781494,
      "learning_rate": 8.984758044352691e-05,
      "loss": 1.9925,
      "step": 172121
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.8018826246261597,
      "learning_rate": 8.984466125592917e-05,
      "loss": 2.0223,
      "step": 172122
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.3107268810272217,
      "learning_rate": 8.984174210740336e-05,
      "loss": 1.7266,
      "step": 172123
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.3131917715072632,
      "learning_rate": 8.983882299794993e-05,
      "loss": 1.9666,
      "step": 172124
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.4389548301696777,
      "learning_rate": 8.983590392756942e-05,
      "loss": 2.0297,
      "step": 172125
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.3693044185638428,
      "learning_rate": 8.983298489626229e-05,
      "loss": 2.0159,
      "step": 172126
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.5552107095718384,
      "learning_rate": 8.983006590402929e-05,
      "loss": 1.7929,
      "step": 172127
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.7216765880584717,
      "learning_rate": 8.982714695087075e-05,
      "loss": 1.9543,
      "step": 172128
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.369627833366394,
      "learning_rate": 8.982422803678741e-05,
      "loss": 2.1565,
      "step": 172129
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.516281008720398,
      "learning_rate": 8.982130916177968e-05,
      "loss": 1.9208,
      "step": 172130
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.3858946561813354,
      "learning_rate": 8.981839032584815e-05,
      "loss": 2.1158,
      "step": 172131
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.4617631435394287,
      "learning_rate": 8.981547152899328e-05,
      "loss": 2.0479,
      "step": 172132
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.5660185813903809,
      "learning_rate": 8.981255277121574e-05,
      "loss": 1.87,
      "step": 172133
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.3443880081176758,
      "learning_rate": 8.980963405251594e-05,
      "loss": 2.1628,
      "step": 172134
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.2453408241271973,
      "learning_rate": 8.980671537289459e-05,
      "loss": 1.9247,
      "step": 172135
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.4426746368408203,
      "learning_rate": 8.980379673235204e-05,
      "loss": 2.0784,
      "step": 172136
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.4468913078308105,
      "learning_rate": 8.980087813088913e-05,
      "loss": 1.8431,
      "step": 172137
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.6822240352630615,
      "learning_rate": 8.979795956850603e-05,
      "loss": 2.0228,
      "step": 172138
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.3640691041946411,
      "learning_rate": 8.979504104520353e-05,
      "loss": 1.9109,
      "step": 172139
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.3897244930267334,
      "learning_rate": 8.9792122560982e-05,
      "loss": 2.031,
      "step": 172140
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.4531656503677368,
      "learning_rate": 8.978920411584218e-05,
      "loss": 2.0098,
      "step": 172141
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.2430877685546875,
      "learning_rate": 8.978628570978443e-05,
      "loss": 2.1413,
      "step": 172142
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.519438624382019,
      "learning_rate": 8.978336734280955e-05,
      "loss": 1.9694,
      "step": 172143
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.3381456136703491,
      "learning_rate": 8.978044901491775e-05,
      "loss": 2.2268,
      "step": 172144
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.2280179262161255,
      "learning_rate": 8.97775307261098e-05,
      "loss": 1.8092,
      "step": 172145
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.3229279518127441,
      "learning_rate": 8.97746124763861e-05,
      "loss": 1.9165,
      "step": 172146
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.1989102363586426,
      "learning_rate": 8.977169426574733e-05,
      "loss": 1.9502,
      "step": 172147
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.268776297569275,
      "learning_rate": 8.97687760941939e-05,
      "loss": 1.9733,
      "step": 172148
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.2892816066741943,
      "learning_rate": 8.976585796172663e-05,
      "loss": 2.0571,
      "step": 172149
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.3141785860061646,
      "learning_rate": 8.976293986834563e-05,
      "loss": 1.9788,
      "step": 172150
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.2142682075500488,
      "learning_rate": 8.97600218140518e-05,
      "loss": 1.9707,
      "step": 172151
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.252400517463684,
      "learning_rate": 8.975710379884543e-05,
      "loss": 2.1996,
      "step": 172152
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.2438023090362549,
      "learning_rate": 8.975418582272726e-05,
      "loss": 2.1255,
      "step": 172153
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.188592791557312,
      "learning_rate": 8.97512678856977e-05,
      "loss": 1.981,
      "step": 172154
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.281229019165039,
      "learning_rate": 8.974834998775743e-05,
      "loss": 2.0294,
      "step": 172155
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.2055999040603638,
      "learning_rate": 8.974543212890689e-05,
      "loss": 2.0481,
      "step": 172156
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.3201093673706055,
      "learning_rate": 8.974251430914666e-05,
      "loss": 2.0387,
      "step": 172157
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.2296185493469238,
      "learning_rate": 8.973959652847717e-05,
      "loss": 2.0906,
      "step": 172158
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.338372826576233,
      "learning_rate": 8.973667878689916e-05,
      "loss": 1.9807,
      "step": 172159
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.341720461845398,
      "learning_rate": 8.973376108441298e-05,
      "loss": 1.9502,
      "step": 172160
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.2140324115753174,
      "learning_rate": 8.973084342101933e-05,
      "loss": 2.0892,
      "step": 172161
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.3021044731140137,
      "learning_rate": 8.972792579671871e-05,
      "loss": 1.9447,
      "step": 172162
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.339745283126831,
      "learning_rate": 8.972500821151164e-05,
      "loss": 2.1996,
      "step": 172163
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.3285318613052368,
      "learning_rate": 8.972209066539853e-05,
      "loss": 1.8971,
      "step": 172164
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.3937137126922607,
      "learning_rate": 8.97191731583802e-05,
      "loss": 1.9254,
      "step": 172165
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.3548756837844849,
      "learning_rate": 8.971625569045692e-05,
      "loss": 1.9111,
      "step": 172166
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.8031065464019775,
      "learning_rate": 8.971333826162947e-05,
      "loss": 1.9875,
      "step": 172167
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.3970742225646973,
      "learning_rate": 8.971042087189824e-05,
      "loss": 2.1892,
      "step": 172168
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.29190194606781,
      "learning_rate": 8.970750352126384e-05,
      "loss": 2.0183,
      "step": 172169
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.2782554626464844,
      "learning_rate": 8.97045862097267e-05,
      "loss": 1.8815,
      "step": 172170
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.4129061698913574,
      "learning_rate": 8.970166893728751e-05,
      "loss": 2.2958,
      "step": 172171
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.4193813800811768,
      "learning_rate": 8.969875170394668e-05,
      "loss": 2.2723,
      "step": 172172
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.366157054901123,
      "learning_rate": 8.969583450970491e-05,
      "loss": 2.0311,
      "step": 172173
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.1940817832946777,
      "learning_rate": 8.969291735456267e-05,
      "loss": 2.1322,
      "step": 172174
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.285629153251648,
      "learning_rate": 8.969000023852044e-05,
      "loss": 1.8341,
      "step": 172175
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.3248604536056519,
      "learning_rate": 8.968708316157878e-05,
      "loss": 2.0217,
      "step": 172176
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.3389945030212402,
      "learning_rate": 8.96841661237383e-05,
      "loss": 2.0503,
      "step": 172177
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.2714476585388184,
      "learning_rate": 8.968124912499945e-05,
      "loss": 1.9094,
      "step": 172178
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.4126172065734863,
      "learning_rate": 8.967833216536287e-05,
      "loss": 2.2512,
      "step": 172179
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.2828850746154785,
      "learning_rate": 8.967541524482909e-05,
      "loss": 2.1018,
      "step": 172180
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.2053427696228027,
      "learning_rate": 8.967249836339864e-05,
      "loss": 1.8727,
      "step": 172181
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.3251874446868896,
      "learning_rate": 8.966958152107194e-05,
      "loss": 2.0622,
      "step": 172182
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.3437691926956177,
      "learning_rate": 8.966666471784971e-05,
      "loss": 2.0057,
      "step": 172183
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.23297917842865,
      "learning_rate": 8.966374795373231e-05,
      "loss": 1.8869,
      "step": 172184
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.138192057609558,
      "learning_rate": 8.966083122872053e-05,
      "loss": 1.8637,
      "step": 172185
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.2142014503479004,
      "learning_rate": 8.965791454281474e-05,
      "loss": 2.1138,
      "step": 172186
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.2802095413208008,
      "learning_rate": 8.965499789601553e-05,
      "loss": 1.9757,
      "step": 172187
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.3418058156967163,
      "learning_rate": 8.96520812883233e-05,
      "loss": 2.0703,
      "step": 172188
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.2172753810882568,
      "learning_rate": 8.964916471973885e-05,
      "loss": 2.0346,
      "step": 172189
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.3042349815368652,
      "learning_rate": 8.964624819026247e-05,
      "loss": 2.1679,
      "step": 172190
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.393075942993164,
      "learning_rate": 8.964333169989491e-05,
      "loss": 1.9554,
      "step": 172191
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.2907594442367554,
      "learning_rate": 8.964041524863665e-05,
      "loss": 2.0966,
      "step": 172192
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.2745229005813599,
      "learning_rate": 8.963749883648821e-05,
      "loss": 1.9662,
      "step": 172193
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.2442386150360107,
      "learning_rate": 8.963458246345005e-05,
      "loss": 2.4007,
      "step": 172194
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.3213920593261719,
      "learning_rate": 8.963166612952286e-05,
      "loss": 2.2198,
      "step": 172195
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.2664626836776733,
      "learning_rate": 8.962874983470702e-05,
      "loss": 2.1516,
      "step": 172196
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.4096953868865967,
      "learning_rate": 8.962583357900326e-05,
      "loss": 2.0465,
      "step": 172197
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.2216540575027466,
      "learning_rate": 8.962291736241203e-05,
      "loss": 2.0435,
      "step": 172198
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.4216105937957764,
      "learning_rate": 8.96200011849338e-05,
      "loss": 2.1275,
      "step": 172199
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.491271734237671,
      "learning_rate": 8.961708504656925e-05,
      "loss": 2.1879,
      "step": 172200
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.3439277410507202,
      "learning_rate": 8.961416894731885e-05,
      "loss": 2.1568,
      "step": 172201
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.2756855487823486,
      "learning_rate": 8.961125288718308e-05,
      "loss": 2.0837,
      "step": 172202
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.2927647829055786,
      "learning_rate": 8.960833686616264e-05,
      "loss": 2.1061,
      "step": 172203
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.3181884288787842,
      "learning_rate": 8.960542088425797e-05,
      "loss": 2.1101,
      "step": 172204
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.1519901752471924,
      "learning_rate": 8.960250494146954e-05,
      "loss": 1.8394,
      "step": 172205
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.2853624820709229,
      "learning_rate": 8.959958903779805e-05,
      "loss": 2.0415,
      "step": 172206
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.2078968286514282,
      "learning_rate": 8.959667317324387e-05,
      "loss": 2.0309,
      "step": 172207
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.1687039136886597,
      "learning_rate": 8.959375734780774e-05,
      "loss": 1.9738,
      "step": 172208
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.3000314235687256,
      "learning_rate": 8.95908415614901e-05,
      "loss": 2.1119,
      "step": 172209
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.2641799449920654,
      "learning_rate": 8.958792581429153e-05,
      "loss": 2.117,
      "step": 172210
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.2233260869979858,
      "learning_rate": 8.958501010621242e-05,
      "loss": 2.0987,
      "step": 172211
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.3566045761108398,
      "learning_rate": 8.958209443725352e-05,
      "loss": 2.07,
      "step": 172212
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.2884732484817505,
      "learning_rate": 8.957917880741518e-05,
      "loss": 2.2015,
      "step": 172213
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.2608294486999512,
      "learning_rate": 8.957626321669816e-05,
      "loss": 1.993,
      "step": 172214
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.1548892259597778,
      "learning_rate": 8.957334766510285e-05,
      "loss": 2.0383,
      "step": 172215
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.268630027770996,
      "learning_rate": 8.957043215262987e-05,
      "loss": 1.885,
      "step": 172216
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.366368055343628,
      "learning_rate": 8.956751667927959e-05,
      "loss": 2.1019,
      "step": 172217
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.3061336278915405,
      "learning_rate": 8.95646012450528e-05,
      "loss": 1.7376,
      "step": 172218
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.2516121864318848,
      "learning_rate": 8.95616858499498e-05,
      "loss": 1.9754,
      "step": 172219
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.228712797164917,
      "learning_rate": 8.955877049397137e-05,
      "loss": 2.021,
      "step": 172220
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.2170238494873047,
      "learning_rate": 8.95558551771178e-05,
      "loss": 2.2291,
      "step": 172221
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.3319857120513916,
      "learning_rate": 8.955293989938993e-05,
      "loss": 2.0901,
      "step": 172222
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.2250266075134277,
      "learning_rate": 8.95500246607881e-05,
      "loss": 2.068,
      "step": 172223
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.256480097770691,
      "learning_rate": 8.954710946131291e-05,
      "loss": 2.0939,
      "step": 172224
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.2449620962142944,
      "learning_rate": 8.954419430096477e-05,
      "loss": 1.9408,
      "step": 172225
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.2976982593536377,
      "learning_rate": 8.954127917974446e-05,
      "loss": 2.0501,
      "step": 172226
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.24492347240448,
      "learning_rate": 8.953836409765226e-05,
      "loss": 2.0003,
      "step": 172227
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.2428056001663208,
      "learning_rate": 8.953544905468898e-05,
      "loss": 2.1245,
      "step": 172228
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.2647007703781128,
      "learning_rate": 8.953253405085505e-05,
      "loss": 1.8724,
      "step": 172229
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.2711691856384277,
      "learning_rate": 8.952961908615096e-05,
      "loss": 1.9962,
      "step": 172230
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.2718108892440796,
      "learning_rate": 8.952670416057722e-05,
      "loss": 1.9851,
      "step": 172231
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.3127856254577637,
      "learning_rate": 8.952378927413452e-05,
      "loss": 2.0331,
      "step": 172232
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.2851507663726807,
      "learning_rate": 8.952087442682321e-05,
      "loss": 2.0535,
      "step": 172233
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.2732733488082886,
      "learning_rate": 8.951795961864406e-05,
      "loss": 1.7549,
      "step": 172234
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.2916909456253052,
      "learning_rate": 8.951504484959749e-05,
      "loss": 2.1584,
      "step": 172235
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.2791844606399536,
      "learning_rate": 8.951213011968404e-05,
      "loss": 1.9159,
      "step": 172236
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.2715402841567993,
      "learning_rate": 8.950921542890417e-05,
      "loss": 2.1168,
      "step": 172237
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.2339085340499878,
      "learning_rate": 8.95063007772586e-05,
      "loss": 1.9782,
      "step": 172238
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.2599996328353882,
      "learning_rate": 8.95033861647477e-05,
      "loss": 1.9559,
      "step": 172239
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.408225178718567,
      "learning_rate": 8.950047159137215e-05,
      "loss": 1.9753,
      "step": 172240
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.3131417036056519,
      "learning_rate": 8.949755705713246e-05,
      "loss": 1.9901,
      "step": 172241
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.3011598587036133,
      "learning_rate": 8.949464256202917e-05,
      "loss": 2.1492,
      "step": 172242
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.2442350387573242,
      "learning_rate": 8.949172810606268e-05,
      "loss": 2.1539,
      "step": 172243
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.4309738874435425,
      "learning_rate": 8.948881368923374e-05,
      "loss": 1.9851,
      "step": 172244
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.4225342273712158,
      "learning_rate": 8.948589931154273e-05,
      "loss": 1.9348,
      "step": 172245
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.1784199476242065,
      "learning_rate": 8.948298497299034e-05,
      "loss": 1.8217,
      "step": 172246
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.4490290880203247,
      "learning_rate": 8.948007067357705e-05,
      "loss": 1.9865,
      "step": 172247
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.279723048210144,
      "learning_rate": 8.947715641330338e-05,
      "loss": 2.0881,
      "step": 172248
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.4190939664840698,
      "learning_rate": 8.947424219216978e-05,
      "loss": 2.0152,
      "step": 172249
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.380387783050537,
      "learning_rate": 8.947132801017703e-05,
      "loss": 2.1066,
      "step": 172250
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.4350264072418213,
      "learning_rate": 8.94684138673254e-05,
      "loss": 2.1164,
      "step": 172251
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.348031997680664,
      "learning_rate": 8.946549976361566e-05,
      "loss": 2.1948,
      "step": 172252
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.4946402311325073,
      "learning_rate": 8.946258569904827e-05,
      "loss": 1.9488,
      "step": 172253
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.177885890007019,
      "learning_rate": 8.945967167362374e-05,
      "loss": 2.2517,
      "step": 172254
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.1868643760681152,
      "learning_rate": 8.945675768734257e-05,
      "loss": 2.0241,
      "step": 172255
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.2436507940292358,
      "learning_rate": 8.945384374020543e-05,
      "loss": 1.9555,
      "step": 172256
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.1937835216522217,
      "learning_rate": 8.94509298322127e-05,
      "loss": 1.8593,
      "step": 172257
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.233198881149292,
      "learning_rate": 8.944801596336511e-05,
      "loss": 2.0188,
      "step": 172258
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.3572216033935547,
      "learning_rate": 8.944510213366312e-05,
      "loss": 2.0838,
      "step": 172259
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.3509169816970825,
      "learning_rate": 8.944218834310729e-05,
      "loss": 1.9488,
      "step": 172260
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.4357990026474,
      "learning_rate": 8.943927459169801e-05,
      "loss": 2.0616,
      "step": 172261
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.3529924154281616,
      "learning_rate": 8.943636087943601e-05,
      "loss": 2.1815,
      "step": 172262
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.359140396118164,
      "learning_rate": 8.94334472063217e-05,
      "loss": 1.9785,
      "step": 172263
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.3388690948486328,
      "learning_rate": 8.943053357235576e-05,
      "loss": 1.887,
      "step": 172264
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.3369442224502563,
      "learning_rate": 8.942761997753869e-05,
      "loss": 1.9139,
      "step": 172265
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.260977029800415,
      "learning_rate": 8.9424706421871e-05,
      "loss": 2.0184,
      "step": 172266
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.2260044813156128,
      "learning_rate": 8.942179290535312e-05,
      "loss": 1.9156,
      "step": 172267
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.1700078248977661,
      "learning_rate": 8.941887942798579e-05,
      "loss": 1.896,
      "step": 172268
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.2980965375900269,
      "learning_rate": 8.941596598976937e-05,
      "loss": 2.0227,
      "step": 172269
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.3845901489257812,
      "learning_rate": 8.941305259070461e-05,
      "loss": 1.9562,
      "step": 172270
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.258919358253479,
      "learning_rate": 8.941013923079193e-05,
      "loss": 1.8748,
      "step": 172271
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.3257840871810913,
      "learning_rate": 8.94072259100319e-05,
      "loss": 2.1038,
      "step": 172272
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.3488421440124512,
      "learning_rate": 8.940431262842491e-05,
      "loss": 2.0108,
      "step": 172273
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.3096386194229126,
      "learning_rate": 8.940139938597177e-05,
      "loss": 2.2023,
      "step": 172274
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.2865525484085083,
      "learning_rate": 8.939848618267275e-05,
      "loss": 2.1967,
      "step": 172275
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.2012466192245483,
      "learning_rate": 8.939557301852863e-05,
      "loss": 1.9244,
      "step": 172276
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.2143534421920776,
      "learning_rate": 8.939265989353986e-05,
      "loss": 1.9351,
      "step": 172277
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.2889797687530518,
      "learning_rate": 8.938974680770698e-05,
      "loss": 1.9085,
      "step": 172278
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.2424041032791138,
      "learning_rate": 8.938683376103043e-05,
      "loss": 2.1434,
      "step": 172279
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.2591818571090698,
      "learning_rate": 8.938392075351093e-05,
      "loss": 1.9407,
      "step": 172280
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.2170860767364502,
      "learning_rate": 8.93810077851488e-05,
      "loss": 2.0658,
      "step": 172281
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.2840666770935059,
      "learning_rate": 8.937809485594486e-05,
      "loss": 2.1779,
      "step": 172282
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.2236168384552002,
      "learning_rate": 8.937518196589949e-05,
      "loss": 1.9705,
      "step": 172283
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.3193796873092651,
      "learning_rate": 8.937226911501316e-05,
      "loss": 2.0877,
      "step": 172284
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.4674485921859741,
      "learning_rate": 8.936935630328662e-05,
      "loss": 2.1002,
      "step": 172285
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.464288353919983,
      "learning_rate": 8.936644353072023e-05,
      "loss": 2.2509,
      "step": 172286
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.2957926988601685,
      "learning_rate": 8.936353079731453e-05,
      "loss": 1.7832,
      "step": 172287
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.1868921518325806,
      "learning_rate": 8.936061810307023e-05,
      "loss": 1.9813,
      "step": 172288
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.292100429534912,
      "learning_rate": 8.935770544798768e-05,
      "loss": 2.0392,
      "step": 172289
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.364630103111267,
      "learning_rate": 8.93547928320676e-05,
      "loss": 2.1477,
      "step": 172290
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.2971031665802002,
      "learning_rate": 8.935188025531042e-05,
      "loss": 2.0738,
      "step": 172291
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.3876497745513916,
      "learning_rate": 8.934896771771672e-05,
      "loss": 2.2115,
      "step": 172292
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.312506914138794,
      "learning_rate": 8.934605521928693e-05,
      "loss": 1.9445,
      "step": 172293
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.3164339065551758,
      "learning_rate": 8.934314276002177e-05,
      "loss": 2.1173,
      "step": 172294
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.4818211793899536,
      "learning_rate": 8.934023033992158e-05,
      "loss": 2.2492,
      "step": 172295
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.786495327949524,
      "learning_rate": 8.933731795898716e-05,
      "loss": 1.9589,
      "step": 172296
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.4279764890670776,
      "learning_rate": 8.933440561721889e-05,
      "loss": 1.9945,
      "step": 172297
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.3818774223327637,
      "learning_rate": 8.933149331461721e-05,
      "loss": 2.0862,
      "step": 172298
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.4106025695800781,
      "learning_rate": 8.93285810511829e-05,
      "loss": 2.2228,
      "step": 172299
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.5299534797668457,
      "learning_rate": 8.932566882691639e-05,
      "loss": 2.0125,
      "step": 172300
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.342786431312561,
      "learning_rate": 8.93227566418181e-05,
      "loss": 1.9561,
      "step": 172301
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.7274514436721802,
      "learning_rate": 8.931984449588879e-05,
      "loss": 2.1052,
      "step": 172302
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.2661864757537842,
      "learning_rate": 8.931693238912889e-05,
      "loss": 2.0173,
      "step": 172303
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.2265714406967163,
      "learning_rate": 8.931402032153889e-05,
      "loss": 2.0285,
      "step": 172304
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.5722382068634033,
      "learning_rate": 8.931110829311947e-05,
      "loss": 2.0697,
      "step": 172305
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.4177266359329224,
      "learning_rate": 8.930819630387096e-05,
      "loss": 1.7712,
      "step": 172306
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.2716729640960693,
      "learning_rate": 8.930528435379416e-05,
      "loss": 2.1319,
      "step": 172307
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.1681550741195679,
      "learning_rate": 8.93023724428895e-05,
      "loss": 1.804,
      "step": 172308
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.3670932054519653,
      "learning_rate": 8.929946057115745e-05,
      "loss": 2.0139,
      "step": 172309
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.3583146333694458,
      "learning_rate": 8.929654873859858e-05,
      "loss": 2.0667,
      "step": 172310
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.3473398685455322,
      "learning_rate": 8.92936369452135e-05,
      "loss": 1.9205,
      "step": 172311
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.3197041749954224,
      "learning_rate": 8.929072519100266e-05,
      "loss": 2.0691,
      "step": 172312
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.3222389221191406,
      "learning_rate": 8.928781347596672e-05,
      "loss": 1.9189,
      "step": 172313
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.4337605237960815,
      "learning_rate": 8.928490180010617e-05,
      "loss": 1.9002,
      "step": 172314
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.52700674533844,
      "learning_rate": 8.928199016342151e-05,
      "loss": 2.2453,
      "step": 172315
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.2076627016067505,
      "learning_rate": 8.927907856591323e-05,
      "loss": 1.9716,
      "step": 172316
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.2151687145233154,
      "learning_rate": 8.927616700758202e-05,
      "loss": 1.8184,
      "step": 172317
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.3093533515930176,
      "learning_rate": 8.927325548842825e-05,
      "loss": 1.9532,
      "step": 172318
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.5828551054000854,
      "learning_rate": 8.927034400845268e-05,
      "loss": 2.0875,
      "step": 172319
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.2535347938537598,
      "learning_rate": 8.926743256765572e-05,
      "loss": 1.9409,
      "step": 172320
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.3188037872314453,
      "learning_rate": 8.926452116603794e-05,
      "loss": 1.8463,
      "step": 172321
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.3867183923721313,
      "learning_rate": 8.926160980359974e-05,
      "loss": 1.9561,
      "step": 172322
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.2861475944519043,
      "learning_rate": 8.92586984803419e-05,
      "loss": 1.9669,
      "step": 172323
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.31710946559906,
      "learning_rate": 8.925578719626476e-05,
      "loss": 1.8096,
      "step": 172324
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.2311203479766846,
      "learning_rate": 8.925287595136901e-05,
      "loss": 2.0436,
      "step": 172325
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.2980093955993652,
      "learning_rate": 8.924996474565515e-05,
      "loss": 1.8649,
      "step": 172326
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.2121094465255737,
      "learning_rate": 8.92470535791237e-05,
      "loss": 1.994,
      "step": 172327
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.3723622560501099,
      "learning_rate": 8.92441424517751e-05,
      "loss": 1.9602,
      "step": 172328
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.2999567985534668,
      "learning_rate": 8.92412313636101e-05,
      "loss": 1.9882,
      "step": 172329
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.2567332983016968,
      "learning_rate": 8.923832031462903e-05,
      "loss": 2.0021,
      "step": 172330
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.2238250970840454,
      "learning_rate": 8.923540930483262e-05,
      "loss": 1.9768,
      "step": 172331
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.2174479961395264,
      "learning_rate": 8.923249833422133e-05,
      "loss": 2.0106,
      "step": 172332
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.4050410985946655,
      "learning_rate": 8.922958740279568e-05,
      "loss": 1.8345,
      "step": 172333
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.5986703634262085,
      "learning_rate": 8.922667651055617e-05,
      "loss": 2.1054,
      "step": 172334
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.3341342210769653,
      "learning_rate": 8.922376565750347e-05,
      "loss": 2.1595,
      "step": 172335
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.2531765699386597,
      "learning_rate": 8.922085484363796e-05,
      "loss": 1.9206,
      "step": 172336
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.2632951736450195,
      "learning_rate": 8.921794406896037e-05,
      "loss": 2.1761,
      "step": 172337
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.4668855667114258,
      "learning_rate": 8.921503333347115e-05,
      "loss": 2.1998,
      "step": 172338
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.3051053285598755,
      "learning_rate": 8.921212263717083e-05,
      "loss": 2.2612,
      "step": 172339
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.3069000244140625,
      "learning_rate": 8.920921198005985e-05,
      "loss": 1.9075,
      "step": 172340
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.8930079936981201,
      "learning_rate": 8.920630136213898e-05,
      "loss": 2.0737,
      "step": 172341
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.3636555671691895,
      "learning_rate": 8.920339078340851e-05,
      "loss": 2.299,
      "step": 172342
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.341304898262024,
      "learning_rate": 8.92004802438692e-05,
      "loss": 2.1378,
      "step": 172343
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.2818981409072876,
      "learning_rate": 8.919756974352153e-05,
      "loss": 1.854,
      "step": 172344
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.3287509679794312,
      "learning_rate": 8.919465928236599e-05,
      "loss": 2.094,
      "step": 172345
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.364000916481018,
      "learning_rate": 8.919174886040306e-05,
      "loss": 2.0053,
      "step": 172346
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.3106963634490967,
      "learning_rate": 8.918883847763344e-05,
      "loss": 2.1435,
      "step": 172347
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.4775687456130981,
      "learning_rate": 8.918592813405753e-05,
      "loss": 1.9819,
      "step": 172348
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.3704171180725098,
      "learning_rate": 8.918301782967601e-05,
      "loss": 2.0097,
      "step": 172349
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.2430751323699951,
      "learning_rate": 8.918010756448935e-05,
      "loss": 2.0003,
      "step": 172350
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.4940582513809204,
      "learning_rate": 8.917719733849809e-05,
      "loss": 2.1546,
      "step": 172351
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.2934892177581787,
      "learning_rate": 8.917428715170264e-05,
      "loss": 2.2581,
      "step": 172352
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.225834846496582,
      "learning_rate": 8.91713770041038e-05,
      "loss": 1.9022,
      "step": 172353
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.340743899345398,
      "learning_rate": 8.916846689570188e-05,
      "loss": 2.129,
      "step": 172354
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.3344911336898804,
      "learning_rate": 8.916555682649761e-05,
      "loss": 2.1662,
      "step": 172355
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.2982858419418335,
      "learning_rate": 8.916264679649138e-05,
      "loss": 1.9895,
      "step": 172356
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.2589964866638184,
      "learning_rate": 8.915973680568396e-05,
      "loss": 2.0147,
      "step": 172357
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.2625049352645874,
      "learning_rate": 8.915682685407551e-05,
      "loss": 2.1762,
      "step": 172358
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.2297943830490112,
      "learning_rate": 8.915391694166691e-05,
      "loss": 2.0231,
      "step": 172359
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.3231416940689087,
      "learning_rate": 8.915100706845849e-05,
      "loss": 2.0084,
      "step": 172360
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.274558186531067,
      "learning_rate": 8.914809723445094e-05,
      "loss": 1.8807,
      "step": 172361
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.187751293182373,
      "learning_rate": 8.914518743964467e-05,
      "loss": 2.1133,
      "step": 172362
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.2047185897827148,
      "learning_rate": 8.914227768404047e-05,
      "loss": 1.8795,
      "step": 172363
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.3472089767456055,
      "learning_rate": 8.913936796763853e-05,
      "loss": 2.1356,
      "step": 172364
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.245953917503357,
      "learning_rate": 8.913645829043961e-05,
      "loss": 1.7774,
      "step": 172365
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.20681893825531,
      "learning_rate": 8.913354865244413e-05,
      "loss": 1.9059,
      "step": 172366
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.5299224853515625,
      "learning_rate": 8.913063905365283e-05,
      "loss": 1.8854,
      "step": 172367
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.5826205015182495,
      "learning_rate": 8.9127729494066e-05,
      "loss": 2.1522,
      "step": 172368
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.2116832733154297,
      "learning_rate": 8.912481997368442e-05,
      "loss": 2.0087,
      "step": 172369
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.420992374420166,
      "learning_rate": 8.91219104925085e-05,
      "loss": 2.0519,
      "step": 172370
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.3822202682495117,
      "learning_rate": 8.911900105053879e-05,
      "loss": 2.0395,
      "step": 172371
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.3268312215805054,
      "learning_rate": 8.911609164777575e-05,
      "loss": 1.8894,
      "step": 172372
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.2204456329345703,
      "learning_rate": 8.91131822842201e-05,
      "loss": 2.0738,
      "step": 172373
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.2872705459594727,
      "learning_rate": 8.911027295987221e-05,
      "loss": 1.8622,
      "step": 172374
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.3431565761566162,
      "learning_rate": 8.910736367473278e-05,
      "loss": 2.007,
      "step": 172375
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.3963321447372437,
      "learning_rate": 8.910445442880228e-05,
      "loss": 1.9817,
      "step": 172376
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.2551207542419434,
      "learning_rate": 8.910154522208125e-05,
      "loss": 1.972,
      "step": 172377
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.2128559350967407,
      "learning_rate": 8.909863605457014e-05,
      "loss": 2.113,
      "step": 172378
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.2337411642074585,
      "learning_rate": 8.909572692626965e-05,
      "loss": 1.9018,
      "step": 172379
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.2665009498596191,
      "learning_rate": 8.909281783718013e-05,
      "loss": 2.142,
      "step": 172380
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.343689203262329,
      "learning_rate": 8.908990878730238e-05,
      "loss": 1.9851,
      "step": 172381
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.3312726020812988,
      "learning_rate": 8.908699977663676e-05,
      "loss": 2.1386,
      "step": 172382
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.42117178440094,
      "learning_rate": 8.908409080518377e-05,
      "loss": 1.9994,
      "step": 172383
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.2749778032302856,
      "learning_rate": 8.908118187294412e-05,
      "loss": 1.8748,
      "step": 172384
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.251960039138794,
      "learning_rate": 8.907827297991827e-05,
      "loss": 1.9764,
      "step": 172385
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.4132438898086548,
      "learning_rate": 8.907536412610661e-05,
      "loss": 2.1639,
      "step": 172386
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.2520101070404053,
      "learning_rate": 8.907245531150995e-05,
      "loss": 2.2401,
      "step": 172387
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.4387825727462769,
      "learning_rate": 8.906954653612872e-05,
      "loss": 2.0668,
      "step": 172388
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.2277737855911255,
      "learning_rate": 8.906663779996332e-05,
      "loss": 1.9006,
      "step": 172389
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.2821489572525024,
      "learning_rate": 8.906372910301452e-05,
      "loss": 2.0358,
      "step": 172390
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.287822961807251,
      "learning_rate": 8.906082044528265e-05,
      "loss": 2.1501,
      "step": 172391
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.3961235284805298,
      "learning_rate": 8.905791182676848e-05,
      "loss": 2.0168,
      "step": 172392
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.2119848728179932,
      "learning_rate": 8.905500324747241e-05,
      "loss": 2.1452,
      "step": 172393
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.272704839706421,
      "learning_rate": 8.905209470739499e-05,
      "loss": 1.9977,
      "step": 172394
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.30078125,
      "learning_rate": 8.904918620653666e-05,
      "loss": 2.0214,
      "step": 172395
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.3681890964508057,
      "learning_rate": 8.904627774489817e-05,
      "loss": 2.1057,
      "step": 172396
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.3884989023208618,
      "learning_rate": 8.904336932247986e-05,
      "loss": 1.8075,
      "step": 172397
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.2672052383422852,
      "learning_rate": 8.904046093928249e-05,
      "loss": 2.001,
      "step": 172398
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.4951688051223755,
      "learning_rate": 8.903755259530648e-05,
      "loss": 2.0865,
      "step": 172399
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.2373732328414917,
      "learning_rate": 8.903464429055235e-05,
      "loss": 2.0174,
      "step": 172400
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.3472697734832764,
      "learning_rate": 8.903173602502059e-05,
      "loss": 2.0875,
      "step": 172401
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.2853857278823853,
      "learning_rate": 8.902882779871188e-05,
      "loss": 1.9384,
      "step": 172402
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.3175253868103027,
      "learning_rate": 8.902591961162659e-05,
      "loss": 2.1669,
      "step": 172403
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.4682587385177612,
      "learning_rate": 8.90230114637655e-05,
      "loss": 2.0005,
      "step": 172404
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.547568678855896,
      "learning_rate": 8.9020103355129e-05,
      "loss": 2.1266,
      "step": 172405
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.358577847480774,
      "learning_rate": 8.901719528571763e-05,
      "loss": 2.2356,
      "step": 172406
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.333755373954773,
      "learning_rate": 8.901428725553187e-05,
      "loss": 2.0002,
      "step": 172407
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.3106907606124878,
      "learning_rate": 8.90113792645724e-05,
      "loss": 2.1239,
      "step": 172408
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.644195556640625,
      "learning_rate": 8.900847131283963e-05,
      "loss": 1.9655,
      "step": 172409
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.431223750114441,
      "learning_rate": 8.900556340033427e-05,
      "loss": 2.0725,
      "step": 172410
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.1607304811477661,
      "learning_rate": 8.900265552705674e-05,
      "loss": 2.1443,
      "step": 172411
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.2177977561950684,
      "learning_rate": 8.899974769300761e-05,
      "loss": 2.2503,
      "step": 172412
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.2806172370910645,
      "learning_rate": 8.89968398981873e-05,
      "loss": 2.3823,
      "step": 172413
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.3043173551559448,
      "learning_rate": 8.899393214259658e-05,
      "loss": 2.1573,
      "step": 172414
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.3366693258285522,
      "learning_rate": 8.899102442623576e-05,
      "loss": 2.0464,
      "step": 172415
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.3448063135147095,
      "learning_rate": 8.898811674910559e-05,
      "loss": 1.9154,
      "step": 172416
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.260377049446106,
      "learning_rate": 8.89852091112065e-05,
      "loss": 1.975,
      "step": 172417
    },
    {
      "epoch": 2.24,
      "grad_norm": 1.4104543924331665,
      "learning_rate": 8.898230151253904e-05,
      "loss": 1.9615,
      "step": 172418
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.3217087984085083,
      "learning_rate": 8.897939395310369e-05,
      "loss": 2.0682,
      "step": 172419
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.3023539781570435,
      "learning_rate": 8.897648643290112e-05,
      "loss": 1.7772,
      "step": 172420
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.449575424194336,
      "learning_rate": 8.897357895193173e-05,
      "loss": 2.048,
      "step": 172421
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.3062775135040283,
      "learning_rate": 8.89706715101962e-05,
      "loss": 2.1169,
      "step": 172422
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.2794907093048096,
      "learning_rate": 8.896776410769495e-05,
      "loss": 2.1728,
      "step": 172423
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.303709864616394,
      "learning_rate": 8.896485674442875e-05,
      "loss": 1.9193,
      "step": 172424
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.3342922925949097,
      "learning_rate": 8.896194942039776e-05,
      "loss": 2.3079,
      "step": 172425
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.2047319412231445,
      "learning_rate": 8.895904213560285e-05,
      "loss": 1.9619,
      "step": 172426
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.3405814170837402,
      "learning_rate": 8.895613489004433e-05,
      "loss": 2.1168,
      "step": 172427
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.3148767948150635,
      "learning_rate": 8.895322768372292e-05,
      "loss": 2.0247,
      "step": 172428
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.2511624097824097,
      "learning_rate": 8.895032051663902e-05,
      "loss": 1.8872,
      "step": 172429
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.2275341749191284,
      "learning_rate": 8.894741338879343e-05,
      "loss": 1.8518,
      "step": 172430
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.1785961389541626,
      "learning_rate": 8.894450630018629e-05,
      "loss": 2.1334,
      "step": 172431
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.267014503479004,
      "learning_rate": 8.894159925081846e-05,
      "loss": 2.2467,
      "step": 172432
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.2247004508972168,
      "learning_rate": 8.893869224069027e-05,
      "loss": 1.886,
      "step": 172433
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.1764713525772095,
      "learning_rate": 8.893578526980247e-05,
      "loss": 2.0125,
      "step": 172434
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.1035057306289673,
      "learning_rate": 8.893287833815542e-05,
      "loss": 2.1521,
      "step": 172435
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.2503052949905396,
      "learning_rate": 8.892997144574985e-05,
      "loss": 1.971,
      "step": 172436
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.4197171926498413,
      "learning_rate": 8.892706459258605e-05,
      "loss": 2.1252,
      "step": 172437
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.2818617820739746,
      "learning_rate": 8.892415777866479e-05,
      "loss": 2.3754,
      "step": 172438
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.2350144386291504,
      "learning_rate": 8.892125100398639e-05,
      "loss": 2.3694,
      "step": 172439
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.2801841497421265,
      "learning_rate": 8.89183442685516e-05,
      "loss": 1.885,
      "step": 172440
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.2441205978393555,
      "learning_rate": 8.89154375723608e-05,
      "loss": 2.4045,
      "step": 172441
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.2631278038024902,
      "learning_rate": 8.89125309154148e-05,
      "loss": 2.0739,
      "step": 172442
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.342136263847351,
      "learning_rate": 8.890962429771374e-05,
      "loss": 1.9618,
      "step": 172443
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.2407708168029785,
      "learning_rate": 8.890671771925849e-05,
      "loss": 2.1881,
      "step": 172444
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.4977484941482544,
      "learning_rate": 8.890381118004935e-05,
      "loss": 2.185,
      "step": 172445
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.3455487489700317,
      "learning_rate": 8.890090468008708e-05,
      "loss": 1.9361,
      "step": 172446
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.3643449544906616,
      "learning_rate": 8.8897998219372e-05,
      "loss": 1.8613,
      "step": 172447
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.2470990419387817,
      "learning_rate": 8.889509179790497e-05,
      "loss": 1.8798,
      "step": 172448
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.551527738571167,
      "learning_rate": 8.889218541568614e-05,
      "loss": 1.96,
      "step": 172449
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.378371238708496,
      "learning_rate": 8.888927907271632e-05,
      "loss": 2.1204,
      "step": 172450
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.33206045627594,
      "learning_rate": 8.888637276899588e-05,
      "loss": 2.1863,
      "step": 172451
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.3257899284362793,
      "learning_rate": 8.888346650452556e-05,
      "loss": 2.0933,
      "step": 172452
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.2787463665008545,
      "learning_rate": 8.888056027930567e-05,
      "loss": 1.9708,
      "step": 172453
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.353454828262329,
      "learning_rate": 8.887765409333707e-05,
      "loss": 1.9624,
      "step": 172454
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.3056899309158325,
      "learning_rate": 8.88747479466199e-05,
      "loss": 2.1489,
      "step": 172455
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.2953591346740723,
      "learning_rate": 8.887184183915498e-05,
      "loss": 2.2559,
      "step": 172456
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.2233855724334717,
      "learning_rate": 8.886893577094271e-05,
      "loss": 2.0347,
      "step": 172457
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.3333008289337158,
      "learning_rate": 8.886602974198378e-05,
      "loss": 1.9716,
      "step": 172458
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.38079035282135,
      "learning_rate": 8.886312375227855e-05,
      "loss": 2.0384,
      "step": 172459
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.3404150009155273,
      "learning_rate": 8.886021780182773e-05,
      "loss": 2.0041,
      "step": 172460
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.3663803339004517,
      "learning_rate": 8.885731189063176e-05,
      "loss": 1.7011,
      "step": 172461
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.2846606969833374,
      "learning_rate": 8.885440601869126e-05,
      "loss": 2.0586,
      "step": 172462
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.3376291990280151,
      "learning_rate": 8.885150018600658e-05,
      "loss": 1.9534,
      "step": 172463
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.37196683883667,
      "learning_rate": 8.884859439257848e-05,
      "loss": 2.101,
      "step": 172464
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.241371989250183,
      "learning_rate": 8.884568863840733e-05,
      "loss": 2.1906,
      "step": 172465
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.2354583740234375,
      "learning_rate": 8.884278292349385e-05,
      "loss": 1.9215,
      "step": 172466
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.2150604724884033,
      "learning_rate": 8.883987724783847e-05,
      "loss": 1.9577,
      "step": 172467
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.3574851751327515,
      "learning_rate": 8.883697161144164e-05,
      "loss": 1.8249,
      "step": 172468
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.3820703029632568,
      "learning_rate": 8.883406601430411e-05,
      "loss": 2.0336,
      "step": 172469
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.3231862783432007,
      "learning_rate": 8.883116045642632e-05,
      "loss": 2.1924,
      "step": 172470
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.1700738668441772,
      "learning_rate": 8.882825493780872e-05,
      "loss": 1.9541,
      "step": 172471
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.1590838432312012,
      "learning_rate": 8.882534945845201e-05,
      "loss": 2.1119,
      "step": 172472
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.2182672023773193,
      "learning_rate": 8.882244401835667e-05,
      "loss": 1.9404,
      "step": 172473
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.480060338973999,
      "learning_rate": 8.88195386175231e-05,
      "loss": 1.931,
      "step": 172474
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.280090093612671,
      "learning_rate": 8.881663325595208e-05,
      "loss": 1.8051,
      "step": 172475
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.2116056680679321,
      "learning_rate": 8.881372793364402e-05,
      "loss": 1.9309,
      "step": 172476
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.2108608484268188,
      "learning_rate": 8.881082265059937e-05,
      "loss": 1.9925,
      "step": 172477
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.219524621963501,
      "learning_rate": 8.88079174068189e-05,
      "loss": 1.9502,
      "step": 172478
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.224257230758667,
      "learning_rate": 8.880501220230301e-05,
      "loss": 2.1935,
      "step": 172479
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.1816353797912598,
      "learning_rate": 8.880210703705218e-05,
      "loss": 1.8984,
      "step": 172480
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.3632776737213135,
      "learning_rate": 8.879920191106709e-05,
      "loss": 2.0915,
      "step": 172481
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.290568232536316,
      "learning_rate": 8.879629682434815e-05,
      "loss": 2.1249,
      "step": 172482
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.3043131828308105,
      "learning_rate": 8.879339177689602e-05,
      "loss": 2.1312,
      "step": 172483
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.2384214401245117,
      "learning_rate": 8.879048676871122e-05,
      "loss": 1.9514,
      "step": 172484
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.3208321332931519,
      "learning_rate": 8.878758179979423e-05,
      "loss": 2.0396,
      "step": 172485
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.181761384010315,
      "learning_rate": 8.878467687014553e-05,
      "loss": 2.1708,
      "step": 172486
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.1966567039489746,
      "learning_rate": 8.878177197976585e-05,
      "loss": 2.1401,
      "step": 172487
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.2135707139968872,
      "learning_rate": 8.877886712865554e-05,
      "loss": 1.9789,
      "step": 172488
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.1222354173660278,
      "learning_rate": 8.877596231681531e-05,
      "loss": 1.9955,
      "step": 172489
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.3065074682235718,
      "learning_rate": 8.877305754424549e-05,
      "loss": 1.9037,
      "step": 172490
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.3798884153366089,
      "learning_rate": 8.877015281094697e-05,
      "loss": 1.9773,
      "step": 172491
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.2529784440994263,
      "learning_rate": 8.876724811691986e-05,
      "loss": 2.2368,
      "step": 172492
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.3129868507385254,
      "learning_rate": 8.8764343462165e-05,
      "loss": 2.0251,
      "step": 172493
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.3025586605072021,
      "learning_rate": 8.876143884668277e-05,
      "loss": 1.8191,
      "step": 172494
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.2354660034179688,
      "learning_rate": 8.875853427047386e-05,
      "loss": 2.0734,
      "step": 172495
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.314907193183899,
      "learning_rate": 8.875562973353863e-05,
      "loss": 2.037,
      "step": 172496
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.3338491916656494,
      "learning_rate": 8.875272523587787e-05,
      "loss": 2.0518,
      "step": 172497
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.2305033206939697,
      "learning_rate": 8.87498207774918e-05,
      "loss": 2.2067,
      "step": 172498
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.359574317932129,
      "learning_rate": 8.874691635838126e-05,
      "loss": 2.3147,
      "step": 172499
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.1890980005264282,
      "learning_rate": 8.874401197854652e-05,
      "loss": 1.8732,
      "step": 172500
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.231469750404358,
      "learning_rate": 8.874110763798834e-05,
      "loss": 2.075,
      "step": 172501
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.2934104204177856,
      "learning_rate": 8.873820333670711e-05,
      "loss": 1.9929,
      "step": 172502
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.2803072929382324,
      "learning_rate": 8.873529907470361e-05,
      "loss": 1.8139,
      "step": 172503
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.3012678623199463,
      "learning_rate": 8.873239485197801e-05,
      "loss": 1.9059,
      "step": 172504
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.257478952407837,
      "learning_rate": 8.872949066853114e-05,
      "loss": 1.9998,
      "step": 172505
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.329715371131897,
      "learning_rate": 8.872658652436337e-05,
      "loss": 2.2447,
      "step": 172506
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.25123929977417,
      "learning_rate": 8.872368241947539e-05,
      "loss": 2.0247,
      "step": 172507
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.2443077564239502,
      "learning_rate": 8.872077835386761e-05,
      "loss": 1.9112,
      "step": 172508
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.3099634647369385,
      "learning_rate": 8.871787432754076e-05,
      "loss": 2.0932,
      "step": 172509
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.410906434059143,
      "learning_rate": 8.871497034049509e-05,
      "loss": 1.9576,
      "step": 172510
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.285886287689209,
      "learning_rate": 8.87120663927314e-05,
      "loss": 2.1019,
      "step": 172511
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.3933215141296387,
      "learning_rate": 8.870916248425e-05,
      "loss": 2.1427,
      "step": 172512
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.5069825649261475,
      "learning_rate": 8.870625861505166e-05,
      "loss": 1.9077,
      "step": 172513
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.3233799934387207,
      "learning_rate": 8.870335478513674e-05,
      "loss": 1.9346,
      "step": 172514
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.1837341785430908,
      "learning_rate": 8.870045099450602e-05,
      "loss": 2.1474,
      "step": 172515
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.3755217790603638,
      "learning_rate": 8.869754724315968e-05,
      "loss": 2.2967,
      "step": 172516
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.2868822813034058,
      "learning_rate": 8.869464353109858e-05,
      "loss": 1.8911,
      "step": 172517
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.4553861618041992,
      "learning_rate": 8.869173985832303e-05,
      "loss": 2.045,
      "step": 172518
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.5256502628326416,
      "learning_rate": 8.868883622483375e-05,
      "loss": 1.9931,
      "step": 172519
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.2471386194229126,
      "learning_rate": 8.868593263063114e-05,
      "loss": 2.0,
      "step": 172520
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.298901081085205,
      "learning_rate": 8.868302907571597e-05,
      "loss": 2.2182,
      "step": 172521
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.3443903923034668,
      "learning_rate": 8.868012556008841e-05,
      "loss": 2.0447,
      "step": 172522
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.4238258600234985,
      "learning_rate": 8.867722208374932e-05,
      "loss": 2.1379,
      "step": 172523
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.3240915536880493,
      "learning_rate": 8.867431864669905e-05,
      "loss": 1.9246,
      "step": 172524
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.2405670881271362,
      "learning_rate": 8.867141524893825e-05,
      "loss": 1.8678,
      "step": 172525
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.1882373094558716,
      "learning_rate": 8.866851189046739e-05,
      "loss": 1.8427,
      "step": 172526
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.3034071922302246,
      "learning_rate": 8.866560857128719e-05,
      "loss": 2.1697,
      "step": 172527
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.3262640237808228,
      "learning_rate": 8.866270529139789e-05,
      "loss": 2.0112,
      "step": 172528
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.2617132663726807,
      "learning_rate": 8.865980205080022e-05,
      "loss": 2.0953,
      "step": 172529
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.1510242223739624,
      "learning_rate": 8.865689884949463e-05,
      "loss": 1.8902,
      "step": 172530
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.287213683128357,
      "learning_rate": 8.865399568748181e-05,
      "loss": 2.1039,
      "step": 172531
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.174656867980957,
      "learning_rate": 8.865109256476208e-05,
      "loss": 1.8027,
      "step": 172532
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.4512335062026978,
      "learning_rate": 8.864818948133632e-05,
      "loss": 2.1758,
      "step": 172533
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.2201690673828125,
      "learning_rate": 8.864528643720465e-05,
      "loss": 2.0087,
      "step": 172534
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.3850717544555664,
      "learning_rate": 8.86423834323679e-05,
      "loss": 1.9007,
      "step": 172535
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.3216981887817383,
      "learning_rate": 8.863948046682644e-05,
      "loss": 2.1144,
      "step": 172536
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.3121082782745361,
      "learning_rate": 8.863657754058096e-05,
      "loss": 1.9848,
      "step": 172537
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.2732375860214233,
      "learning_rate": 8.863367465363183e-05,
      "loss": 2.0693,
      "step": 172538
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.2583621740341187,
      "learning_rate": 8.863077180597992e-05,
      "loss": 2.1413,
      "step": 172539
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.3727917671203613,
      "learning_rate": 8.862786899762533e-05,
      "loss": 2.0612,
      "step": 172540
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.3022657632827759,
      "learning_rate": 8.86249662285689e-05,
      "loss": 1.9798,
      "step": 172541
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.3249695301055908,
      "learning_rate": 8.862206349881098e-05,
      "loss": 1.9398,
      "step": 172542
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.295208215713501,
      "learning_rate": 8.861916080835232e-05,
      "loss": 1.9009,
      "step": 172543
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.2980653047561646,
      "learning_rate": 8.861625815719325e-05,
      "loss": 2.0752,
      "step": 172544
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.3612186908721924,
      "learning_rate": 8.861335554533448e-05,
      "loss": 2.1315,
      "step": 172545
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.1712721586227417,
      "learning_rate": 8.861045297277651e-05,
      "loss": 2.0248,
      "step": 172546
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.2781593799591064,
      "learning_rate": 8.860755043951984e-05,
      "loss": 2.0382,
      "step": 172547
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.2309138774871826,
      "learning_rate": 8.860464794556489e-05,
      "loss": 1.9861,
      "step": 172548
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.2914202213287354,
      "learning_rate": 8.860174549091244e-05,
      "loss": 2.2953,
      "step": 172549
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.7253443002700806,
      "learning_rate": 8.859884307556285e-05,
      "loss": 2.0426,
      "step": 172550
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.4348456859588623,
      "learning_rate": 8.859594069951678e-05,
      "loss": 2.0116,
      "step": 172551
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.272103190422058,
      "learning_rate": 8.859303836277473e-05,
      "loss": 2.1522,
      "step": 172552
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.2840205430984497,
      "learning_rate": 8.859013606533713e-05,
      "loss": 2.1939,
      "step": 172553
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.3077647686004639,
      "learning_rate": 8.85872338072047e-05,
      "loss": 2.0773,
      "step": 172554
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.4918694496154785,
      "learning_rate": 8.858433158837791e-05,
      "loss": 2.066,
      "step": 172555
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.1511955261230469,
      "learning_rate": 8.858142940885717e-05,
      "loss": 1.8826,
      "step": 172556
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.2885551452636719,
      "learning_rate": 8.857852726864326e-05,
      "loss": 2.1583,
      "step": 172557
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.4170016050338745,
      "learning_rate": 8.857562516773657e-05,
      "loss": 2.1717,
      "step": 172558
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.2716894149780273,
      "learning_rate": 8.857272310613758e-05,
      "loss": 2.0305,
      "step": 172559
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.3375517129898071,
      "learning_rate": 8.8569821083847e-05,
      "loss": 2.1324,
      "step": 172560
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.3090708255767822,
      "learning_rate": 8.856691910086526e-05,
      "loss": 1.8445,
      "step": 172561
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.3985857963562012,
      "learning_rate": 8.856401715719286e-05,
      "loss": 2.1412,
      "step": 172562
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.2379602193832397,
      "learning_rate": 8.856111525283048e-05,
      "loss": 2.0012,
      "step": 172563
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.1929999589920044,
      "learning_rate": 8.855821338777856e-05,
      "loss": 1.9508,
      "step": 172564
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.5269094705581665,
      "learning_rate": 8.85553115620376e-05,
      "loss": 1.9016,
      "step": 172565
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.7051485776901245,
      "learning_rate": 8.855240977560826e-05,
      "loss": 1.9555,
      "step": 172566
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.206277847290039,
      "learning_rate": 8.854950802849096e-05,
      "loss": 2.2222,
      "step": 172567
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.208303451538086,
      "learning_rate": 8.854660632068637e-05,
      "loss": 1.8989,
      "step": 172568
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.2543612718582153,
      "learning_rate": 8.854370465219495e-05,
      "loss": 2.087,
      "step": 172569
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.5105791091918945,
      "learning_rate": 8.854080302301728e-05,
      "loss": 1.9971,
      "step": 172570
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.4304263591766357,
      "learning_rate": 8.853790143315373e-05,
      "loss": 2.0224,
      "step": 172571
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.1829689741134644,
      "learning_rate": 8.85349998826051e-05,
      "loss": 1.8324,
      "step": 172572
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.2604148387908936,
      "learning_rate": 8.85320983713717e-05,
      "loss": 2.0108,
      "step": 172573
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.2580567598342896,
      "learning_rate": 8.852919689945427e-05,
      "loss": 2.0008,
      "step": 172574
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.2182106971740723,
      "learning_rate": 8.852629546685325e-05,
      "loss": 2.0363,
      "step": 172575
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.456157922744751,
      "learning_rate": 8.852339407356919e-05,
      "loss": 2.0611,
      "step": 172576
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.2838430404663086,
      "learning_rate": 8.852049271960252e-05,
      "loss": 2.2629,
      "step": 172577
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.2885801792144775,
      "learning_rate": 8.851759140495399e-05,
      "loss": 1.9244,
      "step": 172578
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.4744828939437866,
      "learning_rate": 8.851469012962394e-05,
      "loss": 2.0509,
      "step": 172579
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.1289833784103394,
      "learning_rate": 8.851178889361308e-05,
      "loss": 2.0021,
      "step": 172580
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.3123875856399536,
      "learning_rate": 8.850888769692179e-05,
      "loss": 2.0629,
      "step": 172581
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.213559865951538,
      "learning_rate": 8.850598653955085e-05,
      "loss": 1.9616,
      "step": 172582
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.1850454807281494,
      "learning_rate": 8.85030854215005e-05,
      "loss": 1.8098,
      "step": 172583
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.2929809093475342,
      "learning_rate": 8.850018434277146e-05,
      "loss": 2.2487,
      "step": 172584
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.2330437898635864,
      "learning_rate": 8.849728330336416e-05,
      "loss": 1.8747,
      "step": 172585
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.3886075019836426,
      "learning_rate": 8.84943823032793e-05,
      "loss": 1.9703,
      "step": 172586
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.4060380458831787,
      "learning_rate": 8.849148134251723e-05,
      "loss": 1.9149,
      "step": 172587
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.2186360359191895,
      "learning_rate": 8.848858042107879e-05,
      "loss": 2.0695,
      "step": 172588
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.3114163875579834,
      "learning_rate": 8.848567953896411e-05,
      "loss": 1.8791,
      "step": 172589
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.3189818859100342,
      "learning_rate": 8.848277869617403e-05,
      "loss": 2.088,
      "step": 172590
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.3161749839782715,
      "learning_rate": 8.847987789270889e-05,
      "loss": 2.1571,
      "step": 172591
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.2458425760269165,
      "learning_rate": 8.847697712856947e-05,
      "loss": 1.7994,
      "step": 172592
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.383535623550415,
      "learning_rate": 8.847407640375602e-05,
      "loss": 2.0088,
      "step": 172593
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.289445161819458,
      "learning_rate": 8.847117571826948e-05,
      "loss": 2.1939,
      "step": 172594
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.7532984018325806,
      "learning_rate": 8.846827507210991e-05,
      "loss": 2.1413,
      "step": 172595
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.2792497873306274,
      "learning_rate": 8.846537446527819e-05,
      "loss": 2.0349,
      "step": 172596
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.329723834991455,
      "learning_rate": 8.846247389777464e-05,
      "loss": 1.931,
      "step": 172597
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.2570860385894775,
      "learning_rate": 8.845957336960004e-05,
      "loss": 2.0789,
      "step": 172598
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.4158374071121216,
      "learning_rate": 8.845667288075466e-05,
      "loss": 1.8104,
      "step": 172599
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.6510707139968872,
      "learning_rate": 8.84537724312394e-05,
      "loss": 2.2109,
      "step": 172600
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.267305612564087,
      "learning_rate": 8.845087202105437e-05,
      "loss": 2.2099,
      "step": 172601
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.280264139175415,
      "learning_rate": 8.844797165020042e-05,
      "loss": 1.8667,
      "step": 172602
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.416037917137146,
      "learning_rate": 8.844507131867788e-05,
      "loss": 2.078,
      "step": 172603
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.2745474576950073,
      "learning_rate": 8.844217102648749e-05,
      "loss": 2.0337,
      "step": 172604
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.313396692276001,
      "learning_rate": 8.843927077362962e-05,
      "loss": 2.2154,
      "step": 172605
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.2600171566009521,
      "learning_rate": 8.843637056010505e-05,
      "loss": 1.8606,
      "step": 172606
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.2996348142623901,
      "learning_rate": 8.843347038591399e-05,
      "loss": 1.9345,
      "step": 172607
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.330732822418213,
      "learning_rate": 8.843057025105722e-05,
      "loss": 2.1816,
      "step": 172608
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.2687982320785522,
      "learning_rate": 8.84276701555351e-05,
      "loss": 2.0448,
      "step": 172609
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.8859643936157227,
      "learning_rate": 8.842477009934836e-05,
      "loss": 1.9362,
      "step": 172610
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.2589240074157715,
      "learning_rate": 8.842187008249737e-05,
      "loss": 1.8759,
      "step": 172611
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.2186229228973389,
      "learning_rate": 8.841897010498291e-05,
      "loss": 1.7444,
      "step": 172612
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.3290807008743286,
      "learning_rate": 8.841607016680519e-05,
      "loss": 2.0741,
      "step": 172613
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.3828918933868408,
      "learning_rate": 8.841317026796503e-05,
      "loss": 2.0656,
      "step": 172614
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.3188869953155518,
      "learning_rate": 8.841027040846274e-05,
      "loss": 2.1049,
      "step": 172615
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.487208366394043,
      "learning_rate": 8.840737058829904e-05,
      "loss": 2.2278,
      "step": 172616
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.3360625505447388,
      "learning_rate": 8.840447080747435e-05,
      "loss": 2.2634,
      "step": 172617
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.3882157802581787,
      "learning_rate": 8.840157106598946e-05,
      "loss": 1.9786,
      "step": 172618
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.379312515258789,
      "learning_rate": 8.839867136384448e-05,
      "loss": 2.056,
      "step": 172619
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.2657136917114258,
      "learning_rate": 8.839577170104032e-05,
      "loss": 2.0375,
      "step": 172620
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.459878921508789,
      "learning_rate": 8.839287207757724e-05,
      "loss": 2.0244,
      "step": 172621
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.2413161993026733,
      "learning_rate": 8.838997249345604e-05,
      "loss": 2.2074,
      "step": 172622
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.340293526649475,
      "learning_rate": 8.838707294867707e-05,
      "loss": 1.901,
      "step": 172623
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.1738593578338623,
      "learning_rate": 8.838417344324098e-05,
      "loss": 1.9473,
      "step": 172624
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.3824326992034912,
      "learning_rate": 8.83812739771483e-05,
      "loss": 1.9766,
      "step": 172625
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.240765929222107,
      "learning_rate": 8.837837455039954e-05,
      "loss": 1.9093,
      "step": 172626
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.2810487747192383,
      "learning_rate": 8.837547516299514e-05,
      "loss": 1.9954,
      "step": 172627
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.323102593421936,
      "learning_rate": 8.83725758149358e-05,
      "loss": 2.0716,
      "step": 172628
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.2775605916976929,
      "learning_rate": 8.836967650622193e-05,
      "loss": 2.1319,
      "step": 172629
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.3005506992340088,
      "learning_rate": 8.83667772368542e-05,
      "loss": 2.0606,
      "step": 172630
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.3813860416412354,
      "learning_rate": 8.836387800683307e-05,
      "loss": 2.0578,
      "step": 172631
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.2411681413650513,
      "learning_rate": 8.836097881615915e-05,
      "loss": 2.0599,
      "step": 172632
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.4368640184402466,
      "learning_rate": 8.835807966483276e-05,
      "loss": 1.9361,
      "step": 172633
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.2692968845367432,
      "learning_rate": 8.835518055285472e-05,
      "loss": 1.8693,
      "step": 172634
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.4367283582687378,
      "learning_rate": 8.835228148022536e-05,
      "loss": 1.8787,
      "step": 172635
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.2073585987091064,
      "learning_rate": 8.83493824469454e-05,
      "loss": 1.9953,
      "step": 172636
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.271238088607788,
      "learning_rate": 8.834648345301528e-05,
      "loss": 2.0336,
      "step": 172637
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.2307804822921753,
      "learning_rate": 8.834358449843552e-05,
      "loss": 2.0818,
      "step": 172638
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.3830126523971558,
      "learning_rate": 8.834068558320663e-05,
      "loss": 1.9784,
      "step": 172639
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.386216163635254,
      "learning_rate": 8.833778670732928e-05,
      "loss": 2.0285,
      "step": 172640
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.2504271268844604,
      "learning_rate": 8.833488787080383e-05,
      "loss": 2.0029,
      "step": 172641
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.2755600214004517,
      "learning_rate": 8.833198907363103e-05,
      "loss": 1.9722,
      "step": 172642
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.2698079347610474,
      "learning_rate": 8.832909031581128e-05,
      "loss": 2.1014,
      "step": 172643
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.4029643535614014,
      "learning_rate": 8.832619159734505e-05,
      "loss": 2.087,
      "step": 172644
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.4856343269348145,
      "learning_rate": 8.832329291823311e-05,
      "loss": 2.1377,
      "step": 172645
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.2970004081726074,
      "learning_rate": 8.832039427847586e-05,
      "loss": 2.1009,
      "step": 172646
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.3759582042694092,
      "learning_rate": 8.831749567807373e-05,
      "loss": 2.0719,
      "step": 172647
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.2497221231460571,
      "learning_rate": 8.831459711702748e-05,
      "loss": 1.923,
      "step": 172648
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.3499624729156494,
      "learning_rate": 8.83116985953375e-05,
      "loss": 2.0054,
      "step": 172649
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.285923957824707,
      "learning_rate": 8.830880011300433e-05,
      "loss": 2.0249,
      "step": 172650
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.313137412071228,
      "learning_rate": 8.830590167002863e-05,
      "loss": 1.9342,
      "step": 172651
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.3043735027313232,
      "learning_rate": 8.830300326641075e-05,
      "loss": 1.6829,
      "step": 172652
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.324638843536377,
      "learning_rate": 8.830010490215146e-05,
      "loss": 2.155,
      "step": 172653
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.2935991287231445,
      "learning_rate": 8.829720657725114e-05,
      "loss": 1.9629,
      "step": 172654
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.24795663356781,
      "learning_rate": 8.829430829171042e-05,
      "loss": 2.1023,
      "step": 172655
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.2529984712600708,
      "learning_rate": 8.829141004552963e-05,
      "loss": 1.8976,
      "step": 172656
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.2851002216339111,
      "learning_rate": 8.828851183870959e-05,
      "loss": 2.1982,
      "step": 172657
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.1879823207855225,
      "learning_rate": 8.828561367125061e-05,
      "loss": 2.1175,
      "step": 172658
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.244714379310608,
      "learning_rate": 8.828271554315344e-05,
      "loss": 2.0599,
      "step": 172659
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.2280339002609253,
      "learning_rate": 8.827981745441847e-05,
      "loss": 1.8727,
      "step": 172660
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.320556640625,
      "learning_rate": 8.827691940504634e-05,
      "loss": 2.1239,
      "step": 172661
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.2209583520889282,
      "learning_rate": 8.82740213950374e-05,
      "loss": 1.8614,
      "step": 172662
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.4307365417480469,
      "learning_rate": 8.82711234243924e-05,
      "loss": 2.3129,
      "step": 172663
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.1507914066314697,
      "learning_rate": 8.82682254931117e-05,
      "loss": 1.8789,
      "step": 172664
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.3152077198028564,
      "learning_rate": 8.826532760119603e-05,
      "loss": 2.1687,
      "step": 172665
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.3393851518630981,
      "learning_rate": 8.826242974864574e-05,
      "loss": 2.0618,
      "step": 172666
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.6429604291915894,
      "learning_rate": 8.825953193546167e-05,
      "loss": 2.1827,
      "step": 172667
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.2765555381774902,
      "learning_rate": 8.825663416164394e-05,
      "loss": 2.2276,
      "step": 172668
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.220583200454712,
      "learning_rate": 8.82537364271934e-05,
      "loss": 1.9991,
      "step": 172669
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.3734698295593262,
      "learning_rate": 8.825083873211039e-05,
      "loss": 1.9483,
      "step": 172670
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.2698971033096313,
      "learning_rate": 8.824794107639565e-05,
      "loss": 2.2607,
      "step": 172671
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.3707093000411987,
      "learning_rate": 8.824504346004953e-05,
      "loss": 1.9453,
      "step": 172672
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.393784523010254,
      "learning_rate": 8.82421458830728e-05,
      "loss": 2.2119,
      "step": 172673
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.2212116718292236,
      "learning_rate": 8.823924834546569e-05,
      "loss": 1.8744,
      "step": 172674
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.4086477756500244,
      "learning_rate": 8.823635084722901e-05,
      "loss": 1.9233,
      "step": 172675
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.4307942390441895,
      "learning_rate": 8.823345338836307e-05,
      "loss": 2.3085,
      "step": 172676
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.263317584991455,
      "learning_rate": 8.823055596886864e-05,
      "loss": 2.1688,
      "step": 172677
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.5857279300689697,
      "learning_rate": 8.822765858874602e-05,
      "loss": 2.236,
      "step": 172678
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.3349690437316895,
      "learning_rate": 8.82247612479961e-05,
      "loss": 2.0159,
      "step": 172679
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.4476118087768555,
      "learning_rate": 8.822186394661899e-05,
      "loss": 1.9646,
      "step": 172680
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.3520255088806152,
      "learning_rate": 8.821896668461555e-05,
      "loss": 1.8044,
      "step": 172681
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.453406572341919,
      "learning_rate": 8.821606946198607e-05,
      "loss": 1.9993,
      "step": 172682
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.2498618364334106,
      "learning_rate": 8.821317227873134e-05,
      "loss": 1.8979,
      "step": 172683
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.3564200401306152,
      "learning_rate": 8.821027513485171e-05,
      "loss": 1.9396,
      "step": 172684
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.2392257452011108,
      "learning_rate": 8.820737803034791e-05,
      "loss": 1.8225,
      "step": 172685
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.372047781944275,
      "learning_rate": 8.820448096522022e-05,
      "loss": 1.7637,
      "step": 172686
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.3495768308639526,
      "learning_rate": 8.820158393946939e-05,
      "loss": 1.9927,
      "step": 172687
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.3853081464767456,
      "learning_rate": 8.819868695309579e-05,
      "loss": 2.0803,
      "step": 172688
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.1694320440292358,
      "learning_rate": 8.819579000610015e-05,
      "loss": 2.0091,
      "step": 172689
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.2542988061904907,
      "learning_rate": 8.819289309848282e-05,
      "loss": 1.9802,
      "step": 172690
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.3040251731872559,
      "learning_rate": 8.818999623024453e-05,
      "loss": 2.158,
      "step": 172691
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.2658313512802124,
      "learning_rate": 8.818709940138571e-05,
      "loss": 1.9848,
      "step": 172692
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.3894636631011963,
      "learning_rate": 8.818420261190689e-05,
      "loss": 2.1952,
      "step": 172693
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.4146467447280884,
      "learning_rate": 8.818130586180852e-05,
      "loss": 2.0221,
      "step": 172694
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.2497566938400269,
      "learning_rate": 8.817840915109137e-05,
      "loss": 1.9456,
      "step": 172695
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.3404333591461182,
      "learning_rate": 8.817551247975578e-05,
      "loss": 1.7253,
      "step": 172696
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.2661908864974976,
      "learning_rate": 8.817261584780241e-05,
      "loss": 1.8791,
      "step": 172697
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.2361674308776855,
      "learning_rate": 8.816971925523178e-05,
      "loss": 2.0668,
      "step": 172698
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.3105103969573975,
      "learning_rate": 8.816682270204437e-05,
      "loss": 2.0826,
      "step": 172699
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.2791483402252197,
      "learning_rate": 8.816392618824066e-05,
      "loss": 2.1361,
      "step": 172700
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.341163992881775,
      "learning_rate": 8.816102971382138e-05,
      "loss": 2.3077,
      "step": 172701
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.4651128053665161,
      "learning_rate": 8.815813327878683e-05,
      "loss": 2.3292,
      "step": 172702
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.5194140672683716,
      "learning_rate": 8.815523688313782e-05,
      "loss": 2.0811,
      "step": 172703
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.299167513847351,
      "learning_rate": 8.815234052687476e-05,
      "loss": 2.0106,
      "step": 172704
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.2659894227981567,
      "learning_rate": 8.814944420999816e-05,
      "loss": 2.0657,
      "step": 172705
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.4656699895858765,
      "learning_rate": 8.814654793250848e-05,
      "loss": 2.1269,
      "step": 172706
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.4686611890792847,
      "learning_rate": 8.814365169440643e-05,
      "loss": 1.9445,
      "step": 172707
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.546178936958313,
      "learning_rate": 8.814075549569242e-05,
      "loss": 2.2563,
      "step": 172708
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.2622066736221313,
      "learning_rate": 8.813785933636709e-05,
      "loss": 2.1571,
      "step": 172709
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.2988510131835938,
      "learning_rate": 8.8134963216431e-05,
      "loss": 2.0175,
      "step": 172710
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.4572964906692505,
      "learning_rate": 8.813206713588454e-05,
      "loss": 2.2145,
      "step": 172711
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.151785135269165,
      "learning_rate": 8.812917109472829e-05,
      "loss": 1.8748,
      "step": 172712
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.5661914348602295,
      "learning_rate": 8.81262750929629e-05,
      "loss": 2.0622,
      "step": 172713
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.1959078311920166,
      "learning_rate": 8.812337913058875e-05,
      "loss": 2.0338,
      "step": 172714
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.2080179452896118,
      "learning_rate": 8.812048320760654e-05,
      "loss": 2.0519,
      "step": 172715
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.339805006980896,
      "learning_rate": 8.811758732401674e-05,
      "loss": 1.9182,
      "step": 172716
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.307191252708435,
      "learning_rate": 8.811469147981987e-05,
      "loss": 2.0423,
      "step": 172717
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.3054414987564087,
      "learning_rate": 8.811179567501643e-05,
      "loss": 2.2197,
      "step": 172718
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.2496166229248047,
      "learning_rate": 8.810889990960707e-05,
      "loss": 1.9123,
      "step": 172719
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.2369558811187744,
      "learning_rate": 8.810600418359217e-05,
      "loss": 1.994,
      "step": 172720
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.3296563625335693,
      "learning_rate": 8.810310849697244e-05,
      "loss": 1.9175,
      "step": 172721
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.203513741493225,
      "learning_rate": 8.810021284974834e-05,
      "loss": 1.9736,
      "step": 172722
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.291089653968811,
      "learning_rate": 8.809731724192046e-05,
      "loss": 2.2521,
      "step": 172723
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.3543670177459717,
      "learning_rate": 8.809442167348914e-05,
      "loss": 2.0413,
      "step": 172724
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.242260456085205,
      "learning_rate": 8.809152614445519e-05,
      "loss": 1.9777,
      "step": 172725
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.2402713298797607,
      "learning_rate": 8.808863065481894e-05,
      "loss": 2.1498,
      "step": 172726
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.2350703477859497,
      "learning_rate": 8.80857352045811e-05,
      "loss": 2.0033,
      "step": 172727
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.3380873203277588,
      "learning_rate": 8.808283979374212e-05,
      "loss": 1.9257,
      "step": 172728
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.1891202926635742,
      "learning_rate": 8.807994442230244e-05,
      "loss": 2.2239,
      "step": 172729
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.3814507722854614,
      "learning_rate": 8.807704909026278e-05,
      "loss": 2.1076,
      "step": 172730
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.2538727521896362,
      "learning_rate": 8.807415379762361e-05,
      "loss": 1.829,
      "step": 172731
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.2701377868652344,
      "learning_rate": 8.807125854438535e-05,
      "loss": 1.9294,
      "step": 172732
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.2924784421920776,
      "learning_rate": 8.806836333054875e-05,
      "loss": 2.1391,
      "step": 172733
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.242455244064331,
      "learning_rate": 8.806546815611426e-05,
      "loss": 2.0579,
      "step": 172734
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.3394708633422852,
      "learning_rate": 8.806257302108228e-05,
      "loss": 2.1467,
      "step": 172735
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.3566969633102417,
      "learning_rate": 8.805967792545359e-05,
      "loss": 2.1093,
      "step": 172736
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.2572990655899048,
      "learning_rate": 8.805678286922858e-05,
      "loss": 1.9796,
      "step": 172737
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.3559788465499878,
      "learning_rate": 8.805388785240773e-05,
      "loss": 2.1545,
      "step": 172738
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.2129223346710205,
      "learning_rate": 8.805099287499175e-05,
      "loss": 1.8485,
      "step": 172739
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.133683681488037,
      "learning_rate": 8.80480979369811e-05,
      "loss": 2.0793,
      "step": 172740
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.2159944772720337,
      "learning_rate": 8.80452030383762e-05,
      "loss": 2.1916,
      "step": 172741
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.3727244138717651,
      "learning_rate": 8.804230817917781e-05,
      "loss": 1.9121,
      "step": 172742
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.2414013147354126,
      "learning_rate": 8.803941335938627e-05,
      "loss": 1.8999,
      "step": 172743
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.2949581146240234,
      "learning_rate": 8.803651857900227e-05,
      "loss": 2.0956,
      "step": 172744
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.21483314037323,
      "learning_rate": 8.803362383802628e-05,
      "loss": 1.7981,
      "step": 172745
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.2139569520950317,
      "learning_rate": 8.803072913645886e-05,
      "loss": 2.0252,
      "step": 172746
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.280968189239502,
      "learning_rate": 8.802783447430043e-05,
      "loss": 2.1203,
      "step": 172747
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.276761531829834,
      "learning_rate": 8.802493985155174e-05,
      "loss": 2.1319,
      "step": 172748
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.2491576671600342,
      "learning_rate": 8.802204526821308e-05,
      "loss": 1.8885,
      "step": 172749
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.1977821588516235,
      "learning_rate": 8.801915072428527e-05,
      "loss": 2.1073,
      "step": 172750
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.394659161567688,
      "learning_rate": 8.801625621976857e-05,
      "loss": 2.0925,
      "step": 172751
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.3823461532592773,
      "learning_rate": 8.801336175466383e-05,
      "loss": 2.0324,
      "step": 172752
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.3429887294769287,
      "learning_rate": 8.801046732897125e-05,
      "loss": 2.1359,
      "step": 172753
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.5108277797698975,
      "learning_rate": 8.80075729426916e-05,
      "loss": 1.9578,
      "step": 172754
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.330696702003479,
      "learning_rate": 8.800467859582525e-05,
      "loss": 1.9659,
      "step": 172755
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.1446077823638916,
      "learning_rate": 8.800178428837295e-05,
      "loss": 1.9087,
      "step": 172756
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.2299147844314575,
      "learning_rate": 8.799889002033497e-05,
      "loss": 2.048,
      "step": 172757
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.4177945852279663,
      "learning_rate": 8.799599579171215e-05,
      "loss": 2.1001,
      "step": 172758
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.2149016857147217,
      "learning_rate": 8.799310160250486e-05,
      "loss": 2.072,
      "step": 172759
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.3491954803466797,
      "learning_rate": 8.799020745271363e-05,
      "loss": 2.1354,
      "step": 172760
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.1932692527770996,
      "learning_rate": 8.798731334233896e-05,
      "loss": 1.8455,
      "step": 172761
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.2082068920135498,
      "learning_rate": 8.798441927138153e-05,
      "loss": 1.9476,
      "step": 172762
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.3705575466156006,
      "learning_rate": 8.798152523984172e-05,
      "loss": 2.0061,
      "step": 172763
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.2891660928726196,
      "learning_rate": 8.797863124772021e-05,
      "loss": 1.9961,
      "step": 172764
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.3773647546768188,
      "learning_rate": 8.79757372950175e-05,
      "loss": 1.8841,
      "step": 172765
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.434902310371399,
      "learning_rate": 8.79728433817341e-05,
      "loss": 1.8899,
      "step": 172766
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.4390488862991333,
      "learning_rate": 8.796994950787045e-05,
      "loss": 2.0904,
      "step": 172767
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.2027183771133423,
      "learning_rate": 8.796705567342727e-05,
      "loss": 2.1474,
      "step": 172768
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.593420386314392,
      "learning_rate": 8.796416187840494e-05,
      "loss": 1.9244,
      "step": 172769
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.4893524646759033,
      "learning_rate": 8.796126812280416e-05,
      "loss": 2.0359,
      "step": 172770
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.632362723350525,
      "learning_rate": 8.795837440662539e-05,
      "loss": 2.0058,
      "step": 172771
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.3160133361816406,
      "learning_rate": 8.795548072986918e-05,
      "loss": 2.044,
      "step": 172772
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.2642275094985962,
      "learning_rate": 8.795258709253594e-05,
      "loss": 1.925,
      "step": 172773
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.3255820274353027,
      "learning_rate": 8.794969349462638e-05,
      "loss": 1.9461,
      "step": 172774
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.3544591665267944,
      "learning_rate": 8.79467999361409e-05,
      "loss": 1.9721,
      "step": 172775
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.5801455974578857,
      "learning_rate": 8.794390641708021e-05,
      "loss": 2.0188,
      "step": 172776
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.2053085565567017,
      "learning_rate": 8.794101293744475e-05,
      "loss": 2.0212,
      "step": 172777
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.2638472318649292,
      "learning_rate": 8.793811949723509e-05,
      "loss": 2.1002,
      "step": 172778
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.402758002281189,
      "learning_rate": 8.79352260964516e-05,
      "loss": 2.1223,
      "step": 172779
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.4029276371002197,
      "learning_rate": 8.793233273509506e-05,
      "loss": 2.1134,
      "step": 172780
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.4539732933044434,
      "learning_rate": 8.792943941316581e-05,
      "loss": 1.8706,
      "step": 172781
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.3674607276916504,
      "learning_rate": 8.79265461306646e-05,
      "loss": 2.1198,
      "step": 172782
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.343901515007019,
      "learning_rate": 8.792365288759181e-05,
      "loss": 2.2091,
      "step": 172783
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.321785569190979,
      "learning_rate": 8.792075968394802e-05,
      "loss": 2.1585,
      "step": 172784
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.3240638971328735,
      "learning_rate": 8.791786651973368e-05,
      "loss": 2.289,
      "step": 172785
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.3827236890792847,
      "learning_rate": 8.791497339494951e-05,
      "loss": 1.9909,
      "step": 172786
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.3123726844787598,
      "learning_rate": 8.791208030959585e-05,
      "loss": 1.9717,
      "step": 172787
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.245912790298462,
      "learning_rate": 8.790918726367343e-05,
      "loss": 2.1832,
      "step": 172788
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.3073394298553467,
      "learning_rate": 8.79062942571827e-05,
      "loss": 2.1873,
      "step": 172789
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.279394507408142,
      "learning_rate": 8.790340129012419e-05,
      "loss": 1.9973,
      "step": 172790
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.276930809020996,
      "learning_rate": 8.790050836249837e-05,
      "loss": 2.0966,
      "step": 172791
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.3457857370376587,
      "learning_rate": 8.789761547430591e-05,
      "loss": 1.8844,
      "step": 172792
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.4560153484344482,
      "learning_rate": 8.789472262554719e-05,
      "loss": 2.0697,
      "step": 172793
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.214611291885376,
      "learning_rate": 8.789182981622298e-05,
      "loss": 2.0725,
      "step": 172794
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.336845874786377,
      "learning_rate": 8.788893704633366e-05,
      "loss": 2.0018,
      "step": 172795
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.1936008930206299,
      "learning_rate": 8.788604431587977e-05,
      "loss": 1.9704,
      "step": 172796
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.4016538858413696,
      "learning_rate": 8.78831516248618e-05,
      "loss": 2.0762,
      "step": 172797
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.5104490518569946,
      "learning_rate": 8.788025897328046e-05,
      "loss": 2.1248,
      "step": 172798
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.5671948194503784,
      "learning_rate": 8.787736636113605e-05,
      "loss": 2.0538,
      "step": 172799
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.3864343166351318,
      "learning_rate": 8.787447378842935e-05,
      "loss": 1.7692,
      "step": 172800
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.3529024124145508,
      "learning_rate": 8.787158125516081e-05,
      "loss": 1.9513,
      "step": 172801
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.2989064455032349,
      "learning_rate": 8.786868876133093e-05,
      "loss": 2.1689,
      "step": 172802
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.2260630130767822,
      "learning_rate": 8.78657963069402e-05,
      "loss": 1.9092,
      "step": 172803
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.5185297727584839,
      "learning_rate": 8.786290389198927e-05,
      "loss": 2.2252,
      "step": 172804
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.3272652626037598,
      "learning_rate": 8.786001151647858e-05,
      "loss": 1.9374,
      "step": 172805
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.444444179534912,
      "learning_rate": 8.785711918040877e-05,
      "loss": 1.946,
      "step": 172806
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.3518519401550293,
      "learning_rate": 8.785422688378036e-05,
      "loss": 1.9021,
      "step": 172807
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.8599212169647217,
      "learning_rate": 8.785133462659385e-05,
      "loss": 2.0956,
      "step": 172808
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.814501166343689,
      "learning_rate": 8.784844240884966e-05,
      "loss": 2.214,
      "step": 172809
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.6576091051101685,
      "learning_rate": 8.784555023054857e-05,
      "loss": 2.1699,
      "step": 172810
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.3100078105926514,
      "learning_rate": 8.78426580916909e-05,
      "loss": 2.1083,
      "step": 172811
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.7221769094467163,
      "learning_rate": 8.783976599227738e-05,
      "loss": 2.0603,
      "step": 172812
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.4074578285217285,
      "learning_rate": 8.783687393230844e-05,
      "loss": 2.1184,
      "step": 172813
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.3069432973861694,
      "learning_rate": 8.783398191178454e-05,
      "loss": 2.0991,
      "step": 172814
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.3778483867645264,
      "learning_rate": 8.78310899307064e-05,
      "loss": 2.2271,
      "step": 172815
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.419543981552124,
      "learning_rate": 8.782819798907446e-05,
      "loss": 2.1363,
      "step": 172816
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.3499435186386108,
      "learning_rate": 8.782530608688919e-05,
      "loss": 1.9546,
      "step": 172817
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.2178149223327637,
      "learning_rate": 8.782241422415128e-05,
      "loss": 2.1079,
      "step": 172818
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.3317497968673706,
      "learning_rate": 8.78195224008611e-05,
      "loss": 1.9674,
      "step": 172819
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.474226713180542,
      "learning_rate": 8.781663061701935e-05,
      "loss": 1.9594,
      "step": 172820
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.5041714906692505,
      "learning_rate": 8.781373887262652e-05,
      "loss": 2.1501,
      "step": 172821
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.1142666339874268,
      "learning_rate": 8.781084716768312e-05,
      "loss": 1.8448,
      "step": 172822
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.3377224206924438,
      "learning_rate": 8.78079555021896e-05,
      "loss": 2.0268,
      "step": 172823
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.4675294160842896,
      "learning_rate": 8.780506387614666e-05,
      "loss": 2.0345,
      "step": 172824
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.6504675149917603,
      "learning_rate": 8.780217228955469e-05,
      "loss": 1.9937,
      "step": 172825
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.4899247884750366,
      "learning_rate": 8.779928074241436e-05,
      "loss": 2.2688,
      "step": 172826
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.2503865957260132,
      "learning_rate": 8.779638923472617e-05,
      "loss": 2.0947,
      "step": 172827
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.2867449522018433,
      "learning_rate": 8.779349776649056e-05,
      "loss": 2.0403,
      "step": 172828
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.2547717094421387,
      "learning_rate": 8.779060633770823e-05,
      "loss": 2.0312,
      "step": 172829
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.3156434297561646,
      "learning_rate": 8.778771494837962e-05,
      "loss": 2.1009,
      "step": 172830
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.2457810640335083,
      "learning_rate": 8.778482359850518e-05,
      "loss": 2.0216,
      "step": 172831
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.4599709510803223,
      "learning_rate": 8.778193228808566e-05,
      "loss": 2.0666,
      "step": 172832
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.2539923191070557,
      "learning_rate": 8.777904101712148e-05,
      "loss": 1.9207,
      "step": 172833
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.3307182788848877,
      "learning_rate": 8.777614978561307e-05,
      "loss": 2.0276,
      "step": 172834
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.320984125137329,
      "learning_rate": 8.77732585935612e-05,
      "loss": 2.1604,
      "step": 172835
    },
    {
      "epoch": 2.25,
      "grad_norm": 2.043708324432373,
      "learning_rate": 8.777036744096617e-05,
      "loss": 2.0572,
      "step": 172836
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.2992297410964966,
      "learning_rate": 8.776747632782874e-05,
      "loss": 2.1698,
      "step": 172837
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.240371584892273,
      "learning_rate": 8.776458525414935e-05,
      "loss": 1.9466,
      "step": 172838
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.3544522523880005,
      "learning_rate": 8.776169421992853e-05,
      "loss": 2.0274,
      "step": 172839
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.262082576751709,
      "learning_rate": 8.775880322516671e-05,
      "loss": 2.0288,
      "step": 172840
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.310110330581665,
      "learning_rate": 8.775591226986463e-05,
      "loss": 1.9491,
      "step": 172841
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.2078790664672852,
      "learning_rate": 8.775302135402266e-05,
      "loss": 2.086,
      "step": 172842
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.5266966819763184,
      "learning_rate": 8.775013047764151e-05,
      "loss": 2.1029,
      "step": 172843
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.729817271232605,
      "learning_rate": 8.774723964072157e-05,
      "loss": 2.0363,
      "step": 172844
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.2737616300582886,
      "learning_rate": 8.774434884326346e-05,
      "loss": 1.9421,
      "step": 172845
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.3397190570831299,
      "learning_rate": 8.774145808526758e-05,
      "loss": 2.1196,
      "step": 172846
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.216591238975525,
      "learning_rate": 8.773856736673467e-05,
      "loss": 2.3102,
      "step": 172847
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.3494515419006348,
      "learning_rate": 8.773567668766506e-05,
      "loss": 2.1728,
      "step": 172848
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.191776156425476,
      "learning_rate": 8.773278604805948e-05,
      "loss": 2.0431,
      "step": 172849
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.3091627359390259,
      "learning_rate": 8.77298954479184e-05,
      "loss": 2.0812,
      "step": 172850
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.171125888824463,
      "learning_rate": 8.772700488724235e-05,
      "loss": 2.015,
      "step": 172851
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.3824256658554077,
      "learning_rate": 8.772411436603173e-05,
      "loss": 1.8864,
      "step": 172852
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.2404347658157349,
      "learning_rate": 8.772122388428734e-05,
      "loss": 1.9577,
      "step": 172853
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.395266056060791,
      "learning_rate": 8.771833344200945e-05,
      "loss": 2.2714,
      "step": 172854
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.2379292249679565,
      "learning_rate": 8.771544303919886e-05,
      "loss": 1.8711,
      "step": 172855
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.398343801498413,
      "learning_rate": 8.771255267585592e-05,
      "loss": 2.1672,
      "step": 172856
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.1895420551300049,
      "learning_rate": 8.770966235198128e-05,
      "loss": 2.0086,
      "step": 172857
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.2758468389511108,
      "learning_rate": 8.770677206757531e-05,
      "loss": 1.9652,
      "step": 172858
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.2572507858276367,
      "learning_rate": 8.770388182263872e-05,
      "loss": 2.0796,
      "step": 172859
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.2803162336349487,
      "learning_rate": 8.770099161717194e-05,
      "loss": 2.0679,
      "step": 172860
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.1900289058685303,
      "learning_rate": 8.769810145117561e-05,
      "loss": 1.9902,
      "step": 172861
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.3442481756210327,
      "learning_rate": 8.769521132465024e-05,
      "loss": 1.9642,
      "step": 172862
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.3464409112930298,
      "learning_rate": 8.76923212375963e-05,
      "loss": 2.183,
      "step": 172863
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.215808391571045,
      "learning_rate": 8.768943119001431e-05,
      "loss": 2.0217,
      "step": 172864
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.3337608575820923,
      "learning_rate": 8.768654118190493e-05,
      "loss": 1.9297,
      "step": 172865
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.433844804763794,
      "learning_rate": 8.768365121326855e-05,
      "loss": 1.9722,
      "step": 172866
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.2324994802474976,
      "learning_rate": 8.768076128410588e-05,
      "loss": 1.9292,
      "step": 172867
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.2738722562789917,
      "learning_rate": 8.767787139441737e-05,
      "loss": 1.9483,
      "step": 172868
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.1749117374420166,
      "learning_rate": 8.767498154420354e-05,
      "loss": 1.9592,
      "step": 172869
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.3273255825042725,
      "learning_rate": 8.767209173346483e-05,
      "loss": 1.9446,
      "step": 172870
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.3436559438705444,
      "learning_rate": 8.766920196220201e-05,
      "loss": 2.1482,
      "step": 172871
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.2973718643188477,
      "learning_rate": 8.766631223041539e-05,
      "loss": 2.1145,
      "step": 172872
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.251838207244873,
      "learning_rate": 8.766342253810569e-05,
      "loss": 1.9968,
      "step": 172873
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.2311278581619263,
      "learning_rate": 8.766053288527341e-05,
      "loss": 2.1444,
      "step": 172874
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.2142424583435059,
      "learning_rate": 8.765764327191901e-05,
      "loss": 2.1131,
      "step": 172875
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.2986140251159668,
      "learning_rate": 8.765475369804297e-05,
      "loss": 2.001,
      "step": 172876
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.3192518949508667,
      "learning_rate": 8.765186416364601e-05,
      "loss": 2.154,
      "step": 172877
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.256388545036316,
      "learning_rate": 8.764897466872848e-05,
      "loss": 1.9786,
      "step": 172878
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.2335865497589111,
      "learning_rate": 8.764608521329113e-05,
      "loss": 2.0659,
      "step": 172879
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.280640959739685,
      "learning_rate": 8.764319579733436e-05,
      "loss": 2.0459,
      "step": 172880
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.3385345935821533,
      "learning_rate": 8.764030642085875e-05,
      "loss": 2.1871,
      "step": 172881
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.2755911350250244,
      "learning_rate": 8.76374170838647e-05,
      "loss": 2.0971,
      "step": 172882
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.2568731307983398,
      "learning_rate": 8.763452778635299e-05,
      "loss": 2.008,
      "step": 172883
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.2300368547439575,
      "learning_rate": 8.763163852832389e-05,
      "loss": 2.0509,
      "step": 172884
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.3504301309585571,
      "learning_rate": 8.762874930977822e-05,
      "loss": 2.1146,
      "step": 172885
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.4741551876068115,
      "learning_rate": 8.762586013071622e-05,
      "loss": 2.2558,
      "step": 172886
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.1823670864105225,
      "learning_rate": 8.762297099113881e-05,
      "loss": 1.9641,
      "step": 172887
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.468363642692566,
      "learning_rate": 8.762008189104611e-05,
      "loss": 2.1928,
      "step": 172888
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.3071399927139282,
      "learning_rate": 8.761719283043892e-05,
      "loss": 2.1275,
      "step": 172889
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.225327968597412,
      "learning_rate": 8.761430380931763e-05,
      "loss": 2.07,
      "step": 172890
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.203933835029602,
      "learning_rate": 8.761141482768294e-05,
      "loss": 1.9589,
      "step": 172891
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.276230812072754,
      "learning_rate": 8.760852588553521e-05,
      "loss": 2.0526,
      "step": 172892
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.3251858949661255,
      "learning_rate": 8.760563698287525e-05,
      "loss": 2.059,
      "step": 172893
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.2769417762756348,
      "learning_rate": 8.760274811970321e-05,
      "loss": 1.9586,
      "step": 172894
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.229220986366272,
      "learning_rate": 8.75998592960199e-05,
      "loss": 2.0839,
      "step": 172895
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.2943191528320312,
      "learning_rate": 8.759697051182575e-05,
      "loss": 2.0538,
      "step": 172896
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.3745790719985962,
      "learning_rate": 8.759408176712137e-05,
      "loss": 2.0059,
      "step": 172897
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.4155110120773315,
      "learning_rate": 8.759119306190721e-05,
      "loss": 2.0252,
      "step": 172898
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.2297871112823486,
      "learning_rate": 8.758830439618402e-05,
      "loss": 2.0894,
      "step": 172899
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.178475260734558,
      "learning_rate": 8.7585415769952e-05,
      "loss": 1.8168,
      "step": 172900
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.6331230401992798,
      "learning_rate": 8.758252718321194e-05,
      "loss": 2.0872,
      "step": 172901
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.221683144569397,
      "learning_rate": 8.757963863596422e-05,
      "loss": 1.9953,
      "step": 172902
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.31298828125,
      "learning_rate": 8.757675012820952e-05,
      "loss": 1.9255,
      "step": 172903
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.3255493640899658,
      "learning_rate": 8.757386165994825e-05,
      "loss": 1.8826,
      "step": 172904
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.5057874917984009,
      "learning_rate": 8.757097323118109e-05,
      "loss": 1.9709,
      "step": 172905
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.2578045129776,
      "learning_rate": 8.75680848419085e-05,
      "loss": 1.8121,
      "step": 172906
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.117295742034912,
      "learning_rate": 8.7565196492131e-05,
      "loss": 1.8896,
      "step": 172907
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.1743308305740356,
      "learning_rate": 8.756230818184906e-05,
      "loss": 2.063,
      "step": 172908
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.3270010948181152,
      "learning_rate": 8.755941991106339e-05,
      "loss": 2.1342,
      "step": 172909
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.4157626628875732,
      "learning_rate": 8.755653167977432e-05,
      "loss": 2.133,
      "step": 172910
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.5203592777252197,
      "learning_rate": 8.755364348798263e-05,
      "loss": 2.0093,
      "step": 172911
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.2691271305084229,
      "learning_rate": 8.75507553356887e-05,
      "loss": 1.9613,
      "step": 172912
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.23601496219635,
      "learning_rate": 8.754786722289302e-05,
      "loss": 2.1974,
      "step": 172913
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.271999478340149,
      "learning_rate": 8.754497914959627e-05,
      "loss": 1.9483,
      "step": 172914
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.7297416925430298,
      "learning_rate": 8.754209111579896e-05,
      "loss": 2.1013,
      "step": 172915
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.4581944942474365,
      "learning_rate": 8.753920312150146e-05,
      "loss": 1.959,
      "step": 172916
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.1854101419448853,
      "learning_rate": 8.753631516670451e-05,
      "loss": 2.1604,
      "step": 172917
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.2901909351348877,
      "learning_rate": 8.753342725140861e-05,
      "loss": 2.1099,
      "step": 172918
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.3331636190414429,
      "learning_rate": 8.753053937561416e-05,
      "loss": 2.222,
      "step": 172919
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.5091392993927002,
      "learning_rate": 8.752765153932185e-05,
      "loss": 2.1047,
      "step": 172920
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.3761355876922607,
      "learning_rate": 8.752476374253219e-05,
      "loss": 2.3129,
      "step": 172921
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.297097086906433,
      "learning_rate": 8.75218759852456e-05,
      "loss": 2.0207,
      "step": 172922
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.2348597049713135,
      "learning_rate": 8.751898826746276e-05,
      "loss": 2.0944,
      "step": 172923
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.3265550136566162,
      "learning_rate": 8.75161005891842e-05,
      "loss": 2.0143,
      "step": 172924
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.3005677461624146,
      "learning_rate": 8.751321295041028e-05,
      "loss": 2.1033,
      "step": 172925
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.3762236833572388,
      "learning_rate": 8.751032535114177e-05,
      "loss": 2.0993,
      "step": 172926
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.2633366584777832,
      "learning_rate": 8.750743779137901e-05,
      "loss": 1.9138,
      "step": 172927
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.2697597742080688,
      "learning_rate": 8.75045502711227e-05,
      "loss": 1.8756,
      "step": 172928
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.239716649055481,
      "learning_rate": 8.750166279037334e-05,
      "loss": 2.0238,
      "step": 172929
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.5392276048660278,
      "learning_rate": 8.74987753491314e-05,
      "loss": 2.0641,
      "step": 172930
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.2305314540863037,
      "learning_rate": 8.749588794739739e-05,
      "loss": 2.1671,
      "step": 172931
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.3135402202606201,
      "learning_rate": 8.749300058517198e-05,
      "loss": 2.0269,
      "step": 172932
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.3262290954589844,
      "learning_rate": 8.749011326245553e-05,
      "loss": 2.1181,
      "step": 172933
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.2045376300811768,
      "learning_rate": 8.748722597924878e-05,
      "loss": 2.1428,
      "step": 172934
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.2261744737625122,
      "learning_rate": 8.748433873555219e-05,
      "loss": 2.0474,
      "step": 172935
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.3440250158309937,
      "learning_rate": 8.748145153136622e-05,
      "loss": 2.0957,
      "step": 172936
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.2432411909103394,
      "learning_rate": 8.747856436669142e-05,
      "loss": 1.7881,
      "step": 172937
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.2648093700408936,
      "learning_rate": 8.747567724152841e-05,
      "loss": 1.9193,
      "step": 172938
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.280983328819275,
      "learning_rate": 8.747279015587762e-05,
      "loss": 2.0269,
      "step": 172939
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.2547900676727295,
      "learning_rate": 8.746990310973974e-05,
      "loss": 1.9239,
      "step": 172940
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.3187552690505981,
      "learning_rate": 8.746701610311524e-05,
      "loss": 1.8504,
      "step": 172941
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.2797749042510986,
      "learning_rate": 8.74641291360046e-05,
      "loss": 2.0477,
      "step": 172942
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.2358410358428955,
      "learning_rate": 8.746124220840833e-05,
      "loss": 2.0822,
      "step": 172943
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.5819566249847412,
      "learning_rate": 8.745835532032706e-05,
      "loss": 1.9759,
      "step": 172944
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.6635116338729858,
      "learning_rate": 8.745546847176127e-05,
      "loss": 2.1379,
      "step": 172945
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.325628399848938,
      "learning_rate": 8.745258166271157e-05,
      "loss": 2.1016,
      "step": 172946
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.367525577545166,
      "learning_rate": 8.744969489317845e-05,
      "loss": 1.899,
      "step": 172947
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.268676996231079,
      "learning_rate": 8.744680816316245e-05,
      "loss": 2.0268,
      "step": 172948
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.3426456451416016,
      "learning_rate": 8.744392147266401e-05,
      "loss": 2.0854,
      "step": 172949
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.4362798929214478,
      "learning_rate": 8.744103482168386e-05,
      "loss": 1.925,
      "step": 172950
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.3039151430130005,
      "learning_rate": 8.743814821022233e-05,
      "loss": 2.0125,
      "step": 172951
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.1868441104888916,
      "learning_rate": 8.743526163828017e-05,
      "loss": 1.898,
      "step": 172952
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.3087222576141357,
      "learning_rate": 8.743237510585768e-05,
      "loss": 2.2904,
      "step": 172953
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.358001947402954,
      "learning_rate": 8.742948861295574e-05,
      "loss": 1.8715,
      "step": 172954
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.231331467628479,
      "learning_rate": 8.742660215957447e-05,
      "loss": 2.152,
      "step": 172955
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.1981444358825684,
      "learning_rate": 8.742371574571471e-05,
      "loss": 2.382,
      "step": 172956
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.3746641874313354,
      "learning_rate": 8.74208293713768e-05,
      "loss": 2.1691,
      "step": 172957
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.2319575548171997,
      "learning_rate": 8.741794303656144e-05,
      "loss": 1.9646,
      "step": 172958
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.4471862316131592,
      "learning_rate": 8.741505674126903e-05,
      "loss": 2.0554,
      "step": 172959
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.136772871017456,
      "learning_rate": 8.741217048550036e-05,
      "loss": 2.177,
      "step": 172960
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.3097068071365356,
      "learning_rate": 8.740928426925559e-05,
      "loss": 2.1813,
      "step": 172961
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.31926691532135,
      "learning_rate": 8.740639809253554e-05,
      "loss": 1.8831,
      "step": 172962
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.2873756885528564,
      "learning_rate": 8.740351195534056e-05,
      "loss": 1.9864,
      "step": 172963
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.305985450744629,
      "learning_rate": 8.740062585767138e-05,
      "loss": 1.8523,
      "step": 172964
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.2727032899856567,
      "learning_rate": 8.739773979952835e-05,
      "loss": 2.1338,
      "step": 172965
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.3703255653381348,
      "learning_rate": 8.739485378091226e-05,
      "loss": 1.8255,
      "step": 172966
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.607153058052063,
      "learning_rate": 8.739196780182328e-05,
      "loss": 2.0557,
      "step": 172967
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.2195357084274292,
      "learning_rate": 8.738908186226226e-05,
      "loss": 2.1601,
      "step": 172968
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.2921568155288696,
      "learning_rate": 8.738619596222955e-05,
      "loss": 2.0766,
      "step": 172969
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.2142456769943237,
      "learning_rate": 8.73833101017258e-05,
      "loss": 1.96,
      "step": 172970
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.3947197198867798,
      "learning_rate": 8.738042428075146e-05,
      "loss": 2.1682,
      "step": 172971
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.3647979497909546,
      "learning_rate": 8.73775384993073e-05,
      "loss": 2.0954,
      "step": 172972
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.3599568605422974,
      "learning_rate": 8.737465275739346e-05,
      "loss": 2.139,
      "step": 172973
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.253723382949829,
      "learning_rate": 8.737176705501078e-05,
      "loss": 1.6855,
      "step": 172974
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.3469756841659546,
      "learning_rate": 8.736888139215961e-05,
      "loss": 2.2091,
      "step": 172975
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.3766493797302246,
      "learning_rate": 8.736599576884068e-05,
      "loss": 2.046,
      "step": 172976
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.3047454357147217,
      "learning_rate": 8.736311018505433e-05,
      "loss": 2.1673,
      "step": 172977
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.3686529397964478,
      "learning_rate": 8.736022464080135e-05,
      "loss": 2.1943,
      "step": 172978
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.4088760614395142,
      "learning_rate": 8.735733913608197e-05,
      "loss": 2.1285,
      "step": 172979
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.3404461145401,
      "learning_rate": 8.735445367089697e-05,
      "loss": 1.8927,
      "step": 172980
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.2815639972686768,
      "learning_rate": 8.735156824524666e-05,
      "loss": 1.8878,
      "step": 172981
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.1908047199249268,
      "learning_rate": 8.734868285913182e-05,
      "loss": 1.9392,
      "step": 172982
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.2884538173675537,
      "learning_rate": 8.73457975125528e-05,
      "loss": 2.1025,
      "step": 172983
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.3137271404266357,
      "learning_rate": 8.734291220551038e-05,
      "loss": 1.9807,
      "step": 172984
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.2308400869369507,
      "learning_rate": 8.734002693800477e-05,
      "loss": 1.9428,
      "step": 172985
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.4317697286605835,
      "learning_rate": 8.733714171003671e-05,
      "loss": 1.9919,
      "step": 172986
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.3645920753479004,
      "learning_rate": 8.733425652160661e-05,
      "loss": 1.9409,
      "step": 172987
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.4796439409255981,
      "learning_rate": 8.733137137271518e-05,
      "loss": 1.8806,
      "step": 172988
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.420758843421936,
      "learning_rate": 8.732848626336276e-05,
      "loss": 2.1146,
      "step": 172989
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.572901725769043,
      "learning_rate": 8.732560119355011e-05,
      "loss": 2.0878,
      "step": 172990
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.3346705436706543,
      "learning_rate": 8.732271616327762e-05,
      "loss": 2.0748,
      "step": 172991
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.3200043439865112,
      "learning_rate": 8.731983117254588e-05,
      "loss": 2.088,
      "step": 172992
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.2630293369293213,
      "learning_rate": 8.731694622135527e-05,
      "loss": 1.883,
      "step": 172993
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.2750176191329956,
      "learning_rate": 8.731406130970658e-05,
      "loss": 2.0858,
      "step": 172994
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.3919365406036377,
      "learning_rate": 8.73111764376001e-05,
      "loss": 2.045,
      "step": 172995
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.5537256002426147,
      "learning_rate": 8.730829160503661e-05,
      "loss": 2.058,
      "step": 172996
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.3339980840682983,
      "learning_rate": 8.73054068120165e-05,
      "loss": 2.2336,
      "step": 172997
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.5103107690811157,
      "learning_rate": 8.730252205854025e-05,
      "loss": 1.878,
      "step": 172998
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.1767469644546509,
      "learning_rate": 8.729963734460856e-05,
      "loss": 1.9479,
      "step": 172999
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.3391633033752441,
      "learning_rate": 8.72967526702219e-05,
      "loss": 2.0025,
      "step": 173000
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.5458389520645142,
      "learning_rate": 8.729386803538067e-05,
      "loss": 1.9521,
      "step": 173001
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.3907595872879028,
      "learning_rate": 8.72909834400856e-05,
      "loss": 2.0413,
      "step": 173002
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.2577513456344604,
      "learning_rate": 8.728809888433721e-05,
      "loss": 1.9039,
      "step": 173003
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.3602567911148071,
      "learning_rate": 8.728521436813585e-05,
      "loss": 2.4056,
      "step": 173004
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.496086835861206,
      "learning_rate": 8.728232989148228e-05,
      "loss": 2.1458,
      "step": 173005
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.6219884157180786,
      "learning_rate": 8.727944545437692e-05,
      "loss": 1.9108,
      "step": 173006
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.4892600774765015,
      "learning_rate": 8.727656105682029e-05,
      "loss": 2.0002,
      "step": 173007
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.2502236366271973,
      "learning_rate": 8.727367669881301e-05,
      "loss": 2.2215,
      "step": 173008
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.5652440786361694,
      "learning_rate": 8.727079238035557e-05,
      "loss": 1.7566,
      "step": 173009
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.3981812000274658,
      "learning_rate": 8.726790810144842e-05,
      "loss": 2.0009,
      "step": 173010
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.6733437776565552,
      "learning_rate": 8.726502386209227e-05,
      "loss": 2.1182,
      "step": 173011
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.4509918689727783,
      "learning_rate": 8.726213966228749e-05,
      "loss": 2.2236,
      "step": 173012
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.469222903251648,
      "learning_rate": 8.725925550203476e-05,
      "loss": 2.0205,
      "step": 173013
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.290708303451538,
      "learning_rate": 8.72563713813346e-05,
      "loss": 1.969,
      "step": 173014
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.2865785360336304,
      "learning_rate": 8.725348730018747e-05,
      "loss": 2.009,
      "step": 173015
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.3253806829452515,
      "learning_rate": 8.725060325859382e-05,
      "loss": 2.073,
      "step": 173016
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.3223987817764282,
      "learning_rate": 8.724771925655441e-05,
      "loss": 2.1244,
      "step": 173017
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.3219687938690186,
      "learning_rate": 8.724483529406954e-05,
      "loss": 2.0635,
      "step": 173018
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.2442229986190796,
      "learning_rate": 8.724195137114002e-05,
      "loss": 1.7405,
      "step": 173019
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.5288351774215698,
      "learning_rate": 8.723906748776621e-05,
      "loss": 2.1512,
      "step": 173020
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.295275330543518,
      "learning_rate": 8.723618364394867e-05,
      "loss": 2.0169,
      "step": 173021
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.3113843202590942,
      "learning_rate": 8.723329983968784e-05,
      "loss": 2.0318,
      "step": 173022
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.4187698364257812,
      "learning_rate": 8.723041607498447e-05,
      "loss": 1.8645,
      "step": 173023
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.4601259231567383,
      "learning_rate": 8.722753234983886e-05,
      "loss": 1.9394,
      "step": 173024
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.3382017612457275,
      "learning_rate": 8.722464866425178e-05,
      "loss": 2.1727,
      "step": 173025
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.7301241159439087,
      "learning_rate": 8.722176501822356e-05,
      "loss": 1.9616,
      "step": 173026
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.342773199081421,
      "learning_rate": 8.721888141175502e-05,
      "loss": 1.9091,
      "step": 173027
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.2717795372009277,
      "learning_rate": 8.721599784484633e-05,
      "loss": 1.7952,
      "step": 173028
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.2432664632797241,
      "learning_rate": 8.721311431749826e-05,
      "loss": 1.988,
      "step": 173029
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.2336225509643555,
      "learning_rate": 8.721023082971122e-05,
      "loss": 1.8391,
      "step": 173030
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.3814878463745117,
      "learning_rate": 8.720734738148594e-05,
      "loss": 2.0076,
      "step": 173031
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.3355950117111206,
      "learning_rate": 8.72044639728227e-05,
      "loss": 2.1182,
      "step": 173032
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.2210651636123657,
      "learning_rate": 8.720158060372235e-05,
      "loss": 1.7886,
      "step": 173033
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.2000707387924194,
      "learning_rate": 8.719869727418506e-05,
      "loss": 1.9648,
      "step": 173034
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.105919361114502,
      "learning_rate": 8.719581398421165e-05,
      "loss": 1.721,
      "step": 173035
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.8348448276519775,
      "learning_rate": 8.719293073380246e-05,
      "loss": 2.2539,
      "step": 173036
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.3707045316696167,
      "learning_rate": 8.719004752295822e-05,
      "loss": 1.9182,
      "step": 173037
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.8267136812210083,
      "learning_rate": 8.718716435167926e-05,
      "loss": 1.8633,
      "step": 173038
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.1669526100158691,
      "learning_rate": 8.718428121996642e-05,
      "loss": 1.9907,
      "step": 173039
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.2364305257797241,
      "learning_rate": 8.718139812781983e-05,
      "loss": 1.9752,
      "step": 173040
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.3319941759109497,
      "learning_rate": 8.717851507524036e-05,
      "loss": 2.0472,
      "step": 173041
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.2756879329681396,
      "learning_rate": 8.717563206222831e-05,
      "loss": 1.9266,
      "step": 173042
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.2724378108978271,
      "learning_rate": 8.717274908878444e-05,
      "loss": 1.9932,
      "step": 173043
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.2342183589935303,
      "learning_rate": 8.716986615490907e-05,
      "loss": 1.8046,
      "step": 173044
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.2256814241409302,
      "learning_rate": 8.7166983260603e-05,
      "loss": 1.8844,
      "step": 173045
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.2273374795913696,
      "learning_rate": 8.716410040586646e-05,
      "loss": 2.0964,
      "step": 173046
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.3091975450515747,
      "learning_rate": 8.716121759070018e-05,
      "loss": 2.0264,
      "step": 173047
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.3231383562088013,
      "learning_rate": 8.715833481510462e-05,
      "loss": 1.9199,
      "step": 173048
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.3143882751464844,
      "learning_rate": 8.715545207908039e-05,
      "loss": 1.8609,
      "step": 173049
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.2990596294403076,
      "learning_rate": 8.715256938262787e-05,
      "loss": 2.0463,
      "step": 173050
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.6075875759124756,
      "learning_rate": 8.714968672574791e-05,
      "loss": 2.094,
      "step": 173051
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.3521254062652588,
      "learning_rate": 8.714680410844066e-05,
      "loss": 2.0759,
      "step": 173052
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.5416936874389648,
      "learning_rate": 8.714392153070696e-05,
      "loss": 1.9353,
      "step": 173053
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.3590058088302612,
      "learning_rate": 8.714103899254709e-05,
      "loss": 2.1424,
      "step": 173054
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.437726378440857,
      "learning_rate": 8.71381564939618e-05,
      "loss": 2.0715,
      "step": 173055
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.5892521142959595,
      "learning_rate": 8.713527403495148e-05,
      "loss": 1.9509,
      "step": 173056
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.4043352603912354,
      "learning_rate": 8.713239161551692e-05,
      "loss": 1.735,
      "step": 173057
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.2112438678741455,
      "learning_rate": 8.712950923565825e-05,
      "loss": 2.066,
      "step": 173058
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.3455965518951416,
      "learning_rate": 8.712662689537634e-05,
      "loss": 2.0222,
      "step": 173059
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.2422921657562256,
      "learning_rate": 8.712374459467152e-05,
      "loss": 1.9977,
      "step": 173060
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.3407392501831055,
      "learning_rate": 8.712086233354449e-05,
      "loss": 2.0103,
      "step": 173061
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.6130926609039307,
      "learning_rate": 8.711798011199563e-05,
      "loss": 2.178,
      "step": 173062
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.458720088005066,
      "learning_rate": 8.711509793002575e-05,
      "loss": 2.0077,
      "step": 173063
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.2773069143295288,
      "learning_rate": 8.7112215787635e-05,
      "loss": 2.0572,
      "step": 173064
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.2386611700057983,
      "learning_rate": 8.710933368482419e-05,
      "loss": 2.105,
      "step": 173065
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.388716459274292,
      "learning_rate": 8.710645162159368e-05,
      "loss": 2.2013,
      "step": 173066
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.3686686754226685,
      "learning_rate": 8.710356959794419e-05,
      "loss": 2.0675,
      "step": 173067
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.4508252143859863,
      "learning_rate": 8.710068761387607e-05,
      "loss": 2.2146,
      "step": 173068
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.2754579782485962,
      "learning_rate": 8.709780566939014e-05,
      "loss": 2.0567,
      "step": 173069
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.2785168886184692,
      "learning_rate": 8.709492376448659e-05,
      "loss": 1.7935,
      "step": 173070
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.33909273147583,
      "learning_rate": 8.709204189916617e-05,
      "loss": 2.1458,
      "step": 173071
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.3313528299331665,
      "learning_rate": 8.708916007342926e-05,
      "loss": 1.9517,
      "step": 173072
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.287598729133606,
      "learning_rate": 8.708627828727662e-05,
      "loss": 2.0837,
      "step": 173073
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.215364933013916,
      "learning_rate": 8.708339654070855e-05,
      "loss": 2.0115,
      "step": 173074
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.3041921854019165,
      "learning_rate": 8.708051483372578e-05,
      "loss": 1.9612,
      "step": 173075
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.1958391666412354,
      "learning_rate": 8.707763316632878e-05,
      "loss": 2.0766,
      "step": 173076
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.2478381395339966,
      "learning_rate": 8.707475153851808e-05,
      "loss": 2.2567,
      "step": 173077
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.2004669904708862,
      "learning_rate": 8.707186995029407e-05,
      "loss": 2.0777,
      "step": 173078
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.395338535308838,
      "learning_rate": 8.706898840165753e-05,
      "loss": 2.1333,
      "step": 173079
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.3718801736831665,
      "learning_rate": 8.70661068926088e-05,
      "loss": 2.079,
      "step": 173080
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.2851288318634033,
      "learning_rate": 8.706322542314856e-05,
      "loss": 2.0512,
      "step": 173081
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.2536983489990234,
      "learning_rate": 8.706034399327733e-05,
      "loss": 1.9476,
      "step": 173082
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.545932412147522,
      "learning_rate": 8.705746260299559e-05,
      "loss": 2.354,
      "step": 173083
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.3958251476287842,
      "learning_rate": 8.705458125230377e-05,
      "loss": 2.1063,
      "step": 173084
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.2791472673416138,
      "learning_rate": 8.705169994120263e-05,
      "loss": 1.9021,
      "step": 173085
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.3555790185928345,
      "learning_rate": 8.704881866969253e-05,
      "loss": 1.8303,
      "step": 173086
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.2273468971252441,
      "learning_rate": 8.704593743777414e-05,
      "loss": 2.1991,
      "step": 173087
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.2379045486450195,
      "learning_rate": 8.704305624544791e-05,
      "loss": 1.8957,
      "step": 173088
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.2679589986801147,
      "learning_rate": 8.704017509271435e-05,
      "loss": 2.0178,
      "step": 173089
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.2834694385528564,
      "learning_rate": 8.703729397957412e-05,
      "loss": 1.9597,
      "step": 173090
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.2907774448394775,
      "learning_rate": 8.703441290602767e-05,
      "loss": 2.0095,
      "step": 173091
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.284178614616394,
      "learning_rate": 8.703153187207545e-05,
      "loss": 2.0626,
      "step": 173092
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.2591699361801147,
      "learning_rate": 8.702865087771817e-05,
      "loss": 1.9501,
      "step": 173093
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.457423448562622,
      "learning_rate": 8.70257699229563e-05,
      "loss": 1.8311,
      "step": 173094
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.4197425842285156,
      "learning_rate": 8.702288900779027e-05,
      "loss": 2.0832,
      "step": 173095
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.297751545906067,
      "learning_rate": 8.702000813222076e-05,
      "loss": 1.9632,
      "step": 173096
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.2872099876403809,
      "learning_rate": 8.701712729624822e-05,
      "loss": 2.092,
      "step": 173097
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.1930806636810303,
      "learning_rate": 8.701424649987328e-05,
      "loss": 1.9761,
      "step": 173098
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.2433971166610718,
      "learning_rate": 8.701136574309638e-05,
      "loss": 2.0716,
      "step": 173099
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.3219858407974243,
      "learning_rate": 8.700848502591811e-05,
      "loss": 2.0471,
      "step": 173100
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.26216721534729,
      "learning_rate": 8.700560434833891e-05,
      "loss": 1.8452,
      "step": 173101
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.2443649768829346,
      "learning_rate": 8.70027237103595e-05,
      "loss": 2.0415,
      "step": 173102
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.4145631790161133,
      "learning_rate": 8.699984311198016e-05,
      "loss": 2.2148,
      "step": 173103
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.2844301462173462,
      "learning_rate": 8.699696255320169e-05,
      "loss": 1.8607,
      "step": 173104
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.1603631973266602,
      "learning_rate": 8.699408203402447e-05,
      "loss": 2.0065,
      "step": 173105
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.2315454483032227,
      "learning_rate": 8.699120155444911e-05,
      "loss": 2.0183,
      "step": 173106
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.2410789728164673,
      "learning_rate": 8.6988321114476e-05,
      "loss": 2.1643,
      "step": 173107
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.3200066089630127,
      "learning_rate": 8.698544071410588e-05,
      "loss": 1.9193,
      "step": 173108
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.3602838516235352,
      "learning_rate": 8.698256035333908e-05,
      "loss": 2.1009,
      "step": 173109
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.2400916814804077,
      "learning_rate": 8.697968003217637e-05,
      "loss": 2.2085,
      "step": 173110
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.323476791381836,
      "learning_rate": 8.697679975061804e-05,
      "loss": 1.8876,
      "step": 173111
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.3637492656707764,
      "learning_rate": 8.697391950866494e-05,
      "loss": 2.0369,
      "step": 173112
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.3141571283340454,
      "learning_rate": 8.697103930631721e-05,
      "loss": 2.0001,
      "step": 173113
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.2677271366119385,
      "learning_rate": 8.696815914357568e-05,
      "loss": 2.2006,
      "step": 173114
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.3704620599746704,
      "learning_rate": 8.69652790204407e-05,
      "loss": 2.0149,
      "step": 173115
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.3241627216339111,
      "learning_rate": 8.696239893691299e-05,
      "loss": 2.0206,
      "step": 173116
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.3034965991973877,
      "learning_rate": 8.695951889299291e-05,
      "loss": 1.9562,
      "step": 173117
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.2956939935684204,
      "learning_rate": 8.695663888868124e-05,
      "loss": 1.8712,
      "step": 173118
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.3506184816360474,
      "learning_rate": 8.695375892397819e-05,
      "loss": 1.8924,
      "step": 173119
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.342786431312561,
      "learning_rate": 8.695087899888456e-05,
      "loss": 1.9314,
      "step": 173120
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.2832697629928589,
      "learning_rate": 8.694799911340063e-05,
      "loss": 1.8438,
      "step": 173121
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.269109845161438,
      "learning_rate": 8.694511926752723e-05,
      "loss": 2.0625,
      "step": 173122
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.3573503494262695,
      "learning_rate": 8.694223946126467e-05,
      "loss": 2.1452,
      "step": 173123
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.2862451076507568,
      "learning_rate": 8.693935969461374e-05,
      "loss": 2.0372,
      "step": 173124
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.358325481414795,
      "learning_rate": 8.69364799675746e-05,
      "loss": 2.106,
      "step": 173125
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.3134299516677856,
      "learning_rate": 8.693360028014811e-05,
      "loss": 1.9733,
      "step": 173126
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.2370988130569458,
      "learning_rate": 8.693072063233458e-05,
      "loss": 1.9876,
      "step": 173127
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.2814041376113892,
      "learning_rate": 8.692784102413476e-05,
      "loss": 2.1713,
      "step": 173128
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.3435711860656738,
      "learning_rate": 8.692496145554895e-05,
      "loss": 2.1486,
      "step": 173129
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.2501596212387085,
      "learning_rate": 8.692208192657801e-05,
      "loss": 1.9383,
      "step": 173130
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.2235057353973389,
      "learning_rate": 8.691920243722212e-05,
      "loss": 2.087,
      "step": 173131
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.3258752822875977,
      "learning_rate": 8.691632298748205e-05,
      "loss": 2.2102,
      "step": 173132
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.300806999206543,
      "learning_rate": 8.691344357735814e-05,
      "loss": 1.9683,
      "step": 173133
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.1633245944976807,
      "learning_rate": 8.691056420685115e-05,
      "loss": 2.048,
      "step": 173134
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.2886289358139038,
      "learning_rate": 8.690768487596143e-05,
      "loss": 1.9488,
      "step": 173135
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.2992056608200073,
      "learning_rate": 8.69048055846898e-05,
      "loss": 1.9364,
      "step": 173136
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.3050613403320312,
      "learning_rate": 8.690192633303636e-05,
      "loss": 2.0237,
      "step": 173137
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.3018074035644531,
      "learning_rate": 8.689904712100197e-05,
      "loss": 1.9157,
      "step": 173138
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.1880463361740112,
      "learning_rate": 8.689616794858699e-05,
      "loss": 1.9152,
      "step": 173139
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.293503761291504,
      "learning_rate": 8.689328881579212e-05,
      "loss": 2.0028,
      "step": 173140
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.2941666841506958,
      "learning_rate": 8.689040972261775e-05,
      "loss": 1.9506,
      "step": 173141
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.3917615413665771,
      "learning_rate": 8.688753066906462e-05,
      "loss": 1.9944,
      "step": 173142
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.1953024864196777,
      "learning_rate": 8.688465165513296e-05,
      "loss": 2.0713,
      "step": 173143
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.4441826343536377,
      "learning_rate": 8.688177268082355e-05,
      "loss": 2.0151,
      "step": 173144
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.2206109762191772,
      "learning_rate": 8.687889374613674e-05,
      "loss": 1.8516,
      "step": 173145
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.1612540483474731,
      "learning_rate": 8.687601485107327e-05,
      "loss": 2.0917,
      "step": 173146
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.2948986291885376,
      "learning_rate": 8.687313599563346e-05,
      "loss": 1.9868,
      "step": 173147
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.3032417297363281,
      "learning_rate": 8.687025717981816e-05,
      "loss": 1.9187,
      "step": 173148
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.2359082698822021,
      "learning_rate": 8.686737840362753e-05,
      "loss": 1.9369,
      "step": 173149
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.610897421836853,
      "learning_rate": 8.686449966706234e-05,
      "loss": 1.8938,
      "step": 173150
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.3138351440429688,
      "learning_rate": 8.686162097012301e-05,
      "loss": 2.1469,
      "step": 173151
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.2367455959320068,
      "learning_rate": 8.685874231281019e-05,
      "loss": 1.8408,
      "step": 173152
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.2705281972885132,
      "learning_rate": 8.685586369512427e-05,
      "loss": 2.2357,
      "step": 173153
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.349570631980896,
      "learning_rate": 8.685298511706595e-05,
      "loss": 2.1307,
      "step": 173154
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.2579176425933838,
      "learning_rate": 8.68501065786357e-05,
      "loss": 2.001,
      "step": 173155
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.2586050033569336,
      "learning_rate": 8.684722807983402e-05,
      "loss": 2.1618,
      "step": 173156
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.31374192237854,
      "learning_rate": 8.68443496206614e-05,
      "loss": 1.9049,
      "step": 173157
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.3857157230377197,
      "learning_rate": 8.684147120111848e-05,
      "loss": 2.0935,
      "step": 173158
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.5016391277313232,
      "learning_rate": 8.68385928212057e-05,
      "loss": 2.05,
      "step": 173159
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.2141145467758179,
      "learning_rate": 8.683571448092374e-05,
      "loss": 1.968,
      "step": 173160
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.2822932004928589,
      "learning_rate": 8.683283618027305e-05,
      "loss": 2.1137,
      "step": 173161
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.139412760734558,
      "learning_rate": 8.682995791925416e-05,
      "loss": 1.8515,
      "step": 173162
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.287479043006897,
      "learning_rate": 8.68270796978675e-05,
      "loss": 2.2008,
      "step": 173163
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.1996022462844849,
      "learning_rate": 8.682420151611381e-05,
      "loss": 1.9626,
      "step": 173164
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.2530525922775269,
      "learning_rate": 8.682132337399341e-05,
      "loss": 2.0228,
      "step": 173165
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.3174947500228882,
      "learning_rate": 8.681844527150708e-05,
      "loss": 1.8689,
      "step": 173166
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.1948750019073486,
      "learning_rate": 8.68155672086552e-05,
      "loss": 2.0046,
      "step": 173167
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.24690842628479,
      "learning_rate": 8.681268918543833e-05,
      "loss": 2.006,
      "step": 173168
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.1838185787200928,
      "learning_rate": 8.680981120185691e-05,
      "loss": 1.8667,
      "step": 173169
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.288456678390503,
      "learning_rate": 8.680693325791169e-05,
      "loss": 1.8935,
      "step": 173170
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.3192126750946045,
      "learning_rate": 8.680405535360296e-05,
      "loss": 2.0425,
      "step": 173171
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.38924241065979,
      "learning_rate": 8.680117748893148e-05,
      "loss": 2.0589,
      "step": 173172
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.3607182502746582,
      "learning_rate": 8.67982996638977e-05,
      "loss": 2.1408,
      "step": 173173
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.227717638015747,
      "learning_rate": 8.679542187850206e-05,
      "loss": 2.0567,
      "step": 173174
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.1758767366409302,
      "learning_rate": 8.679254413274523e-05,
      "loss": 2.0143,
      "step": 173175
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.3775238990783691,
      "learning_rate": 8.678966642662771e-05,
      "loss": 2.0054,
      "step": 173176
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.3338671922683716,
      "learning_rate": 8.678678876014993e-05,
      "loss": 1.976,
      "step": 173177
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.2807464599609375,
      "learning_rate": 8.678391113331261e-05,
      "loss": 1.812,
      "step": 173178
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.3137704133987427,
      "learning_rate": 8.678103354611616e-05,
      "loss": 2.1065,
      "step": 173179
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.3799338340759277,
      "learning_rate": 8.677815599856105e-05,
      "loss": 2.0212,
      "step": 173180
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.2700426578521729,
      "learning_rate": 8.6775278490648e-05,
      "loss": 2.1135,
      "step": 173181
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.3234598636627197,
      "learning_rate": 8.677240102237748e-05,
      "loss": 1.9165,
      "step": 173182
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.4618747234344482,
      "learning_rate": 8.676952359374988e-05,
      "loss": 1.9882,
      "step": 173183
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.3285895586013794,
      "learning_rate": 8.676664620476593e-05,
      "loss": 2.2374,
      "step": 173184
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.3370857238769531,
      "learning_rate": 8.676376885542611e-05,
      "loss": 2.0225,
      "step": 173185
    },
    {
      "epoch": 2.25,
      "grad_norm": 1.2551372051239014,
      "learning_rate": 8.676089154573084e-05,
      "loss": 1.9116,
      "step": 173186
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.3089191913604736,
      "learning_rate": 8.675801427568081e-05,
      "loss": 1.9461,
      "step": 173187
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.3507344722747803,
      "learning_rate": 8.675513704527642e-05,
      "loss": 2.1258,
      "step": 173188
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.3343636989593506,
      "learning_rate": 8.675225985451838e-05,
      "loss": 2.108,
      "step": 173189
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.4831295013427734,
      "learning_rate": 8.674938270340708e-05,
      "loss": 2.0882,
      "step": 173190
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.1682308912277222,
      "learning_rate": 8.674650559194312e-05,
      "loss": 1.7523,
      "step": 173191
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.258286952972412,
      "learning_rate": 8.674362852012693e-05,
      "loss": 1.8705,
      "step": 173192
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.324163556098938,
      "learning_rate": 8.67407514879592e-05,
      "loss": 2.1604,
      "step": 173193
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.2726726531982422,
      "learning_rate": 8.67378744954403e-05,
      "loss": 2.2364,
      "step": 173194
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.2385693788528442,
      "learning_rate": 8.673499754257094e-05,
      "loss": 2.2752,
      "step": 173195
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.3987458944320679,
      "learning_rate": 8.673212062935149e-05,
      "loss": 2.0789,
      "step": 173196
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.314163327217102,
      "learning_rate": 8.672924375578275e-05,
      "loss": 1.9398,
      "step": 173197
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.2619824409484863,
      "learning_rate": 8.672636692186486e-05,
      "loss": 1.8308,
      "step": 173198
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.3597331047058105,
      "learning_rate": 8.672349012759869e-05,
      "loss": 2.1364,
      "step": 173199
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.1969547271728516,
      "learning_rate": 8.672061337298453e-05,
      "loss": 2.0395,
      "step": 173200
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.277884840965271,
      "learning_rate": 8.671773665802313e-05,
      "loss": 2.0675,
      "step": 173201
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.5595885515213013,
      "learning_rate": 8.671485998271487e-05,
      "loss": 2.0525,
      "step": 173202
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.2572126388549805,
      "learning_rate": 8.67119833470605e-05,
      "loss": 1.8636,
      "step": 173203
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.366483211517334,
      "learning_rate": 8.670910675106022e-05,
      "loss": 1.945,
      "step": 173204
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.1369293928146362,
      "learning_rate": 8.670623019471483e-05,
      "loss": 1.8061,
      "step": 173205
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.2264111042022705,
      "learning_rate": 8.670335367802469e-05,
      "loss": 2.0772,
      "step": 173206
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.304442048072815,
      "learning_rate": 8.67004772009905e-05,
      "loss": 1.9886,
      "step": 173207
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.4491578340530396,
      "learning_rate": 8.669760076361264e-05,
      "loss": 1.9669,
      "step": 173208
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.177616000175476,
      "learning_rate": 8.66947243658919e-05,
      "loss": 2.1104,
      "step": 173209
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.2811434268951416,
      "learning_rate": 8.669184800782847e-05,
      "loss": 1.8663,
      "step": 173210
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.2858318090438843,
      "learning_rate": 8.668897168942312e-05,
      "loss": 1.8809,
      "step": 173211
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.248304843902588,
      "learning_rate": 8.668609541067624e-05,
      "loss": 2.047,
      "step": 173212
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.2883824110031128,
      "learning_rate": 8.668321917158854e-05,
      "loss": 2.0108,
      "step": 173213
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.4033396244049072,
      "learning_rate": 8.668034297216035e-05,
      "loss": 1.9948,
      "step": 173214
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.3437492847442627,
      "learning_rate": 8.667746681239249e-05,
      "loss": 2.13,
      "step": 173215
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.2429367303848267,
      "learning_rate": 8.667459069228516e-05,
      "loss": 1.9722,
      "step": 173216
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.2131872177124023,
      "learning_rate": 8.667171461183912e-05,
      "loss": 2.1847,
      "step": 173217
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.2771525382995605,
      "learning_rate": 8.666883857105475e-05,
      "loss": 1.938,
      "step": 173218
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.4198664426803589,
      "learning_rate": 8.666596256993273e-05,
      "loss": 2.0247,
      "step": 173219
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.3864020109176636,
      "learning_rate": 8.666308660847349e-05,
      "loss": 2.0389,
      "step": 173220
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.3135603666305542,
      "learning_rate": 8.666021068667767e-05,
      "loss": 1.9609,
      "step": 173221
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.3194363117218018,
      "learning_rate": 8.665733480454578e-05,
      "loss": 2.0577,
      "step": 173222
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.2775622606277466,
      "learning_rate": 8.665445896207828e-05,
      "loss": 2.0254,
      "step": 173223
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.2050312757492065,
      "learning_rate": 8.665158315927568e-05,
      "loss": 1.9884,
      "step": 173224
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.2317146062850952,
      "learning_rate": 8.664870739613866e-05,
      "loss": 2.0423,
      "step": 173225
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.4428731203079224,
      "learning_rate": 8.664583167266758e-05,
      "loss": 2.1593,
      "step": 173226
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.678244709968567,
      "learning_rate": 8.664295598886313e-05,
      "loss": 1.9783,
      "step": 173227
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.2906687259674072,
      "learning_rate": 8.664008034472581e-05,
      "loss": 1.8954,
      "step": 173228
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.567698359489441,
      "learning_rate": 8.663720474025614e-05,
      "loss": 2.112,
      "step": 173229
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.628373146057129,
      "learning_rate": 8.663432917545454e-05,
      "loss": 2.0888,
      "step": 173230
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.678845763206482,
      "learning_rate": 8.663145365032173e-05,
      "loss": 2.0818,
      "step": 173231
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.3057169914245605,
      "learning_rate": 8.662857816485806e-05,
      "loss": 2.0185,
      "step": 173232
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.3937053680419922,
      "learning_rate": 8.662570271906429e-05,
      "loss": 2.1049,
      "step": 173233
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.3126188516616821,
      "learning_rate": 8.662282731294078e-05,
      "loss": 1.9621,
      "step": 173234
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.3079240322113037,
      "learning_rate": 8.661995194648814e-05,
      "loss": 2.0004,
      "step": 173235
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.375314712524414,
      "learning_rate": 8.661707661970681e-05,
      "loss": 2.0932,
      "step": 173236
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.4200046062469482,
      "learning_rate": 8.661420133259744e-05,
      "loss": 2.1209,
      "step": 173237
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.5096983909606934,
      "learning_rate": 8.661132608516046e-05,
      "loss": 2.0399,
      "step": 173238
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.3152941465377808,
      "learning_rate": 8.660845087739656e-05,
      "loss": 2.0421,
      "step": 173239
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.2187873125076294,
      "learning_rate": 8.660557570930614e-05,
      "loss": 1.8642,
      "step": 173240
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.286476492881775,
      "learning_rate": 8.660270058088979e-05,
      "loss": 1.853,
      "step": 173241
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.6404393911361694,
      "learning_rate": 8.659982549214795e-05,
      "loss": 1.9416,
      "step": 173242
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.7364147901535034,
      "learning_rate": 8.659695044308129e-05,
      "loss": 1.9953,
      "step": 173243
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.2867488861083984,
      "learning_rate": 8.65940754336902e-05,
      "loss": 2.0938,
      "step": 173244
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.2842004299163818,
      "learning_rate": 8.659120046397541e-05,
      "loss": 2.1011,
      "step": 173245
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.6047855615615845,
      "learning_rate": 8.658832553393734e-05,
      "loss": 2.0781,
      "step": 173246
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.254414677619934,
      "learning_rate": 8.658545064357653e-05,
      "loss": 1.9509,
      "step": 173247
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.2298015356063843,
      "learning_rate": 8.658257579289343e-05,
      "loss": 1.9344,
      "step": 173248
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.3539094924926758,
      "learning_rate": 8.657970098188874e-05,
      "loss": 2.1119,
      "step": 173249
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.4424927234649658,
      "learning_rate": 8.657682621056282e-05,
      "loss": 2.0217,
      "step": 173250
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.331006646156311,
      "learning_rate": 8.65739514789164e-05,
      "loss": 1.9918,
      "step": 173251
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.3460891246795654,
      "learning_rate": 8.65710767869499e-05,
      "loss": 2.2354,
      "step": 173252
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.3754956722259521,
      "learning_rate": 8.656820213466384e-05,
      "loss": 1.8791,
      "step": 173253
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.3620755672454834,
      "learning_rate": 8.656532752205873e-05,
      "loss": 2.3385,
      "step": 173254
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.2298182249069214,
      "learning_rate": 8.656245294913519e-05,
      "loss": 1.9264,
      "step": 173255
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.2660553455352783,
      "learning_rate": 8.65595784158937e-05,
      "loss": 2.0503,
      "step": 173256
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.2313543558120728,
      "learning_rate": 8.655670392233485e-05,
      "loss": 1.8397,
      "step": 173257
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.2509747743606567,
      "learning_rate": 8.655382946845915e-05,
      "loss": 2.0286,
      "step": 173258
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.2467224597930908,
      "learning_rate": 8.655095505426705e-05,
      "loss": 1.7381,
      "step": 173259
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.4941737651824951,
      "learning_rate": 8.654808067975925e-05,
      "loss": 2.1838,
      "step": 173260
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.3293249607086182,
      "learning_rate": 8.654520634493615e-05,
      "loss": 1.9435,
      "step": 173261
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.26409113407135,
      "learning_rate": 8.654233204979829e-05,
      "loss": 2.0159,
      "step": 173262
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.285981297492981,
      "learning_rate": 8.653945779434627e-05,
      "loss": 1.9857,
      "step": 173263
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.2011268138885498,
      "learning_rate": 8.653658357858065e-05,
      "loss": 2.016,
      "step": 173264
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.3761554956436157,
      "learning_rate": 8.653370940250178e-05,
      "loss": 2.2275,
      "step": 173265
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.286085844039917,
      "learning_rate": 8.653083526611043e-05,
      "loss": 1.9039,
      "step": 173266
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.242324709892273,
      "learning_rate": 8.652796116940704e-05,
      "loss": 2.0569,
      "step": 173267
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.2520073652267456,
      "learning_rate": 8.652508711239202e-05,
      "loss": 2.2115,
      "step": 173268
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.3896980285644531,
      "learning_rate": 8.652221309506612e-05,
      "loss": 2.0865,
      "step": 173269
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.3893533945083618,
      "learning_rate": 8.651933911742975e-05,
      "loss": 1.9878,
      "step": 173270
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.2507758140563965,
      "learning_rate": 8.651646517948339e-05,
      "loss": 1.8176,
      "step": 173271
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.3520421981811523,
      "learning_rate": 8.651359128122776e-05,
      "loss": 1.9518,
      "step": 173272
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.1543880701065063,
      "learning_rate": 8.651071742266317e-05,
      "loss": 1.9368,
      "step": 173273
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.2206237316131592,
      "learning_rate": 8.650784360379033e-05,
      "loss": 1.9585,
      "step": 173274
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.1347633600234985,
      "learning_rate": 8.650496982460974e-05,
      "loss": 1.9244,
      "step": 173275
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.2719119787216187,
      "learning_rate": 8.650209608512193e-05,
      "loss": 2.0667,
      "step": 173276
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.5630877017974854,
      "learning_rate": 8.649922238532728e-05,
      "loss": 2.1435,
      "step": 173277
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.4639551639556885,
      "learning_rate": 8.649634872522656e-05,
      "loss": 2.0582,
      "step": 173278
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.302163004875183,
      "learning_rate": 8.64934751048201e-05,
      "loss": 2.0648,
      "step": 173279
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.3034013509750366,
      "learning_rate": 8.649060152410865e-05,
      "loss": 2.083,
      "step": 173280
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.2683603763580322,
      "learning_rate": 8.648772798309251e-05,
      "loss": 2.0149,
      "step": 173281
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.6597669124603271,
      "learning_rate": 8.648485448177252e-05,
      "loss": 2.1143,
      "step": 173282
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.9641765356063843,
      "learning_rate": 8.648198102014885e-05,
      "loss": 1.7118,
      "step": 173283
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.2893831729888916,
      "learning_rate": 8.647910759822232e-05,
      "loss": 2.1112,
      "step": 173284
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.2851121425628662,
      "learning_rate": 8.647623421599323e-05,
      "loss": 2.0987,
      "step": 173285
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.3146404027938843,
      "learning_rate": 8.647336087346232e-05,
      "loss": 1.9769,
      "step": 173286
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.2906382083892822,
      "learning_rate": 8.647048757062999e-05,
      "loss": 1.9598,
      "step": 173287
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.496532678604126,
      "learning_rate": 8.646761430749692e-05,
      "loss": 2.1247,
      "step": 173288
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.540130376815796,
      "learning_rate": 8.64647410840635e-05,
      "loss": 2.1083,
      "step": 173289
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.237520694732666,
      "learning_rate": 8.646186790033038e-05,
      "loss": 1.9238,
      "step": 173290
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.321158766746521,
      "learning_rate": 8.645899475629789e-05,
      "loss": 2.0536,
      "step": 173291
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.2791944742202759,
      "learning_rate": 8.645612165196681e-05,
      "loss": 2.0716,
      "step": 173292
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.4023631811141968,
      "learning_rate": 8.645324858733746e-05,
      "loss": 2.1835,
      "step": 173293
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.4084091186523438,
      "learning_rate": 8.64503755624106e-05,
      "loss": 1.9941,
      "step": 173294
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.2845228910446167,
      "learning_rate": 8.644750257718665e-05,
      "loss": 1.866,
      "step": 173295
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.3938517570495605,
      "learning_rate": 8.644462963166613e-05,
      "loss": 2.126,
      "step": 173296
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.4020848274230957,
      "learning_rate": 8.644175672584948e-05,
      "loss": 1.982,
      "step": 173297
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.291213035583496,
      "learning_rate": 8.643888385973745e-05,
      "loss": 2.0139,
      "step": 173298
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.3129020929336548,
      "learning_rate": 8.643601103333036e-05,
      "loss": 2.0685,
      "step": 173299
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.3016817569732666,
      "learning_rate": 8.643313824662892e-05,
      "loss": 2.0363,
      "step": 173300
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.3426384925842285,
      "learning_rate": 8.643026549963364e-05,
      "loss": 1.9912,
      "step": 173301
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.3919137716293335,
      "learning_rate": 8.642739279234496e-05,
      "loss": 2.2528,
      "step": 173302
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.2043930292129517,
      "learning_rate": 8.642452012476338e-05,
      "loss": 2.1518,
      "step": 173303
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.7533012628555298,
      "learning_rate": 8.642164749688965e-05,
      "loss": 2.0789,
      "step": 173304
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.34463369846344,
      "learning_rate": 8.6418774908724e-05,
      "loss": 2.0831,
      "step": 173305
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.170480489730835,
      "learning_rate": 8.641590236026727e-05,
      "loss": 2.027,
      "step": 173306
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.3555625677108765,
      "learning_rate": 8.641302985151984e-05,
      "loss": 2.0135,
      "step": 173307
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.2819159030914307,
      "learning_rate": 8.641015738248228e-05,
      "loss": 2.0204,
      "step": 173308
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.507677674293518,
      "learning_rate": 8.640728495315499e-05,
      "loss": 2.0234,
      "step": 173309
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.4089635610580444,
      "learning_rate": 8.640441256353873e-05,
      "loss": 2.1334,
      "step": 173310
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.3288646936416626,
      "learning_rate": 8.64015402136338e-05,
      "loss": 2.2032,
      "step": 173311
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.307147741317749,
      "learning_rate": 8.639866790344094e-05,
      "loss": 1.9882,
      "step": 173312
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.2881730794906616,
      "learning_rate": 8.639579563296065e-05,
      "loss": 1.8812,
      "step": 173313
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.3642058372497559,
      "learning_rate": 8.63929234021934e-05,
      "loss": 1.9931,
      "step": 173314
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.2158948183059692,
      "learning_rate": 8.63900512111396e-05,
      "loss": 2.4138,
      "step": 173315
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.2872753143310547,
      "learning_rate": 8.638717905980004e-05,
      "loss": 2.07,
      "step": 173316
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.181594729423523,
      "learning_rate": 8.638430694817508e-05,
      "loss": 1.9739,
      "step": 173317
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.358540654182434,
      "learning_rate": 8.638143487626534e-05,
      "loss": 2.0115,
      "step": 173318
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.3671189546585083,
      "learning_rate": 8.637856284407138e-05,
      "loss": 2.1204,
      "step": 173319
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.3532105684280396,
      "learning_rate": 8.637569085159364e-05,
      "loss": 2.1544,
      "step": 173320
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.1737947463989258,
      "learning_rate": 8.63728188988326e-05,
      "loss": 1.9684,
      "step": 173321
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.3966283798217773,
      "learning_rate": 8.636994698578898e-05,
      "loss": 2.1328,
      "step": 173322
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.3024476766586304,
      "learning_rate": 8.636707511246313e-05,
      "loss": 1.8435,
      "step": 173323
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.3444583415985107,
      "learning_rate": 8.636420327885579e-05,
      "loss": 1.826,
      "step": 173324
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.1954443454742432,
      "learning_rate": 8.636133148496736e-05,
      "loss": 1.6699,
      "step": 173325
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.3242835998535156,
      "learning_rate": 8.63584597307984e-05,
      "loss": 2.075,
      "step": 173326
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.253039002418518,
      "learning_rate": 8.635558801634933e-05,
      "loss": 2.0797,
      "step": 173327
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.2096245288848877,
      "learning_rate": 8.635271634162087e-05,
      "loss": 1.9142,
      "step": 173328
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.256393551826477,
      "learning_rate": 8.63498447066134e-05,
      "loss": 2.0782,
      "step": 173329
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.3120259046554565,
      "learning_rate": 8.63469731113276e-05,
      "loss": 2.0065,
      "step": 173330
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.2675873041152954,
      "learning_rate": 8.634410155576397e-05,
      "loss": 2.074,
      "step": 173331
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.3397705554962158,
      "learning_rate": 8.634123003992296e-05,
      "loss": 1.8642,
      "step": 173332
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.1789884567260742,
      "learning_rate": 8.633835856380507e-05,
      "loss": 1.9327,
      "step": 173333
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.2798352241516113,
      "learning_rate": 8.633548712741102e-05,
      "loss": 2.105,
      "step": 173334
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.228195309638977,
      "learning_rate": 8.633261573074115e-05,
      "loss": 1.9033,
      "step": 173335
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.2889102697372437,
      "learning_rate": 8.632974437379614e-05,
      "loss": 1.7568,
      "step": 173336
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.2961398363113403,
      "learning_rate": 8.632687305657648e-05,
      "loss": 1.8491,
      "step": 173337
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.2472907304763794,
      "learning_rate": 8.632400177908268e-05,
      "loss": 1.9887,
      "step": 173338
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.3324114084243774,
      "learning_rate": 8.632113054131521e-05,
      "loss": 2.0157,
      "step": 173339
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.4082176685333252,
      "learning_rate": 8.631825934327476e-05,
      "loss": 1.9044,
      "step": 173340
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.2749438285827637,
      "learning_rate": 8.631538818496169e-05,
      "loss": 1.9466,
      "step": 173341
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.1491942405700684,
      "learning_rate": 8.631251706637668e-05,
      "loss": 2.0331,
      "step": 173342
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.2745229005813599,
      "learning_rate": 8.630964598752025e-05,
      "loss": 2.0076,
      "step": 173343
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.2896032333374023,
      "learning_rate": 8.630677494839287e-05,
      "loss": 2.2795,
      "step": 173344
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.3334256410598755,
      "learning_rate": 8.630390394899499e-05,
      "loss": 2.0657,
      "step": 173345
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.2875497341156006,
      "learning_rate": 8.630103298932736e-05,
      "loss": 2.0965,
      "step": 173346
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.201980471611023,
      "learning_rate": 8.62981620693903e-05,
      "loss": 2.2213,
      "step": 173347
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.218604326248169,
      "learning_rate": 8.629529118918454e-05,
      "loss": 1.971,
      "step": 173348
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.3059827089309692,
      "learning_rate": 8.629242034871053e-05,
      "loss": 1.8475,
      "step": 173349
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.7618703842163086,
      "learning_rate": 8.628954954796871e-05,
      "loss": 1.9889,
      "step": 173350
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.2686084508895874,
      "learning_rate": 8.628667878695977e-05,
      "loss": 2.0525,
      "step": 173351
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.3654898405075073,
      "learning_rate": 8.628380806568416e-05,
      "loss": 2.4038,
      "step": 173352
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.3064892292022705,
      "learning_rate": 8.628093738414236e-05,
      "loss": 2.079,
      "step": 173353
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.4017609357833862,
      "learning_rate": 8.627806674233502e-05,
      "loss": 2.14,
      "step": 173354
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.3018845319747925,
      "learning_rate": 8.627519614026258e-05,
      "loss": 1.8346,
      "step": 173355
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.3510397672653198,
      "learning_rate": 8.627232557792568e-05,
      "loss": 1.8688,
      "step": 173356
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.3335206508636475,
      "learning_rate": 8.62694550553248e-05,
      "loss": 2.0983,
      "step": 173357
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.4018993377685547,
      "learning_rate": 8.626658457246036e-05,
      "loss": 1.8714,
      "step": 173358
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.3348125219345093,
      "learning_rate": 8.626371412933307e-05,
      "loss": 1.949,
      "step": 173359
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.2567893266677856,
      "learning_rate": 8.626084372594343e-05,
      "loss": 1.8489,
      "step": 173360
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.301879644393921,
      "learning_rate": 8.625797336229183e-05,
      "loss": 2.0856,
      "step": 173361
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.4131659269332886,
      "learning_rate": 8.625510303837898e-05,
      "loss": 1.9818,
      "step": 173362
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.3172913789749146,
      "learning_rate": 8.625223275420534e-05,
      "loss": 2.1486,
      "step": 173363
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.5612876415252686,
      "learning_rate": 8.624936250977138e-05,
      "loss": 2.119,
      "step": 173364
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.416483759880066,
      "learning_rate": 8.624649230507777e-05,
      "loss": 2.2102,
      "step": 173365
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.2551263570785522,
      "learning_rate": 8.624362214012499e-05,
      "loss": 2.0322,
      "step": 173366
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.2137821912765503,
      "learning_rate": 8.624075201491346e-05,
      "loss": 2.0614,
      "step": 173367
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.4392777681350708,
      "learning_rate": 8.623788192944387e-05,
      "loss": 2.1258,
      "step": 173368
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.2727552652359009,
      "learning_rate": 8.623501188371672e-05,
      "loss": 2.0922,
      "step": 173369
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.3569729328155518,
      "learning_rate": 8.623214187773243e-05,
      "loss": 1.8178,
      "step": 173370
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.312472939491272,
      "learning_rate": 8.62292719114917e-05,
      "loss": 2.0425,
      "step": 173371
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.2662919759750366,
      "learning_rate": 8.622640198499489e-05,
      "loss": 2.0415,
      "step": 173372
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.3813602924346924,
      "learning_rate": 8.622353209824274e-05,
      "loss": 2.0579,
      "step": 173373
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.2359063625335693,
      "learning_rate": 8.622066225123565e-05,
      "loss": 1.9758,
      "step": 173374
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.4679934978485107,
      "learning_rate": 8.621779244397417e-05,
      "loss": 2.2212,
      "step": 173375
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.261704683303833,
      "learning_rate": 8.621492267645875e-05,
      "loss": 1.9976,
      "step": 173376
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.3277933597564697,
      "learning_rate": 8.62120529486901e-05,
      "loss": 2.1737,
      "step": 173377
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.3016165494918823,
      "learning_rate": 8.620918326066859e-05,
      "loss": 2.0749,
      "step": 173378
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.267337441444397,
      "learning_rate": 8.620631361239492e-05,
      "loss": 2.2162,
      "step": 173379
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.363663673400879,
      "learning_rate": 8.620344400386952e-05,
      "loss": 1.9423,
      "step": 173380
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.2991770505905151,
      "learning_rate": 8.620057443509295e-05,
      "loss": 2.0377,
      "step": 173381
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.2812590599060059,
      "learning_rate": 8.619770490606562e-05,
      "loss": 2.1692,
      "step": 173382
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.338253140449524,
      "learning_rate": 8.619483541678826e-05,
      "loss": 2.2047,
      "step": 173383
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.21996009349823,
      "learning_rate": 8.619196596726125e-05,
      "loss": 1.7976,
      "step": 173384
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.4819742441177368,
      "learning_rate": 8.618909655748526e-05,
      "loss": 2.1705,
      "step": 173385
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.3032121658325195,
      "learning_rate": 8.618622718746074e-05,
      "loss": 1.9353,
      "step": 173386
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.4254848957061768,
      "learning_rate": 8.618335785718826e-05,
      "loss": 2.1915,
      "step": 173387
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.3436089754104614,
      "learning_rate": 8.618048856666825e-05,
      "loss": 1.9161,
      "step": 173388
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.4266564846038818,
      "learning_rate": 8.617761931590136e-05,
      "loss": 1.9052,
      "step": 173389
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.3495879173278809,
      "learning_rate": 8.617475010488805e-05,
      "loss": 1.8149,
      "step": 173390
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.21809983253479,
      "learning_rate": 8.617188093362898e-05,
      "loss": 2.0582,
      "step": 173391
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.3094650506973267,
      "learning_rate": 8.616901180212457e-05,
      "loss": 2.2071,
      "step": 173392
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.3963936567306519,
      "learning_rate": 8.616614271037538e-05,
      "loss": 2.0562,
      "step": 173393
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.4641568660736084,
      "learning_rate": 8.616327365838187e-05,
      "loss": 1.8178,
      "step": 173394
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.2682973146438599,
      "learning_rate": 8.616040464614472e-05,
      "loss": 2.0305,
      "step": 173395
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.330631971359253,
      "learning_rate": 8.61575356736643e-05,
      "loss": 2.2295,
      "step": 173396
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.2966872453689575,
      "learning_rate": 8.615466674094131e-05,
      "loss": 2.2075,
      "step": 173397
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.2619398832321167,
      "learning_rate": 8.615179784797623e-05,
      "loss": 1.8566,
      "step": 173398
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.3968013525009155,
      "learning_rate": 8.614892899476953e-05,
      "loss": 2.0184,
      "step": 173399
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.491226077079773,
      "learning_rate": 8.61460601813217e-05,
      "loss": 1.8104,
      "step": 173400
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.2441284656524658,
      "learning_rate": 8.614319140763345e-05,
      "loss": 1.9784,
      "step": 173401
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.4156889915466309,
      "learning_rate": 8.614032267370515e-05,
      "loss": 1.943,
      "step": 173402
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.8335849046707153,
      "learning_rate": 8.613745397953748e-05,
      "loss": 1.9273,
      "step": 173403
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.4201856851577759,
      "learning_rate": 8.613458532513085e-05,
      "loss": 2.1998,
      "step": 173404
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.3904786109924316,
      "learning_rate": 8.61317167104859e-05,
      "loss": 1.943,
      "step": 173405
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.287125825881958,
      "learning_rate": 8.612884813560297e-05,
      "loss": 1.957,
      "step": 173406
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.3043712377548218,
      "learning_rate": 8.612597960048284e-05,
      "loss": 1.9887,
      "step": 173407
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.4202419519424438,
      "learning_rate": 8.612311110512582e-05,
      "loss": 2.1998,
      "step": 173408
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.3842524290084839,
      "learning_rate": 8.612024264953266e-05,
      "loss": 1.9001,
      "step": 173409
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.2464523315429688,
      "learning_rate": 8.611737423370377e-05,
      "loss": 2.2503,
      "step": 173410
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.3357255458831787,
      "learning_rate": 8.611450585763971e-05,
      "loss": 2.0822,
      "step": 173411
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.2161718606948853,
      "learning_rate": 8.61116375213409e-05,
      "loss": 2.145,
      "step": 173412
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.6405562162399292,
      "learning_rate": 8.610876922480803e-05,
      "loss": 2.1303,
      "step": 173413
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.3161312341690063,
      "learning_rate": 8.610590096804154e-05,
      "loss": 2.1302,
      "step": 173414
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.2089650630950928,
      "learning_rate": 8.610303275104207e-05,
      "loss": 1.9509,
      "step": 173415
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.3039264678955078,
      "learning_rate": 8.610016457381008e-05,
      "loss": 2.164,
      "step": 173416
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.345470666885376,
      "learning_rate": 8.609729643634612e-05,
      "loss": 2.0022,
      "step": 173417
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.3857945203781128,
      "learning_rate": 8.60944283386506e-05,
      "loss": 2.0477,
      "step": 173418
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.240723729133606,
      "learning_rate": 8.60915602807243e-05,
      "loss": 2.1316,
      "step": 173419
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.2201262712478638,
      "learning_rate": 8.60886922625675e-05,
      "loss": 2.2414,
      "step": 173420
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.2494620084762573,
      "learning_rate": 8.608582428418091e-05,
      "loss": 2.0085,
      "step": 173421
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.2195684909820557,
      "learning_rate": 8.608295634556492e-05,
      "loss": 1.822,
      "step": 173422
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.289492130279541,
      "learning_rate": 8.608008844672035e-05,
      "loss": 1.8366,
      "step": 173423
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.332596778869629,
      "learning_rate": 8.607722058764736e-05,
      "loss": 2.0391,
      "step": 173424
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.2708748579025269,
      "learning_rate": 8.607435276834674e-05,
      "loss": 1.9422,
      "step": 173425
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.290042757987976,
      "learning_rate": 8.607148498881882e-05,
      "loss": 1.9673,
      "step": 173426
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.32820725440979,
      "learning_rate": 8.606861724906435e-05,
      "loss": 2.176,
      "step": 173427
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.2708874940872192,
      "learning_rate": 8.606574954908367e-05,
      "loss": 2.2829,
      "step": 173428
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.2890667915344238,
      "learning_rate": 8.606288188887757e-05,
      "loss": 2.0355,
      "step": 173429
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.2771267890930176,
      "learning_rate": 8.606001426844626e-05,
      "loss": 1.8891,
      "step": 173430
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.3653557300567627,
      "learning_rate": 8.605714668779053e-05,
      "loss": 1.9598,
      "step": 173431
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.3092565536499023,
      "learning_rate": 8.605427914691072e-05,
      "loss": 2.0389,
      "step": 173432
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.292148232460022,
      "learning_rate": 8.605141164580755e-05,
      "loss": 1.7992,
      "step": 173433
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.303378701210022,
      "learning_rate": 8.604854418448137e-05,
      "loss": 1.992,
      "step": 173434
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.1516307592391968,
      "learning_rate": 8.604567676293286e-05,
      "loss": 1.9448,
      "step": 173435
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.3814098834991455,
      "learning_rate": 8.604280938116254e-05,
      "loss": 2.1973,
      "step": 173436
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.3884806632995605,
      "learning_rate": 8.603994203917087e-05,
      "loss": 2.1328,
      "step": 173437
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.2974475622177124,
      "learning_rate": 8.603707473695834e-05,
      "loss": 2.1883,
      "step": 173438
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.339145302772522,
      "learning_rate": 8.603420747452566e-05,
      "loss": 1.9974,
      "step": 173439
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.1693263053894043,
      "learning_rate": 8.603134025187316e-05,
      "loss": 2.0196,
      "step": 173440
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.2806992530822754,
      "learning_rate": 8.602847306900157e-05,
      "loss": 2.0413,
      "step": 173441
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.214290976524353,
      "learning_rate": 8.602560592591128e-05,
      "loss": 2.119,
      "step": 173442
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.3881421089172363,
      "learning_rate": 8.602273882260283e-05,
      "loss": 2.1871,
      "step": 173443
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.1806325912475586,
      "learning_rate": 8.601987175907686e-05,
      "loss": 2.0621,
      "step": 173444
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.2827091217041016,
      "learning_rate": 8.601700473533386e-05,
      "loss": 2.1156,
      "step": 173445
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.3710973262786865,
      "learning_rate": 8.601413775137421e-05,
      "loss": 2.1393,
      "step": 173446
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.2893904447555542,
      "learning_rate": 8.601127080719869e-05,
      "loss": 2.0264,
      "step": 173447
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.285236120223999,
      "learning_rate": 8.60084039028077e-05,
      "loss": 1.93,
      "step": 173448
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.216230869293213,
      "learning_rate": 8.600553703820168e-05,
      "loss": 2.3133,
      "step": 173449
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.3972679376602173,
      "learning_rate": 8.600267021338138e-05,
      "loss": 2.0528,
      "step": 173450
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.4146268367767334,
      "learning_rate": 8.599980342834726e-05,
      "loss": 2.2551,
      "step": 173451
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.3043568134307861,
      "learning_rate": 8.599693668309968e-05,
      "loss": 1.9445,
      "step": 173452
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.2273119688034058,
      "learning_rate": 8.59940699776394e-05,
      "loss": 1.8877,
      "step": 173453
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.5663739442825317,
      "learning_rate": 8.599120331196688e-05,
      "loss": 2.1656,
      "step": 173454
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.2324161529541016,
      "learning_rate": 8.598833668608251e-05,
      "loss": 1.9463,
      "step": 173455
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.2549320459365845,
      "learning_rate": 8.598547009998709e-05,
      "loss": 2.0405,
      "step": 173456
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.3458225727081299,
      "learning_rate": 8.59826035536809e-05,
      "loss": 2.041,
      "step": 173457
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.240570306777954,
      "learning_rate": 8.597973704716467e-05,
      "loss": 2.0268,
      "step": 173458
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.693831443786621,
      "learning_rate": 8.597687058043883e-05,
      "loss": 2.2877,
      "step": 173459
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.3428030014038086,
      "learning_rate": 8.597400415350394e-05,
      "loss": 2.2986,
      "step": 173460
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.2644144296646118,
      "learning_rate": 8.597113776636045e-05,
      "loss": 2.1136,
      "step": 173461
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.2257936000823975,
      "learning_rate": 8.596827141900904e-05,
      "loss": 2.01,
      "step": 173462
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.2738537788391113,
      "learning_rate": 8.596540511145006e-05,
      "loss": 1.8596,
      "step": 173463
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.2725801467895508,
      "learning_rate": 8.59625388436843e-05,
      "loss": 1.8875,
      "step": 173464
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.3195418119430542,
      "learning_rate": 8.59596726157121e-05,
      "loss": 2.4098,
      "step": 173465
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.2373549938201904,
      "learning_rate": 8.595680642753405e-05,
      "loss": 2.0789,
      "step": 173466
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.3327958583831787,
      "learning_rate": 8.595394027915058e-05,
      "loss": 1.9079,
      "step": 173467
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.1166787147521973,
      "learning_rate": 8.595107417056238e-05,
      "loss": 1.9689,
      "step": 173468
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.290017008781433,
      "learning_rate": 8.594820810176985e-05,
      "loss": 2.1419,
      "step": 173469
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.4137487411499023,
      "learning_rate": 8.594534207277366e-05,
      "loss": 2.1198,
      "step": 173470
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.4144176244735718,
      "learning_rate": 8.59424760835743e-05,
      "loss": 2.0258,
      "step": 173471
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.3421146869659424,
      "learning_rate": 8.593961013417226e-05,
      "loss": 1.7912,
      "step": 173472
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.2515075206756592,
      "learning_rate": 8.5936744224568e-05,
      "loss": 2.1594,
      "step": 173473
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.3357290029525757,
      "learning_rate": 8.593387835476223e-05,
      "loss": 2.0974,
      "step": 173474
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.383001685142517,
      "learning_rate": 8.59310125247553e-05,
      "loss": 1.8972,
      "step": 173475
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.50426185131073,
      "learning_rate": 8.592814673454791e-05,
      "loss": 1.9611,
      "step": 173476
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.3560235500335693,
      "learning_rate": 8.592528098414054e-05,
      "loss": 2.3054,
      "step": 173477
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.2362924814224243,
      "learning_rate": 8.59224152735337e-05,
      "loss": 1.8368,
      "step": 173478
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.3077253103256226,
      "learning_rate": 8.59195496027278e-05,
      "loss": 2.1079,
      "step": 173479
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.4114854335784912,
      "learning_rate": 8.591668397172363e-05,
      "loss": 2.0031,
      "step": 173480
    },
    {
      "epoch": 2.26,
      "grad_norm": 2.037158966064453,
      "learning_rate": 8.591381838052149e-05,
      "loss": 2.1718,
      "step": 173481
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.365800142288208,
      "learning_rate": 8.591095282912209e-05,
      "loss": 2.0097,
      "step": 173482
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.294517159461975,
      "learning_rate": 8.590808731752588e-05,
      "loss": 2.109,
      "step": 173483
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.8014988899230957,
      "learning_rate": 8.590522184573345e-05,
      "loss": 1.6937,
      "step": 173484
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.4667654037475586,
      "learning_rate": 8.590235641374512e-05,
      "loss": 2.0527,
      "step": 173485
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.3095028400421143,
      "learning_rate": 8.589949102156169e-05,
      "loss": 2.1332,
      "step": 173486
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.2796030044555664,
      "learning_rate": 8.589662566918353e-05,
      "loss": 1.8599,
      "step": 173487
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.1622508764266968,
      "learning_rate": 8.589376035661129e-05,
      "loss": 1.9584,
      "step": 173488
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.3765071630477905,
      "learning_rate": 8.589089508384537e-05,
      "loss": 2.2249,
      "step": 173489
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.2134774923324585,
      "learning_rate": 8.588802985088655e-05,
      "loss": 2.0069,
      "step": 173490
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.4335449934005737,
      "learning_rate": 8.5885164657735e-05,
      "loss": 2.2369,
      "step": 173491
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.3076354265213013,
      "learning_rate": 8.588229950439153e-05,
      "loss": 1.9629,
      "step": 173492
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.2479115724563599,
      "learning_rate": 8.587943439085648e-05,
      "loss": 2.0345,
      "step": 173493
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.3073201179504395,
      "learning_rate": 8.58765693171306e-05,
      "loss": 2.0316,
      "step": 173494
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.2609003782272339,
      "learning_rate": 8.58737042832142e-05,
      "loss": 1.9137,
      "step": 173495
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.2900773286819458,
      "learning_rate": 8.58708392891081e-05,
      "loss": 2.1602,
      "step": 173496
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.4352116584777832,
      "learning_rate": 8.586797433481247e-05,
      "loss": 1.977,
      "step": 173497
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.2540730237960815,
      "learning_rate": 8.586510942032812e-05,
      "loss": 2.1108,
      "step": 173498
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.4164196252822876,
      "learning_rate": 8.586224454565544e-05,
      "loss": 2.031,
      "step": 173499
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.227074384689331,
      "learning_rate": 8.585937971079507e-05,
      "loss": 1.9705,
      "step": 173500
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.5591551065444946,
      "learning_rate": 8.585651491574738e-05,
      "loss": 2.0938,
      "step": 173501
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.2031135559082031,
      "learning_rate": 8.585365016051317e-05,
      "loss": 1.9805,
      "step": 173502
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.2140605449676514,
      "learning_rate": 8.585078544509266e-05,
      "loss": 1.9027,
      "step": 173503
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.3000555038452148,
      "learning_rate": 8.58479207694866e-05,
      "loss": 2.0687,
      "step": 173504
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.2899303436279297,
      "learning_rate": 8.584505613369539e-05,
      "loss": 2.0073,
      "step": 173505
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.3424701690673828,
      "learning_rate": 8.58421915377197e-05,
      "loss": 2.1071,
      "step": 173506
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.3991847038269043,
      "learning_rate": 8.583932698155991e-05,
      "loss": 2.0753,
      "step": 173507
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.3320364952087402,
      "learning_rate": 8.58364624652168e-05,
      "loss": 1.7871,
      "step": 173508
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.303499460220337,
      "learning_rate": 8.583359798869055e-05,
      "loss": 2.0675,
      "step": 173509
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.2525705099105835,
      "learning_rate": 8.583073355198198e-05,
      "loss": 2.0417,
      "step": 173510
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.215390920639038,
      "learning_rate": 8.582786915509145e-05,
      "loss": 1.8897,
      "step": 173511
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.4064500331878662,
      "learning_rate": 8.582500479801962e-05,
      "loss": 1.8781,
      "step": 173512
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.2707661390304565,
      "learning_rate": 8.582214048076686e-05,
      "loss": 2.0225,
      "step": 173513
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.3562947511672974,
      "learning_rate": 8.581927620333403e-05,
      "loss": 2.1737,
      "step": 173514
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.3099106550216675,
      "learning_rate": 8.581641196572124e-05,
      "loss": 2.0663,
      "step": 173515
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.3112608194351196,
      "learning_rate": 8.581354776792931e-05,
      "loss": 1.9856,
      "step": 173516
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.3263746500015259,
      "learning_rate": 8.581068360995858e-05,
      "loss": 2.065,
      "step": 173517
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.246498465538025,
      "learning_rate": 8.580781949180979e-05,
      "loss": 1.995,
      "step": 173518
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.3462623357772827,
      "learning_rate": 8.58049554134833e-05,
      "loss": 2.1796,
      "step": 173519
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.3142616748809814,
      "learning_rate": 8.580209137497977e-05,
      "loss": 2.0937,
      "step": 173520
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.374057650566101,
      "learning_rate": 8.579922737629968e-05,
      "loss": 2.0102,
      "step": 173521
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.2509229183197021,
      "learning_rate": 8.579636341744358e-05,
      "loss": 1.988,
      "step": 173522
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.237833023071289,
      "learning_rate": 8.579349949841185e-05,
      "loss": 1.9969,
      "step": 173523
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.3905372619628906,
      "learning_rate": 8.579063561920528e-05,
      "loss": 2.2145,
      "step": 173524
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.318135142326355,
      "learning_rate": 8.578777177982417e-05,
      "loss": 1.9884,
      "step": 173525
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.3675936460494995,
      "learning_rate": 8.578490798026926e-05,
      "loss": 2.2039,
      "step": 173526
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.288107991218567,
      "learning_rate": 8.578204422054098e-05,
      "loss": 2.157,
      "step": 173527
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.3120167255401611,
      "learning_rate": 8.577918050063983e-05,
      "loss": 2.0408,
      "step": 173528
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.170678973197937,
      "learning_rate": 8.57763168205663e-05,
      "loss": 2.0023,
      "step": 173529
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.2492570877075195,
      "learning_rate": 8.577345318032109e-05,
      "loss": 1.9197,
      "step": 173530
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.2817021608352661,
      "learning_rate": 8.577058957990454e-05,
      "loss": 1.8948,
      "step": 173531
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.2063623666763306,
      "learning_rate": 8.576772601931739e-05,
      "loss": 1.9019,
      "step": 173532
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.367461919784546,
      "learning_rate": 8.576486249856004e-05,
      "loss": 2.0087,
      "step": 173533
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.2625443935394287,
      "learning_rate": 8.576199901763296e-05,
      "loss": 2.1317,
      "step": 173534
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.6087229251861572,
      "learning_rate": 8.57591355765369e-05,
      "loss": 1.9693,
      "step": 173535
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.327928066253662,
      "learning_rate": 8.575627217527224e-05,
      "loss": 2.0424,
      "step": 173536
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.3126418590545654,
      "learning_rate": 8.57534088138394e-05,
      "loss": 2.0411,
      "step": 173537
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.1480695009231567,
      "learning_rate": 8.575054549223919e-05,
      "loss": 2.1107,
      "step": 173538
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.4413464069366455,
      "learning_rate": 8.574768221047197e-05,
      "loss": 2.2388,
      "step": 173539
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.3253253698349,
      "learning_rate": 8.574481896853822e-05,
      "loss": 1.9679,
      "step": 173540
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.4836499691009521,
      "learning_rate": 8.574195576643861e-05,
      "loss": 2.0347,
      "step": 173541
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.3919415473937988,
      "learning_rate": 8.573909260417357e-05,
      "loss": 1.9649,
      "step": 173542
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.4671614170074463,
      "learning_rate": 8.573622948174375e-05,
      "loss": 1.8908,
      "step": 173543
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.2018531560897827,
      "learning_rate": 8.573336639914961e-05,
      "loss": 1.872,
      "step": 173544
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.3043131828308105,
      "learning_rate": 8.573050335639168e-05,
      "loss": 1.8391,
      "step": 173545
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.3577033281326294,
      "learning_rate": 8.57276403534704e-05,
      "loss": 2.1344,
      "step": 173546
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.3455228805541992,
      "learning_rate": 8.57247773903865e-05,
      "loss": 1.8367,
      "step": 173547
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.2658084630966187,
      "learning_rate": 8.572191446714029e-05,
      "loss": 2.3198,
      "step": 173548
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.3675570487976074,
      "learning_rate": 8.571905158373256e-05,
      "loss": 2.1258,
      "step": 173549
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.2825255393981934,
      "learning_rate": 8.571618874016365e-05,
      "loss": 2.045,
      "step": 173550
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.135764241218567,
      "learning_rate": 8.571332593643417e-05,
      "loss": 2.1664,
      "step": 173551
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.322420597076416,
      "learning_rate": 8.571046317254453e-05,
      "loss": 2.1042,
      "step": 173552
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.2248165607452393,
      "learning_rate": 8.570760044849549e-05,
      "loss": 1.9625,
      "step": 173553
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.3623368740081787,
      "learning_rate": 8.57047377642873e-05,
      "loss": 2.0003,
      "step": 173554
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.4221465587615967,
      "learning_rate": 8.570187511992078e-05,
      "loss": 2.0142,
      "step": 173555
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.2253797054290771,
      "learning_rate": 8.569901251539622e-05,
      "loss": 1.8739,
      "step": 173556
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.2792539596557617,
      "learning_rate": 8.569614995071443e-05,
      "loss": 2.0152,
      "step": 173557
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.3149287700653076,
      "learning_rate": 8.56932874258756e-05,
      "loss": 1.8728,
      "step": 173558
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.2310316562652588,
      "learning_rate": 8.569042494088053e-05,
      "loss": 2.0051,
      "step": 173559
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.1563628911972046,
      "learning_rate": 8.568756249572955e-05,
      "loss": 2.1481,
      "step": 173560
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.3806488513946533,
      "learning_rate": 8.56847000904234e-05,
      "loss": 2.0152,
      "step": 173561
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.2101112604141235,
      "learning_rate": 8.568183772496242e-05,
      "loss": 1.994,
      "step": 173562
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.1816182136535645,
      "learning_rate": 8.56789753993474e-05,
      "loss": 1.9729,
      "step": 173563
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.4785292148590088,
      "learning_rate": 8.567611311357853e-05,
      "loss": 2.0572,
      "step": 173564
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.40380859375,
      "learning_rate": 8.567325086765662e-05,
      "loss": 2.0244,
      "step": 173565
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.2935501337051392,
      "learning_rate": 8.5670388661582e-05,
      "loss": 2.236,
      "step": 173566
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.2471895217895508,
      "learning_rate": 8.566752649535537e-05,
      "loss": 2.0252,
      "step": 173567
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.290676236152649,
      "learning_rate": 8.566466436897715e-05,
      "loss": 2.0897,
      "step": 173568
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.3170456886291504,
      "learning_rate": 8.566180228244803e-05,
      "loss": 1.9073,
      "step": 173569
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.2556719779968262,
      "learning_rate": 8.565894023576828e-05,
      "loss": 2.0227,
      "step": 173570
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.1663786172866821,
      "learning_rate": 8.565607822893869e-05,
      "loss": 1.8962,
      "step": 173571
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.3746943473815918,
      "learning_rate": 8.565321626195958e-05,
      "loss": 1.8327,
      "step": 173572
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.213834524154663,
      "learning_rate": 8.565035433483167e-05,
      "loss": 1.8301,
      "step": 173573
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.3523027896881104,
      "learning_rate": 8.56474924475553e-05,
      "loss": 2.1832,
      "step": 173574
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.3344740867614746,
      "learning_rate": 8.56446306001313e-05,
      "loss": 1.9597,
      "step": 173575
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.3728675842285156,
      "learning_rate": 8.564176879255981e-05,
      "loss": 1.9796,
      "step": 173576
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.4263250827789307,
      "learning_rate": 8.563890702484168e-05,
      "loss": 2.0243,
      "step": 173577
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.2973896265029907,
      "learning_rate": 8.563604529697723e-05,
      "loss": 1.9001,
      "step": 173578
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.266457438468933,
      "learning_rate": 8.563318360896717e-05,
      "loss": 1.7929,
      "step": 173579
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.1932471990585327,
      "learning_rate": 8.563032196081183e-05,
      "loss": 2.0863,
      "step": 173580
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.1760351657867432,
      "learning_rate": 8.562746035251208e-05,
      "loss": 2.2368,
      "step": 173581
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.3515102863311768,
      "learning_rate": 8.562459878406804e-05,
      "loss": 1.9777,
      "step": 173582
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.328797459602356,
      "learning_rate": 8.562173725548053e-05,
      "loss": 1.9475,
      "step": 173583
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.5616902112960815,
      "learning_rate": 8.561887576674986e-05,
      "loss": 2.0002,
      "step": 173584
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.2930411100387573,
      "learning_rate": 8.561601431787681e-05,
      "loss": 1.9439,
      "step": 173585
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.2738723754882812,
      "learning_rate": 8.561315290886169e-05,
      "loss": 2.0262,
      "step": 173586
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.2124491930007935,
      "learning_rate": 8.561029153970533e-05,
      "loss": 1.9971,
      "step": 173587
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.2727001905441284,
      "learning_rate": 8.560743021040786e-05,
      "loss": 2.0309,
      "step": 173588
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.2720727920532227,
      "learning_rate": 8.560456892097012e-05,
      "loss": 2.2473,
      "step": 173589
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.2479348182678223,
      "learning_rate": 8.560170767139246e-05,
      "loss": 2.2248,
      "step": 173590
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.39173424243927,
      "learning_rate": 8.559884646167554e-05,
      "loss": 2.024,
      "step": 173591
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.272722601890564,
      "learning_rate": 8.55959852918198e-05,
      "loss": 1.9549,
      "step": 173592
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.3052589893341064,
      "learning_rate": 8.559312416182593e-05,
      "loss": 1.7834,
      "step": 173593
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.3490855693817139,
      "learning_rate": 8.559026307169422e-05,
      "loss": 2.0106,
      "step": 173594
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.3264379501342773,
      "learning_rate": 8.558740202142539e-05,
      "loss": 1.9983,
      "step": 173595
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.2895333766937256,
      "learning_rate": 8.558454101101984e-05,
      "loss": 2.1061,
      "step": 173596
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.2868564128875732,
      "learning_rate": 8.558168004047827e-05,
      "loss": 1.9218,
      "step": 173597
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.2688701152801514,
      "learning_rate": 8.557881910980097e-05,
      "loss": 1.9012,
      "step": 173598
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.276671290397644,
      "learning_rate": 8.557595821898885e-05,
      "loss": 2.1976,
      "step": 173599
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.3345277309417725,
      "learning_rate": 8.5573097368042e-05,
      "loss": 2.2472,
      "step": 173600
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.3396466970443726,
      "learning_rate": 8.557023655696124e-05,
      "loss": 1.9113,
      "step": 173601
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.4517213106155396,
      "learning_rate": 8.556737578574696e-05,
      "loss": 1.911,
      "step": 173602
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.3978627920150757,
      "learning_rate": 8.556451505439982e-05,
      "loss": 2.1079,
      "step": 173603
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.4751248359680176,
      "learning_rate": 8.556165436292022e-05,
      "loss": 2.0051,
      "step": 173604
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.3190062046051025,
      "learning_rate": 8.555879371130883e-05,
      "loss": 2.2191,
      "step": 173605
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.2814688682556152,
      "learning_rate": 8.555593309956608e-05,
      "loss": 2.0064,
      "step": 173606
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.3275163173675537,
      "learning_rate": 8.555307252769256e-05,
      "loss": 2.0119,
      "step": 173607
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.2819143533706665,
      "learning_rate": 8.55502119956887e-05,
      "loss": 1.9996,
      "step": 173608
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.4998866319656372,
      "learning_rate": 8.554735150355517e-05,
      "loss": 1.9332,
      "step": 173609
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.2441476583480835,
      "learning_rate": 8.554449105129231e-05,
      "loss": 2.2495,
      "step": 173610
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.2153643369674683,
      "learning_rate": 8.554163063890094e-05,
      "loss": 1.8953,
      "step": 173611
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.2169312238693237,
      "learning_rate": 8.553877026638139e-05,
      "loss": 1.8768,
      "step": 173612
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.5693211555480957,
      "learning_rate": 8.553590993373422e-05,
      "loss": 1.9346,
      "step": 173613
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.306077241897583,
      "learning_rate": 8.553304964095989e-05,
      "loss": 1.9792,
      "step": 173614
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.203242301940918,
      "learning_rate": 8.553018938805914e-05,
      "loss": 1.9347,
      "step": 173615
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.2537087202072144,
      "learning_rate": 8.552732917503224e-05,
      "loss": 2.117,
      "step": 173616
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.273282527923584,
      "learning_rate": 8.552446900187995e-05,
      "loss": 1.9831,
      "step": 173617
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.2917423248291016,
      "learning_rate": 8.552160886860274e-05,
      "loss": 1.9776,
      "step": 173618
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.326858401298523,
      "learning_rate": 8.551874877520103e-05,
      "loss": 1.9376,
      "step": 173619
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.2830300331115723,
      "learning_rate": 8.551588872167549e-05,
      "loss": 2.251,
      "step": 173620
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.5552047491073608,
      "learning_rate": 8.55130287080266e-05,
      "loss": 2.1131,
      "step": 173621
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.3152925968170166,
      "learning_rate": 8.55101687342548e-05,
      "loss": 1.9575,
      "step": 173622
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.3450336456298828,
      "learning_rate": 8.550730880036079e-05,
      "loss": 1.9539,
      "step": 173623
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.285883903503418,
      "learning_rate": 8.550444890634505e-05,
      "loss": 2.2046,
      "step": 173624
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.404123067855835,
      "learning_rate": 8.550158905220799e-05,
      "loss": 1.9407,
      "step": 173625
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.2640458345413208,
      "learning_rate": 8.549872923795029e-05,
      "loss": 1.8277,
      "step": 173626
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.3064204454421997,
      "learning_rate": 8.549586946357244e-05,
      "loss": 1.8668,
      "step": 173627
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.2986128330230713,
      "learning_rate": 8.549300972907486e-05,
      "loss": 2.1034,
      "step": 173628
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.2963355779647827,
      "learning_rate": 8.54901500344583e-05,
      "loss": 1.9937,
      "step": 173629
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.2908047437667847,
      "learning_rate": 8.548729037972315e-05,
      "loss": 1.9482,
      "step": 173630
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.3870584964752197,
      "learning_rate": 8.548443076486984e-05,
      "loss": 2.1219,
      "step": 173631
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.3037227392196655,
      "learning_rate": 8.548157118989916e-05,
      "loss": 1.8805,
      "step": 173632
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.2799891233444214,
      "learning_rate": 8.54787116548114e-05,
      "loss": 1.9661,
      "step": 173633
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.3027369976043701,
      "learning_rate": 8.547585215960729e-05,
      "loss": 1.9496,
      "step": 173634
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.2156150341033936,
      "learning_rate": 8.547299270428726e-05,
      "loss": 1.8325,
      "step": 173635
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.387642741203308,
      "learning_rate": 8.547013328885186e-05,
      "loss": 2.114,
      "step": 173636
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.3135989904403687,
      "learning_rate": 8.546727391330149e-05,
      "loss": 2.0273,
      "step": 173637
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.2865653038024902,
      "learning_rate": 8.546441457763692e-05,
      "loss": 2.1027,
      "step": 173638
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.3956118822097778,
      "learning_rate": 8.546155528185847e-05,
      "loss": 1.9526,
      "step": 173639
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.2670387029647827,
      "learning_rate": 8.545869602596689e-05,
      "loss": 1.9935,
      "step": 173640
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.3791910409927368,
      "learning_rate": 8.545583680996247e-05,
      "loss": 2.2236,
      "step": 173641
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.3000661134719849,
      "learning_rate": 8.545297763384604e-05,
      "loss": 1.8052,
      "step": 173642
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.2204350233078003,
      "learning_rate": 8.545011849761775e-05,
      "loss": 2.0802,
      "step": 173643
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.308461308479309,
      "learning_rate": 8.544725940127845e-05,
      "loss": 1.9765,
      "step": 173644
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.378196358680725,
      "learning_rate": 8.544440034482846e-05,
      "loss": 1.7789,
      "step": 173645
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.3446811437606812,
      "learning_rate": 8.544154132826849e-05,
      "loss": 1.8488,
      "step": 173646
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.3504955768585205,
      "learning_rate": 8.543868235159888e-05,
      "loss": 2.0238,
      "step": 173647
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.2105624675750732,
      "learning_rate": 8.54358234148205e-05,
      "loss": 2.0298,
      "step": 173648
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.2438125610351562,
      "learning_rate": 8.543296451793345e-05,
      "loss": 1.9027,
      "step": 173649
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.2696868181228638,
      "learning_rate": 8.543010566093854e-05,
      "loss": 2.0589,
      "step": 173650
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.407619833946228,
      "learning_rate": 8.542724684383612e-05,
      "loss": 2.0868,
      "step": 173651
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.1975919008255005,
      "learning_rate": 8.542438806662694e-05,
      "loss": 1.9233,
      "step": 173652
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.4095022678375244,
      "learning_rate": 8.54215293293113e-05,
      "loss": 2.0725,
      "step": 173653
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.2384036779403687,
      "learning_rate": 8.541867063189006e-05,
      "loss": 2.1257,
      "step": 173654
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.31088387966156,
      "learning_rate": 8.541581197436335e-05,
      "loss": 1.9341,
      "step": 173655
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.447019338607788,
      "learning_rate": 8.541295335673198e-05,
      "loss": 2.0497,
      "step": 173656
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.3039287328720093,
      "learning_rate": 8.54100947789963e-05,
      "loss": 2.0575,
      "step": 173657
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.28752863407135,
      "learning_rate": 8.540723624115703e-05,
      "loss": 2.0307,
      "step": 173658
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.1887773275375366,
      "learning_rate": 8.540437774321449e-05,
      "loss": 1.9454,
      "step": 173659
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.282016634941101,
      "learning_rate": 8.540151928516956e-05,
      "loss": 2.1244,
      "step": 173660
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.3742396831512451,
      "learning_rate": 8.53986608670223e-05,
      "loss": 1.9428,
      "step": 173661
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.3456714153289795,
      "learning_rate": 8.539580248877358e-05,
      "loss": 2.0426,
      "step": 173662
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.2686995267868042,
      "learning_rate": 8.539294415042378e-05,
      "loss": 1.9361,
      "step": 173663
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.250626564025879,
      "learning_rate": 8.539008585197354e-05,
      "loss": 1.942,
      "step": 173664
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.2211613655090332,
      "learning_rate": 8.538722759342323e-05,
      "loss": 1.9456,
      "step": 173665
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.2534250020980835,
      "learning_rate": 8.538436937477371e-05,
      "loss": 2.1633,
      "step": 173666
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.2393649816513062,
      "learning_rate": 8.53815111960251e-05,
      "loss": 1.9358,
      "step": 173667
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.2023142576217651,
      "learning_rate": 8.537865305717817e-05,
      "loss": 1.9694,
      "step": 173668
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.3914010524749756,
      "learning_rate": 8.537579495823336e-05,
      "loss": 2.1695,
      "step": 173669
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.2592535018920898,
      "learning_rate": 8.53729368991913e-05,
      "loss": 1.862,
      "step": 173670
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.300952672958374,
      "learning_rate": 8.537007888005237e-05,
      "loss": 2.1024,
      "step": 173671
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.347807765007019,
      "learning_rate": 8.536722090081737e-05,
      "loss": 2.2139,
      "step": 173672
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.3266255855560303,
      "learning_rate": 8.536436296148647e-05,
      "loss": 2.1621,
      "step": 173673
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.3207753896713257,
      "learning_rate": 8.536150506206053e-05,
      "loss": 2.1003,
      "step": 173674
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.4156321287155151,
      "learning_rate": 8.53586472025398e-05,
      "loss": 2.2852,
      "step": 173675
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.3065682649612427,
      "learning_rate": 8.535578938292505e-05,
      "loss": 1.9524,
      "step": 173676
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.3054654598236084,
      "learning_rate": 8.53529316032166e-05,
      "loss": 2.0734,
      "step": 173677
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.243701696395874,
      "learning_rate": 8.53500738634153e-05,
      "loss": 1.7911,
      "step": 173678
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.245539665222168,
      "learning_rate": 8.534721616352128e-05,
      "loss": 2.2601,
      "step": 173679
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.366371989250183,
      "learning_rate": 8.534435850353537e-05,
      "loss": 2.3035,
      "step": 173680
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.3663691282272339,
      "learning_rate": 8.53415008834579e-05,
      "loss": 1.8921,
      "step": 173681
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.3381175994873047,
      "learning_rate": 8.533864330328957e-05,
      "loss": 2.0416,
      "step": 173682
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.1614845991134644,
      "learning_rate": 8.533578576303077e-05,
      "loss": 2.0843,
      "step": 173683
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.376396656036377,
      "learning_rate": 8.533292826268226e-05,
      "loss": 2.1717,
      "step": 173684
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.5545002222061157,
      "learning_rate": 8.533007080224426e-05,
      "loss": 2.1652,
      "step": 173685
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.4053795337677002,
      "learning_rate": 8.532721338171749e-05,
      "loss": 1.8279,
      "step": 173686
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.321736454963684,
      "learning_rate": 8.532435600110242e-05,
      "loss": 2.087,
      "step": 173687
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.2441569566726685,
      "learning_rate": 8.532149866039964e-05,
      "loss": 1.8025,
      "step": 173688
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.341858148574829,
      "learning_rate": 8.53186413596096e-05,
      "loss": 2.1486,
      "step": 173689
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.139620304107666,
      "learning_rate": 8.531578409873293e-05,
      "loss": 1.8208,
      "step": 173690
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.2650235891342163,
      "learning_rate": 8.53129268777701e-05,
      "loss": 1.9432,
      "step": 173691
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.4116106033325195,
      "learning_rate": 8.531006969672168e-05,
      "loss": 1.9721,
      "step": 173692
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.2304868698120117,
      "learning_rate": 8.530721255558807e-05,
      "loss": 1.8815,
      "step": 173693
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.1963788270950317,
      "learning_rate": 8.530435545437e-05,
      "loss": 1.7757,
      "step": 173694
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.2938296794891357,
      "learning_rate": 8.530149839306781e-05,
      "loss": 2.2121,
      "step": 173695
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.3232905864715576,
      "learning_rate": 8.529864137168221e-05,
      "loss": 2.0535,
      "step": 173696
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.339638352394104,
      "learning_rate": 8.529578439021365e-05,
      "loss": 2.0433,
      "step": 173697
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.3061174154281616,
      "learning_rate": 8.529292744866264e-05,
      "loss": 1.8929,
      "step": 173698
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.4113810062408447,
      "learning_rate": 8.529007054702965e-05,
      "loss": 2.2446,
      "step": 173699
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.374931812286377,
      "learning_rate": 8.528721368531539e-05,
      "loss": 1.7703,
      "step": 173700
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.4502283334732056,
      "learning_rate": 8.528435686352019e-05,
      "loss": 2.0647,
      "step": 173701
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.2547812461853027,
      "learning_rate": 8.528150008164475e-05,
      "loss": 1.9856,
      "step": 173702
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.1939724683761597,
      "learning_rate": 8.527864333968956e-05,
      "loss": 1.7506,
      "step": 173703
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.2785634994506836,
      "learning_rate": 8.527578663765504e-05,
      "loss": 2.0878,
      "step": 173704
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.349770188331604,
      "learning_rate": 8.527292997554188e-05,
      "loss": 1.9802,
      "step": 173705
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.1907964944839478,
      "learning_rate": 8.527007335335052e-05,
      "loss": 2.016,
      "step": 173706
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.4962701797485352,
      "learning_rate": 8.526721677108143e-05,
      "loss": 2.2606,
      "step": 173707
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.3172051906585693,
      "learning_rate": 8.526436022873533e-05,
      "loss": 1.8938,
      "step": 173708
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.3649451732635498,
      "learning_rate": 8.52615037263126e-05,
      "loss": 2.3001,
      "step": 173709
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.255839228630066,
      "learning_rate": 8.525864726381376e-05,
      "loss": 2.1051,
      "step": 173710
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.2744261026382446,
      "learning_rate": 8.525579084123946e-05,
      "loss": 2.1972,
      "step": 173711
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.1959586143493652,
      "learning_rate": 8.525293445859015e-05,
      "loss": 1.9226,
      "step": 173712
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.2791106700897217,
      "learning_rate": 8.52500781158663e-05,
      "loss": 1.9918,
      "step": 173713
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.3575544357299805,
      "learning_rate": 8.524722181306858e-05,
      "loss": 2.016,
      "step": 173714
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.1668174266815186,
      "learning_rate": 8.52443655501975e-05,
      "loss": 1.9646,
      "step": 173715
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.4165323972702026,
      "learning_rate": 8.524150932725341e-05,
      "loss": 1.9336,
      "step": 173716
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.3598301410675049,
      "learning_rate": 8.523865314423711e-05,
      "loss": 2.1741,
      "step": 173717
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.282020926475525,
      "learning_rate": 8.523579700114888e-05,
      "loss": 2.0001,
      "step": 173718
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.2831664085388184,
      "learning_rate": 8.523294089798947e-05,
      "loss": 2.0169,
      "step": 173719
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.3566893339157104,
      "learning_rate": 8.523008483475932e-05,
      "loss": 2.1434,
      "step": 173720
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.2763162851333618,
      "learning_rate": 8.522722881145893e-05,
      "loss": 2.1203,
      "step": 173721
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.2455651760101318,
      "learning_rate": 8.522437282808876e-05,
      "loss": 2.1424,
      "step": 173722
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.479261875152588,
      "learning_rate": 8.522151688464952e-05,
      "loss": 1.888,
      "step": 173723
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.2203128337860107,
      "learning_rate": 8.521866098114157e-05,
      "loss": 1.9421,
      "step": 173724
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.1962398290634155,
      "learning_rate": 8.521580511756565e-05,
      "loss": 2.143,
      "step": 173725
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.3343008756637573,
      "learning_rate": 8.521294929392204e-05,
      "loss": 2.0152,
      "step": 173726
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.371803879737854,
      "learning_rate": 8.521009351021157e-05,
      "loss": 2.0549,
      "step": 173727
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.233475685119629,
      "learning_rate": 8.520723776643446e-05,
      "loss": 1.9272,
      "step": 173728
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.2631367444992065,
      "learning_rate": 8.520438206259142e-05,
      "loss": 1.8138,
      "step": 173729
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.2517263889312744,
      "learning_rate": 8.520152639868289e-05,
      "loss": 2.0583,
      "step": 173730
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.4079481363296509,
      "learning_rate": 8.519867077470953e-05,
      "loss": 2.075,
      "step": 173731
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.238490104675293,
      "learning_rate": 8.51958151906717e-05,
      "loss": 2.069,
      "step": 173732
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.5350855588912964,
      "learning_rate": 8.51929596465702e-05,
      "loss": 2.1075,
      "step": 173733
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.3406318426132202,
      "learning_rate": 8.51901041424052e-05,
      "loss": 1.8809,
      "step": 173734
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.4351760149002075,
      "learning_rate": 8.51872486781775e-05,
      "loss": 1.9213,
      "step": 173735
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.2726483345031738,
      "learning_rate": 8.518439325388745e-05,
      "loss": 2.0056,
      "step": 173736
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.356898546218872,
      "learning_rate": 8.518153786953579e-05,
      "loss": 2.2345,
      "step": 173737
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.173712134361267,
      "learning_rate": 8.517868252512285e-05,
      "loss": 2.0324,
      "step": 173738
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.3276337385177612,
      "learning_rate": 8.517582722064938e-05,
      "loss": 1.9758,
      "step": 173739
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.2881773710250854,
      "learning_rate": 8.517297195611564e-05,
      "loss": 1.9059,
      "step": 173740
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.207047700881958,
      "learning_rate": 8.517011673152238e-05,
      "loss": 2.1446,
      "step": 173741
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.1296411752700806,
      "learning_rate": 8.516726154686998e-05,
      "loss": 1.7029,
      "step": 173742
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.831699013710022,
      "learning_rate": 8.516440640215909e-05,
      "loss": 1.747,
      "step": 173743
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.2551789283752441,
      "learning_rate": 8.516155129739015e-05,
      "loss": 1.9385,
      "step": 173744
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.2773517370224,
      "learning_rate": 8.51586962325639e-05,
      "loss": 1.9455,
      "step": 173745
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.5960500240325928,
      "learning_rate": 8.515584120768049e-05,
      "loss": 1.9454,
      "step": 173746
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.6097962856292725,
      "learning_rate": 8.515298622274079e-05,
      "loss": 1.9747,
      "step": 173747
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.2921873331069946,
      "learning_rate": 8.515013127774509e-05,
      "loss": 2.0505,
      "step": 173748
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.1585042476654053,
      "learning_rate": 8.514727637269416e-05,
      "loss": 1.9499,
      "step": 173749
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.336787223815918,
      "learning_rate": 8.514442150758827e-05,
      "loss": 1.9827,
      "step": 173750
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.2635189294815063,
      "learning_rate": 8.51415666824283e-05,
      "loss": 1.9898,
      "step": 173751
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.363365650177002,
      "learning_rate": 8.513871189721438e-05,
      "loss": 2.0121,
      "step": 173752
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.3197531700134277,
      "learning_rate": 8.513585715194732e-05,
      "loss": 1.855,
      "step": 173753
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.2113091945648193,
      "learning_rate": 8.513300244662747e-05,
      "loss": 1.9706,
      "step": 173754
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.2910743951797485,
      "learning_rate": 8.513014778125554e-05,
      "loss": 1.9865,
      "step": 173755
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.313827395439148,
      "learning_rate": 8.51272931558319e-05,
      "loss": 2.1251,
      "step": 173756
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.3836157321929932,
      "learning_rate": 8.512443857035723e-05,
      "loss": 2.2336,
      "step": 173757
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.3908588886260986,
      "learning_rate": 8.512158402483198e-05,
      "loss": 2.2395,
      "step": 173758
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.2482644319534302,
      "learning_rate": 8.511872951925667e-05,
      "loss": 1.9882,
      "step": 173759
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.3053792715072632,
      "learning_rate": 8.511587505363177e-05,
      "loss": 2.1553,
      "step": 173760
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.2777423858642578,
      "learning_rate": 8.511302062795797e-05,
      "loss": 1.9558,
      "step": 173761
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.2648736238479614,
      "learning_rate": 8.511016624223562e-05,
      "loss": 1.9526,
      "step": 173762
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.2904937267303467,
      "learning_rate": 8.510731189646545e-05,
      "loss": 2.0586,
      "step": 173763
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.3258943557739258,
      "learning_rate": 8.510445759064787e-05,
      "loss": 1.8737,
      "step": 173764
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.153563380241394,
      "learning_rate": 8.510160332478345e-05,
      "loss": 2.0133,
      "step": 173765
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.4801186323165894,
      "learning_rate": 8.50987490988726e-05,
      "loss": 2.0882,
      "step": 173766
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.5498124361038208,
      "learning_rate": 8.509589491291605e-05,
      "loss": 2.0587,
      "step": 173767
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.2998090982437134,
      "learning_rate": 8.509304076691412e-05,
      "loss": 2.0302,
      "step": 173768
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.2767106294631958,
      "learning_rate": 8.509018666086755e-05,
      "loss": 2.2303,
      "step": 173769
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.3590483665466309,
      "learning_rate": 8.508733259477675e-05,
      "loss": 2.0416,
      "step": 173770
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.31812584400177,
      "learning_rate": 8.508447856864229e-05,
      "loss": 1.7987,
      "step": 173771
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.4271190166473389,
      "learning_rate": 8.50816245824646e-05,
      "loss": 2.1388,
      "step": 173772
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.2967631816864014,
      "learning_rate": 8.507877063624438e-05,
      "loss": 2.1083,
      "step": 173773
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.2657564878463745,
      "learning_rate": 8.507591672998197e-05,
      "loss": 2.078,
      "step": 173774
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.2272392511367798,
      "learning_rate": 8.507306286367809e-05,
      "loss": 2.0527,
      "step": 173775
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.3175883293151855,
      "learning_rate": 8.507020903733323e-05,
      "loss": 2.033,
      "step": 173776
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.2334293127059937,
      "learning_rate": 8.506735525094782e-05,
      "loss": 2.0413,
      "step": 173777
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.209410548210144,
      "learning_rate": 8.506450150452238e-05,
      "loss": 2.024,
      "step": 173778
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.2991005182266235,
      "learning_rate": 8.506164779805759e-05,
      "loss": 2.0616,
      "step": 173779
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.2460179328918457,
      "learning_rate": 8.505879413155382e-05,
      "loss": 2.0342,
      "step": 173780
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.219401240348816,
      "learning_rate": 8.505594050501175e-05,
      "loss": 1.9868,
      "step": 173781
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.4596613645553589,
      "learning_rate": 8.505308691843186e-05,
      "loss": 2.2394,
      "step": 173782
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.3509103059768677,
      "learning_rate": 8.505023337181465e-05,
      "loss": 1.9672,
      "step": 173783
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.560984492301941,
      "learning_rate": 8.504737986516055e-05,
      "loss": 2.0031,
      "step": 173784
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.2742966413497925,
      "learning_rate": 8.504452639847033e-05,
      "loss": 1.8042,
      "step": 173785
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.446515679359436,
      "learning_rate": 8.504167297174425e-05,
      "loss": 1.9906,
      "step": 173786
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.490140676498413,
      "learning_rate": 8.503881958498309e-05,
      "loss": 2.3893,
      "step": 173787
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.4736621379852295,
      "learning_rate": 8.503596623818728e-05,
      "loss": 1.9682,
      "step": 173788
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.3103930950164795,
      "learning_rate": 8.503311293135734e-05,
      "loss": 2.002,
      "step": 173789
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.2999135255813599,
      "learning_rate": 8.503025966449373e-05,
      "loss": 2.0566,
      "step": 173790
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.2888476848602295,
      "learning_rate": 8.502740643759712e-05,
      "loss": 2.0025,
      "step": 173791
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.4235893487930298,
      "learning_rate": 8.502455325066786e-05,
      "loss": 1.9464,
      "step": 173792
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.7147570848464966,
      "learning_rate": 8.502170010370674e-05,
      "loss": 1.9798,
      "step": 173793
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.2059166431427002,
      "learning_rate": 8.50188469967141e-05,
      "loss": 1.857,
      "step": 173794
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.249626636505127,
      "learning_rate": 8.501599392969044e-05,
      "loss": 2.1126,
      "step": 173795
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.2096056938171387,
      "learning_rate": 8.501314090263644e-05,
      "loss": 2.03,
      "step": 173796
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.263565182685852,
      "learning_rate": 8.501028791555259e-05,
      "loss": 1.8826,
      "step": 173797
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.3683364391326904,
      "learning_rate": 8.500743496843927e-05,
      "loss": 2.2123,
      "step": 173798
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.389751672744751,
      "learning_rate": 8.50045820612972e-05,
      "loss": 1.8588,
      "step": 173799
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.2529178857803345,
      "learning_rate": 8.500172919412688e-05,
      "loss": 2.3192,
      "step": 173800
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.4672216176986694,
      "learning_rate": 8.499887636692867e-05,
      "loss": 1.9326,
      "step": 173801
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.2644122838974,
      "learning_rate": 8.499602357970332e-05,
      "loss": 2.1347,
      "step": 173802
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.3298063278198242,
      "learning_rate": 8.499317083245119e-05,
      "loss": 2.0595,
      "step": 173803
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.2100006341934204,
      "learning_rate": 8.4990318125173e-05,
      "loss": 1.9098,
      "step": 173804
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.2790136337280273,
      "learning_rate": 8.498746545786912e-05,
      "loss": 2.2135,
      "step": 173805
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.4425642490386963,
      "learning_rate": 8.498461283054016e-05,
      "loss": 2.0827,
      "step": 173806
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.359053611755371,
      "learning_rate": 8.498176024318654e-05,
      "loss": 2.3075,
      "step": 173807
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.1979604959487915,
      "learning_rate": 8.497890769580892e-05,
      "loss": 1.9698,
      "step": 173808
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.4691283702850342,
      "learning_rate": 8.497605518840771e-05,
      "loss": 2.2079,
      "step": 173809
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.3777207136154175,
      "learning_rate": 8.497320272098361e-05,
      "loss": 1.9227,
      "step": 173810
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.2972795963287354,
      "learning_rate": 8.497035029353704e-05,
      "loss": 1.9481,
      "step": 173811
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.22647225856781,
      "learning_rate": 8.496749790606855e-05,
      "loss": 2.1882,
      "step": 173812
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.236955165863037,
      "learning_rate": 8.496464555857858e-05,
      "loss": 1.9365,
      "step": 173813
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.3551706075668335,
      "learning_rate": 8.49617932510678e-05,
      "loss": 2.0683,
      "step": 173814
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.1300015449523926,
      "learning_rate": 8.495894098353661e-05,
      "loss": 2.0965,
      "step": 173815
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.3085142374038696,
      "learning_rate": 8.495608875598569e-05,
      "loss": 1.9616,
      "step": 173816
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.5984752178192139,
      "learning_rate": 8.495323656841544e-05,
      "loss": 2.0418,
      "step": 173817
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.312743902206421,
      "learning_rate": 8.49503844208266e-05,
      "loss": 1.998,
      "step": 173818
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.2499557733535767,
      "learning_rate": 8.494753231321934e-05,
      "loss": 2.045,
      "step": 173819
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.380223035812378,
      "learning_rate": 8.49446802455945e-05,
      "loss": 1.8537,
      "step": 173820
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.2116785049438477,
      "learning_rate": 8.494182821795243e-05,
      "loss": 2.061,
      "step": 173821
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.3201923370361328,
      "learning_rate": 8.493897623029379e-05,
      "loss": 1.982,
      "step": 173822
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.2730358839035034,
      "learning_rate": 8.4936124282619e-05,
      "loss": 2.178,
      "step": 173823
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.38908052444458,
      "learning_rate": 8.493327237492869e-05,
      "loss": 2.096,
      "step": 173824
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.2672193050384521,
      "learning_rate": 8.493042050722338e-05,
      "loss": 2.0546,
      "step": 173825
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.281639575958252,
      "learning_rate": 8.492756867950356e-05,
      "loss": 1.9765,
      "step": 173826
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.3294717073440552,
      "learning_rate": 8.492471689176968e-05,
      "loss": 1.9489,
      "step": 173827
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.316860318183899,
      "learning_rate": 8.492186514402243e-05,
      "loss": 2.2043,
      "step": 173828
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.3301916122436523,
      "learning_rate": 8.49190134362622e-05,
      "loss": 1.9304,
      "step": 173829
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.3138172626495361,
      "learning_rate": 8.491616176848965e-05,
      "loss": 2.0611,
      "step": 173830
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.249260663986206,
      "learning_rate": 8.491331014070524e-05,
      "loss": 2.1956,
      "step": 173831
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.4370996952056885,
      "learning_rate": 8.491045855290951e-05,
      "loss": 2.1182,
      "step": 173832
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.3588238954544067,
      "learning_rate": 8.49076070051029e-05,
      "loss": 1.9576,
      "step": 173833
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.2816569805145264,
      "learning_rate": 8.490475549728612e-05,
      "loss": 2.0401,
      "step": 173834
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.2458373308181763,
      "learning_rate": 8.490190402945953e-05,
      "loss": 2.0919,
      "step": 173835
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.4080209732055664,
      "learning_rate": 8.489905260162382e-05,
      "loss": 1.9062,
      "step": 173836
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.4101580381393433,
      "learning_rate": 8.489620121377941e-05,
      "loss": 2.0417,
      "step": 173837
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.1789194345474243,
      "learning_rate": 8.489334986592688e-05,
      "loss": 1.9624,
      "step": 173838
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.3514665365219116,
      "learning_rate": 8.489049855806664e-05,
      "loss": 1.9758,
      "step": 173839
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.2128727436065674,
      "learning_rate": 8.488764729019943e-05,
      "loss": 2.1283,
      "step": 173840
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.1818351745605469,
      "learning_rate": 8.488479606232552e-05,
      "loss": 2.0689,
      "step": 173841
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.1810543537139893,
      "learning_rate": 8.488194487444572e-05,
      "loss": 2.1075,
      "step": 173842
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.2488856315612793,
      "learning_rate": 8.487909372656045e-05,
      "loss": 1.7588,
      "step": 173843
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.205094814300537,
      "learning_rate": 8.487624261867017e-05,
      "loss": 2.0759,
      "step": 173844
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.350785732269287,
      "learning_rate": 8.487339155077539e-05,
      "loss": 2.1089,
      "step": 173845
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.2297825813293457,
      "learning_rate": 8.48705405228768e-05,
      "loss": 1.8827,
      "step": 173846
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.429037094116211,
      "learning_rate": 8.48676895349747e-05,
      "loss": 1.8587,
      "step": 173847
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.8166192770004272,
      "learning_rate": 8.486483858706992e-05,
      "loss": 1.9946,
      "step": 173848
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.4158071279525757,
      "learning_rate": 8.486198767916277e-05,
      "loss": 1.9758,
      "step": 173849
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.306358814239502,
      "learning_rate": 8.485913681125386e-05,
      "loss": 2.0872,
      "step": 173850
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.3380991220474243,
      "learning_rate": 8.485628598334362e-05,
      "loss": 1.8681,
      "step": 173851
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.3031327724456787,
      "learning_rate": 8.485343519543275e-05,
      "loss": 2.0167,
      "step": 173852
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.3551751375198364,
      "learning_rate": 8.485058444752155e-05,
      "loss": 2.0304,
      "step": 173853
    },
    {
      "epoch": 2.26,
      "grad_norm": 2.9271507263183594,
      "learning_rate": 8.484773373961083e-05,
      "loss": 2.3178,
      "step": 173854
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.3260447978973389,
      "learning_rate": 8.484488307170096e-05,
      "loss": 2.1403,
      "step": 173855
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.2895036935806274,
      "learning_rate": 8.484203244379247e-05,
      "loss": 1.9826,
      "step": 173856
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.311610221862793,
      "learning_rate": 8.483918185588585e-05,
      "loss": 1.7199,
      "step": 173857
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.362942099571228,
      "learning_rate": 8.483633130798177e-05,
      "loss": 1.9093,
      "step": 173858
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.252014398574829,
      "learning_rate": 8.483348080008058e-05,
      "loss": 2.224,
      "step": 173859
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.3325668573379517,
      "learning_rate": 8.483063033218299e-05,
      "loss": 2.0306,
      "step": 173860
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.3325450420379639,
      "learning_rate": 8.482777990428947e-05,
      "loss": 2.0981,
      "step": 173861
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.3651182651519775,
      "learning_rate": 8.482492951640048e-05,
      "loss": 2.0233,
      "step": 173862
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.3333178758621216,
      "learning_rate": 8.482207916851656e-05,
      "loss": 1.8213,
      "step": 173863
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.2748256921768188,
      "learning_rate": 8.481922886063834e-05,
      "loss": 1.6905,
      "step": 173864
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.2584956884384155,
      "learning_rate": 8.481637859276622e-05,
      "loss": 2.0338,
      "step": 173865
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.321950078010559,
      "learning_rate": 8.481352836490087e-05,
      "loss": 2.073,
      "step": 173866
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.1944873332977295,
      "learning_rate": 8.481067817704273e-05,
      "loss": 2.084,
      "step": 173867
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.3432881832122803,
      "learning_rate": 8.480782802919238e-05,
      "loss": 2.0675,
      "step": 173868
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.2608146667480469,
      "learning_rate": 8.48049779213502e-05,
      "loss": 2.0208,
      "step": 173869
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.2653454542160034,
      "learning_rate": 8.480212785351694e-05,
      "loss": 2.0044,
      "step": 173870
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.239532709121704,
      "learning_rate": 8.479927782569292e-05,
      "loss": 2.0048,
      "step": 173871
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.2614468336105347,
      "learning_rate": 8.479642783787891e-05,
      "loss": 2.1958,
      "step": 173872
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.2572097778320312,
      "learning_rate": 8.479357789007526e-05,
      "loss": 1.9155,
      "step": 173873
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.2947092056274414,
      "learning_rate": 8.479072798228258e-05,
      "loss": 2.0784,
      "step": 173874
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.4439053535461426,
      "learning_rate": 8.478787811450124e-05,
      "loss": 2.0199,
      "step": 173875
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.3161565065383911,
      "learning_rate": 8.4785028286732e-05,
      "loss": 1.8631,
      "step": 173876
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.2782840728759766,
      "learning_rate": 8.478217849897523e-05,
      "loss": 1.998,
      "step": 173877
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.4489760398864746,
      "learning_rate": 8.477932875123158e-05,
      "loss": 2.0352,
      "step": 173878
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.363304615020752,
      "learning_rate": 8.477647904350148e-05,
      "loss": 2.0544,
      "step": 173879
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.3129539489746094,
      "learning_rate": 8.477362937578545e-05,
      "loss": 2.1179,
      "step": 173880
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.3508650064468384,
      "learning_rate": 8.477077974808415e-05,
      "loss": 1.7442,
      "step": 173881
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.36424720287323,
      "learning_rate": 8.4767930160398e-05,
      "loss": 2.014,
      "step": 173882
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.42442786693573,
      "learning_rate": 8.47650806127275e-05,
      "loss": 1.9352,
      "step": 173883
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.49202561378479,
      "learning_rate": 8.476223110507331e-05,
      "loss": 2.1603,
      "step": 173884
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.2205528020858765,
      "learning_rate": 8.475938163743587e-05,
      "loss": 1.8486,
      "step": 173885
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.2776719331741333,
      "learning_rate": 8.475653220981565e-05,
      "loss": 1.9561,
      "step": 173886
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.190940260887146,
      "learning_rate": 8.475368282221334e-05,
      "loss": 1.981,
      "step": 173887
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.3589895963668823,
      "learning_rate": 8.475083347462929e-05,
      "loss": 1.859,
      "step": 173888
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.2412307262420654,
      "learning_rate": 8.47479841670642e-05,
      "loss": 2.045,
      "step": 173889
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.3196715116500854,
      "learning_rate": 8.474513489951855e-05,
      "loss": 1.9457,
      "step": 173890
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.1932791471481323,
      "learning_rate": 8.474228567199276e-05,
      "loss": 1.9904,
      "step": 173891
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.3944547176361084,
      "learning_rate": 8.47394364844875e-05,
      "loss": 2.0315,
      "step": 173892
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.2201505899429321,
      "learning_rate": 8.473658733700326e-05,
      "loss": 1.8238,
      "step": 173893
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.334538459777832,
      "learning_rate": 8.473373822954044e-05,
      "loss": 1.9829,
      "step": 173894
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.2581837177276611,
      "learning_rate": 8.473088916209981e-05,
      "loss": 1.8515,
      "step": 173895
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.2698495388031006,
      "learning_rate": 8.472804013468176e-05,
      "loss": 1.9818,
      "step": 173896
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.3055346012115479,
      "learning_rate": 8.472519114728672e-05,
      "loss": 2.1813,
      "step": 173897
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.4580103158950806,
      "learning_rate": 8.472234219991541e-05,
      "loss": 2.0694,
      "step": 173898
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.3056252002716064,
      "learning_rate": 8.47194932925683e-05,
      "loss": 2.1227,
      "step": 173899
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.2678509950637817,
      "learning_rate": 8.471664442524583e-05,
      "loss": 1.95,
      "step": 173900
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.531558871269226,
      "learning_rate": 8.471379559794866e-05,
      "loss": 1.7777,
      "step": 173901
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.317181944847107,
      "learning_rate": 8.471094681067718e-05,
      "loss": 1.982,
      "step": 173902
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.2550047636032104,
      "learning_rate": 8.470809806343208e-05,
      "loss": 1.8207,
      "step": 173903
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.2843514680862427,
      "learning_rate": 8.470524935621383e-05,
      "loss": 2.2337,
      "step": 173904
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.461787223815918,
      "learning_rate": 8.470240068902293e-05,
      "loss": 1.9925,
      "step": 173905
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.5078628063201904,
      "learning_rate": 8.46995520618598e-05,
      "loss": 2.0863,
      "step": 173906
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.321554183959961,
      "learning_rate": 8.469670347472524e-05,
      "loss": 2.0252,
      "step": 173907
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.272384762763977,
      "learning_rate": 8.469385492761947e-05,
      "loss": 1.9744,
      "step": 173908
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.2892379760742188,
      "learning_rate": 8.469100642054329e-05,
      "loss": 2.2303,
      "step": 173909
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.3503456115722656,
      "learning_rate": 8.468815795349712e-05,
      "loss": 1.9278,
      "step": 173910
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.283227562904358,
      "learning_rate": 8.46853095264815e-05,
      "loss": 1.8074,
      "step": 173911
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.6409072875976562,
      "learning_rate": 8.468246113949685e-05,
      "loss": 2.0581,
      "step": 173912
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.2972244024276733,
      "learning_rate": 8.467961279254386e-05,
      "loss": 1.9031,
      "step": 173913
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.2843101024627686,
      "learning_rate": 8.467676448562295e-05,
      "loss": 2.085,
      "step": 173914
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.3055604696273804,
      "learning_rate": 8.467391621873473e-05,
      "loss": 2.0037,
      "step": 173915
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.3594504594802856,
      "learning_rate": 8.467106799187972e-05,
      "loss": 2.0383,
      "step": 173916
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.4786746501922607,
      "learning_rate": 8.466821980505843e-05,
      "loss": 2.0069,
      "step": 173917
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.270363450050354,
      "learning_rate": 8.466537165827129e-05,
      "loss": 2.0248,
      "step": 173918
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.3556010723114014,
      "learning_rate": 8.466252355151902e-05,
      "loss": 2.073,
      "step": 173919
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.3719134330749512,
      "learning_rate": 8.465967548480191e-05,
      "loss": 2.2415,
      "step": 173920
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.3626490831375122,
      "learning_rate": 8.465682745812079e-05,
      "loss": 2.0839,
      "step": 173921
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.3227529525756836,
      "learning_rate": 8.465397947147598e-05,
      "loss": 2.1109,
      "step": 173922
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.260249376296997,
      "learning_rate": 8.46511315248681e-05,
      "loss": 2.1417,
      "step": 173923
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.3245422840118408,
      "learning_rate": 8.464828361829753e-05,
      "loss": 2.0541,
      "step": 173924
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.2640405893325806,
      "learning_rate": 8.464543575176501e-05,
      "loss": 2.1001,
      "step": 173925
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.3125427961349487,
      "learning_rate": 8.464258792527084e-05,
      "loss": 2.0292,
      "step": 173926
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.3653295040130615,
      "learning_rate": 8.46397401388158e-05,
      "loss": 2.159,
      "step": 173927
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.343941330909729,
      "learning_rate": 8.463689239240031e-05,
      "loss": 2.0663,
      "step": 173928
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.2957185506820679,
      "learning_rate": 8.463404468602488e-05,
      "loss": 2.0594,
      "step": 173929
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.273343801498413,
      "learning_rate": 8.463119701968993e-05,
      "loss": 2.1025,
      "step": 173930
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.317862629890442,
      "learning_rate": 8.46283493933962e-05,
      "loss": 1.882,
      "step": 173931
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.1687755584716797,
      "learning_rate": 8.462550180714407e-05,
      "loss": 1.8986,
      "step": 173932
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.2851884365081787,
      "learning_rate": 8.462265426093417e-05,
      "loss": 2.1585,
      "step": 173933
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.3942593336105347,
      "learning_rate": 8.461980675476702e-05,
      "loss": 2.1533,
      "step": 173934
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.3469356298446655,
      "learning_rate": 8.46169592886431e-05,
      "loss": 1.9547,
      "step": 173935
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.2779592275619507,
      "learning_rate": 8.461411186256287e-05,
      "loss": 1.95,
      "step": 173936
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.2531795501708984,
      "learning_rate": 8.461126447652702e-05,
      "loss": 1.9394,
      "step": 173937
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.3266494274139404,
      "learning_rate": 8.46084171305359e-05,
      "loss": 1.8919,
      "step": 173938
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.24113929271698,
      "learning_rate": 8.460556982459028e-05,
      "loss": 1.8836,
      "step": 173939
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.3033992052078247,
      "learning_rate": 8.46027225586905e-05,
      "loss": 2.5097,
      "step": 173940
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.184088110923767,
      "learning_rate": 8.459987533283717e-05,
      "loss": 2.0251,
      "step": 173941
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.3865015506744385,
      "learning_rate": 8.459702814703069e-05,
      "loss": 2.1091,
      "step": 173942
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.2808136940002441,
      "learning_rate": 8.459418100127176e-05,
      "loss": 2.1632,
      "step": 173943
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.2836166620254517,
      "learning_rate": 8.459133389556078e-05,
      "loss": 2.0342,
      "step": 173944
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.1892119646072388,
      "learning_rate": 8.458848682989841e-05,
      "loss": 1.8117,
      "step": 173945
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.307969570159912,
      "learning_rate": 8.458563980428509e-05,
      "loss": 1.8669,
      "step": 173946
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.2073965072631836,
      "learning_rate": 8.458279281872138e-05,
      "loss": 1.9785,
      "step": 173947
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.2492481470108032,
      "learning_rate": 8.457994587320773e-05,
      "loss": 2.1042,
      "step": 173948
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.6166377067565918,
      "learning_rate": 8.457709896774482e-05,
      "loss": 2.3512,
      "step": 173949
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.4707244634628296,
      "learning_rate": 8.457425210233297e-05,
      "loss": 2.1303,
      "step": 173950
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.2876241207122803,
      "learning_rate": 8.457140527697292e-05,
      "loss": 1.9391,
      "step": 173951
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.2616255283355713,
      "learning_rate": 8.456855849166514e-05,
      "loss": 2.2038,
      "step": 173952
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.5401402711868286,
      "learning_rate": 8.456571174641014e-05,
      "loss": 2.1009,
      "step": 173953
    },
    {
      "epoch": 2.26,
      "grad_norm": 1.3781284093856812,
      "learning_rate": 8.456286504120832e-05,
      "loss": 2.1118,
      "step": 173954
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.3931505680084229,
      "learning_rate": 8.456001837606044e-05,
      "loss": 1.8653,
      "step": 173955
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.2947688102722168,
      "learning_rate": 8.455717175096684e-05,
      "loss": 1.7466,
      "step": 173956
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.2957096099853516,
      "learning_rate": 8.455432516592818e-05,
      "loss": 1.8431,
      "step": 173957
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.4260754585266113,
      "learning_rate": 8.455147862094489e-05,
      "loss": 2.1358,
      "step": 173958
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.3843690156936646,
      "learning_rate": 8.45486321160177e-05,
      "loss": 2.0108,
      "step": 173959
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.275740623474121,
      "learning_rate": 8.454578565114681e-05,
      "loss": 1.9883,
      "step": 173960
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.207785725593567,
      "learning_rate": 8.454293922633302e-05,
      "loss": 2.0485,
      "step": 173961
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.3303471803665161,
      "learning_rate": 8.454009284157667e-05,
      "loss": 1.9648,
      "step": 173962
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.3863393068313599,
      "learning_rate": 8.45372464968785e-05,
      "loss": 2.0062,
      "step": 173963
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.398040771484375,
      "learning_rate": 8.453440019223878e-05,
      "loss": 2.058,
      "step": 173964
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.3144690990447998,
      "learning_rate": 8.453155392765832e-05,
      "loss": 2.1999,
      "step": 173965
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.2358050346374512,
      "learning_rate": 8.45287077031375e-05,
      "loss": 1.9225,
      "step": 173966
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.282974362373352,
      "learning_rate": 8.452586151867684e-05,
      "loss": 2.0096,
      "step": 173967
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.3626351356506348,
      "learning_rate": 8.452301537427682e-05,
      "loss": 1.7499,
      "step": 173968
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.2692875862121582,
      "learning_rate": 8.452016926993814e-05,
      "loss": 1.8084,
      "step": 173969
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.3485167026519775,
      "learning_rate": 8.451732320566112e-05,
      "loss": 1.9383,
      "step": 173970
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.2513245344161987,
      "learning_rate": 8.45144771814465e-05,
      "loss": 2.0823,
      "step": 173971
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.3611018657684326,
      "learning_rate": 8.451163119729469e-05,
      "loss": 1.9311,
      "step": 173972
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.204472541809082,
      "learning_rate": 8.450878525320622e-05,
      "loss": 2.0682,
      "step": 173973
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.2810968160629272,
      "learning_rate": 8.450593934918159e-05,
      "loss": 1.9341,
      "step": 173974
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.2525502443313599,
      "learning_rate": 8.450309348522142e-05,
      "loss": 2.0211,
      "step": 173975
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.2713359594345093,
      "learning_rate": 8.450024766132612e-05,
      "loss": 2.0552,
      "step": 173976
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.3069336414337158,
      "learning_rate": 8.449740187749637e-05,
      "loss": 1.8594,
      "step": 173977
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.2957041263580322,
      "learning_rate": 8.449455613373263e-05,
      "loss": 1.7775,
      "step": 173978
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.5689373016357422,
      "learning_rate": 8.449171043003536e-05,
      "loss": 2.1511,
      "step": 173979
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.4929513931274414,
      "learning_rate": 8.448886476640522e-05,
      "loss": 2.1544,
      "step": 173980
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.4646601676940918,
      "learning_rate": 8.448601914284267e-05,
      "loss": 2.2992,
      "step": 173981
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.3380802869796753,
      "learning_rate": 8.448317355934812e-05,
      "loss": 1.971,
      "step": 173982
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.3341805934906006,
      "learning_rate": 8.448032801592234e-05,
      "loss": 2.1006,
      "step": 173983
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.2663761377334595,
      "learning_rate": 8.44774825125657e-05,
      "loss": 1.8681,
      "step": 173984
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.1133891344070435,
      "learning_rate": 8.44746370492787e-05,
      "loss": 1.9353,
      "step": 173985
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.3590484857559204,
      "learning_rate": 8.447179162606202e-05,
      "loss": 2.1455,
      "step": 173986
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.4063881635665894,
      "learning_rate": 8.446894624291599e-05,
      "loss": 1.9912,
      "step": 173987
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.458065390586853,
      "learning_rate": 8.446610089984137e-05,
      "loss": 2.0651,
      "step": 173988
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.2846533060073853,
      "learning_rate": 8.446325559683857e-05,
      "loss": 2.0927,
      "step": 173989
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.3687262535095215,
      "learning_rate": 8.44604103339081e-05,
      "loss": 2.01,
      "step": 173990
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.2248791456222534,
      "learning_rate": 8.445756511105045e-05,
      "loss": 2.1327,
      "step": 173991
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.195853590965271,
      "learning_rate": 8.445471992826628e-05,
      "loss": 1.9201,
      "step": 173992
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.2896485328674316,
      "learning_rate": 8.445187478555595e-05,
      "loss": 2.0954,
      "step": 173993
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.4980350732803345,
      "learning_rate": 8.444902968292016e-05,
      "loss": 2.1073,
      "step": 173994
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.193183422088623,
      "learning_rate": 8.444618462035939e-05,
      "loss": 1.8434,
      "step": 173995
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.210900902748108,
      "learning_rate": 8.444333959787413e-05,
      "loss": 2.0816,
      "step": 173996
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.2189418077468872,
      "learning_rate": 8.444049461546484e-05,
      "loss": 1.8601,
      "step": 173997
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.2511249780654907,
      "learning_rate": 8.443764967313222e-05,
      "loss": 2.1726,
      "step": 173998
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.2813814878463745,
      "learning_rate": 8.443480477087662e-05,
      "loss": 2.0947,
      "step": 173999
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.231086015701294,
      "learning_rate": 8.443195990869876e-05,
      "loss": 2.0766,
      "step": 174000
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.4662946462631226,
      "learning_rate": 8.442911508659907e-05,
      "loss": 2.0801,
      "step": 174001
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.3247472047805786,
      "learning_rate": 8.442627030457804e-05,
      "loss": 2.1773,
      "step": 174002
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.2889955043792725,
      "learning_rate": 8.442342556263617e-05,
      "loss": 2.005,
      "step": 174003
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.3025940656661987,
      "learning_rate": 8.442058086077413e-05,
      "loss": 2.0755,
      "step": 174004
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.2852869033813477,
      "learning_rate": 8.44177361989923e-05,
      "loss": 2.0191,
      "step": 174005
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.2745107412338257,
      "learning_rate": 8.441489157729137e-05,
      "loss": 1.9977,
      "step": 174006
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.3370436429977417,
      "learning_rate": 8.441204699567179e-05,
      "loss": 1.8255,
      "step": 174007
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.3525296449661255,
      "learning_rate": 8.440920245413408e-05,
      "loss": 1.9681,
      "step": 174008
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.2605934143066406,
      "learning_rate": 8.440635795267867e-05,
      "loss": 1.9699,
      "step": 174009
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.3725073337554932,
      "learning_rate": 8.440351349130625e-05,
      "loss": 2.0819,
      "step": 174010
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.2575215101242065,
      "learning_rate": 8.440066907001725e-05,
      "loss": 1.951,
      "step": 174011
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.2509772777557373,
      "learning_rate": 8.439782468881227e-05,
      "loss": 2.1511,
      "step": 174012
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.3186556100845337,
      "learning_rate": 8.439498034769184e-05,
      "loss": 1.9945,
      "step": 174013
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.3511170148849487,
      "learning_rate": 8.439213604665644e-05,
      "loss": 1.9702,
      "step": 174014
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.340985655784607,
      "learning_rate": 8.438929178570655e-05,
      "loss": 1.8456,
      "step": 174015
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.3077634572982788,
      "learning_rate": 8.438644756484281e-05,
      "loss": 2.0296,
      "step": 174016
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.2542763948440552,
      "learning_rate": 8.438360338406562e-05,
      "loss": 2.2074,
      "step": 174017
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.6221386194229126,
      "learning_rate": 8.438075924337569e-05,
      "loss": 2.1242,
      "step": 174018
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.309447169303894,
      "learning_rate": 8.437791514277347e-05,
      "loss": 1.9235,
      "step": 174019
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.2089592218399048,
      "learning_rate": 8.43750710822594e-05,
      "loss": 2.0433,
      "step": 174020
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.2748160362243652,
      "learning_rate": 8.437222706183404e-05,
      "loss": 2.1126,
      "step": 174021
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.5101977586746216,
      "learning_rate": 8.436938308149805e-05,
      "loss": 1.8981,
      "step": 174022
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.2836552858352661,
      "learning_rate": 8.436653914125172e-05,
      "loss": 1.7953,
      "step": 174023
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.4443656206130981,
      "learning_rate": 8.436369524109585e-05,
      "loss": 2.1886,
      "step": 174024
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.2070711851119995,
      "learning_rate": 8.436085138103074e-05,
      "loss": 2.1577,
      "step": 174025
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.281373143196106,
      "learning_rate": 8.435800756105719e-05,
      "loss": 2.0618,
      "step": 174026
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.2096842527389526,
      "learning_rate": 8.435516378117538e-05,
      "loss": 2.0772,
      "step": 174027
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.368678092956543,
      "learning_rate": 8.43523200413861e-05,
      "loss": 2.0103,
      "step": 174028
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.2265523672103882,
      "learning_rate": 8.434947634168971e-05,
      "loss": 2.1357,
      "step": 174029
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.3799145221710205,
      "learning_rate": 8.434663268208692e-05,
      "loss": 2.2107,
      "step": 174030
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.3594141006469727,
      "learning_rate": 8.43437890625781e-05,
      "loss": 2.0117,
      "step": 174031
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.2202866077423096,
      "learning_rate": 8.434094548316395e-05,
      "loss": 2.0481,
      "step": 174032
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.2598586082458496,
      "learning_rate": 8.433810194384476e-05,
      "loss": 1.9362,
      "step": 174033
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.2866506576538086,
      "learning_rate": 8.433525844462127e-05,
      "loss": 2.0209,
      "step": 174034
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.2199883460998535,
      "learning_rate": 8.433241498549385e-05,
      "loss": 1.9457,
      "step": 174035
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.4585566520690918,
      "learning_rate": 8.432957156646314e-05,
      "loss": 2.0722,
      "step": 174036
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.4476814270019531,
      "learning_rate": 8.432672818752961e-05,
      "loss": 1.9334,
      "step": 174037
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.4444553852081299,
      "learning_rate": 8.432388484869396e-05,
      "loss": 1.9878,
      "step": 174038
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.3581315279006958,
      "learning_rate": 8.432104154995638e-05,
      "loss": 2.18,
      "step": 174039
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.4029139280319214,
      "learning_rate": 8.431819829131772e-05,
      "loss": 1.9755,
      "step": 174040
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.3718039989471436,
      "learning_rate": 8.431535507277825e-05,
      "loss": 2.2144,
      "step": 174041
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.5031988620758057,
      "learning_rate": 8.431251189433873e-05,
      "loss": 1.8968,
      "step": 174042
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.238455057144165,
      "learning_rate": 8.430966875599947e-05,
      "loss": 1.8975,
      "step": 174043
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.263730525970459,
      "learning_rate": 8.430682565776133e-05,
      "loss": 2.3277,
      "step": 174044
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.3819364309310913,
      "learning_rate": 8.430398259962445e-05,
      "loss": 2.0348,
      "step": 174045
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.3843950033187866,
      "learning_rate": 8.430113958158959e-05,
      "loss": 1.9511,
      "step": 174046
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.3371385335922241,
      "learning_rate": 8.429829660365716e-05,
      "loss": 1.8564,
      "step": 174047
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.2876440286636353,
      "learning_rate": 8.429545366582781e-05,
      "loss": 1.8906,
      "step": 174048
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.3754292726516724,
      "learning_rate": 8.42926107681019e-05,
      "loss": 2.143,
      "step": 174049
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.2388718128204346,
      "learning_rate": 8.42897679104802e-05,
      "loss": 2.1323,
      "step": 174050
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.3425627946853638,
      "learning_rate": 8.428692509296307e-05,
      "loss": 1.9152,
      "step": 174051
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.450018286705017,
      "learning_rate": 8.42840823155511e-05,
      "loss": 1.8975,
      "step": 174052
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.435659646987915,
      "learning_rate": 8.428123957824468e-05,
      "loss": 2.1753,
      "step": 174053
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.4298361539840698,
      "learning_rate": 8.427839688104456e-05,
      "loss": 1.9263,
      "step": 174054
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.3262929916381836,
      "learning_rate": 8.427555422395106e-05,
      "loss": 1.9858,
      "step": 174055
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.2204220294952393,
      "learning_rate": 8.42727116069649e-05,
      "loss": 1.926,
      "step": 174056
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.3512850999832153,
      "learning_rate": 8.426986903008653e-05,
      "loss": 2.2543,
      "step": 174057
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.4619730710983276,
      "learning_rate": 8.426702649331642e-05,
      "loss": 2.0365,
      "step": 174058
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.3406256437301636,
      "learning_rate": 8.426418399665508e-05,
      "loss": 2.0171,
      "step": 174059
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.279046893119812,
      "learning_rate": 8.426134154010318e-05,
      "loss": 1.8513,
      "step": 174060
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.3203284740447998,
      "learning_rate": 8.42584991236611e-05,
      "loss": 2.1517,
      "step": 174061
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.315302848815918,
      "learning_rate": 8.425565674732949e-05,
      "loss": 2.0088,
      "step": 174062
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.373424768447876,
      "learning_rate": 8.425281441110887e-05,
      "loss": 1.9853,
      "step": 174063
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.3445930480957031,
      "learning_rate": 8.42499721149996e-05,
      "loss": 2.1651,
      "step": 174064
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.241729497909546,
      "learning_rate": 8.424712985900244e-05,
      "loss": 2.0703,
      "step": 174065
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.6505928039550781,
      "learning_rate": 8.42442876431178e-05,
      "loss": 2.0439,
      "step": 174066
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.338078260421753,
      "learning_rate": 8.424144546734612e-05,
      "loss": 2.0557,
      "step": 174067
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.2996309995651245,
      "learning_rate": 8.423860333168813e-05,
      "loss": 2.2467,
      "step": 174068
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.2178168296813965,
      "learning_rate": 8.423576123614429e-05,
      "loss": 2.1438,
      "step": 174069
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.2654027938842773,
      "learning_rate": 8.423291918071497e-05,
      "loss": 2.2511,
      "step": 174070
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.4878079891204834,
      "learning_rate": 8.423007716540091e-05,
      "loss": 2.0929,
      "step": 174071
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.2899962663650513,
      "learning_rate": 8.422723519020256e-05,
      "loss": 2.2301,
      "step": 174072
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.3130918741226196,
      "learning_rate": 8.422439325512037e-05,
      "loss": 2.0425,
      "step": 174073
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.4327442646026611,
      "learning_rate": 8.422155136015498e-05,
      "loss": 2.0342,
      "step": 174074
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.3196487426757812,
      "learning_rate": 8.421870950530694e-05,
      "loss": 2.0907,
      "step": 174075
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.252160668373108,
      "learning_rate": 8.421586769057657e-05,
      "loss": 2.238,
      "step": 174076
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.3854182958602905,
      "learning_rate": 8.421302591596467e-05,
      "loss": 2.1215,
      "step": 174077
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.2771666049957275,
      "learning_rate": 8.421018418147152e-05,
      "loss": 2.1462,
      "step": 174078
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.2151728868484497,
      "learning_rate": 8.420734248709787e-05,
      "loss": 1.9745,
      "step": 174079
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.2292232513427734,
      "learning_rate": 8.420450083284417e-05,
      "loss": 2.0213,
      "step": 174080
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.415745735168457,
      "learning_rate": 8.42016592187109e-05,
      "loss": 2.1161,
      "step": 174081
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.3846688270568848,
      "learning_rate": 8.419881764469855e-05,
      "loss": 2.285,
      "step": 174082
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.3117836713790894,
      "learning_rate": 8.41959761108078e-05,
      "loss": 2.0528,
      "step": 174083
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.229401707649231,
      "learning_rate": 8.419313461703898e-05,
      "loss": 2.0289,
      "step": 174084
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.2582987546920776,
      "learning_rate": 8.41902931633928e-05,
      "loss": 2.0872,
      "step": 174085
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.4352351427078247,
      "learning_rate": 8.418745174986977e-05,
      "loss": 1.9378,
      "step": 174086
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.253229022026062,
      "learning_rate": 8.418461037647035e-05,
      "loss": 1.9939,
      "step": 174087
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.2544965744018555,
      "learning_rate": 8.418176904319498e-05,
      "loss": 1.9881,
      "step": 174088
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.2351994514465332,
      "learning_rate": 8.417892775004438e-05,
      "loss": 2.0281,
      "step": 174089
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.2631525993347168,
      "learning_rate": 8.417608649701894e-05,
      "loss": 1.8897,
      "step": 174090
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.3427894115447998,
      "learning_rate": 8.417324528411928e-05,
      "loss": 2.0612,
      "step": 174091
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.295520544052124,
      "learning_rate": 8.417040411134583e-05,
      "loss": 2.1148,
      "step": 174092
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.3700729608535767,
      "learning_rate": 8.416756297869935e-05,
      "loss": 2.1454,
      "step": 174093
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.2446398735046387,
      "learning_rate": 8.416472188618001e-05,
      "loss": 1.9631,
      "step": 174094
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.2813799381256104,
      "learning_rate": 8.416188083378864e-05,
      "loss": 2.0386,
      "step": 174095
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.2380461692810059,
      "learning_rate": 8.415903982152552e-05,
      "loss": 1.8861,
      "step": 174096
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.2892078161239624,
      "learning_rate": 8.41561988493914e-05,
      "loss": 2.0371,
      "step": 174097
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.2394684553146362,
      "learning_rate": 8.415335791738667e-05,
      "loss": 2.1077,
      "step": 174098
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.3938466310501099,
      "learning_rate": 8.415051702551206e-05,
      "loss": 2.029,
      "step": 174099
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.3789174556732178,
      "learning_rate": 8.414767617376774e-05,
      "loss": 1.8157,
      "step": 174100
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.307814359664917,
      "learning_rate": 8.414483536215456e-05,
      "loss": 1.9316,
      "step": 174101
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.26777184009552,
      "learning_rate": 8.414199459067281e-05,
      "loss": 1.921,
      "step": 174102
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.3517454862594604,
      "learning_rate": 8.413915385932326e-05,
      "loss": 2.1514,
      "step": 174103
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.276179552078247,
      "learning_rate": 8.41363131681062e-05,
      "loss": 1.8755,
      "step": 174104
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.3955363035202026,
      "learning_rate": 8.413347251702249e-05,
      "loss": 2.0557,
      "step": 174105
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.5743417739868164,
      "learning_rate": 8.413063190607222e-05,
      "loss": 1.9911,
      "step": 174106
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.3139530420303345,
      "learning_rate": 8.412779133525625e-05,
      "loss": 1.9746,
      "step": 174107
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.221354365348816,
      "learning_rate": 8.412495080457492e-05,
      "loss": 2.084,
      "step": 174108
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.2504562139511108,
      "learning_rate": 8.412211031402891e-05,
      "loss": 1.9621,
      "step": 174109
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.3712917566299438,
      "learning_rate": 8.41192698636186e-05,
      "loss": 1.9388,
      "step": 174110
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.405170202255249,
      "learning_rate": 8.411642945334479e-05,
      "loss": 2.0999,
      "step": 174111
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.1620986461639404,
      "learning_rate": 8.411358908320759e-05,
      "loss": 1.7985,
      "step": 174112
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.551588773727417,
      "learning_rate": 8.411074875320785e-05,
      "loss": 2.0037,
      "step": 174113
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.258610725402832,
      "learning_rate": 8.410790846334592e-05,
      "loss": 2.1118,
      "step": 174114
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.2948213815689087,
      "learning_rate": 8.410506821362248e-05,
      "loss": 2.1914,
      "step": 174115
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.5540539026260376,
      "learning_rate": 8.410222800403788e-05,
      "loss": 2.0176,
      "step": 174116
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.4529681205749512,
      "learning_rate": 8.409938783459295e-05,
      "loss": 1.9843,
      "step": 174117
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.3016067743301392,
      "learning_rate": 8.409654770528785e-05,
      "loss": 2.2827,
      "step": 174118
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.363927960395813,
      "learning_rate": 8.409370761612338e-05,
      "loss": 2.0402,
      "step": 174119
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.4115252494812012,
      "learning_rate": 8.409086756709981e-05,
      "loss": 1.9088,
      "step": 174120
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.3567708730697632,
      "learning_rate": 8.408802755821798e-05,
      "loss": 1.7734,
      "step": 174121
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.2030407190322876,
      "learning_rate": 8.408518758947814e-05,
      "loss": 1.9207,
      "step": 174122
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.3137563467025757,
      "learning_rate": 8.408234766088112e-05,
      "loss": 1.9962,
      "step": 174123
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.3422189950942993,
      "learning_rate": 8.407950777242708e-05,
      "loss": 2.0102,
      "step": 174124
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.4791070222854614,
      "learning_rate": 8.407666792411688e-05,
      "loss": 2.1265,
      "step": 174125
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.3270211219787598,
      "learning_rate": 8.407382811595076e-05,
      "loss": 2.1003,
      "step": 174126
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.3521556854248047,
      "learning_rate": 8.407098834792949e-05,
      "loss": 2.2401,
      "step": 174127
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.3712012767791748,
      "learning_rate": 8.40681486200534e-05,
      "loss": 1.9618,
      "step": 174128
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.369035005569458,
      "learning_rate": 8.406530893232332e-05,
      "loss": 2.1345,
      "step": 174129
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.3333021402359009,
      "learning_rate": 8.406246928473937e-05,
      "loss": 2.1026,
      "step": 174130
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.2577366828918457,
      "learning_rate": 8.405962967730242e-05,
      "loss": 2.1866,
      "step": 174131
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.2634421586990356,
      "learning_rate": 8.405679011001272e-05,
      "loss": 2.1621,
      "step": 174132
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.4966380596160889,
      "learning_rate": 8.405395058287106e-05,
      "loss": 2.1658,
      "step": 174133
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.338289499282837,
      "learning_rate": 8.405111109587774e-05,
      "loss": 1.8443,
      "step": 174134
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.299072027206421,
      "learning_rate": 8.404827164903358e-05,
      "loss": 1.9196,
      "step": 174135
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.1300675868988037,
      "learning_rate": 8.404543224233875e-05,
      "loss": 1.8294,
      "step": 174136
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.346459984779358,
      "learning_rate": 8.404259287579404e-05,
      "loss": 1.9403,
      "step": 174137
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.2931181192398071,
      "learning_rate": 8.403975354939977e-05,
      "loss": 1.9994,
      "step": 174138
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.321071743965149,
      "learning_rate": 8.403691426315672e-05,
      "loss": 2.0242,
      "step": 174139
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.254978060722351,
      "learning_rate": 8.403407501706515e-05,
      "loss": 2.0806,
      "step": 174140
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.3205524682998657,
      "learning_rate": 8.403123581112583e-05,
      "loss": 2.0573,
      "step": 174141
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.2986748218536377,
      "learning_rate": 8.402839664533919e-05,
      "loss": 1.9654,
      "step": 174142
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.3015801906585693,
      "learning_rate": 8.402555751970573e-05,
      "loss": 2.2554,
      "step": 174143
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.2445472478866577,
      "learning_rate": 8.402271843422592e-05,
      "loss": 1.974,
      "step": 174144
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.418651819229126,
      "learning_rate": 8.401987938890045e-05,
      "loss": 2.0664,
      "step": 174145
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.2730928659439087,
      "learning_rate": 8.401704038372968e-05,
      "loss": 2.0182,
      "step": 174146
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.4656325578689575,
      "learning_rate": 8.40142014187143e-05,
      "loss": 2.0619,
      "step": 174147
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.2559524774551392,
      "learning_rate": 8.401136249385478e-05,
      "loss": 1.9158,
      "step": 174148
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.2530663013458252,
      "learning_rate": 8.400852360915149e-05,
      "loss": 1.9894,
      "step": 174149
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.2780908346176147,
      "learning_rate": 8.400568476460521e-05,
      "loss": 2.1503,
      "step": 174150
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.265594482421875,
      "learning_rate": 8.400284596021636e-05,
      "loss": 1.9519,
      "step": 174151
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.3081657886505127,
      "learning_rate": 8.400000719598534e-05,
      "loss": 2.0701,
      "step": 174152
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.310868501663208,
      "learning_rate": 8.39971684719129e-05,
      "loss": 2.2688,
      "step": 174153
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.3212641477584839,
      "learning_rate": 8.399432978799948e-05,
      "loss": 2.2893,
      "step": 174154
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.2800014019012451,
      "learning_rate": 8.399149114424549e-05,
      "loss": 1.8629,
      "step": 174155
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.28244948387146,
      "learning_rate": 8.398865254065164e-05,
      "loss": 2.0138,
      "step": 174156
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.2293546199798584,
      "learning_rate": 8.398581397721837e-05,
      "loss": 2.0704,
      "step": 174157
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.2935489416122437,
      "learning_rate": 8.398297545394613e-05,
      "loss": 2.1115,
      "step": 174158
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.2124332189559937,
      "learning_rate": 8.398013697083564e-05,
      "loss": 2.0517,
      "step": 174159
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.27047860622406,
      "learning_rate": 8.397729852788733e-05,
      "loss": 2.0839,
      "step": 174160
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.2679835557937622,
      "learning_rate": 8.397446012510159e-05,
      "loss": 2.0748,
      "step": 174161
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.2305361032485962,
      "learning_rate": 8.397162176247918e-05,
      "loss": 1.9302,
      "step": 174162
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.311766266822815,
      "learning_rate": 8.396878344002045e-05,
      "loss": 2.169,
      "step": 174163
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.2418174743652344,
      "learning_rate": 8.396594515772605e-05,
      "loss": 2.0096,
      "step": 174164
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.446079969406128,
      "learning_rate": 8.39631069155965e-05,
      "loss": 1.9971,
      "step": 174165
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.335724949836731,
      "learning_rate": 8.396026871363228e-05,
      "loss": 1.9749,
      "step": 174166
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.2657867670059204,
      "learning_rate": 8.395743055183383e-05,
      "loss": 2.1561,
      "step": 174167
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.3163448572158813,
      "learning_rate": 8.395459243020186e-05,
      "loss": 2.1658,
      "step": 174168
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.359613299369812,
      "learning_rate": 8.395175434873672e-05,
      "loss": 1.8696,
      "step": 174169
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.202371597290039,
      "learning_rate": 8.394891630743912e-05,
      "loss": 2.0608,
      "step": 174170
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.4669803380966187,
      "learning_rate": 8.394607830630938e-05,
      "loss": 2.1502,
      "step": 174171
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.358768343925476,
      "learning_rate": 8.394324034534835e-05,
      "loss": 2.2426,
      "step": 174172
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.216278076171875,
      "learning_rate": 8.394040242455615e-05,
      "loss": 1.8761,
      "step": 174173
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.3149123191833496,
      "learning_rate": 8.393756454393362e-05,
      "loss": 2.0449,
      "step": 174174
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.312768816947937,
      "learning_rate": 8.39347267034811e-05,
      "loss": 1.9954,
      "step": 174175
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.28616201877594,
      "learning_rate": 8.393188890319922e-05,
      "loss": 2.1749,
      "step": 174176
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.1858843564987183,
      "learning_rate": 8.392905114308844e-05,
      "loss": 1.8817,
      "step": 174177
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.3377723693847656,
      "learning_rate": 8.39262134231495e-05,
      "loss": 1.8591,
      "step": 174178
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.255806565284729,
      "learning_rate": 8.392337574338256e-05,
      "loss": 1.8697,
      "step": 174179
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.3928080797195435,
      "learning_rate": 8.392053810378847e-05,
      "loss": 2.0293,
      "step": 174180
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.2629200220108032,
      "learning_rate": 8.39177005043675e-05,
      "loss": 2.1344,
      "step": 174181
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.3838670253753662,
      "learning_rate": 8.391486294512043e-05,
      "loss": 1.9109,
      "step": 174182
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.5595394372940063,
      "learning_rate": 8.391202542604756e-05,
      "loss": 2.1467,
      "step": 174183
    },
    {
      "epoch": 2.27,
      "grad_norm": 3.1355135440826416,
      "learning_rate": 8.390918794714972e-05,
      "loss": 2.2214,
      "step": 174184
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.390489101409912,
      "learning_rate": 8.390635050842703e-05,
      "loss": 1.9431,
      "step": 174185
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.2536660432815552,
      "learning_rate": 8.390351310988032e-05,
      "loss": 1.7069,
      "step": 174186
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.2598427534103394,
      "learning_rate": 8.390067575150996e-05,
      "loss": 1.9207,
      "step": 174187
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.2686855792999268,
      "learning_rate": 8.389783843331663e-05,
      "loss": 2.1825,
      "step": 174188
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.2439848184585571,
      "learning_rate": 8.389500115530066e-05,
      "loss": 2.1057,
      "step": 174189
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.382153034210205,
      "learning_rate": 8.38921639174629e-05,
      "loss": 2.1409,
      "step": 174190
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.3702017068862915,
      "learning_rate": 8.388932671980346e-05,
      "loss": 1.8642,
      "step": 174191
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.3672469854354858,
      "learning_rate": 8.388648956232316e-05,
      "loss": 2.1404,
      "step": 174192
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.2169346809387207,
      "learning_rate": 8.388365244502238e-05,
      "loss": 2.1379,
      "step": 174193
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.2863727807998657,
      "learning_rate": 8.388081536790178e-05,
      "loss": 1.9772,
      "step": 174194
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.3169103860855103,
      "learning_rate": 8.387797833096173e-05,
      "loss": 1.9909,
      "step": 174195
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.377699851989746,
      "learning_rate": 8.387514133420303e-05,
      "loss": 2.1325,
      "step": 174196
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.2264577150344849,
      "learning_rate": 8.387230437762583e-05,
      "loss": 1.7499,
      "step": 174197
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.165460228919983,
      "learning_rate": 8.386946746123093e-05,
      "loss": 2.0237,
      "step": 174198
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.233052134513855,
      "learning_rate": 8.386663058501869e-05,
      "loss": 1.8431,
      "step": 174199
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.3884421586990356,
      "learning_rate": 8.386379374898984e-05,
      "loss": 2.0121,
      "step": 174200
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.2170079946517944,
      "learning_rate": 8.386095695314467e-05,
      "loss": 2.0813,
      "step": 174201
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.2855316400527954,
      "learning_rate": 8.385812019748404e-05,
      "loss": 2.02,
      "step": 174202
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.2871664762496948,
      "learning_rate": 8.385528348200805e-05,
      "loss": 1.893,
      "step": 174203
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.4482113122940063,
      "learning_rate": 8.385244680671756e-05,
      "loss": 2.3131,
      "step": 174204
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.2018013000488281,
      "learning_rate": 8.384961017161285e-05,
      "loss": 2.1365,
      "step": 174205
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.1211670637130737,
      "learning_rate": 8.38467735766947e-05,
      "loss": 1.9468,
      "step": 174206
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.2593894004821777,
      "learning_rate": 8.384393702196343e-05,
      "loss": 1.9038,
      "step": 174207
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.2839491367340088,
      "learning_rate": 8.384110050741981e-05,
      "loss": 1.8472,
      "step": 174208
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.404987096786499,
      "learning_rate": 8.383826403306405e-05,
      "loss": 1.9641,
      "step": 174209
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.3092249631881714,
      "learning_rate": 8.383542759889691e-05,
      "loss": 1.9818,
      "step": 174210
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.2288464307785034,
      "learning_rate": 8.383259120491875e-05,
      "loss": 2.1379,
      "step": 174211
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.3126835823059082,
      "learning_rate": 8.382975485113032e-05,
      "loss": 2.1276,
      "step": 174212
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.3666658401489258,
      "learning_rate": 8.382691853753189e-05,
      "loss": 1.9851,
      "step": 174213
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.4095591306686401,
      "learning_rate": 8.382408226412432e-05,
      "loss": 1.8927,
      "step": 174214
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.1755768060684204,
      "learning_rate": 8.382124603090775e-05,
      "loss": 1.9299,
      "step": 174215
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.2817893028259277,
      "learning_rate": 8.381840983788297e-05,
      "loss": 2.096,
      "step": 174216
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.3087103366851807,
      "learning_rate": 8.381557368505035e-05,
      "loss": 1.9052,
      "step": 174217
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.419306755065918,
      "learning_rate": 8.381273757241058e-05,
      "loss": 2.0514,
      "step": 174218
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.577613353729248,
      "learning_rate": 8.380990149996399e-05,
      "loss": 1.9765,
      "step": 174219
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.4816088676452637,
      "learning_rate": 8.380706546771146e-05,
      "loss": 2.2278,
      "step": 174220
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.4039268493652344,
      "learning_rate": 8.380422947565302e-05,
      "loss": 2.0398,
      "step": 174221
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.261406421661377,
      "learning_rate": 8.380139352378961e-05,
      "loss": 2.1565,
      "step": 174222
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.2807538509368896,
      "learning_rate": 8.379855761212149e-05,
      "loss": 1.9712,
      "step": 174223
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.4125570058822632,
      "learning_rate": 8.379572174064936e-05,
      "loss": 2.0311,
      "step": 174224
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.4047319889068604,
      "learning_rate": 8.379288590937362e-05,
      "loss": 2.0554,
      "step": 174225
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.3481515645980835,
      "learning_rate": 8.379005011829497e-05,
      "loss": 1.851,
      "step": 174226
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.3207635879516602,
      "learning_rate": 8.378721436741382e-05,
      "loss": 2.118,
      "step": 174227
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.2480754852294922,
      "learning_rate": 8.378437865673072e-05,
      "loss": 1.8842,
      "step": 174228
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.283289909362793,
      "learning_rate": 8.378154298624607e-05,
      "loss": 1.7989,
      "step": 174229
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.4398006200790405,
      "learning_rate": 8.377870735596061e-05,
      "loss": 1.9112,
      "step": 174230
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.5929099321365356,
      "learning_rate": 8.377587176587467e-05,
      "loss": 2.1871,
      "step": 174231
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.2206830978393555,
      "learning_rate": 8.377303621598899e-05,
      "loss": 1.9814,
      "step": 174232
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.232003927230835,
      "learning_rate": 8.377020070630397e-05,
      "loss": 1.8921,
      "step": 174233
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.3501802682876587,
      "learning_rate": 8.376736523682016e-05,
      "loss": 1.9254,
      "step": 174234
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.5444488525390625,
      "learning_rate": 8.376452980753799e-05,
      "loss": 2.0346,
      "step": 174235
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.2918527126312256,
      "learning_rate": 8.376169441845817e-05,
      "loss": 1.9076,
      "step": 174236
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.3597744703292847,
      "learning_rate": 8.375885906958102e-05,
      "loss": 1.9879,
      "step": 174237
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.1596345901489258,
      "learning_rate": 8.375602376090728e-05,
      "loss": 1.9096,
      "step": 174238
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.2838447093963623,
      "learning_rate": 8.375318849243738e-05,
      "loss": 1.7895,
      "step": 174239
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.2917102575302124,
      "learning_rate": 8.375035326417174e-05,
      "loss": 1.9694,
      "step": 174240
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.2650474309921265,
      "learning_rate": 8.374751807611111e-05,
      "loss": 2.0596,
      "step": 174241
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.309831142425537,
      "learning_rate": 8.37446829282559e-05,
      "loss": 1.9209,
      "step": 174242
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.2279691696166992,
      "learning_rate": 8.374184782060652e-05,
      "loss": 2.0864,
      "step": 174243
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.3023526668548584,
      "learning_rate": 8.373901275316371e-05,
      "loss": 1.9296,
      "step": 174244
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.575247049331665,
      "learning_rate": 8.373617772592791e-05,
      "loss": 1.8579,
      "step": 174245
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.3693290948867798,
      "learning_rate": 8.373334273889955e-05,
      "loss": 1.8685,
      "step": 174246
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.3477590084075928,
      "learning_rate": 8.373050779207931e-05,
      "loss": 1.8726,
      "step": 174247
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.32486891746521,
      "learning_rate": 8.372767288546755e-05,
      "loss": 2.0891,
      "step": 174248
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.2571115493774414,
      "learning_rate": 8.372483801906503e-05,
      "loss": 2.0519,
      "step": 174249
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.4660650491714478,
      "learning_rate": 8.372200319287211e-05,
      "loss": 2.0499,
      "step": 174250
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.2680926322937012,
      "learning_rate": 8.371916840688939e-05,
      "loss": 2.0849,
      "step": 174251
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.4170914888381958,
      "learning_rate": 8.371633366111725e-05,
      "loss": 2.1872,
      "step": 174252
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.3354065418243408,
      "learning_rate": 8.371349895555639e-05,
      "loss": 1.7649,
      "step": 174253
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.3027745485305786,
      "learning_rate": 8.37106642902072e-05,
      "loss": 1.9516,
      "step": 174254
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.2667535543441772,
      "learning_rate": 8.370782966507039e-05,
      "loss": 2.1573,
      "step": 174255
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.5578850507736206,
      "learning_rate": 8.370499508014637e-05,
      "loss": 2.184,
      "step": 174256
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.3406651020050049,
      "learning_rate": 8.370216053543569e-05,
      "loss": 2.179,
      "step": 174257
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.3072257041931152,
      "learning_rate": 8.369932603093874e-05,
      "loss": 1.761,
      "step": 174258
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.3306742906570435,
      "learning_rate": 8.369649156665626e-05,
      "loss": 2.0943,
      "step": 174259
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.1798654794692993,
      "learning_rate": 8.36936571425886e-05,
      "loss": 1.9012,
      "step": 174260
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.3487662076950073,
      "learning_rate": 8.369082275873647e-05,
      "loss": 2.0707,
      "step": 174261
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.1673293113708496,
      "learning_rate": 8.368798841510021e-05,
      "loss": 2.0043,
      "step": 174262
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.2937180995941162,
      "learning_rate": 8.368515411168062e-05,
      "loss": 1.9076,
      "step": 174263
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.3615176677703857,
      "learning_rate": 8.368231984847785e-05,
      "loss": 1.9995,
      "step": 174264
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.3988794088363647,
      "learning_rate": 8.367948562549271e-05,
      "loss": 1.858,
      "step": 174265
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.3182923793792725,
      "learning_rate": 8.367665144272557e-05,
      "loss": 2.0656,
      "step": 174266
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.2343844175338745,
      "learning_rate": 8.367381730017709e-05,
      "loss": 2.1605,
      "step": 174267
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.415592074394226,
      "learning_rate": 8.367098319784764e-05,
      "loss": 2.0564,
      "step": 174268
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.4619826078414917,
      "learning_rate": 8.366814913573802e-05,
      "loss": 2.0301,
      "step": 174269
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.4820096492767334,
      "learning_rate": 8.366531511384842e-05,
      "loss": 2.0639,
      "step": 174270
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.2724599838256836,
      "learning_rate": 8.366248113217959e-05,
      "loss": 1.9898,
      "step": 174271
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.3448268175125122,
      "learning_rate": 8.365964719073187e-05,
      "loss": 2.218,
      "step": 174272
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.3265608549118042,
      "learning_rate": 8.365681328950605e-05,
      "loss": 1.9009,
      "step": 174273
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.3433154821395874,
      "learning_rate": 8.365397942850239e-05,
      "loss": 2.1233,
      "step": 174274
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.3354545831680298,
      "learning_rate": 8.365114560772172e-05,
      "loss": 2.1198,
      "step": 174275
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.514760136604309,
      "learning_rate": 8.364831182716421e-05,
      "loss": 1.9955,
      "step": 174276
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.1886788606643677,
      "learning_rate": 8.364547808683067e-05,
      "loss": 2.0507,
      "step": 174277
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.3024985790252686,
      "learning_rate": 8.36426443867214e-05,
      "loss": 1.9775,
      "step": 174278
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.5752147436141968,
      "learning_rate": 8.363981072683716e-05,
      "loss": 1.7747,
      "step": 174279
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.2181737422943115,
      "learning_rate": 8.363697710717825e-05,
      "loss": 2.2406,
      "step": 174280
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.285843014717102,
      "learning_rate": 8.363414352774552e-05,
      "loss": 2.1467,
      "step": 174281
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.3369450569152832,
      "learning_rate": 8.363130998853908e-05,
      "loss": 1.9579,
      "step": 174282
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.3064312934875488,
      "learning_rate": 8.362847648955977e-05,
      "loss": 2.1865,
      "step": 174283
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.349865436553955,
      "learning_rate": 8.362564303080791e-05,
      "loss": 2.1307,
      "step": 174284
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.3958895206451416,
      "learning_rate": 8.362280961228423e-05,
      "loss": 1.8945,
      "step": 174285
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.3329867124557495,
      "learning_rate": 8.361997623398905e-05,
      "loss": 2.0161,
      "step": 174286
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.458625316619873,
      "learning_rate": 8.361714289592312e-05,
      "loss": 1.8399,
      "step": 174287
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.2131086587905884,
      "learning_rate": 8.361430959808681e-05,
      "loss": 1.9604,
      "step": 174288
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.353471279144287,
      "learning_rate": 8.361147634048071e-05,
      "loss": 1.8837,
      "step": 174289
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.4924858808517456,
      "learning_rate": 8.36086431231052e-05,
      "loss": 2.1899,
      "step": 174290
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.3155514001846313,
      "learning_rate": 8.360580994596105e-05,
      "loss": 1.9208,
      "step": 174291
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.3567960262298584,
      "learning_rate": 8.360297680904857e-05,
      "loss": 1.9715,
      "step": 174292
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.734336018562317,
      "learning_rate": 8.360014371236845e-05,
      "loss": 2.0569,
      "step": 174293
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.3122869729995728,
      "learning_rate": 8.359731065592117e-05,
      "loss": 2.0496,
      "step": 174294
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.6237573623657227,
      "learning_rate": 8.359447763970722e-05,
      "loss": 1.7914,
      "step": 174295
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.6463826894760132,
      "learning_rate": 8.359164466372708e-05,
      "loss": 2.0255,
      "step": 174296
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.3580888509750366,
      "learning_rate": 8.35888117279814e-05,
      "loss": 1.9886,
      "step": 174297
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.391000509262085,
      "learning_rate": 8.358597883247056e-05,
      "loss": 1.9641,
      "step": 174298
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.28505277633667,
      "learning_rate": 8.358314597719526e-05,
      "loss": 2.0463,
      "step": 174299
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.3271443843841553,
      "learning_rate": 8.358031316215595e-05,
      "loss": 1.9401,
      "step": 174300
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.62555730342865,
      "learning_rate": 8.357748038735312e-05,
      "loss": 1.9348,
      "step": 174301
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.3269226551055908,
      "learning_rate": 8.357464765278727e-05,
      "loss": 2.0158,
      "step": 174302
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.3849594593048096,
      "learning_rate": 8.357181495845904e-05,
      "loss": 2.037,
      "step": 174303
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.5855457782745361,
      "learning_rate": 8.356898230436881e-05,
      "loss": 2.0814,
      "step": 174304
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.2371773719787598,
      "learning_rate": 8.35661496905173e-05,
      "loss": 2.0246,
      "step": 174305
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.348986268043518,
      "learning_rate": 8.35633171169049e-05,
      "loss": 1.9935,
      "step": 174306
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.5398417711257935,
      "learning_rate": 8.356048458353219e-05,
      "loss": 2.1365,
      "step": 174307
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.5513641834259033,
      "learning_rate": 8.355765209039957e-05,
      "loss": 1.8762,
      "step": 174308
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.674613356590271,
      "learning_rate": 8.355481963750775e-05,
      "loss": 2.0228,
      "step": 174309
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.1590017080307007,
      "learning_rate": 8.355198722485709e-05,
      "loss": 2.026,
      "step": 174310
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.3131910562515259,
      "learning_rate": 8.354915485244831e-05,
      "loss": 2.0739,
      "step": 174311
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.6726375818252563,
      "learning_rate": 8.35463225202818e-05,
      "loss": 2.3298,
      "step": 174312
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.5043121576309204,
      "learning_rate": 8.354349022835811e-05,
      "loss": 1.9046,
      "step": 174313
    },
    {
      "epoch": 2.27,
      "grad_norm": 2.029449462890625,
      "learning_rate": 8.354065797667773e-05,
      "loss": 2.2402,
      "step": 174314
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.2969129085540771,
      "learning_rate": 8.353782576524128e-05,
      "loss": 2.2148,
      "step": 174315
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.2796227931976318,
      "learning_rate": 8.353499359404915e-05,
      "loss": 2.0475,
      "step": 174316
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.344344139099121,
      "learning_rate": 8.353216146310202e-05,
      "loss": 1.9949,
      "step": 174317
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.201488971710205,
      "learning_rate": 8.352932937240037e-05,
      "loss": 1.8347,
      "step": 174318
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.3489046096801758,
      "learning_rate": 8.352649732194472e-05,
      "loss": 2.1095,
      "step": 174319
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.2194349765777588,
      "learning_rate": 8.352366531173544e-05,
      "loss": 1.9168,
      "step": 174320
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.3466860055923462,
      "learning_rate": 8.352083334177334e-05,
      "loss": 1.9558,
      "step": 174321
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.6030590534210205,
      "learning_rate": 8.351800141205867e-05,
      "loss": 2.0978,
      "step": 174322
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.3212707042694092,
      "learning_rate": 8.351516952259221e-05,
      "loss": 2.1687,
      "step": 174323
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.379875659942627,
      "learning_rate": 8.351233767337437e-05,
      "loss": 2.0532,
      "step": 174324
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.135392427444458,
      "learning_rate": 8.350950586440558e-05,
      "loss": 2.0941,
      "step": 174325
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.2067052125930786,
      "learning_rate": 8.350667409568653e-05,
      "loss": 1.9732,
      "step": 174326
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.6190646886825562,
      "learning_rate": 8.350384236721769e-05,
      "loss": 2.1576,
      "step": 174327
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.268837571144104,
      "learning_rate": 8.350101067899949e-05,
      "loss": 2.1297,
      "step": 174328
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.2199281454086304,
      "learning_rate": 8.349817903103261e-05,
      "loss": 1.8772,
      "step": 174329
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.2961933612823486,
      "learning_rate": 8.34953474233175e-05,
      "loss": 1.891,
      "step": 174330
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.2725921869277954,
      "learning_rate": 8.349251585585463e-05,
      "loss": 2.1449,
      "step": 174331
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.325998067855835,
      "learning_rate": 8.348968432864467e-05,
      "loss": 1.8708,
      "step": 174332
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.2612484693527222,
      "learning_rate": 8.348685284168797e-05,
      "loss": 1.8774,
      "step": 174333
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.2238494157791138,
      "learning_rate": 8.348402139498523e-05,
      "loss": 2.0631,
      "step": 174334
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.2858699560165405,
      "learning_rate": 8.34811899885369e-05,
      "loss": 2.2378,
      "step": 174335
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.4967610836029053,
      "learning_rate": 8.34783586223435e-05,
      "loss": 1.924,
      "step": 174336
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.3340235948562622,
      "learning_rate": 8.34755272964055e-05,
      "loss": 2.4273,
      "step": 174337
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.2944204807281494,
      "learning_rate": 8.347269601072356e-05,
      "loss": 1.9773,
      "step": 174338
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.2202560901641846,
      "learning_rate": 8.346986476529802e-05,
      "loss": 1.7786,
      "step": 174339
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.3365082740783691,
      "learning_rate": 8.346703356012964e-05,
      "loss": 2.1101,
      "step": 174340
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.5126367807388306,
      "learning_rate": 8.34642023952188e-05,
      "loss": 1.9613,
      "step": 174341
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.2534583806991577,
      "learning_rate": 8.34613712705661e-05,
      "loss": 1.8657,
      "step": 174342
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.362547755241394,
      "learning_rate": 8.345854018617188e-05,
      "loss": 1.9103,
      "step": 174343
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.2082812786102295,
      "learning_rate": 8.345570914203691e-05,
      "loss": 1.8647,
      "step": 174344
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.2485475540161133,
      "learning_rate": 8.345287813816152e-05,
      "loss": 1.822,
      "step": 174345
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.209826946258545,
      "learning_rate": 8.345004717454644e-05,
      "loss": 1.94,
      "step": 174346
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.4310263395309448,
      "learning_rate": 8.344721625119196e-05,
      "loss": 1.9858,
      "step": 174347
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.2723538875579834,
      "learning_rate": 8.344438536809892e-05,
      "loss": 1.9932,
      "step": 174348
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.28154456615448,
      "learning_rate": 8.344155452526748e-05,
      "loss": 1.9727,
      "step": 174349
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.3559670448303223,
      "learning_rate": 8.343872372269843e-05,
      "loss": 2.128,
      "step": 174350
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.2228224277496338,
      "learning_rate": 8.343589296039213e-05,
      "loss": 1.9442,
      "step": 174351
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.2249751091003418,
      "learning_rate": 8.343306223834925e-05,
      "loss": 2.0611,
      "step": 174352
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.2192623615264893,
      "learning_rate": 8.343023155657018e-05,
      "loss": 1.8908,
      "step": 174353
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.2681736946105957,
      "learning_rate": 8.342740091505563e-05,
      "loss": 1.9907,
      "step": 174354
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.366520881652832,
      "learning_rate": 8.342457031380597e-05,
      "loss": 2.1546,
      "step": 174355
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.3681763410568237,
      "learning_rate": 8.34217397528218e-05,
      "loss": 2.1345,
      "step": 174356
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.2759089469909668,
      "learning_rate": 8.341890923210353e-05,
      "loss": 1.8813,
      "step": 174357
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.2831580638885498,
      "learning_rate": 8.341607875165184e-05,
      "loss": 2.165,
      "step": 174358
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.3194087743759155,
      "learning_rate": 8.34132483114671e-05,
      "loss": 1.9962,
      "step": 174359
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.28744637966156,
      "learning_rate": 8.341041791155007e-05,
      "loss": 2.1332,
      "step": 174360
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.1643848419189453,
      "learning_rate": 8.340758755190108e-05,
      "loss": 2.0039,
      "step": 174361
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.2901906967163086,
      "learning_rate": 8.340475723252071e-05,
      "loss": 1.9898,
      "step": 174362
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.2863065004348755,
      "learning_rate": 8.34019269534094e-05,
      "loss": 1.9942,
      "step": 174363
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.2595069408416748,
      "learning_rate": 8.339909671456785e-05,
      "loss": 1.9367,
      "step": 174364
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.421716570854187,
      "learning_rate": 8.339626651599642e-05,
      "loss": 2.1276,
      "step": 174365
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.2475740909576416,
      "learning_rate": 8.339343635769576e-05,
      "loss": 2.178,
      "step": 174366
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.224039912223816,
      "learning_rate": 8.33906062396664e-05,
      "loss": 2.2619,
      "step": 174367
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.2315101623535156,
      "learning_rate": 8.33877761619088e-05,
      "loss": 1.7801,
      "step": 174368
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.2505321502685547,
      "learning_rate": 8.338494612442341e-05,
      "loss": 2.0032,
      "step": 174369
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.3093175888061523,
      "learning_rate": 8.338211612721092e-05,
      "loss": 2.1811,
      "step": 174370
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.251387596130371,
      "learning_rate": 8.337928617027171e-05,
      "loss": 2.0619,
      "step": 174371
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.2513238191604614,
      "learning_rate": 8.337645625360647e-05,
      "loss": 1.9502,
      "step": 174372
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.2596499919891357,
      "learning_rate": 8.337362637721566e-05,
      "loss": 1.8535,
      "step": 174373
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.4401214122772217,
      "learning_rate": 8.337079654109972e-05,
      "loss": 2.1299,
      "step": 174374
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.415434718132019,
      "learning_rate": 8.33679667452592e-05,
      "loss": 1.8967,
      "step": 174375
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.411420226097107,
      "learning_rate": 8.336513698969472e-05,
      "loss": 2.3061,
      "step": 174376
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.7468366622924805,
      "learning_rate": 8.33623072744067e-05,
      "loss": 2.2302,
      "step": 174377
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.326204776763916,
      "learning_rate": 8.335947759939577e-05,
      "loss": 2.0788,
      "step": 174378
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.750643014907837,
      "learning_rate": 8.335664796466244e-05,
      "loss": 2.146,
      "step": 174379
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.4497008323669434,
      "learning_rate": 8.335381837020716e-05,
      "loss": 2.0763,
      "step": 174380
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.3024601936340332,
      "learning_rate": 8.335098881603043e-05,
      "loss": 1.9545,
      "step": 174381
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.3920711278915405,
      "learning_rate": 8.334815930213292e-05,
      "loss": 1.9026,
      "step": 174382
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.420613169670105,
      "learning_rate": 8.334532982851497e-05,
      "loss": 2.0544,
      "step": 174383
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.4943939447402954,
      "learning_rate": 8.334250039517731e-05,
      "loss": 2.0982,
      "step": 174384
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.5068639516830444,
      "learning_rate": 8.33396710021204e-05,
      "loss": 1.9608,
      "step": 174385
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.3620679378509521,
      "learning_rate": 8.33368416493447e-05,
      "loss": 2.1099,
      "step": 174386
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.3616087436676025,
      "learning_rate": 8.333401233685071e-05,
      "loss": 1.968,
      "step": 174387
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.2243874073028564,
      "learning_rate": 8.33311830646391e-05,
      "loss": 1.9647,
      "step": 174388
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.3599834442138672,
      "learning_rate": 8.332835383271019e-05,
      "loss": 2.0021,
      "step": 174389
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.3405729532241821,
      "learning_rate": 8.332552464106477e-05,
      "loss": 2.034,
      "step": 174390
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.6444096565246582,
      "learning_rate": 8.332269548970318e-05,
      "loss": 2.007,
      "step": 174391
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.4516196250915527,
      "learning_rate": 8.3319866378626e-05,
      "loss": 2.3441,
      "step": 174392
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.2831602096557617,
      "learning_rate": 8.331703730783368e-05,
      "loss": 1.9742,
      "step": 174393
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.2958714962005615,
      "learning_rate": 8.331420827732689e-05,
      "loss": 2.0307,
      "step": 174394
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.3761506080627441,
      "learning_rate": 8.331137928710601e-05,
      "loss": 1.8721,
      "step": 174395
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.5299270153045654,
      "learning_rate": 8.33085503371717e-05,
      "loss": 2.0888,
      "step": 174396
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.2924224138259888,
      "learning_rate": 8.330572142752443e-05,
      "loss": 1.859,
      "step": 174397
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.2520956993103027,
      "learning_rate": 8.33028925581647e-05,
      "loss": 1.8674,
      "step": 174398
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.2539290189743042,
      "learning_rate": 8.330006372909299e-05,
      "loss": 1.9229,
      "step": 174399
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.2289950847625732,
      "learning_rate": 8.329723494030994e-05,
      "loss": 1.6982,
      "step": 174400
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.354490041732788,
      "learning_rate": 8.329440619181596e-05,
      "loss": 1.9678,
      "step": 174401
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.3366551399230957,
      "learning_rate": 8.329157748361171e-05,
      "loss": 2.0016,
      "step": 174402
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.2833629846572876,
      "learning_rate": 8.328874881569769e-05,
      "loss": 1.8766,
      "step": 174403
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.4348028898239136,
      "learning_rate": 8.328592018807435e-05,
      "loss": 2.1405,
      "step": 174404
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.3466930389404297,
      "learning_rate": 8.328309160074216e-05,
      "loss": 2.1303,
      "step": 174405
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.2216062545776367,
      "learning_rate": 8.328026305370183e-05,
      "loss": 2.0829,
      "step": 174406
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.348084807395935,
      "learning_rate": 8.327743454695371e-05,
      "loss": 2.0038,
      "step": 174407
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.299219012260437,
      "learning_rate": 8.32746060804985e-05,
      "loss": 2.0306,
      "step": 174408
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.3812222480773926,
      "learning_rate": 8.327177765433662e-05,
      "loss": 2.1239,
      "step": 174409
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.180632472038269,
      "learning_rate": 8.32689492684685e-05,
      "loss": 2.1909,
      "step": 174410
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.2770801782608032,
      "learning_rate": 8.326612092289489e-05,
      "loss": 2.0749,
      "step": 174411
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.4447462558746338,
      "learning_rate": 8.326329261761618e-05,
      "loss": 2.0993,
      "step": 174412
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.2189223766326904,
      "learning_rate": 8.326046435263283e-05,
      "loss": 2.0616,
      "step": 174413
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.324216365814209,
      "learning_rate": 8.325763612794555e-05,
      "loss": 2.0263,
      "step": 174414
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.3354817628860474,
      "learning_rate": 8.325480794355474e-05,
      "loss": 2.0739,
      "step": 174415
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.3407363891601562,
      "learning_rate": 8.325197979946089e-05,
      "loss": 2.0146,
      "step": 174416
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.2589099407196045,
      "learning_rate": 8.324915169566467e-05,
      "loss": 1.9294,
      "step": 174417
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.134743571281433,
      "learning_rate": 8.324632363216653e-05,
      "loss": 1.9978,
      "step": 174418
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.456658124923706,
      "learning_rate": 8.32434956089669e-05,
      "loss": 2.0884,
      "step": 174419
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.3428364992141724,
      "learning_rate": 8.324066762606649e-05,
      "loss": 2.035,
      "step": 174420
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.2692699432373047,
      "learning_rate": 8.323783968346562e-05,
      "loss": 2.1182,
      "step": 174421
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.404374599456787,
      "learning_rate": 8.323501178116503e-05,
      "loss": 1.9246,
      "step": 174422
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.3853286504745483,
      "learning_rate": 8.323218391916513e-05,
      "loss": 1.8707,
      "step": 174423
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.7931679487228394,
      "learning_rate": 8.322935609746639e-05,
      "loss": 1.7923,
      "step": 174424
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.3116525411605835,
      "learning_rate": 8.322652831606949e-05,
      "loss": 1.9526,
      "step": 174425
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.2930951118469238,
      "learning_rate": 8.322370057497488e-05,
      "loss": 2.042,
      "step": 174426
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.2794138193130493,
      "learning_rate": 8.322087287418295e-05,
      "loss": 1.9901,
      "step": 174427
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.2143059968948364,
      "learning_rate": 8.321804521369446e-05,
      "loss": 2.015,
      "step": 174428
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.229390263557434,
      "learning_rate": 8.321521759350983e-05,
      "loss": 1.7674,
      "step": 174429
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.2319706678390503,
      "learning_rate": 8.321239001362952e-05,
      "loss": 1.9021,
      "step": 174430
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.1808445453643799,
      "learning_rate": 8.320956247405419e-05,
      "loss": 2.1569,
      "step": 174431
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.247826337814331,
      "learning_rate": 8.32067349747842e-05,
      "loss": 2.096,
      "step": 174432
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.411170244216919,
      "learning_rate": 8.320390751582023e-05,
      "loss": 2.0561,
      "step": 174433
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.2544821500778198,
      "learning_rate": 8.32010800971628e-05,
      "loss": 1.9498,
      "step": 174434
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.3596508502960205,
      "learning_rate": 8.319825271881235e-05,
      "loss": 2.1642,
      "step": 174435
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.4837843179702759,
      "learning_rate": 8.319542538076934e-05,
      "loss": 2.0161,
      "step": 174436
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.3182560205459595,
      "learning_rate": 8.31925980830345e-05,
      "loss": 2.204,
      "step": 174437
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.3659759759902954,
      "learning_rate": 8.318977082560816e-05,
      "loss": 2.2352,
      "step": 174438
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.339042067527771,
      "learning_rate": 8.318694360849101e-05,
      "loss": 2.0312,
      "step": 174439
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.2229018211364746,
      "learning_rate": 8.318411643168352e-05,
      "loss": 2.0846,
      "step": 174440
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.4273130893707275,
      "learning_rate": 8.318128929518617e-05,
      "loss": 1.8667,
      "step": 174441
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.197003960609436,
      "learning_rate": 8.317846219899945e-05,
      "loss": 2.1786,
      "step": 174442
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.2949951887130737,
      "learning_rate": 8.317563514312402e-05,
      "loss": 2.0926,
      "step": 174443
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.3753505945205688,
      "learning_rate": 8.317280812756024e-05,
      "loss": 1.8322,
      "step": 174444
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.2051844596862793,
      "learning_rate": 8.316998115230882e-05,
      "loss": 2.0216,
      "step": 174445
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.259721040725708,
      "learning_rate": 8.316715421737021e-05,
      "loss": 2.0446,
      "step": 174446
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.3065452575683594,
      "learning_rate": 8.31643273227449e-05,
      "loss": 2.1993,
      "step": 174447
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.2999937534332275,
      "learning_rate": 8.316150046843336e-05,
      "loss": 2.0367,
      "step": 174448
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.2791839838027954,
      "learning_rate": 8.315867365443625e-05,
      "loss": 1.8581,
      "step": 174449
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.3810619115829468,
      "learning_rate": 8.315584688075399e-05,
      "loss": 1.8844,
      "step": 174450
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.4152902364730835,
      "learning_rate": 8.315302014738721e-05,
      "loss": 1.959,
      "step": 174451
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.1909610033035278,
      "learning_rate": 8.315019345433637e-05,
      "loss": 2.0402,
      "step": 174452
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.3485779762268066,
      "learning_rate": 8.314736680160203e-05,
      "loss": 2.05,
      "step": 174453
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.3628342151641846,
      "learning_rate": 8.314454018918458e-05,
      "loss": 2.0996,
      "step": 174454
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.480010747909546,
      "learning_rate": 8.314171361708476e-05,
      "loss": 1.9134,
      "step": 174455
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.2116994857788086,
      "learning_rate": 8.313888708530287e-05,
      "loss": 1.8179,
      "step": 174456
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.3525793552398682,
      "learning_rate": 8.313606059383968e-05,
      "loss": 2.0579,
      "step": 174457
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.2662792205810547,
      "learning_rate": 8.313323414269555e-05,
      "loss": 1.9971,
      "step": 174458
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.2824256420135498,
      "learning_rate": 8.313040773187106e-05,
      "loss": 2.0753,
      "step": 174459
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.2387827634811401,
      "learning_rate": 8.312758136136666e-05,
      "loss": 1.9161,
      "step": 174460
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.448301076889038,
      "learning_rate": 8.3124755031183e-05,
      "loss": 1.9988,
      "step": 174461
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.2919998168945312,
      "learning_rate": 8.312192874132045e-05,
      "loss": 1.8969,
      "step": 174462
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.2916611433029175,
      "learning_rate": 8.311910249177973e-05,
      "loss": 2.1287,
      "step": 174463
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.3028651475906372,
      "learning_rate": 8.311627628256126e-05,
      "loss": 2.0683,
      "step": 174464
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.4010804891586304,
      "learning_rate": 8.311345011366555e-05,
      "loss": 1.9707,
      "step": 174465
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.2966645956039429,
      "learning_rate": 8.311062398509306e-05,
      "loss": 1.9897,
      "step": 174466
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.3312004804611206,
      "learning_rate": 8.31077978968445e-05,
      "loss": 2.07,
      "step": 174467
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.3586305379867554,
      "learning_rate": 8.31049718489202e-05,
      "loss": 2.1626,
      "step": 174468
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.5247056484222412,
      "learning_rate": 8.31021458413209e-05,
      "loss": 1.8955,
      "step": 174469
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.3424376249313354,
      "learning_rate": 8.309931987404698e-05,
      "loss": 2.0422,
      "step": 174470
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.3536466360092163,
      "learning_rate": 8.309649394709897e-05,
      "loss": 2.0328,
      "step": 174471
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.3816806077957153,
      "learning_rate": 8.309366806047735e-05,
      "loss": 2.0664,
      "step": 174472
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.3030319213867188,
      "learning_rate": 8.30908422141828e-05,
      "loss": 2.1011,
      "step": 174473
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.4065117835998535,
      "learning_rate": 8.308801640821564e-05,
      "loss": 2.1181,
      "step": 174474
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.3596601486206055,
      "learning_rate": 8.308519064257663e-05,
      "loss": 2.0143,
      "step": 174475
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.3639565706253052,
      "learning_rate": 8.308236491726619e-05,
      "loss": 2.2442,
      "step": 174476
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.2648495435714722,
      "learning_rate": 8.307953923228482e-05,
      "loss": 2.1535,
      "step": 174477
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.2335700988769531,
      "learning_rate": 8.307671358763296e-05,
      "loss": 2.0548,
      "step": 174478
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.257541537284851,
      "learning_rate": 8.307388798331131e-05,
      "loss": 1.8033,
      "step": 174479
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.3511320352554321,
      "learning_rate": 8.307106241932026e-05,
      "loss": 2.1065,
      "step": 174480
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.2088594436645508,
      "learning_rate": 8.306823689566048e-05,
      "loss": 2.0803,
      "step": 174481
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.2006034851074219,
      "learning_rate": 8.306541141233238e-05,
      "loss": 2.0431,
      "step": 174482
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.319885015487671,
      "learning_rate": 8.306258596933654e-05,
      "loss": 1.9506,
      "step": 174483
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.4081544876098633,
      "learning_rate": 8.305976056667337e-05,
      "loss": 2.0636,
      "step": 174484
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.3574438095092773,
      "learning_rate": 8.305693520434357e-05,
      "loss": 1.968,
      "step": 174485
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.3382854461669922,
      "learning_rate": 8.305410988234749e-05,
      "loss": 2.073,
      "step": 174486
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.2397816181182861,
      "learning_rate": 8.305128460068583e-05,
      "loss": 1.8463,
      "step": 174487
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.2497960329055786,
      "learning_rate": 8.304845935935895e-05,
      "loss": 2.1637,
      "step": 174488
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.6519854068756104,
      "learning_rate": 8.304563415836765e-05,
      "loss": 2.096,
      "step": 174489
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.4592710733413696,
      "learning_rate": 8.304280899771208e-05,
      "loss": 2.1741,
      "step": 174490
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.3362385034561157,
      "learning_rate": 8.303998387739298e-05,
      "loss": 2.0536,
      "step": 174491
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.4508049488067627,
      "learning_rate": 8.30371587974108e-05,
      "loss": 2.284,
      "step": 174492
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.267760992050171,
      "learning_rate": 8.303433375776622e-05,
      "loss": 2.1854,
      "step": 174493
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.261583924293518,
      "learning_rate": 8.303150875845954e-05,
      "loss": 2.2169,
      "step": 174494
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.486690640449524,
      "learning_rate": 8.30286837994915e-05,
      "loss": 2.2777,
      "step": 174495
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.5016080141067505,
      "learning_rate": 8.302585888086251e-05,
      "loss": 1.8091,
      "step": 174496
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.287894606590271,
      "learning_rate": 8.302303400257309e-05,
      "loss": 1.8969,
      "step": 174497
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.2753472328186035,
      "learning_rate": 8.302020916462369e-05,
      "loss": 2.1718,
      "step": 174498
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.3929399251937866,
      "learning_rate": 8.301738436701507e-05,
      "loss": 1.9793,
      "step": 174499
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.3793901205062866,
      "learning_rate": 8.301455960974747e-05,
      "loss": 2.0611,
      "step": 174500
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.4160486459732056,
      "learning_rate": 8.301173489282168e-05,
      "loss": 2.0869,
      "step": 174501
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.2927660942077637,
      "learning_rate": 8.300891021623812e-05,
      "loss": 2.035,
      "step": 174502
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.511980652809143,
      "learning_rate": 8.300608557999727e-05,
      "loss": 2.1043,
      "step": 174503
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.3542988300323486,
      "learning_rate": 8.30032609840996e-05,
      "loss": 1.9623,
      "step": 174504
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.4329946041107178,
      "learning_rate": 8.300043642854581e-05,
      "loss": 2.3559,
      "step": 174505
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.3736741542816162,
      "learning_rate": 8.299761191333626e-05,
      "loss": 2.218,
      "step": 174506
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.2625012397766113,
      "learning_rate": 8.299478743847163e-05,
      "loss": 1.8878,
      "step": 174507
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.3055546283721924,
      "learning_rate": 8.299196300395237e-05,
      "loss": 2.1604,
      "step": 174508
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.1874243021011353,
      "learning_rate": 8.298913860977889e-05,
      "loss": 2.0587,
      "step": 174509
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.2696607112884521,
      "learning_rate": 8.298631425595194e-05,
      "loss": 2.03,
      "step": 174510
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.2353569269180298,
      "learning_rate": 8.298348994247193e-05,
      "loss": 2.0713,
      "step": 174511
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.1899319887161255,
      "learning_rate": 8.298066566933927e-05,
      "loss": 2.3067,
      "step": 174512
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.1861587762832642,
      "learning_rate": 8.29778414365547e-05,
      "loss": 1.8733,
      "step": 174513
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.3380206823349,
      "learning_rate": 8.297501724411865e-05,
      "loss": 2.0168,
      "step": 174514
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.2533364295959473,
      "learning_rate": 8.297219309203156e-05,
      "loss": 2.0273,
      "step": 174515
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.2017102241516113,
      "learning_rate": 8.296936898029412e-05,
      "loss": 1.7971,
      "step": 174516
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.260803461074829,
      "learning_rate": 8.296654490890678e-05,
      "loss": 1.8709,
      "step": 174517
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.2691293954849243,
      "learning_rate": 8.296372087786993e-05,
      "loss": 1.872,
      "step": 174518
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.2485663890838623,
      "learning_rate": 8.296089688718437e-05,
      "loss": 1.8984,
      "step": 174519
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.2274802923202515,
      "learning_rate": 8.295807293685042e-05,
      "loss": 2.3756,
      "step": 174520
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.2200748920440674,
      "learning_rate": 8.29552490268686e-05,
      "loss": 1.8417,
      "step": 174521
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.269073247909546,
      "learning_rate": 8.295242515723957e-05,
      "loss": 1.9114,
      "step": 174522
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.2410157918930054,
      "learning_rate": 8.294960132796374e-05,
      "loss": 1.8881,
      "step": 174523
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.2044111490249634,
      "learning_rate": 8.294677753904171e-05,
      "loss": 1.954,
      "step": 174524
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.1945414543151855,
      "learning_rate": 8.2943953790474e-05,
      "loss": 2.023,
      "step": 174525
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.2509164810180664,
      "learning_rate": 8.294113008226109e-05,
      "loss": 2.1801,
      "step": 174526
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.2109395265579224,
      "learning_rate": 8.293830641440342e-05,
      "loss": 2.0406,
      "step": 174527
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.3357720375061035,
      "learning_rate": 8.293548278690174e-05,
      "loss": 2.1822,
      "step": 174528
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.2921106815338135,
      "learning_rate": 8.293265919975634e-05,
      "loss": 1.9954,
      "step": 174529
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.2173224687576294,
      "learning_rate": 8.292983565296797e-05,
      "loss": 1.9092,
      "step": 174530
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.2494535446166992,
      "learning_rate": 8.2927012146537e-05,
      "loss": 1.9038,
      "step": 174531
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.5040156841278076,
      "learning_rate": 8.292418868046403e-05,
      "loss": 2.1198,
      "step": 174532
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.3906230926513672,
      "learning_rate": 8.292136525474943e-05,
      "loss": 1.834,
      "step": 174533
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.3071057796478271,
      "learning_rate": 8.291854186939397e-05,
      "loss": 1.9423,
      "step": 174534
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.2603832483291626,
      "learning_rate": 8.291571852439793e-05,
      "loss": 1.9279,
      "step": 174535
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.2379521131515503,
      "learning_rate": 8.291289521976208e-05,
      "loss": 1.9994,
      "step": 174536
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.2370474338531494,
      "learning_rate": 8.29100719554868e-05,
      "loss": 1.9101,
      "step": 174537
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.3057057857513428,
      "learning_rate": 8.290724873157264e-05,
      "loss": 1.9594,
      "step": 174538
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.281502604484558,
      "learning_rate": 8.290442554802003e-05,
      "loss": 2.046,
      "step": 174539
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.5869747400283813,
      "learning_rate": 8.290160240482968e-05,
      "loss": 2.0226,
      "step": 174540
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.3292306661605835,
      "learning_rate": 8.289877930200191e-05,
      "loss": 2.1144,
      "step": 174541
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.4239140748977661,
      "learning_rate": 8.289595623953748e-05,
      "loss": 2.0697,
      "step": 174542
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.320136547088623,
      "learning_rate": 8.289313321743676e-05,
      "loss": 2.0143,
      "step": 174543
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.233546495437622,
      "learning_rate": 8.289031023570032e-05,
      "loss": 2.0255,
      "step": 174544
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.3667676448822021,
      "learning_rate": 8.288748729432855e-05,
      "loss": 2.2021,
      "step": 174545
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.4108095169067383,
      "learning_rate": 8.28846643933222e-05,
      "loss": 2.244,
      "step": 174546
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.2391836643218994,
      "learning_rate": 8.288184153268164e-05,
      "loss": 1.9865,
      "step": 174547
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.2454612255096436,
      "learning_rate": 8.287901871240747e-05,
      "loss": 2.0558,
      "step": 174548
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.259875774383545,
      "learning_rate": 8.287619593250022e-05,
      "loss": 2.1485,
      "step": 174549
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.382105827331543,
      "learning_rate": 8.287337319296041e-05,
      "loss": 2.0082,
      "step": 174550
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.3162511587142944,
      "learning_rate": 8.287055049378841e-05,
      "loss": 2.0218,
      "step": 174551
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.2361127138137817,
      "learning_rate": 8.286772783498498e-05,
      "loss": 2.0448,
      "step": 174552
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.4686613082885742,
      "learning_rate": 8.286490521655044e-05,
      "loss": 2.1012,
      "step": 174553
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.4022804498672485,
      "learning_rate": 8.28620826384855e-05,
      "loss": 1.9339,
      "step": 174554
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.1647911071777344,
      "learning_rate": 8.285926010079053e-05,
      "loss": 2.1785,
      "step": 174555
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.2089319229125977,
      "learning_rate": 8.285643760346626e-05,
      "loss": 1.9604,
      "step": 174556
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.2131575345993042,
      "learning_rate": 8.285361514651294e-05,
      "loss": 1.9286,
      "step": 174557
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.2214521169662476,
      "learning_rate": 8.28507927299313e-05,
      "loss": 2.0415,
      "step": 174558
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.221153974533081,
      "learning_rate": 8.284797035372172e-05,
      "loss": 1.9358,
      "step": 174559
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.3432379961013794,
      "learning_rate": 8.284514801788488e-05,
      "loss": 2.007,
      "step": 174560
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.2062151432037354,
      "learning_rate": 8.284232572242114e-05,
      "loss": 2.0296,
      "step": 174561
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.332237720489502,
      "learning_rate": 8.283950346733129e-05,
      "loss": 1.9721,
      "step": 174562
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.1486790180206299,
      "learning_rate": 8.28366812526155e-05,
      "loss": 1.8974,
      "step": 174563
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.4871697425842285,
      "learning_rate": 8.283385907827454e-05,
      "loss": 2.2745,
      "step": 174564
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.2262474298477173,
      "learning_rate": 8.283103694430881e-05,
      "loss": 2.1323,
      "step": 174565
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.2943501472473145,
      "learning_rate": 8.282821485071896e-05,
      "loss": 2.0732,
      "step": 174566
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.312612533569336,
      "learning_rate": 8.282539279750536e-05,
      "loss": 1.9862,
      "step": 174567
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.6137409210205078,
      "learning_rate": 8.282257078466877e-05,
      "loss": 2.0909,
      "step": 174568
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.2955968379974365,
      "learning_rate": 8.281974881220943e-05,
      "loss": 2.0503,
      "step": 174569
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.4059703350067139,
      "learning_rate": 8.281692688012806e-05,
      "loss": 2.1729,
      "step": 174570
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.2100911140441895,
      "learning_rate": 8.281410498842504e-05,
      "loss": 1.9308,
      "step": 174571
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.2575138807296753,
      "learning_rate": 8.281128313710108e-05,
      "loss": 1.9929,
      "step": 174572
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.3484975099563599,
      "learning_rate": 8.280846132615646e-05,
      "loss": 2.0341,
      "step": 174573
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.3151659965515137,
      "learning_rate": 8.280563955559208e-05,
      "loss": 1.9656,
      "step": 174574
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.3484704494476318,
      "learning_rate": 8.280281782540805e-05,
      "loss": 2.1318,
      "step": 174575
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.2010912895202637,
      "learning_rate": 8.279999613560513e-05,
      "loss": 2.094,
      "step": 174576
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.2165379524230957,
      "learning_rate": 8.279717448618375e-05,
      "loss": 1.9815,
      "step": 174577
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.2607331275939941,
      "learning_rate": 8.279435287714453e-05,
      "loss": 1.7964,
      "step": 174578
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.3218810558319092,
      "learning_rate": 8.279153130848785e-05,
      "loss": 2.0324,
      "step": 174579
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.2902708053588867,
      "learning_rate": 8.278870978021452e-05,
      "loss": 2.0142,
      "step": 174580
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.3700764179229736,
      "learning_rate": 8.278588829232467e-05,
      "loss": 2.1923,
      "step": 174581
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.3709783554077148,
      "learning_rate": 8.278306684481913e-05,
      "loss": 1.9924,
      "step": 174582
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.3616803884506226,
      "learning_rate": 8.278024543769823e-05,
      "loss": 2.1983,
      "step": 174583
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.2009310722351074,
      "learning_rate": 8.277742407096266e-05,
      "loss": 1.747,
      "step": 174584
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.3212883472442627,
      "learning_rate": 8.277460274461278e-05,
      "loss": 2.0927,
      "step": 174585
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.2118476629257202,
      "learning_rate": 8.27717814586493e-05,
      "loss": 2.1448,
      "step": 174586
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.2983931303024292,
      "learning_rate": 8.276896021307265e-05,
      "loss": 2.004,
      "step": 174587
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.2839117050170898,
      "learning_rate": 8.276613900788334e-05,
      "loss": 1.9565,
      "step": 174588
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.2621132135391235,
      "learning_rate": 8.276331784308182e-05,
      "loss": 2.0345,
      "step": 174589
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.2396769523620605,
      "learning_rate": 8.276049671866877e-05,
      "loss": 1.9932,
      "step": 174590
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.3093210458755493,
      "learning_rate": 8.275767563464458e-05,
      "loss": 1.8737,
      "step": 174591
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.2913380861282349,
      "learning_rate": 8.275485459100992e-05,
      "loss": 2.1721,
      "step": 174592
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.4421942234039307,
      "learning_rate": 8.275203358776526e-05,
      "loss": 1.9935,
      "step": 174593
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.358980655670166,
      "learning_rate": 8.274921262491098e-05,
      "loss": 2.0212,
      "step": 174594
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.4298030138015747,
      "learning_rate": 8.274639170244779e-05,
      "loss": 1.9928,
      "step": 174595
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.3445825576782227,
      "learning_rate": 8.274357082037619e-05,
      "loss": 2.1257,
      "step": 174596
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.2344660758972168,
      "learning_rate": 8.274074997869655e-05,
      "loss": 2.0584,
      "step": 174597
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.4703871011734009,
      "learning_rate": 8.27379291774096e-05,
      "loss": 2.0324,
      "step": 174598
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.2587707042694092,
      "learning_rate": 8.273510841651577e-05,
      "loss": 1.9836,
      "step": 174599
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.459701418876648,
      "learning_rate": 8.27322876960155e-05,
      "loss": 1.9749,
      "step": 174600
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.546156406402588,
      "learning_rate": 8.272946701590948e-05,
      "loss": 2.4001,
      "step": 174601
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.297439455986023,
      "learning_rate": 8.272664637619815e-05,
      "loss": 2.1438,
      "step": 174602
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.420045256614685,
      "learning_rate": 8.272382577688193e-05,
      "loss": 2.283,
      "step": 174603
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.3301318883895874,
      "learning_rate": 8.272100521796158e-05,
      "loss": 2.1525,
      "step": 174604
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.3683574199676514,
      "learning_rate": 8.271818469943747e-05,
      "loss": 2.3006,
      "step": 174605
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.2796530723571777,
      "learning_rate": 8.271536422131009e-05,
      "loss": 1.8813,
      "step": 174606
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.4576793909072876,
      "learning_rate": 8.271254378358009e-05,
      "loss": 1.9951,
      "step": 174607
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.1904935836791992,
      "learning_rate": 8.270972338624785e-05,
      "loss": 1.9214,
      "step": 174608
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.2218174934387207,
      "learning_rate": 8.270690302931408e-05,
      "loss": 2.0252,
      "step": 174609
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.4698883295059204,
      "learning_rate": 8.270408271277916e-05,
      "loss": 1.9862,
      "step": 174610
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.3780401945114136,
      "learning_rate": 8.270126243664367e-05,
      "loss": 1.9487,
      "step": 174611
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.1815986633300781,
      "learning_rate": 8.269844220090803e-05,
      "loss": 2.2195,
      "step": 174612
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.3326414823532104,
      "learning_rate": 8.269562200557296e-05,
      "loss": 2.1871,
      "step": 174613
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.386704444885254,
      "learning_rate": 8.269280185063874e-05,
      "loss": 2.1416,
      "step": 174614
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.293620228767395,
      "learning_rate": 8.268998173610615e-05,
      "loss": 2.0104,
      "step": 174615
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.2323894500732422,
      "learning_rate": 8.26871616619756e-05,
      "loss": 2.0574,
      "step": 174616
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.2585872411727905,
      "learning_rate": 8.268434162824759e-05,
      "loss": 2.1506,
      "step": 174617
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.4011634588241577,
      "learning_rate": 8.268152163492261e-05,
      "loss": 1.9304,
      "step": 174618
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.4766696691513062,
      "learning_rate": 8.26787016820013e-05,
      "loss": 2.0493,
      "step": 174619
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.3278862237930298,
      "learning_rate": 8.267588176948404e-05,
      "loss": 2.0532,
      "step": 174620
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.3575665950775146,
      "learning_rate": 8.267306189737149e-05,
      "loss": 2.0601,
      "step": 174621
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.2665135860443115,
      "learning_rate": 8.267024206566406e-05,
      "loss": 1.7794,
      "step": 174622
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.490515947341919,
      "learning_rate": 8.266742227436254e-05,
      "loss": 1.9497,
      "step": 174623
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.2863922119140625,
      "learning_rate": 8.266460252346703e-05,
      "loss": 1.9322,
      "step": 174624
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.3068468570709229,
      "learning_rate": 8.266178281297837e-05,
      "loss": 1.9654,
      "step": 174625
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.2568442821502686,
      "learning_rate": 8.26589631428969e-05,
      "loss": 2.0117,
      "step": 174626
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.2348780632019043,
      "learning_rate": 8.265614351322332e-05,
      "loss": 2.0272,
      "step": 174627
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.3814584016799927,
      "learning_rate": 8.265332392395799e-05,
      "loss": 2.0331,
      "step": 174628
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.3219705820083618,
      "learning_rate": 8.265050437510166e-05,
      "loss": 1.899,
      "step": 174629
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.2876051664352417,
      "learning_rate": 8.264768486665455e-05,
      "loss": 2.0875,
      "step": 174630
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.2895172834396362,
      "learning_rate": 8.264486539861745e-05,
      "loss": 1.8606,
      "step": 174631
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.2362743616104126,
      "learning_rate": 8.264204597099064e-05,
      "loss": 2.1182,
      "step": 174632
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.211002230644226,
      "learning_rate": 8.263922658377489e-05,
      "loss": 1.921,
      "step": 174633
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.301006555557251,
      "learning_rate": 8.26364072369705e-05,
      "loss": 2.2386,
      "step": 174634
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.3236414194107056,
      "learning_rate": 8.263358793057829e-05,
      "loss": 2.1528,
      "step": 174635
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.2177507877349854,
      "learning_rate": 8.263076866459843e-05,
      "loss": 1.9167,
      "step": 174636
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.2479885816574097,
      "learning_rate": 8.262794943903169e-05,
      "loss": 2.1236,
      "step": 174637
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.2848347425460815,
      "learning_rate": 8.262513025387843e-05,
      "loss": 2.3488,
      "step": 174638
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.350817084312439,
      "learning_rate": 8.262231110913938e-05,
      "loss": 2.0964,
      "step": 174639
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.2049165964126587,
      "learning_rate": 8.261949200481484e-05,
      "loss": 1.9834,
      "step": 174640
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.5196058750152588,
      "learning_rate": 8.261667294090559e-05,
      "loss": 2.0291,
      "step": 174641
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.3471195697784424,
      "learning_rate": 8.261385391741185e-05,
      "loss": 2.0543,
      "step": 174642
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.3942757844924927,
      "learning_rate": 8.261103493433438e-05,
      "loss": 1.8709,
      "step": 174643
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.3528741598129272,
      "learning_rate": 8.260821599167355e-05,
      "loss": 1.8069,
      "step": 174644
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.2993148565292358,
      "learning_rate": 8.260539708943002e-05,
      "loss": 2.0146,
      "step": 174645
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.3129993677139282,
      "learning_rate": 8.26025782276042e-05,
      "loss": 2.0043,
      "step": 174646
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.306210994720459,
      "learning_rate": 8.259975940619681e-05,
      "loss": 2.2314,
      "step": 174647
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.363441824913025,
      "learning_rate": 8.259694062520805e-05,
      "loss": 2.1008,
      "step": 174648
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.4695558547973633,
      "learning_rate": 8.259412188463874e-05,
      "loss": 2.257,
      "step": 174649
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.392450213432312,
      "learning_rate": 8.259130318448915e-05,
      "loss": 2.0269,
      "step": 174650
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.2696523666381836,
      "learning_rate": 8.258848452476005e-05,
      "loss": 1.9931,
      "step": 174651
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.2153851985931396,
      "learning_rate": 8.258566590545177e-05,
      "loss": 2.1278,
      "step": 174652
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.2469476461410522,
      "learning_rate": 8.258284732656512e-05,
      "loss": 1.8246,
      "step": 174653
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.2265806198120117,
      "learning_rate": 8.258002878810023e-05,
      "loss": 2.1155,
      "step": 174654
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.377729892730713,
      "learning_rate": 8.257721029005787e-05,
      "loss": 2.0138,
      "step": 174655
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.4770876169204712,
      "learning_rate": 8.257439183243844e-05,
      "loss": 2.0045,
      "step": 174656
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.26948881149292,
      "learning_rate": 8.257157341524265e-05,
      "loss": 2.1901,
      "step": 174657
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.2822322845458984,
      "learning_rate": 8.256875503847082e-05,
      "loss": 1.7385,
      "step": 174658
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.6589442491531372,
      "learning_rate": 8.256593670212372e-05,
      "loss": 2.0744,
      "step": 174659
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.3703229427337646,
      "learning_rate": 8.256311840620155e-05,
      "loss": 1.853,
      "step": 174660
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.2161544561386108,
      "learning_rate": 8.256030015070508e-05,
      "loss": 1.8589,
      "step": 174661
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.316463589668274,
      "learning_rate": 8.255748193563467e-05,
      "loss": 1.991,
      "step": 174662
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.200588345527649,
      "learning_rate": 8.255466376099102e-05,
      "loss": 2.0871,
      "step": 174663
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.3826103210449219,
      "learning_rate": 8.255184562677447e-05,
      "loss": 1.935,
      "step": 174664
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.480484962463379,
      "learning_rate": 8.254902753298582e-05,
      "loss": 2.0931,
      "step": 174665
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.4232938289642334,
      "learning_rate": 8.254620947962521e-05,
      "loss": 2.0623,
      "step": 174666
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.4567078351974487,
      "learning_rate": 8.254339146669348e-05,
      "loss": 2.0376,
      "step": 174667
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.3738698959350586,
      "learning_rate": 8.254057349419093e-05,
      "loss": 1.9757,
      "step": 174668
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.2023929357528687,
      "learning_rate": 8.253775556211828e-05,
      "loss": 1.7208,
      "step": 174669
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.3744759559631348,
      "learning_rate": 8.253493767047588e-05,
      "loss": 1.9445,
      "step": 174670
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.2801754474639893,
      "learning_rate": 8.253211981926446e-05,
      "loss": 2.0428,
      "step": 174671
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.287805199623108,
      "learning_rate": 8.25293020084844e-05,
      "loss": 2.0411,
      "step": 174672
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.2258509397506714,
      "learning_rate": 8.252648423813626e-05,
      "loss": 1.9033,
      "step": 174673
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.4921460151672363,
      "learning_rate": 8.252366650822044e-05,
      "loss": 2.0748,
      "step": 174674
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.1637054681777954,
      "learning_rate": 8.25208488187377e-05,
      "loss": 1.9082,
      "step": 174675
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.236533522605896,
      "learning_rate": 8.251803116968832e-05,
      "loss": 2.1854,
      "step": 174676
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.3966585397720337,
      "learning_rate": 8.251521356107302e-05,
      "loss": 2.0446,
      "step": 174677
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.4253113269805908,
      "learning_rate": 8.251239599289228e-05,
      "loss": 2.1741,
      "step": 174678
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.210374116897583,
      "learning_rate": 8.250957846514657e-05,
      "loss": 2.0962,
      "step": 174679
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.7021713256835938,
      "learning_rate": 8.250676097783635e-05,
      "loss": 1.9593,
      "step": 174680
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.2801339626312256,
      "learning_rate": 8.250394353096231e-05,
      "loss": 1.9002,
      "step": 174681
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.2821553945541382,
      "learning_rate": 8.250112612452483e-05,
      "loss": 1.9963,
      "step": 174682
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.3244709968566895,
      "learning_rate": 8.249830875852456e-05,
      "loss": 1.996,
      "step": 174683
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.3939745426177979,
      "learning_rate": 8.2495491432962e-05,
      "loss": 2.1966,
      "step": 174684
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.2896111011505127,
      "learning_rate": 8.24926741478375e-05,
      "loss": 1.9824,
      "step": 174685
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.2476094961166382,
      "learning_rate": 8.248985690315183e-05,
      "loss": 1.8813,
      "step": 174686
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.2603834867477417,
      "learning_rate": 8.248703969890538e-05,
      "loss": 1.9379,
      "step": 174687
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.497541069984436,
      "learning_rate": 8.248422253509862e-05,
      "loss": 1.8339,
      "step": 174688
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.3080532550811768,
      "learning_rate": 8.248140541173223e-05,
      "loss": 2.1461,
      "step": 174689
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.2881628274917603,
      "learning_rate": 8.247858832880666e-05,
      "loss": 1.8959,
      "step": 174690
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.348730444908142,
      "learning_rate": 8.247577128632234e-05,
      "loss": 2.0392,
      "step": 174691
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.5324225425720215,
      "learning_rate": 8.247295428427998e-05,
      "loss": 1.8988,
      "step": 174692
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.4317134618759155,
      "learning_rate": 8.24701373226799e-05,
      "loss": 1.9848,
      "step": 174693
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.4387147426605225,
      "learning_rate": 8.24673204015228e-05,
      "loss": 1.9359,
      "step": 174694
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.3125836849212646,
      "learning_rate": 8.246450352080914e-05,
      "loss": 2.0843,
      "step": 174695
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.3486751317977905,
      "learning_rate": 8.246168668053948e-05,
      "loss": 1.8736,
      "step": 174696
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.2787039279937744,
      "learning_rate": 8.245886988071415e-05,
      "loss": 1.9234,
      "step": 174697
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.500996470451355,
      "learning_rate": 8.245605312133395e-05,
      "loss": 1.9195,
      "step": 174698
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.3946725130081177,
      "learning_rate": 8.245323640239915e-05,
      "loss": 2.092,
      "step": 174699
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.4994878768920898,
      "learning_rate": 8.245041972391053e-05,
      "loss": 2.0679,
      "step": 174700
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.2387456893920898,
      "learning_rate": 8.244760308586847e-05,
      "loss": 2.1275,
      "step": 174701
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.2845580577850342,
      "learning_rate": 8.24447864882735e-05,
      "loss": 1.8902,
      "step": 174702
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.2612488269805908,
      "learning_rate": 8.244196993112606e-05,
      "loss": 1.9798,
      "step": 174703
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.4600191116333008,
      "learning_rate": 8.243915341442686e-05,
      "loss": 1.9568,
      "step": 174704
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.5044639110565186,
      "learning_rate": 8.243633693817623e-05,
      "loss": 2.0386,
      "step": 174705
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.4466718435287476,
      "learning_rate": 8.24335205023749e-05,
      "loss": 2.0612,
      "step": 174706
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.3751258850097656,
      "learning_rate": 8.243070410702316e-05,
      "loss": 2.1419,
      "step": 174707
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.3572863340377808,
      "learning_rate": 8.242788775212187e-05,
      "loss": 2.2028,
      "step": 174708
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.5766597986221313,
      "learning_rate": 8.242507143767115e-05,
      "loss": 2.0592,
      "step": 174709
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.4176793098449707,
      "learning_rate": 8.242225516367181e-05,
      "loss": 2.0131,
      "step": 174710
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.332634449005127,
      "learning_rate": 8.241943893012422e-05,
      "loss": 1.9276,
      "step": 174711
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.1947906017303467,
      "learning_rate": 8.2416622737029e-05,
      "loss": 1.9078,
      "step": 174712
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.19206964969635,
      "learning_rate": 8.241380658438657e-05,
      "loss": 1.8576,
      "step": 174713
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.2781131267547607,
      "learning_rate": 8.24109904721977e-05,
      "loss": 2.364,
      "step": 174714
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.3459511995315552,
      "learning_rate": 8.240817440046256e-05,
      "loss": 2.1349,
      "step": 174715
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.3672382831573486,
      "learning_rate": 8.240535836918194e-05,
      "loss": 2.17,
      "step": 174716
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.2640737295150757,
      "learning_rate": 8.240254237835619e-05,
      "loss": 1.9052,
      "step": 174717
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.364298939704895,
      "learning_rate": 8.239972642798597e-05,
      "loss": 1.9357,
      "step": 174718
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.5074589252471924,
      "learning_rate": 8.23969105180717e-05,
      "loss": 1.937,
      "step": 174719
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.314982533454895,
      "learning_rate": 8.239409464861411e-05,
      "loss": 1.9164,
      "step": 174720
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.2165874242782593,
      "learning_rate": 8.239127881961338e-05,
      "loss": 2.1078,
      "step": 174721
    },
    {
      "epoch": 2.27,
      "grad_norm": 1.3788115978240967,
      "learning_rate": 8.238846303107031e-05,
      "loss": 2.1488,
      "step": 174722
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.3216710090637207,
      "learning_rate": 8.238564728298526e-05,
      "loss": 1.9671,
      "step": 174723
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.2628285884857178,
      "learning_rate": 8.23828315753589e-05,
      "loss": 1.9541,
      "step": 174724
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.3344653844833374,
      "learning_rate": 8.238001590819157e-05,
      "loss": 1.98,
      "step": 174725
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.3171604871749878,
      "learning_rate": 8.237720028148412e-05,
      "loss": 1.9775,
      "step": 174726
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.3062729835510254,
      "learning_rate": 8.237438469523668e-05,
      "loss": 2.2092,
      "step": 174727
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.2896897792816162,
      "learning_rate": 8.237156914945002e-05,
      "loss": 1.9295,
      "step": 174728
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.2258864641189575,
      "learning_rate": 8.23687536441245e-05,
      "loss": 2.0248,
      "step": 174729
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.228042483329773,
      "learning_rate": 8.236593817926083e-05,
      "loss": 1.9049,
      "step": 174730
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.2954652309417725,
      "learning_rate": 8.236312275485935e-05,
      "loss": 1.9206,
      "step": 174731
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.1922000646591187,
      "learning_rate": 8.236030737092086e-05,
      "loss": 2.0706,
      "step": 174732
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.232980489730835,
      "learning_rate": 8.235749202744552e-05,
      "loss": 2.1982,
      "step": 174733
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.2720333337783813,
      "learning_rate": 8.23546767244341e-05,
      "loss": 1.9817,
      "step": 174734
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.4823518991470337,
      "learning_rate": 8.235186146188698e-05,
      "loss": 2.0847,
      "step": 174735
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.247489094734192,
      "learning_rate": 8.234904623980487e-05,
      "loss": 2.3556,
      "step": 174736
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.2865468263626099,
      "learning_rate": 8.234623105818805e-05,
      "loss": 2.251,
      "step": 174737
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.4196497201919556,
      "learning_rate": 8.234341591703735e-05,
      "loss": 1.952,
      "step": 174738
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.4009878635406494,
      "learning_rate": 8.234060081635298e-05,
      "loss": 1.9189,
      "step": 174739
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.487892746925354,
      "learning_rate": 8.233778575613565e-05,
      "loss": 1.8085,
      "step": 174740
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.344681978225708,
      "learning_rate": 8.233497073638578e-05,
      "loss": 2.0805,
      "step": 174741
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.417876124382019,
      "learning_rate": 8.233215575710402e-05,
      "loss": 2.2146,
      "step": 174742
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.2751307487487793,
      "learning_rate": 8.232934081829073e-05,
      "loss": 2.2917,
      "step": 174743
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.312256932258606,
      "learning_rate": 8.232652591994672e-05,
      "loss": 1.9094,
      "step": 174744
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.3319894075393677,
      "learning_rate": 8.23237110620721e-05,
      "loss": 1.8264,
      "step": 174745
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.2233073711395264,
      "learning_rate": 8.232089624466774e-05,
      "loss": 1.952,
      "step": 174746
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.284749150276184,
      "learning_rate": 8.231808146773393e-05,
      "loss": 1.9234,
      "step": 174747
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.2639204263687134,
      "learning_rate": 8.231526673127139e-05,
      "loss": 1.9536,
      "step": 174748
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.1992998123168945,
      "learning_rate": 8.231245203528044e-05,
      "loss": 1.9899,
      "step": 174749
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.7393903732299805,
      "learning_rate": 8.230963737976191e-05,
      "loss": 1.7557,
      "step": 174750
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.4207193851470947,
      "learning_rate": 8.230682276471596e-05,
      "loss": 2.1125,
      "step": 174751
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.4123157262802124,
      "learning_rate": 8.230400819014335e-05,
      "loss": 1.7513,
      "step": 174752
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.3365613222122192,
      "learning_rate": 8.230119365604446e-05,
      "loss": 1.9987,
      "step": 174753
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.2284774780273438,
      "learning_rate": 8.229837916241997e-05,
      "loss": 1.8495,
      "step": 174754
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.4545385837554932,
      "learning_rate": 8.229556470927023e-05,
      "loss": 1.949,
      "step": 174755
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.3178508281707764,
      "learning_rate": 8.229275029659596e-05,
      "loss": 1.835,
      "step": 174756
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.240673303604126,
      "learning_rate": 8.228993592439756e-05,
      "loss": 1.9617,
      "step": 174757
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.2494213581085205,
      "learning_rate": 8.228712159267556e-05,
      "loss": 1.8119,
      "step": 174758
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.40821373462677,
      "learning_rate": 8.228430730143045e-05,
      "loss": 1.9271,
      "step": 174759
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.2410331964492798,
      "learning_rate": 8.228149305066286e-05,
      "loss": 2.0248,
      "step": 174760
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.1273001432418823,
      "learning_rate": 8.227867884037313e-05,
      "loss": 1.8929,
      "step": 174761
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.3028331995010376,
      "learning_rate": 8.227586467056205e-05,
      "loss": 2.0251,
      "step": 174762
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.2010554075241089,
      "learning_rate": 8.227305054122995e-05,
      "loss": 2.1713,
      "step": 174763
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.4011075496673584,
      "learning_rate": 8.227023645237741e-05,
      "loss": 1.891,
      "step": 174764
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.3060942888259888,
      "learning_rate": 8.226742240400488e-05,
      "loss": 1.9546,
      "step": 174765
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.297328233718872,
      "learning_rate": 8.226460839611301e-05,
      "loss": 1.93,
      "step": 174766
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.2831636667251587,
      "learning_rate": 8.226179442870216e-05,
      "loss": 1.8874,
      "step": 174767
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.2218358516693115,
      "learning_rate": 8.225898050177305e-05,
      "loss": 2.0469,
      "step": 174768
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.2141181230545044,
      "learning_rate": 8.225616661532613e-05,
      "loss": 1.7914,
      "step": 174769
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.2309584617614746,
      "learning_rate": 8.225335276936176e-05,
      "loss": 1.8608,
      "step": 174770
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.2009546756744385,
      "learning_rate": 8.225053896388075e-05,
      "loss": 1.7093,
      "step": 174771
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.4948774576187134,
      "learning_rate": 8.224772519888345e-05,
      "loss": 2.2272,
      "step": 174772
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.412938117980957,
      "learning_rate": 8.22449114743703e-05,
      "loss": 1.9857,
      "step": 174773
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.3456395864486694,
      "learning_rate": 8.224209779034207e-05,
      "loss": 1.8935,
      "step": 174774
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.3197604417800903,
      "learning_rate": 8.223928414679912e-05,
      "loss": 2.0416,
      "step": 174775
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.2512980699539185,
      "learning_rate": 8.223647054374187e-05,
      "loss": 1.9652,
      "step": 174776
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.3541710376739502,
      "learning_rate": 8.223365698117111e-05,
      "loss": 1.767,
      "step": 174777
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.2611196041107178,
      "learning_rate": 8.22308434590871e-05,
      "loss": 2.1714,
      "step": 174778
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.2547205686569214,
      "learning_rate": 8.222802997749061e-05,
      "loss": 2.0891,
      "step": 174779
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.2877562046051025,
      "learning_rate": 8.222521653638203e-05,
      "loss": 1.9347,
      "step": 174780
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.249312400817871,
      "learning_rate": 8.222240313576189e-05,
      "loss": 1.87,
      "step": 174781
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.2873363494873047,
      "learning_rate": 8.221958977563059e-05,
      "loss": 1.9046,
      "step": 174782
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.4820650815963745,
      "learning_rate": 8.221677645598891e-05,
      "loss": 2.134,
      "step": 174783
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.1748507022857666,
      "learning_rate": 8.221396317683714e-05,
      "loss": 2.0338,
      "step": 174784
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.2692569494247437,
      "learning_rate": 8.221114993817601e-05,
      "loss": 2.2111,
      "step": 174785
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.291083574295044,
      "learning_rate": 8.220833674000593e-05,
      "loss": 2.1515,
      "step": 174786
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.2868541479110718,
      "learning_rate": 8.220552358232743e-05,
      "loss": 2.1505,
      "step": 174787
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.354076862335205,
      "learning_rate": 8.220271046514093e-05,
      "loss": 2.022,
      "step": 174788
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.305176019668579,
      "learning_rate": 8.219989738844714e-05,
      "loss": 2.1178,
      "step": 174789
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.3212112188339233,
      "learning_rate": 8.219708435224643e-05,
      "loss": 2.1355,
      "step": 174790
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.3256899118423462,
      "learning_rate": 8.21942713565395e-05,
      "loss": 2.0486,
      "step": 174791
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.283945918083191,
      "learning_rate": 8.219145840132668e-05,
      "loss": 2.208,
      "step": 174792
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.4298065900802612,
      "learning_rate": 8.218864548660872e-05,
      "loss": 2.1226,
      "step": 174793
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.408184289932251,
      "learning_rate": 8.218583261238584e-05,
      "loss": 2.0029,
      "step": 174794
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.1898558139801025,
      "learning_rate": 8.218301977865884e-05,
      "loss": 1.9466,
      "step": 174795
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.2700231075286865,
      "learning_rate": 8.2180206985428e-05,
      "loss": 2.0396,
      "step": 174796
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.2871804237365723,
      "learning_rate": 8.217739423269407e-05,
      "loss": 2.1741,
      "step": 174797
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.3073369264602661,
      "learning_rate": 8.217458152045739e-05,
      "loss": 1.8917,
      "step": 174798
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.2169322967529297,
      "learning_rate": 8.217176884871874e-05,
      "loss": 1.8954,
      "step": 174799
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.5845359563827515,
      "learning_rate": 8.216895621747829e-05,
      "loss": 2.0737,
      "step": 174800
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.2831846475601196,
      "learning_rate": 8.216614362673685e-05,
      "loss": 2.1034,
      "step": 174801
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.3093196153640747,
      "learning_rate": 8.216333107649474e-05,
      "loss": 2.1107,
      "step": 174802
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.3567771911621094,
      "learning_rate": 8.216051856675268e-05,
      "loss": 1.9458,
      "step": 174803
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.4364477396011353,
      "learning_rate": 8.215770609751098e-05,
      "loss": 1.9101,
      "step": 174804
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.2689989805221558,
      "learning_rate": 8.215489366877045e-05,
      "loss": 2.2571,
      "step": 174805
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.2907168865203857,
      "learning_rate": 8.215208128053125e-05,
      "loss": 2.3067,
      "step": 174806
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.2567070722579956,
      "learning_rate": 8.214926893279424e-05,
      "loss": 2.0016,
      "step": 174807
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.3455456495285034,
      "learning_rate": 8.214645662555964e-05,
      "loss": 1.9213,
      "step": 174808
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.4045608043670654,
      "learning_rate": 8.214364435882822e-05,
      "loss": 1.9596,
      "step": 174809
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.3686985969543457,
      "learning_rate": 8.214083213260037e-05,
      "loss": 2.0919,
      "step": 174810
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.6678752899169922,
      "learning_rate": 8.213801994687677e-05,
      "loss": 1.9337,
      "step": 174811
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.27131187915802,
      "learning_rate": 8.213520780165769e-05,
      "loss": 1.9103,
      "step": 174812
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.3205373287200928,
      "learning_rate": 8.213239569694386e-05,
      "loss": 1.8542,
      "step": 174813
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.3365600109100342,
      "learning_rate": 8.212958363273565e-05,
      "loss": 2.0048,
      "step": 174814
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.2112315893173218,
      "learning_rate": 8.212677160903371e-05,
      "loss": 2.0226,
      "step": 174815
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.6220542192459106,
      "learning_rate": 8.212395962583848e-05,
      "loss": 2.0943,
      "step": 174816
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.351205825805664,
      "learning_rate": 8.212114768315068e-05,
      "loss": 2.0159,
      "step": 174817
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.396243691444397,
      "learning_rate": 8.21183357809705e-05,
      "loss": 1.9742,
      "step": 174818
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.264251470565796,
      "learning_rate": 8.211552391929869e-05,
      "loss": 2.1617,
      "step": 174819
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.4396240711212158,
      "learning_rate": 8.211271209813564e-05,
      "loss": 2.0889,
      "step": 174820
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.4358137845993042,
      "learning_rate": 8.210990031748208e-05,
      "loss": 2.1653,
      "step": 174821
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.5868691205978394,
      "learning_rate": 8.210708857733826e-05,
      "loss": 1.9092,
      "step": 174822
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.227966070175171,
      "learning_rate": 8.210427687770497e-05,
      "loss": 2.1719,
      "step": 174823
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.5175122022628784,
      "learning_rate": 8.21014652185826e-05,
      "loss": 1.8745,
      "step": 174824
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.4495456218719482,
      "learning_rate": 8.209865359997167e-05,
      "loss": 2.2267,
      "step": 174825
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.3854185342788696,
      "learning_rate": 8.209584202187264e-05,
      "loss": 2.1975,
      "step": 174826
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.2247117757797241,
      "learning_rate": 8.20930304842862e-05,
      "loss": 1.8597,
      "step": 174827
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.3087866306304932,
      "learning_rate": 8.209021898721264e-05,
      "loss": 2.0583,
      "step": 174828
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.2420799732208252,
      "learning_rate": 8.208740753065275e-05,
      "loss": 2.1621,
      "step": 174829
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.1755123138427734,
      "learning_rate": 8.208459611460695e-05,
      "loss": 2.0405,
      "step": 174830
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.2322245836257935,
      "learning_rate": 8.208178473907571e-05,
      "loss": 1.6745,
      "step": 174831
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.2606793642044067,
      "learning_rate": 8.207897340405946e-05,
      "loss": 1.9292,
      "step": 174832
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.4077463150024414,
      "learning_rate": 8.207616210955896e-05,
      "loss": 2.153,
      "step": 174833
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.367180585861206,
      "learning_rate": 8.207335085557453e-05,
      "loss": 2.087,
      "step": 174834
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.2029389142990112,
      "learning_rate": 8.207053964210688e-05,
      "loss": 1.9185,
      "step": 174835
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.6667674779891968,
      "learning_rate": 8.206772846915639e-05,
      "loss": 2.3238,
      "step": 174836
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.2966647148132324,
      "learning_rate": 8.206491733672365e-05,
      "loss": 1.9642,
      "step": 174837
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.52484929561615,
      "learning_rate": 8.206210624480908e-05,
      "loss": 2.1283,
      "step": 174838
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.3229113817214966,
      "learning_rate": 8.205929519341335e-05,
      "loss": 2.0859,
      "step": 174839
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.232709527015686,
      "learning_rate": 8.205648418253683e-05,
      "loss": 2.028,
      "step": 174840
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.301028847694397,
      "learning_rate": 8.20536732121802e-05,
      "loss": 2.0765,
      "step": 174841
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.2529572248458862,
      "learning_rate": 8.20508622823439e-05,
      "loss": 2.301,
      "step": 174842
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.2003358602523804,
      "learning_rate": 8.204805139302848e-05,
      "loss": 2.1895,
      "step": 174843
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.39136803150177,
      "learning_rate": 8.204524054423434e-05,
      "loss": 2.0399,
      "step": 174844
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.436271071434021,
      "learning_rate": 8.204242973596221e-05,
      "loss": 2.2614,
      "step": 174845
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.2465647459030151,
      "learning_rate": 8.20396189682124e-05,
      "loss": 1.9347,
      "step": 174846
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.2408068180084229,
      "learning_rate": 8.203680824098563e-05,
      "loss": 2.0753,
      "step": 174847
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.2528103590011597,
      "learning_rate": 8.203399755428232e-05,
      "loss": 2.0012,
      "step": 174848
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.3260142803192139,
      "learning_rate": 8.203118690810302e-05,
      "loss": 2.1032,
      "step": 174849
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.1927241086959839,
      "learning_rate": 8.202837630244814e-05,
      "loss": 2.1881,
      "step": 174850
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.3656642436981201,
      "learning_rate": 8.202556573731842e-05,
      "loss": 1.8912,
      "step": 174851
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.299319863319397,
      "learning_rate": 8.202275521271412e-05,
      "loss": 1.9941,
      "step": 174852
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.481611728668213,
      "learning_rate": 8.201994472863603e-05,
      "loss": 2.1564,
      "step": 174853
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.2991238832473755,
      "learning_rate": 8.201713428508456e-05,
      "loss": 1.9853,
      "step": 174854
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.3500038385391235,
      "learning_rate": 8.201432388206008e-05,
      "loss": 2.0494,
      "step": 174855
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.756814956665039,
      "learning_rate": 8.201151351956336e-05,
      "loss": 2.1574,
      "step": 174856
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.2671706676483154,
      "learning_rate": 8.200870319759483e-05,
      "loss": 1.8861,
      "step": 174857
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.5007065534591675,
      "learning_rate": 8.20058929161549e-05,
      "loss": 2.2175,
      "step": 174858
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.380238652229309,
      "learning_rate": 8.200308267524429e-05,
      "loss": 1.925,
      "step": 174859
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.3348418474197388,
      "learning_rate": 8.200027247486339e-05,
      "loss": 2.1424,
      "step": 174860
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.3710646629333496,
      "learning_rate": 8.199746231501271e-05,
      "loss": 2.2397,
      "step": 174861
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.1894662380218506,
      "learning_rate": 8.199465219569287e-05,
      "loss": 2.2269,
      "step": 174862
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.3337485790252686,
      "learning_rate": 8.199184211690438e-05,
      "loss": 1.7529,
      "step": 174863
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.38303804397583,
      "learning_rate": 8.198903207864759e-05,
      "loss": 1.8657,
      "step": 174864
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.7457926273345947,
      "learning_rate": 8.198622208092324e-05,
      "loss": 1.827,
      "step": 174865
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.4029505252838135,
      "learning_rate": 8.198341212373178e-05,
      "loss": 2.0246,
      "step": 174866
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.3085756301879883,
      "learning_rate": 8.198060220707366e-05,
      "loss": 2.0687,
      "step": 174867
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.2372181415557861,
      "learning_rate": 8.197779233094953e-05,
      "loss": 2.1051,
      "step": 174868
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.656640887260437,
      "learning_rate": 8.197498249535973e-05,
      "loss": 1.904,
      "step": 174869
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.4023553133010864,
      "learning_rate": 8.197217270030503e-05,
      "loss": 2.0548,
      "step": 174870
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.2876200675964355,
      "learning_rate": 8.196936294578577e-05,
      "loss": 2.1001,
      "step": 174871
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.3482720851898193,
      "learning_rate": 8.196655323180257e-05,
      "loss": 1.9946,
      "step": 174872
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.2343584299087524,
      "learning_rate": 8.196374355835577e-05,
      "loss": 2.0987,
      "step": 174873
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.4229016304016113,
      "learning_rate": 8.196093392544614e-05,
      "loss": 2.0343,
      "step": 174874
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.4659769535064697,
      "learning_rate": 8.195812433307398e-05,
      "loss": 1.9692,
      "step": 174875
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.5163209438323975,
      "learning_rate": 8.195531478124004e-05,
      "loss": 1.772,
      "step": 174876
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.290375828742981,
      "learning_rate": 8.195250526994462e-05,
      "loss": 2.0582,
      "step": 174877
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.3175926208496094,
      "learning_rate": 8.194969579918851e-05,
      "loss": 2.0633,
      "step": 174878
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.4750640392303467,
      "learning_rate": 8.194688636897189e-05,
      "loss": 1.9267,
      "step": 174879
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.2682358026504517,
      "learning_rate": 8.194407697929556e-05,
      "loss": 1.9924,
      "step": 174880
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.54645574092865,
      "learning_rate": 8.194126763015986e-05,
      "loss": 2.2142,
      "step": 174881
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.212094783782959,
      "learning_rate": 8.193845832156549e-05,
      "loss": 2.109,
      "step": 174882
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.4336048364639282,
      "learning_rate": 8.193564905351277e-05,
      "loss": 1.841,
      "step": 174883
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.2106776237487793,
      "learning_rate": 8.193283982600251e-05,
      "loss": 1.7958,
      "step": 174884
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.5982766151428223,
      "learning_rate": 8.193003063903487e-05,
      "loss": 2.0472,
      "step": 174885
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.3944576978683472,
      "learning_rate": 8.192722149261063e-05,
      "loss": 1.9664,
      "step": 174886
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.323593020439148,
      "learning_rate": 8.192441238673019e-05,
      "loss": 2.0365,
      "step": 174887
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.2934139966964722,
      "learning_rate": 8.192160332139418e-05,
      "loss": 1.9903,
      "step": 174888
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.2331719398498535,
      "learning_rate": 8.191879429660293e-05,
      "loss": 1.9283,
      "step": 174889
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.286042332649231,
      "learning_rate": 8.191598531235724e-05,
      "loss": 2.1697,
      "step": 174890
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.5127562284469604,
      "learning_rate": 8.19131763686575e-05,
      "loss": 1.9695,
      "step": 174891
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.282406210899353,
      "learning_rate": 8.191036746550416e-05,
      "loss": 2.3235,
      "step": 174892
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.1797207593917847,
      "learning_rate": 8.190755860289774e-05,
      "loss": 2.0319,
      "step": 174893
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.25436532497406,
      "learning_rate": 8.19047497808389e-05,
      "loss": 1.9347,
      "step": 174894
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.3133572340011597,
      "learning_rate": 8.190194099932799e-05,
      "loss": 2.1239,
      "step": 174895
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.2659441232681274,
      "learning_rate": 8.189913225836575e-05,
      "loss": 1.9999,
      "step": 174896
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.3828229904174805,
      "learning_rate": 8.189632355795257e-05,
      "loss": 2.0925,
      "step": 174897
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.319060206413269,
      "learning_rate": 8.189351489808895e-05,
      "loss": 1.8326,
      "step": 174898
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.2791659832000732,
      "learning_rate": 8.189070627877536e-05,
      "loss": 2.0692,
      "step": 174899
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.2367595434188843,
      "learning_rate": 8.188789770001251e-05,
      "loss": 1.9874,
      "step": 174900
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.3530670404434204,
      "learning_rate": 8.188508916180074e-05,
      "loss": 2.115,
      "step": 174901
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.376212477684021,
      "learning_rate": 8.18822806641407e-05,
      "loss": 1.9865,
      "step": 174902
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.352474570274353,
      "learning_rate": 8.187947220703291e-05,
      "loss": 1.8687,
      "step": 174903
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.2530078887939453,
      "learning_rate": 8.18766637904778e-05,
      "loss": 2.1805,
      "step": 174904
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.265328049659729,
      "learning_rate": 8.187385541447585e-05,
      "loss": 1.9949,
      "step": 174905
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.3254973888397217,
      "learning_rate": 8.187104707902779e-05,
      "loss": 1.8633,
      "step": 174906
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.3158131837844849,
      "learning_rate": 8.186823878413389e-05,
      "loss": 2.0679,
      "step": 174907
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.2018455266952515,
      "learning_rate": 8.186543052979492e-05,
      "loss": 1.8439,
      "step": 174908
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.2858772277832031,
      "learning_rate": 8.186262231601127e-05,
      "loss": 2.146,
      "step": 174909
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.1714861392974854,
      "learning_rate": 8.185981414278348e-05,
      "loss": 2.0185,
      "step": 174910
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.2347729206085205,
      "learning_rate": 8.185700601011196e-05,
      "loss": 2.0845,
      "step": 174911
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.2381315231323242,
      "learning_rate": 8.185419791799745e-05,
      "loss": 1.9117,
      "step": 174912
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.4592667818069458,
      "learning_rate": 8.185138986644028e-05,
      "loss": 2.1717,
      "step": 174913
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.2022523880004883,
      "learning_rate": 8.184858185544115e-05,
      "loss": 1.9066,
      "step": 174914
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.5011025667190552,
      "learning_rate": 8.184577388500045e-05,
      "loss": 1.9328,
      "step": 174915
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.4528210163116455,
      "learning_rate": 8.184296595511878e-05,
      "loss": 2.0227,
      "step": 174916
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.349167823791504,
      "learning_rate": 8.184015806579648e-05,
      "loss": 2.0603,
      "step": 174917
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.3177803754806519,
      "learning_rate": 8.183735021703436e-05,
      "loss": 2.0833,
      "step": 174918
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.6252638101577759,
      "learning_rate": 8.183454240883268e-05,
      "loss": 1.9886,
      "step": 174919
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.3402423858642578,
      "learning_rate": 8.183173464119215e-05,
      "loss": 2.0084,
      "step": 174920
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.362500548362732,
      "learning_rate": 8.182892691411326e-05,
      "loss": 2.2029,
      "step": 174921
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.2218663692474365,
      "learning_rate": 8.182611922759644e-05,
      "loss": 2.0558,
      "step": 174922
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.3987659215927124,
      "learning_rate": 8.182331158164222e-05,
      "loss": 2.1402,
      "step": 174923
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.3838967084884644,
      "learning_rate": 8.182050397625123e-05,
      "loss": 2.2141,
      "step": 174924
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.4914262294769287,
      "learning_rate": 8.181769641142386e-05,
      "loss": 1.9694,
      "step": 174925
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.4276093244552612,
      "learning_rate": 8.181488888716075e-05,
      "loss": 2.1012,
      "step": 174926
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.3020819425582886,
      "learning_rate": 8.181208140346239e-05,
      "loss": 1.9159,
      "step": 174927
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.3942211866378784,
      "learning_rate": 8.180927396032929e-05,
      "loss": 2.0474,
      "step": 174928
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.425911545753479,
      "learning_rate": 8.180646655776183e-05,
      "loss": 2.0716,
      "step": 174929
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.4208269119262695,
      "learning_rate": 8.18036591957608e-05,
      "loss": 1.9508,
      "step": 174930
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.3478291034698486,
      "learning_rate": 8.180085187432653e-05,
      "loss": 2.1913,
      "step": 174931
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.270336389541626,
      "learning_rate": 8.179804459345963e-05,
      "loss": 1.8475,
      "step": 174932
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.3388382196426392,
      "learning_rate": 8.17952373531606e-05,
      "loss": 2.0646,
      "step": 174933
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.2796169519424438,
      "learning_rate": 8.179243015343e-05,
      "loss": 1.8944,
      "step": 174934
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.4253087043762207,
      "learning_rate": 8.178962299426818e-05,
      "loss": 2.0991,
      "step": 174935
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.4173834323883057,
      "learning_rate": 8.178681587567588e-05,
      "loss": 1.9654,
      "step": 174936
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.2195976972579956,
      "learning_rate": 8.178400879765344e-05,
      "loss": 2.0924,
      "step": 174937
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.2908167839050293,
      "learning_rate": 8.178120176020156e-05,
      "loss": 1.9812,
      "step": 174938
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.32831609249115,
      "learning_rate": 8.177839476332067e-05,
      "loss": 2.1352,
      "step": 174939
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.337562084197998,
      "learning_rate": 8.177558780701122e-05,
      "loss": 2.0353,
      "step": 174940
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.262561559677124,
      "learning_rate": 8.17727808912739e-05,
      "loss": 2.0924,
      "step": 174941
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.3704001903533936,
      "learning_rate": 8.176997401610914e-05,
      "loss": 1.8916,
      "step": 174942
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.4973881244659424,
      "learning_rate": 8.176716718151736e-05,
      "loss": 2.0346,
      "step": 174943
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.236677885055542,
      "learning_rate": 8.176436038749925e-05,
      "loss": 1.8148,
      "step": 174944
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.26691734790802,
      "learning_rate": 8.176155363405533e-05,
      "loss": 1.9492,
      "step": 174945
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.3258305788040161,
      "learning_rate": 8.17587469211859e-05,
      "loss": 2.0031,
      "step": 174946
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.2122471332550049,
      "learning_rate": 8.175594024889176e-05,
      "loss": 1.9561,
      "step": 174947
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.2817835807800293,
      "learning_rate": 8.175313361717329e-05,
      "loss": 2.1157,
      "step": 174948
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.2980103492736816,
      "learning_rate": 8.175032702603097e-05,
      "loss": 1.8407,
      "step": 174949
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.383245825767517,
      "learning_rate": 8.174752047546545e-05,
      "loss": 2.0355,
      "step": 174950
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.2335342168807983,
      "learning_rate": 8.174471396547722e-05,
      "loss": 1.8989,
      "step": 174951
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.3386238813400269,
      "learning_rate": 8.174190749606664e-05,
      "loss": 1.9372,
      "step": 174952
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.3212180137634277,
      "learning_rate": 8.17391010672345e-05,
      "loss": 2.1084,
      "step": 174953
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.3450382947921753,
      "learning_rate": 8.173629467898104e-05,
      "loss": 2.1191,
      "step": 174954
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.3078075647354126,
      "learning_rate": 8.173348833130703e-05,
      "loss": 1.8424,
      "step": 174955
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.2399804592132568,
      "learning_rate": 8.17306820242129e-05,
      "loss": 2.097,
      "step": 174956
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.2596927881240845,
      "learning_rate": 8.172787575769902e-05,
      "loss": 2.0138,
      "step": 174957
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.2193925380706787,
      "learning_rate": 8.172506953176616e-05,
      "loss": 2.4023,
      "step": 174958
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.2797516584396362,
      "learning_rate": 8.172226334641474e-05,
      "loss": 1.9296,
      "step": 174959
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.3268349170684814,
      "learning_rate": 8.171945720164519e-05,
      "loss": 1.7773,
      "step": 174960
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.2825899124145508,
      "learning_rate": 8.17166510974582e-05,
      "loss": 2.0538,
      "step": 174961
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.3152745962142944,
      "learning_rate": 8.171384503385419e-05,
      "loss": 1.9052,
      "step": 174962
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.236496925354004,
      "learning_rate": 8.171103901083362e-05,
      "loss": 2.0726,
      "step": 174963
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.3415273427963257,
      "learning_rate": 8.170823302839718e-05,
      "loss": 2.0261,
      "step": 174964
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.2057560682296753,
      "learning_rate": 8.170542708654528e-05,
      "loss": 2.159,
      "step": 174965
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.332637071609497,
      "learning_rate": 8.170262118527838e-05,
      "loss": 1.998,
      "step": 174966
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.2162418365478516,
      "learning_rate": 8.169981532459719e-05,
      "loss": 1.902,
      "step": 174967
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.2431299686431885,
      "learning_rate": 8.169700950450202e-05,
      "loss": 2.081,
      "step": 174968
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.1726102828979492,
      "learning_rate": 8.169420372499359e-05,
      "loss": 1.9184,
      "step": 174969
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.240199089050293,
      "learning_rate": 8.169139798607232e-05,
      "loss": 2.0916,
      "step": 174970
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.283726453781128,
      "learning_rate": 8.168859228773876e-05,
      "loss": 2.0344,
      "step": 174971
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.2762012481689453,
      "learning_rate": 8.16857866299933e-05,
      "loss": 2.2564,
      "step": 174972
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.2677799463272095,
      "learning_rate": 8.168298101283667e-05,
      "loss": 1.9039,
      "step": 174973
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.3343284130096436,
      "learning_rate": 8.168017543626921e-05,
      "loss": 1.995,
      "step": 174974
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.2910023927688599,
      "learning_rate": 8.167736990029163e-05,
      "loss": 2.2087,
      "step": 174975
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.2806519269943237,
      "learning_rate": 8.167456440490433e-05,
      "loss": 1.7958,
      "step": 174976
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.3406686782836914,
      "learning_rate": 8.167175895010786e-05,
      "loss": 2.0723,
      "step": 174977
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.3965917825698853,
      "learning_rate": 8.166895353590261e-05,
      "loss": 1.9654,
      "step": 174978
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.253035545349121,
      "learning_rate": 8.166614816228934e-05,
      "loss": 1.9287,
      "step": 174979
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.222485899925232,
      "learning_rate": 8.166334282926836e-05,
      "loss": 1.7046,
      "step": 174980
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.329190969467163,
      "learning_rate": 8.16605375368404e-05,
      "loss": 2.2086,
      "step": 174981
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.3306941986083984,
      "learning_rate": 8.165773228500583e-05,
      "loss": 2.0569,
      "step": 174982
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.2798396348953247,
      "learning_rate": 8.165492707376521e-05,
      "loss": 2.1719,
      "step": 174983
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.3111039400100708,
      "learning_rate": 8.165212190311901e-05,
      "loss": 2.0887,
      "step": 174984
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.3025100231170654,
      "learning_rate": 8.164931677306786e-05,
      "loss": 1.785,
      "step": 174985
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.2384119033813477,
      "learning_rate": 8.164651168361214e-05,
      "loss": 2.0083,
      "step": 174986
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.1970878839492798,
      "learning_rate": 8.164370663475252e-05,
      "loss": 2.0571,
      "step": 174987
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.3069311380386353,
      "learning_rate": 8.164090162648949e-05,
      "loss": 1.8898,
      "step": 174988
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.329972505569458,
      "learning_rate": 8.163809665882356e-05,
      "loss": 2.0415,
      "step": 174989
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.2090040445327759,
      "learning_rate": 8.163529173175509e-05,
      "loss": 1.9396,
      "step": 174990
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.3055977821350098,
      "learning_rate": 8.163248684528486e-05,
      "loss": 2.1104,
      "step": 174991
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.3877415657043457,
      "learning_rate": 8.162968199941316e-05,
      "loss": 2.1083,
      "step": 174992
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.2581822872161865,
      "learning_rate": 8.162687719414072e-05,
      "loss": 1.9037,
      "step": 174993
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.336393117904663,
      "learning_rate": 8.162407242946798e-05,
      "loss": 2.1729,
      "step": 174994
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.2854951620101929,
      "learning_rate": 8.162126770539546e-05,
      "loss": 1.9833,
      "step": 174995
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.2297602891921997,
      "learning_rate": 8.161846302192355e-05,
      "loss": 1.9907,
      "step": 174996
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.3430655002593994,
      "learning_rate": 8.1615658379053e-05,
      "loss": 1.828,
      "step": 174997
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.4679991006851196,
      "learning_rate": 8.161285377678411e-05,
      "loss": 1.7616,
      "step": 174998
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.3290069103240967,
      "learning_rate": 8.161004921511762e-05,
      "loss": 2.1744,
      "step": 174999
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.2164912223815918,
      "learning_rate": 8.160724469405393e-05,
      "loss": 1.926,
      "step": 175000
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.2647299766540527,
      "learning_rate": 8.16044402135936e-05,
      "loss": 2.0214,
      "step": 175001
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.4246259927749634,
      "learning_rate": 8.1601635773737e-05,
      "loss": 1.8736,
      "step": 175002
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.2885457277297974,
      "learning_rate": 8.159883137448491e-05,
      "loss": 1.9746,
      "step": 175003
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.2967994213104248,
      "learning_rate": 8.15960270158376e-05,
      "loss": 2.0978,
      "step": 175004
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.362962007522583,
      "learning_rate": 8.15932226977958e-05,
      "loss": 1.8988,
      "step": 175005
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.2079986333847046,
      "learning_rate": 8.159041842035996e-05,
      "loss": 2.2975,
      "step": 175006
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.2244513034820557,
      "learning_rate": 8.158761418353058e-05,
      "loss": 2.0329,
      "step": 175007
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.2916178703308105,
      "learning_rate": 8.158480998730808e-05,
      "loss": 2.0591,
      "step": 175008
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.3703503608703613,
      "learning_rate": 8.158200583169321e-05,
      "loss": 2.1645,
      "step": 175009
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.3379483222961426,
      "learning_rate": 8.157920171668625e-05,
      "loss": 1.6663,
      "step": 175010
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.3559097051620483,
      "learning_rate": 8.157639764228793e-05,
      "loss": 2.2382,
      "step": 175011
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.242967128753662,
      "learning_rate": 8.157359360849867e-05,
      "loss": 1.9385,
      "step": 175012
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.327370524406433,
      "learning_rate": 8.157078961531903e-05,
      "loss": 2.0668,
      "step": 175013
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.3349910974502563,
      "learning_rate": 8.15679856627494e-05,
      "loss": 1.8402,
      "step": 175014
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.3837746381759644,
      "learning_rate": 8.156518175079052e-05,
      "loss": 1.9752,
      "step": 175015
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.2529066801071167,
      "learning_rate": 8.156237787944269e-05,
      "loss": 1.9344,
      "step": 175016
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.398285150527954,
      "learning_rate": 8.15595740487066e-05,
      "loss": 2.0014,
      "step": 175017
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.3817847967147827,
      "learning_rate": 8.155677025858272e-05,
      "loss": 2.0447,
      "step": 175018
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.6139487028121948,
      "learning_rate": 8.155396650907157e-05,
      "loss": 2.0385,
      "step": 175019
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.2563430070877075,
      "learning_rate": 8.155116280017359e-05,
      "loss": 1.8846,
      "step": 175020
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.274240493774414,
      "learning_rate": 8.154835913188946e-05,
      "loss": 2.1446,
      "step": 175021
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.2489908933639526,
      "learning_rate": 8.15455555042195e-05,
      "loss": 2.0664,
      "step": 175022
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.6460013389587402,
      "learning_rate": 8.154275191716442e-05,
      "loss": 1.8809,
      "step": 175023
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.447145938873291,
      "learning_rate": 8.153994837072458e-05,
      "loss": 1.9955,
      "step": 175024
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.27712082862854,
      "learning_rate": 8.15371448649008e-05,
      "loss": 2.0499,
      "step": 175025
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.3408281803131104,
      "learning_rate": 8.153434139969318e-05,
      "loss": 1.9523,
      "step": 175026
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.3862833976745605,
      "learning_rate": 8.153153797510258e-05,
      "loss": 1.9827,
      "step": 175027
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.3512065410614014,
      "learning_rate": 8.152873459112925e-05,
      "loss": 1.9541,
      "step": 175028
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.4509888887405396,
      "learning_rate": 8.152593124777397e-05,
      "loss": 1.9442,
      "step": 175029
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.2956069707870483,
      "learning_rate": 8.152312794503706e-05,
      "loss": 2.1186,
      "step": 175030
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.2079353332519531,
      "learning_rate": 8.152032468291919e-05,
      "loss": 1.8653,
      "step": 175031
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.2025835514068604,
      "learning_rate": 8.151752146142084e-05,
      "loss": 2.0999,
      "step": 175032
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.1957024335861206,
      "learning_rate": 8.151471828054251e-05,
      "loss": 1.9739,
      "step": 175033
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.2731845378875732,
      "learning_rate": 8.15119151402846e-05,
      "loss": 1.9546,
      "step": 175034
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.2026861906051636,
      "learning_rate": 8.150911204064786e-05,
      "loss": 1.9928,
      "step": 175035
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.3974605798721313,
      "learning_rate": 8.15063089816326e-05,
      "loss": 2.0412,
      "step": 175036
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.3127775192260742,
      "learning_rate": 8.150350596323954e-05,
      "loss": 1.966,
      "step": 175037
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.365545392036438,
      "learning_rate": 8.150070298546911e-05,
      "loss": 2.0627,
      "step": 175038
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.3120317459106445,
      "learning_rate": 8.149790004832173e-05,
      "loss": 1.8925,
      "step": 175039
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.3713680505752563,
      "learning_rate": 8.149509715179809e-05,
      "loss": 2.0307,
      "step": 175040
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.3083789348602295,
      "learning_rate": 8.149229429589864e-05,
      "loss": 2.0106,
      "step": 175041
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.2921541929244995,
      "learning_rate": 8.148949148062381e-05,
      "loss": 1.9611,
      "step": 175042
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.3032201528549194,
      "learning_rate": 8.14866887059743e-05,
      "loss": 2.096,
      "step": 175043
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.2228947877883911,
      "learning_rate": 8.148388597195054e-05,
      "loss": 1.932,
      "step": 175044
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.3375710248947144,
      "learning_rate": 8.148108327855293e-05,
      "loss": 2.0302,
      "step": 175045
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.3107702732086182,
      "learning_rate": 8.147828062578225e-05,
      "loss": 2.1527,
      "step": 175046
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.3188914060592651,
      "learning_rate": 8.147547801363885e-05,
      "loss": 1.8048,
      "step": 175047
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.26670503616333,
      "learning_rate": 8.147267544212322e-05,
      "loss": 2.0533,
      "step": 175048
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.2903897762298584,
      "learning_rate": 8.146987291123604e-05,
      "loss": 2.0987,
      "step": 175049
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.343363642692566,
      "learning_rate": 8.146707042097771e-05,
      "loss": 2.2383,
      "step": 175050
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.3469337224960327,
      "learning_rate": 8.146426797134869e-05,
      "loss": 2.1388,
      "step": 175051
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.273134469985962,
      "learning_rate": 8.14614655623497e-05,
      "loss": 2.0779,
      "step": 175052
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.2355505228042603,
      "learning_rate": 8.1458663193981e-05,
      "loss": 2.0914,
      "step": 175053
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.2602964639663696,
      "learning_rate": 8.14558608662434e-05,
      "loss": 2.1052,
      "step": 175054
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.347421646118164,
      "learning_rate": 8.145305857913728e-05,
      "loss": 2.2282,
      "step": 175055
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.33501136302948,
      "learning_rate": 8.145025633266315e-05,
      "loss": 2.095,
      "step": 175056
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.3586682081222534,
      "learning_rate": 8.144745412682144e-05,
      "loss": 2.118,
      "step": 175057
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.4335466623306274,
      "learning_rate": 8.14446519616129e-05,
      "loss": 2.2124,
      "step": 175058
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.1991033554077148,
      "learning_rate": 8.14418498370378e-05,
      "loss": 2.0969,
      "step": 175059
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.501107096672058,
      "learning_rate": 8.143904775309689e-05,
      "loss": 1.9063,
      "step": 175060
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.523910403251648,
      "learning_rate": 8.14362457097906e-05,
      "loss": 2.0205,
      "step": 175061
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.3304553031921387,
      "learning_rate": 8.143344370711942e-05,
      "loss": 2.1398,
      "step": 175062
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.2399686574935913,
      "learning_rate": 8.143064174508379e-05,
      "loss": 1.889,
      "step": 175063
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.372815489768982,
      "learning_rate": 8.142783982368442e-05,
      "loss": 2.0968,
      "step": 175064
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.177618145942688,
      "learning_rate": 8.142503794292169e-05,
      "loss": 1.9998,
      "step": 175065
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.1885727643966675,
      "learning_rate": 8.142223610279622e-05,
      "loss": 1.9648,
      "step": 175066
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.2954553365707397,
      "learning_rate": 8.14194343033085e-05,
      "loss": 2.1698,
      "step": 175067
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.202437400817871,
      "learning_rate": 8.141663254445904e-05,
      "loss": 2.0137,
      "step": 175068
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.2705813646316528,
      "learning_rate": 8.141383082624824e-05,
      "loss": 2.0749,
      "step": 175069
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.3587994575500488,
      "learning_rate": 8.141102914867684e-05,
      "loss": 2.1921,
      "step": 175070
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.3668156862258911,
      "learning_rate": 8.140822751174518e-05,
      "loss": 1.8378,
      "step": 175071
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.300959587097168,
      "learning_rate": 8.140542591545393e-05,
      "loss": 2.1851,
      "step": 175072
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.3502289056777954,
      "learning_rate": 8.140262435980353e-05,
      "loss": 2.1776,
      "step": 175073
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.293080449104309,
      "learning_rate": 8.139982284479451e-05,
      "loss": 2.013,
      "step": 175074
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.2728357315063477,
      "learning_rate": 8.139702137042733e-05,
      "loss": 1.8247,
      "step": 175075
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.6069170236587524,
      "learning_rate": 8.139421993670263e-05,
      "loss": 2.2969,
      "step": 175076
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.3360975980758667,
      "learning_rate": 8.139141854362079e-05,
      "loss": 1.9853,
      "step": 175077
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.2099931240081787,
      "learning_rate": 8.138861719118247e-05,
      "loss": 2.0291,
      "step": 175078
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.3884403705596924,
      "learning_rate": 8.138581587938817e-05,
      "loss": 2.0445,
      "step": 175079
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.351615309715271,
      "learning_rate": 8.138301460823836e-05,
      "loss": 2.0541,
      "step": 175080
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.294783592224121,
      "learning_rate": 8.13802133777335e-05,
      "loss": 2.0601,
      "step": 175081
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.434147834777832,
      "learning_rate": 8.137741218787426e-05,
      "loss": 1.8814,
      "step": 175082
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.2414124011993408,
      "learning_rate": 8.137461103866102e-05,
      "loss": 2.1602,
      "step": 175083
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.4122774600982666,
      "learning_rate": 8.137180993009442e-05,
      "loss": 1.9874,
      "step": 175084
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.293342113494873,
      "learning_rate": 8.136900886217495e-05,
      "loss": 1.7449,
      "step": 175085
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.2429063320159912,
      "learning_rate": 8.136620783490312e-05,
      "loss": 2.0274,
      "step": 175086
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.2057530879974365,
      "learning_rate": 8.136340684827931e-05,
      "loss": 1.7118,
      "step": 175087
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.4530797004699707,
      "learning_rate": 8.136060590230429e-05,
      "loss": 2.1648,
      "step": 175088
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.2279138565063477,
      "learning_rate": 8.135780499697835e-05,
      "loss": 2.1871,
      "step": 175089
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.2184157371520996,
      "learning_rate": 8.135500413230223e-05,
      "loss": 1.8817,
      "step": 175090
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.1889663934707642,
      "learning_rate": 8.135220330827625e-05,
      "loss": 1.8752,
      "step": 175091
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.477678656578064,
      "learning_rate": 8.13494025249012e-05,
      "loss": 2.0505,
      "step": 175092
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.2591913938522339,
      "learning_rate": 8.134660178217726e-05,
      "loss": 2.0739,
      "step": 175093
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.3052316904067993,
      "learning_rate": 8.134380108010516e-05,
      "loss": 1.9674,
      "step": 175094
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.2922017574310303,
      "learning_rate": 8.134100041868532e-05,
      "loss": 2.0732,
      "step": 175095
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.4697763919830322,
      "learning_rate": 8.133819979791839e-05,
      "loss": 2.0895,
      "step": 175096
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.257042646408081,
      "learning_rate": 8.133539921780474e-05,
      "loss": 1.8696,
      "step": 175097
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.35678231716156,
      "learning_rate": 8.133259867834514e-05,
      "loss": 2.0336,
      "step": 175098
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.3263828754425049,
      "learning_rate": 8.132979817953975e-05,
      "loss": 2.1926,
      "step": 175099
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.291373610496521,
      "learning_rate": 8.132699772138938e-05,
      "loss": 2.0332,
      "step": 175100
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.2400497198104858,
      "learning_rate": 8.132419730389434e-05,
      "loss": 1.8374,
      "step": 175101
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.3284235000610352,
      "learning_rate": 8.13213969270554e-05,
      "loss": 2.294,
      "step": 175102
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.3349485397338867,
      "learning_rate": 8.131859659087278e-05,
      "loss": 1.9289,
      "step": 175103
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.1725963354110718,
      "learning_rate": 8.131579629534739e-05,
      "loss": 1.9314,
      "step": 175104
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.291408658027649,
      "learning_rate": 8.131299604047934e-05,
      "loss": 1.8987,
      "step": 175105
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.2743638753890991,
      "learning_rate": 8.13101958262694e-05,
      "loss": 1.7763,
      "step": 175106
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.4550601243972778,
      "learning_rate": 8.130739565271793e-05,
      "loss": 1.9442,
      "step": 175107
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.304176688194275,
      "learning_rate": 8.130459551982564e-05,
      "loss": 2.0195,
      "step": 175108
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.340467095375061,
      "learning_rate": 8.130179542759286e-05,
      "loss": 2.087,
      "step": 175109
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.1567678451538086,
      "learning_rate": 8.12989953760204e-05,
      "loss": 2.0517,
      "step": 175110
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.2723418474197388,
      "learning_rate": 8.12961953651084e-05,
      "loss": 1.8608,
      "step": 175111
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.2885795831680298,
      "learning_rate": 8.129339539485768e-05,
      "loss": 1.8821,
      "step": 175112
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.2375428676605225,
      "learning_rate": 8.12905954652685e-05,
      "loss": 1.986,
      "step": 175113
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.3159611225128174,
      "learning_rate": 8.128779557634166e-05,
      "loss": 1.93,
      "step": 175114
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.3360689878463745,
      "learning_rate": 8.128499572807745e-05,
      "loss": 2.168,
      "step": 175115
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.4035923480987549,
      "learning_rate": 8.128219592047655e-05,
      "loss": 2.1363,
      "step": 175116
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.266836166381836,
      "learning_rate": 8.127939615353944e-05,
      "loss": 1.9982,
      "step": 175117
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.402658462524414,
      "learning_rate": 8.127659642726661e-05,
      "loss": 2.1492,
      "step": 175118
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.4271886348724365,
      "learning_rate": 8.127379674165854e-05,
      "loss": 2.3162,
      "step": 175119
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.3202711343765259,
      "learning_rate": 8.127099709671583e-05,
      "loss": 2.0742,
      "step": 175120
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.3722927570343018,
      "learning_rate": 8.126819749243893e-05,
      "loss": 1.9312,
      "step": 175121
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.3881422281265259,
      "learning_rate": 8.126539792882845e-05,
      "loss": 1.9998,
      "step": 175122
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.3544769287109375,
      "learning_rate": 8.12625984058849e-05,
      "loss": 2.0205,
      "step": 175123
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.5005908012390137,
      "learning_rate": 8.125979892360874e-05,
      "loss": 2.0941,
      "step": 175124
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.2963649034500122,
      "learning_rate": 8.12569994820004e-05,
      "loss": 2.151,
      "step": 175125
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.3953311443328857,
      "learning_rate": 8.125420008106065e-05,
      "loss": 2.0441,
      "step": 175126
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.4554537534713745,
      "learning_rate": 8.125140072078976e-05,
      "loss": 2.1213,
      "step": 175127
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.2295345067977905,
      "learning_rate": 8.124860140118844e-05,
      "loss": 1.8603,
      "step": 175128
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.2933804988861084,
      "learning_rate": 8.124580212225715e-05,
      "loss": 2.2835,
      "step": 175129
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.359266996383667,
      "learning_rate": 8.124300288399633e-05,
      "loss": 2.0872,
      "step": 175130
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.3180432319641113,
      "learning_rate": 8.124020368640661e-05,
      "loss": 1.9006,
      "step": 175131
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.4249672889709473,
      "learning_rate": 8.123740452948849e-05,
      "loss": 2.0541,
      "step": 175132
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.3118255138397217,
      "learning_rate": 8.123460541324237e-05,
      "loss": 2.0381,
      "step": 175133
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.3391779661178589,
      "learning_rate": 8.123180633766895e-05,
      "loss": 1.9344,
      "step": 175134
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.3913971185684204,
      "learning_rate": 8.122900730276869e-05,
      "loss": 1.8795,
      "step": 175135
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.3945130109786987,
      "learning_rate": 8.122620830854196e-05,
      "loss": 2.0606,
      "step": 175136
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.371949315071106,
      "learning_rate": 8.122340935498953e-05,
      "loss": 2.1486,
      "step": 175137
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.3234137296676636,
      "learning_rate": 8.122061044211169e-05,
      "loss": 2.0738,
      "step": 175138
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.7270679473876953,
      "learning_rate": 8.121781156990917e-05,
      "loss": 2.0086,
      "step": 175139
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.6063754558563232,
      "learning_rate": 8.121501273838238e-05,
      "loss": 1.9271,
      "step": 175140
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.4113742113113403,
      "learning_rate": 8.121221394753189e-05,
      "loss": 2.151,
      "step": 175141
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.1235965490341187,
      "learning_rate": 8.120941519735804e-05,
      "loss": 2.0287,
      "step": 175142
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.4502406120300293,
      "learning_rate": 8.120661648786158e-05,
      "loss": 1.8661,
      "step": 175143
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.4501323699951172,
      "learning_rate": 8.120381781904285e-05,
      "loss": 2.003,
      "step": 175144
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.304477572441101,
      "learning_rate": 8.120101919090258e-05,
      "loss": 1.9659,
      "step": 175145
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.24448823928833,
      "learning_rate": 8.119822060344114e-05,
      "loss": 2.1308,
      "step": 175146
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.2950615882873535,
      "learning_rate": 8.11954220566591e-05,
      "loss": 1.8494,
      "step": 175147
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.2299998998641968,
      "learning_rate": 8.119262355055688e-05,
      "loss": 2.2381,
      "step": 175148
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.6649980545043945,
      "learning_rate": 8.118982508513515e-05,
      "loss": 1.9813,
      "step": 175149
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.2660126686096191,
      "learning_rate": 8.118702666039428e-05,
      "loss": 2.1744,
      "step": 175150
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.3201299905776978,
      "learning_rate": 8.118422827633494e-05,
      "loss": 1.8673,
      "step": 175151
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.2502585649490356,
      "learning_rate": 8.118142993295762e-05,
      "loss": 2.132,
      "step": 175152
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.230228304862976,
      "learning_rate": 8.11786316302628e-05,
      "loss": 2.0176,
      "step": 175153
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.359009861946106,
      "learning_rate": 8.117583336825088e-05,
      "loss": 1.9606,
      "step": 175154
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.384713888168335,
      "learning_rate": 8.117303514692261e-05,
      "loss": 2.0917,
      "step": 175155
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.4420422315597534,
      "learning_rate": 8.117023696627832e-05,
      "loss": 1.8754,
      "step": 175156
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.3949060440063477,
      "learning_rate": 8.116743882631871e-05,
      "loss": 1.9466,
      "step": 175157
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.4255101680755615,
      "learning_rate": 8.116464072704409e-05,
      "loss": 2.0668,
      "step": 175158
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.5861737728118896,
      "learning_rate": 8.11618426684553e-05,
      "loss": 1.8985,
      "step": 175159
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.7188977003097534,
      "learning_rate": 8.115904465055246e-05,
      "loss": 1.9011,
      "step": 175160
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.379622459411621,
      "learning_rate": 8.115624667333637e-05,
      "loss": 1.8565,
      "step": 175161
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.3233871459960938,
      "learning_rate": 8.115344873680738e-05,
      "loss": 1.9804,
      "step": 175162
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.5356594324111938,
      "learning_rate": 8.115065084096617e-05,
      "loss": 1.9949,
      "step": 175163
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.556745171546936,
      "learning_rate": 8.114785298581311e-05,
      "loss": 2.1013,
      "step": 175164
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.575162649154663,
      "learning_rate": 8.1145055171349e-05,
      "loss": 2.1965,
      "step": 175165
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.3437776565551758,
      "learning_rate": 8.114225739757395e-05,
      "loss": 1.8886,
      "step": 175166
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.289086937904358,
      "learning_rate": 8.113945966448879e-05,
      "loss": 2.1229,
      "step": 175167
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.2659077644348145,
      "learning_rate": 8.11366619720938e-05,
      "loss": 1.8596,
      "step": 175168
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.303609848022461,
      "learning_rate": 8.113386432038977e-05,
      "loss": 2.0793,
      "step": 175169
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.2420505285263062,
      "learning_rate": 8.113106670937697e-05,
      "loss": 2.0618,
      "step": 175170
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.287553071975708,
      "learning_rate": 8.112826913905624e-05,
      "loss": 2.1522,
      "step": 175171
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.2746869325637817,
      "learning_rate": 8.112547160942773e-05,
      "loss": 1.9248,
      "step": 175172
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.4937800168991089,
      "learning_rate": 8.112267412049219e-05,
      "loss": 2.1791,
      "step": 175173
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.2549774646759033,
      "learning_rate": 8.111987667224998e-05,
      "loss": 1.841,
      "step": 175174
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.2068365812301636,
      "learning_rate": 8.11170792647018e-05,
      "loss": 1.8593,
      "step": 175175
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.2458126544952393,
      "learning_rate": 8.111428189784804e-05,
      "loss": 1.9058,
      "step": 175176
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.1755850315093994,
      "learning_rate": 8.11114845716894e-05,
      "loss": 1.9251,
      "step": 175177
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.2427120208740234,
      "learning_rate": 8.110868728622611e-05,
      "loss": 1.9151,
      "step": 175178
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.4339818954467773,
      "learning_rate": 8.110589004145895e-05,
      "loss": 1.9254,
      "step": 175179
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.1934921741485596,
      "learning_rate": 8.110309283738822e-05,
      "loss": 2.0491,
      "step": 175180
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.1708048582077026,
      "learning_rate": 8.110029567401466e-05,
      "loss": 1.8745,
      "step": 175181
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.2552863359451294,
      "learning_rate": 8.109749855133861e-05,
      "loss": 2.1174,
      "step": 175182
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.2387131452560425,
      "learning_rate": 8.109470146936081e-05,
      "loss": 1.8872,
      "step": 175183
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.2982280254364014,
      "learning_rate": 8.109190442808149e-05,
      "loss": 2.2629,
      "step": 175184
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.3859996795654297,
      "learning_rate": 8.10891074275014e-05,
      "loss": 2.0312,
      "step": 175185
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.3435791730880737,
      "learning_rate": 8.108631046762088e-05,
      "loss": 2.4139,
      "step": 175186
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.378400444984436,
      "learning_rate": 8.108351354844065e-05,
      "loss": 1.9594,
      "step": 175187
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.244482159614563,
      "learning_rate": 8.108071666996104e-05,
      "loss": 1.9268,
      "step": 175188
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.4125252962112427,
      "learning_rate": 8.107791983218279e-05,
      "loss": 2.0536,
      "step": 175189
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.3759170770645142,
      "learning_rate": 8.107512303510616e-05,
      "loss": 1.9506,
      "step": 175190
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.6555558443069458,
      "learning_rate": 8.107232627873187e-05,
      "loss": 1.8675,
      "step": 175191
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.2616499662399292,
      "learning_rate": 8.106952956306029e-05,
      "loss": 1.8373,
      "step": 175192
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.3255738019943237,
      "learning_rate": 8.106673288809208e-05,
      "loss": 2.1601,
      "step": 175193
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.3428791761398315,
      "learning_rate": 8.10639362538276e-05,
      "loss": 2.2075,
      "step": 175194
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.2578459978103638,
      "learning_rate": 8.106113966026769e-05,
      "loss": 2.0947,
      "step": 175195
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.3151557445526123,
      "learning_rate": 8.105834310741246e-05,
      "loss": 1.8309,
      "step": 175196
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.3255770206451416,
      "learning_rate": 8.105554659526268e-05,
      "loss": 2.0608,
      "step": 175197
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.3300896883010864,
      "learning_rate": 8.105275012381873e-05,
      "loss": 1.9089,
      "step": 175198
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.2808195352554321,
      "learning_rate": 8.10499536930813e-05,
      "loss": 2.0707,
      "step": 175199
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.2350138425827026,
      "learning_rate": 8.104715730305072e-05,
      "loss": 2.2686,
      "step": 175200
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.295597791671753,
      "learning_rate": 8.104436095372769e-05,
      "loss": 1.9224,
      "step": 175201
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.2382296323776245,
      "learning_rate": 8.104156464511266e-05,
      "loss": 1.8808,
      "step": 175202
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.2862542867660522,
      "learning_rate": 8.103876837720615e-05,
      "loss": 1.8048,
      "step": 175203
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.2689645290374756,
      "learning_rate": 8.103597215000852e-05,
      "loss": 1.9869,
      "step": 175204
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.3949477672576904,
      "learning_rate": 8.103317596352056e-05,
      "loss": 2.212,
      "step": 175205
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.3783942461013794,
      "learning_rate": 8.103037981774258e-05,
      "loss": 1.9637,
      "step": 175206
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.2276548147201538,
      "learning_rate": 8.10275837126753e-05,
      "loss": 1.7377,
      "step": 175207
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.3740954399108887,
      "learning_rate": 8.102478764831909e-05,
      "loss": 1.9239,
      "step": 175208
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.5490330457687378,
      "learning_rate": 8.102199162467451e-05,
      "loss": 1.8669,
      "step": 175209
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.3123780488967896,
      "learning_rate": 8.1019195641742e-05,
      "loss": 2.1008,
      "step": 175210
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.2488043308258057,
      "learning_rate": 8.101639969952224e-05,
      "loss": 1.9102,
      "step": 175211
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.2322335243225098,
      "learning_rate": 8.101360379801557e-05,
      "loss": 1.9141,
      "step": 175212
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.3878669738769531,
      "learning_rate": 8.101080793722268e-05,
      "loss": 1.9311,
      "step": 175213
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.4025242328643799,
      "learning_rate": 8.100801211714403e-05,
      "loss": 2.1444,
      "step": 175214
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.3879131078720093,
      "learning_rate": 8.100521633778005e-05,
      "loss": 2.0835,
      "step": 175215
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.374405026435852,
      "learning_rate": 8.10024205991314e-05,
      "loss": 2.103,
      "step": 175216
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.232415795326233,
      "learning_rate": 8.099962490119855e-05,
      "loss": 2.0489,
      "step": 175217
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.5380252599716187,
      "learning_rate": 8.099682924398192e-05,
      "loss": 1.9103,
      "step": 175218
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.3817780017852783,
      "learning_rate": 8.099403362748219e-05,
      "loss": 2.0816,
      "step": 175219
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.2609431743621826,
      "learning_rate": 8.09912380516998e-05,
      "loss": 1.8915,
      "step": 175220
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.289917230606079,
      "learning_rate": 8.098844251663518e-05,
      "loss": 2.094,
      "step": 175221
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.3192784786224365,
      "learning_rate": 8.098564702228904e-05,
      "loss": 1.8678,
      "step": 175222
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.6466561555862427,
      "learning_rate": 8.09828515686617e-05,
      "loss": 2.0777,
      "step": 175223
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.2175981998443604,
      "learning_rate": 8.098005615575388e-05,
      "loss": 1.8976,
      "step": 175224
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.3150852918624878,
      "learning_rate": 8.097726078356603e-05,
      "loss": 2.1529,
      "step": 175225
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.3358855247497559,
      "learning_rate": 8.097446545209863e-05,
      "loss": 2.1198,
      "step": 175226
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.2872817516326904,
      "learning_rate": 8.097167016135209e-05,
      "loss": 2.2158,
      "step": 175227
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.436285376548767,
      "learning_rate": 8.096887491132714e-05,
      "loss": 2.1217,
      "step": 175228
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.2632852792739868,
      "learning_rate": 8.096607970202415e-05,
      "loss": 2.1249,
      "step": 175229
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.3778510093688965,
      "learning_rate": 8.09632845334438e-05,
      "loss": 2.2033,
      "step": 175230
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.2744412422180176,
      "learning_rate": 8.096048940558649e-05,
      "loss": 2.0314,
      "step": 175231
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.2909868955612183,
      "learning_rate": 8.095769431845275e-05,
      "loss": 1.8846,
      "step": 175232
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.7805655002593994,
      "learning_rate": 8.095489927204303e-05,
      "loss": 1.9849,
      "step": 175233
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.3506379127502441,
      "learning_rate": 8.095210426635803e-05,
      "loss": 1.9628,
      "step": 175234
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.4478245973587036,
      "learning_rate": 8.094930930139805e-05,
      "loss": 1.8586,
      "step": 175235
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.2577087879180908,
      "learning_rate": 8.094651437716384e-05,
      "loss": 1.947,
      "step": 175236
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.5694528818130493,
      "learning_rate": 8.094371949365571e-05,
      "loss": 2.1522,
      "step": 175237
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.5211182832717896,
      "learning_rate": 8.094092465087447e-05,
      "loss": 1.9628,
      "step": 175238
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.1771409511566162,
      "learning_rate": 8.093812984882026e-05,
      "loss": 1.8594,
      "step": 175239
    },
    {
      "epoch": 2.28,
      "grad_norm": 2.8523099422454834,
      "learning_rate": 8.093533508749385e-05,
      "loss": 2.2137,
      "step": 175240
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.289381980895996,
      "learning_rate": 8.093254036689564e-05,
      "loss": 2.1452,
      "step": 175241
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.2769460678100586,
      "learning_rate": 8.09297456870263e-05,
      "loss": 1.9411,
      "step": 175242
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.1625657081604004,
      "learning_rate": 8.092695104788616e-05,
      "loss": 2.1323,
      "step": 175243
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.3149878978729248,
      "learning_rate": 8.092415644947601e-05,
      "loss": 1.9366,
      "step": 175244
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.2548577785491943,
      "learning_rate": 8.092136189179601e-05,
      "loss": 2.0371,
      "step": 175245
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.2684491872787476,
      "learning_rate": 8.091856737484696e-05,
      "loss": 2.0194,
      "step": 175246
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.4007072448730469,
      "learning_rate": 8.091577289862918e-05,
      "loss": 1.929,
      "step": 175247
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.3012347221374512,
      "learning_rate": 8.09129784631434e-05,
      "loss": 1.9768,
      "step": 175248
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.3006585836410522,
      "learning_rate": 8.091018406838993e-05,
      "loss": 2.0438,
      "step": 175249
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.2988239526748657,
      "learning_rate": 8.09073897143696e-05,
      "loss": 1.854,
      "step": 175250
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.4163925647735596,
      "learning_rate": 8.09045954010825e-05,
      "loss": 2.2944,
      "step": 175251
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.230630874633789,
      "learning_rate": 8.09018011285295e-05,
      "loss": 2.0825,
      "step": 175252
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.1616402864456177,
      "learning_rate": 8.089900689671088e-05,
      "loss": 1.8534,
      "step": 175253
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.3068373203277588,
      "learning_rate": 8.089621270562736e-05,
      "loss": 2.1976,
      "step": 175254
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.3600399494171143,
      "learning_rate": 8.089341855527926e-05,
      "loss": 2.095,
      "step": 175255
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.3748620748519897,
      "learning_rate": 8.089062444566742e-05,
      "loss": 1.9097,
      "step": 175256
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.1962567567825317,
      "learning_rate": 8.088783037679199e-05,
      "loss": 1.6419,
      "step": 175257
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.2565522193908691,
      "learning_rate": 8.08850363486537e-05,
      "loss": 2.0956,
      "step": 175258
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.4515386819839478,
      "learning_rate": 8.088224236125292e-05,
      "loss": 1.831,
      "step": 175259
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.5547453165054321,
      "learning_rate": 8.08794484145904e-05,
      "loss": 2.3089,
      "step": 175260
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.3144093751907349,
      "learning_rate": 8.087665450866641e-05,
      "loss": 2.0334,
      "step": 175261
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.2949540615081787,
      "learning_rate": 8.087386064348178e-05,
      "loss": 2.1156,
      "step": 175262
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.312847375869751,
      "learning_rate": 8.087106681903662e-05,
      "loss": 1.9364,
      "step": 175263
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.2561057806015015,
      "learning_rate": 8.086827303533178e-05,
      "loss": 1.9517,
      "step": 175264
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.3010034561157227,
      "learning_rate": 8.086547929236757e-05,
      "loss": 2.2078,
      "step": 175265
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.3584579229354858,
      "learning_rate": 8.086268559014471e-05,
      "loss": 2.0581,
      "step": 175266
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.214713454246521,
      "learning_rate": 8.08598919286635e-05,
      "loss": 1.8371,
      "step": 175267
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.3060301542282104,
      "learning_rate": 8.085709830792479e-05,
      "loss": 2.0964,
      "step": 175268
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.3402258157730103,
      "learning_rate": 8.085430472792866e-05,
      "loss": 1.9769,
      "step": 175269
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.237032413482666,
      "learning_rate": 8.085151118867597e-05,
      "loss": 2.0994,
      "step": 175270
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.221847653388977,
      "learning_rate": 8.0848717690167e-05,
      "loss": 1.9509,
      "step": 175271
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.3738124370574951,
      "learning_rate": 8.084592423240253e-05,
      "loss": 2.0058,
      "step": 175272
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.3718888759613037,
      "learning_rate": 8.084313081538279e-05,
      "loss": 1.9936,
      "step": 175273
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.3664464950561523,
      "learning_rate": 8.084033743910867e-05,
      "loss": 1.8906,
      "step": 175274
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.382100224494934,
      "learning_rate": 8.083754410358029e-05,
      "loss": 1.8493,
      "step": 175275
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.4537771940231323,
      "learning_rate": 8.083475080879841e-05,
      "loss": 2.193,
      "step": 175276
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.2550920248031616,
      "learning_rate": 8.08319575547634e-05,
      "loss": 1.8899,
      "step": 175277
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.337973713874817,
      "learning_rate": 8.082916434147597e-05,
      "loss": 2.1499,
      "step": 175278
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.4795740842819214,
      "learning_rate": 8.082637116893645e-05,
      "loss": 2.0527,
      "step": 175279
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.4995564222335815,
      "learning_rate": 8.082357803714562e-05,
      "loss": 1.9444,
      "step": 175280
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.2980263233184814,
      "learning_rate": 8.082078494610363e-05,
      "loss": 2.1517,
      "step": 175281
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.3079432249069214,
      "learning_rate": 8.081799189581129e-05,
      "loss": 2.0336,
      "step": 175282
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.3410454988479614,
      "learning_rate": 8.081519888626894e-05,
      "loss": 2.0897,
      "step": 175283
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.3935092687606812,
      "learning_rate": 8.081240591747727e-05,
      "loss": 1.907,
      "step": 175284
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.2663830518722534,
      "learning_rate": 8.080961298943663e-05,
      "loss": 2.2163,
      "step": 175285
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.2812632322311401,
      "learning_rate": 8.080682010214769e-05,
      "loss": 1.9883,
      "step": 175286
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.278115153312683,
      "learning_rate": 8.080402725561092e-05,
      "loss": 1.9672,
      "step": 175287
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.2741694450378418,
      "learning_rate": 8.080123444982681e-05,
      "loss": 1.9529,
      "step": 175288
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.343704342842102,
      "learning_rate": 8.079844168479577e-05,
      "loss": 1.8361,
      "step": 175289
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.2621136903762817,
      "learning_rate": 8.079564896051857e-05,
      "loss": 1.8113,
      "step": 175290
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.393816590309143,
      "learning_rate": 8.079285627699547e-05,
      "loss": 2.0571,
      "step": 175291
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.2583400011062622,
      "learning_rate": 8.079006363422725e-05,
      "loss": 2.0126,
      "step": 175292
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.4768040180206299,
      "learning_rate": 8.078727103221427e-05,
      "loss": 1.901,
      "step": 175293
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.4835617542266846,
      "learning_rate": 8.078447847095707e-05,
      "loss": 2.1616,
      "step": 175294
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.3497458696365356,
      "learning_rate": 8.07816859504561e-05,
      "loss": 1.8582,
      "step": 175295
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.2730225324630737,
      "learning_rate": 8.077889347071203e-05,
      "loss": 1.969,
      "step": 175296
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.3547840118408203,
      "learning_rate": 8.077610103172521e-05,
      "loss": 1.9595,
      "step": 175297
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.292537808418274,
      "learning_rate": 8.077330863349633e-05,
      "loss": 1.9051,
      "step": 175298
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.3394362926483154,
      "learning_rate": 8.077051627602585e-05,
      "loss": 2.1367,
      "step": 175299
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.3210296630859375,
      "learning_rate": 8.07677239593142e-05,
      "loss": 2.101,
      "step": 175300
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.2487988471984863,
      "learning_rate": 8.076493168336201e-05,
      "loss": 2.0094,
      "step": 175301
    },
    {
      "epoch": 2.28,
      "grad_norm": 2.1169674396514893,
      "learning_rate": 8.076213944816978e-05,
      "loss": 2.045,
      "step": 175302
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.158886194229126,
      "learning_rate": 8.075934725373795e-05,
      "loss": 1.9955,
      "step": 175303
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.2704501152038574,
      "learning_rate": 8.075655510006715e-05,
      "loss": 2.0293,
      "step": 175304
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.356719732284546,
      "learning_rate": 8.075376298715783e-05,
      "loss": 1.9615,
      "step": 175305
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.2837883234024048,
      "learning_rate": 8.075097091501044e-05,
      "loss": 2.0507,
      "step": 175306
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.364886999130249,
      "learning_rate": 8.074817888362569e-05,
      "loss": 2.1707,
      "step": 175307
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.412033200263977,
      "learning_rate": 8.074538689300399e-05,
      "loss": 2.1037,
      "step": 175308
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.2805204391479492,
      "learning_rate": 8.07425949431458e-05,
      "loss": 1.9412,
      "step": 175309
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.272897720336914,
      "learning_rate": 8.073980303405174e-05,
      "loss": 2.2277,
      "step": 175310
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.3075896501541138,
      "learning_rate": 8.073701116572231e-05,
      "loss": 2.0219,
      "step": 175311
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.3146202564239502,
      "learning_rate": 8.073421933815792e-05,
      "loss": 1.9599,
      "step": 175312
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.391918659210205,
      "learning_rate": 8.073142755135925e-05,
      "loss": 1.9536,
      "step": 175313
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.2360970973968506,
      "learning_rate": 8.072863580532667e-05,
      "loss": 1.9181,
      "step": 175314
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.398433804512024,
      "learning_rate": 8.072584410006088e-05,
      "loss": 2.1134,
      "step": 175315
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.3701294660568237,
      "learning_rate": 8.072305243556227e-05,
      "loss": 1.9097,
      "step": 175316
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.3737800121307373,
      "learning_rate": 8.072026081183143e-05,
      "loss": 1.9055,
      "step": 175317
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.4078494310379028,
      "learning_rate": 8.07174692288687e-05,
      "loss": 2.0211,
      "step": 175318
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.303313970565796,
      "learning_rate": 8.071467768667483e-05,
      "loss": 2.1199,
      "step": 175319
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.222325086593628,
      "learning_rate": 8.071188618525015e-05,
      "loss": 2.3523,
      "step": 175320
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.4449272155761719,
      "learning_rate": 8.070909472459536e-05,
      "loss": 1.8995,
      "step": 175321
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.1571155786514282,
      "learning_rate": 8.070630330471078e-05,
      "loss": 2.0615,
      "step": 175322
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.3634233474731445,
      "learning_rate": 8.070351192559727e-05,
      "loss": 1.9573,
      "step": 175323
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.4158395528793335,
      "learning_rate": 8.07007205872549e-05,
      "loss": 2.2368,
      "step": 175324
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.2999173402786255,
      "learning_rate": 8.069792928968449e-05,
      "loss": 2.0385,
      "step": 175325
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.4087984561920166,
      "learning_rate": 8.069513803288639e-05,
      "loss": 1.8689,
      "step": 175326
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.2316988706588745,
      "learning_rate": 8.069234681686131e-05,
      "loss": 1.9833,
      "step": 175327
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.2826980352401733,
      "learning_rate": 8.068955564160955e-05,
      "loss": 2.0373,
      "step": 175328
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.3181759119033813,
      "learning_rate": 8.068676450713194e-05,
      "loss": 2.0572,
      "step": 175329
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.2823313474655151,
      "learning_rate": 8.068397341342861e-05,
      "loss": 2.2301,
      "step": 175330
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.23972749710083,
      "learning_rate": 8.068118236050034e-05,
      "loss": 2.0625,
      "step": 175331
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.5290871858596802,
      "learning_rate": 8.067839134834752e-05,
      "loss": 2.1116,
      "step": 175332
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.345328450202942,
      "learning_rate": 8.06756003769708e-05,
      "loss": 2.0426,
      "step": 175333
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.723855972290039,
      "learning_rate": 8.067280944637054e-05,
      "loss": 2.0561,
      "step": 175334
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.191562533378601,
      "learning_rate": 8.067001855654753e-05,
      "loss": 1.9278,
      "step": 175335
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.3785219192504883,
      "learning_rate": 8.066722770750193e-05,
      "loss": 2.0282,
      "step": 175336
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.2023890018463135,
      "learning_rate": 8.066443689923452e-05,
      "loss": 1.9315,
      "step": 175337
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.2096922397613525,
      "learning_rate": 8.066164613174562e-05,
      "loss": 1.8664,
      "step": 175338
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.4373269081115723,
      "learning_rate": 8.065885540503596e-05,
      "loss": 2.1506,
      "step": 175339
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.4006327390670776,
      "learning_rate": 8.065606471910588e-05,
      "loss": 2.1761,
      "step": 175340
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.1821953058242798,
      "learning_rate": 8.065327407395616e-05,
      "loss": 1.805,
      "step": 175341
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.3268131017684937,
      "learning_rate": 8.065048346958694e-05,
      "loss": 2.1595,
      "step": 175342
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.3779932260513306,
      "learning_rate": 8.064769290599903e-05,
      "loss": 2.0385,
      "step": 175343
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.3974782228469849,
      "learning_rate": 8.06449023831928e-05,
      "loss": 1.8691,
      "step": 175344
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.2385432720184326,
      "learning_rate": 8.064211190116888e-05,
      "loss": 2.0525,
      "step": 175345
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.29184091091156,
      "learning_rate": 8.063932145992764e-05,
      "loss": 1.9864,
      "step": 175346
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.3440043926239014,
      "learning_rate": 8.063653105946988e-05,
      "loss": 2.0056,
      "step": 175347
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.3102669715881348,
      "learning_rate": 8.063374069979573e-05,
      "loss": 2.1095,
      "step": 175348
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.3134913444519043,
      "learning_rate": 8.063095038090602e-05,
      "loss": 2.1216,
      "step": 175349
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.287610411643982,
      "learning_rate": 8.062816010280104e-05,
      "loss": 2.0014,
      "step": 175350
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.513202428817749,
      "learning_rate": 8.062536986548155e-05,
      "loss": 2.1427,
      "step": 175351
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.3458954095840454,
      "learning_rate": 8.062257966894787e-05,
      "loss": 2.074,
      "step": 175352
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.3459985256195068,
      "learning_rate": 8.061978951320072e-05,
      "loss": 2.0181,
      "step": 175353
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.2480378150939941,
      "learning_rate": 8.061699939824034e-05,
      "loss": 2.0725,
      "step": 175354
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.3012269735336304,
      "learning_rate": 8.061420932406748e-05,
      "loss": 2.0583,
      "step": 175355
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.3527756929397583,
      "learning_rate": 8.061141929068253e-05,
      "loss": 1.9913,
      "step": 175356
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.3373360633850098,
      "learning_rate": 8.06086292980861e-05,
      "loss": 2.1646,
      "step": 175357
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.3679980039596558,
      "learning_rate": 8.060583934627859e-05,
      "loss": 1.7844,
      "step": 175358
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.341657280921936,
      "learning_rate": 8.060304943526068e-05,
      "loss": 2.0576,
      "step": 175359
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.360065221786499,
      "learning_rate": 8.06002595650328e-05,
      "loss": 2.0798,
      "step": 175360
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.3071906566619873,
      "learning_rate": 8.059746973559551e-05,
      "loss": 2.0764,
      "step": 175361
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.296608328819275,
      "learning_rate": 8.059467994694918e-05,
      "loss": 1.9485,
      "step": 175362
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.2140610218048096,
      "learning_rate": 8.059189019909452e-05,
      "loss": 2.0429,
      "step": 175363
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.2750117778778076,
      "learning_rate": 8.05891004920319e-05,
      "loss": 2.0465,
      "step": 175364
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.4277936220169067,
      "learning_rate": 8.058631082576198e-05,
      "loss": 2.061,
      "step": 175365
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.2353686094284058,
      "learning_rate": 8.058352120028523e-05,
      "loss": 2.0317,
      "step": 175366
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.1586824655532837,
      "learning_rate": 8.058073161560212e-05,
      "loss": 1.8178,
      "step": 175367
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.2135009765625,
      "learning_rate": 8.057794207171313e-05,
      "loss": 2.0989,
      "step": 175368
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.2577794790267944,
      "learning_rate": 8.057515256861888e-05,
      "loss": 2.0036,
      "step": 175369
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.3960027694702148,
      "learning_rate": 8.05723631063198e-05,
      "loss": 1.9873,
      "step": 175370
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.19771409034729,
      "learning_rate": 8.056957368481655e-05,
      "loss": 1.9989,
      "step": 175371
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.2359070777893066,
      "learning_rate": 8.056678430410957e-05,
      "loss": 1.9943,
      "step": 175372
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.3092879056930542,
      "learning_rate": 8.056399496419936e-05,
      "loss": 2.0626,
      "step": 175373
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.25620436668396,
      "learning_rate": 8.056120566508635e-05,
      "loss": 2.0419,
      "step": 175374
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.2534782886505127,
      "learning_rate": 8.055841640677123e-05,
      "loss": 1.963,
      "step": 175375
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.203589677810669,
      "learning_rate": 8.055562718925435e-05,
      "loss": 1.9992,
      "step": 175376
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.3529726266860962,
      "learning_rate": 8.055283801253642e-05,
      "loss": 2.1159,
      "step": 175377
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.5378044843673706,
      "learning_rate": 8.055004887661786e-05,
      "loss": 1.9814,
      "step": 175378
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.2819222211837769,
      "learning_rate": 8.054725978149921e-05,
      "loss": 2.1229,
      "step": 175379
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.3272076845169067,
      "learning_rate": 8.054447072718087e-05,
      "loss": 1.9009,
      "step": 175380
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.259971261024475,
      "learning_rate": 8.054168171366355e-05,
      "loss": 1.9008,
      "step": 175381
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.3393394947052002,
      "learning_rate": 8.053889274094754e-05,
      "loss": 2.1627,
      "step": 175382
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.2835206985473633,
      "learning_rate": 8.053610380903362e-05,
      "loss": 1.9423,
      "step": 175383
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.2523698806762695,
      "learning_rate": 8.053331491792218e-05,
      "loss": 1.9208,
      "step": 175384
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.2562083005905151,
      "learning_rate": 8.053052606761363e-05,
      "loss": 1.9506,
      "step": 175385
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.2692077159881592,
      "learning_rate": 8.05277372581087e-05,
      "loss": 2.1167,
      "step": 175386
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.2083934545516968,
      "learning_rate": 8.052494848940783e-05,
      "loss": 1.894,
      "step": 175387
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.4020036458969116,
      "learning_rate": 8.052215976151138e-05,
      "loss": 2.1899,
      "step": 175388
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.2628557682037354,
      "learning_rate": 8.051937107442014e-05,
      "loss": 2.036,
      "step": 175389
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.2636367082595825,
      "learning_rate": 8.051658242813445e-05,
      "loss": 2.2868,
      "step": 175390
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.3649183511734009,
      "learning_rate": 8.051379382265478e-05,
      "loss": 2.1026,
      "step": 175391
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.2304619550704956,
      "learning_rate": 8.051100525798188e-05,
      "loss": 2.1301,
      "step": 175392
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.2692745923995972,
      "learning_rate": 8.05082167341161e-05,
      "loss": 2.1129,
      "step": 175393
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.2909005880355835,
      "learning_rate": 8.050542825105787e-05,
      "loss": 2.2264,
      "step": 175394
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.3600819110870361,
      "learning_rate": 8.050263980880794e-05,
      "loss": 2.3174,
      "step": 175395
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.4153238534927368,
      "learning_rate": 8.049985140736669e-05,
      "loss": 2.0779,
      "step": 175396
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.2495949268341064,
      "learning_rate": 8.04970630467346e-05,
      "loss": 1.9653,
      "step": 175397
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.3448196649551392,
      "learning_rate": 8.049427472691233e-05,
      "loss": 1.9584,
      "step": 175398
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.2224665880203247,
      "learning_rate": 8.049148644790022e-05,
      "loss": 1.8643,
      "step": 175399
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.2904767990112305,
      "learning_rate": 8.048869820969897e-05,
      "loss": 1.8613,
      "step": 175400
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.2239588499069214,
      "learning_rate": 8.048591001230902e-05,
      "loss": 2.1325,
      "step": 175401
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.3822968006134033,
      "learning_rate": 8.048312185573092e-05,
      "loss": 2.1634,
      "step": 175402
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.185599446296692,
      "learning_rate": 8.048033373996501e-05,
      "loss": 1.8008,
      "step": 175403
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.2640175819396973,
      "learning_rate": 8.047754566501207e-05,
      "loss": 2.0649,
      "step": 175404
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.2707628011703491,
      "learning_rate": 8.047475763087239e-05,
      "loss": 1.8421,
      "step": 175405
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.5010321140289307,
      "learning_rate": 8.047196963754668e-05,
      "loss": 2.1124,
      "step": 175406
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.3148815631866455,
      "learning_rate": 8.04691816850354e-05,
      "loss": 2.0207,
      "step": 175407
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.2092310190200806,
      "learning_rate": 8.046639377333903e-05,
      "loss": 1.9766,
      "step": 175408
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.2513691186904907,
      "learning_rate": 8.046360590245801e-05,
      "loss": 2.1567,
      "step": 175409
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.30474054813385,
      "learning_rate": 8.046081807239306e-05,
      "loss": 2.0795,
      "step": 175410
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.5463593006134033,
      "learning_rate": 8.045803028314448e-05,
      "loss": 2.0796,
      "step": 175411
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.4770786762237549,
      "learning_rate": 8.045524253471298e-05,
      "loss": 2.1183,
      "step": 175412
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.316072702407837,
      "learning_rate": 8.045245482709892e-05,
      "loss": 1.9665,
      "step": 175413
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.3666186332702637,
      "learning_rate": 8.044966716030305e-05,
      "loss": 2.0395,
      "step": 175414
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.9363054037094116,
      "learning_rate": 8.044687953432557e-05,
      "loss": 1.9092,
      "step": 175415
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.7416049242019653,
      "learning_rate": 8.044409194916724e-05,
      "loss": 1.8374,
      "step": 175416
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.4308265447616577,
      "learning_rate": 8.044130440482841e-05,
      "loss": 1.901,
      "step": 175417
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.4895862340927124,
      "learning_rate": 8.04385169013098e-05,
      "loss": 2.2619,
      "step": 175418
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.7364872694015503,
      "learning_rate": 8.043572943861171e-05,
      "loss": 2.242,
      "step": 175419
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.552951455116272,
      "learning_rate": 8.043294201673493e-05,
      "loss": 1.7733,
      "step": 175420
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.1838115453720093,
      "learning_rate": 8.043015463567963e-05,
      "loss": 2.0436,
      "step": 175421
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.2714643478393555,
      "learning_rate": 8.04273672954466e-05,
      "loss": 2.0581,
      "step": 175422
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.5191863775253296,
      "learning_rate": 8.042457999603619e-05,
      "loss": 2.0053,
      "step": 175423
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.1435447931289673,
      "learning_rate": 8.042179273744908e-05,
      "loss": 1.9663,
      "step": 175424
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.6083354949951172,
      "learning_rate": 8.04190055196856e-05,
      "loss": 2.0335,
      "step": 175425
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.43416428565979,
      "learning_rate": 8.041621834274648e-05,
      "loss": 2.0434,
      "step": 175426
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.2363117933273315,
      "learning_rate": 8.041343120663212e-05,
      "loss": 1.9762,
      "step": 175427
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.2160556316375732,
      "learning_rate": 8.041064411134306e-05,
      "loss": 1.954,
      "step": 175428
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.2396773099899292,
      "learning_rate": 8.040785705687971e-05,
      "loss": 1.9469,
      "step": 175429
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.2574259042739868,
      "learning_rate": 8.040507004324277e-05,
      "loss": 2.1522,
      "step": 175430
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.2087364196777344,
      "learning_rate": 8.040228307043257e-05,
      "loss": 2.1091,
      "step": 175431
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.2742929458618164,
      "learning_rate": 8.039949613844985e-05,
      "loss": 1.8724,
      "step": 175432
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.528793454170227,
      "learning_rate": 8.0396709247295e-05,
      "loss": 2.1552,
      "step": 175433
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.2069576978683472,
      "learning_rate": 8.039392239696853e-05,
      "loss": 2.025,
      "step": 175434
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.3361315727233887,
      "learning_rate": 8.03911355874709e-05,
      "loss": 2.0228,
      "step": 175435
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.3009436130523682,
      "learning_rate": 8.03883488188028e-05,
      "loss": 2.1352,
      "step": 175436
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.259531855583191,
      "learning_rate": 8.038556209096454e-05,
      "loss": 2.0735,
      "step": 175437
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.3361780643463135,
      "learning_rate": 8.038277540395685e-05,
      "loss": 2.0765,
      "step": 175438
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.28104567527771,
      "learning_rate": 8.037998875778014e-05,
      "loss": 1.9236,
      "step": 175439
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.2602275609970093,
      "learning_rate": 8.037720215243498e-05,
      "loss": 2.0768,
      "step": 175440
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.3806298971176147,
      "learning_rate": 8.03744155879217e-05,
      "loss": 2.1367,
      "step": 175441
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.2339082956314087,
      "learning_rate": 8.037162906424107e-05,
      "loss": 1.9863,
      "step": 175442
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.3121649026870728,
      "learning_rate": 8.036884258139341e-05,
      "loss": 2.1397,
      "step": 175443
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.2477476596832275,
      "learning_rate": 8.036605613937942e-05,
      "loss": 1.8662,
      "step": 175444
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.379805564880371,
      "learning_rate": 8.036326973819953e-05,
      "loss": 1.9054,
      "step": 175445
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.281725525856018,
      "learning_rate": 8.036048337785425e-05,
      "loss": 2.0268,
      "step": 175446
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.2689114809036255,
      "learning_rate": 8.035769705834399e-05,
      "loss": 1.8604,
      "step": 175447
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.286409854888916,
      "learning_rate": 8.035491077966951e-05,
      "loss": 1.9815,
      "step": 175448
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.2669965028762817,
      "learning_rate": 8.035212454183109e-05,
      "loss": 1.9706,
      "step": 175449
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.3457595109939575,
      "learning_rate": 8.034933834482945e-05,
      "loss": 2.0039,
      "step": 175450
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.359581470489502,
      "learning_rate": 8.0346552188665e-05,
      "loss": 1.9563,
      "step": 175451
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.1845884323120117,
      "learning_rate": 8.034376607333828e-05,
      "loss": 1.9029,
      "step": 175452
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.3242100477218628,
      "learning_rate": 8.034097999884971e-05,
      "loss": 2.0876,
      "step": 175453
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.536933422088623,
      "learning_rate": 8.03381939652e-05,
      "loss": 1.8932,
      "step": 175454
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.2201160192489624,
      "learning_rate": 8.033540797238948e-05,
      "loss": 1.9194,
      "step": 175455
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.3914141654968262,
      "learning_rate": 8.033262202041882e-05,
      "loss": 1.9271,
      "step": 175456
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.2264600992202759,
      "learning_rate": 8.032983610928848e-05,
      "loss": 1.9516,
      "step": 175457
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.274727702140808,
      "learning_rate": 8.032705023899897e-05,
      "loss": 2.01,
      "step": 175458
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.3352171182632446,
      "learning_rate": 8.032426440955072e-05,
      "loss": 2.0745,
      "step": 175459
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.2690881490707397,
      "learning_rate": 8.032147862094442e-05,
      "loss": 2.2276,
      "step": 175460
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.5859500169754028,
      "learning_rate": 8.031869287318042e-05,
      "loss": 1.9886,
      "step": 175461
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.2605987787246704,
      "learning_rate": 8.031590716625942e-05,
      "loss": 1.9614,
      "step": 175462
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.4044193029403687,
      "learning_rate": 8.031312150018184e-05,
      "loss": 2.0786,
      "step": 175463
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.3201619386672974,
      "learning_rate": 8.03103358749482e-05,
      "loss": 2.1722,
      "step": 175464
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.3473273515701294,
      "learning_rate": 8.03075502905589e-05,
      "loss": 1.9856,
      "step": 175465
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.2750784158706665,
      "learning_rate": 8.030476474701468e-05,
      "loss": 2.1897,
      "step": 175466
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.2985649108886719,
      "learning_rate": 8.030197924431589e-05,
      "loss": 2.2485,
      "step": 175467
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.3641642332077026,
      "learning_rate": 8.029919378246315e-05,
      "loss": 2.1623,
      "step": 175468
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.3025531768798828,
      "learning_rate": 8.029640836145696e-05,
      "loss": 2.1412,
      "step": 175469
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.245945692062378,
      "learning_rate": 8.029362298129781e-05,
      "loss": 1.8358,
      "step": 175470
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.4561293125152588,
      "learning_rate": 8.029083764198612e-05,
      "loss": 2.0301,
      "step": 175471
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.3925411701202393,
      "learning_rate": 8.028805234352261e-05,
      "loss": 1.9193,
      "step": 175472
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.4848273992538452,
      "learning_rate": 8.028526708590764e-05,
      "loss": 1.9871,
      "step": 175473
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.333254098892212,
      "learning_rate": 8.028248186914184e-05,
      "loss": 2.1104,
      "step": 175474
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.235329270362854,
      "learning_rate": 8.027969669322567e-05,
      "loss": 1.7936,
      "step": 175475
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.2907154560089111,
      "learning_rate": 8.027691155815958e-05,
      "loss": 1.9753,
      "step": 175476
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.3082072734832764,
      "learning_rate": 8.027412646394425e-05,
      "loss": 1.8802,
      "step": 175477
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.3221826553344727,
      "learning_rate": 8.02713414105801e-05,
      "loss": 1.8573,
      "step": 175478
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.3297860622406006,
      "learning_rate": 8.026855639806759e-05,
      "loss": 2.0429,
      "step": 175479
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.3544508218765259,
      "learning_rate": 8.026577142640736e-05,
      "loss": 2.0488,
      "step": 175480
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.313041090965271,
      "learning_rate": 8.026298649559987e-05,
      "loss": 2.2596,
      "step": 175481
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.3834456205368042,
      "learning_rate": 8.026020160564556e-05,
      "loss": 2.0716,
      "step": 175482
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.3489688634872437,
      "learning_rate": 8.025741675654513e-05,
      "loss": 2.0687,
      "step": 175483
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.4348257780075073,
      "learning_rate": 8.02546319482989e-05,
      "loss": 1.9926,
      "step": 175484
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.417362093925476,
      "learning_rate": 8.025184718090755e-05,
      "loss": 2.1232,
      "step": 175485
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.3496493101119995,
      "learning_rate": 8.024906245437155e-05,
      "loss": 2.1097,
      "step": 175486
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.500430703163147,
      "learning_rate": 8.024627776869141e-05,
      "loss": 2.2419,
      "step": 175487
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.2789497375488281,
      "learning_rate": 8.02434931238675e-05,
      "loss": 1.9488,
      "step": 175488
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.2333338260650635,
      "learning_rate": 8.02407085199006e-05,
      "loss": 2.0221,
      "step": 175489
    },
    {
      "epoch": 2.28,
      "grad_norm": 1.2921247482299805,
      "learning_rate": 8.023792395679097e-05,
      "loss": 2.0715,
      "step": 175490
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.2266908884048462,
      "learning_rate": 8.02351394345394e-05,
      "loss": 1.9435,
      "step": 175491
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.2532551288604736,
      "learning_rate": 8.023235495314623e-05,
      "loss": 2.0529,
      "step": 175492
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.297946572303772,
      "learning_rate": 8.022957051261196e-05,
      "loss": 2.1668,
      "step": 175493
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.2947520017623901,
      "learning_rate": 8.02267861129372e-05,
      "loss": 2.0394,
      "step": 175494
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.5469050407409668,
      "learning_rate": 8.022400175412245e-05,
      "loss": 1.9799,
      "step": 175495
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.386653184890747,
      "learning_rate": 8.022121743616817e-05,
      "loss": 2.1669,
      "step": 175496
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.40531587600708,
      "learning_rate": 8.021843315907494e-05,
      "loss": 2.0666,
      "step": 175497
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.2659088373184204,
      "learning_rate": 8.021564892284318e-05,
      "loss": 1.9632,
      "step": 175498
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.421377182006836,
      "learning_rate": 8.021286472747356e-05,
      "loss": 1.8933,
      "step": 175499
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.360896110534668,
      "learning_rate": 8.021008057296652e-05,
      "loss": 2.1874,
      "step": 175500
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.3368701934814453,
      "learning_rate": 8.020729645932259e-05,
      "loss": 1.9581,
      "step": 175501
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.2755907773971558,
      "learning_rate": 8.020451238654218e-05,
      "loss": 1.9288,
      "step": 175502
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.2119518518447876,
      "learning_rate": 8.020172835462597e-05,
      "loss": 2.0494,
      "step": 175503
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.432829737663269,
      "learning_rate": 8.019894436357435e-05,
      "loss": 1.9993,
      "step": 175504
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.2792662382125854,
      "learning_rate": 8.019616041338797e-05,
      "loss": 2.0184,
      "step": 175505
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.3837230205535889,
      "learning_rate": 8.019337650406726e-05,
      "loss": 2.0058,
      "step": 175506
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.289859414100647,
      "learning_rate": 8.019059263561277e-05,
      "loss": 2.138,
      "step": 175507
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.2523008584976196,
      "learning_rate": 8.018780880802489e-05,
      "loss": 2.0499,
      "step": 175508
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.2100228071212769,
      "learning_rate": 8.018502502130436e-05,
      "loss": 2.1809,
      "step": 175509
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.370201826095581,
      "learning_rate": 8.018224127545145e-05,
      "loss": 1.8246,
      "step": 175510
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.354339361190796,
      "learning_rate": 8.017945757046693e-05,
      "loss": 2.1302,
      "step": 175511
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.3511648178100586,
      "learning_rate": 8.017667390635121e-05,
      "loss": 2.0414,
      "step": 175512
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.344696283340454,
      "learning_rate": 8.017389028310479e-05,
      "loss": 2.2364,
      "step": 175513
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.3192914724349976,
      "learning_rate": 8.017110670072809e-05,
      "loss": 2.1003,
      "step": 175514
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.2844735383987427,
      "learning_rate": 8.016832315922183e-05,
      "loss": 2.0546,
      "step": 175515
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.430374026298523,
      "learning_rate": 8.016553965858635e-05,
      "loss": 1.9915,
      "step": 175516
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.26505708694458,
      "learning_rate": 8.016275619882229e-05,
      "loss": 2.0005,
      "step": 175517
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.499253273010254,
      "learning_rate": 8.015997277993017e-05,
      "loss": 2.3114,
      "step": 175518
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.2605736255645752,
      "learning_rate": 8.015718940191045e-05,
      "loss": 1.9908,
      "step": 175519
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.1970969438552856,
      "learning_rate": 8.015440606476357e-05,
      "loss": 1.9737,
      "step": 175520
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.291095495223999,
      "learning_rate": 8.01516227684902e-05,
      "loss": 1.9951,
      "step": 175521
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.4479076862335205,
      "learning_rate": 8.014883951309072e-05,
      "loss": 2.0352,
      "step": 175522
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.269417643547058,
      "learning_rate": 8.014605629856584e-05,
      "loss": 2.0667,
      "step": 175523
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.2803425788879395,
      "learning_rate": 8.014327312491593e-05,
      "loss": 1.956,
      "step": 175524
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.3219091892242432,
      "learning_rate": 8.014048999214152e-05,
      "loss": 1.9122,
      "step": 175525
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.3357669115066528,
      "learning_rate": 8.013770690024308e-05,
      "loss": 1.9293,
      "step": 175526
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.368714690208435,
      "learning_rate": 8.013492384922127e-05,
      "loss": 1.9939,
      "step": 175527
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.2907483577728271,
      "learning_rate": 8.013214083907644e-05,
      "loss": 2.11,
      "step": 175528
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.354593276977539,
      "learning_rate": 8.012935786980927e-05,
      "loss": 2.1644,
      "step": 175529
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.2337417602539062,
      "learning_rate": 8.012657494142024e-05,
      "loss": 1.9154,
      "step": 175530
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.2400527000427246,
      "learning_rate": 8.012379205390981e-05,
      "loss": 1.9613,
      "step": 175531
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.2157148122787476,
      "learning_rate": 8.012100920727841e-05,
      "loss": 1.7383,
      "step": 175532
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.1784197092056274,
      "learning_rate": 8.011822640152678e-05,
      "loss": 1.9764,
      "step": 175533
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.3118960857391357,
      "learning_rate": 8.01154436366552e-05,
      "loss": 2.0338,
      "step": 175534
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.3183166980743408,
      "learning_rate": 8.011266091266443e-05,
      "loss": 2.1093,
      "step": 175535
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.2921762466430664,
      "learning_rate": 8.010987822955486e-05,
      "loss": 1.971,
      "step": 175536
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.2649999856948853,
      "learning_rate": 8.010709558732703e-05,
      "loss": 2.043,
      "step": 175537
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.3535937070846558,
      "learning_rate": 8.010431298598132e-05,
      "loss": 1.9976,
      "step": 175538
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.2696447372436523,
      "learning_rate": 8.010153042551847e-05,
      "loss": 2.1933,
      "step": 175539
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.2832112312316895,
      "learning_rate": 8.00987479059388e-05,
      "loss": 1.9597,
      "step": 175540
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.391611933708191,
      "learning_rate": 8.009596542724304e-05,
      "loss": 2.1995,
      "step": 175541
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.3065356016159058,
      "learning_rate": 8.009318298943156e-05,
      "loss": 1.9342,
      "step": 175542
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.2583527565002441,
      "learning_rate": 8.009040059250493e-05,
      "loss": 2.0495,
      "step": 175543
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.3174593448638916,
      "learning_rate": 8.008761823646356e-05,
      "loss": 2.1661,
      "step": 175544
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.3275848627090454,
      "learning_rate": 8.008483592130813e-05,
      "loss": 1.9594,
      "step": 175545
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.3093701601028442,
      "learning_rate": 8.0082053647039e-05,
      "loss": 1.8979,
      "step": 175546
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.570083737373352,
      "learning_rate": 8.007927141365684e-05,
      "loss": 1.9817,
      "step": 175547
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.3484992980957031,
      "learning_rate": 8.007648922116213e-05,
      "loss": 2.0652,
      "step": 175548
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.2989965677261353,
      "learning_rate": 8.00737070695553e-05,
      "loss": 2.1716,
      "step": 175549
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.444791316986084,
      "learning_rate": 8.007092495883686e-05,
      "loss": 2.1739,
      "step": 175550
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.4245414733886719,
      "learning_rate": 8.006814288900748e-05,
      "loss": 2.1196,
      "step": 175551
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.4396544694900513,
      "learning_rate": 8.006536086006747e-05,
      "loss": 1.9978,
      "step": 175552
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.405356764793396,
      "learning_rate": 8.006257887201757e-05,
      "loss": 2.1857,
      "step": 175553
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.2110416889190674,
      "learning_rate": 8.00597969248582e-05,
      "loss": 1.7214,
      "step": 175554
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.2028229236602783,
      "learning_rate": 8.005701501858986e-05,
      "loss": 2.0542,
      "step": 175555
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.3107588291168213,
      "learning_rate": 8.005423315321297e-05,
      "loss": 2.0657,
      "step": 175556
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.3309047222137451,
      "learning_rate": 8.005145132872824e-05,
      "loss": 1.8508,
      "step": 175557
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.3957332372665405,
      "learning_rate": 8.004866954513598e-05,
      "loss": 2.3023,
      "step": 175558
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.2162266969680786,
      "learning_rate": 8.004588780243696e-05,
      "loss": 1.8703,
      "step": 175559
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.2881314754486084,
      "learning_rate": 8.004310610063147e-05,
      "loss": 1.9885,
      "step": 175560
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.4641408920288086,
      "learning_rate": 8.004032443972019e-05,
      "loss": 2.0575,
      "step": 175561
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.2504758834838867,
      "learning_rate": 8.003754281970357e-05,
      "loss": 1.8562,
      "step": 175562
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.3380717039108276,
      "learning_rate": 8.00347612405821e-05,
      "loss": 2.174,
      "step": 175563
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.1758068799972534,
      "learning_rate": 8.003197970235625e-05,
      "loss": 1.9933,
      "step": 175564
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.2684868574142456,
      "learning_rate": 8.002919820502672e-05,
      "loss": 1.8047,
      "step": 175565
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.4523789882659912,
      "learning_rate": 8.002641674859381e-05,
      "loss": 1.9889,
      "step": 175566
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.480728268623352,
      "learning_rate": 8.002363533305822e-05,
      "loss": 2.0741,
      "step": 175567
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.263676404953003,
      "learning_rate": 8.002085395842038e-05,
      "loss": 1.9642,
      "step": 175568
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.390334963798523,
      "learning_rate": 8.001807262468082e-05,
      "loss": 2.0951,
      "step": 175569
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.406779408454895,
      "learning_rate": 8.001529133184e-05,
      "loss": 2.0274,
      "step": 175570
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.297197937965393,
      "learning_rate": 8.001251007989855e-05,
      "loss": 1.9702,
      "step": 175571
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.711194396018982,
      "learning_rate": 8.000972886885684e-05,
      "loss": 2.1938,
      "step": 175572
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.5338777303695679,
      "learning_rate": 8.000694769871556e-05,
      "loss": 1.9202,
      "step": 175573
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.2735079526901245,
      "learning_rate": 8.000416656947515e-05,
      "loss": 1.7306,
      "step": 175574
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.1905901432037354,
      "learning_rate": 8.000138548113602e-05,
      "loss": 1.8561,
      "step": 175575
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.3681634664535522,
      "learning_rate": 7.999860443369888e-05,
      "loss": 1.7337,
      "step": 175576
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.2576823234558105,
      "learning_rate": 7.999582342716416e-05,
      "loss": 1.8827,
      "step": 175577
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.2392812967300415,
      "learning_rate": 7.999304246153225e-05,
      "loss": 1.8584,
      "step": 175578
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.4090394973754883,
      "learning_rate": 7.999026153680392e-05,
      "loss": 2.1596,
      "step": 175579
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.371535301208496,
      "learning_rate": 7.998748065297951e-05,
      "loss": 1.8044,
      "step": 175580
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.2713911533355713,
      "learning_rate": 7.998469981005951e-05,
      "loss": 2.1176,
      "step": 175581
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.4452089071273804,
      "learning_rate": 7.998191900804461e-05,
      "loss": 1.9358,
      "step": 175582
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.3646900653839111,
      "learning_rate": 7.997913824693514e-05,
      "loss": 1.8429,
      "step": 175583
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.6053539514541626,
      "learning_rate": 7.997635752673174e-05,
      "loss": 1.9159,
      "step": 175584
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.3548718690872192,
      "learning_rate": 7.997357684743495e-05,
      "loss": 2.0831,
      "step": 175585
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.281899333000183,
      "learning_rate": 7.997079620904517e-05,
      "loss": 1.9409,
      "step": 175586
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.2864484786987305,
      "learning_rate": 7.996801561156291e-05,
      "loss": 1.9303,
      "step": 175587
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.2351933717727661,
      "learning_rate": 7.996523505498886e-05,
      "loss": 1.8607,
      "step": 175588
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.2459620237350464,
      "learning_rate": 7.996245453932331e-05,
      "loss": 2.1521,
      "step": 175589
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.3499314785003662,
      "learning_rate": 7.9959674064567e-05,
      "loss": 1.9571,
      "step": 175590
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.265135407447815,
      "learning_rate": 7.99568936307203e-05,
      "loss": 1.9121,
      "step": 175591
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.3342665433883667,
      "learning_rate": 7.995411323778383e-05,
      "loss": 2.1884,
      "step": 175592
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.4043079614639282,
      "learning_rate": 7.99513328857579e-05,
      "loss": 2.061,
      "step": 175593
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.2450473308563232,
      "learning_rate": 7.994855257464327e-05,
      "loss": 2.1458,
      "step": 175594
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.305187463760376,
      "learning_rate": 7.994577230444028e-05,
      "loss": 1.9778,
      "step": 175595
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.2362885475158691,
      "learning_rate": 7.994299207514963e-05,
      "loss": 2.0407,
      "step": 175596
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.395837664604187,
      "learning_rate": 7.994021188677172e-05,
      "loss": 2.1265,
      "step": 175597
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.4395506381988525,
      "learning_rate": 7.993743173930707e-05,
      "loss": 1.8093,
      "step": 175598
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.3556208610534668,
      "learning_rate": 7.99346516327561e-05,
      "loss": 2.1087,
      "step": 175599
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.251446008682251,
      "learning_rate": 7.993187156711956e-05,
      "loss": 1.9216,
      "step": 175600
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.3514779806137085,
      "learning_rate": 7.992909154239774e-05,
      "loss": 1.951,
      "step": 175601
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.3944703340530396,
      "learning_rate": 7.992631155859131e-05,
      "loss": 1.999,
      "step": 175602
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.453218698501587,
      "learning_rate": 7.992353161570076e-05,
      "loss": 1.9648,
      "step": 175603
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.3342822790145874,
      "learning_rate": 7.992075171372658e-05,
      "loss": 1.9695,
      "step": 175604
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.2693746089935303,
      "learning_rate": 7.991797185266921e-05,
      "loss": 2.1455,
      "step": 175605
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.401149868965149,
      "learning_rate": 7.991519203252933e-05,
      "loss": 1.9412,
      "step": 175606
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.2295829057693481,
      "learning_rate": 7.991241225330726e-05,
      "loss": 2.0912,
      "step": 175607
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.290722370147705,
      "learning_rate": 7.990963251500372e-05,
      "loss": 1.9995,
      "step": 175608
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.3094807863235474,
      "learning_rate": 7.990685281761915e-05,
      "loss": 2.1097,
      "step": 175609
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.1963742971420288,
      "learning_rate": 7.990407316115404e-05,
      "loss": 2.2113,
      "step": 175610
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.231961727142334,
      "learning_rate": 7.99012935456088e-05,
      "loss": 1.9968,
      "step": 175611
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.4751701354980469,
      "learning_rate": 7.989851397098421e-05,
      "loss": 2.3786,
      "step": 175612
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.247396469116211,
      "learning_rate": 7.989573443728052e-05,
      "loss": 1.909,
      "step": 175613
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.3721035718917847,
      "learning_rate": 7.989295494449848e-05,
      "loss": 1.9255,
      "step": 175614
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.2720317840576172,
      "learning_rate": 7.989017549263844e-05,
      "loss": 2.065,
      "step": 175615
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.4402506351470947,
      "learning_rate": 7.988739608170104e-05,
      "loss": 2.1837,
      "step": 175616
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.24245285987854,
      "learning_rate": 7.98846167116866e-05,
      "loss": 1.831,
      "step": 175617
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.3389441967010498,
      "learning_rate": 7.988183738259586e-05,
      "loss": 1.9904,
      "step": 175618
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.367574691772461,
      "learning_rate": 7.987905809442915e-05,
      "loss": 2.0255,
      "step": 175619
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.3630731105804443,
      "learning_rate": 7.987627884718717e-05,
      "loss": 2.2063,
      "step": 175620
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.3303982019424438,
      "learning_rate": 7.987349964087032e-05,
      "loss": 2.2477,
      "step": 175621
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.3627159595489502,
      "learning_rate": 7.987072047547917e-05,
      "loss": 2.1794,
      "step": 175622
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.2717480659484863,
      "learning_rate": 7.986794135101415e-05,
      "loss": 1.9335,
      "step": 175623
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.2295360565185547,
      "learning_rate": 7.986516226747589e-05,
      "loss": 2.0629,
      "step": 175624
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.4843552112579346,
      "learning_rate": 7.986238322486474e-05,
      "loss": 2.169,
      "step": 175625
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.3574708700180054,
      "learning_rate": 7.985960422318144e-05,
      "loss": 1.9352,
      "step": 175626
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.2538038492202759,
      "learning_rate": 7.98568252624263e-05,
      "loss": 1.9319,
      "step": 175627
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.1884294748306274,
      "learning_rate": 7.985404634260011e-05,
      "loss": 2.0709,
      "step": 175628
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.322526454925537,
      "learning_rate": 7.985126746370305e-05,
      "loss": 2.0998,
      "step": 175629
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.4775300025939941,
      "learning_rate": 7.984848862573585e-05,
      "loss": 2.1879,
      "step": 175630
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.3392807245254517,
      "learning_rate": 7.984570982869892e-05,
      "loss": 1.9907,
      "step": 175631
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.2679616212844849,
      "learning_rate": 7.98429310725929e-05,
      "loss": 1.9018,
      "step": 175632
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.508628010749817,
      "learning_rate": 7.984015235741814e-05,
      "loss": 1.9944,
      "step": 175633
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.293056607246399,
      "learning_rate": 7.983737368317544e-05,
      "loss": 2.1643,
      "step": 175634
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.5131142139434814,
      "learning_rate": 7.983459504986494e-05,
      "loss": 1.8977,
      "step": 175635
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.3677600622177124,
      "learning_rate": 7.983181645748745e-05,
      "loss": 1.9566,
      "step": 175636
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.4674854278564453,
      "learning_rate": 7.982903790604326e-05,
      "loss": 2.2092,
      "step": 175637
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.5821280479431152,
      "learning_rate": 7.982625939553312e-05,
      "loss": 2.22,
      "step": 175638
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.3474571704864502,
      "learning_rate": 7.982348092595734e-05,
      "loss": 2.0533,
      "step": 175639
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.737774133682251,
      "learning_rate": 7.982070249731671e-05,
      "loss": 1.9098,
      "step": 175640
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.7595776319503784,
      "learning_rate": 7.981792410961141e-05,
      "loss": 2.0247,
      "step": 175641
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.3497980833053589,
      "learning_rate": 7.981514576284219e-05,
      "loss": 2.1511,
      "step": 175642
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.541366696357727,
      "learning_rate": 7.98123674570094e-05,
      "loss": 1.9873,
      "step": 175643
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.33278226852417,
      "learning_rate": 7.980958919211373e-05,
      "loss": 1.8527,
      "step": 175644
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.3294388055801392,
      "learning_rate": 7.980681096815553e-05,
      "loss": 1.9987,
      "step": 175645
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.3928354978561401,
      "learning_rate": 7.980403278513546e-05,
      "loss": 2.0646,
      "step": 175646
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.7928310632705688,
      "learning_rate": 7.980125464305402e-05,
      "loss": 2.0096,
      "step": 175647
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.5848653316497803,
      "learning_rate": 7.979847654191168e-05,
      "loss": 2.2001,
      "step": 175648
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.2886565923690796,
      "learning_rate": 7.979569848170884e-05,
      "loss": 2.0611,
      "step": 175649
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.3686408996582031,
      "learning_rate": 7.979292046244624e-05,
      "loss": 2.3276,
      "step": 175650
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.2253423929214478,
      "learning_rate": 7.979014248412424e-05,
      "loss": 1.9818,
      "step": 175651
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.358872652053833,
      "learning_rate": 7.978736454674343e-05,
      "loss": 2.0741,
      "step": 175652
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.2399598360061646,
      "learning_rate": 7.978458665030436e-05,
      "loss": 2.2047,
      "step": 175653
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.2091295719146729,
      "learning_rate": 7.978180879480747e-05,
      "loss": 2.1049,
      "step": 175654
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.3568490743637085,
      "learning_rate": 7.977903098025322e-05,
      "loss": 2.1163,
      "step": 175655
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.2992624044418335,
      "learning_rate": 7.977625320664229e-05,
      "loss": 2.0819,
      "step": 175656
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.2987830638885498,
      "learning_rate": 7.977347547397501e-05,
      "loss": 2.1039,
      "step": 175657
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.3908802270889282,
      "learning_rate": 7.977069778225208e-05,
      "loss": 2.0029,
      "step": 175658
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.213480830192566,
      "learning_rate": 7.976792013147397e-05,
      "loss": 1.9043,
      "step": 175659
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.2170621156692505,
      "learning_rate": 7.976514252164106e-05,
      "loss": 2.0551,
      "step": 175660
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.367317795753479,
      "learning_rate": 7.976236495275404e-05,
      "loss": 1.9399,
      "step": 175661
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.3358819484710693,
      "learning_rate": 7.975958742481335e-05,
      "loss": 1.8502,
      "step": 175662
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.3708282709121704,
      "learning_rate": 7.975680993781945e-05,
      "loss": 2.1454,
      "step": 175663
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.5042548179626465,
      "learning_rate": 7.975403249177301e-05,
      "loss": 2.0258,
      "step": 175664
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.4073132276535034,
      "learning_rate": 7.975125508667443e-05,
      "loss": 2.2449,
      "step": 175665
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.281796932220459,
      "learning_rate": 7.974847772252413e-05,
      "loss": 2.1678,
      "step": 175666
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.199980616569519,
      "learning_rate": 7.974570039932288e-05,
      "loss": 2.205,
      "step": 175667
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.3618884086608887,
      "learning_rate": 7.974292311707097e-05,
      "loss": 1.8621,
      "step": 175668
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.6445890665054321,
      "learning_rate": 7.974014587576907e-05,
      "loss": 2.044,
      "step": 175669
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.483447790145874,
      "learning_rate": 7.973736867541764e-05,
      "loss": 2.1105,
      "step": 175670
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.2024905681610107,
      "learning_rate": 7.973459151601723e-05,
      "loss": 2.0231,
      "step": 175671
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.3265284299850464,
      "learning_rate": 7.97318143975682e-05,
      "loss": 2.0284,
      "step": 175672
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.1716994047164917,
      "learning_rate": 7.972903732007126e-05,
      "loss": 2.04,
      "step": 175673
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.3286404609680176,
      "learning_rate": 7.972626028352678e-05,
      "loss": 1.9225,
      "step": 175674
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.4213156700134277,
      "learning_rate": 7.972348328793544e-05,
      "loss": 2.1908,
      "step": 175675
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.2653956413269043,
      "learning_rate": 7.972070633329765e-05,
      "loss": 1.9414,
      "step": 175676
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.3400518894195557,
      "learning_rate": 7.971792941961393e-05,
      "loss": 1.9907,
      "step": 175677
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.2580934762954712,
      "learning_rate": 7.971515254688473e-05,
      "loss": 2.1131,
      "step": 175678
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.393763780593872,
      "learning_rate": 7.971237571511074e-05,
      "loss": 1.9398,
      "step": 175679
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.2796820402145386,
      "learning_rate": 7.970959892429226e-05,
      "loss": 1.9076,
      "step": 175680
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.3547899723052979,
      "learning_rate": 7.970682217443006e-05,
      "loss": 2.0242,
      "step": 175681
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.4575673341751099,
      "learning_rate": 7.970404546552447e-05,
      "loss": 2.1111,
      "step": 175682
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.2797160148620605,
      "learning_rate": 7.970126879757608e-05,
      "loss": 1.8122,
      "step": 175683
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.269659161567688,
      "learning_rate": 7.969849217058529e-05,
      "loss": 2.0186,
      "step": 175684
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.3764432668685913,
      "learning_rate": 7.969571558455278e-05,
      "loss": 1.9578,
      "step": 175685
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.3783928155899048,
      "learning_rate": 7.969293903947893e-05,
      "loss": 2.107,
      "step": 175686
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.1800390481948853,
      "learning_rate": 7.96901625353644e-05,
      "loss": 1.9585,
      "step": 175687
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.3185826539993286,
      "learning_rate": 7.968738607220963e-05,
      "loss": 1.997,
      "step": 175688
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.244411587715149,
      "learning_rate": 7.968460965001514e-05,
      "loss": 1.9476,
      "step": 175689
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.3623377084732056,
      "learning_rate": 7.968183326878135e-05,
      "loss": 1.9966,
      "step": 175690
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.497501254081726,
      "learning_rate": 7.967905692850894e-05,
      "loss": 2.2849,
      "step": 175691
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.2385916709899902,
      "learning_rate": 7.967628062919827e-05,
      "loss": 2.0744,
      "step": 175692
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.4629613161087036,
      "learning_rate": 7.967350437085005e-05,
      "loss": 2.1801,
      "step": 175693
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.2178632020950317,
      "learning_rate": 7.967072815346457e-05,
      "loss": 2.1484,
      "step": 175694
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.3271393775939941,
      "learning_rate": 7.966795197704265e-05,
      "loss": 1.8181,
      "step": 175695
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.426038384437561,
      "learning_rate": 7.966517584158443e-05,
      "loss": 2.1032,
      "step": 175696
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.2739547491073608,
      "learning_rate": 7.966239974709069e-05,
      "loss": 2.203,
      "step": 175697
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.3134933710098267,
      "learning_rate": 7.965962369356181e-05,
      "loss": 2.1215,
      "step": 175698
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.556563377380371,
      "learning_rate": 7.965684768099844e-05,
      "loss": 2.0105,
      "step": 175699
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.2601903676986694,
      "learning_rate": 7.965407170940093e-05,
      "loss": 1.9496,
      "step": 175700
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.2768086194992065,
      "learning_rate": 7.965129577877006e-05,
      "loss": 1.9605,
      "step": 175701
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.2859293222427368,
      "learning_rate": 7.9648519889106e-05,
      "loss": 2.0835,
      "step": 175702
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.3871525526046753,
      "learning_rate": 7.964574404040953e-05,
      "loss": 2.05,
      "step": 175703
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.6469576358795166,
      "learning_rate": 7.9642968232681e-05,
      "loss": 1.8904,
      "step": 175704
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.4229545593261719,
      "learning_rate": 7.964019246592108e-05,
      "loss": 2.0158,
      "step": 175705
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.1769871711730957,
      "learning_rate": 7.963741674013014e-05,
      "loss": 2.2316,
      "step": 175706
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.2353092432022095,
      "learning_rate": 7.963464105530891e-05,
      "loss": 2.0595,
      "step": 175707
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.2860888242721558,
      "learning_rate": 7.963186541145758e-05,
      "loss": 1.9556,
      "step": 175708
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.5945606231689453,
      "learning_rate": 7.962908980857695e-05,
      "loss": 1.8396,
      "step": 175709
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.3626487255096436,
      "learning_rate": 7.962631424666734e-05,
      "loss": 1.9525,
      "step": 175710
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.2262837886810303,
      "learning_rate": 7.962353872572947e-05,
      "loss": 2.0441,
      "step": 175711
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.301384687423706,
      "learning_rate": 7.96207632457636e-05,
      "loss": 1.9082,
      "step": 175712
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.2465424537658691,
      "learning_rate": 7.96179878067706e-05,
      "loss": 2.0347,
      "step": 175713
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.2902754545211792,
      "learning_rate": 7.961521240875061e-05,
      "loss": 1.9894,
      "step": 175714
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.3431241512298584,
      "learning_rate": 7.96124370517044e-05,
      "loss": 2.0377,
      "step": 175715
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.25721275806427,
      "learning_rate": 7.96096617356323e-05,
      "loss": 1.8662,
      "step": 175716
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.1373306512832642,
      "learning_rate": 7.9606886460535e-05,
      "loss": 2.01,
      "step": 175717
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.301407814025879,
      "learning_rate": 7.960411122641286e-05,
      "loss": 2.1101,
      "step": 175718
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.3142588138580322,
      "learning_rate": 7.960133603326667e-05,
      "loss": 2.062,
      "step": 175719
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.3271524906158447,
      "learning_rate": 7.959856088109654e-05,
      "loss": 1.9706,
      "step": 175720
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.281436800956726,
      "learning_rate": 7.959578576990333e-05,
      "loss": 2.1991,
      "step": 175721
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.3972606658935547,
      "learning_rate": 7.95930106996873e-05,
      "loss": 1.9819,
      "step": 175722
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.4590184688568115,
      "learning_rate": 7.959023567044918e-05,
      "loss": 2.1609,
      "step": 175723
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.2883851528167725,
      "learning_rate": 7.958746068218933e-05,
      "loss": 2.1197,
      "step": 175724
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.2465206384658813,
      "learning_rate": 7.95846857349085e-05,
      "loss": 1.9527,
      "step": 175725
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.3796106576919556,
      "learning_rate": 7.958191082860687e-05,
      "loss": 2.05,
      "step": 175726
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.3449050188064575,
      "learning_rate": 7.957913596328518e-05,
      "loss": 1.8098,
      "step": 175727
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.9961436986923218,
      "learning_rate": 7.957636113894381e-05,
      "loss": 1.8735,
      "step": 175728
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.2769660949707031,
      "learning_rate": 7.957358635558346e-05,
      "loss": 1.9533,
      "step": 175729
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.3905038833618164,
      "learning_rate": 7.957081161320447e-05,
      "loss": 2.0417,
      "step": 175730
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.3343144655227661,
      "learning_rate": 7.956803691180748e-05,
      "loss": 1.9085,
      "step": 175731
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.4261256456375122,
      "learning_rate": 7.956526225139295e-05,
      "loss": 2.1634,
      "step": 175732
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.2605801820755005,
      "learning_rate": 7.956248763196142e-05,
      "loss": 2.0034,
      "step": 175733
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.40371572971344,
      "learning_rate": 7.955971305351329e-05,
      "loss": 1.9757,
      "step": 175734
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.388403296470642,
      "learning_rate": 7.955693851604927e-05,
      "loss": 2.1296,
      "step": 175735
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.385861873626709,
      "learning_rate": 7.955416401956967e-05,
      "loss": 1.84,
      "step": 175736
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.270294189453125,
      "learning_rate": 7.955138956407522e-05,
      "loss": 1.9167,
      "step": 175737
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.3085646629333496,
      "learning_rate": 7.95486151495663e-05,
      "loss": 2.2637,
      "step": 175738
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.2336381673812866,
      "learning_rate": 7.954584077604348e-05,
      "loss": 1.9207,
      "step": 175739
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.2717832326889038,
      "learning_rate": 7.954306644350714e-05,
      "loss": 2.0139,
      "step": 175740
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.3731212615966797,
      "learning_rate": 7.954029215195802e-05,
      "loss": 2.1188,
      "step": 175741
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.2207595109939575,
      "learning_rate": 7.95375179013964e-05,
      "loss": 2.2345,
      "step": 175742
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.3242437839508057,
      "learning_rate": 7.953474369182305e-05,
      "loss": 2.0195,
      "step": 175743
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.3219354152679443,
      "learning_rate": 7.953196952323833e-05,
      "loss": 1.9423,
      "step": 175744
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.4465230703353882,
      "learning_rate": 7.952919539564265e-05,
      "loss": 2.0645,
      "step": 175745
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.3358899354934692,
      "learning_rate": 7.95264213090368e-05,
      "loss": 2.0374,
      "step": 175746
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.3561969995498657,
      "learning_rate": 7.952364726342111e-05,
      "loss": 2.0849,
      "step": 175747
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.2748826742172241,
      "learning_rate": 7.952087325879607e-05,
      "loss": 2.0532,
      "step": 175748
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.3665452003479004,
      "learning_rate": 7.951809929516234e-05,
      "loss": 2.0424,
      "step": 175749
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.3002514839172363,
      "learning_rate": 7.951532537252039e-05,
      "loss": 1.9722,
      "step": 175750
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.2330905199050903,
      "learning_rate": 7.951255149087058e-05,
      "loss": 2.0663,
      "step": 175751
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.288954734802246,
      "learning_rate": 7.950977765021365e-05,
      "loss": 1.7506,
      "step": 175752
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.5991233587265015,
      "learning_rate": 7.950700385054999e-05,
      "loss": 1.9394,
      "step": 175753
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.427728533744812,
      "learning_rate": 7.950423009188007e-05,
      "loss": 2.157,
      "step": 175754
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.3307716846466064,
      "learning_rate": 7.950145637420456e-05,
      "loss": 2.4374,
      "step": 175755
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.3383442163467407,
      "learning_rate": 7.94986826975239e-05,
      "loss": 2.1057,
      "step": 175756
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.3552374839782715,
      "learning_rate": 7.94959090618385e-05,
      "loss": 1.818,
      "step": 175757
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.3015767335891724,
      "learning_rate": 7.949313546714909e-05,
      "loss": 1.9318,
      "step": 175758
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.3493582010269165,
      "learning_rate": 7.949036191345594e-05,
      "loss": 1.998,
      "step": 175759
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.2953625917434692,
      "learning_rate": 7.948758840075979e-05,
      "loss": 2.2583,
      "step": 175760
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.2909773588180542,
      "learning_rate": 7.94848149290611e-05,
      "loss": 1.8031,
      "step": 175761
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.408153772354126,
      "learning_rate": 7.94820414983603e-05,
      "loss": 1.9321,
      "step": 175762
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.3421709537506104,
      "learning_rate": 7.947926810865786e-05,
      "loss": 1.9785,
      "step": 175763
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.3891326189041138,
      "learning_rate": 7.947649475995452e-05,
      "loss": 2.1623,
      "step": 175764
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.2155647277832031,
      "learning_rate": 7.947372145225052e-05,
      "loss": 2.0673,
      "step": 175765
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.2812408208847046,
      "learning_rate": 7.947094818554665e-05,
      "loss": 2.0354,
      "step": 175766
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.5073809623718262,
      "learning_rate": 7.94681749598432e-05,
      "loss": 2.2424,
      "step": 175767
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.2541930675506592,
      "learning_rate": 7.946540177514092e-05,
      "loss": 2.2921,
      "step": 175768
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.2356104850769043,
      "learning_rate": 7.946262863144005e-05,
      "loss": 2.0988,
      "step": 175769
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.2455425262451172,
      "learning_rate": 7.945985552874131e-05,
      "loss": 2.0148,
      "step": 175770
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.354848861694336,
      "learning_rate": 7.945708246704504e-05,
      "loss": 1.9088,
      "step": 175771
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.308205246925354,
      "learning_rate": 7.945430944635197e-05,
      "loss": 1.8986,
      "step": 175772
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.3273828029632568,
      "learning_rate": 7.945153646666243e-05,
      "loss": 1.8789,
      "step": 175773
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.2232027053833008,
      "learning_rate": 7.944876352797716e-05,
      "loss": 2.1258,
      "step": 175774
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.3268142938613892,
      "learning_rate": 7.944599063029636e-05,
      "loss": 2.217,
      "step": 175775
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.2621030807495117,
      "learning_rate": 7.944321777362082e-05,
      "loss": 1.8219,
      "step": 175776
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.207442283630371,
      "learning_rate": 7.944044495795085e-05,
      "loss": 2.0255,
      "step": 175777
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.1876863241195679,
      "learning_rate": 7.943767218328715e-05,
      "loss": 1.9612,
      "step": 175778
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.3012460470199585,
      "learning_rate": 7.943489944963008e-05,
      "loss": 2.0407,
      "step": 175779
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.3198118209838867,
      "learning_rate": 7.943212675698037e-05,
      "loss": 1.8848,
      "step": 175780
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.1310549974441528,
      "learning_rate": 7.942935410533825e-05,
      "loss": 1.9124,
      "step": 175781
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.3442573547363281,
      "learning_rate": 7.942658149470443e-05,
      "loss": 1.9475,
      "step": 175782
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.1944631338119507,
      "learning_rate": 7.942380892507931e-05,
      "loss": 2.1277,
      "step": 175783
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.4346450567245483,
      "learning_rate": 7.942103639646352e-05,
      "loss": 1.9641,
      "step": 175784
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.1913670301437378,
      "learning_rate": 7.94182639088575e-05,
      "loss": 1.8305,
      "step": 175785
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.24338960647583,
      "learning_rate": 7.941549146226189e-05,
      "loss": 1.8591,
      "step": 175786
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.263444185256958,
      "learning_rate": 7.941271905667699e-05,
      "loss": 2.0029,
      "step": 175787
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.4281466007232666,
      "learning_rate": 7.940994669210348e-05,
      "loss": 2.1168,
      "step": 175788
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.393025279045105,
      "learning_rate": 7.940717436854176e-05,
      "loss": 2.0094,
      "step": 175789
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.4201070070266724,
      "learning_rate": 7.940440208599247e-05,
      "loss": 2.0993,
      "step": 175790
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.3900090456008911,
      "learning_rate": 7.9401629844456e-05,
      "loss": 2.1237,
      "step": 175791
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.431322455406189,
      "learning_rate": 7.93988576439331e-05,
      "loss": 2.1729,
      "step": 175792
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.286342978477478,
      "learning_rate": 7.939608548442395e-05,
      "loss": 2.18,
      "step": 175793
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.1991387605667114,
      "learning_rate": 7.93933133659293e-05,
      "loss": 1.9814,
      "step": 175794
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.2554630041122437,
      "learning_rate": 7.939054128844951e-05,
      "loss": 2.1163,
      "step": 175795
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.4116617441177368,
      "learning_rate": 7.93877692519853e-05,
      "loss": 2.1213,
      "step": 175796
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.2922425270080566,
      "learning_rate": 7.938499725653695e-05,
      "loss": 1.9195,
      "step": 175797
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.1835182905197144,
      "learning_rate": 7.93822253021053e-05,
      "loss": 1.8851,
      "step": 175798
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.2243399620056152,
      "learning_rate": 7.937945338869046e-05,
      "loss": 2.0856,
      "step": 175799
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.479851484298706,
      "learning_rate": 7.937668151629322e-05,
      "loss": 1.9597,
      "step": 175800
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.340682864189148,
      "learning_rate": 7.937390968491396e-05,
      "loss": 2.0086,
      "step": 175801
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.3189316987991333,
      "learning_rate": 7.937113789455333e-05,
      "loss": 2.2452,
      "step": 175802
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.4699259996414185,
      "learning_rate": 7.936836614521169e-05,
      "loss": 1.913,
      "step": 175803
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.3952430486679077,
      "learning_rate": 7.936559443688979e-05,
      "loss": 2.0267,
      "step": 175804
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.435096025466919,
      "learning_rate": 7.936282276958782e-05,
      "loss": 2.1377,
      "step": 175805
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.441200852394104,
      "learning_rate": 7.936005114330658e-05,
      "loss": 2.1921,
      "step": 175806
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.5846648216247559,
      "learning_rate": 7.935727955804634e-05,
      "loss": 1.9111,
      "step": 175807
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.3292346000671387,
      "learning_rate": 7.935450801380786e-05,
      "loss": 2.0675,
      "step": 175808
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.4293001890182495,
      "learning_rate": 7.935173651059147e-05,
      "loss": 2.0719,
      "step": 175809
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.6016796827316284,
      "learning_rate": 7.934896504839792e-05,
      "loss": 1.8602,
      "step": 175810
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.5466649532318115,
      "learning_rate": 7.934619362722737e-05,
      "loss": 1.757,
      "step": 175811
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.4001089334487915,
      "learning_rate": 7.934342224708063e-05,
      "loss": 2.2296,
      "step": 175812
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.3412903547286987,
      "learning_rate": 7.934065090795805e-05,
      "loss": 2.0135,
      "step": 175813
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.2815661430358887,
      "learning_rate": 7.933787960986024e-05,
      "loss": 2.1735,
      "step": 175814
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.2653687000274658,
      "learning_rate": 7.933510835278765e-05,
      "loss": 1.9856,
      "step": 175815
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.8891834020614624,
      "learning_rate": 7.933233713674099e-05,
      "loss": 1.9088,
      "step": 175816
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.734107255935669,
      "learning_rate": 7.932956596172043e-05,
      "loss": 1.9882,
      "step": 175817
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.4288697242736816,
      "learning_rate": 7.932679482772675e-05,
      "loss": 2.0364,
      "step": 175818
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.304770588874817,
      "learning_rate": 7.93240237347603e-05,
      "loss": 2.1214,
      "step": 175819
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.2932424545288086,
      "learning_rate": 7.932125268282175e-05,
      "loss": 1.9602,
      "step": 175820
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.2066764831542969,
      "learning_rate": 7.931848167191149e-05,
      "loss": 1.8647,
      "step": 175821
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.355063796043396,
      "learning_rate": 7.931571070203018e-05,
      "loss": 1.8428,
      "step": 175822
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.288234829902649,
      "learning_rate": 7.931293977317821e-05,
      "loss": 2.1868,
      "step": 175823
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.3457938432693481,
      "learning_rate": 7.931016888535615e-05,
      "loss": 2.0257,
      "step": 175824
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.203317642211914,
      "learning_rate": 7.930739803856441e-05,
      "loss": 1.6709,
      "step": 175825
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.2943611145019531,
      "learning_rate": 7.930462723280367e-05,
      "loss": 2.0451,
      "step": 175826
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.3298919200897217,
      "learning_rate": 7.930185646807428e-05,
      "loss": 1.8114,
      "step": 175827
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.2922282218933105,
      "learning_rate": 7.929908574437695e-05,
      "loss": 2.0177,
      "step": 175828
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.279550313949585,
      "learning_rate": 7.929631506171208e-05,
      "loss": 2.0329,
      "step": 175829
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.393410325050354,
      "learning_rate": 7.929354442008009e-05,
      "loss": 2.3306,
      "step": 175830
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.276656985282898,
      "learning_rate": 7.929077381948169e-05,
      "loss": 1.9613,
      "step": 175831
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.297210931777954,
      "learning_rate": 7.928800325991731e-05,
      "loss": 1.8945,
      "step": 175832
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.6530710458755493,
      "learning_rate": 7.928523274138734e-05,
      "loss": 2.2042,
      "step": 175833
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.4623888731002808,
      "learning_rate": 7.928246226389253e-05,
      "loss": 2.1389,
      "step": 175834
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.2981852293014526,
      "learning_rate": 7.927969182743326e-05,
      "loss": 1.8875,
      "step": 175835
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.2644085884094238,
      "learning_rate": 7.927692143200995e-05,
      "loss": 1.9865,
      "step": 175836
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.2135311365127563,
      "learning_rate": 7.927415107762337e-05,
      "loss": 1.9697,
      "step": 175837
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.2417809963226318,
      "learning_rate": 7.927138076427384e-05,
      "loss": 1.8761,
      "step": 175838
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.3261948823928833,
      "learning_rate": 7.926861049196186e-05,
      "loss": 2.0473,
      "step": 175839
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.3586336374282837,
      "learning_rate": 7.926584026068812e-05,
      "loss": 2.0487,
      "step": 175840
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.2650128602981567,
      "learning_rate": 7.926307007045301e-05,
      "loss": 1.8749,
      "step": 175841
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.2920448780059814,
      "learning_rate": 7.926029992125695e-05,
      "loss": 2.1748,
      "step": 175842
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.3562637567520142,
      "learning_rate": 7.925752981310069e-05,
      "loss": 1.9553,
      "step": 175843
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.3733080625534058,
      "learning_rate": 7.925475974598455e-05,
      "loss": 1.9964,
      "step": 175844
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.4959628582000732,
      "learning_rate": 7.925198971990915e-05,
      "loss": 2.0131,
      "step": 175845
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.3232755661010742,
      "learning_rate": 7.924921973487502e-05,
      "loss": 2.0067,
      "step": 175846
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.4443703889846802,
      "learning_rate": 7.92464497908826e-05,
      "loss": 1.8804,
      "step": 175847
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.2637571096420288,
      "learning_rate": 7.924367988793233e-05,
      "loss": 2.0968,
      "step": 175848
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.5321530103683472,
      "learning_rate": 7.924091002602496e-05,
      "loss": 2.0081,
      "step": 175849
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.380332589149475,
      "learning_rate": 7.923814020516076e-05,
      "loss": 2.0961,
      "step": 175850
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.3604499101638794,
      "learning_rate": 7.923537042534042e-05,
      "loss": 1.8018,
      "step": 175851
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.3812201023101807,
      "learning_rate": 7.923260068656443e-05,
      "loss": 1.8522,
      "step": 175852
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.4320859909057617,
      "learning_rate": 7.922983098883327e-05,
      "loss": 2.2774,
      "step": 175853
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.308030128479004,
      "learning_rate": 7.922706133214733e-05,
      "loss": 2.0131,
      "step": 175854
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.376554250717163,
      "learning_rate": 7.922429171650734e-05,
      "loss": 2.0151,
      "step": 175855
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.3106926679611206,
      "learning_rate": 7.922152214191364e-05,
      "loss": 2.2922,
      "step": 175856
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.478832483291626,
      "learning_rate": 7.921875260836694e-05,
      "loss": 2.1978,
      "step": 175857
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.2913388013839722,
      "learning_rate": 7.921598311586752e-05,
      "loss": 1.8127,
      "step": 175858
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.3590670824050903,
      "learning_rate": 7.921321366441621e-05,
      "loss": 1.985,
      "step": 175859
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.4137548208236694,
      "learning_rate": 7.921044425401314e-05,
      "loss": 2.0702,
      "step": 175860
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.194373369216919,
      "learning_rate": 7.920767488465913e-05,
      "loss": 2.0818,
      "step": 175861
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.247493863105774,
      "learning_rate": 7.920490555635445e-05,
      "loss": 1.9831,
      "step": 175862
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.2050303220748901,
      "learning_rate": 7.920213626909988e-05,
      "loss": 2.0854,
      "step": 175863
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.440834879875183,
      "learning_rate": 7.91993670228957e-05,
      "loss": 2.1525,
      "step": 175864
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.288028597831726,
      "learning_rate": 7.919659781774272e-05,
      "loss": 1.9666,
      "step": 175865
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.3697032928466797,
      "learning_rate": 7.919382865364102e-05,
      "loss": 2.1399,
      "step": 175866
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.29888916015625,
      "learning_rate": 7.91910595305915e-05,
      "loss": 2.1491,
      "step": 175867
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.3021501302719116,
      "learning_rate": 7.918829044859443e-05,
      "loss": 1.9685,
      "step": 175868
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.4513142108917236,
      "learning_rate": 7.918552140765052e-05,
      "loss": 2.2257,
      "step": 175869
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.2445409297943115,
      "learning_rate": 7.918275240776008e-05,
      "loss": 1.9717,
      "step": 175870
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.2477158308029175,
      "learning_rate": 7.917998344892392e-05,
      "loss": 2.0687,
      "step": 175871
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.313227653503418,
      "learning_rate": 7.917721453114219e-05,
      "loss": 1.9602,
      "step": 175872
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.3462259769439697,
      "learning_rate": 7.91744456544157e-05,
      "loss": 2.242,
      "step": 175873
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.5164517164230347,
      "learning_rate": 7.917167681874472e-05,
      "loss": 2.1184,
      "step": 175874
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.369637370109558,
      "learning_rate": 7.916890802413e-05,
      "loss": 2.0985,
      "step": 175875
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.2979313135147095,
      "learning_rate": 7.916613927057185e-05,
      "loss": 2.1943,
      "step": 175876
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.1876251697540283,
      "learning_rate": 7.916337055807109e-05,
      "loss": 2.1236,
      "step": 175877
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.2560828924179077,
      "learning_rate": 7.916060188662781e-05,
      "loss": 1.8869,
      "step": 175878
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.319685935974121,
      "learning_rate": 7.915783325624286e-05,
      "loss": 2.0538,
      "step": 175879
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.3978241682052612,
      "learning_rate": 7.915506466691655e-05,
      "loss": 1.9901,
      "step": 175880
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.2793102264404297,
      "learning_rate": 7.915229611864954e-05,
      "loss": 2.1529,
      "step": 175881
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.2231696844100952,
      "learning_rate": 7.914952761144223e-05,
      "loss": 1.989,
      "step": 175882
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.3583112955093384,
      "learning_rate": 7.914675914529536e-05,
      "loss": 1.9764,
      "step": 175883
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.2360846996307373,
      "learning_rate": 7.914399072020909e-05,
      "loss": 1.9549,
      "step": 175884
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.2495859861373901,
      "learning_rate": 7.914122233618425e-05,
      "loss": 2.0033,
      "step": 175885
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.2982481718063354,
      "learning_rate": 7.913845399322107e-05,
      "loss": 2.0424,
      "step": 175886
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.210241675376892,
      "learning_rate": 7.913568569132037e-05,
      "loss": 1.9076,
      "step": 175887
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.3561455011367798,
      "learning_rate": 7.91329174304824e-05,
      "loss": 1.9681,
      "step": 175888
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.2207658290863037,
      "learning_rate": 7.913014921070786e-05,
      "loss": 1.8717,
      "step": 175889
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.5971847772598267,
      "learning_rate": 7.912738103199724e-05,
      "loss": 1.9758,
      "step": 175890
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.4325966835021973,
      "learning_rate": 7.912461289435096e-05,
      "loss": 2.1718,
      "step": 175891
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.4733901023864746,
      "learning_rate": 7.912184479776952e-05,
      "loss": 2.0172,
      "step": 175892
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.448198914527893,
      "learning_rate": 7.911907674225358e-05,
      "loss": 2.2698,
      "step": 175893
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.2299606800079346,
      "learning_rate": 7.91163087278035e-05,
      "loss": 1.8752,
      "step": 175894
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.345445990562439,
      "learning_rate": 7.911354075441993e-05,
      "loss": 2.042,
      "step": 175895
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.3044006824493408,
      "learning_rate": 7.911077282210336e-05,
      "loss": 2.0259,
      "step": 175896
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.178482174873352,
      "learning_rate": 7.910800493085426e-05,
      "loss": 1.8685,
      "step": 175897
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.4310015439987183,
      "learning_rate": 7.910523708067303e-05,
      "loss": 2.2132,
      "step": 175898
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.4197936058044434,
      "learning_rate": 7.91024692715604e-05,
      "loss": 2.0603,
      "step": 175899
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.2048267126083374,
      "learning_rate": 7.909970150351674e-05,
      "loss": 2.1037,
      "step": 175900
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.2267004251480103,
      "learning_rate": 7.909693377654266e-05,
      "loss": 2.0764,
      "step": 175901
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.3503344058990479,
      "learning_rate": 7.909416609063865e-05,
      "loss": 2.141,
      "step": 175902
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.4011075496673584,
      "learning_rate": 7.90913984458052e-05,
      "loss": 2.2169,
      "step": 175903
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.171204924583435,
      "learning_rate": 7.908863084204276e-05,
      "loss": 2.0527,
      "step": 175904
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.269618034362793,
      "learning_rate": 7.908586327935198e-05,
      "loss": 2.0413,
      "step": 175905
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.4141976833343506,
      "learning_rate": 7.908309575773323e-05,
      "loss": 1.8802,
      "step": 175906
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.327692985534668,
      "learning_rate": 7.908032827718718e-05,
      "loss": 2.0109,
      "step": 175907
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.1641583442687988,
      "learning_rate": 7.90775608377143e-05,
      "loss": 1.9387,
      "step": 175908
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.3955577611923218,
      "learning_rate": 7.907479343931502e-05,
      "loss": 2.0063,
      "step": 175909
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.2580257654190063,
      "learning_rate": 7.907202608198986e-05,
      "loss": 1.9351,
      "step": 175910
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.573609709739685,
      "learning_rate": 7.906925876573944e-05,
      "loss": 1.8642,
      "step": 175911
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.3270549774169922,
      "learning_rate": 7.906649149056413e-05,
      "loss": 2.1002,
      "step": 175912
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.2292264699935913,
      "learning_rate": 7.906372425646463e-05,
      "loss": 1.8802,
      "step": 175913
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.3702030181884766,
      "learning_rate": 7.906095706344135e-05,
      "loss": 2.0761,
      "step": 175914
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.3539758920669556,
      "learning_rate": 7.905818991149483e-05,
      "loss": 2.1919,
      "step": 175915
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.3320024013519287,
      "learning_rate": 7.905542280062545e-05,
      "loss": 2.1385,
      "step": 175916
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.3610022068023682,
      "learning_rate": 7.905265573083393e-05,
      "loss": 1.9367,
      "step": 175917
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.3215196132659912,
      "learning_rate": 7.90498887021206e-05,
      "loss": 1.7869,
      "step": 175918
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.2689683437347412,
      "learning_rate": 7.904712171448615e-05,
      "loss": 2.1219,
      "step": 175919
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.4300745725631714,
      "learning_rate": 7.904435476793104e-05,
      "loss": 2.1305,
      "step": 175920
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.2569268941879272,
      "learning_rate": 7.904158786245563e-05,
      "loss": 1.8212,
      "step": 175921
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.412222146987915,
      "learning_rate": 7.903882099806067e-05,
      "loss": 1.8716,
      "step": 175922
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.3466256856918335,
      "learning_rate": 7.903605417474654e-05,
      "loss": 2.0774,
      "step": 175923
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.2990082502365112,
      "learning_rate": 7.903328739251372e-05,
      "loss": 2.0085,
      "step": 175924
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.300264835357666,
      "learning_rate": 7.903052065136288e-05,
      "loss": 1.8689,
      "step": 175925
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.4738776683807373,
      "learning_rate": 7.90277539512944e-05,
      "loss": 1.9474,
      "step": 175926
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.4126808643341064,
      "learning_rate": 7.902498729230875e-05,
      "loss": 2.1004,
      "step": 175927
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.356613278388977,
      "learning_rate": 7.902222067440665e-05,
      "loss": 2.0205,
      "step": 175928
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.2916243076324463,
      "learning_rate": 7.901945409758837e-05,
      "loss": 1.9887,
      "step": 175929
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.2555150985717773,
      "learning_rate": 7.901668756185464e-05,
      "loss": 1.9948,
      "step": 175930
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.2984063625335693,
      "learning_rate": 7.901392106720587e-05,
      "loss": 1.8257,
      "step": 175931
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.42489492893219,
      "learning_rate": 7.901115461364261e-05,
      "loss": 2.0691,
      "step": 175932
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.2666866779327393,
      "learning_rate": 7.900838820116525e-05,
      "loss": 1.9068,
      "step": 175933
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.6028140783309937,
      "learning_rate": 7.900562182977446e-05,
      "loss": 2.2779,
      "step": 175934
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.284471035003662,
      "learning_rate": 7.900285549947064e-05,
      "loss": 1.9395,
      "step": 175935
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.6000170707702637,
      "learning_rate": 7.900008921025444e-05,
      "loss": 2.0294,
      "step": 175936
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.6590862274169922,
      "learning_rate": 7.899732296212631e-05,
      "loss": 2.1804,
      "step": 175937
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.343491554260254,
      "learning_rate": 7.899455675508672e-05,
      "loss": 2.2332,
      "step": 175938
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.292049527168274,
      "learning_rate": 7.899179058913615e-05,
      "loss": 2.0516,
      "step": 175939
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.332672119140625,
      "learning_rate": 7.898902446427525e-05,
      "loss": 1.9898,
      "step": 175940
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.2701983451843262,
      "learning_rate": 7.898625838050439e-05,
      "loss": 2.3064,
      "step": 175941
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.4534260034561157,
      "learning_rate": 7.898349233782422e-05,
      "loss": 2.0609,
      "step": 175942
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.4768785238265991,
      "learning_rate": 7.898072633623514e-05,
      "loss": 1.913,
      "step": 175943
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.4849241971969604,
      "learning_rate": 7.897796037573786e-05,
      "loss": 2.044,
      "step": 175944
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.303464651107788,
      "learning_rate": 7.89751944563326e-05,
      "loss": 2.1972,
      "step": 175945
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.2019720077514648,
      "learning_rate": 7.897242857802008e-05,
      "loss": 1.9648,
      "step": 175946
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.1920889616012573,
      "learning_rate": 7.896966274080068e-05,
      "loss": 1.9028,
      "step": 175947
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.5337647199630737,
      "learning_rate": 7.89668969446751e-05,
      "loss": 2.0555,
      "step": 175948
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.3088574409484863,
      "learning_rate": 7.896413118964363e-05,
      "loss": 1.9189,
      "step": 175949
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.4117592573165894,
      "learning_rate": 7.896136547570709e-05,
      "loss": 2.0519,
      "step": 175950
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.3702681064605713,
      "learning_rate": 7.895859980286562e-05,
      "loss": 2.1679,
      "step": 175951
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.2661229372024536,
      "learning_rate": 7.895583417112004e-05,
      "loss": 2.2153,
      "step": 175952
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.2734591960906982,
      "learning_rate": 7.895306858047061e-05,
      "loss": 1.9735,
      "step": 175953
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.2958534955978394,
      "learning_rate": 7.895030303091809e-05,
      "loss": 1.8596,
      "step": 175954
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.3795175552368164,
      "learning_rate": 7.894753752246278e-05,
      "loss": 2.2463,
      "step": 175955
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.3583366870880127,
      "learning_rate": 7.89447720551054e-05,
      "loss": 1.8913,
      "step": 175956
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.4940398931503296,
      "learning_rate": 7.894200662884635e-05,
      "loss": 2.3235,
      "step": 175957
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.2894090414047241,
      "learning_rate": 7.893924124368614e-05,
      "loss": 1.9402,
      "step": 175958
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.3019301891326904,
      "learning_rate": 7.893647589962524e-05,
      "loss": 2.0292,
      "step": 175959
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.246022343635559,
      "learning_rate": 7.893371059666429e-05,
      "loss": 2.1297,
      "step": 175960
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.259872555732727,
      "learning_rate": 7.893094533480366e-05,
      "loss": 2.1254,
      "step": 175961
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.3870128393173218,
      "learning_rate": 7.892818011404403e-05,
      "loss": 2.0753,
      "step": 175962
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.2225446701049805,
      "learning_rate": 7.892541493438581e-05,
      "loss": 1.8613,
      "step": 175963
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.2712100744247437,
      "learning_rate": 7.892264979582952e-05,
      "loss": 1.9724,
      "step": 175964
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.429687261581421,
      "learning_rate": 7.891988469837561e-05,
      "loss": 2.0946,
      "step": 175965
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.3209501504898071,
      "learning_rate": 7.891711964202477e-05,
      "loss": 1.9464,
      "step": 175966
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.4525775909423828,
      "learning_rate": 7.89143546267773e-05,
      "loss": 2.0289,
      "step": 175967
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.2725858688354492,
      "learning_rate": 7.89115896526339e-05,
      "loss": 2.1364,
      "step": 175968
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.3004015684127808,
      "learning_rate": 7.890882471959504e-05,
      "loss": 2.0699,
      "step": 175969
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.2648204565048218,
      "learning_rate": 7.89060598276612e-05,
      "loss": 1.9109,
      "step": 175970
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.2526922225952148,
      "learning_rate": 7.890329497683278e-05,
      "loss": 2.1075,
      "step": 175971
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.5803208351135254,
      "learning_rate": 7.890053016711051e-05,
      "loss": 2.1938,
      "step": 175972
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.4694595336914062,
      "learning_rate": 7.889776539849473e-05,
      "loss": 2.2436,
      "step": 175973
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.3573863506317139,
      "learning_rate": 7.889500067098613e-05,
      "loss": 2.0011,
      "step": 175974
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.303518533706665,
      "learning_rate": 7.889223598458508e-05,
      "loss": 2.0644,
      "step": 175975
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.2907682657241821,
      "learning_rate": 7.888947133929219e-05,
      "loss": 2.082,
      "step": 175976
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.177599549293518,
      "learning_rate": 7.888670673510778e-05,
      "loss": 1.7496,
      "step": 175977
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.2338320016860962,
      "learning_rate": 7.888394217203265e-05,
      "loss": 2.0723,
      "step": 175978
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.2707198858261108,
      "learning_rate": 7.888117765006703e-05,
      "loss": 1.9073,
      "step": 175979
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.2672425508499146,
      "learning_rate": 7.887841316921167e-05,
      "loss": 2.3167,
      "step": 175980
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.2468498945236206,
      "learning_rate": 7.887564872946702e-05,
      "loss": 1.912,
      "step": 175981
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.3307863473892212,
      "learning_rate": 7.887288433083351e-05,
      "loss": 2.1517,
      "step": 175982
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.359075665473938,
      "learning_rate": 7.887011997331165e-05,
      "loss": 2.2375,
      "step": 175983
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.360314130783081,
      "learning_rate": 7.886735565690211e-05,
      "loss": 2.0299,
      "step": 175984
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.2745202779769897,
      "learning_rate": 7.886459138160518e-05,
      "loss": 1.7631,
      "step": 175985
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.1995627880096436,
      "learning_rate": 7.886182714742159e-05,
      "loss": 2.0552,
      "step": 175986
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.2282813787460327,
      "learning_rate": 7.88590629543518e-05,
      "loss": 1.9414,
      "step": 175987
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.4261155128479004,
      "learning_rate": 7.885629880239623e-05,
      "loss": 2.0306,
      "step": 175988
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.3149056434631348,
      "learning_rate": 7.885353469155542e-05,
      "loss": 2.1347,
      "step": 175989
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.2486698627471924,
      "learning_rate": 7.885077062182994e-05,
      "loss": 2.171,
      "step": 175990
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.3067431449890137,
      "learning_rate": 7.884800659322022e-05,
      "loss": 1.8404,
      "step": 175991
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.309133768081665,
      "learning_rate": 7.884524260572691e-05,
      "loss": 1.9377,
      "step": 175992
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.3529458045959473,
      "learning_rate": 7.884247865935047e-05,
      "loss": 1.9426,
      "step": 175993
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.320168137550354,
      "learning_rate": 7.883971475409134e-05,
      "loss": 1.9086,
      "step": 175994
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.5600380897521973,
      "learning_rate": 7.883695088995002e-05,
      "loss": 1.971,
      "step": 175995
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.2738245725631714,
      "learning_rate": 7.883418706692717e-05,
      "loss": 1.8782,
      "step": 175996
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.3251230716705322,
      "learning_rate": 7.883142328502314e-05,
      "loss": 2.1341,
      "step": 175997
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.2675135135650635,
      "learning_rate": 7.882865954423861e-05,
      "loss": 1.9414,
      "step": 175998
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.2623133659362793,
      "learning_rate": 7.882589584457402e-05,
      "loss": 1.9226,
      "step": 175999
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.3171730041503906,
      "learning_rate": 7.882313218602985e-05,
      "loss": 2.2074,
      "step": 176000
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.3641796112060547,
      "learning_rate": 7.882036856860654e-05,
      "loss": 2.1292,
      "step": 176001
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.3551056385040283,
      "learning_rate": 7.88176049923048e-05,
      "loss": 2.2925,
      "step": 176002
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.4857746362686157,
      "learning_rate": 7.881484145712495e-05,
      "loss": 2.1761,
      "step": 176003
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.407861590385437,
      "learning_rate": 7.881207796306769e-05,
      "loss": 1.805,
      "step": 176004
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.2620298862457275,
      "learning_rate": 7.880931451013344e-05,
      "loss": 2.1823,
      "step": 176005
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.41007661819458,
      "learning_rate": 7.880655109832262e-05,
      "loss": 2.1416,
      "step": 176006
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.4556541442871094,
      "learning_rate": 7.880378772763591e-05,
      "loss": 2.0785,
      "step": 176007
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.244530439376831,
      "learning_rate": 7.880102439807378e-05,
      "loss": 2.0606,
      "step": 176008
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.3693909645080566,
      "learning_rate": 7.879826110963661e-05,
      "loss": 2.0187,
      "step": 176009
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.5294908285140991,
      "learning_rate": 7.87954978623251e-05,
      "loss": 1.9858,
      "step": 176010
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.276361107826233,
      "learning_rate": 7.879273465613972e-05,
      "loss": 1.8772,
      "step": 176011
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.219329833984375,
      "learning_rate": 7.878997149108081e-05,
      "loss": 2.0458,
      "step": 176012
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.344429850578308,
      "learning_rate": 7.878720836714915e-05,
      "loss": 1.9817,
      "step": 176013
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.3884835243225098,
      "learning_rate": 7.87844452843451e-05,
      "loss": 2.0755,
      "step": 176014
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.3460620641708374,
      "learning_rate": 7.878168224266911e-05,
      "loss": 2.2664,
      "step": 176015
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.3162603378295898,
      "learning_rate": 7.877891924212189e-05,
      "loss": 1.8448,
      "step": 176016
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.3482836484909058,
      "learning_rate": 7.877615628270383e-05,
      "loss": 1.908,
      "step": 176017
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.209110140800476,
      "learning_rate": 7.877339336441537e-05,
      "loss": 2.1758,
      "step": 176018
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.183713674545288,
      "learning_rate": 7.877063048725718e-05,
      "loss": 2.0042,
      "step": 176019
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.2525259256362915,
      "learning_rate": 7.876786765122966e-05,
      "loss": 1.8383,
      "step": 176020
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.3449461460113525,
      "learning_rate": 7.876510485633342e-05,
      "loss": 2.0514,
      "step": 176021
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.3300809860229492,
      "learning_rate": 7.876234210256893e-05,
      "loss": 2.1371,
      "step": 176022
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.3490509986877441,
      "learning_rate": 7.875957938993663e-05,
      "loss": 2.0125,
      "step": 176023
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.6060614585876465,
      "learning_rate": 7.875681671843716e-05,
      "loss": 2.1819,
      "step": 176024
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.2382453680038452,
      "learning_rate": 7.875405408807099e-05,
      "loss": 1.9047,
      "step": 176025
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.3167471885681152,
      "learning_rate": 7.875129149883851e-05,
      "loss": 1.7559,
      "step": 176026
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.239286184310913,
      "learning_rate": 7.874852895074048e-05,
      "loss": 2.0209,
      "step": 176027
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.4051024913787842,
      "learning_rate": 7.874576644377713e-05,
      "loss": 2.1671,
      "step": 176028
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.2859314680099487,
      "learning_rate": 7.874300397794924e-05,
      "loss": 2.0566,
      "step": 176029
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.988596796989441,
      "learning_rate": 7.874024155325721e-05,
      "loss": 2.0746,
      "step": 176030
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.337470293045044,
      "learning_rate": 7.873747916970152e-05,
      "loss": 1.8129,
      "step": 176031
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.190476894378662,
      "learning_rate": 7.873471682728261e-05,
      "loss": 1.8621,
      "step": 176032
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.2906235456466675,
      "learning_rate": 7.873195452600122e-05,
      "loss": 1.9108,
      "step": 176033
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.2126530408859253,
      "learning_rate": 7.872919226585761e-05,
      "loss": 1.9998,
      "step": 176034
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.3254603147506714,
      "learning_rate": 7.872643004685254e-05,
      "loss": 2.1044,
      "step": 176035
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.3917491436004639,
      "learning_rate": 7.872366786898639e-05,
      "loss": 2.0089,
      "step": 176036
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.345044493675232,
      "learning_rate": 7.872090573225968e-05,
      "loss": 2.0954,
      "step": 176037
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.2994108200073242,
      "learning_rate": 7.871814363667286e-05,
      "loss": 2.1113,
      "step": 176038
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.4672629833221436,
      "learning_rate": 7.87153815822266e-05,
      "loss": 1.934,
      "step": 176039
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.2567055225372314,
      "learning_rate": 7.871261956892123e-05,
      "loss": 2.2216,
      "step": 176040
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.392671823501587,
      "learning_rate": 7.870985759675745e-05,
      "loss": 1.9733,
      "step": 176041
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.3366262912750244,
      "learning_rate": 7.870709566573566e-05,
      "loss": 1.9123,
      "step": 176042
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.2438020706176758,
      "learning_rate": 7.870433377585644e-05,
      "loss": 1.6591,
      "step": 176043
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.2236615419387817,
      "learning_rate": 7.870157192712016e-05,
      "loss": 1.8929,
      "step": 176044
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.2505295276641846,
      "learning_rate": 7.869881011952753e-05,
      "loss": 2.1365,
      "step": 176045
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.3124672174453735,
      "learning_rate": 7.869604835307887e-05,
      "loss": 1.8847,
      "step": 176046
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.4100691080093384,
      "learning_rate": 7.869328662777489e-05,
      "loss": 1.9152,
      "step": 176047
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.2492969036102295,
      "learning_rate": 7.869052494361597e-05,
      "loss": 1.9734,
      "step": 176048
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.3143850564956665,
      "learning_rate": 7.86877633006027e-05,
      "loss": 2.115,
      "step": 176049
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.3625563383102417,
      "learning_rate": 7.868500169873544e-05,
      "loss": 2.0002,
      "step": 176050
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.3579249382019043,
      "learning_rate": 7.86822401380149e-05,
      "loss": 2.0389,
      "step": 176051
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.3473390340805054,
      "learning_rate": 7.867947861844147e-05,
      "loss": 2.0118,
      "step": 176052
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.3335388898849487,
      "learning_rate": 7.867671714001574e-05,
      "loss": 2.0093,
      "step": 176053
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.503899097442627,
      "learning_rate": 7.867395570273819e-05,
      "loss": 1.9463,
      "step": 176054
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.3525910377502441,
      "learning_rate": 7.867119430660937e-05,
      "loss": 1.9869,
      "step": 176055
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.5171185731887817,
      "learning_rate": 7.866843295162963e-05,
      "loss": 2.0738,
      "step": 176056
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.3095698356628418,
      "learning_rate": 7.86656716377997e-05,
      "loss": 1.8072,
      "step": 176057
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.5293992757797241,
      "learning_rate": 7.866291036511989e-05,
      "loss": 2.1211,
      "step": 176058
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.3979051113128662,
      "learning_rate": 7.866014913359093e-05,
      "loss": 1.8278,
      "step": 176059
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.3436193466186523,
      "learning_rate": 7.865738794321326e-05,
      "loss": 2.0779,
      "step": 176060
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.2758172750473022,
      "learning_rate": 7.86546267939873e-05,
      "loss": 2.131,
      "step": 176061
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.3405364751815796,
      "learning_rate": 7.865186568591358e-05,
      "loss": 2.1012,
      "step": 176062
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.3056343793869019,
      "learning_rate": 7.864910461899272e-05,
      "loss": 1.9272,
      "step": 176063
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.2820110321044922,
      "learning_rate": 7.86463435932251e-05,
      "loss": 2.1581,
      "step": 176064
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.4226598739624023,
      "learning_rate": 7.864358260861136e-05,
      "loss": 2.0465,
      "step": 176065
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.2421958446502686,
      "learning_rate": 7.864082166515201e-05,
      "loss": 2.1493,
      "step": 176066
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.3716436624526978,
      "learning_rate": 7.863806076284746e-05,
      "loss": 1.8314,
      "step": 176067
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.4370677471160889,
      "learning_rate": 7.86352999016982e-05,
      "loss": 2.0704,
      "step": 176068
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.3672308921813965,
      "learning_rate": 7.863253908170491e-05,
      "loss": 1.8976,
      "step": 176069
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.6820451021194458,
      "learning_rate": 7.862977830286791e-05,
      "loss": 2.2907,
      "step": 176070
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.2511184215545654,
      "learning_rate": 7.862701756518792e-05,
      "loss": 2.0156,
      "step": 176071
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.2803525924682617,
      "learning_rate": 7.862425686866532e-05,
      "loss": 2.0346,
      "step": 176072
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.2311615943908691,
      "learning_rate": 7.862149621330068e-05,
      "loss": 2.2116,
      "step": 176073
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.6437668800354004,
      "learning_rate": 7.861873559909438e-05,
      "loss": 1.9232,
      "step": 176074
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.2886332273483276,
      "learning_rate": 7.861597502604713e-05,
      "loss": 1.9006,
      "step": 176075
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.4072502851486206,
      "learning_rate": 7.861321449415925e-05,
      "loss": 2.0292,
      "step": 176076
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.4448142051696777,
      "learning_rate": 7.861045400343144e-05,
      "loss": 2.0285,
      "step": 176077
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.2747071981430054,
      "learning_rate": 7.860769355386415e-05,
      "loss": 2.025,
      "step": 176078
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.3922553062438965,
      "learning_rate": 7.860493314545783e-05,
      "loss": 2.1515,
      "step": 176079
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.2480565309524536,
      "learning_rate": 7.860217277821298e-05,
      "loss": 1.8637,
      "step": 176080
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.1729967594146729,
      "learning_rate": 7.859941245213024e-05,
      "loss": 1.9255,
      "step": 176081
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.4438774585723877,
      "learning_rate": 7.859665216720994e-05,
      "loss": 1.9435,
      "step": 176082
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.1926673650741577,
      "learning_rate": 7.859389192345282e-05,
      "loss": 2.0519,
      "step": 176083
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.440437912940979,
      "learning_rate": 7.859113172085926e-05,
      "loss": 1.8571,
      "step": 176084
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.2606219053268433,
      "learning_rate": 7.858837155942978e-05,
      "loss": 2.0421,
      "step": 176085
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.1309970617294312,
      "learning_rate": 7.858561143916484e-05,
      "loss": 1.8904,
      "step": 176086
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.3277008533477783,
      "learning_rate": 7.858285136006509e-05,
      "loss": 1.7855,
      "step": 176087
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.2996602058410645,
      "learning_rate": 7.858009132213086e-05,
      "loss": 2.0583,
      "step": 176088
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.3877358436584473,
      "learning_rate": 7.857733132536289e-05,
      "loss": 1.9667,
      "step": 176089
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.2787374258041382,
      "learning_rate": 7.857457136976146e-05,
      "loss": 1.9718,
      "step": 176090
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.41571843624115,
      "learning_rate": 7.85718114553273e-05,
      "loss": 2.2271,
      "step": 176091
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.2641327381134033,
      "learning_rate": 7.856905158206082e-05,
      "loss": 2.0692,
      "step": 176092
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.3536103963851929,
      "learning_rate": 7.856629174996253e-05,
      "loss": 1.9685,
      "step": 176093
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.232001543045044,
      "learning_rate": 7.856353195903284e-05,
      "loss": 2.0311,
      "step": 176094
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.501310110092163,
      "learning_rate": 7.856077220927246e-05,
      "loss": 2.2501,
      "step": 176095
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.4144971370697021,
      "learning_rate": 7.855801250068174e-05,
      "loss": 1.9221,
      "step": 176096
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.3636025190353394,
      "learning_rate": 7.855525283326136e-05,
      "loss": 2.1298,
      "step": 176097
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.2187659740447998,
      "learning_rate": 7.855249320701175e-05,
      "loss": 2.1171,
      "step": 176098
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.1898165941238403,
      "learning_rate": 7.854973362193337e-05,
      "loss": 2.0716,
      "step": 176099
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.2499850988388062,
      "learning_rate": 7.854697407802671e-05,
      "loss": 2.1615,
      "step": 176100
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.3958863019943237,
      "learning_rate": 7.854421457529243e-05,
      "loss": 2.0424,
      "step": 176101
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.4248539209365845,
      "learning_rate": 7.854145511373087e-05,
      "loss": 1.9575,
      "step": 176102
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.3276753425598145,
      "learning_rate": 7.853869569334276e-05,
      "loss": 2.127,
      "step": 176103
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.262473702430725,
      "learning_rate": 7.853593631412844e-05,
      "loss": 1.9864,
      "step": 176104
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.3300660848617554,
      "learning_rate": 7.853317697608841e-05,
      "loss": 2.0011,
      "step": 176105
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.140324354171753,
      "learning_rate": 7.853041767922332e-05,
      "loss": 2.168,
      "step": 176106
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.4155914783477783,
      "learning_rate": 7.852765842353358e-05,
      "loss": 2.1925,
      "step": 176107
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.2168691158294678,
      "learning_rate": 7.852489920901967e-05,
      "loss": 1.9607,
      "step": 176108
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.2676687240600586,
      "learning_rate": 7.852214003568227e-05,
      "loss": 1.892,
      "step": 176109
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.2328429222106934,
      "learning_rate": 7.851938090352176e-05,
      "loss": 2.1025,
      "step": 176110
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.3102682828903198,
      "learning_rate": 7.851662181253857e-05,
      "loss": 2.0023,
      "step": 176111
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.315534234046936,
      "learning_rate": 7.851386276273345e-05,
      "loss": 2.0105,
      "step": 176112
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.3389836549758911,
      "learning_rate": 7.851110375410665e-05,
      "loss": 2.0256,
      "step": 176113
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.6234228610992432,
      "learning_rate": 7.850834478665895e-05,
      "loss": 1.8557,
      "step": 176114
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.3076993227005005,
      "learning_rate": 7.850558586039071e-05,
      "loss": 1.9024,
      "step": 176115
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.196824312210083,
      "learning_rate": 7.850282697530246e-05,
      "loss": 2.0016,
      "step": 176116
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.3327475786209106,
      "learning_rate": 7.85000681313946e-05,
      "loss": 2.0093,
      "step": 176117
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.3247302770614624,
      "learning_rate": 7.84973093286679e-05,
      "loss": 2.2015,
      "step": 176118
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.5403896570205688,
      "learning_rate": 7.849455056712262e-05,
      "loss": 2.0857,
      "step": 176119
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.2810065746307373,
      "learning_rate": 7.849179184675947e-05,
      "loss": 2.0787,
      "step": 176120
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.2394216060638428,
      "learning_rate": 7.848903316757889e-05,
      "loss": 1.926,
      "step": 176121
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.322005033493042,
      "learning_rate": 7.848627452958135e-05,
      "loss": 1.9446,
      "step": 176122
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.4180113077163696,
      "learning_rate": 7.848351593276733e-05,
      "loss": 2.1395,
      "step": 176123
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.5113540887832642,
      "learning_rate": 7.848075737713748e-05,
      "loss": 2.0609,
      "step": 176124
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.2650266885757446,
      "learning_rate": 7.847799886269214e-05,
      "loss": 1.9217,
      "step": 176125
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.2342510223388672,
      "learning_rate": 7.847524038943203e-05,
      "loss": 2.0142,
      "step": 176126
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.3805644512176514,
      "learning_rate": 7.847248195735759e-05,
      "loss": 2.1228,
      "step": 176127
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.2048298120498657,
      "learning_rate": 7.846972356646924e-05,
      "loss": 2.1637,
      "step": 176128
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.2293519973754883,
      "learning_rate": 7.846696521676749e-05,
      "loss": 2.1307,
      "step": 176129
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.388690710067749,
      "learning_rate": 7.846420690825297e-05,
      "loss": 2.0735,
      "step": 176130
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.4693511724472046,
      "learning_rate": 7.846144864092609e-05,
      "loss": 1.9435,
      "step": 176131
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.3088371753692627,
      "learning_rate": 7.845869041478747e-05,
      "loss": 2.0199,
      "step": 176132
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.1823018789291382,
      "learning_rate": 7.845593222983758e-05,
      "loss": 1.8528,
      "step": 176133
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.3211132287979126,
      "learning_rate": 7.845317408607688e-05,
      "loss": 2.1247,
      "step": 176134
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.5801814794540405,
      "learning_rate": 7.845041598350586e-05,
      "loss": 1.9459,
      "step": 176135
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.3010170459747314,
      "learning_rate": 7.844765792212518e-05,
      "loss": 2.3666,
      "step": 176136
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.3446624279022217,
      "learning_rate": 7.844489990193514e-05,
      "loss": 1.9309,
      "step": 176137
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.2643746137619019,
      "learning_rate": 7.84421419229365e-05,
      "loss": 2.0286,
      "step": 176138
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.2728302478790283,
      "learning_rate": 7.843938398512963e-05,
      "loss": 2.0539,
      "step": 176139
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.2282096147537231,
      "learning_rate": 7.843662608851509e-05,
      "loss": 1.895,
      "step": 176140
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.2923978567123413,
      "learning_rate": 7.843386823309322e-05,
      "loss": 2.1029,
      "step": 176141
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.2378417253494263,
      "learning_rate": 7.84311104188648e-05,
      "loss": 2.155,
      "step": 176142
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.1978790760040283,
      "learning_rate": 7.842835264583014e-05,
      "loss": 2.0293,
      "step": 176143
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.319250464439392,
      "learning_rate": 7.842559491398989e-05,
      "loss": 1.9043,
      "step": 176144
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.324453353881836,
      "learning_rate": 7.842283722334454e-05,
      "loss": 1.8989,
      "step": 176145
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.303796648979187,
      "learning_rate": 7.842007957389455e-05,
      "loss": 1.8407,
      "step": 176146
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.448786973953247,
      "learning_rate": 7.841732196564035e-05,
      "loss": 2.1875,
      "step": 176147
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.3423001766204834,
      "learning_rate": 7.841456439858265e-05,
      "loss": 1.8959,
      "step": 176148
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.4178463220596313,
      "learning_rate": 7.841180687272176e-05,
      "loss": 2.1653,
      "step": 176149
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.244827151298523,
      "learning_rate": 7.840904938805842e-05,
      "loss": 1.7603,
      "step": 176150
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.226468801498413,
      "learning_rate": 7.840629194459299e-05,
      "loss": 2.1276,
      "step": 176151
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.3062794208526611,
      "learning_rate": 7.840353454232604e-05,
      "loss": 1.815,
      "step": 176152
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.3176525831222534,
      "learning_rate": 7.840077718125797e-05,
      "loss": 2.1672,
      "step": 176153
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.2754179239273071,
      "learning_rate": 7.839801986138943e-05,
      "loss": 2.0523,
      "step": 176154
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.329613208770752,
      "learning_rate": 7.839526258272082e-05,
      "loss": 2.0239,
      "step": 176155
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.2074809074401855,
      "learning_rate": 7.83925053452528e-05,
      "loss": 2.2946,
      "step": 176156
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.4220554828643799,
      "learning_rate": 7.838974814898571e-05,
      "loss": 2.2162,
      "step": 176157
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.1891533136367798,
      "learning_rate": 7.838699099392031e-05,
      "loss": 1.9946,
      "step": 176158
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.277578353881836,
      "learning_rate": 7.83842338800568e-05,
      "loss": 2.0545,
      "step": 176159
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.2927262783050537,
      "learning_rate": 7.838147680739589e-05,
      "loss": 2.0847,
      "step": 176160
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.337100863456726,
      "learning_rate": 7.837871977593799e-05,
      "loss": 1.9792,
      "step": 176161
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.2707772254943848,
      "learning_rate": 7.837596278568376e-05,
      "loss": 2.0345,
      "step": 176162
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.420750379562378,
      "learning_rate": 7.837320583663355e-05,
      "loss": 1.8859,
      "step": 176163
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.3065845966339111,
      "learning_rate": 7.83704489287881e-05,
      "loss": 2.0061,
      "step": 176164
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.2534974813461304,
      "learning_rate": 7.83676920621476e-05,
      "loss": 1.9099,
      "step": 176165
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.4094291925430298,
      "learning_rate": 7.836493523671279e-05,
      "loss": 1.8175,
      "step": 176166
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.3605643510818481,
      "learning_rate": 7.836217845248407e-05,
      "loss": 1.9582,
      "step": 176167
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.3455266952514648,
      "learning_rate": 7.835942170946203e-05,
      "loss": 1.8932,
      "step": 176168
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.3631538152694702,
      "learning_rate": 7.835666500764714e-05,
      "loss": 2.0911,
      "step": 176169
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.2678626775741577,
      "learning_rate": 7.835390834704007e-05,
      "loss": 1.959,
      "step": 176170
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.672383189201355,
      "learning_rate": 7.835115172764101e-05,
      "loss": 2.0551,
      "step": 176171
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.4736396074295044,
      "learning_rate": 7.834839514945077e-05,
      "loss": 2.0616,
      "step": 176172
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.3040955066680908,
      "learning_rate": 7.834563861246966e-05,
      "loss": 2.0534,
      "step": 176173
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.290328025817871,
      "learning_rate": 7.834288211669836e-05,
      "loss": 1.7907,
      "step": 176174
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.2910887002944946,
      "learning_rate": 7.834012566213719e-05,
      "loss": 1.8982,
      "step": 176175
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.4833253622055054,
      "learning_rate": 7.833736924878688e-05,
      "loss": 1.9,
      "step": 176176
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.2888374328613281,
      "learning_rate": 7.833461287664785e-05,
      "loss": 2.0531,
      "step": 176177
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.4062793254852295,
      "learning_rate": 7.833185654572059e-05,
      "loss": 1.8841,
      "step": 176178
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.2952971458435059,
      "learning_rate": 7.832910025600555e-05,
      "loss": 1.9084,
      "step": 176179
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.5147027969360352,
      "learning_rate": 7.832634400750339e-05,
      "loss": 2.392,
      "step": 176180
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.2873656749725342,
      "learning_rate": 7.832358780021444e-05,
      "loss": 1.9196,
      "step": 176181
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.198959469795227,
      "learning_rate": 7.832083163413941e-05,
      "loss": 2.0489,
      "step": 176182
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.24153733253479,
      "learning_rate": 7.831807550927874e-05,
      "loss": 2.0604,
      "step": 176183
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.3291630744934082,
      "learning_rate": 7.831531942563294e-05,
      "loss": 2.0887,
      "step": 176184
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.3069819211959839,
      "learning_rate": 7.83125633832024e-05,
      "loss": 2.0523,
      "step": 176185
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.3181378841400146,
      "learning_rate": 7.830980738198782e-05,
      "loss": 2.0574,
      "step": 176186
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.3529912233352661,
      "learning_rate": 7.830705142198954e-05,
      "loss": 1.9668,
      "step": 176187
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.337075114250183,
      "learning_rate": 7.830429550320826e-05,
      "loss": 2.1718,
      "step": 176188
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.2498128414154053,
      "learning_rate": 7.83015396256444e-05,
      "loss": 2.2382,
      "step": 176189
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.2630808353424072,
      "learning_rate": 7.829878378929839e-05,
      "loss": 1.9686,
      "step": 176190
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.340145230293274,
      "learning_rate": 7.829602799417089e-05,
      "loss": 2.03,
      "step": 176191
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.4950670003890991,
      "learning_rate": 7.829327224026235e-05,
      "loss": 2.237,
      "step": 176192
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.4347261190414429,
      "learning_rate": 7.829051652757318e-05,
      "loss": 2.0207,
      "step": 176193
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.4792464971542358,
      "learning_rate": 7.82877608561041e-05,
      "loss": 2.1362,
      "step": 176194
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.1995521783828735,
      "learning_rate": 7.828500522585547e-05,
      "loss": 1.9729,
      "step": 176195
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.3117852210998535,
      "learning_rate": 7.828224963682779e-05,
      "loss": 2.1161,
      "step": 176196
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.49311101436615,
      "learning_rate": 7.827949408902172e-05,
      "loss": 2.1911,
      "step": 176197
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.251582145690918,
      "learning_rate": 7.827673858243765e-05,
      "loss": 1.9129,
      "step": 176198
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.6044737100601196,
      "learning_rate": 7.827398311707603e-05,
      "loss": 2.096,
      "step": 176199
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.6421467065811157,
      "learning_rate": 7.827122769293758e-05,
      "loss": 1.9413,
      "step": 176200
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.4191248416900635,
      "learning_rate": 7.826847231002268e-05,
      "loss": 1.9745,
      "step": 176201
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.4061474800109863,
      "learning_rate": 7.826571696833173e-05,
      "loss": 2.0119,
      "step": 176202
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.2722820043563843,
      "learning_rate": 7.826296166786552e-05,
      "loss": 2.0319,
      "step": 176203
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.358203411102295,
      "learning_rate": 7.82602064086243e-05,
      "loss": 2.0331,
      "step": 176204
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.4153263568878174,
      "learning_rate": 7.825745119060875e-05,
      "loss": 2.1943,
      "step": 176205
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.2802655696868896,
      "learning_rate": 7.825469601381937e-05,
      "loss": 2.022,
      "step": 176206
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.187240481376648,
      "learning_rate": 7.825194087825658e-05,
      "loss": 1.9229,
      "step": 176207
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.49448561668396,
      "learning_rate": 7.824918578392091e-05,
      "loss": 2.0375,
      "step": 176208
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.422149419784546,
      "learning_rate": 7.824643073081296e-05,
      "loss": 2.1053,
      "step": 176209
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.3732627630233765,
      "learning_rate": 7.82436757189331e-05,
      "loss": 1.8612,
      "step": 176210
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.2230050563812256,
      "learning_rate": 7.824092074828199e-05,
      "loss": 2.1114,
      "step": 176211
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.3518147468566895,
      "learning_rate": 7.82381658188601e-05,
      "loss": 2.3789,
      "step": 176212
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.5433284044265747,
      "learning_rate": 7.823541093066792e-05,
      "loss": 2.0068,
      "step": 176213
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.3937592506408691,
      "learning_rate": 7.823265608370588e-05,
      "loss": 2.1025,
      "step": 176214
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.3608487844467163,
      "learning_rate": 7.822990127797466e-05,
      "loss": 2.2183,
      "step": 176215
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.4267618656158447,
      "learning_rate": 7.82271465134746e-05,
      "loss": 2.0225,
      "step": 176216
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.3630374670028687,
      "learning_rate": 7.822439179020641e-05,
      "loss": 2.2044,
      "step": 176217
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.2793899774551392,
      "learning_rate": 7.822163710817047e-05,
      "loss": 2.0872,
      "step": 176218
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.3773576021194458,
      "learning_rate": 7.82188824673673e-05,
      "loss": 2.0511,
      "step": 176219
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.297942042350769,
      "learning_rate": 7.821612786779736e-05,
      "loss": 1.9065,
      "step": 176220
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.3991550207138062,
      "learning_rate": 7.82133733094613e-05,
      "loss": 1.942,
      "step": 176221
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.4288324117660522,
      "learning_rate": 7.821061879235949e-05,
      "loss": 1.7862,
      "step": 176222
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.546355962753296,
      "learning_rate": 7.820786431649257e-05,
      "loss": 1.9415,
      "step": 176223
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.2732927799224854,
      "learning_rate": 7.820510988186095e-05,
      "loss": 1.8949,
      "step": 176224
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.410657286643982,
      "learning_rate": 7.820235548846532e-05,
      "loss": 1.9531,
      "step": 176225
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.2752612829208374,
      "learning_rate": 7.81996011363059e-05,
      "loss": 2.0678,
      "step": 176226
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.2722841501235962,
      "learning_rate": 7.819684682538347e-05,
      "loss": 1.8555,
      "step": 176227
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.653820514678955,
      "learning_rate": 7.819409255569831e-05,
      "loss": 2.0289,
      "step": 176228
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.322000503540039,
      "learning_rate": 7.819133832725117e-05,
      "loss": 2.0076,
      "step": 176229
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.2997397184371948,
      "learning_rate": 7.818858414004233e-05,
      "loss": 2.019,
      "step": 176230
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.3544368743896484,
      "learning_rate": 7.818582999407264e-05,
      "loss": 2.0399,
      "step": 176231
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.345181941986084,
      "learning_rate": 7.818307588934217e-05,
      "loss": 2.079,
      "step": 176232
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.2526142597198486,
      "learning_rate": 7.818032182585178e-05,
      "loss": 1.956,
      "step": 176233
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.2709317207336426,
      "learning_rate": 7.817756780360176e-05,
      "loss": 1.9063,
      "step": 176234
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.4778715372085571,
      "learning_rate": 7.817481382259279e-05,
      "loss": 2.0681,
      "step": 176235
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.4141380786895752,
      "learning_rate": 7.817205988282522e-05,
      "loss": 1.9931,
      "step": 176236
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.2230304479599,
      "learning_rate": 7.816930598429983e-05,
      "loss": 2.1427,
      "step": 176237
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.2297509908676147,
      "learning_rate": 7.816655212701679e-05,
      "loss": 1.9517,
      "step": 176238
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.2761465311050415,
      "learning_rate": 7.816379831097686e-05,
      "loss": 1.9037,
      "step": 176239
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.4422649145126343,
      "learning_rate": 7.816104453618037e-05,
      "loss": 1.8882,
      "step": 176240
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.2219218015670776,
      "learning_rate": 7.815829080262801e-05,
      "loss": 1.9248,
      "step": 176241
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.331538200378418,
      "learning_rate": 7.815553711032013e-05,
      "loss": 2.1009,
      "step": 176242
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.245126485824585,
      "learning_rate": 7.815278345925749e-05,
      "loss": 1.7751,
      "step": 176243
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.4253686666488647,
      "learning_rate": 7.81500298494403e-05,
      "loss": 2.0859,
      "step": 176244
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.2361091375350952,
      "learning_rate": 7.814727628086924e-05,
      "loss": 2.1874,
      "step": 176245
    },
    {
      "epoch": 2.29,
      "grad_norm": 2.020247220993042,
      "learning_rate": 7.814452275354472e-05,
      "loss": 2.1988,
      "step": 176246
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.2549262046813965,
      "learning_rate": 7.814176926746744e-05,
      "loss": 2.1521,
      "step": 176247
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.288195013999939,
      "learning_rate": 7.813901582263768e-05,
      "loss": 2.148,
      "step": 176248
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.2762213945388794,
      "learning_rate": 7.813626241905624e-05,
      "loss": 2.2799,
      "step": 176249
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.4207206964492798,
      "learning_rate": 7.813350905672328e-05,
      "loss": 2.0941,
      "step": 176250
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.374656081199646,
      "learning_rate": 7.813075573563958e-05,
      "loss": 2.2132,
      "step": 176251
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.3605979681015015,
      "learning_rate": 7.812800245580549e-05,
      "loss": 2.1059,
      "step": 176252
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.395342469215393,
      "learning_rate": 7.812524921722168e-05,
      "loss": 1.9571,
      "step": 176253
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.5973913669586182,
      "learning_rate": 7.812249601988846e-05,
      "loss": 2.0001,
      "step": 176254
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.2641867399215698,
      "learning_rate": 7.811974286380664e-05,
      "loss": 2.005,
      "step": 176255
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.4191358089447021,
      "learning_rate": 7.811698974897636e-05,
      "loss": 1.8054,
      "step": 176256
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.3522472381591797,
      "learning_rate": 7.811423667539843e-05,
      "loss": 2.156,
      "step": 176257
    },
    {
      "epoch": 2.29,
      "grad_norm": 1.448311448097229,
      "learning_rate": 7.811148364307315e-05,
      "loss": 1.9359,
      "step": 176258
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.4624112844467163,
      "learning_rate": 7.810873065200124e-05,
      "loss": 1.986,
      "step": 176259
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.419737458229065,
      "learning_rate": 7.810597770218299e-05,
      "loss": 2.1959,
      "step": 176260
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.3613629341125488,
      "learning_rate": 7.810322479361922e-05,
      "loss": 2.0066,
      "step": 176261
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.218418002128601,
      "learning_rate": 7.810047192631008e-05,
      "loss": 2.1723,
      "step": 176262
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.9790743589401245,
      "learning_rate": 7.809771910025635e-05,
      "loss": 1.8453,
      "step": 176263
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.4286603927612305,
      "learning_rate": 7.809496631545833e-05,
      "loss": 1.9057,
      "step": 176264
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.3849940299987793,
      "learning_rate": 7.809221357191671e-05,
      "loss": 2.0213,
      "step": 176265
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.3034725189208984,
      "learning_rate": 7.80894608696319e-05,
      "loss": 1.9229,
      "step": 176266
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.578323483467102,
      "learning_rate": 7.808670820860452e-05,
      "loss": 2.06,
      "step": 176267
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.3615802526474,
      "learning_rate": 7.808395558883501e-05,
      "loss": 2.0217,
      "step": 176268
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.2846804857254028,
      "learning_rate": 7.808120301032384e-05,
      "loss": 2.014,
      "step": 176269
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.60254967212677,
      "learning_rate": 7.807845047307152e-05,
      "loss": 2.0625,
      "step": 176270
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.341464877128601,
      "learning_rate": 7.807569797707868e-05,
      "loss": 1.8888,
      "step": 176271
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.3806248903274536,
      "learning_rate": 7.807294552234569e-05,
      "loss": 1.9666,
      "step": 176272
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.5969487428665161,
      "learning_rate": 7.807019310887317e-05,
      "loss": 2.0201,
      "step": 176273
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.3252371549606323,
      "learning_rate": 7.806744073666164e-05,
      "loss": 1.8958,
      "step": 176274
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.3158892393112183,
      "learning_rate": 7.806468840571145e-05,
      "loss": 1.8478,
      "step": 176275
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.2739347219467163,
      "learning_rate": 7.806193611602333e-05,
      "loss": 2.1523,
      "step": 176276
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.395901083946228,
      "learning_rate": 7.805918386759767e-05,
      "loss": 1.7859,
      "step": 176277
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.3936123847961426,
      "learning_rate": 7.80564316604349e-05,
      "loss": 1.9019,
      "step": 176278
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.3496248722076416,
      "learning_rate": 7.805367949453572e-05,
      "loss": 1.9698,
      "step": 176279
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.2493146657943726,
      "learning_rate": 7.805092736990056e-05,
      "loss": 2.0645,
      "step": 176280
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.3270134925842285,
      "learning_rate": 7.804817528652983e-05,
      "loss": 2.0181,
      "step": 176281
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.3627264499664307,
      "learning_rate": 7.804542324442424e-05,
      "loss": 1.9923,
      "step": 176282
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.3852359056472778,
      "learning_rate": 7.804267124358415e-05,
      "loss": 2.0404,
      "step": 176283
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.3511327505111694,
      "learning_rate": 7.803991928401007e-05,
      "loss": 1.9206,
      "step": 176284
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.4069761037826538,
      "learning_rate": 7.803716736570263e-05,
      "loss": 1.9216,
      "step": 176285
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.3136175870895386,
      "learning_rate": 7.803441548866229e-05,
      "loss": 2.3293,
      "step": 176286
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.3643995523452759,
      "learning_rate": 7.803166365288944e-05,
      "loss": 1.9034,
      "step": 176287
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.3455662727355957,
      "learning_rate": 7.802891185838479e-05,
      "loss": 2.1531,
      "step": 176288
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.3943557739257812,
      "learning_rate": 7.802616010514864e-05,
      "loss": 1.9144,
      "step": 176289
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.3804901838302612,
      "learning_rate": 7.802340839318172e-05,
      "loss": 2.2918,
      "step": 176290
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.259240984916687,
      "learning_rate": 7.802065672248446e-05,
      "loss": 2.2159,
      "step": 176291
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.22969651222229,
      "learning_rate": 7.801790509305731e-05,
      "loss": 1.8979,
      "step": 176292
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.273758888244629,
      "learning_rate": 7.801515350490076e-05,
      "loss": 1.7947,
      "step": 176293
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.299429178237915,
      "learning_rate": 7.801240195801548e-05,
      "loss": 1.9541,
      "step": 176294
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.3522568941116333,
      "learning_rate": 7.800965045240176e-05,
      "loss": 2.1695,
      "step": 176295
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.3281453847885132,
      "learning_rate": 7.800689898806036e-05,
      "loss": 1.9959,
      "step": 176296
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.3215421438217163,
      "learning_rate": 7.800414756499165e-05,
      "loss": 1.8379,
      "step": 176297
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.4732534885406494,
      "learning_rate": 7.800139618319618e-05,
      "loss": 1.9541,
      "step": 176298
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.3346775770187378,
      "learning_rate": 7.799864484267434e-05,
      "loss": 1.8841,
      "step": 176299
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.2828245162963867,
      "learning_rate": 7.799589354342682e-05,
      "loss": 2.1375,
      "step": 176300
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.1934828758239746,
      "learning_rate": 7.799314228545396e-05,
      "loss": 2.1014,
      "step": 176301
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.344281554222107,
      "learning_rate": 7.799039106875647e-05,
      "loss": 2.09,
      "step": 176302
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.2501691579818726,
      "learning_rate": 7.798763989333466e-05,
      "loss": 1.8935,
      "step": 176303
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.469254970550537,
      "learning_rate": 7.798488875918934e-05,
      "loss": 2.0203,
      "step": 176304
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.3863784074783325,
      "learning_rate": 7.798213766632062e-05,
      "loss": 2.1676,
      "step": 176305
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.3251264095306396,
      "learning_rate": 7.797938661472932e-05,
      "loss": 1.9981,
      "step": 176306
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.298736333847046,
      "learning_rate": 7.797663560441574e-05,
      "loss": 1.9138,
      "step": 176307
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.3701286315917969,
      "learning_rate": 7.797388463538057e-05,
      "loss": 2.1805,
      "step": 176308
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.3660794496536255,
      "learning_rate": 7.797113370762417e-05,
      "loss": 1.9894,
      "step": 176309
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.2679176330566406,
      "learning_rate": 7.79683828211473e-05,
      "loss": 2.0029,
      "step": 176310
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.3376516103744507,
      "learning_rate": 7.796563197595014e-05,
      "loss": 2.0027,
      "step": 176311
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.264954686164856,
      "learning_rate": 7.796288117203342e-05,
      "loss": 2.2034,
      "step": 176312
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.254072904586792,
      "learning_rate": 7.796013040939755e-05,
      "loss": 2.181,
      "step": 176313
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.1948506832122803,
      "learning_rate": 7.795737968804312e-05,
      "loss": 1.8112,
      "step": 176314
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.2737525701522827,
      "learning_rate": 7.795462900797054e-05,
      "loss": 2.0128,
      "step": 176315
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.2324613332748413,
      "learning_rate": 7.795187836918056e-05,
      "loss": 1.9994,
      "step": 176316
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.3303741216659546,
      "learning_rate": 7.794912777167334e-05,
      "loss": 2.2554,
      "step": 176317
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.167431354522705,
      "learning_rate": 7.794637721544965e-05,
      "loss": 2.1193,
      "step": 176318
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.3366296291351318,
      "learning_rate": 7.794362670050984e-05,
      "loss": 1.9649,
      "step": 176319
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.3619943857192993,
      "learning_rate": 7.794087622685457e-05,
      "loss": 2.1276,
      "step": 176320
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.3419339656829834,
      "learning_rate": 7.793812579448422e-05,
      "loss": 2.1593,
      "step": 176321
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.2993948459625244,
      "learning_rate": 7.793537540339953e-05,
      "loss": 1.8967,
      "step": 176322
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.4492238759994507,
      "learning_rate": 7.79326250536007e-05,
      "loss": 2.1614,
      "step": 176323
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.2818679809570312,
      "learning_rate": 7.792987474508842e-05,
      "loss": 2.0021,
      "step": 176324
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.3981515169143677,
      "learning_rate": 7.792712447786313e-05,
      "loss": 2.2848,
      "step": 176325
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.2578929662704468,
      "learning_rate": 7.792437425192545e-05,
      "loss": 2.0112,
      "step": 176326
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.4097793102264404,
      "learning_rate": 7.79216240672757e-05,
      "loss": 1.9946,
      "step": 176327
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.3037960529327393,
      "learning_rate": 7.791887392391473e-05,
      "loss": 2.0043,
      "step": 176328
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.3588855266571045,
      "learning_rate": 7.791612382184266e-05,
      "loss": 1.6992,
      "step": 176329
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.3293761014938354,
      "learning_rate": 7.791337376106023e-05,
      "loss": 1.8461,
      "step": 176330
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.3613885641098022,
      "learning_rate": 7.791062374156784e-05,
      "loss": 2.0578,
      "step": 176331
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.2213339805603027,
      "learning_rate": 7.790787376336612e-05,
      "loss": 1.9243,
      "step": 176332
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.292172908782959,
      "learning_rate": 7.790512382645543e-05,
      "loss": 2.0428,
      "step": 176333
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.4778324365615845,
      "learning_rate": 7.790237393083654e-05,
      "loss": 1.8465,
      "step": 176334
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.312584638595581,
      "learning_rate": 7.789962407650963e-05,
      "loss": 2.0159,
      "step": 176335
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.2519813776016235,
      "learning_rate": 7.78968742634755e-05,
      "loss": 2.0915,
      "step": 176336
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.2350842952728271,
      "learning_rate": 7.789412449173438e-05,
      "loss": 2.0626,
      "step": 176337
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.216266393661499,
      "learning_rate": 7.789137476128706e-05,
      "loss": 1.8176,
      "step": 176338
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.2834347486495972,
      "learning_rate": 7.788862507213382e-05,
      "loss": 2.0014,
      "step": 176339
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.3248149156570435,
      "learning_rate": 7.788587542427547e-05,
      "loss": 2.1757,
      "step": 176340
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.1754765510559082,
      "learning_rate": 7.788312581771214e-05,
      "loss": 1.8495,
      "step": 176341
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.2832016944885254,
      "learning_rate": 7.788037625244461e-05,
      "loss": 2.0156,
      "step": 176342
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.3980075120925903,
      "learning_rate": 7.787762672847324e-05,
      "loss": 2.0665,
      "step": 176343
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.3479119539260864,
      "learning_rate": 7.78748772457987e-05,
      "loss": 1.8551,
      "step": 176344
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.3747094869613647,
      "learning_rate": 7.787212780442133e-05,
      "loss": 2.0378,
      "step": 176345
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.4335248470306396,
      "learning_rate": 7.78693784043419e-05,
      "loss": 1.909,
      "step": 176346
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.2726129293441772,
      "learning_rate": 7.786662904556053e-05,
      "loss": 1.988,
      "step": 176347
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.3911296129226685,
      "learning_rate": 7.786387972807809e-05,
      "loss": 2.0041,
      "step": 176348
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.5939279794692993,
      "learning_rate": 7.78611304518948e-05,
      "loss": 1.9595,
      "step": 176349
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.4800633192062378,
      "learning_rate": 7.785838121701146e-05,
      "loss": 1.8049,
      "step": 176350
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.3568058013916016,
      "learning_rate": 7.785563202342834e-05,
      "loss": 1.9983,
      "step": 176351
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.2734507322311401,
      "learning_rate": 7.785288287114612e-05,
      "loss": 2.1881,
      "step": 176352
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.2307171821594238,
      "learning_rate": 7.785013376016525e-05,
      "loss": 2.2012,
      "step": 176353
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.313431978225708,
      "learning_rate": 7.784738469048625e-05,
      "loss": 1.9728,
      "step": 176354
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.527305245399475,
      "learning_rate": 7.784463566210952e-05,
      "loss": 2.0922,
      "step": 176355
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.283390998840332,
      "learning_rate": 7.784188667503575e-05,
      "loss": 2.0808,
      "step": 176356
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.2894471883773804,
      "learning_rate": 7.783913772926526e-05,
      "loss": 2.1393,
      "step": 176357
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.3518320322036743,
      "learning_rate": 7.783638882479878e-05,
      "loss": 2.0785,
      "step": 176358
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.503811240196228,
      "learning_rate": 7.783363996163671e-05,
      "loss": 1.9067,
      "step": 176359
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.3488448858261108,
      "learning_rate": 7.783089113977955e-05,
      "loss": 1.8712,
      "step": 176360
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.297412633895874,
      "learning_rate": 7.782814235922775e-05,
      "loss": 2.0252,
      "step": 176361
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.4326385259628296,
      "learning_rate": 7.782539361998195e-05,
      "loss": 1.9967,
      "step": 176362
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.2701857089996338,
      "learning_rate": 7.782264492204253e-05,
      "loss": 1.8538,
      "step": 176363
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.443282961845398,
      "learning_rate": 7.781989626541016e-05,
      "loss": 1.9799,
      "step": 176364
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.235992431640625,
      "learning_rate": 7.781714765008529e-05,
      "loss": 2.1115,
      "step": 176365
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.303115725517273,
      "learning_rate": 7.781439907606826e-05,
      "loss": 1.982,
      "step": 176366
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.187716007232666,
      "learning_rate": 7.781165054335987e-05,
      "loss": 1.9769,
      "step": 176367
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.4150340557098389,
      "learning_rate": 7.780890205196047e-05,
      "loss": 2.1184,
      "step": 176368
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.3912380933761597,
      "learning_rate": 7.780615360187051e-05,
      "loss": 1.9746,
      "step": 176369
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.3199126720428467,
      "learning_rate": 7.780340519309066e-05,
      "loss": 2.0378,
      "step": 176370
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.3317298889160156,
      "learning_rate": 7.780065682562134e-05,
      "loss": 1.8363,
      "step": 176371
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.2004996538162231,
      "learning_rate": 7.7797908499463e-05,
      "loss": 1.9578,
      "step": 176372
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.354252815246582,
      "learning_rate": 7.779516021461632e-05,
      "loss": 2.2083,
      "step": 176373
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.4588420391082764,
      "learning_rate": 7.779241197108161e-05,
      "loss": 2.1021,
      "step": 176374
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.2228572368621826,
      "learning_rate": 7.778966376885956e-05,
      "loss": 1.9635,
      "step": 176375
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.2697772979736328,
      "learning_rate": 7.778691560795061e-05,
      "loss": 2.0341,
      "step": 176376
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.2639926671981812,
      "learning_rate": 7.77841674883553e-05,
      "loss": 2.0192,
      "step": 176377
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.250961184501648,
      "learning_rate": 7.778141941007399e-05,
      "loss": 1.9748,
      "step": 176378
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.3536609411239624,
      "learning_rate": 7.777867137310742e-05,
      "loss": 1.876,
      "step": 176379
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.3111932277679443,
      "learning_rate": 7.777592337745588e-05,
      "loss": 2.2358,
      "step": 176380
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.2968374490737915,
      "learning_rate": 7.777317542312011e-05,
      "loss": 2.0817,
      "step": 176381
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.3941701650619507,
      "learning_rate": 7.777042751010044e-05,
      "loss": 2.0872,
      "step": 176382
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.3792266845703125,
      "learning_rate": 7.77676796383975e-05,
      "loss": 1.8873,
      "step": 176383
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.198101282119751,
      "learning_rate": 7.776493180801163e-05,
      "loss": 2.0489,
      "step": 176384
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.2684166431427002,
      "learning_rate": 7.776218401894355e-05,
      "loss": 2.1076,
      "step": 176385
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.424569845199585,
      "learning_rate": 7.775943627119358e-05,
      "loss": 1.9833,
      "step": 176386
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.305408239364624,
      "learning_rate": 7.775668856476244e-05,
      "loss": 2.0739,
      "step": 176387
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.3551265001296997,
      "learning_rate": 7.77539408996504e-05,
      "loss": 1.8441,
      "step": 176388
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.2328814268112183,
      "learning_rate": 7.775119327585828e-05,
      "loss": 1.9741,
      "step": 176389
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.3070138692855835,
      "learning_rate": 7.774844569338624e-05,
      "loss": 1.9287,
      "step": 176390
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.290358543395996,
      "learning_rate": 7.774569815223507e-05,
      "loss": 2.0067,
      "step": 176391
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.2473371028900146,
      "learning_rate": 7.774295065240506e-05,
      "loss": 2.0594,
      "step": 176392
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.2461975812911987,
      "learning_rate": 7.774020319389693e-05,
      "loss": 1.98,
      "step": 176393
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.2554587125778198,
      "learning_rate": 7.773745577671097e-05,
      "loss": 2.0184,
      "step": 176394
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.3239022493362427,
      "learning_rate": 7.773470840084803e-05,
      "loss": 1.9498,
      "step": 176395
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.3741390705108643,
      "learning_rate": 7.77319610663082e-05,
      "loss": 1.9213,
      "step": 176396
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.4262012243270874,
      "learning_rate": 7.772921377309231e-05,
      "loss": 2.0835,
      "step": 176397
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.3126100301742554,
      "learning_rate": 7.772646652120064e-05,
      "loss": 2.0814,
      "step": 176398
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.4496716260910034,
      "learning_rate": 7.77237193106339e-05,
      "loss": 2.2829,
      "step": 176399
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.4218930006027222,
      "learning_rate": 7.772097214139245e-05,
      "loss": 1.7949,
      "step": 176400
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.3775039911270142,
      "learning_rate": 7.771822501347703e-05,
      "loss": 2.0396,
      "step": 176401
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.2524346113204956,
      "learning_rate": 7.771547792688783e-05,
      "loss": 1.9475,
      "step": 176402
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.4661247730255127,
      "learning_rate": 7.77127308816256e-05,
      "loss": 2.1179,
      "step": 176403
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.2100086212158203,
      "learning_rate": 7.770998387769067e-05,
      "loss": 2.2542,
      "step": 176404
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.2661974430084229,
      "learning_rate": 7.770723691508376e-05,
      "loss": 1.9117,
      "step": 176405
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.2489962577819824,
      "learning_rate": 7.770448999380515e-05,
      "loss": 2.0595,
      "step": 176406
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.3419491052627563,
      "learning_rate": 7.770174311385565e-05,
      "loss": 1.9989,
      "step": 176407
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.526407241821289,
      "learning_rate": 7.769899627523546e-05,
      "loss": 2.0827,
      "step": 176408
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.6883999109268188,
      "learning_rate": 7.769624947794527e-05,
      "loss": 2.0628,
      "step": 176409
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.2576016187667847,
      "learning_rate": 7.769350272198546e-05,
      "loss": 1.9456,
      "step": 176410
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.2895581722259521,
      "learning_rate": 7.769075600735672e-05,
      "loss": 1.7966,
      "step": 176411
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.2142281532287598,
      "learning_rate": 7.768800933405937e-05,
      "loss": 1.9841,
      "step": 176412
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.3002640008926392,
      "learning_rate": 7.768526270209421e-05,
      "loss": 2.1976,
      "step": 176413
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.3990955352783203,
      "learning_rate": 7.768251611146134e-05,
      "loss": 1.9949,
      "step": 176414
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.2633717060089111,
      "learning_rate": 7.767976956216162e-05,
      "loss": 1.8873,
      "step": 176415
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.285644769668579,
      "learning_rate": 7.767702305419533e-05,
      "loss": 1.9723,
      "step": 176416
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.356097936630249,
      "learning_rate": 7.767427658756314e-05,
      "loss": 1.9769,
      "step": 176417
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.4992635250091553,
      "learning_rate": 7.767153016226543e-05,
      "loss": 2.1839,
      "step": 176418
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.3287369012832642,
      "learning_rate": 7.766878377830295e-05,
      "loss": 1.879,
      "step": 176419
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.2796833515167236,
      "learning_rate": 7.766603743567587e-05,
      "loss": 2.1069,
      "step": 176420
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.2754908800125122,
      "learning_rate": 7.766329113438496e-05,
      "loss": 1.9605,
      "step": 176421
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.2759307622909546,
      "learning_rate": 7.766054487443054e-05,
      "loss": 1.9963,
      "step": 176422
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.2320445775985718,
      "learning_rate": 7.765779865581334e-05,
      "loss": 2.2475,
      "step": 176423
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.2455418109893799,
      "learning_rate": 7.765505247853362e-05,
      "loss": 2.1514,
      "step": 176424
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.294140338897705,
      "learning_rate": 7.765230634259214e-05,
      "loss": 2.1823,
      "step": 176425
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.2891356945037842,
      "learning_rate": 7.764956024798929e-05,
      "loss": 1.9447,
      "step": 176426
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.388065218925476,
      "learning_rate": 7.764681419472558e-05,
      "loss": 1.7834,
      "step": 176427
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.285691499710083,
      "learning_rate": 7.764406818280144e-05,
      "loss": 2.0753,
      "step": 176428
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.3329377174377441,
      "learning_rate": 7.764132221221754e-05,
      "loss": 2.1352,
      "step": 176429
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.4631526470184326,
      "learning_rate": 7.763857628297423e-05,
      "loss": 2.0817,
      "step": 176430
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.5086230039596558,
      "learning_rate": 7.76358303950722e-05,
      "loss": 2.3129,
      "step": 176431
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.383962869644165,
      "learning_rate": 7.763308454851188e-05,
      "loss": 2.1369,
      "step": 176432
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.2772252559661865,
      "learning_rate": 7.763033874329375e-05,
      "loss": 1.8816,
      "step": 176433
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.364996075630188,
      "learning_rate": 7.762759297941825e-05,
      "loss": 1.8072,
      "step": 176434
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.3035576343536377,
      "learning_rate": 7.762484725688608e-05,
      "loss": 2.1414,
      "step": 176435
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.471383810043335,
      "learning_rate": 7.762210157569757e-05,
      "loss": 2.0363,
      "step": 176436
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.3329893350601196,
      "learning_rate": 7.761935593585336e-05,
      "loss": 2.1844,
      "step": 176437
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.4598103761672974,
      "learning_rate": 7.761661033735394e-05,
      "loss": 1.9152,
      "step": 176438
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.359065294265747,
      "learning_rate": 7.761386478019978e-05,
      "loss": 2.025,
      "step": 176439
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.347861647605896,
      "learning_rate": 7.76111192643913e-05,
      "loss": 2.1271,
      "step": 176440
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.411482810974121,
      "learning_rate": 7.76083737899292e-05,
      "loss": 2.2226,
      "step": 176441
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.3529231548309326,
      "learning_rate": 7.760562835681382e-05,
      "loss": 2.1784,
      "step": 176442
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.5575388669967651,
      "learning_rate": 7.760288296504587e-05,
      "loss": 2.2173,
      "step": 176443
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.5682240724563599,
      "learning_rate": 7.76001376146257e-05,
      "loss": 1.9937,
      "step": 176444
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.261565923690796,
      "learning_rate": 7.759739230555387e-05,
      "loss": 1.932,
      "step": 176445
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.3633731603622437,
      "learning_rate": 7.759464703783081e-05,
      "loss": 2.053,
      "step": 176446
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.170928716659546,
      "learning_rate": 7.759190181145718e-05,
      "loss": 2.0357,
      "step": 176447
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.4119043350219727,
      "learning_rate": 7.758915662643332e-05,
      "loss": 2.2365,
      "step": 176448
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.3451863527297974,
      "learning_rate": 7.758641148275995e-05,
      "loss": 1.9336,
      "step": 176449
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.3273676633834839,
      "learning_rate": 7.758366638043742e-05,
      "loss": 2.1775,
      "step": 176450
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.5039536952972412,
      "learning_rate": 7.758092131946623e-05,
      "loss": 2.0387,
      "step": 176451
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.2803183794021606,
      "learning_rate": 7.757817629984704e-05,
      "loss": 1.917,
      "step": 176452
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.2887845039367676,
      "learning_rate": 7.757543132158026e-05,
      "loss": 2.0231,
      "step": 176453
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.233289361000061,
      "learning_rate": 7.757268638466629e-05,
      "loss": 1.9534,
      "step": 176454
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.4179526567459106,
      "learning_rate": 7.756994148910588e-05,
      "loss": 1.9452,
      "step": 176455
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.3004553318023682,
      "learning_rate": 7.756719663489939e-05,
      "loss": 2.2515,
      "step": 176456
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.4252458810806274,
      "learning_rate": 7.756445182204728e-05,
      "loss": 2.2709,
      "step": 176457
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.3765026330947876,
      "learning_rate": 7.756170705055023e-05,
      "loss": 2.058,
      "step": 176458
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.1941388845443726,
      "learning_rate": 7.755896232040865e-05,
      "loss": 1.9273,
      "step": 176459
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.243722677230835,
      "learning_rate": 7.755621763162297e-05,
      "loss": 2.0815,
      "step": 176460
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.1962045431137085,
      "learning_rate": 7.755347298419388e-05,
      "loss": 2.0254,
      "step": 176461
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.2885867357254028,
      "learning_rate": 7.75507283781218e-05,
      "loss": 2.2626,
      "step": 176462
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.1574575901031494,
      "learning_rate": 7.754798381340713e-05,
      "loss": 2.1866,
      "step": 176463
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.1555973291397095,
      "learning_rate": 7.754523929005059e-05,
      "loss": 1.7465,
      "step": 176464
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.43112313747406,
      "learning_rate": 7.754249480805252e-05,
      "loss": 2.2337,
      "step": 176465
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.2761695384979248,
      "learning_rate": 7.753975036741358e-05,
      "loss": 2.3079,
      "step": 176466
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.2655105590820312,
      "learning_rate": 7.753700596813418e-05,
      "loss": 1.9513,
      "step": 176467
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.2555153369903564,
      "learning_rate": 7.753426161021486e-05,
      "loss": 2.0272,
      "step": 176468
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.266263723373413,
      "learning_rate": 7.753151729365603e-05,
      "loss": 2.1614,
      "step": 176469
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.3522162437438965,
      "learning_rate": 7.752877301845837e-05,
      "loss": 2.1385,
      "step": 176470
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.2266829013824463,
      "learning_rate": 7.752602878462224e-05,
      "loss": 1.7889,
      "step": 176471
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.4055312871932983,
      "learning_rate": 7.752328459214832e-05,
      "loss": 2.103,
      "step": 176472
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.3065801858901978,
      "learning_rate": 7.752054044103691e-05,
      "loss": 2.0251,
      "step": 176473
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.2944098711013794,
      "learning_rate": 7.75177963312888e-05,
      "loss": 2.1685,
      "step": 176474
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.2803701162338257,
      "learning_rate": 7.751505226290419e-05,
      "loss": 1.9972,
      "step": 176475
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.331525206565857,
      "learning_rate": 7.751230823588378e-05,
      "loss": 2.0726,
      "step": 176476
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.4820247888565063,
      "learning_rate": 7.750956425022798e-05,
      "loss": 2.2705,
      "step": 176477
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.2417519092559814,
      "learning_rate": 7.750682030593741e-05,
      "loss": 2.024,
      "step": 176478
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.3391773700714111,
      "learning_rate": 7.750407640301247e-05,
      "loss": 1.9989,
      "step": 176479
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.2288217544555664,
      "learning_rate": 7.750133254145386e-05,
      "loss": 2.0939,
      "step": 176480
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.1975765228271484,
      "learning_rate": 7.749858872126177e-05,
      "loss": 2.0123,
      "step": 176481
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.2244375944137573,
      "learning_rate": 7.7495844942437e-05,
      "loss": 1.8707,
      "step": 176482
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.3287451267242432,
      "learning_rate": 7.749310120497987e-05,
      "loss": 2.2795,
      "step": 176483
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.4684903621673584,
      "learning_rate": 7.749035750889108e-05,
      "loss": 2.19,
      "step": 176484
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.3780200481414795,
      "learning_rate": 7.748761385417089e-05,
      "loss": 1.9959,
      "step": 176485
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.3700542449951172,
      "learning_rate": 7.748487024082018e-05,
      "loss": 1.7438,
      "step": 176486
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.2937521934509277,
      "learning_rate": 7.748212666883902e-05,
      "loss": 2.0324,
      "step": 176487
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.3380157947540283,
      "learning_rate": 7.747938313822825e-05,
      "loss": 1.9814,
      "step": 176488
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.3260776996612549,
      "learning_rate": 7.747663964898817e-05,
      "loss": 1.8669,
      "step": 176489
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.4293791055679321,
      "learning_rate": 7.747389620111945e-05,
      "loss": 2.2672,
      "step": 176490
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.280197024345398,
      "learning_rate": 7.747115279462244e-05,
      "loss": 2.1465,
      "step": 176491
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.2550976276397705,
      "learning_rate": 7.746840942949788e-05,
      "loss": 2.306,
      "step": 176492
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.4496235847473145,
      "learning_rate": 7.746566610574611e-05,
      "loss": 2.0876,
      "step": 176493
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.2751919031143188,
      "learning_rate": 7.746292282336766e-05,
      "loss": 2.0804,
      "step": 176494
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.2061986923217773,
      "learning_rate": 7.7460179582363e-05,
      "loss": 1.9207,
      "step": 176495
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.3908514976501465,
      "learning_rate": 7.745743638273275e-05,
      "loss": 2.1201,
      "step": 176496
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.2111480236053467,
      "learning_rate": 7.74546932244773e-05,
      "loss": 2.1134,
      "step": 176497
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.3042536973953247,
      "learning_rate": 7.74519501075973e-05,
      "loss": 1.872,
      "step": 176498
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.296139121055603,
      "learning_rate": 7.744920703209318e-05,
      "loss": 1.9781,
      "step": 176499
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.3103790283203125,
      "learning_rate": 7.744646399796545e-05,
      "loss": 1.9141,
      "step": 176500
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.5886322259902954,
      "learning_rate": 7.744372100521454e-05,
      "loss": 2.0225,
      "step": 176501
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.3571207523345947,
      "learning_rate": 7.744097805384114e-05,
      "loss": 1.8906,
      "step": 176502
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.2763891220092773,
      "learning_rate": 7.743823514384557e-05,
      "loss": 2.1031,
      "step": 176503
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.299256682395935,
      "learning_rate": 7.743549227522851e-05,
      "loss": 1.9747,
      "step": 176504
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.3043115139007568,
      "learning_rate": 7.74327494479904e-05,
      "loss": 2.0747,
      "step": 176505
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.2910680770874023,
      "learning_rate": 7.743000666213177e-05,
      "loss": 2.0963,
      "step": 176506
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.4511677026748657,
      "learning_rate": 7.742726391765297e-05,
      "loss": 1.8943,
      "step": 176507
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.359599232673645,
      "learning_rate": 7.742452121455477e-05,
      "loss": 2.1311,
      "step": 176508
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.3014014959335327,
      "learning_rate": 7.742177855283744e-05,
      "loss": 1.9851,
      "step": 176509
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.316899061203003,
      "learning_rate": 7.74190359325017e-05,
      "loss": 2.0832,
      "step": 176510
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.458609938621521,
      "learning_rate": 7.741629335354795e-05,
      "loss": 2.0848,
      "step": 176511
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.4132593870162964,
      "learning_rate": 7.741355081597675e-05,
      "loss": 2.1211,
      "step": 176512
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.410885214805603,
      "learning_rate": 7.741080831978843e-05,
      "loss": 1.9131,
      "step": 176513
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.3151565790176392,
      "learning_rate": 7.740806586498378e-05,
      "loss": 2.1359,
      "step": 176514
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.378142237663269,
      "learning_rate": 7.740532345156307e-05,
      "loss": 1.8314,
      "step": 176515
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.48744797706604,
      "learning_rate": 7.740258107952697e-05,
      "loss": 2.0669,
      "step": 176516
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.3433997631072998,
      "learning_rate": 7.739983874887597e-05,
      "loss": 1.9542,
      "step": 176517
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.2436213493347168,
      "learning_rate": 7.739709645961051e-05,
      "loss": 1.8696,
      "step": 176518
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.1315008401870728,
      "learning_rate": 7.739435421173107e-05,
      "loss": 2.0477,
      "step": 176519
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.3906995058059692,
      "learning_rate": 7.73916120052383e-05,
      "loss": 2.0305,
      "step": 176520
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.2289382219314575,
      "learning_rate": 7.738886984013254e-05,
      "loss": 1.7855,
      "step": 176521
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.7028270959854126,
      "learning_rate": 7.738612771641448e-05,
      "loss": 2.1253,
      "step": 176522
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.379799246788025,
      "learning_rate": 7.738338563408455e-05,
      "loss": 2.0313,
      "step": 176523
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.2634468078613281,
      "learning_rate": 7.738064359314324e-05,
      "loss": 1.8236,
      "step": 176524
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.528376817703247,
      "learning_rate": 7.737790159359099e-05,
      "loss": 2.18,
      "step": 176525
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.3203907012939453,
      "learning_rate": 7.737515963542848e-05,
      "loss": 2.2294,
      "step": 176526
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.3552813529968262,
      "learning_rate": 7.737241771865604e-05,
      "loss": 2.1242,
      "step": 176527
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.2844451665878296,
      "learning_rate": 7.736967584327437e-05,
      "loss": 1.9909,
      "step": 176528
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.3340702056884766,
      "learning_rate": 7.736693400928386e-05,
      "loss": 2.0471,
      "step": 176529
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.340258240699768,
      "learning_rate": 7.736419221668503e-05,
      "loss": 2.0166,
      "step": 176530
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.3541152477264404,
      "learning_rate": 7.736145046547832e-05,
      "loss": 2.085,
      "step": 176531
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.1901426315307617,
      "learning_rate": 7.735870875566444e-05,
      "loss": 1.981,
      "step": 176532
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.2305169105529785,
      "learning_rate": 7.735596708724365e-05,
      "loss": 2.0179,
      "step": 176533
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.2796807289123535,
      "learning_rate": 7.735322546021669e-05,
      "loss": 2.0991,
      "step": 176534
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.3615902662277222,
      "learning_rate": 7.735048387458397e-05,
      "loss": 2.1235,
      "step": 176535
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.2400859594345093,
      "learning_rate": 7.73477423303459e-05,
      "loss": 1.9556,
      "step": 176536
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.3191269636154175,
      "learning_rate": 7.73450008275032e-05,
      "loss": 2.1259,
      "step": 176537
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.2235453128814697,
      "learning_rate": 7.734225936605625e-05,
      "loss": 2.0943,
      "step": 176538
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.1966185569763184,
      "learning_rate": 7.733951794600547e-05,
      "loss": 1.8703,
      "step": 176539
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.2571595907211304,
      "learning_rate": 7.73367765673516e-05,
      "loss": 2.0421,
      "step": 176540
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.1937389373779297,
      "learning_rate": 7.733403523009501e-05,
      "loss": 2.0765,
      "step": 176541
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.3122962713241577,
      "learning_rate": 7.733129393423616e-05,
      "loss": 1.9244,
      "step": 176542
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.2864304780960083,
      "learning_rate": 7.732855267977567e-05,
      "loss": 1.9208,
      "step": 176543
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.3990455865859985,
      "learning_rate": 7.732581146671402e-05,
      "loss": 2.1256,
      "step": 176544
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.4159682989120483,
      "learning_rate": 7.732307029505163e-05,
      "loss": 1.8609,
      "step": 176545
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.3004329204559326,
      "learning_rate": 7.73203291647892e-05,
      "loss": 2.1273,
      "step": 176546
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.2681092023849487,
      "learning_rate": 7.731758807592707e-05,
      "loss": 1.9684,
      "step": 176547
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.3221800327301025,
      "learning_rate": 7.731484702846573e-05,
      "loss": 1.9857,
      "step": 176548
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.2715500593185425,
      "learning_rate": 7.731210602240587e-05,
      "loss": 2.1317,
      "step": 176549
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.3089665174484253,
      "learning_rate": 7.73093650577478e-05,
      "loss": 1.9527,
      "step": 176550
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.388063669204712,
      "learning_rate": 7.73066241344922e-05,
      "loss": 2.0308,
      "step": 176551
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.3991613388061523,
      "learning_rate": 7.730388325263953e-05,
      "loss": 1.8943,
      "step": 176552
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.3020415306091309,
      "learning_rate": 7.730114241219022e-05,
      "loss": 1.8161,
      "step": 176553
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.339598536491394,
      "learning_rate": 7.729840161314476e-05,
      "loss": 2.1384,
      "step": 176554
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.3422157764434814,
      "learning_rate": 7.729566085550386e-05,
      "loss": 2.0319,
      "step": 176555
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.4064112901687622,
      "learning_rate": 7.729292013926776e-05,
      "loss": 2.087,
      "step": 176556
    },
    {
      "epoch": 2.3,
      "grad_norm": 2.0772483348846436,
      "learning_rate": 7.729017946443725e-05,
      "loss": 1.8999,
      "step": 176557
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.4283301830291748,
      "learning_rate": 7.728743883101258e-05,
      "loss": 2.1757,
      "step": 176558
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.3059824705123901,
      "learning_rate": 7.728469823899446e-05,
      "loss": 1.8756,
      "step": 176559
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.2990561723709106,
      "learning_rate": 7.728195768838334e-05,
      "loss": 1.8758,
      "step": 176560
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.2880113124847412,
      "learning_rate": 7.727921717917971e-05,
      "loss": 2.1406,
      "step": 176561
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.3002469539642334,
      "learning_rate": 7.727647671138395e-05,
      "loss": 1.7293,
      "step": 176562
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.4079283475875854,
      "learning_rate": 7.727373628499683e-05,
      "loss": 1.9456,
      "step": 176563
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.4326255321502686,
      "learning_rate": 7.727099590001863e-05,
      "loss": 2.1138,
      "step": 176564
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.2932226657867432,
      "learning_rate": 7.726825555645003e-05,
      "loss": 2.0457,
      "step": 176565
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.4339264631271362,
      "learning_rate": 7.726551525429148e-05,
      "loss": 2.2544,
      "step": 176566
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.3743185997009277,
      "learning_rate": 7.726277499354347e-05,
      "loss": 2.128,
      "step": 176567
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.2950022220611572,
      "learning_rate": 7.72600347742064e-05,
      "loss": 1.9325,
      "step": 176568
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.5072795152664185,
      "learning_rate": 7.7257294596281e-05,
      "loss": 2.1492,
      "step": 176569
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.5881798267364502,
      "learning_rate": 7.72545544597676e-05,
      "loss": 2.0117,
      "step": 176570
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.7493581771850586,
      "learning_rate": 7.725181436466686e-05,
      "loss": 2.0005,
      "step": 176571
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.443735122680664,
      "learning_rate": 7.724907431097921e-05,
      "loss": 2.0825,
      "step": 176572
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.3759111166000366,
      "learning_rate": 7.724633429870517e-05,
      "loss": 1.9134,
      "step": 176573
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.2486087083816528,
      "learning_rate": 7.724359432784514e-05,
      "loss": 1.9619,
      "step": 176574
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.381903886795044,
      "learning_rate": 7.724085439839985e-05,
      "loss": 1.9705,
      "step": 176575
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.3679441213607788,
      "learning_rate": 7.723811451036955e-05,
      "loss": 1.9484,
      "step": 176576
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.450029730796814,
      "learning_rate": 7.723537466375503e-05,
      "loss": 2.1506,
      "step": 176577
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.358405351638794,
      "learning_rate": 7.723263485855662e-05,
      "loss": 2.2093,
      "step": 176578
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.5519832372665405,
      "learning_rate": 7.722989509477491e-05,
      "loss": 2.069,
      "step": 176579
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.6209734678268433,
      "learning_rate": 7.722715537241023e-05,
      "loss": 1.9778,
      "step": 176580
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.2362116575241089,
      "learning_rate": 7.722441569146337e-05,
      "loss": 2.2569,
      "step": 176581
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.160738229751587,
      "learning_rate": 7.722167605193455e-05,
      "loss": 1.9938,
      "step": 176582
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.3339365720748901,
      "learning_rate": 7.721893645382455e-05,
      "loss": 2.2069,
      "step": 176583
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.5873433351516724,
      "learning_rate": 7.721619689713376e-05,
      "loss": 2.1034,
      "step": 176584
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.3430603742599487,
      "learning_rate": 7.721345738186266e-05,
      "loss": 1.9484,
      "step": 176585
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.4104539155960083,
      "learning_rate": 7.721071790801169e-05,
      "loss": 1.9183,
      "step": 176586
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.356001615524292,
      "learning_rate": 7.720797847558156e-05,
      "loss": 1.8717,
      "step": 176587
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.2991738319396973,
      "learning_rate": 7.720523908457259e-05,
      "loss": 2.0425,
      "step": 176588
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.2950730323791504,
      "learning_rate": 7.720249973498546e-05,
      "loss": 1.9905,
      "step": 176589
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.4442048072814941,
      "learning_rate": 7.719976042682058e-05,
      "loss": 2.2893,
      "step": 176590
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.2131303548812866,
      "learning_rate": 7.719702116007845e-05,
      "loss": 2.3701,
      "step": 176591
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.3106194734573364,
      "learning_rate": 7.719428193475955e-05,
      "loss": 1.9219,
      "step": 176592
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.2867896556854248,
      "learning_rate": 7.719154275086452e-05,
      "loss": 2.023,
      "step": 176593
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.4426405429840088,
      "learning_rate": 7.718880360839366e-05,
      "loss": 2.2778,
      "step": 176594
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.342445731163025,
      "learning_rate": 7.718606450734775e-05,
      "loss": 2.0275,
      "step": 176595
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.2242339849472046,
      "learning_rate": 7.718332544772714e-05,
      "loss": 2.0182,
      "step": 176596
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.3905469179153442,
      "learning_rate": 7.718058642953235e-05,
      "loss": 2.0545,
      "step": 176597
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.2365825176239014,
      "learning_rate": 7.71778474527638e-05,
      "loss": 2.2784,
      "step": 176598
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.4243998527526855,
      "learning_rate": 7.717510851742217e-05,
      "loss": 2.0851,
      "step": 176599
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.505020022392273,
      "learning_rate": 7.717236962350784e-05,
      "loss": 1.9614,
      "step": 176600
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.238498330116272,
      "learning_rate": 7.716963077102143e-05,
      "loss": 2.045,
      "step": 176601
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.3528856039047241,
      "learning_rate": 7.716689195996342e-05,
      "loss": 2.0554,
      "step": 176602
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.2478678226470947,
      "learning_rate": 7.716415319033426e-05,
      "loss": 1.9761,
      "step": 176603
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.221114993095398,
      "learning_rate": 7.716141446213444e-05,
      "loss": 1.8442,
      "step": 176604
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.2889859676361084,
      "learning_rate": 7.715867577536458e-05,
      "loss": 2.0863,
      "step": 176605
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.2373840808868408,
      "learning_rate": 7.715593713002503e-05,
      "loss": 1.9639,
      "step": 176606
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.2929017543792725,
      "learning_rate": 7.715319852611652e-05,
      "loss": 1.8729,
      "step": 176607
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.286834478378296,
      "learning_rate": 7.715045996363943e-05,
      "loss": 1.8201,
      "step": 176608
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.3459192514419556,
      "learning_rate": 7.714772144259426e-05,
      "loss": 1.9939,
      "step": 176609
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.3637738227844238,
      "learning_rate": 7.714498296298146e-05,
      "loss": 2.1638,
      "step": 176610
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.2900093793869019,
      "learning_rate": 7.714224452480168e-05,
      "loss": 2.0286,
      "step": 176611
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.33609139919281,
      "learning_rate": 7.713950612805529e-05,
      "loss": 2.1546,
      "step": 176612
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.3442682027816772,
      "learning_rate": 7.713676777274298e-05,
      "loss": 2.1922,
      "step": 176613
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.2751432657241821,
      "learning_rate": 7.713402945886513e-05,
      "loss": 1.9059,
      "step": 176614
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.3343091011047363,
      "learning_rate": 7.713129118642227e-05,
      "loss": 2.0026,
      "step": 176615
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.343455195426941,
      "learning_rate": 7.71285529554148e-05,
      "loss": 2.0162,
      "step": 176616
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.2214211225509644,
      "learning_rate": 7.712581476584349e-05,
      "loss": 2.0245,
      "step": 176617
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.4319791793823242,
      "learning_rate": 7.712307661770856e-05,
      "loss": 2.107,
      "step": 176618
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.3387198448181152,
      "learning_rate": 7.712033851101075e-05,
      "loss": 2.035,
      "step": 176619
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.3218516111373901,
      "learning_rate": 7.71176004457505e-05,
      "loss": 2.1352,
      "step": 176620
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.826869249343872,
      "learning_rate": 7.711486242192819e-05,
      "loss": 2.0143,
      "step": 176621
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.466318964958191,
      "learning_rate": 7.711212443954451e-05,
      "loss": 1.9564,
      "step": 176622
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.410020112991333,
      "learning_rate": 7.710938649859991e-05,
      "loss": 2.0771,
      "step": 176623
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.3542754650115967,
      "learning_rate": 7.71066485990948e-05,
      "loss": 2.0466,
      "step": 176624
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.4066144227981567,
      "learning_rate": 7.710391074102984e-05,
      "loss": 2.1124,
      "step": 176625
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.392524242401123,
      "learning_rate": 7.710117292440539e-05,
      "loss": 1.9142,
      "step": 176626
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.3069944381713867,
      "learning_rate": 7.709843514922214e-05,
      "loss": 2.1537,
      "step": 176627
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.477523922920227,
      "learning_rate": 7.709569741548049e-05,
      "loss": 2.1219,
      "step": 176628
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.350095510482788,
      "learning_rate": 7.709295972318097e-05,
      "loss": 1.9498,
      "step": 176629
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.2262541055679321,
      "learning_rate": 7.709022207232396e-05,
      "loss": 2.1658,
      "step": 176630
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.4236558675765991,
      "learning_rate": 7.70874844629102e-05,
      "loss": 2.1438,
      "step": 176631
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.3872030973434448,
      "learning_rate": 7.708474689493999e-05,
      "loss": 1.9686,
      "step": 176632
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.253088116645813,
      "learning_rate": 7.708200936841402e-05,
      "loss": 2.0013,
      "step": 176633
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.2420010566711426,
      "learning_rate": 7.70792718833327e-05,
      "loss": 1.9127,
      "step": 176634
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.2167831659317017,
      "learning_rate": 7.707653443969647e-05,
      "loss": 2.0966,
      "step": 176635
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.3249609470367432,
      "learning_rate": 7.707379703750601e-05,
      "loss": 2.1291,
      "step": 176636
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.2320683002471924,
      "learning_rate": 7.707105967676174e-05,
      "loss": 2.0182,
      "step": 176637
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.3882737159729004,
      "learning_rate": 7.706832235746407e-05,
      "loss": 2.0229,
      "step": 176638
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.3291232585906982,
      "learning_rate": 7.706558507961373e-05,
      "loss": 2.2119,
      "step": 176639
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.3771746158599854,
      "learning_rate": 7.706284784321109e-05,
      "loss": 2.0955,
      "step": 176640
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.297590732574463,
      "learning_rate": 7.706011064825656e-05,
      "loss": 2.1364,
      "step": 176641
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.1120144128799438,
      "learning_rate": 7.705737349475088e-05,
      "loss": 2.1342,
      "step": 176642
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.316058874130249,
      "learning_rate": 7.705463638269444e-05,
      "loss": 1.8894,
      "step": 176643
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.2834503650665283,
      "learning_rate": 7.705189931208764e-05,
      "loss": 1.7426,
      "step": 176644
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.4193029403686523,
      "learning_rate": 7.704916228293122e-05,
      "loss": 2.0951,
      "step": 176645
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.3312190771102905,
      "learning_rate": 7.704642529522554e-05,
      "loss": 1.8826,
      "step": 176646
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.1701335906982422,
      "learning_rate": 7.704368834897107e-05,
      "loss": 2.0165,
      "step": 176647
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.5610700845718384,
      "learning_rate": 7.704095144416845e-05,
      "loss": 2.0005,
      "step": 176648
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.3289527893066406,
      "learning_rate": 7.703821458081807e-05,
      "loss": 1.7642,
      "step": 176649
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.2838168144226074,
      "learning_rate": 7.703547775892056e-05,
      "loss": 1.8841,
      "step": 176650
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.2490909099578857,
      "learning_rate": 7.703274097847639e-05,
      "loss": 1.9316,
      "step": 176651
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.3988300561904907,
      "learning_rate": 7.703000423948603e-05,
      "loss": 2.0793,
      "step": 176652
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.4002594947814941,
      "learning_rate": 7.70272675419499e-05,
      "loss": 2.1001,
      "step": 176653
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.389306664466858,
      "learning_rate": 7.70245308858687e-05,
      "loss": 2.0665,
      "step": 176654
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.3007242679595947,
      "learning_rate": 7.702179427124279e-05,
      "loss": 2.0672,
      "step": 176655
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.2211979627609253,
      "learning_rate": 7.701905769807279e-05,
      "loss": 2.2201,
      "step": 176656
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.4979110956192017,
      "learning_rate": 7.701632116635916e-05,
      "loss": 1.8239,
      "step": 176657
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.382267713546753,
      "learning_rate": 7.701358467610243e-05,
      "loss": 1.8425,
      "step": 176658
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.3559032678604126,
      "learning_rate": 7.701084822730297e-05,
      "loss": 1.8871,
      "step": 176659
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.4331446886062622,
      "learning_rate": 7.700811181996149e-05,
      "loss": 2.0678,
      "step": 176660
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.3653507232666016,
      "learning_rate": 7.700537545407834e-05,
      "loss": 2.0425,
      "step": 176661
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.2188031673431396,
      "learning_rate": 7.700263912965418e-05,
      "loss": 2.0536,
      "step": 176662
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.3445136547088623,
      "learning_rate": 7.699990284668944e-05,
      "loss": 1.9978,
      "step": 176663
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.3034257888793945,
      "learning_rate": 7.699716660518464e-05,
      "loss": 2.1515,
      "step": 176664
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.3642152547836304,
      "learning_rate": 7.699443040514016e-05,
      "loss": 2.0881,
      "step": 176665
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.4219624996185303,
      "learning_rate": 7.699169424655672e-05,
      "loss": 2.1684,
      "step": 176666
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.5599188804626465,
      "learning_rate": 7.698895812943468e-05,
      "loss": 2.0836,
      "step": 176667
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.3185111284255981,
      "learning_rate": 7.698622205377464e-05,
      "loss": 2.0611,
      "step": 176668
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.188889741897583,
      "learning_rate": 7.698348601957711e-05,
      "loss": 1.9281,
      "step": 176669
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.362518310546875,
      "learning_rate": 7.698075002684254e-05,
      "loss": 1.9512,
      "step": 176670
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.6446765661239624,
      "learning_rate": 7.697801407557136e-05,
      "loss": 2.0902,
      "step": 176671
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.3961416482925415,
      "learning_rate": 7.69752781657643e-05,
      "loss": 1.9067,
      "step": 176672
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.3260935544967651,
      "learning_rate": 7.697254229742166e-05,
      "loss": 2.1576,
      "step": 176673
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.2212525606155396,
      "learning_rate": 7.696980647054409e-05,
      "loss": 2.1771,
      "step": 176674
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.344538927078247,
      "learning_rate": 7.696707068513203e-05,
      "loss": 2.0739,
      "step": 176675
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.4224164485931396,
      "learning_rate": 7.696433494118605e-05,
      "loss": 1.975,
      "step": 176676
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.403506875038147,
      "learning_rate": 7.69615992387065e-05,
      "loss": 1.9921,
      "step": 176677
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.306847333908081,
      "learning_rate": 7.695886357769409e-05,
      "loss": 2.2853,
      "step": 176678
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.3455684185028076,
      "learning_rate": 7.695612795814912e-05,
      "loss": 1.8363,
      "step": 176679
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.4185694456100464,
      "learning_rate": 7.695339238007233e-05,
      "loss": 2.0414,
      "step": 176680
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.448306918144226,
      "learning_rate": 7.695065684346413e-05,
      "loss": 1.9117,
      "step": 176681
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.2444710731506348,
      "learning_rate": 7.694792134832499e-05,
      "loss": 2.0663,
      "step": 176682
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.2752629518508911,
      "learning_rate": 7.694518589465534e-05,
      "loss": 1.9319,
      "step": 176683
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.2400219440460205,
      "learning_rate": 7.694245048245593e-05,
      "loss": 1.9668,
      "step": 176684
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.2159043550491333,
      "learning_rate": 7.693971511172701e-05,
      "loss": 1.9261,
      "step": 176685
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.4486665725708008,
      "learning_rate": 7.693697978246928e-05,
      "loss": 2.1917,
      "step": 176686
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.2703094482421875,
      "learning_rate": 7.693424449468321e-05,
      "loss": 2.1013,
      "step": 176687
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.2642650604248047,
      "learning_rate": 7.693150924836926e-05,
      "loss": 1.8594,
      "step": 176688
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.2930494546890259,
      "learning_rate": 7.692877404352785e-05,
      "loss": 1.8617,
      "step": 176689
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.327326774597168,
      "learning_rate": 7.692603888015967e-05,
      "loss": 2.0241,
      "step": 176690
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.4769870042800903,
      "learning_rate": 7.69233037582651e-05,
      "loss": 1.9461,
      "step": 176691
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.2465332746505737,
      "learning_rate": 7.692056867784477e-05,
      "loss": 2.0188,
      "step": 176692
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.3198492527008057,
      "learning_rate": 7.691783363889904e-05,
      "loss": 1.9246,
      "step": 176693
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.2889357805252075,
      "learning_rate": 7.691509864142863e-05,
      "loss": 2.1635,
      "step": 176694
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.338073968887329,
      "learning_rate": 7.69123636854338e-05,
      "loss": 2.0302,
      "step": 176695
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.3287798166275024,
      "learning_rate": 7.690962877091521e-05,
      "loss": 2.0215,
      "step": 176696
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.1966012716293335,
      "learning_rate": 7.690689389787328e-05,
      "loss": 1.8631,
      "step": 176697
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.239960789680481,
      "learning_rate": 7.690415906630864e-05,
      "loss": 1.9709,
      "step": 176698
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.4337520599365234,
      "learning_rate": 7.690142427622165e-05,
      "loss": 2.2536,
      "step": 176699
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.264405608177185,
      "learning_rate": 7.689868952761305e-05,
      "loss": 2.127,
      "step": 176700
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.2201274633407593,
      "learning_rate": 7.689595482048303e-05,
      "loss": 1.8411,
      "step": 176701
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.3117706775665283,
      "learning_rate": 7.689322015483236e-05,
      "loss": 2.1282,
      "step": 176702
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.6726471185684204,
      "learning_rate": 7.689048553066136e-05,
      "loss": 2.212,
      "step": 176703
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.2117235660552979,
      "learning_rate": 7.68877509479707e-05,
      "loss": 1.7135,
      "step": 176704
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.3740195035934448,
      "learning_rate": 7.688501640676074e-05,
      "loss": 2.1095,
      "step": 176705
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.3276681900024414,
      "learning_rate": 7.688228190703225e-05,
      "loss": 2.0724,
      "step": 176706
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.5778437852859497,
      "learning_rate": 7.687954744878538e-05,
      "loss": 1.9886,
      "step": 176707
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.485366940498352,
      "learning_rate": 7.68768130320209e-05,
      "loss": 2.1377,
      "step": 176708
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.3987929821014404,
      "learning_rate": 7.687407865673917e-05,
      "loss": 2.1299,
      "step": 176709
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.3063369989395142,
      "learning_rate": 7.687134432294082e-05,
      "loss": 2.0721,
      "step": 176710
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.296787977218628,
      "learning_rate": 7.686861003062622e-05,
      "loss": 1.7973,
      "step": 176711
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.2794420719146729,
      "learning_rate": 7.686587577979603e-05,
      "loss": 2.0725,
      "step": 176712
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.5079913139343262,
      "learning_rate": 7.68631415704507e-05,
      "loss": 1.9197,
      "step": 176713
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.4910681247711182,
      "learning_rate": 7.686040740259071e-05,
      "loss": 2.0888,
      "step": 176714
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.2663947343826294,
      "learning_rate": 7.685767327621652e-05,
      "loss": 2.059,
      "step": 176715
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.2557086944580078,
      "learning_rate": 7.685493919132876e-05,
      "loss": 2.097,
      "step": 176716
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.4621548652648926,
      "learning_rate": 7.685220514792781e-05,
      "loss": 1.9752,
      "step": 176717
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.3087966442108154,
      "learning_rate": 7.684947114601432e-05,
      "loss": 1.791,
      "step": 176718
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.2805057764053345,
      "learning_rate": 7.684673718558874e-05,
      "loss": 2.066,
      "step": 176719
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.4530984163284302,
      "learning_rate": 7.684400326665146e-05,
      "loss": 1.9694,
      "step": 176720
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.288185477256775,
      "learning_rate": 7.684126938920321e-05,
      "loss": 2.029,
      "step": 176721
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.573371171951294,
      "learning_rate": 7.683853555324437e-05,
      "loss": 2.1577,
      "step": 176722
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.6859149932861328,
      "learning_rate": 7.683580175877533e-05,
      "loss": 1.966,
      "step": 176723
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.3067244291305542,
      "learning_rate": 7.683306800579688e-05,
      "loss": 2.163,
      "step": 176724
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.3555914163589478,
      "learning_rate": 7.683033429430932e-05,
      "loss": 2.0526,
      "step": 176725
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.2311288118362427,
      "learning_rate": 7.682760062431316e-05,
      "loss": 1.9125,
      "step": 176726
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.2603974342346191,
      "learning_rate": 7.682486699580903e-05,
      "loss": 1.955,
      "step": 176727
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.3012486696243286,
      "learning_rate": 7.682213340879739e-05,
      "loss": 2.1559,
      "step": 176728
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.3531215190887451,
      "learning_rate": 7.681939986327861e-05,
      "loss": 2.0001,
      "step": 176729
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.423676609992981,
      "learning_rate": 7.681666635925341e-05,
      "loss": 2.2043,
      "step": 176730
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.2744836807250977,
      "learning_rate": 7.681393289672222e-05,
      "loss": 1.7902,
      "step": 176731
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.3439888954162598,
      "learning_rate": 7.681119947568543e-05,
      "loss": 2.0336,
      "step": 176732
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.662390112876892,
      "learning_rate": 7.680846609614376e-05,
      "loss": 1.995,
      "step": 176733
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.3648759126663208,
      "learning_rate": 7.680573275809753e-05,
      "loss": 2.1011,
      "step": 176734
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.2358044385910034,
      "learning_rate": 7.680299946154739e-05,
      "loss": 2.1319,
      "step": 176735
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.2404087781906128,
      "learning_rate": 7.680026620649378e-05,
      "loss": 1.9765,
      "step": 176736
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.3820244073867798,
      "learning_rate": 7.679753299293722e-05,
      "loss": 1.973,
      "step": 176737
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.5807857513427734,
      "learning_rate": 7.679479982087814e-05,
      "loss": 1.8829,
      "step": 176738
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.440467119216919,
      "learning_rate": 7.679206669031722e-05,
      "loss": 1.9534,
      "step": 176739
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.3494497537612915,
      "learning_rate": 7.678933360125475e-05,
      "loss": 2.2056,
      "step": 176740
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.179509162902832,
      "learning_rate": 7.678660055369145e-05,
      "loss": 2.179,
      "step": 176741
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.322206735610962,
      "learning_rate": 7.678386754762772e-05,
      "loss": 1.9383,
      "step": 176742
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.4301261901855469,
      "learning_rate": 7.678113458306413e-05,
      "loss": 1.8653,
      "step": 176743
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.6765048503875732,
      "learning_rate": 7.677840166000101e-05,
      "loss": 2.2435,
      "step": 176744
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.2499713897705078,
      "learning_rate": 7.677566877843909e-05,
      "loss": 2.3696,
      "step": 176745
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.6675373315811157,
      "learning_rate": 7.677293593837871e-05,
      "loss": 2.0166,
      "step": 176746
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.3147622346878052,
      "learning_rate": 7.677020313982053e-05,
      "loss": 2.0076,
      "step": 176747
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.3392802476882935,
      "learning_rate": 7.676747038276501e-05,
      "loss": 2.1429,
      "step": 176748
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.268669843673706,
      "learning_rate": 7.67647376672126e-05,
      "loss": 1.9146,
      "step": 176749
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.2498382329940796,
      "learning_rate": 7.67620049931638e-05,
      "loss": 1.9245,
      "step": 176750
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.3246209621429443,
      "learning_rate": 7.675927236061918e-05,
      "loss": 2.0445,
      "step": 176751
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.3180031776428223,
      "learning_rate": 7.675653976957917e-05,
      "loss": 1.9212,
      "step": 176752
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.2444195747375488,
      "learning_rate": 7.675380722004443e-05,
      "loss": 2.16,
      "step": 176753
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.3312492370605469,
      "learning_rate": 7.675107471201535e-05,
      "loss": 2.1172,
      "step": 176754
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.3172595500946045,
      "learning_rate": 7.67483422454925e-05,
      "loss": 2.0032,
      "step": 176755
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.3112152814865112,
      "learning_rate": 7.674560982047622e-05,
      "loss": 2.0345,
      "step": 176756
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.3464523553848267,
      "learning_rate": 7.674287743696723e-05,
      "loss": 2.1619,
      "step": 176757
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.34456467628479,
      "learning_rate": 7.674014509496588e-05,
      "loss": 2.2045,
      "step": 176758
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.201844334602356,
      "learning_rate": 7.673741279447285e-05,
      "loss": 1.8841,
      "step": 176759
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.2480982542037964,
      "learning_rate": 7.673468053548846e-05,
      "loss": 1.9495,
      "step": 176760
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.3138178586959839,
      "learning_rate": 7.673194831801345e-05,
      "loss": 1.9841,
      "step": 176761
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.2827483415603638,
      "learning_rate": 7.672921614204806e-05,
      "loss": 2.0148,
      "step": 176762
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.1855047941207886,
      "learning_rate": 7.672648400759295e-05,
      "loss": 2.0367,
      "step": 176763
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.2806951999664307,
      "learning_rate": 7.672375191464856e-05,
      "loss": 2.0184,
      "step": 176764
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.2809183597564697,
      "learning_rate": 7.672101986321551e-05,
      "loss": 2.0791,
      "step": 176765
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.317994236946106,
      "learning_rate": 7.671828785329416e-05,
      "loss": 2.1021,
      "step": 176766
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.329723834991455,
      "learning_rate": 7.671555588488526e-05,
      "loss": 1.941,
      "step": 176767
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.1994798183441162,
      "learning_rate": 7.671282395798899e-05,
      "loss": 2.0112,
      "step": 176768
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.3049439191818237,
      "learning_rate": 7.671009207260608e-05,
      "loss": 2.2302,
      "step": 176769
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.2053419351577759,
      "learning_rate": 7.670736022873692e-05,
      "loss": 1.8376,
      "step": 176770
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.2171932458877563,
      "learning_rate": 7.670462842638217e-05,
      "loss": 1.967,
      "step": 176771
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.291178822517395,
      "learning_rate": 7.670189666554213e-05,
      "loss": 1.7702,
      "step": 176772
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.3693163394927979,
      "learning_rate": 7.669916494621762e-05,
      "loss": 2.211,
      "step": 176773
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.4489094018936157,
      "learning_rate": 7.669643326840877e-05,
      "loss": 1.9378,
      "step": 176774
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.4186272621154785,
      "learning_rate": 7.669370163211633e-05,
      "loss": 2.0997,
      "step": 176775
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.2221759557724,
      "learning_rate": 7.66909700373407e-05,
      "loss": 2.0512,
      "step": 176776
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.5452663898468018,
      "learning_rate": 7.668823848408248e-05,
      "loss": 2.0848,
      "step": 176777
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.5172061920166016,
      "learning_rate": 7.668550697234208e-05,
      "loss": 1.9981,
      "step": 176778
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.324878454208374,
      "learning_rate": 7.668277550212021e-05,
      "loss": 1.9344,
      "step": 176779
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.3964639902114868,
      "learning_rate": 7.668004407341705e-05,
      "loss": 2.1629,
      "step": 176780
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.274072527885437,
      "learning_rate": 7.667731268623339e-05,
      "loss": 2.0923,
      "step": 176781
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.3415106534957886,
      "learning_rate": 7.667458134056955e-05,
      "loss": 2.047,
      "step": 176782
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.3029980659484863,
      "learning_rate": 7.66718500364262e-05,
      "loss": 2.1316,
      "step": 176783
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.3370754718780518,
      "learning_rate": 7.666911877380369e-05,
      "loss": 2.1002,
      "step": 176784
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.3344621658325195,
      "learning_rate": 7.666638755270278e-05,
      "loss": 1.8509,
      "step": 176785
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.3128114938735962,
      "learning_rate": 7.666365637312361e-05,
      "loss": 2.0118,
      "step": 176786
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.3227146863937378,
      "learning_rate": 7.666092523506698e-05,
      "loss": 1.8512,
      "step": 176787
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.3366869688034058,
      "learning_rate": 7.665819413853321e-05,
      "loss": 1.9304,
      "step": 176788
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.3381365537643433,
      "learning_rate": 7.6655463083523e-05,
      "loss": 2.0872,
      "step": 176789
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.3658689260482788,
      "learning_rate": 7.665273207003665e-05,
      "loss": 1.9772,
      "step": 176790
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.3215786218643188,
      "learning_rate": 7.665000109807494e-05,
      "loss": 2.1887,
      "step": 176791
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.3472470045089722,
      "learning_rate": 7.664727016763806e-05,
      "loss": 1.9141,
      "step": 176792
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.352759838104248,
      "learning_rate": 7.664453927872674e-05,
      "loss": 1.979,
      "step": 176793
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.3059285879135132,
      "learning_rate": 7.664180843134136e-05,
      "loss": 2.2054,
      "step": 176794
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.1641672849655151,
      "learning_rate": 7.663907762548255e-05,
      "loss": 1.8048,
      "step": 176795
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.3049837350845337,
      "learning_rate": 7.663634686115066e-05,
      "loss": 1.8779,
      "step": 176796
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.4071083068847656,
      "learning_rate": 7.663361613834638e-05,
      "loss": 1.9215,
      "step": 176797
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.360454797744751,
      "learning_rate": 7.663088545707011e-05,
      "loss": 2.1733,
      "step": 176798
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.3236584663391113,
      "learning_rate": 7.66281548173224e-05,
      "loss": 1.9267,
      "step": 176799
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.314746379852295,
      "learning_rate": 7.662542421910363e-05,
      "loss": 2.0389,
      "step": 176800
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.249983549118042,
      "learning_rate": 7.662269366241451e-05,
      "loss": 1.8646,
      "step": 176801
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.32746160030365,
      "learning_rate": 7.661996314725533e-05,
      "loss": 2.1626,
      "step": 176802
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.2450754642486572,
      "learning_rate": 7.661723267362683e-05,
      "loss": 2.0025,
      "step": 176803
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.2822166681289673,
      "learning_rate": 7.661450224152941e-05,
      "loss": 2.1303,
      "step": 176804
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.3990839719772339,
      "learning_rate": 7.661177185096357e-05,
      "loss": 2.3358,
      "step": 176805
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.4255036115646362,
      "learning_rate": 7.66090415019297e-05,
      "loss": 1.8467,
      "step": 176806
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.3538541793823242,
      "learning_rate": 7.660631119442855e-05,
      "loss": 2.1163,
      "step": 176807
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.3586355447769165,
      "learning_rate": 7.660358092846038e-05,
      "loss": 2.0896,
      "step": 176808
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.3370435237884521,
      "learning_rate": 7.660085070402595e-05,
      "loss": 1.9334,
      "step": 176809
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.2757763862609863,
      "learning_rate": 7.659812052112562e-05,
      "loss": 1.9707,
      "step": 176810
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.3549679517745972,
      "learning_rate": 7.65953903797598e-05,
      "loss": 2.0239,
      "step": 176811
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.34518301486969,
      "learning_rate": 7.659266027992923e-05,
      "loss": 1.9692,
      "step": 176812
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.3736611604690552,
      "learning_rate": 7.65899302216343e-05,
      "loss": 1.9867,
      "step": 176813
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.237783432006836,
      "learning_rate": 7.658720020487544e-05,
      "loss": 2.01,
      "step": 176814
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.355055332183838,
      "learning_rate": 7.658447022965332e-05,
      "loss": 2.0144,
      "step": 176815
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.3294779062271118,
      "learning_rate": 7.658174029596836e-05,
      "loss": 2.1519,
      "step": 176816
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.3825489282608032,
      "learning_rate": 7.657901040382098e-05,
      "loss": 2.014,
      "step": 176817
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.3089288473129272,
      "learning_rate": 7.65762805532119e-05,
      "loss": 1.917,
      "step": 176818
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.2568771839141846,
      "learning_rate": 7.657355074414136e-05,
      "loss": 2.0676,
      "step": 176819
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.38785982131958,
      "learning_rate": 7.657082097661015e-05,
      "loss": 2.2158,
      "step": 176820
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.1337991952896118,
      "learning_rate": 7.65680912506186e-05,
      "loss": 1.7887,
      "step": 176821
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.252579927444458,
      "learning_rate": 7.656536156616729e-05,
      "loss": 2.0772,
      "step": 176822
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.2995668649673462,
      "learning_rate": 7.656263192325662e-05,
      "loss": 1.9711,
      "step": 176823
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.3893815279006958,
      "learning_rate": 7.655990232188722e-05,
      "loss": 2.0376,
      "step": 176824
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.3608735799789429,
      "learning_rate": 7.655717276205949e-05,
      "loss": 2.042,
      "step": 176825
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.3192319869995117,
      "learning_rate": 7.65544432437741e-05,
      "loss": 1.8537,
      "step": 176826
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.2866758108139038,
      "learning_rate": 7.655171376703145e-05,
      "loss": 1.9227,
      "step": 176827
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.227785587310791,
      "learning_rate": 7.654898433183205e-05,
      "loss": 2.1273,
      "step": 176828
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.326981544494629,
      "learning_rate": 7.65462549381763e-05,
      "loss": 2.0666,
      "step": 176829
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.277001142501831,
      "learning_rate": 7.654352558606495e-05,
      "loss": 2.1598,
      "step": 176830
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.2984122037887573,
      "learning_rate": 7.654079627549827e-05,
      "loss": 1.8689,
      "step": 176831
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.4252692461013794,
      "learning_rate": 7.653806700647695e-05,
      "loss": 2.1764,
      "step": 176832
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.253517508506775,
      "learning_rate": 7.653533777900133e-05,
      "loss": 2.0703,
      "step": 176833
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.295842170715332,
      "learning_rate": 7.653260859307221e-05,
      "loss": 2.1328,
      "step": 176834
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.2718334197998047,
      "learning_rate": 7.652987944868972e-05,
      "loss": 2.0045,
      "step": 176835
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.2908638715744019,
      "learning_rate": 7.65271503458546e-05,
      "loss": 2.0496,
      "step": 176836
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.445473074913025,
      "learning_rate": 7.652442128456725e-05,
      "loss": 2.1565,
      "step": 176837
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.3504527807235718,
      "learning_rate": 7.652169226482829e-05,
      "loss": 2.0313,
      "step": 176838
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.4473518133163452,
      "learning_rate": 7.65189632866381e-05,
      "loss": 2.0321,
      "step": 176839
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.3422691822052002,
      "learning_rate": 7.651623434999741e-05,
      "loss": 2.0657,
      "step": 176840
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.3658288717269897,
      "learning_rate": 7.651350545490641e-05,
      "loss": 1.9189,
      "step": 176841
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.333596110343933,
      "learning_rate": 7.651077660136584e-05,
      "loss": 2.1284,
      "step": 176842
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.353316307067871,
      "learning_rate": 7.650804778937603e-05,
      "loss": 1.9945,
      "step": 176843
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.3580632209777832,
      "learning_rate": 7.650531901893773e-05,
      "loss": 1.8722,
      "step": 176844
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.3811200857162476,
      "learning_rate": 7.650259029005118e-05,
      "loss": 1.7537,
      "step": 176845
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.4407604932785034,
      "learning_rate": 7.649986160271723e-05,
      "loss": 1.7368,
      "step": 176846
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.4678627252578735,
      "learning_rate": 7.649713295693597e-05,
      "loss": 1.9829,
      "step": 176847
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.32597815990448,
      "learning_rate": 7.649440435270821e-05,
      "loss": 2.1126,
      "step": 176848
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.3061487674713135,
      "learning_rate": 7.649167579003427e-05,
      "loss": 1.8421,
      "step": 176849
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.2227898836135864,
      "learning_rate": 7.648894726891486e-05,
      "loss": 1.9892,
      "step": 176850
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.3145819902420044,
      "learning_rate": 7.648621878935024e-05,
      "loss": 1.9995,
      "step": 176851
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.3642860651016235,
      "learning_rate": 7.648349035134124e-05,
      "loss": 2.0433,
      "step": 176852
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.201104998588562,
      "learning_rate": 7.648076195488802e-05,
      "loss": 1.9917,
      "step": 176853
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.3057225942611694,
      "learning_rate": 7.647803359999134e-05,
      "loss": 2.0353,
      "step": 176854
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.6042364835739136,
      "learning_rate": 7.647530528665148e-05,
      "loss": 2.2321,
      "step": 176855
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.4544343948364258,
      "learning_rate": 7.647257701486919e-05,
      "loss": 2.1415,
      "step": 176856
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.252996802330017,
      "learning_rate": 7.64698487846448e-05,
      "loss": 1.9975,
      "step": 176857
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.2390573024749756,
      "learning_rate": 7.646712059597904e-05,
      "loss": 1.9022,
      "step": 176858
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.3711761236190796,
      "learning_rate": 7.64643924488721e-05,
      "loss": 1.9964,
      "step": 176859
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.290588617324829,
      "learning_rate": 7.646166434332472e-05,
      "loss": 2.1034,
      "step": 176860
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.2193487882614136,
      "learning_rate": 7.645893627933728e-05,
      "loss": 2.045,
      "step": 176861
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.2965583801269531,
      "learning_rate": 7.645620825691039e-05,
      "loss": 2.2131,
      "step": 176862
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.294116735458374,
      "learning_rate": 7.645348027604446e-05,
      "loss": 1.8351,
      "step": 176863
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.2874488830566406,
      "learning_rate": 7.64507523367402e-05,
      "loss": 1.9379,
      "step": 176864
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.2592092752456665,
      "learning_rate": 7.644802443899778e-05,
      "loss": 2.0494,
      "step": 176865
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.3390592336654663,
      "learning_rate": 7.6445296582818e-05,
      "loss": 1.897,
      "step": 176866
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.3106592893600464,
      "learning_rate": 7.644256876820118e-05,
      "loss": 1.8855,
      "step": 176867
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.358062744140625,
      "learning_rate": 7.643984099514798e-05,
      "loss": 2.1463,
      "step": 176868
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.2810933589935303,
      "learning_rate": 7.643711326365873e-05,
      "loss": 1.9405,
      "step": 176869
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.430001974105835,
      "learning_rate": 7.643438557373423e-05,
      "loss": 1.8432,
      "step": 176870
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.3936903476715088,
      "learning_rate": 7.643165792537466e-05,
      "loss": 1.8424,
      "step": 176871
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.2919602394104004,
      "learning_rate": 7.64289303185807e-05,
      "loss": 1.8442,
      "step": 176872
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.4267020225524902,
      "learning_rate": 7.642620275335275e-05,
      "loss": 2.0998,
      "step": 176873
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.4358632564544678,
      "learning_rate": 7.642347522969149e-05,
      "loss": 1.9854,
      "step": 176874
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.3414602279663086,
      "learning_rate": 7.642074774759722e-05,
      "loss": 2.199,
      "step": 176875
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.3069981336593628,
      "learning_rate": 7.641802030707075e-05,
      "loss": 2.0336,
      "step": 176876
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.32976496219635,
      "learning_rate": 7.641529290811217e-05,
      "loss": 2.0883,
      "step": 176877
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.2851338386535645,
      "learning_rate": 7.641256555072234e-05,
      "loss": 2.0717,
      "step": 176878
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.258592963218689,
      "learning_rate": 7.640983823490153e-05,
      "loss": 2.0418,
      "step": 176879
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.2627100944519043,
      "learning_rate": 7.640711096065045e-05,
      "loss": 1.998,
      "step": 176880
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.3886419534683228,
      "learning_rate": 7.64043837279694e-05,
      "loss": 1.92,
      "step": 176881
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.4154456853866577,
      "learning_rate": 7.640165653685909e-05,
      "loss": 1.9656,
      "step": 176882
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.3370453119277954,
      "learning_rate": 7.639892938731996e-05,
      "loss": 2.1647,
      "step": 176883
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.2052042484283447,
      "learning_rate": 7.639620227935248e-05,
      "loss": 2.123,
      "step": 176884
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.3506656885147095,
      "learning_rate": 7.639347521295705e-05,
      "loss": 2.2192,
      "step": 176885
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.3028517961502075,
      "learning_rate": 7.639074818813442e-05,
      "loss": 2.306,
      "step": 176886
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.2874430418014526,
      "learning_rate": 7.638802120488484e-05,
      "loss": 2.1245,
      "step": 176887
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.2827061414718628,
      "learning_rate": 7.638529426320908e-05,
      "loss": 2.0851,
      "step": 176888
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.234331488609314,
      "learning_rate": 7.638256736310748e-05,
      "loss": 2.0021,
      "step": 176889
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.396295189857483,
      "learning_rate": 7.63798405045806e-05,
      "loss": 1.9738,
      "step": 176890
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.3028134107589722,
      "learning_rate": 7.637711368762883e-05,
      "loss": 2.1187,
      "step": 176891
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.223366141319275,
      "learning_rate": 7.637438691225287e-05,
      "loss": 1.7171,
      "step": 176892
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.4034655094146729,
      "learning_rate": 7.637166017845305e-05,
      "loss": 1.945,
      "step": 176893
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.3832910060882568,
      "learning_rate": 7.636893348623004e-05,
      "loss": 2.1324,
      "step": 176894
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.3394105434417725,
      "learning_rate": 7.636620683558427e-05,
      "loss": 2.0354,
      "step": 176895
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.381370186805725,
      "learning_rate": 7.636348022651615e-05,
      "loss": 2.2096,
      "step": 176896
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.1627131700515747,
      "learning_rate": 7.63607536590264e-05,
      "loss": 1.9379,
      "step": 176897
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.2685319185256958,
      "learning_rate": 7.635802713311535e-05,
      "loss": 1.9737,
      "step": 176898
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.5816526412963867,
      "learning_rate": 7.635530064878351e-05,
      "loss": 2.2176,
      "step": 176899
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.4037600755691528,
      "learning_rate": 7.635257420603155e-05,
      "loss": 1.9778,
      "step": 176900
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.215376377105713,
      "learning_rate": 7.634984780485981e-05,
      "loss": 1.8812,
      "step": 176901
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.170387864112854,
      "learning_rate": 7.634712144526883e-05,
      "loss": 2.0778,
      "step": 176902
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.269985318183899,
      "learning_rate": 7.634439512725919e-05,
      "loss": 1.865,
      "step": 176903
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.2122372388839722,
      "learning_rate": 7.634166885083138e-05,
      "loss": 2.0333,
      "step": 176904
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.2402485609054565,
      "learning_rate": 7.633894261598577e-05,
      "loss": 1.8648,
      "step": 176905
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.3040357828140259,
      "learning_rate": 7.633621642272303e-05,
      "loss": 2.0289,
      "step": 176906
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.2907077074050903,
      "learning_rate": 7.633349027104367e-05,
      "loss": 1.9359,
      "step": 176907
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.2664988040924072,
      "learning_rate": 7.6330764160948e-05,
      "loss": 2.0947,
      "step": 176908
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.3512600660324097,
      "learning_rate": 7.632803809243678e-05,
      "loss": 1.9335,
      "step": 176909
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.3834877014160156,
      "learning_rate": 7.632531206551029e-05,
      "loss": 2.1878,
      "step": 176910
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.3851876258850098,
      "learning_rate": 7.632258608016922e-05,
      "loss": 2.1211,
      "step": 176911
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.2589879035949707,
      "learning_rate": 7.631986013641404e-05,
      "loss": 1.9629,
      "step": 176912
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.258602261543274,
      "learning_rate": 7.631713423424521e-05,
      "loss": 2.0611,
      "step": 176913
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.2987496852874756,
      "learning_rate": 7.631440837366316e-05,
      "loss": 1.9934,
      "step": 176914
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.3614997863769531,
      "learning_rate": 7.631168255466857e-05,
      "loss": 2.0352,
      "step": 176915
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.2664886713027954,
      "learning_rate": 7.630895677726175e-05,
      "loss": 1.9475,
      "step": 176916
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.5015935897827148,
      "learning_rate": 7.630623104144343e-05,
      "loss": 2.1473,
      "step": 176917
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.423593521118164,
      "learning_rate": 7.630350534721391e-05,
      "loss": 2.092,
      "step": 176918
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.3432363271713257,
      "learning_rate": 7.630077969457399e-05,
      "loss": 1.9914,
      "step": 176919
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.2935672998428345,
      "learning_rate": 7.629805408352376e-05,
      "loss": 1.9708,
      "step": 176920
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.3056926727294922,
      "learning_rate": 7.629532851406407e-05,
      "loss": 1.9995,
      "step": 176921
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.3018115758895874,
      "learning_rate": 7.62926029861952e-05,
      "loss": 2.044,
      "step": 176922
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.277450680732727,
      "learning_rate": 7.628987749991784e-05,
      "loss": 1.924,
      "step": 176923
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.2449617385864258,
      "learning_rate": 7.628715205523233e-05,
      "loss": 2.0801,
      "step": 176924
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.395022988319397,
      "learning_rate": 7.628442665213943e-05,
      "loss": 2.0636,
      "step": 176925
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.2745634317398071,
      "learning_rate": 7.628170129063931e-05,
      "loss": 2.2508,
      "step": 176926
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.397374153137207,
      "learning_rate": 7.627897597073274e-05,
      "loss": 1.9742,
      "step": 176927
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.4263864755630493,
      "learning_rate": 7.627625069242004e-05,
      "loss": 2.0884,
      "step": 176928
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.2655149698257446,
      "learning_rate": 7.627352545570189e-05,
      "loss": 2.0199,
      "step": 176929
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.3138700723648071,
      "learning_rate": 7.627080026057863e-05,
      "loss": 2.0901,
      "step": 176930
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.244835615158081,
      "learning_rate": 7.626807510705101e-05,
      "loss": 2.1727,
      "step": 176931
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.4771560430526733,
      "learning_rate": 7.626534999511925e-05,
      "loss": 2.0492,
      "step": 176932
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.36404550075531,
      "learning_rate": 7.626262492478402e-05,
      "loss": 2.1411,
      "step": 176933
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.382724642753601,
      "learning_rate": 7.625989989604573e-05,
      "loss": 2.1091,
      "step": 176934
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.3115166425704956,
      "learning_rate": 7.625717490890503e-05,
      "loss": 2.2563,
      "step": 176935
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.445015549659729,
      "learning_rate": 7.625444996336226e-05,
      "loss": 1.9786,
      "step": 176936
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.2497785091400146,
      "learning_rate": 7.625172505941817e-05,
      "loss": 1.9213,
      "step": 176937
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.2628746032714844,
      "learning_rate": 7.624900019707294e-05,
      "loss": 1.9513,
      "step": 176938
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.2210123538970947,
      "learning_rate": 7.624627537632733e-05,
      "loss": 1.9194,
      "step": 176939
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.4500746726989746,
      "learning_rate": 7.62435505971817e-05,
      "loss": 2.0419,
      "step": 176940
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.239048957824707,
      "learning_rate": 7.624082585963666e-05,
      "loss": 2.0171,
      "step": 176941
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.337544560432434,
      "learning_rate": 7.623810116369263e-05,
      "loss": 1.9831,
      "step": 176942
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.4471501111984253,
      "learning_rate": 7.623537650935033e-05,
      "loss": 2.0206,
      "step": 176943
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.323927402496338,
      "learning_rate": 7.623265189660989e-05,
      "loss": 2.0695,
      "step": 176944
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.3578981161117554,
      "learning_rate": 7.622992732547217e-05,
      "loss": 2.047,
      "step": 176945
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.4715664386749268,
      "learning_rate": 7.622720279593738e-05,
      "loss": 2.1821,
      "step": 176946
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.3841626644134521,
      "learning_rate": 7.622447830800629e-05,
      "loss": 2.0425,
      "step": 176947
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.3235357999801636,
      "learning_rate": 7.622175386167922e-05,
      "loss": 2.1095,
      "step": 176948
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.2415331602096558,
      "learning_rate": 7.621902945695689e-05,
      "loss": 1.8725,
      "step": 176949
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.2561115026474,
      "learning_rate": 7.621630509383952e-05,
      "loss": 2.3296,
      "step": 176950
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.2982529401779175,
      "learning_rate": 7.621358077232784e-05,
      "loss": 1.7802,
      "step": 176951
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.4411481618881226,
      "learning_rate": 7.621085649242221e-05,
      "loss": 2.1051,
      "step": 176952
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.3168401718139648,
      "learning_rate": 7.620813225412326e-05,
      "loss": 1.9127,
      "step": 176953
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.2808291912078857,
      "learning_rate": 7.620540805743141e-05,
      "loss": 1.9229,
      "step": 176954
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.2832897901535034,
      "learning_rate": 7.620268390234732e-05,
      "loss": 2.0969,
      "step": 176955
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.3152223825454712,
      "learning_rate": 7.619995978887123e-05,
      "loss": 2.0206,
      "step": 176956
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.4203519821166992,
      "learning_rate": 7.619723571700386e-05,
      "loss": 2.1476,
      "step": 176957
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.2361650466918945,
      "learning_rate": 7.619451168674557e-05,
      "loss": 1.859,
      "step": 176958
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.3115676641464233,
      "learning_rate": 7.619178769809702e-05,
      "loss": 1.9036,
      "step": 176959
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.2015312910079956,
      "learning_rate": 7.618906375105859e-05,
      "loss": 2.0174,
      "step": 176960
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.4899158477783203,
      "learning_rate": 7.61863398456309e-05,
      "loss": 2.0688,
      "step": 176961
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.3768690824508667,
      "learning_rate": 7.618361598181437e-05,
      "loss": 2.0315,
      "step": 176962
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.2577022314071655,
      "learning_rate": 7.618089215960958e-05,
      "loss": 1.9556,
      "step": 176963
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.321458339691162,
      "learning_rate": 7.617816837901688e-05,
      "loss": 1.9696,
      "step": 176964
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.2647920846939087,
      "learning_rate": 7.617544464003696e-05,
      "loss": 2.0701,
      "step": 176965
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.3517159223556519,
      "learning_rate": 7.617272094267017e-05,
      "loss": 1.9909,
      "step": 176966
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.3069708347320557,
      "learning_rate": 7.616999728691717e-05,
      "loss": 2.0893,
      "step": 176967
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.2653372287750244,
      "learning_rate": 7.616727367277841e-05,
      "loss": 1.8673,
      "step": 176968
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.347273826599121,
      "learning_rate": 7.616455010025437e-05,
      "loss": 2.1097,
      "step": 176969
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.38625967502594,
      "learning_rate": 7.616182656934547e-05,
      "loss": 2.1434,
      "step": 176970
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.2745569944381714,
      "learning_rate": 7.615910308005242e-05,
      "loss": 2.0027,
      "step": 176971
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.329681396484375,
      "learning_rate": 7.615637963237554e-05,
      "loss": 2.0634,
      "step": 176972
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.208311915397644,
      "learning_rate": 7.615365622631548e-05,
      "loss": 1.9355,
      "step": 176973
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.1530382633209229,
      "learning_rate": 7.615093286187268e-05,
      "loss": 2.1319,
      "step": 176974
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.3446851968765259,
      "learning_rate": 7.614820953904767e-05,
      "loss": 2.1921,
      "step": 176975
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.2742276191711426,
      "learning_rate": 7.614548625784082e-05,
      "loss": 1.9954,
      "step": 176976
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.543194055557251,
      "learning_rate": 7.614276301825283e-05,
      "loss": 2.1404,
      "step": 176977
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.3871252536773682,
      "learning_rate": 7.614003982028407e-05,
      "loss": 1.9438,
      "step": 176978
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.3189644813537598,
      "learning_rate": 7.613731666393515e-05,
      "loss": 1.8327,
      "step": 176979
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.4026963710784912,
      "learning_rate": 7.613459354920658e-05,
      "loss": 2.1516,
      "step": 176980
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.547452449798584,
      "learning_rate": 7.613187047609868e-05,
      "loss": 2.0073,
      "step": 176981
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.353546142578125,
      "learning_rate": 7.612914744461221e-05,
      "loss": 2.011,
      "step": 176982
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.3918309211730957,
      "learning_rate": 7.612642445474752e-05,
      "loss": 1.7096,
      "step": 176983
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.243029236793518,
      "learning_rate": 7.612370150650511e-05,
      "loss": 1.9135,
      "step": 176984
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.6813366413116455,
      "learning_rate": 7.61209785998856e-05,
      "loss": 2.1227,
      "step": 176985
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.3941187858581543,
      "learning_rate": 7.611825573488942e-05,
      "loss": 1.9688,
      "step": 176986
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.3303171396255493,
      "learning_rate": 7.611553291151702e-05,
      "loss": 1.9408,
      "step": 176987
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.278679370880127,
      "learning_rate": 7.611281012976902e-05,
      "loss": 2.0434,
      "step": 176988
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.4187328815460205,
      "learning_rate": 7.61100873896459e-05,
      "loss": 1.8317,
      "step": 176989
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.2487367391586304,
      "learning_rate": 7.610736469114803e-05,
      "loss": 2.031,
      "step": 176990
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.3149421215057373,
      "learning_rate": 7.610464203427614e-05,
      "loss": 1.6917,
      "step": 176991
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.2432199716567993,
      "learning_rate": 7.610191941903062e-05,
      "loss": 2.1928,
      "step": 176992
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.5775201320648193,
      "learning_rate": 7.609919684541187e-05,
      "loss": 2.1119,
      "step": 176993
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.4080111980438232,
      "learning_rate": 7.609647431342062e-05,
      "loss": 2.1313,
      "step": 176994
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.587727427482605,
      "learning_rate": 7.609375182305713e-05,
      "loss": 1.9941,
      "step": 176995
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.3877310752868652,
      "learning_rate": 7.609102937432219e-05,
      "loss": 1.8814,
      "step": 176996
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.4650509357452393,
      "learning_rate": 7.608830696721611e-05,
      "loss": 1.8659,
      "step": 176997
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.6220935583114624,
      "learning_rate": 7.608558460173945e-05,
      "loss": 1.8642,
      "step": 176998
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.4164445400238037,
      "learning_rate": 7.608286227789261e-05,
      "loss": 1.8906,
      "step": 176999
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.6104892492294312,
      "learning_rate": 7.60801399956763e-05,
      "loss": 2.1059,
      "step": 177000
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.3814321756362915,
      "learning_rate": 7.607741775509084e-05,
      "loss": 2.216,
      "step": 177001
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.394897699356079,
      "learning_rate": 7.607469555613689e-05,
      "loss": 2.1158,
      "step": 177002
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.452216386795044,
      "learning_rate": 7.607197339881477e-05,
      "loss": 1.9043,
      "step": 177003
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.347496509552002,
      "learning_rate": 7.606925128312531e-05,
      "loss": 1.9587,
      "step": 177004
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.475784182548523,
      "learning_rate": 7.606652920906859e-05,
      "loss": 1.7629,
      "step": 177005
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.2495903968811035,
      "learning_rate": 7.606380717664545e-05,
      "loss": 2.0377,
      "step": 177006
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.7536816596984863,
      "learning_rate": 7.606108518585615e-05,
      "loss": 2.0584,
      "step": 177007
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.449949026107788,
      "learning_rate": 7.605836323670144e-05,
      "loss": 2.1026,
      "step": 177008
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.3284891843795776,
      "learning_rate": 7.60556413291816e-05,
      "loss": 1.993,
      "step": 177009
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.5116969347000122,
      "learning_rate": 7.605291946329746e-05,
      "loss": 2.0622,
      "step": 177010
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.2547534704208374,
      "learning_rate": 7.605019763904911e-05,
      "loss": 1.9885,
      "step": 177011
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.4103749990463257,
      "learning_rate": 7.604747585643732e-05,
      "loss": 2.1311,
      "step": 177012
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.3920371532440186,
      "learning_rate": 7.604475411546248e-05,
      "loss": 2.1195,
      "step": 177013
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.5342849493026733,
      "learning_rate": 7.60420324161252e-05,
      "loss": 1.8473,
      "step": 177014
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.498872995376587,
      "learning_rate": 7.603931075842586e-05,
      "loss": 2.1222,
      "step": 177015
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.3734219074249268,
      "learning_rate": 7.603658914236524e-05,
      "loss": 2.3505,
      "step": 177016
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.197248935699463,
      "learning_rate": 7.603386756794345e-05,
      "loss": 1.9034,
      "step": 177017
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.2614123821258545,
      "learning_rate": 7.603114603516129e-05,
      "loss": 2.0389,
      "step": 177018
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.6465747356414795,
      "learning_rate": 7.602842454401904e-05,
      "loss": 1.9138,
      "step": 177019
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.6680444478988647,
      "learning_rate": 7.602570309451745e-05,
      "loss": 2.0905,
      "step": 177020
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.2327526807785034,
      "learning_rate": 7.602298168665682e-05,
      "loss": 2.1698,
      "step": 177021
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.3000965118408203,
      "learning_rate": 7.602026032043792e-05,
      "loss": 1.8872,
      "step": 177022
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.2687896490097046,
      "learning_rate": 7.601753899586094e-05,
      "loss": 2.0983,
      "step": 177023
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.3692677021026611,
      "learning_rate": 7.601481771292658e-05,
      "loss": 2.1754,
      "step": 177024
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.4880729913711548,
      "learning_rate": 7.601209647163521e-05,
      "loss": 2.0042,
      "step": 177025
    },
    {
      "epoch": 2.3,
      "grad_norm": 1.5386425256729126,
      "learning_rate": 7.60093752719875e-05,
      "loss": 1.9506,
      "step": 177026
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.4809588193893433,
      "learning_rate": 7.600665411398381e-05,
      "loss": 2.1375,
      "step": 177027
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.354441523551941,
      "learning_rate": 7.600393299762477e-05,
      "loss": 2.0201,
      "step": 177028
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.3151546716690063,
      "learning_rate": 7.600121192291084e-05,
      "loss": 2.1168,
      "step": 177029
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.526468276977539,
      "learning_rate": 7.599849088984251e-05,
      "loss": 1.9366,
      "step": 177030
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.2792271375656128,
      "learning_rate": 7.59957698984202e-05,
      "loss": 2.0381,
      "step": 177031
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.4247106313705444,
      "learning_rate": 7.59930489486446e-05,
      "loss": 2.1991,
      "step": 177032
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.3173233270645142,
      "learning_rate": 7.599032804051604e-05,
      "loss": 2.0705,
      "step": 177033
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.2828658819198608,
      "learning_rate": 7.598760717403519e-05,
      "loss": 2.1129,
      "step": 177034
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.3039902448654175,
      "learning_rate": 7.598488634920244e-05,
      "loss": 2.1694,
      "step": 177035
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.4777405261993408,
      "learning_rate": 7.598216556601837e-05,
      "loss": 1.8187,
      "step": 177036
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.346182942390442,
      "learning_rate": 7.597944482448336e-05,
      "loss": 2.1399,
      "step": 177037
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.3758624792099,
      "learning_rate": 7.597672412459804e-05,
      "loss": 2.1443,
      "step": 177038
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.4787715673446655,
      "learning_rate": 7.597400346636285e-05,
      "loss": 2.0811,
      "step": 177039
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.2565243244171143,
      "learning_rate": 7.597128284977836e-05,
      "loss": 1.9666,
      "step": 177040
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.2183623313903809,
      "learning_rate": 7.596856227484505e-05,
      "loss": 2.0629,
      "step": 177041
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.272845983505249,
      "learning_rate": 7.596584174156344e-05,
      "loss": 1.9673,
      "step": 177042
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.2929229736328125,
      "learning_rate": 7.596312124993387e-05,
      "loss": 2.2328,
      "step": 177043
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.3165212869644165,
      "learning_rate": 7.59604007999571e-05,
      "loss": 2.1652,
      "step": 177044
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.2708629369735718,
      "learning_rate": 7.595768039163343e-05,
      "loss": 1.8476,
      "step": 177045
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.31477952003479,
      "learning_rate": 7.595496002496358e-05,
      "loss": 2.0744,
      "step": 177046
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.270282506942749,
      "learning_rate": 7.595223969994789e-05,
      "loss": 1.9201,
      "step": 177047
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.44324791431427,
      "learning_rate": 7.594951941658692e-05,
      "loss": 1.9039,
      "step": 177048
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.39485764503479,
      "learning_rate": 7.594679917488106e-05,
      "loss": 2.2376,
      "step": 177049
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.258230209350586,
      "learning_rate": 7.594407897483102e-05,
      "loss": 2.0239,
      "step": 177050
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.203104019165039,
      "learning_rate": 7.594135881643713e-05,
      "loss": 2.0225,
      "step": 177051
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.2956582307815552,
      "learning_rate": 7.593863869970004e-05,
      "loss": 2.1523,
      "step": 177052
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.3194165229797363,
      "learning_rate": 7.593591862462019e-05,
      "loss": 2.1898,
      "step": 177053
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.3981879949569702,
      "learning_rate": 7.59331985911981e-05,
      "loss": 1.975,
      "step": 177054
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.3072001934051514,
      "learning_rate": 7.593047859943415e-05,
      "loss": 2.0697,
      "step": 177055
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.347306489944458,
      "learning_rate": 7.592775864932906e-05,
      "loss": 2.056,
      "step": 177056
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.1951762437820435,
      "learning_rate": 7.592503874088313e-05,
      "loss": 2.0689,
      "step": 177057
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.4184105396270752,
      "learning_rate": 7.592231887409703e-05,
      "loss": 2.0523,
      "step": 177058
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.2642756700515747,
      "learning_rate": 7.591959904897125e-05,
      "loss": 2.1081,
      "step": 177059
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.2735153436660767,
      "learning_rate": 7.59168792655062e-05,
      "loss": 2.0796,
      "step": 177060
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.407320261001587,
      "learning_rate": 7.591415952370237e-05,
      "loss": 2.0059,
      "step": 177061
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.2067513465881348,
      "learning_rate": 7.591143982356041e-05,
      "loss": 1.717,
      "step": 177062
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.3493447303771973,
      "learning_rate": 7.590872016508065e-05,
      "loss": 1.9546,
      "step": 177063
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.3384106159210205,
      "learning_rate": 7.590600054826377e-05,
      "loss": 1.9292,
      "step": 177064
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.4843374490737915,
      "learning_rate": 7.590328097311022e-05,
      "loss": 1.8533,
      "step": 177065
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.321235179901123,
      "learning_rate": 7.590056143962038e-05,
      "loss": 2.0036,
      "step": 177066
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.2799183130264282,
      "learning_rate": 7.589784194779493e-05,
      "loss": 1.9244,
      "step": 177067
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.2969800233840942,
      "learning_rate": 7.589512249763433e-05,
      "loss": 1.9863,
      "step": 177068
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.2812141180038452,
      "learning_rate": 7.589240308913894e-05,
      "loss": 1.9337,
      "step": 177069
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.282909631729126,
      "learning_rate": 7.588968372230949e-05,
      "loss": 2.0486,
      "step": 177070
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.2694859504699707,
      "learning_rate": 7.588696439714638e-05,
      "loss": 2.2608,
      "step": 177071
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.3439282178878784,
      "learning_rate": 7.588424511365e-05,
      "loss": 2.0363,
      "step": 177072
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.3549336194992065,
      "learning_rate": 7.588152587182109e-05,
      "loss": 1.8804,
      "step": 177073
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.3009073734283447,
      "learning_rate": 7.587880667166002e-05,
      "loss": 2.0399,
      "step": 177074
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.508646011352539,
      "learning_rate": 7.587608751316722e-05,
      "loss": 2.049,
      "step": 177075
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.2594071626663208,
      "learning_rate": 7.587336839634335e-05,
      "loss": 1.9068,
      "step": 177076
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.3394216299057007,
      "learning_rate": 7.587064932118888e-05,
      "loss": 2.007,
      "step": 177077
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.3809574842453003,
      "learning_rate": 7.586793028770422e-05,
      "loss": 2.0415,
      "step": 177078
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.407859444618225,
      "learning_rate": 7.586521129588998e-05,
      "loss": 1.8667,
      "step": 177079
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.431254506111145,
      "learning_rate": 7.586249234574658e-05,
      "loss": 2.2385,
      "step": 177080
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.321140170097351,
      "learning_rate": 7.585977343727465e-05,
      "loss": 1.8333,
      "step": 177081
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.358778953552246,
      "learning_rate": 7.585705457047462e-05,
      "loss": 2.1173,
      "step": 177082
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.4180926084518433,
      "learning_rate": 7.585433574534698e-05,
      "loss": 2.0173,
      "step": 177083
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.3696017265319824,
      "learning_rate": 7.585161696189218e-05,
      "loss": 1.8294,
      "step": 177084
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.3646810054779053,
      "learning_rate": 7.584889822011088e-05,
      "loss": 1.9358,
      "step": 177085
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.347403645515442,
      "learning_rate": 7.58461795200034e-05,
      "loss": 1.9048,
      "step": 177086
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.5380055904388428,
      "learning_rate": 7.584346086157046e-05,
      "loss": 2.0171,
      "step": 177087
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.323799967765808,
      "learning_rate": 7.584074224481245e-05,
      "loss": 1.8297,
      "step": 177088
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.2027617692947388,
      "learning_rate": 7.583802366972986e-05,
      "loss": 2.2434,
      "step": 177089
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.3912147283554077,
      "learning_rate": 7.58353051363231e-05,
      "loss": 2.1237,
      "step": 177090
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.302337884902954,
      "learning_rate": 7.583258664459292e-05,
      "loss": 2.0426,
      "step": 177091
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.1835776567459106,
      "learning_rate": 7.582986819453958e-05,
      "loss": 2.1879,
      "step": 177092
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.2780247926712036,
      "learning_rate": 7.582714978616382e-05,
      "loss": 2.0036,
      "step": 177093
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.2483654022216797,
      "learning_rate": 7.582443141946592e-05,
      "loss": 2.1109,
      "step": 177094
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.181994080543518,
      "learning_rate": 7.582171309444657e-05,
      "loss": 1.7756,
      "step": 177095
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.3235787153244019,
      "learning_rate": 7.581899481110618e-05,
      "loss": 2.0575,
      "step": 177096
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.252220630645752,
      "learning_rate": 7.581627656944529e-05,
      "loss": 1.9987,
      "step": 177097
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.3829151391983032,
      "learning_rate": 7.581355836946429e-05,
      "loss": 1.7706,
      "step": 177098
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.3017419576644897,
      "learning_rate": 7.581084021116388e-05,
      "loss": 2.1057,
      "step": 177099
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.2947827577590942,
      "learning_rate": 7.58081220945444e-05,
      "loss": 1.9632,
      "step": 177100
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.7097684144973755,
      "learning_rate": 7.580540401960648e-05,
      "loss": 2.2365,
      "step": 177101
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.393376111984253,
      "learning_rate": 7.580268598635057e-05,
      "loss": 2.1065,
      "step": 177102
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.3150418996810913,
      "learning_rate": 7.579996799477717e-05,
      "loss": 2.0212,
      "step": 177103
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.5956852436065674,
      "learning_rate": 7.579725004488671e-05,
      "loss": 1.9324,
      "step": 177104
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.348145842552185,
      "learning_rate": 7.579453213667985e-05,
      "loss": 1.881,
      "step": 177105
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.4047399759292603,
      "learning_rate": 7.579181427015695e-05,
      "loss": 1.7728,
      "step": 177106
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.3894429206848145,
      "learning_rate": 7.578909644531863e-05,
      "loss": 2.1613,
      "step": 177107
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.4075090885162354,
      "learning_rate": 7.578637866216541e-05,
      "loss": 2.0927,
      "step": 177108
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.555113434791565,
      "learning_rate": 7.57836609206977e-05,
      "loss": 1.9208,
      "step": 177109
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.3314902782440186,
      "learning_rate": 7.578094322091595e-05,
      "loss": 2.0322,
      "step": 177110
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.499822735786438,
      "learning_rate": 7.577822556282085e-05,
      "loss": 1.9449,
      "step": 177111
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.2599575519561768,
      "learning_rate": 7.57755079464127e-05,
      "loss": 2.066,
      "step": 177112
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.338107943534851,
      "learning_rate": 7.577279037169221e-05,
      "loss": 2.2689,
      "step": 177113
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.2949423789978027,
      "learning_rate": 7.577007283865981e-05,
      "loss": 2.2099,
      "step": 177114
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.276218056678772,
      "learning_rate": 7.576735534731599e-05,
      "loss": 1.8232,
      "step": 177115
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.3235300779342651,
      "learning_rate": 7.576463789766117e-05,
      "loss": 1.9352,
      "step": 177116
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.2864289283752441,
      "learning_rate": 7.576192048969598e-05,
      "loss": 2.1156,
      "step": 177117
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.3661017417907715,
      "learning_rate": 7.575920312342082e-05,
      "loss": 1.8687,
      "step": 177118
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.363024353981018,
      "learning_rate": 7.575648579883638e-05,
      "loss": 1.9204,
      "step": 177119
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.3405797481536865,
      "learning_rate": 7.575376851594299e-05,
      "loss": 1.9027,
      "step": 177120
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.4501632452011108,
      "learning_rate": 7.575105127474123e-05,
      "loss": 2.2181,
      "step": 177121
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.3007668256759644,
      "learning_rate": 7.574833407523147e-05,
      "loss": 2.1398,
      "step": 177122
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.3677401542663574,
      "learning_rate": 7.574561691741447e-05,
      "loss": 2.1619,
      "step": 177123
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.2757928371429443,
      "learning_rate": 7.574289980129046e-05,
      "loss": 1.9962,
      "step": 177124
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.413331389427185,
      "learning_rate": 7.574018272686019e-05,
      "loss": 2.1695,
      "step": 177125
    },
    {
      "epoch": 2.31,
      "grad_norm": 2.1562371253967285,
      "learning_rate": 7.573746569412402e-05,
      "loss": 2.0153,
      "step": 177126
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.4058982133865356,
      "learning_rate": 7.57347487030825e-05,
      "loss": 2.1151,
      "step": 177127
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.3319844007492065,
      "learning_rate": 7.573203175373606e-05,
      "loss": 1.9196,
      "step": 177128
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.4622312784194946,
      "learning_rate": 7.572931484608534e-05,
      "loss": 2.1472,
      "step": 177129
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.343943476676941,
      "learning_rate": 7.57265979801307e-05,
      "loss": 1.9905,
      "step": 177130
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.851302146911621,
      "learning_rate": 7.57238811558728e-05,
      "loss": 2.021,
      "step": 177131
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.3283666372299194,
      "learning_rate": 7.572116437331204e-05,
      "loss": 2.1703,
      "step": 177132
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.4000579118728638,
      "learning_rate": 7.571844763244898e-05,
      "loss": 1.9581,
      "step": 177133
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.2855849266052246,
      "learning_rate": 7.571573093328398e-05,
      "loss": 1.8437,
      "step": 177134
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.4424010515213013,
      "learning_rate": 7.571301427581776e-05,
      "loss": 2.0188,
      "step": 177135
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.262719988822937,
      "learning_rate": 7.571029766005065e-05,
      "loss": 2.0002,
      "step": 177136
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.450261116027832,
      "learning_rate": 7.57075810859833e-05,
      "loss": 2.0421,
      "step": 177137
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.5112266540527344,
      "learning_rate": 7.570486455361614e-05,
      "loss": 2.1582,
      "step": 177138
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.422998309135437,
      "learning_rate": 7.570214806294969e-05,
      "loss": 2.0237,
      "step": 177139
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.285423755645752,
      "learning_rate": 7.569943161398436e-05,
      "loss": 1.9703,
      "step": 177140
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.278298258781433,
      "learning_rate": 7.569671520672083e-05,
      "loss": 2.0413,
      "step": 177141
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.355115294456482,
      "learning_rate": 7.569399884115942e-05,
      "loss": 2.0964,
      "step": 177142
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.2161269187927246,
      "learning_rate": 7.56912825173008e-05,
      "loss": 1.876,
      "step": 177143
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.3973006010055542,
      "learning_rate": 7.568856623514544e-05,
      "loss": 2.2054,
      "step": 177144
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.4230070114135742,
      "learning_rate": 7.568584999469382e-05,
      "loss": 1.9713,
      "step": 177145
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.2787268161773682,
      "learning_rate": 7.568313379594629e-05,
      "loss": 2.0362,
      "step": 177146
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.388380527496338,
      "learning_rate": 7.568041763890361e-05,
      "loss": 1.8264,
      "step": 177147
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.2278848886489868,
      "learning_rate": 7.56777015235661e-05,
      "loss": 1.9256,
      "step": 177148
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.4268602132797241,
      "learning_rate": 7.567498544993443e-05,
      "loss": 1.9408,
      "step": 177149
    },
    {
      "epoch": 2.31,
      "grad_norm": 2.1035120487213135,
      "learning_rate": 7.5672269418009e-05,
      "loss": 2.1094,
      "step": 177150
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.2371083498001099,
      "learning_rate": 7.566955342779034e-05,
      "loss": 2.1088,
      "step": 177151
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.1971644163131714,
      "learning_rate": 7.566683747927885e-05,
      "loss": 1.8802,
      "step": 177152
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.2952898740768433,
      "learning_rate": 7.566412157247523e-05,
      "loss": 2.0722,
      "step": 177153
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.4368737936019897,
      "learning_rate": 7.566140570737978e-05,
      "loss": 2.042,
      "step": 177154
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.2958859205245972,
      "learning_rate": 7.565868988399321e-05,
      "loss": 1.9692,
      "step": 177155
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.2542515993118286,
      "learning_rate": 7.565597410231593e-05,
      "loss": 2.1053,
      "step": 177156
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.2595934867858887,
      "learning_rate": 7.565325836234832e-05,
      "loss": 2.1307,
      "step": 177157
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.215964436531067,
      "learning_rate": 7.565054266409112e-05,
      "loss": 1.9872,
      "step": 177158
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.4132695198059082,
      "learning_rate": 7.564782700754471e-05,
      "loss": 2.2815,
      "step": 177159
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.4805673360824585,
      "learning_rate": 7.564511139270951e-05,
      "loss": 2.0237,
      "step": 177160
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.315880537033081,
      "learning_rate": 7.56423958195862e-05,
      "loss": 2.1451,
      "step": 177161
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.2708433866500854,
      "learning_rate": 7.563968028817514e-05,
      "loss": 1.8976,
      "step": 177162
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.3017888069152832,
      "learning_rate": 7.563696479847697e-05,
      "loss": 2.1322,
      "step": 177163
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.2568467855453491,
      "learning_rate": 7.563424935049213e-05,
      "loss": 2.0437,
      "step": 177164
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.446077585220337,
      "learning_rate": 7.563153394422104e-05,
      "loss": 1.9642,
      "step": 177165
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.346820592880249,
      "learning_rate": 7.562881857966437e-05,
      "loss": 2.0964,
      "step": 177166
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.2386889457702637,
      "learning_rate": 7.562610325682252e-05,
      "loss": 2.0934,
      "step": 177167
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.232784628868103,
      "learning_rate": 7.562338797569595e-05,
      "loss": 1.8701,
      "step": 177168
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.2637966871261597,
      "learning_rate": 7.562067273628527e-05,
      "loss": 2.0916,
      "step": 177169
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.638916015625,
      "learning_rate": 7.561795753859098e-05,
      "loss": 1.8648,
      "step": 177170
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.2747352123260498,
      "learning_rate": 7.561524238261344e-05,
      "loss": 2.1731,
      "step": 177171
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.3671497106552124,
      "learning_rate": 7.561252726835339e-05,
      "loss": 1.9159,
      "step": 177172
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.1730681657791138,
      "learning_rate": 7.560981219581116e-05,
      "loss": 2.0233,
      "step": 177173
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.40201735496521,
      "learning_rate": 7.560709716498723e-05,
      "loss": 2.0071,
      "step": 177174
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.4436275959014893,
      "learning_rate": 7.560438217588227e-05,
      "loss": 2.2044,
      "step": 177175
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.3357373476028442,
      "learning_rate": 7.56016672284967e-05,
      "loss": 2.0725,
      "step": 177176
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.2813855409622192,
      "learning_rate": 7.559895232283091e-05,
      "loss": 2.0604,
      "step": 177177
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.318475604057312,
      "learning_rate": 7.559623745888559e-05,
      "loss": 2.2148,
      "step": 177178
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.3267987966537476,
      "learning_rate": 7.55935226366611e-05,
      "loss": 2.196,
      "step": 177179
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.2408201694488525,
      "learning_rate": 7.55908078561581e-05,
      "loss": 1.9794,
      "step": 177180
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.2631614208221436,
      "learning_rate": 7.558809311737699e-05,
      "loss": 2.0391,
      "step": 177181
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.3862693309783936,
      "learning_rate": 7.558537842031828e-05,
      "loss": 1.8314,
      "step": 177182
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.3664730787277222,
      "learning_rate": 7.558266376498243e-05,
      "loss": 2.0786,
      "step": 177183
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.2281842231750488,
      "learning_rate": 7.557994915137005e-05,
      "loss": 1.8851,
      "step": 177184
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.3488783836364746,
      "learning_rate": 7.557723457948152e-05,
      "loss": 1.9536,
      "step": 177185
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.3556452989578247,
      "learning_rate": 7.55745200493175e-05,
      "loss": 2.0315,
      "step": 177186
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.2990622520446777,
      "learning_rate": 7.557180556087845e-05,
      "loss": 2.0947,
      "step": 177187
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.1578537225723267,
      "learning_rate": 7.556909111416476e-05,
      "loss": 1.9653,
      "step": 177188
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.4416512250900269,
      "learning_rate": 7.5566376709177e-05,
      "loss": 2.0937,
      "step": 177189
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.6524810791015625,
      "learning_rate": 7.556366234591573e-05,
      "loss": 2.0254,
      "step": 177190
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.2345855236053467,
      "learning_rate": 7.556094802438136e-05,
      "loss": 2.0423,
      "step": 177191
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.3202403783798218,
      "learning_rate": 7.555823374457452e-05,
      "loss": 2.0921,
      "step": 177192
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.324294924736023,
      "learning_rate": 7.555551950649562e-05,
      "loss": 2.0559,
      "step": 177193
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.3474799394607544,
      "learning_rate": 7.555280531014521e-05,
      "loss": 1.9008,
      "step": 177194
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.5449172258377075,
      "learning_rate": 7.55500911555237e-05,
      "loss": 2.14,
      "step": 177195
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.3078652620315552,
      "learning_rate": 7.554737704263174e-05,
      "loss": 1.8795,
      "step": 177196
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.3957520723342896,
      "learning_rate": 7.554466297146965e-05,
      "loss": 1.772,
      "step": 177197
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.2593194246292114,
      "learning_rate": 7.554194894203812e-05,
      "loss": 2.142,
      "step": 177198
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.3159455060958862,
      "learning_rate": 7.553923495433761e-05,
      "loss": 1.9953,
      "step": 177199
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.2725231647491455,
      "learning_rate": 7.553652100836859e-05,
      "loss": 2.2097,
      "step": 177200
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.3107337951660156,
      "learning_rate": 7.553380710413145e-05,
      "loss": 2.1571,
      "step": 177201
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.3253657817840576,
      "learning_rate": 7.553109324162692e-05,
      "loss": 2.0263,
      "step": 177202
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.2893593311309814,
      "learning_rate": 7.552837942085533e-05,
      "loss": 2.0378,
      "step": 177203
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.312751293182373,
      "learning_rate": 7.552566564181733e-05,
      "loss": 2.1798,
      "step": 177204
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.6779779195785522,
      "learning_rate": 7.552295190451333e-05,
      "loss": 2.0197,
      "step": 177205
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.248832106590271,
      "learning_rate": 7.552023820894387e-05,
      "loss": 1.9908,
      "step": 177206
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.2552284002304077,
      "learning_rate": 7.551752455510933e-05,
      "loss": 2.0837,
      "step": 177207
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.330806016921997,
      "learning_rate": 7.551481094301041e-05,
      "loss": 2.0766,
      "step": 177208
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.3276530504226685,
      "learning_rate": 7.551209737264743e-05,
      "loss": 1.9386,
      "step": 177209
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.302769660949707,
      "learning_rate": 7.55093838440211e-05,
      "loss": 2.1171,
      "step": 177210
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.2281990051269531,
      "learning_rate": 7.550667035713182e-05,
      "loss": 1.8903,
      "step": 177211
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.2999879121780396,
      "learning_rate": 7.550395691198005e-05,
      "loss": 2.1798,
      "step": 177212
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.3059172630310059,
      "learning_rate": 7.550124350856625e-05,
      "loss": 2.0466,
      "step": 177213
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.1934767961502075,
      "learning_rate": 7.549853014689111e-05,
      "loss": 1.9664,
      "step": 177214
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.381996512413025,
      "learning_rate": 7.549581682695495e-05,
      "loss": 2.0341,
      "step": 177215
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.275679349899292,
      "learning_rate": 7.549310354875843e-05,
      "loss": 2.1103,
      "step": 177216
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.4731240272521973,
      "learning_rate": 7.549039031230198e-05,
      "loss": 1.92,
      "step": 177217
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.2326786518096924,
      "learning_rate": 7.548767711758613e-05,
      "loss": 2.0215,
      "step": 177218
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.431748390197754,
      "learning_rate": 7.548496396461123e-05,
      "loss": 1.8865,
      "step": 177219
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.3677164316177368,
      "learning_rate": 7.548225085337802e-05,
      "loss": 2.133,
      "step": 177220
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.2447090148925781,
      "learning_rate": 7.547953778388679e-05,
      "loss": 2.2497,
      "step": 177221
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.3844815492630005,
      "learning_rate": 7.547682475613827e-05,
      "loss": 2.1415,
      "step": 177222
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.5169707536697388,
      "learning_rate": 7.547411177013276e-05,
      "loss": 2.2692,
      "step": 177223
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.3161979913711548,
      "learning_rate": 7.5471398825871e-05,
      "loss": 2.0642,
      "step": 177224
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.2437857389450073,
      "learning_rate": 7.546868592335318e-05,
      "loss": 2.2267,
      "step": 177225
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.2404627799987793,
      "learning_rate": 7.54659730625801e-05,
      "loss": 2.0391,
      "step": 177226
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.2855831384658813,
      "learning_rate": 7.546326024355197e-05,
      "loss": 1.786,
      "step": 177227
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.3073129653930664,
      "learning_rate": 7.54605474662696e-05,
      "loss": 2.0079,
      "step": 177228
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.312784194946289,
      "learning_rate": 7.545783473073326e-05,
      "loss": 1.7676,
      "step": 177229
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.5192320346832275,
      "learning_rate": 7.545512203694369e-05,
      "loss": 1.9834,
      "step": 177230
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.319478988647461,
      "learning_rate": 7.54524093849011e-05,
      "loss": 2.1102,
      "step": 177231
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.4293599128723145,
      "learning_rate": 7.544969677460622e-05,
      "loss": 2.0347,
      "step": 177232
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.3333306312561035,
      "learning_rate": 7.54469842060594e-05,
      "loss": 2.2197,
      "step": 177233
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.3165030479431152,
      "learning_rate": 7.544427167926133e-05,
      "loss": 1.8987,
      "step": 177234
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.3783866167068481,
      "learning_rate": 7.544155919421232e-05,
      "loss": 2.1669,
      "step": 177235
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.453018069267273,
      "learning_rate": 7.543884675091311e-05,
      "loss": 2.0737,
      "step": 177236
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.328477382659912,
      "learning_rate": 7.543613434936388e-05,
      "loss": 1.9641,
      "step": 177237
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.3988703489303589,
      "learning_rate": 7.543342198956542e-05,
      "loss": 2.0957,
      "step": 177238
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.2593401670455933,
      "learning_rate": 7.543070967151803e-05,
      "loss": 2.0047,
      "step": 177239
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.4131523370742798,
      "learning_rate": 7.542799739522241e-05,
      "loss": 2.0975,
      "step": 177240
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.2873042821884155,
      "learning_rate": 7.542528516067889e-05,
      "loss": 1.9777,
      "step": 177241
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.4190703630447388,
      "learning_rate": 7.54225729678881e-05,
      "loss": 2.0858,
      "step": 177242
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.4923001527786255,
      "learning_rate": 7.54198608168505e-05,
      "loss": 2.2578,
      "step": 177243
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.2860819101333618,
      "learning_rate": 7.541714870756659e-05,
      "loss": 2.0556,
      "step": 177244
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.2887827157974243,
      "learning_rate": 7.54144366400368e-05,
      "loss": 2.0571,
      "step": 177245
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.4365023374557495,
      "learning_rate": 7.541172461426178e-05,
      "loss": 2.1664,
      "step": 177246
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.4150243997573853,
      "learning_rate": 7.54090126302419e-05,
      "loss": 2.0072,
      "step": 177247
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.2982014417648315,
      "learning_rate": 7.540630068797777e-05,
      "loss": 2.3146,
      "step": 177248
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.3656573295593262,
      "learning_rate": 7.540358878746985e-05,
      "loss": 1.8092,
      "step": 177249
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.267698884010315,
      "learning_rate": 7.540087692871866e-05,
      "loss": 2.1863,
      "step": 177250
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.3212096691131592,
      "learning_rate": 7.53981651117246e-05,
      "loss": 2.0474,
      "step": 177251
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.3492156267166138,
      "learning_rate": 7.539545333648834e-05,
      "loss": 1.9744,
      "step": 177252
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.3207699060440063,
      "learning_rate": 7.539274160301023e-05,
      "loss": 2.0722,
      "step": 177253
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.5067845582962036,
      "learning_rate": 7.539002991129093e-05,
      "loss": 2.3282,
      "step": 177254
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.3297340869903564,
      "learning_rate": 7.538731826133085e-05,
      "loss": 2.0952,
      "step": 177255
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.4074313640594482,
      "learning_rate": 7.538460665313044e-05,
      "loss": 2.0263,
      "step": 177256
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.3735045194625854,
      "learning_rate": 7.538189508669036e-05,
      "loss": 2.0494,
      "step": 177257
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.3467568159103394,
      "learning_rate": 7.537918356201101e-05,
      "loss": 2.245,
      "step": 177258
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.3965351581573486,
      "learning_rate": 7.537647207909281e-05,
      "loss": 1.9875,
      "step": 177259
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.370938777923584,
      "learning_rate": 7.537376063793648e-05,
      "loss": 2.2612,
      "step": 177260
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.214356780052185,
      "learning_rate": 7.537104923854241e-05,
      "loss": 2.0153,
      "step": 177261
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.2341830730438232,
      "learning_rate": 7.5368337880911e-05,
      "loss": 1.9207,
      "step": 177262
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.2349563837051392,
      "learning_rate": 7.536562656504294e-05,
      "loss": 2.0148,
      "step": 177263
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.34080970287323,
      "learning_rate": 7.536291529093857e-05,
      "loss": 1.9847,
      "step": 177264
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.2804701328277588,
      "learning_rate": 7.536020405859853e-05,
      "loss": 1.6922,
      "step": 177265
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.2988872528076172,
      "learning_rate": 7.535749286802333e-05,
      "loss": 1.8503,
      "step": 177266
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.2781931161880493,
      "learning_rate": 7.535478171921338e-05,
      "loss": 2.2683,
      "step": 177267
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.3896892070770264,
      "learning_rate": 7.535207061216911e-05,
      "loss": 2.189,
      "step": 177268
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.2668596506118774,
      "learning_rate": 7.534935954689124e-05,
      "loss": 2.0705,
      "step": 177269
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.3354713916778564,
      "learning_rate": 7.534664852338009e-05,
      "loss": 2.2183,
      "step": 177270
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.2846258878707886,
      "learning_rate": 7.53439375416363e-05,
      "loss": 2.131,
      "step": 177271
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.2656694650650024,
      "learning_rate": 7.534122660166034e-05,
      "loss": 2.1562,
      "step": 177272
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.2523294687271118,
      "learning_rate": 7.533851570345268e-05,
      "loss": 1.9727,
      "step": 177273
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.6056084632873535,
      "learning_rate": 7.533580484701373e-05,
      "loss": 2.0395,
      "step": 177274
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.3417339324951172,
      "learning_rate": 7.533309403234419e-05,
      "loss": 2.0542,
      "step": 177275
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.2675883769989014,
      "learning_rate": 7.533038325944438e-05,
      "loss": 2.0617,
      "step": 177276
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.4254049062728882,
      "learning_rate": 7.532767252831497e-05,
      "loss": 2.0503,
      "step": 177277
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.2312008142471313,
      "learning_rate": 7.53249618389564e-05,
      "loss": 2.0092,
      "step": 177278
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.2986969947814941,
      "learning_rate": 7.532225119136915e-05,
      "loss": 1.9492,
      "step": 177279
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.3015516996383667,
      "learning_rate": 7.531954058555364e-05,
      "loss": 1.7953,
      "step": 177280
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.281239628791809,
      "learning_rate": 7.531683002151058e-05,
      "loss": 1.8742,
      "step": 177281
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.3473337888717651,
      "learning_rate": 7.531411949924026e-05,
      "loss": 2.1698,
      "step": 177282
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.301195740699768,
      "learning_rate": 7.531140901874338e-05,
      "loss": 1.9152,
      "step": 177283
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.3121559619903564,
      "learning_rate": 7.530869858002035e-05,
      "loss": 1.8534,
      "step": 177284
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.2878731489181519,
      "learning_rate": 7.530598818307166e-05,
      "loss": 1.8614,
      "step": 177285
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.4158626794815063,
      "learning_rate": 7.530327782789777e-05,
      "loss": 2.0238,
      "step": 177286
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.5212740898132324,
      "learning_rate": 7.530056751449932e-05,
      "loss": 1.7874,
      "step": 177287
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.3863309621810913,
      "learning_rate": 7.529785724287663e-05,
      "loss": 1.9892,
      "step": 177288
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.3258261680603027,
      "learning_rate": 7.529514701303041e-05,
      "loss": 2.0034,
      "step": 177289
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.3964130878448486,
      "learning_rate": 7.529243682496098e-05,
      "loss": 1.9806,
      "step": 177290
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.2811050415039062,
      "learning_rate": 7.528972667866909e-05,
      "loss": 2.2994,
      "step": 177291
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.4906165599822998,
      "learning_rate": 7.528701657415493e-05,
      "loss": 1.7572,
      "step": 177292
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.313693642616272,
      "learning_rate": 7.528430651141924e-05,
      "loss": 2.0533,
      "step": 177293
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.3051397800445557,
      "learning_rate": 7.528159649046233e-05,
      "loss": 2.0706,
      "step": 177294
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.4607752561569214,
      "learning_rate": 7.527888651128489e-05,
      "loss": 2.1486,
      "step": 177295
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.3105405569076538,
      "learning_rate": 7.527617657388728e-05,
      "loss": 2.1053,
      "step": 177296
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.1929367780685425,
      "learning_rate": 7.527346667827024e-05,
      "loss": 2.0488,
      "step": 177297
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.282649278640747,
      "learning_rate": 7.527075682443395e-05,
      "loss": 2.2294,
      "step": 177298
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.2316821813583374,
      "learning_rate": 7.526804701237914e-05,
      "loss": 1.9079,
      "step": 177299
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.3424540758132935,
      "learning_rate": 7.526533724210617e-05,
      "loss": 2.2124,
      "step": 177300
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.4045865535736084,
      "learning_rate": 7.526262751361568e-05,
      "loss": 2.0276,
      "step": 177301
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.3048505783081055,
      "learning_rate": 7.525991782690803e-05,
      "loss": 2.0184,
      "step": 177302
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.3307876586914062,
      "learning_rate": 7.525720818198397e-05,
      "loss": 2.0765,
      "step": 177303
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.2091857194900513,
      "learning_rate": 7.525449857884366e-05,
      "loss": 2.0154,
      "step": 177304
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.2805057764053345,
      "learning_rate": 7.525178901748789e-05,
      "loss": 1.9759,
      "step": 177305
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.2664366960525513,
      "learning_rate": 7.524907949791698e-05,
      "loss": 1.9727,
      "step": 177306
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.3592748641967773,
      "learning_rate": 7.524637002013158e-05,
      "loss": 2.1612,
      "step": 177307
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.3778560161590576,
      "learning_rate": 7.524366058413202e-05,
      "loss": 1.9027,
      "step": 177308
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.2722135782241821,
      "learning_rate": 7.52409511899191e-05,
      "loss": 1.9692,
      "step": 177309
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.4878336191177368,
      "learning_rate": 7.523824183749292e-05,
      "loss": 1.8833,
      "step": 177310
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.3864789009094238,
      "learning_rate": 7.523553252685431e-05,
      "loss": 1.9079,
      "step": 177311
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.3792753219604492,
      "learning_rate": 7.523282325800358e-05,
      "loss": 2.1355,
      "step": 177312
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.3422324657440186,
      "learning_rate": 7.523011403094136e-05,
      "loss": 2.0784,
      "step": 177313
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.4171096086502075,
      "learning_rate": 7.522740484566803e-05,
      "loss": 1.8631,
      "step": 177314
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.3064424991607666,
      "learning_rate": 7.522469570218439e-05,
      "loss": 1.9107,
      "step": 177315
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.3688433170318604,
      "learning_rate": 7.522198660049052e-05,
      "loss": 1.8562,
      "step": 177316
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.3948596715927124,
      "learning_rate": 7.521927754058718e-05,
      "loss": 2.0106,
      "step": 177317
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.3423513174057007,
      "learning_rate": 7.521656852247478e-05,
      "loss": 1.7856,
      "step": 177318
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.2734239101409912,
      "learning_rate": 7.521385954615391e-05,
      "loss": 2.1443,
      "step": 177319
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.3288909196853638,
      "learning_rate": 7.521115061162495e-05,
      "loss": 2.0466,
      "step": 177320
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.3923401832580566,
      "learning_rate": 7.520844171888865e-05,
      "loss": 2.1959,
      "step": 177321
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.2165837287902832,
      "learning_rate": 7.52057328679452e-05,
      "loss": 2.0003,
      "step": 177322
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.3155008554458618,
      "learning_rate": 7.520302405879531e-05,
      "loss": 2.1114,
      "step": 177323
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.3471390008926392,
      "learning_rate": 7.520031529143933e-05,
      "loss": 2.1041,
      "step": 177324
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.484038233757019,
      "learning_rate": 7.519760656587797e-05,
      "loss": 1.9327,
      "step": 177325
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.4024732112884521,
      "learning_rate": 7.519489788211149e-05,
      "loss": 2.0267,
      "step": 177326
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.4497145414352417,
      "learning_rate": 7.519218924014064e-05,
      "loss": 1.9175,
      "step": 177327
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.517449975013733,
      "learning_rate": 7.518948063996583e-05,
      "loss": 1.9694,
      "step": 177328
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.5233500003814697,
      "learning_rate": 7.518677208158747e-05,
      "loss": 1.9461,
      "step": 177329
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.5192489624023438,
      "learning_rate": 7.518406356500609e-05,
      "loss": 2.0176,
      "step": 177330
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.3590701818466187,
      "learning_rate": 7.518135509022233e-05,
      "loss": 1.9912,
      "step": 177331
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.2544695138931274,
      "learning_rate": 7.517864665723648e-05,
      "loss": 1.9764,
      "step": 177332
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.193450927734375,
      "learning_rate": 7.51759382660493e-05,
      "loss": 1.8781,
      "step": 177333
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.1768678426742554,
      "learning_rate": 7.51732299166611e-05,
      "loss": 1.8856,
      "step": 177334
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.2146525382995605,
      "learning_rate": 7.517052160907248e-05,
      "loss": 2.1466,
      "step": 177335
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.3299129009246826,
      "learning_rate": 7.516781334328382e-05,
      "loss": 1.8539,
      "step": 177336
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.211647629737854,
      "learning_rate": 7.516510511929574e-05,
      "loss": 1.8993,
      "step": 177337
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.3018382787704468,
      "learning_rate": 7.516239693710867e-05,
      "loss": 2.0452,
      "step": 177338
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.4501359462738037,
      "learning_rate": 7.515968879672326e-05,
      "loss": 2.003,
      "step": 177339
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.251579999923706,
      "learning_rate": 7.515698069813988e-05,
      "loss": 1.9604,
      "step": 177340
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.3918921947479248,
      "learning_rate": 7.515427264135896e-05,
      "loss": 2.2364,
      "step": 177341
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.2179441452026367,
      "learning_rate": 7.51515646263812e-05,
      "loss": 1.8783,
      "step": 177342
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.8762024641036987,
      "learning_rate": 7.514885665320704e-05,
      "loss": 2.1366,
      "step": 177343
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.3446778059005737,
      "learning_rate": 7.514614872183684e-05,
      "loss": 2.0339,
      "step": 177344
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.352608323097229,
      "learning_rate": 7.514344083227129e-05,
      "loss": 1.9664,
      "step": 177345
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.4493684768676758,
      "learning_rate": 7.514073298451083e-05,
      "loss": 2.0123,
      "step": 177346
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.325461506843567,
      "learning_rate": 7.513802517855587e-05,
      "loss": 2.1083,
      "step": 177347
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.4055254459381104,
      "learning_rate": 7.51353174144071e-05,
      "loss": 2.0579,
      "step": 177348
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.5677787065505981,
      "learning_rate": 7.51326096920649e-05,
      "loss": 1.9643,
      "step": 177349
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.3230867385864258,
      "learning_rate": 7.512990201152971e-05,
      "loss": 2.2174,
      "step": 177350
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.5535154342651367,
      "learning_rate": 7.512719437280218e-05,
      "loss": 2.0492,
      "step": 177351
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.3508777618408203,
      "learning_rate": 7.512448677588278e-05,
      "loss": 2.0542,
      "step": 177352
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.371091604232788,
      "learning_rate": 7.512177922077187e-05,
      "loss": 1.8169,
      "step": 177353
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.4262402057647705,
      "learning_rate": 7.511907170747017e-05,
      "loss": 2.1912,
      "step": 177354
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.4675301313400269,
      "learning_rate": 7.5116364235978e-05,
      "loss": 1.9066,
      "step": 177355
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.2001670598983765,
      "learning_rate": 7.511365680629602e-05,
      "loss": 1.9031,
      "step": 177356
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.137941837310791,
      "learning_rate": 7.511094941842466e-05,
      "loss": 1.8226,
      "step": 177357
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.2718605995178223,
      "learning_rate": 7.510824207236439e-05,
      "loss": 2.0025,
      "step": 177358
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.6192207336425781,
      "learning_rate": 7.51055347681157e-05,
      "loss": 1.8795,
      "step": 177359
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.3480268716812134,
      "learning_rate": 7.510282750567921e-05,
      "loss": 2.0628,
      "step": 177360
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.3717020750045776,
      "learning_rate": 7.510012028505523e-05,
      "loss": 2.1125,
      "step": 177361
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.3815232515335083,
      "learning_rate": 7.509741310624451e-05,
      "loss": 2.011,
      "step": 177362
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.3372271060943604,
      "learning_rate": 7.509470596924733e-05,
      "loss": 2.1187,
      "step": 177363
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.3046784400939941,
      "learning_rate": 7.509199887406445e-05,
      "loss": 2.005,
      "step": 177364
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.4114749431610107,
      "learning_rate": 7.508929182069605e-05,
      "loss": 2.2314,
      "step": 177365
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.5927538871765137,
      "learning_rate": 7.50865848091429e-05,
      "loss": 2.0886,
      "step": 177366
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.2348419427871704,
      "learning_rate": 7.508387783940526e-05,
      "loss": 1.9811,
      "step": 177367
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.302729845046997,
      "learning_rate": 7.508117091148392e-05,
      "loss": 1.8603,
      "step": 177368
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.4703787565231323,
      "learning_rate": 7.507846402537913e-05,
      "loss": 2.0965,
      "step": 177369
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.2538073062896729,
      "learning_rate": 7.507575718109168e-05,
      "loss": 2.0379,
      "step": 177370
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.2833771705627441,
      "learning_rate": 7.507305037862168e-05,
      "loss": 1.9096,
      "step": 177371
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.2176117897033691,
      "learning_rate": 7.507034361796996e-05,
      "loss": 2.1031,
      "step": 177372
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.2103121280670166,
      "learning_rate": 7.506763689913683e-05,
      "loss": 1.9553,
      "step": 177373
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.2037543058395386,
      "learning_rate": 7.506493022212295e-05,
      "loss": 2.1131,
      "step": 177374
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.513055682182312,
      "learning_rate": 7.506222358692864e-05,
      "loss": 1.8549,
      "step": 177375
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.355556845664978,
      "learning_rate": 7.505951699355471e-05,
      "loss": 2.0882,
      "step": 177376
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.2505193948745728,
      "learning_rate": 7.505681044200126e-05,
      "loss": 1.8937,
      "step": 177377
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.402411699295044,
      "learning_rate": 7.50541039322691e-05,
      "loss": 1.9543,
      "step": 177378
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.1786075830459595,
      "learning_rate": 7.505139746435856e-05,
      "loss": 2.0078,
      "step": 177379
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.208223819732666,
      "learning_rate": 7.504869103827028e-05,
      "loss": 1.7137,
      "step": 177380
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.3200159072875977,
      "learning_rate": 7.504598465400462e-05,
      "loss": 2.0762,
      "step": 177381
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.279547095298767,
      "learning_rate": 7.50432783115623e-05,
      "loss": 1.8392,
      "step": 177382
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.3028866052627563,
      "learning_rate": 7.504057201094353e-05,
      "loss": 2.1479,
      "step": 177383
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.3122963905334473,
      "learning_rate": 7.503786575214904e-05,
      "loss": 1.8088,
      "step": 177384
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.3422738313674927,
      "learning_rate": 7.503515953517919e-05,
      "loss": 1.8755,
      "step": 177385
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.3352177143096924,
      "learning_rate": 7.503245336003462e-05,
      "loss": 1.9874,
      "step": 177386
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.4239119291305542,
      "learning_rate": 7.502974722671569e-05,
      "loss": 2.1278,
      "step": 177387
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.35031259059906,
      "learning_rate": 7.502704113522313e-05,
      "loss": 2.1207,
      "step": 177388
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.4890594482421875,
      "learning_rate": 7.502433508555711e-05,
      "loss": 1.9882,
      "step": 177389
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.2610708475112915,
      "learning_rate": 7.502162907771845e-05,
      "loss": 1.9459,
      "step": 177390
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.3181159496307373,
      "learning_rate": 7.501892311170739e-05,
      "loss": 2.0004,
      "step": 177391
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.2204447984695435,
      "learning_rate": 7.501621718752467e-05,
      "loss": 2.0682,
      "step": 177392
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.435543417930603,
      "learning_rate": 7.501351130517055e-05,
      "loss": 2.0039,
      "step": 177393
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.4643851518630981,
      "learning_rate": 7.501080546464587e-05,
      "loss": 2.0329,
      "step": 177394
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.3964980840682983,
      "learning_rate": 7.500809966595077e-05,
      "loss": 1.9089,
      "step": 177395
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.4697747230529785,
      "learning_rate": 7.500539390908598e-05,
      "loss": 2.002,
      "step": 177396
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.3363838195800781,
      "learning_rate": 7.500268819405182e-05,
      "loss": 2.0725,
      "step": 177397
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.493994116783142,
      "learning_rate": 7.499998252084903e-05,
      "loss": 2.2028,
      "step": 177398
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.8440451622009277,
      "learning_rate": 7.499727688947789e-05,
      "loss": 1.9249,
      "step": 177399
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.4755457639694214,
      "learning_rate": 7.499457129993921e-05,
      "loss": 2.0701,
      "step": 177400
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.3954713344573975,
      "learning_rate": 7.499186575223307e-05,
      "loss": 1.8965,
      "step": 177401
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.3225467205047607,
      "learning_rate": 7.49891602463603e-05,
      "loss": 2.1095,
      "step": 177402
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.7718268632888794,
      "learning_rate": 7.498645478232117e-05,
      "loss": 1.9326,
      "step": 177403
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.458168864250183,
      "learning_rate": 7.498374936011644e-05,
      "loss": 2.0425,
      "step": 177404
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.330764889717102,
      "learning_rate": 7.498104397974638e-05,
      "loss": 1.9843,
      "step": 177405
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.2112674713134766,
      "learning_rate": 7.497833864121175e-05,
      "loss": 2.0788,
      "step": 177406
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.370553731918335,
      "learning_rate": 7.497563334451273e-05,
      "loss": 1.9206,
      "step": 177407
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.3451381921768188,
      "learning_rate": 7.497292808965008e-05,
      "loss": 1.8913,
      "step": 177408
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.3103396892547607,
      "learning_rate": 7.497022287662408e-05,
      "loss": 1.9377,
      "step": 177409
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.3816273212432861,
      "learning_rate": 7.49675177054355e-05,
      "loss": 1.8488,
      "step": 177410
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.4101086854934692,
      "learning_rate": 7.496481257608461e-05,
      "loss": 2.1345,
      "step": 177411
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.4252458810806274,
      "learning_rate": 7.496210748857219e-05,
      "loss": 2.0699,
      "step": 177412
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.3425054550170898,
      "learning_rate": 7.495940244289837e-05,
      "loss": 1.9249,
      "step": 177413
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.5447559356689453,
      "learning_rate": 7.495669743906394e-05,
      "loss": 2.0902,
      "step": 177414
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.5134072303771973,
      "learning_rate": 7.495399247706921e-05,
      "loss": 2.0109,
      "step": 177415
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.2768036127090454,
      "learning_rate": 7.495128755691488e-05,
      "loss": 1.8598,
      "step": 177416
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.3188883066177368,
      "learning_rate": 7.494858267860125e-05,
      "loss": 2.1662,
      "step": 177417
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.3287566900253296,
      "learning_rate": 7.494587784212902e-05,
      "loss": 2.0066,
      "step": 177418
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.3768975734710693,
      "learning_rate": 7.494317304749862e-05,
      "loss": 1.9949,
      "step": 177419
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.335739254951477,
      "learning_rate": 7.494046829471052e-05,
      "loss": 1.9321,
      "step": 177420
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.4183194637298584,
      "learning_rate": 7.493776358376514e-05,
      "loss": 2.2033,
      "step": 177421
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.3510596752166748,
      "learning_rate": 7.493505891466319e-05,
      "loss": 2.2218,
      "step": 177422
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.196456789970398,
      "learning_rate": 7.493235428740492e-05,
      "loss": 1.9945,
      "step": 177423
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.3981789350509644,
      "learning_rate": 7.492964970199114e-05,
      "loss": 2.0555,
      "step": 177424
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.4145445823669434,
      "learning_rate": 7.492694515842213e-05,
      "loss": 1.9889,
      "step": 177425
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.401668667793274,
      "learning_rate": 7.492424065669836e-05,
      "loss": 1.9115,
      "step": 177426
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.351672649383545,
      "learning_rate": 7.492153619682052e-05,
      "loss": 1.8589,
      "step": 177427
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.2986290454864502,
      "learning_rate": 7.491883177878904e-05,
      "loss": 2.081,
      "step": 177428
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.3976918458938599,
      "learning_rate": 7.491612740260428e-05,
      "loss": 2.1341,
      "step": 177429
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.346840262413025,
      "learning_rate": 7.491342306826698e-05,
      "loss": 1.8738,
      "step": 177430
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.3170101642608643,
      "learning_rate": 7.49107187757775e-05,
      "loss": 2.1138,
      "step": 177431
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.4127709865570068,
      "learning_rate": 7.490801452513628e-05,
      "loss": 2.1719,
      "step": 177432
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.2509349584579468,
      "learning_rate": 7.4905310316344e-05,
      "loss": 2.0313,
      "step": 177433
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.3552931547164917,
      "learning_rate": 7.490260614940105e-05,
      "loss": 2.2232,
      "step": 177434
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.339243769645691,
      "learning_rate": 7.489990202430788e-05,
      "loss": 1.8361,
      "step": 177435
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.1919562816619873,
      "learning_rate": 7.489719794106519e-05,
      "loss": 1.8156,
      "step": 177436
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.2815234661102295,
      "learning_rate": 7.489449389967332e-05,
      "loss": 2.1046,
      "step": 177437
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.380821943283081,
      "learning_rate": 7.489178990013274e-05,
      "loss": 2.0651,
      "step": 177438
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.441206932067871,
      "learning_rate": 7.48890859424441e-05,
      "loss": 2.1511,
      "step": 177439
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.2586759328842163,
      "learning_rate": 7.488638202660776e-05,
      "loss": 1.9608,
      "step": 177440
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.3597885370254517,
      "learning_rate": 7.488367815262436e-05,
      "loss": 1.9493,
      "step": 177441
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.3342456817626953,
      "learning_rate": 7.488097432049436e-05,
      "loss": 1.9629,
      "step": 177442
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.3013091087341309,
      "learning_rate": 7.487827053021822e-05,
      "loss": 1.8784,
      "step": 177443
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.20615816116333,
      "learning_rate": 7.487556678179637e-05,
      "loss": 2.0152,
      "step": 177444
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.4214417934417725,
      "learning_rate": 7.487286307522947e-05,
      "loss": 2.015,
      "step": 177445
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.4597406387329102,
      "learning_rate": 7.487015941051789e-05,
      "loss": 1.8184,
      "step": 177446
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.168471336364746,
      "learning_rate": 7.48674557876623e-05,
      "loss": 2.0664,
      "step": 177447
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.5578807592391968,
      "learning_rate": 7.4864752206663e-05,
      "loss": 1.8768,
      "step": 177448
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.3998860120773315,
      "learning_rate": 7.486204866752074e-05,
      "loss": 2.0266,
      "step": 177449
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.295120120048523,
      "learning_rate": 7.485934517023573e-05,
      "loss": 1.9395,
      "step": 177450
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.228529691696167,
      "learning_rate": 7.485664171480867e-05,
      "loss": 2.0025,
      "step": 177451
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.3459270000457764,
      "learning_rate": 7.485393830123997e-05,
      "loss": 2.1468,
      "step": 177452
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.3880271911621094,
      "learning_rate": 7.485123492953025e-05,
      "loss": 2.0133,
      "step": 177453
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.573649287223816,
      "learning_rate": 7.48485315996798e-05,
      "loss": 2.07,
      "step": 177454
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.329359769821167,
      "learning_rate": 7.484582831168948e-05,
      "loss": 2.2372,
      "step": 177455
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.3704367876052856,
      "learning_rate": 7.48431250655594e-05,
      "loss": 2.0261,
      "step": 177456
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.3660650253295898,
      "learning_rate": 7.484042186129031e-05,
      "loss": 2.211,
      "step": 177457
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.3702629804611206,
      "learning_rate": 7.483771869888253e-05,
      "loss": 1.9948,
      "step": 177458
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.2790669202804565,
      "learning_rate": 7.483501557833677e-05,
      "loss": 1.7999,
      "step": 177459
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.3998013734817505,
      "learning_rate": 7.483231249965334e-05,
      "loss": 2.2877,
      "step": 177460
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.4835538864135742,
      "learning_rate": 7.482960946283304e-05,
      "loss": 2.0798,
      "step": 177461
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.369331955909729,
      "learning_rate": 7.482690646787594e-05,
      "loss": 2.0028,
      "step": 177462
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.3479664325714111,
      "learning_rate": 7.48242035147829e-05,
      "loss": 2.049,
      "step": 177463
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.4166016578674316,
      "learning_rate": 7.482150060355417e-05,
      "loss": 2.1609,
      "step": 177464
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.4234832525253296,
      "learning_rate": 7.481879773419047e-05,
      "loss": 2.2268,
      "step": 177465
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.3300144672393799,
      "learning_rate": 7.481609490669212e-05,
      "loss": 1.9583,
      "step": 177466
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.2991288900375366,
      "learning_rate": 7.48133921210599e-05,
      "loss": 2.0549,
      "step": 177467
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.2469334602355957,
      "learning_rate": 7.481068937729393e-05,
      "loss": 1.9991,
      "step": 177468
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.3623162508010864,
      "learning_rate": 7.480798667539498e-05,
      "loss": 2.0849,
      "step": 177469
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.540366291999817,
      "learning_rate": 7.48052840153634e-05,
      "loss": 2.1921,
      "step": 177470
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.6233396530151367,
      "learning_rate": 7.480258139719988e-05,
      "loss": 2.0106,
      "step": 177471
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.380835771560669,
      "learning_rate": 7.479987882090469e-05,
      "loss": 1.9789,
      "step": 177472
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.274003505706787,
      "learning_rate": 7.479717628647864e-05,
      "loss": 2.0041,
      "step": 177473
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.3570647239685059,
      "learning_rate": 7.479447379392188e-05,
      "loss": 2.151,
      "step": 177474
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.6030397415161133,
      "learning_rate": 7.479177134323514e-05,
      "loss": 2.0588,
      "step": 177475
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.3510606288909912,
      "learning_rate": 7.478906893441878e-05,
      "loss": 2.1048,
      "step": 177476
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.4333933591842651,
      "learning_rate": 7.47863665674735e-05,
      "loss": 2.0014,
      "step": 177477
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.5037091970443726,
      "learning_rate": 7.478366424239959e-05,
      "loss": 2.1593,
      "step": 177478
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.3919323682785034,
      "learning_rate": 7.478096195919782e-05,
      "loss": 2.1032,
      "step": 177479
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.4112980365753174,
      "learning_rate": 7.477825971786831e-05,
      "loss": 2.0168,
      "step": 177480
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.4088276624679565,
      "learning_rate": 7.477555751841189e-05,
      "loss": 1.8682,
      "step": 177481
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.2776962518692017,
      "learning_rate": 7.477285536082886e-05,
      "loss": 1.8766,
      "step": 177482
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.5937669277191162,
      "learning_rate": 7.477015324511991e-05,
      "loss": 2.1104,
      "step": 177483
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.2917574644088745,
      "learning_rate": 7.476745117128533e-05,
      "loss": 1.8152,
      "step": 177484
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.2723183631896973,
      "learning_rate": 7.476474913932592e-05,
      "loss": 2.1657,
      "step": 177485
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.462012767791748,
      "learning_rate": 7.476204714924181e-05,
      "loss": 2.0258,
      "step": 177486
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.2478033304214478,
      "learning_rate": 7.475934520103377e-05,
      "loss": 1.9695,
      "step": 177487
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.1968733072280884,
      "learning_rate": 7.475664329470214e-05,
      "loss": 1.8818,
      "step": 177488
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.3311083316802979,
      "learning_rate": 7.475394143024757e-05,
      "loss": 2.1545,
      "step": 177489
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.3290095329284668,
      "learning_rate": 7.475123960767043e-05,
      "loss": 2.2576,
      "step": 177490
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.4463237524032593,
      "learning_rate": 7.474853782697136e-05,
      "loss": 1.5673,
      "step": 177491
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.2089815139770508,
      "learning_rate": 7.47458360881508e-05,
      "loss": 2.0319,
      "step": 177492
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.3824824094772339,
      "learning_rate": 7.474313439120923e-05,
      "loss": 1.8679,
      "step": 177493
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.384738564491272,
      "learning_rate": 7.47404327361471e-05,
      "loss": 2.0218,
      "step": 177494
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.2826581001281738,
      "learning_rate": 7.473773112296504e-05,
      "loss": 2.1767,
      "step": 177495
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.424088954925537,
      "learning_rate": 7.473502955166341e-05,
      "loss": 1.9663,
      "step": 177496
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.4270576238632202,
      "learning_rate": 7.473232802224287e-05,
      "loss": 2.1837,
      "step": 177497
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.2835354804992676,
      "learning_rate": 7.472962653470383e-05,
      "loss": 1.9179,
      "step": 177498
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.3925741910934448,
      "learning_rate": 7.472692508904683e-05,
      "loss": 2.0931,
      "step": 177499
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.2793724536895752,
      "learning_rate": 7.472422368527226e-05,
      "loss": 1.9388,
      "step": 177500
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.3652193546295166,
      "learning_rate": 7.472152232338077e-05,
      "loss": 2.0591,
      "step": 177501
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.4920063018798828,
      "learning_rate": 7.471882100337276e-05,
      "loss": 2.0761,
      "step": 177502
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.233392596244812,
      "learning_rate": 7.47161197252488e-05,
      "loss": 1.7421,
      "step": 177503
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.362869143486023,
      "learning_rate": 7.471341848900942e-05,
      "loss": 1.9864,
      "step": 177504
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.4868125915527344,
      "learning_rate": 7.471071729465506e-05,
      "loss": 2.2171,
      "step": 177505
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.4467718601226807,
      "learning_rate": 7.470801614218613e-05,
      "loss": 2.0433,
      "step": 177506
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.5438743829727173,
      "learning_rate": 7.47053150316033e-05,
      "loss": 1.9159,
      "step": 177507
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.3159557580947876,
      "learning_rate": 7.470261396290695e-05,
      "loss": 1.8353,
      "step": 177508
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.4171173572540283,
      "learning_rate": 7.46999129360977e-05,
      "loss": 2.0968,
      "step": 177509
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.3634687662124634,
      "learning_rate": 7.469721195117602e-05,
      "loss": 2.0143,
      "step": 177510
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.5727537870407104,
      "learning_rate": 7.469451100814225e-05,
      "loss": 2.2314,
      "step": 177511
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.3632543087005615,
      "learning_rate": 7.469181010699716e-05,
      "loss": 1.8775,
      "step": 177512
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.3794516324996948,
      "learning_rate": 7.468910924774108e-05,
      "loss": 1.9302,
      "step": 177513
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.5654197931289673,
      "learning_rate": 7.468640843037449e-05,
      "loss": 1.9673,
      "step": 177514
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.438065767288208,
      "learning_rate": 7.4683707654898e-05,
      "loss": 2.1845,
      "step": 177515
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.383461594581604,
      "learning_rate": 7.468100692131209e-05,
      "loss": 2.2489,
      "step": 177516
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.4122360944747925,
      "learning_rate": 7.467830622961715e-05,
      "loss": 1.9734,
      "step": 177517
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.2703709602355957,
      "learning_rate": 7.467560557981382e-05,
      "loss": 2.106,
      "step": 177518
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.2693268060684204,
      "learning_rate": 7.467290497190256e-05,
      "loss": 2.3292,
      "step": 177519
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.5374118089675903,
      "learning_rate": 7.467020440588378e-05,
      "loss": 2.0239,
      "step": 177520
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.292198896408081,
      "learning_rate": 7.466750388175816e-05,
      "loss": 1.8714,
      "step": 177521
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.2423940896987915,
      "learning_rate": 7.46648033995261e-05,
      "loss": 2.0742,
      "step": 177522
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.3286950588226318,
      "learning_rate": 7.466210295918802e-05,
      "loss": 1.8405,
      "step": 177523
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.311218500137329,
      "learning_rate": 7.465940256074459e-05,
      "loss": 2.1341,
      "step": 177524
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.4972423315048218,
      "learning_rate": 7.465670220419613e-05,
      "loss": 1.9702,
      "step": 177525
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.5928521156311035,
      "learning_rate": 7.465400188954335e-05,
      "loss": 2.0702,
      "step": 177526
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.3801411390304565,
      "learning_rate": 7.465130161678666e-05,
      "loss": 1.9742,
      "step": 177527
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.3590900897979736,
      "learning_rate": 7.464860138592651e-05,
      "loss": 2.1004,
      "step": 177528
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.2451071739196777,
      "learning_rate": 7.464590119696336e-05,
      "loss": 2.0163,
      "step": 177529
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.1449334621429443,
      "learning_rate": 7.464320104989788e-05,
      "loss": 1.9088,
      "step": 177530
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.3357692956924438,
      "learning_rate": 7.464050094473043e-05,
      "loss": 2.0452,
      "step": 177531
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.248031735420227,
      "learning_rate": 7.46378008814616e-05,
      "loss": 1.8634,
      "step": 177532
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.412585973739624,
      "learning_rate": 7.463510086009188e-05,
      "loss": 2.1165,
      "step": 177533
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.3082605600357056,
      "learning_rate": 7.463240088062175e-05,
      "loss": 1.9232,
      "step": 177534
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.4277443885803223,
      "learning_rate": 7.462970094305163e-05,
      "loss": 1.8045,
      "step": 177535
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.4406908750534058,
      "learning_rate": 7.46270010473822e-05,
      "loss": 1.9374,
      "step": 177536
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.295539379119873,
      "learning_rate": 7.462430119361375e-05,
      "loss": 2.0041,
      "step": 177537
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.4954520463943481,
      "learning_rate": 7.462160138174698e-05,
      "loss": 2.1759,
      "step": 177538
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.537766456604004,
      "learning_rate": 7.461890161178226e-05,
      "loss": 2.1485,
      "step": 177539
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.4157230854034424,
      "learning_rate": 7.461620188372028e-05,
      "loss": 1.887,
      "step": 177540
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.2842382192611694,
      "learning_rate": 7.461350219756123e-05,
      "loss": 1.8896,
      "step": 177541
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.3143813610076904,
      "learning_rate": 7.461080255330588e-05,
      "loss": 2.0133,
      "step": 177542
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.2521473169326782,
      "learning_rate": 7.460810295095455e-05,
      "loss": 1.9611,
      "step": 177543
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.2488905191421509,
      "learning_rate": 7.460540339050794e-05,
      "loss": 2.2962,
      "step": 177544
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.4035645723342896,
      "learning_rate": 7.46027038719663e-05,
      "loss": 1.9282,
      "step": 177545
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.3299415111541748,
      "learning_rate": 7.460000439533046e-05,
      "loss": 1.9929,
      "step": 177546
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.3293291330337524,
      "learning_rate": 7.459730496060059e-05,
      "loss": 1.7491,
      "step": 177547
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.3941571712493896,
      "learning_rate": 7.459460556777743e-05,
      "loss": 2.17,
      "step": 177548
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.3781325817108154,
      "learning_rate": 7.459190621686127e-05,
      "loss": 2.0294,
      "step": 177549
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.4561423063278198,
      "learning_rate": 7.458920690785282e-05,
      "loss": 2.0334,
      "step": 177550
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.3166818618774414,
      "learning_rate": 7.458650764075241e-05,
      "loss": 2.1121,
      "step": 177551
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.27730131149292,
      "learning_rate": 7.458380841556079e-05,
      "loss": 1.9294,
      "step": 177552
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.227846384048462,
      "learning_rate": 7.458110923227814e-05,
      "loss": 2.1141,
      "step": 177553
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.2271467447280884,
      "learning_rate": 7.45784100909052e-05,
      "loss": 2.1281,
      "step": 177554
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.41569983959198,
      "learning_rate": 7.457571099144226e-05,
      "loss": 1.8396,
      "step": 177555
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.3431921005249023,
      "learning_rate": 7.457301193389007e-05,
      "loss": 2.0561,
      "step": 177556
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.4880636930465698,
      "learning_rate": 7.457031291824892e-05,
      "loss": 1.981,
      "step": 177557
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.2348542213439941,
      "learning_rate": 7.456761394451951e-05,
      "loss": 1.7831,
      "step": 177558
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.3191317319869995,
      "learning_rate": 7.45649150127022e-05,
      "loss": 2.0278,
      "step": 177559
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.365963339805603,
      "learning_rate": 7.456221612279758e-05,
      "loss": 2.206,
      "step": 177560
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.3390580415725708,
      "learning_rate": 7.455951727480599e-05,
      "loss": 2.1038,
      "step": 177561
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.3533746004104614,
      "learning_rate": 7.45568184687281e-05,
      "loss": 1.9316,
      "step": 177562
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.5002448558807373,
      "learning_rate": 7.45541197045643e-05,
      "loss": 1.9671,
      "step": 177563
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.3760032653808594,
      "learning_rate": 7.455142098231522e-05,
      "loss": 2.1545,
      "step": 177564
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.2937445640563965,
      "learning_rate": 7.454872230198129e-05,
      "loss": 2.046,
      "step": 177565
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.278122901916504,
      "learning_rate": 7.4546023663563e-05,
      "loss": 2.1093,
      "step": 177566
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.3343873023986816,
      "learning_rate": 7.454332506706076e-05,
      "loss": 1.822,
      "step": 177567
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.4230636358261108,
      "learning_rate": 7.454062651247528e-05,
      "loss": 1.9808,
      "step": 177568
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.3544591665267944,
      "learning_rate": 7.453792799980686e-05,
      "loss": 1.9439,
      "step": 177569
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.2686446905136108,
      "learning_rate": 7.45352295290562e-05,
      "loss": 1.9999,
      "step": 177570
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.3452095985412598,
      "learning_rate": 7.453253110022365e-05,
      "loss": 2.1846,
      "step": 177571
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.2828255891799927,
      "learning_rate": 7.452983271330981e-05,
      "loss": 2.0486,
      "step": 177572
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.3619521856307983,
      "learning_rate": 7.452713436831503e-05,
      "loss": 2.2525,
      "step": 177573
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.2726436853408813,
      "learning_rate": 7.452443606524e-05,
      "loss": 1.9276,
      "step": 177574
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.3379000425338745,
      "learning_rate": 7.452173780408501e-05,
      "loss": 1.8932,
      "step": 177575
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.3147549629211426,
      "learning_rate": 7.45190395848508e-05,
      "loss": 2.056,
      "step": 177576
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.2864834070205688,
      "learning_rate": 7.451634140753775e-05,
      "loss": 1.9428,
      "step": 177577
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.3006091117858887,
      "learning_rate": 7.451364327214639e-05,
      "loss": 2.0313,
      "step": 177578
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.2691426277160645,
      "learning_rate": 7.451094517867711e-05,
      "loss": 2.1009,
      "step": 177579
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.6692924499511719,
      "learning_rate": 7.450824712713057e-05,
      "loss": 2.0634,
      "step": 177580
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.3332377672195435,
      "learning_rate": 7.450554911750711e-05,
      "loss": 1.9728,
      "step": 177581
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.273611068725586,
      "learning_rate": 7.450285114980743e-05,
      "loss": 1.8916,
      "step": 177582
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.4920883178710938,
      "learning_rate": 7.450015322403194e-05,
      "loss": 1.9273,
      "step": 177583
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.3220678567886353,
      "learning_rate": 7.449745534018109e-05,
      "loss": 2.0282,
      "step": 177584
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.3591711521148682,
      "learning_rate": 7.449475749825536e-05,
      "loss": 2.0092,
      "step": 177585
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.2663698196411133,
      "learning_rate": 7.44920596982554e-05,
      "loss": 1.9883,
      "step": 177586
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.454398274421692,
      "learning_rate": 7.44893619401815e-05,
      "loss": 1.9411,
      "step": 177587
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.2324656248092651,
      "learning_rate": 7.448666422403444e-05,
      "loss": 1.8478,
      "step": 177588
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.2978110313415527,
      "learning_rate": 7.44839665498145e-05,
      "loss": 2.0679,
      "step": 177589
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.332190752029419,
      "learning_rate": 7.448126891752228e-05,
      "loss": 2.0345,
      "step": 177590
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.2701101303100586,
      "learning_rate": 7.447857132715815e-05,
      "loss": 1.8791,
      "step": 177591
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.1882792711257935,
      "learning_rate": 7.447587377872279e-05,
      "loss": 1.9095,
      "step": 177592
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.374416708946228,
      "learning_rate": 7.447317627221657e-05,
      "loss": 2.2055,
      "step": 177593
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.375651240348816,
      "learning_rate": 7.447047880764009e-05,
      "loss": 2.1469,
      "step": 177594
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.4360774755477905,
      "learning_rate": 7.446778138499384e-05,
      "loss": 1.6903,
      "step": 177595
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.383571743965149,
      "learning_rate": 7.446508400427825e-05,
      "loss": 1.8157,
      "step": 177596
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.450387954711914,
      "learning_rate": 7.446238666549379e-05,
      "loss": 2.0058,
      "step": 177597
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.4631143808364868,
      "learning_rate": 7.445968936864112e-05,
      "loss": 1.9771,
      "step": 177598
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.358364462852478,
      "learning_rate": 7.445699211372057e-05,
      "loss": 2.0165,
      "step": 177599
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.2294942140579224,
      "learning_rate": 7.445429490073278e-05,
      "loss": 1.994,
      "step": 177600
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.3034913539886475,
      "learning_rate": 7.445159772967825e-05,
      "loss": 1.9897,
      "step": 177601
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.285109519958496,
      "learning_rate": 7.444890060055727e-05,
      "loss": 2.063,
      "step": 177602
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.337146282196045,
      "learning_rate": 7.444620351337062e-05,
      "loss": 1.8746,
      "step": 177603
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.1898517608642578,
      "learning_rate": 7.444350646811868e-05,
      "loss": 1.8888,
      "step": 177604
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.2592463493347168,
      "learning_rate": 7.444080946480188e-05,
      "loss": 1.8951,
      "step": 177605
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.3800740242004395,
      "learning_rate": 7.443811250342085e-05,
      "loss": 2.2879,
      "step": 177606
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.3198319673538208,
      "learning_rate": 7.443541558397602e-05,
      "loss": 2.0504,
      "step": 177607
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.7433102130889893,
      "learning_rate": 7.443271870646784e-05,
      "loss": 1.9954,
      "step": 177608
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.2188011407852173,
      "learning_rate": 7.443002187089698e-05,
      "loss": 2.0594,
      "step": 177609
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.1853246688842773,
      "learning_rate": 7.442732507726372e-05,
      "loss": 2.0225,
      "step": 177610
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.264695405960083,
      "learning_rate": 7.442462832556876e-05,
      "loss": 2.045,
      "step": 177611
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.4632130861282349,
      "learning_rate": 7.442193161581255e-05,
      "loss": 1.8859,
      "step": 177612
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.3156001567840576,
      "learning_rate": 7.441923494799554e-05,
      "loss": 2.0881,
      "step": 177613
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.2788951396942139,
      "learning_rate": 7.441653832211818e-05,
      "loss": 2.0443,
      "step": 177614
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.3942490816116333,
      "learning_rate": 7.44138417381811e-05,
      "loss": 1.9397,
      "step": 177615
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.238664984703064,
      "learning_rate": 7.441114519618471e-05,
      "loss": 1.9206,
      "step": 177616
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.2534890174865723,
      "learning_rate": 7.440844869612959e-05,
      "loss": 2.1351,
      "step": 177617
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.2844158411026,
      "learning_rate": 7.44057522380162e-05,
      "loss": 1.847,
      "step": 177618
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.4245285987854004,
      "learning_rate": 7.440305582184506e-05,
      "loss": 1.9205,
      "step": 177619
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.3577795028686523,
      "learning_rate": 7.440035944761656e-05,
      "loss": 1.9136,
      "step": 177620
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.2250620126724243,
      "learning_rate": 7.439766311533135e-05,
      "loss": 1.8378,
      "step": 177621
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.3484879732131958,
      "learning_rate": 7.439496682498982e-05,
      "loss": 2.107,
      "step": 177622
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.2392497062683105,
      "learning_rate": 7.43922705765926e-05,
      "loss": 1.9151,
      "step": 177623
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.2003051042556763,
      "learning_rate": 7.438957437014004e-05,
      "loss": 2.169,
      "step": 177624
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.2934880256652832,
      "learning_rate": 7.438687820563278e-05,
      "loss": 1.8582,
      "step": 177625
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.3696140050888062,
      "learning_rate": 7.438418208307127e-05,
      "loss": 2.0884,
      "step": 177626
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.4187638759613037,
      "learning_rate": 7.438148600245602e-05,
      "loss": 2.0287,
      "step": 177627
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.4776500463485718,
      "learning_rate": 7.437878996378736e-05,
      "loss": 2.0347,
      "step": 177628
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.343862533569336,
      "learning_rate": 7.43760939670661e-05,
      "loss": 2.007,
      "step": 177629
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.35638427734375,
      "learning_rate": 7.437339801229248e-05,
      "loss": 1.9378,
      "step": 177630
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.236473798751831,
      "learning_rate": 7.437070209946716e-05,
      "loss": 2.1269,
      "step": 177631
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.2785385847091675,
      "learning_rate": 7.436800622859059e-05,
      "loss": 1.7908,
      "step": 177632
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.4137176275253296,
      "learning_rate": 7.43653103996633e-05,
      "loss": 1.8385,
      "step": 177633
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.322274088859558,
      "learning_rate": 7.436261461268566e-05,
      "loss": 1.9553,
      "step": 177634
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.3183979988098145,
      "learning_rate": 7.435991886765835e-05,
      "loss": 1.8465,
      "step": 177635
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.2495043277740479,
      "learning_rate": 7.43572231645817e-05,
      "loss": 1.9373,
      "step": 177636
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.3817580938339233,
      "learning_rate": 7.435452750345638e-05,
      "loss": 2.1552,
      "step": 177637
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.464767575263977,
      "learning_rate": 7.435183188428283e-05,
      "loss": 2.0132,
      "step": 177638
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.4160974025726318,
      "learning_rate": 7.434913630706153e-05,
      "loss": 2.0397,
      "step": 177639
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.3501791954040527,
      "learning_rate": 7.434644077179287e-05,
      "loss": 2.0418,
      "step": 177640
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.5030039548873901,
      "learning_rate": 7.434374527847758e-05,
      "loss": 1.9753,
      "step": 177641
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.378248929977417,
      "learning_rate": 7.434104982711597e-05,
      "loss": 1.8509,
      "step": 177642
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.3402825593948364,
      "learning_rate": 7.43383544177087e-05,
      "loss": 2.1028,
      "step": 177643
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.4791195392608643,
      "learning_rate": 7.433565905025617e-05,
      "loss": 2.0296,
      "step": 177644
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.6739529371261597,
      "learning_rate": 7.433296372475891e-05,
      "loss": 1.958,
      "step": 177645
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.3137102127075195,
      "learning_rate": 7.433026844121735e-05,
      "loss": 2.0329,
      "step": 177646
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.3417402505874634,
      "learning_rate": 7.432757319963211e-05,
      "loss": 1.9506,
      "step": 177647
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.518243432044983,
      "learning_rate": 7.432487800000357e-05,
      "loss": 1.9185,
      "step": 177648
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.328310251235962,
      "learning_rate": 7.432218284233235e-05,
      "loss": 2.0921,
      "step": 177649
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.4077129364013672,
      "learning_rate": 7.431948772661892e-05,
      "loss": 1.997,
      "step": 177650
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.4911065101623535,
      "learning_rate": 7.431679265286375e-05,
      "loss": 2.0436,
      "step": 177651
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.1990511417388916,
      "learning_rate": 7.431409762106726e-05,
      "loss": 1.7699,
      "step": 177652
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.4175121784210205,
      "learning_rate": 7.431140263123013e-05,
      "loss": 2.154,
      "step": 177653
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.4248648881912231,
      "learning_rate": 7.430870768335268e-05,
      "loss": 1.8587,
      "step": 177654
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.3286495208740234,
      "learning_rate": 7.430601277743558e-05,
      "loss": 1.9942,
      "step": 177655
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.2317637205123901,
      "learning_rate": 7.43033179134793e-05,
      "loss": 2.0765,
      "step": 177656
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.3564872741699219,
      "learning_rate": 7.430062309148424e-05,
      "loss": 2.1605,
      "step": 177657
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.301854133605957,
      "learning_rate": 7.42979283114509e-05,
      "loss": 2.048,
      "step": 177658
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.3355233669281006,
      "learning_rate": 7.429523357337991e-05,
      "loss": 2.2625,
      "step": 177659
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.6145353317260742,
      "learning_rate": 7.42925388772716e-05,
      "loss": 1.9302,
      "step": 177660
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.30972158908844,
      "learning_rate": 7.428984422312667e-05,
      "loss": 2.1417,
      "step": 177661
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.390058994293213,
      "learning_rate": 7.428714961094549e-05,
      "loss": 2.4542,
      "step": 177662
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.4583766460418701,
      "learning_rate": 7.428445504072863e-05,
      "loss": 2.0793,
      "step": 177663
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.3311407566070557,
      "learning_rate": 7.428176051247644e-05,
      "loss": 2.0888,
      "step": 177664
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.306575059890747,
      "learning_rate": 7.427906602618961e-05,
      "loss": 1.8397,
      "step": 177665
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.4183443784713745,
      "learning_rate": 7.427637158186848e-05,
      "loss": 2.2973,
      "step": 177666
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.2034037113189697,
      "learning_rate": 7.427367717951373e-05,
      "loss": 2.0859,
      "step": 177667
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.366587519645691,
      "learning_rate": 7.427098281912576e-05,
      "loss": 2.1672,
      "step": 177668
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.2408875226974487,
      "learning_rate": 7.426828850070508e-05,
      "loss": 2.0967,
      "step": 177669
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.447575330734253,
      "learning_rate": 7.426559422425211e-05,
      "loss": 2.1302,
      "step": 177670
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.3739944696426392,
      "learning_rate": 7.42628999897675e-05,
      "loss": 2.1282,
      "step": 177671
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.3275182247161865,
      "learning_rate": 7.426020579725158e-05,
      "loss": 1.8212,
      "step": 177672
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.4612888097763062,
      "learning_rate": 7.425751164670506e-05,
      "loss": 2.1105,
      "step": 177673
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.3180755376815796,
      "learning_rate": 7.425481753812832e-05,
      "loss": 1.8668,
      "step": 177674
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.2606513500213623,
      "learning_rate": 7.425212347152186e-05,
      "loss": 2.0876,
      "step": 177675
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.3570828437805176,
      "learning_rate": 7.424942944688612e-05,
      "loss": 2.1576,
      "step": 177676
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.5849149227142334,
      "learning_rate": 7.424673546422174e-05,
      "loss": 2.3513,
      "step": 177677
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.5085170269012451,
      "learning_rate": 7.424404152352908e-05,
      "loss": 1.8426,
      "step": 177678
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.4015802145004272,
      "learning_rate": 7.424134762480879e-05,
      "loss": 2.0286,
      "step": 177679
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.3012938499450684,
      "learning_rate": 7.423865376806132e-05,
      "loss": 1.9951,
      "step": 177680
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.4447557926177979,
      "learning_rate": 7.423595995328713e-05,
      "loss": 2.1457,
      "step": 177681
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.4295560121536255,
      "learning_rate": 7.423326618048665e-05,
      "loss": 2.0385,
      "step": 177682
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.3491661548614502,
      "learning_rate": 7.423057244966055e-05,
      "loss": 2.0517,
      "step": 177683
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.4059231281280518,
      "learning_rate": 7.422787876080918e-05,
      "loss": 1.928,
      "step": 177684
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.2874231338500977,
      "learning_rate": 7.422518511393316e-05,
      "loss": 2.2441,
      "step": 177685
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.3574069738388062,
      "learning_rate": 7.422249150903297e-05,
      "loss": 2.0585,
      "step": 177686
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.359190821647644,
      "learning_rate": 7.4219797946109e-05,
      "loss": 2.0409,
      "step": 177687
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.3535284996032715,
      "learning_rate": 7.421710442516192e-05,
      "loss": 2.1132,
      "step": 177688
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.4996862411499023,
      "learning_rate": 7.421441094619212e-05,
      "loss": 2.0624,
      "step": 177689
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.2290898561477661,
      "learning_rate": 7.421171750920004e-05,
      "loss": 2.0093,
      "step": 177690
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.3550798892974854,
      "learning_rate": 7.420902411418635e-05,
      "loss": 2.0238,
      "step": 177691
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.4196958541870117,
      "learning_rate": 7.420633076115135e-05,
      "loss": 2.0205,
      "step": 177692
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.4520457983016968,
      "learning_rate": 7.420363745009578e-05,
      "loss": 2.1402,
      "step": 177693
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.2857965230941772,
      "learning_rate": 7.420094418102002e-05,
      "loss": 1.9995,
      "step": 177694
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.570197582244873,
      "learning_rate": 7.419825095392455e-05,
      "loss": 2.1135,
      "step": 177695
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.1920804977416992,
      "learning_rate": 7.419555776880982e-05,
      "loss": 1.7779,
      "step": 177696
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.3466465473175049,
      "learning_rate": 7.419286462567645e-05,
      "loss": 1.9715,
      "step": 177697
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.4029844999313354,
      "learning_rate": 7.419017152452483e-05,
      "loss": 2.2516,
      "step": 177698
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.3045240640640259,
      "learning_rate": 7.418747846535559e-05,
      "loss": 1.8647,
      "step": 177699
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.3269472122192383,
      "learning_rate": 7.418478544816919e-05,
      "loss": 1.9215,
      "step": 177700
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.5074938535690308,
      "learning_rate": 7.418209247296599e-05,
      "loss": 1.8454,
      "step": 177701
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.3376976251602173,
      "learning_rate": 7.41793995397467e-05,
      "loss": 2.1118,
      "step": 177702
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.3109054565429688,
      "learning_rate": 7.417670664851174e-05,
      "loss": 1.8241,
      "step": 177703
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.3261381387710571,
      "learning_rate": 7.417401379926144e-05,
      "loss": 1.9768,
      "step": 177704
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.4431909322738647,
      "learning_rate": 7.41713209919966e-05,
      "loss": 2.0581,
      "step": 177705
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.4825637340545654,
      "learning_rate": 7.416862822671755e-05,
      "loss": 1.9071,
      "step": 177706
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.3257486820220947,
      "learning_rate": 7.416593550342475e-05,
      "loss": 2.0037,
      "step": 177707
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.2614139318466187,
      "learning_rate": 7.416324282211882e-05,
      "loss": 2.0713,
      "step": 177708
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.3783416748046875,
      "learning_rate": 7.416055018280014e-05,
      "loss": 1.9852,
      "step": 177709
    },
    {
      "epoch": 2.31,
      "grad_norm": 2.0078773498535156,
      "learning_rate": 7.415785758546937e-05,
      "loss": 1.8912,
      "step": 177710
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.7095516920089722,
      "learning_rate": 7.415516503012689e-05,
      "loss": 1.9423,
      "step": 177711
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.1758815050125122,
      "learning_rate": 7.415247251677324e-05,
      "loss": 1.908,
      "step": 177712
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.2693979740142822,
      "learning_rate": 7.414978004540885e-05,
      "loss": 2.2422,
      "step": 177713
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.28254234790802,
      "learning_rate": 7.414708761603433e-05,
      "loss": 2.0448,
      "step": 177714
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.3052090406417847,
      "learning_rate": 7.414439522865006e-05,
      "loss": 2.0004,
      "step": 177715
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.3032387495040894,
      "learning_rate": 7.414170288325669e-05,
      "loss": 1.8874,
      "step": 177716
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.3452810049057007,
      "learning_rate": 7.413901057985464e-05,
      "loss": 1.9362,
      "step": 177717
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.2633545398712158,
      "learning_rate": 7.413631831844443e-05,
      "loss": 2.1232,
      "step": 177718
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.258826494216919,
      "learning_rate": 7.413362609902643e-05,
      "loss": 1.8943,
      "step": 177719
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.2783814668655396,
      "learning_rate": 7.413093392160135e-05,
      "loss": 2.0331,
      "step": 177720
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.267780065536499,
      "learning_rate": 7.41282417861695e-05,
      "loss": 2.0632,
      "step": 177721
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.5566281080245972,
      "learning_rate": 7.412554969273156e-05,
      "loss": 1.9426,
      "step": 177722
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.6795246601104736,
      "learning_rate": 7.412285764128796e-05,
      "loss": 1.973,
      "step": 177723
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.3914878368377686,
      "learning_rate": 7.412016563183916e-05,
      "loss": 1.8699,
      "step": 177724
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.3383952379226685,
      "learning_rate": 7.411747366438562e-05,
      "loss": 1.8304,
      "step": 177725
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.5122947692871094,
      "learning_rate": 7.411478173892797e-05,
      "loss": 1.8871,
      "step": 177726
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.494126319885254,
      "learning_rate": 7.411208985546661e-05,
      "loss": 2.0193,
      "step": 177727
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.2650256156921387,
      "learning_rate": 7.41093980140021e-05,
      "loss": 2.1283,
      "step": 177728
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.2763707637786865,
      "learning_rate": 7.410670621453496e-05,
      "loss": 2.0904,
      "step": 177729
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.5406798124313354,
      "learning_rate": 7.410401445706562e-05,
      "loss": 2.0422,
      "step": 177730
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.3646501302719116,
      "learning_rate": 7.410132274159455e-05,
      "loss": 1.9335,
      "step": 177731
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.383908748626709,
      "learning_rate": 7.409863106812237e-05,
      "loss": 2.1086,
      "step": 177732
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.2347699403762817,
      "learning_rate": 7.409593943664943e-05,
      "loss": 2.1303,
      "step": 177733
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.3778806924819946,
      "learning_rate": 7.409324784717643e-05,
      "loss": 2.0416,
      "step": 177734
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.417426586151123,
      "learning_rate": 7.409055629970373e-05,
      "loss": 2.0571,
      "step": 177735
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.2849934101104736,
      "learning_rate": 7.40878647942319e-05,
      "loss": 1.7393,
      "step": 177736
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.2776577472686768,
      "learning_rate": 7.408517333076128e-05,
      "loss": 2.0058,
      "step": 177737
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.4462361335754395,
      "learning_rate": 7.408248190929258e-05,
      "loss": 2.0651,
      "step": 177738
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.2735657691955566,
      "learning_rate": 7.407979052982612e-05,
      "loss": 2.0538,
      "step": 177739
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.7829631567001343,
      "learning_rate": 7.40770991923626e-05,
      "loss": 2.0812,
      "step": 177740
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.231027364730835,
      "learning_rate": 7.407440789690241e-05,
      "loss": 1.98,
      "step": 177741
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.263559103012085,
      "learning_rate": 7.407171664344602e-05,
      "loss": 1.9874,
      "step": 177742
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.3821882009506226,
      "learning_rate": 7.40690254319939e-05,
      "loss": 2.0123,
      "step": 177743
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.3802410364151,
      "learning_rate": 7.406633426254671e-05,
      "loss": 1.97,
      "step": 177744
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.2681291103363037,
      "learning_rate": 7.406364313510475e-05,
      "loss": 1.8544,
      "step": 177745
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.194150447845459,
      "learning_rate": 7.406095204966873e-05,
      "loss": 1.9417,
      "step": 177746
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.3199481964111328,
      "learning_rate": 7.4058261006239e-05,
      "loss": 2.1818,
      "step": 177747
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.5256050825119019,
      "learning_rate": 7.405557000481614e-05,
      "loss": 2.1398,
      "step": 177748
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.2044943571090698,
      "learning_rate": 7.405287904540055e-05,
      "loss": 2.0846,
      "step": 177749
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.3714234828948975,
      "learning_rate": 7.405018812799285e-05,
      "loss": 1.8908,
      "step": 177750
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.2137371301651,
      "learning_rate": 7.404749725259338e-05,
      "loss": 2.0853,
      "step": 177751
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.4265302419662476,
      "learning_rate": 7.404480641920287e-05,
      "loss": 1.9184,
      "step": 177752
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.4346064329147339,
      "learning_rate": 7.404211562782167e-05,
      "loss": 1.8498,
      "step": 177753
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.2993086576461792,
      "learning_rate": 7.403942487845032e-05,
      "loss": 2.0626,
      "step": 177754
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.2763032913208008,
      "learning_rate": 7.403673417108919e-05,
      "loss": 2.3042,
      "step": 177755
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.2692466974258423,
      "learning_rate": 7.403404350573904e-05,
      "loss": 2.1853,
      "step": 177756
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.2757441997528076,
      "learning_rate": 7.403135288240011e-05,
      "loss": 2.0853,
      "step": 177757
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.3782627582550049,
      "learning_rate": 7.40286623010731e-05,
      "loss": 2.0935,
      "step": 177758
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.3861608505249023,
      "learning_rate": 7.402597176175833e-05,
      "loss": 1.9837,
      "step": 177759
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.369315505027771,
      "learning_rate": 7.402328126445658e-05,
      "loss": 2.0689,
      "step": 177760
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.315800428390503,
      "learning_rate": 7.402059080916799e-05,
      "loss": 2.1588,
      "step": 177761
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.4026918411254883,
      "learning_rate": 7.401790039589336e-05,
      "loss": 2.0892,
      "step": 177762
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.3096766471862793,
      "learning_rate": 7.401521002463295e-05,
      "loss": 2.1387,
      "step": 177763
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.2755143642425537,
      "learning_rate": 7.401251969538744e-05,
      "loss": 2.1322,
      "step": 177764
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.2890117168426514,
      "learning_rate": 7.400982940815722e-05,
      "loss": 2.0418,
      "step": 177765
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.4056445360183716,
      "learning_rate": 7.400713916294302e-05,
      "loss": 1.8902,
      "step": 177766
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.2776366472244263,
      "learning_rate": 7.400444895974494e-05,
      "loss": 2.049,
      "step": 177767
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.312434434890747,
      "learning_rate": 7.400175879856383e-05,
      "loss": 1.9399,
      "step": 177768
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.3812072277069092,
      "learning_rate": 7.399906867939997e-05,
      "loss": 2.0357,
      "step": 177769
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.2638752460479736,
      "learning_rate": 7.399637860225399e-05,
      "loss": 2.166,
      "step": 177770
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.3332747220993042,
      "learning_rate": 7.399368856712631e-05,
      "loss": 2.0569,
      "step": 177771
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.260876178741455,
      "learning_rate": 7.399099857401754e-05,
      "loss": 1.9472,
      "step": 177772
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.228315830230713,
      "learning_rate": 7.39883086229281e-05,
      "loss": 2.0492,
      "step": 177773
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.2336632013320923,
      "learning_rate": 7.398561871385853e-05,
      "loss": 1.9634,
      "step": 177774
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.3023042678833008,
      "learning_rate": 7.398292884680918e-05,
      "loss": 2.0037,
      "step": 177775
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.429628849029541,
      "learning_rate": 7.398023902178077e-05,
      "loss": 2.1014,
      "step": 177776
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.2976231575012207,
      "learning_rate": 7.397754923877362e-05,
      "loss": 1.9726,
      "step": 177777
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.3949637413024902,
      "learning_rate": 7.397485949778838e-05,
      "loss": 2.0077,
      "step": 177778
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.3159770965576172,
      "learning_rate": 7.397216979882549e-05,
      "loss": 2.0222,
      "step": 177779
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.5255932807922363,
      "learning_rate": 7.396948014188545e-05,
      "loss": 2.0814,
      "step": 177780
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.2926450967788696,
      "learning_rate": 7.396679052696864e-05,
      "loss": 2.0789,
      "step": 177781
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.4102883338928223,
      "learning_rate": 7.396410095407577e-05,
      "loss": 1.9403,
      "step": 177782
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.2495626211166382,
      "learning_rate": 7.396141142320715e-05,
      "loss": 2.1874,
      "step": 177783
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.3704650402069092,
      "learning_rate": 7.395872193436349e-05,
      "loss": 1.8173,
      "step": 177784
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.3045392036437988,
      "learning_rate": 7.395603248754513e-05,
      "loss": 1.8907,
      "step": 177785
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.2739057540893555,
      "learning_rate": 7.395334308275253e-05,
      "loss": 2.0538,
      "step": 177786
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.3189105987548828,
      "learning_rate": 7.395065371998635e-05,
      "loss": 1.7533,
      "step": 177787
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.5217530727386475,
      "learning_rate": 7.394796439924705e-05,
      "loss": 2.0068,
      "step": 177788
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.195724368095398,
      "learning_rate": 7.394527512053497e-05,
      "loss": 1.8499,
      "step": 177789
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.245951533317566,
      "learning_rate": 7.394258588385082e-05,
      "loss": 1.9318,
      "step": 177790
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.2543166875839233,
      "learning_rate": 7.3939896689195e-05,
      "loss": 2.0313,
      "step": 177791
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.4952995777130127,
      "learning_rate": 7.393720753656795e-05,
      "loss": 2.0072,
      "step": 177792
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.3153679370880127,
      "learning_rate": 7.393451842597033e-05,
      "loss": 2.0927,
      "step": 177793
    },
    {
      "epoch": 2.31,
      "grad_norm": 1.333689570426941,
      "learning_rate": 7.393182935740255e-05,
      "loss": 2.0575,
      "step": 177794
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.235681176185608,
      "learning_rate": 7.392914033086503e-05,
      "loss": 1.9967,
      "step": 177795
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.2871748208999634,
      "learning_rate": 7.392645134635841e-05,
      "loss": 2.0501,
      "step": 177796
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.2989561557769775,
      "learning_rate": 7.392376240388315e-05,
      "loss": 1.9749,
      "step": 177797
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.321139931678772,
      "learning_rate": 7.392107350343965e-05,
      "loss": 2.0135,
      "step": 177798
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.3228768110275269,
      "learning_rate": 7.391838464502856e-05,
      "loss": 2.102,
      "step": 177799
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.5878385305404663,
      "learning_rate": 7.391569582865022e-05,
      "loss": 2.2046,
      "step": 177800
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.3010615110397339,
      "learning_rate": 7.391300705430534e-05,
      "loss": 1.7454,
      "step": 177801
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.4755988121032715,
      "learning_rate": 7.391031832199428e-05,
      "loss": 2.1412,
      "step": 177802
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.394840955734253,
      "learning_rate": 7.390762963171753e-05,
      "loss": 2.2412,
      "step": 177803
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.4253766536712646,
      "learning_rate": 7.390494098347558e-05,
      "loss": 2.1421,
      "step": 177804
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.4506125450134277,
      "learning_rate": 7.390225237726901e-05,
      "loss": 2.1406,
      "step": 177805
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.2804220914840698,
      "learning_rate": 7.389956381309821e-05,
      "loss": 2.0539,
      "step": 177806
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.1933391094207764,
      "learning_rate": 7.389687529096387e-05,
      "loss": 2.3011,
      "step": 177807
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.2722700834274292,
      "learning_rate": 7.389418681086632e-05,
      "loss": 2.012,
      "step": 177808
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.569509744644165,
      "learning_rate": 7.389149837280615e-05,
      "loss": 2.1085,
      "step": 177809
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.245711326599121,
      "learning_rate": 7.388880997678372e-05,
      "loss": 1.932,
      "step": 177810
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.2744373083114624,
      "learning_rate": 7.388612162279969e-05,
      "loss": 2.0011,
      "step": 177811
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.5389326810836792,
      "learning_rate": 7.388343331085444e-05,
      "loss": 2.2416,
      "step": 177812
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.348605751991272,
      "learning_rate": 7.388074504094859e-05,
      "loss": 1.9408,
      "step": 177813
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.3448383808135986,
      "learning_rate": 7.387805681308261e-05,
      "loss": 1.9875,
      "step": 177814
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.376396656036377,
      "learning_rate": 7.387536862725694e-05,
      "loss": 2.1575,
      "step": 177815
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.525607705116272,
      "learning_rate": 7.387268048347203e-05,
      "loss": 2.0501,
      "step": 177816
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.3242590427398682,
      "learning_rate": 7.386999238172854e-05,
      "loss": 2.1217,
      "step": 177817
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.259123682975769,
      "learning_rate": 7.386730432202681e-05,
      "loss": 2.1238,
      "step": 177818
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.3046950101852417,
      "learning_rate": 7.386461630436748e-05,
      "loss": 2.0537,
      "step": 177819
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.3672895431518555,
      "learning_rate": 7.386192832875099e-05,
      "loss": 2.1496,
      "step": 177820
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.313021183013916,
      "learning_rate": 7.385924039517787e-05,
      "loss": 1.895,
      "step": 177821
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.3103082180023193,
      "learning_rate": 7.38565525036485e-05,
      "loss": 1.9693,
      "step": 177822
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.4667035341262817,
      "learning_rate": 7.385386465416351e-05,
      "loss": 2.0551,
      "step": 177823
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.2719929218292236,
      "learning_rate": 7.38511768467233e-05,
      "loss": 2.0801,
      "step": 177824
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.4179848432540894,
      "learning_rate": 7.384848908132851e-05,
      "loss": 2.2299,
      "step": 177825
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.2636274099349976,
      "learning_rate": 7.384580135797943e-05,
      "loss": 1.991,
      "step": 177826
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.4344230890274048,
      "learning_rate": 7.384311367667691e-05,
      "loss": 2.248,
      "step": 177827
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.3524035215377808,
      "learning_rate": 7.384042603742102e-05,
      "loss": 2.0357,
      "step": 177828
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.3160730600357056,
      "learning_rate": 7.383773844021255e-05,
      "loss": 1.9216,
      "step": 177829
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.2986456155776978,
      "learning_rate": 7.383505088505186e-05,
      "loss": 1.7864,
      "step": 177830
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.313850998878479,
      "learning_rate": 7.383236337193955e-05,
      "loss": 1.8998,
      "step": 177831
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.3521488904953003,
      "learning_rate": 7.382967590087601e-05,
      "loss": 2.0662,
      "step": 177832
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.4608036279678345,
      "learning_rate": 7.382698847186196e-05,
      "loss": 1.9268,
      "step": 177833
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.4031782150268555,
      "learning_rate": 7.382430108489759e-05,
      "loss": 1.9128,
      "step": 177834
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.3421604633331299,
      "learning_rate": 7.382161373998364e-05,
      "loss": 2.0911,
      "step": 177835
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.377105951309204,
      "learning_rate": 7.38189264371204e-05,
      "loss": 2.0779,
      "step": 177836
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.3849217891693115,
      "learning_rate": 7.381623917630859e-05,
      "loss": 1.9659,
      "step": 177837
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.27668035030365,
      "learning_rate": 7.381355195754855e-05,
      "loss": 1.857,
      "step": 177838
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.34395170211792,
      "learning_rate": 7.3810864780841e-05,
      "loss": 2.0962,
      "step": 177839
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.25692617893219,
      "learning_rate": 7.380817764618607e-05,
      "loss": 1.9102,
      "step": 177840
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.2906347513198853,
      "learning_rate": 7.380549055358461e-05,
      "loss": 1.6909,
      "step": 177841
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.274190068244934,
      "learning_rate": 7.380280350303687e-05,
      "loss": 1.9649,
      "step": 177842
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.4763094186782837,
      "learning_rate": 7.380011649454354e-05,
      "loss": 1.8053,
      "step": 177843
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.5053887367248535,
      "learning_rate": 7.379742952810496e-05,
      "loss": 1.9524,
      "step": 177844
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.2930943965911865,
      "learning_rate": 7.379474260372187e-05,
      "loss": 2.1541,
      "step": 177845
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.3798725605010986,
      "learning_rate": 7.379205572139443e-05,
      "loss": 1.8821,
      "step": 177846
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.437092661857605,
      "learning_rate": 7.378936888112344e-05,
      "loss": 2.0176,
      "step": 177847
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.2936666011810303,
      "learning_rate": 7.378668208290916e-05,
      "loss": 2.1898,
      "step": 177848
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.395232081413269,
      "learning_rate": 7.37839953267523e-05,
      "loss": 2.087,
      "step": 177849
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.3096197843551636,
      "learning_rate": 7.378130861265316e-05,
      "loss": 1.9578,
      "step": 177850
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.3521684408187866,
      "learning_rate": 7.377862194061256e-05,
      "loss": 2.3516,
      "step": 177851
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.2399686574935913,
      "learning_rate": 7.377593531063057e-05,
      "loss": 2.0,
      "step": 177852
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.582914113998413,
      "learning_rate": 7.377324872270801e-05,
      "loss": 1.9835,
      "step": 177853
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.458038091659546,
      "learning_rate": 7.37705621768452e-05,
      "loss": 1.8545,
      "step": 177854
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.3181737661361694,
      "learning_rate": 7.376787567304279e-05,
      "loss": 1.8991,
      "step": 177855
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.5685900449752808,
      "learning_rate": 7.37651892113011e-05,
      "loss": 2.1389,
      "step": 177856
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.2268788814544678,
      "learning_rate": 7.376250279162091e-05,
      "loss": 2.1371,
      "step": 177857
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.4911869764328003,
      "learning_rate": 7.37598164140024e-05,
      "loss": 2.0895,
      "step": 177858
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.354113221168518,
      "learning_rate": 7.375713007844626e-05,
      "loss": 1.8922,
      "step": 177859
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.3876327276229858,
      "learning_rate": 7.375444378495285e-05,
      "loss": 2.172,
      "step": 177860
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.5262812376022339,
      "learning_rate": 7.375175753352285e-05,
      "loss": 2.1933,
      "step": 177861
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.378074288368225,
      "learning_rate": 7.374907132415661e-05,
      "loss": 1.9645,
      "step": 177862
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.6256728172302246,
      "learning_rate": 7.374638515685474e-05,
      "loss": 1.8026,
      "step": 177863
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.4444128274917603,
      "learning_rate": 7.374369903161771e-05,
      "loss": 2.0068,
      "step": 177864
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.399636149406433,
      "learning_rate": 7.374101294844598e-05,
      "loss": 1.9844,
      "step": 177865
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.3495218753814697,
      "learning_rate": 7.373832690734002e-05,
      "loss": 1.91,
      "step": 177866
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.3471448421478271,
      "learning_rate": 7.373564090830043e-05,
      "loss": 1.91,
      "step": 177867
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.715500831604004,
      "learning_rate": 7.373295495132756e-05,
      "loss": 2.1607,
      "step": 177868
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.8773164749145508,
      "learning_rate": 7.37302690364221e-05,
      "loss": 2.0227,
      "step": 177869
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.4330216646194458,
      "learning_rate": 7.372758316358447e-05,
      "loss": 2.1772,
      "step": 177870
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.1909903287887573,
      "learning_rate": 7.372489733281504e-05,
      "loss": 1.9239,
      "step": 177871
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.4264789819717407,
      "learning_rate": 7.372221154411454e-05,
      "loss": 2.08,
      "step": 177872
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.3306516408920288,
      "learning_rate": 7.371952579748334e-05,
      "loss": 2.1612,
      "step": 177873
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.2716970443725586,
      "learning_rate": 7.371684009292187e-05,
      "loss": 1.8707,
      "step": 177874
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.4776824712753296,
      "learning_rate": 7.371415443043077e-05,
      "loss": 2.1285,
      "step": 177875
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.3162389993667603,
      "learning_rate": 7.371146881001052e-05,
      "loss": 1.9132,
      "step": 177876
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.229804515838623,
      "learning_rate": 7.370878323166149e-05,
      "loss": 1.8687,
      "step": 177877
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.339314579963684,
      "learning_rate": 7.370609769538432e-05,
      "loss": 2.1352,
      "step": 177878
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.3354415893554688,
      "learning_rate": 7.370341220117948e-05,
      "loss": 2.1731,
      "step": 177879
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.5123422145843506,
      "learning_rate": 7.370072674904735e-05,
      "loss": 2.1295,
      "step": 177880
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.4550358057022095,
      "learning_rate": 7.369804133898863e-05,
      "loss": 2.1418,
      "step": 177881
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.2313755750656128,
      "learning_rate": 7.369535597100371e-05,
      "loss": 2.0389,
      "step": 177882
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.2976900339126587,
      "learning_rate": 7.3692670645093e-05,
      "loss": 1.8154,
      "step": 177883
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.222029685974121,
      "learning_rate": 7.368998536125718e-05,
      "loss": 1.9345,
      "step": 177884
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.3438270092010498,
      "learning_rate": 7.368730011949657e-05,
      "loss": 2.0035,
      "step": 177885
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.5706759691238403,
      "learning_rate": 7.368461491981187e-05,
      "loss": 1.9657,
      "step": 177886
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.3385260105133057,
      "learning_rate": 7.368192976220347e-05,
      "loss": 1.9961,
      "step": 177887
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.2915849685668945,
      "learning_rate": 7.367924464667188e-05,
      "loss": 2.1102,
      "step": 177888
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.3163267374038696,
      "learning_rate": 7.367655957321748e-05,
      "loss": 2.1107,
      "step": 177889
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.3487519025802612,
      "learning_rate": 7.367387454184097e-05,
      "loss": 1.9434,
      "step": 177890
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.2864494323730469,
      "learning_rate": 7.367118955254268e-05,
      "loss": 1.8835,
      "step": 177891
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.332485318183899,
      "learning_rate": 7.366850460532329e-05,
      "loss": 1.9943,
      "step": 177892
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.3476526737213135,
      "learning_rate": 7.366581970018308e-05,
      "loss": 2.0068,
      "step": 177893
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.36899733543396,
      "learning_rate": 7.366313483712286e-05,
      "loss": 1.8068,
      "step": 177894
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.399605631828308,
      "learning_rate": 7.366045001614275e-05,
      "loss": 2.2047,
      "step": 177895
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.334622859954834,
      "learning_rate": 7.365776523724353e-05,
      "loss": 1.9421,
      "step": 177896
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.2491949796676636,
      "learning_rate": 7.365508050042554e-05,
      "loss": 1.8701,
      "step": 177897
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.4125005006790161,
      "learning_rate": 7.365239580568938e-05,
      "loss": 1.8543,
      "step": 177898
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.4199875593185425,
      "learning_rate": 7.364971115303547e-05,
      "loss": 1.9147,
      "step": 177899
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.4687352180480957,
      "learning_rate": 7.364702654246448e-05,
      "loss": 1.9961,
      "step": 177900
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.3712449073791504,
      "learning_rate": 7.364434197397664e-05,
      "loss": 1.9686,
      "step": 177901
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.5605045557022095,
      "learning_rate": 7.364165744757267e-05,
      "loss": 1.9353,
      "step": 177902
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.4038634300231934,
      "learning_rate": 7.36389729632529e-05,
      "loss": 2.0294,
      "step": 177903
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.3382681608200073,
      "learning_rate": 7.363628852101799e-05,
      "loss": 1.994,
      "step": 177904
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.2809630632400513,
      "learning_rate": 7.363360412086833e-05,
      "loss": 2.3103,
      "step": 177905
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.2260900735855103,
      "learning_rate": 7.36309197628046e-05,
      "loss": 2.1846,
      "step": 177906
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.3203871250152588,
      "learning_rate": 7.362823544682697e-05,
      "loss": 2.0507,
      "step": 177907
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.5559449195861816,
      "learning_rate": 7.36255511729362e-05,
      "loss": 2.1908,
      "step": 177908
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.6970374584197998,
      "learning_rate": 7.362286694113265e-05,
      "loss": 1.8435,
      "step": 177909
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.2827099561691284,
      "learning_rate": 7.362018275141698e-05,
      "loss": 1.9994,
      "step": 177910
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.524160623550415,
      "learning_rate": 7.361749860378948e-05,
      "loss": 2.0557,
      "step": 177911
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.47612464427948,
      "learning_rate": 7.361481449825091e-05,
      "loss": 2.1571,
      "step": 177912
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.4408663511276245,
      "learning_rate": 7.361213043480148e-05,
      "loss": 1.9537,
      "step": 177913
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.3719396591186523,
      "learning_rate": 7.360944641344192e-05,
      "loss": 2.042,
      "step": 177914
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.3177194595336914,
      "learning_rate": 7.360676243417255e-05,
      "loss": 2.0622,
      "step": 177915
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.4068506956100464,
      "learning_rate": 7.360407849699404e-05,
      "loss": 2.0316,
      "step": 177916
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.2061402797698975,
      "learning_rate": 7.360139460190671e-05,
      "loss": 1.5854,
      "step": 177917
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.737790584564209,
      "learning_rate": 7.359871074891133e-05,
      "loss": 1.904,
      "step": 177918
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.4608609676361084,
      "learning_rate": 7.359602693800805e-05,
      "loss": 2.1473,
      "step": 177919
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.366882085800171,
      "learning_rate": 7.359334316919763e-05,
      "loss": 1.8947,
      "step": 177920
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.2801154851913452,
      "learning_rate": 7.35906594424804e-05,
      "loss": 1.9609,
      "step": 177921
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.3333433866500854,
      "learning_rate": 7.358797575785704e-05,
      "loss": 1.9537,
      "step": 177922
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.4230588674545288,
      "learning_rate": 7.358529211532784e-05,
      "loss": 1.9419,
      "step": 177923
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.926194667816162,
      "learning_rate": 7.358260851489357e-05,
      "loss": 2.0976,
      "step": 177924
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.411942958831787,
      "learning_rate": 7.357992495655443e-05,
      "loss": 2.0629,
      "step": 177925
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.3007733821868896,
      "learning_rate": 7.357724144031111e-05,
      "loss": 2.1764,
      "step": 177926
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.336007833480835,
      "learning_rate": 7.357455796616398e-05,
      "loss": 2.2884,
      "step": 177927
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.525278925895691,
      "learning_rate": 7.357187453411372e-05,
      "loss": 1.9489,
      "step": 177928
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.3443855047225952,
      "learning_rate": 7.356919114416061e-05,
      "loss": 2.1145,
      "step": 177929
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.5579689741134644,
      "learning_rate": 7.356650779630543e-05,
      "loss": 2.0458,
      "step": 177930
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.3212355375289917,
      "learning_rate": 7.356382449054837e-05,
      "loss": 1.9109,
      "step": 177931
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.3656944036483765,
      "learning_rate": 7.356114122689014e-05,
      "loss": 2.059,
      "step": 177932
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.3696781396865845,
      "learning_rate": 7.355845800533107e-05,
      "loss": 2.145,
      "step": 177933
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.4041467905044556,
      "learning_rate": 7.355577482587185e-05,
      "loss": 2.0454,
      "step": 177934
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.3859879970550537,
      "learning_rate": 7.355309168851279e-05,
      "loss": 2.0363,
      "step": 177935
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.34180748462677,
      "learning_rate": 7.355040859325469e-05,
      "loss": 2.0999,
      "step": 177936
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.3387839794158936,
      "learning_rate": 7.354772554009764e-05,
      "loss": 2.1883,
      "step": 177937
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.2961450815200806,
      "learning_rate": 7.354504252904244e-05,
      "loss": 2.2027,
      "step": 177938
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.3967537879943848,
      "learning_rate": 7.35423595600894e-05,
      "loss": 2.08,
      "step": 177939
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.3795888423919678,
      "learning_rate": 7.353967663323919e-05,
      "loss": 1.9275,
      "step": 177940
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.3041284084320068,
      "learning_rate": 7.353699374849215e-05,
      "loss": 1.9821,
      "step": 177941
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.3157265186309814,
      "learning_rate": 7.353431090584904e-05,
      "loss": 1.8969,
      "step": 177942
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.2782821655273438,
      "learning_rate": 7.353162810530999e-05,
      "loss": 2.0097,
      "step": 177943
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.2857681512832642,
      "learning_rate": 7.352894534687579e-05,
      "loss": 2.0208,
      "step": 177944
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.4120280742645264,
      "learning_rate": 7.352626263054675e-05,
      "loss": 1.9885,
      "step": 177945
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.4570506811141968,
      "learning_rate": 7.35235799563235e-05,
      "loss": 2.2595,
      "step": 177946
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.3375636339187622,
      "learning_rate": 7.352089732420644e-05,
      "loss": 1.886,
      "step": 177947
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.3479477167129517,
      "learning_rate": 7.35182147341962e-05,
      "loss": 1.9193,
      "step": 177948
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.3359774351119995,
      "learning_rate": 7.351553218629319e-05,
      "loss": 2.3299,
      "step": 177949
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.260270118713379,
      "learning_rate": 7.35128496804979e-05,
      "loss": 2.0475,
      "step": 177950
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.375870943069458,
      "learning_rate": 7.351016721681078e-05,
      "loss": 1.8349,
      "step": 177951
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.3179078102111816,
      "learning_rate": 7.35074847952325e-05,
      "loss": 1.9217,
      "step": 177952
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.2266697883605957,
      "learning_rate": 7.350480241576336e-05,
      "loss": 1.9468,
      "step": 177953
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.1735824346542358,
      "learning_rate": 7.350212007840402e-05,
      "loss": 2.0802,
      "step": 177954
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.2416101694107056,
      "learning_rate": 7.349943778315491e-05,
      "loss": 1.9018,
      "step": 177955
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.2917771339416504,
      "learning_rate": 7.349675553001648e-05,
      "loss": 1.8825,
      "step": 177956
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.3189717531204224,
      "learning_rate": 7.349407331898933e-05,
      "loss": 2.2125,
      "step": 177957
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.4206591844558716,
      "learning_rate": 7.349139115007392e-05,
      "loss": 2.1082,
      "step": 177958
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.2053769826889038,
      "learning_rate": 7.348870902327064e-05,
      "loss": 2.0326,
      "step": 177959
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.3131380081176758,
      "learning_rate": 7.34860269385802e-05,
      "loss": 1.9946,
      "step": 177960
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.2848780155181885,
      "learning_rate": 7.348334489600297e-05,
      "loss": 2.0401,
      "step": 177961
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.3584198951721191,
      "learning_rate": 7.348066289553936e-05,
      "loss": 1.7992,
      "step": 177962
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.2150182723999023,
      "learning_rate": 7.347798093719005e-05,
      "loss": 1.9389,
      "step": 177963
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.4829872846603394,
      "learning_rate": 7.34752990209555e-05,
      "loss": 2.0863,
      "step": 177964
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.3410950899124146,
      "learning_rate": 7.347261714683604e-05,
      "loss": 1.9919,
      "step": 177965
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.4323710203170776,
      "learning_rate": 7.34699353148324e-05,
      "loss": 1.9236,
      "step": 177966
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.2435870170593262,
      "learning_rate": 7.346725352494498e-05,
      "loss": 1.8457,
      "step": 177967
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.261665940284729,
      "learning_rate": 7.346457177717416e-05,
      "loss": 2.0289,
      "step": 177968
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.1979551315307617,
      "learning_rate": 7.346189007152069e-05,
      "loss": 1.6619,
      "step": 177969
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.5213333368301392,
      "learning_rate": 7.345920840798479e-05,
      "loss": 1.9392,
      "step": 177970
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.286911129951477,
      "learning_rate": 7.345652678656724e-05,
      "loss": 2.0808,
      "step": 177971
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.226067066192627,
      "learning_rate": 7.345384520726835e-05,
      "loss": 1.9729,
      "step": 177972
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.3370000123977661,
      "learning_rate": 7.345116367008868e-05,
      "loss": 2.197,
      "step": 177973
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.2612171173095703,
      "learning_rate": 7.344848217502866e-05,
      "loss": 2.1592,
      "step": 177974
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.2925224304199219,
      "learning_rate": 7.34458007220889e-05,
      "loss": 1.7776,
      "step": 177975
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.4129304885864258,
      "learning_rate": 7.344311931126975e-05,
      "loss": 2.0311,
      "step": 177976
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.255558967590332,
      "learning_rate": 7.344043794257191e-05,
      "loss": 1.9347,
      "step": 177977
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.6154791116714478,
      "learning_rate": 7.343775661599574e-05,
      "loss": 2.0491,
      "step": 177978
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.2818677425384521,
      "learning_rate": 7.343507533154178e-05,
      "loss": 1.8733,
      "step": 177979
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.4091691970825195,
      "learning_rate": 7.343239408921044e-05,
      "loss": 1.9352,
      "step": 177980
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.2952066659927368,
      "learning_rate": 7.342971288900235e-05,
      "loss": 2.1201,
      "step": 177981
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.2396807670593262,
      "learning_rate": 7.34270317309179e-05,
      "loss": 1.8759,
      "step": 177982
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.3419759273529053,
      "learning_rate": 7.342435061495773e-05,
      "loss": 1.8776,
      "step": 177983
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.2864246368408203,
      "learning_rate": 7.342166954112216e-05,
      "loss": 1.8253,
      "step": 177984
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.3736276626586914,
      "learning_rate": 7.341898850941195e-05,
      "loss": 2.1075,
      "step": 177985
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.270171046257019,
      "learning_rate": 7.341630751982725e-05,
      "loss": 2.0425,
      "step": 177986
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.2602078914642334,
      "learning_rate": 7.341362657236881e-05,
      "loss": 2.2812,
      "step": 177987
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.3185725212097168,
      "learning_rate": 7.341094566703698e-05,
      "loss": 1.8696,
      "step": 177988
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.3028161525726318,
      "learning_rate": 7.340826480383241e-05,
      "loss": 2.0353,
      "step": 177989
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.226847767829895,
      "learning_rate": 7.340558398275546e-05,
      "loss": 1.9899,
      "step": 177990
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.3636500835418701,
      "learning_rate": 7.340290320380683e-05,
      "loss": 2.015,
      "step": 177991
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.6074551343917847,
      "learning_rate": 7.340022246698674e-05,
      "loss": 2.1503,
      "step": 177992
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.424492597579956,
      "learning_rate": 7.339754177229588e-05,
      "loss": 2.1829,
      "step": 177993
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.3087579011917114,
      "learning_rate": 7.339486111973461e-05,
      "loss": 1.9328,
      "step": 177994
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.27816903591156,
      "learning_rate": 7.339218050930362e-05,
      "loss": 2.1093,
      "step": 177995
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.2280908823013306,
      "learning_rate": 7.33894999410032e-05,
      "loss": 1.9675,
      "step": 177996
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.3643627166748047,
      "learning_rate": 7.338681941483413e-05,
      "loss": 1.8428,
      "step": 177997
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.4893543720245361,
      "learning_rate": 7.338413893079653e-05,
      "loss": 2.0003,
      "step": 177998
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.4817523956298828,
      "learning_rate": 7.338145848889123e-05,
      "loss": 2.0207,
      "step": 177999
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.4672590494155884,
      "learning_rate": 7.337877808911846e-05,
      "loss": 2.0663,
      "step": 178000
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.348736047744751,
      "learning_rate": 7.3376097731479e-05,
      "loss": 1.9447,
      "step": 178001
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.2253862619400024,
      "learning_rate": 7.337341741597305e-05,
      "loss": 2.0155,
      "step": 178002
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.2964788675308228,
      "learning_rate": 7.337073714260146e-05,
      "loss": 1.7727,
      "step": 178003
    },
    {
      "epoch": 2.32,
      "grad_norm": 2.0437750816345215,
      "learning_rate": 7.336805691136435e-05,
      "loss": 1.881,
      "step": 178004
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.412124514579773,
      "learning_rate": 7.336537672226249e-05,
      "loss": 2.1524,
      "step": 178005
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.427515983581543,
      "learning_rate": 7.33626965752962e-05,
      "loss": 1.9915,
      "step": 178006
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.3175588846206665,
      "learning_rate": 7.336001647046613e-05,
      "loss": 2.1787,
      "step": 178007
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.5656936168670654,
      "learning_rate": 7.335733640777266e-05,
      "loss": 1.9804,
      "step": 178008
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.6229095458984375,
      "learning_rate": 7.335465638721649e-05,
      "loss": 1.9334,
      "step": 178009
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.3066507577896118,
      "learning_rate": 7.335197640879779e-05,
      "loss": 1.8347,
      "step": 178010
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.2184566259384155,
      "learning_rate": 7.33492964725173e-05,
      "loss": 2.2671,
      "step": 178011
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.5846357345581055,
      "learning_rate": 7.33466165783754e-05,
      "loss": 2.1965,
      "step": 178012
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.2758363485336304,
      "learning_rate": 7.334393672637273e-05,
      "loss": 2.0256,
      "step": 178013
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.20795738697052,
      "learning_rate": 7.334125691650962e-05,
      "loss": 1.8425,
      "step": 178014
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.3561768531799316,
      "learning_rate": 7.333857714878679e-05,
      "loss": 2.1534,
      "step": 178015
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.4826608896255493,
      "learning_rate": 7.333589742320446e-05,
      "loss": 2.088,
      "step": 178016
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.254926323890686,
      "learning_rate": 7.333321773976335e-05,
      "loss": 2.1361,
      "step": 178017
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.2925597429275513,
      "learning_rate": 7.333053809846377e-05,
      "loss": 1.9775,
      "step": 178018
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.3565322160720825,
      "learning_rate": 7.33278584993064e-05,
      "loss": 2.086,
      "step": 178019
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.5980762243270874,
      "learning_rate": 7.33251789422916e-05,
      "loss": 2.0016,
      "step": 178020
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.3347526788711548,
      "learning_rate": 7.332249942742007e-05,
      "loss": 2.0723,
      "step": 178021
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.3881416320800781,
      "learning_rate": 7.331981995469201e-05,
      "loss": 1.7859,
      "step": 178022
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.2640973329544067,
      "learning_rate": 7.331714052410818e-05,
      "loss": 1.9681,
      "step": 178023
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.29977548122406,
      "learning_rate": 7.331446113566886e-05,
      "loss": 1.8173,
      "step": 178024
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.3307043313980103,
      "learning_rate": 7.331178178937472e-05,
      "loss": 1.8753,
      "step": 178025
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.272831916809082,
      "learning_rate": 7.330910248522618e-05,
      "loss": 1.989,
      "step": 178026
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.3675862550735474,
      "learning_rate": 7.330642322322379e-05,
      "loss": 1.8654,
      "step": 178027
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.5350278615951538,
      "learning_rate": 7.330374400336804e-05,
      "loss": 1.9554,
      "step": 178028
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.4748932123184204,
      "learning_rate": 7.330106482565939e-05,
      "loss": 2.223,
      "step": 178029
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.2750202417373657,
      "learning_rate": 7.32983856900983e-05,
      "loss": 1.7104,
      "step": 178030
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.6083124876022339,
      "learning_rate": 7.329570659668535e-05,
      "loss": 2.0981,
      "step": 178031
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.5684059858322144,
      "learning_rate": 7.329302754542095e-05,
      "loss": 1.8234,
      "step": 178032
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.3547171354293823,
      "learning_rate": 7.329034853630573e-05,
      "loss": 1.8858,
      "step": 178033
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.294372797012329,
      "learning_rate": 7.328766956934014e-05,
      "loss": 2.1432,
      "step": 178034
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.374447226524353,
      "learning_rate": 7.328499064452462e-05,
      "loss": 1.8992,
      "step": 178035
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.383256435394287,
      "learning_rate": 7.328231176185966e-05,
      "loss": 1.9984,
      "step": 178036
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.4002114534378052,
      "learning_rate": 7.327963292134584e-05,
      "loss": 2.1012,
      "step": 178037
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.4463655948638916,
      "learning_rate": 7.327695412298355e-05,
      "loss": 1.9443,
      "step": 178038
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.249847650527954,
      "learning_rate": 7.327427536677343e-05,
      "loss": 1.8291,
      "step": 178039
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.2480030059814453,
      "learning_rate": 7.327159665271589e-05,
      "loss": 2.023,
      "step": 178040
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.4035981893539429,
      "learning_rate": 7.326891798081147e-05,
      "loss": 2.0261,
      "step": 178041
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.2764836549758911,
      "learning_rate": 7.326623935106052e-05,
      "loss": 1.7689,
      "step": 178042
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.3953230381011963,
      "learning_rate": 7.326356076346377e-05,
      "loss": 1.8309,
      "step": 178043
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.4437813758850098,
      "learning_rate": 7.326088221802152e-05,
      "loss": 2.0186,
      "step": 178044
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.2489798069000244,
      "learning_rate": 7.325820371473442e-05,
      "loss": 2.1359,
      "step": 178045
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.283130407333374,
      "learning_rate": 7.32555252536029e-05,
      "loss": 2.0836,
      "step": 178046
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.374478816986084,
      "learning_rate": 7.325284683462739e-05,
      "loss": 2.0763,
      "step": 178047
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.3324259519577026,
      "learning_rate": 7.325016845780853e-05,
      "loss": 2.0014,
      "step": 178048
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.316361665725708,
      "learning_rate": 7.324749012314674e-05,
      "loss": 2.0191,
      "step": 178049
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.1741669178009033,
      "learning_rate": 7.324481183064243e-05,
      "loss": 1.9299,
      "step": 178050
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.2802562713623047,
      "learning_rate": 7.32421335802963e-05,
      "loss": 2.0403,
      "step": 178051
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.3555881977081299,
      "learning_rate": 7.323945537210871e-05,
      "loss": 2.0329,
      "step": 178052
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.2405803203582764,
      "learning_rate": 7.323677720608012e-05,
      "loss": 2.0377,
      "step": 178053
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.4334012269973755,
      "learning_rate": 7.323409908221118e-05,
      "loss": 1.9249,
      "step": 178054
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.4070830345153809,
      "learning_rate": 7.323142100050219e-05,
      "loss": 1.8448,
      "step": 178055
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.6611607074737549,
      "learning_rate": 7.322874296095386e-05,
      "loss": 2.0005,
      "step": 178056
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.2922683954238892,
      "learning_rate": 7.322606496356662e-05,
      "loss": 2.1029,
      "step": 178057
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.4628218412399292,
      "learning_rate": 7.322338700834089e-05,
      "loss": 1.9581,
      "step": 178058
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.3269723653793335,
      "learning_rate": 7.322070909527716e-05,
      "loss": 2.1159,
      "step": 178059
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.428321361541748,
      "learning_rate": 7.321803122437603e-05,
      "loss": 2.1259,
      "step": 178060
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.3301122188568115,
      "learning_rate": 7.321535339563788e-05,
      "loss": 2.2522,
      "step": 178061
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.3326009511947632,
      "learning_rate": 7.321267560906337e-05,
      "loss": 1.9531,
      "step": 178062
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.3113740682601929,
      "learning_rate": 7.32099978646529e-05,
      "loss": 1.8755,
      "step": 178063
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.3257865905761719,
      "learning_rate": 7.3207320162407e-05,
      "loss": 2.0687,
      "step": 178064
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.1898411512374878,
      "learning_rate": 7.3204642502326e-05,
      "loss": 1.8493,
      "step": 178065
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.4095441102981567,
      "learning_rate": 7.320196488441067e-05,
      "loss": 1.916,
      "step": 178066
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.525780200958252,
      "learning_rate": 7.319928730866125e-05,
      "loss": 2.0412,
      "step": 178067
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.2078903913497925,
      "learning_rate": 7.319660977507846e-05,
      "loss": 2.2412,
      "step": 178068
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.2857582569122314,
      "learning_rate": 7.319393228366265e-05,
      "loss": 2.1153,
      "step": 178069
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.3441165685653687,
      "learning_rate": 7.31912548344145e-05,
      "loss": 2.1051,
      "step": 178070
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.3391854763031006,
      "learning_rate": 7.318857742733422e-05,
      "loss": 1.953,
      "step": 178071
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.2926173210144043,
      "learning_rate": 7.318590006242255e-05,
      "loss": 2.0405,
      "step": 178072
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.3170371055603027,
      "learning_rate": 7.318322273967984e-05,
      "loss": 2.2055,
      "step": 178073
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.2821476459503174,
      "learning_rate": 7.318054545910672e-05,
      "loss": 2.275,
      "step": 178074
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.27925443649292,
      "learning_rate": 7.317786822070351e-05,
      "loss": 1.947,
      "step": 178075
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.3101086616516113,
      "learning_rate": 7.3175191024471e-05,
      "loss": 2.1897,
      "step": 178076
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.271066665649414,
      "learning_rate": 7.317251387040937e-05,
      "loss": 2.1872,
      "step": 178077
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.2243789434432983,
      "learning_rate": 7.31698367585193e-05,
      "loss": 2.1612,
      "step": 178078
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.2481653690338135,
      "learning_rate": 7.316715968880116e-05,
      "loss": 1.7487,
      "step": 178079
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.2921242713928223,
      "learning_rate": 7.316448266125559e-05,
      "loss": 1.9989,
      "step": 178080
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.246687889099121,
      "learning_rate": 7.316180567588298e-05,
      "loss": 2.0117,
      "step": 178081
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.3978431224822998,
      "learning_rate": 7.3159128732684e-05,
      "loss": 2.19,
      "step": 178082
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.434862732887268,
      "learning_rate": 7.31564518316589e-05,
      "loss": 2.0465,
      "step": 178083
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.2313437461853027,
      "learning_rate": 7.315377497280835e-05,
      "loss": 1.9721,
      "step": 178084
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.3493313789367676,
      "learning_rate": 7.31510981561327e-05,
      "loss": 1.9971,
      "step": 178085
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.5560166835784912,
      "learning_rate": 7.314842138163264e-05,
      "loss": 1.9137,
      "step": 178086
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.4574693441390991,
      "learning_rate": 7.314574464930847e-05,
      "loss": 1.9835,
      "step": 178087
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.2984490394592285,
      "learning_rate": 7.314306795916102e-05,
      "loss": 1.8717,
      "step": 178088
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.368859052658081,
      "learning_rate": 7.314039131119031e-05,
      "loss": 2.206,
      "step": 178089
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.418837070465088,
      "learning_rate": 7.313771470539721e-05,
      "loss": 1.9352,
      "step": 178090
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.2313830852508545,
      "learning_rate": 7.313503814178199e-05,
      "loss": 2.0222,
      "step": 178091
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.3406200408935547,
      "learning_rate": 7.313236162034535e-05,
      "loss": 1.8308,
      "step": 178092
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.342307448387146,
      "learning_rate": 7.312968514108759e-05,
      "loss": 1.9764,
      "step": 178093
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.4139692783355713,
      "learning_rate": 7.312700870400939e-05,
      "loss": 2.1313,
      "step": 178094
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.2033822536468506,
      "learning_rate": 7.312433230911119e-05,
      "loss": 2.0823,
      "step": 178095
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.444838285446167,
      "learning_rate": 7.31216559563934e-05,
      "loss": 2.2629,
      "step": 178096
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.220197081565857,
      "learning_rate": 7.311897964585656e-05,
      "loss": 2.1136,
      "step": 178097
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.295022964477539,
      "learning_rate": 7.311630337750123e-05,
      "loss": 2.0318,
      "step": 178098
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.4768497943878174,
      "learning_rate": 7.311362715132775e-05,
      "loss": 2.2566,
      "step": 178099
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.371753215789795,
      "learning_rate": 7.311095096733688e-05,
      "loss": 2.0117,
      "step": 178100
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.3435404300689697,
      "learning_rate": 7.310827482552892e-05,
      "loss": 2.1204,
      "step": 178101
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.2738641500473022,
      "learning_rate": 7.310559872590443e-05,
      "loss": 1.9054,
      "step": 178102
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.3844623565673828,
      "learning_rate": 7.310292266846381e-05,
      "loss": 1.8971,
      "step": 178103
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.3233457803726196,
      "learning_rate": 7.31002466532077e-05,
      "loss": 1.9018,
      "step": 178104
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.4254710674285889,
      "learning_rate": 7.309757068013647e-05,
      "loss": 2.2358,
      "step": 178105
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.2823644876480103,
      "learning_rate": 7.30948947492508e-05,
      "loss": 1.9323,
      "step": 178106
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.1384813785552979,
      "learning_rate": 7.309221886055103e-05,
      "loss": 1.8881,
      "step": 178107
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.3870797157287598,
      "learning_rate": 7.30895430140377e-05,
      "loss": 2.0867,
      "step": 178108
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.421745777130127,
      "learning_rate": 7.308686720971125e-05,
      "loss": 1.991,
      "step": 178109
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.4766703844070435,
      "learning_rate": 7.30841914475723e-05,
      "loss": 2.2454,
      "step": 178110
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.440382719039917,
      "learning_rate": 7.308151572762119e-05,
      "loss": 1.9429,
      "step": 178111
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.3688768148422241,
      "learning_rate": 7.307884004985863e-05,
      "loss": 1.9339,
      "step": 178112
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.3594647645950317,
      "learning_rate": 7.307616441428496e-05,
      "loss": 2.087,
      "step": 178113
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.3959171772003174,
      "learning_rate": 7.307348882090072e-05,
      "loss": 1.9158,
      "step": 178114
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.412224531173706,
      "learning_rate": 7.307081326970636e-05,
      "loss": 2.001,
      "step": 178115
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.2122713327407837,
      "learning_rate": 7.306813776070243e-05,
      "loss": 1.8979,
      "step": 178116
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.4734667539596558,
      "learning_rate": 7.306546229388937e-05,
      "loss": 1.9291,
      "step": 178117
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.3641396760940552,
      "learning_rate": 7.306278686926784e-05,
      "loss": 1.9805,
      "step": 178118
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.3989813327789307,
      "learning_rate": 7.306011148683818e-05,
      "loss": 1.7059,
      "step": 178119
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.3856191635131836,
      "learning_rate": 7.305743614660096e-05,
      "loss": 2.0737,
      "step": 178120
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.2809351682662964,
      "learning_rate": 7.305476084855653e-05,
      "loss": 2.072,
      "step": 178121
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.5392104387283325,
      "learning_rate": 7.305208559270562e-05,
      "loss": 1.9992,
      "step": 178122
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.3150025606155396,
      "learning_rate": 7.304941037904848e-05,
      "loss": 1.8929,
      "step": 178123
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.224582314491272,
      "learning_rate": 7.304673520758586e-05,
      "loss": 2.0564,
      "step": 178124
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.3652722835540771,
      "learning_rate": 7.304406007831812e-05,
      "loss": 2.0493,
      "step": 178125
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.25586998462677,
      "learning_rate": 7.30413849912458e-05,
      "loss": 2.0454,
      "step": 178126
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.316137671470642,
      "learning_rate": 7.303870994636926e-05,
      "loss": 1.8786,
      "step": 178127
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.2433346509933472,
      "learning_rate": 7.303603494368919e-05,
      "loss": 2.1399,
      "step": 178128
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.3842182159423828,
      "learning_rate": 7.303335998320594e-05,
      "loss": 1.9614,
      "step": 178129
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.3393583297729492,
      "learning_rate": 7.303068506492014e-05,
      "loss": 2.1312,
      "step": 178130
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.2466576099395752,
      "learning_rate": 7.302801018883225e-05,
      "loss": 1.9253,
      "step": 178131
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.2470860481262207,
      "learning_rate": 7.302533535494263e-05,
      "loss": 1.9602,
      "step": 178132
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.3418834209442139,
      "learning_rate": 7.302266056325196e-05,
      "loss": 2.2515,
      "step": 178133
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.36831533908844,
      "learning_rate": 7.301998581376066e-05,
      "loss": 1.9575,
      "step": 178134
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.2786990404129028,
      "learning_rate": 7.301731110646918e-05,
      "loss": 2.0911,
      "step": 178135
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.286662220954895,
      "learning_rate": 7.30146364413781e-05,
      "loss": 1.8643,
      "step": 178136
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.332379698753357,
      "learning_rate": 7.301196181848794e-05,
      "loss": 2.1228,
      "step": 178137
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.3746137619018555,
      "learning_rate": 7.300928723779902e-05,
      "loss": 2.0159,
      "step": 178138
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.2755587100982666,
      "learning_rate": 7.300661269931203e-05,
      "loss": 1.8967,
      "step": 178139
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.3332728147506714,
      "learning_rate": 7.300393820302744e-05,
      "loss": 2.1547,
      "step": 178140
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.2329318523406982,
      "learning_rate": 7.300126374894558e-05,
      "loss": 1.9035,
      "step": 178141
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.5373423099517822,
      "learning_rate": 7.299858933706718e-05,
      "loss": 1.9416,
      "step": 178142
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.4149543046951294,
      "learning_rate": 7.29959149673926e-05,
      "loss": 1.9314,
      "step": 178143
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.4218727350234985,
      "learning_rate": 7.299324063992229e-05,
      "loss": 1.9603,
      "step": 178144
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.3035813570022583,
      "learning_rate": 7.299056635465692e-05,
      "loss": 1.9068,
      "step": 178145
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.4202173948287964,
      "learning_rate": 7.298789211159679e-05,
      "loss": 1.9167,
      "step": 178146
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.221682071685791,
      "learning_rate": 7.29852179107426e-05,
      "loss": 1.8506,
      "step": 178147
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.3805041313171387,
      "learning_rate": 7.298254375209474e-05,
      "loss": 2.185,
      "step": 178148
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.2564077377319336,
      "learning_rate": 7.29798696356537e-05,
      "loss": 1.8311,
      "step": 178149
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.3771554231643677,
      "learning_rate": 7.297719556141988e-05,
      "loss": 1.9753,
      "step": 178150
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.3241493701934814,
      "learning_rate": 7.297452152939401e-05,
      "loss": 1.8868,
      "step": 178151
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.3128091096878052,
      "learning_rate": 7.297184753957638e-05,
      "loss": 1.9348,
      "step": 178152
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.3610427379608154,
      "learning_rate": 7.296917359196761e-05,
      "loss": 2.0901,
      "step": 178153
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.2703810930252075,
      "learning_rate": 7.296649968656812e-05,
      "loss": 2.0079,
      "step": 178154
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.3504600524902344,
      "learning_rate": 7.296382582337858e-05,
      "loss": 1.9643,
      "step": 178155
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.3425034284591675,
      "learning_rate": 7.296115200239919e-05,
      "loss": 2.147,
      "step": 178156
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.1794402599334717,
      "learning_rate": 7.29584782236307e-05,
      "loss": 2.149,
      "step": 178157
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.1922452449798584,
      "learning_rate": 7.295580448707342e-05,
      "loss": 1.8414,
      "step": 178158
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.3379714488983154,
      "learning_rate": 7.295313079272803e-05,
      "loss": 2.0609,
      "step": 178159
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.3553413152694702,
      "learning_rate": 7.295045714059483e-05,
      "loss": 1.9092,
      "step": 178160
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.2812609672546387,
      "learning_rate": 7.294778353067454e-05,
      "loss": 1.9537,
      "step": 178161
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.4036258459091187,
      "learning_rate": 7.294510996296752e-05,
      "loss": 2.1353,
      "step": 178162
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.302899956703186,
      "learning_rate": 7.294243643747433e-05,
      "loss": 2.0003,
      "step": 178163
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.3824725151062012,
      "learning_rate": 7.29397629541953e-05,
      "loss": 2.0545,
      "step": 178164
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.2535983324050903,
      "learning_rate": 7.293708951313117e-05,
      "loss": 2.1047,
      "step": 178165
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.231975793838501,
      "learning_rate": 7.293441611428221e-05,
      "loss": 2.0924,
      "step": 178166
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.4167718887329102,
      "learning_rate": 7.293174275764915e-05,
      "loss": 1.9558,
      "step": 178167
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.548028588294983,
      "learning_rate": 7.292906944323231e-05,
      "loss": 2.0317,
      "step": 178168
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.3778443336486816,
      "learning_rate": 7.292639617103231e-05,
      "loss": 1.8681,
      "step": 178169
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.3514268398284912,
      "learning_rate": 7.292372294104944e-05,
      "loss": 2.1917,
      "step": 178170
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.331578254699707,
      "learning_rate": 7.292104975328446e-05,
      "loss": 1.955,
      "step": 178171
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.3200483322143555,
      "learning_rate": 7.29183766077376e-05,
      "loss": 1.8522,
      "step": 178172
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.2356188297271729,
      "learning_rate": 7.291570350440965e-05,
      "loss": 2.3932,
      "step": 178173
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.2470537424087524,
      "learning_rate": 7.291303044330093e-05,
      "loss": 2.0928,
      "step": 178174
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.4204894304275513,
      "learning_rate": 7.291035742441196e-05,
      "loss": 2.1512,
      "step": 178175
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.2994436025619507,
      "learning_rate": 7.290768444774315e-05,
      "loss": 2.3338,
      "step": 178176
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.2621690034866333,
      "learning_rate": 7.290501151329521e-05,
      "loss": 2.0946,
      "step": 178177
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.3713366985321045,
      "learning_rate": 7.290233862106839e-05,
      "loss": 1.8143,
      "step": 178178
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.2322138547897339,
      "learning_rate": 7.28996657710634e-05,
      "loss": 1.9807,
      "step": 178179
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.3522529602050781,
      "learning_rate": 7.289699296328067e-05,
      "loss": 2.0977,
      "step": 178180
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.2321580648422241,
      "learning_rate": 7.289432019772066e-05,
      "loss": 1.848,
      "step": 178181
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.2721995115280151,
      "learning_rate": 7.28916474743838e-05,
      "loss": 1.8745,
      "step": 178182
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.3374452590942383,
      "learning_rate": 7.288897479327079e-05,
      "loss": 2.0084,
      "step": 178183
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.3173835277557373,
      "learning_rate": 7.288630215438188e-05,
      "loss": 2.1365,
      "step": 178184
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.3869937658309937,
      "learning_rate": 7.288362955771777e-05,
      "loss": 1.8532,
      "step": 178185
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.3758766651153564,
      "learning_rate": 7.288095700327892e-05,
      "loss": 2.064,
      "step": 178186
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.4176467657089233,
      "learning_rate": 7.287828449106576e-05,
      "loss": 2.1926,
      "step": 178187
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.2363791465759277,
      "learning_rate": 7.287561202107875e-05,
      "loss": 2.154,
      "step": 178188
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.3338719606399536,
      "learning_rate": 7.287293959331849e-05,
      "loss": 1.9766,
      "step": 178189
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.2721748352050781,
      "learning_rate": 7.28702672077854e-05,
      "loss": 2.1634,
      "step": 178190
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.271120309829712,
      "learning_rate": 7.286759486448008e-05,
      "loss": 2.0406,
      "step": 178191
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.2137199640274048,
      "learning_rate": 7.286492256340296e-05,
      "loss": 1.8626,
      "step": 178192
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.246006965637207,
      "learning_rate": 7.286225030455456e-05,
      "loss": 1.9346,
      "step": 178193
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.5080267190933228,
      "learning_rate": 7.285957808793524e-05,
      "loss": 1.8634,
      "step": 178194
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.2477607727050781,
      "learning_rate": 7.285690591354571e-05,
      "loss": 1.6423,
      "step": 178195
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.3596874475479126,
      "learning_rate": 7.28542337813863e-05,
      "loss": 1.9119,
      "step": 178196
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.2601714134216309,
      "learning_rate": 7.285156169145765e-05,
      "loss": 1.9485,
      "step": 178197
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.2568707466125488,
      "learning_rate": 7.284888964376018e-05,
      "loss": 1.9212,
      "step": 178198
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.4630439281463623,
      "learning_rate": 7.284621763829439e-05,
      "loss": 1.9794,
      "step": 178199
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.316244125366211,
      "learning_rate": 7.284354567506069e-05,
      "loss": 2.0242,
      "step": 178200
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.4851046800613403,
      "learning_rate": 7.284087375405973e-05,
      "loss": 2.0227,
      "step": 178201
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.620205283164978,
      "learning_rate": 7.283820187529187e-05,
      "loss": 1.8378,
      "step": 178202
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.2586003541946411,
      "learning_rate": 7.283553003875779e-05,
      "loss": 2.1416,
      "step": 178203
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.3252424001693726,
      "learning_rate": 7.283285824445781e-05,
      "loss": 1.9316,
      "step": 178204
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.3052068948745728,
      "learning_rate": 7.283018649239253e-05,
      "loss": 2.0509,
      "step": 178205
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.3690022230148315,
      "learning_rate": 7.282751478256232e-05,
      "loss": 2.0491,
      "step": 178206
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.4955888986587524,
      "learning_rate": 7.282484311496786e-05,
      "loss": 2.2226,
      "step": 178207
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.2602665424346924,
      "learning_rate": 7.282217148960942e-05,
      "loss": 2.149,
      "step": 178208
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.428587555885315,
      "learning_rate": 7.281949990648777e-05,
      "loss": 1.9588,
      "step": 178209
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.2575775384902954,
      "learning_rate": 7.281682836560322e-05,
      "loss": 1.9465,
      "step": 178210
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.390580177307129,
      "learning_rate": 7.281415686695635e-05,
      "loss": 1.9423,
      "step": 178211
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.453097939491272,
      "learning_rate": 7.28114854105475e-05,
      "loss": 2.1195,
      "step": 178212
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.4078363180160522,
      "learning_rate": 7.280881399637735e-05,
      "loss": 2.04,
      "step": 178213
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.3776429891586304,
      "learning_rate": 7.280614262444629e-05,
      "loss": 2.137,
      "step": 178214
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.3458964824676514,
      "learning_rate": 7.28034712947549e-05,
      "loss": 1.9178,
      "step": 178215
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.3635320663452148,
      "learning_rate": 7.280080000730367e-05,
      "loss": 2.0065,
      "step": 178216
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.281722903251648,
      "learning_rate": 7.279812876209298e-05,
      "loss": 1.8455,
      "step": 178217
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.4950664043426514,
      "learning_rate": 7.279545755912345e-05,
      "loss": 2.1731,
      "step": 178218
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.2576240301132202,
      "learning_rate": 7.279278639839556e-05,
      "loss": 1.8777,
      "step": 178219
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.328262448310852,
      "learning_rate": 7.27901152799097e-05,
      "loss": 2.1293,
      "step": 178220
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.3656822443008423,
      "learning_rate": 7.27874442036665e-05,
      "loss": 2.085,
      "step": 178221
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.4207582473754883,
      "learning_rate": 7.278477316966642e-05,
      "loss": 1.9868,
      "step": 178222
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.4322423934936523,
      "learning_rate": 7.278210217790989e-05,
      "loss": 1.9333,
      "step": 178223
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.2452749013900757,
      "learning_rate": 7.277943122839749e-05,
      "loss": 1.7871,
      "step": 178224
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.2643882036209106,
      "learning_rate": 7.27767603211297e-05,
      "loss": 1.8378,
      "step": 178225
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.259987473487854,
      "learning_rate": 7.277408945610691e-05,
      "loss": 2.0461,
      "step": 178226
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.3130253553390503,
      "learning_rate": 7.277141863332981e-05,
      "loss": 2.0037,
      "step": 178227
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.2692610025405884,
      "learning_rate": 7.276874785279871e-05,
      "loss": 2.0463,
      "step": 178228
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.2486603260040283,
      "learning_rate": 7.276607711451425e-05,
      "loss": 2.0023,
      "step": 178229
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.3407281637191772,
      "learning_rate": 7.276340641847687e-05,
      "loss": 2.148,
      "step": 178230
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.439976453781128,
      "learning_rate": 7.276073576468698e-05,
      "loss": 2.0016,
      "step": 178231
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.3296321630477905,
      "learning_rate": 7.275806515314527e-05,
      "loss": 1.99,
      "step": 178232
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.3962857723236084,
      "learning_rate": 7.27553945838521e-05,
      "loss": 1.8453,
      "step": 178233
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.3100595474243164,
      "learning_rate": 7.27527240568079e-05,
      "loss": 2.0065,
      "step": 178234
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.2000118494033813,
      "learning_rate": 7.275005357201337e-05,
      "loss": 2.2245,
      "step": 178235
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.6006263494491577,
      "learning_rate": 7.274738312946888e-05,
      "loss": 2.0455,
      "step": 178236
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.3784010410308838,
      "learning_rate": 7.274471272917485e-05,
      "loss": 1.8384,
      "step": 178237
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.2619956731796265,
      "learning_rate": 7.274204237113193e-05,
      "loss": 1.921,
      "step": 178238
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.4595743417739868,
      "learning_rate": 7.273937205534057e-05,
      "loss": 1.8991,
      "step": 178239
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.2790412902832031,
      "learning_rate": 7.273670178180119e-05,
      "loss": 2.219,
      "step": 178240
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.2827810049057007,
      "learning_rate": 7.273403155051444e-05,
      "loss": 2.0814,
      "step": 178241
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.248442530632019,
      "learning_rate": 7.273136136148069e-05,
      "loss": 1.9542,
      "step": 178242
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.2532312870025635,
      "learning_rate": 7.272869121470037e-05,
      "loss": 2.095,
      "step": 178243
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.2630345821380615,
      "learning_rate": 7.272602111017419e-05,
      "loss": 2.1578,
      "step": 178244
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.238646388053894,
      "learning_rate": 7.272335104790245e-05,
      "loss": 2.0686,
      "step": 178245
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.2413907051086426,
      "learning_rate": 7.272068102788583e-05,
      "loss": 1.9594,
      "step": 178246
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.4087927341461182,
      "learning_rate": 7.271801105012469e-05,
      "loss": 1.8946,
      "step": 178247
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.267799735069275,
      "learning_rate": 7.27153411146196e-05,
      "loss": 2.0147,
      "step": 178248
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.3687758445739746,
      "learning_rate": 7.271267122137087e-05,
      "loss": 1.866,
      "step": 178249
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.4922212362289429,
      "learning_rate": 7.271000137037929e-05,
      "loss": 2.1001,
      "step": 178250
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.2759594917297363,
      "learning_rate": 7.27073315616451e-05,
      "loss": 2.1699,
      "step": 178251
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.350407600402832,
      "learning_rate": 7.2704661795169e-05,
      "loss": 1.8132,
      "step": 178252
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.3252031803131104,
      "learning_rate": 7.270199207095141e-05,
      "loss": 1.9274,
      "step": 178253
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.246473789215088,
      "learning_rate": 7.269932238899277e-05,
      "loss": 1.9821,
      "step": 178254
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.3601644039154053,
      "learning_rate": 7.269665274929356e-05,
      "loss": 2.0811,
      "step": 178255
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.2998863458633423,
      "learning_rate": 7.269398315185443e-05,
      "loss": 1.9144,
      "step": 178256
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.2383302450180054,
      "learning_rate": 7.269131359667569e-05,
      "loss": 1.7639,
      "step": 178257
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.4056134223937988,
      "learning_rate": 7.268864408375796e-05,
      "loss": 1.9672,
      "step": 178258
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.4647071361541748,
      "learning_rate": 7.268597461310177e-05,
      "loss": 2.1211,
      "step": 178259
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.3016859292984009,
      "learning_rate": 7.268330518470751e-05,
      "loss": 2.248,
      "step": 178260
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.2566574811935425,
      "learning_rate": 7.268063579857564e-05,
      "loss": 2.0023,
      "step": 178261
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.4254494905471802,
      "learning_rate": 7.26779664547068e-05,
      "loss": 1.9672,
      "step": 178262
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.3172032833099365,
      "learning_rate": 7.267529715310134e-05,
      "loss": 1.8625,
      "step": 178263
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.5504494905471802,
      "learning_rate": 7.267262789375992e-05,
      "loss": 2.0369,
      "step": 178264
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.820091962814331,
      "learning_rate": 7.266995867668295e-05,
      "loss": 2.0996,
      "step": 178265
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.4026191234588623,
      "learning_rate": 7.266728950187094e-05,
      "loss": 2.1065,
      "step": 178266
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.2687610387802124,
      "learning_rate": 7.266462036932429e-05,
      "loss": 1.895,
      "step": 178267
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.3507757186889648,
      "learning_rate": 7.266195127904363e-05,
      "loss": 1.8414,
      "step": 178268
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.4422849416732788,
      "learning_rate": 7.265928223102936e-05,
      "loss": 2.0868,
      "step": 178269
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.2481770515441895,
      "learning_rate": 7.265661322528208e-05,
      "loss": 1.8816,
      "step": 178270
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.304665207862854,
      "learning_rate": 7.265394426180222e-05,
      "loss": 2.0316,
      "step": 178271
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.5400090217590332,
      "learning_rate": 7.265127534059029e-05,
      "loss": 1.9285,
      "step": 178272
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.362706184387207,
      "learning_rate": 7.26486064616467e-05,
      "loss": 1.9975,
      "step": 178273
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.2290529012680054,
      "learning_rate": 7.264593762497212e-05,
      "loss": 1.9548,
      "step": 178274
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.3245551586151123,
      "learning_rate": 7.264326883056685e-05,
      "loss": 2.0313,
      "step": 178275
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.3032277822494507,
      "learning_rate": 7.26406000784316e-05,
      "loss": 2.0181,
      "step": 178276
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.4183669090270996,
      "learning_rate": 7.26379313685667e-05,
      "loss": 2.1397,
      "step": 178277
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.336034893989563,
      "learning_rate": 7.263526270097276e-05,
      "loss": 1.845,
      "step": 178278
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.3726075887680054,
      "learning_rate": 7.26325940756501e-05,
      "loss": 1.983,
      "step": 178279
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.202782392501831,
      "learning_rate": 7.262992549259944e-05,
      "loss": 2.276,
      "step": 178280
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.4038385152816772,
      "learning_rate": 7.262725695182106e-05,
      "loss": 1.7696,
      "step": 178281
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.2597984075546265,
      "learning_rate": 7.262458845331566e-05,
      "loss": 1.7225,
      "step": 178282
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.503416895866394,
      "learning_rate": 7.262191999708361e-05,
      "loss": 1.9572,
      "step": 178283
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.2689697742462158,
      "learning_rate": 7.261925158312548e-05,
      "loss": 1.8589,
      "step": 178284
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.278198480606079,
      "learning_rate": 7.261658321144163e-05,
      "loss": 1.8623,
      "step": 178285
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.4995273351669312,
      "learning_rate": 7.261391488203271e-05,
      "loss": 1.902,
      "step": 178286
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.337456464767456,
      "learning_rate": 7.26112465948991e-05,
      "loss": 2.0757,
      "step": 178287
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.4897098541259766,
      "learning_rate": 7.260857835004144e-05,
      "loss": 1.9674,
      "step": 178288
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.2985973358154297,
      "learning_rate": 7.260591014746012e-05,
      "loss": 1.9892,
      "step": 178289
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.4929689168930054,
      "learning_rate": 7.260324198715566e-05,
      "loss": 2.0003,
      "step": 178290
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.2825299501419067,
      "learning_rate": 7.260057386912844e-05,
      "loss": 2.0216,
      "step": 178291
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.2928345203399658,
      "learning_rate": 7.259790579337917e-05,
      "loss": 2.0833,
      "step": 178292
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.3109216690063477,
      "learning_rate": 7.259523775990815e-05,
      "loss": 2.0297,
      "step": 178293
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.2913904190063477,
      "learning_rate": 7.259256976871608e-05,
      "loss": 1.9077,
      "step": 178294
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.3401066064834595,
      "learning_rate": 7.258990181980325e-05,
      "loss": 2.0326,
      "step": 178295
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.2624008655548096,
      "learning_rate": 7.258723391317041e-05,
      "loss": 2.2202,
      "step": 178296
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.4355343580245972,
      "learning_rate": 7.25845660488177e-05,
      "loss": 1.9953,
      "step": 178297
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.4543700218200684,
      "learning_rate": 7.258189822674595e-05,
      "loss": 2.109,
      "step": 178298
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.2094683647155762,
      "learning_rate": 7.25792304469554e-05,
      "loss": 1.9475,
      "step": 178299
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.364922046661377,
      "learning_rate": 7.257656270944674e-05,
      "loss": 1.8481,
      "step": 178300
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.3264538049697876,
      "learning_rate": 7.257389501422035e-05,
      "loss": 1.9496,
      "step": 178301
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.259779691696167,
      "learning_rate": 7.257122736127691e-05,
      "loss": 1.9999,
      "step": 178302
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.3889753818511963,
      "learning_rate": 7.256855975061657e-05,
      "loss": 1.9128,
      "step": 178303
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.441507339477539,
      "learning_rate": 7.256589218224015e-05,
      "loss": 1.9706,
      "step": 178304
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.2905383110046387,
      "learning_rate": 7.256322465614794e-05,
      "loss": 2.2493,
      "step": 178305
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.2590644359588623,
      "learning_rate": 7.25605571723406e-05,
      "loss": 1.9415,
      "step": 178306
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.2370073795318604,
      "learning_rate": 7.255788973081843e-05,
      "loss": 2.0212,
      "step": 178307
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.2491354942321777,
      "learning_rate": 7.255522233158218e-05,
      "loss": 1.92,
      "step": 178308
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.3006378412246704,
      "learning_rate": 7.255255497463216e-05,
      "loss": 1.9266,
      "step": 178309
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.3477612733840942,
      "learning_rate": 7.254988765996895e-05,
      "loss": 2.1984,
      "step": 178310
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.325171709060669,
      "learning_rate": 7.254722038759288e-05,
      "loss": 1.9703,
      "step": 178311
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.404147744178772,
      "learning_rate": 7.25445531575047e-05,
      "loss": 1.891,
      "step": 178312
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.2742217779159546,
      "learning_rate": 7.254188596970467e-05,
      "loss": 1.7931,
      "step": 178313
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.2223751544952393,
      "learning_rate": 7.253921882419352e-05,
      "loss": 2.0503,
      "step": 178314
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.3482412099838257,
      "learning_rate": 7.253655172097159e-05,
      "loss": 1.7373,
      "step": 178315
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.257519245147705,
      "learning_rate": 7.253388466003935e-05,
      "loss": 2.068,
      "step": 178316
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.4154927730560303,
      "learning_rate": 7.25312176413974e-05,
      "loss": 2.1582,
      "step": 178317
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.3795104026794434,
      "learning_rate": 7.252855066504622e-05,
      "loss": 2.0127,
      "step": 178318
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.241740345954895,
      "learning_rate": 7.252588373098619e-05,
      "loss": 1.823,
      "step": 178319
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.4237279891967773,
      "learning_rate": 7.252321683921796e-05,
      "loss": 2.0323,
      "step": 178320
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.2304826974868774,
      "learning_rate": 7.252054998974199e-05,
      "loss": 1.8151,
      "step": 178321
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.4309046268463135,
      "learning_rate": 7.25178831825586e-05,
      "loss": 1.8046,
      "step": 178322
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.357983112335205,
      "learning_rate": 7.251521641766856e-05,
      "loss": 2.2766,
      "step": 178323
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.3749659061431885,
      "learning_rate": 7.251254969507224e-05,
      "loss": 2.0611,
      "step": 178324
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.3904261589050293,
      "learning_rate": 7.250988301477003e-05,
      "loss": 1.9746,
      "step": 178325
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.2976034879684448,
      "learning_rate": 7.250721637676262e-05,
      "loss": 2.0415,
      "step": 178326
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.3962819576263428,
      "learning_rate": 7.250454978105042e-05,
      "loss": 1.9004,
      "step": 178327
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.353981852531433,
      "learning_rate": 7.25018832276338e-05,
      "loss": 1.9209,
      "step": 178328
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.4211457967758179,
      "learning_rate": 7.24992167165135e-05,
      "loss": 1.9881,
      "step": 178329
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.3653767108917236,
      "learning_rate": 7.249655024768976e-05,
      "loss": 2.0117,
      "step": 178330
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.397461175918579,
      "learning_rate": 7.249388382116331e-05,
      "loss": 2.2476,
      "step": 178331
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.3057029247283936,
      "learning_rate": 7.249121743693457e-05,
      "loss": 1.927,
      "step": 178332
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.505210041999817,
      "learning_rate": 7.248855109500398e-05,
      "loss": 2.0108,
      "step": 178333
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.3501628637313843,
      "learning_rate": 7.248588479537197e-05,
      "loss": 1.9471,
      "step": 178334
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.4627498388290405,
      "learning_rate": 7.248321853803926e-05,
      "loss": 2.2556,
      "step": 178335
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.3670573234558105,
      "learning_rate": 7.248055232300607e-05,
      "loss": 2.0025,
      "step": 178336
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.6041741371154785,
      "learning_rate": 7.247788615027315e-05,
      "loss": 1.9133,
      "step": 178337
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.366129755973816,
      "learning_rate": 7.247522001984089e-05,
      "loss": 2.1268,
      "step": 178338
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.360122561454773,
      "learning_rate": 7.24725539317098e-05,
      "loss": 2.086,
      "step": 178339
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.3320643901824951,
      "learning_rate": 7.246988788588026e-05,
      "loss": 2.1367,
      "step": 178340
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.3041841983795166,
      "learning_rate": 7.246722188235291e-05,
      "loss": 2.0281,
      "step": 178341
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.287895679473877,
      "learning_rate": 7.246455592112816e-05,
      "loss": 1.8784,
      "step": 178342
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.2860432863235474,
      "learning_rate": 7.246189000220661e-05,
      "loss": 2.1895,
      "step": 178343
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.4122370481491089,
      "learning_rate": 7.245922412558869e-05,
      "loss": 2.1799,
      "step": 178344
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.251563549041748,
      "learning_rate": 7.245655829127489e-05,
      "loss": 2.0471,
      "step": 178345
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.604533314704895,
      "learning_rate": 7.245389249926562e-05,
      "loss": 1.9647,
      "step": 178346
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.5012072324752808,
      "learning_rate": 7.245122674956158e-05,
      "loss": 2.0118,
      "step": 178347
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.311541199684143,
      "learning_rate": 7.244856104216305e-05,
      "loss": 2.1088,
      "step": 178348
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.2197452783584595,
      "learning_rate": 7.244589537707069e-05,
      "loss": 1.9252,
      "step": 178349
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.2222158908843994,
      "learning_rate": 7.2443229754285e-05,
      "loss": 1.8453,
      "step": 178350
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.3437023162841797,
      "learning_rate": 7.244056417380635e-05,
      "loss": 2.0684,
      "step": 178351
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.4325655698776245,
      "learning_rate": 7.243789863563522e-05,
      "loss": 2.0663,
      "step": 178352
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.2561237812042236,
      "learning_rate": 7.243523313977227e-05,
      "loss": 1.877,
      "step": 178353
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.2948449850082397,
      "learning_rate": 7.243256768621782e-05,
      "loss": 1.8868,
      "step": 178354
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.3133612871170044,
      "learning_rate": 7.242990227497254e-05,
      "loss": 1.9973,
      "step": 178355
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.4668751955032349,
      "learning_rate": 7.242723690603685e-05,
      "loss": 2.2505,
      "step": 178356
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.4060293436050415,
      "learning_rate": 7.242457157941121e-05,
      "loss": 2.2178,
      "step": 178357
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.3343830108642578,
      "learning_rate": 7.242190629509605e-05,
      "loss": 2.0233,
      "step": 178358
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.3991180658340454,
      "learning_rate": 7.241924105309206e-05,
      "loss": 1.9463,
      "step": 178359
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.312235951423645,
      "learning_rate": 7.241657585339954e-05,
      "loss": 1.9891,
      "step": 178360
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.237144112586975,
      "learning_rate": 7.241391069601917e-05,
      "loss": 1.9257,
      "step": 178361
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.583006501197815,
      "learning_rate": 7.241124558095123e-05,
      "loss": 1.9991,
      "step": 178362
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.431154727935791,
      "learning_rate": 7.240858050819651e-05,
      "loss": 1.8789,
      "step": 178363
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.4126895666122437,
      "learning_rate": 7.240591547775518e-05,
      "loss": 1.9969,
      "step": 178364
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.4658552408218384,
      "learning_rate": 7.240325048962797e-05,
      "loss": 2.1997,
      "step": 178365
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.2216317653656006,
      "learning_rate": 7.240058554381521e-05,
      "loss": 2.1286,
      "step": 178366
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.3553979396820068,
      "learning_rate": 7.239792064031754e-05,
      "loss": 1.9288,
      "step": 178367
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.2993335723876953,
      "learning_rate": 7.239525577913531e-05,
      "loss": 1.9867,
      "step": 178368
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.2459266185760498,
      "learning_rate": 7.239259096026932e-05,
      "loss": 1.9542,
      "step": 178369
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.319834589958191,
      "learning_rate": 7.238992618371963e-05,
      "loss": 2.0373,
      "step": 178370
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.4233980178833008,
      "learning_rate": 7.238726144948702e-05,
      "loss": 1.8995,
      "step": 178371
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.3485667705535889,
      "learning_rate": 7.238459675757187e-05,
      "loss": 2.0109,
      "step": 178372
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.2317652702331543,
      "learning_rate": 7.238193210797476e-05,
      "loss": 1.9,
      "step": 178373
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.332403302192688,
      "learning_rate": 7.23792675006961e-05,
      "loss": 2.0,
      "step": 178374
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.3852497339248657,
      "learning_rate": 7.237660293573659e-05,
      "loss": 2.196,
      "step": 178375
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.318037509918213,
      "learning_rate": 7.23739384130964e-05,
      "loss": 1.9688,
      "step": 178376
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.2209134101867676,
      "learning_rate": 7.237127393277626e-05,
      "loss": 1.7767,
      "step": 178377
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.2179069519042969,
      "learning_rate": 7.236860949477652e-05,
      "loss": 1.9473,
      "step": 178378
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.488644003868103,
      "learning_rate": 7.236594509909783e-05,
      "loss": 1.9485,
      "step": 178379
    },
    {
      "epoch": 2.32,
      "grad_norm": 2.5927846431732178,
      "learning_rate": 7.236328074574052e-05,
      "loss": 2.1141,
      "step": 178380
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.3975332975387573,
      "learning_rate": 7.236061643470537e-05,
      "loss": 2.129,
      "step": 178381
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.283448338508606,
      "learning_rate": 7.235795216599248e-05,
      "loss": 1.8744,
      "step": 178382
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.7583924531936646,
      "learning_rate": 7.235528793960264e-05,
      "loss": 2.0845,
      "step": 178383
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.3353897333145142,
      "learning_rate": 7.235262375553618e-05,
      "loss": 1.9575,
      "step": 178384
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.2545405626296997,
      "learning_rate": 7.234995961379374e-05,
      "loss": 1.8737,
      "step": 178385
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.358559012413025,
      "learning_rate": 7.234729551437564e-05,
      "loss": 1.9452,
      "step": 178386
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.4676764011383057,
      "learning_rate": 7.234463145728265e-05,
      "loss": 2.0188,
      "step": 178387
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.540794014930725,
      "learning_rate": 7.234196744251492e-05,
      "loss": 2.0762,
      "step": 178388
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.2800712585449219,
      "learning_rate": 7.233930347007322e-05,
      "loss": 1.9026,
      "step": 178389
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.3881175518035889,
      "learning_rate": 7.233663953995782e-05,
      "loss": 1.915,
      "step": 178390
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.3504317998886108,
      "learning_rate": 7.233397565216946e-05,
      "loss": 2.1496,
      "step": 178391
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.309094786643982,
      "learning_rate": 7.233131180670842e-05,
      "loss": 2.1208,
      "step": 178392
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.4243439435958862,
      "learning_rate": 7.232864800357533e-05,
      "loss": 2.056,
      "step": 178393
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.2433561086654663,
      "learning_rate": 7.232598424277067e-05,
      "loss": 2.1391,
      "step": 178394
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.4196624755859375,
      "learning_rate": 7.232332052429491e-05,
      "loss": 1.9737,
      "step": 178395
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.1947228908538818,
      "learning_rate": 7.232065684814846e-05,
      "loss": 2.06,
      "step": 178396
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.2703120708465576,
      "learning_rate": 7.231799321433198e-05,
      "loss": 2.0611,
      "step": 178397
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.3450698852539062,
      "learning_rate": 7.231532962284581e-05,
      "loss": 1.9085,
      "step": 178398
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.343124508857727,
      "learning_rate": 7.231266607369061e-05,
      "loss": 2.2486,
      "step": 178399
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.2793761491775513,
      "learning_rate": 7.231000256686674e-05,
      "loss": 2.0628,
      "step": 178400
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.2738615274429321,
      "learning_rate": 7.230733910237469e-05,
      "loss": 1.9038,
      "step": 178401
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.3182429075241089,
      "learning_rate": 7.230467568021506e-05,
      "loss": 2.1682,
      "step": 178402
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.489161729812622,
      "learning_rate": 7.230201230038831e-05,
      "loss": 1.8895,
      "step": 178403
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.1854853630065918,
      "learning_rate": 7.229934896289483e-05,
      "loss": 1.9759,
      "step": 178404
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.216115951538086,
      "learning_rate": 7.229668566773529e-05,
      "loss": 1.9598,
      "step": 178405
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.2812540531158447,
      "learning_rate": 7.229402241491008e-05,
      "loss": 2.0646,
      "step": 178406
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.253339171409607,
      "learning_rate": 7.22913592044196e-05,
      "loss": 1.9381,
      "step": 178407
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.278763771057129,
      "learning_rate": 7.228869603626456e-05,
      "loss": 2.1856,
      "step": 178408
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.6054331064224243,
      "learning_rate": 7.228603291044537e-05,
      "loss": 1.7591,
      "step": 178409
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.2755963802337646,
      "learning_rate": 7.22833698269624e-05,
      "loss": 1.9882,
      "step": 178410
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.4055274724960327,
      "learning_rate": 7.228070678581636e-05,
      "loss": 2.3264,
      "step": 178411
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.3664381504058838,
      "learning_rate": 7.22780437870076e-05,
      "loss": 2.0995,
      "step": 178412
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.2284177541732788,
      "learning_rate": 7.22753808305366e-05,
      "loss": 2.1314,
      "step": 178413
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.3822524547576904,
      "learning_rate": 7.227271791640397e-05,
      "loss": 1.9725,
      "step": 178414
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.2806966304779053,
      "learning_rate": 7.227005504461006e-05,
      "loss": 1.9079,
      "step": 178415
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.3569977283477783,
      "learning_rate": 7.226739221515553e-05,
      "loss": 1.9638,
      "step": 178416
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.2973014116287231,
      "learning_rate": 7.22647294280408e-05,
      "loss": 2.0347,
      "step": 178417
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.3023548126220703,
      "learning_rate": 7.226206668326632e-05,
      "loss": 2.1222,
      "step": 178418
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.3214671611785889,
      "learning_rate": 7.225940398083258e-05,
      "loss": 2.0846,
      "step": 178419
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.3893312215805054,
      "learning_rate": 7.225674132074017e-05,
      "loss": 2.033,
      "step": 178420
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.246940016746521,
      "learning_rate": 7.225407870298945e-05,
      "loss": 1.9566,
      "step": 178421
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.2908570766448975,
      "learning_rate": 7.225141612758109e-05,
      "loss": 1.7732,
      "step": 178422
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.3146001100540161,
      "learning_rate": 7.22487535945155e-05,
      "loss": 1.9486,
      "step": 178423
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.2412863969802856,
      "learning_rate": 7.224609110379314e-05,
      "loss": 2.1949,
      "step": 178424
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.4297362565994263,
      "learning_rate": 7.224342865541449e-05,
      "loss": 1.8553,
      "step": 178425
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.6079972982406616,
      "learning_rate": 7.224076624938011e-05,
      "loss": 2.0642,
      "step": 178426
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.2932312488555908,
      "learning_rate": 7.223810388569044e-05,
      "loss": 1.9637,
      "step": 178427
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.202097773551941,
      "learning_rate": 7.223544156434606e-05,
      "loss": 2.0244,
      "step": 178428
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.3395562171936035,
      "learning_rate": 7.223277928534733e-05,
      "loss": 1.9521,
      "step": 178429
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.5723509788513184,
      "learning_rate": 7.2230117048695e-05,
      "loss": 2.2658,
      "step": 178430
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.3072116374969482,
      "learning_rate": 7.22274548543892e-05,
      "loss": 1.9228,
      "step": 178431
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.2589768171310425,
      "learning_rate": 7.222479270243072e-05,
      "loss": 1.7158,
      "step": 178432
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.2267136573791504,
      "learning_rate": 7.222213059281987e-05,
      "loss": 1.8942,
      "step": 178433
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.5110520124435425,
      "learning_rate": 7.221946852555731e-05,
      "loss": 2.0019,
      "step": 178434
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.3407312631607056,
      "learning_rate": 7.221680650064337e-05,
      "loss": 2.072,
      "step": 178435
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.674676537513733,
      "learning_rate": 7.22141445180788e-05,
      "loss": 1.9328,
      "step": 178436
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.2516579627990723,
      "learning_rate": 7.221148257786371e-05,
      "loss": 1.9126,
      "step": 178437
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.3798340559005737,
      "learning_rate": 7.220882067999894e-05,
      "loss": 2.03,
      "step": 178438
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.401854395866394,
      "learning_rate": 7.220615882448474e-05,
      "loss": 2.0023,
      "step": 178439
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.2783564329147339,
      "learning_rate": 7.220349701132179e-05,
      "loss": 2.0254,
      "step": 178440
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.4541910886764526,
      "learning_rate": 7.220083524051046e-05,
      "loss": 1.8876,
      "step": 178441
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.3182297945022583,
      "learning_rate": 7.219817351205142e-05,
      "loss": 1.966,
      "step": 178442
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.3414825201034546,
      "learning_rate": 7.219551182594489e-05,
      "loss": 2.0087,
      "step": 178443
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.5376955270767212,
      "learning_rate": 7.219285018219163e-05,
      "loss": 1.8827,
      "step": 178444
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.5104488134384155,
      "learning_rate": 7.219018858079189e-05,
      "loss": 2.2856,
      "step": 178445
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.345246434211731,
      "learning_rate": 7.21875270217464e-05,
      "loss": 2.1079,
      "step": 178446
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.324770212173462,
      "learning_rate": 7.218486550505545e-05,
      "loss": 2.2898,
      "step": 178447
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.4255808591842651,
      "learning_rate": 7.218220403071983e-05,
      "loss": 2.0917,
      "step": 178448
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.445896863937378,
      "learning_rate": 7.217954259873964e-05,
      "loss": 2.0271,
      "step": 178449
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.3126842975616455,
      "learning_rate": 7.21768812091157e-05,
      "loss": 2.1565,
      "step": 178450
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.3021436929702759,
      "learning_rate": 7.217421986184823e-05,
      "loss": 2.1587,
      "step": 178451
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.324807047843933,
      "learning_rate": 7.217155855693802e-05,
      "loss": 2.1573,
      "step": 178452
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.2315105199813843,
      "learning_rate": 7.216889729438528e-05,
      "loss": 1.7806,
      "step": 178453
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.3304508924484253,
      "learning_rate": 7.216623607419084e-05,
      "loss": 1.9698,
      "step": 178454
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.3042865991592407,
      "learning_rate": 7.216357489635481e-05,
      "loss": 2.04,
      "step": 178455
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.354621410369873,
      "learning_rate": 7.2160913760878e-05,
      "loss": 2.033,
      "step": 178456
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.3671905994415283,
      "learning_rate": 7.215825266776064e-05,
      "loss": 2.0751,
      "step": 178457
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.3987410068511963,
      "learning_rate": 7.215559161700349e-05,
      "loss": 2.1493,
      "step": 178458
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.304187297821045,
      "learning_rate": 7.21529306086068e-05,
      "loss": 2.0964,
      "step": 178459
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.3720616102218628,
      "learning_rate": 7.215026964257138e-05,
      "loss": 2.1701,
      "step": 178460
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.3645449876785278,
      "learning_rate": 7.214760871889735e-05,
      "loss": 2.1142,
      "step": 178461
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.3112494945526123,
      "learning_rate": 7.214494783758542e-05,
      "loss": 1.6802,
      "step": 178462
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.5611555576324463,
      "learning_rate": 7.214228699863601e-05,
      "loss": 1.9454,
      "step": 178463
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.4447098970413208,
      "learning_rate": 7.213962620204972e-05,
      "loss": 2.1236,
      "step": 178464
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.4217627048492432,
      "learning_rate": 7.21369654478269e-05,
      "loss": 1.9779,
      "step": 178465
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.4420874118804932,
      "learning_rate": 7.213430473596827e-05,
      "loss": 2.0325,
      "step": 178466
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.2592628002166748,
      "learning_rate": 7.2131644066474e-05,
      "loss": 2.024,
      "step": 178467
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.3556479215621948,
      "learning_rate": 7.212898343934485e-05,
      "loss": 2.4112,
      "step": 178468
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.3117743730545044,
      "learning_rate": 7.212632285458116e-05,
      "loss": 2.0834,
      "step": 178469
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.2885991334915161,
      "learning_rate": 7.212366231218356e-05,
      "loss": 2.0658,
      "step": 178470
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.4176530838012695,
      "learning_rate": 7.212100181215238e-05,
      "loss": 2.0909,
      "step": 178471
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.3311221599578857,
      "learning_rate": 7.211834135448838e-05,
      "loss": 2.0361,
      "step": 178472
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.3519618511199951,
      "learning_rate": 7.211568093919171e-05,
      "loss": 1.8283,
      "step": 178473
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.3648244142532349,
      "learning_rate": 7.211302056626316e-05,
      "loss": 1.9893,
      "step": 178474
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.4493216276168823,
      "learning_rate": 7.211036023570297e-05,
      "loss": 2.0606,
      "step": 178475
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.2347261905670166,
      "learning_rate": 7.210769994751185e-05,
      "loss": 2.2362,
      "step": 178476
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.1826952695846558,
      "learning_rate": 7.210503970169015e-05,
      "loss": 2.1217,
      "step": 178477
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.2681653499603271,
      "learning_rate": 7.21023794982385e-05,
      "loss": 1.9869,
      "step": 178478
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.5370442867279053,
      "learning_rate": 7.209971933715732e-05,
      "loss": 2.0706,
      "step": 178479
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.4427598714828491,
      "learning_rate": 7.20970592184471e-05,
      "loss": 2.1913,
      "step": 178480
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.3359668254852295,
      "learning_rate": 7.209439914210824e-05,
      "loss": 1.9488,
      "step": 178481
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.3057935237884521,
      "learning_rate": 7.209173910814144e-05,
      "loss": 1.9977,
      "step": 178482
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.2872133255004883,
      "learning_rate": 7.2089079116547e-05,
      "loss": 2.0106,
      "step": 178483
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.382692575454712,
      "learning_rate": 7.20864191673256e-05,
      "loss": 2.0274,
      "step": 178484
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.317342758178711,
      "learning_rate": 7.208375926047763e-05,
      "loss": 1.8682,
      "step": 178485
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.44180428981781,
      "learning_rate": 7.208109939600354e-05,
      "loss": 1.9524,
      "step": 178486
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.3397040367126465,
      "learning_rate": 7.207843957390384e-05,
      "loss": 1.9013,
      "step": 178487
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.3654940128326416,
      "learning_rate": 7.207577979417914e-05,
      "loss": 2.1877,
      "step": 178488
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.2611582279205322,
      "learning_rate": 7.207312005682978e-05,
      "loss": 2.0105,
      "step": 178489
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.3927730321884155,
      "learning_rate": 7.207046036185642e-05,
      "loss": 1.9295,
      "step": 178490
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.3115265369415283,
      "learning_rate": 7.206780070925942e-05,
      "loss": 2.0629,
      "step": 178491
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.4759639501571655,
      "learning_rate": 7.206514109903927e-05,
      "loss": 2.168,
      "step": 178492
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.4333151578903198,
      "learning_rate": 7.206248153119662e-05,
      "loss": 1.8276,
      "step": 178493
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.3511213064193726,
      "learning_rate": 7.205982200573179e-05,
      "loss": 1.9751,
      "step": 178494
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.299474835395813,
      "learning_rate": 7.20571625226453e-05,
      "loss": 1.963,
      "step": 178495
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.3140151500701904,
      "learning_rate": 7.205450308193779e-05,
      "loss": 1.9815,
      "step": 178496
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.327513337135315,
      "learning_rate": 7.20518436836096e-05,
      "loss": 2.1287,
      "step": 178497
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.2687381505966187,
      "learning_rate": 7.204918432766121e-05,
      "loss": 1.8336,
      "step": 178498
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.3556910753250122,
      "learning_rate": 7.20465250140933e-05,
      "loss": 2.095,
      "step": 178499
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.358625888824463,
      "learning_rate": 7.20438657429061e-05,
      "loss": 1.9534,
      "step": 178500
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.2951546907424927,
      "learning_rate": 7.204120651410039e-05,
      "loss": 2.0862,
      "step": 178501
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.4474183320999146,
      "learning_rate": 7.203854732767648e-05,
      "loss": 1.9302,
      "step": 178502
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.2243480682373047,
      "learning_rate": 7.203588818363493e-05,
      "loss": 2.1164,
      "step": 178503
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.4086414575576782,
      "learning_rate": 7.203322908197609e-05,
      "loss": 2.0645,
      "step": 178504
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.5547152757644653,
      "learning_rate": 7.20305700227007e-05,
      "loss": 2.4249,
      "step": 178505
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.4319700002670288,
      "learning_rate": 7.202791100580902e-05,
      "loss": 2.126,
      "step": 178506
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.3074452877044678,
      "learning_rate": 7.202525203130178e-05,
      "loss": 1.9208,
      "step": 178507
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.3541761636734009,
      "learning_rate": 7.20225930991793e-05,
      "loss": 1.8306,
      "step": 178508
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.3622713088989258,
      "learning_rate": 7.201993420944215e-05,
      "loss": 2.0215,
      "step": 178509
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.3565174341201782,
      "learning_rate": 7.201727536209073e-05,
      "loss": 2.2951,
      "step": 178510
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.3441898822784424,
      "learning_rate": 7.201461655712567e-05,
      "loss": 2.2483,
      "step": 178511
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.347305417060852,
      "learning_rate": 7.20119577945473e-05,
      "loss": 2.1036,
      "step": 178512
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.629795789718628,
      "learning_rate": 7.20092990743563e-05,
      "loss": 2.0102,
      "step": 178513
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.2625118494033813,
      "learning_rate": 7.200664039655298e-05,
      "loss": 1.9063,
      "step": 178514
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.1507718563079834,
      "learning_rate": 7.200398176113813e-05,
      "loss": 2.0485,
      "step": 178515
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.4085122346878052,
      "learning_rate": 7.200132316811186e-05,
      "loss": 1.9597,
      "step": 178516
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.3386427164077759,
      "learning_rate": 7.199866461747493e-05,
      "loss": 1.9092,
      "step": 178517
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.258583903312683,
      "learning_rate": 7.199600610922766e-05,
      "loss": 1.9733,
      "step": 178518
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.2670378684997559,
      "learning_rate": 7.199334764337074e-05,
      "loss": 2.0528,
      "step": 178519
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.2359386682510376,
      "learning_rate": 7.199068921990444e-05,
      "loss": 1.736,
      "step": 178520
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.3088148832321167,
      "learning_rate": 7.198803083882957e-05,
      "loss": 2.0989,
      "step": 178521
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.2535147666931152,
      "learning_rate": 7.198537250014626e-05,
      "loss": 1.9961,
      "step": 178522
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.3011125326156616,
      "learning_rate": 7.198271420385527e-05,
      "loss": 1.8751,
      "step": 178523
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.3496352434158325,
      "learning_rate": 7.198005594995689e-05,
      "loss": 1.9648,
      "step": 178524
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.376929759979248,
      "learning_rate": 7.197739773845181e-05,
      "loss": 2.0508,
      "step": 178525
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.2842010259628296,
      "learning_rate": 7.197473956934035e-05,
      "loss": 2.0583,
      "step": 178526
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.2836945056915283,
      "learning_rate": 7.197208144262328e-05,
      "loss": 2.0058,
      "step": 178527
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.270105004310608,
      "learning_rate": 7.19694233583007e-05,
      "loss": 2.1673,
      "step": 178528
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.6758873462677002,
      "learning_rate": 7.196676531637345e-05,
      "loss": 2.2598,
      "step": 178529
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.385922908782959,
      "learning_rate": 7.196410731684174e-05,
      "loss": 1.8534,
      "step": 178530
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.2838401794433594,
      "learning_rate": 7.196144935970632e-05,
      "loss": 1.8216,
      "step": 178531
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.2244988679885864,
      "learning_rate": 7.195879144496748e-05,
      "loss": 1.8646,
      "step": 178532
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.3076539039611816,
      "learning_rate": 7.195613357262597e-05,
      "loss": 1.836,
      "step": 178533
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.2327319383621216,
      "learning_rate": 7.195347574268196e-05,
      "loss": 2.1117,
      "step": 178534
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.3127202987670898,
      "learning_rate": 7.195081795513619e-05,
      "loss": 2.1872,
      "step": 178535
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.5454257726669312,
      "learning_rate": 7.194816020998896e-05,
      "loss": 1.8742,
      "step": 178536
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.3459348678588867,
      "learning_rate": 7.194550250724098e-05,
      "loss": 2.0801,
      "step": 178537
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.4019842147827148,
      "learning_rate": 7.194284484689256e-05,
      "loss": 1.7735,
      "step": 178538
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.3695498704910278,
      "learning_rate": 7.194018722894442e-05,
      "loss": 1.867,
      "step": 178539
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.674275517463684,
      "learning_rate": 7.193752965339672e-05,
      "loss": 2.1272,
      "step": 178540
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.3937709331512451,
      "learning_rate": 7.193487212025024e-05,
      "loss": 2.0138,
      "step": 178541
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.2417142391204834,
      "learning_rate": 7.193221462950528e-05,
      "loss": 2.0546,
      "step": 178542
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.1479600667953491,
      "learning_rate": 7.192955718116253e-05,
      "loss": 1.9719,
      "step": 178543
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.243272066116333,
      "learning_rate": 7.192689977522231e-05,
      "loss": 1.9336,
      "step": 178544
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.3662383556365967,
      "learning_rate": 7.192424241168531e-05,
      "loss": 2.0856,
      "step": 178545
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.542823076248169,
      "learning_rate": 7.192158509055175e-05,
      "loss": 2.1119,
      "step": 178546
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.3121942281723022,
      "learning_rate": 7.191892781182236e-05,
      "loss": 1.9872,
      "step": 178547
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.2627668380737305,
      "learning_rate": 7.191627057549745e-05,
      "loss": 1.7615,
      "step": 178548
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.321433663368225,
      "learning_rate": 7.191361338157772e-05,
      "loss": 1.866,
      "step": 178549
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.3192726373672485,
      "learning_rate": 7.191095623006343e-05,
      "loss": 2.1233,
      "step": 178550
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.3330291509628296,
      "learning_rate": 7.19082991209554e-05,
      "loss": 2.0707,
      "step": 178551
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.9912906885147095,
      "learning_rate": 7.190564205425373e-05,
      "loss": 1.968,
      "step": 178552
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.2402652502059937,
      "learning_rate": 7.190298502995921e-05,
      "loss": 2.1513,
      "step": 178553
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.2782301902770996,
      "learning_rate": 7.190032804807215e-05,
      "loss": 2.068,
      "step": 178554
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.2708873748779297,
      "learning_rate": 7.189767110859321e-05,
      "loss": 1.8523,
      "step": 178555
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.3556535243988037,
      "learning_rate": 7.18950142115227e-05,
      "loss": 2.0049,
      "step": 178556
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.3876389265060425,
      "learning_rate": 7.189235735686138e-05,
      "loss": 2.0299,
      "step": 178557
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.4646481275558472,
      "learning_rate": 7.188970054460941e-05,
      "loss": 2.2275,
      "step": 178558
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.3016594648361206,
      "learning_rate": 7.188704377476755e-05,
      "loss": 1.8344,
      "step": 178559
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.4946362972259521,
      "learning_rate": 7.188438704733611e-05,
      "loss": 2.1758,
      "step": 178560
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.443099856376648,
      "learning_rate": 7.188173036231574e-05,
      "loss": 2.0554,
      "step": 178561
    },
    {
      "epoch": 2.32,
      "grad_norm": 1.4427597522735596,
      "learning_rate": 7.187907371970677e-05,
      "loss": 1.9957,
      "step": 178562
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.2883538007736206,
      "learning_rate": 7.187641711950987e-05,
      "loss": 1.7595,
      "step": 178563
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.3057904243469238,
      "learning_rate": 7.187376056172549e-05,
      "loss": 2.0225,
      "step": 178564
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.4297930002212524,
      "learning_rate": 7.187110404635404e-05,
      "loss": 2.0192,
      "step": 178565
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.392822027206421,
      "learning_rate": 7.1868447573396e-05,
      "loss": 1.8311,
      "step": 178566
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.3660355806350708,
      "learning_rate": 7.1865791142852e-05,
      "loss": 2.1128,
      "step": 178567
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.3193607330322266,
      "learning_rate": 7.186313475472237e-05,
      "loss": 2.0746,
      "step": 178568
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.3502631187438965,
      "learning_rate": 7.186047840900778e-05,
      "loss": 2.2052,
      "step": 178569
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.3238012790679932,
      "learning_rate": 7.185782210570862e-05,
      "loss": 1.9435,
      "step": 178570
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.5202375650405884,
      "learning_rate": 7.185516584482542e-05,
      "loss": 2.0035,
      "step": 178571
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.5134923458099365,
      "learning_rate": 7.185250962635853e-05,
      "loss": 2.0337,
      "step": 178572
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.2791270017623901,
      "learning_rate": 7.184985345030866e-05,
      "loss": 2.0809,
      "step": 178573
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.7844070196151733,
      "learning_rate": 7.184719731667615e-05,
      "loss": 1.8246,
      "step": 178574
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.3137328624725342,
      "learning_rate": 7.184454122546162e-05,
      "loss": 2.1082,
      "step": 178575
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.335541844367981,
      "learning_rate": 7.184188517666548e-05,
      "loss": 1.9815,
      "step": 178576
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.195952296257019,
      "learning_rate": 7.183922917028818e-05,
      "loss": 1.864,
      "step": 178577
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.208834171295166,
      "learning_rate": 7.183657320633035e-05,
      "loss": 1.8279,
      "step": 178578
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.254862904548645,
      "learning_rate": 7.183391728479242e-05,
      "loss": 2.0075,
      "step": 178579
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.287490963935852,
      "learning_rate": 7.183126140567476e-05,
      "loss": 2.0409,
      "step": 178580
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.2187645435333252,
      "learning_rate": 7.182860556897809e-05,
      "loss": 2.0195,
      "step": 178581
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.3058936595916748,
      "learning_rate": 7.182594977470276e-05,
      "loss": 2.1345,
      "step": 178582
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.293968677520752,
      "learning_rate": 7.182329402284923e-05,
      "loss": 1.9227,
      "step": 178583
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.3185068368911743,
      "learning_rate": 7.182063831341813e-05,
      "loss": 1.9631,
      "step": 178584
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.3891154527664185,
      "learning_rate": 7.18179826464099e-05,
      "loss": 2.0069,
      "step": 178585
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.3678778409957886,
      "learning_rate": 7.181532702182493e-05,
      "loss": 2.0781,
      "step": 178586
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.3085746765136719,
      "learning_rate": 7.181267143966386e-05,
      "loss": 1.8692,
      "step": 178587
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.3243744373321533,
      "learning_rate": 7.181001589992712e-05,
      "loss": 2.3352,
      "step": 178588
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.3151386976242065,
      "learning_rate": 7.180736040261514e-05,
      "loss": 2.0031,
      "step": 178589
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.2474727630615234,
      "learning_rate": 7.180470494772856e-05,
      "loss": 2.1274,
      "step": 178590
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.3059018850326538,
      "learning_rate": 7.180204953526771e-05,
      "loss": 2.152,
      "step": 178591
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.3507282733917236,
      "learning_rate": 7.179939416523323e-05,
      "loss": 1.8319,
      "step": 178592
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.3934496641159058,
      "learning_rate": 7.179673883762556e-05,
      "loss": 2.0184,
      "step": 178593
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.3059258460998535,
      "learning_rate": 7.17940835524452e-05,
      "loss": 1.9492,
      "step": 178594
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.3064122200012207,
      "learning_rate": 7.179142830969254e-05,
      "loss": 1.774,
      "step": 178595
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.2497605085372925,
      "learning_rate": 7.178877310936823e-05,
      "loss": 2.0015,
      "step": 178596
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.2908791303634644,
      "learning_rate": 7.178611795147261e-05,
      "loss": 2.1551,
      "step": 178597
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.3535305261611938,
      "learning_rate": 7.178346283600635e-05,
      "loss": 2.1307,
      "step": 178598
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.2140578031539917,
      "learning_rate": 7.178080776296976e-05,
      "loss": 1.9464,
      "step": 178599
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.304517388343811,
      "learning_rate": 7.177815273236362e-05,
      "loss": 2.0324,
      "step": 178600
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.3116894960403442,
      "learning_rate": 7.177549774418805e-05,
      "loss": 1.9468,
      "step": 178601
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.3851968050003052,
      "learning_rate": 7.17728427984438e-05,
      "loss": 1.9509,
      "step": 178602
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.2858723402023315,
      "learning_rate": 7.17701878951312e-05,
      "loss": 1.9206,
      "step": 178603
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.3210418224334717,
      "learning_rate": 7.176753303425094e-05,
      "loss": 2.0823,
      "step": 178604
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.227701187133789,
      "learning_rate": 7.176487821580329e-05,
      "loss": 1.9437,
      "step": 178605
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.242228388786316,
      "learning_rate": 7.176222343978903e-05,
      "loss": 2.0315,
      "step": 178606
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.5283169746398926,
      "learning_rate": 7.175956870620831e-05,
      "loss": 2.0455,
      "step": 178607
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.2702503204345703,
      "learning_rate": 7.17569140150619e-05,
      "loss": 1.9864,
      "step": 178608
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.2824146747589111,
      "learning_rate": 7.175425936635007e-05,
      "loss": 1.8658,
      "step": 178609
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.4003357887268066,
      "learning_rate": 7.175160476007355e-05,
      "loss": 2.0191,
      "step": 178610
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.2168017625808716,
      "learning_rate": 7.17489501962326e-05,
      "loss": 1.7314,
      "step": 178611
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.3742259740829468,
      "learning_rate": 7.174629567482802e-05,
      "loss": 2.1941,
      "step": 178612
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.2031464576721191,
      "learning_rate": 7.174364119585995e-05,
      "loss": 1.859,
      "step": 178613
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.3509408235549927,
      "learning_rate": 7.17409867593291e-05,
      "loss": 2.018,
      "step": 178614
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.4292035102844238,
      "learning_rate": 7.173833236523583e-05,
      "loss": 2.1067,
      "step": 178615
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.368351697921753,
      "learning_rate": 7.173567801358083e-05,
      "loss": 2.0544,
      "step": 178616
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.3744707107543945,
      "learning_rate": 7.173302370436438e-05,
      "loss": 2.0802,
      "step": 178617
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.304279088973999,
      "learning_rate": 7.17303694375872e-05,
      "loss": 2.1294,
      "step": 178618
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.2242023944854736,
      "learning_rate": 7.172771521324953e-05,
      "loss": 1.9937,
      "step": 178619
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.3878883123397827,
      "learning_rate": 7.172506103135206e-05,
      "loss": 1.7691,
      "step": 178620
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.3128492832183838,
      "learning_rate": 7.172240689189511e-05,
      "loss": 1.8893,
      "step": 178621
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.4314517974853516,
      "learning_rate": 7.171975279487938e-05,
      "loss": 2.0887,
      "step": 178622
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.3310272693634033,
      "learning_rate": 7.171709874030516e-05,
      "loss": 2.141,
      "step": 178623
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.3603489398956299,
      "learning_rate": 7.17144447281732e-05,
      "loss": 2.1554,
      "step": 178624
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.3891005516052246,
      "learning_rate": 7.17117907584837e-05,
      "loss": 2.3389,
      "step": 178625
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.2318809032440186,
      "learning_rate": 7.170913683123733e-05,
      "loss": 1.9199,
      "step": 178626
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.2731237411499023,
      "learning_rate": 7.170648294643448e-05,
      "loss": 1.8414,
      "step": 178627
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.3246797323226929,
      "learning_rate": 7.170382910407577e-05,
      "loss": 1.9474,
      "step": 178628
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.3323845863342285,
      "learning_rate": 7.170117530416155e-05,
      "loss": 2.0844,
      "step": 178629
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.3612091541290283,
      "learning_rate": 7.169852154669252e-05,
      "loss": 1.9067,
      "step": 178630
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.181865930557251,
      "learning_rate": 7.169586783166899e-05,
      "loss": 1.9932,
      "step": 178631
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.338366985321045,
      "learning_rate": 7.169321415909151e-05,
      "loss": 1.8608,
      "step": 178632
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.3593153953552246,
      "learning_rate": 7.16905605289605e-05,
      "loss": 1.8311,
      "step": 178633
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.4028503894805908,
      "learning_rate": 7.168790694127662e-05,
      "loss": 2.1345,
      "step": 178634
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.4122378826141357,
      "learning_rate": 7.168525339604018e-05,
      "loss": 2.0414,
      "step": 178635
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.2551707029342651,
      "learning_rate": 7.168259989325181e-05,
      "loss": 1.895,
      "step": 178636
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.5374804735183716,
      "learning_rate": 7.1679946432912e-05,
      "loss": 2.1255,
      "step": 178637
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.8651642799377441,
      "learning_rate": 7.167729301502118e-05,
      "loss": 1.7999,
      "step": 178638
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.4373748302459717,
      "learning_rate": 7.167463963957979e-05,
      "loss": 1.8145,
      "step": 178639
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.2888593673706055,
      "learning_rate": 7.167198630658847e-05,
      "loss": 2.0006,
      "step": 178640
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.3264650106430054,
      "learning_rate": 7.166933301604753e-05,
      "loss": 2.0613,
      "step": 178641
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.322609543800354,
      "learning_rate": 7.166667976795769e-05,
      "loss": 2.0922,
      "step": 178642
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.5270626544952393,
      "learning_rate": 7.16640265623193e-05,
      "loss": 2.3772,
      "step": 178643
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.2054367065429688,
      "learning_rate": 7.166137339913288e-05,
      "loss": 2.2651,
      "step": 178644
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.3616859912872314,
      "learning_rate": 7.165872027839884e-05,
      "loss": 1.906,
      "step": 178645
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.2161900997161865,
      "learning_rate": 7.165606720011785e-05,
      "loss": 1.9897,
      "step": 178646
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.5003776550292969,
      "learning_rate": 7.165341416429021e-05,
      "loss": 2.0694,
      "step": 178647
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.3360912799835205,
      "learning_rate": 7.165076117091659e-05,
      "loss": 1.9655,
      "step": 178648
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.410931944847107,
      "learning_rate": 7.164810821999745e-05,
      "loss": 1.8305,
      "step": 178649
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.966896414756775,
      "learning_rate": 7.164545531153317e-05,
      "loss": 2.0986,
      "step": 178650
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.3989026546478271,
      "learning_rate": 7.164280244552427e-05,
      "loss": 2.0202,
      "step": 178651
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.2378733158111572,
      "learning_rate": 7.164014962197135e-05,
      "loss": 1.9264,
      "step": 178652
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.5350934267044067,
      "learning_rate": 7.163749684087475e-05,
      "loss": 2.1119,
      "step": 178653
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.497108817100525,
      "learning_rate": 7.163484410223517e-05,
      "loss": 2.0416,
      "step": 178654
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.2890186309814453,
      "learning_rate": 7.163219140605295e-05,
      "loss": 2.1804,
      "step": 178655
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.3004591464996338,
      "learning_rate": 7.162953875232861e-05,
      "loss": 2.1322,
      "step": 178656
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.36883544921875,
      "learning_rate": 7.162688614106257e-05,
      "loss": 2.1746,
      "step": 178657
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.331109642982483,
      "learning_rate": 7.16242335722555e-05,
      "loss": 1.8617,
      "step": 178658
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.5350350141525269,
      "learning_rate": 7.162158104590769e-05,
      "loss": 1.9486,
      "step": 178659
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.7244069576263428,
      "learning_rate": 7.161892856201986e-05,
      "loss": 2.3087,
      "step": 178660
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.4281119108200073,
      "learning_rate": 7.161627612059236e-05,
      "loss": 2.0476,
      "step": 178661
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.265602946281433,
      "learning_rate": 7.161362372162564e-05,
      "loss": 1.9418,
      "step": 178662
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.471093773841858,
      "learning_rate": 7.161097136512031e-05,
      "loss": 1.9883,
      "step": 178663
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.4914804697036743,
      "learning_rate": 7.160831905107681e-05,
      "loss": 1.9008,
      "step": 178664
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.583559274673462,
      "learning_rate": 7.160566677949558e-05,
      "loss": 2.0126,
      "step": 178665
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.1851762533187866,
      "learning_rate": 7.160301455037722e-05,
      "loss": 1.9996,
      "step": 178666
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.4901809692382812,
      "learning_rate": 7.160036236372219e-05,
      "loss": 2.2413,
      "step": 178667
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.2833492755889893,
      "learning_rate": 7.15977102195309e-05,
      "loss": 2.0245,
      "step": 178668
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.269557237625122,
      "learning_rate": 7.159505811780397e-05,
      "loss": 2.0329,
      "step": 178669
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.3382558822631836,
      "learning_rate": 7.159240605854184e-05,
      "loss": 2.2614,
      "step": 178670
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.488753080368042,
      "learning_rate": 7.158975404174488e-05,
      "loss": 2.1506,
      "step": 178671
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.231408953666687,
      "learning_rate": 7.158710206741381e-05,
      "loss": 1.8948,
      "step": 178672
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.330393671989441,
      "learning_rate": 7.1584450135549e-05,
      "loss": 2.0931,
      "step": 178673
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.2766749858856201,
      "learning_rate": 7.158179824615085e-05,
      "loss": 2.0232,
      "step": 178674
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.3431384563446045,
      "learning_rate": 7.157914639922007e-05,
      "loss": 2.162,
      "step": 178675
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.4334710836410522,
      "learning_rate": 7.157649459475696e-05,
      "loss": 2.2234,
      "step": 178676
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.3504992723464966,
      "learning_rate": 7.157384283276215e-05,
      "loss": 1.8456,
      "step": 178677
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.3448724746704102,
      "learning_rate": 7.157119111323609e-05,
      "loss": 2.0023,
      "step": 178678
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.3466684818267822,
      "learning_rate": 7.156853943617925e-05,
      "loss": 2.1767,
      "step": 178679
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.2492493391036987,
      "learning_rate": 7.156588780159205e-05,
      "loss": 2.1497,
      "step": 178680
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.3137320280075073,
      "learning_rate": 7.156323620947514e-05,
      "loss": 1.9598,
      "step": 178681
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.3842483758926392,
      "learning_rate": 7.156058465982885e-05,
      "loss": 1.9236,
      "step": 178682
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.283194661140442,
      "learning_rate": 7.155793315265384e-05,
      "loss": 1.9104,
      "step": 178683
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.2320257425308228,
      "learning_rate": 7.155528168795054e-05,
      "loss": 1.9519,
      "step": 178684
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.3118387460708618,
      "learning_rate": 7.15526302657194e-05,
      "loss": 2.0738,
      "step": 178685
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.396119236946106,
      "learning_rate": 7.154997888596088e-05,
      "loss": 2.1141,
      "step": 178686
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.2916079759597778,
      "learning_rate": 7.15473275486756e-05,
      "loss": 1.8331,
      "step": 178687
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.3799148797988892,
      "learning_rate": 7.15446762538639e-05,
      "loss": 2.1927,
      "step": 178688
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.433937430381775,
      "learning_rate": 7.154202500152647e-05,
      "loss": 1.8496,
      "step": 178689
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.4213616847991943,
      "learning_rate": 7.153937379166359e-05,
      "loss": 2.057,
      "step": 178690
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.3209009170532227,
      "learning_rate": 7.153672262427593e-05,
      "loss": 2.0061,
      "step": 178691
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.2605750560760498,
      "learning_rate": 7.15340714993639e-05,
      "loss": 2.0052,
      "step": 178692
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.4488126039505005,
      "learning_rate": 7.153142041692803e-05,
      "loss": 2.0878,
      "step": 178693
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.3303900957107544,
      "learning_rate": 7.152876937696867e-05,
      "loss": 2.1377,
      "step": 178694
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.3581916093826294,
      "learning_rate": 7.152611837948649e-05,
      "loss": 2.1365,
      "step": 178695
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.2965058088302612,
      "learning_rate": 7.152346742448185e-05,
      "loss": 1.9302,
      "step": 178696
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.3331360816955566,
      "learning_rate": 7.152081651195543e-05,
      "loss": 2.1621,
      "step": 178697
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.3579035997390747,
      "learning_rate": 7.151816564190755e-05,
      "loss": 2.03,
      "step": 178698
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.307368278503418,
      "learning_rate": 7.151551481433876e-05,
      "loss": 1.9461,
      "step": 178699
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.2127569913864136,
      "learning_rate": 7.15128640292495e-05,
      "loss": 2.1277,
      "step": 178700
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.2605818510055542,
      "learning_rate": 7.15102132866404e-05,
      "loss": 1.9149,
      "step": 178701
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.238341212272644,
      "learning_rate": 7.150756258651174e-05,
      "loss": 1.9512,
      "step": 178702
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.29432213306427,
      "learning_rate": 7.150491192886422e-05,
      "loss": 1.9945,
      "step": 178703
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.200462818145752,
      "learning_rate": 7.150226131369831e-05,
      "loss": 2.0159,
      "step": 178704
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.2951725721359253,
      "learning_rate": 7.14996107410144e-05,
      "loss": 2.1406,
      "step": 178705
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.324349045753479,
      "learning_rate": 7.149696021081294e-05,
      "loss": 2.0726,
      "step": 178706
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.7546181678771973,
      "learning_rate": 7.149430972309457e-05,
      "loss": 2.0835,
      "step": 178707
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.397781252861023,
      "learning_rate": 7.149165927785969e-05,
      "loss": 2.1245,
      "step": 178708
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.2862517833709717,
      "learning_rate": 7.148900887510889e-05,
      "loss": 1.9768,
      "step": 178709
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.431002140045166,
      "learning_rate": 7.148635851484257e-05,
      "loss": 2.043,
      "step": 178710
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.4016772508621216,
      "learning_rate": 7.148370819706129e-05,
      "loss": 2.0296,
      "step": 178711
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.4035749435424805,
      "learning_rate": 7.148105792176537e-05,
      "loss": 1.9092,
      "step": 178712
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.3599001169204712,
      "learning_rate": 7.147840768895556e-05,
      "loss": 2.1696,
      "step": 178713
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.2202621698379517,
      "learning_rate": 7.147575749863214e-05,
      "loss": 1.8558,
      "step": 178714
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.2474747896194458,
      "learning_rate": 7.147310735079576e-05,
      "loss": 1.9402,
      "step": 178715
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.4106353521347046,
      "learning_rate": 7.147045724544686e-05,
      "loss": 1.9843,
      "step": 178716
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.3055419921875,
      "learning_rate": 7.14678071825859e-05,
      "loss": 1.9959,
      "step": 178717
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.3306424617767334,
      "learning_rate": 7.146515716221329e-05,
      "loss": 2.2431,
      "step": 178718
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.5159673690795898,
      "learning_rate": 7.146250718432973e-05,
      "loss": 2.0708,
      "step": 178719
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.5107972621917725,
      "learning_rate": 7.145985724893551e-05,
      "loss": 2.1201,
      "step": 178720
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.3993215560913086,
      "learning_rate": 7.145720735603132e-05,
      "loss": 2.1098,
      "step": 178721
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.3082200288772583,
      "learning_rate": 7.145455750561751e-05,
      "loss": 2.025,
      "step": 178722
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.2832108736038208,
      "learning_rate": 7.145190769769463e-05,
      "loss": 1.9908,
      "step": 178723
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.3838520050048828,
      "learning_rate": 7.14492579322631e-05,
      "loss": 2.1001,
      "step": 178724
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.4688268899917603,
      "learning_rate": 7.144660820932353e-05,
      "loss": 2.231,
      "step": 178725
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.2489601373672485,
      "learning_rate": 7.144395852887624e-05,
      "loss": 1.9628,
      "step": 178726
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.2599198818206787,
      "learning_rate": 7.144130889092194e-05,
      "loss": 2.0786,
      "step": 178727
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.4361162185668945,
      "learning_rate": 7.143865929546103e-05,
      "loss": 2.1589,
      "step": 178728
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.1926493644714355,
      "learning_rate": 7.143600974249397e-05,
      "loss": 2.0094,
      "step": 178729
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.531556487083435,
      "learning_rate": 7.143336023202119e-05,
      "loss": 2.2285,
      "step": 178730
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.302404522895813,
      "learning_rate": 7.143071076404334e-05,
      "loss": 1.9435,
      "step": 178731
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.2812020778656006,
      "learning_rate": 7.142806133856074e-05,
      "loss": 2.0175,
      "step": 178732
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.305878758430481,
      "learning_rate": 7.142541195557408e-05,
      "loss": 1.9402,
      "step": 178733
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.277600646018982,
      "learning_rate": 7.142276261508375e-05,
      "loss": 1.9358,
      "step": 178734
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.3464099168777466,
      "learning_rate": 7.142011331709024e-05,
      "loss": 2.0671,
      "step": 178735
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.4972032308578491,
      "learning_rate": 7.141746406159396e-05,
      "loss": 1.9956,
      "step": 178736
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.4320074319839478,
      "learning_rate": 7.141481484859559e-05,
      "loss": 1.9442,
      "step": 178737
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.3890432119369507,
      "learning_rate": 7.141216567809542e-05,
      "loss": 2.0127,
      "step": 178738
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.5944619178771973,
      "learning_rate": 7.140951655009411e-05,
      "loss": 2.3022,
      "step": 178739
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.2581548690795898,
      "learning_rate": 7.140686746459212e-05,
      "loss": 2.1081,
      "step": 178740
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.3661510944366455,
      "learning_rate": 7.140421842158989e-05,
      "loss": 1.9945,
      "step": 178741
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.3111776113510132,
      "learning_rate": 7.140156942108784e-05,
      "loss": 1.9521,
      "step": 178742
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.2534077167510986,
      "learning_rate": 7.139892046308666e-05,
      "loss": 1.9118,
      "step": 178743
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.3025023937225342,
      "learning_rate": 7.139627154758664e-05,
      "loss": 1.9374,
      "step": 178744
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.2275917530059814,
      "learning_rate": 7.139362267458847e-05,
      "loss": 2.01,
      "step": 178745
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.3896926641464233,
      "learning_rate": 7.139097384409253e-05,
      "loss": 1.9681,
      "step": 178746
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.3301492929458618,
      "learning_rate": 7.138832505609932e-05,
      "loss": 1.8937,
      "step": 178747
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.1649919748306274,
      "learning_rate": 7.138567631060924e-05,
      "loss": 1.8638,
      "step": 178748
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.4128659963607788,
      "learning_rate": 7.138302760762297e-05,
      "loss": 1.9588,
      "step": 178749
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.3887152671813965,
      "learning_rate": 7.138037894714082e-05,
      "loss": 2.0582,
      "step": 178750
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.3176575899124146,
      "learning_rate": 7.137773032916346e-05,
      "loss": 1.9394,
      "step": 178751
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.409381628036499,
      "learning_rate": 7.137508175369131e-05,
      "loss": 1.9008,
      "step": 178752
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.7419333457946777,
      "learning_rate": 7.137243322072474e-05,
      "loss": 1.9524,
      "step": 178753
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.3156754970550537,
      "learning_rate": 7.136978473026445e-05,
      "loss": 1.8967,
      "step": 178754
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.3492523431777954,
      "learning_rate": 7.136713628231084e-05,
      "loss": 1.9057,
      "step": 178755
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.2932052612304688,
      "learning_rate": 7.136448787686429e-05,
      "loss": 1.9178,
      "step": 178756
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.4035168886184692,
      "learning_rate": 7.136183951392551e-05,
      "loss": 1.8695,
      "step": 178757
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.5971155166625977,
      "learning_rate": 7.135919119349476e-05,
      "loss": 1.9845,
      "step": 178758
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.3005084991455078,
      "learning_rate": 7.135654291557277e-05,
      "loss": 2.1415,
      "step": 178759
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.42377507686615,
      "learning_rate": 7.13538946801599e-05,
      "loss": 2.0228,
      "step": 178760
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.3064560890197754,
      "learning_rate": 7.135124648725658e-05,
      "loss": 1.8534,
      "step": 178761
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.2743034362792969,
      "learning_rate": 7.134859833686344e-05,
      "loss": 2.0805,
      "step": 178762
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.3816145658493042,
      "learning_rate": 7.134595022898092e-05,
      "loss": 2.0798,
      "step": 178763
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.307555079460144,
      "learning_rate": 7.134330216360945e-05,
      "loss": 1.9453,
      "step": 178764
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.270106315612793,
      "learning_rate": 7.134065414074963e-05,
      "loss": 1.7885,
      "step": 178765
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.5310418605804443,
      "learning_rate": 7.13380061604019e-05,
      "loss": 2.176,
      "step": 178766
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.5090993642807007,
      "learning_rate": 7.133535822256668e-05,
      "loss": 2.1213,
      "step": 178767
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.4846231937408447,
      "learning_rate": 7.13327103272446e-05,
      "loss": 2.0403,
      "step": 178768
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.4189058542251587,
      "learning_rate": 7.133006247443612e-05,
      "loss": 1.8508,
      "step": 178769
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.3324779272079468,
      "learning_rate": 7.132741466414157e-05,
      "loss": 1.9853,
      "step": 178770
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.2557719945907593,
      "learning_rate": 7.132476689636168e-05,
      "loss": 1.9577,
      "step": 178771
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.3903731107711792,
      "learning_rate": 7.132211917109684e-05,
      "loss": 2.0037,
      "step": 178772
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.3092896938323975,
      "learning_rate": 7.131947148834743e-05,
      "loss": 2.0741,
      "step": 178773
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.438581109046936,
      "learning_rate": 7.131682384811414e-05,
      "loss": 2.1546,
      "step": 178774
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.5119930505752563,
      "learning_rate": 7.131417625039727e-05,
      "loss": 2.1759,
      "step": 178775
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.406686544418335,
      "learning_rate": 7.131152869519753e-05,
      "loss": 2.0024,
      "step": 178776
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.2276403903961182,
      "learning_rate": 7.130888118251528e-05,
      "loss": 1.7852,
      "step": 178777
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.482681393623352,
      "learning_rate": 7.130623371235101e-05,
      "loss": 1.985,
      "step": 178778
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.278684377670288,
      "learning_rate": 7.130358628470514e-05,
      "loss": 1.8279,
      "step": 178779
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.4869526624679565,
      "learning_rate": 7.130093889957835e-05,
      "loss": 2.028,
      "step": 178780
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.3486676216125488,
      "learning_rate": 7.129829155697096e-05,
      "loss": 2.0167,
      "step": 178781
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.3447848558425903,
      "learning_rate": 7.129564425688361e-05,
      "loss": 1.8674,
      "step": 178782
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.4528346061706543,
      "learning_rate": 7.129299699931669e-05,
      "loss": 2.1704,
      "step": 178783
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.246845006942749,
      "learning_rate": 7.129034978427075e-05,
      "loss": 1.942,
      "step": 178784
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.3107707500457764,
      "learning_rate": 7.128770261174617e-05,
      "loss": 1.8629,
      "step": 178785
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.44302237033844,
      "learning_rate": 7.128505548174361e-05,
      "loss": 1.9962,
      "step": 178786
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.2051664590835571,
      "learning_rate": 7.128240839426335e-05,
      "loss": 1.7776,
      "step": 178787
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.3118880987167358,
      "learning_rate": 7.127976134930613e-05,
      "loss": 2.1139,
      "step": 178788
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.2441726922988892,
      "learning_rate": 7.127711434687229e-05,
      "loss": 1.9541,
      "step": 178789
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.50974440574646,
      "learning_rate": 7.12744673869624e-05,
      "loss": 1.8073,
      "step": 178790
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.4509282112121582,
      "learning_rate": 7.127182046957676e-05,
      "loss": 2.1162,
      "step": 178791
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.358471155166626,
      "learning_rate": 7.126917359471613e-05,
      "loss": 1.998,
      "step": 178792
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.3210723400115967,
      "learning_rate": 7.126652676238079e-05,
      "loss": 2.1002,
      "step": 178793
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.2035967111587524,
      "learning_rate": 7.12638799725714e-05,
      "loss": 1.8558,
      "step": 178794
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.3087806701660156,
      "learning_rate": 7.126123322528836e-05,
      "loss": 1.892,
      "step": 178795
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.4189739227294922,
      "learning_rate": 7.12585865205322e-05,
      "loss": 2.0037,
      "step": 178796
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.4271384477615356,
      "learning_rate": 7.125593985830329e-05,
      "loss": 1.8871,
      "step": 178797
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.2451047897338867,
      "learning_rate": 7.12532932386023e-05,
      "loss": 2.0702,
      "step": 178798
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.2077356576919556,
      "learning_rate": 7.125064666142955e-05,
      "loss": 1.8266,
      "step": 178799
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.2840564250946045,
      "learning_rate": 7.124800012678571e-05,
      "loss": 1.8571,
      "step": 178800
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.3691117763519287,
      "learning_rate": 7.124535363467119e-05,
      "loss": 2.0224,
      "step": 178801
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.496820092201233,
      "learning_rate": 7.124270718508648e-05,
      "loss": 2.0071,
      "step": 178802
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.383604645729065,
      "learning_rate": 7.124006077803198e-05,
      "loss": 1.953,
      "step": 178803
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.477643370628357,
      "learning_rate": 7.123741441350836e-05,
      "loss": 1.9405,
      "step": 178804
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.3929213285446167,
      "learning_rate": 7.123476809151593e-05,
      "loss": 2.0501,
      "step": 178805
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.3171714544296265,
      "learning_rate": 7.123212181205537e-05,
      "loss": 1.8674,
      "step": 178806
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.2294903993606567,
      "learning_rate": 7.122947557512708e-05,
      "loss": 1.8541,
      "step": 178807
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.3257324695587158,
      "learning_rate": 7.122682938073154e-05,
      "loss": 1.9967,
      "step": 178808
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.3216331005096436,
      "learning_rate": 7.122418322886915e-05,
      "loss": 1.9109,
      "step": 178809
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.356092929840088,
      "learning_rate": 7.122153711954059e-05,
      "loss": 2.0126,
      "step": 178810
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.3155301809310913,
      "learning_rate": 7.121889105274622e-05,
      "loss": 2.05,
      "step": 178811
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.2762196063995361,
      "learning_rate": 7.121624502848661e-05,
      "loss": 2.0072,
      "step": 178812
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.3254339694976807,
      "learning_rate": 7.121359904676227e-05,
      "loss": 1.8137,
      "step": 178813
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.2670711278915405,
      "learning_rate": 7.121095310757361e-05,
      "loss": 1.9711,
      "step": 178814
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.3539574146270752,
      "learning_rate": 7.120830721092106e-05,
      "loss": 2.075,
      "step": 178815
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.2012414932250977,
      "learning_rate": 7.12056613568053e-05,
      "loss": 2.0735,
      "step": 178816
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.3036099672317505,
      "learning_rate": 7.120301554522663e-05,
      "loss": 2.2102,
      "step": 178817
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.4107333421707153,
      "learning_rate": 7.120036977618575e-05,
      "loss": 1.9978,
      "step": 178818
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.398747444152832,
      "learning_rate": 7.119772404968303e-05,
      "loss": 2.03,
      "step": 178819
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.319627285003662,
      "learning_rate": 7.119507836571896e-05,
      "loss": 1.9417,
      "step": 178820
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.3054890632629395,
      "learning_rate": 7.119243272429398e-05,
      "loss": 1.9979,
      "step": 178821
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.275531530380249,
      "learning_rate": 7.118978712540873e-05,
      "loss": 2.2553,
      "step": 178822
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.3533015251159668,
      "learning_rate": 7.11871415690635e-05,
      "loss": 1.9642,
      "step": 178823
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.338064432144165,
      "learning_rate": 7.1184496055259e-05,
      "loss": 1.9324,
      "step": 178824
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.2709542512893677,
      "learning_rate": 7.118185058399555e-05,
      "loss": 2.0148,
      "step": 178825
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.2530039548873901,
      "learning_rate": 7.117920515527388e-05,
      "loss": 1.9918,
      "step": 178826
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.2300779819488525,
      "learning_rate": 7.117655976909414e-05,
      "loss": 2.0288,
      "step": 178827
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.2864519357681274,
      "learning_rate": 7.117391442545709e-05,
      "loss": 2.3679,
      "step": 178828
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.2413101196289062,
      "learning_rate": 7.117126912436301e-05,
      "loss": 1.804,
      "step": 178829
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.3529326915740967,
      "learning_rate": 7.116862386581264e-05,
      "loss": 2.0236,
      "step": 178830
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.3870490789413452,
      "learning_rate": 7.116597864980624e-05,
      "loss": 2.0303,
      "step": 178831
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.24777090549469,
      "learning_rate": 7.116333347634457e-05,
      "loss": 2.1684,
      "step": 178832
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.3788177967071533,
      "learning_rate": 7.11606883454278e-05,
      "loss": 1.9384,
      "step": 178833
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.3312287330627441,
      "learning_rate": 7.115804325705664e-05,
      "loss": 2.1556,
      "step": 178834
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.3725413084030151,
      "learning_rate": 7.115539821123144e-05,
      "loss": 1.8667,
      "step": 178835
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.3427482843399048,
      "learning_rate": 7.115275320795288e-05,
      "loss": 2.1186,
      "step": 178836
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.2435131072998047,
      "learning_rate": 7.115010824722124e-05,
      "loss": 1.832,
      "step": 178837
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.3191436529159546,
      "learning_rate": 7.114746332903721e-05,
      "loss": 1.9934,
      "step": 178838
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.3349984884262085,
      "learning_rate": 7.114481845340117e-05,
      "loss": 2.0328,
      "step": 178839
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.4427173137664795,
      "learning_rate": 7.114217362031362e-05,
      "loss": 2.0606,
      "step": 178840
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.3182530403137207,
      "learning_rate": 7.113952882977499e-05,
      "loss": 2.2035,
      "step": 178841
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.6476962566375732,
      "learning_rate": 7.113688408178593e-05,
      "loss": 2.0643,
      "step": 178842
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.312986969947815,
      "learning_rate": 7.113423937634676e-05,
      "loss": 2.106,
      "step": 178843
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.3640928268432617,
      "learning_rate": 7.113159471345813e-05,
      "loss": 2.0288,
      "step": 178844
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.474779725074768,
      "learning_rate": 7.112895009312046e-05,
      "loss": 2.0616,
      "step": 178845
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.680890440940857,
      "learning_rate": 7.112630551533416e-05,
      "loss": 1.8436,
      "step": 178846
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.455203652381897,
      "learning_rate": 7.11236609800999e-05,
      "loss": 1.9903,
      "step": 178847
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.3387954235076904,
      "learning_rate": 7.112101648741802e-05,
      "loss": 2.0545,
      "step": 178848
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.2078940868377686,
      "learning_rate": 7.111837203728902e-05,
      "loss": 1.8732,
      "step": 178849
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.310767650604248,
      "learning_rate": 7.111572762971351e-05,
      "loss": 1.9342,
      "step": 178850
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.6312638521194458,
      "learning_rate": 7.111308326469192e-05,
      "loss": 1.8654,
      "step": 178851
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.3852170705795288,
      "learning_rate": 7.111043894222462e-05,
      "loss": 1.9966,
      "step": 178852
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.3041067123413086,
      "learning_rate": 7.110779466231232e-05,
      "loss": 1.986,
      "step": 178853
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.1507145166397095,
      "learning_rate": 7.110515042495539e-05,
      "loss": 1.8773,
      "step": 178854
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.3139476776123047,
      "learning_rate": 7.110250623015423e-05,
      "loss": 1.9089,
      "step": 178855
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.3572273254394531,
      "learning_rate": 7.109986207790954e-05,
      "loss": 2.085,
      "step": 178856
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.3918061256408691,
      "learning_rate": 7.10972179682217e-05,
      "loss": 2.0787,
      "step": 178857
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.4824848175048828,
      "learning_rate": 7.109457390109114e-05,
      "loss": 2.0767,
      "step": 178858
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.2673190832138062,
      "learning_rate": 7.109192987651848e-05,
      "loss": 2.2547,
      "step": 178859
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.4131914377212524,
      "learning_rate": 7.108928589450406e-05,
      "loss": 2.0283,
      "step": 178860
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.2783191204071045,
      "learning_rate": 7.108664195504858e-05,
      "loss": 2.075,
      "step": 178861
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.3402303457260132,
      "learning_rate": 7.10839980581524e-05,
      "loss": 2.0113,
      "step": 178862
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.544149398803711,
      "learning_rate": 7.108135420381601e-05,
      "loss": 2.1821,
      "step": 178863
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.254884958267212,
      "learning_rate": 7.107871039203984e-05,
      "loss": 2.1268,
      "step": 178864
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.333096981048584,
      "learning_rate": 7.107606662282456e-05,
      "loss": 2.0103,
      "step": 178865
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.2787420749664307,
      "learning_rate": 7.107342289617046e-05,
      "loss": 2.1615,
      "step": 178866
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.3644846677780151,
      "learning_rate": 7.107077921207823e-05,
      "loss": 1.9152,
      "step": 178867
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.4053058624267578,
      "learning_rate": 7.106813557054827e-05,
      "loss": 2.1095,
      "step": 178868
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.299259901046753,
      "learning_rate": 7.106549197158104e-05,
      "loss": 1.9124,
      "step": 178869
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.3525471687316895,
      "learning_rate": 7.106284841517698e-05,
      "loss": 2.0577,
      "step": 178870
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.3593292236328125,
      "learning_rate": 7.106020490133676e-05,
      "loss": 2.1357,
      "step": 178871
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.4195059537887573,
      "learning_rate": 7.105756143006065e-05,
      "loss": 1.9789,
      "step": 178872
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.3525128364562988,
      "learning_rate": 7.105491800134936e-05,
      "loss": 1.9891,
      "step": 178873
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.2704306840896606,
      "learning_rate": 7.105227461520328e-05,
      "loss": 1.966,
      "step": 178874
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.3832347393035889,
      "learning_rate": 7.104963127162294e-05,
      "loss": 2.0583,
      "step": 178875
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.5567891597747803,
      "learning_rate": 7.10469879706087e-05,
      "loss": 1.9571,
      "step": 178876
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.3379206657409668,
      "learning_rate": 7.10443447121612e-05,
      "loss": 1.9605,
      "step": 178877
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.375701665878296,
      "learning_rate": 7.10417014962808e-05,
      "loss": 2.004,
      "step": 178878
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.355291485786438,
      "learning_rate": 7.103905832296818e-05,
      "loss": 2.3222,
      "step": 178879
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.303577184677124,
      "learning_rate": 7.103641519222368e-05,
      "loss": 1.8933,
      "step": 178880
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.4201446771621704,
      "learning_rate": 7.103377210404785e-05,
      "loss": 1.9567,
      "step": 178881
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.4761340618133545,
      "learning_rate": 7.10311290584411e-05,
      "loss": 1.9985,
      "step": 178882
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.3607755899429321,
      "learning_rate": 7.102848605540404e-05,
      "loss": 2.0627,
      "step": 178883
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.2988343238830566,
      "learning_rate": 7.102584309493706e-05,
      "loss": 2.0629,
      "step": 178884
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.4288169145584106,
      "learning_rate": 7.102320017704076e-05,
      "loss": 2.1195,
      "step": 178885
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.361540675163269,
      "learning_rate": 7.102055730171554e-05,
      "loss": 1.8232,
      "step": 178886
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.3027147054672241,
      "learning_rate": 7.101791446896197e-05,
      "loss": 1.8997,
      "step": 178887
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.3829798698425293,
      "learning_rate": 7.101527167878037e-05,
      "loss": 2.0437,
      "step": 178888
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.3304429054260254,
      "learning_rate": 7.101262893117146e-05,
      "loss": 2.1832,
      "step": 178889
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.4171099662780762,
      "learning_rate": 7.100998622613554e-05,
      "loss": 2.0879,
      "step": 178890
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.2088128328323364,
      "learning_rate": 7.100734356367326e-05,
      "loss": 1.9531,
      "step": 178891
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.3696578741073608,
      "learning_rate": 7.100470094378498e-05,
      "loss": 2.1448,
      "step": 178892
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.4036451578140259,
      "learning_rate": 7.100205836647137e-05,
      "loss": 2.0181,
      "step": 178893
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.3024866580963135,
      "learning_rate": 7.099941583173267e-05,
      "loss": 2.0582,
      "step": 178894
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.1944843530654907,
      "learning_rate": 7.09967733395696e-05,
      "loss": 1.9583,
      "step": 178895
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.95248544216156,
      "learning_rate": 7.099413088998241e-05,
      "loss": 1.9802,
      "step": 178896
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.6517130136489868,
      "learning_rate": 7.099148848297184e-05,
      "loss": 1.984,
      "step": 178897
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.380849838256836,
      "learning_rate": 7.098884611853822e-05,
      "loss": 1.9852,
      "step": 178898
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.2858325242996216,
      "learning_rate": 7.098620379668224e-05,
      "loss": 1.8707,
      "step": 178899
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.409056544303894,
      "learning_rate": 7.09835615174041e-05,
      "loss": 2.1953,
      "step": 178900
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.5658897161483765,
      "learning_rate": 7.09809192807045e-05,
      "loss": 2.0111,
      "step": 178901
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.5663915872573853,
      "learning_rate": 7.09782770865838e-05,
      "loss": 2.0427,
      "step": 178902
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.470563530921936,
      "learning_rate": 7.097563493504262e-05,
      "loss": 1.8731,
      "step": 178903
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.2801759243011475,
      "learning_rate": 7.097299282608133e-05,
      "loss": 2.1161,
      "step": 178904
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.265105962753296,
      "learning_rate": 7.097035075970066e-05,
      "loss": 1.8077,
      "step": 178905
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.3273634910583496,
      "learning_rate": 7.096770873590075e-05,
      "loss": 2.096,
      "step": 178906
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.5369071960449219,
      "learning_rate": 7.096506675468236e-05,
      "loss": 2.0458,
      "step": 178907
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.935295820236206,
      "learning_rate": 7.096242481604576e-05,
      "loss": 1.9893,
      "step": 178908
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.3680967092514038,
      "learning_rate": 7.09597829199917e-05,
      "loss": 2.1517,
      "step": 178909
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.2565360069274902,
      "learning_rate": 7.095714106652042e-05,
      "loss": 1.8812,
      "step": 178910
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.3388738632202148,
      "learning_rate": 7.095449925563274e-05,
      "loss": 2.0301,
      "step": 178911
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.202884316444397,
      "learning_rate": 7.095185748732876e-05,
      "loss": 2.2915,
      "step": 178912
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.3312524557113647,
      "learning_rate": 7.094921576160921e-05,
      "loss": 1.8811,
      "step": 178913
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.4346469640731812,
      "learning_rate": 7.094657407847448e-05,
      "loss": 1.8589,
      "step": 178914
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.2452210187911987,
      "learning_rate": 7.094393243792517e-05,
      "loss": 2.1128,
      "step": 178915
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.2688053846359253,
      "learning_rate": 7.09412908399616e-05,
      "loss": 2.1877,
      "step": 178916
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.3235619068145752,
      "learning_rate": 7.093864928458456e-05,
      "loss": 2.0087,
      "step": 178917
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.668294072151184,
      "learning_rate": 7.09360077717942e-05,
      "loss": 1.9806,
      "step": 178918
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.5573517084121704,
      "learning_rate": 7.093336630159123e-05,
      "loss": 1.9775,
      "step": 178919
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.3013156652450562,
      "learning_rate": 7.093072487397595e-05,
      "loss": 2.0139,
      "step": 178920
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.3005661964416504,
      "learning_rate": 7.092808348894908e-05,
      "loss": 2.0688,
      "step": 178921
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.2514357566833496,
      "learning_rate": 7.092544214651095e-05,
      "loss": 2.0493,
      "step": 178922
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.3032087087631226,
      "learning_rate": 7.092280084666215e-05,
      "loss": 2.1983,
      "step": 178923
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.4630321264266968,
      "learning_rate": 7.092015958940314e-05,
      "loss": 1.9447,
      "step": 178924
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.5851383209228516,
      "learning_rate": 7.09175183747344e-05,
      "loss": 2.1359,
      "step": 178925
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.4234731197357178,
      "learning_rate": 7.091487720265632e-05,
      "loss": 1.9955,
      "step": 178926
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.229240894317627,
      "learning_rate": 7.09122360731696e-05,
      "loss": 1.9959,
      "step": 178927
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.2564036846160889,
      "learning_rate": 7.090959498627452e-05,
      "loss": 2.0318,
      "step": 178928
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.4817721843719482,
      "learning_rate": 7.090695394197174e-05,
      "loss": 2.0109,
      "step": 178929
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.3370150327682495,
      "learning_rate": 7.090431294026171e-05,
      "loss": 1.8855,
      "step": 178930
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.430917501449585,
      "learning_rate": 7.090167198114489e-05,
      "loss": 1.9394,
      "step": 178931
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.3879523277282715,
      "learning_rate": 7.089903106462165e-05,
      "loss": 2.1721,
      "step": 178932
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.397988200187683,
      "learning_rate": 7.089639019069271e-05,
      "loss": 2.1408,
      "step": 178933
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.4828453063964844,
      "learning_rate": 7.089374935935839e-05,
      "loss": 2.1127,
      "step": 178934
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.3109992742538452,
      "learning_rate": 7.089110857061931e-05,
      "loss": 1.987,
      "step": 178935
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.3643518686294556,
      "learning_rate": 7.088846782447592e-05,
      "loss": 2.0153,
      "step": 178936
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.3474067449569702,
      "learning_rate": 7.08858271209286e-05,
      "loss": 2.1046,
      "step": 178937
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.3201416730880737,
      "learning_rate": 7.0883186459978e-05,
      "loss": 2.0524,
      "step": 178938
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.4722238779067993,
      "learning_rate": 7.088054584162454e-05,
      "loss": 2.2062,
      "step": 178939
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.313156008720398,
      "learning_rate": 7.087790526586863e-05,
      "loss": 1.9124,
      "step": 178940
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.3198305368423462,
      "learning_rate": 7.087526473271093e-05,
      "loss": 2.2325,
      "step": 178941
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.3540537357330322,
      "learning_rate": 7.087262424215183e-05,
      "loss": 1.8833,
      "step": 178942
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.3981890678405762,
      "learning_rate": 7.086998379419176e-05,
      "loss": 2.0636,
      "step": 178943
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.2713881731033325,
      "learning_rate": 7.086734338883136e-05,
      "loss": 1.9572,
      "step": 178944
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.2012075185775757,
      "learning_rate": 7.0864703026071e-05,
      "loss": 1.8863,
      "step": 178945
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.4411330223083496,
      "learning_rate": 7.086206270591126e-05,
      "loss": 1.9474,
      "step": 178946
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.359139084815979,
      "learning_rate": 7.085942242835261e-05,
      "loss": 2.2755,
      "step": 178947
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.3274919986724854,
      "learning_rate": 7.085678219339552e-05,
      "loss": 2.1852,
      "step": 178948
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.2517986297607422,
      "learning_rate": 7.085414200104037e-05,
      "loss": 1.9129,
      "step": 178949
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.3225452899932861,
      "learning_rate": 7.085150185128788e-05,
      "loss": 2.0122,
      "step": 178950
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.3832836151123047,
      "learning_rate": 7.084886174413832e-05,
      "loss": 1.8971,
      "step": 178951
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.5760869979858398,
      "learning_rate": 7.084622167959235e-05,
      "loss": 1.9633,
      "step": 178952
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.3938654661178589,
      "learning_rate": 7.084358165765042e-05,
      "loss": 2.0456,
      "step": 178953
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.2931416034698486,
      "learning_rate": 7.084094167831298e-05,
      "loss": 1.8302,
      "step": 178954
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.3662251234054565,
      "learning_rate": 7.083830174158047e-05,
      "loss": 2.2256,
      "step": 178955
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.336300015449524,
      "learning_rate": 7.083566184745349e-05,
      "loss": 2.0363,
      "step": 178956
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.3427813053131104,
      "learning_rate": 7.083302199593241e-05,
      "loss": 1.9513,
      "step": 178957
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.2402135133743286,
      "learning_rate": 7.083038218701793e-05,
      "loss": 1.7834,
      "step": 178958
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.3364077806472778,
      "learning_rate": 7.082774242071028e-05,
      "loss": 2.0851,
      "step": 178959
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.2988333702087402,
      "learning_rate": 7.082510269701025e-05,
      "loss": 2.0905,
      "step": 178960
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.304510235786438,
      "learning_rate": 7.082246301591801e-05,
      "loss": 2.0616,
      "step": 178961
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.2404245138168335,
      "learning_rate": 7.081982337743425e-05,
      "loss": 1.986,
      "step": 178962
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.3119852542877197,
      "learning_rate": 7.081718378155934e-05,
      "loss": 2.0547,
      "step": 178963
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.3921077251434326,
      "learning_rate": 7.081454422829394e-05,
      "loss": 1.8012,
      "step": 178964
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.3703043460845947,
      "learning_rate": 7.081190471763837e-05,
      "loss": 2.1309,
      "step": 178965
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.256638526916504,
      "learning_rate": 7.080926524959333e-05,
      "loss": 1.9322,
      "step": 178966
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.168048620223999,
      "learning_rate": 7.080662582415903e-05,
      "loss": 2.2168,
      "step": 178967
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.3387477397918701,
      "learning_rate": 7.080398644133616e-05,
      "loss": 1.9316,
      "step": 178968
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.2464611530303955,
      "learning_rate": 7.08013471011251e-05,
      "loss": 2.1861,
      "step": 178969
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.2953376770019531,
      "learning_rate": 7.079870780352646e-05,
      "loss": 1.9931,
      "step": 178970
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.3340202569961548,
      "learning_rate": 7.079606854854061e-05,
      "loss": 2.0302,
      "step": 178971
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.3262784481048584,
      "learning_rate": 7.079342933616823e-05,
      "loss": 1.8847,
      "step": 178972
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.3653991222381592,
      "learning_rate": 7.079079016640952e-05,
      "loss": 1.9618,
      "step": 178973
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.3120603561401367,
      "learning_rate": 7.078815103926524e-05,
      "loss": 1.8032,
      "step": 178974
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.581261396408081,
      "learning_rate": 7.078551195473567e-05,
      "loss": 2.1698,
      "step": 178975
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.549004316329956,
      "learning_rate": 7.078287291282147e-05,
      "loss": 1.9272,
      "step": 178976
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.3404173851013184,
      "learning_rate": 7.078023391352297e-05,
      "loss": 1.8427,
      "step": 178977
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.7066341638565063,
      "learning_rate": 7.077759495684097e-05,
      "loss": 1.996,
      "step": 178978
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.4223840236663818,
      "learning_rate": 7.077495604277553e-05,
      "loss": 2.2679,
      "step": 178979
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.4762974977493286,
      "learning_rate": 7.077231717132744e-05,
      "loss": 2.067,
      "step": 178980
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.3270295858383179,
      "learning_rate": 7.076967834249703e-05,
      "loss": 1.8273,
      "step": 178981
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.317411184310913,
      "learning_rate": 7.076703955628496e-05,
      "loss": 2.025,
      "step": 178982
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.3029956817626953,
      "learning_rate": 7.07644008126915e-05,
      "loss": 1.9884,
      "step": 178983
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.350435733795166,
      "learning_rate": 7.076176211171747e-05,
      "loss": 2.1567,
      "step": 178984
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.3387084007263184,
      "learning_rate": 7.075912345336295e-05,
      "loss": 1.9955,
      "step": 178985
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.2925647497177124,
      "learning_rate": 7.075648483762878e-05,
      "loss": 1.9494,
      "step": 178986
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.3961262702941895,
      "learning_rate": 7.075384626451517e-05,
      "loss": 2.0601,
      "step": 178987
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.3082236051559448,
      "learning_rate": 7.075120773402286e-05,
      "loss": 2.0747,
      "step": 178988
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.2525218725204468,
      "learning_rate": 7.074856924615214e-05,
      "loss": 2.0576,
      "step": 178989
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.3181746006011963,
      "learning_rate": 7.074593080090375e-05,
      "loss": 2.0845,
      "step": 178990
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.9174162149429321,
      "learning_rate": 7.074329239827784e-05,
      "loss": 2.0248,
      "step": 178991
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.4342535734176636,
      "learning_rate": 7.074065403827518e-05,
      "loss": 2.1442,
      "step": 178992
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.258406162261963,
      "learning_rate": 7.073801572089607e-05,
      "loss": 2.1122,
      "step": 178993
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.3719682693481445,
      "learning_rate": 7.073537744614118e-05,
      "loss": 1.9413,
      "step": 178994
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.3036292791366577,
      "learning_rate": 7.073273921401082e-05,
      "loss": 1.9244,
      "step": 178995
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.4263818264007568,
      "learning_rate": 7.073010102450572e-05,
      "loss": 1.9369,
      "step": 178996
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.307779312133789,
      "learning_rate": 7.072746287762607e-05,
      "loss": 2.0651,
      "step": 178997
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.2821199893951416,
      "learning_rate": 7.072482477337263e-05,
      "loss": 2.0268,
      "step": 178998
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.486783742904663,
      "learning_rate": 7.072218671174564e-05,
      "loss": 2.1776,
      "step": 178999
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.3801416158676147,
      "learning_rate": 7.071954869274583e-05,
      "loss": 1.75,
      "step": 179000
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.4473408460617065,
      "learning_rate": 7.071691071637348e-05,
      "loss": 2.2342,
      "step": 179001
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.433829665184021,
      "learning_rate": 7.071427278262937e-05,
      "loss": 2.0002,
      "step": 179002
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.3557837009429932,
      "learning_rate": 7.071163489151366e-05,
      "loss": 1.9234,
      "step": 179003
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.3718054294586182,
      "learning_rate": 7.070899704302707e-05,
      "loss": 2.0294,
      "step": 179004
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.341606855392456,
      "learning_rate": 7.070635923716989e-05,
      "loss": 2.0659,
      "step": 179005
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.893262267112732,
      "learning_rate": 7.070372147394281e-05,
      "loss": 1.7964,
      "step": 179006
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.3185985088348389,
      "learning_rate": 7.070108375334614e-05,
      "loss": 2.1138,
      "step": 179007
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.321714162826538,
      "learning_rate": 7.069844607538059e-05,
      "loss": 2.1303,
      "step": 179008
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.3948390483856201,
      "learning_rate": 7.069580844004651e-05,
      "loss": 2.0516,
      "step": 179009
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.353094458580017,
      "learning_rate": 7.069317084734439e-05,
      "loss": 1.9075,
      "step": 179010
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.5815609693527222,
      "learning_rate": 7.069053329727467e-05,
      "loss": 2.1193,
      "step": 179011
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.9348750114440918,
      "learning_rate": 7.068789578983799e-05,
      "loss": 1.9765,
      "step": 179012
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.3253265619277954,
      "learning_rate": 7.06852583250347e-05,
      "loss": 2.0106,
      "step": 179013
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.2676783800125122,
      "learning_rate": 7.06826209028654e-05,
      "loss": 1.9969,
      "step": 179014
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.3799058198928833,
      "learning_rate": 7.067998352333055e-05,
      "loss": 2.1454,
      "step": 179015
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.4201536178588867,
      "learning_rate": 7.067734618643061e-05,
      "loss": 1.9282,
      "step": 179016
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.2690197229385376,
      "learning_rate": 7.0674708892166e-05,
      "loss": 2.0795,
      "step": 179017
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.4121861457824707,
      "learning_rate": 7.067207164053735e-05,
      "loss": 2.0252,
      "step": 179018
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.57881498336792,
      "learning_rate": 7.066943443154503e-05,
      "loss": 2.1934,
      "step": 179019
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.2204605340957642,
      "learning_rate": 7.066679726518968e-05,
      "loss": 1.9562,
      "step": 179020
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.3432022333145142,
      "learning_rate": 7.066416014147168e-05,
      "loss": 1.9266,
      "step": 179021
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.2899420261383057,
      "learning_rate": 7.066152306039147e-05,
      "loss": 1.9662,
      "step": 179022
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.2991863489151,
      "learning_rate": 7.06588860219497e-05,
      "loss": 2.1198,
      "step": 179023
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.3821159601211548,
      "learning_rate": 7.065624902614679e-05,
      "loss": 1.9322,
      "step": 179024
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.507850170135498,
      "learning_rate": 7.065361207298309e-05,
      "loss": 2.2383,
      "step": 179025
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.3536632061004639,
      "learning_rate": 7.065097516245932e-05,
      "loss": 1.9977,
      "step": 179026
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.2243337631225586,
      "learning_rate": 7.064833829457584e-05,
      "loss": 1.8677,
      "step": 179027
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.3652440309524536,
      "learning_rate": 7.064570146933313e-05,
      "loss": 1.8011,
      "step": 179028
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.5870333909988403,
      "learning_rate": 7.064306468673177e-05,
      "loss": 2.0611,
      "step": 179029
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.4465534687042236,
      "learning_rate": 7.064042794677216e-05,
      "loss": 1.983,
      "step": 179030
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.3047088384628296,
      "learning_rate": 7.063779124945477e-05,
      "loss": 2.1007,
      "step": 179031
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.2140841484069824,
      "learning_rate": 7.063515459478025e-05,
      "loss": 2.183,
      "step": 179032
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.428580641746521,
      "learning_rate": 7.063251798274895e-05,
      "loss": 2.097,
      "step": 179033
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.390657901763916,
      "learning_rate": 7.062988141336133e-05,
      "loss": 2.1059,
      "step": 179034
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.4547245502471924,
      "learning_rate": 7.062724488661801e-05,
      "loss": 2.2863,
      "step": 179035
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.3231768608093262,
      "learning_rate": 7.062460840251933e-05,
      "loss": 2.0407,
      "step": 179036
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.3375704288482666,
      "learning_rate": 7.062197196106598e-05,
      "loss": 2.0054,
      "step": 179037
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.2845239639282227,
      "learning_rate": 7.061933556225831e-05,
      "loss": 2.0117,
      "step": 179038
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.248044729232788,
      "learning_rate": 7.061669920609683e-05,
      "loss": 1.9075,
      "step": 179039
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.4150660037994385,
      "learning_rate": 7.061406289258198e-05,
      "loss": 2.1344,
      "step": 179040
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.4250651597976685,
      "learning_rate": 7.061142662171437e-05,
      "loss": 2.1818,
      "step": 179041
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.4830583333969116,
      "learning_rate": 7.060879039349432e-05,
      "loss": 2.147,
      "step": 179042
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.3168251514434814,
      "learning_rate": 7.060615420792255e-05,
      "loss": 2.0019,
      "step": 179043
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.3506755828857422,
      "learning_rate": 7.060351806499933e-05,
      "loss": 1.8942,
      "step": 179044
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.394891619682312,
      "learning_rate": 7.06008819647254e-05,
      "loss": 2.023,
      "step": 179045
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.3388612270355225,
      "learning_rate": 7.059824590710095e-05,
      "loss": 2.0108,
      "step": 179046
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.422863245010376,
      "learning_rate": 7.059560989212668e-05,
      "loss": 2.0383,
      "step": 179047
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.1854510307312012,
      "learning_rate": 7.059297391980294e-05,
      "loss": 1.9257,
      "step": 179048
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.2818644046783447,
      "learning_rate": 7.059033799013039e-05,
      "loss": 2.0561,
      "step": 179049
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.3267580270767212,
      "learning_rate": 7.058770210310934e-05,
      "loss": 2.2112,
      "step": 179050
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.3363664150238037,
      "learning_rate": 7.058506625874051e-05,
      "loss": 1.9487,
      "step": 179051
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.2980291843414307,
      "learning_rate": 7.058243045702412e-05,
      "loss": 2.1863,
      "step": 179052
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.3684840202331543,
      "learning_rate": 7.057979469796084e-05,
      "loss": 1.9367,
      "step": 179053
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.4124583005905151,
      "learning_rate": 7.057715898155101e-05,
      "loss": 2.0531,
      "step": 179054
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.3416296243667603,
      "learning_rate": 7.057452330779534e-05,
      "loss": 2.1332,
      "step": 179055
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.2976062297821045,
      "learning_rate": 7.057188767669411e-05,
      "loss": 2.1284,
      "step": 179056
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.3960916996002197,
      "learning_rate": 7.056925208824806e-05,
      "loss": 2.088,
      "step": 179057
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.4711846113204956,
      "learning_rate": 7.056661654245731e-05,
      "loss": 2.1834,
      "step": 179058
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.4965100288391113,
      "learning_rate": 7.056398103932268e-05,
      "loss": 1.8175,
      "step": 179059
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.3402827978134155,
      "learning_rate": 7.056134557884443e-05,
      "loss": 2.248,
      "step": 179060
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.5083428621292114,
      "learning_rate": 7.055871016102326e-05,
      "loss": 2.0675,
      "step": 179061
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.2984697818756104,
      "learning_rate": 7.055607478585946e-05,
      "loss": 2.1104,
      "step": 179062
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.3931565284729004,
      "learning_rate": 7.05534394533538e-05,
      "loss": 1.7745,
      "step": 179063
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.2692219018936157,
      "learning_rate": 7.055080416350641e-05,
      "loss": 1.9628,
      "step": 179064
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.5227488279342651,
      "learning_rate": 7.054816891631804e-05,
      "loss": 2.2147,
      "step": 179065
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.555679440498352,
      "learning_rate": 7.054553371178902e-05,
      "loss": 2.0113,
      "step": 179066
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.3316643238067627,
      "learning_rate": 7.054289854992e-05,
      "loss": 2.0408,
      "step": 179067
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.253641128540039,
      "learning_rate": 7.05402634307113e-05,
      "loss": 2.0387,
      "step": 179068
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.5130506753921509,
      "learning_rate": 7.053762835416363e-05,
      "loss": 2.0362,
      "step": 179069
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.3165760040283203,
      "learning_rate": 7.053499332027721e-05,
      "loss": 1.9413,
      "step": 179070
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.3310805559158325,
      "learning_rate": 7.053235832905276e-05,
      "loss": 1.8579,
      "step": 179071
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.4443806409835815,
      "learning_rate": 7.052972338049059e-05,
      "loss": 1.9671,
      "step": 179072
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.2883977890014648,
      "learning_rate": 7.052708847459135e-05,
      "loss": 1.9263,
      "step": 179073
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.285441517829895,
      "learning_rate": 7.052445361135536e-05,
      "loss": 2.0632,
      "step": 179074
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.268189787864685,
      "learning_rate": 7.05218187907834e-05,
      "loss": 2.1059,
      "step": 179075
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.3090816736221313,
      "learning_rate": 7.051918401287558e-05,
      "loss": 2.0685,
      "step": 179076
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.314294457435608,
      "learning_rate": 7.051654927763267e-05,
      "loss": 1.8913,
      "step": 179077
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.2994918823242188,
      "learning_rate": 7.051391458505497e-05,
      "loss": 2.0407,
      "step": 179078
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.179563283920288,
      "learning_rate": 7.051127993514316e-05,
      "loss": 1.9452,
      "step": 179079
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.2407110929489136,
      "learning_rate": 7.050864532789753e-05,
      "loss": 2.0668,
      "step": 179080
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.2853556871414185,
      "learning_rate": 7.050601076331887e-05,
      "loss": 1.9056,
      "step": 179081
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.2733633518218994,
      "learning_rate": 7.05033762414073e-05,
      "loss": 1.8176,
      "step": 179082
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.3027217388153076,
      "learning_rate": 7.050074176216355e-05,
      "loss": 1.7998,
      "step": 179083
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.2787859439849854,
      "learning_rate": 7.0498107325588e-05,
      "loss": 2.0201,
      "step": 179084
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.3173611164093018,
      "learning_rate": 7.049547293168124e-05,
      "loss": 2.0677,
      "step": 179085
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.3233007192611694,
      "learning_rate": 7.049283858044359e-05,
      "loss": 1.9743,
      "step": 179086
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.3339580297470093,
      "learning_rate": 7.049020427187587e-05,
      "loss": 1.9286,
      "step": 179087
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.2527340650558472,
      "learning_rate": 7.048757000597816e-05,
      "loss": 1.8966,
      "step": 179088
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.327210783958435,
      "learning_rate": 7.048493578275124e-05,
      "loss": 2.0535,
      "step": 179089
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.3280706405639648,
      "learning_rate": 7.048230160219543e-05,
      "loss": 2.2656,
      "step": 179090
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.2695773839950562,
      "learning_rate": 7.047966746431133e-05,
      "loss": 1.9068,
      "step": 179091
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.3209455013275146,
      "learning_rate": 7.047703336909935e-05,
      "loss": 1.9996,
      "step": 179092
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.2776237726211548,
      "learning_rate": 7.04743993165601e-05,
      "loss": 1.9529,
      "step": 179093
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.2808541059494019,
      "learning_rate": 7.047176530669398e-05,
      "loss": 2.1986,
      "step": 179094
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.3101166486740112,
      "learning_rate": 7.046913133950152e-05,
      "loss": 2.0595,
      "step": 179095
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.295613169670105,
      "learning_rate": 7.046649741498308e-05,
      "loss": 1.8538,
      "step": 179096
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.4233410358428955,
      "learning_rate": 7.046386353313932e-05,
      "loss": 2.0073,
      "step": 179097
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.3284732103347778,
      "learning_rate": 7.046122969397058e-05,
      "loss": 1.8418,
      "step": 179098
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.3040456771850586,
      "learning_rate": 7.045859589747754e-05,
      "loss": 1.9495,
      "step": 179099
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.3890715837478638,
      "learning_rate": 7.045596214366054e-05,
      "loss": 2.0249,
      "step": 179100
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.2866206169128418,
      "learning_rate": 7.045332843252012e-05,
      "loss": 2.2227,
      "step": 179101
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.3623929023742676,
      "learning_rate": 7.045069476405672e-05,
      "loss": 2.0526,
      "step": 179102
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.3224784135818481,
      "learning_rate": 7.044806113827089e-05,
      "loss": 1.904,
      "step": 179103
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.385898232460022,
      "learning_rate": 7.044542755516307e-05,
      "loss": 1.8333,
      "step": 179104
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.3550628423690796,
      "learning_rate": 7.044279401473383e-05,
      "loss": 2.0607,
      "step": 179105
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.214394450187683,
      "learning_rate": 7.04401605169836e-05,
      "loss": 2.0657,
      "step": 179106
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.3708035945892334,
      "learning_rate": 7.043752706191282e-05,
      "loss": 1.9863,
      "step": 179107
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.2927947044372559,
      "learning_rate": 7.043489364952211e-05,
      "loss": 2.0754,
      "step": 179108
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.3272687196731567,
      "learning_rate": 7.043226027981188e-05,
      "loss": 1.7802,
      "step": 179109
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.53487229347229,
      "learning_rate": 7.042962695278255e-05,
      "loss": 1.8531,
      "step": 179110
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.3711391687393188,
      "learning_rate": 7.042699366843477e-05,
      "loss": 1.8328,
      "step": 179111
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.3922650814056396,
      "learning_rate": 7.042436042676896e-05,
      "loss": 2.1211,
      "step": 179112
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.3207855224609375,
      "learning_rate": 7.042172722778546e-05,
      "loss": 1.982,
      "step": 179113
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.3577789068222046,
      "learning_rate": 7.041909407148502e-05,
      "loss": 2.0811,
      "step": 179114
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.3657307624816895,
      "learning_rate": 7.041646095786801e-05,
      "loss": 1.9806,
      "step": 179115
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.2297322750091553,
      "learning_rate": 7.041382788693482e-05,
      "loss": 1.8412,
      "step": 179116
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.3491113185882568,
      "learning_rate": 7.04111948586861e-05,
      "loss": 1.9795,
      "step": 179117
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.3356127738952637,
      "learning_rate": 7.040856187312228e-05,
      "loss": 2.0345,
      "step": 179118
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.4958463907241821,
      "learning_rate": 7.040592893024376e-05,
      "loss": 1.9698,
      "step": 179119
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.3168526887893677,
      "learning_rate": 7.040329603005123e-05,
      "loss": 2.139,
      "step": 179120
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.2641263008117676,
      "learning_rate": 7.040066317254495e-05,
      "loss": 1.8174,
      "step": 179121
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.2722817659378052,
      "learning_rate": 7.039803035772558e-05,
      "loss": 1.9862,
      "step": 179122
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.6556262969970703,
      "learning_rate": 7.039539758559358e-05,
      "loss": 2.0799,
      "step": 179123
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.3605432510375977,
      "learning_rate": 7.03927648561494e-05,
      "loss": 1.9139,
      "step": 179124
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.3791627883911133,
      "learning_rate": 7.039013216939347e-05,
      "loss": 2.0751,
      "step": 179125
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.4531031847000122,
      "learning_rate": 7.038749952532642e-05,
      "loss": 2.1818,
      "step": 179126
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.4244104623794556,
      "learning_rate": 7.038486692394856e-05,
      "loss": 1.9579,
      "step": 179127
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.4223301410675049,
      "learning_rate": 7.038223436526062e-05,
      "loss": 2.204,
      "step": 179128
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.3332382440567017,
      "learning_rate": 7.037960184926293e-05,
      "loss": 1.9114,
      "step": 179129
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.4670798778533936,
      "learning_rate": 7.037696937595599e-05,
      "loss": 2.1745,
      "step": 179130
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.5162262916564941,
      "learning_rate": 7.037433694534026e-05,
      "loss": 1.9927,
      "step": 179131
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.3311505317687988,
      "learning_rate": 7.037170455741632e-05,
      "loss": 2.0872,
      "step": 179132
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.3035320043563843,
      "learning_rate": 7.036907221218456e-05,
      "loss": 2.1308,
      "step": 179133
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.4102370738983154,
      "learning_rate": 7.036643990964559e-05,
      "loss": 1.931,
      "step": 179134
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.2552804946899414,
      "learning_rate": 7.036380764979976e-05,
      "loss": 1.935,
      "step": 179135
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.4223535060882568,
      "learning_rate": 7.036117543264779e-05,
      "loss": 2.0598,
      "step": 179136
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.2519865036010742,
      "learning_rate": 7.035854325818986e-05,
      "loss": 1.8744,
      "step": 179137
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.3967176675796509,
      "learning_rate": 7.03559111264267e-05,
      "loss": 2.099,
      "step": 179138
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.249025821685791,
      "learning_rate": 7.035327903735862e-05,
      "loss": 1.9435,
      "step": 179139
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.4513039588928223,
      "learning_rate": 7.035064699098626e-05,
      "loss": 2.0976,
      "step": 179140
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.2681998014450073,
      "learning_rate": 7.034801498730999e-05,
      "loss": 2.0249,
      "step": 179141
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.388649582862854,
      "learning_rate": 7.034538302633052e-05,
      "loss": 1.9577,
      "step": 179142
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.2998420000076294,
      "learning_rate": 7.034275110804802e-05,
      "loss": 2.0221,
      "step": 179143
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.184411883354187,
      "learning_rate": 7.034011923246321e-05,
      "loss": 2.0274,
      "step": 179144
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.5065449476242065,
      "learning_rate": 7.033748739957642e-05,
      "loss": 1.9099,
      "step": 179145
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.3479135036468506,
      "learning_rate": 7.033485560938835e-05,
      "loss": 2.1553,
      "step": 179146
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.4710087776184082,
      "learning_rate": 7.033222386189924e-05,
      "loss": 2.4436,
      "step": 179147
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.4994590282440186,
      "learning_rate": 7.032959215710988e-05,
      "loss": 1.9827,
      "step": 179148
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.2348960638046265,
      "learning_rate": 7.03269604950204e-05,
      "loss": 2.1384,
      "step": 179149
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.2347056865692139,
      "learning_rate": 7.03243288756316e-05,
      "loss": 1.9418,
      "step": 179150
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.2324684858322144,
      "learning_rate": 7.032169729894373e-05,
      "loss": 2.0316,
      "step": 179151
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.273053765296936,
      "learning_rate": 7.031906576495749e-05,
      "loss": 2.0171,
      "step": 179152
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.3889859914779663,
      "learning_rate": 7.031643427367317e-05,
      "loss": 2.1088,
      "step": 179153
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.2669953107833862,
      "learning_rate": 7.031380282509154e-05,
      "loss": 1.9694,
      "step": 179154
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.4988007545471191,
      "learning_rate": 7.031117141921274e-05,
      "loss": 1.88,
      "step": 179155
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.2697360515594482,
      "learning_rate": 7.030854005603752e-05,
      "loss": 2.0236,
      "step": 179156
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.2941259145736694,
      "learning_rate": 7.030590873556618e-05,
      "loss": 1.8952,
      "step": 179157
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.5130888223648071,
      "learning_rate": 7.03032774577994e-05,
      "loss": 2.2096,
      "step": 179158
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.4088327884674072,
      "learning_rate": 7.030064622273747e-05,
      "loss": 2.0124,
      "step": 179159
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.3243463039398193,
      "learning_rate": 7.029801503038111e-05,
      "loss": 2.0179,
      "step": 179160
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.4963338375091553,
      "learning_rate": 7.029538388073065e-05,
      "loss": 2.1731,
      "step": 179161
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.3067445755004883,
      "learning_rate": 7.029275277378663e-05,
      "loss": 2.3286,
      "step": 179162
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.2397438287734985,
      "learning_rate": 7.029012170954945e-05,
      "loss": 2.1104,
      "step": 179163
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.3015081882476807,
      "learning_rate": 7.028749068801974e-05,
      "loss": 1.825,
      "step": 179164
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.3092539310455322,
      "learning_rate": 7.028485970919784e-05,
      "loss": 1.9125,
      "step": 179165
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.3589133024215698,
      "learning_rate": 7.028222877308441e-05,
      "loss": 2.2631,
      "step": 179166
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.321999430656433,
      "learning_rate": 7.027959787967984e-05,
      "loss": 2.2232,
      "step": 179167
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.6002613306045532,
      "learning_rate": 7.027696702898464e-05,
      "loss": 1.9215,
      "step": 179168
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.3549585342407227,
      "learning_rate": 7.027433622099919e-05,
      "loss": 2.2211,
      "step": 179169
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.3219101428985596,
      "learning_rate": 7.02717054557242e-05,
      "loss": 1.9942,
      "step": 179170
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.3604974746704102,
      "learning_rate": 7.02690747331599e-05,
      "loss": 2.0208,
      "step": 179171
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.3997581005096436,
      "learning_rate": 7.026644405330702e-05,
      "loss": 2.1458,
      "step": 179172
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.2512848377227783,
      "learning_rate": 7.026381341616595e-05,
      "loss": 2.0039,
      "step": 179173
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.2294286489486694,
      "learning_rate": 7.026118282173717e-05,
      "loss": 2.2046,
      "step": 179174
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.2498571872711182,
      "learning_rate": 7.025855227002109e-05,
      "loss": 1.9513,
      "step": 179175
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.3379976749420166,
      "learning_rate": 7.025592176101837e-05,
      "loss": 1.884,
      "step": 179176
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.1776938438415527,
      "learning_rate": 7.025329129472931e-05,
      "loss": 2.1287,
      "step": 179177
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.3651905059814453,
      "learning_rate": 7.025066087115462e-05,
      "loss": 1.9657,
      "step": 179178
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.4190402030944824,
      "learning_rate": 7.024803049029462e-05,
      "loss": 2.1011,
      "step": 179179
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.2953972816467285,
      "learning_rate": 7.02454001521499e-05,
      "loss": 2.0767,
      "step": 179180
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.340091347694397,
      "learning_rate": 7.024276985672077e-05,
      "loss": 1.9495,
      "step": 179181
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.3163996934890747,
      "learning_rate": 7.024013960400794e-05,
      "loss": 2.0091,
      "step": 179182
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.3125466108322144,
      "learning_rate": 7.023750939401174e-05,
      "loss": 1.9223,
      "step": 179183
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.4551727771759033,
      "learning_rate": 7.02348792267328e-05,
      "loss": 2.0536,
      "step": 179184
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.3313753604888916,
      "learning_rate": 7.023224910217152e-05,
      "loss": 2.0307,
      "step": 179185
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.366959810256958,
      "learning_rate": 7.022961902032841e-05,
      "loss": 2.1717,
      "step": 179186
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.3587182760238647,
      "learning_rate": 7.022698898120389e-05,
      "loss": 1.9741,
      "step": 179187
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.3332369327545166,
      "learning_rate": 7.022435898479856e-05,
      "loss": 2.1083,
      "step": 179188
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.3349683284759521,
      "learning_rate": 7.02217290311128e-05,
      "loss": 2.1827,
      "step": 179189
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.3774138689041138,
      "learning_rate": 7.02190991201472e-05,
      "loss": 2.0371,
      "step": 179190
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.3199715614318848,
      "learning_rate": 7.021646925190226e-05,
      "loss": 2.1315,
      "step": 179191
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.660805344581604,
      "learning_rate": 7.021383942637838e-05,
      "loss": 2.1595,
      "step": 179192
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.4274601936340332,
      "learning_rate": 7.021120964357603e-05,
      "loss": 2.0941,
      "step": 179193
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.303725242614746,
      "learning_rate": 7.02085799034958e-05,
      "loss": 2.0174,
      "step": 179194
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.4587864875793457,
      "learning_rate": 7.02059502061381e-05,
      "loss": 2.1509,
      "step": 179195
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.4712203741073608,
      "learning_rate": 7.020332055150348e-05,
      "loss": 1.897,
      "step": 179196
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.2864184379577637,
      "learning_rate": 7.020069093959243e-05,
      "loss": 1.9499,
      "step": 179197
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.212124228477478,
      "learning_rate": 7.019806137040533e-05,
      "loss": 1.8261,
      "step": 179198
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.378915786743164,
      "learning_rate": 7.019543184394282e-05,
      "loss": 2.2772,
      "step": 179199
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.4235297441482544,
      "learning_rate": 7.019280236020532e-05,
      "loss": 1.8488,
      "step": 179200
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.3758952617645264,
      "learning_rate": 7.019017291919322e-05,
      "loss": 2.1578,
      "step": 179201
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.8341255187988281,
      "learning_rate": 7.018754352090722e-05,
      "loss": 2.1473,
      "step": 179202
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.3540791273117065,
      "learning_rate": 7.018491416534767e-05,
      "loss": 2.0345,
      "step": 179203
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.3624894618988037,
      "learning_rate": 7.018228485251502e-05,
      "loss": 2.065,
      "step": 179204
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.4249705076217651,
      "learning_rate": 7.017965558240987e-05,
      "loss": 1.9687,
      "step": 179205
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.2318243980407715,
      "learning_rate": 7.017702635503259e-05,
      "loss": 1.9706,
      "step": 179206
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.337078332901001,
      "learning_rate": 7.017439717038383e-05,
      "loss": 2.067,
      "step": 179207
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.3271844387054443,
      "learning_rate": 7.017176802846399e-05,
      "loss": 1.869,
      "step": 179208
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.3860745429992676,
      "learning_rate": 7.016913892927353e-05,
      "loss": 1.9896,
      "step": 179209
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.2597018480300903,
      "learning_rate": 7.016650987281293e-05,
      "loss": 2.011,
      "step": 179210
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.2746376991271973,
      "learning_rate": 7.016388085908275e-05,
      "loss": 1.9248,
      "step": 179211
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.3595833778381348,
      "learning_rate": 7.01612518880834e-05,
      "loss": 2.0933,
      "step": 179212
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.21162748336792,
      "learning_rate": 7.015862295981548e-05,
      "loss": 2.1019,
      "step": 179213
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.2895393371582031,
      "learning_rate": 7.015599407427942e-05,
      "loss": 1.9281,
      "step": 179214
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.3156147003173828,
      "learning_rate": 7.015336523147568e-05,
      "loss": 1.9625,
      "step": 179215
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.4083017110824585,
      "learning_rate": 7.015073643140469e-05,
      "loss": 2.0232,
      "step": 179216
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.3010730743408203,
      "learning_rate": 7.014810767406711e-05,
      "loss": 1.809,
      "step": 179217
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.394793152809143,
      "learning_rate": 7.014547895946322e-05,
      "loss": 2.2619,
      "step": 179218
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.445648431777954,
      "learning_rate": 7.014285028759375e-05,
      "loss": 2.1541,
      "step": 179219
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.381850004196167,
      "learning_rate": 7.014022165845895e-05,
      "loss": 1.9227,
      "step": 179220
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.4454001188278198,
      "learning_rate": 7.013759307205961e-05,
      "loss": 1.8983,
      "step": 179221
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.6191074848175049,
      "learning_rate": 7.013496452839585e-05,
      "loss": 2.0777,
      "step": 179222
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.3899320363998413,
      "learning_rate": 7.013233602746842e-05,
      "loss": 2.0457,
      "step": 179223
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.2109479904174805,
      "learning_rate": 7.012970756927766e-05,
      "loss": 1.9004,
      "step": 179224
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.2344163656234741,
      "learning_rate": 7.01270791538242e-05,
      "loss": 1.9142,
      "step": 179225
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.1870713233947754,
      "learning_rate": 7.012445078110837e-05,
      "loss": 1.7052,
      "step": 179226
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.3904404640197754,
      "learning_rate": 7.012182245113084e-05,
      "loss": 2.0661,
      "step": 179227
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.2687851190567017,
      "learning_rate": 7.011919416389199e-05,
      "loss": 2.1222,
      "step": 179228
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.2949886322021484,
      "learning_rate": 7.011656591939233e-05,
      "loss": 1.9304,
      "step": 179229
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.3838920593261719,
      "learning_rate": 7.011393771763223e-05,
      "loss": 1.8811,
      "step": 179230
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.3158859014511108,
      "learning_rate": 7.011130955861238e-05,
      "loss": 2.0999,
      "step": 179231
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.329515814781189,
      "learning_rate": 7.010868144233312e-05,
      "loss": 1.9871,
      "step": 179232
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.3465505838394165,
      "learning_rate": 7.010605336879506e-05,
      "loss": 1.9201,
      "step": 179233
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.2372080087661743,
      "learning_rate": 7.010342533799864e-05,
      "loss": 1.8493,
      "step": 179234
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.3074501752853394,
      "learning_rate": 7.01007973499443e-05,
      "loss": 2.2979,
      "step": 179235
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.218888759613037,
      "learning_rate": 7.009816940463251e-05,
      "loss": 1.9964,
      "step": 179236
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.4297038316726685,
      "learning_rate": 7.009554150206389e-05,
      "loss": 2.0031,
      "step": 179237
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.3384071588516235,
      "learning_rate": 7.009291364223875e-05,
      "loss": 2.1323,
      "step": 179238
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.38417387008667,
      "learning_rate": 7.009028582515777e-05,
      "loss": 2.0833,
      "step": 179239
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.3243845701217651,
      "learning_rate": 7.008765805082135e-05,
      "loss": 2.0397,
      "step": 179240
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.3558205366134644,
      "learning_rate": 7.008503031922996e-05,
      "loss": 1.9997,
      "step": 179241
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.3061343431472778,
      "learning_rate": 7.008240263038401e-05,
      "loss": 2.0091,
      "step": 179242
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.3777037858963013,
      "learning_rate": 7.00797749842842e-05,
      "loss": 1.9183,
      "step": 179243
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.2422536611557007,
      "learning_rate": 7.007714738093081e-05,
      "loss": 2.0983,
      "step": 179244
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.3156691789627075,
      "learning_rate": 7.007451982032449e-05,
      "loss": 2.0499,
      "step": 179245
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.2799642086029053,
      "learning_rate": 7.007189230246565e-05,
      "loss": 1.8249,
      "step": 179246
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.453019618988037,
      "learning_rate": 7.006926482735477e-05,
      "loss": 2.0435,
      "step": 179247
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.3675596714019775,
      "learning_rate": 7.006663739499233e-05,
      "loss": 2.2183,
      "step": 179248
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.2186652421951294,
      "learning_rate": 7.006401000537889e-05,
      "loss": 1.9376,
      "step": 179249
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.3200126886367798,
      "learning_rate": 7.006138265851478e-05,
      "loss": 2.0385,
      "step": 179250
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.3572109937667847,
      "learning_rate": 7.005875535440072e-05,
      "loss": 2.0647,
      "step": 179251
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.3789522647857666,
      "learning_rate": 7.005612809303705e-05,
      "loss": 2.2339,
      "step": 179252
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.33650541305542,
      "learning_rate": 7.005350087442434e-05,
      "loss": 1.9537,
      "step": 179253
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.3068995475769043,
      "learning_rate": 7.00508736985629e-05,
      "loss": 1.9995,
      "step": 179254
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.4370830059051514,
      "learning_rate": 7.004824656545342e-05,
      "loss": 2.1258,
      "step": 179255
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.3355555534362793,
      "learning_rate": 7.004561947509623e-05,
      "loss": 1.9391,
      "step": 179256
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.2952687740325928,
      "learning_rate": 7.004299242749201e-05,
      "loss": 2.1694,
      "step": 179257
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.2328040599822998,
      "learning_rate": 7.004036542264112e-05,
      "loss": 1.8474,
      "step": 179258
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.3908274173736572,
      "learning_rate": 7.003773846054406e-05,
      "loss": 2.0911,
      "step": 179259
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.4639880657196045,
      "learning_rate": 7.003511154120125e-05,
      "loss": 1.9148,
      "step": 179260
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.3769766092300415,
      "learning_rate": 7.003248466461337e-05,
      "loss": 2.0473,
      "step": 179261
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.3216475248336792,
      "learning_rate": 7.002985783078064e-05,
      "loss": 2.1133,
      "step": 179262
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.2814527750015259,
      "learning_rate": 7.002723103970382e-05,
      "loss": 1.9837,
      "step": 179263
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.2955511808395386,
      "learning_rate": 7.00246042913833e-05,
      "loss": 1.9767,
      "step": 179264
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.308271050453186,
      "learning_rate": 7.00219775858195e-05,
      "loss": 2.0411,
      "step": 179265
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.311547875404358,
      "learning_rate": 7.001935092301292e-05,
      "loss": 2.2245,
      "step": 179266
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.2996606826782227,
      "learning_rate": 7.001672430296415e-05,
      "loss": 2.2237,
      "step": 179267
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.3169575929641724,
      "learning_rate": 7.001409772567353e-05,
      "loss": 1.9677,
      "step": 179268
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.533176064491272,
      "learning_rate": 7.001147119114169e-05,
      "loss": 1.8881,
      "step": 179269
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.3956272602081299,
      "learning_rate": 7.000884469936908e-05,
      "loss": 2.1228,
      "step": 179270
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.3133875131607056,
      "learning_rate": 7.000621825035618e-05,
      "loss": 2.1293,
      "step": 179271
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.3796788454055786,
      "learning_rate": 7.000359184410336e-05,
      "loss": 1.9938,
      "step": 179272
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.291050672531128,
      "learning_rate": 7.00009654806113e-05,
      "loss": 2.051,
      "step": 179273
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.227317452430725,
      "learning_rate": 6.999833915988032e-05,
      "loss": 2.1135,
      "step": 179274
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.39450204372406,
      "learning_rate": 6.999571288191108e-05,
      "loss": 1.9092,
      "step": 179275
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.3072439432144165,
      "learning_rate": 6.999308664670398e-05,
      "loss": 2.3133,
      "step": 179276
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.4259591102600098,
      "learning_rate": 6.99904604542595e-05,
      "loss": 2.1399,
      "step": 179277
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.3033220767974854,
      "learning_rate": 6.998783430457805e-05,
      "loss": 1.9442,
      "step": 179278
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.3436461687088013,
      "learning_rate": 6.998520819766029e-05,
      "loss": 2.0843,
      "step": 179279
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.2810195684432983,
      "learning_rate": 6.998258213350656e-05,
      "loss": 2.0433,
      "step": 179280
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.36111581325531,
      "learning_rate": 6.997995611211746e-05,
      "loss": 1.9899,
      "step": 179281
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.2825020551681519,
      "learning_rate": 6.997733013349344e-05,
      "loss": 1.9843,
      "step": 179282
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.4758306741714478,
      "learning_rate": 6.997470419763487e-05,
      "loss": 1.87,
      "step": 179283
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.2971596717834473,
      "learning_rate": 6.997207830454248e-05,
      "loss": 2.0068,
      "step": 179284
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.7545889616012573,
      "learning_rate": 6.99694524542166e-05,
      "loss": 2.0356,
      "step": 179285
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.3066672086715698,
      "learning_rate": 6.996682664665765e-05,
      "loss": 1.9639,
      "step": 179286
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.300657033920288,
      "learning_rate": 6.996420088186629e-05,
      "loss": 2.0673,
      "step": 179287
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.4905322790145874,
      "learning_rate": 6.996157515984295e-05,
      "loss": 1.9256,
      "step": 179288
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.293483018875122,
      "learning_rate": 6.995894948058795e-05,
      "loss": 2.1586,
      "step": 179289
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.322028636932373,
      "learning_rate": 6.995632384410205e-05,
      "loss": 1.9814,
      "step": 179290
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.493097186088562,
      "learning_rate": 6.995369825038554e-05,
      "loss": 2.137,
      "step": 179291
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.3169587850570679,
      "learning_rate": 6.995107269943906e-05,
      "loss": 2.2017,
      "step": 179292
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.3470081090927124,
      "learning_rate": 6.994844719126303e-05,
      "loss": 1.9266,
      "step": 179293
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.4020912647247314,
      "learning_rate": 6.99458217258578e-05,
      "loss": 1.9943,
      "step": 179294
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.5043487548828125,
      "learning_rate": 6.994319630322411e-05,
      "loss": 2.0604,
      "step": 179295
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.2844974994659424,
      "learning_rate": 6.994057092336231e-05,
      "loss": 2.0669,
      "step": 179296
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.243057370185852,
      "learning_rate": 6.993794558627278e-05,
      "loss": 1.9247,
      "step": 179297
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.3278591632843018,
      "learning_rate": 6.993532029195625e-05,
      "loss": 2.0231,
      "step": 179298
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.35619056224823,
      "learning_rate": 6.993269504041309e-05,
      "loss": 1.9251,
      "step": 179299
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.3458640575408936,
      "learning_rate": 6.993006983164369e-05,
      "loss": 1.9934,
      "step": 179300
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.419280767440796,
      "learning_rate": 6.99274446656487e-05,
      "loss": 1.8191,
      "step": 179301
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.2370551824569702,
      "learning_rate": 6.992481954242857e-05,
      "loss": 2.0088,
      "step": 179302
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.4687546491622925,
      "learning_rate": 6.992219446198368e-05,
      "loss": 2.1261,
      "step": 179303
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.3389755487442017,
      "learning_rate": 6.991956942431465e-05,
      "loss": 2.2003,
      "step": 179304
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.3772846460342407,
      "learning_rate": 6.991694442942186e-05,
      "loss": 1.8386,
      "step": 179305
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.554847002029419,
      "learning_rate": 6.991431947730592e-05,
      "loss": 2.1982,
      "step": 179306
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.6373236179351807,
      "learning_rate": 6.991169456796723e-05,
      "loss": 1.8738,
      "step": 179307
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.307752251625061,
      "learning_rate": 6.990906970140633e-05,
      "loss": 2.1477,
      "step": 179308
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.8286100625991821,
      "learning_rate": 6.99064448776236e-05,
      "loss": 2.2391,
      "step": 179309
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.483609914779663,
      "learning_rate": 6.99038200966197e-05,
      "loss": 1.9565,
      "step": 179310
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.2877082824707031,
      "learning_rate": 6.990119535839489e-05,
      "loss": 1.8719,
      "step": 179311
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.5586801767349243,
      "learning_rate": 6.98985706629499e-05,
      "loss": 2.1606,
      "step": 179312
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.247637391090393,
      "learning_rate": 6.98959460102851e-05,
      "loss": 1.9934,
      "step": 179313
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.3756158351898193,
      "learning_rate": 6.9893321400401e-05,
      "loss": 2.0123,
      "step": 179314
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.696597933769226,
      "learning_rate": 6.989069683329798e-05,
      "loss": 2.0811,
      "step": 179315
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.3673635721206665,
      "learning_rate": 6.988807230897671e-05,
      "loss": 2.1407,
      "step": 179316
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.378574013710022,
      "learning_rate": 6.988544782743753e-05,
      "loss": 2.1173,
      "step": 179317
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.2901842594146729,
      "learning_rate": 6.988282338868104e-05,
      "loss": 2.1385,
      "step": 179318
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.4415916204452515,
      "learning_rate": 6.98801989927077e-05,
      "loss": 1.895,
      "step": 179319
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.401057243347168,
      "learning_rate": 6.987757463951797e-05,
      "loss": 2.0422,
      "step": 179320
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.853000283241272,
      "learning_rate": 6.987495032911225e-05,
      "loss": 2.0348,
      "step": 179321
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.4457284212112427,
      "learning_rate": 6.987232606149121e-05,
      "loss": 2.2659,
      "step": 179322
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.2614070177078247,
      "learning_rate": 6.986970183665518e-05,
      "loss": 2.0125,
      "step": 179323
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.338996171951294,
      "learning_rate": 6.986707765460479e-05,
      "loss": 2.0377,
      "step": 179324
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.36632239818573,
      "learning_rate": 6.986445351534043e-05,
      "loss": 2.0579,
      "step": 179325
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.4910465478897095,
      "learning_rate": 6.986182941886265e-05,
      "loss": 1.9051,
      "step": 179326
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.6019856929779053,
      "learning_rate": 6.985920536517179e-05,
      "loss": 2.0954,
      "step": 179327
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.3887518644332886,
      "learning_rate": 6.985658135426856e-05,
      "loss": 2.0211,
      "step": 179328
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.4418866634368896,
      "learning_rate": 6.985395738615321e-05,
      "loss": 2.0221,
      "step": 179329
    },
    {
      "epoch": 2.33,
      "grad_norm": 1.3524197340011597,
      "learning_rate": 6.985133346082645e-05,
      "loss": 1.8217,
      "step": 179330
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.4666963815689087,
      "learning_rate": 6.98487095782887e-05,
      "loss": 1.8692,
      "step": 179331
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.3150736093521118,
      "learning_rate": 6.984608573854041e-05,
      "loss": 2.0109,
      "step": 179332
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.3724008798599243,
      "learning_rate": 6.984346194158197e-05,
      "loss": 1.9162,
      "step": 179333
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.2373881340026855,
      "learning_rate": 6.984083818741407e-05,
      "loss": 1.9136,
      "step": 179334
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.2977505922317505,
      "learning_rate": 6.983821447603702e-05,
      "loss": 1.9679,
      "step": 179335
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.3502918481826782,
      "learning_rate": 6.983559080745149e-05,
      "loss": 2.1392,
      "step": 179336
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.3444362878799438,
      "learning_rate": 6.983296718165783e-05,
      "loss": 1.8637,
      "step": 179337
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.3459811210632324,
      "learning_rate": 6.983034359865662e-05,
      "loss": 2.0596,
      "step": 179338
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.3972969055175781,
      "learning_rate": 6.982772005844818e-05,
      "loss": 2.0974,
      "step": 179339
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.30553138256073,
      "learning_rate": 6.982509656103321e-05,
      "loss": 1.9931,
      "step": 179340
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.2534459829330444,
      "learning_rate": 6.9822473106412e-05,
      "loss": 1.9236,
      "step": 179341
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.426953673362732,
      "learning_rate": 6.981984969458519e-05,
      "loss": 2.1238,
      "step": 179342
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.2657856941223145,
      "learning_rate": 6.981722632555324e-05,
      "loss": 1.9844,
      "step": 179343
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.2455620765686035,
      "learning_rate": 6.981460299931664e-05,
      "loss": 2.1922,
      "step": 179344
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.3024922609329224,
      "learning_rate": 6.981197971587573e-05,
      "loss": 2.0441,
      "step": 179345
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.3534091711044312,
      "learning_rate": 6.980935647523123e-05,
      "loss": 2.041,
      "step": 179346
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.317966341972351,
      "learning_rate": 6.980673327738342e-05,
      "loss": 1.8356,
      "step": 179347
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.4391882419586182,
      "learning_rate": 6.980411012233298e-05,
      "loss": 1.9631,
      "step": 179348
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.518356442451477,
      "learning_rate": 6.980148701008027e-05,
      "loss": 1.9416,
      "step": 179349
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.4243179559707642,
      "learning_rate": 6.979886394062582e-05,
      "loss": 2.0823,
      "step": 179350
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.347918152809143,
      "learning_rate": 6.979624091397004e-05,
      "loss": 2.0685,
      "step": 179351
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.387911319732666,
      "learning_rate": 6.979361793011356e-05,
      "loss": 1.8766,
      "step": 179352
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.3212870359420776,
      "learning_rate": 6.979099498905672e-05,
      "loss": 1.939,
      "step": 179353
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.3196913003921509,
      "learning_rate": 6.978837209080016e-05,
      "loss": 2.0327,
      "step": 179354
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.3040026426315308,
      "learning_rate": 6.978574923534427e-05,
      "loss": 2.0015,
      "step": 179355
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.210103988647461,
      "learning_rate": 6.978312642268957e-05,
      "loss": 1.8981,
      "step": 179356
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.3259072303771973,
      "learning_rate": 6.978050365283644e-05,
      "loss": 1.8997,
      "step": 179357
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.2833973169326782,
      "learning_rate": 6.977788092578553e-05,
      "loss": 1.9228,
      "step": 179358
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.3255783319473267,
      "learning_rate": 6.977525824153719e-05,
      "loss": 1.9795,
      "step": 179359
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.4762916564941406,
      "learning_rate": 6.977263560009207e-05,
      "loss": 2.276,
      "step": 179360
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.2173000574111938,
      "learning_rate": 6.977001300145045e-05,
      "loss": 2.0585,
      "step": 179361
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.6804206371307373,
      "learning_rate": 6.976739044561312e-05,
      "loss": 2.1955,
      "step": 179362
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.3134804964065552,
      "learning_rate": 6.976476793258023e-05,
      "loss": 1.8733,
      "step": 179363
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.2991700172424316,
      "learning_rate": 6.976214546235245e-05,
      "loss": 2.1182,
      "step": 179364
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.4029289484024048,
      "learning_rate": 6.975952303493017e-05,
      "loss": 2.2219,
      "step": 179365
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.3562672138214111,
      "learning_rate": 6.975690065031404e-05,
      "loss": 2.0405,
      "step": 179366
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.337258219718933,
      "learning_rate": 6.975427830850434e-05,
      "loss": 2.1452,
      "step": 179367
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.3232381343841553,
      "learning_rate": 6.975165600950177e-05,
      "loss": 2.1601,
      "step": 179368
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.3812440633773804,
      "learning_rate": 6.974903375330672e-05,
      "loss": 2.0788,
      "step": 179369
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.2024030685424805,
      "learning_rate": 6.974641153991966e-05,
      "loss": 2.06,
      "step": 179370
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.540435791015625,
      "learning_rate": 6.974378936934098e-05,
      "loss": 2.0779,
      "step": 179371
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.3951122760772705,
      "learning_rate": 6.97411672415714e-05,
      "loss": 2.1334,
      "step": 179372
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.4358375072479248,
      "learning_rate": 6.973854515661118e-05,
      "loss": 1.9083,
      "step": 179373
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.2785158157348633,
      "learning_rate": 6.9735923114461e-05,
      "loss": 2.0773,
      "step": 179374
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.3463069200515747,
      "learning_rate": 6.973330111512126e-05,
      "loss": 2.1443,
      "step": 179375
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.5837790966033936,
      "learning_rate": 6.973067915859244e-05,
      "loss": 1.7741,
      "step": 179376
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.5452736616134644,
      "learning_rate": 6.972805724487493e-05,
      "loss": 2.1361,
      "step": 179377
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.3486026525497437,
      "learning_rate": 6.972543537396945e-05,
      "loss": 2.0612,
      "step": 179378
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.4934964179992676,
      "learning_rate": 6.972281354587625e-05,
      "loss": 1.8545,
      "step": 179379
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.4616453647613525,
      "learning_rate": 6.972019176059601e-05,
      "loss": 1.9909,
      "step": 179380
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.3711382150650024,
      "learning_rate": 6.971757001812912e-05,
      "loss": 1.9778,
      "step": 179381
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.325343370437622,
      "learning_rate": 6.971494831847601e-05,
      "loss": 2.0511,
      "step": 179382
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.3077882528305054,
      "learning_rate": 6.971232666163733e-05,
      "loss": 2.1325,
      "step": 179383
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.2972251176834106,
      "learning_rate": 6.970970504761346e-05,
      "loss": 2.0776,
      "step": 179384
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.454533338546753,
      "learning_rate": 6.970708347640481e-05,
      "loss": 1.9339,
      "step": 179385
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.2757552862167358,
      "learning_rate": 6.970446194801208e-05,
      "loss": 1.9907,
      "step": 179386
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.3157010078430176,
      "learning_rate": 6.970184046243564e-05,
      "loss": 2.0914,
      "step": 179387
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.4032363891601562,
      "learning_rate": 6.969921901967583e-05,
      "loss": 2.1289,
      "step": 179388
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.3137367963790894,
      "learning_rate": 6.969659761973342e-05,
      "loss": 2.1133,
      "step": 179389
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.2798577547073364,
      "learning_rate": 6.969397626260865e-05,
      "loss": 2.1108,
      "step": 179390
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.3769481182098389,
      "learning_rate": 6.969135494830223e-05,
      "loss": 1.7159,
      "step": 179391
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.2318025827407837,
      "learning_rate": 6.96887336768145e-05,
      "loss": 2.0621,
      "step": 179392
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.3147168159484863,
      "learning_rate": 6.968611244814602e-05,
      "loss": 2.2589,
      "step": 179393
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.3038280010223389,
      "learning_rate": 6.968349126229711e-05,
      "loss": 1.9513,
      "step": 179394
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.2402348518371582,
      "learning_rate": 6.968087011926852e-05,
      "loss": 1.8735,
      "step": 179395
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.3638365268707275,
      "learning_rate": 6.967824901906048e-05,
      "loss": 2.0725,
      "step": 179396
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.32875657081604,
      "learning_rate": 6.967562796167371e-05,
      "loss": 1.8361,
      "step": 179397
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.243914246559143,
      "learning_rate": 6.967300694710859e-05,
      "loss": 2.0791,
      "step": 179398
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.300691843032837,
      "learning_rate": 6.967038597536559e-05,
      "loss": 1.8876,
      "step": 179399
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.4701677560806274,
      "learning_rate": 6.966776504644512e-05,
      "loss": 2.0072,
      "step": 179400
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.201422095298767,
      "learning_rate": 6.966514416034783e-05,
      "loss": 1.8201,
      "step": 179401
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.221970558166504,
      "learning_rate": 6.966252331707411e-05,
      "loss": 1.9814,
      "step": 179402
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.3000420331954956,
      "learning_rate": 6.965990251662455e-05,
      "loss": 2.0953,
      "step": 179403
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.321811556816101,
      "learning_rate": 6.965728175899953e-05,
      "loss": 2.0019,
      "step": 179404
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.3121565580368042,
      "learning_rate": 6.965466104419958e-05,
      "loss": 2.0018,
      "step": 179405
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.483526349067688,
      "learning_rate": 6.96520403722251e-05,
      "loss": 2.2717,
      "step": 179406
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.593668818473816,
      "learning_rate": 6.96494197430767e-05,
      "loss": 2.0901,
      "step": 179407
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.2808517217636108,
      "learning_rate": 6.96467991567548e-05,
      "loss": 2.014,
      "step": 179408
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.2873561382293701,
      "learning_rate": 6.964417861325996e-05,
      "loss": 2.0427,
      "step": 179409
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.5253338813781738,
      "learning_rate": 6.96415581125926e-05,
      "loss": 2.0655,
      "step": 179410
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.2834595441818237,
      "learning_rate": 6.963893765475325e-05,
      "loss": 2.0638,
      "step": 179411
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.4566597938537598,
      "learning_rate": 6.963631723974227e-05,
      "loss": 2.1735,
      "step": 179412
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.3354021310806274,
      "learning_rate": 6.963369686756032e-05,
      "loss": 2.0785,
      "step": 179413
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.2507320642471313,
      "learning_rate": 6.963107653820776e-05,
      "loss": 1.8042,
      "step": 179414
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.399922490119934,
      "learning_rate": 6.96284562516852e-05,
      "loss": 2.1389,
      "step": 179415
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.3161723613739014,
      "learning_rate": 6.962583600799306e-05,
      "loss": 2.04,
      "step": 179416
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.2257565259933472,
      "learning_rate": 6.962321580713184e-05,
      "loss": 2.0076,
      "step": 179417
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.3459707498550415,
      "learning_rate": 6.96205956491019e-05,
      "loss": 2.0352,
      "step": 179418
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.3390594720840454,
      "learning_rate": 6.961797553390396e-05,
      "loss": 2.0084,
      "step": 179419
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.4513827562332153,
      "learning_rate": 6.961535546153826e-05,
      "loss": 1.915,
      "step": 179420
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.2891182899475098,
      "learning_rate": 6.961273543200553e-05,
      "loss": 2.1299,
      "step": 179421
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.7216627597808838,
      "learning_rate": 6.961011544530614e-05,
      "loss": 1.8479,
      "step": 179422
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.3168164491653442,
      "learning_rate": 6.960749550144055e-05,
      "loss": 1.9817,
      "step": 179423
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.3008581399917603,
      "learning_rate": 6.960487560040923e-05,
      "loss": 1.9703,
      "step": 179424
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.1523679494857788,
      "learning_rate": 6.960225574221278e-05,
      "loss": 1.716,
      "step": 179425
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.30255925655365,
      "learning_rate": 6.959963592685152e-05,
      "loss": 1.9468,
      "step": 179426
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.252308964729309,
      "learning_rate": 6.959701615432615e-05,
      "loss": 1.859,
      "step": 179427
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.2337672710418701,
      "learning_rate": 6.959439642463693e-05,
      "loss": 1.8618,
      "step": 179428
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.2532739639282227,
      "learning_rate": 6.959177673778465e-05,
      "loss": 2.0278,
      "step": 179429
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.2276912927627563,
      "learning_rate": 6.958915709376944e-05,
      "loss": 1.981,
      "step": 179430
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.5162534713745117,
      "learning_rate": 6.958653749259203e-05,
      "loss": 2.2177,
      "step": 179431
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.413859486579895,
      "learning_rate": 6.958391793425275e-05,
      "loss": 1.9088,
      "step": 179432
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.3298817873001099,
      "learning_rate": 6.958129841875224e-05,
      "loss": 1.7394,
      "step": 179433
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.3444366455078125,
      "learning_rate": 6.957867894609085e-05,
      "loss": 2.185,
      "step": 179434
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.3023067712783813,
      "learning_rate": 6.957605951626931e-05,
      "loss": 1.9642,
      "step": 179435
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.2819578647613525,
      "learning_rate": 6.957344012928776e-05,
      "loss": 2.1178,
      "step": 179436
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.402151346206665,
      "learning_rate": 6.957082078514691e-05,
      "loss": 2.1276,
      "step": 179437
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.3572478294372559,
      "learning_rate": 6.956820148384715e-05,
      "loss": 1.7646,
      "step": 179438
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.3656749725341797,
      "learning_rate": 6.956558222538906e-05,
      "loss": 1.9007,
      "step": 179439
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.3168213367462158,
      "learning_rate": 6.956296300977304e-05,
      "loss": 1.8599,
      "step": 179440
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.5193828344345093,
      "learning_rate": 6.956034383699973e-05,
      "loss": 2.0032,
      "step": 179441
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.2832844257354736,
      "learning_rate": 6.955772470706936e-05,
      "loss": 1.9036,
      "step": 179442
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.4953522682189941,
      "learning_rate": 6.955510561998266e-05,
      "loss": 1.9909,
      "step": 179443
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.4920541048049927,
      "learning_rate": 6.955248657573988e-05,
      "loss": 2.1348,
      "step": 179444
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.5491766929626465,
      "learning_rate": 6.954986757434176e-05,
      "loss": 2.2684,
      "step": 179445
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.3208749294281006,
      "learning_rate": 6.95472486157886e-05,
      "loss": 1.9195,
      "step": 179446
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.3637545108795166,
      "learning_rate": 6.95446297000811e-05,
      "loss": 1.8601,
      "step": 179447
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.2954611778259277,
      "learning_rate": 6.954201082721943e-05,
      "loss": 2.2292,
      "step": 179448
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.337887167930603,
      "learning_rate": 6.953939199720435e-05,
      "loss": 1.8619,
      "step": 179449
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.4257863759994507,
      "learning_rate": 6.953677321003617e-05,
      "loss": 2.0945,
      "step": 179450
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.405564546585083,
      "learning_rate": 6.953415446571554e-05,
      "loss": 1.8236,
      "step": 179451
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.4313570261001587,
      "learning_rate": 6.953153576424278e-05,
      "loss": 1.8799,
      "step": 179452
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.257063865661621,
      "learning_rate": 6.95289171056185e-05,
      "loss": 1.8628,
      "step": 179453
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.413693904876709,
      "learning_rate": 6.952629848984318e-05,
      "loss": 2.0691,
      "step": 179454
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.4532915353775024,
      "learning_rate": 6.952367991691729e-05,
      "loss": 2.0707,
      "step": 179455
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.3623625040054321,
      "learning_rate": 6.952106138684118e-05,
      "loss": 2.1465,
      "step": 179456
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.381453037261963,
      "learning_rate": 6.951844289961554e-05,
      "loss": 1.9978,
      "step": 179457
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.3681215047836304,
      "learning_rate": 6.95158244552407e-05,
      "loss": 2.0559,
      "step": 179458
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.397275447845459,
      "learning_rate": 6.951320605371731e-05,
      "loss": 2.0679,
      "step": 179459
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.4209553003311157,
      "learning_rate": 6.951058769504576e-05,
      "loss": 1.9755,
      "step": 179460
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.3313120603561401,
      "learning_rate": 6.950796937922653e-05,
      "loss": 2.1406,
      "step": 179461
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.381724238395691,
      "learning_rate": 6.950535110626003e-05,
      "loss": 2.1438,
      "step": 179462
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.5176128149032593,
      "learning_rate": 6.950273287614695e-05,
      "loss": 2.069,
      "step": 179463
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.3738787174224854,
      "learning_rate": 6.950011468888756e-05,
      "loss": 2.1299,
      "step": 179464
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.2402409315109253,
      "learning_rate": 6.949749654448255e-05,
      "loss": 1.973,
      "step": 179465
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.32049560546875,
      "learning_rate": 6.94948784429323e-05,
      "loss": 1.9717,
      "step": 179466
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.6465785503387451,
      "learning_rate": 6.949226038423718e-05,
      "loss": 1.892,
      "step": 179467
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.2407058477401733,
      "learning_rate": 6.948964236839793e-05,
      "loss": 2.0034,
      "step": 179468
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.3435429334640503,
      "learning_rate": 6.948702439541491e-05,
      "loss": 2.032,
      "step": 179469
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.2923431396484375,
      "learning_rate": 6.948440646528848e-05,
      "loss": 1.9909,
      "step": 179470
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.344683289527893,
      "learning_rate": 6.948178857801937e-05,
      "loss": 2.1481,
      "step": 179471
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.3613741397857666,
      "learning_rate": 6.947917073360792e-05,
      "loss": 2.1766,
      "step": 179472
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.451889991760254,
      "learning_rate": 6.947655293205459e-05,
      "loss": 2.1858,
      "step": 179473
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.3684260845184326,
      "learning_rate": 6.947393517335998e-05,
      "loss": 2.0833,
      "step": 179474
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.2478187084197998,
      "learning_rate": 6.947131745752451e-05,
      "loss": 2.024,
      "step": 179475
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.4914880990982056,
      "learning_rate": 6.946869978454863e-05,
      "loss": 2.0318,
      "step": 179476
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.3046492338180542,
      "learning_rate": 6.946608215443293e-05,
      "loss": 2.4269,
      "step": 179477
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.3055157661437988,
      "learning_rate": 6.946346456717781e-05,
      "loss": 2.0851,
      "step": 179478
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.2091420888900757,
      "learning_rate": 6.946084702278376e-05,
      "loss": 1.97,
      "step": 179479
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.495123267173767,
      "learning_rate": 6.945822952125133e-05,
      "loss": 2.2155,
      "step": 179480
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.301991581916809,
      "learning_rate": 6.94556120625809e-05,
      "loss": 1.9981,
      "step": 179481
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.2976250648498535,
      "learning_rate": 6.945299464677309e-05,
      "loss": 2.1806,
      "step": 179482
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.4273380041122437,
      "learning_rate": 6.945037727382832e-05,
      "loss": 1.9074,
      "step": 179483
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.506995677947998,
      "learning_rate": 6.944775994374708e-05,
      "loss": 1.8755,
      "step": 179484
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.2612464427947998,
      "learning_rate": 6.94451426565298e-05,
      "loss": 1.868,
      "step": 179485
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.229069709777832,
      "learning_rate": 6.944252541217705e-05,
      "loss": 1.9862,
      "step": 179486
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.248602032661438,
      "learning_rate": 6.943990821068926e-05,
      "loss": 2.0423,
      "step": 179487
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.3079769611358643,
      "learning_rate": 6.9437291052067e-05,
      "loss": 1.9254,
      "step": 179488
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.4132087230682373,
      "learning_rate": 6.943467393631069e-05,
      "loss": 2.0043,
      "step": 179489
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.3257685899734497,
      "learning_rate": 6.943205686342083e-05,
      "loss": 2.109,
      "step": 179490
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.2688475847244263,
      "learning_rate": 6.942943983339784e-05,
      "loss": 2.2908,
      "step": 179491
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.282737135887146,
      "learning_rate": 6.942682284624232e-05,
      "loss": 1.8505,
      "step": 179492
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.352721929550171,
      "learning_rate": 6.942420590195466e-05,
      "loss": 1.79,
      "step": 179493
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.308951497077942,
      "learning_rate": 6.942158900053546e-05,
      "loss": 1.8984,
      "step": 179494
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.2318778038024902,
      "learning_rate": 6.941897214198504e-05,
      "loss": 2.1299,
      "step": 179495
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.4687086343765259,
      "learning_rate": 6.941635532630416e-05,
      "loss": 2.0423,
      "step": 179496
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.390872597694397,
      "learning_rate": 6.941373855349299e-05,
      "loss": 2.2395,
      "step": 179497
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.1849600076675415,
      "learning_rate": 6.941112182355223e-05,
      "loss": 2.042,
      "step": 179498
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.2691547870635986,
      "learning_rate": 6.94085051364822e-05,
      "loss": 1.9271,
      "step": 179499
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.4087682962417603,
      "learning_rate": 6.940588849228356e-05,
      "loss": 2.1182,
      "step": 179500
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.2861573696136475,
      "learning_rate": 6.940327189095665e-05,
      "loss": 1.9346,
      "step": 179501
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.2371325492858887,
      "learning_rate": 6.940065533250218e-05,
      "loss": 1.909,
      "step": 179502
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.3562161922454834,
      "learning_rate": 6.939803881692032e-05,
      "loss": 2.0602,
      "step": 179503
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.363263487815857,
      "learning_rate": 6.939542234421182e-05,
      "loss": 1.9916,
      "step": 179504
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.3018194437026978,
      "learning_rate": 6.939280591437696e-05,
      "loss": 2.0174,
      "step": 179505
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.3065215349197388,
      "learning_rate": 6.939018952741641e-05,
      "loss": 1.8829,
      "step": 179506
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.431992769241333,
      "learning_rate": 6.938757318333049e-05,
      "loss": 2.1255,
      "step": 179507
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.2833857536315918,
      "learning_rate": 6.938495688211995e-05,
      "loss": 2.0696,
      "step": 179508
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.288967490196228,
      "learning_rate": 6.938234062378494e-05,
      "loss": 1.9944,
      "step": 179509
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.2881654500961304,
      "learning_rate": 6.937972440832616e-05,
      "loss": 2.0331,
      "step": 179510
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.3015153408050537,
      "learning_rate": 6.9377108235744e-05,
      "loss": 1.9658,
      "step": 179511
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.3136645555496216,
      "learning_rate": 6.937449210603905e-05,
      "loss": 2.0558,
      "step": 179512
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.2779300212860107,
      "learning_rate": 6.937187601921166e-05,
      "loss": 2.185,
      "step": 179513
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.3118231296539307,
      "learning_rate": 6.936925997526256e-05,
      "loss": 2.0611,
      "step": 179514
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.1823734045028687,
      "learning_rate": 6.93666439741919e-05,
      "loss": 1.945,
      "step": 179515
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.2933841943740845,
      "learning_rate": 6.936402801600039e-05,
      "loss": 1.9682,
      "step": 179516
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.5688141584396362,
      "learning_rate": 6.93614121006884e-05,
      "loss": 1.9741,
      "step": 179517
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.3106440305709839,
      "learning_rate": 6.935879622825657e-05,
      "loss": 2.0842,
      "step": 179518
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.3432228565216064,
      "learning_rate": 6.93561803987052e-05,
      "loss": 1.9387,
      "step": 179519
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.2308393716812134,
      "learning_rate": 6.935356461203503e-05,
      "loss": 2.1163,
      "step": 179520
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.3789055347442627,
      "learning_rate": 6.935094886824623e-05,
      "loss": 2.1689,
      "step": 179521
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.5602149963378906,
      "learning_rate": 6.934833316733952e-05,
      "loss": 2.0585,
      "step": 179522
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.52245032787323,
      "learning_rate": 6.934571750931521e-05,
      "loss": 1.9536,
      "step": 179523
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.3029953241348267,
      "learning_rate": 6.9343101894174e-05,
      "loss": 2.1785,
      "step": 179524
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.2773531675338745,
      "learning_rate": 6.934048632191615e-05,
      "loss": 1.8226,
      "step": 179525
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.545393466949463,
      "learning_rate": 6.933787079254247e-05,
      "loss": 2.0778,
      "step": 179526
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.2982310056686401,
      "learning_rate": 6.933525530605302e-05,
      "loss": 1.9326,
      "step": 179527
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.4662939310073853,
      "learning_rate": 6.93326398624486e-05,
      "loss": 1.8248,
      "step": 179528
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.3206675052642822,
      "learning_rate": 6.933002446172953e-05,
      "loss": 1.95,
      "step": 179529
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.4026488065719604,
      "learning_rate": 6.932740910389643e-05,
      "loss": 2.0855,
      "step": 179530
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.238945484161377,
      "learning_rate": 6.932479378894961e-05,
      "loss": 1.9206,
      "step": 179531
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.3905391693115234,
      "learning_rate": 6.932217851688987e-05,
      "loss": 1.9305,
      "step": 179532
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.2211796045303345,
      "learning_rate": 6.931956328771731e-05,
      "loss": 1.9729,
      "step": 179533
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.4910451173782349,
      "learning_rate": 6.931694810143269e-05,
      "loss": 2.187,
      "step": 179534
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.3991159200668335,
      "learning_rate": 6.931433295803629e-05,
      "loss": 2.0251,
      "step": 179535
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.338836431503296,
      "learning_rate": 6.931171785752882e-05,
      "loss": 1.985,
      "step": 179536
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.266837477684021,
      "learning_rate": 6.930910279991056e-05,
      "loss": 2.0198,
      "step": 179537
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.3466042280197144,
      "learning_rate": 6.930648778518227e-05,
      "loss": 2.1229,
      "step": 179538
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.3781547546386719,
      "learning_rate": 6.930387281334411e-05,
      "loss": 2.076,
      "step": 179539
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.2920836210250854,
      "learning_rate": 6.930125788439677e-05,
      "loss": 2.0281,
      "step": 179540
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.2876280546188354,
      "learning_rate": 6.929864299834062e-05,
      "loss": 2.0224,
      "step": 179541
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.3549423217773438,
      "learning_rate": 6.929602815517625e-05,
      "loss": 2.1208,
      "step": 179542
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.251777172088623,
      "learning_rate": 6.929341335490405e-05,
      "loss": 2.0912,
      "step": 179543
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.3737754821777344,
      "learning_rate": 6.929079859752464e-05,
      "loss": 2.101,
      "step": 179544
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.257893681526184,
      "learning_rate": 6.928818388303841e-05,
      "loss": 1.8855,
      "step": 179545
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.3786876201629639,
      "learning_rate": 6.92855692114459e-05,
      "loss": 2.081,
      "step": 179546
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.3124958276748657,
      "learning_rate": 6.928295458274745e-05,
      "loss": 2.0189,
      "step": 179547
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.3023040294647217,
      "learning_rate": 6.928033999694373e-05,
      "loss": 1.9035,
      "step": 179548
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.2280042171478271,
      "learning_rate": 6.927772545403508e-05,
      "loss": 1.8397,
      "step": 179549
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.3591011762619019,
      "learning_rate": 6.927511095402212e-05,
      "loss": 1.8718,
      "step": 179550
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.257840633392334,
      "learning_rate": 6.92724964969053e-05,
      "loss": 1.8811,
      "step": 179551
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.3129191398620605,
      "learning_rate": 6.926988208268497e-05,
      "loss": 2.2286,
      "step": 179552
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.742307424545288,
      "learning_rate": 6.926726771136181e-05,
      "loss": 1.9007,
      "step": 179553
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.4845679998397827,
      "learning_rate": 6.926465338293624e-05,
      "loss": 2.1686,
      "step": 179554
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.3762632608413696,
      "learning_rate": 6.926203909740862e-05,
      "loss": 1.8991,
      "step": 179555
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.4020190238952637,
      "learning_rate": 6.925942485477965e-05,
      "loss": 2.1903,
      "step": 179556
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.2836840152740479,
      "learning_rate": 6.925681065504968e-05,
      "loss": 2.0698,
      "step": 179557
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.6530048847198486,
      "learning_rate": 6.925419649821916e-05,
      "loss": 2.0052,
      "step": 179558
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.3799946308135986,
      "learning_rate": 6.925158238428873e-05,
      "loss": 2.1516,
      "step": 179559
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.279737949371338,
      "learning_rate": 6.924896831325876e-05,
      "loss": 2.0859,
      "step": 179560
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.4133172035217285,
      "learning_rate": 6.92463542851297e-05,
      "loss": 2.0485,
      "step": 179561
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.3235927820205688,
      "learning_rate": 6.924374029990219e-05,
      "loss": 2.2308,
      "step": 179562
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.3389501571655273,
      "learning_rate": 6.92411263575766e-05,
      "loss": 1.9483,
      "step": 179563
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.4428644180297852,
      "learning_rate": 6.923851245815336e-05,
      "loss": 1.8461,
      "step": 179564
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.4146817922592163,
      "learning_rate": 6.923589860163314e-05,
      "loss": 2.2235,
      "step": 179565
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.609031319618225,
      "learning_rate": 6.923328478801621e-05,
      "loss": 2.3216,
      "step": 179566
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.2650848627090454,
      "learning_rate": 6.923067101730329e-05,
      "loss": 2.0334,
      "step": 179567
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.1632760763168335,
      "learning_rate": 6.922805728949473e-05,
      "loss": 2.003,
      "step": 179568
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.2624146938323975,
      "learning_rate": 6.922544360459102e-05,
      "loss": 1.8182,
      "step": 179569
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.4197994470596313,
      "learning_rate": 6.922282996259256e-05,
      "loss": 2.1251,
      "step": 179570
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.4108195304870605,
      "learning_rate": 6.922021636350005e-05,
      "loss": 2.0353,
      "step": 179571
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.2621926069259644,
      "learning_rate": 6.921760280731375e-05,
      "loss": 2.1819,
      "step": 179572
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.263289451599121,
      "learning_rate": 6.921498929403435e-05,
      "loss": 1.9971,
      "step": 179573
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.2316017150878906,
      "learning_rate": 6.921237582366223e-05,
      "loss": 2.1812,
      "step": 179574
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.3277201652526855,
      "learning_rate": 6.92097623961979e-05,
      "loss": 1.9659,
      "step": 179575
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.4066082239151,
      "learning_rate": 6.920714901164173e-05,
      "loss": 2.086,
      "step": 179576
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.3312345743179321,
      "learning_rate": 6.920453566999442e-05,
      "loss": 2.0241,
      "step": 179577
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.4022819995880127,
      "learning_rate": 6.920192237125621e-05,
      "loss": 2.1803,
      "step": 179578
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.1819437742233276,
      "learning_rate": 6.919930911542785e-05,
      "loss": 1.9453,
      "step": 179579
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.4357883930206299,
      "learning_rate": 6.919669590250959e-05,
      "loss": 1.9776,
      "step": 179580
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.3160637617111206,
      "learning_rate": 6.91940827325022e-05,
      "loss": 1.905,
      "step": 179581
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.2826508283615112,
      "learning_rate": 6.91914696054058e-05,
      "loss": 1.7789,
      "step": 179582
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.2674944400787354,
      "learning_rate": 6.918885652122117e-05,
      "loss": 2.0365,
      "step": 179583
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.2372782230377197,
      "learning_rate": 6.918624347994862e-05,
      "loss": 1.877,
      "step": 179584
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.268042802810669,
      "learning_rate": 6.918363048158872e-05,
      "loss": 2.2883,
      "step": 179585
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.2420356273651123,
      "learning_rate": 6.918101752614192e-05,
      "loss": 1.9205,
      "step": 179586
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.4372063875198364,
      "learning_rate": 6.917840461360887e-05,
      "loss": 1.9027,
      "step": 179587
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.6973594427108765,
      "learning_rate": 6.917579174398973e-05,
      "loss": 1.9277,
      "step": 179588
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.3555189371109009,
      "learning_rate": 6.917317891728528e-05,
      "loss": 1.8797,
      "step": 179589
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.4468433856964111,
      "learning_rate": 6.917056613349579e-05,
      "loss": 1.9466,
      "step": 179590
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.4238922595977783,
      "learning_rate": 6.916795339262193e-05,
      "loss": 1.8362,
      "step": 179591
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.4118313789367676,
      "learning_rate": 6.916534069466406e-05,
      "loss": 1.941,
      "step": 179592
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.5206714868545532,
      "learning_rate": 6.916272803962284e-05,
      "loss": 1.9138,
      "step": 179593
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.7183845043182373,
      "learning_rate": 6.916011542749846e-05,
      "loss": 1.9536,
      "step": 179594
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.2464046478271484,
      "learning_rate": 6.915750285829166e-05,
      "loss": 1.7138,
      "step": 179595
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.4005388021469116,
      "learning_rate": 6.915489033200275e-05,
      "loss": 2.1186,
      "step": 179596
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.4791423082351685,
      "learning_rate": 6.915227784863239e-05,
      "loss": 1.9445,
      "step": 179597
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.5736043453216553,
      "learning_rate": 6.914966540818092e-05,
      "loss": 2.0939,
      "step": 179598
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.7287262678146362,
      "learning_rate": 6.914705301064902e-05,
      "loss": 1.8762,
      "step": 179599
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.5184807777404785,
      "learning_rate": 6.914444065603689e-05,
      "loss": 2.0017,
      "step": 179600
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.3383257389068604,
      "learning_rate": 6.914182834434523e-05,
      "loss": 2.2021,
      "step": 179601
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.4742541313171387,
      "learning_rate": 6.91392160755744e-05,
      "loss": 1.9618,
      "step": 179602
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.7511540651321411,
      "learning_rate": 6.913660384972503e-05,
      "loss": 1.9782,
      "step": 179603
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.3354880809783936,
      "learning_rate": 6.913399166679744e-05,
      "loss": 2.0839,
      "step": 179604
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.2872698307037354,
      "learning_rate": 6.913137952679234e-05,
      "loss": 2.0807,
      "step": 179605
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.5524789094924927,
      "learning_rate": 6.912876742970994e-05,
      "loss": 1.9664,
      "step": 179606
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.5293182134628296,
      "learning_rate": 6.912615537555094e-05,
      "loss": 1.8552,
      "step": 179607
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.3166825771331787,
      "learning_rate": 6.912354336431564e-05,
      "loss": 1.94,
      "step": 179608
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.3357715606689453,
      "learning_rate": 6.912093139600473e-05,
      "loss": 2.1568,
      "step": 179609
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.3402986526489258,
      "learning_rate": 6.911831947061848e-05,
      "loss": 1.9926,
      "step": 179610
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.2342191934585571,
      "learning_rate": 6.911570758815771e-05,
      "loss": 2.2378,
      "step": 179611
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.3313158750534058,
      "learning_rate": 6.911309574862249e-05,
      "loss": 2.1154,
      "step": 179612
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.3707109689712524,
      "learning_rate": 6.91104839520136e-05,
      "loss": 1.9293,
      "step": 179613
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.223802924156189,
      "learning_rate": 6.910787219833135e-05,
      "loss": 2.0252,
      "step": 179614
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.5070444345474243,
      "learning_rate": 6.910526048757638e-05,
      "loss": 2.0048,
      "step": 179615
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.31509530544281,
      "learning_rate": 6.910264881974906e-05,
      "loss": 1.926,
      "step": 179616
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.3054981231689453,
      "learning_rate": 6.910003719485001e-05,
      "loss": 1.8413,
      "step": 179617
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.2858541011810303,
      "learning_rate": 6.90974256128795e-05,
      "loss": 1.9909,
      "step": 179618
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.3605072498321533,
      "learning_rate": 6.90948140738382e-05,
      "loss": 1.8615,
      "step": 179619
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.2876747846603394,
      "learning_rate": 6.909220257772645e-05,
      "loss": 2.0311,
      "step": 179620
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.375776767730713,
      "learning_rate": 6.908959112454492e-05,
      "loss": 2.0762,
      "step": 179621
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.3119499683380127,
      "learning_rate": 6.908697971429391e-05,
      "loss": 2.0893,
      "step": 179622
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.3073025941848755,
      "learning_rate": 6.908436834697416e-05,
      "loss": 2.0401,
      "step": 179623
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.3647277355194092,
      "learning_rate": 6.908175702258581e-05,
      "loss": 1.9763,
      "step": 179624
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.3064121007919312,
      "learning_rate": 6.907914574112957e-05,
      "loss": 1.9501,
      "step": 179625
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.3875198364257812,
      "learning_rate": 6.907653450260583e-05,
      "loss": 2.0197,
      "step": 179626
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.3219339847564697,
      "learning_rate": 6.907392330701517e-05,
      "loss": 1.9079,
      "step": 179627
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.2997850179672241,
      "learning_rate": 6.907131215435799e-05,
      "loss": 2.171,
      "step": 179628
    },
    {
      "epoch": 2.34,
      "grad_norm": 2.6975598335266113,
      "learning_rate": 6.906870104463488e-05,
      "loss": 2.0545,
      "step": 179629
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.3793734312057495,
      "learning_rate": 6.906608997784627e-05,
      "loss": 1.8393,
      "step": 179630
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.4149435758590698,
      "learning_rate": 6.90634789539926e-05,
      "loss": 1.921,
      "step": 179631
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.3294416666030884,
      "learning_rate": 6.906086797307433e-05,
      "loss": 2.0926,
      "step": 179632
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.5704760551452637,
      "learning_rate": 6.905825703509205e-05,
      "loss": 2.0099,
      "step": 179633
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.2645936012268066,
      "learning_rate": 6.905564614004615e-05,
      "loss": 2.0539,
      "step": 179634
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.4637361764907837,
      "learning_rate": 6.905303528793724e-05,
      "loss": 2.1572,
      "step": 179635
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.3488596677780151,
      "learning_rate": 6.905042447876575e-05,
      "loss": 2.0726,
      "step": 179636
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.3853201866149902,
      "learning_rate": 6.904781371253212e-05,
      "loss": 2.3207,
      "step": 179637
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.2581124305725098,
      "learning_rate": 6.904520298923678e-05,
      "loss": 1.898,
      "step": 179638
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.4144823551177979,
      "learning_rate": 6.904259230888038e-05,
      "loss": 2.0035,
      "step": 179639
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.4526983499526978,
      "learning_rate": 6.903998167146326e-05,
      "loss": 2.0785,
      "step": 179640
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.3377444744110107,
      "learning_rate": 6.903737107698604e-05,
      "loss": 1.8208,
      "step": 179641
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.3647068738937378,
      "learning_rate": 6.90347605254491e-05,
      "loss": 2.0228,
      "step": 179642
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.4429445266723633,
      "learning_rate": 6.90321500168529e-05,
      "loss": 2.1062,
      "step": 179643
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.532560110092163,
      "learning_rate": 6.902953955119806e-05,
      "loss": 1.9928,
      "step": 179644
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.662950873374939,
      "learning_rate": 6.9026929128485e-05,
      "loss": 2.0146,
      "step": 179645
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.3378273248672485,
      "learning_rate": 6.90243187487141e-05,
      "loss": 2.0375,
      "step": 179646
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.2749103307724,
      "learning_rate": 6.902170841188604e-05,
      "loss": 2.3241,
      "step": 179647
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.3352108001708984,
      "learning_rate": 6.90190981180012e-05,
      "loss": 2.121,
      "step": 179648
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.3875133991241455,
      "learning_rate": 6.901648786705997e-05,
      "loss": 2.0147,
      "step": 179649
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.1760303974151611,
      "learning_rate": 6.901387765906303e-05,
      "loss": 1.786,
      "step": 179650
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.3402093648910522,
      "learning_rate": 6.90112674940107e-05,
      "loss": 2.0132,
      "step": 179651
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.2753591537475586,
      "learning_rate": 6.900865737190359e-05,
      "loss": 2.198,
      "step": 179652
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.4809627532958984,
      "learning_rate": 6.900604729274215e-05,
      "loss": 1.8237,
      "step": 179653
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.6663856506347656,
      "learning_rate": 6.900343725652683e-05,
      "loss": 2.1515,
      "step": 179654
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.2845654487609863,
      "learning_rate": 6.900082726325808e-05,
      "loss": 1.9459,
      "step": 179655
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.333262324333191,
      "learning_rate": 6.899821731293649e-05,
      "loss": 2.0139,
      "step": 179656
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.3461703062057495,
      "learning_rate": 6.89956074055624e-05,
      "loss": 1.9787,
      "step": 179657
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.4381837844848633,
      "learning_rate": 6.899299754113649e-05,
      "loss": 2.0621,
      "step": 179658
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.3861985206604004,
      "learning_rate": 6.899038771965917e-05,
      "loss": 1.932,
      "step": 179659
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.2898766994476318,
      "learning_rate": 6.898777794113085e-05,
      "loss": 2.0014,
      "step": 179660
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.2718877792358398,
      "learning_rate": 6.898516820555202e-05,
      "loss": 1.9305,
      "step": 179661
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.229655385017395,
      "learning_rate": 6.898255851292325e-05,
      "loss": 2.0593,
      "step": 179662
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.266013264656067,
      "learning_rate": 6.897994886324493e-05,
      "loss": 1.9487,
      "step": 179663
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.338922142982483,
      "learning_rate": 6.897733925651765e-05,
      "loss": 2.1136,
      "step": 179664
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.4133985042572021,
      "learning_rate": 6.897472969274178e-05,
      "loss": 2.0102,
      "step": 179665
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.2438819408416748,
      "learning_rate": 6.897212017191806e-05,
      "loss": 1.9118,
      "step": 179666
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.4333845376968384,
      "learning_rate": 6.896951069404656e-05,
      "loss": 1.939,
      "step": 179667
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.2937787771224976,
      "learning_rate": 6.896690125912812e-05,
      "loss": 2.1366,
      "step": 179668
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.4234410524368286,
      "learning_rate": 6.896429186716299e-05,
      "loss": 1.8942,
      "step": 179669
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.2279016971588135,
      "learning_rate": 6.896168251815186e-05,
      "loss": 1.8635,
      "step": 179670
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.3448116779327393,
      "learning_rate": 6.895907321209501e-05,
      "loss": 2.1618,
      "step": 179671
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.3691660165786743,
      "learning_rate": 6.895646394899322e-05,
      "loss": 1.9241,
      "step": 179672
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.2535909414291382,
      "learning_rate": 6.895385472884658e-05,
      "loss": 2.036,
      "step": 179673
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.2806655168533325,
      "learning_rate": 6.895124555165591e-05,
      "loss": 2.0793,
      "step": 179674
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.297621250152588,
      "learning_rate": 6.894863641742144e-05,
      "loss": 1.7793,
      "step": 179675
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.3214397430419922,
      "learning_rate": 6.894602732614388e-05,
      "loss": 2.0544,
      "step": 179676
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.546075701713562,
      "learning_rate": 6.894341827782356e-05,
      "loss": 2.1054,
      "step": 179677
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.4164797067642212,
      "learning_rate": 6.894080927246114e-05,
      "loss": 1.9695,
      "step": 179678
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.309822916984558,
      "learning_rate": 6.893820031005685e-05,
      "loss": 2.08,
      "step": 179679
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.230994701385498,
      "learning_rate": 6.893559139061137e-05,
      "loss": 2.0349,
      "step": 179680
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.3010872602462769,
      "learning_rate": 6.893298251412507e-05,
      "loss": 2.1432,
      "step": 179681
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.4305238723754883,
      "learning_rate": 6.893037368059856e-05,
      "loss": 2.1687,
      "step": 179682
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.373318076133728,
      "learning_rate": 6.892776489003216e-05,
      "loss": 1.9806,
      "step": 179683
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.5365309715270996,
      "learning_rate": 6.892515614242662e-05,
      "loss": 1.8947,
      "step": 179684
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.2992335557937622,
      "learning_rate": 6.892254743778212e-05,
      "loss": 2.0996,
      "step": 179685
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.3679405450820923,
      "learning_rate": 6.891993877609932e-05,
      "loss": 2.0512,
      "step": 179686
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.3053499460220337,
      "learning_rate": 6.891733015737861e-05,
      "loss": 2.0643,
      "step": 179687
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.601982593536377,
      "learning_rate": 6.891472158162058e-05,
      "loss": 2.2182,
      "step": 179688
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.4840389490127563,
      "learning_rate": 6.891211304882561e-05,
      "loss": 2.0806,
      "step": 179689
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.2376700639724731,
      "learning_rate": 6.890950455899441e-05,
      "loss": 1.9404,
      "step": 179690
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.3925455808639526,
      "learning_rate": 6.890689611212714e-05,
      "loss": 2.3399,
      "step": 179691
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.3297879695892334,
      "learning_rate": 6.89042877082245e-05,
      "loss": 2.1168,
      "step": 179692
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.4597675800323486,
      "learning_rate": 6.890167934728684e-05,
      "loss": 1.9957,
      "step": 179693
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.6546088457107544,
      "learning_rate": 6.889907102931479e-05,
      "loss": 1.9901,
      "step": 179694
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.4972102642059326,
      "learning_rate": 6.889646275430871e-05,
      "loss": 1.9609,
      "step": 179695
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.5041948556900024,
      "learning_rate": 6.88938545222692e-05,
      "loss": 1.9862,
      "step": 179696
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.36822509765625,
      "learning_rate": 6.889124633319668e-05,
      "loss": 2.0636,
      "step": 179697
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.3881466388702393,
      "learning_rate": 6.888863818709166e-05,
      "loss": 2.1615,
      "step": 179698
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.5177428722381592,
      "learning_rate": 6.88860300839545e-05,
      "loss": 2.1145,
      "step": 179699
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.5895999670028687,
      "learning_rate": 6.888342202378591e-05,
      "loss": 2.3196,
      "step": 179700
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.8690180778503418,
      "learning_rate": 6.888081400658614e-05,
      "loss": 2.1026,
      "step": 179701
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.4153811931610107,
      "learning_rate": 6.887820603235586e-05,
      "loss": 1.9237,
      "step": 179702
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.2966588735580444,
      "learning_rate": 6.887559810109551e-05,
      "loss": 1.8824,
      "step": 179703
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.5030802488327026,
      "learning_rate": 6.887299021280554e-05,
      "loss": 1.9085,
      "step": 179704
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.4784173965454102,
      "learning_rate": 6.887038236748637e-05,
      "loss": 1.9245,
      "step": 179705
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.3837387561798096,
      "learning_rate": 6.886777456513865e-05,
      "loss": 1.9989,
      "step": 179706
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.3481491804122925,
      "learning_rate": 6.886516680576267e-05,
      "loss": 2.0399,
      "step": 179707
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.5429638624191284,
      "learning_rate": 6.88625590893591e-05,
      "loss": 2.0418,
      "step": 179708
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.2714674472808838,
      "learning_rate": 6.885995141592834e-05,
      "loss": 1.8103,
      "step": 179709
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.3442810773849487,
      "learning_rate": 6.885734378547091e-05,
      "loss": 2.2218,
      "step": 179710
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.2787078619003296,
      "learning_rate": 6.885473619798714e-05,
      "loss": 1.7899,
      "step": 179711
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.4728703498840332,
      "learning_rate": 6.885212865347776e-05,
      "loss": 2.0011,
      "step": 179712
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.4585753679275513,
      "learning_rate": 6.8849521151943e-05,
      "loss": 1.899,
      "step": 179713
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.4488177299499512,
      "learning_rate": 6.884691369338359e-05,
      "loss": 1.7163,
      "step": 179714
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.4507429599761963,
      "learning_rate": 6.884430627779992e-05,
      "loss": 2.0292,
      "step": 179715
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.264721393585205,
      "learning_rate": 6.884169890519243e-05,
      "loss": 2.2455,
      "step": 179716
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.3109797239303589,
      "learning_rate": 6.883909157556157e-05,
      "loss": 1.9231,
      "step": 179717
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.3389006853103638,
      "learning_rate": 6.883648428890796e-05,
      "loss": 2.0761,
      "step": 179718
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.4526265859603882,
      "learning_rate": 6.88338770452319e-05,
      "loss": 2.2118,
      "step": 179719
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.3507758378982544,
      "learning_rate": 6.883126984453409e-05,
      "loss": 2.0285,
      "step": 179720
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.3140580654144287,
      "learning_rate": 6.882866268681491e-05,
      "loss": 2.0342,
      "step": 179721
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.3141112327575684,
      "learning_rate": 6.882605557207485e-05,
      "loss": 1.8863,
      "step": 179722
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.232206106185913,
      "learning_rate": 6.882344850031432e-05,
      "loss": 1.9125,
      "step": 179723
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.4267791509628296,
      "learning_rate": 6.882084147153397e-05,
      "loss": 2.0911,
      "step": 179724
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.3445600271224976,
      "learning_rate": 6.881823448573406e-05,
      "loss": 1.8185,
      "step": 179725
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.3279720544815063,
      "learning_rate": 6.881562754291531e-05,
      "loss": 2.0892,
      "step": 179726
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.2713146209716797,
      "learning_rate": 6.881302064307807e-05,
      "loss": 1.9771,
      "step": 179727
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.4447085857391357,
      "learning_rate": 6.881041378622283e-05,
      "loss": 2.1006,
      "step": 179728
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.2724324464797974,
      "learning_rate": 6.880780697235013e-05,
      "loss": 2.2212,
      "step": 179729
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.9628568887710571,
      "learning_rate": 6.880520020146045e-05,
      "loss": 1.8182,
      "step": 179730
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.3518165349960327,
      "learning_rate": 6.880259347355415e-05,
      "loss": 1.8661,
      "step": 179731
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.2087570428848267,
      "learning_rate": 6.879998678863192e-05,
      "loss": 2.2272,
      "step": 179732
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.4262794256210327,
      "learning_rate": 6.879738014669411e-05,
      "loss": 2.019,
      "step": 179733
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.3040924072265625,
      "learning_rate": 6.879477354774114e-05,
      "loss": 1.9933,
      "step": 179734
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.488899827003479,
      "learning_rate": 6.87921669917737e-05,
      "loss": 2.0836,
      "step": 179735
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.457772970199585,
      "learning_rate": 6.878956047879205e-05,
      "loss": 1.8394,
      "step": 179736
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.3318283557891846,
      "learning_rate": 6.878695400879688e-05,
      "loss": 2.0086,
      "step": 179737
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.3022507429122925,
      "learning_rate": 6.878434758178858e-05,
      "loss": 1.908,
      "step": 179738
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.3564610481262207,
      "learning_rate": 6.878174119776765e-05,
      "loss": 2.2286,
      "step": 179739
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.3722529411315918,
      "learning_rate": 6.877913485673445e-05,
      "loss": 1.9776,
      "step": 179740
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.1811085939407349,
      "learning_rate": 6.877652855868968e-05,
      "loss": 2.1063,
      "step": 179741
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.411297082901001,
      "learning_rate": 6.87739223036336e-05,
      "loss": 1.9107,
      "step": 179742
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.3660856485366821,
      "learning_rate": 6.877131609156691e-05,
      "loss": 1.8059,
      "step": 179743
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.39159095287323,
      "learning_rate": 6.876870992248999e-05,
      "loss": 2.2794,
      "step": 179744
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.2628211975097656,
      "learning_rate": 6.876610379640336e-05,
      "loss": 1.9192,
      "step": 179745
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.3192253112792969,
      "learning_rate": 6.876349771330737e-05,
      "loss": 1.9663,
      "step": 179746
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.3268547058105469,
      "learning_rate": 6.876089167320272e-05,
      "loss": 1.9892,
      "step": 179747
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.3392668962478638,
      "learning_rate": 6.875828567608969e-05,
      "loss": 1.9575,
      "step": 179748
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.9033527374267578,
      "learning_rate": 6.875567972196892e-05,
      "loss": 1.8427,
      "step": 179749
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.1981741189956665,
      "learning_rate": 6.875307381084078e-05,
      "loss": 1.7987,
      "step": 179750
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.2263245582580566,
      "learning_rate": 6.875046794270596e-05,
      "loss": 2.1688,
      "step": 179751
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.5298497676849365,
      "learning_rate": 6.874786211756463e-05,
      "loss": 2.096,
      "step": 179752
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.3575834035873413,
      "learning_rate": 6.874525633541751e-05,
      "loss": 1.9055,
      "step": 179753
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.374790072441101,
      "learning_rate": 6.874265059626496e-05,
      "loss": 2.0002,
      "step": 179754
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.4433794021606445,
      "learning_rate": 6.874004490010758e-05,
      "loss": 1.979,
      "step": 179755
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.220565915107727,
      "learning_rate": 6.873743924694572e-05,
      "loss": 2.0322,
      "step": 179756
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.3744752407073975,
      "learning_rate": 6.873483363678009e-05,
      "loss": 1.9452,
      "step": 179757
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.381583333015442,
      "learning_rate": 6.873222806961082e-05,
      "loss": 2.0071,
      "step": 179758
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.5131205320358276,
      "learning_rate": 6.872962254543871e-05,
      "loss": 2.1074,
      "step": 179759
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.3351991176605225,
      "learning_rate": 6.872701706426407e-05,
      "loss": 1.9313,
      "step": 179760
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.3567103147506714,
      "learning_rate": 6.872441162608749e-05,
      "loss": 2.0475,
      "step": 179761
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.430171012878418,
      "learning_rate": 6.872180623090933e-05,
      "loss": 2.0827,
      "step": 179762
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.3901182413101196,
      "learning_rate": 6.871920087873027e-05,
      "loss": 2.0761,
      "step": 179763
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.3304040431976318,
      "learning_rate": 6.871659556955064e-05,
      "loss": 2.244,
      "step": 179764
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.343117594718933,
      "learning_rate": 6.871399030337094e-05,
      "loss": 1.6964,
      "step": 179765
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.3561468124389648,
      "learning_rate": 6.871138508019162e-05,
      "loss": 2.1224,
      "step": 179766
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.3034108877182007,
      "learning_rate": 6.870877990001331e-05,
      "loss": 1.9116,
      "step": 179767
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.2216873168945312,
      "learning_rate": 6.870617476283633e-05,
      "loss": 1.8064,
      "step": 179768
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.2966585159301758,
      "learning_rate": 6.870356966866128e-05,
      "loss": 1.9099,
      "step": 179769
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.313812017440796,
      "learning_rate": 6.870096461748864e-05,
      "loss": 2.2795,
      "step": 179770
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.3625357151031494,
      "learning_rate": 6.869835960931887e-05,
      "loss": 1.8784,
      "step": 179771
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.365524411201477,
      "learning_rate": 6.869575464415232e-05,
      "loss": 2.0209,
      "step": 179772
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.570399522781372,
      "learning_rate": 6.869314972198968e-05,
      "loss": 1.9985,
      "step": 179773
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.3213647603988647,
      "learning_rate": 6.869054484283127e-05,
      "loss": 1.9625,
      "step": 179774
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.4914060831069946,
      "learning_rate": 6.868794000667776e-05,
      "loss": 2.0118,
      "step": 179775
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.3019276857376099,
      "learning_rate": 6.868533521352953e-05,
      "loss": 2.0606,
      "step": 179776
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.5231289863586426,
      "learning_rate": 6.868273046338702e-05,
      "loss": 1.8363,
      "step": 179777
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.5201464891433716,
      "learning_rate": 6.868012575625071e-05,
      "loss": 1.7794,
      "step": 179778
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.3609024286270142,
      "learning_rate": 6.867752109212119e-05,
      "loss": 2.0323,
      "step": 179779
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.4777365922927856,
      "learning_rate": 6.867491647099881e-05,
      "loss": 1.9543,
      "step": 179780
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.4825141429901123,
      "learning_rate": 6.867231189288422e-05,
      "loss": 2.0141,
      "step": 179781
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.337058663368225,
      "learning_rate": 6.866970735777784e-05,
      "loss": 2.1039,
      "step": 179782
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.4093605279922485,
      "learning_rate": 6.866710286568008e-05,
      "loss": 1.8526,
      "step": 179783
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.263295292854309,
      "learning_rate": 6.866449841659142e-05,
      "loss": 2.0838,
      "step": 179784
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.2835867404937744,
      "learning_rate": 6.866189401051246e-05,
      "loss": 2.26,
      "step": 179785
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.311193585395813,
      "learning_rate": 6.865928964744354e-05,
      "loss": 1.9628,
      "step": 179786
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.2628287076950073,
      "learning_rate": 6.865668532738534e-05,
      "loss": 1.8841,
      "step": 179787
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.2574818134307861,
      "learning_rate": 6.865408105033819e-05,
      "loss": 2.1306,
      "step": 179788
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.4136743545532227,
      "learning_rate": 6.865147681630265e-05,
      "loss": 2.0544,
      "step": 179789
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.278878927230835,
      "learning_rate": 6.864887262527905e-05,
      "loss": 2.0651,
      "step": 179790
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.3236737251281738,
      "learning_rate": 6.864626847726808e-05,
      "loss": 2.2314,
      "step": 179791
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.335288643836975,
      "learning_rate": 6.864366437227005e-05,
      "loss": 2.0991,
      "step": 179792
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.1919409036636353,
      "learning_rate": 6.864106031028564e-05,
      "loss": 1.9582,
      "step": 179793
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.390757441520691,
      "learning_rate": 6.863845629131521e-05,
      "loss": 1.9947,
      "step": 179794
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.4313360452651978,
      "learning_rate": 6.863585231535925e-05,
      "loss": 2.0573,
      "step": 179795
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.3560593128204346,
      "learning_rate": 6.863324838241818e-05,
      "loss": 2.0977,
      "step": 179796
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.3027656078338623,
      "learning_rate": 6.863064449249264e-05,
      "loss": 2.146,
      "step": 179797
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.232753872871399,
      "learning_rate": 6.862804064558292e-05,
      "loss": 1.9404,
      "step": 179798
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.2992786169052124,
      "learning_rate": 6.862543684168972e-05,
      "loss": 2.1078,
      "step": 179799
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.2711379528045654,
      "learning_rate": 6.862283308081341e-05,
      "loss": 1.9285,
      "step": 179800
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.3837579488754272,
      "learning_rate": 6.86202293629545e-05,
      "loss": 1.8136,
      "step": 179801
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.3060803413391113,
      "learning_rate": 6.861762568811336e-05,
      "loss": 1.8971,
      "step": 179802
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.3011401891708374,
      "learning_rate": 6.861502205629065e-05,
      "loss": 2.0153,
      "step": 179803
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.398177146911621,
      "learning_rate": 6.861241846748671e-05,
      "loss": 2.1398,
      "step": 179804
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.3870983123779297,
      "learning_rate": 6.860981492170217e-05,
      "loss": 1.7111,
      "step": 179805
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.254767894744873,
      "learning_rate": 6.860721141893744e-05,
      "loss": 1.9755,
      "step": 179806
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.4897056818008423,
      "learning_rate": 6.860460795919298e-05,
      "loss": 1.9969,
      "step": 179807
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.2940092086791992,
      "learning_rate": 6.860200454246922e-05,
      "loss": 1.9271,
      "step": 179808
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.3523118495941162,
      "learning_rate": 6.859940116876677e-05,
      "loss": 1.9865,
      "step": 179809
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.3146905899047852,
      "learning_rate": 6.859679783808601e-05,
      "loss": 2.0065,
      "step": 179810
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.426148772239685,
      "learning_rate": 6.859419455042753e-05,
      "loss": 2.0168,
      "step": 179811
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.3660019636154175,
      "learning_rate": 6.859159130579177e-05,
      "loss": 1.8981,
      "step": 179812
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.2955769300460815,
      "learning_rate": 6.858898810417913e-05,
      "loss": 2.0859,
      "step": 179813
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.348487377166748,
      "learning_rate": 6.858638494559024e-05,
      "loss": 1.7498,
      "step": 179814
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.3337273597717285,
      "learning_rate": 6.85837818300255e-05,
      "loss": 2.1207,
      "step": 179815
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.3584052324295044,
      "learning_rate": 6.858117875748532e-05,
      "loss": 1.9935,
      "step": 179816
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.3812586069107056,
      "learning_rate": 6.857857572797035e-05,
      "loss": 1.8756,
      "step": 179817
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.307856798171997,
      "learning_rate": 6.8575972741481e-05,
      "loss": 1.9998,
      "step": 179818
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.273895263671875,
      "learning_rate": 6.857336979801767e-05,
      "loss": 1.9526,
      "step": 179819
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.7624894380569458,
      "learning_rate": 6.857076689758098e-05,
      "loss": 1.9131,
      "step": 179820
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.416218638420105,
      "learning_rate": 6.856816404017136e-05,
      "loss": 1.9495,
      "step": 179821
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.3532673120498657,
      "learning_rate": 6.856556122578921e-05,
      "loss": 1.9919,
      "step": 179822
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.3570202589035034,
      "learning_rate": 6.85629584544352e-05,
      "loss": 2.0971,
      "step": 179823
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.2511430978775024,
      "learning_rate": 6.856035572610967e-05,
      "loss": 2.0282,
      "step": 179824
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.302086353302002,
      "learning_rate": 6.855775304081306e-05,
      "loss": 1.9691,
      "step": 179825
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.4249613285064697,
      "learning_rate": 6.855515039854602e-05,
      "loss": 2.116,
      "step": 179826
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.272099494934082,
      "learning_rate": 6.855254779930886e-05,
      "loss": 1.9785,
      "step": 179827
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.379875898361206,
      "learning_rate": 6.854994524310223e-05,
      "loss": 2.0915,
      "step": 179828
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.4783740043640137,
      "learning_rate": 6.854734272992656e-05,
      "loss": 1.9556,
      "step": 179829
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.2087026834487915,
      "learning_rate": 6.854474025978221e-05,
      "loss": 1.9385,
      "step": 179830
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.2834672927856445,
      "learning_rate": 6.854213783266983e-05,
      "loss": 2.1291,
      "step": 179831
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.595556378364563,
      "learning_rate": 6.853953544858983e-05,
      "loss": 2.0498,
      "step": 179832
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.2512938976287842,
      "learning_rate": 6.853693310754265e-05,
      "loss": 1.9392,
      "step": 179833
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.359432578086853,
      "learning_rate": 6.853433080952886e-05,
      "loss": 2.2368,
      "step": 179834
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.6091686487197876,
      "learning_rate": 6.853172855454887e-05,
      "loss": 1.9414,
      "step": 179835
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.4203879833221436,
      "learning_rate": 6.852912634260328e-05,
      "loss": 2.1829,
      "step": 179836
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.311781644821167,
      "learning_rate": 6.852652417369246e-05,
      "loss": 2.1673,
      "step": 179837
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.3395761251449585,
      "learning_rate": 6.852392204781696e-05,
      "loss": 1.9189,
      "step": 179838
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.3037679195404053,
      "learning_rate": 6.852131996497712e-05,
      "loss": 1.9294,
      "step": 179839
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.3394775390625,
      "learning_rate": 6.851871792517364e-05,
      "loss": 2.0555,
      "step": 179840
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.3613061904907227,
      "learning_rate": 6.851611592840683e-05,
      "loss": 2.101,
      "step": 179841
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.5191938877105713,
      "learning_rate": 6.851351397467731e-05,
      "loss": 2.2797,
      "step": 179842
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.2532234191894531,
      "learning_rate": 6.851091206398551e-05,
      "loss": 2.0296,
      "step": 179843
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.4255083799362183,
      "learning_rate": 6.850831019633188e-05,
      "loss": 2.0699,
      "step": 179844
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.3386520147323608,
      "learning_rate": 6.850570837171686e-05,
      "loss": 1.9203,
      "step": 179845
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.2766014337539673,
      "learning_rate": 6.850310659014108e-05,
      "loss": 2.1001,
      "step": 179846
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.4444935321807861,
      "learning_rate": 6.850050485160483e-05,
      "loss": 2.0921,
      "step": 179847
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.2152748107910156,
      "learning_rate": 6.849790315610882e-05,
      "loss": 1.7148,
      "step": 179848
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.3455713987350464,
      "learning_rate": 6.849530150365342e-05,
      "loss": 1.9821,
      "step": 179849
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.2778730392456055,
      "learning_rate": 6.849269989423909e-05,
      "loss": 1.8554,
      "step": 179850
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.4245580434799194,
      "learning_rate": 6.849009832786627e-05,
      "loss": 2.2759,
      "step": 179851
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.3740063905715942,
      "learning_rate": 6.84874968045356e-05,
      "loss": 1.8881,
      "step": 179852
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.278428554534912,
      "learning_rate": 6.848489532424738e-05,
      "loss": 1.9408,
      "step": 179853
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.4459046125411987,
      "learning_rate": 6.84822938870023e-05,
      "loss": 1.8805,
      "step": 179854
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.3192647695541382,
      "learning_rate": 6.847969249280067e-05,
      "loss": 1.9965,
      "step": 179855
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.2424533367156982,
      "learning_rate": 6.847709114164309e-05,
      "loss": 1.9242,
      "step": 179856
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.340004801750183,
      "learning_rate": 6.847448983352988e-05,
      "loss": 2.1087,
      "step": 179857
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.3568722009658813,
      "learning_rate": 6.847188856846173e-05,
      "loss": 2.0014,
      "step": 179858
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.4464157819747925,
      "learning_rate": 6.846928734643895e-05,
      "loss": 1.9529,
      "step": 179859
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.3061144351959229,
      "learning_rate": 6.846668616746217e-05,
      "loss": 2.0795,
      "step": 179860
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.3622487783432007,
      "learning_rate": 6.846408503153178e-05,
      "loss": 2.0417,
      "step": 179861
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.4059885740280151,
      "learning_rate": 6.846148393864834e-05,
      "loss": 1.985,
      "step": 179862
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.314327359199524,
      "learning_rate": 6.845888288881216e-05,
      "loss": 2.1668,
      "step": 179863
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.3472959995269775,
      "learning_rate": 6.845628188202394e-05,
      "loss": 1.9225,
      "step": 179864
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.3068231344223022,
      "learning_rate": 6.845368091828399e-05,
      "loss": 2.105,
      "step": 179865
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.3545150756835938,
      "learning_rate": 6.845107999759294e-05,
      "loss": 2.0101,
      "step": 179866
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.421539545059204,
      "learning_rate": 6.844847911995121e-05,
      "loss": 2.223,
      "step": 179867
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.5806303024291992,
      "learning_rate": 6.844587828535929e-05,
      "loss": 1.7784,
      "step": 179868
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.2915945053100586,
      "learning_rate": 6.844327749381757e-05,
      "loss": 2.0836,
      "step": 179869
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.41459321975708,
      "learning_rate": 6.844067674532666e-05,
      "loss": 2.3161,
      "step": 179870
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.3743442296981812,
      "learning_rate": 6.843807603988696e-05,
      "loss": 2.166,
      "step": 179871
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.451812982559204,
      "learning_rate": 6.843547537749905e-05,
      "loss": 2.2254,
      "step": 179872
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.2450447082519531,
      "learning_rate": 6.843287475816336e-05,
      "loss": 2.0109,
      "step": 179873
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.3202179670333862,
      "learning_rate": 6.843027418188039e-05,
      "loss": 1.943,
      "step": 179874
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.3385963439941406,
      "learning_rate": 6.84276736486505e-05,
      "loss": 2.0093,
      "step": 179875
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.3774088621139526,
      "learning_rate": 6.842507315847436e-05,
      "loss": 1.9673,
      "step": 179876
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.3146971464157104,
      "learning_rate": 6.84224727113523e-05,
      "loss": 1.8741,
      "step": 179877
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.4500669240951538,
      "learning_rate": 6.841987230728495e-05,
      "loss": 2.0613,
      "step": 179878
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.242920160293579,
      "learning_rate": 6.84172719462727e-05,
      "loss": 1.8965,
      "step": 179879
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.2076966762542725,
      "learning_rate": 6.841467162831607e-05,
      "loss": 1.9894,
      "step": 179880
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.230534315109253,
      "learning_rate": 6.841207135341544e-05,
      "loss": 2.0681,
      "step": 179881
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.5325614213943481,
      "learning_rate": 6.840947112157147e-05,
      "loss": 1.9577,
      "step": 179882
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.3787658214569092,
      "learning_rate": 6.840687093278445e-05,
      "loss": 2.1273,
      "step": 179883
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.320004940032959,
      "learning_rate": 6.840427078705506e-05,
      "loss": 1.9118,
      "step": 179884
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.3737735748291016,
      "learning_rate": 6.840167068438369e-05,
      "loss": 1.9279,
      "step": 179885
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.4296141862869263,
      "learning_rate": 6.839907062477081e-05,
      "loss": 2.0955,
      "step": 179886
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.52510404586792,
      "learning_rate": 6.839647060821683e-05,
      "loss": 1.9369,
      "step": 179887
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.4580055475234985,
      "learning_rate": 6.839387063472239e-05,
      "loss": 2.0969,
      "step": 179888
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.3608617782592773,
      "learning_rate": 6.839127070428785e-05,
      "loss": 2.0277,
      "step": 179889
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.4878978729248047,
      "learning_rate": 6.838867081691383e-05,
      "loss": 1.9139,
      "step": 179890
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.4215667247772217,
      "learning_rate": 6.838607097260069e-05,
      "loss": 1.9446,
      "step": 179891
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.2229598760604858,
      "learning_rate": 6.838347117134898e-05,
      "loss": 1.9688,
      "step": 179892
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.1727406978607178,
      "learning_rate": 6.838087141315906e-05,
      "loss": 2.0234,
      "step": 179893
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.3349133729934692,
      "learning_rate": 6.837827169803159e-05,
      "loss": 1.8601,
      "step": 179894
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.4301023483276367,
      "learning_rate": 6.837567202596689e-05,
      "loss": 1.8403,
      "step": 179895
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.398174524307251,
      "learning_rate": 6.83730723969656e-05,
      "loss": 2.0428,
      "step": 179896
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.3411566019058228,
      "learning_rate": 6.837047281102805e-05,
      "loss": 1.9675,
      "step": 179897
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.355959415435791,
      "learning_rate": 6.83678732681549e-05,
      "loss": 2.0733,
      "step": 179898
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.4279714822769165,
      "learning_rate": 6.836527376834655e-05,
      "loss": 2.1038,
      "step": 179899
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.3666870594024658,
      "learning_rate": 6.836267431160343e-05,
      "loss": 2.0774,
      "step": 179900
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.6235237121582031,
      "learning_rate": 6.836007489792599e-05,
      "loss": 1.9322,
      "step": 179901
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.4822640419006348,
      "learning_rate": 6.835747552731487e-05,
      "loss": 1.8362,
      "step": 179902
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.2623589038848877,
      "learning_rate": 6.835487619977039e-05,
      "loss": 2.0741,
      "step": 179903
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.2868884801864624,
      "learning_rate": 6.835227691529317e-05,
      "loss": 1.9455,
      "step": 179904
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.34140944480896,
      "learning_rate": 6.834967767388366e-05,
      "loss": 1.957,
      "step": 179905
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.421524167060852,
      "learning_rate": 6.834707847554231e-05,
      "loss": 2.0515,
      "step": 179906
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.2603514194488525,
      "learning_rate": 6.834447932026953e-05,
      "loss": 2.0186,
      "step": 179907
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.2495156526565552,
      "learning_rate": 6.834188020806597e-05,
      "loss": 2.0199,
      "step": 179908
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.2248883247375488,
      "learning_rate": 6.833928113893193e-05,
      "loss": 1.888,
      "step": 179909
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.2530465126037598,
      "learning_rate": 6.833668211286809e-05,
      "loss": 1.9746,
      "step": 179910
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.7885799407958984,
      "learning_rate": 6.833408312987484e-05,
      "loss": 2.1598,
      "step": 179911
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.2547889947891235,
      "learning_rate": 6.833148418995253e-05,
      "loss": 1.5915,
      "step": 179912
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.4464131593704224,
      "learning_rate": 6.832888529310192e-05,
      "loss": 2.0713,
      "step": 179913
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.198980689048767,
      "learning_rate": 6.832628643932329e-05,
      "loss": 1.8487,
      "step": 179914
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.2730355262756348,
      "learning_rate": 6.83236876286171e-05,
      "loss": 2.1015,
      "step": 179915
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.4068541526794434,
      "learning_rate": 6.832108886098401e-05,
      "loss": 2.0469,
      "step": 179916
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.3222287893295288,
      "learning_rate": 6.83184901364244e-05,
      "loss": 1.9434,
      "step": 179917
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.2882710695266724,
      "learning_rate": 6.831589145493866e-05,
      "loss": 2.0705,
      "step": 179918
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.291589617729187,
      "learning_rate": 6.831329281652748e-05,
      "loss": 2.1639,
      "step": 179919
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.3027868270874023,
      "learning_rate": 6.83106942211912e-05,
      "loss": 2.2398,
      "step": 179920
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.234877347946167,
      "learning_rate": 6.830809566893027e-05,
      "loss": 1.91,
      "step": 179921
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.3901065587997437,
      "learning_rate": 6.83054971597453e-05,
      "loss": 2.0337,
      "step": 179922
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.2899996042251587,
      "learning_rate": 6.830289869363676e-05,
      "loss": 1.9289,
      "step": 179923
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.4679235219955444,
      "learning_rate": 6.830030027060495e-05,
      "loss": 2.1391,
      "step": 179924
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.2862060070037842,
      "learning_rate": 6.829770189065061e-05,
      "loss": 2.1783,
      "step": 179925
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.279660940170288,
      "learning_rate": 6.829510355377399e-05,
      "loss": 2.0942,
      "step": 179926
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.2290103435516357,
      "learning_rate": 6.829250525997577e-05,
      "loss": 1.7836,
      "step": 179927
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.3329650163650513,
      "learning_rate": 6.828990700925637e-05,
      "loss": 1.8045,
      "step": 179928
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.3834009170532227,
      "learning_rate": 6.828730880161623e-05,
      "loss": 1.9815,
      "step": 179929
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.2582062482833862,
      "learning_rate": 6.828471063705576e-05,
      "loss": 1.8905,
      "step": 179930
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.3211863040924072,
      "learning_rate": 6.828211251557564e-05,
      "loss": 2.0712,
      "step": 179931
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.223789930343628,
      "learning_rate": 6.827951443717615e-05,
      "loss": 2.1566,
      "step": 179932
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.3649563789367676,
      "learning_rate": 6.827691640185797e-05,
      "loss": 1.9637,
      "step": 179933
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.269864559173584,
      "learning_rate": 6.82743184096215e-05,
      "loss": 1.9858,
      "step": 179934
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.2729250192642212,
      "learning_rate": 6.827172046046719e-05,
      "loss": 2.0492,
      "step": 179935
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.323669195175171,
      "learning_rate": 6.826912255439546e-05,
      "loss": 1.8167,
      "step": 179936
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.2574193477630615,
      "learning_rate": 6.826652469140693e-05,
      "loss": 2.0557,
      "step": 179937
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.3050966262817383,
      "learning_rate": 6.826392687150197e-05,
      "loss": 2.0126,
      "step": 179938
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.302207112312317,
      "learning_rate": 6.826132909468121e-05,
      "loss": 1.9816,
      "step": 179939
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.5288856029510498,
      "learning_rate": 6.825873136094503e-05,
      "loss": 1.8867,
      "step": 179940
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.4137496948242188,
      "learning_rate": 6.825613367029394e-05,
      "loss": 2.0688,
      "step": 179941
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.24053156375885,
      "learning_rate": 6.825353602272833e-05,
      "loss": 1.7438,
      "step": 179942
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.332824945449829,
      "learning_rate": 6.825093841824881e-05,
      "loss": 2.4293,
      "step": 179943
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.3173840045928955,
      "learning_rate": 6.824834085685576e-05,
      "loss": 1.9975,
      "step": 179944
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.2418906688690186,
      "learning_rate": 6.824574333854982e-05,
      "loss": 2.0542,
      "step": 179945
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.3349968194961548,
      "learning_rate": 6.824314586333135e-05,
      "loss": 2.0069,
      "step": 179946
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.3781161308288574,
      "learning_rate": 6.824054843120083e-05,
      "loss": 2.1172,
      "step": 179947
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.3252230882644653,
      "learning_rate": 6.823795104215871e-05,
      "loss": 1.9284,
      "step": 179948
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.3147499561309814,
      "learning_rate": 6.823535369620562e-05,
      "loss": 1.9966,
      "step": 179949
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.379073977470398,
      "learning_rate": 6.823275639334185e-05,
      "loss": 2.1049,
      "step": 179950
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.2980364561080933,
      "learning_rate": 6.823015913356806e-05,
      "loss": 2.0469,
      "step": 179951
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.395863652229309,
      "learning_rate": 6.822756191688471e-05,
      "loss": 2.0939,
      "step": 179952
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.220254898071289,
      "learning_rate": 6.822496474329219e-05,
      "loss": 1.9412,
      "step": 179953
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.2864381074905396,
      "learning_rate": 6.822236761279093e-05,
      "loss": 2.0667,
      "step": 179954
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.2514266967773438,
      "learning_rate": 6.82197705253816e-05,
      "loss": 1.8435,
      "step": 179955
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.336626410484314,
      "learning_rate": 6.821717348106453e-05,
      "loss": 1.6554,
      "step": 179956
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.355899453163147,
      "learning_rate": 6.821457647984033e-05,
      "loss": 2.0706,
      "step": 179957
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.2659566402435303,
      "learning_rate": 6.821197952170943e-05,
      "loss": 1.9522,
      "step": 179958
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.322311282157898,
      "learning_rate": 6.820938260667228e-05,
      "loss": 1.9643,
      "step": 179959
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.4483568668365479,
      "learning_rate": 6.820678573472929e-05,
      "loss": 2.0313,
      "step": 179960
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.3772553205490112,
      "learning_rate": 6.820418890588114e-05,
      "loss": 1.9326,
      "step": 179961
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.3967546224594116,
      "learning_rate": 6.820159212012811e-05,
      "loss": 2.2448,
      "step": 179962
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.2935616970062256,
      "learning_rate": 6.819899537747085e-05,
      "loss": 1.9925,
      "step": 179963
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.4357861280441284,
      "learning_rate": 6.81963986779097e-05,
      "loss": 2.182,
      "step": 179964
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.3665276765823364,
      "learning_rate": 6.81938020214454e-05,
      "loss": 2.0879,
      "step": 179965
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.3453762531280518,
      "learning_rate": 6.819120540807806e-05,
      "loss": 2.0343,
      "step": 179966
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.3018513917922974,
      "learning_rate": 6.818860883780843e-05,
      "loss": 2.0037,
      "step": 179967
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.41732919216156,
      "learning_rate": 6.818601231063686e-05,
      "loss": 2.129,
      "step": 179968
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.269274115562439,
      "learning_rate": 6.818341582656394e-05,
      "loss": 1.9782,
      "step": 179969
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.4102504253387451,
      "learning_rate": 6.818081938559002e-05,
      "loss": 2.1281,
      "step": 179970
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.2112599611282349,
      "learning_rate": 6.817822298771585e-05,
      "loss": 2.0635,
      "step": 179971
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.341353416442871,
      "learning_rate": 6.817562663294154e-05,
      "loss": 1.9545,
      "step": 179972
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.27033531665802,
      "learning_rate": 6.817303032126786e-05,
      "loss": 2.0053,
      "step": 179973
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.3674089908599854,
      "learning_rate": 6.817043405269506e-05,
      "loss": 1.9416,
      "step": 179974
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.4047071933746338,
      "learning_rate": 6.81678378272239e-05,
      "loss": 2.0578,
      "step": 179975
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.3633155822753906,
      "learning_rate": 6.816524164485458e-05,
      "loss": 2.2874,
      "step": 179976
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.2981438636779785,
      "learning_rate": 6.816264550558791e-05,
      "loss": 2.0229,
      "step": 179977
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.426254391670227,
      "learning_rate": 6.8160049409424e-05,
      "loss": 2.0526,
      "step": 179978
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.4383238554000854,
      "learning_rate": 6.815745335636362e-05,
      "loss": 1.9805,
      "step": 179979
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.5177186727523804,
      "learning_rate": 6.815485734640701e-05,
      "loss": 2.1657,
      "step": 179980
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.3712029457092285,
      "learning_rate": 6.815226137955493e-05,
      "loss": 2.0865,
      "step": 179981
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.3479371070861816,
      "learning_rate": 6.81496654558076e-05,
      "loss": 1.9575,
      "step": 179982
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.279508352279663,
      "learning_rate": 6.814706957516579e-05,
      "loss": 2.2089,
      "step": 179983
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.357204794883728,
      "learning_rate": 6.814447373762967e-05,
      "loss": 2.0473,
      "step": 179984
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.392004370689392,
      "learning_rate": 6.814187794319996e-05,
      "loss": 2.0841,
      "step": 179985
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.3380544185638428,
      "learning_rate": 6.813928219187694e-05,
      "loss": 1.9952,
      "step": 179986
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.4323513507843018,
      "learning_rate": 6.813668648366131e-05,
      "loss": 1.9361,
      "step": 179987
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.257767915725708,
      "learning_rate": 6.813409081855336e-05,
      "loss": 1.9869,
      "step": 179988
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.3103296756744385,
      "learning_rate": 6.813149519655371e-05,
      "loss": 2.2614,
      "step": 179989
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.39712393283844,
      "learning_rate": 6.812889961766283e-05,
      "loss": 1.8662,
      "step": 179990
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.5129965543746948,
      "learning_rate": 6.812630408188117e-05,
      "loss": 1.9286,
      "step": 179991
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.2979353666305542,
      "learning_rate": 6.812370858920912e-05,
      "loss": 1.8692,
      "step": 179992
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.3116230964660645,
      "learning_rate": 6.812111313964732e-05,
      "loss": 1.9597,
      "step": 179993
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.2891523838043213,
      "learning_rate": 6.811851773319609e-05,
      "loss": 1.9605,
      "step": 179994
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.5468029975891113,
      "learning_rate": 6.811592236985608e-05,
      "loss": 1.9349,
      "step": 179995
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.6184675693511963,
      "learning_rate": 6.811332704962773e-05,
      "loss": 1.8703,
      "step": 179996
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.3496406078338623,
      "learning_rate": 6.811073177251137e-05,
      "loss": 1.9932,
      "step": 179997
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.4402974843978882,
      "learning_rate": 6.81081365385077e-05,
      "loss": 1.8636,
      "step": 179998
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.4656840562820435,
      "learning_rate": 6.810554134761712e-05,
      "loss": 2.0378,
      "step": 179999
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.3164515495300293,
      "learning_rate": 6.810294619984002e-05,
      "loss": 1.9192,
      "step": 180000
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.3302687406539917,
      "learning_rate": 6.810035109517702e-05,
      "loss": 2.0086,
      "step": 180001
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.3798598051071167,
      "learning_rate": 6.809775603362857e-05,
      "loss": 2.0565,
      "step": 180002
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.255446195602417,
      "learning_rate": 6.8095161015195e-05,
      "loss": 1.9677,
      "step": 180003
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.2847795486450195,
      "learning_rate": 6.809256603987704e-05,
      "loss": 2.1415,
      "step": 180004
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.3126164674758911,
      "learning_rate": 6.808997110767502e-05,
      "loss": 2.0309,
      "step": 180005
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.546947717666626,
      "learning_rate": 6.808737621858939e-05,
      "loss": 1.9938,
      "step": 180006
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.3026626110076904,
      "learning_rate": 6.808478137262076e-05,
      "loss": 2.1957,
      "step": 180007
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.3054203987121582,
      "learning_rate": 6.808218656976957e-05,
      "loss": 1.877,
      "step": 180008
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.3688366413116455,
      "learning_rate": 6.807959181003615e-05,
      "loss": 2.0272,
      "step": 180009
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.4444851875305176,
      "learning_rate": 6.807699709342122e-05,
      "loss": 1.83,
      "step": 180010
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.5343676805496216,
      "learning_rate": 6.80744024199251e-05,
      "loss": 2.1224,
      "step": 180011
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.3419450521469116,
      "learning_rate": 6.807180778954838e-05,
      "loss": 2.0507,
      "step": 180012
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.3853881359100342,
      "learning_rate": 6.80692132022915e-05,
      "loss": 2.0628,
      "step": 180013
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.287530779838562,
      "learning_rate": 6.806661865815493e-05,
      "loss": 1.9249,
      "step": 180014
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.4764026403427124,
      "learning_rate": 6.806402415713908e-05,
      "loss": 2.1358,
      "step": 180015
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.3099685907363892,
      "learning_rate": 6.806142969924455e-05,
      "loss": 1.8457,
      "step": 180016
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.3291151523590088,
      "learning_rate": 6.805883528447173e-05,
      "loss": 1.9508,
      "step": 180017
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.2820632457733154,
      "learning_rate": 6.805624091282123e-05,
      "loss": 2.0884,
      "step": 180018
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.394869327545166,
      "learning_rate": 6.805364658429345e-05,
      "loss": 2.0638,
      "step": 180019
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.3219633102416992,
      "learning_rate": 6.805105229888886e-05,
      "loss": 2.0298,
      "step": 180020
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.2903740406036377,
      "learning_rate": 6.804845805660789e-05,
      "loss": 2.0832,
      "step": 180021
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.2348837852478027,
      "learning_rate": 6.804586385745117e-05,
      "loss": 2.0444,
      "step": 180022
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.497122049331665,
      "learning_rate": 6.804326970141905e-05,
      "loss": 2.1055,
      "step": 180023
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.461044430732727,
      "learning_rate": 6.80406755885121e-05,
      "loss": 2.0186,
      "step": 180024
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.3690379858016968,
      "learning_rate": 6.803808151873079e-05,
      "loss": 1.9017,
      "step": 180025
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.3821659088134766,
      "learning_rate": 6.803548749207556e-05,
      "loss": 2.0,
      "step": 180026
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.3244441747665405,
      "learning_rate": 6.803289350854686e-05,
      "loss": 2.1513,
      "step": 180027
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.2869749069213867,
      "learning_rate": 6.803029956814528e-05,
      "loss": 1.9873,
      "step": 180028
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.307067632675171,
      "learning_rate": 6.802770567087117e-05,
      "loss": 1.9141,
      "step": 180029
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.263594627380371,
      "learning_rate": 6.802511181672518e-05,
      "loss": 1.7176,
      "step": 180030
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.2632558345794678,
      "learning_rate": 6.80225180057076e-05,
      "loss": 2.0169,
      "step": 180031
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.3192884922027588,
      "learning_rate": 6.80199242378192e-05,
      "loss": 1.9984,
      "step": 180032
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.4014451503753662,
      "learning_rate": 6.80173305130601e-05,
      "loss": 2.2169,
      "step": 180033
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.726896047592163,
      "learning_rate": 6.801473683143106e-05,
      "loss": 1.9631,
      "step": 180034
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.3372750282287598,
      "learning_rate": 6.801214319293235e-05,
      "loss": 1.9998,
      "step": 180035
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.361227035522461,
      "learning_rate": 6.800954959756465e-05,
      "loss": 1.8844,
      "step": 180036
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.3393555879592896,
      "learning_rate": 6.800695604532827e-05,
      "loss": 2.2358,
      "step": 180037
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.3331149816513062,
      "learning_rate": 6.800436253622394e-05,
      "loss": 2.1242,
      "step": 180038
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.4194118976593018,
      "learning_rate": 6.800176907025182e-05,
      "loss": 2.2149,
      "step": 180039
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.3633859157562256,
      "learning_rate": 6.799917564741264e-05,
      "loss": 2.1081,
      "step": 180040
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.3008642196655273,
      "learning_rate": 6.799658226770668e-05,
      "loss": 2.1108,
      "step": 180041
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.2922954559326172,
      "learning_rate": 6.799398893113464e-05,
      "loss": 2.2579,
      "step": 180042
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.4345256090164185,
      "learning_rate": 6.79913956376968e-05,
      "loss": 2.1425,
      "step": 180043
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.388266682624817,
      "learning_rate": 6.798880238739395e-05,
      "loss": 1.9834,
      "step": 180044
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.708296298980713,
      "learning_rate": 6.798620918022614e-05,
      "loss": 1.9623,
      "step": 180045
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.2341983318328857,
      "learning_rate": 6.798361601619418e-05,
      "loss": 2.0873,
      "step": 180046
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.4550530910491943,
      "learning_rate": 6.798102289529837e-05,
      "loss": 1.9761,
      "step": 180047
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.3644134998321533,
      "learning_rate": 6.797842981753935e-05,
      "loss": 1.9677,
      "step": 180048
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.3184844255447388,
      "learning_rate": 6.797583678291742e-05,
      "loss": 2.0879,
      "step": 180049
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.2601227760314941,
      "learning_rate": 6.797324379143332e-05,
      "loss": 2.1987,
      "step": 180050
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.4582927227020264,
      "learning_rate": 6.797065084308723e-05,
      "loss": 1.8531,
      "step": 180051
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.3113594055175781,
      "learning_rate": 6.796805793787985e-05,
      "loss": 2.2518,
      "step": 180052
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.3514190912246704,
      "learning_rate": 6.796546507581152e-05,
      "loss": 1.9794,
      "step": 180053
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.3016566038131714,
      "learning_rate": 6.796287225688288e-05,
      "loss": 1.9666,
      "step": 180054
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.2965714931488037,
      "learning_rate": 6.79602794810942e-05,
      "loss": 2.1201,
      "step": 180055
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.3875592947006226,
      "learning_rate": 6.795768674844626e-05,
      "loss": 1.9465,
      "step": 180056
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.3731627464294434,
      "learning_rate": 6.795509405893921e-05,
      "loss": 1.8891,
      "step": 180057
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.322326898574829,
      "learning_rate": 6.795250141257378e-05,
      "loss": 1.7371,
      "step": 180058
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.3031240701675415,
      "learning_rate": 6.794990880935027e-05,
      "loss": 2.0768,
      "step": 180059
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.3646690845489502,
      "learning_rate": 6.794731624926932e-05,
      "loss": 1.9744,
      "step": 180060
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.3349438905715942,
      "learning_rate": 6.794472373233129e-05,
      "loss": 1.931,
      "step": 180061
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.3735097646713257,
      "learning_rate": 6.794213125853686e-05,
      "loss": 1.9295,
      "step": 180062
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.2996788024902344,
      "learning_rate": 6.793953882788621e-05,
      "loss": 2.0352,
      "step": 180063
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.3628205060958862,
      "learning_rate": 6.793694644038004e-05,
      "loss": 1.9542,
      "step": 180064
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.361059308052063,
      "learning_rate": 6.793435409601875e-05,
      "loss": 1.8122,
      "step": 180065
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.3418272733688354,
      "learning_rate": 6.793176179480288e-05,
      "loss": 1.9346,
      "step": 180066
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.3398691415786743,
      "learning_rate": 6.792916953673279e-05,
      "loss": 2.1417,
      "step": 180067
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.3497986793518066,
      "learning_rate": 6.792657732180924e-05,
      "loss": 2.2076,
      "step": 180068
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.317199945449829,
      "learning_rate": 6.792398515003235e-05,
      "loss": 2.19,
      "step": 180069
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.4200888872146606,
      "learning_rate": 6.792139302140283e-05,
      "loss": 2.0108,
      "step": 180070
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.5375009775161743,
      "learning_rate": 6.791880093592102e-05,
      "loss": 1.9826,
      "step": 180071
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.3528672456741333,
      "learning_rate": 6.791620889358761e-05,
      "loss": 1.993,
      "step": 180072
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.2520625591278076,
      "learning_rate": 6.791361689440283e-05,
      "loss": 1.9949,
      "step": 180073
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.2402609586715698,
      "learning_rate": 6.791102493836739e-05,
      "loss": 1.8158,
      "step": 180074
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.4039331674575806,
      "learning_rate": 6.790843302548167e-05,
      "loss": 2.028,
      "step": 180075
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.4047001600265503,
      "learning_rate": 6.790584115574617e-05,
      "loss": 1.9106,
      "step": 180076
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.2831604480743408,
      "learning_rate": 6.790324932916124e-05,
      "loss": 2.0497,
      "step": 180077
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.27181875705719,
      "learning_rate": 6.790065754572758e-05,
      "loss": 1.8911,
      "step": 180078
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.3543753623962402,
      "learning_rate": 6.789806580544549e-05,
      "loss": 2.087,
      "step": 180079
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.2529832124710083,
      "learning_rate": 6.789547410831562e-05,
      "loss": 1.9327,
      "step": 180080
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.3624281883239746,
      "learning_rate": 6.789288245433836e-05,
      "loss": 2.0973,
      "step": 180081
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.2708643674850464,
      "learning_rate": 6.789029084351418e-05,
      "loss": 2.0431,
      "step": 180082
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.4577842950820923,
      "learning_rate": 6.788769927584348e-05,
      "loss": 2.1054,
      "step": 180083
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.2806059122085571,
      "learning_rate": 6.788510775132694e-05,
      "loss": 1.8365,
      "step": 180084
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.505218267440796,
      "learning_rate": 6.788251626996487e-05,
      "loss": 2.2426,
      "step": 180085
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.4593528509140015,
      "learning_rate": 6.787992483175788e-05,
      "loss": 1.9703,
      "step": 180086
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.4092859029769897,
      "learning_rate": 6.78773334367064e-05,
      "loss": 2.0041,
      "step": 180087
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.3448662757873535,
      "learning_rate": 6.787474208481084e-05,
      "loss": 2.2489,
      "step": 180088
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.3516525030136108,
      "learning_rate": 6.787215077607182e-05,
      "loss": 2.0824,
      "step": 180089
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.3615727424621582,
      "learning_rate": 6.786955951048975e-05,
      "loss": 2.1296,
      "step": 180090
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.2820583581924438,
      "learning_rate": 6.786696828806502e-05,
      "loss": 1.9142,
      "step": 180091
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.5336377620697021,
      "learning_rate": 6.786437710879829e-05,
      "loss": 1.8837,
      "step": 180092
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.6615296602249146,
      "learning_rate": 6.786178597268995e-05,
      "loss": 2.2827,
      "step": 180093
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.4578605890274048,
      "learning_rate": 6.785919487974043e-05,
      "loss": 1.859,
      "step": 180094
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.3221389055252075,
      "learning_rate": 6.785660382995032e-05,
      "loss": 1.8991,
      "step": 180095
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.2373160123825073,
      "learning_rate": 6.785401282331998e-05,
      "loss": 2.0997,
      "step": 180096
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.4935804605484009,
      "learning_rate": 6.785142185985004e-05,
      "loss": 2.1054,
      "step": 180097
    },
    {
      "epoch": 2.34,
      "grad_norm": 1.4084328413009644,
      "learning_rate": 6.784883093954088e-05,
      "loss": 1.9914,
      "step": 180098
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.5389361381530762,
      "learning_rate": 6.784624006239305e-05,
      "loss": 1.7939,
      "step": 180099
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.5205433368682861,
      "learning_rate": 6.784364922840687e-05,
      "loss": 1.9123,
      "step": 180100
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.3614296913146973,
      "learning_rate": 6.784105843758302e-05,
      "loss": 1.9381,
      "step": 180101
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.6129096746444702,
      "learning_rate": 6.783846768992182e-05,
      "loss": 2.0639,
      "step": 180102
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.6481753587722778,
      "learning_rate": 6.783587698542393e-05,
      "loss": 1.913,
      "step": 180103
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.594475507736206,
      "learning_rate": 6.783328632408974e-05,
      "loss": 2.0123,
      "step": 180104
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.739608883857727,
      "learning_rate": 6.783069570591969e-05,
      "loss": 2.1291,
      "step": 180105
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.3477545976638794,
      "learning_rate": 6.782810513091424e-05,
      "loss": 1.849,
      "step": 180106
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.4133350849151611,
      "learning_rate": 6.7825514599074e-05,
      "loss": 1.941,
      "step": 180107
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.3521512746810913,
      "learning_rate": 6.78229241103993e-05,
      "loss": 2.0066,
      "step": 180108
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.7234488725662231,
      "learning_rate": 6.782033366489077e-05,
      "loss": 1.9641,
      "step": 180109
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.4533941745758057,
      "learning_rate": 6.781774326254878e-05,
      "loss": 1.9533,
      "step": 180110
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.338004231452942,
      "learning_rate": 6.781515290337398e-05,
      "loss": 1.9473,
      "step": 180111
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.3213284015655518,
      "learning_rate": 6.781256258736657e-05,
      "loss": 1.8696,
      "step": 180112
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.291464924812317,
      "learning_rate": 6.780997231452727e-05,
      "loss": 2.2605,
      "step": 180113
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.3807075023651123,
      "learning_rate": 6.780738208485639e-05,
      "loss": 2.1142,
      "step": 180114
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.4849249124526978,
      "learning_rate": 6.78047918983546e-05,
      "loss": 2.1583,
      "step": 180115
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.4384244680404663,
      "learning_rate": 6.780220175502219e-05,
      "loss": 2.0403,
      "step": 180116
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.3482073545455933,
      "learning_rate": 6.77996116548599e-05,
      "loss": 2.0956,
      "step": 180117
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.2230342626571655,
      "learning_rate": 6.779702159786789e-05,
      "loss": 1.9032,
      "step": 180118
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.3089773654937744,
      "learning_rate": 6.779443158404688e-05,
      "loss": 1.9905,
      "step": 180119
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.383634090423584,
      "learning_rate": 6.779184161339717e-05,
      "loss": 1.9702,
      "step": 180120
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.410749077796936,
      "learning_rate": 6.778925168591942e-05,
      "loss": 1.9789,
      "step": 180121
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.446619987487793,
      "learning_rate": 6.778666180161398e-05,
      "loss": 2.2575,
      "step": 180122
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.3023767471313477,
      "learning_rate": 6.778407196048152e-05,
      "loss": 1.899,
      "step": 180123
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.2841626405715942,
      "learning_rate": 6.778148216252221e-05,
      "loss": 2.0709,
      "step": 180124
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.5857123136520386,
      "learning_rate": 6.777889240773682e-05,
      "loss": 1.9634,
      "step": 180125
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.3383888006210327,
      "learning_rate": 6.777630269612563e-05,
      "loss": 2.3355,
      "step": 180126
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.345862865447998,
      "learning_rate": 6.777371302768926e-05,
      "loss": 1.9659,
      "step": 180127
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.2982063293457031,
      "learning_rate": 6.777112340242807e-05,
      "loss": 2.0121,
      "step": 180128
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.38102388381958,
      "learning_rate": 6.776853382034281e-05,
      "loss": 2.1934,
      "step": 180129
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.3175772428512573,
      "learning_rate": 6.776594428143358e-05,
      "loss": 2.0118,
      "step": 180130
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.2975934743881226,
      "learning_rate": 6.77633547857011e-05,
      "loss": 2.0594,
      "step": 180131
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.4027000665664673,
      "learning_rate": 6.776076533314575e-05,
      "loss": 2.0763,
      "step": 180132
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.159753441810608,
      "learning_rate": 6.77581759237681e-05,
      "loss": 1.9455,
      "step": 180133
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.272651195526123,
      "learning_rate": 6.775558655756854e-05,
      "loss": 1.8771,
      "step": 180134
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.3317289352416992,
      "learning_rate": 6.775299723454774e-05,
      "loss": 1.9907,
      "step": 180135
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.3136279582977295,
      "learning_rate": 6.775040795470591e-05,
      "loss": 2.0682,
      "step": 180136
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.5931682586669922,
      "learning_rate": 6.774781871804373e-05,
      "loss": 1.8416,
      "step": 180137
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.3461867570877075,
      "learning_rate": 6.774522952456152e-05,
      "loss": 1.9933,
      "step": 180138
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.1706936359405518,
      "learning_rate": 6.774264037425996e-05,
      "loss": 1.814,
      "step": 180139
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.451064109802246,
      "learning_rate": 6.774005126713932e-05,
      "loss": 1.98,
      "step": 180140
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.3642959594726562,
      "learning_rate": 6.773746220320039e-05,
      "loss": 2.1513,
      "step": 180141
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.262163758277893,
      "learning_rate": 6.773487318244323e-05,
      "loss": 2.2209,
      "step": 180142
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.3116662502288818,
      "learning_rate": 6.773228420486869e-05,
      "loss": 1.9525,
      "step": 180143
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.255285382270813,
      "learning_rate": 6.772969527047698e-05,
      "loss": 2.0232,
      "step": 180144
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.3125628232955933,
      "learning_rate": 6.772710637926881e-05,
      "loss": 1.9879,
      "step": 180145
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.4019218683242798,
      "learning_rate": 6.772451753124445e-05,
      "loss": 1.9469,
      "step": 180146
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.2242517471313477,
      "learning_rate": 6.772192872640466e-05,
      "loss": 2.036,
      "step": 180147
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.2882179021835327,
      "learning_rate": 6.771933996474957e-05,
      "loss": 2.0079,
      "step": 180148
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.3757258653640747,
      "learning_rate": 6.771675124627992e-05,
      "loss": 2.0441,
      "step": 180149
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.2429195642471313,
      "learning_rate": 6.771416257099605e-05,
      "loss": 1.7379,
      "step": 180150
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.2127968072891235,
      "learning_rate": 6.771157393889858e-05,
      "loss": 2.2006,
      "step": 180151
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.4134217500686646,
      "learning_rate": 6.770898534998782e-05,
      "loss": 2.1435,
      "step": 180152
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.3560683727264404,
      "learning_rate": 6.770639680426451e-05,
      "loss": 2.0566,
      "step": 180153
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.4535045623779297,
      "learning_rate": 6.77038083017288e-05,
      "loss": 2.3002,
      "step": 180154
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.210615634918213,
      "learning_rate": 6.770121984238144e-05,
      "loss": 2.0615,
      "step": 180155
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.2645457983016968,
      "learning_rate": 6.769863142622272e-05,
      "loss": 1.8506,
      "step": 180156
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.4026762247085571,
      "learning_rate": 6.769604305325327e-05,
      "loss": 1.8936,
      "step": 180157
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.2980679273605347,
      "learning_rate": 6.769345472347347e-05,
      "loss": 1.8997,
      "step": 180158
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.331101655960083,
      "learning_rate": 6.769086643688387e-05,
      "loss": 2.006,
      "step": 180159
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.2755730152130127,
      "learning_rate": 6.768827819348496e-05,
      "loss": 1.9811,
      "step": 180160
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.275092363357544,
      "learning_rate": 6.768568999327718e-05,
      "loss": 2.0424,
      "step": 180161
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.3901883363723755,
      "learning_rate": 6.768310183626091e-05,
      "loss": 1.9724,
      "step": 180162
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.2625788450241089,
      "learning_rate": 6.768051372243685e-05,
      "loss": 1.9783,
      "step": 180163
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.484684705734253,
      "learning_rate": 6.767792565180526e-05,
      "loss": 1.9499,
      "step": 180164
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.311087727546692,
      "learning_rate": 6.767533762436684e-05,
      "loss": 1.8329,
      "step": 180165
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.7584089040756226,
      "learning_rate": 6.767274964012192e-05,
      "loss": 2.1335,
      "step": 180166
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.2283803224563599,
      "learning_rate": 6.767016169907106e-05,
      "loss": 1.8836,
      "step": 180167
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.357288122177124,
      "learning_rate": 6.766757380121459e-05,
      "loss": 1.9063,
      "step": 180168
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.3281733989715576,
      "learning_rate": 6.76649859465532e-05,
      "loss": 1.8856,
      "step": 180169
    },
    {
      "epoch": 2.35,
      "grad_norm": 2.106612205505371,
      "learning_rate": 6.766239813508721e-05,
      "loss": 2.1054,
      "step": 180170
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.3003548383712769,
      "learning_rate": 6.765981036681721e-05,
      "loss": 2.0126,
      "step": 180171
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.4709898233413696,
      "learning_rate": 6.765722264174366e-05,
      "loss": 2.1109,
      "step": 180172
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.3476285934448242,
      "learning_rate": 6.765463495986695e-05,
      "loss": 1.9618,
      "step": 180173
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.466712236404419,
      "learning_rate": 6.765204732118768e-05,
      "loss": 2.0109,
      "step": 180174
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.2953153848648071,
      "learning_rate": 6.76494597257063e-05,
      "loss": 1.9803,
      "step": 180175
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.4100228548049927,
      "learning_rate": 6.76468721734232e-05,
      "loss": 1.8947,
      "step": 180176
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.3613998889923096,
      "learning_rate": 6.7644284664339e-05,
      "loss": 2.1556,
      "step": 180177
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.285050630569458,
      "learning_rate": 6.764169719845412e-05,
      "loss": 1.8824,
      "step": 180178
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.3925795555114746,
      "learning_rate": 6.763910977576894e-05,
      "loss": 1.764,
      "step": 180179
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.4016164541244507,
      "learning_rate": 6.763652239628411e-05,
      "loss": 2.045,
      "step": 180180
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.3371378183364868,
      "learning_rate": 6.763393505999995e-05,
      "loss": 1.9064,
      "step": 180181
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.2919812202453613,
      "learning_rate": 6.763134776691714e-05,
      "loss": 1.8788,
      "step": 180182
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.2670364379882812,
      "learning_rate": 6.762876051703604e-05,
      "loss": 1.9988,
      "step": 180183
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.4181452989578247,
      "learning_rate": 6.762617331035715e-05,
      "loss": 2.0231,
      "step": 180184
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.6475787162780762,
      "learning_rate": 6.762358614688083e-05,
      "loss": 2.0872,
      "step": 180185
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.483837366104126,
      "learning_rate": 6.762099902660778e-05,
      "loss": 2.1244,
      "step": 180186
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.4251688718795776,
      "learning_rate": 6.761841194953827e-05,
      "loss": 2.0609,
      "step": 180187
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.34059476852417,
      "learning_rate": 6.761582491567298e-05,
      "loss": 1.9106,
      "step": 180188
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.3688983917236328,
      "learning_rate": 6.761323792501228e-05,
      "loss": 2.2393,
      "step": 180189
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.5806925296783447,
      "learning_rate": 6.76106509775567e-05,
      "loss": 2.141,
      "step": 180190
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.4100524187088013,
      "learning_rate": 6.760806407330655e-05,
      "loss": 1.9757,
      "step": 180191
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.2288240194320679,
      "learning_rate": 6.760547721226259e-05,
      "loss": 2.1953,
      "step": 180192
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.3687784671783447,
      "learning_rate": 6.760289039442504e-05,
      "loss": 1.9393,
      "step": 180193
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.4326804876327515,
      "learning_rate": 6.760030361979458e-05,
      "loss": 2.2039,
      "step": 180194
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.3689935207366943,
      "learning_rate": 6.759771688837152e-05,
      "loss": 2.089,
      "step": 180195
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.2461880445480347,
      "learning_rate": 6.759513020015664e-05,
      "loss": 1.9159,
      "step": 180196
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.3258495330810547,
      "learning_rate": 6.759254355515001e-05,
      "loss": 2.2672,
      "step": 180197
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.3333901166915894,
      "learning_rate": 6.75899569533524e-05,
      "loss": 2.1445,
      "step": 180198
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.3434500694274902,
      "learning_rate": 6.758737039476412e-05,
      "loss": 1.8608,
      "step": 180199
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.3255013227462769,
      "learning_rate": 6.758478387938584e-05,
      "loss": 2.3053,
      "step": 180200
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.3456491231918335,
      "learning_rate": 6.758219740721783e-05,
      "loss": 1.8246,
      "step": 180201
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.368105411529541,
      "learning_rate": 6.757961097826086e-05,
      "loss": 1.9728,
      "step": 180202
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.3530774116516113,
      "learning_rate": 6.757702459251506e-05,
      "loss": 1.852,
      "step": 180203
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.3228543996810913,
      "learning_rate": 6.757443824998115e-05,
      "loss": 1.9723,
      "step": 180204
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.3949530124664307,
      "learning_rate": 6.757185195065944e-05,
      "loss": 1.9667,
      "step": 180205
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.3160288333892822,
      "learning_rate": 6.756926569455063e-05,
      "loss": 1.8792,
      "step": 180206
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.314249038696289,
      "learning_rate": 6.756667948165497e-05,
      "loss": 2.0382,
      "step": 180207
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.2849993705749512,
      "learning_rate": 6.756409331197323e-05,
      "loss": 1.9757,
      "step": 180208
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.4630694389343262,
      "learning_rate": 6.756150718550555e-05,
      "loss": 2.0716,
      "step": 180209
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.3372195959091187,
      "learning_rate": 6.755892110225264e-05,
      "loss": 1.9325,
      "step": 180210
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.3917285203933716,
      "learning_rate": 6.755633506221486e-05,
      "loss": 1.9858,
      "step": 180211
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.474223017692566,
      "learning_rate": 6.755374906539281e-05,
      "loss": 1.8828,
      "step": 180212
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.4523433446884155,
      "learning_rate": 6.75511631117868e-05,
      "loss": 2.0061,
      "step": 180213
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.39930260181427,
      "learning_rate": 6.754857720139761e-05,
      "loss": 2.0107,
      "step": 180214
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.2953486442565918,
      "learning_rate": 6.754599133422535e-05,
      "loss": 2.027,
      "step": 180215
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.1834543943405151,
      "learning_rate": 6.75434055102708e-05,
      "loss": 1.9777,
      "step": 180216
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.4651775360107422,
      "learning_rate": 6.75408197295342e-05,
      "loss": 2.0074,
      "step": 180217
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.2916202545166016,
      "learning_rate": 6.753823399201624e-05,
      "loss": 1.9893,
      "step": 180218
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.3743345737457275,
      "learning_rate": 6.753564829771722e-05,
      "loss": 1.8965,
      "step": 180219
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.392218828201294,
      "learning_rate": 6.753306264663788e-05,
      "loss": 1.9794,
      "step": 180220
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.3292980194091797,
      "learning_rate": 6.753047703877836e-05,
      "loss": 2.1045,
      "step": 180221
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.2270798683166504,
      "learning_rate": 6.75278914741394e-05,
      "loss": 2.1808,
      "step": 180222
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.4629368782043457,
      "learning_rate": 6.75253059527213e-05,
      "loss": 1.9711,
      "step": 180223
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.4401816129684448,
      "learning_rate": 6.752272047452473e-05,
      "loss": 1.7681,
      "step": 180224
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.525713562965393,
      "learning_rate": 6.752013503955002e-05,
      "loss": 1.9055,
      "step": 180225
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.3508691787719727,
      "learning_rate": 6.751754964779775e-05,
      "loss": 2.1101,
      "step": 180226
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.4613888263702393,
      "learning_rate": 6.751496429926838e-05,
      "loss": 1.8654,
      "step": 180227
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.6409547328948975,
      "learning_rate": 6.751237899396235e-05,
      "loss": 2.0359,
      "step": 180228
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.2665009498596191,
      "learning_rate": 6.750979373188007e-05,
      "loss": 1.975,
      "step": 180229
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.4059230089187622,
      "learning_rate": 6.750720851302219e-05,
      "loss": 1.9944,
      "step": 180230
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.3665415048599243,
      "learning_rate": 6.750462333738905e-05,
      "loss": 2.3007,
      "step": 180231
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.246322512626648,
      "learning_rate": 6.750203820498124e-05,
      "loss": 1.9644,
      "step": 180232
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.2464268207550049,
      "learning_rate": 6.749945311579923e-05,
      "loss": 1.9437,
      "step": 180233
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.2975761890411377,
      "learning_rate": 6.749686806984343e-05,
      "loss": 2.1253,
      "step": 180234
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.2466055154800415,
      "learning_rate": 6.749428306711426e-05,
      "loss": 1.8763,
      "step": 180235
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.2642040252685547,
      "learning_rate": 6.749169810761238e-05,
      "loss": 2.0475,
      "step": 180236
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.224773645401001,
      "learning_rate": 6.748911319133812e-05,
      "loss": 2.0825,
      "step": 180237
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.4143717288970947,
      "learning_rate": 6.748652831829208e-05,
      "loss": 2.2286,
      "step": 180238
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.375456690788269,
      "learning_rate": 6.748394348847473e-05,
      "loss": 1.9489,
      "step": 180239
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.3996660709381104,
      "learning_rate": 6.748135870188646e-05,
      "loss": 1.9492,
      "step": 180240
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.232107162475586,
      "learning_rate": 6.747877395852771e-05,
      "loss": 1.9066,
      "step": 180241
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.3582727909088135,
      "learning_rate": 6.747618925839914e-05,
      "loss": 2.2691,
      "step": 180242
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.3287400007247925,
      "learning_rate": 6.747360460150108e-05,
      "loss": 1.9533,
      "step": 180243
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.3546841144561768,
      "learning_rate": 6.747101998783412e-05,
      "loss": 2.1723,
      "step": 180244
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.4882895946502686,
      "learning_rate": 6.746843541739868e-05,
      "loss": 2.0529,
      "step": 180245
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.3143703937530518,
      "learning_rate": 6.746585089019526e-05,
      "loss": 2.0015,
      "step": 180246
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.3171440362930298,
      "learning_rate": 6.746326640622423e-05,
      "loss": 1.9545,
      "step": 180247
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.263610601425171,
      "learning_rate": 6.746068196548627e-05,
      "loss": 2.1332,
      "step": 180248
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.3893191814422607,
      "learning_rate": 6.745809756798168e-05,
      "loss": 2.0626,
      "step": 180249
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.5443427562713623,
      "learning_rate": 6.745551321371108e-05,
      "loss": 2.1285,
      "step": 180250
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.3677457571029663,
      "learning_rate": 6.745292890267489e-05,
      "loss": 2.0061,
      "step": 180251
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.2676420211791992,
      "learning_rate": 6.74503446348736e-05,
      "loss": 2.0279,
      "step": 180252
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.4250682592391968,
      "learning_rate": 6.744776041030759e-05,
      "loss": 1.9618,
      "step": 180253
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.384925127029419,
      "learning_rate": 6.744517622897754e-05,
      "loss": 1.9962,
      "step": 180254
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.4432752132415771,
      "learning_rate": 6.744259209088372e-05,
      "loss": 2.0495,
      "step": 180255
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.4010045528411865,
      "learning_rate": 6.74400079960268e-05,
      "loss": 2.0062,
      "step": 180256
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.221956729888916,
      "learning_rate": 6.743742394440715e-05,
      "loss": 1.9572,
      "step": 180257
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.2945786714553833,
      "learning_rate": 6.743483993602522e-05,
      "loss": 1.8561,
      "step": 180258
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.317046880722046,
      "learning_rate": 6.74322559708816e-05,
      "loss": 2.1559,
      "step": 180259
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.1989353895187378,
      "learning_rate": 6.742967204897672e-05,
      "loss": 2.1853,
      "step": 180260
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.4104269742965698,
      "learning_rate": 6.7427088170311e-05,
      "loss": 1.9,
      "step": 180261
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.397709846496582,
      "learning_rate": 6.742450433488505e-05,
      "loss": 1.9629,
      "step": 180262
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.2643146514892578,
      "learning_rate": 6.742192054269924e-05,
      "loss": 2.0263,
      "step": 180263
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.6127490997314453,
      "learning_rate": 6.741933679375404e-05,
      "loss": 2.1138,
      "step": 180264
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.3063822984695435,
      "learning_rate": 6.741675308805003e-05,
      "loss": 1.9259,
      "step": 180265
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.6080631017684937,
      "learning_rate": 6.741416942558764e-05,
      "loss": 2.0253,
      "step": 180266
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.2486273050308228,
      "learning_rate": 6.741158580636729e-05,
      "loss": 2.0171,
      "step": 180267
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.3837615251541138,
      "learning_rate": 6.740900223038957e-05,
      "loss": 2.1102,
      "step": 180268
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.451678991317749,
      "learning_rate": 6.740641869765491e-05,
      "loss": 2.0594,
      "step": 180269
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.298383116722107,
      "learning_rate": 6.740383520816372e-05,
      "loss": 1.9974,
      "step": 180270
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.2103731632232666,
      "learning_rate": 6.740125176191661e-05,
      "loss": 2.1377,
      "step": 180271
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.3055351972579956,
      "learning_rate": 6.739866835891393e-05,
      "loss": 2.0822,
      "step": 180272
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.463870644569397,
      "learning_rate": 6.739608499915632e-05,
      "loss": 2.1926,
      "step": 180273
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.241349220275879,
      "learning_rate": 6.739350168264414e-05,
      "loss": 2.2942,
      "step": 180274
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.3060250282287598,
      "learning_rate": 6.739091840937794e-05,
      "loss": 1.9794,
      "step": 180275
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.3617982864379883,
      "learning_rate": 6.738833517935805e-05,
      "loss": 2.1277,
      "step": 180276
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.4200327396392822,
      "learning_rate": 6.738575199258514e-05,
      "loss": 1.9464,
      "step": 180277
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.372502088546753,
      "learning_rate": 6.738316884905952e-05,
      "loss": 2.2496,
      "step": 180278
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.2198652029037476,
      "learning_rate": 6.738058574878188e-05,
      "loss": 1.8964,
      "step": 180279
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.4511770009994507,
      "learning_rate": 6.737800269175248e-05,
      "loss": 1.9735,
      "step": 180280
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.3240416049957275,
      "learning_rate": 6.737541967797205e-05,
      "loss": 1.8112,
      "step": 180281
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.3387261629104614,
      "learning_rate": 6.737283670744075e-05,
      "loss": 1.9512,
      "step": 180282
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.4289226531982422,
      "learning_rate": 6.737025378015932e-05,
      "loss": 2.0217,
      "step": 180283
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.2585418224334717,
      "learning_rate": 6.736767089612806e-05,
      "loss": 1.9184,
      "step": 180284
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.4299383163452148,
      "learning_rate": 6.736508805534764e-05,
      "loss": 1.8787,
      "step": 180285
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.4842257499694824,
      "learning_rate": 6.736250525781837e-05,
      "loss": 2.2284,
      "step": 180286
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.3772717714309692,
      "learning_rate": 6.735992250354095e-05,
      "loss": 1.9406,
      "step": 180287
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.2485414743423462,
      "learning_rate": 6.735733979251558e-05,
      "loss": 1.9753,
      "step": 180288
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.4351166486740112,
      "learning_rate": 6.73547571247429e-05,
      "loss": 2.0661,
      "step": 180289
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.1993892192840576,
      "learning_rate": 6.735217450022333e-05,
      "loss": 1.8013,
      "step": 180290
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.3148014545440674,
      "learning_rate": 6.73495919189574e-05,
      "loss": 2.2616,
      "step": 180291
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.3709840774536133,
      "learning_rate": 6.734700938094557e-05,
      "loss": 2.0955,
      "step": 180292
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.5073434114456177,
      "learning_rate": 6.734442688618837e-05,
      "loss": 2.2078,
      "step": 180293
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.432976245880127,
      "learning_rate": 6.734184443468622e-05,
      "loss": 1.9602,
      "step": 180294
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.3224525451660156,
      "learning_rate": 6.733926202643961e-05,
      "loss": 2.0777,
      "step": 180295
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.3194189071655273,
      "learning_rate": 6.733667966144897e-05,
      "loss": 2.0451,
      "step": 180296
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.3772825002670288,
      "learning_rate": 6.733409733971489e-05,
      "loss": 1.9,
      "step": 180297
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.3799985647201538,
      "learning_rate": 6.73315150612377e-05,
      "loss": 2.054,
      "step": 180298
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.4288095235824585,
      "learning_rate": 6.732893282601809e-05,
      "loss": 1.9895,
      "step": 180299
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.3859033584594727,
      "learning_rate": 6.73263506340564e-05,
      "loss": 2.0295,
      "step": 180300
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.4401942491531372,
      "learning_rate": 6.732376848535314e-05,
      "loss": 2.0973,
      "step": 180301
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.4461774826049805,
      "learning_rate": 6.732118637990869e-05,
      "loss": 1.9056,
      "step": 180302
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.3055884838104248,
      "learning_rate": 6.731860431772372e-05,
      "loss": 1.9716,
      "step": 180303
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.8626830577850342,
      "learning_rate": 6.731602229879853e-05,
      "loss": 2.1381,
      "step": 180304
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.8185415267944336,
      "learning_rate": 6.731344032313379e-05,
      "loss": 1.8005,
      "step": 180305
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.385851502418518,
      "learning_rate": 6.731085839072984e-05,
      "loss": 2.0429,
      "step": 180306
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.4565752744674683,
      "learning_rate": 6.730827650158718e-05,
      "loss": 2.0577,
      "step": 180307
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.3764702081680298,
      "learning_rate": 6.730569465570627e-05,
      "loss": 1.9889,
      "step": 180308
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.304539680480957,
      "learning_rate": 6.730311285308768e-05,
      "loss": 2.2718,
      "step": 180309
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.5043359994888306,
      "learning_rate": 6.730053109373173e-05,
      "loss": 1.8576,
      "step": 180310
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.2770671844482422,
      "learning_rate": 6.729794937763913e-05,
      "loss": 2.1454,
      "step": 180311
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.4085665941238403,
      "learning_rate": 6.729536770481022e-05,
      "loss": 2.1285,
      "step": 180312
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.443091630935669,
      "learning_rate": 6.729278607524546e-05,
      "loss": 2.1369,
      "step": 180313
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.3343905210494995,
      "learning_rate": 6.729020448894532e-05,
      "loss": 1.9668,
      "step": 180314
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.3814862966537476,
      "learning_rate": 6.728762294591039e-05,
      "loss": 1.8985,
      "step": 180315
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.4282481670379639,
      "learning_rate": 6.7285041446141e-05,
      "loss": 2.0071,
      "step": 180316
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.322698712348938,
      "learning_rate": 6.728245998963781e-05,
      "loss": 1.796,
      "step": 180317
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.3983465433120728,
      "learning_rate": 6.727987857640117e-05,
      "loss": 1.9722,
      "step": 180318
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.4967339038848877,
      "learning_rate": 6.727729720643162e-05,
      "loss": 2.0634,
      "step": 180319
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.2142059803009033,
      "learning_rate": 6.727471587972954e-05,
      "loss": 2.0738,
      "step": 180320
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.289288878440857,
      "learning_rate": 6.727213459629554e-05,
      "loss": 1.8321,
      "step": 180321
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.481406807899475,
      "learning_rate": 6.726955335613e-05,
      "loss": 1.8913,
      "step": 180322
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.4287221431732178,
      "learning_rate": 6.726697215923346e-05,
      "loss": 2.1254,
      "step": 180323
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.4271926879882812,
      "learning_rate": 6.726439100560643e-05,
      "loss": 2.1344,
      "step": 180324
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.4693689346313477,
      "learning_rate": 6.726180989524935e-05,
      "loss": 2.2552,
      "step": 180325
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.36963951587677,
      "learning_rate": 6.72592288281626e-05,
      "loss": 1.8372,
      "step": 180326
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.3636360168457031,
      "learning_rate": 6.725664780434682e-05,
      "loss": 2.0739,
      "step": 180327
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.303647518157959,
      "learning_rate": 6.725406682380237e-05,
      "loss": 1.982,
      "step": 180328
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.399916648864746,
      "learning_rate": 6.725148588652983e-05,
      "loss": 1.9164,
      "step": 180329
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.2911206483840942,
      "learning_rate": 6.724890499252965e-05,
      "loss": 2.1796,
      "step": 180330
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.4078274965286255,
      "learning_rate": 6.724632414180229e-05,
      "loss": 2.2256,
      "step": 180331
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.3865114450454712,
      "learning_rate": 6.724374333434814e-05,
      "loss": 1.9311,
      "step": 180332
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.3752739429473877,
      "learning_rate": 6.724116257016789e-05,
      "loss": 1.8389,
      "step": 180333
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.3840317726135254,
      "learning_rate": 6.723858184926177e-05,
      "loss": 1.9387,
      "step": 180334
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.341654896736145,
      "learning_rate": 6.723600117163052e-05,
      "loss": 2.2294,
      "step": 180335
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.290747046470642,
      "learning_rate": 6.723342053727448e-05,
      "loss": 1.9096,
      "step": 180336
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.2720621824264526,
      "learning_rate": 6.723083994619412e-05,
      "loss": 2.455,
      "step": 180337
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.3434898853302002,
      "learning_rate": 6.722825939838987e-05,
      "loss": 1.9697,
      "step": 180338
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.2603498697280884,
      "learning_rate": 6.722567889386235e-05,
      "loss": 1.769,
      "step": 180339
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.341673731803894,
      "learning_rate": 6.72230984326119e-05,
      "loss": 1.8888,
      "step": 180340
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.3042958974838257,
      "learning_rate": 6.722051801463913e-05,
      "loss": 1.9432,
      "step": 180341
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.3781912326812744,
      "learning_rate": 6.721793763994449e-05,
      "loss": 2.0023,
      "step": 180342
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.2986974716186523,
      "learning_rate": 6.721535730852835e-05,
      "loss": 2.2011,
      "step": 180343
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.3611193895339966,
      "learning_rate": 6.721277702039133e-05,
      "loss": 1.82,
      "step": 180344
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.4452638626098633,
      "learning_rate": 6.721019677553386e-05,
      "loss": 2.0376,
      "step": 180345
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.1321187019348145,
      "learning_rate": 6.720761657395633e-05,
      "loss": 1.7839,
      "step": 180346
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.3754637241363525,
      "learning_rate": 6.720503641565938e-05,
      "loss": 2.0702,
      "step": 180347
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.321183443069458,
      "learning_rate": 6.72024563006434e-05,
      "loss": 2.1213,
      "step": 180348
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.410474419593811,
      "learning_rate": 6.719987622890881e-05,
      "loss": 1.7348,
      "step": 180349
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.4009168148040771,
      "learning_rate": 6.719729620045623e-05,
      "loss": 2.0984,
      "step": 180350
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.224591851234436,
      "learning_rate": 6.719471621528606e-05,
      "loss": 1.865,
      "step": 180351
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.2629265785217285,
      "learning_rate": 6.71921362733987e-05,
      "loss": 1.965,
      "step": 180352
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.166706919670105,
      "learning_rate": 6.718955637479481e-05,
      "loss": 2.0746,
      "step": 180353
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.3064064979553223,
      "learning_rate": 6.718697651947478e-05,
      "loss": 2.0966,
      "step": 180354
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.380919337272644,
      "learning_rate": 6.718439670743898e-05,
      "loss": 2.1287,
      "step": 180355
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.2998441457748413,
      "learning_rate": 6.718181693868813e-05,
      "loss": 2.0333,
      "step": 180356
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.4146989583969116,
      "learning_rate": 6.717923721322244e-05,
      "loss": 2.2077,
      "step": 180357
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.4889047145843506,
      "learning_rate": 6.717665753104262e-05,
      "loss": 1.92,
      "step": 180358
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.3720675706863403,
      "learning_rate": 6.717407789214907e-05,
      "loss": 2.0169,
      "step": 180359
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.4259532690048218,
      "learning_rate": 6.717149829654216e-05,
      "loss": 2.2006,
      "step": 180360
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.4344300031661987,
      "learning_rate": 6.716891874422255e-05,
      "loss": 2.0501,
      "step": 180361
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.3885208368301392,
      "learning_rate": 6.716633923519064e-05,
      "loss": 2.0308,
      "step": 180362
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.3213692903518677,
      "learning_rate": 6.71637597694468e-05,
      "loss": 2.413,
      "step": 180363
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.357003092765808,
      "learning_rate": 6.716118034699171e-05,
      "loss": 1.9205,
      "step": 180364
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.3597421646118164,
      "learning_rate": 6.715860096782575e-05,
      "loss": 1.8904,
      "step": 180365
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.366390585899353,
      "learning_rate": 6.71560216319493e-05,
      "loss": 2.0211,
      "step": 180366
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.2151095867156982,
      "learning_rate": 6.715344233936305e-05,
      "loss": 2.0634,
      "step": 180367
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.3836255073547363,
      "learning_rate": 6.715086309006734e-05,
      "loss": 2.2037,
      "step": 180368
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.2456018924713135,
      "learning_rate": 6.714828388406263e-05,
      "loss": 2.0137,
      "step": 180369
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.331644892692566,
      "learning_rate": 6.714570472134954e-05,
      "loss": 2.0664,
      "step": 180370
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.4023958444595337,
      "learning_rate": 6.714312560192834e-05,
      "loss": 2.0043,
      "step": 180371
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.2514957189559937,
      "learning_rate": 6.714054652579971e-05,
      "loss": 1.6676,
      "step": 180372
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.2602204084396362,
      "learning_rate": 6.713796749296408e-05,
      "loss": 2.1611,
      "step": 180373
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.3213015794754028,
      "learning_rate": 6.71353885034219e-05,
      "loss": 1.8236,
      "step": 180374
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.3276337385177612,
      "learning_rate": 6.713280955717354e-05,
      "loss": 2.1567,
      "step": 180375
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.3658816814422607,
      "learning_rate": 6.713023065421966e-05,
      "loss": 1.9888,
      "step": 180376
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.4321850538253784,
      "learning_rate": 6.71276517945606e-05,
      "loss": 1.9962,
      "step": 180377
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.3565043210983276,
      "learning_rate": 6.712507297819698e-05,
      "loss": 2.0341,
      "step": 180378
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.4035565853118896,
      "learning_rate": 6.712249420512922e-05,
      "loss": 2.018,
      "step": 180379
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.3389837741851807,
      "learning_rate": 6.711991547535777e-05,
      "loss": 1.9711,
      "step": 180380
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.44123375415802,
      "learning_rate": 6.711733678888305e-05,
      "loss": 1.9837,
      "step": 180381
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.2841740846633911,
      "learning_rate": 6.71147581457057e-05,
      "loss": 1.9303,
      "step": 180382
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.3300966024398804,
      "learning_rate": 6.711217954582602e-05,
      "loss": 2.0501,
      "step": 180383
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.291114330291748,
      "learning_rate": 6.710960098924465e-05,
      "loss": 1.8673,
      "step": 180384
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.293809413909912,
      "learning_rate": 6.710702247596203e-05,
      "loss": 2.0611,
      "step": 180385
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.2741923332214355,
      "learning_rate": 6.71044440059786e-05,
      "loss": 2.0996,
      "step": 180386
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.293544054031372,
      "learning_rate": 6.710186557929478e-05,
      "loss": 2.0411,
      "step": 180387
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.2584137916564941,
      "learning_rate": 6.709928719591118e-05,
      "loss": 1.9589,
      "step": 180388
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.278300404548645,
      "learning_rate": 6.709670885582814e-05,
      "loss": 2.0125,
      "step": 180389
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.3453460931777954,
      "learning_rate": 6.70941305590463e-05,
      "loss": 1.8731,
      "step": 180390
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.3033664226531982,
      "learning_rate": 6.709155230556605e-05,
      "loss": 1.9935,
      "step": 180391
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.3023754358291626,
      "learning_rate": 6.708897409538791e-05,
      "loss": 2.1337,
      "step": 180392
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.29462468624115,
      "learning_rate": 6.708639592851221e-05,
      "loss": 2.0251,
      "step": 180393
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.3108887672424316,
      "learning_rate": 6.708381780493966e-05,
      "loss": 1.8267,
      "step": 180394
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.4410873651504517,
      "learning_rate": 6.708123972467052e-05,
      "loss": 2.2388,
      "step": 180395
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.3573763370513916,
      "learning_rate": 6.707866168770545e-05,
      "loss": 2.0962,
      "step": 180396
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.2709765434265137,
      "learning_rate": 6.707608369404486e-05,
      "loss": 2.0254,
      "step": 180397
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.3340559005737305,
      "learning_rate": 6.707350574368925e-05,
      "loss": 1.7734,
      "step": 180398
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.3287142515182495,
      "learning_rate": 6.707092783663896e-05,
      "loss": 1.9299,
      "step": 180399
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.2469160556793213,
      "learning_rate": 6.706834997289465e-05,
      "loss": 1.9508,
      "step": 180400
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.378400444984436,
      "learning_rate": 6.706577215245668e-05,
      "loss": 2.0665,
      "step": 180401
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.4208097457885742,
      "learning_rate": 6.706319437532565e-05,
      "loss": 1.9988,
      "step": 180402
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.3989207744598389,
      "learning_rate": 6.706061664150196e-05,
      "loss": 1.9764,
      "step": 180403
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.3411728143692017,
      "learning_rate": 6.705803895098612e-05,
      "loss": 2.1589,
      "step": 180404
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.2481849193572998,
      "learning_rate": 6.705546130377852e-05,
      "loss": 2.2323,
      "step": 180405
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.3324201107025146,
      "learning_rate": 6.705288369987976e-05,
      "loss": 1.9816,
      "step": 180406
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.3072292804718018,
      "learning_rate": 6.705030613929015e-05,
      "loss": 1.9843,
      "step": 180407
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.5037798881530762,
      "learning_rate": 6.70477286220104e-05,
      "loss": 2.1504,
      "step": 180408
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.401850700378418,
      "learning_rate": 6.70451511480409e-05,
      "loss": 2.3345,
      "step": 180409
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.3287510871887207,
      "learning_rate": 6.704257371738209e-05,
      "loss": 1.871,
      "step": 180410
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.387499213218689,
      "learning_rate": 6.703999633003438e-05,
      "loss": 1.8096,
      "step": 180411
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.3553900718688965,
      "learning_rate": 6.703741898599838e-05,
      "loss": 1.8494,
      "step": 180412
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.2455708980560303,
      "learning_rate": 6.703484168527448e-05,
      "loss": 1.9145,
      "step": 180413
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.3140877485275269,
      "learning_rate": 6.703226442786326e-05,
      "loss": 2.0908,
      "step": 180414
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.349331021308899,
      "learning_rate": 6.702968721376515e-05,
      "loss": 2.3752,
      "step": 180415
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.3238496780395508,
      "learning_rate": 6.702711004298065e-05,
      "loss": 1.7572,
      "step": 180416
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.3957408666610718,
      "learning_rate": 6.702453291551009e-05,
      "loss": 1.8233,
      "step": 180417
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.465563178062439,
      "learning_rate": 6.702195583135415e-05,
      "loss": 1.8641,
      "step": 180418
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.465287685394287,
      "learning_rate": 6.701937879051312e-05,
      "loss": 1.84,
      "step": 180419
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.3495416641235352,
      "learning_rate": 6.701680179298771e-05,
      "loss": 1.8388,
      "step": 180420
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.397624135017395,
      "learning_rate": 6.701422483877826e-05,
      "loss": 2.0271,
      "step": 180421
    },
    {
      "epoch": 2.35,
      "grad_norm": 2.157174825668335,
      "learning_rate": 6.701164792788525e-05,
      "loss": 1.9578,
      "step": 180422
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.533450961112976,
      "learning_rate": 6.700907106030911e-05,
      "loss": 2.1674,
      "step": 180423
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.679205060005188,
      "learning_rate": 6.700649423605047e-05,
      "loss": 2.1214,
      "step": 180424
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.6124449968338013,
      "learning_rate": 6.700391745510961e-05,
      "loss": 1.9576,
      "step": 180425
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.2345129251480103,
      "learning_rate": 6.700134071748724e-05,
      "loss": 2.058,
      "step": 180426
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.2116565704345703,
      "learning_rate": 6.699876402318361e-05,
      "loss": 1.8116,
      "step": 180427
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.285832166671753,
      "learning_rate": 6.699618737219947e-05,
      "loss": 1.9108,
      "step": 180428
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.357024073600769,
      "learning_rate": 6.699361076453498e-05,
      "loss": 1.9956,
      "step": 180429
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.547420620918274,
      "learning_rate": 6.699103420019087e-05,
      "loss": 2.0478,
      "step": 180430
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.4035885334014893,
      "learning_rate": 6.698845767916742e-05,
      "loss": 2.2895,
      "step": 180431
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.3686895370483398,
      "learning_rate": 6.698588120146531e-05,
      "loss": 1.9042,
      "step": 180432
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.4142229557037354,
      "learning_rate": 6.698330476708484e-05,
      "loss": 1.8193,
      "step": 180433
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.2908954620361328,
      "learning_rate": 6.698072837602663e-05,
      "loss": 2.0564,
      "step": 180434
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.6821256875991821,
      "learning_rate": 6.697815202829116e-05,
      "loss": 1.7815,
      "step": 180435
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.708572268486023,
      "learning_rate": 6.697557572387881e-05,
      "loss": 1.9417,
      "step": 180436
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.5105552673339844,
      "learning_rate": 6.697299946279002e-05,
      "loss": 2.0713,
      "step": 180437
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.2826331853866577,
      "learning_rate": 6.69704232450254e-05,
      "loss": 1.9552,
      "step": 180438
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.3396408557891846,
      "learning_rate": 6.696784707058533e-05,
      "loss": 2.0388,
      "step": 180439
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.7923001050949097,
      "learning_rate": 6.696527093947044e-05,
      "loss": 2.0827,
      "step": 180440
    },
    {
      "epoch": 2.35,
      "grad_norm": 2.0638091564178467,
      "learning_rate": 6.696269485168109e-05,
      "loss": 2.1285,
      "step": 180441
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.448477864265442,
      "learning_rate": 6.696011880721768e-05,
      "loss": 2.028,
      "step": 180442
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.3923335075378418,
      "learning_rate": 6.695754280608085e-05,
      "loss": 2.1823,
      "step": 180443
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.380370855331421,
      "learning_rate": 6.695496684827102e-05,
      "loss": 2.0011,
      "step": 180444
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.6101748943328857,
      "learning_rate": 6.69523909337886e-05,
      "loss": 2.1208,
      "step": 180445
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.4564827680587769,
      "learning_rate": 6.694981506263419e-05,
      "loss": 1.8779,
      "step": 180446
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.517210602760315,
      "learning_rate": 6.694723923480823e-05,
      "loss": 1.991,
      "step": 180447
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.327397346496582,
      "learning_rate": 6.694466345031106e-05,
      "loss": 1.8782,
      "step": 180448
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.3604592084884644,
      "learning_rate": 6.694208770914341e-05,
      "loss": 1.8978,
      "step": 180449
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.2623264789581299,
      "learning_rate": 6.693951201130559e-05,
      "loss": 1.9028,
      "step": 180450
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.2980281114578247,
      "learning_rate": 6.693693635679802e-05,
      "loss": 1.9561,
      "step": 180451
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.5382661819458008,
      "learning_rate": 6.693436074562138e-05,
      "loss": 2.154,
      "step": 180452
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.4896210432052612,
      "learning_rate": 6.693178517777605e-05,
      "loss": 1.7592,
      "step": 180453
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.4132039546966553,
      "learning_rate": 6.69292096532624e-05,
      "loss": 1.9679,
      "step": 180454
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.3213506937026978,
      "learning_rate": 6.692663417208108e-05,
      "loss": 2.0377,
      "step": 180455
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.4853219985961914,
      "learning_rate": 6.692405873423243e-05,
      "loss": 1.9373,
      "step": 180456
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.5832290649414062,
      "learning_rate": 6.692148333971708e-05,
      "loss": 1.7415,
      "step": 180457
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.7774537801742554,
      "learning_rate": 6.691890798853543e-05,
      "loss": 2.1506,
      "step": 180458
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.391050934791565,
      "learning_rate": 6.691633268068795e-05,
      "loss": 2.3858,
      "step": 180459
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.421239972114563,
      "learning_rate": 6.691375741617505e-05,
      "loss": 2.2026,
      "step": 180460
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.7058027982711792,
      "learning_rate": 6.691118219499735e-05,
      "loss": 2.1873,
      "step": 180461
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.3806638717651367,
      "learning_rate": 6.690860701715519e-05,
      "loss": 2.0538,
      "step": 180462
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.4051566123962402,
      "learning_rate": 6.690603188264922e-05,
      "loss": 1.9304,
      "step": 180463
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.2885043621063232,
      "learning_rate": 6.690345679147978e-05,
      "loss": 1.8638,
      "step": 180464
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.2757587432861328,
      "learning_rate": 6.690088174364742e-05,
      "loss": 2.1643,
      "step": 180465
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.4438711404800415,
      "learning_rate": 6.689830673915245e-05,
      "loss": 1.9495,
      "step": 180466
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.3938692808151245,
      "learning_rate": 6.689573177799562e-05,
      "loss": 1.9779,
      "step": 180467
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.3889068365097046,
      "learning_rate": 6.689315686017719e-05,
      "loss": 2.2731,
      "step": 180468
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.274810552597046,
      "learning_rate": 6.689058198569778e-05,
      "loss": 2.034,
      "step": 180469
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.3459190130233765,
      "learning_rate": 6.688800715455782e-05,
      "loss": 1.8751,
      "step": 180470
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.3147815465927124,
      "learning_rate": 6.688543236675779e-05,
      "loss": 2.0613,
      "step": 180471
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.2240872383117676,
      "learning_rate": 6.688285762229805e-05,
      "loss": 1.9903,
      "step": 180472
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.2892228364944458,
      "learning_rate": 6.68802829211793e-05,
      "loss": 2.1494,
      "step": 180473
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.4073699712753296,
      "learning_rate": 6.68777082634018e-05,
      "loss": 1.9468,
      "step": 180474
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.2577664852142334,
      "learning_rate": 6.687513364896626e-05,
      "loss": 1.9261,
      "step": 180475
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.3174527883529663,
      "learning_rate": 6.687255907787297e-05,
      "loss": 2.1408,
      "step": 180476
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.3885250091552734,
      "learning_rate": 6.686998455012251e-05,
      "loss": 2.074,
      "step": 180477
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.3763391971588135,
      "learning_rate": 6.686741006571525e-05,
      "loss": 2.0508,
      "step": 180478
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.1783396005630493,
      "learning_rate": 6.686483562465179e-05,
      "loss": 2.0214,
      "step": 180479
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.3733102083206177,
      "learning_rate": 6.686226122693248e-05,
      "loss": 2.0334,
      "step": 180480
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.3443503379821777,
      "learning_rate": 6.685968687255798e-05,
      "loss": 2.0058,
      "step": 180481
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.2275620698928833,
      "learning_rate": 6.685711256152863e-05,
      "loss": 2.1089,
      "step": 180482
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.4025237560272217,
      "learning_rate": 6.685453829384498e-05,
      "loss": 1.945,
      "step": 180483
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.2583913803100586,
      "learning_rate": 6.685196406950739e-05,
      "loss": 2.1214,
      "step": 180484
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.403975486755371,
      "learning_rate": 6.684938988851649e-05,
      "loss": 2.0858,
      "step": 180485
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.3285722732543945,
      "learning_rate": 6.684681575087259e-05,
      "loss": 2.2011,
      "step": 180486
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.354388952255249,
      "learning_rate": 6.684424165657637e-05,
      "loss": 2.0349,
      "step": 180487
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.251841425895691,
      "learning_rate": 6.684166760562822e-05,
      "loss": 1.8566,
      "step": 180488
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.4122802019119263,
      "learning_rate": 6.683909359802858e-05,
      "loss": 1.9152,
      "step": 180489
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.3865710496902466,
      "learning_rate": 6.683651963377787e-05,
      "loss": 2.2476,
      "step": 180490
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.3561967611312866,
      "learning_rate": 6.683394571287676e-05,
      "loss": 2.3668,
      "step": 180491
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.198868751525879,
      "learning_rate": 6.683137183532553e-05,
      "loss": 1.8845,
      "step": 180492
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.2761536836624146,
      "learning_rate": 6.682879800112483e-05,
      "loss": 2.1164,
      "step": 180493
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.2669094800949097,
      "learning_rate": 6.682622421027495e-05,
      "loss": 1.839,
      "step": 180494
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.405287504196167,
      "learning_rate": 6.682365046277665e-05,
      "loss": 1.8651,
      "step": 180495
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.4125988483428955,
      "learning_rate": 6.682107675863008e-05,
      "loss": 2.0981,
      "step": 180496
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.3037134408950806,
      "learning_rate": 6.681850309783597e-05,
      "loss": 1.9973,
      "step": 180497
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.266899824142456,
      "learning_rate": 6.681592948039462e-05,
      "loss": 2.109,
      "step": 180498
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.3593807220458984,
      "learning_rate": 6.681335590630665e-05,
      "loss": 1.917,
      "step": 180499
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.290705919265747,
      "learning_rate": 6.681078237557239e-05,
      "loss": 2.2022,
      "step": 180500
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.4237024784088135,
      "learning_rate": 6.680820888819259e-05,
      "loss": 2.0434,
      "step": 180501
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.3523095846176147,
      "learning_rate": 6.680563544416739e-05,
      "loss": 1.8275,
      "step": 180502
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.246042013168335,
      "learning_rate": 6.680306204349749e-05,
      "loss": 1.9641,
      "step": 180503
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.604856014251709,
      "learning_rate": 6.680048868618325e-05,
      "loss": 2.13,
      "step": 180504
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.235886812210083,
      "learning_rate": 6.679791537222526e-05,
      "loss": 2.0848,
      "step": 180505
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.4823421239852905,
      "learning_rate": 6.679534210162386e-05,
      "loss": 2.2126,
      "step": 180506
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.2919483184814453,
      "learning_rate": 6.679276887437974e-05,
      "loss": 2.1449,
      "step": 180507
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.3160395622253418,
      "learning_rate": 6.679019569049314e-05,
      "loss": 2.0397,
      "step": 180508
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.3361470699310303,
      "learning_rate": 6.678762254996469e-05,
      "loss": 1.8695,
      "step": 180509
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.2723729610443115,
      "learning_rate": 6.678504945279476e-05,
      "loss": 2.344,
      "step": 180510
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.3677985668182373,
      "learning_rate": 6.678247639898394e-05,
      "loss": 2.1979,
      "step": 180511
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.2556679248809814,
      "learning_rate": 6.67799033885326e-05,
      "loss": 1.7995,
      "step": 180512
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.4458755254745483,
      "learning_rate": 6.677733042144146e-05,
      "loss": 1.8948,
      "step": 180513
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.2747368812561035,
      "learning_rate": 6.677475749771062e-05,
      "loss": 1.8229,
      "step": 180514
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.2861932516098022,
      "learning_rate": 6.677218461734087e-05,
      "loss": 2.0778,
      "step": 180515
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.4744067192077637,
      "learning_rate": 6.676961178033246e-05,
      "loss": 2.0406,
      "step": 180516
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.388685941696167,
      "learning_rate": 6.676703898668609e-05,
      "loss": 1.9235,
      "step": 180517
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.3725076913833618,
      "learning_rate": 6.676446623640205e-05,
      "loss": 1.9347,
      "step": 180518
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.371117353439331,
      "learning_rate": 6.676189352948095e-05,
      "loss": 2.0489,
      "step": 180519
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.2868611812591553,
      "learning_rate": 6.675932086592323e-05,
      "loss": 2.1072,
      "step": 180520
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.403625249862671,
      "learning_rate": 6.675674824572936e-05,
      "loss": 2.0007,
      "step": 180521
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.4349274635314941,
      "learning_rate": 6.675417566889972e-05,
      "loss": 2.2267,
      "step": 180522
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.5571177005767822,
      "learning_rate": 6.6751603135435e-05,
      "loss": 2.1611,
      "step": 180523
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.336305022239685,
      "learning_rate": 6.674903064533544e-05,
      "loss": 2.1393,
      "step": 180524
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.643924593925476,
      "learning_rate": 6.674645819860174e-05,
      "loss": 2.0864,
      "step": 180525
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.4880503416061401,
      "learning_rate": 6.674388579523425e-05,
      "loss": 2.2225,
      "step": 180526
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.2688565254211426,
      "learning_rate": 6.674131343523351e-05,
      "loss": 1.964,
      "step": 180527
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.242352843284607,
      "learning_rate": 6.673874111859985e-05,
      "loss": 1.9399,
      "step": 180528
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.3868669271469116,
      "learning_rate": 6.673616884533396e-05,
      "loss": 1.8958,
      "step": 180529
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.5125224590301514,
      "learning_rate": 6.673359661543615e-05,
      "loss": 1.9184,
      "step": 180530
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.3834208250045776,
      "learning_rate": 6.673102442890701e-05,
      "loss": 1.9366,
      "step": 180531
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.3197424411773682,
      "learning_rate": 6.672845228574701e-05,
      "loss": 1.9948,
      "step": 180532
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.197853446006775,
      "learning_rate": 6.672588018595649e-05,
      "loss": 1.9197,
      "step": 180533
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.4854387044906616,
      "learning_rate": 6.672330812953615e-05,
      "loss": 2.1488,
      "step": 180534
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.6597974300384521,
      "learning_rate": 6.67207361164863e-05,
      "loss": 2.1379,
      "step": 180535
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.5137619972229004,
      "learning_rate": 6.671816414680744e-05,
      "loss": 1.6997,
      "step": 180536
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.4548146724700928,
      "learning_rate": 6.671559222050011e-05,
      "loss": 2.0965,
      "step": 180537
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.429569125175476,
      "learning_rate": 6.67130203375648e-05,
      "loss": 1.9884,
      "step": 180538
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.306485652923584,
      "learning_rate": 6.671044849800182e-05,
      "loss": 2.0144,
      "step": 180539
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.6371785402297974,
      "learning_rate": 6.670787670181187e-05,
      "loss": 2.1463,
      "step": 180540
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.7713358402252197,
      "learning_rate": 6.670530494899526e-05,
      "loss": 2.0955,
      "step": 180541
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.5587810277938843,
      "learning_rate": 6.67027332395526e-05,
      "loss": 1.874,
      "step": 180542
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.396079421043396,
      "learning_rate": 6.670016157348434e-05,
      "loss": 2.0577,
      "step": 180543
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.2940772771835327,
      "learning_rate": 6.669758995079087e-05,
      "loss": 1.9144,
      "step": 180544
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.2471641302108765,
      "learning_rate": 6.669501837147269e-05,
      "loss": 1.983,
      "step": 180545
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.7650682926177979,
      "learning_rate": 6.669244683553037e-05,
      "loss": 2.0794,
      "step": 180546
    },
    {
      "epoch": 2.35,
      "grad_norm": 2.348818778991699,
      "learning_rate": 6.668987534296425e-05,
      "loss": 2.0294,
      "step": 180547
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.438517451286316,
      "learning_rate": 6.668730389377494e-05,
      "loss": 2.1662,
      "step": 180548
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.4908781051635742,
      "learning_rate": 6.668473248796289e-05,
      "loss": 2.116,
      "step": 180549
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.312436580657959,
      "learning_rate": 6.668216112552858e-05,
      "loss": 1.8149,
      "step": 180550
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.3417608737945557,
      "learning_rate": 6.667958980647233e-05,
      "loss": 1.808,
      "step": 180551
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.5431675910949707,
      "learning_rate": 6.667701853079486e-05,
      "loss": 1.9913,
      "step": 180552
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.309965968132019,
      "learning_rate": 6.667444729849645e-05,
      "loss": 1.7157,
      "step": 180553
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.591264247894287,
      "learning_rate": 6.667187610957773e-05,
      "loss": 2.1507,
      "step": 180554
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.4136685132980347,
      "learning_rate": 6.666930496403912e-05,
      "loss": 2.1522,
      "step": 180555
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.3602138757705688,
      "learning_rate": 6.666673386188108e-05,
      "loss": 2.1051,
      "step": 180556
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.627307415008545,
      "learning_rate": 6.666416280310404e-05,
      "loss": 2.1329,
      "step": 180557
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.5325461626052856,
      "learning_rate": 6.666159178770862e-05,
      "loss": 1.9262,
      "step": 180558
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.308349370956421,
      "learning_rate": 6.66590208156951e-05,
      "loss": 1.8906,
      "step": 180559
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.4659290313720703,
      "learning_rate": 6.665644988706419e-05,
      "loss": 2.0676,
      "step": 180560
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.3568300008773804,
      "learning_rate": 6.665387900181617e-05,
      "loss": 2.2071,
      "step": 180561
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.308736801147461,
      "learning_rate": 6.665130815995175e-05,
      "loss": 1.9464,
      "step": 180562
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.3659498691558838,
      "learning_rate": 6.664873736147108e-05,
      "loss": 2.1442,
      "step": 180563
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.4911288022994995,
      "learning_rate": 6.664616660637492e-05,
      "loss": 1.8836,
      "step": 180564
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.3798835277557373,
      "learning_rate": 6.664359589466355e-05,
      "loss": 1.8615,
      "step": 180565
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.4935194253921509,
      "learning_rate": 6.664102522633761e-05,
      "loss": 1.9768,
      "step": 180566
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.2536941766738892,
      "learning_rate": 6.663845460139743e-05,
      "loss": 2.0617,
      "step": 180567
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.3139387369155884,
      "learning_rate": 6.663588401984374e-05,
      "loss": 1.9645,
      "step": 180568
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.330045461654663,
      "learning_rate": 6.663331348167669e-05,
      "loss": 2.2416,
      "step": 180569
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.568943738937378,
      "learning_rate": 6.663074298689699e-05,
      "loss": 1.8082,
      "step": 180570
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.2796630859375,
      "learning_rate": 6.662817253550493e-05,
      "loss": 2.009,
      "step": 180571
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.3928426504135132,
      "learning_rate": 6.662560212750121e-05,
      "loss": 2.0002,
      "step": 180572
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.4573789834976196,
      "learning_rate": 6.662303176288609e-05,
      "loss": 2.0936,
      "step": 180573
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.3201732635498047,
      "learning_rate": 6.662046144166034e-05,
      "loss": 1.9569,
      "step": 180574
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.389857292175293,
      "learning_rate": 6.661789116382407e-05,
      "loss": 2.0597,
      "step": 180575
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.3684239387512207,
      "learning_rate": 6.661532092937806e-05,
      "loss": 1.9261,
      "step": 180576
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.322906255722046,
      "learning_rate": 6.661275073832253e-05,
      "loss": 2.039,
      "step": 180577
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.3573009967803955,
      "learning_rate": 6.661018059065822e-05,
      "loss": 1.9107,
      "step": 180578
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.3605865240097046,
      "learning_rate": 6.660761048638538e-05,
      "loss": 1.8839,
      "step": 180579
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.3546829223632812,
      "learning_rate": 6.660504042550476e-05,
      "loss": 2.0304,
      "step": 180580
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.3099685907363892,
      "learning_rate": 6.66024704080165e-05,
      "loss": 2.0516,
      "step": 180581
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.4657807350158691,
      "learning_rate": 6.659990043392134e-05,
      "loss": 1.9903,
      "step": 180582
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.3009966611862183,
      "learning_rate": 6.659733050321956e-05,
      "loss": 2.0614,
      "step": 180583
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.3690307140350342,
      "learning_rate": 6.659476061591184e-05,
      "loss": 2.159,
      "step": 180584
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.3476121425628662,
      "learning_rate": 6.659219077199846e-05,
      "loss": 1.9884,
      "step": 180585
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.3741761445999146,
      "learning_rate": 6.658962097148018e-05,
      "loss": 1.9882,
      "step": 180586
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.2551418542861938,
      "learning_rate": 6.658705121435715e-05,
      "loss": 1.8885,
      "step": 180587
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.3763511180877686,
      "learning_rate": 6.658448150063006e-05,
      "loss": 2.2354,
      "step": 180588
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.252428412437439,
      "learning_rate": 6.658191183029923e-05,
      "loss": 1.8126,
      "step": 180589
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.2724988460540771,
      "learning_rate": 6.657934220336532e-05,
      "loss": 1.8386,
      "step": 180590
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.324156641960144,
      "learning_rate": 6.657677261982861e-05,
      "loss": 1.9879,
      "step": 180591
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.3677884340286255,
      "learning_rate": 6.657420307968988e-05,
      "loss": 1.9344,
      "step": 180592
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.3408024311065674,
      "learning_rate": 6.657163358294923e-05,
      "loss": 2.0744,
      "step": 180593
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.3481796979904175,
      "learning_rate": 6.65690641296074e-05,
      "loss": 2.1023,
      "step": 180594
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.4137929677963257,
      "learning_rate": 6.656649471966473e-05,
      "loss": 2.1697,
      "step": 180595
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.4953113794326782,
      "learning_rate": 6.656392535312181e-05,
      "loss": 2.1467,
      "step": 180596
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.261473298072815,
      "learning_rate": 6.656135602997899e-05,
      "loss": 1.8401,
      "step": 180597
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.3442734479904175,
      "learning_rate": 6.655878675023698e-05,
      "loss": 2.1317,
      "step": 180598
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.3082363605499268,
      "learning_rate": 6.655621751389592e-05,
      "loss": 2.2094,
      "step": 180599
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.6548792123794556,
      "learning_rate": 6.65536483209566e-05,
      "loss": 2.1662,
      "step": 180600
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.3392597436904907,
      "learning_rate": 6.655107917141922e-05,
      "loss": 1.8272,
      "step": 180601
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.2656171321868896,
      "learning_rate": 6.654851006528452e-05,
      "loss": 1.9343,
      "step": 180602
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.2095980644226074,
      "learning_rate": 6.654594100255277e-05,
      "loss": 1.8921,
      "step": 180603
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.3884514570236206,
      "learning_rate": 6.654337198322461e-05,
      "loss": 2.1285,
      "step": 180604
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.2385858297348022,
      "learning_rate": 6.654080300730045e-05,
      "loss": 1.991,
      "step": 180605
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.4475388526916504,
      "learning_rate": 6.653823407478078e-05,
      "loss": 2.1834,
      "step": 180606
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.2448666095733643,
      "learning_rate": 6.653566518566595e-05,
      "loss": 2.0935,
      "step": 180607
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.3356566429138184,
      "learning_rate": 6.653309633995662e-05,
      "loss": 1.9684,
      "step": 180608
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.3258156776428223,
      "learning_rate": 6.653052753765316e-05,
      "loss": 2.26,
      "step": 180609
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.1575936079025269,
      "learning_rate": 6.652795877875613e-05,
      "loss": 1.9168,
      "step": 180610
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.3049098253250122,
      "learning_rate": 6.652539006326595e-05,
      "loss": 2.0697,
      "step": 180611
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.3683209419250488,
      "learning_rate": 6.652282139118311e-05,
      "loss": 1.9445,
      "step": 180612
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.272017478942871,
      "learning_rate": 6.652025276250802e-05,
      "loss": 1.9024,
      "step": 180613
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.4295402765274048,
      "learning_rate": 6.651768417724131e-05,
      "loss": 2.077,
      "step": 180614
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.2918976545333862,
      "learning_rate": 6.651511563538326e-05,
      "loss": 2.0305,
      "step": 180615
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.2109266519546509,
      "learning_rate": 6.651254713693457e-05,
      "loss": 2.0239,
      "step": 180616
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.3365027904510498,
      "learning_rate": 6.650997868189559e-05,
      "loss": 1.9186,
      "step": 180617
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.29283607006073,
      "learning_rate": 6.65074102702667e-05,
      "loss": 1.7864,
      "step": 180618
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.3225069046020508,
      "learning_rate": 6.650484190204863e-05,
      "loss": 1.9633,
      "step": 180619
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.2958616018295288,
      "learning_rate": 6.650227357724167e-05,
      "loss": 1.7503,
      "step": 180620
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.3300433158874512,
      "learning_rate": 6.649970529584628e-05,
      "loss": 1.9873,
      "step": 180621
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.4328380823135376,
      "learning_rate": 6.64971370578631e-05,
      "loss": 2.2415,
      "step": 180622
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.3551921844482422,
      "learning_rate": 6.64945688632925e-05,
      "loss": 2.1202,
      "step": 180623
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.4349777698516846,
      "learning_rate": 6.649200071213486e-05,
      "loss": 2.1732,
      "step": 180624
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.3018141984939575,
      "learning_rate": 6.648943260439087e-05,
      "loss": 1.8292,
      "step": 180625
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.310083031654358,
      "learning_rate": 6.648686454006081e-05,
      "loss": 1.897,
      "step": 180626
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.348695158958435,
      "learning_rate": 6.648429651914536e-05,
      "loss": 1.9088,
      "step": 180627
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.4453809261322021,
      "learning_rate": 6.648172854164489e-05,
      "loss": 2.0552,
      "step": 180628
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.3095955848693848,
      "learning_rate": 6.647916060755984e-05,
      "loss": 2.1909,
      "step": 180629
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.4448466300964355,
      "learning_rate": 6.647659271689064e-05,
      "loss": 2.0425,
      "step": 180630
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.363774299621582,
      "learning_rate": 6.647402486963794e-05,
      "loss": 2.0433,
      "step": 180631
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.2585928440093994,
      "learning_rate": 6.647145706580204e-05,
      "loss": 1.9608,
      "step": 180632
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.4353383779525757,
      "learning_rate": 6.646888930538362e-05,
      "loss": 1.9664,
      "step": 180633
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.345630168914795,
      "learning_rate": 6.6466321588383e-05,
      "loss": 2.0715,
      "step": 180634
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.2637066841125488,
      "learning_rate": 6.646375391480075e-05,
      "loss": 1.7345,
      "step": 180635
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.4295722246170044,
      "learning_rate": 6.646118628463716e-05,
      "loss": 2.1616,
      "step": 180636
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.3709393739700317,
      "learning_rate": 6.645861869789295e-05,
      "loss": 2.0202,
      "step": 180637
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.4122610092163086,
      "learning_rate": 6.645605115456839e-05,
      "loss": 1.957,
      "step": 180638
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.2249350547790527,
      "learning_rate": 6.645348365466418e-05,
      "loss": 2.0081,
      "step": 180639
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.3476463556289673,
      "learning_rate": 6.645091619818056e-05,
      "loss": 2.2567,
      "step": 180640
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.296457290649414,
      "learning_rate": 6.644834878511831e-05,
      "loss": 2.1114,
      "step": 180641
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.27023184299469,
      "learning_rate": 6.644578141547752e-05,
      "loss": 2.1685,
      "step": 180642
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.339049220085144,
      "learning_rate": 6.644321408925898e-05,
      "loss": 2.011,
      "step": 180643
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.4242987632751465,
      "learning_rate": 6.644064680646296e-05,
      "loss": 1.911,
      "step": 180644
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.2847051620483398,
      "learning_rate": 6.643807956709014e-05,
      "loss": 2.0365,
      "step": 180645
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.3408446311950684,
      "learning_rate": 6.643551237114079e-05,
      "loss": 2.0057,
      "step": 180646
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.4427123069763184,
      "learning_rate": 6.643294521861563e-05,
      "loss": 1.9761,
      "step": 180647
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.2551687955856323,
      "learning_rate": 6.643037810951488e-05,
      "loss": 1.9201,
      "step": 180648
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.2420474290847778,
      "learning_rate": 6.642781104383918e-05,
      "loss": 1.8421,
      "step": 180649
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.2679433822631836,
      "learning_rate": 6.642524402158888e-05,
      "loss": 1.8514,
      "step": 180650
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.2173100709915161,
      "learning_rate": 6.642267704276461e-05,
      "loss": 1.8093,
      "step": 180651
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.3763883113861084,
      "learning_rate": 6.64201101073667e-05,
      "loss": 2.193,
      "step": 180652
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.2338273525238037,
      "learning_rate": 6.641754321539588e-05,
      "loss": 1.8329,
      "step": 180653
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.3056049346923828,
      "learning_rate": 6.641497636685227e-05,
      "loss": 2.1204,
      "step": 180654
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.3140647411346436,
      "learning_rate": 6.641240956173662e-05,
      "loss": 1.7887,
      "step": 180655
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.2882963418960571,
      "learning_rate": 6.640984280004919e-05,
      "loss": 2.0723,
      "step": 180656
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.4016077518463135,
      "learning_rate": 6.640727608179068e-05,
      "loss": 1.9062,
      "step": 180657
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.3105041980743408,
      "learning_rate": 6.640470940696141e-05,
      "loss": 1.8221,
      "step": 180658
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.6199615001678467,
      "learning_rate": 6.640214277556206e-05,
      "loss": 1.7592,
      "step": 180659
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.723555564880371,
      "learning_rate": 6.639957618759281e-05,
      "loss": 1.8848,
      "step": 180660
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.3811309337615967,
      "learning_rate": 6.639700964305435e-05,
      "loss": 1.9233,
      "step": 180661
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.3538923263549805,
      "learning_rate": 6.639444314194701e-05,
      "loss": 1.8762,
      "step": 180662
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.2546563148498535,
      "learning_rate": 6.639187668427146e-05,
      "loss": 2.0549,
      "step": 180663
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.3619097471237183,
      "learning_rate": 6.638931027002798e-05,
      "loss": 2.1173,
      "step": 180664
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.2811535596847534,
      "learning_rate": 6.638674389921732e-05,
      "loss": 2.1026,
      "step": 180665
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.3113878965377808,
      "learning_rate": 6.638417757183956e-05,
      "loss": 1.7939,
      "step": 180666
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.2957849502563477,
      "learning_rate": 6.638161128789549e-05,
      "loss": 1.9346,
      "step": 180667
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.3886507749557495,
      "learning_rate": 6.637904504738543e-05,
      "loss": 1.9224,
      "step": 180668
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.341893196105957,
      "learning_rate": 6.637647885030999e-05,
      "loss": 1.9352,
      "step": 180669
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.3399847745895386,
      "learning_rate": 6.637391269666947e-05,
      "loss": 2.0339,
      "step": 180670
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.2995041608810425,
      "learning_rate": 6.637134658646465e-05,
      "loss": 2.1329,
      "step": 180671
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.5486997365951538,
      "learning_rate": 6.636878051969561e-05,
      "loss": 1.9932,
      "step": 180672
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.3443059921264648,
      "learning_rate": 6.636621449636313e-05,
      "loss": 2.0106,
      "step": 180673
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.462380051612854,
      "learning_rate": 6.636364851646746e-05,
      "loss": 1.8876,
      "step": 180674
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.364349126815796,
      "learning_rate": 6.636108258000932e-05,
      "loss": 2.0518,
      "step": 180675
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.4487476348876953,
      "learning_rate": 6.635851668698896e-05,
      "loss": 2.0698,
      "step": 180676
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.4394069910049438,
      "learning_rate": 6.635595083740713e-05,
      "loss": 1.8939,
      "step": 180677
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.2460236549377441,
      "learning_rate": 6.635338503126397e-05,
      "loss": 2.0054,
      "step": 180678
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.2545831203460693,
      "learning_rate": 6.635081926856025e-05,
      "loss": 2.1186,
      "step": 180679
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.346840262413025,
      "learning_rate": 6.634825354929619e-05,
      "loss": 1.8772,
      "step": 180680
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.3730026483535767,
      "learning_rate": 6.63456878734725e-05,
      "loss": 2.0576,
      "step": 180681
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.2964754104614258,
      "learning_rate": 6.634312224108948e-05,
      "loss": 2.1679,
      "step": 180682
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.4202121496200562,
      "learning_rate": 6.634055665214783e-05,
      "loss": 1.9384,
      "step": 180683
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.3515077829360962,
      "learning_rate": 6.633799110664775e-05,
      "loss": 2.0327,
      "step": 180684
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.4691137075424194,
      "learning_rate": 6.633542560458988e-05,
      "loss": 2.0351,
      "step": 180685
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.4147167205810547,
      "learning_rate": 6.633286014597464e-05,
      "loss": 2.1506,
      "step": 180686
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.3336520195007324,
      "learning_rate": 6.633029473080259e-05,
      "loss": 2.1698,
      "step": 180687
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.2620247602462769,
      "learning_rate": 6.632772935907406e-05,
      "loss": 1.8843,
      "step": 180688
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.3241409063339233,
      "learning_rate": 6.63251640307898e-05,
      "loss": 2.0312,
      "step": 180689
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.2989604473114014,
      "learning_rate": 6.632259874594992e-05,
      "loss": 1.999,
      "step": 180690
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.2955257892608643,
      "learning_rate": 6.632003350455514e-05,
      "loss": 1.963,
      "step": 180691
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.3944766521453857,
      "learning_rate": 6.631746830660582e-05,
      "loss": 1.8872,
      "step": 180692
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.228864312171936,
      "learning_rate": 6.631490315210259e-05,
      "loss": 2.083,
      "step": 180693
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.466165542602539,
      "learning_rate": 6.631233804104576e-05,
      "loss": 1.8347,
      "step": 180694
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.3248413801193237,
      "learning_rate": 6.630977297343594e-05,
      "loss": 1.9138,
      "step": 180695
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.2919596433639526,
      "learning_rate": 6.630720794927354e-05,
      "loss": 1.9921,
      "step": 180696
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.3284460306167603,
      "learning_rate": 6.6304642968559e-05,
      "loss": 2.0473,
      "step": 180697
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.397891879081726,
      "learning_rate": 6.630207803129283e-05,
      "loss": 1.9181,
      "step": 180698
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.3465150594711304,
      "learning_rate": 6.629951313747555e-05,
      "loss": 2.0868,
      "step": 180699
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.388556957244873,
      "learning_rate": 6.629694828710753e-05,
      "loss": 1.9814,
      "step": 180700
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.4452524185180664,
      "learning_rate": 6.629438348018942e-05,
      "loss": 1.9985,
      "step": 180701
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.369173526763916,
      "learning_rate": 6.629181871672158e-05,
      "loss": 1.9011,
      "step": 180702
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.3217040300369263,
      "learning_rate": 6.628925399670442e-05,
      "loss": 2.1162,
      "step": 180703
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.2056819200515747,
      "learning_rate": 6.62866893201386e-05,
      "loss": 2.1216,
      "step": 180704
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.3048595190048218,
      "learning_rate": 6.628412468702445e-05,
      "loss": 2.0139,
      "step": 180705
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.292261004447937,
      "learning_rate": 6.628156009736245e-05,
      "loss": 2.0249,
      "step": 180706
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.2141914367675781,
      "learning_rate": 6.627899555115318e-05,
      "loss": 2.0146,
      "step": 180707
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.330851435661316,
      "learning_rate": 6.627643104839708e-05,
      "loss": 1.865,
      "step": 180708
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.3747998476028442,
      "learning_rate": 6.627386658909453e-05,
      "loss": 2.2298,
      "step": 180709
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.2888656854629517,
      "learning_rate": 6.627130217324612e-05,
      "loss": 1.9887,
      "step": 180710
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.3443456888198853,
      "learning_rate": 6.626873780085232e-05,
      "loss": 2.2754,
      "step": 180711
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.4465190172195435,
      "learning_rate": 6.626617347191346e-05,
      "loss": 1.8997,
      "step": 180712
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.320094108581543,
      "learning_rate": 6.626360918643025e-05,
      "loss": 2.0319,
      "step": 180713
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.271358609199524,
      "learning_rate": 6.626104494440303e-05,
      "loss": 1.9498,
      "step": 180714
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.3454996347427368,
      "learning_rate": 6.625848074583222e-05,
      "loss": 2.0855,
      "step": 180715
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.2701433897018433,
      "learning_rate": 6.625591659071841e-05,
      "loss": 1.765,
      "step": 180716
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.2826905250549316,
      "learning_rate": 6.625335247906201e-05,
      "loss": 1.9926,
      "step": 180717
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.4015278816223145,
      "learning_rate": 6.625078841086358e-05,
      "loss": 1.8619,
      "step": 180718
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.5543830394744873,
      "learning_rate": 6.624822438612357e-05,
      "loss": 2.1485,
      "step": 180719
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.547143816947937,
      "learning_rate": 6.62456604048424e-05,
      "loss": 2.1207,
      "step": 180720
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.5444109439849854,
      "learning_rate": 6.624309646702048e-05,
      "loss": 1.9297,
      "step": 180721
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.312629222869873,
      "learning_rate": 6.624053257265846e-05,
      "loss": 2.0642,
      "step": 180722
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.3158180713653564,
      "learning_rate": 6.623796872175668e-05,
      "loss": 2.0313,
      "step": 180723
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.5195740461349487,
      "learning_rate": 6.623540491431577e-05,
      "loss": 1.8437,
      "step": 180724
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.433447003364563,
      "learning_rate": 6.623284115033599e-05,
      "loss": 2.0603,
      "step": 180725
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.5144644975662231,
      "learning_rate": 6.623027742981814e-05,
      "loss": 2.1586,
      "step": 180726
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.3361924886703491,
      "learning_rate": 6.62277137527623e-05,
      "loss": 1.9089,
      "step": 180727
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.3387444019317627,
      "learning_rate": 6.622515011916926e-05,
      "loss": 2.2198,
      "step": 180728
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.4092689752578735,
      "learning_rate": 6.622258652903925e-05,
      "loss": 1.9698,
      "step": 180729
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.408281922340393,
      "learning_rate": 6.622002298237296e-05,
      "loss": 1.9218,
      "step": 180730
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.4519559144973755,
      "learning_rate": 6.621745947917071e-05,
      "loss": 1.7821,
      "step": 180731
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.5429000854492188,
      "learning_rate": 6.621489601943324e-05,
      "loss": 2.0088,
      "step": 180732
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.33177649974823,
      "learning_rate": 6.621233260316065e-05,
      "loss": 1.9848,
      "step": 180733
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.4594646692276,
      "learning_rate": 6.620976923035367e-05,
      "loss": 2.349,
      "step": 180734
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.3568599224090576,
      "learning_rate": 6.620720590101266e-05,
      "loss": 2.0224,
      "step": 180735
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.6589117050170898,
      "learning_rate": 6.620464261513819e-05,
      "loss": 2.0815,
      "step": 180736
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.2742624282836914,
      "learning_rate": 6.62020793727306e-05,
      "loss": 2.0214,
      "step": 180737
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.4010878801345825,
      "learning_rate": 6.619951617379063e-05,
      "loss": 1.9546,
      "step": 180738
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.476874589920044,
      "learning_rate": 6.619695301831844e-05,
      "loss": 2.0373,
      "step": 180739
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.494947910308838,
      "learning_rate": 6.619438990631472e-05,
      "loss": 1.9437,
      "step": 180740
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.5300850868225098,
      "learning_rate": 6.619182683777978e-05,
      "loss": 2.1245,
      "step": 180741
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.268222451210022,
      "learning_rate": 6.61892638127143e-05,
      "loss": 2.1468,
      "step": 180742
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.4069194793701172,
      "learning_rate": 6.618670083111854e-05,
      "loss": 1.9966,
      "step": 180743
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.2917882204055786,
      "learning_rate": 6.618413789299328e-05,
      "loss": 1.8975,
      "step": 180744
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.1497288942337036,
      "learning_rate": 6.618157499833861e-05,
      "loss": 1.9443,
      "step": 180745
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.3316192626953125,
      "learning_rate": 6.61790121471553e-05,
      "loss": 1.9268,
      "step": 180746
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.3686519861221313,
      "learning_rate": 6.617644933944366e-05,
      "loss": 2.0251,
      "step": 180747
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.4562299251556396,
      "learning_rate": 6.617388657520427e-05,
      "loss": 2.1876,
      "step": 180748
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.4384607076644897,
      "learning_rate": 6.61713238544375e-05,
      "loss": 2.1445,
      "step": 180749
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.3583118915557861,
      "learning_rate": 6.616876117714407e-05,
      "loss": 2.0617,
      "step": 180750
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.3236669301986694,
      "learning_rate": 6.616619854332412e-05,
      "loss": 1.9448,
      "step": 180751
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.2909095287322998,
      "learning_rate": 6.616363595297837e-05,
      "loss": 2.0749,
      "step": 180752
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.416171908378601,
      "learning_rate": 6.616107340610714e-05,
      "loss": 1.9631,
      "step": 180753
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.2883670330047607,
      "learning_rate": 6.615851090271103e-05,
      "loss": 2.0326,
      "step": 180754
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.3464394807815552,
      "learning_rate": 6.61559484427904e-05,
      "loss": 2.1063,
      "step": 180755
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.4381308555603027,
      "learning_rate": 6.615338602634598e-05,
      "loss": 2.0108,
      "step": 180756
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.3288848400115967,
      "learning_rate": 6.615082365337788e-05,
      "loss": 1.9841,
      "step": 180757
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.306233525276184,
      "learning_rate": 6.614826132388682e-05,
      "loss": 2.1681,
      "step": 180758
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.3394970893859863,
      "learning_rate": 6.614569903787316e-05,
      "loss": 1.8703,
      "step": 180759
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.3521857261657715,
      "learning_rate": 6.614313679533747e-05,
      "loss": 2.012,
      "step": 180760
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.400443434715271,
      "learning_rate": 6.614057459628013e-05,
      "loss": 2.0553,
      "step": 180761
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.3543950319290161,
      "learning_rate": 6.613801244070176e-05,
      "loss": 2.144,
      "step": 180762
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.2605453729629517,
      "learning_rate": 6.613545032860274e-05,
      "loss": 1.9416,
      "step": 180763
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.2933850288391113,
      "learning_rate": 6.613288825998355e-05,
      "loss": 1.9128,
      "step": 180764
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.4377645254135132,
      "learning_rate": 6.61303262348446e-05,
      "loss": 2.0142,
      "step": 180765
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.2654850482940674,
      "learning_rate": 6.612776425318652e-05,
      "loss": 2.0454,
      "step": 180766
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.2996704578399658,
      "learning_rate": 6.612520231500962e-05,
      "loss": 2.2045,
      "step": 180767
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.3450069427490234,
      "learning_rate": 6.612264042031453e-05,
      "loss": 2.1968,
      "step": 180768
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.3783038854599,
      "learning_rate": 6.612007856910166e-05,
      "loss": 1.9179,
      "step": 180769
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.3563597202301025,
      "learning_rate": 6.611751676137148e-05,
      "loss": 1.9425,
      "step": 180770
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.302924633026123,
      "learning_rate": 6.611495499712439e-05,
      "loss": 1.9767,
      "step": 180771
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.4396804571151733,
      "learning_rate": 6.611239327636103e-05,
      "loss": 2.0447,
      "step": 180772
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.3910130262374878,
      "learning_rate": 6.61098315990817e-05,
      "loss": 1.9516,
      "step": 180773
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.3130708932876587,
      "learning_rate": 6.610726996528705e-05,
      "loss": 1.9454,
      "step": 180774
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.2557530403137207,
      "learning_rate": 6.610470837497748e-05,
      "loss": 1.9473,
      "step": 180775
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.3055318593978882,
      "learning_rate": 6.610214682815346e-05,
      "loss": 1.9175,
      "step": 180776
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.3493748903274536,
      "learning_rate": 6.609958532481538e-05,
      "loss": 2.0153,
      "step": 180777
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.4383165836334229,
      "learning_rate": 6.60970238649639e-05,
      "loss": 1.8371,
      "step": 180778
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.3865203857421875,
      "learning_rate": 6.609446244859928e-05,
      "loss": 2.2468,
      "step": 180779
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.4371094703674316,
      "learning_rate": 6.609190107572222e-05,
      "loss": 2.0847,
      "step": 180780
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.3081161975860596,
      "learning_rate": 6.60893397463331e-05,
      "loss": 2.0488,
      "step": 180781
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.2271283864974976,
      "learning_rate": 6.60867784604324e-05,
      "loss": 2.0094,
      "step": 180782
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.4021555185317993,
      "learning_rate": 6.608421721802048e-05,
      "loss": 2.0036,
      "step": 180783
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.6824569702148438,
      "learning_rate": 6.608165601909799e-05,
      "loss": 2.1916,
      "step": 180784
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.3346186876296997,
      "learning_rate": 6.607909486366527e-05,
      "loss": 2.1339,
      "step": 180785
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.258272409439087,
      "learning_rate": 6.607653375172294e-05,
      "loss": 2.2084,
      "step": 180786
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.429404854774475,
      "learning_rate": 6.607397268327139e-05,
      "loss": 1.8104,
      "step": 180787
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.3283149003982544,
      "learning_rate": 6.607141165831102e-05,
      "loss": 1.9013,
      "step": 180788
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.3579236268997192,
      "learning_rate": 6.60688506768425e-05,
      "loss": 2.0156,
      "step": 180789
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.297115445137024,
      "learning_rate": 6.606628973886615e-05,
      "loss": 2.1814,
      "step": 180790
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.2952182292938232,
      "learning_rate": 6.606372884438245e-05,
      "loss": 2.0083,
      "step": 180791
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.3633277416229248,
      "learning_rate": 6.606116799339199e-05,
      "loss": 2.1528,
      "step": 180792
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.2416422367095947,
      "learning_rate": 6.605860718589516e-05,
      "loss": 1.7736,
      "step": 180793
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.2647606134414673,
      "learning_rate": 6.605604642189238e-05,
      "loss": 2.0519,
      "step": 180794
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.3204939365386963,
      "learning_rate": 6.605348570138432e-05,
      "loss": 1.9183,
      "step": 180795
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.2983933687210083,
      "learning_rate": 6.605092502437128e-05,
      "loss": 1.8658,
      "step": 180796
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.4811997413635254,
      "learning_rate": 6.604836439085373e-05,
      "loss": 2.1718,
      "step": 180797
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.3768256902694702,
      "learning_rate": 6.604580380083227e-05,
      "loss": 2.0325,
      "step": 180798
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.2830569744110107,
      "learning_rate": 6.604324325430733e-05,
      "loss": 1.9009,
      "step": 180799
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.3030431270599365,
      "learning_rate": 6.604068275127926e-05,
      "loss": 1.9767,
      "step": 180800
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.3538241386413574,
      "learning_rate": 6.603812229174873e-05,
      "loss": 2.0778,
      "step": 180801
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.2828218936920166,
      "learning_rate": 6.603556187571607e-05,
      "loss": 2.1101,
      "step": 180802
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.2576991319656372,
      "learning_rate": 6.603300150318188e-05,
      "loss": 2.0052,
      "step": 180803
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.3170785903930664,
      "learning_rate": 6.60304411741466e-05,
      "loss": 1.7741,
      "step": 180804
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.2605886459350586,
      "learning_rate": 6.602788088861065e-05,
      "loss": 2.0918,
      "step": 180805
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.3000189065933228,
      "learning_rate": 6.602532064657444e-05,
      "loss": 2.1417,
      "step": 180806
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.2823190689086914,
      "learning_rate": 6.602276044803863e-05,
      "loss": 2.0172,
      "step": 180807
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.4063498973846436,
      "learning_rate": 6.602020029300353e-05,
      "loss": 1.8612,
      "step": 180808
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.3736135959625244,
      "learning_rate": 6.60176401814698e-05,
      "loss": 2.225,
      "step": 180809
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.4552658796310425,
      "learning_rate": 6.601508011343777e-05,
      "loss": 2.2717,
      "step": 180810
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.4210153818130493,
      "learning_rate": 6.601252008890797e-05,
      "loss": 2.0125,
      "step": 180811
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.2999281883239746,
      "learning_rate": 6.600996010788078e-05,
      "loss": 2.0605,
      "step": 180812
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.3163655996322632,
      "learning_rate": 6.600740017035682e-05,
      "loss": 2.0069,
      "step": 180813
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.3447414636611938,
      "learning_rate": 6.600484027633643e-05,
      "loss": 2.1387,
      "step": 180814
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.316267967224121,
      "learning_rate": 6.600228042582023e-05,
      "loss": 2.186,
      "step": 180815
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.2972283363342285,
      "learning_rate": 6.599972061880857e-05,
      "loss": 2.1549,
      "step": 180816
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.2827156782150269,
      "learning_rate": 6.599716085530211e-05,
      "loss": 2.1752,
      "step": 180817
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.356076955795288,
      "learning_rate": 6.599460113530105e-05,
      "loss": 2.1136,
      "step": 180818
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.4191027879714966,
      "learning_rate": 6.599204145880609e-05,
      "loss": 2.0242,
      "step": 180819
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.4613386392593384,
      "learning_rate": 6.598948182581752e-05,
      "loss": 1.8892,
      "step": 180820
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.2706093788146973,
      "learning_rate": 6.598692223633606e-05,
      "loss": 2.1331,
      "step": 180821
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.255927562713623,
      "learning_rate": 6.598436269036193e-05,
      "loss": 2.173,
      "step": 180822
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.3367552757263184,
      "learning_rate": 6.598180318789589e-05,
      "loss": 2.2487,
      "step": 180823
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.396545171737671,
      "learning_rate": 6.597924372893809e-05,
      "loss": 1.9226,
      "step": 180824
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.505342960357666,
      "learning_rate": 6.597668431348924e-05,
      "loss": 2.1011,
      "step": 180825
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.3288089036941528,
      "learning_rate": 6.597412494154965e-05,
      "loss": 2.0318,
      "step": 180826
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.270910620689392,
      "learning_rate": 6.597156561312e-05,
      "loss": 1.9471,
      "step": 180827
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.3761165142059326,
      "learning_rate": 6.596900632820054e-05,
      "loss": 2.0366,
      "step": 180828
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.5970332622528076,
      "learning_rate": 6.596644708679198e-05,
      "loss": 2.0662,
      "step": 180829
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.337570071220398,
      "learning_rate": 6.596388788889466e-05,
      "loss": 2.1965,
      "step": 180830
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.3009655475616455,
      "learning_rate": 6.596132873450904e-05,
      "loss": 1.9262,
      "step": 180831
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.331203818321228,
      "learning_rate": 6.595876962363557e-05,
      "loss": 1.9804,
      "step": 180832
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.423108696937561,
      "learning_rate": 6.595621055627488e-05,
      "loss": 2.0456,
      "step": 180833
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.356017827987671,
      "learning_rate": 6.595365153242725e-05,
      "loss": 1.9913,
      "step": 180834
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.4365357160568237,
      "learning_rate": 6.595109255209331e-05,
      "loss": 2.0431,
      "step": 180835
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.3929188251495361,
      "learning_rate": 6.594853361527351e-05,
      "loss": 2.2089,
      "step": 180836
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.3001359701156616,
      "learning_rate": 6.594597472196829e-05,
      "loss": 2.1197,
      "step": 180837
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.3525854349136353,
      "learning_rate": 6.594341587217807e-05,
      "loss": 1.8525,
      "step": 180838
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.3037561178207397,
      "learning_rate": 6.594085706590345e-05,
      "loss": 1.9487,
      "step": 180839
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.4431370496749878,
      "learning_rate": 6.593829830314476e-05,
      "loss": 1.8798,
      "step": 180840
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.3704909086227417,
      "learning_rate": 6.593573958390265e-05,
      "loss": 2.0322,
      "step": 180841
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.1767433881759644,
      "learning_rate": 6.593318090817748e-05,
      "loss": 1.7716,
      "step": 180842
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.3522080183029175,
      "learning_rate": 6.593062227596978e-05,
      "loss": 1.9906,
      "step": 180843
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.3419840335845947,
      "learning_rate": 6.592806368727989e-05,
      "loss": 2.0228,
      "step": 180844
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.4236242771148682,
      "learning_rate": 6.59255051421085e-05,
      "loss": 1.7534,
      "step": 180845
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.5207518339157104,
      "learning_rate": 6.592294664045587e-05,
      "loss": 2.0959,
      "step": 180846
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.3328913450241089,
      "learning_rate": 6.592038818232268e-05,
      "loss": 2.1112,
      "step": 180847
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.2762542963027954,
      "learning_rate": 6.591782976770928e-05,
      "loss": 1.8768,
      "step": 180848
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.2658804655075073,
      "learning_rate": 6.591527139661621e-05,
      "loss": 2.1756,
      "step": 180849
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.3659039735794067,
      "learning_rate": 6.591271306904382e-05,
      "loss": 2.2894,
      "step": 180850
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.4317007064819336,
      "learning_rate": 6.591015478499275e-05,
      "loss": 1.822,
      "step": 180851
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.224485158920288,
      "learning_rate": 6.59075965444633e-05,
      "loss": 1.9475,
      "step": 180852
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.322191834449768,
      "learning_rate": 6.590503834745618e-05,
      "loss": 1.9375,
      "step": 180853
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.4398890733718872,
      "learning_rate": 6.59024801939717e-05,
      "loss": 1.999,
      "step": 180854
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.4001154899597168,
      "learning_rate": 6.589992208401034e-05,
      "loss": 2.0729,
      "step": 180855
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.3288156986236572,
      "learning_rate": 6.589736401757256e-05,
      "loss": 1.9639,
      "step": 180856
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.4285565614700317,
      "learning_rate": 6.589480599465894e-05,
      "loss": 2.0856,
      "step": 180857
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.2889020442962646,
      "learning_rate": 6.589224801526983e-05,
      "loss": 1.963,
      "step": 180858
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.2372868061065674,
      "learning_rate": 6.588969007940584e-05,
      "loss": 2.057,
      "step": 180859
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.4622788429260254,
      "learning_rate": 6.58871321870674e-05,
      "loss": 2.0491,
      "step": 180860
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.2771130800247192,
      "learning_rate": 6.588457433825494e-05,
      "loss": 1.9646,
      "step": 180861
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.330941081047058,
      "learning_rate": 6.588201653296887e-05,
      "loss": 2.2095,
      "step": 180862
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.3859516382217407,
      "learning_rate": 6.587945877120985e-05,
      "loss": 2.1345,
      "step": 180863
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.2580472230911255,
      "learning_rate": 6.587690105297817e-05,
      "loss": 2.0516,
      "step": 180864
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.3029093742370605,
      "learning_rate": 6.587434337827448e-05,
      "loss": 2.1935,
      "step": 180865
    },
    {
      "epoch": 2.35,
      "grad_norm": 1.37795090675354,
      "learning_rate": 6.587178574709916e-05,
      "loss": 2.1378,
      "step": 180866
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.4254666566848755,
      "learning_rate": 6.586922815945269e-05,
      "loss": 2.1425,
      "step": 180867
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.3887544870376587,
      "learning_rate": 6.586667061533548e-05,
      "loss": 2.3396,
      "step": 180868
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.2939146757125854,
      "learning_rate": 6.586411311474816e-05,
      "loss": 1.8827,
      "step": 180869
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.3685747385025024,
      "learning_rate": 6.586155565769103e-05,
      "loss": 2.078,
      "step": 180870
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.449021816253662,
      "learning_rate": 6.585899824416476e-05,
      "loss": 2.0435,
      "step": 180871
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.266644835472107,
      "learning_rate": 6.585644087416968e-05,
      "loss": 2.0632,
      "step": 180872
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.438382863998413,
      "learning_rate": 6.585388354770632e-05,
      "loss": 2.0521,
      "step": 180873
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.4469764232635498,
      "learning_rate": 6.585132626477507e-05,
      "loss": 2.0952,
      "step": 180874
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.3638559579849243,
      "learning_rate": 6.584876902537657e-05,
      "loss": 2.0437,
      "step": 180875
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.502586007118225,
      "learning_rate": 6.58462118295111e-05,
      "loss": 1.9113,
      "step": 180876
    },
    {
      "epoch": 2.36,
      "grad_norm": 2.6738133430480957,
      "learning_rate": 6.584365467717936e-05,
      "loss": 2.0569,
      "step": 180877
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.3876391649246216,
      "learning_rate": 6.584109756838169e-05,
      "loss": 1.9457,
      "step": 180878
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.3163892030715942,
      "learning_rate": 6.583854050311845e-05,
      "loss": 2.1578,
      "step": 180879
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.5311129093170166,
      "learning_rate": 6.58359834813904e-05,
      "loss": 2.0979,
      "step": 180880
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.3591668605804443,
      "learning_rate": 6.583342650319782e-05,
      "loss": 2.121,
      "step": 180881
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.3179925680160522,
      "learning_rate": 6.583086956854114e-05,
      "loss": 2.0395,
      "step": 180882
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.3683018684387207,
      "learning_rate": 6.582831267742102e-05,
      "loss": 1.973,
      "step": 180883
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.3487522602081299,
      "learning_rate": 6.582575582983782e-05,
      "loss": 2.1476,
      "step": 180884
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.4004210233688354,
      "learning_rate": 6.582319902579198e-05,
      "loss": 2.0186,
      "step": 180885
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.3426626920700073,
      "learning_rate": 6.582064226528406e-05,
      "loss": 2.1643,
      "step": 180886
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.3841831684112549,
      "learning_rate": 6.581808554831446e-05,
      "loss": 1.7033,
      "step": 180887
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.3365284204483032,
      "learning_rate": 6.581552887488377e-05,
      "loss": 1.8224,
      "step": 180888
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.295340895652771,
      "learning_rate": 6.58129722449924e-05,
      "loss": 2.0015,
      "step": 180889
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.318602204322815,
      "learning_rate": 6.581041565864082e-05,
      "loss": 1.8094,
      "step": 180890
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.3748271465301514,
      "learning_rate": 6.58078591158294e-05,
      "loss": 2.0179,
      "step": 180891
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.4423259496688843,
      "learning_rate": 6.580530261655882e-05,
      "loss": 2.2687,
      "step": 180892
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.429041862487793,
      "learning_rate": 6.580274616082938e-05,
      "loss": 1.9482,
      "step": 180893
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.274129867553711,
      "learning_rate": 6.580018974864171e-05,
      "loss": 2.0446,
      "step": 180894
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.3807036876678467,
      "learning_rate": 6.57976333799962e-05,
      "loss": 1.9379,
      "step": 180895
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.389784812927246,
      "learning_rate": 6.579507705489324e-05,
      "loss": 2.0886,
      "step": 180896
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.3094192743301392,
      "learning_rate": 6.579252077333348e-05,
      "loss": 1.9836,
      "step": 180897
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.4091798067092896,
      "learning_rate": 6.578996453531731e-05,
      "loss": 2.1395,
      "step": 180898
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.3028998374938965,
      "learning_rate": 6.578740834084516e-05,
      "loss": 1.8785,
      "step": 180899
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.3769036531448364,
      "learning_rate": 6.57848521899176e-05,
      "loss": 1.9786,
      "step": 180900
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.375959873199463,
      "learning_rate": 6.578229608253498e-05,
      "loss": 2.0237,
      "step": 180901
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.4003604650497437,
      "learning_rate": 6.577974001869793e-05,
      "loss": 2.1573,
      "step": 180902
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.2389099597930908,
      "learning_rate": 6.577718399840687e-05,
      "loss": 1.9315,
      "step": 180903
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.4827433824539185,
      "learning_rate": 6.577462802166225e-05,
      "loss": 1.964,
      "step": 180904
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.4848170280456543,
      "learning_rate": 6.577207208846442e-05,
      "loss": 2.0907,
      "step": 180905
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.2507396936416626,
      "learning_rate": 6.57695161988141e-05,
      "loss": 1.9702,
      "step": 180906
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.3565762042999268,
      "learning_rate": 6.576696035271157e-05,
      "loss": 1.9789,
      "step": 180907
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.4064934253692627,
      "learning_rate": 6.576440455015748e-05,
      "loss": 1.9629,
      "step": 180908
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.2526658773422241,
      "learning_rate": 6.576184879115219e-05,
      "loss": 1.9665,
      "step": 180909
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.23983633518219,
      "learning_rate": 6.575929307569619e-05,
      "loss": 2.1568,
      "step": 180910
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.3491753339767456,
      "learning_rate": 6.575673740378984e-05,
      "loss": 1.7562,
      "step": 180911
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.281938076019287,
      "learning_rate": 6.575418177543388e-05,
      "loss": 1.964,
      "step": 180912
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.2724061012268066,
      "learning_rate": 6.575162619062854e-05,
      "loss": 1.8514,
      "step": 180913
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.2218705415725708,
      "learning_rate": 6.574907064937446e-05,
      "loss": 2.0767,
      "step": 180914
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.2243767976760864,
      "learning_rate": 6.574651515167207e-05,
      "loss": 1.6759,
      "step": 180915
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.262887716293335,
      "learning_rate": 6.574395969752182e-05,
      "loss": 1.8975,
      "step": 180916
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.4067589044570923,
      "learning_rate": 6.57414042869241e-05,
      "loss": 2.0982,
      "step": 180917
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.2741426229476929,
      "learning_rate": 6.573884891987956e-05,
      "loss": 2.1588,
      "step": 180918
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.3654476404190063,
      "learning_rate": 6.573629359638849e-05,
      "loss": 2.074,
      "step": 180919
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.406651258468628,
      "learning_rate": 6.573373831645158e-05,
      "loss": 2.2095,
      "step": 180920
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.395033597946167,
      "learning_rate": 6.573118308006916e-05,
      "loss": 1.8626,
      "step": 180921
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.369248390197754,
      "learning_rate": 6.572862788724173e-05,
      "loss": 1.9665,
      "step": 180922
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.3921531438827515,
      "learning_rate": 6.572607273796974e-05,
      "loss": 1.9832,
      "step": 180923
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.4058363437652588,
      "learning_rate": 6.572351763225373e-05,
      "loss": 2.1205,
      "step": 180924
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.520973563194275,
      "learning_rate": 6.572096257009408e-05,
      "loss": 1.9496,
      "step": 180925
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.2681825160980225,
      "learning_rate": 6.57184075514914e-05,
      "loss": 1.8208,
      "step": 180926
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.384368658065796,
      "learning_rate": 6.571585257644609e-05,
      "loss": 2.0999,
      "step": 180927
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.6965575218200684,
      "learning_rate": 6.571329764495862e-05,
      "loss": 2.0489,
      "step": 180928
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.4220993518829346,
      "learning_rate": 6.57107427570294e-05,
      "loss": 2.0367,
      "step": 180929
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.4014275074005127,
      "learning_rate": 6.570818791265904e-05,
      "loss": 2.0028,
      "step": 180930
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.565810203552246,
      "learning_rate": 6.57056331118479e-05,
      "loss": 1.903,
      "step": 180931
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.3126931190490723,
      "learning_rate": 6.570307835459655e-05,
      "loss": 1.9686,
      "step": 180932
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.4569721221923828,
      "learning_rate": 6.570052364090546e-05,
      "loss": 1.79,
      "step": 180933
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.4618237018585205,
      "learning_rate": 6.569796897077501e-05,
      "loss": 2.1352,
      "step": 180934
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.6625027656555176,
      "learning_rate": 6.56954143442057e-05,
      "loss": 2.0249,
      "step": 180935
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.2762720584869385,
      "learning_rate": 6.56928597611981e-05,
      "loss": 2.0776,
      "step": 180936
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.3425666093826294,
      "learning_rate": 6.569030522175252e-05,
      "loss": 2.0149,
      "step": 180937
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.3812142610549927,
      "learning_rate": 6.568775072586968e-05,
      "loss": 2.0258,
      "step": 180938
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.6355422735214233,
      "learning_rate": 6.568519627354983e-05,
      "loss": 2.0625,
      "step": 180939
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.3362749814987183,
      "learning_rate": 6.568264186479359e-05,
      "loss": 1.8456,
      "step": 180940
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.271607756614685,
      "learning_rate": 6.568008749960125e-05,
      "loss": 1.9772,
      "step": 180941
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.3137915134429932,
      "learning_rate": 6.567753317797348e-05,
      "loss": 2.0802,
      "step": 180942
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.3412553071975708,
      "learning_rate": 6.567497889991062e-05,
      "loss": 2.1191,
      "step": 180943
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.3624438047409058,
      "learning_rate": 6.567242466541327e-05,
      "loss": 2.1413,
      "step": 180944
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.592376708984375,
      "learning_rate": 6.566987047448184e-05,
      "loss": 1.9692,
      "step": 180945
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.3926080465316772,
      "learning_rate": 6.566731632711684e-05,
      "loss": 1.8852,
      "step": 180946
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.412290334701538,
      "learning_rate": 6.566476222331857e-05,
      "loss": 2.0044,
      "step": 180947
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.2352403402328491,
      "learning_rate": 6.566220816308777e-05,
      "loss": 2.2272,
      "step": 180948
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.3324166536331177,
      "learning_rate": 6.565965414642471e-05,
      "loss": 2.1232,
      "step": 180949
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.2869110107421875,
      "learning_rate": 6.565710017333e-05,
      "loss": 2.1883,
      "step": 180950
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.3817824125289917,
      "learning_rate": 6.565454624380407e-05,
      "loss": 1.8255,
      "step": 180951
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.384613275527954,
      "learning_rate": 6.56519923578474e-05,
      "loss": 2.1029,
      "step": 180952
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.263985514640808,
      "learning_rate": 6.564943851546033e-05,
      "loss": 2.1106,
      "step": 180953
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.3311333656311035,
      "learning_rate": 6.564688471664355e-05,
      "loss": 2.0185,
      "step": 180954
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.2907743453979492,
      "learning_rate": 6.564433096139738e-05,
      "loss": 1.8272,
      "step": 180955
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.6443020105361938,
      "learning_rate": 6.564177724972238e-05,
      "loss": 2.1395,
      "step": 180956
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.467320442199707,
      "learning_rate": 6.563922358161904e-05,
      "loss": 2.0566,
      "step": 180957
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.3104976415634155,
      "learning_rate": 6.56366699570878e-05,
      "loss": 2.1103,
      "step": 180958
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.6056076288223267,
      "learning_rate": 6.563411637612902e-05,
      "loss": 1.8575,
      "step": 180959
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.3311431407928467,
      "learning_rate": 6.563156283874338e-05,
      "loss": 1.9445,
      "step": 180960
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.4421800374984741,
      "learning_rate": 6.562900934493115e-05,
      "loss": 2.0295,
      "step": 180961
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.277653694152832,
      "learning_rate": 6.562645589469303e-05,
      "loss": 1.8582,
      "step": 180962
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.3239455223083496,
      "learning_rate": 6.562390248802926e-05,
      "loss": 1.8774,
      "step": 180963
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.3901394605636597,
      "learning_rate": 6.562134912494052e-05,
      "loss": 2.0956,
      "step": 180964
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.4292844533920288,
      "learning_rate": 6.56187958054272e-05,
      "loss": 2.158,
      "step": 180965
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.4697755575180054,
      "learning_rate": 6.56162425294898e-05,
      "loss": 2.1325,
      "step": 180966
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.3001738786697388,
      "learning_rate": 6.561368929712862e-05,
      "loss": 1.9012,
      "step": 180967
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.2801408767700195,
      "learning_rate": 6.56111361083444e-05,
      "loss": 2.0062,
      "step": 180968
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.2434700727462769,
      "learning_rate": 6.560858296313741e-05,
      "loss": 2.1747,
      "step": 180969
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.3998795747756958,
      "learning_rate": 6.560602986150831e-05,
      "loss": 2.037,
      "step": 180970
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.3015425205230713,
      "learning_rate": 6.560347680345744e-05,
      "loss": 2.0735,
      "step": 180971
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.5600121021270752,
      "learning_rate": 6.560092378898534e-05,
      "loss": 2.1728,
      "step": 180972
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.220509648323059,
      "learning_rate": 6.559837081809236e-05,
      "loss": 1.901,
      "step": 180973
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.3039941787719727,
      "learning_rate": 6.559581789077914e-05,
      "loss": 1.8207,
      "step": 180974
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.3328865766525269,
      "learning_rate": 6.559326500704599e-05,
      "loss": 1.8838,
      "step": 180975
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.318678379058838,
      "learning_rate": 6.559071216689356e-05,
      "loss": 2.2094,
      "step": 180976
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.1818914413452148,
      "learning_rate": 6.558815937032226e-05,
      "loss": 2.0131,
      "step": 180977
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.3835240602493286,
      "learning_rate": 6.558560661733246e-05,
      "loss": 2.0027,
      "step": 180978
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.2518565654754639,
      "learning_rate": 6.558305390792478e-05,
      "loss": 2.0892,
      "step": 180979
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.326300859451294,
      "learning_rate": 6.558050124209966e-05,
      "loss": 2.0856,
      "step": 180980
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.3120286464691162,
      "learning_rate": 6.557794861985748e-05,
      "loss": 2.1136,
      "step": 180981
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.2863181829452515,
      "learning_rate": 6.557539604119884e-05,
      "loss": 1.9345,
      "step": 180982
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.4772677421569824,
      "learning_rate": 6.557284350612416e-05,
      "loss": 2.0249,
      "step": 180983
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.3344093561172485,
      "learning_rate": 6.557029101463386e-05,
      "loss": 2.0876,
      "step": 180984
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.3085455894470215,
      "learning_rate": 6.556773856672852e-05,
      "loss": 1.9931,
      "step": 180985
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.304063320159912,
      "learning_rate": 6.556518616240849e-05,
      "loss": 2.0563,
      "step": 180986
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.219481348991394,
      "learning_rate": 6.556263380167439e-05,
      "loss": 2.0339,
      "step": 180987
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.2514313459396362,
      "learning_rate": 6.556008148452662e-05,
      "loss": 2.0633,
      "step": 180988
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.283553123474121,
      "learning_rate": 6.555752921096567e-05,
      "loss": 1.8763,
      "step": 180989
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.2315391302108765,
      "learning_rate": 6.555497698099193e-05,
      "loss": 1.9449,
      "step": 180990
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.3709369897842407,
      "learning_rate": 6.555242479460601e-05,
      "loss": 1.9874,
      "step": 180991
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.4296882152557373,
      "learning_rate": 6.554987265180824e-05,
      "loss": 2.1488,
      "step": 180992
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.2935510873794556,
      "learning_rate": 6.554732055259925e-05,
      "loss": 2.0767,
      "step": 180993
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.2965729236602783,
      "learning_rate": 6.554476849697945e-05,
      "loss": 1.9487,
      "step": 180994
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.3567173480987549,
      "learning_rate": 6.554221648494931e-05,
      "loss": 1.9113,
      "step": 180995
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.2585433721542358,
      "learning_rate": 6.553966451650919e-05,
      "loss": 2.0992,
      "step": 180996
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.3200496435165405,
      "learning_rate": 6.553711259165978e-05,
      "loss": 2.1123,
      "step": 180997
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.309311866760254,
      "learning_rate": 6.553456071040134e-05,
      "loss": 2.1802,
      "step": 180998
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.5289044380187988,
      "learning_rate": 6.553200887273456e-05,
      "loss": 2.0091,
      "step": 180999
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.2854374647140503,
      "learning_rate": 6.552945707865978e-05,
      "loss": 1.8726,
      "step": 181000
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.4110239744186401,
      "learning_rate": 6.552690532817753e-05,
      "loss": 2.0708,
      "step": 181001
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.5066986083984375,
      "learning_rate": 6.552435362128815e-05,
      "loss": 1.9417,
      "step": 181002
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.350705623626709,
      "learning_rate": 6.55218019579923e-05,
      "loss": 1.9241,
      "step": 181003
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.4240964651107788,
      "learning_rate": 6.551925033829029e-05,
      "loss": 1.9941,
      "step": 181004
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.2152042388916016,
      "learning_rate": 6.551669876218276e-05,
      "loss": 2.1111,
      "step": 181005
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.3737437725067139,
      "learning_rate": 6.551414722967012e-05,
      "loss": 2.1196,
      "step": 181006
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.4795582294464111,
      "learning_rate": 6.551159574075281e-05,
      "loss": 2.051,
      "step": 181007
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.273695468902588,
      "learning_rate": 6.550904429543125e-05,
      "loss": 1.8927,
      "step": 181008
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.399713158607483,
      "learning_rate": 6.550649289370607e-05,
      "loss": 1.8147,
      "step": 181009
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.2600693702697754,
      "learning_rate": 6.550394153557757e-05,
      "loss": 2.3672,
      "step": 181010
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.3900500535964966,
      "learning_rate": 6.550139022104639e-05,
      "loss": 2.101,
      "step": 181011
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.3060429096221924,
      "learning_rate": 6.549883895011291e-05,
      "loss": 1.9831,
      "step": 181012
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.5463855266571045,
      "learning_rate": 6.549628772277765e-05,
      "loss": 2.1804,
      "step": 181013
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.2253957986831665,
      "learning_rate": 6.549373653904096e-05,
      "loss": 1.997,
      "step": 181014
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.4417575597763062,
      "learning_rate": 6.54911853989035e-05,
      "loss": 2.0425,
      "step": 181015
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.325708031654358,
      "learning_rate": 6.548863430236559e-05,
      "loss": 1.9273,
      "step": 181016
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.2948436737060547,
      "learning_rate": 6.548608324942783e-05,
      "loss": 1.8621,
      "step": 181017
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.4397333860397339,
      "learning_rate": 6.548353224009063e-05,
      "loss": 2.0203,
      "step": 181018
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.4592021703720093,
      "learning_rate": 6.548098127435449e-05,
      "loss": 2.164,
      "step": 181019
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.3418042659759521,
      "learning_rate": 6.547843035221976e-05,
      "loss": 1.7114,
      "step": 181020
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.438125729560852,
      "learning_rate": 6.547587947368712e-05,
      "loss": 2.1116,
      "step": 181021
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.343207836151123,
      "learning_rate": 6.547332863875683e-05,
      "loss": 1.8693,
      "step": 181022
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.275283694267273,
      "learning_rate": 6.547077784742958e-05,
      "loss": 2.0658,
      "step": 181023
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.4682042598724365,
      "learning_rate": 6.546822709970574e-05,
      "loss": 2.1324,
      "step": 181024
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.2943657636642456,
      "learning_rate": 6.546567639558575e-05,
      "loss": 1.76,
      "step": 181025
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.3562606573104858,
      "learning_rate": 6.54631257350701e-05,
      "loss": 2.1094,
      "step": 181026
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.3051815032958984,
      "learning_rate": 6.546057511815931e-05,
      "loss": 2.0582,
      "step": 181027
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.3136169910430908,
      "learning_rate": 6.545802454485375e-05,
      "loss": 2.1689,
      "step": 181028
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.4605860710144043,
      "learning_rate": 6.545547401515408e-05,
      "loss": 2.0921,
      "step": 181029
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.3859319686889648,
      "learning_rate": 6.545292352906055e-05,
      "loss": 1.9945,
      "step": 181030
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.2714051008224487,
      "learning_rate": 6.545037308657394e-05,
      "loss": 2.1118,
      "step": 181031
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.4342318773269653,
      "learning_rate": 6.544782268769436e-05,
      "loss": 2.0544,
      "step": 181032
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.3062723875045776,
      "learning_rate": 6.544527233242253e-05,
      "loss": 2.0665,
      "step": 181033
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.4272364377975464,
      "learning_rate": 6.544272202075877e-05,
      "loss": 2.0171,
      "step": 181034
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.338972806930542,
      "learning_rate": 6.544017175270374e-05,
      "loss": 2.0229,
      "step": 181035
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.3196078538894653,
      "learning_rate": 6.543762152825772e-05,
      "loss": 1.9785,
      "step": 181036
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.2960978746414185,
      "learning_rate": 6.543507134742142e-05,
      "loss": 1.902,
      "step": 181037
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.3804311752319336,
      "learning_rate": 6.5432521210195e-05,
      "loss": 1.9347,
      "step": 181038
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.3176454305648804,
      "learning_rate": 6.542997111657923e-05,
      "loss": 1.9203,
      "step": 181039
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.4084532260894775,
      "learning_rate": 6.542742106657433e-05,
      "loss": 1.7596,
      "step": 181040
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.2543145418167114,
      "learning_rate": 6.542487106018099e-05,
      "loss": 1.9337,
      "step": 181041
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.2679182291030884,
      "learning_rate": 6.54223210973995e-05,
      "loss": 2.0523,
      "step": 181042
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.4384492635726929,
      "learning_rate": 6.541977117823061e-05,
      "loss": 1.8646,
      "step": 181043
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.416393518447876,
      "learning_rate": 6.541722130267446e-05,
      "loss": 2.0541,
      "step": 181044
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.4875471591949463,
      "learning_rate": 6.541467147073177e-05,
      "loss": 1.8594,
      "step": 181045
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.5922574996948242,
      "learning_rate": 6.541212168240279e-05,
      "loss": 1.8454,
      "step": 181046
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.545466423034668,
      "learning_rate": 6.540957193768826e-05,
      "loss": 2.0521,
      "step": 181047
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.2127387523651123,
      "learning_rate": 6.54070222365884e-05,
      "loss": 1.8942,
      "step": 181048
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.3185126781463623,
      "learning_rate": 6.54044725791039e-05,
      "loss": 2.0426,
      "step": 181049
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.4679361581802368,
      "learning_rate": 6.54019229652351e-05,
      "loss": 2.0084,
      "step": 181050
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.4200479984283447,
      "learning_rate": 6.539937339498255e-05,
      "loss": 1.9322,
      "step": 181051
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.4506053924560547,
      "learning_rate": 6.53968238683466e-05,
      "loss": 2.0911,
      "step": 181052
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.3104937076568604,
      "learning_rate": 6.539427438532787e-05,
      "loss": 1.9646,
      "step": 181053
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.6876871585845947,
      "learning_rate": 6.539172494592672e-05,
      "loss": 2.0942,
      "step": 181054
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.2641618251800537,
      "learning_rate": 6.538917555014374e-05,
      "loss": 2.2789,
      "step": 181055
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.2942243814468384,
      "learning_rate": 6.538662619797936e-05,
      "loss": 2.1699,
      "step": 181056
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.3365322351455688,
      "learning_rate": 6.538407688943404e-05,
      "loss": 1.9565,
      "step": 181057
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.2700071334838867,
      "learning_rate": 6.53815276245081e-05,
      "loss": 1.9314,
      "step": 181058
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.3503073453903198,
      "learning_rate": 6.537897840320232e-05,
      "loss": 2.0535,
      "step": 181059
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.3617312908172607,
      "learning_rate": 6.53764292255169e-05,
      "loss": 1.8181,
      "step": 181060
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.2886443138122559,
      "learning_rate": 6.537388009145253e-05,
      "loss": 2.0437,
      "step": 181061
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.4436756372451782,
      "learning_rate": 6.537133100100957e-05,
      "loss": 1.9547,
      "step": 181062
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.514274001121521,
      "learning_rate": 6.536878195418844e-05,
      "loss": 2.024,
      "step": 181063
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.3136028051376343,
      "learning_rate": 6.536623295098977e-05,
      "loss": 2.1513,
      "step": 181064
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.2820497751235962,
      "learning_rate": 6.536368399141391e-05,
      "loss": 1.8325,
      "step": 181065
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.4295847415924072,
      "learning_rate": 6.536113507546131e-05,
      "loss": 1.7899,
      "step": 181066
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.2685482501983643,
      "learning_rate": 6.535858620313258e-05,
      "loss": 2.0191,
      "step": 181067
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.4040449857711792,
      "learning_rate": 6.535603737442811e-05,
      "loss": 2.258,
      "step": 181068
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.3418320417404175,
      "learning_rate": 6.535348858934832e-05,
      "loss": 1.9149,
      "step": 181069
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.383202314376831,
      "learning_rate": 6.535093984789384e-05,
      "loss": 1.9095,
      "step": 181070
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.2795991897583008,
      "learning_rate": 6.534839115006495e-05,
      "loss": 2.1359,
      "step": 181071
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.3589601516723633,
      "learning_rate": 6.53458424958623e-05,
      "loss": 2.1314,
      "step": 181072
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.3730357885360718,
      "learning_rate": 6.534329388528628e-05,
      "loss": 2.0678,
      "step": 181073
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.5598398447036743,
      "learning_rate": 6.534074531833741e-05,
      "loss": 2.0451,
      "step": 181074
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.3387784957885742,
      "learning_rate": 6.5338196795016e-05,
      "loss": 2.0335,
      "step": 181075
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.532808780670166,
      "learning_rate": 6.533564831532276e-05,
      "loss": 2.1174,
      "step": 181076
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.5205305814743042,
      "learning_rate": 6.533309987925795e-05,
      "loss": 1.9537,
      "step": 181077
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.3092454671859741,
      "learning_rate": 6.533055148682223e-05,
      "loss": 2.1197,
      "step": 181078
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.4197880029678345,
      "learning_rate": 6.532800313801599e-05,
      "loss": 1.9889,
      "step": 181079
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.4959964752197266,
      "learning_rate": 6.532545483283971e-05,
      "loss": 2.1704,
      "step": 181080
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.6121487617492676,
      "learning_rate": 6.53229065712938e-05,
      "loss": 2.0297,
      "step": 181081
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.4304687976837158,
      "learning_rate": 6.532035835337884e-05,
      "loss": 2.0083,
      "step": 181082
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.4693089723587036,
      "learning_rate": 6.531781017909518e-05,
      "loss": 2.0403,
      "step": 181083
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.4596095085144043,
      "learning_rate": 6.531526204844347e-05,
      "loss": 1.9443,
      "step": 181084
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.3709970712661743,
      "learning_rate": 6.531271396142405e-05,
      "loss": 1.8273,
      "step": 181085
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.3308327198028564,
      "learning_rate": 6.531016591803747e-05,
      "loss": 2.1306,
      "step": 181086
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.3172836303710938,
      "learning_rate": 6.530761791828405e-05,
      "loss": 1.8611,
      "step": 181087
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.409522533416748,
      "learning_rate": 6.530506996216445e-05,
      "loss": 2.0332,
      "step": 181088
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.3754292726516724,
      "learning_rate": 6.530252204967899e-05,
      "loss": 2.0036,
      "step": 181089
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.350586175918579,
      "learning_rate": 6.529997418082829e-05,
      "loss": 1.9484,
      "step": 181090
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.4033106565475464,
      "learning_rate": 6.52974263556128e-05,
      "loss": 1.9545,
      "step": 181091
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.4083173274993896,
      "learning_rate": 6.529487857403292e-05,
      "loss": 2.1639,
      "step": 181092
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.411940574645996,
      "learning_rate": 6.529233083608907e-05,
      "loss": 1.8278,
      "step": 181093
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.4376301765441895,
      "learning_rate": 6.528978314178188e-05,
      "loss": 2.0467,
      "step": 181094
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.2127761840820312,
      "learning_rate": 6.528723549111169e-05,
      "loss": 1.971,
      "step": 181095
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.4522831439971924,
      "learning_rate": 6.528468788407909e-05,
      "loss": 2.0092,
      "step": 181096
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.4720710515975952,
      "learning_rate": 6.528214032068442e-05,
      "loss": 1.9365,
      "step": 181097
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.4860690832138062,
      "learning_rate": 6.52795928009284e-05,
      "loss": 2.0207,
      "step": 181098
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.4492640495300293,
      "learning_rate": 6.527704532481118e-05,
      "loss": 1.9192,
      "step": 181099
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.3691983222961426,
      "learning_rate": 6.52744978923335e-05,
      "loss": 2.0772,
      "step": 181100
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.5947299003601074,
      "learning_rate": 6.527195050349558e-05,
      "loss": 2.031,
      "step": 181101
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.2860171794891357,
      "learning_rate": 6.526940315829814e-05,
      "loss": 1.8957,
      "step": 181102
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.2787402868270874,
      "learning_rate": 6.526685585674146e-05,
      "loss": 2.3035,
      "step": 181103
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.3081979751586914,
      "learning_rate": 6.526430859882629e-05,
      "loss": 1.9331,
      "step": 181104
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.2151576280593872,
      "learning_rate": 6.52617613845527e-05,
      "loss": 1.9351,
      "step": 181105
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.224472165107727,
      "learning_rate": 6.525921421392152e-05,
      "loss": 1.7316,
      "step": 181106
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.3655568361282349,
      "learning_rate": 6.525666708693296e-05,
      "loss": 2.0583,
      "step": 181107
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.3268423080444336,
      "learning_rate": 6.525412000358772e-05,
      "loss": 2.121,
      "step": 181108
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.3350191116333008,
      "learning_rate": 6.525157296388611e-05,
      "loss": 2.0115,
      "step": 181109
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.343565583229065,
      "learning_rate": 6.524902596782878e-05,
      "loss": 2.1228,
      "step": 181110
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.4401257038116455,
      "learning_rate": 6.524647901541595e-05,
      "loss": 2.0338,
      "step": 181111
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.295019507408142,
      "learning_rate": 6.524393210664832e-05,
      "loss": 1.822,
      "step": 181112
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.3004801273345947,
      "learning_rate": 6.524138524152619e-05,
      "loss": 2.1612,
      "step": 181113
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.3885016441345215,
      "learning_rate": 6.523883842005017e-05,
      "loss": 1.9913,
      "step": 181114
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.4060238599777222,
      "learning_rate": 6.523629164222061e-05,
      "loss": 2.2635,
      "step": 181115
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.3543781042099,
      "learning_rate": 6.523374490803824e-05,
      "loss": 1.9615,
      "step": 181116
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.3034173250198364,
      "learning_rate": 6.523119821750319e-05,
      "loss": 2.022,
      "step": 181117
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.341288447380066,
      "learning_rate": 6.522865157061615e-05,
      "loss": 2.2146,
      "step": 181118
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.3761669397354126,
      "learning_rate": 6.522610496737747e-05,
      "loss": 1.9523,
      "step": 181119
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.3035913705825806,
      "learning_rate": 6.522355840778776e-05,
      "loss": 2.078,
      "step": 181120
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.408003568649292,
      "learning_rate": 6.522101189184735e-05,
      "loss": 2.106,
      "step": 181121
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.3193353414535522,
      "learning_rate": 6.521846541955692e-05,
      "loss": 1.9097,
      "step": 181122
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.265721082687378,
      "learning_rate": 6.521591899091667e-05,
      "loss": 1.9091,
      "step": 181123
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.923565149307251,
      "learning_rate": 6.52133726059273e-05,
      "loss": 2.1017,
      "step": 181124
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.2912709712982178,
      "learning_rate": 6.521082626458912e-05,
      "loss": 1.9568,
      "step": 181125
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.2792483568191528,
      "learning_rate": 6.520827996690272e-05,
      "loss": 1.9572,
      "step": 181126
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.487092137336731,
      "learning_rate": 6.520573371286851e-05,
      "loss": 1.7275,
      "step": 181127
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.3949065208435059,
      "learning_rate": 6.52031875024871e-05,
      "loss": 2.0436,
      "step": 181128
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.4560940265655518,
      "learning_rate": 6.520064133575871e-05,
      "loss": 1.9852,
      "step": 181129
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.3886151313781738,
      "learning_rate": 6.519809521268403e-05,
      "loss": 2.265,
      "step": 181130
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.5056090354919434,
      "learning_rate": 6.519554913326337e-05,
      "loss": 2.0589,
      "step": 181131
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.3454276323318481,
      "learning_rate": 6.519300309749738e-05,
      "loss": 2.0356,
      "step": 181132
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.2902616262435913,
      "learning_rate": 6.519045710538636e-05,
      "loss": 1.9581,
      "step": 181133
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.3823524713516235,
      "learning_rate": 6.518791115693103e-05,
      "loss": 1.8711,
      "step": 181134
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.4165593385696411,
      "learning_rate": 6.518536525213151e-05,
      "loss": 1.8838,
      "step": 181135
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.3511043787002563,
      "learning_rate": 6.518281939098859e-05,
      "loss": 2.0519,
      "step": 181136
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.3116320371627808,
      "learning_rate": 6.51802735735025e-05,
      "loss": 1.9882,
      "step": 181137
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.5001344680786133,
      "learning_rate": 6.517772779967392e-05,
      "loss": 2.0944,
      "step": 181138
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.3216372728347778,
      "learning_rate": 6.517518206950315e-05,
      "loss": 2.1529,
      "step": 181139
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.3868745565414429,
      "learning_rate": 6.517263638299082e-05,
      "loss": 2.0264,
      "step": 181140
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.3121498823165894,
      "learning_rate": 6.517009074013734e-05,
      "loss": 2.0786,
      "step": 181141
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.2979068756103516,
      "learning_rate": 6.516754514094318e-05,
      "loss": 1.7903,
      "step": 181142
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.3833087682724,
      "learning_rate": 6.516499958540872e-05,
      "loss": 2.1323,
      "step": 181143
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.3673913478851318,
      "learning_rate": 6.51624540735346e-05,
      "loss": 1.9477,
      "step": 181144
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.2954198122024536,
      "learning_rate": 6.515990860532115e-05,
      "loss": 2.0471,
      "step": 181145
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.6128140687942505,
      "learning_rate": 6.515736318076895e-05,
      "loss": 2.1118,
      "step": 181146
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.3851264715194702,
      "learning_rate": 6.515481779987846e-05,
      "loss": 1.9815,
      "step": 181147
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.3340345621109009,
      "learning_rate": 6.515227246265003e-05,
      "loss": 2.0597,
      "step": 181148
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.3415478467941284,
      "learning_rate": 6.514972716908429e-05,
      "loss": 1.9907,
      "step": 181149
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.478391408920288,
      "learning_rate": 6.514718191918168e-05,
      "loss": 1.9472,
      "step": 181150
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.9110417366027832,
      "learning_rate": 6.514463671294254e-05,
      "loss": 1.9076,
      "step": 181151
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.3195366859436035,
      "learning_rate": 6.514209155036754e-05,
      "loss": 1.8899,
      "step": 181152
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.3119966983795166,
      "learning_rate": 6.513954643145706e-05,
      "loss": 1.9162,
      "step": 181153
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.294306755065918,
      "learning_rate": 6.513700135621147e-05,
      "loss": 1.9982,
      "step": 181154
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.2406864166259766,
      "learning_rate": 6.513445632463142e-05,
      "loss": 1.9349,
      "step": 181155
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.2971481084823608,
      "learning_rate": 6.513191133671735e-05,
      "loss": 1.901,
      "step": 181156
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.471140742301941,
      "learning_rate": 6.512936639246958e-05,
      "loss": 1.7824,
      "step": 181157
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.2752927541732788,
      "learning_rate": 6.512682149188877e-05,
      "loss": 1.8618,
      "step": 181158
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.3782720565795898,
      "learning_rate": 6.512427663497534e-05,
      "loss": 1.9899,
      "step": 181159
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.2900030612945557,
      "learning_rate": 6.512173182172964e-05,
      "loss": 1.9131,
      "step": 181160
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.4128847122192383,
      "learning_rate": 6.511918705215234e-05,
      "loss": 2.2113,
      "step": 181161
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.5038357973098755,
      "learning_rate": 6.511664232624371e-05,
      "loss": 2.039,
      "step": 181162
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.7120141983032227,
      "learning_rate": 6.511409764400443e-05,
      "loss": 2.0845,
      "step": 181163
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.5859310626983643,
      "learning_rate": 6.511155300543487e-05,
      "loss": 1.6882,
      "step": 181164
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.404543399810791,
      "learning_rate": 6.510900841053551e-05,
      "loss": 1.9717,
      "step": 181165
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.358313798904419,
      "learning_rate": 6.510646385930676e-05,
      "loss": 2.0506,
      "step": 181166
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.3907047510147095,
      "learning_rate": 6.51039193517492e-05,
      "loss": 2.0612,
      "step": 181167
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.3230985403060913,
      "learning_rate": 6.510137488786318e-05,
      "loss": 1.7428,
      "step": 181168
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.4228003025054932,
      "learning_rate": 6.509883046764933e-05,
      "loss": 1.7958,
      "step": 181169
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.4146071672439575,
      "learning_rate": 6.509628609110796e-05,
      "loss": 1.9062,
      "step": 181170
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.2850619554519653,
      "learning_rate": 6.50937417582398e-05,
      "loss": 2.0323,
      "step": 181171
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.4010318517684937,
      "learning_rate": 6.509119746904497e-05,
      "loss": 2.1155,
      "step": 181172
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.2618515491485596,
      "learning_rate": 6.50886532235242e-05,
      "loss": 2.2108,
      "step": 181173
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.3763281106948853,
      "learning_rate": 6.508610902167781e-05,
      "loss": 1.9109,
      "step": 181174
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.5310063362121582,
      "learning_rate": 6.508356486350645e-05,
      "loss": 2.0774,
      "step": 181175
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.3997703790664673,
      "learning_rate": 6.508102074901038e-05,
      "loss": 1.8602,
      "step": 181176
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.2928625345230103,
      "learning_rate": 6.507847667819036e-05,
      "loss": 2.1074,
      "step": 181177
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.452171802520752,
      "learning_rate": 6.507593265104652e-05,
      "loss": 2.1014,
      "step": 181178
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.3393436670303345,
      "learning_rate": 6.507338866757956e-05,
      "loss": 1.8404,
      "step": 181179
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.3198344707489014,
      "learning_rate": 6.507084472778985e-05,
      "loss": 1.8348,
      "step": 181180
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.3306690454483032,
      "learning_rate": 6.506830083167795e-05,
      "loss": 2.2424,
      "step": 181181
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.3550199270248413,
      "learning_rate": 6.506575697924421e-05,
      "loss": 1.9812,
      "step": 181182
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.2613258361816406,
      "learning_rate": 6.506321317048934e-05,
      "loss": 1.7978,
      "step": 181183
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.3162754774093628,
      "learning_rate": 6.506066940541352e-05,
      "loss": 2.0819,
      "step": 181184
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.2793229818344116,
      "learning_rate": 6.50581256840174e-05,
      "loss": 2.1363,
      "step": 181185
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.2173748016357422,
      "learning_rate": 6.505558200630134e-05,
      "loss": 2.0268,
      "step": 181186
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.5063501596450806,
      "learning_rate": 6.505303837226598e-05,
      "loss": 2.197,
      "step": 181187
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.565293788909912,
      "learning_rate": 6.50504947819116e-05,
      "loss": 1.9089,
      "step": 181188
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.4577668905258179,
      "learning_rate": 6.504795123523895e-05,
      "loss": 1.914,
      "step": 181189
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.4184253215789795,
      "learning_rate": 6.504540773224815e-05,
      "loss": 2.3085,
      "step": 181190
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.3911341428756714,
      "learning_rate": 6.504286427293992e-05,
      "loss": 2.031,
      "step": 181191
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.3099409341812134,
      "learning_rate": 6.504032085731458e-05,
      "loss": 1.984,
      "step": 181192
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.3806872367858887,
      "learning_rate": 6.503777748537275e-05,
      "loss": 2.0738,
      "step": 181193
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.8217660188674927,
      "learning_rate": 6.503523415711476e-05,
      "loss": 2.1276,
      "step": 181194
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.296674370765686,
      "learning_rate": 6.503269087254131e-05,
      "loss": 2.1504,
      "step": 181195
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.38331139087677,
      "learning_rate": 6.503014763165257e-05,
      "loss": 2.0165,
      "step": 181196
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.4724422693252563,
      "learning_rate": 6.50276044344492e-05,
      "loss": 2.0053,
      "step": 181197
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.2998276948928833,
      "learning_rate": 6.502506128093159e-05,
      "loss": 2.0656,
      "step": 181198
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.2320412397384644,
      "learning_rate": 6.502251817110034e-05,
      "loss": 1.9177,
      "step": 181199
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.3470054864883423,
      "learning_rate": 6.501997510495573e-05,
      "loss": 2.0866,
      "step": 181200
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.7016934156417847,
      "learning_rate": 6.501743208249856e-05,
      "loss": 2.0069,
      "step": 181201
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.43336021900177,
      "learning_rate": 6.501488910372886e-05,
      "loss": 2.0006,
      "step": 181202
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.3136603832244873,
      "learning_rate": 6.501234616864745e-05,
      "loss": 2.1104,
      "step": 181203
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.3111268281936646,
      "learning_rate": 6.50098032772546e-05,
      "loss": 2.0129,
      "step": 181204
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.3227636814117432,
      "learning_rate": 6.500726042955094e-05,
      "loss": 1.9381,
      "step": 181205
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.4276539087295532,
      "learning_rate": 6.500471762553676e-05,
      "loss": 1.9341,
      "step": 181206
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.4336129426956177,
      "learning_rate": 6.500217486521282e-05,
      "loss": 1.889,
      "step": 181207
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.2680189609527588,
      "learning_rate": 6.499963214857924e-05,
      "loss": 2.0343,
      "step": 181208
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.3208599090576172,
      "learning_rate": 6.499708947563674e-05,
      "loss": 1.9824,
      "step": 181209
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.2888820171356201,
      "learning_rate": 6.499454684638566e-05,
      "loss": 2.1163,
      "step": 181210
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.3395676612854004,
      "learning_rate": 6.49920042608266e-05,
      "loss": 2.4095,
      "step": 181211
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.388177752494812,
      "learning_rate": 6.498946171895986e-05,
      "loss": 1.8758,
      "step": 181212
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.4016486406326294,
      "learning_rate": 6.498691922078623e-05,
      "loss": 1.992,
      "step": 181213
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.396300196647644,
      "learning_rate": 6.498437676630575e-05,
      "loss": 2.082,
      "step": 181214
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.3925995826721191,
      "learning_rate": 6.498183435551921e-05,
      "loss": 2.0017,
      "step": 181215
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.3359614610671997,
      "learning_rate": 6.497929198842693e-05,
      "loss": 2.1049,
      "step": 181216
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.3468879461288452,
      "learning_rate": 6.49767496650295e-05,
      "loss": 2.193,
      "step": 181217
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.4711661338806152,
      "learning_rate": 6.49742073853272e-05,
      "loss": 2.1008,
      "step": 181218
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.4276957511901855,
      "learning_rate": 6.497166514932086e-05,
      "loss": 1.8438,
      "step": 181219
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.3113603591918945,
      "learning_rate": 6.496912295701053e-05,
      "loss": 2.0748,
      "step": 181220
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.454981803894043,
      "learning_rate": 6.496658080839698e-05,
      "loss": 2.2001,
      "step": 181221
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.3899555206298828,
      "learning_rate": 6.496403870348048e-05,
      "loss": 2.1227,
      "step": 181222
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.4691256284713745,
      "learning_rate": 6.496149664226169e-05,
      "loss": 1.9478,
      "step": 181223
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.3085371255874634,
      "learning_rate": 6.495895462474088e-05,
      "loss": 1.8522,
      "step": 181224
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.3229238986968994,
      "learning_rate": 6.495641265091877e-05,
      "loss": 2.076,
      "step": 181225
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.4571973085403442,
      "learning_rate": 6.495387072079565e-05,
      "loss": 2.2241,
      "step": 181226
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.266048550605774,
      "learning_rate": 6.495132883437207e-05,
      "loss": 1.8715,
      "step": 181227
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.4068323373794556,
      "learning_rate": 6.49487869916484e-05,
      "loss": 2.253,
      "step": 181228
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.336705207824707,
      "learning_rate": 6.494624519262528e-05,
      "loss": 2.1163,
      "step": 181229
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.4960657358169556,
      "learning_rate": 6.4943703437303e-05,
      "loss": 1.9269,
      "step": 181230
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.4781363010406494,
      "learning_rate": 6.494116172568219e-05,
      "loss": 2.1231,
      "step": 181231
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.3597050905227661,
      "learning_rate": 6.493862005776325e-05,
      "loss": 2.0032,
      "step": 181232
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.328518033027649,
      "learning_rate": 6.493607843354657e-05,
      "loss": 1.8435,
      "step": 181233
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.3136906623840332,
      "learning_rate": 6.493353685303282e-05,
      "loss": 1.9026,
      "step": 181234
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.3110835552215576,
      "learning_rate": 6.493099531622235e-05,
      "loss": 2.0104,
      "step": 181235
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.2716420888900757,
      "learning_rate": 6.492845382311555e-05,
      "loss": 2.0899,
      "step": 181236
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.4311546087265015,
      "learning_rate": 6.492591237371309e-05,
      "loss": 2.1285,
      "step": 181237
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.8112941980361938,
      "learning_rate": 6.492337096801533e-05,
      "loss": 1.8766,
      "step": 181238
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.2245335578918457,
      "learning_rate": 6.492082960602268e-05,
      "loss": 2.1584,
      "step": 181239
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.2260099649429321,
      "learning_rate": 6.491828828773577e-05,
      "loss": 2.0283,
      "step": 181240
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.2572864294052124,
      "learning_rate": 6.491574701315497e-05,
      "loss": 2.1282,
      "step": 181241
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.2896474599838257,
      "learning_rate": 6.491320578228071e-05,
      "loss": 1.8894,
      "step": 181242
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.3765658140182495,
      "learning_rate": 6.491066459511359e-05,
      "loss": 2.2949,
      "step": 181243
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.2778288125991821,
      "learning_rate": 6.490812345165399e-05,
      "loss": 2.1497,
      "step": 181244
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.2488785982131958,
      "learning_rate": 6.490558235190238e-05,
      "loss": 1.9462,
      "step": 181245
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.3994672298431396,
      "learning_rate": 6.49030412958593e-05,
      "loss": 2.092,
      "step": 181246
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.4466408491134644,
      "learning_rate": 6.490050028352512e-05,
      "loss": 2.2201,
      "step": 181247
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.4561777114868164,
      "learning_rate": 6.489795931490044e-05,
      "loss": 2.0738,
      "step": 181248
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.4100035429000854,
      "learning_rate": 6.48954183899857e-05,
      "loss": 2.1708,
      "step": 181249
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.3556593656539917,
      "learning_rate": 6.48928775087813e-05,
      "loss": 2.0095,
      "step": 181250
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.4837346076965332,
      "learning_rate": 6.489033667128769e-05,
      "loss": 1.9512,
      "step": 181251
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.5479223728179932,
      "learning_rate": 6.488779587750549e-05,
      "loss": 2.2051,
      "step": 181252
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.3671536445617676,
      "learning_rate": 6.488525512743499e-05,
      "loss": 2.0515,
      "step": 181253
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.2487530708312988,
      "learning_rate": 6.488271442107686e-05,
      "loss": 1.9874,
      "step": 181254
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.5394818782806396,
      "learning_rate": 6.488017375843145e-05,
      "loss": 2.1327,
      "step": 181255
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.384678602218628,
      "learning_rate": 6.487763313949928e-05,
      "loss": 2.139,
      "step": 181256
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.3412145376205444,
      "learning_rate": 6.487509256428072e-05,
      "loss": 1.9299,
      "step": 181257
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.309136152267456,
      "learning_rate": 6.487255203277635e-05,
      "loss": 2.1925,
      "step": 181258
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.2399568557739258,
      "learning_rate": 6.487001154498657e-05,
      "loss": 2.2744,
      "step": 181259
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.5657227039337158,
      "learning_rate": 6.486747110091197e-05,
      "loss": 1.7428,
      "step": 181260
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.4460523128509521,
      "learning_rate": 6.486493070055286e-05,
      "loss": 1.968,
      "step": 181261
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.3207734823226929,
      "learning_rate": 6.486239034390995e-05,
      "loss": 1.8765,
      "step": 181262
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.3520805835723877,
      "learning_rate": 6.485985003098341e-05,
      "loss": 1.978,
      "step": 181263
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.305235743522644,
      "learning_rate": 6.485730976177396e-05,
      "loss": 2.1765,
      "step": 181264
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.337833285331726,
      "learning_rate": 6.485476953628186e-05,
      "loss": 1.8637,
      "step": 181265
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.4330304861068726,
      "learning_rate": 6.48522293545078e-05,
      "loss": 2.1918,
      "step": 181266
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.3730953931808472,
      "learning_rate": 6.484968921645205e-05,
      "loss": 2.0132,
      "step": 181267
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.486240267753601,
      "learning_rate": 6.484714912211538e-05,
      "loss": 1.8798,
      "step": 181268
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.301222562789917,
      "learning_rate": 6.484460907149787e-05,
      "loss": 1.9061,
      "step": 181269
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.4905474185943604,
      "learning_rate": 6.484206906460027e-05,
      "loss": 1.944,
      "step": 181270
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.5408152341842651,
      "learning_rate": 6.483952910142288e-05,
      "loss": 2.0241,
      "step": 181271
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.7585505247116089,
      "learning_rate": 6.483698918196636e-05,
      "loss": 1.9159,
      "step": 181272
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.4479448795318604,
      "learning_rate": 6.4834449306231e-05,
      "loss": 2.1995,
      "step": 181273
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.4303622245788574,
      "learning_rate": 6.483190947421752e-05,
      "loss": 1.9581,
      "step": 181274
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.4846183061599731,
      "learning_rate": 6.482936968592605e-05,
      "loss": 1.9233,
      "step": 181275
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.285382866859436,
      "learning_rate": 6.482682994135733e-05,
      "loss": 2.0521,
      "step": 181276
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.3637357950210571,
      "learning_rate": 6.482429024051167e-05,
      "loss": 1.9845,
      "step": 181277
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.4749057292938232,
      "learning_rate": 6.482175058338967e-05,
      "loss": 1.914,
      "step": 181278
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.6927319765090942,
      "learning_rate": 6.481921096999168e-05,
      "loss": 1.9214,
      "step": 181279
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.3662593364715576,
      "learning_rate": 6.48166714003184e-05,
      "loss": 1.9992,
      "step": 181280
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.411306381225586,
      "learning_rate": 6.481413187436996e-05,
      "loss": 2.3075,
      "step": 181281
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.437475562095642,
      "learning_rate": 6.481159239214712e-05,
      "loss": 2.0175,
      "step": 181282
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.2121399641036987,
      "learning_rate": 6.480905295365014e-05,
      "loss": 2.1341,
      "step": 181283
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.512317180633545,
      "learning_rate": 6.480651355887971e-05,
      "loss": 2.1171,
      "step": 181284
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.554787039756775,
      "learning_rate": 6.480397420783607e-05,
      "loss": 2.1714,
      "step": 181285
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.3595689535140991,
      "learning_rate": 6.480143490052e-05,
      "loss": 2.0212,
      "step": 181286
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.3273340463638306,
      "learning_rate": 6.479889563693159e-05,
      "loss": 1.93,
      "step": 181287
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.3594164848327637,
      "learning_rate": 6.479635641707162e-05,
      "loss": 1.9198,
      "step": 181288
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.2830358743667603,
      "learning_rate": 6.479381724094035e-05,
      "loss": 1.9401,
      "step": 181289
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.4090813398361206,
      "learning_rate": 6.479127810853846e-05,
      "loss": 2.1042,
      "step": 181290
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.3420672416687012,
      "learning_rate": 6.478873901986618e-05,
      "loss": 2.1829,
      "step": 181291
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.2316097021102905,
      "learning_rate": 6.478619997492431e-05,
      "loss": 1.8505,
      "step": 181292
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.328077793121338,
      "learning_rate": 6.478366097371293e-05,
      "loss": 2.1105,
      "step": 181293
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.426536202430725,
      "learning_rate": 6.478112201623284e-05,
      "loss": 2.0445,
      "step": 181294
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.3808265924453735,
      "learning_rate": 6.477858310248428e-05,
      "loss": 2.3125,
      "step": 181295
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.4559123516082764,
      "learning_rate": 6.477604423246789e-05,
      "loss": 2.017,
      "step": 181296
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.3899437189102173,
      "learning_rate": 6.477350540618398e-05,
      "loss": 2.0707,
      "step": 181297
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.1986863613128662,
      "learning_rate": 6.477096662363322e-05,
      "loss": 2.009,
      "step": 181298
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.4271714687347412,
      "learning_rate": 6.476842788481596e-05,
      "loss": 2.1769,
      "step": 181299
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.375427007675171,
      "learning_rate": 6.47658891897327e-05,
      "loss": 2.0791,
      "step": 181300
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.3055400848388672,
      "learning_rate": 6.476335053838383e-05,
      "loss": 2.0079,
      "step": 181301
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.371282935142517,
      "learning_rate": 6.476081193076997e-05,
      "loss": 1.856,
      "step": 181302
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.3654686212539673,
      "learning_rate": 6.475827336689144e-05,
      "loss": 2.1618,
      "step": 181303
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.3427543640136719,
      "learning_rate": 6.475573484674884e-05,
      "loss": 1.9201,
      "step": 181304
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.4648590087890625,
      "learning_rate": 6.47531963703426e-05,
      "loss": 1.9636,
      "step": 181305
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.348929762840271,
      "learning_rate": 6.47506579376732e-05,
      "loss": 2.0329,
      "step": 181306
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.4025673866271973,
      "learning_rate": 6.474811954874098e-05,
      "loss": 2.3125,
      "step": 181307
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.4751348495483398,
      "learning_rate": 6.474558120354665e-05,
      "loss": 2.0693,
      "step": 181308
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.4783568382263184,
      "learning_rate": 6.474304290209042e-05,
      "loss": 2.011,
      "step": 181309
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.3463114500045776,
      "learning_rate": 6.474050464437305e-05,
      "loss": 2.052,
      "step": 181310
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.402127742767334,
      "learning_rate": 6.473796643039482e-05,
      "loss": 1.7707,
      "step": 181311
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.2797695398330688,
      "learning_rate": 6.473542826015623e-05,
      "loss": 2.2167,
      "step": 181312
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.4308888912200928,
      "learning_rate": 6.473289013365772e-05,
      "loss": 1.9786,
      "step": 181313
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.3617875576019287,
      "learning_rate": 6.473035205089986e-05,
      "loss": 2.1724,
      "step": 181314
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.4674973487854004,
      "learning_rate": 6.472781401188301e-05,
      "loss": 2.0993,
      "step": 181315
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.4240434169769287,
      "learning_rate": 6.472527601660776e-05,
      "loss": 1.85,
      "step": 181316
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.3334406614303589,
      "learning_rate": 6.472273806507453e-05,
      "loss": 2.158,
      "step": 181317
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.3725348711013794,
      "learning_rate": 6.472020015728378e-05,
      "loss": 2.0526,
      "step": 181318
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.2888309955596924,
      "learning_rate": 6.471766229323589e-05,
      "loss": 1.91,
      "step": 181319
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.400430679321289,
      "learning_rate": 6.471512447293156e-05,
      "loss": 1.9721,
      "step": 181320
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.3682570457458496,
      "learning_rate": 6.4712586696371e-05,
      "loss": 1.8986,
      "step": 181321
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.3175010681152344,
      "learning_rate": 6.47100489635549e-05,
      "loss": 2.0451,
      "step": 181322
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.2964556217193604,
      "learning_rate": 6.470751127448366e-05,
      "loss": 2.0003,
      "step": 181323
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.4218417406082153,
      "learning_rate": 6.470497362915765e-05,
      "loss": 1.8202,
      "step": 181324
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.3348208665847778,
      "learning_rate": 6.470243602757752e-05,
      "loss": 2.0229,
      "step": 181325
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.5936636924743652,
      "learning_rate": 6.469989846974363e-05,
      "loss": 1.9383,
      "step": 181326
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.4451671838760376,
      "learning_rate": 6.469736095565638e-05,
      "loss": 2.2351,
      "step": 181327
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.4176629781723022,
      "learning_rate": 6.469482348531642e-05,
      "loss": 2.3665,
      "step": 181328
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.5332121849060059,
      "learning_rate": 6.469228605872415e-05,
      "loss": 1.9479,
      "step": 181329
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.4597647190093994,
      "learning_rate": 6.468974867587995e-05,
      "loss": 2.0075,
      "step": 181330
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.2068911790847778,
      "learning_rate": 6.468721133678442e-05,
      "loss": 2.0464,
      "step": 181331
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.3515557050704956,
      "learning_rate": 6.468467404143792e-05,
      "loss": 1.8744,
      "step": 181332
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.4797428846359253,
      "learning_rate": 6.468213678984107e-05,
      "loss": 1.8915,
      "step": 181333
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.3956128358840942,
      "learning_rate": 6.467959958199421e-05,
      "loss": 1.8905,
      "step": 181334
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.3110333681106567,
      "learning_rate": 6.46770624178979e-05,
      "loss": 1.7992,
      "step": 181335
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.3505795001983643,
      "learning_rate": 6.467452529755244e-05,
      "loss": 2.1526,
      "step": 181336
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.296779990196228,
      "learning_rate": 6.467198822095853e-05,
      "loss": 2.1546,
      "step": 181337
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.46621572971344,
      "learning_rate": 6.466945118811649e-05,
      "loss": 2.0588,
      "step": 181338
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.3921992778778076,
      "learning_rate": 6.466691419902689e-05,
      "loss": 2.0493,
      "step": 181339
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.4404237270355225,
      "learning_rate": 6.466437725369015e-05,
      "loss": 2.1659,
      "step": 181340
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.2894999980926514,
      "learning_rate": 6.466184035210677e-05,
      "loss": 1.9528,
      "step": 181341
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.4477102756500244,
      "learning_rate": 6.46593034942771e-05,
      "loss": 1.9121,
      "step": 181342
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.307059645652771,
      "learning_rate": 6.46567666802018e-05,
      "loss": 2.0177,
      "step": 181343
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.4380180835723877,
      "learning_rate": 6.465422990988113e-05,
      "loss": 1.9439,
      "step": 181344
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.4036751985549927,
      "learning_rate": 6.465169318331579e-05,
      "loss": 1.7619,
      "step": 181345
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.246665358543396,
      "learning_rate": 6.464915650050609e-05,
      "loss": 2.2165,
      "step": 181346
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.1973319053649902,
      "learning_rate": 6.464661986145267e-05,
      "loss": 1.9475,
      "step": 181347
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.332582950592041,
      "learning_rate": 6.464408326615573e-05,
      "loss": 1.9759,
      "step": 181348
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.2505804300308228,
      "learning_rate": 6.464154671461598e-05,
      "loss": 2.067,
      "step": 181349
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.3663047552108765,
      "learning_rate": 6.463901020683374e-05,
      "loss": 2.1242,
      "step": 181350
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.3483413457870483,
      "learning_rate": 6.463647374280965e-05,
      "loss": 1.9458,
      "step": 181351
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.4216318130493164,
      "learning_rate": 6.463393732254397e-05,
      "loss": 1.9484,
      "step": 181352
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.3609395027160645,
      "learning_rate": 6.463140094603745e-05,
      "loss": 1.8973,
      "step": 181353
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.293201208114624,
      "learning_rate": 6.462886461329026e-05,
      "loss": 1.8905,
      "step": 181354
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.4766088724136353,
      "learning_rate": 6.462632832430306e-05,
      "loss": 1.977,
      "step": 181355
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.4537138938903809,
      "learning_rate": 6.462379207907619e-05,
      "loss": 2.0904,
      "step": 181356
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.285888433456421,
      "learning_rate": 6.462125587761028e-05,
      "loss": 1.9241,
      "step": 181357
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.2967315912246704,
      "learning_rate": 6.461871971990566e-05,
      "loss": 2.1605,
      "step": 181358
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.5156621932983398,
      "learning_rate": 6.461618360596301e-05,
      "loss": 2.208,
      "step": 181359
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.4543969631195068,
      "learning_rate": 6.461364753578248e-05,
      "loss": 2.1167,
      "step": 181360
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.5146996974945068,
      "learning_rate": 6.461111150936482e-05,
      "loss": 2.1164,
      "step": 181361
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.330858826637268,
      "learning_rate": 6.460857552671032e-05,
      "loss": 1.9892,
      "step": 181362
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.8336106538772583,
      "learning_rate": 6.46060395878196e-05,
      "loss": 2.0959,
      "step": 181363
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.3322882652282715,
      "learning_rate": 6.460350369269299e-05,
      "loss": 2.0728,
      "step": 181364
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.4000643491744995,
      "learning_rate": 6.460096784133108e-05,
      "loss": 1.8017,
      "step": 181365
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.4681800603866577,
      "learning_rate": 6.459843203373432e-05,
      "loss": 2.0259,
      "step": 181366
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.413684368133545,
      "learning_rate": 6.459589626990317e-05,
      "loss": 1.896,
      "step": 181367
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.432064414024353,
      "learning_rate": 6.459336054983797e-05,
      "loss": 1.9519,
      "step": 181368
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.2403982877731323,
      "learning_rate": 6.459082487353942e-05,
      "loss": 1.9611,
      "step": 181369
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.3342657089233398,
      "learning_rate": 6.458828924100777e-05,
      "loss": 2.0279,
      "step": 181370
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.2387287616729736,
      "learning_rate": 6.458575365224369e-05,
      "loss": 1.945,
      "step": 181371
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.2891520261764526,
      "learning_rate": 6.45832181072476e-05,
      "loss": 1.8288,
      "step": 181372
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.3550184965133667,
      "learning_rate": 6.458068260601988e-05,
      "loss": 2.0803,
      "step": 181373
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.3934693336486816,
      "learning_rate": 6.4578147148561e-05,
      "loss": 2.0493,
      "step": 181374
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.338671326637268,
      "learning_rate": 6.457561173487156e-05,
      "loss": 1.7766,
      "step": 181375
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.5109046697616577,
      "learning_rate": 6.457307636495186e-05,
      "loss": 1.9944,
      "step": 181376
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.3714557886123657,
      "learning_rate": 6.457054103880257e-05,
      "loss": 1.9311,
      "step": 181377
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.242769718170166,
      "learning_rate": 6.456800575642407e-05,
      "loss": 2.0854,
      "step": 181378
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.5983495712280273,
      "learning_rate": 6.456547051781682e-05,
      "loss": 2.0903,
      "step": 181379
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.5540395975112915,
      "learning_rate": 6.456293532298118e-05,
      "loss": 1.9821,
      "step": 181380
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.4708491563796997,
      "learning_rate": 6.456040017191785e-05,
      "loss": 2.1547,
      "step": 181381
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.682705283164978,
      "learning_rate": 6.45578650646271e-05,
      "loss": 1.9611,
      "step": 181382
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.600739598274231,
      "learning_rate": 6.455533000110955e-05,
      "loss": 1.8196,
      "step": 181383
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.2394927740097046,
      "learning_rate": 6.455279498136562e-05,
      "loss": 1.8817,
      "step": 181384
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.4258103370666504,
      "learning_rate": 6.455026000539579e-05,
      "loss": 2.1975,
      "step": 181385
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.6017869710922241,
      "learning_rate": 6.45477250732004e-05,
      "loss": 1.9973,
      "step": 181386
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.4092464447021484,
      "learning_rate": 6.454519018478014e-05,
      "loss": 2.1244,
      "step": 181387
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.4562230110168457,
      "learning_rate": 6.454265534013529e-05,
      "loss": 1.8804,
      "step": 181388
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.3229484558105469,
      "learning_rate": 6.454012053926645e-05,
      "loss": 2.259,
      "step": 181389
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.2507202625274658,
      "learning_rate": 6.453758578217406e-05,
      "loss": 1.8274,
      "step": 181390
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.5471998453140259,
      "learning_rate": 6.453505106885859e-05,
      "loss": 1.996,
      "step": 181391
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.3682512044906616,
      "learning_rate": 6.453251639932044e-05,
      "loss": 1.8335,
      "step": 181392
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.306979775428772,
      "learning_rate": 6.452998177356019e-05,
      "loss": 2.0584,
      "step": 181393
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.3246009349822998,
      "learning_rate": 6.452744719157816e-05,
      "loss": 2.0141,
      "step": 181394
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.3616924285888672,
      "learning_rate": 6.452491265337505e-05,
      "loss": 2.2256,
      "step": 181395
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.23550546169281,
      "learning_rate": 6.452237815895118e-05,
      "loss": 2.0386,
      "step": 181396
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.4299606084823608,
      "learning_rate": 6.451984370830706e-05,
      "loss": 2.1586,
      "step": 181397
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.373621940612793,
      "learning_rate": 6.451730930144305e-05,
      "loss": 1.9602,
      "step": 181398
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.3459763526916504,
      "learning_rate": 6.451477493835979e-05,
      "loss": 2.1585,
      "step": 181399
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.383209228515625,
      "learning_rate": 6.45122406190576e-05,
      "loss": 2.0041,
      "step": 181400
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.3133145570755005,
      "learning_rate": 6.450970634353714e-05,
      "loss": 1.9506,
      "step": 181401
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.369336485862732,
      "learning_rate": 6.450717211179874e-05,
      "loss": 2.0957,
      "step": 181402
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.4684821367263794,
      "learning_rate": 6.450463792384293e-05,
      "loss": 2.058,
      "step": 181403
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.4361659288406372,
      "learning_rate": 6.450210377967002e-05,
      "loss": 2.0369,
      "step": 181404
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.4035682678222656,
      "learning_rate": 6.449956967928077e-05,
      "loss": 2.1461,
      "step": 181405
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.417149305343628,
      "learning_rate": 6.449703562267534e-05,
      "loss": 2.0197,
      "step": 181406
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.3045028448104858,
      "learning_rate": 6.449450160985449e-05,
      "loss": 2.1671,
      "step": 181407
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.2825247049331665,
      "learning_rate": 6.449196764081854e-05,
      "loss": 1.829,
      "step": 181408
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.2737109661102295,
      "learning_rate": 6.448943371556787e-05,
      "loss": 1.9729,
      "step": 181409
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.366160273551941,
      "learning_rate": 6.448689983410319e-05,
      "loss": 1.9836,
      "step": 181410
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.366660475730896,
      "learning_rate": 6.448436599642482e-05,
      "loss": 1.8418,
      "step": 181411
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.3261855840682983,
      "learning_rate": 6.448183220253318e-05,
      "loss": 1.8712,
      "step": 181412
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.2925944328308105,
      "learning_rate": 6.447929845242889e-05,
      "loss": 1.9243,
      "step": 181413
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.3450599908828735,
      "learning_rate": 6.447676474611233e-05,
      "loss": 1.9098,
      "step": 181414
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.4249927997589111,
      "learning_rate": 6.447423108358392e-05,
      "loss": 1.9869,
      "step": 181415
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.4388943910598755,
      "learning_rate": 6.447169746484423e-05,
      "loss": 1.9862,
      "step": 181416
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.4058425426483154,
      "learning_rate": 6.446916388989377e-05,
      "loss": 1.9969,
      "step": 181417
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.3047198057174683,
      "learning_rate": 6.446663035873281e-05,
      "loss": 2.0481,
      "step": 181418
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.3935203552246094,
      "learning_rate": 6.446409687136204e-05,
      "loss": 1.9245,
      "step": 181419
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.5122658014297485,
      "learning_rate": 6.446156342778187e-05,
      "loss": 2.0346,
      "step": 181420
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.366601586341858,
      "learning_rate": 6.445903002799261e-05,
      "loss": 2.1609,
      "step": 181421
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.5582982301712036,
      "learning_rate": 6.445649667199496e-05,
      "loss": 1.9405,
      "step": 181422
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.4015257358551025,
      "learning_rate": 6.445396335978925e-05,
      "loss": 2.075,
      "step": 181423
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.333126425743103,
      "learning_rate": 6.445143009137604e-05,
      "loss": 2.0263,
      "step": 181424
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.2207247018814087,
      "learning_rate": 6.444889686675578e-05,
      "loss": 1.9324,
      "step": 181425
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.246242642402649,
      "learning_rate": 6.444636368592887e-05,
      "loss": 1.7901,
      "step": 181426
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.4821431636810303,
      "learning_rate": 6.444383054889576e-05,
      "loss": 1.9999,
      "step": 181427
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.3966867923736572,
      "learning_rate": 6.44412974556571e-05,
      "loss": 2.0755,
      "step": 181428
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.5290935039520264,
      "learning_rate": 6.443876440621317e-05,
      "loss": 2.0122,
      "step": 181429
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.4684710502624512,
      "learning_rate": 6.443623140056458e-05,
      "loss": 2.11,
      "step": 181430
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.3812121152877808,
      "learning_rate": 6.443369843871168e-05,
      "loss": 2.1109,
      "step": 181431
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.2684400081634521,
      "learning_rate": 6.443116552065505e-05,
      "loss": 2.1521,
      "step": 181432
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.4040600061416626,
      "learning_rate": 6.442863264639512e-05,
      "loss": 1.7834,
      "step": 181433
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.2689118385314941,
      "learning_rate": 6.442609981593238e-05,
      "loss": 1.884,
      "step": 181434
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.336939811706543,
      "learning_rate": 6.442356702926718e-05,
      "loss": 1.9334,
      "step": 181435
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.3300193548202515,
      "learning_rate": 6.442103428640015e-05,
      "loss": 2.3018,
      "step": 181436
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.3795616626739502,
      "learning_rate": 6.441850158733165e-05,
      "loss": 1.9722,
      "step": 181437
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.4080920219421387,
      "learning_rate": 6.441596893206224e-05,
      "loss": 1.9061,
      "step": 181438
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.2803030014038086,
      "learning_rate": 6.441343632059238e-05,
      "loss": 1.9414,
      "step": 181439
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.3005554676055908,
      "learning_rate": 6.441090375292249e-05,
      "loss": 1.9997,
      "step": 181440
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.2910139560699463,
      "learning_rate": 6.440837122905297e-05,
      "loss": 1.9211,
      "step": 181441
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.418184757232666,
      "learning_rate": 6.44058387489845e-05,
      "loss": 1.986,
      "step": 181442
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.360669493675232,
      "learning_rate": 6.440330631271734e-05,
      "loss": 2.0237,
      "step": 181443
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.3062968254089355,
      "learning_rate": 6.440077392025211e-05,
      "loss": 1.877,
      "step": 181444
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.2767897844314575,
      "learning_rate": 6.439824157158925e-05,
      "loss": 2.0076,
      "step": 181445
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.297989845275879,
      "learning_rate": 6.439570926672918e-05,
      "loss": 2.0061,
      "step": 181446
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.3630175590515137,
      "learning_rate": 6.439317700567233e-05,
      "loss": 1.9605,
      "step": 181447
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.4782600402832031,
      "learning_rate": 6.439064478841932e-05,
      "loss": 2.059,
      "step": 181448
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.2886706590652466,
      "learning_rate": 6.438811261497043e-05,
      "loss": 1.9255,
      "step": 181449
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.3471250534057617,
      "learning_rate": 6.438558048532634e-05,
      "loss": 2.0198,
      "step": 181450
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.5023701190948486,
      "learning_rate": 6.43830483994874e-05,
      "loss": 1.8399,
      "step": 181451
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.3239500522613525,
      "learning_rate": 6.438051635745413e-05,
      "loss": 1.9071,
      "step": 181452
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.3987950086593628,
      "learning_rate": 6.437798435922687e-05,
      "loss": 2.1885,
      "step": 181453
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.2531758546829224,
      "learning_rate": 6.437545240480628e-05,
      "loss": 1.9212,
      "step": 181454
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.3998054265975952,
      "learning_rate": 6.437292049419264e-05,
      "loss": 2.1302,
      "step": 181455
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.296911358833313,
      "learning_rate": 6.43703886273866e-05,
      "loss": 2.2634,
      "step": 181456
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.4155561923980713,
      "learning_rate": 6.436785680438859e-05,
      "loss": 2.0387,
      "step": 181457
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.390516757965088,
      "learning_rate": 6.436532502519903e-05,
      "loss": 1.9697,
      "step": 181458
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.4701216220855713,
      "learning_rate": 6.436279328981832e-05,
      "loss": 1.9335,
      "step": 181459
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.3436373472213745,
      "learning_rate": 6.436026159824707e-05,
      "loss": 1.9905,
      "step": 181460
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.3282067775726318,
      "learning_rate": 6.435772995048564e-05,
      "loss": 2.0151,
      "step": 181461
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.2477688789367676,
      "learning_rate": 6.435519834653463e-05,
      "loss": 1.9098,
      "step": 181462
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.3532582521438599,
      "learning_rate": 6.435266678639445e-05,
      "loss": 2.3191,
      "step": 181463
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.2369803190231323,
      "learning_rate": 6.435013527006551e-05,
      "loss": 1.6664,
      "step": 181464
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.3623102903366089,
      "learning_rate": 6.434760379754831e-05,
      "loss": 2.0676,
      "step": 181465
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.4074591398239136,
      "learning_rate": 6.434507236884339e-05,
      "loss": 2.0014,
      "step": 181466
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.351412057876587,
      "learning_rate": 6.434254098395108e-05,
      "loss": 2.2033,
      "step": 181467
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.2696207761764526,
      "learning_rate": 6.434000964287204e-05,
      "loss": 2.0341,
      "step": 181468
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.3113585710525513,
      "learning_rate": 6.433747834560665e-05,
      "loss": 2.1845,
      "step": 181469
    },
    {
      "epoch": 2.36,
      "grad_norm": 2.030109167098999,
      "learning_rate": 6.433494709215534e-05,
      "loss": 2.2354,
      "step": 181470
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.3176438808441162,
      "learning_rate": 6.433241588251853e-05,
      "loss": 2.1201,
      "step": 181471
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.3007968664169312,
      "learning_rate": 6.432988471669686e-05,
      "loss": 2.0035,
      "step": 181472
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.3774257898330688,
      "learning_rate": 6.432735359469065e-05,
      "loss": 2.1451,
      "step": 181473
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.374180793762207,
      "learning_rate": 6.432482251650051e-05,
      "loss": 2.0516,
      "step": 181474
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.3701691627502441,
      "learning_rate": 6.432229148212682e-05,
      "loss": 1.923,
      "step": 181475
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.306092619895935,
      "learning_rate": 6.431976049157009e-05,
      "loss": 2.0949,
      "step": 181476
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.3500490188598633,
      "learning_rate": 6.431722954483067e-05,
      "loss": 2.0565,
      "step": 181477
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.4504426717758179,
      "learning_rate": 6.431469864190918e-05,
      "loss": 1.9306,
      "step": 181478
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.5728741884231567,
      "learning_rate": 6.4312167782806e-05,
      "loss": 2.0375,
      "step": 181479
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.332507848739624,
      "learning_rate": 6.43096369675217e-05,
      "loss": 1.8611,
      "step": 181480
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.3634308576583862,
      "learning_rate": 6.43071061960567e-05,
      "loss": 1.8876,
      "step": 181481
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.3955079317092896,
      "learning_rate": 6.430457546841142e-05,
      "loss": 1.8526,
      "step": 181482
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.393983244895935,
      "learning_rate": 6.430204478458633e-05,
      "loss": 1.7246,
      "step": 181483
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.365760087966919,
      "learning_rate": 6.429951414458201e-05,
      "loss": 1.8958,
      "step": 181484
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.3099892139434814,
      "learning_rate": 6.429698354839874e-05,
      "loss": 1.9907,
      "step": 181485
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.29142165184021,
      "learning_rate": 6.429445299603725e-05,
      "loss": 1.9718,
      "step": 181486
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.3775358200073242,
      "learning_rate": 6.429192248749783e-05,
      "loss": 2.018,
      "step": 181487
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.3342548608779907,
      "learning_rate": 6.428939202278101e-05,
      "loss": 1.9409,
      "step": 181488
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.2860277891159058,
      "learning_rate": 6.428686160188715e-05,
      "loss": 2.2219,
      "step": 181489
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.3876230716705322,
      "learning_rate": 6.428433122481687e-05,
      "loss": 1.9391,
      "step": 181490
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.324592113494873,
      "learning_rate": 6.428180089157054e-05,
      "loss": 2.1089,
      "step": 181491
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.2855015993118286,
      "learning_rate": 6.427927060214876e-05,
      "loss": 2.2008,
      "step": 181492
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.3725565671920776,
      "learning_rate": 6.427674035655188e-05,
      "loss": 2.2499,
      "step": 181493
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.3208473920822144,
      "learning_rate": 6.427421015478035e-05,
      "loss": 2.0003,
      "step": 181494
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.4143766164779663,
      "learning_rate": 6.427167999683476e-05,
      "loss": 1.9432,
      "step": 181495
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.3003640174865723,
      "learning_rate": 6.426914988271551e-05,
      "loss": 1.8954,
      "step": 181496
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.257521390914917,
      "learning_rate": 6.426661981242301e-05,
      "loss": 2.1093,
      "step": 181497
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.4931514263153076,
      "learning_rate": 6.426408978595784e-05,
      "loss": 1.9592,
      "step": 181498
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.3325423002243042,
      "learning_rate": 6.426155980332038e-05,
      "loss": 2.2559,
      "step": 181499
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.300653338432312,
      "learning_rate": 6.42590298645112e-05,
      "loss": 1.8838,
      "step": 181500
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.3102521896362305,
      "learning_rate": 6.425649996953076e-05,
      "loss": 2.1244,
      "step": 181501
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.359917402267456,
      "learning_rate": 6.425397011837946e-05,
      "loss": 2.1027,
      "step": 181502
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.254460096359253,
      "learning_rate": 6.42514403110577e-05,
      "loss": 2.0101,
      "step": 181503
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.518609881401062,
      "learning_rate": 6.424891054756615e-05,
      "loss": 2.0615,
      "step": 181504
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.3772668838500977,
      "learning_rate": 6.42463808279051e-05,
      "loss": 2.0903,
      "step": 181505
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.2283657789230347,
      "learning_rate": 6.424385115207516e-05,
      "loss": 1.9817,
      "step": 181506
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.2816153764724731,
      "learning_rate": 6.424132152007675e-05,
      "loss": 2.0942,
      "step": 181507
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.3549612760543823,
      "learning_rate": 6.423879193191024e-05,
      "loss": 2.0185,
      "step": 181508
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.4544488191604614,
      "learning_rate": 6.423626238757629e-05,
      "loss": 2.1665,
      "step": 181509
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.442134976387024,
      "learning_rate": 6.423373288707527e-05,
      "loss": 2.1666,
      "step": 181510
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.2565944194793701,
      "learning_rate": 6.423120343040754e-05,
      "loss": 2.1065,
      "step": 181511
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.3128079175949097,
      "learning_rate": 6.422867401757375e-05,
      "loss": 2.0735,
      "step": 181512
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.3578698635101318,
      "learning_rate": 6.422614464857435e-05,
      "loss": 2.0195,
      "step": 181513
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.337470293045044,
      "learning_rate": 6.422361532340965e-05,
      "loss": 1.8492,
      "step": 181514
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.3413255214691162,
      "learning_rate": 6.422108604208031e-05,
      "loss": 2.0094,
      "step": 181515
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.3125154972076416,
      "learning_rate": 6.421855680458664e-05,
      "loss": 1.7687,
      "step": 181516
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.4090207815170288,
      "learning_rate": 6.42160276109293e-05,
      "loss": 1.8909,
      "step": 181517
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.2760822772979736,
      "learning_rate": 6.421349846110863e-05,
      "loss": 2.0786,
      "step": 181518
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.4618510007858276,
      "learning_rate": 6.421096935512512e-05,
      "loss": 2.033,
      "step": 181519
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.573864459991455,
      "learning_rate": 6.420844029297915e-05,
      "loss": 1.98,
      "step": 181520
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.3247209787368774,
      "learning_rate": 6.420591127467138e-05,
      "loss": 2.2247,
      "step": 181521
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.3351860046386719,
      "learning_rate": 6.420338230020207e-05,
      "loss": 1.9536,
      "step": 181522
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.372637152671814,
      "learning_rate": 6.420085336957192e-05,
      "loss": 1.9263,
      "step": 181523
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.2926232814788818,
      "learning_rate": 6.419832448278129e-05,
      "loss": 1.9277,
      "step": 181524
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.3567163944244385,
      "learning_rate": 6.419579563983061e-05,
      "loss": 2.0154,
      "step": 181525
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.2668994665145874,
      "learning_rate": 6.41932668407203e-05,
      "loss": 2.0993,
      "step": 181526
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.322279691696167,
      "learning_rate": 6.419073808545101e-05,
      "loss": 1.962,
      "step": 181527
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.3513506650924683,
      "learning_rate": 6.418820937402303e-05,
      "loss": 2.1559,
      "step": 181528
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.2748483419418335,
      "learning_rate": 6.4185680706437e-05,
      "loss": 1.8251,
      "step": 181529
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.3484481573104858,
      "learning_rate": 6.418315208269331e-05,
      "loss": 2.313,
      "step": 181530
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.243383526802063,
      "learning_rate": 6.418062350279242e-05,
      "loss": 2.0085,
      "step": 181531
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.5023444890975952,
      "learning_rate": 6.417809496673468e-05,
      "loss": 2.0854,
      "step": 181532
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.3989368677139282,
      "learning_rate": 6.417556647452082e-05,
      "loss": 2.025,
      "step": 181533
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.293843150138855,
      "learning_rate": 6.417303802615107e-05,
      "loss": 1.9818,
      "step": 181534
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.4461947679519653,
      "learning_rate": 6.417050962162609e-05,
      "loss": 2.0099,
      "step": 181535
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.3421944379806519,
      "learning_rate": 6.416798126094623e-05,
      "loss": 1.8549,
      "step": 181536
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.4198819398880005,
      "learning_rate": 6.416545294411203e-05,
      "loss": 2.0047,
      "step": 181537
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.243888258934021,
      "learning_rate": 6.416292467112384e-05,
      "loss": 2.1552,
      "step": 181538
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.4342259168624878,
      "learning_rate": 6.416039644198229e-05,
      "loss": 2.0243,
      "step": 181539
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.3673169612884521,
      "learning_rate": 6.415786825668771e-05,
      "loss": 2.1185,
      "step": 181540
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.2937623262405396,
      "learning_rate": 6.41553401152407e-05,
      "loss": 2.0631,
      "step": 181541
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.4408167600631714,
      "learning_rate": 6.415281201764166e-05,
      "loss": 1.778,
      "step": 181542
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.302043080329895,
      "learning_rate": 6.415028396389106e-05,
      "loss": 2.2295,
      "step": 181543
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.3454989194869995,
      "learning_rate": 6.414775595398931e-05,
      "loss": 2.1345,
      "step": 181544
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.3068599700927734,
      "learning_rate": 6.414522798793702e-05,
      "loss": 2.3307,
      "step": 181545
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.428507685661316,
      "learning_rate": 6.414270006573449e-05,
      "loss": 1.7958,
      "step": 181546
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.2179323434829712,
      "learning_rate": 6.414017218738237e-05,
      "loss": 1.8267,
      "step": 181547
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.3206310272216797,
      "learning_rate": 6.413764435288106e-05,
      "loss": 2.2642,
      "step": 181548
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.253193736076355,
      "learning_rate": 6.413511656223099e-05,
      "loss": 2.0328,
      "step": 181549
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.2668168544769287,
      "learning_rate": 6.413258881543261e-05,
      "loss": 1.9204,
      "step": 181550
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.361510992050171,
      "learning_rate": 6.41300611124865e-05,
      "loss": 2.2928,
      "step": 181551
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.400101900100708,
      "learning_rate": 6.412753345339297e-05,
      "loss": 2.0092,
      "step": 181552
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.2893850803375244,
      "learning_rate": 6.412500583815266e-05,
      "loss": 1.9237,
      "step": 181553
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.4383327960968018,
      "learning_rate": 6.412247826676596e-05,
      "loss": 2.0001,
      "step": 181554
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.3412350416183472,
      "learning_rate": 6.411995073923337e-05,
      "loss": 2.0756,
      "step": 181555
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.3034460544586182,
      "learning_rate": 6.411742325555525e-05,
      "loss": 2.0631,
      "step": 181556
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.2899599075317383,
      "learning_rate": 6.411489581573221e-05,
      "loss": 1.7244,
      "step": 181557
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.4526653289794922,
      "learning_rate": 6.41123684197646e-05,
      "loss": 2.0472,
      "step": 181558
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.302059292793274,
      "learning_rate": 6.410984106765304e-05,
      "loss": 1.9019,
      "step": 181559
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.4091460704803467,
      "learning_rate": 6.410731375939783e-05,
      "loss": 2.2597,
      "step": 181560
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.2813560962677002,
      "learning_rate": 6.410478649499968e-05,
      "loss": 2.1036,
      "step": 181561
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.3291969299316406,
      "learning_rate": 6.410225927445875e-05,
      "loss": 1.8251,
      "step": 181562
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.338570237159729,
      "learning_rate": 6.409973209777572e-05,
      "loss": 2.113,
      "step": 181563
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.2722667455673218,
      "learning_rate": 6.409720496495092e-05,
      "loss": 2.0549,
      "step": 181564
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.4423372745513916,
      "learning_rate": 6.4094677875985e-05,
      "loss": 1.9287,
      "step": 181565
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.4471967220306396,
      "learning_rate": 6.409215083087826e-05,
      "loss": 1.9753,
      "step": 181566
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.4271677732467651,
      "learning_rate": 6.408962382963141e-05,
      "loss": 1.8476,
      "step": 181567
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.3600366115570068,
      "learning_rate": 6.408709687224457e-05,
      "loss": 2.2046,
      "step": 181568
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.4028990268707275,
      "learning_rate": 6.408456995871848e-05,
      "loss": 2.075,
      "step": 181569
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.273427963256836,
      "learning_rate": 6.408204308905343e-05,
      "loss": 2.0802,
      "step": 181570
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.3063788414001465,
      "learning_rate": 6.407951626325008e-05,
      "loss": 2.0162,
      "step": 181571
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.6844762563705444,
      "learning_rate": 6.40769894813087e-05,
      "loss": 2.0003,
      "step": 181572
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.406596064567566,
      "learning_rate": 6.407446274323006e-05,
      "loss": 2.1257,
      "step": 181573
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.3553985357284546,
      "learning_rate": 6.407193604901424e-05,
      "loss": 1.9707,
      "step": 181574
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.3765895366668701,
      "learning_rate": 6.406940939866197e-05,
      "loss": 2.1562,
      "step": 181575
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.3265419006347656,
      "learning_rate": 6.40668827921736e-05,
      "loss": 2.0543,
      "step": 181576
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.5068222284317017,
      "learning_rate": 6.406435622954971e-05,
      "loss": 1.9907,
      "step": 181577
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.3847241401672363,
      "learning_rate": 6.406182971079061e-05,
      "loss": 1.9989,
      "step": 181578
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.3722153902053833,
      "learning_rate": 6.405930323589709e-05,
      "loss": 1.8854,
      "step": 181579
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.3831379413604736,
      "learning_rate": 6.40567768048692e-05,
      "loss": 1.9778,
      "step": 181580
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.4563597440719604,
      "learning_rate": 6.40542504177077e-05,
      "loss": 2.0139,
      "step": 181581
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.3031882047653198,
      "learning_rate": 6.405172407441287e-05,
      "loss": 1.8615,
      "step": 181582
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.412776231765747,
      "learning_rate": 6.404919777498535e-05,
      "loss": 2.0025,
      "step": 181583
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.2806196212768555,
      "learning_rate": 6.40466715194255e-05,
      "loss": 2.0103,
      "step": 181584
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.316747784614563,
      "learning_rate": 6.404414530773388e-05,
      "loss": 1.9378,
      "step": 181585
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.3233891725540161,
      "learning_rate": 6.404161913991092e-05,
      "loss": 2.1832,
      "step": 181586
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.3572434186935425,
      "learning_rate": 6.403909301595707e-05,
      "loss": 1.9707,
      "step": 181587
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.4715311527252197,
      "learning_rate": 6.403656693587273e-05,
      "loss": 2.055,
      "step": 181588
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.3553820848464966,
      "learning_rate": 6.403404089965853e-05,
      "loss": 2.0171,
      "step": 181589
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.4173473119735718,
      "learning_rate": 6.403151490731477e-05,
      "loss": 2.0815,
      "step": 181590
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.327265739440918,
      "learning_rate": 6.402898895884207e-05,
      "loss": 2.0257,
      "step": 181591
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.2944982051849365,
      "learning_rate": 6.402646305424087e-05,
      "loss": 1.8937,
      "step": 181592
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.6987085342407227,
      "learning_rate": 6.402393719351148e-05,
      "loss": 2.0659,
      "step": 181593
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.4025115966796875,
      "learning_rate": 6.402141137665461e-05,
      "loss": 2.0407,
      "step": 181594
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.3556126356124878,
      "learning_rate": 6.40188856036706e-05,
      "loss": 1.9521,
      "step": 181595
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.3771072626113892,
      "learning_rate": 6.401635987455986e-05,
      "loss": 2.0797,
      "step": 181596
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.492663860321045,
      "learning_rate": 6.4013834189323e-05,
      "loss": 2.1734,
      "step": 181597
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.5655055046081543,
      "learning_rate": 6.401130854796042e-05,
      "loss": 2.1211,
      "step": 181598
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.3058874607086182,
      "learning_rate": 6.400878295047255e-05,
      "loss": 1.9304,
      "step": 181599
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.3288875818252563,
      "learning_rate": 6.400625739685994e-05,
      "loss": 1.9433,
      "step": 181600
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.5024175643920898,
      "learning_rate": 6.400373188712307e-05,
      "loss": 1.8958,
      "step": 181601
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.3572964668273926,
      "learning_rate": 6.400120642126223e-05,
      "loss": 1.9369,
      "step": 181602
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.2672922611236572,
      "learning_rate": 6.399868099927813e-05,
      "loss": 2.0392,
      "step": 181603
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.5530004501342773,
      "learning_rate": 6.399615562117113e-05,
      "loss": 1.9503,
      "step": 181604
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.5528241395950317,
      "learning_rate": 6.39936302869416e-05,
      "loss": 1.9598,
      "step": 181605
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.2695584297180176,
      "learning_rate": 6.39911049965902e-05,
      "loss": 2.444,
      "step": 181606
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.3010361194610596,
      "learning_rate": 6.398857975011721e-05,
      "loss": 2.0688,
      "step": 181607
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.4552353620529175,
      "learning_rate": 6.398605454752331e-05,
      "loss": 1.8926,
      "step": 181608
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.3961026668548584,
      "learning_rate": 6.398352938880885e-05,
      "loss": 1.9199,
      "step": 181609
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.2889457941055298,
      "learning_rate": 6.398100427397429e-05,
      "loss": 1.961,
      "step": 181610
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.5460201501846313,
      "learning_rate": 6.397847920302006e-05,
      "loss": 2.2602,
      "step": 181611
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.3611009120941162,
      "learning_rate": 6.397595417594673e-05,
      "loss": 1.9728,
      "step": 181612
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.4005179405212402,
      "learning_rate": 6.397342919275467e-05,
      "loss": 1.9564,
      "step": 181613
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.3992156982421875,
      "learning_rate": 6.39709042534445e-05,
      "loss": 2.1784,
      "step": 181614
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.4109385013580322,
      "learning_rate": 6.396837935801657e-05,
      "loss": 1.8846,
      "step": 181615
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.5151891708374023,
      "learning_rate": 6.396585450647139e-05,
      "loss": 2.0439,
      "step": 181616
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.390721082687378,
      "learning_rate": 6.396332969880932e-05,
      "loss": 2.0277,
      "step": 181617
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.263355016708374,
      "learning_rate": 6.3960804935031e-05,
      "loss": 1.9189,
      "step": 181618
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.481733798980713,
      "learning_rate": 6.395828021513677e-05,
      "loss": 2.232,
      "step": 181619
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.4111964702606201,
      "learning_rate": 6.395575553912721e-05,
      "loss": 2.1157,
      "step": 181620
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.4374312162399292,
      "learning_rate": 6.395323090700272e-05,
      "loss": 2.0925,
      "step": 181621
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.4923663139343262,
      "learning_rate": 6.39507063187638e-05,
      "loss": 2.022,
      "step": 181622
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.4744259119033813,
      "learning_rate": 6.39481817744108e-05,
      "loss": 2.327,
      "step": 181623
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.4585494995117188,
      "learning_rate": 6.394565727394439e-05,
      "loss": 1.6693,
      "step": 181624
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.436318039894104,
      "learning_rate": 6.394313281736483e-05,
      "loss": 2.0625,
      "step": 181625
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.4024723768234253,
      "learning_rate": 6.39406084046728e-05,
      "loss": 2.0741,
      "step": 181626
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.338029384613037,
      "learning_rate": 6.393808403586859e-05,
      "loss": 1.9084,
      "step": 181627
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.2202662229537964,
      "learning_rate": 6.39355597109529e-05,
      "loss": 1.9123,
      "step": 181628
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.3307826519012451,
      "learning_rate": 6.393303542992589e-05,
      "loss": 1.938,
      "step": 181629
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.3108892440795898,
      "learning_rate": 6.393051119278825e-05,
      "loss": 1.8327,
      "step": 181630
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.353366732597351,
      "learning_rate": 6.392798699954032e-05,
      "loss": 1.9251,
      "step": 181631
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.2249587774276733,
      "learning_rate": 6.39254628501827e-05,
      "loss": 1.8082,
      "step": 181632
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.3139698505401611,
      "learning_rate": 6.392293874471573e-05,
      "loss": 2.1182,
      "step": 181633
    },
    {
      "epoch": 2.36,
      "grad_norm": 1.458943486213684,
      "learning_rate": 6.392041468314008e-05,
      "loss": 2.0495,
      "step": 181634
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.4298415184020996,
      "learning_rate": 6.391789066545592e-05,
      "loss": 2.062,
      "step": 181635
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.275270700454712,
      "learning_rate": 6.391536669166399e-05,
      "loss": 2.1197,
      "step": 181636
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.2754477262496948,
      "learning_rate": 6.391284276176454e-05,
      "loss": 1.9826,
      "step": 181637
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.2690013647079468,
      "learning_rate": 6.391031887575825e-05,
      "loss": 1.9746,
      "step": 181638
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.3081251382827759,
      "learning_rate": 6.390779503364539e-05,
      "loss": 2.0937,
      "step": 181639
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.3290798664093018,
      "learning_rate": 6.39052712354267e-05,
      "loss": 1.8429,
      "step": 181640
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.2835166454315186,
      "learning_rate": 6.390274748110231e-05,
      "loss": 2.1035,
      "step": 181641
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.1955150365829468,
      "learning_rate": 6.390022377067294e-05,
      "loss": 1.9521,
      "step": 181642
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.2990351915359497,
      "learning_rate": 6.389770010413889e-05,
      "loss": 1.9744,
      "step": 181643
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.509308934211731,
      "learning_rate": 6.38951764815008e-05,
      "loss": 1.8579,
      "step": 181644
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.3222280740737915,
      "learning_rate": 6.389265290275899e-05,
      "loss": 1.9616,
      "step": 181645
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.430324673652649,
      "learning_rate": 6.389012936791413e-05,
      "loss": 2.0327,
      "step": 181646
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.38779878616333,
      "learning_rate": 6.38876058769664e-05,
      "loss": 2.1788,
      "step": 181647
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.5035440921783447,
      "learning_rate": 6.38850824299165e-05,
      "loss": 2.1687,
      "step": 181648
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.3624757528305054,
      "learning_rate": 6.388255902676475e-05,
      "loss": 1.9714,
      "step": 181649
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.306929588317871,
      "learning_rate": 6.388003566751173e-05,
      "loss": 2.2196,
      "step": 181650
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.331425666809082,
      "learning_rate": 6.387751235215784e-05,
      "loss": 1.9455,
      "step": 181651
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.267859697341919,
      "learning_rate": 6.387498908070372e-05,
      "loss": 2.015,
      "step": 181652
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.3603086471557617,
      "learning_rate": 6.387246585314953e-05,
      "loss": 2.2527,
      "step": 181653
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.3373427391052246,
      "learning_rate": 6.386994266949598e-05,
      "loss": 1.9505,
      "step": 181654
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.3291312456130981,
      "learning_rate": 6.386741952974337e-05,
      "loss": 2.2316,
      "step": 181655
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.2550032138824463,
      "learning_rate": 6.386489643389236e-05,
      "loss": 2.0907,
      "step": 181656
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.3263622522354126,
      "learning_rate": 6.386237338194326e-05,
      "loss": 2.0002,
      "step": 181657
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.378374457359314,
      "learning_rate": 6.385985037389675e-05,
      "loss": 2.054,
      "step": 181658
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.2969639301300049,
      "learning_rate": 6.385732740975302e-05,
      "loss": 2.1352,
      "step": 181659
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.2705235481262207,
      "learning_rate": 6.385480448951269e-05,
      "loss": 1.9104,
      "step": 181660
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.388895869255066,
      "learning_rate": 6.385228161317618e-05,
      "loss": 2.1286,
      "step": 181661
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.2521250247955322,
      "learning_rate": 6.384975878074405e-05,
      "loss": 1.7821,
      "step": 181662
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.2350808382034302,
      "learning_rate": 6.384723599221664e-05,
      "loss": 1.9648,
      "step": 181663
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.3061822652816772,
      "learning_rate": 6.384471324759463e-05,
      "loss": 1.9437,
      "step": 181664
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.3539973497390747,
      "learning_rate": 6.384219054687818e-05,
      "loss": 1.862,
      "step": 181665
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.439483642578125,
      "learning_rate": 6.383966789006801e-05,
      "loss": 2.0884,
      "step": 181666
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.3491642475128174,
      "learning_rate": 6.383714527716444e-05,
      "loss": 2.0859,
      "step": 181667
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.395224690437317,
      "learning_rate": 6.383462270816808e-05,
      "loss": 1.8465,
      "step": 181668
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.2903690338134766,
      "learning_rate": 6.383210018307921e-05,
      "loss": 2.0635,
      "step": 181669
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.2425376176834106,
      "learning_rate": 6.382957770189854e-05,
      "loss": 2.1016,
      "step": 181670
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.279941439628601,
      "learning_rate": 6.382705526462638e-05,
      "loss": 1.9263,
      "step": 181671
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.3403395414352417,
      "learning_rate": 6.382453287126323e-05,
      "loss": 2.1111,
      "step": 181672
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.2460871934890747,
      "learning_rate": 6.382201052180946e-05,
      "loss": 1.9347,
      "step": 181673
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.5823757648468018,
      "learning_rate": 6.381948821626573e-05,
      "loss": 2.1304,
      "step": 181674
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.2856122255325317,
      "learning_rate": 6.381696595463235e-05,
      "loss": 1.8479,
      "step": 181675
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.3570852279663086,
      "learning_rate": 6.381444373690994e-05,
      "loss": 1.9922,
      "step": 181676
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.2759199142456055,
      "learning_rate": 6.381192156309887e-05,
      "loss": 2.0338,
      "step": 181677
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.422849416732788,
      "learning_rate": 6.380939943319954e-05,
      "loss": 2.0282,
      "step": 181678
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.5697495937347412,
      "learning_rate": 6.380687734721259e-05,
      "loss": 2.042,
      "step": 181679
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.3501540422439575,
      "learning_rate": 6.380435530513837e-05,
      "loss": 2.0461,
      "step": 181680
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.2877156734466553,
      "learning_rate": 6.380183330697733e-05,
      "loss": 2.0177,
      "step": 181681
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.4166451692581177,
      "learning_rate": 6.379931135273006e-05,
      "loss": 1.9689,
      "step": 181682
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.5128253698349,
      "learning_rate": 6.379678944239696e-05,
      "loss": 1.9572,
      "step": 181683
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.3480829000473022,
      "learning_rate": 6.37942675759784e-05,
      "loss": 2.187,
      "step": 181684
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.5031615495681763,
      "learning_rate": 6.379174575347505e-05,
      "loss": 1.9415,
      "step": 181685
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.230622410774231,
      "learning_rate": 6.378922397488723e-05,
      "loss": 1.9934,
      "step": 181686
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.5795409679412842,
      "learning_rate": 6.378670224021542e-05,
      "loss": 2.0519,
      "step": 181687
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.3074076175689697,
      "learning_rate": 6.378418054946017e-05,
      "loss": 2.184,
      "step": 181688
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.7082082033157349,
      "learning_rate": 6.378165890262191e-05,
      "loss": 2.0527,
      "step": 181689
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.2795734405517578,
      "learning_rate": 6.377913729970103e-05,
      "loss": 1.9346,
      "step": 181690
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.4177855253219604,
      "learning_rate": 6.377661574069813e-05,
      "loss": 2.0594,
      "step": 181691
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.5810587406158447,
      "learning_rate": 6.377409422561353e-05,
      "loss": 2.0822,
      "step": 181692
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.4051783084869385,
      "learning_rate": 6.377157275444787e-05,
      "loss": 2.18,
      "step": 181693
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.3220486640930176,
      "learning_rate": 6.376905132720154e-05,
      "loss": 1.9322,
      "step": 181694
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.3078187704086304,
      "learning_rate": 6.376652994387502e-05,
      "loss": 1.8766,
      "step": 181695
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.4336239099502563,
      "learning_rate": 6.376400860446867e-05,
      "loss": 2.0701,
      "step": 181696
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.442482352256775,
      "learning_rate": 6.376148730898314e-05,
      "loss": 1.9331,
      "step": 181697
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.2944471836090088,
      "learning_rate": 6.375896605741872e-05,
      "loss": 2.334,
      "step": 181698
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.3862963914871216,
      "learning_rate": 6.375644484977602e-05,
      "loss": 2.0475,
      "step": 181699
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.3345341682434082,
      "learning_rate": 6.37539236860555e-05,
      "loss": 2.305,
      "step": 181700
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.3920774459838867,
      "learning_rate": 6.375140256625753e-05,
      "loss": 1.9422,
      "step": 181701
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.3559402227401733,
      "learning_rate": 6.374888149038261e-05,
      "loss": 2.0818,
      "step": 181702
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.4481850862503052,
      "learning_rate": 6.374636045843127e-05,
      "loss": 1.8035,
      "step": 181703
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.3655930757522583,
      "learning_rate": 6.374383947040388e-05,
      "loss": 2.1574,
      "step": 181704
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.3946399688720703,
      "learning_rate": 6.374131852630108e-05,
      "loss": 1.9722,
      "step": 181705
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.2434425354003906,
      "learning_rate": 6.37387976261231e-05,
      "loss": 1.9903,
      "step": 181706
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.3796045780181885,
      "learning_rate": 6.373627676987074e-05,
      "loss": 1.999,
      "step": 181707
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.2550541162490845,
      "learning_rate": 6.373375595754405e-05,
      "loss": 2.2343,
      "step": 181708
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.471763253211975,
      "learning_rate": 6.373123518914384e-05,
      "loss": 2.0567,
      "step": 181709
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.35675048828125,
      "learning_rate": 6.372871446467035e-05,
      "loss": 1.9177,
      "step": 181710
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.5328946113586426,
      "learning_rate": 6.372619378412425e-05,
      "loss": 2.2072,
      "step": 181711
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.4041575193405151,
      "learning_rate": 6.372367314750582e-05,
      "loss": 1.9976,
      "step": 181712
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.3517396450042725,
      "learning_rate": 6.37211525548158e-05,
      "loss": 1.9766,
      "step": 181713
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.3294819593429565,
      "learning_rate": 6.371863200605428e-05,
      "loss": 2.1358,
      "step": 181714
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.3105558156967163,
      "learning_rate": 6.371611150122203e-05,
      "loss": 1.9875,
      "step": 181715
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.2572214603424072,
      "learning_rate": 6.371359104031934e-05,
      "loss": 1.9712,
      "step": 181716
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.230745553970337,
      "learning_rate": 6.371107062334681e-05,
      "loss": 1.8591,
      "step": 181717
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.3795785903930664,
      "learning_rate": 6.370855025030481e-05,
      "loss": 1.8808,
      "step": 181718
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.5916959047317505,
      "learning_rate": 6.3706029921194e-05,
      "loss": 2.0232,
      "step": 181719
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.3818575143814087,
      "learning_rate": 6.370350963601451e-05,
      "loss": 1.8965,
      "step": 181720
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.3020071983337402,
      "learning_rate": 6.370098939476709e-05,
      "loss": 2.1564,
      "step": 181721
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.4505723714828491,
      "learning_rate": 6.369846919745205e-05,
      "loss": 2.1555,
      "step": 181722
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.3949475288391113,
      "learning_rate": 6.369594904407001e-05,
      "loss": 2.0154,
      "step": 181723
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.2661707401275635,
      "learning_rate": 6.369342893462125e-05,
      "loss": 1.9865,
      "step": 181724
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.2624307870864868,
      "learning_rate": 6.369090886910652e-05,
      "loss": 2.0398,
      "step": 181725
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.2292828559875488,
      "learning_rate": 6.368838884752595e-05,
      "loss": 2.1559,
      "step": 181726
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.437670111656189,
      "learning_rate": 6.368586886988026e-05,
      "loss": 2.0674,
      "step": 181727
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.3179293870925903,
      "learning_rate": 6.368334893616972e-05,
      "loss": 2.276,
      "step": 181728
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.418969988822937,
      "learning_rate": 6.368082904639497e-05,
      "loss": 1.9144,
      "step": 181729
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.4218138456344604,
      "learning_rate": 6.367830920055639e-05,
      "loss": 1.9676,
      "step": 181730
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.4644821882247925,
      "learning_rate": 6.367578939865459e-05,
      "loss": 2.0448,
      "step": 181731
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.350919246673584,
      "learning_rate": 6.367326964068977e-05,
      "loss": 1.8882,
      "step": 181732
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.433129072189331,
      "learning_rate": 6.367074992666264e-05,
      "loss": 2.0766,
      "step": 181733
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.37324857711792,
      "learning_rate": 6.36682302565735e-05,
      "loss": 2.0542,
      "step": 181734
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.3843051195144653,
      "learning_rate": 6.366571063042298e-05,
      "loss": 2.0679,
      "step": 181735
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.5132043361663818,
      "learning_rate": 6.366319104821138e-05,
      "loss": 1.8753,
      "step": 181736
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.3249918222427368,
      "learning_rate": 6.36606715099394e-05,
      "loss": 1.935,
      "step": 181737
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.2915397882461548,
      "learning_rate": 6.365815201560723e-05,
      "loss": 1.94,
      "step": 181738
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.1904650926589966,
      "learning_rate": 6.365563256521553e-05,
      "loss": 1.978,
      "step": 181739
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.3657433986663818,
      "learning_rate": 6.365311315876462e-05,
      "loss": 2.0384,
      "step": 181740
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.46929931640625,
      "learning_rate": 6.365059379625518e-05,
      "loss": 2.1826,
      "step": 181741
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.389641284942627,
      "learning_rate": 6.364807447768746e-05,
      "loss": 2.157,
      "step": 181742
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.3291243314743042,
      "learning_rate": 6.364555520306217e-05,
      "loss": 2.0726,
      "step": 181743
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.4966590404510498,
      "learning_rate": 6.364303597237948e-05,
      "loss": 1.9848,
      "step": 181744
    },
    {
      "epoch": 2.37,
      "grad_norm": 2.072343587875366,
      "learning_rate": 6.364051678564006e-05,
      "loss": 1.924,
      "step": 181745
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.3592758178710938,
      "learning_rate": 6.36379976428443e-05,
      "loss": 2.1595,
      "step": 181746
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.363615870475769,
      "learning_rate": 6.363547854399275e-05,
      "loss": 2.1425,
      "step": 181747
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.4010772705078125,
      "learning_rate": 6.363295948908575e-05,
      "loss": 1.9028,
      "step": 181748
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.3127824068069458,
      "learning_rate": 6.363044047812399e-05,
      "loss": 1.9054,
      "step": 181749
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.3784490823745728,
      "learning_rate": 6.362792151110765e-05,
      "loss": 2.0354,
      "step": 181750
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.37153959274292,
      "learning_rate": 6.362540258803743e-05,
      "loss": 2.1385,
      "step": 181751
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.3117563724517822,
      "learning_rate": 6.36228837089136e-05,
      "loss": 2.0518,
      "step": 181752
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.3126749992370605,
      "learning_rate": 6.362036487373687e-05,
      "loss": 1.9824,
      "step": 181753
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.3638560771942139,
      "learning_rate": 6.361784608250745e-05,
      "loss": 1.9876,
      "step": 181754
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.406522274017334,
      "learning_rate": 6.361532733522604e-05,
      "loss": 1.9954,
      "step": 181755
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.4083904027938843,
      "learning_rate": 6.361280863189297e-05,
      "loss": 2.1387,
      "step": 181756
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.3865232467651367,
      "learning_rate": 6.361028997250878e-05,
      "loss": 2.0294,
      "step": 181757
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.355390191078186,
      "learning_rate": 6.360777135707378e-05,
      "loss": 2.0037,
      "step": 181758
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.3375260829925537,
      "learning_rate": 6.360525278558868e-05,
      "loss": 2.0772,
      "step": 181759
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.3679981231689453,
      "learning_rate": 6.360273425805372e-05,
      "loss": 1.9821,
      "step": 181760
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.461122751235962,
      "learning_rate": 6.360021577446954e-05,
      "loss": 1.9061,
      "step": 181761
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.3434090614318848,
      "learning_rate": 6.359769733483657e-05,
      "loss": 1.787,
      "step": 181762
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.325628399848938,
      "learning_rate": 6.359517893915524e-05,
      "loss": 1.9513,
      "step": 181763
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.4331347942352295,
      "learning_rate": 6.359266058742596e-05,
      "loss": 1.9345,
      "step": 181764
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.4605728387832642,
      "learning_rate": 6.359014227964934e-05,
      "loss": 2.0865,
      "step": 181765
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.2570927143096924,
      "learning_rate": 6.358762401582568e-05,
      "loss": 2.1285,
      "step": 181766
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.4467473030090332,
      "learning_rate": 6.358510579595565e-05,
      "loss": 1.9443,
      "step": 181767
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.3673973083496094,
      "learning_rate": 6.35825876200396e-05,
      "loss": 1.9473,
      "step": 181768
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.3636192083358765,
      "learning_rate": 6.358006948807791e-05,
      "loss": 1.9095,
      "step": 181769
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.3837820291519165,
      "learning_rate": 6.357755140007126e-05,
      "loss": 1.8987,
      "step": 181770
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.2943443059921265,
      "learning_rate": 6.357503335602e-05,
      "loss": 1.7351,
      "step": 181771
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.3701964616775513,
      "learning_rate": 6.35725153559245e-05,
      "loss": 1.8907,
      "step": 181772
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.4632201194763184,
      "learning_rate": 6.356999739978544e-05,
      "loss": 2.0408,
      "step": 181773
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.3379437923431396,
      "learning_rate": 6.356747948760316e-05,
      "loss": 1.9818,
      "step": 181774
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.2923974990844727,
      "learning_rate": 6.356496161937807e-05,
      "loss": 2.0265,
      "step": 181775
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.1901620626449585,
      "learning_rate": 6.35624437951108e-05,
      "loss": 2.2048,
      "step": 181776
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.3060462474822998,
      "learning_rate": 6.355992601480166e-05,
      "loss": 1.9402,
      "step": 181777
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.2109190225601196,
      "learning_rate": 6.355740827845128e-05,
      "loss": 1.7314,
      "step": 181778
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.3951518535614014,
      "learning_rate": 6.355489058606001e-05,
      "loss": 1.9611,
      "step": 181779
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.3105272054672241,
      "learning_rate": 6.355237293762839e-05,
      "loss": 2.0448,
      "step": 181780
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.3323194980621338,
      "learning_rate": 6.354985533315676e-05,
      "loss": 2.1191,
      "step": 181781
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.3110454082489014,
      "learning_rate": 6.354733777264573e-05,
      "loss": 2.1993,
      "step": 181782
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.4206444025039673,
      "learning_rate": 6.354482025609563e-05,
      "loss": 1.9962,
      "step": 181783
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.3452903032302856,
      "learning_rate": 6.354230278350714e-05,
      "loss": 2.0208,
      "step": 181784
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.3696393966674805,
      "learning_rate": 6.353978535488055e-05,
      "loss": 2.0998,
      "step": 181785
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.3826779127120972,
      "learning_rate": 6.353726797021638e-05,
      "loss": 2.0879,
      "step": 181786
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.3478888273239136,
      "learning_rate": 6.353475062951503e-05,
      "loss": 1.8711,
      "step": 181787
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.236660361289978,
      "learning_rate": 6.353223333277713e-05,
      "loss": 2.0518,
      "step": 181788
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.355656385421753,
      "learning_rate": 6.352971608000295e-05,
      "loss": 2.1398,
      "step": 181789
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.3629485368728638,
      "learning_rate": 6.352719887119316e-05,
      "loss": 1.8471,
      "step": 181790
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.44931161403656,
      "learning_rate": 6.352468170634803e-05,
      "loss": 1.9569,
      "step": 181791
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.3605141639709473,
      "learning_rate": 6.35221645854683e-05,
      "loss": 2.0091,
      "step": 181792
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.3142502307891846,
      "learning_rate": 6.351964750855406e-05,
      "loss": 1.8961,
      "step": 181793
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.4675840139389038,
      "learning_rate": 6.351713047560609e-05,
      "loss": 2.0681,
      "step": 181794
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.5105226039886475,
      "learning_rate": 6.351461348662469e-05,
      "loss": 1.9541,
      "step": 181795
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.6320345401763916,
      "learning_rate": 6.351209654161045e-05,
      "loss": 2.0451,
      "step": 181796
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.2519869804382324,
      "learning_rate": 6.350957964056367e-05,
      "loss": 2.2564,
      "step": 181797
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.3098037242889404,
      "learning_rate": 6.350706278348512e-05,
      "loss": 1.9197,
      "step": 181798
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.191579818725586,
      "learning_rate": 6.35045459703749e-05,
      "loss": 2.1007,
      "step": 181799
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.350382924079895,
      "learning_rate": 6.350202920123375e-05,
      "loss": 2.0232,
      "step": 181800
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.2856202125549316,
      "learning_rate": 6.349951247606191e-05,
      "loss": 2.0155,
      "step": 181801
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.5865126848220825,
      "learning_rate": 6.349699579486007e-05,
      "loss": 2.228,
      "step": 181802
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.4961068630218506,
      "learning_rate": 6.349447915762856e-05,
      "loss": 1.8913,
      "step": 181803
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.226433515548706,
      "learning_rate": 6.349196256436802e-05,
      "loss": 1.7783,
      "step": 181804
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.3293468952178955,
      "learning_rate": 6.348944601507863e-05,
      "loss": 1.9362,
      "step": 181805
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.198467493057251,
      "learning_rate": 6.34869295097611e-05,
      "loss": 1.851,
      "step": 181806
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.4721113443374634,
      "learning_rate": 6.348441304841573e-05,
      "loss": 2.3093,
      "step": 181807
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.5455290079116821,
      "learning_rate": 6.348189663104315e-05,
      "loss": 2.0171,
      "step": 181808
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.6112879514694214,
      "learning_rate": 6.347938025764366e-05,
      "loss": 2.0869,
      "step": 181809
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.3359919786453247,
      "learning_rate": 6.3476863928218e-05,
      "loss": 2.1094,
      "step": 181810
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.3185516595840454,
      "learning_rate": 6.34743476427663e-05,
      "loss": 2.0898,
      "step": 181811
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.2812917232513428,
      "learning_rate": 6.347183140128925e-05,
      "loss": 2.0809,
      "step": 181812
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.243878960609436,
      "learning_rate": 6.346931520378717e-05,
      "loss": 1.9367,
      "step": 181813
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.2561633586883545,
      "learning_rate": 6.34667990502607e-05,
      "loss": 1.7669,
      "step": 181814
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.2673630714416504,
      "learning_rate": 6.346428294071011e-05,
      "loss": 1.9934,
      "step": 181815
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.503491997718811,
      "learning_rate": 6.346176687513617e-05,
      "loss": 2.0153,
      "step": 181816
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.4626058340072632,
      "learning_rate": 6.345925085353897e-05,
      "loss": 2.1071,
      "step": 181817
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.4859164953231812,
      "learning_rate": 6.345673487591926e-05,
      "loss": 1.9821,
      "step": 181818
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.3864006996154785,
      "learning_rate": 6.34542189422773e-05,
      "loss": 2.01,
      "step": 181819
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.2200168371200562,
      "learning_rate": 6.345170305261377e-05,
      "loss": 2.0641,
      "step": 181820
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.3304988145828247,
      "learning_rate": 6.344918720692895e-05,
      "loss": 1.9044,
      "step": 181821
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.3864307403564453,
      "learning_rate": 6.344667140522356e-05,
      "loss": 2.0205,
      "step": 181822
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.2607351541519165,
      "learning_rate": 6.344415564749772e-05,
      "loss": 1.8775,
      "step": 181823
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.4120713472366333,
      "learning_rate": 6.344163993375217e-05,
      "loss": 2.0236,
      "step": 181824
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.2874923944473267,
      "learning_rate": 6.343912426398716e-05,
      "loss": 2.1996,
      "step": 181825
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.225443959236145,
      "learning_rate": 6.343660863820342e-05,
      "loss": 2.0296,
      "step": 181826
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.4716757535934448,
      "learning_rate": 6.343409305640118e-05,
      "loss": 1.9753,
      "step": 181827
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.4153484106063843,
      "learning_rate": 6.343157751858106e-05,
      "loss": 1.9228,
      "step": 181828
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.3461679220199585,
      "learning_rate": 6.342906202474354e-05,
      "loss": 2.0364,
      "step": 181829
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.3529552221298218,
      "learning_rate": 6.342654657488896e-05,
      "loss": 1.9014,
      "step": 181830
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.3421790599822998,
      "learning_rate": 6.34240311690178e-05,
      "loss": 1.9216,
      "step": 181831
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.3006058931350708,
      "learning_rate": 6.342151580713065e-05,
      "loss": 1.783,
      "step": 181832
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.3758877515792847,
      "learning_rate": 6.341900048922782e-05,
      "loss": 2.0527,
      "step": 181833
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.2369272708892822,
      "learning_rate": 6.341648521530998e-05,
      "loss": 2.1815,
      "step": 181834
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.2750062942504883,
      "learning_rate": 6.341396998537744e-05,
      "loss": 1.9786,
      "step": 181835
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.4296363592147827,
      "learning_rate": 6.341145479943073e-05,
      "loss": 2.1022,
      "step": 181836
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.346982479095459,
      "learning_rate": 6.340893965747023e-05,
      "loss": 1.8815,
      "step": 181837
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.3983311653137207,
      "learning_rate": 6.340642455949651e-05,
      "loss": 1.827,
      "step": 181838
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.3704769611358643,
      "learning_rate": 6.340390950550997e-05,
      "loss": 2.0085,
      "step": 181839
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.6650390625,
      "learning_rate": 6.340139449551114e-05,
      "loss": 2.0623,
      "step": 181840
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.2658329010009766,
      "learning_rate": 6.339887952950049e-05,
      "loss": 1.9485,
      "step": 181841
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.3488798141479492,
      "learning_rate": 6.339636460747846e-05,
      "loss": 2.267,
      "step": 181842
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.263003945350647,
      "learning_rate": 6.33938497294454e-05,
      "loss": 1.8966,
      "step": 181843
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.635358214378357,
      "learning_rate": 6.3391334895402e-05,
      "loss": 1.8027,
      "step": 181844
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.3220365047454834,
      "learning_rate": 6.338882010534852e-05,
      "loss": 2.0175,
      "step": 181845
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.6383596658706665,
      "learning_rate": 6.33863053592856e-05,
      "loss": 1.9879,
      "step": 181846
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.2689610719680786,
      "learning_rate": 6.338379065721366e-05,
      "loss": 2.0141,
      "step": 181847
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.2923940420150757,
      "learning_rate": 6.338127599913313e-05,
      "loss": 1.9625,
      "step": 181848
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.2932031154632568,
      "learning_rate": 6.337876138504437e-05,
      "loss": 1.8763,
      "step": 181849
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.3193228244781494,
      "learning_rate": 6.337624681494809e-05,
      "loss": 2.3047,
      "step": 181850
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.5436077117919922,
      "learning_rate": 6.337373228884453e-05,
      "loss": 2.2089,
      "step": 181851
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.4974572658538818,
      "learning_rate": 6.337121780673436e-05,
      "loss": 2.1806,
      "step": 181852
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.3156368732452393,
      "learning_rate": 6.336870336861791e-05,
      "loss": 2.1617,
      "step": 181853
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.2561856508255005,
      "learning_rate": 6.336618897449564e-05,
      "loss": 1.969,
      "step": 181854
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.299203872680664,
      "learning_rate": 6.336367462436814e-05,
      "loss": 2.1503,
      "step": 181855
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.2744032144546509,
      "learning_rate": 6.336116031823577e-05,
      "loss": 2.057,
      "step": 181856
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.4439823627471924,
      "learning_rate": 6.335864605609898e-05,
      "loss": 2.2255,
      "step": 181857
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.287523865699768,
      "learning_rate": 6.335613183795836e-05,
      "loss": 2.0935,
      "step": 181858
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.3548637628555298,
      "learning_rate": 6.335361766381429e-05,
      "loss": 1.8179,
      "step": 181859
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.5267528295516968,
      "learning_rate": 6.335110353366716e-05,
      "loss": 2.1205,
      "step": 181860
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.3252925872802734,
      "learning_rate": 6.334858944751763e-05,
      "loss": 1.8864,
      "step": 181861
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.365312099456787,
      "learning_rate": 6.334607540536607e-05,
      "loss": 2.1447,
      "step": 181862
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.2385138273239136,
      "learning_rate": 6.334356140721285e-05,
      "loss": 1.9278,
      "step": 181863
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.305911898612976,
      "learning_rate": 6.33410474530586e-05,
      "loss": 2.1528,
      "step": 181864
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.404685139656067,
      "learning_rate": 6.333853354290373e-05,
      "loss": 1.9944,
      "step": 181865
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.402239441871643,
      "learning_rate": 6.333601967674858e-05,
      "loss": 1.8902,
      "step": 181866
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.7190593481063843,
      "learning_rate": 6.333350585459385e-05,
      "loss": 2.1656,
      "step": 181867
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.3845558166503906,
      "learning_rate": 6.333099207643982e-05,
      "loss": 2.0382,
      "step": 181868
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.372294306755066,
      "learning_rate": 6.332847834228705e-05,
      "loss": 1.8973,
      "step": 181869
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.3008651733398438,
      "learning_rate": 6.332596465213602e-05,
      "loss": 1.9782,
      "step": 181870
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.3135132789611816,
      "learning_rate": 6.332345100598716e-05,
      "loss": 2.1015,
      "step": 181871
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.3780629634857178,
      "learning_rate": 6.332093740384087e-05,
      "loss": 1.8754,
      "step": 181872
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.4397616386413574,
      "learning_rate": 6.331842384569774e-05,
      "loss": 2.1397,
      "step": 181873
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.3481566905975342,
      "learning_rate": 6.331591033155811e-05,
      "loss": 1.9887,
      "step": 181874
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.371619701385498,
      "learning_rate": 6.33133968614226e-05,
      "loss": 2.1577,
      "step": 181875
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.3269602060317993,
      "learning_rate": 6.331088343529154e-05,
      "loss": 1.9389,
      "step": 181876
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.3475252389907837,
      "learning_rate": 6.330837005316558e-05,
      "loss": 2.0944,
      "step": 181877
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.3168359994888306,
      "learning_rate": 6.330585671504493e-05,
      "loss": 1.8223,
      "step": 181878
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.4735486507415771,
      "learning_rate": 6.330334342093027e-05,
      "loss": 2.0072,
      "step": 181879
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.3885833024978638,
      "learning_rate": 6.330083017082186e-05,
      "loss": 2.2045,
      "step": 181880
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.284970998764038,
      "learning_rate": 6.329831696472041e-05,
      "loss": 2.0825,
      "step": 181881
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.5531593561172485,
      "learning_rate": 6.329580380262618e-05,
      "loss": 2.1249,
      "step": 181882
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.2600663900375366,
      "learning_rate": 6.329329068453991e-05,
      "loss": 1.9421,
      "step": 181883
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.3653143644332886,
      "learning_rate": 6.329077761046171e-05,
      "loss": 2.0555,
      "step": 181884
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.4536809921264648,
      "learning_rate": 6.32882645803923e-05,
      "loss": 1.9365,
      "step": 181885
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.3789063692092896,
      "learning_rate": 6.328575159433199e-05,
      "loss": 1.9875,
      "step": 181886
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.3168747425079346,
      "learning_rate": 6.32832386522814e-05,
      "loss": 1.8101,
      "step": 181887
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.2998555898666382,
      "learning_rate": 6.328072575424086e-05,
      "loss": 1.6828,
      "step": 181888
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.3550093173980713,
      "learning_rate": 6.327821290021106e-05,
      "loss": 1.9299,
      "step": 181889
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.3109140396118164,
      "learning_rate": 6.327570009019218e-05,
      "loss": 2.1238,
      "step": 181890
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.3878861665725708,
      "learning_rate": 6.327318732418484e-05,
      "loss": 2.1285,
      "step": 181891
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.3486381769180298,
      "learning_rate": 6.327067460218943e-05,
      "loss": 2.0749,
      "step": 181892
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.5221391916275024,
      "learning_rate": 6.326816192420655e-05,
      "loss": 2.261,
      "step": 181893
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.3867628574371338,
      "learning_rate": 6.32656492902365e-05,
      "loss": 2.2312,
      "step": 181894
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.3151711225509644,
      "learning_rate": 6.32631367002799e-05,
      "loss": 2.0691,
      "step": 181895
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.301780104637146,
      "learning_rate": 6.326062415433717e-05,
      "loss": 1.9873,
      "step": 181896
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.2263907194137573,
      "learning_rate": 6.325811165240876e-05,
      "loss": 1.8509,
      "step": 181897
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.4995392560958862,
      "learning_rate": 6.325559919449505e-05,
      "loss": 2.164,
      "step": 181898
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.5546278953552246,
      "learning_rate": 6.325308678059668e-05,
      "loss": 2.073,
      "step": 181899
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.508263111114502,
      "learning_rate": 6.325057441071395e-05,
      "loss": 1.8945,
      "step": 181900
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.279664397239685,
      "learning_rate": 6.324806208484748e-05,
      "loss": 1.9577,
      "step": 181901
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.3471629619598389,
      "learning_rate": 6.324554980299765e-05,
      "loss": 2.1092,
      "step": 181902
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.321165680885315,
      "learning_rate": 6.324303756516496e-05,
      "loss": 1.9519,
      "step": 181903
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.3484556674957275,
      "learning_rate": 6.324052537134979e-05,
      "loss": 2.2211,
      "step": 181904
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.5539828538894653,
      "learning_rate": 6.323801322155272e-05,
      "loss": 2.2566,
      "step": 181905
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.4857733249664307,
      "learning_rate": 6.323550111577413e-05,
      "loss": 2.0623,
      "step": 181906
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.409970998764038,
      "learning_rate": 6.32329890540146e-05,
      "loss": 1.9444,
      "step": 181907
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.384743094444275,
      "learning_rate": 6.323047703627451e-05,
      "loss": 1.9456,
      "step": 181908
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.484608769416809,
      "learning_rate": 6.322796506255438e-05,
      "loss": 1.8592,
      "step": 181909
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.5701626539230347,
      "learning_rate": 6.322545313285453e-05,
      "loss": 2.0197,
      "step": 181910
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.6314114332199097,
      "learning_rate": 6.32229412471756e-05,
      "loss": 1.9315,
      "step": 181911
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.6700788736343384,
      "learning_rate": 6.322042940551797e-05,
      "loss": 2.1437,
      "step": 181912
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.2968131303787231,
      "learning_rate": 6.321791760788215e-05,
      "loss": 2.0438,
      "step": 181913
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.3802756071090698,
      "learning_rate": 6.321540585426862e-05,
      "loss": 1.9617,
      "step": 181914
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.713454008102417,
      "learning_rate": 6.321289414467784e-05,
      "loss": 1.8297,
      "step": 181915
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.269657015800476,
      "learning_rate": 6.321038247911015e-05,
      "loss": 1.9184,
      "step": 181916
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.3618086576461792,
      "learning_rate": 6.320787085756616e-05,
      "loss": 1.9317,
      "step": 181917
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.4078482389450073,
      "learning_rate": 6.320535928004628e-05,
      "loss": 1.9628,
      "step": 181918
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.3122224807739258,
      "learning_rate": 6.320284774655103e-05,
      "loss": 2.2157,
      "step": 181919
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.3658490180969238,
      "learning_rate": 6.320033625708087e-05,
      "loss": 1.9609,
      "step": 181920
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.2986057996749878,
      "learning_rate": 6.319782481163624e-05,
      "loss": 1.7953,
      "step": 181921
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.2835618257522583,
      "learning_rate": 6.31953134102175e-05,
      "loss": 1.8502,
      "step": 181922
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.5627566576004028,
      "learning_rate": 6.319280205282529e-05,
      "loss": 2.0325,
      "step": 181923
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.3687939643859863,
      "learning_rate": 6.319029073945997e-05,
      "loss": 2.1315,
      "step": 181924
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.2509689331054688,
      "learning_rate": 6.318777947012209e-05,
      "loss": 1.6745,
      "step": 181925
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.406713843345642,
      "learning_rate": 6.318526824481209e-05,
      "loss": 2.1994,
      "step": 181926
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.5403306484222412,
      "learning_rate": 6.318275706353042e-05,
      "loss": 1.9058,
      "step": 181927
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.3998736143112183,
      "learning_rate": 6.318024592627747e-05,
      "loss": 2.1632,
      "step": 181928
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.2298004627227783,
      "learning_rate": 6.317773483305384e-05,
      "loss": 1.9956,
      "step": 181929
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.3753546476364136,
      "learning_rate": 6.317522378385987e-05,
      "loss": 1.9234,
      "step": 181930
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.2838631868362427,
      "learning_rate": 6.317271277869617e-05,
      "loss": 2.0531,
      "step": 181931
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.4707494974136353,
      "learning_rate": 6.317020181756316e-05,
      "loss": 2.1634,
      "step": 181932
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.3108515739440918,
      "learning_rate": 6.316769090046126e-05,
      "loss": 1.8957,
      "step": 181933
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.3779102563858032,
      "learning_rate": 6.316518002739086e-05,
      "loss": 2.1549,
      "step": 181934
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.3382066488265991,
      "learning_rate": 6.316266919835265e-05,
      "loss": 2.1562,
      "step": 181935
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.4260839223861694,
      "learning_rate": 6.316015841334684e-05,
      "loss": 1.8489,
      "step": 181936
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.4019849300384521,
      "learning_rate": 6.315764767237414e-05,
      "loss": 2.0024,
      "step": 181937
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.3729244470596313,
      "learning_rate": 6.315513697543489e-05,
      "loss": 2.228,
      "step": 181938
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.2328399419784546,
      "learning_rate": 6.315262632252949e-05,
      "loss": 2.0415,
      "step": 181939
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.3414520025253296,
      "learning_rate": 6.315011571365859e-05,
      "loss": 1.9557,
      "step": 181940
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.3945918083190918,
      "learning_rate": 6.314760514882253e-05,
      "loss": 2.1685,
      "step": 181941
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.3769680261611938,
      "learning_rate": 6.314509462802172e-05,
      "loss": 2.1829,
      "step": 181942
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.5113552808761597,
      "learning_rate": 6.314258415125681e-05,
      "loss": 1.8641,
      "step": 181943
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.270479440689087,
      "learning_rate": 6.314007371852814e-05,
      "loss": 1.8684,
      "step": 181944
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.3553402423858643,
      "learning_rate": 6.313756332983612e-05,
      "loss": 1.9878,
      "step": 181945
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.4152966737747192,
      "learning_rate": 6.313505298518141e-05,
      "loss": 2.2478,
      "step": 181946
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.3023552894592285,
      "learning_rate": 6.313254268456434e-05,
      "loss": 2.071,
      "step": 181947
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.263737678527832,
      "learning_rate": 6.313003242798533e-05,
      "loss": 1.8585,
      "step": 181948
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.262041687965393,
      "learning_rate": 6.312752221544501e-05,
      "loss": 2.0187,
      "step": 181949
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.4420864582061768,
      "learning_rate": 6.312501204694371e-05,
      "loss": 1.8363,
      "step": 181950
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.2415236234664917,
      "learning_rate": 6.31225019224819e-05,
      "loss": 1.9495,
      "step": 181951
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.2488824129104614,
      "learning_rate": 6.311999184206015e-05,
      "loss": 2.3489,
      "step": 181952
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.5114625692367554,
      "learning_rate": 6.311748180567881e-05,
      "loss": 2.1604,
      "step": 181953
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.4186713695526123,
      "learning_rate": 6.311497181333847e-05,
      "loss": 2.1218,
      "step": 181954
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.5329822301864624,
      "learning_rate": 6.31124618650395e-05,
      "loss": 1.8851,
      "step": 181955
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.512878179550171,
      "learning_rate": 6.310995196078243e-05,
      "loss": 1.9718,
      "step": 181956
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.2457207441329956,
      "learning_rate": 6.310744210056761e-05,
      "loss": 2.1698,
      "step": 181957
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.3557263612747192,
      "learning_rate": 6.310493228439565e-05,
      "loss": 2.0127,
      "step": 181958
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.383775234222412,
      "learning_rate": 6.310242251226689e-05,
      "loss": 2.1602,
      "step": 181959
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.2448283433914185,
      "learning_rate": 6.309991278418194e-05,
      "loss": 2.1643,
      "step": 181960
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.3636139631271362,
      "learning_rate": 6.309740310014109e-05,
      "loss": 2.02,
      "step": 181961
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.4796487092971802,
      "learning_rate": 6.3094893460145e-05,
      "loss": 1.9783,
      "step": 181962
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.3285677433013916,
      "learning_rate": 6.309238386419405e-05,
      "loss": 2.013,
      "step": 181963
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.4765480756759644,
      "learning_rate": 6.308987431228867e-05,
      "loss": 1.9941,
      "step": 181964
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.3754088878631592,
      "learning_rate": 6.308736480442929e-05,
      "loss": 2.0358,
      "step": 181965
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.3592076301574707,
      "learning_rate": 6.308485534061654e-05,
      "loss": 2.1007,
      "step": 181966
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.4638445377349854,
      "learning_rate": 6.308234592085068e-05,
      "loss": 2.0835,
      "step": 181967
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.4306235313415527,
      "learning_rate": 6.307983654513237e-05,
      "loss": 2.193,
      "step": 181968
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.356655478477478,
      "learning_rate": 6.307732721346202e-05,
      "loss": 1.9846,
      "step": 181969
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.2645572423934937,
      "learning_rate": 6.307481792584003e-05,
      "loss": 1.9573,
      "step": 181970
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.3310173749923706,
      "learning_rate": 6.307230868226684e-05,
      "loss": 1.9965,
      "step": 181971
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.3129019737243652,
      "learning_rate": 6.306979948274304e-05,
      "loss": 1.795,
      "step": 181972
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.4674087762832642,
      "learning_rate": 6.306729032726898e-05,
      "loss": 1.9572,
      "step": 181973
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.3150954246520996,
      "learning_rate": 6.306478121584525e-05,
      "loss": 1.8426,
      "step": 181974
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.5006413459777832,
      "learning_rate": 6.306227214847224e-05,
      "loss": 2.0707,
      "step": 181975
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.3633747100830078,
      "learning_rate": 6.305976312515048e-05,
      "loss": 1.873,
      "step": 181976
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.3001625537872314,
      "learning_rate": 6.305725414588023e-05,
      "loss": 2.0422,
      "step": 181977
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.3786519765853882,
      "learning_rate": 6.305474521066224e-05,
      "loss": 1.909,
      "step": 181978
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.4122415781021118,
      "learning_rate": 6.305223631949673e-05,
      "loss": 2.0743,
      "step": 181979
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.4214179515838623,
      "learning_rate": 6.30497274723844e-05,
      "loss": 2.0704,
      "step": 181980
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.4544535875320435,
      "learning_rate": 6.304721866932557e-05,
      "loss": 2.1819,
      "step": 181981
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.5374274253845215,
      "learning_rate": 6.304470991032073e-05,
      "loss": 2.1887,
      "step": 181982
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.3335713148117065,
      "learning_rate": 6.304220119537028e-05,
      "loss": 1.9686,
      "step": 181983
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.2712043523788452,
      "learning_rate": 6.303969252447488e-05,
      "loss": 1.9489,
      "step": 181984
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.498364806175232,
      "learning_rate": 6.303718389763475e-05,
      "loss": 2.1367,
      "step": 181985
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.3761004209518433,
      "learning_rate": 6.303467531485054e-05,
      "loss": 1.9976,
      "step": 181986
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.2990931272506714,
      "learning_rate": 6.30321667761227e-05,
      "loss": 1.9346,
      "step": 181987
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.3123334646224976,
      "learning_rate": 6.302965828145166e-05,
      "loss": 1.9666,
      "step": 181988
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.4318327903747559,
      "learning_rate": 6.302714983083776e-05,
      "loss": 2.1221,
      "step": 181989
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.354860782623291,
      "learning_rate": 6.30246414242817e-05,
      "loss": 2.0856,
      "step": 181990
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.3872764110565186,
      "learning_rate": 6.302213306178373e-05,
      "loss": 1.9528,
      "step": 181991
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.3541896343231201,
      "learning_rate": 6.30196247433445e-05,
      "loss": 2.1135,
      "step": 181992
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.3230518102645874,
      "learning_rate": 6.30171164689644e-05,
      "loss": 2.0973,
      "step": 181993
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.3935651779174805,
      "learning_rate": 6.301460823864389e-05,
      "loss": 2.1638,
      "step": 181994
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.3417143821716309,
      "learning_rate": 6.301210005238334e-05,
      "loss": 1.9669,
      "step": 181995
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.328431487083435,
      "learning_rate": 6.300959191018344e-05,
      "loss": 2.211,
      "step": 181996
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.4928025007247925,
      "learning_rate": 6.30070838120444e-05,
      "loss": 1.952,
      "step": 181997
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.433108925819397,
      "learning_rate": 6.300457575796695e-05,
      "loss": 2.2262,
      "step": 181998
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.271811604499817,
      "learning_rate": 6.300206774795138e-05,
      "loss": 1.9005,
      "step": 181999
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.6276367902755737,
      "learning_rate": 6.29995597819982e-05,
      "loss": 1.899,
      "step": 182000
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.2768281698226929,
      "learning_rate": 6.299705186010781e-05,
      "loss": 1.7945,
      "step": 182001
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.355849027633667,
      "learning_rate": 6.299454398228084e-05,
      "loss": 2.0741,
      "step": 182002
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.3593509197235107,
      "learning_rate": 6.299203614851754e-05,
      "loss": 1.9447,
      "step": 182003
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.3366554975509644,
      "learning_rate": 6.298952835881861e-05,
      "loss": 1.7299,
      "step": 182004
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.5560564994812012,
      "learning_rate": 6.298702061318436e-05,
      "loss": 2.1456,
      "step": 182005
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.3807100057601929,
      "learning_rate": 6.298451291161533e-05,
      "loss": 1.9959,
      "step": 182006
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.3119359016418457,
      "learning_rate": 6.298200525411184e-05,
      "loss": 1.9926,
      "step": 182007
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.4889857769012451,
      "learning_rate": 6.297949764067458e-05,
      "loss": 2.0223,
      "step": 182008
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.268286943435669,
      "learning_rate": 6.297699007130383e-05,
      "loss": 1.9479,
      "step": 182009
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.2366517782211304,
      "learning_rate": 6.29744825460002e-05,
      "loss": 1.8092,
      "step": 182010
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.2301503419876099,
      "learning_rate": 6.29719750647641e-05,
      "loss": 2.2581,
      "step": 182011
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.5443013906478882,
      "learning_rate": 6.296946762759596e-05,
      "loss": 2.0117,
      "step": 182012
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.2929012775421143,
      "learning_rate": 6.296696023449621e-05,
      "loss": 1.9292,
      "step": 182013
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.2406377792358398,
      "learning_rate": 6.296445288546545e-05,
      "loss": 1.8115,
      "step": 182014
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.243664264678955,
      "learning_rate": 6.296194558050399e-05,
      "loss": 2.1484,
      "step": 182015
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.3539354801177979,
      "learning_rate": 6.295943831961245e-05,
      "loss": 2.0936,
      "step": 182016
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.474765419960022,
      "learning_rate": 6.295693110279123e-05,
      "loss": 2.007,
      "step": 182017
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.393773078918457,
      "learning_rate": 6.295442393004079e-05,
      "loss": 1.9376,
      "step": 182018
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.4847253561019897,
      "learning_rate": 6.295191680136153e-05,
      "loss": 2.118,
      "step": 182019
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.3378442525863647,
      "learning_rate": 6.294940971675404e-05,
      "loss": 2.1796,
      "step": 182020
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.3575778007507324,
      "learning_rate": 6.294690267621867e-05,
      "loss": 2.2024,
      "step": 182021
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.446907639503479,
      "learning_rate": 6.294439567975605e-05,
      "loss": 1.9475,
      "step": 182022
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.2243448495864868,
      "learning_rate": 6.29418887273665e-05,
      "loss": 1.7327,
      "step": 182023
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.458929181098938,
      "learning_rate": 6.293938181905053e-05,
      "loss": 1.7195,
      "step": 182024
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.3517026901245117,
      "learning_rate": 6.293687495480854e-05,
      "loss": 2.0369,
      "step": 182025
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.2474730014801025,
      "learning_rate": 6.293436813464112e-05,
      "loss": 1.9444,
      "step": 182026
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.2737298011779785,
      "learning_rate": 6.293186135854861e-05,
      "loss": 1.95,
      "step": 182027
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.288188099861145,
      "learning_rate": 6.292935462653162e-05,
      "loss": 2.0093,
      "step": 182028
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.842484712600708,
      "learning_rate": 6.292684793859048e-05,
      "loss": 1.9075,
      "step": 182029
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.4977786540985107,
      "learning_rate": 6.292434129472577e-05,
      "loss": 2.1828,
      "step": 182030
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.3686131238937378,
      "learning_rate": 6.29218346949379e-05,
      "loss": 2.0177,
      "step": 182031
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.3290743827819824,
      "learning_rate": 6.291932813922734e-05,
      "loss": 2.1003,
      "step": 182032
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.3740694522857666,
      "learning_rate": 6.291682162759446e-05,
      "loss": 2.0323,
      "step": 182033
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.3878446817398071,
      "learning_rate": 6.291431516003991e-05,
      "loss": 2.2586,
      "step": 182034
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.1678005456924438,
      "learning_rate": 6.291180873656399e-05,
      "loss": 1.7491,
      "step": 182035
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.2580243349075317,
      "learning_rate": 6.290930235716733e-05,
      "loss": 1.9291,
      "step": 182036
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.312317967414856,
      "learning_rate": 6.290679602185027e-05,
      "loss": 1.9803,
      "step": 182037
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.2204196453094482,
      "learning_rate": 6.290428973061326e-05,
      "loss": 1.9599,
      "step": 182038
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.3336352109909058,
      "learning_rate": 6.290178348345688e-05,
      "loss": 2.0464,
      "step": 182039
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.2994341850280762,
      "learning_rate": 6.289927728038154e-05,
      "loss": 2.0417,
      "step": 182040
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.3125611543655396,
      "learning_rate": 6.289677112138764e-05,
      "loss": 2.0687,
      "step": 182041
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.3131647109985352,
      "learning_rate": 6.289426500647579e-05,
      "loss": 1.9569,
      "step": 182042
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.393042802810669,
      "learning_rate": 6.289175893564633e-05,
      "loss": 2.0231,
      "step": 182043
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.5161458253860474,
      "learning_rate": 6.288925290889972e-05,
      "loss": 1.9945,
      "step": 182044
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.4680744409561157,
      "learning_rate": 6.288674692623653e-05,
      "loss": 2.173,
      "step": 182045
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.4587146043777466,
      "learning_rate": 6.288424098765719e-05,
      "loss": 1.9988,
      "step": 182046
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.5346463918685913,
      "learning_rate": 6.288173509316208e-05,
      "loss": 2.1757,
      "step": 182047
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.532135009765625,
      "learning_rate": 6.287922924275177e-05,
      "loss": 2.346,
      "step": 182048
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.3053061962127686,
      "learning_rate": 6.28767234364267e-05,
      "loss": 2.0247,
      "step": 182049
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.3359562158584595,
      "learning_rate": 6.287421767418726e-05,
      "loss": 2.0177,
      "step": 182050
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.4946180582046509,
      "learning_rate": 6.287171195603406e-05,
      "loss": 1.9762,
      "step": 182051
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.3132894039154053,
      "learning_rate": 6.28692062819674e-05,
      "loss": 1.9601,
      "step": 182052
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.4447764158248901,
      "learning_rate": 6.28667006519879e-05,
      "loss": 1.9358,
      "step": 182053
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.2634612321853638,
      "learning_rate": 6.286419506609597e-05,
      "loss": 2.0206,
      "step": 182054
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.4400153160095215,
      "learning_rate": 6.286168952429204e-05,
      "loss": 1.9997,
      "step": 182055
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.4917854070663452,
      "learning_rate": 6.285918402657653e-05,
      "loss": 2.1004,
      "step": 182056
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.6015777587890625,
      "learning_rate": 6.285667857295006e-05,
      "loss": 2.0271,
      "step": 182057
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.3294413089752197,
      "learning_rate": 6.285417316341294e-05,
      "loss": 2.0442,
      "step": 182058
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.4063167572021484,
      "learning_rate": 6.285166779796577e-05,
      "loss": 2.0829,
      "step": 182059
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.474565029144287,
      "learning_rate": 6.284916247660896e-05,
      "loss": 1.9186,
      "step": 182060
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.6111314296722412,
      "learning_rate": 6.284665719934295e-05,
      "loss": 1.9483,
      "step": 182061
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.60257089138031,
      "learning_rate": 6.284415196616815e-05,
      "loss": 1.7611,
      "step": 182062
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.275691032409668,
      "learning_rate": 6.28416467770852e-05,
      "loss": 1.8875,
      "step": 182063
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.4374395608901978,
      "learning_rate": 6.283914163209434e-05,
      "loss": 2.108,
      "step": 182064
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.341382622718811,
      "learning_rate": 6.283663653119629e-05,
      "loss": 1.7784,
      "step": 182065
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.2884913682937622,
      "learning_rate": 6.283413147439137e-05,
      "loss": 2.0153,
      "step": 182066
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.3289216756820679,
      "learning_rate": 6.283162646168004e-05,
      "loss": 1.9251,
      "step": 182067
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.4860066175460815,
      "learning_rate": 6.282912149306271e-05,
      "loss": 2.0644,
      "step": 182068
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.3677927255630493,
      "learning_rate": 6.282661656854004e-05,
      "loss": 1.9788,
      "step": 182069
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.2913587093353271,
      "learning_rate": 6.282411168811226e-05,
      "loss": 1.9875,
      "step": 182070
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.408721923828125,
      "learning_rate": 6.282160685178008e-05,
      "loss": 1.9245,
      "step": 182071
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.3050309419631958,
      "learning_rate": 6.281910205954378e-05,
      "loss": 1.7835,
      "step": 182072
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.279340386390686,
      "learning_rate": 6.281659731140391e-05,
      "loss": 1.9587,
      "step": 182073
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.3462450504302979,
      "learning_rate": 6.281409260736084e-05,
      "loss": 1.7924,
      "step": 182074
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.3811304569244385,
      "learning_rate": 6.281158794741519e-05,
      "loss": 2.0621,
      "step": 182075
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.2997984886169434,
      "learning_rate": 6.280908333156727e-05,
      "loss": 2.0541,
      "step": 182076
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.3119152784347534,
      "learning_rate": 6.280657875981768e-05,
      "loss": 1.9327,
      "step": 182077
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.489019513130188,
      "learning_rate": 6.280407423216684e-05,
      "loss": 1.9867,
      "step": 182078
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.3334256410598755,
      "learning_rate": 6.28015697486152e-05,
      "loss": 2.1132,
      "step": 182079
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.3581960201263428,
      "learning_rate": 6.279906530916311e-05,
      "loss": 2.1062,
      "step": 182080
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.4289579391479492,
      "learning_rate": 6.279656091381127e-05,
      "loss": 2.11,
      "step": 182081
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.35507071018219,
      "learning_rate": 6.279405656255993e-05,
      "loss": 2.039,
      "step": 182082
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.2771049737930298,
      "learning_rate": 6.279155225540975e-05,
      "loss": 1.9596,
      "step": 182083
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.3218891620635986,
      "learning_rate": 6.278904799236109e-05,
      "loss": 1.9113,
      "step": 182084
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.3054226636886597,
      "learning_rate": 6.278654377341442e-05,
      "loss": 1.9624,
      "step": 182085
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.3247127532958984,
      "learning_rate": 6.278403959857012e-05,
      "loss": 2.0859,
      "step": 182086
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.4185377359390259,
      "learning_rate": 6.278153546782883e-05,
      "loss": 2.0263,
      "step": 182087
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.2961983680725098,
      "learning_rate": 6.277903138119085e-05,
      "loss": 1.9973,
      "step": 182088
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.2742955684661865,
      "learning_rate": 6.277652733865683e-05,
      "loss": 1.9264,
      "step": 182089
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.2875807285308838,
      "learning_rate": 6.277402334022712e-05,
      "loss": 2.1617,
      "step": 182090
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.3131392002105713,
      "learning_rate": 6.277151938590217e-05,
      "loss": 2.1196,
      "step": 182091
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.2859960794448853,
      "learning_rate": 6.27690154756824e-05,
      "loss": 2.0232,
      "step": 182092
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.2980509996414185,
      "learning_rate": 6.276651160956846e-05,
      "loss": 1.9716,
      "step": 182093
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.5104533433914185,
      "learning_rate": 6.27640077875606e-05,
      "loss": 2.0542,
      "step": 182094
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.3534530401229858,
      "learning_rate": 6.276150400965949e-05,
      "loss": 1.9034,
      "step": 182095
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.3669259548187256,
      "learning_rate": 6.275900027586536e-05,
      "loss": 2.1762,
      "step": 182096
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.487754225730896,
      "learning_rate": 6.275649658617903e-05,
      "loss": 1.8493,
      "step": 182097
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.3962996006011963,
      "learning_rate": 6.275399294060058e-05,
      "loss": 1.9006,
      "step": 182098
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.315958857536316,
      "learning_rate": 6.27514893391307e-05,
      "loss": 2.1679,
      "step": 182099
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.3500733375549316,
      "learning_rate": 6.27489857817697e-05,
      "loss": 2.0661,
      "step": 182100
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.5080645084381104,
      "learning_rate": 6.274648226851824e-05,
      "loss": 2.1715,
      "step": 182101
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.4460835456848145,
      "learning_rate": 6.274397879937661e-05,
      "loss": 2.2634,
      "step": 182102
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.3700642585754395,
      "learning_rate": 6.274147537434553e-05,
      "loss": 2.1207,
      "step": 182103
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.3102705478668213,
      "learning_rate": 6.273897199342511e-05,
      "loss": 1.8889,
      "step": 182104
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.3936952352523804,
      "learning_rate": 6.273646865661607e-05,
      "loss": 2.1464,
      "step": 182105
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.4090666770935059,
      "learning_rate": 6.273396536391874e-05,
      "loss": 1.8931,
      "step": 182106
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.2239407300949097,
      "learning_rate": 6.273146211533369e-05,
      "loss": 1.9957,
      "step": 182107
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.715385913848877,
      "learning_rate": 6.272895891086128e-05,
      "loss": 2.165,
      "step": 182108
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.3006267547607422,
      "learning_rate": 6.272645575050219e-05,
      "loss": 2.0542,
      "step": 182109
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.4819473028182983,
      "learning_rate": 6.272395263425658e-05,
      "loss": 2.0852,
      "step": 182110
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.3395776748657227,
      "learning_rate": 6.272144956212514e-05,
      "loss": 1.9091,
      "step": 182111
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.609250545501709,
      "learning_rate": 6.271894653410819e-05,
      "loss": 1.9993,
      "step": 182112
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.3516591787338257,
      "learning_rate": 6.271644355020632e-05,
      "loss": 1.9415,
      "step": 182113
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.287466287612915,
      "learning_rate": 6.271394061041988e-05,
      "loss": 1.8782,
      "step": 182114
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.3264260292053223,
      "learning_rate": 6.27114377147495e-05,
      "loss": 2.1205,
      "step": 182115
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.5336097478866577,
      "learning_rate": 6.270893486319552e-05,
      "loss": 2.0723,
      "step": 182116
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.3637055158615112,
      "learning_rate": 6.270643205575843e-05,
      "loss": 1.9277,
      "step": 182117
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.223023533821106,
      "learning_rate": 6.27039292924386e-05,
      "loss": 1.8542,
      "step": 182118
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.3728065490722656,
      "learning_rate": 6.270142657323669e-05,
      "loss": 2.0005,
      "step": 182119
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.4563469886779785,
      "learning_rate": 6.269892389815298e-05,
      "loss": 1.8833,
      "step": 182120
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.3560463190078735,
      "learning_rate": 6.26964212671881e-05,
      "loss": 1.9399,
      "step": 182121
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.4280506372451782,
      "learning_rate": 6.269391868034244e-05,
      "loss": 1.8697,
      "step": 182122
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.2957488298416138,
      "learning_rate": 6.269141613761636e-05,
      "loss": 1.839,
      "step": 182123
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.383970022201538,
      "learning_rate": 6.268891363901052e-05,
      "loss": 2.0724,
      "step": 182124
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.301944375038147,
      "learning_rate": 6.268641118452529e-05,
      "loss": 1.986,
      "step": 182125
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.3440250158309937,
      "learning_rate": 6.268390877416104e-05,
      "loss": 1.8976,
      "step": 182126
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.4260503053665161,
      "learning_rate": 6.268140640791845e-05,
      "loss": 2.1401,
      "step": 182127
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.360821008682251,
      "learning_rate": 6.267890408579783e-05,
      "loss": 2.2421,
      "step": 182128
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.2557305097579956,
      "learning_rate": 6.267640180779959e-05,
      "loss": 2.1536,
      "step": 182129
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.2995439767837524,
      "learning_rate": 6.267389957392439e-05,
      "loss": 2.0997,
      "step": 182130
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.3034380674362183,
      "learning_rate": 6.267139738417259e-05,
      "loss": 1.9129,
      "step": 182131
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.3639217615127563,
      "learning_rate": 6.266889523854461e-05,
      "loss": 2.3326,
      "step": 182132
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.2279305458068848,
      "learning_rate": 6.266639313704099e-05,
      "loss": 1.8196,
      "step": 182133
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.2991379499435425,
      "learning_rate": 6.26638910796622e-05,
      "loss": 1.9923,
      "step": 182134
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.2886450290679932,
      "learning_rate": 6.266138906640855e-05,
      "loss": 1.8764,
      "step": 182135
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.2846331596374512,
      "learning_rate": 6.265888709728075e-05,
      "loss": 1.9558,
      "step": 182136
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.4040430784225464,
      "learning_rate": 6.265638517227907e-05,
      "loss": 2.0163,
      "step": 182137
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.5911670923233032,
      "learning_rate": 6.26538832914041e-05,
      "loss": 1.9611,
      "step": 182138
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.3742256164550781,
      "learning_rate": 6.265138145465625e-05,
      "loss": 2.2579,
      "step": 182139
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.2466458082199097,
      "learning_rate": 6.264887966203602e-05,
      "loss": 1.8347,
      "step": 182140
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.4638981819152832,
      "learning_rate": 6.264637791354372e-05,
      "loss": 2.022,
      "step": 182141
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.538847804069519,
      "learning_rate": 6.264387620918006e-05,
      "loss": 2.1155,
      "step": 182142
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.4362492561340332,
      "learning_rate": 6.264137454894528e-05,
      "loss": 1.7913,
      "step": 182143
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.2372645139694214,
      "learning_rate": 6.263887293284002e-05,
      "loss": 2.1059,
      "step": 182144
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.2334569692611694,
      "learning_rate": 6.263637136086472e-05,
      "loss": 2.006,
      "step": 182145
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.4678438901901245,
      "learning_rate": 6.263386983301976e-05,
      "loss": 1.8417,
      "step": 182146
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.3850246667861938,
      "learning_rate": 6.263136834930556e-05,
      "loss": 2.2679,
      "step": 182147
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.3682047128677368,
      "learning_rate": 6.262886690972278e-05,
      "loss": 2.0337,
      "step": 182148
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.395188570022583,
      "learning_rate": 6.262636551427167e-05,
      "loss": 2.1373,
      "step": 182149
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.3630173206329346,
      "learning_rate": 6.262386416295288e-05,
      "loss": 1.9226,
      "step": 182150
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.2504687309265137,
      "learning_rate": 6.26213628557668e-05,
      "loss": 2.0665,
      "step": 182151
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.486463189125061,
      "learning_rate": 6.261886159271388e-05,
      "loss": 2.082,
      "step": 182152
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.3386168479919434,
      "learning_rate": 6.261636037379454e-05,
      "loss": 1.8113,
      "step": 182153
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.3588167428970337,
      "learning_rate": 6.261385919900939e-05,
      "loss": 2.1303,
      "step": 182154
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.469193458557129,
      "learning_rate": 6.261135806835867e-05,
      "loss": 1.9863,
      "step": 182155
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.2909339666366577,
      "learning_rate": 6.260885698184311e-05,
      "loss": 2.0992,
      "step": 182156
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.2960149049758911,
      "learning_rate": 6.2606355939463e-05,
      "loss": 1.8102,
      "step": 182157
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.4347028732299805,
      "learning_rate": 6.26038549412189e-05,
      "loss": 2.0173,
      "step": 182158
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.286150574684143,
      "learning_rate": 6.260135398711109e-05,
      "loss": 1.7645,
      "step": 182159
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.3047339916229248,
      "learning_rate": 6.259885307714027e-05,
      "loss": 2.0466,
      "step": 182160
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.252778172492981,
      "learning_rate": 6.259635221130676e-05,
      "loss": 2.0184,
      "step": 182161
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.3560428619384766,
      "learning_rate": 6.259385138961113e-05,
      "loss": 1.9724,
      "step": 182162
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.3622198104858398,
      "learning_rate": 6.25913506120537e-05,
      "loss": 1.9383,
      "step": 182163
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.3341575860977173,
      "learning_rate": 6.258884987863515e-05,
      "loss": 2.1802,
      "step": 182164
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.2963652610778809,
      "learning_rate": 6.258634918935567e-05,
      "loss": 1.6656,
      "step": 182165
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.354392170906067,
      "learning_rate": 6.258384854421598e-05,
      "loss": 1.9304,
      "step": 182166
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.413824200630188,
      "learning_rate": 6.258134794321632e-05,
      "loss": 2.0184,
      "step": 182167
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.545878291130066,
      "learning_rate": 6.257884738635737e-05,
      "loss": 2.0969,
      "step": 182168
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.2442799806594849,
      "learning_rate": 6.257634687363941e-05,
      "loss": 1.9033,
      "step": 182169
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.475547432899475,
      "learning_rate": 6.257384640506312e-05,
      "loss": 2.0569,
      "step": 182170
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.4729377031326294,
      "learning_rate": 6.257134598062872e-05,
      "loss": 1.8441,
      "step": 182171
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.330057144165039,
      "learning_rate": 6.256884560033684e-05,
      "loss": 1.911,
      "step": 182172
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.4202016592025757,
      "learning_rate": 6.256634526418782e-05,
      "loss": 2.087,
      "step": 182173
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.2940829992294312,
      "learning_rate": 6.256384497218227e-05,
      "loss": 1.953,
      "step": 182174
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.293222427368164,
      "learning_rate": 6.256134472432053e-05,
      "loss": 1.9159,
      "step": 182175
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.4085794687271118,
      "learning_rate": 6.255884452060326e-05,
      "loss": 1.7673,
      "step": 182176
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.3175781965255737,
      "learning_rate": 6.255634436103063e-05,
      "loss": 2.2515,
      "step": 182177
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.3613147735595703,
      "learning_rate": 6.255384424560331e-05,
      "loss": 2.1394,
      "step": 182178
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.4004321098327637,
      "learning_rate": 6.255134417432164e-05,
      "loss": 1.8105,
      "step": 182179
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.3094472885131836,
      "learning_rate": 6.254884414718627e-05,
      "loss": 1.9738,
      "step": 182180
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.3776990175247192,
      "learning_rate": 6.254634416419745e-05,
      "loss": 2.1638,
      "step": 182181
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.4880785942077637,
      "learning_rate": 6.254384422535591e-05,
      "loss": 1.9825,
      "step": 182182
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.4378644227981567,
      "learning_rate": 6.254134433066179e-05,
      "loss": 1.9386,
      "step": 182183
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.300445795059204,
      "learning_rate": 6.253884448011579e-05,
      "loss": 2.2604,
      "step": 182184
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.1723145246505737,
      "learning_rate": 6.253634467371824e-05,
      "loss": 1.8203,
      "step": 182185
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.6444532871246338,
      "learning_rate": 6.253384491146975e-05,
      "loss": 1.913,
      "step": 182186
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.2459877729415894,
      "learning_rate": 6.25313451933706e-05,
      "loss": 2.2178,
      "step": 182187
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.3405046463012695,
      "learning_rate": 6.25288455194215e-05,
      "loss": 2.0121,
      "step": 182188
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.4248229265213013,
      "learning_rate": 6.252634588962265e-05,
      "loss": 1.9881,
      "step": 182189
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.3823753595352173,
      "learning_rate": 6.252384630397469e-05,
      "loss": 1.7654,
      "step": 182190
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.440346598625183,
      "learning_rate": 6.252134676247794e-05,
      "loss": 1.951,
      "step": 182191
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.348785161972046,
      "learning_rate": 6.251884726513304e-05,
      "loss": 1.994,
      "step": 182192
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.3055115938186646,
      "learning_rate": 6.251634781194032e-05,
      "loss": 2.057,
      "step": 182193
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.432961344718933,
      "learning_rate": 6.251384840290043e-05,
      "loss": 2.1298,
      "step": 182194
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.6499273777008057,
      "learning_rate": 6.251134903801354e-05,
      "loss": 2.0908,
      "step": 182195
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.3810423612594604,
      "learning_rate": 6.250884971728036e-05,
      "loss": 2.1144,
      "step": 182196
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.3492542505264282,
      "learning_rate": 6.250635044070117e-05,
      "loss": 2.1266,
      "step": 182197
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.2846992015838623,
      "learning_rate": 6.250385120827663e-05,
      "loss": 1.983,
      "step": 182198
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.375826358795166,
      "learning_rate": 6.250135202000702e-05,
      "loss": 1.9965,
      "step": 182199
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.3406517505645752,
      "learning_rate": 6.249885287589297e-05,
      "loss": 2.0854,
      "step": 182200
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.2493255138397217,
      "learning_rate": 6.249635377593489e-05,
      "loss": 1.8665,
      "step": 182201
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.4511672258377075,
      "learning_rate": 6.24938547201332e-05,
      "loss": 1.9335,
      "step": 182202
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.2290065288543701,
      "learning_rate": 6.249135570848831e-05,
      "loss": 2.0352,
      "step": 182203
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.364478349685669,
      "learning_rate": 6.248885674100083e-05,
      "loss": 2.1545,
      "step": 182204
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.4074366092681885,
      "learning_rate": 6.248635781767108e-05,
      "loss": 1.8582,
      "step": 182205
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.3791223764419556,
      "learning_rate": 6.248385893849969e-05,
      "loss": 2.3083,
      "step": 182206
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.49299156665802,
      "learning_rate": 6.248136010348701e-05,
      "loss": 2.0418,
      "step": 182207
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.3690016269683838,
      "learning_rate": 6.247886131263354e-05,
      "loss": 2.0721,
      "step": 182208
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.5210288763046265,
      "learning_rate": 6.247636256593965e-05,
      "loss": 1.9174,
      "step": 182209
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.3846403360366821,
      "learning_rate": 6.2473863863406e-05,
      "loss": 2.0568,
      "step": 182210
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.2490698099136353,
      "learning_rate": 6.247136520503283e-05,
      "loss": 1.9014,
      "step": 182211
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.411769986152649,
      "learning_rate": 6.24688665908208e-05,
      "loss": 2.1304,
      "step": 182212
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.384965181350708,
      "learning_rate": 6.24663680207703e-05,
      "loss": 1.9536,
      "step": 182213
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.354332447052002,
      "learning_rate": 6.24638694948817e-05,
      "loss": 1.9639,
      "step": 182214
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.3915749788284302,
      "learning_rate": 6.246137101315565e-05,
      "loss": 1.9376,
      "step": 182215
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.3527674674987793,
      "learning_rate": 6.245887257559247e-05,
      "loss": 2.0591,
      "step": 182216
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.2500139474868774,
      "learning_rate": 6.245637418219261e-05,
      "loss": 1.8563,
      "step": 182217
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.4753358364105225,
      "learning_rate": 6.245387583295669e-05,
      "loss": 2.0532,
      "step": 182218
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.5655531883239746,
      "learning_rate": 6.245137752788507e-05,
      "loss": 1.9331,
      "step": 182219
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.269322156906128,
      "learning_rate": 6.244887926697815e-05,
      "loss": 2.0605,
      "step": 182220
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.26163911819458,
      "learning_rate": 6.244638105023655e-05,
      "loss": 1.9468,
      "step": 182221
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.2366721630096436,
      "learning_rate": 6.244388287766054e-05,
      "loss": 2.1651,
      "step": 182222
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.2887487411499023,
      "learning_rate": 6.244138474925082e-05,
      "loss": 2.2462,
      "step": 182223
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.412414312362671,
      "learning_rate": 6.243888666500772e-05,
      "loss": 2.0149,
      "step": 182224
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.4644170999526978,
      "learning_rate": 6.243638862493168e-05,
      "loss": 2.1799,
      "step": 182225
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.4385733604431152,
      "learning_rate": 6.243389062902316e-05,
      "loss": 2.1323,
      "step": 182226
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.4969784021377563,
      "learning_rate": 6.243139267728271e-05,
      "loss": 2.0564,
      "step": 182227
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.2483967542648315,
      "learning_rate": 6.24288947697107e-05,
      "loss": 1.9258,
      "step": 182228
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.326338291168213,
      "learning_rate": 6.242639690630771e-05,
      "loss": 2.0008,
      "step": 182229
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.2928367853164673,
      "learning_rate": 6.242389908707413e-05,
      "loss": 1.9747,
      "step": 182230
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.295738697052002,
      "learning_rate": 6.242140131201044e-05,
      "loss": 1.8775,
      "step": 182231
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.3875837326049805,
      "learning_rate": 6.241890358111703e-05,
      "loss": 1.8688,
      "step": 182232
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.415172815322876,
      "learning_rate": 6.241640589439448e-05,
      "loss": 1.8444,
      "step": 182233
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.3229992389678955,
      "learning_rate": 6.241390825184313e-05,
      "loss": 1.9415,
      "step": 182234
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.291791319847107,
      "learning_rate": 6.241141065346362e-05,
      "loss": 2.0829,
      "step": 182235
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.2481722831726074,
      "learning_rate": 6.240891309925624e-05,
      "loss": 2.1259,
      "step": 182236
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.3699761629104614,
      "learning_rate": 6.240641558922166e-05,
      "loss": 2.0106,
      "step": 182237
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.234597086906433,
      "learning_rate": 6.240391812336007e-05,
      "loss": 2.0355,
      "step": 182238
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.5930551290512085,
      "learning_rate": 6.240142070167215e-05,
      "loss": 2.0194,
      "step": 182239
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.4895079135894775,
      "learning_rate": 6.23989233241582e-05,
      "loss": 2.096,
      "step": 182240
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.2762682437896729,
      "learning_rate": 6.239642599081888e-05,
      "loss": 2.0969,
      "step": 182241
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.2872982025146484,
      "learning_rate": 6.239392870165445e-05,
      "loss": 1.8401,
      "step": 182242
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.3022515773773193,
      "learning_rate": 6.239143145666565e-05,
      "loss": 2.0736,
      "step": 182243
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.3108106851577759,
      "learning_rate": 6.238893425585258e-05,
      "loss": 1.9301,
      "step": 182244
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.3320542573928833,
      "learning_rate": 6.238643709921602e-05,
      "loss": 2.1163,
      "step": 182245
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.4838861227035522,
      "learning_rate": 6.238393998675617e-05,
      "loss": 2.1012,
      "step": 182246
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.4745094776153564,
      "learning_rate": 6.238144291847376e-05,
      "loss": 2.1517,
      "step": 182247
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.255493402481079,
      "learning_rate": 6.237894589436901e-05,
      "loss": 1.9225,
      "step": 182248
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.2807940244674683,
      "learning_rate": 6.237644891444267e-05,
      "loss": 1.9652,
      "step": 182249
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.2915130853652954,
      "learning_rate": 6.237395197869488e-05,
      "loss": 1.9149,
      "step": 182250
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.5324119329452515,
      "learning_rate": 6.237145508712634e-05,
      "loss": 2.0385,
      "step": 182251
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.3372471332550049,
      "learning_rate": 6.236895823973734e-05,
      "loss": 2.0688,
      "step": 182252
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.4448575973510742,
      "learning_rate": 6.236646143652852e-05,
      "loss": 2.0404,
      "step": 182253
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.3494720458984375,
      "learning_rate": 6.23639646775002e-05,
      "loss": 2.1759,
      "step": 182254
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.4269211292266846,
      "learning_rate": 6.236146796265302e-05,
      "loss": 2.1143,
      "step": 182255
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.532127857208252,
      "learning_rate": 6.235897129198722e-05,
      "loss": 1.9666,
      "step": 182256
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.3208986520767212,
      "learning_rate": 6.235647466550339e-05,
      "loss": 1.9444,
      "step": 182257
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.374279260635376,
      "learning_rate": 6.235397808320197e-05,
      "loss": 2.0109,
      "step": 182258
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.4218908548355103,
      "learning_rate": 6.235148154508347e-05,
      "loss": 2.1063,
      "step": 182259
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.2880321741104126,
      "learning_rate": 6.234898505114823e-05,
      "loss": 2.0794,
      "step": 182260
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.3826242685317993,
      "learning_rate": 6.234648860139697e-05,
      "loss": 2.1184,
      "step": 182261
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.3929409980773926,
      "learning_rate": 6.234399219582985e-05,
      "loss": 2.2395,
      "step": 182262
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.3419972658157349,
      "learning_rate": 6.234149583444753e-05,
      "loss": 1.8902,
      "step": 182263
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.207047939300537,
      "learning_rate": 6.233899951725031e-05,
      "loss": 1.696,
      "step": 182264
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.3058736324310303,
      "learning_rate": 6.233650324423885e-05,
      "loss": 1.9304,
      "step": 182265
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.2680914402008057,
      "learning_rate": 6.233400701541344e-05,
      "loss": 1.7821,
      "step": 182266
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.3978344202041626,
      "learning_rate": 6.233151083077481e-05,
      "loss": 2.1441,
      "step": 182267
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.3303148746490479,
      "learning_rate": 6.232901469032305e-05,
      "loss": 2.2474,
      "step": 182268
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.7138853073120117,
      "learning_rate": 6.232651859405889e-05,
      "loss": 2.0463,
      "step": 182269
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.3587971925735474,
      "learning_rate": 6.232402254198267e-05,
      "loss": 1.9355,
      "step": 182270
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.7231637239456177,
      "learning_rate": 6.232152653409492e-05,
      "loss": 2.0616,
      "step": 182271
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.468995451927185,
      "learning_rate": 6.231903057039607e-05,
      "loss": 2.1336,
      "step": 182272
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.4453905820846558,
      "learning_rate": 6.231653465088673e-05,
      "loss": 2.1872,
      "step": 182273
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.4516416788101196,
      "learning_rate": 6.231403877556704e-05,
      "loss": 2.0397,
      "step": 182274
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.345098853111267,
      "learning_rate": 6.231154294443777e-05,
      "loss": 1.9669,
      "step": 182275
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.5098941326141357,
      "learning_rate": 6.230904715749922e-05,
      "loss": 2.0531,
      "step": 182276
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.5555365085601807,
      "learning_rate": 6.230655141475196e-05,
      "loss": 1.8663,
      "step": 182277
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.554398775100708,
      "learning_rate": 6.230405571619631e-05,
      "loss": 2.1839,
      "step": 182278
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.46483314037323,
      "learning_rate": 6.230156006183298e-05,
      "loss": 2.0143,
      "step": 182279
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.2780401706695557,
      "learning_rate": 6.229906445166213e-05,
      "loss": 1.9152,
      "step": 182280
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.4475685358047485,
      "learning_rate": 6.229656888568442e-05,
      "loss": 2.0075,
      "step": 182281
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.739597201347351,
      "learning_rate": 6.229407336390021e-05,
      "loss": 2.144,
      "step": 182282
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.5650700330734253,
      "learning_rate": 6.229157788631011e-05,
      "loss": 2.1635,
      "step": 182283
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.5209778547286987,
      "learning_rate": 6.228908245291442e-05,
      "loss": 2.1107,
      "step": 182284
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.2974745035171509,
      "learning_rate": 6.228658706371372e-05,
      "loss": 2.336,
      "step": 182285
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.2942540645599365,
      "learning_rate": 6.228409171870845e-05,
      "loss": 1.942,
      "step": 182286
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.4028393030166626,
      "learning_rate": 6.228159641789907e-05,
      "loss": 1.9876,
      "step": 182287
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.305500864982605,
      "learning_rate": 6.227910116128591e-05,
      "loss": 2.1838,
      "step": 182288
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.3482624292373657,
      "learning_rate": 6.227660594886965e-05,
      "loss": 2.0054,
      "step": 182289
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.283382534980774,
      "learning_rate": 6.227411078065058e-05,
      "loss": 1.9163,
      "step": 182290
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.305160403251648,
      "learning_rate": 6.227161565662934e-05,
      "loss": 1.9133,
      "step": 182291
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.3187226057052612,
      "learning_rate": 6.226912057680627e-05,
      "loss": 1.8396,
      "step": 182292
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.559169888496399,
      "learning_rate": 6.226662554118186e-05,
      "loss": 2.1226,
      "step": 182293
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.4025744199752808,
      "learning_rate": 6.22641305497565e-05,
      "loss": 2.0683,
      "step": 182294
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.391693353652954,
      "learning_rate": 6.226163560253078e-05,
      "loss": 2.0401,
      "step": 182295
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.2740925550460815,
      "learning_rate": 6.225914069950507e-05,
      "loss": 1.8134,
      "step": 182296
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.5700410604476929,
      "learning_rate": 6.225664584067994e-05,
      "loss": 1.9146,
      "step": 182297
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.6072334051132202,
      "learning_rate": 6.225415102605576e-05,
      "loss": 2.113,
      "step": 182298
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.3344409465789795,
      "learning_rate": 6.225165625563298e-05,
      "loss": 1.9452,
      "step": 182299
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.9751139879226685,
      "learning_rate": 6.224916152941217e-05,
      "loss": 1.9514,
      "step": 182300
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.4505860805511475,
      "learning_rate": 6.224666684739371e-05,
      "loss": 1.9077,
      "step": 182301
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.4614230394363403,
      "learning_rate": 6.224417220957803e-05,
      "loss": 1.8584,
      "step": 182302
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.5275421142578125,
      "learning_rate": 6.224167761596571e-05,
      "loss": 1.8323,
      "step": 182303
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.4269295930862427,
      "learning_rate": 6.223918306655714e-05,
      "loss": 2.0214,
      "step": 182304
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.305420994758606,
      "learning_rate": 6.223668856135274e-05,
      "loss": 1.931,
      "step": 182305
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.3784042596817017,
      "learning_rate": 6.223419410035309e-05,
      "loss": 2.0557,
      "step": 182306
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.3458317518234253,
      "learning_rate": 6.22316996835586e-05,
      "loss": 2.1677,
      "step": 182307
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.2675516605377197,
      "learning_rate": 6.222920531096966e-05,
      "loss": 1.9944,
      "step": 182308
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.4574593305587769,
      "learning_rate": 6.222671098258685e-05,
      "loss": 2.2286,
      "step": 182309
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.3986456394195557,
      "learning_rate": 6.22242166984106e-05,
      "loss": 1.975,
      "step": 182310
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.3970927000045776,
      "learning_rate": 6.222172245844128e-05,
      "loss": 2.1621,
      "step": 182311
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.3108458518981934,
      "learning_rate": 6.221922826267949e-05,
      "loss": 2.0109,
      "step": 182312
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.4274742603302002,
      "learning_rate": 6.221673411112557e-05,
      "loss": 1.927,
      "step": 182313
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.3193961381912231,
      "learning_rate": 6.221424000378011e-05,
      "loss": 2.1655,
      "step": 182314
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.330175757408142,
      "learning_rate": 6.221174594064353e-05,
      "loss": 2.1637,
      "step": 182315
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.4176522493362427,
      "learning_rate": 6.220925192171629e-05,
      "loss": 2.096,
      "step": 182316
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.4911717176437378,
      "learning_rate": 6.220675794699871e-05,
      "loss": 1.9159,
      "step": 182317
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.3409372568130493,
      "learning_rate": 6.22042640164915e-05,
      "loss": 2.282,
      "step": 182318
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.407810926437378,
      "learning_rate": 6.220177013019492e-05,
      "loss": 1.9961,
      "step": 182319
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.363765835762024,
      "learning_rate": 6.219927628810958e-05,
      "loss": 2.0148,
      "step": 182320
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.3566194772720337,
      "learning_rate": 6.219678249023581e-05,
      "loss": 2.0219,
      "step": 182321
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.3973021507263184,
      "learning_rate": 6.219428873657434e-05,
      "loss": 1.8816,
      "step": 182322
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.4173812866210938,
      "learning_rate": 6.219179502712524e-05,
      "loss": 2.1232,
      "step": 182323
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.3234306573867798,
      "learning_rate": 6.218930136188927e-05,
      "loss": 2.316,
      "step": 182324
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.3033902645111084,
      "learning_rate": 6.218680774086674e-05,
      "loss": 1.8526,
      "step": 182325
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.2598854303359985,
      "learning_rate": 6.218431416405824e-05,
      "loss": 1.8734,
      "step": 182326
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.2774244546890259,
      "learning_rate": 6.218182063146406e-05,
      "loss": 1.9189,
      "step": 182327
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.3920137882232666,
      "learning_rate": 6.217932714308496e-05,
      "loss": 2.0644,
      "step": 182328
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.5265140533447266,
      "learning_rate": 6.217683369892101e-05,
      "loss": 1.8663,
      "step": 182329
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.2269890308380127,
      "learning_rate": 6.2174340298973e-05,
      "loss": 1.81,
      "step": 182330
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.7266509532928467,
      "learning_rate": 6.217184694324119e-05,
      "loss": 1.9952,
      "step": 182331
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.375443935394287,
      "learning_rate": 6.216935363172618e-05,
      "loss": 2.0785,
      "step": 182332
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.5238093137741089,
      "learning_rate": 6.216686036442829e-05,
      "loss": 1.9394,
      "step": 182333
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.2986952066421509,
      "learning_rate": 6.216436714134826e-05,
      "loss": 2.1847,
      "step": 182334
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.290372610092163,
      "learning_rate": 6.216187396248617e-05,
      "loss": 2.1998,
      "step": 182335
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.355916976928711,
      "learning_rate": 6.215938082784277e-05,
      "loss": 1.9455,
      "step": 182336
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.259263515472412,
      "learning_rate": 6.215688773741836e-05,
      "loss": 1.7538,
      "step": 182337
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.2813820838928223,
      "learning_rate": 6.215439469121355e-05,
      "loss": 2.1399,
      "step": 182338
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.2409192323684692,
      "learning_rate": 6.215190168922864e-05,
      "loss": 2.1551,
      "step": 182339
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.3376508951187134,
      "learning_rate": 6.214940873146434e-05,
      "loss": 2.2227,
      "step": 182340
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.295649528503418,
      "learning_rate": 6.214691581792082e-05,
      "loss": 2.3954,
      "step": 182341
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.6112834215164185,
      "learning_rate": 6.214442294859872e-05,
      "loss": 2.0233,
      "step": 182342
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.4341050386428833,
      "learning_rate": 6.21419301234984e-05,
      "loss": 2.0768,
      "step": 182343
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.3945891857147217,
      "learning_rate": 6.213943734262049e-05,
      "loss": 2.14,
      "step": 182344
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.338232398033142,
      "learning_rate": 6.213694460596523e-05,
      "loss": 2.1795,
      "step": 182345
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.3651224374771118,
      "learning_rate": 6.213445191353337e-05,
      "loss": 2.0938,
      "step": 182346
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.478908896446228,
      "learning_rate": 6.213195926532506e-05,
      "loss": 2.0008,
      "step": 182347
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.2795162200927734,
      "learning_rate": 6.212946666134097e-05,
      "loss": 1.8437,
      "step": 182348
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.4387339353561401,
      "learning_rate": 6.212697410158146e-05,
      "loss": 2.1571,
      "step": 182349
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.3855581283569336,
      "learning_rate": 6.212448158604709e-05,
      "loss": 2.0101,
      "step": 182350
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.291354775428772,
      "learning_rate": 6.212198911473816e-05,
      "loss": 2.1136,
      "step": 182351
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.2402622699737549,
      "learning_rate": 6.211949668765545e-05,
      "loss": 1.9851,
      "step": 182352
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.4175559282302856,
      "learning_rate": 6.211700430479905e-05,
      "loss": 1.9782,
      "step": 182353
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.297722578048706,
      "learning_rate": 6.211451196616964e-05,
      "loss": 1.8676,
      "step": 182354
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.354692816734314,
      "learning_rate": 6.211201967176758e-05,
      "loss": 1.9926,
      "step": 182355
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.6018182039260864,
      "learning_rate": 6.210952742159344e-05,
      "loss": 2.2768,
      "step": 182356
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.3539206981658936,
      "learning_rate": 6.210703521564756e-05,
      "loss": 2.2578,
      "step": 182357
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.3889212608337402,
      "learning_rate": 6.210454305393063e-05,
      "loss": 2.1921,
      "step": 182358
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.4597054719924927,
      "learning_rate": 6.210205093644283e-05,
      "loss": 2.0446,
      "step": 182359
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.3558205366134644,
      "learning_rate": 6.20995588631848e-05,
      "loss": 2.0156,
      "step": 182360
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.2988649606704712,
      "learning_rate": 6.209706683415687e-05,
      "loss": 2.1736,
      "step": 182361
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.297729730606079,
      "learning_rate": 6.209457484935966e-05,
      "loss": 2.0531,
      "step": 182362
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.324478268623352,
      "learning_rate": 6.209208290879349e-05,
      "loss": 2.0444,
      "step": 182363
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.2907910346984863,
      "learning_rate": 6.208959101245898e-05,
      "loss": 2.1286,
      "step": 182364
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.3491016626358032,
      "learning_rate": 6.208709916035648e-05,
      "loss": 1.8315,
      "step": 182365
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.4516874551773071,
      "learning_rate": 6.208460735248651e-05,
      "loss": 2.0183,
      "step": 182366
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.2998526096343994,
      "learning_rate": 6.208211558884942e-05,
      "loss": 2.0701,
      "step": 182367
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.2411047220230103,
      "learning_rate": 6.207962386944582e-05,
      "loss": 1.9462,
      "step": 182368
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.4811586141586304,
      "learning_rate": 6.207713219427606e-05,
      "loss": 2.0761,
      "step": 182369
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.272773265838623,
      "learning_rate": 6.207464056334068e-05,
      "loss": 2.1089,
      "step": 182370
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.3416575193405151,
      "learning_rate": 6.207214897664016e-05,
      "loss": 1.8134,
      "step": 182371
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.5030715465545654,
      "learning_rate": 6.206965743417491e-05,
      "loss": 1.9814,
      "step": 182372
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.4325658082962036,
      "learning_rate": 6.206716593594532e-05,
      "loss": 2.0281,
      "step": 182373
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.2883553504943848,
      "learning_rate": 6.2064674481952e-05,
      "loss": 1.9061,
      "step": 182374
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.3164098262786865,
      "learning_rate": 6.206218307219526e-05,
      "loss": 2.0751,
      "step": 182375
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.426740050315857,
      "learning_rate": 6.205969170667575e-05,
      "loss": 1.7714,
      "step": 182376
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.3421132564544678,
      "learning_rate": 6.205720038539384e-05,
      "loss": 1.9286,
      "step": 182377
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.2970446348190308,
      "learning_rate": 6.205470910834999e-05,
      "loss": 1.9647,
      "step": 182378
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.4541648626327515,
      "learning_rate": 6.205221787554457e-05,
      "loss": 2.0342,
      "step": 182379
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.5322442054748535,
      "learning_rate": 6.204972668697821e-05,
      "loss": 2.1367,
      "step": 182380
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.3498412370681763,
      "learning_rate": 6.204723554265122e-05,
      "loss": 2.0838,
      "step": 182381
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.2735694646835327,
      "learning_rate": 6.204474444256424e-05,
      "loss": 2.0198,
      "step": 182382
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.2856645584106445,
      "learning_rate": 6.204225338671761e-05,
      "loss": 1.8408,
      "step": 182383
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.4939765930175781,
      "learning_rate": 6.203976237511175e-05,
      "loss": 2.1085,
      "step": 182384
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.229280948638916,
      "learning_rate": 6.203727140774729e-05,
      "loss": 1.9882,
      "step": 182385
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.237251877784729,
      "learning_rate": 6.203478048462454e-05,
      "loss": 1.8126,
      "step": 182386
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.2277731895446777,
      "learning_rate": 6.203228960574399e-05,
      "loss": 1.9215,
      "step": 182387
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.2993385791778564,
      "learning_rate": 6.202979877110618e-05,
      "loss": 1.9253,
      "step": 182388
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.4472953081130981,
      "learning_rate": 6.202730798071151e-05,
      "loss": 1.9311,
      "step": 182389
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.2589802742004395,
      "learning_rate": 6.202481723456041e-05,
      "loss": 2.0675,
      "step": 182390
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.3368432521820068,
      "learning_rate": 6.202232653265345e-05,
      "loss": 2.0006,
      "step": 182391
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.2663235664367676,
      "learning_rate": 6.201983587499104e-05,
      "loss": 1.9059,
      "step": 182392
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.39216947555542,
      "learning_rate": 6.201734526157354e-05,
      "loss": 2.1419,
      "step": 182393
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.305918574333191,
      "learning_rate": 6.20148546924016e-05,
      "loss": 2.1128,
      "step": 182394
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.2856303453445435,
      "learning_rate": 6.201236416747558e-05,
      "loss": 1.8561,
      "step": 182395
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.3425229787826538,
      "learning_rate": 6.200987368679587e-05,
      "loss": 1.946,
      "step": 182396
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.3272240161895752,
      "learning_rate": 6.200738325036312e-05,
      "loss": 1.848,
      "step": 182397
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.2462230920791626,
      "learning_rate": 6.200489285817758e-05,
      "loss": 2.0464,
      "step": 182398
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.4252073764801025,
      "learning_rate": 6.200240251023995e-05,
      "loss": 1.9494,
      "step": 182399
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.623697280883789,
      "learning_rate": 6.199991220655052e-05,
      "loss": 2.298,
      "step": 182400
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.551148533821106,
      "learning_rate": 6.199742194710983e-05,
      "loss": 1.9864,
      "step": 182401
    },
    {
      "epoch": 2.37,
      "grad_norm": 1.4127429723739624,
      "learning_rate": 6.199493173191821e-05,
      "loss": 2.0536,
      "step": 182402
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.3775757551193237,
      "learning_rate": 6.199244156097629e-05,
      "loss": 1.9831,
      "step": 182403
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.653396487236023,
      "learning_rate": 6.198995143428442e-05,
      "loss": 2.0024,
      "step": 182404
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.484144926071167,
      "learning_rate": 6.198746135184318e-05,
      "loss": 1.9644,
      "step": 182405
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.3637336492538452,
      "learning_rate": 6.198497131365297e-05,
      "loss": 2.0971,
      "step": 182406
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.353035569190979,
      "learning_rate": 6.198248131971423e-05,
      "loss": 2.2278,
      "step": 182407
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.5330809354782104,
      "learning_rate": 6.19799913700274e-05,
      "loss": 2.0051,
      "step": 182408
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.8874191045761108,
      "learning_rate": 6.197750146459303e-05,
      "loss": 2.1247,
      "step": 182409
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.2785942554473877,
      "learning_rate": 6.197501160341143e-05,
      "loss": 1.85,
      "step": 182410
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.2659343481063843,
      "learning_rate": 6.197252178648329e-05,
      "loss": 1.9373,
      "step": 182411
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.4783592224121094,
      "learning_rate": 6.197003201380887e-05,
      "loss": 2.1594,
      "step": 182412
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.4363973140716553,
      "learning_rate": 6.196754228538883e-05,
      "loss": 2.2079,
      "step": 182413
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.2497258186340332,
      "learning_rate": 6.19650526012234e-05,
      "loss": 1.9484,
      "step": 182414
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.294011116027832,
      "learning_rate": 6.196256296131324e-05,
      "loss": 2.0857,
      "step": 182415
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.410510778427124,
      "learning_rate": 6.196007336565865e-05,
      "loss": 1.753,
      "step": 182416
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.380090355873108,
      "learning_rate": 6.195758381426024e-05,
      "loss": 1.9583,
      "step": 182417
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.303983449935913,
      "learning_rate": 6.195509430711835e-05,
      "loss": 1.8788,
      "step": 182418
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.4461147785186768,
      "learning_rate": 6.195260484423365e-05,
      "loss": 2.072,
      "step": 182419
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.4219485521316528,
      "learning_rate": 6.195011542560628e-05,
      "loss": 2.0352,
      "step": 182420
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.2942984104156494,
      "learning_rate": 6.194762605123697e-05,
      "loss": 1.8861,
      "step": 182421
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.2741252183914185,
      "learning_rate": 6.194513672112602e-05,
      "loss": 1.9991,
      "step": 182422
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.4263935089111328,
      "learning_rate": 6.194264743527401e-05,
      "loss": 2.0696,
      "step": 182423
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.2894690036773682,
      "learning_rate": 6.194015819368132e-05,
      "loss": 1.9787,
      "step": 182424
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.231039047241211,
      "learning_rate": 6.193766899634857e-05,
      "loss": 2.007,
      "step": 182425
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.2959011793136597,
      "learning_rate": 6.193517984327597e-05,
      "loss": 2.1201,
      "step": 182426
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.4899327754974365,
      "learning_rate": 6.193269073446418e-05,
      "loss": 2.0232,
      "step": 182427
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.4916948080062866,
      "learning_rate": 6.19302016699135e-05,
      "loss": 2.0901,
      "step": 182428
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.4412949085235596,
      "learning_rate": 6.19277126496246e-05,
      "loss": 2.0685,
      "step": 182429
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.3987809419631958,
      "learning_rate": 6.192522367359775e-05,
      "loss": 2.215,
      "step": 182430
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.5115760564804077,
      "learning_rate": 6.192273474183356e-05,
      "loss": 1.9492,
      "step": 182431
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.4247905015945435,
      "learning_rate": 6.192024585433243e-05,
      "loss": 1.8824,
      "step": 182432
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.2021350860595703,
      "learning_rate": 6.191775701109482e-05,
      "loss": 1.957,
      "step": 182433
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.4025548696517944,
      "learning_rate": 6.191526821212111e-05,
      "loss": 1.7674,
      "step": 182434
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.3271681070327759,
      "learning_rate": 6.191277945741193e-05,
      "loss": 1.9155,
      "step": 182435
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.344835877418518,
      "learning_rate": 6.191029074696757e-05,
      "loss": 2.2269,
      "step": 182436
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.5345699787139893,
      "learning_rate": 6.190780208078867e-05,
      "loss": 2.1036,
      "step": 182437
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.362465739250183,
      "learning_rate": 6.190531345887562e-05,
      "loss": 1.771,
      "step": 182438
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.3769665956497192,
      "learning_rate": 6.190282488122884e-05,
      "loss": 1.9417,
      "step": 182439
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.3139901161193848,
      "learning_rate": 6.190033634784878e-05,
      "loss": 2.0688,
      "step": 182440
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.4820141792297363,
      "learning_rate": 6.189784785873597e-05,
      "loss": 2.0852,
      "step": 182441
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.2734968662261963,
      "learning_rate": 6.189535941389078e-05,
      "loss": 2.002,
      "step": 182442
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.3562085628509521,
      "learning_rate": 6.189287101331383e-05,
      "loss": 2.1031,
      "step": 182443
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.4420586824417114,
      "learning_rate": 6.18903826570055e-05,
      "loss": 1.9289,
      "step": 182444
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.287408709526062,
      "learning_rate": 6.188789434496621e-05,
      "loss": 2.1375,
      "step": 182445
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.4296562671661377,
      "learning_rate": 6.188540607719641e-05,
      "loss": 1.9736,
      "step": 182446
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.2412446737289429,
      "learning_rate": 6.188291785369667e-05,
      "loss": 1.9626,
      "step": 182447
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.301011085510254,
      "learning_rate": 6.188042967446731e-05,
      "loss": 2.1497,
      "step": 182448
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.430809736251831,
      "learning_rate": 6.187794153950895e-05,
      "loss": 2.0762,
      "step": 182449
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.403366208076477,
      "learning_rate": 6.187545344882196e-05,
      "loss": 2.1297,
      "step": 182450
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.2371840476989746,
      "learning_rate": 6.187296540240684e-05,
      "loss": 1.9614,
      "step": 182451
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.4324716329574585,
      "learning_rate": 6.187047740026396e-05,
      "loss": 2.0925,
      "step": 182452
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.3824182748794556,
      "learning_rate": 6.186798944239391e-05,
      "loss": 2.1183,
      "step": 182453
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.3262548446655273,
      "learning_rate": 6.186550152879704e-05,
      "loss": 1.97,
      "step": 182454
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.384508728981018,
      "learning_rate": 6.186301365947394e-05,
      "loss": 2.0593,
      "step": 182455
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.4120820760726929,
      "learning_rate": 6.186052583442497e-05,
      "loss": 2.0427,
      "step": 182456
    },
    {
      "epoch": 2.38,
      "grad_norm": 2.081542730331421,
      "learning_rate": 6.185803805365066e-05,
      "loss": 1.9335,
      "step": 182457
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.3224222660064697,
      "learning_rate": 6.185555031715136e-05,
      "loss": 2.0769,
      "step": 182458
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.2661277055740356,
      "learning_rate": 6.185306262492767e-05,
      "loss": 2.0002,
      "step": 182459
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.3488373756408691,
      "learning_rate": 6.18505749769799e-05,
      "loss": 2.3016,
      "step": 182460
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.4113401174545288,
      "learning_rate": 6.184808737330872e-05,
      "loss": 2.0364,
      "step": 182461
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.399245262145996,
      "learning_rate": 6.184559981391445e-05,
      "loss": 1.8621,
      "step": 182462
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.4019997119903564,
      "learning_rate": 6.184311229879759e-05,
      "loss": 2.1576,
      "step": 182463
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.229888677597046,
      "learning_rate": 6.184062482795849e-05,
      "loss": 2.3098,
      "step": 182464
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.3659883737564087,
      "learning_rate": 6.183813740139782e-05,
      "loss": 2.0259,
      "step": 182465
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.3894397020339966,
      "learning_rate": 6.183565001911585e-05,
      "loss": 1.9572,
      "step": 182466
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.3928064107894897,
      "learning_rate": 6.183316268111318e-05,
      "loss": 2.1581,
      "step": 182467
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.3587449789047241,
      "learning_rate": 6.183067538739026e-05,
      "loss": 2.1561,
      "step": 182468
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.3186875581741333,
      "learning_rate": 6.182818813794749e-05,
      "loss": 2.0653,
      "step": 182469
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.3485016822814941,
      "learning_rate": 6.18257009327853e-05,
      "loss": 2.2499,
      "step": 182470
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.3365809917449951,
      "learning_rate": 6.182321377190424e-05,
      "loss": 2.0428,
      "step": 182471
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.4080761671066284,
      "learning_rate": 6.182072665530471e-05,
      "loss": 2.1757,
      "step": 182472
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.3367255926132202,
      "learning_rate": 6.181823958298725e-05,
      "loss": 2.0887,
      "step": 182473
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.3531944751739502,
      "learning_rate": 6.18157525549523e-05,
      "loss": 2.2785,
      "step": 182474
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.3103082180023193,
      "learning_rate": 6.18132655712002e-05,
      "loss": 2.2566,
      "step": 182475
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.2201151847839355,
      "learning_rate": 6.18107786317316e-05,
      "loss": 1.7931,
      "step": 182476
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.4747940301895142,
      "learning_rate": 6.180829173654686e-05,
      "loss": 2.0989,
      "step": 182477
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.2802038192749023,
      "learning_rate": 6.180580488564638e-05,
      "loss": 2.0972,
      "step": 182478
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.2670955657958984,
      "learning_rate": 6.180331807903078e-05,
      "loss": 2.1111,
      "step": 182479
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.4170327186584473,
      "learning_rate": 6.180083131670047e-05,
      "loss": 1.8164,
      "step": 182480
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.3976680040359497,
      "learning_rate": 6.179834459865574e-05,
      "loss": 2.1046,
      "step": 182481
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.3987623453140259,
      "learning_rate": 6.17958579248973e-05,
      "loss": 2.0887,
      "step": 182482
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.3374359607696533,
      "learning_rate": 6.179337129542545e-05,
      "loss": 2.0219,
      "step": 182483
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.4155361652374268,
      "learning_rate": 6.179088471024078e-05,
      "loss": 2.0573,
      "step": 182484
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.4275215864181519,
      "learning_rate": 6.178839816934367e-05,
      "loss": 1.8998,
      "step": 182485
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.563873052597046,
      "learning_rate": 6.178591167273458e-05,
      "loss": 2.1499,
      "step": 182486
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.4677852392196655,
      "learning_rate": 6.178342522041395e-05,
      "loss": 2.0898,
      "step": 182487
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.344322919845581,
      "learning_rate": 6.178093881238232e-05,
      "loss": 2.0887,
      "step": 182488
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.283199667930603,
      "learning_rate": 6.177845244864003e-05,
      "loss": 1.9906,
      "step": 182489
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.4462428092956543,
      "learning_rate": 6.17759661291877e-05,
      "loss": 2.0331,
      "step": 182490
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.2649273872375488,
      "learning_rate": 6.177347985402576e-05,
      "loss": 2.0071,
      "step": 182491
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.25168776512146,
      "learning_rate": 6.177099362315458e-05,
      "loss": 2.1917,
      "step": 182492
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.3917946815490723,
      "learning_rate": 6.17685074365746e-05,
      "loss": 1.8814,
      "step": 182493
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.3084983825683594,
      "learning_rate": 6.176602129428642e-05,
      "loss": 2.1501,
      "step": 182494
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.327864170074463,
      "learning_rate": 6.176353519629039e-05,
      "loss": 1.8222,
      "step": 182495
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.446980595588684,
      "learning_rate": 6.176104914258706e-05,
      "loss": 2.2739,
      "step": 182496
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.470700979232788,
      "learning_rate": 6.17585631331768e-05,
      "loss": 2.0183,
      "step": 182497
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.3634823560714722,
      "learning_rate": 6.175607716806017e-05,
      "loss": 2.0131,
      "step": 182498
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.3314543962478638,
      "learning_rate": 6.175359124723761e-05,
      "loss": 1.8494,
      "step": 182499
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.2625669240951538,
      "learning_rate": 6.175110537070953e-05,
      "loss": 1.9806,
      "step": 182500
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.299064040184021,
      "learning_rate": 6.174861953847633e-05,
      "loss": 1.8867,
      "step": 182501
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.7174614667892456,
      "learning_rate": 6.174613375053867e-05,
      "loss": 2.0941,
      "step": 182502
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.2587426900863647,
      "learning_rate": 6.174364800689682e-05,
      "loss": 1.9485,
      "step": 182503
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.285705804824829,
      "learning_rate": 6.174116230755137e-05,
      "loss": 1.7911,
      "step": 182504
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.2945711612701416,
      "learning_rate": 6.173867665250277e-05,
      "loss": 2.1085,
      "step": 182505
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.4263263940811157,
      "learning_rate": 6.173619104175141e-05,
      "loss": 2.1717,
      "step": 182506
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.4403780698776245,
      "learning_rate": 6.173370547529775e-05,
      "loss": 2.1136,
      "step": 182507
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.3968168497085571,
      "learning_rate": 6.173121995314237e-05,
      "loss": 2.1526,
      "step": 182508
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.353674054145813,
      "learning_rate": 6.172873447528554e-05,
      "loss": 2.0837,
      "step": 182509
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.3083686828613281,
      "learning_rate": 6.172624904172794e-05,
      "loss": 2.1412,
      "step": 182510
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.4172176122665405,
      "learning_rate": 6.17237636524699e-05,
      "loss": 1.8818,
      "step": 182511
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.2814093828201294,
      "learning_rate": 6.172127830751194e-05,
      "loss": 2.0894,
      "step": 182512
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.3270573616027832,
      "learning_rate": 6.171879300685439e-05,
      "loss": 1.9273,
      "step": 182513
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.339257001876831,
      "learning_rate": 6.171630775049791e-05,
      "loss": 1.9989,
      "step": 182514
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.2719604969024658,
      "learning_rate": 6.17138225384428e-05,
      "loss": 2.1396,
      "step": 182515
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.4324647188186646,
      "learning_rate": 6.171133737068965e-05,
      "loss": 2.0068,
      "step": 182516
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.2868987321853638,
      "learning_rate": 6.170885224723886e-05,
      "loss": 1.9959,
      "step": 182517
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.3966494798660278,
      "learning_rate": 6.17063671680909e-05,
      "loss": 1.9201,
      "step": 182518
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.4876388311386108,
      "learning_rate": 6.170388213324616e-05,
      "loss": 2.0518,
      "step": 182519
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.3312591314315796,
      "learning_rate": 6.170139714270522e-05,
      "loss": 1.9864,
      "step": 182520
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.4095890522003174,
      "learning_rate": 6.16989121964684e-05,
      "loss": 2.1543,
      "step": 182521
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.346622347831726,
      "learning_rate": 6.169642729453636e-05,
      "loss": 1.8604,
      "step": 182522
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.4521578550338745,
      "learning_rate": 6.169394243690943e-05,
      "loss": 1.8207,
      "step": 182523
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.2873972654342651,
      "learning_rate": 6.169145762358813e-05,
      "loss": 2.0981,
      "step": 182524
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.3267933130264282,
      "learning_rate": 6.168897285457275e-05,
      "loss": 1.8616,
      "step": 182525
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.364113688468933,
      "learning_rate": 6.1686488129864e-05,
      "loss": 1.9338,
      "step": 182526
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.3109710216522217,
      "learning_rate": 6.168400344946217e-05,
      "loss": 2.0861,
      "step": 182527
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.38394033908844,
      "learning_rate": 6.168151881336783e-05,
      "loss": 1.8712,
      "step": 182528
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.6571968793869019,
      "learning_rate": 6.16790342215814e-05,
      "loss": 2.1101,
      "step": 182529
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.4866636991500854,
      "learning_rate": 6.167654967410334e-05,
      "loss": 2.0023,
      "step": 182530
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.562992811203003,
      "learning_rate": 6.167406517093404e-05,
      "loss": 1.8859,
      "step": 182531
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.3235524892807007,
      "learning_rate": 6.16715807120741e-05,
      "loss": 1.9242,
      "step": 182532
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.582177996635437,
      "learning_rate": 6.166909629752381e-05,
      "loss": 1.8965,
      "step": 182533
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.4642643928527832,
      "learning_rate": 6.166661192728384e-05,
      "loss": 2.0088,
      "step": 182534
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.3483296632766724,
      "learning_rate": 6.166412760135454e-05,
      "loss": 2.1895,
      "step": 182535
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.3338274955749512,
      "learning_rate": 6.166164331973638e-05,
      "loss": 1.9462,
      "step": 182536
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.505715012550354,
      "learning_rate": 6.165915908242975e-05,
      "loss": 1.8678,
      "step": 182537
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.2637447118759155,
      "learning_rate": 6.165667488943524e-05,
      "loss": 2.29,
      "step": 182538
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.3225675821304321,
      "learning_rate": 6.165419074075318e-05,
      "loss": 1.9941,
      "step": 182539
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.3613612651824951,
      "learning_rate": 6.165170663638418e-05,
      "loss": 2.0241,
      "step": 182540
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.4610685110092163,
      "learning_rate": 6.164922257632866e-05,
      "loss": 1.991,
      "step": 182541
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.6787736415863037,
      "learning_rate": 6.164673856058702e-05,
      "loss": 2.1872,
      "step": 182542
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.3194398880004883,
      "learning_rate": 6.164425458915969e-05,
      "loss": 1.8173,
      "step": 182543
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.4911152124404907,
      "learning_rate": 6.164177066204721e-05,
      "loss": 2.1354,
      "step": 182544
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.4336186647415161,
      "learning_rate": 6.163928677925e-05,
      "loss": 2.1066,
      "step": 182545
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.5042383670806885,
      "learning_rate": 6.163680294076862e-05,
      "loss": 2.0209,
      "step": 182546
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.517168641090393,
      "learning_rate": 6.163431914660347e-05,
      "loss": 1.9945,
      "step": 182547
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.2454440593719482,
      "learning_rate": 6.163183539675498e-05,
      "loss": 1.9794,
      "step": 182548
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.1901681423187256,
      "learning_rate": 6.162935169122354e-05,
      "loss": 1.9365,
      "step": 182549
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.4423097372055054,
      "learning_rate": 6.162686803000982e-05,
      "loss": 2.185,
      "step": 182550
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.3216030597686768,
      "learning_rate": 6.162438441311407e-05,
      "loss": 1.9249,
      "step": 182551
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.2747347354888916,
      "learning_rate": 6.16219008405369e-05,
      "loss": 2.0954,
      "step": 182552
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.3641763925552368,
      "learning_rate": 6.161941731227873e-05,
      "loss": 2.0397,
      "step": 182553
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.4322534799575806,
      "learning_rate": 6.161693382834003e-05,
      "loss": 2.1423,
      "step": 182554
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.283927321434021,
      "learning_rate": 6.161445038872115e-05,
      "loss": 2.1028,
      "step": 182555
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.3592898845672607,
      "learning_rate": 6.161196699342272e-05,
      "loss": 1.8985,
      "step": 182556
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.3976020812988281,
      "learning_rate": 6.160948364244506e-05,
      "loss": 1.9828,
      "step": 182557
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.2508571147918701,
      "learning_rate": 6.160700033578875e-05,
      "loss": 1.8734,
      "step": 182558
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.4691678285598755,
      "learning_rate": 6.160451707345422e-05,
      "loss": 2.0653,
      "step": 182559
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.363538146018982,
      "learning_rate": 6.160203385544183e-05,
      "loss": 2.1716,
      "step": 182560
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.4200291633605957,
      "learning_rate": 6.15995506817522e-05,
      "loss": 2.0552,
      "step": 182561
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.3133565187454224,
      "learning_rate": 6.159706755238571e-05,
      "loss": 2.0413,
      "step": 182562
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.3140618801116943,
      "learning_rate": 6.159458446734276e-05,
      "loss": 1.8851,
      "step": 182563
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.461376667022705,
      "learning_rate": 6.159210142662392e-05,
      "loss": 1.9832,
      "step": 182564
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.3242692947387695,
      "learning_rate": 6.158961843022956e-05,
      "loss": 2.1141,
      "step": 182565
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.3497505187988281,
      "learning_rate": 6.158713547816024e-05,
      "loss": 2.1094,
      "step": 182566
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.3697656393051147,
      "learning_rate": 6.15846525704164e-05,
      "loss": 2.0194,
      "step": 182567
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.3251124620437622,
      "learning_rate": 6.158216970699841e-05,
      "loss": 2.1162,
      "step": 182568
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.4299683570861816,
      "learning_rate": 6.157968688790685e-05,
      "loss": 2.3444,
      "step": 182569
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.2755006551742554,
      "learning_rate": 6.157720411314212e-05,
      "loss": 1.8889,
      "step": 182570
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.409451961517334,
      "learning_rate": 6.15747213827046e-05,
      "loss": 1.9783,
      "step": 182571
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.3642634153366089,
      "learning_rate": 6.157223869659495e-05,
      "loss": 1.8946,
      "step": 182572
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.293829083442688,
      "learning_rate": 6.156975605481353e-05,
      "loss": 2.073,
      "step": 182573
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.3584635257720947,
      "learning_rate": 6.156727345736068e-05,
      "loss": 1.9723,
      "step": 182574
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.4757908582687378,
      "learning_rate": 6.156479090423708e-05,
      "loss": 1.9548,
      "step": 182575
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.3014836311340332,
      "learning_rate": 6.156230839544307e-05,
      "loss": 2.0462,
      "step": 182576
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.3760441541671753,
      "learning_rate": 6.155982593097904e-05,
      "loss": 1.8804,
      "step": 182577
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.4376184940338135,
      "learning_rate": 6.155734351084563e-05,
      "loss": 1.9156,
      "step": 182578
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.4286739826202393,
      "learning_rate": 6.155486113504321e-05,
      "loss": 2.0549,
      "step": 182579
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.3837416172027588,
      "learning_rate": 6.155237880357216e-05,
      "loss": 2.1189,
      "step": 182580
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.3378359079360962,
      "learning_rate": 6.154989651643309e-05,
      "loss": 1.9819,
      "step": 182581
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.5164680480957031,
      "learning_rate": 6.154741427362636e-05,
      "loss": 1.8186,
      "step": 182582
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.3359158039093018,
      "learning_rate": 6.15449320751525e-05,
      "loss": 1.9027,
      "step": 182583
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.3873249292373657,
      "learning_rate": 6.154244992101197e-05,
      "loss": 2.0123,
      "step": 182584
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.4400641918182373,
      "learning_rate": 6.153996781120518e-05,
      "loss": 1.9047,
      "step": 182585
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.3080711364746094,
      "learning_rate": 6.153748574573253e-05,
      "loss": 1.8453,
      "step": 182586
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.399898648262024,
      "learning_rate": 6.153500372459465e-05,
      "loss": 2.1053,
      "step": 182587
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.294861912727356,
      "learning_rate": 6.153252174779183e-05,
      "loss": 2.2621,
      "step": 182588
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.4436849355697632,
      "learning_rate": 6.153003981532472e-05,
      "loss": 1.8685,
      "step": 182589
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.313395380973816,
      "learning_rate": 6.152755792719364e-05,
      "loss": 2.0168,
      "step": 182590
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.3343936204910278,
      "learning_rate": 6.152507608339912e-05,
      "loss": 2.0709,
      "step": 182591
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.3366621732711792,
      "learning_rate": 6.152259428394149e-05,
      "loss": 1.9048,
      "step": 182592
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.278023600578308,
      "learning_rate": 6.15201125288214e-05,
      "loss": 2.1687,
      "step": 182593
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.2839654684066772,
      "learning_rate": 6.151763081803914e-05,
      "loss": 2.1313,
      "step": 182594
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.2952402830123901,
      "learning_rate": 6.151514915159533e-05,
      "loss": 2.1931,
      "step": 182595
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.4074175357818604,
      "learning_rate": 6.151266752949037e-05,
      "loss": 1.8578,
      "step": 182596
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.4261419773101807,
      "learning_rate": 6.151018595172469e-05,
      "loss": 1.9574,
      "step": 182597
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.290608525276184,
      "learning_rate": 6.150770441829868e-05,
      "loss": 2.0147,
      "step": 182598
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.4047913551330566,
      "learning_rate": 6.1505222929213e-05,
      "loss": 2.0239,
      "step": 182599
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.58104407787323,
      "learning_rate": 6.150274148446792e-05,
      "loss": 2.0651,
      "step": 182600
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.2607938051223755,
      "learning_rate": 6.150026008406403e-05,
      "loss": 1.9119,
      "step": 182601
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.3143128156661987,
      "learning_rate": 6.149777872800178e-05,
      "loss": 2.1959,
      "step": 182602
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.370776653289795,
      "learning_rate": 6.149529741628155e-05,
      "loss": 1.8082,
      "step": 182603
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.2237083911895752,
      "learning_rate": 6.149281614890382e-05,
      "loss": 1.8625,
      "step": 182604
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.3740897178649902,
      "learning_rate": 6.149033492586913e-05,
      "loss": 2.0587,
      "step": 182605
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.4585858583450317,
      "learning_rate": 6.148785374717779e-05,
      "loss": 1.9874,
      "step": 182606
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.3814637660980225,
      "learning_rate": 6.148537261283048e-05,
      "loss": 2.1381,
      "step": 182607
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.1946666240692139,
      "learning_rate": 6.148289152282753e-05,
      "loss": 1.7624,
      "step": 182608
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.293989658355713,
      "learning_rate": 6.148041047716937e-05,
      "loss": 1.9349,
      "step": 182609
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.3183434009552002,
      "learning_rate": 6.147792947585648e-05,
      "loss": 1.8946,
      "step": 182610
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.3883026838302612,
      "learning_rate": 6.147544851888942e-05,
      "loss": 2.1326,
      "step": 182611
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.4378435611724854,
      "learning_rate": 6.147296760626848e-05,
      "loss": 1.9489,
      "step": 182612
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.4758566617965698,
      "learning_rate": 6.147048673799426e-05,
      "loss": 2.0613,
      "step": 182613
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.4012800455093384,
      "learning_rate": 6.146800591406724e-05,
      "loss": 2.1568,
      "step": 182614
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.483679175376892,
      "learning_rate": 6.146552513448778e-05,
      "loss": 1.9168,
      "step": 182615
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.3285090923309326,
      "learning_rate": 6.146304439925634e-05,
      "loss": 1.9399,
      "step": 182616
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.360885500907898,
      "learning_rate": 6.146056370837346e-05,
      "loss": 2.0042,
      "step": 182617
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.4136894941329956,
      "learning_rate": 6.145808306183952e-05,
      "loss": 1.6797,
      "step": 182618
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.4500830173492432,
      "learning_rate": 6.145560245965508e-05,
      "loss": 2.3009,
      "step": 182619
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.3460891246795654,
      "learning_rate": 6.145312190182058e-05,
      "loss": 1.7879,
      "step": 182620
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.3566828966140747,
      "learning_rate": 6.145064138833643e-05,
      "loss": 1.8808,
      "step": 182621
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.31247878074646,
      "learning_rate": 6.144816091920301e-05,
      "loss": 2.0549,
      "step": 182622
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.2833926677703857,
      "learning_rate": 6.144568049442096e-05,
      "loss": 2.221,
      "step": 182623
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.4841258525848389,
      "learning_rate": 6.144320011399062e-05,
      "loss": 1.947,
      "step": 182624
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.4043539762496948,
      "learning_rate": 6.144071977791253e-05,
      "loss": 2.1787,
      "step": 182625
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.2997627258300781,
      "learning_rate": 6.143823948618712e-05,
      "loss": 1.8811,
      "step": 182626
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.4626212120056152,
      "learning_rate": 6.143575923881487e-05,
      "loss": 2.0792,
      "step": 182627
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.257608413696289,
      "learning_rate": 6.143327903579612e-05,
      "loss": 2.1941,
      "step": 182628
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.358162522315979,
      "learning_rate": 6.14307988771315e-05,
      "loss": 1.8962,
      "step": 182629
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.2794698476791382,
      "learning_rate": 6.142831876282131e-05,
      "loss": 1.9859,
      "step": 182630
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.3529260158538818,
      "learning_rate": 6.142583869286622e-05,
      "loss": 2.103,
      "step": 182631
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.4806607961654663,
      "learning_rate": 6.142335866726647e-05,
      "loss": 2.0174,
      "step": 182632
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.3991934061050415,
      "learning_rate": 6.142087868602276e-05,
      "loss": 1.9289,
      "step": 182633
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.3360116481781006,
      "learning_rate": 6.141839874913529e-05,
      "loss": 2.172,
      "step": 182634
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.3360419273376465,
      "learning_rate": 6.141591885660469e-05,
      "loss": 2.2252,
      "step": 182635
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.3784465789794922,
      "learning_rate": 6.141343900843128e-05,
      "loss": 1.775,
      "step": 182636
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.406978964805603,
      "learning_rate": 6.141095920461573e-05,
      "loss": 2.0069,
      "step": 182637
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.3827811479568481,
      "learning_rate": 6.140847944515832e-05,
      "loss": 1.9839,
      "step": 182638
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.4983322620391846,
      "learning_rate": 6.14059997300597e-05,
      "loss": 2.0176,
      "step": 182639
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.4418519735336304,
      "learning_rate": 6.140352005932005e-05,
      "loss": 2.0673,
      "step": 182640
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.3729454278945923,
      "learning_rate": 6.14010404329401e-05,
      "loss": 1.8526,
      "step": 182641
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.3530981540679932,
      "learning_rate": 6.139856085092009e-05,
      "loss": 2.1662,
      "step": 182642
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.2545579671859741,
      "learning_rate": 6.13960813132607e-05,
      "loss": 2.039,
      "step": 182643
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.2431706190109253,
      "learning_rate": 6.139360181996218e-05,
      "loss": 1.8709,
      "step": 182644
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.495648741722107,
      "learning_rate": 6.139112237102519e-05,
      "loss": 1.9552,
      "step": 182645
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.4816707372665405,
      "learning_rate": 6.138864296645006e-05,
      "loss": 1.8375,
      "step": 182646
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.3504481315612793,
      "learning_rate": 6.138616360623732e-05,
      "loss": 2.0307,
      "step": 182647
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.3167359828948975,
      "learning_rate": 6.13836842903873e-05,
      "loss": 1.9953,
      "step": 182648
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.3670954704284668,
      "learning_rate": 6.138120501890066e-05,
      "loss": 2.1768,
      "step": 182649
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.2395392656326294,
      "learning_rate": 6.137872579177767e-05,
      "loss": 1.8161,
      "step": 182650
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.6406993865966797,
      "learning_rate": 6.137624660901895e-05,
      "loss": 2.016,
      "step": 182651
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.2265379428863525,
      "learning_rate": 6.137376747062486e-05,
      "loss": 1.8926,
      "step": 182652
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.3623803853988647,
      "learning_rate": 6.137128837659594e-05,
      "loss": 1.9322,
      "step": 182653
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.5781704187393188,
      "learning_rate": 6.136880932693248e-05,
      "loss": 1.9593,
      "step": 182654
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.2503376007080078,
      "learning_rate": 6.136633032163518e-05,
      "loss": 1.9404,
      "step": 182655
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.3907229900360107,
      "learning_rate": 6.136385136070428e-05,
      "loss": 1.9531,
      "step": 182656
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.4305238723754883,
      "learning_rate": 6.136137244414047e-05,
      "loss": 2.0549,
      "step": 182657
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.4538685083389282,
      "learning_rate": 6.135889357194402e-05,
      "loss": 1.8652,
      "step": 182658
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.5300703048706055,
      "learning_rate": 6.135641474411542e-05,
      "loss": 2.1829,
      "step": 182659
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.3510687351226807,
      "learning_rate": 6.135393596065523e-05,
      "loss": 1.9527,
      "step": 182660
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.4061223268508911,
      "learning_rate": 6.135145722156385e-05,
      "loss": 2.0443,
      "step": 182661
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.3027982711791992,
      "learning_rate": 6.134897852684168e-05,
      "loss": 2.2431,
      "step": 182662
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.3247634172439575,
      "learning_rate": 6.134649987648927e-05,
      "loss": 1.9577,
      "step": 182663
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.3232533931732178,
      "learning_rate": 6.134402127050711e-05,
      "loss": 2.0005,
      "step": 182664
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.242244005203247,
      "learning_rate": 6.134154270889549e-05,
      "loss": 2.0725,
      "step": 182665
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.3359043598175049,
      "learning_rate": 6.133906419165505e-05,
      "loss": 2.0582,
      "step": 182666
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.3581767082214355,
      "learning_rate": 6.133658571878612e-05,
      "loss": 2.005,
      "step": 182667
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.208077073097229,
      "learning_rate": 6.133410729028932e-05,
      "loss": 1.7776,
      "step": 182668
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.3148560523986816,
      "learning_rate": 6.133162890616499e-05,
      "loss": 1.9103,
      "step": 182669
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.2670522928237915,
      "learning_rate": 6.132915056641363e-05,
      "loss": 2.0234,
      "step": 182670
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.3722227811813354,
      "learning_rate": 6.132667227103558e-05,
      "loss": 1.828,
      "step": 182671
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.3974446058273315,
      "learning_rate": 6.13241940200315e-05,
      "loss": 1.9872,
      "step": 182672
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.4567394256591797,
      "learning_rate": 6.13217158134017e-05,
      "loss": 2.4731,
      "step": 182673
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.3276771306991577,
      "learning_rate": 6.131923765114675e-05,
      "loss": 2.113,
      "step": 182674
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.3607432842254639,
      "learning_rate": 6.131675953326708e-05,
      "loss": 2.1515,
      "step": 182675
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.3995698690414429,
      "learning_rate": 6.131428145976313e-05,
      "loss": 2.1553,
      "step": 182676
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.519595742225647,
      "learning_rate": 6.131180343063526e-05,
      "loss": 2.0395,
      "step": 182677
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.4047452211380005,
      "learning_rate": 6.130932544588414e-05,
      "loss": 2.0338,
      "step": 182678
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.4949008226394653,
      "learning_rate": 6.130684750551001e-05,
      "loss": 2.2473,
      "step": 182679
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.3119733333587646,
      "learning_rate": 6.130436960951352e-05,
      "loss": 2.0205,
      "step": 182680
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.2424756288528442,
      "learning_rate": 6.130189175789511e-05,
      "loss": 1.9674,
      "step": 182681
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.3860952854156494,
      "learning_rate": 6.129941395065512e-05,
      "loss": 2.0554,
      "step": 182682
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.411623239517212,
      "learning_rate": 6.129693618779403e-05,
      "loss": 2.2723,
      "step": 182683
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.354628562927246,
      "learning_rate": 6.129445846931241e-05,
      "loss": 1.8136,
      "step": 182684
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.4067217111587524,
      "learning_rate": 6.12919807952106e-05,
      "loss": 2.0795,
      "step": 182685
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.2668668031692505,
      "learning_rate": 6.128950316548918e-05,
      "loss": 2.1246,
      "step": 182686
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.284817099571228,
      "learning_rate": 6.128702558014853e-05,
      "loss": 2.0372,
      "step": 182687
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.3228614330291748,
      "learning_rate": 6.128454803918914e-05,
      "loss": 2.116,
      "step": 182688
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.3304195404052734,
      "learning_rate": 6.128207054261137e-05,
      "loss": 2.0741,
      "step": 182689
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.52936851978302,
      "learning_rate": 6.127959309041588e-05,
      "loss": 2.0843,
      "step": 182690
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.291073203086853,
      "learning_rate": 6.127711568260292e-05,
      "loss": 2.1538,
      "step": 182691
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.3638707399368286,
      "learning_rate": 6.127463831917312e-05,
      "loss": 1.9879,
      "step": 182692
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.2443948984146118,
      "learning_rate": 6.127216100012688e-05,
      "loss": 1.9152,
      "step": 182693
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.303591251373291,
      "learning_rate": 6.126968372546464e-05,
      "loss": 2.0432,
      "step": 182694
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.2771550416946411,
      "learning_rate": 6.126720649518684e-05,
      "loss": 1.9921,
      "step": 182695
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.4733333587646484,
      "learning_rate": 6.126472930929399e-05,
      "loss": 2.0148,
      "step": 182696
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.3029190301895142,
      "learning_rate": 6.126225216778651e-05,
      "loss": 2.134,
      "step": 182697
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.4430307149887085,
      "learning_rate": 6.125977507066492e-05,
      "loss": 1.8306,
      "step": 182698
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.3561818599700928,
      "learning_rate": 6.125729801792959e-05,
      "loss": 2.1241,
      "step": 182699
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.4154572486877441,
      "learning_rate": 6.12548210095812e-05,
      "loss": 2.2281,
      "step": 182700
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.3920621871948242,
      "learning_rate": 6.125234404561985e-05,
      "loss": 2.0675,
      "step": 182701
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.3589552640914917,
      "learning_rate": 6.124986712604631e-05,
      "loss": 1.9831,
      "step": 182702
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.4689608812332153,
      "learning_rate": 6.124739025086082e-05,
      "loss": 2.105,
      "step": 182703
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.4055137634277344,
      "learning_rate": 6.124491342006406e-05,
      "loss": 1.9497,
      "step": 182704
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.3867192268371582,
      "learning_rate": 6.124243663365625e-05,
      "loss": 2.167,
      "step": 182705
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.298518180847168,
      "learning_rate": 6.123995989163818e-05,
      "loss": 1.9174,
      "step": 182706
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.610588550567627,
      "learning_rate": 6.123748319400993e-05,
      "loss": 2.0279,
      "step": 182707
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.3435660600662231,
      "learning_rate": 6.123500654077223e-05,
      "loss": 2.1521,
      "step": 182708
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.3391225337982178,
      "learning_rate": 6.123252993192537e-05,
      "loss": 1.901,
      "step": 182709
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.340300440788269,
      "learning_rate": 6.123005336746996e-05,
      "loss": 2.2265,
      "step": 182710
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.196722149848938,
      "learning_rate": 6.12275768474063e-05,
      "loss": 2.0766,
      "step": 182711
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.387679100036621,
      "learning_rate": 6.12251003717351e-05,
      "loss": 2.1611,
      "step": 182712
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.3995264768600464,
      "learning_rate": 6.122262394045653e-05,
      "loss": 1.8683,
      "step": 182713
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.337772011756897,
      "learning_rate": 6.122014755357124e-05,
      "loss": 1.9352,
      "step": 182714
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.3358256816864014,
      "learning_rate": 6.121767121107955e-05,
      "loss": 1.9953,
      "step": 182715
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.4479421377182007,
      "learning_rate": 6.12151949129821e-05,
      "loss": 2.1471,
      "step": 182716
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.4200505018234253,
      "learning_rate": 6.121271865927915e-05,
      "loss": 2.0401,
      "step": 182717
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.3924728631973267,
      "learning_rate": 6.121024244997144e-05,
      "loss": 1.9024,
      "step": 182718
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.2776281833648682,
      "learning_rate": 6.12077662850591e-05,
      "loss": 1.9972,
      "step": 182719
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.3175303936004639,
      "learning_rate": 6.12052901645428e-05,
      "loss": 1.9604,
      "step": 182720
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.3677037954330444,
      "learning_rate": 6.120281408842287e-05,
      "loss": 2.1023,
      "step": 182721
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.3267052173614502,
      "learning_rate": 6.120033805669995e-05,
      "loss": 2.123,
      "step": 182722
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.386142611503601,
      "learning_rate": 6.119786206937428e-05,
      "loss": 2.0366,
      "step": 182723
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.4111380577087402,
      "learning_rate": 6.119538612644664e-05,
      "loss": 1.8399,
      "step": 182724
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.4184900522232056,
      "learning_rate": 6.119291022791707e-05,
      "loss": 2.1924,
      "step": 182725
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.3361438512802124,
      "learning_rate": 6.119043437378636e-05,
      "loss": 1.9072,
      "step": 182726
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.3092831373214722,
      "learning_rate": 6.118795856405476e-05,
      "loss": 2.0916,
      "step": 182727
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.237385630607605,
      "learning_rate": 6.118548279872292e-05,
      "loss": 1.9903,
      "step": 182728
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.218687891960144,
      "learning_rate": 6.118300707779113e-05,
      "loss": 1.9507,
      "step": 182729
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.3406691551208496,
      "learning_rate": 6.118053140125998e-05,
      "loss": 2.0332,
      "step": 182730
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.3376686573028564,
      "learning_rate": 6.117805576912992e-05,
      "loss": 2.065,
      "step": 182731
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.2005563974380493,
      "learning_rate": 6.117558018140134e-05,
      "loss": 1.9241,
      "step": 182732
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.393376111984253,
      "learning_rate": 6.117310463807465e-05,
      "loss": 1.9832,
      "step": 182733
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.4196836948394775,
      "learning_rate": 6.117062913915046e-05,
      "loss": 2.0443,
      "step": 182734
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.3520883321762085,
      "learning_rate": 6.116815368462909e-05,
      "loss": 2.0674,
      "step": 182735
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.3772882223129272,
      "learning_rate": 6.116567827451114e-05,
      "loss": 1.8993,
      "step": 182736
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.4291532039642334,
      "learning_rate": 6.116320290879701e-05,
      "loss": 1.8774,
      "step": 182737
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.3038588762283325,
      "learning_rate": 6.116072758748714e-05,
      "loss": 2.1464,
      "step": 182738
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.3691486120224,
      "learning_rate": 6.115825231058193e-05,
      "loss": 2.104,
      "step": 182739
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.4834129810333252,
      "learning_rate": 6.1155777078082e-05,
      "loss": 2.0075,
      "step": 182740
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.3165252208709717,
      "learning_rate": 6.115330188998762e-05,
      "loss": 1.8055,
      "step": 182741
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.4774008989334106,
      "learning_rate": 6.115082674629941e-05,
      "loss": 1.8921,
      "step": 182742
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.4650176763534546,
      "learning_rate": 6.114835164701781e-05,
      "loss": 2.0287,
      "step": 182743
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.3162516355514526,
      "learning_rate": 6.114587659214312e-05,
      "loss": 2.0581,
      "step": 182744
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.4258290529251099,
      "learning_rate": 6.114340158167601e-05,
      "loss": 1.8215,
      "step": 182745
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.326409101486206,
      "learning_rate": 6.114092661561689e-05,
      "loss": 1.9569,
      "step": 182746
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.3753225803375244,
      "learning_rate": 6.113845169396607e-05,
      "loss": 1.9063,
      "step": 182747
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.2364321947097778,
      "learning_rate": 6.113597681672419e-05,
      "loss": 1.8906,
      "step": 182748
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.3158490657806396,
      "learning_rate": 6.113350198389167e-05,
      "loss": 1.985,
      "step": 182749
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.3238438367843628,
      "learning_rate": 6.113102719546884e-05,
      "loss": 2.1651,
      "step": 182750
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.3745075464248657,
      "learning_rate": 6.112855245145636e-05,
      "loss": 1.9718,
      "step": 182751
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.267372965812683,
      "learning_rate": 6.112607775185458e-05,
      "loss": 2.0065,
      "step": 182752
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.6714696884155273,
      "learning_rate": 6.112360309666388e-05,
      "loss": 2.0011,
      "step": 182753
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.323177695274353,
      "learning_rate": 6.11211284858849e-05,
      "loss": 2.0116,
      "step": 182754
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.322482705116272,
      "learning_rate": 6.111865391951801e-05,
      "loss": 2.1387,
      "step": 182755
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.507168173789978,
      "learning_rate": 6.11161793975636e-05,
      "loss": 2.0702,
      "step": 182756
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.3117016553878784,
      "learning_rate": 6.111370492002227e-05,
      "loss": 2.1276,
      "step": 182757
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.325466513633728,
      "learning_rate": 6.111123048689433e-05,
      "loss": 1.9897,
      "step": 182758
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.3375186920166016,
      "learning_rate": 6.110875609818042e-05,
      "loss": 2.0373,
      "step": 182759
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.4104772806167603,
      "learning_rate": 6.11062817538809e-05,
      "loss": 1.9129,
      "step": 182760
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.3993889093399048,
      "learning_rate": 6.110380745399622e-05,
      "loss": 2.1395,
      "step": 182761
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.3383543491363525,
      "learning_rate": 6.110133319852676e-05,
      "loss": 1.9604,
      "step": 182762
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.3228977918624878,
      "learning_rate": 6.109885898747317e-05,
      "loss": 2.0729,
      "step": 182763
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.4254618883132935,
      "learning_rate": 6.109638482083573e-05,
      "loss": 2.1391,
      "step": 182764
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.2697045803070068,
      "learning_rate": 6.109391069861504e-05,
      "loss": 2.1605,
      "step": 182765
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.4080605506896973,
      "learning_rate": 6.109143662081143e-05,
      "loss": 2.0099,
      "step": 182766
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.579167127609253,
      "learning_rate": 6.108896258742562e-05,
      "loss": 1.9689,
      "step": 182767
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.384639024734497,
      "learning_rate": 6.108648859845769e-05,
      "loss": 1.9818,
      "step": 182768
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.2747265100479126,
      "learning_rate": 6.10840146539084e-05,
      "loss": 1.9685,
      "step": 182769
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.5481464862823486,
      "learning_rate": 6.108154075377798e-05,
      "loss": 1.8922,
      "step": 182770
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.2726610898971558,
      "learning_rate": 6.107906689806714e-05,
      "loss": 1.9235,
      "step": 182771
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.2252076864242554,
      "learning_rate": 6.107659308677612e-05,
      "loss": 1.9048,
      "step": 182772
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.2519174814224243,
      "learning_rate": 6.107411931990562e-05,
      "loss": 1.893,
      "step": 182773
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.3535596132278442,
      "learning_rate": 6.10716455974558e-05,
      "loss": 1.9661,
      "step": 182774
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.5662680864334106,
      "learning_rate": 6.106917191942732e-05,
      "loss": 1.9694,
      "step": 182775
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.4238855838775635,
      "learning_rate": 6.106669828582054e-05,
      "loss": 2.1039,
      "step": 182776
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.4183567762374878,
      "learning_rate": 6.106422469663605e-05,
      "loss": 1.9844,
      "step": 182777
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.3687676191329956,
      "learning_rate": 6.106175115187414e-05,
      "loss": 1.9508,
      "step": 182778
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.1709506511688232,
      "learning_rate": 6.105927765153552e-05,
      "loss": 2.229,
      "step": 182779
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.5748904943466187,
      "learning_rate": 6.105680419562034e-05,
      "loss": 1.7834,
      "step": 182780
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.607847809791565,
      "learning_rate": 6.105433078412931e-05,
      "loss": 1.9992,
      "step": 182781
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.2544660568237305,
      "learning_rate": 6.105185741706269e-05,
      "loss": 1.9033,
      "step": 182782
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.354036808013916,
      "learning_rate": 6.10493840944211e-05,
      "loss": 1.8733,
      "step": 182783
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.3624242544174194,
      "learning_rate": 6.10469108162049e-05,
      "loss": 1.9458,
      "step": 182784
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.3247212171554565,
      "learning_rate": 6.104443758241473e-05,
      "loss": 1.9855,
      "step": 182785
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.3865222930908203,
      "learning_rate": 6.104196439305073e-05,
      "loss": 1.9678,
      "step": 182786
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.6713171005249023,
      "learning_rate": 6.103949124811365e-05,
      "loss": 1.9554,
      "step": 182787
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.8203402757644653,
      "learning_rate": 6.103701814760373e-05,
      "loss": 1.9454,
      "step": 182788
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.3867826461791992,
      "learning_rate": 6.103454509152165e-05,
      "loss": 2.2393,
      "step": 182789
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.3994405269622803,
      "learning_rate": 6.1032072079867656e-05,
      "loss": 1.9531,
      "step": 182790
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.6429880857467651,
      "learning_rate": 6.1029599112642493e-05,
      "loss": 1.7752,
      "step": 182791
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.6679853200912476,
      "learning_rate": 6.1027126189846255e-05,
      "loss": 2.0106,
      "step": 182792
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.3747382164001465,
      "learning_rate": 6.102465331147967e-05,
      "loss": 1.8394,
      "step": 182793
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.2331624031066895,
      "learning_rate": 6.102218047754306e-05,
      "loss": 1.9499,
      "step": 182794
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.3021247386932373,
      "learning_rate": 6.101970768803697e-05,
      "loss": 2.0151,
      "step": 182795
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.4103610515594482,
      "learning_rate": 6.101723494296177e-05,
      "loss": 2.1043,
      "step": 182796
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.3770779371261597,
      "learning_rate": 6.101476224231816e-05,
      "loss": 1.9633,
      "step": 182797
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.3746000528335571,
      "learning_rate": 6.1012289586106245e-05,
      "loss": 1.8614,
      "step": 182798
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.543290138244629,
      "learning_rate": 6.100981697432672e-05,
      "loss": 1.8386,
      "step": 182799
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.4381017684936523,
      "learning_rate": 6.100734440697992e-05,
      "loss": 1.9998,
      "step": 182800
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.452726125717163,
      "learning_rate": 6.100487188406641e-05,
      "loss": 1.8416,
      "step": 182801
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.3429551124572754,
      "learning_rate": 6.1002399405586556e-05,
      "loss": 1.978,
      "step": 182802
    },
    {
      "epoch": 2.38,
      "grad_norm": 2.083173990249634,
      "learning_rate": 6.099992697154103e-05,
      "loss": 1.9329,
      "step": 182803
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.550098180770874,
      "learning_rate": 6.099745458192995e-05,
      "loss": 1.9305,
      "step": 182804
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.4196022748947144,
      "learning_rate": 6.099498223675406e-05,
      "loss": 1.9963,
      "step": 182805
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.3483587503433228,
      "learning_rate": 6.099250993601364e-05,
      "loss": 2.153,
      "step": 182806
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.6020828485488892,
      "learning_rate": 6.099003767970926e-05,
      "loss": 1.9353,
      "step": 182807
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.4890859127044678,
      "learning_rate": 6.0987565467841304e-05,
      "loss": 1.9481,
      "step": 182808
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.3732296228408813,
      "learning_rate": 6.098509330041041e-05,
      "loss": 2.118,
      "step": 182809
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.2377293109893799,
      "learning_rate": 6.0982621177416755e-05,
      "loss": 1.9475,
      "step": 182810
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.5247057676315308,
      "learning_rate": 6.0980149098861e-05,
      "loss": 1.889,
      "step": 182811
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.4811829328536987,
      "learning_rate": 6.097767706474349e-05,
      "loss": 2.0055,
      "step": 182812
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.2623597383499146,
      "learning_rate": 6.0975205075064804e-05,
      "loss": 2.0071,
      "step": 182813
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.399570107460022,
      "learning_rate": 6.097273312982527e-05,
      "loss": 2.1257,
      "step": 182814
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.2503538131713867,
      "learning_rate": 6.097026122902559e-05,
      "loss": 2.0272,
      "step": 182815
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.5421563386917114,
      "learning_rate": 6.096778937266585e-05,
      "loss": 1.8805,
      "step": 182816
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.5038176774978638,
      "learning_rate": 6.096531756074684e-05,
      "loss": 2.181,
      "step": 182817
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.3773083686828613,
      "learning_rate": 6.096284579326878e-05,
      "loss": 2.0738,
      "step": 182818
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.3253103494644165,
      "learning_rate": 6.09603740702323e-05,
      "loss": 1.9088,
      "step": 182819
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.3215157985687256,
      "learning_rate": 6.095790239163777e-05,
      "loss": 1.9841,
      "step": 182820
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.3051680326461792,
      "learning_rate": 6.0955430757485726e-05,
      "loss": 2.1152,
      "step": 182821
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.2653348445892334,
      "learning_rate": 6.095295916777656e-05,
      "loss": 1.9802,
      "step": 182822
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.4031132459640503,
      "learning_rate": 6.095048762251078e-05,
      "loss": 1.9164,
      "step": 182823
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.328419804573059,
      "learning_rate": 6.094801612168875e-05,
      "loss": 2.0226,
      "step": 182824
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.2440319061279297,
      "learning_rate": 6.094554466531106e-05,
      "loss": 1.9056,
      "step": 182825
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.3243955373764038,
      "learning_rate": 6.094307325337803e-05,
      "loss": 1.9301,
      "step": 182826
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.6254018545150757,
      "learning_rate": 6.094060188589024e-05,
      "loss": 2.0327,
      "step": 182827
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.2456250190734863,
      "learning_rate": 6.093813056284816e-05,
      "loss": 1.8741,
      "step": 182828
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.283193588256836,
      "learning_rate": 6.093565928425207e-05,
      "loss": 2.1978,
      "step": 182829
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.2877053022384644,
      "learning_rate": 6.093318805010266e-05,
      "loss": 1.7699,
      "step": 182830
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.313193917274475,
      "learning_rate": 6.0930716860400286e-05,
      "loss": 2.0293,
      "step": 182831
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.2960854768753052,
      "learning_rate": 6.0928245715145296e-05,
      "loss": 2.2819,
      "step": 182832
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.3656474351882935,
      "learning_rate": 6.0925774614338354e-05,
      "loss": 2.0084,
      "step": 182833
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.2965569496154785,
      "learning_rate": 6.092330355797982e-05,
      "loss": 2.0763,
      "step": 182834
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.3303316831588745,
      "learning_rate": 6.0920832546070095e-05,
      "loss": 1.9955,
      "step": 182835
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.4143143892288208,
      "learning_rate": 6.0918361578609754e-05,
      "loss": 1.9969,
      "step": 182836
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.3203707933425903,
      "learning_rate": 6.091589065559922e-05,
      "loss": 1.9518,
      "step": 182837
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.2654558420181274,
      "learning_rate": 6.0913419777038865e-05,
      "loss": 2.0257,
      "step": 182838
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.326058268547058,
      "learning_rate": 6.0910948942929286e-05,
      "loss": 2.1252,
      "step": 182839
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.3222310543060303,
      "learning_rate": 6.090847815327088e-05,
      "loss": 2.078,
      "step": 182840
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.4824632406234741,
      "learning_rate": 6.0906007408063986e-05,
      "loss": 1.7788,
      "step": 182841
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.4016348123550415,
      "learning_rate": 6.0903536707309295e-05,
      "loss": 2.1356,
      "step": 182842
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.3198870420455933,
      "learning_rate": 6.090106605100705e-05,
      "loss": 1.9004,
      "step": 182843
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.4709776639938354,
      "learning_rate": 6.089859543915791e-05,
      "loss": 2.1815,
      "step": 182844
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.3478280305862427,
      "learning_rate": 6.089612487176222e-05,
      "loss": 2.1323,
      "step": 182845
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.311434030532837,
      "learning_rate": 6.0893654348820463e-05,
      "loss": 1.8777,
      "step": 182846
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.3480030298233032,
      "learning_rate": 6.089118387033302e-05,
      "loss": 2.0713,
      "step": 182847
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.3258249759674072,
      "learning_rate": 6.088871343630047e-05,
      "loss": 1.9847,
      "step": 182848
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.50735604763031,
      "learning_rate": 6.088624304672317e-05,
      "loss": 2.2434,
      "step": 182849
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.458270788192749,
      "learning_rate": 6.0883772701601705e-05,
      "loss": 1.9705,
      "step": 182850
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.3777999877929688,
      "learning_rate": 6.088130240093645e-05,
      "loss": 2.1043,
      "step": 182851
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.2541295289993286,
      "learning_rate": 6.087883214472786e-05,
      "loss": 2.0025,
      "step": 182852
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.3300933837890625,
      "learning_rate": 6.0876361932976344e-05,
      "loss": 2.0038,
      "step": 182853
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.394054889678955,
      "learning_rate": 6.08738917656825e-05,
      "loss": 1.9588,
      "step": 182854
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.3330801725387573,
      "learning_rate": 6.087142164284666e-05,
      "loss": 1.9796,
      "step": 182855
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.376768708229065,
      "learning_rate": 6.086895156446939e-05,
      "loss": 1.8324,
      "step": 182856
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.3562816381454468,
      "learning_rate": 6.0866481530551025e-05,
      "loss": 2.0622,
      "step": 182857
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.2497669458389282,
      "learning_rate": 6.086401154109226e-05,
      "loss": 1.9483,
      "step": 182858
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.3351495265960693,
      "learning_rate": 6.0861541596093235e-05,
      "loss": 1.8782,
      "step": 182859
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.3180665969848633,
      "learning_rate": 6.085907169555464e-05,
      "loss": 2.2187,
      "step": 182860
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.441657543182373,
      "learning_rate": 6.085660183947675e-05,
      "loss": 2.1663,
      "step": 182861
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.4828232526779175,
      "learning_rate": 6.0854132027860226e-05,
      "loss": 2.0173,
      "step": 182862
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.3359402418136597,
      "learning_rate": 6.085166226070537e-05,
      "loss": 2.2396,
      "step": 182863
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.503035545349121,
      "learning_rate": 6.084919253801285e-05,
      "loss": 2.0217,
      "step": 182864
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.3940759897232056,
      "learning_rate": 6.084672285978283e-05,
      "loss": 1.9478,
      "step": 182865
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.5405206680297852,
      "learning_rate": 6.084425322601597e-05,
      "loss": 2.1535,
      "step": 182866
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.3471741676330566,
      "learning_rate": 6.084178363671265e-05,
      "loss": 2.1521,
      "step": 182867
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.3489713668823242,
      "learning_rate": 6.083931409187339e-05,
      "loss": 2.0029,
      "step": 182868
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.3816280364990234,
      "learning_rate": 6.083684459149857e-05,
      "loss": 1.9649,
      "step": 182869
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.4124011993408203,
      "learning_rate": 6.0834375135588806e-05,
      "loss": 2.1442,
      "step": 182870
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.3805514574050903,
      "learning_rate": 6.0831905724144347e-05,
      "loss": 1.9609,
      "step": 182871
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.71595299243927,
      "learning_rate": 6.082943635716578e-05,
      "loss": 2.0658,
      "step": 182872
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.4030978679656982,
      "learning_rate": 6.082696703465348e-05,
      "loss": 1.9716,
      "step": 182873
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.3299428224563599,
      "learning_rate": 6.0824497756608045e-05,
      "loss": 1.908,
      "step": 182874
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.3950916528701782,
      "learning_rate": 6.082202852302978e-05,
      "loss": 2.0769,
      "step": 182875
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.4509354829788208,
      "learning_rate": 6.081955933391933e-05,
      "loss": 2.0127,
      "step": 182876
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.3214800357818604,
      "learning_rate": 6.0817090189276915e-05,
      "loss": 2.2558,
      "step": 182877
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.3660383224487305,
      "learning_rate": 6.08146210891032e-05,
      "loss": 1.9777,
      "step": 182878
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.345358967781067,
      "learning_rate": 6.081215203339848e-05,
      "loss": 2.0538,
      "step": 182879
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.308287262916565,
      "learning_rate": 6.0809683022163344e-05,
      "loss": 2.0759,
      "step": 182880
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.3767014741897583,
      "learning_rate": 6.0807214055398145e-05,
      "loss": 2.0348,
      "step": 182881
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.2808860540390015,
      "learning_rate": 6.080474513310357e-05,
      "loss": 1.8636,
      "step": 182882
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.31488037109375,
      "learning_rate": 6.080227625527976e-05,
      "loss": 2.2559,
      "step": 182883
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.2291488647460938,
      "learning_rate": 6.079980742192737e-05,
      "loss": 1.8594,
      "step": 182884
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.3764406442642212,
      "learning_rate": 6.079733863304674e-05,
      "loss": 2.1271,
      "step": 182885
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.3799551725387573,
      "learning_rate": 6.079486988863848e-05,
      "loss": 2.0176,
      "step": 182886
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.256278395652771,
      "learning_rate": 6.07924011887029e-05,
      "loss": 1.8724,
      "step": 182887
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.615288257598877,
      "learning_rate": 6.078993253324068e-05,
      "loss": 2.0266,
      "step": 182888
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.521829605102539,
      "learning_rate": 6.0787463922251986e-05,
      "loss": 2.1968,
      "step": 182889
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.3844189643859863,
      "learning_rate": 6.0784995355737454e-05,
      "loss": 2.0692,
      "step": 182890
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.2616816759109497,
      "learning_rate": 6.078252683369748e-05,
      "loss": 2.1104,
      "step": 182891
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.2407573461532593,
      "learning_rate": 6.0780058356132586e-05,
      "loss": 2.0151,
      "step": 182892
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.4393858909606934,
      "learning_rate": 6.0777589923043125e-05,
      "loss": 1.8388,
      "step": 182893
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.4028007984161377,
      "learning_rate": 6.077512153442982e-05,
      "loss": 1.9746,
      "step": 182894
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.3219091892242432,
      "learning_rate": 6.077265319029273e-05,
      "loss": 1.8678,
      "step": 182895
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.6159112453460693,
      "learning_rate": 6.07701848906326e-05,
      "loss": 2.0849,
      "step": 182896
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.297772765159607,
      "learning_rate": 6.0767716635449756e-05,
      "loss": 1.8592,
      "step": 182897
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.5577380657196045,
      "learning_rate": 6.076524842474477e-05,
      "loss": 2.3415,
      "step": 182898
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.5478688478469849,
      "learning_rate": 6.076278025851797e-05,
      "loss": 2.026,
      "step": 182899
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.575782060623169,
      "learning_rate": 6.076031213677e-05,
      "loss": 2.18,
      "step": 182900
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.2307418584823608,
      "learning_rate": 6.075784405950114e-05,
      "loss": 2.0472,
      "step": 182901
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.4588712453842163,
      "learning_rate": 6.0755376026711935e-05,
      "loss": 2.0119,
      "step": 182902
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.509592056274414,
      "learning_rate": 6.075290803840275e-05,
      "loss": 2.2442,
      "step": 182903
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.5158601999282837,
      "learning_rate": 6.075044009457421e-05,
      "loss": 2.0686,
      "step": 182904
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.3447768688201904,
      "learning_rate": 6.074797219522657e-05,
      "loss": 2.1275,
      "step": 182905
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.4806116819381714,
      "learning_rate": 6.0745504340360505e-05,
      "loss": 2.0389,
      "step": 182906
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.3587919473648071,
      "learning_rate": 6.074303652997636e-05,
      "loss": 1.9039,
      "step": 182907
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.2177621126174927,
      "learning_rate": 6.074056876407457e-05,
      "loss": 1.9993,
      "step": 182908
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.4757752418518066,
      "learning_rate": 6.073810104265556e-05,
      "loss": 2.0332,
      "step": 182909
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.3035231828689575,
      "learning_rate": 6.073563336571997e-05,
      "loss": 2.1521,
      "step": 182910
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.3174805641174316,
      "learning_rate": 6.073316573326803e-05,
      "loss": 2.1258,
      "step": 182911
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.4705959558486938,
      "learning_rate": 6.0730698145300403e-05,
      "loss": 2.0846,
      "step": 182912
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.425466775894165,
      "learning_rate": 6.072823060181743e-05,
      "loss": 1.909,
      "step": 182913
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.3666104078292847,
      "learning_rate": 6.07257631028196e-05,
      "loss": 2.2259,
      "step": 182914
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.3226853609085083,
      "learning_rate": 6.0723295648307324e-05,
      "loss": 2.1029,
      "step": 182915
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.3247489929199219,
      "learning_rate": 6.072082823828116e-05,
      "loss": 2.137,
      "step": 182916
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.3144125938415527,
      "learning_rate": 6.071836087274141e-05,
      "loss": 2.0955,
      "step": 182917
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.4951374530792236,
      "learning_rate": 6.071589355168874e-05,
      "loss": 2.057,
      "step": 182918
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.2681560516357422,
      "learning_rate": 6.071342627512352e-05,
      "loss": 2.1168,
      "step": 182919
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.3352277278900146,
      "learning_rate": 6.071095904304608e-05,
      "loss": 2.0693,
      "step": 182920
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.271653413772583,
      "learning_rate": 6.070849185545709e-05,
      "loss": 1.807,
      "step": 182921
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.2951958179473877,
      "learning_rate": 6.0706024712356874e-05,
      "loss": 1.9295,
      "step": 182922
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.4054114818572998,
      "learning_rate": 6.0703557613745865e-05,
      "loss": 2.1886,
      "step": 182923
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.2472994327545166,
      "learning_rate": 6.070109055962468e-05,
      "loss": 2.2736,
      "step": 182924
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.3591458797454834,
      "learning_rate": 6.069862354999366e-05,
      "loss": 2.2434,
      "step": 182925
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.3062165975570679,
      "learning_rate": 6.0696156584853194e-05,
      "loss": 2.0951,
      "step": 182926
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.431058645248413,
      "learning_rate": 6.0693689664203936e-05,
      "loss": 2.0918,
      "step": 182927
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.599480390548706,
      "learning_rate": 6.0691222788046155e-05,
      "loss": 2.207,
      "step": 182928
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.649470567703247,
      "learning_rate": 6.068875595638045e-05,
      "loss": 1.9955,
      "step": 182929
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.5751909017562866,
      "learning_rate": 6.0686289169207224e-05,
      "loss": 2.1638,
      "step": 182930
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.3009849786758423,
      "learning_rate": 6.068382242652694e-05,
      "loss": 1.9725,
      "step": 182931
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.4818115234375,
      "learning_rate": 6.068135572834e-05,
      "loss": 2.2397,
      "step": 182932
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.2958917617797852,
      "learning_rate": 6.067888907464696e-05,
      "loss": 2.3532,
      "step": 182933
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.4405072927474976,
      "learning_rate": 6.067642246544817e-05,
      "loss": 2.0853,
      "step": 182934
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.4796010255813599,
      "learning_rate": 6.067395590074422e-05,
      "loss": 2.0924,
      "step": 182935
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.3325397968292236,
      "learning_rate": 6.067148938053551e-05,
      "loss": 2.1522,
      "step": 182936
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.3003798723220825,
      "learning_rate": 6.066902290482247e-05,
      "loss": 1.9898,
      "step": 182937
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.248979926109314,
      "learning_rate": 6.066655647360551e-05,
      "loss": 2.253,
      "step": 182938
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.560652732849121,
      "learning_rate": 6.066409008688522e-05,
      "loss": 1.9847,
      "step": 182939
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.4100515842437744,
      "learning_rate": 6.066162374466193e-05,
      "loss": 1.8792,
      "step": 182940
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.3535034656524658,
      "learning_rate": 6.065915744693622e-05,
      "loss": 1.9163,
      "step": 182941
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.3751156330108643,
      "learning_rate": 6.065669119370845e-05,
      "loss": 2.2745,
      "step": 182942
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.409584879875183,
      "learning_rate": 6.065422498497925e-05,
      "loss": 1.9971,
      "step": 182943
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.3294721841812134,
      "learning_rate": 6.065175882074879e-05,
      "loss": 2.0006,
      "step": 182944
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.6235331296920776,
      "learning_rate": 6.064929270101776e-05,
      "loss": 1.9324,
      "step": 182945
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.2595009803771973,
      "learning_rate": 6.064682662578647e-05,
      "loss": 2.0217,
      "step": 182946
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.349345326423645,
      "learning_rate": 6.0644360595055555e-05,
      "loss": 2.0642,
      "step": 182947
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.3964701890945435,
      "learning_rate": 6.064189460882527e-05,
      "loss": 1.967,
      "step": 182948
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.3899496793746948,
      "learning_rate": 6.0639428667096326e-05,
      "loss": 2.1093,
      "step": 182949
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.482208490371704,
      "learning_rate": 6.063696276986888e-05,
      "loss": 2.1211,
      "step": 182950
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.4997397661209106,
      "learning_rate": 6.063449691714364e-05,
      "loss": 2.0767,
      "step": 182951
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.3439263105392456,
      "learning_rate": 6.0632031108920864e-05,
      "loss": 2.1297,
      "step": 182952
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.4668461084365845,
      "learning_rate": 6.0629565345201194e-05,
      "loss": 2.0486,
      "step": 182953
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.357826590538025,
      "learning_rate": 6.062709962598492e-05,
      "loss": 1.9631,
      "step": 182954
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.367295742034912,
      "learning_rate": 6.0624633951272785e-05,
      "loss": 2.035,
      "step": 182955
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.3922654390335083,
      "learning_rate": 6.0622168321064846e-05,
      "loss": 1.7847,
      "step": 182956
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.2247815132141113,
      "learning_rate": 6.0619702735361876e-05,
      "loss": 1.8274,
      "step": 182957
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.3396327495574951,
      "learning_rate": 6.06172371941641e-05,
      "loss": 2.017,
      "step": 182958
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.451316475868225,
      "learning_rate": 6.06147716974722e-05,
      "loss": 2.0644,
      "step": 182959
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.5581045150756836,
      "learning_rate": 6.0612306245286456e-05,
      "loss": 2.0317,
      "step": 182960
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.1957138776779175,
      "learning_rate": 6.0609840837607584e-05,
      "loss": 1.9327,
      "step": 182961
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.2954858541488647,
      "learning_rate": 6.060737547443567e-05,
      "loss": 1.6956,
      "step": 182962
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.3974812030792236,
      "learning_rate": 6.060491015577146e-05,
      "loss": 1.8644,
      "step": 182963
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.3859992027282715,
      "learning_rate": 6.060244488161521e-05,
      "loss": 2.1851,
      "step": 182964
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.8500375747680664,
      "learning_rate": 6.059997965196759e-05,
      "loss": 2.0082,
      "step": 182965
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.4310370683670044,
      "learning_rate": 6.059751446682887e-05,
      "loss": 2.1474,
      "step": 182966
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.4795957803726196,
      "learning_rate": 6.059504932619967e-05,
      "loss": 2.1697,
      "step": 182967
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.363490343093872,
      "learning_rate": 6.0592584230080375e-05,
      "loss": 1.9124,
      "step": 182968
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.3706309795379639,
      "learning_rate": 6.059011917847144e-05,
      "loss": 2.0594,
      "step": 182969
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.3014272451400757,
      "learning_rate": 6.0587654171373234e-05,
      "loss": 2.0062,
      "step": 182970
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.6221199035644531,
      "learning_rate": 6.058518920878636e-05,
      "loss": 1.8319,
      "step": 182971
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.3930654525756836,
      "learning_rate": 6.058272429071117e-05,
      "loss": 1.9549,
      "step": 182972
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.322707176208496,
      "learning_rate": 6.0580259417148244e-05,
      "loss": 2.0818,
      "step": 182973
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.5378423929214478,
      "learning_rate": 6.057779458809795e-05,
      "loss": 2.0648,
      "step": 182974
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.4578438997268677,
      "learning_rate": 6.057532980356078e-05,
      "loss": 1.8489,
      "step": 182975
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.3735666275024414,
      "learning_rate": 6.05728650635371e-05,
      "loss": 1.8578,
      "step": 182976
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.4424155950546265,
      "learning_rate": 6.057040036802751e-05,
      "loss": 1.8976,
      "step": 182977
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.3008397817611694,
      "learning_rate": 6.0567935717032325e-05,
      "loss": 1.9532,
      "step": 182978
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.3020493984222412,
      "learning_rate": 6.0565471110552154e-05,
      "loss": 1.8864,
      "step": 182979
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.3562912940979004,
      "learning_rate": 6.056300654858741e-05,
      "loss": 1.9406,
      "step": 182980
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.4099371433258057,
      "learning_rate": 6.0560542031138495e-05,
      "loss": 1.9765,
      "step": 182981
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.432675838470459,
      "learning_rate": 6.055807755820583e-05,
      "loss": 2.1115,
      "step": 182982
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.2793816328048706,
      "learning_rate": 6.0555613129789993e-05,
      "loss": 1.9539,
      "step": 182983
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.582518219947815,
      "learning_rate": 6.0553148745891344e-05,
      "loss": 1.7507,
      "step": 182984
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.4104328155517578,
      "learning_rate": 6.055068440651045e-05,
      "loss": 2.1254,
      "step": 182985
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.2920743227005005,
      "learning_rate": 6.054822011164768e-05,
      "loss": 2.0952,
      "step": 182986
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.4857542514801025,
      "learning_rate": 6.054575586130354e-05,
      "loss": 2.0478,
      "step": 182987
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.359834909439087,
      "learning_rate": 6.054329165547838e-05,
      "loss": 2.0893,
      "step": 182988
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.2499605417251587,
      "learning_rate": 6.054082749417284e-05,
      "loss": 1.8248,
      "step": 182989
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.4181575775146484,
      "learning_rate": 6.053836337738719e-05,
      "loss": 1.9376,
      "step": 182990
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.5384349822998047,
      "learning_rate": 6.0535899305122025e-05,
      "loss": 1.9517,
      "step": 182991
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.3569447994232178,
      "learning_rate": 6.0533435277377786e-05,
      "loss": 1.9261,
      "step": 182992
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.3813445568084717,
      "learning_rate": 6.053097129415489e-05,
      "loss": 1.8709,
      "step": 182993
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.3033157587051392,
      "learning_rate": 6.052850735545376e-05,
      "loss": 2.0282,
      "step": 182994
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.4875829219818115,
      "learning_rate": 6.052604346127494e-05,
      "loss": 2.1457,
      "step": 182995
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.4645497798919678,
      "learning_rate": 6.0523579611618775e-05,
      "loss": 2.0937,
      "step": 182996
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.2542482614517212,
      "learning_rate": 6.05211158064859e-05,
      "loss": 1.8512,
      "step": 182997
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.3891891241073608,
      "learning_rate": 6.051865204587667e-05,
      "loss": 1.9811,
      "step": 182998
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.4057339429855347,
      "learning_rate": 6.051618832979153e-05,
      "loss": 1.9826,
      "step": 182999
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.6085067987442017,
      "learning_rate": 6.0513724658230864e-05,
      "loss": 1.913,
      "step": 183000
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.5566656589508057,
      "learning_rate": 6.051126103119533e-05,
      "loss": 2.0727,
      "step": 183001
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.5742287635803223,
      "learning_rate": 6.0508797448685134e-05,
      "loss": 2.2197,
      "step": 183002
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.437544822692871,
      "learning_rate": 6.050633391070099e-05,
      "loss": 2.1115,
      "step": 183003
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.4609918594360352,
      "learning_rate": 6.050387041724326e-05,
      "loss": 1.8531,
      "step": 183004
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.4023702144622803,
      "learning_rate": 6.050140696831225e-05,
      "loss": 2.0897,
      "step": 183005
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.445778727531433,
      "learning_rate": 6.049894356390868e-05,
      "loss": 1.9987,
      "step": 183006
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.3846622705459595,
      "learning_rate": 6.049648020403284e-05,
      "loss": 1.9148,
      "step": 183007
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.3611091375350952,
      "learning_rate": 6.049401688868517e-05,
      "loss": 1.8566,
      "step": 183008
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.5017004013061523,
      "learning_rate": 6.049155361786625e-05,
      "loss": 2.1662,
      "step": 183009
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.6523408889770508,
      "learning_rate": 6.0489090391576477e-05,
      "loss": 1.8697,
      "step": 183010
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.3570635318756104,
      "learning_rate": 6.048662720981622e-05,
      "loss": 1.9049,
      "step": 183011
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.3100062608718872,
      "learning_rate": 6.04841640725861e-05,
      "loss": 2.039,
      "step": 183012
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.4451504945755005,
      "learning_rate": 6.0481700979886407e-05,
      "loss": 1.9998,
      "step": 183013
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.504152536392212,
      "learning_rate": 6.047923793171779e-05,
      "loss": 2.0509,
      "step": 183014
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.3019965887069702,
      "learning_rate": 6.0476774928080575e-05,
      "loss": 1.9634,
      "step": 183015
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.3926969766616821,
      "learning_rate": 6.0474311968975274e-05,
      "loss": 2.112,
      "step": 183016
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.4821219444274902,
      "learning_rate": 6.047184905440222e-05,
      "loss": 2.1765,
      "step": 183017
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.3949542045593262,
      "learning_rate": 6.046938618436208e-05,
      "loss": 1.9423,
      "step": 183018
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.4438287019729614,
      "learning_rate": 6.046692335885508e-05,
      "loss": 2.0356,
      "step": 183019
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.3225812911987305,
      "learning_rate": 6.046446057788192e-05,
      "loss": 1.8367,
      "step": 183020
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.543663740158081,
      "learning_rate": 6.046199784144291e-05,
      "loss": 2.0929,
      "step": 183021
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.253827452659607,
      "learning_rate": 6.045953514953857e-05,
      "loss": 2.0559,
      "step": 183022
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.3074698448181152,
      "learning_rate": 6.045707250216921e-05,
      "loss": 1.9343,
      "step": 183023
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.3213037252426147,
      "learning_rate": 6.045460989933549e-05,
      "loss": 2.1345,
      "step": 183024
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.2525312900543213,
      "learning_rate": 6.0452147341037716e-05,
      "loss": 1.7016,
      "step": 183025
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.2856848239898682,
      "learning_rate": 6.044968482727648e-05,
      "loss": 1.9705,
      "step": 183026
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.5479331016540527,
      "learning_rate": 6.044722235805208e-05,
      "loss": 1.9922,
      "step": 183027
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.3108114004135132,
      "learning_rate": 6.044475993336516e-05,
      "loss": 2.1336,
      "step": 183028
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.3334040641784668,
      "learning_rate": 6.044229755321604e-05,
      "loss": 2.0737,
      "step": 183029
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.387405276298523,
      "learning_rate": 6.043983521760527e-05,
      "loss": 1.9887,
      "step": 183030
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.3045374155044556,
      "learning_rate": 6.0437372926533164e-05,
      "loss": 2.0354,
      "step": 183031
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.3160443305969238,
      "learning_rate": 6.0434910680000334e-05,
      "loss": 1.9827,
      "step": 183032
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.5808162689208984,
      "learning_rate": 6.043244847800711e-05,
      "loss": 1.8129,
      "step": 183033
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.3299673795700073,
      "learning_rate": 6.0429986320554094e-05,
      "loss": 2.2592,
      "step": 183034
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.3340580463409424,
      "learning_rate": 6.042752420764167e-05,
      "loss": 1.9759,
      "step": 183035
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.316669225692749,
      "learning_rate": 6.04250621392703e-05,
      "loss": 2.0377,
      "step": 183036
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.3149065971374512,
      "learning_rate": 6.0422600115440324e-05,
      "loss": 1.9459,
      "step": 183037
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.2218672037124634,
      "learning_rate": 6.042013813615242e-05,
      "loss": 2.0001,
      "step": 183038
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.1985780000686646,
      "learning_rate": 6.041767620140682e-05,
      "loss": 2.1156,
      "step": 183039
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.3770946264266968,
      "learning_rate": 6.041521431120422e-05,
      "loss": 2.0705,
      "step": 183040
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.3216544389724731,
      "learning_rate": 6.041275246554492e-05,
      "loss": 2.0191,
      "step": 183041
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.3991172313690186,
      "learning_rate": 6.041029066442943e-05,
      "loss": 2.0298,
      "step": 183042
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.3035025596618652,
      "learning_rate": 6.04078289078581e-05,
      "loss": 2.2623,
      "step": 183043
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.3885504007339478,
      "learning_rate": 6.040536719583154e-05,
      "loss": 2.028,
      "step": 183044
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.5668628215789795,
      "learning_rate": 6.040290552835009e-05,
      "loss": 1.7916,
      "step": 183045
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.392051339149475,
      "learning_rate": 6.040044390541433e-05,
      "loss": 1.988,
      "step": 183046
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.281750202178955,
      "learning_rate": 6.039798232702467e-05,
      "loss": 1.9809,
      "step": 183047
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.3968549966812134,
      "learning_rate": 6.0395520793181516e-05,
      "loss": 2.193,
      "step": 183048
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.3427479267120361,
      "learning_rate": 6.0393059303885325e-05,
      "loss": 1.9479,
      "step": 183049
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.5132675170898438,
      "learning_rate": 6.039059785913664e-05,
      "loss": 1.9754,
      "step": 183050
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.470542311668396,
      "learning_rate": 6.038813645893578e-05,
      "loss": 2.4661,
      "step": 183051
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.3190264701843262,
      "learning_rate": 6.038567510328338e-05,
      "loss": 2.0611,
      "step": 183052
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.282182216644287,
      "learning_rate": 6.038321379217979e-05,
      "loss": 2.1135,
      "step": 183053
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.4278473854064941,
      "learning_rate": 6.038075252562549e-05,
      "loss": 2.0765,
      "step": 183054
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.4296340942382812,
      "learning_rate": 6.037829130362085e-05,
      "loss": 2.0692,
      "step": 183055
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.3152685165405273,
      "learning_rate": 6.0375830126166514e-05,
      "loss": 1.9927,
      "step": 183056
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.2324519157409668,
      "learning_rate": 6.03733689932627e-05,
      "loss": 1.9377,
      "step": 183057
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.3621424436569214,
      "learning_rate": 6.0370907904910127e-05,
      "loss": 1.9887,
      "step": 183058
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.272233247756958,
      "learning_rate": 6.036844686110911e-05,
      "loss": 1.9686,
      "step": 183059
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.2470743656158447,
      "learning_rate": 6.036598586186012e-05,
      "loss": 1.8386,
      "step": 183060
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.3833876848220825,
      "learning_rate": 6.0363524907163563e-05,
      "loss": 1.9243,
      "step": 183061
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.2868131399154663,
      "learning_rate": 6.036106399702e-05,
      "loss": 1.8412,
      "step": 183062
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.3586336374282837,
      "learning_rate": 6.035860313142976e-05,
      "loss": 1.9084,
      "step": 183063
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.3560802936553955,
      "learning_rate": 6.035614231039349e-05,
      "loss": 2.0321,
      "step": 183064
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.317095160484314,
      "learning_rate": 6.03536815339115e-05,
      "loss": 1.8511,
      "step": 183065
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.5821532011032104,
      "learning_rate": 6.035122080198428e-05,
      "loss": 1.767,
      "step": 183066
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.4778459072113037,
      "learning_rate": 6.034876011461225e-05,
      "loss": 2.0853,
      "step": 183067
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.4729324579238892,
      "learning_rate": 6.034629947179595e-05,
      "loss": 1.9107,
      "step": 183068
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.375709891319275,
      "learning_rate": 6.0343838873535734e-05,
      "loss": 1.868,
      "step": 183069
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.5413284301757812,
      "learning_rate": 6.034137831983221e-05,
      "loss": 1.8163,
      "step": 183070
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.3866527080535889,
      "learning_rate": 6.033891781068575e-05,
      "loss": 2.1195,
      "step": 183071
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.4153993129730225,
      "learning_rate": 6.033645734609678e-05,
      "loss": 2.2819,
      "step": 183072
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.447867751121521,
      "learning_rate": 6.0333996926065734e-05,
      "loss": 1.9677,
      "step": 183073
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.245670199394226,
      "learning_rate": 6.033153655059321e-05,
      "loss": 1.8428,
      "step": 183074
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.40263831615448,
      "learning_rate": 6.032907621967947e-05,
      "loss": 1.9755,
      "step": 183075
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.3602499961853027,
      "learning_rate": 6.032661593332516e-05,
      "loss": 2.0413,
      "step": 183076
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.4032167196273804,
      "learning_rate": 6.032415569153067e-05,
      "loss": 2.0008,
      "step": 183077
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.3576682806015015,
      "learning_rate": 6.032169549429644e-05,
      "loss": 1.8735,
      "step": 183078
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.3511383533477783,
      "learning_rate": 6.031923534162286e-05,
      "loss": 1.9428,
      "step": 183079
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.2969458103179932,
      "learning_rate": 6.031677523351054e-05,
      "loss": 1.7819,
      "step": 183080
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.2989813089370728,
      "learning_rate": 6.031431516995977e-05,
      "loss": 1.9194,
      "step": 183081
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.387250304222107,
      "learning_rate": 6.031185515097115e-05,
      "loss": 2.1154,
      "step": 183082
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.3610790967941284,
      "learning_rate": 6.0309395176545094e-05,
      "loss": 2.1113,
      "step": 183083
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.3288733959197998,
      "learning_rate": 6.0306935246682055e-05,
      "loss": 2.0312,
      "step": 183084
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.2667768001556396,
      "learning_rate": 6.0304475361382395e-05,
      "loss": 2.048,
      "step": 183085
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.2646896839141846,
      "learning_rate": 6.0302015520646766e-05,
      "loss": 1.92,
      "step": 183086
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.5167814493179321,
      "learning_rate": 6.029955572447541e-05,
      "loss": 2.0284,
      "step": 183087
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.39505136013031,
      "learning_rate": 6.0297095972868984e-05,
      "loss": 2.0282,
      "step": 183088
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.2708008289337158,
      "learning_rate": 6.0294636265827835e-05,
      "loss": 1.8874,
      "step": 183089
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.3133318424224854,
      "learning_rate": 6.0292176603352335e-05,
      "loss": 2.0753,
      "step": 183090
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.469177484512329,
      "learning_rate": 6.028971698544316e-05,
      "loss": 2.1276,
      "step": 183091
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.4286298751831055,
      "learning_rate": 6.028725741210066e-05,
      "loss": 2.0854,
      "step": 183092
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.1930269002914429,
      "learning_rate": 6.0284797883325186e-05,
      "loss": 1.8737,
      "step": 183093
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.2316147089004517,
      "learning_rate": 6.028233839911739e-05,
      "loss": 1.9124,
      "step": 183094
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.4044538736343384,
      "learning_rate": 6.027987895947751e-05,
      "loss": 2.1246,
      "step": 183095
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.3493772745132446,
      "learning_rate": 6.027741956440625e-05,
      "loss": 1.9991,
      "step": 183096
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.368159294128418,
      "learning_rate": 6.027496021390393e-05,
      "loss": 2.2221,
      "step": 183097
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.3176124095916748,
      "learning_rate": 6.0272500907971e-05,
      "loss": 1.927,
      "step": 183098
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.2706329822540283,
      "learning_rate": 6.027004164660788e-05,
      "loss": 2.0286,
      "step": 183099
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.400550365447998,
      "learning_rate": 6.0267582429815176e-05,
      "loss": 1.9891,
      "step": 183100
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.4931882619857788,
      "learning_rate": 6.026512325759315e-05,
      "loss": 1.9234,
      "step": 183101
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.3556230068206787,
      "learning_rate": 6.026266412994244e-05,
      "loss": 1.8485,
      "step": 183102
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.2555458545684814,
      "learning_rate": 6.026020504686344e-05,
      "loss": 1.8753,
      "step": 183103
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.5673094987869263,
      "learning_rate": 6.0257746008356525e-05,
      "loss": 1.9337,
      "step": 183104
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.43440842628479,
      "learning_rate": 6.0255287014422284e-05,
      "loss": 1.876,
      "step": 183105
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.5932598114013672,
      "learning_rate": 6.0252828065061085e-05,
      "loss": 2.1114,
      "step": 183106
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.40683114528656,
      "learning_rate": 6.025036916027337e-05,
      "loss": 2.237,
      "step": 183107
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.3083553314208984,
      "learning_rate": 6.02479103000597e-05,
      "loss": 2.0596,
      "step": 183108
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.2942135334014893,
      "learning_rate": 6.024545148442047e-05,
      "loss": 1.9938,
      "step": 183109
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.3237920999526978,
      "learning_rate": 6.0242992713356055e-05,
      "loss": 1.973,
      "step": 183110
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.4459325075149536,
      "learning_rate": 6.024053398686708e-05,
      "loss": 1.9569,
      "step": 183111
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.50868821144104,
      "learning_rate": 6.023807530495385e-05,
      "loss": 2.2031,
      "step": 183112
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.4220283031463623,
      "learning_rate": 6.0235616667616956e-05,
      "loss": 1.8144,
      "step": 183113
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.485608696937561,
      "learning_rate": 6.023315807485678e-05,
      "loss": 2.1554,
      "step": 183114
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.3853901624679565,
      "learning_rate": 6.0230699526673773e-05,
      "loss": 1.9479,
      "step": 183115
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.2817813158035278,
      "learning_rate": 6.0228241023068345e-05,
      "loss": 2.1422,
      "step": 183116
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.3605587482452393,
      "learning_rate": 6.022578256404109e-05,
      "loss": 2.2476,
      "step": 183117
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.4241704940795898,
      "learning_rate": 6.022332414959231e-05,
      "loss": 2.1626,
      "step": 183118
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.417582392692566,
      "learning_rate": 6.02208657797226e-05,
      "loss": 1.9096,
      "step": 183119
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.336714267730713,
      "learning_rate": 6.021840745443237e-05,
      "loss": 2.1182,
      "step": 183120
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.3182674646377563,
      "learning_rate": 6.0215949173722075e-05,
      "loss": 1.9295,
      "step": 183121
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.4382376670837402,
      "learning_rate": 6.021349093759209e-05,
      "loss": 1.8831,
      "step": 183122
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.3126130104064941,
      "learning_rate": 6.021103274604301e-05,
      "loss": 1.7149,
      "step": 183123
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.2717504501342773,
      "learning_rate": 6.020857459907513e-05,
      "loss": 2.0729,
      "step": 183124
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.5604501962661743,
      "learning_rate": 6.020611649668909e-05,
      "loss": 2.2181,
      "step": 183125
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.3075896501541138,
      "learning_rate": 6.020365843888526e-05,
      "loss": 1.987,
      "step": 183126
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.4360206127166748,
      "learning_rate": 6.02012004256641e-05,
      "loss": 2.2777,
      "step": 183127
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.4161460399627686,
      "learning_rate": 6.0198742457026e-05,
      "loss": 1.9468,
      "step": 183128
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.290270209312439,
      "learning_rate": 6.019628453297155e-05,
      "loss": 1.8043,
      "step": 183129
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.3525651693344116,
      "learning_rate": 6.0193826653501034e-05,
      "loss": 2.0649,
      "step": 183130
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.3446580171585083,
      "learning_rate": 6.0191368818615114e-05,
      "loss": 1.9628,
      "step": 183131
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.3717957735061646,
      "learning_rate": 6.0188911028314135e-05,
      "loss": 2.1876,
      "step": 183132
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.411065936088562,
      "learning_rate": 6.018645328259859e-05,
      "loss": 2.0188,
      "step": 183133
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.5715415477752686,
      "learning_rate": 6.018399558146881e-05,
      "loss": 2.186,
      "step": 183134
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.5404735803604126,
      "learning_rate": 6.018153792492544e-05,
      "loss": 2.1857,
      "step": 183135
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.401770830154419,
      "learning_rate": 6.017908031296873e-05,
      "loss": 2.1199,
      "step": 183136
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.5861361026763916,
      "learning_rate": 6.017662274559939e-05,
      "loss": 2.0841,
      "step": 183137
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.3613786697387695,
      "learning_rate": 6.017416522281772e-05,
      "loss": 2.2257,
      "step": 183138
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.3599127531051636,
      "learning_rate": 6.017170774462421e-05,
      "loss": 1.9945,
      "step": 183139
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.338409185409546,
      "learning_rate": 6.01692503110192e-05,
      "loss": 2.0506,
      "step": 183140
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.5899022817611694,
      "learning_rate": 6.016679292200336e-05,
      "loss": 2.0061,
      "step": 183141
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.3091213703155518,
      "learning_rate": 6.016433557757695e-05,
      "loss": 2.1538,
      "step": 183142
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.3442109823226929,
      "learning_rate": 6.01618782777406e-05,
      "loss": 2.1467,
      "step": 183143
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.3560277223587036,
      "learning_rate": 6.015942102249466e-05,
      "loss": 1.88,
      "step": 183144
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.428547739982605,
      "learning_rate": 6.015696381183964e-05,
      "loss": 2.2952,
      "step": 183145
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.3496620655059814,
      "learning_rate": 6.015450664577589e-05,
      "loss": 2.1629,
      "step": 183146
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.3492789268493652,
      "learning_rate": 6.0152049524304e-05,
      "loss": 2.1065,
      "step": 183147
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.4200888872146606,
      "learning_rate": 6.014959244742431e-05,
      "loss": 2.0315,
      "step": 183148
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.328504204750061,
      "learning_rate": 6.0147135415137415e-05,
      "loss": 1.7139,
      "step": 183149
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.3556842803955078,
      "learning_rate": 6.014467842744372e-05,
      "loss": 2.1002,
      "step": 183150
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.3187204599380493,
      "learning_rate": 6.014222148434361e-05,
      "loss": 2.2386,
      "step": 183151
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.509616732597351,
      "learning_rate": 6.0139764585837536e-05,
      "loss": 2.0034,
      "step": 183152
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.3392118215560913,
      "learning_rate": 6.0137307731926086e-05,
      "loss": 1.8587,
      "step": 183153
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.301528811454773,
      "learning_rate": 6.0134850922609566e-05,
      "loss": 2.101,
      "step": 183154
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.4686859846115112,
      "learning_rate": 6.013239415788857e-05,
      "loss": 2.0675,
      "step": 183155
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.6595160961151123,
      "learning_rate": 6.0129937437763476e-05,
      "loss": 2.081,
      "step": 183156
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.4985532760620117,
      "learning_rate": 6.012748076223477e-05,
      "loss": 2.1186,
      "step": 183157
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.3052423000335693,
      "learning_rate": 6.012502413130282e-05,
      "loss": 2.0024,
      "step": 183158
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.3555608987808228,
      "learning_rate": 6.0122567544968234e-05,
      "loss": 1.8044,
      "step": 183159
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.4704889059066772,
      "learning_rate": 6.0120111003231306e-05,
      "loss": 1.8843,
      "step": 183160
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.3429994583129883,
      "learning_rate": 6.0117654506092675e-05,
      "loss": 1.842,
      "step": 183161
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.5012067556381226,
      "learning_rate": 6.011519805355259e-05,
      "loss": 1.9184,
      "step": 183162
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.2670221328735352,
      "learning_rate": 6.011274164561181e-05,
      "loss": 1.8362,
      "step": 183163
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.3972426652908325,
      "learning_rate": 6.011028528227041e-05,
      "loss": 2.0904,
      "step": 183164
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.3045557737350464,
      "learning_rate": 6.010782896352914e-05,
      "loss": 1.9994,
      "step": 183165
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.5318830013275146,
      "learning_rate": 6.0105372689388256e-05,
      "loss": 2.0309,
      "step": 183166
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.3634282350540161,
      "learning_rate": 6.0102916459848394e-05,
      "loss": 1.8895,
      "step": 183167
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.3860117197036743,
      "learning_rate": 6.010046027490986e-05,
      "loss": 2.0886,
      "step": 183168
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.5580018758773804,
      "learning_rate": 6.0098004134573305e-05,
      "loss": 2.0127,
      "step": 183169
    },
    {
      "epoch": 2.38,
      "grad_norm": 1.5194370746612549,
      "learning_rate": 6.0095548038838946e-05,
      "loss": 2.0482,
      "step": 183170
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.4034651517868042,
      "learning_rate": 6.009309198770741e-05,
      "loss": 1.8937,
      "step": 183171
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.4532889127731323,
      "learning_rate": 6.0090635981179025e-05,
      "loss": 1.9385,
      "step": 183172
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.2877553701400757,
      "learning_rate": 6.00881800192544e-05,
      "loss": 2.1181,
      "step": 183173
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.3628098964691162,
      "learning_rate": 6.008572410193382e-05,
      "loss": 1.8389,
      "step": 183174
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.2529762983322144,
      "learning_rate": 6.00832682292179e-05,
      "loss": 2.015,
      "step": 183175
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.4838558435440063,
      "learning_rate": 6.008081240110706e-05,
      "loss": 1.9712,
      "step": 183176
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.2774631977081299,
      "learning_rate": 6.007835661760171e-05,
      "loss": 1.8609,
      "step": 183177
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.3591840267181396,
      "learning_rate": 6.007590087870224e-05,
      "loss": 2.0588,
      "step": 183178
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.3744113445281982,
      "learning_rate": 6.007344518440926e-05,
      "loss": 2.0713,
      "step": 183179
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.3729362487792969,
      "learning_rate": 6.00709895347231e-05,
      "loss": 2.0371,
      "step": 183180
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.4261506795883179,
      "learning_rate": 6.0068533929644355e-05,
      "loss": 2.1323,
      "step": 183181
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.485068440437317,
      "learning_rate": 6.0066078369173364e-05,
      "loss": 2.1207,
      "step": 183182
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.32207190990448,
      "learning_rate": 6.006362285331065e-05,
      "loss": 1.9822,
      "step": 183183
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.438251256942749,
      "learning_rate": 6.006116738205653e-05,
      "loss": 1.8448,
      "step": 183184
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.231726050376892,
      "learning_rate": 6.0058711955411685e-05,
      "loss": 2.0559,
      "step": 183185
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.4332929849624634,
      "learning_rate": 6.005625657337636e-05,
      "loss": 1.9023,
      "step": 183186
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.4194246530532837,
      "learning_rate": 6.005380123595115e-05,
      "loss": 2.1607,
      "step": 183187
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.3893994092941284,
      "learning_rate": 6.005134594313652e-05,
      "loss": 1.9069,
      "step": 183188
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.328190565109253,
      "learning_rate": 6.004889069493277e-05,
      "loss": 1.7162,
      "step": 183189
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.3137177228927612,
      "learning_rate": 6.004643549134054e-05,
      "loss": 1.9912,
      "step": 183190
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.2964898347854614,
      "learning_rate": 6.004398033236022e-05,
      "loss": 2.1462,
      "step": 183191
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.4483343362808228,
      "learning_rate": 6.004152521799215e-05,
      "loss": 2.0892,
      "step": 183192
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.3610039949417114,
      "learning_rate": 6.003907014823699e-05,
      "loss": 1.9632,
      "step": 183193
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.373044490814209,
      "learning_rate": 6.003661512309508e-05,
      "loss": 2.272,
      "step": 183194
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.4382061958312988,
      "learning_rate": 6.003416014256681e-05,
      "loss": 1.9398,
      "step": 183195
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.2638204097747803,
      "learning_rate": 6.003170520665283e-05,
      "loss": 1.8682,
      "step": 183196
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.3325738906860352,
      "learning_rate": 6.002925031535345e-05,
      "loss": 2.025,
      "step": 183197
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.3739416599273682,
      "learning_rate": 6.002679546866912e-05,
      "loss": 2.0977,
      "step": 183198
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.301891565322876,
      "learning_rate": 6.0024340666600405e-05,
      "loss": 2.1365,
      "step": 183199
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.333193063735962,
      "learning_rate": 6.002188590914769e-05,
      "loss": 2.4199,
      "step": 183200
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.4115084409713745,
      "learning_rate": 6.001943119631133e-05,
      "loss": 2.0095,
      "step": 183201
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.4208720922470093,
      "learning_rate": 6.001697652809201e-05,
      "loss": 1.9605,
      "step": 183202
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.2757818698883057,
      "learning_rate": 6.001452190448997e-05,
      "loss": 2.0872,
      "step": 183203
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.2680381536483765,
      "learning_rate": 6.001206732550583e-05,
      "loss": 1.9232,
      "step": 183204
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.4434620141983032,
      "learning_rate": 6.000961279113997e-05,
      "loss": 2.0329,
      "step": 183205
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.2969715595245361,
      "learning_rate": 6.0007158301392885e-05,
      "loss": 2.0065,
      "step": 183206
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.5881903171539307,
      "learning_rate": 6.0004703856264915e-05,
      "loss": 1.7916,
      "step": 183207
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.2913264036178589,
      "learning_rate": 6.000224945575665e-05,
      "loss": 1.8905,
      "step": 183208
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.263515591621399,
      "learning_rate": 5.999979509986842e-05,
      "loss": 1.9736,
      "step": 183209
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.3665741682052612,
      "learning_rate": 5.9997340788600866e-05,
      "loss": 1.8489,
      "step": 183210
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.4180338382720947,
      "learning_rate": 5.9994886521954325e-05,
      "loss": 1.9014,
      "step": 183211
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.3145524263381958,
      "learning_rate": 5.999243229992925e-05,
      "loss": 2.1818,
      "step": 183212
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.3813211917877197,
      "learning_rate": 5.998997812252605e-05,
      "loss": 1.9011,
      "step": 183213
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.4847081899642944,
      "learning_rate": 5.9987523989745325e-05,
      "loss": 2.0332,
      "step": 183214
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.4670403003692627,
      "learning_rate": 5.998506990158737e-05,
      "loss": 2.2666,
      "step": 183215
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.4238466024398804,
      "learning_rate": 5.998261585805278e-05,
      "loss": 2.1448,
      "step": 183216
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.2144402265548706,
      "learning_rate": 5.998016185914197e-05,
      "loss": 1.7481,
      "step": 183217
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.5559301376342773,
      "learning_rate": 5.99777079048554e-05,
      "loss": 2.1341,
      "step": 183218
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.3426458835601807,
      "learning_rate": 5.9975253995193395e-05,
      "loss": 2.0247,
      "step": 183219
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.2203916311264038,
      "learning_rate": 5.99728001301566e-05,
      "loss": 1.8698,
      "step": 183220
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.4762665033340454,
      "learning_rate": 5.997034630974533e-05,
      "loss": 1.9021,
      "step": 183221
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.4054689407348633,
      "learning_rate": 5.996789253396017e-05,
      "loss": 1.8529,
      "step": 183222
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.2734832763671875,
      "learning_rate": 5.9965438802801514e-05,
      "loss": 2.0953,
      "step": 183223
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.520205020904541,
      "learning_rate": 5.996298511626982e-05,
      "loss": 2.0766,
      "step": 183224
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.4343687295913696,
      "learning_rate": 5.9960531474365435e-05,
      "loss": 2.0163,
      "step": 183225
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.3543106317520142,
      "learning_rate": 5.995807787708902e-05,
      "loss": 1.727,
      "step": 183226
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.4197561740875244,
      "learning_rate": 5.995562432444083e-05,
      "loss": 2.1063,
      "step": 183227
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.729278326034546,
      "learning_rate": 5.9953170816421514e-05,
      "loss": 1.9978,
      "step": 183228
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.3764063119888306,
      "learning_rate": 5.995071735303136e-05,
      "loss": 2.2271,
      "step": 183229
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.3160730600357056,
      "learning_rate": 5.994826393427108e-05,
      "loss": 1.9785,
      "step": 183230
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.4621365070343018,
      "learning_rate": 5.994581056014076e-05,
      "loss": 1.9865,
      "step": 183231
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.364012598991394,
      "learning_rate": 5.994335723064111e-05,
      "loss": 1.8494,
      "step": 183232
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.4175845384597778,
      "learning_rate": 5.9940903945772454e-05,
      "loss": 2.0068,
      "step": 183233
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.392075538635254,
      "learning_rate": 5.99384507055354e-05,
      "loss": 1.9882,
      "step": 183234
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.3142595291137695,
      "learning_rate": 5.993599750993024e-05,
      "loss": 1.7314,
      "step": 183235
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.3706824779510498,
      "learning_rate": 5.993354435895769e-05,
      "loss": 2.012,
      "step": 183236
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.682692050933838,
      "learning_rate": 5.993109125261786e-05,
      "loss": 1.8363,
      "step": 183237
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.562246322631836,
      "learning_rate": 5.9928638190911426e-05,
      "loss": 2.0865,
      "step": 183238
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.4079293012619019,
      "learning_rate": 5.992618517383873e-05,
      "loss": 1.9993,
      "step": 183239
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.486804723739624,
      "learning_rate": 5.9923732201400357e-05,
      "loss": 1.8578,
      "step": 183240
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.3223514556884766,
      "learning_rate": 5.992127927359661e-05,
      "loss": 1.8717,
      "step": 183241
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.385855793952942,
      "learning_rate": 5.99188263904282e-05,
      "loss": 2.0001,
      "step": 183242
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.4841169118881226,
      "learning_rate": 5.991637355189525e-05,
      "loss": 2.0791,
      "step": 183243
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.3783067464828491,
      "learning_rate": 5.991392075799846e-05,
      "loss": 1.84,
      "step": 183244
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.2419939041137695,
      "learning_rate": 5.99114680087381e-05,
      "loss": 1.9869,
      "step": 183245
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.4422987699508667,
      "learning_rate": 5.9909015304114826e-05,
      "loss": 2.1446,
      "step": 183246
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.3371868133544922,
      "learning_rate": 5.990656264412892e-05,
      "loss": 1.8885,
      "step": 183247
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.3901304006576538,
      "learning_rate": 5.9904110028781104e-05,
      "loss": 1.8515,
      "step": 183248
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.3512024879455566,
      "learning_rate": 5.990165745807145e-05,
      "loss": 2.039,
      "step": 183249
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.3561112880706787,
      "learning_rate": 5.989920493200069e-05,
      "loss": 2.0578,
      "step": 183250
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.4236295223236084,
      "learning_rate": 5.9896752450569155e-05,
      "loss": 1.8775,
      "step": 183251
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.2944318056106567,
      "learning_rate": 5.989430001377738e-05,
      "loss": 2.0274,
      "step": 183252
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.4071166515350342,
      "learning_rate": 5.989184762162572e-05,
      "loss": 2.2019,
      "step": 183253
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.409393310546875,
      "learning_rate": 5.9889395274114894e-05,
      "loss": 1.8627,
      "step": 183254
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.522779941558838,
      "learning_rate": 5.9886942971244954e-05,
      "loss": 2.0892,
      "step": 183255
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.3264724016189575,
      "learning_rate": 5.988449071301667e-05,
      "loss": 2.0307,
      "step": 183256
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.2669974565505981,
      "learning_rate": 5.9882038499430315e-05,
      "loss": 1.9787,
      "step": 183257
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.455051302909851,
      "learning_rate": 5.987958633048648e-05,
      "loss": 2.1313,
      "step": 183258
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.585536003112793,
      "learning_rate": 5.987713420618551e-05,
      "loss": 1.9521,
      "step": 183259
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.2879180908203125,
      "learning_rate": 5.987468212652805e-05,
      "loss": 1.9347,
      "step": 183260
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.49132239818573,
      "learning_rate": 5.987223009151425e-05,
      "loss": 1.9746,
      "step": 183261
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.2341978549957275,
      "learning_rate": 5.986977810114484e-05,
      "loss": 1.9875,
      "step": 183262
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.3328495025634766,
      "learning_rate": 5.986732615542008e-05,
      "loss": 2.107,
      "step": 183263
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.308493971824646,
      "learning_rate": 5.986487425434061e-05,
      "loss": 2.2379,
      "step": 183264
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.8340585231781006,
      "learning_rate": 5.9862422397906695e-05,
      "loss": 2.1552,
      "step": 183265
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.6258375644683838,
      "learning_rate": 5.9859970586119e-05,
      "loss": 1.9919,
      "step": 183266
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.4087731838226318,
      "learning_rate": 5.9857518818977824e-05,
      "loss": 1.8398,
      "step": 183267
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.3812631368637085,
      "learning_rate": 5.98550670964837e-05,
      "loss": 1.9482,
      "step": 183268
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.4664177894592285,
      "learning_rate": 5.9852615418636965e-05,
      "loss": 2.0967,
      "step": 183269
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.3941280841827393,
      "learning_rate": 5.985016378543824e-05,
      "loss": 2.1606,
      "step": 183270
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.3639429807662964,
      "learning_rate": 5.984771219688781e-05,
      "loss": 2.0362,
      "step": 183271
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.2183606624603271,
      "learning_rate": 5.9845260652986324e-05,
      "loss": 2.0044,
      "step": 183272
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.3542307615280151,
      "learning_rate": 5.984280915373413e-05,
      "loss": 1.7943,
      "step": 183273
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.386580467224121,
      "learning_rate": 5.9840357699131616e-05,
      "loss": 2.0591,
      "step": 183274
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.5657259225845337,
      "learning_rate": 5.983790628917938e-05,
      "loss": 1.9756,
      "step": 183275
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.3406798839569092,
      "learning_rate": 5.983545492387784e-05,
      "loss": 2.2342,
      "step": 183276
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.3213096857070923,
      "learning_rate": 5.98330036032273e-05,
      "loss": 2.0943,
      "step": 183277
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.4436534643173218,
      "learning_rate": 5.9830552327228456e-05,
      "loss": 2.145,
      "step": 183278
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.3782025575637817,
      "learning_rate": 5.982810109588162e-05,
      "loss": 2.0616,
      "step": 183279
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.5379503965377808,
      "learning_rate": 5.98256499091872e-05,
      "loss": 2.106,
      "step": 183280
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.368798851966858,
      "learning_rate": 5.9823198767145795e-05,
      "loss": 2.0194,
      "step": 183281
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.430633544921875,
      "learning_rate": 5.9820747669757805e-05,
      "loss": 2.1523,
      "step": 183282
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.6053152084350586,
      "learning_rate": 5.9818296617023595e-05,
      "loss": 1.9187,
      "step": 183283
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.651294469833374,
      "learning_rate": 5.981584560894377e-05,
      "loss": 2.208,
      "step": 183284
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.3154659271240234,
      "learning_rate": 5.981339464551872e-05,
      "loss": 2.0199,
      "step": 183285
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.3095178604125977,
      "learning_rate": 5.981094372674882e-05,
      "loss": 1.798,
      "step": 183286
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.3622585535049438,
      "learning_rate": 5.980849285263466e-05,
      "loss": 1.9917,
      "step": 183287
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.592720866203308,
      "learning_rate": 5.980604202317658e-05,
      "loss": 2.1222,
      "step": 183288
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.3191852569580078,
      "learning_rate": 5.980359123837515e-05,
      "loss": 1.9923,
      "step": 183289
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.3917354345321655,
      "learning_rate": 5.980114049823076e-05,
      "loss": 2.2898,
      "step": 183290
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.3137192726135254,
      "learning_rate": 5.979868980274388e-05,
      "loss": 1.8275,
      "step": 183291
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.4853075742721558,
      "learning_rate": 5.979623915191488e-05,
      "loss": 2.017,
      "step": 183292
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.5330091714859009,
      "learning_rate": 5.979378854574439e-05,
      "loss": 2.1609,
      "step": 183293
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.3533872365951538,
      "learning_rate": 5.979133798423268e-05,
      "loss": 2.0658,
      "step": 183294
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.4173109531402588,
      "learning_rate": 5.978888746738038e-05,
      "loss": 2.1568,
      "step": 183295
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.3658661842346191,
      "learning_rate": 5.978643699518781e-05,
      "loss": 2.0485,
      "step": 183296
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.2699729204177856,
      "learning_rate": 5.9783986567655525e-05,
      "loss": 1.8552,
      "step": 183297
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.3418158292770386,
      "learning_rate": 5.978153618478385e-05,
      "loss": 1.8908,
      "step": 183298
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.3596593141555786,
      "learning_rate": 5.9779085846573385e-05,
      "loss": 1.7515,
      "step": 183299
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.33721923828125,
      "learning_rate": 5.977663555302445e-05,
      "loss": 1.8743,
      "step": 183300
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.333324670791626,
      "learning_rate": 5.977418530413767e-05,
      "loss": 2.0689,
      "step": 183301
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.3172047138214111,
      "learning_rate": 5.977173509991329e-05,
      "loss": 2.0494,
      "step": 183302
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.3793278932571411,
      "learning_rate": 5.9769284940352044e-05,
      "loss": 2.1127,
      "step": 183303
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.5586791038513184,
      "learning_rate": 5.9766834825454076e-05,
      "loss": 2.1371,
      "step": 183304
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.363487958908081,
      "learning_rate": 5.976438475522009e-05,
      "loss": 2.0589,
      "step": 183305
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.2332245111465454,
      "learning_rate": 5.976193472965033e-05,
      "loss": 2.0676,
      "step": 183306
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.3625938892364502,
      "learning_rate": 5.9759484748745445e-05,
      "loss": 2.0156,
      "step": 183307
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.3370238542556763,
      "learning_rate": 5.975703481250573e-05,
      "loss": 2.0245,
      "step": 183308
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.4065605401992798,
      "learning_rate": 5.975458492093186e-05,
      "loss": 1.939,
      "step": 183309
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.3249105215072632,
      "learning_rate": 5.975213507402402e-05,
      "loss": 2.063,
      "step": 183310
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.392470121383667,
      "learning_rate": 5.9749685271782854e-05,
      "loss": 2.0857,
      "step": 183311
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.3050705194473267,
      "learning_rate": 5.974723551420869e-05,
      "loss": 1.9839,
      "step": 183312
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.3781239986419678,
      "learning_rate": 5.9744785801302096e-05,
      "loss": 2.0125,
      "step": 183313
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.2636151313781738,
      "learning_rate": 5.974233613306344e-05,
      "loss": 2.0892,
      "step": 183314
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.2258893251419067,
      "learning_rate": 5.973988650949335e-05,
      "loss": 2.0121,
      "step": 183315
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.2649019956588745,
      "learning_rate": 5.973743693059203e-05,
      "loss": 2.0889,
      "step": 183316
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.3513882160186768,
      "learning_rate": 5.973498739636011e-05,
      "loss": 1.8552,
      "step": 183317
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.3167905807495117,
      "learning_rate": 5.97325379067979e-05,
      "loss": 2.2601,
      "step": 183318
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.384208083152771,
      "learning_rate": 5.9730088461906046e-05,
      "loss": 2.0032,
      "step": 183319
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.3451464176177979,
      "learning_rate": 5.9727639061684795e-05,
      "loss": 2.0333,
      "step": 183320
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.3913893699645996,
      "learning_rate": 5.972518970613491e-05,
      "loss": 2.1352,
      "step": 183321
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.3523967266082764,
      "learning_rate": 5.9722740395256466e-05,
      "loss": 2.1045,
      "step": 183322
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.507502555847168,
      "learning_rate": 5.972029112905018e-05,
      "loss": 1.8597,
      "step": 183323
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.427889347076416,
      "learning_rate": 5.971784190751636e-05,
      "loss": 2.224,
      "step": 183324
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.3599437475204468,
      "learning_rate": 5.9715392730655607e-05,
      "loss": 1.9386,
      "step": 183325
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.3675531148910522,
      "learning_rate": 5.971294359846819e-05,
      "loss": 1.8659,
      "step": 183326
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.3391152620315552,
      "learning_rate": 5.9710494510954865e-05,
      "loss": 2.1702,
      "step": 183327
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.400224208831787,
      "learning_rate": 5.970804546811571e-05,
      "loss": 1.9873,
      "step": 183328
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.3270604610443115,
      "learning_rate": 5.970559646995148e-05,
      "loss": 1.9408,
      "step": 183329
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.5606886148452759,
      "learning_rate": 5.970314751646241e-05,
      "loss": 2.0624,
      "step": 183330
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.4659360647201538,
      "learning_rate": 5.970069860764914e-05,
      "loss": 2.2642,
      "step": 183331
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.2435946464538574,
      "learning_rate": 5.969824974351197e-05,
      "loss": 1.9175,
      "step": 183332
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.4554134607315063,
      "learning_rate": 5.96958009240516e-05,
      "loss": 1.8136,
      "step": 183333
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.2711145877838135,
      "learning_rate": 5.9693352149268146e-05,
      "loss": 2.0251,
      "step": 183334
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.4459679126739502,
      "learning_rate": 5.969090341916233e-05,
      "loss": 2.0024,
      "step": 183335
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.5198655128479004,
      "learning_rate": 5.96884547337344e-05,
      "loss": 1.884,
      "step": 183336
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.3935673236846924,
      "learning_rate": 5.968600609298504e-05,
      "loss": 1.8659,
      "step": 183337
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.3171030282974243,
      "learning_rate": 5.968355749691447e-05,
      "loss": 2.0407,
      "step": 183338
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.4360071420669556,
      "learning_rate": 5.968110894552346e-05,
      "loss": 2.0041,
      "step": 183339
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.2675186395645142,
      "learning_rate": 5.967866043881208e-05,
      "loss": 1.9884,
      "step": 183340
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.7645021677017212,
      "learning_rate": 5.9676211976781054e-05,
      "loss": 1.8345,
      "step": 183341
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.4471999406814575,
      "learning_rate": 5.9673763559430686e-05,
      "loss": 2.0118,
      "step": 183342
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.4686672687530518,
      "learning_rate": 5.967131518676158e-05,
      "loss": 1.9109,
      "step": 183343
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.36799955368042,
      "learning_rate": 5.9668866858774065e-05,
      "loss": 2.1225,
      "step": 183344
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.4182759523391724,
      "learning_rate": 5.9666418575468775e-05,
      "loss": 2.1322,
      "step": 183345
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.454648494720459,
      "learning_rate": 5.966397033684585e-05,
      "loss": 2.0542,
      "step": 183346
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.3420672416687012,
      "learning_rate": 5.966152214290604e-05,
      "loss": 1.6804,
      "step": 183347
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.4257404804229736,
      "learning_rate": 5.965907399364962e-05,
      "loss": 2.0399,
      "step": 183348
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.4261897802352905,
      "learning_rate": 5.965662588907716e-05,
      "loss": 2.1185,
      "step": 183349
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.4453991651535034,
      "learning_rate": 5.965417782918902e-05,
      "loss": 2.2267,
      "step": 183350
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.3683509826660156,
      "learning_rate": 5.965172981398577e-05,
      "loss": 2.1285,
      "step": 183351
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.3564362525939941,
      "learning_rate": 5.964928184346781e-05,
      "loss": 2.0627,
      "step": 183352
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.4216252565383911,
      "learning_rate": 5.964683391763557e-05,
      "loss": 1.9072,
      "step": 183353
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.3384026288986206,
      "learning_rate": 5.964438603648946e-05,
      "loss": 1.948,
      "step": 183354
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.3348183631896973,
      "learning_rate": 5.964193820003006e-05,
      "loss": 2.0188,
      "step": 183355
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.1794837713241577,
      "learning_rate": 5.9639490408257686e-05,
      "loss": 1.8311,
      "step": 183356
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.4766818284988403,
      "learning_rate": 5.963704266117296e-05,
      "loss": 1.9972,
      "step": 183357
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.4296249151229858,
      "learning_rate": 5.963459495877623e-05,
      "loss": 1.7956,
      "step": 183358
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.3910479545593262,
      "learning_rate": 5.963214730106798e-05,
      "loss": 2.3968,
      "step": 183359
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.4399820566177368,
      "learning_rate": 5.962969968804855e-05,
      "loss": 2.0098,
      "step": 183360
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.302534818649292,
      "learning_rate": 5.962725211971857e-05,
      "loss": 1.9029,
      "step": 183361
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.3024822473526,
      "learning_rate": 5.962480459607838e-05,
      "loss": 1.8906,
      "step": 183362
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.4562910795211792,
      "learning_rate": 5.962235711712854e-05,
      "loss": 2.0012,
      "step": 183363
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.3893158435821533,
      "learning_rate": 5.961990968286942e-05,
      "loss": 2.018,
      "step": 183364
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.4579486846923828,
      "learning_rate": 5.961746229330144e-05,
      "loss": 1.8838,
      "step": 183365
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.3738043308258057,
      "learning_rate": 5.9615014948425186e-05,
      "loss": 2.0636,
      "step": 183366
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.5159680843353271,
      "learning_rate": 5.961256764824102e-05,
      "loss": 1.8526,
      "step": 183367
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.493551254272461,
      "learning_rate": 5.961012039274936e-05,
      "loss": 2.0471,
      "step": 183368
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.420419692993164,
      "learning_rate": 5.960767318195079e-05,
      "loss": 1.9972,
      "step": 183369
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.3388152122497559,
      "learning_rate": 5.96052260158457e-05,
      "loss": 2.0168,
      "step": 183370
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.3478318452835083,
      "learning_rate": 5.960277889443442e-05,
      "loss": 1.8587,
      "step": 183371
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.3604705333709717,
      "learning_rate": 5.960033181771763e-05,
      "loss": 2.0472,
      "step": 183372
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.5074431896209717,
      "learning_rate": 5.959788478569558e-05,
      "loss": 1.9773,
      "step": 183373
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.4354465007781982,
      "learning_rate": 5.9595437798368915e-05,
      "loss": 2.119,
      "step": 183374
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.3272053003311157,
      "learning_rate": 5.9592990855737996e-05,
      "loss": 1.9404,
      "step": 183375
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.2596267461776733,
      "learning_rate": 5.95905439578033e-05,
      "loss": 1.9494,
      "step": 183376
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.4352357387542725,
      "learning_rate": 5.958809710456514e-05,
      "loss": 2.108,
      "step": 183377
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.5489627122879028,
      "learning_rate": 5.958565029602419e-05,
      "loss": 1.9561,
      "step": 183378
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.30838143825531,
      "learning_rate": 5.958320353218072e-05,
      "loss": 2.0359,
      "step": 183379
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.3272453546524048,
      "learning_rate": 5.9580756813035366e-05,
      "loss": 2.0448,
      "step": 183380
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.371932029724121,
      "learning_rate": 5.957831013858846e-05,
      "loss": 2.0385,
      "step": 183381
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.3443409204483032,
      "learning_rate": 5.957586350884053e-05,
      "loss": 2.1734,
      "step": 183382
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.3160254955291748,
      "learning_rate": 5.957341692379187e-05,
      "loss": 2.0251,
      "step": 183383
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.32539963722229,
      "learning_rate": 5.957097038344313e-05,
      "loss": 2.1805,
      "step": 183384
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.5113744735717773,
      "learning_rate": 5.956852388779459e-05,
      "loss": 1.8962,
      "step": 183385
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.2037478685379028,
      "learning_rate": 5.9566077436846905e-05,
      "loss": 2.0863,
      "step": 183386
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.3042738437652588,
      "learning_rate": 5.956363103060036e-05,
      "loss": 2.1011,
      "step": 183387
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.408054232597351,
      "learning_rate": 5.9561184669055615e-05,
      "loss": 1.8761,
      "step": 183388
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.5078903436660767,
      "learning_rate": 5.955873835221282e-05,
      "loss": 1.9139,
      "step": 183389
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.3506070375442505,
      "learning_rate": 5.955629208007266e-05,
      "loss": 2.0077,
      "step": 183390
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.4524248838424683,
      "learning_rate": 5.955384585263548e-05,
      "loss": 1.9347,
      "step": 183391
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.310532808303833,
      "learning_rate": 5.955139966990184e-05,
      "loss": 2.0056,
      "step": 183392
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.317456603050232,
      "learning_rate": 5.9548953531872045e-05,
      "loss": 1.8434,
      "step": 183393
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.3894551992416382,
      "learning_rate": 5.954650743854682e-05,
      "loss": 2.1655,
      "step": 183394
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.2708228826522827,
      "learning_rate": 5.954406138992627e-05,
      "loss": 1.9764,
      "step": 183395
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.3211942911148071,
      "learning_rate": 5.954161538601109e-05,
      "loss": 2.0514,
      "step": 183396
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.3037748336791992,
      "learning_rate": 5.953916942680158e-05,
      "loss": 2.0464,
      "step": 183397
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.275968074798584,
      "learning_rate": 5.953672351229838e-05,
      "loss": 1.7645,
      "step": 183398
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.3893018960952759,
      "learning_rate": 5.953427764250175e-05,
      "loss": 1.9835,
      "step": 183399
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.4021276235580444,
      "learning_rate": 5.953183181741239e-05,
      "loss": 1.8983,
      "step": 183400
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.3565258979797363,
      "learning_rate": 5.952938603703044e-05,
      "loss": 2.0015,
      "step": 183401
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.336120367050171,
      "learning_rate": 5.9526940301356586e-05,
      "loss": 2.1783,
      "step": 183402
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.3734091520309448,
      "learning_rate": 5.952449461039117e-05,
      "loss": 2.1872,
      "step": 183403
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.4073189496994019,
      "learning_rate": 5.9522048964134725e-05,
      "loss": 2.1232,
      "step": 183404
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.589415192604065,
      "learning_rate": 5.951960336258762e-05,
      "loss": 2.1815,
      "step": 183405
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.3806709051132202,
      "learning_rate": 5.9517157805750515e-05,
      "loss": 2.1,
      "step": 183406
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.292844533920288,
      "learning_rate": 5.9514712293623545e-05,
      "loss": 2.1768,
      "step": 183407
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.3467470407485962,
      "learning_rate": 5.951226682620741e-05,
      "loss": 2.027,
      "step": 183408
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.2768068313598633,
      "learning_rate": 5.950982140350241e-05,
      "loss": 1.9651,
      "step": 183409
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.3576844930648804,
      "learning_rate": 5.9507376025509176e-05,
      "loss": 1.9924,
      "step": 183410
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.3479313850402832,
      "learning_rate": 5.950493069222795e-05,
      "loss": 2.0848,
      "step": 183411
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.403206706047058,
      "learning_rate": 5.950248540365945e-05,
      "loss": 2.028,
      "step": 183412
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.4499448537826538,
      "learning_rate": 5.950004015980385e-05,
      "loss": 1.9403,
      "step": 183413
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.3228909969329834,
      "learning_rate": 5.949759496066179e-05,
      "loss": 2.0365,
      "step": 183414
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.2695896625518799,
      "learning_rate": 5.9495149806233586e-05,
      "loss": 2.0524,
      "step": 183415
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.3298789262771606,
      "learning_rate": 5.949270469651985e-05,
      "loss": 2.1941,
      "step": 183416
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.2293661832809448,
      "learning_rate": 5.949025963152089e-05,
      "loss": 2.0558,
      "step": 183417
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.2434520721435547,
      "learning_rate": 5.948781461123738e-05,
      "loss": 2.0478,
      "step": 183418
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.342499852180481,
      "learning_rate": 5.9485369635669475e-05,
      "loss": 2.0956,
      "step": 183419
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.2246006727218628,
      "learning_rate": 5.948292470481787e-05,
      "loss": 2.1378,
      "step": 183420
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.3793632984161377,
      "learning_rate": 5.9480479818682826e-05,
      "loss": 2.005,
      "step": 183421
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.2417941093444824,
      "learning_rate": 5.947803497726501e-05,
      "loss": 1.7583,
      "step": 183422
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.3970999717712402,
      "learning_rate": 5.947559018056466e-05,
      "loss": 2.2227,
      "step": 183423
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.2955199480056763,
      "learning_rate": 5.9473145428582504e-05,
      "loss": 1.8299,
      "step": 183424
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.3428083658218384,
      "learning_rate": 5.9470700721318675e-05,
      "loss": 1.8714,
      "step": 183425
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.4624593257904053,
      "learning_rate": 5.9468256058773876e-05,
      "loss": 2.2253,
      "step": 183426
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.4260557889938354,
      "learning_rate": 5.946581144094838e-05,
      "loss": 1.9077,
      "step": 183427
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.4937986135482788,
      "learning_rate": 5.9463366867842796e-05,
      "loss": 2.1632,
      "step": 183428
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.2964093685150146,
      "learning_rate": 5.946092233945745e-05,
      "loss": 1.8895,
      "step": 183429
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.4092096090316772,
      "learning_rate": 5.945847785579293e-05,
      "loss": 2.0934,
      "step": 183430
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.28985595703125,
      "learning_rate": 5.9456033416849604e-05,
      "loss": 2.0009,
      "step": 183431
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.426503300666809,
      "learning_rate": 5.945358902262797e-05,
      "loss": 2.0007,
      "step": 183432
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.3224998712539673,
      "learning_rate": 5.945114467312836e-05,
      "loss": 1.9761,
      "step": 183433
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.290642499923706,
      "learning_rate": 5.944870036835141e-05,
      "loss": 2.0222,
      "step": 183434
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.3213495016098022,
      "learning_rate": 5.944625610829742e-05,
      "loss": 2.1596,
      "step": 183435
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.2000356912612915,
      "learning_rate": 5.944381189296695e-05,
      "loss": 2.0077,
      "step": 183436
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.2806456089019775,
      "learning_rate": 5.9441367722360444e-05,
      "loss": 1.9737,
      "step": 183437
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.2994049787521362,
      "learning_rate": 5.9438923596478325e-05,
      "loss": 1.9706,
      "step": 183438
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.2133889198303223,
      "learning_rate": 5.943647951532097e-05,
      "loss": 2.016,
      "step": 183439
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.4064370393753052,
      "learning_rate": 5.9434035478889e-05,
      "loss": 1.7674,
      "step": 183440
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.3915939331054688,
      "learning_rate": 5.943159148718272e-05,
      "loss": 2.0672,
      "step": 183441
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.40609610080719,
      "learning_rate": 5.9429147540202694e-05,
      "loss": 2.1792,
      "step": 183442
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.3795946836471558,
      "learning_rate": 5.9426703637949355e-05,
      "loss": 1.9636,
      "step": 183443
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.258217215538025,
      "learning_rate": 5.942425978042311e-05,
      "loss": 2.0815,
      "step": 183444
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.2476739883422852,
      "learning_rate": 5.9421815967624386e-05,
      "loss": 1.8661,
      "step": 183445
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.7929024696350098,
      "learning_rate": 5.941937219955375e-05,
      "loss": 1.8795,
      "step": 183446
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.2730046510696411,
      "learning_rate": 5.94169284762115e-05,
      "loss": 2.0621,
      "step": 183447
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.2813386917114258,
      "learning_rate": 5.941448479759827e-05,
      "loss": 2.1716,
      "step": 183448
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.2737435102462769,
      "learning_rate": 5.9412041163714466e-05,
      "loss": 1.9207,
      "step": 183449
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.410482406616211,
      "learning_rate": 5.940959757456038e-05,
      "loss": 2.0439,
      "step": 183450
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.3539185523986816,
      "learning_rate": 5.9407154030136704e-05,
      "loss": 1.9421,
      "step": 183451
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.324665904045105,
      "learning_rate": 5.9404710530443756e-05,
      "loss": 2.0006,
      "step": 183452
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.522705316543579,
      "learning_rate": 5.940226707548196e-05,
      "loss": 2.1767,
      "step": 183453
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.272378921508789,
      "learning_rate": 5.939982366525188e-05,
      "loss": 2.0462,
      "step": 183454
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.2939438819885254,
      "learning_rate": 5.9397380299753915e-05,
      "loss": 2.1583,
      "step": 183455
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.2858637571334839,
      "learning_rate": 5.939493697898844e-05,
      "loss": 2.2456,
      "step": 183456
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.4215593338012695,
      "learning_rate": 5.939249370295608e-05,
      "loss": 1.9706,
      "step": 183457
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.3754379749298096,
      "learning_rate": 5.9390050471657106e-05,
      "loss": 2.0707,
      "step": 183458
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.4772164821624756,
      "learning_rate": 5.938760728509214e-05,
      "loss": 2.1435,
      "step": 183459
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.3308433294296265,
      "learning_rate": 5.9385164143261566e-05,
      "loss": 1.975,
      "step": 183460
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.390188217163086,
      "learning_rate": 5.938272104616584e-05,
      "loss": 1.8714,
      "step": 183461
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.3164336681365967,
      "learning_rate": 5.938027799380533e-05,
      "loss": 2.2002,
      "step": 183462
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.3553627729415894,
      "learning_rate": 5.9377834986180625e-05,
      "loss": 1.9697,
      "step": 183463
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.5797230005264282,
      "learning_rate": 5.937539202329204e-05,
      "loss": 2.0646,
      "step": 183464
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.3954241275787354,
      "learning_rate": 5.9372949105140244e-05,
      "loss": 1.7744,
      "step": 183465
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.2620184421539307,
      "learning_rate": 5.937050623172552e-05,
      "loss": 2.0316,
      "step": 183466
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.4381942749023438,
      "learning_rate": 5.9368063403048346e-05,
      "loss": 1.923,
      "step": 183467
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.4072582721710205,
      "learning_rate": 5.9365620619109155e-05,
      "loss": 1.9866,
      "step": 183468
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.2969298362731934,
      "learning_rate": 5.9363177879908476e-05,
      "loss": 2.163,
      "step": 183469
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.4227023124694824,
      "learning_rate": 5.9360735185446674e-05,
      "loss": 1.9858,
      "step": 183470
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.3320000171661377,
      "learning_rate": 5.935829253572432e-05,
      "loss": 2.1129,
      "step": 183471
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.3894795179367065,
      "learning_rate": 5.93558499307417e-05,
      "loss": 2.098,
      "step": 183472
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.30885910987854,
      "learning_rate": 5.9353407370499576e-05,
      "loss": 2.0515,
      "step": 183473
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.5546855926513672,
      "learning_rate": 5.9350964854998014e-05,
      "loss": 1.9873,
      "step": 183474
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.349032998085022,
      "learning_rate": 5.934852238423771e-05,
      "loss": 1.9015,
      "step": 183475
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.4375721216201782,
      "learning_rate": 5.9346079958219005e-05,
      "loss": 1.7996,
      "step": 183476
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.4149441719055176,
      "learning_rate": 5.934363757694248e-05,
      "loss": 2.0909,
      "step": 183477
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.361810326576233,
      "learning_rate": 5.9341195240408435e-05,
      "loss": 1.9242,
      "step": 183478
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.4314556121826172,
      "learning_rate": 5.9338752948617565e-05,
      "loss": 1.9363,
      "step": 183479
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.321906328201294,
      "learning_rate": 5.933631070157e-05,
      "loss": 1.8554,
      "step": 183480
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.382698655128479,
      "learning_rate": 5.9333868499266444e-05,
      "loss": 1.8362,
      "step": 183481
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.3085741996765137,
      "learning_rate": 5.933142634170717e-05,
      "loss": 1.9731,
      "step": 183482
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.295770287513733,
      "learning_rate": 5.932898422889283e-05,
      "loss": 1.9143,
      "step": 183483
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.4567859172821045,
      "learning_rate": 5.9326542160823696e-05,
      "loss": 1.9954,
      "step": 183484
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.4138203859329224,
      "learning_rate": 5.932410013750044e-05,
      "loss": 1.8972,
      "step": 183485
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.360365390777588,
      "learning_rate": 5.9321658158923225e-05,
      "loss": 2.0461,
      "step": 183486
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.2561863660812378,
      "learning_rate": 5.9319216225092714e-05,
      "loss": 2.0762,
      "step": 183487
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.273257851600647,
      "learning_rate": 5.931677433600928e-05,
      "loss": 2.002,
      "step": 183488
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.297037959098816,
      "learning_rate": 5.9314332491673444e-05,
      "loss": 2.1024,
      "step": 183489
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.4074530601501465,
      "learning_rate": 5.931189069208552e-05,
      "loss": 1.9126,
      "step": 183490
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.3677048683166504,
      "learning_rate": 5.930944893724626e-05,
      "loss": 2.1037,
      "step": 183491
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.4076040983200073,
      "learning_rate": 5.930700722715571e-05,
      "loss": 1.8952,
      "step": 183492
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.349012017250061,
      "learning_rate": 5.9304565561814666e-05,
      "loss": 2.0855,
      "step": 183493
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.3524041175842285,
      "learning_rate": 5.930212394122332e-05,
      "loss": 1.8533,
      "step": 183494
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.3762376308441162,
      "learning_rate": 5.929968236538235e-05,
      "loss": 1.9514,
      "step": 183495
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.4256187677383423,
      "learning_rate": 5.929724083429205e-05,
      "loss": 2.1421,
      "step": 183496
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.3213160037994385,
      "learning_rate": 5.929479934795298e-05,
      "loss": 2.0625,
      "step": 183497
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.3194109201431274,
      "learning_rate": 5.929235790636558e-05,
      "loss": 2.0202,
      "step": 183498
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.4069706201553345,
      "learning_rate": 5.928991650953026e-05,
      "loss": 1.9624,
      "step": 183499
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.4143904447555542,
      "learning_rate": 5.9287475157447395e-05,
      "loss": 2.1972,
      "step": 183500
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.2928354740142822,
      "learning_rate": 5.9285033850117634e-05,
      "loss": 1.915,
      "step": 183501
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.5106533765792847,
      "learning_rate": 5.9282592587541245e-05,
      "loss": 2.0673,
      "step": 183502
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.3575643301010132,
      "learning_rate": 5.928015136971885e-05,
      "loss": 2.1129,
      "step": 183503
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.3710862398147583,
      "learning_rate": 5.927771019665083e-05,
      "loss": 1.7646,
      "step": 183504
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.3745017051696777,
      "learning_rate": 5.927526906833761e-05,
      "loss": 2.1242,
      "step": 183505
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.3192657232284546,
      "learning_rate": 5.927282798477958e-05,
      "loss": 2.063,
      "step": 183506
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.2581591606140137,
      "learning_rate": 5.9270386945977355e-05,
      "loss": 1.976,
      "step": 183507
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.3835941553115845,
      "learning_rate": 5.926794595193123e-05,
      "loss": 1.7175,
      "step": 183508
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.387230634689331,
      "learning_rate": 5.926550500264183e-05,
      "loss": 2.2248,
      "step": 183509
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.2982542514801025,
      "learning_rate": 5.92630640981095e-05,
      "loss": 1.9205,
      "step": 183510
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.3662543296813965,
      "learning_rate": 5.926062323833474e-05,
      "loss": 2.1218,
      "step": 183511
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.3515361547470093,
      "learning_rate": 5.925818242331787e-05,
      "loss": 2.0864,
      "step": 183512
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.25645911693573,
      "learning_rate": 5.925574165305953e-05,
      "loss": 1.866,
      "step": 183513
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.542245626449585,
      "learning_rate": 5.925330092755999e-05,
      "loss": 1.7959,
      "step": 183514
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.2910046577453613,
      "learning_rate": 5.925086024681991e-05,
      "loss": 2.2096,
      "step": 183515
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.376853346824646,
      "learning_rate": 5.924841961083966e-05,
      "loss": 2.0669,
      "step": 183516
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.3023320436477661,
      "learning_rate": 5.924597901961964e-05,
      "loss": 1.9982,
      "step": 183517
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.3071374893188477,
      "learning_rate": 5.924353847316028e-05,
      "loss": 2.0085,
      "step": 183518
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.4216128587722778,
      "learning_rate": 5.924109797146215e-05,
      "loss": 1.9753,
      "step": 183519
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.4849462509155273,
      "learning_rate": 5.923865751452558e-05,
      "loss": 1.8749,
      "step": 183520
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.55113685131073,
      "learning_rate": 5.923621710235117e-05,
      "loss": 2.0663,
      "step": 183521
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.3831313848495483,
      "learning_rate": 5.923377673493929e-05,
      "loss": 2.3036,
      "step": 183522
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.3726531267166138,
      "learning_rate": 5.923133641229036e-05,
      "loss": 1.9908,
      "step": 183523
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.5054234266281128,
      "learning_rate": 5.922889613440484e-05,
      "loss": 2.0213,
      "step": 183524
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.4019502401351929,
      "learning_rate": 5.922645590128326e-05,
      "loss": 1.9189,
      "step": 183525
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.4686051607131958,
      "learning_rate": 5.9224015712925956e-05,
      "loss": 2.06,
      "step": 183526
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.3233720064163208,
      "learning_rate": 5.9221575569333535e-05,
      "loss": 1.8871,
      "step": 183527
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.4440292119979858,
      "learning_rate": 5.921913547050637e-05,
      "loss": 1.8934,
      "step": 183528
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.3828976154327393,
      "learning_rate": 5.921669541644494e-05,
      "loss": 2.0421,
      "step": 183529
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.531874418258667,
      "learning_rate": 5.921425540714956e-05,
      "loss": 1.7838,
      "step": 183530
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.2982019186019897,
      "learning_rate": 5.921181544262087e-05,
      "loss": 2.2607,
      "step": 183531
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.3358047008514404,
      "learning_rate": 5.920937552285917e-05,
      "loss": 1.9242,
      "step": 183532
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.3454358577728271,
      "learning_rate": 5.9206935647865096e-05,
      "loss": 1.8255,
      "step": 183533
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.2905418872833252,
      "learning_rate": 5.920449581763898e-05,
      "loss": 1.9548,
      "step": 183534
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.342921257019043,
      "learning_rate": 5.9202056032181214e-05,
      "loss": 2.0677,
      "step": 183535
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.3499399423599243,
      "learning_rate": 5.91996162914924e-05,
      "loss": 1.9411,
      "step": 183536
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.3323829174041748,
      "learning_rate": 5.9197176595572945e-05,
      "loss": 2.1938,
      "step": 183537
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.3535923957824707,
      "learning_rate": 5.9194736944423186e-05,
      "loss": 1.9532,
      "step": 183538
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.4218107461929321,
      "learning_rate": 5.919229733804377e-05,
      "loss": 2.1078,
      "step": 183539
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.276731014251709,
      "learning_rate": 5.918985777643501e-05,
      "loss": 1.8771,
      "step": 183540
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.3769440650939941,
      "learning_rate": 5.9187418259597375e-05,
      "loss": 2.1992,
      "step": 183541
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.3467916250228882,
      "learning_rate": 5.9184978787531386e-05,
      "loss": 1.8012,
      "step": 183542
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.3862630128860474,
      "learning_rate": 5.918253936023746e-05,
      "loss": 2.2604,
      "step": 183543
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.1622681617736816,
      "learning_rate": 5.918009997771598e-05,
      "loss": 1.7047,
      "step": 183544
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.3813111782073975,
      "learning_rate": 5.917766063996755e-05,
      "loss": 2.0764,
      "step": 183545
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.2758654356002808,
      "learning_rate": 5.917522134699251e-05,
      "loss": 2.0014,
      "step": 183546
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.2483112812042236,
      "learning_rate": 5.917278209879128e-05,
      "loss": 1.8118,
      "step": 183547
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.579340934753418,
      "learning_rate": 5.917034289536448e-05,
      "loss": 2.1888,
      "step": 183548
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.4109833240509033,
      "learning_rate": 5.9167903736712355e-05,
      "loss": 2.1344,
      "step": 183549
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.3496928215026855,
      "learning_rate": 5.916546462283551e-05,
      "loss": 1.9218,
      "step": 183550
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.3434698581695557,
      "learning_rate": 5.916302555373439e-05,
      "loss": 1.8965,
      "step": 183551
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.3557324409484863,
      "learning_rate": 5.9160586529409415e-05,
      "loss": 1.887,
      "step": 183552
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.7151689529418945,
      "learning_rate": 5.9158147549860925e-05,
      "loss": 2.2407,
      "step": 183553
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.3548634052276611,
      "learning_rate": 5.915570861508958e-05,
      "loss": 1.9528,
      "step": 183554
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.3454095125198364,
      "learning_rate": 5.915326972509565e-05,
      "loss": 2.0786,
      "step": 183555
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.3378664255142212,
      "learning_rate": 5.915083087987977e-05,
      "loss": 1.9028,
      "step": 183556
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.2878749370574951,
      "learning_rate": 5.9148392079442206e-05,
      "loss": 1.7871,
      "step": 183557
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.5084599256515503,
      "learning_rate": 5.9145953323783654e-05,
      "loss": 2.0495,
      "step": 183558
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.3517789840698242,
      "learning_rate": 5.914351461290425e-05,
      "loss": 1.9813,
      "step": 183559
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.5703059434890747,
      "learning_rate": 5.914107594680473e-05,
      "loss": 1.9259,
      "step": 183560
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.2875428199768066,
      "learning_rate": 5.9138637325485317e-05,
      "loss": 2.0844,
      "step": 183561
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.5374810695648193,
      "learning_rate": 5.913619874894665e-05,
      "loss": 2.2273,
      "step": 183562
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.2335076332092285,
      "learning_rate": 5.913376021718906e-05,
      "loss": 1.7619,
      "step": 183563
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.3931570053100586,
      "learning_rate": 5.913132173021312e-05,
      "loss": 2.041,
      "step": 183564
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.281083106994629,
      "learning_rate": 5.912888328801919e-05,
      "loss": 1.9917,
      "step": 183565
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.488442063331604,
      "learning_rate": 5.912644489060777e-05,
      "loss": 1.9779,
      "step": 183566
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.401117205619812,
      "learning_rate": 5.912400653797923e-05,
      "loss": 2.1409,
      "step": 183567
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.4053181409835815,
      "learning_rate": 5.912156823013413e-05,
      "loss": 2.0149,
      "step": 183568
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.4229035377502441,
      "learning_rate": 5.911912996707281e-05,
      "loss": 1.9324,
      "step": 183569
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.4021267890930176,
      "learning_rate": 5.9116691748795867e-05,
      "loss": 1.7511,
      "step": 183570
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.3000932931900024,
      "learning_rate": 5.9114253575303696e-05,
      "loss": 1.9092,
      "step": 183571
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.3883867263793945,
      "learning_rate": 5.911181544659673e-05,
      "loss": 2.0804,
      "step": 183572
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.388140320777893,
      "learning_rate": 5.9109377362675314e-05,
      "loss": 1.9375,
      "step": 183573
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.306819200515747,
      "learning_rate": 5.9106939323540106e-05,
      "loss": 1.9563,
      "step": 183574
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.390036940574646,
      "learning_rate": 5.91045013291914e-05,
      "loss": 2.156,
      "step": 183575
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.4329049587249756,
      "learning_rate": 5.910206337962977e-05,
      "loss": 1.8428,
      "step": 183576
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.3652554750442505,
      "learning_rate": 5.909962547485565e-05,
      "loss": 2.053,
      "step": 183577
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.2645453214645386,
      "learning_rate": 5.9097187614869434e-05,
      "loss": 2.0086,
      "step": 183578
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.2377229928970337,
      "learning_rate": 5.909474979967152e-05,
      "loss": 1.853,
      "step": 183579
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.2606171369552612,
      "learning_rate": 5.9092312029262525e-05,
      "loss": 1.8773,
      "step": 183580
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.3270161151885986,
      "learning_rate": 5.9089874303642725e-05,
      "loss": 2.0647,
      "step": 183581
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.3036707639694214,
      "learning_rate": 5.908743662281273e-05,
      "loss": 1.9394,
      "step": 183582
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.369278907775879,
      "learning_rate": 5.908499898677298e-05,
      "loss": 2.0534,
      "step": 183583
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.320470929145813,
      "learning_rate": 5.908256139552383e-05,
      "loss": 2.1537,
      "step": 183584
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.4873026609420776,
      "learning_rate": 5.908012384906572e-05,
      "loss": 2.0841,
      "step": 183585
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.3962935209274292,
      "learning_rate": 5.907768634739925e-05,
      "loss": 2.191,
      "step": 183586
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.2688624858856201,
      "learning_rate": 5.907524889052467e-05,
      "loss": 2.2078,
      "step": 183587
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.2844175100326538,
      "learning_rate": 5.907281147844267e-05,
      "loss": 2.045,
      "step": 183588
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.5647574663162231,
      "learning_rate": 5.907037411115354e-05,
      "loss": 2.0321,
      "step": 183589
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.417036533355713,
      "learning_rate": 5.906793678865781e-05,
      "loss": 1.9993,
      "step": 183590
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.4238157272338867,
      "learning_rate": 5.9065499510955816e-05,
      "loss": 2.0076,
      "step": 183591
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.3665519952774048,
      "learning_rate": 5.9063062278048156e-05,
      "loss": 2.0557,
      "step": 183592
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.3952765464782715,
      "learning_rate": 5.906062508993516e-05,
      "loss": 1.8626,
      "step": 183593
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.3669123649597168,
      "learning_rate": 5.9058187946617406e-05,
      "loss": 1.9743,
      "step": 183594
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.3703627586364746,
      "learning_rate": 5.9055750848095284e-05,
      "loss": 2.1174,
      "step": 183595
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.3508914709091187,
      "learning_rate": 5.905331379436926e-05,
      "loss": 2.2101,
      "step": 183596
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.4171630144119263,
      "learning_rate": 5.90508767854397e-05,
      "loss": 2.086,
      "step": 183597
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.3087376356124878,
      "learning_rate": 5.904843982130718e-05,
      "loss": 2.0473,
      "step": 183598
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.6676090955734253,
      "learning_rate": 5.9046002901972055e-05,
      "loss": 1.8257,
      "step": 183599
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.36369788646698,
      "learning_rate": 5.904356602743486e-05,
      "loss": 1.9138,
      "step": 183600
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.4421907663345337,
      "learning_rate": 5.9041129197696066e-05,
      "loss": 1.9166,
      "step": 183601
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.251899242401123,
      "learning_rate": 5.903869241275604e-05,
      "loss": 2.2146,
      "step": 183602
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.2996703386306763,
      "learning_rate": 5.903625567261521e-05,
      "loss": 2.0046,
      "step": 183603
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.3636380434036255,
      "learning_rate": 5.903381897727417e-05,
      "loss": 2.2134,
      "step": 183604
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.3478670120239258,
      "learning_rate": 5.90313823267332e-05,
      "loss": 2.0137,
      "step": 183605
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.329221248626709,
      "learning_rate": 5.902894572099292e-05,
      "loss": 2.0532,
      "step": 183606
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.4545117616653442,
      "learning_rate": 5.902650916005371e-05,
      "loss": 2.2081,
      "step": 183607
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.5362612009048462,
      "learning_rate": 5.902407264391603e-05,
      "loss": 2.1402,
      "step": 183608
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.3641469478607178,
      "learning_rate": 5.902163617258024e-05,
      "loss": 2.0117,
      "step": 183609
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.4662501811981201,
      "learning_rate": 5.901919974604695e-05,
      "loss": 2.0893,
      "step": 183610
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.5386022329330444,
      "learning_rate": 5.9016763364316486e-05,
      "loss": 2.0385,
      "step": 183611
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.5410352945327759,
      "learning_rate": 5.9014327027389416e-05,
      "loss": 2.1393,
      "step": 183612
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.2820353507995605,
      "learning_rate": 5.901189073526611e-05,
      "loss": 2.0428,
      "step": 183613
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.517248511314392,
      "learning_rate": 5.9009454487947064e-05,
      "loss": 2.0637,
      "step": 183614
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.3803725242614746,
      "learning_rate": 5.9007018285432636e-05,
      "loss": 2.1008,
      "step": 183615
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.35922372341156,
      "learning_rate": 5.900458212772341e-05,
      "loss": 1.82,
      "step": 183616
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.348713994026184,
      "learning_rate": 5.900214601481974e-05,
      "loss": 2.0078,
      "step": 183617
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.3732486963272095,
      "learning_rate": 5.899970994672216e-05,
      "loss": 1.9442,
      "step": 183618
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.3007491827011108,
      "learning_rate": 5.899727392343111e-05,
      "loss": 2.0968,
      "step": 183619
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.4546360969543457,
      "learning_rate": 5.8994837944946914e-05,
      "loss": 2.0923,
      "step": 183620
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.3510342836380005,
      "learning_rate": 5.8992402011270214e-05,
      "loss": 2.2524,
      "step": 183621
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.3077744245529175,
      "learning_rate": 5.89899661224014e-05,
      "loss": 1.969,
      "step": 183622
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.3221646547317505,
      "learning_rate": 5.898753027834078e-05,
      "loss": 2.0645,
      "step": 183623
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.3562921285629272,
      "learning_rate": 5.898509447908905e-05,
      "loss": 2.2313,
      "step": 183624
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.4554505348205566,
      "learning_rate": 5.8982658724646504e-05,
      "loss": 2.1471,
      "step": 183625
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.297598123550415,
      "learning_rate": 5.898022301501358e-05,
      "loss": 2.075,
      "step": 183626
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.3184280395507812,
      "learning_rate": 5.897778735019085e-05,
      "loss": 2.0952,
      "step": 183627
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.6507411003112793,
      "learning_rate": 5.897535173017868e-05,
      "loss": 1.8353,
      "step": 183628
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.3089358806610107,
      "learning_rate": 5.897291615497749e-05,
      "loss": 1.9893,
      "step": 183629
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.2524839639663696,
      "learning_rate": 5.897048062458786e-05,
      "loss": 1.7726,
      "step": 183630
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.3863426446914673,
      "learning_rate": 5.896804513901008e-05,
      "loss": 2.0997,
      "step": 183631
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.4055743217468262,
      "learning_rate": 5.8965609698244754e-05,
      "loss": 2.13,
      "step": 183632
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.3250147104263306,
      "learning_rate": 5.896317430229228e-05,
      "loss": 2.0133,
      "step": 183633
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.464941382408142,
      "learning_rate": 5.896073895115303e-05,
      "loss": 1.8038,
      "step": 183634
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.267928957939148,
      "learning_rate": 5.895830364482763e-05,
      "loss": 1.887,
      "step": 183635
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.2746561765670776,
      "learning_rate": 5.8955868383316386e-05,
      "loss": 2.1436,
      "step": 183636
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.4861780405044556,
      "learning_rate": 5.895343316661976e-05,
      "loss": 1.9392,
      "step": 183637
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.3539140224456787,
      "learning_rate": 5.895099799473828e-05,
      "loss": 1.9998,
      "step": 183638
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.4161850214004517,
      "learning_rate": 5.894856286767239e-05,
      "loss": 1.8444,
      "step": 183639
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.3623838424682617,
      "learning_rate": 5.8946127785422415e-05,
      "loss": 1.9115,
      "step": 183640
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.3235114812850952,
      "learning_rate": 5.8943692747989027e-05,
      "loss": 2.0464,
      "step": 183641
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.3037428855895996,
      "learning_rate": 5.894125775537252e-05,
      "loss": 1.9965,
      "step": 183642
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.4839752912521362,
      "learning_rate": 5.893882280757329e-05,
      "loss": 1.9565,
      "step": 183643
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.4037606716156006,
      "learning_rate": 5.8936387904591985e-05,
      "loss": 2.2286,
      "step": 183644
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.3803908824920654,
      "learning_rate": 5.893395304642896e-05,
      "loss": 1.8641,
      "step": 183645
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.4742025136947632,
      "learning_rate": 5.8931518233084554e-05,
      "loss": 2.0323,
      "step": 183646
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.245962142944336,
      "learning_rate": 5.8929083464559426e-05,
      "loss": 1.9605,
      "step": 183647
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.4366453886032104,
      "learning_rate": 5.8926648740853875e-05,
      "loss": 2.1203,
      "step": 183648
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.3078899383544922,
      "learning_rate": 5.892421406196844e-05,
      "loss": 2.028,
      "step": 183649
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.3980456590652466,
      "learning_rate": 5.8921779427903594e-05,
      "loss": 1.8323,
      "step": 183650
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.3454625606536865,
      "learning_rate": 5.891934483865972e-05,
      "loss": 1.7382,
      "step": 183651
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.554353952407837,
      "learning_rate": 5.891691029423723e-05,
      "loss": 2.1669,
      "step": 183652
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.3233957290649414,
      "learning_rate": 5.8914475794636687e-05,
      "loss": 1.8735,
      "step": 183653
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.2626392841339111,
      "learning_rate": 5.891204133985842e-05,
      "loss": 1.9047,
      "step": 183654
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.4868255853652954,
      "learning_rate": 5.890960692990303e-05,
      "loss": 2.2958,
      "step": 183655
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.3629074096679688,
      "learning_rate": 5.890717256477092e-05,
      "loss": 2.0761,
      "step": 183656
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.375626564025879,
      "learning_rate": 5.890473824446249e-05,
      "loss": 2.0637,
      "step": 183657
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.3347482681274414,
      "learning_rate": 5.890230396897817e-05,
      "loss": 2.2336,
      "step": 183658
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.3969156742095947,
      "learning_rate": 5.889986973831853e-05,
      "loss": 2.0829,
      "step": 183659
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.2293514013290405,
      "learning_rate": 5.8897435552483866e-05,
      "loss": 1.8711,
      "step": 183660
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.3333526849746704,
      "learning_rate": 5.889500141147481e-05,
      "loss": 1.9947,
      "step": 183661
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.3055026531219482,
      "learning_rate": 5.889256731529173e-05,
      "loss": 2.0304,
      "step": 183662
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.516596794128418,
      "learning_rate": 5.889013326393506e-05,
      "loss": 2.1142,
      "step": 183663
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.3998844623565674,
      "learning_rate": 5.888769925740516e-05,
      "loss": 1.8836,
      "step": 183664
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.2911728620529175,
      "learning_rate": 5.888526529570271e-05,
      "loss": 2.0301,
      "step": 183665
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.3808650970458984,
      "learning_rate": 5.888283137882797e-05,
      "loss": 2.0633,
      "step": 183666
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.3522664308547974,
      "learning_rate": 5.88803975067815e-05,
      "loss": 1.9573,
      "step": 183667
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.6242725849151611,
      "learning_rate": 5.887796367956373e-05,
      "loss": 2.0157,
      "step": 183668
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.3640267848968506,
      "learning_rate": 5.887552989717511e-05,
      "loss": 1.9946,
      "step": 183669
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.3866603374481201,
      "learning_rate": 5.887309615961599e-05,
      "loss": 1.9194,
      "step": 183670
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.5587291717529297,
      "learning_rate": 5.887066246688702e-05,
      "loss": 2.1296,
      "step": 183671
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.4373704195022583,
      "learning_rate": 5.8868228818988415e-05,
      "loss": 2.1116,
      "step": 183672
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.695807933807373,
      "learning_rate": 5.886579521592088e-05,
      "loss": 2.0145,
      "step": 183673
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.5369449853897095,
      "learning_rate": 5.8863361657684725e-05,
      "loss": 2.1928,
      "step": 183674
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.4827616214752197,
      "learning_rate": 5.886092814428044e-05,
      "loss": 1.929,
      "step": 183675
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.6572589874267578,
      "learning_rate": 5.885849467570836e-05,
      "loss": 1.9785,
      "step": 183676
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.3692702054977417,
      "learning_rate": 5.885606125196914e-05,
      "loss": 1.9438,
      "step": 183677
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.2421308755874634,
      "learning_rate": 5.885362787306304e-05,
      "loss": 1.6912,
      "step": 183678
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.4937773942947388,
      "learning_rate": 5.8851194538990666e-05,
      "loss": 1.9809,
      "step": 183679
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.3524410724639893,
      "learning_rate": 5.884876124975243e-05,
      "loss": 1.7814,
      "step": 183680
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.5963990688323975,
      "learning_rate": 5.8846328005348765e-05,
      "loss": 1.9543,
      "step": 183681
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.5966192483901978,
      "learning_rate": 5.884389480578e-05,
      "loss": 2.2751,
      "step": 183682
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.6248761415481567,
      "learning_rate": 5.884146165104684e-05,
      "loss": 1.8769,
      "step": 183683
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.345339298248291,
      "learning_rate": 5.883902854114948e-05,
      "loss": 1.796,
      "step": 183684
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.4145854711532593,
      "learning_rate": 5.883659547608859e-05,
      "loss": 1.8711,
      "step": 183685
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.3364644050598145,
      "learning_rate": 5.883416245586453e-05,
      "loss": 1.9136,
      "step": 183686
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.46027672290802,
      "learning_rate": 5.883172948047777e-05,
      "loss": 2.0605,
      "step": 183687
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.5102735757827759,
      "learning_rate": 5.882929654992865e-05,
      "loss": 2.006,
      "step": 183688
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.3882321119308472,
      "learning_rate": 5.8826863664217795e-05,
      "loss": 1.8138,
      "step": 183689
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.3670296669006348,
      "learning_rate": 5.88244308233455e-05,
      "loss": 1.9546,
      "step": 183690
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.405324101448059,
      "learning_rate": 5.8821998027312376e-05,
      "loss": 2.1,
      "step": 183691
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.345680832862854,
      "learning_rate": 5.881956527611879e-05,
      "loss": 1.8823,
      "step": 183692
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.3351216316223145,
      "learning_rate": 5.881713256976523e-05,
      "loss": 2.1783,
      "step": 183693
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.3977614641189575,
      "learning_rate": 5.881469990825199e-05,
      "loss": 2.1379,
      "step": 183694
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.4570504426956177,
      "learning_rate": 5.8812267291579765e-05,
      "loss": 1.9539,
      "step": 183695
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.2892619371414185,
      "learning_rate": 5.880983471974879e-05,
      "loss": 1.7573,
      "step": 183696
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.2873133420944214,
      "learning_rate": 5.880740219275972e-05,
      "loss": 1.947,
      "step": 183697
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.3108148574829102,
      "learning_rate": 5.8804969710612815e-05,
      "loss": 1.91,
      "step": 183698
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.4109338521957397,
      "learning_rate": 5.880253727330877e-05,
      "loss": 1.8823,
      "step": 183699
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.4143381118774414,
      "learning_rate": 5.880010488084775e-05,
      "loss": 1.8839,
      "step": 183700
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.3739345073699951,
      "learning_rate": 5.87976725332304e-05,
      "loss": 2.0005,
      "step": 183701
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.3193036317825317,
      "learning_rate": 5.879524023045704e-05,
      "loss": 2.1668,
      "step": 183702
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.4213708639144897,
      "learning_rate": 5.8792807972528274e-05,
      "loss": 1.9923,
      "step": 183703
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.3400152921676636,
      "learning_rate": 5.879037575944444e-05,
      "loss": 2.057,
      "step": 183704
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.5205401182174683,
      "learning_rate": 5.878794359120613e-05,
      "loss": 1.9251,
      "step": 183705
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.3455761671066284,
      "learning_rate": 5.8785511467813585e-05,
      "loss": 1.9811,
      "step": 183706
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.4467219114303589,
      "learning_rate": 5.878307938926739e-05,
      "loss": 1.8705,
      "step": 183707
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.2152495384216309,
      "learning_rate": 5.878064735556793e-05,
      "loss": 2.0701,
      "step": 183708
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.3965392112731934,
      "learning_rate": 5.87782153667158e-05,
      "loss": 2.1092,
      "step": 183709
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.7333850860595703,
      "learning_rate": 5.877578342271125e-05,
      "loss": 2.0917,
      "step": 183710
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.413757085800171,
      "learning_rate": 5.87733515235549e-05,
      "loss": 1.9474,
      "step": 183711
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.3579516410827637,
      "learning_rate": 5.877091966924717e-05,
      "loss": 2.0095,
      "step": 183712
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.437853217124939,
      "learning_rate": 5.876848785978847e-05,
      "loss": 2.1464,
      "step": 183713
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.3185672760009766,
      "learning_rate": 5.876605609517916e-05,
      "loss": 2.1389,
      "step": 183714
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.9709385633468628,
      "learning_rate": 5.87636243754199e-05,
      "loss": 1.9938,
      "step": 183715
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.35321044921875,
      "learning_rate": 5.876119270051094e-05,
      "loss": 1.9901,
      "step": 183716
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.3364498615264893,
      "learning_rate": 5.87587610704529e-05,
      "loss": 1.9069,
      "step": 183717
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.5144461393356323,
      "learning_rate": 5.875632948524618e-05,
      "loss": 2.017,
      "step": 183718
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.3860297203063965,
      "learning_rate": 5.8753897944891116e-05,
      "loss": 1.9719,
      "step": 183719
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.4768942594528198,
      "learning_rate": 5.875146644938834e-05,
      "loss": 2.0279,
      "step": 183720
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.5355643033981323,
      "learning_rate": 5.874903499873822e-05,
      "loss": 1.9307,
      "step": 183721
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.460369348526001,
      "learning_rate": 5.8746603592941155e-05,
      "loss": 2.1313,
      "step": 183722
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.33745539188385,
      "learning_rate": 5.874417223199771e-05,
      "loss": 2.0011,
      "step": 183723
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.2624362707138062,
      "learning_rate": 5.8741740915908255e-05,
      "loss": 1.9656,
      "step": 183724
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.4312337636947632,
      "learning_rate": 5.873930964467322e-05,
      "loss": 2.1609,
      "step": 183725
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.2000232934951782,
      "learning_rate": 5.8736878418293163e-05,
      "loss": 1.9209,
      "step": 183726
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.5115375518798828,
      "learning_rate": 5.873444723676847e-05,
      "loss": 2.187,
      "step": 183727
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.502408742904663,
      "learning_rate": 5.873201610009955e-05,
      "loss": 1.6936,
      "step": 183728
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.3642957210540771,
      "learning_rate": 5.8729585008286964e-05,
      "loss": 1.8791,
      "step": 183729
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.2898486852645874,
      "learning_rate": 5.8727153961331086e-05,
      "loss": 2.0121,
      "step": 183730
    },
    {
      "epoch": 2.39,
      "grad_norm": 2.1252634525299072,
      "learning_rate": 5.87247229592323e-05,
      "loss": 2.0278,
      "step": 183731
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.7155463695526123,
      "learning_rate": 5.8722292001991256e-05,
      "loss": 2.1372,
      "step": 183732
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.287577509880066,
      "learning_rate": 5.87198610896082e-05,
      "loss": 2.1727,
      "step": 183733
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.431341290473938,
      "learning_rate": 5.871743022208373e-05,
      "loss": 1.9728,
      "step": 183734
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.241459846496582,
      "learning_rate": 5.871499939941827e-05,
      "loss": 2.0641,
      "step": 183735
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.2296808958053589,
      "learning_rate": 5.8712568621612235e-05,
      "loss": 2.0246,
      "step": 183736
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.3583245277404785,
      "learning_rate": 5.8710137888666006e-05,
      "loss": 2.1096,
      "step": 183737
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.462350606918335,
      "learning_rate": 5.8707707200580234e-05,
      "loss": 1.9563,
      "step": 183738
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.5036044120788574,
      "learning_rate": 5.870527655735514e-05,
      "loss": 2.2667,
      "step": 183739
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.634659767150879,
      "learning_rate": 5.870284595899136e-05,
      "loss": 2.1568,
      "step": 183740
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.448838710784912,
      "learning_rate": 5.87004154054893e-05,
      "loss": 2.0824,
      "step": 183741
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.3105696439743042,
      "learning_rate": 5.869798489684938e-05,
      "loss": 1.8575,
      "step": 183742
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.3909046649932861,
      "learning_rate": 5.8695554433071976e-05,
      "loss": 1.9431,
      "step": 183743
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.4472088813781738,
      "learning_rate": 5.869312401415772e-05,
      "loss": 1.9824,
      "step": 183744
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.714586615562439,
      "learning_rate": 5.869069364010685e-05,
      "loss": 2.0006,
      "step": 183745
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.2760826349258423,
      "learning_rate": 5.868826331092005e-05,
      "loss": 1.8969,
      "step": 183746
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.2504295110702515,
      "learning_rate": 5.868583302659763e-05,
      "loss": 1.9727,
      "step": 183747
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.4567464590072632,
      "learning_rate": 5.868340278714009e-05,
      "loss": 1.9677,
      "step": 183748
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.4861507415771484,
      "learning_rate": 5.868097259254779e-05,
      "loss": 1.9965,
      "step": 183749
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.688360333442688,
      "learning_rate": 5.867854244282131e-05,
      "loss": 1.7311,
      "step": 183750
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.3147878646850586,
      "learning_rate": 5.867611233796097e-05,
      "loss": 2.0142,
      "step": 183751
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.7532302141189575,
      "learning_rate": 5.867368227796737e-05,
      "loss": 1.8902,
      "step": 183752
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.389426827430725,
      "learning_rate": 5.867125226284092e-05,
      "loss": 1.9817,
      "step": 183753
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.5880508422851562,
      "learning_rate": 5.866882229258201e-05,
      "loss": 2.2376,
      "step": 183754
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.3810925483703613,
      "learning_rate": 5.866639236719104e-05,
      "loss": 1.8871,
      "step": 183755
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.5137648582458496,
      "learning_rate": 5.8663962486668616e-05,
      "loss": 1.9805,
      "step": 183756
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.5120421648025513,
      "learning_rate": 5.8661532651015064e-05,
      "loss": 1.9561,
      "step": 183757
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.4705456495285034,
      "learning_rate": 5.8659102860230955e-05,
      "loss": 2.0502,
      "step": 183758
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.710449457168579,
      "learning_rate": 5.8656673114316654e-05,
      "loss": 2.0059,
      "step": 183759
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.7181600332260132,
      "learning_rate": 5.865424341327266e-05,
      "loss": 2.0392,
      "step": 183760
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.4917635917663574,
      "learning_rate": 5.865181375709931e-05,
      "loss": 2.1672,
      "step": 183761
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.3893508911132812,
      "learning_rate": 5.864938414579723e-05,
      "loss": 1.9891,
      "step": 183762
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.3478553295135498,
      "learning_rate": 5.864695457936669e-05,
      "loss": 1.9503,
      "step": 183763
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.7524826526641846,
      "learning_rate": 5.864452505780836e-05,
      "loss": 2.131,
      "step": 183764
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.5965992212295532,
      "learning_rate": 5.864209558112243e-05,
      "loss": 2.2371,
      "step": 183765
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.3990577459335327,
      "learning_rate": 5.863966614930967e-05,
      "loss": 2.0479,
      "step": 183766
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.4238721132278442,
      "learning_rate": 5.863723676237022e-05,
      "loss": 2.1247,
      "step": 183767
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.508529543876648,
      "learning_rate": 5.863480742030471e-05,
      "loss": 1.933,
      "step": 183768
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.7367229461669922,
      "learning_rate": 5.8632378123113465e-05,
      "loss": 1.9787,
      "step": 183769
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.4298712015151978,
      "learning_rate": 5.8629948870797125e-05,
      "loss": 2.0984,
      "step": 183770
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.686780333518982,
      "learning_rate": 5.8627519663355925e-05,
      "loss": 1.9971,
      "step": 183771
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.3272485733032227,
      "learning_rate": 5.862509050079063e-05,
      "loss": 1.908,
      "step": 183772
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.4027141332626343,
      "learning_rate": 5.862266138310127e-05,
      "loss": 1.8436,
      "step": 183773
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.286456823348999,
      "learning_rate": 5.862023231028864e-05,
      "loss": 1.9916,
      "step": 183774
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.3543161153793335,
      "learning_rate": 5.8617803282352946e-05,
      "loss": 1.9187,
      "step": 183775
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.3874149322509766,
      "learning_rate": 5.861537429929486e-05,
      "loss": 2.1934,
      "step": 183776
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.2413796186447144,
      "learning_rate": 5.8612945361114674e-05,
      "loss": 1.9451,
      "step": 183777
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.3043615818023682,
      "learning_rate": 5.861051646781302e-05,
      "loss": 1.8627,
      "step": 183778
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.2923228740692139,
      "learning_rate": 5.860808761939007e-05,
      "loss": 2.0264,
      "step": 183779
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.3701421022415161,
      "learning_rate": 5.860565881584655e-05,
      "loss": 1.8113,
      "step": 183780
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.398891568183899,
      "learning_rate": 5.860323005718266e-05,
      "loss": 2.0478,
      "step": 183781
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.3418834209442139,
      "learning_rate": 5.860080134339911e-05,
      "loss": 2.0537,
      "step": 183782
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.3313300609588623,
      "learning_rate": 5.859837267449612e-05,
      "loss": 2.1337,
      "step": 183783
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.3562989234924316,
      "learning_rate": 5.859594405047443e-05,
      "loss": 1.8541,
      "step": 183784
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.5562775135040283,
      "learning_rate": 5.859351547133414e-05,
      "loss": 2.0625,
      "step": 183785
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.4589903354644775,
      "learning_rate": 5.859108693707595e-05,
      "loss": 2.0787,
      "step": 183786
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.3793401718139648,
      "learning_rate": 5.858865844770019e-05,
      "loss": 2.0464,
      "step": 183787
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.3662760257720947,
      "learning_rate": 5.858623000320739e-05,
      "loss": 1.9253,
      "step": 183788
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.5186704397201538,
      "learning_rate": 5.858380160359789e-05,
      "loss": 1.8389,
      "step": 183789
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.3575313091278076,
      "learning_rate": 5.858137324887239e-05,
      "loss": 2.1121,
      "step": 183790
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.4774410724639893,
      "learning_rate": 5.8578944939031015e-05,
      "loss": 1.9354,
      "step": 183791
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.278684377670288,
      "learning_rate": 5.857651667407444e-05,
      "loss": 2.0921,
      "step": 183792
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.469834566116333,
      "learning_rate": 5.8574088454002984e-05,
      "loss": 2.0205,
      "step": 183793
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.6774334907531738,
      "learning_rate": 5.8571660278817236e-05,
      "loss": 1.9448,
      "step": 183794
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.3269134759902954,
      "learning_rate": 5.856923214851748e-05,
      "loss": 1.9938,
      "step": 183795
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.4324074983596802,
      "learning_rate": 5.8566804063104375e-05,
      "loss": 2.0454,
      "step": 183796
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.3547943830490112,
      "learning_rate": 5.856437602257821e-05,
      "loss": 1.874,
      "step": 183797
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.4855680465698242,
      "learning_rate": 5.8561948026939534e-05,
      "loss": 1.9897,
      "step": 183798
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.5462210178375244,
      "learning_rate": 5.855952007618865e-05,
      "loss": 1.7334,
      "step": 183799
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.2817462682724,
      "learning_rate": 5.8557092170326156e-05,
      "loss": 1.8558,
      "step": 183800
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.4070574045181274,
      "learning_rate": 5.855466430935243e-05,
      "loss": 2.0908,
      "step": 183801
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.302424669265747,
      "learning_rate": 5.855223649326799e-05,
      "loss": 1.8865,
      "step": 183802
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.6367555856704712,
      "learning_rate": 5.854980872207324e-05,
      "loss": 1.99,
      "step": 183803
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.2394858598709106,
      "learning_rate": 5.854738099576869e-05,
      "loss": 2.287,
      "step": 183804
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.3100378513336182,
      "learning_rate": 5.854495331435463e-05,
      "loss": 2.1343,
      "step": 183805
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.2343683242797852,
      "learning_rate": 5.8542525677831684e-05,
      "loss": 1.8684,
      "step": 183806
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.43845534324646,
      "learning_rate": 5.854009808620017e-05,
      "loss": 2.0351,
      "step": 183807
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.2165347337722778,
      "learning_rate": 5.853767053946071e-05,
      "loss": 2.0773,
      "step": 183808
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.3637776374816895,
      "learning_rate": 5.853524303761364e-05,
      "loss": 1.9201,
      "step": 183809
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.3743339776992798,
      "learning_rate": 5.853281558065933e-05,
      "loss": 1.9883,
      "step": 183810
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.3832241296768188,
      "learning_rate": 5.8530388168598444e-05,
      "loss": 2.1247,
      "step": 183811
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.3062098026275635,
      "learning_rate": 5.852796080143127e-05,
      "loss": 1.8886,
      "step": 183812
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.336987853050232,
      "learning_rate": 5.852553347915826e-05,
      "loss": 1.8988,
      "step": 183813
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.4076684713363647,
      "learning_rate": 5.8523106201780004e-05,
      "loss": 2.2095,
      "step": 183814
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.4143794775009155,
      "learning_rate": 5.852067896929683e-05,
      "loss": 2.0548,
      "step": 183815
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.2512675523757935,
      "learning_rate": 5.851825178170915e-05,
      "loss": 2.068,
      "step": 183816
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.3110607862472534,
      "learning_rate": 5.8515824639017585e-05,
      "loss": 2.2038,
      "step": 183817
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.3720240592956543,
      "learning_rate": 5.8513397541222374e-05,
      "loss": 1.9778,
      "step": 183818
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.2811700105667114,
      "learning_rate": 5.8510970488324185e-05,
      "loss": 2.0764,
      "step": 183819
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.3571339845657349,
      "learning_rate": 5.850854348032334e-05,
      "loss": 1.9931,
      "step": 183820
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.3489717245101929,
      "learning_rate": 5.850611651722032e-05,
      "loss": 1.961,
      "step": 183821
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.4549089670181274,
      "learning_rate": 5.8503689599015524e-05,
      "loss": 2.021,
      "step": 183822
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.4259716272354126,
      "learning_rate": 5.850126272570951e-05,
      "loss": 1.8099,
      "step": 183823
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.2979084253311157,
      "learning_rate": 5.8498835897302576e-05,
      "loss": 2.0455,
      "step": 183824
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.4679532051086426,
      "learning_rate": 5.849640911379536e-05,
      "loss": 2.2219,
      "step": 183825
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.543365716934204,
      "learning_rate": 5.849398237518823e-05,
      "loss": 2.0544,
      "step": 183826
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.4566599130630493,
      "learning_rate": 5.8491555681481616e-05,
      "loss": 2.1001,
      "step": 183827
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.4082649946212769,
      "learning_rate": 5.848912903267592e-05,
      "loss": 2.0653,
      "step": 183828
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.3386766910552979,
      "learning_rate": 5.8486702428771706e-05,
      "loss": 1.9023,
      "step": 183829
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.4570767879486084,
      "learning_rate": 5.84842758697693e-05,
      "loss": 2.1052,
      "step": 183830
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.408860445022583,
      "learning_rate": 5.848184935566932e-05,
      "loss": 2.3768,
      "step": 183831
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.5876675844192505,
      "learning_rate": 5.8479422886472016e-05,
      "loss": 1.8114,
      "step": 183832
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.325870156288147,
      "learning_rate": 5.8476996462178125e-05,
      "loss": 2.1432,
      "step": 183833
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.5023863315582275,
      "learning_rate": 5.8474570082787756e-05,
      "loss": 2.0585,
      "step": 183834
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.3152483701705933,
      "learning_rate": 5.8472143748301625e-05,
      "loss": 2.0549,
      "step": 183835
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.4132969379425049,
      "learning_rate": 5.8469717458719976e-05,
      "loss": 1.9565,
      "step": 183836
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.4679049253463745,
      "learning_rate": 5.846729121404344e-05,
      "loss": 2.1653,
      "step": 183837
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.3329392671585083,
      "learning_rate": 5.846486501427232e-05,
      "loss": 1.9671,
      "step": 183838
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.5536293983459473,
      "learning_rate": 5.8462438859407314e-05,
      "loss": 2.0263,
      "step": 183839
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.3139785528182983,
      "learning_rate": 5.846001274944855e-05,
      "loss": 1.9391,
      "step": 183840
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.6786524057388306,
      "learning_rate": 5.845758668439666e-05,
      "loss": 2.0383,
      "step": 183841
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.2894893884658813,
      "learning_rate": 5.845516066425202e-05,
      "loss": 2.1177,
      "step": 183842
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.3623191118240356,
      "learning_rate": 5.845273468901522e-05,
      "loss": 1.9525,
      "step": 183843
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.4091920852661133,
      "learning_rate": 5.84503087586865e-05,
      "loss": 1.9061,
      "step": 183844
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.3459051847457886,
      "learning_rate": 5.844788287326659e-05,
      "loss": 2.2803,
      "step": 183845
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.69234037399292,
      "learning_rate": 5.844545703275566e-05,
      "loss": 1.8778,
      "step": 183846
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.3269842863082886,
      "learning_rate": 5.844303123715434e-05,
      "loss": 1.9831,
      "step": 183847
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.3193761110305786,
      "learning_rate": 5.844060548646292e-05,
      "loss": 1.8857,
      "step": 183848
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.3099926710128784,
      "learning_rate": 5.8438179780682046e-05,
      "loss": 1.8253,
      "step": 183849
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.3237156867980957,
      "learning_rate": 5.8435754119811985e-05,
      "loss": 1.8731,
      "step": 183850
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.3606513738632202,
      "learning_rate": 5.843332850385343e-05,
      "loss": 2.0041,
      "step": 183851
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.7751532793045044,
      "learning_rate": 5.843090293280655e-05,
      "loss": 1.9647,
      "step": 183852
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.4265044927597046,
      "learning_rate": 5.842847740667197e-05,
      "loss": 1.9803,
      "step": 183853
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.342218041419983,
      "learning_rate": 5.8426051925450037e-05,
      "loss": 2.1543,
      "step": 183854
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.193725824356079,
      "learning_rate": 5.842362648914134e-05,
      "loss": 2.1976,
      "step": 183855
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.4924256801605225,
      "learning_rate": 5.8421201097746184e-05,
      "loss": 1.9964,
      "step": 183856
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.483697533607483,
      "learning_rate": 5.8418775751265236e-05,
      "loss": 2.174,
      "step": 183857
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.4021151065826416,
      "learning_rate": 5.841635044969862e-05,
      "loss": 2.069,
      "step": 183858
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.3559410572052002,
      "learning_rate": 5.841392519304705e-05,
      "loss": 1.9401,
      "step": 183859
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.48159658908844,
      "learning_rate": 5.8411499981310816e-05,
      "loss": 1.9593,
      "step": 183860
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.4355300664901733,
      "learning_rate": 5.840907481449052e-05,
      "loss": 2.0387,
      "step": 183861
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.2544643878936768,
      "learning_rate": 5.8406649692586496e-05,
      "loss": 2.0519,
      "step": 183862
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.5161995887756348,
      "learning_rate": 5.840422461559934e-05,
      "loss": 2.1571,
      "step": 183863
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.602289080619812,
      "learning_rate": 5.840179958352928e-05,
      "loss": 2.1655,
      "step": 183864
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.4204485416412354,
      "learning_rate": 5.8399374596376936e-05,
      "loss": 2.1068,
      "step": 183865
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.3997098207473755,
      "learning_rate": 5.839694965414266e-05,
      "loss": 1.949,
      "step": 183866
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.3142319917678833,
      "learning_rate": 5.839452475682701e-05,
      "loss": 2.129,
      "step": 183867
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.3965059518814087,
      "learning_rate": 5.83920999044303e-05,
      "loss": 1.9431,
      "step": 183868
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.2984745502471924,
      "learning_rate": 5.838967509695323e-05,
      "loss": 1.8099,
      "step": 183869
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.4872491359710693,
      "learning_rate": 5.838725033439592e-05,
      "loss": 2.1805,
      "step": 183870
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.469950795173645,
      "learning_rate": 5.838482561675904e-05,
      "loss": 2.1104,
      "step": 183871
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.460517406463623,
      "learning_rate": 5.8382400944042904e-05,
      "loss": 2.0273,
      "step": 183872
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.4676284790039062,
      "learning_rate": 5.8379976316248134e-05,
      "loss": 1.9425,
      "step": 183873
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.4223936796188354,
      "learning_rate": 5.8377551733375025e-05,
      "loss": 1.8909,
      "step": 183874
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.4530503749847412,
      "learning_rate": 5.837512719542422e-05,
      "loss": 1.9534,
      "step": 183875
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.3922237157821655,
      "learning_rate": 5.837270270239588e-05,
      "loss": 2.0129,
      "step": 183876
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.4559612274169922,
      "learning_rate": 5.83702782542907e-05,
      "loss": 2.1959,
      "step": 183877
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.2758816480636597,
      "learning_rate": 5.836785385110899e-05,
      "loss": 1.9652,
      "step": 183878
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.324513554573059,
      "learning_rate": 5.836542949285131e-05,
      "loss": 1.9591,
      "step": 183879
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.3910385370254517,
      "learning_rate": 5.8363005179518e-05,
      "loss": 2.1348,
      "step": 183880
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.2674390077590942,
      "learning_rate": 5.8360580911109645e-05,
      "loss": 2.1289,
      "step": 183881
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.303043007850647,
      "learning_rate": 5.835815668762659e-05,
      "loss": 2.0397,
      "step": 183882
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.6771228313446045,
      "learning_rate": 5.835573250906933e-05,
      "loss": 2.1574,
      "step": 183883
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.6114816665649414,
      "learning_rate": 5.835330837543824e-05,
      "loss": 1.938,
      "step": 183884
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.3446842432022095,
      "learning_rate": 5.83508842867339e-05,
      "loss": 2.127,
      "step": 183885
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.3546687364578247,
      "learning_rate": 5.83484602429566e-05,
      "loss": 2.1273,
      "step": 183886
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.4985724687576294,
      "learning_rate": 5.8346036244106986e-05,
      "loss": 2.0745,
      "step": 183887
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.484526515007019,
      "learning_rate": 5.834361229018537e-05,
      "loss": 2.1203,
      "step": 183888
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.6417137384414673,
      "learning_rate": 5.834118838119228e-05,
      "loss": 2.0683,
      "step": 183889
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.3314814567565918,
      "learning_rate": 5.833876451712802e-05,
      "loss": 2.1019,
      "step": 183890
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.3362057209014893,
      "learning_rate": 5.833634069799322e-05,
      "loss": 1.8418,
      "step": 183891
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.4273639917373657,
      "learning_rate": 5.833391692378822e-05,
      "loss": 1.9072,
      "step": 183892
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.2953476905822754,
      "learning_rate": 5.833149319451354e-05,
      "loss": 2.0589,
      "step": 183893
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.3968236446380615,
      "learning_rate": 5.832906951016961e-05,
      "loss": 2.0947,
      "step": 183894
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.4801255464553833,
      "learning_rate": 5.832664587075678e-05,
      "loss": 2.2553,
      "step": 183895
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.5166810750961304,
      "learning_rate": 5.832422227627568e-05,
      "loss": 1.9951,
      "step": 183896
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.495621919631958,
      "learning_rate": 5.8321798726726665e-05,
      "loss": 1.934,
      "step": 183897
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.4844768047332764,
      "learning_rate": 5.831937522211011e-05,
      "loss": 1.9938,
      "step": 183898
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.26394522190094,
      "learning_rate": 5.831695176242661e-05,
      "loss": 2.0596,
      "step": 183899
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.279049277305603,
      "learning_rate": 5.831452834767657e-05,
      "loss": 1.8959,
      "step": 183900
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.413266658782959,
      "learning_rate": 5.8312104977860355e-05,
      "loss": 2.2137,
      "step": 183901
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.3664301633834839,
      "learning_rate": 5.830968165297852e-05,
      "loss": 1.7751,
      "step": 183902
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.3395179510116577,
      "learning_rate": 5.830725837303145e-05,
      "loss": 1.9375,
      "step": 183903
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.3736244440078735,
      "learning_rate": 5.830483513801967e-05,
      "loss": 2.0784,
      "step": 183904
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.230024814605713,
      "learning_rate": 5.830241194794357e-05,
      "loss": 1.8903,
      "step": 183905
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.4193695783615112,
      "learning_rate": 5.829998880280363e-05,
      "loss": 2.1333,
      "step": 183906
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.5252641439437866,
      "learning_rate": 5.8297565702600216e-05,
      "loss": 1.8954,
      "step": 183907
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.3664768934249878,
      "learning_rate": 5.829514264733392e-05,
      "loss": 2.0288,
      "step": 183908
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.3382138013839722,
      "learning_rate": 5.829271963700503e-05,
      "loss": 1.821,
      "step": 183909
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.3163455724716187,
      "learning_rate": 5.829029667161418e-05,
      "loss": 1.9417,
      "step": 183910
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.3529653549194336,
      "learning_rate": 5.828787375116171e-05,
      "loss": 2.0697,
      "step": 183911
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.4199470281600952,
      "learning_rate": 5.828545087564812e-05,
      "loss": 1.9626,
      "step": 183912
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.3021879196166992,
      "learning_rate": 5.828302804507372e-05,
      "loss": 2.0064,
      "step": 183913
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.2982736825942993,
      "learning_rate": 5.828060525943914e-05,
      "loss": 1.9468,
      "step": 183914
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.2614996433258057,
      "learning_rate": 5.8278182518744706e-05,
      "loss": 1.6777,
      "step": 183915
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.5689204931259155,
      "learning_rate": 5.8275759822990964e-05,
      "loss": 2.1331,
      "step": 183916
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.352506160736084,
      "learning_rate": 5.827333717217827e-05,
      "loss": 2.0263,
      "step": 183917
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.4873929023742676,
      "learning_rate": 5.827091456630729e-05,
      "loss": 2.0652,
      "step": 183918
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.583964228630066,
      "learning_rate": 5.826849200537813e-05,
      "loss": 2.0434,
      "step": 183919
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.6714640855789185,
      "learning_rate": 5.826606948939152e-05,
      "loss": 1.9444,
      "step": 183920
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.3732762336730957,
      "learning_rate": 5.826364701834769e-05,
      "loss": 2.0978,
      "step": 183921
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.372539758682251,
      "learning_rate": 5.8261224592247344e-05,
      "loss": 1.9728,
      "step": 183922
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.386946678161621,
      "learning_rate": 5.8258802211090684e-05,
      "loss": 1.9219,
      "step": 183923
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.4636996984481812,
      "learning_rate": 5.825637987487843e-05,
      "loss": 2.1538,
      "step": 183924
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.4549331665039062,
      "learning_rate": 5.8253957583610767e-05,
      "loss": 1.9173,
      "step": 183925
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.3973453044891357,
      "learning_rate": 5.825153533728828e-05,
      "loss": 2.0432,
      "step": 183926
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.2307261228561401,
      "learning_rate": 5.824911313591134e-05,
      "loss": 2.1558,
      "step": 183927
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.2441232204437256,
      "learning_rate": 5.824669097948055e-05,
      "loss": 2.1192,
      "step": 183928
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.3067152500152588,
      "learning_rate": 5.824426886799617e-05,
      "loss": 2.167,
      "step": 183929
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.282339096069336,
      "learning_rate": 5.82418468014589e-05,
      "loss": 2.0641,
      "step": 183930
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.2043054103851318,
      "learning_rate": 5.823942477986888e-05,
      "loss": 2.0696,
      "step": 183931
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.2812342643737793,
      "learning_rate": 5.823700280322677e-05,
      "loss": 2.0863,
      "step": 183932
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.38186514377594,
      "learning_rate": 5.823458087153291e-05,
      "loss": 1.9054,
      "step": 183933
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.5014606714248657,
      "learning_rate": 5.8232158984787894e-05,
      "loss": 2.1221,
      "step": 183934
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.4750021696090698,
      "learning_rate": 5.8229737142991986e-05,
      "loss": 2.0785,
      "step": 183935
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.4731636047363281,
      "learning_rate": 5.822731534614589e-05,
      "loss": 1.9929,
      "step": 183936
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.4034392833709717,
      "learning_rate": 5.822489359424977e-05,
      "loss": 1.969,
      "step": 183937
    },
    {
      "epoch": 2.39,
      "grad_norm": 1.3804144859313965,
      "learning_rate": 5.8222471887304256e-05,
      "loss": 1.8751,
      "step": 183938
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.3702480792999268,
      "learning_rate": 5.822005022530966e-05,
      "loss": 1.9249,
      "step": 183939
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.456992268562317,
      "learning_rate": 5.821762860826663e-05,
      "loss": 2.1488,
      "step": 183940
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.4378820657730103,
      "learning_rate": 5.821520703617541e-05,
      "loss": 1.8025,
      "step": 183941
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.390967607498169,
      "learning_rate": 5.8212785509036706e-05,
      "loss": 1.9905,
      "step": 183942
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.4527013301849365,
      "learning_rate": 5.821036402685064e-05,
      "loss": 1.9988,
      "step": 183943
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.3591166734695435,
      "learning_rate": 5.820794258961795e-05,
      "loss": 2.1384,
      "step": 183944
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.4345141649246216,
      "learning_rate": 5.8205521197338864e-05,
      "loss": 2.0803,
      "step": 183945
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.6473196744918823,
      "learning_rate": 5.820309985001399e-05,
      "loss": 1.8897,
      "step": 183946
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.5269944667816162,
      "learning_rate": 5.820067854764368e-05,
      "loss": 1.9357,
      "step": 183947
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.499902606010437,
      "learning_rate": 5.819825729022859e-05,
      "loss": 1.9355,
      "step": 183948
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.3033469915390015,
      "learning_rate": 5.8195836077768834e-05,
      "loss": 2.0576,
      "step": 183949
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.6642844676971436,
      "learning_rate": 5.819341491026511e-05,
      "loss": 2.0936,
      "step": 183950
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.7179185152053833,
      "learning_rate": 5.819099378771773e-05,
      "loss": 2.1057,
      "step": 183951
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.5094660520553589,
      "learning_rate": 5.818857271012726e-05,
      "loss": 2.1178,
      "step": 183952
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.5307515859603882,
      "learning_rate": 5.818615167749405e-05,
      "loss": 2.087,
      "step": 183953
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.364491581916809,
      "learning_rate": 5.818373068981875e-05,
      "loss": 1.9722,
      "step": 183954
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.218845248222351,
      "learning_rate": 5.818130974710149e-05,
      "loss": 2.0806,
      "step": 183955
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.5480546951293945,
      "learning_rate": 5.8178888849342996e-05,
      "loss": 2.0172,
      "step": 183956
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.5202174186706543,
      "learning_rate": 5.817646799654351e-05,
      "loss": 2.0687,
      "step": 183957
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.555301547050476,
      "learning_rate": 5.817404718870365e-05,
      "loss": 2.0162,
      "step": 183958
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.2413655519485474,
      "learning_rate": 5.817162642582374e-05,
      "loss": 2.0441,
      "step": 183959
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.3447046279907227,
      "learning_rate": 5.816920570790445e-05,
      "loss": 1.97,
      "step": 183960
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.4653236865997314,
      "learning_rate": 5.816678503494591e-05,
      "loss": 1.8553,
      "step": 183961
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.4664967060089111,
      "learning_rate": 5.816436440694879e-05,
      "loss": 1.9713,
      "step": 183962
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.4948534965515137,
      "learning_rate": 5.8161943823913414e-05,
      "loss": 1.9089,
      "step": 183963
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.4496201276779175,
      "learning_rate": 5.815952328584037e-05,
      "loss": 2.0292,
      "step": 183964
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.338769555091858,
      "learning_rate": 5.8157102792729995e-05,
      "loss": 2.0202,
      "step": 183965
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.2652753591537476,
      "learning_rate": 5.815468234458281e-05,
      "loss": 1.8681,
      "step": 183966
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.4155769348144531,
      "learning_rate": 5.815226194139927e-05,
      "loss": 2.1579,
      "step": 183967
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.423443078994751,
      "learning_rate": 5.8149841583179754e-05,
      "loss": 1.9932,
      "step": 183968
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.2936731576919556,
      "learning_rate": 5.8147421269924676e-05,
      "loss": 1.8455,
      "step": 183969
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.3693701028823853,
      "learning_rate": 5.814500100163466e-05,
      "loss": 2.2039,
      "step": 183970
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.4959471225738525,
      "learning_rate": 5.814258077830998e-05,
      "loss": 2.0744,
      "step": 183971
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.3255820274353027,
      "learning_rate": 5.8140160599951194e-05,
      "loss": 1.8191,
      "step": 183972
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.526039719581604,
      "learning_rate": 5.813774046655874e-05,
      "loss": 2.0447,
      "step": 183973
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.6768542528152466,
      "learning_rate": 5.813532037813308e-05,
      "loss": 1.8374,
      "step": 183974
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.4835970401763916,
      "learning_rate": 5.8132900334674515e-05,
      "loss": 2.0459,
      "step": 183975
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.579781174659729,
      "learning_rate": 5.8130480336183684e-05,
      "loss": 1.8826,
      "step": 183976
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.4348137378692627,
      "learning_rate": 5.812806038266088e-05,
      "loss": 2.1826,
      "step": 183977
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.426537275314331,
      "learning_rate": 5.812564047410674e-05,
      "loss": 1.7794,
      "step": 183978
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.6141479015350342,
      "learning_rate": 5.812322061052159e-05,
      "loss": 2.052,
      "step": 183979
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.382775902748108,
      "learning_rate": 5.812080079190584e-05,
      "loss": 2.1223,
      "step": 183980
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.3954027891159058,
      "learning_rate": 5.811838101826005e-05,
      "loss": 2.046,
      "step": 183981
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.176488995552063,
      "learning_rate": 5.8115961289584624e-05,
      "loss": 1.9538,
      "step": 183982
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.4506118297576904,
      "learning_rate": 5.8113541605879954e-05,
      "loss": 2.0423,
      "step": 183983
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.3291277885437012,
      "learning_rate": 5.8111121967146614e-05,
      "loss": 1.9427,
      "step": 183984
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.3814648389816284,
      "learning_rate": 5.8108702373384964e-05,
      "loss": 1.9327,
      "step": 183985
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.7090357542037964,
      "learning_rate": 5.8106282824595404e-05,
      "loss": 2.075,
      "step": 183986
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.3622909784317017,
      "learning_rate": 5.810386332077851e-05,
      "loss": 2.1749,
      "step": 183987
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.3605648279190063,
      "learning_rate": 5.8101443861934706e-05,
      "loss": 2.0254,
      "step": 183988
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.4387058019638062,
      "learning_rate": 5.809902444806432e-05,
      "loss": 2.0342,
      "step": 183989
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.3397972583770752,
      "learning_rate": 5.809660507916797e-05,
      "loss": 2.0862,
      "step": 183990
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.4074352979660034,
      "learning_rate": 5.8094185755246034e-05,
      "loss": 2.0993,
      "step": 183991
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.5101089477539062,
      "learning_rate": 5.809176647629886e-05,
      "loss": 1.9857,
      "step": 183992
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.52143132686615,
      "learning_rate": 5.8089347242327075e-05,
      "loss": 1.8312,
      "step": 183993
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.3332515954971313,
      "learning_rate": 5.808692805333098e-05,
      "loss": 1.8737,
      "step": 183994
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.3531287908554077,
      "learning_rate": 5.808450890931118e-05,
      "loss": 2.061,
      "step": 183995
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.373421311378479,
      "learning_rate": 5.8082089810268e-05,
      "loss": 1.8977,
      "step": 183996
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.2824993133544922,
      "learning_rate": 5.8079670756201935e-05,
      "loss": 2.0786,
      "step": 183997
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.2957559823989868,
      "learning_rate": 5.807725174711336e-05,
      "loss": 1.8599,
      "step": 183998
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.3224700689315796,
      "learning_rate": 5.8074832783002877e-05,
      "loss": 2.1165,
      "step": 183999
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.5489413738250732,
      "learning_rate": 5.807241386387075e-05,
      "loss": 1.9813,
      "step": 184000
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.2679204940795898,
      "learning_rate": 5.806999498971764e-05,
      "loss": 1.8765,
      "step": 184001
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.4247300624847412,
      "learning_rate": 5.8067576160543786e-05,
      "loss": 2.1539,
      "step": 184002
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.3826701641082764,
      "learning_rate": 5.806515737634988e-05,
      "loss": 1.7357,
      "step": 184003
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.2660870552062988,
      "learning_rate": 5.8062738637136095e-05,
      "loss": 1.9356,
      "step": 184004
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.3964769840240479,
      "learning_rate": 5.8060319942903066e-05,
      "loss": 1.8898,
      "step": 184005
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.3782213926315308,
      "learning_rate": 5.805790129365112e-05,
      "loss": 1.9994,
      "step": 184006
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.381980299949646,
      "learning_rate": 5.805548268938086e-05,
      "loss": 2.0967,
      "step": 184007
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.2562446594238281,
      "learning_rate": 5.805306413009259e-05,
      "loss": 1.8665,
      "step": 184008
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.333467721939087,
      "learning_rate": 5.805064561578696e-05,
      "loss": 1.8219,
      "step": 184009
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.4034820795059204,
      "learning_rate": 5.8048227146464124e-05,
      "loss": 2.0141,
      "step": 184010
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.3550548553466797,
      "learning_rate": 5.804580872212477e-05,
      "loss": 2.0308,
      "step": 184011
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.3239569664001465,
      "learning_rate": 5.804339034276919e-05,
      "loss": 1.9698,
      "step": 184012
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.363226294517517,
      "learning_rate": 5.8040972008398e-05,
      "loss": 2.1316,
      "step": 184013
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.3474498987197876,
      "learning_rate": 5.8038553719011495e-05,
      "loss": 2.0709,
      "step": 184014
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.4156724214553833,
      "learning_rate": 5.80361354746103e-05,
      "loss": 1.8722,
      "step": 184015
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.5613912343978882,
      "learning_rate": 5.803371727519463e-05,
      "loss": 1.9824,
      "step": 184016
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.3905980587005615,
      "learning_rate": 5.803129912076513e-05,
      "loss": 2.1828,
      "step": 184017
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.3262038230895996,
      "learning_rate": 5.802888101132212e-05,
      "loss": 2.2482,
      "step": 184018
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.2384536266326904,
      "learning_rate": 5.802646294686615e-05,
      "loss": 1.9668,
      "step": 184019
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.5279600620269775,
      "learning_rate": 5.8024044927397596e-05,
      "loss": 2.1618,
      "step": 184020
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.3557255268096924,
      "learning_rate": 5.802162695291706e-05,
      "loss": 1.8301,
      "step": 184021
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.4368036985397339,
      "learning_rate": 5.801920902342473e-05,
      "loss": 2.0957,
      "step": 184022
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.4223965406417847,
      "learning_rate": 5.801679113892125e-05,
      "loss": 1.9695,
      "step": 184023
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.5078896284103394,
      "learning_rate": 5.801437329940698e-05,
      "loss": 2.035,
      "step": 184024
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.3999043703079224,
      "learning_rate": 5.8011955504882455e-05,
      "loss": 1.9817,
      "step": 184025
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.4035780429840088,
      "learning_rate": 5.800953775534801e-05,
      "loss": 2.0041,
      "step": 184026
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.3943681716918945,
      "learning_rate": 5.8007120050804315e-05,
      "loss": 2.1085,
      "step": 184027
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.3537707328796387,
      "learning_rate": 5.800470239125149e-05,
      "loss": 1.8957,
      "step": 184028
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.2455689907073975,
      "learning_rate": 5.8002284776690244e-05,
      "loss": 1.9477,
      "step": 184029
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.4125910997390747,
      "learning_rate": 5.799986720712088e-05,
      "loss": 2.2353,
      "step": 184030
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.3214390277862549,
      "learning_rate": 5.799744968254398e-05,
      "loss": 2.0491,
      "step": 184031
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.449560523033142,
      "learning_rate": 5.799503220295983e-05,
      "loss": 1.7612,
      "step": 184032
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.354675054550171,
      "learning_rate": 5.799261476836906e-05,
      "loss": 2.0462,
      "step": 184033
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.3042269945144653,
      "learning_rate": 5.799019737877203e-05,
      "loss": 2.118,
      "step": 184034
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.328783631324768,
      "learning_rate": 5.798778003416921e-05,
      "loss": 2.005,
      "step": 184035
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.343846082687378,
      "learning_rate": 5.798536273456092e-05,
      "loss": 2.0496,
      "step": 184036
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.4072115421295166,
      "learning_rate": 5.798294547994782e-05,
      "loss": 2.1003,
      "step": 184037
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.365709900856018,
      "learning_rate": 5.798052827033019e-05,
      "loss": 2.1391,
      "step": 184038
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.4113496541976929,
      "learning_rate": 5.797811110570859e-05,
      "loss": 2.1824,
      "step": 184039
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.374537706375122,
      "learning_rate": 5.7975693986083437e-05,
      "loss": 2.0307,
      "step": 184040
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.4690812826156616,
      "learning_rate": 5.7973276911455194e-05,
      "loss": 2.0632,
      "step": 184041
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.3995418548583984,
      "learning_rate": 5.797085988182418e-05,
      "loss": 2.1185,
      "step": 184042
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.4329758882522583,
      "learning_rate": 5.796844289719105e-05,
      "loss": 1.8679,
      "step": 184043
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.3521645069122314,
      "learning_rate": 5.7966025957556085e-05,
      "loss": 1.8402,
      "step": 184044
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.5627671480178833,
      "learning_rate": 5.796360906291986e-05,
      "loss": 1.7905,
      "step": 184045
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.2567017078399658,
      "learning_rate": 5.79611922132828e-05,
      "loss": 1.8412,
      "step": 184046
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.306875228881836,
      "learning_rate": 5.795877540864529e-05,
      "loss": 1.8291,
      "step": 184047
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.429741382598877,
      "learning_rate": 5.795635864900774e-05,
      "loss": 2.0771,
      "step": 184048
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.279708981513977,
      "learning_rate": 5.7953941934370765e-05,
      "loss": 1.8946,
      "step": 184049
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.6244531869888306,
      "learning_rate": 5.795152526473462e-05,
      "loss": 1.8786,
      "step": 184050
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.4954712390899658,
      "learning_rate": 5.794910864009995e-05,
      "loss": 2.0712,
      "step": 184051
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.4129914045333862,
      "learning_rate": 5.794669206046712e-05,
      "loss": 2.1716,
      "step": 184052
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.6974951028823853,
      "learning_rate": 5.794427552583655e-05,
      "loss": 1.8516,
      "step": 184053
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.3945538997650146,
      "learning_rate": 5.794185903620865e-05,
      "loss": 2.1561,
      "step": 184054
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.4423893690109253,
      "learning_rate": 5.793944259158398e-05,
      "loss": 2.0903,
      "step": 184055
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.5705289840698242,
      "learning_rate": 5.793702619196289e-05,
      "loss": 1.8452,
      "step": 184056
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.4843780994415283,
      "learning_rate": 5.793460983734592e-05,
      "loss": 2.1409,
      "step": 184057
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.3444350957870483,
      "learning_rate": 5.793219352773349e-05,
      "loss": 1.8829,
      "step": 184058
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.6106988191604614,
      "learning_rate": 5.792977726312607e-05,
      "loss": 1.9867,
      "step": 184059
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.2369869947433472,
      "learning_rate": 5.792736104352393e-05,
      "loss": 1.7855,
      "step": 184060
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.3877182006835938,
      "learning_rate": 5.7924944868927806e-05,
      "loss": 2.0623,
      "step": 184061
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.46466863155365,
      "learning_rate": 5.792252873933787e-05,
      "loss": 2.0371,
      "step": 184062
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.2815223932266235,
      "learning_rate": 5.79201126547548e-05,
      "loss": 1.9076,
      "step": 184063
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.3937854766845703,
      "learning_rate": 5.791769661517897e-05,
      "loss": 1.9043,
      "step": 184064
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.3329970836639404,
      "learning_rate": 5.791528062061073e-05,
      "loss": 1.9821,
      "step": 184065
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.3559093475341797,
      "learning_rate": 5.7912864671050686e-05,
      "loss": 1.8565,
      "step": 184066
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.4654120206832886,
      "learning_rate": 5.7910448766499174e-05,
      "loss": 2.0879,
      "step": 184067
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.6783535480499268,
      "learning_rate": 5.790803290695663e-05,
      "loss": 1.9364,
      "step": 184068
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.3361080884933472,
      "learning_rate": 5.790561709242364e-05,
      "loss": 2.0235,
      "step": 184069
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.3510726690292358,
      "learning_rate": 5.790320132290055e-05,
      "loss": 2.1205,
      "step": 184070
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.5519260168075562,
      "learning_rate": 5.790078559838775e-05,
      "loss": 1.9306,
      "step": 184071
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.3192278146743774,
      "learning_rate": 5.7898369918885854e-05,
      "loss": 2.061,
      "step": 184072
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.4223923683166504,
      "learning_rate": 5.789595428439522e-05,
      "loss": 1.9277,
      "step": 184073
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.4167953729629517,
      "learning_rate": 5.7893538694916206e-05,
      "loss": 1.8921,
      "step": 184074
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.276085615158081,
      "learning_rate": 5.7891123150449425e-05,
      "loss": 2.1429,
      "step": 184075
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.3880103826522827,
      "learning_rate": 5.788870765099527e-05,
      "loss": 2.2439,
      "step": 184076
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.3604393005371094,
      "learning_rate": 5.7886292196554077e-05,
      "loss": 2.2684,
      "step": 184077
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.3354161977767944,
      "learning_rate": 5.7883876787126506e-05,
      "loss": 1.7548,
      "step": 184078
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.3903065919876099,
      "learning_rate": 5.788146142271279e-05,
      "loss": 2.3271,
      "step": 184079
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.3917365074157715,
      "learning_rate": 5.787904610331354e-05,
      "loss": 1.9579,
      "step": 184080
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.4731577634811401,
      "learning_rate": 5.787663082892918e-05,
      "loss": 2.0637,
      "step": 184081
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.2760792970657349,
      "learning_rate": 5.787421559956011e-05,
      "loss": 1.8002,
      "step": 184082
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.3549267053604126,
      "learning_rate": 5.787180041520674e-05,
      "loss": 1.9071,
      "step": 184083
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.3538806438446045,
      "learning_rate": 5.7869385275869616e-05,
      "loss": 1.9154,
      "step": 184084
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.2866624593734741,
      "learning_rate": 5.786697018154909e-05,
      "loss": 1.9867,
      "step": 184085
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.3115020990371704,
      "learning_rate": 5.7864555132245714e-05,
      "loss": 1.9616,
      "step": 184086
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.4049391746520996,
      "learning_rate": 5.78621401279599e-05,
      "loss": 1.9924,
      "step": 184087
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.3395397663116455,
      "learning_rate": 5.7859725168692106e-05,
      "loss": 1.9178,
      "step": 184088
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.3210792541503906,
      "learning_rate": 5.7857310254442666e-05,
      "loss": 1.8745,
      "step": 184089
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.2538589239120483,
      "learning_rate": 5.785489538521219e-05,
      "loss": 2.0817,
      "step": 184090
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.4506243467330933,
      "learning_rate": 5.7852480561000994e-05,
      "loss": 2.0196,
      "step": 184091
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.2895575761795044,
      "learning_rate": 5.7850065781809684e-05,
      "loss": 2.1184,
      "step": 184092
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.3429533243179321,
      "learning_rate": 5.784765104763853e-05,
      "loss": 2.1102,
      "step": 184093
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.461287260055542,
      "learning_rate": 5.78452363584882e-05,
      "loss": 2.0654,
      "step": 184094
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.5989853143692017,
      "learning_rate": 5.784282171435889e-05,
      "loss": 2.0636,
      "step": 184095
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.3576713800430298,
      "learning_rate": 5.784040711525123e-05,
      "loss": 2.0349,
      "step": 184096
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.4554672241210938,
      "learning_rate": 5.7837992561165523e-05,
      "loss": 2.1872,
      "step": 184097
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.2964125871658325,
      "learning_rate": 5.7835578052102405e-05,
      "loss": 1.8091,
      "step": 184098
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.373064637184143,
      "learning_rate": 5.783316358806214e-05,
      "loss": 1.8598,
      "step": 184099
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.463291049003601,
      "learning_rate": 5.783074916904532e-05,
      "loss": 2.1399,
      "step": 184100
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.3361213207244873,
      "learning_rate": 5.7828334795052354e-05,
      "loss": 2.1766,
      "step": 184101
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.3441580533981323,
      "learning_rate": 5.7825920466083673e-05,
      "loss": 2.0135,
      "step": 184102
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.3310022354125977,
      "learning_rate": 5.7823506182139645e-05,
      "loss": 1.9659,
      "step": 184103
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.363388180732727,
      "learning_rate": 5.782109194322087e-05,
      "loss": 2.1258,
      "step": 184104
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.3158576488494873,
      "learning_rate": 5.781867774932767e-05,
      "loss": 1.7977,
      "step": 184105
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.2039536237716675,
      "learning_rate": 5.781626360046059e-05,
      "loss": 1.9528,
      "step": 184106
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.3038170337677002,
      "learning_rate": 5.781384949662006e-05,
      "loss": 2.103,
      "step": 184107
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.447635293006897,
      "learning_rate": 5.7811435437806505e-05,
      "loss": 2.3275,
      "step": 184108
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.531867504119873,
      "learning_rate": 5.780902142402031e-05,
      "loss": 1.97,
      "step": 184109
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.6488755941390991,
      "learning_rate": 5.780660745526206e-05,
      "loss": 1.8901,
      "step": 184110
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.4999123811721802,
      "learning_rate": 5.780419353153205e-05,
      "loss": 2.0861,
      "step": 184111
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.3344053030014038,
      "learning_rate": 5.780177965283089e-05,
      "loss": 2.1192,
      "step": 184112
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.498668909072876,
      "learning_rate": 5.779936581915895e-05,
      "loss": 2.0354,
      "step": 184113
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.286940336227417,
      "learning_rate": 5.7796952030516684e-05,
      "loss": 1.8393,
      "step": 184114
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.4975345134735107,
      "learning_rate": 5.779453828690443e-05,
      "loss": 2.0538,
      "step": 184115
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.290406584739685,
      "learning_rate": 5.7792124588322865e-05,
      "loss": 1.9188,
      "step": 184116
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.3075528144836426,
      "learning_rate": 5.7789710934772205e-05,
      "loss": 2.2236,
      "step": 184117
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.3140509128570557,
      "learning_rate": 5.7787297326253124e-05,
      "loss": 2.0003,
      "step": 184118
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.3613799810409546,
      "learning_rate": 5.778488376276592e-05,
      "loss": 1.9184,
      "step": 184119
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.397133231163025,
      "learning_rate": 5.77824702443111e-05,
      "loss": 2.0773,
      "step": 184120
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.3086397647857666,
      "learning_rate": 5.7780056770889014e-05,
      "loss": 2.0203,
      "step": 184121
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.3112599849700928,
      "learning_rate": 5.7777643342500244e-05,
      "loss": 1.7515,
      "step": 184122
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.339117169380188,
      "learning_rate": 5.777522995914512e-05,
      "loss": 1.8452,
      "step": 184123
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.3268036842346191,
      "learning_rate": 5.7772816620824234e-05,
      "loss": 2.0387,
      "step": 184124
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.3729190826416016,
      "learning_rate": 5.7770403327537964e-05,
      "loss": 1.9963,
      "step": 184125
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.219869613647461,
      "learning_rate": 5.776799007928673e-05,
      "loss": 1.9385,
      "step": 184126
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.4393386840820312,
      "learning_rate": 5.7765576876070906e-05,
      "loss": 1.8894,
      "step": 184127
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.4812448024749756,
      "learning_rate": 5.776316371789113e-05,
      "loss": 2.0389,
      "step": 184128
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.3760074377059937,
      "learning_rate": 5.7760750604747685e-05,
      "loss": 1.878,
      "step": 184129
    },
    {
      "epoch": 2.4,
      "grad_norm": 2.226032257080078,
      "learning_rate": 5.775833753664115e-05,
      "loss": 2.1285,
      "step": 184130
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.4026856422424316,
      "learning_rate": 5.7755924513571925e-05,
      "loss": 2.0024,
      "step": 184131
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.455875039100647,
      "learning_rate": 5.775351153554044e-05,
      "loss": 1.9665,
      "step": 184132
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.3743634223937988,
      "learning_rate": 5.77510986025471e-05,
      "loss": 1.8876,
      "step": 184133
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.3739954233169556,
      "learning_rate": 5.774868571459246e-05,
      "loss": 2.1068,
      "step": 184134
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.3613427877426147,
      "learning_rate": 5.7746272871676825e-05,
      "loss": 2.3291,
      "step": 184135
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.2409799098968506,
      "learning_rate": 5.774386007380079e-05,
      "loss": 2.0765,
      "step": 184136
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.2738513946533203,
      "learning_rate": 5.7741447320964776e-05,
      "loss": 1.9941,
      "step": 184137
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.6482359170913696,
      "learning_rate": 5.773903461316919e-05,
      "loss": 1.9146,
      "step": 184138
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.385356068611145,
      "learning_rate": 5.773662195041444e-05,
      "loss": 2.0595,
      "step": 184139
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.3005255460739136,
      "learning_rate": 5.773420933270106e-05,
      "loss": 2.0504,
      "step": 184140
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.481111764907837,
      "learning_rate": 5.773179676002943e-05,
      "loss": 2.0331,
      "step": 184141
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.406406044960022,
      "learning_rate": 5.7729384232400057e-05,
      "loss": 1.9574,
      "step": 184142
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.378365397453308,
      "learning_rate": 5.772697174981339e-05,
      "loss": 2.07,
      "step": 184143
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.4387199878692627,
      "learning_rate": 5.772455931226986e-05,
      "loss": 2.1145,
      "step": 184144
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.431526780128479,
      "learning_rate": 5.772214691976983e-05,
      "loss": 2.1386,
      "step": 184145
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.3243355751037598,
      "learning_rate": 5.77197345723139e-05,
      "loss": 1.949,
      "step": 184146
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.3153092861175537,
      "learning_rate": 5.771732226990238e-05,
      "loss": 1.9608,
      "step": 184147
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.4569849967956543,
      "learning_rate": 5.771491001253582e-05,
      "loss": 2.0823,
      "step": 184148
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.331162691116333,
      "learning_rate": 5.771249780021466e-05,
      "loss": 1.9459,
      "step": 184149
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.3165092468261719,
      "learning_rate": 5.771008563293934e-05,
      "loss": 2.1602,
      "step": 184150
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.3885434865951538,
      "learning_rate": 5.770767351071018e-05,
      "loss": 2.0082,
      "step": 184151
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.3141719102859497,
      "learning_rate": 5.7705261433527825e-05,
      "loss": 2.0447,
      "step": 184152
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.3167287111282349,
      "learning_rate": 5.7702849401392535e-05,
      "loss": 2.022,
      "step": 184153
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.2489228248596191,
      "learning_rate": 5.770043741430498e-05,
      "loss": 2.0578,
      "step": 184154
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.3484348058700562,
      "learning_rate": 5.769802547226545e-05,
      "loss": 2.1987,
      "step": 184155
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.4660917520523071,
      "learning_rate": 5.769561357527439e-05,
      "loss": 2.0557,
      "step": 184156
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.6378846168518066,
      "learning_rate": 5.7693201723332325e-05,
      "loss": 2.0858,
      "step": 184157
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.4328842163085938,
      "learning_rate": 5.76907899164397e-05,
      "loss": 1.981,
      "step": 184158
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.3328063488006592,
      "learning_rate": 5.768837815459682e-05,
      "loss": 1.9731,
      "step": 184159
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.3713083267211914,
      "learning_rate": 5.768596643780435e-05,
      "loss": 2.0216,
      "step": 184160
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.409570574760437,
      "learning_rate": 5.768355476606265e-05,
      "loss": 1.9424,
      "step": 184161
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.210935115814209,
      "learning_rate": 5.768114313937203e-05,
      "loss": 1.977,
      "step": 184162
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.3024389743804932,
      "learning_rate": 5.767873155773318e-05,
      "loss": 1.9208,
      "step": 184163
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.5310324430465698,
      "learning_rate": 5.767632002114633e-05,
      "loss": 2.1009,
      "step": 184164
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.4839524030685425,
      "learning_rate": 5.76739085296121e-05,
      "loss": 1.8111,
      "step": 184165
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.2994030714035034,
      "learning_rate": 5.7671497083130866e-05,
      "loss": 2.0,
      "step": 184166
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.323244571685791,
      "learning_rate": 5.7669085681703e-05,
      "loss": 2.1402,
      "step": 184167
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.4966832399368286,
      "learning_rate": 5.766667432532909e-05,
      "loss": 1.9406,
      "step": 184168
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.4507735967636108,
      "learning_rate": 5.766426301400954e-05,
      "loss": 1.9116,
      "step": 184169
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.3838157653808594,
      "learning_rate": 5.7661851747744684e-05,
      "loss": 1.898,
      "step": 184170
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.6037734746932983,
      "learning_rate": 5.765944052653515e-05,
      "loss": 2.0648,
      "step": 184171
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.3374629020690918,
      "learning_rate": 5.765702935038131e-05,
      "loss": 2.0175,
      "step": 184172
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.3303245306015015,
      "learning_rate": 5.76546182192835e-05,
      "loss": 1.9473,
      "step": 184173
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.624242901802063,
      "learning_rate": 5.765220713324238e-05,
      "loss": 2.1568,
      "step": 184174
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.4620540142059326,
      "learning_rate": 5.7649796092258246e-05,
      "loss": 2.0603,
      "step": 184175
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.4488879442214966,
      "learning_rate": 5.7647385096331545e-05,
      "loss": 1.9304,
      "step": 184176
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.3827126026153564,
      "learning_rate": 5.764497414546286e-05,
      "loss": 2.0972,
      "step": 184177
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.4877904653549194,
      "learning_rate": 5.764256323965244e-05,
      "loss": 2.1306,
      "step": 184178
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.4406846761703491,
      "learning_rate": 5.764015237890094e-05,
      "loss": 2.1997,
      "step": 184179
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.796157956123352,
      "learning_rate": 5.76377415632087e-05,
      "loss": 2.0712,
      "step": 184180
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.4162474870681763,
      "learning_rate": 5.763533079257617e-05,
      "loss": 2.1444,
      "step": 184181
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.3828448057174683,
      "learning_rate": 5.763292006700374e-05,
      "loss": 2.1209,
      "step": 184182
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.444112777709961,
      "learning_rate": 5.763050938649203e-05,
      "loss": 1.9356,
      "step": 184183
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.4386483430862427,
      "learning_rate": 5.762809875104127e-05,
      "loss": 2.2012,
      "step": 184184
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.4946659803390503,
      "learning_rate": 5.7625688160652104e-05,
      "loss": 2.0402,
      "step": 184185
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.3714158535003662,
      "learning_rate": 5.7623277615324894e-05,
      "loss": 1.8738,
      "step": 184186
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.4354802370071411,
      "learning_rate": 5.76208671150601e-05,
      "loss": 1.9664,
      "step": 184187
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.3144972324371338,
      "learning_rate": 5.761845665985809e-05,
      "loss": 1.9679,
      "step": 184188
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.3308684825897217,
      "learning_rate": 5.761604624971947e-05,
      "loss": 2.0892,
      "step": 184189
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.3044904470443726,
      "learning_rate": 5.7613635884644506e-05,
      "loss": 1.9919,
      "step": 184190
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.4071986675262451,
      "learning_rate": 5.761122556463382e-05,
      "loss": 2.0232,
      "step": 184191
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.5916823148727417,
      "learning_rate": 5.760881528968779e-05,
      "loss": 2.2282,
      "step": 184192
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.3380087614059448,
      "learning_rate": 5.7606405059806847e-05,
      "loss": 2.1072,
      "step": 184193
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.3595551252365112,
      "learning_rate": 5.760399487499139e-05,
      "loss": 2.132,
      "step": 184194
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.360275149345398,
      "learning_rate": 5.760158473524201e-05,
      "loss": 1.984,
      "step": 184195
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.4011693000793457,
      "learning_rate": 5.7599174640558954e-05,
      "loss": 2.1257,
      "step": 184196
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.225622534751892,
      "learning_rate": 5.759676459094291e-05,
      "loss": 1.9928,
      "step": 184197
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.4094138145446777,
      "learning_rate": 5.759435458639415e-05,
      "loss": 2.0367,
      "step": 184198
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.6680335998535156,
      "learning_rate": 5.759194462691321e-05,
      "loss": 1.8624,
      "step": 184199
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.342264175415039,
      "learning_rate": 5.7589534712500454e-05,
      "loss": 1.9455,
      "step": 184200
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.403304100036621,
      "learning_rate": 5.7587124843156405e-05,
      "loss": 1.9348,
      "step": 184201
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.258917212486267,
      "learning_rate": 5.758471501888145e-05,
      "loss": 2.283,
      "step": 184202
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.4890047311782837,
      "learning_rate": 5.75823052396761e-05,
      "loss": 2.0462,
      "step": 184203
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.472874402999878,
      "learning_rate": 5.757989550554079e-05,
      "loss": 2.0793,
      "step": 184204
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.3382794857025146,
      "learning_rate": 5.757748581647598e-05,
      "loss": 1.9951,
      "step": 184205
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.2220021486282349,
      "learning_rate": 5.7575076172482e-05,
      "loss": 2.057,
      "step": 184206
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.3829638957977295,
      "learning_rate": 5.7572666573559475e-05,
      "loss": 2.0046,
      "step": 184207
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.4002723693847656,
      "learning_rate": 5.7570257019708663e-05,
      "loss": 1.9903,
      "step": 184208
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.3085089921951294,
      "learning_rate": 5.756784751093023e-05,
      "loss": 2.3072,
      "step": 184209
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.29519522190094,
      "learning_rate": 5.756543804722448e-05,
      "loss": 2.0219,
      "step": 184210
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.344063401222229,
      "learning_rate": 5.756302862859193e-05,
      "loss": 1.777,
      "step": 184211
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.376111626625061,
      "learning_rate": 5.756061925503287e-05,
      "loss": 1.8353,
      "step": 184212
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.3574491739273071,
      "learning_rate": 5.755820992654798e-05,
      "loss": 1.9839,
      "step": 184213
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.3706557750701904,
      "learning_rate": 5.7555800643137475e-05,
      "loss": 1.8504,
      "step": 184214
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.3335988521575928,
      "learning_rate": 5.755339140480204e-05,
      "loss": 2.0131,
      "step": 184215
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.3601782321929932,
      "learning_rate": 5.755098221154198e-05,
      "loss": 1.9651,
      "step": 184216
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.3001923561096191,
      "learning_rate": 5.754857306335777e-05,
      "loss": 1.993,
      "step": 184217
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.7319632768630981,
      "learning_rate": 5.7546163960249804e-05,
      "loss": 2.098,
      "step": 184218
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.1964401006698608,
      "learning_rate": 5.7543754902218645e-05,
      "loss": 1.7603,
      "step": 184219
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.492684006690979,
      "learning_rate": 5.754134588926459e-05,
      "loss": 1.8003,
      "step": 184220
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.4052931070327759,
      "learning_rate": 5.753893692138828e-05,
      "loss": 2.0636,
      "step": 184221
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.3028931617736816,
      "learning_rate": 5.7536527998590046e-05,
      "loss": 2.0357,
      "step": 184222
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.4058579206466675,
      "learning_rate": 5.753411912087035e-05,
      "loss": 1.8453,
      "step": 184223
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.4593243598937988,
      "learning_rate": 5.753171028822956e-05,
      "loss": 2.0779,
      "step": 184224
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.2725998163223267,
      "learning_rate": 5.752930150066825e-05,
      "loss": 2.182,
      "step": 184225
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.3428057432174683,
      "learning_rate": 5.752689275818677e-05,
      "loss": 1.7715,
      "step": 184226
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.3477095365524292,
      "learning_rate": 5.7524484060785704e-05,
      "loss": 1.9546,
      "step": 184227
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.2354103326797485,
      "learning_rate": 5.7522075408465404e-05,
      "loss": 2.1321,
      "step": 184228
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.618572473526001,
      "learning_rate": 5.751966680122632e-05,
      "loss": 1.9754,
      "step": 184229
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.3259035348892212,
      "learning_rate": 5.751725823906883e-05,
      "loss": 2.0141,
      "step": 184230
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.4180521965026855,
      "learning_rate": 5.751484972199355e-05,
      "loss": 1.7747,
      "step": 184231
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.3196215629577637,
      "learning_rate": 5.751244125000074e-05,
      "loss": 2.2793,
      "step": 184232
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.3117287158966064,
      "learning_rate": 5.751003282309104e-05,
      "loss": 1.954,
      "step": 184233
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.3804214000701904,
      "learning_rate": 5.750762444126473e-05,
      "loss": 2.2288,
      "step": 184234
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.4143295288085938,
      "learning_rate": 5.7505216104522466e-05,
      "loss": 1.8498,
      "step": 184235
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.188782811164856,
      "learning_rate": 5.7502807812864404e-05,
      "loss": 2.0089,
      "step": 184236
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.5237231254577637,
      "learning_rate": 5.750039956629124e-05,
      "loss": 2.1193,
      "step": 184237
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.6579668521881104,
      "learning_rate": 5.749799136480325e-05,
      "loss": 2.1344,
      "step": 184238
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.3706356287002563,
      "learning_rate": 5.749558320840103e-05,
      "loss": 1.981,
      "step": 184239
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.487739086151123,
      "learning_rate": 5.7493175097084875e-05,
      "loss": 1.9934,
      "step": 184240
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.382861852645874,
      "learning_rate": 5.749076703085542e-05,
      "loss": 1.9373,
      "step": 184241
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.2889436483383179,
      "learning_rate": 5.748835900971297e-05,
      "loss": 1.84,
      "step": 184242
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.416007399559021,
      "learning_rate": 5.748595103365805e-05,
      "loss": 1.7744,
      "step": 184243
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.631176471710205,
      "learning_rate": 5.7483543102690965e-05,
      "loss": 2.0486,
      "step": 184244
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.5818941593170166,
      "learning_rate": 5.7481135216812344e-05,
      "loss": 2.158,
      "step": 184245
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.3720299005508423,
      "learning_rate": 5.747872737602249e-05,
      "loss": 2.2065,
      "step": 184246
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.314016580581665,
      "learning_rate": 5.747631958032201e-05,
      "loss": 1.908,
      "step": 184247
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.39820396900177,
      "learning_rate": 5.747391182971125e-05,
      "loss": 2.0338,
      "step": 184248
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.6474692821502686,
      "learning_rate": 5.747150412419066e-05,
      "loss": 1.833,
      "step": 184249
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.3667075634002686,
      "learning_rate": 5.746909646376063e-05,
      "loss": 1.9721,
      "step": 184250
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.1970913410186768,
      "learning_rate": 5.746668884842174e-05,
      "loss": 1.8906,
      "step": 184251
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.2971971035003662,
      "learning_rate": 5.7464281278174306e-05,
      "loss": 1.8842,
      "step": 184252
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.3700194358825684,
      "learning_rate": 5.74618737530189e-05,
      "loss": 1.9198,
      "step": 184253
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.3361951112747192,
      "learning_rate": 5.7459466272955935e-05,
      "loss": 2.1977,
      "step": 184254
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.4929176568984985,
      "learning_rate": 5.745705883798573e-05,
      "loss": 2.2321,
      "step": 184255
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.6576794385910034,
      "learning_rate": 5.745465144810895e-05,
      "loss": 2.098,
      "step": 184256
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.3871521949768066,
      "learning_rate": 5.7452244103325896e-05,
      "loss": 1.8733,
      "step": 184257
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.4605580568313599,
      "learning_rate": 5.7449836803636975e-05,
      "loss": 2.1819,
      "step": 184258
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.3723095655441284,
      "learning_rate": 5.744742954904278e-05,
      "loss": 1.8454,
      "step": 184259
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.3977848291397095,
      "learning_rate": 5.744502233954368e-05,
      "loss": 1.887,
      "step": 184260
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.428244948387146,
      "learning_rate": 5.744261517514007e-05,
      "loss": 2.0136,
      "step": 184261
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.5564497709274292,
      "learning_rate": 5.7440208055832524e-05,
      "loss": 1.9974,
      "step": 184262
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.3619410991668701,
      "learning_rate": 5.7437800981621374e-05,
      "loss": 2.1698,
      "step": 184263
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.5212445259094238,
      "learning_rate": 5.7435393952507147e-05,
      "loss": 2.0385,
      "step": 184264
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.321349859237671,
      "learning_rate": 5.743298696849028e-05,
      "loss": 1.8532,
      "step": 184265
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.3241137266159058,
      "learning_rate": 5.7430580029571205e-05,
      "loss": 1.8677,
      "step": 184266
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.2519093751907349,
      "learning_rate": 5.742817313575029e-05,
      "loss": 2.15,
      "step": 184267
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.4160985946655273,
      "learning_rate": 5.742576628702813e-05,
      "loss": 1.95,
      "step": 184268
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.3691329956054688,
      "learning_rate": 5.742335948340503e-05,
      "loss": 1.982,
      "step": 184269
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.3230550289154053,
      "learning_rate": 5.7420952724881553e-05,
      "loss": 1.7473,
      "step": 184270
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.3848670721054077,
      "learning_rate": 5.741854601145813e-05,
      "loss": 2.0013,
      "step": 184271
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.4460281133651733,
      "learning_rate": 5.741613934313517e-05,
      "loss": 1.8699,
      "step": 184272
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.5443440675735474,
      "learning_rate": 5.741373271991303e-05,
      "loss": 1.8787,
      "step": 184273
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.3910902738571167,
      "learning_rate": 5.741132614179235e-05,
      "loss": 2.2366,
      "step": 184274
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.3173388242721558,
      "learning_rate": 5.740891960877342e-05,
      "loss": 2.2292,
      "step": 184275
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.3415553569793701,
      "learning_rate": 5.740651312085681e-05,
      "loss": 2.2279,
      "step": 184276
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.3857299089431763,
      "learning_rate": 5.7404106678042926e-05,
      "loss": 2.2025,
      "step": 184277
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.3949748277664185,
      "learning_rate": 5.74017002803322e-05,
      "loss": 1.7905,
      "step": 184278
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.4520047903060913,
      "learning_rate": 5.739929392772499e-05,
      "loss": 2.0679,
      "step": 184279
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.3781784772872925,
      "learning_rate": 5.7396887620221895e-05,
      "loss": 2.137,
      "step": 184280
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.5446184873580933,
      "learning_rate": 5.739448135782323e-05,
      "loss": 1.9786,
      "step": 184281
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.4901750087738037,
      "learning_rate": 5.7392075140529583e-05,
      "loss": 2.0654,
      "step": 184282
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.6084611415863037,
      "learning_rate": 5.7389668968341316e-05,
      "loss": 2.0171,
      "step": 184283
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.4290238618850708,
      "learning_rate": 5.738726284125894e-05,
      "loss": 2.0869,
      "step": 184284
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.3067185878753662,
      "learning_rate": 5.738485675928272e-05,
      "loss": 2.1883,
      "step": 184285
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.3938604593276978,
      "learning_rate": 5.7382450722413345e-05,
      "loss": 1.9363,
      "step": 184286
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.4262759685516357,
      "learning_rate": 5.738004473065105e-05,
      "loss": 1.9721,
      "step": 184287
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.6482921838760376,
      "learning_rate": 5.737763878399648e-05,
      "loss": 2.0244,
      "step": 184288
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.3754637241363525,
      "learning_rate": 5.737523288245e-05,
      "loss": 1.9918,
      "step": 184289
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.5367181301116943,
      "learning_rate": 5.7372827026012024e-05,
      "loss": 2.0613,
      "step": 184290
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.3408095836639404,
      "learning_rate": 5.737042121468294e-05,
      "loss": 1.9494,
      "step": 184291
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.5733391046524048,
      "learning_rate": 5.736801544846337e-05,
      "loss": 2.0676,
      "step": 184292
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.2972837686538696,
      "learning_rate": 5.7365609727353544e-05,
      "loss": 2.0925,
      "step": 184293
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.4343806505203247,
      "learning_rate": 5.736320405135414e-05,
      "loss": 2.0273,
      "step": 184294
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.5249439477920532,
      "learning_rate": 5.736079842046552e-05,
      "loss": 2.1161,
      "step": 184295
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.3743466138839722,
      "learning_rate": 5.735839283468807e-05,
      "loss": 1.9092,
      "step": 184296
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.517351508140564,
      "learning_rate": 5.735598729402221e-05,
      "loss": 1.9765,
      "step": 184297
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.2710398435592651,
      "learning_rate": 5.735358179846853e-05,
      "loss": 2.2647,
      "step": 184298
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.3409067392349243,
      "learning_rate": 5.735117634802733e-05,
      "loss": 1.9463,
      "step": 184299
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.4084479808807373,
      "learning_rate": 5.7348770942699207e-05,
      "loss": 2.144,
      "step": 184300
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.6787493228912354,
      "learning_rate": 5.7346365582484434e-05,
      "loss": 1.9671,
      "step": 184301
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.3200082778930664,
      "learning_rate": 5.7343960267383706e-05,
      "loss": 2.132,
      "step": 184302
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.2487088441848755,
      "learning_rate": 5.7341554997397187e-05,
      "loss": 2.2392,
      "step": 184303
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.3320120573043823,
      "learning_rate": 5.733914977252552e-05,
      "loss": 1.7597,
      "step": 184304
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.2926766872406006,
      "learning_rate": 5.733674459276899e-05,
      "loss": 2.1564,
      "step": 184305
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.3555067777633667,
      "learning_rate": 5.733433945812821e-05,
      "loss": 1.9705,
      "step": 184306
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.4137511253356934,
      "learning_rate": 5.73319343686035e-05,
      "loss": 1.9126,
      "step": 184307
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.376717448234558,
      "learning_rate": 5.7329529324195545e-05,
      "loss": 1.8201,
      "step": 184308
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.3022035360336304,
      "learning_rate": 5.7327124324904425e-05,
      "loss": 2.024,
      "step": 184309
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.2633763551712036,
      "learning_rate": 5.732471937073085e-05,
      "loss": 2.0079,
      "step": 184310
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.536848783493042,
      "learning_rate": 5.732231446167512e-05,
      "loss": 2.08,
      "step": 184311
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.4130064249038696,
      "learning_rate": 5.731990959773783e-05,
      "loss": 2.0392,
      "step": 184312
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.4853816032409668,
      "learning_rate": 5.731750477891925e-05,
      "loss": 2.0487,
      "step": 184313
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.2126775979995728,
      "learning_rate": 5.731510000522012e-05,
      "loss": 1.9305,
      "step": 184314
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.2797962427139282,
      "learning_rate": 5.731269527664052e-05,
      "loss": 1.9537,
      "step": 184315
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.3441191911697388,
      "learning_rate": 5.731029059318113e-05,
      "loss": 2.2644,
      "step": 184316
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.3659374713897705,
      "learning_rate": 5.7307885954842284e-05,
      "loss": 2.1128,
      "step": 184317
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.4090269804000854,
      "learning_rate": 5.730548136162457e-05,
      "loss": 2.2532,
      "step": 184318
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.2773287296295166,
      "learning_rate": 5.730307681352824e-05,
      "loss": 2.0251,
      "step": 184319
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.308430790901184,
      "learning_rate": 5.730067231055404e-05,
      "loss": 2.0297,
      "step": 184320
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.3349895477294922,
      "learning_rate": 5.7298267852702054e-05,
      "loss": 1.9812,
      "step": 184321
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.4570170640945435,
      "learning_rate": 5.729586343997297e-05,
      "loss": 2.1206,
      "step": 184322
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.4192683696746826,
      "learning_rate": 5.7293459072367084e-05,
      "loss": 2.0029,
      "step": 184323
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.4036580324172974,
      "learning_rate": 5.7291054749885016e-05,
      "loss": 1.9695,
      "step": 184324
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.3033781051635742,
      "learning_rate": 5.728865047252704e-05,
      "loss": 1.8523,
      "step": 184325
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.3056477308273315,
      "learning_rate": 5.7286246240293774e-05,
      "loss": 2.0908,
      "step": 184326
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.3596999645233154,
      "learning_rate": 5.728384205318558e-05,
      "loss": 2.0301,
      "step": 184327
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.5164012908935547,
      "learning_rate": 5.728143791120289e-05,
      "loss": 2.0183,
      "step": 184328
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.3533616065979004,
      "learning_rate": 5.72790338143461e-05,
      "loss": 2.0054,
      "step": 184329
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.2620631456375122,
      "learning_rate": 5.727662976261578e-05,
      "loss": 1.8959,
      "step": 184330
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.3646719455718994,
      "learning_rate": 5.7274225756012226e-05,
      "loss": 2.1744,
      "step": 184331
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.3170244693756104,
      "learning_rate": 5.7271821794536075e-05,
      "loss": 1.9138,
      "step": 184332
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.3623639345169067,
      "learning_rate": 5.7269417878187663e-05,
      "loss": 1.9671,
      "step": 184333
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.2731963396072388,
      "learning_rate": 5.7267014006967446e-05,
      "loss": 2.0209,
      "step": 184334
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.472377896308899,
      "learning_rate": 5.72646101808758e-05,
      "loss": 2.125,
      "step": 184335
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.426120638847351,
      "learning_rate": 5.726220639991332e-05,
      "loss": 1.9782,
      "step": 184336
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.3216826915740967,
      "learning_rate": 5.725980266408031e-05,
      "loss": 1.9557,
      "step": 184337
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.395695686340332,
      "learning_rate": 5.7257398973377364e-05,
      "loss": 1.9315,
      "step": 184338
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.300646424293518,
      "learning_rate": 5.7254995327804854e-05,
      "loss": 1.9349,
      "step": 184339
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.3513288497924805,
      "learning_rate": 5.725259172736314e-05,
      "loss": 2.1559,
      "step": 184340
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.3308980464935303,
      "learning_rate": 5.7250188172052825e-05,
      "loss": 1.9823,
      "step": 184341
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.3253992795944214,
      "learning_rate": 5.7247784661874276e-05,
      "loss": 1.9149,
      "step": 184342
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.257417917251587,
      "learning_rate": 5.724538119682786e-05,
      "loss": 2.205,
      "step": 184343
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.4056992530822754,
      "learning_rate": 5.72429777769142e-05,
      "loss": 1.9348,
      "step": 184344
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.212460994720459,
      "learning_rate": 5.7240574402133646e-05,
      "loss": 2.2432,
      "step": 184345
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.398808479309082,
      "learning_rate": 5.7238171072486584e-05,
      "loss": 2.1592,
      "step": 184346
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.4105348587036133,
      "learning_rate": 5.723576778797362e-05,
      "loss": 1.9616,
      "step": 184347
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.3478933572769165,
      "learning_rate": 5.723336454859502e-05,
      "loss": 2.0077,
      "step": 184348
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.358967900276184,
      "learning_rate": 5.7230961354351376e-05,
      "loss": 2.0297,
      "step": 184349
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.4716084003448486,
      "learning_rate": 5.7228558205243105e-05,
      "loss": 2.0536,
      "step": 184350
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.3701940774917603,
      "learning_rate": 5.7226155101270624e-05,
      "loss": 1.9229,
      "step": 184351
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.3080297708511353,
      "learning_rate": 5.7223752042434265e-05,
      "loss": 2.1759,
      "step": 184352
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.3533570766448975,
      "learning_rate": 5.7221349028734706e-05,
      "loss": 2.0698,
      "step": 184353
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.4362399578094482,
      "learning_rate": 5.721894606017221e-05,
      "loss": 2.0445,
      "step": 184354
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.5541276931762695,
      "learning_rate": 5.721654313674737e-05,
      "loss": 2.0273,
      "step": 184355
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.4483025074005127,
      "learning_rate": 5.721414025846056e-05,
      "loss": 2.0848,
      "step": 184356
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.4851654767990112,
      "learning_rate": 5.721173742531221e-05,
      "loss": 1.9926,
      "step": 184357
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.3716696500778198,
      "learning_rate": 5.720933463730272e-05,
      "loss": 1.96,
      "step": 184358
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.4260592460632324,
      "learning_rate": 5.720693189443265e-05,
      "loss": 2.1978,
      "step": 184359
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.5058443546295166,
      "learning_rate": 5.7204529196702345e-05,
      "loss": 2.0815,
      "step": 184360
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.4200645685195923,
      "learning_rate": 5.720212654411236e-05,
      "loss": 1.8029,
      "step": 184361
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.2902401685714722,
      "learning_rate": 5.71997239366631e-05,
      "loss": 2.2274,
      "step": 184362
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.411422610282898,
      "learning_rate": 5.7197321374355e-05,
      "loss": 1.8879,
      "step": 184363
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.3645319938659668,
      "learning_rate": 5.719491885718839e-05,
      "loss": 2.0233,
      "step": 184364
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.5365303754806519,
      "learning_rate": 5.7192516385163935e-05,
      "loss": 2.1839,
      "step": 184365
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.320340871810913,
      "learning_rate": 5.719011395828187e-05,
      "loss": 1.8528,
      "step": 184366
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.2640432119369507,
      "learning_rate": 5.7187711576542826e-05,
      "loss": 2.0096,
      "step": 184367
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.334058403968811,
      "learning_rate": 5.718530923994711e-05,
      "loss": 1.9698,
      "step": 184368
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.3972996473312378,
      "learning_rate": 5.718290694849538e-05,
      "loss": 2.123,
      "step": 184369
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.4442119598388672,
      "learning_rate": 5.71805047021878e-05,
      "loss": 1.9142,
      "step": 184370
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.5986014604568481,
      "learning_rate": 5.717810250102498e-05,
      "loss": 1.863,
      "step": 184371
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.28297758102417,
      "learning_rate": 5.717570034500728e-05,
      "loss": 1.9421,
      "step": 184372
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.4162712097167969,
      "learning_rate": 5.7173298234135267e-05,
      "loss": 1.9122,
      "step": 184373
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.238952398300171,
      "learning_rate": 5.717089616840927e-05,
      "loss": 1.955,
      "step": 184374
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.4522385597229004,
      "learning_rate": 5.716849414782993e-05,
      "loss": 2.0305,
      "step": 184375
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.3628805875778198,
      "learning_rate": 5.716609217239738e-05,
      "loss": 2.2549,
      "step": 184376
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.369779348373413,
      "learning_rate": 5.7163690242112315e-05,
      "loss": 2.0069,
      "step": 184377
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.3349417448043823,
      "learning_rate": 5.7161288356975034e-05,
      "loss": 1.7721,
      "step": 184378
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.4486749172210693,
      "learning_rate": 5.715888651698613e-05,
      "loss": 2.25,
      "step": 184379
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.3205357789993286,
      "learning_rate": 5.7156484722145925e-05,
      "loss": 1.9657,
      "step": 184380
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.3649288415908813,
      "learning_rate": 5.7154082972455026e-05,
      "loss": 1.9886,
      "step": 184381
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.251435399055481,
      "learning_rate": 5.715168126791365e-05,
      "loss": 2.1012,
      "step": 184382
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.275484323501587,
      "learning_rate": 5.714927960852239e-05,
      "loss": 2.0379,
      "step": 184383
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.3533868789672852,
      "learning_rate": 5.714687799428161e-05,
      "loss": 1.8265,
      "step": 184384
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.4579867124557495,
      "learning_rate": 5.714447642519192e-05,
      "loss": 2.1984,
      "step": 184385
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.3598576784133911,
      "learning_rate": 5.7142074901253544e-05,
      "loss": 1.9392,
      "step": 184386
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.2771289348602295,
      "learning_rate": 5.713967342246718e-05,
      "loss": 1.9877,
      "step": 184387
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.3394818305969238,
      "learning_rate": 5.7137271988833e-05,
      "loss": 2.0805,
      "step": 184388
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.4299782514572144,
      "learning_rate": 5.713487060035167e-05,
      "loss": 2.1423,
      "step": 184389
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.7458419799804688,
      "learning_rate": 5.7132469257023485e-05,
      "loss": 1.989,
      "step": 184390
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.404160737991333,
      "learning_rate": 5.7130067958849015e-05,
      "loss": 2.0895,
      "step": 184391
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.393248200416565,
      "learning_rate": 5.7127666705828556e-05,
      "loss": 2.1308,
      "step": 184392
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.428348422050476,
      "learning_rate": 5.7125265497962814e-05,
      "loss": 2.0563,
      "step": 184393
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.3727805614471436,
      "learning_rate": 5.712286433525195e-05,
      "loss": 2.1429,
      "step": 184394
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.3849509954452515,
      "learning_rate": 5.712046321769656e-05,
      "loss": 2.1843,
      "step": 184395
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.4343457221984863,
      "learning_rate": 5.711806214529703e-05,
      "loss": 1.9841,
      "step": 184396
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.325878381729126,
      "learning_rate": 5.711566111805387e-05,
      "loss": 1.9321,
      "step": 184397
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.349669098854065,
      "learning_rate": 5.711326013596742e-05,
      "loss": 1.9412,
      "step": 184398
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.3634824752807617,
      "learning_rate": 5.711085919903838e-05,
      "loss": 2.0129,
      "step": 184399
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.2956736087799072,
      "learning_rate": 5.7108458307266855e-05,
      "loss": 2.0428,
      "step": 184400
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.2491947412490845,
      "learning_rate": 5.710605746065354e-05,
      "loss": 2.1414,
      "step": 184401
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.3821079730987549,
      "learning_rate": 5.71036566591987e-05,
      "loss": 2.2833,
      "step": 184402
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.3092573881149292,
      "learning_rate": 5.710125590290294e-05,
      "loss": 1.8595,
      "step": 184403
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.26339590549469,
      "learning_rate": 5.709885519176659e-05,
      "loss": 1.8923,
      "step": 184404
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.3668346405029297,
      "learning_rate": 5.709645452579028e-05,
      "loss": 2.1797,
      "step": 184405
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.2439930438995361,
      "learning_rate": 5.709405390497418e-05,
      "loss": 1.8884,
      "step": 184406
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.5070927143096924,
      "learning_rate": 5.7091653329318955e-05,
      "loss": 2.1186,
      "step": 184407
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.6611201763153076,
      "learning_rate": 5.7089252798824905e-05,
      "loss": 2.1579,
      "step": 184408
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.3795796632766724,
      "learning_rate": 5.7086852313492626e-05,
      "loss": 2.0365,
      "step": 184409
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.3163975477218628,
      "learning_rate": 5.708445187332242e-05,
      "loss": 2.1139,
      "step": 184410
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.3699536323547363,
      "learning_rate": 5.7082051478314924e-05,
      "loss": 2.0974,
      "step": 184411
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.641991376876831,
      "learning_rate": 5.7079651128470335e-05,
      "loss": 2.0011,
      "step": 184412
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.5108085870742798,
      "learning_rate": 5.707725082378929e-05,
      "loss": 2.05,
      "step": 184413
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.4843921661376953,
      "learning_rate": 5.707485056427208e-05,
      "loss": 2.1243,
      "step": 184414
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.3836249113082886,
      "learning_rate": 5.707245034991934e-05,
      "loss": 2.0284,
      "step": 184415
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.4236972332000732,
      "learning_rate": 5.707005018073134e-05,
      "loss": 2.0846,
      "step": 184416
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.433651328086853,
      "learning_rate": 5.7067650056708645e-05,
      "loss": 1.8855,
      "step": 184417
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.3998037576675415,
      "learning_rate": 5.706524997785169e-05,
      "loss": 1.9499,
      "step": 184418
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.357022762298584,
      "learning_rate": 5.706284994416091e-05,
      "loss": 1.9631,
      "step": 184419
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.3556486368179321,
      "learning_rate": 5.706044995563662e-05,
      "loss": 1.9443,
      "step": 184420
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.2282819747924805,
      "learning_rate": 5.7058050012279476e-05,
      "loss": 2.0256,
      "step": 184421
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.3156734704971313,
      "learning_rate": 5.7055650114089734e-05,
      "loss": 1.9713,
      "step": 184422
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.3226121664047241,
      "learning_rate": 5.7053250261068e-05,
      "loss": 2.1446,
      "step": 184423
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.3477773666381836,
      "learning_rate": 5.705085045321469e-05,
      "loss": 1.9177,
      "step": 184424
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.304715871810913,
      "learning_rate": 5.704845069053009e-05,
      "loss": 2.2021,
      "step": 184425
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.3633191585540771,
      "learning_rate": 5.704605097301489e-05,
      "loss": 1.9459,
      "step": 184426
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.3612407445907593,
      "learning_rate": 5.704365130066939e-05,
      "loss": 2.0228,
      "step": 184427
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.306423544883728,
      "learning_rate": 5.7041251673493994e-05,
      "loss": 2.0895,
      "step": 184428
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.39060378074646,
      "learning_rate": 5.70388520914893e-05,
      "loss": 2.0269,
      "step": 184429
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.4048285484313965,
      "learning_rate": 5.703645255465563e-05,
      "loss": 1.9664,
      "step": 184430
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.3326860666275024,
      "learning_rate": 5.7034053062993435e-05,
      "loss": 2.0271,
      "step": 184431
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.4505540132522583,
      "learning_rate": 5.703165361650324e-05,
      "loss": 2.246,
      "step": 184432
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.3130286931991577,
      "learning_rate": 5.702925421518547e-05,
      "loss": 2.211,
      "step": 184433
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.3418859243392944,
      "learning_rate": 5.7026854859040475e-05,
      "loss": 2.0875,
      "step": 184434
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.1874264478683472,
      "learning_rate": 5.702445554806884e-05,
      "loss": 2.1121,
      "step": 184435
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.3805075883865356,
      "learning_rate": 5.7022056282270935e-05,
      "loss": 2.1508,
      "step": 184436
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.5398848056793213,
      "learning_rate": 5.701965706164716e-05,
      "loss": 2.3347,
      "step": 184437
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.2717212438583374,
      "learning_rate": 5.7017257886198087e-05,
      "loss": 2.013,
      "step": 184438
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.3942952156066895,
      "learning_rate": 5.701485875592401e-05,
      "loss": 2.1521,
      "step": 184439
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.3381413221359253,
      "learning_rate": 5.701245967082556e-05,
      "loss": 2.0419,
      "step": 184440
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.258164882659912,
      "learning_rate": 5.701006063090304e-05,
      "loss": 2.1565,
      "step": 184441
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.4629188776016235,
      "learning_rate": 5.700766163615695e-05,
      "loss": 1.9658,
      "step": 184442
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.4928264617919922,
      "learning_rate": 5.700526268658766e-05,
      "loss": 2.0185,
      "step": 184443
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.3099974393844604,
      "learning_rate": 5.700286378219576e-05,
      "loss": 1.8397,
      "step": 184444
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.3608438968658447,
      "learning_rate": 5.7000464922981525e-05,
      "loss": 2.1193,
      "step": 184445
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.3172569274902344,
      "learning_rate": 5.699806610894555e-05,
      "loss": 2.034,
      "step": 184446
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.3500070571899414,
      "learning_rate": 5.699566734008817e-05,
      "loss": 1.8676,
      "step": 184447
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.3951666355133057,
      "learning_rate": 5.699326861641005e-05,
      "loss": 1.8954,
      "step": 184448
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.4537028074264526,
      "learning_rate": 5.699086993791129e-05,
      "loss": 2.0574,
      "step": 184449
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.3503791093826294,
      "learning_rate": 5.6988471304592586e-05,
      "loss": 1.9311,
      "step": 184450
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.5145783424377441,
      "learning_rate": 5.698607271645424e-05,
      "loss": 2.0715,
      "step": 184451
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.389001488685608,
      "learning_rate": 5.698367417349686e-05,
      "loss": 1.9505,
      "step": 184452
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.4116957187652588,
      "learning_rate": 5.6981275675720704e-05,
      "loss": 1.9418,
      "step": 184453
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.3874096870422363,
      "learning_rate": 5.697887722312647e-05,
      "loss": 2.0075,
      "step": 184454
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.5303020477294922,
      "learning_rate": 5.697647881571433e-05,
      "loss": 2.1147,
      "step": 184455
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.443832516670227,
      "learning_rate": 5.6974080453484885e-05,
      "loss": 2.1631,
      "step": 184456
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.6263700723648071,
      "learning_rate": 5.697168213643849e-05,
      "loss": 2.1497,
      "step": 184457
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.474193811416626,
      "learning_rate": 5.696928386457572e-05,
      "loss": 1.9177,
      "step": 184458
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.4448330402374268,
      "learning_rate": 5.696688563789688e-05,
      "loss": 1.8112,
      "step": 184459
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.3064864873886108,
      "learning_rate": 5.6964487456402654e-05,
      "loss": 1.8906,
      "step": 184460
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.4106892347335815,
      "learning_rate": 5.696208932009312e-05,
      "loss": 1.7922,
      "step": 184461
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.3460652828216553,
      "learning_rate": 5.695969122896904e-05,
      "loss": 2.0983,
      "step": 184462
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.324906587600708,
      "learning_rate": 5.695729318303062e-05,
      "loss": 1.9989,
      "step": 184463
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.3226473331451416,
      "learning_rate": 5.695489518227855e-05,
      "loss": 1.9847,
      "step": 184464
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.411171555519104,
      "learning_rate": 5.6952497226713036e-05,
      "loss": 2.1809,
      "step": 184465
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.562773585319519,
      "learning_rate": 5.695009931633481e-05,
      "loss": 1.8789,
      "step": 184466
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.451973557472229,
      "learning_rate": 5.6947701451144e-05,
      "loss": 1.94,
      "step": 184467
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.4994511604309082,
      "learning_rate": 5.6945303631141284e-05,
      "loss": 2.2438,
      "step": 184468
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.4719496965408325,
      "learning_rate": 5.694290585632691e-05,
      "loss": 1.9703,
      "step": 184469
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.3433732986450195,
      "learning_rate": 5.694050812670157e-05,
      "loss": 2.1073,
      "step": 184470
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.3511015176773071,
      "learning_rate": 5.693811044226547e-05,
      "loss": 1.8898,
      "step": 184471
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.3476253747940063,
      "learning_rate": 5.693571280301933e-05,
      "loss": 2.289,
      "step": 184472
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.3138073682785034,
      "learning_rate": 5.693331520896324e-05,
      "loss": 2.1029,
      "step": 184473
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.4077192544937134,
      "learning_rate": 5.693091766009793e-05,
      "loss": 2.2676,
      "step": 184474
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.3409816026687622,
      "learning_rate": 5.6928520156423706e-05,
      "loss": 2.0284,
      "step": 184475
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.3686010837554932,
      "learning_rate": 5.69261226979411e-05,
      "loss": 2.0445,
      "step": 184476
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.2813646793365479,
      "learning_rate": 5.692372528465044e-05,
      "loss": 1.9369,
      "step": 184477
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.3505022525787354,
      "learning_rate": 5.6921327916552404e-05,
      "loss": 1.9759,
      "step": 184478
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.3104609251022339,
      "learning_rate": 5.691893059364715e-05,
      "loss": 2.073,
      "step": 184479
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.337249994277954,
      "learning_rate": 5.691653331593531e-05,
      "loss": 1.9524,
      "step": 184480
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.3759536743164062,
      "learning_rate": 5.691413608341722e-05,
      "loss": 2.1013,
      "step": 184481
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.415914535522461,
      "learning_rate": 5.691173889609344e-05,
      "loss": 1.9377,
      "step": 184482
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.5365179777145386,
      "learning_rate": 5.690934175396428e-05,
      "loss": 1.9632,
      "step": 184483
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.3284189701080322,
      "learning_rate": 5.690694465703043e-05,
      "loss": 1.9677,
      "step": 184484
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.38048255443573,
      "learning_rate": 5.690454760529199e-05,
      "loss": 1.9625,
      "step": 184485
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.3712612390518188,
      "learning_rate": 5.6902150598749676e-05,
      "loss": 1.9222,
      "step": 184486
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.3650647401809692,
      "learning_rate": 5.689975363740376e-05,
      "loss": 2.0369,
      "step": 184487
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.2986863851547241,
      "learning_rate": 5.689735672125484e-05,
      "loss": 1.7273,
      "step": 184488
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.3387372493743896,
      "learning_rate": 5.6894959850303224e-05,
      "loss": 1.9466,
      "step": 184489
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.253413200378418,
      "learning_rate": 5.689256302454955e-05,
      "loss": 2.1079,
      "step": 184490
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.3677451610565186,
      "learning_rate": 5.689016624399403e-05,
      "loss": 2.1916,
      "step": 184491
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.4366087913513184,
      "learning_rate": 5.688776950863725e-05,
      "loss": 2.2923,
      "step": 184492
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.2586318254470825,
      "learning_rate": 5.688537281847958e-05,
      "loss": 2.1875,
      "step": 184493
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.3077991008758545,
      "learning_rate": 5.6882976173521555e-05,
      "loss": 2.0419,
      "step": 184494
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.2859323024749756,
      "learning_rate": 5.688057957376354e-05,
      "loss": 2.0097,
      "step": 184495
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.35152268409729,
      "learning_rate": 5.687818301920607e-05,
      "loss": 2.1736,
      "step": 184496
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.3285225629806519,
      "learning_rate": 5.687578650984951e-05,
      "loss": 2.0364,
      "step": 184497
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.4000499248504639,
      "learning_rate": 5.6873390045694354e-05,
      "loss": 2.0031,
      "step": 184498
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.4599556922912598,
      "learning_rate": 5.687099362674095e-05,
      "loss": 1.8229,
      "step": 184499
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.4108400344848633,
      "learning_rate": 5.686859725298988e-05,
      "loss": 2.1034,
      "step": 184500
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.2838703393936157,
      "learning_rate": 5.686620092444145e-05,
      "loss": 1.9771,
      "step": 184501
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.5086729526519775,
      "learning_rate": 5.6863804641096266e-05,
      "loss": 1.7683,
      "step": 184502
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.3432157039642334,
      "learning_rate": 5.686140840295472e-05,
      "loss": 2.0016,
      "step": 184503
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.3008970022201538,
      "learning_rate": 5.685901221001718e-05,
      "loss": 2.0361,
      "step": 184504
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.3375798463821411,
      "learning_rate": 5.6856616062284085e-05,
      "loss": 2.0327,
      "step": 184505
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.289399266242981,
      "learning_rate": 5.685421995975603e-05,
      "loss": 1.8917,
      "step": 184506
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.2643376588821411,
      "learning_rate": 5.6851823902433245e-05,
      "loss": 1.9751,
      "step": 184507
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.2388699054718018,
      "learning_rate": 5.6849427890316404e-05,
      "loss": 2.2201,
      "step": 184508
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.5674904584884644,
      "learning_rate": 5.6847031923405827e-05,
      "loss": 1.9604,
      "step": 184509
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.4145222902297974,
      "learning_rate": 5.6844636001701926e-05,
      "loss": 1.896,
      "step": 184510
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.4182761907577515,
      "learning_rate": 5.6842240125205265e-05,
      "loss": 2.2151,
      "step": 184511
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.269553542137146,
      "learning_rate": 5.68398442939162e-05,
      "loss": 2.1024,
      "step": 184512
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.2521569728851318,
      "learning_rate": 5.683744850783515e-05,
      "loss": 2.0315,
      "step": 184513
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.3731037378311157,
      "learning_rate": 5.683505276696266e-05,
      "loss": 2.1557,
      "step": 184514
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.4513379335403442,
      "learning_rate": 5.683265707129911e-05,
      "loss": 1.9772,
      "step": 184515
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.268252968788147,
      "learning_rate": 5.6830261420844935e-05,
      "loss": 1.8531,
      "step": 184516
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.3946317434310913,
      "learning_rate": 5.682786581560063e-05,
      "loss": 1.9974,
      "step": 184517
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.301680088043213,
      "learning_rate": 5.682547025556665e-05,
      "loss": 2.1069,
      "step": 184518
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.4282503128051758,
      "learning_rate": 5.682307474074331e-05,
      "loss": 1.8881,
      "step": 184519
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.261386752128601,
      "learning_rate": 5.682067927113124e-05,
      "loss": 1.8242,
      "step": 184520
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.3966593742370605,
      "learning_rate": 5.681828384673077e-05,
      "loss": 2.0262,
      "step": 184521
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.211382269859314,
      "learning_rate": 5.681588846754234e-05,
      "loss": 1.9707,
      "step": 184522
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.2860807180404663,
      "learning_rate": 5.6813493133566465e-05,
      "loss": 2.3136,
      "step": 184523
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.430525779724121,
      "learning_rate": 5.6811097844803496e-05,
      "loss": 2.2394,
      "step": 184524
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.3359425067901611,
      "learning_rate": 5.6808702601253994e-05,
      "loss": 1.8369,
      "step": 184525
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.4567995071411133,
      "learning_rate": 5.680630740291836e-05,
      "loss": 2.1764,
      "step": 184526
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.305067777633667,
      "learning_rate": 5.680391224979702e-05,
      "loss": 2.0451,
      "step": 184527
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.330918550491333,
      "learning_rate": 5.680151714189032e-05,
      "loss": 2.0068,
      "step": 184528
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.4043046236038208,
      "learning_rate": 5.679912207919891e-05,
      "loss": 2.0919,
      "step": 184529
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.4851152896881104,
      "learning_rate": 5.6796727061723066e-05,
      "loss": 2.0301,
      "step": 184530
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.4475055932998657,
      "learning_rate": 5.679433208946339e-05,
      "loss": 2.1386,
      "step": 184531
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.3397762775421143,
      "learning_rate": 5.679193716242021e-05,
      "loss": 2.1272,
      "step": 184532
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.4329222440719604,
      "learning_rate": 5.678954228059399e-05,
      "loss": 2.216,
      "step": 184533
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.4553030729293823,
      "learning_rate": 5.678714744398514e-05,
      "loss": 2.0427,
      "step": 184534
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.3949174880981445,
      "learning_rate": 5.678475265259421e-05,
      "loss": 2.0534,
      "step": 184535
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.2579761743545532,
      "learning_rate": 5.678235790642151e-05,
      "loss": 1.926,
      "step": 184536
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.3041008710861206,
      "learning_rate": 5.677996320546765e-05,
      "loss": 1.782,
      "step": 184537
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.4362632036209106,
      "learning_rate": 5.6777568549732876e-05,
      "loss": 2.0934,
      "step": 184538
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.2731293439865112,
      "learning_rate": 5.677517393921793e-05,
      "loss": 1.9298,
      "step": 184539
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.3220027685165405,
      "learning_rate": 5.6772779373922914e-05,
      "loss": 1.8586,
      "step": 184540
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.391478180885315,
      "learning_rate": 5.67703848538485e-05,
      "loss": 1.7826,
      "step": 184541
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.4405715465545654,
      "learning_rate": 5.676799037899497e-05,
      "loss": 2.135,
      "step": 184542
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.376073956489563,
      "learning_rate": 5.676559594936296e-05,
      "loss": 2.038,
      "step": 184543
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.3360607624053955,
      "learning_rate": 5.676320156495272e-05,
      "loss": 2.064,
      "step": 184544
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.314880132675171,
      "learning_rate": 5.676080722576497e-05,
      "loss": 1.9122,
      "step": 184545
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.3744760751724243,
      "learning_rate": 5.675841293179981e-05,
      "loss": 1.8234,
      "step": 184546
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.310307502746582,
      "learning_rate": 5.6756018683057946e-05,
      "loss": 1.9311,
      "step": 184547
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.257421851158142,
      "learning_rate": 5.6753624479539636e-05,
      "loss": 2.0612,
      "step": 184548
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.3279885053634644,
      "learning_rate": 5.675123032124549e-05,
      "loss": 1.9151,
      "step": 184549
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.3982447385787964,
      "learning_rate": 5.6748836208175795e-05,
      "loss": 1.9142,
      "step": 184550
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.4488039016723633,
      "learning_rate": 5.674644214033126e-05,
      "loss": 1.8959,
      "step": 184551
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.261222004890442,
      "learning_rate": 5.6744048117712015e-05,
      "loss": 2.065,
      "step": 184552
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.5300086736679077,
      "learning_rate": 5.674165414031869e-05,
      "loss": 2.1701,
      "step": 184553
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.3396997451782227,
      "learning_rate": 5.673926020815163e-05,
      "loss": 2.1165,
      "step": 184554
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.594178318977356,
      "learning_rate": 5.673686632121138e-05,
      "loss": 1.8254,
      "step": 184555
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.3321322202682495,
      "learning_rate": 5.673447247949829e-05,
      "loss": 1.8687,
      "step": 184556
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.366042137145996,
      "learning_rate": 5.673207868301299e-05,
      "loss": 2.0614,
      "step": 184557
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.3991707563400269,
      "learning_rate": 5.6729684931755645e-05,
      "loss": 2.0027,
      "step": 184558
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.41793954372406,
      "learning_rate": 5.6727291225726925e-05,
      "loss": 1.8714,
      "step": 184559
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.3512885570526123,
      "learning_rate": 5.672489756492709e-05,
      "loss": 2.3245,
      "step": 184560
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.3787751197814941,
      "learning_rate": 5.6722503949356803e-05,
      "loss": 2.0571,
      "step": 184561
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.3193562030792236,
      "learning_rate": 5.6720110379016303e-05,
      "loss": 1.9399,
      "step": 184562
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.2903825044631958,
      "learning_rate": 5.671771685390619e-05,
      "loss": 2.0293,
      "step": 184563
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.2661770582199097,
      "learning_rate": 5.6715323374026865e-05,
      "loss": 1.9634,
      "step": 184564
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.4533348083496094,
      "learning_rate": 5.6712929939378725e-05,
      "loss": 2.0733,
      "step": 184565
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.3552510738372803,
      "learning_rate": 5.6710536549962206e-05,
      "loss": 2.0505,
      "step": 184566
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.3858458995819092,
      "learning_rate": 5.6708143205777836e-05,
      "loss": 2.2271,
      "step": 184567
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.3518073558807373,
      "learning_rate": 5.670574990682595e-05,
      "loss": 1.956,
      "step": 184568
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.4368106126785278,
      "learning_rate": 5.6703356653107115e-05,
      "loss": 2.0713,
      "step": 184569
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.4349472522735596,
      "learning_rate": 5.670096344462176e-05,
      "loss": 2.0184,
      "step": 184570
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.482823371887207,
      "learning_rate": 5.669857028137026e-05,
      "loss": 2.2709,
      "step": 184571
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.705263376235962,
      "learning_rate": 5.6696177163353006e-05,
      "loss": 1.8599,
      "step": 184572
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.372023105621338,
      "learning_rate": 5.66937840905706e-05,
      "loss": 2.0855,
      "step": 184573
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.432376503944397,
      "learning_rate": 5.6691391063023384e-05,
      "loss": 2.1436,
      "step": 184574
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.3533189296722412,
      "learning_rate": 5.6688998080711845e-05,
      "loss": 1.9962,
      "step": 184575
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.2804996967315674,
      "learning_rate": 5.668660514363645e-05,
      "loss": 1.8425,
      "step": 184576
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.2675248384475708,
      "learning_rate": 5.6684212251797614e-05,
      "loss": 1.7696,
      "step": 184577
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.2630091905593872,
      "learning_rate": 5.668181940519568e-05,
      "loss": 2.093,
      "step": 184578
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.4372930526733398,
      "learning_rate": 5.667942660383127e-05,
      "loss": 1.8822,
      "step": 184579
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.3101227283477783,
      "learning_rate": 5.667703384770467e-05,
      "loss": 2.0188,
      "step": 184580
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.264710783958435,
      "learning_rate": 5.667464113681648e-05,
      "loss": 1.974,
      "step": 184581
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.6196049451828003,
      "learning_rate": 5.667224847116707e-05,
      "loss": 1.911,
      "step": 184582
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.425995111465454,
      "learning_rate": 5.666985585075687e-05,
      "loss": 2.3664,
      "step": 184583
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.4186562299728394,
      "learning_rate": 5.666746327558629e-05,
      "loss": 2.0617,
      "step": 184584
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.3283473253250122,
      "learning_rate": 5.666507074565588e-05,
      "loss": 1.9841,
      "step": 184585
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.1865618228912354,
      "learning_rate": 5.666267826096592e-05,
      "loss": 2.0011,
      "step": 184586
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.394586205482483,
      "learning_rate": 5.666028582151707e-05,
      "loss": 1.9435,
      "step": 184587
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.3564358949661255,
      "learning_rate": 5.665789342730966e-05,
      "loss": 2.0226,
      "step": 184588
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.4717497825622559,
      "learning_rate": 5.665550107834417e-05,
      "loss": 1.9498,
      "step": 184589
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.3538382053375244,
      "learning_rate": 5.665310877462092e-05,
      "loss": 1.8995,
      "step": 184590
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.3627281188964844,
      "learning_rate": 5.66507165161405e-05,
      "loss": 2.0019,
      "step": 184591
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.522217869758606,
      "learning_rate": 5.664832430290327e-05,
      "loss": 2.0175,
      "step": 184592
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.3698577880859375,
      "learning_rate": 5.664593213490974e-05,
      "loss": 2.0412,
      "step": 184593
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.3426456451416016,
      "learning_rate": 5.664354001216036e-05,
      "loss": 2.0902,
      "step": 184594
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.370002031326294,
      "learning_rate": 5.6641147934655515e-05,
      "loss": 1.8789,
      "step": 184595
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.3183127641677856,
      "learning_rate": 5.663875590239562e-05,
      "loss": 1.9858,
      "step": 184596
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.4148832559585571,
      "learning_rate": 5.663636391538125e-05,
      "loss": 1.812,
      "step": 184597
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.4141639471054077,
      "learning_rate": 5.6633971973612705e-05,
      "loss": 1.9498,
      "step": 184598
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.3986358642578125,
      "learning_rate": 5.6631580077090556e-05,
      "loss": 2.0363,
      "step": 184599
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.2857214212417603,
      "learning_rate": 5.662918822581518e-05,
      "loss": 1.9809,
      "step": 184600
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.3239054679870605,
      "learning_rate": 5.662679641978698e-05,
      "loss": 2.0755,
      "step": 184601
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.339284896850586,
      "learning_rate": 5.662440465900652e-05,
      "loss": 1.869,
      "step": 184602
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.466477632522583,
      "learning_rate": 5.6622012943474195e-05,
      "loss": 2.1297,
      "step": 184603
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.3907980918884277,
      "learning_rate": 5.661962127319034e-05,
      "loss": 1.9361,
      "step": 184604
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.307802677154541,
      "learning_rate": 5.661722964815556e-05,
      "loss": 1.9512,
      "step": 184605
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.4284874200820923,
      "learning_rate": 5.661483806837022e-05,
      "loss": 2.0125,
      "step": 184606
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.3183516263961792,
      "learning_rate": 5.661244653383471e-05,
      "loss": 2.0275,
      "step": 184607
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.4100563526153564,
      "learning_rate": 5.661005504454964e-05,
      "loss": 2.0066,
      "step": 184608
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.31365168094635,
      "learning_rate": 5.660766360051527e-05,
      "loss": 1.961,
      "step": 184609
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.3148783445358276,
      "learning_rate": 5.6605272201732174e-05,
      "loss": 1.9595,
      "step": 184610
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.3423985242843628,
      "learning_rate": 5.660288084820078e-05,
      "loss": 1.984,
      "step": 184611
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.367592692375183,
      "learning_rate": 5.6600489539921526e-05,
      "loss": 1.8673,
      "step": 184612
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.4268593788146973,
      "learning_rate": 5.659809827689473e-05,
      "loss": 1.8864,
      "step": 184613
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.5210840702056885,
      "learning_rate": 5.659570705912101e-05,
      "loss": 1.9203,
      "step": 184614
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.3773194551467896,
      "learning_rate": 5.659331588660069e-05,
      "loss": 2.2325,
      "step": 184615
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.54058039188385,
      "learning_rate": 5.6590924759334333e-05,
      "loss": 2.1787,
      "step": 184616
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.4227641820907593,
      "learning_rate": 5.658853367732231e-05,
      "loss": 2.1745,
      "step": 184617
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.389070987701416,
      "learning_rate": 5.6586142640565094e-05,
      "loss": 2.0883,
      "step": 184618
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.330918312072754,
      "learning_rate": 5.6583751649063e-05,
      "loss": 2.0474,
      "step": 184619
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.3481976985931396,
      "learning_rate": 5.658136070281668e-05,
      "loss": 1.8066,
      "step": 184620
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.4523016214370728,
      "learning_rate": 5.657896980182642e-05,
      "loss": 2.2664,
      "step": 184621
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.2932066917419434,
      "learning_rate": 5.657657894609276e-05,
      "loss": 2.2101,
      "step": 184622
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.3893702030181885,
      "learning_rate": 5.657418813561607e-05,
      "loss": 2.0624,
      "step": 184623
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.4700316190719604,
      "learning_rate": 5.6571797370396975e-05,
      "loss": 1.8711,
      "step": 184624
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.4156205654144287,
      "learning_rate": 5.656940665043561e-05,
      "loss": 2.1173,
      "step": 184625
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.3092858791351318,
      "learning_rate": 5.6567015975732675e-05,
      "loss": 1.9087,
      "step": 184626
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.3878499269485474,
      "learning_rate": 5.656462534628844e-05,
      "loss": 2.2002,
      "step": 184627
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.2850347757339478,
      "learning_rate": 5.6562234762103534e-05,
      "loss": 2.1295,
      "step": 184628
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.5539482831954956,
      "learning_rate": 5.655984422317822e-05,
      "loss": 2.1455,
      "step": 184629
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.485447883605957,
      "learning_rate": 5.65574537295131e-05,
      "loss": 1.9653,
      "step": 184630
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.3919589519500732,
      "learning_rate": 5.6555063281108547e-05,
      "loss": 1.8325,
      "step": 184631
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.3587554693222046,
      "learning_rate": 5.6552672877964986e-05,
      "loss": 1.9999,
      "step": 184632
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.437281847000122,
      "learning_rate": 5.655028252008282e-05,
      "loss": 2.0246,
      "step": 184633
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.387589693069458,
      "learning_rate": 5.654789220746262e-05,
      "loss": 2.0497,
      "step": 184634
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.2937560081481934,
      "learning_rate": 5.654550194010471e-05,
      "loss": 1.9357,
      "step": 184635
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.4089802503585815,
      "learning_rate": 5.6543111718009623e-05,
      "loss": 2.0582,
      "step": 184636
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.369383692741394,
      "learning_rate": 5.65407215411778e-05,
      "loss": 1.8389,
      "step": 184637
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.370928406715393,
      "learning_rate": 5.653833140960964e-05,
      "loss": 1.8272,
      "step": 184638
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.3491522073745728,
      "learning_rate": 5.65359413233055e-05,
      "loss": 2.0353,
      "step": 184639
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.3424781560897827,
      "learning_rate": 5.6533551282266055e-05,
      "loss": 2.2346,
      "step": 184640
    },
    {
      "epoch": 2.4,
      "grad_norm": 2.117067813873291,
      "learning_rate": 5.65311612864915e-05,
      "loss": 1.8067,
      "step": 184641
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.3684377670288086,
      "learning_rate": 5.652877133598247e-05,
      "loss": 2.238,
      "step": 184642
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.2772908210754395,
      "learning_rate": 5.6526381430739364e-05,
      "loss": 1.9875,
      "step": 184643
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.3540246486663818,
      "learning_rate": 5.6523991570762586e-05,
      "loss": 1.9673,
      "step": 184644
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.2605960369110107,
      "learning_rate": 5.652160175605253e-05,
      "loss": 1.836,
      "step": 184645
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.4421427249908447,
      "learning_rate": 5.6519211986609756e-05,
      "loss": 2.0122,
      "step": 184646
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.440307855606079,
      "learning_rate": 5.651682226243461e-05,
      "loss": 1.8539,
      "step": 184647
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.429086685180664,
      "learning_rate": 5.651443258352763e-05,
      "loss": 1.9283,
      "step": 184648
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.423421025276184,
      "learning_rate": 5.651204294988923e-05,
      "loss": 2.1357,
      "step": 184649
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.3049705028533936,
      "learning_rate": 5.6509653361519845e-05,
      "loss": 1.8418,
      "step": 184650
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.5351983308792114,
      "learning_rate": 5.650726381841982e-05,
      "loss": 2.0349,
      "step": 184651
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.6179447174072266,
      "learning_rate": 5.650487432058976e-05,
      "loss": 2.1912,
      "step": 184652
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.2970986366271973,
      "learning_rate": 5.650248486802997e-05,
      "loss": 1.827,
      "step": 184653
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.3777331113815308,
      "learning_rate": 5.650009546074105e-05,
      "loss": 1.9511,
      "step": 184654
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.4412086009979248,
      "learning_rate": 5.649770609872337e-05,
      "loss": 1.9329,
      "step": 184655
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.440490961074829,
      "learning_rate": 5.649531678197735e-05,
      "loss": 1.9711,
      "step": 184656
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.4799593687057495,
      "learning_rate": 5.649292751050335e-05,
      "loss": 2.0925,
      "step": 184657
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.3640918731689453,
      "learning_rate": 5.6490538284302036e-05,
      "loss": 1.8322,
      "step": 184658
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.3694112300872803,
      "learning_rate": 5.64881491033736e-05,
      "loss": 1.9851,
      "step": 184659
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.3011852502822876,
      "learning_rate": 5.6485759967718725e-05,
      "loss": 1.9896,
      "step": 184660
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.400132656097412,
      "learning_rate": 5.648337087733773e-05,
      "loss": 2.0615,
      "step": 184661
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.3174911737442017,
      "learning_rate": 5.6480981832231085e-05,
      "loss": 1.9534,
      "step": 184662
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.4391210079193115,
      "learning_rate": 5.647859283239913e-05,
      "loss": 1.9439,
      "step": 184663
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.2832483053207397,
      "learning_rate": 5.6476203877842485e-05,
      "loss": 1.9284,
      "step": 184664
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.2678459882736206,
      "learning_rate": 5.6473814968561426e-05,
      "loss": 2.2395,
      "step": 184665
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.2895623445510864,
      "learning_rate": 5.647142610455655e-05,
      "loss": 1.8547,
      "step": 184666
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.3884433507919312,
      "learning_rate": 5.646903728582826e-05,
      "loss": 2.0602,
      "step": 184667
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.306406855583191,
      "learning_rate": 5.646664851237694e-05,
      "loss": 2.1151,
      "step": 184668
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.303964376449585,
      "learning_rate": 5.646425978420301e-05,
      "loss": 2.062,
      "step": 184669
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.356887698173523,
      "learning_rate": 5.646187110130707e-05,
      "loss": 1.9684,
      "step": 184670
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.266595721244812,
      "learning_rate": 5.6459482463689364e-05,
      "loss": 1.7928,
      "step": 184671
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.2490028142929077,
      "learning_rate": 5.645709387135051e-05,
      "loss": 1.8872,
      "step": 184672
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.4225910902023315,
      "learning_rate": 5.645470532429091e-05,
      "loss": 2.1087,
      "step": 184673
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.3899372816085815,
      "learning_rate": 5.645231682251095e-05,
      "loss": 2.0463,
      "step": 184674
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.4337363243103027,
      "learning_rate": 5.6449928366011e-05,
      "loss": 2.1378,
      "step": 184675
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.3758275508880615,
      "learning_rate": 5.644753995479173e-05,
      "loss": 1.7845,
      "step": 184676
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.5614314079284668,
      "learning_rate": 5.6445151588853346e-05,
      "loss": 1.786,
      "step": 184677
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.3915268182754517,
      "learning_rate": 5.64427632681965e-05,
      "loss": 1.9672,
      "step": 184678
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.4806439876556396,
      "learning_rate": 5.6440374992821566e-05,
      "loss": 1.8164,
      "step": 184679
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.4270670413970947,
      "learning_rate": 5.6437986762728916e-05,
      "loss": 1.7852,
      "step": 184680
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.3140674829483032,
      "learning_rate": 5.6435598577919006e-05,
      "loss": 2.1319,
      "step": 184681
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.2072136402130127,
      "learning_rate": 5.643321043839237e-05,
      "loss": 2.0314,
      "step": 184682
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.4628013372421265,
      "learning_rate": 5.643082234414932e-05,
      "loss": 2.1074,
      "step": 184683
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.329092025756836,
      "learning_rate": 5.642843429519047e-05,
      "loss": 1.9293,
      "step": 184684
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.3876625299453735,
      "learning_rate": 5.642604629151617e-05,
      "loss": 1.9393,
      "step": 184685
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.3149725198745728,
      "learning_rate": 5.6423658333126774e-05,
      "loss": 2.0947,
      "step": 184686
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.3863492012023926,
      "learning_rate": 5.642127042002292e-05,
      "loss": 1.9181,
      "step": 184687
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.6990636587142944,
      "learning_rate": 5.6418882552204916e-05,
      "loss": 2.2035,
      "step": 184688
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.328626275062561,
      "learning_rate": 5.641649472967318e-05,
      "loss": 2.0158,
      "step": 184689
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.2631324529647827,
      "learning_rate": 5.641410695242828e-05,
      "loss": 1.9166,
      "step": 184690
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.2667378187179565,
      "learning_rate": 5.641171922047063e-05,
      "loss": 1.8566,
      "step": 184691
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.4341766834259033,
      "learning_rate": 5.640933153380055e-05,
      "loss": 1.9845,
      "step": 184692
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.287459135055542,
      "learning_rate": 5.640694389241861e-05,
      "loss": 2.1712,
      "step": 184693
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.344280481338501,
      "learning_rate": 5.640455629632528e-05,
      "loss": 2.2424,
      "step": 184694
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.4822885990142822,
      "learning_rate": 5.640216874552081e-05,
      "loss": 1.8753,
      "step": 184695
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.3481419086456299,
      "learning_rate": 5.6399781240005894e-05,
      "loss": 1.8692,
      "step": 184696
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.3764811754226685,
      "learning_rate": 5.6397393779780744e-05,
      "loss": 2.0658,
      "step": 184697
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.3383151292800903,
      "learning_rate": 5.6395006364846026e-05,
      "loss": 2.1979,
      "step": 184698
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.374107837677002,
      "learning_rate": 5.639261899520206e-05,
      "loss": 1.9751,
      "step": 184699
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.438114047050476,
      "learning_rate": 5.639023167084922e-05,
      "loss": 2.1288,
      "step": 184700
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.2594220638275146,
      "learning_rate": 5.638784439178812e-05,
      "loss": 2.1124,
      "step": 184701
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.4779523611068726,
      "learning_rate": 5.638545715801912e-05,
      "loss": 2.0844,
      "step": 184702
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.336973786354065,
      "learning_rate": 5.638306996954256e-05,
      "loss": 2.0274,
      "step": 184703
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.3102391958236694,
      "learning_rate": 5.63806828263591e-05,
      "loss": 1.9216,
      "step": 184704
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.3072636127471924,
      "learning_rate": 5.6378295728469045e-05,
      "loss": 1.9357,
      "step": 184705
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.5345749855041504,
      "learning_rate": 5.6375908675872795e-05,
      "loss": 1.9946,
      "step": 184706
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.2587997913360596,
      "learning_rate": 5.637352166857094e-05,
      "loss": 1.8718,
      "step": 184707
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.374577283859253,
      "learning_rate": 5.637113470656376e-05,
      "loss": 1.9702,
      "step": 184708
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.4029030799865723,
      "learning_rate": 5.636874778985188e-05,
      "loss": 2.0123,
      "step": 184709
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.3342394828796387,
      "learning_rate": 5.636636091843563e-05,
      "loss": 1.9766,
      "step": 184710
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.341589331626892,
      "learning_rate": 5.636397409231548e-05,
      "loss": 1.9986,
      "step": 184711
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.3815099000930786,
      "learning_rate": 5.6361587311491796e-05,
      "loss": 1.9973,
      "step": 184712
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.2484605312347412,
      "learning_rate": 5.635920057596518e-05,
      "loss": 1.9845,
      "step": 184713
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.3153454065322876,
      "learning_rate": 5.63568138857359e-05,
      "loss": 2.1385,
      "step": 184714
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.3155107498168945,
      "learning_rate": 5.6354427240804546e-05,
      "loss": 2.1122,
      "step": 184715
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.4269092082977295,
      "learning_rate": 5.6352040641171525e-05,
      "loss": 2.0392,
      "step": 184716
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.5310070514678955,
      "learning_rate": 5.6349654086837236e-05,
      "loss": 2.2524,
      "step": 184717
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.3731465339660645,
      "learning_rate": 5.634726757780208e-05,
      "loss": 1.9546,
      "step": 184718
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.439894437789917,
      "learning_rate": 5.634488111406665e-05,
      "loss": 2.1451,
      "step": 184719
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.2848236560821533,
      "learning_rate": 5.6342494695631224e-05,
      "loss": 1.6911,
      "step": 184720
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.319166898727417,
      "learning_rate": 5.6340108322496426e-05,
      "loss": 1.9571,
      "step": 184721
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.349753975868225,
      "learning_rate": 5.6337721994662555e-05,
      "loss": 1.9757,
      "step": 184722
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.2764989137649536,
      "learning_rate": 5.633533571213015e-05,
      "loss": 1.9711,
      "step": 184723
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.4225918054580688,
      "learning_rate": 5.63329494748995e-05,
      "loss": 1.9466,
      "step": 184724
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.3194242715835571,
      "learning_rate": 5.633056328297122e-05,
      "loss": 2.0119,
      "step": 184725
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.2976292371749878,
      "learning_rate": 5.632817713634563e-05,
      "loss": 1.9641,
      "step": 184726
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.4946695566177368,
      "learning_rate": 5.6325791035023305e-05,
      "loss": 2.1008,
      "step": 184727
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.423108458518982,
      "learning_rate": 5.632340497900464e-05,
      "loss": 2.1621,
      "step": 184728
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.2920854091644287,
      "learning_rate": 5.632101896829e-05,
      "loss": 2.0166,
      "step": 184729
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.340180516242981,
      "learning_rate": 5.6318633002879855e-05,
      "loss": 2.284,
      "step": 184730
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.3428616523742676,
      "learning_rate": 5.631624708277473e-05,
      "loss": 1.9913,
      "step": 184731
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.3706032037734985,
      "learning_rate": 5.6313861207974943e-05,
      "loss": 2.1241,
      "step": 184732
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.3378612995147705,
      "learning_rate": 5.6311475378481076e-05,
      "loss": 1.7225,
      "step": 184733
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.4505170583724976,
      "learning_rate": 5.630908959429353e-05,
      "loss": 2.2445,
      "step": 184734
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.264810562133789,
      "learning_rate": 5.630670385541268e-05,
      "loss": 1.9133,
      "step": 184735
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.45807945728302,
      "learning_rate": 5.630431816183899e-05,
      "loss": 2.1456,
      "step": 184736
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.3534191846847534,
      "learning_rate": 5.6301932513572954e-05,
      "loss": 2.2735,
      "step": 184737
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.3868745565414429,
      "learning_rate": 5.6299546910614945e-05,
      "loss": 1.9359,
      "step": 184738
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.3428627252578735,
      "learning_rate": 5.629716135296553e-05,
      "loss": 1.8853,
      "step": 184739
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.5551234483718872,
      "learning_rate": 5.629477584062506e-05,
      "loss": 1.9107,
      "step": 184740
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.4754197597503662,
      "learning_rate": 5.629239037359399e-05,
      "loss": 1.9829,
      "step": 184741
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.388734221458435,
      "learning_rate": 5.6290004951872704e-05,
      "loss": 2.1774,
      "step": 184742
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.2753701210021973,
      "learning_rate": 5.6287619575461775e-05,
      "loss": 2.1769,
      "step": 184743
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.3583753108978271,
      "learning_rate": 5.62852342443615e-05,
      "loss": 2.0415,
      "step": 184744
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.359154462814331,
      "learning_rate": 5.628284895857248e-05,
      "loss": 2.0128,
      "step": 184745
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.4034396409988403,
      "learning_rate": 5.628046371809505e-05,
      "loss": 2.0669,
      "step": 184746
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.4058228731155396,
      "learning_rate": 5.62780785229297e-05,
      "loss": 2.2484,
      "step": 184747
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.3514999151229858,
      "learning_rate": 5.6275693373076815e-05,
      "loss": 1.9334,
      "step": 184748
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.3325779438018799,
      "learning_rate": 5.627330826853691e-05,
      "loss": 1.9007,
      "step": 184749
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.6230311393737793,
      "learning_rate": 5.6270923209310325e-05,
      "loss": 2.0079,
      "step": 184750
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.3154886960983276,
      "learning_rate": 5.626853819539766e-05,
      "loss": 2.0158,
      "step": 184751
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.358543872833252,
      "learning_rate": 5.6266153226799284e-05,
      "loss": 2.0948,
      "step": 184752
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.3034694194793701,
      "learning_rate": 5.626376830351562e-05,
      "loss": 1.9734,
      "step": 184753
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.3242212533950806,
      "learning_rate": 5.626138342554708e-05,
      "loss": 1.912,
      "step": 184754
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.3023183345794678,
      "learning_rate": 5.625899859289419e-05,
      "loss": 1.9812,
      "step": 184755
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.4000747203826904,
      "learning_rate": 5.625661380555728e-05,
      "loss": 2.0237,
      "step": 184756
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.3332701921463013,
      "learning_rate": 5.625422906353696e-05,
      "loss": 2.0945,
      "step": 184757
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.4992427825927734,
      "learning_rate": 5.625184436683355e-05,
      "loss": 1.9821,
      "step": 184758
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.5243721008300781,
      "learning_rate": 5.624945971544756e-05,
      "loss": 1.8771,
      "step": 184759
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.335844874382019,
      "learning_rate": 5.624707510937929e-05,
      "loss": 1.8557,
      "step": 184760
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.3741636276245117,
      "learning_rate": 5.6244690548629365e-05,
      "loss": 1.8887,
      "step": 184761
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.380470633506775,
      "learning_rate": 5.624230603319809e-05,
      "loss": 2.0268,
      "step": 184762
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.2242459058761597,
      "learning_rate": 5.6239921563086064e-05,
      "loss": 1.9684,
      "step": 184763
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.302621841430664,
      "learning_rate": 5.623753713829355e-05,
      "loss": 1.9729,
      "step": 184764
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.2659578323364258,
      "learning_rate": 5.6235152758821224e-05,
      "loss": 1.8489,
      "step": 184765
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.5233244895935059,
      "learning_rate": 5.623276842466924e-05,
      "loss": 1.9121,
      "step": 184766
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.422717809677124,
      "learning_rate": 5.623038413583823e-05,
      "loss": 2.0657,
      "step": 184767
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.3008787631988525,
      "learning_rate": 5.6227999892328544e-05,
      "loss": 2.0837,
      "step": 184768
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.389275074005127,
      "learning_rate": 5.622561569414077e-05,
      "loss": 1.9309,
      "step": 184769
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.359086036682129,
      "learning_rate": 5.6223231541275174e-05,
      "loss": 1.9832,
      "step": 184770
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.3970704078674316,
      "learning_rate": 5.622084743373232e-05,
      "loss": 2.0949,
      "step": 184771
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.2959647178649902,
      "learning_rate": 5.6218463371512645e-05,
      "loss": 2.0467,
      "step": 184772
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.234228491783142,
      "learning_rate": 5.621607935461655e-05,
      "loss": 1.8814,
      "step": 184773
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.2541152238845825,
      "learning_rate": 5.62136953830444e-05,
      "loss": 1.8608,
      "step": 184774
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.345360517501831,
      "learning_rate": 5.6211311456796823e-05,
      "loss": 1.7551,
      "step": 184775
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.4362071752548218,
      "learning_rate": 5.620892757587406e-05,
      "loss": 1.8904,
      "step": 184776
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.4128199815750122,
      "learning_rate": 5.620654374027678e-05,
      "loss": 1.886,
      "step": 184777
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.4524580240249634,
      "learning_rate": 5.6204159950005266e-05,
      "loss": 1.9539,
      "step": 184778
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.3955867290496826,
      "learning_rate": 5.620177620506e-05,
      "loss": 2.0313,
      "step": 184779
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.2838423252105713,
      "learning_rate": 5.619939250544138e-05,
      "loss": 2.2429,
      "step": 184780
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.2828227281570435,
      "learning_rate": 5.619700885114996e-05,
      "loss": 1.9056,
      "step": 184781
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.39385187625885,
      "learning_rate": 5.619462524218602e-05,
      "loss": 2.3444,
      "step": 184782
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.4361425638198853,
      "learning_rate": 5.619224167855022e-05,
      "loss": 2.0588,
      "step": 184783
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.3041425943374634,
      "learning_rate": 5.618985816024286e-05,
      "loss": 2.0251,
      "step": 184784
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.4261927604675293,
      "learning_rate": 5.618747468726431e-05,
      "loss": 2.0501,
      "step": 184785
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.409796118736267,
      "learning_rate": 5.61850912596152e-05,
      "loss": 2.012,
      "step": 184786
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.3624123334884644,
      "learning_rate": 5.6182707877295895e-05,
      "loss": 2.1306,
      "step": 184787
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.3901430368423462,
      "learning_rate": 5.6180324540306766e-05,
      "loss": 1.9369,
      "step": 184788
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.4588648080825806,
      "learning_rate": 5.617794124864837e-05,
      "loss": 2.07,
      "step": 184789
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.2850792407989502,
      "learning_rate": 5.6175558002321083e-05,
      "loss": 1.9841,
      "step": 184790
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.2521393299102783,
      "learning_rate": 5.617317480132534e-05,
      "loss": 2.2535,
      "step": 184791
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.4781484603881836,
      "learning_rate": 5.617079164566163e-05,
      "loss": 2.1688,
      "step": 184792
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.3413443565368652,
      "learning_rate": 5.616840853533032e-05,
      "loss": 2.0669,
      "step": 184793
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.412914752960205,
      "learning_rate": 5.6166025470331956e-05,
      "loss": 2.0026,
      "step": 184794
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.3258517980575562,
      "learning_rate": 5.6163642450666956e-05,
      "loss": 1.823,
      "step": 184795
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.4517793655395508,
      "learning_rate": 5.616125947633573e-05,
      "loss": 1.9802,
      "step": 184796
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.4904696941375732,
      "learning_rate": 5.6158876547338673e-05,
      "loss": 1.9953,
      "step": 184797
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.46457839012146,
      "learning_rate": 5.615649366367632e-05,
      "loss": 2.1685,
      "step": 184798
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.4061461687088013,
      "learning_rate": 5.615411082534903e-05,
      "loss": 1.9419,
      "step": 184799
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.5878859758377075,
      "learning_rate": 5.615172803235738e-05,
      "loss": 2.1801,
      "step": 184800
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.4401015043258667,
      "learning_rate": 5.6149345284701696e-05,
      "loss": 2.0057,
      "step": 184801
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.396331787109375,
      "learning_rate": 5.614696258238248e-05,
      "loss": 1.8254,
      "step": 184802
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.335776925086975,
      "learning_rate": 5.6144579925400036e-05,
      "loss": 2.0175,
      "step": 184803
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.359634518623352,
      "learning_rate": 5.614219731375502e-05,
      "loss": 1.8655,
      "step": 184804
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.3436907529830933,
      "learning_rate": 5.6139814747447676e-05,
      "loss": 2.0058,
      "step": 184805
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.408443808555603,
      "learning_rate": 5.613743222647866e-05,
      "loss": 1.8325,
      "step": 184806
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.5801023244857788,
      "learning_rate": 5.613504975084825e-05,
      "loss": 2.1175,
      "step": 184807
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.2933367490768433,
      "learning_rate": 5.6132667320556964e-05,
      "loss": 1.9828,
      "step": 184808
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.2727478742599487,
      "learning_rate": 5.613028493560515e-05,
      "loss": 1.9311,
      "step": 184809
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.3783999681472778,
      "learning_rate": 5.612790259599337e-05,
      "loss": 1.8696,
      "step": 184810
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.545684814453125,
      "learning_rate": 5.612552030172195e-05,
      "loss": 1.9167,
      "step": 184811
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.4218961000442505,
      "learning_rate": 5.612313805279149e-05,
      "loss": 2.1842,
      "step": 184812
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.50127112865448,
      "learning_rate": 5.612075584920234e-05,
      "loss": 1.9692,
      "step": 184813
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.4079312086105347,
      "learning_rate": 5.6118373690954944e-05,
      "loss": 2.1563,
      "step": 184814
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.2019009590148926,
      "learning_rate": 5.6115991578049644e-05,
      "loss": 1.967,
      "step": 184815
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.3362596035003662,
      "learning_rate": 5.611360951048708e-05,
      "loss": 1.7667,
      "step": 184816
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.4044572114944458,
      "learning_rate": 5.611122748826754e-05,
      "loss": 2.0744,
      "step": 184817
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.5710368156433105,
      "learning_rate": 5.6108845511391565e-05,
      "loss": 2.0986,
      "step": 184818
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.3838636875152588,
      "learning_rate": 5.610646357985959e-05,
      "loss": 1.8347,
      "step": 184819
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.3357568979263306,
      "learning_rate": 5.610408169367203e-05,
      "loss": 2.1454,
      "step": 184820
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.4361059665679932,
      "learning_rate": 5.6101699852829244e-05,
      "loss": 1.7603,
      "step": 184821
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.3211605548858643,
      "learning_rate": 5.609931805733181e-05,
      "loss": 1.9771,
      "step": 184822
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.4140363931655884,
      "learning_rate": 5.609693630718004e-05,
      "loss": 2.024,
      "step": 184823
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.568613886833191,
      "learning_rate": 5.6094554602374566e-05,
      "loss": 1.8775,
      "step": 184824
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.314874291419983,
      "learning_rate": 5.609217294291568e-05,
      "loss": 1.9313,
      "step": 184825
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.3996225595474243,
      "learning_rate": 5.608979132880389e-05,
      "loss": 1.9653,
      "step": 184826
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.4074225425720215,
      "learning_rate": 5.6087409760039525e-05,
      "loss": 2.0993,
      "step": 184827
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.2268671989440918,
      "learning_rate": 5.6085028236623186e-05,
      "loss": 1.9308,
      "step": 184828
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.294275164604187,
      "learning_rate": 5.6082646758555175e-05,
      "loss": 2.166,
      "step": 184829
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.3856903314590454,
      "learning_rate": 5.608026532583608e-05,
      "loss": 2.1167,
      "step": 184830
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.2473366260528564,
      "learning_rate": 5.6077883938466184e-05,
      "loss": 2.0186,
      "step": 184831
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.403716802597046,
      "learning_rate": 5.607550259644618e-05,
      "loss": 2.0677,
      "step": 184832
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.2841943502426147,
      "learning_rate": 5.607312129977619e-05,
      "loss": 2.0058,
      "step": 184833
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.3210989236831665,
      "learning_rate": 5.60707400484569e-05,
      "loss": 2.0266,
      "step": 184834
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.5011056661605835,
      "learning_rate": 5.606835884248856e-05,
      "loss": 1.988,
      "step": 184835
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.3219623565673828,
      "learning_rate": 5.606597768187181e-05,
      "loss": 2.2109,
      "step": 184836
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.418321132659912,
      "learning_rate": 5.6063596566606914e-05,
      "loss": 2.026,
      "step": 184837
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.2471120357513428,
      "learning_rate": 5.606121549669454e-05,
      "loss": 1.7783,
      "step": 184838
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.3351002931594849,
      "learning_rate": 5.605883447213486e-05,
      "loss": 2.0688,
      "step": 184839
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.4436373710632324,
      "learning_rate": 5.605645349292853e-05,
      "loss": 2.0138,
      "step": 184840
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.3810933828353882,
      "learning_rate": 5.605407255907582e-05,
      "loss": 2.046,
      "step": 184841
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.317296028137207,
      "learning_rate": 5.6051691670577335e-05,
      "loss": 2.0782,
      "step": 184842
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.3600813150405884,
      "learning_rate": 5.604931082743337e-05,
      "loss": 1.883,
      "step": 184843
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.3312866687774658,
      "learning_rate": 5.6046930029644624e-05,
      "loss": 1.8011,
      "step": 184844
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.2703386545181274,
      "learning_rate": 5.604454927721116e-05,
      "loss": 1.8527,
      "step": 184845
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.4679298400878906,
      "learning_rate": 5.604216857013372e-05,
      "loss": 2.049,
      "step": 184846
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.7275630235671997,
      "learning_rate": 5.6039787908412556e-05,
      "loss": 1.7482,
      "step": 184847
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.4828987121582031,
      "learning_rate": 5.6037407292048286e-05,
      "loss": 1.8958,
      "step": 184848
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.4308432340621948,
      "learning_rate": 5.60350267210412e-05,
      "loss": 2.1386,
      "step": 184849
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.4002525806427002,
      "learning_rate": 5.603264619539196e-05,
      "loss": 2.0955,
      "step": 184850
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.33895742893219,
      "learning_rate": 5.603026571510071e-05,
      "loss": 2.0323,
      "step": 184851
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.411162257194519,
      "learning_rate": 5.6027885280168106e-05,
      "loss": 2.0825,
      "step": 184852
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.6660966873168945,
      "learning_rate": 5.602550489059442e-05,
      "loss": 1.9461,
      "step": 184853
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.3700259923934937,
      "learning_rate": 5.602312454638032e-05,
      "loss": 2.027,
      "step": 184854
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.3739960193634033,
      "learning_rate": 5.602074424752603e-05,
      "loss": 2.0603,
      "step": 184855
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.2395144701004028,
      "learning_rate": 5.601836399403222e-05,
      "loss": 1.9403,
      "step": 184856
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.3846213817596436,
      "learning_rate": 5.601598378589906e-05,
      "loss": 2.074,
      "step": 184857
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.297804355621338,
      "learning_rate": 5.6013603623127225e-05,
      "loss": 1.9267,
      "step": 184858
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.3105562925338745,
      "learning_rate": 5.6011223505716996e-05,
      "loss": 1.9835,
      "step": 184859
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.2533490657806396,
      "learning_rate": 5.6008843433668945e-05,
      "loss": 1.962,
      "step": 184860
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.3793271780014038,
      "learning_rate": 5.600646340698338e-05,
      "loss": 2.0251,
      "step": 184861
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.5464670658111572,
      "learning_rate": 5.6004083425660925e-05,
      "loss": 2.1238,
      "step": 184862
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.2953683137893677,
      "learning_rate": 5.600170348970189e-05,
      "loss": 2.0731,
      "step": 184863
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.5245585441589355,
      "learning_rate": 5.599932359910673e-05,
      "loss": 2.0856,
      "step": 184864
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.5700123310089111,
      "learning_rate": 5.599694375387585e-05,
      "loss": 1.9762,
      "step": 184865
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.2862317562103271,
      "learning_rate": 5.5994563954009806e-05,
      "loss": 1.9545,
      "step": 184866
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.3600677251815796,
      "learning_rate": 5.599218419950892e-05,
      "loss": 2.0029,
      "step": 184867
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.3588967323303223,
      "learning_rate": 5.598980449037377e-05,
      "loss": 1.9124,
      "step": 184868
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.414587140083313,
      "learning_rate": 5.59874248266047e-05,
      "loss": 1.9016,
      "step": 184869
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.4810611009597778,
      "learning_rate": 5.5985045208202104e-05,
      "loss": 1.8564,
      "step": 184870
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.3638485670089722,
      "learning_rate": 5.5982665635166554e-05,
      "loss": 1.9151,
      "step": 184871
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.5675020217895508,
      "learning_rate": 5.598028610749848e-05,
      "loss": 2.2976,
      "step": 184872
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.4548569917678833,
      "learning_rate": 5.597790662519816e-05,
      "loss": 1.8559,
      "step": 184873
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.3737326860427856,
      "learning_rate": 5.597552718826623e-05,
      "loss": 1.6927,
      "step": 184874
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.3615611791610718,
      "learning_rate": 5.5973147796703086e-05,
      "loss": 1.9317,
      "step": 184875
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.297074794769287,
      "learning_rate": 5.597076845050905e-05,
      "loss": 2.0656,
      "step": 184876
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.3535643815994263,
      "learning_rate": 5.596838914968472e-05,
      "loss": 1.8559,
      "step": 184877
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.3990157842636108,
      "learning_rate": 5.596600989423047e-05,
      "loss": 2.2331,
      "step": 184878
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.4628750085830688,
      "learning_rate": 5.5963630684146685e-05,
      "loss": 2.0141,
      "step": 184879
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.2499339580535889,
      "learning_rate": 5.5961251519433944e-05,
      "loss": 1.6206,
      "step": 184880
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.266737699508667,
      "learning_rate": 5.595887240009261e-05,
      "loss": 1.8653,
      "step": 184881
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.4621435403823853,
      "learning_rate": 5.595649332612305e-05,
      "loss": 2.1171,
      "step": 184882
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.4178287982940674,
      "learning_rate": 5.5954114297525864e-05,
      "loss": 2.1341,
      "step": 184883
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.26067316532135,
      "learning_rate": 5.595173531430132e-05,
      "loss": 1.8446,
      "step": 184884
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.341833472251892,
      "learning_rate": 5.594935637645004e-05,
      "loss": 1.8515,
      "step": 184885
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.3650890588760376,
      "learning_rate": 5.59469774839724e-05,
      "loss": 1.8946,
      "step": 184886
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.3574298620224,
      "learning_rate": 5.59445986368688e-05,
      "loss": 2.0282,
      "step": 184887
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.4140045642852783,
      "learning_rate": 5.594221983513967e-05,
      "loss": 1.9987,
      "step": 184888
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.4065862894058228,
      "learning_rate": 5.593984107878551e-05,
      "loss": 2.094,
      "step": 184889
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.4686640501022339,
      "learning_rate": 5.5937462367806683e-05,
      "loss": 2.2522,
      "step": 184890
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.3940720558166504,
      "learning_rate": 5.59350837022038e-05,
      "loss": 2.0023,
      "step": 184891
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.5187665224075317,
      "learning_rate": 5.5932705081977144e-05,
      "loss": 2.1741,
      "step": 184892
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.3821005821228027,
      "learning_rate": 5.593032650712723e-05,
      "loss": 2.1563,
      "step": 184893
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.2742674350738525,
      "learning_rate": 5.5927947977654385e-05,
      "loss": 2.1992,
      "step": 184894
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.350053071975708,
      "learning_rate": 5.592556949355924e-05,
      "loss": 1.9495,
      "step": 184895
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.2829371690750122,
      "learning_rate": 5.592319105484203e-05,
      "loss": 1.9944,
      "step": 184896
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.3998810052871704,
      "learning_rate": 5.5920812661503426e-05,
      "loss": 2.1955,
      "step": 184897
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.334168553352356,
      "learning_rate": 5.591843431354365e-05,
      "loss": 1.9343,
      "step": 184898
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.3746423721313477,
      "learning_rate": 5.59160560109634e-05,
      "loss": 1.816,
      "step": 184899
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.4282416105270386,
      "learning_rate": 5.59136777537628e-05,
      "loss": 1.9475,
      "step": 184900
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.3794630765914917,
      "learning_rate": 5.591129954194252e-05,
      "loss": 2.0061,
      "step": 184901
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.5478595495224,
      "learning_rate": 5.590892137550288e-05,
      "loss": 1.8134,
      "step": 184902
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.3746792078018188,
      "learning_rate": 5.590654325444447e-05,
      "loss": 1.7944,
      "step": 184903
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.3551217317581177,
      "learning_rate": 5.5904165178767554e-05,
      "loss": 2.0285,
      "step": 184904
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.3915938138961792,
      "learning_rate": 5.590178714847281e-05,
      "loss": 2.32,
      "step": 184905
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.4465891122817993,
      "learning_rate": 5.589940916356036e-05,
      "loss": 2.1275,
      "step": 184906
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.2984483242034912,
      "learning_rate": 5.5897031224030906e-05,
      "loss": 1.8722,
      "step": 184907
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.4181183576583862,
      "learning_rate": 5.589465332988475e-05,
      "loss": 2.0863,
      "step": 184908
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.3231990337371826,
      "learning_rate": 5.589227548112243e-05,
      "loss": 1.7674,
      "step": 184909
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.3813198804855347,
      "learning_rate": 5.58898976777443e-05,
      "loss": 2.0329,
      "step": 184910
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.6020832061767578,
      "learning_rate": 5.5887519919750965e-05,
      "loss": 1.9988,
      "step": 184911
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.3479853868484497,
      "learning_rate": 5.588514220714263e-05,
      "loss": 1.9818,
      "step": 184912
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.4110113382339478,
      "learning_rate": 5.588276453991989e-05,
      "loss": 1.9785,
      "step": 184913
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.3193498849868774,
      "learning_rate": 5.588038691808311e-05,
      "loss": 2.0825,
      "step": 184914
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.4131029844284058,
      "learning_rate": 5.5878009341632823e-05,
      "loss": 1.9633,
      "step": 184915
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.265167474746704,
      "learning_rate": 5.587563181056937e-05,
      "loss": 2.0159,
      "step": 184916
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.4713374376296997,
      "learning_rate": 5.58732543248934e-05,
      "loss": 2.0223,
      "step": 184917
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.3303090333938599,
      "learning_rate": 5.587087688460503e-05,
      "loss": 1.9039,
      "step": 184918
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.5348938703536987,
      "learning_rate": 5.5868499489704956e-05,
      "loss": 2.1727,
      "step": 184919
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.2798192501068115,
      "learning_rate": 5.586612214019347e-05,
      "loss": 2.1729,
      "step": 184920
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.363369345664978,
      "learning_rate": 5.586374483607115e-05,
      "loss": 2.0651,
      "step": 184921
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.2584197521209717,
      "learning_rate": 5.586136757733828e-05,
      "loss": 1.8895,
      "step": 184922
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.451592206954956,
      "learning_rate": 5.585899036399554e-05,
      "loss": 2.2581,
      "step": 184923
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.5536667108535767,
      "learning_rate": 5.5856613196043084e-05,
      "loss": 1.9556,
      "step": 184924
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.3299564123153687,
      "learning_rate": 5.585423607348156e-05,
      "loss": 2.1646,
      "step": 184925
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.4138528108596802,
      "learning_rate": 5.585185899631126e-05,
      "loss": 1.8039,
      "step": 184926
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.351698637008667,
      "learning_rate": 5.584948196453282e-05,
      "loss": 2.1816,
      "step": 184927
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.505188226699829,
      "learning_rate": 5.5847104978146465e-05,
      "loss": 1.8859,
      "step": 184928
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.827713966369629,
      "learning_rate": 5.58447280371529e-05,
      "loss": 2.0246,
      "step": 184929
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.2978601455688477,
      "learning_rate": 5.584235114155222e-05,
      "loss": 1.9824,
      "step": 184930
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.3249878883361816,
      "learning_rate": 5.583997429134517e-05,
      "loss": 1.8944,
      "step": 184931
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.4326953887939453,
      "learning_rate": 5.5837597486532005e-05,
      "loss": 2.0217,
      "step": 184932
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.4391921758651733,
      "learning_rate": 5.583522072711333e-05,
      "loss": 1.989,
      "step": 184933
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.3868645429611206,
      "learning_rate": 5.5832844013089375e-05,
      "loss": 2.0631,
      "step": 184934
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.3667280673980713,
      "learning_rate": 5.583046734446087e-05,
      "loss": 2.1899,
      "step": 184935
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.3215090036392212,
      "learning_rate": 5.5828090721227956e-05,
      "loss": 2.0739,
      "step": 184936
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.4349421262741089,
      "learning_rate": 5.5825714143391295e-05,
      "loss": 1.9593,
      "step": 184937
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.4387794733047485,
      "learning_rate": 5.582333761095112e-05,
      "loss": 2.1484,
      "step": 184938
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.558538556098938,
      "learning_rate": 5.5820961123908106e-05,
      "loss": 1.8229,
      "step": 184939
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.3997480869293213,
      "learning_rate": 5.58185846822625e-05,
      "loss": 2.0632,
      "step": 184940
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.395245909690857,
      "learning_rate": 5.5816208286014984e-05,
      "loss": 1.9154,
      "step": 184941
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.5061050653457642,
      "learning_rate": 5.5813831935165684e-05,
      "loss": 1.9933,
      "step": 184942
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.3659042119979858,
      "learning_rate": 5.5811455629715304e-05,
      "loss": 2.1222,
      "step": 184943
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.387943148612976,
      "learning_rate": 5.580907936966408e-05,
      "loss": 2.3018,
      "step": 184944
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.4728903770446777,
      "learning_rate": 5.580670315501263e-05,
      "loss": 2.0119,
      "step": 184945
    },
    {
      "epoch": 2.41,
      "grad_norm": 2.4235568046569824,
      "learning_rate": 5.580432698576127e-05,
      "loss": 2.1541,
      "step": 184946
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.6216408014297485,
      "learning_rate": 5.580195086191056e-05,
      "loss": 1.9793,
      "step": 184947
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.3483349084854126,
      "learning_rate": 5.57995747834609e-05,
      "loss": 2.0957,
      "step": 184948
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.3542146682739258,
      "learning_rate": 5.579719875041269e-05,
      "loss": 1.9538,
      "step": 184949
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.4622948169708252,
      "learning_rate": 5.579482276276629e-05,
      "loss": 1.8614,
      "step": 184950
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.4994087219238281,
      "learning_rate": 5.579244682052234e-05,
      "loss": 2.213,
      "step": 184951
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.4778456687927246,
      "learning_rate": 5.5790070923681105e-05,
      "loss": 1.8784,
      "step": 184952
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.5024733543395996,
      "learning_rate": 5.578769507224319e-05,
      "loss": 1.915,
      "step": 184953
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.3301630020141602,
      "learning_rate": 5.578531926620895e-05,
      "loss": 2.0533,
      "step": 184954
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.395498514175415,
      "learning_rate": 5.578294350557873e-05,
      "loss": 2.064,
      "step": 184955
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.3805571794509888,
      "learning_rate": 5.5780567790353184e-05,
      "loss": 1.9128,
      "step": 184956
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.4134483337402344,
      "learning_rate": 5.577819212053262e-05,
      "loss": 2.157,
      "step": 184957
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.4435508251190186,
      "learning_rate": 5.577581649611743e-05,
      "loss": 2.2588,
      "step": 184958
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.3644338846206665,
      "learning_rate": 5.577344091710819e-05,
      "loss": 2.0918,
      "step": 184959
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.3928335905075073,
      "learning_rate": 5.5771065383505297e-05,
      "loss": 1.8941,
      "step": 184960
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.3062455654144287,
      "learning_rate": 5.5768689895309085e-05,
      "loss": 2.0614,
      "step": 184961
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.4779014587402344,
      "learning_rate": 5.576631445252014e-05,
      "loss": 2.0868,
      "step": 184962
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.3199113607406616,
      "learning_rate": 5.576393905513889e-05,
      "loss": 2.0447,
      "step": 184963
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.4838268756866455,
      "learning_rate": 5.576156370316564e-05,
      "loss": 1.9758,
      "step": 184964
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.4277158975601196,
      "learning_rate": 5.575918839660099e-05,
      "loss": 1.886,
      "step": 184965
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.3694894313812256,
      "learning_rate": 5.5756813135445335e-05,
      "loss": 2.1067,
      "step": 184966
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.3801898956298828,
      "learning_rate": 5.575443791969898e-05,
      "loss": 1.9769,
      "step": 184967
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.3143517971038818,
      "learning_rate": 5.57520627493626e-05,
      "loss": 1.9554,
      "step": 184968
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.2519725561141968,
      "learning_rate": 5.574968762443643e-05,
      "loss": 2.0035,
      "step": 184969
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.451154112815857,
      "learning_rate": 5.574731254492111e-05,
      "loss": 2.0353,
      "step": 184970
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.3096654415130615,
      "learning_rate": 5.574493751081696e-05,
      "loss": 1.8054,
      "step": 184971
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.3195167779922485,
      "learning_rate": 5.574256252212441e-05,
      "loss": 2.143,
      "step": 184972
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.2985241413116455,
      "learning_rate": 5.5740187578843874e-05,
      "loss": 1.979,
      "step": 184973
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.3751505613327026,
      "learning_rate": 5.5737812680975946e-05,
      "loss": 1.9423,
      "step": 184974
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.413467526435852,
      "learning_rate": 5.5735437828520855e-05,
      "loss": 1.8935,
      "step": 184975
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.4471487998962402,
      "learning_rate": 5.5733063021479274e-05,
      "loss": 1.9221,
      "step": 184976
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.2826557159423828,
      "learning_rate": 5.57306882598515e-05,
      "loss": 2.0874,
      "step": 184977
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.3861725330352783,
      "learning_rate": 5.572831354363799e-05,
      "loss": 2.2368,
      "step": 184978
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.2639775276184082,
      "learning_rate": 5.5725938872839126e-05,
      "loss": 2.0264,
      "step": 184979
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.4265644550323486,
      "learning_rate": 5.572356424745553e-05,
      "loss": 1.8696,
      "step": 184980
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.4257858991622925,
      "learning_rate": 5.572118966748744e-05,
      "loss": 2.0512,
      "step": 184981
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.3914101123809814,
      "learning_rate": 5.571881513293546e-05,
      "loss": 2.2204,
      "step": 184982
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.4961849451065063,
      "learning_rate": 5.571644064379988e-05,
      "loss": 1.9055,
      "step": 184983
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.4037374258041382,
      "learning_rate": 5.5714066200081396e-05,
      "loss": 2.0122,
      "step": 184984
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.4902423620224,
      "learning_rate": 5.571169180178012e-05,
      "loss": 2.1592,
      "step": 184985
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.8129745721817017,
      "learning_rate": 5.570931744889672e-05,
      "loss": 1.9242,
      "step": 184986
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.3704789876937866,
      "learning_rate": 5.5706943141431515e-05,
      "loss": 2.003,
      "step": 184987
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.4034173488616943,
      "learning_rate": 5.5704568879385046e-05,
      "loss": 1.9283,
      "step": 184988
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.3973643779754639,
      "learning_rate": 5.570219466275765e-05,
      "loss": 2.1109,
      "step": 184989
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.2962509393692017,
      "learning_rate": 5.569982049154998e-05,
      "loss": 2.2205,
      "step": 184990
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.4326595067977905,
      "learning_rate": 5.5697446365762154e-05,
      "loss": 2.1456,
      "step": 184991
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.5806964635849,
      "learning_rate": 5.5695072285394895e-05,
      "loss": 2.0425,
      "step": 184992
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.516680121421814,
      "learning_rate": 5.569269825044843e-05,
      "loss": 2.0752,
      "step": 184993
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.3830335140228271,
      "learning_rate": 5.56903242609234e-05,
      "loss": 1.9901,
      "step": 184994
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.3696305751800537,
      "learning_rate": 5.568795031682007e-05,
      "loss": 1.9983,
      "step": 184995
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.3462389707565308,
      "learning_rate": 5.568557641813911e-05,
      "loss": 1.8974,
      "step": 184996
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.540531873703003,
      "learning_rate": 5.568320256488065e-05,
      "loss": 2.0881,
      "step": 184997
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.3356093168258667,
      "learning_rate": 5.568082875704538e-05,
      "loss": 2.048,
      "step": 184998
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.4511667490005493,
      "learning_rate": 5.567845499463358e-05,
      "loss": 2.1714,
      "step": 184999
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.4136524200439453,
      "learning_rate": 5.567608127764585e-05,
      "loss": 2.0615,
      "step": 185000
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.4339464902877808,
      "learning_rate": 5.567370760608244e-05,
      "loss": 1.9853,
      "step": 185001
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.3543447256088257,
      "learning_rate": 5.56713339799441e-05,
      "loss": 1.7823,
      "step": 185002
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.3607158660888672,
      "learning_rate": 5.566896039923089e-05,
      "loss": 1.9902,
      "step": 185003
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.4011836051940918,
      "learning_rate": 5.566658686394351e-05,
      "loss": 2.0836,
      "step": 185004
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.3641111850738525,
      "learning_rate": 5.566421337408226e-05,
      "loss": 1.9685,
      "step": 185005
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.3392921686172485,
      "learning_rate": 5.56618399296477e-05,
      "loss": 2.209,
      "step": 185006
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.4181751012802124,
      "learning_rate": 5.565946653064017e-05,
      "loss": 2.0198,
      "step": 185007
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.4172435998916626,
      "learning_rate": 5.565709317706028e-05,
      "loss": 2.0826,
      "step": 185008
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.2592743635177612,
      "learning_rate": 5.56547198689082e-05,
      "loss": 1.9622,
      "step": 185009
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.3567098379135132,
      "learning_rate": 5.5652346606184626e-05,
      "loss": 1.9368,
      "step": 185010
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.183385968208313,
      "learning_rate": 5.564997338888978e-05,
      "loss": 1.6563,
      "step": 185011
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.312455415725708,
      "learning_rate": 5.564760021702432e-05,
      "loss": 1.9816,
      "step": 185012
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.4040039777755737,
      "learning_rate": 5.5645227090588494e-05,
      "loss": 2.088,
      "step": 185013
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.323939323425293,
      "learning_rate": 5.5642854009582995e-05,
      "loss": 1.9554,
      "step": 185014
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.2821428775787354,
      "learning_rate": 5.564048097400792e-05,
      "loss": 1.8165,
      "step": 185015
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.3936578035354614,
      "learning_rate": 5.5638107983864e-05,
      "loss": 1.8158,
      "step": 185016
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.3022537231445312,
      "learning_rate": 5.563573503915147e-05,
      "loss": 2.0642,
      "step": 185017
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.29957115650177,
      "learning_rate": 5.5633362139870975e-05,
      "loss": 1.9809,
      "step": 185018
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.3635369539260864,
      "learning_rate": 5.5630989286022765e-05,
      "loss": 2.0417,
      "step": 185019
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.4993566274642944,
      "learning_rate": 5.562861647760751e-05,
      "loss": 1.8977,
      "step": 185020
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.4291528463363647,
      "learning_rate": 5.562624371462535e-05,
      "loss": 2.2387,
      "step": 185021
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.258957862854004,
      "learning_rate": 5.562387099707698e-05,
      "loss": 2.0391,
      "step": 185022
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.399129867553711,
      "learning_rate": 5.5621498324962636e-05,
      "loss": 1.8254,
      "step": 185023
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.5775883197784424,
      "learning_rate": 5.5619125698282977e-05,
      "loss": 2.1158,
      "step": 185024
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.3795939683914185,
      "learning_rate": 5.561675311703824e-05,
      "loss": 1.8413,
      "step": 185025
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.4211241006851196,
      "learning_rate": 5.561438058122912e-05,
      "loss": 2.0816,
      "step": 185026
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.406484842300415,
      "learning_rate": 5.561200809085577e-05,
      "loss": 1.7647,
      "step": 185027
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.4277992248535156,
      "learning_rate": 5.560963564591879e-05,
      "loss": 2.1881,
      "step": 185028
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.4083799123764038,
      "learning_rate": 5.560726324641858e-05,
      "loss": 1.9477,
      "step": 185029
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.3638627529144287,
      "learning_rate": 5.560489089235561e-05,
      "loss": 2.1046,
      "step": 185030
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.337219476699829,
      "learning_rate": 5.560251858373024e-05,
      "loss": 1.8131,
      "step": 185031
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.633641004562378,
      "learning_rate": 5.560014632054308e-05,
      "loss": 1.9715,
      "step": 185032
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.498449683189392,
      "learning_rate": 5.559777410279445e-05,
      "loss": 1.9722,
      "step": 185033
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.5560047626495361,
      "learning_rate": 5.55954019304848e-05,
      "loss": 2.0278,
      "step": 185034
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.3568933010101318,
      "learning_rate": 5.55930298036145e-05,
      "loss": 2.1347,
      "step": 185035
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.4065935611724854,
      "learning_rate": 5.559065772218415e-05,
      "loss": 1.9359,
      "step": 185036
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.3315942287445068,
      "learning_rate": 5.558828568619406e-05,
      "loss": 1.947,
      "step": 185037
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.4293923377990723,
      "learning_rate": 5.558591369564475e-05,
      "loss": 1.7078,
      "step": 185038
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.3213181495666504,
      "learning_rate": 5.5583541750536656e-05,
      "loss": 2.0504,
      "step": 185039
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.3042137622833252,
      "learning_rate": 5.5581169850870214e-05,
      "loss": 2.0554,
      "step": 185040
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.338767647743225,
      "learning_rate": 5.5578797996645755e-05,
      "loss": 2.0283,
      "step": 185041
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.341802954673767,
      "learning_rate": 5.5576426187863874e-05,
      "loss": 1.9742,
      "step": 185042
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.3138973712921143,
      "learning_rate": 5.5574054424524885e-05,
      "loss": 1.9404,
      "step": 185043
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.323083758354187,
      "learning_rate": 5.5571682706629375e-05,
      "loss": 1.8961,
      "step": 185044
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.2538913488388062,
      "learning_rate": 5.556931103417768e-05,
      "loss": 1.8019,
      "step": 185045
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.413482427597046,
      "learning_rate": 5.55669394071702e-05,
      "loss": 1.8942,
      "step": 185046
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.436901330947876,
      "learning_rate": 5.556456782560753e-05,
      "loss": 2.0737,
      "step": 185047
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.4863260984420776,
      "learning_rate": 5.5562196289490014e-05,
      "loss": 1.9755,
      "step": 185048
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.3813796043395996,
      "learning_rate": 5.555982479881801e-05,
      "loss": 1.96,
      "step": 185049
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.3962286710739136,
      "learning_rate": 5.555745335359215e-05,
      "loss": 1.971,
      "step": 185050
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.3278436660766602,
      "learning_rate": 5.555508195381274e-05,
      "loss": 2.1133,
      "step": 185051
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.4150971174240112,
      "learning_rate": 5.555271059948021e-05,
      "loss": 2.0229,
      "step": 185052
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.3605693578720093,
      "learning_rate": 5.555033929059509e-05,
      "loss": 2.1513,
      "step": 185053
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.4292716979980469,
      "learning_rate": 5.5547968027157716e-05,
      "loss": 2.2001,
      "step": 185054
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.391458511352539,
      "learning_rate": 5.554559680916869e-05,
      "loss": 2.0362,
      "step": 185055
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.6040946245193481,
      "learning_rate": 5.554322563662834e-05,
      "loss": 2.056,
      "step": 185056
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.4419045448303223,
      "learning_rate": 5.5540854509537104e-05,
      "loss": 1.9755,
      "step": 185057
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.436281442642212,
      "learning_rate": 5.553848342789534e-05,
      "loss": 1.8918,
      "step": 185058
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.451132893562317,
      "learning_rate": 5.553611239170369e-05,
      "loss": 1.715,
      "step": 185059
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.4274928569793701,
      "learning_rate": 5.553374140096242e-05,
      "loss": 2.007,
      "step": 185060
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.3731062412261963,
      "learning_rate": 5.553137045567212e-05,
      "loss": 2.111,
      "step": 185061
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.4044203758239746,
      "learning_rate": 5.5528999555833164e-05,
      "loss": 2.0713,
      "step": 185062
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.3810607194900513,
      "learning_rate": 5.5526628701445955e-05,
      "loss": 1.908,
      "step": 185063
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.57387113571167,
      "learning_rate": 5.552425789251088e-05,
      "loss": 2.021,
      "step": 185064
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.366500973701477,
      "learning_rate": 5.5521887129028555e-05,
      "loss": 2.0823,
      "step": 185065
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.5048857927322388,
      "learning_rate": 5.551951641099923e-05,
      "loss": 1.749,
      "step": 185066
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.300762414932251,
      "learning_rate": 5.551714573842355e-05,
      "loss": 2.0063,
      "step": 185067
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.4238574504852295,
      "learning_rate": 5.5514775111301746e-05,
      "loss": 2.1904,
      "step": 185068
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.4236749410629272,
      "learning_rate": 5.5512404529634515e-05,
      "loss": 1.7636,
      "step": 185069
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.4290757179260254,
      "learning_rate": 5.551003399342202e-05,
      "loss": 1.8313,
      "step": 185070
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.3844176530838013,
      "learning_rate": 5.550766350266487e-05,
      "loss": 2.1504,
      "step": 185071
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.6294385194778442,
      "learning_rate": 5.550529305736339e-05,
      "loss": 2.283,
      "step": 185072
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.5604573488235474,
      "learning_rate": 5.550292265751815e-05,
      "loss": 2.0376,
      "step": 185073
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.42000150680542,
      "learning_rate": 5.5500552303129474e-05,
      "loss": 2.3678,
      "step": 185074
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.3361238241195679,
      "learning_rate": 5.549818199419801e-05,
      "loss": 2.1383,
      "step": 185075
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.2376949787139893,
      "learning_rate": 5.5495811730723914e-05,
      "loss": 2.3512,
      "step": 185076
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.6027982234954834,
      "learning_rate": 5.549344151270783e-05,
      "loss": 1.9581,
      "step": 185077
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.6877549886703491,
      "learning_rate": 5.549107134015004e-05,
      "loss": 2.0225,
      "step": 185078
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.4507321119308472,
      "learning_rate": 5.5488701213051154e-05,
      "loss": 2.2491,
      "step": 185079
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.3966789245605469,
      "learning_rate": 5.548633113141148e-05,
      "loss": 2.1256,
      "step": 185080
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.3287606239318848,
      "learning_rate": 5.5483961095231636e-05,
      "loss": 1.9015,
      "step": 185081
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.4203487634658813,
      "learning_rate": 5.548159110451179e-05,
      "loss": 1.903,
      "step": 185082
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.3282642364501953,
      "learning_rate": 5.547922115925265e-05,
      "loss": 2.0026,
      "step": 185083
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.4124863147735596,
      "learning_rate": 5.5476851259454415e-05,
      "loss": 2.0045,
      "step": 185084
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.2561664581298828,
      "learning_rate": 5.5474481405117746e-05,
      "loss": 1.8885,
      "step": 185085
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.3310118913650513,
      "learning_rate": 5.547211159624291e-05,
      "loss": 1.9717,
      "step": 185086
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.2948189973831177,
      "learning_rate": 5.546974183283057e-05,
      "loss": 1.9513,
      "step": 185087
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.2861472368240356,
      "learning_rate": 5.546737211488087e-05,
      "loss": 2.0623,
      "step": 185088
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.4245656728744507,
      "learning_rate": 5.546500244239447e-05,
      "loss": 1.8914,
      "step": 185089
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.4952647686004639,
      "learning_rate": 5.546263281537166e-05,
      "loss": 1.8775,
      "step": 185090
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.744638442993164,
      "learning_rate": 5.5460263233813064e-05,
      "loss": 2.1079,
      "step": 185091
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.4789165258407593,
      "learning_rate": 5.5457893697718925e-05,
      "loss": 1.9163,
      "step": 185092
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.4921644926071167,
      "learning_rate": 5.545552420708995e-05,
      "loss": 1.9676,
      "step": 185093
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.5374711751937866,
      "learning_rate": 5.545315476192624e-05,
      "loss": 2.0651,
      "step": 185094
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.3568669557571411,
      "learning_rate": 5.5450785362228504e-05,
      "loss": 1.9133,
      "step": 185095
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.3670704364776611,
      "learning_rate": 5.544841600799699e-05,
      "loss": 2.0324,
      "step": 185096
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.322134256362915,
      "learning_rate": 5.544604669923231e-05,
      "loss": 2.1269,
      "step": 185097
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.3349307775497437,
      "learning_rate": 5.544367743593475e-05,
      "loss": 1.9544,
      "step": 185098
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.3461705446243286,
      "learning_rate": 5.5441308218104895e-05,
      "loss": 2.1808,
      "step": 185099
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.2164493799209595,
      "learning_rate": 5.5438939045743136e-05,
      "loss": 1.9857,
      "step": 185100
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.5701614618301392,
      "learning_rate": 5.543656991884987e-05,
      "loss": 2.0499,
      "step": 185101
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.616464614868164,
      "learning_rate": 5.5434200837425505e-05,
      "loss": 1.8347,
      "step": 185102
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.3073015213012695,
      "learning_rate": 5.5431831801470596e-05,
      "loss": 2.1734,
      "step": 185103
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.5159969329833984,
      "learning_rate": 5.542946281098548e-05,
      "loss": 2.2449,
      "step": 185104
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.272796630859375,
      "learning_rate": 5.54270938659707e-05,
      "loss": 1.9245,
      "step": 185105
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.3653860092163086,
      "learning_rate": 5.5424724966426636e-05,
      "loss": 2.1241,
      "step": 185106
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.3710386753082275,
      "learning_rate": 5.5422356112353736e-05,
      "loss": 2.0115,
      "step": 185107
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.3548976182937622,
      "learning_rate": 5.541998730375237e-05,
      "loss": 2.159,
      "step": 185108
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.3545581102371216,
      "learning_rate": 5.541761854062309e-05,
      "loss": 2.1899,
      "step": 185109
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.6822330951690674,
      "learning_rate": 5.5415249822966266e-05,
      "loss": 1.8761,
      "step": 185110
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.4541332721710205,
      "learning_rate": 5.541288115078241e-05,
      "loss": 2.0374,
      "step": 185111
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.3187462091445923,
      "learning_rate": 5.54105125240719e-05,
      "loss": 1.8466,
      "step": 185112
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.5185080766677856,
      "learning_rate": 5.5408143942835226e-05,
      "loss": 2.0227,
      "step": 185113
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.46473228931427,
      "learning_rate": 5.540577540707271e-05,
      "loss": 1.7504,
      "step": 185114
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.3770488500595093,
      "learning_rate": 5.540340691678494e-05,
      "loss": 2.0481,
      "step": 185115
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.4589638710021973,
      "learning_rate": 5.540103847197224e-05,
      "loss": 1.9423,
      "step": 185116
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.355981469154358,
      "learning_rate": 5.5398670072635155e-05,
      "loss": 1.9898,
      "step": 185117
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.299633264541626,
      "learning_rate": 5.539630171877409e-05,
      "loss": 2.1657,
      "step": 185118
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.4990280866622925,
      "learning_rate": 5.539393341038946e-05,
      "loss": 2.1185,
      "step": 185119
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.268587589263916,
      "learning_rate": 5.539156514748164e-05,
      "loss": 1.9513,
      "step": 185120
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.341361165046692,
      "learning_rate": 5.538919693005125e-05,
      "loss": 1.8735,
      "step": 185121
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.373348593711853,
      "learning_rate": 5.538682875809851e-05,
      "loss": 2.0825,
      "step": 185122
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.2676280736923218,
      "learning_rate": 5.538446063162406e-05,
      "loss": 2.0931,
      "step": 185123
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.3438847064971924,
      "learning_rate": 5.5382092550628266e-05,
      "loss": 2.0957,
      "step": 185124
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.4673233032226562,
      "learning_rate": 5.5379724515111554e-05,
      "loss": 2.0332,
      "step": 185125
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.342231273651123,
      "learning_rate": 5.5377356525074303e-05,
      "loss": 1.9957,
      "step": 185126
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.5574421882629395,
      "learning_rate": 5.53749885805171e-05,
      "loss": 2.0428,
      "step": 185127
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.3853299617767334,
      "learning_rate": 5.5372620681440216e-05,
      "loss": 2.1741,
      "step": 185128
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.269993782043457,
      "learning_rate": 5.537025282784425e-05,
      "loss": 2.0855,
      "step": 185129
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.4097315073013306,
      "learning_rate": 5.5367885019729574e-05,
      "loss": 2.2892,
      "step": 185130
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.3406014442443848,
      "learning_rate": 5.5365517257096546e-05,
      "loss": 1.841,
      "step": 185131
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.6015571355819702,
      "learning_rate": 5.536314953994577e-05,
      "loss": 2.111,
      "step": 185132
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.8057904243469238,
      "learning_rate": 5.536078186827761e-05,
      "loss": 1.9291,
      "step": 185133
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.8418550491333008,
      "learning_rate": 5.535841424209241e-05,
      "loss": 2.033,
      "step": 185134
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.3167091608047485,
      "learning_rate": 5.5356046661390785e-05,
      "loss": 1.9298,
      "step": 185135
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.6026531457901,
      "learning_rate": 5.535367912617308e-05,
      "loss": 1.9592,
      "step": 185136
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.5455467700958252,
      "learning_rate": 5.535131163643969e-05,
      "loss": 2.0312,
      "step": 185137
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.4806361198425293,
      "learning_rate": 5.534894419219115e-05,
      "loss": 2.049,
      "step": 185138
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.5616744756698608,
      "learning_rate": 5.534657679342789e-05,
      "loss": 2.2109,
      "step": 185139
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.6068751811981201,
      "learning_rate": 5.534420944015021e-05,
      "loss": 1.9964,
      "step": 185140
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.37566077709198,
      "learning_rate": 5.534184213235875e-05,
      "loss": 2.0546,
      "step": 185141
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.516126275062561,
      "learning_rate": 5.533947487005387e-05,
      "loss": 1.9898,
      "step": 185142
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.463814377784729,
      "learning_rate": 5.533710765323594e-05,
      "loss": 2.007,
      "step": 185143
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.614783763885498,
      "learning_rate": 5.533474048190552e-05,
      "loss": 1.9679,
      "step": 185144
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.3841757774353027,
      "learning_rate": 5.5332373356062875e-05,
      "loss": 1.8681,
      "step": 185145
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.3533018827438354,
      "learning_rate": 5.5330006275708686e-05,
      "loss": 1.9798,
      "step": 185146
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.4965438842773438,
      "learning_rate": 5.5327639240843236e-05,
      "loss": 2.204,
      "step": 185147
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.300750970840454,
      "learning_rate": 5.532527225146703e-05,
      "loss": 2.1281,
      "step": 185148
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.4305790662765503,
      "learning_rate": 5.5322905307580376e-05,
      "loss": 2.0315,
      "step": 185149
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.6000474691390991,
      "learning_rate": 5.53205384091839e-05,
      "loss": 1.9078,
      "step": 185150
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.401119351387024,
      "learning_rate": 5.5318171556277866e-05,
      "loss": 2.0948,
      "step": 185151
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.3462917804718018,
      "learning_rate": 5.531580474886288e-05,
      "loss": 2.0942,
      "step": 185152
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.4275083541870117,
      "learning_rate": 5.5313437986939235e-05,
      "loss": 2.0438,
      "step": 185153
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.3235334157943726,
      "learning_rate": 5.53110712705076e-05,
      "loss": 2.2631,
      "step": 185154
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.500506043434143,
      "learning_rate": 5.5308704599568076e-05,
      "loss": 1.997,
      "step": 185155
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.3224059343338013,
      "learning_rate": 5.5306337974121395e-05,
      "loss": 2.1373,
      "step": 185156
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.4239025115966797,
      "learning_rate": 5.530397139416779e-05,
      "loss": 2.0309,
      "step": 185157
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.3356072902679443,
      "learning_rate": 5.530160485970786e-05,
      "loss": 2.0164,
      "step": 185158
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.3142845630645752,
      "learning_rate": 5.529923837074194e-05,
      "loss": 1.9383,
      "step": 185159
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.3385874032974243,
      "learning_rate": 5.5296871927270626e-05,
      "loss": 2.0033,
      "step": 185160
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.381767749786377,
      "learning_rate": 5.529450552929412e-05,
      "loss": 2.0988,
      "step": 185161
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.3777512311935425,
      "learning_rate": 5.5292139176813024e-05,
      "loss": 1.974,
      "step": 185162
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.4858123064041138,
      "learning_rate": 5.52897728698277e-05,
      "loss": 2.1961,
      "step": 185163
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.243804931640625,
      "learning_rate": 5.528740660833868e-05,
      "loss": 1.9454,
      "step": 185164
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.4566441774368286,
      "learning_rate": 5.528504039234627e-05,
      "loss": 1.8873,
      "step": 185165
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.4133244752883911,
      "learning_rate": 5.5282674221851096e-05,
      "loss": 2.0627,
      "step": 185166
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.424743890762329,
      "learning_rate": 5.528030809685346e-05,
      "loss": 1.8736,
      "step": 185167
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.4598026275634766,
      "learning_rate": 5.527794201735384e-05,
      "loss": 2.0327,
      "step": 185168
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.2167805433273315,
      "learning_rate": 5.5275575983352615e-05,
      "loss": 2.0029,
      "step": 185169
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.3096638917922974,
      "learning_rate": 5.527320999485033e-05,
      "loss": 2.0968,
      "step": 185170
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.4010040760040283,
      "learning_rate": 5.527084405184732e-05,
      "loss": 2.2652,
      "step": 185171
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.6399073600769043,
      "learning_rate": 5.526847815434414e-05,
      "loss": 2.0808,
      "step": 185172
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.4816467761993408,
      "learning_rate": 5.526611230234117e-05,
      "loss": 1.9559,
      "step": 185173
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.4894347190856934,
      "learning_rate": 5.526374649583887e-05,
      "loss": 2.0563,
      "step": 185174
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.3503257036209106,
      "learning_rate": 5.526138073483757e-05,
      "loss": 1.9819,
      "step": 185175
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.4731963872909546,
      "learning_rate": 5.525901501933787e-05,
      "loss": 2.0926,
      "step": 185176
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.5053092241287231,
      "learning_rate": 5.525664934934007e-05,
      "loss": 2.0984,
      "step": 185177
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.4075002670288086,
      "learning_rate": 5.525428372484475e-05,
      "loss": 2.1078,
      "step": 185178
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.2129995822906494,
      "learning_rate": 5.525191814585229e-05,
      "loss": 1.9833,
      "step": 185179
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.482573390007019,
      "learning_rate": 5.52495526123631e-05,
      "loss": 2.0314,
      "step": 185180
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.2780336141586304,
      "learning_rate": 5.524718712437758e-05,
      "loss": 1.9723,
      "step": 185181
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.3455573320388794,
      "learning_rate": 5.524482168189629e-05,
      "loss": 2.1241,
      "step": 185182
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.3711483478546143,
      "learning_rate": 5.524245628491953e-05,
      "loss": 1.8387,
      "step": 185183
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.3129675388336182,
      "learning_rate": 5.524009093344791e-05,
      "loss": 1.889,
      "step": 185184
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.4056147336959839,
      "learning_rate": 5.523772562748175e-05,
      "loss": 2.0568,
      "step": 185185
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.476191759109497,
      "learning_rate": 5.5235360367021565e-05,
      "loss": 2.0442,
      "step": 185186
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.2685866355895996,
      "learning_rate": 5.523299515206764e-05,
      "loss": 2.0742,
      "step": 185187
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.420343279838562,
      "learning_rate": 5.523062998262058e-05,
      "loss": 2.174,
      "step": 185188
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.3216956853866577,
      "learning_rate": 5.522826485868072e-05,
      "loss": 1.8995,
      "step": 185189
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.9282091856002808,
      "learning_rate": 5.5225899780248614e-05,
      "loss": 2.1015,
      "step": 185190
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.3677936792373657,
      "learning_rate": 5.522353474732462e-05,
      "loss": 1.9007,
      "step": 185191
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.337770700454712,
      "learning_rate": 5.522116975990921e-05,
      "loss": 2.0464,
      "step": 185192
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.4575225114822388,
      "learning_rate": 5.521880481800271e-05,
      "loss": 2.1667,
      "step": 185193
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.4081790447235107,
      "learning_rate": 5.521643992160576e-05,
      "loss": 1.8331,
      "step": 185194
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.35024094581604,
      "learning_rate": 5.521407507071858e-05,
      "loss": 2.1794,
      "step": 185195
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.6871143579483032,
      "learning_rate": 5.5211710265341825e-05,
      "loss": 2.041,
      "step": 185196
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.4470981359481812,
      "learning_rate": 5.520934550547583e-05,
      "loss": 2.1967,
      "step": 185197
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.3002159595489502,
      "learning_rate": 5.5206980791121035e-05,
      "loss": 2.0984,
      "step": 185198
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.5726888179779053,
      "learning_rate": 5.52046161222778e-05,
      "loss": 2.0359,
      "step": 185199
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.3753520250320435,
      "learning_rate": 5.520225149894676e-05,
      "loss": 1.8963,
      "step": 185200
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.5506327152252197,
      "learning_rate": 5.519988692112811e-05,
      "loss": 2.0239,
      "step": 185201
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.3289064168930054,
      "learning_rate": 5.5197522388822555e-05,
      "loss": 1.9688,
      "step": 185202
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.227722406387329,
      "learning_rate": 5.519515790203036e-05,
      "loss": 1.8018,
      "step": 185203
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.2883005142211914,
      "learning_rate": 5.519279346075203e-05,
      "loss": 1.7581,
      "step": 185204
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.2624214887619019,
      "learning_rate": 5.519042906498788e-05,
      "loss": 2.0518,
      "step": 185205
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.392585039138794,
      "learning_rate": 5.518806471473857e-05,
      "loss": 1.8327,
      "step": 185206
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.34414803981781,
      "learning_rate": 5.518570041000431e-05,
      "loss": 2.0281,
      "step": 185207
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.4622195959091187,
      "learning_rate": 5.5183336150785704e-05,
      "loss": 2.0364,
      "step": 185208
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.3498667478561401,
      "learning_rate": 5.518097193708316e-05,
      "loss": 2.0246,
      "step": 185209
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.2962638139724731,
      "learning_rate": 5.517860776889711e-05,
      "loss": 2.1553,
      "step": 185210
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.2603617906570435,
      "learning_rate": 5.517624364622792e-05,
      "loss": 1.999,
      "step": 185211
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.3196150064468384,
      "learning_rate": 5.517387956907611e-05,
      "loss": 1.8955,
      "step": 185212
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.543839693069458,
      "learning_rate": 5.517151553744206e-05,
      "loss": 1.8519,
      "step": 185213
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.2586764097213745,
      "learning_rate": 5.51691515513263e-05,
      "loss": 1.9791,
      "step": 185214
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.353188395500183,
      "learning_rate": 5.5166787610729227e-05,
      "loss": 2.0087,
      "step": 185215
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.3269654512405396,
      "learning_rate": 5.516442371565117e-05,
      "loss": 1.9643,
      "step": 185216
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.3547221422195435,
      "learning_rate": 5.516205986609278e-05,
      "loss": 2.1329,
      "step": 185217
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.3614275455474854,
      "learning_rate": 5.515969606205437e-05,
      "loss": 1.8977,
      "step": 185218
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.303421139717102,
      "learning_rate": 5.515733230353632e-05,
      "loss": 1.9201,
      "step": 185219
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.3193659782409668,
      "learning_rate": 5.5154968590539216e-05,
      "loss": 1.8923,
      "step": 185220
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.3037301301956177,
      "learning_rate": 5.515260492306344e-05,
      "loss": 1.939,
      "step": 185221
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.2478777170181274,
      "learning_rate": 5.515024130110931e-05,
      "loss": 1.9801,
      "step": 185222
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.3207356929779053,
      "learning_rate": 5.514787772467747e-05,
      "loss": 2.0275,
      "step": 185223
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.4076849222183228,
      "learning_rate": 5.514551419376828e-05,
      "loss": 2.0536,
      "step": 185224
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.3242104053497314,
      "learning_rate": 5.5143150708382044e-05,
      "loss": 2.0344,
      "step": 185225
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.4737820625305176,
      "learning_rate": 5.514078726851943e-05,
      "loss": 1.9608,
      "step": 185226
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.2214877605438232,
      "learning_rate": 5.513842387418073e-05,
      "loss": 1.9599,
      "step": 185227
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.303139328956604,
      "learning_rate": 5.513606052536638e-05,
      "loss": 1.9151,
      "step": 185228
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.3843361139297485,
      "learning_rate": 5.513369722207691e-05,
      "loss": 2.2429,
      "step": 185229
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.3327900171279907,
      "learning_rate": 5.513133396431263e-05,
      "loss": 2.1362,
      "step": 185230
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.3802924156188965,
      "learning_rate": 5.512897075207413e-05,
      "loss": 2.0362,
      "step": 185231
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.247417688369751,
      "learning_rate": 5.512660758536182e-05,
      "loss": 1.8539,
      "step": 185232
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.4173569679260254,
      "learning_rate": 5.512424446417598e-05,
      "loss": 2.031,
      "step": 185233
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.3556255102157593,
      "learning_rate": 5.512188138851723e-05,
      "loss": 2.1676,
      "step": 185234
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.3475781679153442,
      "learning_rate": 5.5119518358385965e-05,
      "loss": 1.9931,
      "step": 185235
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.3823118209838867,
      "learning_rate": 5.511715537378254e-05,
      "loss": 2.0932,
      "step": 185236
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.4425063133239746,
      "learning_rate": 5.5114792434707535e-05,
      "loss": 1.997,
      "step": 185237
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.4169915914535522,
      "learning_rate": 5.511242954116121e-05,
      "loss": 1.9948,
      "step": 185238
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.4914172887802124,
      "learning_rate": 5.51100666931442e-05,
      "loss": 2.002,
      "step": 185239
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.3569444417953491,
      "learning_rate": 5.510770389065683e-05,
      "loss": 2.3185,
      "step": 185240
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.5673152208328247,
      "learning_rate": 5.510534113369958e-05,
      "loss": 1.9417,
      "step": 185241
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.4716825485229492,
      "learning_rate": 5.51029784222728e-05,
      "loss": 1.9676,
      "step": 185242
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.2688665390014648,
      "learning_rate": 5.510061575637708e-05,
      "loss": 2.0618,
      "step": 185243
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.3035897016525269,
      "learning_rate": 5.509825313601266e-05,
      "loss": 2.0389,
      "step": 185244
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.3011064529418945,
      "learning_rate": 5.509589056118022e-05,
      "loss": 2.0963,
      "step": 185245
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.3478055000305176,
      "learning_rate": 5.5093528031880064e-05,
      "loss": 2.0175,
      "step": 185246
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.2889723777770996,
      "learning_rate": 5.5091165548112645e-05,
      "loss": 1.9077,
      "step": 185247
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.4000765085220337,
      "learning_rate": 5.508880310987831e-05,
      "loss": 1.9852,
      "step": 185248
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.3415580987930298,
      "learning_rate": 5.5086440717177685e-05,
      "loss": 2.0656,
      "step": 185249
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.3540886640548706,
      "learning_rate": 5.508407837001103e-05,
      "loss": 2.0862,
      "step": 185250
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.4834787845611572,
      "learning_rate": 5.5081716068378925e-05,
      "loss": 2.087,
      "step": 185251
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.374664306640625,
      "learning_rate": 5.507935381228179e-05,
      "loss": 1.9732,
      "step": 185252
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.3456851243972778,
      "learning_rate": 5.507699160172e-05,
      "loss": 1.8207,
      "step": 185253
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.4791100025177002,
      "learning_rate": 5.507462943669395e-05,
      "loss": 1.8716,
      "step": 185254
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.3588488101959229,
      "learning_rate": 5.507226731720421e-05,
      "loss": 2.0754,
      "step": 185255
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.3737608194351196,
      "learning_rate": 5.506990524325111e-05,
      "loss": 2.0269,
      "step": 185256
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.4204148054122925,
      "learning_rate": 5.506754321483522e-05,
      "loss": 1.8136,
      "step": 185257
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.328587532043457,
      "learning_rate": 5.5065181231956866e-05,
      "loss": 1.9713,
      "step": 185258
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.476319670677185,
      "learning_rate": 5.506281929461652e-05,
      "loss": 2.0541,
      "step": 185259
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.4528579711914062,
      "learning_rate": 5.506045740281455e-05,
      "loss": 1.9949,
      "step": 185260
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.3895655870437622,
      "learning_rate": 5.505809555655155e-05,
      "loss": 2.0303,
      "step": 185261
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.3326694965362549,
      "learning_rate": 5.505573375582779e-05,
      "loss": 1.9827,
      "step": 185262
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.3355380296707153,
      "learning_rate": 5.50533720006439e-05,
      "loss": 2.1187,
      "step": 185263
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.3359196186065674,
      "learning_rate": 5.505101029100019e-05,
      "loss": 1.9162,
      "step": 185264
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.2767143249511719,
      "learning_rate": 5.5048648626897105e-05,
      "loss": 2.0784,
      "step": 185265
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.3339149951934814,
      "learning_rate": 5.5046287008335034e-05,
      "loss": 2.0086,
      "step": 185266
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.4467487335205078,
      "learning_rate": 5.5043925435314563e-05,
      "loss": 2.032,
      "step": 185267
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.3213127851486206,
      "learning_rate": 5.5041563907835964e-05,
      "loss": 1.8782,
      "step": 185268
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.4131463766098022,
      "learning_rate": 5.503920242589983e-05,
      "loss": 1.8949,
      "step": 185269
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.2648690938949585,
      "learning_rate": 5.503684098950657e-05,
      "loss": 1.8832,
      "step": 185270
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.5111658573150635,
      "learning_rate": 5.503447959865655e-05,
      "loss": 1.8399,
      "step": 185271
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.3642730712890625,
      "learning_rate": 5.503211825335019e-05,
      "loss": 2.0506,
      "step": 185272
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.2550278902053833,
      "learning_rate": 5.502975695358803e-05,
      "loss": 1.9652,
      "step": 185273
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.570178508758545,
      "learning_rate": 5.502739569937041e-05,
      "loss": 1.8855,
      "step": 185274
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.504445195198059,
      "learning_rate": 5.50250344906979e-05,
      "loss": 1.8922,
      "step": 185275
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.46689772605896,
      "learning_rate": 5.5022673327570846e-05,
      "loss": 2.0825,
      "step": 185276
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.3745425939559937,
      "learning_rate": 5.5020312209989725e-05,
      "loss": 2.0209,
      "step": 185277
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.4197317361831665,
      "learning_rate": 5.501795113795484e-05,
      "loss": 2.0384,
      "step": 185278
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.5898127555847168,
      "learning_rate": 5.5015590111466854e-05,
      "loss": 2.0815,
      "step": 185279
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.363914132118225,
      "learning_rate": 5.5013229130526005e-05,
      "loss": 2.0105,
      "step": 185280
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.4946333169937134,
      "learning_rate": 5.501086819513288e-05,
      "loss": 1.788,
      "step": 185281
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.2984182834625244,
      "learning_rate": 5.50085073052879e-05,
      "loss": 1.7859,
      "step": 185282
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.2795644998550415,
      "learning_rate": 5.500614646099144e-05,
      "loss": 1.8531,
      "step": 185283
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.2867323160171509,
      "learning_rate": 5.500378566224388e-05,
      "loss": 2.1454,
      "step": 185284
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.5468007326126099,
      "learning_rate": 5.500142490904582e-05,
      "loss": 1.9588,
      "step": 185285
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.4804731607437134,
      "learning_rate": 5.499906420139753e-05,
      "loss": 1.9133,
      "step": 185286
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.5184513330459595,
      "learning_rate": 5.4996703539299626e-05,
      "loss": 2.0082,
      "step": 185287
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.2675840854644775,
      "learning_rate": 5.499434292275245e-05,
      "loss": 1.8849,
      "step": 185288
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.3131103515625,
      "learning_rate": 5.499198235175648e-05,
      "loss": 1.9472,
      "step": 185289
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.4326179027557373,
      "learning_rate": 5.498962182631204e-05,
      "loss": 1.9003,
      "step": 185290
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.2906737327575684,
      "learning_rate": 5.498726134641972e-05,
      "loss": 1.6919,
      "step": 185291
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.335286259651184,
      "learning_rate": 5.49849009120798e-05,
      "loss": 1.8975,
      "step": 185292
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.4260830879211426,
      "learning_rate": 5.498254052329294e-05,
      "loss": 2.1783,
      "step": 185293
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.5857603549957275,
      "learning_rate": 5.498018018005941e-05,
      "loss": 1.915,
      "step": 185294
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.2798289060592651,
      "learning_rate": 5.4977819882379706e-05,
      "loss": 1.9295,
      "step": 185295
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.4283778667449951,
      "learning_rate": 5.497545963025416e-05,
      "loss": 1.9063,
      "step": 185296
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.357537865638733,
      "learning_rate": 5.4973099423683374e-05,
      "loss": 1.886,
      "step": 185297
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.3690029382705688,
      "learning_rate": 5.497073926266765e-05,
      "loss": 1.9475,
      "step": 185298
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.4070063829421997,
      "learning_rate": 5.496837914720755e-05,
      "loss": 1.9252,
      "step": 185299
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.5418556928634644,
      "learning_rate": 5.496601907730341e-05,
      "loss": 2.088,
      "step": 185300
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.4704031944274902,
      "learning_rate": 5.4963659052955834e-05,
      "loss": 1.8305,
      "step": 185301
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.4985520839691162,
      "learning_rate": 5.496129907416501e-05,
      "loss": 2.0986,
      "step": 185302
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.3476790189743042,
      "learning_rate": 5.495893914093158e-05,
      "loss": 1.8119,
      "step": 185303
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.2816100120544434,
      "learning_rate": 5.495657925325581e-05,
      "loss": 1.8167,
      "step": 185304
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.3639405965805054,
      "learning_rate": 5.495421941113832e-05,
      "loss": 2.0743,
      "step": 185305
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.4342716932296753,
      "learning_rate": 5.49518596145794e-05,
      "loss": 1.9705,
      "step": 185306
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.5401694774627686,
      "learning_rate": 5.4949499863579624e-05,
      "loss": 2.0766,
      "step": 185307
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.395936131477356,
      "learning_rate": 5.494714015813938e-05,
      "loss": 2.0021,
      "step": 185308
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.4690486192703247,
      "learning_rate": 5.494478049825909e-05,
      "loss": 2.0001,
      "step": 185309
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.2941550016403198,
      "learning_rate": 5.4942420883939087e-05,
      "loss": 1.8803,
      "step": 185310
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.375336766242981,
      "learning_rate": 5.4940061315180004e-05,
      "loss": 2.0145,
      "step": 185311
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.3099530935287476,
      "learning_rate": 5.493770179198211e-05,
      "loss": 1.7982,
      "step": 185312
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.5198477506637573,
      "learning_rate": 5.4935342314346006e-05,
      "loss": 1.919,
      "step": 185313
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.4954562187194824,
      "learning_rate": 5.4932982882272027e-05,
      "loss": 2.1854,
      "step": 185314
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.3799811601638794,
      "learning_rate": 5.493062349576059e-05,
      "loss": 1.8047,
      "step": 185315
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.3707607984542847,
      "learning_rate": 5.492826415481222e-05,
      "loss": 1.8975,
      "step": 185316
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.4345964193344116,
      "learning_rate": 5.492590485942733e-05,
      "loss": 1.9148,
      "step": 185317
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.3282667398452759,
      "learning_rate": 5.492354560960626e-05,
      "loss": 2.0176,
      "step": 185318
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.377274513244629,
      "learning_rate": 5.49211864053496e-05,
      "loss": 2.1419,
      "step": 185319
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.2884477376937866,
      "learning_rate": 5.4918827246657736e-05,
      "loss": 1.8188,
      "step": 185320
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.6198936700820923,
      "learning_rate": 5.491646813353099e-05,
      "loss": 1.9966,
      "step": 185321
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.3466423749923706,
      "learning_rate": 5.491410906596999e-05,
      "loss": 2.1726,
      "step": 185322
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.4365416765213013,
      "learning_rate": 5.491175004397508e-05,
      "loss": 2.3247,
      "step": 185323
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.5257922410964966,
      "learning_rate": 5.490939106754662e-05,
      "loss": 2.2801,
      "step": 185324
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.5172479152679443,
      "learning_rate": 5.4907032136685235e-05,
      "loss": 1.8958,
      "step": 185325
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.38948392868042,
      "learning_rate": 5.490467325139121e-05,
      "loss": 2.1584,
      "step": 185326
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.4298760890960693,
      "learning_rate": 5.4902314411664995e-05,
      "loss": 2.0437,
      "step": 185327
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.3053560256958008,
      "learning_rate": 5.489995561750713e-05,
      "loss": 1.8828,
      "step": 185328
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.4603867530822754,
      "learning_rate": 5.489759686891792e-05,
      "loss": 1.9553,
      "step": 185329
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.7270383834838867,
      "learning_rate": 5.489523816589795e-05,
      "loss": 1.9719,
      "step": 185330
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.383012056350708,
      "learning_rate": 5.489287950844756e-05,
      "loss": 1.8914,
      "step": 185331
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.3787373304367065,
      "learning_rate": 5.4890520896567224e-05,
      "loss": 2.0022,
      "step": 185332
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.3929178714752197,
      "learning_rate": 5.48881623302573e-05,
      "loss": 1.9069,
      "step": 185333
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.3634414672851562,
      "learning_rate": 5.4885803809518357e-05,
      "loss": 2.183,
      "step": 185334
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.4221104383468628,
      "learning_rate": 5.48834453343507e-05,
      "loss": 1.9477,
      "step": 185335
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.5893564224243164,
      "learning_rate": 5.488108690475491e-05,
      "loss": 1.7375,
      "step": 185336
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.4116371870040894,
      "learning_rate": 5.487872852073134e-05,
      "loss": 2.101,
      "step": 185337
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.432039737701416,
      "learning_rate": 5.487637018228048e-05,
      "loss": 1.9598,
      "step": 185338
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.4501458406448364,
      "learning_rate": 5.487401188940263e-05,
      "loss": 2.0905,
      "step": 185339
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.349108338356018,
      "learning_rate": 5.48716536420984e-05,
      "loss": 1.821,
      "step": 185340
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.3193529844284058,
      "learning_rate": 5.486929544036808e-05,
      "loss": 1.9302,
      "step": 185341
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.4084590673446655,
      "learning_rate": 5.4866937284212266e-05,
      "loss": 2.1618,
      "step": 185342
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.3466814756393433,
      "learning_rate": 5.486457917363133e-05,
      "loss": 1.8973,
      "step": 185343
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.330370545387268,
      "learning_rate": 5.486222110862567e-05,
      "loss": 1.8268,
      "step": 185344
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.3323185443878174,
      "learning_rate": 5.4859863089195685e-05,
      "loss": 2.1289,
      "step": 185345
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.3414710760116577,
      "learning_rate": 5.485750511534195e-05,
      "loss": 2.0813,
      "step": 185346
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.4378958940505981,
      "learning_rate": 5.4855147187064785e-05,
      "loss": 1.9415,
      "step": 185347
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.683832049369812,
      "learning_rate": 5.4852789304364695e-05,
      "loss": 2.0713,
      "step": 185348
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.3890156745910645,
      "learning_rate": 5.485043146724215e-05,
      "loss": 2.0855,
      "step": 185349
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.6152629852294922,
      "learning_rate": 5.4848073675697515e-05,
      "loss": 2.1892,
      "step": 185350
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.8735198974609375,
      "learning_rate": 5.484571592973118e-05,
      "loss": 2.0234,
      "step": 185351
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.3171379566192627,
      "learning_rate": 5.484335822934373e-05,
      "loss": 2.1037,
      "step": 185352
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.3927274942398071,
      "learning_rate": 5.484100057453544e-05,
      "loss": 2.083,
      "step": 185353
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.3506258726119995,
      "learning_rate": 5.483864296530693e-05,
      "loss": 2.0291,
      "step": 185354
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.373414397239685,
      "learning_rate": 5.4836285401658566e-05,
      "loss": 2.0231,
      "step": 185355
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.2772377729415894,
      "learning_rate": 5.483392788359073e-05,
      "loss": 1.9447,
      "step": 185356
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.4401507377624512,
      "learning_rate": 5.483157041110381e-05,
      "loss": 2.0025,
      "step": 185357
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.3603577613830566,
      "learning_rate": 5.482921298419842e-05,
      "loss": 2.1638,
      "step": 185358
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.3922154903411865,
      "learning_rate": 5.4826855602874845e-05,
      "loss": 1.9459,
      "step": 185359
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.2699183225631714,
      "learning_rate": 5.482449826713363e-05,
      "loss": 1.7713,
      "step": 185360
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.5508332252502441,
      "learning_rate": 5.4822140976975195e-05,
      "loss": 1.857,
      "step": 185361
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.2686692476272583,
      "learning_rate": 5.481978373239996e-05,
      "loss": 1.987,
      "step": 185362
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.416472315788269,
      "learning_rate": 5.481742653340826e-05,
      "loss": 1.9781,
      "step": 185363
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.3288612365722656,
      "learning_rate": 5.4815069380000696e-05,
      "loss": 2.0061,
      "step": 185364
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.278928279876709,
      "learning_rate": 5.4812712272177575e-05,
      "loss": 1.9968,
      "step": 185365
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.4194945096969604,
      "learning_rate": 5.4810355209939484e-05,
      "loss": 2.1331,
      "step": 185366
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.3795456886291504,
      "learning_rate": 5.480799819328666e-05,
      "loss": 2.1436,
      "step": 185367
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.5006734132766724,
      "learning_rate": 5.480564122221984e-05,
      "loss": 1.8549,
      "step": 185368
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.3252583742141724,
      "learning_rate": 5.480328429673912e-05,
      "loss": 1.8219,
      "step": 185369
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.3737088441848755,
      "learning_rate": 5.480092741684521e-05,
      "loss": 1.8868,
      "step": 185370
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.4192205667495728,
      "learning_rate": 5.479857058253832e-05,
      "loss": 2.0439,
      "step": 185371
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.4321911334991455,
      "learning_rate": 5.479621379381911e-05,
      "loss": 1.9848,
      "step": 185372
    },
    {
      "epoch": 2.41,
      "grad_norm": 2.0222816467285156,
      "learning_rate": 5.479385705068779e-05,
      "loss": 1.7718,
      "step": 185373
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.3110918998718262,
      "learning_rate": 5.479150035314511e-05,
      "loss": 1.815,
      "step": 185374
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.404829502105713,
      "learning_rate": 5.4789143701191165e-05,
      "loss": 2.0701,
      "step": 185375
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.3309327363967896,
      "learning_rate": 5.478678709482661e-05,
      "loss": 2.1231,
      "step": 185376
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.5520319938659668,
      "learning_rate": 5.478443053405176e-05,
      "loss": 1.8846,
      "step": 185377
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.3326705694198608,
      "learning_rate": 5.478207401886717e-05,
      "loss": 1.9857,
      "step": 185378
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.328482985496521,
      "learning_rate": 5.477971754927315e-05,
      "loss": 1.9375,
      "step": 185379
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.4042000770568848,
      "learning_rate": 5.477736112527036e-05,
      "loss": 2.0809,
      "step": 185380
    },
    {
      "epoch": 2.41,
      "grad_norm": 2.071610689163208,
      "learning_rate": 5.477500474685896e-05,
      "loss": 1.757,
      "step": 185381
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.2481377124786377,
      "learning_rate": 5.477264841403957e-05,
      "loss": 2.1054,
      "step": 185382
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.2867662906646729,
      "learning_rate": 5.47702921268125e-05,
      "loss": 2.0301,
      "step": 185383
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.3536866903305054,
      "learning_rate": 5.476793588517833e-05,
      "loss": 1.8752,
      "step": 185384
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.4273382425308228,
      "learning_rate": 5.476557968913735e-05,
      "loss": 1.9132,
      "step": 185385
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.2980945110321045,
      "learning_rate": 5.4763223538690234e-05,
      "loss": 1.8339,
      "step": 185386
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.2962943315505981,
      "learning_rate": 5.476086743383715e-05,
      "loss": 1.9332,
      "step": 185387
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.3147356510162354,
      "learning_rate": 5.4758511374578696e-05,
      "loss": 2.1005,
      "step": 185388
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.2862379550933838,
      "learning_rate": 5.475615536091517e-05,
      "loss": 1.7837,
      "step": 185389
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.4806993007659912,
      "learning_rate": 5.4753799392847194e-05,
      "loss": 1.9879,
      "step": 185390
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.3513778448104858,
      "learning_rate": 5.475144347037506e-05,
      "loss": 2.0602,
      "step": 185391
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.395204782485962,
      "learning_rate": 5.474908759349931e-05,
      "loss": 2.0345,
      "step": 185392
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.5466676950454712,
      "learning_rate": 5.474673176222035e-05,
      "loss": 1.8653,
      "step": 185393
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.3716356754302979,
      "learning_rate": 5.474437597653859e-05,
      "loss": 2.0425,
      "step": 185394
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.4061710834503174,
      "learning_rate": 5.474202023645442e-05,
      "loss": 1.9362,
      "step": 185395
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.443650245666504,
      "learning_rate": 5.473966454196841e-05,
      "loss": 1.9545,
      "step": 185396
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.365712285041809,
      "learning_rate": 5.4737308893080854e-05,
      "loss": 1.9213,
      "step": 185397
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.4595699310302734,
      "learning_rate": 5.473495328979233e-05,
      "loss": 2.0434,
      "step": 185398
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.483324646949768,
      "learning_rate": 5.473259773210319e-05,
      "loss": 2.0677,
      "step": 185399
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.3378132581710815,
      "learning_rate": 5.473024222001384e-05,
      "loss": 1.8735,
      "step": 185400
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.2789257764816284,
      "learning_rate": 5.4727886753524816e-05,
      "loss": 1.9971,
      "step": 185401
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.2942562103271484,
      "learning_rate": 5.4725531332636555e-05,
      "loss": 1.8872,
      "step": 185402
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.5051798820495605,
      "learning_rate": 5.472317595734934e-05,
      "loss": 1.871,
      "step": 185403
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.505197286605835,
      "learning_rate": 5.472082062766379e-05,
      "loss": 1.8005,
      "step": 185404
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.3124456405639648,
      "learning_rate": 5.4718465343580286e-05,
      "loss": 2.0513,
      "step": 185405
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.3336870670318604,
      "learning_rate": 5.471611010509914e-05,
      "loss": 2.1123,
      "step": 185406
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.377482295036316,
      "learning_rate": 5.471375491222102e-05,
      "loss": 2.0321,
      "step": 185407
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.2871575355529785,
      "learning_rate": 5.4711399764946215e-05,
      "loss": 1.9544,
      "step": 185408
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.2567137479782104,
      "learning_rate": 5.47090446632751e-05,
      "loss": 2.0523,
      "step": 185409
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.272900104522705,
      "learning_rate": 5.470668960720831e-05,
      "loss": 1.9679,
      "step": 185410
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.4301952123641968,
      "learning_rate": 5.470433459674617e-05,
      "loss": 2.2397,
      "step": 185411
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.355910062789917,
      "learning_rate": 5.470197963188905e-05,
      "loss": 1.9524,
      "step": 185412
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.2972079515457153,
      "learning_rate": 5.469962471263755e-05,
      "loss": 1.8189,
      "step": 185413
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.4389692544937134,
      "learning_rate": 5.469726983899193e-05,
      "loss": 2.1882,
      "step": 185414
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.481044054031372,
      "learning_rate": 5.4694915010952764e-05,
      "loss": 1.8867,
      "step": 185415
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.3018338680267334,
      "learning_rate": 5.4692560228520486e-05,
      "loss": 1.8964,
      "step": 185416
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.3929898738861084,
      "learning_rate": 5.469020549169549e-05,
      "loss": 2.0489,
      "step": 185417
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.2892451286315918,
      "learning_rate": 5.468785080047812e-05,
      "loss": 1.9025,
      "step": 185418
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.4536489248275757,
      "learning_rate": 5.468549615486899e-05,
      "loss": 1.8452,
      "step": 185419
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.3077319860458374,
      "learning_rate": 5.4683141554868384e-05,
      "loss": 1.8554,
      "step": 185420
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.3090161085128784,
      "learning_rate": 5.468078700047689e-05,
      "loss": 1.9661,
      "step": 185421
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.5258506536483765,
      "learning_rate": 5.4678432491694876e-05,
      "loss": 2.2578,
      "step": 185422
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.311540961265564,
      "learning_rate": 5.467607802852275e-05,
      "loss": 2.1181,
      "step": 185423
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.427904486656189,
      "learning_rate": 5.467372361096091e-05,
      "loss": 2.0131,
      "step": 185424
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.391387701034546,
      "learning_rate": 5.4671369239009945e-05,
      "loss": 2.065,
      "step": 185425
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.260677695274353,
      "learning_rate": 5.466901491267009e-05,
      "loss": 1.7214,
      "step": 185426
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.3157049417495728,
      "learning_rate": 5.4666660631941995e-05,
      "loss": 1.7306,
      "step": 185427
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.329060435295105,
      "learning_rate": 5.4664306396826006e-05,
      "loss": 2.1999,
      "step": 185428
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.3440182209014893,
      "learning_rate": 5.466195220732257e-05,
      "loss": 2.2428,
      "step": 185429
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.437720537185669,
      "learning_rate": 5.465959806343201e-05,
      "loss": 2.1886,
      "step": 185430
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.5509364604949951,
      "learning_rate": 5.465724396515493e-05,
      "loss": 2.1319,
      "step": 185431
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.398635983467102,
      "learning_rate": 5.465488991249163e-05,
      "loss": 1.9134,
      "step": 185432
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.3380290269851685,
      "learning_rate": 5.4652535905442674e-05,
      "loss": 2.0489,
      "step": 185433
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.321236252784729,
      "learning_rate": 5.4650181944008396e-05,
      "loss": 1.9158,
      "step": 185434
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.3988384008407593,
      "learning_rate": 5.464782802818943e-05,
      "loss": 2.1712,
      "step": 185435
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.3730504512786865,
      "learning_rate": 5.4645474157985915e-05,
      "loss": 1.9882,
      "step": 185436
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.3126347064971924,
      "learning_rate": 5.464312033339847e-05,
      "loss": 2.0562,
      "step": 185437
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.5860819816589355,
      "learning_rate": 5.4640766554427476e-05,
      "loss": 1.8041,
      "step": 185438
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.4387894868850708,
      "learning_rate": 5.463841282107345e-05,
      "loss": 2.1451,
      "step": 185439
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.4016295671463013,
      "learning_rate": 5.4636059133336706e-05,
      "loss": 2.3187,
      "step": 185440
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.5098429918289185,
      "learning_rate": 5.4633705491217905e-05,
      "loss": 1.9727,
      "step": 185441
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.4394346475601196,
      "learning_rate": 5.463135189471715e-05,
      "loss": 2.0562,
      "step": 185442
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.5281206369400024,
      "learning_rate": 5.462899834383516e-05,
      "loss": 1.9718,
      "step": 185443
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.5077077150344849,
      "learning_rate": 5.462664483857219e-05,
      "loss": 2.0968,
      "step": 185444
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.3986033201217651,
      "learning_rate": 5.4624291378928856e-05,
      "loss": 2.0217,
      "step": 185445
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.3221937417984009,
      "learning_rate": 5.4621937964905396e-05,
      "loss": 2.1876,
      "step": 185446
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.3494614362716675,
      "learning_rate": 5.461958459650251e-05,
      "loss": 2.1628,
      "step": 185447
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.4110546112060547,
      "learning_rate": 5.46172312737203e-05,
      "loss": 2.1156,
      "step": 185448
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.3203281164169312,
      "learning_rate": 5.461487799655949e-05,
      "loss": 1.9836,
      "step": 185449
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.352724313735962,
      "learning_rate": 5.4612524765020295e-05,
      "loss": 1.8322,
      "step": 185450
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.3745756149291992,
      "learning_rate": 5.461017157910337e-05,
      "loss": 2.1193,
      "step": 185451
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.3881365060806274,
      "learning_rate": 5.460781843880895e-05,
      "loss": 2.1309,
      "step": 185452
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.3403383493423462,
      "learning_rate": 5.460546534413773e-05,
      "loss": 1.9774,
      "step": 185453
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.3312678337097168,
      "learning_rate": 5.460311229508986e-05,
      "loss": 2.0993,
      "step": 185454
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.3991299867630005,
      "learning_rate": 5.460075929166595e-05,
      "loss": 2.0278,
      "step": 185455
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.333958625793457,
      "learning_rate": 5.4598406333866315e-05,
      "loss": 2.057,
      "step": 185456
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.4530155658721924,
      "learning_rate": 5.459605342169152e-05,
      "loss": 2.1376,
      "step": 185457
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.3138772249221802,
      "learning_rate": 5.459370055514193e-05,
      "loss": 1.998,
      "step": 185458
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.3460005521774292,
      "learning_rate": 5.45913477342181e-05,
      "loss": 2.078,
      "step": 185459
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.2353794574737549,
      "learning_rate": 5.458899495892025e-05,
      "loss": 2.0884,
      "step": 185460
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.324404001235962,
      "learning_rate": 5.458664222924903e-05,
      "loss": 1.9648,
      "step": 185461
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.5184776782989502,
      "learning_rate": 5.4584289545204684e-05,
      "loss": 2.1141,
      "step": 185462
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.3264840841293335,
      "learning_rate": 5.458193690678784e-05,
      "loss": 1.985,
      "step": 185463
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.408506989479065,
      "learning_rate": 5.457958431399876e-05,
      "loss": 2.1137,
      "step": 185464
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.349415898323059,
      "learning_rate": 5.4577231766838146e-05,
      "loss": 1.8742,
      "step": 185465
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.4092849493026733,
      "learning_rate": 5.457487926530607e-05,
      "loss": 1.9882,
      "step": 185466
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.2589397430419922,
      "learning_rate": 5.457252680940326e-05,
      "loss": 2.0005,
      "step": 185467
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.3957550525665283,
      "learning_rate": 5.457017439912995e-05,
      "loss": 2.0553,
      "step": 185468
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.3308194875717163,
      "learning_rate": 5.4567822034486773e-05,
      "loss": 2.1964,
      "step": 185469
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.2966878414154053,
      "learning_rate": 5.4565469715473965e-05,
      "loss": 1.973,
      "step": 185470
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.309446096420288,
      "learning_rate": 5.456311744209225e-05,
      "loss": 2.163,
      "step": 185471
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.454697847366333,
      "learning_rate": 5.456076521434174e-05,
      "loss": 2.0,
      "step": 185472
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.3383815288543701,
      "learning_rate": 5.4558413032223056e-05,
      "loss": 1.8739,
      "step": 185473
    },
    {
      "epoch": 2.41,
      "grad_norm": 1.5593981742858887,
      "learning_rate": 5.4556060895736544e-05,
      "loss": 2.0125,
      "step": 185474
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.2924716472625732,
      "learning_rate": 5.4553708804882786e-05,
      "loss": 2.1456,
      "step": 185475
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.485028624534607,
      "learning_rate": 5.4551356759662035e-05,
      "loss": 2.1798,
      "step": 185476
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.4122283458709717,
      "learning_rate": 5.454900476007488e-05,
      "loss": 2.0298,
      "step": 185477
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.3781890869140625,
      "learning_rate": 5.4546652806121716e-05,
      "loss": 2.1202,
      "step": 185478
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.3977850675582886,
      "learning_rate": 5.454430089780295e-05,
      "loss": 2.1055,
      "step": 185479
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.37112295627594,
      "learning_rate": 5.454194903511898e-05,
      "loss": 2.0896,
      "step": 185480
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.40058171749115,
      "learning_rate": 5.453959721807034e-05,
      "loss": 2.1097,
      "step": 185481
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.3176960945129395,
      "learning_rate": 5.453724544665736e-05,
      "loss": 2.1797,
      "step": 185482
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.384885311126709,
      "learning_rate": 5.453489372088061e-05,
      "loss": 1.894,
      "step": 185483
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.312171459197998,
      "learning_rate": 5.453254204074049e-05,
      "loss": 2.0327,
      "step": 185484
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.3116148710250854,
      "learning_rate": 5.453019040623736e-05,
      "loss": 1.8879,
      "step": 185485
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.371792197227478,
      "learning_rate": 5.452783881737166e-05,
      "loss": 1.8531,
      "step": 185486
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.3444972038269043,
      "learning_rate": 5.452548727414392e-05,
      "loss": 2.0553,
      "step": 185487
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.4214680194854736,
      "learning_rate": 5.452313577655444e-05,
      "loss": 2.0896,
      "step": 185488
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.6693038940429688,
      "learning_rate": 5.452078432460385e-05,
      "loss": 1.9358,
      "step": 185489
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.6190375089645386,
      "learning_rate": 5.451843291829245e-05,
      "loss": 2.1451,
      "step": 185490
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.4747519493103027,
      "learning_rate": 5.451608155762065e-05,
      "loss": 1.9828,
      "step": 185491
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.3629363775253296,
      "learning_rate": 5.451373024258904e-05,
      "loss": 1.8986,
      "step": 185492
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.4695677757263184,
      "learning_rate": 5.4511378973197916e-05,
      "loss": 2.1287,
      "step": 185493
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.3614614009857178,
      "learning_rate": 5.450902774944772e-05,
      "loss": 1.9452,
      "step": 185494
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.3643935918807983,
      "learning_rate": 5.450667657133898e-05,
      "loss": 1.9506,
      "step": 185495
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.4562733173370361,
      "learning_rate": 5.45043254388721e-05,
      "loss": 2.0049,
      "step": 185496
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.3025884628295898,
      "learning_rate": 5.450197435204744e-05,
      "loss": 2.0279,
      "step": 185497
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.3095831871032715,
      "learning_rate": 5.4499623310865536e-05,
      "loss": 1.9508,
      "step": 185498
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.3252767324447632,
      "learning_rate": 5.4497272315326725e-05,
      "loss": 2.0762,
      "step": 185499
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.4115384817123413,
      "learning_rate": 5.44949213654316e-05,
      "loss": 2.2391,
      "step": 185500
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.3052669763565063,
      "learning_rate": 5.449257046118046e-05,
      "loss": 2.2287,
      "step": 185501
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.221346139907837,
      "learning_rate": 5.449021960257381e-05,
      "loss": 2.0728,
      "step": 185502
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.3234058618545532,
      "learning_rate": 5.4487868789612024e-05,
      "loss": 1.7719,
      "step": 185503
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.3023267984390259,
      "learning_rate": 5.4485518022295616e-05,
      "loss": 1.9257,
      "step": 185504
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.266826868057251,
      "learning_rate": 5.44831673006249e-05,
      "loss": 2.0938,
      "step": 185505
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.4023948907852173,
      "learning_rate": 5.448081662460047e-05,
      "loss": 1.9543,
      "step": 185506
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.3624545335769653,
      "learning_rate": 5.447846599422272e-05,
      "loss": 1.8937,
      "step": 185507
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.5852303504943848,
      "learning_rate": 5.4476115409492027e-05,
      "loss": 1.8205,
      "step": 185508
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.306687355041504,
      "learning_rate": 5.447376487040882e-05,
      "loss": 1.8992,
      "step": 185509
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.2775983810424805,
      "learning_rate": 5.447141437697363e-05,
      "loss": 1.9565,
      "step": 185510
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.2803577184677124,
      "learning_rate": 5.446906392918676e-05,
      "loss": 1.7746,
      "step": 185511
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.3201205730438232,
      "learning_rate": 5.446671352704881e-05,
      "loss": 2.0399,
      "step": 185512
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.323385238647461,
      "learning_rate": 5.4464363170560044e-05,
      "loss": 1.9809,
      "step": 185513
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.3225171566009521,
      "learning_rate": 5.446201285972113e-05,
      "loss": 2.0682,
      "step": 185514
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.454661250114441,
      "learning_rate": 5.445966259453223e-05,
      "loss": 1.9536,
      "step": 185515
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.47539222240448,
      "learning_rate": 5.445731237499398e-05,
      "loss": 1.9932,
      "step": 185516
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.3829923868179321,
      "learning_rate": 5.445496220110671e-05,
      "loss": 1.9216,
      "step": 185517
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.4378057718276978,
      "learning_rate": 5.445261207287093e-05,
      "loss": 2.0882,
      "step": 185518
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.456688404083252,
      "learning_rate": 5.445026199028699e-05,
      "loss": 1.7545,
      "step": 185519
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.3166478872299194,
      "learning_rate": 5.4447911953355505e-05,
      "loss": 1.961,
      "step": 185520
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.30916166305542,
      "learning_rate": 5.444556196207667e-05,
      "loss": 1.9893,
      "step": 185521
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.5790791511535645,
      "learning_rate": 5.444321201645111e-05,
      "loss": 1.997,
      "step": 185522
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.4007174968719482,
      "learning_rate": 5.44408621164791e-05,
      "loss": 2.1063,
      "step": 185523
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.425062656402588,
      "learning_rate": 5.443851226216127e-05,
      "loss": 1.9509,
      "step": 185524
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.3953427076339722,
      "learning_rate": 5.443616245349786e-05,
      "loss": 1.7687,
      "step": 185525
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.4146593809127808,
      "learning_rate": 5.4433812690489556e-05,
      "loss": 2.0224,
      "step": 185526
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.3929628133773804,
      "learning_rate": 5.44314629731365e-05,
      "loss": 1.8878,
      "step": 185527
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.2827922105789185,
      "learning_rate": 5.4429113301439354e-05,
      "loss": 1.8368,
      "step": 185528
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.626415729522705,
      "learning_rate": 5.4426763675398354e-05,
      "loss": 2.0088,
      "step": 185529
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.5559446811676025,
      "learning_rate": 5.442441409501417e-05,
      "loss": 2.1811,
      "step": 185530
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.3028568029403687,
      "learning_rate": 5.442206456028703e-05,
      "loss": 2.0807,
      "step": 185531
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.3837361335754395,
      "learning_rate": 5.441971507121763e-05,
      "loss": 1.9174,
      "step": 185532
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.3930106163024902,
      "learning_rate": 5.441736562780608e-05,
      "loss": 2.0476,
      "step": 185533
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.7446399927139282,
      "learning_rate": 5.441501623005307e-05,
      "loss": 2.0679,
      "step": 185534
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.330003261566162,
      "learning_rate": 5.4412666877958834e-05,
      "loss": 1.9772,
      "step": 185535
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.321111798286438,
      "learning_rate": 5.441031757152401e-05,
      "loss": 2.0718,
      "step": 185536
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.4408245086669922,
      "learning_rate": 5.440796831074886e-05,
      "loss": 1.7392,
      "step": 185537
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.360568881034851,
      "learning_rate": 5.440561909563409e-05,
      "loss": 2.0397,
      "step": 185538
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.2797664403915405,
      "learning_rate": 5.440326992617976e-05,
      "loss": 1.8711,
      "step": 185539
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.257523536682129,
      "learning_rate": 5.4400920802386604e-05,
      "loss": 1.9377,
      "step": 185540
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.237305998802185,
      "learning_rate": 5.4398571724254856e-05,
      "loss": 1.9725,
      "step": 185541
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.3759865760803223,
      "learning_rate": 5.4396222691785116e-05,
      "loss": 1.9634,
      "step": 185542
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.4783248901367188,
      "learning_rate": 5.4393873704977716e-05,
      "loss": 1.9887,
      "step": 185543
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.4240387678146362,
      "learning_rate": 5.439152476383326e-05,
      "loss": 2.0713,
      "step": 185544
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.6304659843444824,
      "learning_rate": 5.438917586835191e-05,
      "loss": 2.0148,
      "step": 185545
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.198058843612671,
      "learning_rate": 5.4386827018534293e-05,
      "loss": 1.9794,
      "step": 185546
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.3279039859771729,
      "learning_rate": 5.438447821438075e-05,
      "loss": 1.8904,
      "step": 185547
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.3519517183303833,
      "learning_rate": 5.438212945589186e-05,
      "loss": 1.9044,
      "step": 185548
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.4114738702774048,
      "learning_rate": 5.437978074306789e-05,
      "loss": 1.9519,
      "step": 185549
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.32816743850708,
      "learning_rate": 5.43774320759095e-05,
      "loss": 1.9903,
      "step": 185550
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.4262444972991943,
      "learning_rate": 5.437508345441681e-05,
      "loss": 2.0267,
      "step": 185551
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.3093639612197876,
      "learning_rate": 5.437273487859053e-05,
      "loss": 1.8556,
      "step": 185552
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.4289851188659668,
      "learning_rate": 5.437038634843092e-05,
      "loss": 2.1428,
      "step": 185553
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.3798942565917969,
      "learning_rate": 5.436803786393854e-05,
      "loss": 2.0502,
      "step": 185554
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.331830382347107,
      "learning_rate": 5.436568942511374e-05,
      "loss": 2.0094,
      "step": 185555
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.3071221113204956,
      "learning_rate": 5.436334103195713e-05,
      "loss": 1.9699,
      "step": 185556
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.2260944843292236,
      "learning_rate": 5.436099268446886e-05,
      "loss": 1.8379,
      "step": 185557
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.3543565273284912,
      "learning_rate": 5.4358644382649565e-05,
      "loss": 1.9801,
      "step": 185558
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.3754034042358398,
      "learning_rate": 5.4356296126499596e-05,
      "loss": 1.8902,
      "step": 185559
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.4331878423690796,
      "learning_rate": 5.43539479160195e-05,
      "loss": 2.1244,
      "step": 185560
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.341060996055603,
      "learning_rate": 5.435159975120954e-05,
      "loss": 1.9613,
      "step": 185561
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.380070686340332,
      "learning_rate": 5.4349251632070375e-05,
      "loss": 2.0852,
      "step": 185562
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.3720489740371704,
      "learning_rate": 5.4346903558602285e-05,
      "loss": 1.9557,
      "step": 185563
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.4255433082580566,
      "learning_rate": 5.434455553080576e-05,
      "loss": 1.9745,
      "step": 185564
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.3111495971679688,
      "learning_rate": 5.4342207548681125e-05,
      "loss": 2.1298,
      "step": 185565
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.2524454593658447,
      "learning_rate": 5.4339859612228996e-05,
      "loss": 2.1194,
      "step": 185566
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.3988221883773804,
      "learning_rate": 5.433751172144967e-05,
      "loss": 1.953,
      "step": 185567
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.3995639085769653,
      "learning_rate": 5.43351638763437e-05,
      "loss": 1.9456,
      "step": 185568
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.3224470615386963,
      "learning_rate": 5.433281607691143e-05,
      "loss": 1.9039,
      "step": 185569
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.4813120365142822,
      "learning_rate": 5.4330468323153366e-05,
      "loss": 2.1685,
      "step": 185570
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.3179445266723633,
      "learning_rate": 5.4328120615069824e-05,
      "loss": 2.0137,
      "step": 185571
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.2305686473846436,
      "learning_rate": 5.432577295266138e-05,
      "loss": 2.0232,
      "step": 185572
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.397815227508545,
      "learning_rate": 5.432342533592834e-05,
      "loss": 1.9426,
      "step": 185573
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.273218035697937,
      "learning_rate": 5.432107776487132e-05,
      "loss": 2.0559,
      "step": 185574
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.3852136135101318,
      "learning_rate": 5.4318730239490605e-05,
      "loss": 1.9635,
      "step": 185575
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.2947663068771362,
      "learning_rate": 5.4316382759786614e-05,
      "loss": 1.9848,
      "step": 185576
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.2447348833084106,
      "learning_rate": 5.431403532575995e-05,
      "loss": 1.9188,
      "step": 185577
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.4282282590866089,
      "learning_rate": 5.431168793741092e-05,
      "loss": 2.0018,
      "step": 185578
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.3395459651947021,
      "learning_rate": 5.430934059473988e-05,
      "loss": 2.1264,
      "step": 185579
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.291790246963501,
      "learning_rate": 5.4306993297747505e-05,
      "loss": 1.844,
      "step": 185580
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.2962965965270996,
      "learning_rate": 5.4304646046434056e-05,
      "loss": 1.9629,
      "step": 185581
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.4381060600280762,
      "learning_rate": 5.430229884079996e-05,
      "loss": 2.1717,
      "step": 185582
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.347115397453308,
      "learning_rate": 5.4299951680845765e-05,
      "loss": 2.0141,
      "step": 185583
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.3879755735397339,
      "learning_rate": 5.429760456657183e-05,
      "loss": 1.8933,
      "step": 185584
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.284531831741333,
      "learning_rate": 5.429525749797855e-05,
      "loss": 1.9263,
      "step": 185585
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.2340350151062012,
      "learning_rate": 5.429291047506653e-05,
      "loss": 2.0222,
      "step": 185586
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.6811139583587646,
      "learning_rate": 5.4290563497836036e-05,
      "loss": 2.1483,
      "step": 185587
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.3775402307510376,
      "learning_rate": 5.428821656628753e-05,
      "loss": 2.1177,
      "step": 185588
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.5404911041259766,
      "learning_rate": 5.4285869680421556e-05,
      "loss": 2.064,
      "step": 185589
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.5656418800354004,
      "learning_rate": 5.428352284023837e-05,
      "loss": 2.0404,
      "step": 185590
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.2378201484680176,
      "learning_rate": 5.428117604573864e-05,
      "loss": 1.867,
      "step": 185591
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.4202426671981812,
      "learning_rate": 5.4278829296922635e-05,
      "loss": 1.8484,
      "step": 185592
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.4213675260543823,
      "learning_rate": 5.427648259379085e-05,
      "loss": 1.9265,
      "step": 185593
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.4294579029083252,
      "learning_rate": 5.427413593634362e-05,
      "loss": 2.1429,
      "step": 185594
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.5143325328826904,
      "learning_rate": 5.427178932458155e-05,
      "loss": 2.1675,
      "step": 185595
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.3387846946716309,
      "learning_rate": 5.4269442758504934e-05,
      "loss": 1.8641,
      "step": 185596
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.3931058645248413,
      "learning_rate": 5.426709623811431e-05,
      "loss": 2.0202,
      "step": 185597
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.753029704093933,
      "learning_rate": 5.426474976341003e-05,
      "loss": 2.1042,
      "step": 185598
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.3794502019882202,
      "learning_rate": 5.426240333439271e-05,
      "loss": 2.0557,
      "step": 185599
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.3120437860488892,
      "learning_rate": 5.426005695106248e-05,
      "loss": 2.0647,
      "step": 185600
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.399733304977417,
      "learning_rate": 5.4257710613420035e-05,
      "loss": 2.0147,
      "step": 185601
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.3488692045211792,
      "learning_rate": 5.425536432146564e-05,
      "loss": 2.1213,
      "step": 185602
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.5999811887741089,
      "learning_rate": 5.425301807519987e-05,
      "loss": 2.0141,
      "step": 185603
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.547181487083435,
      "learning_rate": 5.425067187462305e-05,
      "loss": 2.0312,
      "step": 185604
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.4003877639770508,
      "learning_rate": 5.424832571973582e-05,
      "loss": 2.2606,
      "step": 185605
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.2276164293289185,
      "learning_rate": 5.4245979610538304e-05,
      "loss": 1.952,
      "step": 185606
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.407220721244812,
      "learning_rate": 5.4243633547031205e-05,
      "loss": 2.1018,
      "step": 185607
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.3087295293807983,
      "learning_rate": 5.424128752921473e-05,
      "loss": 2.0341,
      "step": 185608
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.4497829675674438,
      "learning_rate": 5.4238941557089563e-05,
      "loss": 2.0665,
      "step": 185609
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.4155348539352417,
      "learning_rate": 5.423659563065589e-05,
      "loss": 1.9837,
      "step": 185610
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.4602183103561401,
      "learning_rate": 5.423424974991446e-05,
      "loss": 1.9493,
      "step": 185611
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.3275583982467651,
      "learning_rate": 5.423190391486535e-05,
      "loss": 1.8896,
      "step": 185612
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.4063401222229004,
      "learning_rate": 5.4229558125509254e-05,
      "loss": 2.1139,
      "step": 185613
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.4677869081497192,
      "learning_rate": 5.422721238184647e-05,
      "loss": 1.9495,
      "step": 185614
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.5148557424545288,
      "learning_rate": 5.4224866683877546e-05,
      "loss": 2.1121,
      "step": 185615
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.3920228481292725,
      "learning_rate": 5.422252103160276e-05,
      "loss": 1.9367,
      "step": 185616
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.4952106475830078,
      "learning_rate": 5.422017542502283e-05,
      "loss": 2.2001,
      "step": 185617
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.3907229900360107,
      "learning_rate": 5.421782986413784e-05,
      "loss": 1.9909,
      "step": 185618
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.3841782808303833,
      "learning_rate": 5.421548434894847e-05,
      "loss": 1.9987,
      "step": 185619
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.3799488544464111,
      "learning_rate": 5.421313887945501e-05,
      "loss": 1.9747,
      "step": 185620
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.3843014240264893,
      "learning_rate": 5.4210793455658034e-05,
      "loss": 2.0414,
      "step": 185621
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.4634449481964111,
      "learning_rate": 5.420844807755783e-05,
      "loss": 2.0337,
      "step": 185622
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.3454198837280273,
      "learning_rate": 5.420610274515508e-05,
      "loss": 2.0724,
      "step": 185623
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.2922664880752563,
      "learning_rate": 5.42037574584499e-05,
      "loss": 1.9103,
      "step": 185624
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.5359190702438354,
      "learning_rate": 5.420141221744297e-05,
      "loss": 1.8454,
      "step": 185625
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.3617157936096191,
      "learning_rate": 5.4199067022134554e-05,
      "loss": 2.0118,
      "step": 185626
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.3400721549987793,
      "learning_rate": 5.419672187252525e-05,
      "loss": 1.7372,
      "step": 185627
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.4845081567764282,
      "learning_rate": 5.419437676861532e-05,
      "loss": 2.0474,
      "step": 185628
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.3358209133148193,
      "learning_rate": 5.419203171040547e-05,
      "loss": 1.9195,
      "step": 185629
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.3756047487258911,
      "learning_rate": 5.41896866978958e-05,
      "loss": 1.986,
      "step": 185630
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.3214964866638184,
      "learning_rate": 5.4187341731086965e-05,
      "loss": 2.0156,
      "step": 185631
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.246528148651123,
      "learning_rate": 5.418499680997931e-05,
      "loss": 2.0122,
      "step": 185632
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.3448195457458496,
      "learning_rate": 5.4182651934573326e-05,
      "loss": 2.1445,
      "step": 185633
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.307797908782959,
      "learning_rate": 5.4180307104869394e-05,
      "loss": 1.9768,
      "step": 185634
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.309705138206482,
      "learning_rate": 5.4177962320868066e-05,
      "loss": 1.8351,
      "step": 185635
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.247853398323059,
      "learning_rate": 5.417561758256964e-05,
      "loss": 2.0142,
      "step": 185636
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.3717069625854492,
      "learning_rate": 5.417327288997466e-05,
      "loss": 1.8936,
      "step": 185637
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.3383667469024658,
      "learning_rate": 5.417092824308342e-05,
      "loss": 2.1216,
      "step": 185638
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.3839259147644043,
      "learning_rate": 5.4168583641896525e-05,
      "loss": 1.7969,
      "step": 185639
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.341687798500061,
      "learning_rate": 5.416623908641423e-05,
      "loss": 2.014,
      "step": 185640
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.3439836502075195,
      "learning_rate": 5.4163894576637176e-05,
      "loss": 1.9365,
      "step": 185641
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.345205545425415,
      "learning_rate": 5.416155011256567e-05,
      "loss": 2.2249,
      "step": 185642
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.3701196908950806,
      "learning_rate": 5.415920569420019e-05,
      "loss": 2.0858,
      "step": 185643
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.324316143989563,
      "learning_rate": 5.4156861321541055e-05,
      "loss": 2.0148,
      "step": 185644
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.3235466480255127,
      "learning_rate": 5.4154516994588894e-05,
      "loss": 1.8755,
      "step": 185645
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.4472652673721313,
      "learning_rate": 5.4152172713343965e-05,
      "loss": 1.833,
      "step": 185646
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.556723952293396,
      "learning_rate": 5.414982847780688e-05,
      "loss": 1.901,
      "step": 185647
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.4149242639541626,
      "learning_rate": 5.4147484287977994e-05,
      "loss": 1.9827,
      "step": 185648
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.2567825317382812,
      "learning_rate": 5.414514014385771e-05,
      "loss": 2.1012,
      "step": 185649
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.3179128170013428,
      "learning_rate": 5.41427960454464e-05,
      "loss": 2.0215,
      "step": 185650
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.303415298461914,
      "learning_rate": 5.414045199274466e-05,
      "loss": 2.1876,
      "step": 185651
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.384786605834961,
      "learning_rate": 5.413810798575279e-05,
      "loss": 2.0081,
      "step": 185652
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.336613655090332,
      "learning_rate": 5.4135764024471384e-05,
      "loss": 2.0282,
      "step": 185653
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.5496734380722046,
      "learning_rate": 5.413342010890075e-05,
      "loss": 1.9255,
      "step": 185654
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.3195663690567017,
      "learning_rate": 5.4131076239041346e-05,
      "loss": 1.9874,
      "step": 185655
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.2479643821716309,
      "learning_rate": 5.412873241489355e-05,
      "loss": 1.983,
      "step": 185656
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.4249215126037598,
      "learning_rate": 5.412638863645795e-05,
      "loss": 2.0187,
      "step": 185657
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.408132553100586,
      "learning_rate": 5.412404490373482e-05,
      "loss": 2.1266,
      "step": 185658
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.4595459699630737,
      "learning_rate": 5.412170121672472e-05,
      "loss": 1.9679,
      "step": 185659
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.3882986307144165,
      "learning_rate": 5.411935757542809e-05,
      "loss": 2.0685,
      "step": 185660
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.3226816654205322,
      "learning_rate": 5.411701397984519e-05,
      "loss": 2.0048,
      "step": 185661
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.3693809509277344,
      "learning_rate": 5.411467042997666e-05,
      "loss": 2.1417,
      "step": 185662
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.467406988143921,
      "learning_rate": 5.411232692582289e-05,
      "loss": 1.975,
      "step": 185663
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.3124192953109741,
      "learning_rate": 5.4109983467384156e-05,
      "loss": 2.0463,
      "step": 185664
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.3203715085983276,
      "learning_rate": 5.4107640054661116e-05,
      "loss": 1.8298,
      "step": 185665
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.3891265392303467,
      "learning_rate": 5.410529668765411e-05,
      "loss": 1.9216,
      "step": 185666
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.306077480316162,
      "learning_rate": 5.41029533663635e-05,
      "loss": 2.0521,
      "step": 185667
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.372304081916809,
      "learning_rate": 5.410061009078985e-05,
      "loss": 2.1549,
      "step": 185668
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.3015804290771484,
      "learning_rate": 5.4098266860933534e-05,
      "loss": 1.9337,
      "step": 185669
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.440001368522644,
      "learning_rate": 5.409592367679495e-05,
      "loss": 2.0002,
      "step": 185670
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.395707607269287,
      "learning_rate": 5.409358053837463e-05,
      "loss": 1.9097,
      "step": 185671
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.4628270864486694,
      "learning_rate": 5.409123744567296e-05,
      "loss": 2.1733,
      "step": 185672
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.2577179670333862,
      "learning_rate": 5.40888943986903e-05,
      "loss": 1.8815,
      "step": 185673
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.2541569471359253,
      "learning_rate": 5.4086551397427226e-05,
      "loss": 2.0259,
      "step": 185674
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.3197580575942993,
      "learning_rate": 5.408420844188401e-05,
      "loss": 1.8682,
      "step": 185675
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.376003623008728,
      "learning_rate": 5.4081865532061265e-05,
      "loss": 1.994,
      "step": 185676
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.404096245765686,
      "learning_rate": 5.407952266795938e-05,
      "loss": 1.8569,
      "step": 185677
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.436328411102295,
      "learning_rate": 5.407717984957871e-05,
      "loss": 1.9819,
      "step": 185678
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.4029418230056763,
      "learning_rate": 5.407483707691967e-05,
      "loss": 1.9854,
      "step": 185679
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.4670326709747314,
      "learning_rate": 5.407249434998283e-05,
      "loss": 2.1301,
      "step": 185680
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.4028398990631104,
      "learning_rate": 5.4070151668768516e-05,
      "loss": 1.9144,
      "step": 185681
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.411685585975647,
      "learning_rate": 5.406780903327726e-05,
      "loss": 1.9944,
      "step": 185682
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.4841833114624023,
      "learning_rate": 5.406546644350935e-05,
      "loss": 2.1681,
      "step": 185683
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.4735394716262817,
      "learning_rate": 5.4063123899465476e-05,
      "loss": 2.0682,
      "step": 185684
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.4247854948043823,
      "learning_rate": 5.406078140114576e-05,
      "loss": 2.1014,
      "step": 185685
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.2602988481521606,
      "learning_rate": 5.405843894855083e-05,
      "loss": 1.9561,
      "step": 185686
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.5056467056274414,
      "learning_rate": 5.405609654168103e-05,
      "loss": 2.1747,
      "step": 185687
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.3233206272125244,
      "learning_rate": 5.4053754180536916e-05,
      "loss": 2.1228,
      "step": 185688
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.405507206916809,
      "learning_rate": 5.405141186511879e-05,
      "loss": 2.1395,
      "step": 185689
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.5746126174926758,
      "learning_rate": 5.4049069595427285e-05,
      "loss": 2.1314,
      "step": 185690
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.3682938814163208,
      "learning_rate": 5.404672737146257e-05,
      "loss": 1.9996,
      "step": 185691
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.58865487575531,
      "learning_rate": 5.404438519322525e-05,
      "loss": 1.9769,
      "step": 185692
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.3763976097106934,
      "learning_rate": 5.404204306071568e-05,
      "loss": 1.9662,
      "step": 185693
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.3297492265701294,
      "learning_rate": 5.40397009739344e-05,
      "loss": 2.0977,
      "step": 185694
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.4402368068695068,
      "learning_rate": 5.4037358932881705e-05,
      "loss": 2.0264,
      "step": 185695
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.3342523574829102,
      "learning_rate": 5.403501693755826e-05,
      "loss": 2.0566,
      "step": 185696
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.3914756774902344,
      "learning_rate": 5.403267498796421e-05,
      "loss": 1.8824,
      "step": 185697
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.348118782043457,
      "learning_rate": 5.4030333084100174e-05,
      "loss": 2.1131,
      "step": 185698
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.3489843606948853,
      "learning_rate": 5.40279912259665e-05,
      "loss": 2.0483,
      "step": 185699
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.3960217237472534,
      "learning_rate": 5.402564941356373e-05,
      "loss": 1.9608,
      "step": 185700
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.3816859722137451,
      "learning_rate": 5.4023307646892154e-05,
      "loss": 2.0699,
      "step": 185701
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.4604716300964355,
      "learning_rate": 5.402096592595236e-05,
      "loss": 2.126,
      "step": 185702
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.3372504711151123,
      "learning_rate": 5.401862425074472e-05,
      "loss": 1.9465,
      "step": 185703
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.4332951307296753,
      "learning_rate": 5.4016282621269635e-05,
      "loss": 2.04,
      "step": 185704
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.4354487657546997,
      "learning_rate": 5.401394103752753e-05,
      "loss": 1.8923,
      "step": 185705
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.4013177156448364,
      "learning_rate": 5.4011599499518945e-05,
      "loss": 2.1454,
      "step": 185706
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.273093342781067,
      "learning_rate": 5.400925800724414e-05,
      "loss": 1.9784,
      "step": 185707
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.339451551437378,
      "learning_rate": 5.400691656070378e-05,
      "loss": 2.1824,
      "step": 185708
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.3911488056182861,
      "learning_rate": 5.400457515989814e-05,
      "loss": 1.8453,
      "step": 185709
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.398498773574829,
      "learning_rate": 5.400223380482772e-05,
      "loss": 1.8934,
      "step": 185710
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.3206369876861572,
      "learning_rate": 5.3999892495492816e-05,
      "loss": 1.8772,
      "step": 185711
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.489419937133789,
      "learning_rate": 5.3997551231894086e-05,
      "loss": 2.0561,
      "step": 185712
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.2732292413711548,
      "learning_rate": 5.399521001403177e-05,
      "loss": 1.9562,
      "step": 185713
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.4431407451629639,
      "learning_rate": 5.399286884190644e-05,
      "loss": 2.1683,
      "step": 185714
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.327836036682129,
      "learning_rate": 5.399052771551853e-05,
      "loss": 1.8985,
      "step": 185715
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.3991130590438843,
      "learning_rate": 5.398818663486839e-05,
      "loss": 1.9885,
      "step": 185716
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.4591556787490845,
      "learning_rate": 5.39858455999564e-05,
      "loss": 1.9895,
      "step": 185717
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.2611769437789917,
      "learning_rate": 5.398350461078319e-05,
      "loss": 1.9154,
      "step": 185718
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.41907799243927,
      "learning_rate": 5.398116366734899e-05,
      "loss": 2.1878,
      "step": 185719
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.3343058824539185,
      "learning_rate": 5.397882276965444e-05,
      "loss": 1.9802,
      "step": 185720
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.2824921607971191,
      "learning_rate": 5.3976481917699865e-05,
      "loss": 1.909,
      "step": 185721
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.36406409740448,
      "learning_rate": 5.3974141111485705e-05,
      "loss": 1.9776,
      "step": 185722
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.3513808250427246,
      "learning_rate": 5.397180035101229e-05,
      "loss": 2.1085,
      "step": 185723
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.2533758878707886,
      "learning_rate": 5.3969459636280274e-05,
      "loss": 2.0502,
      "step": 185724
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.3633878231048584,
      "learning_rate": 5.396711896728988e-05,
      "loss": 1.8413,
      "step": 185725
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.3239949941635132,
      "learning_rate": 5.3964778344041725e-05,
      "loss": 1.8112,
      "step": 185726
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.2804152965545654,
      "learning_rate": 5.396243776653618e-05,
      "loss": 2.0513,
      "step": 185727
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.4623521566390991,
      "learning_rate": 5.396009723477365e-05,
      "loss": 2.1304,
      "step": 185728
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.4231737852096558,
      "learning_rate": 5.3957756748754486e-05,
      "loss": 1.8452,
      "step": 185729
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.3757599592208862,
      "learning_rate": 5.3955416308479336e-05,
      "loss": 2.0532,
      "step": 185730
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.4435782432556152,
      "learning_rate": 5.395307591394843e-05,
      "loss": 2.0618,
      "step": 185731
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.435976505279541,
      "learning_rate": 5.395073556516233e-05,
      "loss": 2.0602,
      "step": 185732
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.3522071838378906,
      "learning_rate": 5.394839526212147e-05,
      "loss": 1.7489,
      "step": 185733
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.3805062770843506,
      "learning_rate": 5.394605500482625e-05,
      "loss": 2.0346,
      "step": 185734
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.499138593673706,
      "learning_rate": 5.3943714793277006e-05,
      "loss": 2.194,
      "step": 185735
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.4071511030197144,
      "learning_rate": 5.3941374627474364e-05,
      "loss": 2.0911,
      "step": 185736
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.4454432725906372,
      "learning_rate": 5.393903450741856e-05,
      "loss": 2.0964,
      "step": 185737
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.3803985118865967,
      "learning_rate": 5.393669443311024e-05,
      "loss": 1.9698,
      "step": 185738
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.2322766780853271,
      "learning_rate": 5.3934354404549716e-05,
      "loss": 1.846,
      "step": 185739
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.295800805091858,
      "learning_rate": 5.393201442173747e-05,
      "loss": 2.3525,
      "step": 185740
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.384237289428711,
      "learning_rate": 5.392967448467379e-05,
      "loss": 2.3026,
      "step": 185741
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.2830615043640137,
      "learning_rate": 5.3927334593359354e-05,
      "loss": 2.1026,
      "step": 185742
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.4100627899169922,
      "learning_rate": 5.392499474779435e-05,
      "loss": 2.082,
      "step": 185743
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.3719624280929565,
      "learning_rate": 5.392265494797942e-05,
      "loss": 2.0458,
      "step": 185744
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.3228940963745117,
      "learning_rate": 5.392031519391492e-05,
      "loss": 2.104,
      "step": 185745
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.3778870105743408,
      "learning_rate": 5.39179754856013e-05,
      "loss": 1.9116,
      "step": 185746
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.4190819263458252,
      "learning_rate": 5.3915635823038874e-05,
      "loss": 2.0232,
      "step": 185747
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.3253185749053955,
      "learning_rate": 5.391329620622825e-05,
      "loss": 1.9887,
      "step": 185748
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.3369958400726318,
      "learning_rate": 5.391095663516973e-05,
      "loss": 1.788,
      "step": 185749
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.502402663230896,
      "learning_rate": 5.390861710986385e-05,
      "loss": 2.1317,
      "step": 185750
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.479657769203186,
      "learning_rate": 5.3906277630311036e-05,
      "loss": 2.0661,
      "step": 185751
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.2310072183609009,
      "learning_rate": 5.390393819651162e-05,
      "loss": 1.931,
      "step": 185752
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.5069547891616821,
      "learning_rate": 5.390159880846614e-05,
      "loss": 1.8781,
      "step": 185753
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.6224309206008911,
      "learning_rate": 5.389925946617503e-05,
      "loss": 1.5961,
      "step": 185754
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.4563919305801392,
      "learning_rate": 5.389692016963861e-05,
      "loss": 2.113,
      "step": 185755
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.503594994544983,
      "learning_rate": 5.38945809188575e-05,
      "loss": 1.8252,
      "step": 185756
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.3222485780715942,
      "learning_rate": 5.389224171383202e-05,
      "loss": 1.8915,
      "step": 185757
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.40272855758667,
      "learning_rate": 5.3889902554562506e-05,
      "loss": 2.1425,
      "step": 185758
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.3829761743545532,
      "learning_rate": 5.3887563441049616e-05,
      "loss": 1.8721,
      "step": 185759
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.6880055665969849,
      "learning_rate": 5.38852243732936e-05,
      "loss": 2.0122,
      "step": 185760
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.5853577852249146,
      "learning_rate": 5.3882885351295013e-05,
      "loss": 2.0891,
      "step": 185761
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.2182458639144897,
      "learning_rate": 5.3880546375054255e-05,
      "loss": 1.635,
      "step": 185762
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.276714563369751,
      "learning_rate": 5.387820744457176e-05,
      "loss": 1.925,
      "step": 185763
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.3100697994232178,
      "learning_rate": 5.38758685598479e-05,
      "loss": 2.0825,
      "step": 185764
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.505719780921936,
      "learning_rate": 5.38735297208832e-05,
      "loss": 2.0281,
      "step": 185765
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.3416681289672852,
      "learning_rate": 5.3871190927678e-05,
      "loss": 1.9423,
      "step": 185766
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.4800673723220825,
      "learning_rate": 5.386885218023286e-05,
      "loss": 2.0647,
      "step": 185767
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.3803257942199707,
      "learning_rate": 5.386651347854815e-05,
      "loss": 1.9912,
      "step": 185768
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.221512794494629,
      "learning_rate": 5.3864174822624196e-05,
      "loss": 2.0763,
      "step": 185769
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.4943962097167969,
      "learning_rate": 5.386183621246164e-05,
      "loss": 2.0714,
      "step": 185770
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.3840149641036987,
      "learning_rate": 5.3859497648060815e-05,
      "loss": 2.1516,
      "step": 185771
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.6282566785812378,
      "learning_rate": 5.385715912942208e-05,
      "loss": 2.0922,
      "step": 185772
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.3347246646881104,
      "learning_rate": 5.385482065654604e-05,
      "loss": 1.8802,
      "step": 185773
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.3610872030258179,
      "learning_rate": 5.3852482229432925e-05,
      "loss": 1.943,
      "step": 185774
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.3126378059387207,
      "learning_rate": 5.385014384808337e-05,
      "loss": 2.1295,
      "step": 185775
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.339171290397644,
      "learning_rate": 5.384780551249771e-05,
      "loss": 2.0745,
      "step": 185776
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.568658471107483,
      "learning_rate": 5.384546722267641e-05,
      "loss": 2.0993,
      "step": 185777
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.4366796016693115,
      "learning_rate": 5.3843128978619796e-05,
      "loss": 2.1941,
      "step": 185778
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.407118558883667,
      "learning_rate": 5.384079078032848e-05,
      "loss": 1.9615,
      "step": 185779
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.3945043087005615,
      "learning_rate": 5.383845262780272e-05,
      "loss": 2.0708,
      "step": 185780
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.441561222076416,
      "learning_rate": 5.3836114521043084e-05,
      "loss": 1.9416,
      "step": 185781
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.5730894804000854,
      "learning_rate": 5.383377646005e-05,
      "loss": 2.0215,
      "step": 185782
    },
    {
      "epoch": 2.42,
      "grad_norm": 5.408425807952881,
      "learning_rate": 5.383143844482385e-05,
      "loss": 2.1246,
      "step": 185783
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.3098807334899902,
      "learning_rate": 5.3829100475365016e-05,
      "loss": 2.1382,
      "step": 185784
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.2693949937820435,
      "learning_rate": 5.382676255167408e-05,
      "loss": 1.869,
      "step": 185785
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.3461823463439941,
      "learning_rate": 5.382442467375132e-05,
      "loss": 1.9614,
      "step": 185786
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.4808634519577026,
      "learning_rate": 5.382208684159729e-05,
      "loss": 2.1797,
      "step": 185787
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.3237165212631226,
      "learning_rate": 5.3819749055212416e-05,
      "loss": 1.8759,
      "step": 185788
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.3547905683517456,
      "learning_rate": 5.381741131459709e-05,
      "loss": 2.0863,
      "step": 185789
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.390055537223816,
      "learning_rate": 5.381507361975166e-05,
      "loss": 1.9905,
      "step": 185790
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.3918275833129883,
      "learning_rate": 5.381273597067675e-05,
      "loss": 2.0262,
      "step": 185791
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.539724349975586,
      "learning_rate": 5.381039836737262e-05,
      "loss": 2.0437,
      "step": 185792
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.4410622119903564,
      "learning_rate": 5.380806080983985e-05,
      "loss": 1.8941,
      "step": 185793
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.3403196334838867,
      "learning_rate": 5.380572329807883e-05,
      "loss": 2.0554,
      "step": 185794
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.4575778245925903,
      "learning_rate": 5.3803385832089964e-05,
      "loss": 2.0158,
      "step": 185795
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.389171838760376,
      "learning_rate": 5.380104841187361e-05,
      "loss": 1.9949,
      "step": 185796
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.44191312789917,
      "learning_rate": 5.379871103743035e-05,
      "loss": 2.0094,
      "step": 185797
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.4882876873016357,
      "learning_rate": 5.379637370876051e-05,
      "loss": 2.0824,
      "step": 185798
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.2925218343734741,
      "learning_rate": 5.379403642586465e-05,
      "loss": 2.0761,
      "step": 185799
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.3655091524124146,
      "learning_rate": 5.379169918874311e-05,
      "loss": 2.4462,
      "step": 185800
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.4646706581115723,
      "learning_rate": 5.378936199739635e-05,
      "loss": 1.9506,
      "step": 185801
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.2834025621414185,
      "learning_rate": 5.378702485182471e-05,
      "loss": 1.8175,
      "step": 185802
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.4162986278533936,
      "learning_rate": 5.378468775202879e-05,
      "loss": 2.0982,
      "step": 185803
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.2829657793045044,
      "learning_rate": 5.378235069800888e-05,
      "loss": 1.9764,
      "step": 185804
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.2905956506729126,
      "learning_rate": 5.3780013689765525e-05,
      "loss": 2.1814,
      "step": 185805
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.3941972255706787,
      "learning_rate": 5.377767672729915e-05,
      "loss": 2.1843,
      "step": 185806
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.3857057094573975,
      "learning_rate": 5.377533981061012e-05,
      "loss": 2.066,
      "step": 185807
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.581197738647461,
      "learning_rate": 5.377300293969884e-05,
      "loss": 1.9512,
      "step": 185808
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.3951830863952637,
      "learning_rate": 5.377066611456587e-05,
      "loss": 2.1743,
      "step": 185809
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.3499830961227417,
      "learning_rate": 5.3768329335211525e-05,
      "loss": 2.1921,
      "step": 185810
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.449487566947937,
      "learning_rate": 5.376599260163639e-05,
      "loss": 2.0271,
      "step": 185811
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.387506127357483,
      "learning_rate": 5.3763655913840765e-05,
      "loss": 2.0558,
      "step": 185812
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.3717504739761353,
      "learning_rate": 5.376131927182512e-05,
      "loss": 1.9946,
      "step": 185813
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.4177017211914062,
      "learning_rate": 5.3758982675589856e-05,
      "loss": 2.2055,
      "step": 185814
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.407014012336731,
      "learning_rate": 5.37566461251355e-05,
      "loss": 2.1374,
      "step": 185815
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.2571783065795898,
      "learning_rate": 5.375430962046237e-05,
      "loss": 1.9862,
      "step": 185816
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.3773781061172485,
      "learning_rate": 5.3751973161571e-05,
      "loss": 2.0782,
      "step": 185817
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.4464918375015259,
      "learning_rate": 5.374963674846182e-05,
      "loss": 2.1621,
      "step": 185818
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.2114142179489136,
      "learning_rate": 5.3747300381135214e-05,
      "loss": 2.1572,
      "step": 185819
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.4218838214874268,
      "learning_rate": 5.374496405959159e-05,
      "loss": 1.8379,
      "step": 185820
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.4443767070770264,
      "learning_rate": 5.374262778383147e-05,
      "loss": 2.0097,
      "step": 185821
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.4707640409469604,
      "learning_rate": 5.3740291553855166e-05,
      "loss": 2.1242,
      "step": 185822
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.4819940328598022,
      "learning_rate": 5.37379553696633e-05,
      "loss": 1.826,
      "step": 185823
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.3871210813522339,
      "learning_rate": 5.3735619231256176e-05,
      "loss": 2.0517,
      "step": 185824
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.3083622455596924,
      "learning_rate": 5.373328313863423e-05,
      "loss": 2.0396,
      "step": 185825
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.3352293968200684,
      "learning_rate": 5.373094709179786e-05,
      "loss": 1.9713,
      "step": 185826
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.310819387435913,
      "learning_rate": 5.372861109074763e-05,
      "loss": 2.154,
      "step": 185827
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.4643696546554565,
      "learning_rate": 5.372627513548381e-05,
      "loss": 1.7789,
      "step": 185828
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.53439462184906,
      "learning_rate": 5.372393922600703e-05,
      "loss": 1.975,
      "step": 185829
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.46359384059906,
      "learning_rate": 5.372160336231759e-05,
      "loss": 2.0181,
      "step": 185830
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.3105170726776123,
      "learning_rate": 5.3719267544415955e-05,
      "loss": 1.8503,
      "step": 185831
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.4263280630111694,
      "learning_rate": 5.3716931772302495e-05,
      "loss": 2.1439,
      "step": 185832
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.3185369968414307,
      "learning_rate": 5.371459604597777e-05,
      "loss": 2.0524,
      "step": 185833
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.4183670282363892,
      "learning_rate": 5.3712260365442094e-05,
      "loss": 2.0918,
      "step": 185834
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.3054231405258179,
      "learning_rate": 5.370992473069602e-05,
      "loss": 1.9017,
      "step": 185835
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.5954846143722534,
      "learning_rate": 5.370758914173985e-05,
      "loss": 2.026,
      "step": 185836
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.3385347127914429,
      "learning_rate": 5.370525359857415e-05,
      "loss": 2.2333,
      "step": 185837
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.356455683708191,
      "learning_rate": 5.370291810119929e-05,
      "loss": 2.1341,
      "step": 185838
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.4585950374603271,
      "learning_rate": 5.370058264961573e-05,
      "loss": 1.8444,
      "step": 185839
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.3388664722442627,
      "learning_rate": 5.3698247243823776e-05,
      "loss": 2.3341,
      "step": 185840
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.5313059091567993,
      "learning_rate": 5.369591188382406e-05,
      "loss": 1.912,
      "step": 185841
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.3446569442749023,
      "learning_rate": 5.369357656961685e-05,
      "loss": 1.9976,
      "step": 185842
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.3617833852767944,
      "learning_rate": 5.369124130120274e-05,
      "loss": 1.8615,
      "step": 185843
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.2881678342819214,
      "learning_rate": 5.368890607858206e-05,
      "loss": 2.066,
      "step": 185844
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.233744740486145,
      "learning_rate": 5.3686570901755185e-05,
      "loss": 2.0686,
      "step": 185845
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.2283515930175781,
      "learning_rate": 5.368423577072272e-05,
      "loss": 2.1148,
      "step": 185846
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.5407521724700928,
      "learning_rate": 5.368190068548498e-05,
      "loss": 2.0078,
      "step": 185847
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.2623989582061768,
      "learning_rate": 5.3679565646042374e-05,
      "loss": 2.1168,
      "step": 185848
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.4234646558761597,
      "learning_rate": 5.367723065239544e-05,
      "loss": 1.7821,
      "step": 185849
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.3300514221191406,
      "learning_rate": 5.367489570454457e-05,
      "loss": 1.9716,
      "step": 185850
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.7263303995132446,
      "learning_rate": 5.36725608024901e-05,
      "loss": 1.887,
      "step": 185851
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.4115879535675049,
      "learning_rate": 5.3670225946232637e-05,
      "loss": 1.9336,
      "step": 185852
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.5153758525848389,
      "learning_rate": 5.3667891135772534e-05,
      "loss": 1.7917,
      "step": 185853
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.2633382081985474,
      "learning_rate": 5.366555637111013e-05,
      "loss": 1.9921,
      "step": 185854
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.4326324462890625,
      "learning_rate": 5.3663221652246055e-05,
      "loss": 1.9494,
      "step": 185855
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.5691909790039062,
      "learning_rate": 5.366088697918062e-05,
      "loss": 1.9504,
      "step": 185856
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.455330729484558,
      "learning_rate": 5.365855235191417e-05,
      "loss": 2.0246,
      "step": 185857
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.5109163522720337,
      "learning_rate": 5.365621777044736e-05,
      "loss": 2.1947,
      "step": 185858
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.3525705337524414,
      "learning_rate": 5.3653883234780414e-05,
      "loss": 2.1387,
      "step": 185859
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.321187138557434,
      "learning_rate": 5.365154874491393e-05,
      "loss": 2.072,
      "step": 185860
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.3559982776641846,
      "learning_rate": 5.364921430084831e-05,
      "loss": 2.0092,
      "step": 185861
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.5284366607666016,
      "learning_rate": 5.3646879902583915e-05,
      "loss": 2.2839,
      "step": 185862
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.8302769660949707,
      "learning_rate": 5.3644545550121154e-05,
      "loss": 2.2331,
      "step": 185863
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.6007986068725586,
      "learning_rate": 5.364221124346059e-05,
      "loss": 1.8605,
      "step": 185864
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.4279403686523438,
      "learning_rate": 5.363987698260252e-05,
      "loss": 2.1129,
      "step": 185865
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.3696900606155396,
      "learning_rate": 5.36375427675475e-05,
      "loss": 2.1233,
      "step": 185866
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.640013337135315,
      "learning_rate": 5.3635208598295955e-05,
      "loss": 2.1744,
      "step": 185867
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.3630682229995728,
      "learning_rate": 5.3632874474848264e-05,
      "loss": 1.803,
      "step": 185868
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.2845157384872437,
      "learning_rate": 5.363054039720477e-05,
      "loss": 1.9878,
      "step": 185869
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.2804940938949585,
      "learning_rate": 5.36282063653661e-05,
      "loss": 1.9246,
      "step": 185870
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.383388638496399,
      "learning_rate": 5.3625872379332526e-05,
      "loss": 2.1005,
      "step": 185871
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.2236833572387695,
      "learning_rate": 5.362353843910461e-05,
      "loss": 2.0242,
      "step": 185872
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.2676202058792114,
      "learning_rate": 5.362120454468275e-05,
      "loss": 2.0055,
      "step": 185873
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.3197187185287476,
      "learning_rate": 5.361887069606735e-05,
      "loss": 2.0148,
      "step": 185874
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.339827537536621,
      "learning_rate": 5.361653689325878e-05,
      "loss": 2.1142,
      "step": 185875
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.277203917503357,
      "learning_rate": 5.361420313625763e-05,
      "loss": 2.0281,
      "step": 185876
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.3242207765579224,
      "learning_rate": 5.361186942506417e-05,
      "loss": 2.0449,
      "step": 185877
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.4356931447982788,
      "learning_rate": 5.3609535759678966e-05,
      "loss": 1.8357,
      "step": 185878
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.294970989227295,
      "learning_rate": 5.3607202140102424e-05,
      "loss": 1.9141,
      "step": 185879
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.4446983337402344,
      "learning_rate": 5.360486856633497e-05,
      "loss": 2.0373,
      "step": 185880
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.3380461931228638,
      "learning_rate": 5.3602535038376904e-05,
      "loss": 1.9503,
      "step": 185881
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.3949272632598877,
      "learning_rate": 5.3600201556228896e-05,
      "loss": 2.2328,
      "step": 185882
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.4567468166351318,
      "learning_rate": 5.3597868119891174e-05,
      "loss": 1.8667,
      "step": 185883
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.4568103551864624,
      "learning_rate": 5.359553472936431e-05,
      "loss": 2.2018,
      "step": 185884
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.3222647905349731,
      "learning_rate": 5.3593201384648696e-05,
      "loss": 2.0128,
      "step": 185885
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.3583545684814453,
      "learning_rate": 5.3590868085744774e-05,
      "loss": 1.9541,
      "step": 185886
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.3734376430511475,
      "learning_rate": 5.358853483265287e-05,
      "loss": 1.9503,
      "step": 185887
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.2947667837142944,
      "learning_rate": 5.3586201625373595e-05,
      "loss": 1.9357,
      "step": 185888
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.4505641460418701,
      "learning_rate": 5.358386846390723e-05,
      "loss": 1.8858,
      "step": 185889
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.4625401496887207,
      "learning_rate": 5.3581535348254355e-05,
      "loss": 2.162,
      "step": 185890
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.3110629320144653,
      "learning_rate": 5.35792022784153e-05,
      "loss": 1.9088,
      "step": 185891
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.377418875694275,
      "learning_rate": 5.357686925439053e-05,
      "loss": 1.896,
      "step": 185892
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.4411240816116333,
      "learning_rate": 5.3574536276180376e-05,
      "loss": 2.0305,
      "step": 185893
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.4090608358383179,
      "learning_rate": 5.357220334378548e-05,
      "loss": 1.8897,
      "step": 185894
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.2721179723739624,
      "learning_rate": 5.3569870457206065e-05,
      "loss": 1.9012,
      "step": 185895
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.3565154075622559,
      "learning_rate": 5.3567537616442735e-05,
      "loss": 2.1948,
      "step": 185896
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.2954721450805664,
      "learning_rate": 5.356520482149589e-05,
      "loss": 2.0469,
      "step": 185897
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.4553154706954956,
      "learning_rate": 5.356287207236589e-05,
      "loss": 2.0758,
      "step": 185898
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.4205970764160156,
      "learning_rate": 5.356053936905311e-05,
      "loss": 2.0467,
      "step": 185899
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.2735408544540405,
      "learning_rate": 5.355820671155818e-05,
      "loss": 1.925,
      "step": 185900
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.4662123918533325,
      "learning_rate": 5.355587409988137e-05,
      "loss": 2.239,
      "step": 185901
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.814787745475769,
      "learning_rate": 5.355354153402324e-05,
      "loss": 2.0463,
      "step": 185902
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.3252642154693604,
      "learning_rate": 5.355120901398409e-05,
      "loss": 1.7878,
      "step": 185903
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.3613404035568237,
      "learning_rate": 5.354887653976455e-05,
      "loss": 1.842,
      "step": 185904
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.3466051816940308,
      "learning_rate": 5.354654411136476e-05,
      "loss": 1.9488,
      "step": 185905
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.3226730823516846,
      "learning_rate": 5.3544211728785425e-05,
      "loss": 2.1451,
      "step": 185906
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.4206955432891846,
      "learning_rate": 5.3541879392026796e-05,
      "loss": 1.9847,
      "step": 185907
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.3310014009475708,
      "learning_rate": 5.3539547101089486e-05,
      "loss": 1.9682,
      "step": 185908
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.3011633157730103,
      "learning_rate": 5.3537214855973724e-05,
      "loss": 2.0382,
      "step": 185909
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.3563158512115479,
      "learning_rate": 5.353488265668017e-05,
      "loss": 1.9927,
      "step": 185910
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.3237732648849487,
      "learning_rate": 5.3532550503209026e-05,
      "loss": 1.8885,
      "step": 185911
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.3092341423034668,
      "learning_rate": 5.353021839556086e-05,
      "loss": 2.2046,
      "step": 185912
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.4300906658172607,
      "learning_rate": 5.3527886333736044e-05,
      "loss": 2.1719,
      "step": 185913
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.3602875471115112,
      "learning_rate": 5.352555431773511e-05,
      "loss": 2.0183,
      "step": 185914
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.3533432483673096,
      "learning_rate": 5.3523222347558384e-05,
      "loss": 2.0313,
      "step": 185915
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.397312879562378,
      "learning_rate": 5.3520890423206463e-05,
      "loss": 2.2234,
      "step": 185916
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.3735040426254272,
      "learning_rate": 5.351855854467949e-05,
      "loss": 1.9733,
      "step": 185917
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.3407716751098633,
      "learning_rate": 5.3516226711978164e-05,
      "loss": 1.9476,
      "step": 185918
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.2907078266143799,
      "learning_rate": 5.3513894925102774e-05,
      "loss": 2.2171,
      "step": 185919
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.57168447971344,
      "learning_rate": 5.3511563184053866e-05,
      "loss": 2.0675,
      "step": 185920
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.3544948101043701,
      "learning_rate": 5.3509231488831697e-05,
      "loss": 1.7856,
      "step": 185921
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.2871438264846802,
      "learning_rate": 5.350689983943694e-05,
      "loss": 1.9709,
      "step": 185922
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.382999300956726,
      "learning_rate": 5.3504568235869885e-05,
      "loss": 1.8917,
      "step": 185923
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.477901577949524,
      "learning_rate": 5.3502236678130975e-05,
      "loss": 1.9412,
      "step": 185924
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.4161527156829834,
      "learning_rate": 5.3499905166220566e-05,
      "loss": 1.9919,
      "step": 185925
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.2689827680587769,
      "learning_rate": 5.3497573700139274e-05,
      "loss": 1.9526,
      "step": 185926
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.3514816761016846,
      "learning_rate": 5.3495242279887384e-05,
      "loss": 1.8466,
      "step": 185927
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.2919963598251343,
      "learning_rate": 5.3492910905465405e-05,
      "loss": 1.9523,
      "step": 185928
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.3721362352371216,
      "learning_rate": 5.3490579576873796e-05,
      "loss": 1.9861,
      "step": 185929
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.4136669635772705,
      "learning_rate": 5.3488248294112925e-05,
      "loss": 2.1975,
      "step": 185930
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.336490511894226,
      "learning_rate": 5.348591705718312e-05,
      "loss": 1.9594,
      "step": 185931
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.3525818586349487,
      "learning_rate": 5.3483585866085066e-05,
      "loss": 2.0681,
      "step": 185932
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.349358320236206,
      "learning_rate": 5.348125472081898e-05,
      "loss": 1.796,
      "step": 185933
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.434694766998291,
      "learning_rate": 5.347892362138546e-05,
      "loss": 2.2545,
      "step": 185934
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.3531330823898315,
      "learning_rate": 5.347659256778485e-05,
      "loss": 2.0043,
      "step": 185935
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.4378552436828613,
      "learning_rate": 5.347426156001754e-05,
      "loss": 1.956,
      "step": 185936
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.3979259729385376,
      "learning_rate": 5.347193059808407e-05,
      "loss": 1.7962,
      "step": 185937
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.2536299228668213,
      "learning_rate": 5.3469599681984835e-05,
      "loss": 2.013,
      "step": 185938
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.4019416570663452,
      "learning_rate": 5.3467268811720165e-05,
      "loss": 1.8735,
      "step": 185939
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.4122663736343384,
      "learning_rate": 5.346493798729067e-05,
      "loss": 2.1113,
      "step": 185940
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.346253514289856,
      "learning_rate": 5.3462607208696705e-05,
      "loss": 1.9307,
      "step": 185941
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.4070547819137573,
      "learning_rate": 5.346027647593861e-05,
      "loss": 1.9555,
      "step": 185942
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.3601791858673096,
      "learning_rate": 5.345794578901699e-05,
      "loss": 2.0887,
      "step": 185943
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.4333133697509766,
      "learning_rate": 5.345561514793209e-05,
      "loss": 2.0198,
      "step": 185944
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.28226637840271,
      "learning_rate": 5.345328455268454e-05,
      "loss": 1.9783,
      "step": 185945
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.3649944067001343,
      "learning_rate": 5.3450954003274685e-05,
      "loss": 1.9576,
      "step": 185946
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.1922593116760254,
      "learning_rate": 5.3448623499702925e-05,
      "loss": 2.1753,
      "step": 185947
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.5178347826004028,
      "learning_rate": 5.344629304196967e-05,
      "loss": 2.0807,
      "step": 185948
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.3078629970550537,
      "learning_rate": 5.3443962630075474e-05,
      "loss": 2.0205,
      "step": 185949
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.5102590322494507,
      "learning_rate": 5.344163226402062e-05,
      "loss": 1.923,
      "step": 185950
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.554307222366333,
      "learning_rate": 5.3439301943805694e-05,
      "loss": 1.7658,
      "step": 185951
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.2953846454620361,
      "learning_rate": 5.343697166943106e-05,
      "loss": 2.1451,
      "step": 185952
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.4363054037094116,
      "learning_rate": 5.343464144089713e-05,
      "loss": 1.8861,
      "step": 185953
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.366101861000061,
      "learning_rate": 5.3432311258204296e-05,
      "loss": 2.0856,
      "step": 185954
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.51317298412323,
      "learning_rate": 5.3429981121353126e-05,
      "loss": 2.0004,
      "step": 185955
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.4026011228561401,
      "learning_rate": 5.342765103034389e-05,
      "loss": 1.8623,
      "step": 185956
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.271865725517273,
      "learning_rate": 5.342532098517721e-05,
      "loss": 2.1641,
      "step": 185957
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.2978137731552124,
      "learning_rate": 5.34229909858534e-05,
      "loss": 2.0427,
      "step": 185958
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.300230860710144,
      "learning_rate": 5.3420661032372917e-05,
      "loss": 2.1298,
      "step": 185959
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.350715160369873,
      "learning_rate": 5.341833112473609e-05,
      "loss": 2.0579,
      "step": 185960
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.4060503244400024,
      "learning_rate": 5.341600126294353e-05,
      "loss": 1.9437,
      "step": 185961
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.3758338689804077,
      "learning_rate": 5.341367144699553e-05,
      "loss": 1.988,
      "step": 185962
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.3625246286392212,
      "learning_rate": 5.341134167689266e-05,
      "loss": 2.4585,
      "step": 185963
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.285212516784668,
      "learning_rate": 5.340901195263518e-05,
      "loss": 1.9288,
      "step": 185964
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.383705973625183,
      "learning_rate": 5.34066822742238e-05,
      "loss": 1.9354,
      "step": 185965
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.4446816444396973,
      "learning_rate": 5.340435264165861e-05,
      "loss": 2.0482,
      "step": 185966
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.3621959686279297,
      "learning_rate": 5.340202305494028e-05,
      "loss": 2.0262,
      "step": 185967
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.3833491802215576,
      "learning_rate": 5.339969351406908e-05,
      "loss": 1.9793,
      "step": 185968
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.386206865310669,
      "learning_rate": 5.339736401904563e-05,
      "loss": 2.1203,
      "step": 185969
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.3378114700317383,
      "learning_rate": 5.339503456987018e-05,
      "loss": 2.1943,
      "step": 185970
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.4237171411514282,
      "learning_rate": 5.339270516654338e-05,
      "loss": 1.9653,
      "step": 185971
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.427750825881958,
      "learning_rate": 5.339037580906541e-05,
      "loss": 1.9398,
      "step": 185972
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.3161404132843018,
      "learning_rate": 5.3388046497436866e-05,
      "loss": 1.7824,
      "step": 185973
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.3265447616577148,
      "learning_rate": 5.3385717231658076e-05,
      "loss": 2.3105,
      "step": 185974
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.3529746532440186,
      "learning_rate": 5.3383388011729646e-05,
      "loss": 1.9131,
      "step": 185975
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.3751221895217896,
      "learning_rate": 5.33810588376518e-05,
      "loss": 2.0047,
      "step": 185976
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.3077281713485718,
      "learning_rate": 5.337872970942522e-05,
      "loss": 1.9192,
      "step": 185977
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.505894422531128,
      "learning_rate": 5.337640062705002e-05,
      "loss": 2.069,
      "step": 185978
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.4830191135406494,
      "learning_rate": 5.337407159052691e-05,
      "loss": 2.1075,
      "step": 185979
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.479419231414795,
      "learning_rate": 5.337174259985613e-05,
      "loss": 2.1657,
      "step": 185980
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.5953260660171509,
      "learning_rate": 5.336941365503826e-05,
      "loss": 1.782,
      "step": 185981
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.3767313957214355,
      "learning_rate": 5.336708475607362e-05,
      "loss": 1.9282,
      "step": 185982
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.3569046258926392,
      "learning_rate": 5.336475590296283e-05,
      "loss": 1.9186,
      "step": 185983
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.4481308460235596,
      "learning_rate": 5.3362427095706025e-05,
      "loss": 2.1864,
      "step": 185984
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.4410539865493774,
      "learning_rate": 5.336009833430388e-05,
      "loss": 2.0223,
      "step": 185985
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.550715446472168,
      "learning_rate": 5.335776961875669e-05,
      "loss": 1.8106,
      "step": 185986
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.3457412719726562,
      "learning_rate": 5.335544094906501e-05,
      "loss": 1.8724,
      "step": 185987
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.3403717279434204,
      "learning_rate": 5.335311232522915e-05,
      "loss": 2.0073,
      "step": 185988
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.2912378311157227,
      "learning_rate": 5.3350783747249746e-05,
      "loss": 1.9051,
      "step": 185989
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.3695704936981201,
      "learning_rate": 5.334845521512693e-05,
      "loss": 1.9368,
      "step": 185990
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.403198480606079,
      "learning_rate": 5.334612672886136e-05,
      "loss": 1.8922,
      "step": 185991
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.4695017337799072,
      "learning_rate": 5.334379828845334e-05,
      "loss": 2.1392,
      "step": 185992
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.3997701406478882,
      "learning_rate": 5.334146989390341e-05,
      "loss": 2.0051,
      "step": 185993
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.3362747430801392,
      "learning_rate": 5.3339141545211924e-05,
      "loss": 1.8675,
      "step": 185994
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.2982021570205688,
      "learning_rate": 5.333681324237946e-05,
      "loss": 1.9055,
      "step": 185995
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.164695143699646,
      "learning_rate": 5.333448498540621e-05,
      "loss": 1.772,
      "step": 185996
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.5270366668701172,
      "learning_rate": 5.333215677429281e-05,
      "loss": 1.8644,
      "step": 185997
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.3082659244537354,
      "learning_rate": 5.332982860903956e-05,
      "loss": 2.1054,
      "step": 185998
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.4170459508895874,
      "learning_rate": 5.332750048964699e-05,
      "loss": 2.0164,
      "step": 185999
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.4477262496948242,
      "learning_rate": 5.332517241611544e-05,
      "loss": 2.0544,
      "step": 186000
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.3801201581954956,
      "learning_rate": 5.3322844388445526e-05,
      "loss": 2.0284,
      "step": 186001
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.4254049062728882,
      "learning_rate": 5.3320516406637434e-05,
      "loss": 2.1082,
      "step": 186002
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.4276551008224487,
      "learning_rate": 5.331818847069179e-05,
      "loss": 1.8927,
      "step": 186003
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.7398476600646973,
      "learning_rate": 5.331586058060886e-05,
      "loss": 2.1128,
      "step": 186004
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.3060375452041626,
      "learning_rate": 5.331353273638924e-05,
      "loss": 1.8096,
      "step": 186005
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.3923698663711548,
      "learning_rate": 5.33112049380332e-05,
      "loss": 1.9217,
      "step": 186006
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.3664376735687256,
      "learning_rate": 5.330887718554138e-05,
      "loss": 1.9754,
      "step": 186007
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.5719503164291382,
      "learning_rate": 5.3306549478914064e-05,
      "loss": 1.8459,
      "step": 186008
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.406240701675415,
      "learning_rate": 5.330422181815173e-05,
      "loss": 2.1199,
      "step": 186009
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.4961042404174805,
      "learning_rate": 5.33018942032547e-05,
      "loss": 2.1041,
      "step": 186010
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.387967586517334,
      "learning_rate": 5.3299566634223625e-05,
      "loss": 2.0527,
      "step": 186011
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.3529469966888428,
      "learning_rate": 5.329723911105872e-05,
      "loss": 2.0141,
      "step": 186012
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.2887134552001953,
      "learning_rate": 5.329491163376056e-05,
      "loss": 1.8526,
      "step": 186013
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.417785406112671,
      "learning_rate": 5.329258420232958e-05,
      "loss": 2.1523,
      "step": 186014
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.5043542385101318,
      "learning_rate": 5.329025681676614e-05,
      "loss": 2.1205,
      "step": 186015
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.4666844606399536,
      "learning_rate": 5.328792947707065e-05,
      "loss": 1.8496,
      "step": 186016
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.44847571849823,
      "learning_rate": 5.328560218324363e-05,
      "loss": 2.0482,
      "step": 186017
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.4150594472885132,
      "learning_rate": 5.328327493528541e-05,
      "loss": 1.8533,
      "step": 186018
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.439914584159851,
      "learning_rate": 5.328094773319658e-05,
      "loss": 1.9396,
      "step": 186019
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.4931715726852417,
      "learning_rate": 5.327862057697745e-05,
      "loss": 2.1542,
      "step": 186020
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.32758629322052,
      "learning_rate": 5.327629346662843e-05,
      "loss": 2.0925,
      "step": 186021
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.5522289276123047,
      "learning_rate": 5.3273966402150046e-05,
      "loss": 1.9015,
      "step": 186022
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.4556300640106201,
      "learning_rate": 5.3271639383542705e-05,
      "loss": 1.9589,
      "step": 186023
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.4315457344055176,
      "learning_rate": 5.326931241080677e-05,
      "loss": 2.2708,
      "step": 186024
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.3181942701339722,
      "learning_rate": 5.326698548394278e-05,
      "loss": 1.9369,
      "step": 186025
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.2973452806472778,
      "learning_rate": 5.3264658602951124e-05,
      "loss": 2.1601,
      "step": 186026
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.349798321723938,
      "learning_rate": 5.326233176783218e-05,
      "loss": 1.9591,
      "step": 186027
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.4793208837509155,
      "learning_rate": 5.326000497858647e-05,
      "loss": 1.8266,
      "step": 186028
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.4498528242111206,
      "learning_rate": 5.3257678235214365e-05,
      "loss": 1.7935,
      "step": 186029
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.3268382549285889,
      "learning_rate": 5.3255351537716265e-05,
      "loss": 2.0792,
      "step": 186030
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.3009015321731567,
      "learning_rate": 5.325302488609273e-05,
      "loss": 1.8716,
      "step": 186031
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.3841885328292847,
      "learning_rate": 5.32506982803441e-05,
      "loss": 2.1364,
      "step": 186032
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.8691571950912476,
      "learning_rate": 5.324837172047075e-05,
      "loss": 2.0665,
      "step": 186033
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.4423387050628662,
      "learning_rate": 5.324604520647327e-05,
      "loss": 1.8258,
      "step": 186034
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.278913974761963,
      "learning_rate": 5.3243718738351914e-05,
      "loss": 2.198,
      "step": 186035
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.5058746337890625,
      "learning_rate": 5.32413923161073e-05,
      "loss": 1.9503,
      "step": 186036
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.3163633346557617,
      "learning_rate": 5.323906593973979e-05,
      "loss": 2.0583,
      "step": 186037
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.505495309829712,
      "learning_rate": 5.323673960924974e-05,
      "loss": 2.1914,
      "step": 186038
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.5853142738342285,
      "learning_rate": 5.3234413324637603e-05,
      "loss": 2.0625,
      "step": 186039
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.5320324897766113,
      "learning_rate": 5.323208708590393e-05,
      "loss": 1.997,
      "step": 186040
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.4147721529006958,
      "learning_rate": 5.322976089304899e-05,
      "loss": 2.21,
      "step": 186041
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.4142839908599854,
      "learning_rate": 5.3227434746073346e-05,
      "loss": 2.0464,
      "step": 186042
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.4854527711868286,
      "learning_rate": 5.322510864497731e-05,
      "loss": 2.1209,
      "step": 186043
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.6530016660690308,
      "learning_rate": 5.322278258976157e-05,
      "loss": 1.8969,
      "step": 186044
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.3931716680526733,
      "learning_rate": 5.32204565804262e-05,
      "loss": 2.0942,
      "step": 186045
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.3206250667572021,
      "learning_rate": 5.321813061697185e-05,
      "loss": 1.9069,
      "step": 186046
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.3124809265136719,
      "learning_rate": 5.3215804699398876e-05,
      "loss": 2.2915,
      "step": 186047
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.3543431758880615,
      "learning_rate": 5.3213478827707804e-05,
      "loss": 2.1854,
      "step": 186048
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.4080798625946045,
      "learning_rate": 5.321115300189892e-05,
      "loss": 2.1446,
      "step": 186049
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.4291304349899292,
      "learning_rate": 5.3208827221972904e-05,
      "loss": 2.0432,
      "step": 186050
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.267136573791504,
      "learning_rate": 5.3206501487929855e-05,
      "loss": 1.9429,
      "step": 186051
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.51178777217865,
      "learning_rate": 5.320417579977047e-05,
      "loss": 1.9758,
      "step": 186052
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.3443167209625244,
      "learning_rate": 5.320185015749501e-05,
      "loss": 2.1194,
      "step": 186053
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.4266022443771362,
      "learning_rate": 5.3199524561104044e-05,
      "loss": 2.1289,
      "step": 186054
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.262694239616394,
      "learning_rate": 5.319719901059791e-05,
      "loss": 1.9349,
      "step": 186055
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.3369454145431519,
      "learning_rate": 5.319487350597718e-05,
      "loss": 1.9155,
      "step": 186056
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.239526629447937,
      "learning_rate": 5.319254804724207e-05,
      "loss": 2.0284,
      "step": 186057
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.3678791522979736,
      "learning_rate": 5.3190222634393184e-05,
      "loss": 2.4059,
      "step": 186058
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.380387544631958,
      "learning_rate": 5.31878972674308e-05,
      "loss": 1.8583,
      "step": 186059
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.3601007461547852,
      "learning_rate": 5.3185571946355545e-05,
      "loss": 1.9633,
      "step": 186060
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.41380774974823,
      "learning_rate": 5.318324667116768e-05,
      "loss": 1.8661,
      "step": 186061
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.4625643491744995,
      "learning_rate": 5.3180921441867844e-05,
      "loss": 2.1443,
      "step": 186062
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.5515090227127075,
      "learning_rate": 5.317859625845616e-05,
      "loss": 1.9901,
      "step": 186063
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.1890900135040283,
      "learning_rate": 5.317627112093334e-05,
      "loss": 1.6184,
      "step": 186064
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.3382030725479126,
      "learning_rate": 5.317394602929962e-05,
      "loss": 1.7811,
      "step": 186065
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.5094382762908936,
      "learning_rate": 5.3171620983555616e-05,
      "loss": 2.026,
      "step": 186066
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.3389127254486084,
      "learning_rate": 5.316929598370158e-05,
      "loss": 1.7761,
      "step": 186067
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.320194959640503,
      "learning_rate": 5.316697102973819e-05,
      "loss": 1.9069,
      "step": 186068
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.3620017766952515,
      "learning_rate": 5.3164646121665565e-05,
      "loss": 2.0137,
      "step": 186069
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.277979850769043,
      "learning_rate": 5.3162321259484364e-05,
      "loss": 2.213,
      "step": 186070
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.3065147399902344,
      "learning_rate": 5.315999644319489e-05,
      "loss": 1.9475,
      "step": 186071
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.3012967109680176,
      "learning_rate": 5.315767167279766e-05,
      "loss": 2.1006,
      "step": 186072
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.4072413444519043,
      "learning_rate": 5.315534694829307e-05,
      "loss": 2.1192,
      "step": 186073
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.280552864074707,
      "learning_rate": 5.3153022269681654e-05,
      "loss": 2.0301,
      "step": 186074
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.5436594486236572,
      "learning_rate": 5.315069763696364e-05,
      "loss": 1.9968,
      "step": 186075
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.386929988861084,
      "learning_rate": 5.31483730501396e-05,
      "loss": 2.2222,
      "step": 186076
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.2677174806594849,
      "learning_rate": 5.3146048509209896e-05,
      "loss": 2.0195,
      "step": 186077
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.3618701696395874,
      "learning_rate": 5.314372401417507e-05,
      "loss": 2.228,
      "step": 186078
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.3007069826126099,
      "learning_rate": 5.31413995650354e-05,
      "loss": 1.9613,
      "step": 186079
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.453716516494751,
      "learning_rate": 5.3139075161791586e-05,
      "loss": 1.8244,
      "step": 186080
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.427065372467041,
      "learning_rate": 5.3136750804443696e-05,
      "loss": 2.0242,
      "step": 186081
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.3344943523406982,
      "learning_rate": 5.313442649299242e-05,
      "loss": 2.0406,
      "step": 186082
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.3736066818237305,
      "learning_rate": 5.313210222743807e-05,
      "loss": 1.869,
      "step": 186083
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.4210354089736938,
      "learning_rate": 5.3129778007781164e-05,
      "loss": 1.8868,
      "step": 186084
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.35380220413208,
      "learning_rate": 5.3127453834022014e-05,
      "loss": 2.1847,
      "step": 186085
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.4371777772903442,
      "learning_rate": 5.3125129706161316e-05,
      "loss": 1.8264,
      "step": 186086
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.3440313339233398,
      "learning_rate": 5.312280562419914e-05,
      "loss": 2.1282,
      "step": 186087
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.7770256996154785,
      "learning_rate": 5.312048158813618e-05,
      "loss": 2.1074,
      "step": 186088
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.4273673295974731,
      "learning_rate": 5.31181575979727e-05,
      "loss": 1.8465,
      "step": 186089
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.4891996383666992,
      "learning_rate": 5.311583365370927e-05,
      "loss": 2.0245,
      "step": 186090
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.290953516960144,
      "learning_rate": 5.311350975534623e-05,
      "loss": 1.8155,
      "step": 186091
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.4624656438827515,
      "learning_rate": 5.3111185902884165e-05,
      "loss": 2.1945,
      "step": 186092
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.315940022468567,
      "learning_rate": 5.310886209632329e-05,
      "loss": 2.127,
      "step": 186093
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.3386991024017334,
      "learning_rate": 5.310653833566416e-05,
      "loss": 1.8717,
      "step": 186094
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.3541206121444702,
      "learning_rate": 5.310421462090715e-05,
      "loss": 1.9613,
      "step": 186095
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.2770073413848877,
      "learning_rate": 5.310189095205278e-05,
      "loss": 1.9395,
      "step": 186096
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.423014521598816,
      "learning_rate": 5.309956732910137e-05,
      "loss": 1.9683,
      "step": 186097
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.419547438621521,
      "learning_rate": 5.3097243752053464e-05,
      "loss": 2.0669,
      "step": 186098
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.3459728956222534,
      "learning_rate": 5.3094920220909445e-05,
      "loss": 2.02,
      "step": 186099
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.3480199575424194,
      "learning_rate": 5.3092596735669734e-05,
      "loss": 1.8783,
      "step": 186100
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.2552434206008911,
      "learning_rate": 5.309027329633471e-05,
      "loss": 1.8608,
      "step": 186101
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.3777323961257935,
      "learning_rate": 5.308794990290496e-05,
      "loss": 1.9567,
      "step": 186102
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.2840945720672607,
      "learning_rate": 5.308562655538072e-05,
      "loss": 1.9354,
      "step": 186103
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.4736039638519287,
      "learning_rate": 5.3083303253762596e-05,
      "loss": 2.0318,
      "step": 186104
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.2488222122192383,
      "learning_rate": 5.308097999805095e-05,
      "loss": 2.0498,
      "step": 186105
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.3460338115692139,
      "learning_rate": 5.307865678824612e-05,
      "loss": 1.8668,
      "step": 186106
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.3447295427322388,
      "learning_rate": 5.307633362434873e-05,
      "loss": 1.8718,
      "step": 186107
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.5424237251281738,
      "learning_rate": 5.307401050635912e-05,
      "loss": 2.1475,
      "step": 186108
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.3723558187484741,
      "learning_rate": 5.307168743427762e-05,
      "loss": 2.014,
      "step": 186109
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.4812617301940918,
      "learning_rate": 5.306936440810483e-05,
      "loss": 2.0028,
      "step": 186110
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.8906198740005493,
      "learning_rate": 5.306704142784108e-05,
      "loss": 2.0581,
      "step": 186111
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.407008409500122,
      "learning_rate": 5.306471849348678e-05,
      "loss": 1.9112,
      "step": 186112
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.3664441108703613,
      "learning_rate": 5.306239560504249e-05,
      "loss": 2.0558,
      "step": 186113
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.4948492050170898,
      "learning_rate": 5.306007276250857e-05,
      "loss": 1.9157,
      "step": 186114
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.6471669673919678,
      "learning_rate": 5.305774996588536e-05,
      "loss": 2.0328,
      "step": 186115
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.3275611400604248,
      "learning_rate": 5.305542721517342e-05,
      "loss": 2.0476,
      "step": 186116
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.664900541305542,
      "learning_rate": 5.305310451037319e-05,
      "loss": 2.1482,
      "step": 186117
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.4534236192703247,
      "learning_rate": 5.305078185148494e-05,
      "loss": 2.1518,
      "step": 186118
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.5148495435714722,
      "learning_rate": 5.3048459238509285e-05,
      "loss": 2.2313,
      "step": 186119
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.8841369152069092,
      "learning_rate": 5.304613667144651e-05,
      "loss": 2.0237,
      "step": 186120
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.4491130113601685,
      "learning_rate": 5.3043814150297215e-05,
      "loss": 1.7198,
      "step": 186121
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.891093134880066,
      "learning_rate": 5.304149167506172e-05,
      "loss": 1.8538,
      "step": 186122
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.3694100379943848,
      "learning_rate": 5.303916924574047e-05,
      "loss": 1.8313,
      "step": 186123
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.3312278985977173,
      "learning_rate": 5.303684686233385e-05,
      "loss": 2.1069,
      "step": 186124
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.560659646987915,
      "learning_rate": 5.3034524524842405e-05,
      "loss": 2.0582,
      "step": 186125
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.6105263233184814,
      "learning_rate": 5.30322022332664e-05,
      "loss": 2.0946,
      "step": 186126
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.7716587781906128,
      "learning_rate": 5.302987998760646e-05,
      "loss": 1.9654,
      "step": 186127
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.340255856513977,
      "learning_rate": 5.302755778786286e-05,
      "loss": 2.1981,
      "step": 186128
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.4138153791427612,
      "learning_rate": 5.3025235634036264e-05,
      "loss": 1.8455,
      "step": 186129
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.361464262008667,
      "learning_rate": 5.3022913526126776e-05,
      "loss": 2.2642,
      "step": 186130
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.344215750694275,
      "learning_rate": 5.302059146413505e-05,
      "loss": 2.0662,
      "step": 186131
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.6032642126083374,
      "learning_rate": 5.3018269448061403e-05,
      "loss": 1.9698,
      "step": 186132
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.3990696668624878,
      "learning_rate": 5.301594747790639e-05,
      "loss": 1.9246,
      "step": 186133
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.6878142356872559,
      "learning_rate": 5.30136255536703e-05,
      "loss": 1.8992,
      "step": 186134
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.5434647798538208,
      "learning_rate": 5.301130367535379e-05,
      "loss": 1.9812,
      "step": 186135
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.3665525913238525,
      "learning_rate": 5.300898184295698e-05,
      "loss": 1.9842,
      "step": 186136
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.3937071561813354,
      "learning_rate": 5.300666005648056e-05,
      "loss": 2.225,
      "step": 186137
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.458979606628418,
      "learning_rate": 5.300433831592479e-05,
      "loss": 1.9167,
      "step": 186138
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.4215260744094849,
      "learning_rate": 5.300201662129021e-05,
      "loss": 2.1721,
      "step": 186139
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.3253337144851685,
      "learning_rate": 5.2999694972577166e-05,
      "loss": 1.9727,
      "step": 186140
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.6128958463668823,
      "learning_rate": 5.299737336978632e-05,
      "loss": 2.1429,
      "step": 186141
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.458297610282898,
      "learning_rate": 5.299505181291774e-05,
      "loss": 1.9259,
      "step": 186142
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.3957575559616089,
      "learning_rate": 5.299273030197213e-05,
      "loss": 2.1028,
      "step": 186143
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.3247593641281128,
      "learning_rate": 5.299040883694975e-05,
      "loss": 1.9568,
      "step": 186144
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.3612868785858154,
      "learning_rate": 5.29880874178512e-05,
      "loss": 2.0314,
      "step": 186145
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.3607773780822754,
      "learning_rate": 5.298576604467675e-05,
      "loss": 2.0609,
      "step": 186146
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.4523251056671143,
      "learning_rate": 5.298344471742706e-05,
      "loss": 1.9558,
      "step": 186147
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.3465479612350464,
      "learning_rate": 5.298112343610227e-05,
      "loss": 2.0179,
      "step": 186148
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.4760768413543701,
      "learning_rate": 5.297880220070301e-05,
      "loss": 2.0348,
      "step": 186149
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.3198283910751343,
      "learning_rate": 5.2976481011229575e-05,
      "loss": 1.8722,
      "step": 186150
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.5622570514678955,
      "learning_rate": 5.297415986768257e-05,
      "loss": 1.8635,
      "step": 186151
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.2537989616394043,
      "learning_rate": 5.2971838770062234e-05,
      "loss": 1.9214,
      "step": 186152
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.3373786211013794,
      "learning_rate": 5.296951771836926e-05,
      "loss": 2.2606,
      "step": 186153
    },
    {
      "epoch": 2.42,
      "grad_norm": 2.1112027168273926,
      "learning_rate": 5.2967196712603776e-05,
      "loss": 1.8709,
      "step": 186154
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.2760282754898071,
      "learning_rate": 5.2964875752766424e-05,
      "loss": 1.8388,
      "step": 186155
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.3527253866195679,
      "learning_rate": 5.296255483885746e-05,
      "loss": 1.8164,
      "step": 186156
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.3500385284423828,
      "learning_rate": 5.296023397087753e-05,
      "loss": 2.1121,
      "step": 186157
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.8350083827972412,
      "learning_rate": 5.295791314882686e-05,
      "loss": 2.042,
      "step": 186158
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.4015439748764038,
      "learning_rate": 5.295559237270615e-05,
      "loss": 2.0644,
      "step": 186159
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.4844706058502197,
      "learning_rate": 5.2953271642515494e-05,
      "loss": 2.0843,
      "step": 186160
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.3774100542068481,
      "learning_rate": 5.295095095825557e-05,
      "loss": 1.8451,
      "step": 186161
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.3102102279663086,
      "learning_rate": 5.294863031992663e-05,
      "loss": 1.9513,
      "step": 186162
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.4415736198425293,
      "learning_rate": 5.294630972752929e-05,
      "loss": 2.0842,
      "step": 186163
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.3847113847732544,
      "learning_rate": 5.294398918106384e-05,
      "loss": 1.8498,
      "step": 186164
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.4295789003372192,
      "learning_rate": 5.2941668680530814e-05,
      "loss": 2.039,
      "step": 186165
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.3673347234725952,
      "learning_rate": 5.293934822593061e-05,
      "loss": 2.1905,
      "step": 186166
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.403769850730896,
      "learning_rate": 5.2937027817263634e-05,
      "loss": 2.1398,
      "step": 186167
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.3969674110412598,
      "learning_rate": 5.2934707454530245e-05,
      "loss": 2.07,
      "step": 186168
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.596451759338379,
      "learning_rate": 5.2932387137731046e-05,
      "loss": 2.0753,
      "step": 186169
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.4177531003952026,
      "learning_rate": 5.2930066866866336e-05,
      "loss": 1.9083,
      "step": 186170
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.4530375003814697,
      "learning_rate": 5.292774664193662e-05,
      "loss": 1.855,
      "step": 186171
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.5443874597549438,
      "learning_rate": 5.2925426462942324e-05,
      "loss": 2.0974,
      "step": 186172
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.740959882736206,
      "learning_rate": 5.2923106329883844e-05,
      "loss": 1.9702,
      "step": 186173
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.5663331747055054,
      "learning_rate": 5.292078624276156e-05,
      "loss": 2.0284,
      "step": 186174
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.2376136779785156,
      "learning_rate": 5.2918466201576025e-05,
      "loss": 1.6897,
      "step": 186175
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.404383897781372,
      "learning_rate": 5.291614620632755e-05,
      "loss": 2.2258,
      "step": 186176
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.5943183898925781,
      "learning_rate": 5.291382625701669e-05,
      "loss": 2.2276,
      "step": 186177
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.2503248453140259,
      "learning_rate": 5.291150635364382e-05,
      "loss": 2.1341,
      "step": 186178
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.3378959894180298,
      "learning_rate": 5.290918649620936e-05,
      "loss": 1.9496,
      "step": 186179
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.3684440851211548,
      "learning_rate": 5.2906866684713666e-05,
      "loss": 2.0796,
      "step": 186180
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.4369471073150635,
      "learning_rate": 5.290454691915732e-05,
      "loss": 1.8733,
      "step": 186181
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.3659296035766602,
      "learning_rate": 5.2902227199540624e-05,
      "loss": 2.0322,
      "step": 186182
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.3837363719940186,
      "learning_rate": 5.2899907525864146e-05,
      "loss": 1.9903,
      "step": 186183
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.3195222616195679,
      "learning_rate": 5.2897587898128256e-05,
      "loss": 1.989,
      "step": 186184
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.3473024368286133,
      "learning_rate": 5.289526831633332e-05,
      "loss": 1.9561,
      "step": 186185
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.3637669086456299,
      "learning_rate": 5.2892948780479794e-05,
      "loss": 1.9822,
      "step": 186186
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.328525185585022,
      "learning_rate": 5.2890629290568185e-05,
      "loss": 2.1028,
      "step": 186187
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.3109623193740845,
      "learning_rate": 5.2888309846598797e-05,
      "loss": 2.1502,
      "step": 186188
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.4392169713974,
      "learning_rate": 5.288599044857219e-05,
      "loss": 1.8683,
      "step": 186189
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.4465627670288086,
      "learning_rate": 5.2883671096488766e-05,
      "loss": 2.0919,
      "step": 186190
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.5189212560653687,
      "learning_rate": 5.2881351790348924e-05,
      "loss": 1.9287,
      "step": 186191
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.4005351066589355,
      "learning_rate": 5.2879032530153064e-05,
      "loss": 1.7286,
      "step": 186192
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.4245001077651978,
      "learning_rate": 5.287671331590169e-05,
      "loss": 1.8906,
      "step": 186193
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.3344063758850098,
      "learning_rate": 5.2874394147595125e-05,
      "loss": 2.2056,
      "step": 186194
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.2430201768875122,
      "learning_rate": 5.2872075025233975e-05,
      "loss": 1.9664,
      "step": 186195
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.4639755487442017,
      "learning_rate": 5.286975594881857e-05,
      "loss": 1.8032,
      "step": 186196
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.2891268730163574,
      "learning_rate": 5.2867436918349245e-05,
      "loss": 1.8691,
      "step": 186197
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.3200843334197998,
      "learning_rate": 5.286511793382661e-05,
      "loss": 2.1313,
      "step": 186198
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.3842968940734863,
      "learning_rate": 5.286279899525101e-05,
      "loss": 2.0777,
      "step": 186199
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.3964388370513916,
      "learning_rate": 5.2860480102622824e-05,
      "loss": 2.1867,
      "step": 186200
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.3977508544921875,
      "learning_rate": 5.285816125594262e-05,
      "loss": 2.2684,
      "step": 186201
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.4080655574798584,
      "learning_rate": 5.285584245521073e-05,
      "loss": 2.1869,
      "step": 186202
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.3455129861831665,
      "learning_rate": 5.285352370042752e-05,
      "loss": 2.1357,
      "step": 186203
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.4580436944961548,
      "learning_rate": 5.285120499159361e-05,
      "loss": 1.9927,
      "step": 186204
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.3841524124145508,
      "learning_rate": 5.284888632870922e-05,
      "loss": 2.1685,
      "step": 186205
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.3918046951293945,
      "learning_rate": 5.2846567711774965e-05,
      "loss": 1.7938,
      "step": 186206
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.4800193309783936,
      "learning_rate": 5.284424914079123e-05,
      "loss": 2.0785,
      "step": 186207
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.3932703733444214,
      "learning_rate": 5.2841930615758396e-05,
      "loss": 2.0601,
      "step": 186208
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.4242122173309326,
      "learning_rate": 5.2839612136676844e-05,
      "loss": 2.0787,
      "step": 186209
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.3468365669250488,
      "learning_rate": 5.2837293703547135e-05,
      "loss": 2.1127,
      "step": 186210
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.3777892589569092,
      "learning_rate": 5.283497531636957e-05,
      "loss": 2.0593,
      "step": 186211
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.513368010520935,
      "learning_rate": 5.283265697514471e-05,
      "loss": 1.9216,
      "step": 186212
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.4684005975723267,
      "learning_rate": 5.283033867987295e-05,
      "loss": 2.1678,
      "step": 186213
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.4577205181121826,
      "learning_rate": 5.282802043055467e-05,
      "loss": 2.0315,
      "step": 186214
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.3967363834381104,
      "learning_rate": 5.282570222719027e-05,
      "loss": 2.2198,
      "step": 186215
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.3539520502090454,
      "learning_rate": 5.2823384069780305e-05,
      "loss": 1.8612,
      "step": 186216
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.725441575050354,
      "learning_rate": 5.282106595832508e-05,
      "loss": 1.8647,
      "step": 186217
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.3799352645874023,
      "learning_rate": 5.281874789282513e-05,
      "loss": 2.1219,
      "step": 186218
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.5102390050888062,
      "learning_rate": 5.281642987328079e-05,
      "loss": 2.1106,
      "step": 186219
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.3062732219696045,
      "learning_rate": 5.281411189969268e-05,
      "loss": 1.8272,
      "step": 186220
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.5578844547271729,
      "learning_rate": 5.281179397206094e-05,
      "loss": 2.0482,
      "step": 186221
    },
    {
      "epoch": 2.42,
      "grad_norm": 2.0365426540374756,
      "learning_rate": 5.2809476090386246e-05,
      "loss": 1.7742,
      "step": 186222
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.3004175424575806,
      "learning_rate": 5.2807158254668855e-05,
      "loss": 2.1925,
      "step": 186223
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.324465036392212,
      "learning_rate": 5.280484046490937e-05,
      "loss": 1.9646,
      "step": 186224
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.3506841659545898,
      "learning_rate": 5.280252272110802e-05,
      "loss": 1.6393,
      "step": 186225
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.3567430973052979,
      "learning_rate": 5.2800205023265505e-05,
      "loss": 2.1009,
      "step": 186226
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.5433417558670044,
      "learning_rate": 5.2797887371381964e-05,
      "loss": 2.0121,
      "step": 186227
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.4003243446350098,
      "learning_rate": 5.2795569765458025e-05,
      "loss": 1.8867,
      "step": 186228
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.3784748315811157,
      "learning_rate": 5.2793252205493985e-05,
      "loss": 1.8399,
      "step": 186229
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.339056372642517,
      "learning_rate": 5.279093469149042e-05,
      "loss": 1.7252,
      "step": 186230
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.3578168153762817,
      "learning_rate": 5.2788617223447615e-05,
      "loss": 2.1423,
      "step": 186231
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.2851866483688354,
      "learning_rate": 5.278629980136615e-05,
      "loss": 1.8638,
      "step": 186232
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.7882288694381714,
      "learning_rate": 5.2783982425246385e-05,
      "loss": 2.3902,
      "step": 186233
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.5530450344085693,
      "learning_rate": 5.278166509508872e-05,
      "loss": 2.0638,
      "step": 186234
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.3362928628921509,
      "learning_rate": 5.2779347810893555e-05,
      "loss": 2.0696,
      "step": 186235
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.4089467525482178,
      "learning_rate": 5.277703057266146e-05,
      "loss": 1.8434,
      "step": 186236
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.4829710721969604,
      "learning_rate": 5.277471338039266e-05,
      "loss": 1.972,
      "step": 186237
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.4399009943008423,
      "learning_rate": 5.277239623408783e-05,
      "loss": 1.9042,
      "step": 186238
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.4201322793960571,
      "learning_rate": 5.277007913374727e-05,
      "loss": 1.9215,
      "step": 186239
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.6480427980422974,
      "learning_rate": 5.27677620793714e-05,
      "loss": 2.0111,
      "step": 186240
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.324561357498169,
      "learning_rate": 5.2765445070960596e-05,
      "loss": 2.0117,
      "step": 186241
    },
    {
      "epoch": 2.42,
      "grad_norm": 1.3308329582214355,
      "learning_rate": 5.2763128108515464e-05,
      "loss": 2.0642,
      "step": 186242
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.457661509513855,
      "learning_rate": 5.2760811192036255e-05,
      "loss": 2.0401,
      "step": 186243
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.3784202337265015,
      "learning_rate": 5.275849432152351e-05,
      "loss": 1.8266,
      "step": 186244
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.3579025268554688,
      "learning_rate": 5.275617749697767e-05,
      "loss": 1.8732,
      "step": 186245
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.5112422704696655,
      "learning_rate": 5.275386071839912e-05,
      "loss": 2.0377,
      "step": 186246
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.3403301239013672,
      "learning_rate": 5.27515439857882e-05,
      "loss": 1.8559,
      "step": 186247
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.432918667793274,
      "learning_rate": 5.274922729914555e-05,
      "loss": 1.9491,
      "step": 186248
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.4365438222885132,
      "learning_rate": 5.2746910658471354e-05,
      "loss": 1.9935,
      "step": 186249
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.3136385679244995,
      "learning_rate": 5.274459406376629e-05,
      "loss": 1.947,
      "step": 186250
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.3822321891784668,
      "learning_rate": 5.2742277515030654e-05,
      "loss": 2.1929,
      "step": 186251
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.3152062892913818,
      "learning_rate": 5.273996101226491e-05,
      "loss": 2.0422,
      "step": 186252
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.2821890115737915,
      "learning_rate": 5.2737644555469396e-05,
      "loss": 1.8946,
      "step": 186253
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.3648494482040405,
      "learning_rate": 5.2735328144644706e-05,
      "loss": 2.0096,
      "step": 186254
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.281656265258789,
      "learning_rate": 5.2733011779791076e-05,
      "loss": 2.1135,
      "step": 186255
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.3243993520736694,
      "learning_rate": 5.273069546090914e-05,
      "loss": 1.9685,
      "step": 186256
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.3905460834503174,
      "learning_rate": 5.272837918799924e-05,
      "loss": 2.0535,
      "step": 186257
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.382143497467041,
      "learning_rate": 5.272606296106182e-05,
      "loss": 1.7594,
      "step": 186258
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.4217735528945923,
      "learning_rate": 5.2723746780097196e-05,
      "loss": 1.9597,
      "step": 186259
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.4485174417495728,
      "learning_rate": 5.2721430645105966e-05,
      "loss": 1.8826,
      "step": 186260
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.398982048034668,
      "learning_rate": 5.2719114556088426e-05,
      "loss": 1.7887,
      "step": 186261
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.3217612504959106,
      "learning_rate": 5.271679851304514e-05,
      "loss": 2.2056,
      "step": 186262
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.2880598306655884,
      "learning_rate": 5.271448251597645e-05,
      "loss": 1.8468,
      "step": 186263
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.4475326538085938,
      "learning_rate": 5.271216656488285e-05,
      "loss": 1.8268,
      "step": 186264
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.4037888050079346,
      "learning_rate": 5.2709850659764606e-05,
      "loss": 2.2156,
      "step": 186265
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.4582332372665405,
      "learning_rate": 5.270753480062239e-05,
      "loss": 1.9881,
      "step": 186266
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.3051420450210571,
      "learning_rate": 5.270521898745639e-05,
      "loss": 2.0202,
      "step": 186267
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.3116021156311035,
      "learning_rate": 5.270290322026728e-05,
      "loss": 1.9421,
      "step": 186268
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.295351266860962,
      "learning_rate": 5.2700587499055325e-05,
      "loss": 1.9028,
      "step": 186269
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.3676586151123047,
      "learning_rate": 5.269827182382102e-05,
      "loss": 2.107,
      "step": 186270
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.4150428771972656,
      "learning_rate": 5.269595619456468e-05,
      "loss": 2.039,
      "step": 186271
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.4131367206573486,
      "learning_rate": 5.269364061128692e-05,
      "loss": 2.2315,
      "step": 186272
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.291905403137207,
      "learning_rate": 5.269132507398802e-05,
      "loss": 2.0358,
      "step": 186273
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.4171857833862305,
      "learning_rate": 5.268900958266854e-05,
      "loss": 1.8994,
      "step": 186274
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.3832371234893799,
      "learning_rate": 5.268669413732881e-05,
      "loss": 2.0864,
      "step": 186275
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.361700415611267,
      "learning_rate": 5.268437873796934e-05,
      "loss": 1.9304,
      "step": 186276
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.4067518711090088,
      "learning_rate": 5.268206338459041e-05,
      "loss": 1.8539,
      "step": 186277
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.4634976387023926,
      "learning_rate": 5.267974807719261e-05,
      "loss": 2.0581,
      "step": 186278
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.4512860774993896,
      "learning_rate": 5.2677432815776285e-05,
      "loss": 1.905,
      "step": 186279
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.3723405599594116,
      "learning_rate": 5.267511760034192e-05,
      "loss": 1.9869,
      "step": 186280
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.2715107202529907,
      "learning_rate": 5.2672802430889936e-05,
      "loss": 1.9323,
      "step": 186281
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.2971113920211792,
      "learning_rate": 5.26704873074207e-05,
      "loss": 2.0257,
      "step": 186282
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.4713902473449707,
      "learning_rate": 5.2668172229934725e-05,
      "loss": 2.2474,
      "step": 186283
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.410185694694519,
      "learning_rate": 5.2665857198432426e-05,
      "loss": 2.0527,
      "step": 186284
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.418168067932129,
      "learning_rate": 5.266354221291414e-05,
      "loss": 1.9872,
      "step": 186285
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.4211901426315308,
      "learning_rate": 5.266122727338044e-05,
      "loss": 2.1022,
      "step": 186286
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.4085253477096558,
      "learning_rate": 5.2658912379831696e-05,
      "loss": 2.2934,
      "step": 186287
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.3380833864212036,
      "learning_rate": 5.265659753226823e-05,
      "loss": 2.0381,
      "step": 186288
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.4503098726272583,
      "learning_rate": 5.265428273069068e-05,
      "loss": 2.0815,
      "step": 186289
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.532068133354187,
      "learning_rate": 5.2651967975099274e-05,
      "loss": 1.8959,
      "step": 186290
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.276960015296936,
      "learning_rate": 5.2649653265494586e-05,
      "loss": 1.8974,
      "step": 186291
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.3885210752487183,
      "learning_rate": 5.2647338601877046e-05,
      "loss": 1.9173,
      "step": 186292
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.343534231185913,
      "learning_rate": 5.264502398424702e-05,
      "loss": 1.8903,
      "step": 186293
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.5785502195358276,
      "learning_rate": 5.2642709412604834e-05,
      "loss": 2.1136,
      "step": 186294
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.4654062986373901,
      "learning_rate": 5.2640394886951166e-05,
      "loss": 1.873,
      "step": 186295
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.2274601459503174,
      "learning_rate": 5.263808040728621e-05,
      "loss": 1.7969,
      "step": 186296
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.3253893852233887,
      "learning_rate": 5.26357659736106e-05,
      "loss": 1.9562,
      "step": 186297
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.435910701751709,
      "learning_rate": 5.263345158592467e-05,
      "loss": 1.915,
      "step": 186298
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.4369081258773804,
      "learning_rate": 5.263113724422875e-05,
      "loss": 2.0068,
      "step": 186299
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.2197203636169434,
      "learning_rate": 5.262882294852347e-05,
      "loss": 2.0571,
      "step": 186300
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.382863998413086,
      "learning_rate": 5.262650869880918e-05,
      "loss": 2.0442,
      "step": 186301
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.657395839691162,
      "learning_rate": 5.262419449508619e-05,
      "loss": 2.0812,
      "step": 186302
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.4605132341384888,
      "learning_rate": 5.2621880337355114e-05,
      "loss": 1.8626,
      "step": 186303
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.4067491292953491,
      "learning_rate": 5.261956622561622e-05,
      "loss": 2.0233,
      "step": 186304
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.3929332494735718,
      "learning_rate": 5.2617252159870093e-05,
      "loss": 2.2499,
      "step": 186305
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.4172312021255493,
      "learning_rate": 5.261493814011708e-05,
      "loss": 2.0944,
      "step": 186306
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.3898745775222778,
      "learning_rate": 5.261262416635764e-05,
      "loss": 2.2286,
      "step": 186307
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.3351036310195923,
      "learning_rate": 5.261031023859208e-05,
      "loss": 1.8993,
      "step": 186308
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.6962789297103882,
      "learning_rate": 5.260799635682102e-05,
      "loss": 2.1418,
      "step": 186309
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.368753433227539,
      "learning_rate": 5.260568252104474e-05,
      "loss": 1.9713,
      "step": 186310
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.530738353729248,
      "learning_rate": 5.260336873126381e-05,
      "loss": 2.1363,
      "step": 186311
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.3909114599227905,
      "learning_rate": 5.260105498747857e-05,
      "loss": 1.9634,
      "step": 186312
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.3380131721496582,
      "learning_rate": 5.259874128968948e-05,
      "loss": 1.8398,
      "step": 186313
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.367331624031067,
      "learning_rate": 5.2596427637896864e-05,
      "loss": 2.0406,
      "step": 186314
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.320441484451294,
      "learning_rate": 5.259411403210132e-05,
      "loss": 1.8315,
      "step": 186315
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.6969285011291504,
      "learning_rate": 5.2591800472303146e-05,
      "loss": 1.9916,
      "step": 186316
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.3957815170288086,
      "learning_rate": 5.258948695850288e-05,
      "loss": 2.0474,
      "step": 186317
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.4011391401290894,
      "learning_rate": 5.258717349070089e-05,
      "loss": 1.9376,
      "step": 186318
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.321940541267395,
      "learning_rate": 5.258486006889763e-05,
      "loss": 2.075,
      "step": 186319
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.6116034984588623,
      "learning_rate": 5.258254669309344e-05,
      "loss": 1.9791,
      "step": 186320
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.497752070426941,
      "learning_rate": 5.258023336328889e-05,
      "loss": 2.0749,
      "step": 186321
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.6958636045455933,
      "learning_rate": 5.257792007948428e-05,
      "loss": 1.8847,
      "step": 186322
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.5271186828613281,
      "learning_rate": 5.2575606841680175e-05,
      "loss": 2.0301,
      "step": 186323
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.3353928327560425,
      "learning_rate": 5.2573293649876936e-05,
      "loss": 1.9004,
      "step": 186324
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.6232056617736816,
      "learning_rate": 5.2570980504075e-05,
      "loss": 1.9539,
      "step": 186325
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.427236557006836,
      "learning_rate": 5.25686674042747e-05,
      "loss": 2.0763,
      "step": 186326
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.4506083726882935,
      "learning_rate": 5.256635435047664e-05,
      "loss": 2.1235,
      "step": 186327
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.58123779296875,
      "learning_rate": 5.256404134268111e-05,
      "loss": 1.7856,
      "step": 186328
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.5181472301483154,
      "learning_rate": 5.2561728380888645e-05,
      "loss": 1.9837,
      "step": 186329
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.4501311779022217,
      "learning_rate": 5.2559415465099616e-05,
      "loss": 2.0414,
      "step": 186330
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.541469931602478,
      "learning_rate": 5.255710259531449e-05,
      "loss": 2.1395,
      "step": 186331
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.7200233936309814,
      "learning_rate": 5.255478977153357e-05,
      "loss": 1.8855,
      "step": 186332
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.3590282201766968,
      "learning_rate": 5.2552476993757475e-05,
      "loss": 1.9482,
      "step": 186333
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.3870532512664795,
      "learning_rate": 5.255016426198645e-05,
      "loss": 2.0387,
      "step": 186334
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.3942320346832275,
      "learning_rate": 5.254785157622112e-05,
      "loss": 2.0166,
      "step": 186335
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.3707317113876343,
      "learning_rate": 5.2545538936461796e-05,
      "loss": 1.8944,
      "step": 186336
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.4660897254943848,
      "learning_rate": 5.254322634270897e-05,
      "loss": 1.9406,
      "step": 186337
    },
    {
      "epoch": 2.43,
      "grad_norm": 2.4936139583587646,
      "learning_rate": 5.2540913794962904e-05,
      "loss": 1.9993,
      "step": 186338
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.5355935096740723,
      "learning_rate": 5.2538601293224234e-05,
      "loss": 2.1255,
      "step": 186339
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.3572776317596436,
      "learning_rate": 5.253628883749327e-05,
      "loss": 2.0426,
      "step": 186340
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.5549594163894653,
      "learning_rate": 5.253397642777053e-05,
      "loss": 2.1906,
      "step": 186341
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.2979985475540161,
      "learning_rate": 5.253166406405639e-05,
      "loss": 2.1879,
      "step": 186342
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.4233298301696777,
      "learning_rate": 5.252935174635131e-05,
      "loss": 1.8853,
      "step": 186343
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.3742860555648804,
      "learning_rate": 5.2527039474655594e-05,
      "loss": 2.0654,
      "step": 186344
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.4771512746810913,
      "learning_rate": 5.252472724896985e-05,
      "loss": 1.8817,
      "step": 186345
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.4161181449890137,
      "learning_rate": 5.2522415069294354e-05,
      "loss": 2.2072,
      "step": 186346
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.4265989065170288,
      "learning_rate": 5.252010293562969e-05,
      "loss": 1.9393,
      "step": 186347
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.432103157043457,
      "learning_rate": 5.251779084797622e-05,
      "loss": 2.0199,
      "step": 186348
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.263729453086853,
      "learning_rate": 5.2515478806334385e-05,
      "loss": 2.1582,
      "step": 186349
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.5269081592559814,
      "learning_rate": 5.2513166810704474e-05,
      "loss": 2.193,
      "step": 186350
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.328826665878296,
      "learning_rate": 5.251085486108716e-05,
      "loss": 1.8838,
      "step": 186351
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.3193855285644531,
      "learning_rate": 5.250854295748264e-05,
      "loss": 2.2443,
      "step": 186352
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.4955317974090576,
      "learning_rate": 5.250623109989154e-05,
      "loss": 2.0453,
      "step": 186353
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.411767601966858,
      "learning_rate": 5.25039192883142e-05,
      "loss": 2.1296,
      "step": 186354
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.3571141958236694,
      "learning_rate": 5.250160752275106e-05,
      "loss": 1.7691,
      "step": 186355
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.5511448383331299,
      "learning_rate": 5.2499295803202435e-05,
      "loss": 2.0115,
      "step": 186356
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.3986947536468506,
      "learning_rate": 5.249698412966897e-05,
      "loss": 1.9243,
      "step": 186357
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.336489200592041,
      "learning_rate": 5.24946725021509e-05,
      "loss": 1.91,
      "step": 186358
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.307016372680664,
      "learning_rate": 5.249236092064882e-05,
      "loss": 1.9125,
      "step": 186359
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.5199284553527832,
      "learning_rate": 5.24900493851631e-05,
      "loss": 1.9493,
      "step": 186360
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.3547675609588623,
      "learning_rate": 5.248773789569414e-05,
      "loss": 2.0555,
      "step": 186361
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.3565338850021362,
      "learning_rate": 5.24854264522423e-05,
      "loss": 1.9199,
      "step": 186362
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.4270178079605103,
      "learning_rate": 5.2483115054808175e-05,
      "loss": 2.1107,
      "step": 186363
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.5784558057785034,
      "learning_rate": 5.248080370339202e-05,
      "loss": 1.9157,
      "step": 186364
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.3992879390716553,
      "learning_rate": 5.247849239799444e-05,
      "loss": 2.1118,
      "step": 186365
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.482603907585144,
      "learning_rate": 5.247618113861573e-05,
      "loss": 2.0083,
      "step": 186366
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.246692180633545,
      "learning_rate": 5.247386992525643e-05,
      "loss": 1.9278,
      "step": 186367
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.3862338066101074,
      "learning_rate": 5.247155875791689e-05,
      "loss": 2.0646,
      "step": 186368
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.3102498054504395,
      "learning_rate": 5.2469247636597563e-05,
      "loss": 2.0684,
      "step": 186369
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.4226937294006348,
      "learning_rate": 5.246693656129883e-05,
      "loss": 1.8538,
      "step": 186370
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.2593094110488892,
      "learning_rate": 5.246462553202122e-05,
      "loss": 1.8473,
      "step": 186371
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.365517020225525,
      "learning_rate": 5.2462314548765026e-05,
      "loss": 1.7457,
      "step": 186372
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.3712950944900513,
      "learning_rate": 5.2460003611530856e-05,
      "loss": 2.1593,
      "step": 186373
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.4173835515975952,
      "learning_rate": 5.2457692720319037e-05,
      "loss": 1.9546,
      "step": 186374
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.3371254205703735,
      "learning_rate": 5.245538187513e-05,
      "loss": 1.8381,
      "step": 186375
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.27103590965271,
      "learning_rate": 5.245307107596411e-05,
      "loss": 2.0737,
      "step": 186376
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.3908650875091553,
      "learning_rate": 5.2450760322821915e-05,
      "loss": 1.9619,
      "step": 186377
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.3033734560012817,
      "learning_rate": 5.244844961570377e-05,
      "loss": 1.9781,
      "step": 186378
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.7947591543197632,
      "learning_rate": 5.2446138954610173e-05,
      "loss": 1.9042,
      "step": 186379
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.478800654411316,
      "learning_rate": 5.244382833954153e-05,
      "loss": 2.0414,
      "step": 186380
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.4918314218521118,
      "learning_rate": 5.244151777049816e-05,
      "loss": 2.1853,
      "step": 186381
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.4204413890838623,
      "learning_rate": 5.2439207247480655e-05,
      "loss": 2.1457,
      "step": 186382
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.4787038564682007,
      "learning_rate": 5.243689677048939e-05,
      "loss": 2.1112,
      "step": 186383
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.563469648361206,
      "learning_rate": 5.243458633952471e-05,
      "loss": 2.236,
      "step": 186384
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.5882768630981445,
      "learning_rate": 5.243227595458718e-05,
      "loss": 2.0287,
      "step": 186385
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.4139224290847778,
      "learning_rate": 5.242996561567716e-05,
      "loss": 1.7792,
      "step": 186386
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.2912267446517944,
      "learning_rate": 5.242765532279502e-05,
      "loss": 1.8788,
      "step": 186387
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.3432968854904175,
      "learning_rate": 5.2425345075941305e-05,
      "loss": 2.2563,
      "step": 186388
    },
    {
      "epoch": 2.43,
      "grad_norm": 2.1557350158691406,
      "learning_rate": 5.2423034875116334e-05,
      "loss": 2.0303,
      "step": 186389
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.580115795135498,
      "learning_rate": 5.242072472032067e-05,
      "loss": 2.0007,
      "step": 186390
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.3701461553573608,
      "learning_rate": 5.241841461155466e-05,
      "loss": 1.8224,
      "step": 186391
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.4225386381149292,
      "learning_rate": 5.241610454881873e-05,
      "loss": 1.8391,
      "step": 186392
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.4225901365280151,
      "learning_rate": 5.241379453211327e-05,
      "loss": 1.9188,
      "step": 186393
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.3683565855026245,
      "learning_rate": 5.2411484561438836e-05,
      "loss": 1.8565,
      "step": 186394
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.3927531242370605,
      "learning_rate": 5.240917463679567e-05,
      "loss": 1.9654,
      "step": 186395
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.3111447095870972,
      "learning_rate": 5.240686475818442e-05,
      "loss": 2.1986,
      "step": 186396
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.4150108098983765,
      "learning_rate": 5.240455492560538e-05,
      "loss": 1.9866,
      "step": 186397
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.3298289775848389,
      "learning_rate": 5.240224513905902e-05,
      "loss": 1.9393,
      "step": 186398
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.3231024742126465,
      "learning_rate": 5.23999353985457e-05,
      "loss": 1.9013,
      "step": 186399
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.3201642036437988,
      "learning_rate": 5.239762570406596e-05,
      "loss": 1.9257,
      "step": 186400
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.4461582899093628,
      "learning_rate": 5.239531605562009e-05,
      "loss": 2.0712,
      "step": 186401
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.449289083480835,
      "learning_rate": 5.2393006453208706e-05,
      "loss": 1.9814,
      "step": 186402
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.344209909439087,
      "learning_rate": 5.239069689683213e-05,
      "loss": 1.8609,
      "step": 186403
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.552304983139038,
      "learning_rate": 5.238838738649076e-05,
      "loss": 1.7709,
      "step": 186404
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.3903599977493286,
      "learning_rate": 5.238607792218503e-05,
      "loss": 1.8973,
      "step": 186405
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.34507417678833,
      "learning_rate": 5.238376850391548e-05,
      "loss": 2.0168,
      "step": 186406
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.4195090532302856,
      "learning_rate": 5.2381459131682365e-05,
      "loss": 2.0808,
      "step": 186407
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.367271900177002,
      "learning_rate": 5.2379149805486295e-05,
      "loss": 2.0734,
      "step": 186408
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.3325462341308594,
      "learning_rate": 5.2376840525327604e-05,
      "loss": 2.3407,
      "step": 186409
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.4182698726654053,
      "learning_rate": 5.2374531291206745e-05,
      "loss": 2.0181,
      "step": 186410
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.33285391330719,
      "learning_rate": 5.237222210312406e-05,
      "loss": 2.0872,
      "step": 186411
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.422552227973938,
      "learning_rate": 5.236991296108012e-05,
      "loss": 2.1266,
      "step": 186412
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.2778366804122925,
      "learning_rate": 5.2367603865075215e-05,
      "loss": 1.9069,
      "step": 186413
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.4567328691482544,
      "learning_rate": 5.236529481510995e-05,
      "loss": 2.0775,
      "step": 186414
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.452778935432434,
      "learning_rate": 5.236298581118462e-05,
      "loss": 2.1125,
      "step": 186415
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.322513222694397,
      "learning_rate": 5.2360676853299666e-05,
      "loss": 2.1554,
      "step": 186416
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.3530513048171997,
      "learning_rate": 5.235836794145552e-05,
      "loss": 2.1301,
      "step": 186417
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.3257536888122559,
      "learning_rate": 5.2356059075652666e-05,
      "loss": 1.8444,
      "step": 186418
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.4394155740737915,
      "learning_rate": 5.235375025589142e-05,
      "loss": 1.9848,
      "step": 186419
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.4159616231918335,
      "learning_rate": 5.2351441482172354e-05,
      "loss": 2.1266,
      "step": 186420
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.4031345844268799,
      "learning_rate": 5.234913275449585e-05,
      "loss": 1.9526,
      "step": 186421
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.4543230533599854,
      "learning_rate": 5.2346824072862313e-05,
      "loss": 1.9153,
      "step": 186422
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.416656732559204,
      "learning_rate": 5.2344515437272086e-05,
      "loss": 2.04,
      "step": 186423
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.2614514827728271,
      "learning_rate": 5.2342206847725784e-05,
      "loss": 1.6796,
      "step": 186424
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.5034306049346924,
      "learning_rate": 5.233989830422366e-05,
      "loss": 1.8436,
      "step": 186425
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.3527944087982178,
      "learning_rate": 5.23375898067663e-05,
      "loss": 2.0883,
      "step": 186426
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.360283613204956,
      "learning_rate": 5.233528135535404e-05,
      "loss": 2.1487,
      "step": 186427
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.426196813583374,
      "learning_rate": 5.233297294998735e-05,
      "loss": 2.1623,
      "step": 186428
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.362358808517456,
      "learning_rate": 5.2330664590666525e-05,
      "loss": 1.9513,
      "step": 186429
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.3750104904174805,
      "learning_rate": 5.23283562773922e-05,
      "loss": 2.2619,
      "step": 186430
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.3292044401168823,
      "learning_rate": 5.2326048010164644e-05,
      "loss": 1.9334,
      "step": 186431
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.4189661741256714,
      "learning_rate": 5.232373978898442e-05,
      "loss": 2.0352,
      "step": 186432
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.2591910362243652,
      "learning_rate": 5.2321431613851826e-05,
      "loss": 1.8443,
      "step": 186433
    },
    {
      "epoch": 2.43,
      "grad_norm": 2.00858736038208,
      "learning_rate": 5.231912348476747e-05,
      "loss": 2.0751,
      "step": 186434
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.5472071170806885,
      "learning_rate": 5.231681540173154e-05,
      "loss": 2.1137,
      "step": 186435
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.3813399076461792,
      "learning_rate": 5.2314507364744674e-05,
      "loss": 1.8567,
      "step": 186436
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.3026267290115356,
      "learning_rate": 5.2312199373807106e-05,
      "loss": 2.1399,
      "step": 186437
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.3396247625350952,
      "learning_rate": 5.230989142891947e-05,
      "loss": 2.007,
      "step": 186438
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.2848756313323975,
      "learning_rate": 5.230758353008203e-05,
      "loss": 1.9251,
      "step": 186439
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.4804314374923706,
      "learning_rate": 5.230527567729542e-05,
      "loss": 2.0232,
      "step": 186440
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.4048436880111694,
      "learning_rate": 5.230296787055981e-05,
      "loss": 1.8609,
      "step": 186441
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.4501482248306274,
      "learning_rate": 5.230066010987582e-05,
      "loss": 1.8329,
      "step": 186442
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.5922929048538208,
      "learning_rate": 5.2298352395243736e-05,
      "loss": 1.8851,
      "step": 186443
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.4856189489364624,
      "learning_rate": 5.229604472666411e-05,
      "loss": 1.9292,
      "step": 186444
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.5448062419891357,
      "learning_rate": 5.229373710413728e-05,
      "loss": 1.9491,
      "step": 186445
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.5391498804092407,
      "learning_rate": 5.2291429527663875e-05,
      "loss": 1.9687,
      "step": 186446
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.3217073678970337,
      "learning_rate": 5.2289121997244e-05,
      "loss": 2.1418,
      "step": 186447
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.5895830392837524,
      "learning_rate": 5.228681451287835e-05,
      "loss": 1.9218,
      "step": 186448
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.460092306137085,
      "learning_rate": 5.228450707456716e-05,
      "loss": 2.0953,
      "step": 186449
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.5142079591751099,
      "learning_rate": 5.228219968231106e-05,
      "loss": 2.0108,
      "step": 186450
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.350893259048462,
      "learning_rate": 5.227989233611026e-05,
      "loss": 2.0542,
      "step": 186451
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.3346521854400635,
      "learning_rate": 5.227758503596542e-05,
      "loss": 1.8157,
      "step": 186452
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.3660097122192383,
      "learning_rate": 5.2275277781876836e-05,
      "loss": 2.0048,
      "step": 186453
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.4184366464614868,
      "learning_rate": 5.2272970573844944e-05,
      "loss": 1.961,
      "step": 186454
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.316833257675171,
      "learning_rate": 5.227066341187011e-05,
      "loss": 1.9799,
      "step": 186455
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.455842137336731,
      "learning_rate": 5.226835629595293e-05,
      "loss": 1.96,
      "step": 186456
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.4303086996078491,
      "learning_rate": 5.2266049226093644e-05,
      "loss": 2.0805,
      "step": 186457
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.4886462688446045,
      "learning_rate": 5.226374220229285e-05,
      "loss": 2.0503,
      "step": 186458
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.261199712753296,
      "learning_rate": 5.2261435224550904e-05,
      "loss": 1.9031,
      "step": 186459
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.5085653066635132,
      "learning_rate": 5.2259128292868254e-05,
      "loss": 1.8822,
      "step": 186460
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.4288359880447388,
      "learning_rate": 5.225682140724523e-05,
      "loss": 2.1377,
      "step": 186461
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.3029778003692627,
      "learning_rate": 5.2254514567682394e-05,
      "loss": 1.89,
      "step": 186462
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.453765869140625,
      "learning_rate": 5.225220777418004e-05,
      "loss": 1.9506,
      "step": 186463
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.3776472806930542,
      "learning_rate": 5.224990102673878e-05,
      "loss": 2.1436,
      "step": 186464
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.347536563873291,
      "learning_rate": 5.224759432535894e-05,
      "loss": 1.984,
      "step": 186465
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.5337328910827637,
      "learning_rate": 5.2245287670040856e-05,
      "loss": 2.135,
      "step": 186466
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.452043890953064,
      "learning_rate": 5.2242981060785126e-05,
      "loss": 2.0267,
      "step": 186467
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.3960052728652954,
      "learning_rate": 5.224067449759212e-05,
      "loss": 1.914,
      "step": 186468
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.535478115081787,
      "learning_rate": 5.223836798046216e-05,
      "loss": 2.0891,
      "step": 186469
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.5322984457015991,
      "learning_rate": 5.223606150939582e-05,
      "loss": 2.0056,
      "step": 186470
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.2973415851593018,
      "learning_rate": 5.22337550843935e-05,
      "loss": 2.0981,
      "step": 186471
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.3716338872909546,
      "learning_rate": 5.2231448705455504e-05,
      "loss": 2.0181,
      "step": 186472
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.338504433631897,
      "learning_rate": 5.222914237258246e-05,
      "loss": 1.9765,
      "step": 186473
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.4781020879745483,
      "learning_rate": 5.2226836085774694e-05,
      "loss": 1.8125,
      "step": 186474
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.4343823194503784,
      "learning_rate": 5.222452984503255e-05,
      "loss": 1.9765,
      "step": 186475
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.4218332767486572,
      "learning_rate": 5.222222365035662e-05,
      "loss": 1.7627,
      "step": 186476
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.4847370386123657,
      "learning_rate": 5.221991750174724e-05,
      "loss": 1.9201,
      "step": 186477
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.3850374221801758,
      "learning_rate": 5.2217611399204815e-05,
      "loss": 2.0346,
      "step": 186478
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.4457685947418213,
      "learning_rate": 5.2215305342729874e-05,
      "loss": 2.1234,
      "step": 186479
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.6882976293563843,
      "learning_rate": 5.221299933232268e-05,
      "loss": 1.9831,
      "step": 186480
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.2654908895492554,
      "learning_rate": 5.221069336798387e-05,
      "loss": 1.8244,
      "step": 186481
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.2752429246902466,
      "learning_rate": 5.220838744971377e-05,
      "loss": 1.9715,
      "step": 186482
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.2224091291427612,
      "learning_rate": 5.2206081577512826e-05,
      "loss": 1.7879,
      "step": 186483
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.5067024230957031,
      "learning_rate": 5.2203775751381326e-05,
      "loss": 1.7511,
      "step": 186484
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.3564872741699219,
      "learning_rate": 5.220146997131994e-05,
      "loss": 1.9224,
      "step": 186485
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.3061639070510864,
      "learning_rate": 5.2199164237328875e-05,
      "loss": 2.2724,
      "step": 186486
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.2997233867645264,
      "learning_rate": 5.219685854940875e-05,
      "loss": 1.8721,
      "step": 186487
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.5030052661895752,
      "learning_rate": 5.219455290755991e-05,
      "loss": 2.1139,
      "step": 186488
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.3647339344024658,
      "learning_rate": 5.219224731178278e-05,
      "loss": 2.0157,
      "step": 186489
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.4575152397155762,
      "learning_rate": 5.218994176207772e-05,
      "loss": 2.1297,
      "step": 186490
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.438833475112915,
      "learning_rate": 5.2187636258445283e-05,
      "loss": 2.0577,
      "step": 186491
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.3126566410064697,
      "learning_rate": 5.218533080088579e-05,
      "loss": 2.0774,
      "step": 186492
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.5487241744995117,
      "learning_rate": 5.218302538939978e-05,
      "loss": 2.2683,
      "step": 186493
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.3875712156295776,
      "learning_rate": 5.218072002398761e-05,
      "loss": 2.2305,
      "step": 186494
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.4822543859481812,
      "learning_rate": 5.217841470464975e-05,
      "loss": 2.1103,
      "step": 186495
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.494244933128357,
      "learning_rate": 5.21761094313865e-05,
      "loss": 2.106,
      "step": 186496
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.509372353553772,
      "learning_rate": 5.21738042041985e-05,
      "loss": 2.1711,
      "step": 186497
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.2703202962875366,
      "learning_rate": 5.2171499023085975e-05,
      "loss": 1.964,
      "step": 186498
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.4389384984970093,
      "learning_rate": 5.216919388804952e-05,
      "loss": 2.0452,
      "step": 186499
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.409941554069519,
      "learning_rate": 5.216688879908941e-05,
      "loss": 1.9475,
      "step": 186500
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.3639804124832153,
      "learning_rate": 5.2164583756206315e-05,
      "loss": 2.0254,
      "step": 186501
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.4604250192642212,
      "learning_rate": 5.216227875940032e-05,
      "loss": 2.1253,
      "step": 186502
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.5093785524368286,
      "learning_rate": 5.215997380867214e-05,
      "loss": 1.9793,
      "step": 186503
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.3245652914047241,
      "learning_rate": 5.215766890402203e-05,
      "loss": 2.0929,
      "step": 186504
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.466847538948059,
      "learning_rate": 5.215536404545056e-05,
      "loss": 1.9575,
      "step": 186505
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.464787244796753,
      "learning_rate": 5.2153059232958e-05,
      "loss": 1.8434,
      "step": 186506
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.478479027748108,
      "learning_rate": 5.215075446654502e-05,
      "loss": 1.9216,
      "step": 186507
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.745398759841919,
      "learning_rate": 5.214844974621174e-05,
      "loss": 1.893,
      "step": 186508
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.4413542747497559,
      "learning_rate": 5.21461450719588e-05,
      "loss": 2.2158,
      "step": 186509
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.2436388731002808,
      "learning_rate": 5.214384044378653e-05,
      "loss": 2.1251,
      "step": 186510
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.2839373350143433,
      "learning_rate": 5.214153586169547e-05,
      "loss": 2.0039,
      "step": 186511
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.398849606513977,
      "learning_rate": 5.2139231325685885e-05,
      "loss": 2.0688,
      "step": 186512
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.33293879032135,
      "learning_rate": 5.213692683575843e-05,
      "loss": 1.8302,
      "step": 186513
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.44338858127594,
      "learning_rate": 5.213462239191328e-05,
      "loss": 1.9213,
      "step": 186514
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.3395251035690308,
      "learning_rate": 5.213231799415104e-05,
      "loss": 2.1324,
      "step": 186515
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.4746861457824707,
      "learning_rate": 5.213001364247204e-05,
      "loss": 1.9425,
      "step": 186516
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.3636234998703003,
      "learning_rate": 5.2127709336876764e-05,
      "loss": 2.0218,
      "step": 186517
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.1835389137268066,
      "learning_rate": 5.21254050773656e-05,
      "loss": 1.8926,
      "step": 186518
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.3650827407836914,
      "learning_rate": 5.212310086393914e-05,
      "loss": 2.0571,
      "step": 186519
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.4603639841079712,
      "learning_rate": 5.212079669659751e-05,
      "loss": 1.9904,
      "step": 186520
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.306948184967041,
      "learning_rate": 5.211849257534139e-05,
      "loss": 2.0505,
      "step": 186521
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.3044520616531372,
      "learning_rate": 5.211618850017103e-05,
      "loss": 1.9039,
      "step": 186522
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.3362730741500854,
      "learning_rate": 5.211388447108704e-05,
      "loss": 1.8608,
      "step": 186523
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.4583162069320679,
      "learning_rate": 5.211158048808969e-05,
      "loss": 1.9189,
      "step": 186524
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.3962945938110352,
      "learning_rate": 5.210927655117963e-05,
      "loss": 2.0258,
      "step": 186525
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.2868421077728271,
      "learning_rate": 5.210697266035698e-05,
      "loss": 1.9623,
      "step": 186526
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.322949767112732,
      "learning_rate": 5.21046688156224e-05,
      "loss": 2.0445,
      "step": 186527
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.404141902923584,
      "learning_rate": 5.210236501697614e-05,
      "loss": 1.9742,
      "step": 186528
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.3803694248199463,
      "learning_rate": 5.2100061264418823e-05,
      "loss": 1.9773,
      "step": 186529
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.2173027992248535,
      "learning_rate": 5.20977575579507e-05,
      "loss": 2.1075,
      "step": 186530
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.3260244131088257,
      "learning_rate": 5.209545389757245e-05,
      "loss": 1.739,
      "step": 186531
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.4459272623062134,
      "learning_rate": 5.2093150283284194e-05,
      "loss": 2.0457,
      "step": 186532
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.3664950132369995,
      "learning_rate": 5.209084671508653e-05,
      "loss": 2.0567,
      "step": 186533
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.2852202653884888,
      "learning_rate": 5.2088543192979834e-05,
      "loss": 1.8825,
      "step": 186534
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.3241556882858276,
      "learning_rate": 5.2086239716964604e-05,
      "loss": 1.9855,
      "step": 186535
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.4464128017425537,
      "learning_rate": 5.208393628704114e-05,
      "loss": 1.9799,
      "step": 186536
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.3878768682479858,
      "learning_rate": 5.208163290321014e-05,
      "loss": 1.9082,
      "step": 186537
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.270907998085022,
      "learning_rate": 5.207932956547166e-05,
      "loss": 2.1703,
      "step": 186538
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.534985065460205,
      "learning_rate": 5.2077026273826386e-05,
      "loss": 2.1246,
      "step": 186539
    },
    {
      "epoch": 2.43,
      "grad_norm": 2.0011796951293945,
      "learning_rate": 5.207472302827463e-05,
      "loss": 1.867,
      "step": 186540
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.3550631999969482,
      "learning_rate": 5.207241982881691e-05,
      "loss": 1.9514,
      "step": 186541
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.3945298194885254,
      "learning_rate": 5.207011667545352e-05,
      "loss": 1.861,
      "step": 186542
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.285294532775879,
      "learning_rate": 5.206781356818509e-05,
      "loss": 1.9607,
      "step": 186543
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.3443013429641724,
      "learning_rate": 5.206551050701192e-05,
      "loss": 1.8757,
      "step": 186544
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.289209246635437,
      "learning_rate": 5.206320749193445e-05,
      "loss": 2.0792,
      "step": 186545
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.4368445873260498,
      "learning_rate": 5.2060904522953e-05,
      "loss": 2.0566,
      "step": 186546
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.2554014921188354,
      "learning_rate": 5.2058601600068205e-05,
      "loss": 1.9024,
      "step": 186547
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.4085948467254639,
      "learning_rate": 5.205629872328034e-05,
      "loss": 2.1377,
      "step": 186548
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.3949240446090698,
      "learning_rate": 5.205399589258997e-05,
      "loss": 1.963,
      "step": 186549
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.356533169746399,
      "learning_rate": 5.205169310799743e-05,
      "loss": 1.9871,
      "step": 186550
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.278721809387207,
      "learning_rate": 5.2049390369503076e-05,
      "loss": 1.9404,
      "step": 186551
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.4036803245544434,
      "learning_rate": 5.204708767710748e-05,
      "loss": 2.0808,
      "step": 186552
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.4127522706985474,
      "learning_rate": 5.204478503081104e-05,
      "loss": 1.8284,
      "step": 186553
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.358034372329712,
      "learning_rate": 5.204248243061406e-05,
      "loss": 1.8857,
      "step": 186554
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.3431626558303833,
      "learning_rate": 5.2040179876517174e-05,
      "loss": 1.9165,
      "step": 186555
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.4171884059906006,
      "learning_rate": 5.203787736852067e-05,
      "loss": 1.9329,
      "step": 186556
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.2625869512557983,
      "learning_rate": 5.203557490662493e-05,
      "loss": 1.7323,
      "step": 186557
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.3332984447479248,
      "learning_rate": 5.2033272490830547e-05,
      "loss": 2.1251,
      "step": 186558
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.417075514793396,
      "learning_rate": 5.203097012113785e-05,
      "loss": 2.0475,
      "step": 186559
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.5186735391616821,
      "learning_rate": 5.202866779754718e-05,
      "loss": 1.8687,
      "step": 186560
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.2756670713424683,
      "learning_rate": 5.202636552005915e-05,
      "loss": 1.9713,
      "step": 186561
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.389535665512085,
      "learning_rate": 5.202406328867412e-05,
      "loss": 1.9368,
      "step": 186562
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.3761425018310547,
      "learning_rate": 5.202176110339241e-05,
      "loss": 1.9135,
      "step": 186563
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.6070200204849243,
      "learning_rate": 5.201945896421462e-05,
      "loss": 2.114,
      "step": 186564
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.8138141632080078,
      "learning_rate": 5.201715687114099e-05,
      "loss": 2.0901,
      "step": 186565
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.2705857753753662,
      "learning_rate": 5.201485482417217e-05,
      "loss": 1.9771,
      "step": 186566
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.3629965782165527,
      "learning_rate": 5.201255282330844e-05,
      "loss": 1.6856,
      "step": 186567
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.382595419883728,
      "learning_rate": 5.201025086855026e-05,
      "loss": 1.9026,
      "step": 186568
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.4468666315078735,
      "learning_rate": 5.200794895989797e-05,
      "loss": 1.8324,
      "step": 186569
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.4630296230316162,
      "learning_rate": 5.2005647097352164e-05,
      "loss": 1.8955,
      "step": 186570
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.4302176237106323,
      "learning_rate": 5.2003345280913144e-05,
      "loss": 1.9289,
      "step": 186571
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.3334038257598877,
      "learning_rate": 5.200104351058144e-05,
      "loss": 1.895,
      "step": 186572
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.3722749948501587,
      "learning_rate": 5.199874178635732e-05,
      "loss": 2.0738,
      "step": 186573
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.4088175296783447,
      "learning_rate": 5.1996440108241485e-05,
      "loss": 2.0529,
      "step": 186574
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.417470097541809,
      "learning_rate": 5.199413847623407e-05,
      "loss": 1.9592,
      "step": 186575
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.4599982500076294,
      "learning_rate": 5.199183689033567e-05,
      "loss": 2.0588,
      "step": 186576
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.5483697652816772,
      "learning_rate": 5.198953535054662e-05,
      "loss": 1.9852,
      "step": 186577
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.349699854850769,
      "learning_rate": 5.1987233856867447e-05,
      "loss": 1.9718,
      "step": 186578
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.3353687524795532,
      "learning_rate": 5.198493240929846e-05,
      "loss": 1.7779,
      "step": 186579
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.3383489847183228,
      "learning_rate": 5.198263100784029e-05,
      "loss": 1.7987,
      "step": 186580
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.5693438053131104,
      "learning_rate": 5.19803296524931e-05,
      "loss": 1.9137,
      "step": 186581
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.3036671876907349,
      "learning_rate": 5.1978028343257526e-05,
      "loss": 1.8221,
      "step": 186582
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.249337077140808,
      "learning_rate": 5.197572708013387e-05,
      "loss": 1.8209,
      "step": 186583
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.327642560005188,
      "learning_rate": 5.197342586312262e-05,
      "loss": 2.1062,
      "step": 186584
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.4567370414733887,
      "learning_rate": 5.197112469222415e-05,
      "loss": 1.9798,
      "step": 186585
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.3900363445281982,
      "learning_rate": 5.19688235674391e-05,
      "loss": 2.1409,
      "step": 186586
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.3640708923339844,
      "learning_rate": 5.196652248876756e-05,
      "loss": 1.995,
      "step": 186587
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.3215538263320923,
      "learning_rate": 5.19642214562102e-05,
      "loss": 1.9605,
      "step": 186588
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.3372098207473755,
      "learning_rate": 5.196192046976728e-05,
      "loss": 2.0136,
      "step": 186589
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.272222638130188,
      "learning_rate": 5.195961952943941e-05,
      "loss": 1.9009,
      "step": 186590
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.2701828479766846,
      "learning_rate": 5.195731863522685e-05,
      "loss": 1.9192,
      "step": 186591
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.2780296802520752,
      "learning_rate": 5.195501778713027e-05,
      "loss": 2.1995,
      "step": 186592
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.364164113998413,
      "learning_rate": 5.1952716985149765e-05,
      "loss": 1.9416,
      "step": 186593
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.4220530986785889,
      "learning_rate": 5.195041622928604e-05,
      "loss": 1.9937,
      "step": 186594
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.3956544399261475,
      "learning_rate": 5.1948115519539326e-05,
      "loss": 2.1816,
      "step": 186595
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.4324922561645508,
      "learning_rate": 5.194581485591018e-05,
      "loss": 2.1018,
      "step": 186596
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.463376760482788,
      "learning_rate": 5.1943514238398954e-05,
      "loss": 2.0027,
      "step": 186597
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.3575457334518433,
      "learning_rate": 5.1941213667006264e-05,
      "loss": 2.0204,
      "step": 186598
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.304229736328125,
      "learning_rate": 5.193891314173222e-05,
      "loss": 2.1125,
      "step": 186599
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.3481067419052124,
      "learning_rate": 5.193661266257748e-05,
      "loss": 2.0198,
      "step": 186600
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.3863695859909058,
      "learning_rate": 5.193431222954232e-05,
      "loss": 2.1039,
      "step": 186601
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.4647127389907837,
      "learning_rate": 5.1932011842627367e-05,
      "loss": 2.0107,
      "step": 186602
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.4449868202209473,
      "learning_rate": 5.192971150183285e-05,
      "loss": 2.0141,
      "step": 186603
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.4418023824691772,
      "learning_rate": 5.192741120715941e-05,
      "loss": 2.2537,
      "step": 186604
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.6308625936508179,
      "learning_rate": 5.1925110958607206e-05,
      "loss": 1.8454,
      "step": 186605
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.3941172361373901,
      "learning_rate": 5.192281075617691e-05,
      "loss": 1.8954,
      "step": 186606
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.505808711051941,
      "learning_rate": 5.1920510599868726e-05,
      "loss": 2.1792,
      "step": 186607
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.3713281154632568,
      "learning_rate": 5.191821048968331e-05,
      "loss": 2.0885,
      "step": 186608
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.488804817199707,
      "learning_rate": 5.19159104256209e-05,
      "loss": 2.059,
      "step": 186609
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.4023643732070923,
      "learning_rate": 5.191361040768216e-05,
      "loss": 1.9149,
      "step": 186610
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.3232207298278809,
      "learning_rate": 5.1911310435867196e-05,
      "loss": 2.0842,
      "step": 186611
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.3939616680145264,
      "learning_rate": 5.1909010510176694e-05,
      "loss": 1.9552,
      "step": 186612
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.3991464376449585,
      "learning_rate": 5.19067106306109e-05,
      "loss": 2.0149,
      "step": 186613
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.4606090784072876,
      "learning_rate": 5.190441079717041e-05,
      "loss": 2.0539,
      "step": 186614
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.3748459815979004,
      "learning_rate": 5.1902111009855495e-05,
      "loss": 1.7044,
      "step": 186615
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.3242888450622559,
      "learning_rate": 5.189981126866685e-05,
      "loss": 2.0307,
      "step": 186616
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.4163084030151367,
      "learning_rate": 5.18975115736045e-05,
      "loss": 2.046,
      "step": 186617
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.3509140014648438,
      "learning_rate": 5.189521192466919e-05,
      "loss": 2.0269,
      "step": 186618
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.4029905796051025,
      "learning_rate": 5.189291232186118e-05,
      "loss": 1.9447,
      "step": 186619
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.336325764656067,
      "learning_rate": 5.189061276518104e-05,
      "loss": 2.1387,
      "step": 186620
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.4938074350357056,
      "learning_rate": 5.188831325462903e-05,
      "loss": 2.1386,
      "step": 186621
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.304337501525879,
      "learning_rate": 5.188601379020579e-05,
      "loss": 2.1534,
      "step": 186622
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.522902250289917,
      "learning_rate": 5.1883714371911524e-05,
      "loss": 2.14,
      "step": 186623
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.4145609140396118,
      "learning_rate": 5.188141499974682e-05,
      "loss": 2.0562,
      "step": 186624
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.417794108390808,
      "learning_rate": 5.1879115673711945e-05,
      "loss": 1.9857,
      "step": 186625
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.3030929565429688,
      "learning_rate": 5.1876816393807506e-05,
      "loss": 1.9645,
      "step": 186626
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.3460559844970703,
      "learning_rate": 5.18745171600338e-05,
      "loss": 2.1495,
      "step": 186627
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.4882291555404663,
      "learning_rate": 5.187221797239136e-05,
      "loss": 2.0832,
      "step": 186628
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.36434006690979,
      "learning_rate": 5.186991883088055e-05,
      "loss": 1.9897,
      "step": 186629
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.4931674003601074,
      "learning_rate": 5.186761973550184e-05,
      "loss": 2.089,
      "step": 186630
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.297894835472107,
      "learning_rate": 5.186532068625553e-05,
      "loss": 2.0083,
      "step": 186631
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.2526955604553223,
      "learning_rate": 5.1863021683142215e-05,
      "loss": 1.9363,
      "step": 186632
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.3678009510040283,
      "learning_rate": 5.186072272616216e-05,
      "loss": 2.0012,
      "step": 186633
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.2798877954483032,
      "learning_rate": 5.1858423815315976e-05,
      "loss": 2.008,
      "step": 186634
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.571340799331665,
      "learning_rate": 5.185612495060398e-05,
      "loss": 2.128,
      "step": 186635
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.3976895809173584,
      "learning_rate": 5.185382613202662e-05,
      "loss": 2.0955,
      "step": 186636
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.3467353582382202,
      "learning_rate": 5.185152735958425e-05,
      "loss": 1.9175,
      "step": 186637
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.3458616733551025,
      "learning_rate": 5.1849228633277415e-05,
      "loss": 2.0238,
      "step": 186638
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.680625557899475,
      "learning_rate": 5.184692995310644e-05,
      "loss": 1.9166,
      "step": 186639
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.3711973428726196,
      "learning_rate": 5.18446313190719e-05,
      "loss": 1.8907,
      "step": 186640
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.3554939031600952,
      "learning_rate": 5.184233273117408e-05,
      "loss": 2.1436,
      "step": 186641
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.3364050388336182,
      "learning_rate": 5.184003418941342e-05,
      "loss": 2.1041,
      "step": 186642
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.5189659595489502,
      "learning_rate": 5.183773569379042e-05,
      "loss": 1.9571,
      "step": 186643
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.3088297843933105,
      "learning_rate": 5.183543724430551e-05,
      "loss": 1.9226,
      "step": 186644
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.4144688844680786,
      "learning_rate": 5.183313884095896e-05,
      "loss": 1.975,
      "step": 186645
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.435987114906311,
      "learning_rate": 5.1830840483751414e-05,
      "loss": 2.1213,
      "step": 186646
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.2824313640594482,
      "learning_rate": 5.1828542172683186e-05,
      "loss": 1.91,
      "step": 186647
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.4871824979782104,
      "learning_rate": 5.1826243907754686e-05,
      "loss": 2.2345,
      "step": 186648
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.4242868423461914,
      "learning_rate": 5.182394568896641e-05,
      "loss": 2.0044,
      "step": 186649
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.3805474042892456,
      "learning_rate": 5.182164751631865e-05,
      "loss": 2.297,
      "step": 186650
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.4077311754226685,
      "learning_rate": 5.1819349389812024e-05,
      "loss": 2.0214,
      "step": 186651
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.4199182987213135,
      "learning_rate": 5.181705130944689e-05,
      "loss": 1.9858,
      "step": 186652
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.5089656114578247,
      "learning_rate": 5.181475327522364e-05,
      "loss": 2.0252,
      "step": 186653
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.4082951545715332,
      "learning_rate": 5.181245528714262e-05,
      "loss": 1.9644,
      "step": 186654
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.3330206871032715,
      "learning_rate": 5.181015734520442e-05,
      "loss": 1.8969,
      "step": 186655
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.3805181980133057,
      "learning_rate": 5.180785944940934e-05,
      "loss": 2.0248,
      "step": 186656
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.6018731594085693,
      "learning_rate": 5.180556159975795e-05,
      "loss": 2.0217,
      "step": 186657
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.3689972162246704,
      "learning_rate": 5.180326379625058e-05,
      "loss": 2.0326,
      "step": 186658
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.2757986783981323,
      "learning_rate": 5.1800966038887636e-05,
      "loss": 1.9754,
      "step": 186659
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.3765347003936768,
      "learning_rate": 5.179866832766955e-05,
      "loss": 2.1232,
      "step": 186660
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.3018735647201538,
      "learning_rate": 5.1796370662596845e-05,
      "loss": 2.0514,
      "step": 186661
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.2676788568496704,
      "learning_rate": 5.179407304366979e-05,
      "loss": 2.1179,
      "step": 186662
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.4219882488250732,
      "learning_rate": 5.1791775470888966e-05,
      "loss": 1.8706,
      "step": 186663
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.2474499940872192,
      "learning_rate": 5.1789477944254684e-05,
      "loss": 1.9522,
      "step": 186664
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.3683111667633057,
      "learning_rate": 5.1787180463767564e-05,
      "loss": 1.9427,
      "step": 186665
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.463505506515503,
      "learning_rate": 5.178488302942773e-05,
      "loss": 2.0071,
      "step": 186666
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.3357552289962769,
      "learning_rate": 5.178258564123587e-05,
      "loss": 1.9942,
      "step": 186667
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.4282476902008057,
      "learning_rate": 5.1780288299192207e-05,
      "loss": 2.1085,
      "step": 186668
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.327399492263794,
      "learning_rate": 5.177799100329739e-05,
      "loss": 1.7841,
      "step": 186669
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.195670485496521,
      "learning_rate": 5.177569375355163e-05,
      "loss": 1.923,
      "step": 186670
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.319661259651184,
      "learning_rate": 5.177339654995561e-05,
      "loss": 1.8973,
      "step": 186671
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.2993786334991455,
      "learning_rate": 5.177109939250945e-05,
      "loss": 1.9604,
      "step": 186672
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.2506217956542969,
      "learning_rate": 5.1768802281213804e-05,
      "loss": 1.9481,
      "step": 186673
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.3611820936203003,
      "learning_rate": 5.1766505216068943e-05,
      "loss": 2.0605,
      "step": 186674
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.3492491245269775,
      "learning_rate": 5.1764208197075466e-05,
      "loss": 1.8889,
      "step": 186675
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.352298378944397,
      "learning_rate": 5.1761911224233596e-05,
      "loss": 1.919,
      "step": 186676
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.4865686893463135,
      "learning_rate": 5.175961429754405e-05,
      "loss": 1.9377,
      "step": 186677
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.430888295173645,
      "learning_rate": 5.175731741700695e-05,
      "loss": 2.1372,
      "step": 186678
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.4788671731948853,
      "learning_rate": 5.1755020582622895e-05,
      "loss": 2.0525,
      "step": 186679
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.377429485321045,
      "learning_rate": 5.1752723794392194e-05,
      "loss": 2.1313,
      "step": 186680
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.355229139328003,
      "learning_rate": 5.175042705231544e-05,
      "loss": 1.9224,
      "step": 186681
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.3760491609573364,
      "learning_rate": 5.1748130356392855e-05,
      "loss": 2.0137,
      "step": 186682
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.3793162107467651,
      "learning_rate": 5.174583370662516e-05,
      "loss": 2.0873,
      "step": 186683
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.464245080947876,
      "learning_rate": 5.1743537103012445e-05,
      "loss": 2.0028,
      "step": 186684
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.2951908111572266,
      "learning_rate": 5.1741240545555376e-05,
      "loss": 2.1893,
      "step": 186685
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.3194377422332764,
      "learning_rate": 5.1738944034254216e-05,
      "loss": 2.0481,
      "step": 186686
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.4522695541381836,
      "learning_rate": 5.1736647569109535e-05,
      "loss": 1.7999,
      "step": 186687
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.3761780261993408,
      "learning_rate": 5.173435115012163e-05,
      "loss": 1.9146,
      "step": 186688
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.5381526947021484,
      "learning_rate": 5.173205477729114e-05,
      "loss": 2.0264,
      "step": 186689
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.4090769290924072,
      "learning_rate": 5.17297584506182e-05,
      "loss": 2.0689,
      "step": 186690
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.2571653127670288,
      "learning_rate": 5.172746217010346e-05,
      "loss": 2.0382,
      "step": 186691
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.3736281394958496,
      "learning_rate": 5.17251659357472e-05,
      "loss": 1.9893,
      "step": 186692
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.7524189949035645,
      "learning_rate": 5.172286974755002e-05,
      "loss": 1.8884,
      "step": 186693
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.3618971109390259,
      "learning_rate": 5.172057360551215e-05,
      "loss": 2.1274,
      "step": 186694
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.3609559535980225,
      "learning_rate": 5.171827750963425e-05,
      "loss": 1.9752,
      "step": 186695
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.3975625038146973,
      "learning_rate": 5.1715981459916467e-05,
      "loss": 2.1052,
      "step": 186696
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.4796098470687866,
      "learning_rate": 5.171368545635942e-05,
      "loss": 2.0847,
      "step": 186697
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.5960661172866821,
      "learning_rate": 5.171138949896345e-05,
      "loss": 1.9633,
      "step": 186698
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.4451303482055664,
      "learning_rate": 5.1709093587729054e-05,
      "loss": 1.8635,
      "step": 186699
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.4702056646347046,
      "learning_rate": 5.1706797722656604e-05,
      "loss": 1.9083,
      "step": 186700
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.6950582265853882,
      "learning_rate": 5.170450190374659e-05,
      "loss": 1.9341,
      "step": 186701
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.662381887435913,
      "learning_rate": 5.170220613099938e-05,
      "loss": 1.9714,
      "step": 186702
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.3485527038574219,
      "learning_rate": 5.1699910404415444e-05,
      "loss": 1.8666,
      "step": 186703
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.3685559034347534,
      "learning_rate": 5.1697614723995085e-05,
      "loss": 2.0927,
      "step": 186704
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.2416034936904907,
      "learning_rate": 5.169531908973893e-05,
      "loss": 1.7022,
      "step": 186705
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.3693748712539673,
      "learning_rate": 5.169302350164721e-05,
      "loss": 1.9798,
      "step": 186706
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.329513669013977,
      "learning_rate": 5.16907279597205e-05,
      "loss": 1.8267,
      "step": 186707
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.4048453569412231,
      "learning_rate": 5.168843246395919e-05,
      "loss": 2.0196,
      "step": 186708
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.3964221477508545,
      "learning_rate": 5.1686137014363694e-05,
      "loss": 1.7147,
      "step": 186709
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.4772648811340332,
      "learning_rate": 5.1683841610934365e-05,
      "loss": 1.9148,
      "step": 186710
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.4126900434494019,
      "learning_rate": 5.168154625367173e-05,
      "loss": 2.116,
      "step": 186711
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.3287233114242554,
      "learning_rate": 5.1679250942576144e-05,
      "loss": 2.0992,
      "step": 186712
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.2623627185821533,
      "learning_rate": 5.167695567764816e-05,
      "loss": 2.07,
      "step": 186713
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.4026806354522705,
      "learning_rate": 5.1674660458888074e-05,
      "loss": 2.1394,
      "step": 186714
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.3088425397872925,
      "learning_rate": 5.167236528629639e-05,
      "loss": 1.8065,
      "step": 186715
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.486947774887085,
      "learning_rate": 5.167007015987341e-05,
      "loss": 2.1078,
      "step": 186716
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.447410225868225,
      "learning_rate": 5.1667775079619733e-05,
      "loss": 2.0672,
      "step": 186717
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.4151815176010132,
      "learning_rate": 5.1665480045535624e-05,
      "loss": 2.117,
      "step": 186718
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.4751766920089722,
      "learning_rate": 5.166318505762168e-05,
      "loss": 1.9829,
      "step": 186719
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.3267035484313965,
      "learning_rate": 5.166089011587824e-05,
      "loss": 1.8371,
      "step": 186720
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.3402458429336548,
      "learning_rate": 5.16585952203057e-05,
      "loss": 2.154,
      "step": 186721
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.355026364326477,
      "learning_rate": 5.165630037090446e-05,
      "loss": 1.9071,
      "step": 186722
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.3424185514450073,
      "learning_rate": 5.165400556767508e-05,
      "loss": 2.0213,
      "step": 186723
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.475236177444458,
      "learning_rate": 5.165171081061781e-05,
      "loss": 2.0908,
      "step": 186724
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.6097795963287354,
      "learning_rate": 5.164941609973326e-05,
      "loss": 2.0173,
      "step": 186725
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.4656696319580078,
      "learning_rate": 5.164712143502179e-05,
      "loss": 2.1232,
      "step": 186726
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.500428318977356,
      "learning_rate": 5.164482681648371e-05,
      "loss": 2.1724,
      "step": 186727
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.551357626914978,
      "learning_rate": 5.1642532244119625e-05,
      "loss": 2.043,
      "step": 186728
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.422982931137085,
      "learning_rate": 5.164023771792991e-05,
      "loss": 1.7822,
      "step": 186729
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.234918475151062,
      "learning_rate": 5.163794323791486e-05,
      "loss": 1.7506,
      "step": 186730
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.5311492681503296,
      "learning_rate": 5.163564880407507e-05,
      "loss": 2.1587,
      "step": 186731
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.2754592895507812,
      "learning_rate": 5.163335441641091e-05,
      "loss": 2.0725,
      "step": 186732
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.3081461191177368,
      "learning_rate": 5.1631060074922746e-05,
      "loss": 2.0608,
      "step": 186733
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.4984204769134521,
      "learning_rate": 5.162876577961112e-05,
      "loss": 1.9812,
      "step": 186734
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.454384446144104,
      "learning_rate": 5.1626471530476276e-05,
      "loss": 2.1793,
      "step": 186735
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.5565522909164429,
      "learning_rate": 5.16241773275189e-05,
      "loss": 1.8489,
      "step": 186736
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.4305615425109863,
      "learning_rate": 5.162188317073922e-05,
      "loss": 2.0694,
      "step": 186737
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.4753575325012207,
      "learning_rate": 5.161958906013777e-05,
      "loss": 2.0301,
      "step": 186738
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.6605448722839355,
      "learning_rate": 5.1617294995714844e-05,
      "loss": 1.9641,
      "step": 186739
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.527309775352478,
      "learning_rate": 5.161500097747101e-05,
      "loss": 2.094,
      "step": 186740
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.3194596767425537,
      "learning_rate": 5.161270700540654e-05,
      "loss": 2.0037,
      "step": 186741
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.6067694425582886,
      "learning_rate": 5.161041307952206e-05,
      "loss": 2.1254,
      "step": 186742
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.5143624544143677,
      "learning_rate": 5.1608119199817874e-05,
      "loss": 2.062,
      "step": 186743
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.3348300457000732,
      "learning_rate": 5.160582536629445e-05,
      "loss": 1.9674,
      "step": 186744
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.4332870244979858,
      "learning_rate": 5.1603531578952115e-05,
      "loss": 2.0233,
      "step": 186745
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.4745490550994873,
      "learning_rate": 5.160123783779144e-05,
      "loss": 2.1045,
      "step": 186746
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.403237223625183,
      "learning_rate": 5.159894414281273e-05,
      "loss": 2.2635,
      "step": 186747
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.388709545135498,
      "learning_rate": 5.15966504940165e-05,
      "loss": 2.036,
      "step": 186748
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.3389015197753906,
      "learning_rate": 5.1594356891403096e-05,
      "loss": 1.9567,
      "step": 186749
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.3544962406158447,
      "learning_rate": 5.159206333497312e-05,
      "loss": 1.6811,
      "step": 186750
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.511301040649414,
      "learning_rate": 5.1589769824726735e-05,
      "loss": 2.0271,
      "step": 186751
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.4501588344573975,
      "learning_rate": 5.158747636066457e-05,
      "loss": 2.0077,
      "step": 186752
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.4538042545318604,
      "learning_rate": 5.15851829427869e-05,
      "loss": 1.932,
      "step": 186753
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.3536957502365112,
      "learning_rate": 5.158288957109431e-05,
      "loss": 2.1503,
      "step": 186754
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.4464443922042847,
      "learning_rate": 5.158059624558708e-05,
      "loss": 1.9874,
      "step": 186755
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.4969128370285034,
      "learning_rate": 5.157830296626587e-05,
      "loss": 2.0618,
      "step": 186756
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.3641176223754883,
      "learning_rate": 5.157600973313079e-05,
      "loss": 2.0598,
      "step": 186757
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.3489713668823242,
      "learning_rate": 5.1573716546182485e-05,
      "loss": 2.3759,
      "step": 186758
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.3484464883804321,
      "learning_rate": 5.157142340542124e-05,
      "loss": 1.897,
      "step": 186759
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.387772798538208,
      "learning_rate": 5.1569130310847616e-05,
      "loss": 2.0355,
      "step": 186760
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.4867321252822876,
      "learning_rate": 5.1566837262461944e-05,
      "loss": 2.0262,
      "step": 186761
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.3118557929992676,
      "learning_rate": 5.1564544260264796e-05,
      "loss": 1.9718,
      "step": 186762
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.3573932647705078,
      "learning_rate": 5.156225130425636e-05,
      "loss": 2.0838,
      "step": 186763
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.3014156818389893,
      "learning_rate": 5.155995839443725e-05,
      "loss": 2.0551,
      "step": 186764
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.302007794380188,
      "learning_rate": 5.155766553080779e-05,
      "loss": 2.075,
      "step": 186765
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.478888988494873,
      "learning_rate": 5.155537271336851e-05,
      "loss": 2.0127,
      "step": 186766
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.4184759855270386,
      "learning_rate": 5.155307994211969e-05,
      "loss": 2.2472,
      "step": 186767
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.3919835090637207,
      "learning_rate": 5.1550787217061916e-05,
      "loss": 1.9941,
      "step": 186768
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.344836711883545,
      "learning_rate": 5.154849453819556e-05,
      "loss": 2.274,
      "step": 186769
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.4747895002365112,
      "learning_rate": 5.154620190552102e-05,
      "loss": 2.0769,
      "step": 186770
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.3856878280639648,
      "learning_rate": 5.1543909319038637e-05,
      "loss": 2.152,
      "step": 186771
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.3195372819900513,
      "learning_rate": 5.1541616778749025e-05,
      "loss": 2.0314,
      "step": 186772
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.5647832155227661,
      "learning_rate": 5.153932428465244e-05,
      "loss": 2.1463,
      "step": 186773
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.3090862035751343,
      "learning_rate": 5.153703183674946e-05,
      "loss": 1.9564,
      "step": 186774
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.3835563659667969,
      "learning_rate": 5.153473943504044e-05,
      "loss": 2.1617,
      "step": 186775
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.4408015012741089,
      "learning_rate": 5.15324470795258e-05,
      "loss": 1.89,
      "step": 186776
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.371025562286377,
      "learning_rate": 5.1530154770205864e-05,
      "loss": 2.0423,
      "step": 186777
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.5665837526321411,
      "learning_rate": 5.152786250708128e-05,
      "loss": 2.0761,
      "step": 186778
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.428760290145874,
      "learning_rate": 5.152557029015225e-05,
      "loss": 1.9701,
      "step": 186779
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.3618378639221191,
      "learning_rate": 5.1523278119419396e-05,
      "loss": 2.0223,
      "step": 186780
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.3799264430999756,
      "learning_rate": 5.152098599488306e-05,
      "loss": 1.9689,
      "step": 186781
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.5542023181915283,
      "learning_rate": 5.151869391654366e-05,
      "loss": 2.2265,
      "step": 186782
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.4888845682144165,
      "learning_rate": 5.151640188440155e-05,
      "loss": 2.0114,
      "step": 186783
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.5558007955551147,
      "learning_rate": 5.1514109898457314e-05,
      "loss": 2.1655,
      "step": 186784
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.2810684442520142,
      "learning_rate": 5.151181795871122e-05,
      "loss": 1.9054,
      "step": 186785
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.3727185726165771,
      "learning_rate": 5.150952606516384e-05,
      "loss": 2.0547,
      "step": 186786
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.36147141456604,
      "learning_rate": 5.1507234217815505e-05,
      "loss": 1.8592,
      "step": 186787
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.3435834646224976,
      "learning_rate": 5.1504942416666685e-05,
      "loss": 1.9128,
      "step": 186788
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.5960251092910767,
      "learning_rate": 5.150265066171774e-05,
      "loss": 1.9294,
      "step": 186789
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.3565293550491333,
      "learning_rate": 5.150035895296917e-05,
      "loss": 1.9521,
      "step": 186790
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.3554409742355347,
      "learning_rate": 5.1498067290421344e-05,
      "loss": 2.2027,
      "step": 186791
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.260355830192566,
      "learning_rate": 5.1495775674074756e-05,
      "loss": 1.812,
      "step": 186792
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.383297085762024,
      "learning_rate": 5.149348410392982e-05,
      "loss": 1.879,
      "step": 186793
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.4908853769302368,
      "learning_rate": 5.1491192579986916e-05,
      "loss": 2.0121,
      "step": 186794
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.551862120628357,
      "learning_rate": 5.148890110224643e-05,
      "loss": 2.1133,
      "step": 186795
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.6003389358520508,
      "learning_rate": 5.148660967070891e-05,
      "loss": 2.0085,
      "step": 186796
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.4204703569412231,
      "learning_rate": 5.1484318285374636e-05,
      "loss": 2.0761,
      "step": 186797
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.4530211687088013,
      "learning_rate": 5.1482026946244206e-05,
      "loss": 2.1446,
      "step": 186798
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.6492459774017334,
      "learning_rate": 5.147973565331798e-05,
      "loss": 1.9324,
      "step": 186799
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.4331055879592896,
      "learning_rate": 5.147744440659632e-05,
      "loss": 2.1275,
      "step": 186800
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.3973292112350464,
      "learning_rate": 5.1475153206079677e-05,
      "loss": 2.0822,
      "step": 186801
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.408241629600525,
      "learning_rate": 5.1472862051768535e-05,
      "loss": 1.9689,
      "step": 186802
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.286804437637329,
      "learning_rate": 5.1470570943663204e-05,
      "loss": 2.0103,
      "step": 186803
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.3514660596847534,
      "learning_rate": 5.146827988176424e-05,
      "loss": 1.9096,
      "step": 186804
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.3571736812591553,
      "learning_rate": 5.146598886607205e-05,
      "loss": 1.9832,
      "step": 186805
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.3686437606811523,
      "learning_rate": 5.1463697896586995e-05,
      "loss": 1.9,
      "step": 186806
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.434037685394287,
      "learning_rate": 5.1461406973309484e-05,
      "loss": 1.9459,
      "step": 186807
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.4650623798370361,
      "learning_rate": 5.145911609624004e-05,
      "loss": 2.1727,
      "step": 186808
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.3821110725402832,
      "learning_rate": 5.1456825265378966e-05,
      "loss": 1.8727,
      "step": 186809
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.3792411088943481,
      "learning_rate": 5.145453448072686e-05,
      "loss": 2.1435,
      "step": 186810
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.3781558275222778,
      "learning_rate": 5.145224374228403e-05,
      "loss": 2.0997,
      "step": 186811
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.3143092393875122,
      "learning_rate": 5.1449953050050826e-05,
      "loss": 2.0239,
      "step": 186812
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.286584734916687,
      "learning_rate": 5.1447662404027867e-05,
      "loss": 1.9575,
      "step": 186813
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.4143459796905518,
      "learning_rate": 5.1445371804215474e-05,
      "loss": 1.8566,
      "step": 186814
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.4267512559890747,
      "learning_rate": 5.144308125061401e-05,
      "loss": 1.942,
      "step": 186815
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.280609369277954,
      "learning_rate": 5.144079074322403e-05,
      "loss": 2.0885,
      "step": 186816
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.4044469594955444,
      "learning_rate": 5.1438500282045904e-05,
      "loss": 1.958,
      "step": 186817
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.4831879138946533,
      "learning_rate": 5.143620986707998e-05,
      "loss": 2.0146,
      "step": 186818
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.4162989854812622,
      "learning_rate": 5.1433919498326824e-05,
      "loss": 2.1421,
      "step": 186819
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.4978899955749512,
      "learning_rate": 5.1431629175786804e-05,
      "loss": 2.0677,
      "step": 186820
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.2541592121124268,
      "learning_rate": 5.142933889946025e-05,
      "loss": 2.0691,
      "step": 186821
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.522212028503418,
      "learning_rate": 5.142704866934776e-05,
      "loss": 1.9765,
      "step": 186822
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.3386908769607544,
      "learning_rate": 5.142475848544967e-05,
      "loss": 2.0817,
      "step": 186823
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.4763720035552979,
      "learning_rate": 5.142246834776634e-05,
      "loss": 2.3427,
      "step": 186824
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.344783067703247,
      "learning_rate": 5.142017825629835e-05,
      "loss": 2.0929,
      "step": 186825
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.2782763242721558,
      "learning_rate": 5.141788821104595e-05,
      "loss": 2.1506,
      "step": 186826
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.4458779096603394,
      "learning_rate": 5.141559821200972e-05,
      "loss": 1.9901,
      "step": 186827
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.396803379058838,
      "learning_rate": 5.141330825919002e-05,
      "loss": 1.9779,
      "step": 186828
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.3217629194259644,
      "learning_rate": 5.141101835258732e-05,
      "loss": 1.858,
      "step": 186829
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.4128607511520386,
      "learning_rate": 5.140872849220191e-05,
      "loss": 1.9936,
      "step": 186830
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.390148639678955,
      "learning_rate": 5.140643867803437e-05,
      "loss": 1.9387,
      "step": 186831
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.5424110889434814,
      "learning_rate": 5.1404148910084985e-05,
      "loss": 1.6923,
      "step": 186832
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.4769353866577148,
      "learning_rate": 5.1401859188354365e-05,
      "loss": 2.1525,
      "step": 186833
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.2875277996063232,
      "learning_rate": 5.139956951284274e-05,
      "loss": 2.0635,
      "step": 186834
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.3151880502700806,
      "learning_rate": 5.139727988355071e-05,
      "loss": 2.1427,
      "step": 186835
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.357435941696167,
      "learning_rate": 5.1394990300478613e-05,
      "loss": 2.2198,
      "step": 186836
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.2992687225341797,
      "learning_rate": 5.1392700763626873e-05,
      "loss": 1.8861,
      "step": 186837
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.4304373264312744,
      "learning_rate": 5.139041127299586e-05,
      "loss": 1.9907,
      "step": 186838
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.1922825574874878,
      "learning_rate": 5.1388121828586136e-05,
      "loss": 1.9121,
      "step": 186839
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.3118596076965332,
      "learning_rate": 5.138583243039798e-05,
      "loss": 2.083,
      "step": 186840
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.3290131092071533,
      "learning_rate": 5.138354307843194e-05,
      "loss": 1.8356,
      "step": 186841
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.4210759401321411,
      "learning_rate": 5.138125377268844e-05,
      "loss": 2.1087,
      "step": 186842
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.3318464756011963,
      "learning_rate": 5.137896451316782e-05,
      "loss": 1.9235,
      "step": 186843
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.5255728960037231,
      "learning_rate": 5.1376675299870464e-05,
      "loss": 2.3175,
      "step": 186844
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.236009120941162,
      "learning_rate": 5.1374386132796994e-05,
      "loss": 2.0561,
      "step": 186845
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.3691095113754272,
      "learning_rate": 5.137209701194759e-05,
      "loss": 2.071,
      "step": 186846
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.443662166595459,
      "learning_rate": 5.1369807937322936e-05,
      "loss": 1.9601,
      "step": 186847
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.5000009536743164,
      "learning_rate": 5.1367518908923275e-05,
      "loss": 1.94,
      "step": 186848
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.3643989562988281,
      "learning_rate": 5.136522992674914e-05,
      "loss": 2.0815,
      "step": 186849
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.32047438621521,
      "learning_rate": 5.1362940990800797e-05,
      "loss": 1.9953,
      "step": 186850
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.3107805252075195,
      "learning_rate": 5.136065210107883e-05,
      "loss": 1.8004,
      "step": 186851
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.6163409948349,
      "learning_rate": 5.135836325758357e-05,
      "loss": 1.9897,
      "step": 186852
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.454397439956665,
      "learning_rate": 5.135607446031555e-05,
      "loss": 1.8349,
      "step": 186853
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.3855304718017578,
      "learning_rate": 5.135378570927513e-05,
      "loss": 2.0106,
      "step": 186854
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.7062302827835083,
      "learning_rate": 5.135149700446273e-05,
      "loss": 2.0896,
      "step": 186855
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.3673224449157715,
      "learning_rate": 5.134920834587871e-05,
      "loss": 1.9578,
      "step": 186856
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.4568475484848022,
      "learning_rate": 5.134691973352365e-05,
      "loss": 1.8606,
      "step": 186857
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.4408462047576904,
      "learning_rate": 5.1344631167397774e-05,
      "loss": 2.1067,
      "step": 186858
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.605216145515442,
      "learning_rate": 5.134234264750175e-05,
      "loss": 2.1989,
      "step": 186859
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.4250444173812866,
      "learning_rate": 5.1340054173835854e-05,
      "loss": 1.9626,
      "step": 186860
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.3291046619415283,
      "learning_rate": 5.133776574640054e-05,
      "loss": 1.9533,
      "step": 186861
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.410404920578003,
      "learning_rate": 5.133547736519614e-05,
      "loss": 1.9825,
      "step": 186862
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.3598014116287231,
      "learning_rate": 5.133318903022326e-05,
      "loss": 2.0282,
      "step": 186863
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.332787275314331,
      "learning_rate": 5.133090074148217e-05,
      "loss": 2.1294,
      "step": 186864
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.4304810762405396,
      "learning_rate": 5.132861249897343e-05,
      "loss": 2.1242,
      "step": 186865
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.546195387840271,
      "learning_rate": 5.132632430269738e-05,
      "loss": 1.9975,
      "step": 186866
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.3747899532318115,
      "learning_rate": 5.132403615265447e-05,
      "loss": 2.0819,
      "step": 186867
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.3382346630096436,
      "learning_rate": 5.132174804884505e-05,
      "loss": 2.1523,
      "step": 186868
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.4369051456451416,
      "learning_rate": 5.131945999126967e-05,
      "loss": 2.0021,
      "step": 186869
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.5522010326385498,
      "learning_rate": 5.131717197992862e-05,
      "loss": 1.9981,
      "step": 186870
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.4218177795410156,
      "learning_rate": 5.1314884014822475e-05,
      "loss": 2.1292,
      "step": 186871
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.3188986778259277,
      "learning_rate": 5.131259609595162e-05,
      "loss": 1.9726,
      "step": 186872
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.3223509788513184,
      "learning_rate": 5.131030822331641e-05,
      "loss": 2.1463,
      "step": 186873
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.371766448020935,
      "learning_rate": 5.1308020396917246e-05,
      "loss": 1.9326,
      "step": 186874
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.2387877702713013,
      "learning_rate": 5.13057326167547e-05,
      "loss": 1.8125,
      "step": 186875
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.3082575798034668,
      "learning_rate": 5.1303444882829e-05,
      "loss": 1.8875,
      "step": 186876
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.3677603006362915,
      "learning_rate": 5.130115719514082e-05,
      "loss": 2.1805,
      "step": 186877
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.405129313468933,
      "learning_rate": 5.1298869553690414e-05,
      "loss": 2.2532,
      "step": 186878
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.4383713006973267,
      "learning_rate": 5.1296581958478225e-05,
      "loss": 1.9873,
      "step": 186879
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.3247380256652832,
      "learning_rate": 5.129429440950464e-05,
      "loss": 2.0737,
      "step": 186880
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.352778673171997,
      "learning_rate": 5.129200690677021e-05,
      "loss": 1.8725,
      "step": 186881
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.5483043193817139,
      "learning_rate": 5.1289719450275223e-05,
      "loss": 1.9612,
      "step": 186882
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.6406854391098022,
      "learning_rate": 5.128743204002025e-05,
      "loss": 2.0049,
      "step": 186883
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.2986129522323608,
      "learning_rate": 5.1285144676005653e-05,
      "loss": 1.9453,
      "step": 186884
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.2748041152954102,
      "learning_rate": 5.1282857358231835e-05,
      "loss": 2.1852,
      "step": 186885
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.3741952180862427,
      "learning_rate": 5.1280570086699124e-05,
      "loss": 1.9785,
      "step": 186886
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.3573088645935059,
      "learning_rate": 5.127828286140816e-05,
      "loss": 1.8225,
      "step": 186887
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.6596177816390991,
      "learning_rate": 5.127599568235917e-05,
      "loss": 1.9874,
      "step": 186888
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.4176971912384033,
      "learning_rate": 5.1273708549552725e-05,
      "loss": 2.1121,
      "step": 186889
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.2937356233596802,
      "learning_rate": 5.1271421462989194e-05,
      "loss": 1.8634,
      "step": 186890
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.3770227432250977,
      "learning_rate": 5.1269134422669034e-05,
      "loss": 2.0142,
      "step": 186891
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.326505422592163,
      "learning_rate": 5.126684742859255e-05,
      "loss": 2.0461,
      "step": 186892
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.4906654357910156,
      "learning_rate": 5.126456048076031e-05,
      "loss": 2.0497,
      "step": 186893
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.3522393703460693,
      "learning_rate": 5.1262273579172606e-05,
      "loss": 2.0032,
      "step": 186894
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.3047146797180176,
      "learning_rate": 5.125998672383005e-05,
      "loss": 2.0456,
      "step": 186895
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.4225982427597046,
      "learning_rate": 5.125769991473293e-05,
      "loss": 1.92,
      "step": 186896
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.2559106349945068,
      "learning_rate": 5.125541315188162e-05,
      "loss": 1.8724,
      "step": 186897
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.395238995552063,
      "learning_rate": 5.1253126435276716e-05,
      "loss": 1.9785,
      "step": 186898
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.4847315549850464,
      "learning_rate": 5.125083976491855e-05,
      "loss": 2.049,
      "step": 186899
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.424548864364624,
      "learning_rate": 5.124855314080746e-05,
      "loss": 2.1307,
      "step": 186900
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.3722128868103027,
      "learning_rate": 5.124626656294404e-05,
      "loss": 2.0013,
      "step": 186901
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.4908581972122192,
      "learning_rate": 5.124398003132859e-05,
      "loss": 2.0093,
      "step": 186902
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.48800790309906,
      "learning_rate": 5.1241693545961614e-05,
      "loss": 2.108,
      "step": 186903
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.313975214958191,
      "learning_rate": 5.123940710684351e-05,
      "loss": 2.1138,
      "step": 186904
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.3885692358016968,
      "learning_rate": 5.123712071397471e-05,
      "loss": 2.0869,
      "step": 186905
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.422561526298523,
      "learning_rate": 5.1234834367355546e-05,
      "loss": 1.9672,
      "step": 186906
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.4132683277130127,
      "learning_rate": 5.123254806698659e-05,
      "loss": 1.97,
      "step": 186907
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.6161140203475952,
      "learning_rate": 5.1230261812868134e-05,
      "loss": 1.8699,
      "step": 186908
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.5748978853225708,
      "learning_rate": 5.1227975605000746e-05,
      "loss": 2.0574,
      "step": 186909
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.2599767446517944,
      "learning_rate": 5.1225689443384766e-05,
      "loss": 1.9281,
      "step": 186910
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.4390482902526855,
      "learning_rate": 5.122340332802055e-05,
      "loss": 2.0514,
      "step": 186911
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.3526637554168701,
      "learning_rate": 5.1221117258908675e-05,
      "loss": 2.0567,
      "step": 186912
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.605674147605896,
      "learning_rate": 5.1218831236049464e-05,
      "loss": 1.9392,
      "step": 186913
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.5906074047088623,
      "learning_rate": 5.121654525944332e-05,
      "loss": 1.9017,
      "step": 186914
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.6265041828155518,
      "learning_rate": 5.121425932909081e-05,
      "loss": 2.1786,
      "step": 186915
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.4577080011367798,
      "learning_rate": 5.121197344499224e-05,
      "loss": 2.179,
      "step": 186916
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.2741963863372803,
      "learning_rate": 5.1209687607147995e-05,
      "loss": 1.9589,
      "step": 186917
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.3643215894699097,
      "learning_rate": 5.120740181555862e-05,
      "loss": 1.8812,
      "step": 186918
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.629240870475769,
      "learning_rate": 5.120511607022452e-05,
      "loss": 2.0186,
      "step": 186919
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.3499524593353271,
      "learning_rate": 5.120283037114598e-05,
      "loss": 2.0161,
      "step": 186920
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.5196731090545654,
      "learning_rate": 5.1200544718323634e-05,
      "loss": 1.8408,
      "step": 186921
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.498138189315796,
      "learning_rate": 5.1198259111757786e-05,
      "loss": 1.9427,
      "step": 186922
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.3235212564468384,
      "learning_rate": 5.1195973551448774e-05,
      "loss": 2.0321,
      "step": 186923
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.3850994110107422,
      "learning_rate": 5.119368803739726e-05,
      "loss": 2.139,
      "step": 186924
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.4909420013427734,
      "learning_rate": 5.119140256960341e-05,
      "loss": 2.0667,
      "step": 186925
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.4185327291488647,
      "learning_rate": 5.118911714806789e-05,
      "loss": 2.1266,
      "step": 186926
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.3153884410858154,
      "learning_rate": 5.1186831772791005e-05,
      "loss": 2.012,
      "step": 186927
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.598536491394043,
      "learning_rate": 5.118454644377314e-05,
      "loss": 2.0356,
      "step": 186928
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.4557677507400513,
      "learning_rate": 5.118226116101475e-05,
      "loss": 2.1949,
      "step": 186929
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.4884635210037231,
      "learning_rate": 5.1179975924516316e-05,
      "loss": 1.9447,
      "step": 186930
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.3989661931991577,
      "learning_rate": 5.117769073427814e-05,
      "loss": 2.0623,
      "step": 186931
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.3347575664520264,
      "learning_rate": 5.117540559030083e-05,
      "loss": 2.0459,
      "step": 186932
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.4670089483261108,
      "learning_rate": 5.1173120492584686e-05,
      "loss": 2.0077,
      "step": 186933
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.6621830463409424,
      "learning_rate": 5.117083544113016e-05,
      "loss": 1.9055,
      "step": 186934
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.3067132234573364,
      "learning_rate": 5.116855043593761e-05,
      "loss": 1.938,
      "step": 186935
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.3865948915481567,
      "learning_rate": 5.1166265477007576e-05,
      "loss": 1.9807,
      "step": 186936
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.3642046451568604,
      "learning_rate": 5.116398056434037e-05,
      "loss": 2.07,
      "step": 186937
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.3062537908554077,
      "learning_rate": 5.1161695697936556e-05,
      "loss": 2.1053,
      "step": 186938
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.355352520942688,
      "learning_rate": 5.115941087779647e-05,
      "loss": 1.9133,
      "step": 186939
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.5347318649291992,
      "learning_rate": 5.1157126103920544e-05,
      "loss": 1.9225,
      "step": 186940
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.6029163599014282,
      "learning_rate": 5.1154841376309145e-05,
      "loss": 2.1589,
      "step": 186941
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.474352478981018,
      "learning_rate": 5.1152556694962834e-05,
      "loss": 1.9369,
      "step": 186942
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.4901564121246338,
      "learning_rate": 5.115027205988188e-05,
      "loss": 1.9086,
      "step": 186943
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.422040343284607,
      "learning_rate": 5.1147987471066853e-05,
      "loss": 2.1083,
      "step": 186944
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.3330464363098145,
      "learning_rate": 5.114570292851812e-05,
      "loss": 2.1062,
      "step": 186945
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.3372029066085815,
      "learning_rate": 5.114341843223611e-05,
      "loss": 2.0784,
      "step": 186946
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.4543757438659668,
      "learning_rate": 5.114113398222115e-05,
      "loss": 2.1433,
      "step": 186947
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.5336084365844727,
      "learning_rate": 5.113884957847382e-05,
      "loss": 1.91,
      "step": 186948
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.3760050535202026,
      "learning_rate": 5.1136565220994405e-05,
      "loss": 2.0186,
      "step": 186949
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.362363576889038,
      "learning_rate": 5.113428090978349e-05,
      "loss": 1.9553,
      "step": 186950
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.3988817930221558,
      "learning_rate": 5.113199664484139e-05,
      "loss": 2.1138,
      "step": 186951
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.635636806488037,
      "learning_rate": 5.1129712426168546e-05,
      "loss": 2.3112,
      "step": 186952
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.3581589460372925,
      "learning_rate": 5.112742825376532e-05,
      "loss": 2.1489,
      "step": 186953
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.46414053440094,
      "learning_rate": 5.112514412763229e-05,
      "loss": 1.9788,
      "step": 186954
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.3778408765792847,
      "learning_rate": 5.112286004776971e-05,
      "loss": 2.0627,
      "step": 186955
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.4271416664123535,
      "learning_rate": 5.1120576014178146e-05,
      "loss": 1.9053,
      "step": 186956
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.4078426361083984,
      "learning_rate": 5.111829202685798e-05,
      "loss": 1.9833,
      "step": 186957
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.4478459358215332,
      "learning_rate": 5.1116008085809627e-05,
      "loss": 2.0154,
      "step": 186958
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.399268627166748,
      "learning_rate": 5.1113724191033425e-05,
      "loss": 1.9077,
      "step": 186959
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.4100373983383179,
      "learning_rate": 5.111144034252994e-05,
      "loss": 1.93,
      "step": 186960
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.3262966871261597,
      "learning_rate": 5.1109156540299476e-05,
      "loss": 1.9184,
      "step": 186961
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.396077036857605,
      "learning_rate": 5.11068727843426e-05,
      "loss": 1.9802,
      "step": 186962
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.2113755941390991,
      "learning_rate": 5.110458907465964e-05,
      "loss": 1.9654,
      "step": 186963
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.373537302017212,
      "learning_rate": 5.1102305411251065e-05,
      "loss": 1.8892,
      "step": 186964
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.469707727432251,
      "learning_rate": 5.110002179411717e-05,
      "loss": 2.0925,
      "step": 186965
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.387673020362854,
      "learning_rate": 5.109773822325857e-05,
      "loss": 2.0858,
      "step": 186966
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.3473834991455078,
      "learning_rate": 5.109545469867548e-05,
      "loss": 2.1066,
      "step": 186967
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.3570464849472046,
      "learning_rate": 5.1093171220368565e-05,
      "loss": 2.033,
      "step": 186968
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.539657473564148,
      "learning_rate": 5.109088778833804e-05,
      "loss": 2.0466,
      "step": 186969
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.4813560247421265,
      "learning_rate": 5.108860440258456e-05,
      "loss": 2.1364,
      "step": 186970
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.495753288269043,
      "learning_rate": 5.108632106310827e-05,
      "loss": 1.8794,
      "step": 186971
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.5435816049575806,
      "learning_rate": 5.108403776990979e-05,
      "loss": 2.3405,
      "step": 186972
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.383975625038147,
      "learning_rate": 5.1081754522989424e-05,
      "loss": 2.0719,
      "step": 186973
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.2553409337997437,
      "learning_rate": 5.107947132234771e-05,
      "loss": 1.9802,
      "step": 186974
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.3149605989456177,
      "learning_rate": 5.107718816798497e-05,
      "loss": 1.9219,
      "step": 186975
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.505831003189087,
      "learning_rate": 5.107490505990182e-05,
      "loss": 2.0374,
      "step": 186976
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.4798623323440552,
      "learning_rate": 5.107262199809841e-05,
      "loss": 1.9285,
      "step": 186977
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.613182544708252,
      "learning_rate": 5.107033898257538e-05,
      "loss": 2.0362,
      "step": 186978
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.661820888519287,
      "learning_rate": 5.1068056013332955e-05,
      "loss": 2.2577,
      "step": 186979
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.3451224565505981,
      "learning_rate": 5.106577309037179e-05,
      "loss": 2.0354,
      "step": 186980
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.5185858011245728,
      "learning_rate": 5.106349021369209e-05,
      "loss": 1.8459,
      "step": 186981
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.6291112899780273,
      "learning_rate": 5.1061207383294576e-05,
      "loss": 1.9309,
      "step": 186982
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.3759993314743042,
      "learning_rate": 5.105892459917931e-05,
      "loss": 2.0195,
      "step": 186983
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.4442073106765747,
      "learning_rate": 5.105664186134695e-05,
      "loss": 2.056,
      "step": 186984
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.3744087219238281,
      "learning_rate": 5.10543591697978e-05,
      "loss": 1.9797,
      "step": 186985
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.4116050004959106,
      "learning_rate": 5.105207652453242e-05,
      "loss": 2.0476,
      "step": 186986
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.391972303390503,
      "learning_rate": 5.1049793925551093e-05,
      "loss": 1.9646,
      "step": 186987
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.5790623426437378,
      "learning_rate": 5.1047511372854375e-05,
      "loss": 2.0723,
      "step": 186988
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.3950424194335938,
      "learning_rate": 5.10452288664426e-05,
      "loss": 2.1729,
      "step": 186989
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.3951396942138672,
      "learning_rate": 5.104294640631627e-05,
      "loss": 1.7652,
      "step": 186990
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.404352068901062,
      "learning_rate": 5.104066399247564e-05,
      "loss": 2.0308,
      "step": 186991
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.4038574695587158,
      "learning_rate": 5.103838162492132e-05,
      "loss": 2.0433,
      "step": 186992
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.4065992832183838,
      "learning_rate": 5.103609930365361e-05,
      "loss": 1.8368,
      "step": 186993
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.3498122692108154,
      "learning_rate": 5.103381702867304e-05,
      "loss": 1.9026,
      "step": 186994
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.3890767097473145,
      "learning_rate": 5.103153479998001e-05,
      "loss": 1.9824,
      "step": 186995
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.353705644607544,
      "learning_rate": 5.1029252617574824e-05,
      "loss": 2.1941,
      "step": 186996
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.3299493789672852,
      "learning_rate": 5.1026970481458104e-05,
      "loss": 2.1126,
      "step": 186997
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.6866838932037354,
      "learning_rate": 5.102468839163013e-05,
      "loss": 2.1878,
      "step": 186998
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.430567979812622,
      "learning_rate": 5.1022406348091325e-05,
      "loss": 1.9796,
      "step": 186999
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.293877363204956,
      "learning_rate": 5.1020124350842194e-05,
      "loss": 1.9917,
      "step": 187000
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.3340938091278076,
      "learning_rate": 5.101784239988317e-05,
      "loss": 1.8672,
      "step": 187001
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.256674885749817,
      "learning_rate": 5.1015560495214516e-05,
      "loss": 1.7728,
      "step": 187002
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.3781601190567017,
      "learning_rate": 5.1013278636836864e-05,
      "loss": 2.0003,
      "step": 187003
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.443786382675171,
      "learning_rate": 5.101099682475052e-05,
      "loss": 2.0276,
      "step": 187004
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.2517255544662476,
      "learning_rate": 5.100871505895587e-05,
      "loss": 1.9166,
      "step": 187005
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.415295124053955,
      "learning_rate": 5.1006433339453465e-05,
      "loss": 2.117,
      "step": 187006
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.3903330564498901,
      "learning_rate": 5.10041516662437e-05,
      "loss": 2.1962,
      "step": 187007
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.4964691400527954,
      "learning_rate": 5.100187003932687e-05,
      "loss": 2.2379,
      "step": 187008
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.333775520324707,
      "learning_rate": 5.0999588458703536e-05,
      "loss": 1.936,
      "step": 187009
    },
    {
      "epoch": 2.43,
      "grad_norm": 1.3766369819641113,
      "learning_rate": 5.099730692437404e-05,
      "loss": 1.9321,
      "step": 187010
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.285272240638733,
      "learning_rate": 5.099502543633891e-05,
      "loss": 1.8737,
      "step": 187011
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.4077246189117432,
      "learning_rate": 5.099274399459848e-05,
      "loss": 2.0177,
      "step": 187012
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.328148365020752,
      "learning_rate": 5.099046259915322e-05,
      "loss": 1.9627,
      "step": 187013
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.467543601989746,
      "learning_rate": 5.098818125000347e-05,
      "loss": 2.0776,
      "step": 187014
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.3223931789398193,
      "learning_rate": 5.0985899947149775e-05,
      "loss": 2.087,
      "step": 187015
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.6751811504364014,
      "learning_rate": 5.0983618690592454e-05,
      "loss": 2.1347,
      "step": 187016
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.3769311904907227,
      "learning_rate": 5.098133748033203e-05,
      "loss": 2.0023,
      "step": 187017
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.388744592666626,
      "learning_rate": 5.0979056316368873e-05,
      "loss": 2.014,
      "step": 187018
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.2998563051223755,
      "learning_rate": 5.097677519870345e-05,
      "loss": 2.034,
      "step": 187019
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.5008026361465454,
      "learning_rate": 5.0974494127336016e-05,
      "loss": 1.7874,
      "step": 187020
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.4922184944152832,
      "learning_rate": 5.097221310226722e-05,
      "loss": 2.0023,
      "step": 187021
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.541938304901123,
      "learning_rate": 5.0969932123497325e-05,
      "loss": 1.9742,
      "step": 187022
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.3159904479980469,
      "learning_rate": 5.096765119102689e-05,
      "loss": 1.9402,
      "step": 187023
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.5719448328018188,
      "learning_rate": 5.0965370304856254e-05,
      "loss": 1.7949,
      "step": 187024
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.3497453927993774,
      "learning_rate": 5.096308946498588e-05,
      "loss": 1.804,
      "step": 187025
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.3314296007156372,
      "learning_rate": 5.096080867141608e-05,
      "loss": 1.9144,
      "step": 187026
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.4159564971923828,
      "learning_rate": 5.095852792414743e-05,
      "loss": 1.9821,
      "step": 187027
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.355406403541565,
      "learning_rate": 5.0956247223180246e-05,
      "loss": 2.1396,
      "step": 187028
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.5438100099563599,
      "learning_rate": 5.095396656851506e-05,
      "loss": 2.0748,
      "step": 187029
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.4534810781478882,
      "learning_rate": 5.095168596015223e-05,
      "loss": 1.6994,
      "step": 187030
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.5269256830215454,
      "learning_rate": 5.094940539809217e-05,
      "loss": 2.0038,
      "step": 187031
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.2777924537658691,
      "learning_rate": 5.0947124882335266e-05,
      "loss": 2.0552,
      "step": 187032
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.4786295890808105,
      "learning_rate": 5.094484441288206e-05,
      "loss": 2.2081,
      "step": 187033
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.6260490417480469,
      "learning_rate": 5.094256398973281e-05,
      "loss": 2.02,
      "step": 187034
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.6745843887329102,
      "learning_rate": 5.094028361288812e-05,
      "loss": 2.0409,
      "step": 187035
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.5043150186538696,
      "learning_rate": 5.0938003282348296e-05,
      "loss": 2.1993,
      "step": 187036
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.5980134010314941,
      "learning_rate": 5.0935722998113926e-05,
      "loss": 1.8578,
      "step": 187037
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.360432505607605,
      "learning_rate": 5.093344276018515e-05,
      "loss": 1.9593,
      "step": 187038
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.573897123336792,
      "learning_rate": 5.093116256856263e-05,
      "loss": 2.0383,
      "step": 187039
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.3688524961471558,
      "learning_rate": 5.0928882423246606e-05,
      "loss": 1.9469,
      "step": 187040
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.590289831161499,
      "learning_rate": 5.092660232423771e-05,
      "loss": 1.8647,
      "step": 187041
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.6986732482910156,
      "learning_rate": 5.09243222715362e-05,
      "loss": 2.0881,
      "step": 187042
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.4534958600997925,
      "learning_rate": 5.0922042265142684e-05,
      "loss": 1.8602,
      "step": 187043
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.4147101640701294,
      "learning_rate": 5.091976230505732e-05,
      "loss": 2.0892,
      "step": 187044
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.5508614778518677,
      "learning_rate": 5.0917482391280716e-05,
      "loss": 1.9903,
      "step": 187045
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.3132644891738892,
      "learning_rate": 5.0915202523813234e-05,
      "loss": 1.9741,
      "step": 187046
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.4363758563995361,
      "learning_rate": 5.0912922702655346e-05,
      "loss": 1.8357,
      "step": 187047
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.3482801914215088,
      "learning_rate": 5.0910642927807413e-05,
      "loss": 2.0495,
      "step": 187048
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.3539446592330933,
      "learning_rate": 5.090836319927003e-05,
      "loss": 1.9908,
      "step": 187049
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.3055434226989746,
      "learning_rate": 5.090608351704331e-05,
      "loss": 1.9644,
      "step": 187050
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.4405760765075684,
      "learning_rate": 5.0903803881127935e-05,
      "loss": 2.089,
      "step": 187051
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.3153800964355469,
      "learning_rate": 5.090152429152419e-05,
      "loss": 1.9491,
      "step": 187052
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.3103610277175903,
      "learning_rate": 5.089924474823263e-05,
      "loss": 2.2346,
      "step": 187053
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.420276165008545,
      "learning_rate": 5.089696525125352e-05,
      "loss": 2.0676,
      "step": 187054
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.3122931718826294,
      "learning_rate": 5.08946858005875e-05,
      "loss": 1.9354,
      "step": 187055
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.5251401662826538,
      "learning_rate": 5.089240639623473e-05,
      "loss": 2.0591,
      "step": 187056
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.4362144470214844,
      "learning_rate": 5.0890127038195815e-05,
      "loss": 1.9498,
      "step": 187057
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.304714322090149,
      "learning_rate": 5.088784772647109e-05,
      "loss": 1.9263,
      "step": 187058
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.3469079732894897,
      "learning_rate": 5.0885568461061045e-05,
      "loss": 1.9492,
      "step": 187059
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.3510465621948242,
      "learning_rate": 5.088328924196605e-05,
      "loss": 1.9826,
      "step": 187060
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.3612996339797974,
      "learning_rate": 5.088101006918668e-05,
      "loss": 2.0335,
      "step": 187061
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.452689528465271,
      "learning_rate": 5.087873094272306e-05,
      "loss": 2.2241,
      "step": 187062
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.3229572772979736,
      "learning_rate": 5.0876451862575886e-05,
      "loss": 1.9262,
      "step": 187063
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.3686443567276,
      "learning_rate": 5.087417282874543e-05,
      "loss": 1.8443,
      "step": 187064
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.4200224876403809,
      "learning_rate": 5.087189384123219e-05,
      "loss": 2.1804,
      "step": 187065
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.3937245607376099,
      "learning_rate": 5.086961490003654e-05,
      "loss": 2.1852,
      "step": 187066
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.4440128803253174,
      "learning_rate": 5.0867336005159066e-05,
      "loss": 2.0725,
      "step": 187067
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.3510282039642334,
      "learning_rate": 5.0865057156599876e-05,
      "loss": 2.1807,
      "step": 187068
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.4475548267364502,
      "learning_rate": 5.08627783543597e-05,
      "loss": 2.0043,
      "step": 187069
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.3716437816619873,
      "learning_rate": 5.086049959843874e-05,
      "loss": 1.9585,
      "step": 187070
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.5017788410186768,
      "learning_rate": 5.085822088883756e-05,
      "loss": 1.9904,
      "step": 187071
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.5199081897735596,
      "learning_rate": 5.08559422255565e-05,
      "loss": 1.9415,
      "step": 187072
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.333430528640747,
      "learning_rate": 5.0853663608596116e-05,
      "loss": 1.9114,
      "step": 187073
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.6168383359909058,
      "learning_rate": 5.085138503795672e-05,
      "loss": 1.7856,
      "step": 187074
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.4544475078582764,
      "learning_rate": 5.084910651363873e-05,
      "loss": 1.7964,
      "step": 187075
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.3624590635299683,
      "learning_rate": 5.084682803564255e-05,
      "loss": 2.0953,
      "step": 187076
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.2986141443252563,
      "learning_rate": 5.084454960396869e-05,
      "loss": 1.7826,
      "step": 187077
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.2876981496810913,
      "learning_rate": 5.08422712186175e-05,
      "loss": 1.9599,
      "step": 187078
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.3421002626419067,
      "learning_rate": 5.0839992879589464e-05,
      "loss": 1.9823,
      "step": 187079
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.653877854347229,
      "learning_rate": 5.0837714586885e-05,
      "loss": 1.8086,
      "step": 187080
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.4815843105316162,
      "learning_rate": 5.0835436340504524e-05,
      "loss": 2.1535,
      "step": 187081
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.5240532159805298,
      "learning_rate": 5.083315814044835e-05,
      "loss": 1.7785,
      "step": 187082
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.2496472597122192,
      "learning_rate": 5.083087998671705e-05,
      "loss": 2.1839,
      "step": 187083
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.247658133506775,
      "learning_rate": 5.082860187931097e-05,
      "loss": 2.1663,
      "step": 187084
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.3767404556274414,
      "learning_rate": 5.0826323818230596e-05,
      "loss": 1.9762,
      "step": 187085
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.4408485889434814,
      "learning_rate": 5.08240458034763e-05,
      "loss": 1.9717,
      "step": 187086
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.4865083694458008,
      "learning_rate": 5.082176783504848e-05,
      "loss": 1.8515,
      "step": 187087
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.460045337677002,
      "learning_rate": 5.081948991294766e-05,
      "loss": 1.8192,
      "step": 187088
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.407418131828308,
      "learning_rate": 5.0817212037174194e-05,
      "loss": 2.0145,
      "step": 187089
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.4343737363815308,
      "learning_rate": 5.0814934207728433e-05,
      "loss": 1.9382,
      "step": 187090
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.4699139595031738,
      "learning_rate": 5.0812656424610944e-05,
      "loss": 2.0462,
      "step": 187091
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.5698912143707275,
      "learning_rate": 5.08103786878221e-05,
      "loss": 2.021,
      "step": 187092
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.7566920518875122,
      "learning_rate": 5.080810099736225e-05,
      "loss": 2.083,
      "step": 187093
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.3385276794433594,
      "learning_rate": 5.080582335323192e-05,
      "loss": 2.124,
      "step": 187094
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.438330054283142,
      "learning_rate": 5.080354575543143e-05,
      "loss": 2.1281,
      "step": 187095
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.3852945566177368,
      "learning_rate": 5.080126820396134e-05,
      "loss": 1.9711,
      "step": 187096
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.3487730026245117,
      "learning_rate": 5.079899069882203e-05,
      "loss": 1.9256,
      "step": 187097
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.3266111612319946,
      "learning_rate": 5.0796713240013857e-05,
      "loss": 2.0226,
      "step": 187098
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.5258454084396362,
      "learning_rate": 5.0794435827537215e-05,
      "loss": 1.9016,
      "step": 187099
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.3981833457946777,
      "learning_rate": 5.079215846139265e-05,
      "loss": 1.9618,
      "step": 187100
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.5424561500549316,
      "learning_rate": 5.078988114158046e-05,
      "loss": 1.9158,
      "step": 187101
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.2370309829711914,
      "learning_rate": 5.0787603868101235e-05,
      "loss": 1.8724,
      "step": 187102
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.4699736833572388,
      "learning_rate": 5.078532664095528e-05,
      "loss": 2.0267,
      "step": 187103
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.2973740100860596,
      "learning_rate": 5.078304946014303e-05,
      "loss": 1.8353,
      "step": 187104
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.3377851247787476,
      "learning_rate": 5.078077232566485e-05,
      "loss": 1.951,
      "step": 187105
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.2883042097091675,
      "learning_rate": 5.0778495237521274e-05,
      "loss": 2.029,
      "step": 187106
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.3998606204986572,
      "learning_rate": 5.0776218195712634e-05,
      "loss": 1.9113,
      "step": 187107
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.3404825925827026,
      "learning_rate": 5.07739412002395e-05,
      "loss": 1.7846,
      "step": 187108
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.3162014484405518,
      "learning_rate": 5.0771664251102065e-05,
      "loss": 1.916,
      "step": 187109
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.3819342851638794,
      "learning_rate": 5.076938734830106e-05,
      "loss": 1.8962,
      "step": 187110
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.373808741569519,
      "learning_rate": 5.076711049183659e-05,
      "loss": 2.2496,
      "step": 187111
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.255725383758545,
      "learning_rate": 5.0764833681709296e-05,
      "loss": 1.9021,
      "step": 187112
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.6110020875930786,
      "learning_rate": 5.076255691791943e-05,
      "loss": 2.092,
      "step": 187113
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.3906913995742798,
      "learning_rate": 5.076028020046756e-05,
      "loss": 1.9167,
      "step": 187114
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.4107714891433716,
      "learning_rate": 5.075800352935403e-05,
      "loss": 2.1571,
      "step": 187115
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.3331303596496582,
      "learning_rate": 5.075572690457943e-05,
      "loss": 2.2716,
      "step": 187116
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.3439620733261108,
      "learning_rate": 5.0753450326143895e-05,
      "loss": 1.8952,
      "step": 187117
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.3883707523345947,
      "learning_rate": 5.0751173794048064e-05,
      "loss": 2.0086,
      "step": 187118
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.3875362873077393,
      "learning_rate": 5.074889730829223e-05,
      "loss": 1.8898,
      "step": 187119
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.5611016750335693,
      "learning_rate": 5.074662086887695e-05,
      "loss": 1.8609,
      "step": 187120
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.313396692276001,
      "learning_rate": 5.074434447580251e-05,
      "loss": 2.0272,
      "step": 187121
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.2910329103469849,
      "learning_rate": 5.0742068129069534e-05,
      "loss": 1.8754,
      "step": 187122
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.3058850765228271,
      "learning_rate": 5.073979182867819e-05,
      "loss": 1.9525,
      "step": 187123
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.3690019845962524,
      "learning_rate": 5.073751557462907e-05,
      "loss": 2.0258,
      "step": 187124
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.3354377746582031,
      "learning_rate": 5.0735239366922484e-05,
      "loss": 2.1425,
      "step": 187125
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.4843523502349854,
      "learning_rate": 5.0732963205558995e-05,
      "loss": 2.0392,
      "step": 187126
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.3147467374801636,
      "learning_rate": 5.07306870905389e-05,
      "loss": 1.8296,
      "step": 187127
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.454216718673706,
      "learning_rate": 5.0728411021862805e-05,
      "loss": 1.9352,
      "step": 187128
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.313873529434204,
      "learning_rate": 5.0726134999530863e-05,
      "loss": 1.9712,
      "step": 187129
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.4028853178024292,
      "learning_rate": 5.072385902354372e-05,
      "loss": 2.1192,
      "step": 187130
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.3556972742080688,
      "learning_rate": 5.072158309390164e-05,
      "loss": 1.8962,
      "step": 187131
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.3383749723434448,
      "learning_rate": 5.071930721060519e-05,
      "loss": 2.0561,
      "step": 187132
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.3584949970245361,
      "learning_rate": 5.0717031373654626e-05,
      "loss": 1.9377,
      "step": 187133
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.5343645811080933,
      "learning_rate": 5.071475558305066e-05,
      "loss": 1.9633,
      "step": 187134
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.3817564249038696,
      "learning_rate": 5.071247983879335e-05,
      "loss": 1.9234,
      "step": 187135
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.322932481765747,
      "learning_rate": 5.0710204140883404e-05,
      "loss": 1.8335,
      "step": 187136
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.3527944087982178,
      "learning_rate": 5.0707928489321046e-05,
      "loss": 2.1376,
      "step": 187137
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.294967770576477,
      "learning_rate": 5.0705652884106854e-05,
      "loss": 1.9821,
      "step": 187138
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.5346897840499878,
      "learning_rate": 5.070337732524111e-05,
      "loss": 1.8743,
      "step": 187139
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.3843400478363037,
      "learning_rate": 5.0701101812724465e-05,
      "loss": 2.0027,
      "step": 187140
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.5996603965759277,
      "learning_rate": 5.0698826346557074e-05,
      "loss": 1.8938,
      "step": 187141
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.4261233806610107,
      "learning_rate": 5.069655092673951e-05,
      "loss": 1.9291,
      "step": 187142
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.3749053478240967,
      "learning_rate": 5.06942755532721e-05,
      "loss": 1.9022,
      "step": 187143
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.4038063287734985,
      "learning_rate": 5.0692000226155414e-05,
      "loss": 2.0074,
      "step": 187144
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.3186402320861816,
      "learning_rate": 5.068972494538972e-05,
      "loss": 1.9254,
      "step": 187145
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.2526966333389282,
      "learning_rate": 5.068744971097565e-05,
      "loss": 1.853,
      "step": 187146
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.360903024673462,
      "learning_rate": 5.068517452291336e-05,
      "loss": 1.7742,
      "step": 187147
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.333363652229309,
      "learning_rate": 5.068289938120344e-05,
      "loss": 2.1787,
      "step": 187148
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.466799020767212,
      "learning_rate": 5.0680624285846226e-05,
      "loss": 1.9791,
      "step": 187149
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.4034043550491333,
      "learning_rate": 5.067834923684228e-05,
      "loss": 1.8921,
      "step": 187150
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.3846806287765503,
      "learning_rate": 5.0676074234191847e-05,
      "loss": 2.0495,
      "step": 187151
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.4349260330200195,
      "learning_rate": 5.0673799277895565e-05,
      "loss": 1.9839,
      "step": 187152
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.3236969709396362,
      "learning_rate": 5.067152436795361e-05,
      "loss": 2.2537,
      "step": 187153
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.4394105672836304,
      "learning_rate": 5.066924950436657e-05,
      "loss": 2.1177,
      "step": 187154
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.3587720394134521,
      "learning_rate": 5.066697468713475e-05,
      "loss": 1.9275,
      "step": 187155
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.447448492050171,
      "learning_rate": 5.0664699916258755e-05,
      "loss": 2.2007,
      "step": 187156
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.2828459739685059,
      "learning_rate": 5.066242519173881e-05,
      "loss": 1.9907,
      "step": 187157
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.4071611166000366,
      "learning_rate": 5.066015051357548e-05,
      "loss": 2.006,
      "step": 187158
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.4541212320327759,
      "learning_rate": 5.065787588176917e-05,
      "loss": 2.3131,
      "step": 187159
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.387607455253601,
      "learning_rate": 5.0655601296320246e-05,
      "loss": 2.0787,
      "step": 187160
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.4195382595062256,
      "learning_rate": 5.0653326757229074e-05,
      "loss": 2.1882,
      "step": 187161
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.3896815776824951,
      "learning_rate": 5.065105226449625e-05,
      "loss": 2.0039,
      "step": 187162
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.3371013402938843,
      "learning_rate": 5.064877781812201e-05,
      "loss": 2.0373,
      "step": 187163
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.3648128509521484,
      "learning_rate": 5.0646503418106954e-05,
      "loss": 2.2476,
      "step": 187164
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.3651522397994995,
      "learning_rate": 5.064422906445141e-05,
      "loss": 1.8234,
      "step": 187165
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.5458133220672607,
      "learning_rate": 5.064195475715582e-05,
      "loss": 2.08,
      "step": 187166
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.68533456325531,
      "learning_rate": 5.063968049622051e-05,
      "loss": 2.0469,
      "step": 187167
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.344582200050354,
      "learning_rate": 5.063740628164605e-05,
      "loss": 2.0531,
      "step": 187168
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.3705358505249023,
      "learning_rate": 5.0635132113432775e-05,
      "loss": 1.9377,
      "step": 187169
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.412185788154602,
      "learning_rate": 5.0632857991581176e-05,
      "loss": 2.0751,
      "step": 187170
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.535529613494873,
      "learning_rate": 5.063058391609166e-05,
      "loss": 2.0612,
      "step": 187171
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.4041836261749268,
      "learning_rate": 5.062830988696452e-05,
      "loss": 2.1206,
      "step": 187172
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.48569655418396,
      "learning_rate": 5.06260359042004e-05,
      "loss": 2.0729,
      "step": 187173
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.2601003646850586,
      "learning_rate": 5.062376196779956e-05,
      "loss": 1.82,
      "step": 187174
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.3421515226364136,
      "learning_rate": 5.0621488077762426e-05,
      "loss": 1.9889,
      "step": 187175
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.4145268201828003,
      "learning_rate": 5.0619214234089545e-05,
      "loss": 2.0636,
      "step": 187176
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.335778832435608,
      "learning_rate": 5.061694043678124e-05,
      "loss": 1.9921,
      "step": 187177
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.4262624979019165,
      "learning_rate": 5.061466668583788e-05,
      "loss": 2.1924,
      "step": 187178
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.4120641946792603,
      "learning_rate": 5.0612392981260064e-05,
      "loss": 1.9244,
      "step": 187179
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.4075342416763306,
      "learning_rate": 5.0610119323047995e-05,
      "loss": 2.1111,
      "step": 187180
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.3229814767837524,
      "learning_rate": 5.06078457112023e-05,
      "loss": 1.9626,
      "step": 187181
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.418683648109436,
      "learning_rate": 5.060557214572335e-05,
      "loss": 1.954,
      "step": 187182
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.3809524774551392,
      "learning_rate": 5.060329862661148e-05,
      "loss": 2.0552,
      "step": 187183
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.4836435317993164,
      "learning_rate": 5.060102515386711e-05,
      "loss": 2.0898,
      "step": 187184
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.329149603843689,
      "learning_rate": 5.0598751727490794e-05,
      "loss": 1.9826,
      "step": 187185
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.5739473104476929,
      "learning_rate": 5.0596478347482814e-05,
      "loss": 1.9514,
      "step": 187186
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.3564029932022095,
      "learning_rate": 5.0594205013843715e-05,
      "loss": 2.0436,
      "step": 187187
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.4941297769546509,
      "learning_rate": 5.059193172657389e-05,
      "loss": 1.9666,
      "step": 187188
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.2883967161178589,
      "learning_rate": 5.05896584856737e-05,
      "loss": 1.9633,
      "step": 187189
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.38229501247406,
      "learning_rate": 5.058738529114352e-05,
      "loss": 1.9988,
      "step": 187190
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.4052579402923584,
      "learning_rate": 5.058511214298395e-05,
      "loss": 1.8622,
      "step": 187191
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.3521003723144531,
      "learning_rate": 5.058283904119522e-05,
      "loss": 2.0794,
      "step": 187192
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.3043619394302368,
      "learning_rate": 5.0580565985777966e-05,
      "loss": 1.7724,
      "step": 187193
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.3670405149459839,
      "learning_rate": 5.057829297673238e-05,
      "loss": 1.8787,
      "step": 187194
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.245242714881897,
      "learning_rate": 5.057602001405917e-05,
      "loss": 2.0105,
      "step": 187195
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.4213781356811523,
      "learning_rate": 5.05737470977584e-05,
      "loss": 1.8929,
      "step": 187196
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.347176194190979,
      "learning_rate": 5.05714742278308e-05,
      "loss": 1.9462,
      "step": 187197
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.2570465803146362,
      "learning_rate": 5.0569201404276574e-05,
      "loss": 1.9634,
      "step": 187198
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.5910494327545166,
      "learning_rate": 5.056692862709632e-05,
      "loss": 2.0501,
      "step": 187199
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.4198275804519653,
      "learning_rate": 5.0564655896290296e-05,
      "loss": 2.1993,
      "step": 187200
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.3892598152160645,
      "learning_rate": 5.0562383211859145e-05,
      "loss": 1.765,
      "step": 187201
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.3463571071624756,
      "learning_rate": 5.0560110573803034e-05,
      "loss": 1.8107,
      "step": 187202
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.59792160987854,
      "learning_rate": 5.055783798212259e-05,
      "loss": 2.1076,
      "step": 187203
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.5214799642562866,
      "learning_rate": 5.055556543681806e-05,
      "loss": 2.0242,
      "step": 187204
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.4600995779037476,
      "learning_rate": 5.0553292937890025e-05,
      "loss": 2.0544,
      "step": 187205
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.3317664861679077,
      "learning_rate": 5.055102048533879e-05,
      "loss": 2.1079,
      "step": 187206
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.2925597429275513,
      "learning_rate": 5.054874807916496e-05,
      "loss": 1.9696,
      "step": 187207
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.394207239151001,
      "learning_rate": 5.05464757193687e-05,
      "loss": 2.0336,
      "step": 187208
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.2931849956512451,
      "learning_rate": 5.05442034059506e-05,
      "loss": 1.9861,
      "step": 187209
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.3239731788635254,
      "learning_rate": 5.054193113891101e-05,
      "loss": 1.927,
      "step": 187210
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.3071436882019043,
      "learning_rate": 5.053965891825041e-05,
      "loss": 2.118,
      "step": 187211
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.2606056928634644,
      "learning_rate": 5.053738674396915e-05,
      "loss": 1.849,
      "step": 187212
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.4793192148208618,
      "learning_rate": 5.053511461606786e-05,
      "loss": 2.2373,
      "step": 187213
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.4110301733016968,
      "learning_rate": 5.0532842534546637e-05,
      "loss": 1.9002,
      "step": 187214
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.3190836906433105,
      "learning_rate": 5.0530570499406145e-05,
      "loss": 1.8784,
      "step": 187215
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.4333354234695435,
      "learning_rate": 5.052829851064665e-05,
      "loss": 1.9246,
      "step": 187216
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.460875153541565,
      "learning_rate": 5.052602656826872e-05,
      "loss": 2.0481,
      "step": 187217
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.3207675218582153,
      "learning_rate": 5.0523754672272655e-05,
      "loss": 2.0648,
      "step": 187218
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.316483497619629,
      "learning_rate": 5.052148282265906e-05,
      "loss": 1.9327,
      "step": 187219
    },
    {
      "epoch": 2.44,
      "grad_norm": 2.0626211166381836,
      "learning_rate": 5.051921101942808e-05,
      "loss": 2.0451,
      "step": 187220
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.695580005645752,
      "learning_rate": 5.051693926258038e-05,
      "loss": 2.0502,
      "step": 187221
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.3982678651809692,
      "learning_rate": 5.051466755211624e-05,
      "loss": 2.001,
      "step": 187222
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.3812816143035889,
      "learning_rate": 5.051239588803616e-05,
      "loss": 1.8459,
      "step": 187223
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.6248831748962402,
      "learning_rate": 5.051012427034048e-05,
      "loss": 2.0241,
      "step": 187224
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.4955092668533325,
      "learning_rate": 5.050785269902983e-05,
      "loss": 2.0478,
      "step": 187225
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.5768178701400757,
      "learning_rate": 5.050558117410434e-05,
      "loss": 1.755,
      "step": 187226
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.3871668577194214,
      "learning_rate": 5.0503309695564645e-05,
      "loss": 2.2595,
      "step": 187227
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.5167763233184814,
      "learning_rate": 5.050103826341098e-05,
      "loss": 2.0183,
      "step": 187228
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.457603931427002,
      "learning_rate": 5.0498766877644005e-05,
      "loss": 2.03,
      "step": 187229
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.358927607536316,
      "learning_rate": 5.0496495538263925e-05,
      "loss": 2.0282,
      "step": 187230
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.5749146938323975,
      "learning_rate": 5.049422424527141e-05,
      "loss": 1.958,
      "step": 187231
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.554100751876831,
      "learning_rate": 5.049195299866655e-05,
      "loss": 2.0792,
      "step": 187232
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.575731635093689,
      "learning_rate": 5.048968179845005e-05,
      "loss": 1.9949,
      "step": 187233
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.5765358209609985,
      "learning_rate": 5.048741064462215e-05,
      "loss": 2.0666,
      "step": 187234
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.3632261753082275,
      "learning_rate": 5.048513953718344e-05,
      "loss": 2.1024,
      "step": 187235
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.6037044525146484,
      "learning_rate": 5.048286847613415e-05,
      "loss": 1.962,
      "step": 187236
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.297386884689331,
      "learning_rate": 5.048059746147489e-05,
      "loss": 1.9831,
      "step": 187237
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.3247896432876587,
      "learning_rate": 5.047832649320599e-05,
      "loss": 2.0068,
      "step": 187238
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.2671808004379272,
      "learning_rate": 5.0476055571327876e-05,
      "loss": 1.9068,
      "step": 187239
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.2774845361709595,
      "learning_rate": 5.047378469584092e-05,
      "loss": 1.8454,
      "step": 187240
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.3212323188781738,
      "learning_rate": 5.047151386674566e-05,
      "loss": 1.8846,
      "step": 187241
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.4050730466842651,
      "learning_rate": 5.046924308404239e-05,
      "loss": 2.1185,
      "step": 187242
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.3664249181747437,
      "learning_rate": 5.046697234773167e-05,
      "loss": 1.8478,
      "step": 187243
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.2329249382019043,
      "learning_rate": 5.0464701657813845e-05,
      "loss": 1.9399,
      "step": 187244
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.4230290651321411,
      "learning_rate": 5.046243101428934e-05,
      "loss": 2.1271,
      "step": 187245
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.2276309728622437,
      "learning_rate": 5.046016041715849e-05,
      "loss": 1.8775,
      "step": 187246
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.3529517650604248,
      "learning_rate": 5.04578898664219e-05,
      "loss": 1.9255,
      "step": 187247
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.2934486865997314,
      "learning_rate": 5.045561936207982e-05,
      "loss": 1.9151,
      "step": 187248
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.3264895677566528,
      "learning_rate": 5.045334890413284e-05,
      "loss": 1.932,
      "step": 187249
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.3716998100280762,
      "learning_rate": 5.045107849258128e-05,
      "loss": 1.9278,
      "step": 187250
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.332270622253418,
      "learning_rate": 5.044880812742557e-05,
      "loss": 2.0932,
      "step": 187251
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.3464174270629883,
      "learning_rate": 5.0446537808666075e-05,
      "loss": 2.1088,
      "step": 187252
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.4321398735046387,
      "learning_rate": 5.044426753630334e-05,
      "loss": 1.8406,
      "step": 187253
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.4576911926269531,
      "learning_rate": 5.0441997310337653e-05,
      "loss": 2.1262,
      "step": 187254
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.2949495315551758,
      "learning_rate": 5.043972713076959e-05,
      "loss": 2.039,
      "step": 187255
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.4323288202285767,
      "learning_rate": 5.043745699759951e-05,
      "loss": 1.951,
      "step": 187256
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.474035620689392,
      "learning_rate": 5.043518691082772e-05,
      "loss": 1.8649,
      "step": 187257
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.5405733585357666,
      "learning_rate": 5.043291687045481e-05,
      "loss": 1.7828,
      "step": 187258
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.4833323955535889,
      "learning_rate": 5.043064687648115e-05,
      "loss": 2.0443,
      "step": 187259
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.3106048107147217,
      "learning_rate": 5.042837692890708e-05,
      "loss": 2.008,
      "step": 187260
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.3125158548355103,
      "learning_rate": 5.042610702773312e-05,
      "loss": 1.9923,
      "step": 187261
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.3558896780014038,
      "learning_rate": 5.042383717295968e-05,
      "loss": 1.8383,
      "step": 187262
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.3666578531265259,
      "learning_rate": 5.042156736458709e-05,
      "loss": 2.1015,
      "step": 187263
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.2629148960113525,
      "learning_rate": 5.041929760261592e-05,
      "loss": 1.9175,
      "step": 187264
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.378338098526001,
      "learning_rate": 5.041702788704649e-05,
      "loss": 2.2014,
      "step": 187265
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.2080986499786377,
      "learning_rate": 5.0414758217879184e-05,
      "loss": 2.05,
      "step": 187266
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.4165490865707397,
      "learning_rate": 5.0412488595114554e-05,
      "loss": 2.0886,
      "step": 187267
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.4307125806808472,
      "learning_rate": 5.041021901875297e-05,
      "loss": 1.9994,
      "step": 187268
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.3112577199935913,
      "learning_rate": 5.040794948879477e-05,
      "loss": 2.0048,
      "step": 187269
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.4060577154159546,
      "learning_rate": 5.040568000524049e-05,
      "loss": 2.0207,
      "step": 187270
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.5028860569000244,
      "learning_rate": 5.040341056809045e-05,
      "loss": 1.8399,
      "step": 187271
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.3927125930786133,
      "learning_rate": 5.040114117734519e-05,
      "loss": 2.1822,
      "step": 187272
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.6397035121917725,
      "learning_rate": 5.039887183300508e-05,
      "loss": 2.1007,
      "step": 187273
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.4888964891433716,
      "learning_rate": 5.039660253507052e-05,
      "loss": 1.9932,
      "step": 187274
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.4614105224609375,
      "learning_rate": 5.039433328354186e-05,
      "loss": 2.0499,
      "step": 187275
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.4193185567855835,
      "learning_rate": 5.039206407841972e-05,
      "loss": 1.9129,
      "step": 187276
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.2358585596084595,
      "learning_rate": 5.0389794919704295e-05,
      "loss": 1.9604,
      "step": 187277
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.4003475904464722,
      "learning_rate": 5.038752580739621e-05,
      "loss": 1.7567,
      "step": 187278
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.3228288888931274,
      "learning_rate": 5.0385256741495705e-05,
      "loss": 2.0069,
      "step": 187279
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.2875815629959106,
      "learning_rate": 5.038298772200348e-05,
      "loss": 2.0612,
      "step": 187280
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.5056332349777222,
      "learning_rate": 5.038071874891959e-05,
      "loss": 2.244,
      "step": 187281
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.4351742267608643,
      "learning_rate": 5.037844982224472e-05,
      "loss": 1.9792,
      "step": 187282
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.3724414110183716,
      "learning_rate": 5.037618094197912e-05,
      "loss": 2.0024,
      "step": 187283
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.2905572652816772,
      "learning_rate": 5.037391210812339e-05,
      "loss": 1.9034,
      "step": 187284
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.272032380104065,
      "learning_rate": 5.037164332067778e-05,
      "loss": 2.0765,
      "step": 187285
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.275252342224121,
      "learning_rate": 5.036937457964294e-05,
      "loss": 2.1047,
      "step": 187286
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.1895509958267212,
      "learning_rate": 5.036710588501901e-05,
      "loss": 1.9557,
      "step": 187287
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.350089430809021,
      "learning_rate": 5.036483723680659e-05,
      "loss": 1.9635,
      "step": 187288
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.2273614406585693,
      "learning_rate": 5.036256863500601e-05,
      "loss": 1.8003,
      "step": 187289
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.412469506263733,
      "learning_rate": 5.0360300079617797e-05,
      "loss": 1.8453,
      "step": 187290
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.3251267671585083,
      "learning_rate": 5.035803157064227e-05,
      "loss": 1.986,
      "step": 187291
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.362197756767273,
      "learning_rate": 5.035576310808004e-05,
      "loss": 2.128,
      "step": 187292
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.2226147651672363,
      "learning_rate": 5.0353494691931215e-05,
      "loss": 2.1106,
      "step": 187293
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.4023160934448242,
      "learning_rate": 5.035122632219647e-05,
      "loss": 2.0004,
      "step": 187294
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.3436675071716309,
      "learning_rate": 5.0348957998876096e-05,
      "loss": 1.9994,
      "step": 187295
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.3744933605194092,
      "learning_rate": 5.034668972197063e-05,
      "loss": 2.1496,
      "step": 187296
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.3639427423477173,
      "learning_rate": 5.0344421491480327e-05,
      "loss": 1.9123,
      "step": 187297
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.3122479915618896,
      "learning_rate": 5.034215330740587e-05,
      "loss": 1.9922,
      "step": 187298
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.4115495681762695,
      "learning_rate": 5.0339885169747376e-05,
      "loss": 1.9522,
      "step": 187299
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.3382786512374878,
      "learning_rate": 5.033761707850549e-05,
      "loss": 1.9607,
      "step": 187300
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.386120080947876,
      "learning_rate": 5.033534903368044e-05,
      "loss": 2.0772,
      "step": 187301
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.4205248355865479,
      "learning_rate": 5.0333081035272894e-05,
      "loss": 1.9016,
      "step": 187302
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.3969565629959106,
      "learning_rate": 5.0330813083283015e-05,
      "loss": 2.0546,
      "step": 187303
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.4412462711334229,
      "learning_rate": 5.032854517771144e-05,
      "loss": 1.8758,
      "step": 187304
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.3559008836746216,
      "learning_rate": 5.032627731855854e-05,
      "loss": 2.1241,
      "step": 187305
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.631927490234375,
      "learning_rate": 5.032400950582467e-05,
      "loss": 2.095,
      "step": 187306
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.460838794708252,
      "learning_rate": 5.03217417395102e-05,
      "loss": 2.0071,
      "step": 187307
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.6053407192230225,
      "learning_rate": 5.0319474019615705e-05,
      "loss": 1.8242,
      "step": 187308
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.3674663305282593,
      "learning_rate": 5.031720634614143e-05,
      "loss": 2.0883,
      "step": 187309
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.3197981119155884,
      "learning_rate": 5.031493871908804e-05,
      "loss": 2.1734,
      "step": 187310
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.4257709980010986,
      "learning_rate": 5.031267113845577e-05,
      "loss": 2.113,
      "step": 187311
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.332679033279419,
      "learning_rate": 5.031040360424511e-05,
      "loss": 2.0087,
      "step": 187312
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.514288067817688,
      "learning_rate": 5.030813611645638e-05,
      "loss": 2.0758,
      "step": 187313
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.302956461906433,
      "learning_rate": 5.0305868675090146e-05,
      "loss": 2.0801,
      "step": 187314
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.39216148853302,
      "learning_rate": 5.0303601280146674e-05,
      "loss": 1.9253,
      "step": 187315
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.3166885375976562,
      "learning_rate": 5.030133393162658e-05,
      "loss": 1.961,
      "step": 187316
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.4028613567352295,
      "learning_rate": 5.029906662953017e-05,
      "loss": 1.7865,
      "step": 187317
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.4899729490280151,
      "learning_rate": 5.0296799373857866e-05,
      "loss": 1.917,
      "step": 187318
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.2500890493392944,
      "learning_rate": 5.029453216461006e-05,
      "loss": 1.8651,
      "step": 187319
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.255232334136963,
      "learning_rate": 5.0292265001787243e-05,
      "loss": 2.0595,
      "step": 187320
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.4442793130874634,
      "learning_rate": 5.028999788538976e-05,
      "loss": 1.8862,
      "step": 187321
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.3976116180419922,
      "learning_rate": 5.0287730815418146e-05,
      "loss": 2.0456,
      "step": 187322
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.2880405187606812,
      "learning_rate": 5.028546379187276e-05,
      "loss": 2.0715,
      "step": 187323
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.4816323518753052,
      "learning_rate": 5.0283196814754035e-05,
      "loss": 2.0177,
      "step": 187324
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.2832098007202148,
      "learning_rate": 5.02809298840623e-05,
      "loss": 1.8301,
      "step": 187325
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.453931212425232,
      "learning_rate": 5.0278662999798104e-05,
      "loss": 1.9842,
      "step": 187326
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.3539519309997559,
      "learning_rate": 5.027639616196176e-05,
      "loss": 1.9664,
      "step": 187327
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.3480356931686401,
      "learning_rate": 5.027412937055382e-05,
      "loss": 2.0412,
      "step": 187328
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.3106276988983154,
      "learning_rate": 5.027186262557463e-05,
      "loss": 2.197,
      "step": 187329
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.5819997787475586,
      "learning_rate": 5.026959592702464e-05,
      "loss": 2.0851,
      "step": 187330
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.3066353797912598,
      "learning_rate": 5.026732927490414e-05,
      "loss": 2.1599,
      "step": 187331
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.5602458715438843,
      "learning_rate": 5.026506266921374e-05,
      "loss": 2.1081,
      "step": 187332
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.4595286846160889,
      "learning_rate": 5.02627961099537e-05,
      "loss": 2.186,
      "step": 187333
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.3656541109085083,
      "learning_rate": 5.026052959712461e-05,
      "loss": 2.0033,
      "step": 187334
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.41374933719635,
      "learning_rate": 5.025826313072678e-05,
      "loss": 2.1041,
      "step": 187335
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.5120223760604858,
      "learning_rate": 5.0255996710760684e-05,
      "loss": 2.0898,
      "step": 187336
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.3367663621902466,
      "learning_rate": 5.02537303372266e-05,
      "loss": 1.8355,
      "step": 187337
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.405892014503479,
      "learning_rate": 5.025146401012519e-05,
      "loss": 2.2072,
      "step": 187338
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.3020445108413696,
      "learning_rate": 5.024919772945663e-05,
      "loss": 2.1668,
      "step": 187339
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.3664958477020264,
      "learning_rate": 5.024693149522152e-05,
      "loss": 2.1958,
      "step": 187340
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.4411710500717163,
      "learning_rate": 5.0244665307420276e-05,
      "loss": 2.0708,
      "step": 187341
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.4318218231201172,
      "learning_rate": 5.024239916605315e-05,
      "loss": 2.0706,
      "step": 187342
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.3615928888320923,
      "learning_rate": 5.0240133071120746e-05,
      "loss": 2.1272,
      "step": 187343
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.3885879516601562,
      "learning_rate": 5.0237867022623435e-05,
      "loss": 1.8546,
      "step": 187344
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.379239797592163,
      "learning_rate": 5.023560102056151e-05,
      "loss": 2.0853,
      "step": 187345
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.384747862815857,
      "learning_rate": 5.023333506493561e-05,
      "loss": 2.0429,
      "step": 187346
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.5399186611175537,
      "learning_rate": 5.023106915574603e-05,
      "loss": 2.1776,
      "step": 187347
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.6222429275512695,
      "learning_rate": 5.0228803292993166e-05,
      "loss": 2.078,
      "step": 187348
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.3469181060791016,
      "learning_rate": 5.0226537476677534e-05,
      "loss": 2.1419,
      "step": 187349
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.331416368484497,
      "learning_rate": 5.0224271706799514e-05,
      "loss": 2.2349,
      "step": 187350
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.4332398176193237,
      "learning_rate": 5.022200598335941e-05,
      "loss": 2.0303,
      "step": 187351
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.4504276514053345,
      "learning_rate": 5.021974030635784e-05,
      "loss": 1.9258,
      "step": 187352
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.3258802890777588,
      "learning_rate": 5.021747467579514e-05,
      "loss": 1.9801,
      "step": 187353
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.5036293268203735,
      "learning_rate": 5.021520909167167e-05,
      "loss": 2.3146,
      "step": 187354
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.2869644165039062,
      "learning_rate": 5.021294355398795e-05,
      "loss": 1.9243,
      "step": 187355
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.3315705060958862,
      "learning_rate": 5.0210678062744315e-05,
      "loss": 2.0696,
      "step": 187356
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.5078922510147095,
      "learning_rate": 5.02084126179413e-05,
      "loss": 1.984,
      "step": 187357
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.6724640130996704,
      "learning_rate": 5.0206147219579226e-05,
      "loss": 1.9562,
      "step": 187358
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.3192435503005981,
      "learning_rate": 5.020388186765858e-05,
      "loss": 2.1701,
      "step": 187359
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.392594337463379,
      "learning_rate": 5.0201616562179644e-05,
      "loss": 2.0165,
      "step": 187360
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.582149863243103,
      "learning_rate": 5.0199351303143034e-05,
      "loss": 2.0758,
      "step": 187361
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.3287203311920166,
      "learning_rate": 5.0197086090549e-05,
      "loss": 1.8974,
      "step": 187362
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.3831868171691895,
      "learning_rate": 5.0194820924398115e-05,
      "loss": 2.0477,
      "step": 187363
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.2678968906402588,
      "learning_rate": 5.019255580469075e-05,
      "loss": 2.1107,
      "step": 187364
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.3918992280960083,
      "learning_rate": 5.0190290731427294e-05,
      "loss": 2.1055,
      "step": 187365
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.4574763774871826,
      "learning_rate": 5.0188025704608094e-05,
      "loss": 1.7571,
      "step": 187366
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.4176559448242188,
      "learning_rate": 5.0185760724233736e-05,
      "loss": 1.932,
      "step": 187367
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.4278773069381714,
      "learning_rate": 5.018349579030449e-05,
      "loss": 2.2452,
      "step": 187368
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.3444777727127075,
      "learning_rate": 5.0181230902820904e-05,
      "loss": 1.9869,
      "step": 187369
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.3049315214157104,
      "learning_rate": 5.017896606178329e-05,
      "loss": 1.9446,
      "step": 187370
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.3992873430252075,
      "learning_rate": 5.0176701267192186e-05,
      "loss": 2.0574,
      "step": 187371
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.4452122449874878,
      "learning_rate": 5.0174436519047936e-05,
      "loss": 2.1814,
      "step": 187372
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.3405983448028564,
      "learning_rate": 5.017217181735099e-05,
      "loss": 2.2609,
      "step": 187373
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.3798099756240845,
      "learning_rate": 5.01699071621017e-05,
      "loss": 1.9358,
      "step": 187374
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.521244764328003,
      "learning_rate": 5.016764255330058e-05,
      "loss": 2.1013,
      "step": 187375
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.381025791168213,
      "learning_rate": 5.016537799094794e-05,
      "loss": 1.9838,
      "step": 187376
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.4810985326766968,
      "learning_rate": 5.016311347504435e-05,
      "loss": 2.1563,
      "step": 187377
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.4101197719573975,
      "learning_rate": 5.016084900559016e-05,
      "loss": 2.2575,
      "step": 187378
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.4082980155944824,
      "learning_rate": 5.015858458258579e-05,
      "loss": 2.1322,
      "step": 187379
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.5377130508422852,
      "learning_rate": 5.015632020603156e-05,
      "loss": 1.934,
      "step": 187380
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.6546164751052856,
      "learning_rate": 5.015405587592808e-05,
      "loss": 1.9076,
      "step": 187381
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.6207538843154907,
      "learning_rate": 5.01517915922756e-05,
      "loss": 2.0488,
      "step": 187382
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.2605955600738525,
      "learning_rate": 5.0149527355074706e-05,
      "loss": 1.9028,
      "step": 187383
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.4449834823608398,
      "learning_rate": 5.014726316432571e-05,
      "loss": 2.0812,
      "step": 187384
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.688326120376587,
      "learning_rate": 5.014499902002906e-05,
      "loss": 2.0456,
      "step": 187385
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.3707841634750366,
      "learning_rate": 5.014273492218509e-05,
      "loss": 1.995,
      "step": 187386
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.6887327432632446,
      "learning_rate": 5.014047087079439e-05,
      "loss": 2.062,
      "step": 187387
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.6229807138442993,
      "learning_rate": 5.01382068658572e-05,
      "loss": 1.9199,
      "step": 187388
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.4782636165618896,
      "learning_rate": 5.0135942907374115e-05,
      "loss": 2.0977,
      "step": 187389
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.3833706378936768,
      "learning_rate": 5.013367899534547e-05,
      "loss": 2.2097,
      "step": 187390
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.342397689819336,
      "learning_rate": 5.01314151297717e-05,
      "loss": 1.7397,
      "step": 187391
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.3459032773971558,
      "learning_rate": 5.012915131065317e-05,
      "loss": 1.8612,
      "step": 187392
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.3110017776489258,
      "learning_rate": 5.012688753799038e-05,
      "loss": 1.9079,
      "step": 187393
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.6517161130905151,
      "learning_rate": 5.012462381178366e-05,
      "loss": 1.8481,
      "step": 187394
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.434517502784729,
      "learning_rate": 5.012236013203355e-05,
      "loss": 2.0625,
      "step": 187395
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.3248971700668335,
      "learning_rate": 5.012009649874044e-05,
      "loss": 1.908,
      "step": 187396
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.3582693338394165,
      "learning_rate": 5.01178329119047e-05,
      "loss": 2.0554,
      "step": 187397
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.4664143323898315,
      "learning_rate": 5.0115569371526696e-05,
      "loss": 1.9424,
      "step": 187398
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.3782196044921875,
      "learning_rate": 5.0113305877607e-05,
      "loss": 2.1039,
      "step": 187399
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.3319474458694458,
      "learning_rate": 5.011104243014588e-05,
      "loss": 1.9893,
      "step": 187400
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.4534389972686768,
      "learning_rate": 5.0108779029143916e-05,
      "loss": 2.1436,
      "step": 187401
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.3541905879974365,
      "learning_rate": 5.0106515674601434e-05,
      "loss": 1.9741,
      "step": 187402
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.6093798875808716,
      "learning_rate": 5.0104252366518884e-05,
      "loss": 2.062,
      "step": 187403
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.3399044275283813,
      "learning_rate": 5.01019891048966e-05,
      "loss": 1.8318,
      "step": 187404
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.3978347778320312,
      "learning_rate": 5.009972588973512e-05,
      "loss": 1.8127,
      "step": 187405
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.3720356225967407,
      "learning_rate": 5.009746272103477e-05,
      "loss": 2.0143,
      "step": 187406
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.422905683517456,
      "learning_rate": 5.0095199598796096e-05,
      "loss": 1.9904,
      "step": 187407
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.2856976985931396,
      "learning_rate": 5.009293652301942e-05,
      "loss": 1.9353,
      "step": 187408
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.2978954315185547,
      "learning_rate": 5.009067349370521e-05,
      "loss": 2.0173,
      "step": 187409
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.5900503396987915,
      "learning_rate": 5.008841051085377e-05,
      "loss": 2.2637,
      "step": 187410
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.4055529832839966,
      "learning_rate": 5.00861475744657e-05,
      "loss": 2.1462,
      "step": 187411
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.3311930894851685,
      "learning_rate": 5.008388468454122e-05,
      "loss": 1.9315,
      "step": 187412
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.3726701736450195,
      "learning_rate": 5.008162184108098e-05,
      "loss": 2.2166,
      "step": 187413
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.3612631559371948,
      "learning_rate": 5.007935904408527e-05,
      "loss": 2.0298,
      "step": 187414
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.3940534591674805,
      "learning_rate": 5.007709629355452e-05,
      "loss": 1.9813,
      "step": 187415
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.3049697875976562,
      "learning_rate": 5.007483358948907e-05,
      "loss": 1.8425,
      "step": 187416
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.3244539499282837,
      "learning_rate": 5.0072570931889525e-05,
      "loss": 2.1818,
      "step": 187417
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.5574151277542114,
      "learning_rate": 5.007030832075613e-05,
      "loss": 1.9617,
      "step": 187418
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.2879489660263062,
      "learning_rate": 5.0068045756089444e-05,
      "loss": 2.1339,
      "step": 187419
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.3402369022369385,
      "learning_rate": 5.0065783237889824e-05,
      "loss": 2.0094,
      "step": 187420
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.2182276248931885,
      "learning_rate": 5.006352076615769e-05,
      "loss": 1.8732,
      "step": 187421
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.3643138408660889,
      "learning_rate": 5.006125834089343e-05,
      "loss": 2.1026,
      "step": 187422
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.2501741647720337,
      "learning_rate": 5.0058995962097524e-05,
      "loss": 2.0046,
      "step": 187423
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.588585376739502,
      "learning_rate": 5.005673362977032e-05,
      "loss": 2.0874,
      "step": 187424
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.432242512702942,
      "learning_rate": 5.005447134391237e-05,
      "loss": 1.9021,
      "step": 187425
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.5696382522583008,
      "learning_rate": 5.005220910452399e-05,
      "loss": 1.92,
      "step": 187426
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.326948881149292,
      "learning_rate": 5.004994691160563e-05,
      "loss": 1.8667,
      "step": 187427
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.4173541069030762,
      "learning_rate": 5.004768476515764e-05,
      "loss": 1.9716,
      "step": 187428
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.460645318031311,
      "learning_rate": 5.004542266518057e-05,
      "loss": 2.0,
      "step": 187429
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.5410282611846924,
      "learning_rate": 5.004316061167469e-05,
      "loss": 2.0178,
      "step": 187430
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.3717349767684937,
      "learning_rate": 5.0040898604640576e-05,
      "loss": 1.9912,
      "step": 187431
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.46149742603302,
      "learning_rate": 5.0038636644078524e-05,
      "loss": 1.897,
      "step": 187432
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.28738534450531,
      "learning_rate": 5.003637472998909e-05,
      "loss": 1.9818,
      "step": 187433
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.4311679601669312,
      "learning_rate": 5.003411286237259e-05,
      "loss": 2.0783,
      "step": 187434
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.4798130989074707,
      "learning_rate": 5.0031851041229475e-05,
      "loss": 1.9643,
      "step": 187435
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.4235365390777588,
      "learning_rate": 5.0029589266560085e-05,
      "loss": 1.9891,
      "step": 187436
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.4507066011428833,
      "learning_rate": 5.002732753836496e-05,
      "loss": 1.9795,
      "step": 187437
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.5744044780731201,
      "learning_rate": 5.002506585664442e-05,
      "loss": 1.8943,
      "step": 187438
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.347065806388855,
      "learning_rate": 5.002280422139904e-05,
      "loss": 1.9259,
      "step": 187439
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.4401906728744507,
      "learning_rate": 5.002054263262911e-05,
      "loss": 2.1069,
      "step": 187440
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.3836042881011963,
      "learning_rate": 5.001828109033501e-05,
      "loss": 1.8555,
      "step": 187441
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.2875463962554932,
      "learning_rate": 5.0016019594517296e-05,
      "loss": 2.0626,
      "step": 187442
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.349358320236206,
      "learning_rate": 5.001375814517634e-05,
      "loss": 2.0824,
      "step": 187443
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.449373722076416,
      "learning_rate": 5.001149674231244e-05,
      "loss": 2.0038,
      "step": 187444
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.413053274154663,
      "learning_rate": 5.000923538592623e-05,
      "loss": 2.0945,
      "step": 187445
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.4097853899002075,
      "learning_rate": 5.0006974076018e-05,
      "loss": 2.0707,
      "step": 187446
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.6218328475952148,
      "learning_rate": 5.0004712812588134e-05,
      "loss": 2.0801,
      "step": 187447
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.3411945104599,
      "learning_rate": 5.000245159563718e-05,
      "loss": 2.0565,
      "step": 187448
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.3361576795578003,
      "learning_rate": 5.0000190425165486e-05,
      "loss": 1.8539,
      "step": 187449
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.4580681324005127,
      "learning_rate": 4.999792930117341e-05,
      "loss": 2.1156,
      "step": 187450
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.4614571332931519,
      "learning_rate": 4.999566822366152e-05,
      "loss": 2.2071,
      "step": 187451
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.3906351327896118,
      "learning_rate": 4.999340719263012e-05,
      "loss": 2.0275,
      "step": 187452
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.7241439819335938,
      "learning_rate": 4.9991146208079636e-05,
      "loss": 1.8331,
      "step": 187453
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.4421052932739258,
      "learning_rate": 4.998888527001057e-05,
      "loss": 1.9036,
      "step": 187454
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.2169206142425537,
      "learning_rate": 4.998662437842319e-05,
      "loss": 2.0671,
      "step": 187455
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.29326331615448,
      "learning_rate": 4.998436353331813e-05,
      "loss": 2.0507,
      "step": 187456
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.3364369869232178,
      "learning_rate": 4.9982102734695686e-05,
      "loss": 1.9148,
      "step": 187457
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.5016287565231323,
      "learning_rate": 4.997984198255626e-05,
      "loss": 2.1594,
      "step": 187458
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.3875482082366943,
      "learning_rate": 4.997758127690025e-05,
      "loss": 1.9382,
      "step": 187459
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.3943729400634766,
      "learning_rate": 4.997532061772819e-05,
      "loss": 2.1629,
      "step": 187460
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.2907509803771973,
      "learning_rate": 4.9973060005040376e-05,
      "loss": 1.7206,
      "step": 187461
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.515378713607788,
      "learning_rate": 4.9970799438837385e-05,
      "loss": 1.9648,
      "step": 187462
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.2897474765777588,
      "learning_rate": 4.996853891911951e-05,
      "loss": 1.8568,
      "step": 187463
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.858492136001587,
      "learning_rate": 4.9966278445887206e-05,
      "loss": 1.8692,
      "step": 187464
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.4009393453598022,
      "learning_rate": 4.996401801914083e-05,
      "loss": 1.9768,
      "step": 187465
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.5439914464950562,
      "learning_rate": 4.9961757638880925e-05,
      "loss": 2.1699,
      "step": 187466
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.4009219408035278,
      "learning_rate": 4.995949730510777e-05,
      "loss": 2.1134,
      "step": 187467
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.3993936777114868,
      "learning_rate": 4.995723701782194e-05,
      "loss": 2.1066,
      "step": 187468
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.3693015575408936,
      "learning_rate": 4.995497677702378e-05,
      "loss": 1.9295,
      "step": 187469
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.4322277307510376,
      "learning_rate": 4.995271658271374e-05,
      "loss": 2.0404,
      "step": 187470
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.5145913362503052,
      "learning_rate": 4.995045643489211e-05,
      "loss": 1.9447,
      "step": 187471
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.2742455005645752,
      "learning_rate": 4.994819633355949e-05,
      "loss": 2.0356,
      "step": 187472
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.3274542093276978,
      "learning_rate": 4.9945936278716114e-05,
      "loss": 1.9207,
      "step": 187473
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.4258171319961548,
      "learning_rate": 4.994367627036262e-05,
      "loss": 2.2316,
      "step": 187474
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.2744699716567993,
      "learning_rate": 4.9941416308499306e-05,
      "loss": 2.0166,
      "step": 187475
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.2121957540512085,
      "learning_rate": 4.99391563931266e-05,
      "loss": 1.8622,
      "step": 187476
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.4034513235092163,
      "learning_rate": 4.9936896524244875e-05,
      "loss": 2.0598,
      "step": 187477
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.3695095777511597,
      "learning_rate": 4.993463670185466e-05,
      "loss": 2.1123,
      "step": 187478
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.4555914402008057,
      "learning_rate": 4.993237692595622e-05,
      "loss": 2.0702,
      "step": 187479
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.4234646558761597,
      "learning_rate": 4.993011719655019e-05,
      "loss": 1.9843,
      "step": 187480
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.4960875511169434,
      "learning_rate": 4.992785751363684e-05,
      "loss": 2.04,
      "step": 187481
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.3719604015350342,
      "learning_rate": 4.992559787721663e-05,
      "loss": 1.9779,
      "step": 187482
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.4325759410858154,
      "learning_rate": 4.9923338287289894e-05,
      "loss": 2.0942,
      "step": 187483
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.3181606531143188,
      "learning_rate": 4.9921078743857204e-05,
      "loss": 1.8466,
      "step": 187484
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.4504114389419556,
      "learning_rate": 4.991881924691882e-05,
      "loss": 1.9979,
      "step": 187485
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.3355764150619507,
      "learning_rate": 4.991655979647534e-05,
      "loss": 2.2123,
      "step": 187486
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.2887758016586304,
      "learning_rate": 4.991430039252707e-05,
      "loss": 2.0627,
      "step": 187487
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.4889330863952637,
      "learning_rate": 4.991204103507448e-05,
      "loss": 1.931,
      "step": 187488
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.4456915855407715,
      "learning_rate": 4.990978172411789e-05,
      "loss": 1.9001,
      "step": 187489
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.4470206499099731,
      "learning_rate": 4.990752245965785e-05,
      "loss": 2.1142,
      "step": 187490
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.3669283390045166,
      "learning_rate": 4.990526324169467e-05,
      "loss": 2.1872,
      "step": 187491
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.3287118673324585,
      "learning_rate": 4.990300407022887e-05,
      "loss": 2.0057,
      "step": 187492
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.3206843137741089,
      "learning_rate": 4.990074494526084e-05,
      "loss": 2.007,
      "step": 187493
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.305295705795288,
      "learning_rate": 4.989848586679095e-05,
      "loss": 1.8975,
      "step": 187494
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.498388409614563,
      "learning_rate": 4.989622683481963e-05,
      "loss": 1.9681,
      "step": 187495
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.4734584093093872,
      "learning_rate": 4.989396784934735e-05,
      "loss": 1.9273,
      "step": 187496
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.5083955526351929,
      "learning_rate": 4.989170891037444e-05,
      "loss": 2.2312,
      "step": 187497
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.5429683923721313,
      "learning_rate": 4.9889450017901466e-05,
      "loss": 2.1951,
      "step": 187498
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.237423300743103,
      "learning_rate": 4.9887191171928696e-05,
      "loss": 1.888,
      "step": 187499
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.4774246215820312,
      "learning_rate": 4.988493237245677e-05,
      "loss": 1.9748,
      "step": 187500
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.3175805807113647,
      "learning_rate": 4.98826736194858e-05,
      "loss": 2.0865,
      "step": 187501
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.3594748973846436,
      "learning_rate": 4.9880414913016444e-05,
      "loss": 2.094,
      "step": 187502
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.322615146636963,
      "learning_rate": 4.987815625304895e-05,
      "loss": 1.9622,
      "step": 187503
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.5935487747192383,
      "learning_rate": 4.987589763958389e-05,
      "loss": 1.8868,
      "step": 187504
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.5574429035186768,
      "learning_rate": 4.987363907262157e-05,
      "loss": 1.9767,
      "step": 187505
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.3813927173614502,
      "learning_rate": 4.987138055216262e-05,
      "loss": 1.9028,
      "step": 187506
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.395113229751587,
      "learning_rate": 4.9869122078207135e-05,
      "loss": 1.8865,
      "step": 187507
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.5625382661819458,
      "learning_rate": 4.986686365075578e-05,
      "loss": 2.0212,
      "step": 187508
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.3433762788772583,
      "learning_rate": 4.986460526980883e-05,
      "loss": 2.1502,
      "step": 187509
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.3501886129379272,
      "learning_rate": 4.9862346935366847e-05,
      "loss": 2.097,
      "step": 187510
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.747170090675354,
      "learning_rate": 4.9860088647430095e-05,
      "loss": 1.8404,
      "step": 187511
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.3891520500183105,
      "learning_rate": 4.985783040599921e-05,
      "loss": 2.0638,
      "step": 187512
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.4365674257278442,
      "learning_rate": 4.985557221107432e-05,
      "loss": 1.9412,
      "step": 187513
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.5263354778289795,
      "learning_rate": 4.98533140626561e-05,
      "loss": 2.0434,
      "step": 187514
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.3133108615875244,
      "learning_rate": 4.9851055960744815e-05,
      "loss": 1.8247,
      "step": 187515
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.4676392078399658,
      "learning_rate": 4.984879790534099e-05,
      "loss": 2.1453,
      "step": 187516
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.88089120388031,
      "learning_rate": 4.9846539896444924e-05,
      "loss": 1.8156,
      "step": 187517
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.5014480352401733,
      "learning_rate": 4.984428193405716e-05,
      "loss": 2.0473,
      "step": 187518
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.3178528547286987,
      "learning_rate": 4.984202401817809e-05,
      "loss": 1.8292,
      "step": 187519
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.329519510269165,
      "learning_rate": 4.983976614880813e-05,
      "loss": 2.0805,
      "step": 187520
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.29849112033844,
      "learning_rate": 4.983750832594755e-05,
      "loss": 1.9589,
      "step": 187521
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.367860198020935,
      "learning_rate": 4.9835250549597006e-05,
      "loss": 1.9533,
      "step": 187522
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.5601484775543213,
      "learning_rate": 4.983299281975676e-05,
      "loss": 2.1068,
      "step": 187523
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.36040461063385,
      "learning_rate": 4.9830735136427334e-05,
      "loss": 1.8841,
      "step": 187524
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.5349482297897339,
      "learning_rate": 4.982847749960908e-05,
      "loss": 2.0596,
      "step": 187525
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.449851632118225,
      "learning_rate": 4.9826219909302445e-05,
      "loss": 1.8345,
      "step": 187526
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.693626880645752,
      "learning_rate": 4.982396236550777e-05,
      "loss": 2.1477,
      "step": 187527
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.5379973649978638,
      "learning_rate": 4.982170486822563e-05,
      "loss": 1.989,
      "step": 187528
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.5877784490585327,
      "learning_rate": 4.981944741745628e-05,
      "loss": 1.9295,
      "step": 187529
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.3953534364700317,
      "learning_rate": 4.981719001320026e-05,
      "loss": 1.9849,
      "step": 187530
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.7417069673538208,
      "learning_rate": 4.9814932655458e-05,
      "loss": 1.8871,
      "step": 187531
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.390626072883606,
      "learning_rate": 4.981267534422973e-05,
      "loss": 2.0682,
      "step": 187532
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.3765029907226562,
      "learning_rate": 4.9810418079516115e-05,
      "loss": 1.9829,
      "step": 187533
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.3170632123947144,
      "learning_rate": 4.980816086131746e-05,
      "loss": 2.1551,
      "step": 187534
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.4525854587554932,
      "learning_rate": 4.980590368963413e-05,
      "loss": 2.2065,
      "step": 187535
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.4842956066131592,
      "learning_rate": 4.980364656446666e-05,
      "loss": 1.8057,
      "step": 187536
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.4977589845657349,
      "learning_rate": 4.980138948581541e-05,
      "loss": 1.8869,
      "step": 187537
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.2622191905975342,
      "learning_rate": 4.979913245368076e-05,
      "loss": 1.9946,
      "step": 187538
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.3775421380996704,
      "learning_rate": 4.9796875468063226e-05,
      "loss": 2.2249,
      "step": 187539
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.4595597982406616,
      "learning_rate": 4.979461852896311e-05,
      "loss": 1.8967,
      "step": 187540
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.410230278968811,
      "learning_rate": 4.9792361636380964e-05,
      "loss": 2.0529,
      "step": 187541
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.6748220920562744,
      "learning_rate": 4.9790104790317167e-05,
      "loss": 2.1912,
      "step": 187542
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.485129714012146,
      "learning_rate": 4.978784799077209e-05,
      "loss": 2.0061,
      "step": 187543
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.4821040630340576,
      "learning_rate": 4.9785591237746104e-05,
      "loss": 2.2085,
      "step": 187544
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.6955004930496216,
      "learning_rate": 4.9783334531239774e-05,
      "loss": 1.9472,
      "step": 187545
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.4025230407714844,
      "learning_rate": 4.978107787125336e-05,
      "loss": 2.1282,
      "step": 187546
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.1987208127975464,
      "learning_rate": 4.977882125778744e-05,
      "loss": 1.7117,
      "step": 187547
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.3777875900268555,
      "learning_rate": 4.977656469084237e-05,
      "loss": 2.1155,
      "step": 187548
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.2736543416976929,
      "learning_rate": 4.977430817041859e-05,
      "loss": 1.9446,
      "step": 187549
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.4164059162139893,
      "learning_rate": 4.9772051696516355e-05,
      "loss": 1.77,
      "step": 187550
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.3029385805130005,
      "learning_rate": 4.976979526913635e-05,
      "loss": 2.0111,
      "step": 187551
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.380373239517212,
      "learning_rate": 4.976753888827876e-05,
      "loss": 2.0272,
      "step": 187552
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.3912078142166138,
      "learning_rate": 4.9765282553944186e-05,
      "loss": 1.9723,
      "step": 187553
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.5259370803833008,
      "learning_rate": 4.9763026266132963e-05,
      "loss": 2.1426,
      "step": 187554
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.3468550443649292,
      "learning_rate": 4.9760770024845534e-05,
      "loss": 1.9394,
      "step": 187555
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.3143621683120728,
      "learning_rate": 4.9758513830082216e-05,
      "loss": 2.1339,
      "step": 187556
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.4690874814987183,
      "learning_rate": 4.975625768184358e-05,
      "loss": 1.9477,
      "step": 187557
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.3956798315048218,
      "learning_rate": 4.97540015801299e-05,
      "loss": 1.8478,
      "step": 187558
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.3678998947143555,
      "learning_rate": 4.975174552494177e-05,
      "loss": 1.9046,
      "step": 187559
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.6199270486831665,
      "learning_rate": 4.9749489516279516e-05,
      "loss": 2.0167,
      "step": 187560
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.4523048400878906,
      "learning_rate": 4.974723355414355e-05,
      "loss": 2.0071,
      "step": 187561
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.3327224254608154,
      "learning_rate": 4.97449776385342e-05,
      "loss": 1.8666,
      "step": 187562
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.4255847930908203,
      "learning_rate": 4.97427217694521e-05,
      "loss": 2.1286,
      "step": 187563
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.4394487142562866,
      "learning_rate": 4.974046594689745e-05,
      "loss": 2.0905,
      "step": 187564
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.379051685333252,
      "learning_rate": 4.9738210170870843e-05,
      "loss": 1.9637,
      "step": 187565
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.411397933959961,
      "learning_rate": 4.9735954441372586e-05,
      "loss": 1.9573,
      "step": 187566
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.332241415977478,
      "learning_rate": 4.973369875840324e-05,
      "loss": 1.8553,
      "step": 187567
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.3843505382537842,
      "learning_rate": 4.973144312196301e-05,
      "loss": 2.0431,
      "step": 187568
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.3204237222671509,
      "learning_rate": 4.972918753205249e-05,
      "loss": 1.7847,
      "step": 187569
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.3274539709091187,
      "learning_rate": 4.972693198867196e-05,
      "loss": 2.14,
      "step": 187570
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.498572826385498,
      "learning_rate": 4.9724676491822e-05,
      "loss": 1.9986,
      "step": 187571
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.4435428380966187,
      "learning_rate": 4.972242104150286e-05,
      "loss": 2.1431,
      "step": 187572
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.3744220733642578,
      "learning_rate": 4.972016563771519e-05,
      "loss": 1.9783,
      "step": 187573
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.4304251670837402,
      "learning_rate": 4.9717910280459136e-05,
      "loss": 2.0212,
      "step": 187574
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.5052287578582764,
      "learning_rate": 4.971565496973533e-05,
      "loss": 2.0604,
      "step": 187575
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.4843144416809082,
      "learning_rate": 4.9713399705543996e-05,
      "loss": 1.965,
      "step": 187576
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.4126465320587158,
      "learning_rate": 4.971114448788578e-05,
      "loss": 1.902,
      "step": 187577
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.3375985622406006,
      "learning_rate": 4.9708889316760904e-05,
      "loss": 1.9065,
      "step": 187578
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.4017237424850464,
      "learning_rate": 4.9706634192170006e-05,
      "loss": 1.7916,
      "step": 187579
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.401471734046936,
      "learning_rate": 4.9704379114113214e-05,
      "loss": 1.8441,
      "step": 187580
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.43125581741333,
      "learning_rate": 4.97021240825912e-05,
      "loss": 1.9896,
      "step": 187581
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.3872781991958618,
      "learning_rate": 4.969986909760419e-05,
      "loss": 1.9862,
      "step": 187582
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.4208370447158813,
      "learning_rate": 4.9697614159152797e-05,
      "loss": 1.979,
      "step": 187583
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.3347952365875244,
      "learning_rate": 4.969535926723725e-05,
      "loss": 2.0123,
      "step": 187584
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.5584577322006226,
      "learning_rate": 4.96931044218582e-05,
      "loss": 1.7992,
      "step": 187585
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.3436890840530396,
      "learning_rate": 4.9690849623015795e-05,
      "loss": 2.023,
      "step": 187586
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.501030445098877,
      "learning_rate": 4.968859487071066e-05,
      "loss": 1.976,
      "step": 187587
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.518367886543274,
      "learning_rate": 4.968634016494307e-05,
      "loss": 2.0727,
      "step": 187588
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.3379319906234741,
      "learning_rate": 4.968408550571359e-05,
      "loss": 1.8996,
      "step": 187589
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.420754075050354,
      "learning_rate": 4.968183089302248e-05,
      "loss": 2.0594,
      "step": 187590
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.6712424755096436,
      "learning_rate": 4.967957632687037e-05,
      "loss": 2.0473,
      "step": 187591
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.5135836601257324,
      "learning_rate": 4.967732180725744e-05,
      "loss": 2.1108,
      "step": 187592
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.4884617328643799,
      "learning_rate": 4.9675067334184284e-05,
      "loss": 1.881,
      "step": 187593
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.504260540008545,
      "learning_rate": 4.967281290765116e-05,
      "loss": 1.926,
      "step": 187594
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.3266137838363647,
      "learning_rate": 4.967055852765871e-05,
      "loss": 2.0337,
      "step": 187595
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.4695947170257568,
      "learning_rate": 4.9668304194207095e-05,
      "loss": 2.0282,
      "step": 187596
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.5838366746902466,
      "learning_rate": 4.9666049907297054e-05,
      "loss": 2.0645,
      "step": 187597
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.4304920434951782,
      "learning_rate": 4.966379566692869e-05,
      "loss": 2.0721,
      "step": 187598
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.417360544204712,
      "learning_rate": 4.9661541473102586e-05,
      "loss": 1.8665,
      "step": 187599
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.431886076927185,
      "learning_rate": 4.9659287325819055e-05,
      "loss": 1.8154,
      "step": 187600
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.4599742889404297,
      "learning_rate": 4.965703322507869e-05,
      "loss": 1.9678,
      "step": 187601
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.4851611852645874,
      "learning_rate": 4.96547791708817e-05,
      "loss": 1.8931,
      "step": 187602
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.3613966703414917,
      "learning_rate": 4.9652525163228705e-05,
      "loss": 2.0795,
      "step": 187603
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.49424147605896,
      "learning_rate": 4.965027120212002e-05,
      "loss": 2.1036,
      "step": 187604
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.3712884187698364,
      "learning_rate": 4.9648017287556094e-05,
      "loss": 1.8555,
      "step": 187605
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.4476070404052734,
      "learning_rate": 4.9645763419537245e-05,
      "loss": 1.8728,
      "step": 187606
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.3298698663711548,
      "learning_rate": 4.9643509598064056e-05,
      "loss": 2.0434,
      "step": 187607
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.3164472579956055,
      "learning_rate": 4.964125582313677e-05,
      "loss": 1.9124,
      "step": 187608
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.4641207456588745,
      "learning_rate": 4.9639002094755986e-05,
      "loss": 1.9316,
      "step": 187609
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.3470410108566284,
      "learning_rate": 4.963674841292203e-05,
      "loss": 1.989,
      "step": 187610
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.371008276939392,
      "learning_rate": 4.9634494777635346e-05,
      "loss": 1.9261,
      "step": 187611
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.4619743824005127,
      "learning_rate": 4.963224118889626e-05,
      "loss": 2.169,
      "step": 187612
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.4289437532424927,
      "learning_rate": 4.962998764670534e-05,
      "loss": 2.104,
      "step": 187613
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.5114457607269287,
      "learning_rate": 4.962773415106285e-05,
      "loss": 2.1231,
      "step": 187614
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.27274751663208,
      "learning_rate": 4.962548070196935e-05,
      "loss": 1.8359,
      "step": 187615
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.3449519872665405,
      "learning_rate": 4.962322729942523e-05,
      "loss": 2.1154,
      "step": 187616
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.444957971572876,
      "learning_rate": 4.962097394343076e-05,
      "loss": 2.2528,
      "step": 187617
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.3633289337158203,
      "learning_rate": 4.96187206339866e-05,
      "loss": 2.1253,
      "step": 187618
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.293919563293457,
      "learning_rate": 4.961646737109303e-05,
      "loss": 2.1208,
      "step": 187619
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.8225128650665283,
      "learning_rate": 4.961421415475039e-05,
      "loss": 1.9749,
      "step": 187620
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.3707770109176636,
      "learning_rate": 4.9611960984959275e-05,
      "loss": 1.9186,
      "step": 187621
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.2753064632415771,
      "learning_rate": 4.960970786172006e-05,
      "loss": 2.0681,
      "step": 187622
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.4793283939361572,
      "learning_rate": 4.9607454785033005e-05,
      "loss": 2.2631,
      "step": 187623
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.6809223890304565,
      "learning_rate": 4.960520175489875e-05,
      "loss": 2.0219,
      "step": 187624
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.3860654830932617,
      "learning_rate": 4.960294877131755e-05,
      "loss": 2.0033,
      "step": 187625
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.4473834037780762,
      "learning_rate": 4.960069583428995e-05,
      "loss": 2.1699,
      "step": 187626
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.333133339881897,
      "learning_rate": 4.959844294381628e-05,
      "loss": 1.9725,
      "step": 187627
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.3438267707824707,
      "learning_rate": 4.959619009989702e-05,
      "loss": 2.053,
      "step": 187628
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.4170701503753662,
      "learning_rate": 4.959393730253247e-05,
      "loss": 2.3235,
      "step": 187629
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.4412723779678345,
      "learning_rate": 4.959168455172321e-05,
      "loss": 1.9494,
      "step": 187630
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.4028069972991943,
      "learning_rate": 4.958943184746951e-05,
      "loss": 1.8935,
      "step": 187631
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.3958863019943237,
      "learning_rate": 4.958717918977193e-05,
      "loss": 1.974,
      "step": 187632
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.4246408939361572,
      "learning_rate": 4.958492657863081e-05,
      "loss": 2.0842,
      "step": 187633
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.3231830596923828,
      "learning_rate": 4.9582674014046616e-05,
      "loss": 1.9522,
      "step": 187634
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.312804937362671,
      "learning_rate": 4.9580421496019615e-05,
      "loss": 1.9293,
      "step": 187635
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.3936353921890259,
      "learning_rate": 4.957816902455044e-05,
      "loss": 1.8416,
      "step": 187636
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.263909101486206,
      "learning_rate": 4.957591659963932e-05,
      "loss": 1.8096,
      "step": 187637
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.3614239692687988,
      "learning_rate": 4.957366422128686e-05,
      "loss": 1.8305,
      "step": 187638
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.5075843334197998,
      "learning_rate": 4.957141188949329e-05,
      "loss": 1.9789,
      "step": 187639
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.386505126953125,
      "learning_rate": 4.95691596042593e-05,
      "loss": 2.1645,
      "step": 187640
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.6314258575439453,
      "learning_rate": 4.956690736558494e-05,
      "loss": 2.0414,
      "step": 187641
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.2788904905319214,
      "learning_rate": 4.9564655173470904e-05,
      "loss": 2.0045,
      "step": 187642
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.4231557846069336,
      "learning_rate": 4.956240302791749e-05,
      "loss": 1.8836,
      "step": 187643
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.4392306804656982,
      "learning_rate": 4.9560150928925194e-05,
      "loss": 2.1446,
      "step": 187644
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.4008288383483887,
      "learning_rate": 4.9557898876494315e-05,
      "loss": 1.8219,
      "step": 187645
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.5463758707046509,
      "learning_rate": 4.95556468706255e-05,
      "loss": 2.1022,
      "step": 187646
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.3986612558364868,
      "learning_rate": 4.955339491131887e-05,
      "loss": 1.7836,
      "step": 187647
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.5542675256729126,
      "learning_rate": 4.955114299857506e-05,
      "loss": 2.1241,
      "step": 187648
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.354467749595642,
      "learning_rate": 4.9548891132394365e-05,
      "loss": 1.9632,
      "step": 187649
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.4479042291641235,
      "learning_rate": 4.954663931277733e-05,
      "loss": 2.0894,
      "step": 187650
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.5726048946380615,
      "learning_rate": 4.954438753972421e-05,
      "loss": 1.8451,
      "step": 187651
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.5815802812576294,
      "learning_rate": 4.954213581323567e-05,
      "loss": 2.1232,
      "step": 187652
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.5004380941390991,
      "learning_rate": 4.953988413331186e-05,
      "loss": 1.9058,
      "step": 187653
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.6042320728302002,
      "learning_rate": 4.953763249995336e-05,
      "loss": 1.9163,
      "step": 187654
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.368704080581665,
      "learning_rate": 4.953538091316045e-05,
      "loss": 1.9949,
      "step": 187655
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.3634697198867798,
      "learning_rate": 4.9533129372933754e-05,
      "loss": 1.9928,
      "step": 187656
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.2943274974822998,
      "learning_rate": 4.9530877879273475e-05,
      "loss": 1.9866,
      "step": 187657
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.5133752822875977,
      "learning_rate": 4.952862643218028e-05,
      "loss": 1.9446,
      "step": 187658
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.3835474252700806,
      "learning_rate": 4.952637503165431e-05,
      "loss": 1.8677,
      "step": 187659
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.473211646080017,
      "learning_rate": 4.952412367769618e-05,
      "loss": 1.906,
      "step": 187660
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.300403356552124,
      "learning_rate": 4.952187237030617e-05,
      "loss": 1.7644,
      "step": 187661
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.6225663423538208,
      "learning_rate": 4.951962110948484e-05,
      "loss": 1.8413,
      "step": 187662
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.3372746706008911,
      "learning_rate": 4.951736989523249e-05,
      "loss": 1.8636,
      "step": 187663
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.3922618627548218,
      "learning_rate": 4.95151187275497e-05,
      "loss": 2.0164,
      "step": 187664
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.3163890838623047,
      "learning_rate": 4.951286760643665e-05,
      "loss": 2.1843,
      "step": 187665
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.3689905405044556,
      "learning_rate": 4.951061653189398e-05,
      "loss": 2.0449,
      "step": 187666
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.3717825412750244,
      "learning_rate": 4.950836550392192e-05,
      "loss": 1.8758,
      "step": 187667
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.4099595546722412,
      "learning_rate": 4.950611452252105e-05,
      "loss": 2.0719,
      "step": 187668
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.5306767225265503,
      "learning_rate": 4.950386358769165e-05,
      "loss": 2.0765,
      "step": 187669
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.3814250230789185,
      "learning_rate": 4.9501612699434366e-05,
      "loss": 1.9275,
      "step": 187670
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.4382760524749756,
      "learning_rate": 4.949936185774933e-05,
      "loss": 2.0647,
      "step": 187671
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.6733237504959106,
      "learning_rate": 4.949711106263714e-05,
      "loss": 2.2032,
      "step": 187672
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.3676663637161255,
      "learning_rate": 4.9494860314098096e-05,
      "loss": 1.978,
      "step": 187673
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.3830024003982544,
      "learning_rate": 4.949260961213276e-05,
      "loss": 1.8789,
      "step": 187674
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.51369309425354,
      "learning_rate": 4.949035895674141e-05,
      "loss": 1.9431,
      "step": 187675
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.3279905319213867,
      "learning_rate": 4.948810834792466e-05,
      "loss": 1.92,
      "step": 187676
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.3620212078094482,
      "learning_rate": 4.948585778568266e-05,
      "loss": 1.8647,
      "step": 187677
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.2601635456085205,
      "learning_rate": 4.9483607270016066e-05,
      "loss": 1.7475,
      "step": 187678
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.4499460458755493,
      "learning_rate": 4.9481356800925085e-05,
      "loss": 1.8991,
      "step": 187679
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.5435324907302856,
      "learning_rate": 4.947910637841035e-05,
      "loss": 1.9185,
      "step": 187680
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.4299484491348267,
      "learning_rate": 4.947685600247209e-05,
      "loss": 1.9158,
      "step": 187681
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.4129244089126587,
      "learning_rate": 4.947460567311097e-05,
      "loss": 1.7943,
      "step": 187682
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.3430668115615845,
      "learning_rate": 4.947235539032712e-05,
      "loss": 2.1752,
      "step": 187683
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.5308564901351929,
      "learning_rate": 4.947010515412112e-05,
      "loss": 2.0556,
      "step": 187684
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.3728229999542236,
      "learning_rate": 4.9467854964493326e-05,
      "loss": 1.8887,
      "step": 187685
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.4740978479385376,
      "learning_rate": 4.946560482144424e-05,
      "loss": 2.1733,
      "step": 187686
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.3421751260757446,
      "learning_rate": 4.946335472497412e-05,
      "loss": 2.0339,
      "step": 187687
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.3481539487838745,
      "learning_rate": 4.9461104675083684e-05,
      "loss": 1.9827,
      "step": 187688
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.3065528869628906,
      "learning_rate": 4.945885467177298e-05,
      "loss": 1.9418,
      "step": 187689
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.372972846031189,
      "learning_rate": 4.9456604715042725e-05,
      "loss": 1.8857,
      "step": 187690
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.3113895654678345,
      "learning_rate": 4.9454354804893105e-05,
      "loss": 1.904,
      "step": 187691
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.356575846672058,
      "learning_rate": 4.945210494132472e-05,
      "loss": 1.8842,
      "step": 187692
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.4425196647644043,
      "learning_rate": 4.9449855124337876e-05,
      "loss": 2.0533,
      "step": 187693
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.4891575574874878,
      "learning_rate": 4.9447605353933074e-05,
      "loss": 1.9226,
      "step": 187694
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.3376768827438354,
      "learning_rate": 4.9445355630110706e-05,
      "loss": 2.0188,
      "step": 187695
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.3244038820266724,
      "learning_rate": 4.944310595287118e-05,
      "loss": 2.122,
      "step": 187696
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.3784751892089844,
      "learning_rate": 4.944085632221482e-05,
      "loss": 2.0088,
      "step": 187697
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.4053728580474854,
      "learning_rate": 4.943860673814223e-05,
      "loss": 2.0219,
      "step": 187698
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.365552306175232,
      "learning_rate": 4.9436357200653644e-05,
      "loss": 1.8641,
      "step": 187699
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.2106728553771973,
      "learning_rate": 4.943410770974966e-05,
      "loss": 1.8985,
      "step": 187700
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.4813721179962158,
      "learning_rate": 4.9431858265430576e-05,
      "loss": 2.1993,
      "step": 187701
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.7631049156188965,
      "learning_rate": 4.9429608867696804e-05,
      "loss": 2.0974,
      "step": 187702
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.4706244468688965,
      "learning_rate": 4.942735951654883e-05,
      "loss": 2.2303,
      "step": 187703
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.366002082824707,
      "learning_rate": 4.9425110211987054e-05,
      "loss": 1.795,
      "step": 187704
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.3785098791122437,
      "learning_rate": 4.9422860954011814e-05,
      "loss": 2.1322,
      "step": 187705
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.4928984642028809,
      "learning_rate": 4.942061174262367e-05,
      "loss": 2.0146,
      "step": 187706
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.5963619947433472,
      "learning_rate": 4.941836257782297e-05,
      "loss": 1.8263,
      "step": 187707
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.461693286895752,
      "learning_rate": 4.941611345961003e-05,
      "loss": 2.0353,
      "step": 187708
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.2307581901550293,
      "learning_rate": 4.941386438798546e-05,
      "loss": 2.1851,
      "step": 187709
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.5221312046051025,
      "learning_rate": 4.941161536294955e-05,
      "loss": 1.8458,
      "step": 187710
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.5130566358566284,
      "learning_rate": 4.9409366384502714e-05,
      "loss": 1.9819,
      "step": 187711
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.506246566772461,
      "learning_rate": 4.940711745264547e-05,
      "loss": 1.923,
      "step": 187712
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.2837107181549072,
      "learning_rate": 4.9404868567378167e-05,
      "loss": 1.9034,
      "step": 187713
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.411816954612732,
      "learning_rate": 4.9402619728701156e-05,
      "loss": 1.9751,
      "step": 187714
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.3419197797775269,
      "learning_rate": 4.940037093661501e-05,
      "loss": 1.9434,
      "step": 187715
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.4722193479537964,
      "learning_rate": 4.939812219111999e-05,
      "loss": 1.8705,
      "step": 187716
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.3823802471160889,
      "learning_rate": 4.939587349221668e-05,
      "loss": 1.9797,
      "step": 187717
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.5813391208648682,
      "learning_rate": 4.9393624839905355e-05,
      "loss": 1.9036,
      "step": 187718
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.3540006875991821,
      "learning_rate": 4.939137623418653e-05,
      "loss": 1.9156,
      "step": 187719
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.3729979991912842,
      "learning_rate": 4.9389127675060504e-05,
      "loss": 2.0586,
      "step": 187720
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.3102236986160278,
      "learning_rate": 4.938687916252784e-05,
      "loss": 1.8219,
      "step": 187721
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.4994782209396362,
      "learning_rate": 4.93846306965888e-05,
      "loss": 2.1145,
      "step": 187722
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.3249733448028564,
      "learning_rate": 4.938238227724396e-05,
      "loss": 1.9533,
      "step": 187723
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.3041030168533325,
      "learning_rate": 4.938013390449358e-05,
      "loss": 2.0476,
      "step": 187724
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.3537272214889526,
      "learning_rate": 4.937788557833836e-05,
      "loss": 2.0571,
      "step": 187725
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.2461588382720947,
      "learning_rate": 4.937563729877834e-05,
      "loss": 1.8874,
      "step": 187726
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.3711661100387573,
      "learning_rate": 4.93733890658142e-05,
      "loss": 2.0007,
      "step": 187727
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.3191331624984741,
      "learning_rate": 4.93711408794462e-05,
      "loss": 1.8807,
      "step": 187728
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.5400547981262207,
      "learning_rate": 4.9368892739674895e-05,
      "loss": 2.0195,
      "step": 187729
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.5229662656784058,
      "learning_rate": 4.936664464650061e-05,
      "loss": 2.106,
      "step": 187730
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.4372080564498901,
      "learning_rate": 4.936439659992392e-05,
      "loss": 1.9504,
      "step": 187731
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.4213486909866333,
      "learning_rate": 4.936214859994495e-05,
      "loss": 2.1503,
      "step": 187732
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.4213957786560059,
      "learning_rate": 4.935990064656441e-05,
      "loss": 1.8743,
      "step": 187733
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.4115036725997925,
      "learning_rate": 4.9357652739782495e-05,
      "loss": 2.0403,
      "step": 187734
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.6256600618362427,
      "learning_rate": 4.93554048795998e-05,
      "loss": 1.9289,
      "step": 187735
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.376335859298706,
      "learning_rate": 4.935315706601657e-05,
      "loss": 2.1623,
      "step": 187736
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.3801053762435913,
      "learning_rate": 4.935090929903349e-05,
      "loss": 1.9421,
      "step": 187737
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.4583203792572021,
      "learning_rate": 4.934866157865067e-05,
      "loss": 1.9873,
      "step": 187738
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.4125224351882935,
      "learning_rate": 4.9346413904868744e-05,
      "loss": 2.1492,
      "step": 187739
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.498140811920166,
      "learning_rate": 4.9344166277687935e-05,
      "loss": 2.0721,
      "step": 187740
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.5455584526062012,
      "learning_rate": 4.934191869710889e-05,
      "loss": 2.0991,
      "step": 187741
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.3402444124221802,
      "learning_rate": 4.933967116313182e-05,
      "loss": 1.9449,
      "step": 187742
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.483116865158081,
      "learning_rate": 4.933742367575741e-05,
      "loss": 1.9863,
      "step": 187743
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.382434368133545,
      "learning_rate": 4.9335176234985726e-05,
      "loss": 2.2735,
      "step": 187744
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.349833607673645,
      "learning_rate": 4.933292884081747e-05,
      "loss": 1.7949,
      "step": 187745
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.4436229467391968,
      "learning_rate": 4.933068149325285e-05,
      "loss": 2.1128,
      "step": 187746
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.3773633241653442,
      "learning_rate": 4.9328434192292464e-05,
      "loss": 2.1495,
      "step": 187747
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.4149389266967773,
      "learning_rate": 4.9326186937936595e-05,
      "loss": 2.0384,
      "step": 187748
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.520833969116211,
      "learning_rate": 4.932393973018588e-05,
      "loss": 1.8613,
      "step": 187749
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.3481193780899048,
      "learning_rate": 4.9321692569040384e-05,
      "loss": 1.9453,
      "step": 187750
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.25544273853302,
      "learning_rate": 4.931944545450084e-05,
      "loss": 2.0926,
      "step": 187751
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.6364545822143555,
      "learning_rate": 4.9317198386567444e-05,
      "loss": 1.976,
      "step": 187752
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.5665658712387085,
      "learning_rate": 4.9314951365240773e-05,
      "loss": 2.175,
      "step": 187753
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.4439592361450195,
      "learning_rate": 4.931270439052115e-05,
      "loss": 2.08,
      "step": 187754
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.370883584022522,
      "learning_rate": 4.931045746240915e-05,
      "loss": 1.7918,
      "step": 187755
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.343998908996582,
      "learning_rate": 4.9308210580904896e-05,
      "loss": 1.8225,
      "step": 187756
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.3433659076690674,
      "learning_rate": 4.9305963746009094e-05,
      "loss": 1.9936,
      "step": 187757
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.4336854219436646,
      "learning_rate": 4.9303716957721936e-05,
      "loss": 2.1842,
      "step": 187758
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.3894692659378052,
      "learning_rate": 4.930147021604407e-05,
      "loss": 2.0935,
      "step": 187759
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.3539639711380005,
      "learning_rate": 4.9299223520975685e-05,
      "loss": 1.9834,
      "step": 187760
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.490580677986145,
      "learning_rate": 4.9296976872517445e-05,
      "loss": 1.9592,
      "step": 187761
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.3275502920150757,
      "learning_rate": 4.9294730270669524e-05,
      "loss": 1.9986,
      "step": 187762
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.3325483798980713,
      "learning_rate": 4.9292483715432484e-05,
      "loss": 1.9723,
      "step": 187763
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.4371604919433594,
      "learning_rate": 4.9290237206806624e-05,
      "loss": 2.1387,
      "step": 187764
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.3636428117752075,
      "learning_rate": 4.928799074479255e-05,
      "loss": 2.0934,
      "step": 187765
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.3548545837402344,
      "learning_rate": 4.928574432939049e-05,
      "loss": 2.1617,
      "step": 187766
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.3892076015472412,
      "learning_rate": 4.9283497960601006e-05,
      "loss": 2.2012,
      "step": 187767
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.2243859767913818,
      "learning_rate": 4.928125163842444e-05,
      "loss": 1.7427,
      "step": 187768
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.4950010776519775,
      "learning_rate": 4.927900536286121e-05,
      "loss": 2.1435,
      "step": 187769
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.3343048095703125,
      "learning_rate": 4.9276759133911704e-05,
      "loss": 2.2136,
      "step": 187770
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.4041411876678467,
      "learning_rate": 4.927451295157644e-05,
      "loss": 2.0426,
      "step": 187771
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.25835120677948,
      "learning_rate": 4.9272266815855724e-05,
      "loss": 1.7476,
      "step": 187772
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.3190836906433105,
      "learning_rate": 4.927002072675008e-05,
      "loss": 1.9941,
      "step": 187773
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.3642791509628296,
      "learning_rate": 4.9267774684259885e-05,
      "loss": 2.1171,
      "step": 187774
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.3437328338623047,
      "learning_rate": 4.9265528688385534e-05,
      "loss": 2.0259,
      "step": 187775
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.301199197769165,
      "learning_rate": 4.9263282739127364e-05,
      "loss": 2.0301,
      "step": 187776
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.427266001701355,
      "learning_rate": 4.9261036836486e-05,
      "loss": 1.9833,
      "step": 187777
    },
    {
      "epoch": 2.44,
      "grad_norm": 1.4275864362716675,
      "learning_rate": 4.925879098046162e-05,
      "loss": 1.8285,
      "step": 187778
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.4793603420257568,
      "learning_rate": 4.925654517105487e-05,
      "loss": 1.9051,
      "step": 187779
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.4725462198257446,
      "learning_rate": 4.925429940826604e-05,
      "loss": 2.0045,
      "step": 187780
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.349141001701355,
      "learning_rate": 4.9252053692095585e-05,
      "loss": 2.0363,
      "step": 187781
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.4408856630325317,
      "learning_rate": 4.9249808022543835e-05,
      "loss": 2.1633,
      "step": 187782
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.5190526247024536,
      "learning_rate": 4.924756239961136e-05,
      "loss": 2.0941,
      "step": 187783
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.387448787689209,
      "learning_rate": 4.92453168232984e-05,
      "loss": 1.8441,
      "step": 187784
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.3896781206130981,
      "learning_rate": 4.924307129360554e-05,
      "loss": 1.8218,
      "step": 187785
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.3468961715698242,
      "learning_rate": 4.924082581053316e-05,
      "loss": 1.9759,
      "step": 187786
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.357017993927002,
      "learning_rate": 4.923858037408155e-05,
      "loss": 1.9935,
      "step": 187787
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.5190634727478027,
      "learning_rate": 4.923633498425128e-05,
      "loss": 2.1524,
      "step": 187788
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.3292139768600464,
      "learning_rate": 4.923408964104272e-05,
      "loss": 1.9884,
      "step": 187789
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.335763931274414,
      "learning_rate": 4.923184434445619e-05,
      "loss": 2.063,
      "step": 187790
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.4070932865142822,
      "learning_rate": 4.922959909449231e-05,
      "loss": 2.0058,
      "step": 187791
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.6479853391647339,
      "learning_rate": 4.9227353891151336e-05,
      "loss": 2.0809,
      "step": 187792
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.4278572797775269,
      "learning_rate": 4.922510873443369e-05,
      "loss": 2.0558,
      "step": 187793
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.3678996562957764,
      "learning_rate": 4.922286362433989e-05,
      "loss": 1.9512,
      "step": 187794
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.3829855918884277,
      "learning_rate": 4.9220618560870296e-05,
      "loss": 2.0171,
      "step": 187795
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.3136523962020874,
      "learning_rate": 4.9218373544025236e-05,
      "loss": 2.1311,
      "step": 187796
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.3643380403518677,
      "learning_rate": 4.921612857380528e-05,
      "loss": 1.9909,
      "step": 187797
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.4176496267318726,
      "learning_rate": 4.9213883650210795e-05,
      "loss": 2.039,
      "step": 187798
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.4777977466583252,
      "learning_rate": 4.921163877324211e-05,
      "loss": 1.9328,
      "step": 187799
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.3847864866256714,
      "learning_rate": 4.92093939428998e-05,
      "loss": 1.9321,
      "step": 187800
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.3236711025238037,
      "learning_rate": 4.920714915918412e-05,
      "loss": 1.8805,
      "step": 187801
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.3808037042617798,
      "learning_rate": 4.9204904422095645e-05,
      "loss": 2.0089,
      "step": 187802
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.4004747867584229,
      "learning_rate": 4.920265973163467e-05,
      "loss": 1.9256,
      "step": 187803
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.3687769174575806,
      "learning_rate": 4.9200415087801706e-05,
      "loss": 1.7846,
      "step": 187804
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.4723851680755615,
      "learning_rate": 4.9198170490597035e-05,
      "loss": 1.927,
      "step": 187805
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.4209823608398438,
      "learning_rate": 4.9195925940021196e-05,
      "loss": 2.0544,
      "step": 187806
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.2749204635620117,
      "learning_rate": 4.919368143607453e-05,
      "loss": 1.7628,
      "step": 187807
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.3400862216949463,
      "learning_rate": 4.919143697875756e-05,
      "loss": 1.9369,
      "step": 187808
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.4717154502868652,
      "learning_rate": 4.9189192568070627e-05,
      "loss": 2.0637,
      "step": 187809
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.4620881080627441,
      "learning_rate": 4.918694820401415e-05,
      "loss": 1.7113,
      "step": 187810
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.4358714818954468,
      "learning_rate": 4.918470388658852e-05,
      "loss": 2.0786,
      "step": 187811
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.4778425693511963,
      "learning_rate": 4.918245961579421e-05,
      "loss": 2.0299,
      "step": 187812
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.3732188940048218,
      "learning_rate": 4.918021539163157e-05,
      "loss": 2.0201,
      "step": 187813
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.6133675575256348,
      "learning_rate": 4.917797121410113e-05,
      "loss": 2.2111,
      "step": 187814
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.5243885517120361,
      "learning_rate": 4.9175727083203184e-05,
      "loss": 1.9915,
      "step": 187815
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.3559739589691162,
      "learning_rate": 4.917348299893834e-05,
      "loss": 1.8945,
      "step": 187816
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.2245234251022339,
      "learning_rate": 4.917123896130672e-05,
      "loss": 1.9713,
      "step": 187817
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.300936222076416,
      "learning_rate": 4.9168994970309004e-05,
      "loss": 2.0527,
      "step": 187818
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.4455634355545044,
      "learning_rate": 4.9166751025945416e-05,
      "loss": 2.0933,
      "step": 187819
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.388521432876587,
      "learning_rate": 4.916450712821652e-05,
      "loss": 1.8391,
      "step": 187820
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.533182144165039,
      "learning_rate": 4.9162263277122625e-05,
      "loss": 1.9158,
      "step": 187821
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.3032574653625488,
      "learning_rate": 4.916001947266432e-05,
      "loss": 1.9567,
      "step": 187822
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.3178142309188843,
      "learning_rate": 4.9157775714841784e-05,
      "loss": 1.8842,
      "step": 187823
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.405336856842041,
      "learning_rate": 4.915553200365564e-05,
      "loss": 1.9228,
      "step": 187824
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.3838833570480347,
      "learning_rate": 4.915328833910612e-05,
      "loss": 1.9283,
      "step": 187825
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.4978208541870117,
      "learning_rate": 4.91510447211938e-05,
      "loss": 2.0912,
      "step": 187826
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.3607555627822876,
      "learning_rate": 4.9148801149919006e-05,
      "loss": 1.9117,
      "step": 187827
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.3452814817428589,
      "learning_rate": 4.91465576252823e-05,
      "loss": 1.8685,
      "step": 187828
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.4476537704467773,
      "learning_rate": 4.9144314147283825e-05,
      "loss": 2.0552,
      "step": 187829
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.4906425476074219,
      "learning_rate": 4.9142070715924244e-05,
      "loss": 1.9302,
      "step": 187830
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.4701993465423584,
      "learning_rate": 4.9139827331203815e-05,
      "loss": 1.8286,
      "step": 187831
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.4835829734802246,
      "learning_rate": 4.913758399312312e-05,
      "loss": 1.9572,
      "step": 187832
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.3218919038772583,
      "learning_rate": 4.913534070168238e-05,
      "loss": 2.2688,
      "step": 187833
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.4967538118362427,
      "learning_rate": 4.9133097456882196e-05,
      "loss": 2.0061,
      "step": 187834
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.3621443510055542,
      "learning_rate": 4.913085425872291e-05,
      "loss": 2.2104,
      "step": 187835
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.3979912996292114,
      "learning_rate": 4.912861110720494e-05,
      "loss": 2.1179,
      "step": 187836
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.4744246006011963,
      "learning_rate": 4.912636800232863e-05,
      "loss": 2.0087,
      "step": 187837
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.427317500114441,
      "learning_rate": 4.912412494409451e-05,
      "loss": 2.1987,
      "step": 187838
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.2956136465072632,
      "learning_rate": 4.912188193250292e-05,
      "loss": 2.1455,
      "step": 187839
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.3599884510040283,
      "learning_rate": 4.9119638967554345e-05,
      "loss": 2.1412,
      "step": 187840
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.4194493293762207,
      "learning_rate": 4.911739604924916e-05,
      "loss": 2.0207,
      "step": 187841
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.4276173114776611,
      "learning_rate": 4.9115153177587794e-05,
      "loss": 2.1761,
      "step": 187842
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.4396233558654785,
      "learning_rate": 4.911291035257059e-05,
      "loss": 2.2421,
      "step": 187843
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.5247968435287476,
      "learning_rate": 4.9110667574198106e-05,
      "loss": 2.1804,
      "step": 187844
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.3954404592514038,
      "learning_rate": 4.910842484247061e-05,
      "loss": 1.9741,
      "step": 187845
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.4117976427078247,
      "learning_rate": 4.910618215738866e-05,
      "loss": 2.1287,
      "step": 187846
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.362888216972351,
      "learning_rate": 4.910393951895264e-05,
      "loss": 2.0743,
      "step": 187847
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.4349583387374878,
      "learning_rate": 4.91016969271629e-05,
      "loss": 2.0973,
      "step": 187848
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.5947918891906738,
      "learning_rate": 4.909945438201982e-05,
      "loss": 1.854,
      "step": 187849
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.4444907903671265,
      "learning_rate": 4.909721188352396e-05,
      "loss": 2.4013,
      "step": 187850
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.2968988418579102,
      "learning_rate": 4.9094969431675615e-05,
      "loss": 1.9274,
      "step": 187851
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.3428932428359985,
      "learning_rate": 4.909272702647529e-05,
      "loss": 1.8528,
      "step": 187852
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.4197896718978882,
      "learning_rate": 4.909048466792338e-05,
      "loss": 2.0651,
      "step": 187853
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.623397946357727,
      "learning_rate": 4.908824235602029e-05,
      "loss": 2.044,
      "step": 187854
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.3045014142990112,
      "learning_rate": 4.9086000090766354e-05,
      "loss": 2.0848,
      "step": 187855
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.3469243049621582,
      "learning_rate": 4.908375787216213e-05,
      "loss": 2.1229,
      "step": 187856
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.379990816116333,
      "learning_rate": 4.908151570020793e-05,
      "loss": 2.0263,
      "step": 187857
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.5039246082305908,
      "learning_rate": 4.907927357490424e-05,
      "loss": 1.9965,
      "step": 187858
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.4307750463485718,
      "learning_rate": 4.907703149625147e-05,
      "loss": 2.061,
      "step": 187859
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.3527888059616089,
      "learning_rate": 4.9074789464250054e-05,
      "loss": 1.981,
      "step": 187860
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.6470451354980469,
      "learning_rate": 4.907254747890025e-05,
      "loss": 1.8187,
      "step": 187861
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.3272823095321655,
      "learning_rate": 4.907030554020269e-05,
      "loss": 1.9397,
      "step": 187862
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.3756163120269775,
      "learning_rate": 4.906806364815762e-05,
      "loss": 2.009,
      "step": 187863
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.3664201498031616,
      "learning_rate": 4.9065821802765594e-05,
      "loss": 1.9901,
      "step": 187864
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.3672691583633423,
      "learning_rate": 4.906358000402698e-05,
      "loss": 1.7227,
      "step": 187865
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.3450502157211304,
      "learning_rate": 4.9061338251942186e-05,
      "loss": 2.1644,
      "step": 187866
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.3661909103393555,
      "learning_rate": 4.9059096546511534e-05,
      "loss": 2.1061,
      "step": 187867
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.3368852138519287,
      "learning_rate": 4.9056854887735634e-05,
      "loss": 2.0566,
      "step": 187868
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.5270265340805054,
      "learning_rate": 4.905461327561471e-05,
      "loss": 2.0305,
      "step": 187869
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.3086127042770386,
      "learning_rate": 4.9052371710149366e-05,
      "loss": 1.9294,
      "step": 187870
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.434556007385254,
      "learning_rate": 4.90501301913399e-05,
      "loss": 2.0942,
      "step": 187871
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.3976870775222778,
      "learning_rate": 4.904788871918674e-05,
      "loss": 1.9815,
      "step": 187872
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.6565138101577759,
      "learning_rate": 4.904564729369027e-05,
      "loss": 2.0061,
      "step": 187873
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.3426752090454102,
      "learning_rate": 4.9043405914851005e-05,
      "loss": 1.7805,
      "step": 187874
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.382019281387329,
      "learning_rate": 4.904116458266926e-05,
      "loss": 1.9617,
      "step": 187875
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.275980830192566,
      "learning_rate": 4.9038923297145547e-05,
      "loss": 1.8458,
      "step": 187876
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.3639039993286133,
      "learning_rate": 4.903668205828025e-05,
      "loss": 1.9155,
      "step": 187877
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.2845814228057861,
      "learning_rate": 4.90344408660737e-05,
      "loss": 1.8557,
      "step": 187878
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.5081892013549805,
      "learning_rate": 4.9032199720526454e-05,
      "loss": 2.219,
      "step": 187879
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.323451280593872,
      "learning_rate": 4.902995862163886e-05,
      "loss": 1.6917,
      "step": 187880
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.7104722261428833,
      "learning_rate": 4.902771756941123e-05,
      "loss": 2.044,
      "step": 187881
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.609148383140564,
      "learning_rate": 4.902547656384418e-05,
      "loss": 1.978,
      "step": 187882
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.4433788061141968,
      "learning_rate": 4.902323560493804e-05,
      "loss": 2.1039,
      "step": 187883
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.2704156637191772,
      "learning_rate": 4.9020994692693137e-05,
      "loss": 2.0093,
      "step": 187884
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.4120780229568481,
      "learning_rate": 4.9018753827110045e-05,
      "loss": 1.953,
      "step": 187885
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.6371463537216187,
      "learning_rate": 4.901651300818906e-05,
      "loss": 2.0679,
      "step": 187886
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.6422429084777832,
      "learning_rate": 4.9014272235930686e-05,
      "loss": 2.1367,
      "step": 187887
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.738511562347412,
      "learning_rate": 4.901203151033528e-05,
      "loss": 1.841,
      "step": 187888
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.4122086763381958,
      "learning_rate": 4.900979083140332e-05,
      "loss": 2.0394,
      "step": 187889
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.4198777675628662,
      "learning_rate": 4.900755019913506e-05,
      "loss": 2.1275,
      "step": 187890
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.394288420677185,
      "learning_rate": 4.900530961353114e-05,
      "loss": 1.9863,
      "step": 187891
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.546596884727478,
      "learning_rate": 4.900306907459176e-05,
      "loss": 1.9851,
      "step": 187892
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.502223253250122,
      "learning_rate": 4.900082858231755e-05,
      "loss": 1.9231,
      "step": 187893
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.4952921867370605,
      "learning_rate": 4.8998588136708816e-05,
      "loss": 2.0695,
      "step": 187894
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.4312580823898315,
      "learning_rate": 4.899634773776599e-05,
      "loss": 1.8744,
      "step": 187895
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.3493006229400635,
      "learning_rate": 4.899410738548942e-05,
      "loss": 2.0864,
      "step": 187896
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.366825819015503,
      "learning_rate": 4.899186707987963e-05,
      "loss": 2.1241,
      "step": 187897
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.2797077894210815,
      "learning_rate": 4.898962682093691e-05,
      "loss": 1.7541,
      "step": 187898
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.4881540536880493,
      "learning_rate": 4.898738660866186e-05,
      "loss": 2.0636,
      "step": 187899
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.8232911825180054,
      "learning_rate": 4.898514644305471e-05,
      "loss": 2.0775,
      "step": 187900
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.316946268081665,
      "learning_rate": 4.898290632411603e-05,
      "loss": 2.0114,
      "step": 187901
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.315394401550293,
      "learning_rate": 4.8980666251846155e-05,
      "loss": 2.0478,
      "step": 187902
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.4358313083648682,
      "learning_rate": 4.897842622624551e-05,
      "loss": 1.9478,
      "step": 187903
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.3030985593795776,
      "learning_rate": 4.897618624731447e-05,
      "loss": 2.032,
      "step": 187904
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.456168532371521,
      "learning_rate": 4.897394631505353e-05,
      "loss": 2.1844,
      "step": 187905
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.4896084070205688,
      "learning_rate": 4.897170642946303e-05,
      "loss": 2.1255,
      "step": 187906
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.3495792150497437,
      "learning_rate": 4.8969466590543496e-05,
      "loss": 2.0741,
      "step": 187907
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.3744627237319946,
      "learning_rate": 4.896722679829529e-05,
      "loss": 2.0254,
      "step": 187908
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.5181663036346436,
      "learning_rate": 4.896498705271878e-05,
      "loss": 1.8999,
      "step": 187909
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.197770357131958,
      "learning_rate": 4.896274735381438e-05,
      "loss": 2.1214,
      "step": 187910
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.4356082677841187,
      "learning_rate": 4.8960507701582576e-05,
      "loss": 2.2531,
      "step": 187911
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.6284767389297485,
      "learning_rate": 4.89582680960237e-05,
      "loss": 2.0713,
      "step": 187912
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.5051801204681396,
      "learning_rate": 4.895602853713833e-05,
      "loss": 1.9174,
      "step": 187913
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.3853410482406616,
      "learning_rate": 4.895378902492672e-05,
      "loss": 1.9122,
      "step": 187914
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.3254677057266235,
      "learning_rate": 4.895154955938937e-05,
      "loss": 1.7363,
      "step": 187915
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.2745389938354492,
      "learning_rate": 4.894931014052656e-05,
      "loss": 1.8247,
      "step": 187916
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.3813886642456055,
      "learning_rate": 4.89470707683389e-05,
      "loss": 1.9009,
      "step": 187917
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.483660101890564,
      "learning_rate": 4.894483144282668e-05,
      "loss": 2.0813,
      "step": 187918
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.322298526763916,
      "learning_rate": 4.894259216399039e-05,
      "loss": 2.1253,
      "step": 187919
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.5362690687179565,
      "learning_rate": 4.894035293183043e-05,
      "loss": 1.9169,
      "step": 187920
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.4798599481582642,
      "learning_rate": 4.8938113746347165e-05,
      "loss": 2.1191,
      "step": 187921
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.3608198165893555,
      "learning_rate": 4.893587460754099e-05,
      "loss": 2.1107,
      "step": 187922
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.4767789840698242,
      "learning_rate": 4.8933635515412485e-05,
      "loss": 2.169,
      "step": 187923
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.2967816591262817,
      "learning_rate": 4.893139646996184e-05,
      "loss": 1.8775,
      "step": 187924
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.791115641593933,
      "learning_rate": 4.8929157471189685e-05,
      "loss": 1.8881,
      "step": 187925
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.481558918952942,
      "learning_rate": 4.8926918519096325e-05,
      "loss": 1.9667,
      "step": 187926
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.3261287212371826,
      "learning_rate": 4.892467961368216e-05,
      "loss": 2.1231,
      "step": 187927
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.3410425186157227,
      "learning_rate": 4.8922440754947625e-05,
      "loss": 2.1006,
      "step": 187928
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.3166429996490479,
      "learning_rate": 4.892020194289318e-05,
      "loss": 1.8643,
      "step": 187929
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.2383980751037598,
      "learning_rate": 4.8917963177519126e-05,
      "loss": 1.8691,
      "step": 187930
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.5220544338226318,
      "learning_rate": 4.891572445882607e-05,
      "loss": 1.9317,
      "step": 187931
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.482255220413208,
      "learning_rate": 4.8913485786814296e-05,
      "loss": 2.3054,
      "step": 187932
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.34806489944458,
      "learning_rate": 4.891124716148426e-05,
      "loss": 1.9135,
      "step": 187933
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.4439144134521484,
      "learning_rate": 4.8909008582836274e-05,
      "loss": 2.0675,
      "step": 187934
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.5531768798828125,
      "learning_rate": 4.8906770050870944e-05,
      "loss": 2.1235,
      "step": 187935
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.2904939651489258,
      "learning_rate": 4.890453156558848e-05,
      "loss": 1.9183,
      "step": 187936
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.448376178741455,
      "learning_rate": 4.89022931269895e-05,
      "loss": 2.0679,
      "step": 187937
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.350096344947815,
      "learning_rate": 4.890005473507431e-05,
      "loss": 1.8637,
      "step": 187938
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.2455257177352905,
      "learning_rate": 4.8897816389843345e-05,
      "loss": 2.0071,
      "step": 187939
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.3495694398880005,
      "learning_rate": 4.889557809129694e-05,
      "loss": 2.1449,
      "step": 187940
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.2595373392105103,
      "learning_rate": 4.889333983943565e-05,
      "loss": 2.0825,
      "step": 187941
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.3422473669052124,
      "learning_rate": 4.889110163425979e-05,
      "loss": 1.8306,
      "step": 187942
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.4112604856491089,
      "learning_rate": 4.8888863475769846e-05,
      "loss": 1.9134,
      "step": 187943
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.287194848060608,
      "learning_rate": 4.8886625363966224e-05,
      "loss": 2.0665,
      "step": 187944
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.4012904167175293,
      "learning_rate": 4.8884387298849324e-05,
      "loss": 2.3045,
      "step": 187945
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.2985481023788452,
      "learning_rate": 4.888214928041948e-05,
      "loss": 2.0163,
      "step": 187946
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.3294813632965088,
      "learning_rate": 4.8879911308677254e-05,
      "loss": 2.1027,
      "step": 187947
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.334432601928711,
      "learning_rate": 4.887767338362292e-05,
      "loss": 1.891,
      "step": 187948
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.3354438543319702,
      "learning_rate": 4.8875435505257036e-05,
      "loss": 2.0971,
      "step": 187949
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.3409104347229004,
      "learning_rate": 4.8873197673579936e-05,
      "loss": 1.8641,
      "step": 187950
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.3524179458618164,
      "learning_rate": 4.8870959888592085e-05,
      "loss": 1.9379,
      "step": 187951
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.3620860576629639,
      "learning_rate": 4.886872215029376e-05,
      "loss": 1.8702,
      "step": 187952
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.4383018016815186,
      "learning_rate": 4.886648445868555e-05,
      "loss": 1.9694,
      "step": 187953
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.3964440822601318,
      "learning_rate": 4.886424681376776e-05,
      "loss": 1.9232,
      "step": 187954
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.5053597688674927,
      "learning_rate": 4.886200921554089e-05,
      "loss": 1.8425,
      "step": 187955
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.1641929149627686,
      "learning_rate": 4.885977166400533e-05,
      "loss": 1.8869,
      "step": 187956
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.3697627782821655,
      "learning_rate": 4.885753415916146e-05,
      "loss": 1.8922,
      "step": 187957
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.3881484270095825,
      "learning_rate": 4.885529670100967e-05,
      "loss": 1.9174,
      "step": 187958
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.3902560472488403,
      "learning_rate": 4.8853059289550464e-05,
      "loss": 2.0551,
      "step": 187959
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.3760064840316772,
      "learning_rate": 4.885082192478414e-05,
      "loss": 2.1256,
      "step": 187960
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.4123892784118652,
      "learning_rate": 4.8848584606711304e-05,
      "loss": 1.8958,
      "step": 187961
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.3370811939239502,
      "learning_rate": 4.8846347335332214e-05,
      "loss": 1.8732,
      "step": 187962
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.277347445487976,
      "learning_rate": 4.884411011064727e-05,
      "loss": 2.0221,
      "step": 187963
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.3579163551330566,
      "learning_rate": 4.884187293265704e-05,
      "loss": 1.8499,
      "step": 187964
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.4708592891693115,
      "learning_rate": 4.8839635801361834e-05,
      "loss": 2.1097,
      "step": 187965
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.3527827262878418,
      "learning_rate": 4.8837398716762e-05,
      "loss": 1.7635,
      "step": 187966
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.5332551002502441,
      "learning_rate": 4.8835161678858116e-05,
      "loss": 2.1317,
      "step": 187967
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.427596926689148,
      "learning_rate": 4.883292468765044e-05,
      "loss": 1.8445,
      "step": 187968
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.346442699432373,
      "learning_rate": 4.8830687743139516e-05,
      "loss": 1.8479,
      "step": 187969
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.2874908447265625,
      "learning_rate": 4.882845084532574e-05,
      "loss": 1.878,
      "step": 187970
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.2694730758666992,
      "learning_rate": 4.8826213994209475e-05,
      "loss": 1.9717,
      "step": 187971
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.3984920978546143,
      "learning_rate": 4.8823977189791084e-05,
      "loss": 2.0738,
      "step": 187972
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.3987661600112915,
      "learning_rate": 4.8821740432071144e-05,
      "loss": 1.7964,
      "step": 187973
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.4161640405654907,
      "learning_rate": 4.881950372104991e-05,
      "loss": 1.9198,
      "step": 187974
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.3623759746551514,
      "learning_rate": 4.881726705672796e-05,
      "loss": 1.9264,
      "step": 187975
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.4333783388137817,
      "learning_rate": 4.881503043910559e-05,
      "loss": 1.8636,
      "step": 187976
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.4516956806182861,
      "learning_rate": 4.8812793868183186e-05,
      "loss": 2.2237,
      "step": 187977
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.4658222198486328,
      "learning_rate": 4.88105573439613e-05,
      "loss": 1.9914,
      "step": 187978
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.636805534362793,
      "learning_rate": 4.880832086644029e-05,
      "loss": 2.025,
      "step": 187979
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.2717771530151367,
      "learning_rate": 4.880608443562045e-05,
      "loss": 1.9725,
      "step": 187980
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.8531426191329956,
      "learning_rate": 4.8803848051502395e-05,
      "loss": 2.0695,
      "step": 187981
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.4754396677017212,
      "learning_rate": 4.8801611714086444e-05,
      "loss": 1.9684,
      "step": 187982
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.4839285612106323,
      "learning_rate": 4.8799375423372936e-05,
      "loss": 1.9162,
      "step": 187983
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.4341933727264404,
      "learning_rate": 4.879713917936243e-05,
      "loss": 1.8635,
      "step": 187984
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.4281727075576782,
      "learning_rate": 4.879490298205524e-05,
      "loss": 1.9169,
      "step": 187985
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.5104142427444458,
      "learning_rate": 4.8792666831451856e-05,
      "loss": 1.9578,
      "step": 187986
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.3760342597961426,
      "learning_rate": 4.879043072755268e-05,
      "loss": 1.9767,
      "step": 187987
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.4332528114318848,
      "learning_rate": 4.8788194670358075e-05,
      "loss": 1.9871,
      "step": 187988
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.3413496017456055,
      "learning_rate": 4.878595865986844e-05,
      "loss": 1.9321,
      "step": 187989
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.5315431356430054,
      "learning_rate": 4.878372269608428e-05,
      "loss": 2.0931,
      "step": 187990
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.3858716487884521,
      "learning_rate": 4.878148677900593e-05,
      "loss": 2.1235,
      "step": 187991
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.4794836044311523,
      "learning_rate": 4.877925090863392e-05,
      "loss": 1.8647,
      "step": 187992
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.3740347623825073,
      "learning_rate": 4.877701508496858e-05,
      "loss": 1.9189,
      "step": 187993
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.286229133605957,
      "learning_rate": 4.8774779308010306e-05,
      "loss": 1.6893,
      "step": 187994
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.3168096542358398,
      "learning_rate": 4.8772543577759505e-05,
      "loss": 2.2374,
      "step": 187995
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.3717683553695679,
      "learning_rate": 4.8770307894216676e-05,
      "loss": 1.9446,
      "step": 187996
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.5229425430297852,
      "learning_rate": 4.876807225738215e-05,
      "loss": 2.1776,
      "step": 187997
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.639754056930542,
      "learning_rate": 4.876583666725643e-05,
      "loss": 2.0732,
      "step": 187998
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.4220576286315918,
      "learning_rate": 4.876360112383988e-05,
      "loss": 2.0306,
      "step": 187999
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.4514635801315308,
      "learning_rate": 4.876136562713293e-05,
      "loss": 2.0752,
      "step": 188000
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.349391222000122,
      "learning_rate": 4.875913017713592e-05,
      "loss": 2.0311,
      "step": 188001
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.4202462434768677,
      "learning_rate": 4.875689477384937e-05,
      "loss": 1.8863,
      "step": 188002
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.3064403533935547,
      "learning_rate": 4.8754659417273635e-05,
      "loss": 2.0677,
      "step": 188003
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.4264543056488037,
      "learning_rate": 4.8752424107409195e-05,
      "loss": 1.9014,
      "step": 188004
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.3972008228302002,
      "learning_rate": 4.875018884425642e-05,
      "loss": 2.0227,
      "step": 188005
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.3796066045761108,
      "learning_rate": 4.874795362781572e-05,
      "loss": 1.944,
      "step": 188006
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.36611807346344,
      "learning_rate": 4.8745718458087445e-05,
      "loss": 1.9688,
      "step": 188007
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.4505747556686401,
      "learning_rate": 4.874348333507218e-05,
      "loss": 1.8691,
      "step": 188008
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.3937008380889893,
      "learning_rate": 4.874124825877017e-05,
      "loss": 1.9925,
      "step": 188009
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.3607732057571411,
      "learning_rate": 4.8739013229181966e-05,
      "loss": 2.0256,
      "step": 188010
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.3766279220581055,
      "learning_rate": 4.8736778246307895e-05,
      "loss": 2.1596,
      "step": 188011
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.3805830478668213,
      "learning_rate": 4.8734543310148456e-05,
      "loss": 2.1528,
      "step": 188012
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.4934273958206177,
      "learning_rate": 4.873230842070388e-05,
      "loss": 1.7637,
      "step": 188013
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.6300283670425415,
      "learning_rate": 4.873007357797484e-05,
      "loss": 2.2491,
      "step": 188014
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.2370774745941162,
      "learning_rate": 4.872783878196149e-05,
      "loss": 1.8995,
      "step": 188015
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.4981762170791626,
      "learning_rate": 4.872560403266451e-05,
      "loss": 2.1415,
      "step": 188016
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.2519667148590088,
      "learning_rate": 4.872336933008413e-05,
      "loss": 1.6741,
      "step": 188017
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.6516907215118408,
      "learning_rate": 4.872113467422084e-05,
      "loss": 2.218,
      "step": 188018
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.4352644681930542,
      "learning_rate": 4.8718900065074985e-05,
      "loss": 1.751,
      "step": 188019
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.4648300409317017,
      "learning_rate": 4.871666550264709e-05,
      "loss": 2.0336,
      "step": 188020
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.3590457439422607,
      "learning_rate": 4.8714430986937424e-05,
      "loss": 2.0748,
      "step": 188021
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.2120282649993896,
      "learning_rate": 4.871219651794655e-05,
      "loss": 1.9668,
      "step": 188022
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.399243950843811,
      "learning_rate": 4.870996209567485e-05,
      "loss": 1.9848,
      "step": 188023
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.3986096382141113,
      "learning_rate": 4.87077277201227e-05,
      "loss": 2.1147,
      "step": 188024
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.328478217124939,
      "learning_rate": 4.870549339129045e-05,
      "loss": 2.1008,
      "step": 188025
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.458478569984436,
      "learning_rate": 4.8703259109178664e-05,
      "loss": 2.0949,
      "step": 188026
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.4411715269088745,
      "learning_rate": 4.8701024873787634e-05,
      "loss": 1.9937,
      "step": 188027
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.2808074951171875,
      "learning_rate": 4.869879068511789e-05,
      "loss": 2.057,
      "step": 188028
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.1802223920822144,
      "learning_rate": 4.869655654316979e-05,
      "loss": 1.7199,
      "step": 188029
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.356658697128296,
      "learning_rate": 4.8694322447943704e-05,
      "loss": 1.8893,
      "step": 188030
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.4343892335891724,
      "learning_rate": 4.8692088399440045e-05,
      "loss": 1.9543,
      "step": 188031
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.3337676525115967,
      "learning_rate": 4.8689854397659347e-05,
      "loss": 1.823,
      "step": 188032
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.3978136777877808,
      "learning_rate": 4.8687620442601873e-05,
      "loss": 1.8791,
      "step": 188033
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.2971677780151367,
      "learning_rate": 4.868538653426819e-05,
      "loss": 1.9111,
      "step": 188034
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.3481281995773315,
      "learning_rate": 4.868315267265857e-05,
      "loss": 2.2311,
      "step": 188035
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.4093321561813354,
      "learning_rate": 4.868091885777363e-05,
      "loss": 1.9041,
      "step": 188036
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.5429807901382446,
      "learning_rate": 4.867868508961349e-05,
      "loss": 1.9947,
      "step": 188037
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.375529408454895,
      "learning_rate": 4.86764513681788e-05,
      "loss": 2.042,
      "step": 188038
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.229056715965271,
      "learning_rate": 4.8674217693469874e-05,
      "loss": 1.9897,
      "step": 188039
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.5191432237625122,
      "learning_rate": 4.86719840654872e-05,
      "loss": 2.2314,
      "step": 188040
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.3479747772216797,
      "learning_rate": 4.866975048423105e-05,
      "loss": 1.9254,
      "step": 188041
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.3807134628295898,
      "learning_rate": 4.866751694970212e-05,
      "loss": 2.0499,
      "step": 188042
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.4074589014053345,
      "learning_rate": 4.8665283461900486e-05,
      "loss": 1.9687,
      "step": 188043
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.4477770328521729,
      "learning_rate": 4.86630500208268e-05,
      "loss": 1.9564,
      "step": 188044
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.346186876296997,
      "learning_rate": 4.86608166264813e-05,
      "loss": 2.0741,
      "step": 188045
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.3887134790420532,
      "learning_rate": 4.8658583278864594e-05,
      "loss": 2.0105,
      "step": 188046
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.3307589292526245,
      "learning_rate": 4.865634997797694e-05,
      "loss": 1.9887,
      "step": 188047
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.3735665082931519,
      "learning_rate": 4.865411672381887e-05,
      "loss": 2.0583,
      "step": 188048
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.3333951234817505,
      "learning_rate": 4.8651883516390724e-05,
      "loss": 1.9899,
      "step": 188049
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.4758957624435425,
      "learning_rate": 4.864965035569296e-05,
      "loss": 1.9075,
      "step": 188050
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.3822295665740967,
      "learning_rate": 4.8647417241725914e-05,
      "loss": 1.9401,
      "step": 188051
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.3416991233825684,
      "learning_rate": 4.864518417449009e-05,
      "loss": 1.937,
      "step": 188052
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.4390361309051514,
      "learning_rate": 4.864295115398582e-05,
      "loss": 1.9105,
      "step": 188053
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.3362537622451782,
      "learning_rate": 4.8640718180213665e-05,
      "loss": 1.8324,
      "step": 188054
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.3602840900421143,
      "learning_rate": 4.863848525317393e-05,
      "loss": 1.8163,
      "step": 188055
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.2260363101959229,
      "learning_rate": 4.863625237286701e-05,
      "loss": 1.9315,
      "step": 188056
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.3379813432693481,
      "learning_rate": 4.863401953929334e-05,
      "loss": 2.0099,
      "step": 188057
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.4060503244400024,
      "learning_rate": 4.8631786752453397e-05,
      "loss": 1.8964,
      "step": 188058
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.4066053628921509,
      "learning_rate": 4.862955401234746e-05,
      "loss": 2.0296,
      "step": 188059
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.3063796758651733,
      "learning_rate": 4.862732131897614e-05,
      "loss": 2.0158,
      "step": 188060
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.703682780265808,
      "learning_rate": 4.862508867233971e-05,
      "loss": 2.0398,
      "step": 188061
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.4850492477416992,
      "learning_rate": 4.862285607243859e-05,
      "loss": 2.0168,
      "step": 188062
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.5438027381896973,
      "learning_rate": 4.8620623519273246e-05,
      "loss": 1.7972,
      "step": 188063
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.4249078035354614,
      "learning_rate": 4.861839101284413e-05,
      "loss": 2.1546,
      "step": 188064
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.550553560256958,
      "learning_rate": 4.861615855315149e-05,
      "loss": 1.904,
      "step": 188065
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.3931751251220703,
      "learning_rate": 4.861392614019592e-05,
      "loss": 2.1416,
      "step": 188066
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.3293508291244507,
      "learning_rate": 4.861169377397778e-05,
      "loss": 2.2398,
      "step": 188067
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.3403109312057495,
      "learning_rate": 4.860946145449741e-05,
      "loss": 2.1066,
      "step": 188068
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.3452640771865845,
      "learning_rate": 4.860722918175536e-05,
      "loss": 1.9136,
      "step": 188069
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.3378918170928955,
      "learning_rate": 4.8604996955751885e-05,
      "loss": 2.065,
      "step": 188070
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.256639838218689,
      "learning_rate": 4.8602764776487554e-05,
      "loss": 1.9454,
      "step": 188071
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.3176771402359009,
      "learning_rate": 4.860053264396271e-05,
      "loss": 1.9707,
      "step": 188072
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.2745308876037598,
      "learning_rate": 4.8598300558177804e-05,
      "loss": 2.0736,
      "step": 188073
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.3794893026351929,
      "learning_rate": 4.859606851913312e-05,
      "loss": 1.9073,
      "step": 188074
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.486966609954834,
      "learning_rate": 4.8593836526829266e-05,
      "loss": 2.1616,
      "step": 188075
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.302536964416504,
      "learning_rate": 4.859160458126646e-05,
      "loss": 1.7557,
      "step": 188076
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.4542202949523926,
      "learning_rate": 4.858937268244531e-05,
      "loss": 2.0697,
      "step": 188077
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.3029394149780273,
      "learning_rate": 4.8587140830366155e-05,
      "loss": 1.959,
      "step": 188078
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.3834410905838013,
      "learning_rate": 4.8584909025029384e-05,
      "loss": 1.9035,
      "step": 188079
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.406498670578003,
      "learning_rate": 4.8582677266435354e-05,
      "loss": 2.0082,
      "step": 188080
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.4306436777114868,
      "learning_rate": 4.85804455545846e-05,
      "loss": 2.0565,
      "step": 188081
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.5680369138717651,
      "learning_rate": 4.857821388947746e-05,
      "loss": 2.0032,
      "step": 188082
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.4728184938430786,
      "learning_rate": 4.8575982271114424e-05,
      "loss": 2.1225,
      "step": 188083
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.4116477966308594,
      "learning_rate": 4.857375069949583e-05,
      "loss": 2.0605,
      "step": 188084
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.3360143899917603,
      "learning_rate": 4.857151917462218e-05,
      "loss": 1.9122,
      "step": 188085
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.4894047975540161,
      "learning_rate": 4.85692876964937e-05,
      "loss": 1.9663,
      "step": 188086
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.439426302909851,
      "learning_rate": 4.856705626511106e-05,
      "loss": 2.1582,
      "step": 188087
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.3879551887512207,
      "learning_rate": 4.856482488047443e-05,
      "loss": 2.0488,
      "step": 188088
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.3903121948242188,
      "learning_rate": 4.856259354258444e-05,
      "loss": 1.9767,
      "step": 188089
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.3922021389007568,
      "learning_rate": 4.856036225144142e-05,
      "loss": 1.9653,
      "step": 188090
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.516701579093933,
      "learning_rate": 4.855813100704577e-05,
      "loss": 1.9802,
      "step": 188091
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.4796786308288574,
      "learning_rate": 4.8555899809397834e-05,
      "loss": 2.0479,
      "step": 188092
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.44121515750885,
      "learning_rate": 4.8553668658498167e-05,
      "loss": 2.0762,
      "step": 188093
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.436743140220642,
      "learning_rate": 4.8551437554347074e-05,
      "loss": 2.0461,
      "step": 188094
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.4413104057312012,
      "learning_rate": 4.854920649694505e-05,
      "loss": 2.1709,
      "step": 188095
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.4559789896011353,
      "learning_rate": 4.85469754862925e-05,
      "loss": 2.097,
      "step": 188096
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.3937424421310425,
      "learning_rate": 4.854474452238981e-05,
      "loss": 1.9359,
      "step": 188097
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.4632349014282227,
      "learning_rate": 4.854251360523733e-05,
      "loss": 1.7815,
      "step": 188098
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.3023322820663452,
      "learning_rate": 4.854028273483559e-05,
      "loss": 1.8548,
      "step": 188099
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.335953712463379,
      "learning_rate": 4.853805191118491e-05,
      "loss": 2.0684,
      "step": 188100
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.3002737760543823,
      "learning_rate": 4.853582113428581e-05,
      "loss": 1.8346,
      "step": 188101
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.4298514127731323,
      "learning_rate": 4.8533590404138615e-05,
      "loss": 2.1097,
      "step": 188102
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.3511322736740112,
      "learning_rate": 4.853135972074388e-05,
      "loss": 1.8471,
      "step": 188103
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.6905901432037354,
      "learning_rate": 4.852912908410178e-05,
      "loss": 2.295,
      "step": 188104
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.4075947999954224,
      "learning_rate": 4.852689849421292e-05,
      "loss": 2.1422,
      "step": 188105
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.4865349531173706,
      "learning_rate": 4.852466795107759e-05,
      "loss": 2.0696,
      "step": 188106
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.4048757553100586,
      "learning_rate": 4.852243745469636e-05,
      "loss": 2.1208,
      "step": 188107
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.3172262907028198,
      "learning_rate": 4.852020700506947e-05,
      "loss": 2.1123,
      "step": 188108
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.3912099599838257,
      "learning_rate": 4.851797660219754e-05,
      "loss": 2.0503,
      "step": 188109
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.4058806896209717,
      "learning_rate": 4.851574624608074e-05,
      "loss": 1.7838,
      "step": 188110
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.4500685930252075,
      "learning_rate": 4.851351593671971e-05,
      "loss": 2.1751,
      "step": 188111
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.2910033464431763,
      "learning_rate": 4.851128567411465e-05,
      "loss": 1.8787,
      "step": 188112
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.4076491594314575,
      "learning_rate": 4.8509055458266186e-05,
      "loss": 2.0168,
      "step": 188113
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.4827274084091187,
      "learning_rate": 4.850682528917459e-05,
      "loss": 1.8203,
      "step": 188114
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.2451297044754028,
      "learning_rate": 4.850459516684042e-05,
      "loss": 2.0758,
      "step": 188115
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.4089534282684326,
      "learning_rate": 4.850236509126385e-05,
      "loss": 1.9785,
      "step": 188116
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.3092231750488281,
      "learning_rate": 4.850013506244554e-05,
      "loss": 2.1855,
      "step": 188117
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.3671094179153442,
      "learning_rate": 4.84979050803857e-05,
      "loss": 1.8022,
      "step": 188118
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.4586851596832275,
      "learning_rate": 4.8495675145084955e-05,
      "loss": 2.2356,
      "step": 188119
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.3035950660705566,
      "learning_rate": 4.849344525654351e-05,
      "loss": 2.0317,
      "step": 188120
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.3267470598220825,
      "learning_rate": 4.849121541476205e-05,
      "loss": 2.235,
      "step": 188121
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.311786413192749,
      "learning_rate": 4.848898561974066e-05,
      "loss": 1.9859,
      "step": 188122
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.2324953079223633,
      "learning_rate": 4.8486755871479966e-05,
      "loss": 1.9682,
      "step": 188123
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.3699415922164917,
      "learning_rate": 4.84845261699803e-05,
      "loss": 1.9281,
      "step": 188124
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.4198256731033325,
      "learning_rate": 4.848229651524216e-05,
      "loss": 2.0702,
      "step": 188125
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.4178462028503418,
      "learning_rate": 4.8480066907265844e-05,
      "loss": 2.1192,
      "step": 188126
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.650465726852417,
      "learning_rate": 4.847783734605196e-05,
      "loss": 1.9971,
      "step": 188127
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.3792979717254639,
      "learning_rate": 4.847560783160067e-05,
      "loss": 1.9228,
      "step": 188128
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.2902379035949707,
      "learning_rate": 4.847337836391258e-05,
      "loss": 1.9187,
      "step": 188129
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.3778209686279297,
      "learning_rate": 4.847114894298794e-05,
      "loss": 1.8669,
      "step": 188130
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.431936502456665,
      "learning_rate": 4.846891956882736e-05,
      "loss": 1.9625,
      "step": 188131
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.4652068614959717,
      "learning_rate": 4.846669024143107e-05,
      "loss": 1.8153,
      "step": 188132
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.4164977073669434,
      "learning_rate": 4.846446096079971e-05,
      "loss": 2.0066,
      "step": 188133
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.4088759422302246,
      "learning_rate": 4.846223172693344e-05,
      "loss": 2.0942,
      "step": 188134
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.4993953704833984,
      "learning_rate": 4.846000253983283e-05,
      "loss": 2.0259,
      "step": 188135
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.392745852470398,
      "learning_rate": 4.845777339949821e-05,
      "loss": 2.0117,
      "step": 188136
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.2214877605438232,
      "learning_rate": 4.8455544305930083e-05,
      "loss": 2.0844,
      "step": 188137
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.49109947681427,
      "learning_rate": 4.8453315259128744e-05,
      "loss": 1.9271,
      "step": 188138
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.4761104583740234,
      "learning_rate": 4.84510862590948e-05,
      "loss": 2.2902,
      "step": 188139
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.5036778450012207,
      "learning_rate": 4.84488573058285e-05,
      "loss": 2.0314,
      "step": 188140
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.362908124923706,
      "learning_rate": 4.844662839933034e-05,
      "loss": 2.0163,
      "step": 188141
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.3591421842575073,
      "learning_rate": 4.844439953960059e-05,
      "loss": 1.9054,
      "step": 188142
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.3776499032974243,
      "learning_rate": 4.8442170726639865e-05,
      "loss": 2.2962,
      "step": 188143
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.4433039426803589,
      "learning_rate": 4.8439941960448425e-05,
      "loss": 1.878,
      "step": 188144
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.478054404258728,
      "learning_rate": 4.843771324102681e-05,
      "loss": 2.2068,
      "step": 188145
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.5228855609893799,
      "learning_rate": 4.843548456837538e-05,
      "loss": 1.9964,
      "step": 188146
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.5199304819107056,
      "learning_rate": 4.8433255942494475e-05,
      "loss": 2.0238,
      "step": 188147
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.4033671617507935,
      "learning_rate": 4.843102736338462e-05,
      "loss": 1.9795,
      "step": 188148
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.3739038705825806,
      "learning_rate": 4.8428798831046224e-05,
      "loss": 2.2477,
      "step": 188149
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.3912204504013062,
      "learning_rate": 4.842657034547958e-05,
      "loss": 2.0807,
      "step": 188150
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.4367378950119019,
      "learning_rate": 4.842434190668522e-05,
      "loss": 1.8816,
      "step": 188151
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.4467772245407104,
      "learning_rate": 4.8422113514663576e-05,
      "loss": 1.8189,
      "step": 188152
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.386242389678955,
      "learning_rate": 4.841988516941492e-05,
      "loss": 1.9978,
      "step": 188153
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.29606294631958,
      "learning_rate": 4.841765687093982e-05,
      "loss": 1.9379,
      "step": 188154
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.484503149986267,
      "learning_rate": 4.841542861923863e-05,
      "loss": 1.8351,
      "step": 188155
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.42628812789917,
      "learning_rate": 4.841320041431169e-05,
      "loss": 1.8227,
      "step": 188156
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.416838526725769,
      "learning_rate": 4.841097225615954e-05,
      "loss": 2.074,
      "step": 188157
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.3494529724121094,
      "learning_rate": 4.840874414478257e-05,
      "loss": 1.8536,
      "step": 188158
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.2253613471984863,
      "learning_rate": 4.8406516080181055e-05,
      "loss": 1.9954,
      "step": 188159
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.3343578577041626,
      "learning_rate": 4.8404288062355625e-05,
      "loss": 2.1364,
      "step": 188160
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.2772128582000732,
      "learning_rate": 4.8402060091306505e-05,
      "loss": 1.906,
      "step": 188161
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.4531259536743164,
      "learning_rate": 4.839983216703427e-05,
      "loss": 2.2033,
      "step": 188162
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.2765626907348633,
      "learning_rate": 4.839760428953925e-05,
      "loss": 1.8016,
      "step": 188163
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.2769156694412231,
      "learning_rate": 4.8395376458821846e-05,
      "loss": 1.8861,
      "step": 188164
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.3419342041015625,
      "learning_rate": 4.839314867488242e-05,
      "loss": 1.8804,
      "step": 188165
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.5391335487365723,
      "learning_rate": 4.8390920937721545e-05,
      "loss": 2.0443,
      "step": 188166
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.389007568359375,
      "learning_rate": 4.838869324733948e-05,
      "loss": 1.8448,
      "step": 188167
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.4335687160491943,
      "learning_rate": 4.8386465603736766e-05,
      "loss": 1.8787,
      "step": 188168
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.2412887811660767,
      "learning_rate": 4.83842380069137e-05,
      "loss": 1.9738,
      "step": 188169
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.431774377822876,
      "learning_rate": 4.838201045687088e-05,
      "loss": 1.9307,
      "step": 188170
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.274175763130188,
      "learning_rate": 4.837978295360847e-05,
      "loss": 1.9494,
      "step": 188171
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.4614821672439575,
      "learning_rate": 4.837755549712707e-05,
      "loss": 1.8471,
      "step": 188172
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.4129871129989624,
      "learning_rate": 4.8375328087426954e-05,
      "loss": 2.0136,
      "step": 188173
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.3178056478500366,
      "learning_rate": 4.837310072450868e-05,
      "loss": 2.1018,
      "step": 188174
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.4176620244979858,
      "learning_rate": 4.8370873408372555e-05,
      "loss": 1.9315,
      "step": 188175
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.4824297428131104,
      "learning_rate": 4.8368646139019176e-05,
      "loss": 1.9732,
      "step": 188176
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.4022955894470215,
      "learning_rate": 4.836641891644863e-05,
      "loss": 1.9066,
      "step": 188177
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.4083902835845947,
      "learning_rate": 4.836419174066164e-05,
      "loss": 1.9918,
      "step": 188178
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.5846446752548218,
      "learning_rate": 4.8361964611658385e-05,
      "loss": 2.1491,
      "step": 188179
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.3815444707870483,
      "learning_rate": 4.835973752943948e-05,
      "loss": 2.0908,
      "step": 188180
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.4761728048324585,
      "learning_rate": 4.835751049400518e-05,
      "loss": 2.0329,
      "step": 188181
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.3217188119888306,
      "learning_rate": 4.8355283505356093e-05,
      "loss": 2.066,
      "step": 188182
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.4193007946014404,
      "learning_rate": 4.835305656349238e-05,
      "loss": 1.909,
      "step": 188183
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.3710969686508179,
      "learning_rate": 4.835082966841467e-05,
      "loss": 1.9349,
      "step": 188184
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.540306568145752,
      "learning_rate": 4.8348602820123204e-05,
      "loss": 1.9052,
      "step": 188185
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.3203610181808472,
      "learning_rate": 4.8346376018618546e-05,
      "loss": 2.0066,
      "step": 188186
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.352545976638794,
      "learning_rate": 4.8344149263900966e-05,
      "loss": 1.7598,
      "step": 188187
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.2510488033294678,
      "learning_rate": 4.834192255597112e-05,
      "loss": 1.9524,
      "step": 188188
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.3786900043487549,
      "learning_rate": 4.833969589482912e-05,
      "loss": 1.8886,
      "step": 188189
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.3085044622421265,
      "learning_rate": 4.833746928047559e-05,
      "loss": 1.9813,
      "step": 188190
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.4163711071014404,
      "learning_rate": 4.83352427129108e-05,
      "loss": 2.1386,
      "step": 188191
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.3927572965621948,
      "learning_rate": 4.833301619213532e-05,
      "loss": 2.3703,
      "step": 188192
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.2991400957107544,
      "learning_rate": 4.833078971814941e-05,
      "loss": 1.9098,
      "step": 188193
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.4672048091888428,
      "learning_rate": 4.8328563290953705e-05,
      "loss": 1.9786,
      "step": 188194
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.3897645473480225,
      "learning_rate": 4.832633691054833e-05,
      "loss": 2.0783,
      "step": 188195
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.3327386379241943,
      "learning_rate": 4.832411057693391e-05,
      "loss": 1.9827,
      "step": 188196
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.3498176336288452,
      "learning_rate": 4.832188429011068e-05,
      "loss": 2.0424,
      "step": 188197
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.5320738554000854,
      "learning_rate": 4.831965805007926e-05,
      "loss": 2.1978,
      "step": 188198
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.3293335437774658,
      "learning_rate": 4.8317431856839906e-05,
      "loss": 1.9275,
      "step": 188199
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.3991639614105225,
      "learning_rate": 4.8315205710393226e-05,
      "loss": 2.0934,
      "step": 188200
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.4320834875106812,
      "learning_rate": 4.831297961073934e-05,
      "loss": 2.1862,
      "step": 188201
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.4479063749313354,
      "learning_rate": 4.8310753557878935e-05,
      "loss": 2.2111,
      "step": 188202
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.3338524103164673,
      "learning_rate": 4.830852755181219e-05,
      "loss": 2.2197,
      "step": 188203
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.4247956275939941,
      "learning_rate": 4.8306301592539755e-05,
      "loss": 2.0625,
      "step": 188204
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.4747830629348755,
      "learning_rate": 4.830407568006185e-05,
      "loss": 2.0475,
      "step": 188205
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.498888373374939,
      "learning_rate": 4.830184981437908e-05,
      "loss": 2.2871,
      "step": 188206
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.3553389310836792,
      "learning_rate": 4.829962399549164e-05,
      "loss": 2.1889,
      "step": 188207
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.2616302967071533,
      "learning_rate": 4.8297398223400106e-05,
      "loss": 2.1068,
      "step": 188208
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.429761528968811,
      "learning_rate": 4.8295172498104773e-05,
      "loss": 2.0326,
      "step": 188209
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.2990976572036743,
      "learning_rate": 4.829294681960617e-05,
      "loss": 2.0377,
      "step": 188210
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.3529889583587646,
      "learning_rate": 4.82907211879046e-05,
      "loss": 2.0231,
      "step": 188211
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.3823689222335815,
      "learning_rate": 4.82884956030007e-05,
      "loss": 2.0341,
      "step": 188212
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.4282183647155762,
      "learning_rate": 4.828627006489456e-05,
      "loss": 2.0984,
      "step": 188213
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.4714293479919434,
      "learning_rate": 4.828404457358682e-05,
      "loss": 2.0241,
      "step": 188214
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.2738741636276245,
      "learning_rate": 4.828181912907778e-05,
      "loss": 1.9753,
      "step": 188215
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.2170493602752686,
      "learning_rate": 4.827959373136794e-05,
      "loss": 1.9107,
      "step": 188216
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.42538321018219,
      "learning_rate": 4.827736838045763e-05,
      "loss": 2.2134,
      "step": 188217
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.5429565906524658,
      "learning_rate": 4.827514307634744e-05,
      "loss": 2.0762,
      "step": 188218
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.3885620832443237,
      "learning_rate": 4.8272917819037495e-05,
      "loss": 1.9648,
      "step": 188219
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.314474105834961,
      "learning_rate": 4.827069260852847e-05,
      "loss": 1.6827,
      "step": 188220
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.2629314661026,
      "learning_rate": 4.826846744482058e-05,
      "loss": 2.0908,
      "step": 188221
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.4525213241577148,
      "learning_rate": 4.8266242327914416e-05,
      "loss": 1.9984,
      "step": 188222
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.4336177110671997,
      "learning_rate": 4.8264017257810216e-05,
      "loss": 2.1615,
      "step": 188223
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.4285807609558105,
      "learning_rate": 4.826179223450858e-05,
      "loss": 2.1473,
      "step": 188224
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.3500478267669678,
      "learning_rate": 4.825956725800983e-05,
      "loss": 1.9568,
      "step": 188225
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.312363624572754,
      "learning_rate": 4.8257342328314386e-05,
      "loss": 1.8273,
      "step": 188226
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.3990027904510498,
      "learning_rate": 4.8255117445422566e-05,
      "loss": 1.9761,
      "step": 188227
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.4156341552734375,
      "learning_rate": 4.8252892609334934e-05,
      "loss": 1.8515,
      "step": 188228
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.3391449451446533,
      "learning_rate": 4.825066782005177e-05,
      "loss": 2.0722,
      "step": 188229
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.321267008781433,
      "learning_rate": 4.8248443077573664e-05,
      "loss": 1.8526,
      "step": 188230
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.3943322896957397,
      "learning_rate": 4.824621838190088e-05,
      "loss": 2.0941,
      "step": 188231
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.289908528327942,
      "learning_rate": 4.824399373303386e-05,
      "loss": 1.8463,
      "step": 188232
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.3260635137557983,
      "learning_rate": 4.8241769130973064e-05,
      "loss": 2.0209,
      "step": 188233
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.417651891708374,
      "learning_rate": 4.8239544575718894e-05,
      "loss": 1.9836,
      "step": 188234
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.4533101320266724,
      "learning_rate": 4.823732006727164e-05,
      "loss": 2.0952,
      "step": 188235
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.3551108837127686,
      "learning_rate": 4.8235095605631926e-05,
      "loss": 1.8851,
      "step": 188236
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.4412531852722168,
      "learning_rate": 4.823287119080006e-05,
      "loss": 2.2459,
      "step": 188237
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.2948484420776367,
      "learning_rate": 4.8230646822776354e-05,
      "loss": 2.0207,
      "step": 188238
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.3038016557693481,
      "learning_rate": 4.822842250156144e-05,
      "loss": 2.0936,
      "step": 188239
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.2742540836334229,
      "learning_rate": 4.822619822715558e-05,
      "loss": 2.1448,
      "step": 188240
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.3223286867141724,
      "learning_rate": 4.822397399955917e-05,
      "loss": 1.9351,
      "step": 188241
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.4231693744659424,
      "learning_rate": 4.822174981877272e-05,
      "loss": 2.0418,
      "step": 188242
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.5174857378005981,
      "learning_rate": 4.821952568479664e-05,
      "loss": 2.0631,
      "step": 188243
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.5316684246063232,
      "learning_rate": 4.82173015976312e-05,
      "loss": 2.0137,
      "step": 188244
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.3966622352600098,
      "learning_rate": 4.8215077557276985e-05,
      "loss": 2.0009,
      "step": 188245
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.236730933189392,
      "learning_rate": 4.8212853563734286e-05,
      "loss": 1.7876,
      "step": 188246
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.3080329895019531,
      "learning_rate": 4.821062961700365e-05,
      "loss": 2.1234,
      "step": 188247
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.2962013483047485,
      "learning_rate": 4.82084057170854e-05,
      "loss": 2.0842,
      "step": 188248
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.2957946062088013,
      "learning_rate": 4.820618186397993e-05,
      "loss": 1.787,
      "step": 188249
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.4598453044891357,
      "learning_rate": 4.820395805768765e-05,
      "loss": 1.9243,
      "step": 188250
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.3356122970581055,
      "learning_rate": 4.820173429820906e-05,
      "loss": 1.9608,
      "step": 188251
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.272251009941101,
      "learning_rate": 4.819951058554449e-05,
      "loss": 1.8161,
      "step": 188252
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.4210805892944336,
      "learning_rate": 4.81972869196944e-05,
      "loss": 1.9843,
      "step": 188253
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.3393182754516602,
      "learning_rate": 4.819506330065923e-05,
      "loss": 1.9175,
      "step": 188254
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.199088454246521,
      "learning_rate": 4.819283972843931e-05,
      "loss": 1.7531,
      "step": 188255
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.354967713356018,
      "learning_rate": 4.819061620303505e-05,
      "loss": 2.1526,
      "step": 188256
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.277459740638733,
      "learning_rate": 4.8188392724446964e-05,
      "loss": 1.9742,
      "step": 188257
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.3710014820098877,
      "learning_rate": 4.818616929267537e-05,
      "loss": 1.9761,
      "step": 188258
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.423475742340088,
      "learning_rate": 4.818394590772076e-05,
      "loss": 2.3691,
      "step": 188259
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.3969221115112305,
      "learning_rate": 4.8181722569583456e-05,
      "loss": 1.9571,
      "step": 188260
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.3731783628463745,
      "learning_rate": 4.8179499278264044e-05,
      "loss": 1.8956,
      "step": 188261
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.495309591293335,
      "learning_rate": 4.817727603376268e-05,
      "loss": 2.0641,
      "step": 188262
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.2528085708618164,
      "learning_rate": 4.817505283608e-05,
      "loss": 2.063,
      "step": 188263
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.4074549674987793,
      "learning_rate": 4.817282968521626e-05,
      "loss": 1.8991,
      "step": 188264
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.3679249286651611,
      "learning_rate": 4.817060658117201e-05,
      "loss": 2.2587,
      "step": 188265
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.5089683532714844,
      "learning_rate": 4.816838352394751e-05,
      "loss": 1.9863,
      "step": 188266
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.2529473304748535,
      "learning_rate": 4.8166160513543415e-05,
      "loss": 1.9724,
      "step": 188267
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.450014591217041,
      "learning_rate": 4.816393754995984e-05,
      "loss": 2.1179,
      "step": 188268
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.364551305770874,
      "learning_rate": 4.8161714633197415e-05,
      "loss": 2.0672,
      "step": 188269
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.4570831060409546,
      "learning_rate": 4.815949176325643e-05,
      "loss": 1.8729,
      "step": 188270
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.367999792098999,
      "learning_rate": 4.815726894013739e-05,
      "loss": 1.8965,
      "step": 188271
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.4132646322250366,
      "learning_rate": 4.81550461638406e-05,
      "loss": 1.8061,
      "step": 188272
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.3819466829299927,
      "learning_rate": 4.815282343436673e-05,
      "loss": 2.0725,
      "step": 188273
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.4360508918762207,
      "learning_rate": 4.8150600751715794e-05,
      "loss": 1.9137,
      "step": 188274
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.3818272352218628,
      "learning_rate": 4.814837811588854e-05,
      "loss": 2.1066,
      "step": 188275
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.3574217557907104,
      "learning_rate": 4.814615552688517e-05,
      "loss": 2.1678,
      "step": 188276
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.3507897853851318,
      "learning_rate": 4.814393298470623e-05,
      "loss": 1.9118,
      "step": 188277
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.403733491897583,
      "learning_rate": 4.814171048935205e-05,
      "loss": 2.0425,
      "step": 188278
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.3665637969970703,
      "learning_rate": 4.81394880408232e-05,
      "loss": 2.1216,
      "step": 188279
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.2879043817520142,
      "learning_rate": 4.8137265639119835e-05,
      "loss": 1.8353,
      "step": 188280
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.3991138935089111,
      "learning_rate": 4.813504328424258e-05,
      "loss": 1.89,
      "step": 188281
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.3997870683670044,
      "learning_rate": 4.81328209761917e-05,
      "loss": 1.8169,
      "step": 188282
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.2756582498550415,
      "learning_rate": 4.813059871496776e-05,
      "loss": 2.0126,
      "step": 188283
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.5126146078109741,
      "learning_rate": 4.812837650057103e-05,
      "loss": 2.0712,
      "step": 188284
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.5256108045578003,
      "learning_rate": 4.812615433300211e-05,
      "loss": 1.807,
      "step": 188285
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.3045700788497925,
      "learning_rate": 4.812393221226116e-05,
      "loss": 1.9778,
      "step": 188286
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.5480014085769653,
      "learning_rate": 4.812171013834882e-05,
      "loss": 2.2174,
      "step": 188287
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.3822604417800903,
      "learning_rate": 4.811948811126529e-05,
      "loss": 1.9308,
      "step": 188288
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.5013734102249146,
      "learning_rate": 4.811726613101123e-05,
      "loss": 2.0195,
      "step": 188289
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.5818390846252441,
      "learning_rate": 4.8115044197586816e-05,
      "loss": 1.9453,
      "step": 188290
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.4069112539291382,
      "learning_rate": 4.811282231099274e-05,
      "loss": 2.0048,
      "step": 188291
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.2752034664154053,
      "learning_rate": 4.811060047122907e-05,
      "loss": 1.8321,
      "step": 188292
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.331531047821045,
      "learning_rate": 4.8108378678296464e-05,
      "loss": 2.0673,
      "step": 188293
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.4152190685272217,
      "learning_rate": 4.810615693219521e-05,
      "loss": 2.2691,
      "step": 188294
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.4363970756530762,
      "learning_rate": 4.810393523292585e-05,
      "loss": 2.0581,
      "step": 188295
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.4094479084014893,
      "learning_rate": 4.810171358048866e-05,
      "loss": 1.9296,
      "step": 188296
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.313965082168579,
      "learning_rate": 4.8099491974884254e-05,
      "loss": 2.1611,
      "step": 188297
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.3830125331878662,
      "learning_rate": 4.8097270416112744e-05,
      "loss": 1.9823,
      "step": 188298
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.3574310541152954,
      "learning_rate": 4.8095048904174804e-05,
      "loss": 2.1238,
      "step": 188299
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.6643821001052856,
      "learning_rate": 4.809282743907067e-05,
      "loss": 2.0618,
      "step": 188300
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.4200619459152222,
      "learning_rate": 4.8090606020800906e-05,
      "loss": 1.9638,
      "step": 188301
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.3045947551727295,
      "learning_rate": 4.8088384649365776e-05,
      "loss": 2.0977,
      "step": 188302
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.3603206872940063,
      "learning_rate": 4.808616332476585e-05,
      "loss": 2.032,
      "step": 188303
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.3625431060791016,
      "learning_rate": 4.8083942047001455e-05,
      "loss": 1.7798,
      "step": 188304
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.3465481996536255,
      "learning_rate": 4.8081720816073e-05,
      "loss": 1.9502,
      "step": 188305
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.318894624710083,
      "learning_rate": 4.8079499631980847e-05,
      "loss": 2.0516,
      "step": 188306
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.3830829858779907,
      "learning_rate": 4.807727849472556e-05,
      "loss": 2.0119,
      "step": 188307
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.3650333881378174,
      "learning_rate": 4.807505740430737e-05,
      "loss": 1.9766,
      "step": 188308
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.2996760606765747,
      "learning_rate": 4.807283636072685e-05,
      "loss": 1.8331,
      "step": 188309
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.4687976837158203,
      "learning_rate": 4.807061536398436e-05,
      "loss": 1.8667,
      "step": 188310
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.880205512046814,
      "learning_rate": 4.806839441408028e-05,
      "loss": 2.0078,
      "step": 188311
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.3092831373214722,
      "learning_rate": 4.8066173511014994e-05,
      "loss": 1.9116,
      "step": 188312
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.3364781141281128,
      "learning_rate": 4.806395265478901e-05,
      "loss": 2.04,
      "step": 188313
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.541904330253601,
      "learning_rate": 4.806173184540265e-05,
      "loss": 2.2496,
      "step": 188314
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.5317089557647705,
      "learning_rate": 4.80595110828564e-05,
      "loss": 1.7915,
      "step": 188315
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.4271830320358276,
      "learning_rate": 4.805729036715067e-05,
      "loss": 1.9093,
      "step": 188316
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.403202772140503,
      "learning_rate": 4.805506969828584e-05,
      "loss": 2.05,
      "step": 188317
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.465707540512085,
      "learning_rate": 4.805284907626228e-05,
      "loss": 1.8253,
      "step": 188318
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.3066442012786865,
      "learning_rate": 4.805062850108048e-05,
      "loss": 2.2012,
      "step": 188319
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.367822289466858,
      "learning_rate": 4.804840797274078e-05,
      "loss": 1.9204,
      "step": 188320
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.2730600833892822,
      "learning_rate": 4.80461874912437e-05,
      "loss": 2.0397,
      "step": 188321
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.2397541999816895,
      "learning_rate": 4.804396705658959e-05,
      "loss": 2.0685,
      "step": 188322
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.324338674545288,
      "learning_rate": 4.804174666877878e-05,
      "loss": 1.8978,
      "step": 188323
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.3454697132110596,
      "learning_rate": 4.803952632781186e-05,
      "loss": 1.9942,
      "step": 188324
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.480685830116272,
      "learning_rate": 4.803730603368914e-05,
      "loss": 1.8905,
      "step": 188325
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.4840635061264038,
      "learning_rate": 4.803508578641098e-05,
      "loss": 1.8767,
      "step": 188326
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.5785331726074219,
      "learning_rate": 4.803286558597791e-05,
      "loss": 1.9764,
      "step": 188327
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.460250735282898,
      "learning_rate": 4.803064543239027e-05,
      "loss": 2.0415,
      "step": 188328
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.4003626108169556,
      "learning_rate": 4.8028425325648455e-05,
      "loss": 1.9388,
      "step": 188329
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.4213886260986328,
      "learning_rate": 4.8026205265752973e-05,
      "loss": 2.2285,
      "step": 188330
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.3768001794815063,
      "learning_rate": 4.802398525270408e-05,
      "loss": 1.8653,
      "step": 188331
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.395910620689392,
      "learning_rate": 4.802176528650238e-05,
      "loss": 2.1412,
      "step": 188332
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.428992748260498,
      "learning_rate": 4.801954536714817e-05,
      "loss": 1.9087,
      "step": 188333
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.376326560974121,
      "learning_rate": 4.801732549464189e-05,
      "loss": 1.9225,
      "step": 188334
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.3113681077957153,
      "learning_rate": 4.80151056689839e-05,
      "loss": 2.1716,
      "step": 188335
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.2804821729660034,
      "learning_rate": 4.80128858901747e-05,
      "loss": 2.0115,
      "step": 188336
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.4045518636703491,
      "learning_rate": 4.801066615821459e-05,
      "loss": 1.8,
      "step": 188337
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.4350314140319824,
      "learning_rate": 4.8008446473104146e-05,
      "loss": 2.2129,
      "step": 188338
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.3313918113708496,
      "learning_rate": 4.800622683484369e-05,
      "loss": 2.0713,
      "step": 188339
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.4246492385864258,
      "learning_rate": 4.800400724343362e-05,
      "loss": 2.1451,
      "step": 188340
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.648897647857666,
      "learning_rate": 4.800178769887427e-05,
      "loss": 1.9756,
      "step": 188341
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.3049038648605347,
      "learning_rate": 4.799956820116625e-05,
      "loss": 1.8407,
      "step": 188342
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.3368427753448486,
      "learning_rate": 4.799734875030978e-05,
      "loss": 2.174,
      "step": 188343
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.3736907243728638,
      "learning_rate": 4.799512934630544e-05,
      "loss": 1.8177,
      "step": 188344
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.4303593635559082,
      "learning_rate": 4.7992909989153486e-05,
      "loss": 1.9942,
      "step": 188345
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.4557929039001465,
      "learning_rate": 4.799069067885455e-05,
      "loss": 1.9187,
      "step": 188346
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.4101061820983887,
      "learning_rate": 4.7988471415408766e-05,
      "loss": 1.8725,
      "step": 188347
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.3885986804962158,
      "learning_rate": 4.7986252198816746e-05,
      "loss": 2.1569,
      "step": 188348
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.5505709648132324,
      "learning_rate": 4.7984033029078774e-05,
      "loss": 2.1836,
      "step": 188349
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.3331042528152466,
      "learning_rate": 4.798181390619539e-05,
      "loss": 1.932,
      "step": 188350
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.4618905782699585,
      "learning_rate": 4.7979594830166855e-05,
      "loss": 1.9792,
      "step": 188351
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.328539490699768,
      "learning_rate": 4.7977375800993844e-05,
      "loss": 2.1082,
      "step": 188352
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.4221229553222656,
      "learning_rate": 4.797515681867642e-05,
      "loss": 1.9038,
      "step": 188353
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.4337878227233887,
      "learning_rate": 4.797293788321528e-05,
      "loss": 2.0745,
      "step": 188354
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.3649916648864746,
      "learning_rate": 4.797071899461066e-05,
      "loss": 1.8772,
      "step": 188355
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.3274410963058472,
      "learning_rate": 4.796850015286309e-05,
      "loss": 2.1302,
      "step": 188356
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.345542311668396,
      "learning_rate": 4.796628135797287e-05,
      "loss": 2.0119,
      "step": 188357
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.3114008903503418,
      "learning_rate": 4.79640626099406e-05,
      "loss": 2.0703,
      "step": 188358
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.3796420097351074,
      "learning_rate": 4.796184390876645e-05,
      "loss": 2.2138,
      "step": 188359
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.3742815256118774,
      "learning_rate": 4.795962525445102e-05,
      "loss": 1.7786,
      "step": 188360
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.3279703855514526,
      "learning_rate": 4.795740664699457e-05,
      "loss": 1.8006,
      "step": 188361
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.4294719696044922,
      "learning_rate": 4.7955188086397664e-05,
      "loss": 1.8742,
      "step": 188362
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.4990575313568115,
      "learning_rate": 4.795296957266058e-05,
      "loss": 2.3827,
      "step": 188363
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.2314643859863281,
      "learning_rate": 4.795075110578395e-05,
      "loss": 1.8865,
      "step": 188364
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.4534599781036377,
      "learning_rate": 4.794853268576786e-05,
      "loss": 2.1797,
      "step": 188365
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.3112475872039795,
      "learning_rate": 4.794631431261299e-05,
      "loss": 1.8373,
      "step": 188366
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.6040301322937012,
      "learning_rate": 4.7944095986319594e-05,
      "loss": 2.0942,
      "step": 188367
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.4109314680099487,
      "learning_rate": 4.794187770688822e-05,
      "loss": 1.8911,
      "step": 188368
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.4657965898513794,
      "learning_rate": 4.7939659474319124e-05,
      "loss": 2.2143,
      "step": 188369
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.2553163766860962,
      "learning_rate": 4.7937441288612875e-05,
      "loss": 1.9196,
      "step": 188370
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.5248147249221802,
      "learning_rate": 4.793522314976981e-05,
      "loss": 2.2203,
      "step": 188371
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.464648723602295,
      "learning_rate": 4.793300505779035e-05,
      "loss": 1.9704,
      "step": 188372
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.477553367614746,
      "learning_rate": 4.793078701267484e-05,
      "loss": 1.8716,
      "step": 188373
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.3222482204437256,
      "learning_rate": 4.792856901442385e-05,
      "loss": 1.9686,
      "step": 188374
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.47632896900177,
      "learning_rate": 4.7926351063037596e-05,
      "loss": 2.0617,
      "step": 188375
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.3281974792480469,
      "learning_rate": 4.7924133158516654e-05,
      "loss": 1.9954,
      "step": 188376
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.3506814241409302,
      "learning_rate": 4.7921915300861394e-05,
      "loss": 1.9803,
      "step": 188377
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.449442982673645,
      "learning_rate": 4.7919697490072175e-05,
      "loss": 2.0194,
      "step": 188378
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.3305896520614624,
      "learning_rate": 4.79174797261494e-05,
      "loss": 2.227,
      "step": 188379
    },
    {
      "epoch": 2.45,
      "grad_norm": 2.18912935256958,
      "learning_rate": 4.7915262009093605e-05,
      "loss": 2.1418,
      "step": 188380
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.52060067653656,
      "learning_rate": 4.791304433890502e-05,
      "loss": 2.0079,
      "step": 188381
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.3930065631866455,
      "learning_rate": 4.7910826715584235e-05,
      "loss": 1.8814,
      "step": 188382
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.2797774076461792,
      "learning_rate": 4.79086091391316e-05,
      "loss": 2.0947,
      "step": 188383
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.3403972387313843,
      "learning_rate": 4.790639160954751e-05,
      "loss": 2.1866,
      "step": 188384
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.251950979232788,
      "learning_rate": 4.79041741268323e-05,
      "loss": 2.0355,
      "step": 188385
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.4325236082077026,
      "learning_rate": 4.790195669098656e-05,
      "loss": 1.9714,
      "step": 188386
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.6371301412582397,
      "learning_rate": 4.789973930201052e-05,
      "loss": 2.1567,
      "step": 188387
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.5248093605041504,
      "learning_rate": 4.789752195990473e-05,
      "loss": 1.816,
      "step": 188388
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.4661288261413574,
      "learning_rate": 4.789530466466958e-05,
      "loss": 1.8813,
      "step": 188389
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.444940447807312,
      "learning_rate": 4.789308741630543e-05,
      "loss": 2.1152,
      "step": 188390
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.5646263360977173,
      "learning_rate": 4.789087021481263e-05,
      "loss": 1.9818,
      "step": 188391
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.3768792152404785,
      "learning_rate": 4.788865306019176e-05,
      "loss": 2.0987,
      "step": 188392
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.3885681629180908,
      "learning_rate": 4.7886435952443106e-05,
      "loss": 1.9726,
      "step": 188393
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.4483845233917236,
      "learning_rate": 4.788421889156715e-05,
      "loss": 1.8732,
      "step": 188394
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.3494445085525513,
      "learning_rate": 4.78820018775643e-05,
      "loss": 1.8951,
      "step": 188395
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.3530982732772827,
      "learning_rate": 4.787978491043496e-05,
      "loss": 1.9968,
      "step": 188396
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.290060043334961,
      "learning_rate": 4.787756799017942e-05,
      "loss": 1.8758,
      "step": 188397
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.2415786981582642,
      "learning_rate": 4.7875351116798286e-05,
      "loss": 2.0156,
      "step": 188398
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.359922170639038,
      "learning_rate": 4.7873134290291825e-05,
      "loss": 1.955,
      "step": 188399
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.510764718055725,
      "learning_rate": 4.787091751066057e-05,
      "loss": 2.1969,
      "step": 188400
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.4447249174118042,
      "learning_rate": 4.786870077790488e-05,
      "loss": 2.1274,
      "step": 188401
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.4209367036819458,
      "learning_rate": 4.7866484092025125e-05,
      "loss": 2.035,
      "step": 188402
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.420105218887329,
      "learning_rate": 4.786426745302171e-05,
      "loss": 2.052,
      "step": 188403
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.5166523456573486,
      "learning_rate": 4.7862050860895163e-05,
      "loss": 1.9324,
      "step": 188404
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.4299274682998657,
      "learning_rate": 4.7859834315645715e-05,
      "loss": 1.9582,
      "step": 188405
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.4181667566299438,
      "learning_rate": 4.7857617817273977e-05,
      "loss": 2.1698,
      "step": 188406
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.36360502243042,
      "learning_rate": 4.785540136578027e-05,
      "loss": 2.0046,
      "step": 188407
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.4117785692214966,
      "learning_rate": 4.7853184961164925e-05,
      "loss": 1.9626,
      "step": 188408
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.4001851081848145,
      "learning_rate": 4.785096860342852e-05,
      "loss": 2.1211,
      "step": 188409
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.5150991678237915,
      "learning_rate": 4.784875229257134e-05,
      "loss": 2.1268,
      "step": 188410
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.4851852655410767,
      "learning_rate": 4.784653602859381e-05,
      "loss": 2.1076,
      "step": 188411
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.370974063873291,
      "learning_rate": 4.78443198114964e-05,
      "loss": 2.014,
      "step": 188412
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.2986500263214111,
      "learning_rate": 4.7842103641279526e-05,
      "loss": 1.8765,
      "step": 188413
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.4691904783248901,
      "learning_rate": 4.783988751794349e-05,
      "loss": 2.0293,
      "step": 188414
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.3479810953140259,
      "learning_rate": 4.783767144148881e-05,
      "loss": 1.9985,
      "step": 188415
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.3727226257324219,
      "learning_rate": 4.78354554119159e-05,
      "loss": 1.8563,
      "step": 188416
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.4193562269210815,
      "learning_rate": 4.783323942922509e-05,
      "loss": 2.1782,
      "step": 188417
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.4150125980377197,
      "learning_rate": 4.783102349341688e-05,
      "loss": 1.7511,
      "step": 188418
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.5121299028396606,
      "learning_rate": 4.782880760449163e-05,
      "loss": 2.1247,
      "step": 188419
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.3523764610290527,
      "learning_rate": 4.78265917624497e-05,
      "loss": 1.9547,
      "step": 188420
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.3994228839874268,
      "learning_rate": 4.7824375967291684e-05,
      "loss": 1.9053,
      "step": 188421
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.3223294019699097,
      "learning_rate": 4.782216021901776e-05,
      "loss": 2.0114,
      "step": 188422
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.3532358407974243,
      "learning_rate": 4.781994451762853e-05,
      "loss": 1.8254,
      "step": 188423
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.2651110887527466,
      "learning_rate": 4.781772886312436e-05,
      "loss": 1.8768,
      "step": 188424
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.274962306022644,
      "learning_rate": 4.7815513255505624e-05,
      "loss": 2.0589,
      "step": 188425
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.3249619007110596,
      "learning_rate": 4.7813297694772646e-05,
      "loss": 1.9961,
      "step": 188426
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.4487046003341675,
      "learning_rate": 4.781108218092603e-05,
      "loss": 2.0746,
      "step": 188427
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.6878101825714111,
      "learning_rate": 4.780886671396601e-05,
      "loss": 1.8303,
      "step": 188428
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.3905378580093384,
      "learning_rate": 4.780665129389318e-05,
      "loss": 2.0388,
      "step": 188429
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.4065947532653809,
      "learning_rate": 4.780443592070775e-05,
      "loss": 2.1421,
      "step": 188430
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.3437048196792603,
      "learning_rate": 4.7802220594410415e-05,
      "loss": 2.0168,
      "step": 188431
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.3656384944915771,
      "learning_rate": 4.7800005315001236e-05,
      "loss": 2.0757,
      "step": 188432
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.3080878257751465,
      "learning_rate": 4.779779008248088e-05,
      "loss": 1.9743,
      "step": 188433
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.3803294897079468,
      "learning_rate": 4.7795574896849587e-05,
      "loss": 2.1338,
      "step": 188434
    },
    {
      "epoch": 2.45,
      "grad_norm": 2.0288689136505127,
      "learning_rate": 4.779335975810795e-05,
      "loss": 1.8507,
      "step": 188435
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.418200969696045,
      "learning_rate": 4.779114466625621e-05,
      "loss": 2.1188,
      "step": 188436
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.4881001710891724,
      "learning_rate": 4.7788929621294925e-05,
      "loss": 2.0394,
      "step": 188437
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.5466736555099487,
      "learning_rate": 4.778671462322447e-05,
      "loss": 2.0107,
      "step": 188438
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.3728286027908325,
      "learning_rate": 4.778449967204516e-05,
      "loss": 1.9218,
      "step": 188439
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.3195960521697998,
      "learning_rate": 4.778228476775745e-05,
      "loss": 1.8976,
      "step": 188440
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.6194361448287964,
      "learning_rate": 4.778006991036183e-05,
      "loss": 2.0893,
      "step": 188441
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.380815863609314,
      "learning_rate": 4.77778550998586e-05,
      "loss": 1.9804,
      "step": 188442
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.3454067707061768,
      "learning_rate": 4.777564033624829e-05,
      "loss": 2.0511,
      "step": 188443
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.629335880279541,
      "learning_rate": 4.7773425619531235e-05,
      "loss": 1.9212,
      "step": 188444
    },
    {
      "epoch": 2.45,
      "grad_norm": 2.116166591644287,
      "learning_rate": 4.777121094970787e-05,
      "loss": 2.0062,
      "step": 188445
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.557902216911316,
      "learning_rate": 4.776899632677853e-05,
      "loss": 2.14,
      "step": 188446
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.5449274778366089,
      "learning_rate": 4.776678175074378e-05,
      "loss": 2.1405,
      "step": 188447
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.3983125686645508,
      "learning_rate": 4.776456722160388e-05,
      "loss": 1.7734,
      "step": 188448
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.3914263248443604,
      "learning_rate": 4.7762352739359336e-05,
      "loss": 1.9904,
      "step": 188449
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.366806149482727,
      "learning_rate": 4.7760138304010575e-05,
      "loss": 2.009,
      "step": 188450
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.3789756298065186,
      "learning_rate": 4.775792391555794e-05,
      "loss": 1.9789,
      "step": 188451
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.3726308345794678,
      "learning_rate": 4.775570957400182e-05,
      "loss": 1.8061,
      "step": 188452
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.5177888870239258,
      "learning_rate": 4.775349527934273e-05,
      "loss": 1.9691,
      "step": 188453
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.2605745792388916,
      "learning_rate": 4.7751281031580945e-05,
      "loss": 1.8191,
      "step": 188454
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.4206899404525757,
      "learning_rate": 4.774906683071705e-05,
      "loss": 2.0925,
      "step": 188455
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.6180415153503418,
      "learning_rate": 4.774685267675134e-05,
      "loss": 1.7954,
      "step": 188456
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.3293672800064087,
      "learning_rate": 4.774463856968429e-05,
      "loss": 1.9571,
      "step": 188457
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.3783084154129028,
      "learning_rate": 4.774242450951619e-05,
      "loss": 2.085,
      "step": 188458
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.465557336807251,
      "learning_rate": 4.774021049624757e-05,
      "loss": 2.0427,
      "step": 188459
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.3483301401138306,
      "learning_rate": 4.7737996529878765e-05,
      "loss": 1.9907,
      "step": 188460
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.2656697034835815,
      "learning_rate": 4.773578261041031e-05,
      "loss": 1.9381,
      "step": 188461
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.3791885375976562,
      "learning_rate": 4.773356873784251e-05,
      "loss": 1.7972,
      "step": 188462
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.3781577348709106,
      "learning_rate": 4.773135491217582e-05,
      "loss": 1.9675,
      "step": 188463
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.356334924697876,
      "learning_rate": 4.772914113341055e-05,
      "loss": 1.9281,
      "step": 188464
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.551405668258667,
      "learning_rate": 4.7726927401547264e-05,
      "loss": 2.0009,
      "step": 188465
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.520882487297058,
      "learning_rate": 4.772471371658626e-05,
      "loss": 2.0259,
      "step": 188466
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.3404613733291626,
      "learning_rate": 4.7722500078528034e-05,
      "loss": 1.881,
      "step": 188467
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.3099541664123535,
      "learning_rate": 4.772028648737296e-05,
      "loss": 1.8278,
      "step": 188468
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.4561879634857178,
      "learning_rate": 4.7718072943121466e-05,
      "loss": 2.083,
      "step": 188469
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.2925430536270142,
      "learning_rate": 4.771585944577385e-05,
      "loss": 2.1499,
      "step": 188470
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.2607510089874268,
      "learning_rate": 4.7713645995330717e-05,
      "loss": 1.8937,
      "step": 188471
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.389405608177185,
      "learning_rate": 4.77114325917923e-05,
      "loss": 1.9982,
      "step": 188472
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.4448720216751099,
      "learning_rate": 4.7709219235159154e-05,
      "loss": 2.0039,
      "step": 188473
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.4590256214141846,
      "learning_rate": 4.770700592543163e-05,
      "loss": 1.9829,
      "step": 188474
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.6439951658248901,
      "learning_rate": 4.770479266261011e-05,
      "loss": 2.0526,
      "step": 188475
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.4410600662231445,
      "learning_rate": 4.770257944669498e-05,
      "loss": 2.1868,
      "step": 188476
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.3404409885406494,
      "learning_rate": 4.7700366277686784e-05,
      "loss": 1.8222,
      "step": 188477
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.3654536008834839,
      "learning_rate": 4.769815315558577e-05,
      "loss": 1.7433,
      "step": 188478
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.5551412105560303,
      "learning_rate": 4.76959400803925e-05,
      "loss": 2.1768,
      "step": 188479
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.4372541904449463,
      "learning_rate": 4.769372705210735e-05,
      "loss": 2.0097,
      "step": 188480
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.3942846059799194,
      "learning_rate": 4.769151407073066e-05,
      "loss": 2.1884,
      "step": 188481
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.416934847831726,
      "learning_rate": 4.768930113626283e-05,
      "loss": 2.1977,
      "step": 188482
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.3848105669021606,
      "learning_rate": 4.76870882487044e-05,
      "loss": 2.1473,
      "step": 188483
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.3453810214996338,
      "learning_rate": 4.7684875408055615e-05,
      "loss": 1.9415,
      "step": 188484
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.4460487365722656,
      "learning_rate": 4.7682662614317034e-05,
      "loss": 1.7956,
      "step": 188485
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.49404776096344,
      "learning_rate": 4.7680449867489036e-05,
      "loss": 1.8265,
      "step": 188486
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.4178801774978638,
      "learning_rate": 4.767823716757198e-05,
      "loss": 2.0962,
      "step": 188487
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.4343973398208618,
      "learning_rate": 4.7676024514566255e-05,
      "loss": 1.9584,
      "step": 188488
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.3650388717651367,
      "learning_rate": 4.767381190847238e-05,
      "loss": 1.9391,
      "step": 188489
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.514085292816162,
      "learning_rate": 4.7671599349290614e-05,
      "loss": 2.1458,
      "step": 188490
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.3416008949279785,
      "learning_rate": 4.766938683702155e-05,
      "loss": 2.1733,
      "step": 188491
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.363232970237732,
      "learning_rate": 4.766717437166553e-05,
      "loss": 1.8873,
      "step": 188492
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.44879949092865,
      "learning_rate": 4.7664961953222844e-05,
      "loss": 2.2302,
      "step": 188493
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.4648770093917847,
      "learning_rate": 4.7662749581694104e-05,
      "loss": 1.9319,
      "step": 188494
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.4462933540344238,
      "learning_rate": 4.76605372570796e-05,
      "loss": 2.0531,
      "step": 188495
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.511307954788208,
      "learning_rate": 4.76583249793797e-05,
      "loss": 2.211,
      "step": 188496
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.796109676361084,
      "learning_rate": 4.7656112748594944e-05,
      "loss": 2.1365,
      "step": 188497
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.3217381238937378,
      "learning_rate": 4.765390056472569e-05,
      "loss": 1.9914,
      "step": 188498
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.3367379903793335,
      "learning_rate": 4.765168842777227e-05,
      "loss": 2.072,
      "step": 188499
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.3980298042297363,
      "learning_rate": 4.764947633773526e-05,
      "loss": 2.0721,
      "step": 188500
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.4730910062789917,
      "learning_rate": 4.764726429461492e-05,
      "loss": 1.9869,
      "step": 188501
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.4505401849746704,
      "learning_rate": 4.764505229841168e-05,
      "loss": 1.9136,
      "step": 188502
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.2455573081970215,
      "learning_rate": 4.7642840349126076e-05,
      "loss": 2.0525,
      "step": 188503
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.4117294549942017,
      "learning_rate": 4.764062844675831e-05,
      "loss": 1.8921,
      "step": 188504
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.359869360923767,
      "learning_rate": 4.7638416591309016e-05,
      "loss": 1.8302,
      "step": 188505
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.3930772542953491,
      "learning_rate": 4.763620478277852e-05,
      "loss": 2.1101,
      "step": 188506
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.2956929206848145,
      "learning_rate": 4.763399302116713e-05,
      "loss": 1.8041,
      "step": 188507
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.3621107339859009,
      "learning_rate": 4.76317813064754e-05,
      "loss": 2.1386,
      "step": 188508
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.1902532577514648,
      "learning_rate": 4.7629569638703714e-05,
      "loss": 1.7935,
      "step": 188509
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.3885157108306885,
      "learning_rate": 4.762735801785236e-05,
      "loss": 2.1505,
      "step": 188510
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.364298939704895,
      "learning_rate": 4.7625146443921936e-05,
      "loss": 1.9917,
      "step": 188511
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.6660962104797363,
      "learning_rate": 4.7622934916912746e-05,
      "loss": 1.7985,
      "step": 188512
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.3411585092544556,
      "learning_rate": 4.7620723436825156e-05,
      "loss": 2.075,
      "step": 188513
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.4034522771835327,
      "learning_rate": 4.76185120036597e-05,
      "loss": 2.0209,
      "step": 188514
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.3178319931030273,
      "learning_rate": 4.761630061741665e-05,
      "loss": 2.0175,
      "step": 188515
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.4884744882583618,
      "learning_rate": 4.761408927809659e-05,
      "loss": 1.8858,
      "step": 188516
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.2794811725616455,
      "learning_rate": 4.76118779856998e-05,
      "loss": 1.9135,
      "step": 188517
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.4332059621810913,
      "learning_rate": 4.760966674022674e-05,
      "loss": 2.079,
      "step": 188518
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.509407877922058,
      "learning_rate": 4.7607455541677744e-05,
      "loss": 1.8703,
      "step": 188519
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.311726450920105,
      "learning_rate": 4.7605244390053376e-05,
      "loss": 1.807,
      "step": 188520
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.5691168308258057,
      "learning_rate": 4.760303328535384e-05,
      "loss": 2.2905,
      "step": 188521
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.4303346872329712,
      "learning_rate": 4.760082222757977e-05,
      "loss": 2.0075,
      "step": 188522
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.4753438234329224,
      "learning_rate": 4.7598611216731464e-05,
      "loss": 2.0468,
      "step": 188523
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.3196651935577393,
      "learning_rate": 4.759640025280932e-05,
      "loss": 1.8707,
      "step": 188524
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.4435546398162842,
      "learning_rate": 4.7594189335813736e-05,
      "loss": 1.9195,
      "step": 188525
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.3500158786773682,
      "learning_rate": 4.759197846574522e-05,
      "loss": 2.0249,
      "step": 188526
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.4107599258422852,
      "learning_rate": 4.7589767642604025e-05,
      "loss": 1.9374,
      "step": 188527
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.3320996761322021,
      "learning_rate": 4.758755686639073e-05,
      "loss": 1.948,
      "step": 188528
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.3605822324752808,
      "learning_rate": 4.75853461371057e-05,
      "loss": 1.8813,
      "step": 188529
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.3555728197097778,
      "learning_rate": 4.7583135454749286e-05,
      "loss": 2.1492,
      "step": 188530
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.2983646392822266,
      "learning_rate": 4.7580924819321874e-05,
      "loss": 1.8538,
      "step": 188531
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.321803331375122,
      "learning_rate": 4.7578714230824024e-05,
      "loss": 2.0355,
      "step": 188532
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.3322200775146484,
      "learning_rate": 4.757650368925596e-05,
      "loss": 2.1645,
      "step": 188533
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.458465576171875,
      "learning_rate": 4.7574293194618265e-05,
      "loss": 1.9597,
      "step": 188534
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.4928944110870361,
      "learning_rate": 4.7572082746911256e-05,
      "loss": 2.0436,
      "step": 188535
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.5660054683685303,
      "learning_rate": 4.756987234613541e-05,
      "loss": 2.0324,
      "step": 188536
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.3290334939956665,
      "learning_rate": 4.756766199229098e-05,
      "loss": 1.9214,
      "step": 188537
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.3121986389160156,
      "learning_rate": 4.7565451685378556e-05,
      "loss": 1.9034,
      "step": 188538
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.293560266494751,
      "learning_rate": 4.7563241425398416e-05,
      "loss": 2.0376,
      "step": 188539
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.3532507419586182,
      "learning_rate": 4.756103121235113e-05,
      "loss": 1.9411,
      "step": 188540
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.3635138273239136,
      "learning_rate": 4.7558821046236974e-05,
      "loss": 1.9615,
      "step": 188541
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.2896195650100708,
      "learning_rate": 4.755661092705644e-05,
      "loss": 1.9969,
      "step": 188542
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.468924880027771,
      "learning_rate": 4.7554400854809794e-05,
      "loss": 1.999,
      "step": 188543
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.4180076122283936,
      "learning_rate": 4.755219082949764e-05,
      "loss": 2.0195,
      "step": 188544
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.3327479362487793,
      "learning_rate": 4.754998085112021e-05,
      "loss": 1.9416,
      "step": 188545
    },
    {
      "epoch": 2.45,
      "grad_norm": 1.5338152647018433,
      "learning_rate": 4.754777091967806e-05,
      "loss": 1.9488,
      "step": 188546
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.2892292737960815,
      "learning_rate": 4.7545561035171574e-05,
      "loss": 2.0792,
      "step": 188547
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.5728424787521362,
      "learning_rate": 4.754335119760114e-05,
      "loss": 1.997,
      "step": 188548
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.512948989868164,
      "learning_rate": 4.75411414069671e-05,
      "loss": 2.01,
      "step": 188549
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.2279106378555298,
      "learning_rate": 4.7538931663269974e-05,
      "loss": 1.913,
      "step": 188550
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.3852410316467285,
      "learning_rate": 4.753672196651004e-05,
      "loss": 1.8355,
      "step": 188551
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.4769010543823242,
      "learning_rate": 4.753451231668789e-05,
      "loss": 2.078,
      "step": 188552
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.3200565576553345,
      "learning_rate": 4.7532302713803824e-05,
      "loss": 1.9872,
      "step": 188553
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.5841659307479858,
      "learning_rate": 4.7530093157858286e-05,
      "loss": 2.1963,
      "step": 188554
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.4245704412460327,
      "learning_rate": 4.7527883648851594e-05,
      "loss": 2.0821,
      "step": 188555
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.3672935962677002,
      "learning_rate": 4.7525674186784324e-05,
      "loss": 1.8792,
      "step": 188556
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.450383186340332,
      "learning_rate": 4.7523464771656706e-05,
      "loss": 2.0188,
      "step": 188557
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.3993959426879883,
      "learning_rate": 4.75212554034693e-05,
      "loss": 2.1946,
      "step": 188558
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.7509347200393677,
      "learning_rate": 4.751904608222245e-05,
      "loss": 2.21,
      "step": 188559
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.2883306741714478,
      "learning_rate": 4.7516836807916624e-05,
      "loss": 2.1341,
      "step": 188560
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.4423062801361084,
      "learning_rate": 4.751462758055208e-05,
      "loss": 2.1018,
      "step": 188561
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.3316450119018555,
      "learning_rate": 4.751241840012942e-05,
      "loss": 1.9682,
      "step": 188562
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.5507320165634155,
      "learning_rate": 4.751020926664887e-05,
      "loss": 2.1131,
      "step": 188563
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.2724313735961914,
      "learning_rate": 4.7508000180111005e-05,
      "loss": 1.9552,
      "step": 188564
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.4791221618652344,
      "learning_rate": 4.750579114051619e-05,
      "loss": 1.978,
      "step": 188565
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.432511806488037,
      "learning_rate": 4.750358214786483e-05,
      "loss": 2.2249,
      "step": 188566
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.4376319646835327,
      "learning_rate": 4.7501373202157245e-05,
      "loss": 2.0825,
      "step": 188567
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.343284010887146,
      "learning_rate": 4.749916430339398e-05,
      "loss": 1.9649,
      "step": 188568
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.3581726551055908,
      "learning_rate": 4.7496955451575293e-05,
      "loss": 2.1615,
      "step": 188569
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.3371752500534058,
      "learning_rate": 4.7494746646701785e-05,
      "loss": 1.9863,
      "step": 188570
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.4585516452789307,
      "learning_rate": 4.7492537888773695e-05,
      "loss": 1.9728,
      "step": 188571
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.367663860321045,
      "learning_rate": 4.7490329177791654e-05,
      "loss": 2.2447,
      "step": 188572
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.3153831958770752,
      "learning_rate": 4.748812051375579e-05,
      "loss": 1.8253,
      "step": 188573
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.2576228380203247,
      "learning_rate": 4.748591189666671e-05,
      "loss": 1.9644,
      "step": 188574
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.5077533721923828,
      "learning_rate": 4.7483703326524705e-05,
      "loss": 2.151,
      "step": 188575
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.3606030941009521,
      "learning_rate": 4.748149480333031e-05,
      "loss": 1.8028,
      "step": 188576
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.424999713897705,
      "learning_rate": 4.74792863270838e-05,
      "loss": 1.9505,
      "step": 188577
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.3940554857254028,
      "learning_rate": 4.7477077897785796e-05,
      "loss": 1.8437,
      "step": 188578
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.471816062927246,
      "learning_rate": 4.7474869515436445e-05,
      "loss": 2.0806,
      "step": 188579
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.4614708423614502,
      "learning_rate": 4.747266118003633e-05,
      "loss": 2.0368,
      "step": 188580
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.4603019952774048,
      "learning_rate": 4.747045289158576e-05,
      "loss": 2.3072,
      "step": 188581
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.2598756551742554,
      "learning_rate": 4.746824465008527e-05,
      "loss": 1.8344,
      "step": 188582
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.382022738456726,
      "learning_rate": 4.7466036455535126e-05,
      "loss": 1.9675,
      "step": 188583
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.4494229555130005,
      "learning_rate": 4.746382830793586e-05,
      "loss": 2.1147,
      "step": 188584
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.4039521217346191,
      "learning_rate": 4.746162020728786e-05,
      "loss": 1.878,
      "step": 188585
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.341583251953125,
      "learning_rate": 4.7459412153591516e-05,
      "loss": 1.9626,
      "step": 188586
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.3454794883728027,
      "learning_rate": 4.745720414684717e-05,
      "loss": 1.9004,
      "step": 188587
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.4153329133987427,
      "learning_rate": 4.7454996187055335e-05,
      "loss": 2.112,
      "step": 188588
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.3469210863113403,
      "learning_rate": 4.745278827421635e-05,
      "loss": 2.0493,
      "step": 188589
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.2809542417526245,
      "learning_rate": 4.74505804083307e-05,
      "loss": 1.978,
      "step": 188590
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.3591554164886475,
      "learning_rate": 4.744837258939875e-05,
      "loss": 2.227,
      "step": 188591
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.4085184335708618,
      "learning_rate": 4.744616481742089e-05,
      "loss": 1.9939,
      "step": 188592
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.3848402500152588,
      "learning_rate": 4.744395709239759e-05,
      "loss": 2.0902,
      "step": 188593
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.4038541316986084,
      "learning_rate": 4.744174941432923e-05,
      "loss": 2.0116,
      "step": 188594
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.3049410581588745,
      "learning_rate": 4.743954178321615e-05,
      "loss": 2.0862,
      "step": 188595
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.718634009361267,
      "learning_rate": 4.7437334199058875e-05,
      "loss": 1.8839,
      "step": 188596
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.5011951923370361,
      "learning_rate": 4.74351266618578e-05,
      "loss": 2.1273,
      "step": 188597
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.5168260335922241,
      "learning_rate": 4.74329191716132e-05,
      "loss": 2.0043,
      "step": 188598
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.4034174680709839,
      "learning_rate": 4.743071172832571e-05,
      "loss": 1.9907,
      "step": 188599
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.3117612600326538,
      "learning_rate": 4.7428504331995586e-05,
      "loss": 2.1079,
      "step": 188600
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.4930075407028198,
      "learning_rate": 4.742629698262317e-05,
      "loss": 2.0262,
      "step": 188601
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.4132226705551147,
      "learning_rate": 4.7424089680209085e-05,
      "loss": 2.0325,
      "step": 188602
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.3694894313812256,
      "learning_rate": 4.742188242475361e-05,
      "loss": 2.0578,
      "step": 188603
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.3281667232513428,
      "learning_rate": 4.7419675216257135e-05,
      "loss": 1.7401,
      "step": 188604
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.5425924062728882,
      "learning_rate": 4.7417468054720135e-05,
      "loss": 2.0793,
      "step": 188605
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.4337992668151855,
      "learning_rate": 4.741526094014296e-05,
      "loss": 2.1345,
      "step": 188606
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.3787025213241577,
      "learning_rate": 4.74130538725261e-05,
      "loss": 2.0993,
      "step": 188607
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.4612902402877808,
      "learning_rate": 4.7410846851869935e-05,
      "loss": 2.1085,
      "step": 188608
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.432020664215088,
      "learning_rate": 4.740863987817487e-05,
      "loss": 2.1108,
      "step": 188609
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.5310288667678833,
      "learning_rate": 4.7406432951441245e-05,
      "loss": 2.0015,
      "step": 188610
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.4060477018356323,
      "learning_rate": 4.740422607166958e-05,
      "loss": 2.2283,
      "step": 188611
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.4646531343460083,
      "learning_rate": 4.740201923886016e-05,
      "loss": 2.2049,
      "step": 188612
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.5806323289871216,
      "learning_rate": 4.739981245301356e-05,
      "loss": 2.0965,
      "step": 188613
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.2650152444839478,
      "learning_rate": 4.7397605714130105e-05,
      "loss": 2.283,
      "step": 188614
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.3455231189727783,
      "learning_rate": 4.7395399022210214e-05,
      "loss": 2.1066,
      "step": 188615
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.3113726377487183,
      "learning_rate": 4.739319237725419e-05,
      "loss": 2.1398,
      "step": 188616
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.3687114715576172,
      "learning_rate": 4.739098577926263e-05,
      "loss": 2.1259,
      "step": 188617
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.400853157043457,
      "learning_rate": 4.7388779228235764e-05,
      "loss": 2.0986,
      "step": 188618
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.4341368675231934,
      "learning_rate": 4.738657272417417e-05,
      "loss": 2.192,
      "step": 188619
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.2984400987625122,
      "learning_rate": 4.738436626707817e-05,
      "loss": 2.0056,
      "step": 188620
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.3500021696090698,
      "learning_rate": 4.7382159856948196e-05,
      "loss": 1.8923,
      "step": 188621
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.4516268968582153,
      "learning_rate": 4.7379953493784594e-05,
      "loss": 1.7968,
      "step": 188622
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.4232991933822632,
      "learning_rate": 4.7377747177587885e-05,
      "loss": 1.7389,
      "step": 188623
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.6020742654800415,
      "learning_rate": 4.7375540908358346e-05,
      "loss": 1.9546,
      "step": 188624
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.488913655281067,
      "learning_rate": 4.7373334686096534e-05,
      "loss": 1.9842,
      "step": 188625
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.3352299928665161,
      "learning_rate": 4.737112851080278e-05,
      "loss": 2.1514,
      "step": 188626
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.4961453676223755,
      "learning_rate": 4.7368922382477524e-05,
      "loss": 2.0132,
      "step": 188627
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.3658246994018555,
      "learning_rate": 4.7366716301121065e-05,
      "loss": 2.15,
      "step": 188628
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.350420355796814,
      "learning_rate": 4.7364510266733966e-05,
      "loss": 1.9298,
      "step": 188629
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.4773200750350952,
      "learning_rate": 4.736230427931653e-05,
      "loss": 1.9053,
      "step": 188630
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.4211061000823975,
      "learning_rate": 4.7360098338869254e-05,
      "loss": 1.9473,
      "step": 188631
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.2624554634094238,
      "learning_rate": 4.7357892445392544e-05,
      "loss": 2.0979,
      "step": 188632
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.2569113969802856,
      "learning_rate": 4.7355686598886724e-05,
      "loss": 1.7325,
      "step": 188633
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.4678709506988525,
      "learning_rate": 4.7353480799352194e-05,
      "loss": 1.9864,
      "step": 188634
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.3658828735351562,
      "learning_rate": 4.73512750467895e-05,
      "loss": 1.8022,
      "step": 188635
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.4963643550872803,
      "learning_rate": 4.734906934119892e-05,
      "loss": 2.2156,
      "step": 188636
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.3253462314605713,
      "learning_rate": 4.7346863682580936e-05,
      "loss": 1.8281,
      "step": 188637
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.3295066356658936,
      "learning_rate": 4.734465807093591e-05,
      "loss": 2.147,
      "step": 188638
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.4036904573440552,
      "learning_rate": 4.7342452506264415e-05,
      "loss": 1.8736,
      "step": 188639
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.36783766746521,
      "learning_rate": 4.734024698856658e-05,
      "loss": 1.8771,
      "step": 188640
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.3748350143432617,
      "learning_rate": 4.7338041517843064e-05,
      "loss": 1.9313,
      "step": 188641
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.5094047784805298,
      "learning_rate": 4.733583609409407e-05,
      "loss": 1.9123,
      "step": 188642
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.4468772411346436,
      "learning_rate": 4.733363071732017e-05,
      "loss": 2.0067,
      "step": 188643
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.3188196420669556,
      "learning_rate": 4.7331425387521694e-05,
      "loss": 1.8975,
      "step": 188644
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.362899899482727,
      "learning_rate": 4.732922010469917e-05,
      "loss": 1.9355,
      "step": 188645
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.2637155055999756,
      "learning_rate": 4.732701486885281e-05,
      "loss": 1.9386,
      "step": 188646
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.408552885055542,
      "learning_rate": 4.7324809679983165e-05,
      "loss": 2.0426,
      "step": 188647
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.5265353918075562,
      "learning_rate": 4.732260453809058e-05,
      "loss": 1.9975,
      "step": 188648
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.4248356819152832,
      "learning_rate": 4.732039944317552e-05,
      "loss": 2.1147,
      "step": 188649
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.2119767665863037,
      "learning_rate": 4.7318194395238306e-05,
      "loss": 1.9034,
      "step": 188650
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.3567349910736084,
      "learning_rate": 4.731598939427959e-05,
      "loss": 1.8328,
      "step": 188651
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.384914755821228,
      "learning_rate": 4.731378444029942e-05,
      "loss": 2.0072,
      "step": 188652
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.288510799407959,
      "learning_rate": 4.7311579533298475e-05,
      "loss": 2.096,
      "step": 188653
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.5240933895111084,
      "learning_rate": 4.730937467327698e-05,
      "loss": 2.0657,
      "step": 188654
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.3224895000457764,
      "learning_rate": 4.7307169860235536e-05,
      "loss": 1.9036,
      "step": 188655
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.3441545963287354,
      "learning_rate": 4.730496509417438e-05,
      "loss": 1.8922,
      "step": 188656
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.3823535442352295,
      "learning_rate": 4.730276037509415e-05,
      "loss": 2.1667,
      "step": 188657
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.2638734579086304,
      "learning_rate": 4.730055570299497e-05,
      "loss": 2.0822,
      "step": 188658
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.4292078018188477,
      "learning_rate": 4.7298351077877436e-05,
      "loss": 2.0699,
      "step": 188659
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.5825740098953247,
      "learning_rate": 4.729614649974186e-05,
      "loss": 2.2674,
      "step": 188660
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.3210277557373047,
      "learning_rate": 4.7293941968588765e-05,
      "loss": 1.9279,
      "step": 188661
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.554344892501831,
      "learning_rate": 4.729173748441842e-05,
      "loss": 2.0583,
      "step": 188662
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.470475673675537,
      "learning_rate": 4.728953304723143e-05,
      "loss": 1.9425,
      "step": 188663
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.4457743167877197,
      "learning_rate": 4.728732865702799e-05,
      "loss": 2.2317,
      "step": 188664
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.4649935960769653,
      "learning_rate": 4.728512431380866e-05,
      "loss": 2.2225,
      "step": 188665
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.4562773704528809,
      "learning_rate": 4.728292001757372e-05,
      "loss": 1.8404,
      "step": 188666
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.3499785661697388,
      "learning_rate": 4.728071576832373e-05,
      "loss": 1.8019,
      "step": 188667
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.35597825050354,
      "learning_rate": 4.7278511566058954e-05,
      "loss": 2.0857,
      "step": 188668
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.4548554420471191,
      "learning_rate": 4.727630741077992e-05,
      "loss": 2.1211,
      "step": 188669
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.451148509979248,
      "learning_rate": 4.727410330248701e-05,
      "loss": 1.9757,
      "step": 188670
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.570523977279663,
      "learning_rate": 4.7271899241180646e-05,
      "loss": 1.9248,
      "step": 188671
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.4652570486068726,
      "learning_rate": 4.726969522686109e-05,
      "loss": 1.9237,
      "step": 188672
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.3434455394744873,
      "learning_rate": 4.726749125952895e-05,
      "loss": 1.9264,
      "step": 188673
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.3896534442901611,
      "learning_rate": 4.72652873391845e-05,
      "loss": 2.1664,
      "step": 188674
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.3281177282333374,
      "learning_rate": 4.7263083465828255e-05,
      "loss": 1.9291,
      "step": 188675
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.385912299156189,
      "learning_rate": 4.7260879639460594e-05,
      "loss": 2.0529,
      "step": 188676
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.3650344610214233,
      "learning_rate": 4.725867586008184e-05,
      "loss": 1.89,
      "step": 188677
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.2823877334594727,
      "learning_rate": 4.72564721276925e-05,
      "loss": 1.8915,
      "step": 188678
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.4838511943817139,
      "learning_rate": 4.7254268442292977e-05,
      "loss": 1.9395,
      "step": 188679
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.5524818897247314,
      "learning_rate": 4.7252064803883594e-05,
      "loss": 1.9232,
      "step": 188680
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.361480712890625,
      "learning_rate": 4.724986121246489e-05,
      "loss": 1.9828,
      "step": 188681
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.4170558452606201,
      "learning_rate": 4.72476576680372e-05,
      "loss": 2.0299,
      "step": 188682
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.4815211296081543,
      "learning_rate": 4.724545417060088e-05,
      "loss": 2.0997,
      "step": 188683
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.536192536354065,
      "learning_rate": 4.724325072015648e-05,
      "loss": 1.8356,
      "step": 188684
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.4707255363464355,
      "learning_rate": 4.724104731670432e-05,
      "loss": 2.0244,
      "step": 188685
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.4073071479797363,
      "learning_rate": 4.723884396024473e-05,
      "loss": 1.9254,
      "step": 188686
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.3407763242721558,
      "learning_rate": 4.723664065077829e-05,
      "loss": 2.1246,
      "step": 188687
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.367842674255371,
      "learning_rate": 4.723443738830536e-05,
      "loss": 1.9,
      "step": 188688
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.3544808626174927,
      "learning_rate": 4.72322341728262e-05,
      "loss": 2.1153,
      "step": 188689
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.3254690170288086,
      "learning_rate": 4.723003100434145e-05,
      "loss": 1.9988,
      "step": 188690
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.3187237977981567,
      "learning_rate": 4.722782788285131e-05,
      "loss": 1.9228,
      "step": 188691
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.3011248111724854,
      "learning_rate": 4.7225624808356376e-05,
      "loss": 1.9247,
      "step": 188692
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.3170417547225952,
      "learning_rate": 4.7223421780856984e-05,
      "loss": 2.0854,
      "step": 188693
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.5763046741485596,
      "learning_rate": 4.72212188003535e-05,
      "loss": 2.2163,
      "step": 188694
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.433992862701416,
      "learning_rate": 4.721901586684632e-05,
      "loss": 1.9484,
      "step": 188695
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.2944408655166626,
      "learning_rate": 4.721681298033595e-05,
      "loss": 2.0285,
      "step": 188696
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.3579181432724,
      "learning_rate": 4.7214610140822684e-05,
      "loss": 2.0707,
      "step": 188697
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.2876018285751343,
      "learning_rate": 4.721240734830706e-05,
      "loss": 1.9711,
      "step": 188698
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.3910329341888428,
      "learning_rate": 4.7210204602789406e-05,
      "loss": 1.976,
      "step": 188699
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.5986404418945312,
      "learning_rate": 4.720800190427016e-05,
      "loss": 1.9811,
      "step": 188700
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.35708749294281,
      "learning_rate": 4.720579925274964e-05,
      "loss": 1.7882,
      "step": 188701
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.406575083732605,
      "learning_rate": 4.7203596648228405e-05,
      "loss": 2.0058,
      "step": 188702
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.282830834388733,
      "learning_rate": 4.7201394090706734e-05,
      "loss": 2.057,
      "step": 188703
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.2935855388641357,
      "learning_rate": 4.719919158018517e-05,
      "loss": 1.8052,
      "step": 188704
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.5260438919067383,
      "learning_rate": 4.7196989116663974e-05,
      "loss": 1.8646,
      "step": 188705
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.3426834344863892,
      "learning_rate": 4.719478670014375e-05,
      "loss": 2.0508,
      "step": 188706
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.445906400680542,
      "learning_rate": 4.719258433062466e-05,
      "loss": 1.9099,
      "step": 188707
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.4061453342437744,
      "learning_rate": 4.7190382008107344e-05,
      "loss": 2.2376,
      "step": 188708
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.483770489692688,
      "learning_rate": 4.718817973259199e-05,
      "loss": 2.1038,
      "step": 188709
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.4872335195541382,
      "learning_rate": 4.718597750407921e-05,
      "loss": 2.2149,
      "step": 188710
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.57893967628479,
      "learning_rate": 4.718377532256926e-05,
      "loss": 2.29,
      "step": 188711
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.3266708850860596,
      "learning_rate": 4.718157318806278e-05,
      "loss": 2.0441,
      "step": 188712
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.365773320198059,
      "learning_rate": 4.717937110055984e-05,
      "loss": 2.1474,
      "step": 188713
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.4390103816986084,
      "learning_rate": 4.717716906006113e-05,
      "loss": 2.1407,
      "step": 188714
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.326058268547058,
      "learning_rate": 4.717496706656685e-05,
      "loss": 1.9014,
      "step": 188715
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.429516315460205,
      "learning_rate": 4.717276512007764e-05,
      "loss": 2.0263,
      "step": 188716
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.4319076538085938,
      "learning_rate": 4.717056322059366e-05,
      "loss": 1.83,
      "step": 188717
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.489168643951416,
      "learning_rate": 4.7168361368115613e-05,
      "loss": 1.8581,
      "step": 188718
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.3789808750152588,
      "learning_rate": 4.7166159562643604e-05,
      "loss": 2.1041,
      "step": 188719
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.2869795560836792,
      "learning_rate": 4.7163957804178223e-05,
      "loss": 1.9805,
      "step": 188720
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.3772368431091309,
      "learning_rate": 4.7161756092719804e-05,
      "loss": 2.1953,
      "step": 188721
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.4009004831314087,
      "learning_rate": 4.715955442826882e-05,
      "loss": 2.0887,
      "step": 188722
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.447507619857788,
      "learning_rate": 4.71573528108256e-05,
      "loss": 1.8243,
      "step": 188723
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.3171519041061401,
      "learning_rate": 4.715515124039074e-05,
      "loss": 2.048,
      "step": 188724
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.4653396606445312,
      "learning_rate": 4.715294971696435e-05,
      "loss": 2.0397,
      "step": 188725
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.5194454193115234,
      "learning_rate": 4.7150748240547085e-05,
      "loss": 2.1288,
      "step": 188726
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.326917290687561,
      "learning_rate": 4.714854681113922e-05,
      "loss": 1.8524,
      "step": 188727
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.402974009513855,
      "learning_rate": 4.714634542874125e-05,
      "loss": 2.184,
      "step": 188728
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.314542293548584,
      "learning_rate": 4.7144144093353475e-05,
      "loss": 2.0241,
      "step": 188729
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.3083182573318481,
      "learning_rate": 4.7141942804976526e-05,
      "loss": 1.9518,
      "step": 188730
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.4616999626159668,
      "learning_rate": 4.7139741563610543e-05,
      "loss": 1.9796,
      "step": 188731
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.4196981191635132,
      "learning_rate": 4.713754036925612e-05,
      "loss": 2.0948,
      "step": 188732
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.3274914026260376,
      "learning_rate": 4.7135339221913494e-05,
      "loss": 2.0021,
      "step": 188733
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.3593062162399292,
      "learning_rate": 4.7133138121583294e-05,
      "loss": 2.0033,
      "step": 188734
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.3127026557922363,
      "learning_rate": 4.713093706826572e-05,
      "loss": 1.8678,
      "step": 188735
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.378101110458374,
      "learning_rate": 4.7128736061961436e-05,
      "loss": 2.1686,
      "step": 188736
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.414862871170044,
      "learning_rate": 4.712653510267055e-05,
      "loss": 1.9395,
      "step": 188737
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.28194260597229,
      "learning_rate": 4.7124334190393686e-05,
      "loss": 2.0133,
      "step": 188738
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.4498002529144287,
      "learning_rate": 4.712213332513109e-05,
      "loss": 2.2593,
      "step": 188739
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.5697094202041626,
      "learning_rate": 4.711993250688334e-05,
      "loss": 2.1747,
      "step": 188740
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.3378901481628418,
      "learning_rate": 4.711773173565072e-05,
      "loss": 1.8812,
      "step": 188741
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.4467540979385376,
      "learning_rate": 4.7115531011433825e-05,
      "loss": 2.0266,
      "step": 188742
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.2819069623947144,
      "learning_rate": 4.711333033423276e-05,
      "loss": 1.9077,
      "step": 188743
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.4386098384857178,
      "learning_rate": 4.711112970404818e-05,
      "loss": 1.9119,
      "step": 188744
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.551099419593811,
      "learning_rate": 4.7108929120880355e-05,
      "loss": 2.1662,
      "step": 188745
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.3074039220809937,
      "learning_rate": 4.7106728584729826e-05,
      "loss": 1.8549,
      "step": 188746
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.397971272468567,
      "learning_rate": 4.710452809559685e-05,
      "loss": 2.0764,
      "step": 188747
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.2844821214675903,
      "learning_rate": 4.710232765348206e-05,
      "loss": 1.9595,
      "step": 188748
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.4732179641723633,
      "learning_rate": 4.71001272583856e-05,
      "loss": 1.9359,
      "step": 188749
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.629758596420288,
      "learning_rate": 4.709792691030803e-05,
      "loss": 2.0768,
      "step": 188750
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.3635309934616089,
      "learning_rate": 4.709572660924967e-05,
      "loss": 1.7658,
      "step": 188751
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.311057686805725,
      "learning_rate": 4.7093526355211076e-05,
      "loss": 1.9105,
      "step": 188752
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.4388771057128906,
      "learning_rate": 4.70913261481925e-05,
      "loss": 1.8071,
      "step": 188753
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.3432188034057617,
      "learning_rate": 4.7089125988194476e-05,
      "loss": 1.9248,
      "step": 188754
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.3478448390960693,
      "learning_rate": 4.708692587521737e-05,
      "loss": 1.9215,
      "step": 188755
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.4649630784988403,
      "learning_rate": 4.708472580926156e-05,
      "loss": 1.9862,
      "step": 188756
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.3957797288894653,
      "learning_rate": 4.708252579032743e-05,
      "loss": 1.887,
      "step": 188757
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.285706877708435,
      "learning_rate": 4.708032581841549e-05,
      "loss": 2.217,
      "step": 188758
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.3626439571380615,
      "learning_rate": 4.707812589352603e-05,
      "loss": 2.1477,
      "step": 188759
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.5226784944534302,
      "learning_rate": 4.707592601565956e-05,
      "loss": 1.8987,
      "step": 188760
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.4212474822998047,
      "learning_rate": 4.707372618481647e-05,
      "loss": 2.0761,
      "step": 188761
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.3143335580825806,
      "learning_rate": 4.707152640099717e-05,
      "loss": 1.9083,
      "step": 188762
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.2948719263076782,
      "learning_rate": 4.7069326664201954e-05,
      "loss": 1.6921,
      "step": 188763
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.463191032409668,
      "learning_rate": 4.706712697443139e-05,
      "loss": 2.1387,
      "step": 188764
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.463356614112854,
      "learning_rate": 4.706492733168574e-05,
      "loss": 1.8884,
      "step": 188765
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.407976746559143,
      "learning_rate": 4.706272773596561e-05,
      "loss": 2.0626,
      "step": 188766
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.3784350156784058,
      "learning_rate": 4.7060528187271264e-05,
      "loss": 1.8806,
      "step": 188767
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.408164381980896,
      "learning_rate": 4.705832868560306e-05,
      "loss": 1.8937,
      "step": 188768
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.2349412441253662,
      "learning_rate": 4.705612923096158e-05,
      "loss": 1.9149,
      "step": 188769
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.4134327173233032,
      "learning_rate": 4.705392982334715e-05,
      "loss": 1.9699,
      "step": 188770
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.3542262315750122,
      "learning_rate": 4.705173046276006e-05,
      "loss": 1.9884,
      "step": 188771
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.4391764402389526,
      "learning_rate": 4.7049531149200926e-05,
      "loss": 2.0625,
      "step": 188772
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.3738895654678345,
      "learning_rate": 4.704733188267003e-05,
      "loss": 2.1674,
      "step": 188773
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.314780354499817,
      "learning_rate": 4.704513266316776e-05,
      "loss": 1.9299,
      "step": 188774
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.417216420173645,
      "learning_rate": 4.704293349069467e-05,
      "loss": 2.2282,
      "step": 188775
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.6210441589355469,
      "learning_rate": 4.7040734365250986e-05,
      "loss": 2.1277,
      "step": 188776
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.4924739599227905,
      "learning_rate": 4.7038535286837255e-05,
      "loss": 1.9991,
      "step": 188777
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.3177392482757568,
      "learning_rate": 4.7036336255453835e-05,
      "loss": 2.1135,
      "step": 188778
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.325491189956665,
      "learning_rate": 4.703413727110116e-05,
      "loss": 1.9693,
      "step": 188779
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.444539189338684,
      "learning_rate": 4.703193833377953e-05,
      "loss": 1.9766,
      "step": 188780
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.4701207876205444,
      "learning_rate": 4.702973944348951e-05,
      "loss": 2.089,
      "step": 188781
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.4228076934814453,
      "learning_rate": 4.702754060023137e-05,
      "loss": 1.8394,
      "step": 188782
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.382377028465271,
      "learning_rate": 4.702534180400564e-05,
      "loss": 1.991,
      "step": 188783
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.3956917524337769,
      "learning_rate": 4.702314305481269e-05,
      "loss": 2.0618,
      "step": 188784
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.4362412691116333,
      "learning_rate": 4.702094435265292e-05,
      "loss": 1.8997,
      "step": 188785
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.3860281705856323,
      "learning_rate": 4.701874569752666e-05,
      "loss": 1.9264,
      "step": 188786
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.3904281854629517,
      "learning_rate": 4.701654708943444e-05,
      "loss": 2.1101,
      "step": 188787
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.3401528596878052,
      "learning_rate": 4.7014348528376563e-05,
      "loss": 2.0271,
      "step": 188788
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.4321306943893433,
      "learning_rate": 4.701215001435357e-05,
      "loss": 1.8916,
      "step": 188789
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.2259790897369385,
      "learning_rate": 4.700995154736571e-05,
      "loss": 1.8389,
      "step": 188790
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.4008971452713013,
      "learning_rate": 4.700775312741363e-05,
      "loss": 1.9337,
      "step": 188791
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.3255195617675781,
      "learning_rate": 4.700555475449742e-05,
      "loss": 2.0246,
      "step": 188792
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.2767831087112427,
      "learning_rate": 4.700335642861776e-05,
      "loss": 2.0304,
      "step": 188793
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.3383818864822388,
      "learning_rate": 4.700115814977483e-05,
      "loss": 1.8897,
      "step": 188794
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.2957085371017456,
      "learning_rate": 4.699895991796928e-05,
      "loss": 1.9607,
      "step": 188795
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.403020977973938,
      "learning_rate": 4.6996761733201305e-05,
      "loss": 2.0945,
      "step": 188796
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.4356056451797485,
      "learning_rate": 4.6994563595471566e-05,
      "loss": 1.8813,
      "step": 188797
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.377365231513977,
      "learning_rate": 4.6992365504780135e-05,
      "loss": 2.1925,
      "step": 188798
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.3209528923034668,
      "learning_rate": 4.699016746112772e-05,
      "loss": 1.9281,
      "step": 188799
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.4680306911468506,
      "learning_rate": 4.6987969464514495e-05,
      "loss": 1.8346,
      "step": 188800
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.4639837741851807,
      "learning_rate": 4.6985771514941084e-05,
      "loss": 1.7904,
      "step": 188801
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.3542754650115967,
      "learning_rate": 4.698357361240771e-05,
      "loss": 1.9905,
      "step": 188802
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.671555995941162,
      "learning_rate": 4.698137575691501e-05,
      "loss": 1.7799,
      "step": 188803
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.2436176538467407,
      "learning_rate": 4.6979177948463106e-05,
      "loss": 1.8981,
      "step": 188804
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.5300657749176025,
      "learning_rate": 4.6976980187052616e-05,
      "loss": 1.7493,
      "step": 188805
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.33640456199646,
      "learning_rate": 4.697478247268379e-05,
      "loss": 2.0253,
      "step": 188806
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.5199693441390991,
      "learning_rate": 4.697258480535724e-05,
      "loss": 1.9473,
      "step": 188807
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.2903679609298706,
      "learning_rate": 4.697038718507319e-05,
      "loss": 1.9683,
      "step": 188808
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.300469160079956,
      "learning_rate": 4.696818961183224e-05,
      "loss": 2.0003,
      "step": 188809
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.3636757135391235,
      "learning_rate": 4.696599208563453e-05,
      "loss": 1.9525,
      "step": 188810
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.5155842304229736,
      "learning_rate": 4.696379460648072e-05,
      "loss": 1.8708,
      "step": 188811
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.5140793323516846,
      "learning_rate": 4.6961597174371015e-05,
      "loss": 2.0676,
      "step": 188812
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.3149937391281128,
      "learning_rate": 4.6959399789306006e-05,
      "loss": 1.8251,
      "step": 188813
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.43379807472229,
      "learning_rate": 4.695720245128597e-05,
      "loss": 1.8782,
      "step": 188814
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.347438931465149,
      "learning_rate": 4.6955005160311496e-05,
      "loss": 2.0908,
      "step": 188815
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.3362706899642944,
      "learning_rate": 4.695280791638273e-05,
      "loss": 1.9028,
      "step": 188816
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.346371054649353,
      "learning_rate": 4.6950610719500256e-05,
      "loss": 2.1715,
      "step": 188817
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.3825654983520508,
      "learning_rate": 4.694841356966439e-05,
      "loss": 2.044,
      "step": 188818
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.5058842897415161,
      "learning_rate": 4.6946216466875655e-05,
      "loss": 2.042,
      "step": 188819
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.3426096439361572,
      "learning_rate": 4.694401941113435e-05,
      "loss": 2.065,
      "step": 188820
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.4556682109832764,
      "learning_rate": 4.6941822402441046e-05,
      "loss": 2.0631,
      "step": 188821
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.2156294584274292,
      "learning_rate": 4.6939625440795867e-05,
      "loss": 1.752,
      "step": 188822
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.4276208877563477,
      "learning_rate": 4.693742852619949e-05,
      "loss": 1.8934,
      "step": 188823
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.523900032043457,
      "learning_rate": 4.693523165865215e-05,
      "loss": 1.9454,
      "step": 188824
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.2993404865264893,
      "learning_rate": 4.69330348381544e-05,
      "loss": 2.0884,
      "step": 188825
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.3707557916641235,
      "learning_rate": 4.693083806470652e-05,
      "loss": 1.9562,
      "step": 188826
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.623781442642212,
      "learning_rate": 4.69286413383091e-05,
      "loss": 1.9612,
      "step": 188827
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.408280372619629,
      "learning_rate": 4.6926444658962284e-05,
      "loss": 2.0547,
      "step": 188828
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.4834219217300415,
      "learning_rate": 4.6924248026666656e-05,
      "loss": 1.9092,
      "step": 188829
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.397438645362854,
      "learning_rate": 4.692205144142256e-05,
      "loss": 2.014,
      "step": 188830
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.5679700374603271,
      "learning_rate": 4.691985490323049e-05,
      "loss": 2.1412,
      "step": 188831
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.4185807704925537,
      "learning_rate": 4.691765841209072e-05,
      "loss": 2.0878,
      "step": 188832
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.4560261964797974,
      "learning_rate": 4.691546196800388e-05,
      "loss": 2.1645,
      "step": 188833
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.3709979057312012,
      "learning_rate": 4.691326557097009e-05,
      "loss": 2.1108,
      "step": 188834
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.4834794998168945,
      "learning_rate": 4.6911069220989975e-05,
      "loss": 1.9163,
      "step": 188835
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.3855078220367432,
      "learning_rate": 4.690887291806381e-05,
      "loss": 1.9495,
      "step": 188836
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.3927202224731445,
      "learning_rate": 4.6906676662192113e-05,
      "loss": 1.9537,
      "step": 188837
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.3813068866729736,
      "learning_rate": 4.690448045337517e-05,
      "loss": 1.9748,
      "step": 188838
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.5163021087646484,
      "learning_rate": 4.690228429161356e-05,
      "loss": 2.2101,
      "step": 188839
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.464514970779419,
      "learning_rate": 4.690008817690757e-05,
      "loss": 2.1161,
      "step": 188840
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.3439034223556519,
      "learning_rate": 4.6897892109257637e-05,
      "loss": 1.912,
      "step": 188841
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.4068673849105835,
      "learning_rate": 4.689569608866407e-05,
      "loss": 1.8763,
      "step": 188842
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.2880579233169556,
      "learning_rate": 4.689350011512746e-05,
      "loss": 1.8357,
      "step": 188843
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.3429162502288818,
      "learning_rate": 4.689130418864807e-05,
      "loss": 2.0505,
      "step": 188844
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.4529707431793213,
      "learning_rate": 4.688910830922644e-05,
      "loss": 2.1069,
      "step": 188845
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.4312567710876465,
      "learning_rate": 4.688691247686287e-05,
      "loss": 1.9498,
      "step": 188846
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.3530141115188599,
      "learning_rate": 4.688471669155782e-05,
      "loss": 1.7631,
      "step": 188847
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.567339539527893,
      "learning_rate": 4.68825209533116e-05,
      "loss": 1.9302,
      "step": 188848
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.3226677179336548,
      "learning_rate": 4.688032526212476e-05,
      "loss": 1.8987,
      "step": 188849
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.2995939254760742,
      "learning_rate": 4.687812961799762e-05,
      "loss": 1.9386,
      "step": 188850
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.409506916999817,
      "learning_rate": 4.687593402093063e-05,
      "loss": 1.8334,
      "step": 188851
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.6061636209487915,
      "learning_rate": 4.6873738470924225e-05,
      "loss": 1.9729,
      "step": 188852
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.7796690464019775,
      "learning_rate": 4.687154296797868e-05,
      "loss": 2.0158,
      "step": 188853
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.3924506902694702,
      "learning_rate": 4.6869347512094555e-05,
      "loss": 1.982,
      "step": 188854
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.5199213027954102,
      "learning_rate": 4.686715210327219e-05,
      "loss": 2.0116,
      "step": 188855
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.4557826519012451,
      "learning_rate": 4.686495674151194e-05,
      "loss": 2.0533,
      "step": 188856
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.401843786239624,
      "learning_rate": 4.6862761426814374e-05,
      "loss": 2.0035,
      "step": 188857
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.4169723987579346,
      "learning_rate": 4.686056615917977e-05,
      "loss": 1.9759,
      "step": 188858
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.4128937721252441,
      "learning_rate": 4.685837093860848e-05,
      "loss": 2.015,
      "step": 188859
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.3436775207519531,
      "learning_rate": 4.6856175765101075e-05,
      "loss": 2.0639,
      "step": 188860
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.2594263553619385,
      "learning_rate": 4.685398063865793e-05,
      "loss": 1.85,
      "step": 188861
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.377238154411316,
      "learning_rate": 4.68517855592793e-05,
      "loss": 1.8763,
      "step": 188862
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.4335789680480957,
      "learning_rate": 4.684959052696575e-05,
      "loss": 2.1005,
      "step": 188863
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.4221707582473755,
      "learning_rate": 4.684739554171769e-05,
      "loss": 2.0297,
      "step": 188864
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.416425347328186,
      "learning_rate": 4.684520060353538e-05,
      "loss": 2.0359,
      "step": 188865
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.4642921686172485,
      "learning_rate": 4.684300571241939e-05,
      "loss": 1.9924,
      "step": 188866
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.4045860767364502,
      "learning_rate": 4.684081086836998e-05,
      "loss": 1.8814,
      "step": 188867
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.4148412942886353,
      "learning_rate": 4.683861607138776e-05,
      "loss": 1.9875,
      "step": 188868
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.4819844961166382,
      "learning_rate": 4.6836421321472987e-05,
      "loss": 2.1023,
      "step": 188869
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.432795524597168,
      "learning_rate": 4.6834226618626126e-05,
      "loss": 2.042,
      "step": 188870
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.3708076477050781,
      "learning_rate": 4.683203196284745e-05,
      "loss": 1.8218,
      "step": 188871
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.3344964981079102,
      "learning_rate": 4.682983735413759e-05,
      "loss": 2.1544,
      "step": 188872
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.4451040029525757,
      "learning_rate": 4.6827642792496745e-05,
      "loss": 1.9891,
      "step": 188873
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.362246036529541,
      "learning_rate": 4.6825448277925515e-05,
      "loss": 2.1653,
      "step": 188874
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.650557518005371,
      "learning_rate": 4.68232538104241e-05,
      "loss": 2.0557,
      "step": 188875
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.4133838415145874,
      "learning_rate": 4.68210593899932e-05,
      "loss": 2.1823,
      "step": 188876
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.412225604057312,
      "learning_rate": 4.681886501663291e-05,
      "loss": 2.1705,
      "step": 188877
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.4438064098358154,
      "learning_rate": 4.6816670690343805e-05,
      "loss": 1.9754,
      "step": 188878
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.27080237865448,
      "learning_rate": 4.681447641112621e-05,
      "loss": 2.2943,
      "step": 188879
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.3230088949203491,
      "learning_rate": 4.681228217898067e-05,
      "loss": 2.158,
      "step": 188880
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.267286777496338,
      "learning_rate": 4.6810087993907395e-05,
      "loss": 1.9273,
      "step": 188881
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.3880829811096191,
      "learning_rate": 4.680789385590708e-05,
      "loss": 1.947,
      "step": 188882
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.3502731323242188,
      "learning_rate": 4.68056997649798e-05,
      "loss": 2.0257,
      "step": 188883
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.471650242805481,
      "learning_rate": 4.68035057211262e-05,
      "loss": 1.8077,
      "step": 188884
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.4338458776474,
      "learning_rate": 4.6801311724346546e-05,
      "loss": 1.8447,
      "step": 188885
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.3485982418060303,
      "learning_rate": 4.679911777464137e-05,
      "loss": 2.0472,
      "step": 188886
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.4128016233444214,
      "learning_rate": 4.679692387201094e-05,
      "loss": 2.028,
      "step": 188887
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.47105872631073,
      "learning_rate": 4.679473001645588e-05,
      "loss": 1.9123,
      "step": 188888
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.3287729024887085,
      "learning_rate": 4.6792536207976304e-05,
      "loss": 1.9382,
      "step": 188889
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.463533639907837,
      "learning_rate": 4.679034244657287e-05,
      "loss": 1.9245,
      "step": 188890
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.6577712297439575,
      "learning_rate": 4.678814873224581e-05,
      "loss": 2.25,
      "step": 188891
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.328033208847046,
      "learning_rate": 4.67859550649957e-05,
      "loss": 2.0818,
      "step": 188892
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.434577465057373,
      "learning_rate": 4.6783761444822785e-05,
      "loss": 2.0329,
      "step": 188893
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.8121190071105957,
      "learning_rate": 4.678156787172769e-05,
      "loss": 1.9938,
      "step": 188894
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.3723098039627075,
      "learning_rate": 4.677937434571053e-05,
      "loss": 1.7969,
      "step": 188895
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.3927044868469238,
      "learning_rate": 4.677718086677195e-05,
      "loss": 2.1706,
      "step": 188896
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.420896291732788,
      "learning_rate": 4.677498743491217e-05,
      "loss": 1.9355,
      "step": 188897
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.4141732454299927,
      "learning_rate": 4.67727940501318e-05,
      "loss": 2.0935,
      "step": 188898
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.2732715606689453,
      "learning_rate": 4.677060071243107e-05,
      "loss": 2.0142,
      "step": 188899
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.3003301620483398,
      "learning_rate": 4.676840742181052e-05,
      "loss": 1.9096,
      "step": 188900
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.4109688997268677,
      "learning_rate": 4.6766214178270536e-05,
      "loss": 2.0384,
      "step": 188901
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.5885616540908813,
      "learning_rate": 4.676402098181146e-05,
      "loss": 1.9787,
      "step": 188902
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.5166372060775757,
      "learning_rate": 4.676182783243369e-05,
      "loss": 2.1627,
      "step": 188903
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.4854753017425537,
      "learning_rate": 4.6759634730137716e-05,
      "loss": 2.1987,
      "step": 188904
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.4614300727844238,
      "learning_rate": 4.6757441674923855e-05,
      "loss": 1.8622,
      "step": 188905
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.5782532691955566,
      "learning_rate": 4.675524866679263e-05,
      "loss": 1.9148,
      "step": 188906
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.5771992206573486,
      "learning_rate": 4.675305570574438e-05,
      "loss": 1.9331,
      "step": 188907
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.567399501800537,
      "learning_rate": 4.6750862791779534e-05,
      "loss": 1.9921,
      "step": 188908
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.293877363204956,
      "learning_rate": 4.674866992489842e-05,
      "loss": 2.0241,
      "step": 188909
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.42171049118042,
      "learning_rate": 4.674647710510154e-05,
      "loss": 1.8592,
      "step": 188910
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.3978830575942993,
      "learning_rate": 4.6744284332389236e-05,
      "loss": 1.9128,
      "step": 188911
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.3896150588989258,
      "learning_rate": 4.6742091606762e-05,
      "loss": 2.1388,
      "step": 188912
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.4307533502578735,
      "learning_rate": 4.67398989282202e-05,
      "loss": 1.8993,
      "step": 188913
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.3955494165420532,
      "learning_rate": 4.6737706296764235e-05,
      "loss": 2.1701,
      "step": 188914
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.3643137216567993,
      "learning_rate": 4.673551371239443e-05,
      "loss": 1.8729,
      "step": 188915
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.2753140926361084,
      "learning_rate": 4.673332117511137e-05,
      "loss": 2.0074,
      "step": 188916
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.5242292881011963,
      "learning_rate": 4.673112868491527e-05,
      "loss": 2.0785,
      "step": 188917
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.454929232597351,
      "learning_rate": 4.6728936241806715e-05,
      "loss": 1.7386,
      "step": 188918
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.4707940816879272,
      "learning_rate": 4.672674384578602e-05,
      "loss": 2.1021,
      "step": 188919
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.38017737865448,
      "learning_rate": 4.6724551496853625e-05,
      "loss": 2.1712,
      "step": 188920
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.3913307189941406,
      "learning_rate": 4.672235919500983e-05,
      "loss": 1.9291,
      "step": 188921
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.4365613460540771,
      "learning_rate": 4.67201669402552e-05,
      "loss": 1.9095,
      "step": 188922
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.3479608297348022,
      "learning_rate": 4.6717974732590005e-05,
      "loss": 1.9617,
      "step": 188923
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.4371699094772339,
      "learning_rate": 4.67157825720148e-05,
      "loss": 2.0845,
      "step": 188924
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.3951317071914673,
      "learning_rate": 4.67135904585299e-05,
      "loss": 1.9115,
      "step": 188925
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.42803955078125,
      "learning_rate": 4.6711398392135735e-05,
      "loss": 2.0911,
      "step": 188926
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.559526801109314,
      "learning_rate": 4.67092063728326e-05,
      "loss": 2.1533,
      "step": 188927
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.3786017894744873,
      "learning_rate": 4.670701440062112e-05,
      "loss": 1.9021,
      "step": 188928
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.3476656675338745,
      "learning_rate": 4.670482247550148e-05,
      "loss": 2.0932,
      "step": 188929
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.49357008934021,
      "learning_rate": 4.67026305974743e-05,
      "loss": 2.0205,
      "step": 188930
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.361426591873169,
      "learning_rate": 4.670043876653985e-05,
      "loss": 1.8758,
      "step": 188931
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.4152016639709473,
      "learning_rate": 4.6698246982698595e-05,
      "loss": 1.9764,
      "step": 188932
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.27718985080719,
      "learning_rate": 4.6696055245950834e-05,
      "loss": 2.0088,
      "step": 188933
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.4964237213134766,
      "learning_rate": 4.66938635562971e-05,
      "loss": 2.0512,
      "step": 188934
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.6064236164093018,
      "learning_rate": 4.669167191373774e-05,
      "loss": 2.0377,
      "step": 188935
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.3983732461929321,
      "learning_rate": 4.668948031827323e-05,
      "loss": 1.993,
      "step": 188936
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.3252619504928589,
      "learning_rate": 4.668728876990392e-05,
      "loss": 1.898,
      "step": 188937
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.500795841217041,
      "learning_rate": 4.6685097268630167e-05,
      "loss": 1.825,
      "step": 188938
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.3458284139633179,
      "learning_rate": 4.6682905814452476e-05,
      "loss": 2.0236,
      "step": 188939
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.492487907409668,
      "learning_rate": 4.6680714407371246e-05,
      "loss": 2.1876,
      "step": 188940
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.394386887550354,
      "learning_rate": 4.667852304738677e-05,
      "loss": 2.1246,
      "step": 188941
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.3024052381515503,
      "learning_rate": 4.667633173449963e-05,
      "loss": 1.9744,
      "step": 188942
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.3667659759521484,
      "learning_rate": 4.667414046871011e-05,
      "loss": 1.9155,
      "step": 188943
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.5307886600494385,
      "learning_rate": 4.667194925001858e-05,
      "loss": 1.9589,
      "step": 188944
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.5353206396102905,
      "learning_rate": 4.666975807842561e-05,
      "loss": 1.9949,
      "step": 188945
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.3848366737365723,
      "learning_rate": 4.66675669539315e-05,
      "loss": 2.0666,
      "step": 188946
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.423258900642395,
      "learning_rate": 4.6665375876536616e-05,
      "loss": 2.0666,
      "step": 188947
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.3009824752807617,
      "learning_rate": 4.666318484624149e-05,
      "loss": 1.9597,
      "step": 188948
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.4257488250732422,
      "learning_rate": 4.666099386304645e-05,
      "loss": 2.0158,
      "step": 188949
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.3809949159622192,
      "learning_rate": 4.6658802926951834e-05,
      "loss": 2.2552,
      "step": 188950
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.3091027736663818,
      "learning_rate": 4.6656612037958205e-05,
      "loss": 1.9124,
      "step": 188951
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.3949013948440552,
      "learning_rate": 4.6654421196065836e-05,
      "loss": 2.0723,
      "step": 188952
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.3676856756210327,
      "learning_rate": 4.665223040127526e-05,
      "loss": 2.0472,
      "step": 188953
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.4305390119552612,
      "learning_rate": 4.66500396535868e-05,
      "loss": 1.9278,
      "step": 188954
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.3946244716644287,
      "learning_rate": 4.66478489530009e-05,
      "loss": 1.9611,
      "step": 188955
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.293928861618042,
      "learning_rate": 4.6645658299517855e-05,
      "loss": 1.9762,
      "step": 188956
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.412551999092102,
      "learning_rate": 4.664346769313826e-05,
      "loss": 1.9925,
      "step": 188957
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.4512132406234741,
      "learning_rate": 4.664127713386232e-05,
      "loss": 1.9612,
      "step": 188958
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.2344917058944702,
      "learning_rate": 4.6639086621690633e-05,
      "loss": 1.8635,
      "step": 188959
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.4179569482803345,
      "learning_rate": 4.663689615662347e-05,
      "loss": 2.1952,
      "step": 188960
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.4492071866989136,
      "learning_rate": 4.663470573866143e-05,
      "loss": 2.0028,
      "step": 188961
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.3888602256774902,
      "learning_rate": 4.6632515367804635e-05,
      "loss": 2.049,
      "step": 188962
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.3395802974700928,
      "learning_rate": 4.66303250440537e-05,
      "loss": 2.017,
      "step": 188963
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.3696811199188232,
      "learning_rate": 4.662813476740891e-05,
      "loss": 1.8214,
      "step": 188964
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.4681792259216309,
      "learning_rate": 4.6625944537870804e-05,
      "loss": 2.1049,
      "step": 188965
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.4334741830825806,
      "learning_rate": 4.662375435543962e-05,
      "loss": 1.8732,
      "step": 188966
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.2979904413223267,
      "learning_rate": 4.6621564220115995e-05,
      "loss": 2.1616,
      "step": 188967
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.3075172901153564,
      "learning_rate": 4.6619374131900145e-05,
      "loss": 1.8625,
      "step": 188968
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.4921257495880127,
      "learning_rate": 4.661718409079255e-05,
      "loss": 1.971,
      "step": 188969
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.4177591800689697,
      "learning_rate": 4.6614994096793535e-05,
      "loss": 2.1487,
      "step": 188970
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.3550993204116821,
      "learning_rate": 4.661280414990367e-05,
      "loss": 2.0735,
      "step": 188971
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.3763649463653564,
      "learning_rate": 4.661061425012316e-05,
      "loss": 1.9801,
      "step": 188972
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.3799711465835571,
      "learning_rate": 4.6608424397452624e-05,
      "loss": 1.9749,
      "step": 188973
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.4419000148773193,
      "learning_rate": 4.660623459189235e-05,
      "loss": 2.0826,
      "step": 188974
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.4332143068313599,
      "learning_rate": 4.660404483344274e-05,
      "loss": 1.9829,
      "step": 188975
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.3340458869934082,
      "learning_rate": 4.660185512210419e-05,
      "loss": 1.8789,
      "step": 188976
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.3166959285736084,
      "learning_rate": 4.659966545787719e-05,
      "loss": 1.9631,
      "step": 188977
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.325117826461792,
      "learning_rate": 4.659747584076203e-05,
      "loss": 1.9956,
      "step": 188978
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.2651348114013672,
      "learning_rate": 4.6595286270759224e-05,
      "loss": 2.0557,
      "step": 188979
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.3099462985992432,
      "learning_rate": 4.659309674786916e-05,
      "loss": 1.9294,
      "step": 188980
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.3498218059539795,
      "learning_rate": 4.6590907272092245e-05,
      "loss": 2.0591,
      "step": 188981
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.389926791191101,
      "learning_rate": 4.658871784342878e-05,
      "loss": 2.1136,
      "step": 188982
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.3656060695648193,
      "learning_rate": 4.658652846187929e-05,
      "loss": 1.7923,
      "step": 188983
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.4839084148406982,
      "learning_rate": 4.658433912744412e-05,
      "loss": 2.258,
      "step": 188984
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.4159693717956543,
      "learning_rate": 4.658214984012375e-05,
      "loss": 2.1401,
      "step": 188985
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.350838541984558,
      "learning_rate": 4.657996059991856e-05,
      "loss": 2.2445,
      "step": 188986
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.3310681581497192,
      "learning_rate": 4.657777140682892e-05,
      "loss": 1.9871,
      "step": 188987
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.4775362014770508,
      "learning_rate": 4.657558226085523e-05,
      "loss": 2.1658,
      "step": 188988
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.4351698160171509,
      "learning_rate": 4.657339316199794e-05,
      "loss": 1.9483,
      "step": 188989
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.4928622245788574,
      "learning_rate": 4.65712041102574e-05,
      "loss": 1.9112,
      "step": 188990
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.313212275505066,
      "learning_rate": 4.65690151056341e-05,
      "loss": 1.98,
      "step": 188991
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.444621205329895,
      "learning_rate": 4.6566826148128445e-05,
      "loss": 1.8414,
      "step": 188992
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.5049575567245483,
      "learning_rate": 4.656463723774077e-05,
      "loss": 2.0054,
      "step": 188993
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.444345474243164,
      "learning_rate": 4.656244837447146e-05,
      "loss": 2.0835,
      "step": 188994
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.3192188739776611,
      "learning_rate": 4.656025955832103e-05,
      "loss": 1.934,
      "step": 188995
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.339659333229065,
      "learning_rate": 4.6558070789289744e-05,
      "loss": 1.8884,
      "step": 188996
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.2967103719711304,
      "learning_rate": 4.65558820673782e-05,
      "loss": 1.941,
      "step": 188997
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.3838491439819336,
      "learning_rate": 4.65536933925867e-05,
      "loss": 2.2092,
      "step": 188998
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.4965088367462158,
      "learning_rate": 4.655150476491564e-05,
      "loss": 2.0892,
      "step": 188999
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.4800329208374023,
      "learning_rate": 4.654931618436538e-05,
      "loss": 2.1295,
      "step": 189000
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.4048830270767212,
      "learning_rate": 4.654712765093643e-05,
      "loss": 2.0726,
      "step": 189001
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.3944419622421265,
      "learning_rate": 4.654493916462912e-05,
      "loss": 1.967,
      "step": 189002
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.3275697231292725,
      "learning_rate": 4.654275072544396e-05,
      "loss": 2.0547,
      "step": 189003
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.2938556671142578,
      "learning_rate": 4.654056233338126e-05,
      "loss": 1.6706,
      "step": 189004
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.3522512912750244,
      "learning_rate": 4.653837398844147e-05,
      "loss": 1.7639,
      "step": 189005
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.4208382368087769,
      "learning_rate": 4.653618569062488e-05,
      "loss": 1.9223,
      "step": 189006
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.516311764717102,
      "learning_rate": 4.6533997439932105e-05,
      "loss": 1.926,
      "step": 189007
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.4011150598526,
      "learning_rate": 4.6531809236363357e-05,
      "loss": 1.945,
      "step": 189008
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.4088308811187744,
      "learning_rate": 4.652962107991922e-05,
      "loss": 2.0208,
      "step": 189009
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.4110530614852905,
      "learning_rate": 4.652743297059999e-05,
      "loss": 2.0612,
      "step": 189010
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.5073469877243042,
      "learning_rate": 4.652524490840609e-05,
      "loss": 1.9812,
      "step": 189011
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.3923178911209106,
      "learning_rate": 4.6523056893337895e-05,
      "loss": 2.0247,
      "step": 189012
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.2983766794204712,
      "learning_rate": 4.652086892539587e-05,
      "loss": 1.8861,
      "step": 189013
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.2682384252548218,
      "learning_rate": 4.651868100458038e-05,
      "loss": 1.8141,
      "step": 189014
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.3663835525512695,
      "learning_rate": 4.6516493130891895e-05,
      "loss": 1.8567,
      "step": 189015
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.3047680854797363,
      "learning_rate": 4.6514305304330776e-05,
      "loss": 2.0994,
      "step": 189016
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.4349608421325684,
      "learning_rate": 4.651211752489743e-05,
      "loss": 2.0115,
      "step": 189017
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.430037260055542,
      "learning_rate": 4.6509929792592215e-05,
      "loss": 1.7532,
      "step": 189018
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.4822056293487549,
      "learning_rate": 4.650774210741567e-05,
      "loss": 2.0757,
      "step": 189019
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.3959119319915771,
      "learning_rate": 4.6505554469368025e-05,
      "loss": 2.0558,
      "step": 189020
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.4887691736221313,
      "learning_rate": 4.6503366878449846e-05,
      "loss": 1.9214,
      "step": 189021
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.3544635772705078,
      "learning_rate": 4.6501179334661495e-05,
      "loss": 2.1086,
      "step": 189022
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.3147348165512085,
      "learning_rate": 4.649899183800335e-05,
      "loss": 1.9935,
      "step": 189023
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.4466557502746582,
      "learning_rate": 4.6496804388475774e-05,
      "loss": 2.1149,
      "step": 189024
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.4116054773330688,
      "learning_rate": 4.649461698607926e-05,
      "loss": 1.9159,
      "step": 189025
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.3235118389129639,
      "learning_rate": 4.649242963081414e-05,
      "loss": 2.0319,
      "step": 189026
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.406165599822998,
      "learning_rate": 4.649024232268093e-05,
      "loss": 1.9512,
      "step": 189027
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.3952692747116089,
      "learning_rate": 4.6488055061679973e-05,
      "loss": 2.11,
      "step": 189028
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.428286075592041,
      "learning_rate": 4.6485867847811585e-05,
      "loss": 2.1223,
      "step": 189029
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.3380377292633057,
      "learning_rate": 4.6483680681076366e-05,
      "loss": 2.0645,
      "step": 189030
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.3754314184188843,
      "learning_rate": 4.648149356147457e-05,
      "loss": 2.0902,
      "step": 189031
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.2956479787826538,
      "learning_rate": 4.6479306489006605e-05,
      "loss": 1.7672,
      "step": 189032
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.4075708389282227,
      "learning_rate": 4.6477119463673006e-05,
      "loss": 1.9425,
      "step": 189033
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.3279521465301514,
      "learning_rate": 4.6474932485474005e-05,
      "loss": 1.9779,
      "step": 189034
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.313118815422058,
      "learning_rate": 4.6472745554410165e-05,
      "loss": 2.0082,
      "step": 189035
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.3755812644958496,
      "learning_rate": 4.647055867048185e-05,
      "loss": 2.0533,
      "step": 189036
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.4904847145080566,
      "learning_rate": 4.646837183368933e-05,
      "loss": 2.1308,
      "step": 189037
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.387690544128418,
      "learning_rate": 4.646618504403324e-05,
      "loss": 2.0088,
      "step": 189038
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.5093711614608765,
      "learning_rate": 4.646399830151385e-05,
      "loss": 1.8583,
      "step": 189039
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.4468070268630981,
      "learning_rate": 4.646181160613155e-05,
      "loss": 1.9691,
      "step": 189040
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.4062507152557373,
      "learning_rate": 4.645962495788681e-05,
      "loss": 2.049,
      "step": 189041
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.4721877574920654,
      "learning_rate": 4.645743835678003e-05,
      "loss": 1.9578,
      "step": 189042
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.3977375030517578,
      "learning_rate": 4.645525180281154e-05,
      "loss": 1.9244,
      "step": 189043
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.2621663808822632,
      "learning_rate": 4.645306529598185e-05,
      "loss": 2.0473,
      "step": 189044
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.2841383218765259,
      "learning_rate": 4.645087883629135e-05,
      "loss": 1.9549,
      "step": 189045
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.3894481658935547,
      "learning_rate": 4.644869242374034e-05,
      "loss": 1.8739,
      "step": 189046
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.309417724609375,
      "learning_rate": 4.644650605832936e-05,
      "loss": 2.0624,
      "step": 189047
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.2830803394317627,
      "learning_rate": 4.644431974005877e-05,
      "loss": 1.9163,
      "step": 189048
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.383026361465454,
      "learning_rate": 4.64421334689289e-05,
      "loss": 1.9016,
      "step": 189049
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.7706098556518555,
      "learning_rate": 4.643994724494029e-05,
      "loss": 2.0218,
      "step": 189050
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.7468738555908203,
      "learning_rate": 4.643776106809321e-05,
      "loss": 2.1697,
      "step": 189051
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.4327654838562012,
      "learning_rate": 4.643557493838821e-05,
      "loss": 1.9266,
      "step": 189052
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.6602439880371094,
      "learning_rate": 4.643338885582564e-05,
      "loss": 1.9586,
      "step": 189053
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.4205024242401123,
      "learning_rate": 4.643120282040585e-05,
      "loss": 2.2145,
      "step": 189054
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.636934518814087,
      "learning_rate": 4.642901683212923e-05,
      "loss": 2.1647,
      "step": 189055
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.5689849853515625,
      "learning_rate": 4.6426830890996325e-05,
      "loss": 2.054,
      "step": 189056
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.2648411989212036,
      "learning_rate": 4.642464499700738e-05,
      "loss": 2.0419,
      "step": 189057
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.390112280845642,
      "learning_rate": 4.642245915016295e-05,
      "loss": 1.9646,
      "step": 189058
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.424431562423706,
      "learning_rate": 4.642027335046338e-05,
      "loss": 1.9487,
      "step": 189059
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.4528746604919434,
      "learning_rate": 4.641808759790906e-05,
      "loss": 1.9704,
      "step": 189060
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.4455268383026123,
      "learning_rate": 4.641590189250033e-05,
      "loss": 1.9968,
      "step": 189061
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.5710986852645874,
      "learning_rate": 4.6413716234237756e-05,
      "loss": 2.0067,
      "step": 189062
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.4324003458023071,
      "learning_rate": 4.641153062312156e-05,
      "loss": 2.0001,
      "step": 189063
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.4939298629760742,
      "learning_rate": 4.6409345059152356e-05,
      "loss": 1.8695,
      "step": 189064
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.3580036163330078,
      "learning_rate": 4.640715954233041e-05,
      "loss": 1.9531,
      "step": 189065
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.5749865770339966,
      "learning_rate": 4.640497407265618e-05,
      "loss": 2.0497,
      "step": 189066
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.2501270771026611,
      "learning_rate": 4.6402788650129964e-05,
      "loss": 1.9192,
      "step": 189067
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.5259366035461426,
      "learning_rate": 4.6400603274752334e-05,
      "loss": 1.9071,
      "step": 189068
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.4457926750183105,
      "learning_rate": 4.639841794652356e-05,
      "loss": 2.0559,
      "step": 189069
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.377084493637085,
      "learning_rate": 4.639623266544417e-05,
      "loss": 1.9237,
      "step": 189070
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.526723861694336,
      "learning_rate": 4.639404743151452e-05,
      "loss": 1.9783,
      "step": 189071
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.5075558423995972,
      "learning_rate": 4.6391862244734966e-05,
      "loss": 1.9427,
      "step": 189072
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.414649248123169,
      "learning_rate": 4.6389677105105924e-05,
      "loss": 2.0386,
      "step": 189073
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.483380675315857,
      "learning_rate": 4.6387492012627866e-05,
      "loss": 2.094,
      "step": 189074
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.3841224908828735,
      "learning_rate": 4.638530696730113e-05,
      "loss": 2.0123,
      "step": 189075
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.4064544439315796,
      "learning_rate": 4.638312196912617e-05,
      "loss": 1.904,
      "step": 189076
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.366709589958191,
      "learning_rate": 4.638093701810343e-05,
      "loss": 2.107,
      "step": 189077
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.3259557485580444,
      "learning_rate": 4.637875211423323e-05,
      "loss": 1.9444,
      "step": 189078
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.2546980381011963,
      "learning_rate": 4.6376567257515914e-05,
      "loss": 1.7796,
      "step": 189079
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.4484553337097168,
      "learning_rate": 4.6374382447952085e-05,
      "loss": 2.1599,
      "step": 189080
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.2732839584350586,
      "learning_rate": 4.637219768554197e-05,
      "loss": 2.0853,
      "step": 189081
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.4264657497406006,
      "learning_rate": 4.637001297028613e-05,
      "loss": 2.1645,
      "step": 189082
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.5307725667953491,
      "learning_rate": 4.6367828302184876e-05,
      "loss": 2.112,
      "step": 189083
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.400423526763916,
      "learning_rate": 4.636564368123866e-05,
      "loss": 2.0013,
      "step": 189084
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.420914888381958,
      "learning_rate": 4.636345910744777e-05,
      "loss": 2.1854,
      "step": 189085
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.3739312887191772,
      "learning_rate": 4.636127458081274e-05,
      "loss": 2.0434,
      "step": 189086
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.2671159505844116,
      "learning_rate": 4.6359090101333906e-05,
      "loss": 1.6838,
      "step": 189087
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.5940237045288086,
      "learning_rate": 4.6356905669011745e-05,
      "loss": 1.9956,
      "step": 189088
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.3568456172943115,
      "learning_rate": 4.6354721283846666e-05,
      "loss": 2.2173,
      "step": 189089
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.5114436149597168,
      "learning_rate": 4.6352536945838994e-05,
      "loss": 1.937,
      "step": 189090
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.3526676893234253,
      "learning_rate": 4.635035265498913e-05,
      "loss": 2.0312,
      "step": 189091
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.4296369552612305,
      "learning_rate": 4.634816841129758e-05,
      "loss": 2.0825,
      "step": 189092
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.4219608306884766,
      "learning_rate": 4.634598421476461e-05,
      "loss": 1.9876,
      "step": 189093
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.3821628093719482,
      "learning_rate": 4.634380006539081e-05,
      "loss": 1.9365,
      "step": 189094
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.3989074230194092,
      "learning_rate": 4.6341615963176456e-05,
      "loss": 2.1089,
      "step": 189095
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.3739488124847412,
      "learning_rate": 4.6339431908121984e-05,
      "loss": 1.9846,
      "step": 189096
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.5206762552261353,
      "learning_rate": 4.633724790022775e-05,
      "loss": 1.7978,
      "step": 189097
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.330573558807373,
      "learning_rate": 4.6335063939494286e-05,
      "loss": 2.0376,
      "step": 189098
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.3189127445220947,
      "learning_rate": 4.6332880025921834e-05,
      "loss": 2.1155,
      "step": 189099
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.4992505311965942,
      "learning_rate": 4.6330696159510964e-05,
      "loss": 1.9714,
      "step": 189100
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.3676950931549072,
      "learning_rate": 4.6328512340261925e-05,
      "loss": 1.8954,
      "step": 189101
    },
    {
      "epoch": 2.46,
      "grad_norm": 2.4083244800567627,
      "learning_rate": 4.632632856817536e-05,
      "loss": 1.8645,
      "step": 189102
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.346930742263794,
      "learning_rate": 4.632414484325138e-05,
      "loss": 1.8103,
      "step": 189103
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.3103647232055664,
      "learning_rate": 4.6321961165490595e-05,
      "loss": 1.9439,
      "step": 189104
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.4291396141052246,
      "learning_rate": 4.6319777534893256e-05,
      "loss": 1.9988,
      "step": 189105
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.505595326423645,
      "learning_rate": 4.631759395145995e-05,
      "loss": 2.029,
      "step": 189106
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.3500573635101318,
      "learning_rate": 4.631541041519092e-05,
      "loss": 1.9328,
      "step": 189107
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.5475720167160034,
      "learning_rate": 4.631322692608676e-05,
      "loss": 1.9939,
      "step": 189108
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.4570738077163696,
      "learning_rate": 4.631104348414765e-05,
      "loss": 2.0451,
      "step": 189109
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.350877046585083,
      "learning_rate": 4.630886008937417e-05,
      "loss": 2.0223,
      "step": 189110
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.6641954183578491,
      "learning_rate": 4.630667674176659e-05,
      "loss": 2.0909,
      "step": 189111
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.3766883611679077,
      "learning_rate": 4.6304493441325454e-05,
      "loss": 2.1106,
      "step": 189112
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.3923602104187012,
      "learning_rate": 4.630231018805105e-05,
      "loss": 1.9372,
      "step": 189113
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.4586323499679565,
      "learning_rate": 4.630012698194389e-05,
      "loss": 2.1001,
      "step": 189114
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.5571460723876953,
      "learning_rate": 4.6297943823004324e-05,
      "loss": 1.9112,
      "step": 189115
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.3772084712982178,
      "learning_rate": 4.6295760711232765e-05,
      "loss": 2.0215,
      "step": 189116
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.4448819160461426,
      "learning_rate": 4.629357764662954e-05,
      "loss": 1.9931,
      "step": 189117
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.3017338514328003,
      "learning_rate": 4.629139462919519e-05,
      "loss": 1.8864,
      "step": 189118
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.2762497663497925,
      "learning_rate": 4.6289211658930004e-05,
      "loss": 1.8376,
      "step": 189119
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.4252210855484009,
      "learning_rate": 4.6287028735834485e-05,
      "loss": 1.9488,
      "step": 189120
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.391816258430481,
      "learning_rate": 4.6284845859909004e-05,
      "loss": 2.053,
      "step": 189121
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.4150912761688232,
      "learning_rate": 4.628266303115392e-05,
      "loss": 2.0759,
      "step": 189122
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.4457472562789917,
      "learning_rate": 4.628048024956974e-05,
      "loss": 1.9076,
      "step": 189123
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.3661518096923828,
      "learning_rate": 4.627829751515679e-05,
      "loss": 2.0077,
      "step": 189124
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.564474105834961,
      "learning_rate": 4.6276114827915414e-05,
      "loss": 1.9516,
      "step": 189125
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.3612970113754272,
      "learning_rate": 4.62739321878462e-05,
      "loss": 2.1108,
      "step": 189126
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.2957760095596313,
      "learning_rate": 4.627174959494942e-05,
      "loss": 1.8626,
      "step": 189127
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.4025838375091553,
      "learning_rate": 4.626956704922547e-05,
      "loss": 1.965,
      "step": 189128
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.478180170059204,
      "learning_rate": 4.626738455067485e-05,
      "loss": 2.161,
      "step": 189129
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.353582501411438,
      "learning_rate": 4.626520209929793e-05,
      "loss": 1.9101,
      "step": 189130
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.3034515380859375,
      "learning_rate": 4.6263019695095e-05,
      "loss": 2.0434,
      "step": 189131
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.4983570575714111,
      "learning_rate": 4.626083733806665e-05,
      "loss": 1.9787,
      "step": 189132
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.4326575994491577,
      "learning_rate": 4.625865502821322e-05,
      "loss": 1.9991,
      "step": 189133
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.3836143016815186,
      "learning_rate": 4.625647276553499e-05,
      "loss": 1.8794,
      "step": 189134
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.3982176780700684,
      "learning_rate": 4.6254290550032555e-05,
      "loss": 2.0471,
      "step": 189135
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.6249722242355347,
      "learning_rate": 4.625210838170619e-05,
      "loss": 1.9172,
      "step": 189136
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.439813256263733,
      "learning_rate": 4.624992626055638e-05,
      "loss": 2.2074,
      "step": 189137
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.2544242143630981,
      "learning_rate": 4.6247744186583504e-05,
      "loss": 2.1723,
      "step": 189138
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.7767683267593384,
      "learning_rate": 4.624556215978799e-05,
      "loss": 1.8941,
      "step": 189139
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.2947571277618408,
      "learning_rate": 4.624338018017013e-05,
      "loss": 1.7326,
      "step": 189140
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.4680765867233276,
      "learning_rate": 4.6241198247730474e-05,
      "loss": 1.8759,
      "step": 189141
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.5031871795654297,
      "learning_rate": 4.6239016362469304e-05,
      "loss": 2.0493,
      "step": 189142
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.6497827768325806,
      "learning_rate": 4.62368345243872e-05,
      "loss": 2.0567,
      "step": 189143
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.3242249488830566,
      "learning_rate": 4.623465273348442e-05,
      "loss": 2.0144,
      "step": 189144
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.4416356086730957,
      "learning_rate": 4.623247098976141e-05,
      "loss": 2.1931,
      "step": 189145
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.4138312339782715,
      "learning_rate": 4.6230289293218514e-05,
      "loss": 2.0673,
      "step": 189146
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.4108524322509766,
      "learning_rate": 4.622810764385625e-05,
      "loss": 1.8816,
      "step": 189147
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.3991299867630005,
      "learning_rate": 4.622592604167491e-05,
      "loss": 1.8242,
      "step": 189148
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.2876203060150146,
      "learning_rate": 4.622374448667503e-05,
      "loss": 1.825,
      "step": 189149
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.3814641237258911,
      "learning_rate": 4.622156297885698e-05,
      "loss": 1.858,
      "step": 189150
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.331629753112793,
      "learning_rate": 4.6219381518221074e-05,
      "loss": 1.8983,
      "step": 189151
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.4580693244934082,
      "learning_rate": 4.621720010476774e-05,
      "loss": 1.8678,
      "step": 189152
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.34572434425354,
      "learning_rate": 4.6215018738497486e-05,
      "loss": 1.7841,
      "step": 189153
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.4376177787780762,
      "learning_rate": 4.6212837419410566e-05,
      "loss": 2.2461,
      "step": 189154
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.371133804321289,
      "learning_rate": 4.621065614750756e-05,
      "loss": 2.0559,
      "step": 189155
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.444762110710144,
      "learning_rate": 4.620847492278879e-05,
      "loss": 2.2138,
      "step": 189156
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.4173121452331543,
      "learning_rate": 4.620629374525463e-05,
      "loss": 2.1671,
      "step": 189157
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.3441681861877441,
      "learning_rate": 4.6204112614905465e-05,
      "loss": 2.032,
      "step": 189158
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.4891042709350586,
      "learning_rate": 4.6201931531741786e-05,
      "loss": 1.7609,
      "step": 189159
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.3340476751327515,
      "learning_rate": 4.61997504957639e-05,
      "loss": 1.8184,
      "step": 189160
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.3481643199920654,
      "learning_rate": 4.619756950697237e-05,
      "loss": 1.8189,
      "step": 189161
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.4798636436462402,
      "learning_rate": 4.619538856536745e-05,
      "loss": 2.0386,
      "step": 189162
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.26399827003479,
      "learning_rate": 4.619320767094963e-05,
      "loss": 1.9742,
      "step": 189163
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.4068281650543213,
      "learning_rate": 4.619102682371922e-05,
      "loss": 1.9163,
      "step": 189164
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.549656629562378,
      "learning_rate": 4.618884602367674e-05,
      "loss": 2.0349,
      "step": 189165
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.4108649492263794,
      "learning_rate": 4.618666527082251e-05,
      "loss": 1.7806,
      "step": 189166
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.3807295560836792,
      "learning_rate": 4.6184484565157e-05,
      "loss": 2.0955,
      "step": 189167
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.2568999528884888,
      "learning_rate": 4.618230390668054e-05,
      "loss": 2.0584,
      "step": 189168
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.3772066831588745,
      "learning_rate": 4.61801232953937e-05,
      "loss": 1.9844,
      "step": 189169
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.4077818393707275,
      "learning_rate": 4.617794273129665e-05,
      "loss": 2.0278,
      "step": 189170
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.5054165124893188,
      "learning_rate": 4.617576221438998e-05,
      "loss": 1.8018,
      "step": 189171
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.4391453266143799,
      "learning_rate": 4.6173581744673924e-05,
      "loss": 2.071,
      "step": 189172
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.5346649885177612,
      "learning_rate": 4.617140132214909e-05,
      "loss": 1.8518,
      "step": 189173
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.37566077709198,
      "learning_rate": 4.616922094681571e-05,
      "loss": 1.9677,
      "step": 189174
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.3455560207366943,
      "learning_rate": 4.616704061867438e-05,
      "loss": 2.0401,
      "step": 189175
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.3806877136230469,
      "learning_rate": 4.61648603377253e-05,
      "loss": 2.0244,
      "step": 189176
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.4000537395477295,
      "learning_rate": 4.6162680103968994e-05,
      "loss": 1.9581,
      "step": 189177
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.3777612447738647,
      "learning_rate": 4.616049991740578e-05,
      "loss": 2.0335,
      "step": 189178
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.4317387342453003,
      "learning_rate": 4.615831977803618e-05,
      "loss": 2.0375,
      "step": 189179
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.409135103225708,
      "learning_rate": 4.6156139685860463e-05,
      "loss": 2.2062,
      "step": 189180
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.3270772695541382,
      "learning_rate": 4.6153959640879257e-05,
      "loss": 1.9132,
      "step": 189181
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.4088894128799438,
      "learning_rate": 4.61517796430927e-05,
      "loss": 2.0405,
      "step": 189182
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.5416944026947021,
      "learning_rate": 4.6149599692501395e-05,
      "loss": 2.0954,
      "step": 189183
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.2821855545043945,
      "learning_rate": 4.614741978910557e-05,
      "loss": 1.9695,
      "step": 189184
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.4944911003112793,
      "learning_rate": 4.614523993290582e-05,
      "loss": 2.0242,
      "step": 189185
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.6692792177200317,
      "learning_rate": 4.6143060123902386e-05,
      "loss": 1.8394,
      "step": 189186
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.3205169439315796,
      "learning_rate": 4.61408803620959e-05,
      "loss": 2.2011,
      "step": 189187
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.559396743774414,
      "learning_rate": 4.613870064748646e-05,
      "loss": 1.8539,
      "step": 189188
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.4120479822158813,
      "learning_rate": 4.61365209800747e-05,
      "loss": 1.9553,
      "step": 189189
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.4560195207595825,
      "learning_rate": 4.6134341359860886e-05,
      "loss": 1.9147,
      "step": 189190
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.3807793855667114,
      "learning_rate": 4.6132161786845576e-05,
      "loss": 2.0076,
      "step": 189191
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.5142816305160522,
      "learning_rate": 4.612998226102899e-05,
      "loss": 1.942,
      "step": 189192
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.451366662979126,
      "learning_rate": 4.61278027824118e-05,
      "loss": 2.0116,
      "step": 189193
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.51407790184021,
      "learning_rate": 4.612562335099407e-05,
      "loss": 1.9812,
      "step": 189194
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.4439409971237183,
      "learning_rate": 4.6123443966776474e-05,
      "loss": 2.0789,
      "step": 189195
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.4054807424545288,
      "learning_rate": 4.6121264629759255e-05,
      "loss": 2.0711,
      "step": 189196
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.4360617399215698,
      "learning_rate": 4.6119085339942954e-05,
      "loss": 2.3461,
      "step": 189197
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.3561513423919678,
      "learning_rate": 4.611690609732782e-05,
      "loss": 1.8301,
      "step": 189198
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.3340857028961182,
      "learning_rate": 4.611472690191443e-05,
      "loss": 1.9634,
      "step": 189199
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.3179363012313843,
      "learning_rate": 4.611254775370309e-05,
      "loss": 1.9375,
      "step": 189200
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.4626878499984741,
      "learning_rate": 4.611036865269425e-05,
      "loss": 2.0331,
      "step": 189201
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.4333158731460571,
      "learning_rate": 4.610818959888819e-05,
      "loss": 2.08,
      "step": 189202
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.3772166967391968,
      "learning_rate": 4.61060105922855e-05,
      "loss": 1.8979,
      "step": 189203
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.3739668130874634,
      "learning_rate": 4.610383163288639e-05,
      "loss": 2.037,
      "step": 189204
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.2632986307144165,
      "learning_rate": 4.6101652720691486e-05,
      "loss": 1.9048,
      "step": 189205
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.4926645755767822,
      "learning_rate": 4.609947385570105e-05,
      "loss": 2.0221,
      "step": 189206
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.4444530010223389,
      "learning_rate": 4.6097295037915495e-05,
      "loss": 1.9742,
      "step": 189207
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.5054967403411865,
      "learning_rate": 4.609511626733521e-05,
      "loss": 1.9715,
      "step": 189208
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.3836926221847534,
      "learning_rate": 4.60929375439607e-05,
      "loss": 1.9866,
      "step": 189209
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.3207875490188599,
      "learning_rate": 4.609075886779222e-05,
      "loss": 2.1515,
      "step": 189210
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.3417764902114868,
      "learning_rate": 4.6088580238830355e-05,
      "loss": 2.2415,
      "step": 189211
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.2586491107940674,
      "learning_rate": 4.608640165707539e-05,
      "loss": 1.8235,
      "step": 189212
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.3040974140167236,
      "learning_rate": 4.60842231225277e-05,
      "loss": 1.8445,
      "step": 189213
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.3655149936676025,
      "learning_rate": 4.608204463518781e-05,
      "loss": 1.8086,
      "step": 189214
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.417717695236206,
      "learning_rate": 4.607986619505606e-05,
      "loss": 1.9294,
      "step": 189215
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.3393105268478394,
      "learning_rate": 4.6077687802132814e-05,
      "loss": 1.9814,
      "step": 189216
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.4749000072479248,
      "learning_rate": 4.607550945641857e-05,
      "loss": 1.9312,
      "step": 189217
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.4303929805755615,
      "learning_rate": 4.607333115791366e-05,
      "loss": 2.2327,
      "step": 189218
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.4558414220809937,
      "learning_rate": 4.607115290661846e-05,
      "loss": 2.0304,
      "step": 189219
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.2646907567977905,
      "learning_rate": 4.606897470253349e-05,
      "loss": 2.0325,
      "step": 189220
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.327389121055603,
      "learning_rate": 4.6066796545659024e-05,
      "loss": 1.9494,
      "step": 189221
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.5092517137527466,
      "learning_rate": 4.606461843599563e-05,
      "loss": 2.1698,
      "step": 189222
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.3265937566757202,
      "learning_rate": 4.606244037354356e-05,
      "loss": 1.9269,
      "step": 189223
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.3476316928863525,
      "learning_rate": 4.606026235830333e-05,
      "loss": 1.9298,
      "step": 189224
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.4273934364318848,
      "learning_rate": 4.605808439027521e-05,
      "loss": 2.0185,
      "step": 189225
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.4174342155456543,
      "learning_rate": 4.6055906469459714e-05,
      "loss": 1.8315,
      "step": 189226
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.3941665887832642,
      "learning_rate": 4.6053728595857184e-05,
      "loss": 1.9494,
      "step": 189227
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.3331708908081055,
      "learning_rate": 4.605155076946813e-05,
      "loss": 2.4052,
      "step": 189228
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.4690271615982056,
      "learning_rate": 4.60493729902929e-05,
      "loss": 2.1266,
      "step": 189229
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.6063439846038818,
      "learning_rate": 4.604719525833184e-05,
      "loss": 2.1378,
      "step": 189230
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.3977203369140625,
      "learning_rate": 4.604501757358537e-05,
      "loss": 2.02,
      "step": 189231
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.4473291635513306,
      "learning_rate": 4.6042839936053975e-05,
      "loss": 2.0199,
      "step": 189232
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.3360072374343872,
      "learning_rate": 4.604066234573794e-05,
      "loss": 2.0499,
      "step": 189233
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.4229233264923096,
      "learning_rate": 4.603848480263784e-05,
      "loss": 1.819,
      "step": 189234
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.3955848217010498,
      "learning_rate": 4.6036307306753874e-05,
      "loss": 1.9247,
      "step": 189235
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.2850297689437866,
      "learning_rate": 4.60341298580867e-05,
      "loss": 2.0393,
      "step": 189236
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.4984216690063477,
      "learning_rate": 4.6031952456636456e-05,
      "loss": 1.8856,
      "step": 189237
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.4224263429641724,
      "learning_rate": 4.602977510240371e-05,
      "loss": 2.2106,
      "step": 189238
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.3004488945007324,
      "learning_rate": 4.602759779538876e-05,
      "loss": 1.7846,
      "step": 189239
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.5565950870513916,
      "learning_rate": 4.602542053559214e-05,
      "loss": 2.1345,
      "step": 189240
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.4994581937789917,
      "learning_rate": 4.602324332301411e-05,
      "loss": 1.9964,
      "step": 189241
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.4555667638778687,
      "learning_rate": 4.6021066157655314e-05,
      "loss": 1.9446,
      "step": 189242
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.4529883861541748,
      "learning_rate": 4.6018889039515846e-05,
      "loss": 2.1569,
      "step": 189243
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.299804449081421,
      "learning_rate": 4.6016711968596344e-05,
      "loss": 1.9515,
      "step": 189244
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.2916966676712036,
      "learning_rate": 4.6014534944897065e-05,
      "loss": 2.2996,
      "step": 189245
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.3925939798355103,
      "learning_rate": 4.601235796841851e-05,
      "loss": 2.1323,
      "step": 189246
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.286649465560913,
      "learning_rate": 4.601018103916102e-05,
      "loss": 2.0834,
      "step": 189247
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.407163381576538,
      "learning_rate": 4.6008004157125155e-05,
      "loss": 1.8084,
      "step": 189248
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.3398609161376953,
      "learning_rate": 4.600582732231105e-05,
      "loss": 1.8799,
      "step": 189249
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.3163412809371948,
      "learning_rate": 4.600365053471934e-05,
      "loss": 2.1263,
      "step": 189250
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.3767672777175903,
      "learning_rate": 4.6001473794350286e-05,
      "loss": 2.0818,
      "step": 189251
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.3337045907974243,
      "learning_rate": 4.599929710120439e-05,
      "loss": 2.1292,
      "step": 189252
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.524674415588379,
      "learning_rate": 4.599712045528195e-05,
      "loss": 1.9049,
      "step": 189253
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.433089256286621,
      "learning_rate": 4.599494385658361e-05,
      "loss": 2.1093,
      "step": 189254
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.9430549144744873,
      "learning_rate": 4.5992767305109454e-05,
      "loss": 2.0466,
      "step": 189255
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.4752798080444336,
      "learning_rate": 4.5990590800860096e-05,
      "loss": 2.0037,
      "step": 189256
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.2929288148880005,
      "learning_rate": 4.598841434383582e-05,
      "loss": 2.0558,
      "step": 189257
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.3666623830795288,
      "learning_rate": 4.5986237934037175e-05,
      "loss": 2.0536,
      "step": 189258
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.473315954208374,
      "learning_rate": 4.598406157146442e-05,
      "loss": 2.1969,
      "step": 189259
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.2754195928573608,
      "learning_rate": 4.5981885256118145e-05,
      "loss": 1.8786,
      "step": 189260
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.3729984760284424,
      "learning_rate": 4.5979708987998466e-05,
      "loss": 1.9194,
      "step": 189261
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.522990107536316,
      "learning_rate": 4.597753276710607e-05,
      "loss": 1.8998,
      "step": 189262
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.5666959285736084,
      "learning_rate": 4.5975356593441135e-05,
      "loss": 2.0363,
      "step": 189263
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.3248286247253418,
      "learning_rate": 4.597318046700429e-05,
      "loss": 1.9928,
      "step": 189264
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.7059487104415894,
      "learning_rate": 4.597100438779573e-05,
      "loss": 1.9341,
      "step": 189265
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.4193634986877441,
      "learning_rate": 4.596882835581609e-05,
      "loss": 2.1761,
      "step": 189266
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.309229850769043,
      "learning_rate": 4.596665237106551e-05,
      "loss": 1.8742,
      "step": 189267
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.3567376136779785,
      "learning_rate": 4.5964476433544576e-05,
      "loss": 2.0929,
      "step": 189268
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.3985612392425537,
      "learning_rate": 4.596230054325359e-05,
      "loss": 2.1221,
      "step": 189269
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.3715952634811401,
      "learning_rate": 4.5960124700193104e-05,
      "loss": 2.0777,
      "step": 189270
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.3561650514602661,
      "learning_rate": 4.5957948904363295e-05,
      "loss": 1.8839,
      "step": 189271
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.4584964513778687,
      "learning_rate": 4.595577315576488e-05,
      "loss": 1.9211,
      "step": 189272
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.5422438383102417,
      "learning_rate": 4.595359745439794e-05,
      "loss": 2.0906,
      "step": 189273
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.5639981031417847,
      "learning_rate": 4.595142180026309e-05,
      "loss": 1.9084,
      "step": 189274
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.3314005136489868,
      "learning_rate": 4.594924619336059e-05,
      "loss": 2.1423,
      "step": 189275
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.4188387393951416,
      "learning_rate": 4.594707063369097e-05,
      "loss": 1.9159,
      "step": 189276
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.4709054231643677,
      "learning_rate": 4.594489512125454e-05,
      "loss": 1.9506,
      "step": 189277
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.370703101158142,
      "learning_rate": 4.5942719656051895e-05,
      "loss": 2.0336,
      "step": 189278
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.370722770690918,
      "learning_rate": 4.5940544238083166e-05,
      "loss": 1.8055,
      "step": 189279
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.3590478897094727,
      "learning_rate": 4.593836886734892e-05,
      "loss": 1.9433,
      "step": 189280
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.2860569953918457,
      "learning_rate": 4.593619354384949e-05,
      "loss": 1.9886,
      "step": 189281
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.3765064477920532,
      "learning_rate": 4.593401826758537e-05,
      "loss": 1.8866,
      "step": 189282
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.4240052700042725,
      "learning_rate": 4.5931843038556835e-05,
      "loss": 1.8535,
      "step": 189283
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.6175415515899658,
      "learning_rate": 4.592966785676445e-05,
      "loss": 2.1172,
      "step": 189284
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.482211709022522,
      "learning_rate": 4.5927492722208516e-05,
      "loss": 1.99,
      "step": 189285
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.4048992395401,
      "learning_rate": 4.592531763488946e-05,
      "loss": 1.9317,
      "step": 189286
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.4574682712554932,
      "learning_rate": 4.5923142594807646e-05,
      "loss": 1.9732,
      "step": 189287
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.31504225730896,
      "learning_rate": 4.5920967601963553e-05,
      "loss": 1.7598,
      "step": 189288
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.4555236101150513,
      "learning_rate": 4.591879265635747e-05,
      "loss": 1.9274,
      "step": 189289
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.5370007753372192,
      "learning_rate": 4.591661775798997e-05,
      "loss": 2.0102,
      "step": 189290
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.426538348197937,
      "learning_rate": 4.591444290686138e-05,
      "loss": 1.8723,
      "step": 189291
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.3345870971679688,
      "learning_rate": 4.591226810297207e-05,
      "loss": 1.9607,
      "step": 189292
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.2961370944976807,
      "learning_rate": 4.5910093346322404e-05,
      "loss": 2.0322,
      "step": 189293
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.3266596794128418,
      "learning_rate": 4.5907918636912924e-05,
      "loss": 2.0416,
      "step": 189294
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.448032259941101,
      "learning_rate": 4.590574397474388e-05,
      "loss": 2.0137,
      "step": 189295
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.449405312538147,
      "learning_rate": 4.5903569359815815e-05,
      "loss": 1.8272,
      "step": 189296
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.3886754512786865,
      "learning_rate": 4.5901394792129096e-05,
      "loss": 2.0583,
      "step": 189297
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.397205114364624,
      "learning_rate": 4.5899220271684026e-05,
      "loss": 1.9916,
      "step": 189298
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.2929065227508545,
      "learning_rate": 4.5897045798481155e-05,
      "loss": 1.8521,
      "step": 189299
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.3445029258728027,
      "learning_rate": 4.58948713725208e-05,
      "loss": 2.0429,
      "step": 189300
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.3136736154556274,
      "learning_rate": 4.5892696993803324e-05,
      "loss": 2.0446,
      "step": 189301
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.3712730407714844,
      "learning_rate": 4.5890522662329286e-05,
      "loss": 2.1599,
      "step": 189302
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.401899814605713,
      "learning_rate": 4.588834837809899e-05,
      "loss": 1.9929,
      "step": 189303
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.4217650890350342,
      "learning_rate": 4.588617414111277e-05,
      "loss": 1.9066,
      "step": 189304
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.4357569217681885,
      "learning_rate": 4.5883999951371164e-05,
      "loss": 1.965,
      "step": 189305
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.3906283378601074,
      "learning_rate": 4.5881825808874526e-05,
      "loss": 2.079,
      "step": 189306
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.5756046772003174,
      "learning_rate": 4.58796517136232e-05,
      "loss": 2.0304,
      "step": 189307
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.4116214513778687,
      "learning_rate": 4.587747766561768e-05,
      "loss": 1.8781,
      "step": 189308
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.4564687013626099,
      "learning_rate": 4.5875303664858365e-05,
      "loss": 1.9861,
      "step": 189309
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.3494465351104736,
      "learning_rate": 4.587312971134556e-05,
      "loss": 1.9091,
      "step": 189310
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.3362354040145874,
      "learning_rate": 4.587095580507979e-05,
      "loss": 1.9637,
      "step": 189311
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.4110143184661865,
      "learning_rate": 4.586878194606133e-05,
      "loss": 2.0785,
      "step": 189312
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.3824766874313354,
      "learning_rate": 4.586660813429074e-05,
      "loss": 2.0319,
      "step": 189313
    },
    {
      "epoch": 2.46,
      "grad_norm": 1.4481511116027832,
      "learning_rate": 4.586443436976836e-05,
      "loss": 2.0853,
      "step": 189314
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.4804205894470215,
      "learning_rate": 4.586226065249458e-05,
      "loss": 1.8711,
      "step": 189315
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.419184923171997,
      "learning_rate": 4.58600869824697e-05,
      "loss": 1.9828,
      "step": 189316
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.4649739265441895,
      "learning_rate": 4.585791335969433e-05,
      "loss": 2.2517,
      "step": 189317
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.3595722913742065,
      "learning_rate": 4.5855739784168664e-05,
      "loss": 2.0184,
      "step": 189318
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.3628486394882202,
      "learning_rate": 4.5853566255893306e-05,
      "loss": 2.0451,
      "step": 189319
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.4310251474380493,
      "learning_rate": 4.585139277486851e-05,
      "loss": 2.1488,
      "step": 189320
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.3670358657836914,
      "learning_rate": 4.5849219341094855e-05,
      "loss": 2.0684,
      "step": 189321
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.3349403142929077,
      "learning_rate": 4.58470459545725e-05,
      "loss": 2.0485,
      "step": 189322
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.6129094362258911,
      "learning_rate": 4.5844872615302055e-05,
      "loss": 1.9286,
      "step": 189323
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.5031445026397705,
      "learning_rate": 4.584269932328377e-05,
      "loss": 2.1367,
      "step": 189324
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.40259850025177,
      "learning_rate": 4.584052607851819e-05,
      "loss": 1.8657,
      "step": 189325
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.3185986280441284,
      "learning_rate": 4.583835288100558e-05,
      "loss": 1.9071,
      "step": 189326
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.4001141786575317,
      "learning_rate": 4.583617973074657e-05,
      "loss": 2.118,
      "step": 189327
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.4656243324279785,
      "learning_rate": 4.583400662774126e-05,
      "loss": 1.9739,
      "step": 189328
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.5057002305984497,
      "learning_rate": 4.5831833571990284e-05,
      "loss": 2.0752,
      "step": 189329
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.2433860301971436,
      "learning_rate": 4.5829660563493876e-05,
      "loss": 1.9262,
      "step": 189330
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.4639464616775513,
      "learning_rate": 4.5827487602252635e-05,
      "loss": 1.9958,
      "step": 189331
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.3321198225021362,
      "learning_rate": 4.582531468826675e-05,
      "loss": 1.8532,
      "step": 189332
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.3275140523910522,
      "learning_rate": 4.5823141821536914e-05,
      "loss": 2.0617,
      "step": 189333
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.4176006317138672,
      "learning_rate": 4.58209690020632e-05,
      "loss": 1.7474,
      "step": 189334
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.498956322669983,
      "learning_rate": 4.581879622984626e-05,
      "loss": 1.9138,
      "step": 189335
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.3831958770751953,
      "learning_rate": 4.581662350488631e-05,
      "loss": 1.9287,
      "step": 189336
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.4391928911209106,
      "learning_rate": 4.5814450827183926e-05,
      "loss": 2.1084,
      "step": 189337
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.4043046236038208,
      "learning_rate": 4.581227819673934e-05,
      "loss": 2.0153,
      "step": 189338
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.3477250337600708,
      "learning_rate": 4.581010561355322e-05,
      "loss": 2.0422,
      "step": 189339
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.631212830543518,
      "learning_rate": 4.580793307762566e-05,
      "loss": 1.9547,
      "step": 189340
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.4265189170837402,
      "learning_rate": 4.580576058895723e-05,
      "loss": 1.9256,
      "step": 189341
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.3560271263122559,
      "learning_rate": 4.58035881475483e-05,
      "loss": 1.8654,
      "step": 189342
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.435543179512024,
      "learning_rate": 4.5801415753399295e-05,
      "loss": 2.0222,
      "step": 189343
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.379237413406372,
      "learning_rate": 4.579924340651056e-05,
      "loss": 2.0195,
      "step": 189344
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.4297559261322021,
      "learning_rate": 4.579707110688268e-05,
      "loss": 1.6963,
      "step": 189345
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.3470886945724487,
      "learning_rate": 4.579489885451579e-05,
      "loss": 2.009,
      "step": 189346
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.4121562242507935,
      "learning_rate": 4.57927266494105e-05,
      "loss": 1.9385,
      "step": 189347
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.3878687620162964,
      "learning_rate": 4.5790554491567076e-05,
      "loss": 2.0076,
      "step": 189348
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.258593201637268,
      "learning_rate": 4.578838238098604e-05,
      "loss": 2.2029,
      "step": 189349
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.4106703996658325,
      "learning_rate": 4.578621031766767e-05,
      "loss": 1.956,
      "step": 189350
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.432555079460144,
      "learning_rate": 4.5784038301612594e-05,
      "loss": 1.8999,
      "step": 189351
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.2916308641433716,
      "learning_rate": 4.5781866332820905e-05,
      "loss": 1.8587,
      "step": 189352
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.3779245615005493,
      "learning_rate": 4.5779694411293275e-05,
      "loss": 2.1444,
      "step": 189353
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.4081679582595825,
      "learning_rate": 4.577752253702987e-05,
      "loss": 2.1364,
      "step": 189354
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.408807635307312,
      "learning_rate": 4.577535071003133e-05,
      "loss": 1.8158,
      "step": 189355
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.5439974069595337,
      "learning_rate": 4.577317893029788e-05,
      "loss": 1.8625,
      "step": 189356
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.3753513097763062,
      "learning_rate": 4.5771007197830154e-05,
      "loss": 1.9813,
      "step": 189357
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.3410675525665283,
      "learning_rate": 4.576883551262822e-05,
      "loss": 2.0073,
      "step": 189358
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.2452841997146606,
      "learning_rate": 4.576666387469274e-05,
      "loss": 2.0138,
      "step": 189359
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.3948148488998413,
      "learning_rate": 4.576449228402399e-05,
      "loss": 2.0079,
      "step": 189360
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.3314087390899658,
      "learning_rate": 4.576232074062246e-05,
      "loss": 1.9848,
      "step": 189361
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.488926649093628,
      "learning_rate": 4.5760149244488454e-05,
      "loss": 2.01,
      "step": 189362
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.3722541332244873,
      "learning_rate": 4.575797779562257e-05,
      "loss": 2.0118,
      "step": 189363
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.4921839237213135,
      "learning_rate": 4.575580639402494e-05,
      "loss": 2.0473,
      "step": 189364
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.4001845121383667,
      "learning_rate": 4.575363503969617e-05,
      "loss": 2.0878,
      "step": 189365
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.3368959426879883,
      "learning_rate": 4.575146373263655e-05,
      "loss": 1.9299,
      "step": 189366
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.564218521118164,
      "learning_rate": 4.5749292472846586e-05,
      "loss": 1.8908,
      "step": 189367
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.3803004026412964,
      "learning_rate": 4.574712126032655e-05,
      "loss": 1.8679,
      "step": 189368
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.3579529523849487,
      "learning_rate": 4.574495009507703e-05,
      "loss": 2.083,
      "step": 189369
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.2522485256195068,
      "learning_rate": 4.574277897709829e-05,
      "loss": 2.0485,
      "step": 189370
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.3400782346725464,
      "learning_rate": 4.5740607906390746e-05,
      "loss": 2.0157,
      "step": 189371
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.3164150714874268,
      "learning_rate": 4.573843688295479e-05,
      "loss": 2.1061,
      "step": 189372
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.3627071380615234,
      "learning_rate": 4.5736265906790914e-05,
      "loss": 1.8705,
      "step": 189373
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.6939054727554321,
      "learning_rate": 4.573409497789939e-05,
      "loss": 2.0366,
      "step": 189374
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.6654448509216309,
      "learning_rate": 4.573192409628079e-05,
      "loss": 2.0777,
      "step": 189375
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.4049161672592163,
      "learning_rate": 4.572975326193541e-05,
      "loss": 2.0588,
      "step": 189376
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.3665664196014404,
      "learning_rate": 4.5727582474863675e-05,
      "loss": 2.1782,
      "step": 189377
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.4853347539901733,
      "learning_rate": 4.5725411735065896e-05,
      "loss": 1.9313,
      "step": 189378
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.4540990591049194,
      "learning_rate": 4.572324104254267e-05,
      "loss": 1.8178,
      "step": 189379
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.4182157516479492,
      "learning_rate": 4.572107039729419e-05,
      "loss": 2.0551,
      "step": 189380
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.329996109008789,
      "learning_rate": 4.571889979932103e-05,
      "loss": 1.7165,
      "step": 189381
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.6065477132797241,
      "learning_rate": 4.571672924862355e-05,
      "loss": 1.967,
      "step": 189382
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.4622732400894165,
      "learning_rate": 4.5714558745202026e-05,
      "loss": 2.1036,
      "step": 189383
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.3117786645889282,
      "learning_rate": 4.571238828905708e-05,
      "loss": 2.0648,
      "step": 189384
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.3331540822982788,
      "learning_rate": 4.571021788018898e-05,
      "loss": 2.1575,
      "step": 189385
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.3554062843322754,
      "learning_rate": 4.570804751859807e-05,
      "loss": 2.0458,
      "step": 189386
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.2467517852783203,
      "learning_rate": 4.57058772042849e-05,
      "loss": 1.9278,
      "step": 189387
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.3550935983657837,
      "learning_rate": 4.570370693724982e-05,
      "loss": 1.9334,
      "step": 189388
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.4017775058746338,
      "learning_rate": 4.570153671749315e-05,
      "loss": 2.0363,
      "step": 189389
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.3454240560531616,
      "learning_rate": 4.569936654501543e-05,
      "loss": 1.828,
      "step": 189390
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.3192633390426636,
      "learning_rate": 4.5697196419817016e-05,
      "loss": 2.0884,
      "step": 189391
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.2762367725372314,
      "learning_rate": 4.569502634189822e-05,
      "loss": 1.7471,
      "step": 189392
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.3567655086517334,
      "learning_rate": 4.5692856311259574e-05,
      "loss": 1.9372,
      "step": 189393
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.3842617273330688,
      "learning_rate": 4.569068632790144e-05,
      "loss": 1.841,
      "step": 189394
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.4243783950805664,
      "learning_rate": 4.568851639182412e-05,
      "loss": 2.0455,
      "step": 189395
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.3137550354003906,
      "learning_rate": 4.568634650302817e-05,
      "loss": 1.9471,
      "step": 189396
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.4924721717834473,
      "learning_rate": 4.568417666151391e-05,
      "loss": 1.6882,
      "step": 189397
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.4592299461364746,
      "learning_rate": 4.568200686728178e-05,
      "loss": 2.061,
      "step": 189398
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.3619260787963867,
      "learning_rate": 4.5679837120332174e-05,
      "loss": 1.9638,
      "step": 189399
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.4108320474624634,
      "learning_rate": 4.567766742066551e-05,
      "loss": 1.859,
      "step": 189400
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.4839123487472534,
      "learning_rate": 4.567549776828209e-05,
      "loss": 1.9016,
      "step": 189401
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.3885985612869263,
      "learning_rate": 4.567332816318244e-05,
      "loss": 1.958,
      "step": 189402
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.4082890748977661,
      "learning_rate": 4.5671158605366874e-05,
      "loss": 1.9574,
      "step": 189403
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.303673267364502,
      "learning_rate": 4.5668989094835885e-05,
      "loss": 2.1338,
      "step": 189404
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.333831787109375,
      "learning_rate": 4.56668196315898e-05,
      "loss": 2.1805,
      "step": 189405
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.3651583194732666,
      "learning_rate": 4.5664650215629164e-05,
      "loss": 2.0735,
      "step": 189406
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.3753958940505981,
      "learning_rate": 4.5662480846954164e-05,
      "loss": 2.1607,
      "step": 189407
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.4858906269073486,
      "learning_rate": 4.566031152556534e-05,
      "loss": 1.8194,
      "step": 189408
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.3305644989013672,
      "learning_rate": 4.5658142251463026e-05,
      "loss": 2.2067,
      "step": 189409
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.4188646078109741,
      "learning_rate": 4.565597302464772e-05,
      "loss": 2.246,
      "step": 189410
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.3993581533432007,
      "learning_rate": 4.5653803845119684e-05,
      "loss": 2.0044,
      "step": 189411
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.4773812294006348,
      "learning_rate": 4.565163471287956e-05,
      "loss": 2.0839,
      "step": 189412
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.6395511627197266,
      "learning_rate": 4.5649465627927474e-05,
      "loss": 2.008,
      "step": 189413
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.520961046218872,
      "learning_rate": 4.5647296590264025e-05,
      "loss": 1.9566,
      "step": 189414
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.3362935781478882,
      "learning_rate": 4.564512759988945e-05,
      "loss": 2.0905,
      "step": 189415
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.6854084730148315,
      "learning_rate": 4.564295865680432e-05,
      "loss": 1.9535,
      "step": 189416
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.452321171760559,
      "learning_rate": 4.564078976100892e-05,
      "loss": 1.9117,
      "step": 189417
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.352366328239441,
      "learning_rate": 4.563862091250383e-05,
      "loss": 2.1649,
      "step": 189418
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.3487004041671753,
      "learning_rate": 4.563645211128914e-05,
      "loss": 1.7543,
      "step": 189419
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.452781081199646,
      "learning_rate": 4.563428335736556e-05,
      "loss": 2.0334,
      "step": 189420
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.538156509399414,
      "learning_rate": 4.563211465073329e-05,
      "loss": 2.0445,
      "step": 189421
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.4070583581924438,
      "learning_rate": 4.5629945991392845e-05,
      "loss": 1.9266,
      "step": 189422
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.4741863012313843,
      "learning_rate": 4.5627777379344547e-05,
      "loss": 2.0259,
      "step": 189423
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.4307904243469238,
      "learning_rate": 4.562560881458901e-05,
      "loss": 1.9488,
      "step": 189424
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.5933058261871338,
      "learning_rate": 4.5623440297126314e-05,
      "loss": 1.8811,
      "step": 189425
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.3986023664474487,
      "learning_rate": 4.5621271826957084e-05,
      "loss": 2.0345,
      "step": 189426
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.3770179748535156,
      "learning_rate": 4.5619103404081594e-05,
      "loss": 1.8835,
      "step": 189427
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.3720073699951172,
      "learning_rate": 4.56169350285004e-05,
      "loss": 1.9428,
      "step": 189428
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.3407191038131714,
      "learning_rate": 4.561476670021375e-05,
      "loss": 2.0063,
      "step": 189429
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.3102302551269531,
      "learning_rate": 4.5612598419222225e-05,
      "loss": 1.8073,
      "step": 189430
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.3251055479049683,
      "learning_rate": 4.5610430185526e-05,
      "loss": 2.0944,
      "step": 189431
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.3889918327331543,
      "learning_rate": 4.5608261999125684e-05,
      "loss": 1.8048,
      "step": 189432
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.3154345750808716,
      "learning_rate": 4.56060938600215e-05,
      "loss": 1.8993,
      "step": 189433
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.3394917249679565,
      "learning_rate": 4.560392576821401e-05,
      "loss": 1.9328,
      "step": 189434
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.3648380041122437,
      "learning_rate": 4.560175772370349e-05,
      "loss": 1.9779,
      "step": 189435
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.4485867023468018,
      "learning_rate": 4.55995897264905e-05,
      "loss": 1.8935,
      "step": 189436
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.4181950092315674,
      "learning_rate": 4.559742177657531e-05,
      "loss": 2.1022,
      "step": 189437
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.5449377298355103,
      "learning_rate": 4.5595253873958395e-05,
      "loss": 2.2982,
      "step": 189438
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.3661696910858154,
      "learning_rate": 4.5593086018640035e-05,
      "loss": 2.2177,
      "step": 189439
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.3480592966079712,
      "learning_rate": 4.559091821062081e-05,
      "loss": 2.051,
      "step": 189440
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.5299580097198486,
      "learning_rate": 4.558875044990094e-05,
      "loss": 2.0846,
      "step": 189441
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.3331859111785889,
      "learning_rate": 4.558658273648101e-05,
      "loss": 2.0267,
      "step": 189442
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.427670955657959,
      "learning_rate": 4.5584415070361344e-05,
      "loss": 1.9484,
      "step": 189443
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.3317248821258545,
      "learning_rate": 4.558224745154231e-05,
      "loss": 1.7923,
      "step": 189444
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.4631431102752686,
      "learning_rate": 4.55800798800243e-05,
      "loss": 2.0529,
      "step": 189445
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.648833155632019,
      "learning_rate": 4.557791235580782e-05,
      "loss": 2.1574,
      "step": 189446
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.3051997423171997,
      "learning_rate": 4.557574487889311e-05,
      "loss": 2.127,
      "step": 189447
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.3273411989212036,
      "learning_rate": 4.557357744928079e-05,
      "loss": 2.0766,
      "step": 189448
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.5468593835830688,
      "learning_rate": 4.557141006697109e-05,
      "loss": 1.9228,
      "step": 189449
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.4907110929489136,
      "learning_rate": 4.5569242731964496e-05,
      "loss": 1.8276,
      "step": 189450
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.4589112997055054,
      "learning_rate": 4.5567075444261325e-05,
      "loss": 2.1703,
      "step": 189451
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.35978102684021,
      "learning_rate": 4.5564908203862084e-05,
      "loss": 1.9533,
      "step": 189452
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.4392626285552979,
      "learning_rate": 4.5562741010767065e-05,
      "loss": 2.061,
      "step": 189453
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.5212020874023438,
      "learning_rate": 4.556057386497681e-05,
      "loss": 2.1299,
      "step": 189454
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.380176067352295,
      "learning_rate": 4.555840676649165e-05,
      "loss": 2.0842,
      "step": 189455
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.5860931873321533,
      "learning_rate": 4.555623971531198e-05,
      "loss": 1.7792,
      "step": 189456
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.2964717149734497,
      "learning_rate": 4.555407271143814e-05,
      "loss": 2.0134,
      "step": 189457
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.3967304229736328,
      "learning_rate": 4.555190575487069e-05,
      "loss": 1.9649,
      "step": 189458
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.3470211029052734,
      "learning_rate": 4.554973884560983e-05,
      "loss": 2.0991,
      "step": 189459
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.3021173477172852,
      "learning_rate": 4.5547571983656195e-05,
      "loss": 1.9151,
      "step": 189460
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.41133451461792,
      "learning_rate": 4.554540516901005e-05,
      "loss": 2.0207,
      "step": 189461
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.409341812133789,
      "learning_rate": 4.55432384016718e-05,
      "loss": 1.9424,
      "step": 189462
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.302479863166809,
      "learning_rate": 4.5541071681641806e-05,
      "loss": 1.9454,
      "step": 189463
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.4013532400131226,
      "learning_rate": 4.55389050089206e-05,
      "loss": 1.9915,
      "step": 189464
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.2910562753677368,
      "learning_rate": 4.5536738383508456e-05,
      "loss": 1.9793,
      "step": 189465
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.3816384077072144,
      "learning_rate": 4.5534571805405896e-05,
      "loss": 2.12,
      "step": 189466
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.4759870767593384,
      "learning_rate": 4.553240527461326e-05,
      "loss": 2.0291,
      "step": 189467
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.3867156505584717,
      "learning_rate": 4.5530238791130945e-05,
      "loss": 2.0469,
      "step": 189468
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.4828028678894043,
      "learning_rate": 4.552807235495932e-05,
      "loss": 2.2443,
      "step": 189469
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.3979569673538208,
      "learning_rate": 4.552590596609889e-05,
      "loss": 2.0482,
      "step": 189470
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.4977247714996338,
      "learning_rate": 4.5523739624549904e-05,
      "loss": 1.9093,
      "step": 189471
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.4235807657241821,
      "learning_rate": 4.5521573330312947e-05,
      "loss": 1.9843,
      "step": 189472
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.7399334907531738,
      "learning_rate": 4.5519407083388336e-05,
      "loss": 2.0505,
      "step": 189473
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.3274285793304443,
      "learning_rate": 4.551724088377642e-05,
      "loss": 1.7289,
      "step": 189474
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.2852661609649658,
      "learning_rate": 4.5515074731477687e-05,
      "loss": 2.0837,
      "step": 189475
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.3043465614318848,
      "learning_rate": 4.551290862649251e-05,
      "loss": 2.1467,
      "step": 189476
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.4315863847732544,
      "learning_rate": 4.5510742568821214e-05,
      "loss": 1.944,
      "step": 189477
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.4150081872940063,
      "learning_rate": 4.5508576558464375e-05,
      "loss": 2.0532,
      "step": 189478
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.4316684007644653,
      "learning_rate": 4.550641059542228e-05,
      "loss": 2.2727,
      "step": 189479
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.3348945379257202,
      "learning_rate": 4.550424467969528e-05,
      "loss": 1.8443,
      "step": 189480
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.6100049018859863,
      "learning_rate": 4.5502078811283894e-05,
      "loss": 2.0748,
      "step": 189481
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.462280035018921,
      "learning_rate": 4.549991299018843e-05,
      "loss": 1.9781,
      "step": 189482
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.36863374710083,
      "learning_rate": 4.5497747216409374e-05,
      "loss": 2.1498,
      "step": 189483
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.3627333641052246,
      "learning_rate": 4.549558148994714e-05,
      "loss": 2.0115,
      "step": 189484
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.5149896144866943,
      "learning_rate": 4.5493415810802056e-05,
      "loss": 2.2419,
      "step": 189485
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.4295134544372559,
      "learning_rate": 4.5491250178974456e-05,
      "loss": 2.1396,
      "step": 189486
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.3818026781082153,
      "learning_rate": 4.548908459446494e-05,
      "loss": 1.8883,
      "step": 189487
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.5010030269622803,
      "learning_rate": 4.5486919057273744e-05,
      "loss": 2.003,
      "step": 189488
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.514755129814148,
      "learning_rate": 4.548475356740136e-05,
      "loss": 2.1672,
      "step": 189489
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.4270375967025757,
      "learning_rate": 4.5482588124848216e-05,
      "loss": 2.0976,
      "step": 189490
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.586368203163147,
      "learning_rate": 4.548042272961463e-05,
      "loss": 1.9459,
      "step": 189491
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.4307581186294556,
      "learning_rate": 4.547825738170099e-05,
      "loss": 2.2367,
      "step": 189492
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.3827271461486816,
      "learning_rate": 4.54760920811078e-05,
      "loss": 2.0465,
      "step": 189493
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.4253108501434326,
      "learning_rate": 4.5473926827835314e-05,
      "loss": 2.1653,
      "step": 189494
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.547692894935608,
      "learning_rate": 4.547176162188415e-05,
      "loss": 2.1769,
      "step": 189495
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.4287062883377075,
      "learning_rate": 4.54695964632545e-05,
      "loss": 2.0365,
      "step": 189496
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.3510979413986206,
      "learning_rate": 4.546743135194699e-05,
      "loss": 1.907,
      "step": 189497
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.2846171855926514,
      "learning_rate": 4.5465266287961766e-05,
      "loss": 1.9643,
      "step": 189498
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.3969906568527222,
      "learning_rate": 4.5463101271299386e-05,
      "loss": 1.9421,
      "step": 189499
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.334107756614685,
      "learning_rate": 4.546093630196018e-05,
      "loss": 2.0532,
      "step": 189500
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.3047181367874146,
      "learning_rate": 4.545877137994466e-05,
      "loss": 2.0817,
      "step": 189501
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.478804111480713,
      "learning_rate": 4.545660650525308e-05,
      "loss": 2.0388,
      "step": 189502
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.3754571676254272,
      "learning_rate": 4.545444167788601e-05,
      "loss": 1.9214,
      "step": 189503
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.4446659088134766,
      "learning_rate": 4.5452276897843756e-05,
      "loss": 2.026,
      "step": 189504
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.583998680114746,
      "learning_rate": 4.545011216512671e-05,
      "loss": 2.0821,
      "step": 189505
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.5151867866516113,
      "learning_rate": 4.5447947479735234e-05,
      "loss": 2.0483,
      "step": 189506
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.3917460441589355,
      "learning_rate": 4.5445782841669875e-05,
      "loss": 1.9753,
      "step": 189507
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.3784950971603394,
      "learning_rate": 4.544361825093086e-05,
      "loss": 1.9406,
      "step": 189508
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.3629562854766846,
      "learning_rate": 4.544145370751878e-05,
      "loss": 1.8844,
      "step": 189509
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.355328917503357,
      "learning_rate": 4.543928921143392e-05,
      "loss": 2.081,
      "step": 189510
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.7230957746505737,
      "learning_rate": 4.543712476267669e-05,
      "loss": 1.8906,
      "step": 189511
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.5646629333496094,
      "learning_rate": 4.543496036124744e-05,
      "loss": 2.1261,
      "step": 189512
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.47723388671875,
      "learning_rate": 4.54327960071467e-05,
      "loss": 1.9383,
      "step": 189513
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.411707878112793,
      "learning_rate": 4.5430631700374766e-05,
      "loss": 2.0589,
      "step": 189514
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.3205989599227905,
      "learning_rate": 4.542846744093214e-05,
      "loss": 1.8689,
      "step": 189515
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.632609486579895,
      "learning_rate": 4.542630322881915e-05,
      "loss": 2.0357,
      "step": 189516
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.5359448194503784,
      "learning_rate": 4.542413906403624e-05,
      "loss": 1.8551,
      "step": 189517
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.3078354597091675,
      "learning_rate": 4.5421974946583695e-05,
      "loss": 1.9258,
      "step": 189518
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.3629028797149658,
      "learning_rate": 4.541981087646209e-05,
      "loss": 1.9135,
      "step": 189519
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.7279188632965088,
      "learning_rate": 4.5417646853671696e-05,
      "loss": 2.0454,
      "step": 189520
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.3239752054214478,
      "learning_rate": 4.5415482878213006e-05,
      "loss": 2.0884,
      "step": 189521
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.4665354490280151,
      "learning_rate": 4.541331895008639e-05,
      "loss": 2.1292,
      "step": 189522
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.301504373550415,
      "learning_rate": 4.541115506929227e-05,
      "loss": 1.9426,
      "step": 189523
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.4260952472686768,
      "learning_rate": 4.540899123583093e-05,
      "loss": 2.0762,
      "step": 189524
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.533089518547058,
      "learning_rate": 4.540682744970292e-05,
      "loss": 1.9349,
      "step": 189525
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.6512717008590698,
      "learning_rate": 4.5404663710908554e-05,
      "loss": 2.0798,
      "step": 189526
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.4234232902526855,
      "learning_rate": 4.5402500019448315e-05,
      "loss": 2.1066,
      "step": 189527
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.4648948907852173,
      "learning_rate": 4.5400336375322554e-05,
      "loss": 1.9992,
      "step": 189528
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.4260907173156738,
      "learning_rate": 4.53981727785317e-05,
      "loss": 2.0172,
      "step": 189529
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.808770775794983,
      "learning_rate": 4.539600922907604e-05,
      "loss": 2.2094,
      "step": 189530
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.4272515773773193,
      "learning_rate": 4.539384572695615e-05,
      "loss": 2.0052,
      "step": 189531
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.3724989891052246,
      "learning_rate": 4.539168227217226e-05,
      "loss": 2.2091,
      "step": 189532
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.3319092988967896,
      "learning_rate": 4.538951886472494e-05,
      "loss": 2.074,
      "step": 189533
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.4159293174743652,
      "learning_rate": 4.538735550461452e-05,
      "loss": 2.1474,
      "step": 189534
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.3483827114105225,
      "learning_rate": 4.53851921918414e-05,
      "loss": 1.9842,
      "step": 189535
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.3958524465560913,
      "learning_rate": 4.538302892640588e-05,
      "loss": 1.9664,
      "step": 189536
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.3381335735321045,
      "learning_rate": 4.538086570830857e-05,
      "loss": 2.1158,
      "step": 189537
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.4001938104629517,
      "learning_rate": 4.5378702537549685e-05,
      "loss": 2.0285,
      "step": 189538
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.4322370290756226,
      "learning_rate": 4.5376539414129765e-05,
      "loss": 2.03,
      "step": 189539
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.4069697856903076,
      "learning_rate": 4.5374376338049147e-05,
      "loss": 2.0987,
      "step": 189540
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.6475964784622192,
      "learning_rate": 4.537221330930826e-05,
      "loss": 1.9458,
      "step": 189541
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.4595866203308105,
      "learning_rate": 4.5370050327907404e-05,
      "loss": 1.7071,
      "step": 189542
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.2482930421829224,
      "learning_rate": 4.5367887393847147e-05,
      "loss": 2.179,
      "step": 189543
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.4102181196212769,
      "learning_rate": 4.536572450712772e-05,
      "loss": 1.958,
      "step": 189544
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.4709621667861938,
      "learning_rate": 4.536356166774966e-05,
      "loss": 1.9594,
      "step": 189545
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.3742094039916992,
      "learning_rate": 4.5361398875713365e-05,
      "loss": 1.8616,
      "step": 189546
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.3849163055419922,
      "learning_rate": 4.535923613101916e-05,
      "loss": 2.0126,
      "step": 189547
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.4048126935958862,
      "learning_rate": 4.5357073433667424e-05,
      "loss": 1.9595,
      "step": 189548
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.4883265495300293,
      "learning_rate": 4.535491078365868e-05,
      "loss": 2.0122,
      "step": 189549
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.3745261430740356,
      "learning_rate": 4.53527481809932e-05,
      "loss": 1.8732,
      "step": 189550
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.6127647161483765,
      "learning_rate": 4.535058562567152e-05,
      "loss": 1.7513,
      "step": 189551
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.5218199491500854,
      "learning_rate": 4.5348423117693965e-05,
      "loss": 2.1314,
      "step": 189552
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.4885671138763428,
      "learning_rate": 4.534626065706094e-05,
      "loss": 1.8966,
      "step": 189553
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.4068682193756104,
      "learning_rate": 4.534409824377281e-05,
      "loss": 1.9374,
      "step": 189554
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.3047925233840942,
      "learning_rate": 4.534193587783007e-05,
      "loss": 1.9131,
      "step": 189555
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.4117752313613892,
      "learning_rate": 4.5339773559232994e-05,
      "loss": 2.0993,
      "step": 189556
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.2531025409698486,
      "learning_rate": 4.533761128798214e-05,
      "loss": 1.9799,
      "step": 189557
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.3784141540527344,
      "learning_rate": 4.533544906407785e-05,
      "loss": 1.9421,
      "step": 189558
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.3823515176773071,
      "learning_rate": 4.533328688752043e-05,
      "loss": 1.9162,
      "step": 189559
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.3891334533691406,
      "learning_rate": 4.533112475831039e-05,
      "loss": 2.2615,
      "step": 189560
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.4827678203582764,
      "learning_rate": 4.5328962676448144e-05,
      "loss": 1.9102,
      "step": 189561
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.4584770202636719,
      "learning_rate": 4.5326800641933956e-05,
      "loss": 1.7851,
      "step": 189562
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.384376883506775,
      "learning_rate": 4.5324638654768395e-05,
      "loss": 2.0442,
      "step": 189563
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.3770685195922852,
      "learning_rate": 4.532247671495179e-05,
      "loss": 2.0756,
      "step": 189564
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.4264267683029175,
      "learning_rate": 4.5320314822484515e-05,
      "loss": 1.9917,
      "step": 189565
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.338218331336975,
      "learning_rate": 4.531815297736703e-05,
      "loss": 1.9367,
      "step": 189566
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.350791096687317,
      "learning_rate": 4.531599117959963e-05,
      "loss": 1.9679,
      "step": 189567
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.3154507875442505,
      "learning_rate": 4.531382942918289e-05,
      "loss": 2.0103,
      "step": 189568
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.3069132566452026,
      "learning_rate": 4.531166772611711e-05,
      "loss": 2.1436,
      "step": 189569
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.395814299583435,
      "learning_rate": 4.530950607040265e-05,
      "loss": 1.8313,
      "step": 189570
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.3559386730194092,
      "learning_rate": 4.5307344462040006e-05,
      "loss": 2.1228,
      "step": 189571
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.385492205619812,
      "learning_rate": 4.530518290102952e-05,
      "loss": 1.9575,
      "step": 189572
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.3689223527908325,
      "learning_rate": 4.5303021387371566e-05,
      "loss": 1.9443,
      "step": 189573
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.367476463317871,
      "learning_rate": 4.530085992106662e-05,
      "loss": 1.9865,
      "step": 189574
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.5756689310073853,
      "learning_rate": 4.52986985021151e-05,
      "loss": 1.9643,
      "step": 189575
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.4699801206588745,
      "learning_rate": 4.529653713051727e-05,
      "loss": 2.0242,
      "step": 189576
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.3430380821228027,
      "learning_rate": 4.52943758062737e-05,
      "loss": 1.8411,
      "step": 189577
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.3923360109329224,
      "learning_rate": 4.5292214529384674e-05,
      "loss": 2.006,
      "step": 189578
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.5505048036575317,
      "learning_rate": 4.5290053299850605e-05,
      "loss": 2.1354,
      "step": 189579
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.3063234090805054,
      "learning_rate": 4.528789211767199e-05,
      "loss": 1.903,
      "step": 189580
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.2344995737075806,
      "learning_rate": 4.528573098284909e-05,
      "loss": 1.9497,
      "step": 189581
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.6475492715835571,
      "learning_rate": 4.5283569895382445e-05,
      "loss": 1.983,
      "step": 189582
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.4315294027328491,
      "learning_rate": 4.528140885527238e-05,
      "loss": 2.0328,
      "step": 189583
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.2358404397964478,
      "learning_rate": 4.5279247862519305e-05,
      "loss": 1.9886,
      "step": 189584
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.436567783355713,
      "learning_rate": 4.5277086917123584e-05,
      "loss": 2.057,
      "step": 189585
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.4877575635910034,
      "learning_rate": 4.527492601908571e-05,
      "loss": 2.0501,
      "step": 189586
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.460944652557373,
      "learning_rate": 4.527276516840599e-05,
      "loss": 2.0195,
      "step": 189587
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.3748314380645752,
      "learning_rate": 4.527060436508492e-05,
      "loss": 2.0758,
      "step": 189588
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.35824453830719,
      "learning_rate": 4.526844360912286e-05,
      "loss": 2.0354,
      "step": 189589
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.5727492570877075,
      "learning_rate": 4.5266282900520226e-05,
      "loss": 1.9273,
      "step": 189590
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.420884132385254,
      "learning_rate": 4.52641222392773e-05,
      "loss": 1.8148,
      "step": 189591
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.3448065519332886,
      "learning_rate": 4.526196162539466e-05,
      "loss": 2.0069,
      "step": 189592
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.423388123512268,
      "learning_rate": 4.5259801058872536e-05,
      "loss": 1.9403,
      "step": 189593
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.5080794095993042,
      "learning_rate": 4.5257640539711526e-05,
      "loss": 2.135,
      "step": 189594
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.3428646326065063,
      "learning_rate": 4.525548006791193e-05,
      "loss": 1.9513,
      "step": 189595
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.3694443702697754,
      "learning_rate": 4.525331964347412e-05,
      "loss": 1.954,
      "step": 189596
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.3968067169189453,
      "learning_rate": 4.525115926639848e-05,
      "loss": 2.1122,
      "step": 189597
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.423617959022522,
      "learning_rate": 4.524899893668553e-05,
      "loss": 1.9493,
      "step": 189598
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.476521372795105,
      "learning_rate": 4.5246838654335526e-05,
      "loss": 2.1605,
      "step": 189599
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.3248825073242188,
      "learning_rate": 4.5244678419349e-05,
      "loss": 1.8991,
      "step": 189600
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.3569161891937256,
      "learning_rate": 4.524251823172632e-05,
      "loss": 1.9345,
      "step": 189601
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.5548911094665527,
      "learning_rate": 4.5240358091467856e-05,
      "loss": 2.1951,
      "step": 189602
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.3398810625076294,
      "learning_rate": 4.523819799857393e-05,
      "loss": 2.0195,
      "step": 189603
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.3804329633712769,
      "learning_rate": 4.5236037953045124e-05,
      "loss": 1.7984,
      "step": 189604
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.2584270238876343,
      "learning_rate": 4.523387795488169e-05,
      "loss": 1.9691,
      "step": 189605
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.4034873247146606,
      "learning_rate": 4.523171800408414e-05,
      "loss": 2.1054,
      "step": 189606
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.467623233795166,
      "learning_rate": 4.52295581006528e-05,
      "loss": 1.9937,
      "step": 189607
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.3969537019729614,
      "learning_rate": 4.522739824458813e-05,
      "loss": 2.2838,
      "step": 189608
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.3791766166687012,
      "learning_rate": 4.522523843589041e-05,
      "loss": 2.0068,
      "step": 189609
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.4553173780441284,
      "learning_rate": 4.52230786745602e-05,
      "loss": 1.9789,
      "step": 189610
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.3316959142684937,
      "learning_rate": 4.522091896059776e-05,
      "loss": 2.2709,
      "step": 189611
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.4295800924301147,
      "learning_rate": 4.52187592940036e-05,
      "loss": 2.0637,
      "step": 189612
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.5108839273452759,
      "learning_rate": 4.521659967477812e-05,
      "loss": 1.8206,
      "step": 189613
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.586849570274353,
      "learning_rate": 4.521444010292168e-05,
      "loss": 2.0719,
      "step": 189614
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.48111891746521,
      "learning_rate": 4.521228057843461e-05,
      "loss": 2.1292,
      "step": 189615
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.5170634984970093,
      "learning_rate": 4.521012110131745e-05,
      "loss": 2.0368,
      "step": 189616
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.3873265981674194,
      "learning_rate": 4.520796167157047e-05,
      "loss": 1.9619,
      "step": 189617
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.525928020477295,
      "learning_rate": 4.5205802289194225e-05,
      "loss": 2.0735,
      "step": 189618
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.33826744556427,
      "learning_rate": 4.520364295418902e-05,
      "loss": 2.0285,
      "step": 189619
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.4626846313476562,
      "learning_rate": 4.520148366655526e-05,
      "loss": 1.9241,
      "step": 189620
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.445283055305481,
      "learning_rate": 4.5199324426293303e-05,
      "loss": 2.2265,
      "step": 189621
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.3436977863311768,
      "learning_rate": 4.5197165233403655e-05,
      "loss": 1.8207,
      "step": 189622
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.3385428190231323,
      "learning_rate": 4.519500608788657e-05,
      "loss": 1.8613,
      "step": 189623
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.421294927597046,
      "learning_rate": 4.519284698974267e-05,
      "loss": 1.8966,
      "step": 189624
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.346104621887207,
      "learning_rate": 4.5190687938972206e-05,
      "loss": 2.1769,
      "step": 189625
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.5386333465576172,
      "learning_rate": 4.518852893557558e-05,
      "loss": 2.1958,
      "step": 189626
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.4523870944976807,
      "learning_rate": 4.518636997955315e-05,
      "loss": 2.1363,
      "step": 189627
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.6576480865478516,
      "learning_rate": 4.5184211070905464e-05,
      "loss": 1.9925,
      "step": 189628
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.3288605213165283,
      "learning_rate": 4.518205220963278e-05,
      "loss": 2.209,
      "step": 189629
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.3666075468063354,
      "learning_rate": 4.5179893395735634e-05,
      "loss": 1.9195,
      "step": 189630
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.450676441192627,
      "learning_rate": 4.517773462921436e-05,
      "loss": 2.2039,
      "step": 189631
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.340795874595642,
      "learning_rate": 4.517557591006935e-05,
      "loss": 1.895,
      "step": 189632
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.4222537279129028,
      "learning_rate": 4.517341723830091e-05,
      "loss": 2.2198,
      "step": 189633
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.4135353565216064,
      "learning_rate": 4.5171258613909643e-05,
      "loss": 2.0839,
      "step": 189634
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.2961782217025757,
      "learning_rate": 4.516910003689578e-05,
      "loss": 1.9914,
      "step": 189635
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.5964510440826416,
      "learning_rate": 4.516694150725988e-05,
      "loss": 1.9729,
      "step": 189636
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.543565034866333,
      "learning_rate": 4.516478302500215e-05,
      "loss": 2.0911,
      "step": 189637
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.426851511001587,
      "learning_rate": 4.516262459012325e-05,
      "loss": 1.8783,
      "step": 189638
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.2882311344146729,
      "learning_rate": 4.5160466202623294e-05,
      "loss": 2.1701,
      "step": 189639
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.473223328590393,
      "learning_rate": 4.515830786250286e-05,
      "loss": 1.8496,
      "step": 189640
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.4349982738494873,
      "learning_rate": 4.515614956976227e-05,
      "loss": 1.9625,
      "step": 189641
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.456346035003662,
      "learning_rate": 4.515399132440204e-05,
      "loss": 2.0408,
      "step": 189642
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.5140485763549805,
      "learning_rate": 4.515183312642238e-05,
      "loss": 1.9756,
      "step": 189643
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.4079607725143433,
      "learning_rate": 4.5149674975823915e-05,
      "loss": 1.7972,
      "step": 189644
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.2974967956542969,
      "learning_rate": 4.5147516872606916e-05,
      "loss": 2.0148,
      "step": 189645
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.335343837738037,
      "learning_rate": 4.5145358816771816e-05,
      "loss": 1.9945,
      "step": 189646
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.4504671096801758,
      "learning_rate": 4.514320080831891e-05,
      "loss": 1.923,
      "step": 189647
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.3636096715927124,
      "learning_rate": 4.514104284724877e-05,
      "loss": 2.0734,
      "step": 189648
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.4425462484359741,
      "learning_rate": 4.513888493356167e-05,
      "loss": 1.8435,
      "step": 189649
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.7705049514770508,
      "learning_rate": 4.513672706725813e-05,
      "loss": 1.8893,
      "step": 189650
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.3648773431777954,
      "learning_rate": 4.513456924833848e-05,
      "loss": 1.9705,
      "step": 189651
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.302748441696167,
      "learning_rate": 4.513241147680313e-05,
      "loss": 2.0679,
      "step": 189652
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.3454910516738892,
      "learning_rate": 4.513025375265241e-05,
      "loss": 1.7789,
      "step": 189653
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.6479196548461914,
      "learning_rate": 4.512809607588682e-05,
      "loss": 1.9374,
      "step": 189654
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.8108969926834106,
      "learning_rate": 4.5125938446506694e-05,
      "loss": 2.056,
      "step": 189655
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.4629430770874023,
      "learning_rate": 4.5123780864512524e-05,
      "loss": 1.9684,
      "step": 189656
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.4561879634857178,
      "learning_rate": 4.512162332990466e-05,
      "loss": 1.98,
      "step": 189657
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.5053964853286743,
      "learning_rate": 4.511946584268341e-05,
      "loss": 2.0368,
      "step": 189658
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.4075746536254883,
      "learning_rate": 4.511730840284936e-05,
      "loss": 1.9035,
      "step": 189659
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.4772902727127075,
      "learning_rate": 4.511515101040281e-05,
      "loss": 2.0161,
      "step": 189660
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.5193889141082764,
      "learning_rate": 4.5112993665344076e-05,
      "loss": 1.9386,
      "step": 189661
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.5543968677520752,
      "learning_rate": 4.511083636767374e-05,
      "loss": 1.8302,
      "step": 189662
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.3641400337219238,
      "learning_rate": 4.51086791173921e-05,
      "loss": 1.9871,
      "step": 189663
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.3857561349868774,
      "learning_rate": 4.5106521914499484e-05,
      "loss": 1.9928,
      "step": 189664
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.6679139137268066,
      "learning_rate": 4.510436475899646e-05,
      "loss": 2.227,
      "step": 189665
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.3524402379989624,
      "learning_rate": 4.510220765088326e-05,
      "loss": 1.964,
      "step": 189666
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.6908867359161377,
      "learning_rate": 4.510005059016045e-05,
      "loss": 2.0561,
      "step": 189667
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.5725349187850952,
      "learning_rate": 4.509789357682837e-05,
      "loss": 1.8682,
      "step": 189668
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.592924952507019,
      "learning_rate": 4.509573661088741e-05,
      "loss": 2.0023,
      "step": 189669
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.2893527746200562,
      "learning_rate": 4.5093579692337843e-05,
      "loss": 1.88,
      "step": 189670
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.3925400972366333,
      "learning_rate": 4.50914228211803e-05,
      "loss": 1.9489,
      "step": 189671
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.4670783281326294,
      "learning_rate": 4.5089265997415014e-05,
      "loss": 2.0545,
      "step": 189672
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.3745858669281006,
      "learning_rate": 4.5087109221042486e-05,
      "loss": 2.2235,
      "step": 189673
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.3527781963348389,
      "learning_rate": 4.508495249206311e-05,
      "loss": 2.1616,
      "step": 189674
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.3827018737792969,
      "learning_rate": 4.508279581047722e-05,
      "loss": 2.046,
      "step": 189675
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.4460138082504272,
      "learning_rate": 4.508063917628519e-05,
      "loss": 2.113,
      "step": 189676
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.3987362384796143,
      "learning_rate": 4.5078482589487555e-05,
      "loss": 1.8444,
      "step": 189677
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.4121119976043701,
      "learning_rate": 4.507632605008457e-05,
      "loss": 1.9057,
      "step": 189678
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.4976266622543335,
      "learning_rate": 4.507416955807677e-05,
      "loss": 2.0906,
      "step": 189679
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.4436949491500854,
      "learning_rate": 4.507201311346452e-05,
      "loss": 2.0811,
      "step": 189680
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.4157261848449707,
      "learning_rate": 4.506985671624817e-05,
      "loss": 1.9968,
      "step": 189681
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.4739314317703247,
      "learning_rate": 4.5067700366428095e-05,
      "loss": 1.8767,
      "step": 189682
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.486914038658142,
      "learning_rate": 4.5065544064004775e-05,
      "loss": 2.012,
      "step": 189683
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.3508292436599731,
      "learning_rate": 4.5063387808978546e-05,
      "loss": 1.7659,
      "step": 189684
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.2588611841201782,
      "learning_rate": 4.5061231601349936e-05,
      "loss": 2.0697,
      "step": 189685
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.441315770149231,
      "learning_rate": 4.505907544111921e-05,
      "loss": 1.924,
      "step": 189686
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.4673285484313965,
      "learning_rate": 4.5056919328286834e-05,
      "loss": 1.8651,
      "step": 189687
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.3723269701004028,
      "learning_rate": 4.505476326285314e-05,
      "loss": 1.948,
      "step": 189688
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.6309362649917603,
      "learning_rate": 4.5052607244818606e-05,
      "loss": 1.9525,
      "step": 189689
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.3360698223114014,
      "learning_rate": 4.505045127418359e-05,
      "loss": 1.7546,
      "step": 189690
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.4900240898132324,
      "learning_rate": 4.504829535094852e-05,
      "loss": 2.0024,
      "step": 189691
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.3765203952789307,
      "learning_rate": 4.504613947511383e-05,
      "loss": 2.0742,
      "step": 189692
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.4902031421661377,
      "learning_rate": 4.504398364667986e-05,
      "loss": 1.9771,
      "step": 189693
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.3738794326782227,
      "learning_rate": 4.504182786564694e-05,
      "loss": 1.8697,
      "step": 189694
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.7774808406829834,
      "learning_rate": 4.503967213201563e-05,
      "loss": 2.0857,
      "step": 189695
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.523027777671814,
      "learning_rate": 4.503751644578622e-05,
      "loss": 2.0102,
      "step": 189696
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.3242546319961548,
      "learning_rate": 4.503536080695918e-05,
      "loss": 1.7568,
      "step": 189697
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.5596038103103638,
      "learning_rate": 4.503320521553489e-05,
      "loss": 1.9741,
      "step": 189698
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.3549925088882446,
      "learning_rate": 4.503104967151374e-05,
      "loss": 1.9328,
      "step": 189699
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.3850393295288086,
      "learning_rate": 4.5028894174896077e-05,
      "loss": 1.8567,
      "step": 189700
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.4017921686172485,
      "learning_rate": 4.50267387256824e-05,
      "loss": 1.9529,
      "step": 189701
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.3792803287506104,
      "learning_rate": 4.5024583323873e-05,
      "loss": 1.9822,
      "step": 189702
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.502278447151184,
      "learning_rate": 4.502242796946844e-05,
      "loss": 1.9663,
      "step": 189703
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.4135382175445557,
      "learning_rate": 4.502027266246894e-05,
      "loss": 2.0478,
      "step": 189704
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.5917915105819702,
      "learning_rate": 4.501811740287511e-05,
      "loss": 2.008,
      "step": 189705
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.4144068956375122,
      "learning_rate": 4.501596219068709e-05,
      "loss": 2.0441,
      "step": 189706
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.3309321403503418,
      "learning_rate": 4.501380702590549e-05,
      "loss": 2.0296,
      "step": 189707
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.4955902099609375,
      "learning_rate": 4.501165190853057e-05,
      "loss": 2.0489,
      "step": 189708
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.5687804222106934,
      "learning_rate": 4.5009496838562854e-05,
      "loss": 2.2285,
      "step": 189709
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.424629807472229,
      "learning_rate": 4.500734181600262e-05,
      "loss": 2.1101,
      "step": 189710
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.3781988620758057,
      "learning_rate": 4.500518684085049e-05,
      "loss": 2.0089,
      "step": 189711
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.3634601831436157,
      "learning_rate": 4.5003031913106546e-05,
      "loss": 1.9687,
      "step": 189712
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.2617014646530151,
      "learning_rate": 4.500087703277144e-05,
      "loss": 1.7903,
      "step": 189713
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.397790789604187,
      "learning_rate": 4.4998722199845417e-05,
      "loss": 1.8787,
      "step": 189714
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.400482177734375,
      "learning_rate": 4.4996567414329e-05,
      "loss": 1.8392,
      "step": 189715
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.3752727508544922,
      "learning_rate": 4.499441267622249e-05,
      "loss": 2.0477,
      "step": 189716
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.3730605840682983,
      "learning_rate": 4.499225798552646e-05,
      "loss": 2.0012,
      "step": 189717
    },
    {
      "epoch": 2.47,
      "grad_norm": 2.165656089782715,
      "learning_rate": 4.4990103342241036e-05,
      "loss": 2.0427,
      "step": 189718
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.2811543941497803,
      "learning_rate": 4.4987948746366817e-05,
      "loss": 1.9934,
      "step": 189719
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.3513211011886597,
      "learning_rate": 4.4985794197904114e-05,
      "loss": 1.7846,
      "step": 189720
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.4934314489364624,
      "learning_rate": 4.498363969685341e-05,
      "loss": 2.2079,
      "step": 189721
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.3658438920974731,
      "learning_rate": 4.498148524321502e-05,
      "loss": 2.0914,
      "step": 189722
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.302846908569336,
      "learning_rate": 4.49793308369895e-05,
      "loss": 2.0769,
      "step": 189723
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.3502850532531738,
      "learning_rate": 4.497717647817702e-05,
      "loss": 2.0373,
      "step": 189724
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.4175305366516113,
      "learning_rate": 4.4975022166778154e-05,
      "loss": 1.9314,
      "step": 189725
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.2902241945266724,
      "learning_rate": 4.4972867902793154e-05,
      "loss": 1.8706,
      "step": 189726
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.7923282384872437,
      "learning_rate": 4.4970713686222626e-05,
      "loss": 2.208,
      "step": 189727
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.4262381792068481,
      "learning_rate": 4.496855951706674e-05,
      "loss": 2.1221,
      "step": 189728
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.417109489440918,
      "learning_rate": 4.4966405395326123e-05,
      "loss": 2.0355,
      "step": 189729
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.4571433067321777,
      "learning_rate": 4.4964251321001043e-05,
      "loss": 1.9717,
      "step": 189730
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.5262699127197266,
      "learning_rate": 4.496209729409194e-05,
      "loss": 2.0525,
      "step": 189731
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.384305477142334,
      "learning_rate": 4.49599433145991e-05,
      "loss": 2.2236,
      "step": 189732
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.5109118223190308,
      "learning_rate": 4.49577893825231e-05,
      "loss": 2.0237,
      "step": 189733
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.3481945991516113,
      "learning_rate": 4.4955635497864204e-05,
      "loss": 1.9,
      "step": 189734
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.3386409282684326,
      "learning_rate": 4.495348166062295e-05,
      "loss": 1.9688,
      "step": 189735
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.2798596620559692,
      "learning_rate": 4.4951327870799626e-05,
      "loss": 2.0601,
      "step": 189736
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.32938551902771,
      "learning_rate": 4.494917412839467e-05,
      "loss": 1.8363,
      "step": 189737
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.3295150995254517,
      "learning_rate": 4.494702043340842e-05,
      "loss": 1.7527,
      "step": 189738
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.5977702140808105,
      "learning_rate": 4.494486678584141e-05,
      "loss": 2.0894,
      "step": 189739
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.2130554914474487,
      "learning_rate": 4.494271318569389e-05,
      "loss": 2.0115,
      "step": 189740
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.4695475101470947,
      "learning_rate": 4.494055963296641e-05,
      "loss": 1.9228,
      "step": 189741
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.3224750757217407,
      "learning_rate": 4.493840612765926e-05,
      "loss": 2.1468,
      "step": 189742
    },
    {
      "epoch": 2.47,
      "grad_norm": 2.039306640625,
      "learning_rate": 4.493625266977285e-05,
      "loss": 2.0827,
      "step": 189743
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.30266535282135,
      "learning_rate": 4.493409925930764e-05,
      "loss": 2.2177,
      "step": 189744
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.3835783004760742,
      "learning_rate": 4.4931945896264e-05,
      "loss": 2.1859,
      "step": 189745
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.373063087463379,
      "learning_rate": 4.492979258064229e-05,
      "loss": 1.9683,
      "step": 189746
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.3659653663635254,
      "learning_rate": 4.4927639312442976e-05,
      "loss": 2.1692,
      "step": 189747
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.358389139175415,
      "learning_rate": 4.4925486091666465e-05,
      "loss": 2.103,
      "step": 189748
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.3786016702651978,
      "learning_rate": 4.492333291831302e-05,
      "loss": 1.9677,
      "step": 189749
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.5212147235870361,
      "learning_rate": 4.49211797923832e-05,
      "loss": 2.09,
      "step": 189750
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.3799351453781128,
      "learning_rate": 4.491902671387738e-05,
      "loss": 2.0939,
      "step": 189751
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.3968650102615356,
      "learning_rate": 4.4916873682795864e-05,
      "loss": 2.1401,
      "step": 189752
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.4527153968811035,
      "learning_rate": 4.491472069913917e-05,
      "loss": 2.0022,
      "step": 189753
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.537372350692749,
      "learning_rate": 4.4912567762907645e-05,
      "loss": 2.1083,
      "step": 189754
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.4388322830200195,
      "learning_rate": 4.491041487410162e-05,
      "loss": 1.9981,
      "step": 189755
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.2960246801376343,
      "learning_rate": 4.490826203272162e-05,
      "loss": 2.0784,
      "step": 189756
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.4581899642944336,
      "learning_rate": 4.490610923876792e-05,
      "loss": 2.0351,
      "step": 189757
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.5042405128479004,
      "learning_rate": 4.4903956492241077e-05,
      "loss": 2.1132,
      "step": 189758
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.527732253074646,
      "learning_rate": 4.49018037931414e-05,
      "loss": 2.1121,
      "step": 189759
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.3241052627563477,
      "learning_rate": 4.489965114146929e-05,
      "loss": 2.0411,
      "step": 189760
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.561768889427185,
      "learning_rate": 4.489749853722509e-05,
      "loss": 1.9825,
      "step": 189761
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.390607476234436,
      "learning_rate": 4.489534598040934e-05,
      "loss": 1.8352,
      "step": 189762
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.5023701190948486,
      "learning_rate": 4.489319347102227e-05,
      "loss": 2.0443,
      "step": 189763
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.4088263511657715,
      "learning_rate": 4.489104100906443e-05,
      "loss": 2.0327,
      "step": 189764
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.3332360982894897,
      "learning_rate": 4.488888859453619e-05,
      "loss": 2.0481,
      "step": 189765
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.2874518632888794,
      "learning_rate": 4.4886736227437904e-05,
      "loss": 2.1388,
      "step": 189766
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.371374487876892,
      "learning_rate": 4.488458390776991e-05,
      "loss": 2.0153,
      "step": 189767
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.3497155904769897,
      "learning_rate": 4.488243163553278e-05,
      "loss": 2.0073,
      "step": 189768
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.4456382989883423,
      "learning_rate": 4.488027941072674e-05,
      "loss": 2.035,
      "step": 189769
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.5225385427474976,
      "learning_rate": 4.4878127233352326e-05,
      "loss": 1.9359,
      "step": 189770
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.423171043395996,
      "learning_rate": 4.4875975103409835e-05,
      "loss": 1.816,
      "step": 189771
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.3610501289367676,
      "learning_rate": 4.487382302089983e-05,
      "loss": 1.8639,
      "step": 189772
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.489194631576538,
      "learning_rate": 4.487167098582249e-05,
      "loss": 1.9697,
      "step": 189773
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.5538095235824585,
      "learning_rate": 4.486951899817838e-05,
      "loss": 2.2004,
      "step": 189774
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.4939709901809692,
      "learning_rate": 4.4867367057967784e-05,
      "loss": 2.0933,
      "step": 189775
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.4725193977355957,
      "learning_rate": 4.486521516519122e-05,
      "loss": 2.1946,
      "step": 189776
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.363571286201477,
      "learning_rate": 4.486306331984897e-05,
      "loss": 1.7729,
      "step": 189777
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.4567875862121582,
      "learning_rate": 4.4860911521941613e-05,
      "loss": 2.1116,
      "step": 189778
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.3775509595870972,
      "learning_rate": 4.485875977146928e-05,
      "loss": 1.8872,
      "step": 189779
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.4466358423233032,
      "learning_rate": 4.4856608068432634e-05,
      "loss": 2.0833,
      "step": 189780
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.5857535600662231,
      "learning_rate": 4.485445641283184e-05,
      "loss": 2.1747,
      "step": 189781
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.5673694610595703,
      "learning_rate": 4.4852304804667505e-05,
      "loss": 1.9073,
      "step": 189782
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.305870532989502,
      "learning_rate": 4.485015324393989e-05,
      "loss": 1.6985,
      "step": 189783
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.5946110486984253,
      "learning_rate": 4.4848001730649595e-05,
      "loss": 1.8634,
      "step": 189784
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.33597993850708,
      "learning_rate": 4.484585026479669e-05,
      "loss": 2.023,
      "step": 189785
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.7700562477111816,
      "learning_rate": 4.484369884638187e-05,
      "loss": 1.999,
      "step": 189786
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.3799031972885132,
      "learning_rate": 4.4841547475405335e-05,
      "loss": 2.0917,
      "step": 189787
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.4745137691497803,
      "learning_rate": 4.483939615186766e-05,
      "loss": 2.1144,
      "step": 189788
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.4858818054199219,
      "learning_rate": 4.483724487576906e-05,
      "loss": 2.0559,
      "step": 189789
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.362626314163208,
      "learning_rate": 4.483509364711019e-05,
      "loss": 2.0565,
      "step": 189790
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.291771411895752,
      "learning_rate": 4.483294246589113e-05,
      "loss": 2.0784,
      "step": 189791
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.4067376852035522,
      "learning_rate": 4.4830791332112525e-05,
      "loss": 2.1599,
      "step": 189792
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.4089308977127075,
      "learning_rate": 4.4828640245774635e-05,
      "loss": 2.3155,
      "step": 189793
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.2936153411865234,
      "learning_rate": 4.482648920687797e-05,
      "loss": 1.9648,
      "step": 189794
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.3368771076202393,
      "learning_rate": 4.4824338215422814e-05,
      "loss": 1.7814,
      "step": 189795
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.5109785795211792,
      "learning_rate": 4.4822187271409784e-05,
      "loss": 2.0028,
      "step": 189796
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.4940427541732788,
      "learning_rate": 4.4820036374838994e-05,
      "loss": 2.2505,
      "step": 189797
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.303867220878601,
      "learning_rate": 4.481788552571103e-05,
      "loss": 1.9593,
      "step": 189798
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.5363224744796753,
      "learning_rate": 4.481573472402614e-05,
      "loss": 2.1536,
      "step": 189799
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.5086535215377808,
      "learning_rate": 4.4813583969784936e-05,
      "loss": 1.9668,
      "step": 189800
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.5675784349441528,
      "learning_rate": 4.4811433262987616e-05,
      "loss": 2.091,
      "step": 189801
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.4345341920852661,
      "learning_rate": 4.480928260363481e-05,
      "loss": 2.1215,
      "step": 189802
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.4822375774383545,
      "learning_rate": 4.480713199172662e-05,
      "loss": 2.0167,
      "step": 189803
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.2938729524612427,
      "learning_rate": 4.4804981427263676e-05,
      "loss": 1.828,
      "step": 189804
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.405577301979065,
      "learning_rate": 4.480283091024624e-05,
      "loss": 2.0751,
      "step": 189805
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.5677536725997925,
      "learning_rate": 4.4800680440674866e-05,
      "loss": 1.9633,
      "step": 189806
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.4183070659637451,
      "learning_rate": 4.479853001854976e-05,
      "loss": 2.0773,
      "step": 189807
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.466184139251709,
      "learning_rate": 4.479637964387157e-05,
      "loss": 1.9892,
      "step": 189808
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.2447587251663208,
      "learning_rate": 4.479422931664043e-05,
      "loss": 1.9522,
      "step": 189809
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.3883895874023438,
      "learning_rate": 4.4792079036856935e-05,
      "loss": 1.9801,
      "step": 189810
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.340177059173584,
      "learning_rate": 4.478992880452131e-05,
      "loss": 2.0931,
      "step": 189811
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.5692315101623535,
      "learning_rate": 4.4787778619634144e-05,
      "loss": 2.0035,
      "step": 189812
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.3249313831329346,
      "learning_rate": 4.478562848219565e-05,
      "loss": 2.0473,
      "step": 189813
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.4485124349594116,
      "learning_rate": 4.478347839220649e-05,
      "loss": 2.0711,
      "step": 189814
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.3086720705032349,
      "learning_rate": 4.4781328349666755e-05,
      "loss": 2.1231,
      "step": 189815
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.3413344621658325,
      "learning_rate": 4.477917835457706e-05,
      "loss": 2.1213,
      "step": 189816
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.453513741493225,
      "learning_rate": 4.477702840693768e-05,
      "loss": 1.9859,
      "step": 189817
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.4821377992630005,
      "learning_rate": 4.47748785067491e-05,
      "loss": 1.8736,
      "step": 189818
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.3492803573608398,
      "learning_rate": 4.477272865401164e-05,
      "loss": 2.0329,
      "step": 189819
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.3416224718093872,
      "learning_rate": 4.4770578848725815e-05,
      "loss": 2.005,
      "step": 189820
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.4244366884231567,
      "learning_rate": 4.4768429090891964e-05,
      "loss": 2.0666,
      "step": 189821
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.3759499788284302,
      "learning_rate": 4.4766279380510464e-05,
      "loss": 2.0324,
      "step": 189822
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.410609245300293,
      "learning_rate": 4.47641297175817e-05,
      "loss": 2.0217,
      "step": 189823
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.5411416292190552,
      "learning_rate": 4.476198010210611e-05,
      "loss": 1.9143,
      "step": 189824
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.5832059383392334,
      "learning_rate": 4.475983053408406e-05,
      "loss": 2.1933,
      "step": 189825
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.2754385471343994,
      "learning_rate": 4.4757681013516025e-05,
      "loss": 2.0476,
      "step": 189826
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.4626438617706299,
      "learning_rate": 4.475553154040239e-05,
      "loss": 1.975,
      "step": 189827
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.2366474866867065,
      "learning_rate": 4.4753382114743394e-05,
      "loss": 1.7098,
      "step": 189828
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.3648568391799927,
      "learning_rate": 4.4751232736539675e-05,
      "loss": 1.952,
      "step": 189829
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.456488013267517,
      "learning_rate": 4.47490834057915e-05,
      "loss": 1.9373,
      "step": 189830
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.4182640314102173,
      "learning_rate": 4.474693412249922e-05,
      "loss": 2.0715,
      "step": 189831
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.4264776706695557,
      "learning_rate": 4.4744784886663385e-05,
      "loss": 2.0577,
      "step": 189832
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.7442079782485962,
      "learning_rate": 4.474263569828432e-05,
      "loss": 2.2281,
      "step": 189833
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.3201762437820435,
      "learning_rate": 4.4740486557362334e-05,
      "loss": 1.8216,
      "step": 189834
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.3879404067993164,
      "learning_rate": 4.4738337463897976e-05,
      "loss": 1.9164,
      "step": 189835
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.3780092000961304,
      "learning_rate": 4.473618841789159e-05,
      "loss": 1.9591,
      "step": 189836
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.44803786277771,
      "learning_rate": 4.4734039419343483e-05,
      "loss": 2.0954,
      "step": 189837
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.7022244930267334,
      "learning_rate": 4.47318904682542e-05,
      "loss": 1.961,
      "step": 189838
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.7414034605026245,
      "learning_rate": 4.47297415646241e-05,
      "loss": 2.2653,
      "step": 189839
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.3305574655532837,
      "learning_rate": 4.47275927084535e-05,
      "loss": 1.9054,
      "step": 189840
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.3982652425765991,
      "learning_rate": 4.4725443899742894e-05,
      "loss": 1.9857,
      "step": 189841
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.2819898128509521,
      "learning_rate": 4.4723295138492596e-05,
      "loss": 1.9863,
      "step": 189842
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.5977944135665894,
      "learning_rate": 4.47211464247031e-05,
      "loss": 2.1294,
      "step": 189843
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.309069037437439,
      "learning_rate": 4.471899775837481e-05,
      "loss": 1.9334,
      "step": 189844
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.526641607284546,
      "learning_rate": 4.471684913950805e-05,
      "loss": 2.0764,
      "step": 189845
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.3867441415786743,
      "learning_rate": 4.4714700568103155e-05,
      "loss": 2.1694,
      "step": 189846
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.4402551651000977,
      "learning_rate": 4.471255204416073e-05,
      "loss": 1.9767,
      "step": 189847
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.3755948543548584,
      "learning_rate": 4.471040356768097e-05,
      "loss": 2.23,
      "step": 189848
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.3398773670196533,
      "learning_rate": 4.470825513866442e-05,
      "loss": 2.1117,
      "step": 189849
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.3369067907333374,
      "learning_rate": 4.470610675711136e-05,
      "loss": 2.1803,
      "step": 189850
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.3684808015823364,
      "learning_rate": 4.47039584230224e-05,
      "loss": 2.0181,
      "step": 189851
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.7243083715438843,
      "learning_rate": 4.4701810136397675e-05,
      "loss": 1.8815,
      "step": 189852
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.2849931716918945,
      "learning_rate": 4.469966189723775e-05,
      "loss": 2.003,
      "step": 189853
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.355239748954773,
      "learning_rate": 4.469751370554289e-05,
      "loss": 2.0122,
      "step": 189854
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.3408403396606445,
      "learning_rate": 4.469536556131369e-05,
      "loss": 2.0466,
      "step": 189855
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.4455138444900513,
      "learning_rate": 4.4693217464550355e-05,
      "loss": 1.9173,
      "step": 189856
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.483925461769104,
      "learning_rate": 4.469106941525349e-05,
      "loss": 2.1552,
      "step": 189857
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.250697135925293,
      "learning_rate": 4.468892141342325e-05,
      "loss": 2.1036,
      "step": 189858
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.509480357170105,
      "learning_rate": 4.4686773459060246e-05,
      "loss": 2.1143,
      "step": 189859
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.3431739807128906,
      "learning_rate": 4.4684625552164696e-05,
      "loss": 1.9497,
      "step": 189860
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.395451545715332,
      "learning_rate": 4.468247769273718e-05,
      "loss": 1.7962,
      "step": 189861
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.5329375267028809,
      "learning_rate": 4.468032988077793e-05,
      "loss": 1.9799,
      "step": 189862
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.3049129247665405,
      "learning_rate": 4.4678182116287574e-05,
      "loss": 2.069,
      "step": 189863
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.3693768978118896,
      "learning_rate": 4.4676034399266215e-05,
      "loss": 2.1056,
      "step": 189864
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.3488242626190186,
      "learning_rate": 4.467388672971448e-05,
      "loss": 1.9074,
      "step": 189865
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.5868701934814453,
      "learning_rate": 4.467173910763261e-05,
      "loss": 1.9633,
      "step": 189866
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.3597631454467773,
      "learning_rate": 4.466959153302113e-05,
      "loss": 2.0781,
      "step": 189867
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.390094518661499,
      "learning_rate": 4.466744400588038e-05,
      "loss": 2.1399,
      "step": 189868
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.378418207168579,
      "learning_rate": 4.466529652621086e-05,
      "loss": 2.2671,
      "step": 189869
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.2776259183883667,
      "learning_rate": 4.466314909401276e-05,
      "loss": 2.0315,
      "step": 189870
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.389189600944519,
      "learning_rate": 4.4661001709286695e-05,
      "loss": 2.0169,
      "step": 189871
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.734951138496399,
      "learning_rate": 4.4658854372032884e-05,
      "loss": 1.9305,
      "step": 189872
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.390721321105957,
      "learning_rate": 4.465670708225186e-05,
      "loss": 2.0249,
      "step": 189873
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.3251529932022095,
      "learning_rate": 4.4654559839943904e-05,
      "loss": 2.2031,
      "step": 189874
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.2475810050964355,
      "learning_rate": 4.465241264510967e-05,
      "loss": 2.0825,
      "step": 189875
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.3965587615966797,
      "learning_rate": 4.4650265497749194e-05,
      "loss": 2.0335,
      "step": 189876
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.4754908084869385,
      "learning_rate": 4.464811839786314e-05,
      "loss": 2.0725,
      "step": 189877
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.413175106048584,
      "learning_rate": 4.464597134545175e-05,
      "loss": 1.9792,
      "step": 189878
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.3982603549957275,
      "learning_rate": 4.4643824340515546e-05,
      "loss": 2.0154,
      "step": 189879
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.3810521364212036,
      "learning_rate": 4.4641677383054794e-05,
      "loss": 2.1247,
      "step": 189880
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.5325783491134644,
      "learning_rate": 4.463953047307014e-05,
      "loss": 1.9831,
      "step": 189881
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.3710476160049438,
      "learning_rate": 4.4637383610561705e-05,
      "loss": 1.9367,
      "step": 189882
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.4899640083312988,
      "learning_rate": 4.4635236795530025e-05,
      "loss": 2.0932,
      "step": 189883
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.584523320198059,
      "learning_rate": 4.4633090027975435e-05,
      "loss": 2.1879,
      "step": 189884
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.5597983598709106,
      "learning_rate": 4.4630943307898405e-05,
      "loss": 1.9648,
      "step": 189885
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.2991249561309814,
      "learning_rate": 4.462879663529926e-05,
      "loss": 2.0852,
      "step": 189886
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.3766496181488037,
      "learning_rate": 4.46266500101786e-05,
      "loss": 2.3874,
      "step": 189887
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.419189214706421,
      "learning_rate": 4.4624503432536495e-05,
      "loss": 1.9971,
      "step": 189888
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.437194585800171,
      "learning_rate": 4.4622356902373616e-05,
      "loss": 2.0071,
      "step": 189889
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.5374884605407715,
      "learning_rate": 4.462021041969015e-05,
      "loss": 2.008,
      "step": 189890
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.4616191387176514,
      "learning_rate": 4.461806398448671e-05,
      "loss": 1.9401,
      "step": 189891
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.3735930919647217,
      "learning_rate": 4.461591759676352e-05,
      "loss": 2.3219,
      "step": 189892
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.4762943983078003,
      "learning_rate": 4.461377125652118e-05,
      "loss": 1.9183,
      "step": 189893
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.3493958711624146,
      "learning_rate": 4.461162496375983e-05,
      "loss": 1.8784,
      "step": 189894
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.3374568223953247,
      "learning_rate": 4.4609478718480065e-05,
      "loss": 2.1567,
      "step": 189895
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.5519466400146484,
      "learning_rate": 4.460733252068215e-05,
      "loss": 2.0581,
      "step": 189896
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.6220496892929077,
      "learning_rate": 4.460518637036662e-05,
      "loss": 2.1004,
      "step": 189897
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.4442986249923706,
      "learning_rate": 4.460304026753374e-05,
      "loss": 1.9365,
      "step": 189898
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.3191478252410889,
      "learning_rate": 4.460089421218411e-05,
      "loss": 1.7858,
      "step": 189899
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.3460149765014648,
      "learning_rate": 4.459874820431787e-05,
      "loss": 2.0196,
      "step": 189900
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.4169447422027588,
      "learning_rate": 4.459660224393557e-05,
      "loss": 1.8319,
      "step": 189901
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.3492571115493774,
      "learning_rate": 4.4594456331037567e-05,
      "loss": 2.1132,
      "step": 189902
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.3948906660079956,
      "learning_rate": 4.4592310465624335e-05,
      "loss": 2.0961,
      "step": 189903
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.5152199268341064,
      "learning_rate": 4.4590164647696124e-05,
      "loss": 2.0557,
      "step": 189904
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.9461290836334229,
      "learning_rate": 4.4588018877253494e-05,
      "loss": 2.1378,
      "step": 189905
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.2800155878067017,
      "learning_rate": 4.458587315429678e-05,
      "loss": 1.8553,
      "step": 189906
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.3910529613494873,
      "learning_rate": 4.458372747882638e-05,
      "loss": 2.2773,
      "step": 189907
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.4068453311920166,
      "learning_rate": 4.45815818508426e-05,
      "loss": 2.1026,
      "step": 189908
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.3981657028198242,
      "learning_rate": 4.457943627034602e-05,
      "loss": 2.0701,
      "step": 189909
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.3079049587249756,
      "learning_rate": 4.457729073733684e-05,
      "loss": 1.7662,
      "step": 189910
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.3431254625320435,
      "learning_rate": 4.457514525181566e-05,
      "loss": 2.209,
      "step": 189911
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.4914416074752808,
      "learning_rate": 4.457299981378276e-05,
      "loss": 1.9786,
      "step": 189912
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.3456101417541504,
      "learning_rate": 4.457085442323858e-05,
      "loss": 2.0798,
      "step": 189913
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.514891266822815,
      "learning_rate": 4.456870908018344e-05,
      "loss": 2.132,
      "step": 189914
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.402357816696167,
      "learning_rate": 4.456656378461785e-05,
      "loss": 1.9798,
      "step": 189915
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.2543522119522095,
      "learning_rate": 4.45644185365421e-05,
      "loss": 2.0808,
      "step": 189916
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.3546569347381592,
      "learning_rate": 4.45622733359567e-05,
      "loss": 2.0252,
      "step": 189917
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.4535553455352783,
      "learning_rate": 4.456012818286201e-05,
      "loss": 1.7976,
      "step": 189918
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.3916558027267456,
      "learning_rate": 4.455798307725833e-05,
      "loss": 2.1349,
      "step": 189919
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.397986888885498,
      "learning_rate": 4.4555838019146226e-05,
      "loss": 1.8688,
      "step": 189920
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.3758312463760376,
      "learning_rate": 4.4553693008526004e-05,
      "loss": 2.1113,
      "step": 189921
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.3082466125488281,
      "learning_rate": 4.455154804539799e-05,
      "loss": 1.9473,
      "step": 189922
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.402384877204895,
      "learning_rate": 4.454940312976275e-05,
      "loss": 2.005,
      "step": 189923
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.4631670713424683,
      "learning_rate": 4.4547258261620625e-05,
      "loss": 2.1036,
      "step": 189924
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.3843663930892944,
      "learning_rate": 4.454511344097187e-05,
      "loss": 1.9604,
      "step": 189925
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.2915441989898682,
      "learning_rate": 4.454296866781709e-05,
      "loss": 1.9103,
      "step": 189926
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.3811986446380615,
      "learning_rate": 4.4540823942156524e-05,
      "loss": 2.029,
      "step": 189927
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.5502903461456299,
      "learning_rate": 4.453867926399073e-05,
      "loss": 2.0553,
      "step": 189928
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.3547195196151733,
      "learning_rate": 4.453653463332001e-05,
      "loss": 2.2131,
      "step": 189929
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.4839890003204346,
      "learning_rate": 4.453439005014473e-05,
      "loss": 1.9318,
      "step": 189930
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.4037443399429321,
      "learning_rate": 4.453224551446529e-05,
      "loss": 2.0424,
      "step": 189931
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.4958891868591309,
      "learning_rate": 4.4530101026282225e-05,
      "loss": 1.952,
      "step": 189932
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.5157197713851929,
      "learning_rate": 4.452795658559574e-05,
      "loss": 2.1098,
      "step": 189933
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.5279653072357178,
      "learning_rate": 4.4525812192406386e-05,
      "loss": 1.8543,
      "step": 189934
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.4182960987091064,
      "learning_rate": 4.452366784671451e-05,
      "loss": 1.924,
      "step": 189935
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.2986279726028442,
      "learning_rate": 4.452152354852053e-05,
      "loss": 1.9407,
      "step": 189936
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.469632863998413,
      "learning_rate": 4.451937929782473e-05,
      "loss": 2.055,
      "step": 189937
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.3125090599060059,
      "learning_rate": 4.4517235094627665e-05,
      "loss": 2.3124,
      "step": 189938
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.4571670293807983,
      "learning_rate": 4.451509093892964e-05,
      "loss": 2.1344,
      "step": 189939
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.383386492729187,
      "learning_rate": 4.4512946830731125e-05,
      "loss": 2.231,
      "step": 189940
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.4257460832595825,
      "learning_rate": 4.451080277003237e-05,
      "loss": 2.1331,
      "step": 189941
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.3711446523666382,
      "learning_rate": 4.450865875683406e-05,
      "loss": 1.9849,
      "step": 189942
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.4581828117370605,
      "learning_rate": 4.4506514791136284e-05,
      "loss": 2.0404,
      "step": 189943
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.6049760580062866,
      "learning_rate": 4.450437087293961e-05,
      "loss": 2.2396,
      "step": 189944
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.3304598331451416,
      "learning_rate": 4.450222700224434e-05,
      "loss": 1.9842,
      "step": 189945
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.282301664352417,
      "learning_rate": 4.4500083179051006e-05,
      "loss": 2.2864,
      "step": 189946
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.3534117937088013,
      "learning_rate": 4.449793940335984e-05,
      "loss": 1.8276,
      "step": 189947
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.369689702987671,
      "learning_rate": 4.4495795675171475e-05,
      "loss": 2.0624,
      "step": 189948
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.4064754247665405,
      "learning_rate": 4.4493651994486045e-05,
      "loss": 2.0672,
      "step": 189949
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.5330687761306763,
      "learning_rate": 4.449150836130412e-05,
      "loss": 1.8932,
      "step": 189950
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.669234275817871,
      "learning_rate": 4.448936477562599e-05,
      "loss": 2.1858,
      "step": 189951
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.490908145904541,
      "learning_rate": 4.448722123745216e-05,
      "loss": 1.9672,
      "step": 189952
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.3701207637786865,
      "learning_rate": 4.4485077746782934e-05,
      "loss": 1.7197,
      "step": 189953
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.280758023262024,
      "learning_rate": 4.448293430361888e-05,
      "loss": 2.0329,
      "step": 189954
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.431624412536621,
      "learning_rate": 4.448079090796012e-05,
      "loss": 2.0352,
      "step": 189955
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.5160523653030396,
      "learning_rate": 4.447864755980729e-05,
      "loss": 2.0395,
      "step": 189956
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.4234832525253296,
      "learning_rate": 4.447650425916063e-05,
      "loss": 2.042,
      "step": 189957
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.6478418111801147,
      "learning_rate": 4.4474361006020665e-05,
      "loss": 1.9735,
      "step": 189958
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.8072952032089233,
      "learning_rate": 4.44722178003877e-05,
      "loss": 2.0356,
      "step": 189959
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.2688401937484741,
      "learning_rate": 4.44700746422623e-05,
      "loss": 1.7684,
      "step": 189960
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.185280442237854,
      "learning_rate": 4.446793153164459e-05,
      "loss": 1.6584,
      "step": 189961
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.4471906423568726,
      "learning_rate": 4.446578846853519e-05,
      "loss": 2.0394,
      "step": 189962
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.2571310997009277,
      "learning_rate": 4.446364545293435e-05,
      "loss": 1.893,
      "step": 189963
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.3390636444091797,
      "learning_rate": 4.4461502484842635e-05,
      "loss": 2.0127,
      "step": 189964
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.284416675567627,
      "learning_rate": 4.445935956426025e-05,
      "loss": 1.8585,
      "step": 189965
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.4118239879608154,
      "learning_rate": 4.445721669118784e-05,
      "loss": 1.9489,
      "step": 189966
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.3685742616653442,
      "learning_rate": 4.445507386562551e-05,
      "loss": 1.7816,
      "step": 189967
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.5721815824508667,
      "learning_rate": 4.445293108757392e-05,
      "loss": 1.8824,
      "step": 189968
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.477198600769043,
      "learning_rate": 4.445078835703325e-05,
      "loss": 1.9858,
      "step": 189969
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.396909236907959,
      "learning_rate": 4.444864567400405e-05,
      "loss": 1.9629,
      "step": 189970
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.4637997150421143,
      "learning_rate": 4.444650303848663e-05,
      "loss": 2.1334,
      "step": 189971
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.3990023136138916,
      "learning_rate": 4.4444360450481484e-05,
      "loss": 1.9445,
      "step": 189972
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.399438500404358,
      "learning_rate": 4.444221790998896e-05,
      "loss": 2.0196,
      "step": 189973
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.3084994554519653,
      "learning_rate": 4.444007541700947e-05,
      "loss": 2.1427,
      "step": 189974
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.4930822849273682,
      "learning_rate": 4.443793297154333e-05,
      "loss": 2.0051,
      "step": 189975
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.295905351638794,
      "learning_rate": 4.4435790573591025e-05,
      "loss": 2.0375,
      "step": 189976
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.3182721138000488,
      "learning_rate": 4.443364822315291e-05,
      "loss": 1.8446,
      "step": 189977
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.3173457384109497,
      "learning_rate": 4.443150592022945e-05,
      "loss": 1.9218,
      "step": 189978
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.385789155960083,
      "learning_rate": 4.4429363664820986e-05,
      "loss": 1.9269,
      "step": 189979
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.4762946367263794,
      "learning_rate": 4.4427221456927944e-05,
      "loss": 1.9343,
      "step": 189980
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.388419508934021,
      "learning_rate": 4.4425079296550624e-05,
      "loss": 2.0303,
      "step": 189981
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.3426202535629272,
      "learning_rate": 4.44229371836896e-05,
      "loss": 2.1064,
      "step": 189982
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.3256065845489502,
      "learning_rate": 4.44207951183451e-05,
      "loss": 1.9244,
      "step": 189983
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.462786316871643,
      "learning_rate": 4.441865310051765e-05,
      "loss": 1.9841,
      "step": 189984
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.4060426950454712,
      "learning_rate": 4.44165111302076e-05,
      "loss": 2.0854,
      "step": 189985
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.4286857843399048,
      "learning_rate": 4.4414369207415365e-05,
      "loss": 2.1167,
      "step": 189986
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.5873204469680786,
      "learning_rate": 4.4412227332141224e-05,
      "loss": 2.1477,
      "step": 189987
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.4284311532974243,
      "learning_rate": 4.4410085504385775e-05,
      "loss": 2.0553,
      "step": 189988
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.3546839952468872,
      "learning_rate": 4.440794372414924e-05,
      "loss": 1.967,
      "step": 189989
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.344102382659912,
      "learning_rate": 4.440580199143217e-05,
      "loss": 2.0265,
      "step": 189990
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.4804606437683105,
      "learning_rate": 4.440366030623484e-05,
      "loss": 1.7816,
      "step": 189991
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.4040968418121338,
      "learning_rate": 4.4401518668557746e-05,
      "loss": 1.9018,
      "step": 189992
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.363335132598877,
      "learning_rate": 4.439937707840113e-05,
      "loss": 1.961,
      "step": 189993
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.4346755743026733,
      "learning_rate": 4.4397235535765576e-05,
      "loss": 1.8846,
      "step": 189994
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.4151265621185303,
      "learning_rate": 4.4395094040651335e-05,
      "loss": 2.1537,
      "step": 189995
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.2916806936264038,
      "learning_rate": 4.439295259305895e-05,
      "loss": 2.046,
      "step": 189996
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.3261806964874268,
      "learning_rate": 4.439081119298872e-05,
      "loss": 2.1349,
      "step": 189997
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.364450216293335,
      "learning_rate": 4.438866984044107e-05,
      "loss": 2.0558,
      "step": 189998
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.347161889076233,
      "learning_rate": 4.4386528535416346e-05,
      "loss": 2.0985,
      "step": 189999
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.3677457571029663,
      "learning_rate": 4.438438727791503e-05,
      "loss": 1.9785,
      "step": 190000
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.424075961112976,
      "learning_rate": 4.438224606793741e-05,
      "loss": 2.039,
      "step": 190001
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.3328009843826294,
      "learning_rate": 4.438010490548404e-05,
      "loss": 1.9233,
      "step": 190002
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.3483022451400757,
      "learning_rate": 4.4377963790555216e-05,
      "loss": 1.8658,
      "step": 190003
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.4332904815673828,
      "learning_rate": 4.437582272315131e-05,
      "loss": 1.9037,
      "step": 190004
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.4296894073486328,
      "learning_rate": 4.437368170327282e-05,
      "loss": 1.8636,
      "step": 190005
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.399482011795044,
      "learning_rate": 4.437154073092009e-05,
      "loss": 2.1317,
      "step": 190006
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.3264570236206055,
      "learning_rate": 4.4369399806093466e-05,
      "loss": 2.0572,
      "step": 190007
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.3595675230026245,
      "learning_rate": 4.436725892879342e-05,
      "loss": 1.9666,
      "step": 190008
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.419670820236206,
      "learning_rate": 4.436511809902037e-05,
      "loss": 2.0093,
      "step": 190009
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.2939144372940063,
      "learning_rate": 4.436297731677456e-05,
      "loss": 1.8559,
      "step": 190010
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.3800160884857178,
      "learning_rate": 4.43608365820566e-05,
      "loss": 2.0392,
      "step": 190011
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.4073736667633057,
      "learning_rate": 4.435869589486669e-05,
      "loss": 2.0435,
      "step": 190012
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.4769240617752075,
      "learning_rate": 4.435655525520543e-05,
      "loss": 1.8589,
      "step": 190013
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.2966197729110718,
      "learning_rate": 4.435441466307308e-05,
      "loss": 2.0491,
      "step": 190014
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.4123625755310059,
      "learning_rate": 4.4352274118470075e-05,
      "loss": 1.9339,
      "step": 190015
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.2984977960586548,
      "learning_rate": 4.435013362139676e-05,
      "loss": 1.8933,
      "step": 190016
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.4363449811935425,
      "learning_rate": 4.4347993171853644e-05,
      "loss": 1.7484,
      "step": 190017
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.3764920234680176,
      "learning_rate": 4.4345852769840985e-05,
      "loss": 2.0892,
      "step": 190018
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.3619574308395386,
      "learning_rate": 4.43437124153593e-05,
      "loss": 2.1811,
      "step": 190019
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.277380108833313,
      "learning_rate": 4.4341572108408995e-05,
      "loss": 2.0047,
      "step": 190020
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.5877211093902588,
      "learning_rate": 4.433943184899037e-05,
      "loss": 1.9747,
      "step": 190021
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.436474084854126,
      "learning_rate": 4.433729163710382e-05,
      "loss": 1.8362,
      "step": 190022
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.454936146736145,
      "learning_rate": 4.4335151472749886e-05,
      "loss": 2.1352,
      "step": 190023
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.3478292226791382,
      "learning_rate": 4.4333011355928795e-05,
      "loss": 1.9647,
      "step": 190024
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.4544309377670288,
      "learning_rate": 4.433087128664108e-05,
      "loss": 2.1105,
      "step": 190025
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.4469231367111206,
      "learning_rate": 4.432873126488701e-05,
      "loss": 2.1732,
      "step": 190026
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.421847939491272,
      "learning_rate": 4.4326591290667215e-05,
      "loss": 1.7557,
      "step": 190027
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.5525919198989868,
      "learning_rate": 4.432445136398176e-05,
      "loss": 2.0668,
      "step": 190028
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.4877711534500122,
      "learning_rate": 4.4322311484831316e-05,
      "loss": 1.9267,
      "step": 190029
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.4659568071365356,
      "learning_rate": 4.4320171653216085e-05,
      "loss": 2.0271,
      "step": 190030
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.3749675750732422,
      "learning_rate": 4.431803186913666e-05,
      "loss": 1.9544,
      "step": 190031
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.4336639642715454,
      "learning_rate": 4.431589213259324e-05,
      "loss": 2.1726,
      "step": 190032
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.572360634803772,
      "learning_rate": 4.431375244358647e-05,
      "loss": 1.9207,
      "step": 190033
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.3846800327301025,
      "learning_rate": 4.431161280211647e-05,
      "loss": 1.9802,
      "step": 190034
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.421968936920166,
      "learning_rate": 4.4309473208183845e-05,
      "loss": 2.085,
      "step": 190035
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.3586136102676392,
      "learning_rate": 4.4307333661788826e-05,
      "loss": 1.8607,
      "step": 190036
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.34950852394104,
      "learning_rate": 4.430519416293198e-05,
      "loss": 2.0026,
      "step": 190037
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.4690539836883545,
      "learning_rate": 4.430305471161354e-05,
      "loss": 2.158,
      "step": 190038
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.3315385580062866,
      "learning_rate": 4.4300915307834076e-05,
      "loss": 1.8601,
      "step": 190039
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.4823787212371826,
      "learning_rate": 4.4298775951593915e-05,
      "loss": 1.9357,
      "step": 190040
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.3049261569976807,
      "learning_rate": 4.429663664289339e-05,
      "loss": 1.8056,
      "step": 190041
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.3289320468902588,
      "learning_rate": 4.429449738173291e-05,
      "loss": 1.8215,
      "step": 190042
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.4847168922424316,
      "learning_rate": 4.4292358168112995e-05,
      "loss": 2.1861,
      "step": 190043
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.6358805894851685,
      "learning_rate": 4.429021900203386e-05,
      "loss": 2.0222,
      "step": 190044
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.3439314365386963,
      "learning_rate": 4.428807988349605e-05,
      "loss": 1.7455,
      "step": 190045
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.309755802154541,
      "learning_rate": 4.4285940812499954e-05,
      "loss": 1.9674,
      "step": 190046
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.3047865629196167,
      "learning_rate": 4.4283801789045926e-05,
      "loss": 1.8439,
      "step": 190047
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.666646957397461,
      "learning_rate": 4.4281662813134264e-05,
      "loss": 1.8761,
      "step": 190048
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.51290762424469,
      "learning_rate": 4.427952388476558e-05,
      "loss": 1.9472,
      "step": 190049
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.4285130500793457,
      "learning_rate": 4.4277385003940055e-05,
      "loss": 1.9404,
      "step": 190050
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.4483044147491455,
      "learning_rate": 4.427524617065827e-05,
      "loss": 1.9579,
      "step": 190051
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.5349868535995483,
      "learning_rate": 4.427310738492056e-05,
      "loss": 1.8839,
      "step": 190052
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.2921556234359741,
      "learning_rate": 4.4270968646727276e-05,
      "loss": 2.0575,
      "step": 190053
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.4402945041656494,
      "learning_rate": 4.42688299560788e-05,
      "loss": 2.1014,
      "step": 190054
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.3191428184509277,
      "learning_rate": 4.4266691312975655e-05,
      "loss": 1.9166,
      "step": 190055
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.5863667726516724,
      "learning_rate": 4.426455271741808e-05,
      "loss": 1.9778,
      "step": 190056
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.552115797996521,
      "learning_rate": 4.42624141694066e-05,
      "loss": 1.9682,
      "step": 190057
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.4698413610458374,
      "learning_rate": 4.426027566894159e-05,
      "loss": 1.9923,
      "step": 190058
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.4319055080413818,
      "learning_rate": 4.425813721602342e-05,
      "loss": 2.1822,
      "step": 190059
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.323532223701477,
      "learning_rate": 4.425599881065241e-05,
      "loss": 1.8345,
      "step": 190060
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.2343475818634033,
      "learning_rate": 4.425386045282912e-05,
      "loss": 2.092,
      "step": 190061
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.5804297924041748,
      "learning_rate": 4.4251722142553784e-05,
      "loss": 2.0128,
      "step": 190062
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.2912254333496094,
      "learning_rate": 4.424958387982693e-05,
      "loss": 1.9329,
      "step": 190063
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.3594492673873901,
      "learning_rate": 4.424744566464895e-05,
      "loss": 1.973,
      "step": 190064
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.4046211242675781,
      "learning_rate": 4.4245307497020164e-05,
      "loss": 1.9628,
      "step": 190065
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.3889585733413696,
      "learning_rate": 4.424316937694094e-05,
      "loss": 1.9313,
      "step": 190066
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.3397297859191895,
      "learning_rate": 4.424103130441182e-05,
      "loss": 2.1246,
      "step": 190067
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.3590171337127686,
      "learning_rate": 4.423889327943306e-05,
      "loss": 2.0008,
      "step": 190068
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.3437798023223877,
      "learning_rate": 4.4236755302005166e-05,
      "loss": 1.8683,
      "step": 190069
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.3201961517333984,
      "learning_rate": 4.42346173721285e-05,
      "loss": 1.8509,
      "step": 190070
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.4139692783355713,
      "learning_rate": 4.423247948980343e-05,
      "loss": 2.1049,
      "step": 190071
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.4778919219970703,
      "learning_rate": 4.423034165503032e-05,
      "loss": 1.9049,
      "step": 190072
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.3968992233276367,
      "learning_rate": 4.4228203867809675e-05,
      "loss": 2.0685,
      "step": 190073
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.4661922454833984,
      "learning_rate": 4.422606612814176e-05,
      "loss": 2.04,
      "step": 190074
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.4511996507644653,
      "learning_rate": 4.4223928436027135e-05,
      "loss": 1.8377,
      "step": 190075
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.5218263864517212,
      "learning_rate": 4.422179079146607e-05,
      "loss": 1.7913,
      "step": 190076
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.3983867168426514,
      "learning_rate": 4.4219653194459037e-05,
      "loss": 2.0623,
      "step": 190077
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.3625750541687012,
      "learning_rate": 4.421751564500633e-05,
      "loss": 1.8745,
      "step": 190078
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.4104986190795898,
      "learning_rate": 4.421537814310848e-05,
      "loss": 1.9694,
      "step": 190079
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.2840710878372192,
      "learning_rate": 4.4213240688765764e-05,
      "loss": 1.8635,
      "step": 190080
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.4924569129943848,
      "learning_rate": 4.4211103281978667e-05,
      "loss": 1.8326,
      "step": 190081
    },
    {
      "epoch": 2.47,
      "grad_norm": 1.37246835231781,
      "learning_rate": 4.420896592274756e-05,
      "loss": 2.302,
      "step": 190082
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.4459228515625,
      "learning_rate": 4.420682861107289e-05,
      "loss": 2.1722,
      "step": 190083
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.5089167356491089,
      "learning_rate": 4.420469134695487e-05,
      "loss": 2.0414,
      "step": 190084
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.3556979894638062,
      "learning_rate": 4.4202554130394105e-05,
      "loss": 1.8884,
      "step": 190085
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.4242780208587646,
      "learning_rate": 4.4200416961390874e-05,
      "loss": 2.0044,
      "step": 190086
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.3891167640686035,
      "learning_rate": 4.419827983994566e-05,
      "loss": 1.7906,
      "step": 190087
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.5842580795288086,
      "learning_rate": 4.41961427660588e-05,
      "loss": 2.0278,
      "step": 190088
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.3463008403778076,
      "learning_rate": 4.4194005739730676e-05,
      "loss": 1.9754,
      "step": 190089
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.2951624393463135,
      "learning_rate": 4.419186876096177e-05,
      "loss": 1.7224,
      "step": 190090
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.4523078203201294,
      "learning_rate": 4.418973182975242e-05,
      "loss": 1.9374,
      "step": 190091
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.3912993669509888,
      "learning_rate": 4.418759494610296e-05,
      "loss": 2.1703,
      "step": 190092
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.626387119293213,
      "learning_rate": 4.418545811001392e-05,
      "loss": 1.8175,
      "step": 190093
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.9509049654006958,
      "learning_rate": 4.4183321321485634e-05,
      "loss": 2.0719,
      "step": 190094
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.3748494386672974,
      "learning_rate": 4.41811845805184e-05,
      "loss": 2.0993,
      "step": 190095
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.5276092290878296,
      "learning_rate": 4.4179047887112825e-05,
      "loss": 1.857,
      "step": 190096
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.2845479249954224,
      "learning_rate": 4.4176911241269174e-05,
      "loss": 1.8529,
      "step": 190097
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.3835517168045044,
      "learning_rate": 4.417477464298781e-05,
      "loss": 1.9239,
      "step": 190098
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.349604606628418,
      "learning_rate": 4.417263809226923e-05,
      "loss": 2.0649,
      "step": 190099
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.4748071432113647,
      "learning_rate": 4.41705015891138e-05,
      "loss": 1.9476,
      "step": 190100
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.3287718296051025,
      "learning_rate": 4.4168365133521835e-05,
      "loss": 1.9236,
      "step": 190101
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.495603084564209,
      "learning_rate": 4.4166228725493844e-05,
      "loss": 1.9646,
      "step": 190102
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.4300884008407593,
      "learning_rate": 4.416409236503015e-05,
      "loss": 1.8813,
      "step": 190103
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.4016644954681396,
      "learning_rate": 4.4161956052131234e-05,
      "loss": 1.8797,
      "step": 190104
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.4211212396621704,
      "learning_rate": 4.4159819786797435e-05,
      "loss": 1.9694,
      "step": 190105
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.507775902748108,
      "learning_rate": 4.415768356902909e-05,
      "loss": 2.0037,
      "step": 190106
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.4194809198379517,
      "learning_rate": 4.4155547398826727e-05,
      "loss": 1.9687,
      "step": 190107
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.354401707649231,
      "learning_rate": 4.4153411276190685e-05,
      "loss": 1.8664,
      "step": 190108
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.3717315196990967,
      "learning_rate": 4.4151275201121296e-05,
      "loss": 1.9649,
      "step": 190109
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.3971545696258545,
      "learning_rate": 4.4149139173619054e-05,
      "loss": 1.9978,
      "step": 190110
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.310736894607544,
      "learning_rate": 4.414700319368426e-05,
      "loss": 2.0779,
      "step": 190111
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.3574765920639038,
      "learning_rate": 4.414486726131742e-05,
      "loss": 1.9942,
      "step": 190112
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.35597825050354,
      "learning_rate": 4.414273137651892e-05,
      "loss": 1.7407,
      "step": 190113
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.4111320972442627,
      "learning_rate": 4.414059553928908e-05,
      "loss": 1.9115,
      "step": 190114
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.4158607721328735,
      "learning_rate": 4.4138459749628284e-05,
      "loss": 2.0079,
      "step": 190115
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.469430685043335,
      "learning_rate": 4.413632400753704e-05,
      "loss": 1.8507,
      "step": 190116
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.3104248046875,
      "learning_rate": 4.413418831301561e-05,
      "loss": 2.0003,
      "step": 190117
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.5567858219146729,
      "learning_rate": 4.413205266606452e-05,
      "loss": 2.036,
      "step": 190118
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.3911104202270508,
      "learning_rate": 4.412991706668415e-05,
      "loss": 2.0306,
      "step": 190119
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.3647278547286987,
      "learning_rate": 4.412778151487483e-05,
      "loss": 1.9106,
      "step": 190120
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.4042342901229858,
      "learning_rate": 4.412564601063692e-05,
      "loss": 2.0764,
      "step": 190121
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.5717296600341797,
      "learning_rate": 4.412351055397095e-05,
      "loss": 1.9717,
      "step": 190122
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.3264294862747192,
      "learning_rate": 4.412137514487717e-05,
      "loss": 2.1244,
      "step": 190123
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.36184823513031,
      "learning_rate": 4.4119239783356164e-05,
      "loss": 2.1034,
      "step": 190124
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.4227732419967651,
      "learning_rate": 4.4117104469408206e-05,
      "loss": 1.8278,
      "step": 190125
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.4693342447280884,
      "learning_rate": 4.411496920303369e-05,
      "loss": 1.9782,
      "step": 190126
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.3509241342544556,
      "learning_rate": 4.411283398423295e-05,
      "loss": 2.0744,
      "step": 190127
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.3041908740997314,
      "learning_rate": 4.411069881300656e-05,
      "loss": 1.8467,
      "step": 190128
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.3952736854553223,
      "learning_rate": 4.410856368935475e-05,
      "loss": 2.0311,
      "step": 190129
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.4161347150802612,
      "learning_rate": 4.410642861327804e-05,
      "loss": 1.9637,
      "step": 190130
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.4930016994476318,
      "learning_rate": 4.410429358477682e-05,
      "loss": 2.0578,
      "step": 190131
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.3477752208709717,
      "learning_rate": 4.41021586038514e-05,
      "loss": 1.8716,
      "step": 190132
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.674796462059021,
      "learning_rate": 4.410002367050216e-05,
      "loss": 1.8844,
      "step": 190133
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.4286963939666748,
      "learning_rate": 4.409788878472963e-05,
      "loss": 2.0156,
      "step": 190134
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.4481968879699707,
      "learning_rate": 4.4095753946534074e-05,
      "loss": 1.9787,
      "step": 190135
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.36878502368927,
      "learning_rate": 4.4093619155916e-05,
      "loss": 2.0011,
      "step": 190136
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.3185501098632812,
      "learning_rate": 4.4091484412875754e-05,
      "loss": 2.0645,
      "step": 190137
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.3882451057434082,
      "learning_rate": 4.4089349717413724e-05,
      "loss": 2.0764,
      "step": 190138
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.363053321838379,
      "learning_rate": 4.4087215069530266e-05,
      "loss": 2.0963,
      "step": 190139
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.3827518224716187,
      "learning_rate": 4.408508046922589e-05,
      "loss": 2.0882,
      "step": 190140
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.278488278388977,
      "learning_rate": 4.4082945916500844e-05,
      "loss": 2.0115,
      "step": 190141
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.5552096366882324,
      "learning_rate": 4.408081141135571e-05,
      "loss": 1.9499,
      "step": 190142
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.6088969707489014,
      "learning_rate": 4.407867695379075e-05,
      "loss": 2.0654,
      "step": 190143
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.3897920846939087,
      "learning_rate": 4.407654254380639e-05,
      "loss": 2.0994,
      "step": 190144
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.3114818334579468,
      "learning_rate": 4.407440818140301e-05,
      "loss": 1.7316,
      "step": 190145
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.3285778760910034,
      "learning_rate": 4.4072273866581066e-05,
      "loss": 2.0704,
      "step": 190146
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.4344346523284912,
      "learning_rate": 4.4070139599340825e-05,
      "loss": 2.1084,
      "step": 190147
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.6126943826675415,
      "learning_rate": 4.406800537968289e-05,
      "loss": 2.0493,
      "step": 190148
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.5692696571350098,
      "learning_rate": 4.4065871207607526e-05,
      "loss": 2.234,
      "step": 190149
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.4166021347045898,
      "learning_rate": 4.406373708311514e-05,
      "loss": 1.9662,
      "step": 190150
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.4250613451004028,
      "learning_rate": 4.406160300620608e-05,
      "loss": 1.9686,
      "step": 190151
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.3032755851745605,
      "learning_rate": 4.405946897688086e-05,
      "loss": 1.9755,
      "step": 190152
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.2672019004821777,
      "learning_rate": 4.405733499513975e-05,
      "loss": 1.8214,
      "step": 190153
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.545480728149414,
      "learning_rate": 4.405520106098327e-05,
      "loss": 2.0642,
      "step": 190154
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.3292820453643799,
      "learning_rate": 4.405306717441176e-05,
      "loss": 1.8336,
      "step": 190155
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.259660005569458,
      "learning_rate": 4.405093333542562e-05,
      "loss": 2.0743,
      "step": 190156
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.5466002225875854,
      "learning_rate": 4.404879954402518e-05,
      "loss": 2.0049,
      "step": 190157
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.3613581657409668,
      "learning_rate": 4.4046665800210976e-05,
      "loss": 2.0952,
      "step": 190158
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.3288993835449219,
      "learning_rate": 4.4044532103983244e-05,
      "loss": 1.8744,
      "step": 190159
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.437793493270874,
      "learning_rate": 4.404239845534254e-05,
      "loss": 2.0092,
      "step": 190160
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.5476795434951782,
      "learning_rate": 4.404026485428921e-05,
      "loss": 2.0544,
      "step": 190161
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.4113037586212158,
      "learning_rate": 4.403813130082361e-05,
      "loss": 1.9204,
      "step": 190162
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.3103594779968262,
      "learning_rate": 4.4035997794946074e-05,
      "loss": 1.9002,
      "step": 190163
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.4925546646118164,
      "learning_rate": 4.403386433665714e-05,
      "loss": 2.1813,
      "step": 190164
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.2589203119277954,
      "learning_rate": 4.4031730925957074e-05,
      "loss": 1.6692,
      "step": 190165
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.2442967891693115,
      "learning_rate": 4.402959756284644e-05,
      "loss": 1.8912,
      "step": 190166
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.4207855463027954,
      "learning_rate": 4.40274642473255e-05,
      "loss": 1.961,
      "step": 190167
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.3158987760543823,
      "learning_rate": 4.40253309793947e-05,
      "loss": 2.028,
      "step": 190168
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.2460312843322754,
      "learning_rate": 4.402319775905436e-05,
      "loss": 2.0487,
      "step": 190169
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.3096835613250732,
      "learning_rate": 4.402106458630499e-05,
      "loss": 2.0347,
      "step": 190170
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.3222205638885498,
      "learning_rate": 4.401893146114688e-05,
      "loss": 1.9764,
      "step": 190171
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.4125785827636719,
      "learning_rate": 4.401679838358057e-05,
      "loss": 1.8385,
      "step": 190172
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.3450994491577148,
      "learning_rate": 4.401466535360628e-05,
      "loss": 1.9202,
      "step": 190173
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.2623698711395264,
      "learning_rate": 4.401253237122456e-05,
      "loss": 1.8904,
      "step": 190174
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.3716155290603638,
      "learning_rate": 4.401039943643577e-05,
      "loss": 2.1534,
      "step": 190175
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.3170629739761353,
      "learning_rate": 4.400826654924024e-05,
      "loss": 1.9518,
      "step": 190176
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.3404709100723267,
      "learning_rate": 4.4006133709638344e-05,
      "loss": 1.8125,
      "step": 190177
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.567949891090393,
      "learning_rate": 4.4004000917630644e-05,
      "loss": 2.1028,
      "step": 190178
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.369425892829895,
      "learning_rate": 4.400186817321734e-05,
      "loss": 1.8698,
      "step": 190179
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.4276880025863647,
      "learning_rate": 4.399973547639899e-05,
      "loss": 2.1417,
      "step": 190180
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.3607782125473022,
      "learning_rate": 4.3997602827175906e-05,
      "loss": 2.1914,
      "step": 190181
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.421808123588562,
      "learning_rate": 4.399547022554851e-05,
      "loss": 1.9894,
      "step": 190182
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.3631757497787476,
      "learning_rate": 4.399333767151715e-05,
      "loss": 1.8774,
      "step": 190183
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.5462698936462402,
      "learning_rate": 4.399120516508231e-05,
      "loss": 1.9894,
      "step": 190184
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.3689426183700562,
      "learning_rate": 4.398907270624423e-05,
      "loss": 2.1689,
      "step": 190185
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.4200648069381714,
      "learning_rate": 4.398694029500355e-05,
      "loss": 2.1467,
      "step": 190186
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.4698073863983154,
      "learning_rate": 4.398480793136049e-05,
      "loss": 1.993,
      "step": 190187
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.4764573574066162,
      "learning_rate": 4.398267561531542e-05,
      "loss": 1.9933,
      "step": 190188
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.24297034740448,
      "learning_rate": 4.398054334686888e-05,
      "loss": 2.0497,
      "step": 190189
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.6450506448745728,
      "learning_rate": 4.39784111260212e-05,
      "loss": 2.0315,
      "step": 190190
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.5322924852371216,
      "learning_rate": 4.397627895277267e-05,
      "loss": 1.9975,
      "step": 190191
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.4885014295578003,
      "learning_rate": 4.397414682712387e-05,
      "loss": 1.8851,
      "step": 190192
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.419914960861206,
      "learning_rate": 4.397201474907509e-05,
      "loss": 1.926,
      "step": 190193
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.4395148754119873,
      "learning_rate": 4.39698827186267e-05,
      "loss": 2.0096,
      "step": 190194
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.3625130653381348,
      "learning_rate": 4.396775073577921e-05,
      "loss": 2.0626,
      "step": 190195
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.4111117124557495,
      "learning_rate": 4.396561880053293e-05,
      "loss": 1.877,
      "step": 190196
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.3840724229812622,
      "learning_rate": 4.3963486912888216e-05,
      "loss": 1.7394,
      "step": 190197
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.2845373153686523,
      "learning_rate": 4.3961355072845585e-05,
      "loss": 2.1471,
      "step": 190198
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.505444049835205,
      "learning_rate": 4.395922328040538e-05,
      "loss": 2.069,
      "step": 190199
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.4674965143203735,
      "learning_rate": 4.395709153556792e-05,
      "loss": 2.0394,
      "step": 190200
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.4541521072387695,
      "learning_rate": 4.3954959838333725e-05,
      "loss": 2.1838,
      "step": 190201
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.4009242057800293,
      "learning_rate": 4.3952828188703085e-05,
      "loss": 2.0459,
      "step": 190202
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.423004388809204,
      "learning_rate": 4.395069658667653e-05,
      "loss": 1.9646,
      "step": 190203
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.3922064304351807,
      "learning_rate": 4.394856503225432e-05,
      "loss": 2.1431,
      "step": 190204
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.259946346282959,
      "learning_rate": 4.394643352543694e-05,
      "loss": 2.2033,
      "step": 190205
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.3646780252456665,
      "learning_rate": 4.394430206622468e-05,
      "loss": 2.0187,
      "step": 190206
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.3923784494400024,
      "learning_rate": 4.39421706546181e-05,
      "loss": 2.0609,
      "step": 190207
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.7538312673568726,
      "learning_rate": 4.3940039290617414e-05,
      "loss": 1.815,
      "step": 190208
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.2727419137954712,
      "learning_rate": 4.3937907974223175e-05,
      "loss": 2.2816,
      "step": 190209
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.364916443824768,
      "learning_rate": 4.393577670543569e-05,
      "loss": 2.367,
      "step": 190210
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.3300496339797974,
      "learning_rate": 4.393364548425542e-05,
      "loss": 1.9952,
      "step": 190211
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.4435663223266602,
      "learning_rate": 4.3931514310682635e-05,
      "loss": 2.0723,
      "step": 190212
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.449113130569458,
      "learning_rate": 4.392938318471787e-05,
      "loss": 2.066,
      "step": 190213
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.5056195259094238,
      "learning_rate": 4.392725210636142e-05,
      "loss": 2.1047,
      "step": 190214
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.3167506456375122,
      "learning_rate": 4.392512107561376e-05,
      "loss": 1.8822,
      "step": 190215
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.2713128328323364,
      "learning_rate": 4.3922990092475275e-05,
      "loss": 2.0278,
      "step": 190216
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.3404685258865356,
      "learning_rate": 4.3920859156946345e-05,
      "loss": 1.9727,
      "step": 190217
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.396966576576233,
      "learning_rate": 4.391872826902729e-05,
      "loss": 2.1284,
      "step": 190218
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.5459059476852417,
      "learning_rate": 4.3916597428718656e-05,
      "loss": 2.128,
      "step": 190219
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.4140809774398804,
      "learning_rate": 4.391446663602067e-05,
      "loss": 2.134,
      "step": 190220
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.4587743282318115,
      "learning_rate": 4.3912335890933905e-05,
      "loss": 2.0945,
      "step": 190221
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.3283779621124268,
      "learning_rate": 4.391020519345868e-05,
      "loss": 2.1795,
      "step": 190222
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.423274040222168,
      "learning_rate": 4.3908074543595337e-05,
      "loss": 1.93,
      "step": 190223
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.3063031435012817,
      "learning_rate": 4.390594394134431e-05,
      "loss": 1.8116,
      "step": 190224
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.4828819036483765,
      "learning_rate": 4.3903813386706035e-05,
      "loss": 1.9959,
      "step": 190225
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.464805006980896,
      "learning_rate": 4.39016828796808e-05,
      "loss": 2.0079,
      "step": 190226
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.3319059610366821,
      "learning_rate": 4.3899552420269143e-05,
      "loss": 2.0259,
      "step": 190227
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.3330341577529907,
      "learning_rate": 4.3897422008471405e-05,
      "loss": 1.9553,
      "step": 190228
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.4231764078140259,
      "learning_rate": 4.3895291644287976e-05,
      "loss": 1.8377,
      "step": 190229
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.410093069076538,
      "learning_rate": 4.38931613277192e-05,
      "loss": 2.1003,
      "step": 190230
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.3628607988357544,
      "learning_rate": 4.3891031058765555e-05,
      "loss": 2.1863,
      "step": 190231
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.5266540050506592,
      "learning_rate": 4.3888900837427335e-05,
      "loss": 1.9778,
      "step": 190232
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.6874823570251465,
      "learning_rate": 4.3886770663705086e-05,
      "loss": 2.0985,
      "step": 190233
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.4955641031265259,
      "learning_rate": 4.388464053759911e-05,
      "loss": 2.0837,
      "step": 190234
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.346253752708435,
      "learning_rate": 4.388251045910979e-05,
      "loss": 1.8918,
      "step": 190235
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.5146353244781494,
      "learning_rate": 4.38803804282375e-05,
      "loss": 1.9979,
      "step": 190236
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.4659132957458496,
      "learning_rate": 4.387825044498277e-05,
      "loss": 2.114,
      "step": 190237
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.2961333990097046,
      "learning_rate": 4.3876120509345834e-05,
      "loss": 2.0953,
      "step": 190238
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.82188081741333,
      "learning_rate": 4.3873990621327215e-05,
      "loss": 1.9246,
      "step": 190239
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.4415035247802734,
      "learning_rate": 4.387186078092717e-05,
      "loss": 1.8049,
      "step": 190240
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.4639978408813477,
      "learning_rate": 4.386973098814633e-05,
      "loss": 1.9976,
      "step": 190241
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.2711360454559326,
      "learning_rate": 4.386760124298484e-05,
      "loss": 2.0729,
      "step": 190242
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.3677035570144653,
      "learning_rate": 4.3865471545443224e-05,
      "loss": 1.875,
      "step": 190243
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.470682978630066,
      "learning_rate": 4.3863341895521805e-05,
      "loss": 2.0063,
      "step": 190244
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.3735042810440063,
      "learning_rate": 4.38612122932211e-05,
      "loss": 2.0824,
      "step": 190245
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.3285566568374634,
      "learning_rate": 4.385908273854134e-05,
      "loss": 1.8833,
      "step": 190246
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.4267573356628418,
      "learning_rate": 4.385695323148315e-05,
      "loss": 1.8898,
      "step": 190247
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.473140835762024,
      "learning_rate": 4.385482377204667e-05,
      "loss": 2.1328,
      "step": 190248
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.3020719289779663,
      "learning_rate": 4.3852694360232485e-05,
      "loss": 1.8266,
      "step": 190249
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.5909361839294434,
      "learning_rate": 4.385056499604081e-05,
      "loss": 1.7936,
      "step": 190250
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.360249638557434,
      "learning_rate": 4.384843567947225e-05,
      "loss": 2.1868,
      "step": 190251
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.4221644401550293,
      "learning_rate": 4.384630641052706e-05,
      "loss": 2.0366,
      "step": 190252
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.3708709478378296,
      "learning_rate": 4.384417718920577e-05,
      "loss": 2.1619,
      "step": 190253
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.4198225736618042,
      "learning_rate": 4.3842048015508546e-05,
      "loss": 1.895,
      "step": 190254
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.548750400543213,
      "learning_rate": 4.3839918889435995e-05,
      "loss": 2.2903,
      "step": 190255
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.43109929561615,
      "learning_rate": 4.383778981098839e-05,
      "loss": 2.1023,
      "step": 190256
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.314630389213562,
      "learning_rate": 4.3835660780166217e-05,
      "loss": 1.8257,
      "step": 190257
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.3764234781265259,
      "learning_rate": 4.383353179696978e-05,
      "loss": 2.1476,
      "step": 190258
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.3670475482940674,
      "learning_rate": 4.383140286139968e-05,
      "loss": 1.9103,
      "step": 190259
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.3474525213241577,
      "learning_rate": 4.382927397345598e-05,
      "loss": 1.9858,
      "step": 190260
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.4846755266189575,
      "learning_rate": 4.382714513313935e-05,
      "loss": 2.1168,
      "step": 190261
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.4805828332901,
      "learning_rate": 4.382501634044999e-05,
      "loss": 1.9651,
      "step": 190262
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.2993615865707397,
      "learning_rate": 4.38228875953885e-05,
      "loss": 2.0735,
      "step": 190263
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.3715667724609375,
      "learning_rate": 4.3820758897955075e-05,
      "loss": 1.9418,
      "step": 190264
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.4363350868225098,
      "learning_rate": 4.3818630248150285e-05,
      "loss": 1.8224,
      "step": 190265
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.382485270500183,
      "learning_rate": 4.3816501645974464e-05,
      "loss": 1.9658,
      "step": 190266
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.4135512113571167,
      "learning_rate": 4.381437309142798e-05,
      "loss": 2.0617,
      "step": 190267
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.4380879402160645,
      "learning_rate": 4.381224458451116e-05,
      "loss": 2.0414,
      "step": 190268
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.2834529876708984,
      "learning_rate": 4.381011612522457e-05,
      "loss": 1.9994,
      "step": 190269
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.4017670154571533,
      "learning_rate": 4.3807987713568414e-05,
      "loss": 1.9294,
      "step": 190270
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.389601230621338,
      "learning_rate": 4.380585934954329e-05,
      "loss": 2.0842,
      "step": 190271
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.597591757774353,
      "learning_rate": 4.380373103314947e-05,
      "loss": 1.987,
      "step": 190272
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.313020944595337,
      "learning_rate": 4.380160276438731e-05,
      "loss": 2.1103,
      "step": 190273
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.4262244701385498,
      "learning_rate": 4.379947454325735e-05,
      "loss": 2.0914,
      "step": 190274
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.4210292100906372,
      "learning_rate": 4.379734636975989e-05,
      "loss": 2.0478,
      "step": 190275
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.441941738128662,
      "learning_rate": 4.3795218243895256e-05,
      "loss": 1.9966,
      "step": 190276
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.4205522537231445,
      "learning_rate": 4.3793090165664024e-05,
      "loss": 1.9972,
      "step": 190277
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.4421331882476807,
      "learning_rate": 4.379096213506649e-05,
      "loss": 2.073,
      "step": 190278
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.4763281345367432,
      "learning_rate": 4.378883415210298e-05,
      "loss": 1.9989,
      "step": 190279
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.4290976524353027,
      "learning_rate": 4.3786706216774036e-05,
      "loss": 2.2373,
      "step": 190280
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.3435478210449219,
      "learning_rate": 4.378457832907996e-05,
      "loss": 1.9899,
      "step": 190281
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.3902844190597534,
      "learning_rate": 4.37824504890211e-05,
      "loss": 1.8174,
      "step": 190282
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.3645991086959839,
      "learning_rate": 4.378032269659801e-05,
      "loss": 2.0068,
      "step": 190283
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.3996925354003906,
      "learning_rate": 4.377819495181099e-05,
      "loss": 1.917,
      "step": 190284
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.3414469957351685,
      "learning_rate": 4.377606725466035e-05,
      "loss": 1.9586,
      "step": 190285
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.554545283317566,
      "learning_rate": 4.377393960514665e-05,
      "loss": 2.0885,
      "step": 190286
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.7031186819076538,
      "learning_rate": 4.377181200327017e-05,
      "loss": 2.2563,
      "step": 190287
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.379615306854248,
      "learning_rate": 4.376968444903138e-05,
      "loss": 1.8256,
      "step": 190288
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.276973843574524,
      "learning_rate": 4.376755694243066e-05,
      "loss": 1.9659,
      "step": 190289
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.4804357290267944,
      "learning_rate": 4.3765429483468394e-05,
      "loss": 2.0944,
      "step": 190290
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.35391366481781,
      "learning_rate": 4.376330207214489e-05,
      "loss": 2.0264,
      "step": 190291
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.4521563053131104,
      "learning_rate": 4.3761174708460714e-05,
      "loss": 1.9903,
      "step": 190292
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.2810841798782349,
      "learning_rate": 4.375904739241609e-05,
      "loss": 1.762,
      "step": 190293
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.3820470571517944,
      "learning_rate": 4.375692012401156e-05,
      "loss": 2.0429,
      "step": 190294
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.3720601797103882,
      "learning_rate": 4.375479290324746e-05,
      "loss": 1.8328,
      "step": 190295
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.308240294456482,
      "learning_rate": 4.375266573012418e-05,
      "loss": 2.0914,
      "step": 190296
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.3764135837554932,
      "learning_rate": 4.3750538604642026e-05,
      "loss": 2.0811,
      "step": 190297
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.3959381580352783,
      "learning_rate": 4.37484115268016e-05,
      "loss": 2.1727,
      "step": 190298
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.2988626956939697,
      "learning_rate": 4.374628449660306e-05,
      "loss": 1.8921,
      "step": 190299
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.4411547183990479,
      "learning_rate": 4.374415751404704e-05,
      "loss": 2.1028,
      "step": 190300
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.402929425239563,
      "learning_rate": 4.374203057913378e-05,
      "loss": 2.1427,
      "step": 190301
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.6747275590896606,
      "learning_rate": 4.373990369186374e-05,
      "loss": 2.065,
      "step": 190302
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.3151695728302002,
      "learning_rate": 4.373777685223723e-05,
      "loss": 2.1832,
      "step": 190303
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.3142454624176025,
      "learning_rate": 4.373565006025474e-05,
      "loss": 1.9069,
      "step": 190304
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.4254906177520752,
      "learning_rate": 4.3733523315916564e-05,
      "loss": 2.0098,
      "step": 190305
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.4017115831375122,
      "learning_rate": 4.373139661922325e-05,
      "loss": 2.1086,
      "step": 190306
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.3411093950271606,
      "learning_rate": 4.372926997017502e-05,
      "loss": 2.0416,
      "step": 190307
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.4295270442962646,
      "learning_rate": 4.372714336877251e-05,
      "loss": 1.947,
      "step": 190308
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.32685124874115,
      "learning_rate": 4.372501681501582e-05,
      "loss": 1.9488,
      "step": 190309
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.368988037109375,
      "learning_rate": 4.372289030890556e-05,
      "loss": 1.9919,
      "step": 190310
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.3793280124664307,
      "learning_rate": 4.372076385044197e-05,
      "loss": 2.1146,
      "step": 190311
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.376551628112793,
      "learning_rate": 4.371863743962562e-05,
      "loss": 2.1485,
      "step": 190312
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.315210223197937,
      "learning_rate": 4.371651107645674e-05,
      "loss": 1.9148,
      "step": 190313
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.4258015155792236,
      "learning_rate": 4.3714384760935916e-05,
      "loss": 1.9188,
      "step": 190314
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.457255244255066,
      "learning_rate": 4.371225849306331e-05,
      "loss": 2.099,
      "step": 190315
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.2792282104492188,
      "learning_rate": 4.371013227283946e-05,
      "loss": 2.017,
      "step": 190316
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.457485318183899,
      "learning_rate": 4.3708006100264725e-05,
      "loss": 1.948,
      "step": 190317
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.359235405921936,
      "learning_rate": 4.370587997533954e-05,
      "loss": 1.8672,
      "step": 190318
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.5530604124069214,
      "learning_rate": 4.370375389806421e-05,
      "loss": 1.9143,
      "step": 190319
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.5715749263763428,
      "learning_rate": 4.370162786843933e-05,
      "loss": 2.1221,
      "step": 190320
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.6949498653411865,
      "learning_rate": 4.3699501886464994e-05,
      "loss": 2.232,
      "step": 190321
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.4242528676986694,
      "learning_rate": 4.3697375952141846e-05,
      "loss": 2.0583,
      "step": 190322
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.3991243839263916,
      "learning_rate": 4.369525006547014e-05,
      "loss": 1.9401,
      "step": 190323
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.4636694192886353,
      "learning_rate": 4.369312422645039e-05,
      "loss": 1.879,
      "step": 190324
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.5700445175170898,
      "learning_rate": 4.3690998435082824e-05,
      "loss": 2.0072,
      "step": 190325
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.3649272918701172,
      "learning_rate": 4.36888726913681e-05,
      "loss": 1.9464,
      "step": 190326
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.262111783027649,
      "learning_rate": 4.368674699530633e-05,
      "loss": 1.8418,
      "step": 190327
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.301863431930542,
      "learning_rate": 4.368462134689811e-05,
      "loss": 1.9759,
      "step": 190328
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.5040619373321533,
      "learning_rate": 4.3682495746143666e-05,
      "loss": 1.9417,
      "step": 190329
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.408861517906189,
      "learning_rate": 4.368037019304353e-05,
      "loss": 1.9825,
      "step": 190330
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.3883916139602661,
      "learning_rate": 4.367824468759802e-05,
      "loss": 1.9139,
      "step": 190331
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.5265161991119385,
      "learning_rate": 4.367611922980768e-05,
      "loss": 1.8912,
      "step": 190332
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.3736010789871216,
      "learning_rate": 4.367399381967266e-05,
      "loss": 1.9218,
      "step": 190333
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.4693642854690552,
      "learning_rate": 4.367186845719355e-05,
      "loss": 2.1704,
      "step": 190334
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.6329292058944702,
      "learning_rate": 4.3669743142370616e-05,
      "loss": 2.0101,
      "step": 190335
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.4084007740020752,
      "learning_rate": 4.3667617875204364e-05,
      "loss": 2.0632,
      "step": 190336
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.3153389692306519,
      "learning_rate": 4.366549265569509e-05,
      "loss": 1.802,
      "step": 190337
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.3567743301391602,
      "learning_rate": 4.366336748384336e-05,
      "loss": 1.994,
      "step": 190338
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.459797739982605,
      "learning_rate": 4.366124235964934e-05,
      "loss": 2.0435,
      "step": 190339
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.5030149221420288,
      "learning_rate": 4.365911728311356e-05,
      "loss": 1.7268,
      "step": 190340
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.4080702066421509,
      "learning_rate": 4.365699225423637e-05,
      "loss": 2.0805,
      "step": 190341
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.3531004190444946,
      "learning_rate": 4.365486727301821e-05,
      "loss": 1.983,
      "step": 190342
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.4478813409805298,
      "learning_rate": 4.36527423394594e-05,
      "loss": 1.7194,
      "step": 190343
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.3506850004196167,
      "learning_rate": 4.3650617453560534e-05,
      "loss": 1.8413,
      "step": 190344
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.375467300415039,
      "learning_rate": 4.364849261532171e-05,
      "loss": 2.0407,
      "step": 190345
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.5253099203109741,
      "learning_rate": 4.3646367824743525e-05,
      "loss": 1.9654,
      "step": 190346
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.3076382875442505,
      "learning_rate": 4.3644243081826245e-05,
      "loss": 2.0103,
      "step": 190347
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.4139615297317505,
      "learning_rate": 4.3642118386570444e-05,
      "loss": 1.9538,
      "step": 190348
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.2645297050476074,
      "learning_rate": 4.363999373897632e-05,
      "loss": 1.8855,
      "step": 190349
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.422623872756958,
      "learning_rate": 4.3637869139044436e-05,
      "loss": 2.1116,
      "step": 190350
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.4848695993423462,
      "learning_rate": 4.363574458677509e-05,
      "loss": 1.9166,
      "step": 190351
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.291844367980957,
      "learning_rate": 4.363362008216872e-05,
      "loss": 1.818,
      "step": 190352
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.2910962104797363,
      "learning_rate": 4.3631495625225624e-05,
      "loss": 1.9054,
      "step": 190353
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.4761974811553955,
      "learning_rate": 4.362937121594636e-05,
      "loss": 1.8854,
      "step": 190354
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.4357450008392334,
      "learning_rate": 4.3627246854331146e-05,
      "loss": 2.1261,
      "step": 190355
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.411162257194519,
      "learning_rate": 4.362512254038053e-05,
      "loss": 1.9841,
      "step": 190356
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.5497857332229614,
      "learning_rate": 4.3622998274094854e-05,
      "loss": 1.9547,
      "step": 190357
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.463404655456543,
      "learning_rate": 4.362087405547449e-05,
      "loss": 2.1105,
      "step": 190358
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.4369856119155884,
      "learning_rate": 4.361874988451979e-05,
      "loss": 2.0485,
      "step": 190359
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.5230525732040405,
      "learning_rate": 4.361662576123126e-05,
      "loss": 2.2567,
      "step": 190360
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.374974250793457,
      "learning_rate": 4.3614501685609206e-05,
      "loss": 2.0003,
      "step": 190361
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.380496859550476,
      "learning_rate": 4.3612377657654086e-05,
      "loss": 1.825,
      "step": 190362
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.6961922645568848,
      "learning_rate": 4.361025367736627e-05,
      "loss": 2.1393,
      "step": 190363
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.4182924032211304,
      "learning_rate": 4.360812974474609e-05,
      "loss": 1.8821,
      "step": 190364
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.3542770147323608,
      "learning_rate": 4.360600585979408e-05,
      "loss": 2.0902,
      "step": 190365
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.3681162595748901,
      "learning_rate": 4.3603882022510506e-05,
      "loss": 2.11,
      "step": 190366
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.360495924949646,
      "learning_rate": 4.36017582328958e-05,
      "loss": 2.0795,
      "step": 190367
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.3286890983581543,
      "learning_rate": 4.3599634490950394e-05,
      "loss": 1.9988,
      "step": 190368
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.434120535850525,
      "learning_rate": 4.359751079667466e-05,
      "loss": 1.9667,
      "step": 190369
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.4870648384094238,
      "learning_rate": 4.3595387150068926e-05,
      "loss": 2.0854,
      "step": 190370
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.4222370386123657,
      "learning_rate": 4.359326355113373e-05,
      "loss": 2.0924,
      "step": 190371
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.2931727170944214,
      "learning_rate": 4.35911399998693e-05,
      "loss": 1.7746,
      "step": 190372
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.4450455904006958,
      "learning_rate": 4.35890164962762e-05,
      "loss": 2.0916,
      "step": 190373
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.5122298002243042,
      "learning_rate": 4.358689304035473e-05,
      "loss": 2.0937,
      "step": 190374
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.3408288955688477,
      "learning_rate": 4.35847696321053e-05,
      "loss": 1.8682,
      "step": 190375
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.2897844314575195,
      "learning_rate": 4.358264627152827e-05,
      "loss": 2.0432,
      "step": 190376
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.3149358034133911,
      "learning_rate": 4.3580522958624064e-05,
      "loss": 2.1853,
      "step": 190377
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.3790082931518555,
      "learning_rate": 4.357839969339306e-05,
      "loss": 2.0182,
      "step": 190378
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.363558292388916,
      "learning_rate": 4.3576276475835726e-05,
      "loss": 1.7674,
      "step": 190379
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.4117231369018555,
      "learning_rate": 4.357415330595242e-05,
      "loss": 1.947,
      "step": 190380
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.3714804649353027,
      "learning_rate": 4.3572030183743514e-05,
      "loss": 1.9448,
      "step": 190381
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.5564546585083008,
      "learning_rate": 4.35699071092093e-05,
      "loss": 1.9955,
      "step": 190382
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.5658234357833862,
      "learning_rate": 4.356778408235039e-05,
      "loss": 2.101,
      "step": 190383
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.465178370475769,
      "learning_rate": 4.3565661103167e-05,
      "loss": 1.9092,
      "step": 190384
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.3987751007080078,
      "learning_rate": 4.356353817165965e-05,
      "loss": 1.7494,
      "step": 190385
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.4597601890563965,
      "learning_rate": 4.356141528782859e-05,
      "loss": 2.0248,
      "step": 190386
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.4120553731918335,
      "learning_rate": 4.3559292451674464e-05,
      "loss": 1.9456,
      "step": 190387
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.1904206275939941,
      "learning_rate": 4.355716966319737e-05,
      "loss": 2.0003,
      "step": 190388
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.4192537069320679,
      "learning_rate": 4.355504692239793e-05,
      "loss": 2.0816,
      "step": 190389
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.39111328125,
      "learning_rate": 4.3552924229276355e-05,
      "loss": 1.9912,
      "step": 190390
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.5221251249313354,
      "learning_rate": 4.355080158383317e-05,
      "loss": 2.194,
      "step": 190391
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.387661337852478,
      "learning_rate": 4.354867898606872e-05,
      "loss": 2.0441,
      "step": 190392
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.4004815816879272,
      "learning_rate": 4.3546556435983525e-05,
      "loss": 1.9724,
      "step": 190393
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.3518037796020508,
      "learning_rate": 4.3544433933577725e-05,
      "loss": 2.0337,
      "step": 190394
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.4720689058303833,
      "learning_rate": 4.354231147885192e-05,
      "loss": 2.3115,
      "step": 190395
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.3899725675582886,
      "learning_rate": 4.3540189071806375e-05,
      "loss": 1.836,
      "step": 190396
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.3429737091064453,
      "learning_rate": 4.3538066712441624e-05,
      "loss": 2.0119,
      "step": 190397
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.429553747177124,
      "learning_rate": 4.3535944400757896e-05,
      "loss": 2.0763,
      "step": 190398
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.4545873403549194,
      "learning_rate": 4.353382213675586e-05,
      "loss": 2.1347,
      "step": 190399
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.405339002609253,
      "learning_rate": 4.353169992043555e-05,
      "loss": 2.1434,
      "step": 190400
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.4574607610702515,
      "learning_rate": 4.3529577751797626e-05,
      "loss": 1.9735,
      "step": 190401
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.3842613697052002,
      "learning_rate": 4.352745563084233e-05,
      "loss": 1.9923,
      "step": 190402
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.7249053716659546,
      "learning_rate": 4.352533355757016e-05,
      "loss": 1.9836,
      "step": 190403
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.4963711500167847,
      "learning_rate": 4.352321153198145e-05,
      "loss": 2.2127,
      "step": 190404
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.2891749143600464,
      "learning_rate": 4.3521089554076725e-05,
      "loss": 2.139,
      "step": 190405
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.3912827968597412,
      "learning_rate": 4.351896762385613e-05,
      "loss": 1.9691,
      "step": 190406
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.4317857027053833,
      "learning_rate": 4.351684574132028e-05,
      "loss": 2.0389,
      "step": 190407
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.4439339637756348,
      "learning_rate": 4.351472390646939e-05,
      "loss": 2.2203,
      "step": 190408
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.3300104141235352,
      "learning_rate": 4.351260211930406e-05,
      "loss": 1.9865,
      "step": 190409
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.429779052734375,
      "learning_rate": 4.351048037982452e-05,
      "loss": 1.745,
      "step": 190410
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.4709895849227905,
      "learning_rate": 4.350835868803133e-05,
      "loss": 1.9893,
      "step": 190411
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.5863823890686035,
      "learning_rate": 4.3506237043924664e-05,
      "loss": 2.1083,
      "step": 190412
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.3387845754623413,
      "learning_rate": 4.350411544750508e-05,
      "loss": 1.7961,
      "step": 190413
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.236356258392334,
      "learning_rate": 4.3501993898772856e-05,
      "loss": 1.9218,
      "step": 190414
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.4442942142486572,
      "learning_rate": 4.349987239772852e-05,
      "loss": 2.0567,
      "step": 190415
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.3873399496078491,
      "learning_rate": 4.349775094437233e-05,
      "loss": 1.9772,
      "step": 190416
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.3293707370758057,
      "learning_rate": 4.34956295387049e-05,
      "loss": 2.0886,
      "step": 190417
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.4298607110977173,
      "learning_rate": 4.349350818072632e-05,
      "loss": 1.9579,
      "step": 190418
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.2777550220489502,
      "learning_rate": 4.349138687043723e-05,
      "loss": 1.9772,
      "step": 190419
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.3706262111663818,
      "learning_rate": 4.348926560783785e-05,
      "loss": 2.0236,
      "step": 190420
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.2821227312088013,
      "learning_rate": 4.348714439292873e-05,
      "loss": 2.1988,
      "step": 190421
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.3243714570999146,
      "learning_rate": 4.348502322571013e-05,
      "loss": 2.0857,
      "step": 190422
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.434471845626831,
      "learning_rate": 4.3482902106182636e-05,
      "loss": 2.272,
      "step": 190423
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.4955055713653564,
      "learning_rate": 4.3480781034346366e-05,
      "loss": 2.1424,
      "step": 190424
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.3466389179229736,
      "learning_rate": 4.347866001020195e-05,
      "loss": 1.9661,
      "step": 190425
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.4179494380950928,
      "learning_rate": 4.347653903374961e-05,
      "loss": 2.0375,
      "step": 190426
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.4305952787399292,
      "learning_rate": 4.34744181049899e-05,
      "loss": 1.7252,
      "step": 190427
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.2823809385299683,
      "learning_rate": 4.347229722392306e-05,
      "loss": 1.9523,
      "step": 190428
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.3454660177230835,
      "learning_rate": 4.3470176390549704e-05,
      "loss": 1.9617,
      "step": 190429
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.5407079458236694,
      "learning_rate": 4.346805560486997e-05,
      "loss": 2.1844,
      "step": 190430
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.3240753412246704,
      "learning_rate": 4.346593486688441e-05,
      "loss": 1.8549,
      "step": 190431
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.6086548566818237,
      "learning_rate": 4.3463814176593305e-05,
      "loss": 1.9721,
      "step": 190432
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.5233365297317505,
      "learning_rate": 4.3461693533997175e-05,
      "loss": 2.1859,
      "step": 190433
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.571685552597046,
      "learning_rate": 4.345957293909633e-05,
      "loss": 1.9004,
      "step": 190434
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.7160106897354126,
      "learning_rate": 4.3457452391891235e-05,
      "loss": 2.1567,
      "step": 190435
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.3691061735153198,
      "learning_rate": 4.345533189238225e-05,
      "loss": 1.993,
      "step": 190436
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.384893536567688,
      "learning_rate": 4.345321144056974e-05,
      "loss": 1.8898,
      "step": 190437
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.6549066305160522,
      "learning_rate": 4.3451091036454086e-05,
      "loss": 2.1718,
      "step": 190438
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.3929927349090576,
      "learning_rate": 4.3448970680035775e-05,
      "loss": 2.1858,
      "step": 190439
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.6437410116195679,
      "learning_rate": 4.344685037131505e-05,
      "loss": 2.0547,
      "step": 190440
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.3481836318969727,
      "learning_rate": 4.344473011029249e-05,
      "loss": 1.841,
      "step": 190441
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.600366473197937,
      "learning_rate": 4.3442609896968387e-05,
      "loss": 2.0857,
      "step": 190442
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.3514528274536133,
      "learning_rate": 4.3440489731343156e-05,
      "loss": 2.0327,
      "step": 190443
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.5329859256744385,
      "learning_rate": 4.34383696134171e-05,
      "loss": 2.0514,
      "step": 190444
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.4193581342697144,
      "learning_rate": 4.3436249543190794e-05,
      "loss": 2.0206,
      "step": 190445
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.4723666906356812,
      "learning_rate": 4.343412952066443e-05,
      "loss": 2.1512,
      "step": 190446
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.4342658519744873,
      "learning_rate": 4.343200954583857e-05,
      "loss": 2.0754,
      "step": 190447
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.4854336977005005,
      "learning_rate": 4.342988961871356e-05,
      "loss": 1.9676,
      "step": 190448
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.4479387998580933,
      "learning_rate": 4.342776973928973e-05,
      "loss": 1.8086,
      "step": 190449
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.2652205228805542,
      "learning_rate": 4.3425649907567574e-05,
      "loss": 2.0061,
      "step": 190450
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.414370059967041,
      "learning_rate": 4.342353012354742e-05,
      "loss": 1.9186,
      "step": 190451
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.4535537958145142,
      "learning_rate": 4.342141038722958e-05,
      "loss": 1.8781,
      "step": 190452
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.416397213935852,
      "learning_rate": 4.341929069861465e-05,
      "loss": 1.9545,
      "step": 190453
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.4488869905471802,
      "learning_rate": 4.341717105770293e-05,
      "loss": 2.0126,
      "step": 190454
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.2149813175201416,
      "learning_rate": 4.341505146449471e-05,
      "loss": 1.6858,
      "step": 190455
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.2416999340057373,
      "learning_rate": 4.3412931918990535e-05,
      "loss": 1.9476,
      "step": 190456
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.385364294052124,
      "learning_rate": 4.3410812421190664e-05,
      "loss": 1.95,
      "step": 190457
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.3049547672271729,
      "learning_rate": 4.340869297109567e-05,
      "loss": 2.24,
      "step": 190458
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.3586896657943726,
      "learning_rate": 4.340657356870581e-05,
      "loss": 1.9373,
      "step": 190459
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.370474100112915,
      "learning_rate": 4.340445421402153e-05,
      "loss": 2.0572,
      "step": 190460
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.463171362876892,
      "learning_rate": 4.3402334907043154e-05,
      "loss": 1.7764,
      "step": 190461
    },
    {
      "epoch": 2.48,
      "grad_norm": 2.2100157737731934,
      "learning_rate": 4.340021564777115e-05,
      "loss": 2.0044,
      "step": 190462
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.3220587968826294,
      "learning_rate": 4.339809643620586e-05,
      "loss": 1.7767,
      "step": 190463
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.4426771402359009,
      "learning_rate": 4.339597727234776e-05,
      "loss": 1.988,
      "step": 190464
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.3704109191894531,
      "learning_rate": 4.33938581561972e-05,
      "loss": 2.1966,
      "step": 190465
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.442883014678955,
      "learning_rate": 4.339173908775455e-05,
      "loss": 2.1005,
      "step": 190466
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.3817834854125977,
      "learning_rate": 4.338962006702017e-05,
      "loss": 1.992,
      "step": 190467
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.374182939529419,
      "learning_rate": 4.338750109399456e-05,
      "loss": 2.0225,
      "step": 190468
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.3654656410217285,
      "learning_rate": 4.3385382168678016e-05,
      "loss": 1.9861,
      "step": 190469
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.4304800033569336,
      "learning_rate": 4.3383263291071014e-05,
      "loss": 2.0338,
      "step": 190470
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.3739254474639893,
      "learning_rate": 4.338114446117384e-05,
      "loss": 2.073,
      "step": 190471
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.360164761543274,
      "learning_rate": 4.337902567898708e-05,
      "loss": 1.9638,
      "step": 190472
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.4219683408737183,
      "learning_rate": 4.337690694451088e-05,
      "loss": 1.935,
      "step": 190473
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.7942180633544922,
      "learning_rate": 4.337478825774585e-05,
      "loss": 1.8902,
      "step": 190474
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.3537942171096802,
      "learning_rate": 4.337266961869219e-05,
      "loss": 1.7453,
      "step": 190475
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.4259312152862549,
      "learning_rate": 4.3370551027350464e-05,
      "loss": 2.0123,
      "step": 190476
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.3387686014175415,
      "learning_rate": 4.336843248372094e-05,
      "loss": 2.0265,
      "step": 190477
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.5362699031829834,
      "learning_rate": 4.336631398780421e-05,
      "loss": 1.9697,
      "step": 190478
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.439969778060913,
      "learning_rate": 4.336419553960039e-05,
      "loss": 2.0257,
      "step": 190479
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.2551614046096802,
      "learning_rate": 4.3362077139110064e-05,
      "loss": 2.0903,
      "step": 190480
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.345794916152954,
      "learning_rate": 4.3359958786333506e-05,
      "loss": 1.9876,
      "step": 190481
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.4321997165679932,
      "learning_rate": 4.3357840481271276e-05,
      "loss": 1.9394,
      "step": 190482
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.2869776487350464,
      "learning_rate": 4.335572222392355e-05,
      "loss": 2.1891,
      "step": 190483
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.3869258165359497,
      "learning_rate": 4.335360401429102e-05,
      "loss": 1.9035,
      "step": 190484
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.5076203346252441,
      "learning_rate": 4.3351485852373726e-05,
      "loss": 1.9574,
      "step": 190485
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.4602689743041992,
      "learning_rate": 4.3349367738172325e-05,
      "loss": 1.9795,
      "step": 190486
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.4314950704574585,
      "learning_rate": 4.3347249671687056e-05,
      "loss": 2.0517,
      "step": 190487
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.4225945472717285,
      "learning_rate": 4.334513165291845e-05,
      "loss": 1.9071,
      "step": 190488
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.4402738809585571,
      "learning_rate": 4.3343013681866744e-05,
      "loss": 2.0177,
      "step": 190489
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.35637629032135,
      "learning_rate": 4.334089575853257e-05,
      "loss": 1.9406,
      "step": 190490
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.373904824256897,
      "learning_rate": 4.333877788291599e-05,
      "loss": 2.1398,
      "step": 190491
    },
    {
      "epoch": 2.48,
      "grad_norm": 2.008779287338257,
      "learning_rate": 4.33366600550177e-05,
      "loss": 2.0908,
      "step": 190492
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.3411483764648438,
      "learning_rate": 4.333454227483788e-05,
      "loss": 1.796,
      "step": 190493
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.3779654502868652,
      "learning_rate": 4.3332424542377054e-05,
      "loss": 1.9606,
      "step": 190494
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.3955246210098267,
      "learning_rate": 4.3330306857635554e-05,
      "loss": 2.031,
      "step": 190495
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.4012632369995117,
      "learning_rate": 4.3328189220613884e-05,
      "loss": 1.8775,
      "step": 190496
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.409311294555664,
      "learning_rate": 4.332607163131224e-05,
      "loss": 2.1472,
      "step": 190497
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.3444492816925049,
      "learning_rate": 4.3323954089731196e-05,
      "loss": 2.2653,
      "step": 190498
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.3477803468704224,
      "learning_rate": 4.332183659587101e-05,
      "loss": 2.0405,
      "step": 190499
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.4727791547775269,
      "learning_rate": 4.331971914973218e-05,
      "loss": 1.9552,
      "step": 190500
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.3926893472671509,
      "learning_rate": 4.3317601751315015e-05,
      "loss": 2.0427,
      "step": 190501
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.40300714969635,
      "learning_rate": 4.3315484400620006e-05,
      "loss": 2.1806,
      "step": 190502
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.3955864906311035,
      "learning_rate": 4.3313367097647524e-05,
      "loss": 2.1985,
      "step": 190503
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.4882062673568726,
      "learning_rate": 4.33112498423979e-05,
      "loss": 2.0848,
      "step": 190504
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.3034008741378784,
      "learning_rate": 4.3309132634871505e-05,
      "loss": 1.9152,
      "step": 190505
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.3844584226608276,
      "learning_rate": 4.330701547506886e-05,
      "loss": 2.0509,
      "step": 190506
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.4367340803146362,
      "learning_rate": 4.330489836299022e-05,
      "loss": 2.1459,
      "step": 190507
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.4291479587554932,
      "learning_rate": 4.330278129863609e-05,
      "loss": 1.8054,
      "step": 190508
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.3431031703948975,
      "learning_rate": 4.3300664282006825e-05,
      "loss": 1.8558,
      "step": 190509
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.4750934839248657,
      "learning_rate": 4.3298547313102816e-05,
      "loss": 2.1906,
      "step": 190510
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.5484427213668823,
      "learning_rate": 4.32964303919244e-05,
      "loss": 2.058,
      "step": 190511
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.44571053981781,
      "learning_rate": 4.329431351847207e-05,
      "loss": 2.2851,
      "step": 190512
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.4777144193649292,
      "learning_rate": 4.329219669274613e-05,
      "loss": 2.179,
      "step": 190513
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.2277162075042725,
      "learning_rate": 4.3290079914747075e-05,
      "loss": 2.1039,
      "step": 190514
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.6189091205596924,
      "learning_rate": 4.328796318447522e-05,
      "loss": 2.1188,
      "step": 190515
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.4039015769958496,
      "learning_rate": 4.328584650193101e-05,
      "loss": 1.8166,
      "step": 190516
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.6110985279083252,
      "learning_rate": 4.328372986711472e-05,
      "loss": 2.1216,
      "step": 190517
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.4891204833984375,
      "learning_rate": 4.3281613280026895e-05,
      "loss": 2.0471,
      "step": 190518
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.2713676691055298,
      "learning_rate": 4.327949674066779e-05,
      "loss": 2.2224,
      "step": 190519
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.456827163696289,
      "learning_rate": 4.3277380249037965e-05,
      "loss": 2.0305,
      "step": 190520
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.4161710739135742,
      "learning_rate": 4.3275263805137696e-05,
      "loss": 1.9877,
      "step": 190521
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.807835340499878,
      "learning_rate": 4.3273147408967415e-05,
      "loss": 1.9004,
      "step": 190522
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.3879724740982056,
      "learning_rate": 4.327103106052745e-05,
      "loss": 2.1522,
      "step": 190523
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.3776726722717285,
      "learning_rate": 4.3268914759818274e-05,
      "loss": 1.8336,
      "step": 190524
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.4914685487747192,
      "learning_rate": 4.3266798506840214e-05,
      "loss": 2.1355,
      "step": 190525
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.335979700088501,
      "learning_rate": 4.3264682301593775e-05,
      "loss": 1.5772,
      "step": 190526
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.4266163110733032,
      "learning_rate": 4.3262566144079256e-05,
      "loss": 2.0636,
      "step": 190527
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.4573708772659302,
      "learning_rate": 4.326045003429709e-05,
      "loss": 2.0669,
      "step": 190528
    },
    {
      "epoch": 2.48,
      "grad_norm": 2.227342367172241,
      "learning_rate": 4.3258333972247574e-05,
      "loss": 2.0174,
      "step": 190529
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.6597357988357544,
      "learning_rate": 4.325621795793127e-05,
      "loss": 2.2579,
      "step": 190530
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.391378402709961,
      "learning_rate": 4.325410199134839e-05,
      "loss": 2.0579,
      "step": 190531
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.250762939453125,
      "learning_rate": 4.325198607249949e-05,
      "loss": 2.0099,
      "step": 190532
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.7842603921890259,
      "learning_rate": 4.3249870201384874e-05,
      "loss": 1.9458,
      "step": 190533
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.3819677829742432,
      "learning_rate": 4.324775437800491e-05,
      "loss": 1.9648,
      "step": 190534
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.3386263847351074,
      "learning_rate": 4.3245638602360124e-05,
      "loss": 2.05,
      "step": 190535
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.507581353187561,
      "learning_rate": 4.324352287445079e-05,
      "loss": 2.0361,
      "step": 190536
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.5082453489303589,
      "learning_rate": 4.324140719427727e-05,
      "loss": 2.1943,
      "step": 190537
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.4191069602966309,
      "learning_rate": 4.32392915618401e-05,
      "loss": 1.8774,
      "step": 190538
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.300333023071289,
      "learning_rate": 4.323717597713957e-05,
      "loss": 1.9743,
      "step": 190539
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.5168379545211792,
      "learning_rate": 4.3235060440176025e-05,
      "loss": 2.2067,
      "step": 190540
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.365127444267273,
      "learning_rate": 4.3232944950949996e-05,
      "loss": 2.0601,
      "step": 190541
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.450123906135559,
      "learning_rate": 4.323082950946185e-05,
      "loss": 1.8896,
      "step": 190542
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.6427286863327026,
      "learning_rate": 4.3228714115711816e-05,
      "loss": 2.0882,
      "step": 190543
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.6655573844909668,
      "learning_rate": 4.3226598769700525e-05,
      "loss": 1.9664,
      "step": 190544
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.5209379196166992,
      "learning_rate": 4.322448347142825e-05,
      "loss": 1.7387,
      "step": 190545
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.4423654079437256,
      "learning_rate": 4.322236822089529e-05,
      "loss": 2.0645,
      "step": 190546
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.4270682334899902,
      "learning_rate": 4.3220253018102234e-05,
      "loss": 2.0866,
      "step": 190547
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.6218594312667847,
      "learning_rate": 4.32181378630493e-05,
      "loss": 2.0177,
      "step": 190548
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.5172035694122314,
      "learning_rate": 4.3216022755737034e-05,
      "loss": 1.9288,
      "step": 190549
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.3582080602645874,
      "learning_rate": 4.321390769616575e-05,
      "loss": 2.082,
      "step": 190550
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.5560853481292725,
      "learning_rate": 4.321179268433588e-05,
      "loss": 1.9761,
      "step": 190551
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.6447914838790894,
      "learning_rate": 4.3209677720247685e-05,
      "loss": 2.1115,
      "step": 190552
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.3202561140060425,
      "learning_rate": 4.320756280390173e-05,
      "loss": 2.0632,
      "step": 190553
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.4578264951705933,
      "learning_rate": 4.3205447935298287e-05,
      "loss": 2.0025,
      "step": 190554
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.3222726583480835,
      "learning_rate": 4.320333311443782e-05,
      "loss": 2.0308,
      "step": 190555
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.9488478899002075,
      "learning_rate": 4.3201218341320666e-05,
      "loss": 1.9281,
      "step": 190556
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.607696294784546,
      "learning_rate": 4.319910361594742e-05,
      "loss": 1.9213,
      "step": 190557
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.4338973760604858,
      "learning_rate": 4.319698893831811e-05,
      "loss": 1.8305,
      "step": 190558
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.3310585021972656,
      "learning_rate": 4.3194874308433445e-05,
      "loss": 1.9791,
      "step": 190559
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.3071492910385132,
      "learning_rate": 4.319275972629359e-05,
      "loss": 1.7962,
      "step": 190560
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.4028843641281128,
      "learning_rate": 4.3190645191899145e-05,
      "loss": 1.7375,
      "step": 190561
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.3761565685272217,
      "learning_rate": 4.3188530705250335e-05,
      "loss": 2.0397,
      "step": 190562
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.4484409093856812,
      "learning_rate": 4.318641626634777e-05,
      "loss": 1.9217,
      "step": 190563
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.3890972137451172,
      "learning_rate": 4.318430187519154e-05,
      "loss": 1.9609,
      "step": 190564
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.3793836832046509,
      "learning_rate": 4.3182187531782286e-05,
      "loss": 1.8442,
      "step": 190565
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.3066357374191284,
      "learning_rate": 4.3180073236120236e-05,
      "loss": 2.1972,
      "step": 190566
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.4934865236282349,
      "learning_rate": 4.317795898820589e-05,
      "loss": 2.0266,
      "step": 190567
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.455087661743164,
      "learning_rate": 4.317584478803959e-05,
      "loss": 2.0867,
      "step": 190568
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.3908252716064453,
      "learning_rate": 4.3173730635621796e-05,
      "loss": 1.8846,
      "step": 190569
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.6477006673812866,
      "learning_rate": 4.317161653095287e-05,
      "loss": 1.9592,
      "step": 190570
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.4728726148605347,
      "learning_rate": 4.316950247403318e-05,
      "loss": 1.9936,
      "step": 190571
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.500044822692871,
      "learning_rate": 4.316738846486303e-05,
      "loss": 2.0272,
      "step": 190572
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.4323347806930542,
      "learning_rate": 4.316527450344302e-05,
      "loss": 2.0795,
      "step": 190573
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.4404271841049194,
      "learning_rate": 4.316316058977334e-05,
      "loss": 1.9401,
      "step": 190574
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.3416022062301636,
      "learning_rate": 4.3161046723854565e-05,
      "loss": 1.8141,
      "step": 190575
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.3597230911254883,
      "learning_rate": 4.3158932905686996e-05,
      "loss": 2.0403,
      "step": 190576
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.687117338180542,
      "learning_rate": 4.315681913527099e-05,
      "loss": 1.7897,
      "step": 190577
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.3849974870681763,
      "learning_rate": 4.315470541260696e-05,
      "loss": 2.2177,
      "step": 190578
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.4014137983322144,
      "learning_rate": 4.3152591737695366e-05,
      "loss": 2.0651,
      "step": 190579
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.4249184131622314,
      "learning_rate": 4.3150478110536505e-05,
      "loss": 2.0864,
      "step": 190580
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.4403536319732666,
      "learning_rate": 4.3148364531130875e-05,
      "loss": 1.9562,
      "step": 190581
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.4953434467315674,
      "learning_rate": 4.314625099947878e-05,
      "loss": 1.9968,
      "step": 190582
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.2977149486541748,
      "learning_rate": 4.314413751558069e-05,
      "loss": 1.9938,
      "step": 190583
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.4149401187896729,
      "learning_rate": 4.3142024079436864e-05,
      "loss": 2.223,
      "step": 190584
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.3070257902145386,
      "learning_rate": 4.313991069104788e-05,
      "loss": 1.9884,
      "step": 190585
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.4968804121017456,
      "learning_rate": 4.3137797350413916e-05,
      "loss": 1.9477,
      "step": 190586
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.4151164293289185,
      "learning_rate": 4.313568405753559e-05,
      "loss": 1.9384,
      "step": 190587
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.5555691719055176,
      "learning_rate": 4.313357081241317e-05,
      "loss": 2.1547,
      "step": 190588
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.398118257522583,
      "learning_rate": 4.3131457615047104e-05,
      "loss": 2.0631,
      "step": 190589
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.4252792596817017,
      "learning_rate": 4.3129344465437646e-05,
      "loss": 1.9009,
      "step": 190590
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.4955474138259888,
      "learning_rate": 4.3127231363585377e-05,
      "loss": 1.9864,
      "step": 190591
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.4501646757125854,
      "learning_rate": 4.312511830949051e-05,
      "loss": 2.0612,
      "step": 190592
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.340477705001831,
      "learning_rate": 4.3123005303153614e-05,
      "loss": 2.0107,
      "step": 190593
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.4308992624282837,
      "learning_rate": 4.312089234457501e-05,
      "loss": 1.9633,
      "step": 190594
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.3604612350463867,
      "learning_rate": 4.311877943375507e-05,
      "loss": 1.9606,
      "step": 190595
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.3604350090026855,
      "learning_rate": 4.311666657069416e-05,
      "loss": 1.8776,
      "step": 190596
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.6836398839950562,
      "learning_rate": 4.3114553755392744e-05,
      "loss": 2.1598,
      "step": 190597
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.3935190439224243,
      "learning_rate": 4.3112440987851125e-05,
      "loss": 1.753,
      "step": 190598
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.4039028882980347,
      "learning_rate": 4.31103282680698e-05,
      "loss": 2.212,
      "step": 190599
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.3808016777038574,
      "learning_rate": 4.3108215596049146e-05,
      "loss": 2.0953,
      "step": 190600
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.3746070861816406,
      "learning_rate": 4.310610297178951e-05,
      "loss": 2.1667,
      "step": 190601
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.2922788858413696,
      "learning_rate": 4.310399039529124e-05,
      "loss": 1.7897,
      "step": 190602
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.3944690227508545,
      "learning_rate": 4.310187786655483e-05,
      "loss": 2.0363,
      "step": 190603
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.4566224813461304,
      "learning_rate": 4.3099765385580586e-05,
      "loss": 2.0238,
      "step": 190604
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.4522286653518677,
      "learning_rate": 4.3097652952369e-05,
      "loss": 1.7966,
      "step": 190605
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.3576632738113403,
      "learning_rate": 4.3095540566920436e-05,
      "loss": 2.034,
      "step": 190606
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.5200133323669434,
      "learning_rate": 4.309342822923524e-05,
      "loss": 2.0779,
      "step": 190607
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.3590009212493896,
      "learning_rate": 4.3091315939313764e-05,
      "loss": 2.1224,
      "step": 190608
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.787871241569519,
      "learning_rate": 4.3089203697156514e-05,
      "loss": 2.1504,
      "step": 190609
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.9625803232192993,
      "learning_rate": 4.308709150276376e-05,
      "loss": 1.8014,
      "step": 190610
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.36479651927948,
      "learning_rate": 4.308497935613606e-05,
      "loss": 2.0538,
      "step": 190611
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.6688390970230103,
      "learning_rate": 4.308286725727369e-05,
      "loss": 2.0873,
      "step": 190612
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.442338228225708,
      "learning_rate": 4.3080755206177075e-05,
      "loss": 2.0815,
      "step": 190613
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.375425100326538,
      "learning_rate": 4.307864320284652e-05,
      "loss": 1.9646,
      "step": 190614
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.5760066509246826,
      "learning_rate": 4.307653124728259e-05,
      "loss": 1.8968,
      "step": 190615
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.502771019935608,
      "learning_rate": 4.307441933948548e-05,
      "loss": 1.8956,
      "step": 190616
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.535752534866333,
      "learning_rate": 4.307230747945576e-05,
      "loss": 2.1172,
      "step": 190617
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.7788292169570923,
      "learning_rate": 4.3070195667193764e-05,
      "loss": 1.766,
      "step": 190618
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.436929702758789,
      "learning_rate": 4.306808390269979e-05,
      "loss": 1.9301,
      "step": 190619
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.4451509714126587,
      "learning_rate": 4.306597218597437e-05,
      "loss": 1.8525,
      "step": 190620
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.4549401998519897,
      "learning_rate": 4.306386051701784e-05,
      "loss": 1.8432,
      "step": 190621
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.406717300415039,
      "learning_rate": 4.306174889583054e-05,
      "loss": 1.8248,
      "step": 190622
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.730875015258789,
      "learning_rate": 4.305963732241295e-05,
      "loss": 1.8969,
      "step": 190623
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.3131353855133057,
      "learning_rate": 4.305752579676546e-05,
      "loss": 2.0493,
      "step": 190624
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.5370934009552002,
      "learning_rate": 4.305541431888835e-05,
      "loss": 1.9057,
      "step": 190625
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.4373219013214111,
      "learning_rate": 4.3053302888782126e-05,
      "loss": 2.3541,
      "step": 190626
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.2900378704071045,
      "learning_rate": 4.305119150644716e-05,
      "loss": 1.8637,
      "step": 190627
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.385661244392395,
      "learning_rate": 4.3049080171883774e-05,
      "loss": 2.1833,
      "step": 190628
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.5389654636383057,
      "learning_rate": 4.304696888509245e-05,
      "loss": 2.0714,
      "step": 190629
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.4179470539093018,
      "learning_rate": 4.3044857646073575e-05,
      "loss": 1.9696,
      "step": 190630
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.4343314170837402,
      "learning_rate": 4.304274645482741e-05,
      "loss": 2.0822,
      "step": 190631
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.348862886428833,
      "learning_rate": 4.3040635311354544e-05,
      "loss": 2.3287,
      "step": 190632
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.2610310316085815,
      "learning_rate": 4.303852421565522e-05,
      "loss": 1.9913,
      "step": 190633
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.4712061882019043,
      "learning_rate": 4.303641316772992e-05,
      "loss": 1.9952,
      "step": 190634
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.509837031364441,
      "learning_rate": 4.3034302167579005e-05,
      "loss": 2.0893,
      "step": 190635
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.4159756898880005,
      "learning_rate": 4.303219121520281e-05,
      "loss": 1.9234,
      "step": 190636
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.4672409296035767,
      "learning_rate": 4.303008031060186e-05,
      "loss": 2.0768,
      "step": 190637
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.4081686735153198,
      "learning_rate": 4.302796945377646e-05,
      "loss": 1.9711,
      "step": 190638
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.377504825592041,
      "learning_rate": 4.3025858644726943e-05,
      "loss": 2.0206,
      "step": 190639
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.396061658859253,
      "learning_rate": 4.302374788345381e-05,
      "loss": 2.1759,
      "step": 190640
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.3835890293121338,
      "learning_rate": 4.302163716995742e-05,
      "loss": 2.182,
      "step": 190641
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.3134839534759521,
      "learning_rate": 4.301952650423812e-05,
      "loss": 2.1541,
      "step": 190642
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.5605860948562622,
      "learning_rate": 4.30174158862964e-05,
      "loss": 1.9291,
      "step": 190643
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.38559091091156,
      "learning_rate": 4.30153053161326e-05,
      "loss": 2.1306,
      "step": 190644
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.4683148860931396,
      "learning_rate": 4.3013194793747006e-05,
      "loss": 2.1634,
      "step": 190645
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.432116150856018,
      "learning_rate": 4.30110843191402e-05,
      "loss": 1.9555,
      "step": 190646
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.3775469064712524,
      "learning_rate": 4.30089738923124e-05,
      "loss": 1.8436,
      "step": 190647
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.3728787899017334,
      "learning_rate": 4.300686351326419e-05,
      "loss": 1.9724,
      "step": 190648
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.474280834197998,
      "learning_rate": 4.300475318199582e-05,
      "loss": 2.0002,
      "step": 190649
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.4417041540145874,
      "learning_rate": 4.300264289850773e-05,
      "loss": 2.1033,
      "step": 190650
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.3674408197402954,
      "learning_rate": 4.300053266280022e-05,
      "loss": 1.9745,
      "step": 190651
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.3143298625946045,
      "learning_rate": 4.299842247487383e-05,
      "loss": 2.0872,
      "step": 190652
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.5130916833877563,
      "learning_rate": 4.299631233472885e-05,
      "loss": 2.1349,
      "step": 190653
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.5007261037826538,
      "learning_rate": 4.299420224236574e-05,
      "loss": 1.9366,
      "step": 190654
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.482075572013855,
      "learning_rate": 4.299209219778488e-05,
      "loss": 1.9562,
      "step": 190655
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.4190504550933838,
      "learning_rate": 4.2989982200986625e-05,
      "loss": 2.0829,
      "step": 190656
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.476646900177002,
      "learning_rate": 4.298787225197132e-05,
      "loss": 1.9792,
      "step": 190657
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.3897937536239624,
      "learning_rate": 4.298576235073949e-05,
      "loss": 1.9095,
      "step": 190658
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.39585280418396,
      "learning_rate": 4.298365249729138e-05,
      "loss": 1.9969,
      "step": 190659
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.2999379634857178,
      "learning_rate": 4.298154269162757e-05,
      "loss": 2.0831,
      "step": 190660
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.4975693225860596,
      "learning_rate": 4.29794329337483e-05,
      "loss": 1.8828,
      "step": 190661
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.4894236326217651,
      "learning_rate": 4.297732322365402e-05,
      "loss": 1.8753,
      "step": 190662
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.380777359008789,
      "learning_rate": 4.2975213561345035e-05,
      "loss": 2.1216,
      "step": 190663
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.3300304412841797,
      "learning_rate": 4.297310394682186e-05,
      "loss": 1.8297,
      "step": 190664
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.495780110359192,
      "learning_rate": 4.29709943800848e-05,
      "loss": 1.8511,
      "step": 190665
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.3800387382507324,
      "learning_rate": 4.296888486113434e-05,
      "loss": 2.1784,
      "step": 190666
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.5149611234664917,
      "learning_rate": 4.296677538997083e-05,
      "loss": 2.036,
      "step": 190667
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.3766859769821167,
      "learning_rate": 4.2964665966594626e-05,
      "loss": 2.0867,
      "step": 190668
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.4390274286270142,
      "learning_rate": 4.296255659100609e-05,
      "loss": 2.0918,
      "step": 190669
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.3861268758773804,
      "learning_rate": 4.2960447263205735e-05,
      "loss": 2.1974,
      "step": 190670
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.28279447555542,
      "learning_rate": 4.295833798319378e-05,
      "loss": 1.7979,
      "step": 190671
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.4902303218841553,
      "learning_rate": 4.2956228750970836e-05,
      "loss": 1.7784,
      "step": 190672
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.46793532371521,
      "learning_rate": 4.295411956653716e-05,
      "loss": 2.0115,
      "step": 190673
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.293066143989563,
      "learning_rate": 4.2952010429893194e-05,
      "loss": 2.158,
      "step": 190674
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.3484777212142944,
      "learning_rate": 4.294990134103919e-05,
      "loss": 1.9485,
      "step": 190675
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.3398534059524536,
      "learning_rate": 4.2947792299975736e-05,
      "loss": 2.214,
      "step": 190676
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.6258859634399414,
      "learning_rate": 4.294568330670308e-05,
      "loss": 2.0063,
      "step": 190677
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.2842551469802856,
      "learning_rate": 4.2943574361221736e-05,
      "loss": 2.1387,
      "step": 190678
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.3067537546157837,
      "learning_rate": 4.2941465463532024e-05,
      "loss": 2.0347,
      "step": 190679
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.4778214693069458,
      "learning_rate": 4.293935661363438e-05,
      "loss": 1.9553,
      "step": 190680
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.4474188089370728,
      "learning_rate": 4.293724781152904e-05,
      "loss": 2.2305,
      "step": 190681
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.3993076086044312,
      "learning_rate": 4.293513905721664e-05,
      "loss": 1.9397,
      "step": 190682
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.316940426826477,
      "learning_rate": 4.293303035069734e-05,
      "loss": 1.8494,
      "step": 190683
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.454078197479248,
      "learning_rate": 4.2930921691971745e-05,
      "loss": 1.9438,
      "step": 190684
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.3877726793289185,
      "learning_rate": 4.292881308104011e-05,
      "loss": 1.9762,
      "step": 190685
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.3541547060012817,
      "learning_rate": 4.292670451790289e-05,
      "loss": 2.067,
      "step": 190686
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.3262062072753906,
      "learning_rate": 4.2924596002560365e-05,
      "loss": 2.0666,
      "step": 190687
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.3152440786361694,
      "learning_rate": 4.292248753501311e-05,
      "loss": 2.3012,
      "step": 190688
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.2866746187210083,
      "learning_rate": 4.2920379115261315e-05,
      "loss": 2.0096,
      "step": 190689
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.4277986288070679,
      "learning_rate": 4.2918270743305526e-05,
      "loss": 2.1274,
      "step": 190690
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.3504652976989746,
      "learning_rate": 4.2916162419146106e-05,
      "loss": 2.1213,
      "step": 190691
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.6602423191070557,
      "learning_rate": 4.2914054142783447e-05,
      "loss": 1.9598,
      "step": 190692
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.7250895500183105,
      "learning_rate": 4.2911945914217827e-05,
      "loss": 2.0094,
      "step": 190693
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.3674254417419434,
      "learning_rate": 4.29098377334498e-05,
      "loss": 1.8547,
      "step": 190694
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.3850089311599731,
      "learning_rate": 4.2907729600479615e-05,
      "loss": 2.0895,
      "step": 190695
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.4807863235473633,
      "learning_rate": 4.290562151530779e-05,
      "loss": 2.0991,
      "step": 190696
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.705810785293579,
      "learning_rate": 4.290351347793469e-05,
      "loss": 2.0175,
      "step": 190697
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.4518194198608398,
      "learning_rate": 4.2901405488360663e-05,
      "loss": 2.1566,
      "step": 190698
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.577836513519287,
      "learning_rate": 4.289929754658607e-05,
      "loss": 1.8968,
      "step": 190699
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.4259881973266602,
      "learning_rate": 4.28971896526114e-05,
      "loss": 2.0098,
      "step": 190700
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.3606584072113037,
      "learning_rate": 4.2895081806436925e-05,
      "loss": 1.8684,
      "step": 190701
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.5504884719848633,
      "learning_rate": 4.2892974008063184e-05,
      "loss": 2.0534,
      "step": 190702
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.5836559534072876,
      "learning_rate": 4.2890866257490406e-05,
      "loss": 2.062,
      "step": 190703
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.4759880304336548,
      "learning_rate": 4.288875855471923e-05,
      "loss": 1.8923,
      "step": 190704
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.7593953609466553,
      "learning_rate": 4.2886650899749744e-05,
      "loss": 1.9069,
      "step": 190705
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.3963860273361206,
      "learning_rate": 4.288454329258255e-05,
      "loss": 1.9527,
      "step": 190706
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.3794981241226196,
      "learning_rate": 4.288243573321793e-05,
      "loss": 1.9044,
      "step": 190707
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.430059790611267,
      "learning_rate": 4.288032822165636e-05,
      "loss": 1.8691,
      "step": 190708
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.5965301990509033,
      "learning_rate": 4.287822075789812e-05,
      "loss": 1.9141,
      "step": 190709
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.4809015989303589,
      "learning_rate": 4.287611334194374e-05,
      "loss": 1.8718,
      "step": 190710
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.4781962633132935,
      "learning_rate": 4.287400597379356e-05,
      "loss": 2.3546,
      "step": 190711
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.558563232421875,
      "learning_rate": 4.287189865344793e-05,
      "loss": 2.0454,
      "step": 190712
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.3397895097732544,
      "learning_rate": 4.28697913809072e-05,
      "loss": 2.1435,
      "step": 190713
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.3213634490966797,
      "learning_rate": 4.2867684156171936e-05,
      "loss": 2.1025,
      "step": 190714
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.477108120918274,
      "learning_rate": 4.286557697924232e-05,
      "loss": 2.302,
      "step": 190715
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.4097843170166016,
      "learning_rate": 4.2863469850118935e-05,
      "loss": 1.8923,
      "step": 190716
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.3392714262008667,
      "learning_rate": 4.2861362768802074e-05,
      "loss": 1.917,
      "step": 190717
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.4442461729049683,
      "learning_rate": 4.285925573529208e-05,
      "loss": 1.9147,
      "step": 190718
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.3846105337142944,
      "learning_rate": 4.285714874958946e-05,
      "loss": 1.8055,
      "step": 190719
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.4306985139846802,
      "learning_rate": 4.2855041811694544e-05,
      "loss": 1.8919,
      "step": 190720
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.5968694686889648,
      "learning_rate": 4.2852934921607684e-05,
      "loss": 2.1108,
      "step": 190721
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.283399224281311,
      "learning_rate": 4.285082807932938e-05,
      "loss": 1.8138,
      "step": 190722
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.495376706123352,
      "learning_rate": 4.2848721284859965e-05,
      "loss": 1.8433,
      "step": 190723
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.3183619976043701,
      "learning_rate": 4.284661453819974e-05,
      "loss": 1.8873,
      "step": 190724
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.4561541080474854,
      "learning_rate": 4.284450783934927e-05,
      "loss": 2.0427,
      "step": 190725
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.5457465648651123,
      "learning_rate": 4.284240118830886e-05,
      "loss": 2.1414,
      "step": 190726
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.6731207370758057,
      "learning_rate": 4.284029458507881e-05,
      "loss": 1.9993,
      "step": 190727
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.494456171989441,
      "learning_rate": 4.2838188029659705e-05,
      "loss": 2.032,
      "step": 190728
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.3973982334136963,
      "learning_rate": 4.283608152205179e-05,
      "loss": 2.1831,
      "step": 190729
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.4386086463928223,
      "learning_rate": 4.2833975062255467e-05,
      "loss": 2.025,
      "step": 190730
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.4453763961791992,
      "learning_rate": 4.283186865027123e-05,
      "loss": 2.0447,
      "step": 190731
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.4735074043273926,
      "learning_rate": 4.282976228609935e-05,
      "loss": 2.1797,
      "step": 190732
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.3461623191833496,
      "learning_rate": 4.282765596974031e-05,
      "loss": 2.008,
      "step": 190733
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.4499971866607666,
      "learning_rate": 4.282554970119447e-05,
      "loss": 2.0505,
      "step": 190734
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.2555583715438843,
      "learning_rate": 4.282344348046222e-05,
      "loss": 1.9181,
      "step": 190735
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.4527242183685303,
      "learning_rate": 4.282133730754388e-05,
      "loss": 1.8468,
      "step": 190736
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.3706731796264648,
      "learning_rate": 4.281923118243996e-05,
      "loss": 2.0639,
      "step": 190737
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.6498361825942993,
      "learning_rate": 4.2817125105150726e-05,
      "loss": 2.0237,
      "step": 190738
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.3633227348327637,
      "learning_rate": 4.281501907567675e-05,
      "loss": 1.9732,
      "step": 190739
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.4910222291946411,
      "learning_rate": 4.2812913094018286e-05,
      "loss": 1.7755,
      "step": 190740
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.4756159782409668,
      "learning_rate": 4.281080716017577e-05,
      "loss": 1.9856,
      "step": 190741
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.7095155715942383,
      "learning_rate": 4.2808701274149517e-05,
      "loss": 2.0082,
      "step": 190742
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.328306794166565,
      "learning_rate": 4.280659543594004e-05,
      "loss": 2.0742,
      "step": 190743
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.3298760652542114,
      "learning_rate": 4.280448964554758e-05,
      "loss": 2.2977,
      "step": 190744
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.3542603254318237,
      "learning_rate": 4.280238390297274e-05,
      "loss": 1.9353,
      "step": 190745
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.3609237670898438,
      "learning_rate": 4.2800278208215746e-05,
      "loss": 2.1206,
      "step": 190746
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.302001953125,
      "learning_rate": 4.279817256127707e-05,
      "loss": 1.9461,
      "step": 190747
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.347483515739441,
      "learning_rate": 4.279606696215697e-05,
      "loss": 1.984,
      "step": 190748
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.2761852741241455,
      "learning_rate": 4.2793961410856025e-05,
      "loss": 1.9909,
      "step": 190749
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.5266472101211548,
      "learning_rate": 4.279185590737446e-05,
      "loss": 2.1299,
      "step": 190750
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.3713873624801636,
      "learning_rate": 4.278975045171281e-05,
      "loss": 2.0389,
      "step": 190751
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.4647529125213623,
      "learning_rate": 4.278764504387141e-05,
      "loss": 2.2542,
      "step": 190752
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.4950284957885742,
      "learning_rate": 4.2785539683850626e-05,
      "loss": 2.0368,
      "step": 190753
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.4751393795013428,
      "learning_rate": 4.278343437165082e-05,
      "loss": 2.131,
      "step": 190754
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.430138111114502,
      "learning_rate": 4.278132910727249e-05,
      "loss": 2.0446,
      "step": 190755
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.4178831577301025,
      "learning_rate": 4.2779223890715874e-05,
      "loss": 2.1175,
      "step": 190756
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.5059888362884521,
      "learning_rate": 4.277711872198154e-05,
      "loss": 2.2986,
      "step": 190757
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.298606514930725,
      "learning_rate": 4.277501360106982e-05,
      "loss": 1.9813,
      "step": 190758
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.3835545778274536,
      "learning_rate": 4.277290852798104e-05,
      "loss": 2.0527,
      "step": 190759
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.411427617073059,
      "learning_rate": 4.2770803502715615e-05,
      "loss": 1.9688,
      "step": 190760
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.3282442092895508,
      "learning_rate": 4.276869852527396e-05,
      "loss": 2.1021,
      "step": 190761
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.3414946794509888,
      "learning_rate": 4.276659359565642e-05,
      "loss": 1.7311,
      "step": 190762
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.4042797088623047,
      "learning_rate": 4.276448871386352e-05,
      "loss": 1.9897,
      "step": 190763
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.3365225791931152,
      "learning_rate": 4.276238387989553e-05,
      "loss": 2.1685,
      "step": 190764
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.332890272140503,
      "learning_rate": 4.276027909375286e-05,
      "loss": 1.8352,
      "step": 190765
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.3665441274642944,
      "learning_rate": 4.275817435543586e-05,
      "loss": 1.8868,
      "step": 190766
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.415402889251709,
      "learning_rate": 4.275606966494504e-05,
      "loss": 1.8678,
      "step": 190767
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.4214247465133667,
      "learning_rate": 4.275396502228062e-05,
      "loss": 1.9815,
      "step": 190768
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.482911467552185,
      "learning_rate": 4.2751860427443187e-05,
      "loss": 2.2009,
      "step": 190769
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.475411295890808,
      "learning_rate": 4.2749755880432987e-05,
      "loss": 1.9577,
      "step": 190770
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.3086018562316895,
      "learning_rate": 4.274765138125057e-05,
      "loss": 1.9609,
      "step": 190771
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.4252115488052368,
      "learning_rate": 4.2745546929896126e-05,
      "loss": 1.9502,
      "step": 190772
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.7607827186584473,
      "learning_rate": 4.2743442526370156e-05,
      "loss": 2.1811,
      "step": 190773
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.371224284172058,
      "learning_rate": 4.274133817067299e-05,
      "loss": 2.002,
      "step": 190774
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.473309874534607,
      "learning_rate": 4.2739233862805164e-05,
      "loss": 1.9962,
      "step": 190775
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.3520845174789429,
      "learning_rate": 4.2737129602766854e-05,
      "loss": 2.0303,
      "step": 190776
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.537146806716919,
      "learning_rate": 4.273502539055874e-05,
      "loss": 2.2131,
      "step": 190777
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.4286012649536133,
      "learning_rate": 4.273292122618091e-05,
      "loss": 1.961,
      "step": 190778
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.3762202262878418,
      "learning_rate": 4.273081710963391e-05,
      "loss": 2.0448,
      "step": 190779
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.6428879499435425,
      "learning_rate": 4.272871304091809e-05,
      "loss": 1.9287,
      "step": 190780
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.514447569847107,
      "learning_rate": 4.2726609020033864e-05,
      "loss": 1.9591,
      "step": 190781
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.3990544080734253,
      "learning_rate": 4.272450504698162e-05,
      "loss": 1.9517,
      "step": 190782
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.8343000411987305,
      "learning_rate": 4.272240112176184e-05,
      "loss": 2.0364,
      "step": 190783
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.3319358825683594,
      "learning_rate": 4.27202972443747e-05,
      "loss": 2.0089,
      "step": 190784
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.4421069622039795,
      "learning_rate": 4.271819341482077e-05,
      "loss": 2.0429,
      "step": 190785
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.5704375505447388,
      "learning_rate": 4.2716089633100335e-05,
      "loss": 1.9215,
      "step": 190786
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.4169784784317017,
      "learning_rate": 4.2713985899213874e-05,
      "loss": 2.1064,
      "step": 190787
    },
    {
      "epoch": 2.48,
      "grad_norm": 2.204524517059326,
      "learning_rate": 4.2711882213161687e-05,
      "loss": 2.0562,
      "step": 190788
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.2947338819503784,
      "learning_rate": 4.2709778574944364e-05,
      "loss": 1.7897,
      "step": 190789
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.4363752603530884,
      "learning_rate": 4.2707674984562014e-05,
      "loss": 2.1792,
      "step": 190790
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.4946722984313965,
      "learning_rate": 4.2705571442015227e-05,
      "loss": 2.1003,
      "step": 190791
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.2584776878356934,
      "learning_rate": 4.2703467947304245e-05,
      "loss": 1.7988,
      "step": 190792
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.3480995893478394,
      "learning_rate": 4.270136450042963e-05,
      "loss": 2.0584,
      "step": 190793
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.7052947282791138,
      "learning_rate": 4.2699261101391626e-05,
      "loss": 2.0278,
      "step": 190794
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.51494300365448,
      "learning_rate": 4.2697157750190746e-05,
      "loss": 2.0849,
      "step": 190795
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.4417991638183594,
      "learning_rate": 4.269505444682734e-05,
      "loss": 1.8646,
      "step": 190796
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.342501163482666,
      "learning_rate": 4.269295119130176e-05,
      "loss": 1.868,
      "step": 190797
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.4020624160766602,
      "learning_rate": 4.269084798361435e-05,
      "loss": 1.8829,
      "step": 190798
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.3252086639404297,
      "learning_rate": 4.268874482376564e-05,
      "loss": 1.8764,
      "step": 190799
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.3870481252670288,
      "learning_rate": 4.268664171175587e-05,
      "loss": 1.8536,
      "step": 190800
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.4253369569778442,
      "learning_rate": 4.268453864758559e-05,
      "loss": 2.0705,
      "step": 190801
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.4147655963897705,
      "learning_rate": 4.2682435631255104e-05,
      "loss": 1.9156,
      "step": 190802
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.4452592134475708,
      "learning_rate": 4.2680332662764825e-05,
      "loss": 2.2878,
      "step": 190803
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.406821370124817,
      "learning_rate": 4.267822974211504e-05,
      "loss": 1.9441,
      "step": 190804
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.9068528413772583,
      "learning_rate": 4.267612686930632e-05,
      "loss": 1.9859,
      "step": 190805
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.3696537017822266,
      "learning_rate": 4.267402404433887e-05,
      "loss": 2.0296,
      "step": 190806
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.4684162139892578,
      "learning_rate": 4.2671921267213285e-05,
      "loss": 1.84,
      "step": 190807
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.323921799659729,
      "learning_rate": 4.266981853792982e-05,
      "loss": 2.089,
      "step": 190808
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.3675882816314697,
      "learning_rate": 4.266771585648883e-05,
      "loss": 2.0921,
      "step": 190809
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.3281867504119873,
      "learning_rate": 4.266561322289083e-05,
      "loss": 1.8892,
      "step": 190810
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.3921810388565063,
      "learning_rate": 4.266351063713615e-05,
      "loss": 2.0626,
      "step": 190811
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.3753889799118042,
      "learning_rate": 4.2661408099225146e-05,
      "loss": 2.18,
      "step": 190812
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.4933758974075317,
      "learning_rate": 4.2659305609158266e-05,
      "loss": 2.044,
      "step": 190813
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.6014039516448975,
      "learning_rate": 4.265720316693591e-05,
      "loss": 2.1603,
      "step": 190814
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.3545424938201904,
      "learning_rate": 4.265510077255836e-05,
      "loss": 2.1258,
      "step": 190815
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.3240801095962524,
      "learning_rate": 4.2652998426026165e-05,
      "loss": 1.8467,
      "step": 190816
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.466468334197998,
      "learning_rate": 4.265089612733953e-05,
      "loss": 2.0739,
      "step": 190817
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.3533450365066528,
      "learning_rate": 4.264879387649905e-05,
      "loss": 2.0301,
      "step": 190818
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.4294201135635376,
      "learning_rate": 4.264669167350501e-05,
      "loss": 2.0872,
      "step": 190819
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.5719079971313477,
      "learning_rate": 4.2644589518357816e-05,
      "loss": 2.1897,
      "step": 190820
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.3890275955200195,
      "learning_rate": 4.264248741105776e-05,
      "loss": 2.0449,
      "step": 190821
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.3692940473556519,
      "learning_rate": 4.264038535160542e-05,
      "loss": 1.8157,
      "step": 190822
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.4585108757019043,
      "learning_rate": 4.263828334000101e-05,
      "loss": 2.0821,
      "step": 190823
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.4397027492523193,
      "learning_rate": 4.263618137624506e-05,
      "loss": 2.0615,
      "step": 190824
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.376889705657959,
      "learning_rate": 4.26340794603379e-05,
      "loss": 2.0939,
      "step": 190825
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.544042706489563,
      "learning_rate": 4.263197759227993e-05,
      "loss": 2.2069,
      "step": 190826
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.4126410484313965,
      "learning_rate": 4.262987577207149e-05,
      "loss": 1.9051,
      "step": 190827
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.481610894203186,
      "learning_rate": 4.262777399971303e-05,
      "loss": 2.1016,
      "step": 190828
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.5498394966125488,
      "learning_rate": 4.2625672275204905e-05,
      "loss": 1.9916,
      "step": 190829
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.3138569593429565,
      "learning_rate": 4.2623570598547566e-05,
      "loss": 2.0256,
      "step": 190830
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.4062165021896362,
      "learning_rate": 4.262146896974138e-05,
      "loss": 1.9707,
      "step": 190831
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.5844415426254272,
      "learning_rate": 4.261936738878672e-05,
      "loss": 2.2239,
      "step": 190832
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.7644081115722656,
      "learning_rate": 4.261726585568391e-05,
      "loss": 2.0796,
      "step": 190833
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.535864233970642,
      "learning_rate": 4.261516437043346e-05,
      "loss": 1.9994,
      "step": 190834
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.337144374847412,
      "learning_rate": 4.261306293303567e-05,
      "loss": 1.9679,
      "step": 190835
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.3754053115844727,
      "learning_rate": 4.261096154349103e-05,
      "loss": 2.0262,
      "step": 190836
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.4240256547927856,
      "learning_rate": 4.2608860201799774e-05,
      "loss": 2.0444,
      "step": 190837
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.5823594331741333,
      "learning_rate": 4.2606758907962546e-05,
      "loss": 1.9021,
      "step": 190838
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.4238594770431519,
      "learning_rate": 4.2604657661979435e-05,
      "loss": 2.0917,
      "step": 190839
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.311719298362732,
      "learning_rate": 4.2602556463851075e-05,
      "loss": 1.9814,
      "step": 190840
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.4461641311645508,
      "learning_rate": 4.260045531357767e-05,
      "loss": 1.7669,
      "step": 190841
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.3759434223175049,
      "learning_rate": 4.259835421115976e-05,
      "loss": 1.9379,
      "step": 190842
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.2889297008514404,
      "learning_rate": 4.2596253156597624e-05,
      "loss": 2.111,
      "step": 190843
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.2762196063995361,
      "learning_rate": 4.259415214989181e-05,
      "loss": 2.108,
      "step": 190844
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.3077512979507446,
      "learning_rate": 4.259205119104249e-05,
      "loss": 1.942,
      "step": 190845
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.3327854871749878,
      "learning_rate": 4.258995028005024e-05,
      "loss": 1.9332,
      "step": 190846
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.4210675954818726,
      "learning_rate": 4.258784941691529e-05,
      "loss": 2.0337,
      "step": 190847
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.4416602849960327,
      "learning_rate": 4.2585748601638184e-05,
      "loss": 2.106,
      "step": 190848
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.510888695716858,
      "learning_rate": 4.2583647834219194e-05,
      "loss": 2.0437,
      "step": 190849
    },
    {
      "epoch": 2.48,
      "grad_norm": 1.5147731304168701,
      "learning_rate": 4.258154711465889e-05,
      "loss": 1.8577,
      "step": 190850
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.4745278358459473,
      "learning_rate": 4.25794464429574e-05,
      "loss": 1.8697,
      "step": 190851
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.443981647491455,
      "learning_rate": 4.257734581911533e-05,
      "loss": 2.0865,
      "step": 190852
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.522728681564331,
      "learning_rate": 4.257524524313291e-05,
      "loss": 1.9784,
      "step": 190853
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.5871518850326538,
      "learning_rate": 4.25731447150107e-05,
      "loss": 2.041,
      "step": 190854
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.4027328491210938,
      "learning_rate": 4.257104423474891e-05,
      "loss": 1.9601,
      "step": 190855
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.2486122846603394,
      "learning_rate": 4.256894380234817e-05,
      "loss": 1.9136,
      "step": 190856
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.3964920043945312,
      "learning_rate": 4.2566843417808574e-05,
      "loss": 2.009,
      "step": 190857
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.327004313468933,
      "learning_rate": 4.256474308113073e-05,
      "loss": 2.0819,
      "step": 190858
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.5342416763305664,
      "learning_rate": 4.2562642792314896e-05,
      "loss": 2.0575,
      "step": 190859
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.5137357711791992,
      "learning_rate": 4.256054255136161e-05,
      "loss": 2.0766,
      "step": 190860
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.3018617630004883,
      "learning_rate": 4.255844235827111e-05,
      "loss": 2.0999,
      "step": 190861
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.6183850765228271,
      "learning_rate": 4.255634221304398e-05,
      "loss": 2.0787,
      "step": 190862
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.8595054149627686,
      "learning_rate": 4.2554242115680335e-05,
      "loss": 2.0861,
      "step": 190863
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.5041736364364624,
      "learning_rate": 4.255214206618081e-05,
      "loss": 2.251,
      "step": 190864
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.511559247970581,
      "learning_rate": 4.255004206454562e-05,
      "loss": 2.0991,
      "step": 190865
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.40749990940094,
      "learning_rate": 4.254794211077528e-05,
      "loss": 2.1247,
      "step": 190866
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.5307319164276123,
      "learning_rate": 4.254584220487012e-05,
      "loss": 2.0511,
      "step": 190867
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.4932444095611572,
      "learning_rate": 4.2543742346830636e-05,
      "loss": 2.0468,
      "step": 190868
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.404579997062683,
      "learning_rate": 4.254164253665703e-05,
      "loss": 1.9841,
      "step": 190869
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.3128939867019653,
      "learning_rate": 4.253954277434983e-05,
      "loss": 1.8505,
      "step": 190870
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.345094084739685,
      "learning_rate": 4.253744305990935e-05,
      "loss": 1.8092,
      "step": 190871
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.4666739702224731,
      "learning_rate": 4.2535343393336077e-05,
      "loss": 2.1388,
      "step": 190872
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.402068018913269,
      "learning_rate": 4.253324377463028e-05,
      "loss": 2.225,
      "step": 190873
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.3692680597305298,
      "learning_rate": 4.253114420379255e-05,
      "loss": 1.8566,
      "step": 190874
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.3907241821289062,
      "learning_rate": 4.252904468082301e-05,
      "loss": 2.112,
      "step": 190875
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.450290322303772,
      "learning_rate": 4.252694520572221e-05,
      "loss": 1.9956,
      "step": 190876
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.4226053953170776,
      "learning_rate": 4.252484577849048e-05,
      "loss": 2.0214,
      "step": 190877
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.662839651107788,
      "learning_rate": 4.252274639912829e-05,
      "loss": 1.8139,
      "step": 190878
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.3128033876419067,
      "learning_rate": 4.252064706763595e-05,
      "loss": 2.3215,
      "step": 190879
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.3713104724884033,
      "learning_rate": 4.251854778401391e-05,
      "loss": 2.0196,
      "step": 190880
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.4796392917633057,
      "learning_rate": 4.251644854826255e-05,
      "loss": 1.7897,
      "step": 190881
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.315324068069458,
      "learning_rate": 4.251434936038223e-05,
      "loss": 1.882,
      "step": 190882
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.322440505027771,
      "learning_rate": 4.251225022037328e-05,
      "loss": 2.0142,
      "step": 190883
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.6747808456420898,
      "learning_rate": 4.251015112823627e-05,
      "loss": 2.4684,
      "step": 190884
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.3990379571914673,
      "learning_rate": 4.2508052083971366e-05,
      "loss": 1.9055,
      "step": 190885
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.331547498703003,
      "learning_rate": 4.250595308757917e-05,
      "loss": 2.1414,
      "step": 190886
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.3911633491516113,
      "learning_rate": 4.250385413905998e-05,
      "loss": 1.9286,
      "step": 190887
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.3486895561218262,
      "learning_rate": 4.250175523841419e-05,
      "loss": 2.1012,
      "step": 190888
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.368903636932373,
      "learning_rate": 4.249965638564211e-05,
      "loss": 1.9435,
      "step": 190889
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.340989112854004,
      "learning_rate": 4.249755758074427e-05,
      "loss": 2.0195,
      "step": 190890
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.471288800239563,
      "learning_rate": 4.24954588237209e-05,
      "loss": 2.0422,
      "step": 190891
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.5347013473510742,
      "learning_rate": 4.2493360114572596e-05,
      "loss": 2.2129,
      "step": 190892
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.394223690032959,
      "learning_rate": 4.249126145329963e-05,
      "loss": 1.9547,
      "step": 190893
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.526862621307373,
      "learning_rate": 4.24891628399023e-05,
      "loss": 1.859,
      "step": 190894
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.287774920463562,
      "learning_rate": 4.2487064274381175e-05,
      "loss": 1.94,
      "step": 190895
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.4803050756454468,
      "learning_rate": 4.248496575673659e-05,
      "loss": 1.9331,
      "step": 190896
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.309804081916809,
      "learning_rate": 4.2482867286968836e-05,
      "loss": 1.9071,
      "step": 190897
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.4880651235580444,
      "learning_rate": 4.2480768865078416e-05,
      "loss": 1.8427,
      "step": 190898
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.4258705377578735,
      "learning_rate": 4.24786704910657e-05,
      "loss": 1.8874,
      "step": 190899
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.607690691947937,
      "learning_rate": 4.247657216493102e-05,
      "loss": 1.9769,
      "step": 190900
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.4144150018692017,
      "learning_rate": 4.247447388667484e-05,
      "loss": 1.8999,
      "step": 190901
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.482461929321289,
      "learning_rate": 4.247237565629747e-05,
      "loss": 2.0181,
      "step": 190902
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.4016183614730835,
      "learning_rate": 4.247027747379939e-05,
      "loss": 1.9737,
      "step": 190903
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.2971868515014648,
      "learning_rate": 4.246817933918097e-05,
      "loss": 1.8046,
      "step": 190904
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.3797998428344727,
      "learning_rate": 4.246608125244256e-05,
      "loss": 1.9225,
      "step": 190905
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.380188226699829,
      "learning_rate": 4.246398321358452e-05,
      "loss": 1.9891,
      "step": 190906
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.3962053060531616,
      "learning_rate": 4.246188522260734e-05,
      "loss": 2.0788,
      "step": 190907
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.3894771337509155,
      "learning_rate": 4.245978727951126e-05,
      "loss": 2.0511,
      "step": 190908
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.3671518564224243,
      "learning_rate": 4.245768938429688e-05,
      "loss": 1.9564,
      "step": 190909
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.441127896308899,
      "learning_rate": 4.2455591536964464e-05,
      "loss": 1.9723,
      "step": 190910
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.4283697605133057,
      "learning_rate": 4.245349373751441e-05,
      "loss": 2.084,
      "step": 190911
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.4676053524017334,
      "learning_rate": 4.245139598594705e-05,
      "loss": 2.0323,
      "step": 190912
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.6078912019729614,
      "learning_rate": 4.24492982822629e-05,
      "loss": 1.8698,
      "step": 190913
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.3471660614013672,
      "learning_rate": 4.244720062646221e-05,
      "loss": 1.8373,
      "step": 190914
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.4240049123764038,
      "learning_rate": 4.244510301854555e-05,
      "loss": 1.9872,
      "step": 190915
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.4280270338058472,
      "learning_rate": 4.244300545851311e-05,
      "loss": 2.1299,
      "step": 190916
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.3588287830352783,
      "learning_rate": 4.244090794636551e-05,
      "loss": 2.0291,
      "step": 190917
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.4172203540802002,
      "learning_rate": 4.243881048210287e-05,
      "loss": 2.0723,
      "step": 190918
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.3663287162780762,
      "learning_rate": 4.2436713065725795e-05,
      "loss": 1.9314,
      "step": 190919
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.5479294061660767,
      "learning_rate": 4.243461569723454e-05,
      "loss": 1.851,
      "step": 190920
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.7937220335006714,
      "learning_rate": 4.243251837662962e-05,
      "loss": 2.0053,
      "step": 190921
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.4688211679458618,
      "learning_rate": 4.243042110391126e-05,
      "loss": 2.1378,
      "step": 190922
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.3853868246078491,
      "learning_rate": 4.24283238790801e-05,
      "loss": 2.0974,
      "step": 190923
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.4581907987594604,
      "learning_rate": 4.2426226702136254e-05,
      "loss": 2.1869,
      "step": 190924
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.3519222736358643,
      "learning_rate": 4.242412957308028e-05,
      "loss": 1.9634,
      "step": 190925
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.3613357543945312,
      "learning_rate": 4.242203249191246e-05,
      "loss": 2.0523,
      "step": 190926
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.5750491619110107,
      "learning_rate": 4.2419935458633334e-05,
      "loss": 1.8531,
      "step": 190927
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.3290138244628906,
      "learning_rate": 4.2417838473243104e-05,
      "loss": 1.9957,
      "step": 190928
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.367597222328186,
      "learning_rate": 4.241574153574243e-05,
      "loss": 1.9642,
      "step": 190929
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.3389678001403809,
      "learning_rate": 4.2413644646131386e-05,
      "loss": 2.0625,
      "step": 190930
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.3788878917694092,
      "learning_rate": 4.241154780441056e-05,
      "loss": 2.1514,
      "step": 190931
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.4891122579574585,
      "learning_rate": 4.240945101058023e-05,
      "loss": 1.8444,
      "step": 190932
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.6438779830932617,
      "learning_rate": 4.240735426464089e-05,
      "loss": 2.0005,
      "step": 190933
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.3361916542053223,
      "learning_rate": 4.240525756659284e-05,
      "loss": 1.9672,
      "step": 190934
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.3651549816131592,
      "learning_rate": 4.240316091643665e-05,
      "loss": 1.9458,
      "step": 190935
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.3795504570007324,
      "learning_rate": 4.240106431417245e-05,
      "loss": 1.9513,
      "step": 190936
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.387585163116455,
      "learning_rate": 4.239896775980081e-05,
      "loss": 1.6938,
      "step": 190937
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.3930234909057617,
      "learning_rate": 4.2396871253322016e-05,
      "loss": 2.0219,
      "step": 190938
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.4025338888168335,
      "learning_rate": 4.2394774794736555e-05,
      "loss": 2.0438,
      "step": 190939
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.4393528699874878,
      "learning_rate": 4.239267838404471e-05,
      "loss": 1.9051,
      "step": 190940
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.3946865797042847,
      "learning_rate": 4.2390582021247053e-05,
      "loss": 1.9105,
      "step": 190941
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.6908048391342163,
      "learning_rate": 4.2388485706343724e-05,
      "loss": 1.7805,
      "step": 190942
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.4970346689224243,
      "learning_rate": 4.2386389439335276e-05,
      "loss": 1.7855,
      "step": 190943
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.3507366180419922,
      "learning_rate": 4.238429322022202e-05,
      "loss": 2.1193,
      "step": 190944
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.3902006149291992,
      "learning_rate": 4.2382197049004455e-05,
      "loss": 2.1398,
      "step": 190945
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.4879448413848877,
      "learning_rate": 4.2380100925682834e-05,
      "loss": 2.0339,
      "step": 190946
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.391966700553894,
      "learning_rate": 4.237800485025774e-05,
      "loss": 2.1761,
      "step": 190947
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.4411789178848267,
      "learning_rate": 4.237590882272933e-05,
      "loss": 2.1017,
      "step": 190948
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.4340417385101318,
      "learning_rate": 4.2373812843098144e-05,
      "loss": 2.1689,
      "step": 190949
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.4534050226211548,
      "learning_rate": 4.237171691136447e-05,
      "loss": 2.0877,
      "step": 190950
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.68769109249115,
      "learning_rate": 4.236962102752882e-05,
      "loss": 1.9181,
      "step": 190951
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.4613754749298096,
      "learning_rate": 4.236752519159146e-05,
      "loss": 2.1423,
      "step": 190952
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.305053949356079,
      "learning_rate": 4.2365429403552976e-05,
      "loss": 1.8915,
      "step": 190953
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.3485444784164429,
      "learning_rate": 4.236333366341348e-05,
      "loss": 1.8422,
      "step": 190954
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.6466048955917358,
      "learning_rate": 4.2361237971173566e-05,
      "loss": 2.1132,
      "step": 190955
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.4292187690734863,
      "learning_rate": 4.2359142326833504e-05,
      "loss": 2.0329,
      "step": 190956
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.4907087087631226,
      "learning_rate": 4.2357046730393793e-05,
      "loss": 1.9684,
      "step": 190957
    },
    {
      "epoch": 2.49,
      "grad_norm": 2.1443030834198,
      "learning_rate": 4.235495118185474e-05,
      "loss": 2.0992,
      "step": 190958
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.4774903059005737,
      "learning_rate": 4.235285568121686e-05,
      "loss": 2.2148,
      "step": 190959
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.2934508323669434,
      "learning_rate": 4.235076022848033e-05,
      "loss": 2.0429,
      "step": 190960
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.298396110534668,
      "learning_rate": 4.2348664823645715e-05,
      "loss": 2.073,
      "step": 190961
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.3995229005813599,
      "learning_rate": 4.2346569466713284e-05,
      "loss": 2.1394,
      "step": 190962
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.4546070098876953,
      "learning_rate": 4.2344474157683575e-05,
      "loss": 2.0925,
      "step": 190963
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.4195137023925781,
      "learning_rate": 4.2342378896556805e-05,
      "loss": 2.0776,
      "step": 190964
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.4357675313949585,
      "learning_rate": 4.234028368333359e-05,
      "loss": 2.094,
      "step": 190965
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.4161570072174072,
      "learning_rate": 4.233818851801405e-05,
      "loss": 1.8701,
      "step": 190966
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.3683348894119263,
      "learning_rate": 4.233609340059876e-05,
      "loss": 2.1327,
      "step": 190967
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.4191559553146362,
      "learning_rate": 4.233399833108798e-05,
      "loss": 2.1388,
      "step": 190968
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.4490420818328857,
      "learning_rate": 4.233190330948225e-05,
      "loss": 1.8507,
      "step": 190969
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.4420194625854492,
      "learning_rate": 4.2329808335781835e-05,
      "loss": 2.1018,
      "step": 190970
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.4293137788772583,
      "learning_rate": 4.232771340998723e-05,
      "loss": 1.8073,
      "step": 190971
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.3883936405181885,
      "learning_rate": 4.2325618532098734e-05,
      "loss": 1.6621,
      "step": 190972
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.4300719499588013,
      "learning_rate": 4.232352370211682e-05,
      "loss": 1.9813,
      "step": 190973
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.3803621530532837,
      "learning_rate": 4.232142892004172e-05,
      "loss": 1.9597,
      "step": 190974
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.3620562553405762,
      "learning_rate": 4.2319334185874034e-05,
      "loss": 1.9232,
      "step": 190975
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.4389351606369019,
      "learning_rate": 4.2317239499613956e-05,
      "loss": 2.0626,
      "step": 190976
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.381556749343872,
      "learning_rate": 4.231514486126202e-05,
      "loss": 1.9768,
      "step": 190977
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.4668550491333008,
      "learning_rate": 4.2313050270818595e-05,
      "loss": 2.0749,
      "step": 190978
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.3485783338546753,
      "learning_rate": 4.231095572828395e-05,
      "loss": 1.9163,
      "step": 190979
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.4583700895309448,
      "learning_rate": 4.2308861233658644e-05,
      "loss": 1.9245,
      "step": 190980
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.359325647354126,
      "learning_rate": 4.230676678694298e-05,
      "loss": 2.0964,
      "step": 190981
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.3161048889160156,
      "learning_rate": 4.230467238813726e-05,
      "loss": 1.9108,
      "step": 190982
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.3816156387329102,
      "learning_rate": 4.2302578037242084e-05,
      "loss": 2.0225,
      "step": 190983
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.4444218873977661,
      "learning_rate": 4.2300483734257686e-05,
      "loss": 2.0117,
      "step": 190984
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.4734773635864258,
      "learning_rate": 4.2298389479184426e-05,
      "loss": 1.9584,
      "step": 190985
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.460547685623169,
      "learning_rate": 4.229629527202284e-05,
      "loss": 1.9993,
      "step": 190986
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.3186912536621094,
      "learning_rate": 4.229420111277323e-05,
      "loss": 1.8706,
      "step": 190987
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.507834553718567,
      "learning_rate": 4.2292107001435924e-05,
      "loss": 2.3258,
      "step": 190988
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.3327293395996094,
      "learning_rate": 4.2290012938011465e-05,
      "loss": 1.8727,
      "step": 190989
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.3512489795684814,
      "learning_rate": 4.228791892250014e-05,
      "loss": 1.6609,
      "step": 190990
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.2895630598068237,
      "learning_rate": 4.2285824954902295e-05,
      "loss": 1.9217,
      "step": 190991
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.4357229471206665,
      "learning_rate": 4.228373103521845e-05,
      "loss": 2.0167,
      "step": 190992
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.341680645942688,
      "learning_rate": 4.2281637163448844e-05,
      "loss": 2.0218,
      "step": 190993
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.3769317865371704,
      "learning_rate": 4.227954333959401e-05,
      "loss": 2.0688,
      "step": 190994
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.3613314628601074,
      "learning_rate": 4.2277449563654284e-05,
      "loss": 2.0709,
      "step": 190995
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.323093295097351,
      "learning_rate": 4.227535583563006e-05,
      "loss": 1.8371,
      "step": 190996
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.3556395769119263,
      "learning_rate": 4.227326215552165e-05,
      "loss": 2.0749,
      "step": 190997
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.387511134147644,
      "learning_rate": 4.227116852332954e-05,
      "loss": 1.9442,
      "step": 190998
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.355299472808838,
      "learning_rate": 4.2269074939054024e-05,
      "loss": 2.0903,
      "step": 190999
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.317541480064392,
      "learning_rate": 4.226698140269562e-05,
      "loss": 2.0743,
      "step": 191000
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.4170626401901245,
      "learning_rate": 4.226488791425459e-05,
      "loss": 1.9433,
      "step": 191001
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.347274899482727,
      "learning_rate": 4.226279447373152e-05,
      "loss": 2.045,
      "step": 191002
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.4776191711425781,
      "learning_rate": 4.226070108112652e-05,
      "loss": 2.2141,
      "step": 191003
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.3106307983398438,
      "learning_rate": 4.22586077364402e-05,
      "loss": 1.9482,
      "step": 191004
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.32697594165802,
      "learning_rate": 4.22565144396728e-05,
      "loss": 1.907,
      "step": 191005
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.3821290731430054,
      "learning_rate": 4.225442119082482e-05,
      "loss": 1.9441,
      "step": 191006
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.5331095457077026,
      "learning_rate": 4.225232798989656e-05,
      "loss": 2.0002,
      "step": 191007
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.3741697072982788,
      "learning_rate": 4.225023483688861e-05,
      "loss": 2.1014,
      "step": 191008
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.4594788551330566,
      "learning_rate": 4.224814173180105e-05,
      "loss": 1.9202,
      "step": 191009
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.47273850440979,
      "learning_rate": 4.22460486746345e-05,
      "loss": 2.1313,
      "step": 191010
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.3969552516937256,
      "learning_rate": 4.224395566538922e-05,
      "loss": 1.9765,
      "step": 191011
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.382237195968628,
      "learning_rate": 4.22418627040657e-05,
      "loss": 1.9839,
      "step": 191012
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.4273611307144165,
      "learning_rate": 4.2239769790664216e-05,
      "loss": 2.0451,
      "step": 191013
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.420677661895752,
      "learning_rate": 4.223767692518536e-05,
      "loss": 2.0427,
      "step": 191014
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.4690649509429932,
      "learning_rate": 4.2235584107629274e-05,
      "loss": 1.8899,
      "step": 191015
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.3776320219039917,
      "learning_rate": 4.2233491337996514e-05,
      "loss": 1.8854,
      "step": 191016
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.400547742843628,
      "learning_rate": 4.2231398616287327e-05,
      "loss": 2.0872,
      "step": 191017
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.2936880588531494,
      "learning_rate": 4.222930594250227e-05,
      "loss": 2.1352,
      "step": 191018
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.3927175998687744,
      "learning_rate": 4.222721331664161e-05,
      "loss": 1.8216,
      "step": 191019
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.3788206577301025,
      "learning_rate": 4.2225120738705875e-05,
      "loss": 2.1775,
      "step": 191020
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.419293761253357,
      "learning_rate": 4.222302820869524e-05,
      "loss": 1.9214,
      "step": 191021
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.4873108863830566,
      "learning_rate": 4.222093572661027e-05,
      "loss": 1.9972,
      "step": 191022
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.4492135047912598,
      "learning_rate": 4.22188432924512e-05,
      "loss": 2.1205,
      "step": 191023
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.370478868484497,
      "learning_rate": 4.2216750906218624e-05,
      "loss": 2.0192,
      "step": 191024
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.4431337118148804,
      "learning_rate": 4.221465856791274e-05,
      "loss": 2.0262,
      "step": 191025
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.482486367225647,
      "learning_rate": 4.221256627753416e-05,
      "loss": 1.9977,
      "step": 191026
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.3278875350952148,
      "learning_rate": 4.221047403508297e-05,
      "loss": 1.9745,
      "step": 191027
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.252203106880188,
      "learning_rate": 4.22083818405598e-05,
      "loss": 1.9298,
      "step": 191028
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.3612955808639526,
      "learning_rate": 4.220628969396487e-05,
      "loss": 1.6646,
      "step": 191029
    },
    {
      "epoch": 2.49,
      "grad_norm": 2.6536662578582764,
      "learning_rate": 4.2204197595298764e-05,
      "loss": 1.988,
      "step": 191030
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.493680715560913,
      "learning_rate": 4.2202105544561684e-05,
      "loss": 2.1092,
      "step": 191031
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.502211570739746,
      "learning_rate": 4.2200013541754194e-05,
      "loss": 1.8287,
      "step": 191032
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.2970529794692993,
      "learning_rate": 4.2197921586876504e-05,
      "loss": 1.8581,
      "step": 191033
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.4565036296844482,
      "learning_rate": 4.219582967992914e-05,
      "loss": 2.0172,
      "step": 191034
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.2577886581420898,
      "learning_rate": 4.2193737820912366e-05,
      "loss": 2.1035,
      "step": 191035
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.3858691453933716,
      "learning_rate": 4.219164600982672e-05,
      "loss": 1.999,
      "step": 191036
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.3861933946609497,
      "learning_rate": 4.218955424667243e-05,
      "loss": 1.9181,
      "step": 191037
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.6139366626739502,
      "learning_rate": 4.218746253145003e-05,
      "loss": 1.8734,
      "step": 191038
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.365155577659607,
      "learning_rate": 4.218537086415986e-05,
      "loss": 1.8542,
      "step": 191039
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.322237253189087,
      "learning_rate": 4.218327924480228e-05,
      "loss": 2.0953,
      "step": 191040
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.475320816040039,
      "learning_rate": 4.218118767337766e-05,
      "loss": 2.2558,
      "step": 191041
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.3884012699127197,
      "learning_rate": 4.217909614988646e-05,
      "loss": 2.0573,
      "step": 191042
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.4182056188583374,
      "learning_rate": 4.2177004674328984e-05,
      "loss": 1.7913,
      "step": 191043
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.2369987964630127,
      "learning_rate": 4.2174913246705736e-05,
      "loss": 1.7953,
      "step": 191044
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.4062849283218384,
      "learning_rate": 4.217282186701704e-05,
      "loss": 2.3742,
      "step": 191045
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.5760157108306885,
      "learning_rate": 4.217073053526326e-05,
      "loss": 2.0799,
      "step": 191046
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.4446488618850708,
      "learning_rate": 4.216863925144478e-05,
      "loss": 2.1553,
      "step": 191047
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.5289314985275269,
      "learning_rate": 4.216654801556205e-05,
      "loss": 2.2034,
      "step": 191048
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.4142684936523438,
      "learning_rate": 4.216445682761538e-05,
      "loss": 1.9961,
      "step": 191049
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.3159127235412598,
      "learning_rate": 4.216236568760526e-05,
      "loss": 1.9381,
      "step": 191050
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.378966212272644,
      "learning_rate": 4.216027459553203e-05,
      "loss": 2.0207,
      "step": 191051
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.3582972288131714,
      "learning_rate": 4.2158183551396086e-05,
      "loss": 2.1725,
      "step": 191052
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.3915939331054688,
      "learning_rate": 4.21560925551977e-05,
      "loss": 1.9405,
      "step": 191053
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.5256527662277222,
      "learning_rate": 4.215400160693746e-05,
      "loss": 2.0964,
      "step": 191054
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.3578569889068604,
      "learning_rate": 4.215191070661558e-05,
      "loss": 1.8442,
      "step": 191055
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.492867350578308,
      "learning_rate": 4.214981985423258e-05,
      "loss": 2.157,
      "step": 191056
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.4741071462631226,
      "learning_rate": 4.21477290497888e-05,
      "loss": 1.9085,
      "step": 191057
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.3244516849517822,
      "learning_rate": 4.2145638293284645e-05,
      "loss": 1.9653,
      "step": 191058
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.3927134275436401,
      "learning_rate": 4.2143547584720405e-05,
      "loss": 2.108,
      "step": 191059
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.3640196323394775,
      "learning_rate": 4.214145692409662e-05,
      "loss": 2.0825,
      "step": 191060
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.4631849527359009,
      "learning_rate": 4.2139366311413546e-05,
      "loss": 1.9416,
      "step": 191061
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.3210530281066895,
      "learning_rate": 4.2137275746671665e-05,
      "loss": 2.0552,
      "step": 191062
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.344136118888855,
      "learning_rate": 4.213518522987136e-05,
      "loss": 1.8679,
      "step": 191063
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.4993685483932495,
      "learning_rate": 4.213309476101291e-05,
      "loss": 2.0421,
      "step": 191064
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.3907254934310913,
      "learning_rate": 4.213100434009687e-05,
      "loss": 1.9522,
      "step": 191065
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.367186188697815,
      "learning_rate": 4.212891396712352e-05,
      "loss": 1.8784,
      "step": 191066
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.3255418539047241,
      "learning_rate": 4.212682364209322e-05,
      "loss": 1.8559,
      "step": 191067
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.5659661293029785,
      "learning_rate": 4.2124733365006504e-05,
      "loss": 1.9871,
      "step": 191068
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.6713430881500244,
      "learning_rate": 4.212264313586363e-05,
      "loss": 1.8335,
      "step": 191069
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.744299292564392,
      "learning_rate": 4.2120552954664977e-05,
      "loss": 1.931,
      "step": 191070
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.3353151082992554,
      "learning_rate": 4.211846282141104e-05,
      "loss": 1.9955,
      "step": 191071
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.3237844705581665,
      "learning_rate": 4.2116372736102144e-05,
      "loss": 1.8925,
      "step": 191072
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.317431926727295,
      "learning_rate": 4.2114282698738635e-05,
      "loss": 2.001,
      "step": 191073
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.528524398803711,
      "learning_rate": 4.2112192709320966e-05,
      "loss": 2.1253,
      "step": 191074
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.4743906259536743,
      "learning_rate": 4.211010276784955e-05,
      "loss": 2.233,
      "step": 191075
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.4404268264770508,
      "learning_rate": 4.2108012874324684e-05,
      "loss": 2.2273,
      "step": 191076
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.3060302734375,
      "learning_rate": 4.2105923028746823e-05,
      "loss": 2.0496,
      "step": 191077
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.442185640335083,
      "learning_rate": 4.210383323111631e-05,
      "loss": 1.9615,
      "step": 191078
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.4033253192901611,
      "learning_rate": 4.210174348143361e-05,
      "loss": 1.7678,
      "step": 191079
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.3679702281951904,
      "learning_rate": 4.20996537796991e-05,
      "loss": 1.9647,
      "step": 191080
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.5473883152008057,
      "learning_rate": 4.2097564125913084e-05,
      "loss": 2.0208,
      "step": 191081
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.460209846496582,
      "learning_rate": 4.209547452007596e-05,
      "loss": 2.2264,
      "step": 191082
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.4483954906463623,
      "learning_rate": 4.2093384962188203e-05,
      "loss": 1.9863,
      "step": 191083
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.4928163290023804,
      "learning_rate": 4.2091295452250095e-05,
      "loss": 1.8562,
      "step": 191084
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.3161863088607788,
      "learning_rate": 4.2089205990262164e-05,
      "loss": 2.1785,
      "step": 191085
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.3393723964691162,
      "learning_rate": 4.208711657622471e-05,
      "loss": 2.0865,
      "step": 191086
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.3720550537109375,
      "learning_rate": 4.2085027210138135e-05,
      "loss": 1.8596,
      "step": 191087
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.532724142074585,
      "learning_rate": 4.208293789200273e-05,
      "loss": 1.9422,
      "step": 191088
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.447701334953308,
      "learning_rate": 4.2080848621819074e-05,
      "loss": 1.973,
      "step": 191089
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.4240416288375854,
      "learning_rate": 4.207875939958739e-05,
      "loss": 2.1927,
      "step": 191090
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.383131980895996,
      "learning_rate": 4.207667022530818e-05,
      "loss": 2.2868,
      "step": 191091
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.2881025075912476,
      "learning_rate": 4.2074581098981716e-05,
      "loss": 1.9822,
      "step": 191092
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.390366554260254,
      "learning_rate": 4.2072492020608595e-05,
      "loss": 1.9637,
      "step": 191093
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.42732572555542,
      "learning_rate": 4.2070402990188944e-05,
      "loss": 2.2336,
      "step": 191094
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.2989656925201416,
      "learning_rate": 4.206831400772334e-05,
      "loss": 1.9394,
      "step": 191095
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.2937824726104736,
      "learning_rate": 4.2066225073212035e-05,
      "loss": 1.7478,
      "step": 191096
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.5159471035003662,
      "learning_rate": 4.206413618665554e-05,
      "loss": 2.1011,
      "step": 191097
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.4667787551879883,
      "learning_rate": 4.206204734805415e-05,
      "loss": 1.9015,
      "step": 191098
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.4781928062438965,
      "learning_rate": 4.20599585574084e-05,
      "loss": 1.9719,
      "step": 191099
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.4810905456542969,
      "learning_rate": 4.205786981471846e-05,
      "loss": 1.9853,
      "step": 191100
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.4889118671417236,
      "learning_rate": 4.2055781119984886e-05,
      "loss": 2.1692,
      "step": 191101
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.5001791715621948,
      "learning_rate": 4.205369247320796e-05,
      "loss": 1.9739,
      "step": 191102
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.3962970972061157,
      "learning_rate": 4.2051603874388165e-05,
      "loss": 2.0367,
      "step": 191103
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.306903600692749,
      "learning_rate": 4.2049515323525805e-05,
      "loss": 1.9463,
      "step": 191104
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.4785544872283936,
      "learning_rate": 4.204742682062135e-05,
      "loss": 1.9635,
      "step": 191105
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.4673632383346558,
      "learning_rate": 4.2045338365675175e-05,
      "loss": 1.8973,
      "step": 191106
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.3503916263580322,
      "learning_rate": 4.204324995868763e-05,
      "loss": 1.931,
      "step": 191107
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.3769382238388062,
      "learning_rate": 4.204116159965902e-05,
      "loss": 2.186,
      "step": 191108
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.5987404584884644,
      "learning_rate": 4.2039073288589954e-05,
      "loss": 2.1709,
      "step": 191109
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.5119014978408813,
      "learning_rate": 4.2036985025480584e-05,
      "loss": 2.0715,
      "step": 191110
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.4594358205795288,
      "learning_rate": 4.2034896810331486e-05,
      "loss": 2.0338,
      "step": 191111
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.3548305034637451,
      "learning_rate": 4.203280864314296e-05,
      "loss": 2.3147,
      "step": 191112
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.3808958530426025,
      "learning_rate": 4.2030720523915396e-05,
      "loss": 1.9992,
      "step": 191113
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.3234604597091675,
      "learning_rate": 4.2028632452649135e-05,
      "loss": 2.1474,
      "step": 191114
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.4178245067596436,
      "learning_rate": 4.202654442934471e-05,
      "loss": 2.178,
      "step": 191115
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.5333251953125,
      "learning_rate": 4.202445645400232e-05,
      "loss": 2.2707,
      "step": 191116
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.5592255592346191,
      "learning_rate": 4.202236852662253e-05,
      "loss": 2.092,
      "step": 191117
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.5259960889816284,
      "learning_rate": 4.202028064720567e-05,
      "loss": 1.9916,
      "step": 191118
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.6020132303237915,
      "learning_rate": 4.201819281575208e-05,
      "loss": 2.1135,
      "step": 191119
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.59782874584198,
      "learning_rate": 4.201610503226213e-05,
      "loss": 2.1574,
      "step": 191120
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.3921411037445068,
      "learning_rate": 4.201401729673634e-05,
      "loss": 1.96,
      "step": 191121
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.7138532400131226,
      "learning_rate": 4.201192960917491e-05,
      "loss": 1.9334,
      "step": 191122
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.3345519304275513,
      "learning_rate": 4.200984196957842e-05,
      "loss": 2.0631,
      "step": 191123
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.8799724578857422,
      "learning_rate": 4.2007754377947153e-05,
      "loss": 2.0668,
      "step": 191124
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.511264443397522,
      "learning_rate": 4.200566683428153e-05,
      "loss": 2.0079,
      "step": 191125
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.3783535957336426,
      "learning_rate": 4.200357933858186e-05,
      "loss": 2.1473,
      "step": 191126
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.361340880393982,
      "learning_rate": 4.200149189084866e-05,
      "loss": 2.2045,
      "step": 191127
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.2754932641983032,
      "learning_rate": 4.199940449108219e-05,
      "loss": 1.8801,
      "step": 191128
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.357387661933899,
      "learning_rate": 4.199731713928295e-05,
      "loss": 1.9928,
      "step": 191129
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.5250883102416992,
      "learning_rate": 4.199522983545127e-05,
      "loss": 2.2004,
      "step": 191130
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.4435113668441772,
      "learning_rate": 4.199314257958759e-05,
      "loss": 1.9728,
      "step": 191131
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.3729467391967773,
      "learning_rate": 4.199105537169214e-05,
      "loss": 2.0797,
      "step": 191132
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.4727529287338257,
      "learning_rate": 4.1988968211765514e-05,
      "loss": 2.3204,
      "step": 191133
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.4963587522506714,
      "learning_rate": 4.1986881099807956e-05,
      "loss": 2.0048,
      "step": 191134
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.3203052282333374,
      "learning_rate": 4.1984794035819955e-05,
      "loss": 1.8701,
      "step": 191135
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.3747495412826538,
      "learning_rate": 4.1982707019801854e-05,
      "loss": 2.1521,
      "step": 191136
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.6816452741622925,
      "learning_rate": 4.198062005175404e-05,
      "loss": 1.9876,
      "step": 191137
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.290177583694458,
      "learning_rate": 4.197853313167683e-05,
      "loss": 1.9277,
      "step": 191138
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.38698148727417,
      "learning_rate": 4.197644625957074e-05,
      "loss": 2.001,
      "step": 191139
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.362438678741455,
      "learning_rate": 4.197435943543605e-05,
      "loss": 1.8054,
      "step": 191140
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.4657024145126343,
      "learning_rate": 4.1972272659273276e-05,
      "loss": 2.0141,
      "step": 191141
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.5838520526885986,
      "learning_rate": 4.19701859310827e-05,
      "loss": 2.1924,
      "step": 191142
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.4921796321868896,
      "learning_rate": 4.196809925086475e-05,
      "loss": 1.9804,
      "step": 191143
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.3866877555847168,
      "learning_rate": 4.196601261861973e-05,
      "loss": 1.8411,
      "step": 191144
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.3536145687103271,
      "learning_rate": 4.196392603434816e-05,
      "loss": 1.963,
      "step": 191145
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.493652582168579,
      "learning_rate": 4.1961839498050286e-05,
      "loss": 2.2127,
      "step": 191146
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.295578122138977,
      "learning_rate": 4.195975300972667e-05,
      "loss": 2.0942,
      "step": 191147
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.4224799871444702,
      "learning_rate": 4.1957666569377614e-05,
      "loss": 1.9333,
      "step": 191148
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.490761637687683,
      "learning_rate": 4.195558017700348e-05,
      "loss": 1.6908,
      "step": 191149
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.4318641424179077,
      "learning_rate": 4.19534938326046e-05,
      "loss": 2.117,
      "step": 191150
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.3655742406845093,
      "learning_rate": 4.195140753618152e-05,
      "loss": 2.0623,
      "step": 191151
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.3055897951126099,
      "learning_rate": 4.194932128773448e-05,
      "loss": 1.8022,
      "step": 191152
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.658036231994629,
      "learning_rate": 4.1947235087263977e-05,
      "loss": 2.2003,
      "step": 191153
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.5437815189361572,
      "learning_rate": 4.194514893477039e-05,
      "loss": 2.154,
      "step": 191154
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.288047432899475,
      "learning_rate": 4.194306283025396e-05,
      "loss": 2.1352,
      "step": 191155
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.3296962976455688,
      "learning_rate": 4.1940976773715285e-05,
      "loss": 1.9761,
      "step": 191156
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.4946340322494507,
      "learning_rate": 4.1938890765154635e-05,
      "loss": 1.9727,
      "step": 191157
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.4268900156021118,
      "learning_rate": 4.1936804804572364e-05,
      "loss": 2.1015,
      "step": 191158
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.3609919548034668,
      "learning_rate": 4.1934718891968986e-05,
      "loss": 1.9089,
      "step": 191159
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.3883748054504395,
      "learning_rate": 4.1932633027344794e-05,
      "loss": 2.0198,
      "step": 191160
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.440925121307373,
      "learning_rate": 4.1930547210700126e-05,
      "loss": 2.0753,
      "step": 191161
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.324202537536621,
      "learning_rate": 4.1928461442035545e-05,
      "loss": 1.9836,
      "step": 191162
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.3012044429779053,
      "learning_rate": 4.1926375721351246e-05,
      "loss": 2.0958,
      "step": 191163
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.354986548423767,
      "learning_rate": 4.1924290048647766e-05,
      "loss": 1.9468,
      "step": 191164
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.2590816020965576,
      "learning_rate": 4.192220442392544e-05,
      "loss": 1.8829,
      "step": 191165
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.3612091541290283,
      "learning_rate": 4.1920118847184635e-05,
      "loss": 1.9988,
      "step": 191166
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.3958044052124023,
      "learning_rate": 4.191803331842568e-05,
      "loss": 2.1883,
      "step": 191167
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.4159282445907593,
      "learning_rate": 4.191594783764911e-05,
      "loss": 2.0267,
      "step": 191168
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.3789544105529785,
      "learning_rate": 4.1913862404855194e-05,
      "loss": 2.07,
      "step": 191169
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.4521533250808716,
      "learning_rate": 4.191177702004439e-05,
      "loss": 1.8483,
      "step": 191170
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.3648885488510132,
      "learning_rate": 4.1909691683217106e-05,
      "loss": 1.9102,
      "step": 191171
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.364398717880249,
      "learning_rate": 4.1907606394373575e-05,
      "loss": 2.0886,
      "step": 191172
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.372064471244812,
      "learning_rate": 4.1905521153514386e-05,
      "loss": 2.0602,
      "step": 191173
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.3073937892913818,
      "learning_rate": 4.190343596063982e-05,
      "loss": 1.7868,
      "step": 191174
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.2879040241241455,
      "learning_rate": 4.19013508157502e-05,
      "loss": 1.9816,
      "step": 191175
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.5802512168884277,
      "learning_rate": 4.189926571884606e-05,
      "loss": 2.0839,
      "step": 191176
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.353804349899292,
      "learning_rate": 4.189718066992764e-05,
      "loss": 1.9097,
      "step": 191177
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.3979802131652832,
      "learning_rate": 4.18950956689955e-05,
      "loss": 1.8206,
      "step": 191178
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.4510451555252075,
      "learning_rate": 4.189301071604995e-05,
      "loss": 1.8186,
      "step": 191179
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.4499138593673706,
      "learning_rate": 4.189092581109134e-05,
      "loss": 2.1158,
      "step": 191180
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.5654523372650146,
      "learning_rate": 4.1888840954119976e-05,
      "loss": 2.0491,
      "step": 191181
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.3564777374267578,
      "learning_rate": 4.188675614513646e-05,
      "loss": 2.0473,
      "step": 191182
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.2790247201919556,
      "learning_rate": 4.188467138414099e-05,
      "loss": 2.0992,
      "step": 191183
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.3199881315231323,
      "learning_rate": 4.18825866711341e-05,
      "loss": 2.1453,
      "step": 191184
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.371221661567688,
      "learning_rate": 4.188050200611609e-05,
      "loss": 1.9307,
      "step": 191185
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.3287566900253296,
      "learning_rate": 4.187841738908739e-05,
      "loss": 2.099,
      "step": 191186
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.3957173824310303,
      "learning_rate": 4.1876332820048306e-05,
      "loss": 2.1708,
      "step": 191187
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.3974977731704712,
      "learning_rate": 4.1874248298999335e-05,
      "loss": 1.8643,
      "step": 191188
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.3495689630508423,
      "learning_rate": 4.1872163825940743e-05,
      "loss": 2.2311,
      "step": 191189
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.5880361795425415,
      "learning_rate": 4.187007940087306e-05,
      "loss": 2.0266,
      "step": 191190
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.3640601634979248,
      "learning_rate": 4.186799502379662e-05,
      "loss": 2.1956,
      "step": 191191
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.451008915901184,
      "learning_rate": 4.186591069471176e-05,
      "loss": 1.9197,
      "step": 191192
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.4264516830444336,
      "learning_rate": 4.186382641361884e-05,
      "loss": 2.0195,
      "step": 191193
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.5106799602508545,
      "learning_rate": 4.1861742180518395e-05,
      "loss": 2.0489,
      "step": 191194
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.4622802734375,
      "learning_rate": 4.185965799541063e-05,
      "loss": 1.7644,
      "step": 191195
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.400023341178894,
      "learning_rate": 4.1857573858296144e-05,
      "loss": 1.9406,
      "step": 191196
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.4172413349151611,
      "learning_rate": 4.1855489769175165e-05,
      "loss": 2.2144,
      "step": 191197
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.3117856979370117,
      "learning_rate": 4.185340572804813e-05,
      "loss": 1.9523,
      "step": 191198
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.3361963033676147,
      "learning_rate": 4.1851321734915364e-05,
      "loss": 2.1883,
      "step": 191199
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.4126362800598145,
      "learning_rate": 4.184923778977735e-05,
      "loss": 2.1051,
      "step": 191200
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.3265540599822998,
      "learning_rate": 4.18471538926344e-05,
      "loss": 1.973,
      "step": 191201
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.4945157766342163,
      "learning_rate": 4.184507004348699e-05,
      "loss": 2.4181,
      "step": 191202
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.4141902923583984,
      "learning_rate": 4.184298624233546e-05,
      "loss": 1.9321,
      "step": 191203
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.457234501838684,
      "learning_rate": 4.1840902489180175e-05,
      "loss": 2.0705,
      "step": 191204
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.4516698122024536,
      "learning_rate": 4.183881878402149e-05,
      "loss": 1.9197,
      "step": 191205
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.468737244606018,
      "learning_rate": 4.183673512685991e-05,
      "loss": 1.9414,
      "step": 191206
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.3182580471038818,
      "learning_rate": 4.183465151769567e-05,
      "loss": 1.8503,
      "step": 191207
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.3588541746139526,
      "learning_rate": 4.18325679565293e-05,
      "loss": 2.0917,
      "step": 191208
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.4026885032653809,
      "learning_rate": 4.1830484443361177e-05,
      "loss": 1.81,
      "step": 191209
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.5578194856643677,
      "learning_rate": 4.182840097819159e-05,
      "loss": 2.0665,
      "step": 191210
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.3249157667160034,
      "learning_rate": 4.182631756102094e-05,
      "loss": 1.9845,
      "step": 191211
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.4750137329101562,
      "learning_rate": 4.182423419184973e-05,
      "loss": 1.9056,
      "step": 191212
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.39087975025177,
      "learning_rate": 4.182215087067815e-05,
      "loss": 1.9011,
      "step": 191213
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.4790133237838745,
      "learning_rate": 4.182006759750681e-05,
      "loss": 2.0743,
      "step": 191214
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.439115285873413,
      "learning_rate": 4.1817984372336013e-05,
      "loss": 1.8106,
      "step": 191215
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.2440972328186035,
      "learning_rate": 4.181590119516608e-05,
      "loss": 1.9011,
      "step": 191216
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.3319226503372192,
      "learning_rate": 4.1813818065997384e-05,
      "loss": 1.8037,
      "step": 191217
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.397546648979187,
      "learning_rate": 4.181173498483046e-05,
      "loss": 1.9215,
      "step": 191218
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.5128774642944336,
      "learning_rate": 4.180965195166554e-05,
      "loss": 2.1353,
      "step": 191219
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.3402297496795654,
      "learning_rate": 4.1807568966503145e-05,
      "loss": 2.0227,
      "step": 191220
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.33633553981781,
      "learning_rate": 4.180548602934359e-05,
      "loss": 1.9872,
      "step": 191221
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.4249604940414429,
      "learning_rate": 4.1803403140187275e-05,
      "loss": 2.0434,
      "step": 191222
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.4408687353134155,
      "learning_rate": 4.180132029903449e-05,
      "loss": 1.9472,
      "step": 191223
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.5659079551696777,
      "learning_rate": 4.1799237505885805e-05,
      "loss": 1.967,
      "step": 191224
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.3423571586608887,
      "learning_rate": 4.179715476074146e-05,
      "loss": 2.0854,
      "step": 191225
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.3686513900756836,
      "learning_rate": 4.179507206360194e-05,
      "loss": 1.9624,
      "step": 191226
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.2609546184539795,
      "learning_rate": 4.1792989414467595e-05,
      "loss": 1.9623,
      "step": 191227
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.2755204439163208,
      "learning_rate": 4.1790906813338806e-05,
      "loss": 1.7927,
      "step": 191228
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.3525111675262451,
      "learning_rate": 4.178882426021592e-05,
      "loss": 2.017,
      "step": 191229
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.4012939929962158,
      "learning_rate": 4.178674175509941e-05,
      "loss": 1.8444,
      "step": 191230
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.5268210172653198,
      "learning_rate": 4.178465929798955e-05,
      "loss": 2.1306,
      "step": 191231
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.55527663230896,
      "learning_rate": 4.178257688888687e-05,
      "loss": 1.9565,
      "step": 191232
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.330466389656067,
      "learning_rate": 4.178049452779167e-05,
      "loss": 1.8657,
      "step": 191233
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.386930227279663,
      "learning_rate": 4.177841221470438e-05,
      "loss": 2.0107,
      "step": 191234
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.3916162252426147,
      "learning_rate": 4.1776329949625254e-05,
      "loss": 1.9778,
      "step": 191235
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.4826796054840088,
      "learning_rate": 4.177424773255489e-05,
      "loss": 2.0309,
      "step": 191236
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.3319666385650635,
      "learning_rate": 4.1772165563493456e-05,
      "loss": 1.8952,
      "step": 191237
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.5679129362106323,
      "learning_rate": 4.177008344244155e-05,
      "loss": 1.6844,
      "step": 191238
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.30905020236969,
      "learning_rate": 4.1768001369399376e-05,
      "loss": 1.9569,
      "step": 191239
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.3433926105499268,
      "learning_rate": 4.176591934436749e-05,
      "loss": 1.894,
      "step": 191240
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.4167684316635132,
      "learning_rate": 4.176383736734618e-05,
      "loss": 1.9555,
      "step": 191241
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.3192631006240845,
      "learning_rate": 4.176175543833585e-05,
      "loss": 2.0799,
      "step": 191242
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.5232408046722412,
      "learning_rate": 4.175967355733682e-05,
      "loss": 1.8596,
      "step": 191243
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.5073273181915283,
      "learning_rate": 4.175759172434959e-05,
      "loss": 2.036,
      "step": 191244
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.7184779644012451,
      "learning_rate": 4.1755509939374455e-05,
      "loss": 1.9904,
      "step": 191245
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.4251792430877686,
      "learning_rate": 4.175342820241191e-05,
      "loss": 1.9716,
      "step": 191246
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.404726266860962,
      "learning_rate": 4.175134651346226e-05,
      "loss": 2.1854,
      "step": 191247
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.5346150398254395,
      "learning_rate": 4.1749264872525934e-05,
      "loss": 1.9482,
      "step": 191248
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.3714327812194824,
      "learning_rate": 4.17471832796032e-05,
      "loss": 2.0401,
      "step": 191249
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.358685851097107,
      "learning_rate": 4.174510173469463e-05,
      "loss": 2.0159,
      "step": 191250
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.3209855556488037,
      "learning_rate": 4.174302023780045e-05,
      "loss": 1.8102,
      "step": 191251
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.427445411682129,
      "learning_rate": 4.1740938788921164e-05,
      "loss": 2.1311,
      "step": 191252
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.647351622581482,
      "learning_rate": 4.173885738805714e-05,
      "loss": 1.9249,
      "step": 191253
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.40243661403656,
      "learning_rate": 4.1736776035208664e-05,
      "loss": 1.942,
      "step": 191254
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.3692474365234375,
      "learning_rate": 4.173469473037625e-05,
      "loss": 1.8334,
      "step": 191255
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.3985161781311035,
      "learning_rate": 4.173261347356023e-05,
      "loss": 2.0608,
      "step": 191256
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.3254575729370117,
      "learning_rate": 4.173053226476096e-05,
      "loss": 1.9335,
      "step": 191257
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.4780596494674683,
      "learning_rate": 4.172845110397889e-05,
      "loss": 2.0077,
      "step": 191258
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.4248470067977905,
      "learning_rate": 4.172636999121437e-05,
      "loss": 2.1084,
      "step": 191259
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.532891035079956,
      "learning_rate": 4.1724288926467775e-05,
      "loss": 1.9985,
      "step": 191260
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.3018025159835815,
      "learning_rate": 4.172220790973953e-05,
      "loss": 2.0105,
      "step": 191261
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.4343336820602417,
      "learning_rate": 4.1720126941029954e-05,
      "loss": 1.9918,
      "step": 191262
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.5428147315979004,
      "learning_rate": 4.1718046020339526e-05,
      "loss": 2.2368,
      "step": 191263
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.3027479648590088,
      "learning_rate": 4.1715965147668616e-05,
      "loss": 2.0546,
      "step": 191264
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.4564448595046997,
      "learning_rate": 4.171388432301756e-05,
      "loss": 2.0398,
      "step": 191265
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.5233407020568848,
      "learning_rate": 4.171180354638674e-05,
      "loss": 1.9654,
      "step": 191266
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.3132624626159668,
      "learning_rate": 4.1709722817776624e-05,
      "loss": 2.2011,
      "step": 191267
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.4391146898269653,
      "learning_rate": 4.170764213718747e-05,
      "loss": 2.0738,
      "step": 191268
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.307293176651001,
      "learning_rate": 4.17055615046198e-05,
      "loss": 2.134,
      "step": 191269
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.3078852891921997,
      "learning_rate": 4.170348092007395e-05,
      "loss": 1.8914,
      "step": 191270
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.4451076984405518,
      "learning_rate": 4.170140038355029e-05,
      "loss": 2.1153,
      "step": 191271
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.3914176225662231,
      "learning_rate": 4.169931989504918e-05,
      "loss": 1.7979,
      "step": 191272
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.3947241306304932,
      "learning_rate": 4.1697239454571085e-05,
      "loss": 2.0112,
      "step": 191273
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.4014021158218384,
      "learning_rate": 4.169515906211628e-05,
      "loss": 1.7864,
      "step": 191274
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.4637330770492554,
      "learning_rate": 4.169307871768529e-05,
      "loss": 1.945,
      "step": 191275
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.295836329460144,
      "learning_rate": 4.169099842127842e-05,
      "loss": 1.9234,
      "step": 191276
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.514907956123352,
      "learning_rate": 4.1688918172896103e-05,
      "loss": 1.9115,
      "step": 191277
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.3881996870040894,
      "learning_rate": 4.16868379725386e-05,
      "loss": 2.0153,
      "step": 191278
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.2873629331588745,
      "learning_rate": 4.168475782020645e-05,
      "loss": 2.1973,
      "step": 191279
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.3030898571014404,
      "learning_rate": 4.168267771589992e-05,
      "loss": 2.1381,
      "step": 191280
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.3802831172943115,
      "learning_rate": 4.168059765961953e-05,
      "loss": 1.8845,
      "step": 191281
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.3407225608825684,
      "learning_rate": 4.1678517651365596e-05,
      "loss": 2.0131,
      "step": 191282
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.394558072090149,
      "learning_rate": 4.1676437691138506e-05,
      "loss": 2.1522,
      "step": 191283
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.5071460008621216,
      "learning_rate": 4.167435777893857e-05,
      "loss": 2.1484,
      "step": 191284
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.5483756065368652,
      "learning_rate": 4.1672277914766284e-05,
      "loss": 2.0475,
      "step": 191285
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.4309368133544922,
      "learning_rate": 4.167019809862198e-05,
      "loss": 1.7742,
      "step": 191286
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.4707473516464233,
      "learning_rate": 4.166811833050609e-05,
      "loss": 2.0104,
      "step": 191287
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.3465074300765991,
      "learning_rate": 4.1666038610419004e-05,
      "loss": 1.8579,
      "step": 191288
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.4371910095214844,
      "learning_rate": 4.166395893836104e-05,
      "loss": 2.0398,
      "step": 191289
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.3785898685455322,
      "learning_rate": 4.166187931433259e-05,
      "loss": 1.7777,
      "step": 191290
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.375923752784729,
      "learning_rate": 4.165979973833412e-05,
      "loss": 1.8426,
      "step": 191291
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.3120367527008057,
      "learning_rate": 4.165772021036593e-05,
      "loss": 2.0639,
      "step": 191292
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.5881456136703491,
      "learning_rate": 4.1655640730428485e-05,
      "loss": 2.077,
      "step": 191293
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.4627606868743896,
      "learning_rate": 4.1653561298522156e-05,
      "loss": 2.0494,
      "step": 191294
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.2764456272125244,
      "learning_rate": 4.16514819146473e-05,
      "loss": 2.1862,
      "step": 191295
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.4251086711883545,
      "learning_rate": 4.1649402578804223e-05,
      "loss": 2.1272,
      "step": 191296
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.4639681577682495,
      "learning_rate": 4.164732329099349e-05,
      "loss": 2.0141,
      "step": 191297
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.3859034776687622,
      "learning_rate": 4.164524405121531e-05,
      "loss": 2.0963,
      "step": 191298
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.339890480041504,
      "learning_rate": 4.164316485947023e-05,
      "loss": 1.8393,
      "step": 191299
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.3356906175613403,
      "learning_rate": 4.164108571575857e-05,
      "loss": 2.1319,
      "step": 191300
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.2710974216461182,
      "learning_rate": 4.163900662008072e-05,
      "loss": 1.9224,
      "step": 191301
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.3150520324707031,
      "learning_rate": 4.1636927572436977e-05,
      "loss": 1.9356,
      "step": 191302
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.5256747007369995,
      "learning_rate": 4.163484857282784e-05,
      "loss": 2.2143,
      "step": 191303
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.481690526008606,
      "learning_rate": 4.1632769621253646e-05,
      "loss": 2.2377,
      "step": 191304
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.3580578565597534,
      "learning_rate": 4.1630690717714835e-05,
      "loss": 2.0852,
      "step": 191305
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.3564103841781616,
      "learning_rate": 4.162861186221169e-05,
      "loss": 1.9391,
      "step": 191306
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.443446397781372,
      "learning_rate": 4.162653305474479e-05,
      "loss": 1.8171,
      "step": 191307
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.3330520391464233,
      "learning_rate": 4.1624454295314305e-05,
      "loss": 1.9211,
      "step": 191308
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.356596827507019,
      "learning_rate": 4.162237558392072e-05,
      "loss": 1.9058,
      "step": 191309
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.4529391527175903,
      "learning_rate": 4.162029692056438e-05,
      "loss": 2.0503,
      "step": 191310
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.4717589616775513,
      "learning_rate": 4.1618218305245775e-05,
      "loss": 2.0618,
      "step": 191311
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.3487374782562256,
      "learning_rate": 4.161613973796511e-05,
      "loss": 1.821,
      "step": 191312
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.3356671333312988,
      "learning_rate": 4.161406121872306e-05,
      "loss": 1.9446,
      "step": 191313
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.4417835474014282,
      "learning_rate": 4.161198274751971e-05,
      "loss": 1.9302,
      "step": 191314
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.4651254415512085,
      "learning_rate": 4.16099043243556e-05,
      "loss": 2.1405,
      "step": 191315
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.527479648590088,
      "learning_rate": 4.160782594923103e-05,
      "loss": 2.1413,
      "step": 191316
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.3726913928985596,
      "learning_rate": 4.16057476221465e-05,
      "loss": 1.9658,
      "step": 191317
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.440075159072876,
      "learning_rate": 4.160366934310231e-05,
      "loss": 2.0006,
      "step": 191318
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.3690974712371826,
      "learning_rate": 4.160159111209898e-05,
      "loss": 2.0648,
      "step": 191319
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.3140331506729126,
      "learning_rate": 4.159951292913667e-05,
      "loss": 1.8952,
      "step": 191320
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.457633137702942,
      "learning_rate": 4.1597434794215925e-05,
      "loss": 1.9395,
      "step": 191321
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.3196343183517456,
      "learning_rate": 4.1595356707337046e-05,
      "loss": 1.643,
      "step": 191322
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.5177971124649048,
      "learning_rate": 4.159327866850054e-05,
      "loss": 1.9663,
      "step": 191323
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.3534150123596191,
      "learning_rate": 4.159120067770667e-05,
      "loss": 1.8017,
      "step": 191324
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.2807639837265015,
      "learning_rate": 4.158912273495594e-05,
      "loss": 1.8783,
      "step": 191325
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.3470094203948975,
      "learning_rate": 4.158704484024864e-05,
      "loss": 1.9046,
      "step": 191326
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.384414792060852,
      "learning_rate": 4.158496699358521e-05,
      "loss": 1.8896,
      "step": 191327
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.3093974590301514,
      "learning_rate": 4.158288919496592e-05,
      "loss": 1.8636,
      "step": 191328
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.4139411449432373,
      "learning_rate": 4.1580811444391326e-05,
      "loss": 2.0546,
      "step": 191329
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.4108331203460693,
      "learning_rate": 4.1578733741861704e-05,
      "loss": 2.2695,
      "step": 191330
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.394843578338623,
      "learning_rate": 4.157665608737748e-05,
      "loss": 2.0945,
      "step": 191331
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.5036927461624146,
      "learning_rate": 4.1574578480939093e-05,
      "loss": 1.9794,
      "step": 191332
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.4237855672836304,
      "learning_rate": 4.157250092254684e-05,
      "loss": 2.0912,
      "step": 191333
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.576134204864502,
      "learning_rate": 4.157042341220108e-05,
      "loss": 1.9394,
      "step": 191334
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.2848225831985474,
      "learning_rate": 4.1568345949902296e-05,
      "loss": 1.895,
      "step": 191335
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.6461893320083618,
      "learning_rate": 4.15662685356508e-05,
      "loss": 1.9295,
      "step": 191336
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.488464593887329,
      "learning_rate": 4.1564191169447084e-05,
      "loss": 1.9609,
      "step": 191337
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.3666449785232544,
      "learning_rate": 4.1562113851291424e-05,
      "loss": 1.9294,
      "step": 191338
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.5527400970458984,
      "learning_rate": 4.1560036581184196e-05,
      "loss": 2.2293,
      "step": 191339
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.3782974481582642,
      "learning_rate": 4.1557959359125936e-05,
      "loss": 2.1682,
      "step": 191340
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.5196834802627563,
      "learning_rate": 4.155588218511687e-05,
      "loss": 1.9435,
      "step": 191341
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.3555101156234741,
      "learning_rate": 4.155380505915744e-05,
      "loss": 2.045,
      "step": 191342
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.3530042171478271,
      "learning_rate": 4.1551727981248073e-05,
      "loss": 1.9038,
      "step": 191343
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.4521085023880005,
      "learning_rate": 4.15496509513891e-05,
      "loss": 2.0568,
      "step": 191344
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.3022559881210327,
      "learning_rate": 4.1547573969580896e-05,
      "loss": 1.9473,
      "step": 191345
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.4135252237319946,
      "learning_rate": 4.154549703582392e-05,
      "loss": 1.9061,
      "step": 191346
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.3673804998397827,
      "learning_rate": 4.154342015011844e-05,
      "loss": 2.0177,
      "step": 191347
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.4986287355422974,
      "learning_rate": 4.154134331246501e-05,
      "loss": 2.2649,
      "step": 191348
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.4856188297271729,
      "learning_rate": 4.153926652286389e-05,
      "loss": 1.8958,
      "step": 191349
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.370939016342163,
      "learning_rate": 4.1537189781315526e-05,
      "loss": 1.9986,
      "step": 191350
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.3683476448059082,
      "learning_rate": 4.153511308782019e-05,
      "loss": 2.1195,
      "step": 191351
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.4784828424453735,
      "learning_rate": 4.1533036442378456e-05,
      "loss": 2.0612,
      "step": 191352
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.4728738069534302,
      "learning_rate": 4.153095984499051e-05,
      "loss": 1.8915,
      "step": 191353
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.4006121158599854,
      "learning_rate": 4.152888329565692e-05,
      "loss": 2.0303,
      "step": 191354
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.4769917726516724,
      "learning_rate": 4.1526806794378e-05,
      "loss": 2.3429,
      "step": 191355
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.333698034286499,
      "learning_rate": 4.15247303411541e-05,
      "loss": 2.0396,
      "step": 191356
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.2559188604354858,
      "learning_rate": 4.1522653935985596e-05,
      "loss": 1.8611,
      "step": 191357
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.5427649021148682,
      "learning_rate": 4.152057757887295e-05,
      "loss": 2.0972,
      "step": 191358
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.3999464511871338,
      "learning_rate": 4.151850126981643e-05,
      "loss": 1.9015,
      "step": 191359
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.2762662172317505,
      "learning_rate": 4.15164250088166e-05,
      "loss": 2.1122,
      "step": 191360
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.4279310703277588,
      "learning_rate": 4.151434879587373e-05,
      "loss": 2.046,
      "step": 191361
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.4367977380752563,
      "learning_rate": 4.151227263098822e-05,
      "loss": 1.9588,
      "step": 191362
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.4042150974273682,
      "learning_rate": 4.15101965141604e-05,
      "loss": 2.0273,
      "step": 191363
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.2744297981262207,
      "learning_rate": 4.150812044539077e-05,
      "loss": 2.0212,
      "step": 191364
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.408659815788269,
      "learning_rate": 4.15060444246796e-05,
      "loss": 2.1371,
      "step": 191365
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.4764246940612793,
      "learning_rate": 4.150396845202738e-05,
      "loss": 1.9451,
      "step": 191366
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.3241851329803467,
      "learning_rate": 4.150189252743449e-05,
      "loss": 2.0452,
      "step": 191367
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.4498530626296997,
      "learning_rate": 4.149981665090125e-05,
      "loss": 1.8894,
      "step": 191368
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.533470869064331,
      "learning_rate": 4.1497740822428006e-05,
      "loss": 2.0198,
      "step": 191369
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.5381022691726685,
      "learning_rate": 4.1495665042015315e-05,
      "loss": 1.8566,
      "step": 191370
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.4801645278930664,
      "learning_rate": 4.149358930966334e-05,
      "loss": 2.0394,
      "step": 191371
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.5989733934402466,
      "learning_rate": 4.1491513625372695e-05,
      "loss": 1.9963,
      "step": 191372
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.2868328094482422,
      "learning_rate": 4.148943798914357e-05,
      "loss": 1.9688,
      "step": 191373
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.255563497543335,
      "learning_rate": 4.148736240097656e-05,
      "loss": 1.983,
      "step": 191374
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.4140565395355225,
      "learning_rate": 4.148528686087185e-05,
      "loss": 1.9622,
      "step": 191375
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.610164761543274,
      "learning_rate": 4.148321136882992e-05,
      "loss": 2.1683,
      "step": 191376
    },
    {
      "epoch": 2.49,
      "grad_norm": 2.723435401916504,
      "learning_rate": 4.148113592485107e-05,
      "loss": 2.127,
      "step": 191377
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.6653170585632324,
      "learning_rate": 4.147906052893585e-05,
      "loss": 2.0019,
      "step": 191378
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.2949774265289307,
      "learning_rate": 4.147698518108448e-05,
      "loss": 2.1318,
      "step": 191379
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.4631743431091309,
      "learning_rate": 4.147490988129756e-05,
      "loss": 1.9286,
      "step": 191380
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.3309228420257568,
      "learning_rate": 4.147283462957519e-05,
      "loss": 1.9712,
      "step": 191381
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.4967930316925049,
      "learning_rate": 4.147075942591798e-05,
      "loss": 1.9859,
      "step": 191382
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.4131989479064941,
      "learning_rate": 4.146868427032615e-05,
      "loss": 2.0086,
      "step": 191383
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.4004056453704834,
      "learning_rate": 4.146660916280023e-05,
      "loss": 2.0309,
      "step": 191384
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.3453794717788696,
      "learning_rate": 4.146453410334051e-05,
      "loss": 1.9163,
      "step": 191385
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.4110383987426758,
      "learning_rate": 4.146245909194753e-05,
      "loss": 1.8883,
      "step": 191386
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.406259536743164,
      "learning_rate": 4.1460384128621435e-05,
      "loss": 2.2104,
      "step": 191387
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.3251718282699585,
      "learning_rate": 4.1458309213362795e-05,
      "loss": 2.0128,
      "step": 191388
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.4612010717391968,
      "learning_rate": 4.145623434617187e-05,
      "loss": 1.9902,
      "step": 191389
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.498388648033142,
      "learning_rate": 4.1454159527049156e-05,
      "loss": 1.9979,
      "step": 191390
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.3796679973602295,
      "learning_rate": 4.145208475599496e-05,
      "loss": 2.0433,
      "step": 191391
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.3134280443191528,
      "learning_rate": 4.1450010033009814e-05,
      "loss": 2.3087,
      "step": 191392
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.4389193058013916,
      "learning_rate": 4.144793535809389e-05,
      "loss": 1.9211,
      "step": 191393
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.3348703384399414,
      "learning_rate": 4.1445860731247704e-05,
      "loss": 2.1129,
      "step": 191394
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.321553111076355,
      "learning_rate": 4.144378615247157e-05,
      "loss": 2.0656,
      "step": 191395
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.4598796367645264,
      "learning_rate": 4.144171162176598e-05,
      "loss": 1.8902,
      "step": 191396
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.372025489807129,
      "learning_rate": 4.1439637139131176e-05,
      "loss": 1.729,
      "step": 191397
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.500740647315979,
      "learning_rate": 4.143756270456775e-05,
      "loss": 1.9749,
      "step": 191398
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.3281035423278809,
      "learning_rate": 4.143548831807587e-05,
      "loss": 1.863,
      "step": 191399
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.3785885572433472,
      "learning_rate": 4.143341397965604e-05,
      "loss": 1.9947,
      "step": 191400
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.4848976135253906,
      "learning_rate": 4.1431339689308586e-05,
      "loss": 1.8616,
      "step": 191401
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.4435619115829468,
      "learning_rate": 4.1429265447033975e-05,
      "loss": 2.0292,
      "step": 191402
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.5043601989746094,
      "learning_rate": 4.142719125283248e-05,
      "loss": 2.1301,
      "step": 191403
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.4834873676300049,
      "learning_rate": 4.142511710670466e-05,
      "loss": 2.0067,
      "step": 191404
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.3507028818130493,
      "learning_rate": 4.142304300865069e-05,
      "loss": 1.9458,
      "step": 191405
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.3681617975234985,
      "learning_rate": 4.142096895867113e-05,
      "loss": 1.9486,
      "step": 191406
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.6741082668304443,
      "learning_rate": 4.1418894956766216e-05,
      "loss": 2.0442,
      "step": 191407
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.6501097679138184,
      "learning_rate": 4.141682100293647e-05,
      "loss": 2.0917,
      "step": 191408
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.3327279090881348,
      "learning_rate": 4.141474709718215e-05,
      "loss": 1.9223,
      "step": 191409
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.30039644241333,
      "learning_rate": 4.141267323950382e-05,
      "loss": 1.8248,
      "step": 191410
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.3809446096420288,
      "learning_rate": 4.141059942990168e-05,
      "loss": 1.8478,
      "step": 191411
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.4154683351516724,
      "learning_rate": 4.140852566837621e-05,
      "loss": 1.9432,
      "step": 191412
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.4820297956466675,
      "learning_rate": 4.140645195492772e-05,
      "loss": 1.8055,
      "step": 191413
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.4035816192626953,
      "learning_rate": 4.140437828955674e-05,
      "loss": 2.1881,
      "step": 191414
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.4967259168624878,
      "learning_rate": 4.14023046722635e-05,
      "loss": 1.9193,
      "step": 191415
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.4827063083648682,
      "learning_rate": 4.14002311030485e-05,
      "loss": 2.1032,
      "step": 191416
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.3632961511611938,
      "learning_rate": 4.139815758191207e-05,
      "loss": 2.0632,
      "step": 191417
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.4390246868133545,
      "learning_rate": 4.139608410885462e-05,
      "loss": 1.7568,
      "step": 191418
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.4015977382659912,
      "learning_rate": 4.139401068387644e-05,
      "loss": 2.1379,
      "step": 191419
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.7503973245620728,
      "learning_rate": 4.1391937306978076e-05,
      "loss": 2.0244,
      "step": 191420
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.7200419902801514,
      "learning_rate": 4.1389863978159774e-05,
      "loss": 2.0596,
      "step": 191421
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.4455064535140991,
      "learning_rate": 4.1387790697422016e-05,
      "loss": 2.1627,
      "step": 191422
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.3534510135650635,
      "learning_rate": 4.138571746476517e-05,
      "loss": 1.8314,
      "step": 191423
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.189034342765808,
      "learning_rate": 4.1383644280189556e-05,
      "loss": 1.8977,
      "step": 191424
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.4042612314224243,
      "learning_rate": 4.1381571143695613e-05,
      "loss": 1.7773,
      "step": 191425
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.3393641710281372,
      "learning_rate": 4.1379498055283774e-05,
      "loss": 2.2205,
      "step": 191426
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.6406346559524536,
      "learning_rate": 4.137742501495428e-05,
      "loss": 1.7825,
      "step": 191427
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.4598931074142456,
      "learning_rate": 4.137535202270765e-05,
      "loss": 1.9793,
      "step": 191428
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.3477635383605957,
      "learning_rate": 4.137327907854426e-05,
      "loss": 2.0042,
      "step": 191429
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.3129067420959473,
      "learning_rate": 4.137120618246438e-05,
      "loss": 1.7544,
      "step": 191430
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.4415782690048218,
      "learning_rate": 4.136913333446853e-05,
      "loss": 2.0627,
      "step": 191431
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.4675052165985107,
      "learning_rate": 4.136706053455705e-05,
      "loss": 2.0403,
      "step": 191432
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.4819679260253906,
      "learning_rate": 4.1364987782730244e-05,
      "loss": 1.952,
      "step": 191433
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.5513168573379517,
      "learning_rate": 4.136291507898863e-05,
      "loss": 2.1581,
      "step": 191434
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.5539168119430542,
      "learning_rate": 4.136084242333253e-05,
      "loss": 1.9543,
      "step": 191435
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.3556462526321411,
      "learning_rate": 4.1358769815762296e-05,
      "loss": 1.8249,
      "step": 191436
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.4854122400283813,
      "learning_rate": 4.135669725627839e-05,
      "loss": 1.998,
      "step": 191437
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.3453540802001953,
      "learning_rate": 4.135462474488109e-05,
      "loss": 2.0068,
      "step": 191438
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.277732253074646,
      "learning_rate": 4.1352552281570926e-05,
      "loss": 1.971,
      "step": 191439
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.402039885520935,
      "learning_rate": 4.13504798663482e-05,
      "loss": 2.1823,
      "step": 191440
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.5600712299346924,
      "learning_rate": 4.13484074992133e-05,
      "loss": 2.0187,
      "step": 191441
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.3340051174163818,
      "learning_rate": 4.1346335180166534e-05,
      "loss": 1.9406,
      "step": 191442
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.315997838973999,
      "learning_rate": 4.134426290920844e-05,
      "loss": 1.9086,
      "step": 191443
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.3960585594177246,
      "learning_rate": 4.134219068633927e-05,
      "loss": 1.8827,
      "step": 191444
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.2888399362564087,
      "learning_rate": 4.134011851155954e-05,
      "loss": 2.1593,
      "step": 191445
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.3314694166183472,
      "learning_rate": 4.133804638486947e-05,
      "loss": 1.9893,
      "step": 191446
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.3559768199920654,
      "learning_rate": 4.1335974306269706e-05,
      "loss": 2.2847,
      "step": 191447
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.4940605163574219,
      "learning_rate": 4.133390227576033e-05,
      "loss": 1.948,
      "step": 191448
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.4716507196426392,
      "learning_rate": 4.133183029334193e-05,
      "loss": 2.2188,
      "step": 191449
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.4186480045318604,
      "learning_rate": 4.1329758359014756e-05,
      "loss": 2.19,
      "step": 191450
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.4033775329589844,
      "learning_rate": 4.132768647277931e-05,
      "loss": 1.9962,
      "step": 191451
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.3130533695220947,
      "learning_rate": 4.132561463463589e-05,
      "loss": 1.9828,
      "step": 191452
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.396530270576477,
      "learning_rate": 4.132354284458504e-05,
      "loss": 1.9925,
      "step": 191453
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.364393711090088,
      "learning_rate": 4.132147110262689e-05,
      "loss": 1.6647,
      "step": 191454
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.3542993068695068,
      "learning_rate": 4.1319399408762034e-05,
      "loss": 1.9687,
      "step": 191455
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.2599689960479736,
      "learning_rate": 4.131732776299074e-05,
      "loss": 1.842,
      "step": 191456
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.5042585134506226,
      "learning_rate": 4.131525616531347e-05,
      "loss": 1.9256,
      "step": 191457
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.3425260782241821,
      "learning_rate": 4.1313184615730535e-05,
      "loss": 1.8977,
      "step": 191458
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.4530712366104126,
      "learning_rate": 4.131111311424249e-05,
      "loss": 1.9636,
      "step": 191459
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.2687532901763916,
      "learning_rate": 4.130904166084944e-05,
      "loss": 1.8225,
      "step": 191460
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.3584704399108887,
      "learning_rate": 4.1306970255552016e-05,
      "loss": 2.303,
      "step": 191461
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.3834893703460693,
      "learning_rate": 4.1304898898350424e-05,
      "loss": 1.9159,
      "step": 191462
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.38423490524292,
      "learning_rate": 4.130282758924519e-05,
      "loss": 1.8174,
      "step": 191463
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.3934972286224365,
      "learning_rate": 4.1300756328236586e-05,
      "loss": 1.958,
      "step": 191464
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.2712388038635254,
      "learning_rate": 4.12986851153252e-05,
      "loss": 1.7345,
      "step": 191465
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.4077297449111938,
      "learning_rate": 4.129661395051115e-05,
      "loss": 2.0655,
      "step": 191466
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.365578055381775,
      "learning_rate": 4.1294542833794985e-05,
      "loss": 2.0127,
      "step": 191467
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.4259235858917236,
      "learning_rate": 4.129247176517698e-05,
      "loss": 2.1788,
      "step": 191468
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.5985188484191895,
      "learning_rate": 4.129040074465767e-05,
      "loss": 2.0833,
      "step": 191469
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.4174503087997437,
      "learning_rate": 4.1288329772237283e-05,
      "loss": 1.7843,
      "step": 191470
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.5038648843765259,
      "learning_rate": 4.128625884791642e-05,
      "loss": 2.148,
      "step": 191471
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.3847829103469849,
      "learning_rate": 4.128418797169518e-05,
      "loss": 2.0071,
      "step": 191472
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.374407410621643,
      "learning_rate": 4.128211714357417e-05,
      "loss": 2.0596,
      "step": 191473
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.3948851823806763,
      "learning_rate": 4.128004636355364e-05,
      "loss": 1.9445,
      "step": 191474
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.449944257736206,
      "learning_rate": 4.127797563163407e-05,
      "loss": 1.8459,
      "step": 191475
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.5689855813980103,
      "learning_rate": 4.1275904947815766e-05,
      "loss": 2.0296,
      "step": 191476
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.5155072212219238,
      "learning_rate": 4.1273834312099275e-05,
      "loss": 2.0194,
      "step": 191477
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.2997761964797974,
      "learning_rate": 4.127176372448474e-05,
      "loss": 2.0704,
      "step": 191478
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.3818769454956055,
      "learning_rate": 4.126969318497273e-05,
      "loss": 2.0916,
      "step": 191479
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.3998249769210815,
      "learning_rate": 4.1267622693563504e-05,
      "loss": 1.9789,
      "step": 191480
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.4229873418807983,
      "learning_rate": 4.1265552250257574e-05,
      "loss": 2.1482,
      "step": 191481
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.6302675008773804,
      "learning_rate": 4.126348185505519e-05,
      "loss": 1.944,
      "step": 191482
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.3422582149505615,
      "learning_rate": 4.1261411507956966e-05,
      "loss": 1.8653,
      "step": 191483
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.3770450353622437,
      "learning_rate": 4.125934120896296e-05,
      "loss": 1.921,
      "step": 191484
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.3112730979919434,
      "learning_rate": 4.125727095807381e-05,
      "loss": 1.7085,
      "step": 191485
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.3336224555969238,
      "learning_rate": 4.1255200755289776e-05,
      "loss": 1.9448,
      "step": 191486
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.5664564371109009,
      "learning_rate": 4.1253130600611326e-05,
      "loss": 2.0534,
      "step": 191487
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.351548194885254,
      "learning_rate": 4.1251060494038735e-05,
      "loss": 1.9004,
      "step": 191488
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.369486927986145,
      "learning_rate": 4.1248990435572595e-05,
      "loss": 1.9006,
      "step": 191489
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.681677222251892,
      "learning_rate": 4.1246920425213e-05,
      "loss": 1.8281,
      "step": 191490
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.3655486106872559,
      "learning_rate": 4.124485046296057e-05,
      "loss": 1.9765,
      "step": 191491
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.724586009979248,
      "learning_rate": 4.1242780548815546e-05,
      "loss": 1.9204,
      "step": 191492
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.4309176206588745,
      "learning_rate": 4.1240710682778446e-05,
      "loss": 1.9147,
      "step": 191493
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.3356636762619019,
      "learning_rate": 4.123864086484949e-05,
      "loss": 1.8577,
      "step": 191494
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.6477729082107544,
      "learning_rate": 4.123657109502929e-05,
      "loss": 1.8989,
      "step": 191495
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.6820942163467407,
      "learning_rate": 4.1234501373317976e-05,
      "loss": 1.8261,
      "step": 191496
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.3895646333694458,
      "learning_rate": 4.123243169971614e-05,
      "loss": 1.8703,
      "step": 191497
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.460961937904358,
      "learning_rate": 4.1230362074223986e-05,
      "loss": 2.0688,
      "step": 191498
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.4691423177719116,
      "learning_rate": 4.1228292496842054e-05,
      "loss": 2.1022,
      "step": 191499
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.6678811311721802,
      "learning_rate": 4.1226222967570596e-05,
      "loss": 1.8887,
      "step": 191500
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.3834097385406494,
      "learning_rate": 4.122415348641016e-05,
      "loss": 1.883,
      "step": 191501
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.3974032402038574,
      "learning_rate": 4.122208405336104e-05,
      "loss": 1.963,
      "step": 191502
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.3687059879302979,
      "learning_rate": 4.1220014668423596e-05,
      "loss": 1.8344,
      "step": 191503
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.836348295211792,
      "learning_rate": 4.1217945331598166e-05,
      "loss": 2.0849,
      "step": 191504
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.4669785499572754,
      "learning_rate": 4.1215876042885285e-05,
      "loss": 1.8949,
      "step": 191505
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.2944613695144653,
      "learning_rate": 4.121380680228522e-05,
      "loss": 1.9407,
      "step": 191506
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.6659674644470215,
      "learning_rate": 4.121173760979843e-05,
      "loss": 2.0505,
      "step": 191507
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.519327998161316,
      "learning_rate": 4.1209668465425264e-05,
      "loss": 1.9834,
      "step": 191508
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.4269496202468872,
      "learning_rate": 4.1207599369166034e-05,
      "loss": 2.1257,
      "step": 191509
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.289760708808899,
      "learning_rate": 4.120553032102125e-05,
      "loss": 2.053,
      "step": 191510
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.3466174602508545,
      "learning_rate": 4.120346132099128e-05,
      "loss": 2.0004,
      "step": 191511
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.4701337814331055,
      "learning_rate": 4.120139236907639e-05,
      "loss": 1.8838,
      "step": 191512
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.4824752807617188,
      "learning_rate": 4.119932346527711e-05,
      "loss": 1.9224,
      "step": 191513
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.3695249557495117,
      "learning_rate": 4.119725460959374e-05,
      "loss": 2.0305,
      "step": 191514
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.3421480655670166,
      "learning_rate": 4.119518580202665e-05,
      "loss": 1.8062,
      "step": 191515
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.4253326654434204,
      "learning_rate": 4.1193117042576304e-05,
      "loss": 2.1312,
      "step": 191516
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.4581340551376343,
      "learning_rate": 4.119104833124307e-05,
      "loss": 1.9427,
      "step": 191517
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.4896749258041382,
      "learning_rate": 4.118897966802721e-05,
      "loss": 1.8334,
      "step": 191518
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.2965604066848755,
      "learning_rate": 4.1186911052929294e-05,
      "loss": 1.9339,
      "step": 191519
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.3225691318511963,
      "learning_rate": 4.118484248594962e-05,
      "loss": 1.9013,
      "step": 191520
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.3700335025787354,
      "learning_rate": 4.118277396708849e-05,
      "loss": 1.8576,
      "step": 191521
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.3072563409805298,
      "learning_rate": 4.118070549634643e-05,
      "loss": 1.9083,
      "step": 191522
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.3852794170379639,
      "learning_rate": 4.117863707372369e-05,
      "loss": 2.2514,
      "step": 191523
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.2623094320297241,
      "learning_rate": 4.117656869922081e-05,
      "loss": 1.7815,
      "step": 191524
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.4389820098876953,
      "learning_rate": 4.117450037283808e-05,
      "loss": 2.2666,
      "step": 191525
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.4424822330474854,
      "learning_rate": 4.1172432094575924e-05,
      "loss": 2.0155,
      "step": 191526
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.3502922058105469,
      "learning_rate": 4.117036386443461e-05,
      "loss": 1.9904,
      "step": 191527
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.3380078077316284,
      "learning_rate": 4.1168295682414665e-05,
      "loss": 1.8563,
      "step": 191528
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.3751938343048096,
      "learning_rate": 4.1166227548516364e-05,
      "loss": 2.3025,
      "step": 191529
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.4459202289581299,
      "learning_rate": 4.1164159462740234e-05,
      "loss": 1.9909,
      "step": 191530
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.467779278755188,
      "learning_rate": 4.116209142508654e-05,
      "loss": 2.1493,
      "step": 191531
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.3307398557662964,
      "learning_rate": 4.116002343555572e-05,
      "loss": 1.8178,
      "step": 191532
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.342272400856018,
      "learning_rate": 4.1157955494148076e-05,
      "loss": 1.9617,
      "step": 191533
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.5857502222061157,
      "learning_rate": 4.11558876008641e-05,
      "loss": 1.9632,
      "step": 191534
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.3728455305099487,
      "learning_rate": 4.1153819755704063e-05,
      "loss": 1.8855,
      "step": 191535
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.4819514751434326,
      "learning_rate": 4.11517519586685e-05,
      "loss": 1.8392,
      "step": 191536
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.4658637046813965,
      "learning_rate": 4.114968420975764e-05,
      "loss": 2.1122,
      "step": 191537
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.359094500541687,
      "learning_rate": 4.114761650897208e-05,
      "loss": 1.7513,
      "step": 191538
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.4961832761764526,
      "learning_rate": 4.114554885631193e-05,
      "loss": 2.2349,
      "step": 191539
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.213736653327942,
      "learning_rate": 4.114348125177774e-05,
      "loss": 1.9008,
      "step": 191540
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.295702576637268,
      "learning_rate": 4.114141369536983e-05,
      "loss": 1.9587,
      "step": 191541
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.4429535865783691,
      "learning_rate": 4.113934618708869e-05,
      "loss": 2.0453,
      "step": 191542
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.405207872390747,
      "learning_rate": 4.113727872693454e-05,
      "loss": 2.043,
      "step": 191543
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.3614344596862793,
      "learning_rate": 4.1135211314908034e-05,
      "loss": 2.0802,
      "step": 191544
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.249192237854004,
      "learning_rate": 4.1133143951009195e-05,
      "loss": 1.9915,
      "step": 191545
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.279395341873169,
      "learning_rate": 4.113107663523869e-05,
      "loss": 1.9602,
      "step": 191546
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.5046584606170654,
      "learning_rate": 4.1129009367596724e-05,
      "loss": 2.0907,
      "step": 191547
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.3736350536346436,
      "learning_rate": 4.112694214808382e-05,
      "loss": 2.178,
      "step": 191548
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.5763790607452393,
      "learning_rate": 4.112487497670025e-05,
      "loss": 1.9562,
      "step": 191549
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.2925792932510376,
      "learning_rate": 4.1122807853446584e-05,
      "loss": 1.8146,
      "step": 191550
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.4554228782653809,
      "learning_rate": 4.1120740778322946e-05,
      "loss": 1.8206,
      "step": 191551
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.4730455875396729,
      "learning_rate": 4.111867375132991e-05,
      "loss": 1.9826,
      "step": 191552
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.4082913398742676,
      "learning_rate": 4.1116606772467744e-05,
      "loss": 1.8554,
      "step": 191553
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.4383822679519653,
      "learning_rate": 4.111453984173694e-05,
      "loss": 1.8563,
      "step": 191554
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.489363670349121,
      "learning_rate": 4.1112472959137764e-05,
      "loss": 2.0014,
      "step": 191555
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.3252885341644287,
      "learning_rate": 4.111040612467079e-05,
      "loss": 2.068,
      "step": 191556
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.4920735359191895,
      "learning_rate": 4.1108339338336185e-05,
      "loss": 1.8778,
      "step": 191557
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.3168907165527344,
      "learning_rate": 4.110627260013444e-05,
      "loss": 1.9346,
      "step": 191558
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.5776933431625366,
      "learning_rate": 4.110420591006589e-05,
      "loss": 2.041,
      "step": 191559
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.458868384361267,
      "learning_rate": 4.110213926813104e-05,
      "loss": 2.0083,
      "step": 191560
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.522287130355835,
      "learning_rate": 4.1100072674330085e-05,
      "loss": 1.981,
      "step": 191561
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.374515414237976,
      "learning_rate": 4.109800612866366e-05,
      "loss": 1.9748,
      "step": 191562
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.6775296926498413,
      "learning_rate": 4.1095939631131866e-05,
      "loss": 2.087,
      "step": 191563
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.3480950593948364,
      "learning_rate": 4.1093873181735306e-05,
      "loss": 2.0683,
      "step": 191564
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.9189878702163696,
      "learning_rate": 4.1091806780474204e-05,
      "loss": 2.2735,
      "step": 191565
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.5260021686553955,
      "learning_rate": 4.108974042734909e-05,
      "loss": 1.9026,
      "step": 191566
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.571184754371643,
      "learning_rate": 4.1087674122360216e-05,
      "loss": 2.1148,
      "step": 191567
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.359400749206543,
      "learning_rate": 4.108560786550816e-05,
      "loss": 2.1361,
      "step": 191568
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.3587791919708252,
      "learning_rate": 4.108354165679304e-05,
      "loss": 2.0797,
      "step": 191569
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.3811758756637573,
      "learning_rate": 4.108147549621544e-05,
      "loss": 2.0935,
      "step": 191570
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.4386054277420044,
      "learning_rate": 4.107940938377564e-05,
      "loss": 1.9477,
      "step": 191571
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.4753319025039673,
      "learning_rate": 4.1077343319474134e-05,
      "loss": 2.0708,
      "step": 191572
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.3589553833007812,
      "learning_rate": 4.107527730331115e-05,
      "loss": 2.0906,
      "step": 191573
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.611312985420227,
      "learning_rate": 4.1073211335287227e-05,
      "loss": 2.2093,
      "step": 191574
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.5277584791183472,
      "learning_rate": 4.107114541540266e-05,
      "loss": 2.0441,
      "step": 191575
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.2825913429260254,
      "learning_rate": 4.106907954365789e-05,
      "loss": 2.3215,
      "step": 191576
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.5064830780029297,
      "learning_rate": 4.1067013720053175e-05,
      "loss": 1.9191,
      "step": 191577
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.5110899209976196,
      "learning_rate": 4.106494794458909e-05,
      "loss": 1.9654,
      "step": 191578
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.5280048847198486,
      "learning_rate": 4.106288221726582e-05,
      "loss": 1.8901,
      "step": 191579
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.279958963394165,
      "learning_rate": 4.1060816538083916e-05,
      "loss": 2.0838,
      "step": 191580
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.3856614828109741,
      "learning_rate": 4.1058750907043704e-05,
      "loss": 2.0004,
      "step": 191581
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.5753382444381714,
      "learning_rate": 4.105668532414558e-05,
      "loss": 2.0299,
      "step": 191582
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.5517913103103638,
      "learning_rate": 4.105461978938981e-05,
      "loss": 1.9214,
      "step": 191583
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.3770943880081177,
      "learning_rate": 4.105255430277696e-05,
      "loss": 1.9556,
      "step": 191584
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.4973301887512207,
      "learning_rate": 4.105048886430724e-05,
      "loss": 1.9148,
      "step": 191585
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.3066906929016113,
      "learning_rate": 4.104842347398121e-05,
      "loss": 1.9128,
      "step": 191586
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.3235509395599365,
      "learning_rate": 4.104635813179913e-05,
      "loss": 1.9747,
      "step": 191587
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.3758374452590942,
      "learning_rate": 4.1044292837761475e-05,
      "loss": 2.1211,
      "step": 191588
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.544100046157837,
      "learning_rate": 4.104222759186847e-05,
      "loss": 1.8875,
      "step": 191589
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.6692527532577515,
      "learning_rate": 4.104016239412066e-05,
      "loss": 2.0537,
      "step": 191590
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.4541858434677124,
      "learning_rate": 4.103809724451833e-05,
      "loss": 2.0855,
      "step": 191591
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.385343313217163,
      "learning_rate": 4.103603214306196e-05,
      "loss": 2.3063,
      "step": 191592
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.2810769081115723,
      "learning_rate": 4.1033967089751906e-05,
      "loss": 1.8864,
      "step": 191593
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.355478286743164,
      "learning_rate": 4.1031902084588506e-05,
      "loss": 2.0268,
      "step": 191594
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.3958064317703247,
      "learning_rate": 4.10298371275721e-05,
      "loss": 2.0621,
      "step": 191595
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.303004264831543,
      "learning_rate": 4.102777221870317e-05,
      "loss": 2.029,
      "step": 191596
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.4253252744674683,
      "learning_rate": 4.102570735798203e-05,
      "loss": 1.6887,
      "step": 191597
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.5130399465560913,
      "learning_rate": 4.1023642545409176e-05,
      "loss": 2.1594,
      "step": 191598
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.4432306289672852,
      "learning_rate": 4.102157778098491e-05,
      "loss": 2.0173,
      "step": 191599
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.4613648653030396,
      "learning_rate": 4.1019513064709554e-05,
      "loss": 2.0624,
      "step": 191600
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.656232237815857,
      "learning_rate": 4.101744839658362e-05,
      "loss": 1.9601,
      "step": 191601
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.519748330116272,
      "learning_rate": 4.10153837766074e-05,
      "loss": 1.9722,
      "step": 191602
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.3918503522872925,
      "learning_rate": 4.10133192047813e-05,
      "loss": 2.0558,
      "step": 191603
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.3381918668746948,
      "learning_rate": 4.101125468110575e-05,
      "loss": 1.8966,
      "step": 191604
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.3969329595565796,
      "learning_rate": 4.100919020558109e-05,
      "loss": 1.8348,
      "step": 191605
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.3791221380233765,
      "learning_rate": 4.100712577820764e-05,
      "loss": 1.8922,
      "step": 191606
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.3931691646575928,
      "learning_rate": 4.100506139898594e-05,
      "loss": 1.9866,
      "step": 191607
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.37100088596344,
      "learning_rate": 4.100299706791622e-05,
      "loss": 1.8722,
      "step": 191608
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.3114020824432373,
      "learning_rate": 4.100093278499899e-05,
      "loss": 2.0436,
      "step": 191609
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.5412095785140991,
      "learning_rate": 4.09988685502346e-05,
      "loss": 2.0831,
      "step": 191610
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.4044901132583618,
      "learning_rate": 4.099680436362336e-05,
      "loss": 1.8468,
      "step": 191611
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.3939969539642334,
      "learning_rate": 4.099474022516567e-05,
      "loss": 1.9945,
      "step": 191612
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.4648569822311401,
      "learning_rate": 4.0992676134862023e-05,
      "loss": 1.985,
      "step": 191613
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.4060860872268677,
      "learning_rate": 4.099061209271263e-05,
      "loss": 1.9071,
      "step": 191614
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.3467097282409668,
      "learning_rate": 4.098854809871808e-05,
      "loss": 1.7895,
      "step": 191615
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.3267309665679932,
      "learning_rate": 4.0986484152878606e-05,
      "loss": 2.0988,
      "step": 191616
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.38758385181427,
      "learning_rate": 4.098442025519465e-05,
      "loss": 2.1402,
      "step": 191617
    },
    {
      "epoch": 2.49,
      "grad_norm": 1.3415882587432861,
      "learning_rate": 4.09823564056665e-05,
      "loss": 2.0136,
      "step": 191618
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.4013323783874512,
      "learning_rate": 4.098029260429471e-05,
      "loss": 1.8775,
      "step": 191619
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.5736830234527588,
      "learning_rate": 4.0978228851079485e-05,
      "loss": 1.9302,
      "step": 191620
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.462697148323059,
      "learning_rate": 4.097616514602138e-05,
      "loss": 1.9722,
      "step": 191621
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.5921969413757324,
      "learning_rate": 4.097410148912061e-05,
      "loss": 1.8997,
      "step": 191622
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.4185270071029663,
      "learning_rate": 4.0972037880377796e-05,
      "loss": 1.8217,
      "step": 191623
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.4222462177276611,
      "learning_rate": 4.096997431979305e-05,
      "loss": 2.11,
      "step": 191624
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.4979342222213745,
      "learning_rate": 4.096791080736692e-05,
      "loss": 2.1496,
      "step": 191625
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.419600248336792,
      "learning_rate": 4.096584734309967e-05,
      "loss": 1.8777,
      "step": 191626
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.402554988861084,
      "learning_rate": 4.096378392699182e-05,
      "loss": 1.8718,
      "step": 191627
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.343907356262207,
      "learning_rate": 4.0961720559043655e-05,
      "loss": 2.0956,
      "step": 191628
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.2200233936309814,
      "learning_rate": 4.095965723925573e-05,
      "loss": 1.9199,
      "step": 191629
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.3495638370513916,
      "learning_rate": 4.095759396762814e-05,
      "loss": 2.0349,
      "step": 191630
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.378753662109375,
      "learning_rate": 4.0955530744161467e-05,
      "loss": 1.9679,
      "step": 191631
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.291018009185791,
      "learning_rate": 4.0953467568856035e-05,
      "loss": 2.0675,
      "step": 191632
    },
    {
      "epoch": 2.5,
      "grad_norm": 2.0365207195281982,
      "learning_rate": 4.095140444171228e-05,
      "loss": 1.9761,
      "step": 191633
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.5393860340118408,
      "learning_rate": 4.094934136273049e-05,
      "loss": 2.1198,
      "step": 191634
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.5798144340515137,
      "learning_rate": 4.094727833191125e-05,
      "loss": 2.0399,
      "step": 191635
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.561893343925476,
      "learning_rate": 4.0945215349254645e-05,
      "loss": 2.1406,
      "step": 191636
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.3798829317092896,
      "learning_rate": 4.094315241476132e-05,
      "loss": 2.0138,
      "step": 191637
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.4789279699325562,
      "learning_rate": 4.094108952843143e-05,
      "loss": 2.1037,
      "step": 191638
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.390783429145813,
      "learning_rate": 4.093902669026557e-05,
      "loss": 1.8909,
      "step": 191639
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.442270278930664,
      "learning_rate": 4.0936963900264e-05,
      "loss": 2.0529,
      "step": 191640
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.4045377969741821,
      "learning_rate": 4.093490115842719e-05,
      "loss": 2.1404,
      "step": 191641
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.3879557847976685,
      "learning_rate": 4.093283846475546e-05,
      "loss": 2.0014,
      "step": 191642
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.397578477859497,
      "learning_rate": 4.093077581924919e-05,
      "loss": 1.9091,
      "step": 191643
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.471721887588501,
      "learning_rate": 4.0928713221908736e-05,
      "loss": 2.2369,
      "step": 191644
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.5513414144515991,
      "learning_rate": 4.0926650672734586e-05,
      "loss": 1.9012,
      "step": 191645
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.6609547138214111,
      "learning_rate": 4.0924588171727004e-05,
      "loss": 2.0009,
      "step": 191646
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.4440302848815918,
      "learning_rate": 4.0922525718886486e-05,
      "loss": 1.9957,
      "step": 191647
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.5411804914474487,
      "learning_rate": 4.092046331421338e-05,
      "loss": 2.0383,
      "step": 191648
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.3404922485351562,
      "learning_rate": 4.091840095770804e-05,
      "loss": 1.8782,
      "step": 191649
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.6085035800933838,
      "learning_rate": 4.091633864937077e-05,
      "loss": 2.1491,
      "step": 191650
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.479990005493164,
      "learning_rate": 4.0914276389202135e-05,
      "loss": 2.0283,
      "step": 191651
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.5811257362365723,
      "learning_rate": 4.091221417720237e-05,
      "loss": 1.8348,
      "step": 191652
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.4634796380996704,
      "learning_rate": 4.0910152013371976e-05,
      "loss": 1.7677,
      "step": 191653
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.3444783687591553,
      "learning_rate": 4.090808989771125e-05,
      "loss": 1.8224,
      "step": 191654
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.471692442893982,
      "learning_rate": 4.090602783022062e-05,
      "loss": 1.8394,
      "step": 191655
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.5137702226638794,
      "learning_rate": 4.090396581090036e-05,
      "loss": 1.998,
      "step": 191656
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.3230006694793701,
      "learning_rate": 4.0901903839751036e-05,
      "loss": 2.0381,
      "step": 191657
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.5355592966079712,
      "learning_rate": 4.089984191677288e-05,
      "loss": 2.058,
      "step": 191658
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.343197226524353,
      "learning_rate": 4.089778004196639e-05,
      "loss": 2.0278,
      "step": 191659
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.3550000190734863,
      "learning_rate": 4.0895718215331894e-05,
      "loss": 1.9091,
      "step": 191660
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.3960044384002686,
      "learning_rate": 4.089365643686977e-05,
      "loss": 2.1333,
      "step": 191661
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.3619998693466187,
      "learning_rate": 4.0891594706580346e-05,
      "loss": 1.9192,
      "step": 191662
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.421604871749878,
      "learning_rate": 4.088953302446411e-05,
      "loss": 2.0997,
      "step": 191663
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.5686167478561401,
      "learning_rate": 4.0887471390521354e-05,
      "loss": 2.0556,
      "step": 191664
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.4354026317596436,
      "learning_rate": 4.088540980475259e-05,
      "loss": 1.6713,
      "step": 191665
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.4248565435409546,
      "learning_rate": 4.088334826715809e-05,
      "loss": 1.9523,
      "step": 191666
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.2232402563095093,
      "learning_rate": 4.0881286777738286e-05,
      "loss": 2.3145,
      "step": 191667
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.3985319137573242,
      "learning_rate": 4.087922533649348e-05,
      "loss": 1.9176,
      "step": 191668
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.4270780086517334,
      "learning_rate": 4.087716394342417e-05,
      "loss": 1.9843,
      "step": 191669
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.3321462869644165,
      "learning_rate": 4.087510259853063e-05,
      "loss": 1.9516,
      "step": 191670
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.356876015663147,
      "learning_rate": 4.087304130181335e-05,
      "loss": 2.1213,
      "step": 191671
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.4225126504898071,
      "learning_rate": 4.087098005327266e-05,
      "loss": 2.1771,
      "step": 191672
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.4742738008499146,
      "learning_rate": 4.086891885290897e-05,
      "loss": 1.8645,
      "step": 191673
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.4206117391586304,
      "learning_rate": 4.086685770072254e-05,
      "loss": 1.8044,
      "step": 191674
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.3988560438156128,
      "learning_rate": 4.086479659671395e-05,
      "loss": 1.9577,
      "step": 191675
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.3060497045516968,
      "learning_rate": 4.0862735540883405e-05,
      "loss": 2.0005,
      "step": 191676
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.5811725854873657,
      "learning_rate": 4.086067453323146e-05,
      "loss": 1.8857,
      "step": 191677
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.4740169048309326,
      "learning_rate": 4.085861357375838e-05,
      "loss": 2.232,
      "step": 191678
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.3382021188735962,
      "learning_rate": 4.085655266246459e-05,
      "loss": 1.9642,
      "step": 191679
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.3726801872253418,
      "learning_rate": 4.085449179935036e-05,
      "loss": 1.9955,
      "step": 191680
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.4223757982254028,
      "learning_rate": 4.0852430984416295e-05,
      "loss": 2.1839,
      "step": 191681
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.3232018947601318,
      "learning_rate": 4.0850370217662554e-05,
      "loss": 2.1557,
      "step": 191682
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.3594030141830444,
      "learning_rate": 4.084830949908967e-05,
      "loss": 2.0269,
      "step": 191683
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.4033163785934448,
      "learning_rate": 4.084624882869801e-05,
      "loss": 2.0051,
      "step": 191684
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.2978312969207764,
      "learning_rate": 4.084418820648785e-05,
      "loss": 1.8439,
      "step": 191685
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.4021817445755005,
      "learning_rate": 4.084212763245971e-05,
      "loss": 1.8427,
      "step": 191686
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.3317770957946777,
      "learning_rate": 4.0840067106613894e-05,
      "loss": 1.8502,
      "step": 191687
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.4778281450271606,
      "learning_rate": 4.083800662895077e-05,
      "loss": 2.031,
      "step": 191688
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.7265148162841797,
      "learning_rate": 4.08359461994708e-05,
      "loss": 2.1406,
      "step": 191689
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.3307684659957886,
      "learning_rate": 4.083388581817429e-05,
      "loss": 1.8817,
      "step": 191690
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.3264766931533813,
      "learning_rate": 4.083182548506164e-05,
      "loss": 1.832,
      "step": 191691
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.501365065574646,
      "learning_rate": 4.082976520013328e-05,
      "loss": 1.8355,
      "step": 191692
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.2573695182800293,
      "learning_rate": 4.082770496338957e-05,
      "loss": 1.6832,
      "step": 191693
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.3471993207931519,
      "learning_rate": 4.082564477483083e-05,
      "loss": 1.8945,
      "step": 191694
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.3411487340927124,
      "learning_rate": 4.082358463445753e-05,
      "loss": 2.1422,
      "step": 191695
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.4130610227584839,
      "learning_rate": 4.0821524542270026e-05,
      "loss": 2.2292,
      "step": 191696
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.4335793256759644,
      "learning_rate": 4.081946449826864e-05,
      "loss": 2.2388,
      "step": 191697
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.4986838102340698,
      "learning_rate": 4.081740450245384e-05,
      "loss": 2.146,
      "step": 191698
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.4320857524871826,
      "learning_rate": 4.081534455482597e-05,
      "loss": 1.934,
      "step": 191699
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.4151192903518677,
      "learning_rate": 4.081328465538545e-05,
      "loss": 2.2317,
      "step": 191700
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.5116605758666992,
      "learning_rate": 4.081122480413261e-05,
      "loss": 1.8807,
      "step": 191701
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.378082513809204,
      "learning_rate": 4.08091650010679e-05,
      "loss": 2.0519,
      "step": 191702
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.2737113237380981,
      "learning_rate": 4.0807105246191574e-05,
      "loss": 2.1591,
      "step": 191703
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.4733595848083496,
      "learning_rate": 4.080504553950417e-05,
      "loss": 1.8656,
      "step": 191704
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.4297020435333252,
      "learning_rate": 4.080298588100591e-05,
      "loss": 1.9023,
      "step": 191705
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.3739593029022217,
      "learning_rate": 4.080092627069737e-05,
      "loss": 2.1493,
      "step": 191706
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.3520622253417969,
      "learning_rate": 4.079886670857876e-05,
      "loss": 1.9534,
      "step": 191707
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.599605917930603,
      "learning_rate": 4.079680719465059e-05,
      "loss": 1.9907,
      "step": 191708
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.45722496509552,
      "learning_rate": 4.079474772891318e-05,
      "loss": 1.9303,
      "step": 191709
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.5602257251739502,
      "learning_rate": 4.079268831136692e-05,
      "loss": 2.0585,
      "step": 191710
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.5345792770385742,
      "learning_rate": 4.079062894201214e-05,
      "loss": 2.1844,
      "step": 191711
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.4583014249801636,
      "learning_rate": 4.078856962084931e-05,
      "loss": 1.8089,
      "step": 191712
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.5635271072387695,
      "learning_rate": 4.078651034787873e-05,
      "loss": 1.9644,
      "step": 191713
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.4311624765396118,
      "learning_rate": 4.0784451123100906e-05,
      "loss": 1.985,
      "step": 191714
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.4828323125839233,
      "learning_rate": 4.078239194651616e-05,
      "loss": 1.8696,
      "step": 191715
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.5485575199127197,
      "learning_rate": 4.0780332818124825e-05,
      "loss": 1.799,
      "step": 191716
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.5259807109832764,
      "learning_rate": 4.077827373792728e-05,
      "loss": 1.9176,
      "step": 191717
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.5073843002319336,
      "learning_rate": 4.077621470592398e-05,
      "loss": 1.9195,
      "step": 191718
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.4708319902420044,
      "learning_rate": 4.077415572211523e-05,
      "loss": 1.9517,
      "step": 191719
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.3188157081604004,
      "learning_rate": 4.077209678650153e-05,
      "loss": 1.8999,
      "step": 191720
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.4459325075149536,
      "learning_rate": 4.077003789908321e-05,
      "loss": 1.8812,
      "step": 191721
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.5896834135055542,
      "learning_rate": 4.07679790598606e-05,
      "loss": 2.1211,
      "step": 191722
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.5467848777770996,
      "learning_rate": 4.076592026883404e-05,
      "loss": 1.9178,
      "step": 191723
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.3755137920379639,
      "learning_rate": 4.076386152600409e-05,
      "loss": 2.1459,
      "step": 191724
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.4134693145751953,
      "learning_rate": 4.076180283137096e-05,
      "loss": 1.8463,
      "step": 191725
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.334201455116272,
      "learning_rate": 4.075974418493514e-05,
      "loss": 2.0729,
      "step": 191726
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.4022189378738403,
      "learning_rate": 4.0757685586696995e-05,
      "loss": 1.9096,
      "step": 191727
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.7944390773773193,
      "learning_rate": 4.07556270366569e-05,
      "loss": 2.0368,
      "step": 191728
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.4146993160247803,
      "learning_rate": 4.0753568534815154e-05,
      "loss": 2.0007,
      "step": 191729
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.4236472845077515,
      "learning_rate": 4.075151008117228e-05,
      "loss": 1.9646,
      "step": 191730
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.3236489295959473,
      "learning_rate": 4.0749451675728515e-05,
      "loss": 1.7355,
      "step": 191731
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.5529223680496216,
      "learning_rate": 4.07473933184844e-05,
      "loss": 1.7885,
      "step": 191732
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.500151515007019,
      "learning_rate": 4.074533500944023e-05,
      "loss": 2.1021,
      "step": 191733
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.4542653560638428,
      "learning_rate": 4.07432767485964e-05,
      "loss": 1.9188,
      "step": 191734
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.4333189725875854,
      "learning_rate": 4.0741218535953245e-05,
      "loss": 1.8421,
      "step": 191735
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.5112872123718262,
      "learning_rate": 4.0739160371511234e-05,
      "loss": 1.9153,
      "step": 191736
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.3882437944412231,
      "learning_rate": 4.0737102255270637e-05,
      "loss": 2.2248,
      "step": 191737
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.339388132095337,
      "learning_rate": 4.073504418723198e-05,
      "loss": 2.1681,
      "step": 191738
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.6657084226608276,
      "learning_rate": 4.073298616739556e-05,
      "loss": 2.0749,
      "step": 191739
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.306775689125061,
      "learning_rate": 4.073092819576179e-05,
      "loss": 1.9659,
      "step": 191740
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.3747026920318604,
      "learning_rate": 4.072887027233096e-05,
      "loss": 1.8738,
      "step": 191741
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.8505622148513794,
      "learning_rate": 4.0726812397103605e-05,
      "loss": 2.1504,
      "step": 191742
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.4157291650772095,
      "learning_rate": 4.0724754570079954e-05,
      "loss": 1.8515,
      "step": 191743
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.5638813972473145,
      "learning_rate": 4.072269679126054e-05,
      "loss": 2.2427,
      "step": 191744
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.4026930332183838,
      "learning_rate": 4.0720639060645646e-05,
      "loss": 1.984,
      "step": 191745
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.298325777053833,
      "learning_rate": 4.071858137823571e-05,
      "loss": 2.026,
      "step": 191746
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.5926803350448608,
      "learning_rate": 4.0716523744030995e-05,
      "loss": 2.0791,
      "step": 191747
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.8930532932281494,
      "learning_rate": 4.0714466158032046e-05,
      "loss": 2.1046,
      "step": 191748
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.445152759552002,
      "learning_rate": 4.07124086202391e-05,
      "loss": 2.1245,
      "step": 191749
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.4342180490493774,
      "learning_rate": 4.07103511306527e-05,
      "loss": 1.9201,
      "step": 191750
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.334591031074524,
      "learning_rate": 4.070829368927313e-05,
      "loss": 1.9666,
      "step": 191751
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.3841183185577393,
      "learning_rate": 4.070623629610077e-05,
      "loss": 2.191,
      "step": 191752
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.505488395690918,
      "learning_rate": 4.070417895113598e-05,
      "loss": 2.3414,
      "step": 191753
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.56714928150177,
      "learning_rate": 4.0702121654379195e-05,
      "loss": 2.0013,
      "step": 191754
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.4073344469070435,
      "learning_rate": 4.0700064405830746e-05,
      "loss": 1.888,
      "step": 191755
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.3825172185897827,
      "learning_rate": 4.06980072054911e-05,
      "loss": 2.0091,
      "step": 191756
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.5906071662902832,
      "learning_rate": 4.0695950053360624e-05,
      "loss": 2.081,
      "step": 191757
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.4387174844741821,
      "learning_rate": 4.069389294943962e-05,
      "loss": 2.0339,
      "step": 191758
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.3809062242507935,
      "learning_rate": 4.069183589372849e-05,
      "loss": 1.8974,
      "step": 191759
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.456028938293457,
      "learning_rate": 4.068977888622769e-05,
      "loss": 1.8529,
      "step": 191760
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.321427583694458,
      "learning_rate": 4.0687721926937486e-05,
      "loss": 2.0003,
      "step": 191761
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.541322112083435,
      "learning_rate": 4.0685665015858395e-05,
      "loss": 1.896,
      "step": 191762
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.2884557247161865,
      "learning_rate": 4.068360815299073e-05,
      "loss": 2.0424,
      "step": 191763
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.3095263242721558,
      "learning_rate": 4.06815513383349e-05,
      "loss": 1.8094,
      "step": 191764
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.3977606296539307,
      "learning_rate": 4.067949457189118e-05,
      "loss": 2.0082,
      "step": 191765
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.2620887756347656,
      "learning_rate": 4.067743785366009e-05,
      "loss": 2.0579,
      "step": 191766
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.4413626194000244,
      "learning_rate": 4.067538118364193e-05,
      "loss": 1.9384,
      "step": 191767
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.5108190774917603,
      "learning_rate": 4.067332456183714e-05,
      "loss": 1.9694,
      "step": 191768
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.2886990308761597,
      "learning_rate": 4.067126798824601e-05,
      "loss": 1.8531,
      "step": 191769
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.3180464506149292,
      "learning_rate": 4.0669211462869086e-05,
      "loss": 1.833,
      "step": 191770
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.3605961799621582,
      "learning_rate": 4.066715498570663e-05,
      "loss": 2.2576,
      "step": 191771
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.4362921714782715,
      "learning_rate": 4.066509855675904e-05,
      "loss": 1.9741,
      "step": 191772
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.3947545289993286,
      "learning_rate": 4.066304217602664e-05,
      "loss": 1.9629,
      "step": 191773
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.3880181312561035,
      "learning_rate": 4.0660985843509976e-05,
      "loss": 1.9886,
      "step": 191774
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.509161353111267,
      "learning_rate": 4.065892955920921e-05,
      "loss": 2.1645,
      "step": 191775
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.4209189414978027,
      "learning_rate": 4.0656873323124936e-05,
      "loss": 2.0583,
      "step": 191776
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.5666937828063965,
      "learning_rate": 4.065481713525747e-05,
      "loss": 1.9985,
      "step": 191777
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.4635558128356934,
      "learning_rate": 4.065276099560712e-05,
      "loss": 2.0641,
      "step": 191778
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.486385464668274,
      "learning_rate": 4.065070490417428e-05,
      "loss": 2.1353,
      "step": 191779
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.4291527271270752,
      "learning_rate": 4.064864886095943e-05,
      "loss": 1.9378,
      "step": 191780
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.478432536125183,
      "learning_rate": 4.0646592865962815e-05,
      "loss": 2.0088,
      "step": 191781
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.311047911643982,
      "learning_rate": 4.064453691918499e-05,
      "loss": 1.834,
      "step": 191782
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.4775384664535522,
      "learning_rate": 4.0642481020626225e-05,
      "loss": 2.0501,
      "step": 191783
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.4081617593765259,
      "learning_rate": 4.0640425170286826e-05,
      "loss": 1.9853,
      "step": 191784
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.4304225444793701,
      "learning_rate": 4.063836936816736e-05,
      "loss": 1.981,
      "step": 191785
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.3651009798049927,
      "learning_rate": 4.0636313614268115e-05,
      "loss": 2.0653,
      "step": 191786
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.3623753786087036,
      "learning_rate": 4.063425790858943e-05,
      "loss": 2.0159,
      "step": 191787
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.341201901435852,
      "learning_rate": 4.0632202251131776e-05,
      "loss": 2.0367,
      "step": 191788
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.431030035018921,
      "learning_rate": 4.063014664189548e-05,
      "loss": 1.9558,
      "step": 191789
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.4453763961791992,
      "learning_rate": 4.062809108088087e-05,
      "loss": 2.05,
      "step": 191790
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.39200758934021,
      "learning_rate": 4.06260355680885e-05,
      "loss": 1.8774,
      "step": 191791
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.387880563735962,
      "learning_rate": 4.062398010351855e-05,
      "loss": 2.2627,
      "step": 191792
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.4217537641525269,
      "learning_rate": 4.0621924687171556e-05,
      "loss": 1.9875,
      "step": 191793
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.3954306840896606,
      "learning_rate": 4.061986931904786e-05,
      "loss": 1.7524,
      "step": 191794
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.7719169855117798,
      "learning_rate": 4.061781399914782e-05,
      "loss": 2.0364,
      "step": 191795
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.6466721296310425,
      "learning_rate": 4.0615758727471736e-05,
      "loss": 2.1858,
      "step": 191796
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.8453006744384766,
      "learning_rate": 4.061370350402018e-05,
      "loss": 2.0702,
      "step": 191797
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.6207284927368164,
      "learning_rate": 4.061164832879334e-05,
      "loss": 2.0322,
      "step": 191798
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.3947343826293945,
      "learning_rate": 4.060959320179177e-05,
      "loss": 2.0925,
      "step": 191799
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.6453777551651,
      "learning_rate": 4.060753812301578e-05,
      "loss": 2.1344,
      "step": 191800
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.6013506650924683,
      "learning_rate": 4.060548309246575e-05,
      "loss": 2.0398,
      "step": 191801
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.7528876066207886,
      "learning_rate": 4.060342811014198e-05,
      "loss": 2.0107,
      "step": 191802
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.5286277532577515,
      "learning_rate": 4.0601373176045e-05,
      "loss": 2.1434,
      "step": 191803
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.5997543334960938,
      "learning_rate": 4.0599318290175035e-05,
      "loss": 1.9411,
      "step": 191804
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.497232437133789,
      "learning_rate": 4.059726345253263e-05,
      "loss": 1.9775,
      "step": 191805
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.3048789501190186,
      "learning_rate": 4.059520866311811e-05,
      "loss": 1.9422,
      "step": 191806
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.380187749862671,
      "learning_rate": 4.059315392193181e-05,
      "loss": 2.109,
      "step": 191807
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.7233593463897705,
      "learning_rate": 4.05910992289741e-05,
      "loss": 1.839,
      "step": 191808
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.4645341634750366,
      "learning_rate": 4.058904458424544e-05,
      "loss": 1.9661,
      "step": 191809
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.7235958576202393,
      "learning_rate": 4.058698998774611e-05,
      "loss": 2.2304,
      "step": 191810
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.4613311290740967,
      "learning_rate": 4.058493543947663e-05,
      "loss": 2.2982,
      "step": 191811
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.3251558542251587,
      "learning_rate": 4.0582880939437336e-05,
      "loss": 1.9586,
      "step": 191812
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.3511451482772827,
      "learning_rate": 4.058082648762853e-05,
      "loss": 2.1795,
      "step": 191813
    },
    {
      "epoch": 2.5,
      "grad_norm": 2.228471517562866,
      "learning_rate": 4.057877208405061e-05,
      "loss": 2.1131,
      "step": 191814
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.49013352394104,
      "learning_rate": 4.0576717728704065e-05,
      "loss": 1.986,
      "step": 191815
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.8507624864578247,
      "learning_rate": 4.0574663421589116e-05,
      "loss": 2.105,
      "step": 191816
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.4689455032348633,
      "learning_rate": 4.057260916270632e-05,
      "loss": 1.9923,
      "step": 191817
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.3183529376983643,
      "learning_rate": 4.057055495205597e-05,
      "loss": 2.0673,
      "step": 191818
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.5423964262008667,
      "learning_rate": 4.056850078963844e-05,
      "loss": 2.0038,
      "step": 191819
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.410470962524414,
      "learning_rate": 4.0566446675454055e-05,
      "loss": 2.0843,
      "step": 191820
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.3885366916656494,
      "learning_rate": 4.0564392609503325e-05,
      "loss": 1.9665,
      "step": 191821
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.5597659349441528,
      "learning_rate": 4.0562338591786514e-05,
      "loss": 1.987,
      "step": 191822
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.6810500621795654,
      "learning_rate": 4.056028462230415e-05,
      "loss": 1.9664,
      "step": 191823
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.3531568050384521,
      "learning_rate": 4.0558230701056505e-05,
      "loss": 1.7989,
      "step": 191824
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.4408143758773804,
      "learning_rate": 4.055617682804397e-05,
      "loss": 1.8782,
      "step": 191825
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.3044521808624268,
      "learning_rate": 4.055412300326689e-05,
      "loss": 2.0107,
      "step": 191826
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.413172960281372,
      "learning_rate": 4.055206922672576e-05,
      "loss": 2.1449,
      "step": 191827
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.3013267517089844,
      "learning_rate": 4.055001549842084e-05,
      "loss": 1.9946,
      "step": 191828
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.559005856513977,
      "learning_rate": 4.0547961818352634e-05,
      "loss": 2.2757,
      "step": 191829
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.4331203699111938,
      "learning_rate": 4.054590818652145e-05,
      "loss": 2.0168,
      "step": 191830
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.3350074291229248,
      "learning_rate": 4.054385460292767e-05,
      "loss": 2.0985,
      "step": 191831
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.4083900451660156,
      "learning_rate": 4.054180106757164e-05,
      "loss": 1.9702,
      "step": 191832
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.5432286262512207,
      "learning_rate": 4.053974758045386e-05,
      "loss": 1.9768,
      "step": 191833
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.3684803247451782,
      "learning_rate": 4.0537694141574564e-05,
      "loss": 2.0023,
      "step": 191834
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.3975520133972168,
      "learning_rate": 4.053564075093424e-05,
      "loss": 1.9245,
      "step": 191835
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.3780369758605957,
      "learning_rate": 4.05335874085332e-05,
      "loss": 2.0156,
      "step": 191836
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.334769606590271,
      "learning_rate": 4.0531534114372e-05,
      "loss": 2.0593,
      "step": 191837
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.3621312379837036,
      "learning_rate": 4.052948086845075e-05,
      "loss": 1.9271,
      "step": 191838
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.312448501586914,
      "learning_rate": 4.052742767077004e-05,
      "loss": 1.9073,
      "step": 191839
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.5289655923843384,
      "learning_rate": 4.052537452133011e-05,
      "loss": 1.9762,
      "step": 191840
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.5568732023239136,
      "learning_rate": 4.052332142013146e-05,
      "loss": 2.2714,
      "step": 191841
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.3178462982177734,
      "learning_rate": 4.052126836717438e-05,
      "loss": 1.8496,
      "step": 191842
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.3576858043670654,
      "learning_rate": 4.051921536245942e-05,
      "loss": 2.0411,
      "step": 191843
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.3260998725891113,
      "learning_rate": 4.051716240598667e-05,
      "loss": 2.03,
      "step": 191844
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.3520170450210571,
      "learning_rate": 4.051510949775679e-05,
      "loss": 1.9455,
      "step": 191845
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.4107985496520996,
      "learning_rate": 4.051305663776996e-05,
      "loss": 2.0609,
      "step": 191846
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.4276381731033325,
      "learning_rate": 4.0511003826026734e-05,
      "loss": 1.7657,
      "step": 191847
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.4087777137756348,
      "learning_rate": 4.0508951062527315e-05,
      "loss": 1.8996,
      "step": 191848
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.4222192764282227,
      "learning_rate": 4.0506898347272346e-05,
      "loss": 2.0536,
      "step": 191849
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.4767704010009766,
      "learning_rate": 4.050484568026188e-05,
      "loss": 1.8492,
      "step": 191850
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.397928237915039,
      "learning_rate": 4.050279306149655e-05,
      "loss": 2.1181,
      "step": 191851
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.364314079284668,
      "learning_rate": 4.050074049097657e-05,
      "loss": 2.0251,
      "step": 191852
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.315372109413147,
      "learning_rate": 4.049868796870249e-05,
      "loss": 1.784,
      "step": 191853
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.3703404664993286,
      "learning_rate": 4.049663549467452e-05,
      "loss": 2.1024,
      "step": 191854
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.3310450315475464,
      "learning_rate": 4.049458306889326e-05,
      "loss": 1.921,
      "step": 191855
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.3398916721343994,
      "learning_rate": 4.04925306913588e-05,
      "loss": 2.0968,
      "step": 191856
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.3971405029296875,
      "learning_rate": 4.0490478362071756e-05,
      "loss": 1.9684,
      "step": 191857
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.6001437902450562,
      "learning_rate": 4.048842608103238e-05,
      "loss": 2.0658,
      "step": 191858
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.5056947469711304,
      "learning_rate": 4.0486373848241174e-05,
      "loss": 2.0561,
      "step": 191859
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.5310101509094238,
      "learning_rate": 4.048432166369837e-05,
      "loss": 1.9532,
      "step": 191860
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.2858543395996094,
      "learning_rate": 4.0482269527404474e-05,
      "loss": 2.0821,
      "step": 191861
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.3021833896636963,
      "learning_rate": 4.048021743935985e-05,
      "loss": 2.1686,
      "step": 191862
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.4558236598968506,
      "learning_rate": 4.047816539956487e-05,
      "loss": 1.9222,
      "step": 191863
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.4526218175888062,
      "learning_rate": 4.047611340801978e-05,
      "loss": 1.9526,
      "step": 191864
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.4759845733642578,
      "learning_rate": 4.04740614647252e-05,
      "loss": 1.9235,
      "step": 191865
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.3756319284439087,
      "learning_rate": 4.04720095696813e-05,
      "loss": 2.1182,
      "step": 191866
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.3093737363815308,
      "learning_rate": 4.046995772288863e-05,
      "loss": 1.9029,
      "step": 191867
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.367404580116272,
      "learning_rate": 4.046790592434746e-05,
      "loss": 2.1619,
      "step": 191868
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.5783026218414307,
      "learning_rate": 4.046585417405815e-05,
      "loss": 1.9579,
      "step": 191869
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.424543023109436,
      "learning_rate": 4.046380247202122e-05,
      "loss": 2.0084,
      "step": 191870
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.2865921258926392,
      "learning_rate": 4.046175081823696e-05,
      "loss": 2.0229,
      "step": 191871
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.2976032495498657,
      "learning_rate": 4.04596992127057e-05,
      "loss": 2.0371,
      "step": 191872
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.3293870687484741,
      "learning_rate": 4.045764765542794e-05,
      "loss": 2.1779,
      "step": 191873
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.430160641670227,
      "learning_rate": 4.0455596146404016e-05,
      "loss": 1.9259,
      "step": 191874
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.5173163414001465,
      "learning_rate": 4.045354468563422e-05,
      "loss": 1.8991,
      "step": 191875
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.4780125617980957,
      "learning_rate": 4.045149327311906e-05,
      "loss": 2.1235,
      "step": 191876
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.4463924169540405,
      "learning_rate": 4.0449441908858903e-05,
      "loss": 1.8328,
      "step": 191877
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.472015380859375,
      "learning_rate": 4.0447390592854014e-05,
      "loss": 2.0475,
      "step": 191878
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.3542981147766113,
      "learning_rate": 4.044533932510488e-05,
      "loss": 2.016,
      "step": 191879
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.4605392217636108,
      "learning_rate": 4.044328810561189e-05,
      "loss": 2.0887,
      "step": 191880
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.382833480834961,
      "learning_rate": 4.044123693437533e-05,
      "loss": 2.0614,
      "step": 191881
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.3885953426361084,
      "learning_rate": 4.043918581139573e-05,
      "loss": 2.343,
      "step": 191882
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.535220742225647,
      "learning_rate": 4.0437134736673274e-05,
      "loss": 2.0212,
      "step": 191883
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.3112156391143799,
      "learning_rate": 4.043508371020854e-05,
      "loss": 1.8723,
      "step": 191884
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.4677422046661377,
      "learning_rate": 4.043303273200183e-05,
      "loss": 2.0775,
      "step": 191885
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.4725695848464966,
      "learning_rate": 4.0430981802053534e-05,
      "loss": 1.9781,
      "step": 191886
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.3299610614776611,
      "learning_rate": 4.042893092036392e-05,
      "loss": 1.9505,
      "step": 191887
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.4954837560653687,
      "learning_rate": 4.042688008693354e-05,
      "loss": 1.7994,
      "step": 191888
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.5497653484344482,
      "learning_rate": 4.042482930176266e-05,
      "loss": 1.9869,
      "step": 191889
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.555375337600708,
      "learning_rate": 4.042277856485174e-05,
      "loss": 1.8938,
      "step": 191890
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.4977824687957764,
      "learning_rate": 4.0420727876201155e-05,
      "loss": 2.0563,
      "step": 191891
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.2525651454925537,
      "learning_rate": 4.0418677235811225e-05,
      "loss": 1.8913,
      "step": 191892
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.3701393604278564,
      "learning_rate": 4.0416626643682325e-05,
      "loss": 2.0137,
      "step": 191893
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.4248079061508179,
      "learning_rate": 4.0414576099814924e-05,
      "loss": 2.076,
      "step": 191894
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.527278184890747,
      "learning_rate": 4.041252560420928e-05,
      "loss": 1.8808,
      "step": 191895
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.3498950004577637,
      "learning_rate": 4.041047515686593e-05,
      "loss": 1.9594,
      "step": 191896
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.339431643486023,
      "learning_rate": 4.0408424757785175e-05,
      "loss": 1.7974,
      "step": 191897
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.3056654930114746,
      "learning_rate": 4.0406374406967376e-05,
      "loss": 1.7489,
      "step": 191898
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.6088179349899292,
      "learning_rate": 4.040432410441288e-05,
      "loss": 2.0595,
      "step": 191899
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.4731004238128662,
      "learning_rate": 4.040227385012217e-05,
      "loss": 2.0386,
      "step": 191900
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.3752328157424927,
      "learning_rate": 4.040022364409552e-05,
      "loss": 1.9309,
      "step": 191901
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.4572968482971191,
      "learning_rate": 4.039817348633343e-05,
      "loss": 1.9051,
      "step": 191902
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.239255428314209,
      "learning_rate": 4.039612337683616e-05,
      "loss": 1.8445,
      "step": 191903
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.4667476415634155,
      "learning_rate": 4.039407331560429e-05,
      "loss": 2.1725,
      "step": 191904
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.417966604232788,
      "learning_rate": 4.0392023302637945e-05,
      "loss": 1.9143,
      "step": 191905
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.354282259941101,
      "learning_rate": 4.0389973337937656e-05,
      "loss": 1.9643,
      "step": 191906
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.4585728645324707,
      "learning_rate": 4.038792342150372e-05,
      "loss": 1.9967,
      "step": 191907
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.5642460584640503,
      "learning_rate": 4.038587355333665e-05,
      "loss": 2.1089,
      "step": 191908
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.3768454790115356,
      "learning_rate": 4.038382373343666e-05,
      "loss": 1.9331,
      "step": 191909
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.2878448963165283,
      "learning_rate": 4.0381773961804366e-05,
      "loss": 2.0343,
      "step": 191910
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.3756290674209595,
      "learning_rate": 4.0379724238439823e-05,
      "loss": 2.1029,
      "step": 191911
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.4348266124725342,
      "learning_rate": 4.037767456334371e-05,
      "loss": 2.0487,
      "step": 191912
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.5265675783157349,
      "learning_rate": 4.037562493651622e-05,
      "loss": 2.1557,
      "step": 191913
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.470595121383667,
      "learning_rate": 4.0373575357957845e-05,
      "loss": 2.0184,
      "step": 191914
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.413228988647461,
      "learning_rate": 4.037152582766886e-05,
      "loss": 2.1139,
      "step": 191915
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.6563084125518799,
      "learning_rate": 4.0369476345649834e-05,
      "loss": 1.9614,
      "step": 191916
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.675953984260559,
      "learning_rate": 4.036742691190089e-05,
      "loss": 2.0793,
      "step": 191917
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.511732578277588,
      "learning_rate": 4.036537752642264e-05,
      "loss": 2.0262,
      "step": 191918
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.384873390197754,
      "learning_rate": 4.036332818921527e-05,
      "loss": 2.0293,
      "step": 191919
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.4130773544311523,
      "learning_rate": 4.036127890027936e-05,
      "loss": 2.0516,
      "step": 191920
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.3593628406524658,
      "learning_rate": 4.035922965961507e-05,
      "loss": 1.8099,
      "step": 191921
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.4224567413330078,
      "learning_rate": 4.035718046722306e-05,
      "loss": 2.0303,
      "step": 191922
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.603631854057312,
      "learning_rate": 4.035513132310344e-05,
      "loss": 1.8108,
      "step": 191923
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.450803518295288,
      "learning_rate": 4.035308222725674e-05,
      "loss": 2.1187,
      "step": 191924
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.3308181762695312,
      "learning_rate": 4.035103317968326e-05,
      "loss": 1.9716,
      "step": 191925
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.2260301113128662,
      "learning_rate": 4.034898418038346e-05,
      "loss": 1.8103,
      "step": 191926
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.3334358930587769,
      "learning_rate": 4.0346935229357644e-05,
      "loss": 2.1316,
      "step": 191927
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.4129605293273926,
      "learning_rate": 4.034488632660635e-05,
      "loss": 2.0159,
      "step": 191928
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.4702314138412476,
      "learning_rate": 4.034283747212974e-05,
      "loss": 2.0061,
      "step": 191929
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.5317745208740234,
      "learning_rate": 4.034078866592832e-05,
      "loss": 1.9831,
      "step": 191930
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.5158603191375732,
      "learning_rate": 4.0338739908002406e-05,
      "loss": 1.981,
      "step": 191931
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.4455064535140991,
      "learning_rate": 4.0336691198352475e-05,
      "loss": 2.181,
      "step": 191932
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.4561915397644043,
      "learning_rate": 4.0334642536978834e-05,
      "loss": 2.1868,
      "step": 191933
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.3553434610366821,
      "learning_rate": 4.033259392388197e-05,
      "loss": 1.9039,
      "step": 191934
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.3509560823440552,
      "learning_rate": 4.033054535906206e-05,
      "loss": 2.1166,
      "step": 191935
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.3706438541412354,
      "learning_rate": 4.0328496842519665e-05,
      "loss": 2.0414,
      "step": 191936
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.4602171182632446,
      "learning_rate": 4.032644837425505e-05,
      "loss": 2.2186,
      "step": 191937
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.444027066230774,
      "learning_rate": 4.0324399954268715e-05,
      "loss": 2.0867,
      "step": 191938
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.4194530248641968,
      "learning_rate": 4.0322351582560894e-05,
      "loss": 2.0109,
      "step": 191939
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.2051604986190796,
      "learning_rate": 4.032030325913219e-05,
      "loss": 1.8925,
      "step": 191940
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.4234464168548584,
      "learning_rate": 4.03182549839827e-05,
      "loss": 2.0587,
      "step": 191941
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.3582249879837036,
      "learning_rate": 4.031620675711302e-05,
      "loss": 2.0872,
      "step": 191942
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.3752845525741577,
      "learning_rate": 4.031415857852343e-05,
      "loss": 2.1433,
      "step": 191943
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.2879360914230347,
      "learning_rate": 4.031211044821437e-05,
      "loss": 1.8572,
      "step": 191944
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.6299854516983032,
      "learning_rate": 4.0310062366186134e-05,
      "loss": 2.0594,
      "step": 191945
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.3213765621185303,
      "learning_rate": 4.030801433243921e-05,
      "loss": 2.1374,
      "step": 191946
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.3956842422485352,
      "learning_rate": 4.0305966346973925e-05,
      "loss": 2.2071,
      "step": 191947
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.4024301767349243,
      "learning_rate": 4.030391840979069e-05,
      "loss": 1.9343,
      "step": 191948
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.3367689847946167,
      "learning_rate": 4.030187052088977e-05,
      "loss": 1.9062,
      "step": 191949
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.264833688735962,
      "learning_rate": 4.029982268027172e-05,
      "loss": 1.8953,
      "step": 191950
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.415750503540039,
      "learning_rate": 4.029777488793675e-05,
      "loss": 1.9115,
      "step": 191951
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.343030571937561,
      "learning_rate": 4.029572714388543e-05,
      "loss": 1.8944,
      "step": 191952
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.5295910835266113,
      "learning_rate": 4.0293679448117976e-05,
      "loss": 1.8762,
      "step": 191953
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.3214097023010254,
      "learning_rate": 4.0291631800634804e-05,
      "loss": 2.092,
      "step": 191954
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.388073444366455,
      "learning_rate": 4.028958420143638e-05,
      "loss": 2.0412,
      "step": 191955
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.3617275953292847,
      "learning_rate": 4.028753665052302e-05,
      "loss": 1.9082,
      "step": 191956
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.3979488611221313,
      "learning_rate": 4.0285489147895045e-05,
      "loss": 2.0001,
      "step": 191957
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.5333434343338013,
      "learning_rate": 4.0283441693552976e-05,
      "loss": 1.9967,
      "step": 191958
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.4869896173477173,
      "learning_rate": 4.028139428749712e-05,
      "loss": 2.1326,
      "step": 191959
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.3858650922775269,
      "learning_rate": 4.027934692972776e-05,
      "loss": 2.0838,
      "step": 191960
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.3244086503982544,
      "learning_rate": 4.027729962024545e-05,
      "loss": 1.9448,
      "step": 191961
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.3950961828231812,
      "learning_rate": 4.0275252359050516e-05,
      "loss": 2.0099,
      "step": 191962
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.4211245775222778,
      "learning_rate": 4.0273205146143215e-05,
      "loss": 1.9309,
      "step": 191963
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.5700433254241943,
      "learning_rate": 4.0271157981524124e-05,
      "loss": 2.1593,
      "step": 191964
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.4185097217559814,
      "learning_rate": 4.026911086519351e-05,
      "loss": 2.0696,
      "step": 191965
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.428221583366394,
      "learning_rate": 4.02670637971517e-05,
      "loss": 2.1528,
      "step": 191966
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.3967318534851074,
      "learning_rate": 4.026501677739923e-05,
      "loss": 2.0564,
      "step": 191967
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.419107437133789,
      "learning_rate": 4.026296980593633e-05,
      "loss": 1.9772,
      "step": 191968
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.369974970817566,
      "learning_rate": 4.02609228827635e-05,
      "loss": 1.7924,
      "step": 191969
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.4246784448623657,
      "learning_rate": 4.025887600788108e-05,
      "loss": 2.2827,
      "step": 191970
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.5454366207122803,
      "learning_rate": 4.025682918128943e-05,
      "loss": 1.99,
      "step": 191971
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.5015578269958496,
      "learning_rate": 4.025478240298885e-05,
      "loss": 2.0763,
      "step": 191972
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.3704971075057983,
      "learning_rate": 4.0252735672979905e-05,
      "loss": 1.8426,
      "step": 191973
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.477343201637268,
      "learning_rate": 4.02506889912628e-05,
      "loss": 1.9522,
      "step": 191974
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.49648916721344,
      "learning_rate": 4.024864235783807e-05,
      "loss": 1.9596,
      "step": 191975
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.3826847076416016,
      "learning_rate": 4.0246595772706004e-05,
      "loss": 1.8645,
      "step": 191976
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.444541573524475,
      "learning_rate": 4.024454923586704e-05,
      "loss": 2.095,
      "step": 191977
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.4219906330108643,
      "learning_rate": 4.024250274732141e-05,
      "loss": 1.8922,
      "step": 191978
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.3652390241622925,
      "learning_rate": 4.024045630706969e-05,
      "loss": 2.0157,
      "step": 191979
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.4082365036010742,
      "learning_rate": 4.0238409915112104e-05,
      "loss": 2.154,
      "step": 191980
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.411731481552124,
      "learning_rate": 4.023636357144915e-05,
      "loss": 1.9096,
      "step": 191981
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.4372074604034424,
      "learning_rate": 4.0234317276081094e-05,
      "loss": 1.8778,
      "step": 191982
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.388306736946106,
      "learning_rate": 4.0232271029008546e-05,
      "loss": 1.9948,
      "step": 191983
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.3777648210525513,
      "learning_rate": 4.023022483023156e-05,
      "loss": 1.9485,
      "step": 191984
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.361265778541565,
      "learning_rate": 4.0228178679750746e-05,
      "loss": 1.7907,
      "step": 191985
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.4700329303741455,
      "learning_rate": 4.022613257756636e-05,
      "loss": 1.9352,
      "step": 191986
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.6163121461868286,
      "learning_rate": 4.0224086523678914e-05,
      "loss": 2.0461,
      "step": 191987
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.3597137928009033,
      "learning_rate": 4.022204051808863e-05,
      "loss": 1.9793,
      "step": 191988
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.3296045064926147,
      "learning_rate": 4.021999456079611e-05,
      "loss": 1.9359,
      "step": 191989
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.4894812107086182,
      "learning_rate": 4.021794865180146e-05,
      "loss": 1.9876,
      "step": 191990
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.3868408203125,
      "learning_rate": 4.021590279110527e-05,
      "loss": 1.9579,
      "step": 191991
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.5819498300552368,
      "learning_rate": 4.021385697870778e-05,
      "loss": 1.9043,
      "step": 191992
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.3591303825378418,
      "learning_rate": 4.021181121460952e-05,
      "loss": 1.9496,
      "step": 191993
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.4770870208740234,
      "learning_rate": 4.0209765498810696e-05,
      "loss": 1.9491,
      "step": 191994
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.4828553199768066,
      "learning_rate": 4.020771983131194e-05,
      "loss": 2.2727,
      "step": 191995
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.6576933860778809,
      "learning_rate": 4.0205674212113336e-05,
      "loss": 1.8371,
      "step": 191996
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.4110709428787231,
      "learning_rate": 4.020362864121543e-05,
      "loss": 1.9127,
      "step": 191997
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.4181736707687378,
      "learning_rate": 4.020158311861853e-05,
      "loss": 2.0668,
      "step": 191998
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.5040556192398071,
      "learning_rate": 4.019953764432315e-05,
      "loss": 2.0039,
      "step": 191999
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.4641833305358887,
      "learning_rate": 4.019749221832947e-05,
      "loss": 2.0232,
      "step": 192000
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.3999892473220825,
      "learning_rate": 4.019544684063816e-05,
      "loss": 2.0079,
      "step": 192001
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.5619492530822754,
      "learning_rate": 4.019340151124927e-05,
      "loss": 2.035,
      "step": 192002
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.4900399446487427,
      "learning_rate": 4.019135623016337e-05,
      "loss": 1.9828,
      "step": 192003
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.4395267963409424,
      "learning_rate": 4.018931099738077e-05,
      "loss": 2.1283,
      "step": 192004
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.4119926691055298,
      "learning_rate": 4.0187265812901936e-05,
      "loss": 2.1714,
      "step": 192005
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.3018102645874023,
      "learning_rate": 4.0185220676727125e-05,
      "loss": 1.972,
      "step": 192006
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.3976287841796875,
      "learning_rate": 4.018317558885691e-05,
      "loss": 2.0483,
      "step": 192007
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.5105130672454834,
      "learning_rate": 4.0181130549291416e-05,
      "loss": 1.9928,
      "step": 192008
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.2853680849075317,
      "learning_rate": 4.017908555803126e-05,
      "loss": 1.7948,
      "step": 192009
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.447817087173462,
      "learning_rate": 4.0177040615076615e-05,
      "loss": 1.9682,
      "step": 192010
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.381710410118103,
      "learning_rate": 4.0174995720428004e-05,
      "loss": 2.0725,
      "step": 192011
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.4406541585922241,
      "learning_rate": 4.017295087408575e-05,
      "loss": 2.1511,
      "step": 192012
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.4223637580871582,
      "learning_rate": 4.017090607605036e-05,
      "loss": 2.0975,
      "step": 192013
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.4602965116500854,
      "learning_rate": 4.016886132632196e-05,
      "loss": 1.791,
      "step": 192014
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.3478280305862427,
      "learning_rate": 4.0166816624901146e-05,
      "loss": 1.9674,
      "step": 192015
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.3694289922714233,
      "learning_rate": 4.0164771971788154e-05,
      "loss": 1.6473,
      "step": 192016
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.4943311214447021,
      "learning_rate": 4.0162727366983524e-05,
      "loss": 2.1673,
      "step": 192017
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.458862066268921,
      "learning_rate": 4.0160682810487454e-05,
      "loss": 1.8496,
      "step": 192018
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.5352863073349,
      "learning_rate": 4.015863830230057e-05,
      "loss": 1.996,
      "step": 192019
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.3680083751678467,
      "learning_rate": 4.015659384242298e-05,
      "loss": 2.0295,
      "step": 192020
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.416718602180481,
      "learning_rate": 4.0154549430855246e-05,
      "loss": 1.832,
      "step": 192021
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.4426895380020142,
      "learning_rate": 4.015250506759759e-05,
      "loss": 2.092,
      "step": 192022
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.3445464372634888,
      "learning_rate": 4.015046075265057e-05,
      "loss": 1.9321,
      "step": 192023
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.4066773653030396,
      "learning_rate": 4.0148416486014434e-05,
      "loss": 1.927,
      "step": 192024
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.495558738708496,
      "learning_rate": 4.014637226768972e-05,
      "loss": 1.9686,
      "step": 192025
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.3569390773773193,
      "learning_rate": 4.014432809767656e-05,
      "loss": 2.0067,
      "step": 192026
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.5210671424865723,
      "learning_rate": 4.0142283975975584e-05,
      "loss": 2.1903,
      "step": 192027
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.5005676746368408,
      "learning_rate": 4.014023990258697e-05,
      "loss": 1.9774,
      "step": 192028
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.5731838941574097,
      "learning_rate": 4.013819587751123e-05,
      "loss": 2.0129,
      "step": 192029
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.383239984512329,
      "learning_rate": 4.013615190074869e-05,
      "loss": 2.0454,
      "step": 192030
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.4684665203094482,
      "learning_rate": 4.01341079722998e-05,
      "loss": 1.9465,
      "step": 192031
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.4794505834579468,
      "learning_rate": 4.013206409216486e-05,
      "loss": 1.9783,
      "step": 192032
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.5220916271209717,
      "learning_rate": 4.0130020260344275e-05,
      "loss": 2.0601,
      "step": 192033
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.6077349185943604,
      "learning_rate": 4.012797647683838e-05,
      "loss": 2.0663,
      "step": 192034
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.4449933767318726,
      "learning_rate": 4.012593274164766e-05,
      "loss": 1.8897,
      "step": 192035
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.3370047807693481,
      "learning_rate": 4.012388905477236e-05,
      "loss": 2.0374,
      "step": 192036
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.5495941638946533,
      "learning_rate": 4.012184541621302e-05,
      "loss": 2.1708,
      "step": 192037
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.3853150606155396,
      "learning_rate": 4.011980182596992e-05,
      "loss": 1.905,
      "step": 192038
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.490049958229065,
      "learning_rate": 4.011775828404344e-05,
      "loss": 1.9662,
      "step": 192039
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.3921434879302979,
      "learning_rate": 4.011571479043392e-05,
      "loss": 2.0118,
      "step": 192040
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.3800537586212158,
      "learning_rate": 4.011367134514187e-05,
      "loss": 2.0998,
      "step": 192041
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.5511857271194458,
      "learning_rate": 4.0111627948167513e-05,
      "loss": 1.9492,
      "step": 192042
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.4877424240112305,
      "learning_rate": 4.01095845995114e-05,
      "loss": 2.1452,
      "step": 192043
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.4331823587417603,
      "learning_rate": 4.0107541299173806e-05,
      "loss": 2.0966,
      "step": 192044
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.3520007133483887,
      "learning_rate": 4.010549804715506e-05,
      "loss": 2.1844,
      "step": 192045
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.4333152770996094,
      "learning_rate": 4.0103454843455664e-05,
      "loss": 2.0102,
      "step": 192046
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.3641633987426758,
      "learning_rate": 4.0101411688075944e-05,
      "loss": 2.2093,
      "step": 192047
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.4583042860031128,
      "learning_rate": 4.009936858101621e-05,
      "loss": 1.9548,
      "step": 192048
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.3829411268234253,
      "learning_rate": 4.009732552227699e-05,
      "loss": 1.8591,
      "step": 192049
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.4751794338226318,
      "learning_rate": 4.009528251185858e-05,
      "loss": 1.9662,
      "step": 192050
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.4418076276779175,
      "learning_rate": 4.009323954976129e-05,
      "loss": 2.0992,
      "step": 192051
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.362231969833374,
      "learning_rate": 4.0091196635985645e-05,
      "loss": 2.1081,
      "step": 192052
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.532872200012207,
      "learning_rate": 4.008915377053188e-05,
      "loss": 1.9492,
      "step": 192053
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.3476837873458862,
      "learning_rate": 4.0087110953400524e-05,
      "loss": 1.8147,
      "step": 192054
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.3450363874435425,
      "learning_rate": 4.0085068184591884e-05,
      "loss": 1.903,
      "step": 192055
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.3647059202194214,
      "learning_rate": 4.0083025464106324e-05,
      "loss": 2.0967,
      "step": 192056
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.4658851623535156,
      "learning_rate": 4.008098279194417e-05,
      "loss": 2.1078,
      "step": 192057
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.4431594610214233,
      "learning_rate": 4.0078940168105936e-05,
      "loss": 2.0854,
      "step": 192058
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.472150206565857,
      "learning_rate": 4.0076897592591874e-05,
      "loss": 2.02,
      "step": 192059
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.546105980873108,
      "learning_rate": 4.007485506540249e-05,
      "loss": 1.8753,
      "step": 192060
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.4737598896026611,
      "learning_rate": 4.007281258653808e-05,
      "loss": 2.0535,
      "step": 192061
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.602393388748169,
      "learning_rate": 4.0070770155999055e-05,
      "loss": 2.0801,
      "step": 192062
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.3531594276428223,
      "learning_rate": 4.0068727773785704e-05,
      "loss": 1.8427,
      "step": 192063
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.370997428894043,
      "learning_rate": 4.006668543989856e-05,
      "loss": 2.1005,
      "step": 192064
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.4408631324768066,
      "learning_rate": 4.006464315433786e-05,
      "loss": 2.3075,
      "step": 192065
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.3995894193649292,
      "learning_rate": 4.0062600917104105e-05,
      "loss": 2.0799,
      "step": 192066
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.4036190509796143,
      "learning_rate": 4.006055872819756e-05,
      "loss": 2.0548,
      "step": 192067
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.3994921445846558,
      "learning_rate": 4.005851658761878e-05,
      "loss": 1.7947,
      "step": 192068
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.4414458274841309,
      "learning_rate": 4.005647449536792e-05,
      "loss": 1.8756,
      "step": 192069
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.3585377931594849,
      "learning_rate": 4.005443245144553e-05,
      "loss": 1.8825,
      "step": 192070
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.4585182666778564,
      "learning_rate": 4.0052390455851843e-05,
      "loss": 2.0783,
      "step": 192071
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.3967832326889038,
      "learning_rate": 4.0050348508587404e-05,
      "loss": 2.0399,
      "step": 192072
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.438693881034851,
      "learning_rate": 4.004830660965247e-05,
      "loss": 2.0096,
      "step": 192073
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.322163462638855,
      "learning_rate": 4.004626475904754e-05,
      "loss": 2.0964,
      "step": 192074
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.3718172311782837,
      "learning_rate": 4.004422295677282e-05,
      "loss": 1.9115,
      "step": 192075
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.4388079643249512,
      "learning_rate": 4.004218120282887e-05,
      "loss": 1.8347,
      "step": 192076
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.5139708518981934,
      "learning_rate": 4.004013949721589e-05,
      "loss": 1.9477,
      "step": 192077
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.3348259925842285,
      "learning_rate": 4.003809783993442e-05,
      "loss": 1.9948,
      "step": 192078
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.4714406728744507,
      "learning_rate": 4.003605623098473e-05,
      "loss": 2.1431,
      "step": 192079
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.3942760229110718,
      "learning_rate": 4.003401467036736e-05,
      "loss": 2.0409,
      "step": 192080
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.4833011627197266,
      "learning_rate": 4.003197315808244e-05,
      "loss": 1.892,
      "step": 192081
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.3451029062271118,
      "learning_rate": 4.002993169413056e-05,
      "loss": 2.0077,
      "step": 192082
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.3989933729171753,
      "learning_rate": 4.0027890278511944e-05,
      "loss": 1.8908,
      "step": 192083
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.3304082155227661,
      "learning_rate": 4.002584891122713e-05,
      "loss": 1.8955,
      "step": 192084
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.3616771697998047,
      "learning_rate": 4.002380759227632e-05,
      "loss": 2.0001,
      "step": 192085
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.3917945623397827,
      "learning_rate": 4.0021766321660156e-05,
      "loss": 1.8664,
      "step": 192086
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.60780930519104,
      "learning_rate": 4.001972509937872e-05,
      "loss": 1.9107,
      "step": 192087
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.5059916973114014,
      "learning_rate": 4.0017683925432584e-05,
      "loss": 1.7738,
      "step": 192088
    },
    {
      "epoch": 2.5,
      "grad_norm": 2.0027875900268555,
      "learning_rate": 4.001564279982199e-05,
      "loss": 2.1471,
      "step": 192089
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.4470146894454956,
      "learning_rate": 4.001360172254749e-05,
      "loss": 1.8656,
      "step": 192090
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.4602444171905518,
      "learning_rate": 4.00115606936093e-05,
      "loss": 2.0885,
      "step": 192091
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.4383982419967651,
      "learning_rate": 4.000951971300798e-05,
      "loss": 1.8211,
      "step": 192092
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.3648282289505005,
      "learning_rate": 4.000747878074369e-05,
      "loss": 2.0899,
      "step": 192093
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.4798059463500977,
      "learning_rate": 4.000543789681696e-05,
      "loss": 1.7648,
      "step": 192094
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.57157564163208,
      "learning_rate": 4.000339706122807e-05,
      "loss": 2.1238,
      "step": 192095
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.3698289394378662,
      "learning_rate": 4.0001356273977515e-05,
      "loss": 2.0136,
      "step": 192096
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.3458232879638672,
      "learning_rate": 3.999931553506556e-05,
      "loss": 1.6637,
      "step": 192097
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.5283925533294678,
      "learning_rate": 3.999727484449276e-05,
      "loss": 2.0395,
      "step": 192098
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.304484486579895,
      "learning_rate": 3.999523420225924e-05,
      "loss": 1.9021,
      "step": 192099
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.475093960762024,
      "learning_rate": 3.999319360836558e-05,
      "loss": 2.0165,
      "step": 192100
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.4398952722549438,
      "learning_rate": 3.9991153062812056e-05,
      "loss": 1.8346,
      "step": 192101
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.272921085357666,
      "learning_rate": 3.998911256559912e-05,
      "loss": 1.9361,
      "step": 192102
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.5167789459228516,
      "learning_rate": 3.998707211672706e-05,
      "loss": 2.1738,
      "step": 192103
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.3914462327957153,
      "learning_rate": 3.998503171619639e-05,
      "loss": 1.8737,
      "step": 192104
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.405119776725769,
      "learning_rate": 3.998299136400739e-05,
      "loss": 2.208,
      "step": 192105
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.4542222023010254,
      "learning_rate": 3.998095106016048e-05,
      "loss": 1.848,
      "step": 192106
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.4490050077438354,
      "learning_rate": 3.997891080465594e-05,
      "loss": 2.2111,
      "step": 192107
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.331314206123352,
      "learning_rate": 3.9976870597494295e-05,
      "loss": 2.1926,
      "step": 192108
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.5343295335769653,
      "learning_rate": 3.9974830438675806e-05,
      "loss": 2.1267,
      "step": 192109
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.3364453315734863,
      "learning_rate": 3.997279032820095e-05,
      "loss": 2.0358,
      "step": 192110
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.4461718797683716,
      "learning_rate": 3.9970750266070095e-05,
      "loss": 1.9883,
      "step": 192111
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.3512272834777832,
      "learning_rate": 3.996871025228353e-05,
      "loss": 1.8515,
      "step": 192112
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.4145739078521729,
      "learning_rate": 3.996667028684166e-05,
      "loss": 1.9578,
      "step": 192113
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.4200838804244995,
      "learning_rate": 3.9964630369744954e-05,
      "loss": 2.1106,
      "step": 192114
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.4845119714736938,
      "learning_rate": 3.996259050099367e-05,
      "loss": 1.8281,
      "step": 192115
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.3020910024642944,
      "learning_rate": 3.996055068058832e-05,
      "loss": 1.9701,
      "step": 192116
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.3801738023757935,
      "learning_rate": 3.9958510908529194e-05,
      "loss": 2.1187,
      "step": 192117
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.2834113836288452,
      "learning_rate": 3.995647118481673e-05,
      "loss": 1.9501,
      "step": 192118
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.4055052995681763,
      "learning_rate": 3.9954431509451154e-05,
      "loss": 1.8799,
      "step": 192119
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.566473126411438,
      "learning_rate": 3.9952391882433045e-05,
      "loss": 2.0524,
      "step": 192120
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.5299683809280396,
      "learning_rate": 3.995035230376262e-05,
      "loss": 2.1053,
      "step": 192121
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.310829520225525,
      "learning_rate": 3.9948312773440425e-05,
      "loss": 2.0215,
      "step": 192122
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.341038703918457,
      "learning_rate": 3.994627329146672e-05,
      "loss": 1.8625,
      "step": 192123
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.3393876552581787,
      "learning_rate": 3.9944233857841914e-05,
      "loss": 1.9171,
      "step": 192124
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.4865071773529053,
      "learning_rate": 3.9942194472566324e-05,
      "loss": 2.1244,
      "step": 192125
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.3135297298431396,
      "learning_rate": 3.994015513564046e-05,
      "loss": 2.0547,
      "step": 192126
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.4985305070877075,
      "learning_rate": 3.9938115847064556e-05,
      "loss": 2.1111,
      "step": 192127
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.555730938911438,
      "learning_rate": 3.993607660683914e-05,
      "loss": 2.3542,
      "step": 192128
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.4895135164260864,
      "learning_rate": 3.993403741496451e-05,
      "loss": 2.0795,
      "step": 192129
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.5727660655975342,
      "learning_rate": 3.993199827144098e-05,
      "loss": 2.0481,
      "step": 192130
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.6571470499038696,
      "learning_rate": 3.992995917626907e-05,
      "loss": 1.8924,
      "step": 192131
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.4619568586349487,
      "learning_rate": 3.992792012944907e-05,
      "loss": 2.0151,
      "step": 192132
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.4254764318466187,
      "learning_rate": 3.992588113098134e-05,
      "loss": 1.9927,
      "step": 192133
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.4200475215911865,
      "learning_rate": 3.992384218086636e-05,
      "loss": 2.099,
      "step": 192134
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.4409210681915283,
      "learning_rate": 3.992180327910447e-05,
      "loss": 1.8615,
      "step": 192135
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.4437440633773804,
      "learning_rate": 3.9919764425695935e-05,
      "loss": 1.9909,
      "step": 192136
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.4504839181900024,
      "learning_rate": 3.991772562064128e-05,
      "loss": 1.9894,
      "step": 192137
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.5071135759353638,
      "learning_rate": 3.991568686394085e-05,
      "loss": 2.0363,
      "step": 192138
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.4192620515823364,
      "learning_rate": 3.991364815559491e-05,
      "loss": 1.9986,
      "step": 192139
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.2925772666931152,
      "learning_rate": 3.991160949560402e-05,
      "loss": 1.9611,
      "step": 192140
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.318304181098938,
      "learning_rate": 3.990957088396849e-05,
      "loss": 2.0228,
      "step": 192141
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.4075063467025757,
      "learning_rate": 3.9907532320688576e-05,
      "loss": 2.0344,
      "step": 192142
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.2771176099777222,
      "learning_rate": 3.9905493805764845e-05,
      "loss": 2.018,
      "step": 192143
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.5001471042633057,
      "learning_rate": 3.9903455339197534e-05,
      "loss": 2.006,
      "step": 192144
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.4882086515426636,
      "learning_rate": 3.990141692098714e-05,
      "loss": 2.0947,
      "step": 192145
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.398476004600525,
      "learning_rate": 3.989937855113397e-05,
      "loss": 2.0346,
      "step": 192146
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.3647211790084839,
      "learning_rate": 3.989734022963844e-05,
      "loss": 2.1365,
      "step": 192147
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.3648921251296997,
      "learning_rate": 3.98953019565008e-05,
      "loss": 1.8817,
      "step": 192148
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.2801470756530762,
      "learning_rate": 3.989326373172165e-05,
      "loss": 1.9644,
      "step": 192149
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.4868009090423584,
      "learning_rate": 3.989122555530114e-05,
      "loss": 1.921,
      "step": 192150
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.4283009767532349,
      "learning_rate": 3.9889187427239856e-05,
      "loss": 1.9729,
      "step": 192151
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.4912474155426025,
      "learning_rate": 3.988714934753802e-05,
      "loss": 2.1087,
      "step": 192152
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.3668620586395264,
      "learning_rate": 3.988511131619616e-05,
      "loss": 2.2162,
      "step": 192153
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.4341299533843994,
      "learning_rate": 3.9883073333214485e-05,
      "loss": 2.1395,
      "step": 192154
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.4276551008224487,
      "learning_rate": 3.98810353985935e-05,
      "loss": 1.9796,
      "step": 192155
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.4953914880752563,
      "learning_rate": 3.987899751233349e-05,
      "loss": 2.105,
      "step": 192156
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.466059923171997,
      "learning_rate": 3.9876959674434926e-05,
      "loss": 2.0825,
      "step": 192157
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.3171191215515137,
      "learning_rate": 3.9874921884898116e-05,
      "loss": 1.9365,
      "step": 192158
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.436488389968872,
      "learning_rate": 3.9872884143723546e-05,
      "loss": 1.9144,
      "step": 192159
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.179011344909668,
      "learning_rate": 3.9870846450911434e-05,
      "loss": 1.9543,
      "step": 192160
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.4253724813461304,
      "learning_rate": 3.9868808806462294e-05,
      "loss": 2.0933,
      "step": 192161
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.3289846181869507,
      "learning_rate": 3.986677121037637e-05,
      "loss": 1.9479,
      "step": 192162
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.5183193683624268,
      "learning_rate": 3.9864733662654225e-05,
      "loss": 1.9525,
      "step": 192163
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.440083384513855,
      "learning_rate": 3.986269616329603e-05,
      "loss": 2.2237,
      "step": 192164
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.303858757019043,
      "learning_rate": 3.986065871230244e-05,
      "loss": 1.9718,
      "step": 192165
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.4387199878692627,
      "learning_rate": 3.98586213096735e-05,
      "loss": 2.1461,
      "step": 192166
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.3491878509521484,
      "learning_rate": 3.985658395540984e-05,
      "loss": 1.8709,
      "step": 192167
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.4626905918121338,
      "learning_rate": 3.985454664951169e-05,
      "loss": 2.0433,
      "step": 192168
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.6013062000274658,
      "learning_rate": 3.9852509391979516e-05,
      "loss": 2.2447,
      "step": 192169
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.5062702894210815,
      "learning_rate": 3.9850472182813655e-05,
      "loss": 2.0096,
      "step": 192170
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.4929486513137817,
      "learning_rate": 3.9848435022014534e-05,
      "loss": 1.9948,
      "step": 192171
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.323170781135559,
      "learning_rate": 3.984639790958249e-05,
      "loss": 2.2491,
      "step": 192172
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.4431174993515015,
      "learning_rate": 3.984436084551796e-05,
      "loss": 2.0639,
      "step": 192173
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.5155069828033447,
      "learning_rate": 3.9842323829821176e-05,
      "loss": 2.1366,
      "step": 192174
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.3463099002838135,
      "learning_rate": 3.98402868624927e-05,
      "loss": 1.9636,
      "step": 192175
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.439846396446228,
      "learning_rate": 3.983824994353273e-05,
      "loss": 2.0029,
      "step": 192176
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.805017352104187,
      "learning_rate": 3.9836213072941834e-05,
      "loss": 2.064,
      "step": 192177
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.3817174434661865,
      "learning_rate": 3.983417625072028e-05,
      "loss": 1.8757,
      "step": 192178
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.324405550956726,
      "learning_rate": 3.983213947686847e-05,
      "loss": 1.8412,
      "step": 192179
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.3152838945388794,
      "learning_rate": 3.983010275138671e-05,
      "loss": 2.0395,
      "step": 192180
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.696777582168579,
      "learning_rate": 3.982806607427551e-05,
      "loss": 1.9522,
      "step": 192181
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.3707274198532104,
      "learning_rate": 3.982602944553512e-05,
      "loss": 1.9241,
      "step": 192182
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.5093121528625488,
      "learning_rate": 3.982399286516604e-05,
      "loss": 1.9764,
      "step": 192183
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.4802910089492798,
      "learning_rate": 3.9821956333168604e-05,
      "loss": 1.8263,
      "step": 192184
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.4789844751358032,
      "learning_rate": 3.981991984954317e-05,
      "loss": 1.9702,
      "step": 192185
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.4072247743606567,
      "learning_rate": 3.981788341429004e-05,
      "loss": 2.005,
      "step": 192186
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.2977176904678345,
      "learning_rate": 3.981584702740975e-05,
      "loss": 1.9347,
      "step": 192187
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.4702425003051758,
      "learning_rate": 3.981381068890257e-05,
      "loss": 1.9875,
      "step": 192188
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.4051247835159302,
      "learning_rate": 3.9811774398768955e-05,
      "loss": 2.1522,
      "step": 192189
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.5499829053878784,
      "learning_rate": 3.980973815700925e-05,
      "loss": 1.755,
      "step": 192190
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.4859083890914917,
      "learning_rate": 3.980770196362382e-05,
      "loss": 1.9198,
      "step": 192191
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.3842980861663818,
      "learning_rate": 3.980566581861299e-05,
      "loss": 1.8621,
      "step": 192192
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.3689172267913818,
      "learning_rate": 3.980362972197726e-05,
      "loss": 1.7863,
      "step": 192193
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.3305364847183228,
      "learning_rate": 3.9801593673716904e-05,
      "loss": 1.7929,
      "step": 192194
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.4502851963043213,
      "learning_rate": 3.979955767383238e-05,
      "loss": 1.9631,
      "step": 192195
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.4754751920700073,
      "learning_rate": 3.979752172232403e-05,
      "loss": 1.987,
      "step": 192196
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.5769314765930176,
      "learning_rate": 3.979548581919225e-05,
      "loss": 2.0974,
      "step": 192197
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.4718588590621948,
      "learning_rate": 3.979344996443731e-05,
      "loss": 2.0175,
      "step": 192198
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.4089514017105103,
      "learning_rate": 3.979141415805976e-05,
      "loss": 2.0768,
      "step": 192199
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.3636806011199951,
      "learning_rate": 3.9789378400059857e-05,
      "loss": 2.0953,
      "step": 192200
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.4227229356765747,
      "learning_rate": 3.978734269043805e-05,
      "loss": 1.931,
      "step": 192201
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.2760571241378784,
      "learning_rate": 3.978530702919471e-05,
      "loss": 1.9446,
      "step": 192202
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.6258901357650757,
      "learning_rate": 3.9783271416330175e-05,
      "loss": 2.19,
      "step": 192203
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.3901032209396362,
      "learning_rate": 3.978123585184477e-05,
      "loss": 1.9546,
      "step": 192204
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.5552654266357422,
      "learning_rate": 3.977920033573904e-05,
      "loss": 1.6868,
      "step": 192205
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.455187201499939,
      "learning_rate": 3.97771648680132e-05,
      "loss": 1.8503,
      "step": 192206
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.4852269887924194,
      "learning_rate": 3.977512944866773e-05,
      "loss": 1.9544,
      "step": 192207
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.4167399406433105,
      "learning_rate": 3.977309407770299e-05,
      "loss": 1.9959,
      "step": 192208
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.3868602514266968,
      "learning_rate": 3.977105875511936e-05,
      "loss": 1.9976,
      "step": 192209
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.4423567056655884,
      "learning_rate": 3.976902348091715e-05,
      "loss": 2.1442,
      "step": 192210
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.3050323724746704,
      "learning_rate": 3.976698825509681e-05,
      "loss": 2.02,
      "step": 192211
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.3653424978256226,
      "learning_rate": 3.9764953077658665e-05,
      "loss": 2.1322,
      "step": 192212
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.3356705904006958,
      "learning_rate": 3.976291794860319e-05,
      "loss": 2.1739,
      "step": 192213
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.3423106670379639,
      "learning_rate": 3.976088286793071e-05,
      "loss": 1.9053,
      "step": 192214
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.5868507623672485,
      "learning_rate": 3.97588478356415e-05,
      "loss": 2.1571,
      "step": 192215
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.6112895011901855,
      "learning_rate": 3.9756812851736106e-05,
      "loss": 1.899,
      "step": 192216
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.4057562351226807,
      "learning_rate": 3.9754777916214856e-05,
      "loss": 1.8946,
      "step": 192217
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.3214484453201294,
      "learning_rate": 3.9752743029078025e-05,
      "loss": 1.9086,
      "step": 192218
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.3280857801437378,
      "learning_rate": 3.975070819032613e-05,
      "loss": 1.9205,
      "step": 192219
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.4116156101226807,
      "learning_rate": 3.97486733999595e-05,
      "loss": 2.1326,
      "step": 192220
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.4088044166564941,
      "learning_rate": 3.9746638657978426e-05,
      "loss": 1.9044,
      "step": 192221
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.3754446506500244,
      "learning_rate": 3.974460396438345e-05,
      "loss": 1.881,
      "step": 192222
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.5040255784988403,
      "learning_rate": 3.9742569319174866e-05,
      "loss": 2.2294,
      "step": 192223
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.4249815940856934,
      "learning_rate": 3.9740534722352946e-05,
      "loss": 1.7767,
      "step": 192224
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.4935498237609863,
      "learning_rate": 3.9738500173918286e-05,
      "loss": 1.9132,
      "step": 192225
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.3820487260818481,
      "learning_rate": 3.973646567387112e-05,
      "loss": 1.6867,
      "step": 192226
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.310825228691101,
      "learning_rate": 3.973443122221182e-05,
      "loss": 2.0538,
      "step": 192227
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.4251688718795776,
      "learning_rate": 3.973239681894084e-05,
      "loss": 1.9538,
      "step": 192228
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.3887090682983398,
      "learning_rate": 3.9730362464058484e-05,
      "loss": 1.9932,
      "step": 192229
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.3328027725219727,
      "learning_rate": 3.972832815756519e-05,
      "loss": 2.1997,
      "step": 192230
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.402936577796936,
      "learning_rate": 3.972629389946136e-05,
      "loss": 1.8921,
      "step": 192231
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.5128591060638428,
      "learning_rate": 3.972425968974729e-05,
      "loss": 1.9524,
      "step": 192232
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.33913254737854,
      "learning_rate": 3.9722225528423367e-05,
      "loss": 1.8752,
      "step": 192233
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.4415205717086792,
      "learning_rate": 3.972019141549001e-05,
      "loss": 2.1196,
      "step": 192234
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.408193588256836,
      "learning_rate": 3.971815735094755e-05,
      "loss": 2.0686,
      "step": 192235
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.3715875148773193,
      "learning_rate": 3.971612333479647e-05,
      "loss": 1.8973,
      "step": 192236
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.3095319271087646,
      "learning_rate": 3.9714089367036984e-05,
      "loss": 1.9446,
      "step": 192237
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.4700632095336914,
      "learning_rate": 3.971205544766966e-05,
      "loss": 2.1926,
      "step": 192238
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.5386216640472412,
      "learning_rate": 3.9710021576694754e-05,
      "loss": 2.1192,
      "step": 192239
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.350203037261963,
      "learning_rate": 3.970798775411267e-05,
      "loss": 2.1773,
      "step": 192240
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.3401517868041992,
      "learning_rate": 3.970595397992371e-05,
      "loss": 2.1044,
      "step": 192241
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.442315697669983,
      "learning_rate": 3.970392025412841e-05,
      "loss": 2.2376,
      "step": 192242
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.4123870134353638,
      "learning_rate": 3.9701886576726995e-05,
      "loss": 2.1237,
      "step": 192243
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.4702953100204468,
      "learning_rate": 3.969985294771997e-05,
      "loss": 1.8162,
      "step": 192244
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.3608784675598145,
      "learning_rate": 3.969781936710766e-05,
      "loss": 2.2052,
      "step": 192245
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.4052245616912842,
      "learning_rate": 3.9695785834890445e-05,
      "loss": 1.9784,
      "step": 192246
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.3632529973983765,
      "learning_rate": 3.969375235106862e-05,
      "loss": 2.2579,
      "step": 192247
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.4644970893859863,
      "learning_rate": 3.969171891564271e-05,
      "loss": 2.2162,
      "step": 192248
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.3795275688171387,
      "learning_rate": 3.968968552861296e-05,
      "loss": 2.0306,
      "step": 192249
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.4230237007141113,
      "learning_rate": 3.968765218997989e-05,
      "loss": 2.0702,
      "step": 192250
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.365028977394104,
      "learning_rate": 3.9685618899743775e-05,
      "loss": 1.8308,
      "step": 192251
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.444206714630127,
      "learning_rate": 3.9683585657905045e-05,
      "loss": 1.8133,
      "step": 192252
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.393733024597168,
      "learning_rate": 3.968155246446394e-05,
      "loss": 2.1074,
      "step": 192253
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.3124490976333618,
      "learning_rate": 3.967951931942105e-05,
      "loss": 1.7968,
      "step": 192254
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.4841569662094116,
      "learning_rate": 3.967748622277658e-05,
      "loss": 2.0889,
      "step": 192255
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.2248789072036743,
      "learning_rate": 3.967545317453106e-05,
      "loss": 1.7095,
      "step": 192256
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.3759007453918457,
      "learning_rate": 3.967342017468476e-05,
      "loss": 2.0392,
      "step": 192257
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.4282398223876953,
      "learning_rate": 3.967138722323811e-05,
      "loss": 1.8202,
      "step": 192258
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.511515736579895,
      "learning_rate": 3.9669354320191374e-05,
      "loss": 2.2591,
      "step": 192259
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.4727509021759033,
      "learning_rate": 3.9667321465545086e-05,
      "loss": 1.8598,
      "step": 192260
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.2792418003082275,
      "learning_rate": 3.966528865929949e-05,
      "loss": 1.919,
      "step": 192261
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.4689103364944458,
      "learning_rate": 3.9663255901455106e-05,
      "loss": 2.0221,
      "step": 192262
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.5103156566619873,
      "learning_rate": 3.9661223192012234e-05,
      "loss": 2.1357,
      "step": 192263
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.5406241416931152,
      "learning_rate": 3.9659190530971284e-05,
      "loss": 2.142,
      "step": 192264
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.3170469999313354,
      "learning_rate": 3.9657157918332514e-05,
      "loss": 1.7968,
      "step": 192265
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.6168010234832764,
      "learning_rate": 3.9655125354096426e-05,
      "loss": 2.003,
      "step": 192266
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.375168800354004,
      "learning_rate": 3.965309283826336e-05,
      "loss": 2.0796,
      "step": 192267
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.624833345413208,
      "learning_rate": 3.965106037083373e-05,
      "loss": 2.0933,
      "step": 192268
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.5827449560165405,
      "learning_rate": 3.964902795180789e-05,
      "loss": 2.1518,
      "step": 192269
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.3734720945358276,
      "learning_rate": 3.964699558118619e-05,
      "loss": 2.0299,
      "step": 192270
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.5156582593917847,
      "learning_rate": 3.9644963258968976e-05,
      "loss": 2.0894,
      "step": 192271
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.3834774494171143,
      "learning_rate": 3.964293098515674e-05,
      "loss": 1.8625,
      "step": 192272
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.3704084157943726,
      "learning_rate": 3.964089875974975e-05,
      "loss": 2.1039,
      "step": 192273
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.3933156728744507,
      "learning_rate": 3.9638866582748474e-05,
      "loss": 1.784,
      "step": 192274
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.3773282766342163,
      "learning_rate": 3.9636834454153275e-05,
      "loss": 1.9905,
      "step": 192275
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.3045282363891602,
      "learning_rate": 3.963480237396449e-05,
      "loss": 1.974,
      "step": 192276
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.3418210744857788,
      "learning_rate": 3.9632770342182455e-05,
      "loss": 2.0046,
      "step": 192277
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.4156849384307861,
      "learning_rate": 3.963073835880766e-05,
      "loss": 1.8021,
      "step": 192278
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.313458800315857,
      "learning_rate": 3.962870642384034e-05,
      "loss": 1.6991,
      "step": 192279
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.4372859001159668,
      "learning_rate": 3.9626674537281035e-05,
      "loss": 2.05,
      "step": 192280
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.4336782693862915,
      "learning_rate": 3.962464269913004e-05,
      "loss": 1.9601,
      "step": 192281
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.4121170043945312,
      "learning_rate": 3.9622610909387755e-05,
      "loss": 2.0408,
      "step": 192282
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.353953242301941,
      "learning_rate": 3.9620579168054475e-05,
      "loss": 1.6721,
      "step": 192283
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.4485160112380981,
      "learning_rate": 3.961854747513071e-05,
      "loss": 1.9488,
      "step": 192284
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.469383716583252,
      "learning_rate": 3.9616515830616693e-05,
      "loss": 2.0984,
      "step": 192285
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.364647388458252,
      "learning_rate": 3.961448423451298e-05,
      "loss": 1.8851,
      "step": 192286
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.5110145807266235,
      "learning_rate": 3.9612452686819805e-05,
      "loss": 1.9507,
      "step": 192287
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.377395749092102,
      "learning_rate": 3.961042118753761e-05,
      "loss": 1.991,
      "step": 192288
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.4117721319198608,
      "learning_rate": 3.960838973666669e-05,
      "loss": 1.971,
      "step": 192289
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.4090086221694946,
      "learning_rate": 3.960635833420754e-05,
      "loss": 2.0969,
      "step": 192290
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.2983907461166382,
      "learning_rate": 3.960432698016044e-05,
      "loss": 1.9975,
      "step": 192291
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.825734257698059,
      "learning_rate": 3.960229567452588e-05,
      "loss": 2.2757,
      "step": 192292
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.5013347864151,
      "learning_rate": 3.960026441730412e-05,
      "loss": 2.0818,
      "step": 192293
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.5656875371932983,
      "learning_rate": 3.9598233208495635e-05,
      "loss": 2.0041,
      "step": 192294
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.3929779529571533,
      "learning_rate": 3.9596202048100654e-05,
      "loss": 2.1,
      "step": 192295
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.5814851522445679,
      "learning_rate": 3.959417093611975e-05,
      "loss": 1.8237,
      "step": 192296
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.3411799669265747,
      "learning_rate": 3.959213987255312e-05,
      "loss": 1.9922,
      "step": 192297
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.6438056230545044,
      "learning_rate": 3.9590108857401295e-05,
      "loss": 2.1254,
      "step": 192298
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.4610522985458374,
      "learning_rate": 3.95880778906646e-05,
      "loss": 2.1781,
      "step": 192299
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.368966817855835,
      "learning_rate": 3.958604697234339e-05,
      "loss": 2.1757,
      "step": 192300
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.490566372871399,
      "learning_rate": 3.958401610243798e-05,
      "loss": 2.0241,
      "step": 192301
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.4307119846343994,
      "learning_rate": 3.958198528094887e-05,
      "loss": 1.9305,
      "step": 192302
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.4167112112045288,
      "learning_rate": 3.957995450787633e-05,
      "loss": 1.8791,
      "step": 192303
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.4555277824401855,
      "learning_rate": 3.957792378322087e-05,
      "loss": 1.7875,
      "step": 192304
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.2624320983886719,
      "learning_rate": 3.95758931069827e-05,
      "loss": 1.5724,
      "step": 192305
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.344571828842163,
      "learning_rate": 3.957386247916238e-05,
      "loss": 1.8968,
      "step": 192306
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.5861234664916992,
      "learning_rate": 3.957183189976019e-05,
      "loss": 2.1508,
      "step": 192307
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.449170708656311,
      "learning_rate": 3.95698013687765e-05,
      "loss": 2.1184,
      "step": 192308
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.6904420852661133,
      "learning_rate": 3.9567770886211645e-05,
      "loss": 1.9396,
      "step": 192309
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.456788182258606,
      "learning_rate": 3.9565740452066094e-05,
      "loss": 1.9321,
      "step": 192310
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.3458608388900757,
      "learning_rate": 3.956371006634015e-05,
      "loss": 2.3867,
      "step": 192311
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.3654474020004272,
      "learning_rate": 3.95616797290343e-05,
      "loss": 1.8683,
      "step": 192312
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.6283164024353027,
      "learning_rate": 3.955964944014882e-05,
      "loss": 1.8382,
      "step": 192313
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.4350258111953735,
      "learning_rate": 3.9557619199684074e-05,
      "loss": 1.9025,
      "step": 192314
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.4568086862564087,
      "learning_rate": 3.9555589007640564e-05,
      "loss": 1.9177,
      "step": 192315
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.3993405103683472,
      "learning_rate": 3.955355886401855e-05,
      "loss": 1.9936,
      "step": 192316
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.3191527128219604,
      "learning_rate": 3.9551528768818405e-05,
      "loss": 2.0045,
      "step": 192317
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.336262583732605,
      "learning_rate": 3.95494987220406e-05,
      "loss": 1.9902,
      "step": 192318
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.4230351448059082,
      "learning_rate": 3.954746872368545e-05,
      "loss": 2.0508,
      "step": 192319
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.554681420326233,
      "learning_rate": 3.95454387737533e-05,
      "loss": 2.2433,
      "step": 192320
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.3193185329437256,
      "learning_rate": 3.954340887224462e-05,
      "loss": 1.9776,
      "step": 192321
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.4771333932876587,
      "learning_rate": 3.9541379019159766e-05,
      "loss": 1.9949,
      "step": 192322
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.3172650337219238,
      "learning_rate": 3.953934921449898e-05,
      "loss": 2.0117,
      "step": 192323
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.4873106479644775,
      "learning_rate": 3.9537319458262826e-05,
      "loss": 2.069,
      "step": 192324
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.3162121772766113,
      "learning_rate": 3.9535289750451605e-05,
      "loss": 1.8174,
      "step": 192325
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.5742741823196411,
      "learning_rate": 3.9533260091065644e-05,
      "loss": 2.0859,
      "step": 192326
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.3772224187850952,
      "learning_rate": 3.9531230480105383e-05,
      "loss": 1.9733,
      "step": 192327
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.2924405336380005,
      "learning_rate": 3.952920091757115e-05,
      "loss": 1.8608,
      "step": 192328
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.4430878162384033,
      "learning_rate": 3.9527171403463406e-05,
      "loss": 1.9282,
      "step": 192329
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.3697466850280762,
      "learning_rate": 3.95251419377825e-05,
      "loss": 2.0498,
      "step": 192330
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.34781813621521,
      "learning_rate": 3.952311252052879e-05,
      "loss": 2.0236,
      "step": 192331
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.28971266746521,
      "learning_rate": 3.952108315170257e-05,
      "loss": 1.8925,
      "step": 192332
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.285157561302185,
      "learning_rate": 3.951905383130435e-05,
      "loss": 1.7816,
      "step": 192333
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.5979989767074585,
      "learning_rate": 3.951702455933442e-05,
      "loss": 1.981,
      "step": 192334
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.4272234439849854,
      "learning_rate": 3.951499533579322e-05,
      "loss": 2.0951,
      "step": 192335
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.5152784585952759,
      "learning_rate": 3.951296616068111e-05,
      "loss": 1.9879,
      "step": 192336
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.3845199346542358,
      "learning_rate": 3.951093703399847e-05,
      "loss": 2.1847,
      "step": 192337
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.5147466659545898,
      "learning_rate": 3.950890795574562e-05,
      "loss": 1.9164,
      "step": 192338
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.4479707479476929,
      "learning_rate": 3.950687892592299e-05,
      "loss": 1.872,
      "step": 192339
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.3898379802703857,
      "learning_rate": 3.950484994453093e-05,
      "loss": 1.9798,
      "step": 192340
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.4328495264053345,
      "learning_rate": 3.950282101156988e-05,
      "loss": 2.2127,
      "step": 192341
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.529394507408142,
      "learning_rate": 3.950079212704017e-05,
      "loss": 2.0364,
      "step": 192342
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.4631975889205933,
      "learning_rate": 3.949876329094221e-05,
      "loss": 1.9989,
      "step": 192343
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.3122001886367798,
      "learning_rate": 3.949673450327625e-05,
      "loss": 2.0316,
      "step": 192344
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.4889730215072632,
      "learning_rate": 3.949470576404284e-05,
      "loss": 1.9845,
      "step": 192345
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.497418761253357,
      "learning_rate": 3.949267707324223e-05,
      "loss": 2.0265,
      "step": 192346
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.4531667232513428,
      "learning_rate": 3.9490648430874875e-05,
      "loss": 2.1812,
      "step": 192347
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.4921584129333496,
      "learning_rate": 3.948861983694117e-05,
      "loss": 2.039,
      "step": 192348
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.4362940788269043,
      "learning_rate": 3.948659129144143e-05,
      "loss": 2.0443,
      "step": 192349
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.4057610034942627,
      "learning_rate": 3.948456279437597e-05,
      "loss": 1.8769,
      "step": 192350
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.3072439432144165,
      "learning_rate": 3.9482534345745344e-05,
      "loss": 1.8796,
      "step": 192351
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.318534255027771,
      "learning_rate": 3.948050594554973e-05,
      "loss": 2.1906,
      "step": 192352
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.4461417198181152,
      "learning_rate": 3.9478477593789715e-05,
      "loss": 2.0719,
      "step": 192353
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.7550469636917114,
      "learning_rate": 3.9476449290465515e-05,
      "loss": 2.0021,
      "step": 192354
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.5556467771530151,
      "learning_rate": 3.9474421035577607e-05,
      "loss": 1.8573,
      "step": 192355
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.3339682817459106,
      "learning_rate": 3.9472392829126254e-05,
      "loss": 2.0378,
      "step": 192356
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.5395246744155884,
      "learning_rate": 3.947036467111195e-05,
      "loss": 2.2137,
      "step": 192357
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.3566839694976807,
      "learning_rate": 3.9468336561534944e-05,
      "loss": 2.1376,
      "step": 192358
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.5584958791732788,
      "learning_rate": 3.946630850039578e-05,
      "loss": 2.2387,
      "step": 192359
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.349929928779602,
      "learning_rate": 3.946428048769474e-05,
      "loss": 1.8875,
      "step": 192360
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.384405255317688,
      "learning_rate": 3.946225252343223e-05,
      "loss": 1.9897,
      "step": 192361
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.59697687625885,
      "learning_rate": 3.9460224607608525e-05,
      "loss": 2.0455,
      "step": 192362
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.5273571014404297,
      "learning_rate": 3.9458196740224115e-05,
      "loss": 1.7653,
      "step": 192363
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.3470269441604614,
      "learning_rate": 3.945616892127932e-05,
      "loss": 2.1362,
      "step": 192364
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.3524876832962036,
      "learning_rate": 3.945414115077461e-05,
      "loss": 2.0272,
      "step": 192365
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.4090980291366577,
      "learning_rate": 3.945211342871025e-05,
      "loss": 2.1951,
      "step": 192366
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.270257830619812,
      "learning_rate": 3.945008575508671e-05,
      "loss": 1.9079,
      "step": 192367
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.4451897144317627,
      "learning_rate": 3.944805812990422e-05,
      "loss": 2.1832,
      "step": 192368
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.4873040914535522,
      "learning_rate": 3.944603055316332e-05,
      "loss": 2.1034,
      "step": 192369
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.5328185558319092,
      "learning_rate": 3.944400302486427e-05,
      "loss": 1.9864,
      "step": 192370
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.4829778671264648,
      "learning_rate": 3.9441975545007566e-05,
      "loss": 1.9508,
      "step": 192371
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.4227482080459595,
      "learning_rate": 3.943994811359345e-05,
      "loss": 1.9825,
      "step": 192372
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.4184253215789795,
      "learning_rate": 3.943792073062248e-05,
      "loss": 1.8331,
      "step": 192373
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.491024374961853,
      "learning_rate": 3.943589339609482e-05,
      "loss": 2.0097,
      "step": 192374
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.6660292148590088,
      "learning_rate": 3.943386611001098e-05,
      "loss": 1.9438,
      "step": 192375
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.418234944343567,
      "learning_rate": 3.9431838872371255e-05,
      "loss": 2.0694,
      "step": 192376
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.4916080236434937,
      "learning_rate": 3.9429811683176114e-05,
      "loss": 1.9657,
      "step": 192377
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.4415885210037231,
      "learning_rate": 3.942778454242586e-05,
      "loss": 1.8927,
      "step": 192378
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.2891219854354858,
      "learning_rate": 3.942575745012098e-05,
      "loss": 1.9513,
      "step": 192379
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.544256329536438,
      "learning_rate": 3.9423730406261676e-05,
      "loss": 1.9317,
      "step": 192380
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.6402150392532349,
      "learning_rate": 3.942170341084849e-05,
      "loss": 2.1336,
      "step": 192381
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.5608702898025513,
      "learning_rate": 3.941967646388166e-05,
      "loss": 1.9723,
      "step": 192382
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.3646337985992432,
      "learning_rate": 3.941764956536166e-05,
      "loss": 2.0098,
      "step": 192383
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.5903583765029907,
      "learning_rate": 3.941562271528882e-05,
      "loss": 2.1723,
      "step": 192384
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.4187052249908447,
      "learning_rate": 3.941359591366362e-05,
      "loss": 2.1861,
      "step": 192385
    },
    {
      "epoch": 2.5,
      "grad_norm": 1.4188120365142822,
      "learning_rate": 3.9411569160486264e-05,
      "loss": 2.0405,
      "step": 192386
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.485190987586975,
      "learning_rate": 3.940954245575726e-05,
      "loss": 2.215,
      "step": 192387
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.4220327138900757,
      "learning_rate": 3.9407515799476865e-05,
      "loss": 2.0484,
      "step": 192388
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.3457486629486084,
      "learning_rate": 3.940548919164561e-05,
      "loss": 1.9205,
      "step": 192389
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.3467360734939575,
      "learning_rate": 3.940346263226374e-05,
      "loss": 2.1241,
      "step": 192390
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.3456486463546753,
      "learning_rate": 3.940143612133174e-05,
      "loss": 2.0591,
      "step": 192391
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.5709470510482788,
      "learning_rate": 3.939940965884992e-05,
      "loss": 2.056,
      "step": 192392
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.5947877168655396,
      "learning_rate": 3.939738324481868e-05,
      "loss": 2.0529,
      "step": 192393
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.2733774185180664,
      "learning_rate": 3.939535687923835e-05,
      "loss": 1.9115,
      "step": 192394
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.3629465103149414,
      "learning_rate": 3.939333056210936e-05,
      "loss": 1.9457,
      "step": 192395
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.459113597869873,
      "learning_rate": 3.9391304293432014e-05,
      "loss": 1.7677,
      "step": 192396
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.5100014209747314,
      "learning_rate": 3.938927807320684e-05,
      "loss": 2.1369,
      "step": 192397
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.5516152381896973,
      "learning_rate": 3.938725190143408e-05,
      "loss": 1.8621,
      "step": 192398
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.3632864952087402,
      "learning_rate": 3.9385225778114124e-05,
      "loss": 2.1146,
      "step": 192399
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.3902289867401123,
      "learning_rate": 3.9383199703247414e-05,
      "loss": 1.8296,
      "step": 192400
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.4604510068893433,
      "learning_rate": 3.9381173676834275e-05,
      "loss": 2.1763,
      "step": 192401
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.513529658317566,
      "learning_rate": 3.9379147698875046e-05,
      "loss": 2.0069,
      "step": 192402
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.6584786176681519,
      "learning_rate": 3.937712176937022e-05,
      "loss": 2.1409,
      "step": 192403
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.4384630918502808,
      "learning_rate": 3.9375095888320106e-05,
      "loss": 2.0405,
      "step": 192404
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.4710978269577026,
      "learning_rate": 3.9373070055725e-05,
      "loss": 2.0931,
      "step": 192405
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.2905513048171997,
      "learning_rate": 3.937104427158546e-05,
      "loss": 2.1087,
      "step": 192406
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.352624535560608,
      "learning_rate": 3.936901853590173e-05,
      "loss": 2.1642,
      "step": 192407
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.3969783782958984,
      "learning_rate": 3.9366992848674174e-05,
      "loss": 1.8013,
      "step": 192408
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.5531554222106934,
      "learning_rate": 3.936496720990325e-05,
      "loss": 1.7823,
      "step": 192409
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.5389597415924072,
      "learning_rate": 3.936294161958934e-05,
      "loss": 2.0233,
      "step": 192410
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.5840848684310913,
      "learning_rate": 3.9360916077732696e-05,
      "loss": 2.0316,
      "step": 192411
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.53130304813385,
      "learning_rate": 3.935889058433382e-05,
      "loss": 1.9393,
      "step": 192412
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.4047579765319824,
      "learning_rate": 3.935686513939299e-05,
      "loss": 1.8805,
      "step": 192413
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.949339747428894,
      "learning_rate": 3.935483974291073e-05,
      "loss": 1.9145,
      "step": 192414
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.9357672929763794,
      "learning_rate": 3.93528143948873e-05,
      "loss": 1.9895,
      "step": 192415
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.7523071765899658,
      "learning_rate": 3.935078909532311e-05,
      "loss": 1.9847,
      "step": 192416
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.630696177482605,
      "learning_rate": 3.934876384421849e-05,
      "loss": 1.9882,
      "step": 192417
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.5045459270477295,
      "learning_rate": 3.934673864157387e-05,
      "loss": 1.9103,
      "step": 192418
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.4199563264846802,
      "learning_rate": 3.934471348738959e-05,
      "loss": 2.0363,
      "step": 192419
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.635137915611267,
      "learning_rate": 3.934268838166608e-05,
      "loss": 2.1516,
      "step": 192420
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.443904995918274,
      "learning_rate": 3.93406633244037e-05,
      "loss": 1.9584,
      "step": 192421
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.5331751108169556,
      "learning_rate": 3.933863831560282e-05,
      "loss": 2.0708,
      "step": 192422
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.3985365629196167,
      "learning_rate": 3.933661335526371e-05,
      "loss": 2.0463,
      "step": 192423
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.5687532424926758,
      "learning_rate": 3.933458844338694e-05,
      "loss": 1.9847,
      "step": 192424
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.351148009300232,
      "learning_rate": 3.933256357997273e-05,
      "loss": 1.8372,
      "step": 192425
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.5842454433441162,
      "learning_rate": 3.933053876502156e-05,
      "loss": 1.9428,
      "step": 192426
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.5544718503952026,
      "learning_rate": 3.9328513998533786e-05,
      "loss": 2.0622,
      "step": 192427
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.6738039255142212,
      "learning_rate": 3.932648928050974e-05,
      "loss": 2.1329,
      "step": 192428
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.399649977684021,
      "learning_rate": 3.932446461094977e-05,
      "loss": 1.9617,
      "step": 192429
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.3425406217575073,
      "learning_rate": 3.932243998985436e-05,
      "loss": 1.9828,
      "step": 192430
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.4627617597579956,
      "learning_rate": 3.932041541722375e-05,
      "loss": 2.01,
      "step": 192431
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.3509597778320312,
      "learning_rate": 3.9318390893058473e-05,
      "loss": 2.0262,
      "step": 192432
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.2886557579040527,
      "learning_rate": 3.931636641735886e-05,
      "loss": 2.0542,
      "step": 192433
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.3919848203659058,
      "learning_rate": 3.931434199012521e-05,
      "loss": 2.1723,
      "step": 192434
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.4566413164138794,
      "learning_rate": 3.931231761135789e-05,
      "loss": 2.1371,
      "step": 192435
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.638924479484558,
      "learning_rate": 3.93102932810574e-05,
      "loss": 2.0778,
      "step": 192436
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.5198137760162354,
      "learning_rate": 3.930826899922398e-05,
      "loss": 2.2757,
      "step": 192437
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.4046332836151123,
      "learning_rate": 3.9306244765858155e-05,
      "loss": 1.8554,
      "step": 192438
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.3490203619003296,
      "learning_rate": 3.930422058096012e-05,
      "loss": 2.0206,
      "step": 192439
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.3630194664001465,
      "learning_rate": 3.930219644453052e-05,
      "loss": 1.9778,
      "step": 192440
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.4856690168380737,
      "learning_rate": 3.930017235656941e-05,
      "loss": 2.255,
      "step": 192441
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.3381320238113403,
      "learning_rate": 3.9298148317077406e-05,
      "loss": 1.8768,
      "step": 192442
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.4123897552490234,
      "learning_rate": 3.9296124326054725e-05,
      "loss": 1.8531,
      "step": 192443
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.4372318983078003,
      "learning_rate": 3.929410038350187e-05,
      "loss": 2.0464,
      "step": 192444
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.470174789428711,
      "learning_rate": 3.9292076489419076e-05,
      "loss": 2.0679,
      "step": 192445
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.48896324634552,
      "learning_rate": 3.929005264380698e-05,
      "loss": 1.9509,
      "step": 192446
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.382664680480957,
      "learning_rate": 3.9288028846665644e-05,
      "loss": 2.1625,
      "step": 192447
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.3748490810394287,
      "learning_rate": 3.928600509799563e-05,
      "loss": 2.191,
      "step": 192448
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.4805254936218262,
      "learning_rate": 3.928398139779721e-05,
      "loss": 1.9738,
      "step": 192449
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.3947011232376099,
      "learning_rate": 3.928195774607089e-05,
      "loss": 2.1875,
      "step": 192450
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.4393970966339111,
      "learning_rate": 3.927993414281689e-05,
      "loss": 1.8992,
      "step": 192451
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.3436825275421143,
      "learning_rate": 3.927791058803585e-05,
      "loss": 1.9601,
      "step": 192452
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.3625452518463135,
      "learning_rate": 3.927588708172781e-05,
      "loss": 1.9812,
      "step": 192453
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.388688087463379,
      "learning_rate": 3.9273863623893355e-05,
      "loss": 1.8445,
      "step": 192454
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.4799612760543823,
      "learning_rate": 3.927184021453276e-05,
      "loss": 2.1309,
      "step": 192455
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.3221184015274048,
      "learning_rate": 3.926981685364652e-05,
      "loss": 1.8396,
      "step": 192456
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.4500067234039307,
      "learning_rate": 3.9267793541234874e-05,
      "loss": 1.9966,
      "step": 192457
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.3881583213806152,
      "learning_rate": 3.926577027729838e-05,
      "loss": 2.0266,
      "step": 192458
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.478057622909546,
      "learning_rate": 3.926374706183722e-05,
      "loss": 2.0033,
      "step": 192459
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.485311508178711,
      "learning_rate": 3.9261723894851874e-05,
      "loss": 1.9529,
      "step": 192460
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.4322702884674072,
      "learning_rate": 3.925970077634262e-05,
      "loss": 1.9926,
      "step": 192461
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.3875212669372559,
      "learning_rate": 3.925767770630999e-05,
      "loss": 1.985,
      "step": 192462
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.3441559076309204,
      "learning_rate": 3.925565468475425e-05,
      "loss": 1.7392,
      "step": 192463
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.521712303161621,
      "learning_rate": 3.9253631711675894e-05,
      "loss": 2.0526,
      "step": 192464
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.3762333393096924,
      "learning_rate": 3.9251608787075094e-05,
      "loss": 1.9612,
      "step": 192465
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.40838623046875,
      "learning_rate": 3.924958591095239e-05,
      "loss": 1.8981,
      "step": 192466
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.350298285484314,
      "learning_rate": 3.924756308330807e-05,
      "loss": 2.1058,
      "step": 192467
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.3444104194641113,
      "learning_rate": 3.92455403041426e-05,
      "loss": 2.0889,
      "step": 192468
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.5348118543624878,
      "learning_rate": 3.924351757345625e-05,
      "loss": 2.0204,
      "step": 192469
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.4078105688095093,
      "learning_rate": 3.924149489124956e-05,
      "loss": 2.0951,
      "step": 192470
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.3162033557891846,
      "learning_rate": 3.923947225752269e-05,
      "loss": 2.0421,
      "step": 192471
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.840085506439209,
      "learning_rate": 3.92374496722762e-05,
      "loss": 2.1445,
      "step": 192472
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.4609990119934082,
      "learning_rate": 3.92354271355103e-05,
      "loss": 1.9401,
      "step": 192473
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.3769561052322388,
      "learning_rate": 3.923340464722552e-05,
      "loss": 2.1674,
      "step": 192474
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.4425158500671387,
      "learning_rate": 3.923138220742212e-05,
      "loss": 2.085,
      "step": 192475
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.379031777381897,
      "learning_rate": 3.922935981610058e-05,
      "loss": 1.9299,
      "step": 192476
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.504721999168396,
      "learning_rate": 3.922733747326125e-05,
      "loss": 2.0356,
      "step": 192477
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.4008451700210571,
      "learning_rate": 3.9225315178904485e-05,
      "loss": 1.9134,
      "step": 192478
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.4190216064453125,
      "learning_rate": 3.922329293303056e-05,
      "loss": 2.0351,
      "step": 192479
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.534045934677124,
      "learning_rate": 3.922127073564002e-05,
      "loss": 2.0581,
      "step": 192480
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.2929322719573975,
      "learning_rate": 3.92192485867331e-05,
      "loss": 1.8391,
      "step": 192481
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.3925366401672363,
      "learning_rate": 3.9217226486310325e-05,
      "loss": 1.7676,
      "step": 192482
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.4191802740097046,
      "learning_rate": 3.921520443437201e-05,
      "loss": 1.9371,
      "step": 192483
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.3662692308425903,
      "learning_rate": 3.921318243091847e-05,
      "loss": 2.0319,
      "step": 192484
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.4266884326934814,
      "learning_rate": 3.9211160475950075e-05,
      "loss": 2.0873,
      "step": 192485
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.3217847347259521,
      "learning_rate": 3.92091385694673e-05,
      "loss": 1.8273,
      "step": 192486
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.445976972579956,
      "learning_rate": 3.920711671147041e-05,
      "loss": 2.0948,
      "step": 192487
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.3564393520355225,
      "learning_rate": 3.92050949019599e-05,
      "loss": 1.7152,
      "step": 192488
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.4105141162872314,
      "learning_rate": 3.92030731409361e-05,
      "loss": 1.9653,
      "step": 192489
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.3797415494918823,
      "learning_rate": 3.920105142839932e-05,
      "loss": 1.9203,
      "step": 192490
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.3972961902618408,
      "learning_rate": 3.9199029764350024e-05,
      "loss": 1.8326,
      "step": 192491
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.311113953590393,
      "learning_rate": 3.919700814878857e-05,
      "loss": 1.968,
      "step": 192492
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.3693095445632935,
      "learning_rate": 3.9194986581715226e-05,
      "loss": 1.9962,
      "step": 192493
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.6075552701950073,
      "learning_rate": 3.919296506313053e-05,
      "loss": 2.1974,
      "step": 192494
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.355181097984314,
      "learning_rate": 3.919094359303482e-05,
      "loss": 2.0047,
      "step": 192495
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.5376904010772705,
      "learning_rate": 3.9188922171428315e-05,
      "loss": 1.8634,
      "step": 192496
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.5305085182189941,
      "learning_rate": 3.918690079831163e-05,
      "loss": 2.1936,
      "step": 192497
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.3386050462722778,
      "learning_rate": 3.9184879473684914e-05,
      "loss": 1.9031,
      "step": 192498
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.403209924697876,
      "learning_rate": 3.918285819754875e-05,
      "loss": 1.9465,
      "step": 192499
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.4117275476455688,
      "learning_rate": 3.918083696990339e-05,
      "loss": 1.9285,
      "step": 192500
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.376435399055481,
      "learning_rate": 3.9178815790749254e-05,
      "loss": 2.0702,
      "step": 192501
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.5067187547683716,
      "learning_rate": 3.9176794660086654e-05,
      "loss": 1.9634,
      "step": 192502
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.4668511152267456,
      "learning_rate": 3.917477357791603e-05,
      "loss": 2.1446,
      "step": 192503
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.3858965635299683,
      "learning_rate": 3.917275254423772e-05,
      "loss": 2.0509,
      "step": 192504
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.318000316619873,
      "learning_rate": 3.9170731559052146e-05,
      "loss": 2.0866,
      "step": 192505
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.3683297634124756,
      "learning_rate": 3.916871062235966e-05,
      "loss": 2.0115,
      "step": 192506
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.4423991441726685,
      "learning_rate": 3.916668973416064e-05,
      "loss": 2.0053,
      "step": 192507
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.3849761486053467,
      "learning_rate": 3.9164668894455395e-05,
      "loss": 2.0473,
      "step": 192508
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.5484813451766968,
      "learning_rate": 3.9162648103244434e-05,
      "loss": 2.0527,
      "step": 192509
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.3120094537734985,
      "learning_rate": 3.9160627360528007e-05,
      "loss": 1.8786,
      "step": 192510
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.52265465259552,
      "learning_rate": 3.9158606666306594e-05,
      "loss": 2.3709,
      "step": 192511
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.5673328638076782,
      "learning_rate": 3.9156586020580446e-05,
      "loss": 1.7429,
      "step": 192512
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.3549177646636963,
      "learning_rate": 3.915456542335015e-05,
      "loss": 1.9727,
      "step": 192513
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.3920153379440308,
      "learning_rate": 3.9152544874615784e-05,
      "loss": 2.0425,
      "step": 192514
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.4261059761047363,
      "learning_rate": 3.9150524374378e-05,
      "loss": 1.9634,
      "step": 192515
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.3572216033935547,
      "learning_rate": 3.9148503922636954e-05,
      "loss": 2.0811,
      "step": 192516
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.5434129238128662,
      "learning_rate": 3.914648351939321e-05,
      "loss": 1.9857,
      "step": 192517
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.4084197282791138,
      "learning_rate": 3.914446316464698e-05,
      "loss": 1.7764,
      "step": 192518
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.535538673400879,
      "learning_rate": 3.9142442858398884e-05,
      "loss": 2.0927,
      "step": 192519
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.5754039287567139,
      "learning_rate": 3.9140422600648955e-05,
      "loss": 2.0847,
      "step": 192520
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.35983145236969,
      "learning_rate": 3.913840239139784e-05,
      "loss": 2.0881,
      "step": 192521
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.5203590393066406,
      "learning_rate": 3.9136382230645755e-05,
      "loss": 2.0569,
      "step": 192522
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.2700397968292236,
      "learning_rate": 3.913436211839318e-05,
      "loss": 2.0704,
      "step": 192523
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.326178789138794,
      "learning_rate": 3.913234205464041e-05,
      "loss": 1.7504,
      "step": 192524
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.444236159324646,
      "learning_rate": 3.9130322039387975e-05,
      "loss": 2.0513,
      "step": 192525
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.406606912612915,
      "learning_rate": 3.912830207263601e-05,
      "loss": 2.0817,
      "step": 192526
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.3843188285827637,
      "learning_rate": 3.912628215438511e-05,
      "loss": 2.0414,
      "step": 192527
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.3573583364486694,
      "learning_rate": 3.912426228463546e-05,
      "loss": 1.9728,
      "step": 192528
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.4375851154327393,
      "learning_rate": 3.9122242463387634e-05,
      "loss": 1.8533,
      "step": 192529
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.4446535110473633,
      "learning_rate": 3.912022269064181e-05,
      "loss": 2.163,
      "step": 192530
    },
    {
      "epoch": 2.51,
      "grad_norm": 2.238948106765747,
      "learning_rate": 3.911820296639859e-05,
      "loss": 2.1914,
      "step": 192531
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.5035876035690308,
      "learning_rate": 3.911618329065811e-05,
      "loss": 2.0814,
      "step": 192532
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.397768259048462,
      "learning_rate": 3.911416366342093e-05,
      "loss": 2.021,
      "step": 192533
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.2410131692886353,
      "learning_rate": 3.911214408468725e-05,
      "loss": 2.2194,
      "step": 192534
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.3560030460357666,
      "learning_rate": 3.911012455445763e-05,
      "loss": 2.0735,
      "step": 192535
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.4415146112442017,
      "learning_rate": 3.910810507273232e-05,
      "loss": 2.0448,
      "step": 192536
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.5076789855957031,
      "learning_rate": 3.910608563951184e-05,
      "loss": 2.0975,
      "step": 192537
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.3660374879837036,
      "learning_rate": 3.910406625479633e-05,
      "loss": 1.8433,
      "step": 192538
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.3636940717697144,
      "learning_rate": 3.9102046918586386e-05,
      "loss": 1.7427,
      "step": 192539
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.3608235120773315,
      "learning_rate": 3.910002763088224e-05,
      "loss": 2.0169,
      "step": 192540
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.3436700105667114,
      "learning_rate": 3.909800839168435e-05,
      "loss": 1.9462,
      "step": 192541
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.4911764860153198,
      "learning_rate": 3.909598920099301e-05,
      "loss": 2.2009,
      "step": 192542
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.306110143661499,
      "learning_rate": 3.9093970058808786e-05,
      "loss": 1.7938,
      "step": 192543
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.3348991870880127,
      "learning_rate": 3.9091950965131804e-05,
      "loss": 1.9356,
      "step": 192544
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.3652732372283936,
      "learning_rate": 3.908993191996261e-05,
      "loss": 2.1099,
      "step": 192545
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.4423028230667114,
      "learning_rate": 3.9087912923301456e-05,
      "loss": 2.1647,
      "step": 192546
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.3614851236343384,
      "learning_rate": 3.908589397514883e-05,
      "loss": 2.057,
      "step": 192547
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.7769289016723633,
      "learning_rate": 3.908387507550503e-05,
      "loss": 1.7955,
      "step": 192548
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.5908410549163818,
      "learning_rate": 3.9081856224370567e-05,
      "loss": 1.8378,
      "step": 192549
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.2908169031143188,
      "learning_rate": 3.90798374217456e-05,
      "loss": 2.2204,
      "step": 192550
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.3992087841033936,
      "learning_rate": 3.9077818667630654e-05,
      "loss": 1.6867,
      "step": 192551
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.4513155221939087,
      "learning_rate": 3.9075799962026045e-05,
      "loss": 1.7792,
      "step": 192552
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.5368002653121948,
      "learning_rate": 3.907378130493219e-05,
      "loss": 2.127,
      "step": 192553
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.5185272693634033,
      "learning_rate": 3.907176269634941e-05,
      "loss": 1.8421,
      "step": 192554
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.4770480394363403,
      "learning_rate": 3.9069744136278247e-05,
      "loss": 1.9641,
      "step": 192555
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.3656697273254395,
      "learning_rate": 3.906772562471878e-05,
      "loss": 2.1624,
      "step": 192556
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.5281339883804321,
      "learning_rate": 3.906570716167165e-05,
      "loss": 2.0408,
      "step": 192557
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.4204695224761963,
      "learning_rate": 3.9063688747137036e-05,
      "loss": 1.9762,
      "step": 192558
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.3393950462341309,
      "learning_rate": 3.906167038111552e-05,
      "loss": 2.0779,
      "step": 192559
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.320340871810913,
      "learning_rate": 3.905965206360726e-05,
      "loss": 1.9804,
      "step": 192560
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.3960769176483154,
      "learning_rate": 3.9057633794612796e-05,
      "loss": 1.8514,
      "step": 192561
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.3366608619689941,
      "learning_rate": 3.905561557413246e-05,
      "loss": 1.7975,
      "step": 192562
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.4247387647628784,
      "learning_rate": 3.9053597402166615e-05,
      "loss": 1.9239,
      "step": 192563
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.493249773979187,
      "learning_rate": 3.905157927871556e-05,
      "loss": 2.1763,
      "step": 192564
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.3431990146636963,
      "learning_rate": 3.90495612037798e-05,
      "loss": 2.0088,
      "step": 192565
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.4198905229568481,
      "learning_rate": 3.90475431773596e-05,
      "loss": 1.9214,
      "step": 192566
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.4224869012832642,
      "learning_rate": 3.9045525199455455e-05,
      "loss": 2.2821,
      "step": 192567
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.7559112310409546,
      "learning_rate": 3.9043507270067665e-05,
      "loss": 2.1314,
      "step": 192568
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.5063267946243286,
      "learning_rate": 3.90414893891966e-05,
      "loss": 2.0564,
      "step": 192569
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.3514233827590942,
      "learning_rate": 3.9039471556842595e-05,
      "loss": 1.9839,
      "step": 192570
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.3359854221343994,
      "learning_rate": 3.903745377300614e-05,
      "loss": 2.0643,
      "step": 192571
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.3686083555221558,
      "learning_rate": 3.9035436037687485e-05,
      "loss": 2.1376,
      "step": 192572
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.4422279596328735,
      "learning_rate": 3.9033418350887146e-05,
      "loss": 2.046,
      "step": 192573
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.3752672672271729,
      "learning_rate": 3.90314007126054e-05,
      "loss": 1.7815,
      "step": 192574
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.350968360900879,
      "learning_rate": 3.90293831228426e-05,
      "loss": 1.868,
      "step": 192575
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.3365525007247925,
      "learning_rate": 3.9027365581599204e-05,
      "loss": 1.9822,
      "step": 192576
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.3119922876358032,
      "learning_rate": 3.9025348088875554e-05,
      "loss": 1.943,
      "step": 192577
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.3825424909591675,
      "learning_rate": 3.9023330644671957e-05,
      "loss": 2.0281,
      "step": 192578
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.4227676391601562,
      "learning_rate": 3.9021313248988914e-05,
      "loss": 1.8502,
      "step": 192579
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.5243463516235352,
      "learning_rate": 3.9019295901826764e-05,
      "loss": 2.2691,
      "step": 192580
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.414810299873352,
      "learning_rate": 3.901727860318573e-05,
      "loss": 1.9704,
      "step": 192581
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.4319021701812744,
      "learning_rate": 3.901526135306642e-05,
      "loss": 2.0088,
      "step": 192582
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.363539218902588,
      "learning_rate": 3.901324415146909e-05,
      "loss": 1.9743,
      "step": 192583
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.3576570749282837,
      "learning_rate": 3.901122699839405e-05,
      "loss": 2.0441,
      "step": 192584
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.2469230890274048,
      "learning_rate": 3.900920989384183e-05,
      "loss": 1.6426,
      "step": 192585
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.3978544473648071,
      "learning_rate": 3.9007192837812695e-05,
      "loss": 2.0804,
      "step": 192586
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.3535829782485962,
      "learning_rate": 3.900517583030701e-05,
      "loss": 2.0409,
      "step": 192587
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.4066654443740845,
      "learning_rate": 3.900315887132525e-05,
      "loss": 2.0172,
      "step": 192588
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.383176565170288,
      "learning_rate": 3.9001141960867664e-05,
      "loss": 2.07,
      "step": 192589
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.3435852527618408,
      "learning_rate": 3.899912509893477e-05,
      "loss": 1.9221,
      "step": 192590
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.3122215270996094,
      "learning_rate": 3.899710828552686e-05,
      "loss": 1.9975,
      "step": 192591
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.3912959098815918,
      "learning_rate": 3.8995091520644296e-05,
      "loss": 1.7879,
      "step": 192592
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.4568511247634888,
      "learning_rate": 3.8993074804287416e-05,
      "loss": 1.8563,
      "step": 192593
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.3634846210479736,
      "learning_rate": 3.899105813645672e-05,
      "loss": 2.2007,
      "step": 192594
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.290144681930542,
      "learning_rate": 3.898904151715244e-05,
      "loss": 2.01,
      "step": 192595
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.3696722984313965,
      "learning_rate": 3.8987024946375076e-05,
      "loss": 1.9462,
      "step": 192596
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.3995243310928345,
      "learning_rate": 3.898500842412493e-05,
      "loss": 2.0987,
      "step": 192597
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.3740026950836182,
      "learning_rate": 3.8982991950402495e-05,
      "loss": 2.0843,
      "step": 192598
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.4711569547653198,
      "learning_rate": 3.898097552520791e-05,
      "loss": 2.1181,
      "step": 192599
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.3979763984680176,
      "learning_rate": 3.897895914854178e-05,
      "loss": 1.9509,
      "step": 192600
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.338055968284607,
      "learning_rate": 3.897694282040432e-05,
      "loss": 2.1521,
      "step": 192601
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.4170101881027222,
      "learning_rate": 3.8974926540796015e-05,
      "loss": 2.1105,
      "step": 192602
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.3217072486877441,
      "learning_rate": 3.897291030971715e-05,
      "loss": 1.9228,
      "step": 192603
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.593997836112976,
      "learning_rate": 3.8970894127168275e-05,
      "loss": 2.1464,
      "step": 192604
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.367474913597107,
      "learning_rate": 3.896887799314951e-05,
      "loss": 2.0331,
      "step": 192605
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.3877086639404297,
      "learning_rate": 3.896686190766142e-05,
      "loss": 2.0275,
      "step": 192606
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.3793436288833618,
      "learning_rate": 3.896484587070427e-05,
      "loss": 2.0811,
      "step": 192607
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.3878387212753296,
      "learning_rate": 3.896282988227855e-05,
      "loss": 1.9312,
      "step": 192608
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.4004521369934082,
      "learning_rate": 3.896081394238446e-05,
      "loss": 1.8848,
      "step": 192609
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.4002293348312378,
      "learning_rate": 3.8958798051022654e-05,
      "loss": 1.9899,
      "step": 192610
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.4083712100982666,
      "learning_rate": 3.8956782208193194e-05,
      "loss": 2.0827,
      "step": 192611
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.342844843864441,
      "learning_rate": 3.895476641389664e-05,
      "loss": 2.05,
      "step": 192612
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.427422046661377,
      "learning_rate": 3.895275066813327e-05,
      "loss": 1.8728,
      "step": 192613
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.3956212997436523,
      "learning_rate": 3.895073497090357e-05,
      "loss": 1.9112,
      "step": 192614
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.513059377670288,
      "learning_rate": 3.8948719322207824e-05,
      "loss": 1.9506,
      "step": 192615
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.3631504774093628,
      "learning_rate": 3.894670372204655e-05,
      "loss": 2.0154,
      "step": 192616
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.4356880187988281,
      "learning_rate": 3.894468817041988e-05,
      "loss": 2.1141,
      "step": 192617
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.439253807067871,
      "learning_rate": 3.894267266732839e-05,
      "loss": 1.8907,
      "step": 192618
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.462370753288269,
      "learning_rate": 3.8940657212772305e-05,
      "loss": 2.065,
      "step": 192619
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.459169864654541,
      "learning_rate": 3.893864180675214e-05,
      "loss": 1.8493,
      "step": 192620
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.5391045808792114,
      "learning_rate": 3.893662644926817e-05,
      "loss": 1.9731,
      "step": 192621
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.6482266187667847,
      "learning_rate": 3.8934611140320916e-05,
      "loss": 2.0522,
      "step": 192622
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.5715444087982178,
      "learning_rate": 3.893259587991053e-05,
      "loss": 2.1131,
      "step": 192623
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.491834282875061,
      "learning_rate": 3.893058066803756e-05,
      "loss": 2.0145,
      "step": 192624
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.3031882047653198,
      "learning_rate": 3.892856550470226e-05,
      "loss": 1.8206,
      "step": 192625
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.5761796236038208,
      "learning_rate": 3.892655038990513e-05,
      "loss": 1.8933,
      "step": 192626
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.6024657487869263,
      "learning_rate": 3.892453532364644e-05,
      "loss": 2.01,
      "step": 192627
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.449500322341919,
      "learning_rate": 3.89225203059267e-05,
      "loss": 1.7135,
      "step": 192628
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.467727780342102,
      "learning_rate": 3.892050533674609e-05,
      "loss": 2.1273,
      "step": 192629
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.4760990142822266,
      "learning_rate": 3.8918490416105155e-05,
      "loss": 2.0639,
      "step": 192630
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.4271535873413086,
      "learning_rate": 3.891647554400412e-05,
      "loss": 1.7647,
      "step": 192631
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.4909549951553345,
      "learning_rate": 3.89144607204435e-05,
      "loss": 1.9207,
      "step": 192632
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.4584277868270874,
      "learning_rate": 3.891244594542354e-05,
      "loss": 1.8248,
      "step": 192633
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.5902388095855713,
      "learning_rate": 3.891043121894483e-05,
      "loss": 2.1748,
      "step": 192634
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.619787335395813,
      "learning_rate": 3.8908416541007484e-05,
      "loss": 2.1457,
      "step": 192635
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.465694785118103,
      "learning_rate": 3.890640191161204e-05,
      "loss": 1.7934,
      "step": 192636
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.5412805080413818,
      "learning_rate": 3.890438733075873e-05,
      "loss": 1.9693,
      "step": 192637
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.4867910146713257,
      "learning_rate": 3.890237279844813e-05,
      "loss": 1.9763,
      "step": 192638
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.3836065530776978,
      "learning_rate": 3.8900358314680434e-05,
      "loss": 1.8879,
      "step": 192639
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.5913684368133545,
      "learning_rate": 3.889834387945614e-05,
      "loss": 1.9339,
      "step": 192640
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.482318639755249,
      "learning_rate": 3.8896329492775615e-05,
      "loss": 1.9665,
      "step": 192641
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.7166125774383545,
      "learning_rate": 3.889431515463916e-05,
      "loss": 2.0639,
      "step": 192642
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.3354465961456299,
      "learning_rate": 3.88923008650471e-05,
      "loss": 2.1983,
      "step": 192643
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.3808447122573853,
      "learning_rate": 3.8890286623999975e-05,
      "loss": 2.0187,
      "step": 192644
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.3158093690872192,
      "learning_rate": 3.888827243149799e-05,
      "loss": 2.0703,
      "step": 192645
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.3068561553955078,
      "learning_rate": 3.8886258287541705e-05,
      "loss": 2.1613,
      "step": 192646
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.3714410066604614,
      "learning_rate": 3.8884244192131355e-05,
      "loss": 2.0768,
      "step": 192647
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.484620213508606,
      "learning_rate": 3.888223014526737e-05,
      "loss": 1.8496,
      "step": 192648
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.4283504486083984,
      "learning_rate": 3.888021614695005e-05,
      "loss": 2.0441,
      "step": 192649
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.3303064107894897,
      "learning_rate": 3.887820219717987e-05,
      "loss": 1.9651,
      "step": 192650
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.505610704421997,
      "learning_rate": 3.887618829595711e-05,
      "loss": 2.1438,
      "step": 192651
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.4128563404083252,
      "learning_rate": 3.8874174443282255e-05,
      "loss": 1.8225,
      "step": 192652
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.3061186075210571,
      "learning_rate": 3.887216063915563e-05,
      "loss": 1.822,
      "step": 192653
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.6395772695541382,
      "learning_rate": 3.887014688357761e-05,
      "loss": 2.0,
      "step": 192654
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.4701755046844482,
      "learning_rate": 3.8868133176548485e-05,
      "loss": 2.0686,
      "step": 192655
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.370714545249939,
      "learning_rate": 3.886611951806876e-05,
      "loss": 1.7944,
      "step": 192656
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.4488738775253296,
      "learning_rate": 3.88641059081387e-05,
      "loss": 2.1402,
      "step": 192657
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.3692162036895752,
      "learning_rate": 3.886209234675877e-05,
      "loss": 2.1008,
      "step": 192658
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.510697841644287,
      "learning_rate": 3.886007883392933e-05,
      "loss": 1.7514,
      "step": 192659
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.3893721103668213,
      "learning_rate": 3.885806536965066e-05,
      "loss": 2.0624,
      "step": 192660
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.3706706762313843,
      "learning_rate": 3.885605195392328e-05,
      "loss": 1.9716,
      "step": 192661
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.445777416229248,
      "learning_rate": 3.885403858674747e-05,
      "loss": 2.0926,
      "step": 192662
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.4132214784622192,
      "learning_rate": 3.885202526812355e-05,
      "loss": 1.833,
      "step": 192663
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.4129936695098877,
      "learning_rate": 3.885001199805206e-05,
      "loss": 1.9051,
      "step": 192664
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.4650607109069824,
      "learning_rate": 3.8847998776533276e-05,
      "loss": 2.0391,
      "step": 192665
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.3777110576629639,
      "learning_rate": 3.884598560356751e-05,
      "loss": 1.823,
      "step": 192666
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.4029793739318848,
      "learning_rate": 3.884397247915527e-05,
      "loss": 2.1214,
      "step": 192667
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.3422772884368896,
      "learning_rate": 3.884195940329686e-05,
      "loss": 2.003,
      "step": 192668
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.509655237197876,
      "learning_rate": 3.883994637599261e-05,
      "loss": 2.0744,
      "step": 192669
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.372381329536438,
      "learning_rate": 3.8837933397242995e-05,
      "loss": 1.8284,
      "step": 192670
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.457727074623108,
      "learning_rate": 3.883592046704836e-05,
      "loss": 2.0368,
      "step": 192671
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.3910105228424072,
      "learning_rate": 3.8833907585408966e-05,
      "loss": 2.0299,
      "step": 192672
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.4482189416885376,
      "learning_rate": 3.8831894752325353e-05,
      "loss": 2.1968,
      "step": 192673
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.5325175523757935,
      "learning_rate": 3.882988196779777e-05,
      "loss": 2.0988,
      "step": 192674
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.430897831916809,
      "learning_rate": 3.882786923182668e-05,
      "loss": 2.1284,
      "step": 192675
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.4356334209442139,
      "learning_rate": 3.882585654441241e-05,
      "loss": 1.9052,
      "step": 192676
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.3921324014663696,
      "learning_rate": 3.882384390555537e-05,
      "loss": 1.9518,
      "step": 192677
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.3905359506607056,
      "learning_rate": 3.882183131525586e-05,
      "loss": 1.9565,
      "step": 192678
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.315142273902893,
      "learning_rate": 3.8819818773514336e-05,
      "loss": 1.9083,
      "step": 192679
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.3932809829711914,
      "learning_rate": 3.8817806280331066e-05,
      "loss": 2.195,
      "step": 192680
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.3066680431365967,
      "learning_rate": 3.881579383570659e-05,
      "loss": 1.9456,
      "step": 192681
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.2672619819641113,
      "learning_rate": 3.88137814396411e-05,
      "loss": 1.7696,
      "step": 192682
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.3290619850158691,
      "learning_rate": 3.8811769092135204e-05,
      "loss": 2.066,
      "step": 192683
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.457597017288208,
      "learning_rate": 3.8809756793188996e-05,
      "loss": 2.1276,
      "step": 192684
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.4177017211914062,
      "learning_rate": 3.8807744542803044e-05,
      "loss": 1.8108,
      "step": 192685
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.2994024753570557,
      "learning_rate": 3.8805732340977613e-05,
      "loss": 1.9229,
      "step": 192686
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.4712507724761963,
      "learning_rate": 3.880372018771317e-05,
      "loss": 2.0599,
      "step": 192687
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.4797626733779907,
      "learning_rate": 3.880170808300999e-05,
      "loss": 1.9732,
      "step": 192688
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.4349478483200073,
      "learning_rate": 3.879969602686862e-05,
      "loss": 1.8229,
      "step": 192689
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.3410171270370483,
      "learning_rate": 3.879768401928921e-05,
      "loss": 2.0837,
      "step": 192690
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.4326826333999634,
      "learning_rate": 3.8795672060272285e-05,
      "loss": 1.9825,
      "step": 192691
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.3599640130996704,
      "learning_rate": 3.879366014981815e-05,
      "loss": 1.8836,
      "step": 192692
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.4341418743133545,
      "learning_rate": 3.8791648287927235e-05,
      "loss": 2.0074,
      "step": 192693
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.5199209451675415,
      "learning_rate": 3.878963647459981e-05,
      "loss": 1.8547,
      "step": 192694
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.2944934368133545,
      "learning_rate": 3.878762470983646e-05,
      "loss": 1.9505,
      "step": 192695
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.401746153831482,
      "learning_rate": 3.8785612993637304e-05,
      "loss": 1.8469,
      "step": 192696
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.350318193435669,
      "learning_rate": 3.87836013260029e-05,
      "loss": 2.143,
      "step": 192697
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.6357368230819702,
      "learning_rate": 3.878158970693348e-05,
      "loss": 2.0327,
      "step": 192698
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.3190702199935913,
      "learning_rate": 3.877957813642955e-05,
      "loss": 1.9621,
      "step": 192699
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.632128119468689,
      "learning_rate": 3.8777566614491395e-05,
      "loss": 2.013,
      "step": 192700
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.6170454025268555,
      "learning_rate": 3.877555514111953e-05,
      "loss": 2.1969,
      "step": 192701
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.400192379951477,
      "learning_rate": 3.877354371631408e-05,
      "loss": 1.9412,
      "step": 192702
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.3233627080917358,
      "learning_rate": 3.877153234007562e-05,
      "loss": 1.9519,
      "step": 192703
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.4460198879241943,
      "learning_rate": 3.876952101240444e-05,
      "loss": 2.0614,
      "step": 192704
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.4077790975570679,
      "learning_rate": 3.8767509733300975e-05,
      "loss": 2.0132,
      "step": 192705
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.3540037870407104,
      "learning_rate": 3.8765498502765494e-05,
      "loss": 2.0421,
      "step": 192706
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.3872113227844238,
      "learning_rate": 3.876348732079849e-05,
      "loss": 1.7958,
      "step": 192707
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.4512888193130493,
      "learning_rate": 3.8761476187400306e-05,
      "loss": 1.8447,
      "step": 192708
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.2886590957641602,
      "learning_rate": 3.875946510257128e-05,
      "loss": 2.0603,
      "step": 192709
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.4821090698242188,
      "learning_rate": 3.8757454066311753e-05,
      "loss": 2.0805,
      "step": 192710
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.4508172273635864,
      "learning_rate": 3.8755443078622184e-05,
      "loss": 1.8632,
      "step": 192711
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.4789273738861084,
      "learning_rate": 3.875343213950286e-05,
      "loss": 1.9703,
      "step": 192712
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.4629098176956177,
      "learning_rate": 3.8751421248954286e-05,
      "loss": 2.026,
      "step": 192713
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.5534716844558716,
      "learning_rate": 3.874941040697675e-05,
      "loss": 1.9831,
      "step": 192714
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.378857135772705,
      "learning_rate": 3.874739961357061e-05,
      "loss": 2.132,
      "step": 192715
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.3309985399246216,
      "learning_rate": 3.8745388868736205e-05,
      "loss": 1.8152,
      "step": 192716
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.3691810369491577,
      "learning_rate": 3.874337817247401e-05,
      "loss": 2.136,
      "step": 192717
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.4331499338150024,
      "learning_rate": 3.874136752478434e-05,
      "loss": 2.1653,
      "step": 192718
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.3321789503097534,
      "learning_rate": 3.873935692566761e-05,
      "loss": 1.8119,
      "step": 192719
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.4042848348617554,
      "learning_rate": 3.873734637512416e-05,
      "loss": 1.8505,
      "step": 192720
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.5447659492492676,
      "learning_rate": 3.8735335873154384e-05,
      "loss": 1.9361,
      "step": 192721
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.4619566202163696,
      "learning_rate": 3.873332541975855e-05,
      "loss": 1.9309,
      "step": 192722
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.275403618812561,
      "learning_rate": 3.873131501493724e-05,
      "loss": 1.9706,
      "step": 192723
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.3365204334259033,
      "learning_rate": 3.87293046586906e-05,
      "loss": 1.844,
      "step": 192724
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.448171854019165,
      "learning_rate": 3.87272943510192e-05,
      "loss": 1.9948,
      "step": 192725
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.2998502254486084,
      "learning_rate": 3.872528409192331e-05,
      "loss": 2.0256,
      "step": 192726
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.2846338748931885,
      "learning_rate": 3.872327388140334e-05,
      "loss": 1.7669,
      "step": 192727
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.227905035018921,
      "learning_rate": 3.872126371945957e-05,
      "loss": 1.7702,
      "step": 192728
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.3764400482177734,
      "learning_rate": 3.871925360609251e-05,
      "loss": 2.1882,
      "step": 192729
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.3578486442565918,
      "learning_rate": 3.87172435413024e-05,
      "loss": 2.1133,
      "step": 192730
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.3191334009170532,
      "learning_rate": 3.871523352508976e-05,
      "loss": 2.2395,
      "step": 192731
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.4259204864501953,
      "learning_rate": 3.87132235574549e-05,
      "loss": 2.0441,
      "step": 192732
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.3071521520614624,
      "learning_rate": 3.8711213638398173e-05,
      "loss": 2.0456,
      "step": 192733
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.4255167245864868,
      "learning_rate": 3.870920376791993e-05,
      "loss": 2.1007,
      "step": 192734
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.378597617149353,
      "learning_rate": 3.870719394602062e-05,
      "loss": 2.1344,
      "step": 192735
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.2666822671890259,
      "learning_rate": 3.870518417270049e-05,
      "loss": 1.8989,
      "step": 192736
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.41019606590271,
      "learning_rate": 3.87031744479601e-05,
      "loss": 1.9693,
      "step": 192737
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.5056226253509521,
      "learning_rate": 3.8701164771799674e-05,
      "loss": 2.0024,
      "step": 192738
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.598561406135559,
      "learning_rate": 3.8699155144219666e-05,
      "loss": 2.0644,
      "step": 192739
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.513013482093811,
      "learning_rate": 3.869714556522036e-05,
      "loss": 2.0922,
      "step": 192740
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.3821542263031006,
      "learning_rate": 3.869513603480225e-05,
      "loss": 1.7731,
      "step": 192741
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.3202227354049683,
      "learning_rate": 3.869312655296556e-05,
      "loss": 1.8533,
      "step": 192742
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.3910105228424072,
      "learning_rate": 3.8691117119710834e-05,
      "loss": 2.1511,
      "step": 192743
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.3087143898010254,
      "learning_rate": 3.868910773503835e-05,
      "loss": 1.8119,
      "step": 192744
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.3650524616241455,
      "learning_rate": 3.86870983989485e-05,
      "loss": 1.9159,
      "step": 192745
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.314843773841858,
      "learning_rate": 3.8685089111441594e-05,
      "loss": 1.8584,
      "step": 192746
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.440568208694458,
      "learning_rate": 3.868307987251812e-05,
      "loss": 1.6818,
      "step": 192747
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.3378666639328003,
      "learning_rate": 3.8681070682178316e-05,
      "loss": 1.8721,
      "step": 192748
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.2883663177490234,
      "learning_rate": 3.8679061540422715e-05,
      "loss": 1.8401,
      "step": 192749
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.4138964414596558,
      "learning_rate": 3.867705244725162e-05,
      "loss": 2.0025,
      "step": 192750
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.3371390104293823,
      "learning_rate": 3.867504340266533e-05,
      "loss": 2.047,
      "step": 192751
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.482541799545288,
      "learning_rate": 3.8673034406664306e-05,
      "loss": 2.0605,
      "step": 192752
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.5771446228027344,
      "learning_rate": 3.867102545924895e-05,
      "loss": 2.0397,
      "step": 192753
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.4695162773132324,
      "learning_rate": 3.8669016560419464e-05,
      "loss": 1.9016,
      "step": 192754
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.4688040018081665,
      "learning_rate": 3.8667007710176445e-05,
      "loss": 1.8434,
      "step": 192755
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.4305424690246582,
      "learning_rate": 3.8664998908520164e-05,
      "loss": 1.8536,
      "step": 192756
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.4557594060897827,
      "learning_rate": 3.8662990155450914e-05,
      "loss": 2.0249,
      "step": 192757
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.4689489603042603,
      "learning_rate": 3.86609814509692e-05,
      "loss": 1.8639,
      "step": 192758
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.4144871234893799,
      "learning_rate": 3.865897279507528e-05,
      "loss": 2.212,
      "step": 192759
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.3541207313537598,
      "learning_rate": 3.86569641877697e-05,
      "loss": 2.0225,
      "step": 192760
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.3684589862823486,
      "learning_rate": 3.865495562905269e-05,
      "loss": 2.04,
      "step": 192761
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.4784592390060425,
      "learning_rate": 3.865294711892464e-05,
      "loss": 1.8857,
      "step": 192762
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.4729844331741333,
      "learning_rate": 3.8650938657385886e-05,
      "loss": 1.995,
      "step": 192763
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.344520092010498,
      "learning_rate": 3.8648930244436935e-05,
      "loss": 2.0366,
      "step": 192764
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.3765532970428467,
      "learning_rate": 3.864692188007802e-05,
      "loss": 1.9626,
      "step": 192765
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.3636757135391235,
      "learning_rate": 3.8644913564309635e-05,
      "loss": 1.7133,
      "step": 192766
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.392107367515564,
      "learning_rate": 3.864290529713208e-05,
      "loss": 1.9732,
      "step": 192767
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.4661164283752441,
      "learning_rate": 3.8640897078545755e-05,
      "loss": 1.87,
      "step": 192768
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.4103939533233643,
      "learning_rate": 3.863888890855096e-05,
      "loss": 2.0403,
      "step": 192769
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.2987067699432373,
      "learning_rate": 3.863688078714816e-05,
      "loss": 1.8503,
      "step": 192770
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.3818398714065552,
      "learning_rate": 3.863487271433766e-05,
      "loss": 1.6729,
      "step": 192771
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.4145653247833252,
      "learning_rate": 3.863286469011993e-05,
      "loss": 2.1909,
      "step": 192772
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.4087716341018677,
      "learning_rate": 3.863085671449525e-05,
      "loss": 2.0477,
      "step": 192773
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.2029975652694702,
      "learning_rate": 3.862884878746404e-05,
      "loss": 2.0006,
      "step": 192774
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.3790911436080933,
      "learning_rate": 3.862684090902669e-05,
      "loss": 2.1788,
      "step": 192775
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.3792606592178345,
      "learning_rate": 3.862483307918354e-05,
      "loss": 2.0978,
      "step": 192776
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.3113441467285156,
      "learning_rate": 3.862282529793491e-05,
      "loss": 1.8278,
      "step": 192777
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.5138816833496094,
      "learning_rate": 3.862081756528128e-05,
      "loss": 2.073,
      "step": 192778
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.4168457984924316,
      "learning_rate": 3.861880988122291e-05,
      "loss": 1.7929,
      "step": 192779
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.4166725873947144,
      "learning_rate": 3.861680224576034e-05,
      "loss": 1.9441,
      "step": 192780
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.4841827154159546,
      "learning_rate": 3.861479465889379e-05,
      "loss": 2.0685,
      "step": 192781
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.511879801750183,
      "learning_rate": 3.86127871206237e-05,
      "loss": 2.1122,
      "step": 192782
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.5012657642364502,
      "learning_rate": 3.861077963095037e-05,
      "loss": 2.1301,
      "step": 192783
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.331678032875061,
      "learning_rate": 3.8608772189874306e-05,
      "loss": 1.9652,
      "step": 192784
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.4167362451553345,
      "learning_rate": 3.860676479739573e-05,
      "loss": 2.0895,
      "step": 192785
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.4111665487289429,
      "learning_rate": 3.8604757453515144e-05,
      "loss": 2.178,
      "step": 192786
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.3722132444381714,
      "learning_rate": 3.8602750158232855e-05,
      "loss": 1.9233,
      "step": 192787
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.3706364631652832,
      "learning_rate": 3.8600742911549285e-05,
      "loss": 1.9941,
      "step": 192788
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.4638557434082031,
      "learning_rate": 3.859873571346468e-05,
      "loss": 2.0337,
      "step": 192789
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.42400324344635,
      "learning_rate": 3.8596728563979594e-05,
      "loss": 2.054,
      "step": 192790
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.5045583248138428,
      "learning_rate": 3.859472146309423e-05,
      "loss": 2.086,
      "step": 192791
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.6120408773422241,
      "learning_rate": 3.85927144108091e-05,
      "loss": 2.2104,
      "step": 192792
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.351853847503662,
      "learning_rate": 3.8590707407124525e-05,
      "loss": 1.9875,
      "step": 192793
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.3162806034088135,
      "learning_rate": 3.858870045204087e-05,
      "loss": 2.1578,
      "step": 192794
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.341633915901184,
      "learning_rate": 3.8586693545558467e-05,
      "loss": 2.0701,
      "step": 192795
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.6764708757400513,
      "learning_rate": 3.858468668767776e-05,
      "loss": 1.8408,
      "step": 192796
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.379701018333435,
      "learning_rate": 3.858267987839907e-05,
      "loss": 1.8852,
      "step": 192797
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.3770060539245605,
      "learning_rate": 3.8580673117722835e-05,
      "loss": 2.1786,
      "step": 192798
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.5147595405578613,
      "learning_rate": 3.857866640564936e-05,
      "loss": 2.0964,
      "step": 192799
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.50676429271698,
      "learning_rate": 3.8576659742179105e-05,
      "loss": 1.9582,
      "step": 192800
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.3829578161239624,
      "learning_rate": 3.857465312731227e-05,
      "loss": 2.0121,
      "step": 192801
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.4425252676010132,
      "learning_rate": 3.8572646561049427e-05,
      "loss": 2.1354,
      "step": 192802
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.312485694885254,
      "learning_rate": 3.8570640043390797e-05,
      "loss": 1.9381,
      "step": 192803
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.4222501516342163,
      "learning_rate": 3.8568633574336896e-05,
      "loss": 1.9792,
      "step": 192804
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.508080244064331,
      "learning_rate": 3.856662715388801e-05,
      "loss": 2.0858,
      "step": 192805
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.50655996799469,
      "learning_rate": 3.856462078204451e-05,
      "loss": 2.1867,
      "step": 192806
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.4082578420639038,
      "learning_rate": 3.8562614458806726e-05,
      "loss": 1.9955,
      "step": 192807
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.449750542640686,
      "learning_rate": 3.856060818417517e-05,
      "loss": 2.0148,
      "step": 192808
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.3232619762420654,
      "learning_rate": 3.855860195815006e-05,
      "loss": 2.012,
      "step": 192809
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.3384169340133667,
      "learning_rate": 3.85565957807319e-05,
      "loss": 2.0565,
      "step": 192810
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.3143092393875122,
      "learning_rate": 3.8554589651921e-05,
      "loss": 2.0031,
      "step": 192811
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.3356077671051025,
      "learning_rate": 3.855258357171774e-05,
      "loss": 1.8256,
      "step": 192812
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.4333293437957764,
      "learning_rate": 3.855057754012244e-05,
      "loss": 1.8925,
      "step": 192813
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.379563331604004,
      "learning_rate": 3.8548571557135555e-05,
      "loss": 2.108,
      "step": 192814
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.32911217212677,
      "learning_rate": 3.8546565622757385e-05,
      "loss": 1.9528,
      "step": 192815
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.2485032081604004,
      "learning_rate": 3.8544559736988403e-05,
      "loss": 1.9683,
      "step": 192816
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.3249894380569458,
      "learning_rate": 3.8542553899828935e-05,
      "loss": 1.9365,
      "step": 192817
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.4854607582092285,
      "learning_rate": 3.854054811127931e-05,
      "loss": 1.8208,
      "step": 192818
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.3714014291763306,
      "learning_rate": 3.853854237133987e-05,
      "loss": 2.0576,
      "step": 192819
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.4378819465637207,
      "learning_rate": 3.853653668001115e-05,
      "loss": 2.0682,
      "step": 192820
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.4129911661148071,
      "learning_rate": 3.853453103729335e-05,
      "loss": 2.1583,
      "step": 192821
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.5523245334625244,
      "learning_rate": 3.8532525443186956e-05,
      "loss": 1.8877,
      "step": 192822
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.366722583770752,
      "learning_rate": 3.853051989769231e-05,
      "loss": 2.094,
      "step": 192823
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.412394404411316,
      "learning_rate": 3.8528514400809784e-05,
      "loss": 1.9878,
      "step": 192824
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.4482194185256958,
      "learning_rate": 3.85265089525397e-05,
      "loss": 1.8413,
      "step": 192825
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.50933039188385,
      "learning_rate": 3.85245035528825e-05,
      "loss": 1.8797,
      "step": 192826
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.5508488416671753,
      "learning_rate": 3.852249820183848e-05,
      "loss": 2.2652,
      "step": 192827
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.4682409763336182,
      "learning_rate": 3.852049289940814e-05,
      "loss": 2.0812,
      "step": 192828
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.3783764839172363,
      "learning_rate": 3.851848764559178e-05,
      "loss": 2.0328,
      "step": 192829
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.3639332056045532,
      "learning_rate": 3.851648244038973e-05,
      "loss": 1.9646,
      "step": 192830
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.3718739748001099,
      "learning_rate": 3.8514477283802356e-05,
      "loss": 2.1575,
      "step": 192831
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.328245997428894,
      "learning_rate": 3.8512472175830167e-05,
      "loss": 2.0329,
      "step": 192832
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.4571385383605957,
      "learning_rate": 3.851046711647335e-05,
      "loss": 2.0861,
      "step": 192833
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.4670896530151367,
      "learning_rate": 3.8508462105732455e-05,
      "loss": 1.812,
      "step": 192834
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.4282493591308594,
      "learning_rate": 3.850645714360776e-05,
      "loss": 1.9524,
      "step": 192835
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.462283968925476,
      "learning_rate": 3.850445223009958e-05,
      "loss": 2.0564,
      "step": 192836
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.5817621946334839,
      "learning_rate": 3.850244736520844e-05,
      "loss": 1.8234,
      "step": 192837
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.4217451810836792,
      "learning_rate": 3.850044254893465e-05,
      "loss": 2.1739,
      "step": 192838
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.4098609685897827,
      "learning_rate": 3.849843778127847e-05,
      "loss": 2.1081,
      "step": 192839
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.45704185962677,
      "learning_rate": 3.849643306224043e-05,
      "loss": 2.0939,
      "step": 192840
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.4424066543579102,
      "learning_rate": 3.84944283918208e-05,
      "loss": 2.0299,
      "step": 192841
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.468814492225647,
      "learning_rate": 3.849242377002001e-05,
      "loss": 1.9775,
      "step": 192842
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.447889804840088,
      "learning_rate": 3.849041919683846e-05,
      "loss": 2.0788,
      "step": 192843
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.4822832345962524,
      "learning_rate": 3.848841467227639e-05,
      "loss": 2.0712,
      "step": 192844
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.335521936416626,
      "learning_rate": 3.848641019633433e-05,
      "loss": 2.0878,
      "step": 192845
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.353933572769165,
      "learning_rate": 3.8484405769012574e-05,
      "loss": 1.7801,
      "step": 192846
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.3358043432235718,
      "learning_rate": 3.848240139031146e-05,
      "loss": 1.8348,
      "step": 192847
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.264116644859314,
      "learning_rate": 3.848039706023145e-05,
      "loss": 1.9218,
      "step": 192848
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.4346851110458374,
      "learning_rate": 3.8478392778772885e-05,
      "loss": 1.9176,
      "step": 192849
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.4554226398468018,
      "learning_rate": 3.8476388545936055e-05,
      "loss": 2.0416,
      "step": 192850
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.3224903345108032,
      "learning_rate": 3.8474384361721465e-05,
      "loss": 1.8694,
      "step": 192851
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.5698068141937256,
      "learning_rate": 3.847238022612942e-05,
      "loss": 2.2671,
      "step": 192852
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.4487625360488892,
      "learning_rate": 3.847037613916024e-05,
      "loss": 1.936,
      "step": 192853
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.4653048515319824,
      "learning_rate": 3.8468372100814406e-05,
      "loss": 2.0908,
      "step": 192854
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.3364849090576172,
      "learning_rate": 3.846636811109224e-05,
      "loss": 2.3488,
      "step": 192855
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.449748158454895,
      "learning_rate": 3.846436416999404e-05,
      "loss": 2.398,
      "step": 192856
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.5410469770431519,
      "learning_rate": 3.846236027752032e-05,
      "loss": 1.878,
      "step": 192857
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.388802409172058,
      "learning_rate": 3.8460356433671336e-05,
      "loss": 2.0268,
      "step": 192858
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.6122134923934937,
      "learning_rate": 3.845835263844755e-05,
      "loss": 1.8479,
      "step": 192859
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.3584448099136353,
      "learning_rate": 3.845634889184934e-05,
      "loss": 2.138,
      "step": 192860
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.3639487028121948,
      "learning_rate": 3.8454345193877004e-05,
      "loss": 1.9774,
      "step": 192861
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.4078017473220825,
      "learning_rate": 3.845234154453086e-05,
      "loss": 2.0995,
      "step": 192862
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.4346263408660889,
      "learning_rate": 3.845033794381143e-05,
      "loss": 1.9531,
      "step": 192863
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.4294533729553223,
      "learning_rate": 3.844833439171896e-05,
      "loss": 1.9717,
      "step": 192864
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.5426383018493652,
      "learning_rate": 3.844633088825396e-05,
      "loss": 1.9554,
      "step": 192865
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.4361093044281006,
      "learning_rate": 3.844432743341673e-05,
      "loss": 1.9642,
      "step": 192866
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.4829607009887695,
      "learning_rate": 3.844232402720761e-05,
      "loss": 2.0173,
      "step": 192867
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.250644326210022,
      "learning_rate": 3.8440320669626945e-05,
      "loss": 2.0864,
      "step": 192868
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.2803535461425781,
      "learning_rate": 3.843831736067522e-05,
      "loss": 1.9135,
      "step": 192869
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.432842493057251,
      "learning_rate": 3.84363141003527e-05,
      "loss": 1.9047,
      "step": 192870
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.3388800621032715,
      "learning_rate": 3.8434310888659866e-05,
      "loss": 1.8157,
      "step": 192871
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.4469267129898071,
      "learning_rate": 3.8432307725597046e-05,
      "loss": 2.0941,
      "step": 192872
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.5286545753479004,
      "learning_rate": 3.843030461116458e-05,
      "loss": 1.9566,
      "step": 192873
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.389417290687561,
      "learning_rate": 3.8428301545362795e-05,
      "loss": 1.9648,
      "step": 192874
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.4121887683868408,
      "learning_rate": 3.8426298528192164e-05,
      "loss": 2.0253,
      "step": 192875
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.3625110387802124,
      "learning_rate": 3.842429555965301e-05,
      "loss": 2.0797,
      "step": 192876
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.5990358591079712,
      "learning_rate": 3.842229263974574e-05,
      "loss": 1.9772,
      "step": 192877
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.3750866651535034,
      "learning_rate": 3.8420289768470715e-05,
      "loss": 2.2533,
      "step": 192878
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.3578599691390991,
      "learning_rate": 3.8418286945828315e-05,
      "loss": 1.8021,
      "step": 192879
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.3708378076553345,
      "learning_rate": 3.8416284171818814e-05,
      "loss": 1.9032,
      "step": 192880
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.4375855922698975,
      "learning_rate": 3.8414281446442705e-05,
      "loss": 2.106,
      "step": 192881
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.4888372421264648,
      "learning_rate": 3.84122787697003e-05,
      "loss": 1.8288,
      "step": 192882
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.3712663650512695,
      "learning_rate": 3.841027614159202e-05,
      "loss": 2.0743,
      "step": 192883
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.5029757022857666,
      "learning_rate": 3.8408273562118205e-05,
      "loss": 1.9144,
      "step": 192884
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.338638186454773,
      "learning_rate": 3.840627103127926e-05,
      "loss": 2.0389,
      "step": 192885
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.4080442190170288,
      "learning_rate": 3.8404268549075434e-05,
      "loss": 1.8666,
      "step": 192886
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.5540469884872437,
      "learning_rate": 3.840226611550727e-05,
      "loss": 2.1284,
      "step": 192887
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.3741589784622192,
      "learning_rate": 3.840026373057501e-05,
      "loss": 2.1131,
      "step": 192888
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.7972636222839355,
      "learning_rate": 3.839826139427914e-05,
      "loss": 1.873,
      "step": 192889
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.468093752861023,
      "learning_rate": 3.8396259106619965e-05,
      "loss": 1.8279,
      "step": 192890
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.4328382015228271,
      "learning_rate": 3.839425686759785e-05,
      "loss": 1.8436,
      "step": 192891
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.2936209440231323,
      "learning_rate": 3.839225467721313e-05,
      "loss": 1.9883,
      "step": 192892
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.4589029550552368,
      "learning_rate": 3.8390252535466304e-05,
      "loss": 2.1796,
      "step": 192893
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.4189866781234741,
      "learning_rate": 3.838825044235757e-05,
      "loss": 1.9755,
      "step": 192894
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.5507700443267822,
      "learning_rate": 3.83862483978875e-05,
      "loss": 2.0146,
      "step": 192895
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.3895212411880493,
      "learning_rate": 3.838424640205635e-05,
      "loss": 2.0451,
      "step": 192896
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.5418351888656616,
      "learning_rate": 3.83822444548645e-05,
      "loss": 2.0834,
      "step": 192897
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.389672875404358,
      "learning_rate": 3.838024255631227e-05,
      "loss": 2.0856,
      "step": 192898
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.4497336149215698,
      "learning_rate": 3.837824070640013e-05,
      "loss": 2.0739,
      "step": 192899
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.4754366874694824,
      "learning_rate": 3.837623890512839e-05,
      "loss": 2.0283,
      "step": 192900
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.349377989768982,
      "learning_rate": 3.83742371524975e-05,
      "loss": 2.0237,
      "step": 192901
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.3790243864059448,
      "learning_rate": 3.837223544850777e-05,
      "loss": 1.9826,
      "step": 192902
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.3671116828918457,
      "learning_rate": 3.837023379315957e-05,
      "loss": 1.9816,
      "step": 192903
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.2512656450271606,
      "learning_rate": 3.8368232186453253e-05,
      "loss": 1.9085,
      "step": 192904
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.4496381282806396,
      "learning_rate": 3.836623062838926e-05,
      "loss": 2.205,
      "step": 192905
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.3603458404541016,
      "learning_rate": 3.836422911896786e-05,
      "loss": 1.8858,
      "step": 192906
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.4442998170852661,
      "learning_rate": 3.836222765818955e-05,
      "loss": 2.0554,
      "step": 192907
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.7112619876861572,
      "learning_rate": 3.836022624605456e-05,
      "loss": 2.171,
      "step": 192908
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.4664415121078491,
      "learning_rate": 3.835822488256349e-05,
      "loss": 2.0511,
      "step": 192909
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.6038419008255005,
      "learning_rate": 3.835622356771645e-05,
      "loss": 1.9456,
      "step": 192910
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.415537714958191,
      "learning_rate": 3.8354222301513995e-05,
      "loss": 2.0715,
      "step": 192911
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.7279123067855835,
      "learning_rate": 3.8352221083956326e-05,
      "loss": 2.0919,
      "step": 192912
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.4362785816192627,
      "learning_rate": 3.835021991504401e-05,
      "loss": 2.0265,
      "step": 192913
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.574554443359375,
      "learning_rate": 3.834821879477725e-05,
      "loss": 2.101,
      "step": 192914
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.5027878284454346,
      "learning_rate": 3.8346217723156644e-05,
      "loss": 2.1021,
      "step": 192915
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.4156973361968994,
      "learning_rate": 3.8344216700182264e-05,
      "loss": 1.9111,
      "step": 192916
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.4464476108551025,
      "learning_rate": 3.834221572585473e-05,
      "loss": 2.1756,
      "step": 192917
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.5400564670562744,
      "learning_rate": 3.834021480017422e-05,
      "loss": 1.8914,
      "step": 192918
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.3258912563323975,
      "learning_rate": 3.833821392314126e-05,
      "loss": 1.9806,
      "step": 192919
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.5303922891616821,
      "learning_rate": 3.833621309475612e-05,
      "loss": 1.9953,
      "step": 192920
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.4457017183303833,
      "learning_rate": 3.8334212315019294e-05,
      "loss": 1.9787,
      "step": 192921
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.3857117891311646,
      "learning_rate": 3.833221158393106e-05,
      "loss": 1.9523,
      "step": 192922
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.4548115730285645,
      "learning_rate": 3.833021090149181e-05,
      "loss": 2.1617,
      "step": 192923
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.4407380819320679,
      "learning_rate": 3.832821026770184e-05,
      "loss": 2.114,
      "step": 192924
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.5651695728302002,
      "learning_rate": 3.832620968256169e-05,
      "loss": 1.8836,
      "step": 192925
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.4173895120620728,
      "learning_rate": 3.832420914607156e-05,
      "loss": 1.7557,
      "step": 192926
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.3960013389587402,
      "learning_rate": 3.8322208658231945e-05,
      "loss": 2.1531,
      "step": 192927
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.4909862279891968,
      "learning_rate": 3.832020821904318e-05,
      "loss": 1.9813,
      "step": 192928
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.3438353538513184,
      "learning_rate": 3.8318207828505634e-05,
      "loss": 2.06,
      "step": 192929
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.4083932638168335,
      "learning_rate": 3.83162074866196e-05,
      "loss": 1.9968,
      "step": 192930
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.3805192708969116,
      "learning_rate": 3.831420719338558e-05,
      "loss": 2.0263,
      "step": 192931
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.4574412107467651,
      "learning_rate": 3.8312206948803846e-05,
      "loss": 1.6936,
      "step": 192932
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.2865549325942993,
      "learning_rate": 3.831020675287486e-05,
      "loss": 2.0016,
      "step": 192933
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.3649243116378784,
      "learning_rate": 3.8308206605598955e-05,
      "loss": 2.033,
      "step": 192934
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.674494743347168,
      "learning_rate": 3.830620650697643e-05,
      "loss": 2.0093,
      "step": 192935
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.3980964422225952,
      "learning_rate": 3.830420645700779e-05,
      "loss": 2.0027,
      "step": 192936
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.4303882122039795,
      "learning_rate": 3.830220645569333e-05,
      "loss": 2.0591,
      "step": 192937
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.7079689502716064,
      "learning_rate": 3.830020650303338e-05,
      "loss": 2.2611,
      "step": 192938
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.3586950302124023,
      "learning_rate": 3.829820659902841e-05,
      "loss": 2.0248,
      "step": 192939
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.3793835639953613,
      "learning_rate": 3.829620674367872e-05,
      "loss": 2.2257,
      "step": 192940
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.6456395387649536,
      "learning_rate": 3.829420693698468e-05,
      "loss": 2.1512,
      "step": 192941
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.7182891368865967,
      "learning_rate": 3.8292207178946754e-05,
      "loss": 2.0102,
      "step": 192942
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.4633681774139404,
      "learning_rate": 3.8290207469565136e-05,
      "loss": 2.215,
      "step": 192943
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.648756980895996,
      "learning_rate": 3.828820780884043e-05,
      "loss": 2.2044,
      "step": 192944
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.3411320447921753,
      "learning_rate": 3.8286208196772836e-05,
      "loss": 2.0408,
      "step": 192945
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.5458455085754395,
      "learning_rate": 3.8284208633362825e-05,
      "loss": 2.0392,
      "step": 192946
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.5337053537368774,
      "learning_rate": 3.8282209118610615e-05,
      "loss": 2.0434,
      "step": 192947
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.3388869762420654,
      "learning_rate": 3.8280209652516755e-05,
      "loss": 2.2323,
      "step": 192948
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.45614492893219,
      "learning_rate": 3.827821023508151e-05,
      "loss": 1.7625,
      "step": 192949
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.4978539943695068,
      "learning_rate": 3.8276210866305303e-05,
      "loss": 1.9405,
      "step": 192950
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.4999703168869019,
      "learning_rate": 3.827421154618854e-05,
      "loss": 2.1127,
      "step": 192951
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.325553297996521,
      "learning_rate": 3.8272212274731486e-05,
      "loss": 1.868,
      "step": 192952
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.3045096397399902,
      "learning_rate": 3.827021305193454e-05,
      "loss": 2.1753,
      "step": 192953
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.320916771888733,
      "learning_rate": 3.826821387779817e-05,
      "loss": 2.0169,
      "step": 192954
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.4929914474487305,
      "learning_rate": 3.826621475232261e-05,
      "loss": 1.9533,
      "step": 192955
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.4819364547729492,
      "learning_rate": 3.826421567550836e-05,
      "loss": 2.0,
      "step": 192956
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.3609488010406494,
      "learning_rate": 3.8262216647355716e-05,
      "loss": 1.9625,
      "step": 192957
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.5593723058700562,
      "learning_rate": 3.826021766786508e-05,
      "loss": 2.1961,
      "step": 192958
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.6836262941360474,
      "learning_rate": 3.825821873703675e-05,
      "loss": 2.1084,
      "step": 192959
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.4676234722137451,
      "learning_rate": 3.825621985487123e-05,
      "loss": 2.0284,
      "step": 192960
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.2738951444625854,
      "learning_rate": 3.8254221021368745e-05,
      "loss": 1.9492,
      "step": 192961
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.2926498651504517,
      "learning_rate": 3.82522222365298e-05,
      "loss": 2.0373,
      "step": 192962
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.3873072862625122,
      "learning_rate": 3.825022350035474e-05,
      "loss": 2.0251,
      "step": 192963
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.3814858198165894,
      "learning_rate": 3.8248224812843875e-05,
      "loss": 2.0125,
      "step": 192964
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.3787312507629395,
      "learning_rate": 3.824622617399755e-05,
      "loss": 2.0224,
      "step": 192965
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.370256781578064,
      "learning_rate": 3.8244227583816235e-05,
      "loss": 2.1035,
      "step": 192966
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.330464243888855,
      "learning_rate": 3.824222904230022e-05,
      "loss": 1.7813,
      "step": 192967
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.4503655433654785,
      "learning_rate": 3.824023054944998e-05,
      "loss": 1.8783,
      "step": 192968
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.3569096326828003,
      "learning_rate": 3.823823210526581e-05,
      "loss": 1.9719,
      "step": 192969
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.3410364389419556,
      "learning_rate": 3.823623370974811e-05,
      "loss": 2.0408,
      "step": 192970
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.3520770072937012,
      "learning_rate": 3.823423536289715e-05,
      "loss": 1.9141,
      "step": 192971
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.3964380025863647,
      "learning_rate": 3.823223706471346e-05,
      "loss": 1.9188,
      "step": 192972
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.428139090538025,
      "learning_rate": 3.8230238815197264e-05,
      "loss": 1.9781,
      "step": 192973
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.3955219984054565,
      "learning_rate": 3.8228240614349084e-05,
      "loss": 1.7406,
      "step": 192974
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.3381465673446655,
      "learning_rate": 3.822624246216917e-05,
      "loss": 1.899,
      "step": 192975
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.4463655948638916,
      "learning_rate": 3.8224244358658015e-05,
      "loss": 2.0449,
      "step": 192976
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.4058412313461304,
      "learning_rate": 3.822224630381584e-05,
      "loss": 2.0193,
      "step": 192977
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.464592695236206,
      "learning_rate": 3.822024829764313e-05,
      "loss": 1.8829,
      "step": 192978
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.4768881797790527,
      "learning_rate": 3.821825034014016e-05,
      "loss": 1.8583,
      "step": 192979
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.5122129917144775,
      "learning_rate": 3.821625243130741e-05,
      "loss": 2.0975,
      "step": 192980
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.4967865943908691,
      "learning_rate": 3.821425457114514e-05,
      "loss": 2.052,
      "step": 192981
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.3798794746398926,
      "learning_rate": 3.8212256759653936e-05,
      "loss": 1.9929,
      "step": 192982
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.421712875366211,
      "learning_rate": 3.821025899683386e-05,
      "loss": 2.0998,
      "step": 192983
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.5399025678634644,
      "learning_rate": 3.820826128268553e-05,
      "loss": 2.1275,
      "step": 192984
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.3431708812713623,
      "learning_rate": 3.820626361720912e-05,
      "loss": 2.0107,
      "step": 192985
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.3292547464370728,
      "learning_rate": 3.820426600040518e-05,
      "loss": 1.9857,
      "step": 192986
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.3568949699401855,
      "learning_rate": 3.820226843227397e-05,
      "loss": 2.2025,
      "step": 192987
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.3249655961990356,
      "learning_rate": 3.820027091281603e-05,
      "loss": 2.029,
      "step": 192988
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.4085613489151,
      "learning_rate": 3.819827344203146e-05,
      "loss": 2.0449,
      "step": 192989
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.457564353942871,
      "learning_rate": 3.819627601992081e-05,
      "loss": 1.8621,
      "step": 192990
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.3813953399658203,
      "learning_rate": 3.81942786464844e-05,
      "loss": 2.1112,
      "step": 192991
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.385655403137207,
      "learning_rate": 3.819228132172265e-05,
      "loss": 1.8638,
      "step": 192992
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.4542896747589111,
      "learning_rate": 3.8190284045635833e-05,
      "loss": 1.9563,
      "step": 192993
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.3351597785949707,
      "learning_rate": 3.818828681822451e-05,
      "loss": 2.0871,
      "step": 192994
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.3607357740402222,
      "learning_rate": 3.818628963948882e-05,
      "loss": 1.7748,
      "step": 192995
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.4862637519836426,
      "learning_rate": 3.8184292509429295e-05,
      "loss": 2.1394,
      "step": 192996
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.5132313966751099,
      "learning_rate": 3.8182295428046164e-05,
      "loss": 1.8381,
      "step": 192997
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.3313432931900024,
      "learning_rate": 3.818029839533999e-05,
      "loss": 2.2532,
      "step": 192998
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.402306318283081,
      "learning_rate": 3.8178301411310955e-05,
      "loss": 1.8505,
      "step": 192999
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.29436457157135,
      "learning_rate": 3.817630447595964e-05,
      "loss": 2.1114,
      "step": 193000
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.4130208492279053,
      "learning_rate": 3.817430758928616e-05,
      "loss": 2.2565,
      "step": 193001
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.6504638195037842,
      "learning_rate": 3.81723107512911e-05,
      "loss": 2.0049,
      "step": 193002
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.472370982170105,
      "learning_rate": 3.817031396197467e-05,
      "loss": 1.9798,
      "step": 193003
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.3660284280776978,
      "learning_rate": 3.8168317221337405e-05,
      "loss": 1.9677,
      "step": 193004
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.4447969198226929,
      "learning_rate": 3.8166320529379535e-05,
      "loss": 1.7566,
      "step": 193005
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.2768412828445435,
      "learning_rate": 3.816432388610152e-05,
      "loss": 2.0405,
      "step": 193006
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.4374712705612183,
      "learning_rate": 3.8162327291503705e-05,
      "loss": 2.0814,
      "step": 193007
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.5225554704666138,
      "learning_rate": 3.816033074558648e-05,
      "loss": 2.0866,
      "step": 193008
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.3981496095657349,
      "learning_rate": 3.815833424835011e-05,
      "loss": 1.9632,
      "step": 193009
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.8638641834259033,
      "learning_rate": 3.81563377997951e-05,
      "loss": 2.1601,
      "step": 193010
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.372336983680725,
      "learning_rate": 3.815434139992175e-05,
      "loss": 1.8968,
      "step": 193011
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.2679585218429565,
      "learning_rate": 3.8152345048730456e-05,
      "loss": 2.0256,
      "step": 193012
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.2933876514434814,
      "learning_rate": 3.8150348746221625e-05,
      "loss": 1.7658,
      "step": 193013
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.3510839939117432,
      "learning_rate": 3.8148352492395584e-05,
      "loss": 1.7887,
      "step": 193014
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.5249130725860596,
      "learning_rate": 3.8146356287252634e-05,
      "loss": 1.9203,
      "step": 193015
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.5039494037628174,
      "learning_rate": 3.8144360130793275e-05,
      "loss": 2.0473,
      "step": 193016
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.3824775218963623,
      "learning_rate": 3.8142364023017765e-05,
      "loss": 2.0615,
      "step": 193017
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.34567391872406,
      "learning_rate": 3.814036796392662e-05,
      "loss": 1.9768,
      "step": 193018
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.3316186666488647,
      "learning_rate": 3.8138371953520094e-05,
      "loss": 2.0899,
      "step": 193019
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.5847647190093994,
      "learning_rate": 3.813637599179853e-05,
      "loss": 2.1827,
      "step": 193020
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.4768966436386108,
      "learning_rate": 3.813438007876242e-05,
      "loss": 2.0302,
      "step": 193021
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.5250499248504639,
      "learning_rate": 3.813238421441209e-05,
      "loss": 1.7813,
      "step": 193022
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.4598660469055176,
      "learning_rate": 3.813038839874779e-05,
      "loss": 1.9274,
      "step": 193023
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.344868779182434,
      "learning_rate": 3.812839263177011e-05,
      "loss": 2.0616,
      "step": 193024
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.5158319473266602,
      "learning_rate": 3.812639691347925e-05,
      "loss": 2.1397,
      "step": 193025
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.5209076404571533,
      "learning_rate": 3.812440124387561e-05,
      "loss": 1.8431,
      "step": 193026
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.287124752998352,
      "learning_rate": 3.812240562295963e-05,
      "loss": 1.9936,
      "step": 193027
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.4208223819732666,
      "learning_rate": 3.8120410050731665e-05,
      "loss": 1.9256,
      "step": 193028
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.3517507314682007,
      "learning_rate": 3.811841452719199e-05,
      "loss": 1.7974,
      "step": 193029
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.4197866916656494,
      "learning_rate": 3.81164190523411e-05,
      "loss": 1.8907,
      "step": 193030
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.3584489822387695,
      "learning_rate": 3.811442362617929e-05,
      "loss": 2.0739,
      "step": 193031
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.3647499084472656,
      "learning_rate": 3.8112428248706906e-05,
      "loss": 1.9115,
      "step": 193032
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.2937402725219727,
      "learning_rate": 3.8110432919924436e-05,
      "loss": 1.9645,
      "step": 193033
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.476088047027588,
      "learning_rate": 3.810843763983209e-05,
      "loss": 1.8541,
      "step": 193034
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.2766590118408203,
      "learning_rate": 3.8106442408430424e-05,
      "loss": 2.1,
      "step": 193035
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.4122505187988281,
      "learning_rate": 3.8104447225719714e-05,
      "loss": 2.3118,
      "step": 193036
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.3690366744995117,
      "learning_rate": 3.810245209170032e-05,
      "loss": 2.0308,
      "step": 193037
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.4884001016616821,
      "learning_rate": 3.810045700637254e-05,
      "loss": 2.0747,
      "step": 193038
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.39963698387146,
      "learning_rate": 3.809846196973691e-05,
      "loss": 2.0643,
      "step": 193039
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.3504575490951538,
      "learning_rate": 3.809646698179367e-05,
      "loss": 1.9815,
      "step": 193040
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.4435571432113647,
      "learning_rate": 3.809447204254331e-05,
      "loss": 2.0132,
      "step": 193041
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.422330379486084,
      "learning_rate": 3.8092477151986034e-05,
      "loss": 1.849,
      "step": 193042
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.42864990234375,
      "learning_rate": 3.8090482310122474e-05,
      "loss": 2.0135,
      "step": 193043
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.5266169309616089,
      "learning_rate": 3.808848751695266e-05,
      "loss": 1.9677,
      "step": 193044
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.395530343055725,
      "learning_rate": 3.808649277247726e-05,
      "loss": 2.1742,
      "step": 193045
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.3698776960372925,
      "learning_rate": 3.808449807669641e-05,
      "loss": 1.9251,
      "step": 193046
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.5511431694030762,
      "learning_rate": 3.808250342961071e-05,
      "loss": 1.9789,
      "step": 193047
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.3621065616607666,
      "learning_rate": 3.8080508831220324e-05,
      "loss": 2.2235,
      "step": 193048
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.5270580053329468,
      "learning_rate": 3.807851428152585e-05,
      "loss": 1.9626,
      "step": 193049
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.4223212003707886,
      "learning_rate": 3.8076519780527396e-05,
      "loss": 2.1271,
      "step": 193050
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.5101721286773682,
      "learning_rate": 3.8074525328225516e-05,
      "loss": 2.2257,
      "step": 193051
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.4209537506103516,
      "learning_rate": 3.807253092462045e-05,
      "loss": 2.0422,
      "step": 193052
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.3082873821258545,
      "learning_rate": 3.8070536569712727e-05,
      "loss": 2.1506,
      "step": 193053
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.3804157972335815,
      "learning_rate": 3.806854226350259e-05,
      "loss": 1.9444,
      "step": 193054
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.473908543586731,
      "learning_rate": 3.806654800599056e-05,
      "loss": 1.9408,
      "step": 193055
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.6664284467697144,
      "learning_rate": 3.806455379717678e-05,
      "loss": 1.9422,
      "step": 193056
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.3902313709259033,
      "learning_rate": 3.8062559637061806e-05,
      "loss": 1.9087,
      "step": 193057
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.3754830360412598,
      "learning_rate": 3.806056552564588e-05,
      "loss": 2.0126,
      "step": 193058
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.4139775037765503,
      "learning_rate": 3.80585714629295e-05,
      "loss": 2.065,
      "step": 193059
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.4241305589675903,
      "learning_rate": 3.805657744891293e-05,
      "loss": 1.9302,
      "step": 193060
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.5094506740570068,
      "learning_rate": 3.8054583483596704e-05,
      "loss": 1.9926,
      "step": 193061
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.3149131536483765,
      "learning_rate": 3.805258956698095e-05,
      "loss": 1.8644,
      "step": 193062
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.5236735343933105,
      "learning_rate": 3.805059569906624e-05,
      "loss": 1.9954,
      "step": 193063
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.5006077289581299,
      "learning_rate": 3.804860187985278e-05,
      "loss": 1.9953,
      "step": 193064
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.5412923097610474,
      "learning_rate": 3.8046608109341096e-05,
      "loss": 2.1028,
      "step": 193065
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.5627977848052979,
      "learning_rate": 3.804461438753145e-05,
      "loss": 2.3525,
      "step": 193066
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.4564762115478516,
      "learning_rate": 3.804262071442438e-05,
      "loss": 2.0243,
      "step": 193067
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.4609211683273315,
      "learning_rate": 3.804062709001998e-05,
      "loss": 2.0042,
      "step": 193068
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.325631856918335,
      "learning_rate": 3.803863351431886e-05,
      "loss": 1.9508,
      "step": 193069
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.466861605644226,
      "learning_rate": 3.8036639987321215e-05,
      "loss": 2.1042,
      "step": 193070
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.4187061786651611,
      "learning_rate": 3.8034646509027576e-05,
      "loss": 2.1014,
      "step": 193071
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.389712929725647,
      "learning_rate": 3.8032653079438217e-05,
      "loss": 1.9102,
      "step": 193072
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.336138129234314,
      "learning_rate": 3.8030659698553624e-05,
      "loss": 1.9033,
      "step": 193073
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.3566018342971802,
      "learning_rate": 3.802866636637394e-05,
      "loss": 1.8404,
      "step": 193074
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.3305786848068237,
      "learning_rate": 3.802667308289973e-05,
      "loss": 1.8555,
      "step": 193075
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.3947464227676392,
      "learning_rate": 3.8024679848131255e-05,
      "loss": 1.9818,
      "step": 193076
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.3623509407043457,
      "learning_rate": 3.8022686662069025e-05,
      "loss": 2.158,
      "step": 193077
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.5055100917816162,
      "learning_rate": 3.802069352471323e-05,
      "loss": 2.0677,
      "step": 193078
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.4640758037567139,
      "learning_rate": 3.8018700436064476e-05,
      "loss": 2.0631,
      "step": 193079
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.3919926881790161,
      "learning_rate": 3.8016707396122856e-05,
      "loss": 2.2327,
      "step": 193080
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.355977177619934,
      "learning_rate": 3.8014714404888944e-05,
      "loss": 1.9069,
      "step": 193081
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.4250835180282593,
      "learning_rate": 3.801272146236297e-05,
      "loss": 2.0519,
      "step": 193082
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.480545163154602,
      "learning_rate": 3.801072856854547e-05,
      "loss": 1.9455,
      "step": 193083
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.4506051540374756,
      "learning_rate": 3.8008735723436634e-05,
      "loss": 1.9867,
      "step": 193084
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.374442219734192,
      "learning_rate": 3.800674292703704e-05,
      "loss": 1.8424,
      "step": 193085
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.445635199546814,
      "learning_rate": 3.800475017934681e-05,
      "loss": 1.9075,
      "step": 193086
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.4042959213256836,
      "learning_rate": 3.800275748036652e-05,
      "loss": 2.0428,
      "step": 193087
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.354811429977417,
      "learning_rate": 3.80007648300964e-05,
      "loss": 2.0376,
      "step": 193088
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.3180158138275146,
      "learning_rate": 3.799877222853691e-05,
      "loss": 1.9355,
      "step": 193089
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.6561044454574585,
      "learning_rate": 3.799677967568836e-05,
      "loss": 2.1106,
      "step": 193090
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.5550373792648315,
      "learning_rate": 3.799478717155131e-05,
      "loss": 1.9738,
      "step": 193091
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.419847011566162,
      "learning_rate": 3.79927947161258e-05,
      "loss": 1.8669,
      "step": 193092
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.4425721168518066,
      "learning_rate": 3.799080230941245e-05,
      "loss": 1.8384,
      "step": 193093
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.7575337886810303,
      "learning_rate": 3.7988809951411505e-05,
      "loss": 1.9481,
      "step": 193094
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.3742425441741943,
      "learning_rate": 3.798681764212346e-05,
      "loss": 1.9154,
      "step": 193095
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.2764805555343628,
      "learning_rate": 3.798482538154851e-05,
      "loss": 1.9034,
      "step": 193096
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.4442399740219116,
      "learning_rate": 3.798283316968723e-05,
      "loss": 2.1611,
      "step": 193097
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.43175208568573,
      "learning_rate": 3.798084100653985e-05,
      "loss": 2.1564,
      "step": 193098
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.4023821353912354,
      "learning_rate": 3.7978848892106804e-05,
      "loss": 2.0691,
      "step": 193099
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.6636199951171875,
      "learning_rate": 3.797685682638836e-05,
      "loss": 1.8929,
      "step": 193100
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.3455754518508911,
      "learning_rate": 3.797486480938504e-05,
      "loss": 1.9852,
      "step": 193101
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.4135175943374634,
      "learning_rate": 3.797287284109709e-05,
      "loss": 1.8461,
      "step": 193102
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.4217860698699951,
      "learning_rate": 3.797088092152497e-05,
      "loss": 1.8842,
      "step": 193103
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.6562515497207642,
      "learning_rate": 3.7968889050669014e-05,
      "loss": 2.0748,
      "step": 193104
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.4171993732452393,
      "learning_rate": 3.796689722852949e-05,
      "loss": 2.2008,
      "step": 193105
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.4613739252090454,
      "learning_rate": 3.796490545510697e-05,
      "loss": 2.0565,
      "step": 193106
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.4373849630355835,
      "learning_rate": 3.796291373040173e-05,
      "loss": 2.079,
      "step": 193107
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.5381895303726196,
      "learning_rate": 3.796092205441403e-05,
      "loss": 1.9544,
      "step": 193108
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.491479516029358,
      "learning_rate": 3.7958930427144426e-05,
      "loss": 2.0023,
      "step": 193109
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.4472612142562866,
      "learning_rate": 3.795693884859322e-05,
      "loss": 2.2222,
      "step": 193110
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.7477933168411255,
      "learning_rate": 3.795494731876068e-05,
      "loss": 2.2133,
      "step": 193111
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.4333040714263916,
      "learning_rate": 3.7952955837647325e-05,
      "loss": 1.9035,
      "step": 193112
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.582213044166565,
      "learning_rate": 3.795096440525347e-05,
      "loss": 2.1261,
      "step": 193113
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.3186393976211548,
      "learning_rate": 3.794897302157941e-05,
      "loss": 2.0346,
      "step": 193114
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.3862978219985962,
      "learning_rate": 3.7946981686625645e-05,
      "loss": 1.8031,
      "step": 193115
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.2891607284545898,
      "learning_rate": 3.794499040039247e-05,
      "loss": 2.0403,
      "step": 193116
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.446040391921997,
      "learning_rate": 3.794299916288023e-05,
      "loss": 2.0123,
      "step": 193117
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.3585569858551025,
      "learning_rate": 3.7941007974089374e-05,
      "loss": 2.0356,
      "step": 193118
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.4238744974136353,
      "learning_rate": 3.793901683402015e-05,
      "loss": 1.8717,
      "step": 193119
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.4007859230041504,
      "learning_rate": 3.793702574267312e-05,
      "loss": 1.9157,
      "step": 193120
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.3494722843170166,
      "learning_rate": 3.7935034700048514e-05,
      "loss": 1.8828,
      "step": 193121
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.3789855241775513,
      "learning_rate": 3.793304370614674e-05,
      "loss": 1.8107,
      "step": 193122
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.4313294887542725,
      "learning_rate": 3.7931052760968114e-05,
      "loss": 1.8316,
      "step": 193123
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.3798136711120605,
      "learning_rate": 3.792906186451309e-05,
      "loss": 2.1127,
      "step": 193124
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.5814272165298462,
      "learning_rate": 3.792707101678195e-05,
      "loss": 1.9471,
      "step": 193125
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.536399006843567,
      "learning_rate": 3.792508021777517e-05,
      "loss": 1.7962,
      "step": 193126
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.4148623943328857,
      "learning_rate": 3.7923089467493014e-05,
      "loss": 2.1675,
      "step": 193127
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.6156476736068726,
      "learning_rate": 3.792109876593602e-05,
      "loss": 2.2226,
      "step": 193128
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.4549247026443481,
      "learning_rate": 3.791910811310431e-05,
      "loss": 2.0616,
      "step": 193129
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.3463215827941895,
      "learning_rate": 3.7917117508998466e-05,
      "loss": 1.7347,
      "step": 193130
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.5561602115631104,
      "learning_rate": 3.7915126953618705e-05,
      "loss": 2.1531,
      "step": 193131
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.6023813486099243,
      "learning_rate": 3.791313644696553e-05,
      "loss": 1.865,
      "step": 193132
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.4023149013519287,
      "learning_rate": 3.791114598903918e-05,
      "loss": 1.8857,
      "step": 193133
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.4024536609649658,
      "learning_rate": 3.790915557984022e-05,
      "loss": 1.8196,
      "step": 193134
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.4043824672698975,
      "learning_rate": 3.790716521936879e-05,
      "loss": 1.9081,
      "step": 193135
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.3816760778427124,
      "learning_rate": 3.79051749076254e-05,
      "loss": 1.9096,
      "step": 193136
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.3855458498001099,
      "learning_rate": 3.790318464461034e-05,
      "loss": 1.9105,
      "step": 193137
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.528656005859375,
      "learning_rate": 3.79011944303241e-05,
      "loss": 2.2704,
      "step": 193138
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.6205942630767822,
      "learning_rate": 3.789920426476688e-05,
      "loss": 2.0867,
      "step": 193139
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.4012848138809204,
      "learning_rate": 3.789721414793932e-05,
      "loss": 1.9512,
      "step": 193140
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.2456181049346924,
      "learning_rate": 3.7895224079841436e-05,
      "loss": 2.1665,
      "step": 193141
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.4093893766403198,
      "learning_rate": 3.7893234060473875e-05,
      "loss": 2.1064,
      "step": 193142
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.505406141281128,
      "learning_rate": 3.789124408983684e-05,
      "loss": 1.9788,
      "step": 193143
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.4364200830459595,
      "learning_rate": 3.788925416793082e-05,
      "loss": 2.3037,
      "step": 193144
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.3650814294815063,
      "learning_rate": 3.788726429475608e-05,
      "loss": 1.9157,
      "step": 193145
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.3200310468673706,
      "learning_rate": 3.7885274470313164e-05,
      "loss": 2.1842,
      "step": 193146
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.4969580173492432,
      "learning_rate": 3.78832846946022e-05,
      "loss": 1.8325,
      "step": 193147
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.3238105773925781,
      "learning_rate": 3.788129496762376e-05,
      "loss": 2.1213,
      "step": 193148
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.3212931156158447,
      "learning_rate": 3.787930528937806e-05,
      "loss": 1.9454,
      "step": 193149
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.6193161010742188,
      "learning_rate": 3.7877315659865613e-05,
      "loss": 1.9379,
      "step": 193150
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.3176487684249878,
      "learning_rate": 3.787532607908666e-05,
      "loss": 1.9277,
      "step": 193151
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.5622864961624146,
      "learning_rate": 3.787333654704174e-05,
      "loss": 1.8881,
      "step": 193152
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.4238743782043457,
      "learning_rate": 3.7871347063731016e-05,
      "loss": 2.0395,
      "step": 193153
    },
    {
      "epoch": 2.51,
      "grad_norm": 1.5542643070220947,
      "learning_rate": 3.786935762915501e-05,
      "loss": 2.1486,
      "step": 193154
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.4103407859802246,
      "learning_rate": 3.786736824331398e-05,
      "loss": 1.9525,
      "step": 193155
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.3927288055419922,
      "learning_rate": 3.78653789062084e-05,
      "loss": 2.2352,
      "step": 193156
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.2988313436508179,
      "learning_rate": 3.7863389617838534e-05,
      "loss": 2.0186,
      "step": 193157
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.5108153820037842,
      "learning_rate": 3.786140037820495e-05,
      "loss": 2.1505,
      "step": 193158
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.5336008071899414,
      "learning_rate": 3.785941118730775e-05,
      "loss": 2.1294,
      "step": 193159
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.4269721508026123,
      "learning_rate": 3.78574220451475e-05,
      "loss": 2.0311,
      "step": 193160
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.2947221994400024,
      "learning_rate": 3.785543295172443e-05,
      "loss": 1.9771,
      "step": 193161
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.4582668542861938,
      "learning_rate": 3.785344390703904e-05,
      "loss": 2.0035,
      "step": 193162
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.379019021987915,
      "learning_rate": 3.785145491109159e-05,
      "loss": 2.0848,
      "step": 193163
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.4334944486618042,
      "learning_rate": 3.784946596388263e-05,
      "loss": 2.0077,
      "step": 193164
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.3938322067260742,
      "learning_rate": 3.784747706541228e-05,
      "loss": 2.1117,
      "step": 193165
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.456119418144226,
      "learning_rate": 3.7845488215681107e-05,
      "loss": 1.8884,
      "step": 193166
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.3644076585769653,
      "learning_rate": 3.7843499414689315e-05,
      "loss": 1.9783,
      "step": 193167
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.3083550930023193,
      "learning_rate": 3.784151066243744e-05,
      "loss": 1.9011,
      "step": 193168
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.413493037223816,
      "learning_rate": 3.78395219589257e-05,
      "loss": 2.0382,
      "step": 193169
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.526628851890564,
      "learning_rate": 3.7837533304154713e-05,
      "loss": 1.8462,
      "step": 193170
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.480401635169983,
      "learning_rate": 3.7835544698124495e-05,
      "loss": 2.0415,
      "step": 193171
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.3864322900772095,
      "learning_rate": 3.7833556140835697e-05,
      "loss": 1.9726,
      "step": 193172
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.306828498840332,
      "learning_rate": 3.783156763228851e-05,
      "loss": 1.9909,
      "step": 193173
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.413980484008789,
      "learning_rate": 3.782957917248346e-05,
      "loss": 2.0223,
      "step": 193174
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.4205983877182007,
      "learning_rate": 3.782759076142079e-05,
      "loss": 1.9536,
      "step": 193175
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.3720465898513794,
      "learning_rate": 3.7825602399100964e-05,
      "loss": 1.8947,
      "step": 193176
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.4556996822357178,
      "learning_rate": 3.782361408552428e-05,
      "loss": 1.965,
      "step": 193177
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.4384782314300537,
      "learning_rate": 3.782162582069117e-05,
      "loss": 1.9502,
      "step": 193178
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.3753341436386108,
      "learning_rate": 3.781963760460191e-05,
      "loss": 2.1125,
      "step": 193179
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.3577604293823242,
      "learning_rate": 3.7817649437256984e-05,
      "loss": 2.2664,
      "step": 193180
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.7154473066329956,
      "learning_rate": 3.781566131865663e-05,
      "loss": 1.9095,
      "step": 193181
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.3503659963607788,
      "learning_rate": 3.781367324880136e-05,
      "loss": 2.1438,
      "step": 193182
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.4205478429794312,
      "learning_rate": 3.781168522769149e-05,
      "loss": 2.0773,
      "step": 193183
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.465787649154663,
      "learning_rate": 3.7809697255327365e-05,
      "loss": 1.9422,
      "step": 193184
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.2926905155181885,
      "learning_rate": 3.7807709331709315e-05,
      "loss": 1.9165,
      "step": 193185
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.41677987575531,
      "learning_rate": 3.78057214568378e-05,
      "loss": 1.8996,
      "step": 193186
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.4278192520141602,
      "learning_rate": 3.7803733630713095e-05,
      "loss": 1.8049,
      "step": 193187
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.3628102540969849,
      "learning_rate": 3.780174585333573e-05,
      "loss": 1.9785,
      "step": 193188
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.3400132656097412,
      "learning_rate": 3.779975812470594e-05,
      "loss": 1.9233,
      "step": 193189
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.4005165100097656,
      "learning_rate": 3.779777044482411e-05,
      "loss": 1.9064,
      "step": 193190
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.5601108074188232,
      "learning_rate": 3.7795782813690565e-05,
      "loss": 1.7535,
      "step": 193191
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.3340150117874146,
      "learning_rate": 3.779379523130582e-05,
      "loss": 2.1399,
      "step": 193192
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.4190291166305542,
      "learning_rate": 3.779180769767005e-05,
      "loss": 2.0545,
      "step": 193193
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.4126811027526855,
      "learning_rate": 3.778982021278385e-05,
      "loss": 2.2269,
      "step": 193194
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.5916602611541748,
      "learning_rate": 3.7787832776647455e-05,
      "loss": 2.179,
      "step": 193195
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.459908366203308,
      "learning_rate": 3.778584538926116e-05,
      "loss": 1.9141,
      "step": 193196
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.4261664152145386,
      "learning_rate": 3.778385805062551e-05,
      "loss": 1.9783,
      "step": 193197
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.4102420806884766,
      "learning_rate": 3.7781870760740784e-05,
      "loss": 2.0228,
      "step": 193198
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.4249610900878906,
      "learning_rate": 3.77798835196073e-05,
      "loss": 2.0345,
      "step": 193199
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.389013409614563,
      "learning_rate": 3.777789632722556e-05,
      "loss": 1.8972,
      "step": 193200
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.380372405052185,
      "learning_rate": 3.7775909183595845e-05,
      "loss": 1.9488,
      "step": 193201
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.369394302368164,
      "learning_rate": 3.777392208871847e-05,
      "loss": 2.1214,
      "step": 193202
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.3594609498977661,
      "learning_rate": 3.777193504259396e-05,
      "loss": 1.7687,
      "step": 193203
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.3977437019348145,
      "learning_rate": 3.776994804522252e-05,
      "loss": 2.0009,
      "step": 193204
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.6010105609893799,
      "learning_rate": 3.776796109660465e-05,
      "loss": 1.9638,
      "step": 193205
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.3609199523925781,
      "learning_rate": 3.776597419674068e-05,
      "loss": 1.9048,
      "step": 193206
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.3778488636016846,
      "learning_rate": 3.7763987345630944e-05,
      "loss": 2.0403,
      "step": 193207
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.5545663833618164,
      "learning_rate": 3.776200054327581e-05,
      "loss": 1.9814,
      "step": 193208
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.463637351989746,
      "learning_rate": 3.7760013789675705e-05,
      "loss": 1.92,
      "step": 193209
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.433411717414856,
      "learning_rate": 3.77580270848309e-05,
      "loss": 1.8802,
      "step": 193210
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.3705915212631226,
      "learning_rate": 3.775604042874189e-05,
      "loss": 2.1594,
      "step": 193211
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.3538929224014282,
      "learning_rate": 3.775405382140899e-05,
      "loss": 1.9923,
      "step": 193212
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.382827639579773,
      "learning_rate": 3.775206726283255e-05,
      "loss": 1.9336,
      "step": 193213
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.3485010862350464,
      "learning_rate": 3.7750080753012915e-05,
      "loss": 1.7696,
      "step": 193214
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.3336448669433594,
      "learning_rate": 3.7748094291950534e-05,
      "loss": 2.0168,
      "step": 193215
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.335994839668274,
      "learning_rate": 3.774610787964566e-05,
      "loss": 1.9531,
      "step": 193216
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.4463061094284058,
      "learning_rate": 3.7744121516098815e-05,
      "loss": 2.0847,
      "step": 193217
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.54076087474823,
      "learning_rate": 3.774213520131023e-05,
      "loss": 2.0109,
      "step": 193218
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.3448244333267212,
      "learning_rate": 3.7740148935280444e-05,
      "loss": 1.9754,
      "step": 193219
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.4102457761764526,
      "learning_rate": 3.773816271800959e-05,
      "loss": 2.1087,
      "step": 193220
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.482192039489746,
      "learning_rate": 3.773617654949823e-05,
      "loss": 1.886,
      "step": 193221
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.5310096740722656,
      "learning_rate": 3.7734190429746606e-05,
      "loss": 1.7572,
      "step": 193222
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.4158458709716797,
      "learning_rate": 3.77322043587552e-05,
      "loss": 2.1148,
      "step": 193223
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.3378968238830566,
      "learning_rate": 3.773021833652427e-05,
      "loss": 2.0362,
      "step": 193224
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.4978033304214478,
      "learning_rate": 3.772823236305436e-05,
      "loss": 2.0992,
      "step": 193225
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.445474624633789,
      "learning_rate": 3.7726246438345605e-05,
      "loss": 2.1367,
      "step": 193226
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.4117462635040283,
      "learning_rate": 3.7724260562398555e-05,
      "loss": 2.1318,
      "step": 193227
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.5945119857788086,
      "learning_rate": 3.772227473521343e-05,
      "loss": 1.7737,
      "step": 193228
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.4365243911743164,
      "learning_rate": 3.77202889567908e-05,
      "loss": 1.9874,
      "step": 193229
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.2881592512130737,
      "learning_rate": 3.7718303227130796e-05,
      "loss": 1.9976,
      "step": 193230
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.6038309335708618,
      "learning_rate": 3.771631754623409e-05,
      "loss": 1.8406,
      "step": 193231
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.518458604812622,
      "learning_rate": 3.77143319141007e-05,
      "loss": 1.7826,
      "step": 193232
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.5150840282440186,
      "learning_rate": 3.771234633073128e-05,
      "loss": 2.2111,
      "step": 193233
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.726556420326233,
      "learning_rate": 3.7710360796125984e-05,
      "loss": 1.8242,
      "step": 193234
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.2296956777572632,
      "learning_rate": 3.7708375310285344e-05,
      "loss": 2.0482,
      "step": 193235
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.3226332664489746,
      "learning_rate": 3.770638987320963e-05,
      "loss": 2.0716,
      "step": 193236
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.3092175722122192,
      "learning_rate": 3.770440448489931e-05,
      "loss": 2.0286,
      "step": 193237
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.4345329999923706,
      "learning_rate": 3.770241914535468e-05,
      "loss": 2.0234,
      "step": 193238
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.5321604013442993,
      "learning_rate": 3.770043385457611e-05,
      "loss": 1.9493,
      "step": 193239
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.3534760475158691,
      "learning_rate": 3.7698448612563924e-05,
      "loss": 1.9939,
      "step": 193240
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.2767640352249146,
      "learning_rate": 3.769646341931863e-05,
      "loss": 1.8803,
      "step": 193241
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.2903010845184326,
      "learning_rate": 3.7694478274840436e-05,
      "loss": 1.974,
      "step": 193242
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.4308158159255981,
      "learning_rate": 3.7692493179129856e-05,
      "loss": 1.8763,
      "step": 193243
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.304093360900879,
      "learning_rate": 3.7690508132187206e-05,
      "loss": 1.8168,
      "step": 193244
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.3672704696655273,
      "learning_rate": 3.7688523134012836e-05,
      "loss": 1.8285,
      "step": 193245
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.3132781982421875,
      "learning_rate": 3.768653818460703e-05,
      "loss": 2.3157,
      "step": 193246
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.3945342302322388,
      "learning_rate": 3.768455328397034e-05,
      "loss": 2.1816,
      "step": 193247
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.5265905857086182,
      "learning_rate": 3.768256843210298e-05,
      "loss": 1.8815,
      "step": 193248
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.333912968635559,
      "learning_rate": 3.768058362900543e-05,
      "loss": 1.9848,
      "step": 193249
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.3653370141983032,
      "learning_rate": 3.767859887467801e-05,
      "loss": 2.0425,
      "step": 193250
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.431857705116272,
      "learning_rate": 3.767661416912111e-05,
      "loss": 1.9428,
      "step": 193251
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.3355355262756348,
      "learning_rate": 3.7674629512335e-05,
      "loss": 1.8914,
      "step": 193252
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.35954749584198,
      "learning_rate": 3.76726449043202e-05,
      "loss": 2.1528,
      "step": 193253
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.5613484382629395,
      "learning_rate": 3.767066034507696e-05,
      "loss": 1.8947,
      "step": 193254
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.3935694694519043,
      "learning_rate": 3.7668675834605746e-05,
      "loss": 2.0912,
      "step": 193255
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.5160067081451416,
      "learning_rate": 3.766669137290684e-05,
      "loss": 2.0787,
      "step": 193256
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.4937241077423096,
      "learning_rate": 3.766470695998069e-05,
      "loss": 1.9822,
      "step": 193257
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.412893533706665,
      "learning_rate": 3.766272259582757e-05,
      "loss": 1.8343,
      "step": 193258
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.4458014965057373,
      "learning_rate": 3.766073828044792e-05,
      "loss": 2.1289,
      "step": 193259
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.50443434715271,
      "learning_rate": 3.7658754013842064e-05,
      "loss": 2.0293,
      "step": 193260
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.3379799127578735,
      "learning_rate": 3.765676979601046e-05,
      "loss": 2.0637,
      "step": 193261
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.3988397121429443,
      "learning_rate": 3.76547856269534e-05,
      "loss": 2.0609,
      "step": 193262
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.365493893623352,
      "learning_rate": 3.765280150667126e-05,
      "loss": 1.8604,
      "step": 193263
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.5269367694854736,
      "learning_rate": 3.765081743516437e-05,
      "loss": 1.9357,
      "step": 193264
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.5550745725631714,
      "learning_rate": 3.76488334124332e-05,
      "loss": 1.8145,
      "step": 193265
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.4836907386779785,
      "learning_rate": 3.764684943847801e-05,
      "loss": 2.0375,
      "step": 193266
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.398676872253418,
      "learning_rate": 3.764486551329927e-05,
      "loss": 2.1247,
      "step": 193267
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.5056060552597046,
      "learning_rate": 3.7642881636897314e-05,
      "loss": 2.2304,
      "step": 193268
    },
    {
      "epoch": 2.52,
      "grad_norm": 2.3982670307159424,
      "learning_rate": 3.764089780927251e-05,
      "loss": 1.9551,
      "step": 193269
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.5542081594467163,
      "learning_rate": 3.763891403042511e-05,
      "loss": 1.9912,
      "step": 193270
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.2591495513916016,
      "learning_rate": 3.763693030035571e-05,
      "loss": 1.8564,
      "step": 193271
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.2992063760757446,
      "learning_rate": 3.7634946619064445e-05,
      "loss": 2.0399,
      "step": 193272
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.5596977472305298,
      "learning_rate": 3.76329629865519e-05,
      "loss": 2.2067,
      "step": 193273
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.4895380735397339,
      "learning_rate": 3.76309794028183e-05,
      "loss": 1.8677,
      "step": 193274
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.6233866214752197,
      "learning_rate": 3.762899586786408e-05,
      "loss": 1.931,
      "step": 193275
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.4031988382339478,
      "learning_rate": 3.76270123816895e-05,
      "loss": 2.1223,
      "step": 193276
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.4729593992233276,
      "learning_rate": 3.7625028944295074e-05,
      "loss": 1.7992,
      "step": 193277
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.4228882789611816,
      "learning_rate": 3.762304555568106e-05,
      "loss": 2.1439,
      "step": 193278
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.424617886543274,
      "learning_rate": 3.762106221584793e-05,
      "loss": 1.9847,
      "step": 193279
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.3396309614181519,
      "learning_rate": 3.7619078924796007e-05,
      "loss": 2.1308,
      "step": 193280
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.3729218244552612,
      "learning_rate": 3.76170956825256e-05,
      "loss": 1.9436,
      "step": 193281
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.5413693189620972,
      "learning_rate": 3.761511248903717e-05,
      "loss": 1.9009,
      "step": 193282
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.515432596206665,
      "learning_rate": 3.7613129344331016e-05,
      "loss": 2.3031,
      "step": 193283
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.4793981313705444,
      "learning_rate": 3.761114624840751e-05,
      "loss": 1.9518,
      "step": 193284
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.4518122673034668,
      "learning_rate": 3.7609163201267115e-05,
      "loss": 1.8425,
      "step": 193285
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.464916467666626,
      "learning_rate": 3.760718020291009e-05,
      "loss": 1.9327,
      "step": 193286
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.333779215812683,
      "learning_rate": 3.7605197253336816e-05,
      "loss": 1.8965,
      "step": 193287
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.4803515672683716,
      "learning_rate": 3.760321435254775e-05,
      "loss": 2.0828,
      "step": 193288
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.631225347518921,
      "learning_rate": 3.7601231500543095e-05,
      "loss": 1.7905,
      "step": 193289
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.300992727279663,
      "learning_rate": 3.759924869732345e-05,
      "loss": 1.8295,
      "step": 193290
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.5975112915039062,
      "learning_rate": 3.759726594288901e-05,
      "loss": 2.1886,
      "step": 193291
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.4521363973617554,
      "learning_rate": 3.7595283237240216e-05,
      "loss": 1.8693,
      "step": 193292
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.4543514251708984,
      "learning_rate": 3.759330058037733e-05,
      "loss": 2.0363,
      "step": 193293
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.3612149953842163,
      "learning_rate": 3.759131797230088e-05,
      "loss": 2.0116,
      "step": 193294
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.3921924829483032,
      "learning_rate": 3.758933541301108e-05,
      "loss": 1.8769,
      "step": 193295
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.4617944955825806,
      "learning_rate": 3.758735290250845e-05,
      "loss": 1.7383,
      "step": 193296
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.4343819618225098,
      "learning_rate": 3.7585370440793295e-05,
      "loss": 1.9434,
      "step": 193297
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.3608452081680298,
      "learning_rate": 3.7583388027865945e-05,
      "loss": 1.8395,
      "step": 193298
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.6063135862350464,
      "learning_rate": 3.758140566372677e-05,
      "loss": 1.8394,
      "step": 193299
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.3980317115783691,
      "learning_rate": 3.7579423348376206e-05,
      "loss": 1.965,
      "step": 193300
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.442476511001587,
      "learning_rate": 3.757744108181451e-05,
      "loss": 1.9397,
      "step": 193301
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.3875936269760132,
      "learning_rate": 3.757545886404219e-05,
      "loss": 1.8749,
      "step": 193302
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.436067819595337,
      "learning_rate": 3.75734766950595e-05,
      "loss": 1.9626,
      "step": 193303
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.395334005355835,
      "learning_rate": 3.757149457486689e-05,
      "loss": 2.2334,
      "step": 193304
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.4510002136230469,
      "learning_rate": 3.7569512503464715e-05,
      "loss": 1.7797,
      "step": 193305
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.426984429359436,
      "learning_rate": 3.756753048085331e-05,
      "loss": 1.8908,
      "step": 193306
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.5702095031738281,
      "learning_rate": 3.756554850703298e-05,
      "loss": 2.4582,
      "step": 193307
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.4796499013900757,
      "learning_rate": 3.7563566582004254e-05,
      "loss": 1.8375,
      "step": 193308
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.4154146909713745,
      "learning_rate": 3.7561584705767366e-05,
      "loss": 2.1367,
      "step": 193309
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.3509403467178345,
      "learning_rate": 3.755960287832278e-05,
      "loss": 1.871,
      "step": 193310
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.6001389026641846,
      "learning_rate": 3.75576210996708e-05,
      "loss": 1.9341,
      "step": 193311
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.4081120491027832,
      "learning_rate": 3.755563936981182e-05,
      "loss": 1.8913,
      "step": 193312
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.3586786985397339,
      "learning_rate": 3.755365768874614e-05,
      "loss": 2.0189,
      "step": 193313
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.4135055541992188,
      "learning_rate": 3.7551676056474264e-05,
      "loss": 2.0075,
      "step": 193314
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.4341758489608765,
      "learning_rate": 3.754969447299639e-05,
      "loss": 2.0337,
      "step": 193315
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.5443922281265259,
      "learning_rate": 3.754771293831309e-05,
      "loss": 1.924,
      "step": 193316
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.5770561695098877,
      "learning_rate": 3.754573145242459e-05,
      "loss": 2.178,
      "step": 193317
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.3975918292999268,
      "learning_rate": 3.754375001533132e-05,
      "loss": 2.1808,
      "step": 193318
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.4018571376800537,
      "learning_rate": 3.7541768627033554e-05,
      "loss": 1.9568,
      "step": 193319
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.2529700994491577,
      "learning_rate": 3.753978728753175e-05,
      "loss": 1.9936,
      "step": 193320
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.4802974462509155,
      "learning_rate": 3.753780599682625e-05,
      "loss": 2.1204,
      "step": 193321
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.4306312799453735,
      "learning_rate": 3.7535824754917457e-05,
      "loss": 1.9151,
      "step": 193322
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.4268991947174072,
      "learning_rate": 3.753384356180572e-05,
      "loss": 1.8849,
      "step": 193323
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.4658632278442383,
      "learning_rate": 3.753186241749139e-05,
      "loss": 1.776,
      "step": 193324
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.3464789390563965,
      "learning_rate": 3.7529881321974756e-05,
      "loss": 2.0376,
      "step": 193325
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.3880908489227295,
      "learning_rate": 3.752790027525635e-05,
      "loss": 2.0689,
      "step": 193326
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.386352300643921,
      "learning_rate": 3.752591927733641e-05,
      "loss": 1.9221,
      "step": 193327
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.466642141342163,
      "learning_rate": 3.7523938328215407e-05,
      "loss": 2.1514,
      "step": 193328
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.3461834192276,
      "learning_rate": 3.752195742789367e-05,
      "loss": 1.9979,
      "step": 193329
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.5799193382263184,
      "learning_rate": 3.751997657637156e-05,
      "loss": 1.9442,
      "step": 193330
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.4163486957550049,
      "learning_rate": 3.751799577364935e-05,
      "loss": 2.0123,
      "step": 193331
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.5357224941253662,
      "learning_rate": 3.75160150197276e-05,
      "loss": 2.1229,
      "step": 193332
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.3076080083847046,
      "learning_rate": 3.7514034314606475e-05,
      "loss": 2.0554,
      "step": 193333
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.3840118646621704,
      "learning_rate": 3.751205365828652e-05,
      "loss": 2.1029,
      "step": 193334
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.2976562976837158,
      "learning_rate": 3.751007305076806e-05,
      "loss": 2.0569,
      "step": 193335
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.5035954713821411,
      "learning_rate": 3.75080924920514e-05,
      "loss": 2.2114,
      "step": 193336
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.3394819498062134,
      "learning_rate": 3.75061119821369e-05,
      "loss": 2.1472,
      "step": 193337
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.394951343536377,
      "learning_rate": 3.7504131521025027e-05,
      "loss": 2.1202,
      "step": 193338
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.2627465724945068,
      "learning_rate": 3.7502151108716016e-05,
      "loss": 2.1939,
      "step": 193339
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.3538591861724854,
      "learning_rate": 3.7500170745210366e-05,
      "loss": 1.9239,
      "step": 193340
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.4960867166519165,
      "learning_rate": 3.749819043050841e-05,
      "loss": 2.0493,
      "step": 193341
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.2463531494140625,
      "learning_rate": 3.749621016461049e-05,
      "loss": 2.0243,
      "step": 193342
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.3361375331878662,
      "learning_rate": 3.749422994751688e-05,
      "loss": 2.0159,
      "step": 193343
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.4176971912384033,
      "learning_rate": 3.749224977922818e-05,
      "loss": 2.0097,
      "step": 193344
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.335024118423462,
      "learning_rate": 3.74902696597445e-05,
      "loss": 1.8776,
      "step": 193345
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.3382917642593384,
      "learning_rate": 3.748828958906644e-05,
      "loss": 2.0589,
      "step": 193346
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.3359025716781616,
      "learning_rate": 3.748630956719425e-05,
      "loss": 2.1699,
      "step": 193347
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.3866891860961914,
      "learning_rate": 3.7484329594128285e-05,
      "loss": 1.9789,
      "step": 193348
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.6069016456604004,
      "learning_rate": 3.7482349669868904e-05,
      "loss": 2.1294,
      "step": 193349
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.4683537483215332,
      "learning_rate": 3.748036979441655e-05,
      "loss": 2.0825,
      "step": 193350
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.5091257095336914,
      "learning_rate": 3.7478389967771524e-05,
      "loss": 2.0369,
      "step": 193351
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.413288950920105,
      "learning_rate": 3.747641018993426e-05,
      "loss": 1.7651,
      "step": 193352
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.6498286724090576,
      "learning_rate": 3.747443046090508e-05,
      "loss": 1.9352,
      "step": 193353
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.3878393173217773,
      "learning_rate": 3.7472450780684396e-05,
      "loss": 1.9331,
      "step": 193354
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.4230417013168335,
      "learning_rate": 3.747047114927243e-05,
      "loss": 2.0056,
      "step": 193355
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.4103697538375854,
      "learning_rate": 3.7468491566669755e-05,
      "loss": 2.2079,
      "step": 193356
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.3187689781188965,
      "learning_rate": 3.746651203287654e-05,
      "loss": 1.8407,
      "step": 193357
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.372042179107666,
      "learning_rate": 3.746453254789334e-05,
      "loss": 1.8768,
      "step": 193358
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.445298671722412,
      "learning_rate": 3.7462553111720475e-05,
      "loss": 1.9973,
      "step": 193359
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.392791509628296,
      "learning_rate": 3.746057372435822e-05,
      "loss": 1.9998,
      "step": 193360
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.6357067823410034,
      "learning_rate": 3.7458594385807e-05,
      "loss": 2.0806,
      "step": 193361
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.3603423833847046,
      "learning_rate": 3.745661509606719e-05,
      "loss": 2.082,
      "step": 193362
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.5267152786254883,
      "learning_rate": 3.74546358551391e-05,
      "loss": 1.9118,
      "step": 193363
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.5215485095977783,
      "learning_rate": 3.745265666302324e-05,
      "loss": 1.9005,
      "step": 193364
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.3288010358810425,
      "learning_rate": 3.745067751971986e-05,
      "loss": 1.9211,
      "step": 193365
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.4799859523773193,
      "learning_rate": 3.7448698425229304e-05,
      "loss": 1.8418,
      "step": 193366
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.4363911151885986,
      "learning_rate": 3.7446719379552035e-05,
      "loss": 1.9621,
      "step": 193367
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.5008636713027954,
      "learning_rate": 3.744474038268839e-05,
      "loss": 2.0205,
      "step": 193368
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.2940301895141602,
      "learning_rate": 3.744276143463866e-05,
      "loss": 1.9985,
      "step": 193369
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.4534255266189575,
      "learning_rate": 3.7440782535403346e-05,
      "loss": 1.9867,
      "step": 193370
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.5948573350906372,
      "learning_rate": 3.743880368498269e-05,
      "loss": 2.2357,
      "step": 193371
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.3545775413513184,
      "learning_rate": 3.7436824883377145e-05,
      "loss": 1.9175,
      "step": 193372
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.5314013957977295,
      "learning_rate": 3.743484613058709e-05,
      "loss": 2.1408,
      "step": 193373
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.3047144412994385,
      "learning_rate": 3.743286742661282e-05,
      "loss": 1.9112,
      "step": 193374
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.3755548000335693,
      "learning_rate": 3.7430888771454706e-05,
      "loss": 2.2766,
      "step": 193375
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.4040559530258179,
      "learning_rate": 3.74289101651132e-05,
      "loss": 1.9158,
      "step": 193376
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.3655756711959839,
      "learning_rate": 3.742693160758855e-05,
      "loss": 2.1055,
      "step": 193377
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.289948582649231,
      "learning_rate": 3.742495309888125e-05,
      "loss": 2.0507,
      "step": 193378
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.3232704401016235,
      "learning_rate": 3.74229746389916e-05,
      "loss": 1.8123,
      "step": 193379
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.3191184997558594,
      "learning_rate": 3.742099622791993e-05,
      "loss": 1.6999,
      "step": 193380
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.3428925275802612,
      "learning_rate": 3.741901786566671e-05,
      "loss": 2.2454,
      "step": 193381
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.4676063060760498,
      "learning_rate": 3.741703955223224e-05,
      "loss": 2.2952,
      "step": 193382
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.570093035697937,
      "learning_rate": 3.7415061287616854e-05,
      "loss": 1.8725,
      "step": 193383
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.5448198318481445,
      "learning_rate": 3.741308307182102e-05,
      "loss": 2.0169,
      "step": 193384
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.4541765451431274,
      "learning_rate": 3.741110490484502e-05,
      "loss": 2.228,
      "step": 193385
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.3978147506713867,
      "learning_rate": 3.7409126786689214e-05,
      "loss": 2.0744,
      "step": 193386
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.4994471073150635,
      "learning_rate": 3.740714871735409e-05,
      "loss": 1.8863,
      "step": 193387
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.5060786008834839,
      "learning_rate": 3.740517069683987e-05,
      "loss": 1.9806,
      "step": 193388
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.4512755870819092,
      "learning_rate": 3.740319272514703e-05,
      "loss": 2.2605,
      "step": 193389
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.3622061014175415,
      "learning_rate": 3.740121480227588e-05,
      "loss": 1.9143,
      "step": 193390
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.5286389589309692,
      "learning_rate": 3.739923692822684e-05,
      "loss": 2.1029,
      "step": 193391
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.4137316942214966,
      "learning_rate": 3.7397259103000145e-05,
      "loss": 1.8082,
      "step": 193392
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.422699213027954,
      "learning_rate": 3.739528132659633e-05,
      "loss": 1.7422,
      "step": 193393
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.6364545822143555,
      "learning_rate": 3.7393303599015623e-05,
      "loss": 2.1052,
      "step": 193394
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.3783056735992432,
      "learning_rate": 3.739132592025853e-05,
      "loss": 2.0072,
      "step": 193395
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.3029543161392212,
      "learning_rate": 3.738934829032535e-05,
      "loss": 1.9584,
      "step": 193396
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.403130292892456,
      "learning_rate": 3.7387370709216416e-05,
      "loss": 2.1483,
      "step": 193397
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.4231771230697632,
      "learning_rate": 3.738539317693209e-05,
      "loss": 2.0552,
      "step": 193398
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.5127447843551636,
      "learning_rate": 3.738341569347284e-05,
      "loss": 1.9562,
      "step": 193399
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.4545384645462036,
      "learning_rate": 3.73814382588389e-05,
      "loss": 2.0209,
      "step": 193400
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.5341737270355225,
      "learning_rate": 3.7379460873030774e-05,
      "loss": 1.929,
      "step": 193401
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.3861757516860962,
      "learning_rate": 3.737748353604876e-05,
      "loss": 1.9003,
      "step": 193402
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.4827412366867065,
      "learning_rate": 3.7375506247893216e-05,
      "loss": 2.235,
      "step": 193403
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.3638606071472168,
      "learning_rate": 3.737352900856448e-05,
      "loss": 1.9827,
      "step": 193404
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.2842950820922852,
      "learning_rate": 3.737155181806303e-05,
      "loss": 1.9318,
      "step": 193405
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.3822753429412842,
      "learning_rate": 3.736957467638908e-05,
      "loss": 1.7909,
      "step": 193406
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.3007866144180298,
      "learning_rate": 3.736759758354317e-05,
      "loss": 1.8635,
      "step": 193407
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.4559522867202759,
      "learning_rate": 3.736562053952558e-05,
      "loss": 2.034,
      "step": 193408
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.5043102502822876,
      "learning_rate": 3.7363643544336684e-05,
      "loss": 2.0677,
      "step": 193409
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.461583137512207,
      "learning_rate": 3.736166659797677e-05,
      "loss": 2.0151,
      "step": 193410
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.3985506296157837,
      "learning_rate": 3.7359689700446324e-05,
      "loss": 1.904,
      "step": 193411
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.4944311380386353,
      "learning_rate": 3.7357712851745624e-05,
      "loss": 1.8759,
      "step": 193412
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.6204030513763428,
      "learning_rate": 3.735573605187513e-05,
      "loss": 1.8685,
      "step": 193413
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.7765051126480103,
      "learning_rate": 3.7353759300835206e-05,
      "loss": 2.1421,
      "step": 193414
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.3850173950195312,
      "learning_rate": 3.7351782598626124e-05,
      "loss": 1.7508,
      "step": 193415
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.5327891111373901,
      "learning_rate": 3.734980594524828e-05,
      "loss": 1.8467,
      "step": 193416
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.4215859174728394,
      "learning_rate": 3.7347829340702104e-05,
      "loss": 2.1122,
      "step": 193417
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.4915648698806763,
      "learning_rate": 3.734585278498787e-05,
      "loss": 1.9659,
      "step": 193418
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.4892933368682861,
      "learning_rate": 3.734387627810608e-05,
      "loss": 1.9512,
      "step": 193419
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.4088083505630493,
      "learning_rate": 3.734189982005699e-05,
      "loss": 2.1438,
      "step": 193420
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.3661233186721802,
      "learning_rate": 3.7339923410841e-05,
      "loss": 2.1203,
      "step": 193421
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.3827306032180786,
      "learning_rate": 3.733794705045845e-05,
      "loss": 2.035,
      "step": 193422
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.4109939336776733,
      "learning_rate": 3.733597073890977e-05,
      "loss": 2.0277,
      "step": 193423
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.4623327255249023,
      "learning_rate": 3.733399447619523e-05,
      "loss": 1.9232,
      "step": 193424
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.3076856136322021,
      "learning_rate": 3.733201826231532e-05,
      "loss": 1.9091,
      "step": 193425
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.4747709035873413,
      "learning_rate": 3.7330042097270355e-05,
      "loss": 1.9894,
      "step": 193426
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.3059545755386353,
      "learning_rate": 3.732806598106068e-05,
      "loss": 1.8288,
      "step": 193427
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.5656722784042358,
      "learning_rate": 3.732608991368662e-05,
      "loss": 1.7608,
      "step": 193428
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.3928707838058472,
      "learning_rate": 3.732411389514865e-05,
      "loss": 1.7866,
      "step": 193429
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.428989052772522,
      "learning_rate": 3.732213792544702e-05,
      "loss": 1.8171,
      "step": 193430
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.4052751064300537,
      "learning_rate": 3.7320162004582264e-05,
      "loss": 1.9898,
      "step": 193431
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.572653889656067,
      "learning_rate": 3.7318186132554605e-05,
      "loss": 1.942,
      "step": 193432
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.5063655376434326,
      "learning_rate": 3.731621030936448e-05,
      "loss": 1.9317,
      "step": 193433
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.4728777408599854,
      "learning_rate": 3.7314234535012155e-05,
      "loss": 1.9666,
      "step": 193434
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.5072128772735596,
      "learning_rate": 3.731225880949816e-05,
      "loss": 1.9512,
      "step": 193435
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.3620308637619019,
      "learning_rate": 3.731028313282267e-05,
      "loss": 2.0103,
      "step": 193436
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.5613677501678467,
      "learning_rate": 3.7308307504986246e-05,
      "loss": 1.8281,
      "step": 193437
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.4587361812591553,
      "learning_rate": 3.730633192598912e-05,
      "loss": 2.0049,
      "step": 193438
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.5868034362792969,
      "learning_rate": 3.730435639583179e-05,
      "loss": 1.8827,
      "step": 193439
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.3426674604415894,
      "learning_rate": 3.7302380914514465e-05,
      "loss": 1.9368,
      "step": 193440
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.3547163009643555,
      "learning_rate": 3.73004054820376e-05,
      "loss": 2.1576,
      "step": 193441
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.4394121170043945,
      "learning_rate": 3.72984300984015e-05,
      "loss": 2.0035,
      "step": 193442
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.5054818391799927,
      "learning_rate": 3.7296454763606665e-05,
      "loss": 1.9704,
      "step": 193443
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.4242676496505737,
      "learning_rate": 3.729447947765329e-05,
      "loss": 2.1489,
      "step": 193444
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.5656347274780273,
      "learning_rate": 3.7292504240541955e-05,
      "loss": 1.9103,
      "step": 193445
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.4549014568328857,
      "learning_rate": 3.729052905227278e-05,
      "loss": 2.0418,
      "step": 193446
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.4569268226623535,
      "learning_rate": 3.728855391284633e-05,
      "loss": 2.0752,
      "step": 193447
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.3327261209487915,
      "learning_rate": 3.728657882226281e-05,
      "loss": 1.8582,
      "step": 193448
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.4061232805252075,
      "learning_rate": 3.728460378052276e-05,
      "loss": 1.989,
      "step": 193449
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.4614121913909912,
      "learning_rate": 3.72826287876264e-05,
      "loss": 1.9271,
      "step": 193450
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.4849351644515991,
      "learning_rate": 3.72806538435742e-05,
      "loss": 1.9514,
      "step": 193451
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.5179156064987183,
      "learning_rate": 3.72786789483665e-05,
      "loss": 1.7051,
      "step": 193452
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.3557759523391724,
      "learning_rate": 3.7276704102003664e-05,
      "loss": 2.0136,
      "step": 193453
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.3596845865249634,
      "learning_rate": 3.7274729304485986e-05,
      "loss": 1.8449,
      "step": 193454
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.4015147686004639,
      "learning_rate": 3.7272754555813934e-05,
      "loss": 1.9981,
      "step": 193455
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.2434558868408203,
      "learning_rate": 3.727077985598778e-05,
      "loss": 2.2043,
      "step": 193456
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.4857196807861328,
      "learning_rate": 3.726880520500801e-05,
      "loss": 2.2562,
      "step": 193457
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.501561164855957,
      "learning_rate": 3.7266830602874944e-05,
      "loss": 2.1759,
      "step": 193458
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.3296982049942017,
      "learning_rate": 3.726485604958893e-05,
      "loss": 2.0957,
      "step": 193459
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.393214225769043,
      "learning_rate": 3.726288154515025e-05,
      "loss": 2.079,
      "step": 193460
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.3458904027938843,
      "learning_rate": 3.726090708955946e-05,
      "loss": 1.7421,
      "step": 193461
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.3105627298355103,
      "learning_rate": 3.725893268281672e-05,
      "loss": 1.8712,
      "step": 193462
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.4316319227218628,
      "learning_rate": 3.7256958324922615e-05,
      "loss": 2.0795,
      "step": 193463
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.4138580560684204,
      "learning_rate": 3.7254984015877366e-05,
      "loss": 2.0696,
      "step": 193464
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.4435808658599854,
      "learning_rate": 3.725300975568134e-05,
      "loss": 2.1084,
      "step": 193465
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.3183084726333618,
      "learning_rate": 3.725103554433497e-05,
      "loss": 1.9786,
      "step": 193466
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.3580029010772705,
      "learning_rate": 3.724906138183863e-05,
      "loss": 1.7646,
      "step": 193467
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.3190776109695435,
      "learning_rate": 3.724708726819254e-05,
      "loss": 2.0295,
      "step": 193468
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.4768099784851074,
      "learning_rate": 3.724511320339728e-05,
      "loss": 1.9501,
      "step": 193469
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.3616348505020142,
      "learning_rate": 3.724313918745311e-05,
      "loss": 1.8504,
      "step": 193470
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.4925659894943237,
      "learning_rate": 3.724116522036033e-05,
      "loss": 2.1464,
      "step": 193471
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.3669506311416626,
      "learning_rate": 3.723919130211943e-05,
      "loss": 1.9536,
      "step": 193472
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.2787158489227295,
      "learning_rate": 3.7237217432730725e-05,
      "loss": 2.0097,
      "step": 193473
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.3601112365722656,
      "learning_rate": 3.723524361219451e-05,
      "loss": 1.8177,
      "step": 193474
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.296955943107605,
      "learning_rate": 3.723326984051131e-05,
      "loss": 2.1922,
      "step": 193475
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.402453064918518,
      "learning_rate": 3.723129611768141e-05,
      "loss": 1.8534,
      "step": 193476
    },
    {
      "epoch": 2.52,
      "grad_norm": 2.1953125,
      "learning_rate": 3.722932244370509e-05,
      "loss": 2.0467,
      "step": 193477
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.33412766456604,
      "learning_rate": 3.722734881858286e-05,
      "loss": 1.7067,
      "step": 193478
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.3169797658920288,
      "learning_rate": 3.722537524231498e-05,
      "loss": 1.9778,
      "step": 193479
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.608221173286438,
      "learning_rate": 3.722340171490189e-05,
      "loss": 2.1769,
      "step": 193480
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.591015338897705,
      "learning_rate": 3.722142823634395e-05,
      "loss": 2.1301,
      "step": 193481
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.2706252336502075,
      "learning_rate": 3.7219454806641537e-05,
      "loss": 2.0111,
      "step": 193482
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.3712559938430786,
      "learning_rate": 3.7217481425794906e-05,
      "loss": 2.0586,
      "step": 193483
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.4098581075668335,
      "learning_rate": 3.721550809380456e-05,
      "loss": 1.9943,
      "step": 193484
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.3815875053405762,
      "learning_rate": 3.721353481067073e-05,
      "loss": 2.0725,
      "step": 193485
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.3856748342514038,
      "learning_rate": 3.7211561576393955e-05,
      "loss": 2.0377,
      "step": 193486
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.5737885236740112,
      "learning_rate": 3.7209588390974495e-05,
      "loss": 2.0357,
      "step": 193487
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.452290654182434,
      "learning_rate": 3.7207615254412756e-05,
      "loss": 2.079,
      "step": 193488
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.5362443923950195,
      "learning_rate": 3.7205642166709e-05,
      "loss": 1.946,
      "step": 193489
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.449660062789917,
      "learning_rate": 3.720366912786377e-05,
      "loss": 2.0143,
      "step": 193490
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.2859984636306763,
      "learning_rate": 3.720169613787721e-05,
      "loss": 1.8886,
      "step": 193491
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.4540605545043945,
      "learning_rate": 3.719972319674994e-05,
      "loss": 2.0516,
      "step": 193492
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.4773837327957153,
      "learning_rate": 3.719775030448219e-05,
      "loss": 2.1029,
      "step": 193493
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.292863130569458,
      "learning_rate": 3.719577746107432e-05,
      "loss": 1.9947,
      "step": 193494
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.3938705921173096,
      "learning_rate": 3.719380466652666e-05,
      "loss": 1.8396,
      "step": 193495
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.3272252082824707,
      "learning_rate": 3.719183192083969e-05,
      "loss": 1.8148,
      "step": 193496
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.5008426904678345,
      "learning_rate": 3.718985922401366e-05,
      "loss": 2.0357,
      "step": 193497
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.348387360572815,
      "learning_rate": 3.718788657604909e-05,
      "loss": 1.8967,
      "step": 193498
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.3057509660720825,
      "learning_rate": 3.718591397694623e-05,
      "loss": 1.9079,
      "step": 193499
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.283676028251648,
      "learning_rate": 3.7183941426705445e-05,
      "loss": 1.9854,
      "step": 193500
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.4833080768585205,
      "learning_rate": 3.718196892532708e-05,
      "loss": 1.9426,
      "step": 193501
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.3812180757522583,
      "learning_rate": 3.7179996472811625e-05,
      "loss": 2.0583,
      "step": 193502
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.327150583267212,
      "learning_rate": 3.7178024069159286e-05,
      "loss": 1.9882,
      "step": 193503
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.2900199890136719,
      "learning_rate": 3.717605171437059e-05,
      "loss": 2.1079,
      "step": 193504
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.3859165906906128,
      "learning_rate": 3.7174079408445775e-05,
      "loss": 1.851,
      "step": 193505
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.2614465951919556,
      "learning_rate": 3.7172107151385376e-05,
      "loss": 1.8879,
      "step": 193506
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.4123833179473877,
      "learning_rate": 3.717013494318949e-05,
      "loss": 2.0905,
      "step": 193507
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.3818789720535278,
      "learning_rate": 3.716816278385875e-05,
      "loss": 2.0289,
      "step": 193508
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.5327914953231812,
      "learning_rate": 3.716619067339332e-05,
      "loss": 2.1386,
      "step": 193509
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.5976450443267822,
      "learning_rate": 3.716421861179369e-05,
      "loss": 1.978,
      "step": 193510
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.2656114101409912,
      "learning_rate": 3.716224659906018e-05,
      "loss": 1.9365,
      "step": 193511
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.4545354843139648,
      "learning_rate": 3.7160274635193286e-05,
      "loss": 2.2183,
      "step": 193512
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.491688847541809,
      "learning_rate": 3.71583027201931e-05,
      "loss": 1.7566,
      "step": 193513
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.412243127822876,
      "learning_rate": 3.7156330854060255e-05,
      "loss": 2.1474,
      "step": 193514
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.4296172857284546,
      "learning_rate": 3.715435903679492e-05,
      "loss": 1.9962,
      "step": 193515
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.482506275177002,
      "learning_rate": 3.7152387268397634e-05,
      "loss": 2.0234,
      "step": 193516
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.4163497686386108,
      "learning_rate": 3.7150415548868585e-05,
      "loss": 1.8192,
      "step": 193517
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.490344762802124,
      "learning_rate": 3.714844387820838e-05,
      "loss": 2.1022,
      "step": 193518
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.371622085571289,
      "learning_rate": 3.714647225641711e-05,
      "loss": 1.9467,
      "step": 193519
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.2665956020355225,
      "learning_rate": 3.7144500683495324e-05,
      "loss": 1.6041,
      "step": 193520
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.4831089973449707,
      "learning_rate": 3.714252915944331e-05,
      "loss": 2.1434,
      "step": 193521
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.5262408256530762,
      "learning_rate": 3.714055768426151e-05,
      "loss": 1.9812,
      "step": 193522
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.463675618171692,
      "learning_rate": 3.7138586257950175e-05,
      "loss": 2.1846,
      "step": 193523
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.3655871152877808,
      "learning_rate": 3.713661488050985e-05,
      "loss": 1.9397,
      "step": 193524
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.3319222927093506,
      "learning_rate": 3.713464355194067e-05,
      "loss": 2.0606,
      "step": 193525
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.3195768594741821,
      "learning_rate": 3.7132672272243194e-05,
      "loss": 1.9715,
      "step": 193526
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.4164502620697021,
      "learning_rate": 3.713070104141766e-05,
      "loss": 2.1429,
      "step": 193527
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.5041193962097168,
      "learning_rate": 3.7128729859464553e-05,
      "loss": 2.0967,
      "step": 193528
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.382568359375,
      "learning_rate": 3.71267587263841e-05,
      "loss": 1.9396,
      "step": 193529
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.2106692790985107,
      "learning_rate": 3.7124787642176875e-05,
      "loss": 1.8218,
      "step": 193530
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.4499592781066895,
      "learning_rate": 3.7122816606843e-05,
      "loss": 1.9131,
      "step": 193531
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.3785916566848755,
      "learning_rate": 3.7120845620382985e-05,
      "loss": 1.9466,
      "step": 193532
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.5298709869384766,
      "learning_rate": 3.711887468279715e-05,
      "loss": 1.7599,
      "step": 193533
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.507838249206543,
      "learning_rate": 3.711690379408592e-05,
      "loss": 2.0377,
      "step": 193534
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.359710693359375,
      "learning_rate": 3.711493295424955e-05,
      "loss": 2.0165,
      "step": 193535
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.423519492149353,
      "learning_rate": 3.71129621632886e-05,
      "loss": 1.9198,
      "step": 193536
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.4534735679626465,
      "learning_rate": 3.7110991421203206e-05,
      "loss": 1.8229,
      "step": 193537
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.4309414625167847,
      "learning_rate": 3.7109020727993886e-05,
      "loss": 1.982,
      "step": 193538
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.3661624193191528,
      "learning_rate": 3.710705008366091e-05,
      "loss": 1.779,
      "step": 193539
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.3081122636795044,
      "learning_rate": 3.710507948820476e-05,
      "loss": 1.8558,
      "step": 193540
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.3932923078536987,
      "learning_rate": 3.710310894162568e-05,
      "loss": 1.7988,
      "step": 193541
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.531516432762146,
      "learning_rate": 3.710113844392414e-05,
      "loss": 1.9639,
      "step": 193542
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.4416645765304565,
      "learning_rate": 3.709916799510047e-05,
      "loss": 2.1068,
      "step": 193543
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.4188225269317627,
      "learning_rate": 3.709719759515503e-05,
      "loss": 1.9034,
      "step": 193544
    },
    {
      "epoch": 2.52,
      "grad_norm": 2.119196891784668,
      "learning_rate": 3.709522724408812e-05,
      "loss": 1.9564,
      "step": 193545
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.49312162399292,
      "learning_rate": 3.7093256941900216e-05,
      "loss": 1.9271,
      "step": 193546
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.4353387355804443,
      "learning_rate": 3.709128668859158e-05,
      "loss": 2.1254,
      "step": 193547
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.3789578676223755,
      "learning_rate": 3.708931648416271e-05,
      "loss": 1.9877,
      "step": 193548
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.4121084213256836,
      "learning_rate": 3.708734632861391e-05,
      "loss": 2.1073,
      "step": 193549
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.2829854488372803,
      "learning_rate": 3.708537622194544e-05,
      "loss": 1.9302,
      "step": 193550
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.5191209316253662,
      "learning_rate": 3.7083406164157866e-05,
      "loss": 2.0096,
      "step": 193551
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.393192172050476,
      "learning_rate": 3.7081436155251434e-05,
      "loss": 1.7584,
      "step": 193552
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.5363779067993164,
      "learning_rate": 3.707946619522643e-05,
      "loss": 1.8375,
      "step": 193553
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.5520000457763672,
      "learning_rate": 3.707749628408342e-05,
      "loss": 2.1834,
      "step": 193554
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.3220863342285156,
      "learning_rate": 3.7075526421822644e-05,
      "loss": 1.7855,
      "step": 193555
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.4441826343536377,
      "learning_rate": 3.707355660844443e-05,
      "loss": 1.9642,
      "step": 193556
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.3323867321014404,
      "learning_rate": 3.707158684394929e-05,
      "loss": 2.0009,
      "step": 193557
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.416753888130188,
      "learning_rate": 3.706961712833747e-05,
      "loss": 1.9566,
      "step": 193558
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.540212631225586,
      "learning_rate": 3.7067647461609354e-05,
      "loss": 2.0227,
      "step": 193559
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.3571525812149048,
      "learning_rate": 3.706567784376536e-05,
      "loss": 2.2605,
      "step": 193560
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.467033863067627,
      "learning_rate": 3.7063708274805803e-05,
      "loss": 2.0708,
      "step": 193561
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.5296192169189453,
      "learning_rate": 3.7061738754731044e-05,
      "loss": 2.05,
      "step": 193562
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.3292436599731445,
      "learning_rate": 3.705976928354151e-05,
      "loss": 1.7592,
      "step": 193563
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.2673372030258179,
      "learning_rate": 3.7057799861237466e-05,
      "loss": 2.2109,
      "step": 193564
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.3903459310531616,
      "learning_rate": 3.7055830487819394e-05,
      "loss": 2.2431,
      "step": 193565
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.5247360467910767,
      "learning_rate": 3.705386116328761e-05,
      "loss": 2.1229,
      "step": 193566
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.3045393228530884,
      "learning_rate": 3.705189188764249e-05,
      "loss": 2.0763,
      "step": 193567
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.568505048751831,
      "learning_rate": 3.704992266088434e-05,
      "loss": 2.1977,
      "step": 193568
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.3689953088760376,
      "learning_rate": 3.7047953483013606e-05,
      "loss": 1.9441,
      "step": 193569
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.3438427448272705,
      "learning_rate": 3.7045984354030567e-05,
      "loss": 1.9106,
      "step": 193570
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.377830982208252,
      "learning_rate": 3.704401527393569e-05,
      "loss": 2.0976,
      "step": 193571
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.413623571395874,
      "learning_rate": 3.704204624272924e-05,
      "loss": 1.9571,
      "step": 193572
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.5363502502441406,
      "learning_rate": 3.7040077260411784e-05,
      "loss": 2.1168,
      "step": 193573
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.6032317876815796,
      "learning_rate": 3.7038108326983383e-05,
      "loss": 1.886,
      "step": 193574
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.5849816799163818,
      "learning_rate": 3.703613944244465e-05,
      "loss": 2.0433,
      "step": 193575
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.42218017578125,
      "learning_rate": 3.70341706067958e-05,
      "loss": 2.0526,
      "step": 193576
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.3967117071151733,
      "learning_rate": 3.703220182003731e-05,
      "loss": 1.7491,
      "step": 193577
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.3771556615829468,
      "learning_rate": 3.703023308216945e-05,
      "loss": 2.0123,
      "step": 193578
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.6071816682815552,
      "learning_rate": 3.702826439319274e-05,
      "loss": 1.9024,
      "step": 193579
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.6439765691757202,
      "learning_rate": 3.702629575310733e-05,
      "loss": 2.1319,
      "step": 193580
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.4132943153381348,
      "learning_rate": 3.7024327161913734e-05,
      "loss": 1.9037,
      "step": 193581
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.3695118427276611,
      "learning_rate": 3.7022358619612236e-05,
      "loss": 2.0279,
      "step": 193582
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.713287115097046,
      "learning_rate": 3.702039012620332e-05,
      "loss": 2.1411,
      "step": 193583
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.5711616277694702,
      "learning_rate": 3.7018421681687205e-05,
      "loss": 1.9954,
      "step": 193584
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.359264612197876,
      "learning_rate": 3.701645328606444e-05,
      "loss": 1.7956,
      "step": 193585
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.3597856760025024,
      "learning_rate": 3.701448493933516e-05,
      "loss": 1.8468,
      "step": 193586
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.481749176979065,
      "learning_rate": 3.7012516641499944e-05,
      "loss": 2.0478,
      "step": 193587
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.577408790588379,
      "learning_rate": 3.7010548392558983e-05,
      "loss": 1.8553,
      "step": 193588
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.5687205791473389,
      "learning_rate": 3.700858019251277e-05,
      "loss": 2.1328,
      "step": 193589
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.444875717163086,
      "learning_rate": 3.7006612041361586e-05,
      "loss": 2.0682,
      "step": 193590
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.4544113874435425,
      "learning_rate": 3.700464393910595e-05,
      "loss": 2.0822,
      "step": 193591
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.4382987022399902,
      "learning_rate": 3.7002675885746e-05,
      "loss": 2.0106,
      "step": 193592
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.269315481185913,
      "learning_rate": 3.7000707881282276e-05,
      "loss": 1.9301,
      "step": 193593
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.4256774187088013,
      "learning_rate": 3.699873992571504e-05,
      "loss": 1.8406,
      "step": 193594
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.4486143589019775,
      "learning_rate": 3.699677201904475e-05,
      "loss": 1.9866,
      "step": 193595
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.57060968875885,
      "learning_rate": 3.699480416127165e-05,
      "loss": 1.8776,
      "step": 193596
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.3553987741470337,
      "learning_rate": 3.69928363523963e-05,
      "loss": 1.8173,
      "step": 193597
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.409041166305542,
      "learning_rate": 3.6990868592418846e-05,
      "loss": 2.2515,
      "step": 193598
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.4301064014434814,
      "learning_rate": 3.69889008813398e-05,
      "loss": 2.0739,
      "step": 193599
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.5617109537124634,
      "learning_rate": 3.698693321915944e-05,
      "loss": 2.0879,
      "step": 193600
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.321232795715332,
      "learning_rate": 3.6984965605878205e-05,
      "loss": 1.914,
      "step": 193601
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.3058222532272339,
      "learning_rate": 3.6982998041496384e-05,
      "loss": 2.081,
      "step": 193602
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.3046340942382812,
      "learning_rate": 3.698103052601454e-05,
      "loss": 1.7991,
      "step": 193603
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.2751115560531616,
      "learning_rate": 3.697906305943272e-05,
      "loss": 1.7104,
      "step": 193604
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.4732701778411865,
      "learning_rate": 3.697709564175155e-05,
      "loss": 2.3127,
      "step": 193605
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.4478693008422852,
      "learning_rate": 3.69751282729712e-05,
      "loss": 2.0951,
      "step": 193606
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.4909405708312988,
      "learning_rate": 3.697316095309223e-05,
      "loss": 2.0127,
      "step": 193607
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.4988124370574951,
      "learning_rate": 3.697119368211488e-05,
      "loss": 2.3269,
      "step": 193608
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.3761762380599976,
      "learning_rate": 3.6969226460039636e-05,
      "loss": 2.1254,
      "step": 193609
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.389803409576416,
      "learning_rate": 3.696725928686665e-05,
      "loss": 2.0193,
      "step": 193610
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.3304197788238525,
      "learning_rate": 3.696529216259651e-05,
      "loss": 1.825,
      "step": 193611
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.4577057361602783,
      "learning_rate": 3.696332508722938e-05,
      "loss": 1.9377,
      "step": 193612
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.4417287111282349,
      "learning_rate": 3.696135806076583e-05,
      "loss": 1.8874,
      "step": 193613
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.4262888431549072,
      "learning_rate": 3.695939108320607e-05,
      "loss": 1.9557,
      "step": 193614
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.4719592332839966,
      "learning_rate": 3.695742415455062e-05,
      "loss": 2.1148,
      "step": 193615
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.2962501049041748,
      "learning_rate": 3.695545727479965e-05,
      "loss": 1.9313,
      "step": 193616
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.4068572521209717,
      "learning_rate": 3.695349044395366e-05,
      "loss": 2.0577,
      "step": 193617
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.411937952041626,
      "learning_rate": 3.695152366201295e-05,
      "loss": 1.9864,
      "step": 193618
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.5622042417526245,
      "learning_rate": 3.694955692897799e-05,
      "loss": 2.0419,
      "step": 193619
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.3905929327011108,
      "learning_rate": 3.6947590244848966e-05,
      "loss": 2.0645,
      "step": 193620
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.43950355052948,
      "learning_rate": 3.6945623609626496e-05,
      "loss": 1.8913,
      "step": 193621
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.3985854387283325,
      "learning_rate": 3.694365702331067e-05,
      "loss": 2.1659,
      "step": 193622
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.2832911014556885,
      "learning_rate": 3.6941690485902054e-05,
      "loss": 1.9051,
      "step": 193623
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.412323236465454,
      "learning_rate": 3.693972399740088e-05,
      "loss": 2.0494,
      "step": 193624
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.560333251953125,
      "learning_rate": 3.693775755780766e-05,
      "loss": 2.0294,
      "step": 193625
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.3413392305374146,
      "learning_rate": 3.693579116712258e-05,
      "loss": 1.9506,
      "step": 193626
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.4217545986175537,
      "learning_rate": 3.693382482534617e-05,
      "loss": 1.8549,
      "step": 193627
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.3487628698349,
      "learning_rate": 3.693185853247874e-05,
      "loss": 2.1068,
      "step": 193628
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.4126895666122437,
      "learning_rate": 3.692989228852066e-05,
      "loss": 1.906,
      "step": 193629
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.3836383819580078,
      "learning_rate": 3.692792609347218e-05,
      "loss": 1.9945,
      "step": 193630
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.416396975517273,
      "learning_rate": 3.692595994733385e-05,
      "loss": 1.781,
      "step": 193631
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.3965480327606201,
      "learning_rate": 3.692399385010586e-05,
      "loss": 2.065,
      "step": 193632
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.467883586883545,
      "learning_rate": 3.692202780178878e-05,
      "loss": 1.9359,
      "step": 193633
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.4683834314346313,
      "learning_rate": 3.6920061802382804e-05,
      "loss": 2.0266,
      "step": 193634
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.3375674486160278,
      "learning_rate": 3.691809585188837e-05,
      "loss": 2.1358,
      "step": 193635
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.392836332321167,
      "learning_rate": 3.691612995030578e-05,
      "loss": 2.069,
      "step": 193636
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.4489291906356812,
      "learning_rate": 3.691416409763554e-05,
      "loss": 1.9947,
      "step": 193637
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.3411586284637451,
      "learning_rate": 3.691219829387779e-05,
      "loss": 2.0485,
      "step": 193638
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.398360252380371,
      "learning_rate": 3.691023253903316e-05,
      "loss": 2.0303,
      "step": 193639
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.373514175415039,
      "learning_rate": 3.690826683310183e-05,
      "loss": 1.9848,
      "step": 193640
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.3327581882476807,
      "learning_rate": 3.6906301176084176e-05,
      "loss": 2.1659,
      "step": 193641
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.5407613515853882,
      "learning_rate": 3.6904335567980695e-05,
      "loss": 2.1148,
      "step": 193642
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.3710280656814575,
      "learning_rate": 3.690237000879162e-05,
      "loss": 1.7612,
      "step": 193643
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.492515206336975,
      "learning_rate": 3.690040449851731e-05,
      "loss": 2.0735,
      "step": 193644
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.3605239391326904,
      "learning_rate": 3.689843903715828e-05,
      "loss": 2.0299,
      "step": 193645
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.2316051721572876,
      "learning_rate": 3.689647362471475e-05,
      "loss": 1.8361,
      "step": 193646
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.3967816829681396,
      "learning_rate": 3.6894508261187094e-05,
      "loss": 2.0317,
      "step": 193647
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.5491225719451904,
      "learning_rate": 3.689254294657578e-05,
      "loss": 1.85,
      "step": 193648
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.2950170040130615,
      "learning_rate": 3.689057768088103e-05,
      "loss": 2.0112,
      "step": 193649
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.5174674987792969,
      "learning_rate": 3.688861246410335e-05,
      "loss": 2.0306,
      "step": 193650
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.3451128005981445,
      "learning_rate": 3.6886647296243047e-05,
      "loss": 2.0361,
      "step": 193651
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.3753660917282104,
      "learning_rate": 3.68846821773005e-05,
      "loss": 1.9092,
      "step": 193652
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.487565040588379,
      "learning_rate": 3.688271710727597e-05,
      "loss": 2.0577,
      "step": 193653
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.4049406051635742,
      "learning_rate": 3.6880752086170007e-05,
      "loss": 2.064,
      "step": 193654
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.5086861848831177,
      "learning_rate": 3.687878711398278e-05,
      "loss": 1.8871,
      "step": 193655
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.3697340488433838,
      "learning_rate": 3.687682219071485e-05,
      "loss": 2.0021,
      "step": 193656
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.4651976823806763,
      "learning_rate": 3.687485731636646e-05,
      "loss": 2.0067,
      "step": 193657
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.4055875539779663,
      "learning_rate": 3.6872892490938e-05,
      "loss": 2.0168,
      "step": 193658
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.3811836242675781,
      "learning_rate": 3.687092771442978e-05,
      "loss": 1.9944,
      "step": 193659
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.4505361318588257,
      "learning_rate": 3.6868962986842286e-05,
      "loss": 2.0826,
      "step": 193660
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.4514176845550537,
      "learning_rate": 3.6866998308175766e-05,
      "loss": 2.1298,
      "step": 193661
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.4103810787200928,
      "learning_rate": 3.686503367843068e-05,
      "loss": 2.0754,
      "step": 193662
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.4592326879501343,
      "learning_rate": 3.686306909760729e-05,
      "loss": 1.9535,
      "step": 193663
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.4028136730194092,
      "learning_rate": 3.686110456570617e-05,
      "loss": 2.013,
      "step": 193664
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.4813919067382812,
      "learning_rate": 3.6859140082727414e-05,
      "loss": 1.9179,
      "step": 193665
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.3306819200515747,
      "learning_rate": 3.6857175648671555e-05,
      "loss": 2.0138,
      "step": 193666
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.462303638458252,
      "learning_rate": 3.6855211263538834e-05,
      "loss": 1.9624,
      "step": 193667
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.4646469354629517,
      "learning_rate": 3.6853246927329774e-05,
      "loss": 2.0258,
      "step": 193668
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.289222002029419,
      "learning_rate": 3.685128264004461e-05,
      "loss": 1.906,
      "step": 193669
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.4541263580322266,
      "learning_rate": 3.684931840168388e-05,
      "loss": 2.0898,
      "step": 193670
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.3946526050567627,
      "learning_rate": 3.6847354212247717e-05,
      "loss": 2.1614,
      "step": 193671
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.4127652645111084,
      "learning_rate": 3.684539007173665e-05,
      "loss": 1.9716,
      "step": 193672
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.296159267425537,
      "learning_rate": 3.684342598015094e-05,
      "loss": 2.0519,
      "step": 193673
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.6318260431289673,
      "learning_rate": 3.6841461937491034e-05,
      "loss": 2.0032,
      "step": 193674
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.4178625345230103,
      "learning_rate": 3.683949794375722e-05,
      "loss": 1.9389,
      "step": 193675
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.408774971961975,
      "learning_rate": 3.683753399895007e-05,
      "loss": 2.1274,
      "step": 193676
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.4101520776748657,
      "learning_rate": 3.683557010306962e-05,
      "loss": 1.9786,
      "step": 193677
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.457338571548462,
      "learning_rate": 3.68336062561165e-05,
      "loss": 2.0558,
      "step": 193678
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.5442211627960205,
      "learning_rate": 3.6831642458090905e-05,
      "loss": 1.9745,
      "step": 193679
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.5291093587875366,
      "learning_rate": 3.682967870899334e-05,
      "loss": 2.2682,
      "step": 193680
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.2837536334991455,
      "learning_rate": 3.682771500882403e-05,
      "loss": 1.8591,
      "step": 193681
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.6247457265853882,
      "learning_rate": 3.682575135758355e-05,
      "loss": 2.0987,
      "step": 193682
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.4999924898147583,
      "learning_rate": 3.682378775527201e-05,
      "loss": 1.92,
      "step": 193683
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.407426357269287,
      "learning_rate": 3.682182420188995e-05,
      "loss": 1.9136,
      "step": 193684
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.4168200492858887,
      "learning_rate": 3.681986069743762e-05,
      "loss": 2.1485,
      "step": 193685
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.6511319875717163,
      "learning_rate": 3.6817897241915514e-05,
      "loss": 1.8293,
      "step": 193686
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.437210202217102,
      "learning_rate": 3.6815933835323873e-05,
      "loss": 1.9395,
      "step": 193687
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.4707812070846558,
      "learning_rate": 3.6813970477663226e-05,
      "loss": 2.3051,
      "step": 193688
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.3987680673599243,
      "learning_rate": 3.68120071689337e-05,
      "loss": 2.0719,
      "step": 193689
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.402131199836731,
      "learning_rate": 3.6810043909135864e-05,
      "loss": 2.1316,
      "step": 193690
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.4215587377548218,
      "learning_rate": 3.6808080698269956e-05,
      "loss": 1.9162,
      "step": 193691
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.3777379989624023,
      "learning_rate": 3.6806117536336444e-05,
      "loss": 2.1956,
      "step": 193692
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.4174684286117554,
      "learning_rate": 3.6804154423335585e-05,
      "loss": 2.0985,
      "step": 193693
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.4270821809768677,
      "learning_rate": 3.680219135926792e-05,
      "loss": 1.9473,
      "step": 193694
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.4257863759994507,
      "learning_rate": 3.680022834413357e-05,
      "loss": 1.8884,
      "step": 193695
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.3805725574493408,
      "learning_rate": 3.679826537793309e-05,
      "loss": 1.8828,
      "step": 193696
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.3880043029785156,
      "learning_rate": 3.6796302460666726e-05,
      "loss": 1.8852,
      "step": 193697
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.4412482976913452,
      "learning_rate": 3.679433959233495e-05,
      "loss": 2.0251,
      "step": 193698
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.4185887575149536,
      "learning_rate": 3.679237677293804e-05,
      "loss": 1.9183,
      "step": 193699
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.3550561666488647,
      "learning_rate": 3.679041400247651e-05,
      "loss": 1.8935,
      "step": 193700
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.4214894771575928,
      "learning_rate": 3.678845128095047e-05,
      "loss": 2.0115,
      "step": 193701
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.4856127500534058,
      "learning_rate": 3.678648860836049e-05,
      "loss": 2.132,
      "step": 193702
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.5529096126556396,
      "learning_rate": 3.678452598470679e-05,
      "loss": 2.1143,
      "step": 193703
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.6108211278915405,
      "learning_rate": 3.678256340998992e-05,
      "loss": 1.9734,
      "step": 193704
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.3946597576141357,
      "learning_rate": 3.6780600884210035e-05,
      "loss": 2.1449,
      "step": 193705
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.4016169309616089,
      "learning_rate": 3.67786384073677e-05,
      "loss": 2.0004,
      "step": 193706
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.343399167060852,
      "learning_rate": 3.677667597946319e-05,
      "loss": 2.03,
      "step": 193707
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.5132287740707397,
      "learning_rate": 3.677471360049683e-05,
      "loss": 1.961,
      "step": 193708
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.4352061748504639,
      "learning_rate": 3.677275127046896e-05,
      "loss": 1.8397,
      "step": 193709
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.4258590936660767,
      "learning_rate": 3.6770788989380104e-05,
      "loss": 1.9376,
      "step": 193710
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.4591532945632935,
      "learning_rate": 3.6768826757230444e-05,
      "loss": 2.0168,
      "step": 193711
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.4342864751815796,
      "learning_rate": 3.6766864574020496e-05,
      "loss": 2.0433,
      "step": 193712
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.2106233835220337,
      "learning_rate": 3.6764902439750534e-05,
      "loss": 1.8765,
      "step": 193713
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.4534199237823486,
      "learning_rate": 3.6762940354420956e-05,
      "loss": 2.0973,
      "step": 193714
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.5347206592559814,
      "learning_rate": 3.676097831803206e-05,
      "loss": 2.2174,
      "step": 193715
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.4637742042541504,
      "learning_rate": 3.6759016330584324e-05,
      "loss": 1.9812,
      "step": 193716
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.5362013578414917,
      "learning_rate": 3.6757054392078e-05,
      "loss": 2.0302,
      "step": 193717
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.354835867881775,
      "learning_rate": 3.675509250251356e-05,
      "loss": 1.7908,
      "step": 193718
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.3708804845809937,
      "learning_rate": 3.675313066189134e-05,
      "loss": 2.1465,
      "step": 193719
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.5600204467773438,
      "learning_rate": 3.6751168870211666e-05,
      "loss": 2.0502,
      "step": 193720
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.3982168436050415,
      "learning_rate": 3.674920712747488e-05,
      "loss": 1.9607,
      "step": 193721
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.5115708112716675,
      "learning_rate": 3.674724543368141e-05,
      "loss": 1.9342,
      "step": 193722
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.4583609104156494,
      "learning_rate": 3.674528378883155e-05,
      "loss": 1.9758,
      "step": 193723
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.716792345046997,
      "learning_rate": 3.674332219292577e-05,
      "loss": 2.0079,
      "step": 193724
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.292171835899353,
      "learning_rate": 3.674136064596438e-05,
      "loss": 1.6892,
      "step": 193725
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.3881458044052124,
      "learning_rate": 3.67393991479477e-05,
      "loss": 2.1658,
      "step": 193726
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.372384786605835,
      "learning_rate": 3.673743769887617e-05,
      "loss": 1.865,
      "step": 193727
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.4752044677734375,
      "learning_rate": 3.673547629875012e-05,
      "loss": 1.9119,
      "step": 193728
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.4641484022140503,
      "learning_rate": 3.673351494756985e-05,
      "loss": 2.1321,
      "step": 193729
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.4665428400039673,
      "learning_rate": 3.673155364533587e-05,
      "loss": 2.1479,
      "step": 193730
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.4005845785140991,
      "learning_rate": 3.6729592392048454e-05,
      "loss": 2.2062,
      "step": 193731
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.4452122449874878,
      "learning_rate": 3.67276311877079e-05,
      "loss": 2.1046,
      "step": 193732
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.5191576480865479,
      "learning_rate": 3.6725670032314745e-05,
      "loss": 2.0955,
      "step": 193733
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.4839290380477905,
      "learning_rate": 3.672370892586918e-05,
      "loss": 1.8855,
      "step": 193734
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.7253276109695435,
      "learning_rate": 3.6721747868371686e-05,
      "loss": 1.9891,
      "step": 193735
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.349564552307129,
      "learning_rate": 3.671978685982261e-05,
      "loss": 2.0735,
      "step": 193736
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.4264816045761108,
      "learning_rate": 3.671782590022231e-05,
      "loss": 1.962,
      "step": 193737
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.4256914854049683,
      "learning_rate": 3.6715864989571055e-05,
      "loss": 2.1626,
      "step": 193738
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.5007076263427734,
      "learning_rate": 3.671390412786934e-05,
      "loss": 2.071,
      "step": 193739
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.4700782299041748,
      "learning_rate": 3.671194331511744e-05,
      "loss": 2.1528,
      "step": 193740
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.237947940826416,
      "learning_rate": 3.670998255131582e-05,
      "loss": 2.0135,
      "step": 193741
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.2509797811508179,
      "learning_rate": 3.6708021836464774e-05,
      "loss": 1.9613,
      "step": 193742
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.3549660444259644,
      "learning_rate": 3.670606117056467e-05,
      "loss": 1.9747,
      "step": 193743
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.4256373643875122,
      "learning_rate": 3.670410055361584e-05,
      "loss": 1.9423,
      "step": 193744
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.2645649909973145,
      "learning_rate": 3.670213998561873e-05,
      "loss": 1.567,
      "step": 193745
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.405937910079956,
      "learning_rate": 3.670017946657362e-05,
      "loss": 1.9969,
      "step": 193746
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.4351961612701416,
      "learning_rate": 3.669821899648096e-05,
      "loss": 1.7834,
      "step": 193747
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.5189452171325684,
      "learning_rate": 3.6696258575341e-05,
      "loss": 1.8748,
      "step": 193748
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.4134790897369385,
      "learning_rate": 3.669429820315431e-05,
      "loss": 2.0095,
      "step": 193749
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.4762213230133057,
      "learning_rate": 3.669233787992101e-05,
      "loss": 2.096,
      "step": 193750
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.4467542171478271,
      "learning_rate": 3.669037760564164e-05,
      "loss": 2.0202,
      "step": 193751
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.9912307262420654,
      "learning_rate": 3.668841738031642e-05,
      "loss": 2.2673,
      "step": 193752
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.381614089012146,
      "learning_rate": 3.668645720394586e-05,
      "loss": 1.9957,
      "step": 193753
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.342254638671875,
      "learning_rate": 3.6684497076530216e-05,
      "loss": 1.9272,
      "step": 193754
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.4014573097229004,
      "learning_rate": 3.668253699806998e-05,
      "loss": 2.1383,
      "step": 193755
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.430399775505066,
      "learning_rate": 3.6680576968565346e-05,
      "loss": 1.9903,
      "step": 193756
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.3446582555770874,
      "learning_rate": 3.667861698801682e-05,
      "loss": 1.8572,
      "step": 193757
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.4102282524108887,
      "learning_rate": 3.667665705642463e-05,
      "loss": 2.1263,
      "step": 193758
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.4218177795410156,
      "learning_rate": 3.667469717378928e-05,
      "loss": 1.8759,
      "step": 193759
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.4702752828598022,
      "learning_rate": 3.6672737340111e-05,
      "loss": 1.8981,
      "step": 193760
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.5105409622192383,
      "learning_rate": 3.6670777555390395e-05,
      "loss": 1.9492,
      "step": 193761
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.4273395538330078,
      "learning_rate": 3.666881781962749e-05,
      "loss": 1.8966,
      "step": 193762
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.3848892450332642,
      "learning_rate": 3.666685813282293e-05,
      "loss": 2.1156,
      "step": 193763
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.3799477815628052,
      "learning_rate": 3.6664898494976865e-05,
      "loss": 1.8735,
      "step": 193764
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.5916186571121216,
      "learning_rate": 3.6662938906089875e-05,
      "loss": 2.0343,
      "step": 193765
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.5455679893493652,
      "learning_rate": 3.666097936616213e-05,
      "loss": 1.9138,
      "step": 193766
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.3481632471084595,
      "learning_rate": 3.6659019875194205e-05,
      "loss": 2.0773,
      "step": 193767
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.9064185619354248,
      "learning_rate": 3.6657060433186224e-05,
      "loss": 2.0591,
      "step": 193768
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.46940279006958,
      "learning_rate": 3.665510104013871e-05,
      "loss": 1.9708,
      "step": 193769
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.4434627294540405,
      "learning_rate": 3.6653141696051935e-05,
      "loss": 2.0263,
      "step": 193770
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.4647953510284424,
      "learning_rate": 3.665118240092636e-05,
      "loss": 2.2375,
      "step": 193771
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.4964218139648438,
      "learning_rate": 3.6649223154762255e-05,
      "loss": 1.9937,
      "step": 193772
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.5381439924240112,
      "learning_rate": 3.664726395756008e-05,
      "loss": 1.9623,
      "step": 193773
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.3769396543502808,
      "learning_rate": 3.664530480932018e-05,
      "loss": 1.8735,
      "step": 193774
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.3869949579238892,
      "learning_rate": 3.664334571004287e-05,
      "loss": 1.9463,
      "step": 193775
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.417399287223816,
      "learning_rate": 3.664138665972847e-05,
      "loss": 1.9324,
      "step": 193776
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.414595365524292,
      "learning_rate": 3.663942765837746e-05,
      "loss": 1.8645,
      "step": 193777
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.6407065391540527,
      "learning_rate": 3.663746870599009e-05,
      "loss": 2.1486,
      "step": 193778
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.3927943706512451,
      "learning_rate": 3.663550980256685e-05,
      "loss": 2.034,
      "step": 193779
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.4451359510421753,
      "learning_rate": 3.6633550948108045e-05,
      "loss": 2.039,
      "step": 193780
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.4619051218032837,
      "learning_rate": 3.6631592142614044e-05,
      "loss": 2.087,
      "step": 193781
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.6321252584457397,
      "learning_rate": 3.6629633386085135e-05,
      "loss": 2.1514,
      "step": 193782
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.382371187210083,
      "learning_rate": 3.662767467852179e-05,
      "loss": 1.9149,
      "step": 193783
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.5773018598556519,
      "learning_rate": 3.662571601992431e-05,
      "loss": 1.8209,
      "step": 193784
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.3605742454528809,
      "learning_rate": 3.66237574102931e-05,
      "loss": 1.8577,
      "step": 193785
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.3305100202560425,
      "learning_rate": 3.662179884962852e-05,
      "loss": 2.0929,
      "step": 193786
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.3383501768112183,
      "learning_rate": 3.661984033793093e-05,
      "loss": 1.9061,
      "step": 193787
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.5347095727920532,
      "learning_rate": 3.66178818752006e-05,
      "loss": 1.9299,
      "step": 193788
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.388513445854187,
      "learning_rate": 3.661592346143808e-05,
      "loss": 1.9148,
      "step": 193789
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.6511443853378296,
      "learning_rate": 3.661396509664355e-05,
      "loss": 2.2453,
      "step": 193790
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.4112409353256226,
      "learning_rate": 3.661200678081748e-05,
      "loss": 2.014,
      "step": 193791
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.4875353574752808,
      "learning_rate": 3.661004851396024e-05,
      "loss": 2.0931,
      "step": 193792
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.3528603315353394,
      "learning_rate": 3.6608090296072166e-05,
      "loss": 2.039,
      "step": 193793
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.4517035484313965,
      "learning_rate": 3.660613212715359e-05,
      "loss": 2.0245,
      "step": 193794
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.4531275033950806,
      "learning_rate": 3.66041740072049e-05,
      "loss": 2.018,
      "step": 193795
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.4403448104858398,
      "learning_rate": 3.660221593622644e-05,
      "loss": 2.1384,
      "step": 193796
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.4366304874420166,
      "learning_rate": 3.660025791421868e-05,
      "loss": 1.8079,
      "step": 193797
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.4185912609100342,
      "learning_rate": 3.6598299941181875e-05,
      "loss": 1.8517,
      "step": 193798
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.4221711158752441,
      "learning_rate": 3.659634201711643e-05,
      "loss": 1.9144,
      "step": 193799
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.4759511947631836,
      "learning_rate": 3.659438414202265e-05,
      "loss": 1.912,
      "step": 193800
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.4356786012649536,
      "learning_rate": 3.659242631590099e-05,
      "loss": 1.8881,
      "step": 193801
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.5098410844802856,
      "learning_rate": 3.6590468538751695e-05,
      "loss": 2.0824,
      "step": 193802
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.471914529800415,
      "learning_rate": 3.658851081057529e-05,
      "loss": 2.1203,
      "step": 193803
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.2126942873001099,
      "learning_rate": 3.658655313137204e-05,
      "loss": 1.9248,
      "step": 193804
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.452758550643921,
      "learning_rate": 3.658459550114232e-05,
      "loss": 2.3425,
      "step": 193805
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.342382550239563,
      "learning_rate": 3.658263791988646e-05,
      "loss": 1.9244,
      "step": 193806
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.30947744846344,
      "learning_rate": 3.658068038760489e-05,
      "loss": 2.051,
      "step": 193807
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.3896416425704956,
      "learning_rate": 3.657872290429787e-05,
      "loss": 2.1434,
      "step": 193808
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.390986680984497,
      "learning_rate": 3.6576765469965954e-05,
      "loss": 2.0085,
      "step": 193809
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.4048274755477905,
      "learning_rate": 3.657480808460935e-05,
      "loss": 2.065,
      "step": 193810
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.4427738189697266,
      "learning_rate": 3.657285074822841e-05,
      "loss": 2.1254,
      "step": 193811
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.3836641311645508,
      "learning_rate": 3.657089346082359e-05,
      "loss": 1.9484,
      "step": 193812
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.3621079921722412,
      "learning_rate": 3.656893622239526e-05,
      "loss": 2.16,
      "step": 193813
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.3653463125228882,
      "learning_rate": 3.656697903294365e-05,
      "loss": 1.935,
      "step": 193814
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.421952724456787,
      "learning_rate": 3.656502189246927e-05,
      "loss": 2.0377,
      "step": 193815
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.3912731409072876,
      "learning_rate": 3.656306480097241e-05,
      "loss": 2.1225,
      "step": 193816
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.5384267568588257,
      "learning_rate": 3.656110775845344e-05,
      "loss": 2.0249,
      "step": 193817
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.520565152168274,
      "learning_rate": 3.655915076491275e-05,
      "loss": 1.9061,
      "step": 193818
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.4692407846450806,
      "learning_rate": 3.655719382035068e-05,
      "loss": 1.9975,
      "step": 193819
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.4895457029342651,
      "learning_rate": 3.655523692476757e-05,
      "loss": 1.9989,
      "step": 193820
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.3476115465164185,
      "learning_rate": 3.655328007816385e-05,
      "loss": 2.2301,
      "step": 193821
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.3468295335769653,
      "learning_rate": 3.6551323280539846e-05,
      "loss": 2.1713,
      "step": 193822
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.540557861328125,
      "learning_rate": 3.65493665318959e-05,
      "loss": 1.7151,
      "step": 193823
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.3761587142944336,
      "learning_rate": 3.6547409832232435e-05,
      "loss": 1.8817,
      "step": 193824
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.5406657457351685,
      "learning_rate": 3.6545453181549724e-05,
      "loss": 1.9056,
      "step": 193825
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.3709293603897095,
      "learning_rate": 3.654349657984823e-05,
      "loss": 2.0211,
      "step": 193826
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.5816198587417603,
      "learning_rate": 3.654154002712829e-05,
      "loss": 1.9518,
      "step": 193827
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.341857671737671,
      "learning_rate": 3.6539583523390235e-05,
      "loss": 2.0694,
      "step": 193828
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.542932391166687,
      "learning_rate": 3.65376270686344e-05,
      "loss": 2.1093,
      "step": 193829
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.5931763648986816,
      "learning_rate": 3.6535670662861247e-05,
      "loss": 2.0548,
      "step": 193830
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.5079423189163208,
      "learning_rate": 3.653371430607105e-05,
      "loss": 1.8285,
      "step": 193831
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.4706050157546997,
      "learning_rate": 3.653175799826423e-05,
      "loss": 1.9248,
      "step": 193832
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.4895329475402832,
      "learning_rate": 3.65298017394411e-05,
      "loss": 1.9704,
      "step": 193833
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.3256815671920776,
      "learning_rate": 3.652784552960215e-05,
      "loss": 2.1007,
      "step": 193834
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.4816118478775024,
      "learning_rate": 3.652588936874755e-05,
      "loss": 2.1685,
      "step": 193835
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.3574787378311157,
      "learning_rate": 3.65239332568778e-05,
      "loss": 2.0401,
      "step": 193836
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.3836480379104614,
      "learning_rate": 3.652197719399317e-05,
      "loss": 1.7298,
      "step": 193837
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.3083935976028442,
      "learning_rate": 3.652002118009415e-05,
      "loss": 2.0474,
      "step": 193838
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.3211263418197632,
      "learning_rate": 3.651806521518098e-05,
      "loss": 1.8863,
      "step": 193839
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.5219111442565918,
      "learning_rate": 3.651610929925409e-05,
      "loss": 1.8417,
      "step": 193840
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.3155663013458252,
      "learning_rate": 3.651415343231389e-05,
      "loss": 1.9717,
      "step": 193841
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.6001877784729004,
      "learning_rate": 3.6512197614360624e-05,
      "loss": 2.0179,
      "step": 193842
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.3345041275024414,
      "learning_rate": 3.6510241845394684e-05,
      "loss": 2.1077,
      "step": 193843
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.2317477464675903,
      "learning_rate": 3.650828612541652e-05,
      "loss": 1.7876,
      "step": 193844
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.421376347541809,
      "learning_rate": 3.650633045442637e-05,
      "loss": 2.0235,
      "step": 193845
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.460279107093811,
      "learning_rate": 3.650437483242473e-05,
      "loss": 1.9508,
      "step": 193846
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.5096949338912964,
      "learning_rate": 3.650241925941194e-05,
      "loss": 2.0015,
      "step": 193847
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.5351594686508179,
      "learning_rate": 3.6500463735388265e-05,
      "loss": 1.8966,
      "step": 193848
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.470486044883728,
      "learning_rate": 3.64985082603541e-05,
      "loss": 1.8251,
      "step": 193849
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.3177376985549927,
      "learning_rate": 3.649655283430991e-05,
      "loss": 1.8856,
      "step": 193850
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.481019377708435,
      "learning_rate": 3.64945974572559e-05,
      "loss": 1.8379,
      "step": 193851
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.4666961431503296,
      "learning_rate": 3.6492642129192574e-05,
      "loss": 1.9412,
      "step": 193852
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.5012550354003906,
      "learning_rate": 3.6490686850120255e-05,
      "loss": 1.9283,
      "step": 193853
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.4586042165756226,
      "learning_rate": 3.648873162003928e-05,
      "loss": 2.0148,
      "step": 193854
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.4422080516815186,
      "learning_rate": 3.648677643894998e-05,
      "loss": 1.9881,
      "step": 193855
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.5282847881317139,
      "learning_rate": 3.6484821306852794e-05,
      "loss": 2.1715,
      "step": 193856
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.3716973066329956,
      "learning_rate": 3.648286622374802e-05,
      "loss": 1.9831,
      "step": 193857
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.3550963401794434,
      "learning_rate": 3.6480911189636116e-05,
      "loss": 1.9527,
      "step": 193858
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.4159605503082275,
      "learning_rate": 3.647895620451736e-05,
      "loss": 2.23,
      "step": 193859
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.3807557821273804,
      "learning_rate": 3.6477001268392146e-05,
      "loss": 2.1788,
      "step": 193860
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.402570366859436,
      "learning_rate": 3.647504638126081e-05,
      "loss": 1.9384,
      "step": 193861
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.3265736103057861,
      "learning_rate": 3.647309154312374e-05,
      "loss": 1.8001,
      "step": 193862
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.4907358884811401,
      "learning_rate": 3.647113675398129e-05,
      "loss": 2.1274,
      "step": 193863
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.5031148195266724,
      "learning_rate": 3.6469182013833875e-05,
      "loss": 2.2739,
      "step": 193864
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.4104373455047607,
      "learning_rate": 3.6467227322681805e-05,
      "loss": 1.8914,
      "step": 193865
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.4012317657470703,
      "learning_rate": 3.6465272680525435e-05,
      "loss": 1.9951,
      "step": 193866
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.379841923713684,
      "learning_rate": 3.646331808736511e-05,
      "loss": 1.9337,
      "step": 193867
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.590355396270752,
      "learning_rate": 3.646136354320126e-05,
      "loss": 1.8751,
      "step": 193868
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.3490279912948608,
      "learning_rate": 3.6459409048034185e-05,
      "loss": 1.905,
      "step": 193869
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.300101399421692,
      "learning_rate": 3.6457454601864345e-05,
      "loss": 1.9428,
      "step": 193870
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.7701154947280884,
      "learning_rate": 3.6455500204692014e-05,
      "loss": 1.9115,
      "step": 193871
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.3907065391540527,
      "learning_rate": 3.645354585651762e-05,
      "loss": 1.8991,
      "step": 193872
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.2875382900238037,
      "learning_rate": 3.6451591557341395e-05,
      "loss": 1.7949,
      "step": 193873
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.482784628868103,
      "learning_rate": 3.6449637307163846e-05,
      "loss": 2.0838,
      "step": 193874
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.462945818901062,
      "learning_rate": 3.6447683105985235e-05,
      "loss": 2.0031,
      "step": 193875
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.418280839920044,
      "learning_rate": 3.6445728953806064e-05,
      "loss": 2.1576,
      "step": 193876
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.563678503036499,
      "learning_rate": 3.644377485062656e-05,
      "loss": 1.9536,
      "step": 193877
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.7149348258972168,
      "learning_rate": 3.6441820796447166e-05,
      "loss": 1.7324,
      "step": 193878
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.44509756565094,
      "learning_rate": 3.643986679126814e-05,
      "loss": 1.8159,
      "step": 193879
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.3530113697052002,
      "learning_rate": 3.643791283508998e-05,
      "loss": 1.8144,
      "step": 193880
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.4134551286697388,
      "learning_rate": 3.6435958927912925e-05,
      "loss": 1.8926,
      "step": 193881
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.5183366537094116,
      "learning_rate": 3.6434005069737474e-05,
      "loss": 2.0458,
      "step": 193882
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.4412164688110352,
      "learning_rate": 3.643205126056389e-05,
      "loss": 2.0899,
      "step": 193883
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.3732107877731323,
      "learning_rate": 3.6430097500392575e-05,
      "loss": 1.954,
      "step": 193884
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.5226972103118896,
      "learning_rate": 3.642814378922383e-05,
      "loss": 1.8809,
      "step": 193885
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.3682568073272705,
      "learning_rate": 3.642619012705812e-05,
      "loss": 2.0669,
      "step": 193886
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.5131537914276123,
      "learning_rate": 3.6424236513895676e-05,
      "loss": 1.9201,
      "step": 193887
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.3825812339782715,
      "learning_rate": 3.6422282949737037e-05,
      "loss": 2.0797,
      "step": 193888
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.4384597539901733,
      "learning_rate": 3.6420329434582465e-05,
      "loss": 1.9612,
      "step": 193889
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.4491831064224243,
      "learning_rate": 3.641837596843232e-05,
      "loss": 2.0024,
      "step": 193890
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.4594948291778564,
      "learning_rate": 3.6416422551286914e-05,
      "loss": 1.9762,
      "step": 193891
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.3459765911102295,
      "learning_rate": 3.641446918314671e-05,
      "loss": 2.0014,
      "step": 193892
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.4374672174453735,
      "learning_rate": 3.6412515864012006e-05,
      "loss": 1.9516,
      "step": 193893
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.4747390747070312,
      "learning_rate": 3.641056259388323e-05,
      "loss": 1.9188,
      "step": 193894
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.3862665891647339,
      "learning_rate": 3.640860937276072e-05,
      "loss": 1.9071,
      "step": 193895
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.5578588247299194,
      "learning_rate": 3.640665620064475e-05,
      "loss": 2.1006,
      "step": 193896
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.4479395151138306,
      "learning_rate": 3.640470307753583e-05,
      "loss": 1.9575,
      "step": 193897
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.4675267934799194,
      "learning_rate": 3.640275000343426e-05,
      "loss": 2.0845,
      "step": 193898
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.4222195148468018,
      "learning_rate": 3.640079697834031e-05,
      "loss": 2.0181,
      "step": 193899
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.3660863637924194,
      "learning_rate": 3.6398844002254525e-05,
      "loss": 1.924,
      "step": 193900
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.3528952598571777,
      "learning_rate": 3.6396891075177134e-05,
      "loss": 2.0407,
      "step": 193901
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.3314872980117798,
      "learning_rate": 3.639493819710848e-05,
      "loss": 2.0374,
      "step": 193902
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.350279450416565,
      "learning_rate": 3.639298536804906e-05,
      "loss": 1.9545,
      "step": 193903
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.2740634679794312,
      "learning_rate": 3.639103258799916e-05,
      "loss": 1.9517,
      "step": 193904
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.3229494094848633,
      "learning_rate": 3.638907985695906e-05,
      "loss": 2.1535,
      "step": 193905
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.4989255666732788,
      "learning_rate": 3.638712717492929e-05,
      "loss": 2.0976,
      "step": 193906
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.387593388557434,
      "learning_rate": 3.638517454191008e-05,
      "loss": 1.9923,
      "step": 193907
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.3372803926467896,
      "learning_rate": 3.638322195790187e-05,
      "loss": 1.9959,
      "step": 193908
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.5228338241577148,
      "learning_rate": 3.6381269422905016e-05,
      "loss": 2.3518,
      "step": 193909
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.460735559463501,
      "learning_rate": 3.637931693691979e-05,
      "loss": 2.0997,
      "step": 193910
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.4616577625274658,
      "learning_rate": 3.637736449994669e-05,
      "loss": 1.9056,
      "step": 193911
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.4990758895874023,
      "learning_rate": 3.6375412111986026e-05,
      "loss": 1.9396,
      "step": 193912
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.5237780809402466,
      "learning_rate": 3.637345977303805e-05,
      "loss": 2.1504,
      "step": 193913
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.3458459377288818,
      "learning_rate": 3.63715074831033e-05,
      "loss": 2.1989,
      "step": 193914
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.3484203815460205,
      "learning_rate": 3.636955524218208e-05,
      "loss": 1.9384,
      "step": 193915
    },
    {
      "epoch": 2.52,
      "grad_norm": 2.2313127517700195,
      "learning_rate": 3.636760305027469e-05,
      "loss": 1.9729,
      "step": 193916
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.3109077215194702,
      "learning_rate": 3.6365650907381585e-05,
      "loss": 1.8483,
      "step": 193917
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.3342664241790771,
      "learning_rate": 3.6363698813503074e-05,
      "loss": 1.8233,
      "step": 193918
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.3827401399612427,
      "learning_rate": 3.636174676863945e-05,
      "loss": 2.1013,
      "step": 193919
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.640189528465271,
      "learning_rate": 3.635979477279123e-05,
      "loss": 2.0191,
      "step": 193920
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.4624180793762207,
      "learning_rate": 3.6357842825958694e-05,
      "loss": 1.9835,
      "step": 193921
    },
    {
      "epoch": 2.52,
      "grad_norm": 1.7218292951583862,
      "learning_rate": 3.635589092814218e-05,
      "loss": 2.1315,
      "step": 193922
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.4800546169281006,
      "learning_rate": 3.635393907934213e-05,
      "loss": 1.9286,
      "step": 193923
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.5227172374725342,
      "learning_rate": 3.635198727955877e-05,
      "loss": 2.2932,
      "step": 193924
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.2741413116455078,
      "learning_rate": 3.635003552879263e-05,
      "loss": 1.8915,
      "step": 193925
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.314631462097168,
      "learning_rate": 3.634808382704402e-05,
      "loss": 1.7045,
      "step": 193926
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.2791874408721924,
      "learning_rate": 3.6346132174313257e-05,
      "loss": 2.0052,
      "step": 193927
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.6092277765274048,
      "learning_rate": 3.634418057060066e-05,
      "loss": 1.96,
      "step": 193928
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.3940606117248535,
      "learning_rate": 3.634222901590674e-05,
      "loss": 1.9299,
      "step": 193929
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.3390729427337646,
      "learning_rate": 3.634027751023169e-05,
      "loss": 2.1317,
      "step": 193930
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.369248390197754,
      "learning_rate": 3.6338326053576015e-05,
      "loss": 1.7209,
      "step": 193931
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.479923963546753,
      "learning_rate": 3.6336374645940036e-05,
      "loss": 2.0015,
      "step": 193932
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.5796127319335938,
      "learning_rate": 3.633442328732411e-05,
      "loss": 2.001,
      "step": 193933
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.3664745092391968,
      "learning_rate": 3.633247197772854e-05,
      "loss": 2.0705,
      "step": 193934
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.4335983991622925,
      "learning_rate": 3.633052071715379e-05,
      "loss": 2.0631,
      "step": 193935
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.3840839862823486,
      "learning_rate": 3.632856950560009e-05,
      "loss": 1.8134,
      "step": 193936
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.5708338022232056,
      "learning_rate": 3.632661834306798e-05,
      "loss": 2.11,
      "step": 193937
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.476122498512268,
      "learning_rate": 3.632466722955769e-05,
      "loss": 1.91,
      "step": 193938
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.357216477394104,
      "learning_rate": 3.632271616506965e-05,
      "loss": 1.9116,
      "step": 193939
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.3678514957427979,
      "learning_rate": 3.6320765149604135e-05,
      "loss": 1.8832,
      "step": 193940
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.3662680387496948,
      "learning_rate": 3.6318814183161605e-05,
      "loss": 1.9803,
      "step": 193941
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.3551336526870728,
      "learning_rate": 3.631686326574236e-05,
      "loss": 1.9304,
      "step": 193942
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.5274453163146973,
      "learning_rate": 3.631491239734683e-05,
      "loss": 1.9351,
      "step": 193943
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.3048996925354004,
      "learning_rate": 3.631296157797532e-05,
      "loss": 1.7818,
      "step": 193944
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.3728396892547607,
      "learning_rate": 3.631101080762823e-05,
      "loss": 1.8756,
      "step": 193945
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.3499106168746948,
      "learning_rate": 3.630906008630582e-05,
      "loss": 2.0101,
      "step": 193946
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.4195332527160645,
      "learning_rate": 3.6307109414008625e-05,
      "loss": 2.0737,
      "step": 193947
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.4314302206039429,
      "learning_rate": 3.6305158790736825e-05,
      "loss": 2.0572,
      "step": 193948
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.4993613958358765,
      "learning_rate": 3.630320821649093e-05,
      "loss": 1.9268,
      "step": 193949
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.5298762321472168,
      "learning_rate": 3.630125769127129e-05,
      "loss": 1.9775,
      "step": 193950
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.5501707792282104,
      "learning_rate": 3.62993072150782e-05,
      "loss": 1.9887,
      "step": 193951
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.3753528594970703,
      "learning_rate": 3.629735678791196e-05,
      "loss": 1.9593,
      "step": 193952
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.3709287643432617,
      "learning_rate": 3.62954064097731e-05,
      "loss": 2.1629,
      "step": 193953
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.2683149576187134,
      "learning_rate": 3.629345608066185e-05,
      "loss": 2.1328,
      "step": 193954
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.344903826713562,
      "learning_rate": 3.629150580057869e-05,
      "loss": 2.1945,
      "step": 193955
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.321435570716858,
      "learning_rate": 3.6289555569523875e-05,
      "loss": 1.9875,
      "step": 193956
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.4764765501022339,
      "learning_rate": 3.628760538749784e-05,
      "loss": 1.8898,
      "step": 193957
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.459895372390747,
      "learning_rate": 3.6285655254500855e-05,
      "loss": 2.0919,
      "step": 193958
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.3787771463394165,
      "learning_rate": 3.6283705170533414e-05,
      "loss": 1.9887,
      "step": 193959
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.4077264070510864,
      "learning_rate": 3.628175513559576e-05,
      "loss": 1.9241,
      "step": 193960
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.500957727432251,
      "learning_rate": 3.6279805149688345e-05,
      "loss": 1.9807,
      "step": 193961
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.3093876838684082,
      "learning_rate": 3.627785521281148e-05,
      "loss": 1.8379,
      "step": 193962
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.3829795122146606,
      "learning_rate": 3.627590532496556e-05,
      "loss": 1.948,
      "step": 193963
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.2855132818222046,
      "learning_rate": 3.627395548615085e-05,
      "loss": 1.9859,
      "step": 193964
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.510399341583252,
      "learning_rate": 3.627200569636786e-05,
      "loss": 2.107,
      "step": 193965
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.4726052284240723,
      "learning_rate": 3.627005595561684e-05,
      "loss": 1.9901,
      "step": 193966
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.331095814704895,
      "learning_rate": 3.626810626389823e-05,
      "loss": 1.8532,
      "step": 193967
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.2818224430084229,
      "learning_rate": 3.626615662121237e-05,
      "loss": 1.8234,
      "step": 193968
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.4802864789962769,
      "learning_rate": 3.626420702755962e-05,
      "loss": 1.9834,
      "step": 193969
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.3254739046096802,
      "learning_rate": 3.626225748294025e-05,
      "loss": 2.1051,
      "step": 193970
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.4079481363296509,
      "learning_rate": 3.626030798735475e-05,
      "loss": 1.9384,
      "step": 193971
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.4257973432540894,
      "learning_rate": 3.62583585408034e-05,
      "loss": 1.8026,
      "step": 193972
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.4644826650619507,
      "learning_rate": 3.6256409143286694e-05,
      "loss": 2.2656,
      "step": 193973
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.438681721687317,
      "learning_rate": 3.625445979480479e-05,
      "loss": 2.0659,
      "step": 193974
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.3883811235427856,
      "learning_rate": 3.62525104953583e-05,
      "loss": 1.9444,
      "step": 193975
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.3950507640838623,
      "learning_rate": 3.625056124494732e-05,
      "loss": 2.0729,
      "step": 193976
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.4638822078704834,
      "learning_rate": 3.624861204357241e-05,
      "loss": 1.9245,
      "step": 193977
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.3306512832641602,
      "learning_rate": 3.624666289123378e-05,
      "loss": 2.2306,
      "step": 193978
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.7026604413986206,
      "learning_rate": 3.624471378793195e-05,
      "loss": 1.8964,
      "step": 193979
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.4145869016647339,
      "learning_rate": 3.624276473366714e-05,
      "loss": 2.2013,
      "step": 193980
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.4712673425674438,
      "learning_rate": 3.6240815728439933e-05,
      "loss": 2.0518,
      "step": 193981
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.4370851516723633,
      "learning_rate": 3.623886677225036e-05,
      "loss": 1.8779,
      "step": 193982
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.5873267650604248,
      "learning_rate": 3.623691786509907e-05,
      "loss": 2.0908,
      "step": 193983
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.4211825132369995,
      "learning_rate": 3.6234969006986216e-05,
      "loss": 1.9404,
      "step": 193984
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.5376735925674438,
      "learning_rate": 3.623302019791233e-05,
      "loss": 2.0838,
      "step": 193985
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.4715795516967773,
      "learning_rate": 3.623107143787766e-05,
      "loss": 2.0266,
      "step": 193986
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.4811068773269653,
      "learning_rate": 3.622912272688269e-05,
      "loss": 1.9907,
      "step": 193987
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.4889615774154663,
      "learning_rate": 3.622717406492766e-05,
      "loss": 1.9863,
      "step": 193988
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.5854218006134033,
      "learning_rate": 3.6225225452012995e-05,
      "loss": 1.8956,
      "step": 193989
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.3179638385772705,
      "learning_rate": 3.6223276888139006e-05,
      "loss": 2.0517,
      "step": 193990
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.6438791751861572,
      "learning_rate": 3.6221328373306124e-05,
      "loss": 2.085,
      "step": 193991
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.4227958917617798,
      "learning_rate": 3.621937990751461e-05,
      "loss": 1.8983,
      "step": 193992
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.3885648250579834,
      "learning_rate": 3.621743149076497e-05,
      "loss": 1.8194,
      "step": 193993
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.3292464017868042,
      "learning_rate": 3.621548312305747e-05,
      "loss": 1.9564,
      "step": 193994
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.4915980100631714,
      "learning_rate": 3.621353480439243e-05,
      "loss": 1.8565,
      "step": 193995
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.4165343046188354,
      "learning_rate": 3.621158653477037e-05,
      "loss": 2.0937,
      "step": 193996
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.4675874710083008,
      "learning_rate": 3.62096383141915e-05,
      "loss": 1.9455,
      "step": 193997
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.389096975326538,
      "learning_rate": 3.6207690142656206e-05,
      "loss": 2.0239,
      "step": 193998
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.5577601194381714,
      "learning_rate": 3.6205742020164954e-05,
      "loss": 1.9099,
      "step": 193999
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.4357811212539673,
      "learning_rate": 3.620379394671803e-05,
      "loss": 2.1104,
      "step": 194000
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.372784972190857,
      "learning_rate": 3.6201845922315744e-05,
      "loss": 1.8717,
      "step": 194001
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.3740837574005127,
      "learning_rate": 3.619989794695855e-05,
      "loss": 2.0446,
      "step": 194002
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.6360617876052856,
      "learning_rate": 3.6197950020646805e-05,
      "loss": 2.0726,
      "step": 194003
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.5182888507843018,
      "learning_rate": 3.619600214338075e-05,
      "loss": 2.0108,
      "step": 194004
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.39875328540802,
      "learning_rate": 3.61940543151609e-05,
      "loss": 1.9832,
      "step": 194005
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.5570847988128662,
      "learning_rate": 3.619210653598759e-05,
      "loss": 1.9965,
      "step": 194006
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.5467519760131836,
      "learning_rate": 3.619015880586107e-05,
      "loss": 2.0408,
      "step": 194007
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.4173972606658936,
      "learning_rate": 3.618821112478182e-05,
      "loss": 2.1486,
      "step": 194008
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.5013811588287354,
      "learning_rate": 3.61862634927501e-05,
      "loss": 1.8831,
      "step": 194009
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.4531699419021606,
      "learning_rate": 3.6184315909766416e-05,
      "loss": 1.9953,
      "step": 194010
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.4679853916168213,
      "learning_rate": 3.618236837583106e-05,
      "loss": 2.1005,
      "step": 194011
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.3097542524337769,
      "learning_rate": 3.618042089094434e-05,
      "loss": 1.974,
      "step": 194012
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.3151605129241943,
      "learning_rate": 3.6178473455106614e-05,
      "loss": 1.9625,
      "step": 194013
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.482195258140564,
      "learning_rate": 3.617652606831839e-05,
      "loss": 2.0032,
      "step": 194014
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.3370712995529175,
      "learning_rate": 3.6174578730579826e-05,
      "loss": 1.972,
      "step": 194015
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.3496910333633423,
      "learning_rate": 3.617263144189146e-05,
      "loss": 1.9788,
      "step": 194016
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.6969547271728516,
      "learning_rate": 3.6170684202253564e-05,
      "loss": 1.8894,
      "step": 194017
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.4360268115997314,
      "learning_rate": 3.616873701166656e-05,
      "loss": 1.9899,
      "step": 194018
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.3781659603118896,
      "learning_rate": 3.6166789870130686e-05,
      "loss": 2.1227,
      "step": 194019
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.3920315504074097,
      "learning_rate": 3.616484277764644e-05,
      "loss": 2.0074,
      "step": 194020
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.5832791328430176,
      "learning_rate": 3.616289573421406e-05,
      "loss": 1.9887,
      "step": 194021
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.305153489112854,
      "learning_rate": 3.616094873983404e-05,
      "loss": 1.7889,
      "step": 194022
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.5631139278411865,
      "learning_rate": 3.615900179450671e-05,
      "loss": 2.0794,
      "step": 194023
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.4401618242263794,
      "learning_rate": 3.615705489823238e-05,
      "loss": 2.1585,
      "step": 194024
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.3760020732879639,
      "learning_rate": 3.6155108051011384e-05,
      "loss": 2.0659,
      "step": 194025
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.3919020891189575,
      "learning_rate": 3.615316125284418e-05,
      "loss": 2.1401,
      "step": 194026
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.511107325553894,
      "learning_rate": 3.615121450373103e-05,
      "loss": 1.9769,
      "step": 194027
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.5314033031463623,
      "learning_rate": 3.614926780367241e-05,
      "loss": 1.9271,
      "step": 194028
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.375572919845581,
      "learning_rate": 3.614732115266862e-05,
      "loss": 2.086,
      "step": 194029
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.4281072616577148,
      "learning_rate": 3.6145374550720016e-05,
      "loss": 1.9613,
      "step": 194030
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.4341989755630493,
      "learning_rate": 3.6143427997826913e-05,
      "loss": 1.9281,
      "step": 194031
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.307281255722046,
      "learning_rate": 3.61414814939898e-05,
      "loss": 2.1185,
      "step": 194032
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.330477237701416,
      "learning_rate": 3.613953503920891e-05,
      "loss": 1.9134,
      "step": 194033
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.357943058013916,
      "learning_rate": 3.613758863348472e-05,
      "loss": 1.9276,
      "step": 194034
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.4193052053451538,
      "learning_rate": 3.613564227681751e-05,
      "loss": 2.011,
      "step": 194035
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.290562629699707,
      "learning_rate": 3.613369596920767e-05,
      "loss": 1.9883,
      "step": 194036
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.3604391813278198,
      "learning_rate": 3.613174971065551e-05,
      "loss": 1.9291,
      "step": 194037
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.5129622220993042,
      "learning_rate": 3.612980350116148e-05,
      "loss": 2.07,
      "step": 194038
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.429437279701233,
      "learning_rate": 3.612785734072587e-05,
      "loss": 2.1142,
      "step": 194039
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.6063663959503174,
      "learning_rate": 3.6125911229349094e-05,
      "loss": 2.2139,
      "step": 194040
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.2181715965270996,
      "learning_rate": 3.612396516703147e-05,
      "loss": 1.8663,
      "step": 194041
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.3808069229125977,
      "learning_rate": 3.6122019153773495e-05,
      "loss": 2.0484,
      "step": 194042
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.4554734230041504,
      "learning_rate": 3.6120073189575284e-05,
      "loss": 1.9889,
      "step": 194043
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.3873417377471924,
      "learning_rate": 3.611812727443739e-05,
      "loss": 1.9785,
      "step": 194044
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.501847743988037,
      "learning_rate": 3.611618140836006e-05,
      "loss": 1.9481,
      "step": 194045
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.2465039491653442,
      "learning_rate": 3.6114235591343777e-05,
      "loss": 1.8917,
      "step": 194046
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.370724081993103,
      "learning_rate": 3.611228982338879e-05,
      "loss": 1.9093,
      "step": 194047
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.426840901374817,
      "learning_rate": 3.6110344104495615e-05,
      "loss": 2.2442,
      "step": 194048
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.493790626525879,
      "learning_rate": 3.61083984346644e-05,
      "loss": 1.9085,
      "step": 194049
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.503818154335022,
      "learning_rate": 3.6106452813895635e-05,
      "loss": 2.0679,
      "step": 194050
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.3934799432754517,
      "learning_rate": 3.610450724218966e-05,
      "loss": 2.1658,
      "step": 194051
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.3662984371185303,
      "learning_rate": 3.6102561719546874e-05,
      "loss": 1.8529,
      "step": 194052
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.5199284553527832,
      "learning_rate": 3.610061624596754e-05,
      "loss": 2.0177,
      "step": 194053
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.4036767482757568,
      "learning_rate": 3.609867082145219e-05,
      "loss": 2.1743,
      "step": 194054
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.429456114768982,
      "learning_rate": 3.609672544600096e-05,
      "loss": 2.1989,
      "step": 194055
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.3611167669296265,
      "learning_rate": 3.609478011961442e-05,
      "loss": 2.0525,
      "step": 194056
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.4082106351852417,
      "learning_rate": 3.609283484229276e-05,
      "loss": 1.665,
      "step": 194057
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.3499006032943726,
      "learning_rate": 3.609088961403649e-05,
      "loss": 2.0565,
      "step": 194058
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.4305678606033325,
      "learning_rate": 3.608894443484584e-05,
      "loss": 2.0199,
      "step": 194059
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.690024733543396,
      "learning_rate": 3.608699930472136e-05,
      "loss": 2.1399,
      "step": 194060
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.3548657894134521,
      "learning_rate": 3.608505422366318e-05,
      "loss": 1.897,
      "step": 194061
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.422803521156311,
      "learning_rate": 3.608310919167181e-05,
      "loss": 1.8902,
      "step": 194062
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.3512271642684937,
      "learning_rate": 3.608116420874753e-05,
      "loss": 1.9665,
      "step": 194063
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.5241427421569824,
      "learning_rate": 3.607921927489079e-05,
      "loss": 2.1983,
      "step": 194064
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.3100680112838745,
      "learning_rate": 3.607727439010184e-05,
      "loss": 1.8352,
      "step": 194065
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.471082091331482,
      "learning_rate": 3.607532955438124e-05,
      "loss": 2.1035,
      "step": 194066
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.3977372646331787,
      "learning_rate": 3.607338476772908e-05,
      "loss": 2.0153,
      "step": 194067
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.5569369792938232,
      "learning_rate": 3.607144003014595e-05,
      "loss": 2.18,
      "step": 194068
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.434208631515503,
      "learning_rate": 3.606949534163206e-05,
      "loss": 2.0736,
      "step": 194069
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.420324444770813,
      "learning_rate": 3.6067550702187884e-05,
      "loss": 2.0628,
      "step": 194070
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.5816261768341064,
      "learning_rate": 3.606560611181366e-05,
      "loss": 1.9152,
      "step": 194071
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.3742765188217163,
      "learning_rate": 3.6063661570509885e-05,
      "loss": 2.1138,
      "step": 194072
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.352705717086792,
      "learning_rate": 3.606171707827685e-05,
      "loss": 2.0162,
      "step": 194073
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.523476004600525,
      "learning_rate": 3.605977263511494e-05,
      "loss": 1.7888,
      "step": 194074
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.4962244033813477,
      "learning_rate": 3.6057828241024434e-05,
      "loss": 1.8514,
      "step": 194075
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.3253681659698486,
      "learning_rate": 3.6055883896005846e-05,
      "loss": 1.8706,
      "step": 194076
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.4343316555023193,
      "learning_rate": 3.605393960005933e-05,
      "loss": 2.2734,
      "step": 194077
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.4233577251434326,
      "learning_rate": 3.6051995353185505e-05,
      "loss": 2.1318,
      "step": 194078
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.3548810482025146,
      "learning_rate": 3.605005115538455e-05,
      "loss": 1.9766,
      "step": 194079
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.2999886274337769,
      "learning_rate": 3.6048107006656876e-05,
      "loss": 2.1107,
      "step": 194080
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.4254668951034546,
      "learning_rate": 3.604616290700278e-05,
      "loss": 1.9575,
      "step": 194081
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.3399819135665894,
      "learning_rate": 3.604421885642277e-05,
      "loss": 1.8936,
      "step": 194082
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.3587586879730225,
      "learning_rate": 3.604227485491703e-05,
      "loss": 2.094,
      "step": 194083
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.403965950012207,
      "learning_rate": 3.604033090248607e-05,
      "loss": 2.1463,
      "step": 194084
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.5435453653335571,
      "learning_rate": 3.603838699913022e-05,
      "loss": 1.89,
      "step": 194085
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.4740314483642578,
      "learning_rate": 3.6036443144849746e-05,
      "loss": 1.9658,
      "step": 194086
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.428374171257019,
      "learning_rate": 3.6034499339645154e-05,
      "loss": 2.2482,
      "step": 194087
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.4185104370117188,
      "learning_rate": 3.603255558351673e-05,
      "loss": 1.9147,
      "step": 194088
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.4923617839813232,
      "learning_rate": 3.603061187646476e-05,
      "loss": 1.9182,
      "step": 194089
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.5332973003387451,
      "learning_rate": 3.602866821848975e-05,
      "loss": 1.709,
      "step": 194090
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.4049079418182373,
      "learning_rate": 3.6026724609592e-05,
      "loss": 2.0293,
      "step": 194091
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.532047986984253,
      "learning_rate": 3.602478104977178e-05,
      "loss": 2.0086,
      "step": 194092
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.4398075342178345,
      "learning_rate": 3.60228375390296e-05,
      "loss": 1.9607,
      "step": 194093
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.536149263381958,
      "learning_rate": 3.6020894077365736e-05,
      "loss": 1.9802,
      "step": 194094
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.4072364568710327,
      "learning_rate": 3.6018950664780614e-05,
      "loss": 1.9154,
      "step": 194095
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.4112101793289185,
      "learning_rate": 3.6017007301274525e-05,
      "loss": 2.0751,
      "step": 194096
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.388368844985962,
      "learning_rate": 3.6015063986847884e-05,
      "loss": 2.0035,
      "step": 194097
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.4065810441970825,
      "learning_rate": 3.601312072150094e-05,
      "loss": 1.8949,
      "step": 194098
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.433377981185913,
      "learning_rate": 3.601117750523421e-05,
      "loss": 2.0016,
      "step": 194099
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.417483925819397,
      "learning_rate": 3.6009234338047956e-05,
      "loss": 1.9565,
      "step": 194100
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.632988691329956,
      "learning_rate": 3.60072912199426e-05,
      "loss": 2.2834,
      "step": 194101
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.4837933778762817,
      "learning_rate": 3.6005348150918486e-05,
      "loss": 2.2849,
      "step": 194102
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.4566692113876343,
      "learning_rate": 3.600340513097595e-05,
      "loss": 1.9269,
      "step": 194103
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.348166584968567,
      "learning_rate": 3.6001462160115316e-05,
      "loss": 1.8555,
      "step": 194104
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.3327406644821167,
      "learning_rate": 3.599951923833702e-05,
      "loss": 1.7368,
      "step": 194105
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.4155341386795044,
      "learning_rate": 3.599757636564136e-05,
      "loss": 2.1611,
      "step": 194106
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.4131743907928467,
      "learning_rate": 3.59956335420288e-05,
      "loss": 2.0104,
      "step": 194107
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.3838865756988525,
      "learning_rate": 3.599369076749958e-05,
      "loss": 1.749,
      "step": 194108
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.5087333917617798,
      "learning_rate": 3.599174804205424e-05,
      "loss": 1.9927,
      "step": 194109
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.4388329982757568,
      "learning_rate": 3.598980536569286e-05,
      "loss": 1.9535,
      "step": 194110
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.3488624095916748,
      "learning_rate": 3.598786273841605e-05,
      "loss": 1.9461,
      "step": 194111
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.5382435321807861,
      "learning_rate": 3.598592016022401e-05,
      "loss": 1.9894,
      "step": 194112
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.3312392234802246,
      "learning_rate": 3.5983977631117246e-05,
      "loss": 2.0118,
      "step": 194113
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.4270758628845215,
      "learning_rate": 3.5982035151095944e-05,
      "loss": 1.9778,
      "step": 194114
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.3975872993469238,
      "learning_rate": 3.598009272016075e-05,
      "loss": 2.0581,
      "step": 194115
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.5404912233352661,
      "learning_rate": 3.597815033831165e-05,
      "loss": 2.0426,
      "step": 194116
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.5105830430984497,
      "learning_rate": 3.597620800554929e-05,
      "loss": 2.0982,
      "step": 194117
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.568244457244873,
      "learning_rate": 3.59742657218739e-05,
      "loss": 2.0438,
      "step": 194118
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.3910248279571533,
      "learning_rate": 3.59723234872859e-05,
      "loss": 2.1592,
      "step": 194119
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.4562804698944092,
      "learning_rate": 3.597038130178558e-05,
      "loss": 1.8465,
      "step": 194120
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.2872183322906494,
      "learning_rate": 3.5968439165373486e-05,
      "loss": 1.8932,
      "step": 194121
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.3839459419250488,
      "learning_rate": 3.5966497078049696e-05,
      "loss": 2.2525,
      "step": 194122
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.3855795860290527,
      "learning_rate": 3.59645550398148e-05,
      "loss": 2.1229,
      "step": 194123
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.4342321157455444,
      "learning_rate": 3.596261305066901e-05,
      "loss": 2.1753,
      "step": 194124
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.3781107664108276,
      "learning_rate": 3.5960671110612816e-05,
      "loss": 2.0307,
      "step": 194125
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.4582027196884155,
      "learning_rate": 3.5958729219646456e-05,
      "loss": 1.9867,
      "step": 194126
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.401308298110962,
      "learning_rate": 3.595678737777049e-05,
      "loss": 2.1403,
      "step": 194127
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.4257981777191162,
      "learning_rate": 3.5954845584984995e-05,
      "loss": 1.8488,
      "step": 194128
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.2892800569534302,
      "learning_rate": 3.5952903841290526e-05,
      "loss": 2.1834,
      "step": 194129
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.4334635734558105,
      "learning_rate": 3.5950962146687354e-05,
      "loss": 2.1173,
      "step": 194130
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.3943500518798828,
      "learning_rate": 3.594902050117595e-05,
      "loss": 1.9935,
      "step": 194131
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.434928059577942,
      "learning_rate": 3.594707890475654e-05,
      "loss": 2.1005,
      "step": 194132
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.4111348390579224,
      "learning_rate": 3.59451373574297e-05,
      "loss": 2.0228,
      "step": 194133
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.4071483612060547,
      "learning_rate": 3.5943195859195485e-05,
      "loss": 2.276,
      "step": 194134
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.464146614074707,
      "learning_rate": 3.59412544100545e-05,
      "loss": 1.8772,
      "step": 194135
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.3869444131851196,
      "learning_rate": 3.593931301000694e-05,
      "loss": 1.9199,
      "step": 194136
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.4776445627212524,
      "learning_rate": 3.593737165905332e-05,
      "loss": 2.1138,
      "step": 194137
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.3394871950149536,
      "learning_rate": 3.593543035719385e-05,
      "loss": 2.0888,
      "step": 194138
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.4349174499511719,
      "learning_rate": 3.5933489104429114e-05,
      "loss": 2.0293,
      "step": 194139
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.3812565803527832,
      "learning_rate": 3.593154790075917e-05,
      "loss": 2.017,
      "step": 194140
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.479292869567871,
      "learning_rate": 3.592960674618463e-05,
      "loss": 1.9771,
      "step": 194141
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.4464045763015747,
      "learning_rate": 3.5927665640705705e-05,
      "loss": 1.9058,
      "step": 194142
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.4639666080474854,
      "learning_rate": 3.5925724584322846e-05,
      "loss": 2.2374,
      "step": 194143
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.4803673028945923,
      "learning_rate": 3.592378357703635e-05,
      "loss": 1.7795,
      "step": 194144
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.4416966438293457,
      "learning_rate": 3.592184261884671e-05,
      "loss": 2.1214,
      "step": 194145
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.3260787725448608,
      "learning_rate": 3.591990170975403e-05,
      "loss": 2.0629,
      "step": 194146
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.3294677734375,
      "learning_rate": 3.5917960849758944e-05,
      "loss": 1.8872,
      "step": 194147
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.4260306358337402,
      "learning_rate": 3.591602003886158e-05,
      "loss": 1.9702,
      "step": 194148
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.4547910690307617,
      "learning_rate": 3.591407927706251e-05,
      "loss": 2.0631,
      "step": 194149
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.3677239418029785,
      "learning_rate": 3.5912138564361934e-05,
      "loss": 1.9003,
      "step": 194150
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.3667548894882202,
      "learning_rate": 3.591019790076038e-05,
      "loss": 1.8374,
      "step": 194151
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.4256396293640137,
      "learning_rate": 3.590825728625799e-05,
      "loss": 1.8575,
      "step": 194152
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.3265109062194824,
      "learning_rate": 3.5906316720855324e-05,
      "loss": 1.9409,
      "step": 194153
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.3531688451766968,
      "learning_rate": 3.590437620455258e-05,
      "loss": 2.101,
      "step": 194154
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.5352779626846313,
      "learning_rate": 3.590243573735026e-05,
      "loss": 2.0455,
      "step": 194155
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.2158236503601074,
      "learning_rate": 3.590049531924859e-05,
      "loss": 1.949,
      "step": 194156
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.5667928457260132,
      "learning_rate": 3.589855495024808e-05,
      "loss": 2.1986,
      "step": 194157
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.361337423324585,
      "learning_rate": 3.5896614630349e-05,
      "loss": 1.8961,
      "step": 194158
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.503783106803894,
      "learning_rate": 3.589467435955173e-05,
      "loss": 2.0079,
      "step": 194159
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.3594415187835693,
      "learning_rate": 3.589273413785656e-05,
      "loss": 1.8697,
      "step": 194160
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.3995004892349243,
      "learning_rate": 3.589079396526401e-05,
      "loss": 2.1423,
      "step": 194161
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.3623994588851929,
      "learning_rate": 3.588885384177425e-05,
      "loss": 1.9162,
      "step": 194162
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.3711960315704346,
      "learning_rate": 3.588691376738781e-05,
      "loss": 2.0698,
      "step": 194163
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.746434211730957,
      "learning_rate": 3.5884973742104985e-05,
      "loss": 1.9492,
      "step": 194164
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.5879179239273071,
      "learning_rate": 3.588303376592612e-05,
      "loss": 1.9918,
      "step": 194165
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.307165503501892,
      "learning_rate": 3.58810938388515e-05,
      "loss": 1.613,
      "step": 194166
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.4398870468139648,
      "learning_rate": 3.587915396088168e-05,
      "loss": 2.124,
      "step": 194167
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.3124539852142334,
      "learning_rate": 3.587721413201681e-05,
      "loss": 2.0042,
      "step": 194168
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.658775806427002,
      "learning_rate": 3.587527435225742e-05,
      "loss": 1.9267,
      "step": 194169
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.5225772857666016,
      "learning_rate": 3.587333462160382e-05,
      "loss": 2.0367,
      "step": 194170
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.5734519958496094,
      "learning_rate": 3.5871394940056274e-05,
      "loss": 2.004,
      "step": 194171
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.6007089614868164,
      "learning_rate": 3.586945530761531e-05,
      "loss": 2.136,
      "step": 194172
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.3768202066421509,
      "learning_rate": 3.586751572428117e-05,
      "loss": 2.1633,
      "step": 194173
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.4751492738723755,
      "learning_rate": 3.5865576190054216e-05,
      "loss": 2.0137,
      "step": 194174
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.3588817119598389,
      "learning_rate": 3.5863636704934875e-05,
      "loss": 1.9413,
      "step": 194175
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.3381998538970947,
      "learning_rate": 3.586169726892345e-05,
      "loss": 1.9875,
      "step": 194176
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.3459343910217285,
      "learning_rate": 3.585975788202028e-05,
      "loss": 1.7032,
      "step": 194177
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.3304646015167236,
      "learning_rate": 3.5857818544225855e-05,
      "loss": 2.1271,
      "step": 194178
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.4644228219985962,
      "learning_rate": 3.585587925554035e-05,
      "loss": 1.9137,
      "step": 194179
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.3546578884124756,
      "learning_rate": 3.58539400159643e-05,
      "loss": 2.1107,
      "step": 194180
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.589066743850708,
      "learning_rate": 3.5852000825498e-05,
      "loss": 2.178,
      "step": 194181
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.2579134702682495,
      "learning_rate": 3.585006168414177e-05,
      "loss": 2.1414,
      "step": 194182
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.386249303817749,
      "learning_rate": 3.584812259189593e-05,
      "loss": 1.8982,
      "step": 194183
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.4296414852142334,
      "learning_rate": 3.584618354876101e-05,
      "loss": 1.9404,
      "step": 194184
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.3797814846038818,
      "learning_rate": 3.5844244554737204e-05,
      "loss": 2.1609,
      "step": 194185
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.5028494596481323,
      "learning_rate": 3.584230560982497e-05,
      "loss": 2.1286,
      "step": 194186
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.3091449737548828,
      "learning_rate": 3.584036671402466e-05,
      "loss": 1.9011,
      "step": 194187
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.4548017978668213,
      "learning_rate": 3.5838427867336625e-05,
      "loss": 2.1326,
      "step": 194188
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.508233666419983,
      "learning_rate": 3.5836489069761144e-05,
      "loss": 2.116,
      "step": 194189
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.5276553630828857,
      "learning_rate": 3.583455032129867e-05,
      "loss": 1.7905,
      "step": 194190
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.3487292528152466,
      "learning_rate": 3.583261162194952e-05,
      "loss": 1.8922,
      "step": 194191
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.4670432806015015,
      "learning_rate": 3.5830672971714136e-05,
      "loss": 1.8269,
      "step": 194192
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.441398024559021,
      "learning_rate": 3.5828734370592735e-05,
      "loss": 2.1732,
      "step": 194193
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.311574935913086,
      "learning_rate": 3.582679581858589e-05,
      "loss": 2.1815,
      "step": 194194
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.4328019618988037,
      "learning_rate": 3.582485731569371e-05,
      "loss": 1.9849,
      "step": 194195
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.3720242977142334,
      "learning_rate": 3.5822918861916713e-05,
      "loss": 1.8784,
      "step": 194196
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.5949984788894653,
      "learning_rate": 3.58209804572552e-05,
      "loss": 1.9672,
      "step": 194197
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.448691964149475,
      "learning_rate": 3.5819042101709594e-05,
      "loss": 1.9427,
      "step": 194198
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.3411110639572144,
      "learning_rate": 3.5817103795280164e-05,
      "loss": 1.7088,
      "step": 194199
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.42630934715271,
      "learning_rate": 3.5815165537967416e-05,
      "loss": 1.9353,
      "step": 194200
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.8436509370803833,
      "learning_rate": 3.581322732977154e-05,
      "loss": 1.9621,
      "step": 194201
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.4209198951721191,
      "learning_rate": 3.581128917069298e-05,
      "loss": 1.7868,
      "step": 194202
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.3948168754577637,
      "learning_rate": 3.580935106073206e-05,
      "loss": 1.9875,
      "step": 194203
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.4579830169677734,
      "learning_rate": 3.5807412999889216e-05,
      "loss": 1.9205,
      "step": 194204
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.459122896194458,
      "learning_rate": 3.580547498816472e-05,
      "loss": 2.0057,
      "step": 194205
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.4692747592926025,
      "learning_rate": 3.5803537025559095e-05,
      "loss": 2.1356,
      "step": 194206
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.3911221027374268,
      "learning_rate": 3.5801599112072444e-05,
      "loss": 1.9953,
      "step": 194207
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.3695499897003174,
      "learning_rate": 3.57996612477053e-05,
      "loss": 2.1981,
      "step": 194208
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.3360482454299927,
      "learning_rate": 3.579772343245797e-05,
      "loss": 2.0072,
      "step": 194209
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.4583500623703003,
      "learning_rate": 3.579578566633088e-05,
      "loss": 1.9698,
      "step": 194210
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.388572096824646,
      "learning_rate": 3.579384794932426e-05,
      "loss": 2.0379,
      "step": 194211
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.3987830877304077,
      "learning_rate": 3.579191028143868e-05,
      "loss": 2.0041,
      "step": 194212
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.2874171733856201,
      "learning_rate": 3.5789972662674246e-05,
      "loss": 1.7925,
      "step": 194213
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.3145703077316284,
      "learning_rate": 3.578803509303152e-05,
      "loss": 1.8675,
      "step": 194214
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.460404872894287,
      "learning_rate": 3.578609757251073e-05,
      "loss": 2.0196,
      "step": 194215
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.4148187637329102,
      "learning_rate": 3.5784160101112314e-05,
      "loss": 2.0474,
      "step": 194216
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.462283730506897,
      "learning_rate": 3.57822226788366e-05,
      "loss": 2.153,
      "step": 194217
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.4463502168655396,
      "learning_rate": 3.578028530568403e-05,
      "loss": 1.9433,
      "step": 194218
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.4987140893936157,
      "learning_rate": 3.57783479816548e-05,
      "loss": 1.865,
      "step": 194219
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.4314374923706055,
      "learning_rate": 3.577641070674944e-05,
      "loss": 2.0004,
      "step": 194220
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.2473801374435425,
      "learning_rate": 3.5774473480968144e-05,
      "loss": 2.0114,
      "step": 194221
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.5067347288131714,
      "learning_rate": 3.577253630431143e-05,
      "loss": 2.0927,
      "step": 194222
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.4480814933776855,
      "learning_rate": 3.577059917677951e-05,
      "loss": 2.0183,
      "step": 194223
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.4226137399673462,
      "learning_rate": 3.576866209837297e-05,
      "loss": 2.0796,
      "step": 194224
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.3936375379562378,
      "learning_rate": 3.5766725069091894e-05,
      "loss": 1.7897,
      "step": 194225
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.3612208366394043,
      "learning_rate": 3.576478808893683e-05,
      "loss": 2.0476,
      "step": 194226
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.4229576587677002,
      "learning_rate": 3.5762851157908024e-05,
      "loss": 1.999,
      "step": 194227
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.3475669622421265,
      "learning_rate": 3.576091427600596e-05,
      "loss": 2.0606,
      "step": 194228
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.4107327461242676,
      "learning_rate": 3.5758977443230865e-05,
      "loss": 1.8747,
      "step": 194229
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.5381574630737305,
      "learning_rate": 3.575704065958327e-05,
      "loss": 2.0251,
      "step": 194230
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.2844603061676025,
      "learning_rate": 3.575510392506331e-05,
      "loss": 1.7685,
      "step": 194231
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.3328067064285278,
      "learning_rate": 3.575316723967152e-05,
      "loss": 2.0494,
      "step": 194232
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.2278544902801514,
      "learning_rate": 3.5751230603408165e-05,
      "loss": 2.0799,
      "step": 194233
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.3828951120376587,
      "learning_rate": 3.574929401627371e-05,
      "loss": 1.9959,
      "step": 194234
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.4544700384140015,
      "learning_rate": 3.5747357478268356e-05,
      "loss": 1.9541,
      "step": 194235
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.5249032974243164,
      "learning_rate": 3.5745420989392704e-05,
      "loss": 1.6551,
      "step": 194236
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.3906288146972656,
      "learning_rate": 3.574348454964685e-05,
      "loss": 2.0143,
      "step": 194237
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.4302884340286255,
      "learning_rate": 3.57415481590313e-05,
      "loss": 1.7937,
      "step": 194238
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.6194196939468384,
      "learning_rate": 3.5739611817546344e-05,
      "loss": 1.8082,
      "step": 194239
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.380231499671936,
      "learning_rate": 3.573767552519245e-05,
      "loss": 1.9905,
      "step": 194240
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.4978340864181519,
      "learning_rate": 3.573573928196983e-05,
      "loss": 2.1426,
      "step": 194241
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.3078770637512207,
      "learning_rate": 3.573380308787901e-05,
      "loss": 1.918,
      "step": 194242
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.474868893623352,
      "learning_rate": 3.573186694292025e-05,
      "loss": 2.1011,
      "step": 194243
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.6643860340118408,
      "learning_rate": 3.572993084709393e-05,
      "loss": 2.0528,
      "step": 194244
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.6794317960739136,
      "learning_rate": 3.5727994800400325e-05,
      "loss": 1.9041,
      "step": 194245
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.392076849937439,
      "learning_rate": 3.572605880283993e-05,
      "loss": 1.9479,
      "step": 194246
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.4167829751968384,
      "learning_rate": 3.572412285441303e-05,
      "loss": 2.203,
      "step": 194247
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.4022305011749268,
      "learning_rate": 3.5722186955120026e-05,
      "loss": 1.7063,
      "step": 194248
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.4726512432098389,
      "learning_rate": 3.5720251104961284e-05,
      "loss": 1.8609,
      "step": 194249
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.375510334968567,
      "learning_rate": 3.5718315303937106e-05,
      "loss": 1.9748,
      "step": 194250
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.637869119644165,
      "learning_rate": 3.571637955204786e-05,
      "loss": 2.0166,
      "step": 194251
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.4206892251968384,
      "learning_rate": 3.571444384929397e-05,
      "loss": 2.1133,
      "step": 194252
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.406976580619812,
      "learning_rate": 3.571250819567568e-05,
      "loss": 2.1594,
      "step": 194253
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.390007495880127,
      "learning_rate": 3.571057259119349e-05,
      "loss": 1.8992,
      "step": 194254
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.395901083946228,
      "learning_rate": 3.570863703584772e-05,
      "loss": 1.7699,
      "step": 194255
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.2964388132095337,
      "learning_rate": 3.5706701529638613e-05,
      "loss": 2.0622,
      "step": 194256
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.3836952447891235,
      "learning_rate": 3.570476607256667e-05,
      "loss": 2.1322,
      "step": 194257
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.3837034702301025,
      "learning_rate": 3.5702830664632224e-05,
      "loss": 1.8489,
      "step": 194258
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.4596081972122192,
      "learning_rate": 3.570089530583553e-05,
      "loss": 2.0876,
      "step": 194259
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.517195701599121,
      "learning_rate": 3.569895999617713e-05,
      "loss": 2.1257,
      "step": 194260
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.444933295249939,
      "learning_rate": 3.569702473565723e-05,
      "loss": 1.936,
      "step": 194261
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.3680905103683472,
      "learning_rate": 3.569508952427622e-05,
      "loss": 2.0621,
      "step": 194262
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.412501573562622,
      "learning_rate": 3.569315436203454e-05,
      "loss": 2.0489,
      "step": 194263
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.4460861682891846,
      "learning_rate": 3.569121924893248e-05,
      "loss": 1.9074,
      "step": 194264
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.7083362340927124,
      "learning_rate": 3.5689284184970346e-05,
      "loss": 1.9796,
      "step": 194265
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.4979995489120483,
      "learning_rate": 3.568734917014864e-05,
      "loss": 1.9529,
      "step": 194266
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.5210609436035156,
      "learning_rate": 3.568541420446762e-05,
      "loss": 1.9346,
      "step": 194267
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.5609312057495117,
      "learning_rate": 3.568347928792763e-05,
      "loss": 2.0188,
      "step": 194268
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.5386401414871216,
      "learning_rate": 3.568154442052913e-05,
      "loss": 2.1176,
      "step": 194269
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.3257468938827515,
      "learning_rate": 3.567960960227235e-05,
      "loss": 1.9721,
      "step": 194270
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.3422578573226929,
      "learning_rate": 3.5677674833157797e-05,
      "loss": 1.975,
      "step": 194271
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.6328418254852295,
      "learning_rate": 3.567574011318574e-05,
      "loss": 1.883,
      "step": 194272
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.3430334329605103,
      "learning_rate": 3.5673805442356564e-05,
      "loss": 1.6331,
      "step": 194273
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.319576621055603,
      "learning_rate": 3.567187082067054e-05,
      "loss": 2.0111,
      "step": 194274
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.5240947008132935,
      "learning_rate": 3.5669936248128186e-05,
      "loss": 2.0129,
      "step": 194275
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.3818919658660889,
      "learning_rate": 3.566800172472968e-05,
      "loss": 2.0392,
      "step": 194276
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.3838393688201904,
      "learning_rate": 3.5666067250475595e-05,
      "loss": 1.986,
      "step": 194277
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.5257182121276855,
      "learning_rate": 3.5664132825366065e-05,
      "loss": 1.9446,
      "step": 194278
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.3749130964279175,
      "learning_rate": 3.5662198449401726e-05,
      "loss": 2.1982,
      "step": 194279
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.6035510301589966,
      "learning_rate": 3.566026412258264e-05,
      "loss": 2.0411,
      "step": 194280
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.4682482481002808,
      "learning_rate": 3.565832984490934e-05,
      "loss": 1.9534,
      "step": 194281
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.5539659261703491,
      "learning_rate": 3.565639561638213e-05,
      "loss": 1.9867,
      "step": 194282
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.3742039203643799,
      "learning_rate": 3.56544614370014e-05,
      "loss": 1.9117,
      "step": 194283
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.40043306350708,
      "learning_rate": 3.565252730676746e-05,
      "loss": 1.8644,
      "step": 194284
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.609333872795105,
      "learning_rate": 3.5650593225680844e-05,
      "loss": 1.9172,
      "step": 194285
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.5063735246658325,
      "learning_rate": 3.564865919374161e-05,
      "loss": 2.0227,
      "step": 194286
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.5081020593643188,
      "learning_rate": 3.5646725210950364e-05,
      "loss": 2.0192,
      "step": 194287
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.510349154472351,
      "learning_rate": 3.564479127730731e-05,
      "loss": 1.8988,
      "step": 194288
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.4409040212631226,
      "learning_rate": 3.5642857392812965e-05,
      "loss": 1.8513,
      "step": 194289
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.2927418947219849,
      "learning_rate": 3.5640923557467506e-05,
      "loss": 1.7959,
      "step": 194290
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.6891111135482788,
      "learning_rate": 3.563898977127156e-05,
      "loss": 2.0522,
      "step": 194291
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.5476977825164795,
      "learning_rate": 3.563705603422514e-05,
      "loss": 2.1289,
      "step": 194292
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.4770004749298096,
      "learning_rate": 3.563512234632887e-05,
      "loss": 1.9382,
      "step": 194293
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.4985255002975464,
      "learning_rate": 3.563318870758298e-05,
      "loss": 2.1561,
      "step": 194294
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.3775150775909424,
      "learning_rate": 3.563125511798791e-05,
      "loss": 1.966,
      "step": 194295
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.4827258586883545,
      "learning_rate": 3.562932157754389e-05,
      "loss": 2.0733,
      "step": 194296
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.441432237625122,
      "learning_rate": 3.562738808625155e-05,
      "loss": 2.0948,
      "step": 194297
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.3764621019363403,
      "learning_rate": 3.56254546441109e-05,
      "loss": 1.9866,
      "step": 194298
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.4331010580062866,
      "learning_rate": 3.5623521251122554e-05,
      "loss": 2.0586,
      "step": 194299
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.4215574264526367,
      "learning_rate": 3.56215879072867e-05,
      "loss": 1.978,
      "step": 194300
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.4101877212524414,
      "learning_rate": 3.561965461260389e-05,
      "loss": 2.1029,
      "step": 194301
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.353490948677063,
      "learning_rate": 3.561772136707426e-05,
      "loss": 1.8387,
      "step": 194302
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.6540186405181885,
      "learning_rate": 3.561578817069844e-05,
      "loss": 1.9658,
      "step": 194303
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.4259016513824463,
      "learning_rate": 3.561385502347651e-05,
      "loss": 1.9008,
      "step": 194304
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.6298859119415283,
      "learning_rate": 3.561192192540903e-05,
      "loss": 2.0216,
      "step": 194305
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.3507499694824219,
      "learning_rate": 3.560998887649619e-05,
      "loss": 2.0754,
      "step": 194306
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.4312987327575684,
      "learning_rate": 3.560805587673853e-05,
      "loss": 1.7949,
      "step": 194307
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.3305786848068237,
      "learning_rate": 3.560612292613625e-05,
      "loss": 1.9078,
      "step": 194308
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.447143316268921,
      "learning_rate": 3.560419002468985e-05,
      "loss": 2.2388,
      "step": 194309
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.480336308479309,
      "learning_rate": 3.560225717239963e-05,
      "loss": 1.9614,
      "step": 194310
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.2163629531860352,
      "learning_rate": 3.560032436926592e-05,
      "loss": 2.2117,
      "step": 194311
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.3676968812942505,
      "learning_rate": 3.559839161528906e-05,
      "loss": 2.0969,
      "step": 194312
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.3726481199264526,
      "learning_rate": 3.5596458910469503e-05,
      "loss": 1.9877,
      "step": 194313
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.416494607925415,
      "learning_rate": 3.55945262548075e-05,
      "loss": 2.1427,
      "step": 194314
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.3817349672317505,
      "learning_rate": 3.55925936483035e-05,
      "loss": 1.9885,
      "step": 194315
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.439100980758667,
      "learning_rate": 3.559066109095785e-05,
      "loss": 2.1167,
      "step": 194316
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.4114793539047241,
      "learning_rate": 3.558872858277091e-05,
      "loss": 2.0395,
      "step": 194317
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.272506594657898,
      "learning_rate": 3.558679612374291e-05,
      "loss": 1.7899,
      "step": 194318
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.3463388681411743,
      "learning_rate": 3.558486371387439e-05,
      "loss": 2.2876,
      "step": 194319
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.2687935829162598,
      "learning_rate": 3.558293135316561e-05,
      "loss": 2.0313,
      "step": 194320
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.4852312803268433,
      "learning_rate": 3.558099904161697e-05,
      "loss": 2.1194,
      "step": 194321
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.3307933807373047,
      "learning_rate": 3.557906677922885e-05,
      "loss": 2.2217,
      "step": 194322
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.3539714813232422,
      "learning_rate": 3.557713456600153e-05,
      "loss": 1.9364,
      "step": 194323
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.47822904586792,
      "learning_rate": 3.5575202401935385e-05,
      "loss": 1.8528,
      "step": 194324
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.3707460165023804,
      "learning_rate": 3.5573270287030855e-05,
      "loss": 1.9894,
      "step": 194325
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.3708642721176147,
      "learning_rate": 3.557133822128819e-05,
      "loss": 2.0087,
      "step": 194326
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.3745163679122925,
      "learning_rate": 3.556940620470784e-05,
      "loss": 1.8233,
      "step": 194327
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.3018372058868408,
      "learning_rate": 3.556747423729016e-05,
      "loss": 1.9939,
      "step": 194328
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.3803938627243042,
      "learning_rate": 3.556554231903545e-05,
      "loss": 1.9329,
      "step": 194329
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.2978037595748901,
      "learning_rate": 3.556361044994406e-05,
      "loss": 1.978,
      "step": 194330
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.3759130239486694,
      "learning_rate": 3.5561678630016433e-05,
      "loss": 1.9463,
      "step": 194331
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.5567498207092285,
      "learning_rate": 3.555974685925281e-05,
      "loss": 2.035,
      "step": 194332
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.3510282039642334,
      "learning_rate": 3.55578151376537e-05,
      "loss": 1.9925,
      "step": 194333
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.5067144632339478,
      "learning_rate": 3.555588346521939e-05,
      "loss": 1.8644,
      "step": 194334
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.3092087507247925,
      "learning_rate": 3.5553951841950226e-05,
      "loss": 1.9112,
      "step": 194335
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.5484594106674194,
      "learning_rate": 3.5552020267846525e-05,
      "loss": 1.8505,
      "step": 194336
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.2890737056732178,
      "learning_rate": 3.5550088742908736e-05,
      "loss": 2.0033,
      "step": 194337
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.432198166847229,
      "learning_rate": 3.554815726713711e-05,
      "loss": 2.1109,
      "step": 194338
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.4458768367767334,
      "learning_rate": 3.554622584053216e-05,
      "loss": 1.9648,
      "step": 194339
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.4837323427200317,
      "learning_rate": 3.554429446309415e-05,
      "loss": 1.8281,
      "step": 194340
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.3816593885421753,
      "learning_rate": 3.554236313482337e-05,
      "loss": 2.0705,
      "step": 194341
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.4531127214431763,
      "learning_rate": 3.554043185572036e-05,
      "loss": 1.8607,
      "step": 194342
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.4239795207977295,
      "learning_rate": 3.5538500625785324e-05,
      "loss": 1.9459,
      "step": 194343
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.3809553384780884,
      "learning_rate": 3.553656944501866e-05,
      "loss": 2.0295,
      "step": 194344
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.4495481252670288,
      "learning_rate": 3.553463831342076e-05,
      "loss": 1.7967,
      "step": 194345
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.4266517162322998,
      "learning_rate": 3.5532707230991996e-05,
      "loss": 2.0256,
      "step": 194346
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.4167730808258057,
      "learning_rate": 3.55307761977326e-05,
      "loss": 1.9423,
      "step": 194347
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.412125587463379,
      "learning_rate": 3.552884521364311e-05,
      "loss": 1.828,
      "step": 194348
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.2633328437805176,
      "learning_rate": 3.5526914278723786e-05,
      "loss": 1.8984,
      "step": 194349
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.480492353439331,
      "learning_rate": 3.552498339297496e-05,
      "loss": 1.9242,
      "step": 194350
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.5554496049880981,
      "learning_rate": 3.552305255639707e-05,
      "loss": 1.9869,
      "step": 194351
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.339575171470642,
      "learning_rate": 3.552112176899045e-05,
      "loss": 2.1394,
      "step": 194352
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.4558143615722656,
      "learning_rate": 3.551919103075536e-05,
      "loss": 2.0644,
      "step": 194353
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.3441663980484009,
      "learning_rate": 3.5517260341692336e-05,
      "loss": 1.9943,
      "step": 194354
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.277740478515625,
      "learning_rate": 3.5515329701801574e-05,
      "loss": 1.9803,
      "step": 194355
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.507737398147583,
      "learning_rate": 3.551339911108355e-05,
      "loss": 2.2565,
      "step": 194356
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.354124665260315,
      "learning_rate": 3.5511468569538624e-05,
      "loss": 1.811,
      "step": 194357
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.4786559343338013,
      "learning_rate": 3.5509538077167066e-05,
      "loss": 2.2111,
      "step": 194358
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.5771220922470093,
      "learning_rate": 3.5507607633969234e-05,
      "loss": 1.7958,
      "step": 194359
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.5625394582748413,
      "learning_rate": 3.5505677239945574e-05,
      "loss": 1.8792,
      "step": 194360
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.4529904127120972,
      "learning_rate": 3.550374689509638e-05,
      "loss": 2.0281,
      "step": 194361
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.3860385417938232,
      "learning_rate": 3.550181659942204e-05,
      "loss": 1.8939,
      "step": 194362
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.535492181777954,
      "learning_rate": 3.5499886352922936e-05,
      "loss": 2.0209,
      "step": 194363
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.5089031457901,
      "learning_rate": 3.5497956155599394e-05,
      "loss": 2.0594,
      "step": 194364
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.41788649559021,
      "learning_rate": 3.549602600745172e-05,
      "loss": 2.0988,
      "step": 194365
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.4995849132537842,
      "learning_rate": 3.549409590848037e-05,
      "loss": 1.9454,
      "step": 194366
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.3312393426895142,
      "learning_rate": 3.5492165858685616e-05,
      "loss": 1.8843,
      "step": 194367
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.456233024597168,
      "learning_rate": 3.549023585806793e-05,
      "loss": 1.7633,
      "step": 194368
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.5643755197525024,
      "learning_rate": 3.54883059066275e-05,
      "loss": 2.0256,
      "step": 194369
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.397829532623291,
      "learning_rate": 3.548637600436494e-05,
      "loss": 1.938,
      "step": 194370
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.567793607711792,
      "learning_rate": 3.548444615128033e-05,
      "loss": 2.0024,
      "step": 194371
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.3786391019821167,
      "learning_rate": 3.5482516347374224e-05,
      "loss": 2.1601,
      "step": 194372
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.3606079816818237,
      "learning_rate": 3.5480586592646844e-05,
      "loss": 1.7433,
      "step": 194373
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.3659709692001343,
      "learning_rate": 3.547865688709866e-05,
      "loss": 1.9939,
      "step": 194374
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.4450894594192505,
      "learning_rate": 3.5476727230729965e-05,
      "loss": 1.9802,
      "step": 194375
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.4056055545806885,
      "learning_rate": 3.547479762354116e-05,
      "loss": 1.9825,
      "step": 194376
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.4217313528060913,
      "learning_rate": 3.547286806553259e-05,
      "loss": 1.9875,
      "step": 194377
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.4204429388046265,
      "learning_rate": 3.547093855670464e-05,
      "loss": 1.8971,
      "step": 194378
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.6025359630584717,
      "learning_rate": 3.5469009097057555e-05,
      "loss": 1.8115,
      "step": 194379
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.5432734489440918,
      "learning_rate": 3.546707968659183e-05,
      "loss": 1.9652,
      "step": 194380
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.3484573364257812,
      "learning_rate": 3.546515032530769e-05,
      "loss": 2.086,
      "step": 194381
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.5414544343948364,
      "learning_rate": 3.546322101320568e-05,
      "loss": 2.036,
      "step": 194382
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.6394174098968506,
      "learning_rate": 3.546129175028599e-05,
      "loss": 2.0878,
      "step": 194383
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.445199966430664,
      "learning_rate": 3.5459362536549094e-05,
      "loss": 1.9953,
      "step": 194384
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.4048776626586914,
      "learning_rate": 3.545743337199523e-05,
      "loss": 2.0969,
      "step": 194385
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.405678153038025,
      "learning_rate": 3.5455504256624844e-05,
      "loss": 1.8499,
      "step": 194386
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.4616459608078003,
      "learning_rate": 3.5453575190438255e-05,
      "loss": 1.8909,
      "step": 194387
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.5045075416564941,
      "learning_rate": 3.5451646173435854e-05,
      "loss": 1.9175,
      "step": 194388
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.221694827079773,
      "learning_rate": 3.5449717205618014e-05,
      "loss": 1.8439,
      "step": 194389
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.2778416872024536,
      "learning_rate": 3.544778828698509e-05,
      "loss": 1.8736,
      "step": 194390
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.4390738010406494,
      "learning_rate": 3.544585941753732e-05,
      "loss": 1.8811,
      "step": 194391
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.3740304708480835,
      "learning_rate": 3.5443930597275217e-05,
      "loss": 1.9219,
      "step": 194392
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.5712095499038696,
      "learning_rate": 3.544200182619903e-05,
      "loss": 2.07,
      "step": 194393
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.4499534368515015,
      "learning_rate": 3.544007310430926e-05,
      "loss": 2.089,
      "step": 194394
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.4263123273849487,
      "learning_rate": 3.5438144431606156e-05,
      "loss": 1.9211,
      "step": 194395
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.3780614137649536,
      "learning_rate": 3.5436215808090064e-05,
      "loss": 2.033,
      "step": 194396
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.4156659841537476,
      "learning_rate": 3.5434287233761325e-05,
      "loss": 1.942,
      "step": 194397
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.4750723838806152,
      "learning_rate": 3.543235870862045e-05,
      "loss": 1.9121,
      "step": 194398
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.4134957790374756,
      "learning_rate": 3.5430430232667585e-05,
      "loss": 2.0425,
      "step": 194399
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.4453308582305908,
      "learning_rate": 3.5428501805903274e-05,
      "loss": 2.0564,
      "step": 194400
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.4976825714111328,
      "learning_rate": 3.542657342832779e-05,
      "loss": 2.0695,
      "step": 194401
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.4278253316879272,
      "learning_rate": 3.5424645099941515e-05,
      "loss": 1.8193,
      "step": 194402
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.546299695968628,
      "learning_rate": 3.54227168207447e-05,
      "loss": 1.9403,
      "step": 194403
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.6310375928878784,
      "learning_rate": 3.54207885907379e-05,
      "loss": 1.9081,
      "step": 194404
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.37099289894104,
      "learning_rate": 3.541886040992129e-05,
      "loss": 1.9728,
      "step": 194405
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.3110932111740112,
      "learning_rate": 3.541693227829536e-05,
      "loss": 2.0845,
      "step": 194406
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.5644845962524414,
      "learning_rate": 3.541500419586042e-05,
      "loss": 2.0477,
      "step": 194407
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.390018105506897,
      "learning_rate": 3.541307616261686e-05,
      "loss": 2.0667,
      "step": 194408
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.487153172492981,
      "learning_rate": 3.541114817856489e-05,
      "loss": 1.8522,
      "step": 194409
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.5004003047943115,
      "learning_rate": 3.540922024370506e-05,
      "loss": 1.8967,
      "step": 194410
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.3479797840118408,
      "learning_rate": 3.540729235803762e-05,
      "loss": 2.0816,
      "step": 194411
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.439818263053894,
      "learning_rate": 3.5405364521563e-05,
      "loss": 2.1727,
      "step": 194412
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.3991248607635498,
      "learning_rate": 3.540343673428149e-05,
      "loss": 1.8954,
      "step": 194413
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.43443763256073,
      "learning_rate": 3.540150899619353e-05,
      "loss": 1.8504,
      "step": 194414
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.2471437454223633,
      "learning_rate": 3.539958130729932e-05,
      "loss": 2.1752,
      "step": 194415
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.4138448238372803,
      "learning_rate": 3.539765366759939e-05,
      "loss": 2.0396,
      "step": 194416
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.575595498085022,
      "learning_rate": 3.539572607709398e-05,
      "loss": 2.1989,
      "step": 194417
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.5234085321426392,
      "learning_rate": 3.539379853578355e-05,
      "loss": 2.1062,
      "step": 194418
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.4091018438339233,
      "learning_rate": 3.539187104366843e-05,
      "loss": 2.0541,
      "step": 194419
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.3214175701141357,
      "learning_rate": 3.5389943600748925e-05,
      "loss": 1.8645,
      "step": 194420
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.4555824995040894,
      "learning_rate": 3.5388016207025374e-05,
      "loss": 1.7869,
      "step": 194421
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.3906270265579224,
      "learning_rate": 3.538608886249826e-05,
      "loss": 2.0484,
      "step": 194422
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.3276727199554443,
      "learning_rate": 3.53841615671678e-05,
      "loss": 2.1153,
      "step": 194423
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.4515496492385864,
      "learning_rate": 3.538223432103449e-05,
      "loss": 2.0419,
      "step": 194424
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.3662782907485962,
      "learning_rate": 3.5380307124098584e-05,
      "loss": 1.9077,
      "step": 194425
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.3678010702133179,
      "learning_rate": 3.5378379976360495e-05,
      "loss": 1.9498,
      "step": 194426
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.524775505065918,
      "learning_rate": 3.537645287782052e-05,
      "loss": 1.8686,
      "step": 194427
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.5120551586151123,
      "learning_rate": 3.5374525828479083e-05,
      "loss": 2.093,
      "step": 194428
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.401618480682373,
      "learning_rate": 3.5372598828336466e-05,
      "loss": 2.1185,
      "step": 194429
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.4678568840026855,
      "learning_rate": 3.537067187739315e-05,
      "loss": 2.0137,
      "step": 194430
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.4891074895858765,
      "learning_rate": 3.5368744975649424e-05,
      "loss": 2.116,
      "step": 194431
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.5629589557647705,
      "learning_rate": 3.5366818123105564e-05,
      "loss": 2.1308,
      "step": 194432
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.4341027736663818,
      "learning_rate": 3.536489131976209e-05,
      "loss": 1.8886,
      "step": 194433
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.6417587995529175,
      "learning_rate": 3.5362964565619256e-05,
      "loss": 2.0066,
      "step": 194434
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.629314661026001,
      "learning_rate": 3.53610378606774e-05,
      "loss": 1.9058,
      "step": 194435
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.6101340055465698,
      "learning_rate": 3.535911120493699e-05,
      "loss": 1.9797,
      "step": 194436
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.5269911289215088,
      "learning_rate": 3.535718459839828e-05,
      "loss": 2.0069,
      "step": 194437
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.3682113885879517,
      "learning_rate": 3.535525804106162e-05,
      "loss": 2.0091,
      "step": 194438
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.4275407791137695,
      "learning_rate": 3.53533315329275e-05,
      "loss": 1.9189,
      "step": 194439
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.41494619846344,
      "learning_rate": 3.535140507399609e-05,
      "loss": 2.1192,
      "step": 194440
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.5749292373657227,
      "learning_rate": 3.534947866426795e-05,
      "loss": 2.0144,
      "step": 194441
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.470766305923462,
      "learning_rate": 3.534755230374332e-05,
      "loss": 2.2401,
      "step": 194442
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.424275517463684,
      "learning_rate": 3.53456259924225e-05,
      "loss": 1.973,
      "step": 194443
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.4605737924575806,
      "learning_rate": 3.5343699730306016e-05,
      "loss": 2.1112,
      "step": 194444
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.4827592372894287,
      "learning_rate": 3.534177351739411e-05,
      "loss": 1.9967,
      "step": 194445
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.4976356029510498,
      "learning_rate": 3.533984735368711e-05,
      "loss": 2.0977,
      "step": 194446
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.5543925762176514,
      "learning_rate": 3.5337921239185486e-05,
      "loss": 2.045,
      "step": 194447
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.3667093515396118,
      "learning_rate": 3.533599517388953e-05,
      "loss": 2.1225,
      "step": 194448
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.4558707475662231,
      "learning_rate": 3.533406915779955e-05,
      "loss": 2.0994,
      "step": 194449
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.4050153493881226,
      "learning_rate": 3.533214319091604e-05,
      "loss": 1.8319,
      "step": 194450
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.4193871021270752,
      "learning_rate": 3.533021727323927e-05,
      "loss": 2.1133,
      "step": 194451
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.3587181568145752,
      "learning_rate": 3.532829140476957e-05,
      "loss": 2.1243,
      "step": 194452
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.6233996152877808,
      "learning_rate": 3.5326365585507376e-05,
      "loss": 2.1111,
      "step": 194453
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.4724127054214478,
      "learning_rate": 3.5324439815452954e-05,
      "loss": 1.9093,
      "step": 194454
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.5216424465179443,
      "learning_rate": 3.532251409460677e-05,
      "loss": 2.0074,
      "step": 194455
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.4898228645324707,
      "learning_rate": 3.5320588422969123e-05,
      "loss": 2.1237,
      "step": 194456
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.4425177574157715,
      "learning_rate": 3.531866280054037e-05,
      "loss": 1.9592,
      "step": 194457
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.7749990224838257,
      "learning_rate": 3.53167372273208e-05,
      "loss": 1.8153,
      "step": 194458
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.365033745765686,
      "learning_rate": 3.531481170331093e-05,
      "loss": 2.0302,
      "step": 194459
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.5488084554672241,
      "learning_rate": 3.5312886228511e-05,
      "loss": 2.1361,
      "step": 194460
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.3867512941360474,
      "learning_rate": 3.531096080292143e-05,
      "loss": 2.0651,
      "step": 194461
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.2314300537109375,
      "learning_rate": 3.530903542654253e-05,
      "loss": 1.9186,
      "step": 194462
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.396120309829712,
      "learning_rate": 3.5307110099374705e-05,
      "loss": 2.0062,
      "step": 194463
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.2040822505950928,
      "learning_rate": 3.530518482141821e-05,
      "loss": 1.8498,
      "step": 194464
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.4914673566818237,
      "learning_rate": 3.5303259592673547e-05,
      "loss": 1.9282,
      "step": 194465
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.3520252704620361,
      "learning_rate": 3.530133441314092e-05,
      "loss": 2.0923,
      "step": 194466
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.4361600875854492,
      "learning_rate": 3.5299409282820866e-05,
      "loss": 1.8216,
      "step": 194467
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.3595396280288696,
      "learning_rate": 3.529748420171364e-05,
      "loss": 1.9074,
      "step": 194468
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.3611336946487427,
      "learning_rate": 3.5295559169819585e-05,
      "loss": 1.9135,
      "step": 194469
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.3185632228851318,
      "learning_rate": 3.5293634187139065e-05,
      "loss": 1.9683,
      "step": 194470
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.393717646598816,
      "learning_rate": 3.5291709253672474e-05,
      "loss": 1.9426,
      "step": 194471
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.4017280340194702,
      "learning_rate": 3.5289784369420116e-05,
      "loss": 1.93,
      "step": 194472
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.3972049951553345,
      "learning_rate": 3.5287859534382427e-05,
      "loss": 1.9964,
      "step": 194473
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.4962716102600098,
      "learning_rate": 3.528593474855973e-05,
      "loss": 2.1873,
      "step": 194474
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.4228832721710205,
      "learning_rate": 3.528401001195237e-05,
      "loss": 1.9677,
      "step": 194475
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.348855972290039,
      "learning_rate": 3.528208532456067e-05,
      "loss": 2.0111,
      "step": 194476
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.4314333200454712,
      "learning_rate": 3.528016068638507e-05,
      "loss": 2.2128,
      "step": 194477
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.3747447729110718,
      "learning_rate": 3.5278236097425806e-05,
      "loss": 2.169,
      "step": 194478
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.4605070352554321,
      "learning_rate": 3.52763115576834e-05,
      "loss": 2.0415,
      "step": 194479
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.5653353929519653,
      "learning_rate": 3.5274387067158125e-05,
      "loss": 2.0648,
      "step": 194480
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.3779345750808716,
      "learning_rate": 3.527246262585035e-05,
      "loss": 2.1784,
      "step": 194481
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.310492992401123,
      "learning_rate": 3.5270538233760336e-05,
      "loss": 1.8553,
      "step": 194482
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.4440139532089233,
      "learning_rate": 3.5268613890888585e-05,
      "loss": 2.0072,
      "step": 194483
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.6285271644592285,
      "learning_rate": 3.526668959723533e-05,
      "loss": 1.969,
      "step": 194484
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.5090911388397217,
      "learning_rate": 3.526476535280107e-05,
      "loss": 2.055,
      "step": 194485
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.424993872642517,
      "learning_rate": 3.5262841157586105e-05,
      "loss": 1.7627,
      "step": 194486
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.347946286201477,
      "learning_rate": 3.526091701159077e-05,
      "loss": 1.9888,
      "step": 194487
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.3584709167480469,
      "learning_rate": 3.525899291481533e-05,
      "loss": 2.0225,
      "step": 194488
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.3685895204544067,
      "learning_rate": 3.525706886726035e-05,
      "loss": 2.0977,
      "step": 194489
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.306341528892517,
      "learning_rate": 3.5255144868926e-05,
      "loss": 1.9041,
      "step": 194490
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.4897494316101074,
      "learning_rate": 3.525322091981274e-05,
      "loss": 1.9986,
      "step": 194491
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.4083316326141357,
      "learning_rate": 3.5251297019920945e-05,
      "loss": 2.0054,
      "step": 194492
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.4200057983398438,
      "learning_rate": 3.524937316925094e-05,
      "loss": 1.9822,
      "step": 194493
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.4502671957015991,
      "learning_rate": 3.5247449367802995e-05,
      "loss": 2.1072,
      "step": 194494
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.3233271837234497,
      "learning_rate": 3.5245525615577605e-05,
      "loss": 1.9382,
      "step": 194495
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.6998838186264038,
      "learning_rate": 3.524360191257498e-05,
      "loss": 1.9458,
      "step": 194496
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.3600842952728271,
      "learning_rate": 3.524167825879568e-05,
      "loss": 2.0133,
      "step": 194497
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.4075263738632202,
      "learning_rate": 3.5239754654239936e-05,
      "loss": 1.8699,
      "step": 194498
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.4454774856567383,
      "learning_rate": 3.5237831098908086e-05,
      "loss": 2.125,
      "step": 194499
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.3375779390335083,
      "learning_rate": 3.523590759280049e-05,
      "loss": 1.7294,
      "step": 194500
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.3449305295944214,
      "learning_rate": 3.5233984135917595e-05,
      "loss": 1.8598,
      "step": 194501
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.388540267944336,
      "learning_rate": 3.523206072825965e-05,
      "loss": 1.778,
      "step": 194502
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.4867891073226929,
      "learning_rate": 3.5230137369827095e-05,
      "loss": 2.1631,
      "step": 194503
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.4711166620254517,
      "learning_rate": 3.5228214060620265e-05,
      "loss": 2.2015,
      "step": 194504
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.6100068092346191,
      "learning_rate": 3.522629080063952e-05,
      "loss": 2.0119,
      "step": 194505
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.4438674449920654,
      "learning_rate": 3.522436758988514e-05,
      "loss": 1.7815,
      "step": 194506
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.4513189792633057,
      "learning_rate": 3.522244442835761e-05,
      "loss": 2.1107,
      "step": 194507
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.4745702743530273,
      "learning_rate": 3.522052131605717e-05,
      "loss": 2.0741,
      "step": 194508
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.3971660137176514,
      "learning_rate": 3.521859825298429e-05,
      "loss": 2.1202,
      "step": 194509
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.3671197891235352,
      "learning_rate": 3.521667523913919e-05,
      "loss": 2.1879,
      "step": 194510
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.4960718154907227,
      "learning_rate": 3.521475227452245e-05,
      "loss": 1.9044,
      "step": 194511
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.4635652303695679,
      "learning_rate": 3.5212829359134165e-05,
      "loss": 2.0162,
      "step": 194512
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.41006338596344,
      "learning_rate": 3.5210906492974864e-05,
      "loss": 1.8634,
      "step": 194513
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.3629348278045654,
      "learning_rate": 3.520898367604482e-05,
      "loss": 1.8609,
      "step": 194514
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.4744229316711426,
      "learning_rate": 3.5207060908344466e-05,
      "loss": 1.9628,
      "step": 194515
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.4957815408706665,
      "learning_rate": 3.5205138189874026e-05,
      "loss": 2.299,
      "step": 194516
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.6030662059783936,
      "learning_rate": 3.520321552063404e-05,
      "loss": 1.9543,
      "step": 194517
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.469425082206726,
      "learning_rate": 3.520129290062478e-05,
      "loss": 2.3125,
      "step": 194518
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.4018709659576416,
      "learning_rate": 3.51993703298466e-05,
      "loss": 2.0067,
      "step": 194519
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.3938814401626587,
      "learning_rate": 3.5197447808299775e-05,
      "loss": 1.8175,
      "step": 194520
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.4087393283843994,
      "learning_rate": 3.519552533598483e-05,
      "loss": 1.9239,
      "step": 194521
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.4562854766845703,
      "learning_rate": 3.519360291290194e-05,
      "loss": 2.2025,
      "step": 194522
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.378806710243225,
      "learning_rate": 3.5191680539051635e-05,
      "loss": 2.054,
      "step": 194523
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.2989083528518677,
      "learning_rate": 3.5189758214434215e-05,
      "loss": 1.755,
      "step": 194524
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.453607439994812,
      "learning_rate": 3.518783593905001e-05,
      "loss": 2.1459,
      "step": 194525
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.529843807220459,
      "learning_rate": 3.51859137128993e-05,
      "loss": 2.3481,
      "step": 194526
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.4597187042236328,
      "learning_rate": 3.518399153598259e-05,
      "loss": 2.172,
      "step": 194527
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.3012399673461914,
      "learning_rate": 3.518206940830014e-05,
      "loss": 2.0507,
      "step": 194528
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.4091641902923584,
      "learning_rate": 3.5180147329852405e-05,
      "loss": 1.9582,
      "step": 194529
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.3935766220092773,
      "learning_rate": 3.517822530063965e-05,
      "loss": 2.0028,
      "step": 194530
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.455876350402832,
      "learning_rate": 3.517630332066221e-05,
      "loss": 2.0132,
      "step": 194531
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.4661610126495361,
      "learning_rate": 3.517438138992056e-05,
      "loss": 1.9469,
      "step": 194532
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.5544835329055786,
      "learning_rate": 3.5172459508415017e-05,
      "loss": 1.9385,
      "step": 194533
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.2204264402389526,
      "learning_rate": 3.5170537676145825e-05,
      "loss": 1.9938,
      "step": 194534
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.44045090675354,
      "learning_rate": 3.5168615893113486e-05,
      "loss": 2.1785,
      "step": 194535
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.5643197298049927,
      "learning_rate": 3.5166694159318285e-05,
      "loss": 2.1659,
      "step": 194536
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.51230788230896,
      "learning_rate": 3.5164772474760574e-05,
      "loss": 2.2014,
      "step": 194537
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.397330641746521,
      "learning_rate": 3.516285083944077e-05,
      "loss": 2.1134,
      "step": 194538
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.3229962587356567,
      "learning_rate": 3.516092925335912e-05,
      "loss": 2.0342,
      "step": 194539
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.3562626838684082,
      "learning_rate": 3.515900771651614e-05,
      "loss": 1.7709,
      "step": 194540
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.3923600912094116,
      "learning_rate": 3.515708622891208e-05,
      "loss": 2.1362,
      "step": 194541
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.4684782028198242,
      "learning_rate": 3.515516479054733e-05,
      "loss": 1.9263,
      "step": 194542
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.5957854986190796,
      "learning_rate": 3.515324340142216e-05,
      "loss": 1.9052,
      "step": 194543
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.3284964561462402,
      "learning_rate": 3.515132206153707e-05,
      "loss": 1.8996,
      "step": 194544
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.485906958580017,
      "learning_rate": 3.514940077089229e-05,
      "loss": 2.0469,
      "step": 194545
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.6172969341278076,
      "learning_rate": 3.514747952948829e-05,
      "loss": 2.0337,
      "step": 194546
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.3065980672836304,
      "learning_rate": 3.51455583373254e-05,
      "loss": 1.6452,
      "step": 194547
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.551579236984253,
      "learning_rate": 3.5143637194403915e-05,
      "loss": 2.0475,
      "step": 194548
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.3661394119262695,
      "learning_rate": 3.5141716100724184e-05,
      "loss": 2.0722,
      "step": 194549
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.570050835609436,
      "learning_rate": 3.513979505628666e-05,
      "loss": 2.053,
      "step": 194550
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.506616473197937,
      "learning_rate": 3.513787406109158e-05,
      "loss": 2.1703,
      "step": 194551
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.4150547981262207,
      "learning_rate": 3.513595311513944e-05,
      "loss": 1.9802,
      "step": 194552
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.4018043279647827,
      "learning_rate": 3.513403221843054e-05,
      "loss": 2.132,
      "step": 194553
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.5050030946731567,
      "learning_rate": 3.5132111370965186e-05,
      "loss": 1.9351,
      "step": 194554
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.3938030004501343,
      "learning_rate": 3.513019057274375e-05,
      "loss": 2.085,
      "step": 194555
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.364187240600586,
      "learning_rate": 3.512826982376664e-05,
      "loss": 1.9722,
      "step": 194556
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.4191884994506836,
      "learning_rate": 3.512634912403415e-05,
      "loss": 2.2889,
      "step": 194557
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.5270236730575562,
      "learning_rate": 3.512442847354673e-05,
      "loss": 2.1583,
      "step": 194558
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.5098538398742676,
      "learning_rate": 3.512250787230465e-05,
      "loss": 1.858,
      "step": 194559
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.6305197477340698,
      "learning_rate": 3.5120587320308316e-05,
      "loss": 1.954,
      "step": 194560
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.5591316223144531,
      "learning_rate": 3.511866681755802e-05,
      "loss": 1.8858,
      "step": 194561
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.3708635568618774,
      "learning_rate": 3.511674636405419e-05,
      "loss": 2.0389,
      "step": 194562
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.3065382242202759,
      "learning_rate": 3.511482595979714e-05,
      "loss": 1.949,
      "step": 194563
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.492521047592163,
      "learning_rate": 3.5112905604787265e-05,
      "loss": 1.9403,
      "step": 194564
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.4543930292129517,
      "learning_rate": 3.5110985299024926e-05,
      "loss": 2.129,
      "step": 194565
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.8000239133834839,
      "learning_rate": 3.510906504251043e-05,
      "loss": 1.986,
      "step": 194566
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.510071873664856,
      "learning_rate": 3.510714483524414e-05,
      "loss": 2.0248,
      "step": 194567
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.32981276512146,
      "learning_rate": 3.510522467722645e-05,
      "loss": 2.1099,
      "step": 194568
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.5515718460083008,
      "learning_rate": 3.5103304568457666e-05,
      "loss": 2.0,
      "step": 194569
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.4609395265579224,
      "learning_rate": 3.510138450893823e-05,
      "loss": 2.4904,
      "step": 194570
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.509667158126831,
      "learning_rate": 3.5099464498668415e-05,
      "loss": 1.7526,
      "step": 194571
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.8327351808547974,
      "learning_rate": 3.509754453764865e-05,
      "loss": 1.9056,
      "step": 194572
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.4397737979888916,
      "learning_rate": 3.509562462587918e-05,
      "loss": 1.9538,
      "step": 194573
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.3193154335021973,
      "learning_rate": 3.509370476336049e-05,
      "loss": 1.7799,
      "step": 194574
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.3492605686187744,
      "learning_rate": 3.5091784950092826e-05,
      "loss": 2.1309,
      "step": 194575
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.3701660633087158,
      "learning_rate": 3.5089865186076674e-05,
      "loss": 1.8369,
      "step": 194576
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.5249651670455933,
      "learning_rate": 3.5087945471312215e-05,
      "loss": 1.9092,
      "step": 194577
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.3825360536575317,
      "learning_rate": 3.5086025805800065e-05,
      "loss": 2.1612,
      "step": 194578
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.3838486671447754,
      "learning_rate": 3.5084106189540286e-05,
      "loss": 2.0849,
      "step": 194579
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.5685755014419556,
      "learning_rate": 3.508218662253344e-05,
      "loss": 1.8719,
      "step": 194580
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.4080668687820435,
      "learning_rate": 3.508026710477976e-05,
      "loss": 1.9627,
      "step": 194581
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.368424415588379,
      "learning_rate": 3.507834763627971e-05,
      "loss": 1.8802,
      "step": 194582
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.4344595670700073,
      "learning_rate": 3.5076428217033536e-05,
      "loss": 2.0072,
      "step": 194583
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.3031069040298462,
      "learning_rate": 3.50745088470418e-05,
      "loss": 1.925,
      "step": 194584
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.5339943170547485,
      "learning_rate": 3.507258952630455e-05,
      "loss": 1.951,
      "step": 194585
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.4343703985214233,
      "learning_rate": 3.507067025482241e-05,
      "loss": 2.1306,
      "step": 194586
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.5115269422531128,
      "learning_rate": 3.5068751032595533e-05,
      "loss": 1.7903,
      "step": 194587
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.3850157260894775,
      "learning_rate": 3.506683185962446e-05,
      "loss": 2.0703,
      "step": 194588
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.3549485206604004,
      "learning_rate": 3.506491273590941e-05,
      "loss": 2.0173,
      "step": 194589
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.4094336032867432,
      "learning_rate": 3.50629936614509e-05,
      "loss": 1.9793,
      "step": 194590
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.2874683141708374,
      "learning_rate": 3.5061074636249053e-05,
      "loss": 2.0926,
      "step": 194591
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.3069628477096558,
      "learning_rate": 3.505915566030444e-05,
      "loss": 1.8694,
      "step": 194592
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.4928815364837646,
      "learning_rate": 3.5057236733617254e-05,
      "loss": 2.0757,
      "step": 194593
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.627106785774231,
      "learning_rate": 3.5055317856188e-05,
      "loss": 1.888,
      "step": 194594
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.3890446424484253,
      "learning_rate": 3.5053399028016885e-05,
      "loss": 2.0728,
      "step": 194595
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.4177409410476685,
      "learning_rate": 3.5051480249104455e-05,
      "loss": 2.1505,
      "step": 194596
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.5078288316726685,
      "learning_rate": 3.504956151945087e-05,
      "loss": 2.102,
      "step": 194597
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.4032273292541504,
      "learning_rate": 3.5047642839056636e-05,
      "loss": 2.0602,
      "step": 194598
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.4484772682189941,
      "learning_rate": 3.5045724207921965e-05,
      "loss": 1.8511,
      "step": 194599
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.3841750621795654,
      "learning_rate": 3.504380562604733e-05,
      "loss": 2.0363,
      "step": 194600
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.4493229389190674,
      "learning_rate": 3.504188709343302e-05,
      "loss": 1.7017,
      "step": 194601
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.490181565284729,
      "learning_rate": 3.503996861007951e-05,
      "loss": 2.2381,
      "step": 194602
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.445281982421875,
      "learning_rate": 3.503805017598702e-05,
      "loss": 1.7661,
      "step": 194603
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.3407479524612427,
      "learning_rate": 3.5036131791156e-05,
      "loss": 1.9554,
      "step": 194604
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.425152063369751,
      "learning_rate": 3.50342134555867e-05,
      "loss": 2.08,
      "step": 194605
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.2944567203521729,
      "learning_rate": 3.503229516927957e-05,
      "loss": 1.9278,
      "step": 194606
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.4079755544662476,
      "learning_rate": 3.503037693223489e-05,
      "loss": 2.1279,
      "step": 194607
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.3534523248672485,
      "learning_rate": 3.502845874445315e-05,
      "loss": 2.1167,
      "step": 194608
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.4516085386276245,
      "learning_rate": 3.50265406059346e-05,
      "loss": 1.9748,
      "step": 194609
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.3275368213653564,
      "learning_rate": 3.502462251667963e-05,
      "loss": 2.1111,
      "step": 194610
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.4698597192764282,
      "learning_rate": 3.502270447668854e-05,
      "loss": 2.0214,
      "step": 194611
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.3931914567947388,
      "learning_rate": 3.502078648596176e-05,
      "loss": 1.7739,
      "step": 194612
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.624216079711914,
      "learning_rate": 3.501886854449955e-05,
      "loss": 2.0296,
      "step": 194613
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.4176186323165894,
      "learning_rate": 3.501695065230242e-05,
      "loss": 1.8983,
      "step": 194614
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.555646538734436,
      "learning_rate": 3.501503280937063e-05,
      "loss": 2.1271,
      "step": 194615
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.4032553434371948,
      "learning_rate": 3.5013115015704485e-05,
      "loss": 1.9105,
      "step": 194616
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.3967632055282593,
      "learning_rate": 3.501119727130447e-05,
      "loss": 1.971,
      "step": 194617
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.4518957138061523,
      "learning_rate": 3.500927957617088e-05,
      "loss": 2.1808,
      "step": 194618
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.4052351713180542,
      "learning_rate": 3.500736193030398e-05,
      "loss": 2.0366,
      "step": 194619
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.4339865446090698,
      "learning_rate": 3.5005444333704305e-05,
      "loss": 1.9593,
      "step": 194620
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.325734257698059,
      "learning_rate": 3.5003526786372116e-05,
      "loss": 2.0277,
      "step": 194621
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.6847527027130127,
      "learning_rate": 3.5001609288307686e-05,
      "loss": 2.0327,
      "step": 194622
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.3598450422286987,
      "learning_rate": 3.4999691839511544e-05,
      "loss": 2.0783,
      "step": 194623
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.3440957069396973,
      "learning_rate": 3.499777443998389e-05,
      "loss": 1.9308,
      "step": 194624
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.3668088912963867,
      "learning_rate": 3.4995857089725235e-05,
      "loss": 1.9991,
      "step": 194625
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.5306822061538696,
      "learning_rate": 3.499393978873583e-05,
      "loss": 1.9067,
      "step": 194626
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.4163126945495605,
      "learning_rate": 3.4992022537016076e-05,
      "loss": 2.16,
      "step": 194627
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.4509096145629883,
      "learning_rate": 3.499010533456621e-05,
      "loss": 2.0994,
      "step": 194628
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.621995449066162,
      "learning_rate": 3.4988188181386765e-05,
      "loss": 2.1343,
      "step": 194629
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.4093444347381592,
      "learning_rate": 3.4986271077477976e-05,
      "loss": 1.9674,
      "step": 194630
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.4619193077087402,
      "learning_rate": 3.4984354022840275e-05,
      "loss": 2.0913,
      "step": 194631
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.422078013420105,
      "learning_rate": 3.4982437017474e-05,
      "loss": 1.9179,
      "step": 194632
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.661967158317566,
      "learning_rate": 3.49805200613795e-05,
      "loss": 2.2743,
      "step": 194633
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.3711330890655518,
      "learning_rate": 3.497860315455703e-05,
      "loss": 1.9708,
      "step": 194634
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.5076136589050293,
      "learning_rate": 3.497668629700714e-05,
      "loss": 1.9547,
      "step": 194635
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.2798075675964355,
      "learning_rate": 3.4974769488730006e-05,
      "loss": 2.1821,
      "step": 194636
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.3750015497207642,
      "learning_rate": 3.497285272972612e-05,
      "loss": 1.7176,
      "step": 194637
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.5129057168960571,
      "learning_rate": 3.497093601999579e-05,
      "loss": 1.9839,
      "step": 194638
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.3804091215133667,
      "learning_rate": 3.496901935953937e-05,
      "loss": 1.9616,
      "step": 194639
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.471118688583374,
      "learning_rate": 3.4967102748357144e-05,
      "loss": 2.1675,
      "step": 194640
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.3477363586425781,
      "learning_rate": 3.49651861864496e-05,
      "loss": 1.9504,
      "step": 194641
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.5559015274047852,
      "learning_rate": 3.496326967381697e-05,
      "loss": 1.859,
      "step": 194642
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.3560529947280884,
      "learning_rate": 3.496135321045976e-05,
      "loss": 2.2005,
      "step": 194643
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.3387067317962646,
      "learning_rate": 3.4959436796378135e-05,
      "loss": 1.8961,
      "step": 194644
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.5093125104904175,
      "learning_rate": 3.495752043157269e-05,
      "loss": 2.0842,
      "step": 194645
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.3424715995788574,
      "learning_rate": 3.495560411604353e-05,
      "loss": 2.064,
      "step": 194646
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.5133179426193237,
      "learning_rate": 3.495368784979119e-05,
      "loss": 1.8174,
      "step": 194647
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.4215766191482544,
      "learning_rate": 3.4951771632815896e-05,
      "loss": 2.2276,
      "step": 194648
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.493520736694336,
      "learning_rate": 3.4949855465118116e-05,
      "loss": 2.1921,
      "step": 194649
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.5008087158203125,
      "learning_rate": 3.494793934669812e-05,
      "loss": 1.9208,
      "step": 194650
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.4169303178787231,
      "learning_rate": 3.494602327755644e-05,
      "loss": 1.8984,
      "step": 194651
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.4985613822937012,
      "learning_rate": 3.4944107257693134e-05,
      "loss": 2.1115,
      "step": 194652
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.4201877117156982,
      "learning_rate": 3.494219128710881e-05,
      "loss": 1.8631,
      "step": 194653
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.568463683128357,
      "learning_rate": 3.4940275365803674e-05,
      "loss": 1.9426,
      "step": 194654
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.545790195465088,
      "learning_rate": 3.4938359493778214e-05,
      "loss": 2.1188,
      "step": 194655
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.5855129957199097,
      "learning_rate": 3.493644367103263e-05,
      "loss": 2.0376,
      "step": 194656
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.4081414937973022,
      "learning_rate": 3.49345278975675e-05,
      "loss": 2.2668,
      "step": 194657
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.4924776554107666,
      "learning_rate": 3.4932612173382946e-05,
      "loss": 2.3115,
      "step": 194658
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.409036636352539,
      "learning_rate": 3.4930696498479435e-05,
      "loss": 1.9373,
      "step": 194659
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.4869972467422485,
      "learning_rate": 3.4928780872857276e-05,
      "loss": 2.0389,
      "step": 194660
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.472238302230835,
      "learning_rate": 3.4926865296516924e-05,
      "loss": 1.9699,
      "step": 194661
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.5409610271453857,
      "learning_rate": 3.492494976945862e-05,
      "loss": 2.169,
      "step": 194662
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.457335114479065,
      "learning_rate": 3.492303429168286e-05,
      "loss": 2.1042,
      "step": 194663
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.3165018558502197,
      "learning_rate": 3.492111886318981e-05,
      "loss": 2.1787,
      "step": 194664
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.7701307535171509,
      "learning_rate": 3.491920348398001e-05,
      "loss": 1.9317,
      "step": 194665
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.4587892293930054,
      "learning_rate": 3.491728815405365e-05,
      "loss": 1.9205,
      "step": 194666
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.5119280815124512,
      "learning_rate": 3.491537287341123e-05,
      "loss": 1.9493,
      "step": 194667
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.2456659078598022,
      "learning_rate": 3.4913457642052986e-05,
      "loss": 1.8242,
      "step": 194668
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.3664460182189941,
      "learning_rate": 3.491154245997946e-05,
      "loss": 1.8304,
      "step": 194669
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.28154456615448,
      "learning_rate": 3.490962732719077e-05,
      "loss": 1.9931,
      "step": 194670
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.385880470275879,
      "learning_rate": 3.4907712243687426e-05,
      "loss": 1.8381,
      "step": 194671
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.4987812042236328,
      "learning_rate": 3.490579720946969e-05,
      "loss": 2.0512,
      "step": 194672
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.3862857818603516,
      "learning_rate": 3.4903882224538026e-05,
      "loss": 1.9962,
      "step": 194673
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.5557302236557007,
      "learning_rate": 3.490196728889267e-05,
      "loss": 1.7727,
      "step": 194674
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.313592791557312,
      "learning_rate": 3.49000524025342e-05,
      "loss": 1.8023,
      "step": 194675
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.5079346895217896,
      "learning_rate": 3.489813756546266e-05,
      "loss": 1.9413,
      "step": 194676
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.408959150314331,
      "learning_rate": 3.4896222777678594e-05,
      "loss": 2.1206,
      "step": 194677
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.3881292343139648,
      "learning_rate": 3.4894308039182305e-05,
      "loss": 2.0022,
      "step": 194678
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.388722538948059,
      "learning_rate": 3.4892393349974226e-05,
      "loss": 2.0226,
      "step": 194679
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.3879715204238892,
      "learning_rate": 3.489047871005458e-05,
      "loss": 2.0475,
      "step": 194680
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.392690896987915,
      "learning_rate": 3.4888564119423946e-05,
      "loss": 2.155,
      "step": 194681
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.3258130550384521,
      "learning_rate": 3.488664957808238e-05,
      "loss": 1.8768,
      "step": 194682
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.6220221519470215,
      "learning_rate": 3.488473508603046e-05,
      "loss": 1.756,
      "step": 194683
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.5582361221313477,
      "learning_rate": 3.488282064326844e-05,
      "loss": 1.9586,
      "step": 194684
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.2914834022521973,
      "learning_rate": 3.488090624979673e-05,
      "loss": 2.0196,
      "step": 194685
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.3276710510253906,
      "learning_rate": 3.4878991905615614e-05,
      "loss": 2.101,
      "step": 194686
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.4323958158493042,
      "learning_rate": 3.487707761072558e-05,
      "loss": 2.0619,
      "step": 194687
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.4743454456329346,
      "learning_rate": 3.487516336512688e-05,
      "loss": 1.9474,
      "step": 194688
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.244930624961853,
      "learning_rate": 3.4873249168819884e-05,
      "loss": 1.8384,
      "step": 194689
    },
    {
      "epoch": 2.53,
      "grad_norm": 1.4856679439544678,
      "learning_rate": 3.487133502180489e-05,
      "loss": 2.1201,
      "step": 194690
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.5027042627334595,
      "learning_rate": 3.48694209240824e-05,
      "loss": 2.1075,
      "step": 194691
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.448488473892212,
      "learning_rate": 3.486750687565265e-05,
      "loss": 1.8929,
      "step": 194692
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.4669917821884155,
      "learning_rate": 3.486559287651607e-05,
      "loss": 2.0554,
      "step": 194693
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.4425500631332397,
      "learning_rate": 3.486367892667299e-05,
      "loss": 1.9969,
      "step": 194694
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.3961889743804932,
      "learning_rate": 3.4861765026123746e-05,
      "loss": 1.9906,
      "step": 194695
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.971166729927063,
      "learning_rate": 3.485985117486864e-05,
      "loss": 2.014,
      "step": 194696
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.3665939569473267,
      "learning_rate": 3.4857937372908164e-05,
      "loss": 2.0685,
      "step": 194697
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.5917723178863525,
      "learning_rate": 3.485602362024252e-05,
      "loss": 1.954,
      "step": 194698
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.4747424125671387,
      "learning_rate": 3.485410991687222e-05,
      "loss": 1.975,
      "step": 194699
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.4463549852371216,
      "learning_rate": 3.485219626279758e-05,
      "loss": 2.0516,
      "step": 194700
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.39357328414917,
      "learning_rate": 3.485028265801881e-05,
      "loss": 1.9763,
      "step": 194701
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.3859727382659912,
      "learning_rate": 3.4848369102536474e-05,
      "loss": 2.0593,
      "step": 194702
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.460473895072937,
      "learning_rate": 3.484645559635081e-05,
      "loss": 1.9845,
      "step": 194703
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.4463001489639282,
      "learning_rate": 3.484454213946214e-05,
      "loss": 1.9373,
      "step": 194704
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.4206461906433105,
      "learning_rate": 3.484262873187094e-05,
      "loss": 2.0616,
      "step": 194705
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.3563876152038574,
      "learning_rate": 3.48407153735775e-05,
      "loss": 1.9561,
      "step": 194706
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.4325751066207886,
      "learning_rate": 3.483880206458213e-05,
      "loss": 2.0587,
      "step": 194707
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.322478175163269,
      "learning_rate": 3.483688880488526e-05,
      "loss": 2.0876,
      "step": 194708
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.6000038385391235,
      "learning_rate": 3.483497559448725e-05,
      "loss": 2.0211,
      "step": 194709
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.4468414783477783,
      "learning_rate": 3.483306243338834e-05,
      "loss": 2.0285,
      "step": 194710
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.4667593240737915,
      "learning_rate": 3.483114932158902e-05,
      "loss": 1.788,
      "step": 194711
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.506089448928833,
      "learning_rate": 3.482923625908961e-05,
      "loss": 2.1152,
      "step": 194712
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.3499901294708252,
      "learning_rate": 3.482732324589039e-05,
      "loss": 1.9814,
      "step": 194713
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.5578292608261108,
      "learning_rate": 3.482541028199184e-05,
      "loss": 2.2372,
      "step": 194714
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.4206628799438477,
      "learning_rate": 3.482349736739421e-05,
      "loss": 2.0691,
      "step": 194715
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.4590049982070923,
      "learning_rate": 3.4821584502097924e-05,
      "loss": 2.1221,
      "step": 194716
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.5616884231567383,
      "learning_rate": 3.481967168610332e-05,
      "loss": 2.1305,
      "step": 194717
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.6178501844406128,
      "learning_rate": 3.4817758919410756e-05,
      "loss": 1.9706,
      "step": 194718
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.3843653202056885,
      "learning_rate": 3.4815846202020514e-05,
      "loss": 2.1298,
      "step": 194719
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.4592170715332031,
      "learning_rate": 3.481393353393307e-05,
      "loss": 2.105,
      "step": 194720
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.444192886352539,
      "learning_rate": 3.481202091514865e-05,
      "loss": 1.8816,
      "step": 194721
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.6009576320648193,
      "learning_rate": 3.4810108345667765e-05,
      "loss": 2.017,
      "step": 194722
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.519963026046753,
      "learning_rate": 3.4808195825490605e-05,
      "loss": 2.0094,
      "step": 194723
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.5902851819992065,
      "learning_rate": 3.4806283354617736e-05,
      "loss": 2.0409,
      "step": 194724
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.5268181562423706,
      "learning_rate": 3.4804370933049265e-05,
      "loss": 2.0369,
      "step": 194725
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.3581613302230835,
      "learning_rate": 3.4802458560785714e-05,
      "loss": 1.8983,
      "step": 194726
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.3851069211959839,
      "learning_rate": 3.480054623782733e-05,
      "loss": 2.04,
      "step": 194727
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.5351181030273438,
      "learning_rate": 3.4798633964174635e-05,
      "loss": 2.1128,
      "step": 194728
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.4040591716766357,
      "learning_rate": 3.479672173982777e-05,
      "loss": 2.03,
      "step": 194729
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.6425718069076538,
      "learning_rate": 3.4794809564787365e-05,
      "loss": 1.8986,
      "step": 194730
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.485128402709961,
      "learning_rate": 3.4792897439053445e-05,
      "loss": 2.0218,
      "step": 194731
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.5025818347930908,
      "learning_rate": 3.479098536262662e-05,
      "loss": 1.8885,
      "step": 194732
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.3372124433517456,
      "learning_rate": 3.478907333550709e-05,
      "loss": 1.9688,
      "step": 194733
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.5475021600723267,
      "learning_rate": 3.4787161357695317e-05,
      "loss": 2.1498,
      "step": 194734
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.5423952341079712,
      "learning_rate": 3.4785249429191595e-05,
      "loss": 1.7226,
      "step": 194735
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.4286526441574097,
      "learning_rate": 3.4783337549996406e-05,
      "loss": 1.9936,
      "step": 194736
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.5185593366622925,
      "learning_rate": 3.4781425720109865e-05,
      "loss": 1.99,
      "step": 194737
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.8549745082855225,
      "learning_rate": 3.477951393953252e-05,
      "loss": 1.9346,
      "step": 194738
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.4346935749053955,
      "learning_rate": 3.4777602208264635e-05,
      "loss": 2.2141,
      "step": 194739
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.4092320203781128,
      "learning_rate": 3.4775690526306665e-05,
      "loss": 2.0368,
      "step": 194740
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.3307429552078247,
      "learning_rate": 3.477377889365882e-05,
      "loss": 2.0335,
      "step": 194741
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.4540787935256958,
      "learning_rate": 3.477186731032167e-05,
      "loss": 2.0038,
      "step": 194742
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.48270845413208,
      "learning_rate": 3.47699557762953e-05,
      "loss": 1.963,
      "step": 194743
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.4752087593078613,
      "learning_rate": 3.476804429158029e-05,
      "loss": 2.008,
      "step": 194744
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.430554747581482,
      "learning_rate": 3.4766132856176797e-05,
      "loss": 2.0248,
      "step": 194745
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.4209498167037964,
      "learning_rate": 3.4764221470085395e-05,
      "loss": 2.2057,
      "step": 194746
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.3674184083938599,
      "learning_rate": 3.476231013330625e-05,
      "loss": 1.8565,
      "step": 194747
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.3977758884429932,
      "learning_rate": 3.476039884583992e-05,
      "loss": 2.0514,
      "step": 194748
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.409178614616394,
      "learning_rate": 3.4758487607686516e-05,
      "loss": 2.0639,
      "step": 194749
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.384202003479004,
      "learning_rate": 3.475657641884659e-05,
      "loss": 2.0344,
      "step": 194750
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.428587794303894,
      "learning_rate": 3.475466527932036e-05,
      "loss": 1.879,
      "step": 194751
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.5350587368011475,
      "learning_rate": 3.4752754189108275e-05,
      "loss": 1.9423,
      "step": 194752
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.491263508796692,
      "learning_rate": 3.475084314821065e-05,
      "loss": 1.8013,
      "step": 194753
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.3095886707305908,
      "learning_rate": 3.474893215662794e-05,
      "loss": 1.798,
      "step": 194754
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.3852850198745728,
      "learning_rate": 3.474702121436028e-05,
      "loss": 2.2555,
      "step": 194755
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.4409583806991577,
      "learning_rate": 3.4745110321408245e-05,
      "loss": 2.2539,
      "step": 194756
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.3855055570602417,
      "learning_rate": 3.4743199477772024e-05,
      "loss": 2.2915,
      "step": 194757
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.5974688529968262,
      "learning_rate": 3.4741288683452114e-05,
      "loss": 1.9404,
      "step": 194758
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.3583159446716309,
      "learning_rate": 3.473937793844873e-05,
      "loss": 2.0062,
      "step": 194759
    },
    {
      "epoch": 2.54,
      "grad_norm": 2.0374691486358643,
      "learning_rate": 3.473746724276243e-05,
      "loss": 2.0483,
      "step": 194760
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.3365947008132935,
      "learning_rate": 3.473555659639334e-05,
      "loss": 2.0933,
      "step": 194761
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.5199458599090576,
      "learning_rate": 3.473364599934197e-05,
      "loss": 2.129,
      "step": 194762
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.3917268514633179,
      "learning_rate": 3.473173545160855e-05,
      "loss": 2.0997,
      "step": 194763
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.50693941116333,
      "learning_rate": 3.4729824953193585e-05,
      "loss": 2.0958,
      "step": 194764
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.2254480123519897,
      "learning_rate": 3.472791450409727e-05,
      "loss": 2.0023,
      "step": 194765
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.4878813028335571,
      "learning_rate": 3.4726004104320195e-05,
      "loss": 2.157,
      "step": 194766
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.3809653520584106,
      "learning_rate": 3.47240937538624e-05,
      "loss": 2.1332,
      "step": 194767
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.3384076356887817,
      "learning_rate": 3.472218345272449e-05,
      "loss": 2.1191,
      "step": 194768
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.402855634689331,
      "learning_rate": 3.472027320090667e-05,
      "loss": 1.8171,
      "step": 194769
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.5319634675979614,
      "learning_rate": 3.471836299840942e-05,
      "loss": 1.8223,
      "step": 194770
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.3318227529525757,
      "learning_rate": 3.471645284523299e-05,
      "loss": 1.8913,
      "step": 194771
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.4645668268203735,
      "learning_rate": 3.471454274137781e-05,
      "loss": 2.0043,
      "step": 194772
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.3754911422729492,
      "learning_rate": 3.4712632686844236e-05,
      "loss": 2.1189,
      "step": 194773
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.5490608215332031,
      "learning_rate": 3.471072268163257e-05,
      "loss": 2.031,
      "step": 194774
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.4951914548873901,
      "learning_rate": 3.470881272574313e-05,
      "loss": 2.1267,
      "step": 194775
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.4453383684158325,
      "learning_rate": 3.4706902819176395e-05,
      "loss": 2.1534,
      "step": 194776
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.379239797592163,
      "learning_rate": 3.470499296193258e-05,
      "loss": 2.0843,
      "step": 194777
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.477388858795166,
      "learning_rate": 3.470308315401217e-05,
      "loss": 2.0599,
      "step": 194778
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.3206820487976074,
      "learning_rate": 3.4701173395415504e-05,
      "loss": 1.8661,
      "step": 194779
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.3057674169540405,
      "learning_rate": 3.4699263686142854e-05,
      "loss": 2.0875,
      "step": 194780
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.5658447742462158,
      "learning_rate": 3.469735402619458e-05,
      "loss": 1.9981,
      "step": 194781
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.3975423574447632,
      "learning_rate": 3.469544441557115e-05,
      "loss": 2.0227,
      "step": 194782
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.4125239849090576,
      "learning_rate": 3.4693534854272764e-05,
      "loss": 2.0347,
      "step": 194783
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.6443533897399902,
      "learning_rate": 3.4691625342299926e-05,
      "loss": 2.0201,
      "step": 194784
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.4645318984985352,
      "learning_rate": 3.468971587965293e-05,
      "loss": 1.943,
      "step": 194785
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.4748866558074951,
      "learning_rate": 3.468780646633204e-05,
      "loss": 2.0461,
      "step": 194786
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.4438718557357788,
      "learning_rate": 3.468589710233776e-05,
      "loss": 1.97,
      "step": 194787
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.406369686126709,
      "learning_rate": 3.4683987787670384e-05,
      "loss": 1.9273,
      "step": 194788
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.390062689781189,
      "learning_rate": 3.4682078522330224e-05,
      "loss": 2.126,
      "step": 194789
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.493706226348877,
      "learning_rate": 3.4680169306317704e-05,
      "loss": 1.9761,
      "step": 194790
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.3267194032669067,
      "learning_rate": 3.4678260139633165e-05,
      "loss": 1.9999,
      "step": 194791
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.5467251539230347,
      "learning_rate": 3.46763510222769e-05,
      "loss": 2.1138,
      "step": 194792
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.4476680755615234,
      "learning_rate": 3.4674441954249334e-05,
      "loss": 2.0536,
      "step": 194793
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.3889154195785522,
      "learning_rate": 3.467253293555081e-05,
      "loss": 2.0416,
      "step": 194794
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.334417700767517,
      "learning_rate": 3.467062396618163e-05,
      "loss": 1.9199,
      "step": 194795
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.4876242876052856,
      "learning_rate": 3.4668715046142225e-05,
      "loss": 2.0008,
      "step": 194796
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.4512462615966797,
      "learning_rate": 3.4666806175432925e-05,
      "loss": 2.1641,
      "step": 194797
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.3339111804962158,
      "learning_rate": 3.466489735405403e-05,
      "loss": 1.9167,
      "step": 194798
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.4439014196395874,
      "learning_rate": 3.466298858200598e-05,
      "loss": 1.8759,
      "step": 194799
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.4279078245162964,
      "learning_rate": 3.4661079859289035e-05,
      "loss": 1.7776,
      "step": 194800
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.379021167755127,
      "learning_rate": 3.4659171185903656e-05,
      "loss": 1.7853,
      "step": 194801
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.3622933626174927,
      "learning_rate": 3.4657262561850154e-05,
      "loss": 2.0665,
      "step": 194802
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.4875825643539429,
      "learning_rate": 3.465535398712889e-05,
      "loss": 2.1371,
      "step": 194803
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.5689257383346558,
      "learning_rate": 3.4653445461740095e-05,
      "loss": 2.0179,
      "step": 194804
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.4760185480117798,
      "learning_rate": 3.4651536985684337e-05,
      "loss": 2.0569,
      "step": 194805
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.5312517881393433,
      "learning_rate": 3.464962855896178e-05,
      "loss": 2.1709,
      "step": 194806
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.4992936849594116,
      "learning_rate": 3.464772018157297e-05,
      "loss": 2.1294,
      "step": 194807
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.5203298330307007,
      "learning_rate": 3.4645811853518126e-05,
      "loss": 2.0736,
      "step": 194808
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.330680251121521,
      "learning_rate": 3.4643903574797646e-05,
      "loss": 1.9123,
      "step": 194809
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.4157192707061768,
      "learning_rate": 3.464199534541178e-05,
      "loss": 2.1617,
      "step": 194810
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.4081876277923584,
      "learning_rate": 3.464008716536108e-05,
      "loss": 1.8895,
      "step": 194811
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.561795711517334,
      "learning_rate": 3.4638179034645707e-05,
      "loss": 1.8856,
      "step": 194812
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.462510108947754,
      "learning_rate": 3.4636270953266177e-05,
      "loss": 1.9574,
      "step": 194813
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.3792078495025635,
      "learning_rate": 3.463436292122271e-05,
      "loss": 1.869,
      "step": 194814
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.276123046875,
      "learning_rate": 3.463245493851585e-05,
      "loss": 1.937,
      "step": 194815
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.356268048286438,
      "learning_rate": 3.463054700514573e-05,
      "loss": 1.8998,
      "step": 194816
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.4496229887008667,
      "learning_rate": 3.46286391211128e-05,
      "loss": 2.086,
      "step": 194817
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.35028874874115,
      "learning_rate": 3.462673128641741e-05,
      "loss": 1.936,
      "step": 194818
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.4267346858978271,
      "learning_rate": 3.462482350105995e-05,
      "loss": 2.1904,
      "step": 194819
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.4860448837280273,
      "learning_rate": 3.462291576504069e-05,
      "loss": 2.08,
      "step": 194820
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.7761929035186768,
      "learning_rate": 3.462100807836017e-05,
      "loss": 2.005,
      "step": 194821
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.3586097955703735,
      "learning_rate": 3.461910044101848e-05,
      "loss": 2.0008,
      "step": 194822
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.3925418853759766,
      "learning_rate": 3.461719285301619e-05,
      "loss": 1.8766,
      "step": 194823
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.5427888631820679,
      "learning_rate": 3.461528531435349e-05,
      "loss": 1.9562,
      "step": 194824
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.3723348379135132,
      "learning_rate": 3.4613377825030894e-05,
      "loss": 1.9662,
      "step": 194825
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.4063231945037842,
      "learning_rate": 3.46114703850486e-05,
      "loss": 2.2118,
      "step": 194826
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.484614610671997,
      "learning_rate": 3.4609562994407204e-05,
      "loss": 1.8463,
      "step": 194827
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.4308897256851196,
      "learning_rate": 3.4607655653106734e-05,
      "loss": 1.8079,
      "step": 194828
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.434637427330017,
      "learning_rate": 3.46057483611478e-05,
      "loss": 2.1113,
      "step": 194829
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.3130239248275757,
      "learning_rate": 3.460384111853059e-05,
      "loss": 2.0819,
      "step": 194830
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.4510806798934937,
      "learning_rate": 3.460193392525559e-05,
      "loss": 2.0733,
      "step": 194831
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.503922462463379,
      "learning_rate": 3.460002678132308e-05,
      "loss": 2.1752,
      "step": 194832
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.305875301361084,
      "learning_rate": 3.459811968673353e-05,
      "loss": 1.9246,
      "step": 194833
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.448810338973999,
      "learning_rate": 3.459621264148707e-05,
      "loss": 2.0296,
      "step": 194834
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.4235715866088867,
      "learning_rate": 3.4594305645584256e-05,
      "loss": 1.9423,
      "step": 194835
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.377263069152832,
      "learning_rate": 3.459239869902529e-05,
      "loss": 2.1537,
      "step": 194836
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.5644201040267944,
      "learning_rate": 3.4590491801810694e-05,
      "loss": 1.9663,
      "step": 194837
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.485452651977539,
      "learning_rate": 3.4588584953940654e-05,
      "loss": 2.1126,
      "step": 194838
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.3322689533233643,
      "learning_rate": 3.4586678155415684e-05,
      "loss": 1.9268,
      "step": 194839
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.394243836402893,
      "learning_rate": 3.4584771406236034e-05,
      "loss": 1.9804,
      "step": 194840
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.4354877471923828,
      "learning_rate": 3.458286470640211e-05,
      "loss": 1.9456,
      "step": 194841
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.5242267847061157,
      "learning_rate": 3.458095805591419e-05,
      "loss": 1.9829,
      "step": 194842
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.4071685075759888,
      "learning_rate": 3.457905145477272e-05,
      "loss": 1.9838,
      "step": 194843
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.3079506158828735,
      "learning_rate": 3.457714490297795e-05,
      "loss": 1.9626,
      "step": 194844
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.3460543155670166,
      "learning_rate": 3.457523840053037e-05,
      "loss": 1.9646,
      "step": 194845
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.4050027132034302,
      "learning_rate": 3.457333194743025e-05,
      "loss": 1.8776,
      "step": 194846
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.4896256923675537,
      "learning_rate": 3.457142554367799e-05,
      "loss": 2.0008,
      "step": 194847
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.8111488819122314,
      "learning_rate": 3.456951918927382e-05,
      "loss": 1.9383,
      "step": 194848
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.5280251502990723,
      "learning_rate": 3.456761288421827e-05,
      "loss": 2.0495,
      "step": 194849
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.4039560556411743,
      "learning_rate": 3.456570662851152e-05,
      "loss": 1.9939,
      "step": 194850
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.4413447380065918,
      "learning_rate": 3.4563800422154095e-05,
      "loss": 2.1147,
      "step": 194851
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.4171658754348755,
      "learning_rate": 3.456189426514629e-05,
      "loss": 2.0508,
      "step": 194852
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.546627163887024,
      "learning_rate": 3.455998815748842e-05,
      "loss": 1.7462,
      "step": 194853
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.4076322317123413,
      "learning_rate": 3.4558082099180797e-05,
      "loss": 2.1452,
      "step": 194854
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.3038476705551147,
      "learning_rate": 3.45561760902239e-05,
      "loss": 1.9653,
      "step": 194855
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.3427464962005615,
      "learning_rate": 3.4554270130617956e-05,
      "loss": 1.9831,
      "step": 194856
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.417340636253357,
      "learning_rate": 3.455236422036344e-05,
      "loss": 2.1779,
      "step": 194857
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.2700344324111938,
      "learning_rate": 3.455045835946068e-05,
      "loss": 1.9069,
      "step": 194858
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.4695266485214233,
      "learning_rate": 3.454855254790997e-05,
      "loss": 1.9649,
      "step": 194859
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.4239602088928223,
      "learning_rate": 3.454664678571165e-05,
      "loss": 1.9539,
      "step": 194860
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.6228312253952026,
      "learning_rate": 3.454474107286616e-05,
      "loss": 2.1319,
      "step": 194861
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.3102235794067383,
      "learning_rate": 3.454283540937378e-05,
      "loss": 1.8512,
      "step": 194862
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.7604120969772339,
      "learning_rate": 3.454092979523493e-05,
      "loss": 1.9883,
      "step": 194863
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.3641362190246582,
      "learning_rate": 3.4539024230449965e-05,
      "loss": 1.9767,
      "step": 194864
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.3214399814605713,
      "learning_rate": 3.4537118715019184e-05,
      "loss": 1.8021,
      "step": 194865
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.4543415307998657,
      "learning_rate": 3.453521324894289e-05,
      "loss": 1.9461,
      "step": 194866
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.4875671863555908,
      "learning_rate": 3.453330783222159e-05,
      "loss": 2.1,
      "step": 194867
    },
    {
      "epoch": 2.54,
      "grad_norm": 2.016814708709717,
      "learning_rate": 3.453140246485551e-05,
      "loss": 1.7445,
      "step": 194868
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.545426845550537,
      "learning_rate": 3.4529497146845075e-05,
      "loss": 2.3142,
      "step": 194869
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.346545934677124,
      "learning_rate": 3.452759187819066e-05,
      "loss": 1.9129,
      "step": 194870
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.326916217803955,
      "learning_rate": 3.452568665889257e-05,
      "loss": 1.8744,
      "step": 194871
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.4552204608917236,
      "learning_rate": 3.4523781488951095e-05,
      "loss": 1.9918,
      "step": 194872
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.386288046836853,
      "learning_rate": 3.452187636836671e-05,
      "loss": 1.9005,
      "step": 194873
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.3900610208511353,
      "learning_rate": 3.451997129713967e-05,
      "loss": 1.8283,
      "step": 194874
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.4141621589660645,
      "learning_rate": 3.451806627527045e-05,
      "loss": 2.0089,
      "step": 194875
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.438410758972168,
      "learning_rate": 3.4516161302759316e-05,
      "loss": 2.1432,
      "step": 194876
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.448446273803711,
      "learning_rate": 3.45142563796066e-05,
      "loss": 2.0282,
      "step": 194877
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.5291271209716797,
      "learning_rate": 3.451235150581273e-05,
      "loss": 1.9646,
      "step": 194878
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.397404432296753,
      "learning_rate": 3.451044668137804e-05,
      "loss": 1.8213,
      "step": 194879
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.5198768377304077,
      "learning_rate": 3.450854190630278e-05,
      "loss": 1.9227,
      "step": 194880
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.5883363485336304,
      "learning_rate": 3.450663718058749e-05,
      "loss": 1.9469,
      "step": 194881
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.3510370254516602,
      "learning_rate": 3.450473250423242e-05,
      "loss": 2.0002,
      "step": 194882
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.6333078145980835,
      "learning_rate": 3.450282787723787e-05,
      "loss": 1.8553,
      "step": 194883
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.6634830236434937,
      "learning_rate": 3.45009232996043e-05,
      "loss": 2.0514,
      "step": 194884
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.3009033203125,
      "learning_rate": 3.449901877133198e-05,
      "loss": 2.0047,
      "step": 194885
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.6245458126068115,
      "learning_rate": 3.449711429242137e-05,
      "loss": 2.0167,
      "step": 194886
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.364133358001709,
      "learning_rate": 3.449520986287274e-05,
      "loss": 2.0184,
      "step": 194887
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.339950680732727,
      "learning_rate": 3.449330548268646e-05,
      "loss": 2.0788,
      "step": 194888
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.3393107652664185,
      "learning_rate": 3.449140115186283e-05,
      "loss": 1.9582,
      "step": 194889
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.4627423286437988,
      "learning_rate": 3.448949687040235e-05,
      "loss": 2.0943,
      "step": 194890
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.4037715196609497,
      "learning_rate": 3.4487592638305174e-05,
      "loss": 2.0133,
      "step": 194891
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.236729621887207,
      "learning_rate": 3.448568845557188e-05,
      "loss": 1.9024,
      "step": 194892
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.4532687664031982,
      "learning_rate": 3.4483784322202666e-05,
      "loss": 2.0238,
      "step": 194893
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.4340481758117676,
      "learning_rate": 3.4481880238197934e-05,
      "loss": 2.0093,
      "step": 194894
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.4007209539413452,
      "learning_rate": 3.447997620355798e-05,
      "loss": 1.8637,
      "step": 194895
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.3438351154327393,
      "learning_rate": 3.4478072218283306e-05,
      "loss": 2.0625,
      "step": 194896
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.4137786626815796,
      "learning_rate": 3.447616828237407e-05,
      "loss": 2.1355,
      "step": 194897
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.3277266025543213,
      "learning_rate": 3.4474264395830786e-05,
      "loss": 1.8812,
      "step": 194898
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.5391318798065186,
      "learning_rate": 3.447236055865368e-05,
      "loss": 2.0404,
      "step": 194899
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.5375102758407593,
      "learning_rate": 3.447045677084331e-05,
      "loss": 1.9989,
      "step": 194900
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.4723827838897705,
      "learning_rate": 3.446855303239979e-05,
      "loss": 1.8895,
      "step": 194901
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.4424397945404053,
      "learning_rate": 3.4466649343323616e-05,
      "loss": 1.8357,
      "step": 194902
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.3845409154891968,
      "learning_rate": 3.446474570361505e-05,
      "loss": 2.0585,
      "step": 194903
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.42039954662323,
      "learning_rate": 3.4462842113274556e-05,
      "loss": 1.9695,
      "step": 194904
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.3905608654022217,
      "learning_rate": 3.4460938572302376e-05,
      "loss": 1.9891,
      "step": 194905
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.4656137228012085,
      "learning_rate": 3.4459035080699e-05,
      "loss": 1.9599,
      "step": 194906
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.578946828842163,
      "learning_rate": 3.4457131638464674e-05,
      "loss": 1.9798,
      "step": 194907
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.5024282932281494,
      "learning_rate": 3.445522824559978e-05,
      "loss": 1.8245,
      "step": 194908
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.6671831607818604,
      "learning_rate": 3.445332490210464e-05,
      "loss": 2.0879,
      "step": 194909
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.4057512283325195,
      "learning_rate": 3.445142160797967e-05,
      "loss": 2.0782,
      "step": 194910
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.3750030994415283,
      "learning_rate": 3.4449518363225135e-05,
      "loss": 1.835,
      "step": 194911
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.680571436882019,
      "learning_rate": 3.4447615167841516e-05,
      "loss": 2.1209,
      "step": 194912
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.390012264251709,
      "learning_rate": 3.44457120218291e-05,
      "loss": 1.8602,
      "step": 194913
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.5183072090148926,
      "learning_rate": 3.4443808925188256e-05,
      "loss": 1.9785,
      "step": 194914
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.4181052446365356,
      "learning_rate": 3.444190587791922e-05,
      "loss": 2.1057,
      "step": 194915
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.7384018898010254,
      "learning_rate": 3.4440002880022523e-05,
      "loss": 2.0051,
      "step": 194916
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.4657930135726929,
      "learning_rate": 3.44380999314984e-05,
      "loss": 2.033,
      "step": 194917
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.5121265649795532,
      "learning_rate": 3.443619703234728e-05,
      "loss": 2.1674,
      "step": 194918
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.5148301124572754,
      "learning_rate": 3.44342941825695e-05,
      "loss": 1.9665,
      "step": 194919
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.4297118186950684,
      "learning_rate": 3.443239138216539e-05,
      "loss": 1.7607,
      "step": 194920
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.5588266849517822,
      "learning_rate": 3.4430488631135256e-05,
      "loss": 2.1489,
      "step": 194921
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.3569544553756714,
      "learning_rate": 3.4428585929479594e-05,
      "loss": 1.9239,
      "step": 194922
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.403530478477478,
      "learning_rate": 3.4426683277198566e-05,
      "loss": 2.1228,
      "step": 194923
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.7787415981292725,
      "learning_rate": 3.442478067429271e-05,
      "loss": 1.7851,
      "step": 194924
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.4206271171569824,
      "learning_rate": 3.442287812076232e-05,
      "loss": 2.0558,
      "step": 194925
    },
    {
      "epoch": 2.54,
      "grad_norm": 2.0887439250946045,
      "learning_rate": 3.44209756166077e-05,
      "loss": 2.1152,
      "step": 194926
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.4111948013305664,
      "learning_rate": 3.441907316182919e-05,
      "loss": 2.0465,
      "step": 194927
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.376548409461975,
      "learning_rate": 3.441717075642725e-05,
      "loss": 1.9271,
      "step": 194928
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.2906670570373535,
      "learning_rate": 3.441526840040207e-05,
      "loss": 1.6469,
      "step": 194929
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.385193109512329,
      "learning_rate": 3.44133660937542e-05,
      "loss": 2.0449,
      "step": 194930
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.303693175315857,
      "learning_rate": 3.44114638364839e-05,
      "loss": 1.917,
      "step": 194931
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.354556679725647,
      "learning_rate": 3.440956162859153e-05,
      "loss": 2.0749,
      "step": 194932
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.4175249338150024,
      "learning_rate": 3.440765947007737e-05,
      "loss": 2.0128,
      "step": 194933
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.4046622514724731,
      "learning_rate": 3.44057573609419e-05,
      "loss": 1.9813,
      "step": 194934
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.4128499031066895,
      "learning_rate": 3.4403855301185334e-05,
      "loss": 1.7971,
      "step": 194935
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.422280192375183,
      "learning_rate": 3.440195329080817e-05,
      "loss": 1.958,
      "step": 194936
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.3332654237747192,
      "learning_rate": 3.4400051329810705e-05,
      "loss": 2.0394,
      "step": 194937
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.5913527011871338,
      "learning_rate": 3.439814941819328e-05,
      "loss": 2.0733,
      "step": 194938
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.561485767364502,
      "learning_rate": 3.439624755595621e-05,
      "loss": 2.018,
      "step": 194939
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.3554857969284058,
      "learning_rate": 3.439434574309995e-05,
      "loss": 1.7938,
      "step": 194940
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.436043381690979,
      "learning_rate": 3.439244397962472e-05,
      "loss": 2.0187,
      "step": 194941
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.6232768297195435,
      "learning_rate": 3.439054226553098e-05,
      "loss": 2.0126,
      "step": 194942
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.4836302995681763,
      "learning_rate": 3.4388640600819085e-05,
      "loss": 2.0619,
      "step": 194943
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.571144938468933,
      "learning_rate": 3.4386738985489384e-05,
      "loss": 1.8915,
      "step": 194944
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.5674540996551514,
      "learning_rate": 3.438483741954211e-05,
      "loss": 2.2941,
      "step": 194945
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.4110616445541382,
      "learning_rate": 3.438293590297777e-05,
      "loss": 1.9025,
      "step": 194946
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.4603488445281982,
      "learning_rate": 3.4381034435796596e-05,
      "loss": 1.7807,
      "step": 194947
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.3123365640640259,
      "learning_rate": 3.437913301799905e-05,
      "loss": 1.9643,
      "step": 194948
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.414490818977356,
      "learning_rate": 3.437723164958543e-05,
      "loss": 1.9928,
      "step": 194949
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.4841961860656738,
      "learning_rate": 3.437533033055612e-05,
      "loss": 2.1486,
      "step": 194950
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.401855230331421,
      "learning_rate": 3.43734290609114e-05,
      "loss": 2.0549,
      "step": 194951
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.3352283239364624,
      "learning_rate": 3.43715278406517e-05,
      "loss": 2.0485,
      "step": 194952
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.4892934560775757,
      "learning_rate": 3.4369626669777305e-05,
      "loss": 2.0193,
      "step": 194953
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.477266788482666,
      "learning_rate": 3.436772554828867e-05,
      "loss": 1.9565,
      "step": 194954
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.3713284730911255,
      "learning_rate": 3.43658244761861e-05,
      "loss": 2.1931,
      "step": 194955
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.263486623764038,
      "learning_rate": 3.4363923453469925e-05,
      "loss": 1.8425,
      "step": 194956
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.5058742761611938,
      "learning_rate": 3.436202248014047e-05,
      "loss": 1.8784,
      "step": 194957
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.462040662765503,
      "learning_rate": 3.436012155619818e-05,
      "loss": 2.0684,
      "step": 194958
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.333545207977295,
      "learning_rate": 3.435822068164329e-05,
      "loss": 1.855,
      "step": 194959
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.523051381111145,
      "learning_rate": 3.4356319856476254e-05,
      "loss": 2.1006,
      "step": 194960
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.4232041835784912,
      "learning_rate": 3.435441908069745e-05,
      "loss": 2.0386,
      "step": 194961
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.5112565755844116,
      "learning_rate": 3.435251835430707e-05,
      "loss": 2.2561,
      "step": 194962
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.4418120384216309,
      "learning_rate": 3.435061767730565e-05,
      "loss": 2.1362,
      "step": 194963
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.3665294647216797,
      "learning_rate": 3.434871704969345e-05,
      "loss": 1.8436,
      "step": 194964
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.4551570415496826,
      "learning_rate": 3.434681647147078e-05,
      "loss": 2.0599,
      "step": 194965
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.4575130939483643,
      "learning_rate": 3.4344915942638144e-05,
      "loss": 1.9556,
      "step": 194966
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.3851763010025024,
      "learning_rate": 3.434301546319579e-05,
      "loss": 1.9919,
      "step": 194967
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.5196075439453125,
      "learning_rate": 3.4341115033143996e-05,
      "loss": 1.9618,
      "step": 194968
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.381952166557312,
      "learning_rate": 3.433921465248329e-05,
      "loss": 1.9809,
      "step": 194969
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.3569756746292114,
      "learning_rate": 3.4337314321213914e-05,
      "loss": 1.9067,
      "step": 194970
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.2941125631332397,
      "learning_rate": 3.43354140393362e-05,
      "loss": 2.1068,
      "step": 194971
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.4967024326324463,
      "learning_rate": 3.43335138068506e-05,
      "loss": 1.9993,
      "step": 194972
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.322165608406067,
      "learning_rate": 3.433161362375739e-05,
      "loss": 1.9431,
      "step": 194973
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.4028714895248413,
      "learning_rate": 3.4329713490056976e-05,
      "loss": 1.9682,
      "step": 194974
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.3781077861785889,
      "learning_rate": 3.432781340574968e-05,
      "loss": 2.0791,
      "step": 194975
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.4067590236663818,
      "learning_rate": 3.4325913370835813e-05,
      "loss": 2.2675,
      "step": 194976
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.388238787651062,
      "learning_rate": 3.432401338531583e-05,
      "loss": 2.0266,
      "step": 194977
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.3071529865264893,
      "learning_rate": 3.4322113449190005e-05,
      "loss": 1.9494,
      "step": 194978
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.319185495376587,
      "learning_rate": 3.4320213562458696e-05,
      "loss": 2.0044,
      "step": 194979
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.3907345533370972,
      "learning_rate": 3.4318313725122316e-05,
      "loss": 2.0091,
      "step": 194980
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.4223871231079102,
      "learning_rate": 3.431641393718115e-05,
      "loss": 2.0765,
      "step": 194981
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.3423612117767334,
      "learning_rate": 3.4314514198635545e-05,
      "loss": 2.0868,
      "step": 194982
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.361345887184143,
      "learning_rate": 3.431261450948596e-05,
      "loss": 1.9205,
      "step": 194983
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.4454210996627808,
      "learning_rate": 3.431071486973259e-05,
      "loss": 2.1034,
      "step": 194984
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.4427045583724976,
      "learning_rate": 3.430881527937594e-05,
      "loss": 1.8831,
      "step": 194985
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.2476320266723633,
      "learning_rate": 3.430691573841628e-05,
      "loss": 1.9254,
      "step": 194986
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.4537761211395264,
      "learning_rate": 3.430501624685401e-05,
      "loss": 1.8519,
      "step": 194987
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.3155285120010376,
      "learning_rate": 3.430311680468939e-05,
      "loss": 2.0735,
      "step": 194988
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.4021786451339722,
      "learning_rate": 3.4301217411922886e-05,
      "loss": 2.0904,
      "step": 194989
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.3286982774734497,
      "learning_rate": 3.4299318068554735e-05,
      "loss": 1.9617,
      "step": 194990
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.5537405014038086,
      "learning_rate": 3.429741877458544e-05,
      "loss": 2.0118,
      "step": 194991
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.4007431268692017,
      "learning_rate": 3.429551953001526e-05,
      "loss": 2.0412,
      "step": 194992
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.4142817258834839,
      "learning_rate": 3.429362033484453e-05,
      "loss": 1.8643,
      "step": 194993
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.3678909540176392,
      "learning_rate": 3.429172118907358e-05,
      "loss": 2.0825,
      "step": 194994
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.5419312715530396,
      "learning_rate": 3.428982209270292e-05,
      "loss": 2.0196,
      "step": 194995
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.4898422956466675,
      "learning_rate": 3.428792304573268e-05,
      "loss": 1.9728,
      "step": 194996
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.4119832515716553,
      "learning_rate": 3.428602404816342e-05,
      "loss": 1.9291,
      "step": 194997
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.3444652557373047,
      "learning_rate": 3.4284125099995405e-05,
      "loss": 2.0016,
      "step": 194998
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.402207851409912,
      "learning_rate": 3.428222620122898e-05,
      "loss": 1.9209,
      "step": 194999
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.411864161491394,
      "learning_rate": 3.4280327351864464e-05,
      "loss": 1.9814,
      "step": 195000
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.4726742506027222,
      "learning_rate": 3.4278428551902305e-05,
      "loss": 2.1174,
      "step": 195001
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.5911294221878052,
      "learning_rate": 3.4276529801342725e-05,
      "loss": 1.9781,
      "step": 195002
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.3644442558288574,
      "learning_rate": 3.427463110018622e-05,
      "loss": 1.8316,
      "step": 195003
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.4521013498306274,
      "learning_rate": 3.427273244843307e-05,
      "loss": 1.9765,
      "step": 195004
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.393668532371521,
      "learning_rate": 3.4270833846083634e-05,
      "loss": 1.8992,
      "step": 195005
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.4964244365692139,
      "learning_rate": 3.426893529313821e-05,
      "loss": 2.3106,
      "step": 195006
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.4103312492370605,
      "learning_rate": 3.4267036789597304e-05,
      "loss": 1.997,
      "step": 195007
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.432088851928711,
      "learning_rate": 3.426513833546107e-05,
      "loss": 1.8878,
      "step": 195008
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.5004907846450806,
      "learning_rate": 3.426323993073002e-05,
      "loss": 2.2097,
      "step": 195009
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.394614815711975,
      "learning_rate": 3.426134157540445e-05,
      "loss": 1.9958,
      "step": 195010
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.545961856842041,
      "learning_rate": 3.4259443269484735e-05,
      "loss": 2.1967,
      "step": 195011
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.3290534019470215,
      "learning_rate": 3.4257545012971156e-05,
      "loss": 1.9311,
      "step": 195012
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.4058256149291992,
      "learning_rate": 3.4255646805864155e-05,
      "loss": 2.1854,
      "step": 195013
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.4146034717559814,
      "learning_rate": 3.425374864816397e-05,
      "loss": 1.7193,
      "step": 195014
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.4235100746154785,
      "learning_rate": 3.425185053987109e-05,
      "loss": 1.9077,
      "step": 195015
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.3573894500732422,
      "learning_rate": 3.424995248098583e-05,
      "loss": 1.7905,
      "step": 195016
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.4760202169418335,
      "learning_rate": 3.4248054471508516e-05,
      "loss": 1.9859,
      "step": 195017
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.3064463138580322,
      "learning_rate": 3.424615651143944e-05,
      "loss": 2.1541,
      "step": 195018
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.3710041046142578,
      "learning_rate": 3.424425860077911e-05,
      "loss": 1.829,
      "step": 195019
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.3810169696807861,
      "learning_rate": 3.424236073952769e-05,
      "loss": 2.0114,
      "step": 195020
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.2630648612976074,
      "learning_rate": 3.4240462927685705e-05,
      "loss": 2.0741,
      "step": 195021
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.3503047227859497,
      "learning_rate": 3.4238565165253437e-05,
      "loss": 2.0566,
      "step": 195022
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.3545116186141968,
      "learning_rate": 3.4236667452231216e-05,
      "loss": 1.9788,
      "step": 195023
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.3864972591400146,
      "learning_rate": 3.423476978861937e-05,
      "loss": 1.9347,
      "step": 195024
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.4890356063842773,
      "learning_rate": 3.423287217441836e-05,
      "loss": 2.0054,
      "step": 195025
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.775168776512146,
      "learning_rate": 3.423097460962842e-05,
      "loss": 2.0689,
      "step": 195026
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.4860320091247559,
      "learning_rate": 3.4229077094250004e-05,
      "loss": 2.0109,
      "step": 195027
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.4624147415161133,
      "learning_rate": 3.4227179628283455e-05,
      "loss": 2.0638,
      "step": 195028
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.388708233833313,
      "learning_rate": 3.422528221172904e-05,
      "loss": 1.8046,
      "step": 195029
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.32963228225708,
      "learning_rate": 3.422338484458714e-05,
      "loss": 1.9603,
      "step": 195030
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.3423858880996704,
      "learning_rate": 3.422148752685818e-05,
      "loss": 1.9853,
      "step": 195031
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.3470232486724854,
      "learning_rate": 3.4219590258542393e-05,
      "loss": 1.8777,
      "step": 195032
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.2877652645111084,
      "learning_rate": 3.421769303964025e-05,
      "loss": 1.799,
      "step": 195033
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.3108277320861816,
      "learning_rate": 3.421579587015207e-05,
      "loss": 1.9415,
      "step": 195034
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.470911979675293,
      "learning_rate": 3.4213898750078204e-05,
      "loss": 2.098,
      "step": 195035
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.360546350479126,
      "learning_rate": 3.421200167941891e-05,
      "loss": 2.1441,
      "step": 195036
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.620349407196045,
      "learning_rate": 3.4210104658174684e-05,
      "loss": 2.0235,
      "step": 195037
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.5294581651687622,
      "learning_rate": 3.420820768634577e-05,
      "loss": 1.8654,
      "step": 195038
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.3346620798110962,
      "learning_rate": 3.4206310763932624e-05,
      "loss": 1.7712,
      "step": 195039
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.3892204761505127,
      "learning_rate": 3.420441389093548e-05,
      "loss": 1.9074,
      "step": 195040
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.2963961362838745,
      "learning_rate": 3.420251706735487e-05,
      "loss": 1.9521,
      "step": 195041
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.6781260967254639,
      "learning_rate": 3.420062029319091e-05,
      "loss": 2.0848,
      "step": 195042
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.5437648296356201,
      "learning_rate": 3.419872356844414e-05,
      "loss": 2.1542,
      "step": 195043
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.617400050163269,
      "learning_rate": 3.419682689311478e-05,
      "loss": 1.9761,
      "step": 195044
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.5712050199508667,
      "learning_rate": 3.419493026720329e-05,
      "loss": 2.0225,
      "step": 195045
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.642622947692871,
      "learning_rate": 3.419303369070993e-05,
      "loss": 2.089,
      "step": 195046
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.5270005464553833,
      "learning_rate": 3.4191137163635185e-05,
      "loss": 1.9908,
      "step": 195047
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.5144239664077759,
      "learning_rate": 3.41892406859793e-05,
      "loss": 2.1305,
      "step": 195048
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.260313630104065,
      "learning_rate": 3.418734425774266e-05,
      "loss": 2.1311,
      "step": 195049
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.4494807720184326,
      "learning_rate": 3.418544787892554e-05,
      "loss": 1.9749,
      "step": 195050
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.4052722454071045,
      "learning_rate": 3.418355154952843e-05,
      "loss": 1.71,
      "step": 195051
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.4102972745895386,
      "learning_rate": 3.4181655269551554e-05,
      "loss": 1.862,
      "step": 195052
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.368562936782837,
      "learning_rate": 3.4179759038995414e-05,
      "loss": 2.1597,
      "step": 195053
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.512345790863037,
      "learning_rate": 3.417786285786024e-05,
      "loss": 2.08,
      "step": 195054
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.4260032176971436,
      "learning_rate": 3.417596672614643e-05,
      "loss": 1.9738,
      "step": 195055
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.3773854970932007,
      "learning_rate": 3.417407064385426e-05,
      "loss": 1.9346,
      "step": 195056
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.38102388381958,
      "learning_rate": 3.417217461098422e-05,
      "loss": 1.9324,
      "step": 195057
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.3257861137390137,
      "learning_rate": 3.417027862753654e-05,
      "loss": 2.1393,
      "step": 195058
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.7181645631790161,
      "learning_rate": 3.416838269351166e-05,
      "loss": 2.0491,
      "step": 195059
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.8668073415756226,
      "learning_rate": 3.4166486808909886e-05,
      "loss": 2.1016,
      "step": 195060
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.4449663162231445,
      "learning_rate": 3.416459097373154e-05,
      "loss": 2.1209,
      "step": 195061
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.4637612104415894,
      "learning_rate": 3.41626951879771e-05,
      "loss": 2.074,
      "step": 195062
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.3759877681732178,
      "learning_rate": 3.416079945164678e-05,
      "loss": 2.0098,
      "step": 195063
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.4326667785644531,
      "learning_rate": 3.415890376474096e-05,
      "loss": 1.9458,
      "step": 195064
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.4594162702560425,
      "learning_rate": 3.415700812726008e-05,
      "loss": 2.1568,
      "step": 195065
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.379612922668457,
      "learning_rate": 3.415511253920439e-05,
      "loss": 1.8905,
      "step": 195066
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.3401380777359009,
      "learning_rate": 3.4153217000574265e-05,
      "loss": 2.067,
      "step": 195067
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.4705612659454346,
      "learning_rate": 3.415132151137013e-05,
      "loss": 2.0204,
      "step": 195068
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.5730011463165283,
      "learning_rate": 3.41494260715922e-05,
      "loss": 1.8968,
      "step": 195069
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.4377225637435913,
      "learning_rate": 3.4147530681241e-05,
      "loss": 2.0433,
      "step": 195070
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.390134572982788,
      "learning_rate": 3.414563534031676e-05,
      "loss": 1.9878,
      "step": 195071
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.4134156703948975,
      "learning_rate": 3.414374004881988e-05,
      "loss": 2.1028,
      "step": 195072
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.344361662864685,
      "learning_rate": 3.414184480675066e-05,
      "loss": 1.7811,
      "step": 195073
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.5036801099777222,
      "learning_rate": 3.413994961410951e-05,
      "loss": 2.1416,
      "step": 195074
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.4934037923812866,
      "learning_rate": 3.413805447089672e-05,
      "loss": 2.1725,
      "step": 195075
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.3133587837219238,
      "learning_rate": 3.413615937711275e-05,
      "loss": 2.0286,
      "step": 195076
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.480605125427246,
      "learning_rate": 3.4134264332757873e-05,
      "loss": 2.0833,
      "step": 195077
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.458066463470459,
      "learning_rate": 3.4132369337832464e-05,
      "loss": 1.8963,
      "step": 195078
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.3898831605911255,
      "learning_rate": 3.4130474392336815e-05,
      "loss": 1.9638,
      "step": 195079
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.4609233140945435,
      "learning_rate": 3.412857949627139e-05,
      "loss": 1.9364,
      "step": 195080
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.4531809091567993,
      "learning_rate": 3.4126684649636394e-05,
      "loss": 1.8911,
      "step": 195081
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.4368867874145508,
      "learning_rate": 3.4124789852432354e-05,
      "loss": 2.0541,
      "step": 195082
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.3673287630081177,
      "learning_rate": 3.412289510465951e-05,
      "loss": 1.9684,
      "step": 195083
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.3990850448608398,
      "learning_rate": 3.4121000406318256e-05,
      "loss": 2.0471,
      "step": 195084
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.4103903770446777,
      "learning_rate": 3.411910575740886e-05,
      "loss": 1.9485,
      "step": 195085
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.45490562915802,
      "learning_rate": 3.411721115793179e-05,
      "loss": 2.2014,
      "step": 195086
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.4995954036712646,
      "learning_rate": 3.411531660788731e-05,
      "loss": 2.1721,
      "step": 195087
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.5417377948760986,
      "learning_rate": 3.411342210727585e-05,
      "loss": 2.0858,
      "step": 195088
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.3930264711380005,
      "learning_rate": 3.4111527656097756e-05,
      "loss": 1.912,
      "step": 195089
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.5875777006149292,
      "learning_rate": 3.410963325435332e-05,
      "loss": 2.3654,
      "step": 195090
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.4341238737106323,
      "learning_rate": 3.410773890204287e-05,
      "loss": 1.8838,
      "step": 195091
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.4357783794403076,
      "learning_rate": 3.410584459916688e-05,
      "loss": 1.938,
      "step": 195092
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.4272414445877075,
      "learning_rate": 3.410395034572557e-05,
      "loss": 1.7988,
      "step": 195093
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.571241855621338,
      "learning_rate": 3.410205614171943e-05,
      "loss": 1.805,
      "step": 195094
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.5552409887313843,
      "learning_rate": 3.41001619871487e-05,
      "loss": 2.1089,
      "step": 195095
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.335296869277954,
      "learning_rate": 3.40982678820138e-05,
      "loss": 2.3374,
      "step": 195096
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.3443351984024048,
      "learning_rate": 3.409637382631495e-05,
      "loss": 1.8435,
      "step": 195097
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.607208013534546,
      "learning_rate": 3.409447982005269e-05,
      "loss": 2.0644,
      "step": 195098
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.4658788442611694,
      "learning_rate": 3.4092585863227253e-05,
      "loss": 2.0782,
      "step": 195099
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.632684588432312,
      "learning_rate": 3.4090691955839036e-05,
      "loss": 2.0863,
      "step": 195100
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.6143109798431396,
      "learning_rate": 3.4088798097888405e-05,
      "loss": 2.0232,
      "step": 195101
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.5300190448760986,
      "learning_rate": 3.408690428937569e-05,
      "loss": 1.8242,
      "step": 195102
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.4758082628250122,
      "learning_rate": 3.408501053030117e-05,
      "loss": 1.955,
      "step": 195103
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.6207889318466187,
      "learning_rate": 3.4083116820665336e-05,
      "loss": 2.1017,
      "step": 195104
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.5826135873794556,
      "learning_rate": 3.4081223160468384e-05,
      "loss": 1.9817,
      "step": 195105
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.3190617561340332,
      "learning_rate": 3.407932954971082e-05,
      "loss": 2.058,
      "step": 195106
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.590558648109436,
      "learning_rate": 3.4077435988392876e-05,
      "loss": 1.9067,
      "step": 195107
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.4829425811767578,
      "learning_rate": 3.4075542476515085e-05,
      "loss": 1.8928,
      "step": 195108
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.4028451442718506,
      "learning_rate": 3.4073649014077544e-05,
      "loss": 2.1445,
      "step": 195109
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.3530892133712769,
      "learning_rate": 3.407175560108079e-05,
      "loss": 1.8893,
      "step": 195110
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.4321010112762451,
      "learning_rate": 3.406986223752505e-05,
      "loss": 1.9148,
      "step": 195111
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.4230867624282837,
      "learning_rate": 3.406796892341083e-05,
      "loss": 1.8618,
      "step": 195112
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.4653065204620361,
      "learning_rate": 3.406607565873829e-05,
      "loss": 1.8463,
      "step": 195113
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.324480414390564,
      "learning_rate": 3.406418244350804e-05,
      "loss": 1.8037,
      "step": 195114
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.5893962383270264,
      "learning_rate": 3.406228927772013e-05,
      "loss": 2.1788,
      "step": 195115
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.4410103559494019,
      "learning_rate": 3.406039616137515e-05,
      "loss": 1.6727,
      "step": 195116
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.4264875650405884,
      "learning_rate": 3.405850309447328e-05,
      "loss": 2.0623,
      "step": 195117
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.3387503623962402,
      "learning_rate": 3.405661007701502e-05,
      "loss": 1.9873,
      "step": 195118
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.467442274093628,
      "learning_rate": 3.4054717109000615e-05,
      "loss": 1.9968,
      "step": 195119
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.4096649885177612,
      "learning_rate": 3.405282419043056e-05,
      "loss": 2.0362,
      "step": 195120
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.3226209878921509,
      "learning_rate": 3.405093132130499e-05,
      "loss": 2.2095,
      "step": 195121
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.4135150909423828,
      "learning_rate": 3.404903850162443e-05,
      "loss": 1.801,
      "step": 195122
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.4305559396743774,
      "learning_rate": 3.4047145731389116e-05,
      "loss": 1.8902,
      "step": 195123
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.4675294160842896,
      "learning_rate": 3.4045253010599515e-05,
      "loss": 1.914,
      "step": 195124
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.3184508085250854,
      "learning_rate": 3.404336033925586e-05,
      "loss": 1.962,
      "step": 195125
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.4559385776519775,
      "learning_rate": 3.4041467717358694e-05,
      "loss": 2.0416,
      "step": 195126
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.3512282371520996,
      "learning_rate": 3.403957514490814e-05,
      "loss": 1.9128,
      "step": 195127
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.3990079164505005,
      "learning_rate": 3.403768262190466e-05,
      "loss": 2.0353,
      "step": 195128
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.5652447938919067,
      "learning_rate": 3.403579014834856e-05,
      "loss": 2.0407,
      "step": 195129
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.5518581867218018,
      "learning_rate": 3.403389772424028e-05,
      "loss": 2.0552,
      "step": 195130
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.5549026727676392,
      "learning_rate": 3.4032005349580075e-05,
      "loss": 1.9851,
      "step": 195131
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.4986735582351685,
      "learning_rate": 3.403011302436838e-05,
      "loss": 2.0149,
      "step": 195132
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.3164124488830566,
      "learning_rate": 3.4028220748605525e-05,
      "loss": 1.9202,
      "step": 195133
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.256675124168396,
      "learning_rate": 3.402632852229186e-05,
      "loss": 1.7533,
      "step": 195134
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.3180086612701416,
      "learning_rate": 3.402443634542763e-05,
      "loss": 2.0659,
      "step": 195135
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.587943196296692,
      "learning_rate": 3.402254421801335e-05,
      "loss": 2.0622,
      "step": 195136
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.4849966764450073,
      "learning_rate": 3.4020652140049207e-05,
      "loss": 1.8395,
      "step": 195137
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.6357383728027344,
      "learning_rate": 3.401876011153575e-05,
      "loss": 2.1641,
      "step": 195138
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.3286046981811523,
      "learning_rate": 3.40168681324732e-05,
      "loss": 2.0013,
      "step": 195139
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.5855499505996704,
      "learning_rate": 3.401497620286195e-05,
      "loss": 2.0638,
      "step": 195140
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.4621002674102783,
      "learning_rate": 3.401308432270229e-05,
      "loss": 2.2136,
      "step": 195141
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.430573582649231,
      "learning_rate": 3.4011192491994665e-05,
      "loss": 1.8158,
      "step": 195142
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.3174856901168823,
      "learning_rate": 3.4009300710739315e-05,
      "loss": 1.9213,
      "step": 195143
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.6217551231384277,
      "learning_rate": 3.400740897893671e-05,
      "loss": 1.9557,
      "step": 195144
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.4159595966339111,
      "learning_rate": 3.400551729658715e-05,
      "loss": 1.8309,
      "step": 195145
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.5051106214523315,
      "learning_rate": 3.400362566369093e-05,
      "loss": 2.1925,
      "step": 195146
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.8292241096496582,
      "learning_rate": 3.400173408024851e-05,
      "loss": 1.8841,
      "step": 195147
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.5478851795196533,
      "learning_rate": 3.39998425462602e-05,
      "loss": 1.8171,
      "step": 195148
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.5620313882827759,
      "learning_rate": 3.39979510617263e-05,
      "loss": 1.9675,
      "step": 195149
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.3915272951126099,
      "learning_rate": 3.3996059626647245e-05,
      "loss": 2.0892,
      "step": 195150
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.3748351335525513,
      "learning_rate": 3.399416824102332e-05,
      "loss": 2.0822,
      "step": 195151
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.4514472484588623,
      "learning_rate": 3.3992276904854876e-05,
      "loss": 1.9368,
      "step": 195152
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.5273332595825195,
      "learning_rate": 3.3990385618142335e-05,
      "loss": 1.785,
      "step": 195153
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.3845698833465576,
      "learning_rate": 3.3988494380886e-05,
      "loss": 2.1177,
      "step": 195154
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.4822473526000977,
      "learning_rate": 3.3986603193086174e-05,
      "loss": 1.9305,
      "step": 195155
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.3474258184432983,
      "learning_rate": 3.398471205474328e-05,
      "loss": 2.1496,
      "step": 195156
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.2753989696502686,
      "learning_rate": 3.3982820965857694e-05,
      "loss": 1.9499,
      "step": 195157
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.3582103252410889,
      "learning_rate": 3.3980929926429646e-05,
      "loss": 2.185,
      "step": 195158
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.3892935514450073,
      "learning_rate": 3.3979038936459635e-05,
      "loss": 1.9343,
      "step": 195159
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.5231455564498901,
      "learning_rate": 3.3977147995947864e-05,
      "loss": 2.0421,
      "step": 195160
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.8244926929473877,
      "learning_rate": 3.397525710489486e-05,
      "loss": 2.1662,
      "step": 195161
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.269055724143982,
      "learning_rate": 3.3973366263300825e-05,
      "loss": 1.7833,
      "step": 195162
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.420405387878418,
      "learning_rate": 3.39714754711662e-05,
      "loss": 1.9903,
      "step": 195163
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.4748153686523438,
      "learning_rate": 3.396958472849124e-05,
      "loss": 1.8217,
      "step": 195164
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.3378708362579346,
      "learning_rate": 3.3967694035276415e-05,
      "loss": 1.9607,
      "step": 195165
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.5013946294784546,
      "learning_rate": 3.3965803391521926e-05,
      "loss": 2.1623,
      "step": 195166
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.4275352954864502,
      "learning_rate": 3.396391279722831e-05,
      "loss": 1.8855,
      "step": 195167
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.4388593435287476,
      "learning_rate": 3.396202225239579e-05,
      "loss": 2.1354,
      "step": 195168
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.6443251371383667,
      "learning_rate": 3.396013175702478e-05,
      "loss": 2.1885,
      "step": 195169
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.3328853845596313,
      "learning_rate": 3.395824131111553e-05,
      "loss": 2.0698,
      "step": 195170
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.456183671951294,
      "learning_rate": 3.395635091466855e-05,
      "loss": 1.8962,
      "step": 195171
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.31163489818573,
      "learning_rate": 3.395446056768404e-05,
      "loss": 1.8313,
      "step": 195172
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.4438228607177734,
      "learning_rate": 3.395257027016246e-05,
      "loss": 2.0639,
      "step": 195173
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.3646879196166992,
      "learning_rate": 3.395068002210405e-05,
      "loss": 1.8954,
      "step": 195174
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.3796578645706177,
      "learning_rate": 3.394878982350937e-05,
      "loss": 1.9201,
      "step": 195175
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.3295525312423706,
      "learning_rate": 3.3946899674378524e-05,
      "loss": 1.8175,
      "step": 195176
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.5765219926834106,
      "learning_rate": 3.394500957471201e-05,
      "loss": 1.8425,
      "step": 195177
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.4053128957748413,
      "learning_rate": 3.39431195245101e-05,
      "loss": 1.9823,
      "step": 195178
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.284964680671692,
      "learning_rate": 3.3941229523773217e-05,
      "loss": 1.7425,
      "step": 195179
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.4002324342727661,
      "learning_rate": 3.3939339572501636e-05,
      "loss": 2.1637,
      "step": 195180
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.3981623649597168,
      "learning_rate": 3.393744967069588e-05,
      "loss": 1.7754,
      "step": 195181
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.496716856956482,
      "learning_rate": 3.393555981835606e-05,
      "loss": 2.2122,
      "step": 195182
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.3493646383285522,
      "learning_rate": 3.393367001548267e-05,
      "loss": 2.0658,
      "step": 195183
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.345617413520813,
      "learning_rate": 3.393178026207602e-05,
      "loss": 2.0028,
      "step": 195184
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.3163540363311768,
      "learning_rate": 3.3929890558136485e-05,
      "loss": 1.9265,
      "step": 195185
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.4670321941375732,
      "learning_rate": 3.392800090366439e-05,
      "loss": 2.0378,
      "step": 195186
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.417199730873108,
      "learning_rate": 3.392611129866019e-05,
      "loss": 2.0999,
      "step": 195187
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.3993726968765259,
      "learning_rate": 3.392422174312405e-05,
      "loss": 1.8641,
      "step": 195188
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.5129873752593994,
      "learning_rate": 3.392233223705647e-05,
      "loss": 2.0343,
      "step": 195189
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.6048458814620972,
      "learning_rate": 3.392044278045769e-05,
      "loss": 2.1347,
      "step": 195190
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.3573517799377441,
      "learning_rate": 3.3918553373328175e-05,
      "loss": 2.0141,
      "step": 195191
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.3286149501800537,
      "learning_rate": 3.391666401566818e-05,
      "loss": 2.1515,
      "step": 195192
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.3670802116394043,
      "learning_rate": 3.391477470747822e-05,
      "loss": 2.2654,
      "step": 195193
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.3783400058746338,
      "learning_rate": 3.3912885448758385e-05,
      "loss": 2.2333,
      "step": 195194
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.5480804443359375,
      "learning_rate": 3.391099623950925e-05,
      "loss": 1.8136,
      "step": 195195
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.6098884344100952,
      "learning_rate": 3.3909107079731004e-05,
      "loss": 2.0092,
      "step": 195196
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.485986590385437,
      "learning_rate": 3.390721796942415e-05,
      "loss": 1.7288,
      "step": 195197
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.3740397691726685,
      "learning_rate": 3.390532890858889e-05,
      "loss": 2.0131,
      "step": 195198
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.2970569133758545,
      "learning_rate": 3.3903439897225793e-05,
      "loss": 1.8398,
      "step": 195199
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.4465618133544922,
      "learning_rate": 3.390155093533495e-05,
      "loss": 2.0823,
      "step": 195200
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.298201560974121,
      "learning_rate": 3.389966202291691e-05,
      "loss": 1.8278,
      "step": 195201
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.4409399032592773,
      "learning_rate": 3.3897773159971854e-05,
      "loss": 1.9131,
      "step": 195202
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.3484468460083008,
      "learning_rate": 3.389588434650029e-05,
      "loss": 2.0254,
      "step": 195203
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.5867295265197754,
      "learning_rate": 3.3893995582502455e-05,
      "loss": 2.0099,
      "step": 195204
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.3935470581054688,
      "learning_rate": 3.389210686797884e-05,
      "loss": 1.9875,
      "step": 195205
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.4374988079071045,
      "learning_rate": 3.389021820292962e-05,
      "loss": 2.0496,
      "step": 195206
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.385025143623352,
      "learning_rate": 3.388832958735529e-05,
      "loss": 2.1448,
      "step": 195207
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.4543615579605103,
      "learning_rate": 3.388644102125605e-05,
      "loss": 1.7561,
      "step": 195208
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.435351014137268,
      "learning_rate": 3.3884552504632434e-05,
      "loss": 1.9426,
      "step": 195209
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.4705514907836914,
      "learning_rate": 3.388266403748464e-05,
      "loss": 1.8857,
      "step": 195210
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.37489914894104,
      "learning_rate": 3.38807756198132e-05,
      "loss": 2.0465,
      "step": 195211
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.4064509868621826,
      "learning_rate": 3.387888725161823e-05,
      "loss": 1.8501,
      "step": 195212
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.45144784450531,
      "learning_rate": 3.387699893290024e-05,
      "loss": 1.9777,
      "step": 195213
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.5143489837646484,
      "learning_rate": 3.38751106636595e-05,
      "loss": 2.3618,
      "step": 195214
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.692014217376709,
      "learning_rate": 3.387322244389645e-05,
      "loss": 1.9724,
      "step": 195215
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.3172410726547241,
      "learning_rate": 3.387133427361133e-05,
      "loss": 1.8069,
      "step": 195216
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.3576087951660156,
      "learning_rate": 3.386944615280466e-05,
      "loss": 2.2043,
      "step": 195217
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.3985151052474976,
      "learning_rate": 3.386755808147657e-05,
      "loss": 1.9119,
      "step": 195218
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.453676462173462,
      "learning_rate": 3.386567005962762e-05,
      "loss": 1.813,
      "step": 195219
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.38906729221344,
      "learning_rate": 3.3863782087257977e-05,
      "loss": 1.8327,
      "step": 195220
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.30418062210083,
      "learning_rate": 3.386189416436812e-05,
      "loss": 1.8531,
      "step": 195221
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.4681322574615479,
      "learning_rate": 3.386000629095835e-05,
      "loss": 1.7766,
      "step": 195222
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.4066822528839111,
      "learning_rate": 3.385811846702904e-05,
      "loss": 1.9588,
      "step": 195223
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.3643697500228882,
      "learning_rate": 3.385623069258058e-05,
      "loss": 1.6514,
      "step": 195224
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.3039138317108154,
      "learning_rate": 3.385434296761324e-05,
      "loss": 1.9988,
      "step": 195225
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.4822100400924683,
      "learning_rate": 3.385245529212731e-05,
      "loss": 1.83,
      "step": 195226
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.480573058128357,
      "learning_rate": 3.385056766612334e-05,
      "loss": 1.9464,
      "step": 195227
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.3547577857971191,
      "learning_rate": 3.384868008960149e-05,
      "loss": 1.7242,
      "step": 195228
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.4064565896987915,
      "learning_rate": 3.3846792562562286e-05,
      "loss": 2.0561,
      "step": 195229
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.2384717464447021,
      "learning_rate": 3.384490508500597e-05,
      "loss": 1.8738,
      "step": 195230
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.3286694288253784,
      "learning_rate": 3.3843017656932836e-05,
      "loss": 1.8348,
      "step": 195231
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.3897536993026733,
      "learning_rate": 3.384113027834339e-05,
      "loss": 1.9514,
      "step": 195232
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.777783989906311,
      "learning_rate": 3.383924294923789e-05,
      "loss": 2.076,
      "step": 195233
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.4901710748672485,
      "learning_rate": 3.383735566961664e-05,
      "loss": 2.1402,
      "step": 195234
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.4456204175949097,
      "learning_rate": 3.383546843948011e-05,
      "loss": 2.0468,
      "step": 195235
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.349679946899414,
      "learning_rate": 3.383358125882859e-05,
      "loss": 2.0809,
      "step": 195236
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.4554431438446045,
      "learning_rate": 3.383169412766236e-05,
      "loss": 1.9924,
      "step": 195237
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.362838864326477,
      "learning_rate": 3.382980704598191e-05,
      "loss": 1.8103,
      "step": 195238
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.4772120714187622,
      "learning_rate": 3.382792001378754e-05,
      "loss": 1.9643,
      "step": 195239
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.5392460823059082,
      "learning_rate": 3.382603303107951e-05,
      "loss": 1.8848,
      "step": 195240
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.439637303352356,
      "learning_rate": 3.3824146097858305e-05,
      "loss": 1.945,
      "step": 195241
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.328663945198059,
      "learning_rate": 3.3822259214124214e-05,
      "loss": 1.9549,
      "step": 195242
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.3323078155517578,
      "learning_rate": 3.3820372379877535e-05,
      "loss": 1.94,
      "step": 195243
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.5860615968704224,
      "learning_rate": 3.3818485595118736e-05,
      "loss": 1.9404,
      "step": 195244
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.4718730449676514,
      "learning_rate": 3.381659885984802e-05,
      "loss": 2.0539,
      "step": 195245
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.314960241317749,
      "learning_rate": 3.3814712174065914e-05,
      "loss": 1.8948,
      "step": 195246
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.3214876651763916,
      "learning_rate": 3.381282553777266e-05,
      "loss": 2.0432,
      "step": 195247
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.4953986406326294,
      "learning_rate": 3.381093895096865e-05,
      "loss": 2.0966,
      "step": 195248
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.471362590789795,
      "learning_rate": 3.380905241365411e-05,
      "loss": 1.8561,
      "step": 195249
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.478866457939148,
      "learning_rate": 3.380716592582959e-05,
      "loss": 1.9741,
      "step": 195250
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.368411660194397,
      "learning_rate": 3.380527948749525e-05,
      "loss": 1.9653,
      "step": 195251
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.3337793350219727,
      "learning_rate": 3.380339309865162e-05,
      "loss": 1.8112,
      "step": 195252
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.5743610858917236,
      "learning_rate": 3.380150675929897e-05,
      "loss": 2.1019,
      "step": 195253
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.3152109384536743,
      "learning_rate": 3.379962046943763e-05,
      "loss": 1.9601,
      "step": 195254
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.4406194686889648,
      "learning_rate": 3.3797734229067893e-05,
      "loss": 2.1033,
      "step": 195255
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.4914604425430298,
      "learning_rate": 3.379584803819027e-05,
      "loss": 2.176,
      "step": 195256
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.396501064300537,
      "learning_rate": 3.379396189680493e-05,
      "loss": 1.8972,
      "step": 195257
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.54098641872406,
      "learning_rate": 3.37920758049124e-05,
      "loss": 2.0139,
      "step": 195258
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.4994889497756958,
      "learning_rate": 3.379018976251291e-05,
      "loss": 2.064,
      "step": 195259
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.3928180932998657,
      "learning_rate": 3.378830376960693e-05,
      "loss": 1.904,
      "step": 195260
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.3656315803527832,
      "learning_rate": 3.378641782619462e-05,
      "loss": 1.9967,
      "step": 195261
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.3426153659820557,
      "learning_rate": 3.378453193227649e-05,
      "loss": 1.8969,
      "step": 195262
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.5025566816329956,
      "learning_rate": 3.3782646087852804e-05,
      "loss": 2.038,
      "step": 195263
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.4424234628677368,
      "learning_rate": 3.3780760292923994e-05,
      "loss": 2.143,
      "step": 195264
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.4971833229064941,
      "learning_rate": 3.377887454749032e-05,
      "loss": 2.2285,
      "step": 195265
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.411226511001587,
      "learning_rate": 3.377698885155229e-05,
      "loss": 1.9805,
      "step": 195266
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.4301135540008545,
      "learning_rate": 3.3775103205110034e-05,
      "loss": 2.2864,
      "step": 195267
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.5499939918518066,
      "learning_rate": 3.377321760816405e-05,
      "loss": 1.9775,
      "step": 195268
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.4337115287780762,
      "learning_rate": 3.3771332060714604e-05,
      "loss": 2.0451,
      "step": 195269
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.5036789178848267,
      "learning_rate": 3.376944656276214e-05,
      "loss": 1.9947,
      "step": 195270
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.5042273998260498,
      "learning_rate": 3.3767561114306905e-05,
      "loss": 1.9461,
      "step": 195271
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.6401785612106323,
      "learning_rate": 3.376567571534945e-05,
      "loss": 1.961,
      "step": 195272
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.4592820405960083,
      "learning_rate": 3.376379036588983e-05,
      "loss": 1.8618,
      "step": 195273
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.4220150709152222,
      "learning_rate": 3.376190506592862e-05,
      "loss": 2.0041,
      "step": 195274
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.3796932697296143,
      "learning_rate": 3.376001981546604e-05,
      "loss": 1.7828,
      "step": 195275
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.484106421470642,
      "learning_rate": 3.3758134614502573e-05,
      "loss": 1.8804,
      "step": 195276
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.500718116760254,
      "learning_rate": 3.375624946303841e-05,
      "loss": 2.0625,
      "step": 195277
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.4235961437225342,
      "learning_rate": 3.375436436107412e-05,
      "loss": 1.987,
      "step": 195278
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.5629186630249023,
      "learning_rate": 3.3752479308609795e-05,
      "loss": 2.0308,
      "step": 195279
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.4199949502944946,
      "learning_rate": 3.375059430564597e-05,
      "loss": 1.8976,
      "step": 195280
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.4598667621612549,
      "learning_rate": 3.374870935218286e-05,
      "loss": 2.1632,
      "step": 195281
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.3935093879699707,
      "learning_rate": 3.3746824448220975e-05,
      "loss": 1.9094,
      "step": 195282
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.4083802700042725,
      "learning_rate": 3.3744939593760504e-05,
      "loss": 1.8525,
      "step": 195283
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.4648106098175049,
      "learning_rate": 3.374305478880199e-05,
      "loss": 2.145,
      "step": 195284
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.417606234550476,
      "learning_rate": 3.374117003334555e-05,
      "loss": 1.8974,
      "step": 195285
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.499070167541504,
      "learning_rate": 3.373928532739175e-05,
      "loss": 2.0838,
      "step": 195286
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.5069100856781006,
      "learning_rate": 3.3737400670940754e-05,
      "loss": 1.9338,
      "step": 195287
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.4218006134033203,
      "learning_rate": 3.3735516063993054e-05,
      "loss": 1.8803,
      "step": 195288
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.443001627922058,
      "learning_rate": 3.373363150654889e-05,
      "loss": 2.152,
      "step": 195289
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.407079815864563,
      "learning_rate": 3.373174699860879e-05,
      "loss": 1.9558,
      "step": 195290
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.4511030912399292,
      "learning_rate": 3.372986254017286e-05,
      "loss": 2.1489,
      "step": 195291
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.3940143585205078,
      "learning_rate": 3.372797813124166e-05,
      "loss": 1.9397,
      "step": 195292
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.3601773977279663,
      "learning_rate": 3.372609377181537e-05,
      "loss": 2.0282,
      "step": 195293
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.4648936986923218,
      "learning_rate": 3.37242094618945e-05,
      "loss": 2.1771,
      "step": 195294
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.3812357187271118,
      "learning_rate": 3.372232520147927e-05,
      "loss": 1.9545,
      "step": 195295
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.409307837486267,
      "learning_rate": 3.372044099057021e-05,
      "loss": 2.1395,
      "step": 195296
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.3386610746383667,
      "learning_rate": 3.371855682916741e-05,
      "loss": 2.0566,
      "step": 195297
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.3823816776275635,
      "learning_rate": 3.371667271727141e-05,
      "loss": 2.0099,
      "step": 195298
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.3138768672943115,
      "learning_rate": 3.3714788654882444e-05,
      "loss": 1.9758,
      "step": 195299
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.3877402544021606,
      "learning_rate": 3.3712904642001004e-05,
      "loss": 2.0648,
      "step": 195300
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.3379212617874146,
      "learning_rate": 3.37110206786273e-05,
      "loss": 1.9962,
      "step": 195301
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.465572476387024,
      "learning_rate": 3.3709136764761854e-05,
      "loss": 2.1526,
      "step": 195302
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.4765228033065796,
      "learning_rate": 3.3707252900404815e-05,
      "loss": 1.8654,
      "step": 195303
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.448009729385376,
      "learning_rate": 3.3705369085556636e-05,
      "loss": 2.0857,
      "step": 195304
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.3321114778518677,
      "learning_rate": 3.370348532021761e-05,
      "loss": 1.9144,
      "step": 195305
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.6184027194976807,
      "learning_rate": 3.370160160438823e-05,
      "loss": 1.9234,
      "step": 195306
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.4340357780456543,
      "learning_rate": 3.369971793806867e-05,
      "loss": 2.0539,
      "step": 195307
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.5015597343444824,
      "learning_rate": 3.369783432125941e-05,
      "loss": 1.9635,
      "step": 195308
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.3222804069519043,
      "learning_rate": 3.3695950753960744e-05,
      "loss": 1.8268,
      "step": 195309
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.2784405946731567,
      "learning_rate": 3.3694067236173036e-05,
      "loss": 1.8451,
      "step": 195310
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.321211338043213,
      "learning_rate": 3.3692183767896566e-05,
      "loss": 1.9209,
      "step": 195311
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.667668104171753,
      "learning_rate": 3.3690300349131814e-05,
      "loss": 1.9666,
      "step": 195312
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.3600085973739624,
      "learning_rate": 3.3688416979879e-05,
      "loss": 2.0407,
      "step": 195313
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.3981579542160034,
      "learning_rate": 3.368653366013857e-05,
      "loss": 1.9666,
      "step": 195314
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.428929090499878,
      "learning_rate": 3.368465038991087e-05,
      "loss": 1.8751,
      "step": 195315
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.3555785417556763,
      "learning_rate": 3.36827671691962e-05,
      "loss": 1.7996,
      "step": 195316
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.4454551935195923,
      "learning_rate": 3.368088399799489e-05,
      "loss": 2.2376,
      "step": 195317
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.3550982475280762,
      "learning_rate": 3.36790008763074e-05,
      "loss": 2.0057,
      "step": 195318
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.3930597305297852,
      "learning_rate": 3.367711780413391e-05,
      "loss": 1.9357,
      "step": 195319
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.5193735361099243,
      "learning_rate": 3.367523478147498e-05,
      "loss": 2.0733,
      "step": 195320
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.4346897602081299,
      "learning_rate": 3.367335180833081e-05,
      "loss": 1.9849,
      "step": 195321
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.3305672407150269,
      "learning_rate": 3.367146888470176e-05,
      "loss": 2.0827,
      "step": 195322
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.5395548343658447,
      "learning_rate": 3.3669586010588266e-05,
      "loss": 1.9763,
      "step": 195323
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.470421314239502,
      "learning_rate": 3.3667703185990605e-05,
      "loss": 1.9775,
      "step": 195324
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.3786786794662476,
      "learning_rate": 3.366582041090909e-05,
      "loss": 1.9074,
      "step": 195325
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.4275306463241577,
      "learning_rate": 3.366393768534421e-05,
      "loss": 1.9418,
      "step": 195326
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.4430781602859497,
      "learning_rate": 3.366205500929622e-05,
      "loss": 1.8263,
      "step": 195327
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.4170640707015991,
      "learning_rate": 3.366017238276542e-05,
      "loss": 2.0329,
      "step": 195328
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.3724168539047241,
      "learning_rate": 3.365828980575227e-05,
      "loss": 2.037,
      "step": 195329
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.4912827014923096,
      "learning_rate": 3.3656407278257016e-05,
      "loss": 2.2388,
      "step": 195330
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.3225568532943726,
      "learning_rate": 3.365452480028012e-05,
      "loss": 1.8919,
      "step": 195331
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.3436564207077026,
      "learning_rate": 3.36526423718219e-05,
      "loss": 1.9352,
      "step": 195332
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.3276290893554688,
      "learning_rate": 3.365075999288265e-05,
      "loss": 1.9989,
      "step": 195333
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.5182597637176514,
      "learning_rate": 3.364887766346271e-05,
      "loss": 2.0337,
      "step": 195334
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.4418246746063232,
      "learning_rate": 3.364699538356254e-05,
      "loss": 1.881,
      "step": 195335
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.640108346939087,
      "learning_rate": 3.364511315318235e-05,
      "loss": 2.1095,
      "step": 195336
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.6346697807312012,
      "learning_rate": 3.364323097232261e-05,
      "loss": 2.2921,
      "step": 195337
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.373281717300415,
      "learning_rate": 3.364134884098363e-05,
      "loss": 1.9843,
      "step": 195338
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.3143502473831177,
      "learning_rate": 3.3639466759165736e-05,
      "loss": 1.8255,
      "step": 195339
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.3347904682159424,
      "learning_rate": 3.363758472686926e-05,
      "loss": 1.9837,
      "step": 195340
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.520353078842163,
      "learning_rate": 3.363570274409464e-05,
      "loss": 2.0612,
      "step": 195341
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.3822576999664307,
      "learning_rate": 3.363382081084211e-05,
      "loss": 1.972,
      "step": 195342
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.5930819511413574,
      "learning_rate": 3.363193892711212e-05,
      "loss": 2.0166,
      "step": 195343
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.3489032983779907,
      "learning_rate": 3.363005709290493e-05,
      "loss": 1.7089,
      "step": 195344
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.4391365051269531,
      "learning_rate": 3.362817530822108e-05,
      "loss": 1.9935,
      "step": 195345
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.4863680601119995,
      "learning_rate": 3.3626293573060657e-05,
      "loss": 2.1823,
      "step": 195346
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.4690707921981812,
      "learning_rate": 3.362441188742419e-05,
      "loss": 2.1023,
      "step": 195347
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.4533087015151978,
      "learning_rate": 3.36225302513119e-05,
      "loss": 1.8022,
      "step": 195348
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.3130379915237427,
      "learning_rate": 3.3620648664724256e-05,
      "loss": 1.9977,
      "step": 195349
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.3635082244873047,
      "learning_rate": 3.361876712766154e-05,
      "loss": 2.1339,
      "step": 195350
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.3480459451675415,
      "learning_rate": 3.361688564012424e-05,
      "loss": 2.025,
      "step": 195351
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.397459864616394,
      "learning_rate": 3.361500420211245e-05,
      "loss": 2.0027,
      "step": 195352
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.3757600784301758,
      "learning_rate": 3.361312281362671e-05,
      "loss": 2.0228,
      "step": 195353
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.3335237503051758,
      "learning_rate": 3.3611241474667296e-05,
      "loss": 2.1119,
      "step": 195354
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.4699029922485352,
      "learning_rate": 3.3609360185234594e-05,
      "loss": 1.9472,
      "step": 195355
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.4436818361282349,
      "learning_rate": 3.3607478945328904e-05,
      "loss": 1.9753,
      "step": 195356
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.647239327430725,
      "learning_rate": 3.360559775495076e-05,
      "loss": 1.9527,
      "step": 195357
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.4084725379943848,
      "learning_rate": 3.360371661410021e-05,
      "loss": 1.9561,
      "step": 195358
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.4554872512817383,
      "learning_rate": 3.360183552277783e-05,
      "loss": 1.998,
      "step": 195359
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.4286718368530273,
      "learning_rate": 3.3599954480983835e-05,
      "loss": 1.8855,
      "step": 195360
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.481814980506897,
      "learning_rate": 3.359807348871869e-05,
      "loss": 1.772,
      "step": 195361
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.3880705833435059,
      "learning_rate": 3.3596192545982625e-05,
      "loss": 2.2682,
      "step": 195362
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.4648396968841553,
      "learning_rate": 3.359431165277621e-05,
      "loss": 1.6753,
      "step": 195363
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.3719170093536377,
      "learning_rate": 3.359243080909947e-05,
      "loss": 1.7996,
      "step": 195364
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.4488049745559692,
      "learning_rate": 3.3590550014953045e-05,
      "loss": 2.0097,
      "step": 195365
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.4250929355621338,
      "learning_rate": 3.3588669270337065e-05,
      "loss": 1.9556,
      "step": 195366
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.464026689529419,
      "learning_rate": 3.3586788575252034e-05,
      "loss": 1.9873,
      "step": 195367
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.4660108089447021,
      "learning_rate": 3.358490792969821e-05,
      "loss": 1.8825,
      "step": 195368
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.4112297296524048,
      "learning_rate": 3.358302733367611e-05,
      "loss": 2.0357,
      "step": 195369
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.6610554456710815,
      "learning_rate": 3.358114678718581e-05,
      "loss": 1.9501,
      "step": 195370
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.419458031654358,
      "learning_rate": 3.357926629022789e-05,
      "loss": 1.9028,
      "step": 195371
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.5266324281692505,
      "learning_rate": 3.357738584280252e-05,
      "loss": 2.0384,
      "step": 195372
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.324433445930481,
      "learning_rate": 3.357550544491022e-05,
      "loss": 1.9725,
      "step": 195373
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.319562554359436,
      "learning_rate": 3.357362509655121e-05,
      "loss": 2.1478,
      "step": 195374
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.3200651407241821,
      "learning_rate": 3.357174479772593e-05,
      "loss": 1.9348,
      "step": 195375
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.3842793703079224,
      "learning_rate": 3.356986454843473e-05,
      "loss": 2.216,
      "step": 195376
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.652937650680542,
      "learning_rate": 3.356798434867788e-05,
      "loss": 2.2664,
      "step": 195377
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.3532432317733765,
      "learning_rate": 3.3566104198455735e-05,
      "loss": 1.9393,
      "step": 195378
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.419108510017395,
      "learning_rate": 3.3564224097768736e-05,
      "loss": 2.1017,
      "step": 195379
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.545840859413147,
      "learning_rate": 3.3562344046617105e-05,
      "loss": 1.9217,
      "step": 195380
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.462725043296814,
      "learning_rate": 3.3560464045001356e-05,
      "loss": 2.1481,
      "step": 195381
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.2672909498214722,
      "learning_rate": 3.3558584092921714e-05,
      "loss": 1.8844,
      "step": 195382
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.4404646158218384,
      "learning_rate": 3.355670419037858e-05,
      "loss": 2.1852,
      "step": 195383
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.5519381761550903,
      "learning_rate": 3.355482433737222e-05,
      "loss": 1.9557,
      "step": 195384
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.4803147315979004,
      "learning_rate": 3.3552944533903104e-05,
      "loss": 1.9556,
      "step": 195385
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.4307355880737305,
      "learning_rate": 3.355106477997146e-05,
      "loss": 2.0452,
      "step": 195386
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.3690704107284546,
      "learning_rate": 3.3549185075577754e-05,
      "loss": 2.0788,
      "step": 195387
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.5573790073394775,
      "learning_rate": 3.3547305420722326e-05,
      "loss": 2.0115,
      "step": 195388
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.5169963836669922,
      "learning_rate": 3.3545425815405436e-05,
      "loss": 1.9511,
      "step": 195389
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.4427094459533691,
      "learning_rate": 3.354354625962745e-05,
      "loss": 2.0182,
      "step": 195390
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.4944685697555542,
      "learning_rate": 3.3541666753388804e-05,
      "loss": 1.8609,
      "step": 195391
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.3651710748672485,
      "learning_rate": 3.35397872966897e-05,
      "loss": 1.8948,
      "step": 195392
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.4513800144195557,
      "learning_rate": 3.3537907889530666e-05,
      "loss": 1.9048,
      "step": 195393
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.4328668117523193,
      "learning_rate": 3.353602853191197e-05,
      "loss": 1.9199,
      "step": 195394
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.53934907913208,
      "learning_rate": 3.353414922383395e-05,
      "loss": 1.9544,
      "step": 195395
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.4837806224822998,
      "learning_rate": 3.3532269965296933e-05,
      "loss": 1.9296,
      "step": 195396
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.4659837484359741,
      "learning_rate": 3.353039075630132e-05,
      "loss": 1.9013,
      "step": 195397
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.3746305704116821,
      "learning_rate": 3.352851159684737e-05,
      "loss": 2.09,
      "step": 195398
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.622889518737793,
      "learning_rate": 3.3526632486935564e-05,
      "loss": 2.0986,
      "step": 195399
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.532430648803711,
      "learning_rate": 3.35247534265662e-05,
      "loss": 2.0768,
      "step": 195400
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.5432188510894775,
      "learning_rate": 3.352287441573963e-05,
      "loss": 1.8135,
      "step": 195401
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.360718846321106,
      "learning_rate": 3.35209954544561e-05,
      "loss": 2.0265,
      "step": 195402
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.4493868350982666,
      "learning_rate": 3.3519116542716106e-05,
      "loss": 1.9442,
      "step": 195403
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.3947672843933105,
      "learning_rate": 3.351723768051992e-05,
      "loss": 2.1158,
      "step": 195404
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.5031720399856567,
      "learning_rate": 3.3515358867867924e-05,
      "loss": 1.9651,
      "step": 195405
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.392372488975525,
      "learning_rate": 3.351348010476047e-05,
      "loss": 2.0102,
      "step": 195406
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.3848711252212524,
      "learning_rate": 3.351160139119782e-05,
      "loss": 2.0414,
      "step": 195407
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.288464069366455,
      "learning_rate": 3.3509722727180464e-05,
      "loss": 1.9262,
      "step": 195408
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.2873305082321167,
      "learning_rate": 3.350784411270868e-05,
      "loss": 2.1647,
      "step": 195409
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.4011125564575195,
      "learning_rate": 3.3505965547782765e-05,
      "loss": 1.9196,
      "step": 195410
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.483812689781189,
      "learning_rate": 3.350408703240315e-05,
      "loss": 1.8467,
      "step": 195411
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.4401671886444092,
      "learning_rate": 3.35022085665702e-05,
      "loss": 2.138,
      "step": 195412
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.4185264110565186,
      "learning_rate": 3.350033015028411e-05,
      "loss": 2.032,
      "step": 195413
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.4657843112945557,
      "learning_rate": 3.349845178354547e-05,
      "loss": 1.9935,
      "step": 195414
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.3573986291885376,
      "learning_rate": 3.3496573466354447e-05,
      "loss": 1.8422,
      "step": 195415
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.5143730640411377,
      "learning_rate": 3.34946951987114e-05,
      "loss": 2.06,
      "step": 195416
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.374651551246643,
      "learning_rate": 3.349281698061674e-05,
      "loss": 1.9106,
      "step": 195417
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.5041438341140747,
      "learning_rate": 3.3490938812070855e-05,
      "loss": 2.3036,
      "step": 195418
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.3463650941848755,
      "learning_rate": 3.348906069307393e-05,
      "loss": 2.0979,
      "step": 195419
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.4620050191879272,
      "learning_rate": 3.348718262362651e-05,
      "loss": 1.832,
      "step": 195420
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.365421175956726,
      "learning_rate": 3.348530460372878e-05,
      "loss": 2.1287,
      "step": 195421
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.3684557676315308,
      "learning_rate": 3.3483426633381215e-05,
      "loss": 1.9546,
      "step": 195422
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.3970407247543335,
      "learning_rate": 3.348154871258412e-05,
      "loss": 2.029,
      "step": 195423
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.401476263999939,
      "learning_rate": 3.3479670841337845e-05,
      "loss": 1.9557,
      "step": 195424
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.324021339416504,
      "learning_rate": 3.347779301964264e-05,
      "loss": 1.9542,
      "step": 195425
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.302895426750183,
      "learning_rate": 3.347591524749903e-05,
      "loss": 1.9547,
      "step": 195426
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.476887583732605,
      "learning_rate": 3.347403752490722e-05,
      "loss": 2.1763,
      "step": 195427
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.3260899782180786,
      "learning_rate": 3.3472159851867665e-05,
      "loss": 1.8853,
      "step": 195428
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.3562779426574707,
      "learning_rate": 3.347028222838061e-05,
      "loss": 1.9371,
      "step": 195429
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.3242124319076538,
      "learning_rate": 3.346840465444658e-05,
      "loss": 1.8825,
      "step": 195430
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.4072283506393433,
      "learning_rate": 3.346652713006571e-05,
      "loss": 1.8979,
      "step": 195431
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.464828610420227,
      "learning_rate": 3.3464649655238476e-05,
      "loss": 1.922,
      "step": 195432
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.3679125308990479,
      "learning_rate": 3.3462772229965126e-05,
      "loss": 1.9494,
      "step": 195433
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.569912075996399,
      "learning_rate": 3.346089485424618e-05,
      "loss": 2.0962,
      "step": 195434
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.6163623332977295,
      "learning_rate": 3.345901752808179e-05,
      "loss": 2.2311,
      "step": 195435
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.4072202444076538,
      "learning_rate": 3.3457140251472526e-05,
      "loss": 2.0133,
      "step": 195436
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.3531770706176758,
      "learning_rate": 3.3455263024418486e-05,
      "loss": 1.9583,
      "step": 195437
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.4266088008880615,
      "learning_rate": 3.345338584692021e-05,
      "loss": 2.1585,
      "step": 195438
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.4117416143417358,
      "learning_rate": 3.345150871897793e-05,
      "loss": 1.9474,
      "step": 195439
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.3055269718170166,
      "learning_rate": 3.34496316405921e-05,
      "loss": 2.021,
      "step": 195440
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.3949166536331177,
      "learning_rate": 3.3447754611762935e-05,
      "loss": 2.1029,
      "step": 195441
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.5645153522491455,
      "learning_rate": 3.3445877632490955e-05,
      "loss": 1.9393,
      "step": 195442
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.630940318107605,
      "learning_rate": 3.3444000702776406e-05,
      "loss": 2.1745,
      "step": 195443
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.303864598274231,
      "learning_rate": 3.344212382261965e-05,
      "loss": 2.1739,
      "step": 195444
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.4086092710494995,
      "learning_rate": 3.344024699202098e-05,
      "loss": 1.9332,
      "step": 195445
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.3720967769622803,
      "learning_rate": 3.3438370210980836e-05,
      "loss": 2.0857,
      "step": 195446
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.4930508136749268,
      "learning_rate": 3.343649347949948e-05,
      "loss": 1.806,
      "step": 195447
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.332634449005127,
      "learning_rate": 3.343461679757739e-05,
      "loss": 2.0989,
      "step": 195448
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.4972822666168213,
      "learning_rate": 3.343274016521481e-05,
      "loss": 2.3456,
      "step": 195449
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.395849347114563,
      "learning_rate": 3.343086358241216e-05,
      "loss": 1.85,
      "step": 195450
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.4680699110031128,
      "learning_rate": 3.342898704916963e-05,
      "loss": 1.9032,
      "step": 195451
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.4298197031021118,
      "learning_rate": 3.342711056548779e-05,
      "loss": 2.0868,
      "step": 195452
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.3734147548675537,
      "learning_rate": 3.342523413136677e-05,
      "loss": 1.8692,
      "step": 195453
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.4037063121795654,
      "learning_rate": 3.342335774680714e-05,
      "loss": 1.8182,
      "step": 195454
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.3148912191390991,
      "learning_rate": 3.3421481411809136e-05,
      "loss": 1.9544,
      "step": 195455
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.3016345500946045,
      "learning_rate": 3.341960512637308e-05,
      "loss": 2.0483,
      "step": 195456
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.3797563314437866,
      "learning_rate": 3.341772889049932e-05,
      "loss": 2.1261,
      "step": 195457
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.2766329050064087,
      "learning_rate": 3.3415852704188276e-05,
      "loss": 1.8385,
      "step": 195458
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.3841966390609741,
      "learning_rate": 3.3413976567440216e-05,
      "loss": 2.0415,
      "step": 195459
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.382897973060608,
      "learning_rate": 3.341210048025558e-05,
      "loss": 1.8108,
      "step": 195460
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.578269362449646,
      "learning_rate": 3.341022444263466e-05,
      "loss": 1.9467,
      "step": 195461
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.8705896139144897,
      "learning_rate": 3.3408348454577797e-05,
      "loss": 2.2007,
      "step": 195462
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.5293463468551636,
      "learning_rate": 3.340647251608531e-05,
      "loss": 2.123,
      "step": 195463
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.5342108011245728,
      "learning_rate": 3.3404596627157686e-05,
      "loss": 1.9409,
      "step": 195464
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.3939433097839355,
      "learning_rate": 3.340272078779508e-05,
      "loss": 1.8655,
      "step": 195465
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.613258957862854,
      "learning_rate": 3.3400844997998024e-05,
      "loss": 1.8676,
      "step": 195466
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.6019484996795654,
      "learning_rate": 3.339896925776675e-05,
      "loss": 1.7793,
      "step": 195467
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.4333171844482422,
      "learning_rate": 3.3397093567101663e-05,
      "loss": 1.8899,
      "step": 195468
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.4332832098007202,
      "learning_rate": 3.339521792600302e-05,
      "loss": 2.1187,
      "step": 195469
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.6271533966064453,
      "learning_rate": 3.339334233447131e-05,
      "loss": 1.8105,
      "step": 195470
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.3282289505004883,
      "learning_rate": 3.339146679250674e-05,
      "loss": 2.0673,
      "step": 195471
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.4068669080734253,
      "learning_rate": 3.338959130010979e-05,
      "loss": 1.8856,
      "step": 195472
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.3586938381195068,
      "learning_rate": 3.338771585728075e-05,
      "loss": 2.0883,
      "step": 195473
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.3208873271942139,
      "learning_rate": 3.3385840464019994e-05,
      "loss": 1.7597,
      "step": 195474
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.4968876838684082,
      "learning_rate": 3.338396512032776e-05,
      "loss": 2.1,
      "step": 195475
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.4119526147842407,
      "learning_rate": 3.3382089826204505e-05,
      "loss": 1.899,
      "step": 195476
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.4100168943405151,
      "learning_rate": 3.338021458165053e-05,
      "loss": 2.0451,
      "step": 195477
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.4464099407196045,
      "learning_rate": 3.337833938666628e-05,
      "loss": 2.2147,
      "step": 195478
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.5453523397445679,
      "learning_rate": 3.3376464241252e-05,
      "loss": 2.0243,
      "step": 195479
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.4373222589492798,
      "learning_rate": 3.337458914540807e-05,
      "loss": 2.0247,
      "step": 195480
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.4765024185180664,
      "learning_rate": 3.337271409913479e-05,
      "loss": 2.2011,
      "step": 195481
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.4428602457046509,
      "learning_rate": 3.3370839102432625e-05,
      "loss": 1.9486,
      "step": 195482
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.4696531295776367,
      "learning_rate": 3.336896415530177e-05,
      "loss": 1.9094,
      "step": 195483
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.5976685285568237,
      "learning_rate": 3.336708925774273e-05,
      "loss": 1.8884,
      "step": 195484
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.3941854238510132,
      "learning_rate": 3.336521440975577e-05,
      "loss": 1.9044,
      "step": 195485
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.6552425622940063,
      "learning_rate": 3.336333961134126e-05,
      "loss": 1.9951,
      "step": 195486
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.3782809972763062,
      "learning_rate": 3.336146486249949e-05,
      "loss": 2.0187,
      "step": 195487
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.455742597579956,
      "learning_rate": 3.33595901632309e-05,
      "loss": 1.891,
      "step": 195488
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.4405194520950317,
      "learning_rate": 3.335771551353572e-05,
      "loss": 1.8418,
      "step": 195489
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.35361909866333,
      "learning_rate": 3.335584091341446e-05,
      "loss": 1.963,
      "step": 195490
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.474394679069519,
      "learning_rate": 3.335396636286737e-05,
      "loss": 2.0328,
      "step": 195491
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.5038203001022339,
      "learning_rate": 3.335209186189476e-05,
      "loss": 2.0574,
      "step": 195492
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.374937653541565,
      "learning_rate": 3.3350217410497094e-05,
      "loss": 2.0964,
      "step": 195493
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.5295921564102173,
      "learning_rate": 3.334834300867464e-05,
      "loss": 1.9,
      "step": 195494
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.3978360891342163,
      "learning_rate": 3.3346468656427694e-05,
      "loss": 2.1902,
      "step": 195495
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.6186089515686035,
      "learning_rate": 3.334459435375676e-05,
      "loss": 2.2872,
      "step": 195496
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.2727755308151245,
      "learning_rate": 3.3342720100662103e-05,
      "loss": 1.927,
      "step": 195497
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.3685626983642578,
      "learning_rate": 3.3340845897143986e-05,
      "loss": 1.9598,
      "step": 195498
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.3994630575180054,
      "learning_rate": 3.333897174320288e-05,
      "loss": 1.8998,
      "step": 195499
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.3990322351455688,
      "learning_rate": 3.333709763883915e-05,
      "loss": 1.9005,
      "step": 195500
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.4201364517211914,
      "learning_rate": 3.3335223584053e-05,
      "loss": 1.9772,
      "step": 195501
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.3192479610443115,
      "learning_rate": 3.333334957884492e-05,
      "loss": 2.0481,
      "step": 195502
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.4363890886306763,
      "learning_rate": 3.333147562321517e-05,
      "loss": 2.0355,
      "step": 195503
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.6430692672729492,
      "learning_rate": 3.332960171716411e-05,
      "loss": 2.1232,
      "step": 195504
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.518680453300476,
      "learning_rate": 3.332772786069219e-05,
      "loss": 2.4087,
      "step": 195505
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.3324428796768188,
      "learning_rate": 3.33258540537996e-05,
      "loss": 1.8774,
      "step": 195506
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.4676017761230469,
      "learning_rate": 3.332398029648683e-05,
      "loss": 1.8306,
      "step": 195507
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.4312719106674194,
      "learning_rate": 3.3322106588754164e-05,
      "loss": 2.0378,
      "step": 195508
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.3890281915664673,
      "learning_rate": 3.332023293060186e-05,
      "loss": 1.9134,
      "step": 195509
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.3730542659759521,
      "learning_rate": 3.331835932203047e-05,
      "loss": 1.8039,
      "step": 195510
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.3576064109802246,
      "learning_rate": 3.331648576304019e-05,
      "loss": 2.195,
      "step": 195511
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.4163233041763306,
      "learning_rate": 3.331461225363138e-05,
      "loss": 1.9006,
      "step": 195512
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.3362220525741577,
      "learning_rate": 3.331273879380447e-05,
      "loss": 2.1311,
      "step": 195513
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.576237678527832,
      "learning_rate": 3.3310865383559704e-05,
      "loss": 1.9051,
      "step": 195514
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.4695206880569458,
      "learning_rate": 3.330899202289751e-05,
      "loss": 1.9702,
      "step": 195515
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.4437732696533203,
      "learning_rate": 3.3307118711818246e-05,
      "loss": 1.9109,
      "step": 195516
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.4180059432983398,
      "learning_rate": 3.330524545032219e-05,
      "loss": 1.9751,
      "step": 195517
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.409590482711792,
      "learning_rate": 3.3303372238409675e-05,
      "loss": 1.9035,
      "step": 195518
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.4364063739776611,
      "learning_rate": 3.3301499076081194e-05,
      "loss": 1.9964,
      "step": 195519
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.4847006797790527,
      "learning_rate": 3.3299625963336884e-05,
      "loss": 2.0458,
      "step": 195520
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.3470367193222046,
      "learning_rate": 3.329775290017731e-05,
      "loss": 2.1212,
      "step": 195521
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.3269013166427612,
      "learning_rate": 3.329587988660267e-05,
      "loss": 1.7979,
      "step": 195522
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.5269373655319214,
      "learning_rate": 3.3294006922613406e-05,
      "loss": 2.1966,
      "step": 195523
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.5174810886383057,
      "learning_rate": 3.3292134008209734e-05,
      "loss": 2.0752,
      "step": 195524
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.6203112602233887,
      "learning_rate": 3.329026114339217e-05,
      "loss": 2.0316,
      "step": 195525
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.4860345125198364,
      "learning_rate": 3.328838832816091e-05,
      "loss": 1.9942,
      "step": 195526
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.3858277797698975,
      "learning_rate": 3.328651556251641e-05,
      "loss": 1.9571,
      "step": 195527
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.496717929840088,
      "learning_rate": 3.328464284645902e-05,
      "loss": 2.0268,
      "step": 195528
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.4161386489868164,
      "learning_rate": 3.328277017998903e-05,
      "loss": 2.0187,
      "step": 195529
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.4543488025665283,
      "learning_rate": 3.3280897563106765e-05,
      "loss": 1.6912,
      "step": 195530
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.475900411605835,
      "learning_rate": 3.327902499581264e-05,
      "loss": 2.0645,
      "step": 195531
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.4270727634429932,
      "learning_rate": 3.327715247810695e-05,
      "loss": 1.9764,
      "step": 195532
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.3221412897109985,
      "learning_rate": 3.3275280009990125e-05,
      "loss": 1.8634,
      "step": 195533
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.4070740938186646,
      "learning_rate": 3.3273407591462465e-05,
      "loss": 2.1846,
      "step": 195534
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.4426649808883667,
      "learning_rate": 3.32715352225243e-05,
      "loss": 2.0389,
      "step": 195535
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.399040937423706,
      "learning_rate": 3.3269662903175945e-05,
      "loss": 2.1247,
      "step": 195536
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.506508231163025,
      "learning_rate": 3.3267790633417844e-05,
      "loss": 1.7318,
      "step": 195537
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.4246838092803955,
      "learning_rate": 3.326591841325025e-05,
      "loss": 1.8948,
      "step": 195538
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.386243462562561,
      "learning_rate": 3.326404624267358e-05,
      "loss": 2.0465,
      "step": 195539
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.5329604148864746,
      "learning_rate": 3.326217412168821e-05,
      "loss": 2.2499,
      "step": 195540
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.5064419507980347,
      "learning_rate": 3.326030205029441e-05,
      "loss": 2.0979,
      "step": 195541
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.3926326036453247,
      "learning_rate": 3.32584300284925e-05,
      "loss": 2.1393,
      "step": 195542
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.4615004062652588,
      "learning_rate": 3.325655805628292e-05,
      "loss": 1.9693,
      "step": 195543
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.3916727304458618,
      "learning_rate": 3.325468613366594e-05,
      "loss": 2.0362,
      "step": 195544
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.5740714073181152,
      "learning_rate": 3.3252814260642026e-05,
      "loss": 2.1862,
      "step": 195545
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.429441213607788,
      "learning_rate": 3.3250942437211436e-05,
      "loss": 2.037,
      "step": 195546
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.7044167518615723,
      "learning_rate": 3.324907066337451e-05,
      "loss": 1.9644,
      "step": 195547
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.549294114112854,
      "learning_rate": 3.324719893913158e-05,
      "loss": 2.1002,
      "step": 195548
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.7865667343139648,
      "learning_rate": 3.324532726448308e-05,
      "loss": 2.088,
      "step": 195549
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.476283073425293,
      "learning_rate": 3.324345563942927e-05,
      "loss": 1.7098,
      "step": 195550
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.6407828330993652,
      "learning_rate": 3.324158406397059e-05,
      "loss": 2.1334,
      "step": 195551
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.4336190223693848,
      "learning_rate": 3.323971253810731e-05,
      "loss": 2.1084,
      "step": 195552
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.4884400367736816,
      "learning_rate": 3.323784106183982e-05,
      "loss": 1.9251,
      "step": 195553
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.4578866958618164,
      "learning_rate": 3.323596963516839e-05,
      "loss": 1.8671,
      "step": 195554
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.33790922164917,
      "learning_rate": 3.323409825809349e-05,
      "loss": 1.8432,
      "step": 195555
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.3311727046966553,
      "learning_rate": 3.323222693061535e-05,
      "loss": 2.2661,
      "step": 195556
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.38663911819458,
      "learning_rate": 3.323035565273443e-05,
      "loss": 2.1712,
      "step": 195557
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.4249162673950195,
      "learning_rate": 3.322848442445104e-05,
      "loss": 1.9909,
      "step": 195558
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.3826614618301392,
      "learning_rate": 3.322661324576548e-05,
      "loss": 1.9554,
      "step": 195559
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.4779996871948242,
      "learning_rate": 3.3224742116678106e-05,
      "loss": 2.061,
      "step": 195560
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.4771673679351807,
      "learning_rate": 3.322287103718929e-05,
      "loss": 2.0503,
      "step": 195561
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.462106466293335,
      "learning_rate": 3.322100000729937e-05,
      "loss": 2.0241,
      "step": 195562
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.3639756441116333,
      "learning_rate": 3.321912902700874e-05,
      "loss": 1.911,
      "step": 195563
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.2972716093063354,
      "learning_rate": 3.321725809631773e-05,
      "loss": 1.8931,
      "step": 195564
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.4009279012680054,
      "learning_rate": 3.321538721522665e-05,
      "loss": 1.9208,
      "step": 195565
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.4172499179840088,
      "learning_rate": 3.3213516383735795e-05,
      "loss": 2.0126,
      "step": 195566
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.4200626611709595,
      "learning_rate": 3.321164560184566e-05,
      "loss": 2.2569,
      "step": 195567
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.4127891063690186,
      "learning_rate": 3.320977486955645e-05,
      "loss": 2.1076,
      "step": 195568
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.2716470956802368,
      "learning_rate": 3.320790418686863e-05,
      "loss": 1.9988,
      "step": 195569
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.326354742050171,
      "learning_rate": 3.320603355378253e-05,
      "loss": 1.778,
      "step": 195570
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.452754020690918,
      "learning_rate": 3.320416297029842e-05,
      "loss": 1.8742,
      "step": 195571
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.2236902713775635,
      "learning_rate": 3.320229243641667e-05,
      "loss": 1.9376,
      "step": 195572
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.4496694803237915,
      "learning_rate": 3.32004219521377e-05,
      "loss": 1.8271,
      "step": 195573
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.4811639785766602,
      "learning_rate": 3.319855151746172e-05,
      "loss": 2.0073,
      "step": 195574
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.4168977737426758,
      "learning_rate": 3.3196681132389266e-05,
      "loss": 1.9212,
      "step": 195575
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.4073460102081299,
      "learning_rate": 3.31948107969205e-05,
      "loss": 1.838,
      "step": 195576
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.3866039514541626,
      "learning_rate": 3.319294051105592e-05,
      "loss": 1.8702,
      "step": 195577
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.3686025142669678,
      "learning_rate": 3.319107027479583e-05,
      "loss": 2.074,
      "step": 195578
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.4520152807235718,
      "learning_rate": 3.3189200088140525e-05,
      "loss": 2.0308,
      "step": 195579
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.4144372940063477,
      "learning_rate": 3.3187329951090334e-05,
      "loss": 1.9487,
      "step": 195580
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.4014453887939453,
      "learning_rate": 3.3185459863645734e-05,
      "loss": 2.0698,
      "step": 195581
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.4096647500991821,
      "learning_rate": 3.3183589825806886e-05,
      "loss": 1.9413,
      "step": 195582
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.536033272743225,
      "learning_rate": 3.318171983757436e-05,
      "loss": 2.2252,
      "step": 195583
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.3758984804153442,
      "learning_rate": 3.317984989894835e-05,
      "loss": 2.1882,
      "step": 195584
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.455795168876648,
      "learning_rate": 3.3177980009929264e-05,
      "loss": 2.1499,
      "step": 195585
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.3917953968048096,
      "learning_rate": 3.317611017051736e-05,
      "loss": 2.1538,
      "step": 195586
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.4298651218414307,
      "learning_rate": 3.3174240380713135e-05,
      "loss": 1.9941,
      "step": 195587
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.4923194646835327,
      "learning_rate": 3.3172370640516766e-05,
      "loss": 2.124,
      "step": 195588
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.4410110712051392,
      "learning_rate": 3.3170500949928745e-05,
      "loss": 1.9162,
      "step": 195589
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.3926806449890137,
      "learning_rate": 3.3168631308949377e-05,
      "loss": 2.194,
      "step": 195590
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.5167120695114136,
      "learning_rate": 3.316676171757896e-05,
      "loss": 1.9243,
      "step": 195591
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.415783405303955,
      "learning_rate": 3.316489217581793e-05,
      "loss": 2.0856,
      "step": 195592
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.4395751953125,
      "learning_rate": 3.316302268366654e-05,
      "loss": 2.0629,
      "step": 195593
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.4997676610946655,
      "learning_rate": 3.316115324112517e-05,
      "loss": 2.2078,
      "step": 195594
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.4740965366363525,
      "learning_rate": 3.3159283848194216e-05,
      "loss": 1.9205,
      "step": 195595
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.3982994556427002,
      "learning_rate": 3.315741450487398e-05,
      "loss": 1.9109,
      "step": 195596
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.4215599298477173,
      "learning_rate": 3.315554521116479e-05,
      "loss": 1.8576,
      "step": 195597
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.359854817390442,
      "learning_rate": 3.315367596706705e-05,
      "loss": 2.0328,
      "step": 195598
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.4026532173156738,
      "learning_rate": 3.315180677258109e-05,
      "loss": 1.8844,
      "step": 195599
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.3332165479660034,
      "learning_rate": 3.314993762770721e-05,
      "loss": 1.8817,
      "step": 195600
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.4362413883209229,
      "learning_rate": 3.314806853244581e-05,
      "loss": 1.9714,
      "step": 195601
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.2670197486877441,
      "learning_rate": 3.3146199486797263e-05,
      "loss": 1.9771,
      "step": 195602
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.3770300149917603,
      "learning_rate": 3.314433049076179e-05,
      "loss": 1.9357,
      "step": 195603
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.306784987449646,
      "learning_rate": 3.314246154433987e-05,
      "loss": 2.1019,
      "step": 195604
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.4385684728622437,
      "learning_rate": 3.314059264753176e-05,
      "loss": 1.7476,
      "step": 195605
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.3184049129486084,
      "learning_rate": 3.313872380033793e-05,
      "loss": 1.895,
      "step": 195606
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.4350110292434692,
      "learning_rate": 3.3136855002758645e-05,
      "loss": 2.0437,
      "step": 195607
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.7416223287582397,
      "learning_rate": 3.3134986254794235e-05,
      "loss": 1.9879,
      "step": 195608
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.4018186330795288,
      "learning_rate": 3.313311755644501e-05,
      "loss": 1.9301,
      "step": 195609
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.2796247005462646,
      "learning_rate": 3.313124890771143e-05,
      "loss": 2.1645,
      "step": 195610
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.3544074296951294,
      "learning_rate": 3.3129380308593754e-05,
      "loss": 1.8824,
      "step": 195611
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.4177587032318115,
      "learning_rate": 3.3127511759092395e-05,
      "loss": 2.0607,
      "step": 195612
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.3895372152328491,
      "learning_rate": 3.312564325920771e-05,
      "loss": 2.0833,
      "step": 195613
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.3985707759857178,
      "learning_rate": 3.312377480893998e-05,
      "loss": 1.7601,
      "step": 195614
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.4340856075286865,
      "learning_rate": 3.312190640828951e-05,
      "loss": 1.9982,
      "step": 195615
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.4013832807540894,
      "learning_rate": 3.312003805725679e-05,
      "loss": 1.9813,
      "step": 195616
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.8853123188018799,
      "learning_rate": 3.3118169755842016e-05,
      "loss": 1.9901,
      "step": 195617
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.453660488128662,
      "learning_rate": 3.311630150404568e-05,
      "loss": 2.0078,
      "step": 195618
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.364694595336914,
      "learning_rate": 3.311443330186806e-05,
      "loss": 1.947,
      "step": 195619
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.2645078897476196,
      "learning_rate": 3.311256514930951e-05,
      "loss": 1.7937,
      "step": 195620
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.434197187423706,
      "learning_rate": 3.3110697046370303e-05,
      "loss": 2.0167,
      "step": 195621
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.4294078350067139,
      "learning_rate": 3.310882899305094e-05,
      "loss": 1.9202,
      "step": 195622
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.4393216371536255,
      "learning_rate": 3.310696098935158e-05,
      "loss": 2.1183,
      "step": 195623
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.3027760982513428,
      "learning_rate": 3.3105093035272766e-05,
      "loss": 1.9032,
      "step": 195624
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.3428514003753662,
      "learning_rate": 3.310322513081476e-05,
      "loss": 2.0569,
      "step": 195625
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.4712176322937012,
      "learning_rate": 3.310135727597789e-05,
      "loss": 2.0188,
      "step": 195626
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.4140433073043823,
      "learning_rate": 3.309948947076246e-05,
      "loss": 1.9513,
      "step": 195627
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.4110032320022583,
      "learning_rate": 3.309762171516891e-05,
      "loss": 1.996,
      "step": 195628
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.5653856992721558,
      "learning_rate": 3.309575400919753e-05,
      "loss": 2.1037,
      "step": 195629
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.4053395986557007,
      "learning_rate": 3.309388635284872e-05,
      "loss": 1.9288,
      "step": 195630
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.5962071418762207,
      "learning_rate": 3.3092018746122785e-05,
      "loss": 2.0511,
      "step": 195631
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.4870749711990356,
      "learning_rate": 3.309015118902012e-05,
      "loss": 2.0229,
      "step": 195632
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.5106943845748901,
      "learning_rate": 3.308828368154093e-05,
      "loss": 1.9314,
      "step": 195633
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.379926323890686,
      "learning_rate": 3.3086416223685775e-05,
      "loss": 2.0548,
      "step": 195634
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.5471298694610596,
      "learning_rate": 3.308454881545479e-05,
      "loss": 2.1204,
      "step": 195635
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.3727673292160034,
      "learning_rate": 3.308268145684852e-05,
      "loss": 2.1447,
      "step": 195636
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.392257809638977,
      "learning_rate": 3.3080814147867174e-05,
      "loss": 1.8764,
      "step": 195637
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.319475769996643,
      "learning_rate": 3.3078946888511173e-05,
      "loss": 2.0753,
      "step": 195638
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.4660073518753052,
      "learning_rate": 3.3077079678780774e-05,
      "loss": 2.0095,
      "step": 195639
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.4278690814971924,
      "learning_rate": 3.307521251867644e-05,
      "loss": 1.9735,
      "step": 195640
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.4988210201263428,
      "learning_rate": 3.3073345408198415e-05,
      "loss": 2.0533,
      "step": 195641
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.4041047096252441,
      "learning_rate": 3.307147834734716e-05,
      "loss": 1.9522,
      "step": 195642
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.5121885538101196,
      "learning_rate": 3.306961133612287e-05,
      "loss": 2.1402,
      "step": 195643
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.3616935014724731,
      "learning_rate": 3.3067744374526116e-05,
      "loss": 1.8656,
      "step": 195644
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.3479654788970947,
      "learning_rate": 3.306587746255697e-05,
      "loss": 2.235,
      "step": 195645
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.5431314706802368,
      "learning_rate": 3.3064010600215984e-05,
      "loss": 2.1771,
      "step": 195646
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.5702760219573975,
      "learning_rate": 3.306214378750337e-05,
      "loss": 1.809,
      "step": 195647
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.4061400890350342,
      "learning_rate": 3.3060277024419624e-05,
      "loss": 1.9905,
      "step": 195648
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.3146867752075195,
      "learning_rate": 3.305841031096491e-05,
      "loss": 1.9288,
      "step": 195649
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.4442927837371826,
      "learning_rate": 3.3056543647139835e-05,
      "loss": 1.9796,
      "step": 195650
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.3826656341552734,
      "learning_rate": 3.3054677032944456e-05,
      "loss": 1.8398,
      "step": 195651
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.465133786201477,
      "learning_rate": 3.3052810468379306e-05,
      "loss": 1.9186,
      "step": 195652
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.5015608072280884,
      "learning_rate": 3.305094395344463e-05,
      "loss": 1.9801,
      "step": 195653
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.3689481019973755,
      "learning_rate": 3.3049077488140885e-05,
      "loss": 2.0499,
      "step": 195654
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.388251543045044,
      "learning_rate": 3.304721107246827e-05,
      "loss": 2.0008,
      "step": 195655
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.396520733833313,
      "learning_rate": 3.3045344706427356e-05,
      "loss": 2.0433,
      "step": 195656
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.5331685543060303,
      "learning_rate": 3.304347839001824e-05,
      "loss": 1.9779,
      "step": 195657
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.4442845582962036,
      "learning_rate": 3.304161212324142e-05,
      "loss": 1.978,
      "step": 195658
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.444974422454834,
      "learning_rate": 3.303974590609717e-05,
      "loss": 2.3224,
      "step": 195659
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.4907472133636475,
      "learning_rate": 3.3037879738585916e-05,
      "loss": 1.9322,
      "step": 195660
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.6700119972229004,
      "learning_rate": 3.303601362070789e-05,
      "loss": 2.0299,
      "step": 195661
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.3804866075515747,
      "learning_rate": 3.303414755246363e-05,
      "loss": 2.1228,
      "step": 195662
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.4182437658309937,
      "learning_rate": 3.303228153385327e-05,
      "loss": 2.0435,
      "step": 195663
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.4273184537887573,
      "learning_rate": 3.303041556487727e-05,
      "loss": 1.9949,
      "step": 195664
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.4472216367721558,
      "learning_rate": 3.3028549645535896e-05,
      "loss": 2.0339,
      "step": 195665
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.6529539823532104,
      "learning_rate": 3.302668377582966e-05,
      "loss": 2.1267,
      "step": 195666
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.628699541091919,
      "learning_rate": 3.3024817955758675e-05,
      "loss": 2.0535,
      "step": 195667
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.5101542472839355,
      "learning_rate": 3.3022952185323525e-05,
      "loss": 2.2444,
      "step": 195668
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.4579535722732544,
      "learning_rate": 3.302108646452444e-05,
      "loss": 1.9751,
      "step": 195669
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.372420072555542,
      "learning_rate": 3.301922079336178e-05,
      "loss": 1.9085,
      "step": 195670
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.438293695449829,
      "learning_rate": 3.3017355171835816e-05,
      "loss": 2.0068,
      "step": 195671
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.4609524011611938,
      "learning_rate": 3.301548959994701e-05,
      "loss": 2.1673,
      "step": 195672
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.4270302057266235,
      "learning_rate": 3.301362407769563e-05,
      "loss": 2.1283,
      "step": 195673
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.5783871412277222,
      "learning_rate": 3.301175860508212e-05,
      "loss": 2.2074,
      "step": 195674
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.375571608543396,
      "learning_rate": 3.300989318210673e-05,
      "loss": 1.9477,
      "step": 195675
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.4525811672210693,
      "learning_rate": 3.30080278087698e-05,
      "loss": 2.0866,
      "step": 195676
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.5100243091583252,
      "learning_rate": 3.30061624850718e-05,
      "loss": 1.8094,
      "step": 195677
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.3322029113769531,
      "learning_rate": 3.300429721101295e-05,
      "loss": 1.8856,
      "step": 195678
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.4674732685089111,
      "learning_rate": 3.3002431986593605e-05,
      "loss": 2.0614,
      "step": 195679
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.3393168449401855,
      "learning_rate": 3.300056681181421e-05,
      "loss": 2.0872,
      "step": 195680
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.422169804573059,
      "learning_rate": 3.2998701686675044e-05,
      "loss": 2.1445,
      "step": 195681
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.2896679639816284,
      "learning_rate": 3.29968366111764e-05,
      "loss": 1.85,
      "step": 195682
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.4344173669815063,
      "learning_rate": 3.299497158531875e-05,
      "loss": 1.9831,
      "step": 195683
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.4079678058624268,
      "learning_rate": 3.299310660910239e-05,
      "loss": 2.101,
      "step": 195684
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.299155354499817,
      "learning_rate": 3.299124168252756e-05,
      "loss": 1.9526,
      "step": 195685
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.5304348468780518,
      "learning_rate": 3.298937680559478e-05,
      "loss": 2.1107,
      "step": 195686
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.472078561782837,
      "learning_rate": 3.2987511978304296e-05,
      "loss": 2.0148,
      "step": 195687
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.4189454317092896,
      "learning_rate": 3.298564720065643e-05,
      "loss": 1.9194,
      "step": 195688
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.475786566734314,
      "learning_rate": 3.298378247265162e-05,
      "loss": 1.8934,
      "step": 195689
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.3117778301239014,
      "learning_rate": 3.2981917794290133e-05,
      "loss": 2.1856,
      "step": 195690
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.440729022026062,
      "learning_rate": 3.2980053165572374e-05,
      "loss": 2.1672,
      "step": 195691
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.4776690006256104,
      "learning_rate": 3.2978188586498666e-05,
      "loss": 1.969,
      "step": 195692
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.4785351753234863,
      "learning_rate": 3.2976324057069384e-05,
      "loss": 1.9728,
      "step": 195693
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.348049283027649,
      "learning_rate": 3.297445957728475e-05,
      "loss": 1.8911,
      "step": 195694
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.4187389612197876,
      "learning_rate": 3.297259514714527e-05,
      "loss": 2.0767,
      "step": 195695
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.4458414316177368,
      "learning_rate": 3.297073076665119e-05,
      "loss": 2.1839,
      "step": 195696
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.3563792705535889,
      "learning_rate": 3.296886643580292e-05,
      "loss": 2.0053,
      "step": 195697
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.403874158859253,
      "learning_rate": 3.296700215460081e-05,
      "loss": 1.9515,
      "step": 195698
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.4063465595245361,
      "learning_rate": 3.296513792304515e-05,
      "loss": 1.7916,
      "step": 195699
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.6200534105300903,
      "learning_rate": 3.2963273741136274e-05,
      "loss": 1.9747,
      "step": 195700
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.4188497066497803,
      "learning_rate": 3.296140960887459e-05,
      "loss": 2.1477,
      "step": 195701
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.323269009590149,
      "learning_rate": 3.295954552626039e-05,
      "loss": 1.9544,
      "step": 195702
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.4639675617218018,
      "learning_rate": 3.2957681493294105e-05,
      "loss": 1.8308,
      "step": 195703
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.4789999723434448,
      "learning_rate": 3.2955817509976014e-05,
      "loss": 2.0058,
      "step": 195704
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.4650081396102905,
      "learning_rate": 3.29539535763065e-05,
      "loss": 1.8734,
      "step": 195705
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.6541098356246948,
      "learning_rate": 3.295208969228581e-05,
      "loss": 2.0036,
      "step": 195706
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.4821178913116455,
      "learning_rate": 3.295022585791444e-05,
      "loss": 1.97,
      "step": 195707
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.795041799545288,
      "learning_rate": 3.294836207319261e-05,
      "loss": 2.0919,
      "step": 195708
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.522260308265686,
      "learning_rate": 3.2946498338120775e-05,
      "loss": 1.7553,
      "step": 195709
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.582241415977478,
      "learning_rate": 3.294463465269916e-05,
      "loss": 2.1428,
      "step": 195710
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.447265386581421,
      "learning_rate": 3.2942771016928285e-05,
      "loss": 2.0445,
      "step": 195711
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.5552562475204468,
      "learning_rate": 3.2940907430808296e-05,
      "loss": 1.9914,
      "step": 195712
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.4341076612472534,
      "learning_rate": 3.293904389433966e-05,
      "loss": 1.9901,
      "step": 195713
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.4110398292541504,
      "learning_rate": 3.293718040752267e-05,
      "loss": 1.9328,
      "step": 195714
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.8028556108474731,
      "learning_rate": 3.293531697035776e-05,
      "loss": 2.1374,
      "step": 195715
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.5217258930206299,
      "learning_rate": 3.293345358284514e-05,
      "loss": 2.1942,
      "step": 195716
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.494489073753357,
      "learning_rate": 3.293159024498533e-05,
      "loss": 1.962,
      "step": 195717
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.626007080078125,
      "learning_rate": 3.29297269567785e-05,
      "loss": 2.0877,
      "step": 195718
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.4457380771636963,
      "learning_rate": 3.292786371822512e-05,
      "loss": 2.0776,
      "step": 195719
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.6095666885375977,
      "learning_rate": 3.292600052932543e-05,
      "loss": 1.9949,
      "step": 195720
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.7791790962219238,
      "learning_rate": 3.292413739007991e-05,
      "loss": 2.1328,
      "step": 195721
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.462729573249817,
      "learning_rate": 3.2922274300488737e-05,
      "loss": 2.2199,
      "step": 195722
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.3814114332199097,
      "learning_rate": 3.292041126055251e-05,
      "loss": 1.9064,
      "step": 195723
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.5332528352737427,
      "learning_rate": 3.2918548270271305e-05,
      "loss": 2.0332,
      "step": 195724
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.3227519989013672,
      "learning_rate": 3.291668532964561e-05,
      "loss": 1.8941,
      "step": 195725
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.592814564704895,
      "learning_rate": 3.2914822438675726e-05,
      "loss": 1.9403,
      "step": 195726
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.4230297803878784,
      "learning_rate": 3.2912959597362054e-05,
      "loss": 2.1464,
      "step": 195727
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.6580607891082764,
      "learning_rate": 3.291109680570483e-05,
      "loss": 1.9252,
      "step": 195728
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.3729249238967896,
      "learning_rate": 3.2909234063704616e-05,
      "loss": 2.0551,
      "step": 195729
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.492000699043274,
      "learning_rate": 3.2907371371361514e-05,
      "loss": 2.0194,
      "step": 195730
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.3736814260482788,
      "learning_rate": 3.290550872867599e-05,
      "loss": 2.0225,
      "step": 195731
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.485993504524231,
      "learning_rate": 3.290364613564834e-05,
      "loss": 1.8921,
      "step": 195732
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.327343463897705,
      "learning_rate": 3.290178359227901e-05,
      "loss": 1.9433,
      "step": 195733
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.2996647357940674,
      "learning_rate": 3.289992109856822e-05,
      "loss": 2.0846,
      "step": 195734
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.29041588306427,
      "learning_rate": 3.289805865451647e-05,
      "loss": 1.863,
      "step": 195735
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.6643749475479126,
      "learning_rate": 3.28961962601239e-05,
      "loss": 2.1371,
      "step": 195736
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.343153476715088,
      "learning_rate": 3.289433391539104e-05,
      "loss": 2.0786,
      "step": 195737
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.3888576030731201,
      "learning_rate": 3.289247162031812e-05,
      "loss": 2.0783,
      "step": 195738
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.3913264274597168,
      "learning_rate": 3.2890609374905576e-05,
      "loss": 1.7719,
      "step": 195739
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.4276142120361328,
      "learning_rate": 3.288874717915364e-05,
      "loss": 2.0805,
      "step": 195740
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.5443189144134521,
      "learning_rate": 3.288688503306285e-05,
      "loss": 2.1458,
      "step": 195741
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.305849313735962,
      "learning_rate": 3.2885022936633334e-05,
      "loss": 1.6742,
      "step": 195742
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.5578348636627197,
      "learning_rate": 3.2883160889865555e-05,
      "loss": 1.914,
      "step": 195743
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.455296516418457,
      "learning_rate": 3.2881298892759825e-05,
      "loss": 1.8107,
      "step": 195744
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.5772230625152588,
      "learning_rate": 3.2879436945316526e-05,
      "loss": 1.9857,
      "step": 195745
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.4771473407745361,
      "learning_rate": 3.287757504753594e-05,
      "loss": 2.1683,
      "step": 195746
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.5026116371154785,
      "learning_rate": 3.287571319941856e-05,
      "loss": 1.8866,
      "step": 195747
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.3925373554229736,
      "learning_rate": 3.287385140096452e-05,
      "loss": 1.9743,
      "step": 195748
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.3362774848937988,
      "learning_rate": 3.2871989652174356e-05,
      "loss": 1.8213,
      "step": 195749
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.4023348093032837,
      "learning_rate": 3.287012795304823e-05,
      "loss": 2.2922,
      "step": 195750
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.4880908727645874,
      "learning_rate": 3.2868266303586677e-05,
      "loss": 1.7037,
      "step": 195751
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.4919706583023071,
      "learning_rate": 3.2866404703789896e-05,
      "loss": 1.8392,
      "step": 195752
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.496575951576233,
      "learning_rate": 3.286454315365832e-05,
      "loss": 1.9161,
      "step": 195753
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.6793477535247803,
      "learning_rate": 3.286268165319231e-05,
      "loss": 2.0289,
      "step": 195754
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.3219407796859741,
      "learning_rate": 3.286082020239214e-05,
      "loss": 2.1949,
      "step": 195755
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.4235990047454834,
      "learning_rate": 3.285895880125815e-05,
      "loss": 1.9812,
      "step": 195756
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.486712098121643,
      "learning_rate": 3.285709744979076e-05,
      "loss": 2.002,
      "step": 195757
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.549275517463684,
      "learning_rate": 3.285523614799024e-05,
      "loss": 1.9311,
      "step": 195758
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.9702659845352173,
      "learning_rate": 3.285337489585702e-05,
      "loss": 2.1736,
      "step": 195759
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.53257417678833,
      "learning_rate": 3.285151369339141e-05,
      "loss": 1.873,
      "step": 195760
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.429558277130127,
      "learning_rate": 3.284965254059373e-05,
      "loss": 2.0971,
      "step": 195761
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.3786487579345703,
      "learning_rate": 3.284779143746429e-05,
      "loss": 2.3117,
      "step": 195762
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.4289884567260742,
      "learning_rate": 3.284593038400355e-05,
      "loss": 2.0681,
      "step": 195763
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.8516602516174316,
      "learning_rate": 3.2844069380211746e-05,
      "loss": 1.9482,
      "step": 195764
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.5533140897750854,
      "learning_rate": 3.284220842608931e-05,
      "loss": 2.0488,
      "step": 195765
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.6445294618606567,
      "learning_rate": 3.2840347521636546e-05,
      "loss": 2.2102,
      "step": 195766
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.567958116531372,
      "learning_rate": 3.283848666685378e-05,
      "loss": 2.0504,
      "step": 195767
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.4477709531784058,
      "learning_rate": 3.2836625861741415e-05,
      "loss": 1.9935,
      "step": 195768
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.3367325067520142,
      "learning_rate": 3.283476510629975e-05,
      "loss": 1.9223,
      "step": 195769
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.4072980880737305,
      "learning_rate": 3.2832904400529124e-05,
      "loss": 2.0677,
      "step": 195770
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.3675919771194458,
      "learning_rate": 3.2831043744429965e-05,
      "loss": 2.0444,
      "step": 195771
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.3769757747650146,
      "learning_rate": 3.282918313800253e-05,
      "loss": 1.8914,
      "step": 195772
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.3442779779434204,
      "learning_rate": 3.282732258124714e-05,
      "loss": 2.1422,
      "step": 195773
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.5951969623565674,
      "learning_rate": 3.282546207416424e-05,
      "loss": 2.0019,
      "step": 195774
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.2805490493774414,
      "learning_rate": 3.2823601616754116e-05,
      "loss": 1.9116,
      "step": 195775
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.3007266521453857,
      "learning_rate": 3.282174120901715e-05,
      "loss": 2.1241,
      "step": 195776
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.2251596450805664,
      "learning_rate": 3.2819880850953686e-05,
      "loss": 1.8649,
      "step": 195777
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.347451090812683,
      "learning_rate": 3.281802054256402e-05,
      "loss": 2.0251,
      "step": 195778
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.3843560218811035,
      "learning_rate": 3.281616028384848e-05,
      "loss": 1.8542,
      "step": 195779
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.4415221214294434,
      "learning_rate": 3.281430007480754e-05,
      "loss": 1.9651,
      "step": 195780
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.393608808517456,
      "learning_rate": 3.2812439915441366e-05,
      "loss": 2.0421,
      "step": 195781
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.3452662229537964,
      "learning_rate": 3.281057980575049e-05,
      "loss": 1.9084,
      "step": 195782
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.2664544582366943,
      "learning_rate": 3.280871974573518e-05,
      "loss": 1.9188,
      "step": 195783
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.3280009031295776,
      "learning_rate": 3.280685973539576e-05,
      "loss": 1.7954,
      "step": 195784
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.5495997667312622,
      "learning_rate": 3.2804999774732545e-05,
      "loss": 2.0231,
      "step": 195785
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.481831669807434,
      "learning_rate": 3.2803139863745955e-05,
      "loss": 1.9608,
      "step": 195786
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.548850655555725,
      "learning_rate": 3.2801280002436256e-05,
      "loss": 2.0002,
      "step": 195787
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.4789403676986694,
      "learning_rate": 3.279942019080393e-05,
      "loss": 1.9477,
      "step": 195788
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.3850171566009521,
      "learning_rate": 3.2797560428849125e-05,
      "loss": 1.8941,
      "step": 195789
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.5412888526916504,
      "learning_rate": 3.279570071657245e-05,
      "loss": 2.1253,
      "step": 195790
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.332895278930664,
      "learning_rate": 3.2793841053973966e-05,
      "loss": 1.9731,
      "step": 195791
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.6224042177200317,
      "learning_rate": 3.279198144105422e-05,
      "loss": 2.1502,
      "step": 195792
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.3625566959381104,
      "learning_rate": 3.279012187781343e-05,
      "loss": 1.9273,
      "step": 195793
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.304298758506775,
      "learning_rate": 3.278826236425206e-05,
      "loss": 2.121,
      "step": 195794
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.3539410829544067,
      "learning_rate": 3.278640290037032e-05,
      "loss": 1.9131,
      "step": 195795
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.468201994895935,
      "learning_rate": 3.278454348616878e-05,
      "loss": 1.9969,
      "step": 195796
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.439768671989441,
      "learning_rate": 3.2782684121647494e-05,
      "loss": 2.2623,
      "step": 195797
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.471581220626831,
      "learning_rate": 3.2780824806807e-05,
      "loss": 1.8724,
      "step": 195798
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.4513088464736938,
      "learning_rate": 3.2778965541647575e-05,
      "loss": 1.9641,
      "step": 195799
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.4174379110336304,
      "learning_rate": 3.27771063261696e-05,
      "loss": 2.1148,
      "step": 195800
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.43947172164917,
      "learning_rate": 3.277524716037339e-05,
      "loss": 2.1093,
      "step": 195801
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.4602452516555786,
      "learning_rate": 3.27733880442594e-05,
      "loss": 1.8013,
      "step": 195802
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.3262203931808472,
      "learning_rate": 3.277152897782776e-05,
      "loss": 1.9424,
      "step": 195803
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.3531620502471924,
      "learning_rate": 3.2769669961078996e-05,
      "loss": 2.0509,
      "step": 195804
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.3674139976501465,
      "learning_rate": 3.2767810994013346e-05,
      "loss": 1.8614,
      "step": 195805
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.3343786001205444,
      "learning_rate": 3.2765952076631253e-05,
      "loss": 1.9911,
      "step": 195806
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.3200088739395142,
      "learning_rate": 3.276409320893295e-05,
      "loss": 2.0644,
      "step": 195807
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.455324649810791,
      "learning_rate": 3.2762234390918976e-05,
      "loss": 2.161,
      "step": 195808
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.5678000450134277,
      "learning_rate": 3.276037562258942e-05,
      "loss": 1.8686,
      "step": 195809
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.5024068355560303,
      "learning_rate": 3.275851690394482e-05,
      "loss": 2.1248,
      "step": 195810
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.521834135055542,
      "learning_rate": 3.2756658234985414e-05,
      "loss": 2.1133,
      "step": 195811
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.3345340490341187,
      "learning_rate": 3.275479961571162e-05,
      "loss": 2.1537,
      "step": 195812
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.4285762310028076,
      "learning_rate": 3.2752941046123724e-05,
      "loss": 1.9917,
      "step": 195813
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.294085144996643,
      "learning_rate": 3.275108252622218e-05,
      "loss": 2.0797,
      "step": 195814
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.6183520555496216,
      "learning_rate": 3.274922405600716e-05,
      "loss": 2.2145,
      "step": 195815
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.3097872734069824,
      "learning_rate": 3.274736563547916e-05,
      "loss": 1.9188,
      "step": 195816
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.3651783466339111,
      "learning_rate": 3.2745507264638415e-05,
      "loss": 1.8606,
      "step": 195817
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.4418596029281616,
      "learning_rate": 3.274364894348539e-05,
      "loss": 1.9545,
      "step": 195818
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.472341775894165,
      "learning_rate": 3.274179067202032e-05,
      "loss": 1.9586,
      "step": 195819
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.5002027750015259,
      "learning_rate": 3.2739932450243665e-05,
      "loss": 2.0576,
      "step": 195820
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.427927851676941,
      "learning_rate": 3.273807427815564e-05,
      "loss": 2.0018,
      "step": 195821
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.534906029701233,
      "learning_rate": 3.273621615575666e-05,
      "loss": 2.0448,
      "step": 195822
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.5258351564407349,
      "learning_rate": 3.273435808304703e-05,
      "loss": 1.7895,
      "step": 195823
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.4408706426620483,
      "learning_rate": 3.273250006002719e-05,
      "loss": 2.1791,
      "step": 195824
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.3277173042297363,
      "learning_rate": 3.2730642086697335e-05,
      "loss": 1.6862,
      "step": 195825
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.4289652109146118,
      "learning_rate": 3.2728784163058064e-05,
      "loss": 1.9183,
      "step": 195826
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.616500735282898,
      "learning_rate": 3.272692628910941e-05,
      "loss": 1.9286,
      "step": 195827
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.451906442642212,
      "learning_rate": 3.2725068464851914e-05,
      "loss": 2.0809,
      "step": 195828
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.4273685216903687,
      "learning_rate": 3.272321069028583e-05,
      "loss": 2.1239,
      "step": 195829
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.3910692930221558,
      "learning_rate": 3.27213529654116e-05,
      "loss": 1.9153,
      "step": 195830
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.4893431663513184,
      "learning_rate": 3.2719495290229446e-05,
      "loss": 1.931,
      "step": 195831
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.4893962144851685,
      "learning_rate": 3.2717637664739945e-05,
      "loss": 1.9383,
      "step": 195832
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.4080950021743774,
      "learning_rate": 3.2715780088943136e-05,
      "loss": 1.8291,
      "step": 195833
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.3233762979507446,
      "learning_rate": 3.2713922562839535e-05,
      "loss": 1.9344,
      "step": 195834
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.4188967943191528,
      "learning_rate": 3.271206508642945e-05,
      "loss": 1.8403,
      "step": 195835
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.4248822927474976,
      "learning_rate": 3.271020765971328e-05,
      "loss": 2.0816,
      "step": 195836
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.627394676208496,
      "learning_rate": 3.2708350282691266e-05,
      "loss": 2.1493,
      "step": 195837
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.415282130241394,
      "learning_rate": 3.270649295536386e-05,
      "loss": 2.1756,
      "step": 195838
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.5448095798492432,
      "learning_rate": 3.270463567773137e-05,
      "loss": 2.0242,
      "step": 195839
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.3212509155273438,
      "learning_rate": 3.270277844979417e-05,
      "loss": 1.8112,
      "step": 195840
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.5100879669189453,
      "learning_rate": 3.270092127155247e-05,
      "loss": 2.1234,
      "step": 195841
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.3285140991210938,
      "learning_rate": 3.269906414300679e-05,
      "loss": 2.0471,
      "step": 195842
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.5718024969100952,
      "learning_rate": 3.2697207064157327e-05,
      "loss": 2.0669,
      "step": 195843
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.4178245067596436,
      "learning_rate": 3.2695350035004574e-05,
      "loss": 2.0879,
      "step": 195844
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.3607431650161743,
      "learning_rate": 3.26934930555488e-05,
      "loss": 2.0366,
      "step": 195845
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.4066699743270874,
      "learning_rate": 3.2691636125790345e-05,
      "loss": 1.942,
      "step": 195846
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.4957386255264282,
      "learning_rate": 3.26897792457295e-05,
      "loss": 1.991,
      "step": 195847
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.475617527961731,
      "learning_rate": 3.268792241536673e-05,
      "loss": 2.03,
      "step": 195848
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.6047275066375732,
      "learning_rate": 3.268606563470224e-05,
      "loss": 1.9426,
      "step": 195849
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.429654836654663,
      "learning_rate": 3.268420890373656e-05,
      "loss": 2.2241,
      "step": 195850
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.3711504936218262,
      "learning_rate": 3.2682352222469896e-05,
      "loss": 2.0904,
      "step": 195851
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.381410837173462,
      "learning_rate": 3.268049559090261e-05,
      "loss": 1.8972,
      "step": 195852
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.493756651878357,
      "learning_rate": 3.26786390090351e-05,
      "loss": 2.0464,
      "step": 195853
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.4039279222488403,
      "learning_rate": 3.2676782476867666e-05,
      "loss": 1.8141,
      "step": 195854
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.4888074398040771,
      "learning_rate": 3.2674925994400616e-05,
      "loss": 1.8103,
      "step": 195855
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.374415636062622,
      "learning_rate": 3.2673069561634404e-05,
      "loss": 2.0483,
      "step": 195856
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.3912328481674194,
      "learning_rate": 3.26712131785693e-05,
      "loss": 1.9902,
      "step": 195857
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.6225357055664062,
      "learning_rate": 3.266935684520561e-05,
      "loss": 1.9002,
      "step": 195858
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.432306170463562,
      "learning_rate": 3.2667500561543825e-05,
      "loss": 1.9794,
      "step": 195859
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.3878729343414307,
      "learning_rate": 3.266564432758415e-05,
      "loss": 2.1822,
      "step": 195860
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.3761663436889648,
      "learning_rate": 3.266378814332695e-05,
      "loss": 1.8568,
      "step": 195861
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.259826898574829,
      "learning_rate": 3.266193200877266e-05,
      "loss": 2.0671,
      "step": 195862
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.485323429107666,
      "learning_rate": 3.2660075923921544e-05,
      "loss": 1.8652,
      "step": 195863
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.2842217683792114,
      "learning_rate": 3.265821988877391e-05,
      "loss": 1.8076,
      "step": 195864
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.4680696725845337,
      "learning_rate": 3.265636390333024e-05,
      "loss": 1.9726,
      "step": 195865
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.4783194065093994,
      "learning_rate": 3.2654507967590716e-05,
      "loss": 1.744,
      "step": 195866
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.4582279920578003,
      "learning_rate": 3.265265208155583e-05,
      "loss": 1.9887,
      "step": 195867
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.3395508527755737,
      "learning_rate": 3.2650796245225855e-05,
      "loss": 1.9358,
      "step": 195868
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.4806407690048218,
      "learning_rate": 3.264894045860115e-05,
      "loss": 2.0435,
      "step": 195869
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.3586797714233398,
      "learning_rate": 3.2647084721682016e-05,
      "loss": 2.0344,
      "step": 195870
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.3198601007461548,
      "learning_rate": 3.264522903446886e-05,
      "loss": 1.8704,
      "step": 195871
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.4796655178070068,
      "learning_rate": 3.264337339696197e-05,
      "loss": 2.0968,
      "step": 195872
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.3014702796936035,
      "learning_rate": 3.2641517809161754e-05,
      "loss": 2.0055,
      "step": 195873
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.401806354522705,
      "learning_rate": 3.263966227106848e-05,
      "loss": 1.9441,
      "step": 195874
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.5689561367034912,
      "learning_rate": 3.263780678268267e-05,
      "loss": 1.8518,
      "step": 195875
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.4276983737945557,
      "learning_rate": 3.263595134400441e-05,
      "loss": 2.1745,
      "step": 195876
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.3995203971862793,
      "learning_rate": 3.263409595503424e-05,
      "loss": 2.1612,
      "step": 195877
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.50981605052948,
      "learning_rate": 3.2632240615772385e-05,
      "loss": 2.1424,
      "step": 195878
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.5378433465957642,
      "learning_rate": 3.26303853262193e-05,
      "loss": 2.1026,
      "step": 195879
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.425890564918518,
      "learning_rate": 3.262853008637521e-05,
      "loss": 1.9252,
      "step": 195880
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.3275951147079468,
      "learning_rate": 3.2626674896240635e-05,
      "loss": 2.0651,
      "step": 195881
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.4597036838531494,
      "learning_rate": 3.262481975581569e-05,
      "loss": 2.1423,
      "step": 195882
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.4897615909576416,
      "learning_rate": 3.262296466510091e-05,
      "loss": 2.0899,
      "step": 195883
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.4992449283599854,
      "learning_rate": 3.2621109624096474e-05,
      "loss": 1.9113,
      "step": 195884
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.4540082216262817,
      "learning_rate": 3.261925463280291e-05,
      "loss": 2.3181,
      "step": 195885
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.3865900039672852,
      "learning_rate": 3.261739969122044e-05,
      "loss": 1.9519,
      "step": 195886
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.5140358209609985,
      "learning_rate": 3.2615544799349505e-05,
      "loss": 2.1298,
      "step": 195887
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.389789342880249,
      "learning_rate": 3.2613689957190306e-05,
      "loss": 2.0357,
      "step": 195888
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.4482287168502808,
      "learning_rate": 3.261183516474331e-05,
      "loss": 1.9999,
      "step": 195889
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.5461140871047974,
      "learning_rate": 3.2609980422008775e-05,
      "loss": 2.1047,
      "step": 195890
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.4562349319458008,
      "learning_rate": 3.260812572898714e-05,
      "loss": 1.9241,
      "step": 195891
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.6028193235397339,
      "learning_rate": 3.260627108567865e-05,
      "loss": 2.3311,
      "step": 195892
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.55134117603302,
      "learning_rate": 3.260441649208382e-05,
      "loss": 2.3171,
      "step": 195893
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.6143358945846558,
      "learning_rate": 3.260256194820276e-05,
      "loss": 1.7235,
      "step": 195894
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.464802861213684,
      "learning_rate": 3.260070745403599e-05,
      "loss": 1.9126,
      "step": 195895
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.5382609367370605,
      "learning_rate": 3.2598853009583726e-05,
      "loss": 2.0877,
      "step": 195896
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.3911279439926147,
      "learning_rate": 3.259699861484646e-05,
      "loss": 2.0529,
      "step": 195897
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.4654194116592407,
      "learning_rate": 3.2595144269824395e-05,
      "loss": 2.1255,
      "step": 195898
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.540970802307129,
      "learning_rate": 3.2593289974518055e-05,
      "loss": 2.2052,
      "step": 195899
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.7590641975402832,
      "learning_rate": 3.259143572892755e-05,
      "loss": 2.0658,
      "step": 195900
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.6130118370056152,
      "learning_rate": 3.258958153305341e-05,
      "loss": 2.0073,
      "step": 195901
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.4565914869308472,
      "learning_rate": 3.258772738689583e-05,
      "loss": 1.9409,
      "step": 195902
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.4160765409469604,
      "learning_rate": 3.2585873290455325e-05,
      "loss": 2.3259,
      "step": 195903
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.407004714012146,
      "learning_rate": 3.258401924373211e-05,
      "loss": 1.9097,
      "step": 195904
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.2757693529129028,
      "learning_rate": 3.2582165246726656e-05,
      "loss": 1.9605,
      "step": 195905
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.3212639093399048,
      "learning_rate": 3.258031129943913e-05,
      "loss": 1.9188,
      "step": 195906
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.4749606847763062,
      "learning_rate": 3.257845740187001e-05,
      "loss": 1.6995,
      "step": 195907
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.674731969833374,
      "learning_rate": 3.2576603554019576e-05,
      "loss": 1.8982,
      "step": 195908
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.481145977973938,
      "learning_rate": 3.2574749755888244e-05,
      "loss": 2.1423,
      "step": 195909
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.3650239706039429,
      "learning_rate": 3.2572896007476235e-05,
      "loss": 2.0261,
      "step": 195910
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.5576025247573853,
      "learning_rate": 3.257104230878406e-05,
      "loss": 1.8074,
      "step": 195911
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.3843995332717896,
      "learning_rate": 3.256918865981194e-05,
      "loss": 2.0582,
      "step": 195912
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.5012140274047852,
      "learning_rate": 3.2567335060560286e-05,
      "loss": 1.994,
      "step": 195913
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.4735833406448364,
      "learning_rate": 3.2565481511029356e-05,
      "loss": 1.843,
      "step": 195914
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.411589503288269,
      "learning_rate": 3.2563628011219586e-05,
      "loss": 1.9021,
      "step": 195915
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.3418232202529907,
      "learning_rate": 3.2561774561131246e-05,
      "loss": 1.9565,
      "step": 195916
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.326068639755249,
      "learning_rate": 3.255992116076477e-05,
      "loss": 1.8782,
      "step": 195917
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.4882570505142212,
      "learning_rate": 3.2558067810120484e-05,
      "loss": 2.0939,
      "step": 195918
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.3107473850250244,
      "learning_rate": 3.255621450919865e-05,
      "loss": 1.9877,
      "step": 195919
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.5258229970932007,
      "learning_rate": 3.255436125799965e-05,
      "loss": 1.9177,
      "step": 195920
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.4993886947631836,
      "learning_rate": 3.2552508056523876e-05,
      "loss": 2.0275,
      "step": 195921
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.3036049604415894,
      "learning_rate": 3.255065490477159e-05,
      "loss": 1.7916,
      "step": 195922
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.497497320175171,
      "learning_rate": 3.254880180274323e-05,
      "loss": 2.1841,
      "step": 195923
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.420738697052002,
      "learning_rate": 3.254694875043913e-05,
      "loss": 2.1085,
      "step": 195924
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.5408989191055298,
      "learning_rate": 3.254509574785955e-05,
      "loss": 2.0113,
      "step": 195925
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.6113629341125488,
      "learning_rate": 3.254324279500486e-05,
      "loss": 1.9589,
      "step": 195926
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.398199439048767,
      "learning_rate": 3.2541389891875505e-05,
      "loss": 2.0552,
      "step": 195927
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.326729655265808,
      "learning_rate": 3.2539537038471664e-05,
      "loss": 2.0138,
      "step": 195928
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.5233405828475952,
      "learning_rate": 3.253768423479385e-05,
      "loss": 1.9779,
      "step": 195929
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.3828301429748535,
      "learning_rate": 3.253583148084232e-05,
      "loss": 2.0592,
      "step": 195930
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.3988964557647705,
      "learning_rate": 3.253397877661742e-05,
      "loss": 2.1688,
      "step": 195931
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.467846155166626,
      "learning_rate": 3.253212612211944e-05,
      "loss": 1.8867,
      "step": 195932
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.625951886177063,
      "learning_rate": 3.253027351734888e-05,
      "loss": 1.987,
      "step": 195933
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.3967645168304443,
      "learning_rate": 3.252842096230591e-05,
      "loss": 2.0124,
      "step": 195934
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.2843220233917236,
      "learning_rate": 3.2526568456991e-05,
      "loss": 1.6757,
      "step": 195935
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.4436006546020508,
      "learning_rate": 3.252471600140447e-05,
      "loss": 1.9037,
      "step": 195936
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.529880404472351,
      "learning_rate": 3.252286359554657e-05,
      "loss": 2.0273,
      "step": 195937
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.6091840267181396,
      "learning_rate": 3.2521011239417785e-05,
      "loss": 2.0884,
      "step": 195938
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.4546757936477661,
      "learning_rate": 3.251915893301838e-05,
      "loss": 2.1698,
      "step": 195939
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.6512354612350464,
      "learning_rate": 3.251730667634868e-05,
      "loss": 1.9254,
      "step": 195940
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.51394784450531,
      "learning_rate": 3.251545446940912e-05,
      "loss": 2.0268,
      "step": 195941
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.5455942153930664,
      "learning_rate": 3.2513602312199985e-05,
      "loss": 2.0426,
      "step": 195942
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.3772085905075073,
      "learning_rate": 3.2511750204721534e-05,
      "loss": 1.9367,
      "step": 195943
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.4100335836410522,
      "learning_rate": 3.250989814697427e-05,
      "loss": 2.1789,
      "step": 195944
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.510815143585205,
      "learning_rate": 3.2508046138958486e-05,
      "loss": 1.9105,
      "step": 195945
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.3650513887405396,
      "learning_rate": 3.2506194180674426e-05,
      "loss": 2.0266,
      "step": 195946
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.5698602199554443,
      "learning_rate": 3.250434227212255e-05,
      "loss": 1.9067,
      "step": 195947
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.7102147340774536,
      "learning_rate": 3.2502490413303196e-05,
      "loss": 1.9892,
      "step": 195948
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.5362385511398315,
      "learning_rate": 3.2500638604216624e-05,
      "loss": 2.0228,
      "step": 195949
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.5125727653503418,
      "learning_rate": 3.24987868448633e-05,
      "loss": 1.8676,
      "step": 195950
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.5313373804092407,
      "learning_rate": 3.2496935135243426e-05,
      "loss": 1.8983,
      "step": 195951
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.312690258026123,
      "learning_rate": 3.249508347535747e-05,
      "loss": 1.8896,
      "step": 195952
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.418550729751587,
      "learning_rate": 3.2493231865205735e-05,
      "loss": 1.8164,
      "step": 195953
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.6175745725631714,
      "learning_rate": 3.249138030478855e-05,
      "loss": 1.9798,
      "step": 195954
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.7291232347488403,
      "learning_rate": 3.248952879410625e-05,
      "loss": 1.8631,
      "step": 195955
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.4661154747009277,
      "learning_rate": 3.248767733315919e-05,
      "loss": 1.8919,
      "step": 195956
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.5054054260253906,
      "learning_rate": 3.2485825921947714e-05,
      "loss": 2.0251,
      "step": 195957
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.6558579206466675,
      "learning_rate": 3.248397456047219e-05,
      "loss": 1.9168,
      "step": 195958
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.3860357999801636,
      "learning_rate": 3.2482123248732914e-05,
      "loss": 1.8435,
      "step": 195959
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.5568267107009888,
      "learning_rate": 3.248027198673039e-05,
      "loss": 2.2212,
      "step": 195960
    },
    {
      "epoch": 2.55,
      "grad_norm": 2.025348424911499,
      "learning_rate": 3.2478420774464706e-05,
      "loss": 1.9131,
      "step": 195961
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.690268635749817,
      "learning_rate": 3.247656961193637e-05,
      "loss": 1.788,
      "step": 195962
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.4520246982574463,
      "learning_rate": 3.247471849914566e-05,
      "loss": 2.1224,
      "step": 195963
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.4321177005767822,
      "learning_rate": 3.2472867436092983e-05,
      "loss": 1.9035,
      "step": 195964
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.4500929117202759,
      "learning_rate": 3.24710164227786e-05,
      "loss": 1.8694,
      "step": 195965
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.425824522972107,
      "learning_rate": 3.246916545920302e-05,
      "loss": 2.06,
      "step": 195966
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.5645800828933716,
      "learning_rate": 3.246731454536638e-05,
      "loss": 2.3583,
      "step": 195967
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.5691710710525513,
      "learning_rate": 3.246546368126913e-05,
      "loss": 1.9867,
      "step": 195968
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.5108497142791748,
      "learning_rate": 3.2463612866911596e-05,
      "loss": 1.8654,
      "step": 195969
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.6808186769485474,
      "learning_rate": 3.2461762102294134e-05,
      "loss": 2.0598,
      "step": 195970
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.5213210582733154,
      "learning_rate": 3.245991138741705e-05,
      "loss": 2.0491,
      "step": 195971
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.4417661428451538,
      "learning_rate": 3.245806072228084e-05,
      "loss": 1.941,
      "step": 195972
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.4629106521606445,
      "learning_rate": 3.245621010688558e-05,
      "loss": 1.9265,
      "step": 195973
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.5917778015136719,
      "learning_rate": 3.245435954123187e-05,
      "loss": 1.8899,
      "step": 195974
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.436754822731018,
      "learning_rate": 3.245250902531983e-05,
      "loss": 2.2746,
      "step": 195975
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.4562963247299194,
      "learning_rate": 3.245065855915e-05,
      "loss": 2.0191,
      "step": 195976
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.4304263591766357,
      "learning_rate": 3.244880814272258e-05,
      "loss": 1.8806,
      "step": 195977
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.420896053314209,
      "learning_rate": 3.244695777603807e-05,
      "loss": 1.8338,
      "step": 195978
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.3766000270843506,
      "learning_rate": 3.244510745909667e-05,
      "loss": 1.7848,
      "step": 195979
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.4929002523422241,
      "learning_rate": 3.244325719189882e-05,
      "loss": 2.0276,
      "step": 195980
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.4194355010986328,
      "learning_rate": 3.244140697444474e-05,
      "loss": 2.1631,
      "step": 195981
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.3377031087875366,
      "learning_rate": 3.24395568067349e-05,
      "loss": 1.9429,
      "step": 195982
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.5736037492752075,
      "learning_rate": 3.2437706688769536e-05,
      "loss": 2.187,
      "step": 195983
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.4061030149459839,
      "learning_rate": 3.243585662054915e-05,
      "loss": 2.0682,
      "step": 195984
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.4430017471313477,
      "learning_rate": 3.243400660207394e-05,
      "loss": 2.0124,
      "step": 195985
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.4800634384155273,
      "learning_rate": 3.243215663334433e-05,
      "loss": 2.0758,
      "step": 195986
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.3299450874328613,
      "learning_rate": 3.243030671436057e-05,
      "loss": 1.87,
      "step": 195987
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.463482141494751,
      "learning_rate": 3.242845684512311e-05,
      "loss": 2.1582,
      "step": 195988
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.5409640073776245,
      "learning_rate": 3.2426607025632225e-05,
      "loss": 1.8305,
      "step": 195989
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.5044097900390625,
      "learning_rate": 3.242475725588831e-05,
      "loss": 2.1135,
      "step": 195990
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.3567590713500977,
      "learning_rate": 3.242290753589171e-05,
      "loss": 2.0579,
      "step": 195991
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.4468908309936523,
      "learning_rate": 3.242105786564271e-05,
      "loss": 2.0817,
      "step": 195992
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.3797080516815186,
      "learning_rate": 3.2419208245141646e-05,
      "loss": 1.9978,
      "step": 195993
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.5423204898834229,
      "learning_rate": 3.241735867438896e-05,
      "loss": 2.0671,
      "step": 195994
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.5161149501800537,
      "learning_rate": 3.241550915338488e-05,
      "loss": 2.1347,
      "step": 195995
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.4306247234344482,
      "learning_rate": 3.241365968212987e-05,
      "loss": 2.177,
      "step": 195996
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.7052534818649292,
      "learning_rate": 3.2411810260624226e-05,
      "loss": 2.104,
      "step": 195997
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.4051133394241333,
      "learning_rate": 3.240996088886826e-05,
      "loss": 2.2067,
      "step": 195998
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.332877516746521,
      "learning_rate": 3.240811156686226e-05,
      "loss": 2.0507,
      "step": 195999
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.370376467704773,
      "learning_rate": 3.240626229460673e-05,
      "loss": 2.0089,
      "step": 196000
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.3898544311523438,
      "learning_rate": 3.240441307210184e-05,
      "loss": 1.9623,
      "step": 196001
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.4807970523834229,
      "learning_rate": 3.2402563899348087e-05,
      "loss": 1.7577,
      "step": 196002
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.4503378868103027,
      "learning_rate": 3.240071477634577e-05,
      "loss": 2.02,
      "step": 196003
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.524748682975769,
      "learning_rate": 3.239886570309519e-05,
      "loss": 1.9153,
      "step": 196004
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.4519366025924683,
      "learning_rate": 3.239701667959668e-05,
      "loss": 1.8522,
      "step": 196005
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.3565599918365479,
      "learning_rate": 3.2395167705850635e-05,
      "loss": 1.9857,
      "step": 196006
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.6559698581695557,
      "learning_rate": 3.2393318781857356e-05,
      "loss": 2.1761,
      "step": 196007
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.4219039678573608,
      "learning_rate": 3.239146990761725e-05,
      "loss": 2.0423,
      "step": 196008
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.3880196809768677,
      "learning_rate": 3.2389621083130615e-05,
      "loss": 2.2003,
      "step": 196009
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.4910372495651245,
      "learning_rate": 3.238777230839781e-05,
      "loss": 1.955,
      "step": 196010
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.3661786317825317,
      "learning_rate": 3.238592358341914e-05,
      "loss": 1.8916,
      "step": 196011
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.3223273754119873,
      "learning_rate": 3.2384074908195004e-05,
      "loss": 1.9286,
      "step": 196012
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.3571679592132568,
      "learning_rate": 3.238222628272567e-05,
      "loss": 1.986,
      "step": 196013
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.5214474201202393,
      "learning_rate": 3.238037770701156e-05,
      "loss": 1.6702,
      "step": 196014
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.293835163116455,
      "learning_rate": 3.237852918105302e-05,
      "loss": 2.0659,
      "step": 196015
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.4896306991577148,
      "learning_rate": 3.2376680704850386e-05,
      "loss": 1.9022,
      "step": 196016
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.461392879486084,
      "learning_rate": 3.237483227840388e-05,
      "loss": 1.9646,
      "step": 196017
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.4894630908966064,
      "learning_rate": 3.237298390171401e-05,
      "loss": 1.8831,
      "step": 196018
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.4618834257125854,
      "learning_rate": 3.2371135574781e-05,
      "loss": 2.0476,
      "step": 196019
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.471729040145874,
      "learning_rate": 3.236928729760533e-05,
      "loss": 2.0767,
      "step": 196020
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.5685272216796875,
      "learning_rate": 3.236743907018725e-05,
      "loss": 2.0206,
      "step": 196021
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.560278058052063,
      "learning_rate": 3.236559089252704e-05,
      "loss": 1.9858,
      "step": 196022
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.484108567237854,
      "learning_rate": 3.236374276462519e-05,
      "loss": 1.9787,
      "step": 196023
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.4283947944641113,
      "learning_rate": 3.236189468648197e-05,
      "loss": 1.9354,
      "step": 196024
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.4637714624404907,
      "learning_rate": 3.236004665809765e-05,
      "loss": 2.0573,
      "step": 196025
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.427901268005371,
      "learning_rate": 3.2358198679472726e-05,
      "loss": 2.0543,
      "step": 196026
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.2769076824188232,
      "learning_rate": 3.2356350750607466e-05,
      "loss": 1.7465,
      "step": 196027
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.5385960340499878,
      "learning_rate": 3.2354502871502166e-05,
      "loss": 2.0735,
      "step": 196028
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.4834747314453125,
      "learning_rate": 3.2352655042157226e-05,
      "loss": 2.0688,
      "step": 196029
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.4741376638412476,
      "learning_rate": 3.235080726257302e-05,
      "loss": 2.0993,
      "step": 196030
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.4569438695907593,
      "learning_rate": 3.234895953274981e-05,
      "loss": 2.0436,
      "step": 196031
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.6303976774215698,
      "learning_rate": 3.234711185268799e-05,
      "loss": 1.922,
      "step": 196032
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.5829530954360962,
      "learning_rate": 3.2345264222387936e-05,
      "loss": 1.9875,
      "step": 196033
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.4423179626464844,
      "learning_rate": 3.234341664184987e-05,
      "loss": 1.9245,
      "step": 196034
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.3436570167541504,
      "learning_rate": 3.2341569111074274e-05,
      "loss": 2.2294,
      "step": 196035
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.6276917457580566,
      "learning_rate": 3.233972163006137e-05,
      "loss": 2.1345,
      "step": 196036
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.4044722318649292,
      "learning_rate": 3.233787419881162e-05,
      "loss": 2.0154,
      "step": 196037
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.324808120727539,
      "learning_rate": 3.2336026817325335e-05,
      "loss": 1.8033,
      "step": 196038
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.4364852905273438,
      "learning_rate": 3.2334179485602805e-05,
      "loss": 2.1487,
      "step": 196039
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.4374711513519287,
      "learning_rate": 3.2332332203644376e-05,
      "loss": 1.8967,
      "step": 196040
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.5658249855041504,
      "learning_rate": 3.233048497145047e-05,
      "loss": 2.1839,
      "step": 196041
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.3712098598480225,
      "learning_rate": 3.2328637789021316e-05,
      "loss": 2.0159,
      "step": 196042
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.450691819190979,
      "learning_rate": 3.232679065635736e-05,
      "loss": 2.1459,
      "step": 196043
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.3359310626983643,
      "learning_rate": 3.232494357345893e-05,
      "loss": 2.1238,
      "step": 196044
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.3550691604614258,
      "learning_rate": 3.2323096540326286e-05,
      "loss": 1.9928,
      "step": 196045
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.428099513053894,
      "learning_rate": 3.2321249556959906e-05,
      "loss": 2.0086,
      "step": 196046
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.4545228481292725,
      "learning_rate": 3.231940262336001e-05,
      "loss": 1.919,
      "step": 196047
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.4378615617752075,
      "learning_rate": 3.231755573952698e-05,
      "loss": 2.1888,
      "step": 196048
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.3898378610610962,
      "learning_rate": 3.23157089054612e-05,
      "loss": 1.8666,
      "step": 196049
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.355302095413208,
      "learning_rate": 3.231386212116291e-05,
      "loss": 1.8012,
      "step": 196050
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.4958900213241577,
      "learning_rate": 3.231201538663262e-05,
      "loss": 2.0559,
      "step": 196051
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.4630142450332642,
      "learning_rate": 3.2310168701870575e-05,
      "loss": 1.8306,
      "step": 196052
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.537358045578003,
      "learning_rate": 3.2308322066877124e-05,
      "loss": 1.8483,
      "step": 196053
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.3979487419128418,
      "learning_rate": 3.230647548165253e-05,
      "loss": 2.082,
      "step": 196054
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.3742026090621948,
      "learning_rate": 3.230462894619729e-05,
      "loss": 2.0306,
      "step": 196055
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.4991238117218018,
      "learning_rate": 3.23027824605116e-05,
      "loss": 2.104,
      "step": 196056
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.444875955581665,
      "learning_rate": 3.230093602459594e-05,
      "loss": 1.9148,
      "step": 196057
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.4233773946762085,
      "learning_rate": 3.229908963845059e-05,
      "loss": 1.8664,
      "step": 196058
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.4527183771133423,
      "learning_rate": 3.22972433020759e-05,
      "loss": 2.0314,
      "step": 196059
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.4162116050720215,
      "learning_rate": 3.2295397015472135e-05,
      "loss": 2.2678,
      "step": 196060
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.4127203226089478,
      "learning_rate": 3.2293550778639755e-05,
      "loss": 2.1633,
      "step": 196061
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.3923289775848389,
      "learning_rate": 3.2291704591579024e-05,
      "loss": 1.9891,
      "step": 196062
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.6900385618209839,
      "learning_rate": 3.228985845429038e-05,
      "loss": 2.1269,
      "step": 196063
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.4449816942214966,
      "learning_rate": 3.2288012366774095e-05,
      "loss": 2.0698,
      "step": 196064
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.3534762859344482,
      "learning_rate": 3.228616632903053e-05,
      "loss": 2.0499,
      "step": 196065
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.3852477073669434,
      "learning_rate": 3.2284320341059944e-05,
      "loss": 1.9597,
      "step": 196066
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.5029224157333374,
      "learning_rate": 3.228247440286281e-05,
      "loss": 2.0618,
      "step": 196067
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.4029576778411865,
      "learning_rate": 3.2280628514439396e-05,
      "loss": 2.054,
      "step": 196068
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.373640775680542,
      "learning_rate": 3.22787826757901e-05,
      "loss": 2.0232,
      "step": 196069
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.4196934700012207,
      "learning_rate": 3.227693688691523e-05,
      "loss": 1.9498,
      "step": 196070
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.488855004310608,
      "learning_rate": 3.227509114781514e-05,
      "loss": 2.2518,
      "step": 196071
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.4371240139007568,
      "learning_rate": 3.2273245458490096e-05,
      "loss": 1.6721,
      "step": 196072
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.6426202058792114,
      "learning_rate": 3.2271399818940566e-05,
      "loss": 1.8473,
      "step": 196073
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.4374696016311646,
      "learning_rate": 3.226955422916679e-05,
      "loss": 2.2075,
      "step": 196074
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.6127445697784424,
      "learning_rate": 3.2267708689169206e-05,
      "loss": 2.0147,
      "step": 196075
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.5713576078414917,
      "learning_rate": 3.226586319894813e-05,
      "loss": 2.1473,
      "step": 196076
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.5611385107040405,
      "learning_rate": 3.226401775850387e-05,
      "loss": 1.9373,
      "step": 196077
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.2918621301651,
      "learning_rate": 3.226217236783672e-05,
      "loss": 1.9988,
      "step": 196078
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.4487582445144653,
      "learning_rate": 3.226032702694713e-05,
      "loss": 2.1216,
      "step": 196079
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.5155420303344727,
      "learning_rate": 3.2258481735835354e-05,
      "loss": 2.0859,
      "step": 196080
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.3121404647827148,
      "learning_rate": 3.225663649450185e-05,
      "loss": 2.0371,
      "step": 196081
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.5913147926330566,
      "learning_rate": 3.22547913029469e-05,
      "loss": 1.7204,
      "step": 196082
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.4669638872146606,
      "learning_rate": 3.22529461611708e-05,
      "loss": 1.934,
      "step": 196083
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.3712750673294067,
      "learning_rate": 3.225110106917391e-05,
      "loss": 1.7648,
      "step": 196084
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.4186292886734009,
      "learning_rate": 3.2249256026956637e-05,
      "loss": 1.9632,
      "step": 196085
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.391882300376892,
      "learning_rate": 3.224741103451921e-05,
      "loss": 2.0848,
      "step": 196086
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.6826167106628418,
      "learning_rate": 3.224556609186213e-05,
      "loss": 2.1304,
      "step": 196087
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.4145305156707764,
      "learning_rate": 3.2243721198985625e-05,
      "loss": 1.9254,
      "step": 196088
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.4988569021224976,
      "learning_rate": 3.22418763558901e-05,
      "loss": 2.0883,
      "step": 196089
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.447131633758545,
      "learning_rate": 3.22400315625758e-05,
      "loss": 1.9277,
      "step": 196090
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.3319002389907837,
      "learning_rate": 3.2238186819043164e-05,
      "loss": 1.9214,
      "step": 196091
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.2854543924331665,
      "learning_rate": 3.2236342125292446e-05,
      "loss": 2.1934,
      "step": 196092
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.4361191987991333,
      "learning_rate": 3.223449748132414e-05,
      "loss": 2.1072,
      "step": 196093
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.9495041370391846,
      "learning_rate": 3.223265288713844e-05,
      "loss": 2.0989,
      "step": 196094
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.2654998302459717,
      "learning_rate": 3.2230808342735794e-05,
      "loss": 2.0378,
      "step": 196095
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.4798929691314697,
      "learning_rate": 3.2228963848116416e-05,
      "loss": 2.0078,
      "step": 196096
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.3954848051071167,
      "learning_rate": 3.222711940328079e-05,
      "loss": 2.1431,
      "step": 196097
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.3673107624053955,
      "learning_rate": 3.222527500822913e-05,
      "loss": 2.0275,
      "step": 196098
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.28287672996521,
      "learning_rate": 3.222343066296192e-05,
      "loss": 1.93,
      "step": 196099
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.4202194213867188,
      "learning_rate": 3.222158636747942e-05,
      "loss": 1.9973,
      "step": 196100
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.4052424430847168,
      "learning_rate": 3.2219742121782e-05,
      "loss": 1.9518,
      "step": 196101
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.4099242687225342,
      "learning_rate": 3.2217897925869916e-05,
      "loss": 2.0733,
      "step": 196102
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.3051759004592896,
      "learning_rate": 3.221605377974364e-05,
      "loss": 1.9642,
      "step": 196103
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.371634602546692,
      "learning_rate": 3.221420968340338e-05,
      "loss": 2.0698,
      "step": 196104
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.498162865638733,
      "learning_rate": 3.2212365636849624e-05,
      "loss": 1.9965,
      "step": 196105
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.277347445487976,
      "learning_rate": 3.221052164008264e-05,
      "loss": 2.039,
      "step": 196106
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.3792325258255005,
      "learning_rate": 3.220867769310277e-05,
      "loss": 1.9166,
      "step": 196107
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.3680375814437866,
      "learning_rate": 3.22068337959103e-05,
      "loss": 2.0362,
      "step": 196108
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.4484903812408447,
      "learning_rate": 3.220498994850571e-05,
      "loss": 1.9733,
      "step": 196109
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.4906671047210693,
      "learning_rate": 3.220314615088919e-05,
      "loss": 2.0341,
      "step": 196110
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.3979963064193726,
      "learning_rate": 3.220130240306125e-05,
      "loss": 1.7993,
      "step": 196111
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.3670483827590942,
      "learning_rate": 3.2199458705022074e-05,
      "loss": 1.9732,
      "step": 196112
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.4532155990600586,
      "learning_rate": 3.219761505677211e-05,
      "loss": 1.935,
      "step": 196113
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.4254928827285767,
      "learning_rate": 3.219577145831168e-05,
      "loss": 2.1248,
      "step": 196114
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.4094387292861938,
      "learning_rate": 3.2193927909641094e-05,
      "loss": 1.9758,
      "step": 196115
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.517785668373108,
      "learning_rate": 3.2192084410760675e-05,
      "loss": 2.0343,
      "step": 196116
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.4084354639053345,
      "learning_rate": 3.2190240961670866e-05,
      "loss": 1.8474,
      "step": 196117
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.3996014595031738,
      "learning_rate": 3.218839756237186e-05,
      "loss": 1.8918,
      "step": 196118
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.4436373710632324,
      "learning_rate": 3.218655421286416e-05,
      "loss": 1.9491,
      "step": 196119
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.4667115211486816,
      "learning_rate": 3.218471091314806e-05,
      "loss": 2.1204,
      "step": 196120
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.3240679502487183,
      "learning_rate": 3.21828676632238e-05,
      "loss": 2.0412,
      "step": 196121
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.352612018585205,
      "learning_rate": 3.218102446309184e-05,
      "loss": 2.0953,
      "step": 196122
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.5182511806488037,
      "learning_rate": 3.217918131275249e-05,
      "loss": 1.8748,
      "step": 196123
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.2827006578445435,
      "learning_rate": 3.217733821220604e-05,
      "loss": 1.8826,
      "step": 196124
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.370206356048584,
      "learning_rate": 3.2175495161452927e-05,
      "loss": 2.1869,
      "step": 196125
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.4772658348083496,
      "learning_rate": 3.217365216049345e-05,
      "loss": 2.1213,
      "step": 196126
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.4196221828460693,
      "learning_rate": 3.21718092093279e-05,
      "loss": 1.8516,
      "step": 196127
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.2756669521331787,
      "learning_rate": 3.2169966307956696e-05,
      "loss": 1.9669,
      "step": 196128
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.5196046829223633,
      "learning_rate": 3.216812345638016e-05,
      "loss": 2.0052,
      "step": 196129
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.4156146049499512,
      "learning_rate": 3.2166280654598584e-05,
      "loss": 2.008,
      "step": 196130
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.4704649448394775,
      "learning_rate": 3.216443790261242e-05,
      "loss": 1.9659,
      "step": 196131
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.383123517036438,
      "learning_rate": 3.2162595200421916e-05,
      "loss": 2.2261,
      "step": 196132
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.2881546020507812,
      "learning_rate": 3.216075254802738e-05,
      "loss": 2.0469,
      "step": 196133
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.2581099271774292,
      "learning_rate": 3.215890994542928e-05,
      "loss": 1.8382,
      "step": 196134
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.4531728029251099,
      "learning_rate": 3.2157067392627846e-05,
      "loss": 1.9464,
      "step": 196135
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.391438364982605,
      "learning_rate": 3.215522488962351e-05,
      "loss": 2.1079,
      "step": 196136
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.502493143081665,
      "learning_rate": 3.215338243641661e-05,
      "loss": 2.0591,
      "step": 196137
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.3626587390899658,
      "learning_rate": 3.2151540033007405e-05,
      "loss": 2.0811,
      "step": 196138
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.4657403230667114,
      "learning_rate": 3.214969767939627e-05,
      "loss": 2.0233,
      "step": 196139
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.485485553741455,
      "learning_rate": 3.21478553755836e-05,
      "loss": 2.0083,
      "step": 196140
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.5598177909851074,
      "learning_rate": 3.214601312156963e-05,
      "loss": 1.738,
      "step": 196141
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.3915523290634155,
      "learning_rate": 3.214417091735485e-05,
      "loss": 1.948,
      "step": 196142
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.32008957862854,
      "learning_rate": 3.2142328762939505e-05,
      "loss": 2.0142,
      "step": 196143
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.360876441001892,
      "learning_rate": 3.214048665832399e-05,
      "loss": 1.9556,
      "step": 196144
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.4287066459655762,
      "learning_rate": 3.213864460350851e-05,
      "loss": 1.9585,
      "step": 196145
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.5679887533187866,
      "learning_rate": 3.2136802598493625e-05,
      "loss": 2.0923,
      "step": 196146
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.406976580619812,
      "learning_rate": 3.213496064327947e-05,
      "loss": 2.1029,
      "step": 196147
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.5102342367172241,
      "learning_rate": 3.213311873786655e-05,
      "loss": 1.9565,
      "step": 196148
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.3805434703826904,
      "learning_rate": 3.213127688225512e-05,
      "loss": 1.9649,
      "step": 196149
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.34510338306427,
      "learning_rate": 3.2129435076445584e-05,
      "loss": 1.9238,
      "step": 196150
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.6034018993377686,
      "learning_rate": 3.212759332043815e-05,
      "loss": 2.1354,
      "step": 196151
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.3811383247375488,
      "learning_rate": 3.212575161423331e-05,
      "loss": 2.011,
      "step": 196152
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.507946252822876,
      "learning_rate": 3.212390995783133e-05,
      "loss": 2.2563,
      "step": 196153
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.5550943613052368,
      "learning_rate": 3.212206835123258e-05,
      "loss": 1.6754,
      "step": 196154
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.5593211650848389,
      "learning_rate": 3.212022679443742e-05,
      "loss": 2.2208,
      "step": 196155
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.6420100927352905,
      "learning_rate": 3.211838528744616e-05,
      "loss": 1.7547,
      "step": 196156
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.3906280994415283,
      "learning_rate": 3.211654383025909e-05,
      "loss": 1.9684,
      "step": 196157
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.532397747039795,
      "learning_rate": 3.2114702422876684e-05,
      "loss": 1.8916,
      "step": 196158
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.350641131401062,
      "learning_rate": 3.2112861065299134e-05,
      "loss": 1.9791,
      "step": 196159
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.3883295059204102,
      "learning_rate": 3.2111019757526945e-05,
      "loss": 2.2071,
      "step": 196160
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.5765068531036377,
      "learning_rate": 3.210917849956035e-05,
      "loss": 1.8336,
      "step": 196161
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.515075445175171,
      "learning_rate": 3.2107337291399715e-05,
      "loss": 1.9536,
      "step": 196162
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.5694547891616821,
      "learning_rate": 3.210549613304534e-05,
      "loss": 2.1836,
      "step": 196163
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.4526374340057373,
      "learning_rate": 3.2103655024497656e-05,
      "loss": 2.1758,
      "step": 196164
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.3035134077072144,
      "learning_rate": 3.2101813965756894e-05,
      "loss": 1.8856,
      "step": 196165
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.4694788455963135,
      "learning_rate": 3.2099972956823526e-05,
      "loss": 2.1772,
      "step": 196166
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.5765986442565918,
      "learning_rate": 3.209813199769785e-05,
      "loss": 2.1502,
      "step": 196167
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.405446171760559,
      "learning_rate": 3.2096291088380165e-05,
      "loss": 1.9589,
      "step": 196168
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.344144582748413,
      "learning_rate": 3.2094450228870775e-05,
      "loss": 2.0223,
      "step": 196169
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.421661376953125,
      "learning_rate": 3.209260941917017e-05,
      "loss": 1.8739,
      "step": 196170
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.4342292547225952,
      "learning_rate": 3.209076865927852e-05,
      "loss": 1.892,
      "step": 196171
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.4915887117385864,
      "learning_rate": 3.208892794919633e-05,
      "loss": 2.1742,
      "step": 196172
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.332770824432373,
      "learning_rate": 3.2087087288923796e-05,
      "loss": 2.0481,
      "step": 196173
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.5097652673721313,
      "learning_rate": 3.208524667846145e-05,
      "loss": 1.9923,
      "step": 196174
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.4167335033416748,
      "learning_rate": 3.2083406117809396e-05,
      "loss": 1.9515,
      "step": 196175
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.499739646911621,
      "learning_rate": 3.208156560696816e-05,
      "loss": 1.7322,
      "step": 196176
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.3390190601348877,
      "learning_rate": 3.207972514593795e-05,
      "loss": 2.1291,
      "step": 196177
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.5177525281906128,
      "learning_rate": 3.207788473471923e-05,
      "loss": 2.1821,
      "step": 196178
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.630926489830017,
      "learning_rate": 3.207604437331226e-05,
      "loss": 2.2509,
      "step": 196179
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.4744043350219727,
      "learning_rate": 3.207420406171748e-05,
      "loss": 2.3619,
      "step": 196180
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.5026085376739502,
      "learning_rate": 3.207236379993508e-05,
      "loss": 1.9675,
      "step": 196181
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.4010764360427856,
      "learning_rate": 3.207052358796555e-05,
      "loss": 1.8565,
      "step": 196182
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.4857463836669922,
      "learning_rate": 3.2068683425809094e-05,
      "loss": 2.0038,
      "step": 196183
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.5847716331481934,
      "learning_rate": 3.206684331346619e-05,
      "loss": 1.9333,
      "step": 196184
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.6837551593780518,
      "learning_rate": 3.206500325093705e-05,
      "loss": 1.7038,
      "step": 196185
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.4409189224243164,
      "learning_rate": 3.206316323822219e-05,
      "loss": 2.0798,
      "step": 196186
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.4493783712387085,
      "learning_rate": 3.206132327532175e-05,
      "loss": 2.1206,
      "step": 196187
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.2519371509552002,
      "learning_rate": 3.205948336223623e-05,
      "loss": 1.8779,
      "step": 196188
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.5123177766799927,
      "learning_rate": 3.2057643498965825e-05,
      "loss": 1.9109,
      "step": 196189
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.4339112043380737,
      "learning_rate": 3.2055803685511036e-05,
      "loss": 1.9588,
      "step": 196190
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.575081467628479,
      "learning_rate": 3.20539639218721e-05,
      "loss": 1.9764,
      "step": 196191
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.34734046459198,
      "learning_rate": 3.205212420804949e-05,
      "loss": 1.9332,
      "step": 196192
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.4385011196136475,
      "learning_rate": 3.205028454404329e-05,
      "loss": 2.0156,
      "step": 196193
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.4915982484817505,
      "learning_rate": 3.204844492985411e-05,
      "loss": 2.0548,
      "step": 196194
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.4654855728149414,
      "learning_rate": 3.204660536548211e-05,
      "loss": 2.0467,
      "step": 196195
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.4021542072296143,
      "learning_rate": 3.204476585092773e-05,
      "loss": 1.8761,
      "step": 196196
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.3731659650802612,
      "learning_rate": 3.204292638619127e-05,
      "loss": 1.8343,
      "step": 196197
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.4161795377731323,
      "learning_rate": 3.204108697127312e-05,
      "loss": 1.9754,
      "step": 196198
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.3735365867614746,
      "learning_rate": 3.2039247606173625e-05,
      "loss": 1.9508,
      "step": 196199
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.454207420349121,
      "learning_rate": 3.203740829089308e-05,
      "loss": 2.0666,
      "step": 196200
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.4968984127044678,
      "learning_rate": 3.203556902543174e-05,
      "loss": 1.8882,
      "step": 196201
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.4317809343338013,
      "learning_rate": 3.203372980979015e-05,
      "loss": 2.0624,
      "step": 196202
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.4182982444763184,
      "learning_rate": 3.2031890643968447e-05,
      "loss": 2.0511,
      "step": 196203
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.536134123802185,
      "learning_rate": 3.203005152796716e-05,
      "loss": 1.8981,
      "step": 196204
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.2993863821029663,
      "learning_rate": 3.202821246178655e-05,
      "loss": 2.1682,
      "step": 196205
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.3068559169769287,
      "learning_rate": 3.2026373445426954e-05,
      "loss": 1.9392,
      "step": 196206
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.4251737594604492,
      "learning_rate": 3.202453447888864e-05,
      "loss": 1.8884,
      "step": 196207
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.6873880624771118,
      "learning_rate": 3.202269556217211e-05,
      "loss": 2.0941,
      "step": 196208
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.4110074043273926,
      "learning_rate": 3.202085669527752e-05,
      "loss": 1.901,
      "step": 196209
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.5011703968048096,
      "learning_rate": 3.201901787820538e-05,
      "loss": 2.1566,
      "step": 196210
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.480051875114441,
      "learning_rate": 3.201717911095598e-05,
      "loss": 1.8321,
      "step": 196211
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.3400248289108276,
      "learning_rate": 3.20153403935296e-05,
      "loss": 2.042,
      "step": 196212
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.3487259149551392,
      "learning_rate": 3.201350172592666e-05,
      "loss": 2.0057,
      "step": 196213
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.6194405555725098,
      "learning_rate": 3.201166310814747e-05,
      "loss": 1.6527,
      "step": 196214
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.302293300628662,
      "learning_rate": 3.200982454019232e-05,
      "loss": 2.0342,
      "step": 196215
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.4532357454299927,
      "learning_rate": 3.200798602206165e-05,
      "loss": 2.0359,
      "step": 196216
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.4115564823150635,
      "learning_rate": 3.200614755375576e-05,
      "loss": 1.9417,
      "step": 196217
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.3399847745895386,
      "learning_rate": 3.200430913527491e-05,
      "loss": 1.9166,
      "step": 196218
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.384021520614624,
      "learning_rate": 3.2002470766619605e-05,
      "loss": 1.8855,
      "step": 196219
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.5204893350601196,
      "learning_rate": 3.200063244779004e-05,
      "loss": 2.1403,
      "step": 196220
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.4404083490371704,
      "learning_rate": 3.199879417878666e-05,
      "loss": 2.1242,
      "step": 196221
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.4069849252700806,
      "learning_rate": 3.199695595960975e-05,
      "loss": 1.8563,
      "step": 196222
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.5511391162872314,
      "learning_rate": 3.1995117790259684e-05,
      "loss": 2.0529,
      "step": 196223
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.4012831449508667,
      "learning_rate": 3.199327967073673e-05,
      "loss": 1.8532,
      "step": 196224
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.4194940328598022,
      "learning_rate": 3.1991441601041344e-05,
      "loss": 1.9812,
      "step": 196225
    },
    {
      "epoch": 2.55,
      "grad_norm": 1.378427505493164,
      "learning_rate": 3.198960358117374e-05,
      "loss": 2.1,
      "step": 196226
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.3261899948120117,
      "learning_rate": 3.198776561113441e-05,
      "loss": 1.9673,
      "step": 196227
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.4104586839675903,
      "learning_rate": 3.1985927690923585e-05,
      "loss": 2.0053,
      "step": 196228
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.3307366371154785,
      "learning_rate": 3.1984089820541634e-05,
      "loss": 1.8663,
      "step": 196229
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.5613621473312378,
      "learning_rate": 3.198225199998886e-05,
      "loss": 1.9654,
      "step": 196230
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.3805005550384521,
      "learning_rate": 3.198041422926569e-05,
      "loss": 1.9399,
      "step": 196231
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.6749731302261353,
      "learning_rate": 3.197857650837237e-05,
      "loss": 2.1681,
      "step": 196232
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.5462204217910767,
      "learning_rate": 3.197673883730935e-05,
      "loss": 1.9652,
      "step": 196233
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.5409293174743652,
      "learning_rate": 3.19749012160769e-05,
      "loss": 1.9413,
      "step": 196234
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.5630571842193604,
      "learning_rate": 3.1973063644675386e-05,
      "loss": 2.0844,
      "step": 196235
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.4591771364212036,
      "learning_rate": 3.1971226123105086e-05,
      "loss": 2.0236,
      "step": 196236
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.4959907531738281,
      "learning_rate": 3.1969388651366425e-05,
      "loss": 1.9843,
      "step": 196237
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.6381219625473022,
      "learning_rate": 3.1967551229459695e-05,
      "loss": 2.1161,
      "step": 196238
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.5246020555496216,
      "learning_rate": 3.1965713857385276e-05,
      "loss": 2.1059,
      "step": 196239
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.4745752811431885,
      "learning_rate": 3.1963876535143464e-05,
      "loss": 2.0282,
      "step": 196240
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.358879566192627,
      "learning_rate": 3.1962039262734716e-05,
      "loss": 2.1864,
      "step": 196241
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.3793585300445557,
      "learning_rate": 3.196020204015921e-05,
      "loss": 1.9253,
      "step": 196242
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.4912899732589722,
      "learning_rate": 3.1958364867417375e-05,
      "loss": 1.8088,
      "step": 196243
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.6072015762329102,
      "learning_rate": 3.195652774450951e-05,
      "loss": 2.2035,
      "step": 196244
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.446192979812622,
      "learning_rate": 3.1954690671436045e-05,
      "loss": 2.0046,
      "step": 196245
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.5909459590911865,
      "learning_rate": 3.195285364819718e-05,
      "loss": 2.0582,
      "step": 196246
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.3978630304336548,
      "learning_rate": 3.195101667479346e-05,
      "loss": 1.8597,
      "step": 196247
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.6236824989318848,
      "learning_rate": 3.194917975122501e-05,
      "loss": 2.0548,
      "step": 196248
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.5865164995193481,
      "learning_rate": 3.194734287749232e-05,
      "loss": 2.0849,
      "step": 196249
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.4749284982681274,
      "learning_rate": 3.194550605359561e-05,
      "loss": 1.9842,
      "step": 196250
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.5397272109985352,
      "learning_rate": 3.194366927953535e-05,
      "loss": 2.1387,
      "step": 196251
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.4115427732467651,
      "learning_rate": 3.194183255531177e-05,
      "loss": 1.9348,
      "step": 196252
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.4220073223114014,
      "learning_rate": 3.1939995880925394e-05,
      "loss": 2.1634,
      "step": 196253
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.5182372331619263,
      "learning_rate": 3.1938159256376287e-05,
      "loss": 2.0652,
      "step": 196254
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.5183292627334595,
      "learning_rate": 3.1936322681665005e-05,
      "loss": 2.0146,
      "step": 196255
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.4563257694244385,
      "learning_rate": 3.1934486156791794e-05,
      "loss": 1.8647,
      "step": 196256
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.3995115756988525,
      "learning_rate": 3.193264968175705e-05,
      "loss": 2.0962,
      "step": 196257
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.3558480739593506,
      "learning_rate": 3.1930813256561006e-05,
      "loss": 1.7593,
      "step": 196258
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.4133033752441406,
      "learning_rate": 3.192897688120423e-05,
      "loss": 2.0715,
      "step": 196259
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.3735878467559814,
      "learning_rate": 3.1927140555686814e-05,
      "loss": 1.9837,
      "step": 196260
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.5294091701507568,
      "learning_rate": 3.192530428000923e-05,
      "loss": 2.2266,
      "step": 196261
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.3945729732513428,
      "learning_rate": 3.1923468054171754e-05,
      "loss": 1.9822,
      "step": 196262
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.4081010818481445,
      "learning_rate": 3.1921631878174806e-05,
      "loss": 2.2468,
      "step": 196263
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.4941896200180054,
      "learning_rate": 3.191979575201865e-05,
      "loss": 1.9682,
      "step": 196264
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.4430347681045532,
      "learning_rate": 3.1917959675703765e-05,
      "loss": 1.8564,
      "step": 196265
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.436295509338379,
      "learning_rate": 3.191612364923028e-05,
      "loss": 2.0469,
      "step": 196266
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.385662317276001,
      "learning_rate": 3.1914287672598714e-05,
      "loss": 1.9456,
      "step": 196267
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.4327725172042847,
      "learning_rate": 3.191245174580925e-05,
      "loss": 1.9344,
      "step": 196268
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.546433687210083,
      "learning_rate": 3.191061586886242e-05,
      "loss": 1.9159,
      "step": 196269
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.4235727787017822,
      "learning_rate": 3.1908780041758386e-05,
      "loss": 2.1416,
      "step": 196270
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.3820781707763672,
      "learning_rate": 3.190694426449771e-05,
      "loss": 2.132,
      "step": 196271
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.4569222927093506,
      "learning_rate": 3.1905108537080474e-05,
      "loss": 2.1055,
      "step": 196272
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.3826441764831543,
      "learning_rate": 3.190327285950719e-05,
      "loss": 2.0271,
      "step": 196273
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.4440968036651611,
      "learning_rate": 3.190143723177807e-05,
      "loss": 1.9958,
      "step": 196274
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.5471550226211548,
      "learning_rate": 3.189960165389361e-05,
      "loss": 2.0579,
      "step": 196275
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.4438101053237915,
      "learning_rate": 3.189776612585402e-05,
      "loss": 2.2182,
      "step": 196276
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.4573333263397217,
      "learning_rate": 3.189593064765982e-05,
      "loss": 1.8614,
      "step": 196277
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.5391203165054321,
      "learning_rate": 3.189409521931108e-05,
      "loss": 1.8712,
      "step": 196278
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.4949742555618286,
      "learning_rate": 3.189225984080837e-05,
      "loss": 1.8839,
      "step": 196279
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.3723413944244385,
      "learning_rate": 3.189042451215188e-05,
      "loss": 2.2208,
      "step": 196280
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.4947129487991333,
      "learning_rate": 3.188858923334209e-05,
      "loss": 2.0467,
      "step": 196281
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.3316000699996948,
      "learning_rate": 3.188675400437922e-05,
      "loss": 1.9868,
      "step": 196282
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.512181282043457,
      "learning_rate": 3.188491882526372e-05,
      "loss": 1.9452,
      "step": 196283
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.3226168155670166,
      "learning_rate": 3.188308369599587e-05,
      "loss": 2.0839,
      "step": 196284
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.4316391944885254,
      "learning_rate": 3.188124861657602e-05,
      "loss": 1.9719,
      "step": 196285
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.4104914665222168,
      "learning_rate": 3.187941358700442e-05,
      "loss": 2.0167,
      "step": 196286
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.4552428722381592,
      "learning_rate": 3.187757860728159e-05,
      "loss": 2.1155,
      "step": 196287
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.4288249015808105,
      "learning_rate": 3.18757436774077e-05,
      "loss": 2.0735,
      "step": 196288
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.553389072418213,
      "learning_rate": 3.187390879738325e-05,
      "loss": 1.9407,
      "step": 196289
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.4576648473739624,
      "learning_rate": 3.1872073967208486e-05,
      "loss": 2.0512,
      "step": 196290
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.4512758255004883,
      "learning_rate": 3.187023918688375e-05,
      "loss": 1.7853,
      "step": 196291
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.4357647895812988,
      "learning_rate": 3.186840445640937e-05,
      "loss": 2.0584,
      "step": 196292
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.4011105298995972,
      "learning_rate": 3.186656977578578e-05,
      "loss": 2.0391,
      "step": 196293
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.3409184217453003,
      "learning_rate": 3.1864735145013175e-05,
      "loss": 1.9771,
      "step": 196294
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.273403286933899,
      "learning_rate": 3.186290056409203e-05,
      "loss": 2.0145,
      "step": 196295
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.3234983682632446,
      "learning_rate": 3.186106603302264e-05,
      "loss": 2.007,
      "step": 196296
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.447516679763794,
      "learning_rate": 3.185923155180528e-05,
      "loss": 2.0518,
      "step": 196297
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.4436947107315063,
      "learning_rate": 3.185739712044043e-05,
      "loss": 1.9162,
      "step": 196298
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.5377697944641113,
      "learning_rate": 3.185556273892831e-05,
      "loss": 2.0652,
      "step": 196299
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.4734715223312378,
      "learning_rate": 3.185372840726928e-05,
      "loss": 1.8796,
      "step": 196300
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.6672749519348145,
      "learning_rate": 3.185189412546376e-05,
      "loss": 2.0023,
      "step": 196301
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.3338876962661743,
      "learning_rate": 3.1850059893512006e-05,
      "loss": 1.813,
      "step": 196302
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.456925392150879,
      "learning_rate": 3.1848225711414366e-05,
      "loss": 1.9456,
      "step": 196303
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.3697967529296875,
      "learning_rate": 3.1846391579171215e-05,
      "loss": 1.8749,
      "step": 196304
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.4045811891555786,
      "learning_rate": 3.184455749678292e-05,
      "loss": 1.9364,
      "step": 196305
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.3433901071548462,
      "learning_rate": 3.184272346424971e-05,
      "loss": 1.8383,
      "step": 196306
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.3325786590576172,
      "learning_rate": 3.184088948157208e-05,
      "loss": 1.9659,
      "step": 196307
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.5222491025924683,
      "learning_rate": 3.183905554875028e-05,
      "loss": 2.028,
      "step": 196308
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.418378472328186,
      "learning_rate": 3.18372216657846e-05,
      "loss": 2.2367,
      "step": 196309
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.468791127204895,
      "learning_rate": 3.183538783267546e-05,
      "loss": 1.9776,
      "step": 196310
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.452410101890564,
      "learning_rate": 3.183355404942318e-05,
      "loss": 1.8008,
      "step": 196311
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.4135555028915405,
      "learning_rate": 3.183172031602815e-05,
      "loss": 1.8332,
      "step": 196312
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.4704030752182007,
      "learning_rate": 3.182988663249064e-05,
      "loss": 2.068,
      "step": 196313
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.3902286291122437,
      "learning_rate": 3.182805299881105e-05,
      "loss": 2.0313,
      "step": 196314
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.649812936782837,
      "learning_rate": 3.1826219414989605e-05,
      "loss": 2.0024,
      "step": 196315
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.4663931131362915,
      "learning_rate": 3.182438588102682e-05,
      "loss": 2.3404,
      "step": 196316
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.5750644207000732,
      "learning_rate": 3.182255239692284e-05,
      "loss": 1.9309,
      "step": 196317
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.2822248935699463,
      "learning_rate": 3.1820718962678214e-05,
      "loss": 1.9134,
      "step": 196318
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.419788122177124,
      "learning_rate": 3.1818885578293076e-05,
      "loss": 2.1935,
      "step": 196319
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.427506446838379,
      "learning_rate": 3.181705224376802e-05,
      "loss": 1.9026,
      "step": 196320
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.6355557441711426,
      "learning_rate": 3.1815218959103107e-05,
      "loss": 2.1507,
      "step": 196321
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.4255176782608032,
      "learning_rate": 3.1813385724298844e-05,
      "loss": 2.1119,
      "step": 196322
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.384618878364563,
      "learning_rate": 3.18115525393555e-05,
      "loss": 2.0714,
      "step": 196323
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.449259877204895,
      "learning_rate": 3.1809719404273505e-05,
      "loss": 1.7165,
      "step": 196324
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.3981709480285645,
      "learning_rate": 3.1807886319053085e-05,
      "loss": 1.9867,
      "step": 196325
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.3650089502334595,
      "learning_rate": 3.1806053283694785e-05,
      "loss": 2.0027,
      "step": 196326
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.4433889389038086,
      "learning_rate": 3.1804220298198665e-05,
      "loss": 1.951,
      "step": 196327
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.2668129205703735,
      "learning_rate": 3.180238736256526e-05,
      "loss": 1.9593,
      "step": 196328
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.4787441492080688,
      "learning_rate": 3.180055447679477e-05,
      "loss": 2.062,
      "step": 196329
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.4504138231277466,
      "learning_rate": 3.179872164088773e-05,
      "loss": 2.0131,
      "step": 196330
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.442104458808899,
      "learning_rate": 3.179688885484426e-05,
      "loss": 1.9409,
      "step": 196331
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.429563283920288,
      "learning_rate": 3.1795056118664975e-05,
      "loss": 1.9697,
      "step": 196332
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.4850716590881348,
      "learning_rate": 3.1793223432349904e-05,
      "loss": 1.9507,
      "step": 196333
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.4398518800735474,
      "learning_rate": 3.179139079589961e-05,
      "loss": 2.0842,
      "step": 196334
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.606241226196289,
      "learning_rate": 3.178955820931429e-05,
      "loss": 1.9149,
      "step": 196335
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.2973130941390991,
      "learning_rate": 3.178772567259439e-05,
      "loss": 2.01,
      "step": 196336
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.3104192018508911,
      "learning_rate": 3.1785893185740194e-05,
      "loss": 2.0318,
      "step": 196337
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.5484243631362915,
      "learning_rate": 3.1784060748752185e-05,
      "loss": 2.0117,
      "step": 196338
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.5048786401748657,
      "learning_rate": 3.178222836163045e-05,
      "loss": 1.8895,
      "step": 196339
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.3587366342544556,
      "learning_rate": 3.178039602437549e-05,
      "loss": 2.0306,
      "step": 196340
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.4416353702545166,
      "learning_rate": 3.1778563736987606e-05,
      "loss": 1.9682,
      "step": 196341
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.457273006439209,
      "learning_rate": 3.177673149946717e-05,
      "loss": 2.0323,
      "step": 196342
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.3130162954330444,
      "learning_rate": 3.1774899311814475e-05,
      "loss": 2.0251,
      "step": 196343
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.5461797714233398,
      "learning_rate": 3.1773067174029985e-05,
      "loss": 2.1723,
      "step": 196344
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.3501547574996948,
      "learning_rate": 3.177123508611384e-05,
      "loss": 2.1059,
      "step": 196345
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.4451287984848022,
      "learning_rate": 3.176940304806654e-05,
      "loss": 2.043,
      "step": 196346
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.4569991827011108,
      "learning_rate": 3.176757105988832e-05,
      "loss": 2.136,
      "step": 196347
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.3384228944778442,
      "learning_rate": 3.176573912157964e-05,
      "loss": 1.8298,
      "step": 196348
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.363808035850525,
      "learning_rate": 3.17639072331407e-05,
      "loss": 1.8777,
      "step": 196349
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.444318413734436,
      "learning_rate": 3.1762075394572027e-05,
      "loss": 2.0086,
      "step": 196350
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.3601478338241577,
      "learning_rate": 3.1760243605873735e-05,
      "loss": 2.1153,
      "step": 196351
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.337941288948059,
      "learning_rate": 3.175841186704632e-05,
      "loss": 1.9705,
      "step": 196352
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.4698723554611206,
      "learning_rate": 3.175658017809004e-05,
      "loss": 2.1206,
      "step": 196353
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.3280524015426636,
      "learning_rate": 3.175474853900534e-05,
      "loss": 1.7944,
      "step": 196354
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.4077616930007935,
      "learning_rate": 3.1752916949792405e-05,
      "loss": 1.9654,
      "step": 196355
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.486279845237732,
      "learning_rate": 3.175108541045182e-05,
      "loss": 1.912,
      "step": 196356
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.467431902885437,
      "learning_rate": 3.174925392098363e-05,
      "loss": 2.017,
      "step": 196357
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.4162019491195679,
      "learning_rate": 3.174742248138839e-05,
      "loss": 2.2318,
      "step": 196358
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.6083955764770508,
      "learning_rate": 3.174559109166631e-05,
      "loss": 1.7962,
      "step": 196359
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.3632090091705322,
      "learning_rate": 3.1743759751817845e-05,
      "loss": 2.1638,
      "step": 196360
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.5160043239593506,
      "learning_rate": 3.1741928461843215e-05,
      "loss": 2.1033,
      "step": 196361
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.5018396377563477,
      "learning_rate": 3.174009722174292e-05,
      "loss": 2.1304,
      "step": 196362
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.6402004957199097,
      "learning_rate": 3.1738266031517135e-05,
      "loss": 2.0134,
      "step": 196363
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.4603098630905151,
      "learning_rate": 3.173643489116629e-05,
      "loss": 2.2268,
      "step": 196364
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.6507784128189087,
      "learning_rate": 3.173460380069067e-05,
      "loss": 2.0626,
      "step": 196365
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.4567346572875977,
      "learning_rate": 3.17327727600907e-05,
      "loss": 2.1252,
      "step": 196366
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.484597086906433,
      "learning_rate": 3.173094176936659e-05,
      "loss": 1.9731,
      "step": 196367
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.3817507028579712,
      "learning_rate": 3.172911082851892e-05,
      "loss": 1.9147,
      "step": 196368
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.4669358730316162,
      "learning_rate": 3.172727993754772e-05,
      "loss": 1.9288,
      "step": 196369
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.3702051639556885,
      "learning_rate": 3.172544909645356e-05,
      "loss": 1.9209,
      "step": 196370
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.3973426818847656,
      "learning_rate": 3.172361830523666e-05,
      "loss": 1.9319,
      "step": 196371
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.5429677963256836,
      "learning_rate": 3.1721787563897436e-05,
      "loss": 1.9522,
      "step": 196372
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.3865054845809937,
      "learning_rate": 3.171995687243615e-05,
      "loss": 1.9127,
      "step": 196373
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.6840391159057617,
      "learning_rate": 3.171812623085326e-05,
      "loss": 1.9334,
      "step": 196374
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.6274911165237427,
      "learning_rate": 3.1716295639149005e-05,
      "loss": 1.8713,
      "step": 196375
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.5195281505584717,
      "learning_rate": 3.171446509732378e-05,
      "loss": 1.9278,
      "step": 196376
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.4810872077941895,
      "learning_rate": 3.171263460537783e-05,
      "loss": 2.0916,
      "step": 196377
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.5124937295913696,
      "learning_rate": 3.171080416331161e-05,
      "loss": 1.9832,
      "step": 196378
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.3051916360855103,
      "learning_rate": 3.1708973771125355e-05,
      "loss": 2.016,
      "step": 196379
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.4375227689743042,
      "learning_rate": 3.170714342881957e-05,
      "loss": 2.1965,
      "step": 196380
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.3557144403457642,
      "learning_rate": 3.170531313639445e-05,
      "loss": 2.1291,
      "step": 196381
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.4927982091903687,
      "learning_rate": 3.170348289385033e-05,
      "loss": 1.8646,
      "step": 196382
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.5274769067764282,
      "learning_rate": 3.170165270118764e-05,
      "loss": 1.9735,
      "step": 196383
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.4005674123764038,
      "learning_rate": 3.169982255840672e-05,
      "loss": 1.8923,
      "step": 196384
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.3028168678283691,
      "learning_rate": 3.169799246550776e-05,
      "loss": 1.7027,
      "step": 196385
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.441931128501892,
      "learning_rate": 3.16961624224913e-05,
      "loss": 1.967,
      "step": 196386
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.529191493988037,
      "learning_rate": 3.169433242935757e-05,
      "loss": 2.0131,
      "step": 196387
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.3133443593978882,
      "learning_rate": 3.1692502486106876e-05,
      "loss": 1.6827,
      "step": 196388
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.5300917625427246,
      "learning_rate": 3.169067259273967e-05,
      "loss": 1.95,
      "step": 196389
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.450994849205017,
      "learning_rate": 3.1688842749256236e-05,
      "loss": 2.1357,
      "step": 196390
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.4352452754974365,
      "learning_rate": 3.1687012955656835e-05,
      "loss": 1.865,
      "step": 196391
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.499111533164978,
      "learning_rate": 3.1685183211941955e-05,
      "loss": 1.9064,
      "step": 196392
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.3873727321624756,
      "learning_rate": 3.1683353518111873e-05,
      "loss": 1.9634,
      "step": 196393
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.280469536781311,
      "learning_rate": 3.1681523874166856e-05,
      "loss": 1.9268,
      "step": 196394
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.5054707527160645,
      "learning_rate": 3.167969428010737e-05,
      "loss": 1.965,
      "step": 196395
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.3384259939193726,
      "learning_rate": 3.167786473593361e-05,
      "loss": 1.888,
      "step": 196396
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.4064199924468994,
      "learning_rate": 3.167603524164608e-05,
      "loss": 1.9368,
      "step": 196397
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.4003493785858154,
      "learning_rate": 3.1674205797245045e-05,
      "loss": 1.8502,
      "step": 196398
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.4980329275131226,
      "learning_rate": 3.1672376402730834e-05,
      "loss": 1.997,
      "step": 196399
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.372533917427063,
      "learning_rate": 3.1670547058103726e-05,
      "loss": 2.239,
      "step": 196400
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.4582277536392212,
      "learning_rate": 3.16687177633642e-05,
      "loss": 2.0132,
      "step": 196401
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.4339576959609985,
      "learning_rate": 3.166688851851248e-05,
      "loss": 1.9634,
      "step": 196402
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.3310385942459106,
      "learning_rate": 3.166505932354898e-05,
      "loss": 1.9003,
      "step": 196403
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.4395942687988281,
      "learning_rate": 3.1663230178473976e-05,
      "loss": 1.9998,
      "step": 196404
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.3744345903396606,
      "learning_rate": 3.166140108328793e-05,
      "loss": 2.0792,
      "step": 196405
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.3109378814697266,
      "learning_rate": 3.165957203799101e-05,
      "loss": 2.1318,
      "step": 196406
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.3137743473052979,
      "learning_rate": 3.1657743042583685e-05,
      "loss": 2.0212,
      "step": 196407
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.512803316116333,
      "learning_rate": 3.165591409706618e-05,
      "loss": 2.1742,
      "step": 196408
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.4884401559829712,
      "learning_rate": 3.1654085201438975e-05,
      "loss": 1.9715,
      "step": 196409
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.5158276557922363,
      "learning_rate": 3.1652256355702286e-05,
      "loss": 1.9334,
      "step": 196410
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.4071635007858276,
      "learning_rate": 3.165042755985663e-05,
      "loss": 1.9926,
      "step": 196411
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.6377934217453003,
      "learning_rate": 3.1648598813902096e-05,
      "loss": 1.9856,
      "step": 196412
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.5557689666748047,
      "learning_rate": 3.164677011783922e-05,
      "loss": 1.9939,
      "step": 196413
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.5260980129241943,
      "learning_rate": 3.1644941471668194e-05,
      "loss": 1.8477,
      "step": 196414
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.4004428386688232,
      "learning_rate": 3.16431128753895e-05,
      "loss": 1.8939,
      "step": 196415
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.3180055618286133,
      "learning_rate": 3.164128432900339e-05,
      "loss": 1.8276,
      "step": 196416
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.4782934188842773,
      "learning_rate": 3.163945583251034e-05,
      "loss": 2.0847,
      "step": 196417
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.395362377166748,
      "learning_rate": 3.1637627385910444e-05,
      "loss": 2.173,
      "step": 196418
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.4835166931152344,
      "learning_rate": 3.163579898920424e-05,
      "loss": 2.0123,
      "step": 196419
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.3517138957977295,
      "learning_rate": 3.1633970642391956e-05,
      "loss": 1.957,
      "step": 196420
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.4446132183074951,
      "learning_rate": 3.163214234547406e-05,
      "loss": 1.9681,
      "step": 196421
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.447317361831665,
      "learning_rate": 3.163031409845073e-05,
      "loss": 1.8707,
      "step": 196422
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.3729935884475708,
      "learning_rate": 3.162848590132251e-05,
      "loss": 2.0828,
      "step": 196423
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.4377260208129883,
      "learning_rate": 3.1626657754089515e-05,
      "loss": 2.1888,
      "step": 196424
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.3398449420928955,
      "learning_rate": 3.1624829656752236e-05,
      "loss": 2.0938,
      "step": 196425
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.4779082536697388,
      "learning_rate": 3.1623001609310916e-05,
      "loss": 2.0551,
      "step": 196426
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.4899382591247559,
      "learning_rate": 3.162117361176602e-05,
      "loss": 2.13,
      "step": 196427
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.3731967210769653,
      "learning_rate": 3.1619345664117734e-05,
      "loss": 2.1864,
      "step": 196428
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.6114990711212158,
      "learning_rate": 3.1617517766366605e-05,
      "loss": 2.3695,
      "step": 196429
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.3235419988632202,
      "learning_rate": 3.16156899185127e-05,
      "loss": 1.9943,
      "step": 196430
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.5085771083831787,
      "learning_rate": 3.161386212055661e-05,
      "loss": 1.9284,
      "step": 196431
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.3622362613677979,
      "learning_rate": 3.1612034372498474e-05,
      "loss": 1.9584,
      "step": 196432
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.3063201904296875,
      "learning_rate": 3.1610206674338787e-05,
      "loss": 1.872,
      "step": 196433
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.4004498720169067,
      "learning_rate": 3.160837902607779e-05,
      "loss": 2.0999,
      "step": 196434
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.3723453283309937,
      "learning_rate": 3.160655142771598e-05,
      "loss": 2.1688,
      "step": 196435
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.4641773700714111,
      "learning_rate": 3.1604723879253446e-05,
      "loss": 2.1243,
      "step": 196436
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.6831337213516235,
      "learning_rate": 3.1602896380690736e-05,
      "loss": 1.9384,
      "step": 196437
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.3819189071655273,
      "learning_rate": 3.160106893202804e-05,
      "loss": 1.9666,
      "step": 196438
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.401777982711792,
      "learning_rate": 3.159924153326583e-05,
      "loss": 2.098,
      "step": 196439
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.391263484954834,
      "learning_rate": 3.159741418440433e-05,
      "loss": 1.9032,
      "step": 196440
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.475584626197815,
      "learning_rate": 3.159558688544399e-05,
      "loss": 1.9242,
      "step": 196441
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.3376884460449219,
      "learning_rate": 3.15937596363851e-05,
      "loss": 1.8686,
      "step": 196442
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.4915140867233276,
      "learning_rate": 3.159193243722802e-05,
      "loss": 2.357,
      "step": 196443
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.419870376586914,
      "learning_rate": 3.159010528797299e-05,
      "loss": 1.923,
      "step": 196444
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.4026309251785278,
      "learning_rate": 3.158827818862047e-05,
      "loss": 2.0007,
      "step": 196445
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.5434027910232544,
      "learning_rate": 3.158645113917071e-05,
      "loss": 2.1522,
      "step": 196446
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.3744455575942993,
      "learning_rate": 3.158462413962416e-05,
      "loss": 1.9369,
      "step": 196447
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.3926723003387451,
      "learning_rate": 3.158279718998109e-05,
      "loss": 1.8975,
      "step": 196448
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.318373203277588,
      "learning_rate": 3.158097029024187e-05,
      "loss": 1.8721,
      "step": 196449
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.556502103805542,
      "learning_rate": 3.157914344040673e-05,
      "loss": 1.9304,
      "step": 196450
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.5129824876785278,
      "learning_rate": 3.1577316640476134e-05,
      "loss": 2.0255,
      "step": 196451
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.4593645334243774,
      "learning_rate": 3.157548989045036e-05,
      "loss": 1.9312,
      "step": 196452
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.4067360162734985,
      "learning_rate": 3.157366319032979e-05,
      "loss": 1.9592,
      "step": 196453
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.4972567558288574,
      "learning_rate": 3.157183654011477e-05,
      "loss": 2.0954,
      "step": 196454
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.733711838722229,
      "learning_rate": 3.1570009939805595e-05,
      "loss": 1.9026,
      "step": 196455
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.5396854877471924,
      "learning_rate": 3.15681833894026e-05,
      "loss": 1.8656,
      "step": 196456
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.3199032545089722,
      "learning_rate": 3.1566356888906187e-05,
      "loss": 1.9841,
      "step": 196457
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.5364291667938232,
      "learning_rate": 3.156453043831658e-05,
      "loss": 1.909,
      "step": 196458
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.536756992340088,
      "learning_rate": 3.1562704037634256e-05,
      "loss": 2.0762,
      "step": 196459
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.4410914182662964,
      "learning_rate": 3.1560877686859507e-05,
      "loss": 1.9674,
      "step": 196460
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.4723503589630127,
      "learning_rate": 3.155905138599264e-05,
      "loss": 1.7784,
      "step": 196461
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.517846941947937,
      "learning_rate": 3.1557225135033984e-05,
      "loss": 2.197,
      "step": 196462
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.4323410987854004,
      "learning_rate": 3.1555398933983936e-05,
      "loss": 1.9789,
      "step": 196463
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.6520307064056396,
      "learning_rate": 3.155357278284276e-05,
      "loss": 1.952,
      "step": 196464
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.6489802598953247,
      "learning_rate": 3.15517466816109e-05,
      "loss": 2.0515,
      "step": 196465
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.4516041278839111,
      "learning_rate": 3.1549920630288616e-05,
      "loss": 2.1157,
      "step": 196466
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.4373043775558472,
      "learning_rate": 3.1548094628876245e-05,
      "loss": 1.8415,
      "step": 196467
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.361613392829895,
      "learning_rate": 3.154626867737421e-05,
      "loss": 1.8402,
      "step": 196468
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.3099784851074219,
      "learning_rate": 3.1544442775782756e-05,
      "loss": 1.7543,
      "step": 196469
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.389448881149292,
      "learning_rate": 3.1542616924102235e-05,
      "loss": 1.8263,
      "step": 196470
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.3905361890792847,
      "learning_rate": 3.154079112233303e-05,
      "loss": 1.9747,
      "step": 196471
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.4219474792480469,
      "learning_rate": 3.15389653704755e-05,
      "loss": 2.0511,
      "step": 196472
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.4171366691589355,
      "learning_rate": 3.1537139668529874e-05,
      "loss": 1.8574,
      "step": 196473
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.4632128477096558,
      "learning_rate": 3.153531401649659e-05,
      "loss": 2.0392,
      "step": 196474
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.377943992614746,
      "learning_rate": 3.1533488414375985e-05,
      "loss": 2.0159,
      "step": 196475
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.3267921209335327,
      "learning_rate": 3.153166286216832e-05,
      "loss": 1.8116,
      "step": 196476
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.2939804792404175,
      "learning_rate": 3.152983735987402e-05,
      "loss": 1.9869,
      "step": 196477
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.4051038026809692,
      "learning_rate": 3.152801190749343e-05,
      "loss": 1.8896,
      "step": 196478
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.5912456512451172,
      "learning_rate": 3.152618650502675e-05,
      "loss": 2.154,
      "step": 196479
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.3579511642456055,
      "learning_rate": 3.152436115247451e-05,
      "loss": 1.6694,
      "step": 196480
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.4416097402572632,
      "learning_rate": 3.1522535849836875e-05,
      "loss": 1.9879,
      "step": 196481
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.4319466352462769,
      "learning_rate": 3.152071059711434e-05,
      "loss": 1.8444,
      "step": 196482
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.4897847175598145,
      "learning_rate": 3.151888539430718e-05,
      "loss": 2.1184,
      "step": 196483
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.305562138557434,
      "learning_rate": 3.151706024141573e-05,
      "loss": 1.8489,
      "step": 196484
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.5381786823272705,
      "learning_rate": 3.151523513844024e-05,
      "loss": 2.0973,
      "step": 196485
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.3482214212417603,
      "learning_rate": 3.151341008538123e-05,
      "loss": 2.0202,
      "step": 196486
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.438328504562378,
      "learning_rate": 3.151158508223885e-05,
      "loss": 1.8526,
      "step": 196487
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.5241693258285522,
      "learning_rate": 3.150976012901362e-05,
      "loss": 1.9851,
      "step": 196488
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.6422407627105713,
      "learning_rate": 3.1507935225705786e-05,
      "loss": 1.9918,
      "step": 196489
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.3210054636001587,
      "learning_rate": 3.1506110372315684e-05,
      "loss": 1.8996,
      "step": 196490
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.482548475265503,
      "learning_rate": 3.1504285568843626e-05,
      "loss": 2.0727,
      "step": 196491
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.5871663093566895,
      "learning_rate": 3.150246081529003e-05,
      "loss": 1.9383,
      "step": 196492
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.477229356765747,
      "learning_rate": 3.1500636111655144e-05,
      "loss": 1.9397,
      "step": 196493
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.44998037815094,
      "learning_rate": 3.149881145793942e-05,
      "loss": 1.9252,
      "step": 196494
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.3811644315719604,
      "learning_rate": 3.149698685414307e-05,
      "loss": 2.0117,
      "step": 196495
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.4187647104263306,
      "learning_rate": 3.149516230026662e-05,
      "loss": 1.9078,
      "step": 196496
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.572960615158081,
      "learning_rate": 3.149333779631017e-05,
      "loss": 2.0813,
      "step": 196497
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.8145076036453247,
      "learning_rate": 3.149151334227422e-05,
      "loss": 2.134,
      "step": 196498
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.538129210472107,
      "learning_rate": 3.148968893815904e-05,
      "loss": 2.0509,
      "step": 196499
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.4465529918670654,
      "learning_rate": 3.148786458396503e-05,
      "loss": 1.8659,
      "step": 196500
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.3701448440551758,
      "learning_rate": 3.148604027969245e-05,
      "loss": 2.0084,
      "step": 196501
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.7468783855438232,
      "learning_rate": 3.14842160253418e-05,
      "loss": 1.8653,
      "step": 196502
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.5532917976379395,
      "learning_rate": 3.148239182091315e-05,
      "loss": 2.0107,
      "step": 196503
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.4832676649093628,
      "learning_rate": 3.148056766640711e-05,
      "loss": 1.8199,
      "step": 196504
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.440522313117981,
      "learning_rate": 3.147874356182379e-05,
      "loss": 2.1385,
      "step": 196505
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.404768705368042,
      "learning_rate": 3.1476919507163744e-05,
      "loss": 1.7496,
      "step": 196506
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.3909128904342651,
      "learning_rate": 3.1475095502427126e-05,
      "loss": 1.9726,
      "step": 196507
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.3009804487228394,
      "learning_rate": 3.1473271547614406e-05,
      "loss": 1.9893,
      "step": 196508
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.3979525566101074,
      "learning_rate": 3.147144764272589e-05,
      "loss": 1.8356,
      "step": 196509
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.5748460292816162,
      "learning_rate": 3.14696237877619e-05,
      "loss": 2.1062,
      "step": 196510
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.4461169242858887,
      "learning_rate": 3.146779998272271e-05,
      "loss": 1.9801,
      "step": 196511
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.6245533227920532,
      "learning_rate": 3.146597622760878e-05,
      "loss": 1.9585,
      "step": 196512
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.4348224401474,
      "learning_rate": 3.1464152522420314e-05,
      "loss": 2.04,
      "step": 196513
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.6263965368270874,
      "learning_rate": 3.146232886715785e-05,
      "loss": 1.7565,
      "step": 196514
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.6214444637298584,
      "learning_rate": 3.1460505261821545e-05,
      "loss": 2.1507,
      "step": 196515
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.2905333042144775,
      "learning_rate": 3.1458681706411836e-05,
      "loss": 1.9121,
      "step": 196516
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.6237622499465942,
      "learning_rate": 3.145685820092896e-05,
      "loss": 2.1464,
      "step": 196517
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.6141531467437744,
      "learning_rate": 3.145503474537341e-05,
      "loss": 1.8858,
      "step": 196518
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.5051895380020142,
      "learning_rate": 3.1453211339745324e-05,
      "loss": 2.0119,
      "step": 196519
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.6513322591781616,
      "learning_rate": 3.145138798404526e-05,
      "loss": 1.9438,
      "step": 196520
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.3812626600265503,
      "learning_rate": 3.144956467827343e-05,
      "loss": 2.0204,
      "step": 196521
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.4457939863204956,
      "learning_rate": 3.14477414224302e-05,
      "loss": 1.8878,
      "step": 196522
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.434493064880371,
      "learning_rate": 3.144591821651585e-05,
      "loss": 1.9616,
      "step": 196523
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.6526073217391968,
      "learning_rate": 3.144409506053084e-05,
      "loss": 1.9534,
      "step": 196524
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.3480117321014404,
      "learning_rate": 3.144227195447535e-05,
      "loss": 1.9958,
      "step": 196525
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.3431099653244019,
      "learning_rate": 3.144044889834989e-05,
      "loss": 1.8964,
      "step": 196526
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.5094445943832397,
      "learning_rate": 3.143862589215472e-05,
      "loss": 1.885,
      "step": 196527
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.5701037645339966,
      "learning_rate": 3.143680293589018e-05,
      "loss": 2.2905,
      "step": 196528
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.5185894966125488,
      "learning_rate": 3.143498002955657e-05,
      "loss": 2.1482,
      "step": 196529
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.4003562927246094,
      "learning_rate": 3.143315717315431e-05,
      "loss": 1.8404,
      "step": 196530
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.4250651597976685,
      "learning_rate": 3.143133436668362e-05,
      "loss": 2.1732,
      "step": 196531
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.3818740844726562,
      "learning_rate": 3.1429511610144976e-05,
      "loss": 1.997,
      "step": 196532
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.6147557497024536,
      "learning_rate": 3.142768890353866e-05,
      "loss": 1.999,
      "step": 196533
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.2859679460525513,
      "learning_rate": 3.142586624686504e-05,
      "loss": 1.9546,
      "step": 196534
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.4144586324691772,
      "learning_rate": 3.1424043640124306e-05,
      "loss": 1.8649,
      "step": 196535
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.4755648374557495,
      "learning_rate": 3.1422221083317e-05,
      "loss": 2.0053,
      "step": 196536
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.4186238050460815,
      "learning_rate": 3.1420398576443315e-05,
      "loss": 2.0235,
      "step": 196537
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.4706140756607056,
      "learning_rate": 3.141857611950372e-05,
      "loss": 2.1124,
      "step": 196538
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.6547603607177734,
      "learning_rate": 3.141675371249845e-05,
      "loss": 2.0497,
      "step": 196539
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.4849309921264648,
      "learning_rate": 3.141493135542791e-05,
      "loss": 1.8823,
      "step": 196540
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.4284439086914062,
      "learning_rate": 3.141310904829232e-05,
      "loss": 1.9618,
      "step": 196541
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.4054028987884521,
      "learning_rate": 3.141128679109215e-05,
      "loss": 1.9488,
      "step": 196542
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.543096899986267,
      "learning_rate": 3.140946458382767e-05,
      "loss": 2.3317,
      "step": 196543
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.4594331979751587,
      "learning_rate": 3.140764242649928e-05,
      "loss": 1.8559,
      "step": 196544
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.4686009883880615,
      "learning_rate": 3.140582031910728e-05,
      "loss": 2.2101,
      "step": 196545
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.592869758605957,
      "learning_rate": 3.1403998261651995e-05,
      "loss": 2.0226,
      "step": 196546
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.5046268701553345,
      "learning_rate": 3.140217625413374e-05,
      "loss": 2.3076,
      "step": 196547
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.3912913799285889,
      "learning_rate": 3.140035429655293e-05,
      "loss": 2.0375,
      "step": 196548
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.433601975440979,
      "learning_rate": 3.139853238890981e-05,
      "loss": 2.1366,
      "step": 196549
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.3352679014205933,
      "learning_rate": 3.139671053120485e-05,
      "loss": 1.9847,
      "step": 196550
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.3386107683181763,
      "learning_rate": 3.1394888723438305e-05,
      "loss": 1.8592,
      "step": 196551
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.3381178379058838,
      "learning_rate": 3.1393066965610516e-05,
      "loss": 1.7845,
      "step": 196552
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.4839065074920654,
      "learning_rate": 3.139124525772178e-05,
      "loss": 1.8467,
      "step": 196553
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.5241637229919434,
      "learning_rate": 3.138942359977252e-05,
      "loss": 1.9866,
      "step": 196554
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.4080849885940552,
      "learning_rate": 3.138760199176299e-05,
      "loss": 1.9024,
      "step": 196555
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.320360541343689,
      "learning_rate": 3.138578043369361e-05,
      "loss": 1.9916,
      "step": 196556
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.4876614809036255,
      "learning_rate": 3.1383958925564714e-05,
      "loss": 1.9881,
      "step": 196557
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.3731459379196167,
      "learning_rate": 3.138213746737653e-05,
      "loss": 1.9958,
      "step": 196558
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.42916738986969,
      "learning_rate": 3.138031605912954e-05,
      "loss": 1.9119,
      "step": 196559
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.4143078327178955,
      "learning_rate": 3.137849470082403e-05,
      "loss": 2.0607,
      "step": 196560
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.5017613172531128,
      "learning_rate": 3.1376673392460275e-05,
      "loss": 2.1491,
      "step": 196561
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.4641661643981934,
      "learning_rate": 3.137485213403873e-05,
      "loss": 1.8971,
      "step": 196562
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.9904279708862305,
      "learning_rate": 3.137303092555964e-05,
      "loss": 2.0325,
      "step": 196563
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.5048305988311768,
      "learning_rate": 3.137120976702333e-05,
      "loss": 2.0861,
      "step": 196564
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.5064871311187744,
      "learning_rate": 3.136938865843027e-05,
      "loss": 2.0206,
      "step": 196565
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.4951761960983276,
      "learning_rate": 3.136756759978063e-05,
      "loss": 2.1741,
      "step": 196566
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.334470272064209,
      "learning_rate": 3.136574659107487e-05,
      "loss": 2.19,
      "step": 196567
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.5955320596694946,
      "learning_rate": 3.136392563231329e-05,
      "loss": 2.0079,
      "step": 196568
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.4817959070205688,
      "learning_rate": 3.136210472349626e-05,
      "loss": 1.8809,
      "step": 196569
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.4079627990722656,
      "learning_rate": 3.136028386462401e-05,
      "loss": 1.9647,
      "step": 196570
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.3502252101898193,
      "learning_rate": 3.135846305569701e-05,
      "loss": 1.9792,
      "step": 196571
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.3919693231582642,
      "learning_rate": 3.13566422967155e-05,
      "loss": 1.8894,
      "step": 196572
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.5131235122680664,
      "learning_rate": 3.135482158767989e-05,
      "loss": 1.9899,
      "step": 196573
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.4116977453231812,
      "learning_rate": 3.13530009285905e-05,
      "loss": 1.8619,
      "step": 196574
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.5122630596160889,
      "learning_rate": 3.1351180319447625e-05,
      "loss": 1.7883,
      "step": 196575
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.5028464794158936,
      "learning_rate": 3.134935976025166e-05,
      "loss": 2.1339,
      "step": 196576
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.5310969352722168,
      "learning_rate": 3.134753925100294e-05,
      "loss": 2.135,
      "step": 196577
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.5335273742675781,
      "learning_rate": 3.13457187917017e-05,
      "loss": 2.0626,
      "step": 196578
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.582219123840332,
      "learning_rate": 3.1343898382348475e-05,
      "loss": 2.1448,
      "step": 196579
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.4824289083480835,
      "learning_rate": 3.13420780229434e-05,
      "loss": 2.0572,
      "step": 196580
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.4726059436798096,
      "learning_rate": 3.134025771348696e-05,
      "loss": 2.1809,
      "step": 196581
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.5155929327011108,
      "learning_rate": 3.133843745397944e-05,
      "loss": 1.8786,
      "step": 196582
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.4635698795318604,
      "learning_rate": 3.1336617244421156e-05,
      "loss": 1.9664,
      "step": 196583
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.3292957544326782,
      "learning_rate": 3.133479708481242e-05,
      "loss": 2.0771,
      "step": 196584
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.4798787832260132,
      "learning_rate": 3.13329769751537e-05,
      "loss": 2.0154,
      "step": 196585
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.4511083364486694,
      "learning_rate": 3.133115691544516e-05,
      "loss": 1.9173,
      "step": 196586
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.4635008573532104,
      "learning_rate": 3.132933690568732e-05,
      "loss": 1.8023,
      "step": 196587
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.5034818649291992,
      "learning_rate": 3.1327516945880396e-05,
      "loss": 2.1336,
      "step": 196588
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.427255392074585,
      "learning_rate": 3.1325697036024774e-05,
      "loss": 1.901,
      "step": 196589
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.3392301797866821,
      "learning_rate": 3.1323877176120705e-05,
      "loss": 1.8679,
      "step": 196590
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.3306924104690552,
      "learning_rate": 3.1322057366168676e-05,
      "loss": 1.8928,
      "step": 196591
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.5092840194702148,
      "learning_rate": 3.132023760616888e-05,
      "loss": 2.0177,
      "step": 196592
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.363205075263977,
      "learning_rate": 3.1318417896121774e-05,
      "loss": 1.9069,
      "step": 196593
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.4853595495224,
      "learning_rate": 3.131659823602767e-05,
      "loss": 1.9874,
      "step": 196594
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.3504711389541626,
      "learning_rate": 3.131477862588687e-05,
      "loss": 1.8833,
      "step": 196595
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.5331218242645264,
      "learning_rate": 3.131295906569965e-05,
      "loss": 2.029,
      "step": 196596
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.5077428817749023,
      "learning_rate": 3.13111395554665e-05,
      "loss": 2.0587,
      "step": 196597
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.4093613624572754,
      "learning_rate": 3.130932009518763e-05,
      "loss": 1.8034,
      "step": 196598
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.5747113227844238,
      "learning_rate": 3.130750068486346e-05,
      "loss": 2.1675,
      "step": 196599
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.4461314678192139,
      "learning_rate": 3.130568132449433e-05,
      "loss": 1.8997,
      "step": 196600
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.3420549631118774,
      "learning_rate": 3.130386201408052e-05,
      "loss": 1.9501,
      "step": 196601
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.7084693908691406,
      "learning_rate": 3.130204275362234e-05,
      "loss": 2.0448,
      "step": 196602
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.374502420425415,
      "learning_rate": 3.130022354312026e-05,
      "loss": 2.1395,
      "step": 196603
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.4656254053115845,
      "learning_rate": 3.129840438257446e-05,
      "loss": 1.9667,
      "step": 196604
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.3927046060562134,
      "learning_rate": 3.129658527198543e-05,
      "loss": 1.8989,
      "step": 196605
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.5427250862121582,
      "learning_rate": 3.129476621135344e-05,
      "loss": 1.7576,
      "step": 196606
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.4161213636398315,
      "learning_rate": 3.129294720067878e-05,
      "loss": 2.2175,
      "step": 196607
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.5400341749191284,
      "learning_rate": 3.129112823996184e-05,
      "loss": 1.9372,
      "step": 196608
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.4328173398971558,
      "learning_rate": 3.128930932920296e-05,
      "loss": 1.9443,
      "step": 196609
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.4932905435562134,
      "learning_rate": 3.128749046840245e-05,
      "loss": 2.0623,
      "step": 196610
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.6941725015640259,
      "learning_rate": 3.1285671657560715e-05,
      "loss": 2.2687,
      "step": 196611
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.6080687046051025,
      "learning_rate": 3.128385289667802e-05,
      "loss": 2.0108,
      "step": 196612
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.5359090566635132,
      "learning_rate": 3.1282034185754755e-05,
      "loss": 2.1991,
      "step": 196613
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.3974478244781494,
      "learning_rate": 3.128021552479116e-05,
      "loss": 1.939,
      "step": 196614
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.588989019393921,
      "learning_rate": 3.127839691378774e-05,
      "loss": 1.8272,
      "step": 196615
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.5432887077331543,
      "learning_rate": 3.127657835274465e-05,
      "loss": 2.0562,
      "step": 196616
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.5196360349655151,
      "learning_rate": 3.127475984166237e-05,
      "loss": 2.0595,
      "step": 196617
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.3381839990615845,
      "learning_rate": 3.127294138054118e-05,
      "loss": 1.6878,
      "step": 196618
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.4933987855911255,
      "learning_rate": 3.1271122969381435e-05,
      "loss": 2.0777,
      "step": 196619
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.395595669746399,
      "learning_rate": 3.126930460818342e-05,
      "loss": 1.9462,
      "step": 196620
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.3913942575454712,
      "learning_rate": 3.126748629694754e-05,
      "loss": 1.9475,
      "step": 196621
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.4114985466003418,
      "learning_rate": 3.126566803567406e-05,
      "loss": 1.9018,
      "step": 196622
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.4512243270874023,
      "learning_rate": 3.126384982436345e-05,
      "loss": 2.0958,
      "step": 196623
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.3613276481628418,
      "learning_rate": 3.126203166301594e-05,
      "loss": 2.0509,
      "step": 196624
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.3891865015029907,
      "learning_rate": 3.1260213551631865e-05,
      "loss": 1.9103,
      "step": 196625
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.448270559310913,
      "learning_rate": 3.125839549021158e-05,
      "loss": 2.0876,
      "step": 196626
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.6786895990371704,
      "learning_rate": 3.1256577478755475e-05,
      "loss": 2.0681,
      "step": 196627
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.501920461654663,
      "learning_rate": 3.125475951726376e-05,
      "loss": 2.0746,
      "step": 196628
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.4130909442901611,
      "learning_rate": 3.125294160573695e-05,
      "loss": 2.0177,
      "step": 196629
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.480739712715149,
      "learning_rate": 3.125112374417527e-05,
      "loss": 2.0239,
      "step": 196630
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.3859776258468628,
      "learning_rate": 3.124930593257908e-05,
      "loss": 1.8592,
      "step": 196631
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.3015581369400024,
      "learning_rate": 3.1247488170948666e-05,
      "loss": 2.1475,
      "step": 196632
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.5072067975997925,
      "learning_rate": 3.1245670459284475e-05,
      "loss": 2.0515,
      "step": 196633
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.561638593673706,
      "learning_rate": 3.1243852797586724e-05,
      "loss": 2.0626,
      "step": 196634
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.4982774257659912,
      "learning_rate": 3.124203518585586e-05,
      "loss": 2.0824,
      "step": 196635
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.6331453323364258,
      "learning_rate": 3.12402176240922e-05,
      "loss": 2.0031,
      "step": 196636
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.3412857055664062,
      "learning_rate": 3.1238400112296066e-05,
      "loss": 1.7988,
      "step": 196637
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.5422407388687134,
      "learning_rate": 3.12365826504677e-05,
      "loss": 1.7846,
      "step": 196638
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.3255174160003662,
      "learning_rate": 3.1234765238607594e-05,
      "loss": 2.0985,
      "step": 196639
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.3701874017715454,
      "learning_rate": 3.1232947876715955e-05,
      "loss": 1.8883,
      "step": 196640
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.437811255455017,
      "learning_rate": 3.123113056479327e-05,
      "loss": 2.1239,
      "step": 196641
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.4859015941619873,
      "learning_rate": 3.122931330283969e-05,
      "loss": 2.0256,
      "step": 196642
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.4406687021255493,
      "learning_rate": 3.122749609085573e-05,
      "loss": 2.0529,
      "step": 196643
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.4947007894515991,
      "learning_rate": 3.122567892884167e-05,
      "loss": 2.2939,
      "step": 196644
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.5450040102005005,
      "learning_rate": 3.12238618167978e-05,
      "loss": 2.0254,
      "step": 196645
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.4479219913482666,
      "learning_rate": 3.122204475472446e-05,
      "loss": 1.9991,
      "step": 196646
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.3986589908599854,
      "learning_rate": 3.122022774262205e-05,
      "loss": 2.0442,
      "step": 196647
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.3598731756210327,
      "learning_rate": 3.121841078049083e-05,
      "loss": 1.9071,
      "step": 196648
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.3899176120758057,
      "learning_rate": 3.121659386833124e-05,
      "loss": 1.9584,
      "step": 196649
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.4224121570587158,
      "learning_rate": 3.1214777006143545e-05,
      "loss": 1.9313,
      "step": 196650
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.2915931940078735,
      "learning_rate": 3.12129601939281e-05,
      "loss": 1.6539,
      "step": 196651
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.392616868019104,
      "learning_rate": 3.121114343168519e-05,
      "loss": 2.2274,
      "step": 196652
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.3845511674880981,
      "learning_rate": 3.1209326719415274e-05,
      "loss": 1.9179,
      "step": 196653
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.3834822177886963,
      "learning_rate": 3.120751005711851e-05,
      "loss": 2.0313,
      "step": 196654
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.4322377443313599,
      "learning_rate": 3.1205693444795445e-05,
      "loss": 2.1819,
      "step": 196655
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.402671456336975,
      "learning_rate": 3.12038768824463e-05,
      "loss": 1.912,
      "step": 196656
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.4278467893600464,
      "learning_rate": 3.120206037007138e-05,
      "loss": 2.0722,
      "step": 196657
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.9076036214828491,
      "learning_rate": 3.120024390767112e-05,
      "loss": 1.9175,
      "step": 196658
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.7258371114730835,
      "learning_rate": 3.1198427495245815e-05,
      "loss": 2.0297,
      "step": 196659
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.476722240447998,
      "learning_rate": 3.1196611132795734e-05,
      "loss": 1.9244,
      "step": 196660
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.419150948524475,
      "learning_rate": 3.119479482032134e-05,
      "loss": 1.7419,
      "step": 196661
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.746841549873352,
      "learning_rate": 3.119297855782291e-05,
      "loss": 1.8301,
      "step": 196662
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.9099847078323364,
      "learning_rate": 3.11911623453007e-05,
      "loss": 2.0758,
      "step": 196663
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.4589135646820068,
      "learning_rate": 3.1189346182755214e-05,
      "loss": 1.6721,
      "step": 196664
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.3456615209579468,
      "learning_rate": 3.1187530070186615e-05,
      "loss": 1.9393,
      "step": 196665
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.4030879735946655,
      "learning_rate": 3.1185714007595406e-05,
      "loss": 2.1401,
      "step": 196666
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.4294875860214233,
      "learning_rate": 3.118389799498186e-05,
      "loss": 1.9268,
      "step": 196667
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.480912685394287,
      "learning_rate": 3.1182082032346255e-05,
      "loss": 1.942,
      "step": 196668
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.396471619606018,
      "learning_rate": 3.1180266119688944e-05,
      "loss": 2.0961,
      "step": 196669
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.4857021570205688,
      "learning_rate": 3.117845025701039e-05,
      "loss": 2.0051,
      "step": 196670
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.4270225763320923,
      "learning_rate": 3.117663444431072e-05,
      "loss": 1.8837,
      "step": 196671
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.467968463897705,
      "learning_rate": 3.117481868159047e-05,
      "loss": 2.1686,
      "step": 196672
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.5415669679641724,
      "learning_rate": 3.117300296884991e-05,
      "loss": 2.0604,
      "step": 196673
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.4744646549224854,
      "learning_rate": 3.1171187306089343e-05,
      "loss": 2.0519,
      "step": 196674
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.2875399589538574,
      "learning_rate": 3.1169371693309086e-05,
      "loss": 2.0622,
      "step": 196675
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.6515129804611206,
      "learning_rate": 3.116755613050955e-05,
      "loss": 2.1165,
      "step": 196676
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.6700289249420166,
      "learning_rate": 3.1165740617691006e-05,
      "loss": 1.9696,
      "step": 196677
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.4789090156555176,
      "learning_rate": 3.1163925154853876e-05,
      "loss": 1.9476,
      "step": 196678
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.5938098430633545,
      "learning_rate": 3.1162109741998465e-05,
      "loss": 2.0733,
      "step": 196679
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.4396125078201294,
      "learning_rate": 3.1160294379125073e-05,
      "loss": 2.0808,
      "step": 196680
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.4203085899353027,
      "learning_rate": 3.1158479066234e-05,
      "loss": 2.1022,
      "step": 196681
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.3492861986160278,
      "learning_rate": 3.115666380332568e-05,
      "loss": 1.9903,
      "step": 196682
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.417336106300354,
      "learning_rate": 3.1154848590400375e-05,
      "loss": 1.9962,
      "step": 196683
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.5993475914001465,
      "learning_rate": 3.115303342745852e-05,
      "loss": 1.9725,
      "step": 196684
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.4078336954116821,
      "learning_rate": 3.115121831450038e-05,
      "loss": 2.0964,
      "step": 196685
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.4816453456878662,
      "learning_rate": 3.11494032515263e-05,
      "loss": 2.1599,
      "step": 196686
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.4072927236557007,
      "learning_rate": 3.114758823853656e-05,
      "loss": 1.8077,
      "step": 196687
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.3490986824035645,
      "learning_rate": 3.114577327553164e-05,
      "loss": 1.9462,
      "step": 196688
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.3858586549758911,
      "learning_rate": 3.1143958362511735e-05,
      "loss": 2.0836,
      "step": 196689
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.3890843391418457,
      "learning_rate": 3.1142143499477276e-05,
      "loss": 1.9401,
      "step": 196690
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.4632680416107178,
      "learning_rate": 3.1140328686428576e-05,
      "loss": 2.1728,
      "step": 196691
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.612346887588501,
      "learning_rate": 3.113851392336598e-05,
      "loss": 1.8265,
      "step": 196692
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.5187501907348633,
      "learning_rate": 3.113669921028974e-05,
      "loss": 2.0582,
      "step": 196693
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.3192448616027832,
      "learning_rate": 3.1134884547200315e-05,
      "loss": 2.0513,
      "step": 196694
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.5076617002487183,
      "learning_rate": 3.113306993409793e-05,
      "loss": 1.9674,
      "step": 196695
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.4662724733352661,
      "learning_rate": 3.1131255370983065e-05,
      "loss": 1.9792,
      "step": 196696
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.5323541164398193,
      "learning_rate": 3.1129440857855945e-05,
      "loss": 1.8001,
      "step": 196697
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.410720705986023,
      "learning_rate": 3.1127626394716975e-05,
      "loss": 2.0706,
      "step": 196698
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.3976378440856934,
      "learning_rate": 3.112581198156635e-05,
      "loss": 2.0217,
      "step": 196699
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.5315946340560913,
      "learning_rate": 3.1123997618404607e-05,
      "loss": 1.9755,
      "step": 196700
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.4427331686019897,
      "learning_rate": 3.1122183305231906e-05,
      "loss": 2.088,
      "step": 196701
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.5686991214752197,
      "learning_rate": 3.112036904204875e-05,
      "loss": 2.0063,
      "step": 196702
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.5843359231948853,
      "learning_rate": 3.111855482885538e-05,
      "loss": 1.9128,
      "step": 196703
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.4929980039596558,
      "learning_rate": 3.1116740665652154e-05,
      "loss": 1.9315,
      "step": 196704
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.4893771409988403,
      "learning_rate": 3.1114926552439334e-05,
      "loss": 2.0916,
      "step": 196705
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.451925277709961,
      "learning_rate": 3.111311248921736e-05,
      "loss": 2.036,
      "step": 196706
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.4259262084960938,
      "learning_rate": 3.11112984759865e-05,
      "loss": 1.827,
      "step": 196707
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.3838099241256714,
      "learning_rate": 3.110948451274719e-05,
      "loss": 1.9695,
      "step": 196708
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.3701709508895874,
      "learning_rate": 3.110767059949965e-05,
      "loss": 2.1556,
      "step": 196709
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.3680872917175293,
      "learning_rate": 3.110585673624435e-05,
      "loss": 2.0735,
      "step": 196710
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.4286479949951172,
      "learning_rate": 3.1104042922981466e-05,
      "loss": 1.7641,
      "step": 196711
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.465555191040039,
      "learning_rate": 3.110222915971146e-05,
      "loss": 1.8689,
      "step": 196712
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.4549938440322876,
      "learning_rate": 3.1100415446434555e-05,
      "loss": 2.1155,
      "step": 196713
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.2702300548553467,
      "learning_rate": 3.109860178315123e-05,
      "loss": 2.1696,
      "step": 196714
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.3508304357528687,
      "learning_rate": 3.109678816986172e-05,
      "loss": 2.1646,
      "step": 196715
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.4675711393356323,
      "learning_rate": 3.109497460656648e-05,
      "loss": 1.8619,
      "step": 196716
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.3681310415267944,
      "learning_rate": 3.1093161093265614e-05,
      "loss": 1.9159,
      "step": 196717
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.5108522176742554,
      "learning_rate": 3.109134762995973e-05,
      "loss": 1.9002,
      "step": 196718
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.4523106813430786,
      "learning_rate": 3.1089534216648945e-05,
      "loss": 1.9618,
      "step": 196719
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.4458343982696533,
      "learning_rate": 3.108772085333374e-05,
      "loss": 2.0578,
      "step": 196720
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.2936664819717407,
      "learning_rate": 3.1085907540014376e-05,
      "loss": 1.8303,
      "step": 196721
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.3964836597442627,
      "learning_rate": 3.108409427669132e-05,
      "loss": 2.0256,
      "step": 196722
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.3990812301635742,
      "learning_rate": 3.1082281063364665e-05,
      "loss": 2.0539,
      "step": 196723
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.6851543188095093,
      "learning_rate": 3.108046790003499e-05,
      "loss": 1.9723,
      "step": 196724
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.497926115989685,
      "learning_rate": 3.107865478670245e-05,
      "loss": 2.1219,
      "step": 196725
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.458017110824585,
      "learning_rate": 3.107684172336752e-05,
      "loss": 2.0207,
      "step": 196726
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.4698790311813354,
      "learning_rate": 3.107502871003042e-05,
      "loss": 1.7778,
      "step": 196727
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.382325291633606,
      "learning_rate": 3.1073215746691636e-05,
      "loss": 1.9743,
      "step": 196728
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.4833863973617554,
      "learning_rate": 3.107140283335139e-05,
      "loss": 1.9601,
      "step": 196729
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.3553529977798462,
      "learning_rate": 3.1069589970010044e-05,
      "loss": 1.7507,
      "step": 196730
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.419842004776001,
      "learning_rate": 3.10677771566679e-05,
      "loss": 2.0262,
      "step": 196731
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.3636823892593384,
      "learning_rate": 3.106596439332537e-05,
      "loss": 2.0188,
      "step": 196732
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.4361342191696167,
      "learning_rate": 3.106415167998274e-05,
      "loss": 1.986,
      "step": 196733
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.4191064834594727,
      "learning_rate": 3.106233901664038e-05,
      "loss": 2.2516,
      "step": 196734
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.4540284872055054,
      "learning_rate": 3.106052640329859e-05,
      "loss": 2.1643,
      "step": 196735
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.3615294694900513,
      "learning_rate": 3.105871383995777e-05,
      "loss": 1.9394,
      "step": 196736
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.3140755891799927,
      "learning_rate": 3.105690132661812e-05,
      "loss": 1.7993,
      "step": 196737
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.5088194608688354,
      "learning_rate": 3.1055088863280144e-05,
      "loss": 2.161,
      "step": 196738
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.4139806032180786,
      "learning_rate": 3.1053276449944034e-05,
      "loss": 1.8168,
      "step": 196739
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.7293537855148315,
      "learning_rate": 3.105146408661029e-05,
      "loss": 2.1632,
      "step": 196740
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.4433966875076294,
      "learning_rate": 3.104965177327911e-05,
      "loss": 1.9954,
      "step": 196741
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.3881711959838867,
      "learning_rate": 3.1047839509950844e-05,
      "loss": 1.779,
      "step": 196742
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.3603118658065796,
      "learning_rate": 3.104602729662591e-05,
      "loss": 2.055,
      "step": 196743
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.4641242027282715,
      "learning_rate": 3.1044215133304606e-05,
      "loss": 2.0631,
      "step": 196744
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.3927514553070068,
      "learning_rate": 3.104240301998717e-05,
      "loss": 1.8356,
      "step": 196745
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.4330637454986572,
      "learning_rate": 3.104059095667414e-05,
      "loss": 2.0639,
      "step": 196746
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.3823574781417847,
      "learning_rate": 3.1038778943365704e-05,
      "loss": 1.8449,
      "step": 196747
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.5092915296554565,
      "learning_rate": 3.103696698006217e-05,
      "loss": 1.9087,
      "step": 196748
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.3879709243774414,
      "learning_rate": 3.1035155066764005e-05,
      "loss": 1.9164,
      "step": 196749
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.4317688941955566,
      "learning_rate": 3.103334320347151e-05,
      "loss": 1.9487,
      "step": 196750
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.5031553506851196,
      "learning_rate": 3.103153139018491e-05,
      "loss": 1.8691,
      "step": 196751
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.3786579370498657,
      "learning_rate": 3.10297196269047e-05,
      "loss": 2.2663,
      "step": 196752
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.5180829763412476,
      "learning_rate": 3.102790791363114e-05,
      "loss": 1.9643,
      "step": 196753
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.358006238937378,
      "learning_rate": 3.1026096250364494e-05,
      "loss": 2.0147,
      "step": 196754
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.3708889484405518,
      "learning_rate": 3.1024284637105227e-05,
      "loss": 1.9472,
      "step": 196755
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.454047679901123,
      "learning_rate": 3.1022473073853585e-05,
      "loss": 2.0283,
      "step": 196756
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.5438134670257568,
      "learning_rate": 3.102066156061e-05,
      "loss": 2.0541,
      "step": 196757
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.561651587486267,
      "learning_rate": 3.101885009737472e-05,
      "loss": 2.1118,
      "step": 196758
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.4482676982879639,
      "learning_rate": 3.1017038684148175e-05,
      "loss": 1.8063,
      "step": 196759
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.336905598640442,
      "learning_rate": 3.1015227320930526e-05,
      "loss": 2.0315,
      "step": 196760
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.320756196975708,
      "learning_rate": 3.1013416007722305e-05,
      "loss": 2.0162,
      "step": 196761
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.3635669946670532,
      "learning_rate": 3.101160474452371e-05,
      "loss": 2.153,
      "step": 196762
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.633915901184082,
      "learning_rate": 3.100979353133518e-05,
      "loss": 2.2001,
      "step": 196763
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.3731801509857178,
      "learning_rate": 3.100798236815701e-05,
      "loss": 2.0542,
      "step": 196764
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.4389370679855347,
      "learning_rate": 3.100617125498954e-05,
      "loss": 1.8459,
      "step": 196765
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.4561489820480347,
      "learning_rate": 3.100436019183307e-05,
      "loss": 2.0114,
      "step": 196766
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.474108099937439,
      "learning_rate": 3.1002549178687984e-05,
      "loss": 1.9907,
      "step": 196767
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.4986971616744995,
      "learning_rate": 3.1000738215554564e-05,
      "loss": 1.8248,
      "step": 196768
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.6366592645645142,
      "learning_rate": 3.099892730243324e-05,
      "loss": 1.987,
      "step": 196769
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.306174874305725,
      "learning_rate": 3.0997116439324276e-05,
      "loss": 1.7687,
      "step": 196770
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.5945484638214111,
      "learning_rate": 3.099530562622804e-05,
      "loss": 2.0078,
      "step": 196771
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.4010037183761597,
      "learning_rate": 3.09934948631448e-05,
      "loss": 1.9361,
      "step": 196772
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.4243985414505005,
      "learning_rate": 3.0991684150075014e-05,
      "loss": 1.9802,
      "step": 196773
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.4274715185165405,
      "learning_rate": 3.098987348701886e-05,
      "loss": 2.156,
      "step": 196774
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.4480273723602295,
      "learning_rate": 3.098806287397686e-05,
      "loss": 2.183,
      "step": 196775
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.424311637878418,
      "learning_rate": 3.098625231094919e-05,
      "loss": 1.8693,
      "step": 196776
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.4257192611694336,
      "learning_rate": 3.098444179793634e-05,
      "loss": 1.9571,
      "step": 196777
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.328366756439209,
      "learning_rate": 3.098263133493849e-05,
      "loss": 2.0869,
      "step": 196778
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.3883596658706665,
      "learning_rate": 3.098082092195606e-05,
      "loss": 1.8941,
      "step": 196779
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.4081512689590454,
      "learning_rate": 3.097901055898936e-05,
      "loss": 2.1191,
      "step": 196780
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.4808588027954102,
      "learning_rate": 3.097720024603878e-05,
      "loss": 2.0235,
      "step": 196781
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.6629871129989624,
      "learning_rate": 3.097538998310456e-05,
      "loss": 2.0463,
      "step": 196782
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.4671000242233276,
      "learning_rate": 3.09735797701872e-05,
      "loss": 1.9764,
      "step": 196783
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.3735780715942383,
      "learning_rate": 3.097176960728682e-05,
      "loss": 1.905,
      "step": 196784
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.3825994729995728,
      "learning_rate": 3.0969959494403907e-05,
      "loss": 2.0324,
      "step": 196785
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.5159789323806763,
      "learning_rate": 3.0968149431538716e-05,
      "loss": 1.9763,
      "step": 196786
    },
    {
      "epoch": 2.56,
      "grad_norm": 2.9267706871032715,
      "learning_rate": 3.0966339418691674e-05,
      "loss": 2.0609,
      "step": 196787
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.402081847190857,
      "learning_rate": 3.096452945586303e-05,
      "loss": 1.786,
      "step": 196788
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.4355814456939697,
      "learning_rate": 3.096271954305327e-05,
      "loss": 2.0944,
      "step": 196789
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.468934416770935,
      "learning_rate": 3.09609096802625e-05,
      "loss": 2.0936,
      "step": 196790
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.4926838874816895,
      "learning_rate": 3.095909986749122e-05,
      "loss": 2.0109,
      "step": 196791
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.5136487483978271,
      "learning_rate": 3.0957290104739664e-05,
      "loss": 2.2003,
      "step": 196792
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.4014246463775635,
      "learning_rate": 3.095548039200829e-05,
      "loss": 2.0708,
      "step": 196793
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.4365391731262207,
      "learning_rate": 3.095367072929731e-05,
      "loss": 1.8136,
      "step": 196794
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.331319808959961,
      "learning_rate": 3.095186111660725e-05,
      "loss": 1.946,
      "step": 196795
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.4105186462402344,
      "learning_rate": 3.09500515539382e-05,
      "loss": 1.9543,
      "step": 196796
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.5439609289169312,
      "learning_rate": 3.094824204129065e-05,
      "loss": 2.1689,
      "step": 196797
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.4316554069519043,
      "learning_rate": 3.094643257866488e-05,
      "loss": 1.8444,
      "step": 196798
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.4789626598358154,
      "learning_rate": 3.094462316606127e-05,
      "loss": 1.9685,
      "step": 196799
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.3518236875534058,
      "learning_rate": 3.0942813803480106e-05,
      "loss": 1.9734,
      "step": 196800
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.4438880681991577,
      "learning_rate": 3.094100449092183e-05,
      "loss": 2.0039,
      "step": 196801
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.4173213243484497,
      "learning_rate": 3.09391952283866e-05,
      "loss": 2.0201,
      "step": 196802
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.458080530166626,
      "learning_rate": 3.093738601587493e-05,
      "loss": 1.9697,
      "step": 196803
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.5358717441558838,
      "learning_rate": 3.093557685338701e-05,
      "loss": 2.0694,
      "step": 196804
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.415310263633728,
      "learning_rate": 3.093376774092331e-05,
      "loss": 2.262,
      "step": 196805
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.670599102973938,
      "learning_rate": 3.093195867848402e-05,
      "loss": 1.9225,
      "step": 196806
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.3998260498046875,
      "learning_rate": 3.0930149666069696e-05,
      "loss": 2.1173,
      "step": 196807
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.3327842950820923,
      "learning_rate": 3.092834070368041e-05,
      "loss": 2.0991,
      "step": 196808
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.5186344385147095,
      "learning_rate": 3.092653179131668e-05,
      "loss": 1.759,
      "step": 196809
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.2990829944610596,
      "learning_rate": 3.092472292897873e-05,
      "loss": 2.0149,
      "step": 196810
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.4618992805480957,
      "learning_rate": 3.0922914116667e-05,
      "loss": 2.1702,
      "step": 196811
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.3538297414779663,
      "learning_rate": 3.092110535438175e-05,
      "loss": 1.8496,
      "step": 196812
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.3484913110733032,
      "learning_rate": 3.091929664212345e-05,
      "loss": 2.1015,
      "step": 196813
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.367588996887207,
      "learning_rate": 3.0917487979892194e-05,
      "loss": 1.9452,
      "step": 196814
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.4489712715148926,
      "learning_rate": 3.091567936768856e-05,
      "loss": 1.976,
      "step": 196815
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.4413425922393799,
      "learning_rate": 3.091387080551267e-05,
      "loss": 1.9501,
      "step": 196816
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.2963656187057495,
      "learning_rate": 3.091206229336507e-05,
      "loss": 1.9182,
      "step": 196817
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.4822169542312622,
      "learning_rate": 3.091025383124591e-05,
      "loss": 1.9482,
      "step": 196818
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.4346058368682861,
      "learning_rate": 3.0908445419155704e-05,
      "loss": 1.7257,
      "step": 196819
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.288118600845337,
      "learning_rate": 3.090663705709468e-05,
      "loss": 1.7176,
      "step": 196820
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.403288722038269,
      "learning_rate": 3.090482874506319e-05,
      "loss": 1.9752,
      "step": 196821
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.4063433408737183,
      "learning_rate": 3.0903020483061494e-05,
      "loss": 1.7941,
      "step": 196822
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.36643385887146,
      "learning_rate": 3.0901212271090104e-05,
      "loss": 2.0834,
      "step": 196823
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.4346452951431274,
      "learning_rate": 3.089940410914916e-05,
      "loss": 2.0401,
      "step": 196824
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.2816050052642822,
      "learning_rate": 3.08975959972392e-05,
      "loss": 1.8441,
      "step": 196825
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.5005186796188354,
      "learning_rate": 3.089578793536042e-05,
      "loss": 2.2671,
      "step": 196826
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.3836580514907837,
      "learning_rate": 3.0893979923513146e-05,
      "loss": 2.1126,
      "step": 196827
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.4782506227493286,
      "learning_rate": 3.089217196169782e-05,
      "loss": 2.1772,
      "step": 196828
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.3908820152282715,
      "learning_rate": 3.089036404991471e-05,
      "loss": 2.0762,
      "step": 196829
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.4763003587722778,
      "learning_rate": 3.08885561881641e-05,
      "loss": 1.9905,
      "step": 196830
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.425255298614502,
      "learning_rate": 3.088674837644648e-05,
      "loss": 2.0439,
      "step": 196831
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.5343689918518066,
      "learning_rate": 3.088494061476207e-05,
      "loss": 2.1185,
      "step": 196832
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.339538812637329,
      "learning_rate": 3.088313290311116e-05,
      "loss": 2.1024,
      "step": 196833
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.6814544200897217,
      "learning_rate": 3.088132524149424e-05,
      "loss": 1.9087,
      "step": 196834
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.4979462623596191,
      "learning_rate": 3.0879517629911557e-05,
      "loss": 1.8401,
      "step": 196835
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.397863507270813,
      "learning_rate": 3.0877710068363384e-05,
      "loss": 2.1491,
      "step": 196836
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.4202980995178223,
      "learning_rate": 3.087590255685015e-05,
      "loss": 2.1534,
      "step": 196837
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.3426611423492432,
      "learning_rate": 3.087409509537223e-05,
      "loss": 2.1688,
      "step": 196838
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.3092015981674194,
      "learning_rate": 3.0872287683929784e-05,
      "loss": 2.074,
      "step": 196839
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.5286303758621216,
      "learning_rate": 3.087048032252335e-05,
      "loss": 2.016,
      "step": 196840
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.4221798181533813,
      "learning_rate": 3.086867301115309e-05,
      "loss": 2.2651,
      "step": 196841
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.4351528882980347,
      "learning_rate": 3.08668657498195e-05,
      "loss": 1.9864,
      "step": 196842
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.2589433193206787,
      "learning_rate": 3.0865058538522855e-05,
      "loss": 2.1022,
      "step": 196843
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.4165383577346802,
      "learning_rate": 3.086325137726345e-05,
      "loss": 2.1663,
      "step": 196844
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.3009417057037354,
      "learning_rate": 3.086144426604159e-05,
      "loss": 2.1031,
      "step": 196845
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.3482431173324585,
      "learning_rate": 3.085963720485774e-05,
      "loss": 2.0479,
      "step": 196846
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.4672366380691528,
      "learning_rate": 3.085783019371212e-05,
      "loss": 2.2275,
      "step": 196847
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.3951330184936523,
      "learning_rate": 3.085602323260514e-05,
      "loss": 1.7828,
      "step": 196848
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.5124845504760742,
      "learning_rate": 3.085421632153704e-05,
      "loss": 1.9617,
      "step": 196849
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.457287311553955,
      "learning_rate": 3.0852409460508375e-05,
      "loss": 2.0692,
      "step": 196850
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.452681303024292,
      "learning_rate": 3.085060264951919e-05,
      "loss": 2.0466,
      "step": 196851
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.432022213935852,
      "learning_rate": 3.084879588857e-05,
      "loss": 2.0114,
      "step": 196852
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.3906866312026978,
      "learning_rate": 3.0846989177661055e-05,
      "loss": 1.8234,
      "step": 196853
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.388981819152832,
      "learning_rate": 3.084518251679281e-05,
      "loss": 2.0034,
      "step": 196854
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.3586634397506714,
      "learning_rate": 3.084337590596544e-05,
      "loss": 2.0291,
      "step": 196855
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.4713070392608643,
      "learning_rate": 3.084156934517951e-05,
      "loss": 1.8735,
      "step": 196856
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.3972978591918945,
      "learning_rate": 3.083976283443508e-05,
      "loss": 2.0067,
      "step": 196857
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.4000142812728882,
      "learning_rate": 3.0837956373732686e-05,
      "loss": 2.0407,
      "step": 196858
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.5378403663635254,
      "learning_rate": 3.083614996307253e-05,
      "loss": 2.0743,
      "step": 196859
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.425022006034851,
      "learning_rate": 3.083434360245508e-05,
      "loss": 2.147,
      "step": 196860
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.4710941314697266,
      "learning_rate": 3.0832537291880535e-05,
      "loss": 1.8925,
      "step": 196861
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.437157392501831,
      "learning_rate": 3.083073103134942e-05,
      "loss": 1.8001,
      "step": 196862
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.5342516899108887,
      "learning_rate": 3.082892482086184e-05,
      "loss": 1.9694,
      "step": 196863
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.4690965414047241,
      "learning_rate": 3.0827118660418305e-05,
      "loss": 1.9342,
      "step": 196864
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.3875218629837036,
      "learning_rate": 3.082531255001903e-05,
      "loss": 1.8933,
      "step": 196865
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.4890378713607788,
      "learning_rate": 3.08235064896645e-05,
      "loss": 2.1165,
      "step": 196866
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.5936650037765503,
      "learning_rate": 3.082170047935486e-05,
      "loss": 2.2003,
      "step": 196867
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.3907564878463745,
      "learning_rate": 3.0819894519090694e-05,
      "loss": 1.9918,
      "step": 196868
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.3187386989593506,
      "learning_rate": 3.081808860887206e-05,
      "loss": 1.8017,
      "step": 196869
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.5783796310424805,
      "learning_rate": 3.081628274869946e-05,
      "loss": 2.2942,
      "step": 196870
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.5956323146820068,
      "learning_rate": 3.081447693857316e-05,
      "loss": 2.1945,
      "step": 196871
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.3854234218597412,
      "learning_rate": 3.08126711784936e-05,
      "loss": 1.9629,
      "step": 196872
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.3303653001785278,
      "learning_rate": 3.081086546846097e-05,
      "loss": 1.9045,
      "step": 196873
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.5926896333694458,
      "learning_rate": 3.08090598084758e-05,
      "loss": 1.8626,
      "step": 196874
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.5314544439315796,
      "learning_rate": 3.08072541985382e-05,
      "loss": 2.0766,
      "step": 196875
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.5819544792175293,
      "learning_rate": 3.0805448638648666e-05,
      "loss": 1.8933,
      "step": 196876
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.464926838874817,
      "learning_rate": 3.08036431288074e-05,
      "loss": 2.1475,
      "step": 196877
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.5398519039154053,
      "learning_rate": 3.0801837669014894e-05,
      "loss": 2.0097,
      "step": 196878
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.4379035234451294,
      "learning_rate": 3.0800032259271355e-05,
      "loss": 1.9849,
      "step": 196879
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.402449369430542,
      "learning_rate": 3.0798226899577285e-05,
      "loss": 2.0426,
      "step": 196880
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.4884110689163208,
      "learning_rate": 3.0796421589932806e-05,
      "loss": 1.9373,
      "step": 196881
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.400433897972107,
      "learning_rate": 3.079461633033839e-05,
      "loss": 2.0278,
      "step": 196882
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.5527992248535156,
      "learning_rate": 3.079281112079428e-05,
      "loss": 2.0887,
      "step": 196883
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.3652392625808716,
      "learning_rate": 3.079100596130093e-05,
      "loss": 2.1075,
      "step": 196884
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.4881513118743896,
      "learning_rate": 3.078920085185854e-05,
      "loss": 1.9303,
      "step": 196885
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.3917434215545654,
      "learning_rate": 3.0787395792467683e-05,
      "loss": 2.0038,
      "step": 196886
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.4854686260223389,
      "learning_rate": 3.0785590783128355e-05,
      "loss": 2.0497,
      "step": 196887
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.739762544631958,
      "learning_rate": 3.078378582384116e-05,
      "loss": 1.9277,
      "step": 196888
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.3666026592254639,
      "learning_rate": 3.0781980914606284e-05,
      "loss": 2.2513,
      "step": 196889
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.3811825513839722,
      "learning_rate": 3.078017605542418e-05,
      "loss": 1.9063,
      "step": 196890
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.4531930685043335,
      "learning_rate": 3.0778371246295033e-05,
      "loss": 1.846,
      "step": 196891
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.410082221031189,
      "learning_rate": 3.077656648721941e-05,
      "loss": 1.9945,
      "step": 196892
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.4281423091888428,
      "learning_rate": 3.077476177819739e-05,
      "loss": 1.8018,
      "step": 196893
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.48000967502594,
      "learning_rate": 3.077295711922949e-05,
      "loss": 2.1287,
      "step": 196894
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.4301947355270386,
      "learning_rate": 3.077115251031589e-05,
      "loss": 2.002,
      "step": 196895
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.6748789548873901,
      "learning_rate": 3.076934795145711e-05,
      "loss": 2.0492,
      "step": 196896
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.5601314306259155,
      "learning_rate": 3.07675434426533e-05,
      "loss": 1.8835,
      "step": 196897
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.7419289350509644,
      "learning_rate": 3.076573898390501e-05,
      "loss": 1.8331,
      "step": 196898
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.4794402122497559,
      "learning_rate": 3.076393457521235e-05,
      "loss": 2.004,
      "step": 196899
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.294735312461853,
      "learning_rate": 3.076213021657581e-05,
      "loss": 1.7457,
      "step": 196900
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.4368221759796143,
      "learning_rate": 3.0760325907995595e-05,
      "loss": 2.0857,
      "step": 196901
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.5486253499984741,
      "learning_rate": 3.0758521649472214e-05,
      "loss": 2.138,
      "step": 196902
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.4386202096939087,
      "learning_rate": 3.0756717441005816e-05,
      "loss": 2.0824,
      "step": 196903
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.475247859954834,
      "learning_rate": 3.0754913282596884e-05,
      "loss": 2.004,
      "step": 196904
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.4977504014968872,
      "learning_rate": 3.07531091742457e-05,
      "loss": 1.9592,
      "step": 196905
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.4898455142974854,
      "learning_rate": 3.0751305115952615e-05,
      "loss": 2.1244,
      "step": 196906
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.73770272731781,
      "learning_rate": 3.074950110771789e-05,
      "loss": 1.9228,
      "step": 196907
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.4826395511627197,
      "learning_rate": 3.074769714954195e-05,
      "loss": 2.1206,
      "step": 196908
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.4400653839111328,
      "learning_rate": 3.074589324142504e-05,
      "loss": 2.0442,
      "step": 196909
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.5079200267791748,
      "learning_rate": 3.074408938336761e-05,
      "loss": 2.1736,
      "step": 196910
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.371969223022461,
      "learning_rate": 3.074228557536994e-05,
      "loss": 2.0496,
      "step": 196911
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.3083198070526123,
      "learning_rate": 3.074048181743229e-05,
      "loss": 1.8758,
      "step": 196912
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.3410826921463013,
      "learning_rate": 3.073867810955517e-05,
      "loss": 2.007,
      "step": 196913
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.4267699718475342,
      "learning_rate": 3.073687445173877e-05,
      "loss": 1.8542,
      "step": 196914
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.4416775703430176,
      "learning_rate": 3.073507084398342e-05,
      "loss": 2.0594,
      "step": 196915
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.2893497943878174,
      "learning_rate": 3.073326728628956e-05,
      "loss": 2.018,
      "step": 196916
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.386396050453186,
      "learning_rate": 3.073146377865746e-05,
      "loss": 1.94,
      "step": 196917
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.3323718309402466,
      "learning_rate": 3.072966032108741e-05,
      "loss": 1.8835,
      "step": 196918
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.501881718635559,
      "learning_rate": 3.072785691357988e-05,
      "loss": 1.9607,
      "step": 196919
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.3931729793548584,
      "learning_rate": 3.072605355613511e-05,
      "loss": 1.9836,
      "step": 196920
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.2784812450408936,
      "learning_rate": 3.072425024875339e-05,
      "loss": 1.799,
      "step": 196921
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.5616247653961182,
      "learning_rate": 3.072244699143516e-05,
      "loss": 1.8715,
      "step": 196922
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.3510736227035522,
      "learning_rate": 3.072064378418074e-05,
      "loss": 1.9311,
      "step": 196923
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.531159520149231,
      "learning_rate": 3.0718840626990346e-05,
      "loss": 1.9553,
      "step": 196924
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.5031828880310059,
      "learning_rate": 3.071703751986447e-05,
      "loss": 1.9396,
      "step": 196925
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.4441893100738525,
      "learning_rate": 3.071523446280332e-05,
      "loss": 2.0164,
      "step": 196926
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.362831473350525,
      "learning_rate": 3.071343145580738e-05,
      "loss": 1.849,
      "step": 196927
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.4756208658218384,
      "learning_rate": 3.071162849887686e-05,
      "loss": 1.9604,
      "step": 196928
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.312270998954773,
      "learning_rate": 3.070982559201213e-05,
      "loss": 2.0139,
      "step": 196929
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.3866255283355713,
      "learning_rate": 3.0708022735213477e-05,
      "loss": 1.9231,
      "step": 196930
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.4007270336151123,
      "learning_rate": 3.070621992848138e-05,
      "loss": 1.9081,
      "step": 196931
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.4886529445648193,
      "learning_rate": 3.070441717181597e-05,
      "loss": 1.9891,
      "step": 196932
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.4227477312088013,
      "learning_rate": 3.070261446521777e-05,
      "loss": 1.8787,
      "step": 196933
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.5280832052230835,
      "learning_rate": 3.0700811808687054e-05,
      "loss": 2.1022,
      "step": 196934
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.4589844942092896,
      "learning_rate": 3.0699009202224125e-05,
      "loss": 1.9985,
      "step": 196935
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.3529309034347534,
      "learning_rate": 3.0697206645829274e-05,
      "loss": 1.8571,
      "step": 196936
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.281846523284912,
      "learning_rate": 3.069540413950295e-05,
      "loss": 1.9307,
      "step": 196937
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.387929081916809,
      "learning_rate": 3.069360168324536e-05,
      "loss": 1.9519,
      "step": 196938
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.2896562814712524,
      "learning_rate": 3.0691799277056996e-05,
      "loss": 1.9929,
      "step": 196939
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.4494621753692627,
      "learning_rate": 3.0689996920938044e-05,
      "loss": 2.1552,
      "step": 196940
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.419196367263794,
      "learning_rate": 3.068819461488904e-05,
      "loss": 2.1736,
      "step": 196941
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.3794713020324707,
      "learning_rate": 3.068639235891005e-05,
      "loss": 1.735,
      "step": 196942
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.4918618202209473,
      "learning_rate": 3.068459015300161e-05,
      "loss": 1.8913,
      "step": 196943
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.5905653238296509,
      "learning_rate": 3.0682787997163914e-05,
      "loss": 1.866,
      "step": 196944
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.424424409866333,
      "learning_rate": 3.0680985891397435e-05,
      "loss": 2.0629,
      "step": 196945
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.5373212099075317,
      "learning_rate": 3.06791838357024e-05,
      "loss": 1.8274,
      "step": 196946
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.4463341236114502,
      "learning_rate": 3.0677381830079285e-05,
      "loss": 2.0372,
      "step": 196947
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.4274969100952148,
      "learning_rate": 3.0675579874528244e-05,
      "loss": 1.9293,
      "step": 196948
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.4088062047958374,
      "learning_rate": 3.067377796904972e-05,
      "loss": 2.0155,
      "step": 196949
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.3797651529312134,
      "learning_rate": 3.0671976113644006e-05,
      "loss": 1.9738,
      "step": 196950
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.461463451385498,
      "learning_rate": 3.067017430831147e-05,
      "loss": 2.0911,
      "step": 196951
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.4890507459640503,
      "learning_rate": 3.066837255305238e-05,
      "loss": 2.0307,
      "step": 196952
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.5675843954086304,
      "learning_rate": 3.066657084786727e-05,
      "loss": 1.9154,
      "step": 196953
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.4935556650161743,
      "learning_rate": 3.06647691927562e-05,
      "loss": 2.0336,
      "step": 196954
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.5076875686645508,
      "learning_rate": 3.0662967587719676e-05,
      "loss": 2.0821,
      "step": 196955
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.4128825664520264,
      "learning_rate": 3.066116603275793e-05,
      "loss": 1.8937,
      "step": 196956
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.4807112216949463,
      "learning_rate": 3.0659364527871436e-05,
      "loss": 2.0489,
      "step": 196957
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.620173692703247,
      "learning_rate": 3.0657563073060374e-05,
      "loss": 1.8985,
      "step": 196958
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.3744066953659058,
      "learning_rate": 3.06557616683253e-05,
      "loss": 2.0687,
      "step": 196959
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.4461933374404907,
      "learning_rate": 3.065396031366626e-05,
      "loss": 1.8328,
      "step": 196960
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.5692143440246582,
      "learning_rate": 3.06521590090838e-05,
      "loss": 2.1151,
      "step": 196961
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.3568799495697021,
      "learning_rate": 3.0650357754578145e-05,
      "loss": 2.1756,
      "step": 196962
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.7688181400299072,
      "learning_rate": 3.0648556550149705e-05,
      "loss": 1.9396,
      "step": 196963
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.4958436489105225,
      "learning_rate": 3.064675539579874e-05,
      "loss": 1.9295,
      "step": 196964
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.6058628559112549,
      "learning_rate": 3.0644954291525745e-05,
      "loss": 2.1375,
      "step": 196965
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.439107894897461,
      "learning_rate": 3.064315323733083e-05,
      "loss": 1.8393,
      "step": 196966
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.3845068216323853,
      "learning_rate": 3.0641352233214486e-05,
      "loss": 1.9891,
      "step": 196967
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.3163203001022339,
      "learning_rate": 3.063955127917692e-05,
      "loss": 1.8798,
      "step": 196968
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.4745302200317383,
      "learning_rate": 3.063775037521863e-05,
      "loss": 2.002,
      "step": 196969
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.6128513813018799,
      "learning_rate": 3.063594952133981e-05,
      "loss": 1.957,
      "step": 196970
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.356197714805603,
      "learning_rate": 3.0634148717540966e-05,
      "loss": 2.1147,
      "step": 196971
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.5063345432281494,
      "learning_rate": 3.063234796382219e-05,
      "loss": 1.91,
      "step": 196972
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.323754072189331,
      "learning_rate": 3.063054726018399e-05,
      "loss": 1.9866,
      "step": 196973
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.4364858865737915,
      "learning_rate": 3.062874660662663e-05,
      "loss": 1.8877,
      "step": 196974
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.3541666269302368,
      "learning_rate": 3.062694600315051e-05,
      "loss": 1.7981,
      "step": 196975
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.4567846059799194,
      "learning_rate": 3.062514544975586e-05,
      "loss": 1.9892,
      "step": 196976
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.4130710363388062,
      "learning_rate": 3.062334494644315e-05,
      "loss": 2.0596,
      "step": 196977
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.3787307739257812,
      "learning_rate": 3.062154449321265e-05,
      "loss": 1.8902,
      "step": 196978
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.4140897989273071,
      "learning_rate": 3.061974409006468e-05,
      "loss": 2.1432,
      "step": 196979
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.4366871118545532,
      "learning_rate": 3.0617943736999554e-05,
      "loss": 1.9591,
      "step": 196980
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.3453514575958252,
      "learning_rate": 3.061614343401766e-05,
      "loss": 2.0129,
      "step": 196981
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.351643443107605,
      "learning_rate": 3.061434318111927e-05,
      "loss": 1.9482,
      "step": 196982
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.503507137298584,
      "learning_rate": 3.0612542978304786e-05,
      "loss": 1.9801,
      "step": 196983
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.4243792295455933,
      "learning_rate": 3.061074282557454e-05,
      "loss": 2.0222,
      "step": 196984
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.2991987466812134,
      "learning_rate": 3.0608942722928864e-05,
      "loss": 1.8723,
      "step": 196985
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.4624632596969604,
      "learning_rate": 3.060714267036799e-05,
      "loss": 1.9011,
      "step": 196986
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.3092424869537354,
      "learning_rate": 3.060534266789238e-05,
      "loss": 1.8442,
      "step": 196987
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.3529562950134277,
      "learning_rate": 3.0603542715502284e-05,
      "loss": 1.6688,
      "step": 196988
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.426417350769043,
      "learning_rate": 3.060174281319812e-05,
      "loss": 1.941,
      "step": 196989
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.420127272605896,
      "learning_rate": 3.059994296098018e-05,
      "loss": 2.1887,
      "step": 196990
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.493494987487793,
      "learning_rate": 3.059814315884882e-05,
      "loss": 1.8692,
      "step": 196991
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.5531830787658691,
      "learning_rate": 3.059634340680426e-05,
      "loss": 1.9096,
      "step": 196992
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.427476406097412,
      "learning_rate": 3.0594543704847e-05,
      "loss": 2.148,
      "step": 196993
    },
    {
      "epoch": 2.56,
      "grad_norm": 1.5083205699920654,
      "learning_rate": 3.059274405297724e-05,
      "loss": 2.0771,
      "step": 196994
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.441023826599121,
      "learning_rate": 3.059094445119541e-05,
      "loss": 1.9607,
      "step": 196995
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.3563138246536255,
      "learning_rate": 3.058914489950185e-05,
      "loss": 2.2361,
      "step": 196996
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.4299753904342651,
      "learning_rate": 3.0587345397896825e-05,
      "loss": 2.1106,
      "step": 196997
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.500829815864563,
      "learning_rate": 3.058554594638063e-05,
      "loss": 1.8973,
      "step": 196998
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.6038634777069092,
      "learning_rate": 3.058374654495377e-05,
      "loss": 2.0972,
      "step": 196999
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.2726035118103027,
      "learning_rate": 3.058194719361637e-05,
      "loss": 1.7066,
      "step": 197000
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.5675734281539917,
      "learning_rate": 3.0580147892368947e-05,
      "loss": 1.9417,
      "step": 197001
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.2850749492645264,
      "learning_rate": 3.057834864121178e-05,
      "loss": 1.7646,
      "step": 197002
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.3882131576538086,
      "learning_rate": 3.0576549440145115e-05,
      "loss": 1.7411,
      "step": 197003
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.428335189819336,
      "learning_rate": 3.057475028916938e-05,
      "loss": 1.8702,
      "step": 197004
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.465126872062683,
      "learning_rate": 3.0572951188284914e-05,
      "loss": 1.936,
      "step": 197005
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.552445411682129,
      "learning_rate": 3.057115213749197e-05,
      "loss": 2.0052,
      "step": 197006
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.4792171716690063,
      "learning_rate": 3.0569353136791e-05,
      "loss": 1.6826,
      "step": 197007
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.4775004386901855,
      "learning_rate": 3.0567554186182256e-05,
      "loss": 1.754,
      "step": 197008
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.385693073272705,
      "learning_rate": 3.056575528566604e-05,
      "loss": 1.9771,
      "step": 197009
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.3294867277145386,
      "learning_rate": 3.0563956435242785e-05,
      "loss": 1.983,
      "step": 197010
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.447677731513977,
      "learning_rate": 3.05621576349127e-05,
      "loss": 2.0007,
      "step": 197011
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.3698399066925049,
      "learning_rate": 3.0560358884676275e-05,
      "loss": 1.9319,
      "step": 197012
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.4411565065383911,
      "learning_rate": 3.055856018453378e-05,
      "loss": 1.8955,
      "step": 197013
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.2987438440322876,
      "learning_rate": 3.0556761534485506e-05,
      "loss": 1.8239,
      "step": 197014
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.4294358491897583,
      "learning_rate": 3.0554962934531767e-05,
      "loss": 1.9501,
      "step": 197015
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.434820294380188,
      "learning_rate": 3.0553164384673026e-05,
      "loss": 1.8869,
      "step": 197016
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.5299185514450073,
      "learning_rate": 3.0551365884909473e-05,
      "loss": 2.0775,
      "step": 197017
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.5204293727874756,
      "learning_rate": 3.054956743524155e-05,
      "loss": 2.0104,
      "step": 197018
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.3470646142959595,
      "learning_rate": 3.0547769035669555e-05,
      "loss": 1.7326,
      "step": 197019
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.4619438648223877,
      "learning_rate": 3.054597068619382e-05,
      "loss": 1.9609,
      "step": 197020
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.3988178968429565,
      "learning_rate": 3.054417238681461e-05,
      "loss": 1.8162,
      "step": 197021
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.4983844757080078,
      "learning_rate": 3.05423741375324e-05,
      "loss": 2.0394,
      "step": 197022
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.3591386079788208,
      "learning_rate": 3.054057593834738e-05,
      "loss": 1.907,
      "step": 197023
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.5536139011383057,
      "learning_rate": 3.0538777789259984e-05,
      "loss": 2.0724,
      "step": 197024
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.397357702255249,
      "learning_rate": 3.053697969027048e-05,
      "loss": 1.9634,
      "step": 197025
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.4136536121368408,
      "learning_rate": 3.053518164137937e-05,
      "loss": 1.9092,
      "step": 197026
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.4133031368255615,
      "learning_rate": 3.053338364258672e-05,
      "loss": 1.8177,
      "step": 197027
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.5731874704360962,
      "learning_rate": 3.053158569389306e-05,
      "loss": 1.9626,
      "step": 197028
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.3469972610473633,
      "learning_rate": 3.052978779529862e-05,
      "loss": 2.1458,
      "step": 197029
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.608238935470581,
      "learning_rate": 3.0527989946803846e-05,
      "loss": 1.8959,
      "step": 197030
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.6731274127960205,
      "learning_rate": 3.0526192148408925e-05,
      "loss": 1.905,
      "step": 197031
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.6388291120529175,
      "learning_rate": 3.0524394400114396e-05,
      "loss": 1.9096,
      "step": 197032
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.5888645648956299,
      "learning_rate": 3.052259670192032e-05,
      "loss": 2.1341,
      "step": 197033
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.4715385437011719,
      "learning_rate": 3.0520799053827275e-05,
      "loss": 2.2333,
      "step": 197034
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.388762354850769,
      "learning_rate": 3.051900145583544e-05,
      "loss": 2.017,
      "step": 197035
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.5271658897399902,
      "learning_rate": 3.0517203907945277e-05,
      "loss": 2.0093,
      "step": 197036
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.5469492673873901,
      "learning_rate": 3.051540641015696e-05,
      "loss": 1.9648,
      "step": 197037
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.511113166809082,
      "learning_rate": 3.051360896247107e-05,
      "loss": 1.7542,
      "step": 197038
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.4599229097366333,
      "learning_rate": 3.0511811564887633e-05,
      "loss": 1.9273,
      "step": 197039
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.4552475214004517,
      "learning_rate": 3.051001421740722e-05,
      "loss": 2.0444,
      "step": 197040
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.5340265035629272,
      "learning_rate": 3.0508216920030027e-05,
      "loss": 2.1008,
      "step": 197041
    },
    {
      "epoch": 2.57,
      "grad_norm": 2.351529598236084,
      "learning_rate": 3.050641967275649e-05,
      "loss": 1.9922,
      "step": 197042
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.7238247394561768,
      "learning_rate": 3.0504622475586838e-05,
      "loss": 1.9692,
      "step": 197043
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.6897298097610474,
      "learning_rate": 3.050282532852154e-05,
      "loss": 2.1938,
      "step": 197044
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.4590587615966797,
      "learning_rate": 3.0501028231560832e-05,
      "loss": 1.9666,
      "step": 197045
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.5410711765289307,
      "learning_rate": 3.0499231184705074e-05,
      "loss": 2.1401,
      "step": 197046
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.5288565158843994,
      "learning_rate": 3.049743418795457e-05,
      "loss": 2.0888,
      "step": 197047
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.4415854215621948,
      "learning_rate": 3.049563724130969e-05,
      "loss": 1.8273,
      "step": 197048
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.4391882419586182,
      "learning_rate": 3.0493840344770726e-05,
      "loss": 2.0942,
      "step": 197049
    },
    {
      "epoch": 2.57,
      "grad_norm": 2.0383498668670654,
      "learning_rate": 3.0492043498338116e-05,
      "loss": 1.8288,
      "step": 197050
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.3904476165771484,
      "learning_rate": 3.0490246702012088e-05,
      "loss": 1.8243,
      "step": 197051
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.8599085807800293,
      "learning_rate": 3.0488449955793016e-05,
      "loss": 2.0482,
      "step": 197052
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.5027605295181274,
      "learning_rate": 3.048665325968119e-05,
      "loss": 2.0952,
      "step": 197053
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.346337914466858,
      "learning_rate": 3.0484856613677023e-05,
      "loss": 2.0044,
      "step": 197054
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.437451720237732,
      "learning_rate": 3.048306001778077e-05,
      "loss": 2.0011,
      "step": 197055
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.5133121013641357,
      "learning_rate": 3.048126347199287e-05,
      "loss": 2.0097,
      "step": 197056
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.4423383474349976,
      "learning_rate": 3.047946697631355e-05,
      "loss": 1.9255,
      "step": 197057
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.8422223329544067,
      "learning_rate": 3.0477670530743214e-05,
      "loss": 1.9525,
      "step": 197058
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.5092062950134277,
      "learning_rate": 3.0475874135282098e-05,
      "loss": 2.0795,
      "step": 197059
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.435609221458435,
      "learning_rate": 3.0474077789930662e-05,
      "loss": 2.1001,
      "step": 197060
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.3124064207077026,
      "learning_rate": 3.0472281494689143e-05,
      "loss": 1.7579,
      "step": 197061
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.4575800895690918,
      "learning_rate": 3.047048524955794e-05,
      "loss": 2.1118,
      "step": 197062
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.6178865432739258,
      "learning_rate": 3.0468689054537387e-05,
      "loss": 1.794,
      "step": 197063
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.3690074682235718,
      "learning_rate": 3.0466892909627784e-05,
      "loss": 1.8225,
      "step": 197064
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.6727895736694336,
      "learning_rate": 3.0465096814829425e-05,
      "loss": 2.005,
      "step": 197065
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.674876093864441,
      "learning_rate": 3.0463300770142752e-05,
      "loss": 2.1854,
      "step": 197066
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.3912882804870605,
      "learning_rate": 3.0461504775567958e-05,
      "loss": 2.1499,
      "step": 197067
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.4532628059387207,
      "learning_rate": 3.0459708831105544e-05,
      "loss": 2.094,
      "step": 197068
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.5519506931304932,
      "learning_rate": 3.0457912936755747e-05,
      "loss": 1.9229,
      "step": 197069
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.4212157726287842,
      "learning_rate": 3.0456117092518896e-05,
      "loss": 1.9355,
      "step": 197070
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.3575685024261475,
      "learning_rate": 3.045432129839529e-05,
      "loss": 1.9189,
      "step": 197071
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.42327880859375,
      "learning_rate": 3.0452525554385397e-05,
      "loss": 1.9089,
      "step": 197072
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.4668684005737305,
      "learning_rate": 3.0450729860489386e-05,
      "loss": 2.0851,
      "step": 197073
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.3906564712524414,
      "learning_rate": 3.044893421670772e-05,
      "loss": 1.9214,
      "step": 197074
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.5233168601989746,
      "learning_rate": 3.04471386230407e-05,
      "loss": 1.9219,
      "step": 197075
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.460203766822815,
      "learning_rate": 3.044534307948866e-05,
      "loss": 1.9829,
      "step": 197076
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.5099525451660156,
      "learning_rate": 3.044354758605183e-05,
      "loss": 2.0027,
      "step": 197077
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.4981242418289185,
      "learning_rate": 3.0441752142730712e-05,
      "loss": 2.236,
      "step": 197078
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.3521300554275513,
      "learning_rate": 3.0439956749525506e-05,
      "loss": 1.9237,
      "step": 197079
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.4620765447616577,
      "learning_rate": 3.0438161406436647e-05,
      "loss": 1.7888,
      "step": 197080
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.3461650609970093,
      "learning_rate": 3.0436366113464426e-05,
      "loss": 2.0929,
      "step": 197081
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.3358148336410522,
      "learning_rate": 3.0434570870609153e-05,
      "loss": 2.1296,
      "step": 197082
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.3166918754577637,
      "learning_rate": 3.0432775677871156e-05,
      "loss": 1.9849,
      "step": 197083
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.4871994256973267,
      "learning_rate": 3.043098053525084e-05,
      "loss": 2.0023,
      "step": 197084
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.4761449098587036,
      "learning_rate": 3.0429185442748427e-05,
      "loss": 2.0045,
      "step": 197085
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.4271032810211182,
      "learning_rate": 3.0427390400364392e-05,
      "loss": 2.1716,
      "step": 197086
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.4243189096450806,
      "learning_rate": 3.0425595408098968e-05,
      "loss": 2.1562,
      "step": 197087
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.3672008514404297,
      "learning_rate": 3.0423800465952487e-05,
      "loss": 2.0527,
      "step": 197088
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.357562780380249,
      "learning_rate": 3.0422005573925345e-05,
      "loss": 2.0052,
      "step": 197089
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.430979609489441,
      "learning_rate": 3.0420210732017848e-05,
      "loss": 1.888,
      "step": 197090
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.4895646572113037,
      "learning_rate": 3.0418415940230256e-05,
      "loss": 1.9135,
      "step": 197091
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.385805368423462,
      "learning_rate": 3.041662119856304e-05,
      "loss": 2.0151,
      "step": 197092
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.4004746675491333,
      "learning_rate": 3.0414826507016467e-05,
      "loss": 2.013,
      "step": 197093
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.473922610282898,
      "learning_rate": 3.04130318655908e-05,
      "loss": 1.9903,
      "step": 197094
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.45177161693573,
      "learning_rate": 3.0411237274286503e-05,
      "loss": 1.9397,
      "step": 197095
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.311671257019043,
      "learning_rate": 3.040944273310385e-05,
      "loss": 2.0013,
      "step": 197096
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.5188170671463013,
      "learning_rate": 3.0407648242043105e-05,
      "loss": 2.0373,
      "step": 197097
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.5316648483276367,
      "learning_rate": 3.0405853801104728e-05,
      "loss": 1.9984,
      "step": 197098
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.5570729970932007,
      "learning_rate": 3.040405941028896e-05,
      "loss": 2.2456,
      "step": 197099
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.5456910133361816,
      "learning_rate": 3.0402265069596167e-05,
      "loss": 2.1287,
      "step": 197100
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.4595156908035278,
      "learning_rate": 3.0400470779026708e-05,
      "loss": 1.8604,
      "step": 197101
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.4676553010940552,
      "learning_rate": 3.0398676538580823e-05,
      "loss": 1.9727,
      "step": 197102
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.489276647567749,
      "learning_rate": 3.0396882348258976e-05,
      "loss": 2.0669,
      "step": 197103
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.428304672241211,
      "learning_rate": 3.0395088208061436e-05,
      "loss": 1.929,
      "step": 197104
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.4663079977035522,
      "learning_rate": 3.0393294117988565e-05,
      "loss": 2.1855,
      "step": 197105
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.422374963760376,
      "learning_rate": 3.0391500078040597e-05,
      "loss": 1.9132,
      "step": 197106
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.4196780920028687,
      "learning_rate": 3.0389706088217968e-05,
      "loss": 2.0527,
      "step": 197107
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.4613112211227417,
      "learning_rate": 3.0387912148520977e-05,
      "loss": 1.8486,
      "step": 197108
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.3576085567474365,
      "learning_rate": 3.0386118258949986e-05,
      "loss": 2.0909,
      "step": 197109
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.5133309364318848,
      "learning_rate": 3.0384324419505236e-05,
      "loss": 2.0824,
      "step": 197110
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.2878795862197876,
      "learning_rate": 3.038253063018722e-05,
      "loss": 2.0561,
      "step": 197111
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.4066179990768433,
      "learning_rate": 3.038073689099617e-05,
      "loss": 2.0107,
      "step": 197112
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.3752731084823608,
      "learning_rate": 3.0378943201932427e-05,
      "loss": 2.0287,
      "step": 197113
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.451918125152588,
      "learning_rate": 3.037714956299625e-05,
      "loss": 1.743,
      "step": 197114
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.3336296081542969,
      "learning_rate": 3.037535597418814e-05,
      "loss": 1.9866,
      "step": 197115
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.4471238851547241,
      "learning_rate": 3.037356243550827e-05,
      "loss": 2.0498,
      "step": 197116
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.342207908630371,
      "learning_rate": 3.0371768946957097e-05,
      "loss": 2.2108,
      "step": 197117
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.4458805322647095,
      "learning_rate": 3.0369975508534892e-05,
      "loss": 1.8437,
      "step": 197118
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.4660265445709229,
      "learning_rate": 3.0368182120242025e-05,
      "loss": 1.9193,
      "step": 197119
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.5467585325241089,
      "learning_rate": 3.0366388782078722e-05,
      "loss": 1.9211,
      "step": 197120
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.3374181985855103,
      "learning_rate": 3.0364595494045485e-05,
      "loss": 1.8618,
      "step": 197121
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.4839437007904053,
      "learning_rate": 3.036280225614248e-05,
      "loss": 1.9397,
      "step": 197122
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.3303039073944092,
      "learning_rate": 3.036100906837018e-05,
      "loss": 1.9807,
      "step": 197123
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.3577457666397095,
      "learning_rate": 3.0359215930728842e-05,
      "loss": 1.9558,
      "step": 197124
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.477200984954834,
      "learning_rate": 3.0357422843218838e-05,
      "loss": 2.0088,
      "step": 197125
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.526227593421936,
      "learning_rate": 3.035562980584043e-05,
      "loss": 1.8569,
      "step": 197126
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.2931268215179443,
      "learning_rate": 3.0353836818594025e-05,
      "loss": 2.0974,
      "step": 197127
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.362838864326477,
      "learning_rate": 3.0352043881479915e-05,
      "loss": 1.8024,
      "step": 197128
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.5085963010787964,
      "learning_rate": 3.035025099449847e-05,
      "loss": 2.066,
      "step": 197129
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.4287034273147583,
      "learning_rate": 3.0348458157650024e-05,
      "loss": 1.9705,
      "step": 197130
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.447485089302063,
      "learning_rate": 3.0346665370934874e-05,
      "loss": 1.8849,
      "step": 197131
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.5252649784088135,
      "learning_rate": 3.0344872634353323e-05,
      "loss": 2.1075,
      "step": 197132
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.2982280254364014,
      "learning_rate": 3.0343079947905803e-05,
      "loss": 1.9956,
      "step": 197133
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.3584578037261963,
      "learning_rate": 3.0341287311592543e-05,
      "loss": 1.982,
      "step": 197134
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.5540114641189575,
      "learning_rate": 3.033949472541398e-05,
      "loss": 2.128,
      "step": 197135
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.4063133001327515,
      "learning_rate": 3.0337702189370415e-05,
      "loss": 1.9696,
      "step": 197136
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.4111278057098389,
      "learning_rate": 3.0335909703462147e-05,
      "loss": 1.8575,
      "step": 197137
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.5494322776794434,
      "learning_rate": 3.033411726768944e-05,
      "loss": 1.8534,
      "step": 197138
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.3716981410980225,
      "learning_rate": 3.0332324882052793e-05,
      "loss": 1.9996,
      "step": 197139
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.4693549871444702,
      "learning_rate": 3.033053254655241e-05,
      "loss": 2.1078,
      "step": 197140
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.363141655921936,
      "learning_rate": 3.032874026118872e-05,
      "loss": 1.9847,
      "step": 197141
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.5357840061187744,
      "learning_rate": 3.0326948025961994e-05,
      "loss": 2.0748,
      "step": 197142
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.7836862802505493,
      "learning_rate": 3.0325155840872594e-05,
      "loss": 2.007,
      "step": 197143
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.3972059488296509,
      "learning_rate": 3.032336370592079e-05,
      "loss": 2.0385,
      "step": 197144
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.3567336797714233,
      "learning_rate": 3.032157162110701e-05,
      "loss": 1.9001,
      "step": 197145
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.4793063402175903,
      "learning_rate": 3.031977958643146e-05,
      "loss": 2.025,
      "step": 197146
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.399871587753296,
      "learning_rate": 3.0317987601894634e-05,
      "loss": 1.8935,
      "step": 197147
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.2453563213348389,
      "learning_rate": 3.0316195667496802e-05,
      "loss": 1.9059,
      "step": 197148
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.5981144905090332,
      "learning_rate": 3.031440378323826e-05,
      "loss": 1.668,
      "step": 197149
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.3504812717437744,
      "learning_rate": 3.031261194911928e-05,
      "loss": 1.8851,
      "step": 197150
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.383441686630249,
      "learning_rate": 3.0310820165140358e-05,
      "loss": 2.0162,
      "step": 197151
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.4383078813552856,
      "learning_rate": 3.0309028431301696e-05,
      "loss": 2.192,
      "step": 197152
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.352458119392395,
      "learning_rate": 3.0307236747603726e-05,
      "loss": 1.7863,
      "step": 197153
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.5261591672897339,
      "learning_rate": 3.0305445114046745e-05,
      "loss": 2.2036,
      "step": 197154
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.32819402217865,
      "learning_rate": 3.0303653530631055e-05,
      "loss": 2.0542,
      "step": 197155
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.6678407192230225,
      "learning_rate": 3.0301861997356923e-05,
      "loss": 2.1059,
      "step": 197156
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.39970064163208,
      "learning_rate": 3.0300070514224882e-05,
      "loss": 1.8864,
      "step": 197157
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.465932011604309,
      "learning_rate": 3.0298279081235068e-05,
      "loss": 1.9625,
      "step": 197158
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.5172843933105469,
      "learning_rate": 3.029648769838794e-05,
      "loss": 2.052,
      "step": 197159
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.4292609691619873,
      "learning_rate": 3.0294696365683767e-05,
      "loss": 1.9016,
      "step": 197160
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.361055612564087,
      "learning_rate": 3.0292905083122954e-05,
      "loss": 1.891,
      "step": 197161
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.4417670965194702,
      "learning_rate": 3.0291113850705697e-05,
      "loss": 1.932,
      "step": 197162
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.6345607042312622,
      "learning_rate": 3.0289322668432457e-05,
      "loss": 2.0831,
      "step": 197163
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.5497514009475708,
      "learning_rate": 3.0287531536303477e-05,
      "loss": 2.0535,
      "step": 197164
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.4203286170959473,
      "learning_rate": 3.0285740454319186e-05,
      "loss": 2.0721,
      "step": 197165
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.355414867401123,
      "learning_rate": 3.0283949422479882e-05,
      "loss": 1.7567,
      "step": 197166
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.4787874221801758,
      "learning_rate": 3.0282158440785864e-05,
      "loss": 1.9011,
      "step": 197167
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.2982832193374634,
      "learning_rate": 3.028036750923747e-05,
      "loss": 1.9338,
      "step": 197168
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.4667822122573853,
      "learning_rate": 3.027857662783506e-05,
      "loss": 2.0142,
      "step": 197169
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.414989948272705,
      "learning_rate": 3.0276785796578905e-05,
      "loss": 2.0963,
      "step": 197170
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.5146008729934692,
      "learning_rate": 3.0274995015469438e-05,
      "loss": 2.0542,
      "step": 197171
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.4608533382415771,
      "learning_rate": 3.0273204284506957e-05,
      "loss": 2.0291,
      "step": 197172
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.411015510559082,
      "learning_rate": 3.027141360369173e-05,
      "loss": 2.0539,
      "step": 197173
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.468430519104004,
      "learning_rate": 3.0269622973024188e-05,
      "loss": 1.9863,
      "step": 197174
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.4390130043029785,
      "learning_rate": 3.0267832392504598e-05,
      "loss": 1.9676,
      "step": 197175
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.6165881156921387,
      "learning_rate": 3.026604186213326e-05,
      "loss": 2.0587,
      "step": 197176
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.3175477981567383,
      "learning_rate": 3.0264251381910642e-05,
      "loss": 2.1003,
      "step": 197177
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.467484951019287,
      "learning_rate": 3.0262460951836907e-05,
      "loss": 1.9046,
      "step": 197178
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.3958832025527954,
      "learning_rate": 3.0260670571912527e-05,
      "loss": 1.9326,
      "step": 197179
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.4401458501815796,
      "learning_rate": 3.0258880242137796e-05,
      "loss": 1.8956,
      "step": 197180
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.5211925506591797,
      "learning_rate": 3.0257089962513013e-05,
      "loss": 1.8304,
      "step": 197181
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.4434716701507568,
      "learning_rate": 3.0255299733038486e-05,
      "loss": 1.8898,
      "step": 197182
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.4414498805999756,
      "learning_rate": 3.025350955371464e-05,
      "loss": 1.9654,
      "step": 197183
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.3505045175552368,
      "learning_rate": 3.0251719424541677e-05,
      "loss": 2.0419,
      "step": 197184
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.4922881126403809,
      "learning_rate": 3.0249929345520097e-05,
      "loss": 2.0963,
      "step": 197185
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.3701967000961304,
      "learning_rate": 3.0248139316650165e-05,
      "loss": 1.9287,
      "step": 197186
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.452121376991272,
      "learning_rate": 3.024634933793212e-05,
      "loss": 1.8447,
      "step": 197187
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.4827404022216797,
      "learning_rate": 3.024455940936642e-05,
      "loss": 2.0337,
      "step": 197188
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.569325566291809,
      "learning_rate": 3.0242769530953336e-05,
      "loss": 2.0647,
      "step": 197189
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.3379267454147339,
      "learning_rate": 3.024097970269317e-05,
      "loss": 2.1334,
      "step": 197190
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.4121946096420288,
      "learning_rate": 3.0239189924586383e-05,
      "loss": 1.9608,
      "step": 197191
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.48236083984375,
      "learning_rate": 3.0237400196633176e-05,
      "loss": 1.98,
      "step": 197192
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.5161253213882446,
      "learning_rate": 3.023561051883392e-05,
      "loss": 1.9862,
      "step": 197193
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.4441479444503784,
      "learning_rate": 3.0233820891188975e-05,
      "loss": 1.9319,
      "step": 197194
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.4748950004577637,
      "learning_rate": 3.023203131369868e-05,
      "loss": 1.9739,
      "step": 197195
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.434434413909912,
      "learning_rate": 3.0230241786363263e-05,
      "loss": 1.7483,
      "step": 197196
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.402480125427246,
      "learning_rate": 3.0228452309183193e-05,
      "loss": 2.1396,
      "step": 197197
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.4262404441833496,
      "learning_rate": 3.022666288215877e-05,
      "loss": 2.2142,
      "step": 197198
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.345589518547058,
      "learning_rate": 3.0224873505290226e-05,
      "loss": 1.9001,
      "step": 197199
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.5020521879196167,
      "learning_rate": 3.0223084178578027e-05,
      "loss": 2.0633,
      "step": 197200
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.48814058303833,
      "learning_rate": 3.022129490202241e-05,
      "loss": 1.9282,
      "step": 197201
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.587382197380066,
      "learning_rate": 3.0219505675623767e-05,
      "loss": 1.9098,
      "step": 197202
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.3935672044754028,
      "learning_rate": 3.0217716499382437e-05,
      "loss": 1.8516,
      "step": 197203
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.5017763376235962,
      "learning_rate": 3.021592737329872e-05,
      "loss": 1.9864,
      "step": 197204
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.4608198404312134,
      "learning_rate": 3.0214138297372915e-05,
      "loss": 1.9585,
      "step": 197205
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.5080516338348389,
      "learning_rate": 3.021234927160545e-05,
      "loss": 1.9723,
      "step": 197206
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.384405493736267,
      "learning_rate": 3.0210560295996533e-05,
      "loss": 1.9066,
      "step": 197207
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.4015644788742065,
      "learning_rate": 3.0208771370546626e-05,
      "loss": 2.1052,
      "step": 197208
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.520128607749939,
      "learning_rate": 3.0206982495255995e-05,
      "loss": 2.0277,
      "step": 197209
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.4452158212661743,
      "learning_rate": 3.020519367012497e-05,
      "loss": 2.1356,
      "step": 197210
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.4427366256713867,
      "learning_rate": 3.020340489515386e-05,
      "loss": 1.9992,
      "step": 197211
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.4606488943099976,
      "learning_rate": 3.020161617034306e-05,
      "loss": 2.0995,
      "step": 197212
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.518552303314209,
      "learning_rate": 3.019982749569283e-05,
      "loss": 2.1517,
      "step": 197213
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.3543751239776611,
      "learning_rate": 3.019803887120361e-05,
      "loss": 1.9103,
      "step": 197214
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.44076406955719,
      "learning_rate": 3.0196250296875667e-05,
      "loss": 2.2437,
      "step": 197215
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.3817379474639893,
      "learning_rate": 3.01944617727093e-05,
      "loss": 1.9173,
      "step": 197216
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.325599193572998,
      "learning_rate": 3.019267329870484e-05,
      "loss": 2.0357,
      "step": 197217
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.4181429147720337,
      "learning_rate": 3.019088487486272e-05,
      "loss": 2.1314,
      "step": 197218
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.4365572929382324,
      "learning_rate": 3.0189096501183175e-05,
      "loss": 1.848,
      "step": 197219
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.4711846113204956,
      "learning_rate": 3.018730817766657e-05,
      "loss": 1.8766,
      "step": 197220
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.3494843244552612,
      "learning_rate": 3.0185519904313272e-05,
      "loss": 1.8171,
      "step": 197221
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.3908897638320923,
      "learning_rate": 3.018373168112358e-05,
      "loss": 2.218,
      "step": 197222
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.447833776473999,
      "learning_rate": 3.0181943508097763e-05,
      "loss": 2.0318,
      "step": 197223
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.4300686120986938,
      "learning_rate": 3.0180155385236283e-05,
      "loss": 1.9591,
      "step": 197224
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.399451494216919,
      "learning_rate": 3.017836731253931e-05,
      "loss": 1.8218,
      "step": 197225
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.4713891744613647,
      "learning_rate": 3.0176579290007375e-05,
      "loss": 2.1123,
      "step": 197226
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.6216614246368408,
      "learning_rate": 3.017479131764068e-05,
      "loss": 1.8758,
      "step": 197227
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.553703784942627,
      "learning_rate": 3.0173003395439587e-05,
      "loss": 2.0936,
      "step": 197228
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.526947021484375,
      "learning_rate": 3.017121552340437e-05,
      "loss": 1.9234,
      "step": 197229
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.3369724750518799,
      "learning_rate": 3.0169427701535486e-05,
      "loss": 1.9756,
      "step": 197230
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.4345808029174805,
      "learning_rate": 3.0167639929833144e-05,
      "loss": 2.1543,
      "step": 197231
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.4433151483535767,
      "learning_rate": 3.016585220829777e-05,
      "loss": 2.0844,
      "step": 197232
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.4470995664596558,
      "learning_rate": 3.0164064536929665e-05,
      "loss": 2.0459,
      "step": 197233
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.6018171310424805,
      "learning_rate": 3.0162276915729168e-05,
      "loss": 1.9322,
      "step": 197234
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.399956464767456,
      "learning_rate": 3.0160489344696538e-05,
      "loss": 2.1016,
      "step": 197235
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.4874824285507202,
      "learning_rate": 3.015870182383221e-05,
      "loss": 2.0719,
      "step": 197236
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.4870638847351074,
      "learning_rate": 3.015691435313642e-05,
      "loss": 2.0803,
      "step": 197237
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.458685278892517,
      "learning_rate": 3.0155126932609597e-05,
      "loss": 1.8111,
      "step": 197238
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.3402189016342163,
      "learning_rate": 3.0153339562252042e-05,
      "loss": 1.8114,
      "step": 197239
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.4771229028701782,
      "learning_rate": 3.0151552242064093e-05,
      "loss": 2.1458,
      "step": 197240
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.354335904121399,
      "learning_rate": 3.0149764972045976e-05,
      "loss": 2.035,
      "step": 197241
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.394515037536621,
      "learning_rate": 3.0147977752198194e-05,
      "loss": 2.2564,
      "step": 197242
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.2976760864257812,
      "learning_rate": 3.014619058252091e-05,
      "loss": 1.8701,
      "step": 197243
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.3971117734909058,
      "learning_rate": 3.0144403463014633e-05,
      "loss": 1.967,
      "step": 197244
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.4001679420471191,
      "learning_rate": 3.014261639367952e-05,
      "loss": 1.8739,
      "step": 197245
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.403059959411621,
      "learning_rate": 3.0140829374516106e-05,
      "loss": 1.9196,
      "step": 197246
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.4446980953216553,
      "learning_rate": 3.0139042405524527e-05,
      "loss": 2.1056,
      "step": 197247
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.3773548603057861,
      "learning_rate": 3.0137255486705213e-05,
      "loss": 2.1089,
      "step": 197248
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.39126455783844,
      "learning_rate": 3.0135468618058433e-05,
      "loss": 2.0741,
      "step": 197249
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.4403489828109741,
      "learning_rate": 3.013368179958462e-05,
      "loss": 1.9693,
      "step": 197250
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.4860036373138428,
      "learning_rate": 3.0131895031284005e-05,
      "loss": 1.9097,
      "step": 197251
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.4079689979553223,
      "learning_rate": 3.0130108313157052e-05,
      "loss": 2.1483,
      "step": 197252
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.4460936784744263,
      "learning_rate": 3.01283216452039e-05,
      "loss": 2.092,
      "step": 197253
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.4135717153549194,
      "learning_rate": 3.0126535027425046e-05,
      "loss": 2.2533,
      "step": 197254
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.6035569906234741,
      "learning_rate": 3.0124748459820724e-05,
      "loss": 2.2208,
      "step": 197255
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.4868507385253906,
      "learning_rate": 3.0122961942391367e-05,
      "loss": 1.8841,
      "step": 197256
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.512266755104065,
      "learning_rate": 3.0121175475137173e-05,
      "loss": 2.0179,
      "step": 197257
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.44899582862854,
      "learning_rate": 3.011938905805864e-05,
      "loss": 2.2298,
      "step": 197258
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.409500241279602,
      "learning_rate": 3.0117602691155907e-05,
      "loss": 2.0526,
      "step": 197259
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.3892189264297485,
      "learning_rate": 3.011581637442947e-05,
      "loss": 1.9797,
      "step": 197260
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.4910321235656738,
      "learning_rate": 3.011403010787956e-05,
      "loss": 1.8521,
      "step": 197261
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.4615405797958374,
      "learning_rate": 3.0112243891506582e-05,
      "loss": 1.9282,
      "step": 197262
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.464499831199646,
      "learning_rate": 3.0110457725310767e-05,
      "loss": 2.0198,
      "step": 197263
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.473960280418396,
      "learning_rate": 3.0108671609292577e-05,
      "loss": 2.0432,
      "step": 197264
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.5032157897949219,
      "learning_rate": 3.0106885543452287e-05,
      "loss": 2.0558,
      "step": 197265
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.4289252758026123,
      "learning_rate": 3.0105099527790222e-05,
      "loss": 1.9212,
      "step": 197266
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.4018731117248535,
      "learning_rate": 3.0103313562306653e-05,
      "loss": 1.9425,
      "step": 197267
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.4219720363616943,
      "learning_rate": 3.010152764700201e-05,
      "loss": 1.8264,
      "step": 197268
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.6466021537780762,
      "learning_rate": 3.0099741781876562e-05,
      "loss": 1.8669,
      "step": 197269
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.4040547609329224,
      "learning_rate": 3.009795596693071e-05,
      "loss": 1.7993,
      "step": 197270
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.4251877069473267,
      "learning_rate": 3.0096170202164714e-05,
      "loss": 2.1222,
      "step": 197271
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.4374148845672607,
      "learning_rate": 3.0094384487578914e-05,
      "loss": 1.9312,
      "step": 197272
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.3182998895645142,
      "learning_rate": 3.0092598823173708e-05,
      "loss": 2.1459,
      "step": 197273
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.4358893632888794,
      "learning_rate": 3.0090813208949393e-05,
      "loss": 1.7934,
      "step": 197274
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.5291173458099365,
      "learning_rate": 3.0089027644906237e-05,
      "loss": 2.1187,
      "step": 197275
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.3679031133651733,
      "learning_rate": 3.0087242131044674e-05,
      "loss": 2.0969,
      "step": 197276
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.5077893733978271,
      "learning_rate": 3.0085456667364972e-05,
      "loss": 2.0203,
      "step": 197277
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.3781803846359253,
      "learning_rate": 3.0083671253867425e-05,
      "loss": 1.8513,
      "step": 197278
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.3833732604980469,
      "learning_rate": 3.0081885890552504e-05,
      "loss": 2.0076,
      "step": 197279
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.3133575916290283,
      "learning_rate": 3.008010057742044e-05,
      "loss": 2.0112,
      "step": 197280
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.3337767124176025,
      "learning_rate": 3.0078315314471503e-05,
      "loss": 2.0616,
      "step": 197281
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.486092448234558,
      "learning_rate": 3.0076530101706188e-05,
      "loss": 2.1144,
      "step": 197282
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.404960036277771,
      "learning_rate": 3.007474493912473e-05,
      "loss": 1.8274,
      "step": 197283
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.350860357284546,
      "learning_rate": 3.007295982672743e-05,
      "loss": 1.8976,
      "step": 197284
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.3704599142074585,
      "learning_rate": 3.0071174764514682e-05,
      "loss": 1.9577,
      "step": 197285
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.4514944553375244,
      "learning_rate": 3.0069389752486795e-05,
      "loss": 2.047,
      "step": 197286
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.3863043785095215,
      "learning_rate": 3.006760479064413e-05,
      "loss": 1.961,
      "step": 197287
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.7042063474655151,
      "learning_rate": 3.0065819878986985e-05,
      "loss": 2.0102,
      "step": 197288
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.4256075620651245,
      "learning_rate": 3.0064035017515697e-05,
      "loss": 2.0295,
      "step": 197289
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.365590214729309,
      "learning_rate": 3.0062250206230566e-05,
      "loss": 1.8718,
      "step": 197290
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.5586841106414795,
      "learning_rate": 3.0060465445131988e-05,
      "loss": 1.8807,
      "step": 197291
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.485416293144226,
      "learning_rate": 3.005868073422023e-05,
      "loss": 2.1497,
      "step": 197292
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.4436326026916504,
      "learning_rate": 3.0056896073495696e-05,
      "loss": 1.8878,
      "step": 197293
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.524088740348816,
      "learning_rate": 3.0055111462958715e-05,
      "loss": 1.8063,
      "step": 197294
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.38200843334198,
      "learning_rate": 3.0053326902609554e-05,
      "loss": 2.004,
      "step": 197295
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.4167362451553345,
      "learning_rate": 3.0051542392448515e-05,
      "loss": 2.0627,
      "step": 197296
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.6173795461654663,
      "learning_rate": 3.0049757932476058e-05,
      "loss": 2.2824,
      "step": 197297
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.5064692497253418,
      "learning_rate": 3.0047973522692394e-05,
      "loss": 2.0664,
      "step": 197298
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.5749024152755737,
      "learning_rate": 3.0046189163097976e-05,
      "loss": 2.072,
      "step": 197299
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.296849012374878,
      "learning_rate": 3.004440485369305e-05,
      "loss": 2.0849,
      "step": 197300
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.429473638534546,
      "learning_rate": 3.0042620594477972e-05,
      "loss": 2.0585,
      "step": 197301
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.4253417253494263,
      "learning_rate": 3.0040836385452982e-05,
      "loss": 2.0381,
      "step": 197302
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.4545650482177734,
      "learning_rate": 3.0039052226618577e-05,
      "loss": 2.0105,
      "step": 197303
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.5406385660171509,
      "learning_rate": 3.0037268117974956e-05,
      "loss": 1.9713,
      "step": 197304
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.506426453590393,
      "learning_rate": 3.003548405952255e-05,
      "loss": 2.0607,
      "step": 197305
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.6465691328048706,
      "learning_rate": 3.0033700051261665e-05,
      "loss": 1.7861,
      "step": 197306
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.455830454826355,
      "learning_rate": 3.0031916093192598e-05,
      "loss": 1.87,
      "step": 197307
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.4644207954406738,
      "learning_rate": 3.0030132185315613e-05,
      "loss": 1.9235,
      "step": 197308
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.7888362407684326,
      "learning_rate": 3.0028348327631212e-05,
      "loss": 2.0571,
      "step": 197309
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.5096381902694702,
      "learning_rate": 3.002656452013956e-05,
      "loss": 2.0193,
      "step": 197310
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.40959894657135,
      "learning_rate": 3.0024780762841128e-05,
      "loss": 1.8268,
      "step": 197311
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.382615327835083,
      "learning_rate": 3.0022997055736142e-05,
      "loss": 2.0303,
      "step": 197312
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.4705592393875122,
      "learning_rate": 3.0021213398825072e-05,
      "loss": 2.2262,
      "step": 197313
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.411836862564087,
      "learning_rate": 3.0019429792108053e-05,
      "loss": 1.9129,
      "step": 197314
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.2945103645324707,
      "learning_rate": 3.0017646235585546e-05,
      "loss": 1.8772,
      "step": 197315
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.3513442277908325,
      "learning_rate": 3.001586272925782e-05,
      "loss": 1.8583,
      "step": 197316
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.3963898420333862,
      "learning_rate": 3.0014079273125313e-05,
      "loss": 1.9818,
      "step": 197317
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.406165599822998,
      "learning_rate": 3.001229586718822e-05,
      "loss": 1.887,
      "step": 197318
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.3740496635437012,
      "learning_rate": 3.0010512511447038e-05,
      "loss": 1.7579,
      "step": 197319
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.3901575803756714,
      "learning_rate": 3.0008729205901904e-05,
      "loss": 2.1395,
      "step": 197320
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.4587126970291138,
      "learning_rate": 3.0006945950553284e-05,
      "loss": 1.9964,
      "step": 197321
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.3252286911010742,
      "learning_rate": 3.0005162745401412e-05,
      "loss": 1.8304,
      "step": 197322
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.4488807916641235,
      "learning_rate": 3.0003379590446753e-05,
      "loss": 2.0766,
      "step": 197323
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.4952266216278076,
      "learning_rate": 3.0001596485689505e-05,
      "loss": 2.1533,
      "step": 197324
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.651964545249939,
      "learning_rate": 2.9999813431130137e-05,
      "loss": 2.1211,
      "step": 197325
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.3803621530532837,
      "learning_rate": 2.9998030426768814e-05,
      "loss": 2.0673,
      "step": 197326
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.5739617347717285,
      "learning_rate": 2.9996247472606005e-05,
      "loss": 2.1563,
      "step": 197327
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.359518051147461,
      "learning_rate": 2.999446456864194e-05,
      "loss": 2.012,
      "step": 197328
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.3574553728103638,
      "learning_rate": 2.9992681714877054e-05,
      "loss": 1.9704,
      "step": 197329
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.5424094200134277,
      "learning_rate": 2.9990898911311546e-05,
      "loss": 2.1487,
      "step": 197330
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.5098706483840942,
      "learning_rate": 2.998911615794598e-05,
      "loss": 1.9942,
      "step": 197331
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.4897994995117188,
      "learning_rate": 2.9987333454780392e-05,
      "loss": 1.9396,
      "step": 197332
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.4047062397003174,
      "learning_rate": 2.9985550801815316e-05,
      "loss": 1.9068,
      "step": 197333
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.593912959098816,
      "learning_rate": 2.9983768199050982e-05,
      "loss": 2.0909,
      "step": 197334
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.4492679834365845,
      "learning_rate": 2.9981985646487794e-05,
      "loss": 2.1211,
      "step": 197335
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.3863189220428467,
      "learning_rate": 2.9980203144126013e-05,
      "loss": 1.8314,
      "step": 197336
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.611864447593689,
      "learning_rate": 2.997842069196611e-05,
      "loss": 2.0122,
      "step": 197337
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.5039108991622925,
      "learning_rate": 2.9976638290008215e-05,
      "loss": 1.9823,
      "step": 197338
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.4511805772781372,
      "learning_rate": 2.997485593825283e-05,
      "loss": 1.9571,
      "step": 197339
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.4792546033859253,
      "learning_rate": 2.9973073636700117e-05,
      "loss": 2.056,
      "step": 197340
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.4438992738723755,
      "learning_rate": 2.9971291385350583e-05,
      "loss": 1.9325,
      "step": 197341
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.7899349927902222,
      "learning_rate": 2.9969509184204455e-05,
      "loss": 2.0867,
      "step": 197342
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.4007810354232788,
      "learning_rate": 2.996772703326217e-05,
      "loss": 1.8476,
      "step": 197343
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.3266652822494507,
      "learning_rate": 2.996594493252389e-05,
      "loss": 2.0294,
      "step": 197344
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.4263020753860474,
      "learning_rate": 2.9964162881990085e-05,
      "loss": 1.8846,
      "step": 197345
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.463643193244934,
      "learning_rate": 2.9962380881660987e-05,
      "loss": 2.0429,
      "step": 197346
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.3987704515457153,
      "learning_rate": 2.9960598931537027e-05,
      "loss": 1.889,
      "step": 197347
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.3770824670791626,
      "learning_rate": 2.9958817031618444e-05,
      "loss": 1.8895,
      "step": 197348
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.3802305459976196,
      "learning_rate": 2.9957035181905666e-05,
      "loss": 2.2354,
      "step": 197349
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.4499417543411255,
      "learning_rate": 2.995525338239896e-05,
      "loss": 2.1134,
      "step": 197350
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.467984676361084,
      "learning_rate": 2.995347163309869e-05,
      "loss": 2.1246,
      "step": 197351
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.423774003982544,
      "learning_rate": 2.9951689934005097e-05,
      "loss": 1.8555,
      "step": 197352
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.704485535621643,
      "learning_rate": 2.994990828511864e-05,
      "loss": 1.8961,
      "step": 197353
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.3054602146148682,
      "learning_rate": 2.9948126686439556e-05,
      "loss": 2.0074,
      "step": 197354
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.410434603691101,
      "learning_rate": 2.994634513796824e-05,
      "loss": 2.0105,
      "step": 197355
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.3505021333694458,
      "learning_rate": 2.994456363970503e-05,
      "loss": 1.8366,
      "step": 197356
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.3247977495193481,
      "learning_rate": 2.9942782191650195e-05,
      "loss": 1.7774,
      "step": 197357
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.4018306732177734,
      "learning_rate": 2.9941000793804025e-05,
      "loss": 1.8459,
      "step": 197358
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.4030144214630127,
      "learning_rate": 2.9939219446167027e-05,
      "loss": 1.85,
      "step": 197359
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.4323707818984985,
      "learning_rate": 2.9937438148739334e-05,
      "loss": 2.0539,
      "step": 197360
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.269987940788269,
      "learning_rate": 2.9935656901521443e-05,
      "loss": 2.1158,
      "step": 197361
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.5060856342315674,
      "learning_rate": 2.993387570451359e-05,
      "loss": 2.1366,
      "step": 197362
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.3931381702423096,
      "learning_rate": 2.993209455771607e-05,
      "loss": 1.982,
      "step": 197363
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.5455704927444458,
      "learning_rate": 2.993031346112935e-05,
      "loss": 2.0874,
      "step": 197364
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.4036022424697876,
      "learning_rate": 2.992853241475367e-05,
      "loss": 1.9653,
      "step": 197365
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.6170891523361206,
      "learning_rate": 2.9926751418589322e-05,
      "loss": 2.0996,
      "step": 197366
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.566086769104004,
      "learning_rate": 2.9924970472636745e-05,
      "loss": 1.925,
      "step": 197367
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.448004961013794,
      "learning_rate": 2.9923189576896202e-05,
      "loss": 1.8337,
      "step": 197368
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.5226155519485474,
      "learning_rate": 2.992140873136799e-05,
      "loss": 2.1367,
      "step": 197369
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.446042537689209,
      "learning_rate": 2.9919627936052547e-05,
      "loss": 2.0369,
      "step": 197370
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.426722526550293,
      "learning_rate": 2.991784719095007e-05,
      "loss": 2.0457,
      "step": 197371
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.4202454090118408,
      "learning_rate": 2.9916066496061064e-05,
      "loss": 2.0209,
      "step": 197372
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.400612235069275,
      "learning_rate": 2.9914285851385722e-05,
      "loss": 1.7401,
      "step": 197373
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.5510228872299194,
      "learning_rate": 2.9912505256924414e-05,
      "loss": 1.8414,
      "step": 197374
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.4203150272369385,
      "learning_rate": 2.99107247126774e-05,
      "loss": 2.075,
      "step": 197375
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.4063429832458496,
      "learning_rate": 2.9908944218645158e-05,
      "loss": 1.9127,
      "step": 197376
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.4316656589508057,
      "learning_rate": 2.990716377482788e-05,
      "loss": 2.0987,
      "step": 197377
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.4038803577423096,
      "learning_rate": 2.9905383381226033e-05,
      "loss": 1.8015,
      "step": 197378
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.278852939605713,
      "learning_rate": 2.9903603037839853e-05,
      "loss": 1.8938,
      "step": 197379
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.4647433757781982,
      "learning_rate": 2.9901822744669702e-05,
      "loss": 2.0973,
      "step": 197380
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.3377439975738525,
      "learning_rate": 2.990004250171585e-05,
      "loss": 2.0627,
      "step": 197381
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.3385405540466309,
      "learning_rate": 2.9898262308978694e-05,
      "loss": 1.8731,
      "step": 197382
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.419184684753418,
      "learning_rate": 2.9896482166458537e-05,
      "loss": 1.9873,
      "step": 197383
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.3174710273742676,
      "learning_rate": 2.9894702074155775e-05,
      "loss": 1.9199,
      "step": 197384
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.3087700605392456,
      "learning_rate": 2.989292203207061e-05,
      "loss": 1.9907,
      "step": 197385
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.3045092821121216,
      "learning_rate": 2.9891142040203576e-05,
      "loss": 2.0947,
      "step": 197386
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.4756824970245361,
      "learning_rate": 2.988936209855477e-05,
      "loss": 2.2325,
      "step": 197387
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.4202296733856201,
      "learning_rate": 2.988758220712466e-05,
      "loss": 2.1186,
      "step": 197388
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.4125713109970093,
      "learning_rate": 2.9885802365913513e-05,
      "loss": 2.0291,
      "step": 197389
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.5406014919281006,
      "learning_rate": 2.988402257492173e-05,
      "loss": 2.131,
      "step": 197390
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.4216055870056152,
      "learning_rate": 2.988224283414957e-05,
      "loss": 1.8119,
      "step": 197391
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.4188967943191528,
      "learning_rate": 2.988046314359751e-05,
      "loss": 1.9557,
      "step": 197392
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.4015734195709229,
      "learning_rate": 2.9878683503265675e-05,
      "loss": 1.9913,
      "step": 197393
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.415432095527649,
      "learning_rate": 2.9876903913154504e-05,
      "loss": 2.1749,
      "step": 197394
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.339465618133545,
      "learning_rate": 2.9875124373264293e-05,
      "loss": 2.1572,
      "step": 197395
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.4788708686828613,
      "learning_rate": 2.987334488359544e-05,
      "loss": 2.298,
      "step": 197396
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.4863611459732056,
      "learning_rate": 2.987156544414818e-05,
      "loss": 2.0768,
      "step": 197397
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.4748893976211548,
      "learning_rate": 2.9869786054923017e-05,
      "loss": 2.0578,
      "step": 197398
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.410487174987793,
      "learning_rate": 2.9868006715920046e-05,
      "loss": 1.7568,
      "step": 197399
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.3520365953445435,
      "learning_rate": 2.9866227427139734e-05,
      "loss": 1.9958,
      "step": 197400
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.381231427192688,
      "learning_rate": 2.9864448188582378e-05,
      "loss": 1.9988,
      "step": 197401
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.4970570802688599,
      "learning_rate": 2.9862669000248353e-05,
      "loss": 2.0117,
      "step": 197402
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.4290015697479248,
      "learning_rate": 2.9860889862137883e-05,
      "loss": 2.0955,
      "step": 197403
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.5172299146652222,
      "learning_rate": 2.9859110774251503e-05,
      "loss": 1.8622,
      "step": 197404
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.460015058517456,
      "learning_rate": 2.985733173658932e-05,
      "loss": 1.8311,
      "step": 197405
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.4657402038574219,
      "learning_rate": 2.985555274915179e-05,
      "loss": 1.9332,
      "step": 197406
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.4070450067520142,
      "learning_rate": 2.9853773811939185e-05,
      "loss": 1.907,
      "step": 197407
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.3582464456558228,
      "learning_rate": 2.9851994924951872e-05,
      "loss": 1.9211,
      "step": 197408
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.4699074029922485,
      "learning_rate": 2.9850216088190148e-05,
      "loss": 2.0568,
      "step": 197409
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.524678349494934,
      "learning_rate": 2.9848437301654482e-05,
      "loss": 1.9458,
      "step": 197410
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.3896840810775757,
      "learning_rate": 2.9846658565344973e-05,
      "loss": 1.8164,
      "step": 197411
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.3870524168014526,
      "learning_rate": 2.984487987926212e-05,
      "loss": 2.0856,
      "step": 197412
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.4497891664505005,
      "learning_rate": 2.9843101243406154e-05,
      "loss": 2.1021,
      "step": 197413
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.3799270391464233,
      "learning_rate": 2.984132265777751e-05,
      "loss": 1.9707,
      "step": 197414
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.380974292755127,
      "learning_rate": 2.9839544122376425e-05,
      "loss": 1.9649,
      "step": 197415
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.4845794439315796,
      "learning_rate": 2.983776563720336e-05,
      "loss": 2.0802,
      "step": 197416
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.3517287969589233,
      "learning_rate": 2.9835987202258415e-05,
      "loss": 1.9188,
      "step": 197417
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.5126500129699707,
      "learning_rate": 2.983420881754216e-05,
      "loss": 1.92,
      "step": 197418
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.4133349657058716,
      "learning_rate": 2.983243048305476e-05,
      "loss": 1.8863,
      "step": 197419
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.4542531967163086,
      "learning_rate": 2.9830652198796645e-05,
      "loss": 2.1369,
      "step": 197420
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.4188978672027588,
      "learning_rate": 2.9828873964768085e-05,
      "loss": 1.7493,
      "step": 197421
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.3795976638793945,
      "learning_rate": 2.9827095780969545e-05,
      "loss": 1.7917,
      "step": 197422
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.7064430713653564,
      "learning_rate": 2.9825317647401126e-05,
      "loss": 1.8876,
      "step": 197423
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.4213528633117676,
      "learning_rate": 2.9823539564063326e-05,
      "loss": 1.8936,
      "step": 197424
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.4345792531967163,
      "learning_rate": 2.982176153095638e-05,
      "loss": 1.9967,
      "step": 197425
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.499293327331543,
      "learning_rate": 2.9819983548080752e-05,
      "loss": 1.831,
      "step": 197426
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.4429117441177368,
      "learning_rate": 2.9818205615436608e-05,
      "loss": 2.2723,
      "step": 197427
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.4520821571350098,
      "learning_rate": 2.9816427733024483e-05,
      "loss": 2.0746,
      "step": 197428
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.3954041004180908,
      "learning_rate": 2.9814649900844444e-05,
      "loss": 1.9263,
      "step": 197429
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.4383436441421509,
      "learning_rate": 2.9812872118897057e-05,
      "loss": 1.9432,
      "step": 197430
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.4234918355941772,
      "learning_rate": 2.9811094387182453e-05,
      "loss": 2.2934,
      "step": 197431
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.66958487033844,
      "learning_rate": 2.980931670570117e-05,
      "loss": 1.8875,
      "step": 197432
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.4941562414169312,
      "learning_rate": 2.9807539074453368e-05,
      "loss": 1.8632,
      "step": 197433
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.4716265201568604,
      "learning_rate": 2.9805761493439482e-05,
      "loss": 1.8917,
      "step": 197434
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.4611073732376099,
      "learning_rate": 2.9803983962659817e-05,
      "loss": 1.9322,
      "step": 197435
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.398194670677185,
      "learning_rate": 2.9802206482114698e-05,
      "loss": 2.2086,
      "step": 197436
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.4963451623916626,
      "learning_rate": 2.98004290518044e-05,
      "loss": 1.8796,
      "step": 197437
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.3910834789276123,
      "learning_rate": 2.9798651671729344e-05,
      "loss": 1.9776,
      "step": 197438
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.378557801246643,
      "learning_rate": 2.9796874341889743e-05,
      "loss": 2.2023,
      "step": 197439
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.5461852550506592,
      "learning_rate": 2.9795097062286087e-05,
      "loss": 1.8527,
      "step": 197440
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.4779272079467773,
      "learning_rate": 2.9793319832918617e-05,
      "loss": 2.0897,
      "step": 197441
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.3399165868759155,
      "learning_rate": 2.9791542653787658e-05,
      "loss": 1.9493,
      "step": 197442
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.5817439556121826,
      "learning_rate": 2.9789765524893515e-05,
      "loss": 1.9387,
      "step": 197443
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.3509039878845215,
      "learning_rate": 2.9787988446236622e-05,
      "loss": 1.9683,
      "step": 197444
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.546836495399475,
      "learning_rate": 2.9786211417817175e-05,
      "loss": 1.7669,
      "step": 197445
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.4145276546478271,
      "learning_rate": 2.9784434439635607e-05,
      "loss": 2.1093,
      "step": 197446
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.4767730236053467,
      "learning_rate": 2.978265751169222e-05,
      "loss": 2.2147,
      "step": 197447
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.4302830696105957,
      "learning_rate": 2.978088063398728e-05,
      "loss": 1.8783,
      "step": 197448
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.6442070007324219,
      "learning_rate": 2.9779103806521254e-05,
      "loss": 1.7615,
      "step": 197449
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.4687801599502563,
      "learning_rate": 2.9777327029294374e-05,
      "loss": 2.0921,
      "step": 197450
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.6419391632080078,
      "learning_rate": 2.977555030230694e-05,
      "loss": 1.9328,
      "step": 197451
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.3821276426315308,
      "learning_rate": 2.9773773625559384e-05,
      "loss": 2.0535,
      "step": 197452
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.3798202276229858,
      "learning_rate": 2.9771996999052005e-05,
      "loss": 1.7944,
      "step": 197453
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.5265495777130127,
      "learning_rate": 2.977022042278504e-05,
      "loss": 1.9686,
      "step": 197454
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.4855655431747437,
      "learning_rate": 2.9768443896758954e-05,
      "loss": 2.0214,
      "step": 197455
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.7097244262695312,
      "learning_rate": 2.9766667420973946e-05,
      "loss": 2.0487,
      "step": 197456
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.3409909009933472,
      "learning_rate": 2.976489099543048e-05,
      "loss": 2.1229,
      "step": 197457
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.4999232292175293,
      "learning_rate": 2.9763114620128824e-05,
      "loss": 2.1627,
      "step": 197458
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.644059181213379,
      "learning_rate": 2.976133829506928e-05,
      "loss": 2.2061,
      "step": 197459
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.4952478408813477,
      "learning_rate": 2.975956202025218e-05,
      "loss": 2.0927,
      "step": 197460
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.391913652420044,
      "learning_rate": 2.9757785795677923e-05,
      "loss": 2.0238,
      "step": 197461
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.3457046747207642,
      "learning_rate": 2.9756009621346743e-05,
      "loss": 2.0623,
      "step": 197462
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.6879326105117798,
      "learning_rate": 2.975423349725907e-05,
      "loss": 1.9999,
      "step": 197463
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.377982497215271,
      "learning_rate": 2.9752457423415212e-05,
      "loss": 1.867,
      "step": 197464
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.508705735206604,
      "learning_rate": 2.9750681399815456e-05,
      "loss": 2.2913,
      "step": 197465
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.3906320333480835,
      "learning_rate": 2.974890542646008e-05,
      "loss": 1.912,
      "step": 197466
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.6219850778579712,
      "learning_rate": 2.9747129503349544e-05,
      "loss": 2.1761,
      "step": 197467
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.4168024063110352,
      "learning_rate": 2.974535363048408e-05,
      "loss": 1.8983,
      "step": 197468
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.307330846786499,
      "learning_rate": 2.9743577807864095e-05,
      "loss": 2.1026,
      "step": 197469
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.426070213317871,
      "learning_rate": 2.9741802035489846e-05,
      "loss": 1.8058,
      "step": 197470
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.3811559677124023,
      "learning_rate": 2.9740026313361774e-05,
      "loss": 1.8442,
      "step": 197471
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.3291102647781372,
      "learning_rate": 2.9738250641480043e-05,
      "loss": 2.0831,
      "step": 197472
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.4250797033309937,
      "learning_rate": 2.9736475019845152e-05,
      "loss": 2.0154,
      "step": 197473
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.43016517162323,
      "learning_rate": 2.9734699448457266e-05,
      "loss": 2.0585,
      "step": 197474
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.4940797090530396,
      "learning_rate": 2.9732923927316855e-05,
      "loss": 2.0153,
      "step": 197475
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.472663164138794,
      "learning_rate": 2.9731148456424146e-05,
      "loss": 2.0321,
      "step": 197476
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.405911922454834,
      "learning_rate": 2.9729373035779647e-05,
      "loss": 2.0764,
      "step": 197477
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.6123175621032715,
      "learning_rate": 2.972759766538342e-05,
      "loss": 2.0976,
      "step": 197478
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.3864130973815918,
      "learning_rate": 2.9725822345235995e-05,
      "loss": 2.0154,
      "step": 197479
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.3781611919403076,
      "learning_rate": 2.9724047075337608e-05,
      "loss": 1.9963,
      "step": 197480
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.4198455810546875,
      "learning_rate": 2.972227185568866e-05,
      "loss": 1.9344,
      "step": 197481
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.461679220199585,
      "learning_rate": 2.9720496686289384e-05,
      "loss": 2.0142,
      "step": 197482
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.456590175628662,
      "learning_rate": 2.971872156714028e-05,
      "loss": 1.876,
      "step": 197483
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.3219044208526611,
      "learning_rate": 2.971694649824148e-05,
      "loss": 2.0104,
      "step": 197484
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.3420007228851318,
      "learning_rate": 2.9715171479593413e-05,
      "loss": 1.8404,
      "step": 197485
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.4416213035583496,
      "learning_rate": 2.9713396511196386e-05,
      "loss": 2.1249,
      "step": 197486
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.448946237564087,
      "learning_rate": 2.9711621593050795e-05,
      "loss": 2.0388,
      "step": 197487
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.448854923248291,
      "learning_rate": 2.970984672515684e-05,
      "loss": 2.0493,
      "step": 197488
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.3571182489395142,
      "learning_rate": 2.9708071907515018e-05,
      "loss": 2.0023,
      "step": 197489
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.5858914852142334,
      "learning_rate": 2.97062971401255e-05,
      "loss": 2.1714,
      "step": 197490
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.3914896249771118,
      "learning_rate": 2.9704522422988718e-05,
      "loss": 1.8172,
      "step": 197491
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.4410187005996704,
      "learning_rate": 2.9702747756104907e-05,
      "loss": 1.8773,
      "step": 197492
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.5757548809051514,
      "learning_rate": 2.9700973139474527e-05,
      "loss": 2.0311,
      "step": 197493
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.4450311660766602,
      "learning_rate": 2.9699198573097782e-05,
      "loss": 1.8267,
      "step": 197494
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.2887629270553589,
      "learning_rate": 2.9697424056975173e-05,
      "loss": 2.0288,
      "step": 197495
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.399118185043335,
      "learning_rate": 2.969564959110676e-05,
      "loss": 1.9352,
      "step": 197496
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.4342403411865234,
      "learning_rate": 2.9693875175493153e-05,
      "loss": 2.1056,
      "step": 197497
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.3787482976913452,
      "learning_rate": 2.9692100810134444e-05,
      "loss": 1.7096,
      "step": 197498
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.420433521270752,
      "learning_rate": 2.9690326495031168e-05,
      "loss": 1.9804,
      "step": 197499
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.5194323062896729,
      "learning_rate": 2.9688552230183495e-05,
      "loss": 1.9275,
      "step": 197500
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.352971076965332,
      "learning_rate": 2.9686778015591916e-05,
      "loss": 1.8733,
      "step": 197501
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.3612749576568604,
      "learning_rate": 2.968500385125657e-05,
      "loss": 2.2454,
      "step": 197502
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.4305243492126465,
      "learning_rate": 2.968322973717796e-05,
      "loss": 2.0584,
      "step": 197503
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.5532119274139404,
      "learning_rate": 2.968145567335625e-05,
      "loss": 2.0664,
      "step": 197504
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.4594476222991943,
      "learning_rate": 2.967968165979193e-05,
      "loss": 1.7635,
      "step": 197505
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.43149733543396,
      "learning_rate": 2.9677907696485216e-05,
      "loss": 2.042,
      "step": 197506
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.3352268934249878,
      "learning_rate": 2.9676133783436594e-05,
      "loss": 2.0279,
      "step": 197507
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.3342055082321167,
      "learning_rate": 2.967435992064614e-05,
      "loss": 1.8787,
      "step": 197508
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.4834202527999878,
      "learning_rate": 2.9672586108114414e-05,
      "loss": 2.127,
      "step": 197509
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.4983360767364502,
      "learning_rate": 2.9670812345841587e-05,
      "loss": 1.9187,
      "step": 197510
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.3207225799560547,
      "learning_rate": 2.9669038633828125e-05,
      "loss": 1.8688,
      "step": 197511
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.3536568880081177,
      "learning_rate": 2.9667264972074222e-05,
      "loss": 2.1229,
      "step": 197512
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.4638891220092773,
      "learning_rate": 2.966549136058035e-05,
      "loss": 2.0287,
      "step": 197513
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.3392839431762695,
      "learning_rate": 2.9663717799346742e-05,
      "loss": 1.7449,
      "step": 197514
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.4090250730514526,
      "learning_rate": 2.966194428837376e-05,
      "loss": 2.0719,
      "step": 197515
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.4866834878921509,
      "learning_rate": 2.966017082766168e-05,
      "loss": 2.248,
      "step": 197516
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.2982062101364136,
      "learning_rate": 2.9658397417210922e-05,
      "loss": 1.8052,
      "step": 197517
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.4386632442474365,
      "learning_rate": 2.965662405702173e-05,
      "loss": 2.0468,
      "step": 197518
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.446850299835205,
      "learning_rate": 2.9654850747094528e-05,
      "loss": 2.0175,
      "step": 197519
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.549298644065857,
      "learning_rate": 2.9653077487429588e-05,
      "loss": 1.8053,
      "step": 197520
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.3217030763626099,
      "learning_rate": 2.965130427802721e-05,
      "loss": 1.9622,
      "step": 197521
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.338860273361206,
      "learning_rate": 2.964953111888776e-05,
      "loss": 1.9766,
      "step": 197522
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.6867177486419678,
      "learning_rate": 2.964775801001157e-05,
      "loss": 2.0191,
      "step": 197523
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.4922000169754028,
      "learning_rate": 2.9645984951398936e-05,
      "loss": 1.8093,
      "step": 197524
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.3747096061706543,
      "learning_rate": 2.9644211943050268e-05,
      "loss": 1.9918,
      "step": 197525
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.3829948902130127,
      "learning_rate": 2.9642438984965856e-05,
      "loss": 2.0238,
      "step": 197526
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.5146969556808472,
      "learning_rate": 2.9640666077146e-05,
      "loss": 2.0692,
      "step": 197527
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.47048020362854,
      "learning_rate": 2.9638893219591007e-05,
      "loss": 1.987,
      "step": 197528
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.5585095882415771,
      "learning_rate": 2.9637120412301306e-05,
      "loss": 2.075,
      "step": 197529
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.401668906211853,
      "learning_rate": 2.9635347655277098e-05,
      "loss": 2.1869,
      "step": 197530
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.4737290143966675,
      "learning_rate": 2.9633574948518847e-05,
      "loss": 2.0625,
      "step": 197531
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.4293467998504639,
      "learning_rate": 2.9631802292026818e-05,
      "loss": 1.7997,
      "step": 197532
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.2680349349975586,
      "learning_rate": 2.963002968580128e-05,
      "loss": 1.7491,
      "step": 197533
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.374422311782837,
      "learning_rate": 2.962825712984267e-05,
      "loss": 1.9251,
      "step": 197534
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.4110493659973145,
      "learning_rate": 2.9626484624151282e-05,
      "loss": 1.875,
      "step": 197535
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.5145395994186401,
      "learning_rate": 2.962471216872738e-05,
      "loss": 1.9989,
      "step": 197536
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.4806928634643555,
      "learning_rate": 2.9622939763571407e-05,
      "loss": 2.16,
      "step": 197537
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.4746612310409546,
      "learning_rate": 2.9621167408683622e-05,
      "loss": 1.984,
      "step": 197538
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.6886494159698486,
      "learning_rate": 2.9619395104064292e-05,
      "loss": 1.6906,
      "step": 197539
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.3466435670852661,
      "learning_rate": 2.9617622849713918e-05,
      "loss": 1.8568,
      "step": 197540
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.4876353740692139,
      "learning_rate": 2.96158506456327e-05,
      "loss": 2.0697,
      "step": 197541
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.5604695081710815,
      "learning_rate": 2.9614078491820968e-05,
      "loss": 1.7735,
      "step": 197542
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.4296681880950928,
      "learning_rate": 2.9612306388279128e-05,
      "loss": 1.9409,
      "step": 197543
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.3271815776824951,
      "learning_rate": 2.961053433500744e-05,
      "loss": 1.7044,
      "step": 197544
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.3214551210403442,
      "learning_rate": 2.9608762332006242e-05,
      "loss": 2.0224,
      "step": 197545
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.4471615552902222,
      "learning_rate": 2.9606990379275898e-05,
      "loss": 1.8689,
      "step": 197546
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.3954615592956543,
      "learning_rate": 2.9605218476816705e-05,
      "loss": 1.9213,
      "step": 197547
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.4468631744384766,
      "learning_rate": 2.9603446624629035e-05,
      "loss": 1.8734,
      "step": 197548
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.5695687532424927,
      "learning_rate": 2.9601674822713185e-05,
      "loss": 1.8463,
      "step": 197549
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.3713828325271606,
      "learning_rate": 2.9599903071069485e-05,
      "loss": 1.9566,
      "step": 197550
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.4972550868988037,
      "learning_rate": 2.959813136969824e-05,
      "loss": 1.8778,
      "step": 197551
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.6125239133834839,
      "learning_rate": 2.959635971859985e-05,
      "loss": 1.834,
      "step": 197552
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.6211508512496948,
      "learning_rate": 2.959458811777451e-05,
      "loss": 1.9287,
      "step": 197553
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.4159951210021973,
      "learning_rate": 2.9592816567222755e-05,
      "loss": 1.8988,
      "step": 197554
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.39076828956604,
      "learning_rate": 2.9591045066944685e-05,
      "loss": 2.0486,
      "step": 197555
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.5991307497024536,
      "learning_rate": 2.95892736169409e-05,
      "loss": 2.0913,
      "step": 197556
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.387514591217041,
      "learning_rate": 2.9587502217211435e-05,
      "loss": 1.9225,
      "step": 197557
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.3723119497299194,
      "learning_rate": 2.9585730867756817e-05,
      "loss": 2.122,
      "step": 197558
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.5081888437271118,
      "learning_rate": 2.9583959568577288e-05,
      "loss": 2.0913,
      "step": 197559
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.3781194686889648,
      "learning_rate": 2.958218831967324e-05,
      "loss": 2.3166,
      "step": 197560
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.417932391166687,
      "learning_rate": 2.9580417121044907e-05,
      "loss": 2.1237,
      "step": 197561
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.3395886421203613,
      "learning_rate": 2.9578645972692794e-05,
      "loss": 1.8426,
      "step": 197562
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.3809419870376587,
      "learning_rate": 2.957687487461703e-05,
      "loss": 1.9187,
      "step": 197563
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.5033894777297974,
      "learning_rate": 2.9575103826818048e-05,
      "loss": 1.9752,
      "step": 197564
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.4732149839401245,
      "learning_rate": 2.9573332829296116e-05,
      "loss": 2.0858,
      "step": 197565
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.637256383895874,
      "learning_rate": 2.9571561882051664e-05,
      "loss": 2.0753,
      "step": 197566
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.4590164422988892,
      "learning_rate": 2.9569790985084896e-05,
      "loss": 2.0118,
      "step": 197567
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.3320996761322021,
      "learning_rate": 2.956802013839634e-05,
      "loss": 1.8689,
      "step": 197568
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.5042284727096558,
      "learning_rate": 2.9566249341986072e-05,
      "loss": 2.1884,
      "step": 197569
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.4533418416976929,
      "learning_rate": 2.956447859585461e-05,
      "loss": 2.0901,
      "step": 197570
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.575216293334961,
      "learning_rate": 2.9562707900002136e-05,
      "loss": 2.1743,
      "step": 197571
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.644436240196228,
      "learning_rate": 2.956093725442914e-05,
      "loss": 2.0663,
      "step": 197572
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.3626184463500977,
      "learning_rate": 2.955916665913579e-05,
      "loss": 1.7543,
      "step": 197573
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.7682955265045166,
      "learning_rate": 2.955739611412262e-05,
      "loss": 1.9223,
      "step": 197574
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.3464821577072144,
      "learning_rate": 2.9555625619389733e-05,
      "loss": 2.0678,
      "step": 197575
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.460221767425537,
      "learning_rate": 2.9553855174937625e-05,
      "loss": 2.2803,
      "step": 197576
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.4240554571151733,
      "learning_rate": 2.955208478076646e-05,
      "loss": 2.2496,
      "step": 197577
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.4327503442764282,
      "learning_rate": 2.9550314436876777e-05,
      "loss": 2.016,
      "step": 197578
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.409786343574524,
      "learning_rate": 2.9548544143268705e-05,
      "loss": 2.1072,
      "step": 197579
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.9470669031143188,
      "learning_rate": 2.9546773899942744e-05,
      "loss": 1.9646,
      "step": 197580
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.416020154953003,
      "learning_rate": 2.954500370689913e-05,
      "loss": 1.953,
      "step": 197581
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.5627268552780151,
      "learning_rate": 2.9543233564138192e-05,
      "loss": 1.7438,
      "step": 197582
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.517984390258789,
      "learning_rate": 2.954146347166023e-05,
      "loss": 2.0936,
      "step": 197583
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.6512678861618042,
      "learning_rate": 2.953969342946565e-05,
      "loss": 1.97,
      "step": 197584
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.6801812648773193,
      "learning_rate": 2.953792343755471e-05,
      "loss": 1.7851,
      "step": 197585
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.4549365043640137,
      "learning_rate": 2.953615349592785e-05,
      "loss": 1.9409,
      "step": 197586
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.399860143661499,
      "learning_rate": 2.95343836045853e-05,
      "loss": 1.9448,
      "step": 197587
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.4311308860778809,
      "learning_rate": 2.9532613763527423e-05,
      "loss": 1.8872,
      "step": 197588
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.3939638137817383,
      "learning_rate": 2.953084397275446e-05,
      "loss": 1.9013,
      "step": 197589
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.3640238046646118,
      "learning_rate": 2.9529074232266904e-05,
      "loss": 1.9736,
      "step": 197590
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.4817631244659424,
      "learning_rate": 2.9527304542064923e-05,
      "loss": 2.1456,
      "step": 197591
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.3241949081420898,
      "learning_rate": 2.9525534902148985e-05,
      "loss": 1.9689,
      "step": 197592
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.462520956993103,
      "learning_rate": 2.9523765312519353e-05,
      "loss": 1.9511,
      "step": 197593
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.5930585861206055,
      "learning_rate": 2.9521995773176366e-05,
      "loss": 1.9564,
      "step": 197594
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.5815937519073486,
      "learning_rate": 2.9520226284120285e-05,
      "loss": 1.7453,
      "step": 197595
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.2919676303863525,
      "learning_rate": 2.9518456845351545e-05,
      "loss": 2.277,
      "step": 197596
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.50688636302948,
      "learning_rate": 2.9516687456870414e-05,
      "loss": 1.8854,
      "step": 197597
    },
    {
      "epoch": 2.57,
      "grad_norm": 2.1190237998962402,
      "learning_rate": 2.9514918118677254e-05,
      "loss": 1.8201,
      "step": 197598
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.3386613130569458,
      "learning_rate": 2.951314883077237e-05,
      "loss": 2.0007,
      "step": 197599
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.397844910621643,
      "learning_rate": 2.9511379593156125e-05,
      "loss": 1.8152,
      "step": 197600
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.6367405652999878,
      "learning_rate": 2.9509610405828756e-05,
      "loss": 2.1016,
      "step": 197601
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.3777683973312378,
      "learning_rate": 2.9507841268790687e-05,
      "loss": 1.8752,
      "step": 197602
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.4167667627334595,
      "learning_rate": 2.9506072182042195e-05,
      "loss": 2.3531,
      "step": 197603
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.5001310110092163,
      "learning_rate": 2.9504303145583676e-05,
      "loss": 1.8855,
      "step": 197604
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.397234559059143,
      "learning_rate": 2.9502534159415426e-05,
      "loss": 1.726,
      "step": 197605
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.4262137413024902,
      "learning_rate": 2.950076522353775e-05,
      "loss": 2.0439,
      "step": 197606
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.500693678855896,
      "learning_rate": 2.949899633795091e-05,
      "loss": 2.0974,
      "step": 197607
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.3864946365356445,
      "learning_rate": 2.9497227502655375e-05,
      "loss": 2.1347,
      "step": 197608
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.4265698194503784,
      "learning_rate": 2.9495458717651376e-05,
      "loss": 1.9595,
      "step": 197609
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.3893654346466064,
      "learning_rate": 2.9493689982939318e-05,
      "loss": 2.0476,
      "step": 197610
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.6430760622024536,
      "learning_rate": 2.9491921298519495e-05,
      "loss": 2.0176,
      "step": 197611
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.6622531414031982,
      "learning_rate": 2.949015266439224e-05,
      "loss": 1.9766,
      "step": 197612
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.4600343704223633,
      "learning_rate": 2.9488384080557794e-05,
      "loss": 2.0728,
      "step": 197613
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.495801568031311,
      "learning_rate": 2.9486615547016646e-05,
      "loss": 1.9355,
      "step": 197614
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.422903299331665,
      "learning_rate": 2.9484847063768968e-05,
      "loss": 1.8918,
      "step": 197615
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.3746010065078735,
      "learning_rate": 2.9483078630815226e-05,
      "loss": 1.9312,
      "step": 197616
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.4148699045181274,
      "learning_rate": 2.948131024815569e-05,
      "loss": 1.7724,
      "step": 197617
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.3368184566497803,
      "learning_rate": 2.9479541915790616e-05,
      "loss": 1.968,
      "step": 197618
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.4074788093566895,
      "learning_rate": 2.9477773633720446e-05,
      "loss": 2.0318,
      "step": 197619
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.3913071155548096,
      "learning_rate": 2.9476005401945513e-05,
      "loss": 2.0712,
      "step": 197620
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.537844181060791,
      "learning_rate": 2.947423722046598e-05,
      "loss": 1.9295,
      "step": 197621
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.3821420669555664,
      "learning_rate": 2.947246908928238e-05,
      "loss": 2.0654,
      "step": 197622
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.4425373077392578,
      "learning_rate": 2.9470701008394982e-05,
      "loss": 1.9746,
      "step": 197623
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.3807975053787231,
      "learning_rate": 2.946893297780398e-05,
      "loss": 1.8876,
      "step": 197624
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.4262796640396118,
      "learning_rate": 2.9467164997509885e-05,
      "loss": 2.1316,
      "step": 197625
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.4114418029785156,
      "learning_rate": 2.9465397067512987e-05,
      "loss": 1.9742,
      "step": 197626
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.7054916620254517,
      "learning_rate": 2.9463629187813486e-05,
      "loss": 2.0372,
      "step": 197627
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.3728519678115845,
      "learning_rate": 2.9461861358411854e-05,
      "loss": 1.8447,
      "step": 197628
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.3866286277770996,
      "learning_rate": 2.9460093579308384e-05,
      "loss": 2.0835,
      "step": 197629
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.3620166778564453,
      "learning_rate": 2.9458325850503316e-05,
      "loss": 1.916,
      "step": 197630
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.376188039779663,
      "learning_rate": 2.9456558171997115e-05,
      "loss": 2.0686,
      "step": 197631
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.6531912088394165,
      "learning_rate": 2.9454790543790006e-05,
      "loss": 2.1176,
      "step": 197632
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.4678555727005005,
      "learning_rate": 2.94530229658824e-05,
      "loss": 1.9707,
      "step": 197633
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.4581834077835083,
      "learning_rate": 2.945125543827459e-05,
      "loss": 1.8773,
      "step": 197634
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.417984127998352,
      "learning_rate": 2.944948796096691e-05,
      "loss": 1.8591,
      "step": 197635
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.6906852722167969,
      "learning_rate": 2.9447720533959596e-05,
      "loss": 2.2019,
      "step": 197636
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.4288147687911987,
      "learning_rate": 2.9445953157253144e-05,
      "loss": 1.9039,
      "step": 197637
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.393584966659546,
      "learning_rate": 2.944418583084769e-05,
      "loss": 2.0056,
      "step": 197638
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.309238076210022,
      "learning_rate": 2.944241855474376e-05,
      "loss": 2.0164,
      "step": 197639
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.3564772605895996,
      "learning_rate": 2.94406513289416e-05,
      "loss": 1.9287,
      "step": 197640
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.4715676307678223,
      "learning_rate": 2.9438884153441465e-05,
      "loss": 2.0246,
      "step": 197641
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.3878370523452759,
      "learning_rate": 2.9437117028243795e-05,
      "loss": 1.8086,
      "step": 197642
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.3355299234390259,
      "learning_rate": 2.9435349953348887e-05,
      "loss": 1.8245,
      "step": 197643
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.362722635269165,
      "learning_rate": 2.9433582928757003e-05,
      "loss": 1.9094,
      "step": 197644
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.3681387901306152,
      "learning_rate": 2.9431815954468552e-05,
      "loss": 1.783,
      "step": 197645
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.372913122177124,
      "learning_rate": 2.9430049030483793e-05,
      "loss": 1.8654,
      "step": 197646
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.3314905166625977,
      "learning_rate": 2.942828215680316e-05,
      "loss": 2.0723,
      "step": 197647
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.4616211652755737,
      "learning_rate": 2.942651533342689e-05,
      "loss": 2.1082,
      "step": 197648
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.3700560331344604,
      "learning_rate": 2.9424748560355382e-05,
      "loss": 1.7879,
      "step": 197649
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.5323375463485718,
      "learning_rate": 2.9422981837588832e-05,
      "loss": 2.068,
      "step": 197650
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.576432228088379,
      "learning_rate": 2.942121516512771e-05,
      "loss": 2.1578,
      "step": 197651
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.327278971672058,
      "learning_rate": 2.9419448542972246e-05,
      "loss": 2.1817,
      "step": 197652
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.3799939155578613,
      "learning_rate": 2.9417681971122873e-05,
      "loss": 1.9967,
      "step": 197653
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.4567155838012695,
      "learning_rate": 2.9415915449579862e-05,
      "loss": 2.1795,
      "step": 197654
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.6098374128341675,
      "learning_rate": 2.9414148978343544e-05,
      "loss": 1.9813,
      "step": 197655
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.3376765251159668,
      "learning_rate": 2.9412382557414182e-05,
      "loss": 1.7504,
      "step": 197656
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.4076495170593262,
      "learning_rate": 2.941061618679221e-05,
      "loss": 1.8827,
      "step": 197657
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.4296079874038696,
      "learning_rate": 2.940884986647787e-05,
      "loss": 2.1741,
      "step": 197658
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.5962094068527222,
      "learning_rate": 2.940708359647158e-05,
      "loss": 2.2695,
      "step": 197659
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.4722124338150024,
      "learning_rate": 2.940531737677362e-05,
      "loss": 2.1142,
      "step": 197660
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.3505462408065796,
      "learning_rate": 2.9403551207384345e-05,
      "loss": 1.8674,
      "step": 197661
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.5593212842941284,
      "learning_rate": 2.9401785088303963e-05,
      "loss": 2.0272,
      "step": 197662
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.479506492614746,
      "learning_rate": 2.940001901953297e-05,
      "loss": 1.8565,
      "step": 197663
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.5021826028823853,
      "learning_rate": 2.939825300107157e-05,
      "loss": 2.0177,
      "step": 197664
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.4485267400741577,
      "learning_rate": 2.939648703292019e-05,
      "loss": 1.9976,
      "step": 197665
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.433820128440857,
      "learning_rate": 2.9394721115079135e-05,
      "loss": 1.8308,
      "step": 197666
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.4180759191513062,
      "learning_rate": 2.93929552475487e-05,
      "loss": 2.0113,
      "step": 197667
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.3185285329818726,
      "learning_rate": 2.9391189430329154e-05,
      "loss": 1.7464,
      "step": 197668
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.3934801816940308,
      "learning_rate": 2.938942366342093e-05,
      "loss": 1.9327,
      "step": 197669
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.5417630672454834,
      "learning_rate": 2.9387657946824293e-05,
      "loss": 1.8559,
      "step": 197670
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.5024082660675049,
      "learning_rate": 2.9385892280539647e-05,
      "loss": 2.0439,
      "step": 197671
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.4987527132034302,
      "learning_rate": 2.9384126664567288e-05,
      "loss": 1.9163,
      "step": 197672
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.4773606061935425,
      "learning_rate": 2.9382361098907514e-05,
      "loss": 1.9968,
      "step": 197673
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.5221223831176758,
      "learning_rate": 2.9380595583560595e-05,
      "loss": 2.0717,
      "step": 197674
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.3809398412704468,
      "learning_rate": 2.937883011852703e-05,
      "loss": 1.8459,
      "step": 197675
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.3808913230895996,
      "learning_rate": 2.9377064703806952e-05,
      "loss": 2.2264,
      "step": 197676
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.6507742404937744,
      "learning_rate": 2.937529933940086e-05,
      "loss": 1.8311,
      "step": 197677
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.4718799591064453,
      "learning_rate": 2.937353402530902e-05,
      "loss": 1.9973,
      "step": 197678
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.4060513973236084,
      "learning_rate": 2.9371768761531732e-05,
      "loss": 1.9534,
      "step": 197679
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.3700625896453857,
      "learning_rate": 2.9370003548069267e-05,
      "loss": 2.0066,
      "step": 197680
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.5542935132980347,
      "learning_rate": 2.9368238384922117e-05,
      "loss": 1.8342,
      "step": 197681
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.3183938264846802,
      "learning_rate": 2.9366473272090453e-05,
      "loss": 2.1299,
      "step": 197682
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.4328151941299438,
      "learning_rate": 2.9364708209574707e-05,
      "loss": 1.9957,
      "step": 197683
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.4973547458648682,
      "learning_rate": 2.936294319737521e-05,
      "loss": 1.9114,
      "step": 197684
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.3509011268615723,
      "learning_rate": 2.93611782354922e-05,
      "loss": 1.7693,
      "step": 197685
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.4040632247924805,
      "learning_rate": 2.935941332392604e-05,
      "loss": 1.9775,
      "step": 197686
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.3132022619247437,
      "learning_rate": 2.935764846267713e-05,
      "loss": 1.8837,
      "step": 197687
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.5161569118499756,
      "learning_rate": 2.935588365174567e-05,
      "loss": 2.028,
      "step": 197688
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.4261811971664429,
      "learning_rate": 2.9354118891132095e-05,
      "loss": 2.0423,
      "step": 197689
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.4460054636001587,
      "learning_rate": 2.9352354180836734e-05,
      "loss": 2.0567,
      "step": 197690
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.491420030593872,
      "learning_rate": 2.9350589520859858e-05,
      "loss": 2.1153,
      "step": 197691
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.4184157848358154,
      "learning_rate": 2.9348824911201797e-05,
      "loss": 1.996,
      "step": 197692
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.3765314817428589,
      "learning_rate": 2.9347060351862917e-05,
      "loss": 1.7422,
      "step": 197693
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.4422649145126343,
      "learning_rate": 2.9345295842843486e-05,
      "loss": 1.9081,
      "step": 197694
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.4586857557296753,
      "learning_rate": 2.9343531384143905e-05,
      "loss": 2.1839,
      "step": 197695
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.4486134052276611,
      "learning_rate": 2.93417669757645e-05,
      "loss": 2.0638,
      "step": 197696
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.421756386756897,
      "learning_rate": 2.934000261770555e-05,
      "loss": 1.8884,
      "step": 197697
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.3287967443466187,
      "learning_rate": 2.9338238309967377e-05,
      "loss": 2.0379,
      "step": 197698
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.5363143682479858,
      "learning_rate": 2.9336474052550353e-05,
      "loss": 1.9889,
      "step": 197699
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.535982608795166,
      "learning_rate": 2.933470984545474e-05,
      "loss": 1.8159,
      "step": 197700
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.6002705097198486,
      "learning_rate": 2.933294568868101e-05,
      "loss": 1.8876,
      "step": 197701
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.498378038406372,
      "learning_rate": 2.9331181582229356e-05,
      "loss": 2.0905,
      "step": 197702
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.5632911920547485,
      "learning_rate": 2.9329417526100153e-05,
      "loss": 2.0249,
      "step": 197703
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.458997130393982,
      "learning_rate": 2.932765352029366e-05,
      "loss": 1.9162,
      "step": 197704
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.5133322477340698,
      "learning_rate": 2.932588956481031e-05,
      "loss": 2.0258,
      "step": 197705
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.3226191997528076,
      "learning_rate": 2.9324125659650344e-05,
      "loss": 2.0076,
      "step": 197706
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.5509363412857056,
      "learning_rate": 2.9322361804814188e-05,
      "loss": 2.0106,
      "step": 197707
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.3761327266693115,
      "learning_rate": 2.9320598000302075e-05,
      "loss": 2.137,
      "step": 197708
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.3148984909057617,
      "learning_rate": 2.931883424611441e-05,
      "loss": 1.9475,
      "step": 197709
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.376525640487671,
      "learning_rate": 2.9317070542251488e-05,
      "loss": 1.958,
      "step": 197710
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.3898941278457642,
      "learning_rate": 2.931530688871364e-05,
      "loss": 2.0676,
      "step": 197711
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.2967133522033691,
      "learning_rate": 2.9313543285501106e-05,
      "loss": 1.9881,
      "step": 197712
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.3441845178604126,
      "learning_rate": 2.9311779732614384e-05,
      "loss": 1.8959,
      "step": 197713
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.419927954673767,
      "learning_rate": 2.9310016230053636e-05,
      "loss": 1.9141,
      "step": 197714
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.6035746335983276,
      "learning_rate": 2.9308252777819297e-05,
      "loss": 1.9577,
      "step": 197715
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.5409740209579468,
      "learning_rate": 2.9306489375911702e-05,
      "loss": 2.2725,
      "step": 197716
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.4891752004623413,
      "learning_rate": 2.930472602433105e-05,
      "loss": 1.9635,
      "step": 197717
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.4881770610809326,
      "learning_rate": 2.9302962723077837e-05,
      "loss": 1.9968,
      "step": 197718
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.326821208000183,
      "learning_rate": 2.9301199472152337e-05,
      "loss": 2.0441,
      "step": 197719
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.605609655380249,
      "learning_rate": 2.9299436271554776e-05,
      "loss": 2.2012,
      "step": 197720
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.3648154735565186,
      "learning_rate": 2.9297673121285593e-05,
      "loss": 1.9711,
      "step": 197721
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.6347737312316895,
      "learning_rate": 2.9295910021345114e-05,
      "loss": 2.11,
      "step": 197722
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.424822449684143,
      "learning_rate": 2.929414697173358e-05,
      "loss": 1.9357,
      "step": 197723
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.4398927688598633,
      "learning_rate": 2.9292383972451416e-05,
      "loss": 2.0247,
      "step": 197724
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.3045576810836792,
      "learning_rate": 2.9290621023498894e-05,
      "loss": 2.2351,
      "step": 197725
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.4763402938842773,
      "learning_rate": 2.9288858124876314e-05,
      "loss": 2.0253,
      "step": 197726
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.5125352144241333,
      "learning_rate": 2.9287095276584103e-05,
      "loss": 2.119,
      "step": 197727
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.5580140352249146,
      "learning_rate": 2.9285332478622536e-05,
      "loss": 2.2029,
      "step": 197728
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.400450587272644,
      "learning_rate": 2.9283569730991873e-05,
      "loss": 2.0706,
      "step": 197729
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.4901646375656128,
      "learning_rate": 2.928180703369255e-05,
      "loss": 2.0723,
      "step": 197730
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.4464877843856812,
      "learning_rate": 2.92800443867248e-05,
      "loss": 1.8829,
      "step": 197731
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.3980664014816284,
      "learning_rate": 2.9278281790089086e-05,
      "loss": 1.9258,
      "step": 197732
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.3856711387634277,
      "learning_rate": 2.927651924378561e-05,
      "loss": 2.0385,
      "step": 197733
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.3253589868545532,
      "learning_rate": 2.9274756747814743e-05,
      "loss": 1.9287,
      "step": 197734
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.4247528314590454,
      "learning_rate": 2.9272994302176777e-05,
      "loss": 1.8716,
      "step": 197735
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.332000970840454,
      "learning_rate": 2.927123190687212e-05,
      "loss": 1.938,
      "step": 197736
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.3622692823410034,
      "learning_rate": 2.9269469561900993e-05,
      "loss": 1.8801,
      "step": 197737
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.3972967863082886,
      "learning_rate": 2.9267707267263842e-05,
      "loss": 1.8154,
      "step": 197738
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.451858401298523,
      "learning_rate": 2.9265945022960922e-05,
      "loss": 2.2357,
      "step": 197739
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.5415931940078735,
      "learning_rate": 2.9264182828992576e-05,
      "loss": 2.1326,
      "step": 197740
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.3720802068710327,
      "learning_rate": 2.9262420685359068e-05,
      "loss": 1.9353,
      "step": 197741
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.4184201955795288,
      "learning_rate": 2.926065859206086e-05,
      "loss": 2.0488,
      "step": 197742
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.361146330833435,
      "learning_rate": 2.925889654909812e-05,
      "loss": 2.2774,
      "step": 197743
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.4221140146255493,
      "learning_rate": 2.925713455647135e-05,
      "loss": 2.0034,
      "step": 197744
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.509908676147461,
      "learning_rate": 2.9255372614180783e-05,
      "loss": 1.8703,
      "step": 197745
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.5770574808120728,
      "learning_rate": 2.925361072222675e-05,
      "loss": 2.1224,
      "step": 197746
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.4369207620620728,
      "learning_rate": 2.9251848880609518e-05,
      "loss": 1.7735,
      "step": 197747
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.3061851263046265,
      "learning_rate": 2.9250087089329556e-05,
      "loss": 1.9541,
      "step": 197748
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.419324278831482,
      "learning_rate": 2.9248325348387024e-05,
      "loss": 2.1579,
      "step": 197749
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.5029160976409912,
      "learning_rate": 2.924656365778243e-05,
      "loss": 1.9258,
      "step": 197750
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.4723020792007446,
      "learning_rate": 2.9244802017515968e-05,
      "loss": 2.0391,
      "step": 197751
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.4369646310806274,
      "learning_rate": 2.924304042758804e-05,
      "loss": 1.9282,
      "step": 197752
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.3497049808502197,
      "learning_rate": 2.9241278887998876e-05,
      "loss": 1.9634,
      "step": 197753
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.3845195770263672,
      "learning_rate": 2.9239517398748946e-05,
      "loss": 1.9038,
      "step": 197754
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.4577808380126953,
      "learning_rate": 2.9237755959838417e-05,
      "loss": 1.9882,
      "step": 197755
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.3788394927978516,
      "learning_rate": 2.923599457126775e-05,
      "loss": 1.9964,
      "step": 197756
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.475313425064087,
      "learning_rate": 2.9234233233037252e-05,
      "loss": 1.787,
      "step": 197757
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.4624500274658203,
      "learning_rate": 2.923247194514722e-05,
      "loss": 2.1959,
      "step": 197758
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.4735352993011475,
      "learning_rate": 2.9230710707597916e-05,
      "loss": 1.9804,
      "step": 197759
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.399817943572998,
      "learning_rate": 2.9228949520389776e-05,
      "loss": 2.1344,
      "step": 197760
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.3812628984451294,
      "learning_rate": 2.9227188383523038e-05,
      "loss": 1.9131,
      "step": 197761
    },
    {
      "epoch": 2.57,
      "grad_norm": 1.4067590236663818,
      "learning_rate": 2.9225427296998127e-05,
      "loss": 1.9064,
      "step": 197762
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.4342963695526123,
      "learning_rate": 2.922366626081535e-05,
      "loss": 1.9245,
      "step": 197763
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.4114845991134644,
      "learning_rate": 2.9221905274974965e-05,
      "loss": 2.0303,
      "step": 197764
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.5009005069732666,
      "learning_rate": 2.9220144339477312e-05,
      "loss": 2.004,
      "step": 197765
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.3791652917861938,
      "learning_rate": 2.9218383454322792e-05,
      "loss": 1.8327,
      "step": 197766
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.5005457401275635,
      "learning_rate": 2.921662261951163e-05,
      "loss": 1.9372,
      "step": 197767
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.3685616254806519,
      "learning_rate": 2.9214861835044267e-05,
      "loss": 1.9012,
      "step": 197768
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.4707640409469604,
      "learning_rate": 2.9213101100920966e-05,
      "loss": 2.1907,
      "step": 197769
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.3967411518096924,
      "learning_rate": 2.9211340417142058e-05,
      "loss": 1.8874,
      "step": 197770
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.4994250535964966,
      "learning_rate": 2.9209579783707848e-05,
      "loss": 1.9106,
      "step": 197771
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.3588709831237793,
      "learning_rate": 2.920781920061873e-05,
      "loss": 1.7458,
      "step": 197772
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.5395599603652954,
      "learning_rate": 2.920605866787491e-05,
      "loss": 2.0439,
      "step": 197773
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.4777915477752686,
      "learning_rate": 2.9204298185476884e-05,
      "loss": 2.0935,
      "step": 197774
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.5071996450424194,
      "learning_rate": 2.9202537753424814e-05,
      "loss": 2.1486,
      "step": 197775
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.3595930337905884,
      "learning_rate": 2.920077737171921e-05,
      "loss": 2.0583,
      "step": 197776
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.5444004535675049,
      "learning_rate": 2.9199017040360194e-05,
      "loss": 1.8842,
      "step": 197777
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.311052918434143,
      "learning_rate": 2.919725675934824e-05,
      "loss": 1.89,
      "step": 197778
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.4000577926635742,
      "learning_rate": 2.919549652868358e-05,
      "loss": 1.9655,
      "step": 197779
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.5540012121200562,
      "learning_rate": 2.9193736348366647e-05,
      "loss": 2.1464,
      "step": 197780
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.353901982307434,
      "learning_rate": 2.919197621839764e-05,
      "loss": 1.8516,
      "step": 197781
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.4104489088058472,
      "learning_rate": 2.919021613877709e-05,
      "loss": 1.7616,
      "step": 197782
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.4869314432144165,
      "learning_rate": 2.9188456109505066e-05,
      "loss": 1.8681,
      "step": 197783
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.3696905374526978,
      "learning_rate": 2.9186696130582067e-05,
      "loss": 1.9696,
      "step": 197784
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.479023814201355,
      "learning_rate": 2.9184936202008326e-05,
      "loss": 1.8917,
      "step": 197785
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.5886536836624146,
      "learning_rate": 2.9183176323784273e-05,
      "loss": 1.8713,
      "step": 197786
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.4138951301574707,
      "learning_rate": 2.918141649591015e-05,
      "loss": 1.9651,
      "step": 197787
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.627356767654419,
      "learning_rate": 2.917965671838638e-05,
      "loss": 1.8863,
      "step": 197788
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.5514971017837524,
      "learning_rate": 2.9177896991213133e-05,
      "loss": 1.9148,
      "step": 197789
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.3599716424942017,
      "learning_rate": 2.917613731439088e-05,
      "loss": 2.1537,
      "step": 197790
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.40963613986969,
      "learning_rate": 2.9174377687919848e-05,
      "loss": 1.8454,
      "step": 197791
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.4707609415054321,
      "learning_rate": 2.9172618111800473e-05,
      "loss": 2.0301,
      "step": 197792
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.463457465171814,
      "learning_rate": 2.9170858586032952e-05,
      "loss": 2.0167,
      "step": 197793
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.6024694442749023,
      "learning_rate": 2.9169099110617755e-05,
      "loss": 2.2578,
      "step": 197794
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.5648901462554932,
      "learning_rate": 2.9167339685555112e-05,
      "loss": 1.9829,
      "step": 197795
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.4265693426132202,
      "learning_rate": 2.916558031084536e-05,
      "loss": 2.2249,
      "step": 197796
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.361310601234436,
      "learning_rate": 2.9163820986488795e-05,
      "loss": 1.7863,
      "step": 197797
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.381103754043579,
      "learning_rate": 2.9162061712485853e-05,
      "loss": 1.9146,
      "step": 197798
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.5320110321044922,
      "learning_rate": 2.9160302488836763e-05,
      "loss": 2.0188,
      "step": 197799
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.4308080673217773,
      "learning_rate": 2.9158543315541894e-05,
      "loss": 1.8155,
      "step": 197800
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.3920303583145142,
      "learning_rate": 2.9156784192601612e-05,
      "loss": 2.1317,
      "step": 197801
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.5143018960952759,
      "learning_rate": 2.9155025120016152e-05,
      "loss": 2.063,
      "step": 197802
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.3821582794189453,
      "learning_rate": 2.9153266097785842e-05,
      "loss": 1.7144,
      "step": 197803
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.4613747596740723,
      "learning_rate": 2.9151507125911122e-05,
      "loss": 2.0245,
      "step": 197804
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.4445165395736694,
      "learning_rate": 2.914974820439219e-05,
      "loss": 2.0406,
      "step": 197805
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.437853455543518,
      "learning_rate": 2.9147989333229503e-05,
      "loss": 2.0754,
      "step": 197806
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.6288604736328125,
      "learning_rate": 2.9146230512423275e-05,
      "loss": 2.0316,
      "step": 197807
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.4725521802902222,
      "learning_rate": 2.9144471741973864e-05,
      "loss": 2.0039,
      "step": 197808
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.4911807775497437,
      "learning_rate": 2.9142713021881636e-05,
      "loss": 2.0049,
      "step": 197809
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.4118634462356567,
      "learning_rate": 2.9140954352146896e-05,
      "loss": 2.0367,
      "step": 197810
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.6347873210906982,
      "learning_rate": 2.9139195732769904e-05,
      "loss": 1.8974,
      "step": 197811
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.3752461671829224,
      "learning_rate": 2.913743716375113e-05,
      "loss": 1.9423,
      "step": 197812
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.6198723316192627,
      "learning_rate": 2.913567864509081e-05,
      "loss": 2.1253,
      "step": 197813
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.3288674354553223,
      "learning_rate": 2.9133920176789204e-05,
      "loss": 2.0194,
      "step": 197814
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.4308478832244873,
      "learning_rate": 2.9132161758846784e-05,
      "loss": 1.9073,
      "step": 197815
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.4381110668182373,
      "learning_rate": 2.9130403391263746e-05,
      "loss": 1.964,
      "step": 197816
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.3982024192810059,
      "learning_rate": 2.9128645074040557e-05,
      "loss": 1.9502,
      "step": 197817
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.4018234014511108,
      "learning_rate": 2.912688680717745e-05,
      "loss": 1.9138,
      "step": 197818
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.3631287813186646,
      "learning_rate": 2.9125128590674763e-05,
      "loss": 1.9187,
      "step": 197819
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.6122442483901978,
      "learning_rate": 2.9123370424532755e-05,
      "loss": 2.1271,
      "step": 197820
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.3692995309829712,
      "learning_rate": 2.912161230875193e-05,
      "loss": 1.8612,
      "step": 197821
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.5428903102874756,
      "learning_rate": 2.911985424333242e-05,
      "loss": 1.9434,
      "step": 197822
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.438315987586975,
      "learning_rate": 2.911809622827469e-05,
      "loss": 1.9575,
      "step": 197823
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.444982647895813,
      "learning_rate": 2.9116338263579043e-05,
      "loss": 2.059,
      "step": 197824
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.6061021089553833,
      "learning_rate": 2.9114580349245775e-05,
      "loss": 1.7914,
      "step": 197825
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.458592176437378,
      "learning_rate": 2.9112822485275157e-05,
      "loss": 2.0765,
      "step": 197826
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.5312660932540894,
      "learning_rate": 2.911106467166765e-05,
      "loss": 1.9716,
      "step": 197827
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.5739789009094238,
      "learning_rate": 2.9109306908423458e-05,
      "loss": 1.9985,
      "step": 197828
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.4449505805969238,
      "learning_rate": 2.910754919554298e-05,
      "loss": 1.7924,
      "step": 197829
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.7374507188796997,
      "learning_rate": 2.9105791533026545e-05,
      "loss": 1.9541,
      "step": 197830
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.419068694114685,
      "learning_rate": 2.910403392087446e-05,
      "loss": 2.114,
      "step": 197831
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.4004662036895752,
      "learning_rate": 2.9102276359086986e-05,
      "loss": 1.979,
      "step": 197832
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.4560409784317017,
      "learning_rate": 2.910051884766459e-05,
      "loss": 2.0469,
      "step": 197833
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.424856185913086,
      "learning_rate": 2.909876138660744e-05,
      "loss": 1.6896,
      "step": 197834
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.3968329429626465,
      "learning_rate": 2.9097003975916e-05,
      "loss": 2.0924,
      "step": 197835
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.4035993814468384,
      "learning_rate": 2.909524661559054e-05,
      "loss": 2.0895,
      "step": 197836
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.6666966676712036,
      "learning_rate": 2.909348930563139e-05,
      "loss": 2.2115,
      "step": 197837
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.4021329879760742,
      "learning_rate": 2.9091732046038818e-05,
      "loss": 2.074,
      "step": 197838
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.413439393043518,
      "learning_rate": 2.908997483681329e-05,
      "loss": 1.9916,
      "step": 197839
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.3703818321228027,
      "learning_rate": 2.9088217677954973e-05,
      "loss": 1.8873,
      "step": 197840
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.4000221490859985,
      "learning_rate": 2.90864605694643e-05,
      "loss": 1.8009,
      "step": 197841
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.403926968574524,
      "learning_rate": 2.9084703511341533e-05,
      "loss": 1.7762,
      "step": 197842
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.4258471727371216,
      "learning_rate": 2.9082946503587177e-05,
      "loss": 1.9635,
      "step": 197843
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.4559494256973267,
      "learning_rate": 2.9081189546201266e-05,
      "loss": 2.001,
      "step": 197844
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.4136804342269897,
      "learning_rate": 2.907943263918433e-05,
      "loss": 2.3412,
      "step": 197845
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.4543851613998413,
      "learning_rate": 2.9077675782536602e-05,
      "loss": 1.8615,
      "step": 197846
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.3516757488250732,
      "learning_rate": 2.9075918976258482e-05,
      "loss": 2.094,
      "step": 197847
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.2987346649169922,
      "learning_rate": 2.907416222035024e-05,
      "loss": 1.9145,
      "step": 197848
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.382601261138916,
      "learning_rate": 2.9072405514812304e-05,
      "loss": 2.1023,
      "step": 197849
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.4445230960845947,
      "learning_rate": 2.907064885964484e-05,
      "loss": 1.892,
      "step": 197850
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.4406992197036743,
      "learning_rate": 2.9068892254848287e-05,
      "loss": 2.2583,
      "step": 197851
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.5136964321136475,
      "learning_rate": 2.9067135700422905e-05,
      "loss": 1.9574,
      "step": 197852
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.4754395484924316,
      "learning_rate": 2.90653791963691e-05,
      "loss": 1.9316,
      "step": 197853
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.4504982233047485,
      "learning_rate": 2.9063622742687098e-05,
      "loss": 1.8441,
      "step": 197854
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.4627504348754883,
      "learning_rate": 2.9061866339377405e-05,
      "loss": 2.1467,
      "step": 197855
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.5128684043884277,
      "learning_rate": 2.9060109986440118e-05,
      "loss": 2.1036,
      "step": 197856
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.530582308769226,
      "learning_rate": 2.90583536838757e-05,
      "loss": 2.0689,
      "step": 197857
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.2505502700805664,
      "learning_rate": 2.9056597431684427e-05,
      "loss": 1.9752,
      "step": 197858
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.4429991245269775,
      "learning_rate": 2.9054841229866687e-05,
      "loss": 1.9107,
      "step": 197859
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.3995641469955444,
      "learning_rate": 2.9053085078422722e-05,
      "loss": 2.1686,
      "step": 197860
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.466921329498291,
      "learning_rate": 2.9051328977352995e-05,
      "loss": 1.9782,
      "step": 197861
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.398991584777832,
      "learning_rate": 2.904957292665764e-05,
      "loss": 2.1173,
      "step": 197862
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.459633469581604,
      "learning_rate": 2.904781692633712e-05,
      "loss": 1.8134,
      "step": 197863
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.6591243743896484,
      "learning_rate": 2.9046060976391704e-05,
      "loss": 2.1517,
      "step": 197864
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.3758584260940552,
      "learning_rate": 2.904430507682176e-05,
      "loss": 2.1912,
      "step": 197865
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.5731884241104126,
      "learning_rate": 2.9042549227627554e-05,
      "loss": 2.0605,
      "step": 197866
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.3519829511642456,
      "learning_rate": 2.9040793428809585e-05,
      "loss": 1.8439,
      "step": 197867
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.3890477418899536,
      "learning_rate": 2.9039037680367914e-05,
      "loss": 1.7882,
      "step": 197868
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.7031688690185547,
      "learning_rate": 2.903728198230305e-05,
      "loss": 1.9877,
      "step": 197869
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.313349723815918,
      "learning_rate": 2.9035526334615224e-05,
      "loss": 1.9003,
      "step": 197870
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.5104665756225586,
      "learning_rate": 2.903377073730486e-05,
      "loss": 2.0457,
      "step": 197871
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.369957447052002,
      "learning_rate": 2.903201519037217e-05,
      "loss": 2.039,
      "step": 197872
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.383484125137329,
      "learning_rate": 2.903025969381768e-05,
      "loss": 2.0399,
      "step": 197873
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.36809504032135,
      "learning_rate": 2.902850424764146e-05,
      "loss": 2.0275,
      "step": 197874
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.4449894428253174,
      "learning_rate": 2.9026748851844005e-05,
      "loss": 2.0059,
      "step": 197875
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.3761835098266602,
      "learning_rate": 2.902499350642552e-05,
      "loss": 2.0061,
      "step": 197876
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.49285089969635,
      "learning_rate": 2.9023238211386502e-05,
      "loss": 2.1707,
      "step": 197877
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.4153199195861816,
      "learning_rate": 2.9021482966727082e-05,
      "loss": 2.0308,
      "step": 197878
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.6602623462677002,
      "learning_rate": 2.9019727772447766e-05,
      "loss": 1.6902,
      "step": 197879
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.6317105293273926,
      "learning_rate": 2.901797262854878e-05,
      "loss": 2.1108,
      "step": 197880
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.2902354001998901,
      "learning_rate": 2.9016217535030494e-05,
      "loss": 1.7757,
      "step": 197881
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.5309468507766724,
      "learning_rate": 2.9014462491893107e-05,
      "loss": 1.8923,
      "step": 197882
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.424355149269104,
      "learning_rate": 2.9012707499137156e-05,
      "loss": 1.8993,
      "step": 197883
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.4273946285247803,
      "learning_rate": 2.901095255676277e-05,
      "loss": 2.2829,
      "step": 197884
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.3520441055297852,
      "learning_rate": 2.9009197664770412e-05,
      "loss": 2.0329,
      "step": 197885
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.4339410066604614,
      "learning_rate": 2.900744282316039e-05,
      "loss": 1.7958,
      "step": 197886
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.6215603351593018,
      "learning_rate": 2.9005688031932962e-05,
      "loss": 1.8839,
      "step": 197887
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.3969939947128296,
      "learning_rate": 2.9003933291088465e-05,
      "loss": 1.9081,
      "step": 197888
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.4418225288391113,
      "learning_rate": 2.90021786006273e-05,
      "loss": 1.9185,
      "step": 197889
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.3213648796081543,
      "learning_rate": 2.90004239605497e-05,
      "loss": 1.9311,
      "step": 197890
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.4100550413131714,
      "learning_rate": 2.8998669370856065e-05,
      "loss": 1.9114,
      "step": 197891
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.4902983903884888,
      "learning_rate": 2.8996914831546724e-05,
      "loss": 2.0054,
      "step": 197892
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.3428529500961304,
      "learning_rate": 2.899516034262188e-05,
      "loss": 1.9633,
      "step": 197893
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.4570326805114746,
      "learning_rate": 2.8993405904082035e-05,
      "loss": 1.869,
      "step": 197894
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.3704029321670532,
      "learning_rate": 2.8991651515927417e-05,
      "loss": 1.8363,
      "step": 197895
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.3408443927764893,
      "learning_rate": 2.8989897178158327e-05,
      "loss": 2.0753,
      "step": 197896
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.4527089595794678,
      "learning_rate": 2.8988142890775168e-05,
      "loss": 2.1445,
      "step": 197897
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.36798095703125,
      "learning_rate": 2.898638865377827e-05,
      "loss": 1.9519,
      "step": 197898
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.5010992288589478,
      "learning_rate": 2.89846344671678e-05,
      "loss": 2.1093,
      "step": 197899
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.3436766862869263,
      "learning_rate": 2.8982880330944292e-05,
      "loss": 1.9005,
      "step": 197900
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.4205965995788574,
      "learning_rate": 2.8981126245107944e-05,
      "loss": 2.0675,
      "step": 197901
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.4194285869598389,
      "learning_rate": 2.8979372209659156e-05,
      "loss": 1.8728,
      "step": 197902
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.3806145191192627,
      "learning_rate": 2.897761822459823e-05,
      "loss": 1.9944,
      "step": 197903
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.4119690656661987,
      "learning_rate": 2.8975864289925466e-05,
      "loss": 2.2609,
      "step": 197904
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.5366793870925903,
      "learning_rate": 2.8974110405641126e-05,
      "loss": 2.1455,
      "step": 197905
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.4508565664291382,
      "learning_rate": 2.8972356571745715e-05,
      "loss": 2.2568,
      "step": 197906
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.3676249980926514,
      "learning_rate": 2.8970602788239393e-05,
      "loss": 2.1022,
      "step": 197907
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.402951955795288,
      "learning_rate": 2.89688490551226e-05,
      "loss": 2.4313,
      "step": 197908
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.5337711572647095,
      "learning_rate": 2.8967095372395598e-05,
      "loss": 1.9839,
      "step": 197909
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.3892279863357544,
      "learning_rate": 2.8965341740058755e-05,
      "loss": 2.0425,
      "step": 197910
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.253603458404541,
      "learning_rate": 2.8963588158112305e-05,
      "loss": 1.9026,
      "step": 197911
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.4352508783340454,
      "learning_rate": 2.8961834626556713e-05,
      "loss": 1.9712,
      "step": 197912
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.4115681648254395,
      "learning_rate": 2.8960081145392144e-05,
      "loss": 1.9291,
      "step": 197913
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.4089773893356323,
      "learning_rate": 2.89583277146191e-05,
      "loss": 1.8949,
      "step": 197914
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.5062048435211182,
      "learning_rate": 2.8956574334237746e-05,
      "loss": 1.898,
      "step": 197915
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.668725848197937,
      "learning_rate": 2.8954821004248585e-05,
      "loss": 2.1103,
      "step": 197916
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.4724459648132324,
      "learning_rate": 2.8953067724651745e-05,
      "loss": 1.9636,
      "step": 197917
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.456220269203186,
      "learning_rate": 2.8951314495447697e-05,
      "loss": 1.7473,
      "step": 197918
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.3996684551239014,
      "learning_rate": 2.8949561316636638e-05,
      "loss": 1.9978,
      "step": 197919
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.5541937351226807,
      "learning_rate": 2.894780818821907e-05,
      "loss": 2.188,
      "step": 197920
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.5090702772140503,
      "learning_rate": 2.894605511019512e-05,
      "loss": 2.0495,
      "step": 197921
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.536093831062317,
      "learning_rate": 2.8944302082565362e-05,
      "loss": 1.9144,
      "step": 197922
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.384158730506897,
      "learning_rate": 2.894254910532986e-05,
      "loss": 1.9092,
      "step": 197923
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.47287118434906,
      "learning_rate": 2.8940796178489078e-05,
      "loss": 2.2759,
      "step": 197924
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.3665491342544556,
      "learning_rate": 2.8939043302043285e-05,
      "loss": 1.8447,
      "step": 197925
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.3575024604797363,
      "learning_rate": 2.8937290475992913e-05,
      "loss": 1.8326,
      "step": 197926
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.388798713684082,
      "learning_rate": 2.893553770033813e-05,
      "loss": 2.0226,
      "step": 197927
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.503399133682251,
      "learning_rate": 2.89337849750795e-05,
      "loss": 1.9343,
      "step": 197928
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.4289387464523315,
      "learning_rate": 2.8932032300217056e-05,
      "loss": 1.8257,
      "step": 197929
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.432328701019287,
      "learning_rate": 2.89302796757513e-05,
      "loss": 2.1847,
      "step": 197930
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.3989367485046387,
      "learning_rate": 2.8928527101682497e-05,
      "loss": 1.964,
      "step": 197931
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.395643949508667,
      "learning_rate": 2.8926774578011046e-05,
      "loss": 2.1079,
      "step": 197932
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.4056119918823242,
      "learning_rate": 2.892502210473715e-05,
      "loss": 1.8612,
      "step": 197933
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.3796769380569458,
      "learning_rate": 2.892326968186134e-05,
      "loss": 1.9427,
      "step": 197934
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.3867933750152588,
      "learning_rate": 2.8921517309383714e-05,
      "loss": 1.9495,
      "step": 197935
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.3997139930725098,
      "learning_rate": 2.891976498730474e-05,
      "loss": 1.9764,
      "step": 197936
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.399896502494812,
      "learning_rate": 2.8918012715624618e-05,
      "loss": 1.7574,
      "step": 197937
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.3599144220352173,
      "learning_rate": 2.8916260494343844e-05,
      "loss": 2.0015,
      "step": 197938
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.5324177742004395,
      "learning_rate": 2.891450832346256e-05,
      "loss": 1.875,
      "step": 197939
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.4598193168640137,
      "learning_rate": 2.8912756202981324e-05,
      "loss": 2.007,
      "step": 197940
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.4797369241714478,
      "learning_rate": 2.8911004132900207e-05,
      "loss": 1.9223,
      "step": 197941
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.5512133836746216,
      "learning_rate": 2.8909252113219706e-05,
      "loss": 1.8234,
      "step": 197942
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.3867658376693726,
      "learning_rate": 2.8907500143940053e-05,
      "loss": 2.0965,
      "step": 197943
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.4194362163543701,
      "learning_rate": 2.8905748225061653e-05,
      "loss": 2.0063,
      "step": 197944
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.494274616241455,
      "learning_rate": 2.8903996356584735e-05,
      "loss": 2.0405,
      "step": 197945
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.4998509883880615,
      "learning_rate": 2.89022445385098e-05,
      "loss": 2.1204,
      "step": 197946
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.449946403503418,
      "learning_rate": 2.8900492770836947e-05,
      "loss": 2.0405,
      "step": 197947
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.495587944984436,
      "learning_rate": 2.8898741053566642e-05,
      "loss": 2.0158,
      "step": 197948
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.5131680965423584,
      "learning_rate": 2.8896989386699154e-05,
      "loss": 2.0253,
      "step": 197949
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.3795192241668701,
      "learning_rate": 2.889523777023488e-05,
      "loss": 1.9121,
      "step": 197950
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.3993428945541382,
      "learning_rate": 2.889348620417402e-05,
      "loss": 1.9828,
      "step": 197951
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.464158535003662,
      "learning_rate": 2.889173468851711e-05,
      "loss": 2.0628,
      "step": 197952
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.4541155099868774,
      "learning_rate": 2.8889983223264247e-05,
      "loss": 2.0817,
      "step": 197953
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.390622615814209,
      "learning_rate": 2.8888231808415896e-05,
      "loss": 1.9689,
      "step": 197954
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.4033441543579102,
      "learning_rate": 2.888648044397226e-05,
      "loss": 2.0225,
      "step": 197955
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.3751780986785889,
      "learning_rate": 2.888472912993384e-05,
      "loss": 1.74,
      "step": 197956
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.4983757734298706,
      "learning_rate": 2.8882977866300793e-05,
      "loss": 2.1256,
      "step": 197957
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.3327860832214355,
      "learning_rate": 2.888122665307363e-05,
      "loss": 1.8962,
      "step": 197958
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.310550332069397,
      "learning_rate": 2.8879475490252446e-05,
      "loss": 1.8728,
      "step": 197959
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.5440068244934082,
      "learning_rate": 2.887772437783774e-05,
      "loss": 1.8646,
      "step": 197960
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.296305537223816,
      "learning_rate": 2.887597331582975e-05,
      "loss": 1.8514,
      "step": 197961
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.44209885597229,
      "learning_rate": 2.8874222304228867e-05,
      "loss": 1.7462,
      "step": 197962
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.3519452810287476,
      "learning_rate": 2.887247134303533e-05,
      "loss": 2.0238,
      "step": 197963
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.4738672971725464,
      "learning_rate": 2.8870720432249572e-05,
      "loss": 2.0153,
      "step": 197964
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.4776802062988281,
      "learning_rate": 2.886896957187189e-05,
      "loss": 1.8983,
      "step": 197965
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.381839394569397,
      "learning_rate": 2.8867218761902554e-05,
      "loss": 2.0345,
      "step": 197966
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.319183588027954,
      "learning_rate": 2.8865468002341862e-05,
      "loss": 1.9207,
      "step": 197967
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.4026561975479126,
      "learning_rate": 2.886371729319028e-05,
      "loss": 2.0132,
      "step": 197968
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.4508720636367798,
      "learning_rate": 2.8861966634447975e-05,
      "loss": 2.0338,
      "step": 197969
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.3483009338378906,
      "learning_rate": 2.8860216026115414e-05,
      "loss": 1.8391,
      "step": 197970
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.310307502746582,
      "learning_rate": 2.885846546819286e-05,
      "loss": 2.0427,
      "step": 197971
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.5869338512420654,
      "learning_rate": 2.8856714960680648e-05,
      "loss": 1.9368,
      "step": 197972
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.5711597204208374,
      "learning_rate": 2.8854964503579016e-05,
      "loss": 1.8746,
      "step": 197973
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.479769229888916,
      "learning_rate": 2.885321409688842e-05,
      "loss": 1.8422,
      "step": 197974
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.4358257055282593,
      "learning_rate": 2.885146374060907e-05,
      "loss": 2.017,
      "step": 197975
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.5035887956619263,
      "learning_rate": 2.8849713434741427e-05,
      "loss": 2.0358,
      "step": 197976
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.66872239112854,
      "learning_rate": 2.8847963179285726e-05,
      "loss": 2.2062,
      "step": 197977
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.3910146951675415,
      "learning_rate": 2.8846212974242267e-05,
      "loss": 1.9878,
      "step": 197978
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.4391998052597046,
      "learning_rate": 2.8844462819611446e-05,
      "loss": 1.8204,
      "step": 197979
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.492437720298767,
      "learning_rate": 2.884271271539357e-05,
      "loss": 1.8889,
      "step": 197980
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.3426649570465088,
      "learning_rate": 2.8840962661588897e-05,
      "loss": 1.9506,
      "step": 197981
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.4728626012802124,
      "learning_rate": 2.8839212658197864e-05,
      "loss": 1.9266,
      "step": 197982
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.463839054107666,
      "learning_rate": 2.883746270522074e-05,
      "loss": 1.9796,
      "step": 197983
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.673628807067871,
      "learning_rate": 2.8835712802657784e-05,
      "loss": 2.2447,
      "step": 197984
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.3906733989715576,
      "learning_rate": 2.8833962950509438e-05,
      "loss": 1.7855,
      "step": 197985
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.4904768466949463,
      "learning_rate": 2.8832213148775995e-05,
      "loss": 2.0242,
      "step": 197986
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.5310732126235962,
      "learning_rate": 2.8830463397457725e-05,
      "loss": 1.8947,
      "step": 197987
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.4466789960861206,
      "learning_rate": 2.8828713696554996e-05,
      "loss": 1.9467,
      "step": 197988
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.8280041217803955,
      "learning_rate": 2.882696404606817e-05,
      "loss": 1.924,
      "step": 197989
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.4897931814193726,
      "learning_rate": 2.8825214445997448e-05,
      "loss": 2.0434,
      "step": 197990
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.5103769302368164,
      "learning_rate": 2.882346489634333e-05,
      "loss": 2.0125,
      "step": 197991
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.3238857984542847,
      "learning_rate": 2.882171539710595e-05,
      "loss": 2.1544,
      "step": 197992
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.3931255340576172,
      "learning_rate": 2.8819965948285805e-05,
      "loss": 1.8553,
      "step": 197993
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.4571894407272339,
      "learning_rate": 2.8818216549883166e-05,
      "loss": 2.1975,
      "step": 197994
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.455759048461914,
      "learning_rate": 2.8816467201898296e-05,
      "loss": 1.9266,
      "step": 197995
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.6566028594970703,
      "learning_rate": 2.8814717904331562e-05,
      "loss": 2.1735,
      "step": 197996
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.4679036140441895,
      "learning_rate": 2.8812968657183298e-05,
      "loss": 2.044,
      "step": 197997
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.419988751411438,
      "learning_rate": 2.8811219460453803e-05,
      "loss": 2.2065,
      "step": 197998
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.5451617240905762,
      "learning_rate": 2.8809470314143446e-05,
      "loss": 2.1342,
      "step": 197999
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.5022681951522827,
      "learning_rate": 2.880772121825249e-05,
      "loss": 2.2133,
      "step": 198000
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.4910616874694824,
      "learning_rate": 2.88059721727814e-05,
      "loss": 2.0439,
      "step": 198001
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.4933502674102783,
      "learning_rate": 2.8804223177730312e-05,
      "loss": 2.1508,
      "step": 198002
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.3726003170013428,
      "learning_rate": 2.880247423309966e-05,
      "loss": 2.2693,
      "step": 198003
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.4556076526641846,
      "learning_rate": 2.880072533888971e-05,
      "loss": 2.0328,
      "step": 198004
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.325927972793579,
      "learning_rate": 2.879897649510089e-05,
      "loss": 1.9692,
      "step": 198005
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.2672866582870483,
      "learning_rate": 2.8797227701733373e-05,
      "loss": 1.93,
      "step": 198006
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.4576280117034912,
      "learning_rate": 2.8795478958787722e-05,
      "loss": 1.9991,
      "step": 198007
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.5140471458435059,
      "learning_rate": 2.879373026626397e-05,
      "loss": 1.8178,
      "step": 198008
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.5097616910934448,
      "learning_rate": 2.879198162416265e-05,
      "loss": 2.0165,
      "step": 198009
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.747017741203308,
      "learning_rate": 2.8790233032483935e-05,
      "loss": 1.9552,
      "step": 198010
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.3457478284835815,
      "learning_rate": 2.878848449122835e-05,
      "loss": 1.9715,
      "step": 198011
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.3703134059906006,
      "learning_rate": 2.8786736000395993e-05,
      "loss": 2.0397,
      "step": 198012
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.429113745689392,
      "learning_rate": 2.8784987559987438e-05,
      "loss": 1.9404,
      "step": 198013
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.4039286375045776,
      "learning_rate": 2.8783239170002747e-05,
      "loss": 1.9436,
      "step": 198014
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.4806239604949951,
      "learning_rate": 2.8781490830442457e-05,
      "loss": 2.0324,
      "step": 198015
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.416192650794983,
      "learning_rate": 2.877974254130673e-05,
      "loss": 2.0321,
      "step": 198016
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.3646023273468018,
      "learning_rate": 2.877799430259603e-05,
      "loss": 1.9407,
      "step": 198017
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.6799285411834717,
      "learning_rate": 2.8776246114310562e-05,
      "loss": 2.0622,
      "step": 198018
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.481742024421692,
      "learning_rate": 2.877449797645083e-05,
      "loss": 1.9513,
      "step": 198019
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.5090227127075195,
      "learning_rate": 2.8772749889016922e-05,
      "loss": 1.8354,
      "step": 198020
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.4010432958602905,
      "learning_rate": 2.8771001852009312e-05,
      "loss": 1.9748,
      "step": 198021
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.2665657997131348,
      "learning_rate": 2.8769253865428264e-05,
      "loss": 1.8233,
      "step": 198022
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.5307447910308838,
      "learning_rate": 2.876750592927418e-05,
      "loss": 2.069,
      "step": 198023
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.4455167055130005,
      "learning_rate": 2.8765758043547292e-05,
      "loss": 1.908,
      "step": 198024
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.8490748405456543,
      "learning_rate": 2.8764010208248066e-05,
      "loss": 2.298,
      "step": 198025
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.3624417781829834,
      "learning_rate": 2.8762262423376635e-05,
      "loss": 1.9709,
      "step": 198026
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.4206920862197876,
      "learning_rate": 2.8760514688933467e-05,
      "loss": 2.0152,
      "step": 198027
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.491342544555664,
      "learning_rate": 2.875876700491876e-05,
      "loss": 2.4433,
      "step": 198028
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.3328711986541748,
      "learning_rate": 2.875701937133301e-05,
      "loss": 1.8604,
      "step": 198029
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.3878813982009888,
      "learning_rate": 2.875527178817639e-05,
      "loss": 1.9679,
      "step": 198030
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.4623759984970093,
      "learning_rate": 2.8753524255449366e-05,
      "loss": 2.2753,
      "step": 198031
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.4636940956115723,
      "learning_rate": 2.8751776773152103e-05,
      "loss": 1.9977,
      "step": 198032
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.4879757165908813,
      "learning_rate": 2.875002934128503e-05,
      "loss": 1.9399,
      "step": 198033
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.6372830867767334,
      "learning_rate": 2.8748281959848417e-05,
      "loss": 2.0641,
      "step": 198034
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.6711463928222656,
      "learning_rate": 2.8746534628842667e-05,
      "loss": 1.7924,
      "step": 198035
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.3413374423980713,
      "learning_rate": 2.8744787348268005e-05,
      "loss": 1.8088,
      "step": 198036
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.3853094577789307,
      "learning_rate": 2.8743040118124906e-05,
      "loss": 1.9777,
      "step": 198037
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.3956800699234009,
      "learning_rate": 2.8741292938413494e-05,
      "loss": 1.9764,
      "step": 198038
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.3501834869384766,
      "learning_rate": 2.873954580913428e-05,
      "loss": 2.1424,
      "step": 198039
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.349134087562561,
      "learning_rate": 2.8737798730287388e-05,
      "loss": 1.9851,
      "step": 198040
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.4691730737686157,
      "learning_rate": 2.8736051701873354e-05,
      "loss": 1.9679,
      "step": 198041
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.4105093479156494,
      "learning_rate": 2.8734304723892342e-05,
      "loss": 2.0763,
      "step": 198042
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.294942021369934,
      "learning_rate": 2.8732557796344824e-05,
      "loss": 1.989,
      "step": 198043
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.5075206756591797,
      "learning_rate": 2.8730810919231027e-05,
      "loss": 1.8641,
      "step": 198044
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.8198609352111816,
      "learning_rate": 2.872906409255129e-05,
      "loss": 2.1261,
      "step": 198045
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.4073450565338135,
      "learning_rate": 2.8727317316305875e-05,
      "loss": 1.9561,
      "step": 198046
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.4872530698776245,
      "learning_rate": 2.872557059049522e-05,
      "loss": 2.0996,
      "step": 198047
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.4267271757125854,
      "learning_rate": 2.872382391511958e-05,
      "loss": 2.0228,
      "step": 198048
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.5382322072982788,
      "learning_rate": 2.8722077290179337e-05,
      "loss": 1.8035,
      "step": 198049
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.460591197013855,
      "learning_rate": 2.872033071567478e-05,
      "loss": 2.1342,
      "step": 198050
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.586110234260559,
      "learning_rate": 2.871858419160624e-05,
      "loss": 1.9303,
      "step": 198051
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.377629280090332,
      "learning_rate": 2.8716837717973993e-05,
      "loss": 2.0022,
      "step": 198052
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.3428276777267456,
      "learning_rate": 2.8715091294778437e-05,
      "loss": 1.9867,
      "step": 198053
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.460798740386963,
      "learning_rate": 2.871334492201983e-05,
      "loss": 2.0181,
      "step": 198054
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.4004714488983154,
      "learning_rate": 2.871159859969858e-05,
      "loss": 2.2305,
      "step": 198055
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.328691005706787,
      "learning_rate": 2.8709852327814954e-05,
      "loss": 1.8318,
      "step": 198056
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.5961486101150513,
      "learning_rate": 2.870810610636931e-05,
      "loss": 1.9404,
      "step": 198057
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.3395740985870361,
      "learning_rate": 2.870635993536189e-05,
      "loss": 1.9937,
      "step": 198058
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.3908023834228516,
      "learning_rate": 2.870461381479312e-05,
      "loss": 1.9571,
      "step": 198059
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.3443183898925781,
      "learning_rate": 2.8702867744663238e-05,
      "loss": 1.8665,
      "step": 198060
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.4830774068832397,
      "learning_rate": 2.8701121724972677e-05,
      "loss": 1.8877,
      "step": 198061
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.3720555305480957,
      "learning_rate": 2.8699375755721664e-05,
      "loss": 1.9212,
      "step": 198062
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.4807422161102295,
      "learning_rate": 2.869762983691054e-05,
      "loss": 2.0951,
      "step": 198063
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.294213891029358,
      "learning_rate": 2.8695883968539703e-05,
      "loss": 2.0871,
      "step": 198064
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.3461575508117676,
      "learning_rate": 2.8694138150609415e-05,
      "loss": 1.8156,
      "step": 198065
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.5342397689819336,
      "learning_rate": 2.8692392383119946e-05,
      "loss": 1.9562,
      "step": 198066
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.4305111169815063,
      "learning_rate": 2.8690646666071726e-05,
      "loss": 2.0903,
      "step": 198067
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.4238425493240356,
      "learning_rate": 2.868890099946506e-05,
      "loss": 1.9002,
      "step": 198068
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.5306459665298462,
      "learning_rate": 2.8687155383300175e-05,
      "loss": 1.9578,
      "step": 198069
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.5208584070205688,
      "learning_rate": 2.8685409817577542e-05,
      "loss": 1.758,
      "step": 198070
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.4071005582809448,
      "learning_rate": 2.8683664302297395e-05,
      "loss": 1.8909,
      "step": 198071
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.379485845565796,
      "learning_rate": 2.8681918837460028e-05,
      "loss": 2.0861,
      "step": 198072
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.456244945526123,
      "learning_rate": 2.868017342306588e-05,
      "loss": 2.0881,
      "step": 198073
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.3749058246612549,
      "learning_rate": 2.8678428059115212e-05,
      "loss": 2.0988,
      "step": 198074
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.409380316734314,
      "learning_rate": 2.867668274560826e-05,
      "loss": 1.9542,
      "step": 198075
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.4841728210449219,
      "learning_rate": 2.8674937482545522e-05,
      "loss": 1.985,
      "step": 198076
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.4427456855773926,
      "learning_rate": 2.867319226992717e-05,
      "loss": 2.0044,
      "step": 198077
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.5306386947631836,
      "learning_rate": 2.8671447107753633e-05,
      "loss": 1.8775,
      "step": 198078
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.5821073055267334,
      "learning_rate": 2.8669701996025206e-05,
      "loss": 2.2182,
      "step": 198079
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.469534993171692,
      "learning_rate": 2.8667956934742196e-05,
      "loss": 2.0029,
      "step": 198080
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.533341884613037,
      "learning_rate": 2.8666211923904904e-05,
      "loss": 2.2324,
      "step": 198081
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.3400228023529053,
      "learning_rate": 2.866446696351372e-05,
      "loss": 1.7582,
      "step": 198082
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.5593875646591187,
      "learning_rate": 2.8662722053568887e-05,
      "loss": 1.9446,
      "step": 198083
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.4718717336654663,
      "learning_rate": 2.8660977194070833e-05,
      "loss": 1.9245,
      "step": 198084
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.4685373306274414,
      "learning_rate": 2.8659232385019825e-05,
      "loss": 2.0362,
      "step": 198085
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.4395495653152466,
      "learning_rate": 2.8657487626416165e-05,
      "loss": 1.9792,
      "step": 198086
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.3649383783340454,
      "learning_rate": 2.865574291826015e-05,
      "loss": 2.1086,
      "step": 198087
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.399502158164978,
      "learning_rate": 2.8653998260552246e-05,
      "loss": 1.926,
      "step": 198088
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.401758074760437,
      "learning_rate": 2.865225365329259e-05,
      "loss": 1.9258,
      "step": 198089
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.4374151229858398,
      "learning_rate": 2.865050909648168e-05,
      "loss": 2.0745,
      "step": 198090
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.4026663303375244,
      "learning_rate": 2.864876459011971e-05,
      "loss": 2.1607,
      "step": 198091
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.459751844406128,
      "learning_rate": 2.8647020134207156e-05,
      "loss": 2.0974,
      "step": 198092
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.3708969354629517,
      "learning_rate": 2.8645275728744112e-05,
      "loss": 2.0326,
      "step": 198093
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.3897759914398193,
      "learning_rate": 2.864353137373111e-05,
      "loss": 2.0374,
      "step": 198094
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.5084813833236694,
      "learning_rate": 2.8641787069168353e-05,
      "loss": 2.0882,
      "step": 198095
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.329262137413025,
      "learning_rate": 2.8640042815056242e-05,
      "loss": 1.9252,
      "step": 198096
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.281046748161316,
      "learning_rate": 2.863829861139504e-05,
      "loss": 1.9435,
      "step": 198097
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.3301829099655151,
      "learning_rate": 2.8636554458185214e-05,
      "loss": 1.8337,
      "step": 198098
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.406504511833191,
      "learning_rate": 2.863481035542683e-05,
      "loss": 1.9618,
      "step": 198099
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.4188671112060547,
      "learning_rate": 2.8633066303120456e-05,
      "loss": 2.1905,
      "step": 198100
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.5461084842681885,
      "learning_rate": 2.8631322301266223e-05,
      "loss": 1.9027,
      "step": 198101
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.6175503730773926,
      "learning_rate": 2.8629578349864636e-05,
      "loss": 2.1268,
      "step": 198102
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.489595890045166,
      "learning_rate": 2.8627834448915855e-05,
      "loss": 2.1776,
      "step": 198103
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.362136721611023,
      "learning_rate": 2.862609059842038e-05,
      "loss": 1.9749,
      "step": 198104
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.3345881700515747,
      "learning_rate": 2.862434679837835e-05,
      "loss": 2.0926,
      "step": 198105
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.4869612455368042,
      "learning_rate": 2.8622603048790227e-05,
      "loss": 2.2783,
      "step": 198106
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.4614216089248657,
      "learning_rate": 2.8620859349656244e-05,
      "loss": 1.9323,
      "step": 198107
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.4614827632904053,
      "learning_rate": 2.8619115700976803e-05,
      "loss": 1.991,
      "step": 198108
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.3620256185531616,
      "learning_rate": 2.8617372102752135e-05,
      "loss": 1.8969,
      "step": 198109
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.3731211423873901,
      "learning_rate": 2.8615628554982673e-05,
      "loss": 2.0193,
      "step": 198110
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.3361533880233765,
      "learning_rate": 2.8613885057668716e-05,
      "loss": 2.0119,
      "step": 198111
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.4586278200149536,
      "learning_rate": 2.8612141610810534e-05,
      "loss": 2.0513,
      "step": 198112
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.3803184032440186,
      "learning_rate": 2.8610398214408424e-05,
      "loss": 1.8413,
      "step": 198113
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.385009765625,
      "learning_rate": 2.860865486846282e-05,
      "loss": 2.137,
      "step": 198114
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.3068511486053467,
      "learning_rate": 2.8606911572973955e-05,
      "loss": 1.8861,
      "step": 198115
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.4261404275894165,
      "learning_rate": 2.860516832794223e-05,
      "loss": 2.1975,
      "step": 198116
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.3922228813171387,
      "learning_rate": 2.860342513336791e-05,
      "loss": 2.0822,
      "step": 198117
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.2857393026351929,
      "learning_rate": 2.8601681989251357e-05,
      "loss": 1.8527,
      "step": 198118
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.3784226179122925,
      "learning_rate": 2.8599938895592812e-05,
      "loss": 2.0314,
      "step": 198119
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.2420514822006226,
      "learning_rate": 2.8598195852392704e-05,
      "loss": 1.902,
      "step": 198120
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.4791361093521118,
      "learning_rate": 2.8596452859651297e-05,
      "loss": 1.9374,
      "step": 198121
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.4588977098464966,
      "learning_rate": 2.8594709917368964e-05,
      "loss": 2.1444,
      "step": 198122
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.541982889175415,
      "learning_rate": 2.8592967025545997e-05,
      "loss": 2.1404,
      "step": 198123
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.5347007513046265,
      "learning_rate": 2.8591224184182704e-05,
      "loss": 2.0084,
      "step": 198124
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.4931074380874634,
      "learning_rate": 2.858948139327938e-05,
      "loss": 2.08,
      "step": 198125
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.4437663555145264,
      "learning_rate": 2.8587738652836456e-05,
      "loss": 1.9442,
      "step": 198126
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.3744630813598633,
      "learning_rate": 2.8585995962854135e-05,
      "loss": 1.871,
      "step": 198127
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.4808664321899414,
      "learning_rate": 2.8584253323332884e-05,
      "loss": 1.9479,
      "step": 198128
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.6083797216415405,
      "learning_rate": 2.85825107342729e-05,
      "loss": 1.9111,
      "step": 198129
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.4244517087936401,
      "learning_rate": 2.8580768195674587e-05,
      "loss": 2.1866,
      "step": 198130
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.5621163845062256,
      "learning_rate": 2.857902570753817e-05,
      "loss": 1.8561,
      "step": 198131
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.3706588745117188,
      "learning_rate": 2.857728326986406e-05,
      "loss": 1.99,
      "step": 198132
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.3097411394119263,
      "learning_rate": 2.8575540882652547e-05,
      "loss": 2.0355,
      "step": 198133
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.3785446882247925,
      "learning_rate": 2.8573798545904002e-05,
      "loss": 2.173,
      "step": 198134
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.4807419776916504,
      "learning_rate": 2.8572056259618692e-05,
      "loss": 2.0759,
      "step": 198135
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.4657824039459229,
      "learning_rate": 2.8570314023796982e-05,
      "loss": 2.1793,
      "step": 198136
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.3342344760894775,
      "learning_rate": 2.8568571838439104e-05,
      "loss": 2.0319,
      "step": 198137
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.4445853233337402,
      "learning_rate": 2.8566829703545524e-05,
      "loss": 2.0878,
      "step": 198138
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.485952615737915,
      "learning_rate": 2.856508761911641e-05,
      "loss": 1.9704,
      "step": 198139
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.459635615348816,
      "learning_rate": 2.8563345585152265e-05,
      "loss": 1.8522,
      "step": 198140
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.3630657196044922,
      "learning_rate": 2.8561603601653282e-05,
      "loss": 1.8769,
      "step": 198141
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.3570696115493774,
      "learning_rate": 2.8559861668619864e-05,
      "loss": 1.8314,
      "step": 198142
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.6229298114776611,
      "learning_rate": 2.855811978605218e-05,
      "loss": 2.1221,
      "step": 198143
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.4079211950302124,
      "learning_rate": 2.855637795395076e-05,
      "loss": 2.1039,
      "step": 198144
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.481063723564148,
      "learning_rate": 2.855463617231577e-05,
      "loss": 2.0826,
      "step": 198145
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.4207868576049805,
      "learning_rate": 2.8552894441147643e-05,
      "loss": 2.0622,
      "step": 198146
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.4429329633712769,
      "learning_rate": 2.8551152760446682e-05,
      "loss": 1.9838,
      "step": 198147
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.3543503284454346,
      "learning_rate": 2.854941113021315e-05,
      "loss": 1.9134,
      "step": 198148
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.4384667873382568,
      "learning_rate": 2.8547669550447383e-05,
      "loss": 1.9083,
      "step": 198149
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.5481674671173096,
      "learning_rate": 2.8545928021149776e-05,
      "loss": 2.0374,
      "step": 198150
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.6680909395217896,
      "learning_rate": 2.8544186542320534e-05,
      "loss": 1.7814,
      "step": 198151
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.5491914749145508,
      "learning_rate": 2.854244511396012e-05,
      "loss": 2.0769,
      "step": 198152
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.4355478286743164,
      "learning_rate": 2.854070373606877e-05,
      "loss": 2.0779,
      "step": 198153
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.4562804698944092,
      "learning_rate": 2.853896240864678e-05,
      "loss": 2.0403,
      "step": 198154
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.5259777307510376,
      "learning_rate": 2.8537221131694588e-05,
      "loss": 2.2504,
      "step": 198155
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.3477857112884521,
      "learning_rate": 2.8535479905212456e-05,
      "loss": 2.1305,
      "step": 198156
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.4104851484298706,
      "learning_rate": 2.853373872920065e-05,
      "loss": 1.78,
      "step": 198157
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.4928662776947021,
      "learning_rate": 2.8531997603659574e-05,
      "loss": 1.9625,
      "step": 198158
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.5631811618804932,
      "learning_rate": 2.8530256528589555e-05,
      "loss": 2.0023,
      "step": 198159
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.4802838563919067,
      "learning_rate": 2.85285155039908e-05,
      "loss": 1.8366,
      "step": 198160
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.3244315385818481,
      "learning_rate": 2.8526774529863806e-05,
      "loss": 2.2573,
      "step": 198161
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.422742247581482,
      "learning_rate": 2.8525033606208737e-05,
      "loss": 1.9534,
      "step": 198162
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.5096427202224731,
      "learning_rate": 2.8523292733026026e-05,
      "loss": 2.0134,
      "step": 198163
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.4357035160064697,
      "learning_rate": 2.8521551910315976e-05,
      "loss": 1.8131,
      "step": 198164
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.4234098196029663,
      "learning_rate": 2.8519811138078918e-05,
      "loss": 1.9234,
      "step": 198165
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.5188968181610107,
      "learning_rate": 2.8518070416315052e-05,
      "loss": 1.8974,
      "step": 198166
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.4751856327056885,
      "learning_rate": 2.851632974502488e-05,
      "loss": 1.9835,
      "step": 198167
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.4187718629837036,
      "learning_rate": 2.85145891242086e-05,
      "loss": 2.2314,
      "step": 198168
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.5280206203460693,
      "learning_rate": 2.8512848553866642e-05,
      "loss": 1.7704,
      "step": 198169
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.5039067268371582,
      "learning_rate": 2.8511108033999243e-05,
      "loss": 1.9918,
      "step": 198170
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.5225977897644043,
      "learning_rate": 2.8509367564606767e-05,
      "loss": 1.9574,
      "step": 198171
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.5146706104278564,
      "learning_rate": 2.8507627145689482e-05,
      "loss": 2.0246,
      "step": 198172
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.3779208660125732,
      "learning_rate": 2.850588677724779e-05,
      "loss": 2.0658,
      "step": 198173
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.5884267091751099,
      "learning_rate": 2.8504146459281918e-05,
      "loss": 2.0615,
      "step": 198174
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.4410290718078613,
      "learning_rate": 2.850240619179234e-05,
      "loss": 2.1659,
      "step": 198175
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.7054404020309448,
      "learning_rate": 2.850066597477918e-05,
      "loss": 2.1581,
      "step": 198176
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.3336174488067627,
      "learning_rate": 2.849892580824298e-05,
      "loss": 1.966,
      "step": 198177
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.305172085762024,
      "learning_rate": 2.849718569218393e-05,
      "loss": 1.8783,
      "step": 198178
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.5426967144012451,
      "learning_rate": 2.8495445626602375e-05,
      "loss": 2.0323,
      "step": 198179
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.4406906366348267,
      "learning_rate": 2.8493705611498575e-05,
      "loss": 2.2391,
      "step": 198180
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.6837468147277832,
      "learning_rate": 2.8491965646872996e-05,
      "loss": 1.9357,
      "step": 198181
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.4141582250595093,
      "learning_rate": 2.8490225732725836e-05,
      "loss": 2.1791,
      "step": 198182
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.4868450164794922,
      "learning_rate": 2.848848586905753e-05,
      "loss": 1.9847,
      "step": 198183
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.4797252416610718,
      "learning_rate": 2.8486746055868313e-05,
      "loss": 1.9675,
      "step": 198184
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.37924325466156,
      "learning_rate": 2.8485006293158553e-05,
      "loss": 1.9079,
      "step": 198185
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.410420536994934,
      "learning_rate": 2.8483266580928476e-05,
      "loss": 2.0542,
      "step": 198186
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.5894818305969238,
      "learning_rate": 2.8481526919178554e-05,
      "loss": 1.7217,
      "step": 198187
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.4291045665740967,
      "learning_rate": 2.8479787307908984e-05,
      "loss": 2.0001,
      "step": 198188
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.5004525184631348,
      "learning_rate": 2.8478047747120237e-05,
      "loss": 2.0479,
      "step": 198189
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.5523542165756226,
      "learning_rate": 2.8476308236812505e-05,
      "loss": 1.9207,
      "step": 198190
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.4635428190231323,
      "learning_rate": 2.847456877698616e-05,
      "loss": 2.0975,
      "step": 198191
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.3674606084823608,
      "learning_rate": 2.847282936764147e-05,
      "loss": 2.0032,
      "step": 198192
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.5213674306869507,
      "learning_rate": 2.847109000877886e-05,
      "loss": 2.0675,
      "step": 198193
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.614791750907898,
      "learning_rate": 2.8469350700398543e-05,
      "loss": 2.0844,
      "step": 198194
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.5119017362594604,
      "learning_rate": 2.846761144250097e-05,
      "loss": 1.8719,
      "step": 198195
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.6053301095962524,
      "learning_rate": 2.8465872235086383e-05,
      "loss": 2.0254,
      "step": 198196
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.6363964080810547,
      "learning_rate": 2.8464133078155084e-05,
      "loss": 2.0179,
      "step": 198197
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.3519535064697266,
      "learning_rate": 2.846239397170743e-05,
      "loss": 2.0261,
      "step": 198198
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.5617669820785522,
      "learning_rate": 2.8460654915743763e-05,
      "loss": 1.9848,
      "step": 198199
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.5676956176757812,
      "learning_rate": 2.8458915910264313e-05,
      "loss": 1.9324,
      "step": 198200
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.5089250802993774,
      "learning_rate": 2.845717695526958e-05,
      "loss": 2.0471,
      "step": 198201
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.5010358095169067,
      "learning_rate": 2.845543805075976e-05,
      "loss": 1.9205,
      "step": 198202
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.3546439409255981,
      "learning_rate": 2.845369919673519e-05,
      "loss": 2.1521,
      "step": 198203
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.4496105909347534,
      "learning_rate": 2.8451960393196172e-05,
      "loss": 2.1081,
      "step": 198204
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.3727481365203857,
      "learning_rate": 2.8450221640143068e-05,
      "loss": 1.846,
      "step": 198205
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.493114709854126,
      "learning_rate": 2.844848293757618e-05,
      "loss": 1.8576,
      "step": 198206
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.4248744249343872,
      "learning_rate": 2.8446744285495904e-05,
      "loss": 2.1101,
      "step": 198207
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.4318028688430786,
      "learning_rate": 2.844500568390248e-05,
      "loss": 1.8414,
      "step": 198208
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.406422734260559,
      "learning_rate": 2.8443267132796266e-05,
      "loss": 1.9296,
      "step": 198209
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.4149610996246338,
      "learning_rate": 2.8441528632177536e-05,
      "loss": 1.8662,
      "step": 198210
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.6057250499725342,
      "learning_rate": 2.8439790182046684e-05,
      "loss": 2.0541,
      "step": 198211
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.5341782569885254,
      "learning_rate": 2.8438051782403948e-05,
      "loss": 1.9002,
      "step": 198212
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.5212516784667969,
      "learning_rate": 2.843631343324976e-05,
      "loss": 2.1003,
      "step": 198213
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.434858798980713,
      "learning_rate": 2.8434575134584413e-05,
      "loss": 1.8815,
      "step": 198214
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.3680157661437988,
      "learning_rate": 2.8432836886408184e-05,
      "loss": 2.0994,
      "step": 198215
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.5299328565597534,
      "learning_rate": 2.843109868872133e-05,
      "loss": 1.9204,
      "step": 198216
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.3764169216156006,
      "learning_rate": 2.8429360541524358e-05,
      "loss": 1.9202,
      "step": 198217
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.2904844284057617,
      "learning_rate": 2.8427622444817433e-05,
      "loss": 2.0096,
      "step": 198218
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.393918514251709,
      "learning_rate": 2.8425884398600984e-05,
      "loss": 1.8366,
      "step": 198219
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.4035297632217407,
      "learning_rate": 2.8424146402875314e-05,
      "loss": 2.0035,
      "step": 198220
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.529895305633545,
      "learning_rate": 2.842240845764069e-05,
      "loss": 1.9616,
      "step": 198221
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.3690569400787354,
      "learning_rate": 2.842067056289744e-05,
      "loss": 2.176,
      "step": 198222
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.423978567123413,
      "learning_rate": 2.8418932718645938e-05,
      "loss": 1.9651,
      "step": 198223
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.4797697067260742,
      "learning_rate": 2.8417194924886442e-05,
      "loss": 2.3123,
      "step": 198224
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.420231819152832,
      "learning_rate": 2.841545718161936e-05,
      "loss": 1.9851,
      "step": 198225
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.584699273109436,
      "learning_rate": 2.8413719488844988e-05,
      "loss": 2.1706,
      "step": 198226
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.579521894454956,
      "learning_rate": 2.8411981846563626e-05,
      "loss": 1.9167,
      "step": 198227
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.5147719383239746,
      "learning_rate": 2.8410244254775572e-05,
      "loss": 1.905,
      "step": 198228
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.6057329177856445,
      "learning_rate": 2.8408506713481195e-05,
      "loss": 2.0022,
      "step": 198229
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.3919236660003662,
      "learning_rate": 2.840676922268076e-05,
      "loss": 2.034,
      "step": 198230
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.4534969329833984,
      "learning_rate": 2.84050317823747e-05,
      "loss": 2.2207,
      "step": 198231
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.418218970298767,
      "learning_rate": 2.840329439256328e-05,
      "loss": 1.7849,
      "step": 198232
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.4189625978469849,
      "learning_rate": 2.8401557053246803e-05,
      "loss": 1.954,
      "step": 198233
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.441882848739624,
      "learning_rate": 2.8399819764425535e-05,
      "loss": 1.8408,
      "step": 198234
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.5971026420593262,
      "learning_rate": 2.8398082526099942e-05,
      "loss": 2.2401,
      "step": 198235
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.4945365190505981,
      "learning_rate": 2.8396345338270188e-05,
      "loss": 1.9853,
      "step": 198236
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.3729546070098877,
      "learning_rate": 2.8394608200936774e-05,
      "loss": 2.0244,
      "step": 198237
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.4620447158813477,
      "learning_rate": 2.8392871114099903e-05,
      "loss": 1.7079,
      "step": 198238
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.3639289140701294,
      "learning_rate": 2.839113407775987e-05,
      "loss": 2.2341,
      "step": 198239
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.5242259502410889,
      "learning_rate": 2.8389397091917144e-05,
      "loss": 1.7788,
      "step": 198240
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.4760208129882812,
      "learning_rate": 2.8387660156571923e-05,
      "loss": 2.1059,
      "step": 198241
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.3598809242248535,
      "learning_rate": 2.838592327172451e-05,
      "loss": 1.9358,
      "step": 198242
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.4235421419143677,
      "learning_rate": 2.8384186437375332e-05,
      "loss": 1.8864,
      "step": 198243
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.5624363422393799,
      "learning_rate": 2.8382449653524596e-05,
      "loss": 2.0651,
      "step": 198244
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.7043700218200684,
      "learning_rate": 2.8380712920172765e-05,
      "loss": 2.1012,
      "step": 198245
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.4781498908996582,
      "learning_rate": 2.8378976237320073e-05,
      "loss": 2.2236,
      "step": 198246
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.3161356449127197,
      "learning_rate": 2.8377239604966882e-05,
      "loss": 1.8219,
      "step": 198247
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.4386006593704224,
      "learning_rate": 2.83755030231134e-05,
      "loss": 2.0216,
      "step": 198248
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.3194068670272827,
      "learning_rate": 2.837376649176012e-05,
      "loss": 1.9596,
      "step": 198249
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.2930572032928467,
      "learning_rate": 2.8372030010907244e-05,
      "loss": 1.7816,
      "step": 198250
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.4178388118743896,
      "learning_rate": 2.8370293580555138e-05,
      "loss": 1.9874,
      "step": 198251
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.5018877983093262,
      "learning_rate": 2.8368557200704166e-05,
      "loss": 1.9228,
      "step": 198252
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.394284963607788,
      "learning_rate": 2.8366820871354535e-05,
      "loss": 1.894,
      "step": 198253
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.5482425689697266,
      "learning_rate": 2.8365084592506702e-05,
      "loss": 2.2412,
      "step": 198254
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.5200964212417603,
      "learning_rate": 2.836334836416091e-05,
      "loss": 2.2632,
      "step": 198255
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.4776906967163086,
      "learning_rate": 2.836161218631745e-05,
      "loss": 1.9635,
      "step": 198256
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.4940857887268066,
      "learning_rate": 2.835987605897676e-05,
      "loss": 1.9656,
      "step": 198257
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.406901240348816,
      "learning_rate": 2.8358139982139106e-05,
      "loss": 1.8557,
      "step": 198258
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.369291067123413,
      "learning_rate": 2.8356403955804717e-05,
      "loss": 1.7907,
      "step": 198259
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.5142796039581299,
      "learning_rate": 2.8354667979974065e-05,
      "loss": 2.1793,
      "step": 198260
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.349708914756775,
      "learning_rate": 2.8352932054647383e-05,
      "loss": 2.0106,
      "step": 198261
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.3958889245986938,
      "learning_rate": 2.835119617982503e-05,
      "loss": 1.8817,
      "step": 198262
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.6837035417556763,
      "learning_rate": 2.834946035550735e-05,
      "loss": 1.7677,
      "step": 198263
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.474990725517273,
      "learning_rate": 2.8347724581694596e-05,
      "loss": 2.086,
      "step": 198264
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.4256209135055542,
      "learning_rate": 2.8345988858387115e-05,
      "loss": 2.1078,
      "step": 198265
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.38456130027771,
      "learning_rate": 2.8344253185585265e-05,
      "loss": 2.021,
      "step": 198266
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.545224666595459,
      "learning_rate": 2.8342517563289312e-05,
      "loss": 2.2032,
      "step": 198267
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.3858405351638794,
      "learning_rate": 2.834078199149966e-05,
      "loss": 1.8661,
      "step": 198268
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.392458438873291,
      "learning_rate": 2.8339046470216575e-05,
      "loss": 2.0427,
      "step": 198269
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.8286935091018677,
      "learning_rate": 2.833731099944042e-05,
      "loss": 2.2188,
      "step": 198270
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.4378210306167603,
      "learning_rate": 2.83355755791714e-05,
      "loss": 1.8616,
      "step": 198271
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.4291105270385742,
      "learning_rate": 2.8333840209410006e-05,
      "loss": 2.0365,
      "step": 198272
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.4142558574676514,
      "learning_rate": 2.833210489015638e-05,
      "loss": 2.043,
      "step": 198273
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.558448076248169,
      "learning_rate": 2.8330369621411052e-05,
      "loss": 2.0432,
      "step": 198274
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.4501527547836304,
      "learning_rate": 2.832863440317419e-05,
      "loss": 2.2912,
      "step": 198275
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.4033972024917603,
      "learning_rate": 2.8326899235446186e-05,
      "loss": 2.1113,
      "step": 198276
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.420989751815796,
      "learning_rate": 2.8325164118227283e-05,
      "loss": 1.9229,
      "step": 198277
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.5853230953216553,
      "learning_rate": 2.8323429051517943e-05,
      "loss": 1.9963,
      "step": 198278
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.4829238653182983,
      "learning_rate": 2.83216940353183e-05,
      "loss": 2.1225,
      "step": 198279
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.415359616279602,
      "learning_rate": 2.8319959069628884e-05,
      "loss": 2.0233,
      "step": 198280
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.3814934492111206,
      "learning_rate": 2.8318224154449897e-05,
      "loss": 1.9744,
      "step": 198281
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.4016129970550537,
      "learning_rate": 2.8316489289781673e-05,
      "loss": 1.7572,
      "step": 198282
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.4023070335388184,
      "learning_rate": 2.831475447562448e-05,
      "loss": 2.0479,
      "step": 198283
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.4764344692230225,
      "learning_rate": 2.831301971197878e-05,
      "loss": 1.8657,
      "step": 198284
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.5200247764587402,
      "learning_rate": 2.8311284998844742e-05,
      "loss": 1.9531,
      "step": 198285
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.5153568983078003,
      "learning_rate": 2.830955033622283e-05,
      "loss": 1.99,
      "step": 198286
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.4492533206939697,
      "learning_rate": 2.8307815724113313e-05,
      "loss": 2.2622,
      "step": 198287
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.4187943935394287,
      "learning_rate": 2.8306081162516493e-05,
      "loss": 1.9288,
      "step": 198288
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.7358862161636353,
      "learning_rate": 2.830434665143263e-05,
      "loss": 1.8239,
      "step": 198289
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.5078774690628052,
      "learning_rate": 2.8302612190862193e-05,
      "loss": 2.125,
      "step": 198290
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.5693861246109009,
      "learning_rate": 2.8300877780805355e-05,
      "loss": 2.1269,
      "step": 198291
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.4324803352355957,
      "learning_rate": 2.8299143421262572e-05,
      "loss": 1.9667,
      "step": 198292
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.5567599534988403,
      "learning_rate": 2.8297409112234114e-05,
      "loss": 1.8997,
      "step": 198293
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.3891903162002563,
      "learning_rate": 2.8295674853720286e-05,
      "loss": 2.1633,
      "step": 198294
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.5147031545639038,
      "learning_rate": 2.829394064572138e-05,
      "loss": 1.9369,
      "step": 198295
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.7274012565612793,
      "learning_rate": 2.8292206488237834e-05,
      "loss": 2.3135,
      "step": 198296
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.5533814430236816,
      "learning_rate": 2.8290472381269814e-05,
      "loss": 2.0323,
      "step": 198297
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.3663161993026733,
      "learning_rate": 2.8288738324817782e-05,
      "loss": 2.1691,
      "step": 198298
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.4672472476959229,
      "learning_rate": 2.828700431888198e-05,
      "loss": 2.1254,
      "step": 198299
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.5257989168167114,
      "learning_rate": 2.8285270363462796e-05,
      "loss": 2.1171,
      "step": 198300
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.6197690963745117,
      "learning_rate": 2.8283536458560408e-05,
      "loss": 2.1359,
      "step": 198301
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.4724081754684448,
      "learning_rate": 2.828180260417534e-05,
      "loss": 1.9942,
      "step": 198302
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.3759955167770386,
      "learning_rate": 2.8280068800307733e-05,
      "loss": 1.9881,
      "step": 198303
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.3563652038574219,
      "learning_rate": 2.8278335046958045e-05,
      "loss": 1.9696,
      "step": 198304
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.2729837894439697,
      "learning_rate": 2.8276601344126548e-05,
      "loss": 2.0164,
      "step": 198305
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.3676851987838745,
      "learning_rate": 2.8274867691813573e-05,
      "loss": 1.8658,
      "step": 198306
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.3885701894760132,
      "learning_rate": 2.8273134090019357e-05,
      "loss": 2.0592,
      "step": 198307
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.3064653873443604,
      "learning_rate": 2.827140053874436e-05,
      "loss": 2.0438,
      "step": 198308
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.3175219297409058,
      "learning_rate": 2.8269667037988785e-05,
      "loss": 1.911,
      "step": 198309
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.301167368888855,
      "learning_rate": 2.8267933587753066e-05,
      "loss": 2.0607,
      "step": 198310
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.377092957496643,
      "learning_rate": 2.82662001880374e-05,
      "loss": 1.8989,
      "step": 198311
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.460652470588684,
      "learning_rate": 2.826446683884229e-05,
      "loss": 2.2136,
      "step": 198312
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.5680254697799683,
      "learning_rate": 2.8262733540167835e-05,
      "loss": 2.1843,
      "step": 198313
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.432801604270935,
      "learning_rate": 2.8261000292014537e-05,
      "loss": 1.846,
      "step": 198314
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.4611810445785522,
      "learning_rate": 2.8259267094382552e-05,
      "loss": 1.971,
      "step": 198315
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.4455105066299438,
      "learning_rate": 2.8257533947272394e-05,
      "loss": 1.8422,
      "step": 198316
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.491466999053955,
      "learning_rate": 2.825580085068422e-05,
      "loss": 2.1266,
      "step": 198317
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.4487239122390747,
      "learning_rate": 2.825406780461853e-05,
      "loss": 1.9748,
      "step": 198318
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.4607930183410645,
      "learning_rate": 2.825233480907543e-05,
      "loss": 1.8352,
      "step": 198319
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.6767843961715698,
      "learning_rate": 2.8250601864055412e-05,
      "loss": 2.2901,
      "step": 198320
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.4409360885620117,
      "learning_rate": 2.824886896955868e-05,
      "loss": 1.8816,
      "step": 198321
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.5863726139068604,
      "learning_rate": 2.8247136125585666e-05,
      "loss": 2.021,
      "step": 198322
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.420316457748413,
      "learning_rate": 2.8245403332136574e-05,
      "loss": 1.9935,
      "step": 198323
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.4067989587783813,
      "learning_rate": 2.8243670589211863e-05,
      "loss": 2.1642,
      "step": 198324
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.5533978939056396,
      "learning_rate": 2.8241937896811772e-05,
      "loss": 1.9475,
      "step": 198325
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.3479163646697998,
      "learning_rate": 2.8240205254936664e-05,
      "loss": 2.1162,
      "step": 198326
    },
    {
      "epoch": 2.58,
      "grad_norm": 2.302030324935913,
      "learning_rate": 2.8238472663586744e-05,
      "loss": 2.1671,
      "step": 198327
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.360648274421692,
      "learning_rate": 2.8236740122762504e-05,
      "loss": 1.87,
      "step": 198328
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.3798949718475342,
      "learning_rate": 2.8235007632464078e-05,
      "loss": 2.1117,
      "step": 198329
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.364282250404358,
      "learning_rate": 2.8233275192692007e-05,
      "loss": 1.9538,
      "step": 198330
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.4956670999526978,
      "learning_rate": 2.823154280344645e-05,
      "loss": 2.084,
      "step": 198331
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.320723295211792,
      "learning_rate": 2.8229810464727807e-05,
      "loss": 2.043,
      "step": 198332
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.3484301567077637,
      "learning_rate": 2.8228078176536318e-05,
      "loss": 2.0079,
      "step": 198333
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.6173206567764282,
      "learning_rate": 2.8226345938872407e-05,
      "loss": 1.9999,
      "step": 198334
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.3319581747055054,
      "learning_rate": 2.8224613751736313e-05,
      "loss": 1.8615,
      "step": 198335
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.514229655265808,
      "learning_rate": 2.8222881615128433e-05,
      "loss": 2.1133,
      "step": 198336
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.3748573064804077,
      "learning_rate": 2.8221149529049038e-05,
      "loss": 1.9533,
      "step": 198337
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.4000868797302246,
      "learning_rate": 2.821941749349842e-05,
      "loss": 2.083,
      "step": 198338
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.4666391611099243,
      "learning_rate": 2.8217685508476983e-05,
      "loss": 1.8947,
      "step": 198339
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.3947621583938599,
      "learning_rate": 2.821595357398503e-05,
      "loss": 2.0178,
      "step": 198340
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.438696026802063,
      "learning_rate": 2.821422169002282e-05,
      "loss": 2.2106,
      "step": 198341
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.4320564270019531,
      "learning_rate": 2.8212489856590726e-05,
      "loss": 1.7751,
      "step": 198342
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.4160577058792114,
      "learning_rate": 2.8210758073689077e-05,
      "loss": 2.2668,
      "step": 198343
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.334996223449707,
      "learning_rate": 2.820902634131814e-05,
      "loss": 2.1856,
      "step": 198344
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.323835849761963,
      "learning_rate": 2.8207294659478318e-05,
      "loss": 1.8289,
      "step": 198345
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.4218977689743042,
      "learning_rate": 2.820556302816984e-05,
      "loss": 2.1146,
      "step": 198346
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.3893954753875732,
      "learning_rate": 2.8203831447393144e-05,
      "loss": 1.9676,
      "step": 198347
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.282027006149292,
      "learning_rate": 2.8202099917148457e-05,
      "loss": 1.8,
      "step": 198348
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.3309682607650757,
      "learning_rate": 2.820036843743615e-05,
      "loss": 1.9319,
      "step": 198349
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.3722652196884155,
      "learning_rate": 2.8198637008256453e-05,
      "loss": 2.0569,
      "step": 198350
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.4254705905914307,
      "learning_rate": 2.8196905629609833e-05,
      "loss": 2.2433,
      "step": 198351
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.502228021621704,
      "learning_rate": 2.819517430149649e-05,
      "loss": 1.9939,
      "step": 198352
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.5060477256774902,
      "learning_rate": 2.819344302391683e-05,
      "loss": 1.8748,
      "step": 198353
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.4077478647232056,
      "learning_rate": 2.8191711796871175e-05,
      "loss": 1.9713,
      "step": 198354
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.4651522636413574,
      "learning_rate": 2.8189980620359766e-05,
      "loss": 2.3048,
      "step": 198355
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.4873158931732178,
      "learning_rate": 2.818824949438293e-05,
      "loss": 2.1079,
      "step": 198356
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.4398021697998047,
      "learning_rate": 2.8186518418941108e-05,
      "loss": 2.0733,
      "step": 198357
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.4598206281661987,
      "learning_rate": 2.8184787394034458e-05,
      "loss": 2.1003,
      "step": 198358
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.3307602405548096,
      "learning_rate": 2.8183056419663454e-05,
      "loss": 2.0213,
      "step": 198359
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.426285982131958,
      "learning_rate": 2.8181325495828356e-05,
      "loss": 2.126,
      "step": 198360
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.3943655490875244,
      "learning_rate": 2.8179594622529467e-05,
      "loss": 1.8878,
      "step": 198361
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.3624893426895142,
      "learning_rate": 2.8177863799767088e-05,
      "loss": 1.964,
      "step": 198362
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.3867520093917847,
      "learning_rate": 2.8176133027541615e-05,
      "loss": 1.969,
      "step": 198363
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.6063822507858276,
      "learning_rate": 2.8174402305853282e-05,
      "loss": 1.8697,
      "step": 198364
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.4271262884140015,
      "learning_rate": 2.8172671634702492e-05,
      "loss": 1.8663,
      "step": 198365
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.4536410570144653,
      "learning_rate": 2.8170941014089543e-05,
      "loss": 2.015,
      "step": 198366
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.4122625589370728,
      "learning_rate": 2.8169210444014768e-05,
      "loss": 1.8654,
      "step": 198367
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.2997924089431763,
      "learning_rate": 2.8167479924478396e-05,
      "loss": 1.8763,
      "step": 198368
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.5171990394592285,
      "learning_rate": 2.816574945548087e-05,
      "loss": 2.039,
      "step": 198369
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.5401053428649902,
      "learning_rate": 2.8164019037022413e-05,
      "loss": 2.0522,
      "step": 198370
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.4469574689865112,
      "learning_rate": 2.816228866910346e-05,
      "loss": 1.9705,
      "step": 198371
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.3928141593933105,
      "learning_rate": 2.8160558351724283e-05,
      "loss": 1.8143,
      "step": 198372
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.5932230949401855,
      "learning_rate": 2.8158828084885175e-05,
      "loss": 2.063,
      "step": 198373
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.3106961250305176,
      "learning_rate": 2.815709786858641e-05,
      "loss": 2.0797,
      "step": 198374
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.4388843774795532,
      "learning_rate": 2.815536770282841e-05,
      "loss": 1.9801,
      "step": 198375
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.4855817556381226,
      "learning_rate": 2.8153637587611423e-05,
      "loss": 1.9555,
      "step": 198376
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.4989609718322754,
      "learning_rate": 2.8151907522935868e-05,
      "loss": 1.8802,
      "step": 198377
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.5827271938323975,
      "learning_rate": 2.8150177508801952e-05,
      "loss": 1.864,
      "step": 198378
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.4180231094360352,
      "learning_rate": 2.814844754521014e-05,
      "loss": 2.121,
      "step": 198379
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.334803819656372,
      "learning_rate": 2.8146717632160566e-05,
      "loss": 1.8078,
      "step": 198380
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.3940051794052124,
      "learning_rate": 2.8144987769653692e-05,
      "loss": 2.0443,
      "step": 198381
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.4241271018981934,
      "learning_rate": 2.8143257957689756e-05,
      "loss": 1.8627,
      "step": 198382
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.4769854545593262,
      "learning_rate": 2.814152819626919e-05,
      "loss": 1.8505,
      "step": 198383
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.6582584381103516,
      "learning_rate": 2.813979848539216e-05,
      "loss": 1.9909,
      "step": 198384
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.3350661993026733,
      "learning_rate": 2.81380688250592e-05,
      "loss": 2.0532,
      "step": 198385
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.4680068492889404,
      "learning_rate": 2.8136339215270375e-05,
      "loss": 1.8881,
      "step": 198386
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.439427375793457,
      "learning_rate": 2.813460965602622e-05,
      "loss": 1.987,
      "step": 198387
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.3895390033721924,
      "learning_rate": 2.81328801473269e-05,
      "loss": 1.8675,
      "step": 198388
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.4106252193450928,
      "learning_rate": 2.8131150689172876e-05,
      "loss": 1.8549,
      "step": 198389
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.3982874155044556,
      "learning_rate": 2.8129421281564326e-05,
      "loss": 1.9166,
      "step": 198390
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.4022918939590454,
      "learning_rate": 2.812769192450177e-05,
      "loss": 1.7342,
      "step": 198391
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.4147894382476807,
      "learning_rate": 2.812596261798532e-05,
      "loss": 2.1981,
      "step": 198392
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.3086124658584595,
      "learning_rate": 2.81242333620154e-05,
      "loss": 1.951,
      "step": 198393
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.4565174579620361,
      "learning_rate": 2.8122504156592283e-05,
      "loss": 1.8379,
      "step": 198394
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.4924789667129517,
      "learning_rate": 2.81207750017164e-05,
      "loss": 2.2396,
      "step": 198395
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.4655762910842896,
      "learning_rate": 2.8119045897387914e-05,
      "loss": 1.8764,
      "step": 198396
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.420425295829773,
      "learning_rate": 2.811731684360733e-05,
      "loss": 1.8478,
      "step": 198397
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.4193782806396484,
      "learning_rate": 2.8115587840374777e-05,
      "loss": 2.0462,
      "step": 198398
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.5222915410995483,
      "learning_rate": 2.8113858887690722e-05,
      "loss": 1.9065,
      "step": 198399
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.4096131324768066,
      "learning_rate": 2.8112129985555398e-05,
      "loss": 2.0391,
      "step": 198400
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.4523247480392456,
      "learning_rate": 2.8110401133969174e-05,
      "loss": 2.0323,
      "step": 198401
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.3569220304489136,
      "learning_rate": 2.810867233293235e-05,
      "loss": 2.0431,
      "step": 198402
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.3427077531814575,
      "learning_rate": 2.8106943582445318e-05,
      "loss": 2.0941,
      "step": 198403
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.4957479238510132,
      "learning_rate": 2.8105214882508254e-05,
      "loss": 1.9933,
      "step": 198404
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.4508999586105347,
      "learning_rate": 2.810348623312162e-05,
      "loss": 2.1617,
      "step": 198405
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.4302897453308105,
      "learning_rate": 2.8101757634285615e-05,
      "loss": 1.8433,
      "step": 198406
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.4025461673736572,
      "learning_rate": 2.8100029086000707e-05,
      "loss": 1.992,
      "step": 198407
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.4263088703155518,
      "learning_rate": 2.8098300588267064e-05,
      "loss": 1.9995,
      "step": 198408
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.4401044845581055,
      "learning_rate": 2.8096572141085116e-05,
      "loss": 2.0093,
      "step": 198409
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.528820514678955,
      "learning_rate": 2.8094843744455165e-05,
      "loss": 1.9756,
      "step": 198410
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.4203810691833496,
      "learning_rate": 2.809311539837751e-05,
      "loss": 1.9455,
      "step": 198411
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.409727931022644,
      "learning_rate": 2.8091387102852447e-05,
      "loss": 1.9347,
      "step": 198412
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.411974310874939,
      "learning_rate": 2.8089658857880348e-05,
      "loss": 2.0998,
      "step": 198413
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.5092977285385132,
      "learning_rate": 2.8087930663461478e-05,
      "loss": 2.1218,
      "step": 198414
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.4202848672866821,
      "learning_rate": 2.8086202519596235e-05,
      "loss": 1.8087,
      "step": 198415
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.3690836429595947,
      "learning_rate": 2.808447442628492e-05,
      "loss": 1.9903,
      "step": 198416
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.4660601615905762,
      "learning_rate": 2.8082746383527832e-05,
      "loss": 2.1121,
      "step": 198417
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.3585841655731201,
      "learning_rate": 2.8081018391325204e-05,
      "loss": 1.8758,
      "step": 198418
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.5324825048446655,
      "learning_rate": 2.8079290449677537e-05,
      "loss": 1.8696,
      "step": 198419
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.4344412088394165,
      "learning_rate": 2.8077562558584998e-05,
      "loss": 2.3725,
      "step": 198420
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.4221149682998657,
      "learning_rate": 2.807583471804805e-05,
      "loss": 1.9794,
      "step": 198421
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.346835970878601,
      "learning_rate": 2.8074106928066898e-05,
      "loss": 1.8804,
      "step": 198422
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.489312767982483,
      "learning_rate": 2.807237918864187e-05,
      "loss": 2.1155,
      "step": 198423
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.4705768823623657,
      "learning_rate": 2.8070651499773367e-05,
      "loss": 2.0147,
      "step": 198424
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.4755932092666626,
      "learning_rate": 2.806892386146169e-05,
      "loss": 1.9491,
      "step": 198425
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.4644099473953247,
      "learning_rate": 2.806719627370704e-05,
      "loss": 2.167,
      "step": 198426
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.4359652996063232,
      "learning_rate": 2.806546873650991e-05,
      "loss": 1.873,
      "step": 198427
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.5343947410583496,
      "learning_rate": 2.806374124987054e-05,
      "loss": 1.9115,
      "step": 198428
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.3533166646957397,
      "learning_rate": 2.8062013813789197e-05,
      "loss": 1.9663,
      "step": 198429
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.2916831970214844,
      "learning_rate": 2.806028642826631e-05,
      "loss": 1.8834,
      "step": 198430
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.3382649421691895,
      "learning_rate": 2.8058559093302147e-05,
      "loss": 1.9227,
      "step": 198431
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.3012995719909668,
      "learning_rate": 2.8056831808896973e-05,
      "loss": 1.9452,
      "step": 198432
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.3486595153808594,
      "learning_rate": 2.8055104575051223e-05,
      "loss": 2.1234,
      "step": 198433
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.474277138710022,
      "learning_rate": 2.8053377391765164e-05,
      "loss": 2.0605,
      "step": 198434
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.3953359127044678,
      "learning_rate": 2.805165025903906e-05,
      "loss": 1.9404,
      "step": 198435
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.4694331884384155,
      "learning_rate": 2.8049923176873344e-05,
      "loss": 2.0392,
      "step": 198436
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.326198697090149,
      "learning_rate": 2.804819614526822e-05,
      "loss": 1.921,
      "step": 198437
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.4764294624328613,
      "learning_rate": 2.804646916422415e-05,
      "loss": 1.8842,
      "step": 198438
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.434287667274475,
      "learning_rate": 2.804474223374137e-05,
      "loss": 2.1259,
      "step": 198439
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.4519065618515015,
      "learning_rate": 2.8043015353820173e-05,
      "loss": 2.1181,
      "step": 198440
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.5253652334213257,
      "learning_rate": 2.804128852446087e-05,
      "loss": 2.0796,
      "step": 198441
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.5045363903045654,
      "learning_rate": 2.803956174566392e-05,
      "loss": 1.7385,
      "step": 198442
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.3754373788833618,
      "learning_rate": 2.8037835017429455e-05,
      "loss": 2.0363,
      "step": 198443
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.348724603652954,
      "learning_rate": 2.8036108339757945e-05,
      "loss": 1.844,
      "step": 198444
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.2973679304122925,
      "learning_rate": 2.8034381712649624e-05,
      "loss": 2.0098,
      "step": 198445
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.367285132408142,
      "learning_rate": 2.803265513610492e-05,
      "loss": 2.0164,
      "step": 198446
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.4372913837432861,
      "learning_rate": 2.8030928610124005e-05,
      "loss": 2.0735,
      "step": 198447
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.61031973361969,
      "learning_rate": 2.8029202134707307e-05,
      "loss": 2.1216,
      "step": 198448
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.3352981805801392,
      "learning_rate": 2.8027475709855064e-05,
      "loss": 2.0683,
      "step": 198449
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.4327406883239746,
      "learning_rate": 2.8025749335567703e-05,
      "loss": 1.6404,
      "step": 198450
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.4326543807983398,
      "learning_rate": 2.802402301184543e-05,
      "loss": 2.0174,
      "step": 198451
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.5768320560455322,
      "learning_rate": 2.8022296738688743e-05,
      "loss": 2.0801,
      "step": 198452
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.490044116973877,
      "learning_rate": 2.802057051609774e-05,
      "loss": 2.0525,
      "step": 198453
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.3816512823104858,
      "learning_rate": 2.8018844344072888e-05,
      "loss": 1.9447,
      "step": 198454
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.5674099922180176,
      "learning_rate": 2.801711822261442e-05,
      "loss": 2.022,
      "step": 198455
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.6350059509277344,
      "learning_rate": 2.8015392151722772e-05,
      "loss": 2.1393,
      "step": 198456
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.488453984260559,
      "learning_rate": 2.8013666131398106e-05,
      "loss": 1.8637,
      "step": 198457
    },
    {
      "epoch": 2.58,
      "grad_norm": 2.277723550796509,
      "learning_rate": 2.8011940161640955e-05,
      "loss": 1.9055,
      "step": 198458
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.2565397024154663,
      "learning_rate": 2.8010214242451424e-05,
      "loss": 2.1418,
      "step": 198459
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.4634343385696411,
      "learning_rate": 2.8008488373829973e-05,
      "loss": 1.8501,
      "step": 198460
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.5343272686004639,
      "learning_rate": 2.800676255577684e-05,
      "loss": 1.9987,
      "step": 198461
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.6454070806503296,
      "learning_rate": 2.8005036788292424e-05,
      "loss": 2.0495,
      "step": 198462
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.5159591436386108,
      "learning_rate": 2.8003311071376956e-05,
      "loss": 2.071,
      "step": 198463
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.44575035572052,
      "learning_rate": 2.8001585405030903e-05,
      "loss": 1.9287,
      "step": 198464
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.4255576133728027,
      "learning_rate": 2.7999859789254398e-05,
      "loss": 1.8452,
      "step": 198465
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.452518343925476,
      "learning_rate": 2.7998134224047907e-05,
      "loss": 1.8286,
      "step": 198466
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.5361018180847168,
      "learning_rate": 2.7996408709411632e-05,
      "loss": 1.9555,
      "step": 198467
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.4850821495056152,
      "learning_rate": 2.7994683245346005e-05,
      "loss": 2.1143,
      "step": 198468
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.5925244092941284,
      "learning_rate": 2.799295783185126e-05,
      "loss": 1.8983,
      "step": 198469
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.4728026390075684,
      "learning_rate": 2.7991232468927893e-05,
      "loss": 2.0113,
      "step": 198470
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.5672394037246704,
      "learning_rate": 2.798950715657594e-05,
      "loss": 2.1287,
      "step": 198471
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.5832546949386597,
      "learning_rate": 2.7987781894795934e-05,
      "loss": 1.9551,
      "step": 198472
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.5040968656539917,
      "learning_rate": 2.798605668358811e-05,
      "loss": 2.1043,
      "step": 198473
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.5530680418014526,
      "learning_rate": 2.798433152295283e-05,
      "loss": 2.006,
      "step": 198474
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.5151625871658325,
      "learning_rate": 2.7982606412890362e-05,
      "loss": 1.8163,
      "step": 198475
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.6340155601501465,
      "learning_rate": 2.7980881353401173e-05,
      "loss": 2.0055,
      "step": 198476
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.5134268999099731,
      "learning_rate": 2.7979156344485367e-05,
      "loss": 2.1611,
      "step": 198477
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.518376111984253,
      "learning_rate": 2.79774313861434e-05,
      "loss": 2.1617,
      "step": 198478
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.3630616664886475,
      "learning_rate": 2.7975706478375515e-05,
      "loss": 2.1044,
      "step": 198479
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.2990703582763672,
      "learning_rate": 2.7973981621182174e-05,
      "loss": 1.956,
      "step": 198480
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.4683197736740112,
      "learning_rate": 2.797225681456351e-05,
      "loss": 2.0261,
      "step": 198481
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.3554284572601318,
      "learning_rate": 2.7970532058520056e-05,
      "loss": 1.8754,
      "step": 198482
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.3768742084503174,
      "learning_rate": 2.7968807353051913e-05,
      "loss": 2.0525,
      "step": 198483
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.6024894714355469,
      "learning_rate": 2.7967082698159547e-05,
      "loss": 2.0877,
      "step": 198484
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.4490158557891846,
      "learning_rate": 2.7965358093843194e-05,
      "loss": 1.9987,
      "step": 198485
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.3698917627334595,
      "learning_rate": 2.796363354010328e-05,
      "loss": 1.9675,
      "step": 198486
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.399388074874878,
      "learning_rate": 2.7961909036939976e-05,
      "loss": 2.0694,
      "step": 198487
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.4214723110198975,
      "learning_rate": 2.7960184584353817e-05,
      "loss": 2.0289,
      "step": 198488
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.4531984329223633,
      "learning_rate": 2.7958460182344866e-05,
      "loss": 2.1231,
      "step": 198489
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.4089947938919067,
      "learning_rate": 2.7956735830913658e-05,
      "loss": 1.9304,
      "step": 198490
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.340863585472107,
      "learning_rate": 2.7955011530060356e-05,
      "loss": 1.9221,
      "step": 198491
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.386733055114746,
      "learning_rate": 2.795328727978543e-05,
      "loss": 1.7932,
      "step": 198492
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.3454298973083496,
      "learning_rate": 2.7951563080089046e-05,
      "loss": 1.8864,
      "step": 198493
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.4062589406967163,
      "learning_rate": 2.7949838930971702e-05,
      "loss": 1.9655,
      "step": 198494
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.3853753805160522,
      "learning_rate": 2.7948114832433534e-05,
      "loss": 1.8922,
      "step": 198495
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.5447174310684204,
      "learning_rate": 2.794639078447497e-05,
      "loss": 2.0489,
      "step": 198496
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.3649121522903442,
      "learning_rate": 2.794466678709628e-05,
      "loss": 1.883,
      "step": 198497
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.373365879058838,
      "learning_rate": 2.7942942840297868e-05,
      "loss": 2.1575,
      "step": 198498
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.4118250608444214,
      "learning_rate": 2.7941218944079923e-05,
      "loss": 2.094,
      "step": 198499
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.424575924873352,
      "learning_rate": 2.7939495098442886e-05,
      "loss": 1.7285,
      "step": 198500
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.4777284860610962,
      "learning_rate": 2.793777130338706e-05,
      "loss": 2.0162,
      "step": 198501
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.4582207202911377,
      "learning_rate": 2.7936047558912732e-05,
      "loss": 1.9394,
      "step": 198502
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.3500442504882812,
      "learning_rate": 2.793432386502018e-05,
      "loss": 2.1239,
      "step": 198503
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.4308379888534546,
      "learning_rate": 2.79326002217098e-05,
      "loss": 2.1432,
      "step": 198504
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.4638468027114868,
      "learning_rate": 2.7930876628981857e-05,
      "loss": 1.9541,
      "step": 198505
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.5764437913894653,
      "learning_rate": 2.7929153086836752e-05,
      "loss": 2.14,
      "step": 198506
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.4002149105072021,
      "learning_rate": 2.7927429595274754e-05,
      "loss": 1.8247,
      "step": 198507
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.4697469472885132,
      "learning_rate": 2.7925706154296124e-05,
      "loss": 2.0394,
      "step": 198508
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.374572992324829,
      "learning_rate": 2.7923982763901298e-05,
      "loss": 1.9584,
      "step": 198509
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.4987876415252686,
      "learning_rate": 2.792225942409051e-05,
      "loss": 2.0436,
      "step": 198510
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.4703115224838257,
      "learning_rate": 2.792053613486409e-05,
      "loss": 2.0757,
      "step": 198511
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.4071460962295532,
      "learning_rate": 2.7918812896222408e-05,
      "loss": 1.9094,
      "step": 198512
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.5634881258010864,
      "learning_rate": 2.7917089708165796e-05,
      "loss": 1.94,
      "step": 198513
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.3107243776321411,
      "learning_rate": 2.7915366570694453e-05,
      "loss": 2.0174,
      "step": 198514
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.5554476976394653,
      "learning_rate": 2.791364348380881e-05,
      "loss": 2.0563,
      "step": 198515
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.4250693321228027,
      "learning_rate": 2.7911920447509205e-05,
      "loss": 1.9892,
      "step": 198516
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.4224261045455933,
      "learning_rate": 2.79101974617958e-05,
      "loss": 2.2478,
      "step": 198517
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.5354069471359253,
      "learning_rate": 2.7908474526669133e-05,
      "loss": 2.2928,
      "step": 198518
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.315451979637146,
      "learning_rate": 2.79067516421294e-05,
      "loss": 1.9959,
      "step": 198519
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.3564400672912598,
      "learning_rate": 2.7905028808176867e-05,
      "loss": 2.083,
      "step": 198520
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.5589977502822876,
      "learning_rate": 2.7903306024812e-05,
      "loss": 1.9976,
      "step": 198521
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.4601634740829468,
      "learning_rate": 2.790158329203497e-05,
      "loss": 1.8418,
      "step": 198522
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.5738636255264282,
      "learning_rate": 2.7899860609846237e-05,
      "loss": 1.8684,
      "step": 198523
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.5274957418441772,
      "learning_rate": 2.789813797824607e-05,
      "loss": 1.8654,
      "step": 198524
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.5690032243728638,
      "learning_rate": 2.789641539723474e-05,
      "loss": 2.0345,
      "step": 198525
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.3981248140335083,
      "learning_rate": 2.7894692866812573e-05,
      "loss": 1.7557,
      "step": 198526
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.5561726093292236,
      "learning_rate": 2.7892970386979975e-05,
      "loss": 2.0128,
      "step": 198527
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.453076958656311,
      "learning_rate": 2.7891247957737138e-05,
      "loss": 1.9545,
      "step": 198528
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.3298181295394897,
      "learning_rate": 2.7889525579084537e-05,
      "loss": 1.8227,
      "step": 198529
    },
    {
      "epoch": 2.58,
      "grad_norm": 1.4511159658432007,
      "learning_rate": 2.78878032510224e-05,
      "loss": 1.9208,
      "step": 198530
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.4962375164031982,
      "learning_rate": 2.7886080973551063e-05,
      "loss": 2.1732,
      "step": 198531
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.3693972826004028,
      "learning_rate": 2.7884358746670753e-05,
      "loss": 2.1309,
      "step": 198532
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.4972717761993408,
      "learning_rate": 2.7882636570381946e-05,
      "loss": 1.9996,
      "step": 198533
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.407660722732544,
      "learning_rate": 2.7880914444684865e-05,
      "loss": 2.0185,
      "step": 198534
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.4292563199996948,
      "learning_rate": 2.7879192369579883e-05,
      "loss": 1.9677,
      "step": 198535
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.4947364330291748,
      "learning_rate": 2.7877470345067267e-05,
      "loss": 2.0881,
      "step": 198536
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.4340052604675293,
      "learning_rate": 2.7875748371147445e-05,
      "loss": 1.9513,
      "step": 198537
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.396624207496643,
      "learning_rate": 2.787402644782055e-05,
      "loss": 1.9815,
      "step": 198538
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.449391484260559,
      "learning_rate": 2.787230457508709e-05,
      "loss": 2.0752,
      "step": 198539
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.4742302894592285,
      "learning_rate": 2.7870582752947256e-05,
      "loss": 1.9035,
      "step": 198540
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.358201265335083,
      "learning_rate": 2.7868860981401454e-05,
      "loss": 1.9725,
      "step": 198541
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.5534709692001343,
      "learning_rate": 2.7867139260449912e-05,
      "loss": 1.8999,
      "step": 198542
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.4573780298233032,
      "learning_rate": 2.7865417590093132e-05,
      "loss": 1.9803,
      "step": 198543
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.504015564918518,
      "learning_rate": 2.7863695970331178e-05,
      "loss": 1.9786,
      "step": 198544
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.4898560047149658,
      "learning_rate": 2.7861974401164556e-05,
      "loss": 2.145,
      "step": 198545
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.6910308599472046,
      "learning_rate": 2.7860252882593457e-05,
      "loss": 1.9841,
      "step": 198546
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.4319607019424438,
      "learning_rate": 2.7858531414618358e-05,
      "loss": 1.957,
      "step": 198547
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.4187439680099487,
      "learning_rate": 2.785680999723945e-05,
      "loss": 1.8898,
      "step": 198548
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.4547991752624512,
      "learning_rate": 2.78550886304572e-05,
      "loss": 2.1376,
      "step": 198549
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.4437185525894165,
      "learning_rate": 2.7853367314271714e-05,
      "loss": 2.2249,
      "step": 198550
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.422052264213562,
      "learning_rate": 2.785164604868345e-05,
      "loss": 2.0859,
      "step": 198551
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.4070614576339722,
      "learning_rate": 2.784992483369268e-05,
      "loss": 2.0687,
      "step": 198552
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.5049424171447754,
      "learning_rate": 2.7848203669299774e-05,
      "loss": 2.1305,
      "step": 198553
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.4565569162368774,
      "learning_rate": 2.784648255550499e-05,
      "loss": 2.1252,
      "step": 198554
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.457737684249878,
      "learning_rate": 2.7844761492308798e-05,
      "loss": 1.9451,
      "step": 198555
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.460344672203064,
      "learning_rate": 2.78430404797113e-05,
      "loss": 1.8116,
      "step": 198556
    },
    {
      "epoch": 2.59,
      "grad_norm": 2.1246769428253174,
      "learning_rate": 2.7841319517712957e-05,
      "loss": 2.1661,
      "step": 198557
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.3706958293914795,
      "learning_rate": 2.7839598606313974e-05,
      "loss": 1.9605,
      "step": 198558
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.4008991718292236,
      "learning_rate": 2.7837877745514846e-05,
      "loss": 1.6886,
      "step": 198559
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.4297715425491333,
      "learning_rate": 2.783615693531571e-05,
      "loss": 2.0491,
      "step": 198560
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.443983554840088,
      "learning_rate": 2.7834436175717067e-05,
      "loss": 1.9323,
      "step": 198561
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.5154132843017578,
      "learning_rate": 2.7832715466719047e-05,
      "loss": 2.0605,
      "step": 198562
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.4224728345870972,
      "learning_rate": 2.7830994808322148e-05,
      "loss": 1.8504,
      "step": 198563
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.4112684726715088,
      "learning_rate": 2.782927420052651e-05,
      "loss": 2.0161,
      "step": 198564
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.3959510326385498,
      "learning_rate": 2.7827553643332622e-05,
      "loss": 2.0687,
      "step": 198565
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.509131669998169,
      "learning_rate": 2.782583313674066e-05,
      "loss": 1.9566,
      "step": 198566
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.4467718601226807,
      "learning_rate": 2.782411268075112e-05,
      "loss": 2.0921,
      "step": 198567
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.6376302242279053,
      "learning_rate": 2.7822392275364104e-05,
      "loss": 2.0526,
      "step": 198568
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.3784432411193848,
      "learning_rate": 2.782067192058014e-05,
      "loss": 1.887,
      "step": 198569
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.4041966199874878,
      "learning_rate": 2.7818951616399332e-05,
      "loss": 1.8199,
      "step": 198570
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.3710678815841675,
      "learning_rate": 2.7817231362822213e-05,
      "loss": 1.7811,
      "step": 198571
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.376572847366333,
      "learning_rate": 2.7815511159848946e-05,
      "loss": 1.8695,
      "step": 198572
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.5830124616622925,
      "learning_rate": 2.7813791007480036e-05,
      "loss": 1.8735,
      "step": 198573
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.592782974243164,
      "learning_rate": 2.7812070905715545e-05,
      "loss": 2.0862,
      "step": 198574
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.3352183103561401,
      "learning_rate": 2.7810350854556006e-05,
      "loss": 2.0756,
      "step": 198575
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.4202619791030884,
      "learning_rate": 2.780863085400159e-05,
      "loss": 1.8334,
      "step": 198576
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.5343753099441528,
      "learning_rate": 2.780691090405276e-05,
      "loss": 1.8095,
      "step": 198577
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.3404192924499512,
      "learning_rate": 2.780519100470968e-05,
      "loss": 1.9289,
      "step": 198578
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.4151718616485596,
      "learning_rate": 2.7803471155972823e-05,
      "loss": 2.2303,
      "step": 198579
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.4819473028182983,
      "learning_rate": 2.780175135784245e-05,
      "loss": 2.0143,
      "step": 198580
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.3200886249542236,
      "learning_rate": 2.7800031610318863e-05,
      "loss": 2.0371,
      "step": 198581
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.5497092008590698,
      "learning_rate": 2.7798311913402327e-05,
      "loss": 1.8059,
      "step": 198582
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.600119948387146,
      "learning_rate": 2.7796592267093275e-05,
      "loss": 2.2359,
      "step": 198583
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.5694310665130615,
      "learning_rate": 2.7794872671391942e-05,
      "loss": 2.1705,
      "step": 198584
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.4219059944152832,
      "learning_rate": 2.779315312629873e-05,
      "loss": 2.1266,
      "step": 198585
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.4714469909667969,
      "learning_rate": 2.7791433631813896e-05,
      "loss": 1.8774,
      "step": 198586
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.4150234460830688,
      "learning_rate": 2.778971418793775e-05,
      "loss": 2.1281,
      "step": 198587
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.6035517454147339,
      "learning_rate": 2.778799479467062e-05,
      "loss": 2.0582,
      "step": 198588
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.335881233215332,
      "learning_rate": 2.7786275452012875e-05,
      "loss": 2.0111,
      "step": 198589
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.4156804084777832,
      "learning_rate": 2.7784556159964745e-05,
      "loss": 2.1694,
      "step": 198590
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.6279535293579102,
      "learning_rate": 2.7782836918526664e-05,
      "loss": 2.191,
      "step": 198591
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.6370797157287598,
      "learning_rate": 2.7781117727698897e-05,
      "loss": 2.0506,
      "step": 198592
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.6042721271514893,
      "learning_rate": 2.777939858748175e-05,
      "loss": 1.9911,
      "step": 198593
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.399430751800537,
      "learning_rate": 2.7777679497875517e-05,
      "loss": 1.8968,
      "step": 198594
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.4900227785110474,
      "learning_rate": 2.7775960458880598e-05,
      "loss": 1.9028,
      "step": 198595
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.3462361097335815,
      "learning_rate": 2.7774241470497195e-05,
      "loss": 1.9574,
      "step": 198596
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.453983187675476,
      "learning_rate": 2.7772522532725772e-05,
      "loss": 2.1163,
      "step": 198597
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.411001205444336,
      "learning_rate": 2.7770803645566564e-05,
      "loss": 2.1831,
      "step": 198598
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.6923418045043945,
      "learning_rate": 2.7769084809019838e-05,
      "loss": 1.9911,
      "step": 198599
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.4679465293884277,
      "learning_rate": 2.7767366023086057e-05,
      "loss": 1.8977,
      "step": 198600
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.528082251548767,
      "learning_rate": 2.7765647287765457e-05,
      "loss": 2.0732,
      "step": 198601
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.4264569282531738,
      "learning_rate": 2.7763928603058273e-05,
      "loss": 2.0221,
      "step": 198602
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.5724854469299316,
      "learning_rate": 2.7762209968964998e-05,
      "loss": 1.8709,
      "step": 198603
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.39519464969635,
      "learning_rate": 2.776049138548587e-05,
      "loss": 2.0482,
      "step": 198604
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.3747637271881104,
      "learning_rate": 2.775877285262116e-05,
      "loss": 1.8851,
      "step": 198605
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.420910358428955,
      "learning_rate": 2.7757054370371258e-05,
      "loss": 1.869,
      "step": 198606
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.6707450151443481,
      "learning_rate": 2.7755335938736435e-05,
      "loss": 2.1738,
      "step": 198607
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.4725700616836548,
      "learning_rate": 2.7753617557717055e-05,
      "loss": 2.2017,
      "step": 198608
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.3624080419540405,
      "learning_rate": 2.7751899227313456e-05,
      "loss": 1.6742,
      "step": 198609
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.460785150527954,
      "learning_rate": 2.7750180947525867e-05,
      "loss": 1.8886,
      "step": 198610
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.3652092218399048,
      "learning_rate": 2.774846271835466e-05,
      "loss": 2.0981,
      "step": 198611
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.4910520315170288,
      "learning_rate": 2.774674453980016e-05,
      "loss": 1.9662,
      "step": 198612
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.4634337425231934,
      "learning_rate": 2.774502641186267e-05,
      "loss": 2.0632,
      "step": 198613
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.4506018161773682,
      "learning_rate": 2.7743308334542525e-05,
      "loss": 1.7987,
      "step": 198614
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.4436962604522705,
      "learning_rate": 2.774159030784009e-05,
      "loss": 2.0349,
      "step": 198615
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.5285884141921997,
      "learning_rate": 2.7739872331755597e-05,
      "loss": 2.0216,
      "step": 198616
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.5430841445922852,
      "learning_rate": 2.7738154406289347e-05,
      "loss": 1.8905,
      "step": 198617
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.2790840864181519,
      "learning_rate": 2.7736436531441776e-05,
      "loss": 1.9756,
      "step": 198618
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.5611238479614258,
      "learning_rate": 2.7734718707213076e-05,
      "loss": 1.8832,
      "step": 198619
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.3382073640823364,
      "learning_rate": 2.773300093360369e-05,
      "loss": 1.9729,
      "step": 198620
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.3474146127700806,
      "learning_rate": 2.773128321061384e-05,
      "loss": 1.8701,
      "step": 198621
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.3902820348739624,
      "learning_rate": 2.772956553824397e-05,
      "loss": 1.952,
      "step": 198622
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.3182212114334106,
      "learning_rate": 2.7727847916494237e-05,
      "loss": 1.8044,
      "step": 198623
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.4508475065231323,
      "learning_rate": 2.772613034536505e-05,
      "loss": 2.0661,
      "step": 198624
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.4673972129821777,
      "learning_rate": 2.77244128248567e-05,
      "loss": 1.8794,
      "step": 198625
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.3579493761062622,
      "learning_rate": 2.772269535496956e-05,
      "loss": 1.9708,
      "step": 198626
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.3945386409759521,
      "learning_rate": 2.7720977935703858e-05,
      "loss": 1.9777,
      "step": 198627
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.4265154600143433,
      "learning_rate": 2.7719260567060065e-05,
      "loss": 2.0381,
      "step": 198628
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.5917232036590576,
      "learning_rate": 2.7717543249038276e-05,
      "loss": 2.1859,
      "step": 198629
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.5380160808563232,
      "learning_rate": 2.7715825981639027e-05,
      "loss": 1.8383,
      "step": 198630
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.4645826816558838,
      "learning_rate": 2.7714108764862486e-05,
      "loss": 2.0201,
      "step": 198631
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.4601048231124878,
      "learning_rate": 2.7712391598709048e-05,
      "loss": 1.8171,
      "step": 198632
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.4357651472091675,
      "learning_rate": 2.7710674483178985e-05,
      "loss": 2.1097,
      "step": 198633
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.4149309396743774,
      "learning_rate": 2.770895741827279e-05,
      "loss": 2.0458,
      "step": 198634
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.425915002822876,
      "learning_rate": 2.770724040399047e-05,
      "loss": 2.0585,
      "step": 198635
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.428808331489563,
      "learning_rate": 2.7705523440332623e-05,
      "loss": 2.0589,
      "step": 198636
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.54005765914917,
      "learning_rate": 2.7703806527299376e-05,
      "loss": 1.9026,
      "step": 198637
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.4439940452575684,
      "learning_rate": 2.7702089664891204e-05,
      "loss": 1.8578,
      "step": 198638
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.386139154434204,
      "learning_rate": 2.7700372853108267e-05,
      "loss": 1.929,
      "step": 198639
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.4499064683914185,
      "learning_rate": 2.76986560919511e-05,
      "loss": 1.9283,
      "step": 198640
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.3700511455535889,
      "learning_rate": 2.7696939381419803e-05,
      "loss": 1.958,
      "step": 198641
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.4350740909576416,
      "learning_rate": 2.769522272151481e-05,
      "loss": 1.9075,
      "step": 198642
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.4684300422668457,
      "learning_rate": 2.7693506112236353e-05,
      "loss": 1.9864,
      "step": 198643
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.352641224861145,
      "learning_rate": 2.76917895535849e-05,
      "loss": 2.0226,
      "step": 198644
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.480826735496521,
      "learning_rate": 2.7690073045560612e-05,
      "loss": 2.0751,
      "step": 198645
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.530251145362854,
      "learning_rate": 2.7688356588163928e-05,
      "loss": 2.1835,
      "step": 198646
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.4550954103469849,
      "learning_rate": 2.7686640181395148e-05,
      "loss": 1.8972,
      "step": 198647
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.4867185354232788,
      "learning_rate": 2.7684923825254534e-05,
      "loss": 1.9494,
      "step": 198648
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.4622535705566406,
      "learning_rate": 2.768320751974239e-05,
      "loss": 2.0698,
      "step": 198649
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.4819843769073486,
      "learning_rate": 2.7681491264859113e-05,
      "loss": 1.9169,
      "step": 198650
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.3343696594238281,
      "learning_rate": 2.767977506060497e-05,
      "loss": 1.8943,
      "step": 198651
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.297558069229126,
      "learning_rate": 2.767805890698033e-05,
      "loss": 1.9891,
      "step": 198652
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.4548909664154053,
      "learning_rate": 2.767634280398545e-05,
      "loss": 1.9126,
      "step": 198653
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.4548070430755615,
      "learning_rate": 2.767462675162071e-05,
      "loss": 1.9322,
      "step": 198654
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.436518669128418,
      "learning_rate": 2.7672910749886333e-05,
      "loss": 1.8168,
      "step": 198655
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.2993632555007935,
      "learning_rate": 2.7671194798782793e-05,
      "loss": 2.285,
      "step": 198656
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.4811491966247559,
      "learning_rate": 2.7669478898310214e-05,
      "loss": 1.9736,
      "step": 198657
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.3692444562911987,
      "learning_rate": 2.7667763048469104e-05,
      "loss": 1.9682,
      "step": 198658
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.331662893295288,
      "learning_rate": 2.7666047249259693e-05,
      "loss": 1.8987,
      "step": 198659
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.4071667194366455,
      "learning_rate": 2.7664331500682314e-05,
      "loss": 1.8957,
      "step": 198660
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.4469537734985352,
      "learning_rate": 2.76626158027372e-05,
      "loss": 2.027,
      "step": 198661
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.4410951137542725,
      "learning_rate": 2.7660900155424815e-05,
      "loss": 2.0435,
      "step": 198662
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.371345043182373,
      "learning_rate": 2.765918455874533e-05,
      "loss": 1.9401,
      "step": 198663
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.5085433721542358,
      "learning_rate": 2.765746901269921e-05,
      "loss": 2.0586,
      "step": 198664
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.4297603368759155,
      "learning_rate": 2.7655753517286716e-05,
      "loss": 1.9595,
      "step": 198665
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.4510772228240967,
      "learning_rate": 2.7654038072508155e-05,
      "loss": 1.896,
      "step": 198666
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.4701992273330688,
      "learning_rate": 2.7652322678363825e-05,
      "loss": 1.9054,
      "step": 198667
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.5220229625701904,
      "learning_rate": 2.765060733485409e-05,
      "loss": 2.0103,
      "step": 198668
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.3756595849990845,
      "learning_rate": 2.7648892041979187e-05,
      "loss": 2.0499,
      "step": 198669
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.3412195444107056,
      "learning_rate": 2.764717679973958e-05,
      "loss": 1.8672,
      "step": 198670
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.4470375776290894,
      "learning_rate": 2.7645461608135465e-05,
      "loss": 2.149,
      "step": 198671
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.4999595880508423,
      "learning_rate": 2.7643746467167216e-05,
      "loss": 2.0766,
      "step": 198672
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.4542975425720215,
      "learning_rate": 2.7642031376835094e-05,
      "loss": 1.7874,
      "step": 198673
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.5896977186203003,
      "learning_rate": 2.76403163371395e-05,
      "loss": 1.919,
      "step": 198674
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.4417481422424316,
      "learning_rate": 2.7638601348080667e-05,
      "loss": 1.8117,
      "step": 198675
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.335428237915039,
      "learning_rate": 2.7636886409659033e-05,
      "loss": 1.6535,
      "step": 198676
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.3996418714523315,
      "learning_rate": 2.763517152187482e-05,
      "loss": 2.135,
      "step": 198677
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.4597043991088867,
      "learning_rate": 2.7633456684728372e-05,
      "loss": 1.9855,
      "step": 198678
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.4201440811157227,
      "learning_rate": 2.763174189821995e-05,
      "loss": 1.933,
      "step": 198679
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.6577759981155396,
      "learning_rate": 2.7630027162349953e-05,
      "loss": 2.036,
      "step": 198680
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.3802680969238281,
      "learning_rate": 2.7628312477118654e-05,
      "loss": 2.05,
      "step": 198681
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.6494072675704956,
      "learning_rate": 2.7626597842526443e-05,
      "loss": 1.8801,
      "step": 198682
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.4614967107772827,
      "learning_rate": 2.7624883258573593e-05,
      "loss": 1.871,
      "step": 198683
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.3505221605300903,
      "learning_rate": 2.762316872526037e-05,
      "loss": 1.9245,
      "step": 198684
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.3818069696426392,
      "learning_rate": 2.7621454242587204e-05,
      "loss": 2.0439,
      "step": 198685
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.5068917274475098,
      "learning_rate": 2.761973981055433e-05,
      "loss": 2.0147,
      "step": 198686
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.4751533269882202,
      "learning_rate": 2.7618025429162017e-05,
      "loss": 1.8635,
      "step": 198687
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.5425724983215332,
      "learning_rate": 2.7616311098410726e-05,
      "loss": 2.0882,
      "step": 198688
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.5448986291885376,
      "learning_rate": 2.7614596818300726e-05,
      "loss": 2.0767,
      "step": 198689
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.5480331182479858,
      "learning_rate": 2.761288258883225e-05,
      "loss": 2.0647,
      "step": 198690
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.3788344860076904,
      "learning_rate": 2.7611168410005736e-05,
      "loss": 1.827,
      "step": 198691
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.4088120460510254,
      "learning_rate": 2.760945428182141e-05,
      "loss": 2.1509,
      "step": 198692
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.4149070978164673,
      "learning_rate": 2.7607740204279606e-05,
      "loss": 1.8968,
      "step": 198693
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.54098379611969,
      "learning_rate": 2.7606026177380724e-05,
      "loss": 2.0084,
      "step": 198694
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.4009333848953247,
      "learning_rate": 2.7604312201125035e-05,
      "loss": 2.1099,
      "step": 198695
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.4584463834762573,
      "learning_rate": 2.7602598275512767e-05,
      "loss": 1.9773,
      "step": 198696
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.612745761871338,
      "learning_rate": 2.7600884400544356e-05,
      "loss": 2.0785,
      "step": 198697
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.408036231994629,
      "learning_rate": 2.7599170576220064e-05,
      "loss": 1.7657,
      "step": 198698
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.5517877340316772,
      "learning_rate": 2.7597456802540262e-05,
      "loss": 1.9554,
      "step": 198699
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.4091261625289917,
      "learning_rate": 2.7595743079505252e-05,
      "loss": 1.9916,
      "step": 198700
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.46780526638031,
      "learning_rate": 2.7594029407115327e-05,
      "loss": 2.0045,
      "step": 198701
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.4657039642333984,
      "learning_rate": 2.7592315785370756e-05,
      "loss": 1.9842,
      "step": 198702
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.5408706665039062,
      "learning_rate": 2.759060221427197e-05,
      "loss": 1.9035,
      "step": 198703
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.4585965871810913,
      "learning_rate": 2.7588888693819178e-05,
      "loss": 1.8972,
      "step": 198704
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.4094421863555908,
      "learning_rate": 2.7587175224012802e-05,
      "loss": 1.8527,
      "step": 198705
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.4135818481445312,
      "learning_rate": 2.758546180485308e-05,
      "loss": 1.9981,
      "step": 198706
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.3402599096298218,
      "learning_rate": 2.7583748436340447e-05,
      "loss": 1.8985,
      "step": 198707
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.4295316934585571,
      "learning_rate": 2.758203511847503e-05,
      "loss": 1.9003,
      "step": 198708
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.5735557079315186,
      "learning_rate": 2.7580321851257336e-05,
      "loss": 2.1449,
      "step": 198709
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.505350947380066,
      "learning_rate": 2.7578608634687526e-05,
      "loss": 1.9553,
      "step": 198710
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.3605295419692993,
      "learning_rate": 2.7576895468766068e-05,
      "loss": 1.8703,
      "step": 198711
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.3700097799301147,
      "learning_rate": 2.757518235349313e-05,
      "loss": 2.07,
      "step": 198712
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.4498876333236694,
      "learning_rate": 2.7573469288869177e-05,
      "loss": 1.9066,
      "step": 198713
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.4946503639221191,
      "learning_rate": 2.7571756274894442e-05,
      "loss": 2.2042,
      "step": 198714
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.3753383159637451,
      "learning_rate": 2.757004331156929e-05,
      "loss": 1.9282,
      "step": 198715
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.4706826210021973,
      "learning_rate": 2.7568330398893924e-05,
      "loss": 1.7539,
      "step": 198716
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.5532989501953125,
      "learning_rate": 2.7566617536868806e-05,
      "loss": 2.0266,
      "step": 198717
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.4916313886642456,
      "learning_rate": 2.756490472549414e-05,
      "loss": 1.9837,
      "step": 198718
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.370660662651062,
      "learning_rate": 2.7563191964770394e-05,
      "loss": 1.8519,
      "step": 198719
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.4316329956054688,
      "learning_rate": 2.7561479254697762e-05,
      "loss": 2.006,
      "step": 198720
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.6923078298568726,
      "learning_rate": 2.7559766595276578e-05,
      "loss": 2.001,
      "step": 198721
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.3674383163452148,
      "learning_rate": 2.7558053986507145e-05,
      "loss": 1.8647,
      "step": 198722
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.372124195098877,
      "learning_rate": 2.7556341428389862e-05,
      "loss": 2.0548,
      "step": 198723
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.4880344867706299,
      "learning_rate": 2.755462892092496e-05,
      "loss": 2.0396,
      "step": 198724
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.4072234630584717,
      "learning_rate": 2.755291646411284e-05,
      "loss": 2.0695,
      "step": 198725
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.6801505088806152,
      "learning_rate": 2.755120405795377e-05,
      "loss": 2.0196,
      "step": 198726
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.3458753824234009,
      "learning_rate": 2.7549491702448045e-05,
      "loss": 1.9578,
      "step": 198727
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.5701922178268433,
      "learning_rate": 2.7547779397596005e-05,
      "loss": 2.1705,
      "step": 198728
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.7306320667266846,
      "learning_rate": 2.754606714339801e-05,
      "loss": 1.8463,
      "step": 198729
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.4980428218841553,
      "learning_rate": 2.7544354939854295e-05,
      "loss": 1.9432,
      "step": 198730
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.3989124298095703,
      "learning_rate": 2.7542642786965265e-05,
      "loss": 2.1193,
      "step": 198731
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.5107309818267822,
      "learning_rate": 2.754093068473121e-05,
      "loss": 2.1647,
      "step": 198732
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.4384468793869019,
      "learning_rate": 2.7539218633152437e-05,
      "loss": 1.8431,
      "step": 198733
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.3857821226119995,
      "learning_rate": 2.7537506632229246e-05,
      "loss": 2.0473,
      "step": 198734
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.3775441646575928,
      "learning_rate": 2.7535794681962e-05,
      "loss": 1.9824,
      "step": 198735
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.3782902956008911,
      "learning_rate": 2.753408278235093e-05,
      "loss": 1.9156,
      "step": 198736
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.3306902647018433,
      "learning_rate": 2.7532370933396476e-05,
      "loss": 2.0207,
      "step": 198737
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.4678945541381836,
      "learning_rate": 2.7530659135098897e-05,
      "loss": 2.1277,
      "step": 198738
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.3513855934143066,
      "learning_rate": 2.7528947387458534e-05,
      "loss": 2.1159,
      "step": 198739
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.3969115018844604,
      "learning_rate": 2.7527235690475613e-05,
      "loss": 2.1209,
      "step": 198740
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.2956064939498901,
      "learning_rate": 2.752552404415057e-05,
      "loss": 1.8472,
      "step": 198741
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.322113275527954,
      "learning_rate": 2.752381244848364e-05,
      "loss": 1.9191,
      "step": 198742
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.30927312374115,
      "learning_rate": 2.752210090347522e-05,
      "loss": 1.8267,
      "step": 198743
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.3279609680175781,
      "learning_rate": 2.7520389409125575e-05,
      "loss": 2.139,
      "step": 198744
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.4653587341308594,
      "learning_rate": 2.7518677965435043e-05,
      "loss": 1.8498,
      "step": 198745
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.5271259546279907,
      "learning_rate": 2.7516966572403854e-05,
      "loss": 2.024,
      "step": 198746
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.5271878242492676,
      "learning_rate": 2.7515255230032506e-05,
      "loss": 1.9374,
      "step": 198747
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.428239107131958,
      "learning_rate": 2.7513543938321135e-05,
      "loss": 1.9477,
      "step": 198748
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.3413605690002441,
      "learning_rate": 2.7511832697270174e-05,
      "loss": 2.0028,
      "step": 198749
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.335440754890442,
      "learning_rate": 2.7510121506879956e-05,
      "loss": 2.0895,
      "step": 198750
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.4464339017868042,
      "learning_rate": 2.7508410367150713e-05,
      "loss": 1.9159,
      "step": 198751
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.5526354312896729,
      "learning_rate": 2.7506699278082746e-05,
      "loss": 2.1919,
      "step": 198752
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.3941320180892944,
      "learning_rate": 2.7504988239676486e-05,
      "loss": 1.9481,
      "step": 198753
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.4289555549621582,
      "learning_rate": 2.7503277251932164e-05,
      "loss": 1.7931,
      "step": 198754
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.4953579902648926,
      "learning_rate": 2.7501566314850154e-05,
      "loss": 2.0704,
      "step": 198755
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.4447633028030396,
      "learning_rate": 2.749985542843075e-05,
      "loss": 2.1184,
      "step": 198756
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.5138667821884155,
      "learning_rate": 2.7498144592674253e-05,
      "loss": 2.0078,
      "step": 198757
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.6189802885055542,
      "learning_rate": 2.7496433807580965e-05,
      "loss": 1.9237,
      "step": 198758
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.2980389595031738,
      "learning_rate": 2.7494723073151282e-05,
      "loss": 2.0063,
      "step": 198759
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.4181462526321411,
      "learning_rate": 2.7493012389385405e-05,
      "loss": 2.2071,
      "step": 198760
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.5238820314407349,
      "learning_rate": 2.74913017562838e-05,
      "loss": 1.9985,
      "step": 198761
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.599100112915039,
      "learning_rate": 2.7489591173846703e-05,
      "loss": 2.1338,
      "step": 198762
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.483076572418213,
      "learning_rate": 2.748788064207441e-05,
      "loss": 1.9168,
      "step": 198763
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.5909576416015625,
      "learning_rate": 2.7486170160967223e-05,
      "loss": 2.2463,
      "step": 198764
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.4899052381515503,
      "learning_rate": 2.7484459730525577e-05,
      "loss": 2.107,
      "step": 198765
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.369280457496643,
      "learning_rate": 2.7482749350749633e-05,
      "loss": 1.9104,
      "step": 198766
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.4241701364517212,
      "learning_rate": 2.7481039021639862e-05,
      "loss": 2.2537,
      "step": 198767
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.440221905708313,
      "learning_rate": 2.7479328743196492e-05,
      "loss": 2.1118,
      "step": 198768
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.4762107133865356,
      "learning_rate": 2.7477618515419798e-05,
      "loss": 1.9745,
      "step": 198769
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.3703681230545044,
      "learning_rate": 2.747590833831024e-05,
      "loss": 2.1909,
      "step": 198770
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.3718301057815552,
      "learning_rate": 2.747419821186805e-05,
      "loss": 1.9217,
      "step": 198771
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.4318878650665283,
      "learning_rate": 2.7472488136093497e-05,
      "loss": 2.0364,
      "step": 198772
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.5718721151351929,
      "learning_rate": 2.747077811098698e-05,
      "loss": 2.0024,
      "step": 198773
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.5309895277023315,
      "learning_rate": 2.74690681365488e-05,
      "loss": 2.0259,
      "step": 198774
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.2852922677993774,
      "learning_rate": 2.7467358212779222e-05,
      "loss": 2.2327,
      "step": 198775
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.4275619983673096,
      "learning_rate": 2.7465648339678682e-05,
      "loss": 2.068,
      "step": 198776
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.4198092222213745,
      "learning_rate": 2.7463938517247375e-05,
      "loss": 1.8951,
      "step": 198777
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.4829978942871094,
      "learning_rate": 2.7462228745485636e-05,
      "loss": 1.8706,
      "step": 198778
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.4268527030944824,
      "learning_rate": 2.746051902439387e-05,
      "loss": 1.7553,
      "step": 198779
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.3519089221954346,
      "learning_rate": 2.74588093539723e-05,
      "loss": 1.8692,
      "step": 198780
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.3015681505203247,
      "learning_rate": 2.74570997342213e-05,
      "loss": 1.759,
      "step": 198781
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.3054754734039307,
      "learning_rate": 2.74553901651412e-05,
      "loss": 1.9976,
      "step": 198782
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.390291452407837,
      "learning_rate": 2.74536806467322e-05,
      "loss": 2.0874,
      "step": 198783
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.5867618322372437,
      "learning_rate": 2.7451971178994804e-05,
      "loss": 1.9499,
      "step": 198784
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.4168404340744019,
      "learning_rate": 2.7450261761929203e-05,
      "loss": 2.1802,
      "step": 198785
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.4849098920822144,
      "learning_rate": 2.7448552395535705e-05,
      "loss": 2.0706,
      "step": 198786
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.4785728454589844,
      "learning_rate": 2.7446843079814706e-05,
      "loss": 2.1784,
      "step": 198787
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.2877826690673828,
      "learning_rate": 2.7445133814766473e-05,
      "loss": 1.9775,
      "step": 198788
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.4761712551116943,
      "learning_rate": 2.7443424600391306e-05,
      "loss": 2.0642,
      "step": 198789
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.563238263130188,
      "learning_rate": 2.744171543668957e-05,
      "loss": 2.2357,
      "step": 198790
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.5708800554275513,
      "learning_rate": 2.7440006323661533e-05,
      "loss": 2.0318,
      "step": 198791
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.571124792098999,
      "learning_rate": 2.7438297261307628e-05,
      "loss": 1.932,
      "step": 198792
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.4750516414642334,
      "learning_rate": 2.7436588249628055e-05,
      "loss": 2.1488,
      "step": 198793
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.464526891708374,
      "learning_rate": 2.7434879288623146e-05,
      "loss": 2.0034,
      "step": 198794
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.4053288698196411,
      "learning_rate": 2.7433170378293234e-05,
      "loss": 1.9733,
      "step": 198795
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.5928875207901,
      "learning_rate": 2.7431461518638653e-05,
      "loss": 2.0467,
      "step": 198796
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.4312517642974854,
      "learning_rate": 2.7429752709659668e-05,
      "loss": 1.9095,
      "step": 198797
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.6013517379760742,
      "learning_rate": 2.7428043951356683e-05,
      "loss": 2.099,
      "step": 198798
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.398902416229248,
      "learning_rate": 2.742633524372999e-05,
      "loss": 1.9469,
      "step": 198799
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.2791975736618042,
      "learning_rate": 2.7424626586779865e-05,
      "loss": 2.0701,
      "step": 198800
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.4084327220916748,
      "learning_rate": 2.7422917980506598e-05,
      "loss": 2.1041,
      "step": 198801
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.4076670408248901,
      "learning_rate": 2.7421209424910628e-05,
      "loss": 1.7921,
      "step": 198802
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.4302051067352295,
      "learning_rate": 2.741950091999212e-05,
      "loss": 2.0896,
      "step": 198803
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.477409839630127,
      "learning_rate": 2.7417792465751544e-05,
      "loss": 1.8723,
      "step": 198804
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.4007914066314697,
      "learning_rate": 2.741608406218916e-05,
      "loss": 1.7658,
      "step": 198805
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.4768669605255127,
      "learning_rate": 2.7414375709305235e-05,
      "loss": 1.8911,
      "step": 198806
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.3771111965179443,
      "learning_rate": 2.7412667407100107e-05,
      "loss": 1.9762,
      "step": 198807
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.3826048374176025,
      "learning_rate": 2.7410959155574142e-05,
      "loss": 1.9773,
      "step": 198808
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.466129183769226,
      "learning_rate": 2.7409250954727567e-05,
      "loss": 2.0188,
      "step": 198809
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.3494911193847656,
      "learning_rate": 2.740754280456082e-05,
      "loss": 2.0824,
      "step": 198810
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.325955867767334,
      "learning_rate": 2.740583470507417e-05,
      "loss": 1.9009,
      "step": 198811
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.470727562904358,
      "learning_rate": 2.740412665626788e-05,
      "loss": 2.0016,
      "step": 198812
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.3434934616088867,
      "learning_rate": 2.740241865814228e-05,
      "loss": 2.2029,
      "step": 198813
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.5614607334136963,
      "learning_rate": 2.740071071069777e-05,
      "loss": 2.1059,
      "step": 198814
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.6280508041381836,
      "learning_rate": 2.7399002813934557e-05,
      "loss": 2.088,
      "step": 198815
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.584943175315857,
      "learning_rate": 2.7397294967853066e-05,
      "loss": 1.7843,
      "step": 198816
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.3544071912765503,
      "learning_rate": 2.739558717245357e-05,
      "loss": 1.8899,
      "step": 198817
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.4303241968154907,
      "learning_rate": 2.73938794277364e-05,
      "loss": 2.066,
      "step": 198818
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.403175950050354,
      "learning_rate": 2.7392171733701785e-05,
      "loss": 1.7062,
      "step": 198819
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.6426302194595337,
      "learning_rate": 2.739046409035016e-05,
      "loss": 2.1758,
      "step": 198820
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.5021815299987793,
      "learning_rate": 2.738875649768173e-05,
      "loss": 1.7989,
      "step": 198821
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.5117379426956177,
      "learning_rate": 2.7387048955696922e-05,
      "loss": 1.9933,
      "step": 198822
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.59230375289917,
      "learning_rate": 2.7385341464396038e-05,
      "loss": 2.1885,
      "step": 198823
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.5624842643737793,
      "learning_rate": 2.7383634023779345e-05,
      "loss": 1.907,
      "step": 198824
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.3730053901672363,
      "learning_rate": 2.738192663384714e-05,
      "loss": 2.0275,
      "step": 198825
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.505080223083496,
      "learning_rate": 2.7380219294599825e-05,
      "loss": 1.8477,
      "step": 198826
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.6166516542434692,
      "learning_rate": 2.7378512006037636e-05,
      "loss": 2.0871,
      "step": 198827
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.4804307222366333,
      "learning_rate": 2.7376804768160966e-05,
      "loss": 1.9151,
      "step": 198828
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.3672850131988525,
      "learning_rate": 2.7375097580970083e-05,
      "loss": 1.8761,
      "step": 198829
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.609714388847351,
      "learning_rate": 2.7373390444465327e-05,
      "loss": 2.0566,
      "step": 198830
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.3450367450714111,
      "learning_rate": 2.7371683358646958e-05,
      "loss": 2.0557,
      "step": 198831
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.3008291721343994,
      "learning_rate": 2.7369976323515376e-05,
      "loss": 1.796,
      "step": 198832
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.3744713068008423,
      "learning_rate": 2.7368269339070815e-05,
      "loss": 2.055,
      "step": 198833
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.3882052898406982,
      "learning_rate": 2.7366562405313708e-05,
      "loss": 1.7445,
      "step": 198834
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.66732656955719,
      "learning_rate": 2.736485552224429e-05,
      "loss": 1.9953,
      "step": 198835
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.5380462408065796,
      "learning_rate": 2.736314868986289e-05,
      "loss": 2.0681,
      "step": 198836
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.5493563413619995,
      "learning_rate": 2.736144190816978e-05,
      "loss": 2.0251,
      "step": 198837
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.3900032043457031,
      "learning_rate": 2.7359735177165388e-05,
      "loss": 1.9359,
      "step": 198838
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.5344693660736084,
      "learning_rate": 2.735802849684988e-05,
      "loss": 1.8688,
      "step": 198839
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.3690069913864136,
      "learning_rate": 2.735632186722376e-05,
      "loss": 1.909,
      "step": 198840
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.382770299911499,
      "learning_rate": 2.7354615288287198e-05,
      "loss": 1.9904,
      "step": 198841
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.4014610052108765,
      "learning_rate": 2.7352908760040582e-05,
      "loss": 2.0165,
      "step": 198842
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.5681637525558472,
      "learning_rate": 2.7351202282484152e-05,
      "loss": 2.3078,
      "step": 198843
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.5265125036239624,
      "learning_rate": 2.734949585561834e-05,
      "loss": 2.1647,
      "step": 198844
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.4268149137496948,
      "learning_rate": 2.7347789479443348e-05,
      "loss": 2.0059,
      "step": 198845
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.4772789478302002,
      "learning_rate": 2.7346083153959608e-05,
      "loss": 2.0312,
      "step": 198846
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.4983665943145752,
      "learning_rate": 2.7344376879167317e-05,
      "loss": 1.9776,
      "step": 198847
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.3695931434631348,
      "learning_rate": 2.734267065506698e-05,
      "loss": 1.832,
      "step": 198848
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.3933111429214478,
      "learning_rate": 2.7340964481658654e-05,
      "loss": 2.0136,
      "step": 198849
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.5892093181610107,
      "learning_rate": 2.7339258358942852e-05,
      "loss": 2.1798,
      "step": 198850
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.386654019355774,
      "learning_rate": 2.7337552286919762e-05,
      "loss": 2.0827,
      "step": 198851
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.3709384202957153,
      "learning_rate": 2.7335846265589823e-05,
      "loss": 2.0432,
      "step": 198852
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.4568763971328735,
      "learning_rate": 2.7334140294953267e-05,
      "loss": 1.8785,
      "step": 198853
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.463724970817566,
      "learning_rate": 2.7332434375010494e-05,
      "loss": 2.0333,
      "step": 198854
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.539980411529541,
      "learning_rate": 2.7330728505761735e-05,
      "loss": 1.9288,
      "step": 198855
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.4922289848327637,
      "learning_rate": 2.7329022687207356e-05,
      "loss": 2.0545,
      "step": 198856
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.6567540168762207,
      "learning_rate": 2.7327316919347632e-05,
      "loss": 2.0615,
      "step": 198857
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.5142847299575806,
      "learning_rate": 2.7325611202182917e-05,
      "loss": 2.0459,
      "step": 198858
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.3993664979934692,
      "learning_rate": 2.7323905535713487e-05,
      "loss": 1.8605,
      "step": 198859
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.4036163091659546,
      "learning_rate": 2.7322199919939734e-05,
      "loss": 1.6899,
      "step": 198860
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.3366353511810303,
      "learning_rate": 2.7320494354861967e-05,
      "loss": 2.1131,
      "step": 198861
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.5106886625289917,
      "learning_rate": 2.7318788840480443e-05,
      "loss": 1.8735,
      "step": 198862
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.503891944885254,
      "learning_rate": 2.7317083376795434e-05,
      "loss": 1.9154,
      "step": 198863
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.4367891550064087,
      "learning_rate": 2.7315377963807406e-05,
      "loss": 2.0363,
      "step": 198864
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.4142245054244995,
      "learning_rate": 2.7313672601516523e-05,
      "loss": 1.9909,
      "step": 198865
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.4222488403320312,
      "learning_rate": 2.731196728992322e-05,
      "loss": 2.2275,
      "step": 198866
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.4494611024856567,
      "learning_rate": 2.7310262029027797e-05,
      "loss": 2.0432,
      "step": 198867
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.4401637315750122,
      "learning_rate": 2.7308556818830485e-05,
      "loss": 1.8001,
      "step": 198868
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.399196982383728,
      "learning_rate": 2.7306851659331687e-05,
      "loss": 1.871,
      "step": 198869
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.403748631477356,
      "learning_rate": 2.730514655053173e-05,
      "loss": 1.9273,
      "step": 198870
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.3504619598388672,
      "learning_rate": 2.7303441492430822e-05,
      "loss": 1.9944,
      "step": 198871
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.3873714208602905,
      "learning_rate": 2.730173648502939e-05,
      "loss": 2.065,
      "step": 198872
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.3690048456192017,
      "learning_rate": 2.730003152832774e-05,
      "loss": 2.0388,
      "step": 198873
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.4075379371643066,
      "learning_rate": 2.7298326622326094e-05,
      "loss": 1.8777,
      "step": 198874
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.3874820470809937,
      "learning_rate": 2.7296621767024895e-05,
      "loss": 1.9315,
      "step": 198875
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.6744242906570435,
      "learning_rate": 2.7294916962424407e-05,
      "loss": 2.0074,
      "step": 198876
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.5658128261566162,
      "learning_rate": 2.7293212208524862e-05,
      "loss": 2.0639,
      "step": 198877
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.3610725402832031,
      "learning_rate": 2.7291507505326758e-05,
      "loss": 2.0536,
      "step": 198878
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.4412102699279785,
      "learning_rate": 2.7289802852830267e-05,
      "loss": 1.9368,
      "step": 198879
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.3806980848312378,
      "learning_rate": 2.728809825103572e-05,
      "loss": 1.8882,
      "step": 198880
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.5673373937606812,
      "learning_rate": 2.728639369994351e-05,
      "loss": 2.174,
      "step": 198881
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.2836908102035522,
      "learning_rate": 2.7284689199553844e-05,
      "loss": 1.8233,
      "step": 198882
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.3943942785263062,
      "learning_rate": 2.7282984749867154e-05,
      "loss": 1.9649,
      "step": 198883
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.5231612920761108,
      "learning_rate": 2.728128035088374e-05,
      "loss": 2.1112,
      "step": 198884
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.3332033157348633,
      "learning_rate": 2.727957600260383e-05,
      "loss": 1.8776,
      "step": 198885
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.467840313911438,
      "learning_rate": 2.7277871705027764e-05,
      "loss": 1.9277,
      "step": 198886
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.4391344785690308,
      "learning_rate": 2.7276167458155972e-05,
      "loss": 1.9167,
      "step": 198887
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.3890005350112915,
      "learning_rate": 2.7274463261988588e-05,
      "loss": 2.0198,
      "step": 198888
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.4565702676773071,
      "learning_rate": 2.727275911652611e-05,
      "loss": 1.9584,
      "step": 198889
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.352958083152771,
      "learning_rate": 2.727105502176877e-05,
      "loss": 1.9587,
      "step": 198890
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.3408076763153076,
      "learning_rate": 2.7269350977716907e-05,
      "loss": 1.8462,
      "step": 198891
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.6670830249786377,
      "learning_rate": 2.726764698437075e-05,
      "loss": 1.9018,
      "step": 198892
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.4409736394882202,
      "learning_rate": 2.7265943041730732e-05,
      "loss": 1.8316,
      "step": 198893
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.4531302452087402,
      "learning_rate": 2.7264239149797084e-05,
      "loss": 1.8893,
      "step": 198894
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.5833942890167236,
      "learning_rate": 2.726253530857021e-05,
      "loss": 1.765,
      "step": 198895
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.3602802753448486,
      "learning_rate": 2.7260831518050375e-05,
      "loss": 1.7977,
      "step": 198896
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.504350185394287,
      "learning_rate": 2.725912777823791e-05,
      "loss": 2.1335,
      "step": 198897
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.7120031118392944,
      "learning_rate": 2.725742408913305e-05,
      "loss": 2.1261,
      "step": 198898
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.4002652168273926,
      "learning_rate": 2.7255720450736262e-05,
      "loss": 1.9988,
      "step": 198899
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.443723440170288,
      "learning_rate": 2.7254016863047712e-05,
      "loss": 1.8088,
      "step": 198900
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.4251078367233276,
      "learning_rate": 2.7252313326067832e-05,
      "loss": 2.1313,
      "step": 198901
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.3695056438446045,
      "learning_rate": 2.7250609839796922e-05,
      "loss": 2.062,
      "step": 198902
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.4940651655197144,
      "learning_rate": 2.7248906404235248e-05,
      "loss": 1.9711,
      "step": 198903
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.5092726945877075,
      "learning_rate": 2.7247203019383113e-05,
      "loss": 2.2098,
      "step": 198904
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.4039891958236694,
      "learning_rate": 2.7245499685240912e-05,
      "loss": 1.8785,
      "step": 198905
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.3778572082519531,
      "learning_rate": 2.7243796401808883e-05,
      "loss": 1.7469,
      "step": 198906
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.4779824018478394,
      "learning_rate": 2.724209316908742e-05,
      "loss": 1.9875,
      "step": 198907
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.383671760559082,
      "learning_rate": 2.724038998707683e-05,
      "loss": 2.0375,
      "step": 198908
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.5063092708587646,
      "learning_rate": 2.723868685577737e-05,
      "loss": 1.852,
      "step": 198909
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.3700686693191528,
      "learning_rate": 2.7236983775189348e-05,
      "loss": 2.1356,
      "step": 198910
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.5087183713912964,
      "learning_rate": 2.723528074531316e-05,
      "loss": 2.0421,
      "step": 198911
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.4342024326324463,
      "learning_rate": 2.723357776614904e-05,
      "loss": 2.0665,
      "step": 198912
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.4196422100067139,
      "learning_rate": 2.723187483769739e-05,
      "loss": 2.0963,
      "step": 198913
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.7238291501998901,
      "learning_rate": 2.7230171959958434e-05,
      "loss": 2.1069,
      "step": 198914
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.308069109916687,
      "learning_rate": 2.722846913293265e-05,
      "loss": 1.7326,
      "step": 198915
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.5673846006393433,
      "learning_rate": 2.7226766356620134e-05,
      "loss": 1.8054,
      "step": 198916
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.4717998504638672,
      "learning_rate": 2.7225063631021348e-05,
      "loss": 1.8148,
      "step": 198917
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.648366928100586,
      "learning_rate": 2.7223360956136532e-05,
      "loss": 1.9518,
      "step": 198918
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.5270118713378906,
      "learning_rate": 2.7221658331966113e-05,
      "loss": 2.0005,
      "step": 198919
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.3728830814361572,
      "learning_rate": 2.721995575851026e-05,
      "loss": 2.1188,
      "step": 198920
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.4748814105987549,
      "learning_rate": 2.721825323576947e-05,
      "loss": 2.0006,
      "step": 198921
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.5504872798919678,
      "learning_rate": 2.7216550763743882e-05,
      "loss": 2.1011,
      "step": 198922
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.3954644203186035,
      "learning_rate": 2.7214848342433893e-05,
      "loss": 1.9899,
      "step": 198923
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.468868374824524,
      "learning_rate": 2.72131459718398e-05,
      "loss": 1.7993,
      "step": 198924
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.436887502670288,
      "learning_rate": 2.7211443651961974e-05,
      "loss": 2.0992,
      "step": 198925
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.4953577518463135,
      "learning_rate": 2.720974138280061e-05,
      "loss": 1.9051,
      "step": 198926
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.4240387678146362,
      "learning_rate": 2.7208039164356245e-05,
      "loss": 1.8761,
      "step": 198927
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.346285343170166,
      "learning_rate": 2.7206336996628943e-05,
      "loss": 2.0885,
      "step": 198928
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.3861463069915771,
      "learning_rate": 2.720463487961917e-05,
      "loss": 2.0615,
      "step": 198929
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.3998730182647705,
      "learning_rate": 2.720293281332716e-05,
      "loss": 2.0072,
      "step": 198930
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.4192811250686646,
      "learning_rate": 2.7201230797753348e-05,
      "loss": 1.889,
      "step": 198931
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.4281944036483765,
      "learning_rate": 2.7199528832897898e-05,
      "loss": 1.9937,
      "step": 198932
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.4090988636016846,
      "learning_rate": 2.719782691876131e-05,
      "loss": 1.8289,
      "step": 198933
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.561091423034668,
      "learning_rate": 2.7196125055343686e-05,
      "loss": 1.9843,
      "step": 198934
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.3075337409973145,
      "learning_rate": 2.7194423242645525e-05,
      "loss": 1.78,
      "step": 198935
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.3780872821807861,
      "learning_rate": 2.719272148066699e-05,
      "loss": 2.1865,
      "step": 198936
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.7036724090576172,
      "learning_rate": 2.719101976940855e-05,
      "loss": 1.9963,
      "step": 198937
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.3486101627349854,
      "learning_rate": 2.7189318108870405e-05,
      "loss": 2.1411,
      "step": 198938
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.4276670217514038,
      "learning_rate": 2.7187616499053023e-05,
      "loss": 1.9076,
      "step": 198939
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.586766004562378,
      "learning_rate": 2.71859149399565e-05,
      "loss": 2.0561,
      "step": 198940
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.4027321338653564,
      "learning_rate": 2.7184213431581303e-05,
      "loss": 2.0051,
      "step": 198941
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.3472503423690796,
      "learning_rate": 2.7182511973927667e-05,
      "loss": 2.0684,
      "step": 198942
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.3443158864974976,
      "learning_rate": 2.718081056699599e-05,
      "loss": 2.1424,
      "step": 198943
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.6606462001800537,
      "learning_rate": 2.717910921078651e-05,
      "loss": 1.8455,
      "step": 198944
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.3981002569198608,
      "learning_rate": 2.7177407905299653e-05,
      "loss": 2.1904,
      "step": 198945
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.407125473022461,
      "learning_rate": 2.7175706650535657e-05,
      "loss": 1.8326,
      "step": 198946
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.290484070777893,
      "learning_rate": 2.7174005446494853e-05,
      "loss": 1.9528,
      "step": 198947
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.3425763845443726,
      "learning_rate": 2.717230429317747e-05,
      "loss": 2.034,
      "step": 198948
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.4933925867080688,
      "learning_rate": 2.7170603190583984e-05,
      "loss": 2.1295,
      "step": 198949
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.3123670816421509,
      "learning_rate": 2.7168902138714556e-05,
      "loss": 1.9134,
      "step": 198950
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.339980959892273,
      "learning_rate": 2.716720113756965e-05,
      "loss": 1.8936,
      "step": 198951
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.4272077083587646,
      "learning_rate": 2.7165500187149503e-05,
      "loss": 1.9751,
      "step": 198952
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.452471137046814,
      "learning_rate": 2.7163799287454413e-05,
      "loss": 2.0229,
      "step": 198953
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.449567437171936,
      "learning_rate": 2.716209843848475e-05,
      "loss": 2.1464,
      "step": 198954
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.3333557844161987,
      "learning_rate": 2.7160397640240804e-05,
      "loss": 1.8646,
      "step": 198955
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.3529150485992432,
      "learning_rate": 2.7158696892722854e-05,
      "loss": 2.2271,
      "step": 198956
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.4647822380065918,
      "learning_rate": 2.715699619593129e-05,
      "loss": 2.0007,
      "step": 198957
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.3347688913345337,
      "learning_rate": 2.7155295549866387e-05,
      "loss": 2.0201,
      "step": 198958
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.5994374752044678,
      "learning_rate": 2.7153594954528436e-05,
      "loss": 2.2271,
      "step": 198959
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.4325883388519287,
      "learning_rate": 2.7151894409917807e-05,
      "loss": 1.9738,
      "step": 198960
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.340224027633667,
      "learning_rate": 2.7150193916034803e-05,
      "loss": 1.857,
      "step": 198961
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.4793953895568848,
      "learning_rate": 2.7148493472879684e-05,
      "loss": 2.1693,
      "step": 198962
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.5272341966629028,
      "learning_rate": 2.7146793080452856e-05,
      "loss": 2.1228,
      "step": 198963
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.6181001663208008,
      "learning_rate": 2.714509273875458e-05,
      "loss": 1.9888,
      "step": 198964
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.415380597114563,
      "learning_rate": 2.714339244778513e-05,
      "loss": 2.0097,
      "step": 198965
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.5072404146194458,
      "learning_rate": 2.714169220754493e-05,
      "loss": 1.7331,
      "step": 198966
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.5215027332305908,
      "learning_rate": 2.7139992018034184e-05,
      "loss": 2.0765,
      "step": 198967
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.572853446006775,
      "learning_rate": 2.7138291879253328e-05,
      "loss": 1.9962,
      "step": 198968
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.5500235557556152,
      "learning_rate": 2.7136591791202623e-05,
      "loss": 2.2054,
      "step": 198969
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.730840802192688,
      "learning_rate": 2.713489175388237e-05,
      "loss": 1.9659,
      "step": 198970
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.5795823335647583,
      "learning_rate": 2.7133191767292807e-05,
      "loss": 1.9543,
      "step": 198971
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.4711087942123413,
      "learning_rate": 2.7131491831434427e-05,
      "loss": 1.9005,
      "step": 198972
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.337304711341858,
      "learning_rate": 2.71297919463074e-05,
      "loss": 1.9225,
      "step": 198973
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.4199719429016113,
      "learning_rate": 2.7128092111912124e-05,
      "loss": 1.9244,
      "step": 198974
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.4378681182861328,
      "learning_rate": 2.7126392328248902e-05,
      "loss": 1.8685,
      "step": 198975
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.4318889379501343,
      "learning_rate": 2.712469259531803e-05,
      "loss": 1.8705,
      "step": 198976
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.3393079042434692,
      "learning_rate": 2.7122992913119778e-05,
      "loss": 2.0944,
      "step": 198977
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.3334420919418335,
      "learning_rate": 2.7121293281654543e-05,
      "loss": 1.9375,
      "step": 198978
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.4923992156982422,
      "learning_rate": 2.7119593700922594e-05,
      "loss": 1.9612,
      "step": 198979
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.4631174802780151,
      "learning_rate": 2.7117894170924293e-05,
      "loss": 1.945,
      "step": 198980
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.5056358575820923,
      "learning_rate": 2.7116194691659878e-05,
      "loss": 2.0623,
      "step": 198981
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.4986615180969238,
      "learning_rate": 2.7114495263129808e-05,
      "loss": 2.0095,
      "step": 198982
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.5132993459701538,
      "learning_rate": 2.7112795885334227e-05,
      "loss": 1.881,
      "step": 198983
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.3213783502578735,
      "learning_rate": 2.711109655827356e-05,
      "loss": 2.1396,
      "step": 198984
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.414620041847229,
      "learning_rate": 2.7109397281948042e-05,
      "loss": 1.9827,
      "step": 198985
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.3714207410812378,
      "learning_rate": 2.7107698056358105e-05,
      "loss": 2.1741,
      "step": 198986
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.4897130727767944,
      "learning_rate": 2.710599888150392e-05,
      "loss": 2.0956,
      "step": 198987
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.3773177862167358,
      "learning_rate": 2.7104299757386017e-05,
      "loss": 1.8433,
      "step": 198988
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.508763313293457,
      "learning_rate": 2.7102600684004462e-05,
      "loss": 2.0457,
      "step": 198989
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.4750083684921265,
      "learning_rate": 2.710090166135972e-05,
      "loss": 1.869,
      "step": 198990
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.3803837299346924,
      "learning_rate": 2.7099202689452027e-05,
      "loss": 1.9188,
      "step": 198991
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.378282904624939,
      "learning_rate": 2.7097503768281815e-05,
      "loss": 1.9987,
      "step": 198992
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.5328989028930664,
      "learning_rate": 2.709580489784925e-05,
      "loss": 2.0499,
      "step": 198993
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.5413665771484375,
      "learning_rate": 2.709410607815483e-05,
      "loss": 1.8464,
      "step": 198994
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.4596136808395386,
      "learning_rate": 2.709240730919866e-05,
      "loss": 1.9018,
      "step": 198995
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.4914859533309937,
      "learning_rate": 2.7090708590981237e-05,
      "loss": 2.2216,
      "step": 198996
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.3426084518432617,
      "learning_rate": 2.708900992350276e-05,
      "loss": 1.9849,
      "step": 198997
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.4384480714797974,
      "learning_rate": 2.7087311306763625e-05,
      "loss": 1.9569,
      "step": 198998
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.4817852973937988,
      "learning_rate": 2.708561274076404e-05,
      "loss": 2.1464,
      "step": 198999
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.3634296655654907,
      "learning_rate": 2.70839142255045e-05,
      "loss": 1.9236,
      "step": 199000
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.3952622413635254,
      "learning_rate": 2.7082215760985138e-05,
      "loss": 1.9842,
      "step": 199001
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.4441282749176025,
      "learning_rate": 2.7080517347206353e-05,
      "loss": 2.1744,
      "step": 199002
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.4355511665344238,
      "learning_rate": 2.707881898416845e-05,
      "loss": 1.9434,
      "step": 199003
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.5298793315887451,
      "learning_rate": 2.7077120671871755e-05,
      "loss": 1.9227,
      "step": 199004
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.6387990713119507,
      "learning_rate": 2.7075422410316538e-05,
      "loss": 1.82,
      "step": 199005
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.3394689559936523,
      "learning_rate": 2.7073724199503266e-05,
      "loss": 2.0536,
      "step": 199006
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.5045853853225708,
      "learning_rate": 2.707202603943204e-05,
      "loss": 1.9083,
      "step": 199007
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.5029243230819702,
      "learning_rate": 2.7070327930103286e-05,
      "loss": 1.9636,
      "step": 199008
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.414109706878662,
      "learning_rate": 2.7068629871517312e-05,
      "loss": 2.0014,
      "step": 199009
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.507348895072937,
      "learning_rate": 2.7066931863674446e-05,
      "loss": 1.9757,
      "step": 199010
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.5455697774887085,
      "learning_rate": 2.7065233906574955e-05,
      "loss": 1.9216,
      "step": 199011
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.731665849685669,
      "learning_rate": 2.7063536000219278e-05,
      "loss": 1.8727,
      "step": 199012
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.363269567489624,
      "learning_rate": 2.7061838144607572e-05,
      "loss": 2.0876,
      "step": 199013
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.5065593719482422,
      "learning_rate": 2.7060140339740244e-05,
      "loss": 2.0317,
      "step": 199014
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.418436050415039,
      "learning_rate": 2.7058442585617522e-05,
      "loss": 1.9364,
      "step": 199015
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.487500548362732,
      "learning_rate": 2.7056744882239877e-05,
      "loss": 2.1393,
      "step": 199016
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.4029455184936523,
      "learning_rate": 2.7055047229607474e-05,
      "loss": 1.8925,
      "step": 199017
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.4525010585784912,
      "learning_rate": 2.7053349627720777e-05,
      "loss": 1.8048,
      "step": 199018
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.4734517335891724,
      "learning_rate": 2.7051652076579923e-05,
      "loss": 2.0364,
      "step": 199019
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.7339855432510376,
      "learning_rate": 2.7049954576185375e-05,
      "loss": 2.101,
      "step": 199020
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.404524326324463,
      "learning_rate": 2.7048257126537333e-05,
      "loss": 1.9058,
      "step": 199021
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.5060776472091675,
      "learning_rate": 2.704655972763623e-05,
      "loss": 2.0582,
      "step": 199022
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.4476629495620728,
      "learning_rate": 2.704486237948227e-05,
      "loss": 1.9727,
      "step": 199023
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.3845587968826294,
      "learning_rate": 2.7043165082075914e-05,
      "loss": 2.1623,
      "step": 199024
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.6582340002059937,
      "learning_rate": 2.7041467835417298e-05,
      "loss": 2.0086,
      "step": 199025
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.5810626745224,
      "learning_rate": 2.7039770639506886e-05,
      "loss": 1.8918,
      "step": 199026
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.4013051986694336,
      "learning_rate": 2.7038073494344848e-05,
      "loss": 1.9869,
      "step": 199027
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.6345497369766235,
      "learning_rate": 2.7036376399931646e-05,
      "loss": 2.1426,
      "step": 199028
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.4194581508636475,
      "learning_rate": 2.703467935626752e-05,
      "loss": 2.0769,
      "step": 199029
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.3792933225631714,
      "learning_rate": 2.7032982363352795e-05,
      "loss": 1.9239,
      "step": 199030
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.3821016550064087,
      "learning_rate": 2.7031285421187843e-05,
      "loss": 2.2282,
      "step": 199031
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.5058685541152954,
      "learning_rate": 2.7029588529772894e-05,
      "loss": 2.0871,
      "step": 199032
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.9051094055175781,
      "learning_rate": 2.702789168910825e-05,
      "loss": 2.0417,
      "step": 199033
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.7318758964538574,
      "learning_rate": 2.7026194899194343e-05,
      "loss": 2.0372,
      "step": 199034
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.5652735233306885,
      "learning_rate": 2.7024498160031337e-05,
      "loss": 2.2561,
      "step": 199035
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.4864670038223267,
      "learning_rate": 2.7022801471619703e-05,
      "loss": 2.2628,
      "step": 199036
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.432802438735962,
      "learning_rate": 2.7021104833959707e-05,
      "loss": 1.8211,
      "step": 199037
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.580580711364746,
      "learning_rate": 2.701940824705161e-05,
      "loss": 2.1747,
      "step": 199038
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.4187240600585938,
      "learning_rate": 2.701771171089572e-05,
      "loss": 1.9805,
      "step": 199039
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.373923659324646,
      "learning_rate": 2.7016015225492427e-05,
      "loss": 1.9532,
      "step": 199040
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.6047980785369873,
      "learning_rate": 2.701431879084194e-05,
      "loss": 2.092,
      "step": 199041
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.4484879970550537,
      "learning_rate": 2.7012622406944718e-05,
      "loss": 1.9466,
      "step": 199042
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.4292776584625244,
      "learning_rate": 2.7010926073801032e-05,
      "loss": 1.7951,
      "step": 199043
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.4003081321716309,
      "learning_rate": 2.700922979141108e-05,
      "loss": 1.983,
      "step": 199044
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.3517216444015503,
      "learning_rate": 2.700753355977533e-05,
      "loss": 1.9827,
      "step": 199045
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.4078909158706665,
      "learning_rate": 2.700583737889401e-05,
      "loss": 1.9868,
      "step": 199046
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.4588664770126343,
      "learning_rate": 2.700414124876743e-05,
      "loss": 1.931,
      "step": 199047
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.733034372329712,
      "learning_rate": 2.7002445169395982e-05,
      "loss": 1.9713,
      "step": 199048
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.351704716682434,
      "learning_rate": 2.700074914077993e-05,
      "loss": 2.0309,
      "step": 199049
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.3812192678451538,
      "learning_rate": 2.699905316291955e-05,
      "loss": 2.0531,
      "step": 199050
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.438240885734558,
      "learning_rate": 2.6997357235815232e-05,
      "loss": 1.9193,
      "step": 199051
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.3487046957015991,
      "learning_rate": 2.6995661359467213e-05,
      "loss": 1.9858,
      "step": 199052
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.4712783098220825,
      "learning_rate": 2.6993965533875895e-05,
      "loss": 2.1626,
      "step": 199053
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.6400331258773804,
      "learning_rate": 2.6992269759041573e-05,
      "loss": 1.8531,
      "step": 199054
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.3837815523147583,
      "learning_rate": 2.6990574034964552e-05,
      "loss": 2.3522,
      "step": 199055
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.4025444984436035,
      "learning_rate": 2.6988878361645062e-05,
      "loss": 2.1817,
      "step": 199056
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.420972466468811,
      "learning_rate": 2.6987182739083533e-05,
      "loss": 2.1028,
      "step": 199057
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.3949239253997803,
      "learning_rate": 2.6985487167280207e-05,
      "loss": 1.9544,
      "step": 199058
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.564227819442749,
      "learning_rate": 2.6983791646235475e-05,
      "loss": 2.0778,
      "step": 199059
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.4301859140396118,
      "learning_rate": 2.698209617594964e-05,
      "loss": 1.7516,
      "step": 199060
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.390737771987915,
      "learning_rate": 2.6980400756422936e-05,
      "loss": 2.0194,
      "step": 199061
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.428507924079895,
      "learning_rate": 2.697870538765573e-05,
      "loss": 1.9793,
      "step": 199062
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.4183036088943481,
      "learning_rate": 2.697701006964835e-05,
      "loss": 1.9678,
      "step": 199063
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.4139130115509033,
      "learning_rate": 2.697531480240107e-05,
      "loss": 2.0755,
      "step": 199064
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.5787911415100098,
      "learning_rate": 2.6973619585914252e-05,
      "loss": 1.9722,
      "step": 199065
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.5778048038482666,
      "learning_rate": 2.6971924420188163e-05,
      "loss": 2.0185,
      "step": 199066
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.435319423675537,
      "learning_rate": 2.6970229305223268e-05,
      "loss": 2.159,
      "step": 199067
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.628607988357544,
      "learning_rate": 2.6968534241019634e-05,
      "loss": 2.007,
      "step": 199068
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.6408551931381226,
      "learning_rate": 2.6966839227577764e-05,
      "loss": 1.9983,
      "step": 199069
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.5496954917907715,
      "learning_rate": 2.6965144264897852e-05,
      "loss": 1.9322,
      "step": 199070
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.5366368293762207,
      "learning_rate": 2.6963449352980338e-05,
      "loss": 2.0595,
      "step": 199071
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.4597513675689697,
      "learning_rate": 2.6961754491825416e-05,
      "loss": 2.1098,
      "step": 199072
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.4326146841049194,
      "learning_rate": 2.6960059681433555e-05,
      "loss": 2.0044,
      "step": 199073
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.3520303964614868,
      "learning_rate": 2.6958364921804853e-05,
      "loss": 2.0553,
      "step": 199074
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.4894345998764038,
      "learning_rate": 2.6956670212939845e-05,
      "loss": 1.8624,
      "step": 199075
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.4764634370803833,
      "learning_rate": 2.6954975554838666e-05,
      "loss": 1.9666,
      "step": 199076
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.4676986932754517,
      "learning_rate": 2.6953280947501745e-05,
      "loss": 2.0625,
      "step": 199077
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.5473583936691284,
      "learning_rate": 2.6951586390929347e-05,
      "loss": 1.9083,
      "step": 199078
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.3835493326187134,
      "learning_rate": 2.6949891885121877e-05,
      "loss": 2.0388,
      "step": 199079
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.4300028085708618,
      "learning_rate": 2.6948197430079497e-05,
      "loss": 2.0182,
      "step": 199080
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.6092039346694946,
      "learning_rate": 2.6946503025802645e-05,
      "loss": 1.9011,
      "step": 199081
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.3487162590026855,
      "learning_rate": 2.694480867229152e-05,
      "loss": 2.1683,
      "step": 199082
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.2904072999954224,
      "learning_rate": 2.694311436954658e-05,
      "loss": 1.7597,
      "step": 199083
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.3562707901000977,
      "learning_rate": 2.6941420117568e-05,
      "loss": 1.8719,
      "step": 199084
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.35395348072052,
      "learning_rate": 2.6939725916356313e-05,
      "loss": 1.8181,
      "step": 199085
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.3708008527755737,
      "learning_rate": 2.6938031765911548e-05,
      "loss": 2.0733,
      "step": 199086
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.4942951202392578,
      "learning_rate": 2.6936337666234208e-05,
      "loss": 2.0033,
      "step": 199087
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.4212191104888916,
      "learning_rate": 2.6934643617324526e-05,
      "loss": 1.9017,
      "step": 199088
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.371268630027771,
      "learning_rate": 2.6932949619182865e-05,
      "loss": 2.0765,
      "step": 199089
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.345044732093811,
      "learning_rate": 2.6931255671809492e-05,
      "loss": 1.8419,
      "step": 199090
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.4027750492095947,
      "learning_rate": 2.692956177520488e-05,
      "loss": 2.0387,
      "step": 199091
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.383805274963379,
      "learning_rate": 2.6927867929369086e-05,
      "loss": 1.972,
      "step": 199092
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.458629846572876,
      "learning_rate": 2.6926174134302615e-05,
      "loss": 2.0606,
      "step": 199093
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.2974960803985596,
      "learning_rate": 2.6924480390005666e-05,
      "loss": 1.9473,
      "step": 199094
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.519974708557129,
      "learning_rate": 2.692278669647864e-05,
      "loss": 1.9557,
      "step": 199095
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.3728835582733154,
      "learning_rate": 2.6921093053721797e-05,
      "loss": 1.9404,
      "step": 199096
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.5500479936599731,
      "learning_rate": 2.6919399461735582e-05,
      "loss": 2.1154,
      "step": 199097
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.311711311340332,
      "learning_rate": 2.6917705920520115e-05,
      "loss": 1.9963,
      "step": 199098
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.314431071281433,
      "learning_rate": 2.691601243007584e-05,
      "loss": 1.9053,
      "step": 199099
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.56600022315979,
      "learning_rate": 2.691431899040295e-05,
      "loss": 2.1896,
      "step": 199100
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.3523602485656738,
      "learning_rate": 2.6912625601501947e-05,
      "loss": 2.0377,
      "step": 199101
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.5128988027572632,
      "learning_rate": 2.6910932263372963e-05,
      "loss": 1.9241,
      "step": 199102
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.4295847415924072,
      "learning_rate": 2.69092389760165e-05,
      "loss": 1.9232,
      "step": 199103
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.7961422204971313,
      "learning_rate": 2.6907545739432657e-05,
      "loss": 1.7643,
      "step": 199104
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.3968924283981323,
      "learning_rate": 2.6905852553621898e-05,
      "loss": 2.0264,
      "step": 199105
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.3893232345581055,
      "learning_rate": 2.690415941858446e-05,
      "loss": 1.9765,
      "step": 199106
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.8512965440750122,
      "learning_rate": 2.690246633432074e-05,
      "loss": 2.13,
      "step": 199107
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.4772562980651855,
      "learning_rate": 2.690077330083097e-05,
      "loss": 2.0098,
      "step": 199108
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.4916285276412964,
      "learning_rate": 2.6899080318115553e-05,
      "loss": 1.9975,
      "step": 199109
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.3746463060379028,
      "learning_rate": 2.6897387386174717e-05,
      "loss": 2.1032,
      "step": 199110
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.3355170488357544,
      "learning_rate": 2.6895694505008837e-05,
      "loss": 1.9137,
      "step": 199111
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.3775601387023926,
      "learning_rate": 2.6894001674618137e-05,
      "loss": 2.1419,
      "step": 199112
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.3008053302764893,
      "learning_rate": 2.6892308895003088e-05,
      "loss": 2.1204,
      "step": 199113
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.432486891746521,
      "learning_rate": 2.6890616166163826e-05,
      "loss": 1.9097,
      "step": 199114
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.4255561828613281,
      "learning_rate": 2.688892348810081e-05,
      "loss": 1.8258,
      "step": 199115
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.4392478466033936,
      "learning_rate": 2.6887230860814313e-05,
      "loss": 1.9286,
      "step": 199116
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.5387569665908813,
      "learning_rate": 2.688553828430463e-05,
      "loss": 1.8763,
      "step": 199117
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.3583920001983643,
      "learning_rate": 2.6883845758572e-05,
      "loss": 2.1082,
      "step": 199118
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.3204855918884277,
      "learning_rate": 2.6882153283616914e-05,
      "loss": 1.9257,
      "step": 199119
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.4824484586715698,
      "learning_rate": 2.6880460859439512e-05,
      "loss": 1.8961,
      "step": 199120
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.4913818836212158,
      "learning_rate": 2.6878768486040258e-05,
      "loss": 2.1463,
      "step": 199121
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.3941599130630493,
      "learning_rate": 2.6877076163419386e-05,
      "loss": 2.1274,
      "step": 199122
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.324249505996704,
      "learning_rate": 2.6875383891577197e-05,
      "loss": 1.8484,
      "step": 199123
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.3249726295471191,
      "learning_rate": 2.687369167051402e-05,
      "loss": 1.7102,
      "step": 199124
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.433680534362793,
      "learning_rate": 2.6871999500230225e-05,
      "loss": 1.9528,
      "step": 199125
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.5044511556625366,
      "learning_rate": 2.6870307380726007e-05,
      "loss": 2.1983,
      "step": 199126
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.4258975982666016,
      "learning_rate": 2.6868615312001808e-05,
      "loss": 1.6868,
      "step": 199127
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.4985086917877197,
      "learning_rate": 2.6866923294057884e-05,
      "loss": 1.886,
      "step": 199128
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.3794547319412231,
      "learning_rate": 2.6865231326894544e-05,
      "loss": 1.8709,
      "step": 199129
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.3523801565170288,
      "learning_rate": 2.6863539410512114e-05,
      "loss": 1.7897,
      "step": 199130
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.506583333015442,
      "learning_rate": 2.686184754491093e-05,
      "loss": 1.9934,
      "step": 199131
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.4631520509719849,
      "learning_rate": 2.686015573009126e-05,
      "loss": 1.8563,
      "step": 199132
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.657982587814331,
      "learning_rate": 2.6858463966053466e-05,
      "loss": 1.9151,
      "step": 199133
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.4196172952651978,
      "learning_rate": 2.6856772252797854e-05,
      "loss": 1.9871,
      "step": 199134
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.6395299434661865,
      "learning_rate": 2.6855080590324652e-05,
      "loss": 1.8508,
      "step": 199135
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.2185180187225342,
      "learning_rate": 2.6853388978634295e-05,
      "loss": 1.9131,
      "step": 199136
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.4707480669021606,
      "learning_rate": 2.685169741772708e-05,
      "loss": 1.8757,
      "step": 199137
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.4487982988357544,
      "learning_rate": 2.6850005907603212e-05,
      "loss": 1.8157,
      "step": 199138
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.4181394577026367,
      "learning_rate": 2.6848314448263154e-05,
      "loss": 2.039,
      "step": 199139
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.4297822713851929,
      "learning_rate": 2.6846623039707137e-05,
      "loss": 2.072,
      "step": 199140
    },
    {
      "epoch": 2.59,
      "grad_norm": 2.222067356109619,
      "learning_rate": 2.6844931681935434e-05,
      "loss": 1.8808,
      "step": 199141
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.3605380058288574,
      "learning_rate": 2.6843240374948472e-05,
      "loss": 1.8553,
      "step": 199142
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.4838266372680664,
      "learning_rate": 2.684154911874645e-05,
      "loss": 1.9985,
      "step": 199143
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.4737006425857544,
      "learning_rate": 2.683985791332981e-05,
      "loss": 2.0842,
      "step": 199144
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.3262016773223877,
      "learning_rate": 2.6838166758698775e-05,
      "loss": 1.9453,
      "step": 199145
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.3593271970748901,
      "learning_rate": 2.6836475654853685e-05,
      "loss": 1.9659,
      "step": 199146
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.7769496440887451,
      "learning_rate": 2.68347846017948e-05,
      "loss": 1.9736,
      "step": 199147
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.414526104927063,
      "learning_rate": 2.683309359952256e-05,
      "loss": 1.747,
      "step": 199148
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.7636947631835938,
      "learning_rate": 2.6831402648037126e-05,
      "loss": 2.1812,
      "step": 199149
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.395409345626831,
      "learning_rate": 2.6829711747338967e-05,
      "loss": 1.9274,
      "step": 199150
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.4817414283752441,
      "learning_rate": 2.682802089742825e-05,
      "loss": 2.1582,
      "step": 199151
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.4746202230453491,
      "learning_rate": 2.682633009830547e-05,
      "loss": 2.0994,
      "step": 199152
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.483088493347168,
      "learning_rate": 2.68246393499707e-05,
      "loss": 1.9731,
      "step": 199153
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.4791005849838257,
      "learning_rate": 2.682294865242447e-05,
      "loss": 2.052,
      "step": 199154
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.6045386791229248,
      "learning_rate": 2.6821258005666947e-05,
      "loss": 2.2193,
      "step": 199155
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.5937910079956055,
      "learning_rate": 2.6819567409698594e-05,
      "loss": 2.2324,
      "step": 199156
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.6642568111419678,
      "learning_rate": 2.681787686451955e-05,
      "loss": 1.9576,
      "step": 199157
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.6145281791687012,
      "learning_rate": 2.6816186370130343e-05,
      "loss": 1.9721,
      "step": 199158
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.4968922138214111,
      "learning_rate": 2.6814495926531043e-05,
      "loss": 2.303,
      "step": 199159
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.3026843070983887,
      "learning_rate": 2.6812805533722147e-05,
      "loss": 1.8656,
      "step": 199160
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.5921710729599,
      "learning_rate": 2.6811115191703857e-05,
      "loss": 2.2761,
      "step": 199161
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.4264137744903564,
      "learning_rate": 2.6809424900476573e-05,
      "loss": 2.1299,
      "step": 199162
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.582598090171814,
      "learning_rate": 2.6807734660040525e-05,
      "loss": 2.1544,
      "step": 199163
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.4326460361480713,
      "learning_rate": 2.6806044470396214e-05,
      "loss": 1.8762,
      "step": 199164
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.5523326396942139,
      "learning_rate": 2.6804354331543676e-05,
      "loss": 1.9924,
      "step": 199165
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.3941658735275269,
      "learning_rate": 2.680266424348344e-05,
      "loss": 1.9022,
      "step": 199166
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.3677890300750732,
      "learning_rate": 2.6800974206215675e-05,
      "loss": 2.0554,
      "step": 199167
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.5337954759597778,
      "learning_rate": 2.6799284219740813e-05,
      "loss": 2.0148,
      "step": 199168
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.4363296031951904,
      "learning_rate": 2.6797594284059088e-05,
      "loss": 2.1264,
      "step": 199169
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.5880299806594849,
      "learning_rate": 2.679590439917093e-05,
      "loss": 1.7295,
      "step": 199170
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.4135831594467163,
      "learning_rate": 2.679421456507648e-05,
      "loss": 1.8607,
      "step": 199171
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.351595163345337,
      "learning_rate": 2.6792524781776225e-05,
      "loss": 1.9146,
      "step": 199172
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.361624002456665,
      "learning_rate": 2.679083504927031e-05,
      "loss": 1.9566,
      "step": 199173
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.6912728548049927,
      "learning_rate": 2.6789145367559194e-05,
      "loss": 2.2946,
      "step": 199174
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.3986661434173584,
      "learning_rate": 2.678745573664308e-05,
      "loss": 1.9396,
      "step": 199175
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.4730520248413086,
      "learning_rate": 2.6785766156522372e-05,
      "loss": 2.062,
      "step": 199176
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.5871257781982422,
      "learning_rate": 2.678407662719736e-05,
      "loss": 1.9895,
      "step": 199177
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.5867352485656738,
      "learning_rate": 2.678238714866835e-05,
      "loss": 2.0374,
      "step": 199178
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.4147164821624756,
      "learning_rate": 2.6780697720935607e-05,
      "loss": 1.9019,
      "step": 199179
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.5200128555297852,
      "learning_rate": 2.6779008343999532e-05,
      "loss": 2.0085,
      "step": 199180
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.5317424535751343,
      "learning_rate": 2.6777319017860354e-05,
      "loss": 2.0341,
      "step": 199181
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.4547022581100464,
      "learning_rate": 2.6775629742518478e-05,
      "loss": 1.9197,
      "step": 199182
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.3046993017196655,
      "learning_rate": 2.677394051797417e-05,
      "loss": 1.9291,
      "step": 199183
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.3333975076675415,
      "learning_rate": 2.6772251344227726e-05,
      "loss": 2.1044,
      "step": 199184
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.3869916200637817,
      "learning_rate": 2.6770562221279446e-05,
      "loss": 1.8707,
      "step": 199185
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.440491795539856,
      "learning_rate": 2.6768873149129732e-05,
      "loss": 2.0145,
      "step": 199186
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.455522894859314,
      "learning_rate": 2.6767184127778752e-05,
      "loss": 2.0383,
      "step": 199187
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.5498020648956299,
      "learning_rate": 2.6765495157227e-05,
      "loss": 1.8834,
      "step": 199188
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.4577027559280396,
      "learning_rate": 2.676380623747468e-05,
      "loss": 1.7905,
      "step": 199189
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.5478079319000244,
      "learning_rate": 2.6762117368522164e-05,
      "loss": 2.111,
      "step": 199190
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.400687336921692,
      "learning_rate": 2.6760428550369638e-05,
      "loss": 1.8955,
      "step": 199191
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.8123679161071777,
      "learning_rate": 2.675873978301758e-05,
      "loss": 1.9406,
      "step": 199192
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.6560949087142944,
      "learning_rate": 2.675705106646615e-05,
      "loss": 1.9062,
      "step": 199193
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.3800368309020996,
      "learning_rate": 2.6755362400715818e-05,
      "loss": 1.8864,
      "step": 199194
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.3703327178955078,
      "learning_rate": 2.6753673785766818e-05,
      "loss": 1.8385,
      "step": 199195
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.6430869102478027,
      "learning_rate": 2.675198522161944e-05,
      "loss": 2.3038,
      "step": 199196
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.5397359132766724,
      "learning_rate": 2.6750296708273998e-05,
      "loss": 2.0797,
      "step": 199197
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.3515945672988892,
      "learning_rate": 2.6748608245730918e-05,
      "loss": 1.8771,
      "step": 199198
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.5682719945907593,
      "learning_rate": 2.6746919833990332e-05,
      "loss": 2.1538,
      "step": 199199
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.5529037714004517,
      "learning_rate": 2.674523147305271e-05,
      "loss": 1.8361,
      "step": 199200
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.5016515254974365,
      "learning_rate": 2.6743543162918314e-05,
      "loss": 2.0983,
      "step": 199201
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.3872617483139038,
      "learning_rate": 2.674185490358748e-05,
      "loss": 1.9952,
      "step": 199202
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.4290415048599243,
      "learning_rate": 2.6740166695060406e-05,
      "loss": 2.0671,
      "step": 199203
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.4354788064956665,
      "learning_rate": 2.673847853733756e-05,
      "loss": 1.8952,
      "step": 199204
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.5149040222167969,
      "learning_rate": 2.6736790430419108e-05,
      "loss": 2.1774,
      "step": 199205
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.5117233991622925,
      "learning_rate": 2.6735102374305516e-05,
      "loss": 1.7756,
      "step": 199206
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.4334633350372314,
      "learning_rate": 2.673341436899705e-05,
      "loss": 2.0819,
      "step": 199207
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.3412822484970093,
      "learning_rate": 2.673172641449398e-05,
      "loss": 1.9715,
      "step": 199208
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.4462766647338867,
      "learning_rate": 2.673003851079657e-05,
      "loss": 2.2242,
      "step": 199209
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.4921540021896362,
      "learning_rate": 2.6728350657905286e-05,
      "loss": 1.8783,
      "step": 199210
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.4871054887771606,
      "learning_rate": 2.672666285582029e-05,
      "loss": 1.9823,
      "step": 199211
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.3789982795715332,
      "learning_rate": 2.6724975104542023e-05,
      "loss": 1.9134,
      "step": 199212
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.4134411811828613,
      "learning_rate": 2.672328740407075e-05,
      "loss": 2.1588,
      "step": 199213
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.5245167016983032,
      "learning_rate": 2.672159975440673e-05,
      "loss": 1.908,
      "step": 199214
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.4177284240722656,
      "learning_rate": 2.671991215555037e-05,
      "loss": 1.9146,
      "step": 199215
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.4589534997940063,
      "learning_rate": 2.6718224607501903e-05,
      "loss": 2.0161,
      "step": 199216
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.3247686624526978,
      "learning_rate": 2.671653711026166e-05,
      "loss": 1.9467,
      "step": 199217
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.3842172622680664,
      "learning_rate": 2.6714849663830008e-05,
      "loss": 1.811,
      "step": 199218
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.630861759185791,
      "learning_rate": 2.6713162268207243e-05,
      "loss": 1.9879,
      "step": 199219
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.518749713897705,
      "learning_rate": 2.6711474923393607e-05,
      "loss": 2.0356,
      "step": 199220
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.345041275024414,
      "learning_rate": 2.670978762938949e-05,
      "loss": 1.8008,
      "step": 199221
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.4820770025253296,
      "learning_rate": 2.67081003861952e-05,
      "loss": 2.0213,
      "step": 199222
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.4373668432235718,
      "learning_rate": 2.6706413193810992e-05,
      "loss": 2.1683,
      "step": 199223
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.4722037315368652,
      "learning_rate": 2.6704726052237247e-05,
      "loss": 1.9909,
      "step": 199224
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.3849530220031738,
      "learning_rate": 2.6703038961474287e-05,
      "loss": 1.9974,
      "step": 199225
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.4219508171081543,
      "learning_rate": 2.670135192152232e-05,
      "loss": 1.8873,
      "step": 199226
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.372673749923706,
      "learning_rate": 2.669966493238177e-05,
      "loss": 1.8844,
      "step": 199227
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.3083325624465942,
      "learning_rate": 2.6697977994052877e-05,
      "loss": 1.9388,
      "step": 199228
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.4235038757324219,
      "learning_rate": 2.669629110653604e-05,
      "loss": 2.118,
      "step": 199229
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.4962536096572876,
      "learning_rate": 2.669460426983152e-05,
      "loss": 2.174,
      "step": 199230
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.3320902585983276,
      "learning_rate": 2.6692917483939623e-05,
      "loss": 1.8755,
      "step": 199231
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.400996446609497,
      "learning_rate": 2.6691230748860614e-05,
      "loss": 1.9806,
      "step": 199232
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.3617697954177856,
      "learning_rate": 2.6689544064594925e-05,
      "loss": 1.9629,
      "step": 199233
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.4392597675323486,
      "learning_rate": 2.6687857431142757e-05,
      "loss": 2.0097,
      "step": 199234
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.3748459815979004,
      "learning_rate": 2.668617084850454e-05,
      "loss": 2.0046,
      "step": 199235
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.3237576484680176,
      "learning_rate": 2.6684484316680443e-05,
      "loss": 1.9185,
      "step": 199236
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.365652084350586,
      "learning_rate": 2.6682797835671e-05,
      "loss": 1.9424,
      "step": 199237
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.534415602684021,
      "learning_rate": 2.6681111405476242e-05,
      "loss": 2.0235,
      "step": 199238
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.391047716140747,
      "learning_rate": 2.66794250260967e-05,
      "loss": 1.9283,
      "step": 199239
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.610804796218872,
      "learning_rate": 2.6677738697532582e-05,
      "loss": 2.086,
      "step": 199240
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.5065374374389648,
      "learning_rate": 2.6676052419784246e-05,
      "loss": 2.1061,
      "step": 199241
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.4028085470199585,
      "learning_rate": 2.667436619285196e-05,
      "loss": 1.8136,
      "step": 199242
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.5052545070648193,
      "learning_rate": 2.6672680016736092e-05,
      "loss": 2.0527,
      "step": 199243
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.4776142835617065,
      "learning_rate": 2.6670993891436976e-05,
      "loss": 2.073,
      "step": 199244
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.4481552839279175,
      "learning_rate": 2.6669307816954843e-05,
      "loss": 1.9448,
      "step": 199245
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.4866302013397217,
      "learning_rate": 2.6667621793289994e-05,
      "loss": 1.9882,
      "step": 199246
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.425032615661621,
      "learning_rate": 2.6665935820442863e-05,
      "loss": 1.8505,
      "step": 199247
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.4734035730361938,
      "learning_rate": 2.666424989841365e-05,
      "loss": 2.2485,
      "step": 199248
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.4286837577819824,
      "learning_rate": 2.6662564027202747e-05,
      "loss": 2.0684,
      "step": 199249
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.3685011863708496,
      "learning_rate": 2.6660878206810433e-05,
      "loss": 1.9763,
      "step": 199250
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.4716960191726685,
      "learning_rate": 2.665919243723703e-05,
      "loss": 2.2671,
      "step": 199251
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.5056058168411255,
      "learning_rate": 2.665750671848278e-05,
      "loss": 2.0194,
      "step": 199252
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.3134804964065552,
      "learning_rate": 2.6655821050548142e-05,
      "loss": 1.9833,
      "step": 199253
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.4719691276550293,
      "learning_rate": 2.6654135433433256e-05,
      "loss": 1.8751,
      "step": 199254
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.3338303565979004,
      "learning_rate": 2.6652449867138616e-05,
      "loss": 1.9246,
      "step": 199255
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.4865103960037231,
      "learning_rate": 2.665076435166442e-05,
      "loss": 2.1487,
      "step": 199256
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.5598978996276855,
      "learning_rate": 2.664907888701101e-05,
      "loss": 1.8024,
      "step": 199257
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.5426959991455078,
      "learning_rate": 2.6647393473178648e-05,
      "loss": 1.6994,
      "step": 199258
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.4404841661453247,
      "learning_rate": 2.664570811016773e-05,
      "loss": 1.8545,
      "step": 199259
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.4406299591064453,
      "learning_rate": 2.6644022797978492e-05,
      "loss": 1.9643,
      "step": 199260
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.4444055557250977,
      "learning_rate": 2.664233753661137e-05,
      "loss": 1.9894,
      "step": 199261
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.4073607921600342,
      "learning_rate": 2.664065232606656e-05,
      "loss": 1.8227,
      "step": 199262
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.4481977224349976,
      "learning_rate": 2.6638967166344426e-05,
      "loss": 1.8453,
      "step": 199263
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.4309226274490356,
      "learning_rate": 2.663728205744521e-05,
      "loss": 2.0592,
      "step": 199264
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.3364040851593018,
      "learning_rate": 2.663559699936937e-05,
      "loss": 1.846,
      "step": 199265
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.3601195812225342,
      "learning_rate": 2.663391199211704e-05,
      "loss": 1.9383,
      "step": 199266
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.4826127290725708,
      "learning_rate": 2.6632227035688692e-05,
      "loss": 2.1714,
      "step": 199267
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.5213634967803955,
      "learning_rate": 2.6630542130084586e-05,
      "loss": 2.0914,
      "step": 199268
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.4192081689834595,
      "learning_rate": 2.6628857275304994e-05,
      "loss": 2.0464,
      "step": 199269
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.3772398233413696,
      "learning_rate": 2.662717247135021e-05,
      "loss": 1.966,
      "step": 199270
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.3689024448394775,
      "learning_rate": 2.662548771822067e-05,
      "loss": 1.9727,
      "step": 199271
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.6005054712295532,
      "learning_rate": 2.6623803015916544e-05,
      "loss": 1.9489,
      "step": 199272
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.3524281978607178,
      "learning_rate": 2.662211836443826e-05,
      "loss": 2.0214,
      "step": 199273
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.3711942434310913,
      "learning_rate": 2.662043376378612e-05,
      "loss": 1.8669,
      "step": 199274
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.2821192741394043,
      "learning_rate": 2.6618749213960357e-05,
      "loss": 1.8566,
      "step": 199275
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.4232680797576904,
      "learning_rate": 2.66170647149613e-05,
      "loss": 1.9683,
      "step": 199276
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.4751075506210327,
      "learning_rate": 2.6615380266789355e-05,
      "loss": 1.9766,
      "step": 199277
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.4057337045669556,
      "learning_rate": 2.6613695869444686e-05,
      "loss": 1.9524,
      "step": 199278
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.5121384859085083,
      "learning_rate": 2.6612011522927756e-05,
      "loss": 2.1428,
      "step": 199279
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.3813982009887695,
      "learning_rate": 2.6610327227238837e-05,
      "loss": 1.9255,
      "step": 199280
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.3302605152130127,
      "learning_rate": 2.6608642982378193e-05,
      "loss": 1.9859,
      "step": 199281
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.3638029098510742,
      "learning_rate": 2.6606958788346123e-05,
      "loss": 2.0343,
      "step": 199282
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.4836708307266235,
      "learning_rate": 2.6605274645143027e-05,
      "loss": 1.9761,
      "step": 199283
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.3762664794921875,
      "learning_rate": 2.6603590552769104e-05,
      "loss": 2.0122,
      "step": 199284
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.519993543624878,
      "learning_rate": 2.6601906511224824e-05,
      "loss": 1.8022,
      "step": 199285
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.4907640218734741,
      "learning_rate": 2.6600222520510382e-05,
      "loss": 1.9404,
      "step": 199286
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.536652684211731,
      "learning_rate": 2.6598538580626117e-05,
      "loss": 2.0122,
      "step": 199287
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.3400336503982544,
      "learning_rate": 2.6596854691572288e-05,
      "loss": 1.8398,
      "step": 199288
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.5142706632614136,
      "learning_rate": 2.6595170853349333e-05,
      "loss": 2.1442,
      "step": 199289
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.4798089265823364,
      "learning_rate": 2.6593487065957453e-05,
      "loss": 1.981,
      "step": 199290
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.4816443920135498,
      "learning_rate": 2.6591803329397042e-05,
      "loss": 2.1404,
      "step": 199291
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.4620293378829956,
      "learning_rate": 2.6590119643668373e-05,
      "loss": 1.9756,
      "step": 199292
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.6072545051574707,
      "learning_rate": 2.6588436008771774e-05,
      "loss": 2.0227,
      "step": 199293
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.6478127241134644,
      "learning_rate": 2.658675242470748e-05,
      "loss": 2.097,
      "step": 199294
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.576827049255371,
      "learning_rate": 2.6585068891475925e-05,
      "loss": 2.0012,
      "step": 199295
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.402727723121643,
      "learning_rate": 2.658338540907734e-05,
      "loss": 2.0416,
      "step": 199296
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.494738221168518,
      "learning_rate": 2.6581701977512094e-05,
      "loss": 2.0746,
      "step": 199297
    },
    {
      "epoch": 2.59,
      "grad_norm": 1.3419909477233887,
      "learning_rate": 2.658001859678045e-05,
      "loss": 1.8169,
      "step": 199298
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.422484040260315,
      "learning_rate": 2.6578335266882712e-05,
      "loss": 2.1421,
      "step": 199299
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.377720832824707,
      "learning_rate": 2.6576651987819274e-05,
      "loss": 1.9331,
      "step": 199300
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.749165654182434,
      "learning_rate": 2.657496875959041e-05,
      "loss": 2.0941,
      "step": 199301
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.5237857103347778,
      "learning_rate": 2.6573285582196347e-05,
      "loss": 1.9626,
      "step": 199302
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.4046604633331299,
      "learning_rate": 2.6571602455637552e-05,
      "loss": 2.0657,
      "step": 199303
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.547837495803833,
      "learning_rate": 2.656991937991423e-05,
      "loss": 1.9259,
      "step": 199304
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.2745378017425537,
      "learning_rate": 2.6568236355026673e-05,
      "loss": 1.9218,
      "step": 199305
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.5632694959640503,
      "learning_rate": 2.656655338097532e-05,
      "loss": 1.8727,
      "step": 199306
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.5277067422866821,
      "learning_rate": 2.6564870457760367e-05,
      "loss": 2.1284,
      "step": 199307
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.885688304901123,
      "learning_rate": 2.656318758538215e-05,
      "loss": 1.9451,
      "step": 199308
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.5783677101135254,
      "learning_rate": 2.6561504763841034e-05,
      "loss": 2.0991,
      "step": 199309
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.6779922246932983,
      "learning_rate": 2.655982199313722e-05,
      "loss": 2.1163,
      "step": 199310
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.3805848360061646,
      "learning_rate": 2.655813927327117e-05,
      "loss": 2.0843,
      "step": 199311
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.4587334394454956,
      "learning_rate": 2.655645660424316e-05,
      "loss": 2.0826,
      "step": 199312
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.3673352003097534,
      "learning_rate": 2.6554773986053346e-05,
      "loss": 1.8743,
      "step": 199313
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.4199855327606201,
      "learning_rate": 2.6553091418702267e-05,
      "loss": 2.0229,
      "step": 199314
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.4302098751068115,
      "learning_rate": 2.655140890219012e-05,
      "loss": 2.0036,
      "step": 199315
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.4324003458023071,
      "learning_rate": 2.6549726436517137e-05,
      "loss": 2.1609,
      "step": 199316
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.5410304069519043,
      "learning_rate": 2.6548044021683824e-05,
      "loss": 2.1273,
      "step": 199317
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.4967150688171387,
      "learning_rate": 2.6546361657690374e-05,
      "loss": 2.0177,
      "step": 199318
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.3761262893676758,
      "learning_rate": 2.6544679344537058e-05,
      "loss": 1.7551,
      "step": 199319
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.3844664096832275,
      "learning_rate": 2.6542997082224305e-05,
      "loss": 2.1937,
      "step": 199320
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.5682264566421509,
      "learning_rate": 2.654131487075235e-05,
      "loss": 1.8537,
      "step": 199321
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.3389307260513306,
      "learning_rate": 2.6539632710121496e-05,
      "loss": 1.9639,
      "step": 199322
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.5577510595321655,
      "learning_rate": 2.6537950600332137e-05,
      "loss": 2.1638,
      "step": 199323
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.3724960088729858,
      "learning_rate": 2.653626854138454e-05,
      "loss": 1.9734,
      "step": 199324
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.6608145236968994,
      "learning_rate": 2.6534586533278947e-05,
      "loss": 2.2847,
      "step": 199325
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.3168275356292725,
      "learning_rate": 2.6532904576015813e-05,
      "loss": 2.0523,
      "step": 199326
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.3845150470733643,
      "learning_rate": 2.6531222669595277e-05,
      "loss": 2.0231,
      "step": 199327
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.383863925933838,
      "learning_rate": 2.6529540814017836e-05,
      "loss": 1.9935,
      "step": 199328
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.5788525342941284,
      "learning_rate": 2.6527859009283723e-05,
      "loss": 2.1293,
      "step": 199329
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.620936393737793,
      "learning_rate": 2.652617725539321e-05,
      "loss": 2.2489,
      "step": 199330
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.5023432970046997,
      "learning_rate": 2.6524495552346624e-05,
      "loss": 2.1028,
      "step": 199331
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.3555172681808472,
      "learning_rate": 2.6522813900144334e-05,
      "loss": 1.9906,
      "step": 199332
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.6345211267471313,
      "learning_rate": 2.652113229878654e-05,
      "loss": 2.2614,
      "step": 199333
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.4811618328094482,
      "learning_rate": 2.651945074827371e-05,
      "loss": 1.9679,
      "step": 199334
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.462632417678833,
      "learning_rate": 2.6517769248606068e-05,
      "loss": 1.9299,
      "step": 199335
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.5157470703125,
      "learning_rate": 2.6516087799783958e-05,
      "loss": 1.9085,
      "step": 199336
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.488356113433838,
      "learning_rate": 2.6514406401807576e-05,
      "loss": 2.0469,
      "step": 199337
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.3859864473342896,
      "learning_rate": 2.651272505467742e-05,
      "loss": 2.3076,
      "step": 199338
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.467498779296875,
      "learning_rate": 2.6511043758393624e-05,
      "loss": 2.0466,
      "step": 199339
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.4276947975158691,
      "learning_rate": 2.6509362512956654e-05,
      "loss": 1.7993,
      "step": 199340
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.4896894693374634,
      "learning_rate": 2.650768131836678e-05,
      "loss": 2.2545,
      "step": 199341
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.4345062971115112,
      "learning_rate": 2.6506000174624265e-05,
      "loss": 2.1702,
      "step": 199342
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.8428138494491577,
      "learning_rate": 2.6504319081729408e-05,
      "loss": 1.8841,
      "step": 199343
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.6351544857025146,
      "learning_rate": 2.650263803968258e-05,
      "loss": 2.1195,
      "step": 199344
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.482190728187561,
      "learning_rate": 2.650095704848404e-05,
      "loss": 2.0758,
      "step": 199345
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.5363296270370483,
      "learning_rate": 2.6499276108134194e-05,
      "loss": 2.0529,
      "step": 199346
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.5050016641616821,
      "learning_rate": 2.6497595218633307e-05,
      "loss": 1.9706,
      "step": 199347
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.4560449123382568,
      "learning_rate": 2.6495914379981677e-05,
      "loss": 2.106,
      "step": 199348
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.4691804647445679,
      "learning_rate": 2.6494233592179536e-05,
      "loss": 1.9615,
      "step": 199349
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.4456554651260376,
      "learning_rate": 2.6492552855227357e-05,
      "loss": 2.2792,
      "step": 199350
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.441328525543213,
      "learning_rate": 2.649087216912533e-05,
      "loss": 2.182,
      "step": 199351
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.434205174446106,
      "learning_rate": 2.6489191533873865e-05,
      "loss": 2.0629,
      "step": 199352
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.4188421964645386,
      "learning_rate": 2.648751094947319e-05,
      "loss": 1.8498,
      "step": 199353
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.4366297721862793,
      "learning_rate": 2.648583041592367e-05,
      "loss": 1.9936,
      "step": 199354
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.4081424474716187,
      "learning_rate": 2.648414993322554e-05,
      "loss": 1.9958,
      "step": 199355
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.5265194177627563,
      "learning_rate": 2.6482469501379233e-05,
      "loss": 1.8427,
      "step": 199356
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.4856348037719727,
      "learning_rate": 2.648078912038495e-05,
      "loss": 2.0227,
      "step": 199357
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.4135233163833618,
      "learning_rate": 2.6479108790243085e-05,
      "loss": 1.998,
      "step": 199358
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.5799838304519653,
      "learning_rate": 2.6477428510953945e-05,
      "loss": 1.8428,
      "step": 199359
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.6405147314071655,
      "learning_rate": 2.6475748282517793e-05,
      "loss": 1.8612,
      "step": 199360
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.6716657876968384,
      "learning_rate": 2.6474068104934897e-05,
      "loss": 2.0709,
      "step": 199361
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.4316163063049316,
      "learning_rate": 2.647238797820572e-05,
      "loss": 1.7628,
      "step": 199362
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.4001892805099487,
      "learning_rate": 2.6470707902330402e-05,
      "loss": 2.1101,
      "step": 199363
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.2576441764831543,
      "learning_rate": 2.6469027877309436e-05,
      "loss": 1.9012,
      "step": 199364
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.538502812385559,
      "learning_rate": 2.6467347903142986e-05,
      "loss": 2.0823,
      "step": 199365
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.4022120237350464,
      "learning_rate": 2.6465667979831463e-05,
      "loss": 1.9272,
      "step": 199366
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.4249664545059204,
      "learning_rate": 2.646398810737509e-05,
      "loss": 2.0126,
      "step": 199367
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.581096887588501,
      "learning_rate": 2.6462308285774237e-05,
      "loss": 1.9526,
      "step": 199368
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.5044338703155518,
      "learning_rate": 2.646062851502917e-05,
      "loss": 1.9462,
      "step": 199369
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.5972844362258911,
      "learning_rate": 2.645894879514029e-05,
      "loss": 1.9053,
      "step": 199370
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.336504340171814,
      "learning_rate": 2.645726912610786e-05,
      "loss": 1.8977,
      "step": 199371
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.3992525339126587,
      "learning_rate": 2.6455589507932185e-05,
      "loss": 2.0229,
      "step": 199372
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.4416152238845825,
      "learning_rate": 2.6453909940613526e-05,
      "loss": 2.084,
      "step": 199373
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.4491639137268066,
      "learning_rate": 2.645223042415232e-05,
      "loss": 1.9785,
      "step": 199374
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.3784602880477905,
      "learning_rate": 2.645055095854873e-05,
      "loss": 1.9694,
      "step": 199375
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.5951532125473022,
      "learning_rate": 2.6448871543803196e-05,
      "loss": 2.1352,
      "step": 199376
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.4148881435394287,
      "learning_rate": 2.6447192179915942e-05,
      "loss": 1.9759,
      "step": 199377
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.325696349143982,
      "learning_rate": 2.644551286688741e-05,
      "loss": 1.7611,
      "step": 199378
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.3984960317611694,
      "learning_rate": 2.6443833604717722e-05,
      "loss": 2.065,
      "step": 199379
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.476527452468872,
      "learning_rate": 2.6442154393407356e-05,
      "loss": 1.9491,
      "step": 199380
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.4117463827133179,
      "learning_rate": 2.6440475232956504e-05,
      "loss": 2.0588,
      "step": 199381
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.4315155744552612,
      "learning_rate": 2.643879612336557e-05,
      "loss": 1.9489,
      "step": 199382
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.500150442123413,
      "learning_rate": 2.643711706463475e-05,
      "loss": 2.0083,
      "step": 199383
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.3188515901565552,
      "learning_rate": 2.643543805676458e-05,
      "loss": 2.0325,
      "step": 199384
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.5401188135147095,
      "learning_rate": 2.6433759099755127e-05,
      "loss": 1.9231,
      "step": 199385
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.4853218793869019,
      "learning_rate": 2.6432080193606818e-05,
      "loss": 1.9113,
      "step": 199386
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.482387661933899,
      "learning_rate": 2.6430401338319928e-05,
      "loss": 2.0322,
      "step": 199387
    },
    {
      "epoch": 2.6,
      "grad_norm": 2.0138537883758545,
      "learning_rate": 2.642872253389482e-05,
      "loss": 2.2987,
      "step": 199388
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.5604361295700073,
      "learning_rate": 2.6427043780331726e-05,
      "loss": 1.9699,
      "step": 199389
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.4048564434051514,
      "learning_rate": 2.6425365077631078e-05,
      "loss": 2.0969,
      "step": 199390
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.279671549797058,
      "learning_rate": 2.6423686425793113e-05,
      "loss": 2.0945,
      "step": 199391
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.458052396774292,
      "learning_rate": 2.6422007824818127e-05,
      "loss": 2.0774,
      "step": 199392
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.3488363027572632,
      "learning_rate": 2.642032927470642e-05,
      "loss": 1.8856,
      "step": 199393
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.3414589166641235,
      "learning_rate": 2.6418650775458394e-05,
      "loss": 1.736,
      "step": 199394
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.5206938982009888,
      "learning_rate": 2.641697232707425e-05,
      "loss": 2.014,
      "step": 199395
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.5637680292129517,
      "learning_rate": 2.6415293929554416e-05,
      "loss": 1.9996,
      "step": 199396
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.3633248805999756,
      "learning_rate": 2.6413615582899128e-05,
      "loss": 2.0253,
      "step": 199397
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.462772250175476,
      "learning_rate": 2.6411937287108653e-05,
      "loss": 1.9544,
      "step": 199398
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.4417963027954102,
      "learning_rate": 2.6410259042183425e-05,
      "loss": 1.9661,
      "step": 199399
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.3623833656311035,
      "learning_rate": 2.6408580848123707e-05,
      "loss": 1.7856,
      "step": 199400
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.613344669342041,
      "learning_rate": 2.6406902704929735e-05,
      "loss": 2.139,
      "step": 199401
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.4253425598144531,
      "learning_rate": 2.6405224612601937e-05,
      "loss": 2.0222,
      "step": 199402
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.4223812818527222,
      "learning_rate": 2.640354657114059e-05,
      "loss": 1.8911,
      "step": 199403
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.4147133827209473,
      "learning_rate": 2.6401868580545914e-05,
      "loss": 2.118,
      "step": 199404
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.265696406364441,
      "learning_rate": 2.6400190640818352e-05,
      "loss": 1.8976,
      "step": 199405
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.4346636533737183,
      "learning_rate": 2.639851275195817e-05,
      "loss": 1.8943,
      "step": 199406
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.4507800340652466,
      "learning_rate": 2.6396834913965593e-05,
      "loss": 1.9456,
      "step": 199407
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.5000776052474976,
      "learning_rate": 2.6395157126841093e-05,
      "loss": 2.0625,
      "step": 199408
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.3860009908676147,
      "learning_rate": 2.639347939058487e-05,
      "loss": 1.8264,
      "step": 199409
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.3797781467437744,
      "learning_rate": 2.6391801705197226e-05,
      "loss": 1.9424,
      "step": 199410
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.5193718671798706,
      "learning_rate": 2.6390124070678552e-05,
      "loss": 2.0806,
      "step": 199411
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.4715288877487183,
      "learning_rate": 2.638844648702906e-05,
      "loss": 2.0312,
      "step": 199412
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.4267144203186035,
      "learning_rate": 2.6386768954249204e-05,
      "loss": 1.9433,
      "step": 199413
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.4168943166732788,
      "learning_rate": 2.6385091472339193e-05,
      "loss": 2.0201,
      "step": 199414
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.4815399646759033,
      "learning_rate": 2.6383414041299355e-05,
      "loss": 2.1528,
      "step": 199415
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.546260952949524,
      "learning_rate": 2.638173666112996e-05,
      "loss": 1.8395,
      "step": 199416
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.5147998332977295,
      "learning_rate": 2.6380059331831404e-05,
      "loss": 2.0172,
      "step": 199417
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.4325854778289795,
      "learning_rate": 2.6378382053403925e-05,
      "loss": 2.1802,
      "step": 199418
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.404115915298462,
      "learning_rate": 2.6376704825847917e-05,
      "loss": 1.9721,
      "step": 199419
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.44722580909729,
      "learning_rate": 2.637502764916365e-05,
      "loss": 1.9399,
      "step": 199420
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.2858712673187256,
      "learning_rate": 2.6373350523351423e-05,
      "loss": 1.7854,
      "step": 199421
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.4874887466430664,
      "learning_rate": 2.63716734484115e-05,
      "loss": 2.0114,
      "step": 199422
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.3807646036148071,
      "learning_rate": 2.636999642434432e-05,
      "loss": 2.1175,
      "step": 199423
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.4735667705535889,
      "learning_rate": 2.6368319451150044e-05,
      "loss": 2.2205,
      "step": 199424
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.3539128303527832,
      "learning_rate": 2.636664252882914e-05,
      "loss": 2.2033,
      "step": 199425
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.3634538650512695,
      "learning_rate": 2.636496565738181e-05,
      "loss": 1.8923,
      "step": 199426
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.538725733757019,
      "learning_rate": 2.6363288836808415e-05,
      "loss": 2.1539,
      "step": 199427
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.5659273862838745,
      "learning_rate": 2.6361612067109195e-05,
      "loss": 2.1311,
      "step": 199428
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.5331215858459473,
      "learning_rate": 2.6359935348284577e-05,
      "loss": 1.7539,
      "step": 199429
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.462556004524231,
      "learning_rate": 2.6358258680334766e-05,
      "loss": 1.9806,
      "step": 199430
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.3915129899978638,
      "learning_rate": 2.6356582063260156e-05,
      "loss": 1.7697,
      "step": 199431
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.5189343690872192,
      "learning_rate": 2.6354905497061052e-05,
      "loss": 2.186,
      "step": 199432
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.4088622331619263,
      "learning_rate": 2.6353228981737685e-05,
      "loss": 1.7683,
      "step": 199433
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.4919947385787964,
      "learning_rate": 2.635155251729042e-05,
      "loss": 1.9774,
      "step": 199434
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.3997530937194824,
      "learning_rate": 2.6349876103719592e-05,
      "loss": 1.8258,
      "step": 199435
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.4114019870758057,
      "learning_rate": 2.6348199741025432e-05,
      "loss": 2.0486,
      "step": 199436
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.4974373579025269,
      "learning_rate": 2.6346523429208344e-05,
      "loss": 1.7403,
      "step": 199437
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.2739882469177246,
      "learning_rate": 2.6344847168268656e-05,
      "loss": 1.9547,
      "step": 199438
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.524497389793396,
      "learning_rate": 2.634317095820657e-05,
      "loss": 1.9754,
      "step": 199439
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.5044596195220947,
      "learning_rate": 2.6341494799022424e-05,
      "loss": 2.0609,
      "step": 199440
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.5652529001235962,
      "learning_rate": 2.633981869071664e-05,
      "loss": 2.2042,
      "step": 199441
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.501062273979187,
      "learning_rate": 2.633814263328936e-05,
      "loss": 2.0713,
      "step": 199442
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.5810627937316895,
      "learning_rate": 2.6336466626741048e-05,
      "loss": 2.1817,
      "step": 199443
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.3826578855514526,
      "learning_rate": 2.6334790671071905e-05,
      "loss": 1.9408,
      "step": 199444
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.5426406860351562,
      "learning_rate": 2.633311476628239e-05,
      "loss": 1.9611,
      "step": 199445
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.508697509765625,
      "learning_rate": 2.6331438912372615e-05,
      "loss": 1.6907,
      "step": 199446
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.6552282571792603,
      "learning_rate": 2.6329763109343038e-05,
      "loss": 2.1456,
      "step": 199447
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.5118125677108765,
      "learning_rate": 2.6328087357193862e-05,
      "loss": 1.9032,
      "step": 199448
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.4199331998825073,
      "learning_rate": 2.6326411655925515e-05,
      "loss": 1.9129,
      "step": 199449
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.4506237506866455,
      "learning_rate": 2.6324736005538238e-05,
      "loss": 2.0492,
      "step": 199450
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.4962308406829834,
      "learning_rate": 2.6323060406032425e-05,
      "loss": 1.8708,
      "step": 199451
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.47383713722229,
      "learning_rate": 2.632138485740821e-05,
      "loss": 1.9571,
      "step": 199452
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.4094098806381226,
      "learning_rate": 2.6319709359666096e-05,
      "loss": 2.0121,
      "step": 199453
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.371351957321167,
      "learning_rate": 2.6318033912806246e-05,
      "loss": 1.9193,
      "step": 199454
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.396055817604065,
      "learning_rate": 2.6316358516829095e-05,
      "loss": 2.076,
      "step": 199455
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.4217495918273926,
      "learning_rate": 2.6314683171734874e-05,
      "loss": 1.8661,
      "step": 199456
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.597865343093872,
      "learning_rate": 2.6313007877523985e-05,
      "loss": 1.9673,
      "step": 199457
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.3869234323501587,
      "learning_rate": 2.631133263419659e-05,
      "loss": 2.018,
      "step": 199458
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.5054304599761963,
      "learning_rate": 2.630965744175313e-05,
      "loss": 2.226,
      "step": 199459
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.4509550333023071,
      "learning_rate": 2.630798230019383e-05,
      "loss": 2.0653,
      "step": 199460
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.4247512817382812,
      "learning_rate": 2.6306307209519063e-05,
      "loss": 1.9566,
      "step": 199461
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.387790322303772,
      "learning_rate": 2.630463216972909e-05,
      "loss": 1.9676,
      "step": 199462
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.3210084438323975,
      "learning_rate": 2.630295718082438e-05,
      "loss": 1.8268,
      "step": 199463
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.565854787826538,
      "learning_rate": 2.6301282242804965e-05,
      "loss": 2.0705,
      "step": 199464
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.3680121898651123,
      "learning_rate": 2.629960735567138e-05,
      "loss": 2.0054,
      "step": 199465
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.516257882118225,
      "learning_rate": 2.6297932519423824e-05,
      "loss": 2.2172,
      "step": 199466
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.3887085914611816,
      "learning_rate": 2.62962577340627e-05,
      "loss": 1.9191,
      "step": 199467
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.5358318090438843,
      "learning_rate": 2.6294582999588197e-05,
      "loss": 1.8749,
      "step": 199468
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.434956669807434,
      "learning_rate": 2.6292908316000825e-05,
      "loss": 1.9908,
      "step": 199469
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.504606008529663,
      "learning_rate": 2.629123368330065e-05,
      "loss": 2.0765,
      "step": 199470
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.2613753080368042,
      "learning_rate": 2.628955910148817e-05,
      "loss": 1.8645,
      "step": 199471
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.4108511209487915,
      "learning_rate": 2.6287884570563544e-05,
      "loss": 1.8467,
      "step": 199472
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.4408462047576904,
      "learning_rate": 2.628621009052725e-05,
      "loss": 2.306,
      "step": 199473
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.4913053512573242,
      "learning_rate": 2.6284535661379447e-05,
      "loss": 2.1032,
      "step": 199474
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.5437740087509155,
      "learning_rate": 2.628286128312057e-05,
      "loss": 1.9583,
      "step": 199475
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.5077931880950928,
      "learning_rate": 2.6281186955750887e-05,
      "loss": 2.0235,
      "step": 199476
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.3686012029647827,
      "learning_rate": 2.6279512679270663e-05,
      "loss": 2.1102,
      "step": 199477
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.4770416021347046,
      "learning_rate": 2.6277838453680233e-05,
      "loss": 2.0937,
      "step": 199478
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.4862205982208252,
      "learning_rate": 2.627616427897996e-05,
      "loss": 1.967,
      "step": 199479
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.4296717643737793,
      "learning_rate": 2.6274490155170046e-05,
      "loss": 2.0051,
      "step": 199480
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.3165456056594849,
      "learning_rate": 2.627281608225096e-05,
      "loss": 2.0306,
      "step": 199481
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.2768768072128296,
      "learning_rate": 2.6271142060222895e-05,
      "loss": 1.9235,
      "step": 199482
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.3838655948638916,
      "learning_rate": 2.6269468089086188e-05,
      "loss": 2.1235,
      "step": 199483
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.5173981189727783,
      "learning_rate": 2.6267794168841107e-05,
      "loss": 2.0169,
      "step": 199484
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.3780949115753174,
      "learning_rate": 2.626612029948808e-05,
      "loss": 1.9035,
      "step": 199485
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.4574490785598755,
      "learning_rate": 2.6264446481027278e-05,
      "loss": 2.0245,
      "step": 199486
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.3837412595748901,
      "learning_rate": 2.6262772713459133e-05,
      "loss": 1.9187,
      "step": 199487
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.394294023513794,
      "learning_rate": 2.6261098996783913e-05,
      "loss": 1.9652,
      "step": 199488
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.414096474647522,
      "learning_rate": 2.625942533100188e-05,
      "loss": 1.8515,
      "step": 199489
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.497340202331543,
      "learning_rate": 2.6257751716113438e-05,
      "loss": 2.0284,
      "step": 199490
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.4783388376235962,
      "learning_rate": 2.625607815211882e-05,
      "loss": 1.9498,
      "step": 199491
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.4709057807922363,
      "learning_rate": 2.6254404639018323e-05,
      "loss": 2.0906,
      "step": 199492
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.4946229457855225,
      "learning_rate": 2.625273117681238e-05,
      "loss": 1.8989,
      "step": 199493
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.4428982734680176,
      "learning_rate": 2.6251057765501194e-05,
      "loss": 1.8334,
      "step": 199494
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.363692045211792,
      "learning_rate": 2.624938440508506e-05,
      "loss": 2.2022,
      "step": 199495
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.5006331205368042,
      "learning_rate": 2.6247711095564384e-05,
      "loss": 1.988,
      "step": 199496
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.4378918409347534,
      "learning_rate": 2.624603783693936e-05,
      "loss": 1.9271,
      "step": 199497
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.4447938203811646,
      "learning_rate": 2.6244364629210423e-05,
      "loss": 2.0211,
      "step": 199498
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.472800850868225,
      "learning_rate": 2.624269147237784e-05,
      "loss": 1.9851,
      "step": 199499
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.490706443786621,
      "learning_rate": 2.6241018366441914e-05,
      "loss": 1.9599,
      "step": 199500
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.5322856903076172,
      "learning_rate": 2.6239345311402872e-05,
      "loss": 2.0894,
      "step": 199501
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.5300023555755615,
      "learning_rate": 2.6237672307261183e-05,
      "loss": 2.0634,
      "step": 199502
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.3251047134399414,
      "learning_rate": 2.6235999354017012e-05,
      "loss": 1.8796,
      "step": 199503
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.4532737731933594,
      "learning_rate": 2.6234326451670797e-05,
      "loss": 1.9615,
      "step": 199504
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.4521645307540894,
      "learning_rate": 2.6232653600222764e-05,
      "loss": 2.0498,
      "step": 199505
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.6678717136383057,
      "learning_rate": 2.6230980799673284e-05,
      "loss": 1.8922,
      "step": 199506
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.3564563989639282,
      "learning_rate": 2.6229308050022556e-05,
      "loss": 1.9891,
      "step": 199507
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.5654574632644653,
      "learning_rate": 2.6227635351271016e-05,
      "loss": 2.1354,
      "step": 199508
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.3507695198059082,
      "learning_rate": 2.622596270341889e-05,
      "loss": 2.0093,
      "step": 199509
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.5818508863449097,
      "learning_rate": 2.6224290106466585e-05,
      "loss": 1.985,
      "step": 199510
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.5259672403335571,
      "learning_rate": 2.6222617560414262e-05,
      "loss": 2.1224,
      "step": 199511
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.4594396352767944,
      "learning_rate": 2.6220945065262457e-05,
      "loss": 2.0109,
      "step": 199512
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.630530595779419,
      "learning_rate": 2.6219272621011233e-05,
      "loss": 1.9821,
      "step": 199513
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.4879292249679565,
      "learning_rate": 2.621760022766106e-05,
      "loss": 1.8947,
      "step": 199514
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.3894195556640625,
      "learning_rate": 2.6215927885212173e-05,
      "loss": 1.8322,
      "step": 199515
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.4384100437164307,
      "learning_rate": 2.6214255593664967e-05,
      "loss": 1.9858,
      "step": 199516
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.4993304014205933,
      "learning_rate": 2.6212583353019613e-05,
      "loss": 1.9628,
      "step": 199517
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.531517505645752,
      "learning_rate": 2.621091116327664e-05,
      "loss": 1.8234,
      "step": 199518
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.44345223903656,
      "learning_rate": 2.6209239024436114e-05,
      "loss": 1.9265,
      "step": 199519
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.3610632419586182,
      "learning_rate": 2.6207566936498502e-05,
      "loss": 1.8946,
      "step": 199520
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.4918447732925415,
      "learning_rate": 2.6205894899464042e-05,
      "loss": 2.0336,
      "step": 199521
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.4256842136383057,
      "learning_rate": 2.6204222913333095e-05,
      "loss": 2.1797,
      "step": 199522
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.4067869186401367,
      "learning_rate": 2.620255097810593e-05,
      "loss": 2.058,
      "step": 199523
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.497025966644287,
      "learning_rate": 2.6200879093782977e-05,
      "loss": 2.0294,
      "step": 199524
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.3755946159362793,
      "learning_rate": 2.6199207260364342e-05,
      "loss": 2.1042,
      "step": 199525
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.5419243574142456,
      "learning_rate": 2.6197535477850483e-05,
      "loss": 2.0022,
      "step": 199526
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.5811079740524292,
      "learning_rate": 2.6195863746241607e-05,
      "loss": 1.8242,
      "step": 199527
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.779554843902588,
      "learning_rate": 2.6194192065538177e-05,
      "loss": 1.9404,
      "step": 199528
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.4519083499908447,
      "learning_rate": 2.6192520435740326e-05,
      "loss": 2.0134,
      "step": 199529
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.5333505868911743,
      "learning_rate": 2.619084885684859e-05,
      "loss": 2.1113,
      "step": 199530
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.345096230506897,
      "learning_rate": 2.6189177328863032e-05,
      "loss": 1.9758,
      "step": 199531
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.53742253780365,
      "learning_rate": 2.6187505851784118e-05,
      "loss": 2.0376,
      "step": 199532
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.5092341899871826,
      "learning_rate": 2.6185834425612052e-05,
      "loss": 1.958,
      "step": 199533
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.3936619758605957,
      "learning_rate": 2.6184163050347294e-05,
      "loss": 1.8223,
      "step": 199534
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.7162535190582275,
      "learning_rate": 2.6182491725989985e-05,
      "loss": 1.8739,
      "step": 199535
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.3411686420440674,
      "learning_rate": 2.618082045254065e-05,
      "loss": 1.8288,
      "step": 199536
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.422807216644287,
      "learning_rate": 2.6179149229999365e-05,
      "loss": 1.8362,
      "step": 199537
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.6158186197280884,
      "learning_rate": 2.617747805836655e-05,
      "loss": 2.1387,
      "step": 199538
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.388951301574707,
      "learning_rate": 2.617580693764252e-05,
      "loss": 1.8457,
      "step": 199539
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.3193327188491821,
      "learning_rate": 2.6174135867827596e-05,
      "loss": 2.0444,
      "step": 199540
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.4837992191314697,
      "learning_rate": 2.6172464848922016e-05,
      "loss": 2.1273,
      "step": 199541
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.56955087184906,
      "learning_rate": 2.6170793880926243e-05,
      "loss": 1.9668,
      "step": 199542
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.6447398662567139,
      "learning_rate": 2.6169122963840417e-05,
      "loss": 2.0633,
      "step": 199543
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.7526527643203735,
      "learning_rate": 2.616745209766493e-05,
      "loss": 1.8969,
      "step": 199544
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.5890285968780518,
      "learning_rate": 2.6165781282400057e-05,
      "loss": 1.7913,
      "step": 199545
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.7092134952545166,
      "learning_rate": 2.6164110518046187e-05,
      "loss": 1.9359,
      "step": 199546
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.6198731660842896,
      "learning_rate": 2.6162439804603497e-05,
      "loss": 1.9241,
      "step": 199547
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.8360111713409424,
      "learning_rate": 2.616076914207251e-05,
      "loss": 1.828,
      "step": 199548
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.6719194650650024,
      "learning_rate": 2.6159098530453305e-05,
      "loss": 2.0791,
      "step": 199549
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.3692870140075684,
      "learning_rate": 2.6157427969746336e-05,
      "loss": 1.9573,
      "step": 199550
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.5167858600616455,
      "learning_rate": 2.615575745995181e-05,
      "loss": 1.923,
      "step": 199551
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.5735046863555908,
      "learning_rate": 2.6154087001070157e-05,
      "loss": 2.1004,
      "step": 199552
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.384894609451294,
      "learning_rate": 2.6152416593101578e-05,
      "loss": 1.9996,
      "step": 199553
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.4352436065673828,
      "learning_rate": 2.6150746236046537e-05,
      "loss": 1.9324,
      "step": 199554
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.578501582145691,
      "learning_rate": 2.6149075929905138e-05,
      "loss": 1.9979,
      "step": 199555
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.4653162956237793,
      "learning_rate": 2.6147405674677846e-05,
      "loss": 1.9316,
      "step": 199556
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.4250929355621338,
      "learning_rate": 2.6145735470364858e-05,
      "loss": 2.0651,
      "step": 199557
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.6206284761428833,
      "learning_rate": 2.614406531696661e-05,
      "loss": 1.9131,
      "step": 199558
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.8118048906326294,
      "learning_rate": 2.6142395214483303e-05,
      "loss": 2.0032,
      "step": 199559
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.5448249578475952,
      "learning_rate": 2.6140725162915332e-05,
      "loss": 1.91,
      "step": 199560
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.3697383403778076,
      "learning_rate": 2.6139055162263e-05,
      "loss": 1.9978,
      "step": 199561
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.4366647005081177,
      "learning_rate": 2.6137385212526574e-05,
      "loss": 2.0866,
      "step": 199562
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.486009955406189,
      "learning_rate": 2.6135715313706318e-05,
      "loss": 1.9057,
      "step": 199563
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.4807549715042114,
      "learning_rate": 2.6134045465802632e-05,
      "loss": 2.0324,
      "step": 199564
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.4689500331878662,
      "learning_rate": 2.6132375668815784e-05,
      "loss": 2.0927,
      "step": 199565
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.5883735418319702,
      "learning_rate": 2.613070592274611e-05,
      "loss": 1.8561,
      "step": 199566
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.356512427330017,
      "learning_rate": 2.6129036227593936e-05,
      "loss": 1.999,
      "step": 199567
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.4583817720413208,
      "learning_rate": 2.6127366583359533e-05,
      "loss": 2.0636,
      "step": 199568
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.4755003452301025,
      "learning_rate": 2.61256969900432e-05,
      "loss": 2.1707,
      "step": 199569
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.3291339874267578,
      "learning_rate": 2.6124027447645267e-05,
      "loss": 1.8034,
      "step": 199570
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.3670963048934937,
      "learning_rate": 2.612235795616604e-05,
      "loss": 1.955,
      "step": 199571
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.8995106220245361,
      "learning_rate": 2.612068851560588e-05,
      "loss": 2.011,
      "step": 199572
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.4851895570755005,
      "learning_rate": 2.6119019125965025e-05,
      "loss": 2.2502,
      "step": 199573
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.3381246328353882,
      "learning_rate": 2.61173497872438e-05,
      "loss": 2.0811,
      "step": 199574
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.4464181661605835,
      "learning_rate": 2.611568049944255e-05,
      "loss": 2.0093,
      "step": 199575
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.3925570249557495,
      "learning_rate": 2.6114011262561597e-05,
      "loss": 1.8778,
      "step": 199576
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.227239727973938,
      "learning_rate": 2.6112342076601146e-05,
      "loss": 1.9582,
      "step": 199577
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.3834704160690308,
      "learning_rate": 2.611067294156163e-05,
      "loss": 1.7986,
      "step": 199578
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.286676049232483,
      "learning_rate": 2.6109003857443313e-05,
      "loss": 1.9319,
      "step": 199579
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.4827128648757935,
      "learning_rate": 2.6107334824246463e-05,
      "loss": 2.1881,
      "step": 199580
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.4148468971252441,
      "learning_rate": 2.610566584197148e-05,
      "loss": 1.9803,
      "step": 199581
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.4841631650924683,
      "learning_rate": 2.6103996910618598e-05,
      "loss": 2.1165,
      "step": 199582
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.4265074729919434,
      "learning_rate": 2.6102328030188114e-05,
      "loss": 1.8707,
      "step": 199583
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.5429481267929077,
      "learning_rate": 2.6100659200680462e-05,
      "loss": 2.2885,
      "step": 199584
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.3294461965560913,
      "learning_rate": 2.609899042209581e-05,
      "loss": 1.9686,
      "step": 199585
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.4534205198287964,
      "learning_rate": 2.6097321694434526e-05,
      "loss": 2.0509,
      "step": 199586
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.3612594604492188,
      "learning_rate": 2.6095653017696937e-05,
      "loss": 1.7603,
      "step": 199587
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.5170210599899292,
      "learning_rate": 2.6093984391883317e-05,
      "loss": 2.0193,
      "step": 199588
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.5007307529449463,
      "learning_rate": 2.6092315816994023e-05,
      "loss": 1.8931,
      "step": 199589
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.5245234966278076,
      "learning_rate": 2.6090647293029333e-05,
      "loss": 1.9249,
      "step": 199590
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.4868316650390625,
      "learning_rate": 2.608897881998957e-05,
      "loss": 1.9551,
      "step": 199591
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.4901070594787598,
      "learning_rate": 2.6087310397875005e-05,
      "loss": 1.9433,
      "step": 199592
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.4515907764434814,
      "learning_rate": 2.6085642026686005e-05,
      "loss": 2.0141,
      "step": 199593
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.693220853805542,
      "learning_rate": 2.60839737064228e-05,
      "loss": 2.2088,
      "step": 199594
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.375586986541748,
      "learning_rate": 2.6082305437085826e-05,
      "loss": 1.8135,
      "step": 199595
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.3096095323562622,
      "learning_rate": 2.608063721867525e-05,
      "loss": 1.9726,
      "step": 199596
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.5336729288101196,
      "learning_rate": 2.60789690511916e-05,
      "loss": 2.1108,
      "step": 199597
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.4218599796295166,
      "learning_rate": 2.6077300934634882e-05,
      "loss": 1.986,
      "step": 199598
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.5310710668563843,
      "learning_rate": 2.607563286900566e-05,
      "loss": 1.7887,
      "step": 199599
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.4067851305007935,
      "learning_rate": 2.6073964854304064e-05,
      "loss": 2.0679,
      "step": 199600
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.519134283065796,
      "learning_rate": 2.6072296890530563e-05,
      "loss": 1.9004,
      "step": 199601
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.435396671295166,
      "learning_rate": 2.6070628977685327e-05,
      "loss": 1.9012,
      "step": 199602
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.4288825988769531,
      "learning_rate": 2.606896111576885e-05,
      "loss": 1.915,
      "step": 199603
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.3222261667251587,
      "learning_rate": 2.6067293304781202e-05,
      "loss": 2.06,
      "step": 199604
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.5015733242034912,
      "learning_rate": 2.606562554472288e-05,
      "loss": 1.8761,
      "step": 199605
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.48459792137146,
      "learning_rate": 2.6063957835594085e-05,
      "loss": 1.9828,
      "step": 199606
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.3744934797286987,
      "learning_rate": 2.6062290177395217e-05,
      "loss": 2.0785,
      "step": 199607
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.394729733467102,
      "learning_rate": 2.606062257012648e-05,
      "loss": 1.9168,
      "step": 199608
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.435825228691101,
      "learning_rate": 2.605895501378833e-05,
      "loss": 1.9608,
      "step": 199609
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.419021725654602,
      "learning_rate": 2.6057287508380908e-05,
      "loss": 1.8569,
      "step": 199610
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.4481981992721558,
      "learning_rate": 2.605562005390468e-05,
      "loss": 1.88,
      "step": 199611
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.4854536056518555,
      "learning_rate": 2.6053952650359812e-05,
      "loss": 1.8961,
      "step": 199612
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.4419240951538086,
      "learning_rate": 2.6052285297746735e-05,
      "loss": 1.9638,
      "step": 199613
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.4266210794448853,
      "learning_rate": 2.6050617996065647e-05,
      "loss": 1.9198,
      "step": 199614
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.3309720754623413,
      "learning_rate": 2.6048950745317054e-05,
      "loss": 2.0601,
      "step": 199615
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.436508059501648,
      "learning_rate": 2.6047283545500986e-05,
      "loss": 1.9415,
      "step": 199616
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.495848298072815,
      "learning_rate": 2.6045616396617974e-05,
      "loss": 2.0397,
      "step": 199617
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.4401260614395142,
      "learning_rate": 2.6043949298668222e-05,
      "loss": 1.9595,
      "step": 199618
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.5655595064163208,
      "learning_rate": 2.604228225165209e-05,
      "loss": 2.283,
      "step": 199619
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.2943600416183472,
      "learning_rate": 2.604061525556982e-05,
      "loss": 1.7972,
      "step": 199620
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.4163440465927124,
      "learning_rate": 2.6038948310421902e-05,
      "loss": 1.7801,
      "step": 199621
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.4665746688842773,
      "learning_rate": 2.6037281416208413e-05,
      "loss": 2.0214,
      "step": 199622
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.4725905656814575,
      "learning_rate": 2.6035614572929776e-05,
      "loss": 1.8614,
      "step": 199623
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.5396674871444702,
      "learning_rate": 2.6033947780586262e-05,
      "loss": 2.0747,
      "step": 199624
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.4620431661605835,
      "learning_rate": 2.6032281039178272e-05,
      "loss": 2.0381,
      "step": 199625
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.3873810768127441,
      "learning_rate": 2.603061434870597e-05,
      "loss": 2.1609,
      "step": 199626
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.601579189300537,
      "learning_rate": 2.6028947709169888e-05,
      "loss": 1.8478,
      "step": 199627
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.313772439956665,
      "learning_rate": 2.6027281120570063e-05,
      "loss": 2.1573,
      "step": 199628
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.391669750213623,
      "learning_rate": 2.6025614582906995e-05,
      "loss": 1.9227,
      "step": 199629
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.5592056512832642,
      "learning_rate": 2.6023948096180914e-05,
      "loss": 2.0819,
      "step": 199630
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.4671709537506104,
      "learning_rate": 2.6022281660392152e-05,
      "loss": 1.8494,
      "step": 199631
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.4276974201202393,
      "learning_rate": 2.602061527554101e-05,
      "loss": 1.9919,
      "step": 199632
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.3034497499465942,
      "learning_rate": 2.6018948941627893e-05,
      "loss": 1.8882,
      "step": 199633
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.428539752960205,
      "learning_rate": 2.6017282658652928e-05,
      "loss": 1.8537,
      "step": 199634
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.3345530033111572,
      "learning_rate": 2.601561642661658e-05,
      "loss": 1.8087,
      "step": 199635
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.5528857707977295,
      "learning_rate": 2.6013950245519054e-05,
      "loss": 1.8844,
      "step": 199636
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.4875797033309937,
      "learning_rate": 2.601228411536075e-05,
      "loss": 1.9391,
      "step": 199637
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.4210704565048218,
      "learning_rate": 2.6010618036141863e-05,
      "loss": 1.981,
      "step": 199638
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.427587628364563,
      "learning_rate": 2.6008952007862895e-05,
      "loss": 1.8609,
      "step": 199639
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.4459608793258667,
      "learning_rate": 2.6007286030523943e-05,
      "loss": 2.1212,
      "step": 199640
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.4486693143844604,
      "learning_rate": 2.6005620104125447e-05,
      "loss": 1.9729,
      "step": 199641
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.464506983757019,
      "learning_rate": 2.60039542286676e-05,
      "loss": 1.9554,
      "step": 199642
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.3835350275039673,
      "learning_rate": 2.6002288404150872e-05,
      "loss": 2.071,
      "step": 199643
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.4676530361175537,
      "learning_rate": 2.600062263057543e-05,
      "loss": 2.0448,
      "step": 199644
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.315413236618042,
      "learning_rate": 2.5998956907941703e-05,
      "loss": 2.0849,
      "step": 199645
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.4914995431900024,
      "learning_rate": 2.5997291236249962e-05,
      "loss": 1.8055,
      "step": 199646
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.368335485458374,
      "learning_rate": 2.599562561550047e-05,
      "loss": 1.9354,
      "step": 199647
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.4519526958465576,
      "learning_rate": 2.5993960045693497e-05,
      "loss": 1.9183,
      "step": 199648
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.4927012920379639,
      "learning_rate": 2.5992294526829504e-05,
      "loss": 1.9614,
      "step": 199649
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.54438316822052,
      "learning_rate": 2.5990629058908662e-05,
      "loss": 2.3475,
      "step": 199650
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.4107528924942017,
      "learning_rate": 2.598896364193137e-05,
      "loss": 1.8792,
      "step": 199651
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.506578803062439,
      "learning_rate": 2.5987298275897895e-05,
      "loss": 1.8516,
      "step": 199652
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.3378181457519531,
      "learning_rate": 2.5985632960808566e-05,
      "loss": 2.0518,
      "step": 199653
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.3586907386779785,
      "learning_rate": 2.598396769666362e-05,
      "loss": 1.9748,
      "step": 199654
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.4540951251983643,
      "learning_rate": 2.598230248346349e-05,
      "loss": 2.3245,
      "step": 199655
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.3526484966278076,
      "learning_rate": 2.5980637321208376e-05,
      "loss": 2.0216,
      "step": 199656
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.4243886470794678,
      "learning_rate": 2.5978972209898674e-05,
      "loss": 1.9905,
      "step": 199657
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.3578990697860718,
      "learning_rate": 2.5977307149534653e-05,
      "loss": 1.9153,
      "step": 199658
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.4957818984985352,
      "learning_rate": 2.597564214011658e-05,
      "loss": 1.9822,
      "step": 199659
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.4715620279312134,
      "learning_rate": 2.5973977181644856e-05,
      "loss": 2.1644,
      "step": 199660
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.5050419569015503,
      "learning_rate": 2.5972312274119745e-05,
      "loss": 2.1152,
      "step": 199661
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.377493143081665,
      "learning_rate": 2.597064741754148e-05,
      "loss": 1.9574,
      "step": 199662
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.4358843564987183,
      "learning_rate": 2.596898261191053e-05,
      "loss": 1.9917,
      "step": 199663
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.3811028003692627,
      "learning_rate": 2.5967317857227088e-05,
      "loss": 1.9684,
      "step": 199664
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.3908848762512207,
      "learning_rate": 2.5965653153491463e-05,
      "loss": 1.9789,
      "step": 199665
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.4549806118011475,
      "learning_rate": 2.5963988500704046e-05,
      "loss": 1.9107,
      "step": 199666
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.3503303527832031,
      "learning_rate": 2.5962323898865113e-05,
      "loss": 2.079,
      "step": 199667
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.3210266828536987,
      "learning_rate": 2.5960659347974855e-05,
      "loss": 1.873,
      "step": 199668
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.3958463668823242,
      "learning_rate": 2.5958994848033775e-05,
      "loss": 1.8105,
      "step": 199669
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.36248779296875,
      "learning_rate": 2.5957330399042075e-05,
      "loss": 1.9524,
      "step": 199670
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.4896162748336792,
      "learning_rate": 2.5955666001000054e-05,
      "loss": 2.0134,
      "step": 199671
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.746734380722046,
      "learning_rate": 2.5954001653908075e-05,
      "loss": 2.0841,
      "step": 199672
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.3833973407745361,
      "learning_rate": 2.595233735776637e-05,
      "loss": 1.961,
      "step": 199673
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.3494364023208618,
      "learning_rate": 2.595067311257535e-05,
      "loss": 1.8952,
      "step": 199674
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.4631963968276978,
      "learning_rate": 2.5949008918335266e-05,
      "loss": 2.0986,
      "step": 199675
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.4145523309707642,
      "learning_rate": 2.5947344775046465e-05,
      "loss": 2.0344,
      "step": 199676
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.3962000608444214,
      "learning_rate": 2.5945680682709135e-05,
      "loss": 2.0229,
      "step": 199677
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.4694280624389648,
      "learning_rate": 2.594401664132375e-05,
      "loss": 2.085,
      "step": 199678
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.3622370958328247,
      "learning_rate": 2.5942352650890508e-05,
      "loss": 2.0157,
      "step": 199679
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.4646215438842773,
      "learning_rate": 2.5940688711409773e-05,
      "loss": 1.9211,
      "step": 199680
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.785569429397583,
      "learning_rate": 2.5939024822881814e-05,
      "loss": 2.1091,
      "step": 199681
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.4241631031036377,
      "learning_rate": 2.593736098530703e-05,
      "loss": 1.9902,
      "step": 199682
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.3832629919052124,
      "learning_rate": 2.593569719868559e-05,
      "loss": 1.9624,
      "step": 199683
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.4758485555648804,
      "learning_rate": 2.5934033463017923e-05,
      "loss": 2.121,
      "step": 199684
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.5252201557159424,
      "learning_rate": 2.593236977830426e-05,
      "loss": 2.0582,
      "step": 199685
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.4239989519119263,
      "learning_rate": 2.593070614454498e-05,
      "loss": 2.034,
      "step": 199686
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.5153576135635376,
      "learning_rate": 2.5929042561740297e-05,
      "loss": 2.0737,
      "step": 199687
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.4285558462142944,
      "learning_rate": 2.5927379029890693e-05,
      "loss": 2.0067,
      "step": 199688
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.4525668621063232,
      "learning_rate": 2.592571554899623e-05,
      "loss": 1.8636,
      "step": 199689
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.3463783264160156,
      "learning_rate": 2.5924052119057437e-05,
      "loss": 1.934,
      "step": 199690
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.2945646047592163,
      "learning_rate": 2.592238874007445e-05,
      "loss": 1.8177,
      "step": 199691
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.3813337087631226,
      "learning_rate": 2.5920725412047772e-05,
      "loss": 1.8224,
      "step": 199692
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.408584475517273,
      "learning_rate": 2.59190621349775e-05,
      "loss": 1.8488,
      "step": 199693
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.5072672367095947,
      "learning_rate": 2.5917398908864162e-05,
      "loss": 1.9348,
      "step": 199694
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.3614767789840698,
      "learning_rate": 2.5915735733707866e-05,
      "loss": 2.0783,
      "step": 199695
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.3845239877700806,
      "learning_rate": 2.591407260950904e-05,
      "loss": 1.842,
      "step": 199696
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.7061423063278198,
      "learning_rate": 2.591240953626792e-05,
      "loss": 1.8059,
      "step": 199697
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.3794374465942383,
      "learning_rate": 2.5910746513984937e-05,
      "loss": 2.1331,
      "step": 199698
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.328370451927185,
      "learning_rate": 2.5909083542660225e-05,
      "loss": 1.9384,
      "step": 199699
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.37856924533844,
      "learning_rate": 2.5907420622294317e-05,
      "loss": 2.0273,
      "step": 199700
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.3312358856201172,
      "learning_rate": 2.590575775288728e-05,
      "loss": 1.8044,
      "step": 199701
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.777420163154602,
      "learning_rate": 2.590409493443958e-05,
      "loss": 1.9352,
      "step": 199702
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.335586667060852,
      "learning_rate": 2.5902432166951446e-05,
      "loss": 2.1001,
      "step": 199703
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.5423811674118042,
      "learning_rate": 2.5900769450423288e-05,
      "loss": 1.9386,
      "step": 199704
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.4232674837112427,
      "learning_rate": 2.589910678485526e-05,
      "loss": 1.9067,
      "step": 199705
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.3956013917922974,
      "learning_rate": 2.5897444170247904e-05,
      "loss": 1.8334,
      "step": 199706
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.5428433418273926,
      "learning_rate": 2.589578160660125e-05,
      "loss": 2.0159,
      "step": 199707
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.4626426696777344,
      "learning_rate": 2.5894119093915834e-05,
      "loss": 1.8085,
      "step": 199708
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.4442459344863892,
      "learning_rate": 2.589245663219178e-05,
      "loss": 1.8871,
      "step": 199709
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.4548323154449463,
      "learning_rate": 2.5890794221429567e-05,
      "loss": 1.9541,
      "step": 199710
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.4293670654296875,
      "learning_rate": 2.5889131861629386e-05,
      "loss": 2.0735,
      "step": 199711
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.4749393463134766,
      "learning_rate": 2.588746955279164e-05,
      "loss": 2.1617,
      "step": 199712
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.5878938436508179,
      "learning_rate": 2.588580729491656e-05,
      "loss": 2.1003,
      "step": 199713
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.4563883543014526,
      "learning_rate": 2.5884145088004516e-05,
      "loss": 2.0333,
      "step": 199714
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.4422756433486938,
      "learning_rate": 2.5882482932055705e-05,
      "loss": 1.9754,
      "step": 199715
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.4305733442306519,
      "learning_rate": 2.5880820827070592e-05,
      "loss": 1.9132,
      "step": 199716
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.4738693237304688,
      "learning_rate": 2.5879158773049345e-05,
      "loss": 2.0403,
      "step": 199717
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.3425185680389404,
      "learning_rate": 2.58774967699924e-05,
      "loss": 2.0636,
      "step": 199718
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.8942335844039917,
      "learning_rate": 2.5875834817899988e-05,
      "loss": 1.8524,
      "step": 199719
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.416269302368164,
      "learning_rate": 2.587417291677244e-05,
      "loss": 1.757,
      "step": 199720
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.3972227573394775,
      "learning_rate": 2.587251106660999e-05,
      "loss": 1.9415,
      "step": 199721
    },
    {
      "epoch": 2.6,
      "grad_norm": 2.2333827018737793,
      "learning_rate": 2.5870849267413076e-05,
      "loss": 1.9314,
      "step": 199722
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.4702337980270386,
      "learning_rate": 2.586918751918192e-05,
      "loss": 2.1226,
      "step": 199723
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.461306095123291,
      "learning_rate": 2.586752582191687e-05,
      "loss": 2.2043,
      "step": 199724
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.3958783149719238,
      "learning_rate": 2.5865864175618245e-05,
      "loss": 2.0045,
      "step": 199725
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.2964062690734863,
      "learning_rate": 2.586420258028632e-05,
      "loss": 2.0138,
      "step": 199726
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.4139493703842163,
      "learning_rate": 2.5862541035921357e-05,
      "loss": 1.9531,
      "step": 199727
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.5139731168746948,
      "learning_rate": 2.5860879542523794e-05,
      "loss": 1.9689,
      "step": 199728
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.331623911857605,
      "learning_rate": 2.5859218100093792e-05,
      "loss": 1.9376,
      "step": 199729
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.5098377466201782,
      "learning_rate": 2.585755670863182e-05,
      "loss": 2.1589,
      "step": 199730
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.2891855239868164,
      "learning_rate": 2.585589536813808e-05,
      "loss": 1.6864,
      "step": 199731
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.3739663362503052,
      "learning_rate": 2.58542340786129e-05,
      "loss": 2.0098,
      "step": 199732
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.5046112537384033,
      "learning_rate": 2.585257284005655e-05,
      "loss": 2.0871,
      "step": 199733
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.389134407043457,
      "learning_rate": 2.5850911652469432e-05,
      "loss": 1.8781,
      "step": 199734
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.3776640892028809,
      "learning_rate": 2.5849250515851772e-05,
      "loss": 2.1514,
      "step": 199735
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.4891551733016968,
      "learning_rate": 2.584758943020394e-05,
      "loss": 2.0391,
      "step": 199736
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.5832200050354004,
      "learning_rate": 2.5845928395526205e-05,
      "loss": 1.9719,
      "step": 199737
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.5428780317306519,
      "learning_rate": 2.584426741181893e-05,
      "loss": 2.0345,
      "step": 199738
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.443140983581543,
      "learning_rate": 2.5842606479082284e-05,
      "loss": 1.871,
      "step": 199739
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.4126012325286865,
      "learning_rate": 2.584094559731673e-05,
      "loss": 1.9751,
      "step": 199740
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.410902500152588,
      "learning_rate": 2.5839284766522504e-05,
      "loss": 1.9538,
      "step": 199741
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.394119143486023,
      "learning_rate": 2.583762398669994e-05,
      "loss": 1.8055,
      "step": 199742
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.4247499704360962,
      "learning_rate": 2.583596325784937e-05,
      "loss": 1.9286,
      "step": 199743
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.4489270448684692,
      "learning_rate": 2.583430257997099e-05,
      "loss": 2.1525,
      "step": 199744
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.400306224822998,
      "learning_rate": 2.583264195306527e-05,
      "loss": 1.9292,
      "step": 199745
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.6166164875030518,
      "learning_rate": 2.583098137713241e-05,
      "loss": 2.0454,
      "step": 199746
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.561668038368225,
      "learning_rate": 2.5829320852172707e-05,
      "loss": 1.9126,
      "step": 199747
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.5854220390319824,
      "learning_rate": 2.5827660378186564e-05,
      "loss": 1.8327,
      "step": 199748
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.4253895282745361,
      "learning_rate": 2.5825999955174213e-05,
      "loss": 1.8821,
      "step": 199749
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.5298405885696411,
      "learning_rate": 2.5824339583135956e-05,
      "loss": 2.4143,
      "step": 199750
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.4298999309539795,
      "learning_rate": 2.5822679262072155e-05,
      "loss": 1.9126,
      "step": 199751
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.5827065706253052,
      "learning_rate": 2.5821018991983112e-05,
      "loss": 1.9225,
      "step": 199752
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.422206997871399,
      "learning_rate": 2.5819358772869057e-05,
      "loss": 1.813,
      "step": 199753
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.3492554426193237,
      "learning_rate": 2.581769860473043e-05,
      "loss": 1.907,
      "step": 199754
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.4270849227905273,
      "learning_rate": 2.5816038487567457e-05,
      "loss": 1.896,
      "step": 199755
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.4059813022613525,
      "learning_rate": 2.581437842138041e-05,
      "loss": 2.1714,
      "step": 199756
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.466637372970581,
      "learning_rate": 2.5812718406169686e-05,
      "loss": 2.0797,
      "step": 199757
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.4052646160125732,
      "learning_rate": 2.5811058441935516e-05,
      "loss": 1.9202,
      "step": 199758
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.749928593635559,
      "learning_rate": 2.5809398528678306e-05,
      "loss": 1.8853,
      "step": 199759
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.439946174621582,
      "learning_rate": 2.580773866639828e-05,
      "loss": 1.8327,
      "step": 199760
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.434889554977417,
      "learning_rate": 2.580607885509578e-05,
      "loss": 2.0659,
      "step": 199761
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.323991060256958,
      "learning_rate": 2.5804419094771035e-05,
      "loss": 1.8584,
      "step": 199762
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.2593756914138794,
      "learning_rate": 2.580275938542451e-05,
      "loss": 1.6296,
      "step": 199763
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.393810749053955,
      "learning_rate": 2.580109972705634e-05,
      "loss": 1.9817,
      "step": 199764
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.4846378564834595,
      "learning_rate": 2.5799440119667024e-05,
      "loss": 1.8316,
      "step": 199765
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.4411286115646362,
      "learning_rate": 2.579778056325673e-05,
      "loss": 1.7533,
      "step": 199766
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.3291670083999634,
      "learning_rate": 2.5796121057825824e-05,
      "loss": 2.1727,
      "step": 199767
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.4304465055465698,
      "learning_rate": 2.5794461603374506e-05,
      "loss": 2.0039,
      "step": 199768
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.4447306394577026,
      "learning_rate": 2.5792802199903272e-05,
      "loss": 1.8812,
      "step": 199769
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.4877737760543823,
      "learning_rate": 2.579114284741226e-05,
      "loss": 1.9202,
      "step": 199770
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.3549623489379883,
      "learning_rate": 2.57894835459019e-05,
      "loss": 1.9083,
      "step": 199771
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.597766637802124,
      "learning_rate": 2.5787824295372396e-05,
      "loss": 1.9041,
      "step": 199772
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.3557522296905518,
      "learning_rate": 2.578616509582421e-05,
      "loss": 1.9976,
      "step": 199773
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.3220975399017334,
      "learning_rate": 2.578450594725747e-05,
      "loss": 1.8547,
      "step": 199774
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.3739813566207886,
      "learning_rate": 2.5782846849672623e-05,
      "loss": 1.9209,
      "step": 199775
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.5687686204910278,
      "learning_rate": 2.5781187803069824e-05,
      "loss": 1.9593,
      "step": 199776
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.3798353672027588,
      "learning_rate": 2.577952880744958e-05,
      "loss": 2.0825,
      "step": 199777
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.513669490814209,
      "learning_rate": 2.5777869862812017e-05,
      "loss": 1.9169,
      "step": 199778
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.387170672416687,
      "learning_rate": 2.577621096915761e-05,
      "loss": 1.9918,
      "step": 199779
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.4756062030792236,
      "learning_rate": 2.5774552126486547e-05,
      "loss": 2.0077,
      "step": 199780
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.5828498601913452,
      "learning_rate": 2.577289333479917e-05,
      "loss": 1.9699,
      "step": 199781
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.359780192375183,
      "learning_rate": 2.5771234594095748e-05,
      "loss": 2.0451,
      "step": 199782
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.4469561576843262,
      "learning_rate": 2.576957590437667e-05,
      "loss": 1.9234,
      "step": 199783
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.345574140548706,
      "learning_rate": 2.576791726564218e-05,
      "loss": 1.8811,
      "step": 199784
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.6231646537780762,
      "learning_rate": 2.5766258677892637e-05,
      "loss": 2.0571,
      "step": 199785
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.4026622772216797,
      "learning_rate": 2.5764600141128344e-05,
      "loss": 2.0208,
      "step": 199786
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.2960458993911743,
      "learning_rate": 2.5762941655349567e-05,
      "loss": 2.1184,
      "step": 199787
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.3624281883239746,
      "learning_rate": 2.5761283220556606e-05,
      "loss": 2.0028,
      "step": 199788
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.3610007762908936,
      "learning_rate": 2.5759624836749827e-05,
      "loss": 1.7023,
      "step": 199789
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.4849085807800293,
      "learning_rate": 2.5757966503929496e-05,
      "loss": 1.9322,
      "step": 199790
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.4203870296478271,
      "learning_rate": 2.5756308222095947e-05,
      "loss": 1.8909,
      "step": 199791
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.4775373935699463,
      "learning_rate": 2.575464999124951e-05,
      "loss": 2.1383,
      "step": 199792
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.4586403369903564,
      "learning_rate": 2.5752991811390422e-05,
      "loss": 1.9221,
      "step": 199793
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.5417325496673584,
      "learning_rate": 2.5751333682519015e-05,
      "loss": 1.8682,
      "step": 199794
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.7032790184020996,
      "learning_rate": 2.574967560463562e-05,
      "loss": 2.0907,
      "step": 199795
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.5115786790847778,
      "learning_rate": 2.5748017577740544e-05,
      "loss": 1.882,
      "step": 199796
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.5345267057418823,
      "learning_rate": 2.574635960183411e-05,
      "loss": 1.9753,
      "step": 199797
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.4666025638580322,
      "learning_rate": 2.5744701676916592e-05,
      "loss": 1.6674,
      "step": 199798
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.3822311162948608,
      "learning_rate": 2.5743043802988316e-05,
      "loss": 1.9275,
      "step": 199799
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.4628349542617798,
      "learning_rate": 2.5741385980049557e-05,
      "loss": 2.076,
      "step": 199800
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.2580814361572266,
      "learning_rate": 2.5739728208100674e-05,
      "loss": 1.8709,
      "step": 199801
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.4291588068008423,
      "learning_rate": 2.5738070487141904e-05,
      "loss": 1.8161,
      "step": 199802
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.3440510034561157,
      "learning_rate": 2.573641281717368e-05,
      "loss": 1.8959,
      "step": 199803
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.352919578552246,
      "learning_rate": 2.57347551981962e-05,
      "loss": 2.0859,
      "step": 199804
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.3939399719238281,
      "learning_rate": 2.573309763020983e-05,
      "loss": 1.9782,
      "step": 199805
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.571548581123352,
      "learning_rate": 2.573144011321481e-05,
      "loss": 1.8248,
      "step": 199806
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.4723601341247559,
      "learning_rate": 2.572978264721153e-05,
      "loss": 2.1322,
      "step": 199807
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.6146535873413086,
      "learning_rate": 2.572812523220019e-05,
      "loss": 1.9484,
      "step": 199808
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.4371941089630127,
      "learning_rate": 2.5726467868181268e-05,
      "loss": 2.134,
      "step": 199809
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.3512688875198364,
      "learning_rate": 2.572481055515495e-05,
      "loss": 2.1901,
      "step": 199810
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.4056788682937622,
      "learning_rate": 2.5723153293121544e-05,
      "loss": 1.8007,
      "step": 199811
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.3906139135360718,
      "learning_rate": 2.572149608208135e-05,
      "loss": 1.9909,
      "step": 199812
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.4976249933242798,
      "learning_rate": 2.5719838922034765e-05,
      "loss": 1.8962,
      "step": 199813
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.3179566860198975,
      "learning_rate": 2.5718181812981988e-05,
      "loss": 2.0407,
      "step": 199814
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.4040937423706055,
      "learning_rate": 2.571652475492342e-05,
      "loss": 2.0834,
      "step": 199815
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.4855223894119263,
      "learning_rate": 2.5714867747859324e-05,
      "loss": 2.0398,
      "step": 199816
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.4086798429489136,
      "learning_rate": 2.5713210791790008e-05,
      "loss": 1.9831,
      "step": 199817
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.420837640762329,
      "learning_rate": 2.5711553886715764e-05,
      "loss": 1.8829,
      "step": 199818
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.4596881866455078,
      "learning_rate": 2.570989703263693e-05,
      "loss": 2.0525,
      "step": 199819
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.3232520818710327,
      "learning_rate": 2.5708240229553767e-05,
      "loss": 1.8732,
      "step": 199820
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.3141061067581177,
      "learning_rate": 2.570658347746668e-05,
      "loss": 1.7783,
      "step": 199821
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.4364207983016968,
      "learning_rate": 2.5704926776375935e-05,
      "loss": 1.7339,
      "step": 199822
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.4334903955459595,
      "learning_rate": 2.5703270126281794e-05,
      "loss": 1.8956,
      "step": 199823
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.5481281280517578,
      "learning_rate": 2.5701613527184528e-05,
      "loss": 2.0157,
      "step": 199824
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.5543959140777588,
      "learning_rate": 2.5699956979084602e-05,
      "loss": 2.1356,
      "step": 199825
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.484821081161499,
      "learning_rate": 2.5698300481982148e-05,
      "loss": 2.0258,
      "step": 199826
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.5005595684051514,
      "learning_rate": 2.56966440358776e-05,
      "loss": 1.8815,
      "step": 199827
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.4351308345794678,
      "learning_rate": 2.5694987640771258e-05,
      "loss": 2.0163,
      "step": 199828
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.5316752195358276,
      "learning_rate": 2.5693331296663388e-05,
      "loss": 2.0241,
      "step": 199829
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.3623600006103516,
      "learning_rate": 2.5691675003554257e-05,
      "loss": 2.0164,
      "step": 199830
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.4479376077651978,
      "learning_rate": 2.5690018761444265e-05,
      "loss": 1.9781,
      "step": 199831
    },
    {
      "epoch": 2.6,
      "grad_norm": 2.2136282920837402,
      "learning_rate": 2.5688362570333608e-05,
      "loss": 1.9569,
      "step": 199832
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.459312915802002,
      "learning_rate": 2.5686706430222724e-05,
      "loss": 2.0893,
      "step": 199833
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.4189825057983398,
      "learning_rate": 2.568505034111188e-05,
      "loss": 1.978,
      "step": 199834
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.405602216720581,
      "learning_rate": 2.568339430300127e-05,
      "loss": 1.9295,
      "step": 199835
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.4810203313827515,
      "learning_rate": 2.56817383158914e-05,
      "loss": 1.8516,
      "step": 199836
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.4489407539367676,
      "learning_rate": 2.568008237978243e-05,
      "loss": 2.0446,
      "step": 199837
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.4517431259155273,
      "learning_rate": 2.5678426494674664e-05,
      "loss": 2.103,
      "step": 199838
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.385290265083313,
      "learning_rate": 2.5676770660568536e-05,
      "loss": 1.8566,
      "step": 199839
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.5190963745117188,
      "learning_rate": 2.567511487746424e-05,
      "loss": 2.0184,
      "step": 199840
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.4337224960327148,
      "learning_rate": 2.5673459145362084e-05,
      "loss": 1.9803,
      "step": 199841
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.4631974697113037,
      "learning_rate": 2.5671803464262464e-05,
      "loss": 1.7531,
      "step": 199842
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.4843426942825317,
      "learning_rate": 2.567014783416561e-05,
      "loss": 2.1052,
      "step": 199843
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.4977447986602783,
      "learning_rate": 2.566849225507186e-05,
      "loss": 1.9589,
      "step": 199844
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.4530811309814453,
      "learning_rate": 2.5666836726981543e-05,
      "loss": 1.7811,
      "step": 199845
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.5414385795593262,
      "learning_rate": 2.5665181249894863e-05,
      "loss": 1.9352,
      "step": 199846
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.5573577880859375,
      "learning_rate": 2.5663525823812282e-05,
      "loss": 1.8947,
      "step": 199847
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.481971263885498,
      "learning_rate": 2.5661870448734e-05,
      "loss": 1.9341,
      "step": 199848
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.462598204612732,
      "learning_rate": 2.5660215124660322e-05,
      "loss": 2.1363,
      "step": 199849
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.47903311252594,
      "learning_rate": 2.565855985159164e-05,
      "loss": 2.0067,
      "step": 199850
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.4201375246047974,
      "learning_rate": 2.565690462952823e-05,
      "loss": 1.9258,
      "step": 199851
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.640390157699585,
      "learning_rate": 2.5655249458470283e-05,
      "loss": 1.898,
      "step": 199852
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.520751953125,
      "learning_rate": 2.5653594338418304e-05,
      "loss": 2.2111,
      "step": 199853
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.4748588800430298,
      "learning_rate": 2.5651939269372457e-05,
      "loss": 2.1477,
      "step": 199854
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.5059466361999512,
      "learning_rate": 2.5650284251333077e-05,
      "loss": 2.182,
      "step": 199855
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.5298995971679688,
      "learning_rate": 2.5648629284300527e-05,
      "loss": 1.8417,
      "step": 199856
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.324478268623352,
      "learning_rate": 2.564697436827501e-05,
      "loss": 1.9072,
      "step": 199857
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.4307811260223389,
      "learning_rate": 2.5645319503256957e-05,
      "loss": 1.9399,
      "step": 199858
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.4134063720703125,
      "learning_rate": 2.5643664689246635e-05,
      "loss": 1.9561,
      "step": 199859
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.4660075902938843,
      "learning_rate": 2.564200992624431e-05,
      "loss": 2.1233,
      "step": 199860
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.3655226230621338,
      "learning_rate": 2.5640355214250285e-05,
      "loss": 1.9149,
      "step": 199861
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.4932547807693481,
      "learning_rate": 2.5638700553264956e-05,
      "loss": 2.1222,
      "step": 199862
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.6072413921356201,
      "learning_rate": 2.5637045943288492e-05,
      "loss": 2.1555,
      "step": 199863
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.4800418615341187,
      "learning_rate": 2.5635391384321356e-05,
      "loss": 2.078,
      "step": 199864
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.3967586755752563,
      "learning_rate": 2.5633736876363753e-05,
      "loss": 2.014,
      "step": 199865
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.6031076908111572,
      "learning_rate": 2.5632082419416044e-05,
      "loss": 2.2341,
      "step": 199866
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.470784068107605,
      "learning_rate": 2.5630428013478433e-05,
      "loss": 2.0317,
      "step": 199867
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.3075547218322754,
      "learning_rate": 2.5628773658551383e-05,
      "loss": 1.818,
      "step": 199868
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.5493987798690796,
      "learning_rate": 2.562711935463503e-05,
      "loss": 2.2256,
      "step": 199869
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.5251989364624023,
      "learning_rate": 2.5625465101729868e-05,
      "loss": 1.9895,
      "step": 199870
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.3918180465698242,
      "learning_rate": 2.5623810899836105e-05,
      "loss": 1.9736,
      "step": 199871
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.362038493156433,
      "learning_rate": 2.5622156748954036e-05,
      "loss": 2.0801,
      "step": 199872
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.4062880277633667,
      "learning_rate": 2.5620502649083928e-05,
      "loss": 1.9998,
      "step": 199873
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.4823479652404785,
      "learning_rate": 2.561884860022625e-05,
      "loss": 2.2028,
      "step": 199874
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.4334921836853027,
      "learning_rate": 2.5617194602381097e-05,
      "loss": 2.2162,
      "step": 199875
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.4841965436935425,
      "learning_rate": 2.561554065554897e-05,
      "loss": 1.9284,
      "step": 199876
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.4259788990020752,
      "learning_rate": 2.5613886759730106e-05,
      "loss": 1.7835,
      "step": 199877
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.375296950340271,
      "learning_rate": 2.561223291492477e-05,
      "loss": 2.0647,
      "step": 199878
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.5163789987564087,
      "learning_rate": 2.5610579121133258e-05,
      "loss": 2.2951,
      "step": 199879
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.4151561260223389,
      "learning_rate": 2.560892537835597e-05,
      "loss": 2.2778,
      "step": 199880
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.7169851064682007,
      "learning_rate": 2.560727168659311e-05,
      "loss": 2.0403,
      "step": 199881
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.4037443399429321,
      "learning_rate": 2.560561804584511e-05,
      "loss": 1.9527,
      "step": 199882
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.4470863342285156,
      "learning_rate": 2.560396445611217e-05,
      "loss": 2.1658,
      "step": 199883
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.3380444049835205,
      "learning_rate": 2.5602310917394654e-05,
      "loss": 2.1067,
      "step": 199884
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.3990334272384644,
      "learning_rate": 2.5600657429692793e-05,
      "loss": 1.9623,
      "step": 199885
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.3811103105545044,
      "learning_rate": 2.5599003993006995e-05,
      "loss": 1.9361,
      "step": 199886
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.384841799736023,
      "learning_rate": 2.5597350607337487e-05,
      "loss": 1.8918,
      "step": 199887
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.3830803632736206,
      "learning_rate": 2.5595697272684634e-05,
      "loss": 2.0299,
      "step": 199888
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.4701828956604004,
      "learning_rate": 2.559404398904874e-05,
      "loss": 2.0393,
      "step": 199889
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.498603343963623,
      "learning_rate": 2.5592390756430104e-05,
      "loss": 1.8108,
      "step": 199890
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.494046926498413,
      "learning_rate": 2.5590737574828955e-05,
      "loss": 1.997,
      "step": 199891
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.3599438667297363,
      "learning_rate": 2.5589084444245733e-05,
      "loss": 2.032,
      "step": 199892
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.4644246101379395,
      "learning_rate": 2.55874313646806e-05,
      "loss": 2.2084,
      "step": 199893
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.4682815074920654,
      "learning_rate": 2.5585778336134022e-05,
      "loss": 1.922,
      "step": 199894
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.6014642715454102,
      "learning_rate": 2.5584125358606234e-05,
      "loss": 2.0178,
      "step": 199895
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.414842128753662,
      "learning_rate": 2.55824724320975e-05,
      "loss": 1.9722,
      "step": 199896
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.477430820465088,
      "learning_rate": 2.5580819556608158e-05,
      "loss": 1.875,
      "step": 199897
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.3245837688446045,
      "learning_rate": 2.5579166732138533e-05,
      "loss": 2.2707,
      "step": 199898
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.4471744298934937,
      "learning_rate": 2.55775139586889e-05,
      "loss": 2.0549,
      "step": 199899
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.4698233604431152,
      "learning_rate": 2.557586123625962e-05,
      "loss": 2.0503,
      "step": 199900
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.4198447465896606,
      "learning_rate": 2.5574208564850963e-05,
      "loss": 1.891,
      "step": 199901
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.4978654384613037,
      "learning_rate": 2.5572555944463258e-05,
      "loss": 1.8235,
      "step": 199902
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.508795976638794,
      "learning_rate": 2.557090337509674e-05,
      "loss": 2.0348,
      "step": 199903
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.3889118432998657,
      "learning_rate": 2.556925085675181e-05,
      "loss": 1.9474,
      "step": 199904
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.4181747436523438,
      "learning_rate": 2.55675983894287e-05,
      "loss": 1.8715,
      "step": 199905
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.6622732877731323,
      "learning_rate": 2.556594597312781e-05,
      "loss": 2.0993,
      "step": 199906
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.4485076665878296,
      "learning_rate": 2.5564293607849372e-05,
      "loss": 1.935,
      "step": 199907
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.32147216796875,
      "learning_rate": 2.5562641293593722e-05,
      "loss": 1.9127,
      "step": 199908
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.40264093875885,
      "learning_rate": 2.556098903036109e-05,
      "loss": 2.1051,
      "step": 199909
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.670295238494873,
      "learning_rate": 2.5559336818151942e-05,
      "loss": 1.9826,
      "step": 199910
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.6963080167770386,
      "learning_rate": 2.5557684656966415e-05,
      "loss": 2.0892,
      "step": 199911
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.4667648077011108,
      "learning_rate": 2.555603254680494e-05,
      "loss": 2.0054,
      "step": 199912
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.4633252620697021,
      "learning_rate": 2.5554380487667746e-05,
      "loss": 2.1495,
      "step": 199913
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.4726357460021973,
      "learning_rate": 2.555272847955524e-05,
      "loss": 1.9801,
      "step": 199914
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.4499049186706543,
      "learning_rate": 2.5551076522467585e-05,
      "loss": 1.9992,
      "step": 199915
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.4525877237319946,
      "learning_rate": 2.5549424616405244e-05,
      "loss": 2.1842,
      "step": 199916
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.5247423648834229,
      "learning_rate": 2.5547772761368358e-05,
      "loss": 2.1676,
      "step": 199917
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.5006927251815796,
      "learning_rate": 2.5546120957357387e-05,
      "loss": 2.0804,
      "step": 199918
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.278417944908142,
      "learning_rate": 2.5544469204372497e-05,
      "loss": 2.074,
      "step": 199919
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.4876052141189575,
      "learning_rate": 2.554281750241416e-05,
      "loss": 1.9694,
      "step": 199920
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.4331011772155762,
      "learning_rate": 2.554116585148257e-05,
      "loss": 1.9353,
      "step": 199921
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.4076781272888184,
      "learning_rate": 2.5539514251578065e-05,
      "loss": 2.0811,
      "step": 199922
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.4522351026535034,
      "learning_rate": 2.5537862702700906e-05,
      "loss": 2.0852,
      "step": 199923
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.3999308347702026,
      "learning_rate": 2.5536211204851464e-05,
      "loss": 1.9072,
      "step": 199924
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.4358137845993042,
      "learning_rate": 2.5534559758030004e-05,
      "loss": 2.053,
      "step": 199925
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.3445441722869873,
      "learning_rate": 2.5532908362236858e-05,
      "loss": 1.9089,
      "step": 199926
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.325803279876709,
      "learning_rate": 2.553125701747236e-05,
      "loss": 1.9126,
      "step": 199927
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.4794858694076538,
      "learning_rate": 2.552960572373678e-05,
      "loss": 2.1773,
      "step": 199928
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.3763140439987183,
      "learning_rate": 2.5527954481030344e-05,
      "loss": 1.9249,
      "step": 199929
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.493906855583191,
      "learning_rate": 2.5526303289353524e-05,
      "loss": 2.154,
      "step": 199930
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.5916990041732788,
      "learning_rate": 2.552465214870648e-05,
      "loss": 2.0743,
      "step": 199931
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.4971332550048828,
      "learning_rate": 2.5523001059089654e-05,
      "loss": 2.0007,
      "step": 199932
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.4876466989517212,
      "learning_rate": 2.5521350020503272e-05,
      "loss": 1.9563,
      "step": 199933
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.4379305839538574,
      "learning_rate": 2.5519699032947606e-05,
      "loss": 2.0558,
      "step": 199934
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.5973771810531616,
      "learning_rate": 2.551804809642305e-05,
      "loss": 2.1679,
      "step": 199935
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.4761605262756348,
      "learning_rate": 2.5516397210929873e-05,
      "loss": 2.0043,
      "step": 199936
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.4309489727020264,
      "learning_rate": 2.551474637646834e-05,
      "loss": 2.0393,
      "step": 199937
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.4013967514038086,
      "learning_rate": 2.5513095593038823e-05,
      "loss": 2.0185,
      "step": 199938
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.4876283407211304,
      "learning_rate": 2.5511444860641617e-05,
      "loss": 1.88,
      "step": 199939
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.3306416273117065,
      "learning_rate": 2.5509794179276988e-05,
      "loss": 1.7383,
      "step": 199940
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.6096822023391724,
      "learning_rate": 2.550814354894527e-05,
      "loss": 1.9498,
      "step": 199941
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.4390571117401123,
      "learning_rate": 2.5506492969646762e-05,
      "loss": 2.118,
      "step": 199942
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.3956652879714966,
      "learning_rate": 2.55048424413818e-05,
      "loss": 1.8378,
      "step": 199943
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.6199344396591187,
      "learning_rate": 2.550319196415068e-05,
      "loss": 1.9827,
      "step": 199944
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.4505976438522339,
      "learning_rate": 2.5501541537953707e-05,
      "loss": 2.014,
      "step": 199945
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.4572407007217407,
      "learning_rate": 2.5499891162791142e-05,
      "loss": 2.0261,
      "step": 199946
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.6680607795715332,
      "learning_rate": 2.5498240838663352e-05,
      "loss": 2.086,
      "step": 199947
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.379595160484314,
      "learning_rate": 2.5496590565570575e-05,
      "loss": 2.0496,
      "step": 199948
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.386439323425293,
      "learning_rate": 2.5494940343513204e-05,
      "loss": 2.1241,
      "step": 199949
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.3459073305130005,
      "learning_rate": 2.5493290172491542e-05,
      "loss": 1.9175,
      "step": 199950
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.4029167890548706,
      "learning_rate": 2.5491640052505825e-05,
      "loss": 2.1368,
      "step": 199951
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.4704669713974,
      "learning_rate": 2.5489989983556347e-05,
      "loss": 1.9586,
      "step": 199952
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.372147798538208,
      "learning_rate": 2.548833996564351e-05,
      "loss": 2.0845,
      "step": 199953
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.4793846607208252,
      "learning_rate": 2.548668999876755e-05,
      "loss": 2.0853,
      "step": 199954
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.4612666368484497,
      "learning_rate": 2.5485040082928832e-05,
      "loss": 2.0635,
      "step": 199955
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.4491629600524902,
      "learning_rate": 2.548339021812762e-05,
      "loss": 2.1494,
      "step": 199956
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.6187366247177124,
      "learning_rate": 2.5481740404364215e-05,
      "loss": 1.9348,
      "step": 199957
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.5148284435272217,
      "learning_rate": 2.5480090641638916e-05,
      "loss": 2.1827,
      "step": 199958
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.5418094396591187,
      "learning_rate": 2.547844092995206e-05,
      "loss": 2.1001,
      "step": 199959
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.4955642223358154,
      "learning_rate": 2.5476791269303943e-05,
      "loss": 2.1153,
      "step": 199960
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.3294917345046997,
      "learning_rate": 2.54751416596949e-05,
      "loss": 1.9931,
      "step": 199961
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.4368170499801636,
      "learning_rate": 2.547349210112519e-05,
      "loss": 2.0269,
      "step": 199962
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.4296154975891113,
      "learning_rate": 2.547184259359516e-05,
      "loss": 1.7871,
      "step": 199963
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.4438341856002808,
      "learning_rate": 2.5470193137105032e-05,
      "loss": 1.9814,
      "step": 199964
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.4293454885482788,
      "learning_rate": 2.5468543731655245e-05,
      "loss": 2.1044,
      "step": 199965
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.3955620527267456,
      "learning_rate": 2.5466894377245993e-05,
      "loss": 1.88,
      "step": 199966
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.3950997591018677,
      "learning_rate": 2.5465245073877683e-05,
      "loss": 1.8966,
      "step": 199967
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.5159443616867065,
      "learning_rate": 2.546359582155054e-05,
      "loss": 2.04,
      "step": 199968
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.5139719247817993,
      "learning_rate": 2.5461946620264907e-05,
      "loss": 2.1949,
      "step": 199969
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.5531033277511597,
      "learning_rate": 2.546029747002104e-05,
      "loss": 2.2039,
      "step": 199970
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.4300107955932617,
      "learning_rate": 2.5458648370819313e-05,
      "loss": 2.033,
      "step": 199971
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.3884007930755615,
      "learning_rate": 2.5456999322659953e-05,
      "loss": 2.0887,
      "step": 199972
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.3885606527328491,
      "learning_rate": 2.5455350325543398e-05,
      "loss": 1.8321,
      "step": 199973
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.2805081605911255,
      "learning_rate": 2.5453701379469882e-05,
      "loss": 1.9488,
      "step": 199974
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.5302369594573975,
      "learning_rate": 2.5452052484439666e-05,
      "loss": 1.8735,
      "step": 199975
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.4720301628112793,
      "learning_rate": 2.5450403640453088e-05,
      "loss": 2.3618,
      "step": 199976
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.453286051750183,
      "learning_rate": 2.5448754847510478e-05,
      "loss": 1.8385,
      "step": 199977
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.3406771421432495,
      "learning_rate": 2.5447106105612102e-05,
      "loss": 1.9121,
      "step": 199978
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.3616068363189697,
      "learning_rate": 2.544545741475833e-05,
      "loss": 1.9472,
      "step": 199979
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.355796456336975,
      "learning_rate": 2.544380877494936e-05,
      "loss": 1.9331,
      "step": 199980
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.5890960693359375,
      "learning_rate": 2.5442160186185688e-05,
      "loss": 2.1649,
      "step": 199981
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.4402525424957275,
      "learning_rate": 2.544051164846742e-05,
      "loss": 1.9459,
      "step": 199982
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.3687316179275513,
      "learning_rate": 2.5438863161794985e-05,
      "loss": 2.0205,
      "step": 199983
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.4482474327087402,
      "learning_rate": 2.5437214726168585e-05,
      "loss": 2.1922,
      "step": 199984
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.2929673194885254,
      "learning_rate": 2.543556634158862e-05,
      "loss": 1.8274,
      "step": 199985
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.3653643131256104,
      "learning_rate": 2.5433918008055354e-05,
      "loss": 1.9435,
      "step": 199986
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.5955637693405151,
      "learning_rate": 2.5432269725569187e-05,
      "loss": 2.1199,
      "step": 199987
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.4059429168701172,
      "learning_rate": 2.5430621494130254e-05,
      "loss": 1.8674,
      "step": 199988
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.322691559791565,
      "learning_rate": 2.5428973313738987e-05,
      "loss": 1.942,
      "step": 199989
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.4326064586639404,
      "learning_rate": 2.5427325184395618e-05,
      "loss": 2.0377,
      "step": 199990
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.4596986770629883,
      "learning_rate": 2.542567710610055e-05,
      "loss": 1.9522,
      "step": 199991
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.3528350591659546,
      "learning_rate": 2.5424029078853948e-05,
      "loss": 2.0533,
      "step": 199992
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.5404164791107178,
      "learning_rate": 2.542238110265634e-05,
      "loss": 1.9824,
      "step": 199993
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.5682899951934814,
      "learning_rate": 2.5420733177507767e-05,
      "loss": 1.7119,
      "step": 199994
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.5615897178649902,
      "learning_rate": 2.5419085303408727e-05,
      "loss": 2.0477,
      "step": 199995
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.3528623580932617,
      "learning_rate": 2.5417437480359414e-05,
      "loss": 1.8044,
      "step": 199996
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.3732690811157227,
      "learning_rate": 2.5415789708360235e-05,
      "loss": 2.0449,
      "step": 199997
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.3962492942810059,
      "learning_rate": 2.5414141987411386e-05,
      "loss": 2.1399,
      "step": 199998
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.4632850885391235,
      "learning_rate": 2.541249431751333e-05,
      "loss": 1.9025,
      "step": 199999
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.5003315210342407,
      "learning_rate": 2.5410846698666175e-05,
      "loss": 2.07,
      "step": 200000
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.463654637336731,
      "learning_rate": 2.5409199130870383e-05,
      "loss": 2.0703,
      "step": 200001
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.5160467624664307,
      "learning_rate": 2.5407551614126154e-05,
      "loss": 1.988,
      "step": 200002
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.4482239484786987,
      "learning_rate": 2.5405904148433885e-05,
      "loss": 2.0565,
      "step": 200003
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.422995686531067,
      "learning_rate": 2.5404256733793783e-05,
      "loss": 1.8686,
      "step": 200004
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.4748480319976807,
      "learning_rate": 2.5402609370206274e-05,
      "loss": 2.0612,
      "step": 200005
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.4331748485565186,
      "learning_rate": 2.5400962057671626e-05,
      "loss": 2.0929,
      "step": 200006
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.4582105875015259,
      "learning_rate": 2.5399314796190107e-05,
      "loss": 1.9211,
      "step": 200007
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.3060688972473145,
      "learning_rate": 2.539766758576198e-05,
      "loss": 2.0762,
      "step": 200008
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.4540174007415771,
      "learning_rate": 2.5396020426387655e-05,
      "loss": 2.0182,
      "step": 200009
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.4157549142837524,
      "learning_rate": 2.539437331806735e-05,
      "loss": 2.0837,
      "step": 200010
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.428907871246338,
      "learning_rate": 2.5392726260801478e-05,
      "loss": 2.0496,
      "step": 200011
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.4596437215805054,
      "learning_rate": 2.5391079254590264e-05,
      "loss": 1.8218,
      "step": 200012
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.3934202194213867,
      "learning_rate": 2.5389432299434043e-05,
      "loss": 2.2033,
      "step": 200013
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.400911808013916,
      "learning_rate": 2.5387785395333048e-05,
      "loss": 1.8532,
      "step": 200014
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.3869822025299072,
      "learning_rate": 2.5386138542287716e-05,
      "loss": 1.9041,
      "step": 200015
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.4089045524597168,
      "learning_rate": 2.538449174029824e-05,
      "loss": 1.806,
      "step": 200016
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.462307333946228,
      "learning_rate": 2.5382844989364993e-05,
      "loss": 2.2642,
      "step": 200017
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.4495652914047241,
      "learning_rate": 2.538119828948827e-05,
      "loss": 2.0829,
      "step": 200018
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.3441803455352783,
      "learning_rate": 2.5379551640668338e-05,
      "loss": 2.0726,
      "step": 200019
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.4012978076934814,
      "learning_rate": 2.5377905042905566e-05,
      "loss": 2.04,
      "step": 200020
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.3732517957687378,
      "learning_rate": 2.537625849620022e-05,
      "loss": 1.7394,
      "step": 200021
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.4535515308380127,
      "learning_rate": 2.537461200055253e-05,
      "loss": 1.9831,
      "step": 200022
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.4357644319534302,
      "learning_rate": 2.5372965555962997e-05,
      "loss": 2.0091,
      "step": 200023
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.5550856590270996,
      "learning_rate": 2.537131916243176e-05,
      "loss": 1.8126,
      "step": 200024
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.653423547744751,
      "learning_rate": 2.5369672819959174e-05,
      "loss": 1.966,
      "step": 200025
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.3510777950286865,
      "learning_rate": 2.536802652854555e-05,
      "loss": 1.7656,
      "step": 200026
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.4226346015930176,
      "learning_rate": 2.5366380288191244e-05,
      "loss": 2.0414,
      "step": 200027
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.405897855758667,
      "learning_rate": 2.5364734098896434e-05,
      "loss": 1.8466,
      "step": 200028
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.6546077728271484,
      "learning_rate": 2.5363087960661542e-05,
      "loss": 2.0486,
      "step": 200029
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.555274486541748,
      "learning_rate": 2.5361441873486877e-05,
      "loss": 1.8932,
      "step": 200030
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.3930995464324951,
      "learning_rate": 2.5359795837372632e-05,
      "loss": 1.9301,
      "step": 200031
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.5932211875915527,
      "learning_rate": 2.535814985231921e-05,
      "loss": 2.1265,
      "step": 200032
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.4576423168182373,
      "learning_rate": 2.5356503918326875e-05,
      "loss": 1.8967,
      "step": 200033
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.4591891765594482,
      "learning_rate": 2.5354858035395998e-05,
      "loss": 2.0752,
      "step": 200034
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.5003376007080078,
      "learning_rate": 2.535321220352684e-05,
      "loss": 2.0933,
      "step": 200035
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.4158827066421509,
      "learning_rate": 2.535156642271967e-05,
      "loss": 2.066,
      "step": 200036
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.7049882411956787,
      "learning_rate": 2.5349920692974823e-05,
      "loss": 2.1155,
      "step": 200037
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.2721302509307861,
      "learning_rate": 2.5348275014292628e-05,
      "loss": 2.0404,
      "step": 200038
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.4854265451431274,
      "learning_rate": 2.5346629386673324e-05,
      "loss": 1.9395,
      "step": 200039
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.5985187292099,
      "learning_rate": 2.5344983810117336e-05,
      "loss": 1.9179,
      "step": 200040
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.4800540208816528,
      "learning_rate": 2.5343338284624902e-05,
      "loss": 2.0878,
      "step": 200041
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.4642387628555298,
      "learning_rate": 2.5341692810196324e-05,
      "loss": 1.8842,
      "step": 200042
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.492815375328064,
      "learning_rate": 2.534004738683183e-05,
      "loss": 1.9977,
      "step": 200043
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.412122130393982,
      "learning_rate": 2.533840201453189e-05,
      "loss": 1.8608,
      "step": 200044
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.3997304439544678,
      "learning_rate": 2.5336756693296634e-05,
      "loss": 2.0562,
      "step": 200045
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.4899815320968628,
      "learning_rate": 2.5335111423126565e-05,
      "loss": 2.1239,
      "step": 200046
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.4996726512908936,
      "learning_rate": 2.533346620402178e-05,
      "loss": 2.0509,
      "step": 200047
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.4763962030410767,
      "learning_rate": 2.5331821035982845e-05,
      "loss": 1.9725,
      "step": 200048
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.5537627935409546,
      "learning_rate": 2.5330175919009767e-05,
      "loss": 2.0214,
      "step": 200049
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.4497604370117188,
      "learning_rate": 2.532853085310307e-05,
      "loss": 1.8241,
      "step": 200050
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.437497854232788,
      "learning_rate": 2.532688583826289e-05,
      "loss": 1.9572,
      "step": 200051
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.4232635498046875,
      "learning_rate": 2.5325240874489726e-05,
      "loss": 2.0059,
      "step": 200052
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.3516101837158203,
      "learning_rate": 2.5323595961783715e-05,
      "loss": 1.744,
      "step": 200053
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.5561710596084595,
      "learning_rate": 2.532195110014532e-05,
      "loss": 2.0267,
      "step": 200054
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.3949477672576904,
      "learning_rate": 2.5320306289574678e-05,
      "loss": 1.8966,
      "step": 200055
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.5017473697662354,
      "learning_rate": 2.5318661530072214e-05,
      "loss": 2.0117,
      "step": 200056
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.4331594705581665,
      "learning_rate": 2.531701682163817e-05,
      "loss": 2.0996,
      "step": 200057
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.4907515048980713,
      "learning_rate": 2.5315372164272907e-05,
      "loss": 2.0814,
      "step": 200058
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.5099153518676758,
      "learning_rate": 2.531372755797666e-05,
      "loss": 2.0773,
      "step": 200059
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.3946176767349243,
      "learning_rate": 2.531208300274986e-05,
      "loss": 1.868,
      "step": 200060
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.4997406005859375,
      "learning_rate": 2.5310438498592644e-05,
      "loss": 2.0176,
      "step": 200061
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.4082422256469727,
      "learning_rate": 2.5308794045505443e-05,
      "loss": 2.0745,
      "step": 200062
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.4029662609100342,
      "learning_rate": 2.5307149643488488e-05,
      "loss": 2.077,
      "step": 200063
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.5444250106811523,
      "learning_rate": 2.530550529254215e-05,
      "loss": 1.9879,
      "step": 200064
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.4425909519195557,
      "learning_rate": 2.5303860992666656e-05,
      "loss": 1.8756,
      "step": 200065
    },
    {
      "epoch": 2.6,
      "grad_norm": 1.5397155284881592,
      "learning_rate": 2.530221674386248e-05,
      "loss": 2.1546,
      "step": 200066
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.3792598247528076,
      "learning_rate": 2.5300572546129683e-05,
      "loss": 1.7856,
      "step": 200067
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.8023830652236938,
      "learning_rate": 2.5298928399468764e-05,
      "loss": 1.776,
      "step": 200068
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.4973317384719849,
      "learning_rate": 2.5297284303879896e-05,
      "loss": 2.1818,
      "step": 200069
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.4302210807800293,
      "learning_rate": 2.5295640259363503e-05,
      "loss": 1.8226,
      "step": 200070
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.4397612810134888,
      "learning_rate": 2.529399626591979e-05,
      "loss": 2.0766,
      "step": 200071
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.398813247680664,
      "learning_rate": 2.5292352323549192e-05,
      "loss": 1.8273,
      "step": 200072
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.5063934326171875,
      "learning_rate": 2.529070843225184e-05,
      "loss": 1.9703,
      "step": 200073
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.527836561203003,
      "learning_rate": 2.5289064592028196e-05,
      "loss": 1.7993,
      "step": 200074
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.513257622718811,
      "learning_rate": 2.5287420802878434e-05,
      "loss": 1.9007,
      "step": 200075
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.4316667318344116,
      "learning_rate": 2.5285777064802982e-05,
      "loss": 2.1298,
      "step": 200076
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.3867846727371216,
      "learning_rate": 2.5284133377802042e-05,
      "loss": 1.9237,
      "step": 200077
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.4453996419906616,
      "learning_rate": 2.5282489741876044e-05,
      "loss": 1.9142,
      "step": 200078
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.428362488746643,
      "learning_rate": 2.528084615702516e-05,
      "loss": 1.8182,
      "step": 200079
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.3371840715408325,
      "learning_rate": 2.5279202623249782e-05,
      "loss": 2.0466,
      "step": 200080
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.3690080642700195,
      "learning_rate": 2.527755914055012e-05,
      "loss": 1.8235,
      "step": 200081
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.460734248161316,
      "learning_rate": 2.5275915708926598e-05,
      "loss": 1.9652,
      "step": 200082
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.521607756614685,
      "learning_rate": 2.527427232837942e-05,
      "loss": 2.0893,
      "step": 200083
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.4612699747085571,
      "learning_rate": 2.5272628998909084e-05,
      "loss": 2.0586,
      "step": 200084
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.4010272026062012,
      "learning_rate": 2.527098572051559e-05,
      "loss": 1.9731,
      "step": 200085
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.4478991031646729,
      "learning_rate": 2.5269342493199506e-05,
      "loss": 2.0323,
      "step": 200086
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.51187264919281,
      "learning_rate": 2.5267699316960965e-05,
      "loss": 1.8294,
      "step": 200087
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.6006592512130737,
      "learning_rate": 2.52660561918004e-05,
      "loss": 2.1273,
      "step": 200088
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.4150835275650024,
      "learning_rate": 2.5264413117717973e-05,
      "loss": 2.0604,
      "step": 200089
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.3006805181503296,
      "learning_rate": 2.5262770094714224e-05,
      "loss": 1.8843,
      "step": 200090
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.4064704179763794,
      "learning_rate": 2.5261127122789215e-05,
      "loss": 2.0959,
      "step": 200091
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.5628880262374878,
      "learning_rate": 2.5259484201943382e-05,
      "loss": 1.8833,
      "step": 200092
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.526384711265564,
      "learning_rate": 2.525784133217692e-05,
      "loss": 1.9286,
      "step": 200093
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.3342576026916504,
      "learning_rate": 2.52561985134903e-05,
      "loss": 2.0043,
      "step": 200094
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.456506371498108,
      "learning_rate": 2.5254555745883685e-05,
      "loss": 1.847,
      "step": 200095
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.3540090322494507,
      "learning_rate": 2.525291302935748e-05,
      "loss": 2.0163,
      "step": 200096
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.382951259613037,
      "learning_rate": 2.5251270363911945e-05,
      "loss": 2.0206,
      "step": 200097
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.5034984350204468,
      "learning_rate": 2.5249627749547387e-05,
      "loss": 2.1087,
      "step": 200098
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.510561466217041,
      "learning_rate": 2.524798518626403e-05,
      "loss": 1.9982,
      "step": 200099
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.6054155826568604,
      "learning_rate": 2.524634267406235e-05,
      "loss": 2.0328,
      "step": 200100
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.4886034727096558,
      "learning_rate": 2.524470021294247e-05,
      "loss": 2.0924,
      "step": 200101
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.5218734741210938,
      "learning_rate": 2.5243057802904863e-05,
      "loss": 2.1142,
      "step": 200102
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.513901948928833,
      "learning_rate": 2.5241415443949764e-05,
      "loss": 2.0605,
      "step": 200103
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.3387068510055542,
      "learning_rate": 2.5239773136077433e-05,
      "loss": 2.1533,
      "step": 200104
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.4873754978179932,
      "learning_rate": 2.5238130879288243e-05,
      "loss": 2.1623,
      "step": 200105
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.4205785989761353,
      "learning_rate": 2.5236488673582455e-05,
      "loss": 1.8398,
      "step": 200106
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.4837422370910645,
      "learning_rate": 2.5234846518960373e-05,
      "loss": 1.9413,
      "step": 200107
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.3504931926727295,
      "learning_rate": 2.5233204415422357e-05,
      "loss": 1.723,
      "step": 200108
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.42503023147583,
      "learning_rate": 2.5231562362968682e-05,
      "loss": 2.0406,
      "step": 200109
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.3830939531326294,
      "learning_rate": 2.5229920361599577e-05,
      "loss": 2.1446,
      "step": 200110
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.3931924104690552,
      "learning_rate": 2.5228278411315472e-05,
      "loss": 2.3541,
      "step": 200111
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.4146251678466797,
      "learning_rate": 2.5226636512116637e-05,
      "loss": 1.8153,
      "step": 200112
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.3594605922698975,
      "learning_rate": 2.5224994664003273e-05,
      "loss": 2.055,
      "step": 200113
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.4902747869491577,
      "learning_rate": 2.5223352866975842e-05,
      "loss": 2.0243,
      "step": 200114
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.4992382526397705,
      "learning_rate": 2.5221711121034584e-05,
      "loss": 1.915,
      "step": 200115
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.490910291671753,
      "learning_rate": 2.5220069426179723e-05,
      "loss": 1.9795,
      "step": 200116
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.509064793586731,
      "learning_rate": 2.52184277824117e-05,
      "loss": 1.9992,
      "step": 200117
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.454733967781067,
      "learning_rate": 2.521678618973071e-05,
      "loss": 2.0953,
      "step": 200118
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.6788159608840942,
      "learning_rate": 2.5215144648137152e-05,
      "loss": 2.0491,
      "step": 200119
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.6402167081832886,
      "learning_rate": 2.52135031576313e-05,
      "loss": 1.9333,
      "step": 200120
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.447386384010315,
      "learning_rate": 2.5211861718213444e-05,
      "loss": 1.9915,
      "step": 200121
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.4941136837005615,
      "learning_rate": 2.5210220329883825e-05,
      "loss": 2.1565,
      "step": 200122
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.6101338863372803,
      "learning_rate": 2.520857899264287e-05,
      "loss": 1.9241,
      "step": 200123
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.300122857093811,
      "learning_rate": 2.5206937706490783e-05,
      "loss": 1.9752,
      "step": 200124
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.5052317380905151,
      "learning_rate": 2.5205296471427926e-05,
      "loss": 2.0045,
      "step": 200125
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.4360984563827515,
      "learning_rate": 2.520365528745457e-05,
      "loss": 2.1131,
      "step": 200126
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.4812514781951904,
      "learning_rate": 2.5202014154571148e-05,
      "loss": 2.0557,
      "step": 200127
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.5451984405517578,
      "learning_rate": 2.5200373072777758e-05,
      "loss": 2.0096,
      "step": 200128
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.3570345640182495,
      "learning_rate": 2.5198732042074865e-05,
      "loss": 1.9679,
      "step": 200129
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.3270024061203003,
      "learning_rate": 2.5197091062462637e-05,
      "loss": 1.8715,
      "step": 200130
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.3317958116531372,
      "learning_rate": 2.5195450133941543e-05,
      "loss": 2.1046,
      "step": 200131
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.5059785842895508,
      "learning_rate": 2.519380925651171e-05,
      "loss": 2.0441,
      "step": 200132
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.4781607389450073,
      "learning_rate": 2.5192168430173676e-05,
      "loss": 1.9979,
      "step": 200133
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.4387762546539307,
      "learning_rate": 2.5190527654927473e-05,
      "loss": 1.9823,
      "step": 200134
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.4148813486099243,
      "learning_rate": 2.51888869307736e-05,
      "loss": 1.9042,
      "step": 200135
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.3778986930847168,
      "learning_rate": 2.518724625771226e-05,
      "loss": 2.039,
      "step": 200136
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.4385250806808472,
      "learning_rate": 2.5185605635743843e-05,
      "loss": 2.0263,
      "step": 200137
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.4359588623046875,
      "learning_rate": 2.5183965064868527e-05,
      "loss": 2.0215,
      "step": 200138
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.6095112562179565,
      "learning_rate": 2.518232454508684e-05,
      "loss": 1.8939,
      "step": 200139
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.4530774354934692,
      "learning_rate": 2.518068407639885e-05,
      "loss": 1.9294,
      "step": 200140
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.5546847581863403,
      "learning_rate": 2.5179043658804986e-05,
      "loss": 2.0142,
      "step": 200141
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.61212956905365,
      "learning_rate": 2.517740329230549e-05,
      "loss": 1.9649,
      "step": 200142
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.548797845840454,
      "learning_rate": 2.5175762976900716e-05,
      "loss": 1.9254,
      "step": 200143
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.5746608972549438,
      "learning_rate": 2.517412271259094e-05,
      "loss": 2.0061,
      "step": 200144
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.4129832983016968,
      "learning_rate": 2.517248249937659e-05,
      "loss": 1.8043,
      "step": 200145
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.5273659229278564,
      "learning_rate": 2.5170842337257736e-05,
      "loss": 1.921,
      "step": 200146
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.6300773620605469,
      "learning_rate": 2.5169202226234875e-05,
      "loss": 1.6569,
      "step": 200147
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.3533074855804443,
      "learning_rate": 2.516756216630821e-05,
      "loss": 2.1201,
      "step": 200148
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.4160313606262207,
      "learning_rate": 2.51659221574781e-05,
      "loss": 1.9464,
      "step": 200149
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.5016670227050781,
      "learning_rate": 2.5164282199744822e-05,
      "loss": 2.0062,
      "step": 200150
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.3609941005706787,
      "learning_rate": 2.516264229310877e-05,
      "loss": 1.7516,
      "step": 200151
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.3552734851837158,
      "learning_rate": 2.5161002437570076e-05,
      "loss": 1.9562,
      "step": 200152
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.387125015258789,
      "learning_rate": 2.5159362633129177e-05,
      "loss": 1.8308,
      "step": 200153
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.5220229625701904,
      "learning_rate": 2.51577228797863e-05,
      "loss": 1.7253,
      "step": 200154
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.4206383228302002,
      "learning_rate": 2.515608317754185e-05,
      "loss": 1.9377,
      "step": 200155
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.5487480163574219,
      "learning_rate": 2.5154443526395996e-05,
      "loss": 1.7716,
      "step": 200156
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.4707136154174805,
      "learning_rate": 2.5152803926349264e-05,
      "loss": 2.175,
      "step": 200157
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.414820909500122,
      "learning_rate": 2.5151164377401654e-05,
      "loss": 2.0835,
      "step": 200158
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.4323093891143799,
      "learning_rate": 2.514952487955374e-05,
      "loss": 2.0189,
      "step": 200159
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.4242258071899414,
      "learning_rate": 2.5147885432805615e-05,
      "loss": 1.7995,
      "step": 200160
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.6784043312072754,
      "learning_rate": 2.514624603715778e-05,
      "loss": 2.0808,
      "step": 200161
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.504380226135254,
      "learning_rate": 2.514460669261037e-05,
      "loss": 1.9196,
      "step": 200162
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.475786805152893,
      "learning_rate": 2.5142967399163883e-05,
      "loss": 2.1652,
      "step": 200163
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.4999171495437622,
      "learning_rate": 2.5141328156818385e-05,
      "loss": 1.9989,
      "step": 200164
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.8184726238250732,
      "learning_rate": 2.5139688965574378e-05,
      "loss": 2.071,
      "step": 200165
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.4786903858184814,
      "learning_rate": 2.5138049825432028e-05,
      "loss": 1.9547,
      "step": 200166
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.4587997198104858,
      "learning_rate": 2.513641073639173e-05,
      "loss": 1.9435,
      "step": 200167
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.6310601234436035,
      "learning_rate": 2.5134771698453725e-05,
      "loss": 2.1402,
      "step": 200168
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.496363639831543,
      "learning_rate": 2.5133132711618477e-05,
      "loss": 2.0551,
      "step": 200169
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.4995943307876587,
      "learning_rate": 2.5131493775886048e-05,
      "loss": 2.0294,
      "step": 200170
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.4497456550598145,
      "learning_rate": 2.512985489125694e-05,
      "loss": 2.0071,
      "step": 200171
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.4160237312316895,
      "learning_rate": 2.5128216057731287e-05,
      "loss": 2.1824,
      "step": 200172
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.6799845695495605,
      "learning_rate": 2.512657727530959e-05,
      "loss": 2.0101,
      "step": 200173
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.6906814575195312,
      "learning_rate": 2.5124938543991946e-05,
      "loss": 2.05,
      "step": 200174
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.4306120872497559,
      "learning_rate": 2.5123299863778888e-05,
      "loss": 1.9766,
      "step": 200175
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.4160144329071045,
      "learning_rate": 2.512166123467052e-05,
      "loss": 2.1682,
      "step": 200176
    },
    {
      "epoch": 2.61,
      "grad_norm": 2.2906646728515625,
      "learning_rate": 2.512002265666724e-05,
      "loss": 1.9645,
      "step": 200177
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.4089908599853516,
      "learning_rate": 2.511838412976931e-05,
      "loss": 2.0221,
      "step": 200178
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.5213617086410522,
      "learning_rate": 2.51167456539771e-05,
      "loss": 1.9671,
      "step": 200179
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.4410359859466553,
      "learning_rate": 2.511510722929081e-05,
      "loss": 2.2254,
      "step": 200180
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.428720235824585,
      "learning_rate": 2.5113468855710873e-05,
      "loss": 1.9029,
      "step": 200181
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.4836074113845825,
      "learning_rate": 2.5111830533237555e-05,
      "loss": 2.0967,
      "step": 200182
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.5260164737701416,
      "learning_rate": 2.5110192261871122e-05,
      "loss": 2.0239,
      "step": 200183
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.4275776147842407,
      "learning_rate": 2.5108554041611838e-05,
      "loss": 1.7996,
      "step": 200184
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.532011866569519,
      "learning_rate": 2.510691587246011e-05,
      "loss": 2.3439,
      "step": 200185
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.6820298433303833,
      "learning_rate": 2.510527775441613e-05,
      "loss": 2.1145,
      "step": 200186
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.487338662147522,
      "learning_rate": 2.5103639687480337e-05,
      "loss": 2.0944,
      "step": 200187
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.543295979499817,
      "learning_rate": 2.510200167165296e-05,
      "loss": 2.2087,
      "step": 200188
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.4531069993972778,
      "learning_rate": 2.5100363706934267e-05,
      "loss": 2.0912,
      "step": 200189
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.4494173526763916,
      "learning_rate": 2.5098725793324657e-05,
      "loss": 1.9467,
      "step": 200190
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.4571930170059204,
      "learning_rate": 2.50970879308244e-05,
      "loss": 1.9712,
      "step": 200191
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.4399185180664062,
      "learning_rate": 2.5095450119433692e-05,
      "loss": 1.8936,
      "step": 200192
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.4839540719985962,
      "learning_rate": 2.5093812359152998e-05,
      "loss": 2.1279,
      "step": 200193
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.268249750137329,
      "learning_rate": 2.5092174649982556e-05,
      "loss": 1.7794,
      "step": 200194
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.4343090057373047,
      "learning_rate": 2.5090536991922627e-05,
      "loss": 1.9463,
      "step": 200195
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.420799970626831,
      "learning_rate": 2.5088899384973615e-05,
      "loss": 1.9805,
      "step": 200196
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.4083985090255737,
      "learning_rate": 2.508726182913575e-05,
      "loss": 2.2118,
      "step": 200197
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.5866773128509521,
      "learning_rate": 2.508562432440927e-05,
      "loss": 1.9474,
      "step": 200198
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.4585752487182617,
      "learning_rate": 2.508398687079467e-05,
      "loss": 1.9255,
      "step": 200199
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.5825862884521484,
      "learning_rate": 2.508234946829212e-05,
      "loss": 1.9751,
      "step": 200200
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.4469293355941772,
      "learning_rate": 2.508071211690188e-05,
      "loss": 2.08,
      "step": 200201
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.481202483177185,
      "learning_rate": 2.507907481662439e-05,
      "loss": 1.9053,
      "step": 200202
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.3693149089813232,
      "learning_rate": 2.5077437567459846e-05,
      "loss": 2.1086,
      "step": 200203
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.3391112089157104,
      "learning_rate": 2.507580036940865e-05,
      "loss": 1.9031,
      "step": 200204
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.473423957824707,
      "learning_rate": 2.5074163222471066e-05,
      "loss": 2.0322,
      "step": 200205
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.4655500650405884,
      "learning_rate": 2.5072526126647362e-05,
      "loss": 1.9548,
      "step": 200206
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.4860284328460693,
      "learning_rate": 2.5070889081937807e-05,
      "loss": 2.0387,
      "step": 200207
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.4888261556625366,
      "learning_rate": 2.506925208834283e-05,
      "loss": 2.0118,
      "step": 200208
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.4695508480072021,
      "learning_rate": 2.5067615145862595e-05,
      "loss": 2.0263,
      "step": 200209
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.3819422721862793,
      "learning_rate": 2.5065978254497575e-05,
      "loss": 1.9646,
      "step": 200210
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.5487687587738037,
      "learning_rate": 2.5064341414247936e-05,
      "loss": 1.8699,
      "step": 200211
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.3793871402740479,
      "learning_rate": 2.506270462511404e-05,
      "loss": 2.1808,
      "step": 200212
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.340694546699524,
      "learning_rate": 2.506106788709612e-05,
      "loss": 1.8598,
      "step": 200213
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.5281195640563965,
      "learning_rate": 2.505943120019461e-05,
      "loss": 1.9492,
      "step": 200214
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.4651151895523071,
      "learning_rate": 2.5057794564409684e-05,
      "loss": 2.2406,
      "step": 200215
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.3978849649429321,
      "learning_rate": 2.505615797974173e-05,
      "loss": 2.0145,
      "step": 200216
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.5383943319320679,
      "learning_rate": 2.5054521446191023e-05,
      "loss": 2.0458,
      "step": 200217
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.477634072303772,
      "learning_rate": 2.5052884963757924e-05,
      "loss": 2.2171,
      "step": 200218
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.311122179031372,
      "learning_rate": 2.5051248532442603e-05,
      "loss": 2.0848,
      "step": 200219
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.5329387187957764,
      "learning_rate": 2.504961215224549e-05,
      "loss": 1.9355,
      "step": 200220
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.4542462825775146,
      "learning_rate": 2.504797582316682e-05,
      "loss": 1.9953,
      "step": 200221
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.3177218437194824,
      "learning_rate": 2.5046339545206927e-05,
      "loss": 1.8033,
      "step": 200222
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.3911283016204834,
      "learning_rate": 2.5044703318366077e-05,
      "loss": 1.7391,
      "step": 200223
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.48915696144104,
      "learning_rate": 2.5043067142644737e-05,
      "loss": 1.8659,
      "step": 200224
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.5386855602264404,
      "learning_rate": 2.5041431018042935e-05,
      "loss": 2.0301,
      "step": 200225
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.5310653448104858,
      "learning_rate": 2.503979494456121e-05,
      "loss": 2.2789,
      "step": 200226
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.4225358963012695,
      "learning_rate": 2.503815892219969e-05,
      "loss": 1.8019,
      "step": 200227
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.3621039390563965,
      "learning_rate": 2.503652295095885e-05,
      "loss": 1.7131,
      "step": 200228
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.6275254487991333,
      "learning_rate": 2.5034887030838848e-05,
      "loss": 2.0241,
      "step": 200229
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.4834483861923218,
      "learning_rate": 2.503325116184015e-05,
      "loss": 2.1107,
      "step": 200230
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.4505335092544556,
      "learning_rate": 2.5031615343962862e-05,
      "loss": 1.9557,
      "step": 200231
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.4880305528640747,
      "learning_rate": 2.502997957720745e-05,
      "loss": 2.0425,
      "step": 200232
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.4385271072387695,
      "learning_rate": 2.5028343861574074e-05,
      "loss": 1.8465,
      "step": 200233
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.5200313329696655,
      "learning_rate": 2.502670819706317e-05,
      "loss": 2.0042,
      "step": 200234
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.8215596675872803,
      "learning_rate": 2.5025072583674976e-05,
      "loss": 2.0254,
      "step": 200235
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.415658950805664,
      "learning_rate": 2.5023437021409886e-05,
      "loss": 1.9671,
      "step": 200236
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.328499674797058,
      "learning_rate": 2.5021801510268037e-05,
      "loss": 1.9706,
      "step": 200237
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.38717782497406,
      "learning_rate": 2.502016605024989e-05,
      "loss": 1.9909,
      "step": 200238
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.3651260137557983,
      "learning_rate": 2.5018530641355617e-05,
      "loss": 1.9432,
      "step": 200239
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.294252634048462,
      "learning_rate": 2.501689528358565e-05,
      "loss": 1.9574,
      "step": 200240
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.2970646619796753,
      "learning_rate": 2.5015259976940183e-05,
      "loss": 1.9304,
      "step": 200241
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.580169677734375,
      "learning_rate": 2.5013624721419655e-05,
      "loss": 2.0142,
      "step": 200242
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.4468555450439453,
      "learning_rate": 2.5011989517024202e-05,
      "loss": 1.9431,
      "step": 200243
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.3549219369888306,
      "learning_rate": 2.5010354363754248e-05,
      "loss": 2.1014,
      "step": 200244
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.4952480792999268,
      "learning_rate": 2.500871926161e-05,
      "loss": 2.0797,
      "step": 200245
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.5130127668380737,
      "learning_rate": 2.500708421059189e-05,
      "loss": 1.9435,
      "step": 200246
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.3558238744735718,
      "learning_rate": 2.5005449210700113e-05,
      "loss": 1.8855,
      "step": 200247
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.29984450340271,
      "learning_rate": 2.500381426193504e-05,
      "loss": 2.0896,
      "step": 200248
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.6267050504684448,
      "learning_rate": 2.500217936429697e-05,
      "loss": 2.2959,
      "step": 200249
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.3882825374603271,
      "learning_rate": 2.500054451778617e-05,
      "loss": 2.0532,
      "step": 200250
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.4084384441375732,
      "learning_rate": 2.4998909722402905e-05,
      "loss": 1.8175,
      "step": 200251
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.4409613609313965,
      "learning_rate": 2.4997274978147607e-05,
      "loss": 1.8122,
      "step": 200252
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.5165082216262817,
      "learning_rate": 2.4995640285020445e-05,
      "loss": 2.0769,
      "step": 200253
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.4112917184829712,
      "learning_rate": 2.4994005643021818e-05,
      "loss": 1.8169,
      "step": 200254
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.4778125286102295,
      "learning_rate": 2.4992371052152027e-05,
      "loss": 2.0018,
      "step": 200255
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.5041589736938477,
      "learning_rate": 2.4990736512411334e-05,
      "loss": 2.0379,
      "step": 200256
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.4743249416351318,
      "learning_rate": 2.498910202379998e-05,
      "loss": 2.002,
      "step": 200257
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.4156107902526855,
      "learning_rate": 2.4987467586318423e-05,
      "loss": 1.8811,
      "step": 200258
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.4958902597427368,
      "learning_rate": 2.498583319996683e-05,
      "loss": 1.8856,
      "step": 200259
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.4504022598266602,
      "learning_rate": 2.4984198864745607e-05,
      "loss": 1.9559,
      "step": 200260
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.3588863611221313,
      "learning_rate": 2.498256458065502e-05,
      "loss": 1.9099,
      "step": 200261
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.4350579977035522,
      "learning_rate": 2.498093034769536e-05,
      "loss": 1.9861,
      "step": 200262
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.5707406997680664,
      "learning_rate": 2.4979296165866868e-05,
      "loss": 2.1868,
      "step": 200263
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.3052144050598145,
      "learning_rate": 2.497766203517001e-05,
      "loss": 1.9153,
      "step": 200264
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.5449732542037964,
      "learning_rate": 2.4976027955604915e-05,
      "loss": 1.9919,
      "step": 200265
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.5137759447097778,
      "learning_rate": 2.4974393927172022e-05,
      "loss": 1.8837,
      "step": 200266
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.3889626264572144,
      "learning_rate": 2.497275994987159e-05,
      "loss": 2.0164,
      "step": 200267
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.5085192918777466,
      "learning_rate": 2.4971126023703892e-05,
      "loss": 2.0295,
      "step": 200268
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.53806471824646,
      "learning_rate": 2.4969492148669224e-05,
      "loss": 1.7277,
      "step": 200269
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.319943904876709,
      "learning_rate": 2.4967858324767987e-05,
      "loss": 1.8001,
      "step": 200270
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.4249036312103271,
      "learning_rate": 2.4966224552000313e-05,
      "loss": 2.0497,
      "step": 200271
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.4354039430618286,
      "learning_rate": 2.4964590830366704e-05,
      "loss": 2.2141,
      "step": 200272
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.4836732149124146,
      "learning_rate": 2.496295715986736e-05,
      "loss": 2.0476,
      "step": 200273
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.5492854118347168,
      "learning_rate": 2.4961323540502576e-05,
      "loss": 2.13,
      "step": 200274
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.5263729095458984,
      "learning_rate": 2.4959689972272655e-05,
      "loss": 1.9167,
      "step": 200275
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.5744190216064453,
      "learning_rate": 2.4958056455177932e-05,
      "loss": 1.982,
      "step": 200276
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.4743374586105347,
      "learning_rate": 2.4956422989218673e-05,
      "loss": 2.2024,
      "step": 200277
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.568204402923584,
      "learning_rate": 2.495478957439524e-05,
      "loss": 1.9606,
      "step": 200278
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.4985802173614502,
      "learning_rate": 2.4953156210707936e-05,
      "loss": 1.9931,
      "step": 200279
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.4602012634277344,
      "learning_rate": 2.4951522898156963e-05,
      "loss": 1.9386,
      "step": 200280
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.3541700839996338,
      "learning_rate": 2.4949889636742716e-05,
      "loss": 1.8693,
      "step": 200281
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.4456660747528076,
      "learning_rate": 2.494825642646553e-05,
      "loss": 1.8355,
      "step": 200282
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.314016342163086,
      "learning_rate": 2.4946623267325572e-05,
      "loss": 2.0475,
      "step": 200283
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.6181010007858276,
      "learning_rate": 2.4944990159323274e-05,
      "loss": 1.8035,
      "step": 200284
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.4383293390274048,
      "learning_rate": 2.4943357102458906e-05,
      "loss": 1.9837,
      "step": 200285
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.3406485319137573,
      "learning_rate": 2.4941724096732695e-05,
      "loss": 2.0269,
      "step": 200286
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.4161226749420166,
      "learning_rate": 2.494009114214508e-05,
      "loss": 2.1173,
      "step": 200287
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.4676555395126343,
      "learning_rate": 2.4938458238696223e-05,
      "loss": 1.918,
      "step": 200288
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.4584203958511353,
      "learning_rate": 2.4936825386386595e-05,
      "loss": 1.9516,
      "step": 200289
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.4167535305023193,
      "learning_rate": 2.4935192585216357e-05,
      "loss": 1.815,
      "step": 200290
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.3080161809921265,
      "learning_rate": 2.493355983518588e-05,
      "loss": 1.9038,
      "step": 200291
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.5732741355895996,
      "learning_rate": 2.4931927136295392e-05,
      "loss": 1.9489,
      "step": 200292
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.408247470855713,
      "learning_rate": 2.4930294488545298e-05,
      "loss": 2.0427,
      "step": 200293
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.276532530784607,
      "learning_rate": 2.4928661891935798e-05,
      "loss": 2.1013,
      "step": 200294
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.4995216131210327,
      "learning_rate": 2.492702934646732e-05,
      "loss": 2.2211,
      "step": 200295
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.4419529438018799,
      "learning_rate": 2.4925396852140067e-05,
      "loss": 2.186,
      "step": 200296
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.429795503616333,
      "learning_rate": 2.4923764408954407e-05,
      "loss": 1.9293,
      "step": 200297
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.4232568740844727,
      "learning_rate": 2.4922132016910536e-05,
      "loss": 1.9333,
      "step": 200298
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.5329798460006714,
      "learning_rate": 2.492049967600892e-05,
      "loss": 2.0271,
      "step": 200299
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.507965087890625,
      "learning_rate": 2.4918867386249696e-05,
      "loss": 1.9233,
      "step": 200300
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.4846632480621338,
      "learning_rate": 2.4917235147633297e-05,
      "loss": 1.8015,
      "step": 200301
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.4120309352874756,
      "learning_rate": 2.491560296015995e-05,
      "loss": 2.2469,
      "step": 200302
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.4379898309707642,
      "learning_rate": 2.4913970823830066e-05,
      "loss": 1.8622,
      "step": 200303
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.532989501953125,
      "learning_rate": 2.4912338738643768e-05,
      "loss": 2.0381,
      "step": 200304
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.4253071546554565,
      "learning_rate": 2.4910706704601523e-05,
      "loss": 2.2523,
      "step": 200305
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.4638019800186157,
      "learning_rate": 2.4909074721703503e-05,
      "loss": 2.0922,
      "step": 200306
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.4233005046844482,
      "learning_rate": 2.490744278995014e-05,
      "loss": 1.8844,
      "step": 200307
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.3119242191314697,
      "learning_rate": 2.4905810909341628e-05,
      "loss": 1.9436,
      "step": 200308
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.469467282295227,
      "learning_rate": 2.4904179079878406e-05,
      "loss": 1.909,
      "step": 200309
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.461276650428772,
      "learning_rate": 2.4902547301560605e-05,
      "loss": 2.0334,
      "step": 200310
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.4157580137252808,
      "learning_rate": 2.490091557438866e-05,
      "loss": 1.8758,
      "step": 200311
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.5337857007980347,
      "learning_rate": 2.4899283898362765e-05,
      "loss": 2.1128,
      "step": 200312
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.4484539031982422,
      "learning_rate": 2.4897652273483326e-05,
      "loss": 2.0034,
      "step": 200313
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.514958381652832,
      "learning_rate": 2.4896020699750575e-05,
      "loss": 2.0787,
      "step": 200314
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.4222500324249268,
      "learning_rate": 2.489438917716491e-05,
      "loss": 1.958,
      "step": 200315
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.4599202871322632,
      "learning_rate": 2.4892757705726563e-05,
      "loss": 1.7989,
      "step": 200316
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.553199291229248,
      "learning_rate": 2.4891126285435837e-05,
      "loss": 1.9833,
      "step": 200317
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.7230616807937622,
      "learning_rate": 2.4889494916292995e-05,
      "loss": 2.0518,
      "step": 200318
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.5436264276504517,
      "learning_rate": 2.488786359829844e-05,
      "loss": 1.9027,
      "step": 200319
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.405762791633606,
      "learning_rate": 2.488623233145237e-05,
      "loss": 2.2081,
      "step": 200320
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.5121954679489136,
      "learning_rate": 2.488460111575519e-05,
      "loss": 1.9148,
      "step": 200321
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.4288156032562256,
      "learning_rate": 2.4882969951207155e-05,
      "loss": 1.7911,
      "step": 200322
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.608563780784607,
      "learning_rate": 2.4881338837808575e-05,
      "loss": 2.1212,
      "step": 200323
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.33146071434021,
      "learning_rate": 2.4879707775559708e-05,
      "loss": 1.8357,
      "step": 200324
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.4113609790802002,
      "learning_rate": 2.487807676446093e-05,
      "loss": 2.1851,
      "step": 200325
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.3877084255218506,
      "learning_rate": 2.4876445804512467e-05,
      "loss": 2.1172,
      "step": 200326
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.3700683116912842,
      "learning_rate": 2.4874814895714722e-05,
      "loss": 1.874,
      "step": 200327
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.5723737478256226,
      "learning_rate": 2.4873184038067962e-05,
      "loss": 1.9608,
      "step": 200328
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.343675971031189,
      "learning_rate": 2.4871553231572417e-05,
      "loss": 2.0721,
      "step": 200329
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.3600170612335205,
      "learning_rate": 2.4869922476228422e-05,
      "loss": 2.0821,
      "step": 200330
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.367268681526184,
      "learning_rate": 2.4868291772036375e-05,
      "loss": 1.8545,
      "step": 200331
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.3488973379135132,
      "learning_rate": 2.4866661118996413e-05,
      "loss": 1.826,
      "step": 200332
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.4156675338745117,
      "learning_rate": 2.486503051710903e-05,
      "loss": 2.1196,
      "step": 200333
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.4406360387802124,
      "learning_rate": 2.4863399966374398e-05,
      "loss": 1.9397,
      "step": 200334
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.5084151029586792,
      "learning_rate": 2.4861769466792845e-05,
      "loss": 2.0006,
      "step": 200335
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.3196306228637695,
      "learning_rate": 2.4860139018364643e-05,
      "loss": 1.8007,
      "step": 200336
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.383994221687317,
      "learning_rate": 2.485850862109019e-05,
      "loss": 2.2571,
      "step": 200337
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.6544132232666016,
      "learning_rate": 2.4856878274969682e-05,
      "loss": 2.074,
      "step": 200338
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.3906983137130737,
      "learning_rate": 2.4855247980003558e-05,
      "loss": 1.7439,
      "step": 200339
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.5142260789871216,
      "learning_rate": 2.4853617736191976e-05,
      "loss": 1.9298,
      "step": 200340
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.4693011045455933,
      "learning_rate": 2.4851987543535346e-05,
      "loss": 2.1444,
      "step": 200341
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.4723467826843262,
      "learning_rate": 2.485035740203386e-05,
      "loss": 1.9442,
      "step": 200342
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.514629602432251,
      "learning_rate": 2.4848727311687923e-05,
      "loss": 2.0149,
      "step": 200343
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.4245507717132568,
      "learning_rate": 2.4847097272497763e-05,
      "loss": 2.0684,
      "step": 200344
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.353061318397522,
      "learning_rate": 2.4845467284463782e-05,
      "loss": 1.9449,
      "step": 200345
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.409453272819519,
      "learning_rate": 2.4843837347586216e-05,
      "loss": 2.0831,
      "step": 200346
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.326230525970459,
      "learning_rate": 2.4842207461865358e-05,
      "loss": 1.9499,
      "step": 200347
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.5507407188415527,
      "learning_rate": 2.4840577627301482e-05,
      "loss": 1.831,
      "step": 200348
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.357271671295166,
      "learning_rate": 2.4838947843895018e-05,
      "loss": 1.9705,
      "step": 200349
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.4167256355285645,
      "learning_rate": 2.4837318111646098e-05,
      "loss": 2.16,
      "step": 200350
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.347960352897644,
      "learning_rate": 2.483568843055519e-05,
      "loss": 1.8698,
      "step": 200351
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.5143605470657349,
      "learning_rate": 2.4834058800622526e-05,
      "loss": 2.0662,
      "step": 200352
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.6235531568527222,
      "learning_rate": 2.483242922184837e-05,
      "loss": 2.1025,
      "step": 200353
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.4083346128463745,
      "learning_rate": 2.4830799694233062e-05,
      "loss": 1.9839,
      "step": 200354
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.5142556428909302,
      "learning_rate": 2.4829170217776894e-05,
      "loss": 2.0697,
      "step": 200355
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.4090148210525513,
      "learning_rate": 2.482754079248017e-05,
      "loss": 1.7339,
      "step": 200356
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.3557289838790894,
      "learning_rate": 2.4825911418343226e-05,
      "loss": 1.9543,
      "step": 200357
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.5365869998931885,
      "learning_rate": 2.4824282095366355e-05,
      "loss": 1.8642,
      "step": 200358
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.421922206878662,
      "learning_rate": 2.4822652823549827e-05,
      "loss": 2.1407,
      "step": 200359
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.4382601976394653,
      "learning_rate": 2.482102360289394e-05,
      "loss": 1.9477,
      "step": 200360
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.3803865909576416,
      "learning_rate": 2.4819394433399063e-05,
      "loss": 1.9731,
      "step": 200361
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.4281299114227295,
      "learning_rate": 2.4817765315065396e-05,
      "loss": 1.9969,
      "step": 200362
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.5394353866577148,
      "learning_rate": 2.4816136247893336e-05,
      "loss": 2.0769,
      "step": 200363
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.3863615989685059,
      "learning_rate": 2.481450723188315e-05,
      "loss": 1.9608,
      "step": 200364
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.429810881614685,
      "learning_rate": 2.4812878267035106e-05,
      "loss": 2.1526,
      "step": 200365
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.5550777912139893,
      "learning_rate": 2.4811249353349604e-05,
      "loss": 2.0959,
      "step": 200366
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.2875823974609375,
      "learning_rate": 2.4809620490826877e-05,
      "loss": 1.9777,
      "step": 200367
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.5311365127563477,
      "learning_rate": 2.4807991679467186e-05,
      "loss": 1.8735,
      "step": 200368
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.4825822114944458,
      "learning_rate": 2.480636291927094e-05,
      "loss": 1.9106,
      "step": 200369
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.708240032196045,
      "learning_rate": 2.4804734210238364e-05,
      "loss": 2.0417,
      "step": 200370
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.5199580192565918,
      "learning_rate": 2.480310555236976e-05,
      "loss": 2.0338,
      "step": 200371
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.5091935396194458,
      "learning_rate": 2.4801476945665466e-05,
      "loss": 2.0128,
      "step": 200372
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.6943844556808472,
      "learning_rate": 2.479984839012581e-05,
      "loss": 1.9633,
      "step": 200373
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.4243113994598389,
      "learning_rate": 2.479821988575099e-05,
      "loss": 1.8805,
      "step": 200374
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.308613896369934,
      "learning_rate": 2.4796591432541445e-05,
      "loss": 2.0672,
      "step": 200375
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.373539686203003,
      "learning_rate": 2.479496303049737e-05,
      "loss": 1.92,
      "step": 200376
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.4815032482147217,
      "learning_rate": 2.479333467961907e-05,
      "loss": 1.9875,
      "step": 200377
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.410493016242981,
      "learning_rate": 2.4791706379906974e-05,
      "loss": 2.005,
      "step": 200378
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.3388413190841675,
      "learning_rate": 2.4790078131361212e-05,
      "loss": 1.9883,
      "step": 200379
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.399157166481018,
      "learning_rate": 2.478844993398219e-05,
      "loss": 2.0603,
      "step": 200380
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.422649621963501,
      "learning_rate": 2.478682178777024e-05,
      "loss": 2.0466,
      "step": 200381
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.8726074695587158,
      "learning_rate": 2.4785193692725525e-05,
      "loss": 1.9922,
      "step": 200382
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.67178213596344,
      "learning_rate": 2.4783565648848514e-05,
      "loss": 2.1118,
      "step": 200383
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.487837314605713,
      "learning_rate": 2.4781937656139438e-05,
      "loss": 2.049,
      "step": 200384
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.3737543821334839,
      "learning_rate": 2.4780309714598535e-05,
      "loss": 2.1934,
      "step": 200385
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.4164268970489502,
      "learning_rate": 2.477868182422623e-05,
      "loss": 1.9217,
      "step": 200386
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.4256153106689453,
      "learning_rate": 2.47770539850227e-05,
      "loss": 1.821,
      "step": 200387
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.4179919958114624,
      "learning_rate": 2.4775426196988335e-05,
      "loss": 2.0466,
      "step": 200388
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.3694772720336914,
      "learning_rate": 2.477379846012344e-05,
      "loss": 1.8641,
      "step": 200389
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.5144538879394531,
      "learning_rate": 2.4772170774428313e-05,
      "loss": 2.0477,
      "step": 200390
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.5888279676437378,
      "learning_rate": 2.4770543139903155e-05,
      "loss": 1.9684,
      "step": 200391
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.2966305017471313,
      "learning_rate": 2.4768915556548396e-05,
      "loss": 1.9045,
      "step": 200392
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.3137718439102173,
      "learning_rate": 2.4767288024364242e-05,
      "loss": 1.687,
      "step": 200393
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.4109543561935425,
      "learning_rate": 2.4765660543351117e-05,
      "loss": 1.9722,
      "step": 200394
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.4604029655456543,
      "learning_rate": 2.476403311350923e-05,
      "loss": 2.0888,
      "step": 200395
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.457542896270752,
      "learning_rate": 2.476240573483891e-05,
      "loss": 2.0303,
      "step": 200396
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.4107519388198853,
      "learning_rate": 2.476077840734042e-05,
      "loss": 1.9594,
      "step": 200397
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.734300136566162,
      "learning_rate": 2.475915113101413e-05,
      "loss": 2.0558,
      "step": 200398
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.3774380683898926,
      "learning_rate": 2.4757523905860277e-05,
      "loss": 2.0657,
      "step": 200399
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.3103201389312744,
      "learning_rate": 2.475589673187922e-05,
      "loss": 2.0227,
      "step": 200400
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.4109668731689453,
      "learning_rate": 2.4754269609071265e-05,
      "loss": 1.9539,
      "step": 200401
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.3625355958938599,
      "learning_rate": 2.475264253743664e-05,
      "loss": 2.1045,
      "step": 200402
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.3432915210723877,
      "learning_rate": 2.4751015516975682e-05,
      "loss": 1.7262,
      "step": 200403
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.5333940982818604,
      "learning_rate": 2.4749388547688754e-05,
      "loss": 1.9556,
      "step": 200404
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.4385267496109009,
      "learning_rate": 2.4747761629576056e-05,
      "loss": 2.0317,
      "step": 200405
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.4831109046936035,
      "learning_rate": 2.4746134762637993e-05,
      "loss": 2.0682,
      "step": 200406
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.589218258857727,
      "learning_rate": 2.4744507946874824e-05,
      "loss": 2.0789,
      "step": 200407
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.8132163286209106,
      "learning_rate": 2.4742881182286855e-05,
      "loss": 1.9959,
      "step": 200408
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.3014675378799438,
      "learning_rate": 2.4741254468874316e-05,
      "loss": 2.1522,
      "step": 200409
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.4446405172348022,
      "learning_rate": 2.4739627806637607e-05,
      "loss": 2.0526,
      "step": 200410
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.6956708431243896,
      "learning_rate": 2.4738001195576996e-05,
      "loss": 2.0303,
      "step": 200411
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.4723505973815918,
      "learning_rate": 2.473637463569278e-05,
      "loss": 1.8461,
      "step": 200412
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.4526523351669312,
      "learning_rate": 2.4734748126985292e-05,
      "loss": 2.2429,
      "step": 200413
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.452319622039795,
      "learning_rate": 2.4733121669454837e-05,
      "loss": 1.9576,
      "step": 200414
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.3166863918304443,
      "learning_rate": 2.4731495263101608e-05,
      "loss": 1.8244,
      "step": 200415
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.4686447381973267,
      "learning_rate": 2.4729868907926042e-05,
      "loss": 2.0345,
      "step": 200416
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.4252625703811646,
      "learning_rate": 2.4728242603928372e-05,
      "loss": 1.8742,
      "step": 200417
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.4144362211227417,
      "learning_rate": 2.472661635110893e-05,
      "loss": 1.9886,
      "step": 200418
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.413838267326355,
      "learning_rate": 2.4724990149468017e-05,
      "loss": 1.9801,
      "step": 200419
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.4167389869689941,
      "learning_rate": 2.4723363999005928e-05,
      "loss": 1.9345,
      "step": 200420
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.5324218273162842,
      "learning_rate": 2.4721737899722905e-05,
      "loss": 2.2095,
      "step": 200421
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.498095154762268,
      "learning_rate": 2.4720111851619374e-05,
      "loss": 1.7138,
      "step": 200422
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.5433908700942993,
      "learning_rate": 2.47184858546955e-05,
      "loss": 1.9344,
      "step": 200423
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.403042197227478,
      "learning_rate": 2.4716859908951725e-05,
      "loss": 1.9808,
      "step": 200424
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.5153313875198364,
      "learning_rate": 2.4715234014388274e-05,
      "loss": 1.9612,
      "step": 200425
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.5189688205718994,
      "learning_rate": 2.471360817100545e-05,
      "loss": 1.9606,
      "step": 200426
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.4715098142623901,
      "learning_rate": 2.471198237880352e-05,
      "loss": 2.2222,
      "step": 200427
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.424851655960083,
      "learning_rate": 2.471035663778288e-05,
      "loss": 2.0405,
      "step": 200428
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.3246023654937744,
      "learning_rate": 2.4708730947943734e-05,
      "loss": 2.0294,
      "step": 200429
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.463841438293457,
      "learning_rate": 2.4707105309286477e-05,
      "loss": 2.0628,
      "step": 200430
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.360931158065796,
      "learning_rate": 2.470547972181135e-05,
      "loss": 1.9128,
      "step": 200431
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.398971676826477,
      "learning_rate": 2.4703854185518677e-05,
      "loss": 1.9051,
      "step": 200432
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.4503434896469116,
      "learning_rate": 2.4702228700408734e-05,
      "loss": 1.9302,
      "step": 200433
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.4208849668502808,
      "learning_rate": 2.4700603266481878e-05,
      "loss": 1.8407,
      "step": 200434
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.4351824522018433,
      "learning_rate": 2.4698977883738313e-05,
      "loss": 1.7674,
      "step": 200435
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.4623217582702637,
      "learning_rate": 2.469735255217844e-05,
      "loss": 2.0004,
      "step": 200436
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.4003647565841675,
      "learning_rate": 2.4695727271802558e-05,
      "loss": 1.7559,
      "step": 200437
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.3499408960342407,
      "learning_rate": 2.469410204261093e-05,
      "loss": 1.9387,
      "step": 200438
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.5945522785186768,
      "learning_rate": 2.4692476864603828e-05,
      "loss": 2.2432,
      "step": 200439
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.4195080995559692,
      "learning_rate": 2.4690851737781615e-05,
      "loss": 2.1976,
      "step": 200440
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.430618405342102,
      "learning_rate": 2.4689226662144522e-05,
      "loss": 1.9788,
      "step": 200441
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.4368348121643066,
      "learning_rate": 2.4687601637692956e-05,
      "loss": 2.0508,
      "step": 200442
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.6835342645645142,
      "learning_rate": 2.4685976664427177e-05,
      "loss": 2.0161,
      "step": 200443
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.3754969835281372,
      "learning_rate": 2.4684351742347454e-05,
      "loss": 1.9434,
      "step": 200444
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.4720277786254883,
      "learning_rate": 2.468272687145405e-05,
      "loss": 1.9662,
      "step": 200445
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.4558411836624146,
      "learning_rate": 2.4681102051747403e-05,
      "loss": 1.9369,
      "step": 200446
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.4832255840301514,
      "learning_rate": 2.4679477283227645e-05,
      "loss": 2.0747,
      "step": 200447
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.3411139249801636,
      "learning_rate": 2.467785256589524e-05,
      "loss": 2.0956,
      "step": 200448
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.607740879058838,
      "learning_rate": 2.467622789975039e-05,
      "loss": 2.1923,
      "step": 200449
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.4046322107315063,
      "learning_rate": 2.467460328479346e-05,
      "loss": 2.0324,
      "step": 200450
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.5295754671096802,
      "learning_rate": 2.4672978721024717e-05,
      "loss": 1.9061,
      "step": 200451
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.5737885236740112,
      "learning_rate": 2.467135420844446e-05,
      "loss": 2.1757,
      "step": 200452
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.4043171405792236,
      "learning_rate": 2.4669729747052957e-05,
      "loss": 1.9054,
      "step": 200453
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.570489525794983,
      "learning_rate": 2.4668105336850575e-05,
      "loss": 1.9431,
      "step": 200454
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.435518741607666,
      "learning_rate": 2.4666480977837576e-05,
      "loss": 1.8992,
      "step": 200455
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.4214253425598145,
      "learning_rate": 2.4664856670014298e-05,
      "loss": 2.0878,
      "step": 200456
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.4248663187026978,
      "learning_rate": 2.466323241338101e-05,
      "loss": 2.1072,
      "step": 200457
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.414893388748169,
      "learning_rate": 2.4661608207938067e-05,
      "loss": 1.9773,
      "step": 200458
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.474961280822754,
      "learning_rate": 2.4659984053685643e-05,
      "loss": 2.0584,
      "step": 200459
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.5865545272827148,
      "learning_rate": 2.4658359950624174e-05,
      "loss": 2.1252,
      "step": 200460
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.4956721067428589,
      "learning_rate": 2.4656735898753887e-05,
      "loss": 2.1478,
      "step": 200461
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.4754878282546997,
      "learning_rate": 2.4655111898075153e-05,
      "loss": 1.973,
      "step": 200462
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.536028265953064,
      "learning_rate": 2.46534879485882e-05,
      "loss": 1.9451,
      "step": 200463
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.3909955024719238,
      "learning_rate": 2.465186405029337e-05,
      "loss": 1.8601,
      "step": 200464
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.4325153827667236,
      "learning_rate": 2.4650240203190952e-05,
      "loss": 2.2079,
      "step": 200465
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.4137699604034424,
      "learning_rate": 2.4648616407281285e-05,
      "loss": 2.1265,
      "step": 200466
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.5077959299087524,
      "learning_rate": 2.4646992662564568e-05,
      "loss": 2.1023,
      "step": 200467
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.5188169479370117,
      "learning_rate": 2.4645368969041235e-05,
      "loss": 1.8746,
      "step": 200468
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.4323508739471436,
      "learning_rate": 2.4643745326711517e-05,
      "loss": 1.8828,
      "step": 200469
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.5810459852218628,
      "learning_rate": 2.4642121735575683e-05,
      "loss": 2.1829,
      "step": 200470
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.4864380359649658,
      "learning_rate": 2.4640498195634128e-05,
      "loss": 2.094,
      "step": 200471
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.6009695529937744,
      "learning_rate": 2.4638874706887092e-05,
      "loss": 1.9027,
      "step": 200472
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.4726667404174805,
      "learning_rate": 2.4637251269334835e-05,
      "loss": 2.0087,
      "step": 200473
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.5020064115524292,
      "learning_rate": 2.4635627882977792e-05,
      "loss": 2.1715,
      "step": 200474
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.3960273265838623,
      "learning_rate": 2.4634004547816132e-05,
      "loss": 2.1043,
      "step": 200475
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.3944141864776611,
      "learning_rate": 2.4632381263850187e-05,
      "loss": 2.216,
      "step": 200476
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.9778887033462524,
      "learning_rate": 2.4630758031080356e-05,
      "loss": 1.99,
      "step": 200477
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.5583840608596802,
      "learning_rate": 2.4629134849506772e-05,
      "loss": 1.8405,
      "step": 200478
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.5213193893432617,
      "learning_rate": 2.46275117191299e-05,
      "loss": 1.883,
      "step": 200479
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.5752850770950317,
      "learning_rate": 2.4625888639949975e-05,
      "loss": 2.0765,
      "step": 200480
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.7400356531143188,
      "learning_rate": 2.4624265611967296e-05,
      "loss": 2.1647,
      "step": 200481
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.5421159267425537,
      "learning_rate": 2.4622642635182098e-05,
      "loss": 2.1036,
      "step": 200482
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.3997877836227417,
      "learning_rate": 2.4621019709594813e-05,
      "loss": 2.0766,
      "step": 200483
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.5671721696853638,
      "learning_rate": 2.461939683520561e-05,
      "loss": 1.9827,
      "step": 200484
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.5293612480163574,
      "learning_rate": 2.4617774012014914e-05,
      "loss": 2.0642,
      "step": 200485
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.3945393562316895,
      "learning_rate": 2.4616151240022997e-05,
      "loss": 1.9841,
      "step": 200486
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.6454261541366577,
      "learning_rate": 2.4614528519230125e-05,
      "loss": 2.1241,
      "step": 200487
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.4648517370224,
      "learning_rate": 2.4612905849636533e-05,
      "loss": 2.0136,
      "step": 200488
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.4248251914978027,
      "learning_rate": 2.4611283231242684e-05,
      "loss": 2.227,
      "step": 200489
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.48532235622406,
      "learning_rate": 2.4609660664048714e-05,
      "loss": 2.1614,
      "step": 200490
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.587173342704773,
      "learning_rate": 2.4608038148055087e-05,
      "loss": 1.8997,
      "step": 200491
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.3149288892745972,
      "learning_rate": 2.4606415683262003e-05,
      "loss": 2.0625,
      "step": 200492
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.557596206665039,
      "learning_rate": 2.4604793269669798e-05,
      "loss": 1.9973,
      "step": 200493
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.4728225469589233,
      "learning_rate": 2.4603170907278735e-05,
      "loss": 1.7617,
      "step": 200494
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.3024202585220337,
      "learning_rate": 2.460154859608915e-05,
      "loss": 1.8277,
      "step": 200495
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.5063002109527588,
      "learning_rate": 2.4599926336101304e-05,
      "loss": 2.1043,
      "step": 200496
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.4120219945907593,
      "learning_rate": 2.459830412731557e-05,
      "loss": 1.948,
      "step": 200497
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.6416393518447876,
      "learning_rate": 2.459668196973221e-05,
      "loss": 2.1674,
      "step": 200498
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.4692760705947876,
      "learning_rate": 2.459505986335153e-05,
      "loss": 2.0257,
      "step": 200499
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.4550226926803589,
      "learning_rate": 2.4593437808173788e-05,
      "loss": 2.003,
      "step": 200500
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.7955957651138306,
      "learning_rate": 2.4591815804199357e-05,
      "loss": 1.9638,
      "step": 200501
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.4183202981948853,
      "learning_rate": 2.4590193851428463e-05,
      "loss": 2.0704,
      "step": 200502
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.4078280925750732,
      "learning_rate": 2.458857194986151e-05,
      "loss": 1.9542,
      "step": 200503
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.4295247793197632,
      "learning_rate": 2.4586950099498738e-05,
      "loss": 2.0377,
      "step": 200504
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.5683612823486328,
      "learning_rate": 2.4585328300340434e-05,
      "loss": 2.0173,
      "step": 200505
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.7035210132598877,
      "learning_rate": 2.458370655238687e-05,
      "loss": 2.0334,
      "step": 200506
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.3693442344665527,
      "learning_rate": 2.4582084855638452e-05,
      "loss": 1.8322,
      "step": 200507
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.3899767398834229,
      "learning_rate": 2.458046321009537e-05,
      "loss": 1.9295,
      "step": 200508
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.4437977075576782,
      "learning_rate": 2.4578841615758026e-05,
      "loss": 1.8355,
      "step": 200509
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.4760428667068481,
      "learning_rate": 2.4577220072626625e-05,
      "loss": 2.1622,
      "step": 200510
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.4662551879882812,
      "learning_rate": 2.4575598580701628e-05,
      "loss": 2.0294,
      "step": 200511
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.4367765188217163,
      "learning_rate": 2.4573977139983137e-05,
      "loss": 1.9424,
      "step": 200512
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.4632517099380493,
      "learning_rate": 2.457235575047155e-05,
      "loss": 2.1188,
      "step": 200513
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.3248928785324097,
      "learning_rate": 2.4570734412167138e-05,
      "loss": 2.1737,
      "step": 200514
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.4240756034851074,
      "learning_rate": 2.4569113125070294e-05,
      "loss": 2.2754,
      "step": 200515
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.4602080583572388,
      "learning_rate": 2.4567491889181192e-05,
      "loss": 2.1236,
      "step": 200516
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.3527060747146606,
      "learning_rate": 2.4565870704500256e-05,
      "loss": 1.9819,
      "step": 200517
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.5466876029968262,
      "learning_rate": 2.456424957102766e-05,
      "loss": 2.0819,
      "step": 200518
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.3952867984771729,
      "learning_rate": 2.4562628488763802e-05,
      "loss": 2.0409,
      "step": 200519
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.470301866531372,
      "learning_rate": 2.4561007457708913e-05,
      "loss": 1.969,
      "step": 200520
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.4090209007263184,
      "learning_rate": 2.4559386477863364e-05,
      "loss": 1.9969,
      "step": 200521
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.4471057653427124,
      "learning_rate": 2.4557765549227417e-05,
      "loss": 2.202,
      "step": 200522
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.4300614595413208,
      "learning_rate": 2.4556144671801438e-05,
      "loss": 2.1368,
      "step": 200523
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.409144639968872,
      "learning_rate": 2.4554523845585595e-05,
      "loss": 2.0919,
      "step": 200524
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.385716199874878,
      "learning_rate": 2.455290307058029e-05,
      "loss": 1.9103,
      "step": 200525
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.458405613899231,
      "learning_rate": 2.455128234678575e-05,
      "loss": 1.908,
      "step": 200526
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.405910611152649,
      "learning_rate": 2.4549661674202413e-05,
      "loss": 1.9495,
      "step": 200527
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.4146569967269897,
      "learning_rate": 2.4548041052830416e-05,
      "loss": 1.966,
      "step": 200528
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.4255436658859253,
      "learning_rate": 2.4546420482670247e-05,
      "loss": 1.9557,
      "step": 200529
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.3693439960479736,
      "learning_rate": 2.4544799963721985e-05,
      "loss": 1.9273,
      "step": 200530
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.3214340209960938,
      "learning_rate": 2.454317949598612e-05,
      "loss": 1.8579,
      "step": 200531
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.4104440212249756,
      "learning_rate": 2.454155907946279e-05,
      "loss": 2.2358,
      "step": 200532
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.3639854192733765,
      "learning_rate": 2.4539938714152464e-05,
      "loss": 1.9425,
      "step": 200533
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.6815435886383057,
      "learning_rate": 2.4538318400055302e-05,
      "loss": 2.0587,
      "step": 200534
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.5933853387832642,
      "learning_rate": 2.4536698137171774e-05,
      "loss": 1.817,
      "step": 200535
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.377838134765625,
      "learning_rate": 2.4535077925501946e-05,
      "loss": 2.0199,
      "step": 200536
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.324467658996582,
      "learning_rate": 2.453345776504635e-05,
      "loss": 1.9301,
      "step": 200537
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.377199411392212,
      "learning_rate": 2.4531837655805087e-05,
      "loss": 2.2789,
      "step": 200538
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.4475977420806885,
      "learning_rate": 2.4530217597778623e-05,
      "loss": 2.0207,
      "step": 200539
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.4673222303390503,
      "learning_rate": 2.4528597590967124e-05,
      "loss": 2.0108,
      "step": 200540
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.3822073936462402,
      "learning_rate": 2.4526977635371026e-05,
      "loss": 2.0918,
      "step": 200541
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.3460508584976196,
      "learning_rate": 2.4525357730990558e-05,
      "loss": 2.1024,
      "step": 200542
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.3960553407669067,
      "learning_rate": 2.4523737877826056e-05,
      "loss": 2.0601,
      "step": 200543
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.5544145107269287,
      "learning_rate": 2.4522118075877685e-05,
      "loss": 2.0197,
      "step": 200544
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.5279743671417236,
      "learning_rate": 2.4520498325145944e-05,
      "loss": 1.9637,
      "step": 200545
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.6482036113739014,
      "learning_rate": 2.4518878625630966e-05,
      "loss": 2.2115,
      "step": 200546
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.4382444620132446,
      "learning_rate": 2.4517258977333186e-05,
      "loss": 2.0223,
      "step": 200547
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.362479567527771,
      "learning_rate": 2.4515639380252873e-05,
      "loss": 2.2326,
      "step": 200548
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.442734718322754,
      "learning_rate": 2.451401983439022e-05,
      "loss": 2.0394,
      "step": 200549
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.5112961530685425,
      "learning_rate": 2.4512400339745697e-05,
      "loss": 1.9715,
      "step": 200550
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.4740033149719238,
      "learning_rate": 2.451078089631947e-05,
      "loss": 2.0893,
      "step": 200551
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.4695461988449097,
      "learning_rate": 2.4509161504111874e-05,
      "loss": 2.2076,
      "step": 200552
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.4374524354934692,
      "learning_rate": 2.4507542163123272e-05,
      "loss": 1.9723,
      "step": 200553
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.4250340461730957,
      "learning_rate": 2.4505922873353902e-05,
      "loss": 2.0487,
      "step": 200554
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.5188688039779663,
      "learning_rate": 2.450430363480406e-05,
      "loss": 2.0854,
      "step": 200555
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.4066696166992188,
      "learning_rate": 2.4502684447474107e-05,
      "loss": 1.758,
      "step": 200556
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.4078799486160278,
      "learning_rate": 2.4501065311364286e-05,
      "loss": 2.0376,
      "step": 200557
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.582528829574585,
      "learning_rate": 2.4499446226474862e-05,
      "loss": 1.8884,
      "step": 200558
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.606237769126892,
      "learning_rate": 2.4497827192806263e-05,
      "loss": 1.8922,
      "step": 200559
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.3101240396499634,
      "learning_rate": 2.4496208210358725e-05,
      "loss": 1.9054,
      "step": 200560
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.3939154148101807,
      "learning_rate": 2.4494589279132482e-05,
      "loss": 2.0339,
      "step": 200561
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.519824504852295,
      "learning_rate": 2.4492970399127964e-05,
      "loss": 1.951,
      "step": 200562
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.3034948110580444,
      "learning_rate": 2.449135157034531e-05,
      "loss": 1.8829,
      "step": 200563
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.5939329862594604,
      "learning_rate": 2.448973279278501e-05,
      "loss": 2.0096,
      "step": 200564
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.4881287813186646,
      "learning_rate": 2.448811406644724e-05,
      "loss": 2.0739,
      "step": 200565
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.4279707670211792,
      "learning_rate": 2.4486495391332328e-05,
      "loss": 1.9868,
      "step": 200566
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.3910185098648071,
      "learning_rate": 2.4484876767440542e-05,
      "loss": 1.9271,
      "step": 200567
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.4699240922927856,
      "learning_rate": 2.448325819477228e-05,
      "loss": 2.0294,
      "step": 200568
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.5128886699676514,
      "learning_rate": 2.4481639673327713e-05,
      "loss": 1.8783,
      "step": 200569
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.3736661672592163,
      "learning_rate": 2.448002120310727e-05,
      "loss": 2.0531,
      "step": 200570
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.4532347917556763,
      "learning_rate": 2.4478402784111184e-05,
      "loss": 1.9914,
      "step": 200571
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.6123535633087158,
      "learning_rate": 2.4476784416339755e-05,
      "loss": 2.1603,
      "step": 200572
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.4435875415802002,
      "learning_rate": 2.4475166099793254e-05,
      "loss": 2.2331,
      "step": 200573
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.270673155784607,
      "learning_rate": 2.4473547834472073e-05,
      "loss": 1.9452,
      "step": 200574
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.5313116312026978,
      "learning_rate": 2.4471929620376385e-05,
      "loss": 1.8899,
      "step": 200575
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.6020057201385498,
      "learning_rate": 2.4470311457506654e-05,
      "loss": 1.9655,
      "step": 200576
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.407188057899475,
      "learning_rate": 2.446869334586301e-05,
      "loss": 2.0811,
      "step": 200577
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.535065770149231,
      "learning_rate": 2.446707528544596e-05,
      "loss": 2.1992,
      "step": 200578
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.39137601852417,
      "learning_rate": 2.446545727625556e-05,
      "loss": 1.8298,
      "step": 200579
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.4635303020477295,
      "learning_rate": 2.4463839318292323e-05,
      "loss": 1.8654,
      "step": 200580
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.3007820844650269,
      "learning_rate": 2.446222141155637e-05,
      "loss": 2.1509,
      "step": 200581
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.478203296661377,
      "learning_rate": 2.4460603556048174e-05,
      "loss": 1.9568,
      "step": 200582
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.327561616897583,
      "learning_rate": 2.4458985751767866e-05,
      "loss": 1.8802,
      "step": 200583
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.302108645439148,
      "learning_rate": 2.445736799871595e-05,
      "loss": 2.0601,
      "step": 200584
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.491956353187561,
      "learning_rate": 2.445575029689252e-05,
      "loss": 2.1564,
      "step": 200585
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.3974677324295044,
      "learning_rate": 2.445413264629801e-05,
      "loss": 1.7872,
      "step": 200586
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.5279109477996826,
      "learning_rate": 2.445251504693262e-05,
      "loss": 1.9789,
      "step": 200587
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.4899688959121704,
      "learning_rate": 2.4450897498796783e-05,
      "loss": 2.0078,
      "step": 200588
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.617055058479309,
      "learning_rate": 2.4449280001890637e-05,
      "loss": 1.9923,
      "step": 200589
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.4368568658828735,
      "learning_rate": 2.4447662556214707e-05,
      "loss": 1.9287,
      "step": 200590
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.499445915222168,
      "learning_rate": 2.4446045161769035e-05,
      "loss": 1.9256,
      "step": 200591
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.3897545337677002,
      "learning_rate": 2.4444427818554114e-05,
      "loss": 2.0895,
      "step": 200592
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.3232574462890625,
      "learning_rate": 2.444281052657011e-05,
      "loss": 1.9603,
      "step": 200593
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.3020076751708984,
      "learning_rate": 2.444119328581746e-05,
      "loss": 1.8071,
      "step": 200594
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.483418583869934,
      "learning_rate": 2.4439576096296333e-05,
      "loss": 2.116,
      "step": 200595
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.5909318923950195,
      "learning_rate": 2.4437958958007186e-05,
      "loss": 2.1072,
      "step": 200596
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.3535616397857666,
      "learning_rate": 2.4436341870950126e-05,
      "loss": 2.0377,
      "step": 200597
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.3467271327972412,
      "learning_rate": 2.4434724835125586e-05,
      "loss": 2.0503,
      "step": 200598
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.39115571975708,
      "learning_rate": 2.4433107850533796e-05,
      "loss": 2.041,
      "step": 200599
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.5271705389022827,
      "learning_rate": 2.4431490917175157e-05,
      "loss": 1.9776,
      "step": 200600
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.5055263042449951,
      "learning_rate": 2.4429874035049834e-05,
      "loss": 2.0468,
      "step": 200601
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.3616236448287964,
      "learning_rate": 2.44282572041583e-05,
      "loss": 1.8327,
      "step": 200602
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.4475655555725098,
      "learning_rate": 2.4426640424500675e-05,
      "loss": 1.8976,
      "step": 200603
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.5012214183807373,
      "learning_rate": 2.4425023696077372e-05,
      "loss": 1.9628,
      "step": 200604
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.3931560516357422,
      "learning_rate": 2.4423407018888618e-05,
      "loss": 1.9195,
      "step": 200605
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.5381327867507935,
      "learning_rate": 2.4421790392934783e-05,
      "loss": 1.9991,
      "step": 200606
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.4717636108398438,
      "learning_rate": 2.4420173818216127e-05,
      "loss": 1.9151,
      "step": 200607
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.4052320718765259,
      "learning_rate": 2.441855729473302e-05,
      "loss": 1.8453,
      "step": 200608
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.4540884494781494,
      "learning_rate": 2.441694082248563e-05,
      "loss": 2.2276,
      "step": 200609
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.4152965545654297,
      "learning_rate": 2.4415324401474388e-05,
      "loss": 2.0333,
      "step": 200610
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.4877378940582275,
      "learning_rate": 2.4413708031699464e-05,
      "loss": 1.992,
      "step": 200611
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.3442145586013794,
      "learning_rate": 2.441209171316132e-05,
      "loss": 1.9914,
      "step": 200612
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.4468227624893188,
      "learning_rate": 2.441047544586009e-05,
      "loss": 1.9288,
      "step": 200613
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.4712862968444824,
      "learning_rate": 2.4408859229796275e-05,
      "loss": 2.2275,
      "step": 200614
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.3494648933410645,
      "learning_rate": 2.4407243064969938e-05,
      "loss": 1.9816,
      "step": 200615
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.4025757312774658,
      "learning_rate": 2.4405626951381552e-05,
      "loss": 1.9276,
      "step": 200616
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.3204445838928223,
      "learning_rate": 2.440401088903131e-05,
      "loss": 2.0357,
      "step": 200617
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.3630485534667969,
      "learning_rate": 2.4402394877919616e-05,
      "loss": 1.8746,
      "step": 200618
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.4224802255630493,
      "learning_rate": 2.440077891804667e-05,
      "loss": 1.934,
      "step": 200619
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.3212873935699463,
      "learning_rate": 2.4399163009412935e-05,
      "loss": 1.9159,
      "step": 200620
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.4396471977233887,
      "learning_rate": 2.439754715201848e-05,
      "loss": 2.0913,
      "step": 200621
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.4341998100280762,
      "learning_rate": 2.4395931345863806e-05,
      "loss": 2.1655,
      "step": 200622
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.3872405290603638,
      "learning_rate": 2.439431559094904e-05,
      "loss": 1.9418,
      "step": 200623
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.3905525207519531,
      "learning_rate": 2.4392699887274624e-05,
      "loss": 1.9244,
      "step": 200624
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.5232739448547363,
      "learning_rate": 2.4391084234840786e-05,
      "loss": 1.9193,
      "step": 200625
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.483968734741211,
      "learning_rate": 2.4389468633647892e-05,
      "loss": 1.8565,
      "step": 200626
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.3494677543640137,
      "learning_rate": 2.4387853083696175e-05,
      "loss": 1.9156,
      "step": 200627
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.529228925704956,
      "learning_rate": 2.438623758498597e-05,
      "loss": 2.0081,
      "step": 200628
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.4362037181854248,
      "learning_rate": 2.438462213751754e-05,
      "loss": 1.9758,
      "step": 200629
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.392405390739441,
      "learning_rate": 2.438300674129122e-05,
      "loss": 2.1245,
      "step": 200630
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.6086206436157227,
      "learning_rate": 2.438139139630728e-05,
      "loss": 2.0571,
      "step": 200631
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.4837032556533813,
      "learning_rate": 2.4379776102566086e-05,
      "loss": 2.0669,
      "step": 200632
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.3325084447860718,
      "learning_rate": 2.4378160860067897e-05,
      "loss": 1.7312,
      "step": 200633
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.5152052640914917,
      "learning_rate": 2.4376545668812952e-05,
      "loss": 2.1749,
      "step": 200634
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.3646408319473267,
      "learning_rate": 2.4374930528801684e-05,
      "loss": 2.1734,
      "step": 200635
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.5340343713760376,
      "learning_rate": 2.437331544003429e-05,
      "loss": 2.0352,
      "step": 200636
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.4528775215148926,
      "learning_rate": 2.437170040251104e-05,
      "loss": 2.1102,
      "step": 200637
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.470292568206787,
      "learning_rate": 2.4370085416232364e-05,
      "loss": 2.0271,
      "step": 200638
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.2849191427230835,
      "learning_rate": 2.43684704811985e-05,
      "loss": 1.9006,
      "step": 200639
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.3465887308120728,
      "learning_rate": 2.4366855597409674e-05,
      "loss": 2.1635,
      "step": 200640
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.4158905744552612,
      "learning_rate": 2.4365240764866324e-05,
      "loss": 1.9418,
      "step": 200641
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.3912701606750488,
      "learning_rate": 2.436362598356868e-05,
      "loss": 2.0737,
      "step": 200642
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.4942830801010132,
      "learning_rate": 2.4362011253516976e-05,
      "loss": 1.9069,
      "step": 200643
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.426727294921875,
      "learning_rate": 2.4360396574711615e-05,
      "loss": 2.0661,
      "step": 200644
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.3935028314590454,
      "learning_rate": 2.435878194715286e-05,
      "loss": 2.0605,
      "step": 200645
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.4704509973526,
      "learning_rate": 2.435716737084098e-05,
      "loss": 2.0956,
      "step": 200646
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.4909553527832031,
      "learning_rate": 2.435555284577634e-05,
      "loss": 2.0085,
      "step": 200647
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.5137072801589966,
      "learning_rate": 2.435393837195917e-05,
      "loss": 2.096,
      "step": 200648
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.7463358640670776,
      "learning_rate": 2.4352323949389842e-05,
      "loss": 1.8237,
      "step": 200649
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.329314112663269,
      "learning_rate": 2.435070957806865e-05,
      "loss": 1.8551,
      "step": 200650
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.4833652973175049,
      "learning_rate": 2.434909525799583e-05,
      "loss": 2.236,
      "step": 200651
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.3914028406143188,
      "learning_rate": 2.4347480989171685e-05,
      "loss": 1.7935,
      "step": 200652
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.5116822719573975,
      "learning_rate": 2.434586677159658e-05,
      "loss": 1.9948,
      "step": 200653
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.3952620029449463,
      "learning_rate": 2.4344252605270742e-05,
      "loss": 1.7889,
      "step": 200654
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.4285138845443726,
      "learning_rate": 2.4342638490194575e-05,
      "loss": 2.0878,
      "step": 200655
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.459471583366394,
      "learning_rate": 2.4341024426368283e-05,
      "loss": 1.9529,
      "step": 200656
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.323071837425232,
      "learning_rate": 2.4339410413792227e-05,
      "loss": 2.0148,
      "step": 200657
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.520634412765503,
      "learning_rate": 2.4337796452466608e-05,
      "loss": 1.8668,
      "step": 200658
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.401999592781067,
      "learning_rate": 2.433618254239186e-05,
      "loss": 1.8317,
      "step": 200659
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.3138267993927002,
      "learning_rate": 2.4334568683568177e-05,
      "loss": 1.8191,
      "step": 200660
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.4872440099716187,
      "learning_rate": 2.4332954875995934e-05,
      "loss": 2.052,
      "step": 200661
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.4059087038040161,
      "learning_rate": 2.433134111967536e-05,
      "loss": 1.9113,
      "step": 200662
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.3879332542419434,
      "learning_rate": 2.4329727414606892e-05,
      "loss": 1.9527,
      "step": 200663
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.5437639951705933,
      "learning_rate": 2.4328113760790625e-05,
      "loss": 1.8837,
      "step": 200664
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.4443023204803467,
      "learning_rate": 2.4326500158226993e-05,
      "loss": 2.0251,
      "step": 200665
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.372995138168335,
      "learning_rate": 2.432488660691626e-05,
      "loss": 2.0163,
      "step": 200666
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.3987497091293335,
      "learning_rate": 2.4323273106858764e-05,
      "loss": 2.0209,
      "step": 200667
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.4766868352890015,
      "learning_rate": 2.432165965805474e-05,
      "loss": 1.9599,
      "step": 200668
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.4283252954483032,
      "learning_rate": 2.432004626050461e-05,
      "loss": 1.9933,
      "step": 200669
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.4800511598587036,
      "learning_rate": 2.4318432914208486e-05,
      "loss": 2.0226,
      "step": 200670
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.54990816116333,
      "learning_rate": 2.4316819619166793e-05,
      "loss": 2.1246,
      "step": 200671
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.511210560798645,
      "learning_rate": 2.43152063753798e-05,
      "loss": 1.9551,
      "step": 200672
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.5303051471710205,
      "learning_rate": 2.4313593182847845e-05,
      "loss": 1.922,
      "step": 200673
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.3025684356689453,
      "learning_rate": 2.4311980041571156e-05,
      "loss": 1.7518,
      "step": 200674
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.4137572050094604,
      "learning_rate": 2.4310366951550163e-05,
      "loss": 2.1068,
      "step": 200675
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.616233229637146,
      "learning_rate": 2.4308753912784972e-05,
      "loss": 1.8686,
      "step": 200676
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.520827293395996,
      "learning_rate": 2.430714092527608e-05,
      "loss": 1.7998,
      "step": 200677
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.5131340026855469,
      "learning_rate": 2.4305527989023588e-05,
      "loss": 1.9397,
      "step": 200678
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.2927523851394653,
      "learning_rate": 2.4303915104027993e-05,
      "loss": 1.921,
      "step": 200679
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.3134030103683472,
      "learning_rate": 2.430230227028943e-05,
      "loss": 2.0496,
      "step": 200680
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.4714454412460327,
      "learning_rate": 2.43006894878084e-05,
      "loss": 1.9492,
      "step": 200681
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.6793996095657349,
      "learning_rate": 2.4299076756584933e-05,
      "loss": 1.964,
      "step": 200682
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.3678158521652222,
      "learning_rate": 2.4297464076619565e-05,
      "loss": 2.0296,
      "step": 200683
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.442712426185608,
      "learning_rate": 2.429585144791243e-05,
      "loss": 1.9337,
      "step": 200684
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.4789413213729858,
      "learning_rate": 2.4294238870463955e-05,
      "loss": 2.0421,
      "step": 200685
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.4365298748016357,
      "learning_rate": 2.4292626344274345e-05,
      "loss": 2.0938,
      "step": 200686
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.5273791551589966,
      "learning_rate": 2.4291013869344033e-05,
      "loss": 1.7717,
      "step": 200687
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.437801718711853,
      "learning_rate": 2.428940144567312e-05,
      "loss": 1.9947,
      "step": 200688
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.2943044900894165,
      "learning_rate": 2.4287789073262066e-05,
      "loss": 1.9356,
      "step": 200689
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.4807838201522827,
      "learning_rate": 2.4286176752111077e-05,
      "loss": 2.0348,
      "step": 200690
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.4473270177841187,
      "learning_rate": 2.4284564482220515e-05,
      "loss": 2.1842,
      "step": 200691
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.4809008836746216,
      "learning_rate": 2.4282952263590615e-05,
      "loss": 1.9884,
      "step": 200692
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.4025682210922241,
      "learning_rate": 2.4281340096221845e-05,
      "loss": 1.9617,
      "step": 200693
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.447524070739746,
      "learning_rate": 2.427972798011424e-05,
      "loss": 1.719,
      "step": 200694
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.398873209953308,
      "learning_rate": 2.4278115915268327e-05,
      "loss": 2.0158,
      "step": 200695
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.6270630359649658,
      "learning_rate": 2.427650390168424e-05,
      "loss": 2.1897,
      "step": 200696
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.539575219154358,
      "learning_rate": 2.4274891939362417e-05,
      "loss": 1.8836,
      "step": 200697
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.3627703189849854,
      "learning_rate": 2.4273280028303056e-05,
      "loss": 2.102,
      "step": 200698
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.4432710409164429,
      "learning_rate": 2.4271668168506586e-05,
      "loss": 1.9726,
      "step": 200699
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.3100731372833252,
      "learning_rate": 2.4270056359973112e-05,
      "loss": 1.9656,
      "step": 200700
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.554185152053833,
      "learning_rate": 2.4268444602703095e-05,
      "loss": 1.8998,
      "step": 200701
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.4607765674591064,
      "learning_rate": 2.426683289669671e-05,
      "loss": 1.8295,
      "step": 200702
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.4486520290374756,
      "learning_rate": 2.4265221241954413e-05,
      "loss": 1.9424,
      "step": 200703
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.682201862335205,
      "learning_rate": 2.4263609638476343e-05,
      "loss": 2.0269,
      "step": 200704
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.501700758934021,
      "learning_rate": 2.4261998086262967e-05,
      "loss": 2.1828,
      "step": 200705
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.4168314933776855,
      "learning_rate": 2.4260386585314418e-05,
      "loss": 1.9351,
      "step": 200706
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.2808067798614502,
      "learning_rate": 2.4258775135631094e-05,
      "loss": 1.9104,
      "step": 200707
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.3910362720489502,
      "learning_rate": 2.4257163737213226e-05,
      "loss": 1.9997,
      "step": 200708
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.5352321863174438,
      "learning_rate": 2.4255552390061184e-05,
      "loss": 2.0806,
      "step": 200709
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.3831087350845337,
      "learning_rate": 2.4253941094175234e-05,
      "loss": 2.0161,
      "step": 200710
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.5164225101470947,
      "learning_rate": 2.4252329849555707e-05,
      "loss": 2.0717,
      "step": 200711
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.3722033500671387,
      "learning_rate": 2.4250718656202873e-05,
      "loss": 1.825,
      "step": 200712
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.379227876663208,
      "learning_rate": 2.424910751411703e-05,
      "loss": 2.0666,
      "step": 200713
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.608412504196167,
      "learning_rate": 2.424749642329844e-05,
      "loss": 2.1285,
      "step": 200714
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.4657793045043945,
      "learning_rate": 2.424588538374751e-05,
      "loss": 1.9973,
      "step": 200715
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.4307441711425781,
      "learning_rate": 2.4244274395464402e-05,
      "loss": 2.4122,
      "step": 200716
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.5349124670028687,
      "learning_rate": 2.4242663458449552e-05,
      "loss": 1.9362,
      "step": 200717
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.4167215824127197,
      "learning_rate": 2.4241052572703224e-05,
      "loss": 1.9705,
      "step": 200718
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.548582911491394,
      "learning_rate": 2.4239441738225652e-05,
      "loss": 2.1282,
      "step": 200719
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.4047918319702148,
      "learning_rate": 2.4237830955017135e-05,
      "loss": 2.1604,
      "step": 200720
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.5161043405532837,
      "learning_rate": 2.423622022307804e-05,
      "loss": 2.0572,
      "step": 200721
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.407556176185608,
      "learning_rate": 2.42346095424086e-05,
      "loss": 1.9516,
      "step": 200722
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.4426194429397583,
      "learning_rate": 2.4232998913009215e-05,
      "loss": 2.0353,
      "step": 200723
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.4841417074203491,
      "learning_rate": 2.423138833488012e-05,
      "loss": 1.9466,
      "step": 200724
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.3445841073989868,
      "learning_rate": 2.4229777808021544e-05,
      "loss": 1.9665,
      "step": 200725
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.3845038414001465,
      "learning_rate": 2.4228167332433922e-05,
      "loss": 2.1378,
      "step": 200726
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.4238145351409912,
      "learning_rate": 2.4226556908117524e-05,
      "loss": 1.8931,
      "step": 200727
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.4809798002243042,
      "learning_rate": 2.422494653507251e-05,
      "loss": 2.1274,
      "step": 200728
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.5932703018188477,
      "learning_rate": 2.422333621329935e-05,
      "loss": 2.0816,
      "step": 200729
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.455021858215332,
      "learning_rate": 2.4221725942798276e-05,
      "loss": 1.7525,
      "step": 200730
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.5141727924346924,
      "learning_rate": 2.4220115723569557e-05,
      "loss": 2.0899,
      "step": 200731
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.4829697608947754,
      "learning_rate": 2.4218505555613554e-05,
      "loss": 2.1506,
      "step": 200732
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.3946516513824463,
      "learning_rate": 2.4216895438930506e-05,
      "loss": 1.8097,
      "step": 200733
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.4779645204544067,
      "learning_rate": 2.4215285373520775e-05,
      "loss": 2.1993,
      "step": 200734
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.6021593809127808,
      "learning_rate": 2.421367535938463e-05,
      "loss": 1.8224,
      "step": 200735
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.5770375728607178,
      "learning_rate": 2.4212065396522373e-05,
      "loss": 1.9041,
      "step": 200736
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.4480571746826172,
      "learning_rate": 2.4210455484934265e-05,
      "loss": 1.9927,
      "step": 200737
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.380210280418396,
      "learning_rate": 2.4208845624620677e-05,
      "loss": 1.8767,
      "step": 200738
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.4097236394882202,
      "learning_rate": 2.4207235815581804e-05,
      "loss": 1.9312,
      "step": 200739
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.4399042129516602,
      "learning_rate": 2.4205626057818085e-05,
      "loss": 2.0582,
      "step": 200740
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.488869071006775,
      "learning_rate": 2.4204016351329745e-05,
      "loss": 1.7225,
      "step": 200741
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.3935915231704712,
      "learning_rate": 2.420240669611706e-05,
      "loss": 2.0648,
      "step": 200742
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.559363842010498,
      "learning_rate": 2.4200797092180324e-05,
      "loss": 2.0963,
      "step": 200743
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.5581958293914795,
      "learning_rate": 2.4199187539519904e-05,
      "loss": 1.858,
      "step": 200744
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.5150706768035889,
      "learning_rate": 2.4197578038136036e-05,
      "loss": 2.0282,
      "step": 200745
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.4338934421539307,
      "learning_rate": 2.419596858802908e-05,
      "loss": 2.1383,
      "step": 200746
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.4838371276855469,
      "learning_rate": 2.419435918919924e-05,
      "loss": 2.2592,
      "step": 200747
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.658429741859436,
      "learning_rate": 2.4192749841646987e-05,
      "loss": 2.0881,
      "step": 200748
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.529387354850769,
      "learning_rate": 2.4191140545372377e-05,
      "loss": 2.0042,
      "step": 200749
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.4927932024002075,
      "learning_rate": 2.418953130037592e-05,
      "loss": 1.9483,
      "step": 200750
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.5573365688323975,
      "learning_rate": 2.4187922106657775e-05,
      "loss": 2.0551,
      "step": 200751
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.5816326141357422,
      "learning_rate": 2.418631296421838e-05,
      "loss": 2.0632,
      "step": 200752
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.3282527923583984,
      "learning_rate": 2.4184703873057866e-05,
      "loss": 2.1723,
      "step": 200753
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.3266000747680664,
      "learning_rate": 2.4183094833176765e-05,
      "loss": 1.8235,
      "step": 200754
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.5043866634368896,
      "learning_rate": 2.4181485844575076e-05,
      "loss": 2.0912,
      "step": 200755
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.59244966506958,
      "learning_rate": 2.4179876907253337e-05,
      "loss": 1.8166,
      "step": 200756
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.4596246480941772,
      "learning_rate": 2.4178268021211746e-05,
      "loss": 1.8666,
      "step": 200757
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.544278860092163,
      "learning_rate": 2.417665918645063e-05,
      "loss": 2.1563,
      "step": 200758
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.3722392320632935,
      "learning_rate": 2.417505040297023e-05,
      "loss": 1.7339,
      "step": 200759
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.3631930351257324,
      "learning_rate": 2.417344167077101e-05,
      "loss": 1.8887,
      "step": 200760
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.526087760925293,
      "learning_rate": 2.4171832989853035e-05,
      "loss": 1.6596,
      "step": 200761
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.5268741846084595,
      "learning_rate": 2.4170224360216807e-05,
      "loss": 2.1575,
      "step": 200762
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.3126202821731567,
      "learning_rate": 2.4168615781862455e-05,
      "loss": 1.8298,
      "step": 200763
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.4340676069259644,
      "learning_rate": 2.4167007254790448e-05,
      "loss": 1.6374,
      "step": 200764
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.6869372129440308,
      "learning_rate": 2.4165398779000923e-05,
      "loss": 2.0846,
      "step": 200765
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.5400351285934448,
      "learning_rate": 2.416379035449437e-05,
      "loss": 2.0264,
      "step": 200766
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.528965711593628,
      "learning_rate": 2.4162181981270866e-05,
      "loss": 2.1159,
      "step": 200767
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.4910492897033691,
      "learning_rate": 2.4160573659330873e-05,
      "loss": 1.9619,
      "step": 200768
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.3883728981018066,
      "learning_rate": 2.4158965388674588e-05,
      "loss": 1.9367,
      "step": 200769
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.4077821969985962,
      "learning_rate": 2.4157357169302417e-05,
      "loss": 1.7858,
      "step": 200770
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.433202862739563,
      "learning_rate": 2.4155749001214554e-05,
      "loss": 1.9609,
      "step": 200771
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.4218523502349854,
      "learning_rate": 2.4154140884411433e-05,
      "loss": 1.859,
      "step": 200772
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.3694385290145874,
      "learning_rate": 2.4152532818893155e-05,
      "loss": 2.0044,
      "step": 200773
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.3768115043640137,
      "learning_rate": 2.4150924804660187e-05,
      "loss": 2.0273,
      "step": 200774
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.4640142917633057,
      "learning_rate": 2.414931684171273e-05,
      "loss": 1.9947,
      "step": 200775
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.3958616256713867,
      "learning_rate": 2.4147708930051146e-05,
      "loss": 2.1125,
      "step": 200776
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.6001862287521362,
      "learning_rate": 2.4146101069675672e-05,
      "loss": 2.0308,
      "step": 200777
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.4051096439361572,
      "learning_rate": 2.4144493260586705e-05,
      "loss": 2.0882,
      "step": 200778
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.5044652223587036,
      "learning_rate": 2.414288550278448e-05,
      "loss": 1.9407,
      "step": 200779
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.411816954612732,
      "learning_rate": 2.41412777962693e-05,
      "loss": 1.9323,
      "step": 200780
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.4421766996383667,
      "learning_rate": 2.413967014104139e-05,
      "loss": 2.0145,
      "step": 200781
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.4834259748458862,
      "learning_rate": 2.413806253710122e-05,
      "loss": 1.925,
      "step": 200782
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.521932601928711,
      "learning_rate": 2.4136454984448894e-05,
      "loss": 2.0092,
      "step": 200783
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.473935842514038,
      "learning_rate": 2.4134847483084875e-05,
      "loss": 2.0581,
      "step": 200784
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.543165922164917,
      "learning_rate": 2.4133240033009393e-05,
      "loss": 2.0474,
      "step": 200785
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.3443924188613892,
      "learning_rate": 2.4131632634222786e-05,
      "loss": 1.8164,
      "step": 200786
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.4126012325286865,
      "learning_rate": 2.413002528672522e-05,
      "loss": 2.1471,
      "step": 200787
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.5291205644607544,
      "learning_rate": 2.4128417990517158e-05,
      "loss": 1.8264,
      "step": 200788
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.4000645875930786,
      "learning_rate": 2.4126810745598768e-05,
      "loss": 1.6458,
      "step": 200789
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.6148957014083862,
      "learning_rate": 2.4125203551970486e-05,
      "loss": 2.0152,
      "step": 200790
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.4402804374694824,
      "learning_rate": 2.4123596409632505e-05,
      "loss": 2.0265,
      "step": 200791
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.5232610702514648,
      "learning_rate": 2.4121989318585165e-05,
      "loss": 2.0503,
      "step": 200792
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.4844366312026978,
      "learning_rate": 2.4120382278828733e-05,
      "loss": 1.8072,
      "step": 200793
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.397274136543274,
      "learning_rate": 2.4118775290363533e-05,
      "loss": 2.1696,
      "step": 200794
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.584741234779358,
      "learning_rate": 2.411716835318984e-05,
      "loss": 1.9569,
      "step": 200795
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.3729743957519531,
      "learning_rate": 2.411556146730802e-05,
      "loss": 1.8697,
      "step": 200796
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.344425916671753,
      "learning_rate": 2.4113954632718335e-05,
      "loss": 1.7548,
      "step": 200797
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.3517762422561646,
      "learning_rate": 2.4112347849421053e-05,
      "loss": 1.9206,
      "step": 200798
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.4839131832122803,
      "learning_rate": 2.4110741117416444e-05,
      "loss": 1.9116,
      "step": 200799
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.4629195928573608,
      "learning_rate": 2.4109134436704935e-05,
      "loss": 1.9036,
      "step": 200800
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.433906078338623,
      "learning_rate": 2.4107527807286665e-05,
      "loss": 1.7517,
      "step": 200801
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.4851160049438477,
      "learning_rate": 2.4105921229162063e-05,
      "loss": 1.9604,
      "step": 200802
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.645390510559082,
      "learning_rate": 2.4104314702331396e-05,
      "loss": 2.065,
      "step": 200803
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.4048823118209839,
      "learning_rate": 2.4102708226794932e-05,
      "loss": 2.0356,
      "step": 200804
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.6972899436950684,
      "learning_rate": 2.4101101802552935e-05,
      "loss": 1.9917,
      "step": 200805
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.4005578756332397,
      "learning_rate": 2.409949542960581e-05,
      "loss": 1.7604,
      "step": 200806
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.4003982543945312,
      "learning_rate": 2.409788910795375e-05,
      "loss": 1.9664,
      "step": 200807
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.2925529479980469,
      "learning_rate": 2.4096282837597125e-05,
      "loss": 1.9185,
      "step": 200808
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.3385456800460815,
      "learning_rate": 2.4094676618536234e-05,
      "loss": 1.7967,
      "step": 200809
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.236757516860962,
      "learning_rate": 2.409307045077128e-05,
      "loss": 1.857,
      "step": 200810
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.4992914199829102,
      "learning_rate": 2.409146433430269e-05,
      "loss": 2.0046,
      "step": 200811
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.3978021144866943,
      "learning_rate": 2.4089858269130736e-05,
      "loss": 2.1328,
      "step": 200812
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.6601791381835938,
      "learning_rate": 2.408825225525558e-05,
      "loss": 1.8646,
      "step": 200813
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.4720910787582397,
      "learning_rate": 2.4086646292677723e-05,
      "loss": 2.2176,
      "step": 200814
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.4497079849243164,
      "learning_rate": 2.4085040381397337e-05,
      "loss": 2.0232,
      "step": 200815
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.8461089134216309,
      "learning_rate": 2.4083434521414712e-05,
      "loss": 2.1157,
      "step": 200816
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.3851860761642456,
      "learning_rate": 2.4081828712730256e-05,
      "loss": 1.9898,
      "step": 200817
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.3796225786209106,
      "learning_rate": 2.4080222955344165e-05,
      "loss": 1.8022,
      "step": 200818
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.5426216125488281,
      "learning_rate": 2.4078617249256738e-05,
      "loss": 2.119,
      "step": 200819
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.3213123083114624,
      "learning_rate": 2.4077011594468343e-05,
      "loss": 2.1287,
      "step": 200820
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.461122989654541,
      "learning_rate": 2.4075405990979248e-05,
      "loss": 1.9936,
      "step": 200821
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.4121267795562744,
      "learning_rate": 2.407380043878968e-05,
      "loss": 1.9664,
      "step": 200822
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.4088189601898193,
      "learning_rate": 2.4072194937900046e-05,
      "loss": 1.9362,
      "step": 200823
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.3006099462509155,
      "learning_rate": 2.4070589488310577e-05,
      "loss": 1.8143,
      "step": 200824
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.4193402528762817,
      "learning_rate": 2.4068984090021604e-05,
      "loss": 1.7479,
      "step": 200825
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.5939747095108032,
      "learning_rate": 2.4067378743033462e-05,
      "loss": 2.0096,
      "step": 200826
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.5184030532836914,
      "learning_rate": 2.4065773447346382e-05,
      "loss": 2.1299,
      "step": 200827
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.4144009351730347,
      "learning_rate": 2.4064168202960632e-05,
      "loss": 1.9148,
      "step": 200828
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.261070966720581,
      "learning_rate": 2.4062563009876612e-05,
      "loss": 2.039,
      "step": 200829
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.4932159185409546,
      "learning_rate": 2.406095786809452e-05,
      "loss": 2.0754,
      "step": 200830
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.4359300136566162,
      "learning_rate": 2.405935277761476e-05,
      "loss": 1.892,
      "step": 200831
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.3521108627319336,
      "learning_rate": 2.4057747738437493e-05,
      "loss": 1.9755,
      "step": 200832
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.322360873222351,
      "learning_rate": 2.4056142750563257e-05,
      "loss": 1.9081,
      "step": 200833
    },
    {
      "epoch": 2.61,
      "grad_norm": 1.3289196491241455,
      "learning_rate": 2.4054537813992048e-05,
      "loss": 1.9939,
      "step": 200834
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.4017201662063599,
      "learning_rate": 2.4052932928724368e-05,
      "loss": 1.9539,
      "step": 200835
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.2899597883224487,
      "learning_rate": 2.4051328094760413e-05,
      "loss": 1.8883,
      "step": 200836
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.4265564680099487,
      "learning_rate": 2.404972331210059e-05,
      "loss": 1.7947,
      "step": 200837
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.421985387802124,
      "learning_rate": 2.4048118580745057e-05,
      "loss": 2.0488,
      "step": 200838
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.6484322547912598,
      "learning_rate": 2.404651390069432e-05,
      "loss": 2.2186,
      "step": 200839
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.5376417636871338,
      "learning_rate": 2.404490927194841e-05,
      "loss": 1.8811,
      "step": 200840
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.2948873043060303,
      "learning_rate": 2.4043304694507824e-05,
      "loss": 2.1691,
      "step": 200841
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.4606261253356934,
      "learning_rate": 2.4041700168372735e-05,
      "loss": 2.2413,
      "step": 200842
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.3951966762542725,
      "learning_rate": 2.404009569354357e-05,
      "loss": 1.6782,
      "step": 200843
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.5556468963623047,
      "learning_rate": 2.40384912700205e-05,
      "loss": 2.0218,
      "step": 200844
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.6112043857574463,
      "learning_rate": 2.403688689780395e-05,
      "loss": 1.9592,
      "step": 200845
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.4710577726364136,
      "learning_rate": 2.403528257689413e-05,
      "loss": 2.1409,
      "step": 200846
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.3803205490112305,
      "learning_rate": 2.403367830729137e-05,
      "loss": 2.1014,
      "step": 200847
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.5592460632324219,
      "learning_rate": 2.4032074088995934e-05,
      "loss": 2.0196,
      "step": 200848
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.3059474229812622,
      "learning_rate": 2.4030469922008155e-05,
      "loss": 1.9436,
      "step": 200849
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.4751148223876953,
      "learning_rate": 2.4028865806328302e-05,
      "loss": 1.9144,
      "step": 200850
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.384259581565857,
      "learning_rate": 2.4027261741956704e-05,
      "loss": 1.9063,
      "step": 200851
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.420547604560852,
      "learning_rate": 2.4025657728893666e-05,
      "loss": 2.2163,
      "step": 200852
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.518776774406433,
      "learning_rate": 2.4024053767139482e-05,
      "loss": 1.9612,
      "step": 200853
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.436058521270752,
      "learning_rate": 2.402244985669436e-05,
      "loss": 1.898,
      "step": 200854
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.245724081993103,
      "learning_rate": 2.4020845997558758e-05,
      "loss": 1.928,
      "step": 200855
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.4305659532546997,
      "learning_rate": 2.4019242189732812e-05,
      "loss": 1.9445,
      "step": 200856
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.4423800706863403,
      "learning_rate": 2.401763843321696e-05,
      "loss": 1.8357,
      "step": 200857
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.4705865383148193,
      "learning_rate": 2.4016034728011457e-05,
      "loss": 1.9675,
      "step": 200858
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.5467514991760254,
      "learning_rate": 2.4014431074116548e-05,
      "loss": 2.0142,
      "step": 200859
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.3886568546295166,
      "learning_rate": 2.4012827471532524e-05,
      "loss": 1.9969,
      "step": 200860
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.3755159378051758,
      "learning_rate": 2.4011223920259794e-05,
      "loss": 1.9079,
      "step": 200861
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.2587251663208008,
      "learning_rate": 2.4009620420298513e-05,
      "loss": 1.8074,
      "step": 200862
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.4662601947784424,
      "learning_rate": 2.4008016971649124e-05,
      "loss": 2.1914,
      "step": 200863
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.4051882028579712,
      "learning_rate": 2.400641357431182e-05,
      "loss": 2.0772,
      "step": 200864
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.3603160381317139,
      "learning_rate": 2.4004810228286975e-05,
      "loss": 2.0245,
      "step": 200865
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.496782660484314,
      "learning_rate": 2.4003206933574782e-05,
      "loss": 2.2356,
      "step": 200866
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.3905425071716309,
      "learning_rate": 2.400160369017564e-05,
      "loss": 2.0285,
      "step": 200867
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.4157203435897827,
      "learning_rate": 2.400000049808979e-05,
      "loss": 2.18,
      "step": 200868
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.3995893001556396,
      "learning_rate": 2.3998397357317554e-05,
      "loss": 1.8229,
      "step": 200869
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.4222793579101562,
      "learning_rate": 2.3996794267859276e-05,
      "loss": 1.9881,
      "step": 200870
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.558739423751831,
      "learning_rate": 2.3995191229715182e-05,
      "loss": 1.6821,
      "step": 200871
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.5130324363708496,
      "learning_rate": 2.399358824288551e-05,
      "loss": 2.1257,
      "step": 200872
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.4037617444992065,
      "learning_rate": 2.399198530737072e-05,
      "loss": 1.8947,
      "step": 200873
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.4740115404129028,
      "learning_rate": 2.3990382423170983e-05,
      "loss": 2.0641,
      "step": 200874
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.487945795059204,
      "learning_rate": 2.39887795902867e-05,
      "loss": 2.0153,
      "step": 200875
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.406872272491455,
      "learning_rate": 2.39871768087181e-05,
      "loss": 2.1562,
      "step": 200876
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.374427318572998,
      "learning_rate": 2.3985574078465485e-05,
      "loss": 1.9705,
      "step": 200877
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.568435788154602,
      "learning_rate": 2.3983971399529124e-05,
      "loss": 2.0127,
      "step": 200878
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.3732084035873413,
      "learning_rate": 2.398236877190941e-05,
      "loss": 1.8287,
      "step": 200879
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.402984380722046,
      "learning_rate": 2.3980766195606548e-05,
      "loss": 1.8354,
      "step": 200880
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.381673812866211,
      "learning_rate": 2.397916367062087e-05,
      "loss": 2.0126,
      "step": 200881
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.3462132215499878,
      "learning_rate": 2.397756119695271e-05,
      "loss": 1.9446,
      "step": 200882
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.3706798553466797,
      "learning_rate": 2.3975958774602334e-05,
      "loss": 1.9619,
      "step": 200883
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.544637680053711,
      "learning_rate": 2.397435640356997e-05,
      "loss": 1.9806,
      "step": 200884
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.4982349872589111,
      "learning_rate": 2.397275408385606e-05,
      "loss": 2.0067,
      "step": 200885
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.34285569190979,
      "learning_rate": 2.3971151815460764e-05,
      "loss": 2.0046,
      "step": 200886
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.4192805290222168,
      "learning_rate": 2.3969549598384486e-05,
      "loss": 2.1365,
      "step": 200887
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.3995625972747803,
      "learning_rate": 2.396794743262749e-05,
      "loss": 1.9937,
      "step": 200888
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.4886001348495483,
      "learning_rate": 2.396634531819004e-05,
      "loss": 2.0376,
      "step": 200889
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.422378420829773,
      "learning_rate": 2.3964743255072438e-05,
      "loss": 2.0916,
      "step": 200890
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.4399757385253906,
      "learning_rate": 2.3963141243275018e-05,
      "loss": 1.7515,
      "step": 200891
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.3316508531570435,
      "learning_rate": 2.3961539282798047e-05,
      "loss": 2.019,
      "step": 200892
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.3215502500534058,
      "learning_rate": 2.3959937373641858e-05,
      "loss": 1.9871,
      "step": 200893
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.4862836599349976,
      "learning_rate": 2.3958335515806713e-05,
      "loss": 2.0528,
      "step": 200894
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.4913922548294067,
      "learning_rate": 2.395673370929292e-05,
      "loss": 1.9496,
      "step": 200895
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.3961209058761597,
      "learning_rate": 2.3955131954100804e-05,
      "loss": 2.2015,
      "step": 200896
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.3196070194244385,
      "learning_rate": 2.3953530250230635e-05,
      "loss": 1.9789,
      "step": 200897
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.6759395599365234,
      "learning_rate": 2.395192859768268e-05,
      "loss": 1.9854,
      "step": 200898
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.6061701774597168,
      "learning_rate": 2.3950326996457302e-05,
      "loss": 1.9753,
      "step": 200899
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.4324376583099365,
      "learning_rate": 2.3948725446554806e-05,
      "loss": 1.8614,
      "step": 200900
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.3610435724258423,
      "learning_rate": 2.3947123947975354e-05,
      "loss": 1.86,
      "step": 200901
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.3228803873062134,
      "learning_rate": 2.394552250071945e-05,
      "loss": 1.9795,
      "step": 200902
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.4801676273345947,
      "learning_rate": 2.3943921104787222e-05,
      "loss": 1.8485,
      "step": 200903
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.404102087020874,
      "learning_rate": 2.3942319760179007e-05,
      "loss": 2.0666,
      "step": 200904
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.669415831565857,
      "learning_rate": 2.3940718466895205e-05,
      "loss": 2.1416,
      "step": 200905
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.4655826091766357,
      "learning_rate": 2.3939117224936012e-05,
      "loss": 2.2162,
      "step": 200906
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.366882085800171,
      "learning_rate": 2.3937516034301696e-05,
      "loss": 1.9113,
      "step": 200907
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.2984520196914673,
      "learning_rate": 2.3935914894992658e-05,
      "loss": 1.9882,
      "step": 200908
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.477277398109436,
      "learning_rate": 2.3934313807009066e-05,
      "loss": 2.0305,
      "step": 200909
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.5404233932495117,
      "learning_rate": 2.393271277035138e-05,
      "loss": 2.157,
      "step": 200910
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.5293378829956055,
      "learning_rate": 2.3931111785019807e-05,
      "loss": 1.968,
      "step": 200911
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.5361191034317017,
      "learning_rate": 2.392951085101461e-05,
      "loss": 2.0928,
      "step": 200912
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.4029290676116943,
      "learning_rate": 2.3927909968336156e-05,
      "loss": 2.1117,
      "step": 200913
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.391275405883789,
      "learning_rate": 2.3926309136984746e-05,
      "loss": 2.1446,
      "step": 200914
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.4504482746124268,
      "learning_rate": 2.3924708356960575e-05,
      "loss": 2.1527,
      "step": 200915
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.5134605169296265,
      "learning_rate": 2.392310762826405e-05,
      "loss": 1.9613,
      "step": 200916
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.3983441591262817,
      "learning_rate": 2.3921506950895464e-05,
      "loss": 1.8863,
      "step": 200917
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.39974844455719,
      "learning_rate": 2.3919906324854988e-05,
      "loss": 1.9604,
      "step": 200918
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.664015531539917,
      "learning_rate": 2.391830575014312e-05,
      "loss": 1.9782,
      "step": 200919
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.392487645149231,
      "learning_rate": 2.391670522675999e-05,
      "loss": 2.0865,
      "step": 200920
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.3910789489746094,
      "learning_rate": 2.3915104754705938e-05,
      "loss": 1.9846,
      "step": 200921
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.413816213607788,
      "learning_rate": 2.3913504333981325e-05,
      "loss": 1.835,
      "step": 200922
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.5846956968307495,
      "learning_rate": 2.3911903964586354e-05,
      "loss": 2.1773,
      "step": 200923
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.6515463590621948,
      "learning_rate": 2.3910303646521423e-05,
      "loss": 2.071,
      "step": 200924
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.4051969051361084,
      "learning_rate": 2.3908703379786797e-05,
      "loss": 1.7502,
      "step": 200925
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.4424024820327759,
      "learning_rate": 2.3907103164382713e-05,
      "loss": 2.0806,
      "step": 200926
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.389714241027832,
      "learning_rate": 2.39055030003095e-05,
      "loss": 2.181,
      "step": 200927
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.3703161478042603,
      "learning_rate": 2.3903902887567494e-05,
      "loss": 1.7677,
      "step": 200928
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.4418013095855713,
      "learning_rate": 2.3902302826156928e-05,
      "loss": 1.9689,
      "step": 200929
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.4408425092697144,
      "learning_rate": 2.3900702816078164e-05,
      "loss": 2.2302,
      "step": 200930
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.3310556411743164,
      "learning_rate": 2.3899102857331477e-05,
      "loss": 2.1973,
      "step": 200931
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.3795701265335083,
      "learning_rate": 2.389750294991716e-05,
      "loss": 2.02,
      "step": 200932
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.7957595586776733,
      "learning_rate": 2.3895903093835476e-05,
      "loss": 2.1455,
      "step": 200933
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.3824862241744995,
      "learning_rate": 2.389430328908677e-05,
      "loss": 2.0793,
      "step": 200934
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.3193352222442627,
      "learning_rate": 2.3892703535671297e-05,
      "loss": 2.0588,
      "step": 200935
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.5234452486038208,
      "learning_rate": 2.3891103833589464e-05,
      "loss": 2.044,
      "step": 200936
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.4039087295532227,
      "learning_rate": 2.3889504182841433e-05,
      "loss": 1.9421,
      "step": 200937
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.4925856590270996,
      "learning_rate": 2.3887904583427576e-05,
      "loss": 2.322,
      "step": 200938
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.4974051713943481,
      "learning_rate": 2.388630503534812e-05,
      "loss": 2.0832,
      "step": 200939
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.442099690437317,
      "learning_rate": 2.3884705538603466e-05,
      "loss": 2.2,
      "step": 200940
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.4049197435379028,
      "learning_rate": 2.3883106093193816e-05,
      "loss": 1.9814,
      "step": 200941
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.456275224685669,
      "learning_rate": 2.3881506699119536e-05,
      "loss": 1.9251,
      "step": 200942
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.3031377792358398,
      "learning_rate": 2.3879907356380924e-05,
      "loss": 1.9502,
      "step": 200943
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.4076915979385376,
      "learning_rate": 2.3878308064978215e-05,
      "loss": 1.808,
      "step": 200944
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.3513383865356445,
      "learning_rate": 2.3876708824911706e-05,
      "loss": 1.9521,
      "step": 200945
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.3084338903427124,
      "learning_rate": 2.38751096361818e-05,
      "loss": 1.5297,
      "step": 200946
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.411927580833435,
      "learning_rate": 2.387351049878866e-05,
      "loss": 2.1895,
      "step": 200947
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.4562519788742065,
      "learning_rate": 2.387191141273269e-05,
      "loss": 1.8076,
      "step": 200948
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.46037757396698,
      "learning_rate": 2.387031237801412e-05,
      "loss": 2.0378,
      "step": 200949
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.442883014678955,
      "learning_rate": 2.386871339463332e-05,
      "loss": 2.0881,
      "step": 200950
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.7072430849075317,
      "learning_rate": 2.3867114462590455e-05,
      "loss": 2.0055,
      "step": 200951
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.4217870235443115,
      "learning_rate": 2.386551558188595e-05,
      "loss": 1.7477,
      "step": 200952
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.569440245628357,
      "learning_rate": 2.3863916752520053e-05,
      "loss": 1.8996,
      "step": 200953
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.4724045991897583,
      "learning_rate": 2.3862317974493083e-05,
      "loss": 1.8594,
      "step": 200954
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.4257676601409912,
      "learning_rate": 2.3860719247805315e-05,
      "loss": 2.0146,
      "step": 200955
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.4015897512435913,
      "learning_rate": 2.385912057245708e-05,
      "loss": 2.0068,
      "step": 200956
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.3984774351119995,
      "learning_rate": 2.3857521948448575e-05,
      "loss": 2.0346,
      "step": 200957
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.4542871713638306,
      "learning_rate": 2.3855923375780207e-05,
      "loss": 1.9506,
      "step": 200958
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.3420255184173584,
      "learning_rate": 2.38543248544522e-05,
      "loss": 1.9022,
      "step": 200959
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.4433298110961914,
      "learning_rate": 2.385272638446496e-05,
      "loss": 2.1822,
      "step": 200960
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.558552861213684,
      "learning_rate": 2.3851127965818683e-05,
      "loss": 1.9148,
      "step": 200961
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.386475682258606,
      "learning_rate": 2.3849529598513706e-05,
      "loss": 1.9553,
      "step": 200962
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.3689669370651245,
      "learning_rate": 2.384793128255026e-05,
      "loss": 1.9778,
      "step": 200963
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.3801721334457397,
      "learning_rate": 2.3846333017928742e-05,
      "loss": 2.1208,
      "step": 200964
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.5595844984054565,
      "learning_rate": 2.384473480464936e-05,
      "loss": 1.8208,
      "step": 200965
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.3534681797027588,
      "learning_rate": 2.3843136642712467e-05,
      "loss": 1.9648,
      "step": 200966
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.3919388055801392,
      "learning_rate": 2.384153853211841e-05,
      "loss": 2.0859,
      "step": 200967
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.4698379039764404,
      "learning_rate": 2.383994047286738e-05,
      "loss": 2.2368,
      "step": 200968
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.3752107620239258,
      "learning_rate": 2.383834246495968e-05,
      "loss": 2.2476,
      "step": 200969
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.457867980003357,
      "learning_rate": 2.383674450839571e-05,
      "loss": 1.8951,
      "step": 200970
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.4946380853652954,
      "learning_rate": 2.3835146603175635e-05,
      "loss": 1.8512,
      "step": 200971
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.4716460704803467,
      "learning_rate": 2.383354874929989e-05,
      "loss": 2.0069,
      "step": 200972
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.4824233055114746,
      "learning_rate": 2.3831950946768673e-05,
      "loss": 1.9506,
      "step": 200973
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.5207792520523071,
      "learning_rate": 2.383035319558232e-05,
      "loss": 1.8417,
      "step": 200974
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.3524945974349976,
      "learning_rate": 2.382875549574106e-05,
      "loss": 1.9235,
      "step": 200975
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.4724855422973633,
      "learning_rate": 2.3827157847245326e-05,
      "loss": 2.0582,
      "step": 200976
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.5379456281661987,
      "learning_rate": 2.3825560250095255e-05,
      "loss": 1.9942,
      "step": 200977
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.6654075384140015,
      "learning_rate": 2.382396270429131e-05,
      "loss": 2.0845,
      "step": 200978
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.4628982543945312,
      "learning_rate": 2.3822365209833627e-05,
      "loss": 2.125,
      "step": 200979
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.332120656967163,
      "learning_rate": 2.382076776672267e-05,
      "loss": 1.9335,
      "step": 200980
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.4570895433425903,
      "learning_rate": 2.3819170374958574e-05,
      "loss": 2.0154,
      "step": 200981
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.4359807968139648,
      "learning_rate": 2.3817573034541736e-05,
      "loss": 1.9592,
      "step": 200982
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.3763251304626465,
      "learning_rate": 2.3815975745472394e-05,
      "loss": 2.1151,
      "step": 200983
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.4226869344711304,
      "learning_rate": 2.381437850775091e-05,
      "loss": 1.8973,
      "step": 200984
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.3417149782180786,
      "learning_rate": 2.381278132137748e-05,
      "loss": 1.9212,
      "step": 200985
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.2968354225158691,
      "learning_rate": 2.381118418635255e-05,
      "loss": 1.9653,
      "step": 200986
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.5612324476242065,
      "learning_rate": 2.3809587102676342e-05,
      "loss": 1.9826,
      "step": 200987
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.5501997470855713,
      "learning_rate": 2.3807990070349124e-05,
      "loss": 2.1029,
      "step": 200988
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.44161856174469,
      "learning_rate": 2.380639308937117e-05,
      "loss": 1.8757,
      "step": 200989
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.5708255767822266,
      "learning_rate": 2.3804796159742833e-05,
      "loss": 1.9949,
      "step": 200990
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.4805659055709839,
      "learning_rate": 2.380319928146439e-05,
      "loss": 1.9485,
      "step": 200991
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.489228367805481,
      "learning_rate": 2.3801602454536207e-05,
      "loss": 1.9998,
      "step": 200992
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.655005931854248,
      "learning_rate": 2.380000567895848e-05,
      "loss": 2.1797,
      "step": 200993
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.4803413152694702,
      "learning_rate": 2.3798408954731507e-05,
      "loss": 2.3648,
      "step": 200994
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.3826991319656372,
      "learning_rate": 2.3796812281855694e-05,
      "loss": 1.9149,
      "step": 200995
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.419202208518982,
      "learning_rate": 2.379521566033127e-05,
      "loss": 2.0215,
      "step": 200996
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.4793643951416016,
      "learning_rate": 2.3793619090158465e-05,
      "loss": 1.8445,
      "step": 200997
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.4027516841888428,
      "learning_rate": 2.3792022571337688e-05,
      "loss": 1.8308,
      "step": 200998
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.4988887310028076,
      "learning_rate": 2.3790426103869197e-05,
      "loss": 2.0125,
      "step": 200999
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.4778681993484497,
      "learning_rate": 2.3788829687753197e-05,
      "loss": 2.0313,
      "step": 201000
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.446249008178711,
      "learning_rate": 2.378723332299015e-05,
      "loss": 1.9045,
      "step": 201001
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.565024495124817,
      "learning_rate": 2.378563700958026e-05,
      "loss": 1.8799,
      "step": 201002
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.4395616054534912,
      "learning_rate": 2.3784040747523792e-05,
      "loss": 2.0321,
      "step": 201003
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.5274035930633545,
      "learning_rate": 2.3782444536821145e-05,
      "loss": 2.0175,
      "step": 201004
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.5464060306549072,
      "learning_rate": 2.3780848377472517e-05,
      "loss": 2.1078,
      "step": 201005
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.6308974027633667,
      "learning_rate": 2.377925226947821e-05,
      "loss": 2.1825,
      "step": 201006
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.428983449935913,
      "learning_rate": 2.3777656212838625e-05,
      "loss": 2.0746,
      "step": 201007
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.5499553680419922,
      "learning_rate": 2.3776060207553926e-05,
      "loss": 2.0421,
      "step": 201008
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.49839186668396,
      "learning_rate": 2.3774464253624513e-05,
      "loss": 2.0898,
      "step": 201009
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.789900541305542,
      "learning_rate": 2.3772868351050657e-05,
      "loss": 1.9914,
      "step": 201010
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.5462489128112793,
      "learning_rate": 2.3771272499832618e-05,
      "loss": 1.7671,
      "step": 201011
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.6051279306411743,
      "learning_rate": 2.3769676699970695e-05,
      "loss": 1.947,
      "step": 201012
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.4767024517059326,
      "learning_rate": 2.376808095146523e-05,
      "loss": 1.877,
      "step": 201013
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.531848669052124,
      "learning_rate": 2.3766485254316448e-05,
      "loss": 1.9958,
      "step": 201014
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.4648823738098145,
      "learning_rate": 2.376488960852472e-05,
      "loss": 2.221,
      "step": 201015
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.42275071144104,
      "learning_rate": 2.376329401409034e-05,
      "loss": 2.115,
      "step": 201016
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.5050350427627563,
      "learning_rate": 2.376169847101358e-05,
      "loss": 2.0535,
      "step": 201017
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.441849946975708,
      "learning_rate": 2.376010297929467e-05,
      "loss": 1.9812,
      "step": 201018
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.5587007999420166,
      "learning_rate": 2.3758507538934012e-05,
      "loss": 2.0468,
      "step": 201019
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.4698978662490845,
      "learning_rate": 2.375691214993184e-05,
      "loss": 1.9766,
      "step": 201020
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.40303373336792,
      "learning_rate": 2.3755316812288516e-05,
      "loss": 1.965,
      "step": 201021
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.4852228164672852,
      "learning_rate": 2.3753721526004275e-05,
      "loss": 1.8517,
      "step": 201022
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.507270336151123,
      "learning_rate": 2.3752126291079454e-05,
      "loss": 1.985,
      "step": 201023
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.5158087015151978,
      "learning_rate": 2.3750531107514248e-05,
      "loss": 2.0305,
      "step": 201024
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.4289216995239258,
      "learning_rate": 2.3748935975309126e-05,
      "loss": 1.8846,
      "step": 201025
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.3977701663970947,
      "learning_rate": 2.374734089446422e-05,
      "loss": 1.9943,
      "step": 201026
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.442770004272461,
      "learning_rate": 2.3745745864979927e-05,
      "loss": 1.7162,
      "step": 201027
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.5023624897003174,
      "learning_rate": 2.374415088685655e-05,
      "loss": 2.1922,
      "step": 201028
    },
    {
      "epoch": 2.62,
      "grad_norm": 2.194317579269409,
      "learning_rate": 2.374255596009432e-05,
      "loss": 2.1459,
      "step": 201029
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.4022283554077148,
      "learning_rate": 2.374096108469351e-05,
      "loss": 1.9094,
      "step": 201030
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.509522795677185,
      "learning_rate": 2.3739366260654547e-05,
      "loss": 2.0302,
      "step": 201031
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.4291740655899048,
      "learning_rate": 2.3737771487977598e-05,
      "loss": 1.9553,
      "step": 201032
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.529316782951355,
      "learning_rate": 2.3736176766663028e-05,
      "loss": 2.2562,
      "step": 201033
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.4522342681884766,
      "learning_rate": 2.3734582096711175e-05,
      "loss": 2.0594,
      "step": 201034
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.4356950521469116,
      "learning_rate": 2.3732987478122234e-05,
      "loss": 2.0599,
      "step": 201035
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.5466786623001099,
      "learning_rate": 2.3731392910896507e-05,
      "loss": 1.9365,
      "step": 201036
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.3068982362747192,
      "learning_rate": 2.3729798395034393e-05,
      "loss": 1.7943,
      "step": 201037
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.4241046905517578,
      "learning_rate": 2.372820393053606e-05,
      "loss": 2.0391,
      "step": 201038
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.4837771654129028,
      "learning_rate": 2.3726609517401906e-05,
      "loss": 1.7942,
      "step": 201039
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.493478775024414,
      "learning_rate": 2.37250151556322e-05,
      "loss": 2.1206,
      "step": 201040
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.458422303199768,
      "learning_rate": 2.3723420845227238e-05,
      "loss": 2.1268,
      "step": 201041
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.5221645832061768,
      "learning_rate": 2.3721826586187252e-05,
      "loss": 1.9529,
      "step": 201042
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.5316017866134644,
      "learning_rate": 2.372023237851265e-05,
      "loss": 2.1228,
      "step": 201043
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.463001012802124,
      "learning_rate": 2.3718638222203624e-05,
      "loss": 1.8452,
      "step": 201044
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.335038661956787,
      "learning_rate": 2.371704411726054e-05,
      "loss": 2.1003,
      "step": 201045
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.4721132516860962,
      "learning_rate": 2.371545006368364e-05,
      "loss": 1.9369,
      "step": 201046
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.4109253883361816,
      "learning_rate": 2.371385606147338e-05,
      "loss": 2.0502,
      "step": 201047
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.7389628887176514,
      "learning_rate": 2.3712262110629797e-05,
      "loss": 2.1344,
      "step": 201048
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.3440101146697998,
      "learning_rate": 2.3710668211153393e-05,
      "loss": 1.8035,
      "step": 201049
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.3452423810958862,
      "learning_rate": 2.37090743630443e-05,
      "loss": 1.8937,
      "step": 201050
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.4154689311981201,
      "learning_rate": 2.3707480566302984e-05,
      "loss": 1.9464,
      "step": 201051
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.4532430171966553,
      "learning_rate": 2.370588682092961e-05,
      "loss": 1.9367,
      "step": 201052
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.4384225606918335,
      "learning_rate": 2.3704293126924645e-05,
      "loss": 2.0763,
      "step": 201053
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.4850773811340332,
      "learning_rate": 2.3702699484288157e-05,
      "loss": 1.8498,
      "step": 201054
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.5456405878067017,
      "learning_rate": 2.3701105893020578e-05,
      "loss": 2.2508,
      "step": 201055
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.4383811950683594,
      "learning_rate": 2.3699512353122142e-05,
      "loss": 1.989,
      "step": 201056
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.461989402770996,
      "learning_rate": 2.3697918864593246e-05,
      "loss": 1.9849,
      "step": 201057
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.3058315515518188,
      "learning_rate": 2.369632542743406e-05,
      "loss": 1.7247,
      "step": 201058
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.5264277458190918,
      "learning_rate": 2.369473204164505e-05,
      "loss": 1.9705,
      "step": 201059
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.4323512315750122,
      "learning_rate": 2.3693138707226313e-05,
      "loss": 1.8829,
      "step": 201060
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.5845394134521484,
      "learning_rate": 2.3691545424178282e-05,
      "loss": 2.0313,
      "step": 201061
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.449291706085205,
      "learning_rate": 2.3689952192501126e-05,
      "loss": 2.0133,
      "step": 201062
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.548116683959961,
      "learning_rate": 2.3688359012195312e-05,
      "loss": 1.8298,
      "step": 201063
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.4552668333053589,
      "learning_rate": 2.3686765883260973e-05,
      "loss": 1.9177,
      "step": 201064
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.6705619096755981,
      "learning_rate": 2.3685172805698605e-05,
      "loss": 2.0554,
      "step": 201065
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.4203541278839111,
      "learning_rate": 2.3683579779508278e-05,
      "loss": 1.9415,
      "step": 201066
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.6025629043579102,
      "learning_rate": 2.3681986804690424e-05,
      "loss": 2.1606,
      "step": 201067
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.4495699405670166,
      "learning_rate": 2.368039388124524e-05,
      "loss": 1.9381,
      "step": 201068
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.4393514394760132,
      "learning_rate": 2.3678801009173164e-05,
      "loss": 1.8427,
      "step": 201069
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.3824118375778198,
      "learning_rate": 2.367720818847436e-05,
      "loss": 1.915,
      "step": 201070
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.3054567575454712,
      "learning_rate": 2.367561541914923e-05,
      "loss": 2.0682,
      "step": 201071
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.4949570894241333,
      "learning_rate": 2.3674022701198005e-05,
      "loss": 1.8771,
      "step": 201072
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.5831681489944458,
      "learning_rate": 2.3672430034621014e-05,
      "loss": 1.9559,
      "step": 201073
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.4436124563217163,
      "learning_rate": 2.3670837419418466e-05,
      "loss": 1.8629,
      "step": 201074
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.3831274509429932,
      "learning_rate": 2.3669244855590786e-05,
      "loss": 2.0306,
      "step": 201075
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.4099196195602417,
      "learning_rate": 2.3667652343138176e-05,
      "loss": 2.0239,
      "step": 201076
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.5603859424591064,
      "learning_rate": 2.3666059882060973e-05,
      "loss": 2.0108,
      "step": 201077
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.4490652084350586,
      "learning_rate": 2.3664467472359504e-05,
      "loss": 1.8964,
      "step": 201078
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.3869004249572754,
      "learning_rate": 2.3662875114033974e-05,
      "loss": 1.973,
      "step": 201079
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.3239696025848389,
      "learning_rate": 2.3661282807084746e-05,
      "loss": 1.8812,
      "step": 201080
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.4035392999649048,
      "learning_rate": 2.3659690551512157e-05,
      "loss": 2.0168,
      "step": 201081
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.4726334810256958,
      "learning_rate": 2.3658098347316367e-05,
      "loss": 1.9119,
      "step": 201082
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.3981187343597412,
      "learning_rate": 2.365650619449778e-05,
      "loss": 1.8004,
      "step": 201083
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.3248639106750488,
      "learning_rate": 2.3654914093056697e-05,
      "loss": 2.126,
      "step": 201084
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.3331973552703857,
      "learning_rate": 2.365332204299332e-05,
      "loss": 2.0055,
      "step": 201085
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.4396322965621948,
      "learning_rate": 2.365173004430807e-05,
      "loss": 2.0046,
      "step": 201086
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.3199849128723145,
      "learning_rate": 2.365013809700116e-05,
      "loss": 1.8814,
      "step": 201087
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.4131237268447876,
      "learning_rate": 2.364854620107288e-05,
      "loss": 1.8724,
      "step": 201088
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.6060703992843628,
      "learning_rate": 2.3646954356523574e-05,
      "loss": 2.047,
      "step": 201089
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.3792229890823364,
      "learning_rate": 2.364536256335353e-05,
      "loss": 1.8961,
      "step": 201090
    },
    {
      "epoch": 2.62,
      "grad_norm": 2.3151090145111084,
      "learning_rate": 2.3643770821562992e-05,
      "loss": 2.0727,
      "step": 201091
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.3416146039962769,
      "learning_rate": 2.3642179131152316e-05,
      "loss": 1.9602,
      "step": 201092
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.5696475505828857,
      "learning_rate": 2.3640587492121744e-05,
      "loss": 2.0336,
      "step": 201093
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.3845049142837524,
      "learning_rate": 2.3638995904471635e-05,
      "loss": 2.0235,
      "step": 201094
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.472490668296814,
      "learning_rate": 2.3637404368202294e-05,
      "loss": 2.0141,
      "step": 201095
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.3974463939666748,
      "learning_rate": 2.3635812883313955e-05,
      "loss": 2.0235,
      "step": 201096
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.371753454208374,
      "learning_rate": 2.3634221449806878e-05,
      "loss": 1.984,
      "step": 201097
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.486464023590088,
      "learning_rate": 2.363263006768147e-05,
      "loss": 1.7898,
      "step": 201098
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.3674023151397705,
      "learning_rate": 2.3631038736937924e-05,
      "loss": 2.0104,
      "step": 201099
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.3507992029190063,
      "learning_rate": 2.362944745757661e-05,
      "loss": 1.9867,
      "step": 201100
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.4547934532165527,
      "learning_rate": 2.362785622959783e-05,
      "loss": 2.0414,
      "step": 201101
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.3758037090301514,
      "learning_rate": 2.362626505300185e-05,
      "loss": 1.8429,
      "step": 201102
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.6038837432861328,
      "learning_rate": 2.3624673927788896e-05,
      "loss": 2.0596,
      "step": 201103
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.3897614479064941,
      "learning_rate": 2.3623082853959374e-05,
      "loss": 2.0101,
      "step": 201104
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.4868674278259277,
      "learning_rate": 2.3621491831513518e-05,
      "loss": 2.1151,
      "step": 201105
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.409778356552124,
      "learning_rate": 2.361990086045169e-05,
      "loss": 1.9608,
      "step": 201106
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.5318247079849243,
      "learning_rate": 2.3618309940774126e-05,
      "loss": 2.0685,
      "step": 201107
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.4330158233642578,
      "learning_rate": 2.3616719072481127e-05,
      "loss": 2.0539,
      "step": 201108
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.4030886888504028,
      "learning_rate": 2.3615128255572958e-05,
      "loss": 2.0205,
      "step": 201109
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.3605140447616577,
      "learning_rate": 2.3613537490050015e-05,
      "loss": 1.9817,
      "step": 201110
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.3475233316421509,
      "learning_rate": 2.3611946775912472e-05,
      "loss": 2.0403,
      "step": 201111
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.590782642364502,
      "learning_rate": 2.361035611316072e-05,
      "loss": 2.033,
      "step": 201112
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.4015016555786133,
      "learning_rate": 2.3608765501795e-05,
      "loss": 1.9504,
      "step": 201113
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.4725372791290283,
      "learning_rate": 2.360717494181571e-05,
      "loss": 1.9702,
      "step": 201114
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.3555527925491333,
      "learning_rate": 2.3605584433222945e-05,
      "loss": 1.9298,
      "step": 201115
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.439651608467102,
      "learning_rate": 2.360399397601721e-05,
      "loss": 1.7428,
      "step": 201116
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.4338105916976929,
      "learning_rate": 2.3602403570198637e-05,
      "loss": 1.8912,
      "step": 201117
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.6219841241836548,
      "learning_rate": 2.3600813215767655e-05,
      "loss": 1.9688,
      "step": 201118
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.699143648147583,
      "learning_rate": 2.3599222912724434e-05,
      "loss": 1.9289,
      "step": 201119
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.4547014236450195,
      "learning_rate": 2.3597632661069442e-05,
      "loss": 1.9505,
      "step": 201120
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.4394114017486572,
      "learning_rate": 2.3596042460802777e-05,
      "loss": 2.0886,
      "step": 201121
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.4740592241287231,
      "learning_rate": 2.359445231192487e-05,
      "loss": 1.9204,
      "step": 201122
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.4754174947738647,
      "learning_rate": 2.3592862214435927e-05,
      "loss": 2.0518,
      "step": 201123
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.453887701034546,
      "learning_rate": 2.3591272168336305e-05,
      "loss": 1.861,
      "step": 201124
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.4476358890533447,
      "learning_rate": 2.358968217362628e-05,
      "loss": 1.8625,
      "step": 201125
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.5874203443527222,
      "learning_rate": 2.358809223030621e-05,
      "loss": 2.3172,
      "step": 201126
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.3818047046661377,
      "learning_rate": 2.3586502338376266e-05,
      "loss": 2.0921,
      "step": 201127
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.2734729051589966,
      "learning_rate": 2.358491249783685e-05,
      "loss": 1.942,
      "step": 201128
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.4212905168533325,
      "learning_rate": 2.3583322708688156e-05,
      "loss": 1.7951,
      "step": 201129
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.3877280950546265,
      "learning_rate": 2.358173297093059e-05,
      "loss": 2.0868,
      "step": 201130
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.394044041633606,
      "learning_rate": 2.3580143284564346e-05,
      "loss": 1.8398,
      "step": 201131
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.5268528461456299,
      "learning_rate": 2.3578553649589892e-05,
      "loss": 2.0381,
      "step": 201132
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.6137356758117676,
      "learning_rate": 2.3576964066007265e-05,
      "loss": 2.0003,
      "step": 201133
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.4373716115951538,
      "learning_rate": 2.3575374533816992e-05,
      "loss": 1.9499,
      "step": 201134
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.4485398530960083,
      "learning_rate": 2.3573785053019177e-05,
      "loss": 1.7282,
      "step": 201135
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.5802143812179565,
      "learning_rate": 2.3572195623614286e-05,
      "loss": 1.9234,
      "step": 201136
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.5590970516204834,
      "learning_rate": 2.357060624560252e-05,
      "loss": 1.9105,
      "step": 201137
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.326533555984497,
      "learning_rate": 2.3569016918984274e-05,
      "loss": 1.7954,
      "step": 201138
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.6008715629577637,
      "learning_rate": 2.356742764375965e-05,
      "loss": 2.1103,
      "step": 201139
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.4725509881973267,
      "learning_rate": 2.3565838419929118e-05,
      "loss": 1.9165,
      "step": 201140
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.376611590385437,
      "learning_rate": 2.356424924749287e-05,
      "loss": 1.8392,
      "step": 201141
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.4963074922561646,
      "learning_rate": 2.3562660126451317e-05,
      "loss": 2.0347,
      "step": 201142
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.6217020750045776,
      "learning_rate": 2.3561071056804616e-05,
      "loss": 2.0136,
      "step": 201143
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.357936978340149,
      "learning_rate": 2.35594820385532e-05,
      "loss": 1.8356,
      "step": 201144
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.5903053283691406,
      "learning_rate": 2.3557893071697244e-05,
      "loss": 1.8766,
      "step": 201145
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.5430481433868408,
      "learning_rate": 2.3556304156237105e-05,
      "loss": 2.0304,
      "step": 201146
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.316838026046753,
      "learning_rate": 2.3554715292173056e-05,
      "loss": 1.9507,
      "step": 201147
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.359679102897644,
      "learning_rate": 2.3553126479505424e-05,
      "loss": 2.0085,
      "step": 201148
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.4155383110046387,
      "learning_rate": 2.3551537718234446e-05,
      "loss": 2.1244,
      "step": 201149
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.451480746269226,
      "learning_rate": 2.3549949008360558e-05,
      "loss": 2.1568,
      "step": 201150
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.4131215810775757,
      "learning_rate": 2.3548360349883822e-05,
      "loss": 1.9387,
      "step": 201151
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.464733362197876,
      "learning_rate": 2.3546771742804737e-05,
      "loss": 1.9159,
      "step": 201152
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.4471924304962158,
      "learning_rate": 2.3545183187123472e-05,
      "loss": 1.7404,
      "step": 201153
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.437728762626648,
      "learning_rate": 2.3543594682840428e-05,
      "loss": 1.9071,
      "step": 201154
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.4862315654754639,
      "learning_rate": 2.35420062299558e-05,
      "loss": 1.9701,
      "step": 201155
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.5133163928985596,
      "learning_rate": 2.3540417828469992e-05,
      "loss": 1.9678,
      "step": 201156
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.2912312746047974,
      "learning_rate": 2.3538829478383238e-05,
      "loss": 2.2152,
      "step": 201157
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.3497949838638306,
      "learning_rate": 2.3537241179695832e-05,
      "loss": 1.9613,
      "step": 201158
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.4544702768325806,
      "learning_rate": 2.353565293240801e-05,
      "loss": 1.741,
      "step": 201159
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.6004332304000854,
      "learning_rate": 2.3534064736520176e-05,
      "loss": 2.2186,
      "step": 201160
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.4853445291519165,
      "learning_rate": 2.3532476592032557e-05,
      "loss": 1.9902,
      "step": 201161
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.6698154211044312,
      "learning_rate": 2.353088849894549e-05,
      "loss": 1.9193,
      "step": 201162
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.5367727279663086,
      "learning_rate": 2.3529300457259236e-05,
      "loss": 2.0076,
      "step": 201163
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.4262869358062744,
      "learning_rate": 2.3527712466974137e-05,
      "loss": 1.9674,
      "step": 201164
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.4239637851715088,
      "learning_rate": 2.3526124528090383e-05,
      "loss": 2.1104,
      "step": 201165
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.400335669517517,
      "learning_rate": 2.3524536640608416e-05,
      "loss": 1.9514,
      "step": 201166
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.5576703548431396,
      "learning_rate": 2.3522948804528396e-05,
      "loss": 2.1006,
      "step": 201167
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.4662081003189087,
      "learning_rate": 2.3521361019850727e-05,
      "loss": 1.979,
      "step": 201168
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.4503928422927856,
      "learning_rate": 2.3519773286575638e-05,
      "loss": 1.8204,
      "step": 201169
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.7446084022521973,
      "learning_rate": 2.351818560470343e-05,
      "loss": 1.8144,
      "step": 201170
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.439182996749878,
      "learning_rate": 2.3516597974234443e-05,
      "loss": 1.9805,
      "step": 201171
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.5305848121643066,
      "learning_rate": 2.3515010395168932e-05,
      "loss": 2.2528,
      "step": 201172
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.555737018585205,
      "learning_rate": 2.3513422867507137e-05,
      "loss": 2.0381,
      "step": 201173
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.5721355676651,
      "learning_rate": 2.351183539124949e-05,
      "loss": 1.9962,
      "step": 201174
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.5818562507629395,
      "learning_rate": 2.351024796639619e-05,
      "loss": 1.8921,
      "step": 201175
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.502501368522644,
      "learning_rate": 2.3508660592947537e-05,
      "loss": 1.9668,
      "step": 201176
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.5140742063522339,
      "learning_rate": 2.3507073270903864e-05,
      "loss": 1.9198,
      "step": 201177
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.4417014122009277,
      "learning_rate": 2.3505486000265405e-05,
      "loss": 2.0669,
      "step": 201178
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.4108158349990845,
      "learning_rate": 2.350389878103256e-05,
      "loss": 1.8585,
      "step": 201179
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.2852210998535156,
      "learning_rate": 2.3502311613205528e-05,
      "loss": 1.9985,
      "step": 201180
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.3156071901321411,
      "learning_rate": 2.350072449678464e-05,
      "loss": 1.9916,
      "step": 201181
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.4820317029953003,
      "learning_rate": 2.3499137431770164e-05,
      "loss": 2.1632,
      "step": 201182
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.4701564311981201,
      "learning_rate": 2.349755041816247e-05,
      "loss": 1.9434,
      "step": 201183
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.5528734922409058,
      "learning_rate": 2.349596345596172e-05,
      "loss": 2.2045,
      "step": 201184
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.6217045783996582,
      "learning_rate": 2.3494376545168346e-05,
      "loss": 2.0897,
      "step": 201185
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.4624912738800049,
      "learning_rate": 2.3492789685782587e-05,
      "loss": 2.0645,
      "step": 201186
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.5586934089660645,
      "learning_rate": 2.3491202877804772e-05,
      "loss": 2.1338,
      "step": 201187
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.3501183986663818,
      "learning_rate": 2.3489616121235067e-05,
      "loss": 1.8692,
      "step": 201188
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.3327436447143555,
      "learning_rate": 2.348802941607394e-05,
      "loss": 2.05,
      "step": 201189
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.3398300409317017,
      "learning_rate": 2.3486442762321556e-05,
      "loss": 2.0602,
      "step": 201190
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.4733028411865234,
      "learning_rate": 2.348485615997828e-05,
      "loss": 1.9334,
      "step": 201191
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.5926604270935059,
      "learning_rate": 2.3483269609044354e-05,
      "loss": 2.0938,
      "step": 201192
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.381565809249878,
      "learning_rate": 2.3481683109520235e-05,
      "loss": 2.0345,
      "step": 201193
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.4878908395767212,
      "learning_rate": 2.3480096661405956e-05,
      "loss": 1.8976,
      "step": 201194
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.3599097728729248,
      "learning_rate": 2.347851026470202e-05,
      "loss": 2.0864,
      "step": 201195
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.468217134475708,
      "learning_rate": 2.3476923919408597e-05,
      "loss": 1.8923,
      "step": 201196
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.4266979694366455,
      "learning_rate": 2.347533762552608e-05,
      "loss": 1.9068,
      "step": 201197
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.5140577554702759,
      "learning_rate": 2.3473751383054673e-05,
      "loss": 2.0197,
      "step": 201198
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.5799610614776611,
      "learning_rate": 2.3472165191994806e-05,
      "loss": 2.1878,
      "step": 201199
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.5594056844711304,
      "learning_rate": 2.3470579052346583e-05,
      "loss": 2.0443,
      "step": 201200
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.4984912872314453,
      "learning_rate": 2.3468992964110434e-05,
      "loss": 2.0749,
      "step": 201201
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.4906980991363525,
      "learning_rate": 2.3467406927286593e-05,
      "loss": 1.9338,
      "step": 201202
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.345194697380066,
      "learning_rate": 2.3465820941875422e-05,
      "loss": 2.0231,
      "step": 201203
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.4333491325378418,
      "learning_rate": 2.346423500787713e-05,
      "loss": 1.969,
      "step": 201204
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.530909776687622,
      "learning_rate": 2.3462649125292177e-05,
      "loss": 2.2587,
      "step": 201205
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.42226243019104,
      "learning_rate": 2.3461063294120597e-05,
      "loss": 2.0966,
      "step": 201206
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.4816707372665405,
      "learning_rate": 2.345947751436289e-05,
      "loss": 2.1943,
      "step": 201207
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.3614836931228638,
      "learning_rate": 2.3457891786019256e-05,
      "loss": 2.0052,
      "step": 201208
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.6121822595596313,
      "learning_rate": 2.3456306109090062e-05,
      "loss": 2.3165,
      "step": 201209
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.4644622802734375,
      "learning_rate": 2.3454720483575472e-05,
      "loss": 1.9198,
      "step": 201210
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.5905287265777588,
      "learning_rate": 2.3453134909476023e-05,
      "loss": 2.1775,
      "step": 201211
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.2767363786697388,
      "learning_rate": 2.345154938679171e-05,
      "loss": 2.0747,
      "step": 201212
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.3456370830535889,
      "learning_rate": 2.3449963915523072e-05,
      "loss": 1.8967,
      "step": 201213
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.4267842769622803,
      "learning_rate": 2.344837849567024e-05,
      "loss": 2.0223,
      "step": 201214
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.4659733772277832,
      "learning_rate": 2.344679312723361e-05,
      "loss": 1.847,
      "step": 201215
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.4228370189666748,
      "learning_rate": 2.3445207810213383e-05,
      "loss": 1.9789,
      "step": 201216
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.5105565786361694,
      "learning_rate": 2.344362254461003e-05,
      "loss": 2.0348,
      "step": 201217
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.4420596361160278,
      "learning_rate": 2.3442037330423648e-05,
      "loss": 1.8535,
      "step": 201218
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.3165032863616943,
      "learning_rate": 2.3440452167654633e-05,
      "loss": 2.0371,
      "step": 201219
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.4000898599624634,
      "learning_rate": 2.343886705630319e-05,
      "loss": 1.981,
      "step": 201220
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.7457096576690674,
      "learning_rate": 2.3437281996369784e-05,
      "loss": 1.9023,
      "step": 201221
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.417027235031128,
      "learning_rate": 2.343569698785451e-05,
      "loss": 2.1974,
      "step": 201222
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.405787467956543,
      "learning_rate": 2.343411203075788e-05,
      "loss": 1.8659,
      "step": 201223
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.4472894668579102,
      "learning_rate": 2.343252712507998e-05,
      "loss": 2.0015,
      "step": 201224
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.4460417032241821,
      "learning_rate": 2.343094227082122e-05,
      "loss": 2.2175,
      "step": 201225
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.5040496587753296,
      "learning_rate": 2.3429357467981857e-05,
      "loss": 1.7295,
      "step": 201226
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.4592643976211548,
      "learning_rate": 2.34277727165622e-05,
      "loss": 2.2573,
      "step": 201227
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.4492669105529785,
      "learning_rate": 2.342618801656251e-05,
      "loss": 1.9089,
      "step": 201228
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.497002363204956,
      "learning_rate": 2.342460336798322e-05,
      "loss": 1.9872,
      "step": 201229
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.3205548524856567,
      "learning_rate": 2.3423018770824433e-05,
      "loss": 1.9331,
      "step": 201230
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.3771816492080688,
      "learning_rate": 2.3421434225086543e-05,
      "loss": 2.039,
      "step": 201231
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.4531090259552002,
      "learning_rate": 2.341984973076979e-05,
      "loss": 1.966,
      "step": 201232
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.5491536855697632,
      "learning_rate": 2.3418265287874572e-05,
      "loss": 2.284,
      "step": 201233
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.4594460725784302,
      "learning_rate": 2.341668089640105e-05,
      "loss": 2.0251,
      "step": 201234
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.5753388404846191,
      "learning_rate": 2.34150965563497e-05,
      "loss": 2.0283,
      "step": 201235
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.3918253183364868,
      "learning_rate": 2.3413512267720614e-05,
      "loss": 1.9946,
      "step": 201236
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.2979774475097656,
      "learning_rate": 2.3411928030514194e-05,
      "loss": 1.8404,
      "step": 201237
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.4425936937332153,
      "learning_rate": 2.3410343844730704e-05,
      "loss": 1.8963,
      "step": 201238
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.7404476404190063,
      "learning_rate": 2.3408759710370483e-05,
      "loss": 1.9323,
      "step": 201239
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.546648621559143,
      "learning_rate": 2.3407175627433727e-05,
      "loss": 2.1552,
      "step": 201240
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.5476038455963135,
      "learning_rate": 2.340559159592087e-05,
      "loss": 2.0374,
      "step": 201241
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.3473602533340454,
      "learning_rate": 2.3404007615832144e-05,
      "loss": 1.7936,
      "step": 201242
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.5613723993301392,
      "learning_rate": 2.3402423687167816e-05,
      "loss": 1.9474,
      "step": 201243
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.4046318531036377,
      "learning_rate": 2.3400839809928186e-05,
      "loss": 1.9002,
      "step": 201244
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.586492896080017,
      "learning_rate": 2.3399255984113553e-05,
      "loss": 2.0624,
      "step": 201245
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.4198180437088013,
      "learning_rate": 2.3397672209724217e-05,
      "loss": 1.966,
      "step": 201246
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.3196814060211182,
      "learning_rate": 2.3396088486760512e-05,
      "loss": 2.0843,
      "step": 201247
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.3328614234924316,
      "learning_rate": 2.339450481522267e-05,
      "loss": 1.8559,
      "step": 201248
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.3969405889511108,
      "learning_rate": 2.3392921195111026e-05,
      "loss": 1.9432,
      "step": 201249
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.4796929359436035,
      "learning_rate": 2.339133762642581e-05,
      "loss": 2.1434,
      "step": 201250
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.4000160694122314,
      "learning_rate": 2.3389754109167424e-05,
      "loss": 2.0574,
      "step": 201251
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.479079246520996,
      "learning_rate": 2.3388170643336067e-05,
      "loss": 1.9808,
      "step": 201252
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.4927067756652832,
      "learning_rate": 2.3386587228932106e-05,
      "loss": 2.0263,
      "step": 201253
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.4372080564498901,
      "learning_rate": 2.3385003865955777e-05,
      "loss": 2.0131,
      "step": 201254
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.365757703781128,
      "learning_rate": 2.3383420554407372e-05,
      "loss": 1.7239,
      "step": 201255
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.520908236503601,
      "learning_rate": 2.3381837294287264e-05,
      "loss": 2.1849,
      "step": 201256
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.35897958278656,
      "learning_rate": 2.3380254085595686e-05,
      "loss": 2.0512,
      "step": 201257
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.3513209819793701,
      "learning_rate": 2.3378670928332865e-05,
      "loss": 2.0071,
      "step": 201258
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.4117615222930908,
      "learning_rate": 2.3377087822499273e-05,
      "loss": 1.9736,
      "step": 201259
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.490089774131775,
      "learning_rate": 2.3375504768095078e-05,
      "loss": 2.1333,
      "step": 201260
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.462920069694519,
      "learning_rate": 2.337392176512054e-05,
      "loss": 1.9587,
      "step": 201261
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.4904814958572388,
      "learning_rate": 2.3372338813576064e-05,
      "loss": 2.2109,
      "step": 201262
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.3661491870880127,
      "learning_rate": 2.337075591346188e-05,
      "loss": 2.0502,
      "step": 201263
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.490058422088623,
      "learning_rate": 2.3369173064778257e-05,
      "loss": 2.0136,
      "step": 201264
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.7158784866333008,
      "learning_rate": 2.3367590267525594e-05,
      "loss": 1.9418,
      "step": 201265
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.5161622762680054,
      "learning_rate": 2.3366007521704088e-05,
      "loss": 2.1016,
      "step": 201266
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.4015817642211914,
      "learning_rate": 2.3364424827314042e-05,
      "loss": 2.0991,
      "step": 201267
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.3983614444732666,
      "learning_rate": 2.336284218435579e-05,
      "loss": 1.9144,
      "step": 201268
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.398966908454895,
      "learning_rate": 2.3361259592829595e-05,
      "loss": 1.8026,
      "step": 201269
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.3451263904571533,
      "learning_rate": 2.3359677052735793e-05,
      "loss": 1.9227,
      "step": 201270
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.417715311050415,
      "learning_rate": 2.335809456407465e-05,
      "loss": 1.8304,
      "step": 201271
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.4489500522613525,
      "learning_rate": 2.3356512126846462e-05,
      "loss": 1.9353,
      "step": 201272
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.3523821830749512,
      "learning_rate": 2.3354929741051465e-05,
      "loss": 2.1584,
      "step": 201273
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.3502923250198364,
      "learning_rate": 2.335334740669006e-05,
      "loss": 1.8579,
      "step": 201274
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.4765733480453491,
      "learning_rate": 2.3351765123762444e-05,
      "loss": 1.9195,
      "step": 201275
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.3227810859680176,
      "learning_rate": 2.335018289226899e-05,
      "loss": 2.0149,
      "step": 201276
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.3620226383209229,
      "learning_rate": 2.334860071220992e-05,
      "loss": 2.3013,
      "step": 201277
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.5130488872528076,
      "learning_rate": 2.3347018583585675e-05,
      "loss": 2.0063,
      "step": 201278
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.4153529405593872,
      "learning_rate": 2.334543650639632e-05,
      "loss": 2.0426,
      "step": 201279
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.3817217350006104,
      "learning_rate": 2.3343854480642354e-05,
      "loss": 2.0516,
      "step": 201280
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.4622693061828613,
      "learning_rate": 2.334227250632391e-05,
      "loss": 2.0017,
      "step": 201281
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.6155767440795898,
      "learning_rate": 2.3340690583441425e-05,
      "loss": 2.0607,
      "step": 201282
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.4733824729919434,
      "learning_rate": 2.333910871199506e-05,
      "loss": 1.9221,
      "step": 201283
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.516812801361084,
      "learning_rate": 2.3337526891985314e-05,
      "loss": 2.2469,
      "step": 201284
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.7053438425064087,
      "learning_rate": 2.3335945123412224e-05,
      "loss": 2.0463,
      "step": 201285
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.4702893495559692,
      "learning_rate": 2.3334363406276223e-05,
      "loss": 2.0514,
      "step": 201286
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.3767632246017456,
      "learning_rate": 2.3332781740577578e-05,
      "loss": 1.9599,
      "step": 201287
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.4761797189712524,
      "learning_rate": 2.333120012631662e-05,
      "loss": 1.9505,
      "step": 201288
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.4671040773391724,
      "learning_rate": 2.3329618563493578e-05,
      "loss": 2.0612,
      "step": 201289
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.4800535440444946,
      "learning_rate": 2.3328037052108893e-05,
      "loss": 1.7846,
      "step": 201290
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.4437823295593262,
      "learning_rate": 2.332645559216263e-05,
      "loss": 2.122,
      "step": 201291
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.4718246459960938,
      "learning_rate": 2.3324874183655252e-05,
      "loss": 1.9976,
      "step": 201292
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.3035731315612793,
      "learning_rate": 2.332329282658696e-05,
      "loss": 1.9882,
      "step": 201293
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.3941847085952759,
      "learning_rate": 2.3321711520958118e-05,
      "loss": 1.8257,
      "step": 201294
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.7120534181594849,
      "learning_rate": 2.3320130266768967e-05,
      "loss": 1.9006,
      "step": 201295
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.6908997297286987,
      "learning_rate": 2.3318549064019897e-05,
      "loss": 2.1689,
      "step": 201296
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.5618194341659546,
      "learning_rate": 2.3316967912711083e-05,
      "loss": 2.0031,
      "step": 201297
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.4430490732192993,
      "learning_rate": 2.3315386812842885e-05,
      "loss": 1.8664,
      "step": 201298
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.6688231229782104,
      "learning_rate": 2.3313805764415506e-05,
      "loss": 1.7234,
      "step": 201299
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.5135475397109985,
      "learning_rate": 2.331222476742941e-05,
      "loss": 2.1278,
      "step": 201300
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.4687107801437378,
      "learning_rate": 2.33106438218847e-05,
      "loss": 1.9289,
      "step": 201301
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.5425430536270142,
      "learning_rate": 2.330906292778191e-05,
      "loss": 2.085,
      "step": 201302
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.3641494512557983,
      "learning_rate": 2.3307482085121066e-05,
      "loss": 1.8147,
      "step": 201303
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.398299217224121,
      "learning_rate": 2.330590129390264e-05,
      "loss": 1.947,
      "step": 201304
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.3395850658416748,
      "learning_rate": 2.3304320554126797e-05,
      "loss": 2.008,
      "step": 201305
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.4519661664962769,
      "learning_rate": 2.3302739865793975e-05,
      "loss": 1.9366,
      "step": 201306
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.4935455322265625,
      "learning_rate": 2.3301159228904333e-05,
      "loss": 2.0574,
      "step": 201307
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.5904076099395752,
      "learning_rate": 2.329957864345834e-05,
      "loss": 1.9598,
      "step": 201308
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.4936089515686035,
      "learning_rate": 2.3297998109456095e-05,
      "loss": 2.1734,
      "step": 201309
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.4923115968704224,
      "learning_rate": 2.3296417626898e-05,
      "loss": 1.7805,
      "step": 201310
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.499977946281433,
      "learning_rate": 2.3294837195784287e-05,
      "loss": 1.9867,
      "step": 201311
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.3605260848999023,
      "learning_rate": 2.3293256816115326e-05,
      "loss": 1.9271,
      "step": 201312
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.4729043245315552,
      "learning_rate": 2.329167648789131e-05,
      "loss": 2.0053,
      "step": 201313
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.5996439456939697,
      "learning_rate": 2.3290096211112674e-05,
      "loss": 1.9329,
      "step": 201314
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.439276099205017,
      "learning_rate": 2.328851598577962e-05,
      "loss": 1.9364,
      "step": 201315
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.3828681707382202,
      "learning_rate": 2.328693581189245e-05,
      "loss": 2.0079,
      "step": 201316
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.230507493019104,
      "learning_rate": 2.3285355689451425e-05,
      "loss": 1.8867,
      "step": 201317
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.320250153541565,
      "learning_rate": 2.3283775618456912e-05,
      "loss": 1.9552,
      "step": 201318
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.4031256437301636,
      "learning_rate": 2.3282195598909114e-05,
      "loss": 2.0236,
      "step": 201319
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.3653488159179688,
      "learning_rate": 2.3280615630808465e-05,
      "loss": 1.95,
      "step": 201320
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.4253517389297485,
      "learning_rate": 2.3279035714155126e-05,
      "loss": 1.8156,
      "step": 201321
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.3060709238052368,
      "learning_rate": 2.3277455848949467e-05,
      "loss": 1.9383,
      "step": 201322
    },
    {
      "epoch": 2.62,
      "grad_norm": 2.055889129638672,
      "learning_rate": 2.3275876035191688e-05,
      "loss": 1.8832,
      "step": 201323
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.2987087965011597,
      "learning_rate": 2.327429627288222e-05,
      "loss": 1.8181,
      "step": 201324
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.3420796394348145,
      "learning_rate": 2.32727165620212e-05,
      "loss": 2.1073,
      "step": 201325
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.5157948732376099,
      "learning_rate": 2.327113690260909e-05,
      "loss": 1.8468,
      "step": 201326
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.6504807472229004,
      "learning_rate": 2.3269557294646056e-05,
      "loss": 2.0164,
      "step": 201327
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.5268696546554565,
      "learning_rate": 2.326797773813247e-05,
      "loss": 2.086,
      "step": 201328
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.5478585958480835,
      "learning_rate": 2.3266398233068528e-05,
      "loss": 2.0759,
      "step": 201329
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.4344929456710815,
      "learning_rate": 2.3264818779454664e-05,
      "loss": 1.8813,
      "step": 201330
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.44571852684021,
      "learning_rate": 2.326323937729101e-05,
      "loss": 2.1561,
      "step": 201331
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.5720782279968262,
      "learning_rate": 2.3261660026578e-05,
      "loss": 1.9619,
      "step": 201332
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.5555917024612427,
      "learning_rate": 2.3260080727315865e-05,
      "loss": 2.1859,
      "step": 201333
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.347949504852295,
      "learning_rate": 2.3258501479504908e-05,
      "loss": 1.9696,
      "step": 201334
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.4968538284301758,
      "learning_rate": 2.3256922283145395e-05,
      "loss": 2.0463,
      "step": 201335
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.407689094543457,
      "learning_rate": 2.325534313823766e-05,
      "loss": 1.9993,
      "step": 201336
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.3700264692306519,
      "learning_rate": 2.3253764044781963e-05,
      "loss": 2.0855,
      "step": 201337
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.5137407779693604,
      "learning_rate": 2.3252185002778644e-05,
      "loss": 1.7831,
      "step": 201338
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.4568513631820679,
      "learning_rate": 2.3250606012227968e-05,
      "loss": 1.7887,
      "step": 201339
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.4421366453170776,
      "learning_rate": 2.32490270731302e-05,
      "loss": 1.9613,
      "step": 201340
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.476596713066101,
      "learning_rate": 2.3247448185485672e-05,
      "loss": 1.8172,
      "step": 201341
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.3793429136276245,
      "learning_rate": 2.324586934929469e-05,
      "loss": 1.749,
      "step": 201342
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.4082921743392944,
      "learning_rate": 2.3244290564557478e-05,
      "loss": 2.0512,
      "step": 201343
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.5735981464385986,
      "learning_rate": 2.3242711831274442e-05,
      "loss": 2.0005,
      "step": 201344
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.6336671113967896,
      "learning_rate": 2.3241133149445777e-05,
      "loss": 1.9371,
      "step": 201345
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.4605293273925781,
      "learning_rate": 2.323955451907179e-05,
      "loss": 2.0347,
      "step": 201346
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.4756426811218262,
      "learning_rate": 2.323797594015281e-05,
      "loss": 1.9249,
      "step": 201347
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.4978179931640625,
      "learning_rate": 2.3236397412689134e-05,
      "loss": 1.9585,
      "step": 201348
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.3064695596694946,
      "learning_rate": 2.3234818936681e-05,
      "loss": 1.889,
      "step": 201349
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.379175066947937,
      "learning_rate": 2.323324051212877e-05,
      "loss": 1.9695,
      "step": 201350
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.4892749786376953,
      "learning_rate": 2.3231662139032716e-05,
      "loss": 1.8287,
      "step": 201351
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.3478078842163086,
      "learning_rate": 2.323008381739303e-05,
      "loss": 2.0106,
      "step": 201352
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.4557294845581055,
      "learning_rate": 2.3228505547210186e-05,
      "loss": 2.0286,
      "step": 201353
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.2887554168701172,
      "learning_rate": 2.3226927328484347e-05,
      "loss": 2.1332,
      "step": 201354
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.4873417615890503,
      "learning_rate": 2.322534916121588e-05,
      "loss": 1.8667,
      "step": 201355
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.355419635772705,
      "learning_rate": 2.3223771045405015e-05,
      "loss": 1.9835,
      "step": 201356
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.4968706369400024,
      "learning_rate": 2.3222192981052124e-05,
      "loss": 2.1464,
      "step": 201357
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.4921950101852417,
      "learning_rate": 2.322061496815737e-05,
      "loss": 2.0395,
      "step": 201358
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.4900598526000977,
      "learning_rate": 2.3219037006721186e-05,
      "loss": 1.8256,
      "step": 201359
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.8354626893997192,
      "learning_rate": 2.3217459096743774e-05,
      "loss": 1.8871,
      "step": 201360
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.436499834060669,
      "learning_rate": 2.3215881238225496e-05,
      "loss": 1.9457,
      "step": 201361
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.4703216552734375,
      "learning_rate": 2.3214303431166625e-05,
      "loss": 1.8974,
      "step": 201362
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.4931176900863647,
      "learning_rate": 2.3212725675567425e-05,
      "loss": 1.9747,
      "step": 201363
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.4305040836334229,
      "learning_rate": 2.3211147971428156e-05,
      "loss": 2.1262,
      "step": 201364
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.36934494972229,
      "learning_rate": 2.3209570318749194e-05,
      "loss": 1.9455,
      "step": 201365
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.6137279272079468,
      "learning_rate": 2.3207992717530765e-05,
      "loss": 1.9567,
      "step": 201366
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.5141829252243042,
      "learning_rate": 2.320641516777324e-05,
      "loss": 2.0094,
      "step": 201367
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.4538495540618896,
      "learning_rate": 2.320483766947685e-05,
      "loss": 1.9286,
      "step": 201368
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.3919416666030884,
      "learning_rate": 2.3203260222641962e-05,
      "loss": 2.3101,
      "step": 201369
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.3015012741088867,
      "learning_rate": 2.320168282726874e-05,
      "loss": 1.958,
      "step": 201370
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.4912382364273071,
      "learning_rate": 2.3200105483357588e-05,
      "loss": 2.1271,
      "step": 201371
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.57008695602417,
      "learning_rate": 2.3198528190908705e-05,
      "loss": 2.1863,
      "step": 201372
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.2829400300979614,
      "learning_rate": 2.3196950949922488e-05,
      "loss": 1.8531,
      "step": 201373
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.6202203035354614,
      "learning_rate": 2.319537376039914e-05,
      "loss": 2.0636,
      "step": 201374
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.7688052654266357,
      "learning_rate": 2.3193796622339122e-05,
      "loss": 1.7946,
      "step": 201375
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.433123230934143,
      "learning_rate": 2.3192219535742474e-05,
      "loss": 1.9706,
      "step": 201376
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.4006928205490112,
      "learning_rate": 2.3190642500609658e-05,
      "loss": 2.0249,
      "step": 201377
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.5100761651992798,
      "learning_rate": 2.3189065516940907e-05,
      "loss": 2.0262,
      "step": 201378
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.3865745067596436,
      "learning_rate": 2.318748858473656e-05,
      "loss": 2.2461,
      "step": 201379
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.3717917203903198,
      "learning_rate": 2.318591170399684e-05,
      "loss": 2.0161,
      "step": 201380
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.471493124961853,
      "learning_rate": 2.3184334874722153e-05,
      "loss": 1.8797,
      "step": 201381
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.5956015586853027,
      "learning_rate": 2.31827580969127e-05,
      "loss": 2.4017,
      "step": 201382
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.4157772064208984,
      "learning_rate": 2.3181181370568812e-05,
      "loss": 1.9373,
      "step": 201383
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.4371010065078735,
      "learning_rate": 2.317960469569069e-05,
      "loss": 2.1404,
      "step": 201384
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.408049464225769,
      "learning_rate": 2.31780280722788e-05,
      "loss": 1.9861,
      "step": 201385
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.3908134698867798,
      "learning_rate": 2.3176451500333238e-05,
      "loss": 2.1061,
      "step": 201386
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.5041388273239136,
      "learning_rate": 2.3174874979854476e-05,
      "loss": 2.024,
      "step": 201387
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.3488714694976807,
      "learning_rate": 2.3173298510842743e-05,
      "loss": 2.0273,
      "step": 201388
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.4236623048782349,
      "learning_rate": 2.3171722093298305e-05,
      "loss": 2.0173,
      "step": 201389
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.48876953125,
      "learning_rate": 2.3170145727221435e-05,
      "loss": 1.9904,
      "step": 201390
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.4195846319198608,
      "learning_rate": 2.3168569412612492e-05,
      "loss": 2.0834,
      "step": 201391
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.4892228841781616,
      "learning_rate": 2.316699314947168e-05,
      "loss": 1.9227,
      "step": 201392
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.4322140216827393,
      "learning_rate": 2.316541693779943e-05,
      "loss": 2.3191,
      "step": 201393
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.4236829280853271,
      "learning_rate": 2.3163840777595944e-05,
      "loss": 2.1163,
      "step": 201394
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.4182767868041992,
      "learning_rate": 2.316226466886152e-05,
      "loss": 2.0141,
      "step": 201395
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.3790092468261719,
      "learning_rate": 2.316068861159639e-05,
      "loss": 2.0284,
      "step": 201396
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.433807134628296,
      "learning_rate": 2.315911260580099e-05,
      "loss": 1.8073,
      "step": 201397
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.5814120769500732,
      "learning_rate": 2.315753665147545e-05,
      "loss": 1.9681,
      "step": 201398
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.8286818265914917,
      "learning_rate": 2.315596074862024e-05,
      "loss": 1.8757,
      "step": 201399
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.4653475284576416,
      "learning_rate": 2.3154384897235556e-05,
      "loss": 1.918,
      "step": 201400
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.4956382513046265,
      "learning_rate": 2.3152809097321667e-05,
      "loss": 2.0294,
      "step": 201401
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.433692455291748,
      "learning_rate": 2.315123334887887e-05,
      "loss": 1.9968,
      "step": 201402
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.5494670867919922,
      "learning_rate": 2.314965765190754e-05,
      "loss": 1.8694,
      "step": 201403
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.6225063800811768,
      "learning_rate": 2.3148082006407866e-05,
      "loss": 2.0061,
      "step": 201404
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.4942818880081177,
      "learning_rate": 2.314650641238025e-05,
      "loss": 2.0704,
      "step": 201405
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.5279191732406616,
      "learning_rate": 2.3144930869824894e-05,
      "loss": 1.9934,
      "step": 201406
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.7077419757843018,
      "learning_rate": 2.3143355378742135e-05,
      "loss": 1.901,
      "step": 201407
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.5306988954544067,
      "learning_rate": 2.31417799391322e-05,
      "loss": 1.9451,
      "step": 201408
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.4633209705352783,
      "learning_rate": 2.3140204550995488e-05,
      "loss": 2.006,
      "step": 201409
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.4277244806289673,
      "learning_rate": 2.3138629214332172e-05,
      "loss": 1.872,
      "step": 201410
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.501466155052185,
      "learning_rate": 2.313705392914268e-05,
      "loss": 1.8934,
      "step": 201411
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.410025715827942,
      "learning_rate": 2.3135478695427245e-05,
      "loss": 1.8269,
      "step": 201412
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.5745333433151245,
      "learning_rate": 2.3133903513186136e-05,
      "loss": 1.7105,
      "step": 201413
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.5797232389450073,
      "learning_rate": 2.313232838241962e-05,
      "loss": 1.9839,
      "step": 201414
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.494322419166565,
      "learning_rate": 2.3130753303128057e-05,
      "loss": 2.1358,
      "step": 201415
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.6476490497589111,
      "learning_rate": 2.312917827531169e-05,
      "loss": 2.1817,
      "step": 201416
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.467238187789917,
      "learning_rate": 2.3127603298970877e-05,
      "loss": 1.9978,
      "step": 201417
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.365361213684082,
      "learning_rate": 2.3126028374105855e-05,
      "loss": 1.9201,
      "step": 201418
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.4632149934768677,
      "learning_rate": 2.3124453500716923e-05,
      "loss": 2.0083,
      "step": 201419
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.5427340269088745,
      "learning_rate": 2.312287867880438e-05,
      "loss": 2.0958,
      "step": 201420
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.3991739749908447,
      "learning_rate": 2.3121303908368527e-05,
      "loss": 1.8049,
      "step": 201421
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.5019406080245972,
      "learning_rate": 2.31197291894096e-05,
      "loss": 1.8054,
      "step": 201422
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.5081290006637573,
      "learning_rate": 2.3118154521928023e-05,
      "loss": 1.8732,
      "step": 201423
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.3886611461639404,
      "learning_rate": 2.311657990592397e-05,
      "loss": 2.0657,
      "step": 201424
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.5444679260253906,
      "learning_rate": 2.3115005341397807e-05,
      "loss": 2.2819,
      "step": 201425
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.5996363162994385,
      "learning_rate": 2.311343082834973e-05,
      "loss": 1.8677,
      "step": 201426
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.439590334892273,
      "learning_rate": 2.3111856366780114e-05,
      "loss": 1.8017,
      "step": 201427
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.534757137298584,
      "learning_rate": 2.3110281956689215e-05,
      "loss": 2.0556,
      "step": 201428
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.6297080516815186,
      "learning_rate": 2.310870759807737e-05,
      "loss": 1.997,
      "step": 201429
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.4725223779678345,
      "learning_rate": 2.310713329094488e-05,
      "loss": 2.0187,
      "step": 201430
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.3049999475479126,
      "learning_rate": 2.3105559035291908e-05,
      "loss": 1.8941,
      "step": 201431
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.3903465270996094,
      "learning_rate": 2.3103984831118893e-05,
      "loss": 1.9255,
      "step": 201432
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.408002257347107,
      "learning_rate": 2.3102410678426097e-05,
      "loss": 1.9768,
      "step": 201433
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.3949145078659058,
      "learning_rate": 2.310083657721372e-05,
      "loss": 1.9056,
      "step": 201434
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.5582637786865234,
      "learning_rate": 2.30992625274822e-05,
      "loss": 1.6882,
      "step": 201435
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.3874247074127197,
      "learning_rate": 2.309768852923173e-05,
      "loss": 1.8066,
      "step": 201436
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.4721282720565796,
      "learning_rate": 2.3096114582462578e-05,
      "loss": 2.0375,
      "step": 201437
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.5611867904663086,
      "learning_rate": 2.3094540687175143e-05,
      "loss": 2.0611,
      "step": 201438
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.447911262512207,
      "learning_rate": 2.3092966843369632e-05,
      "loss": 2.069,
      "step": 201439
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.3214313983917236,
      "learning_rate": 2.309139305104637e-05,
      "loss": 1.9365,
      "step": 201440
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.4261513948440552,
      "learning_rate": 2.308981931020566e-05,
      "loss": 1.946,
      "step": 201441
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.4332748651504517,
      "learning_rate": 2.30882456208478e-05,
      "loss": 1.9052,
      "step": 201442
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.4883167743682861,
      "learning_rate": 2.308667198297299e-05,
      "loss": 1.9915,
      "step": 201443
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.7792072296142578,
      "learning_rate": 2.308509839658167e-05,
      "loss": 1.9571,
      "step": 201444
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.423563003540039,
      "learning_rate": 2.3083524861673996e-05,
      "loss": 1.928,
      "step": 201445
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.440805196762085,
      "learning_rate": 2.308195137825034e-05,
      "loss": 2.1077,
      "step": 201446
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.4546937942504883,
      "learning_rate": 2.3080377946311003e-05,
      "loss": 2.1306,
      "step": 201447
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.3268766403198242,
      "learning_rate": 2.3078804565856214e-05,
      "loss": 1.8984,
      "step": 201448
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.5346194505691528,
      "learning_rate": 2.307723123688634e-05,
      "loss": 2.1889,
      "step": 201449
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.530269980430603,
      "learning_rate": 2.307565795940165e-05,
      "loss": 1.9917,
      "step": 201450
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.4244235754013062,
      "learning_rate": 2.3074084733402342e-05,
      "loss": 2.0927,
      "step": 201451
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.385398268699646,
      "learning_rate": 2.3072511558888852e-05,
      "loss": 2.0782,
      "step": 201452
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.4034794569015503,
      "learning_rate": 2.3070938435861376e-05,
      "loss": 2.0008,
      "step": 201453
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.6674684286117554,
      "learning_rate": 2.306936536432028e-05,
      "loss": 2.0286,
      "step": 201454
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.5493011474609375,
      "learning_rate": 2.3067792344265802e-05,
      "loss": 2.1217,
      "step": 201455
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.5859185457229614,
      "learning_rate": 2.306621937569827e-05,
      "loss": 2.0116,
      "step": 201456
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.559707760810852,
      "learning_rate": 2.3064646458617887e-05,
      "loss": 2.0258,
      "step": 201457
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.6795399188995361,
      "learning_rate": 2.3063073593025082e-05,
      "loss": 1.9449,
      "step": 201458
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.490904450416565,
      "learning_rate": 2.306150077892003e-05,
      "loss": 1.9299,
      "step": 201459
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.3180451393127441,
      "learning_rate": 2.3059928016303086e-05,
      "loss": 2.0986,
      "step": 201460
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.4067280292510986,
      "learning_rate": 2.305835530517456e-05,
      "loss": 2.0382,
      "step": 201461
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.5557584762573242,
      "learning_rate": 2.3056782645534712e-05,
      "loss": 2.0346,
      "step": 201462
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.498303771018982,
      "learning_rate": 2.3055210037383775e-05,
      "loss": 1.8522,
      "step": 201463
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.4398430585861206,
      "learning_rate": 2.3053637480722155e-05,
      "loss": 2.1224,
      "step": 201464
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.4504060745239258,
      "learning_rate": 2.3052064975550045e-05,
      "loss": 1.845,
      "step": 201465
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.3452708721160889,
      "learning_rate": 2.3050492521867848e-05,
      "loss": 2.0077,
      "step": 201466
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.442534327507019,
      "learning_rate": 2.3048920119675795e-05,
      "loss": 1.9749,
      "step": 201467
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.4427319765090942,
      "learning_rate": 2.3047347768974156e-05,
      "loss": 1.8715,
      "step": 201468
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.3763697147369385,
      "learning_rate": 2.3045775469763196e-05,
      "loss": 1.8624,
      "step": 201469
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.4551500082015991,
      "learning_rate": 2.304420322204331e-05,
      "loss": 1.8685,
      "step": 201470
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.4944833517074585,
      "learning_rate": 2.3042631025814674e-05,
      "loss": 2.1615,
      "step": 201471
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.4237598180770874,
      "learning_rate": 2.304105888107768e-05,
      "loss": 1.9233,
      "step": 201472
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.4900755882263184,
      "learning_rate": 2.3039486787832595e-05,
      "loss": 2.056,
      "step": 201473
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.3969086408615112,
      "learning_rate": 2.3037914746079722e-05,
      "loss": 1.9965,
      "step": 201474
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.320294976234436,
      "learning_rate": 2.303634275581926e-05,
      "loss": 1.7973,
      "step": 201475
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.4332798719406128,
      "learning_rate": 2.3034770817051606e-05,
      "loss": 2.1838,
      "step": 201476
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.3998188972473145,
      "learning_rate": 2.3033198929776963e-05,
      "loss": 2.0482,
      "step": 201477
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.5171371698379517,
      "learning_rate": 2.3031627093995763e-05,
      "loss": 2.1928,
      "step": 201478
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.6288167238235474,
      "learning_rate": 2.3030055309708173e-05,
      "loss": 1.8874,
      "step": 201479
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.4470975399017334,
      "learning_rate": 2.302848357691456e-05,
      "loss": 1.8782,
      "step": 201480
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.4292893409729004,
      "learning_rate": 2.3026911895615086e-05,
      "loss": 2.0777,
      "step": 201481
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.5129568576812744,
      "learning_rate": 2.3025340265810222e-05,
      "loss": 1.921,
      "step": 201482
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.4112950563430786,
      "learning_rate": 2.3023768687500133e-05,
      "loss": 1.9781,
      "step": 201483
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.728350043296814,
      "learning_rate": 2.3022197160685186e-05,
      "loss": 1.9736,
      "step": 201484
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.4897950887680054,
      "learning_rate": 2.3020625685365613e-05,
      "loss": 2.1033,
      "step": 201485
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.4841753244400024,
      "learning_rate": 2.301905426154178e-05,
      "loss": 1.959,
      "step": 201486
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.3709968328475952,
      "learning_rate": 2.3017482889213855e-05,
      "loss": 2.0297,
      "step": 201487
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.3385558128356934,
      "learning_rate": 2.3015911568382273e-05,
      "loss": 1.8411,
      "step": 201488
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.475915789604187,
      "learning_rate": 2.3014340299047195e-05,
      "loss": 1.9517,
      "step": 201489
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.4254862070083618,
      "learning_rate": 2.301276908120906e-05,
      "loss": 1.9709,
      "step": 201490
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.395241618156433,
      "learning_rate": 2.301119791486806e-05,
      "loss": 2.1921,
      "step": 201491
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.6533687114715576,
      "learning_rate": 2.3009626800024504e-05,
      "loss": 1.9958,
      "step": 201492
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.4974160194396973,
      "learning_rate": 2.3008055736678653e-05,
      "loss": 1.9989,
      "step": 201493
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.4497281312942505,
      "learning_rate": 2.3006484724830877e-05,
      "loss": 2.0646,
      "step": 201494
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.4764173030853271,
      "learning_rate": 2.300491376448137e-05,
      "loss": 1.9451,
      "step": 201495
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.446953535079956,
      "learning_rate": 2.3003342855630503e-05,
      "loss": 1.9483,
      "step": 201496
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.486830711364746,
      "learning_rate": 2.3001771998278572e-05,
      "loss": 2.0007,
      "step": 201497
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.4481476545333862,
      "learning_rate": 2.3000201192425848e-05,
      "loss": 2.1551,
      "step": 201498
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.5195937156677246,
      "learning_rate": 2.299863043807253e-05,
      "loss": 2.0061,
      "step": 201499
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.3887646198272705,
      "learning_rate": 2.299705973521908e-05,
      "loss": 1.9542,
      "step": 201500
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.5410172939300537,
      "learning_rate": 2.2995489083865637e-05,
      "loss": 2.037,
      "step": 201501
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.4297131299972534,
      "learning_rate": 2.299391848401263e-05,
      "loss": 2.0269,
      "step": 201502
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.5420026779174805,
      "learning_rate": 2.299234793566026e-05,
      "loss": 2.0468,
      "step": 201503
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.499526858329773,
      "learning_rate": 2.299077743880886e-05,
      "loss": 1.9365,
      "step": 201504
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.3503092527389526,
      "learning_rate": 2.2989206993458665e-05,
      "loss": 1.9283,
      "step": 201505
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.3940216302871704,
      "learning_rate": 2.2987636599610004e-05,
      "loss": 2.0534,
      "step": 201506
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.5523512363433838,
      "learning_rate": 2.298606625726318e-05,
      "loss": 1.8949,
      "step": 201507
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.3402355909347534,
      "learning_rate": 2.2984495966418493e-05,
      "loss": 1.7156,
      "step": 201508
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.4630836248397827,
      "learning_rate": 2.2982925727076207e-05,
      "loss": 1.8796,
      "step": 201509
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.4525059461593628,
      "learning_rate": 2.2981355539236658e-05,
      "loss": 2.0377,
      "step": 201510
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.3923070430755615,
      "learning_rate": 2.297978540290001e-05,
      "loss": 2.1581,
      "step": 201511
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.6165963411331177,
      "learning_rate": 2.2978215318066727e-05,
      "loss": 1.942,
      "step": 201512
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.4352269172668457,
      "learning_rate": 2.2976645284736982e-05,
      "loss": 2.0296,
      "step": 201513
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.3608506917953491,
      "learning_rate": 2.297507530291114e-05,
      "loss": 1.969,
      "step": 201514
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.3663288354873657,
      "learning_rate": 2.297350537258943e-05,
      "loss": 1.8734,
      "step": 201515
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.4836748838424683,
      "learning_rate": 2.297193549377222e-05,
      "loss": 1.9142,
      "step": 201516
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.3698176145553589,
      "learning_rate": 2.2970365666459743e-05,
      "loss": 1.8907,
      "step": 201517
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.4169485569000244,
      "learning_rate": 2.2968795890652304e-05,
      "loss": 1.979,
      "step": 201518
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.5750467777252197,
      "learning_rate": 2.296722616635016e-05,
      "loss": 1.8873,
      "step": 201519
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.416061282157898,
      "learning_rate": 2.2965656493553653e-05,
      "loss": 2.2335,
      "step": 201520
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.3548482656478882,
      "learning_rate": 2.2964086872263044e-05,
      "loss": 1.8009,
      "step": 201521
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.4401450157165527,
      "learning_rate": 2.29625173024787e-05,
      "loss": 2.1587,
      "step": 201522
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.5083736181259155,
      "learning_rate": 2.2960947784200823e-05,
      "loss": 2.0294,
      "step": 201523
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.5356780290603638,
      "learning_rate": 2.2959378317429712e-05,
      "loss": 2.1049,
      "step": 201524
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.4020943641662598,
      "learning_rate": 2.2957808902165696e-05,
      "loss": 1.9962,
      "step": 201525
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.4460920095443726,
      "learning_rate": 2.2956239538409082e-05,
      "loss": 1.7028,
      "step": 201526
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.6786900758743286,
      "learning_rate": 2.2954670226160067e-05,
      "loss": 1.8409,
      "step": 201527
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.5197412967681885,
      "learning_rate": 2.295310096541908e-05,
      "loss": 1.9282,
      "step": 201528
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.4671802520751953,
      "learning_rate": 2.2951531756186326e-05,
      "loss": 1.9841,
      "step": 201529
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.4158185720443726,
      "learning_rate": 2.2949962598462067e-05,
      "loss": 2.0215,
      "step": 201530
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.4134366512298584,
      "learning_rate": 2.2948393492246675e-05,
      "loss": 1.8877,
      "step": 201531
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.3828946352005005,
      "learning_rate": 2.2946824437540413e-05,
      "loss": 1.8639,
      "step": 201532
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.3491185903549194,
      "learning_rate": 2.2945255434343546e-05,
      "loss": 1.8264,
      "step": 201533
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.4469000101089478,
      "learning_rate": 2.294368648265641e-05,
      "loss": 2.1072,
      "step": 201534
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.5379095077514648,
      "learning_rate": 2.294211758247927e-05,
      "loss": 1.9355,
      "step": 201535
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.3566852807998657,
      "learning_rate": 2.2940548733812357e-05,
      "loss": 1.9975,
      "step": 201536
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.4934593439102173,
      "learning_rate": 2.293897993665611e-05,
      "loss": 2.095,
      "step": 201537
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.4618675708770752,
      "learning_rate": 2.2937411191010658e-05,
      "loss": 2.0753,
      "step": 201538
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.37098228931427,
      "learning_rate": 2.293584249687643e-05,
      "loss": 1.8593,
      "step": 201539
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.2857449054718018,
      "learning_rate": 2.293427385425367e-05,
      "loss": 2.0679,
      "step": 201540
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.6002140045166016,
      "learning_rate": 2.2932705263142635e-05,
      "loss": 1.8841,
      "step": 201541
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.5265636444091797,
      "learning_rate": 2.293113672354363e-05,
      "loss": 2.278,
      "step": 201542
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.4129091501235962,
      "learning_rate": 2.2929568235456986e-05,
      "loss": 2.0015,
      "step": 201543
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.3743774890899658,
      "learning_rate": 2.29279997988829e-05,
      "loss": 1.7758,
      "step": 201544
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.602235198020935,
      "learning_rate": 2.2926431413821778e-05,
      "loss": 1.9567,
      "step": 201545
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.5492502450942993,
      "learning_rate": 2.2924863080273882e-05,
      "loss": 2.0021,
      "step": 201546
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.426131010055542,
      "learning_rate": 2.2923294798239477e-05,
      "loss": 1.9868,
      "step": 201547
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.4710118770599365,
      "learning_rate": 2.29217265677188e-05,
      "loss": 1.9561,
      "step": 201548
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.476735234260559,
      "learning_rate": 2.2920158388712282e-05,
      "loss": 2.0672,
      "step": 201549
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.369028091430664,
      "learning_rate": 2.291859026122006e-05,
      "loss": 2.0109,
      "step": 201550
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.485883355140686,
      "learning_rate": 2.2917022185242563e-05,
      "loss": 1.8773,
      "step": 201551
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.448589563369751,
      "learning_rate": 2.291545416078002e-05,
      "loss": 1.8592,
      "step": 201552
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.5199756622314453,
      "learning_rate": 2.291388618783274e-05,
      "loss": 1.9716,
      "step": 201553
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.3744118213653564,
      "learning_rate": 2.291231826640092e-05,
      "loss": 1.9863,
      "step": 201554
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.576139211654663,
      "learning_rate": 2.2910750396485024e-05,
      "loss": 2.002,
      "step": 201555
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.384720802307129,
      "learning_rate": 2.290918257808515e-05,
      "loss": 2.0898,
      "step": 201556
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.4748932123184204,
      "learning_rate": 2.290761481120177e-05,
      "loss": 2.0883,
      "step": 201557
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.432701587677002,
      "learning_rate": 2.290604709583508e-05,
      "loss": 2.0643,
      "step": 201558
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.3869508504867554,
      "learning_rate": 2.290447943198541e-05,
      "loss": 2.007,
      "step": 201559
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.4115976095199585,
      "learning_rate": 2.290291181965297e-05,
      "loss": 2.1028,
      "step": 201560
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.4679429531097412,
      "learning_rate": 2.2901344258838118e-05,
      "loss": 1.8551,
      "step": 201561
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.3785974979400635,
      "learning_rate": 2.289977674954112e-05,
      "loss": 1.9061,
      "step": 201562
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.6217769384384155,
      "learning_rate": 2.2898209291762347e-05,
      "loss": 2.0174,
      "step": 201563
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.3302083015441895,
      "learning_rate": 2.2896641885501998e-05,
      "loss": 1.9756,
      "step": 201564
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.465731143951416,
      "learning_rate": 2.289507453076044e-05,
      "loss": 1.996,
      "step": 201565
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.3398278951644897,
      "learning_rate": 2.2893507227537833e-05,
      "loss": 1.9438,
      "step": 201566
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.5624514818191528,
      "learning_rate": 2.2891939975834583e-05,
      "loss": 2.1037,
      "step": 201567
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.4445760250091553,
      "learning_rate": 2.2890372775650957e-05,
      "loss": 1.8045,
      "step": 201568
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.4018189907073975,
      "learning_rate": 2.288880562698725e-05,
      "loss": 1.96,
      "step": 201569
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.4466005563735962,
      "learning_rate": 2.2887238529843765e-05,
      "loss": 2.0491,
      "step": 201570
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.7207289934158325,
      "learning_rate": 2.2885671484220768e-05,
      "loss": 1.9256,
      "step": 201571
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.722048044204712,
      "learning_rate": 2.2884104490118492e-05,
      "loss": 2.0846,
      "step": 201572
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.4291573762893677,
      "learning_rate": 2.2882537547537372e-05,
      "loss": 2.1324,
      "step": 201573
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.4930353164672852,
      "learning_rate": 2.288097065647754e-05,
      "loss": 1.914,
      "step": 201574
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.561640977859497,
      "learning_rate": 2.2879403816939423e-05,
      "loss": 2.0544,
      "step": 201575
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.3928831815719604,
      "learning_rate": 2.2877837028923262e-05,
      "loss": 2.0202,
      "step": 201576
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.4126867055892944,
      "learning_rate": 2.287627029242932e-05,
      "loss": 1.9244,
      "step": 201577
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.6188873052597046,
      "learning_rate": 2.28747036074579e-05,
      "loss": 2.2158,
      "step": 201578
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.4510811567306519,
      "learning_rate": 2.287313697400933e-05,
      "loss": 2.0472,
      "step": 201579
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.4974907636642456,
      "learning_rate": 2.2871570392083817e-05,
      "loss": 2.0501,
      "step": 201580
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.3592294454574585,
      "learning_rate": 2.2870003861681784e-05,
      "loss": 2.1536,
      "step": 201581
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.8720343112945557,
      "learning_rate": 2.286843738280337e-05,
      "loss": 1.9723,
      "step": 201582
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.299545168876648,
      "learning_rate": 2.286687095544908e-05,
      "loss": 1.9343,
      "step": 201583
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.350615382194519,
      "learning_rate": 2.2865304579618937e-05,
      "loss": 2.0555,
      "step": 201584
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.613653302192688,
      "learning_rate": 2.2863738255313447e-05,
      "loss": 1.9466,
      "step": 201585
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.5508679151535034,
      "learning_rate": 2.286217198253274e-05,
      "loss": 2.1946,
      "step": 201586
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.6007965803146362,
      "learning_rate": 2.286060576127725e-05,
      "loss": 1.9381,
      "step": 201587
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.405712604522705,
      "learning_rate": 2.285903959154718e-05,
      "loss": 2.0078,
      "step": 201588
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.457651972770691,
      "learning_rate": 2.285747347334289e-05,
      "loss": 1.7678,
      "step": 201589
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.4036810398101807,
      "learning_rate": 2.2855907406664585e-05,
      "loss": 2.0719,
      "step": 201590
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.3754838705062866,
      "learning_rate": 2.28543413915126e-05,
      "loss": 1.9908,
      "step": 201591
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.442766547203064,
      "learning_rate": 2.2852775427887195e-05,
      "loss": 2.0793,
      "step": 201592
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.47929847240448,
      "learning_rate": 2.2851209515788736e-05,
      "loss": 1.8744,
      "step": 201593
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.6444990634918213,
      "learning_rate": 2.284964365521743e-05,
      "loss": 2.1776,
      "step": 201594
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.4604716300964355,
      "learning_rate": 2.2848077846173706e-05,
      "loss": 1.9611,
      "step": 201595
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.7224143743515015,
      "learning_rate": 2.2846512088657666e-05,
      "loss": 1.9154,
      "step": 201596
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.4636982679367065,
      "learning_rate": 2.2844946382669704e-05,
      "loss": 1.9686,
      "step": 201597
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.4771956205368042,
      "learning_rate": 2.284338072821006e-05,
      "loss": 1.9733,
      "step": 201598
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.3170896768569946,
      "learning_rate": 2.2841815125279127e-05,
      "loss": 1.9864,
      "step": 201599
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.4013978242874146,
      "learning_rate": 2.284024957387708e-05,
      "loss": 2.0752,
      "step": 201600
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.5312689542770386,
      "learning_rate": 2.283868407400431e-05,
      "loss": 1.7743,
      "step": 201601
    },
    {
      "epoch": 2.62,
      "grad_norm": 1.4473683834075928,
      "learning_rate": 2.2837118625661054e-05,
      "loss": 1.9203,
      "step": 201602
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.5580171346664429,
      "learning_rate": 2.283555322884765e-05,
      "loss": 1.8481,
      "step": 201603
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.4703911542892456,
      "learning_rate": 2.283398788356425e-05,
      "loss": 1.925,
      "step": 201604
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.484444499015808,
      "learning_rate": 2.2832422589811307e-05,
      "loss": 1.9119,
      "step": 201605
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.2697196006774902,
      "learning_rate": 2.2830857347589004e-05,
      "loss": 1.8071,
      "step": 201606
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.3971201181411743,
      "learning_rate": 2.2829292156897715e-05,
      "loss": 1.8743,
      "step": 201607
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.5357637405395508,
      "learning_rate": 2.2827727017737706e-05,
      "loss": 1.889,
      "step": 201608
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.434188961982727,
      "learning_rate": 2.2826161930109243e-05,
      "loss": 1.9229,
      "step": 201609
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.383530855178833,
      "learning_rate": 2.2824596894012594e-05,
      "loss": 2.0634,
      "step": 201610
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.4436571598052979,
      "learning_rate": 2.2823031909448118e-05,
      "loss": 1.9755,
      "step": 201611
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.397101640701294,
      "learning_rate": 2.2821466976416026e-05,
      "loss": 1.9116,
      "step": 201612
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.6975553035736084,
      "learning_rate": 2.281990209491671e-05,
      "loss": 1.9509,
      "step": 201613
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.5771740674972534,
      "learning_rate": 2.2818337264950404e-05,
      "loss": 1.9285,
      "step": 201614
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.8600742816925049,
      "learning_rate": 2.2816772486517342e-05,
      "loss": 1.8806,
      "step": 201615
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.494184136390686,
      "learning_rate": 2.281520775961796e-05,
      "loss": 1.7939,
      "step": 201616
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.4326677322387695,
      "learning_rate": 2.281364308425242e-05,
      "loss": 2.1261,
      "step": 201617
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.4504315853118896,
      "learning_rate": 2.2812078460421023e-05,
      "loss": 2.0838,
      "step": 201618
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.5213555097579956,
      "learning_rate": 2.281051388812414e-05,
      "loss": 2.1843,
      "step": 201619
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.3673007488250732,
      "learning_rate": 2.280894936736203e-05,
      "loss": 1.9322,
      "step": 201620
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.3629212379455566,
      "learning_rate": 2.2807384898134896e-05,
      "loss": 1.96,
      "step": 201621
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.5596541166305542,
      "learning_rate": 2.2805820480443172e-05,
      "loss": 1.9093,
      "step": 201622
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.39434814453125,
      "learning_rate": 2.2804256114287022e-05,
      "loss": 2.0277,
      "step": 201623
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.5404804944992065,
      "learning_rate": 2.280269179966685e-05,
      "loss": 1.9034,
      "step": 201624
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.4362938404083252,
      "learning_rate": 2.2801127536582854e-05,
      "loss": 2.0957,
      "step": 201625
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.4770082235336304,
      "learning_rate": 2.27995633250354e-05,
      "loss": 1.9216,
      "step": 201626
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.4504785537719727,
      "learning_rate": 2.2797999165024683e-05,
      "loss": 2.0051,
      "step": 201627
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.5006670951843262,
      "learning_rate": 2.2796435056551077e-05,
      "loss": 2.0225,
      "step": 201628
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.3489805459976196,
      "learning_rate": 2.279487099961481e-05,
      "loss": 2.173,
      "step": 201629
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.4869595766067505,
      "learning_rate": 2.2793306994216286e-05,
      "loss": 1.9461,
      "step": 201630
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.58018159866333,
      "learning_rate": 2.2791743040355704e-05,
      "loss": 2.0539,
      "step": 201631
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.602102279663086,
      "learning_rate": 2.279017913803336e-05,
      "loss": 2.0182,
      "step": 201632
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.5323665142059326,
      "learning_rate": 2.278861528724949e-05,
      "loss": 2.1264,
      "step": 201633
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.5694183111190796,
      "learning_rate": 2.2787051488004528e-05,
      "loss": 2.1454,
      "step": 201634
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.4552868604660034,
      "learning_rate": 2.2785487740298602e-05,
      "loss": 1.5518,
      "step": 201635
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.4520878791809082,
      "learning_rate": 2.2783924044132185e-05,
      "loss": 2.0716,
      "step": 201636
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.5341529846191406,
      "learning_rate": 2.278236039950544e-05,
      "loss": 1.9353,
      "step": 201637
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.3971067667007446,
      "learning_rate": 2.2780796806418666e-05,
      "loss": 1.8976,
      "step": 201638
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.4341472387313843,
      "learning_rate": 2.2779233264872167e-05,
      "loss": 2.013,
      "step": 201639
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.3798030614852905,
      "learning_rate": 2.277766977486627e-05,
      "loss": 1.7512,
      "step": 201640
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.4647549390792847,
      "learning_rate": 2.2776106336401178e-05,
      "loss": 1.827,
      "step": 201641
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.5212178230285645,
      "learning_rate": 2.2774542949477325e-05,
      "loss": 2.1264,
      "step": 201642
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.4288524389266968,
      "learning_rate": 2.2772979614094878e-05,
      "loss": 2.0323,
      "step": 201643
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.563599944114685,
      "learning_rate": 2.27714163302542e-05,
      "loss": 2.0198,
      "step": 201644
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.5720267295837402,
      "learning_rate": 2.276985309795546e-05,
      "loss": 2.0012,
      "step": 201645
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.4286357164382935,
      "learning_rate": 2.2768289917199124e-05,
      "loss": 1.8262,
      "step": 201646
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.371021032333374,
      "learning_rate": 2.2766726787985323e-05,
      "loss": 1.6984,
      "step": 201647
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.667739748954773,
      "learning_rate": 2.2765163710314493e-05,
      "loss": 1.9237,
      "step": 201648
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.418509840965271,
      "learning_rate": 2.27636006841868e-05,
      "loss": 1.888,
      "step": 201649
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.4292402267456055,
      "learning_rate": 2.2762037709602676e-05,
      "loss": 2.2054,
      "step": 201650
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.5042202472686768,
      "learning_rate": 2.276047478656222e-05,
      "loss": 2.0706,
      "step": 201651
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.5230114459991455,
      "learning_rate": 2.2758911915065868e-05,
      "loss": 1.9012,
      "step": 201652
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.4484138488769531,
      "learning_rate": 2.2757349095113818e-05,
      "loss": 2.1648,
      "step": 201653
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.389904260635376,
      "learning_rate": 2.275578632670647e-05,
      "loss": 1.909,
      "step": 201654
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.3739067316055298,
      "learning_rate": 2.2754223609844026e-05,
      "loss": 1.8979,
      "step": 201655
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.326100468635559,
      "learning_rate": 2.275266094452688e-05,
      "loss": 2.1177,
      "step": 201656
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.484676718711853,
      "learning_rate": 2.2751098330755135e-05,
      "loss": 1.9057,
      "step": 201657
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.6366506814956665,
      "learning_rate": 2.2749535768529258e-05,
      "loss": 1.9585,
      "step": 201658
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.6160852909088135,
      "learning_rate": 2.274797325784945e-05,
      "loss": 2.0542,
      "step": 201659
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.4015836715698242,
      "learning_rate": 2.274641079871604e-05,
      "loss": 2.0526,
      "step": 201660
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.5187499523162842,
      "learning_rate": 2.2744848391129265e-05,
      "loss": 2.1953,
      "step": 201661
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.4391154050827026,
      "learning_rate": 2.274328603508959e-05,
      "loss": 1.8218,
      "step": 201662
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.6885472536087036,
      "learning_rate": 2.274172373059705e-05,
      "loss": 1.9565,
      "step": 201663
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.325836420059204,
      "learning_rate": 2.2740161477652073e-05,
      "loss": 1.8198,
      "step": 201664
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.353781819343567,
      "learning_rate": 2.273859927625493e-05,
      "loss": 2.1083,
      "step": 201665
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.4000288248062134,
      "learning_rate": 2.2737037126405956e-05,
      "loss": 2.1744,
      "step": 201666
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.4125624895095825,
      "learning_rate": 2.2735475028105344e-05,
      "loss": 1.9354,
      "step": 201667
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.4466819763183594,
      "learning_rate": 2.273391298135353e-05,
      "loss": 2.1303,
      "step": 201668
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.4206486940383911,
      "learning_rate": 2.2732350986150617e-05,
      "loss": 1.6718,
      "step": 201669
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.3205918073654175,
      "learning_rate": 2.2730789042497065e-05,
      "loss": 1.9963,
      "step": 201670
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.493530511856079,
      "learning_rate": 2.2729227150393014e-05,
      "loss": 2.0217,
      "step": 201671
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.4949662685394287,
      "learning_rate": 2.2727665309838892e-05,
      "loss": 1.9255,
      "step": 201672
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.592223882675171,
      "learning_rate": 2.2726103520834905e-05,
      "loss": 2.0576,
      "step": 201673
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.4117144346237183,
      "learning_rate": 2.2724541783381445e-05,
      "loss": 1.7284,
      "step": 201674
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.4866242408752441,
      "learning_rate": 2.272298009747865e-05,
      "loss": 2.0308,
      "step": 201675
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.3872196674346924,
      "learning_rate": 2.2721418463126917e-05,
      "loss": 1.9732,
      "step": 201676
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.406672716140747,
      "learning_rate": 2.2719856880326447e-05,
      "loss": 1.7519,
      "step": 201677
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.4534800052642822,
      "learning_rate": 2.2718295349077675e-05,
      "loss": 2.2175,
      "step": 201678
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.3120542764663696,
      "learning_rate": 2.271673386938073e-05,
      "loss": 1.8064,
      "step": 201679
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.6070396900177002,
      "learning_rate": 2.2715172441236083e-05,
      "loss": 1.7501,
      "step": 201680
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.783537745475769,
      "learning_rate": 2.27136110646438e-05,
      "loss": 1.766,
      "step": 201681
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.6160005331039429,
      "learning_rate": 2.2712049739604378e-05,
      "loss": 1.8421,
      "step": 201682
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.3930506706237793,
      "learning_rate": 2.2710488466117948e-05,
      "loss": 1.9445,
      "step": 201683
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.3648850917816162,
      "learning_rate": 2.2708927244184917e-05,
      "loss": 1.9142,
      "step": 201684
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.5124232769012451,
      "learning_rate": 2.2707366073805476e-05,
      "loss": 1.8682,
      "step": 201685
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.3370378017425537,
      "learning_rate": 2.2705804954980066e-05,
      "loss": 1.8385,
      "step": 201686
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.463669776916504,
      "learning_rate": 2.270424388770885e-05,
      "loss": 2.1451,
      "step": 201687
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.3264001607894897,
      "learning_rate": 2.2702682871992128e-05,
      "loss": 1.8277,
      "step": 201688
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.5137594938278198,
      "learning_rate": 2.27011219078302e-05,
      "loss": 2.0436,
      "step": 201689
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.4001994132995605,
      "learning_rate": 2.2699560995223397e-05,
      "loss": 2.0301,
      "step": 201690
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.4262210130691528,
      "learning_rate": 2.269800013417192e-05,
      "loss": 2.2341,
      "step": 201691
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.4218600988388062,
      "learning_rate": 2.2696439324676174e-05,
      "loss": 2.1111,
      "step": 201692
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.4231767654418945,
      "learning_rate": 2.2694878566736418e-05,
      "loss": 2.0004,
      "step": 201693
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.5078624486923218,
      "learning_rate": 2.2693317860352888e-05,
      "loss": 2.062,
      "step": 201694
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.3918673992156982,
      "learning_rate": 2.2691757205525884e-05,
      "loss": 1.8271,
      "step": 201695
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.4925862550735474,
      "learning_rate": 2.269019660225574e-05,
      "loss": 1.9921,
      "step": 201696
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.5112853050231934,
      "learning_rate": 2.2688636050542686e-05,
      "loss": 2.2195,
      "step": 201697
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.5523234605789185,
      "learning_rate": 2.2687075550387125e-05,
      "loss": 2.0021,
      "step": 201698
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.4279353618621826,
      "learning_rate": 2.268551510178922e-05,
      "loss": 2.0248,
      "step": 201699
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.3603062629699707,
      "learning_rate": 2.2683954704749307e-05,
      "loss": 2.1803,
      "step": 201700
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.5190268754959106,
      "learning_rate": 2.268239435926772e-05,
      "loss": 1.9667,
      "step": 201701
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.4498056173324585,
      "learning_rate": 2.268083406534469e-05,
      "loss": 1.9858,
      "step": 201702
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.73770010471344,
      "learning_rate": 2.2679273822980482e-05,
      "loss": 1.9875,
      "step": 201703
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.514973521232605,
      "learning_rate": 2.26777136321755e-05,
      "loss": 1.9553,
      "step": 201704
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.4117153882980347,
      "learning_rate": 2.2676153492929973e-05,
      "loss": 2.0605,
      "step": 201705
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.4764894247055054,
      "learning_rate": 2.26745934052441e-05,
      "loss": 2.1232,
      "step": 201706
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.3865954875946045,
      "learning_rate": 2.2673033369118322e-05,
      "loss": 2.352,
      "step": 201707
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.3833087682724,
      "learning_rate": 2.2671473384552864e-05,
      "loss": 1.9191,
      "step": 201708
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.5011433362960815,
      "learning_rate": 2.2669913451547962e-05,
      "loss": 2.0329,
      "step": 201709
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.3932768106460571,
      "learning_rate": 2.2668353570104015e-05,
      "loss": 2.0514,
      "step": 201710
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.4137444496154785,
      "learning_rate": 2.2666793740221258e-05,
      "loss": 1.8987,
      "step": 201711
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.3374500274658203,
      "learning_rate": 2.266523396189992e-05,
      "loss": 1.8836,
      "step": 201712
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.4682190418243408,
      "learning_rate": 2.2663674235140405e-05,
      "loss": 2.0394,
      "step": 201713
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.553121566772461,
      "learning_rate": 2.266211455994288e-05,
      "loss": 2.0778,
      "step": 201714
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.3448816537857056,
      "learning_rate": 2.2660554936307772e-05,
      "loss": 1.7818,
      "step": 201715
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.6234148740768433,
      "learning_rate": 2.2658995364235288e-05,
      "loss": 2.0067,
      "step": 201716
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.4365705251693726,
      "learning_rate": 2.2657435843725758e-05,
      "loss": 2.0606,
      "step": 201717
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.6498947143554688,
      "learning_rate": 2.265587637477938e-05,
      "loss": 1.8132,
      "step": 201718
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.379715085029602,
      "learning_rate": 2.2654316957396558e-05,
      "loss": 2.0379,
      "step": 201719
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.4811115264892578,
      "learning_rate": 2.2652757591577487e-05,
      "loss": 2.1429,
      "step": 201720
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.4107208251953125,
      "learning_rate": 2.265119827732257e-05,
      "loss": 1.98,
      "step": 201721
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.859749674797058,
      "learning_rate": 2.2649639014631936e-05,
      "loss": 1.9525,
      "step": 201722
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.364660620689392,
      "learning_rate": 2.2648079803506093e-05,
      "loss": 2.1014,
      "step": 201723
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.4480282068252563,
      "learning_rate": 2.264652064394513e-05,
      "loss": 1.8727,
      "step": 201724
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.3666009902954102,
      "learning_rate": 2.2644961535949426e-05,
      "loss": 1.9847,
      "step": 201725
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.4122790098190308,
      "learning_rate": 2.2643402479519236e-05,
      "loss": 2.0014,
      "step": 201726
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.5617196559906006,
      "learning_rate": 2.264184347465493e-05,
      "loss": 2.0968,
      "step": 201727
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.5020302534103394,
      "learning_rate": 2.2640284521356677e-05,
      "loss": 2.0195,
      "step": 201728
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.4822810888290405,
      "learning_rate": 2.263872561962491e-05,
      "loss": 1.9602,
      "step": 201729
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.5330870151519775,
      "learning_rate": 2.2637166769459757e-05,
      "loss": 1.9136,
      "step": 201730
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.5702285766601562,
      "learning_rate": 2.263560797086166e-05,
      "loss": 2.0436,
      "step": 201731
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.4814831018447876,
      "learning_rate": 2.2634049223830775e-05,
      "loss": 2.0551,
      "step": 201732
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.5929054021835327,
      "learning_rate": 2.2632490528367507e-05,
      "loss": 2.0993,
      "step": 201733
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.3818503618240356,
      "learning_rate": 2.2630931884472026e-05,
      "loss": 1.882,
      "step": 201734
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.3328222036361694,
      "learning_rate": 2.262937329214479e-05,
      "loss": 1.6148,
      "step": 201735
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.3701391220092773,
      "learning_rate": 2.2627814751385908e-05,
      "loss": 1.8707,
      "step": 201736
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.474233627319336,
      "learning_rate": 2.2626256262195807e-05,
      "loss": 1.915,
      "step": 201737
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.2942219972610474,
      "learning_rate": 2.2624697824574657e-05,
      "loss": 1.9856,
      "step": 201738
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.4077883958816528,
      "learning_rate": 2.2623139438522886e-05,
      "loss": 2.1292,
      "step": 201739
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.419938087463379,
      "learning_rate": 2.2621581104040632e-05,
      "loss": 2.1018,
      "step": 201740
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.5237804651260376,
      "learning_rate": 2.262002282112836e-05,
      "loss": 1.935,
      "step": 201741
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.5096772909164429,
      "learning_rate": 2.2618464589786167e-05,
      "loss": 1.8994,
      "step": 201742
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.3925532102584839,
      "learning_rate": 2.261690641001449e-05,
      "loss": 1.9483,
      "step": 201743
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.4368107318878174,
      "learning_rate": 2.2615348281813528e-05,
      "loss": 1.8711,
      "step": 201744
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.4215877056121826,
      "learning_rate": 2.2613790205183645e-05,
      "loss": 1.9524,
      "step": 201745
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.4511289596557617,
      "learning_rate": 2.2612232180125044e-05,
      "loss": 1.855,
      "step": 201746
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.4175965785980225,
      "learning_rate": 2.261067420663816e-05,
      "loss": 2.1612,
      "step": 201747
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.4586796760559082,
      "learning_rate": 2.2609116284723116e-05,
      "loss": 2.1173,
      "step": 201748
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.5476866960525513,
      "learning_rate": 2.260755841438029e-05,
      "loss": 2.1689,
      "step": 201749
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.4520206451416016,
      "learning_rate": 2.260600059560991e-05,
      "loss": 1.9916,
      "step": 201750
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.5111701488494873,
      "learning_rate": 2.2604442828412373e-05,
      "loss": 1.8313,
      "step": 201751
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.3767846822738647,
      "learning_rate": 2.260288511278785e-05,
      "loss": 2.0478,
      "step": 201752
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.414086937904358,
      "learning_rate": 2.2601327448736806e-05,
      "loss": 2.0345,
      "step": 201753
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.3596347570419312,
      "learning_rate": 2.2599769836259273e-05,
      "loss": 1.9576,
      "step": 201754
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.4215537309646606,
      "learning_rate": 2.2598212275355753e-05,
      "loss": 1.9464,
      "step": 201755
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.551153540611267,
      "learning_rate": 2.2596654766026413e-05,
      "loss": 1.9494,
      "step": 201756
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.573349952697754,
      "learning_rate": 2.259509730827165e-05,
      "loss": 2.1446,
      "step": 201757
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.5135412216186523,
      "learning_rate": 2.259353990209163e-05,
      "loss": 1.9766,
      "step": 201758
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.3540853261947632,
      "learning_rate": 2.259198254748682e-05,
      "loss": 1.9194,
      "step": 201759
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.4884672164916992,
      "learning_rate": 2.2590425244457288e-05,
      "loss": 1.8561,
      "step": 201760
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.507245659828186,
      "learning_rate": 2.2588867993003466e-05,
      "loss": 2.1686,
      "step": 201761
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.3875749111175537,
      "learning_rate": 2.258731079312559e-05,
      "loss": 2.0669,
      "step": 201762
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.3961809873580933,
      "learning_rate": 2.258575364482399e-05,
      "loss": 1.7945,
      "step": 201763
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.4604960680007935,
      "learning_rate": 2.2584196548098898e-05,
      "loss": 1.967,
      "step": 201764
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.4140433073043823,
      "learning_rate": 2.258263950295075e-05,
      "loss": 1.9043,
      "step": 201765
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.4293487071990967,
      "learning_rate": 2.258108250937961e-05,
      "loss": 2.0147,
      "step": 201766
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.472618818283081,
      "learning_rate": 2.2579525567385946e-05,
      "loss": 1.9651,
      "step": 201767
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.3518563508987427,
      "learning_rate": 2.2577968676969926e-05,
      "loss": 1.8681,
      "step": 201768
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.6646326780319214,
      "learning_rate": 2.257641183813198e-05,
      "loss": 2.0005,
      "step": 201769
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.3271377086639404,
      "learning_rate": 2.257485505087221e-05,
      "loss": 1.975,
      "step": 201770
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.4262467622756958,
      "learning_rate": 2.2573298315191145e-05,
      "loss": 1.9311,
      "step": 201771
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.462380051612854,
      "learning_rate": 2.2571741631088858e-05,
      "loss": 2.0656,
      "step": 201772
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.350163221359253,
      "learning_rate": 2.2570184998565745e-05,
      "loss": 2.0603,
      "step": 201773
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.5944479703903198,
      "learning_rate": 2.2568628417622005e-05,
      "loss": 1.9185,
      "step": 201774
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.6368738412857056,
      "learning_rate": 2.256707188825807e-05,
      "loss": 1.9309,
      "step": 201775
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.3844385147094727,
      "learning_rate": 2.256551541047411e-05,
      "loss": 1.8156,
      "step": 201776
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.3874661922454834,
      "learning_rate": 2.256395898427049e-05,
      "loss": 2.0901,
      "step": 201777
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.4606735706329346,
      "learning_rate": 2.256240260964748e-05,
      "loss": 2.1057,
      "step": 201778
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.5225085020065308,
      "learning_rate": 2.256084628660534e-05,
      "loss": 1.9019,
      "step": 201779
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.4101530313491821,
      "learning_rate": 2.2559290015144338e-05,
      "loss": 1.8289,
      "step": 201780
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.533397912979126,
      "learning_rate": 2.2557733795264844e-05,
      "loss": 1.9376,
      "step": 201781
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.6232153177261353,
      "learning_rate": 2.2556177626967055e-05,
      "loss": 2.1,
      "step": 201782
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.405674695968628,
      "learning_rate": 2.255462151025137e-05,
      "loss": 1.7575,
      "step": 201783
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.4012885093688965,
      "learning_rate": 2.2553065445118023e-05,
      "loss": 2.1132,
      "step": 201784
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.3905740976333618,
      "learning_rate": 2.2551509431567217e-05,
      "loss": 1.8198,
      "step": 201785
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.4548124074935913,
      "learning_rate": 2.2549953469599414e-05,
      "loss": 2.0069,
      "step": 201786
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.3962082862854004,
      "learning_rate": 2.254839755921478e-05,
      "loss": 1.7234,
      "step": 201787
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.6497405767440796,
      "learning_rate": 2.254684170041359e-05,
      "loss": 1.8953,
      "step": 201788
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.7195544242858887,
      "learning_rate": 2.254528589319623e-05,
      "loss": 1.9389,
      "step": 201789
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.5092324018478394,
      "learning_rate": 2.2543730137562943e-05,
      "loss": 2.0778,
      "step": 201790
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.3947710990905762,
      "learning_rate": 2.254217443351396e-05,
      "loss": 1.8801,
      "step": 201791
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.477629542350769,
      "learning_rate": 2.254061878104968e-05,
      "loss": 1.8997,
      "step": 201792
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.4427229166030884,
      "learning_rate": 2.2539063180170336e-05,
      "loss": 2.1189,
      "step": 201793
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.531428337097168,
      "learning_rate": 2.2537507630876195e-05,
      "loss": 1.9165,
      "step": 201794
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.552400827407837,
      "learning_rate": 2.253595213316759e-05,
      "loss": 2.4345,
      "step": 201795
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.459382176399231,
      "learning_rate": 2.2534396687044786e-05,
      "loss": 2.0537,
      "step": 201796
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.4970452785491943,
      "learning_rate": 2.253284129250802e-05,
      "loss": 1.9266,
      "step": 201797
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.4989075660705566,
      "learning_rate": 2.2531285949557688e-05,
      "loss": 1.9509,
      "step": 201798
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.3493376970291138,
      "learning_rate": 2.252973065819399e-05,
      "loss": 1.9491,
      "step": 201799
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.4378199577331543,
      "learning_rate": 2.2528175418417295e-05,
      "loss": 2.0187,
      "step": 201800
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.4380801916122437,
      "learning_rate": 2.2526620230227865e-05,
      "loss": 2.0265,
      "step": 201801
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.6175724267959595,
      "learning_rate": 2.252506509362597e-05,
      "loss": 2.0048,
      "step": 201802
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.3820794820785522,
      "learning_rate": 2.2523510008611844e-05,
      "loss": 1.7648,
      "step": 201803
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.48831307888031,
      "learning_rate": 2.2521954975185884e-05,
      "loss": 2.1046,
      "step": 201804
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.4999006986618042,
      "learning_rate": 2.2520399993348293e-05,
      "loss": 2.1037,
      "step": 201805
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.4960781335830688,
      "learning_rate": 2.2518845063099433e-05,
      "loss": 1.9504,
      "step": 201806
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.6230783462524414,
      "learning_rate": 2.2517290184439542e-05,
      "loss": 1.98,
      "step": 201807
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.5392255783081055,
      "learning_rate": 2.2515735357368948e-05,
      "loss": 1.8787,
      "step": 201808
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.559573769569397,
      "learning_rate": 2.2514180581887885e-05,
      "loss": 1.8168,
      "step": 201809
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.4604392051696777,
      "learning_rate": 2.251262585799669e-05,
      "loss": 1.7498,
      "step": 201810
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.5569244623184204,
      "learning_rate": 2.2511071185695595e-05,
      "loss": 2.0094,
      "step": 201811
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.642412781715393,
      "learning_rate": 2.2509516564984998e-05,
      "loss": 1.8968,
      "step": 201812
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.5984218120574951,
      "learning_rate": 2.250796199586503e-05,
      "loss": 2.008,
      "step": 201813
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.591149926185608,
      "learning_rate": 2.2506407478336196e-05,
      "loss": 1.7225,
      "step": 201814
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.4433913230895996,
      "learning_rate": 2.250485301239856e-05,
      "loss": 2.078,
      "step": 201815
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.4136415719985962,
      "learning_rate": 2.2503298598052555e-05,
      "loss": 1.973,
      "step": 201816
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.6419304609298706,
      "learning_rate": 2.250174423529838e-05,
      "loss": 1.9777,
      "step": 201817
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.7311092615127563,
      "learning_rate": 2.2500189924136437e-05,
      "loss": 2.3086,
      "step": 201818
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.2765440940856934,
      "learning_rate": 2.2498635664566855e-05,
      "loss": 1.9021,
      "step": 201819
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.5703397989273071,
      "learning_rate": 2.249708145659014e-05,
      "loss": 1.9509,
      "step": 201820
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.5091615915298462,
      "learning_rate": 2.249552730020635e-05,
      "loss": 1.9886,
      "step": 201821
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.6366641521453857,
      "learning_rate": 2.2493973195415927e-05,
      "loss": 1.7618,
      "step": 201822
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.6066631078720093,
      "learning_rate": 2.2492419142219065e-05,
      "loss": 1.9388,
      "step": 201823
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.3906430006027222,
      "learning_rate": 2.2490865140616166e-05,
      "loss": 2.0714,
      "step": 201824
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.4091627597808838,
      "learning_rate": 2.2489311190607396e-05,
      "loss": 1.974,
      "step": 201825
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.3297938108444214,
      "learning_rate": 2.2487757292193188e-05,
      "loss": 2.0936,
      "step": 201826
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.4820551872253418,
      "learning_rate": 2.2486203445373674e-05,
      "loss": 1.9775,
      "step": 201827
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.5369155406951904,
      "learning_rate": 2.248464965014922e-05,
      "loss": 1.9401,
      "step": 201828
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.4264870882034302,
      "learning_rate": 2.24830959065201e-05,
      "loss": 2.2561,
      "step": 201829
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.5489035844802856,
      "learning_rate": 2.2481542214486638e-05,
      "loss": 1.8252,
      "step": 201830
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.5656541585922241,
      "learning_rate": 2.2479988574049033e-05,
      "loss": 2.1192,
      "step": 201831
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.5815078020095825,
      "learning_rate": 2.247843498520776e-05,
      "loss": 1.9907,
      "step": 201832
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.4847136735916138,
      "learning_rate": 2.247688144796288e-05,
      "loss": 1.8845,
      "step": 201833
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.819115400314331,
      "learning_rate": 2.2475327962314828e-05,
      "loss": 1.7828,
      "step": 201834
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.5068856477737427,
      "learning_rate": 2.24737745282638e-05,
      "loss": 2.0433,
      "step": 201835
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.407686710357666,
      "learning_rate": 2.24722211458102e-05,
      "loss": 2.0807,
      "step": 201836
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.469548225402832,
      "learning_rate": 2.247066781495419e-05,
      "loss": 1.8616,
      "step": 201837
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.5873702764511108,
      "learning_rate": 2.2469114535696243e-05,
      "loss": 2.1027,
      "step": 201838
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.65702223777771,
      "learning_rate": 2.2467561308036387e-05,
      "loss": 1.9244,
      "step": 201839
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.609915852546692,
      "learning_rate": 2.2466008131975157e-05,
      "loss": 1.9505,
      "step": 201840
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.3177874088287354,
      "learning_rate": 2.246445500751265e-05,
      "loss": 1.8961,
      "step": 201841
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.365206241607666,
      "learning_rate": 2.246290193464927e-05,
      "loss": 2.0224,
      "step": 201842
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.384200096130371,
      "learning_rate": 2.2461348913385245e-05,
      "loss": 1.9019,
      "step": 201843
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.4320601224899292,
      "learning_rate": 2.2459795943721014e-05,
      "loss": 1.9605,
      "step": 201844
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.4198254346847534,
      "learning_rate": 2.2458243025656606e-05,
      "loss": 1.858,
      "step": 201845
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.5904886722564697,
      "learning_rate": 2.2456690159192525e-05,
      "loss": 1.8777,
      "step": 201846
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.627129316329956,
      "learning_rate": 2.2455137344328935e-05,
      "loss": 2.1007,
      "step": 201847
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.5153602361679077,
      "learning_rate": 2.2453584581066232e-05,
      "loss": 1.993,
      "step": 201848
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.4290987253189087,
      "learning_rate": 2.2452031869404584e-05,
      "loss": 1.9276,
      "step": 201849
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.3864383697509766,
      "learning_rate": 2.2450479209344397e-05,
      "loss": 1.8641,
      "step": 201850
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.3679534196853638,
      "learning_rate": 2.2448926600885897e-05,
      "loss": 1.9013,
      "step": 201851
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.4580559730529785,
      "learning_rate": 2.2447374044029386e-05,
      "loss": 1.9047,
      "step": 201852
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.3945599794387817,
      "learning_rate": 2.2445821538775098e-05,
      "loss": 2.0321,
      "step": 201853
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.3422304391860962,
      "learning_rate": 2.2444269085123403e-05,
      "loss": 1.9037,
      "step": 201854
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.5290858745574951,
      "learning_rate": 2.2442716683074523e-05,
      "loss": 2.1411,
      "step": 201855
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.4256783723831177,
      "learning_rate": 2.2441164332628835e-05,
      "loss": 2.036,
      "step": 201856
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.4830412864685059,
      "learning_rate": 2.243961203378657e-05,
      "loss": 2.0072,
      "step": 201857
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.4959712028503418,
      "learning_rate": 2.2438059786548024e-05,
      "loss": 2.0965,
      "step": 201858
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.49961256980896,
      "learning_rate": 2.2436507590913434e-05,
      "loss": 2.0136,
      "step": 201859
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.488655686378479,
      "learning_rate": 2.2434955446883197e-05,
      "loss": 1.8496,
      "step": 201860
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.72885000705719,
      "learning_rate": 2.2433403354457446e-05,
      "loss": 1.9375,
      "step": 201861
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.3719556331634521,
      "learning_rate": 2.243185131363665e-05,
      "loss": 1.7833,
      "step": 201862
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.4996848106384277,
      "learning_rate": 2.243029932442101e-05,
      "loss": 1.9512,
      "step": 201863
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.4752562046051025,
      "learning_rate": 2.242874738681082e-05,
      "loss": 1.9449,
      "step": 201864
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.370652437210083,
      "learning_rate": 2.2427195500806316e-05,
      "loss": 2.0848,
      "step": 201865
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.488480806350708,
      "learning_rate": 2.2425643666407866e-05,
      "loss": 2.0951,
      "step": 201866
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.5393527746200562,
      "learning_rate": 2.2424091883615704e-05,
      "loss": 2.0046,
      "step": 201867
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.4861314296722412,
      "learning_rate": 2.242254015243019e-05,
      "loss": 2.1044,
      "step": 201868
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.5510821342468262,
      "learning_rate": 2.2420988472851564e-05,
      "loss": 1.8255,
      "step": 201869
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.3564363718032837,
      "learning_rate": 2.241943684488009e-05,
      "loss": 1.8814,
      "step": 201870
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.580476999282837,
      "learning_rate": 2.2417885268516068e-05,
      "loss": 2.1808,
      "step": 201871
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.416494369506836,
      "learning_rate": 2.241633374375983e-05,
      "loss": 1.9909,
      "step": 201872
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.4174524545669556,
      "learning_rate": 2.2414782270611575e-05,
      "loss": 2.1146,
      "step": 201873
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.4842684268951416,
      "learning_rate": 2.2413230849071738e-05,
      "loss": 2.0202,
      "step": 201874
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.4010339975357056,
      "learning_rate": 2.2411679479140487e-05,
      "loss": 2.1074,
      "step": 201875
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.346601128578186,
      "learning_rate": 2.241012816081812e-05,
      "loss": 1.8332,
      "step": 201876
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.536688208580017,
      "learning_rate": 2.240857689410497e-05,
      "loss": 2.0821,
      "step": 201877
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.4852266311645508,
      "learning_rate": 2.2407025679001332e-05,
      "loss": 1.9219,
      "step": 201878
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.3679455518722534,
      "learning_rate": 2.2405474515507416e-05,
      "loss": 1.9361,
      "step": 201879
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.4397552013397217,
      "learning_rate": 2.240392340362358e-05,
      "loss": 2.0095,
      "step": 201880
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.448194146156311,
      "learning_rate": 2.240237234335013e-05,
      "loss": 2.0336,
      "step": 201881
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.3278425931930542,
      "learning_rate": 2.240082133468726e-05,
      "loss": 2.0168,
      "step": 201882
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.4918816089630127,
      "learning_rate": 2.239927037763537e-05,
      "loss": 2.045,
      "step": 201883
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.5180044174194336,
      "learning_rate": 2.239771947219463e-05,
      "loss": 2.0041,
      "step": 201884
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.4551900625228882,
      "learning_rate": 2.2396168618365473e-05,
      "loss": 1.8627,
      "step": 201885
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.4066509008407593,
      "learning_rate": 2.2394617816148097e-05,
      "loss": 1.943,
      "step": 201886
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.5022666454315186,
      "learning_rate": 2.239306706554277e-05,
      "loss": 1.9125,
      "step": 201887
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.3558837175369263,
      "learning_rate": 2.239151636654979e-05,
      "loss": 2.2406,
      "step": 201888
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.529183268547058,
      "learning_rate": 2.2389965719169524e-05,
      "loss": 1.9667,
      "step": 201889
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.4567514657974243,
      "learning_rate": 2.2388415123402138e-05,
      "loss": 1.9049,
      "step": 201890
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.4847502708435059,
      "learning_rate": 2.2386864579248064e-05,
      "loss": 1.9423,
      "step": 201891
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.5129834413528442,
      "learning_rate": 2.2385314086707472e-05,
      "loss": 1.8606,
      "step": 201892
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.3966608047485352,
      "learning_rate": 2.2383763645780727e-05,
      "loss": 2.2698,
      "step": 201893
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.4805835485458374,
      "learning_rate": 2.2382213256467996e-05,
      "loss": 1.9475,
      "step": 201894
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.3894083499908447,
      "learning_rate": 2.2380662918769743e-05,
      "loss": 1.9496,
      "step": 201895
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.4980676174163818,
      "learning_rate": 2.237911263268607e-05,
      "loss": 1.8959,
      "step": 201896
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.504889965057373,
      "learning_rate": 2.2377562398217443e-05,
      "loss": 2.0885,
      "step": 201897
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.5104198455810547,
      "learning_rate": 2.2376012215363992e-05,
      "loss": 1.9142,
      "step": 201898
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.411577820777893,
      "learning_rate": 2.237446208412622e-05,
      "loss": 1.833,
      "step": 201899
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.456864356994629,
      "learning_rate": 2.2372912004504128e-05,
      "loss": 2.0276,
      "step": 201900
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.4299017190933228,
      "learning_rate": 2.2371361976498248e-05,
      "loss": 1.9759,
      "step": 201901
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.3882808685302734,
      "learning_rate": 2.2369812000108678e-05,
      "loss": 2.1183,
      "step": 201902
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.4541170597076416,
      "learning_rate": 2.2368262075335886e-05,
      "loss": 2.3097,
      "step": 201903
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.6847096681594849,
      "learning_rate": 2.2366712202180005e-05,
      "loss": 2.3252,
      "step": 201904
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.4619059562683105,
      "learning_rate": 2.23651623806415e-05,
      "loss": 2.1041,
      "step": 201905
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.5006215572357178,
      "learning_rate": 2.2363612610720472e-05,
      "loss": 1.98,
      "step": 201906
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.734855055809021,
      "learning_rate": 2.236206289241732e-05,
      "loss": 1.9154,
      "step": 201907
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.378820538520813,
      "learning_rate": 2.2360513225732247e-05,
      "loss": 1.7841,
      "step": 201908
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.4408634901046753,
      "learning_rate": 2.2358963610665648e-05,
      "loss": 1.8919,
      "step": 201909
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.3876228332519531,
      "learning_rate": 2.235741404721769e-05,
      "loss": 1.9256,
      "step": 201910
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.467430830001831,
      "learning_rate": 2.2355864535388877e-05,
      "loss": 2.1005,
      "step": 201911
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.7387723922729492,
      "learning_rate": 2.2354315075179207e-05,
      "loss": 2.0635,
      "step": 201912
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.4620085954666138,
      "learning_rate": 2.2352765666589178e-05,
      "loss": 2.1274,
      "step": 201913
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.4700285196304321,
      "learning_rate": 2.235121630961896e-05,
      "loss": 2.038,
      "step": 201914
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.4961553812026978,
      "learning_rate": 2.2349667004268945e-05,
      "loss": 2.0858,
      "step": 201915
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.3590497970581055,
      "learning_rate": 2.2348117750539306e-05,
      "loss": 2.0676,
      "step": 201916
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.3784899711608887,
      "learning_rate": 2.2346568548430445e-05,
      "loss": 1.8642,
      "step": 201917
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.5106204748153687,
      "learning_rate": 2.234501939794262e-05,
      "loss": 2.1108,
      "step": 201918
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.3160464763641357,
      "learning_rate": 2.2343470299076073e-05,
      "loss": 1.9986,
      "step": 201919
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.4572780132293701,
      "learning_rate": 2.2341921251831062e-05,
      "loss": 2.0016,
      "step": 201920
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.4252207279205322,
      "learning_rate": 2.2340372256207995e-05,
      "loss": 1.7781,
      "step": 201921
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.3965941667556763,
      "learning_rate": 2.2338823312207034e-05,
      "loss": 1.9114,
      "step": 201922
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.4978729486465454,
      "learning_rate": 2.233727441982861e-05,
      "loss": 1.9451,
      "step": 201923
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.535507321357727,
      "learning_rate": 2.2335725579072894e-05,
      "loss": 2.0612,
      "step": 201924
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.5210609436035156,
      "learning_rate": 2.2334176789940218e-05,
      "loss": 2.0255,
      "step": 201925
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.5393232107162476,
      "learning_rate": 2.233262805243081e-05,
      "loss": 1.9516,
      "step": 201926
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.5180320739746094,
      "learning_rate": 2.2331079366545047e-05,
      "loss": 2.1736,
      "step": 201927
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.4561952352523804,
      "learning_rate": 2.232953073228312e-05,
      "loss": 1.9616,
      "step": 201928
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.3679267168045044,
      "learning_rate": 2.232798214964546e-05,
      "loss": 1.9672,
      "step": 201929
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.5201977491378784,
      "learning_rate": 2.2326433618632245e-05,
      "loss": 2.0205,
      "step": 201930
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.5914137363433838,
      "learning_rate": 2.2324885139243796e-05,
      "loss": 1.9987,
      "step": 201931
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.3671077489852905,
      "learning_rate": 2.2323336711480322e-05,
      "loss": 1.9618,
      "step": 201932
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.3553639650344849,
      "learning_rate": 2.232178833534225e-05,
      "loss": 1.5956,
      "step": 201933
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.4868323802947998,
      "learning_rate": 2.232024001082975e-05,
      "loss": 1.9999,
      "step": 201934
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.5971828699111938,
      "learning_rate": 2.2318691737943217e-05,
      "loss": 1.908,
      "step": 201935
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.5662797689437866,
      "learning_rate": 2.2317143516682855e-05,
      "loss": 1.9432,
      "step": 201936
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.4354006052017212,
      "learning_rate": 2.2315595347049e-05,
      "loss": 1.9683,
      "step": 201937
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.4843764305114746,
      "learning_rate": 2.231404722904184e-05,
      "loss": 1.8958,
      "step": 201938
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.2888935804367065,
      "learning_rate": 2.2312499162661823e-05,
      "loss": 1.8817,
      "step": 201939
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.4620329141616821,
      "learning_rate": 2.2310951147909106e-05,
      "loss": 2.0572,
      "step": 201940
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.58242666721344,
      "learning_rate": 2.2309403184784058e-05,
      "loss": 2.1737,
      "step": 201941
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.5307446718215942,
      "learning_rate": 2.230785527328691e-05,
      "loss": 2.1655,
      "step": 201942
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.43256676197052,
      "learning_rate": 2.2306307413418e-05,
      "loss": 2.0331,
      "step": 201943
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.4390274286270142,
      "learning_rate": 2.2304759605177558e-05,
      "loss": 2.1192,
      "step": 201944
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.3429393768310547,
      "learning_rate": 2.2303211848565915e-05,
      "loss": 1.9648,
      "step": 201945
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.4258497953414917,
      "learning_rate": 2.2301664143583342e-05,
      "loss": 1.6696,
      "step": 201946
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.6507309675216675,
      "learning_rate": 2.2300116490230134e-05,
      "loss": 1.7543,
      "step": 201947
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.492760181427002,
      "learning_rate": 2.2298568888506597e-05,
      "loss": 1.9754,
      "step": 201948
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.4606761932373047,
      "learning_rate": 2.2297021338413025e-05,
      "loss": 1.9056,
      "step": 201949
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.4033151865005493,
      "learning_rate": 2.2295473839949583e-05,
      "loss": 1.9678,
      "step": 201950
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.4185078144073486,
      "learning_rate": 2.2293926393116747e-05,
      "loss": 1.9981,
      "step": 201951
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.4245598316192627,
      "learning_rate": 2.2292378997914608e-05,
      "loss": 2.0445,
      "step": 201952
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.6462775468826294,
      "learning_rate": 2.2290831654343667e-05,
      "loss": 2.0001,
      "step": 201953
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.4996916055679321,
      "learning_rate": 2.228928436240406e-05,
      "loss": 2.0244,
      "step": 201954
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.436163067817688,
      "learning_rate": 2.2287737122096118e-05,
      "loss": 2.0558,
      "step": 201955
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.4586689472198486,
      "learning_rate": 2.2286189933420106e-05,
      "loss": 1.7875,
      "step": 201956
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.4865001440048218,
      "learning_rate": 2.228464279637636e-05,
      "loss": 2.0101,
      "step": 201957
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.4459295272827148,
      "learning_rate": 2.2283095710965117e-05,
      "loss": 1.9593,
      "step": 201958
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.535658836364746,
      "learning_rate": 2.22815486771867e-05,
      "loss": 2.074,
      "step": 201959
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.541641354560852,
      "learning_rate": 2.2280001695041416e-05,
      "loss": 2.0685,
      "step": 201960
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.4554023742675781,
      "learning_rate": 2.2278454764529463e-05,
      "loss": 1.9634,
      "step": 201961
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.4134029150009155,
      "learning_rate": 2.2276907885651242e-05,
      "loss": 2.0506,
      "step": 201962
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.4844906330108643,
      "learning_rate": 2.227536105840698e-05,
      "loss": 2.0093,
      "step": 201963
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.4543567895889282,
      "learning_rate": 2.227381428279692e-05,
      "loss": 2.1829,
      "step": 201964
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.4612817764282227,
      "learning_rate": 2.2272267558821456e-05,
      "loss": 1.9238,
      "step": 201965
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.5268739461898804,
      "learning_rate": 2.2270720886480818e-05,
      "loss": 2.1159,
      "step": 201966
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.4473378658294678,
      "learning_rate": 2.2269174265775248e-05,
      "loss": 2.0443,
      "step": 201967
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.4900730848312378,
      "learning_rate": 2.226762769670514e-05,
      "loss": 2.0562,
      "step": 201968
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.4008008241653442,
      "learning_rate": 2.2266081179270656e-05,
      "loss": 2.074,
      "step": 201969
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.340433955192566,
      "learning_rate": 2.2264534713472206e-05,
      "loss": 1.9953,
      "step": 201970
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.3570393323898315,
      "learning_rate": 2.2262988299310013e-05,
      "loss": 1.7558,
      "step": 201971
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.4581713676452637,
      "learning_rate": 2.2261441936784384e-05,
      "loss": 1.9136,
      "step": 201972
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.4803324937820435,
      "learning_rate": 2.225989562589555e-05,
      "loss": 2.1215,
      "step": 201973
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.3900425434112549,
      "learning_rate": 2.225834936664388e-05,
      "loss": 1.801,
      "step": 201974
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.471251130104065,
      "learning_rate": 2.22568031590296e-05,
      "loss": 2.0568,
      "step": 201975
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.4762252569198608,
      "learning_rate": 2.225525700305305e-05,
      "loss": 2.0106,
      "step": 201976
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.3913331031799316,
      "learning_rate": 2.2253710898714494e-05,
      "loss": 1.9784,
      "step": 201977
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.4246630668640137,
      "learning_rate": 2.2252164846014165e-05,
      "loss": 2.0277,
      "step": 201978
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.3553599119186401,
      "learning_rate": 2.225061884495246e-05,
      "loss": 2.0888,
      "step": 201979
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.5214877128601074,
      "learning_rate": 2.2249072895529585e-05,
      "loss": 2.1294,
      "step": 201980
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.479283094406128,
      "learning_rate": 2.2247526997745836e-05,
      "loss": 1.8244,
      "step": 201981
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.4517043828964233,
      "learning_rate": 2.2245981151601545e-05,
      "loss": 2.0117,
      "step": 201982
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.4947439432144165,
      "learning_rate": 2.2244435357096913e-05,
      "loss": 2.2055,
      "step": 201983
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.5111172199249268,
      "learning_rate": 2.224288961423234e-05,
      "loss": 2.1108,
      "step": 201984
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.4031496047973633,
      "learning_rate": 2.224134392300806e-05,
      "loss": 2.0419,
      "step": 201985
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.3849037885665894,
      "learning_rate": 2.223979828342437e-05,
      "loss": 2.1103,
      "step": 201986
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.3288700580596924,
      "learning_rate": 2.2238252695481472e-05,
      "loss": 1.9645,
      "step": 201987
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.371848225593567,
      "learning_rate": 2.2236707159179767e-05,
      "loss": 2.157,
      "step": 201988
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.6074858903884888,
      "learning_rate": 2.2235161674519485e-05,
      "loss": 2.1073,
      "step": 201989
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.3194544315338135,
      "learning_rate": 2.2233616241500962e-05,
      "loss": 1.7251,
      "step": 201990
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.6043603420257568,
      "learning_rate": 2.2232070860124463e-05,
      "loss": 2.0937,
      "step": 201991
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.4032093286514282,
      "learning_rate": 2.2230525530390254e-05,
      "loss": 2.1736,
      "step": 201992
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.4872970581054688,
      "learning_rate": 2.2228980252298567e-05,
      "loss": 1.8503,
      "step": 201993
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.612581491470337,
      "learning_rate": 2.222743502584984e-05,
      "loss": 2.0439,
      "step": 201994
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.4148235321044922,
      "learning_rate": 2.2225889851044197e-05,
      "loss": 1.8226,
      "step": 201995
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.4403283596038818,
      "learning_rate": 2.2224344727882083e-05,
      "loss": 2.2363,
      "step": 201996
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.4760583639144897,
      "learning_rate": 2.2222799656363687e-05,
      "loss": 1.8382,
      "step": 201997
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.4012407064437866,
      "learning_rate": 2.2221254636489316e-05,
      "loss": 1.854,
      "step": 201998
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.4808084964752197,
      "learning_rate": 2.2219709668259232e-05,
      "loss": 2.0073,
      "step": 201999
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.5485398769378662,
      "learning_rate": 2.2218164751673773e-05,
      "loss": 2.0561,
      "step": 202000
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.3905744552612305,
      "learning_rate": 2.2216619886733133e-05,
      "loss": 2.0308,
      "step": 202001
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.5196728706359863,
      "learning_rate": 2.2215075073437748e-05,
      "loss": 1.9965,
      "step": 202002
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.4291859865188599,
      "learning_rate": 2.221353031178782e-05,
      "loss": 1.7806,
      "step": 202003
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.456714153289795,
      "learning_rate": 2.2211985601783643e-05,
      "loss": 2.0703,
      "step": 202004
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.4410247802734375,
      "learning_rate": 2.2210440943425455e-05,
      "loss": 2.2321,
      "step": 202005
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.3923898935317993,
      "learning_rate": 2.220889633671362e-05,
      "loss": 1.8374,
      "step": 202006
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.48159658908844,
      "learning_rate": 2.2207351781648342e-05,
      "loss": 1.8841,
      "step": 202007
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.5064761638641357,
      "learning_rate": 2.2205807278230014e-05,
      "loss": 1.9492,
      "step": 202008
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.7134560346603394,
      "learning_rate": 2.2204262826458906e-05,
      "loss": 1.8088,
      "step": 202009
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.5158984661102295,
      "learning_rate": 2.220271842633522e-05,
      "loss": 1.7294,
      "step": 202010
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.4816701412200928,
      "learning_rate": 2.2201174077859284e-05,
      "loss": 1.9536,
      "step": 202011
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.4065253734588623,
      "learning_rate": 2.2199629781031435e-05,
      "loss": 1.9741,
      "step": 202012
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.5963354110717773,
      "learning_rate": 2.219808553585184e-05,
      "loss": 2.0169,
      "step": 202013
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.714698314666748,
      "learning_rate": 2.2196541342320927e-05,
      "loss": 2.1672,
      "step": 202014
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.4752094745635986,
      "learning_rate": 2.2194997200438934e-05,
      "loss": 1.8562,
      "step": 202015
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.39736807346344,
      "learning_rate": 2.2193453110206127e-05,
      "loss": 1.8071,
      "step": 202016
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.517243504524231,
      "learning_rate": 2.219190907162277e-05,
      "loss": 1.8617,
      "step": 202017
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.5335521697998047,
      "learning_rate": 2.2190365084689198e-05,
      "loss": 1.9485,
      "step": 202018
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.4275239706039429,
      "learning_rate": 2.2188821149405645e-05,
      "loss": 1.8763,
      "step": 202019
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.4813681840896606,
      "learning_rate": 2.2187277265772475e-05,
      "loss": 2.282,
      "step": 202020
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.657721996307373,
      "learning_rate": 2.2185733433789954e-05,
      "loss": 2.0168,
      "step": 202021
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.5702297687530518,
      "learning_rate": 2.218418965345835e-05,
      "loss": 1.9819,
      "step": 202022
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.385598063468933,
      "learning_rate": 2.2182645924777897e-05,
      "loss": 2.0474,
      "step": 202023
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.5275994539260864,
      "learning_rate": 2.218110224774896e-05,
      "loss": 2.0845,
      "step": 202024
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.362221360206604,
      "learning_rate": 2.2179558622371775e-05,
      "loss": 1.9838,
      "step": 202025
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.3892910480499268,
      "learning_rate": 2.2178015048646703e-05,
      "loss": 2.0443,
      "step": 202026
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.384024977684021,
      "learning_rate": 2.217647152657398e-05,
      "loss": 2.0394,
      "step": 202027
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.508899450302124,
      "learning_rate": 2.2174928056153873e-05,
      "loss": 1.9506,
      "step": 202028
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.4983474016189575,
      "learning_rate": 2.2173384637386682e-05,
      "loss": 2.0241,
      "step": 202029
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.4947080612182617,
      "learning_rate": 2.2171841270272705e-05,
      "loss": 1.8216,
      "step": 202030
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.3865416049957275,
      "learning_rate": 2.217029795481221e-05,
      "loss": 1.9896,
      "step": 202031
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.632011890411377,
      "learning_rate": 2.2168754691005563e-05,
      "loss": 2.1307,
      "step": 202032
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.6027708053588867,
      "learning_rate": 2.2167211478852963e-05,
      "loss": 2.1854,
      "step": 202033
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.3979030847549438,
      "learning_rate": 2.2165668318354714e-05,
      "loss": 2.096,
      "step": 202034
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.3924241065979004,
      "learning_rate": 2.2164125209511074e-05,
      "loss": 2.0762,
      "step": 202035
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.4298170804977417,
      "learning_rate": 2.216258215232245e-05,
      "loss": 1.9315,
      "step": 202036
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.4676555395126343,
      "learning_rate": 2.216103914678894e-05,
      "loss": 1.9331,
      "step": 202037
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.4737595319747925,
      "learning_rate": 2.215949619291104e-05,
      "loss": 1.9778,
      "step": 202038
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.522827386856079,
      "learning_rate": 2.215795329068889e-05,
      "loss": 2.0261,
      "step": 202039
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.413598895072937,
      "learning_rate": 2.215641044012285e-05,
      "loss": 2.0035,
      "step": 202040
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.326098084449768,
      "learning_rate": 2.2154867641213126e-05,
      "loss": 1.8431,
      "step": 202041
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.5696589946746826,
      "learning_rate": 2.215332489396008e-05,
      "loss": 1.8039,
      "step": 202042
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.5592002868652344,
      "learning_rate": 2.2151782198363943e-05,
      "loss": 1.913,
      "step": 202043
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.3402241468429565,
      "learning_rate": 2.2150239554425087e-05,
      "loss": 1.8618,
      "step": 202044
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.418811559677124,
      "learning_rate": 2.2148696962143708e-05,
      "loss": 1.9692,
      "step": 202045
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.5353482961654663,
      "learning_rate": 2.2147154421520174e-05,
      "loss": 2.2519,
      "step": 202046
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.4126176834106445,
      "learning_rate": 2.214561193255472e-05,
      "loss": 1.8784,
      "step": 202047
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.4677072763442993,
      "learning_rate": 2.214406949524764e-05,
      "loss": 1.7489,
      "step": 202048
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.4306819438934326,
      "learning_rate": 2.2142527109599173e-05,
      "loss": 2.0723,
      "step": 202049
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.3367832899093628,
      "learning_rate": 2.2140984775609716e-05,
      "loss": 1.8434,
      "step": 202050
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.8401808738708496,
      "learning_rate": 2.2139442493279437e-05,
      "loss": 2.2007,
      "step": 202051
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.590442180633545,
      "learning_rate": 2.2137900262608732e-05,
      "loss": 1.9543,
      "step": 202052
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.4606965780258179,
      "learning_rate": 2.2136358083597838e-05,
      "loss": 2.0356,
      "step": 202053
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.499238133430481,
      "learning_rate": 2.2134815956247053e-05,
      "loss": 2.005,
      "step": 202054
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.505513072013855,
      "learning_rate": 2.213327388055658e-05,
      "loss": 1.9896,
      "step": 202055
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.4553117752075195,
      "learning_rate": 2.2131731856526847e-05,
      "loss": 1.8995,
      "step": 202056
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.7158714532852173,
      "learning_rate": 2.2130189884157988e-05,
      "loss": 2.0995,
      "step": 202057
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.4504008293151855,
      "learning_rate": 2.2128647963450442e-05,
      "loss": 1.965,
      "step": 202058
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.3868972063064575,
      "learning_rate": 2.2127106094404434e-05,
      "loss": 1.998,
      "step": 202059
    },
    {
      "epoch": 2.63,
      "grad_norm": 2.0593972206115723,
      "learning_rate": 2.212556427702017e-05,
      "loss": 2.0522,
      "step": 202060
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.4031089544296265,
      "learning_rate": 2.212402251129808e-05,
      "loss": 1.8534,
      "step": 202061
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.3543968200683594,
      "learning_rate": 2.2122480797238363e-05,
      "loss": 2.0215,
      "step": 202062
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.3379789590835571,
      "learning_rate": 2.2120939134841286e-05,
      "loss": 1.8441,
      "step": 202063
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.4576855897903442,
      "learning_rate": 2.211939752410722e-05,
      "loss": 2.0059,
      "step": 202064
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.6388347148895264,
      "learning_rate": 2.211785596503639e-05,
      "loss": 1.9729,
      "step": 202065
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.497173547744751,
      "learning_rate": 2.211631445762907e-05,
      "loss": 1.7663,
      "step": 202066
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.4842931032180786,
      "learning_rate": 2.211477300188559e-05,
      "loss": 1.8061,
      "step": 202067
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.4338486194610596,
      "learning_rate": 2.2113231597806213e-05,
      "loss": 2.2822,
      "step": 202068
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.408782958984375,
      "learning_rate": 2.2111690245391245e-05,
      "loss": 1.9388,
      "step": 202069
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.412090539932251,
      "learning_rate": 2.211014894464098e-05,
      "loss": 1.8444,
      "step": 202070
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.3623727560043335,
      "learning_rate": 2.210860769555566e-05,
      "loss": 1.983,
      "step": 202071
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.4278682470321655,
      "learning_rate": 2.2107066498135574e-05,
      "loss": 2.0476,
      "step": 202072
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.6499967575073242,
      "learning_rate": 2.210552535238106e-05,
      "loss": 2.111,
      "step": 202073
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.5733673572540283,
      "learning_rate": 2.2103984258292316e-05,
      "loss": 2.1603,
      "step": 202074
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.5221141576766968,
      "learning_rate": 2.2102443215869746e-05,
      "loss": 1.8663,
      "step": 202075
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.517271876335144,
      "learning_rate": 2.210090222511358e-05,
      "loss": 2.0269,
      "step": 202076
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.4464616775512695,
      "learning_rate": 2.209936128602412e-05,
      "loss": 2.2468,
      "step": 202077
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.3783644437789917,
      "learning_rate": 2.2097820398601564e-05,
      "loss": 1.9822,
      "step": 202078
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.529091715812683,
      "learning_rate": 2.2096279562846342e-05,
      "loss": 1.9629,
      "step": 202079
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.4312200546264648,
      "learning_rate": 2.2094738778758593e-05,
      "loss": 1.972,
      "step": 202080
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.3962206840515137,
      "learning_rate": 2.209319804633871e-05,
      "loss": 2.147,
      "step": 202081
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.3673876523971558,
      "learning_rate": 2.2091657365587002e-05,
      "loss": 1.8986,
      "step": 202082
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.3400665521621704,
      "learning_rate": 2.209011673650366e-05,
      "loss": 1.9367,
      "step": 202083
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.490397334098816,
      "learning_rate": 2.2088576159088954e-05,
      "loss": 2.0237,
      "step": 202084
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.4272855520248413,
      "learning_rate": 2.208703563334332e-05,
      "loss": 1.979,
      "step": 202085
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.6871014833450317,
      "learning_rate": 2.2085495159266853e-05,
      "loss": 1.7855,
      "step": 202086
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.351868748664856,
      "learning_rate": 2.208395473686002e-05,
      "loss": 1.7491,
      "step": 202087
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.4197267293930054,
      "learning_rate": 2.2082414366123024e-05,
      "loss": 1.7666,
      "step": 202088
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.4936864376068115,
      "learning_rate": 2.2080874047056163e-05,
      "loss": 1.8719,
      "step": 202089
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.5839871168136597,
      "learning_rate": 2.2079333779659636e-05,
      "loss": 2.1483,
      "step": 202090
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.5789034366607666,
      "learning_rate": 2.2077793563933877e-05,
      "loss": 1.9295,
      "step": 202091
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.3016273975372314,
      "learning_rate": 2.2076253399879052e-05,
      "loss": 2.0126,
      "step": 202092
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.374865174293518,
      "learning_rate": 2.2074713287495528e-05,
      "loss": 2.0557,
      "step": 202093
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.6495481729507446,
      "learning_rate": 2.2073173226783568e-05,
      "loss": 1.8915,
      "step": 202094
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.3300483226776123,
      "learning_rate": 2.2071633217743477e-05,
      "loss": 1.9655,
      "step": 202095
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.4873548746109009,
      "learning_rate": 2.2070093260375422e-05,
      "loss": 1.844,
      "step": 202096
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.6989396810531616,
      "learning_rate": 2.2068553354679863e-05,
      "loss": 2.0766,
      "step": 202097
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.7265725135803223,
      "learning_rate": 2.206701350065697e-05,
      "loss": 2.0452,
      "step": 202098
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.6121747493743896,
      "learning_rate": 2.2065473698307077e-05,
      "loss": 2.0103,
      "step": 202099
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.6564279794692993,
      "learning_rate": 2.2063933947630484e-05,
      "loss": 2.0394,
      "step": 202100
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.406601071357727,
      "learning_rate": 2.2062394248627458e-05,
      "loss": 1.9975,
      "step": 202101
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.3938450813293457,
      "learning_rate": 2.2060854601298194e-05,
      "loss": 2.1006,
      "step": 202102
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.6556397676467896,
      "learning_rate": 2.2059315005643162e-05,
      "loss": 2.0243,
      "step": 202103
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.4100866317749023,
      "learning_rate": 2.2057775461662463e-05,
      "loss": 1.9106,
      "step": 202104
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.3867449760437012,
      "learning_rate": 2.205623596935653e-05,
      "loss": 1.8079,
      "step": 202105
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.5424259901046753,
      "learning_rate": 2.2054696528725592e-05,
      "loss": 1.9058,
      "step": 202106
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.3661489486694336,
      "learning_rate": 2.205315713976995e-05,
      "loss": 1.8043,
      "step": 202107
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.3187263011932373,
      "learning_rate": 2.2051617802489772e-05,
      "loss": 1.9567,
      "step": 202108
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.4052422046661377,
      "learning_rate": 2.205007851688556e-05,
      "loss": 1.8804,
      "step": 202109
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.4375908374786377,
      "learning_rate": 2.2048539282957378e-05,
      "loss": 1.712,
      "step": 202110
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.3680394887924194,
      "learning_rate": 2.2047000100705693e-05,
      "loss": 1.9732,
      "step": 202111
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.261244773864746,
      "learning_rate": 2.2045460970130668e-05,
      "loss": 2.0189,
      "step": 202112
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.3060121536254883,
      "learning_rate": 2.204392189123274e-05,
      "loss": 1.8028,
      "step": 202113
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.3861496448516846,
      "learning_rate": 2.204238286401201e-05,
      "loss": 2.0545,
      "step": 202114
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.4226300716400146,
      "learning_rate": 2.2040843888468872e-05,
      "loss": 1.9982,
      "step": 202115
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.460660457611084,
      "learning_rate": 2.2039304964603565e-05,
      "loss": 1.9364,
      "step": 202116
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.4600685834884644,
      "learning_rate": 2.203776609241642e-05,
      "loss": 2.0697,
      "step": 202117
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.4560370445251465,
      "learning_rate": 2.203622727190767e-05,
      "loss": 2.2719,
      "step": 202118
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.4965016841888428,
      "learning_rate": 2.2034688503077714e-05,
      "loss": 1.9954,
      "step": 202119
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.595746636390686,
      "learning_rate": 2.203314978592665e-05,
      "loss": 2.0854,
      "step": 202120
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.6040468215942383,
      "learning_rate": 2.203161112045495e-05,
      "loss": 2.259,
      "step": 202121
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.5544614791870117,
      "learning_rate": 2.2030072506662743e-05,
      "loss": 2.0441,
      "step": 202122
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.3338268995285034,
      "learning_rate": 2.2028533944550463e-05,
      "loss": 1.8437,
      "step": 202123
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.4181848764419556,
      "learning_rate": 2.2026995434118277e-05,
      "loss": 2.0936,
      "step": 202124
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.6754481792449951,
      "learning_rate": 2.202545697536662e-05,
      "loss": 2.071,
      "step": 202125
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.2947968244552612,
      "learning_rate": 2.202391856829555e-05,
      "loss": 1.8061,
      "step": 202126
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.3796215057373047,
      "learning_rate": 2.2022380212905543e-05,
      "loss": 1.9425,
      "step": 202127
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.3409970998764038,
      "learning_rate": 2.2020841909196797e-05,
      "loss": 1.9426,
      "step": 202128
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.3118252754211426,
      "learning_rate": 2.201930365716964e-05,
      "loss": 1.9065,
      "step": 202129
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.5338726043701172,
      "learning_rate": 2.201776545682431e-05,
      "loss": 2.126,
      "step": 202130
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.5192663669586182,
      "learning_rate": 2.2016227308161172e-05,
      "loss": 1.8642,
      "step": 202131
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.4745113849639893,
      "learning_rate": 2.201468921118049e-05,
      "loss": 1.8787,
      "step": 202132
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.4384671449661255,
      "learning_rate": 2.2013151165882503e-05,
      "loss": 1.8898,
      "step": 202133
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.4782825708389282,
      "learning_rate": 2.201161317226744e-05,
      "loss": 1.9617,
      "step": 202134
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.480493426322937,
      "learning_rate": 2.2010075230335766e-05,
      "loss": 2.1125,
      "step": 202135
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.633291244506836,
      "learning_rate": 2.2008537340087583e-05,
      "loss": 2.0246,
      "step": 202136
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.4963594675064087,
      "learning_rate": 2.2006999501523324e-05,
      "loss": 2.0247,
      "step": 202137
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.7028809785842896,
      "learning_rate": 2.200546171464319e-05,
      "loss": 2.1611,
      "step": 202138
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.430235743522644,
      "learning_rate": 2.200392397944751e-05,
      "loss": 1.9733,
      "step": 202139
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.428938865661621,
      "learning_rate": 2.2002386295936493e-05,
      "loss": 2.2313,
      "step": 202140
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.4486004114151,
      "learning_rate": 2.2000848664110527e-05,
      "loss": 1.8927,
      "step": 202141
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.421231985092163,
      "learning_rate": 2.1999311083969816e-05,
      "loss": 1.9902,
      "step": 202142
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.391663908958435,
      "learning_rate": 2.199777355551473e-05,
      "loss": 1.8768,
      "step": 202143
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.4530928134918213,
      "learning_rate": 2.19962360787455e-05,
      "loss": 1.822,
      "step": 202144
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.5251823663711548,
      "learning_rate": 2.199469865366236e-05,
      "loss": 1.7834,
      "step": 202145
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.484262228012085,
      "learning_rate": 2.1993161280265704e-05,
      "loss": 2.0597,
      "step": 202146
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.7518056631088257,
      "learning_rate": 2.199162395855577e-05,
      "loss": 1.8984,
      "step": 202147
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.409020185470581,
      "learning_rate": 2.199008668853276e-05,
      "loss": 2.1296,
      "step": 202148
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.4662214517593384,
      "learning_rate": 2.1988549470197136e-05,
      "loss": 1.8231,
      "step": 202149
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.4969960451126099,
      "learning_rate": 2.1987012303549067e-05,
      "loss": 2.064,
      "step": 202150
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.5490453243255615,
      "learning_rate": 2.198547518858882e-05,
      "loss": 2.0715,
      "step": 202151
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.4252948760986328,
      "learning_rate": 2.1983938125316758e-05,
      "loss": 1.912,
      "step": 202152
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.4077624082565308,
      "learning_rate": 2.198240111373315e-05,
      "loss": 2.1138,
      "step": 202153
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.4108219146728516,
      "learning_rate": 2.1980864153838195e-05,
      "loss": 1.9593,
      "step": 202154
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.4635323286056519,
      "learning_rate": 2.1979327245632296e-05,
      "loss": 1.9105,
      "step": 202155
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.634448528289795,
      "learning_rate": 2.1977790389115712e-05,
      "loss": 2.0472,
      "step": 202156
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.4132952690124512,
      "learning_rate": 2.1976253584288616e-05,
      "loss": 1.9801,
      "step": 202157
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.4259086847305298,
      "learning_rate": 2.1974716831151475e-05,
      "loss": 2.131,
      "step": 202158
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.5824602842330933,
      "learning_rate": 2.1973180129704383e-05,
      "loss": 1.8395,
      "step": 202159
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.4703726768493652,
      "learning_rate": 2.197164347994781e-05,
      "loss": 2.1724,
      "step": 202160
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.5194497108459473,
      "learning_rate": 2.1970106881881954e-05,
      "loss": 2.0725,
      "step": 202161
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.3879106044769287,
      "learning_rate": 2.1968570335507084e-05,
      "loss": 2.0583,
      "step": 202162
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.487627625465393,
      "learning_rate": 2.1967033840823466e-05,
      "loss": 2.031,
      "step": 202163
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.380529761314392,
      "learning_rate": 2.19654973978315e-05,
      "loss": 1.8675,
      "step": 202164
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.4430814981460571,
      "learning_rate": 2.1963961006531317e-05,
      "loss": 1.8693,
      "step": 202165
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.4884750843048096,
      "learning_rate": 2.1962424666923352e-05,
      "loss": 1.9303,
      "step": 202166
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.4669952392578125,
      "learning_rate": 2.19608883790078e-05,
      "loss": 1.8316,
      "step": 202167
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.5391348600387573,
      "learning_rate": 2.1959352142784968e-05,
      "loss": 2.0981,
      "step": 202168
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.462180733680725,
      "learning_rate": 2.1957815958255088e-05,
      "loss": 1.9255,
      "step": 202169
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.494533896446228,
      "learning_rate": 2.1956279825418555e-05,
      "loss": 2.1246,
      "step": 202170
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.5106685161590576,
      "learning_rate": 2.195474374427557e-05,
      "loss": 2.1537,
      "step": 202171
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.4160728454589844,
      "learning_rate": 2.195320771482647e-05,
      "loss": 1.8884,
      "step": 202172
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.387115478515625,
      "learning_rate": 2.195167173707152e-05,
      "loss": 2.0157,
      "step": 202173
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.644243836402893,
      "learning_rate": 2.1950135811011015e-05,
      "loss": 2.0305,
      "step": 202174
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.6648598909378052,
      "learning_rate": 2.1948599936645163e-05,
      "loss": 2.0063,
      "step": 202175
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.5216200351715088,
      "learning_rate": 2.194706411397439e-05,
      "loss": 2.0966,
      "step": 202176
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.5558247566223145,
      "learning_rate": 2.194552834299883e-05,
      "loss": 1.9342,
      "step": 202177
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.4210857152938843,
      "learning_rate": 2.1943992623718887e-05,
      "loss": 1.8638,
      "step": 202178
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.350278377532959,
      "learning_rate": 2.194245695613479e-05,
      "loss": 1.8525,
      "step": 202179
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.4374721050262451,
      "learning_rate": 2.194092134024691e-05,
      "loss": 2.0562,
      "step": 202180
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.2581077814102173,
      "learning_rate": 2.1939385776055375e-05,
      "loss": 1.8153,
      "step": 202181
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.4026894569396973,
      "learning_rate": 2.193785026356062e-05,
      "loss": 1.9723,
      "step": 202182
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.4029953479766846,
      "learning_rate": 2.1936314802762777e-05,
      "loss": 1.8189,
      "step": 202183
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.454883337020874,
      "learning_rate": 2.1934779393662318e-05,
      "loss": 1.8871,
      "step": 202184
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.3572347164154053,
      "learning_rate": 2.1933244036259335e-05,
      "loss": 2.009,
      "step": 202185
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.399151086807251,
      "learning_rate": 2.1931708730554364e-05,
      "loss": 2.1454,
      "step": 202186
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.4854941368103027,
      "learning_rate": 2.193017347654741e-05,
      "loss": 1.8604,
      "step": 202187
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.4708032608032227,
      "learning_rate": 2.192863827423893e-05,
      "loss": 2.0682,
      "step": 202188
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.3795679807662964,
      "learning_rate": 2.192710312362913e-05,
      "loss": 1.8187,
      "step": 202189
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.4294850826263428,
      "learning_rate": 2.1925568024718344e-05,
      "loss": 1.8779,
      "step": 202190
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.6229993104934692,
      "learning_rate": 2.1924032977506832e-05,
      "loss": 1.9178,
      "step": 202191
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.6122440099716187,
      "learning_rate": 2.192249798199497e-05,
      "loss": 2.0474,
      "step": 202192
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.4688628911972046,
      "learning_rate": 2.192096303818288e-05,
      "loss": 1.9578,
      "step": 202193
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.4581067562103271,
      "learning_rate": 2.1919428146070973e-05,
      "loss": 2.1019,
      "step": 202194
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.4661368131637573,
      "learning_rate": 2.1917893305659472e-05,
      "loss": 2.1921,
      "step": 202195
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.5840092897415161,
      "learning_rate": 2.1916358516948684e-05,
      "loss": 2.0278,
      "step": 202196
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.7795236110687256,
      "learning_rate": 2.1914823779938873e-05,
      "loss": 1.7986,
      "step": 202197
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.3998591899871826,
      "learning_rate": 2.191328909463047e-05,
      "loss": 2.015,
      "step": 202198
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.2936041355133057,
      "learning_rate": 2.191175446102348e-05,
      "loss": 2.0339,
      "step": 202199
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.431362271308899,
      "learning_rate": 2.1910219879118428e-05,
      "loss": 2.1207,
      "step": 202200
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.5759788751602173,
      "learning_rate": 2.1908685348915454e-05,
      "loss": 2.0125,
      "step": 202201
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.623528242111206,
      "learning_rate": 2.1907150870414993e-05,
      "loss": 2.0503,
      "step": 202202
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.5122253894805908,
      "learning_rate": 2.1905616443617134e-05,
      "loss": 2.1397,
      "step": 202203
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.4724600315093994,
      "learning_rate": 2.1904082068522387e-05,
      "loss": 2.0691,
      "step": 202204
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.7880187034606934,
      "learning_rate": 2.190254774513085e-05,
      "loss": 2.0184,
      "step": 202205
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.4133567810058594,
      "learning_rate": 2.1901013473442918e-05,
      "loss": 1.942,
      "step": 202206
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.4566296339035034,
      "learning_rate": 2.189947925345876e-05,
      "loss": 2.0662,
      "step": 202207
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.5699121952056885,
      "learning_rate": 2.1897945085178815e-05,
      "loss": 2.0868,
      "step": 202208
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.491398572921753,
      "learning_rate": 2.189641096860324e-05,
      "loss": 1.8151,
      "step": 202209
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.5166535377502441,
      "learning_rate": 2.1894876903732473e-05,
      "loss": 1.964,
      "step": 202210
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.3456073999404907,
      "learning_rate": 2.189334289056658e-05,
      "loss": 1.9491,
      "step": 202211
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.5005301237106323,
      "learning_rate": 2.1891808929106026e-05,
      "loss": 2.024,
      "step": 202212
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.5302027463912964,
      "learning_rate": 2.1890275019350977e-05,
      "loss": 1.7552,
      "step": 202213
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.5495160818099976,
      "learning_rate": 2.1888741161301838e-05,
      "loss": 2.1199,
      "step": 202214
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.5021328926086426,
      "learning_rate": 2.188720735495877e-05,
      "loss": 2.0796,
      "step": 202215
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.5881766080856323,
      "learning_rate": 2.1885673600322207e-05,
      "loss": 2.0246,
      "step": 202216
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.5715030431747437,
      "learning_rate": 2.1884139897392284e-05,
      "loss": 1.7313,
      "step": 202217
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.5021038055419922,
      "learning_rate": 2.1882606246169366e-05,
      "loss": 2.0554,
      "step": 202218
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.506090521812439,
      "learning_rate": 2.1881072646653686e-05,
      "loss": 1.863,
      "step": 202219
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.4165947437286377,
      "learning_rate": 2.187953909884561e-05,
      "loss": 2.0082,
      "step": 202220
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.4612302780151367,
      "learning_rate": 2.1878005602745342e-05,
      "loss": 1.7086,
      "step": 202221
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.512597680091858,
      "learning_rate": 2.1876472158353243e-05,
      "loss": 1.9742,
      "step": 202222
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.469002366065979,
      "learning_rate": 2.1874938765669548e-05,
      "loss": 2.0501,
      "step": 202223
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.4297339916229248,
      "learning_rate": 2.1873405424694558e-05,
      "loss": 1.9144,
      "step": 202224
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.4106462001800537,
      "learning_rate": 2.1871872135428503e-05,
      "loss": 2.0434,
      "step": 202225
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.3483518362045288,
      "learning_rate": 2.1870338897871787e-05,
      "loss": 1.9815,
      "step": 202226
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.4478622674942017,
      "learning_rate": 2.186880571202454e-05,
      "loss": 2.0915,
      "step": 202227
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.466370940208435,
      "learning_rate": 2.1867272577887195e-05,
      "loss": 2.0548,
      "step": 202228
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.4218412637710571,
      "learning_rate": 2.186573949545999e-05,
      "loss": 1.8195,
      "step": 202229
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.4554228782653809,
      "learning_rate": 2.186420646474315e-05,
      "loss": 2.08,
      "step": 202230
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.5169415473937988,
      "learning_rate": 2.1862673485737013e-05,
      "loss": 1.8991,
      "step": 202231
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.4143190383911133,
      "learning_rate": 2.1861140558441882e-05,
      "loss": 1.9564,
      "step": 202232
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.4691742658615112,
      "learning_rate": 2.185960768285798e-05,
      "loss": 2.1674,
      "step": 202233
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.5111263990402222,
      "learning_rate": 2.185807485898565e-05,
      "loss": 2.1599,
      "step": 202234
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.3579052686691284,
      "learning_rate": 2.1856542086825158e-05,
      "loss": 1.8994,
      "step": 202235
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.3715511560440063,
      "learning_rate": 2.185500936637673e-05,
      "loss": 2.0522,
      "step": 202236
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.5055773258209229,
      "learning_rate": 2.185347669764077e-05,
      "loss": 1.9597,
      "step": 202237
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.512325406074524,
      "learning_rate": 2.1851944080617513e-05,
      "loss": 1.9547,
      "step": 202238
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.414859414100647,
      "learning_rate": 2.1850411515307154e-05,
      "loss": 2.0745,
      "step": 202239
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.4526020288467407,
      "learning_rate": 2.184887900171013e-05,
      "loss": 2.1178,
      "step": 202240
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.577746868133545,
      "learning_rate": 2.184734653982664e-05,
      "loss": 1.9414,
      "step": 202241
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.6281335353851318,
      "learning_rate": 2.1845814129656914e-05,
      "loss": 2.0131,
      "step": 202242
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.4760500192642212,
      "learning_rate": 2.1844281771201355e-05,
      "loss": 2.0022,
      "step": 202243
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.4766411781311035,
      "learning_rate": 2.184274946446013e-05,
      "loss": 1.8215,
      "step": 202244
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.4296464920043945,
      "learning_rate": 2.184121720943367e-05,
      "loss": 1.9885,
      "step": 202245
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.4925999641418457,
      "learning_rate": 2.1839685006122177e-05,
      "loss": 2.2305,
      "step": 202246
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.4146549701690674,
      "learning_rate": 2.1838152854525915e-05,
      "loss": 1.8973,
      "step": 202247
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.4904589653015137,
      "learning_rate": 2.183662075464515e-05,
      "loss": 1.8358,
      "step": 202248
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.5297890901565552,
      "learning_rate": 2.183508870648025e-05,
      "loss": 1.9425,
      "step": 202249
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.4618613719940186,
      "learning_rate": 2.183355671003142e-05,
      "loss": 1.8987,
      "step": 202250
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.416571021080017,
      "learning_rate": 2.183202476529905e-05,
      "loss": 1.8997,
      "step": 202251
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.4353824853897095,
      "learning_rate": 2.1830492872283344e-05,
      "loss": 1.9748,
      "step": 202252
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.3444384336471558,
      "learning_rate": 2.182896103098457e-05,
      "loss": 2.0055,
      "step": 202253
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.4288907051086426,
      "learning_rate": 2.1827429241403028e-05,
      "loss": 2.0603,
      "step": 202254
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.4542351961135864,
      "learning_rate": 2.1825897503539046e-05,
      "loss": 1.8336,
      "step": 202255
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.5000005960464478,
      "learning_rate": 2.1824365817392864e-05,
      "loss": 2.1184,
      "step": 202256
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.4554226398468018,
      "learning_rate": 2.182283418296481e-05,
      "loss": 2.2477,
      "step": 202257
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.4152140617370605,
      "learning_rate": 2.1821302600255085e-05,
      "loss": 2.1149,
      "step": 202258
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.4783064126968384,
      "learning_rate": 2.181977106926416e-05,
      "loss": 1.9466,
      "step": 202259
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.4716445207595825,
      "learning_rate": 2.181823958999206e-05,
      "loss": 1.8819,
      "step": 202260
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.4588537216186523,
      "learning_rate": 2.1816708162439255e-05,
      "loss": 1.9989,
      "step": 202261
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.4512996673583984,
      "learning_rate": 2.1815176786605913e-05,
      "loss": 2.1563,
      "step": 202262
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.3566951751708984,
      "learning_rate": 2.1813645462492468e-05,
      "loss": 1.9066,
      "step": 202263
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.5148990154266357,
      "learning_rate": 2.181211419009905e-05,
      "loss": 2.0164,
      "step": 202264
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.5538395643234253,
      "learning_rate": 2.1810582969426093e-05,
      "loss": 2.0235,
      "step": 202265
    },
    {
      "epoch": 2.63,
      "grad_norm": 2.365520477294922,
      "learning_rate": 2.1809051800473732e-05,
      "loss": 2.0218,
      "step": 202266
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.5706233978271484,
      "learning_rate": 2.180752068324233e-05,
      "loss": 1.9244,
      "step": 202267
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.5062453746795654,
      "learning_rate": 2.1805989617732154e-05,
      "loss": 2.0743,
      "step": 202268
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.6041170358657837,
      "learning_rate": 2.180445860394351e-05,
      "loss": 1.9759,
      "step": 202269
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.4754822254180908,
      "learning_rate": 2.1802927641876656e-05,
      "loss": 2.0574,
      "step": 202270
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.583849310874939,
      "learning_rate": 2.180139673153196e-05,
      "loss": 2.0974,
      "step": 202271
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.4691343307495117,
      "learning_rate": 2.1799865872909528e-05,
      "loss": 2.0018,
      "step": 202272
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.4995230436325073,
      "learning_rate": 2.179833506600982e-05,
      "loss": 1.9111,
      "step": 202273
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.3949946165084839,
      "learning_rate": 2.1796804310833006e-05,
      "loss": 2.1522,
      "step": 202274
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.4066216945648193,
      "learning_rate": 2.1795273607379483e-05,
      "loss": 2.037,
      "step": 202275
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.4745526313781738,
      "learning_rate": 2.1793742955649384e-05,
      "loss": 2.142,
      "step": 202276
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.5040183067321777,
      "learning_rate": 2.1792212355643213e-05,
      "loss": 2.2916,
      "step": 202277
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.4400954246520996,
      "learning_rate": 2.1790681807361e-05,
      "loss": 1.984,
      "step": 202278
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.456758737564087,
      "learning_rate": 2.178915131080321e-05,
      "loss": 2.1955,
      "step": 202279
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.4151712656021118,
      "learning_rate": 2.178762086597001e-05,
      "loss": 2.1822,
      "step": 202280
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.4106818437576294,
      "learning_rate": 2.178609047286177e-05,
      "loss": 1.886,
      "step": 202281
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.507055401802063,
      "learning_rate": 2.1784560131478723e-05,
      "loss": 2.2514,
      "step": 202282
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.3934320211410522,
      "learning_rate": 2.1783029841821296e-05,
      "loss": 2.1287,
      "step": 202283
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.4898489713668823,
      "learning_rate": 2.178149960388953e-05,
      "loss": 2.0337,
      "step": 202284
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.5029760599136353,
      "learning_rate": 2.1779969417683886e-05,
      "loss": 1.9638,
      "step": 202285
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.431194543838501,
      "learning_rate": 2.1778439283204563e-05,
      "loss": 1.8794,
      "step": 202286
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.4897361993789673,
      "learning_rate": 2.1776909200451932e-05,
      "loss": 2.0088,
      "step": 202287
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.447778582572937,
      "learning_rate": 2.1775379169426153e-05,
      "loss": 1.9786,
      "step": 202288
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.6263600587844849,
      "learning_rate": 2.17738491901277e-05,
      "loss": 1.9688,
      "step": 202289
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.5954221487045288,
      "learning_rate": 2.1772319262556636e-05,
      "loss": 2.0921,
      "step": 202290
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.4737910032272339,
      "learning_rate": 2.1770789386713395e-05,
      "loss": 1.8959,
      "step": 202291
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.3687825202941895,
      "learning_rate": 2.1769259562598175e-05,
      "loss": 2.0777,
      "step": 202292
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.785704255104065,
      "learning_rate": 2.1767729790211342e-05,
      "loss": 1.909,
      "step": 202293
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.4622118473052979,
      "learning_rate": 2.1766200069553096e-05,
      "loss": 2.1179,
      "step": 202294
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.3972691297531128,
      "learning_rate": 2.1764670400623873e-05,
      "loss": 1.9731,
      "step": 202295
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.3795822858810425,
      "learning_rate": 2.176314078342374e-05,
      "loss": 1.9676,
      "step": 202296
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.473319172859192,
      "learning_rate": 2.1761611217953123e-05,
      "loss": 1.9464,
      "step": 202297
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.382847785949707,
      "learning_rate": 2.1760081704212262e-05,
      "loss": 1.9216,
      "step": 202298
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.3844467401504517,
      "learning_rate": 2.175855224220149e-05,
      "loss": 1.9493,
      "step": 202299
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.56874680519104,
      "learning_rate": 2.1757022831921e-05,
      "loss": 2.1889,
      "step": 202300
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.5224058628082275,
      "learning_rate": 2.1755493473371233e-05,
      "loss": 1.8449,
      "step": 202301
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.4947222471237183,
      "learning_rate": 2.175396416655225e-05,
      "loss": 2.014,
      "step": 202302
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.3730422258377075,
      "learning_rate": 2.1752434911464556e-05,
      "loss": 1.9025,
      "step": 202303
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.4377468824386597,
      "learning_rate": 2.1750905708108247e-05,
      "loss": 1.9592,
      "step": 202304
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.4829156398773193,
      "learning_rate": 2.1749376556483756e-05,
      "loss": 2.1005,
      "step": 202305
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.5129717588424683,
      "learning_rate": 2.1747847456591285e-05,
      "loss": 1.9028,
      "step": 202306
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.4285690784454346,
      "learning_rate": 2.1746318408431164e-05,
      "loss": 1.9553,
      "step": 202307
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.3995931148529053,
      "learning_rate": 2.174478941200366e-05,
      "loss": 1.9399,
      "step": 202308
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.7117410898208618,
      "learning_rate": 2.1743260467309042e-05,
      "loss": 2.0391,
      "step": 202309
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.4025970697402954,
      "learning_rate": 2.1741731574347577e-05,
      "loss": 1.9358,
      "step": 202310
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.5918688774108887,
      "learning_rate": 2.1740202733119594e-05,
      "loss": 2.0587,
      "step": 202311
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.4574620723724365,
      "learning_rate": 2.1738673943625327e-05,
      "loss": 2.0262,
      "step": 202312
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.412222981452942,
      "learning_rate": 2.1737145205865146e-05,
      "loss": 1.9429,
      "step": 202313
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.5479276180267334,
      "learning_rate": 2.1735616519839284e-05,
      "loss": 2.041,
      "step": 202314
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.3702410459518433,
      "learning_rate": 2.1734087885548034e-05,
      "loss": 1.993,
      "step": 202315
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.3146077394485474,
      "learning_rate": 2.1732559302991604e-05,
      "loss": 2.1881,
      "step": 202316
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.4732909202575684,
      "learning_rate": 2.1731030772170388e-05,
      "loss": 1.8432,
      "step": 202317
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.4378199577331543,
      "learning_rate": 2.172950229308459e-05,
      "loss": 1.8545,
      "step": 202318
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.4216402769088745,
      "learning_rate": 2.1727973865734573e-05,
      "loss": 1.9242,
      "step": 202319
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.3405977487564087,
      "learning_rate": 2.1726445490120604e-05,
      "loss": 1.8616,
      "step": 202320
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.4666242599487305,
      "learning_rate": 2.172491716624285e-05,
      "loss": 2.1862,
      "step": 202321
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.420925498008728,
      "learning_rate": 2.1723388894101745e-05,
      "loss": 2.0575,
      "step": 202322
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.4176336526870728,
      "learning_rate": 2.172186067369752e-05,
      "loss": 1.9789,
      "step": 202323
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.465651035308838,
      "learning_rate": 2.1720332505030412e-05,
      "loss": 1.9941,
      "step": 202324
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.3826383352279663,
      "learning_rate": 2.171880438810082e-05,
      "loss": 1.9277,
      "step": 202325
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.402561068534851,
      "learning_rate": 2.1717276322908906e-05,
      "loss": 2.2505,
      "step": 202326
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.4246853590011597,
      "learning_rate": 2.1715748309455004e-05,
      "loss": 2.1938,
      "step": 202327
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.451817274093628,
      "learning_rate": 2.171422034773942e-05,
      "loss": 2.1189,
      "step": 202328
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.4092130661010742,
      "learning_rate": 2.1712692437762346e-05,
      "loss": 2.101,
      "step": 202329
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.6397428512573242,
      "learning_rate": 2.171116457952422e-05,
      "loss": 2.0365,
      "step": 202330
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.4300365447998047,
      "learning_rate": 2.1709636773025207e-05,
      "loss": 2.004,
      "step": 202331
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.3423749208450317,
      "learning_rate": 2.1708109018265674e-05,
      "loss": 2.0684,
      "step": 202332
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.5332924127578735,
      "learning_rate": 2.1706581315245752e-05,
      "loss": 1.9416,
      "step": 202333
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.3889135122299194,
      "learning_rate": 2.170505366396591e-05,
      "loss": 2.1359,
      "step": 202334
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.3804492950439453,
      "learning_rate": 2.1703526064426313e-05,
      "loss": 1.9858,
      "step": 202335
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.3023062944412231,
      "learning_rate": 2.1701998516627327e-05,
      "loss": 1.8836,
      "step": 202336
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.4203664064407349,
      "learning_rate": 2.170047102056919e-05,
      "loss": 2.0708,
      "step": 202337
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.3816909790039062,
      "learning_rate": 2.1698943576252158e-05,
      "loss": 1.8838,
      "step": 202338
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.403354287147522,
      "learning_rate": 2.169741618367654e-05,
      "loss": 2.0559,
      "step": 202339
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.4125896692276,
      "learning_rate": 2.1695888842842635e-05,
      "loss": 1.9829,
      "step": 202340
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.602805495262146,
      "learning_rate": 2.1694361553750707e-05,
      "loss": 2.1464,
      "step": 202341
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.5474029779434204,
      "learning_rate": 2.1692834316401087e-05,
      "loss": 1.9368,
      "step": 202342
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.4513942003250122,
      "learning_rate": 2.1691307130793945e-05,
      "loss": 2.0443,
      "step": 202343
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.4710758924484253,
      "learning_rate": 2.168977999692978e-05,
      "loss": 2.1301,
      "step": 202344
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.5143637657165527,
      "learning_rate": 2.1688252914808623e-05,
      "loss": 1.9198,
      "step": 202345
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.3214778900146484,
      "learning_rate": 2.168672588443091e-05,
      "loss": 2.088,
      "step": 202346
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.6882054805755615,
      "learning_rate": 2.1685198905796876e-05,
      "loss": 2.0469,
      "step": 202347
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.5617090463638306,
      "learning_rate": 2.1683671978906847e-05,
      "loss": 1.8944,
      "step": 202348
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.420532464981079,
      "learning_rate": 2.1682145103761028e-05,
      "loss": 2.0078,
      "step": 202349
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.4642685651779175,
      "learning_rate": 2.1680618280359818e-05,
      "loss": 1.925,
      "step": 202350
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.585460901260376,
      "learning_rate": 2.1679091508703383e-05,
      "loss": 2.1934,
      "step": 202351
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.653202772140503,
      "learning_rate": 2.167756478879209e-05,
      "loss": 2.075,
      "step": 202352
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.4709056615829468,
      "learning_rate": 2.167603812062614e-05,
      "loss": 2.0735,
      "step": 202353
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.5253061056137085,
      "learning_rate": 2.1674511504205927e-05,
      "loss": 1.9476,
      "step": 202354
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.3803879022598267,
      "learning_rate": 2.1672984939531622e-05,
      "loss": 2.0296,
      "step": 202355
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.2826777696609497,
      "learning_rate": 2.167145842660366e-05,
      "loss": 1.7654,
      "step": 202356
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.4195178747177124,
      "learning_rate": 2.1669931965422104e-05,
      "loss": 1.7723,
      "step": 202357
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.3641738891601562,
      "learning_rate": 2.1668405555987457e-05,
      "loss": 1.8854,
      "step": 202358
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.4199750423431396,
      "learning_rate": 2.1666879198299846e-05,
      "loss": 2.0674,
      "step": 202359
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.3773797750473022,
      "learning_rate": 2.1665352892359645e-05,
      "loss": 2.0412,
      "step": 202360
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.4531680345535278,
      "learning_rate": 2.1663826638167083e-05,
      "loss": 1.7901,
      "step": 202361
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.3845877647399902,
      "learning_rate": 2.1662300435722524e-05,
      "loss": 2.0224,
      "step": 202362
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.370646357536316,
      "learning_rate": 2.166077428502614e-05,
      "loss": 2.0277,
      "step": 202363
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.3529818058013916,
      "learning_rate": 2.1659248186078328e-05,
      "loss": 2.0542,
      "step": 202364
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.4845761060714722,
      "learning_rate": 2.1657722138879252e-05,
      "loss": 1.834,
      "step": 202365
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.4758373498916626,
      "learning_rate": 2.1656196143429282e-05,
      "loss": 1.8126,
      "step": 202366
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.5603642463684082,
      "learning_rate": 2.1654670199728654e-05,
      "loss": 2.1528,
      "step": 202367
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.6041464805603027,
      "learning_rate": 2.165314430777779e-05,
      "loss": 2.1533,
      "step": 202368
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.472773790359497,
      "learning_rate": 2.1651618467576736e-05,
      "loss": 1.9197,
      "step": 202369
    },
    {
      "epoch": 2.63,
      "grad_norm": 1.640303611755371,
      "learning_rate": 2.1650092679125985e-05,
      "loss": 2.2523,
      "step": 202370
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.6527018547058105,
      "learning_rate": 2.1648566942425637e-05,
      "loss": 1.8291,
      "step": 202371
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.4396514892578125,
      "learning_rate": 2.1647041257476163e-05,
      "loss": 1.9548,
      "step": 202372
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.4508048295974731,
      "learning_rate": 2.1645515624277688e-05,
      "loss": 2.0953,
      "step": 202373
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.4306938648223877,
      "learning_rate": 2.1643990042830683e-05,
      "loss": 1.8667,
      "step": 202374
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.3641399145126343,
      "learning_rate": 2.1642464513135215e-05,
      "loss": 2.1286,
      "step": 202375
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.5704749822616577,
      "learning_rate": 2.1640939035191717e-05,
      "loss": 1.9814,
      "step": 202376
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.3816717863082886,
      "learning_rate": 2.1639413609000355e-05,
      "loss": 1.9008,
      "step": 202377
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.4004228115081787,
      "learning_rate": 2.1637888234561528e-05,
      "loss": 1.7915,
      "step": 202378
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.4811574220657349,
      "learning_rate": 2.1636362911875438e-05,
      "loss": 2.244,
      "step": 202379
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.4923884868621826,
      "learning_rate": 2.1634837640942448e-05,
      "loss": 1.8766,
      "step": 202380
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.328619360923767,
      "learning_rate": 2.1633312421762795e-05,
      "loss": 1.8247,
      "step": 202381
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.4888454675674438,
      "learning_rate": 2.1631787254336773e-05,
      "loss": 2.0691,
      "step": 202382
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.480237364768982,
      "learning_rate": 2.163026213866459e-05,
      "loss": 2.0609,
      "step": 202383
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.451908826828003,
      "learning_rate": 2.1628737074746673e-05,
      "loss": 1.9058,
      "step": 202384
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.3987767696380615,
      "learning_rate": 2.1627212062583156e-05,
      "loss": 1.8696,
      "step": 202385
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.4074339866638184,
      "learning_rate": 2.1625687102174437e-05,
      "loss": 1.9079,
      "step": 202386
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.3885176181793213,
      "learning_rate": 2.162416219352079e-05,
      "loss": 1.9754,
      "step": 202387
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.480132818222046,
      "learning_rate": 2.162263733662244e-05,
      "loss": 2.0929,
      "step": 202388
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.924230933189392,
      "learning_rate": 2.1621112531479657e-05,
      "loss": 2.0895,
      "step": 202389
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.497005820274353,
      "learning_rate": 2.1619587778092807e-05,
      "loss": 2.0509,
      "step": 202390
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.695264220237732,
      "learning_rate": 2.161806307646209e-05,
      "loss": 2.0195,
      "step": 202391
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.5840665102005005,
      "learning_rate": 2.1616538426587872e-05,
      "loss": 1.9179,
      "step": 202392
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.461623191833496,
      "learning_rate": 2.1615013828470384e-05,
      "loss": 1.9566,
      "step": 202393
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.4811517000198364,
      "learning_rate": 2.161348928210993e-05,
      "loss": 2.0196,
      "step": 202394
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.5217164754867554,
      "learning_rate": 2.1611964787506742e-05,
      "loss": 2.0113,
      "step": 202395
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.5195074081420898,
      "learning_rate": 2.1610440344661184e-05,
      "loss": 1.8739,
      "step": 202396
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.3855770826339722,
      "learning_rate": 2.160891595357346e-05,
      "loss": 2.0912,
      "step": 202397
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.4133458137512207,
      "learning_rate": 2.160739161424393e-05,
      "loss": 1.9818,
      "step": 202398
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.468080997467041,
      "learning_rate": 2.160586732667283e-05,
      "loss": 1.8618,
      "step": 202399
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.4277693033218384,
      "learning_rate": 2.16043430908605e-05,
      "loss": 1.9119,
      "step": 202400
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.4476779699325562,
      "learning_rate": 2.1602818906807095e-05,
      "loss": 1.8067,
      "step": 202401
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.4110572338104248,
      "learning_rate": 2.160129477451302e-05,
      "loss": 1.8153,
      "step": 202402
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.4081755876541138,
      "learning_rate": 2.159977069397848e-05,
      "loss": 1.9066,
      "step": 202403
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.4358253479003906,
      "learning_rate": 2.1598246665203866e-05,
      "loss": 1.8237,
      "step": 202404
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.5364331007003784,
      "learning_rate": 2.1596722688189383e-05,
      "loss": 1.9453,
      "step": 202405
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.591610312461853,
      "learning_rate": 2.159519876293526e-05,
      "loss": 1.9422,
      "step": 202406
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.4858523607254028,
      "learning_rate": 2.1593674889441903e-05,
      "loss": 2.0583,
      "step": 202407
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.4869621992111206,
      "learning_rate": 2.159215106770954e-05,
      "loss": 1.8728,
      "step": 202408
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.3735147714614868,
      "learning_rate": 2.1590627297738407e-05,
      "loss": 2.1856,
      "step": 202409
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.316765308380127,
      "learning_rate": 2.15891035795289e-05,
      "loss": 1.878,
      "step": 202410
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.3930026292800903,
      "learning_rate": 2.1587579913081187e-05,
      "loss": 1.9898,
      "step": 202411
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.4613020420074463,
      "learning_rate": 2.158605629839557e-05,
      "loss": 2.0287,
      "step": 202412
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.4779984951019287,
      "learning_rate": 2.1584532735472414e-05,
      "loss": 2.0409,
      "step": 202413
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.3982484340667725,
      "learning_rate": 2.158300922431192e-05,
      "loss": 2.0941,
      "step": 202414
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.675925374031067,
      "learning_rate": 2.158148576491442e-05,
      "loss": 2.2354,
      "step": 202415
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.4842830896377563,
      "learning_rate": 2.1579962357280177e-05,
      "loss": 2.0918,
      "step": 202416
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.775017499923706,
      "learning_rate": 2.1578439001409497e-05,
      "loss": 1.929,
      "step": 202417
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.395877718925476,
      "learning_rate": 2.157691569730258e-05,
      "loss": 1.9619,
      "step": 202418
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.3403600454330444,
      "learning_rate": 2.1575392444959816e-05,
      "loss": 1.876,
      "step": 202419
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.461621642112732,
      "learning_rate": 2.157386924438138e-05,
      "loss": 2.1025,
      "step": 202420
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.4456192255020142,
      "learning_rate": 2.157234609556767e-05,
      "loss": 1.9517,
      "step": 202421
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.3346740007400513,
      "learning_rate": 2.1570822998518956e-05,
      "loss": 1.8176,
      "step": 202422
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.4440803527832031,
      "learning_rate": 2.156929995323543e-05,
      "loss": 2.1059,
      "step": 202423
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.382856845855713,
      "learning_rate": 2.1567776959717397e-05,
      "loss": 1.9586,
      "step": 202424
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.4035788774490356,
      "learning_rate": 2.1566254017965225e-05,
      "loss": 1.9654,
      "step": 202425
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.4717556238174438,
      "learning_rate": 2.1564731127979108e-05,
      "loss": 1.9228,
      "step": 202426
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.5619462728500366,
      "learning_rate": 2.1563208289759382e-05,
      "loss": 1.8998,
      "step": 202427
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.406685709953308,
      "learning_rate": 2.1561685503306282e-05,
      "loss": 2.1134,
      "step": 202428
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.3252984285354614,
      "learning_rate": 2.1560162768620204e-05,
      "loss": 2.0722,
      "step": 202429
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.4992444515228271,
      "learning_rate": 2.1558640085701252e-05,
      "loss": 1.8899,
      "step": 202430
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.4339860677719116,
      "learning_rate": 2.1557117454549888e-05,
      "loss": 2.1132,
      "step": 202431
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.4762074947357178,
      "learning_rate": 2.1555594875166216e-05,
      "loss": 2.0359,
      "step": 202432
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.3403483629226685,
      "learning_rate": 2.15540723475507e-05,
      "loss": 2.0287,
      "step": 202433
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.3728094100952148,
      "learning_rate": 2.1552549871703474e-05,
      "loss": 1.8625,
      "step": 202434
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.4663703441619873,
      "learning_rate": 2.1551027447624967e-05,
      "loss": 1.9093,
      "step": 202435
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.4694830179214478,
      "learning_rate": 2.1549505075315288e-05,
      "loss": 2.0386,
      "step": 202436
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.5521807670593262,
      "learning_rate": 2.1547982754774862e-05,
      "loss": 2.0755,
      "step": 202437
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.3767577409744263,
      "learning_rate": 2.154646048600389e-05,
      "loss": 1.9759,
      "step": 202438
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.60097336769104,
      "learning_rate": 2.1544938269002744e-05,
      "loss": 1.8688,
      "step": 202439
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.6729090213775635,
      "learning_rate": 2.1543416103771583e-05,
      "loss": 2.1975,
      "step": 202440
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.3896907567977905,
      "learning_rate": 2.1541893990310842e-05,
      "loss": 2.138,
      "step": 202441
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.3605159521102905,
      "learning_rate": 2.1540371928620626e-05,
      "loss": 1.97,
      "step": 202442
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.4372341632843018,
      "learning_rate": 2.1538849918701396e-05,
      "loss": 2.0514,
      "step": 202443
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.4455983638763428,
      "learning_rate": 2.1537327960553253e-05,
      "loss": 2.0254,
      "step": 202444
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.3258726596832275,
      "learning_rate": 2.1535806054176662e-05,
      "loss": 1.785,
      "step": 202445
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.4884824752807617,
      "learning_rate": 2.153428419957176e-05,
      "loss": 2.2319,
      "step": 202446
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.407691240310669,
      "learning_rate": 2.153276239673891e-05,
      "loss": 2.0284,
      "step": 202447
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.4119545221328735,
      "learning_rate": 2.1531240645678417e-05,
      "loss": 1.9237,
      "step": 202448
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.3691540956497192,
      "learning_rate": 2.1529718946390506e-05,
      "loss": 2.0076,
      "step": 202449
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.7185531854629517,
      "learning_rate": 2.152819729887545e-05,
      "loss": 1.8822,
      "step": 202450
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.4787962436676025,
      "learning_rate": 2.152667570313358e-05,
      "loss": 2.1046,
      "step": 202451
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.4246597290039062,
      "learning_rate": 2.1525154159165127e-05,
      "loss": 1.9425,
      "step": 202452
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.546815276145935,
      "learning_rate": 2.1523632666970425e-05,
      "loss": 2.0251,
      "step": 202453
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.560647964477539,
      "learning_rate": 2.1522111226549745e-05,
      "loss": 1.8697,
      "step": 202454
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.393688440322876,
      "learning_rate": 2.152058983790338e-05,
      "loss": 2.0199,
      "step": 202455
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.385671854019165,
      "learning_rate": 2.1519068501031533e-05,
      "loss": 1.8623,
      "step": 202456
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.447744607925415,
      "learning_rate": 2.1517547215934604e-05,
      "loss": 2.001,
      "step": 202457
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.4142451286315918,
      "learning_rate": 2.151602598261276e-05,
      "loss": 1.9756,
      "step": 202458
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.4896469116210938,
      "learning_rate": 2.15145048010664e-05,
      "loss": 1.8912,
      "step": 202459
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.4937463998794556,
      "learning_rate": 2.1512983671295758e-05,
      "loss": 2.0596,
      "step": 202460
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.5775479078292847,
      "learning_rate": 2.15114625933011e-05,
      "loss": 1.8091,
      "step": 202461
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.46488618850708,
      "learning_rate": 2.1509941567082688e-05,
      "loss": 1.9706,
      "step": 202462
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.478767991065979,
      "learning_rate": 2.1508420592640863e-05,
      "loss": 2.0403,
      "step": 202463
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.5504461526870728,
      "learning_rate": 2.150689966997582e-05,
      "loss": 2.2389,
      "step": 202464
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.3691686391830444,
      "learning_rate": 2.1505378799087957e-05,
      "loss": 1.7956,
      "step": 202465
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.4966380596160889,
      "learning_rate": 2.150385797997751e-05,
      "loss": 2.0534,
      "step": 202466
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.6211696863174438,
      "learning_rate": 2.150233721264475e-05,
      "loss": 2.1144,
      "step": 202467
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.5157455205917358,
      "learning_rate": 2.1500816497089935e-05,
      "loss": 1.7371,
      "step": 202468
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.4460514783859253,
      "learning_rate": 2.14992958333134e-05,
      "loss": 1.8707,
      "step": 202469
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.3168561458587646,
      "learning_rate": 2.149777522131535e-05,
      "loss": 1.8661,
      "step": 202470
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.4745770692825317,
      "learning_rate": 2.1496254661096178e-05,
      "loss": 2.0296,
      "step": 202471
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.6381679773330688,
      "learning_rate": 2.1494734152656122e-05,
      "loss": 1.9736,
      "step": 202472
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.4736179113388062,
      "learning_rate": 2.1493213695995414e-05,
      "loss": 1.9923,
      "step": 202473
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.381313681602478,
      "learning_rate": 2.1491693291114352e-05,
      "loss": 1.9653,
      "step": 202474
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.6798352003097534,
      "learning_rate": 2.1490172938013305e-05,
      "loss": 2.024,
      "step": 202475
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.4465770721435547,
      "learning_rate": 2.148865263669244e-05,
      "loss": 1.9391,
      "step": 202476
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.2489198446273804,
      "learning_rate": 2.1487132387152116e-05,
      "loss": 1.8934,
      "step": 202477
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.4459809064865112,
      "learning_rate": 2.148561218939261e-05,
      "loss": 1.9154,
      "step": 202478
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.4026129245758057,
      "learning_rate": 2.148409204341418e-05,
      "loss": 1.8334,
      "step": 202479
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.4376393556594849,
      "learning_rate": 2.1482571949217065e-05,
      "loss": 1.9962,
      "step": 202480
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.5458139181137085,
      "learning_rate": 2.1481051906801628e-05,
      "loss": 2.1024,
      "step": 202481
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.4841041564941406,
      "learning_rate": 2.1479531916168103e-05,
      "loss": 1.8995,
      "step": 202482
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.4896149635314941,
      "learning_rate": 2.1478011977316823e-05,
      "loss": 1.9498,
      "step": 202483
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.4355292320251465,
      "learning_rate": 2.1476492090248055e-05,
      "loss": 2.2276,
      "step": 202484
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.7464030981063843,
      "learning_rate": 2.147497225496203e-05,
      "loss": 2.171,
      "step": 202485
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.483237385749817,
      "learning_rate": 2.1473452471459052e-05,
      "loss": 1.9243,
      "step": 202486
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.5820163488388062,
      "learning_rate": 2.147193273973945e-05,
      "loss": 1.7813,
      "step": 202487
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.4490458965301514,
      "learning_rate": 2.1470413059803425e-05,
      "loss": 1.8601,
      "step": 202488
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.395617961883545,
      "learning_rate": 2.1468893431651345e-05,
      "loss": 1.8503,
      "step": 202489
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.4713038206100464,
      "learning_rate": 2.1467373855283475e-05,
      "loss": 1.8891,
      "step": 202490
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.4271148443222046,
      "learning_rate": 2.1465854330700018e-05,
      "loss": 1.8186,
      "step": 202491
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.358283519744873,
      "learning_rate": 2.1464334857901366e-05,
      "loss": 1.9921,
      "step": 202492
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.4657763242721558,
      "learning_rate": 2.1462815436887758e-05,
      "loss": 1.8234,
      "step": 202493
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.4322936534881592,
      "learning_rate": 2.146129606765943e-05,
      "loss": 1.8997,
      "step": 202494
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.4189091920852661,
      "learning_rate": 2.145977675021674e-05,
      "loss": 1.8855,
      "step": 202495
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.4241305589675903,
      "learning_rate": 2.1458257484559926e-05,
      "loss": 1.7211,
      "step": 202496
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.455562949180603,
      "learning_rate": 2.1456738270689254e-05,
      "loss": 2.1072,
      "step": 202497
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.3749372959136963,
      "learning_rate": 2.145521910860506e-05,
      "loss": 1.9283,
      "step": 202498
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.7467783689498901,
      "learning_rate": 2.1453699998307605e-05,
      "loss": 2.1282,
      "step": 202499
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.6630642414093018,
      "learning_rate": 2.1452180939797126e-05,
      "loss": 1.9315,
      "step": 202500
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.4416602849960327,
      "learning_rate": 2.1450661933073986e-05,
      "loss": 2.0899,
      "step": 202501
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.6048294305801392,
      "learning_rate": 2.1449142978138424e-05,
      "loss": 2.0822,
      "step": 202502
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.5143674612045288,
      "learning_rate": 2.14476240749907e-05,
      "loss": 2.1054,
      "step": 202503
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.4107918739318848,
      "learning_rate": 2.144610522363115e-05,
      "loss": 2.1504,
      "step": 202504
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.434985637664795,
      "learning_rate": 2.1444586424059974e-05,
      "loss": 1.9857,
      "step": 202505
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.4642137289047241,
      "learning_rate": 2.144306767627757e-05,
      "loss": 1.9856,
      "step": 202506
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.437599539756775,
      "learning_rate": 2.144154898028414e-05,
      "loss": 2.0652,
      "step": 202507
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.5567585229873657,
      "learning_rate": 2.1440030336079983e-05,
      "loss": 2.0302,
      "step": 202508
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.3838403224945068,
      "learning_rate": 2.143851174366533e-05,
      "loss": 2.0368,
      "step": 202509
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.4748636484146118,
      "learning_rate": 2.1436993203040586e-05,
      "loss": 1.9028,
      "step": 202510
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.4035395383834839,
      "learning_rate": 2.1435474714205913e-05,
      "loss": 1.9358,
      "step": 202511
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.4409128427505493,
      "learning_rate": 2.143395627716168e-05,
      "loss": 1.9459,
      "step": 202512
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.359089970588684,
      "learning_rate": 2.143243789190808e-05,
      "loss": 1.7522,
      "step": 202513
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.4702869653701782,
      "learning_rate": 2.1430919558445526e-05,
      "loss": 2.0316,
      "step": 202514
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.428031325340271,
      "learning_rate": 2.1429401276774204e-05,
      "loss": 1.6492,
      "step": 202515
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.4198343753814697,
      "learning_rate": 2.142788304689439e-05,
      "loss": 2.246,
      "step": 202516
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.411449670791626,
      "learning_rate": 2.1426364868806377e-05,
      "loss": 1.9696,
      "step": 202517
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.5949593782424927,
      "learning_rate": 2.1424846742510506e-05,
      "loss": 2.0515,
      "step": 202518
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.5443443059921265,
      "learning_rate": 2.1423328668006935e-05,
      "loss": 2.2429,
      "step": 202519
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.4422646760940552,
      "learning_rate": 2.14218106452961e-05,
      "loss": 1.8127,
      "step": 202520
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.4597184658050537,
      "learning_rate": 2.1420292674378204e-05,
      "loss": 1.9465,
      "step": 202521
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.4303455352783203,
      "learning_rate": 2.1418774755253544e-05,
      "loss": 1.8678,
      "step": 202522
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.5151535272598267,
      "learning_rate": 2.141725688792235e-05,
      "loss": 1.9939,
      "step": 202523
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.4391390085220337,
      "learning_rate": 2.1415739072384965e-05,
      "loss": 2.0212,
      "step": 202524
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.4659770727157593,
      "learning_rate": 2.1414221308641645e-05,
      "loss": 1.9882,
      "step": 202525
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.3510102033615112,
      "learning_rate": 2.1412703596692692e-05,
      "loss": 2.0561,
      "step": 202526
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.4611152410507202,
      "learning_rate": 2.1411185936538378e-05,
      "loss": 1.9199,
      "step": 202527
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.49940824508667,
      "learning_rate": 2.1409668328178997e-05,
      "loss": 1.8888,
      "step": 202528
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.4401109218597412,
      "learning_rate": 2.1408150771614786e-05,
      "loss": 1.8605,
      "step": 202529
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.398762822151184,
      "learning_rate": 2.1406633266846074e-05,
      "loss": 1.937,
      "step": 202530
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.337787389755249,
      "learning_rate": 2.14051158138731e-05,
      "loss": 2.1117,
      "step": 202531
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.5039712190628052,
      "learning_rate": 2.140359841269622e-05,
      "loss": 1.9113,
      "step": 202532
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.439453363418579,
      "learning_rate": 2.1402081063315646e-05,
      "loss": 1.8458,
      "step": 202533
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.6975420713424683,
      "learning_rate": 2.1400563765731705e-05,
      "loss": 2.0806,
      "step": 202534
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.3916597366333008,
      "learning_rate": 2.1399046519944597e-05,
      "loss": 2.003,
      "step": 202535
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.43143630027771,
      "learning_rate": 2.139752932595472e-05,
      "loss": 1.9092,
      "step": 202536
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.5891761779785156,
      "learning_rate": 2.139601218376228e-05,
      "loss": 1.8713,
      "step": 202537
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.4158204793930054,
      "learning_rate": 2.1394495093367604e-05,
      "loss": 2.0581,
      "step": 202538
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.4970223903656006,
      "learning_rate": 2.139297805477096e-05,
      "loss": 1.729,
      "step": 202539
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.3351954221725464,
      "learning_rate": 2.1391461067972615e-05,
      "loss": 2.0451,
      "step": 202540
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.3756251335144043,
      "learning_rate": 2.1389944132972802e-05,
      "loss": 2.0149,
      "step": 202541
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.4344255924224854,
      "learning_rate": 2.138842724977189e-05,
      "loss": 2.087,
      "step": 202542
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.3614999055862427,
      "learning_rate": 2.1386910418370106e-05,
      "loss": 1.9098,
      "step": 202543
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.4770272970199585,
      "learning_rate": 2.138539363876779e-05,
      "loss": 2.0372,
      "step": 202544
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.487337589263916,
      "learning_rate": 2.1383876910965203e-05,
      "loss": 1.943,
      "step": 202545
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.4051839113235474,
      "learning_rate": 2.1382360234962614e-05,
      "loss": 2.0442,
      "step": 202546
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.6199727058410645,
      "learning_rate": 2.1380843610760223e-05,
      "loss": 1.8374,
      "step": 202547
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.5086768865585327,
      "learning_rate": 2.1379327038358464e-05,
      "loss": 2.1486,
      "step": 202548
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.7181020975112915,
      "learning_rate": 2.13778105177575e-05,
      "loss": 2.0477,
      "step": 202549
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.555748462677002,
      "learning_rate": 2.1376294048957733e-05,
      "loss": 2.3107,
      "step": 202550
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.4339478015899658,
      "learning_rate": 2.1374777631959326e-05,
      "loss": 2.0203,
      "step": 202551
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.429232120513916,
      "learning_rate": 2.1373261266762653e-05,
      "loss": 1.9329,
      "step": 202552
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.4308154582977295,
      "learning_rate": 2.1371744953367876e-05,
      "loss": 1.9062,
      "step": 202553
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.551066517829895,
      "learning_rate": 2.1370228691775393e-05,
      "loss": 2.0314,
      "step": 202554
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.4000463485717773,
      "learning_rate": 2.1368712481985407e-05,
      "loss": 1.9816,
      "step": 202555
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.464138150215149,
      "learning_rate": 2.136719632399828e-05,
      "loss": 2.1167,
      "step": 202556
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.339173674583435,
      "learning_rate": 2.136568021781425e-05,
      "loss": 1.9522,
      "step": 202557
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.5295366048812866,
      "learning_rate": 2.1364164163433584e-05,
      "loss": 1.963,
      "step": 202558
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.4428935050964355,
      "learning_rate": 2.1362648160856577e-05,
      "loss": 2.174,
      "step": 202559
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.3787204027175903,
      "learning_rate": 2.1361132210083533e-05,
      "loss": 1.909,
      "step": 202560
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.5495681762695312,
      "learning_rate": 2.1359616311114648e-05,
      "loss": 2.2398,
      "step": 202561
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.3928945064544678,
      "learning_rate": 2.1358100463950355e-05,
      "loss": 1.9099,
      "step": 202562
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.5323138236999512,
      "learning_rate": 2.1356584668590825e-05,
      "loss": 2.0193,
      "step": 202563
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.4309155941009521,
      "learning_rate": 2.1355068925036354e-05,
      "loss": 1.948,
      "step": 202564
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.4279075860977173,
      "learning_rate": 2.1353553233287212e-05,
      "loss": 2.053,
      "step": 202565
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.4271881580352783,
      "learning_rate": 2.1352037593343763e-05,
      "loss": 1.8849,
      "step": 202566
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.3868095874786377,
      "learning_rate": 2.135052200520617e-05,
      "loss": 2.0095,
      "step": 202567
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.5132136344909668,
      "learning_rate": 2.1349006468874808e-05,
      "loss": 2.0699,
      "step": 202568
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.5597478151321411,
      "learning_rate": 2.1347490984349935e-05,
      "loss": 2.2324,
      "step": 202569
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.7129735946655273,
      "learning_rate": 2.1345975551631823e-05,
      "loss": 1.893,
      "step": 202570
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.52335524559021,
      "learning_rate": 2.1344460170720733e-05,
      "loss": 1.9681,
      "step": 202571
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.3849947452545166,
      "learning_rate": 2.134294484161697e-05,
      "loss": 2.0316,
      "step": 202572
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.6618108749389648,
      "learning_rate": 2.13414295643208e-05,
      "loss": 2.2488,
      "step": 202573
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.4619146585464478,
      "learning_rate": 2.1339914338832553e-05,
      "loss": 1.854,
      "step": 202574
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.4212327003479004,
      "learning_rate": 2.1338399165152464e-05,
      "loss": 1.9245,
      "step": 202575
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.4436285495758057,
      "learning_rate": 2.13368840432808e-05,
      "loss": 2.0339,
      "step": 202576
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.2953845262527466,
      "learning_rate": 2.1335368973217927e-05,
      "loss": 2.0805,
      "step": 202577
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.4840023517608643,
      "learning_rate": 2.133385395496404e-05,
      "loss": 1.862,
      "step": 202578
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.482264518737793,
      "learning_rate": 2.1332338988519416e-05,
      "loss": 2.3677,
      "step": 202579
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.4690957069396973,
      "learning_rate": 2.1330824073884412e-05,
      "loss": 1.8802,
      "step": 202580
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.4211207628250122,
      "learning_rate": 2.1329309211059232e-05,
      "loss": 1.9527,
      "step": 202581
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.5181320905685425,
      "learning_rate": 2.1327794400044208e-05,
      "loss": 1.9241,
      "step": 202582
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.3657774925231934,
      "learning_rate": 2.1326279640839638e-05,
      "loss": 2.2059,
      "step": 202583
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.338761806488037,
      "learning_rate": 2.1324764933445793e-05,
      "loss": 2.0391,
      "step": 202584
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.4631386995315552,
      "learning_rate": 2.1323250277862836e-05,
      "loss": 2.1757,
      "step": 202585
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.5600223541259766,
      "learning_rate": 2.1321735674091234e-05,
      "loss": 2.1618,
      "step": 202586
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.4033620357513428,
      "learning_rate": 2.132022112213112e-05,
      "loss": 2.0268,
      "step": 202587
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.5356931686401367,
      "learning_rate": 2.1318706621982896e-05,
      "loss": 2.0708,
      "step": 202588
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.6669600009918213,
      "learning_rate": 2.1317192173646756e-05,
      "loss": 2.0159,
      "step": 202589
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.4118432998657227,
      "learning_rate": 2.1315677777122975e-05,
      "loss": 1.7921,
      "step": 202590
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.6300595998764038,
      "learning_rate": 2.1314163432411945e-05,
      "loss": 2.1727,
      "step": 202591
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.4251843690872192,
      "learning_rate": 2.131264913951384e-05,
      "loss": 2.1235,
      "step": 202592
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.3846198320388794,
      "learning_rate": 2.131113489842895e-05,
      "loss": 2.0306,
      "step": 202593
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.3759685754776,
      "learning_rate": 2.1309620709157615e-05,
      "loss": 1.8886,
      "step": 202594
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.537538766860962,
      "learning_rate": 2.1308106571700066e-05,
      "loss": 2.2134,
      "step": 202595
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.4904776811599731,
      "learning_rate": 2.1306592486056574e-05,
      "loss": 1.8811,
      "step": 202596
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.4890697002410889,
      "learning_rate": 2.13050784522275e-05,
      "loss": 2.2872,
      "step": 202597
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.4593477249145508,
      "learning_rate": 2.1303564470213073e-05,
      "loss": 1.9231,
      "step": 202598
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.577580213546753,
      "learning_rate": 2.1302050540013503e-05,
      "loss": 2.0938,
      "step": 202599
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.536643147468567,
      "learning_rate": 2.1300536661629218e-05,
      "loss": 1.9366,
      "step": 202600
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.4876259565353394,
      "learning_rate": 2.1299022835060388e-05,
      "loss": 1.8244,
      "step": 202601
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.5943210124969482,
      "learning_rate": 2.1297509060307304e-05,
      "loss": 1.754,
      "step": 202602
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.4517263174057007,
      "learning_rate": 2.1295995337370343e-05,
      "loss": 1.9569,
      "step": 202603
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.5586988925933838,
      "learning_rate": 2.129448166624963e-05,
      "loss": 1.9636,
      "step": 202604
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.3500550985336304,
      "learning_rate": 2.1292968046945602e-05,
      "loss": 1.7515,
      "step": 202605
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.574018120765686,
      "learning_rate": 2.1291454479458492e-05,
      "loss": 1.9692,
      "step": 202606
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.636080265045166,
      "learning_rate": 2.128994096378853e-05,
      "loss": 2.059,
      "step": 202607
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.3726669549942017,
      "learning_rate": 2.128842749993599e-05,
      "loss": 1.9504,
      "step": 202608
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.4459052085876465,
      "learning_rate": 2.1286914087901264e-05,
      "loss": 1.8817,
      "step": 202609
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.5261518955230713,
      "learning_rate": 2.128540072768449e-05,
      "loss": 1.9469,
      "step": 202610
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.5856423377990723,
      "learning_rate": 2.128388741928606e-05,
      "loss": 1.9296,
      "step": 202611
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.4440914392471313,
      "learning_rate": 2.1282374162706217e-05,
      "loss": 2.1489,
      "step": 202612
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.4571748971939087,
      "learning_rate": 2.1280860957945256e-05,
      "loss": 1.8171,
      "step": 202613
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.5237236022949219,
      "learning_rate": 2.127934780500341e-05,
      "loss": 1.7519,
      "step": 202614
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.3426809310913086,
      "learning_rate": 2.1277834703881013e-05,
      "loss": 2.2385,
      "step": 202615
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.8348267078399658,
      "learning_rate": 2.12763216545783e-05,
      "loss": 2.0723,
      "step": 202616
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.449220061302185,
      "learning_rate": 2.127480865709563e-05,
      "loss": 1.9603,
      "step": 202617
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.3883271217346191,
      "learning_rate": 2.1273295711433213e-05,
      "loss": 1.9706,
      "step": 202618
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.4366772174835205,
      "learning_rate": 2.1271782817591376e-05,
      "loss": 1.8744,
      "step": 202619
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.4757364988327026,
      "learning_rate": 2.127026997557032e-05,
      "loss": 1.9358,
      "step": 202620
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.378600001335144,
      "learning_rate": 2.126875718537041e-05,
      "loss": 2.2097,
      "step": 202621
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.517831563949585,
      "learning_rate": 2.1267244446991882e-05,
      "loss": 2.21,
      "step": 202622
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.6503570079803467,
      "learning_rate": 2.1265731760435067e-05,
      "loss": 2.1525,
      "step": 202623
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.5279054641723633,
      "learning_rate": 2.12642191257002e-05,
      "loss": 1.9479,
      "step": 202624
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.562933325767517,
      "learning_rate": 2.1262706542787612e-05,
      "loss": 1.8964,
      "step": 202625
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.489186406135559,
      "learning_rate": 2.126119401169747e-05,
      "loss": 2.0833,
      "step": 202626
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.477550983428955,
      "learning_rate": 2.1259681532430207e-05,
      "loss": 1.9533,
      "step": 202627
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.5225050449371338,
      "learning_rate": 2.1258169104985957e-05,
      "loss": 1.9807,
      "step": 202628
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.3350449800491333,
      "learning_rate": 2.1256656729365118e-05,
      "loss": 1.95,
      "step": 202629
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.5547714233398438,
      "learning_rate": 2.125514440556796e-05,
      "loss": 1.8859,
      "step": 202630
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.4801881313323975,
      "learning_rate": 2.125363213359471e-05,
      "loss": 1.9341,
      "step": 202631
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.3611911535263062,
      "learning_rate": 2.125211991344564e-05,
      "loss": 2.0721,
      "step": 202632
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.493715524673462,
      "learning_rate": 2.1250607745121086e-05,
      "loss": 1.7167,
      "step": 202633
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.5317233800888062,
      "learning_rate": 2.1249095628621273e-05,
      "loss": 2.034,
      "step": 202634
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.3248916864395142,
      "learning_rate": 2.1247583563946567e-05,
      "loss": 1.929,
      "step": 202635
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.551011562347412,
      "learning_rate": 2.124607155109721e-05,
      "loss": 1.9393,
      "step": 202636
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.4926953315734863,
      "learning_rate": 2.124455959007343e-05,
      "loss": 2.2008,
      "step": 202637
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.4999059438705444,
      "learning_rate": 2.1243047680875524e-05,
      "loss": 2.0161,
      "step": 202638
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.4412469863891602,
      "learning_rate": 2.1241535823503865e-05,
      "loss": 1.9193,
      "step": 202639
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.4963544607162476,
      "learning_rate": 2.1240024017958578e-05,
      "loss": 2.0342,
      "step": 202640
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.5648711919784546,
      "learning_rate": 2.1238512264240103e-05,
      "loss": 1.8546,
      "step": 202641
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.4076331853866577,
      "learning_rate": 2.123700056234867e-05,
      "loss": 1.9255,
      "step": 202642
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.3904424905776978,
      "learning_rate": 2.1235488912284514e-05,
      "loss": 1.7541,
      "step": 202643
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.3763327598571777,
      "learning_rate": 2.12339773140479e-05,
      "loss": 2.0688,
      "step": 202644
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.5686520338058472,
      "learning_rate": 2.1232465767639195e-05,
      "loss": 2.0942,
      "step": 202645
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.3798799514770508,
      "learning_rate": 2.12309542730586e-05,
      "loss": 2.1031,
      "step": 202646
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.5185469388961792,
      "learning_rate": 2.1229442830306474e-05,
      "loss": 1.8969,
      "step": 202647
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.5045033693313599,
      "learning_rate": 2.1227931439382995e-05,
      "loss": 2.0538,
      "step": 202648
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.473818063735962,
      "learning_rate": 2.122642010028862e-05,
      "loss": 2.0254,
      "step": 202649
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.3675477504730225,
      "learning_rate": 2.122490881302342e-05,
      "loss": 1.9778,
      "step": 202650
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.4536287784576416,
      "learning_rate": 2.12233975775878e-05,
      "loss": 1.9001,
      "step": 202651
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.5949788093566895,
      "learning_rate": 2.122188639398198e-05,
      "loss": 1.9667,
      "step": 202652
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.4750739336013794,
      "learning_rate": 2.1220375262206336e-05,
      "loss": 2.0825,
      "step": 202653
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.7169448137283325,
      "learning_rate": 2.1218864182260998e-05,
      "loss": 1.8586,
      "step": 202654
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.4122272729873657,
      "learning_rate": 2.1217353154146467e-05,
      "loss": 1.9516,
      "step": 202655
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.3330104351043701,
      "learning_rate": 2.1215842177862807e-05,
      "loss": 2.0198,
      "step": 202656
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.5418192148208618,
      "learning_rate": 2.1214331253410387e-05,
      "loss": 2.1913,
      "step": 202657
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.4203757047653198,
      "learning_rate": 2.1212820380789475e-05,
      "loss": 1.9323,
      "step": 202658
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.4427493810653687,
      "learning_rate": 2.12113095600004e-05,
      "loss": 1.9973,
      "step": 202659
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.3942607641220093,
      "learning_rate": 2.1209798791043364e-05,
      "loss": 2.049,
      "step": 202660
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.5374484062194824,
      "learning_rate": 2.1208288073918768e-05,
      "loss": 2.1081,
      "step": 202661
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.5357943773269653,
      "learning_rate": 2.1206777408626707e-05,
      "loss": 2.0586,
      "step": 202662
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.451491117477417,
      "learning_rate": 2.120526679516765e-05,
      "loss": 2.0309,
      "step": 202663
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.540398120880127,
      "learning_rate": 2.1203756233541737e-05,
      "loss": 2.0215,
      "step": 202664
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.3122626543045044,
      "learning_rate": 2.1202245723749356e-05,
      "loss": 2.0589,
      "step": 202665
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.471397876739502,
      "learning_rate": 2.120073526579068e-05,
      "loss": 2.032,
      "step": 202666
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.5290613174438477,
      "learning_rate": 2.119922485966611e-05,
      "loss": 2.2308,
      "step": 202667
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.5440181493759155,
      "learning_rate": 2.1197714505375872e-05,
      "loss": 1.8455,
      "step": 202668
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.4759228229522705,
      "learning_rate": 2.119620420292024e-05,
      "loss": 1.9376,
      "step": 202669
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.580230712890625,
      "learning_rate": 2.1194693952299413e-05,
      "loss": 2.0412,
      "step": 202670
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.6203489303588867,
      "learning_rate": 2.1193183753513855e-05,
      "loss": 1.9596,
      "step": 202671
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.7025198936462402,
      "learning_rate": 2.1191673606563663e-05,
      "loss": 1.9073,
      "step": 202672
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.4219073057174683,
      "learning_rate": 2.1190163511449276e-05,
      "loss": 1.9989,
      "step": 202673
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.5323814153671265,
      "learning_rate": 2.118865346817089e-05,
      "loss": 2.0049,
      "step": 202674
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.4547066688537598,
      "learning_rate": 2.118714347672874e-05,
      "loss": 2.1762,
      "step": 202675
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.5369112491607666,
      "learning_rate": 2.1185633537123226e-05,
      "loss": 1.9962,
      "step": 202676
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.5348604917526245,
      "learning_rate": 2.1184123649354545e-05,
      "loss": 1.9188,
      "step": 202677
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.5939403772354126,
      "learning_rate": 2.1182613813422966e-05,
      "loss": 1.9343,
      "step": 202678
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.4126373529434204,
      "learning_rate": 2.1181104029328855e-05,
      "loss": 1.7353,
      "step": 202679
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.54407799243927,
      "learning_rate": 2.117959429707241e-05,
      "loss": 2.3896,
      "step": 202680
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.37188720703125,
      "learning_rate": 2.1178084616653934e-05,
      "loss": 1.9548,
      "step": 202681
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.4227367639541626,
      "learning_rate": 2.1176574988073723e-05,
      "loss": 2.2004,
      "step": 202682
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.7280056476593018,
      "learning_rate": 2.1175065411332082e-05,
      "loss": 1.8079,
      "step": 202683
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.3978859186172485,
      "learning_rate": 2.1173555886429205e-05,
      "loss": 2.0185,
      "step": 202684
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.4528096914291382,
      "learning_rate": 2.1172046413365464e-05,
      "loss": 1.9764,
      "step": 202685
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.4663200378417969,
      "learning_rate": 2.1170536992141118e-05,
      "loss": 2.043,
      "step": 202686
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.6378401517868042,
      "learning_rate": 2.1169027622756374e-05,
      "loss": 2.1158,
      "step": 202687
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.833745002746582,
      "learning_rate": 2.116751830521163e-05,
      "loss": 1.83,
      "step": 202688
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.5541281700134277,
      "learning_rate": 2.1166009039507048e-05,
      "loss": 2.0951,
      "step": 202689
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.466943621635437,
      "learning_rate": 2.1164499825643032e-05,
      "loss": 2.209,
      "step": 202690
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.4822642803192139,
      "learning_rate": 2.1162990663619784e-05,
      "loss": 1.9492,
      "step": 202691
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.4763774871826172,
      "learning_rate": 2.1161481553437598e-05,
      "loss": 1.9022,
      "step": 202692
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.4210416078567505,
      "learning_rate": 2.115997249509671e-05,
      "loss": 1.8769,
      "step": 202693
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.398680567741394,
      "learning_rate": 2.1158463488597487e-05,
      "loss": 1.8952,
      "step": 202694
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.4363232851028442,
      "learning_rate": 2.115695453394013e-05,
      "loss": 2.0978,
      "step": 202695
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.4536250829696655,
      "learning_rate": 2.1155445631125002e-05,
      "loss": 2.041,
      "step": 202696
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.4886683225631714,
      "learning_rate": 2.115393678015237e-05,
      "loss": 1.9359,
      "step": 202697
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.6328532695770264,
      "learning_rate": 2.1152427981022434e-05,
      "loss": 2.0198,
      "step": 202698
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.3142348527908325,
      "learning_rate": 2.115091923373553e-05,
      "loss": 2.0028,
      "step": 202699
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.4696890115737915,
      "learning_rate": 2.1149410538291954e-05,
      "loss": 2.1852,
      "step": 202700
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.4062156677246094,
      "learning_rate": 2.114790189469191e-05,
      "loss": 1.9056,
      "step": 202701
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.4214988946914673,
      "learning_rate": 2.1146393302935793e-05,
      "loss": 1.7886,
      "step": 202702
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.4678630828857422,
      "learning_rate": 2.1144884763023838e-05,
      "loss": 1.9826,
      "step": 202703
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.3257172107696533,
      "learning_rate": 2.1143376274956282e-05,
      "loss": 1.9501,
      "step": 202704
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.3925796747207642,
      "learning_rate": 2.1141867838733418e-05,
      "loss": 2.0212,
      "step": 202705
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.3934063911437988,
      "learning_rate": 2.114035945435558e-05,
      "loss": 2.119,
      "step": 202706
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.4006074666976929,
      "learning_rate": 2.1138851121822977e-05,
      "loss": 1.9091,
      "step": 202707
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.6619060039520264,
      "learning_rate": 2.1137342841135928e-05,
      "loss": 2.0071,
      "step": 202708
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.4828490018844604,
      "learning_rate": 2.1135834612294776e-05,
      "loss": 2.0025,
      "step": 202709
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.529656171798706,
      "learning_rate": 2.1134326435299688e-05,
      "loss": 2.1813,
      "step": 202710
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.6141655445098877,
      "learning_rate": 2.1132818310150957e-05,
      "loss": 2.0266,
      "step": 202711
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.442269206047058,
      "learning_rate": 2.113131023684895e-05,
      "loss": 1.9313,
      "step": 202712
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.5101019144058228,
      "learning_rate": 2.1129802215393876e-05,
      "loss": 1.9936,
      "step": 202713
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.4552685022354126,
      "learning_rate": 2.112829424578606e-05,
      "loss": 1.8796,
      "step": 202714
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.5168604850769043,
      "learning_rate": 2.1126786328025703e-05,
      "loss": 2.0588,
      "step": 202715
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.3526098728179932,
      "learning_rate": 2.1125278462113236e-05,
      "loss": 1.9755,
      "step": 202716
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.3353207111358643,
      "learning_rate": 2.1123770648048764e-05,
      "loss": 2.2161,
      "step": 202717
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.5771679878234863,
      "learning_rate": 2.1122262885832686e-05,
      "loss": 1.9745,
      "step": 202718
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.4417623281478882,
      "learning_rate": 2.1120755175465165e-05,
      "loss": 1.7428,
      "step": 202719
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.4916599988937378,
      "learning_rate": 2.111924751694667e-05,
      "loss": 1.9019,
      "step": 202720
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.4344866275787354,
      "learning_rate": 2.1117739910277265e-05,
      "loss": 1.8268,
      "step": 202721
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.3820048570632935,
      "learning_rate": 2.1116232355457485e-05,
      "loss": 1.9068,
      "step": 202722
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.4365341663360596,
      "learning_rate": 2.111472485248733e-05,
      "loss": 2.191,
      "step": 202723
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.466212272644043,
      "learning_rate": 2.1113217401367266e-05,
      "loss": 2.0172,
      "step": 202724
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.5132240056991577,
      "learning_rate": 2.111171000209746e-05,
      "loss": 2.032,
      "step": 202725
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.5584285259246826,
      "learning_rate": 2.111020265467831e-05,
      "loss": 2.1262,
      "step": 202726
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.4501560926437378,
      "learning_rate": 2.1108695359110016e-05,
      "loss": 2.0945,
      "step": 202727
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.6204733848571777,
      "learning_rate": 2.110718811539295e-05,
      "loss": 1.9102,
      "step": 202728
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.4503532648086548,
      "learning_rate": 2.1105680923527235e-05,
      "loss": 2.1821,
      "step": 202729
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.3442106246948242,
      "learning_rate": 2.110417378351328e-05,
      "loss": 1.9301,
      "step": 202730
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.626064658164978,
      "learning_rate": 2.110266669535128e-05,
      "loss": 2.1669,
      "step": 202731
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.4316558837890625,
      "learning_rate": 2.11011596590416e-05,
      "loss": 1.9836,
      "step": 202732
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.4633123874664307,
      "learning_rate": 2.1099652674584445e-05,
      "loss": 2.024,
      "step": 202733
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.3988627195358276,
      "learning_rate": 2.109814574198021e-05,
      "loss": 1.9924,
      "step": 202734
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.4721838235855103,
      "learning_rate": 2.1096638861228997e-05,
      "loss": 1.9623,
      "step": 202735
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.7271676063537598,
      "learning_rate": 2.1095132032331242e-05,
      "loss": 2.2162,
      "step": 202736
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.4239503145217896,
      "learning_rate": 2.1093625255287107e-05,
      "loss": 1.9984,
      "step": 202737
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.3466073274612427,
      "learning_rate": 2.109211853009699e-05,
      "loss": 1.8256,
      "step": 202738
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.396353840827942,
      "learning_rate": 2.1090611856761064e-05,
      "loss": 1.9279,
      "step": 202739
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.312817931175232,
      "learning_rate": 2.1089105235279757e-05,
      "loss": 2.0624,
      "step": 202740
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.3871889114379883,
      "learning_rate": 2.108759866565314e-05,
      "loss": 2.2125,
      "step": 202741
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.4920896291732788,
      "learning_rate": 2.1086092147881672e-05,
      "loss": 1.9845,
      "step": 202742
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.4902493953704834,
      "learning_rate": 2.1084585681965494e-05,
      "loss": 2.1118,
      "step": 202743
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.4868042469024658,
      "learning_rate": 2.1083079267905033e-05,
      "loss": 2.0145,
      "step": 202744
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.4175256490707397,
      "learning_rate": 2.108157290570043e-05,
      "loss": 1.8958,
      "step": 202745
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.455628514289856,
      "learning_rate": 2.1080066595352108e-05,
      "loss": 1.9919,
      "step": 202746
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.5306717157363892,
      "learning_rate": 2.1078560336860174e-05,
      "loss": 2.0157,
      "step": 202747
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.3471599817276,
      "learning_rate": 2.1077054130225057e-05,
      "loss": 1.9696,
      "step": 202748
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.5005862712860107,
      "learning_rate": 2.107554797544696e-05,
      "loss": 1.8762,
      "step": 202749
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.3571891784667969,
      "learning_rate": 2.1074041872526183e-05,
      "loss": 1.6777,
      "step": 202750
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.3504072427749634,
      "learning_rate": 2.107253582146299e-05,
      "loss": 1.9103,
      "step": 202751
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.4200800657272339,
      "learning_rate": 2.1071029822257713e-05,
      "loss": 2.1289,
      "step": 202752
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.4236068725585938,
      "learning_rate": 2.106952387491062e-05,
      "loss": 1.9961,
      "step": 202753
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.54393470287323,
      "learning_rate": 2.1068017979421946e-05,
      "loss": 2.0901,
      "step": 202754
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.535044550895691,
      "learning_rate": 2.1066512135791923e-05,
      "loss": 2.1709,
      "step": 202755
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.3293681144714355,
      "learning_rate": 2.1065006344020984e-05,
      "loss": 2.1902,
      "step": 202756
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.4056214094161987,
      "learning_rate": 2.1063500604109264e-05,
      "loss": 1.9684,
      "step": 202757
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.4087094068527222,
      "learning_rate": 2.1061994916057157e-05,
      "loss": 2.1426,
      "step": 202758
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.3926295042037964,
      "learning_rate": 2.10604892798649e-05,
      "loss": 1.915,
      "step": 202759
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.4782052040100098,
      "learning_rate": 2.105898369553276e-05,
      "loss": 2.1072,
      "step": 202760
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.5577287673950195,
      "learning_rate": 2.1057478163060936e-05,
      "loss": 2.0892,
      "step": 202761
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.303160548210144,
      "learning_rate": 2.1055972682449895e-05,
      "loss": 2.0568,
      "step": 202762
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.399819254875183,
      "learning_rate": 2.1054467253699737e-05,
      "loss": 1.9166,
      "step": 202763
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.4487063884735107,
      "learning_rate": 2.1052961876810858e-05,
      "loss": 2.0019,
      "step": 202764
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.5199503898620605,
      "learning_rate": 2.105145655178353e-05,
      "loss": 2.0282,
      "step": 202765
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.4269500970840454,
      "learning_rate": 2.1049951278617914e-05,
      "loss": 2.1022,
      "step": 202766
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.3346130847930908,
      "learning_rate": 2.104844605731445e-05,
      "loss": 1.8409,
      "step": 202767
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.5576030015945435,
      "learning_rate": 2.104694088787333e-05,
      "loss": 2.1678,
      "step": 202768
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.5604627132415771,
      "learning_rate": 2.1045435770294828e-05,
      "loss": 1.9562,
      "step": 202769
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.2588508129119873,
      "learning_rate": 2.104393070457927e-05,
      "loss": 2.1006,
      "step": 202770
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.6441841125488281,
      "learning_rate": 2.1042425690726927e-05,
      "loss": 2.0954,
      "step": 202771
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.4197895526885986,
      "learning_rate": 2.1040920728737998e-05,
      "loss": 1.67,
      "step": 202772
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.3816999197006226,
      "learning_rate": 2.1039415818612913e-05,
      "loss": 1.9075,
      "step": 202773
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.2900044918060303,
      "learning_rate": 2.1037910960351778e-05,
      "loss": 1.9523,
      "step": 202774
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.517231822013855,
      "learning_rate": 2.103640615395502e-05,
      "loss": 1.9209,
      "step": 202775
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.3773396015167236,
      "learning_rate": 2.1034901399422876e-05,
      "loss": 2.1252,
      "step": 202776
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.322087049484253,
      "learning_rate": 2.1033396696755577e-05,
      "loss": 1.8805,
      "step": 202777
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.5690805912017822,
      "learning_rate": 2.1031892045953424e-05,
      "loss": 1.7251,
      "step": 202778
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.4096870422363281,
      "learning_rate": 2.1030387447016717e-05,
      "loss": 1.8321,
      "step": 202779
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.44491708278656,
      "learning_rate": 2.1028882899945722e-05,
      "loss": 1.9211,
      "step": 202780
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.426783800125122,
      "learning_rate": 2.1027378404740736e-05,
      "loss": 1.8981,
      "step": 202781
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.4398810863494873,
      "learning_rate": 2.102587396140203e-05,
      "loss": 1.9075,
      "step": 202782
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.4585108757019043,
      "learning_rate": 2.10243695699299e-05,
      "loss": 2.0032,
      "step": 202783
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.4677734375,
      "learning_rate": 2.1022865230324548e-05,
      "loss": 1.9926,
      "step": 202784
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.4545228481292725,
      "learning_rate": 2.1021360942586374e-05,
      "loss": 2.0067,
      "step": 202785
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.3238921165466309,
      "learning_rate": 2.101985670671551e-05,
      "loss": 2.1078,
      "step": 202786
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.387281060218811,
      "learning_rate": 2.101835252271239e-05,
      "loss": 2.0234,
      "step": 202787
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.393386721611023,
      "learning_rate": 2.1016848390577178e-05,
      "loss": 1.9437,
      "step": 202788
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.4773916006088257,
      "learning_rate": 2.1015344310310312e-05,
      "loss": 1.8423,
      "step": 202789
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.6483433246612549,
      "learning_rate": 2.1013840281911855e-05,
      "loss": 2.0709,
      "step": 202790
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.4151420593261719,
      "learning_rate": 2.1012336305382205e-05,
      "loss": 2.0236,
      "step": 202791
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.3412529230117798,
      "learning_rate": 2.101083238072163e-05,
      "loss": 2.0157,
      "step": 202792
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.4810620546340942,
      "learning_rate": 2.1009328507930434e-05,
      "loss": 1.9849,
      "step": 202793
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.5346912145614624,
      "learning_rate": 2.100782468700881e-05,
      "loss": 2.1342,
      "step": 202794
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.3885754346847534,
      "learning_rate": 2.10063209179572e-05,
      "loss": 1.7918,
      "step": 202795
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.5767719745635986,
      "learning_rate": 2.100481720077569e-05,
      "loss": 1.8925,
      "step": 202796
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.4937896728515625,
      "learning_rate": 2.1003313535464693e-05,
      "loss": 2.2011,
      "step": 202797
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.5069479942321777,
      "learning_rate": 2.1001809922024403e-05,
      "loss": 2.0806,
      "step": 202798
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.6142226457595825,
      "learning_rate": 2.100030636045519e-05,
      "loss": 2.1497,
      "step": 202799
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.420503854751587,
      "learning_rate": 2.0998802850757244e-05,
      "loss": 2.201,
      "step": 202800
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.6695582866668701,
      "learning_rate": 2.0997299392930978e-05,
      "loss": 1.7828,
      "step": 202801
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.3748492002487183,
      "learning_rate": 2.099579598697648e-05,
      "loss": 1.8331,
      "step": 202802
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.5272542238235474,
      "learning_rate": 2.0994292632894193e-05,
      "loss": 2.2223,
      "step": 202803
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.626500129699707,
      "learning_rate": 2.0992789330684278e-05,
      "loss": 1.9035,
      "step": 202804
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.5201116800308228,
      "learning_rate": 2.0991286080347135e-05,
      "loss": 2.0418,
      "step": 202805
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.3780347108840942,
      "learning_rate": 2.09897828818829e-05,
      "loss": 2.1176,
      "step": 202806
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.4314717054367065,
      "learning_rate": 2.0988279735292035e-05,
      "loss": 2.0873,
      "step": 202807
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.4106780290603638,
      "learning_rate": 2.0986776640574644e-05,
      "loss": 1.7783,
      "step": 202808
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.718017816543579,
      "learning_rate": 2.098527359773109e-05,
      "loss": 2.0991,
      "step": 202809
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.4490751028060913,
      "learning_rate": 2.0983770606761608e-05,
      "loss": 1.9778,
      "step": 202810
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.5556846857070923,
      "learning_rate": 2.0982267667666562e-05,
      "loss": 1.9963,
      "step": 202811
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.5427582263946533,
      "learning_rate": 2.0980764780446124e-05,
      "loss": 2.1656,
      "step": 202812
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.4159828424453735,
      "learning_rate": 2.0979261945100722e-05,
      "loss": 2.0046,
      "step": 202813
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.418921947479248,
      "learning_rate": 2.097775916163046e-05,
      "loss": 2.0796,
      "step": 202814
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.2943716049194336,
      "learning_rate": 2.097625643003573e-05,
      "loss": 2.1795,
      "step": 202815
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.4750474691390991,
      "learning_rate": 2.0974753750316743e-05,
      "loss": 2.0461,
      "step": 202816
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.4236085414886475,
      "learning_rate": 2.0973251122473887e-05,
      "loss": 2.3347,
      "step": 202817
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.3190711736679077,
      "learning_rate": 2.0971748546507307e-05,
      "loss": 1.8489,
      "step": 202818
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.407686471939087,
      "learning_rate": 2.0970246022417425e-05,
      "loss": 1.9628,
      "step": 202819
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.61228609085083,
      "learning_rate": 2.0968743550204348e-05,
      "loss": 1.8577,
      "step": 202820
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.363792896270752,
      "learning_rate": 2.0967241129868505e-05,
      "loss": 1.834,
      "step": 202821
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.369002103805542,
      "learning_rate": 2.096573876141007e-05,
      "loss": 2.0149,
      "step": 202822
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.5433237552642822,
      "learning_rate": 2.096423644482943e-05,
      "loss": 2.1487,
      "step": 202823
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.463516354560852,
      "learning_rate": 2.0962734180126728e-05,
      "loss": 1.9047,
      "step": 202824
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.3691002130508423,
      "learning_rate": 2.096123196730246e-05,
      "loss": 1.9663,
      "step": 202825
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.6856002807617188,
      "learning_rate": 2.0959729806356627e-05,
      "loss": 1.9516,
      "step": 202826
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.4413610696792603,
      "learning_rate": 2.0958227697289732e-05,
      "loss": 1.9485,
      "step": 202827
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.5288543701171875,
      "learning_rate": 2.09567256401019e-05,
      "loss": 1.8787,
      "step": 202828
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.5340427160263062,
      "learning_rate": 2.095522363479354e-05,
      "loss": 2.0467,
      "step": 202829
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.460033893585205,
      "learning_rate": 2.095372168136481e-05,
      "loss": 2.1569,
      "step": 202830
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.4210056066513062,
      "learning_rate": 2.0952219779816147e-05,
      "loss": 2.0437,
      "step": 202831
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.448589563369751,
      "learning_rate": 2.0950717930147654e-05,
      "loss": 1.9703,
      "step": 202832
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.3402397632598877,
      "learning_rate": 2.0949216132359726e-05,
      "loss": 1.7975,
      "step": 202833
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.5237829685211182,
      "learning_rate": 2.0947714386452563e-05,
      "loss": 2.0559,
      "step": 202834
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.4478708505630493,
      "learning_rate": 2.0946212692426533e-05,
      "loss": 1.9197,
      "step": 202835
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.4063951969146729,
      "learning_rate": 2.0944711050281803e-05,
      "loss": 1.8524,
      "step": 202836
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.3592969179153442,
      "learning_rate": 2.0943209460018807e-05,
      "loss": 1.7866,
      "step": 202837
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.7376717329025269,
      "learning_rate": 2.0941707921637707e-05,
      "loss": 2.004,
      "step": 202838
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.438535451889038,
      "learning_rate": 2.0940206435138805e-05,
      "loss": 2.0994,
      "step": 202839
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.4723018407821655,
      "learning_rate": 2.0938705000522372e-05,
      "loss": 1.9604,
      "step": 202840
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.339630365371704,
      "learning_rate": 2.09372036177887e-05,
      "loss": 1.9555,
      "step": 202841
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.512602686882019,
      "learning_rate": 2.093570228693806e-05,
      "loss": 1.7899,
      "step": 202842
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.3617424964904785,
      "learning_rate": 2.0934201007970786e-05,
      "loss": 2.0262,
      "step": 202843
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.452930212020874,
      "learning_rate": 2.093269978088711e-05,
      "loss": 2.1092,
      "step": 202844
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.4703662395477295,
      "learning_rate": 2.0931198605687292e-05,
      "loss": 1.8919,
      "step": 202845
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.5152817964553833,
      "learning_rate": 2.092969748237161e-05,
      "loss": 2.1238,
      "step": 202846
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.6572349071502686,
      "learning_rate": 2.092819641094039e-05,
      "loss": 2.0399,
      "step": 202847
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.4399218559265137,
      "learning_rate": 2.0926695391393868e-05,
      "loss": 1.9854,
      "step": 202848
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.3114091157913208,
      "learning_rate": 2.0925194423732373e-05,
      "loss": 2.0362,
      "step": 202849
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.440924048423767,
      "learning_rate": 2.092369350795614e-05,
      "loss": 1.7624,
      "step": 202850
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.6983566284179688,
      "learning_rate": 2.0922192644065438e-05,
      "loss": 1.9555,
      "step": 202851
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.5116610527038574,
      "learning_rate": 2.09206918320606e-05,
      "loss": 2.0678,
      "step": 202852
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.712946891784668,
      "learning_rate": 2.0919191071941856e-05,
      "loss": 1.787,
      "step": 202853
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.5076507329940796,
      "learning_rate": 2.091769036370947e-05,
      "loss": 1.8561,
      "step": 202854
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.4310369491577148,
      "learning_rate": 2.091618970736382e-05,
      "loss": 2.1582,
      "step": 202855
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.3498793840408325,
      "learning_rate": 2.0914689102905092e-05,
      "loss": 2.0208,
      "step": 202856
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.4319846630096436,
      "learning_rate": 2.091318855033356e-05,
      "loss": 2.092,
      "step": 202857
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.4527233839035034,
      "learning_rate": 2.0911688049649587e-05,
      "loss": 1.9103,
      "step": 202858
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.4832565784454346,
      "learning_rate": 2.0910187600853344e-05,
      "loss": 1.9235,
      "step": 202859
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.4117079973220825,
      "learning_rate": 2.0908687203945194e-05,
      "loss": 2.047,
      "step": 202860
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.363987684249878,
      "learning_rate": 2.0907186858925405e-05,
      "loss": 2.0073,
      "step": 202861
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.5011261701583862,
      "learning_rate": 2.0905686565794245e-05,
      "loss": 1.9648,
      "step": 202862
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.3804080486297607,
      "learning_rate": 2.090418632455194e-05,
      "loss": 1.8844,
      "step": 202863
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.4445903301239014,
      "learning_rate": 2.0902686135198866e-05,
      "loss": 2.1146,
      "step": 202864
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.4944509267807007,
      "learning_rate": 2.090118599773518e-05,
      "loss": 1.9848,
      "step": 202865
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.3593146800994873,
      "learning_rate": 2.0899685912161292e-05,
      "loss": 2.0962,
      "step": 202866
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.5310285091400146,
      "learning_rate": 2.0898185878477424e-05,
      "loss": 1.9542,
      "step": 202867
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.3003554344177246,
      "learning_rate": 2.0896685896683852e-05,
      "loss": 1.8963,
      "step": 202868
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.3991111516952515,
      "learning_rate": 2.0895185966780802e-05,
      "loss": 2.1179,
      "step": 202869
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.4553537368774414,
      "learning_rate": 2.0893686088768648e-05,
      "loss": 1.9117,
      "step": 202870
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.5936815738677979,
      "learning_rate": 2.089218626264758e-05,
      "loss": 2.0543,
      "step": 202871
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.3762038946151733,
      "learning_rate": 2.0890686488417972e-05,
      "loss": 1.845,
      "step": 202872
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.3943712711334229,
      "learning_rate": 2.0889186766080025e-05,
      "loss": 1.993,
      "step": 202873
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.5104626417160034,
      "learning_rate": 2.0887687095634097e-05,
      "loss": 1.9645,
      "step": 202874
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.4176493883132935,
      "learning_rate": 2.088618747708036e-05,
      "loss": 1.8964,
      "step": 202875
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.7845109701156616,
      "learning_rate": 2.0884687910419185e-05,
      "loss": 2.1509,
      "step": 202876
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.604365348815918,
      "learning_rate": 2.0883188395650764e-05,
      "loss": 2.0727,
      "step": 202877
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.5415141582489014,
      "learning_rate": 2.0881688932775498e-05,
      "loss": 2.1749,
      "step": 202878
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.509375810623169,
      "learning_rate": 2.0880189521793522e-05,
      "loss": 2.229,
      "step": 202879
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.4365792274475098,
      "learning_rate": 2.08786901627053e-05,
      "loss": 1.873,
      "step": 202880
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.4010522365570068,
      "learning_rate": 2.0877190855510907e-05,
      "loss": 1.8582,
      "step": 202881
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.3973509073257446,
      "learning_rate": 2.087569160021073e-05,
      "loss": 2.0426,
      "step": 202882
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.4412798881530762,
      "learning_rate": 2.087419239680501e-05,
      "loss": 1.9842,
      "step": 202883
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.5864675045013428,
      "learning_rate": 2.0872693245294082e-05,
      "loss": 1.9917,
      "step": 202884
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.4460484981536865,
      "learning_rate": 2.087119414567814e-05,
      "loss": 1.9937,
      "step": 202885
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.4122296571731567,
      "learning_rate": 2.0869695097957616e-05,
      "loss": 2.0262,
      "step": 202886
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.3921605348587036,
      "learning_rate": 2.0868196102132584e-05,
      "loss": 1.9862,
      "step": 202887
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.423411250114441,
      "learning_rate": 2.086669715820347e-05,
      "loss": 2.01,
      "step": 202888
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.4747785329818726,
      "learning_rate": 2.0865198266170447e-05,
      "loss": 1.9519,
      "step": 202889
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.3839778900146484,
      "learning_rate": 2.0863699426033942e-05,
      "loss": 1.9403,
      "step": 202890
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.4110428094863892,
      "learning_rate": 2.086220063779406e-05,
      "loss": 2.0467,
      "step": 202891
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.4437755346298218,
      "learning_rate": 2.086070190145126e-05,
      "loss": 1.9876,
      "step": 202892
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.3967787027359009,
      "learning_rate": 2.085920321700565e-05,
      "loss": 1.7813,
      "step": 202893
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.4895086288452148,
      "learning_rate": 2.0857704584457626e-05,
      "loss": 1.9886,
      "step": 202894
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.4315414428710938,
      "learning_rate": 2.085620600380735e-05,
      "loss": 2.0024,
      "step": 202895
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.3896152973175049,
      "learning_rate": 2.085470747505523e-05,
      "loss": 1.9062,
      "step": 202896
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.394814133644104,
      "learning_rate": 2.0853208998201465e-05,
      "loss": 2.0108,
      "step": 202897
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.3752785921096802,
      "learning_rate": 2.085171057324645e-05,
      "loss": 2.1399,
      "step": 202898
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.5215203762054443,
      "learning_rate": 2.0850212200190252e-05,
      "loss": 2.0976,
      "step": 202899
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.4185690879821777,
      "learning_rate": 2.0848713879033342e-05,
      "loss": 2.0574,
      "step": 202900
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.342495322227478,
      "learning_rate": 2.084721560977588e-05,
      "loss": 2.0161,
      "step": 202901
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.48965322971344,
      "learning_rate": 2.0845717392418204e-05,
      "loss": 2.1468,
      "step": 202902
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.535270094871521,
      "learning_rate": 2.0844219226960545e-05,
      "loss": 2.1034,
      "step": 202903
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.342817783355713,
      "learning_rate": 2.084272111340334e-05,
      "loss": 1.8502,
      "step": 202904
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.338166356086731,
      "learning_rate": 2.0841223051746614e-05,
      "loss": 2.0202,
      "step": 202905
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.5128259658813477,
      "learning_rate": 2.083972504199084e-05,
      "loss": 1.9937,
      "step": 202906
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.3449240922927856,
      "learning_rate": 2.0838227084136183e-05,
      "loss": 1.9324,
      "step": 202907
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.4103065729141235,
      "learning_rate": 2.083672917818301e-05,
      "loss": 2.0419,
      "step": 202908
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.4892064332962036,
      "learning_rate": 2.0835231324131484e-05,
      "loss": 1.8099,
      "step": 202909
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.4417517185211182,
      "learning_rate": 2.0833733521982075e-05,
      "loss": 1.8821,
      "step": 202910
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.3509583473205566,
      "learning_rate": 2.0832235771734885e-05,
      "loss": 2.0285,
      "step": 202911
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.4975529909133911,
      "learning_rate": 2.0830738073390242e-05,
      "loss": 2.0102,
      "step": 202912
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.7068816423416138,
      "learning_rate": 2.0829240426948414e-05,
      "loss": 2.1548,
      "step": 202913
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.4731940031051636,
      "learning_rate": 2.0827742832409766e-05,
      "loss": 2.1875,
      "step": 202914
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.4346882104873657,
      "learning_rate": 2.0826245289774434e-05,
      "loss": 2.2197,
      "step": 202915
    },
    {
      "epoch": 2.64,
      "grad_norm": 2.281261682510376,
      "learning_rate": 2.0824747799042818e-05,
      "loss": 2.0823,
      "step": 202916
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.3872162103652954,
      "learning_rate": 2.082325036021518e-05,
      "loss": 2.0409,
      "step": 202917
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.4704612493515015,
      "learning_rate": 2.082175297329176e-05,
      "loss": 2.005,
      "step": 202918
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.4063217639923096,
      "learning_rate": 2.082025563827279e-05,
      "loss": 2.0542,
      "step": 202919
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.2970876693725586,
      "learning_rate": 2.081875835515866e-05,
      "loss": 1.6985,
      "step": 202920
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.5434041023254395,
      "learning_rate": 2.0817261123949514e-05,
      "loss": 2.0956,
      "step": 202921
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.442914605140686,
      "learning_rate": 2.081576394464578e-05,
      "loss": 1.9748,
      "step": 202922
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.4218575954437256,
      "learning_rate": 2.0814266817247693e-05,
      "loss": 1.8188,
      "step": 202923
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.5187523365020752,
      "learning_rate": 2.0812769741755453e-05,
      "loss": 2.1437,
      "step": 202924
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.508079171180725,
      "learning_rate": 2.0811272718169392e-05,
      "loss": 1.7394,
      "step": 202925
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.4969379901885986,
      "learning_rate": 2.0809775746489808e-05,
      "loss": 2.1302,
      "step": 202926
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.502521276473999,
      "learning_rate": 2.0808278826716908e-05,
      "loss": 2.0439,
      "step": 202927
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.4775956869125366,
      "learning_rate": 2.0806781958851047e-05,
      "loss": 1.9958,
      "step": 202928
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.4225008487701416,
      "learning_rate": 2.08052851428925e-05,
      "loss": 1.9229,
      "step": 202929
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.4879311323165894,
      "learning_rate": 2.0803788378841536e-05,
      "loss": 1.9163,
      "step": 202930
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.555372953414917,
      "learning_rate": 2.0802291666698346e-05,
      "loss": 2.0375,
      "step": 202931
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.4458776712417603,
      "learning_rate": 2.0800795006463334e-05,
      "loss": 1.9901,
      "step": 202932
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.6239731311798096,
      "learning_rate": 2.0799298398136667e-05,
      "loss": 2.0481,
      "step": 202933
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.461187481880188,
      "learning_rate": 2.0797801841718743e-05,
      "loss": 2.2227,
      "step": 202934
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.439862847328186,
      "learning_rate": 2.0796305337209762e-05,
      "loss": 1.9021,
      "step": 202935
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.5072611570358276,
      "learning_rate": 2.0794808884609993e-05,
      "loss": 2.0476,
      "step": 202936
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.5868499279022217,
      "learning_rate": 2.0793312483919768e-05,
      "loss": 1.9647,
      "step": 202937
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.4011229276657104,
      "learning_rate": 2.079181613513935e-05,
      "loss": 1.6629,
      "step": 202938
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.5136044025421143,
      "learning_rate": 2.0790319838268943e-05,
      "loss": 2.0439,
      "step": 202939
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.5276440382003784,
      "learning_rate": 2.0788823593308944e-05,
      "loss": 2.0813,
      "step": 202940
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.362585425376892,
      "learning_rate": 2.0787327400259558e-05,
      "loss": 1.8721,
      "step": 202941
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.5157634019851685,
      "learning_rate": 2.0785831259121045e-05,
      "loss": 1.9002,
      "step": 202942
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.3662335872650146,
      "learning_rate": 2.078433516989377e-05,
      "loss": 1.8238,
      "step": 202943
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.4900885820388794,
      "learning_rate": 2.0782839132577943e-05,
      "loss": 2.144,
      "step": 202944
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.434721827507019,
      "learning_rate": 2.0781343147173822e-05,
      "loss": 1.9718,
      "step": 202945
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.494133710861206,
      "learning_rate": 2.0779847213681743e-05,
      "loss": 1.8671,
      "step": 202946
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.5719122886657715,
      "learning_rate": 2.077835133210197e-05,
      "loss": 1.9247,
      "step": 202947
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.4996271133422852,
      "learning_rate": 2.077685550243474e-05,
      "loss": 1.8405,
      "step": 202948
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.3780317306518555,
      "learning_rate": 2.077535972468042e-05,
      "loss": 1.758,
      "step": 202949
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.5077509880065918,
      "learning_rate": 2.0773863998839168e-05,
      "loss": 2.0034,
      "step": 202950
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.4916876554489136,
      "learning_rate": 2.077236832491136e-05,
      "loss": 2.0268,
      "step": 202951
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.4424207210540771,
      "learning_rate": 2.0770872702897255e-05,
      "loss": 1.7632,
      "step": 202952
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.3412303924560547,
      "learning_rate": 2.0769377132797126e-05,
      "loss": 1.9805,
      "step": 202953
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.3646984100341797,
      "learning_rate": 2.0767881614611172e-05,
      "loss": 1.8753,
      "step": 202954
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.6134713888168335,
      "learning_rate": 2.0766386148339786e-05,
      "loss": 2.0081,
      "step": 202955
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.5342278480529785,
      "learning_rate": 2.0764890733983174e-05,
      "loss": 1.9786,
      "step": 202956
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.4844914674758911,
      "learning_rate": 2.076339537154167e-05,
      "loss": 1.9911,
      "step": 202957
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.4565619230270386,
      "learning_rate": 2.0761900061015468e-05,
      "loss": 1.9498,
      "step": 202958
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.4015419483184814,
      "learning_rate": 2.0760404802405007e-05,
      "loss": 1.9856,
      "step": 202959
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.4128931760787964,
      "learning_rate": 2.075890959571038e-05,
      "loss": 1.8873,
      "step": 202960
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.3280179500579834,
      "learning_rate": 2.0757414440931963e-05,
      "loss": 1.9073,
      "step": 202961
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.52883780002594,
      "learning_rate": 2.0755919338069983e-05,
      "loss": 2.2277,
      "step": 202962
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.7396492958068848,
      "learning_rate": 2.0754424287124804e-05,
      "loss": 2.0089,
      "step": 202963
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.3466391563415527,
      "learning_rate": 2.0752929288096565e-05,
      "loss": 1.925,
      "step": 202964
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.5262699127197266,
      "learning_rate": 2.0751434340985763e-05,
      "loss": 2.0628,
      "step": 202965
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.4151804447174072,
      "learning_rate": 2.074993944579243e-05,
      "loss": 1.9293,
      "step": 202966
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.4759258031845093,
      "learning_rate": 2.0748444602516968e-05,
      "loss": 1.9734,
      "step": 202967
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.519336462020874,
      "learning_rate": 2.0746949811159642e-05,
      "loss": 1.7365,
      "step": 202968
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.4507969617843628,
      "learning_rate": 2.074545507172075e-05,
      "loss": 1.8483,
      "step": 202969
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.3190875053405762,
      "learning_rate": 2.0743960384200532e-05,
      "loss": 1.8614,
      "step": 202970
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.4828799962997437,
      "learning_rate": 2.0742465748599348e-05,
      "loss": 1.9749,
      "step": 202971
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.4057914018630981,
      "learning_rate": 2.074097116491733e-05,
      "loss": 1.9584,
      "step": 202972
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.4785919189453125,
      "learning_rate": 2.0739476633154883e-05,
      "loss": 1.8357,
      "step": 202973
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.4382319450378418,
      "learning_rate": 2.0737982153312172e-05,
      "loss": 2.1513,
      "step": 202974
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.4170145988464355,
      "learning_rate": 2.0736487725389595e-05,
      "loss": 1.9907,
      "step": 202975
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.340555191040039,
      "learning_rate": 2.0734993349387353e-05,
      "loss": 2.0313,
      "step": 202976
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.365308403968811,
      "learning_rate": 2.0733499025305812e-05,
      "loss": 1.9668,
      "step": 202977
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.5485092401504517,
      "learning_rate": 2.0732004753145103e-05,
      "loss": 2.219,
      "step": 202978
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.4274505376815796,
      "learning_rate": 2.0730510532905664e-05,
      "loss": 1.7861,
      "step": 202979
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.406201720237732,
      "learning_rate": 2.0729016364587593e-05,
      "loss": 2.1565,
      "step": 202980
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.3835253715515137,
      "learning_rate": 2.072752224819135e-05,
      "loss": 1.8035,
      "step": 202981
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.5660762786865234,
      "learning_rate": 2.072602818371708e-05,
      "loss": 1.8971,
      "step": 202982
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.5216017961502075,
      "learning_rate": 2.0724534171165173e-05,
      "loss": 2.0534,
      "step": 202983
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.4225916862487793,
      "learning_rate": 2.0723040210535833e-05,
      "loss": 2.0312,
      "step": 202984
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.5470912456512451,
      "learning_rate": 2.0721546301829363e-05,
      "loss": 1.996,
      "step": 202985
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.3902562856674194,
      "learning_rate": 2.0720052445045954e-05,
      "loss": 1.9023,
      "step": 202986
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.5475844144821167,
      "learning_rate": 2.071855864018601e-05,
      "loss": 2.0227,
      "step": 202987
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.3290797472000122,
      "learning_rate": 2.0717064887249736e-05,
      "loss": 1.761,
      "step": 202988
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.6050901412963867,
      "learning_rate": 2.0715571186237455e-05,
      "loss": 1.9198,
      "step": 202989
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.5795807838439941,
      "learning_rate": 2.0714077537149443e-05,
      "loss": 1.9019,
      "step": 202990
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.5024102926254272,
      "learning_rate": 2.071258393998596e-05,
      "loss": 2.1363,
      "step": 202991
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.476246953010559,
      "learning_rate": 2.0711090394747208e-05,
      "loss": 2.0636,
      "step": 202992
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.5309505462646484,
      "learning_rate": 2.070959690143359e-05,
      "loss": 1.7795,
      "step": 202993
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.4968352317810059,
      "learning_rate": 2.07081034600453e-05,
      "loss": 2.1279,
      "step": 202994
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.5068447589874268,
      "learning_rate": 2.0706610070582673e-05,
      "loss": 1.997,
      "step": 202995
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.6850693225860596,
      "learning_rate": 2.0705116733045945e-05,
      "loss": 2.104,
      "step": 202996
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.786873698234558,
      "learning_rate": 2.0703623447435446e-05,
      "loss": 1.9824,
      "step": 202997
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.5758472681045532,
      "learning_rate": 2.0702130213751344e-05,
      "loss": 1.9906,
      "step": 202998
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.6074538230895996,
      "learning_rate": 2.0700637031994038e-05,
      "loss": 2.0634,
      "step": 202999
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.4591177701950073,
      "learning_rate": 2.0699143902163727e-05,
      "loss": 2.1822,
      "step": 203000
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.4156218767166138,
      "learning_rate": 2.0697650824260748e-05,
      "loss": 1.7775,
      "step": 203001
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.497465968132019,
      "learning_rate": 2.069615779828533e-05,
      "loss": 2.0092,
      "step": 203002
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.4018718004226685,
      "learning_rate": 2.0694664824237805e-05,
      "loss": 1.9224,
      "step": 203003
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.4800751209259033,
      "learning_rate": 2.0693171902118343e-05,
      "loss": 2.0821,
      "step": 203004
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.4340895414352417,
      "learning_rate": 2.0691679031927345e-05,
      "loss": 2.042,
      "step": 203005
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.558248519897461,
      "learning_rate": 2.0690186213665005e-05,
      "loss": 2.1221,
      "step": 203006
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.4799373149871826,
      "learning_rate": 2.068869344733166e-05,
      "loss": 1.8826,
      "step": 203007
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.4008086919784546,
      "learning_rate": 2.0687200732927544e-05,
      "loss": 2.2359,
      "step": 203008
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.4038852453231812,
      "learning_rate": 2.0685708070452988e-05,
      "loss": 2.2025,
      "step": 203009
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.3865971565246582,
      "learning_rate": 2.068421545990816e-05,
      "loss": 1.8142,
      "step": 203010
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.425697922706604,
      "learning_rate": 2.068272290129346e-05,
      "loss": 2.0301,
      "step": 203011
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.5908944606781006,
      "learning_rate": 2.0681230394609082e-05,
      "loss": 2.1202,
      "step": 203012
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.406550407409668,
      "learning_rate": 2.0679737939855367e-05,
      "loss": 1.8695,
      "step": 203013
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.3338876962661743,
      "learning_rate": 2.0678245537032544e-05,
      "loss": 1.8777,
      "step": 203014
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.450725793838501,
      "learning_rate": 2.0676753186140948e-05,
      "loss": 1.9515,
      "step": 203015
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.5484519004821777,
      "learning_rate": 2.0675260887180746e-05,
      "loss": 2.0138,
      "step": 203016
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.4842904806137085,
      "learning_rate": 2.0673768640152334e-05,
      "loss": 1.9767,
      "step": 203017
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.5443143844604492,
      "learning_rate": 2.067227644505588e-05,
      "loss": 2.0649,
      "step": 203018
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.6037272214889526,
      "learning_rate": 2.0670784301891817e-05,
      "loss": 2.0659,
      "step": 203019
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.4797375202178955,
      "learning_rate": 2.0669292210660282e-05,
      "loss": 2.0252,
      "step": 203020
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.532062292098999,
      "learning_rate": 2.066780017136157e-05,
      "loss": 1.977,
      "step": 203021
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.4379206895828247,
      "learning_rate": 2.0666308183996018e-05,
      "loss": 2.0133,
      "step": 203022
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.5160653591156006,
      "learning_rate": 2.0664816248563888e-05,
      "loss": 1.8435,
      "step": 203023
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.435121774673462,
      "learning_rate": 2.066332436506538e-05,
      "loss": 1.9846,
      "step": 203024
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.5101944208145142,
      "learning_rate": 2.06618325335009e-05,
      "loss": 2.0294,
      "step": 203025
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.4920676946640015,
      "learning_rate": 2.066034075387064e-05,
      "loss": 2.131,
      "step": 203026
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.463726282119751,
      "learning_rate": 2.0658849026174873e-05,
      "loss": 2.2903,
      "step": 203027
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.3766740560531616,
      "learning_rate": 2.0657357350413928e-05,
      "loss": 2.0427,
      "step": 203028
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.4246058464050293,
      "learning_rate": 2.065586572658807e-05,
      "loss": 2.1116,
      "step": 203029
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.51734459400177,
      "learning_rate": 2.065437415469747e-05,
      "loss": 1.9504,
      "step": 203030
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.5631201267242432,
      "learning_rate": 2.0652882634742595e-05,
      "loss": 2.0826,
      "step": 203031
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.421692967414856,
      "learning_rate": 2.0651391166723574e-05,
      "loss": 1.8741,
      "step": 203032
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.3370106220245361,
      "learning_rate": 2.0649899750640707e-05,
      "loss": 1.7411,
      "step": 203033
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.4849003553390503,
      "learning_rate": 2.0648408386494363e-05,
      "loss": 1.8128,
      "step": 203034
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.361482858657837,
      "learning_rate": 2.0646917074284675e-05,
      "loss": 2.0305,
      "step": 203035
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.5219389200210571,
      "learning_rate": 2.0645425814012075e-05,
      "loss": 1.9552,
      "step": 203036
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.3947489261627197,
      "learning_rate": 2.0643934605676727e-05,
      "loss": 1.9325,
      "step": 203037
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.380433440208435,
      "learning_rate": 2.0642443449278966e-05,
      "loss": 1.7992,
      "step": 203038
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.6328884363174438,
      "learning_rate": 2.064095234481903e-05,
      "loss": 2.0258,
      "step": 203039
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.321752667427063,
      "learning_rate": 2.063946129229721e-05,
      "loss": 2.305,
      "step": 203040
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.3911352157592773,
      "learning_rate": 2.063797029171378e-05,
      "loss": 2.2511,
      "step": 203041
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.3946812152862549,
      "learning_rate": 2.0636479343069036e-05,
      "loss": 1.7953,
      "step": 203042
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.3384398221969604,
      "learning_rate": 2.0634988446363275e-05,
      "loss": 1.8257,
      "step": 203043
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.3481301069259644,
      "learning_rate": 2.0633497601596738e-05,
      "loss": 1.9695,
      "step": 203044
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.3914382457733154,
      "learning_rate": 2.0632006808769618e-05,
      "loss": 1.8571,
      "step": 203045
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.3842966556549072,
      "learning_rate": 2.0630516067882385e-05,
      "loss": 1.8299,
      "step": 203046
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.3812651634216309,
      "learning_rate": 2.0629025378935138e-05,
      "loss": 2.0845,
      "step": 203047
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.438147783279419,
      "learning_rate": 2.0627534741928276e-05,
      "loss": 2.0873,
      "step": 203048
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.465775728225708,
      "learning_rate": 2.0626044156861997e-05,
      "loss": 2.0856,
      "step": 203049
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.472042202949524,
      "learning_rate": 2.0624553623736638e-05,
      "loss": 1.9,
      "step": 203050
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.6308802366256714,
      "learning_rate": 2.0623063142552465e-05,
      "loss": 2.129,
      "step": 203051
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.525216817855835,
      "learning_rate": 2.0621572713309707e-05,
      "loss": 1.9129,
      "step": 203052
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.4502958059310913,
      "learning_rate": 2.062008233600867e-05,
      "loss": 2.0921,
      "step": 203053
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.445393681526184,
      "learning_rate": 2.0618592010649648e-05,
      "loss": 1.9431,
      "step": 203054
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.4968771934509277,
      "learning_rate": 2.061710173723288e-05,
      "loss": 1.8525,
      "step": 203055
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.3091998100280762,
      "learning_rate": 2.061561151575869e-05,
      "loss": 2.0672,
      "step": 203056
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.507744550704956,
      "learning_rate": 2.0614121346227354e-05,
      "loss": 2.0038,
      "step": 203057
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.5563324689865112,
      "learning_rate": 2.0612631228639097e-05,
      "loss": 2.0354,
      "step": 203058
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.5208371877670288,
      "learning_rate": 2.0611141162994193e-05,
      "loss": 2.0447,
      "step": 203059
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.3152954578399658,
      "learning_rate": 2.0609651149293006e-05,
      "loss": 1.8377,
      "step": 203060
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.4788646697998047,
      "learning_rate": 2.06081611875357e-05,
      "loss": 2.059,
      "step": 203061
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.7034327983856201,
      "learning_rate": 2.0606671277722674e-05,
      "loss": 2.1303,
      "step": 203062
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.535588264465332,
      "learning_rate": 2.0605181419854132e-05,
      "loss": 2.0206,
      "step": 203063
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.6264734268188477,
      "learning_rate": 2.0603691613930374e-05,
      "loss": 1.9652,
      "step": 203064
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.5384329557418823,
      "learning_rate": 2.0602201859951596e-05,
      "loss": 2.0605,
      "step": 203065
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.4386998414993286,
      "learning_rate": 2.0600712157918197e-05,
      "loss": 2.0496,
      "step": 203066
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.4311509132385254,
      "learning_rate": 2.0599222507830348e-05,
      "loss": 1.9492,
      "step": 203067
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.4459675550460815,
      "learning_rate": 2.0597732909688447e-05,
      "loss": 1.9345,
      "step": 203068
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.4069851636886597,
      "learning_rate": 2.0596243363492692e-05,
      "loss": 2.0737,
      "step": 203069
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.2825523614883423,
      "learning_rate": 2.0594753869243353e-05,
      "loss": 1.8964,
      "step": 203070
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.4568207263946533,
      "learning_rate": 2.0593264426940692e-05,
      "loss": 1.9888,
      "step": 203071
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.3846604824066162,
      "learning_rate": 2.0591775036585047e-05,
      "loss": 2.1013,
      "step": 203072
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.4431852102279663,
      "learning_rate": 2.0590285698176645e-05,
      "loss": 2.1933,
      "step": 203073
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.500349521636963,
      "learning_rate": 2.0588796411715825e-05,
      "loss": 2.0146,
      "step": 203074
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.339091181755066,
      "learning_rate": 2.0587307177202783e-05,
      "loss": 1.8624,
      "step": 203075
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.3012006282806396,
      "learning_rate": 2.0585817994637886e-05,
      "loss": 2.1136,
      "step": 203076
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.542202353477478,
      "learning_rate": 2.0584328864021272e-05,
      "loss": 1.9891,
      "step": 203077
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.563376784324646,
      "learning_rate": 2.0582839785353366e-05,
      "loss": 1.9127,
      "step": 203078
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.4287039041519165,
      "learning_rate": 2.0581350758634375e-05,
      "loss": 2.0901,
      "step": 203079
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.5753991603851318,
      "learning_rate": 2.0579861783864594e-05,
      "loss": 2.1543,
      "step": 203080
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.4461653232574463,
      "learning_rate": 2.057837286104429e-05,
      "loss": 1.9031,
      "step": 203081
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.4551634788513184,
      "learning_rate": 2.0576883990173766e-05,
      "loss": 1.9726,
      "step": 203082
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.4777215719223022,
      "learning_rate": 2.0575395171253185e-05,
      "loss": 1.9242,
      "step": 203083
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.349756121635437,
      "learning_rate": 2.057390640428298e-05,
      "loss": 2.02,
      "step": 203084
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.4314186573028564,
      "learning_rate": 2.057241768926332e-05,
      "loss": 2.1949,
      "step": 203085
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.425968050956726,
      "learning_rate": 2.057092902619457e-05,
      "loss": 2.0406,
      "step": 203086
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.4396976232528687,
      "learning_rate": 2.0569440415076965e-05,
      "loss": 1.9705,
      "step": 203087
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.4489381313323975,
      "learning_rate": 2.056795185591077e-05,
      "loss": 1.9217,
      "step": 203088
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.5334564447402954,
      "learning_rate": 2.0566463348696216e-05,
      "loss": 2.1195,
      "step": 203089
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.289226770401001,
      "learning_rate": 2.0564974893433672e-05,
      "loss": 2.095,
      "step": 203090
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.4463825225830078,
      "learning_rate": 2.056348649012334e-05,
      "loss": 1.8977,
      "step": 203091
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.3135504722595215,
      "learning_rate": 2.0561998138765578e-05,
      "loss": 1.9256,
      "step": 203092
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.3919992446899414,
      "learning_rate": 2.0560509839360596e-05,
      "loss": 2.0026,
      "step": 203093
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.3389722108840942,
      "learning_rate": 2.0559021591908685e-05,
      "loss": 1.7735,
      "step": 203094
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.3696002960205078,
      "learning_rate": 2.055753339641012e-05,
      "loss": 1.9695,
      "step": 203095
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.512947678565979,
      "learning_rate": 2.0556045252865194e-05,
      "loss": 1.762,
      "step": 203096
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.3670451641082764,
      "learning_rate": 2.0554557161274142e-05,
      "loss": 2.0592,
      "step": 203097
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.391189694404602,
      "learning_rate": 2.0553069121637333e-05,
      "loss": 2.1186,
      "step": 203098
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.38068687915802,
      "learning_rate": 2.0551581133954964e-05,
      "loss": 2.1474,
      "step": 203099
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.4936902523040771,
      "learning_rate": 2.0550093198227335e-05,
      "loss": 2.0502,
      "step": 203100
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.401216983795166,
      "learning_rate": 2.0548605314454682e-05,
      "loss": 2.0318,
      "step": 203101
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.4442710876464844,
      "learning_rate": 2.0547117482637366e-05,
      "loss": 2.1042,
      "step": 203102
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.6098661422729492,
      "learning_rate": 2.0545629702775557e-05,
      "loss": 1.9418,
      "step": 203103
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.3851890563964844,
      "learning_rate": 2.054414197486962e-05,
      "loss": 2.1018,
      "step": 203104
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.3974769115447998,
      "learning_rate": 2.0542654298919824e-05,
      "loss": 1.9368,
      "step": 203105
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.5814013481140137,
      "learning_rate": 2.0541166674926436e-05,
      "loss": 2.0748,
      "step": 203106
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.4603534936904907,
      "learning_rate": 2.053967910288965e-05,
      "loss": 1.8453,
      "step": 203107
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.5671437978744507,
      "learning_rate": 2.053819158280987e-05,
      "loss": 1.9943,
      "step": 203108
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.4734498262405396,
      "learning_rate": 2.0536704114687262e-05,
      "loss": 2.0459,
      "step": 203109
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.5070945024490356,
      "learning_rate": 2.0535216698522227e-05,
      "loss": 1.8566,
      "step": 203110
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.4291551113128662,
      "learning_rate": 2.0533729334314962e-05,
      "loss": 2.0061,
      "step": 203111
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.4203749895095825,
      "learning_rate": 2.053224202206567e-05,
      "loss": 1.9522,
      "step": 203112
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.7106884717941284,
      "learning_rate": 2.0530754761774782e-05,
      "loss": 2.1224,
      "step": 203113
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.441681146621704,
      "learning_rate": 2.052926755344253e-05,
      "loss": 1.8388,
      "step": 203114
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.401149868965149,
      "learning_rate": 2.052778039706908e-05,
      "loss": 1.988,
      "step": 203115
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.48843252658844,
      "learning_rate": 2.0526293292654838e-05,
      "loss": 2.0414,
      "step": 203116
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.6327097415924072,
      "learning_rate": 2.0524806240199997e-05,
      "loss": 2.043,
      "step": 203117
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.5686396360397339,
      "learning_rate": 2.0523319239704926e-05,
      "loss": 2.1222,
      "step": 203118
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.496858835220337,
      "learning_rate": 2.0521832291169827e-05,
      "loss": 2.0064,
      "step": 203119
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.428247332572937,
      "learning_rate": 2.052034539459496e-05,
      "loss": 1.8705,
      "step": 203120
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.635549545288086,
      "learning_rate": 2.0518858549980667e-05,
      "loss": 2.1249,
      "step": 203121
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.6133337020874023,
      "learning_rate": 2.0517371757327206e-05,
      "loss": 2.073,
      "step": 203122
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.4979243278503418,
      "learning_rate": 2.0515885016634782e-05,
      "loss": 1.889,
      "step": 203123
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.416797161102295,
      "learning_rate": 2.051439832790379e-05,
      "loss": 2.0204,
      "step": 203124
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.6130292415618896,
      "learning_rate": 2.051291169113447e-05,
      "loss": 1.9509,
      "step": 203125
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.4951143264770508,
      "learning_rate": 2.0511425106326985e-05,
      "loss": 1.9339,
      "step": 203126
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.511614441871643,
      "learning_rate": 2.0509938573481765e-05,
      "loss": 1.9788,
      "step": 203127
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.4748773574829102,
      "learning_rate": 2.050845209259905e-05,
      "loss": 2.071,
      "step": 203128
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.5132497549057007,
      "learning_rate": 2.0506965663678998e-05,
      "loss": 1.772,
      "step": 203129
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.437363624572754,
      "learning_rate": 2.050547928672205e-05,
      "loss": 2.0289,
      "step": 203130
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.3746896982192993,
      "learning_rate": 2.0503992961728398e-05,
      "loss": 1.8785,
      "step": 203131
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.3645285367965698,
      "learning_rate": 2.0502506688698283e-05,
      "loss": 2.0558,
      "step": 203132
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.4396235942840576,
      "learning_rate": 2.05010204676321e-05,
      "loss": 1.8535,
      "step": 203133
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.4396864175796509,
      "learning_rate": 2.0499534298529985e-05,
      "loss": 2.0561,
      "step": 203134
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.449495792388916,
      "learning_rate": 2.0498048181392333e-05,
      "loss": 1.9384,
      "step": 203135
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.4490406513214111,
      "learning_rate": 2.049656211621935e-05,
      "loss": 1.9385,
      "step": 203136
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.3722032308578491,
      "learning_rate": 2.0495076103011364e-05,
      "loss": 1.7398,
      "step": 203137
    },
    {
      "epoch": 2.64,
      "grad_norm": 1.4823449850082397,
      "learning_rate": 2.0493590141768547e-05,
      "loss": 2.1052,
      "step": 203138
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.385197401046753,
      "learning_rate": 2.0492104232491324e-05,
      "loss": 2.0201,
      "step": 203139
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.4092237949371338,
      "learning_rate": 2.0490618375179836e-05,
      "loss": 1.7282,
      "step": 203140
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.3749868869781494,
      "learning_rate": 2.0489132569834443e-05,
      "loss": 2.0083,
      "step": 203141
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.403955340385437,
      "learning_rate": 2.0487646816455416e-05,
      "loss": 2.1316,
      "step": 203142
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.366623044013977,
      "learning_rate": 2.048616111504302e-05,
      "loss": 1.9997,
      "step": 203143
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.602278709411621,
      "learning_rate": 2.0484675465597456e-05,
      "loss": 2.0734,
      "step": 203144
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.5119813680648804,
      "learning_rate": 2.048318986811912e-05,
      "loss": 1.8911,
      "step": 203145
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.3307794332504272,
      "learning_rate": 2.0481704322608183e-05,
      "loss": 1.9924,
      "step": 203146
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.5789059400558472,
      "learning_rate": 2.0480218829065044e-05,
      "loss": 1.972,
      "step": 203147
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.4271613359451294,
      "learning_rate": 2.0478733387489897e-05,
      "loss": 2.0058,
      "step": 203148
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.559098720550537,
      "learning_rate": 2.0477247997883016e-05,
      "loss": 2.1361,
      "step": 203149
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.4330788850784302,
      "learning_rate": 2.0475762660244667e-05,
      "loss": 2.0399,
      "step": 203150
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.4170767068862915,
      "learning_rate": 2.0474277374575176e-05,
      "loss": 2.1104,
      "step": 203151
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.522325038909912,
      "learning_rate": 2.047279214087475e-05,
      "loss": 1.8364,
      "step": 203152
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.5268187522888184,
      "learning_rate": 2.0471306959143753e-05,
      "loss": 2.0126,
      "step": 203153
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.5443787574768066,
      "learning_rate": 2.046982182938245e-05,
      "loss": 1.9119,
      "step": 203154
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.3742430210113525,
      "learning_rate": 2.046833675159104e-05,
      "loss": 1.8624,
      "step": 203155
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.3613237142562866,
      "learning_rate": 2.0466851725769828e-05,
      "loss": 1.8711,
      "step": 203156
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.4795840978622437,
      "learning_rate": 2.0465366751919143e-05,
      "loss": 2.156,
      "step": 203157
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.5241448879241943,
      "learning_rate": 2.0463881830039153e-05,
      "loss": 1.9472,
      "step": 203158
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.577551245689392,
      "learning_rate": 2.0462396960130255e-05,
      "loss": 1.9767,
      "step": 203159
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.4572349786758423,
      "learning_rate": 2.0460912142192687e-05,
      "loss": 1.9406,
      "step": 203160
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.4073352813720703,
      "learning_rate": 2.0459427376226712e-05,
      "loss": 1.9135,
      "step": 203161
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.5934131145477295,
      "learning_rate": 2.045794266223256e-05,
      "loss": 2.1965,
      "step": 203162
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.486796498298645,
      "learning_rate": 2.0456458000210608e-05,
      "loss": 1.936,
      "step": 203163
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.416513204574585,
      "learning_rate": 2.045497339016101e-05,
      "loss": 1.9748,
      "step": 203164
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.5403560400009155,
      "learning_rate": 2.0453488832084175e-05,
      "loss": 2.1598,
      "step": 203165
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.4043747186660767,
      "learning_rate": 2.0452004325980333e-05,
      "loss": 1.8727,
      "step": 203166
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.3866605758666992,
      "learning_rate": 2.0450519871849714e-05,
      "loss": 1.8865,
      "step": 203167
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.4049475193023682,
      "learning_rate": 2.0449035469692555e-05,
      "loss": 1.8963,
      "step": 203168
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.4971625804901123,
      "learning_rate": 2.0447551119509254e-05,
      "loss": 1.9657,
      "step": 203169
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.3921287059783936,
      "learning_rate": 2.0446066821300013e-05,
      "loss": 1.843,
      "step": 203170
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.4555095434188843,
      "learning_rate": 2.0444582575065127e-05,
      "loss": 2.1818,
      "step": 203171
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.4392342567443848,
      "learning_rate": 2.04430983808049e-05,
      "loss": 2.0232,
      "step": 203172
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.5846915245056152,
      "learning_rate": 2.0441614238519564e-05,
      "loss": 1.9996,
      "step": 203173
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.5243159532546997,
      "learning_rate": 2.044013014820939e-05,
      "loss": 2.1189,
      "step": 203174
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.552833080291748,
      "learning_rate": 2.0438646109874702e-05,
      "loss": 1.8546,
      "step": 203175
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.4403468370437622,
      "learning_rate": 2.043716212351567e-05,
      "loss": 2.1428,
      "step": 203176
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.5824979543685913,
      "learning_rate": 2.043567818913273e-05,
      "loss": 2.0627,
      "step": 203177
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.4388210773468018,
      "learning_rate": 2.0434194306726082e-05,
      "loss": 1.9055,
      "step": 203178
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.5293724536895752,
      "learning_rate": 2.0432710476295987e-05,
      "loss": 2.19,
      "step": 203179
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.343307375907898,
      "learning_rate": 2.0431226697842652e-05,
      "loss": 2.0118,
      "step": 203180
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.3770086765289307,
      "learning_rate": 2.04297429713665e-05,
      "loss": 2.0585,
      "step": 203181
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.391499400138855,
      "learning_rate": 2.0428259296867677e-05,
      "loss": 1.8771,
      "step": 203182
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.5723342895507812,
      "learning_rate": 2.0426775674346572e-05,
      "loss": 2.0639,
      "step": 203183
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.563942551612854,
      "learning_rate": 2.042529210380336e-05,
      "loss": 1.9622,
      "step": 203184
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.56452214717865,
      "learning_rate": 2.0423808585238466e-05,
      "loss": 2.0517,
      "step": 203185
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.5549964904785156,
      "learning_rate": 2.042232511865196e-05,
      "loss": 2.1516,
      "step": 203186
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.5123525857925415,
      "learning_rate": 2.0420841704044277e-05,
      "loss": 1.9827,
      "step": 203187
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.4918519258499146,
      "learning_rate": 2.041935834141558e-05,
      "loss": 2.0334,
      "step": 203188
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.6175734996795654,
      "learning_rate": 2.0417875030766274e-05,
      "loss": 2.251,
      "step": 203189
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.8345963954925537,
      "learning_rate": 2.0416391772096486e-05,
      "loss": 1.9464,
      "step": 203190
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.4790606498718262,
      "learning_rate": 2.041490856540665e-05,
      "loss": 1.8229,
      "step": 203191
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.3656195402145386,
      "learning_rate": 2.0413425410696903e-05,
      "loss": 1.8263,
      "step": 203192
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.4913073778152466,
      "learning_rate": 2.0411942307967612e-05,
      "loss": 1.9102,
      "step": 203193
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.549264669418335,
      "learning_rate": 2.041045925721897e-05,
      "loss": 2.0417,
      "step": 203194
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.449734091758728,
      "learning_rate": 2.040897625845135e-05,
      "loss": 2.0926,
      "step": 203195
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.417763590812683,
      "learning_rate": 2.0407493311664914e-05,
      "loss": 1.9939,
      "step": 203196
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.4551950693130493,
      "learning_rate": 2.0406010416860063e-05,
      "loss": 2.0747,
      "step": 203197
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.5780848264694214,
      "learning_rate": 2.0404527574037032e-05,
      "loss": 2.0268,
      "step": 203198
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.516983985900879,
      "learning_rate": 2.0403044783196053e-05,
      "loss": 1.8614,
      "step": 203199
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.3268465995788574,
      "learning_rate": 2.040156204433736e-05,
      "loss": 2.4376,
      "step": 203200
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.339879035949707,
      "learning_rate": 2.0400079357461386e-05,
      "loss": 2.0744,
      "step": 203201
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.4277927875518799,
      "learning_rate": 2.039859672256823e-05,
      "loss": 2.0089,
      "step": 203202
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.2873623371124268,
      "learning_rate": 2.039711413965832e-05,
      "loss": 2.0465,
      "step": 203203
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.418195366859436,
      "learning_rate": 2.0395631608731866e-05,
      "loss": 1.9541,
      "step": 203204
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.5079472064971924,
      "learning_rate": 2.0394149129789126e-05,
      "loss": 2.0442,
      "step": 203205
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.5956029891967773,
      "learning_rate": 2.0392666702830374e-05,
      "loss": 2.1801,
      "step": 203206
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.368346929550171,
      "learning_rate": 2.0391184327855935e-05,
      "loss": 1.931,
      "step": 203207
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.382225751876831,
      "learning_rate": 2.0389702004866016e-05,
      "loss": 1.9172,
      "step": 203208
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.4184954166412354,
      "learning_rate": 2.0388219733860944e-05,
      "loss": 1.8747,
      "step": 203209
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.480607509613037,
      "learning_rate": 2.0386737514840988e-05,
      "loss": 2.078,
      "step": 203210
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.3682736158370972,
      "learning_rate": 2.0385255347806385e-05,
      "loss": 1.9117,
      "step": 203211
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.600840449333191,
      "learning_rate": 2.0383773232757494e-05,
      "loss": 1.8314,
      "step": 203212
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.4143203496932983,
      "learning_rate": 2.038229116969452e-05,
      "loss": 1.859,
      "step": 203213
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.445610523223877,
      "learning_rate": 2.038080915861773e-05,
      "loss": 2.1573,
      "step": 203214
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.4024062156677246,
      "learning_rate": 2.037932719952745e-05,
      "loss": 2.0285,
      "step": 203215
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.6009232997894287,
      "learning_rate": 2.0377845292423956e-05,
      "loss": 1.9279,
      "step": 203216
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.4416484832763672,
      "learning_rate": 2.0376363437307443e-05,
      "loss": 2.2664,
      "step": 203217
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.5606106519699097,
      "learning_rate": 2.0374881634178275e-05,
      "loss": 2.124,
      "step": 203218
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.4592593908309937,
      "learning_rate": 2.0373399883036656e-05,
      "loss": 2.1758,
      "step": 203219
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.3556525707244873,
      "learning_rate": 2.037191818388295e-05,
      "loss": 1.9207,
      "step": 203220
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.4779269695281982,
      "learning_rate": 2.037043653671736e-05,
      "loss": 1.7498,
      "step": 203221
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.4278916120529175,
      "learning_rate": 2.0368954941540216e-05,
      "loss": 1.9381,
      "step": 203222
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.4082014560699463,
      "learning_rate": 2.0367473398351684e-05,
      "loss": 1.9798,
      "step": 203223
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.5296454429626465,
      "learning_rate": 2.03659919071522e-05,
      "loss": 1.9834,
      "step": 203224
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.5054469108581543,
      "learning_rate": 2.0364510467941897e-05,
      "loss": 1.8862,
      "step": 203225
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.4309475421905518,
      "learning_rate": 2.036302908072114e-05,
      "loss": 2.0129,
      "step": 203226
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.5872856378555298,
      "learning_rate": 2.0361547745490194e-05,
      "loss": 1.9556,
      "step": 203227
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.3126715421676636,
      "learning_rate": 2.0360066462249324e-05,
      "loss": 1.9965,
      "step": 203228
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.4524118900299072,
      "learning_rate": 2.0358585230998737e-05,
      "loss": 1.9042,
      "step": 203229
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.3548399209976196,
      "learning_rate": 2.035710405173879e-05,
      "loss": 1.9474,
      "step": 203230
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.52803373336792,
      "learning_rate": 2.0355622924469728e-05,
      "loss": 1.9484,
      "step": 203231
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.4732277393341064,
      "learning_rate": 2.0354141849191842e-05,
      "loss": 1.8153,
      "step": 203232
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.3871302604675293,
      "learning_rate": 2.035266082590543e-05,
      "loss": 1.8923,
      "step": 203233
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.4327996969223022,
      "learning_rate": 2.0351179854610733e-05,
      "loss": 2.0289,
      "step": 203234
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.340561032295227,
      "learning_rate": 2.0349698935307977e-05,
      "loss": 2.011,
      "step": 203235
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.4478038549423218,
      "learning_rate": 2.0348218067997534e-05,
      "loss": 2.1711,
      "step": 203236
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.6070693731307983,
      "learning_rate": 2.0346737252679602e-05,
      "loss": 2.0888,
      "step": 203237
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.3861868381500244,
      "learning_rate": 2.034525648935451e-05,
      "loss": 1.9032,
      "step": 203238
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.5191069841384888,
      "learning_rate": 2.0343775778022566e-05,
      "loss": 1.9735,
      "step": 203239
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.4205646514892578,
      "learning_rate": 2.034229511868396e-05,
      "loss": 1.9823,
      "step": 203240
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.4580094814300537,
      "learning_rate": 2.0340814511338932e-05,
      "loss": 2.0404,
      "step": 203241
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.5455513000488281,
      "learning_rate": 2.0339333955987914e-05,
      "loss": 1.9483,
      "step": 203242
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.4834798574447632,
      "learning_rate": 2.0337853452631004e-05,
      "loss": 2.1072,
      "step": 203243
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.4344311952590942,
      "learning_rate": 2.0336373001268635e-05,
      "loss": 2.0224,
      "step": 203244
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.4907336235046387,
      "learning_rate": 2.0334892601901008e-05,
      "loss": 1.9588,
      "step": 203245
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.4776787757873535,
      "learning_rate": 2.0333412254528424e-05,
      "loss": 1.9663,
      "step": 203246
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.4191527366638184,
      "learning_rate": 2.0331931959151082e-05,
      "loss": 2.1562,
      "step": 203247
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.398351788520813,
      "learning_rate": 2.0330451715769346e-05,
      "loss": 1.9179,
      "step": 203248
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.469205617904663,
      "learning_rate": 2.032897152438342e-05,
      "loss": 1.8015,
      "step": 203249
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.4860621690750122,
      "learning_rate": 2.0327491384993666e-05,
      "loss": 1.8856,
      "step": 203250
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.3699932098388672,
      "learning_rate": 2.0326011297600254e-05,
      "loss": 2.0332,
      "step": 203251
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.469406247138977,
      "learning_rate": 2.0324531262203612e-05,
      "loss": 2.1923,
      "step": 203252
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.8067259788513184,
      "learning_rate": 2.0323051278803847e-05,
      "loss": 1.8409,
      "step": 203253
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.6308828592300415,
      "learning_rate": 2.0321571347401323e-05,
      "loss": 2.1262,
      "step": 203254
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.4217267036437988,
      "learning_rate": 2.032009146799627e-05,
      "loss": 2.0884,
      "step": 203255
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.4866808652877808,
      "learning_rate": 2.031861164058902e-05,
      "loss": 2.0966,
      "step": 203256
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.483987808227539,
      "learning_rate": 2.0317131865179815e-05,
      "loss": 1.7619,
      "step": 203257
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.4626554250717163,
      "learning_rate": 2.031565214176898e-05,
      "loss": 2.1464,
      "step": 203258
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.4345747232437134,
      "learning_rate": 2.0314172470356684e-05,
      "loss": 2.0,
      "step": 203259
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.521260142326355,
      "learning_rate": 2.031269285094329e-05,
      "loss": 2.0566,
      "step": 203260
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.5102108716964722,
      "learning_rate": 2.0311213283529004e-05,
      "loss": 2.1855,
      "step": 203261
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.416746973991394,
      "learning_rate": 2.030973376811419e-05,
      "loss": 2.0318,
      "step": 203262
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.5543116331100464,
      "learning_rate": 2.0308254304699013e-05,
      "loss": 2.1211,
      "step": 203263
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.488898754119873,
      "learning_rate": 2.030677489328394e-05,
      "loss": 1.9752,
      "step": 203264
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.343869924545288,
      "learning_rate": 2.0305295533869003e-05,
      "loss": 1.9059,
      "step": 203265
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.4620625972747803,
      "learning_rate": 2.030381622645464e-05,
      "loss": 1.8739,
      "step": 203266
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.4295175075531006,
      "learning_rate": 2.0302336971041012e-05,
      "loss": 1.8758,
      "step": 203267
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.4911149740219116,
      "learning_rate": 2.0300857767628555e-05,
      "loss": 2.0292,
      "step": 203268
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.6992285251617432,
      "learning_rate": 2.0299378616217364e-05,
      "loss": 2.0663,
      "step": 203269
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.5070185661315918,
      "learning_rate": 2.029789951680788e-05,
      "loss": 1.9754,
      "step": 203270
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.4145846366882324,
      "learning_rate": 2.029642046940023e-05,
      "loss": 1.8821,
      "step": 203271
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.4111343622207642,
      "learning_rate": 2.0294941473994785e-05,
      "loss": 2.0047,
      "step": 203272
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.439736008644104,
      "learning_rate": 2.0293462530591776e-05,
      "loss": 1.872,
      "step": 203273
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.5646882057189941,
      "learning_rate": 2.02919836391915e-05,
      "loss": 2.0349,
      "step": 203274
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.4595173597335815,
      "learning_rate": 2.0290504799794193e-05,
      "loss": 1.7623,
      "step": 203275
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.4959386587142944,
      "learning_rate": 2.0289026012400257e-05,
      "loss": 2.0152,
      "step": 203276
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.5522421598434448,
      "learning_rate": 2.0287547277009752e-05,
      "loss": 1.8894,
      "step": 203277
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.4558475017547607,
      "learning_rate": 2.028606859362315e-05,
      "loss": 1.6412,
      "step": 203278
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.4236423969268799,
      "learning_rate": 2.0284589962240584e-05,
      "loss": 1.9976,
      "step": 203279
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.5248947143554688,
      "learning_rate": 2.028311138286245e-05,
      "loss": 2.103,
      "step": 203280
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.4774203300476074,
      "learning_rate": 2.0281632855488918e-05,
      "loss": 1.9737,
      "step": 203281
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.4985051155090332,
      "learning_rate": 2.028015438012035e-05,
      "loss": 1.9072,
      "step": 203282
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.4180505275726318,
      "learning_rate": 2.0278675956757017e-05,
      "loss": 2.0532,
      "step": 203283
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.5764403343200684,
      "learning_rate": 2.0277197585399118e-05,
      "loss": 2.0886,
      "step": 203284
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.5978769063949585,
      "learning_rate": 2.0275719266046918e-05,
      "loss": 1.9528,
      "step": 203285
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.4659122228622437,
      "learning_rate": 2.027424099870082e-05,
      "loss": 2.0045,
      "step": 203286
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.3839867115020752,
      "learning_rate": 2.027276278336095e-05,
      "loss": 1.7971,
      "step": 203287
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.6244219541549683,
      "learning_rate": 2.0271284620027717e-05,
      "loss": 1.932,
      "step": 203288
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.6450766324996948,
      "learning_rate": 2.0269806508701313e-05,
      "loss": 2.1246,
      "step": 203289
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.6640771627426147,
      "learning_rate": 2.0268328449382045e-05,
      "loss": 2.0137,
      "step": 203290
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.6946452856063843,
      "learning_rate": 2.0266850442070138e-05,
      "loss": 1.8768,
      "step": 203291
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.5017598867416382,
      "learning_rate": 2.026537248676593e-05,
      "loss": 1.8856,
      "step": 203292
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.442772388458252,
      "learning_rate": 2.026389458346962e-05,
      "loss": 1.9375,
      "step": 203293
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.3646847009658813,
      "learning_rate": 2.0262416732181573e-05,
      "loss": 2.1134,
      "step": 203294
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.4403026103973389,
      "learning_rate": 2.026093893290206e-05,
      "loss": 2.0897,
      "step": 203295
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.5309970378875732,
      "learning_rate": 2.0259461185631243e-05,
      "loss": 1.9153,
      "step": 203296
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.372496485710144,
      "learning_rate": 2.0257983490369523e-05,
      "loss": 2.0035,
      "step": 203297
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.5428122282028198,
      "learning_rate": 2.0256505847117134e-05,
      "loss": 1.8019,
      "step": 203298
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.8056684732437134,
      "learning_rate": 2.0255028255874306e-05,
      "loss": 1.9296,
      "step": 203299
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.4710948467254639,
      "learning_rate": 2.0253550716641375e-05,
      "loss": 2.0993,
      "step": 203300
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.412122368812561,
      "learning_rate": 2.0252073229418608e-05,
      "loss": 1.9754,
      "step": 203301
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.474478006362915,
      "learning_rate": 2.025059579420617e-05,
      "loss": 1.9409,
      "step": 203302
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.4181935787200928,
      "learning_rate": 2.0249118411004528e-05,
      "loss": 2.0442,
      "step": 203303
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.3555724620819092,
      "learning_rate": 2.024764107981378e-05,
      "loss": 2.0559,
      "step": 203304
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.4629942178726196,
      "learning_rate": 2.024616380063433e-05,
      "loss": 1.976,
      "step": 203305
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.5848262310028076,
      "learning_rate": 2.0244686573466406e-05,
      "loss": 1.9863,
      "step": 203306
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.4347596168518066,
      "learning_rate": 2.0243209398310244e-05,
      "loss": 1.9148,
      "step": 203307
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.4873899221420288,
      "learning_rate": 2.0241732275166145e-05,
      "loss": 2.0652,
      "step": 203308
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.3964707851409912,
      "learning_rate": 2.0240255204034404e-05,
      "loss": 1.9926,
      "step": 203309
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.3431650400161743,
      "learning_rate": 2.023877818491526e-05,
      "loss": 2.1191,
      "step": 203310
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.4738292694091797,
      "learning_rate": 2.0237301217809043e-05,
      "loss": 2.095,
      "step": 203311
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.4010751247406006,
      "learning_rate": 2.0235824302715985e-05,
      "loss": 1.7479,
      "step": 203312
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.414856195449829,
      "learning_rate": 2.0234347439636357e-05,
      "loss": 1.9219,
      "step": 203313
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.3932175636291504,
      "learning_rate": 2.023287062857042e-05,
      "loss": 2.0053,
      "step": 203314
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.641297698020935,
      "learning_rate": 2.023139386951851e-05,
      "loss": 1.8718,
      "step": 203315
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.5011919736862183,
      "learning_rate": 2.0229917162480825e-05,
      "loss": 1.9252,
      "step": 203316
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.2750779390335083,
      "learning_rate": 2.0228440507457734e-05,
      "loss": 2.0146,
      "step": 203317
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.396510362625122,
      "learning_rate": 2.02269639044494e-05,
      "loss": 1.9721,
      "step": 203318
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.4664753675460815,
      "learning_rate": 2.022548735345626e-05,
      "loss": 2.0256,
      "step": 203319
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.4201761484146118,
      "learning_rate": 2.022401085447838e-05,
      "loss": 1.8345,
      "step": 203320
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.4022293090820312,
      "learning_rate": 2.022253440751619e-05,
      "loss": 1.7206,
      "step": 203321
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.3969452381134033,
      "learning_rate": 2.0221058012569857e-05,
      "loss": 2.0233,
      "step": 203322
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.3791714906692505,
      "learning_rate": 2.0219581669639752e-05,
      "loss": 1.8077,
      "step": 203323
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.5581849813461304,
      "learning_rate": 2.021810537872607e-05,
      "loss": 2.1796,
      "step": 203324
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.4113249778747559,
      "learning_rate": 2.021662913982921e-05,
      "loss": 2.0614,
      "step": 203325
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.5530091524124146,
      "learning_rate": 2.0215152952949275e-05,
      "loss": 2.0238,
      "step": 203326
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.4065160751342773,
      "learning_rate": 2.0213676818086667e-05,
      "loss": 1.9896,
      "step": 203327
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.4331724643707275,
      "learning_rate": 2.021220073524158e-05,
      "loss": 1.9296,
      "step": 203328
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.3881579637527466,
      "learning_rate": 2.021072470441435e-05,
      "loss": 1.8096,
      "step": 203329
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.7237530946731567,
      "learning_rate": 2.020924872560521e-05,
      "loss": 1.8847,
      "step": 203330
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.8553746938705444,
      "learning_rate": 2.0207772798814526e-05,
      "loss": 2.2314,
      "step": 203331
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.4580378532409668,
      "learning_rate": 2.0206296924042397e-05,
      "loss": 1.9875,
      "step": 203332
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.380476951599121,
      "learning_rate": 2.0204821101289258e-05,
      "loss": 1.9078,
      "step": 203333
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.4430259466171265,
      "learning_rate": 2.0203345330555275e-05,
      "loss": 1.9709,
      "step": 203334
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.4498268365859985,
      "learning_rate": 2.0201869611840815e-05,
      "loss": 1.8162,
      "step": 203335
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.3514459133148193,
      "learning_rate": 2.0200393945146076e-05,
      "loss": 1.9767,
      "step": 203336
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.5562853813171387,
      "learning_rate": 2.019891833047146e-05,
      "loss": 1.8727,
      "step": 203337
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.4493745565414429,
      "learning_rate": 2.0197442767817063e-05,
      "loss": 2.0862,
      "step": 203338
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.3878509998321533,
      "learning_rate": 2.0195967257183253e-05,
      "loss": 2.1517,
      "step": 203339
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.4241173267364502,
      "learning_rate": 2.0194491798570265e-05,
      "loss": 1.7806,
      "step": 203340
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.3810762166976929,
      "learning_rate": 2.0193016391978467e-05,
      "loss": 2.0486,
      "step": 203341
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.3545644283294678,
      "learning_rate": 2.0191541037407987e-05,
      "loss": 1.9142,
      "step": 203342
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.4235694408416748,
      "learning_rate": 2.0190065734859296e-05,
      "loss": 2.0236,
      "step": 203343
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.4145740270614624,
      "learning_rate": 2.0188590484332458e-05,
      "loss": 2.1632,
      "step": 203344
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.4295402765274048,
      "learning_rate": 2.0187115285827903e-05,
      "loss": 1.9851,
      "step": 203345
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.5567431449890137,
      "learning_rate": 2.018564013934577e-05,
      "loss": 2.0201,
      "step": 203346
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.3527247905731201,
      "learning_rate": 2.018416504488649e-05,
      "loss": 1.896,
      "step": 203347
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.4621914625167847,
      "learning_rate": 2.0182690002450163e-05,
      "loss": 2.1003,
      "step": 203348
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.4403094053268433,
      "learning_rate": 2.0181215012037287e-05,
      "loss": 2.2075,
      "step": 203349
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.4143187999725342,
      "learning_rate": 2.0179740073647897e-05,
      "loss": 1.8739,
      "step": 203350
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.5224099159240723,
      "learning_rate": 2.0178265187282427e-05,
      "loss": 2.0572,
      "step": 203351
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.4265395402908325,
      "learning_rate": 2.0176790352941074e-05,
      "loss": 2.1118,
      "step": 203352
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.4361047744750977,
      "learning_rate": 2.017531557062414e-05,
      "loss": 2.0606,
      "step": 203353
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.4853028059005737,
      "learning_rate": 2.017384084033189e-05,
      "loss": 1.9378,
      "step": 203354
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.5028295516967773,
      "learning_rate": 2.0172366162064657e-05,
      "loss": 2.1229,
      "step": 203355
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.4335460662841797,
      "learning_rate": 2.017089153582261e-05,
      "loss": 2.0163,
      "step": 203356
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.3515313863754272,
      "learning_rate": 2.016941696160611e-05,
      "loss": 1.9066,
      "step": 203357
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.4619299173355103,
      "learning_rate": 2.016794243941533e-05,
      "loss": 2.0426,
      "step": 203358
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.541279673576355,
      "learning_rate": 2.0166467969250698e-05,
      "loss": 2.0213,
      "step": 203359
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.5018800497055054,
      "learning_rate": 2.016499355111232e-05,
      "loss": 1.9502,
      "step": 203360
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.6700831651687622,
      "learning_rate": 2.0163519185000655e-05,
      "loss": 1.8536,
      "step": 203361
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.493737816810608,
      "learning_rate": 2.0162044870915805e-05,
      "loss": 2.0067,
      "step": 203362
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.5852651596069336,
      "learning_rate": 2.016057060885814e-05,
      "loss": 1.9548,
      "step": 203363
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.5871708393096924,
      "learning_rate": 2.0159096398827856e-05,
      "loss": 1.9309,
      "step": 203364
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.3209484815597534,
      "learning_rate": 2.0157622240825323e-05,
      "loss": 2.037,
      "step": 203365
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.34442138671875,
      "learning_rate": 2.0156148134850735e-05,
      "loss": 2.0437,
      "step": 203366
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.3618251085281372,
      "learning_rate": 2.0154674080904465e-05,
      "loss": 2.0493,
      "step": 203367
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.402856707572937,
      "learning_rate": 2.0153200078986675e-05,
      "loss": 1.8955,
      "step": 203368
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.662313461303711,
      "learning_rate": 2.01517261290977e-05,
      "loss": 1.9501,
      "step": 203369
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.3819173574447632,
      "learning_rate": 2.0150252231237773e-05,
      "loss": 2.1025,
      "step": 203370
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.254040241241455,
      "learning_rate": 2.0148778385407227e-05,
      "loss": 1.8675,
      "step": 203371
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.3352221250534058,
      "learning_rate": 2.0147304591606293e-05,
      "loss": 2.0084,
      "step": 203372
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.3421446084976196,
      "learning_rate": 2.0145830849835276e-05,
      "loss": 1.9244,
      "step": 203373
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.4061212539672852,
      "learning_rate": 2.0144357160094437e-05,
      "loss": 1.9638,
      "step": 203374
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.4578887224197388,
      "learning_rate": 2.0142883522384046e-05,
      "loss": 2.1145,
      "step": 203375
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.5214197635650635,
      "learning_rate": 2.0141409936704332e-05,
      "loss": 2.1148,
      "step": 203376
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.4468289613723755,
      "learning_rate": 2.0139936403055666e-05,
      "loss": 2.0676,
      "step": 203377
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.3750386238098145,
      "learning_rate": 2.0138462921438215e-05,
      "loss": 1.9618,
      "step": 203378
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.4181307554244995,
      "learning_rate": 2.013698949185234e-05,
      "loss": 1.8623,
      "step": 203379
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.4023643732070923,
      "learning_rate": 2.0135516114298278e-05,
      "loss": 2.0899,
      "step": 203380
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.4068089723587036,
      "learning_rate": 2.0134042788776293e-05,
      "loss": 1.8143,
      "step": 203381
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.538151502609253,
      "learning_rate": 2.013256951528669e-05,
      "loss": 2.0449,
      "step": 203382
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.4865013360977173,
      "learning_rate": 2.013109629382973e-05,
      "loss": 1.9145,
      "step": 203383
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.4193572998046875,
      "learning_rate": 2.0129623124405647e-05,
      "loss": 1.8668,
      "step": 203384
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.3774516582489014,
      "learning_rate": 2.0128150007014776e-05,
      "loss": 2.1549,
      "step": 203385
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.4635283946990967,
      "learning_rate": 2.0126676941657382e-05,
      "loss": 2.0349,
      "step": 203386
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.5206608772277832,
      "learning_rate": 2.01252039283337e-05,
      "loss": 2.0827,
      "step": 203387
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.2957595586776733,
      "learning_rate": 2.0123730967044028e-05,
      "loss": 1.8436,
      "step": 203388
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.3171201944351196,
      "learning_rate": 2.0122258057788664e-05,
      "loss": 1.9855,
      "step": 203389
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.4726171493530273,
      "learning_rate": 2.012078520056778e-05,
      "loss": 1.7674,
      "step": 203390
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.4366480112075806,
      "learning_rate": 2.0119312395381805e-05,
      "loss": 2.1409,
      "step": 203391
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.4470409154891968,
      "learning_rate": 2.0117839642230903e-05,
      "loss": 1.8747,
      "step": 203392
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.4477332830429077,
      "learning_rate": 2.0116366941115346e-05,
      "loss": 2.0789,
      "step": 203393
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.4681978225708008,
      "learning_rate": 2.01148942920355e-05,
      "loss": 1.9981,
      "step": 203394
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.4811575412750244,
      "learning_rate": 2.0113421694991526e-05,
      "loss": 1.9341,
      "step": 203395
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.5240607261657715,
      "learning_rate": 2.0111949149983765e-05,
      "loss": 2.0231,
      "step": 203396
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.4088467359542847,
      "learning_rate": 2.0110476657012508e-05,
      "loss": 1.9019,
      "step": 203397
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.4010146856307983,
      "learning_rate": 2.0109004216077994e-05,
      "loss": 1.7786,
      "step": 203398
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.538942575454712,
      "learning_rate": 2.0107531827180455e-05,
      "loss": 1.8452,
      "step": 203399
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.3710370063781738,
      "learning_rate": 2.010605949032026e-05,
      "loss": 1.9144,
      "step": 203400
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.417819619178772,
      "learning_rate": 2.0104587205497567e-05,
      "loss": 2.0933,
      "step": 203401
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.4712369441986084,
      "learning_rate": 2.0103114972712754e-05,
      "loss": 2.0393,
      "step": 203402
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.3921165466308594,
      "learning_rate": 2.010164279196601e-05,
      "loss": 1.8959,
      "step": 203403
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.5708140134811401,
      "learning_rate": 2.0100170663257774e-05,
      "loss": 2.0189,
      "step": 203404
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.478339433670044,
      "learning_rate": 2.0098698586588113e-05,
      "loss": 1.9798,
      "step": 203405
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.561336874961853,
      "learning_rate": 2.009722656195739e-05,
      "loss": 2.0219,
      "step": 203406
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.5980325937271118,
      "learning_rate": 2.0095754589365876e-05,
      "loss": 1.9824,
      "step": 203407
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.4043809175491333,
      "learning_rate": 2.0094282668813867e-05,
      "loss": 2.075,
      "step": 203408
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.33584725856781,
      "learning_rate": 2.0092810800301563e-05,
      "loss": 2.044,
      "step": 203409
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.3777332305908203,
      "learning_rate": 2.0091338983829398e-05,
      "loss": 1.9158,
      "step": 203410
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.385536551475525,
      "learning_rate": 2.0089867219397436e-05,
      "loss": 1.9788,
      "step": 203411
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.5103397369384766,
      "learning_rate": 2.0088395507006084e-05,
      "loss": 2.0666,
      "step": 203412
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.3813246488571167,
      "learning_rate": 2.0086923846655535e-05,
      "loss": 1.7967,
      "step": 203413
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.537591576576233,
      "learning_rate": 2.008545223834619e-05,
      "loss": 1.9932,
      "step": 203414
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.406808614730835,
      "learning_rate": 2.008398068207818e-05,
      "loss": 1.9522,
      "step": 203415
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.4016989469528198,
      "learning_rate": 2.0082509177851912e-05,
      "loss": 1.9767,
      "step": 203416
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.343404769897461,
      "learning_rate": 2.0081037725667516e-05,
      "loss": 1.9255,
      "step": 203417
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.5047118663787842,
      "learning_rate": 2.0079566325525387e-05,
      "loss": 1.9527,
      "step": 203418
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.5049787759780884,
      "learning_rate": 2.0078094977425728e-05,
      "loss": 1.8831,
      "step": 203419
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.4897481203079224,
      "learning_rate": 2.007662368136884e-05,
      "loss": 1.9791,
      "step": 203420
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.5013506412506104,
      "learning_rate": 2.0075152437354958e-05,
      "loss": 1.8864,
      "step": 203421
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.4630125761032104,
      "learning_rate": 2.0073681245384508e-05,
      "loss": 2.1202,
      "step": 203422
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.438895583152771,
      "learning_rate": 2.007221010545753e-05,
      "loss": 2.001,
      "step": 203423
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.4865903854370117,
      "learning_rate": 2.0070739017574454e-05,
      "loss": 1.9388,
      "step": 203424
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.3585712909698486,
      "learning_rate": 2.0069267981735483e-05,
      "loss": 1.7833,
      "step": 203425
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.6624226570129395,
      "learning_rate": 2.0067796997940944e-05,
      "loss": 1.9066,
      "step": 203426
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.4784284830093384,
      "learning_rate": 2.006632606619104e-05,
      "loss": 2.0483,
      "step": 203427
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.3862684965133667,
      "learning_rate": 2.0064855186486208e-05,
      "loss": 2.1733,
      "step": 203428
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.4231735467910767,
      "learning_rate": 2.006338435882647e-05,
      "loss": 2.1072,
      "step": 203429
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.3866662979125977,
      "learning_rate": 2.006191358321231e-05,
      "loss": 1.973,
      "step": 203430
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.548237681388855,
      "learning_rate": 2.0060442859643875e-05,
      "loss": 1.769,
      "step": 203431
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.6611781120300293,
      "learning_rate": 2.0058972188121514e-05,
      "loss": 2.0692,
      "step": 203432
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.5473625659942627,
      "learning_rate": 2.005750156864545e-05,
      "loss": 2.2076,
      "step": 203433
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.4384340047836304,
      "learning_rate": 2.005603100121609e-05,
      "loss": 1.9542,
      "step": 203434
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.4079176187515259,
      "learning_rate": 2.005456048583346e-05,
      "loss": 1.9434,
      "step": 203435
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.4945967197418213,
      "learning_rate": 2.0053090022498065e-05,
      "loss": 2.0524,
      "step": 203436
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.558351993560791,
      "learning_rate": 2.005161961121e-05,
      "loss": 1.9502,
      "step": 203437
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.389642357826233,
      "learning_rate": 2.0050149251969703e-05,
      "loss": 2.009,
      "step": 203438
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.4408113956451416,
      "learning_rate": 2.0048678944777273e-05,
      "loss": 1.6527,
      "step": 203439
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.4541916847229004,
      "learning_rate": 2.004720868963321e-05,
      "loss": 2.0237,
      "step": 203440
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.527891755104065,
      "learning_rate": 2.0045738486537576e-05,
      "loss": 1.9952,
      "step": 203441
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.5811495780944824,
      "learning_rate": 2.0044268335490743e-05,
      "loss": 1.9546,
      "step": 203442
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.5213457345962524,
      "learning_rate": 2.004279823649291e-05,
      "loss": 1.9911,
      "step": 203443
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.4312620162963867,
      "learning_rate": 2.004132818954447e-05,
      "loss": 1.8688,
      "step": 203444
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.3768835067749023,
      "learning_rate": 2.0039858194645596e-05,
      "loss": 2.0522,
      "step": 203445
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.4766587018966675,
      "learning_rate": 2.0038388251796622e-05,
      "loss": 2.273,
      "step": 203446
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.388339877128601,
      "learning_rate": 2.0036918360997813e-05,
      "loss": 2.1348,
      "step": 203447
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.504223346710205,
      "learning_rate": 2.0035448522249398e-05,
      "loss": 1.8512,
      "step": 203448
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.3595887422561646,
      "learning_rate": 2.003397873555165e-05,
      "loss": 1.9812,
      "step": 203449
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.3810675144195557,
      "learning_rate": 2.003250900090493e-05,
      "loss": 1.7323,
      "step": 203450
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.4664556980133057,
      "learning_rate": 2.0031039318309373e-05,
      "loss": 2.0871,
      "step": 203451
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.394592523574829,
      "learning_rate": 2.002956968776538e-05,
      "loss": 1.9889,
      "step": 203452
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.4978992938995361,
      "learning_rate": 2.0028100109273216e-05,
      "loss": 2.1424,
      "step": 203453
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.4800719022750854,
      "learning_rate": 2.0026630582833082e-05,
      "loss": 2.213,
      "step": 203454
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.5810290575027466,
      "learning_rate": 2.002516110844524e-05,
      "loss": 2.1085,
      "step": 203455
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.49905526638031,
      "learning_rate": 2.002369168611003e-05,
      "loss": 1.9627,
      "step": 203456
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.495116114616394,
      "learning_rate": 2.0022222315827686e-05,
      "loss": 2.0429,
      "step": 203457
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.4264332056045532,
      "learning_rate": 2.0020752997598534e-05,
      "loss": 2.1468,
      "step": 203458
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.4587278366088867,
      "learning_rate": 2.001928373142281e-05,
      "loss": 2.3009,
      "step": 203459
    },
    {
      "epoch": 2.65,
      "grad_norm": 2.0229201316833496,
      "learning_rate": 2.001781451730078e-05,
      "loss": 2.1589,
      "step": 203460
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.4473298788070679,
      "learning_rate": 2.0016345355232675e-05,
      "loss": 1.989,
      "step": 203461
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.410225510597229,
      "learning_rate": 2.001487624521887e-05,
      "loss": 1.8375,
      "step": 203462
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.4454100131988525,
      "learning_rate": 2.0013407187259557e-05,
      "loss": 2.1301,
      "step": 203463
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.4954533576965332,
      "learning_rate": 2.0011938181355037e-05,
      "loss": 1.8618,
      "step": 203464
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.4933733940124512,
      "learning_rate": 2.001046922750561e-05,
      "loss": 2.0734,
      "step": 203465
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.342687964439392,
      "learning_rate": 2.0009000325711478e-05,
      "loss": 2.1237,
      "step": 203466
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.4732823371887207,
      "learning_rate": 2.0007531475973003e-05,
      "loss": 2.1144,
      "step": 203467
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.3271656036376953,
      "learning_rate": 2.0006062678290426e-05,
      "loss": 1.9235,
      "step": 203468
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.6074589490890503,
      "learning_rate": 2.000459393266394e-05,
      "loss": 1.9789,
      "step": 203469
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.3337923288345337,
      "learning_rate": 2.000312523909394e-05,
      "loss": 1.9681,
      "step": 203470
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.4518147706985474,
      "learning_rate": 2.000165659758064e-05,
      "loss": 1.7466,
      "step": 203471
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.25572669506073,
      "learning_rate": 2.000018800812426e-05,
      "loss": 1.9868,
      "step": 203472
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.4711600542068481,
      "learning_rate": 1.9998719470725212e-05,
      "loss": 1.8931,
      "step": 203473
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.3791359663009644,
      "learning_rate": 1.9997250985383652e-05,
      "loss": 2.0237,
      "step": 203474
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.5676226615905762,
      "learning_rate": 1.999578255209985e-05,
      "loss": 1.8936,
      "step": 203475
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.5727102756500244,
      "learning_rate": 1.999431417087417e-05,
      "loss": 2.0624,
      "step": 203476
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.6318320035934448,
      "learning_rate": 1.999284584170685e-05,
      "loss": 1.9468,
      "step": 203477
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.5261714458465576,
      "learning_rate": 1.9991377564598054e-05,
      "loss": 2.1683,
      "step": 203478
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.3837226629257202,
      "learning_rate": 1.998990933954825e-05,
      "loss": 1.9524,
      "step": 203479
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.5766857862472534,
      "learning_rate": 1.99884411665575e-05,
      "loss": 1.8853,
      "step": 203480
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.4616942405700684,
      "learning_rate": 1.9986973045626276e-05,
      "loss": 2.097,
      "step": 203481
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.437914252281189,
      "learning_rate": 1.998550497675474e-05,
      "loss": 1.9802,
      "step": 203482
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.5545364618301392,
      "learning_rate": 1.9984036959943194e-05,
      "loss": 1.9434,
      "step": 203483
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.3908143043518066,
      "learning_rate": 1.998256899519184e-05,
      "loss": 1.8157,
      "step": 203484
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.7024421691894531,
      "learning_rate": 1.9981101082501074e-05,
      "loss": 1.879,
      "step": 203485
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.4080772399902344,
      "learning_rate": 1.997963322187106e-05,
      "loss": 1.959,
      "step": 203486
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.332425594329834,
      "learning_rate": 1.9978165413302173e-05,
      "loss": 1.9019,
      "step": 203487
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.4390422105789185,
      "learning_rate": 1.9976697656794605e-05,
      "loss": 2.0129,
      "step": 203488
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.6013360023498535,
      "learning_rate": 1.997522995234866e-05,
      "loss": 2.0406,
      "step": 203489
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.437686800956726,
      "learning_rate": 1.997376229996457e-05,
      "loss": 1.9494,
      "step": 203490
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.4981114864349365,
      "learning_rate": 1.9972294699642698e-05,
      "loss": 1.9279,
      "step": 203491
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.5445700883865356,
      "learning_rate": 1.9970827151383184e-05,
      "loss": 2.0,
      "step": 203492
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.3963333368301392,
      "learning_rate": 1.9969359655186456e-05,
      "loss": 1.9345,
      "step": 203493
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.5195177793502808,
      "learning_rate": 1.996789221105265e-05,
      "loss": 1.9993,
      "step": 203494
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.5620622634887695,
      "learning_rate": 1.9966424818982196e-05,
      "loss": 2.0331,
      "step": 203495
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.3731591701507568,
      "learning_rate": 1.9964957478975195e-05,
      "loss": 1.9145,
      "step": 203496
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.3273335695266724,
      "learning_rate": 1.9963490191032017e-05,
      "loss": 2.0041,
      "step": 203497
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.4226124286651611,
      "learning_rate": 1.9962022955152857e-05,
      "loss": 2.0439,
      "step": 203498
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.4464397430419922,
      "learning_rate": 1.9960555771338116e-05,
      "loss": 2.1261,
      "step": 203499
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.4944947957992554,
      "learning_rate": 1.9959088639587928e-05,
      "loss": 2.164,
      "step": 203500
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.484142541885376,
      "learning_rate": 1.9957621559902725e-05,
      "loss": 1.9102,
      "step": 203501
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.273669958114624,
      "learning_rate": 1.995615453228261e-05,
      "loss": 1.7744,
      "step": 203502
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.4627147912979126,
      "learning_rate": 1.9954687556727978e-05,
      "loss": 2.0297,
      "step": 203503
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.5739606618881226,
      "learning_rate": 1.9953220633238998e-05,
      "loss": 1.887,
      "step": 203504
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.4462376832962036,
      "learning_rate": 1.9951753761816035e-05,
      "loss": 1.8935,
      "step": 203505
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.4410903453826904,
      "learning_rate": 1.9950286942459292e-05,
      "loss": 1.8354,
      "step": 203506
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.3955755233764648,
      "learning_rate": 1.9948820175169167e-05,
      "loss": 2.2,
      "step": 203507
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.4580280780792236,
      "learning_rate": 1.9947353459945758e-05,
      "loss": 2.0188,
      "step": 203508
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.3108117580413818,
      "learning_rate": 1.9945886796789467e-05,
      "loss": 1.8738,
      "step": 203509
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.371132254600525,
      "learning_rate": 1.9944420185700494e-05,
      "loss": 1.8904,
      "step": 203510
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.3788772821426392,
      "learning_rate": 1.9942953626679135e-05,
      "loss": 1.8924,
      "step": 203511
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.5618308782577515,
      "learning_rate": 1.9941487119725664e-05,
      "loss": 2.0375,
      "step": 203512
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.6259161233901978,
      "learning_rate": 1.9940020664840406e-05,
      "loss": 2.0233,
      "step": 203513
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.5270781517028809,
      "learning_rate": 1.9938554262023564e-05,
      "loss": 1.9995,
      "step": 203514
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.4000747203826904,
      "learning_rate": 1.993708791127544e-05,
      "loss": 1.8135,
      "step": 203515
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.531230092048645,
      "learning_rate": 1.9935621612596264e-05,
      "loss": 2.1651,
      "step": 203516
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.5107982158660889,
      "learning_rate": 1.9934155365986372e-05,
      "loss": 2.0387,
      "step": 203517
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.3617560863494873,
      "learning_rate": 1.993268917144596e-05,
      "loss": 1.9845,
      "step": 203518
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.441695213317871,
      "learning_rate": 1.9931223028975397e-05,
      "loss": 2.0892,
      "step": 203519
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.4425591230392456,
      "learning_rate": 1.9929756938574915e-05,
      "loss": 1.7695,
      "step": 203520
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.541120171546936,
      "learning_rate": 1.992829090024475e-05,
      "loss": 2.092,
      "step": 203521
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.4846996068954468,
      "learning_rate": 1.9926824913985195e-05,
      "loss": 1.9627,
      "step": 203522
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.4938926696777344,
      "learning_rate": 1.992535897979656e-05,
      "loss": 1.9341,
      "step": 203523
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.5340452194213867,
      "learning_rate": 1.9923893097679035e-05,
      "loss": 2.0286,
      "step": 203524
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.3100273609161377,
      "learning_rate": 1.992242726763299e-05,
      "loss": 1.8846,
      "step": 203525
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.4739867448806763,
      "learning_rate": 1.992096148965866e-05,
      "loss": 2.1158,
      "step": 203526
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.4723238945007324,
      "learning_rate": 1.991949576375631e-05,
      "loss": 1.6942,
      "step": 203527
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.455919861793518,
      "learning_rate": 1.991803008992614e-05,
      "loss": 1.9149,
      "step": 203528
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.3576654195785522,
      "learning_rate": 1.9916564468168584e-05,
      "loss": 1.9551,
      "step": 203529
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.4051893949508667,
      "learning_rate": 1.991509889848377e-05,
      "loss": 1.8121,
      "step": 203530
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.414271593093872,
      "learning_rate": 1.991363338087204e-05,
      "loss": 2.0454,
      "step": 203531
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.3526177406311035,
      "learning_rate": 1.9912167915333688e-05,
      "loss": 1.9719,
      "step": 203532
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.4833917617797852,
      "learning_rate": 1.9910702501868947e-05,
      "loss": 1.8064,
      "step": 203533
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.4959818124771118,
      "learning_rate": 1.990923714047802e-05,
      "loss": 1.806,
      "step": 203534
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.6008890867233276,
      "learning_rate": 1.9907771831161335e-05,
      "loss": 1.928,
      "step": 203535
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.3759667873382568,
      "learning_rate": 1.9906306573919028e-05,
      "loss": 2.0359,
      "step": 203536
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.421173095703125,
      "learning_rate": 1.9904841368751435e-05,
      "loss": 1.9808,
      "step": 203537
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.2677302360534668,
      "learning_rate": 1.990337621565885e-05,
      "loss": 1.9668,
      "step": 203538
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.5106687545776367,
      "learning_rate": 1.990191111464151e-05,
      "loss": 1.8303,
      "step": 203539
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.5154094696044922,
      "learning_rate": 1.990044606569965e-05,
      "loss": 2.1433,
      "step": 203540
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.3451932668685913,
      "learning_rate": 1.989898106883363e-05,
      "loss": 1.948,
      "step": 203541
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.6010677814483643,
      "learning_rate": 1.9897516124043656e-05,
      "loss": 1.9697,
      "step": 203542
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.348839521408081,
      "learning_rate": 1.9896051231330024e-05,
      "loss": 2.0177,
      "step": 203543
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.7195706367492676,
      "learning_rate": 1.9894586390693035e-05,
      "loss": 1.8621,
      "step": 203544
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.5199737548828125,
      "learning_rate": 1.9893121602132922e-05,
      "loss": 1.9667,
      "step": 203545
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.405618667602539,
      "learning_rate": 1.9891656865649885e-05,
      "loss": 2.2227,
      "step": 203546
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.406765103340149,
      "learning_rate": 1.989019218124436e-05,
      "loss": 1.9033,
      "step": 203547
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.5824103355407715,
      "learning_rate": 1.988872754891647e-05,
      "loss": 1.9906,
      "step": 203548
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.3736366033554077,
      "learning_rate": 1.988726296866663e-05,
      "loss": 2.0465,
      "step": 203549
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.3994097709655762,
      "learning_rate": 1.9885798440494995e-05,
      "loss": 1.9352,
      "step": 203550
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.4648897647857666,
      "learning_rate": 1.98843339644019e-05,
      "loss": 1.8435,
      "step": 203551
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.7232937812805176,
      "learning_rate": 1.9882869540387546e-05,
      "loss": 2.1355,
      "step": 203552
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.4380053281784058,
      "learning_rate": 1.98814051684523e-05,
      "loss": 1.9738,
      "step": 203553
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.5029484033584595,
      "learning_rate": 1.987994084859633e-05,
      "loss": 1.9339,
      "step": 203554
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.6010444164276123,
      "learning_rate": 1.9878476580820037e-05,
      "loss": 2.1703,
      "step": 203555
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.491105556488037,
      "learning_rate": 1.9877012365123614e-05,
      "loss": 1.9154,
      "step": 203556
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.500731110572815,
      "learning_rate": 1.9875548201507296e-05,
      "loss": 2.0531,
      "step": 203557
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.6661639213562012,
      "learning_rate": 1.9874084089971453e-05,
      "loss": 1.7929,
      "step": 203558
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.3221664428710938,
      "learning_rate": 1.9872620030516285e-05,
      "loss": 2.2599,
      "step": 203559
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.3946698904037476,
      "learning_rate": 1.9871156023142055e-05,
      "loss": 2.1675,
      "step": 203560
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.3620930910110474,
      "learning_rate": 1.9869692067849096e-05,
      "loss": 1.8879,
      "step": 203561
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.4734463691711426,
      "learning_rate": 1.986822816463768e-05,
      "loss": 1.9638,
      "step": 203562
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.3821789026260376,
      "learning_rate": 1.9866764313507967e-05,
      "loss": 2.0378,
      "step": 203563
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.3782638311386108,
      "learning_rate": 1.986530051446036e-05,
      "loss": 2.0943,
      "step": 203564
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.3945618867874146,
      "learning_rate": 1.9863836767495056e-05,
      "loss": 2.1362,
      "step": 203565
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.3972485065460205,
      "learning_rate": 1.986237307261239e-05,
      "loss": 2.1235,
      "step": 203566
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.6407830715179443,
      "learning_rate": 1.9860909429812598e-05,
      "loss": 2.3146,
      "step": 203567
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.5147545337677002,
      "learning_rate": 1.9859445839095945e-05,
      "loss": 2.0321,
      "step": 203568
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.376783847808838,
      "learning_rate": 1.985798230046266e-05,
      "loss": 1.9357,
      "step": 203569
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.4676944017410278,
      "learning_rate": 1.9856518813913113e-05,
      "loss": 2.0438,
      "step": 203570
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.4229668378829956,
      "learning_rate": 1.9855055379447503e-05,
      "loss": 2.0989,
      "step": 203571
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.5382674932479858,
      "learning_rate": 1.9853591997066166e-05,
      "loss": 2.0107,
      "step": 203572
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.433259129524231,
      "learning_rate": 1.9852128666769295e-05,
      "loss": 2.048,
      "step": 203573
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.3407624959945679,
      "learning_rate": 1.985066538855723e-05,
      "loss": 2.2169,
      "step": 203574
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.4153298139572144,
      "learning_rate": 1.9849202162430167e-05,
      "loss": 1.9902,
      "step": 203575
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.4175454378128052,
      "learning_rate": 1.9847738988388472e-05,
      "loss": 1.659,
      "step": 203576
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.3834147453308105,
      "learning_rate": 1.9846275866432348e-05,
      "loss": 1.835,
      "step": 203577
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.361783504486084,
      "learning_rate": 1.9844812796562092e-05,
      "loss": 2.0744,
      "step": 203578
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.6309089660644531,
      "learning_rate": 1.9843349778777973e-05,
      "loss": 1.8906,
      "step": 203579
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.388617753982544,
      "learning_rate": 1.9841886813080288e-05,
      "loss": 1.9142,
      "step": 203580
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.5690791606903076,
      "learning_rate": 1.984042389946927e-05,
      "loss": 2.1559,
      "step": 203581
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.6372466087341309,
      "learning_rate": 1.9838961037945224e-05,
      "loss": 1.9798,
      "step": 203582
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.487468957901001,
      "learning_rate": 1.9837498228508343e-05,
      "loss": 1.9675,
      "step": 203583
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.4043842554092407,
      "learning_rate": 1.9836035471159028e-05,
      "loss": 2.2602,
      "step": 203584
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.3935250043869019,
      "learning_rate": 1.9834572765897418e-05,
      "loss": 2.2769,
      "step": 203585
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.4295793771743774,
      "learning_rate": 1.9833110112723906e-05,
      "loss": 2.1146,
      "step": 203586
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.4124226570129395,
      "learning_rate": 1.9831647511638692e-05,
      "loss": 2.1643,
      "step": 203587
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.552882432937622,
      "learning_rate": 1.983018496264208e-05,
      "loss": 2.1283,
      "step": 203588
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.4639743566513062,
      "learning_rate": 1.9828722465734272e-05,
      "loss": 1.8082,
      "step": 203589
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.4393267631530762,
      "learning_rate": 1.9827260020915625e-05,
      "loss": 1.9796,
      "step": 203590
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.6362131834030151,
      "learning_rate": 1.9825797628186347e-05,
      "loss": 1.8353,
      "step": 203591
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.4178717136383057,
      "learning_rate": 1.98243352875468e-05,
      "loss": 2.1727,
      "step": 203592
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.487540364265442,
      "learning_rate": 1.9822872998997187e-05,
      "loss": 2.0203,
      "step": 203593
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.5364052057266235,
      "learning_rate": 1.982141076253777e-05,
      "loss": 2.0854,
      "step": 203594
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.4255216121673584,
      "learning_rate": 1.981994857816882e-05,
      "loss": 1.6342,
      "step": 203595
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.5273635387420654,
      "learning_rate": 1.9818486445890668e-05,
      "loss": 1.9323,
      "step": 203596
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.4987812042236328,
      "learning_rate": 1.9817024365703516e-05,
      "loss": 2.1657,
      "step": 203597
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.3956470489501953,
      "learning_rate": 1.981556233760769e-05,
      "loss": 1.8232,
      "step": 203598
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.3929359912872314,
      "learning_rate": 1.9814100361603468e-05,
      "loss": 1.9393,
      "step": 203599
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.3805240392684937,
      "learning_rate": 1.9812638437691077e-05,
      "loss": 1.922,
      "step": 203600
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.3851076364517212,
      "learning_rate": 1.9811176565870747e-05,
      "loss": 1.954,
      "step": 203601
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.4562757015228271,
      "learning_rate": 1.980971474614288e-05,
      "loss": 2.087,
      "step": 203602
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.4206892251968384,
      "learning_rate": 1.980825297850761e-05,
      "loss": 2.1384,
      "step": 203603
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.4248559474945068,
      "learning_rate": 1.980679126296534e-05,
      "loss": 2.0197,
      "step": 203604
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.5741347074508667,
      "learning_rate": 1.9805329599516263e-05,
      "loss": 1.8913,
      "step": 203605
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.513511061668396,
      "learning_rate": 1.9803867988160683e-05,
      "loss": 1.9351,
      "step": 203606
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.4383190870285034,
      "learning_rate": 1.9802406428898763e-05,
      "loss": 1.9938,
      "step": 203607
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.3569495677947998,
      "learning_rate": 1.9800944921730942e-05,
      "loss": 1.756,
      "step": 203608
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.4780689477920532,
      "learning_rate": 1.9799483466657386e-05,
      "loss": 1.9103,
      "step": 203609
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.500312328338623,
      "learning_rate": 1.9798022063678422e-05,
      "loss": 1.8405,
      "step": 203610
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.4404815435409546,
      "learning_rate": 1.9796560712794285e-05,
      "loss": 2.0004,
      "step": 203611
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.44222092628479,
      "learning_rate": 1.9795099414005276e-05,
      "loss": 1.953,
      "step": 203612
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.413299798965454,
      "learning_rate": 1.9793638167311566e-05,
      "loss": 1.9913,
      "step": 203613
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.4668707847595215,
      "learning_rate": 1.979217697271358e-05,
      "loss": 2.0239,
      "step": 203614
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.4148391485214233,
      "learning_rate": 1.979071583021149e-05,
      "loss": 2.1236,
      "step": 203615
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.775895595550537,
      "learning_rate": 1.9789254739805628e-05,
      "loss": 1.9662,
      "step": 203616
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.5819580554962158,
      "learning_rate": 1.9787793701496225e-05,
      "loss": 2.0784,
      "step": 203617
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.3853120803833008,
      "learning_rate": 1.9786332715283547e-05,
      "loss": 1.8648,
      "step": 203618
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.502400279045105,
      "learning_rate": 1.9784871781167867e-05,
      "loss": 2.0218,
      "step": 203619
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.4515151977539062,
      "learning_rate": 1.9783410899149477e-05,
      "loss": 1.7239,
      "step": 203620
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.3630648851394653,
      "learning_rate": 1.9781950069228614e-05,
      "loss": 1.9124,
      "step": 203621
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.2727885246276855,
      "learning_rate": 1.9780489291405644e-05,
      "loss": 1.708,
      "step": 203622
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.5204511880874634,
      "learning_rate": 1.9779028565680733e-05,
      "loss": 2.1911,
      "step": 203623
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.407974123954773,
      "learning_rate": 1.9777567892054214e-05,
      "loss": 2.2167,
      "step": 203624
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.3776366710662842,
      "learning_rate": 1.9776107270526255e-05,
      "loss": 2.0319,
      "step": 203625
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.4707262516021729,
      "learning_rate": 1.9774646701097286e-05,
      "loss": 1.8901,
      "step": 203626
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.6067005395889282,
      "learning_rate": 1.977318618376744e-05,
      "loss": 1.9963,
      "step": 203627
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.374893069267273,
      "learning_rate": 1.9771725718537125e-05,
      "loss": 1.801,
      "step": 203628
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.5037444829940796,
      "learning_rate": 1.9770265305406498e-05,
      "loss": 2.0451,
      "step": 203629
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.452409267425537,
      "learning_rate": 1.9768804944375892e-05,
      "loss": 1.887,
      "step": 203630
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.3845937252044678,
      "learning_rate": 1.9767344635445482e-05,
      "loss": 2.0038,
      "step": 203631
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.3698716163635254,
      "learning_rate": 1.9765884378615694e-05,
      "loss": 2.1597,
      "step": 203632
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.5007516145706177,
      "learning_rate": 1.9764424173886627e-05,
      "loss": 1.9659,
      "step": 203633
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.3760151863098145,
      "learning_rate": 1.976296402125872e-05,
      "loss": 1.9202,
      "step": 203634
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.5683388710021973,
      "learning_rate": 1.9761503920732168e-05,
      "loss": 2.2084,
      "step": 203635
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.531187891960144,
      "learning_rate": 1.976004387230724e-05,
      "loss": 2.0825,
      "step": 203636
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.5829277038574219,
      "learning_rate": 1.975858387598417e-05,
      "loss": 1.9946,
      "step": 203637
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.4317487478256226,
      "learning_rate": 1.975712393176332e-05,
      "loss": 1.9676,
      "step": 203638
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.465910792350769,
      "learning_rate": 1.9755664039644824e-05,
      "loss": 1.8224,
      "step": 203639
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.3385753631591797,
      "learning_rate": 1.975420419962912e-05,
      "loss": 1.9007,
      "step": 203640
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.5029425621032715,
      "learning_rate": 1.9752744411716402e-05,
      "loss": 2.0988,
      "step": 203641
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.4810541868209839,
      "learning_rate": 1.9751284675906876e-05,
      "loss": 2.1241,
      "step": 203642
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.4271743297576904,
      "learning_rate": 1.9749824992200934e-05,
      "loss": 2.3101,
      "step": 203643
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.4758825302124023,
      "learning_rate": 1.9748365360598785e-05,
      "loss": 2.0468,
      "step": 203644
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.583710789680481,
      "learning_rate": 1.9746905781100654e-05,
      "loss": 2.0318,
      "step": 203645
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.3835515975952148,
      "learning_rate": 1.9745446253706908e-05,
      "loss": 1.991,
      "step": 203646
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.4304028749465942,
      "learning_rate": 1.9743986778417754e-05,
      "loss": 2.0352,
      "step": 203647
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.4244089126586914,
      "learning_rate": 1.9742527355233516e-05,
      "loss": 1.856,
      "step": 203648
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.3802158832550049,
      "learning_rate": 1.9741067984154435e-05,
      "loss": 2.1503,
      "step": 203649
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.5585488080978394,
      "learning_rate": 1.973960866518074e-05,
      "loss": 1.8072,
      "step": 203650
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.5095140933990479,
      "learning_rate": 1.973814939831273e-05,
      "loss": 1.9628,
      "step": 203651
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.5419785976409912,
      "learning_rate": 1.973669018355074e-05,
      "loss": 1.9071,
      "step": 203652
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.5776289701461792,
      "learning_rate": 1.9735231020894936e-05,
      "loss": 2.0122,
      "step": 203653
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.5144144296646118,
      "learning_rate": 1.9733771910345687e-05,
      "loss": 1.9538,
      "step": 203654
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.4499486684799194,
      "learning_rate": 1.973231285190322e-05,
      "loss": 1.9247,
      "step": 203655
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.4235919713974,
      "learning_rate": 1.9730853845567773e-05,
      "loss": 1.9301,
      "step": 203656
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.4228140115737915,
      "learning_rate": 1.9729394891339677e-05,
      "loss": 2.0281,
      "step": 203657
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.6180864572525024,
      "learning_rate": 1.97279359892192e-05,
      "loss": 1.8311,
      "step": 203658
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.4211771488189697,
      "learning_rate": 1.972647713920651e-05,
      "loss": 1.9463,
      "step": 203659
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.467710256576538,
      "learning_rate": 1.9725018341302034e-05,
      "loss": 1.9651,
      "step": 203660
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.6065616607666016,
      "learning_rate": 1.9723559595505978e-05,
      "loss": 2.1408,
      "step": 203661
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.4697688817977905,
      "learning_rate": 1.9722100901818505e-05,
      "loss": 1.9178,
      "step": 203662
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.4437437057495117,
      "learning_rate": 1.9720642260240082e-05,
      "loss": 1.9701,
      "step": 203663
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.449607253074646,
      "learning_rate": 1.971918367077081e-05,
      "loss": 2.241,
      "step": 203664
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.4642788171768188,
      "learning_rate": 1.971772513341109e-05,
      "loss": 1.8906,
      "step": 203665
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.4402048587799072,
      "learning_rate": 1.9716266648161117e-05,
      "loss": 1.7436,
      "step": 203666
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.3763469457626343,
      "learning_rate": 1.9714808215021192e-05,
      "loss": 1.9105,
      "step": 203667
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.621246576309204,
      "learning_rate": 1.9713349833991554e-05,
      "loss": 2.014,
      "step": 203668
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.4038296937942505,
      "learning_rate": 1.9711891505072498e-05,
      "loss": 1.8803,
      "step": 203669
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.4778032302856445,
      "learning_rate": 1.9710433228264287e-05,
      "loss": 1.9813,
      "step": 203670
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.3233610391616821,
      "learning_rate": 1.970897500356723e-05,
      "loss": 1.7495,
      "step": 203671
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.643131971359253,
      "learning_rate": 1.9707516830981552e-05,
      "loss": 2.2004,
      "step": 203672
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.4579421281814575,
      "learning_rate": 1.9706058710507522e-05,
      "loss": 2.0667,
      "step": 203673
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.3249742984771729,
      "learning_rate": 1.9704600642145406e-05,
      "loss": 1.8321,
      "step": 203674
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.5189791917800903,
      "learning_rate": 1.970314262589554e-05,
      "loss": 1.9451,
      "step": 203675
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.5017290115356445,
      "learning_rate": 1.970168466175812e-05,
      "loss": 2.0283,
      "step": 203676
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.3611812591552734,
      "learning_rate": 1.9700226749733482e-05,
      "loss": 1.8621,
      "step": 203677
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.5201168060302734,
      "learning_rate": 1.9698768889821858e-05,
      "loss": 1.942,
      "step": 203678
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.3731842041015625,
      "learning_rate": 1.9697311082023515e-05,
      "loss": 1.8166,
      "step": 203679
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.4128046035766602,
      "learning_rate": 1.9695853326338684e-05,
      "loss": 1.9965,
      "step": 203680
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.5409454107284546,
      "learning_rate": 1.9694395622767766e-05,
      "loss": 2.1847,
      "step": 203681
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.4015682935714722,
      "learning_rate": 1.969293797131086e-05,
      "loss": 1.9426,
      "step": 203682
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.525144100189209,
      "learning_rate": 1.9691480371968405e-05,
      "loss": 2.1009,
      "step": 203683
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.464192271232605,
      "learning_rate": 1.969002282474059e-05,
      "loss": 2.0489,
      "step": 203684
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.553753137588501,
      "learning_rate": 1.9688565329627694e-05,
      "loss": 2.067,
      "step": 203685
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.4675579071044922,
      "learning_rate": 1.9687107886629938e-05,
      "loss": 1.8454,
      "step": 203686
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.4366350173950195,
      "learning_rate": 1.9685650495747662e-05,
      "loss": 2.0181,
      "step": 203687
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.424679160118103,
      "learning_rate": 1.96841931569811e-05,
      "loss": 2.0123,
      "step": 203688
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.47303307056427,
      "learning_rate": 1.9682735870330547e-05,
      "loss": 1.7282,
      "step": 203689
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.7435590028762817,
      "learning_rate": 1.9681278635796304e-05,
      "loss": 2.1489,
      "step": 203690
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.6045031547546387,
      "learning_rate": 1.9679821453378574e-05,
      "loss": 2.0445,
      "step": 203691
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.4134505987167358,
      "learning_rate": 1.9678364323077624e-05,
      "loss": 2.0102,
      "step": 203692
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.3837296962738037,
      "learning_rate": 1.9676907244893814e-05,
      "loss": 2.1076,
      "step": 203693
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.3810803890228271,
      "learning_rate": 1.9675450218827314e-05,
      "loss": 1.9643,
      "step": 203694
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.5051326751708984,
      "learning_rate": 1.967399324487846e-05,
      "loss": 1.8327,
      "step": 203695
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.42848539352417,
      "learning_rate": 1.9672536323047518e-05,
      "loss": 2.0927,
      "step": 203696
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.6008044481277466,
      "learning_rate": 1.967107945333475e-05,
      "loss": 2.0358,
      "step": 203697
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.3765181303024292,
      "learning_rate": 1.9669622635740357e-05,
      "loss": 1.7058,
      "step": 203698
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.3971185684204102,
      "learning_rate": 1.966816587026474e-05,
      "loss": 2.0883,
      "step": 203699
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.4086737632751465,
      "learning_rate": 1.9666709156908033e-05,
      "loss": 1.9662,
      "step": 203700
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.460736632347107,
      "learning_rate": 1.9665252495670635e-05,
      "loss": 2.0234,
      "step": 203701
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.4724138975143433,
      "learning_rate": 1.9663795886552746e-05,
      "loss": 1.9497,
      "step": 203702
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.432765007019043,
      "learning_rate": 1.9662339329554665e-05,
      "loss": 1.7783,
      "step": 203703
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.3855034112930298,
      "learning_rate": 1.9660882824676594e-05,
      "loss": 2.1998,
      "step": 203704
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.4390802383422852,
      "learning_rate": 1.9659426371918897e-05,
      "loss": 1.9467,
      "step": 203705
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.3865474462509155,
      "learning_rate": 1.9657969971281773e-05,
      "loss": 2.1863,
      "step": 203706
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.39633309841156,
      "learning_rate": 1.9656513622765558e-05,
      "loss": 2.1819,
      "step": 203707
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.4399127960205078,
      "learning_rate": 1.9655057326370482e-05,
      "loss": 2.0373,
      "step": 203708
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.4518334865570068,
      "learning_rate": 1.965360108209685e-05,
      "loss": 1.9844,
      "step": 203709
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.449347734451294,
      "learning_rate": 1.9652144889944854e-05,
      "loss": 2.0745,
      "step": 203710
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.5720634460449219,
      "learning_rate": 1.9650688749914835e-05,
      "loss": 2.1682,
      "step": 203711
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.340308666229248,
      "learning_rate": 1.964923266200702e-05,
      "loss": 1.9608,
      "step": 203712
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.533158540725708,
      "learning_rate": 1.9647776626221744e-05,
      "loss": 1.9094,
      "step": 203713
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.425196886062622,
      "learning_rate": 1.964632064255921e-05,
      "loss": 2.0937,
      "step": 203714
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.470085620880127,
      "learning_rate": 1.9644864711019782e-05,
      "loss": 2.0226,
      "step": 203715
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.3370206356048584,
      "learning_rate": 1.9643408831603592e-05,
      "loss": 2.0036,
      "step": 203716
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.590119481086731,
      "learning_rate": 1.9641953004311005e-05,
      "loss": 2.107,
      "step": 203717
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.5183874368667603,
      "learning_rate": 1.964049722914226e-05,
      "loss": 1.9948,
      "step": 203718
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.263217568397522,
      "learning_rate": 1.963904150609765e-05,
      "loss": 1.9108,
      "step": 203719
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.4172518253326416,
      "learning_rate": 1.9637585835177415e-05,
      "loss": 1.9289,
      "step": 203720
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.4775115251541138,
      "learning_rate": 1.963613021638195e-05,
      "loss": 2.0826,
      "step": 203721
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.4986310005187988,
      "learning_rate": 1.9634674649711292e-05,
      "loss": 2.0449,
      "step": 203722
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.4700952768325806,
      "learning_rate": 1.96332191351659e-05,
      "loss": 2.0231,
      "step": 203723
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.5890636444091797,
      "learning_rate": 1.9631763672745947e-05,
      "loss": 1.9952,
      "step": 203724
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.563126802444458,
      "learning_rate": 1.9630308262451765e-05,
      "loss": 2.0164,
      "step": 203725
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.519476056098938,
      "learning_rate": 1.962885290428359e-05,
      "loss": 1.8566,
      "step": 203726
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.3366519212722778,
      "learning_rate": 1.9627397598241745e-05,
      "loss": 1.9315,
      "step": 203727
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.32831871509552,
      "learning_rate": 1.962594234432644e-05,
      "loss": 1.9369,
      "step": 203728
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.4519603252410889,
      "learning_rate": 1.9624487142537972e-05,
      "loss": 2.1505,
      "step": 203729
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.5908453464508057,
      "learning_rate": 1.9623031992876537e-05,
      "loss": 1.9289,
      "step": 203730
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.46760892868042,
      "learning_rate": 1.962157689534254e-05,
      "loss": 2.0343,
      "step": 203731
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.4215660095214844,
      "learning_rate": 1.9620121849936142e-05,
      "loss": 1.9438,
      "step": 203732
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.5459946393966675,
      "learning_rate": 1.9618666856657718e-05,
      "loss": 2.1653,
      "step": 203733
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.726253628730774,
      "learning_rate": 1.9617211915507425e-05,
      "loss": 1.899,
      "step": 203734
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.5854350328445435,
      "learning_rate": 1.9615757026485635e-05,
      "loss": 2.0866,
      "step": 203735
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.4316378831863403,
      "learning_rate": 1.961430218959251e-05,
      "loss": 1.8961,
      "step": 203736
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.420926809310913,
      "learning_rate": 1.961284740482839e-05,
      "loss": 2.1443,
      "step": 203737
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.4483503103256226,
      "learning_rate": 1.9611392672193537e-05,
      "loss": 2.0158,
      "step": 203738
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.4671632051467896,
      "learning_rate": 1.9609937991688253e-05,
      "loss": 1.9186,
      "step": 203739
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.527152419090271,
      "learning_rate": 1.9608483363312767e-05,
      "loss": 2.0439,
      "step": 203740
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.4553020000457764,
      "learning_rate": 1.960702878706728e-05,
      "loss": 1.975,
      "step": 203741
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.4870564937591553,
      "learning_rate": 1.9605574262952227e-05,
      "loss": 1.8451,
      "step": 203742
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.3359769582748413,
      "learning_rate": 1.9604119790967775e-05,
      "loss": 1.8752,
      "step": 203743
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.3781921863555908,
      "learning_rate": 1.9602665371114156e-05,
      "loss": 1.9179,
      "step": 203744
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.4753316640853882,
      "learning_rate": 1.960121100339177e-05,
      "loss": 1.8078,
      "step": 203745
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.4198189973831177,
      "learning_rate": 1.9599756687800783e-05,
      "loss": 2.1212,
      "step": 203746
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.3013081550598145,
      "learning_rate": 1.959830242434146e-05,
      "loss": 1.8867,
      "step": 203747
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.3059980869293213,
      "learning_rate": 1.959684821301414e-05,
      "loss": 1.9168,
      "step": 203748
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.4870188236236572,
      "learning_rate": 1.9595394053819046e-05,
      "loss": 1.9746,
      "step": 203749
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.5577774047851562,
      "learning_rate": 1.9593939946756486e-05,
      "loss": 2.0139,
      "step": 203750
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.4555702209472656,
      "learning_rate": 1.9592485891826694e-05,
      "loss": 1.8665,
      "step": 203751
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.342240333557129,
      "learning_rate": 1.9591031889029963e-05,
      "loss": 2.0188,
      "step": 203752
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.438559889793396,
      "learning_rate": 1.9589577938366497e-05,
      "loss": 1.8847,
      "step": 203753
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.4571701288223267,
      "learning_rate": 1.9588124039836693e-05,
      "loss": 2.0016,
      "step": 203754
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.55914306640625,
      "learning_rate": 1.9586670193440656e-05,
      "loss": 2.1174,
      "step": 203755
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.516648292541504,
      "learning_rate": 1.958521639917885e-05,
      "loss": 1.8766,
      "step": 203756
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.4212263822555542,
      "learning_rate": 1.9583762657051406e-05,
      "loss": 1.881,
      "step": 203757
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.5496309995651245,
      "learning_rate": 1.9582308967058657e-05,
      "loss": 1.9641,
      "step": 203758
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.3940236568450928,
      "learning_rate": 1.9580855329200806e-05,
      "loss": 2.0501,
      "step": 203759
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.4144384860992432,
      "learning_rate": 1.9579401743478186e-05,
      "loss": 2.0027,
      "step": 203760
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.420992374420166,
      "learning_rate": 1.9577948209891026e-05,
      "loss": 1.9476,
      "step": 203761
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.4901549816131592,
      "learning_rate": 1.9576494728439663e-05,
      "loss": 2.1313,
      "step": 203762
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.456424355506897,
      "learning_rate": 1.957504129912433e-05,
      "loss": 1.9692,
      "step": 203763
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.3205562829971313,
      "learning_rate": 1.957358792194529e-05,
      "loss": 2.0226,
      "step": 203764
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.54996657371521,
      "learning_rate": 1.9572134596902744e-05,
      "loss": 2.2118,
      "step": 203765
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.394163966178894,
      "learning_rate": 1.9570681323997096e-05,
      "loss": 1.9311,
      "step": 203766
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.4472123384475708,
      "learning_rate": 1.9569228103228507e-05,
      "loss": 2.0236,
      "step": 203767
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.3949921131134033,
      "learning_rate": 1.9567774934597314e-05,
      "loss": 2.0791,
      "step": 203768
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.4745779037475586,
      "learning_rate": 1.956632181810378e-05,
      "loss": 2.135,
      "step": 203769
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.4374724626541138,
      "learning_rate": 1.9564868753748174e-05,
      "loss": 1.9088,
      "step": 203770
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.3755080699920654,
      "learning_rate": 1.95634157415307e-05,
      "loss": 2.0007,
      "step": 203771
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.441386103630066,
      "learning_rate": 1.9561962781451712e-05,
      "loss": 2.0237,
      "step": 203772
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.5300068855285645,
      "learning_rate": 1.9560509873511422e-05,
      "loss": 2.2896,
      "step": 203773
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.4691557884216309,
      "learning_rate": 1.9559057017710157e-05,
      "loss": 2.0023,
      "step": 203774
    },
    {
      "epoch": 2.65,
      "grad_norm": 2.5371060371398926,
      "learning_rate": 1.9557604214048185e-05,
      "loss": 1.9775,
      "step": 203775
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.5556827783584595,
      "learning_rate": 1.9556151462525705e-05,
      "loss": 1.7829,
      "step": 203776
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.5259771347045898,
      "learning_rate": 1.9554698763143016e-05,
      "loss": 1.9818,
      "step": 203777
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.4895378351211548,
      "learning_rate": 1.9553246115900423e-05,
      "loss": 1.9521,
      "step": 203778
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.4840898513793945,
      "learning_rate": 1.9551793520798152e-05,
      "loss": 2.0804,
      "step": 203779
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.4768017530441284,
      "learning_rate": 1.9550340977836543e-05,
      "loss": 2.1071,
      "step": 203780
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.3393429517745972,
      "learning_rate": 1.9548888487015758e-05,
      "loss": 1.9631,
      "step": 203781
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.3981242179870605,
      "learning_rate": 1.954743604833623e-05,
      "loss": 1.9244,
      "step": 203782
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.4634196758270264,
      "learning_rate": 1.954598366179806e-05,
      "loss": 1.8868,
      "step": 203783
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.5030860900878906,
      "learning_rate": 1.954453132740158e-05,
      "loss": 2.0941,
      "step": 203784
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.357933759689331,
      "learning_rate": 1.9543079045147058e-05,
      "loss": 2.0193,
      "step": 203785
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.6182541847229004,
      "learning_rate": 1.9541626815034794e-05,
      "loss": 1.8919,
      "step": 203786
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.479906678199768,
      "learning_rate": 1.954017463706502e-05,
      "loss": 2.0074,
      "step": 203787
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.5767098665237427,
      "learning_rate": 1.9538722511238102e-05,
      "loss": 2.0362,
      "step": 203788
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.3932828903198242,
      "learning_rate": 1.9537270437554108e-05,
      "loss": 1.9648,
      "step": 203789
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.4631235599517822,
      "learning_rate": 1.9535818416013506e-05,
      "loss": 1.9662,
      "step": 203790
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.3780827522277832,
      "learning_rate": 1.953436644661642e-05,
      "loss": 2.1378,
      "step": 203791
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.5035616159439087,
      "learning_rate": 1.9532914529363265e-05,
      "loss": 1.6789,
      "step": 203792
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.5519849061965942,
      "learning_rate": 1.9531462664254195e-05,
      "loss": 2.1451,
      "step": 203793
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.4083366394042969,
      "learning_rate": 1.9530010851289578e-05,
      "loss": 1.9213,
      "step": 203794
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.602290153503418,
      "learning_rate": 1.9528559090469552e-05,
      "loss": 1.8011,
      "step": 203795
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.3274214267730713,
      "learning_rate": 1.9527107381794515e-05,
      "loss": 1.9309,
      "step": 203796
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.4732342958450317,
      "learning_rate": 1.95256557252646e-05,
      "loss": 2.0167,
      "step": 203797
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.3823614120483398,
      "learning_rate": 1.9524204120880237e-05,
      "loss": 1.7991,
      "step": 203798
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.3872467279434204,
      "learning_rate": 1.9522752568641563e-05,
      "loss": 1.9109,
      "step": 203799
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.3743518590927124,
      "learning_rate": 1.9521301068549013e-05,
      "loss": 1.8763,
      "step": 203800
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.3850793838500977,
      "learning_rate": 1.9519849620602646e-05,
      "loss": 2.0054,
      "step": 203801
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.4754854440689087,
      "learning_rate": 1.9518398224802867e-05,
      "loss": 2.0074,
      "step": 203802
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.465395212173462,
      "learning_rate": 1.9516946881149874e-05,
      "loss": 1.9779,
      "step": 203803
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.4788955450057983,
      "learning_rate": 1.9515495589644035e-05,
      "loss": 1.9566,
      "step": 203804
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.6323965787887573,
      "learning_rate": 1.9514044350285485e-05,
      "loss": 2.0834,
      "step": 203805
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.365868091583252,
      "learning_rate": 1.9512593163074687e-05,
      "loss": 2.0898,
      "step": 203806
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.6231606006622314,
      "learning_rate": 1.9511142028011707e-05,
      "loss": 1.8645,
      "step": 203807
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.4904510974884033,
      "learning_rate": 1.9509690945096913e-05,
      "loss": 1.9844,
      "step": 203808
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.3412868976593018,
      "learning_rate": 1.9508239914330536e-05,
      "loss": 2.0066,
      "step": 203809
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.438657522201538,
      "learning_rate": 1.9506788935712914e-05,
      "loss": 1.8471,
      "step": 203810
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.4201487302780151,
      "learning_rate": 1.950533800924424e-05,
      "loss": 2.0379,
      "step": 203811
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.3858575820922852,
      "learning_rate": 1.950388713492489e-05,
      "loss": 1.9469,
      "step": 203812
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.555651307106018,
      "learning_rate": 1.9502436312754986e-05,
      "loss": 1.9724,
      "step": 203813
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.4098902940750122,
      "learning_rate": 1.9500985542734905e-05,
      "loss": 1.8806,
      "step": 203814
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.6615740060806274,
      "learning_rate": 1.949953482486487e-05,
      "loss": 1.9437,
      "step": 203815
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.4700661897659302,
      "learning_rate": 1.949808415914519e-05,
      "loss": 1.9579,
      "step": 203816
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.4675854444503784,
      "learning_rate": 1.949663354557606e-05,
      "loss": 2.0053,
      "step": 203817
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.4600729942321777,
      "learning_rate": 1.949518298415784e-05,
      "loss": 1.931,
      "step": 203818
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.539892554283142,
      "learning_rate": 1.9493732474890777e-05,
      "loss": 2.0669,
      "step": 203819
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.5211131572723389,
      "learning_rate": 1.949228201777513e-05,
      "loss": 2.0249,
      "step": 203820
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.489134430885315,
      "learning_rate": 1.9490831612811096e-05,
      "loss": 1.9085,
      "step": 203821
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.3349403142929077,
      "learning_rate": 1.9489381259999082e-05,
      "loss": 2.163,
      "step": 203822
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.4823431968688965,
      "learning_rate": 1.948793095933925e-05,
      "loss": 1.9304,
      "step": 203823
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.5991666316986084,
      "learning_rate": 1.94864807108319e-05,
      "loss": 2.2908,
      "step": 203824
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.3797346353530884,
      "learning_rate": 1.9485030514477363e-05,
      "loss": 1.9727,
      "step": 203825
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.5351994037628174,
      "learning_rate": 1.9483580370275777e-05,
      "loss": 1.9725,
      "step": 203826
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.2875972986221313,
      "learning_rate": 1.9482130278227538e-05,
      "loss": 2.0393,
      "step": 203827
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.4137647151947021,
      "learning_rate": 1.948068023833288e-05,
      "loss": 2.0416,
      "step": 203828
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.3526190519332886,
      "learning_rate": 1.9479230250592004e-05,
      "loss": 2.1864,
      "step": 203829
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.3535288572311401,
      "learning_rate": 1.9477780315005275e-05,
      "loss": 1.9874,
      "step": 203830
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.6592376232147217,
      "learning_rate": 1.947633043157293e-05,
      "loss": 2.182,
      "step": 203831
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.3831149339675903,
      "learning_rate": 1.947488060029516e-05,
      "loss": 1.8292,
      "step": 203832
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.6724827289581299,
      "learning_rate": 1.9473430821172375e-05,
      "loss": 2.1296,
      "step": 203833
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.397403359413147,
      "learning_rate": 1.947198109420477e-05,
      "loss": 2.2082,
      "step": 203834
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.5621623992919922,
      "learning_rate": 1.9470531419392576e-05,
      "loss": 1.958,
      "step": 203835
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.6343716382980347,
      "learning_rate": 1.9469081796736162e-05,
      "loss": 1.8798,
      "step": 203836
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.736670970916748,
      "learning_rate": 1.946763222623573e-05,
      "loss": 2.0328,
      "step": 203837
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.5029246807098389,
      "learning_rate": 1.946618270789151e-05,
      "loss": 1.9519,
      "step": 203838
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.3517248630523682,
      "learning_rate": 1.9464733241703867e-05,
      "loss": 1.7142,
      "step": 203839
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.4104138612747192,
      "learning_rate": 1.9463283827673004e-05,
      "loss": 2.1084,
      "step": 203840
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.393872618675232,
      "learning_rate": 1.946183446579922e-05,
      "loss": 1.8203,
      "step": 203841
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.4401490688323975,
      "learning_rate": 1.9460385156082815e-05,
      "loss": 1.7021,
      "step": 203842
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.4562581777572632,
      "learning_rate": 1.9458935898523985e-05,
      "loss": 2.004,
      "step": 203843
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.4360431432724,
      "learning_rate": 1.9457486693123005e-05,
      "loss": 1.9041,
      "step": 203844
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.5339571237564087,
      "learning_rate": 1.9456037539880233e-05,
      "loss": 1.8641,
      "step": 203845
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.391516923904419,
      "learning_rate": 1.9454588438795805e-05,
      "loss": 1.9769,
      "step": 203846
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.3720524311065674,
      "learning_rate": 1.945313938987012e-05,
      "loss": 1.9434,
      "step": 203847
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.3333326578140259,
      "learning_rate": 1.9451690393103347e-05,
      "loss": 1.8741,
      "step": 203848
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.445330023765564,
      "learning_rate": 1.9450241448495886e-05,
      "loss": 1.9584,
      "step": 203849
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.4313604831695557,
      "learning_rate": 1.9448792556047864e-05,
      "loss": 1.8554,
      "step": 203850
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.439049482345581,
      "learning_rate": 1.9447343715759622e-05,
      "loss": 1.9048,
      "step": 203851
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.3718864917755127,
      "learning_rate": 1.9445894927631354e-05,
      "loss": 1.809,
      "step": 203852
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.2906197309494019,
      "learning_rate": 1.9444446191663467e-05,
      "loss": 1.834,
      "step": 203853
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.3958755731582642,
      "learning_rate": 1.944299750785612e-05,
      "loss": 1.9731,
      "step": 203854
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.5796416997909546,
      "learning_rate": 1.944154887620968e-05,
      "loss": 1.8451,
      "step": 203855
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.4127451181411743,
      "learning_rate": 1.9440100296724248e-05,
      "loss": 1.9283,
      "step": 203856
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.4559518098831177,
      "learning_rate": 1.943865176940026e-05,
      "loss": 1.9943,
      "step": 203857
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.4960883855819702,
      "learning_rate": 1.943720329423788e-05,
      "loss": 1.9927,
      "step": 203858
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.3988574743270874,
      "learning_rate": 1.9435754871237475e-05,
      "loss": 1.7945,
      "step": 203859
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.4610329866409302,
      "learning_rate": 1.943430650039918e-05,
      "loss": 1.9419,
      "step": 203860
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.5939784049987793,
      "learning_rate": 1.9432858181723455e-05,
      "loss": 2.1096,
      "step": 203861
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.3719637393951416,
      "learning_rate": 1.943140991521037e-05,
      "loss": 1.9164,
      "step": 203862
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.4087575674057007,
      "learning_rate": 1.942996170086033e-05,
      "loss": 1.9465,
      "step": 203863
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.4992642402648926,
      "learning_rate": 1.9428513538673497e-05,
      "loss": 1.9021,
      "step": 203864
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.418354868888855,
      "learning_rate": 1.942706542865027e-05,
      "loss": 1.8566,
      "step": 203865
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.3602155447006226,
      "learning_rate": 1.942561737079078e-05,
      "loss": 2.1244,
      "step": 203866
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.3673882484436035,
      "learning_rate": 1.9424169365095464e-05,
      "loss": 1.64,
      "step": 203867
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.4429430961608887,
      "learning_rate": 1.9422721411564392e-05,
      "loss": 2.1858,
      "step": 203868
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.429638147354126,
      "learning_rate": 1.9421273510197986e-05,
      "loss": 1.9307,
      "step": 203869
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.3222143650054932,
      "learning_rate": 1.9419825660996423e-05,
      "loss": 1.7988,
      "step": 203870
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.3626421689987183,
      "learning_rate": 1.9418377863960034e-05,
      "loss": 1.9287,
      "step": 203871
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.467542052268982,
      "learning_rate": 1.9416930119089047e-05,
      "loss": 1.7456,
      "step": 203872
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.3934414386749268,
      "learning_rate": 1.9415482426383834e-05,
      "loss": 2.2412,
      "step": 203873
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.2856618165969849,
      "learning_rate": 1.9414034785844457e-05,
      "loss": 1.9437,
      "step": 203874
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.4688750505447388,
      "learning_rate": 1.9412587197471385e-05,
      "loss": 2.0432,
      "step": 203875
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.4032317399978638,
      "learning_rate": 1.9411139661264752e-05,
      "loss": 1.859,
      "step": 203876
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.3643841743469238,
      "learning_rate": 1.9409692177224923e-05,
      "loss": 1.878,
      "step": 203877
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.566668152809143,
      "learning_rate": 1.94082447453521e-05,
      "loss": 2.2448,
      "step": 203878
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.3591946363449097,
      "learning_rate": 1.9406797365646675e-05,
      "loss": 1.8773,
      "step": 203879
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.4291160106658936,
      "learning_rate": 1.9405350038108726e-05,
      "loss": 2.1935,
      "step": 203880
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.6043636798858643,
      "learning_rate": 1.9403902762738642e-05,
      "loss": 2.1059,
      "step": 203881
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.451205849647522,
      "learning_rate": 1.9402455539536633e-05,
      "loss": 1.852,
      "step": 203882
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.4672037363052368,
      "learning_rate": 1.9401008368503057e-05,
      "loss": 1.8741,
      "step": 203883
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.4656364917755127,
      "learning_rate": 1.9399561249638085e-05,
      "loss": 1.8814,
      "step": 203884
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.3840197324752808,
      "learning_rate": 1.9398114182942147e-05,
      "loss": 1.9098,
      "step": 203885
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.50929856300354,
      "learning_rate": 1.939666716841528e-05,
      "loss": 1.8014,
      "step": 203886
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.468753695487976,
      "learning_rate": 1.9395220206057917e-05,
      "loss": 2.099,
      "step": 203887
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.6110661029815674,
      "learning_rate": 1.9393773295870218e-05,
      "loss": 2.0639,
      "step": 203888
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.4617626667022705,
      "learning_rate": 1.939232643785259e-05,
      "loss": 2.1143,
      "step": 203889
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.2938956022262573,
      "learning_rate": 1.9390879632005163e-05,
      "loss": 1.9308,
      "step": 203890
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.3597488403320312,
      "learning_rate": 1.938943287832837e-05,
      "loss": 1.8397,
      "step": 203891
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.4215840101242065,
      "learning_rate": 1.9387986176822278e-05,
      "loss": 2.0677,
      "step": 203892
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.3861814737319946,
      "learning_rate": 1.938653952748729e-05,
      "loss": 1.86,
      "step": 203893
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.3482177257537842,
      "learning_rate": 1.9385092930323598e-05,
      "loss": 2.064,
      "step": 203894
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.5694422721862793,
      "learning_rate": 1.9383646385331576e-05,
      "loss": 1.813,
      "step": 203895
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.644938588142395,
      "learning_rate": 1.9382199892511386e-05,
      "loss": 1.684,
      "step": 203896
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.5603318214416504,
      "learning_rate": 1.9380753451863428e-05,
      "loss": 1.7755,
      "step": 203897
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.4395244121551514,
      "learning_rate": 1.9379307063387773e-05,
      "loss": 1.7327,
      "step": 203898
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.484663486480713,
      "learning_rate": 1.9377860727084883e-05,
      "loss": 1.9548,
      "step": 203899
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.4250479936599731,
      "learning_rate": 1.9376414442954892e-05,
      "loss": 1.916,
      "step": 203900
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.3889648914337158,
      "learning_rate": 1.9374968210998166e-05,
      "loss": 1.9543,
      "step": 203901
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.5383689403533936,
      "learning_rate": 1.9373522031214872e-05,
      "loss": 2.026,
      "step": 203902
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.3572587966918945,
      "learning_rate": 1.937207590360538e-05,
      "loss": 2.0526,
      "step": 203903
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.4798274040222168,
      "learning_rate": 1.9370629828169913e-05,
      "loss": 1.9035,
      "step": 203904
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.5216330289840698,
      "learning_rate": 1.9369183804908783e-05,
      "loss": 1.8784,
      "step": 203905
    },
    {
      "epoch": 2.65,
      "grad_norm": 1.5981789827346802,
      "learning_rate": 1.936773783382215e-05,
      "loss": 1.9403,
      "step": 203906
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.3826056718826294,
      "learning_rate": 1.936629191491038e-05,
      "loss": 2.0076,
      "step": 203907
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.445467472076416,
      "learning_rate": 1.9364846048173678e-05,
      "loss": 2.0188,
      "step": 203908
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.5959383249282837,
      "learning_rate": 1.9363400233612403e-05,
      "loss": 1.9607,
      "step": 203909
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.4290324449539185,
      "learning_rate": 1.936195447122676e-05,
      "loss": 2.0335,
      "step": 203910
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.3678760528564453,
      "learning_rate": 1.9360508761017013e-05,
      "loss": 2.1084,
      "step": 203911
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.6376854181289673,
      "learning_rate": 1.9359063102983463e-05,
      "loss": 1.8939,
      "step": 203912
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.7010804414749146,
      "learning_rate": 1.9357617497126343e-05,
      "loss": 1.8195,
      "step": 203913
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.4447379112243652,
      "learning_rate": 1.935617194344592e-05,
      "loss": 2.1013,
      "step": 203914
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.5834258794784546,
      "learning_rate": 1.9354726441942524e-05,
      "loss": 1.9572,
      "step": 203915
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.4253581762313843,
      "learning_rate": 1.935328099261636e-05,
      "loss": 1.8906,
      "step": 203916
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.3756808042526245,
      "learning_rate": 1.9351835595467725e-05,
      "loss": 1.8607,
      "step": 203917
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.746638298034668,
      "learning_rate": 1.935039025049688e-05,
      "loss": 2.1694,
      "step": 203918
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.4036716222763062,
      "learning_rate": 1.9348944957704104e-05,
      "loss": 2.0103,
      "step": 203919
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.48103666305542,
      "learning_rate": 1.934749971708962e-05,
      "loss": 2.1122,
      "step": 203920
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.5229438543319702,
      "learning_rate": 1.9346054528653766e-05,
      "loss": 1.9419,
      "step": 203921
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.361425518989563,
      "learning_rate": 1.9344609392396804e-05,
      "loss": 1.8547,
      "step": 203922
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.3569672107696533,
      "learning_rate": 1.9343164308318905e-05,
      "loss": 1.7504,
      "step": 203923
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.3619449138641357,
      "learning_rate": 1.9341719276420464e-05,
      "loss": 1.9554,
      "step": 203924
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.4802006483078003,
      "learning_rate": 1.9340274296701653e-05,
      "loss": 2.0664,
      "step": 203925
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.4291898012161255,
      "learning_rate": 1.9338829369162835e-05,
      "loss": 2.1428,
      "step": 203926
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.4758416414260864,
      "learning_rate": 1.9337384493804208e-05,
      "loss": 2.0043,
      "step": 203927
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.3467977046966553,
      "learning_rate": 1.9335939670626077e-05,
      "loss": 1.9162,
      "step": 203928
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.509771466255188,
      "learning_rate": 1.9334494899628638e-05,
      "loss": 1.7034,
      "step": 203929
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.4056994915008545,
      "learning_rate": 1.933305018081226e-05,
      "loss": 1.9735,
      "step": 203930
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.500099539756775,
      "learning_rate": 1.9331605514177107e-05,
      "loss": 2.144,
      "step": 203931
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.5713340044021606,
      "learning_rate": 1.9330160899723578e-05,
      "loss": 1.9872,
      "step": 203932
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.5056573152542114,
      "learning_rate": 1.9328716337451845e-05,
      "loss": 1.9689,
      "step": 203933
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.4659043550491333,
      "learning_rate": 1.9327271827362234e-05,
      "loss": 2.1112,
      "step": 203934
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.44154953956604,
      "learning_rate": 1.9325827369454915e-05,
      "loss": 1.9491,
      "step": 203935
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.4014196395874023,
      "learning_rate": 1.9324382963730257e-05,
      "loss": 1.8591,
      "step": 203936
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.3917882442474365,
      "learning_rate": 1.9322938610188486e-05,
      "loss": 1.982,
      "step": 203937
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.409104347229004,
      "learning_rate": 1.9321494308829878e-05,
      "loss": 2.0046,
      "step": 203938
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.4168757200241089,
      "learning_rate": 1.932005005965469e-05,
      "loss": 1.9924,
      "step": 203939
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.5217053890228271,
      "learning_rate": 1.9318605862663296e-05,
      "loss": 1.954,
      "step": 203940
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.4285811185836792,
      "learning_rate": 1.9317161717855756e-05,
      "loss": 1.9251,
      "step": 203941
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.3472007513046265,
      "learning_rate": 1.9315717625232507e-05,
      "loss": 1.8917,
      "step": 203942
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.391896367073059,
      "learning_rate": 1.9314273584793717e-05,
      "loss": 2.1858,
      "step": 203943
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.4741541147232056,
      "learning_rate": 1.9312829596539747e-05,
      "loss": 1.8545,
      "step": 203944
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.4160372018814087,
      "learning_rate": 1.93113856604708e-05,
      "loss": 1.9585,
      "step": 203945
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.3054157495498657,
      "learning_rate": 1.9309941776587245e-05,
      "loss": 1.7673,
      "step": 203946
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.5072871446609497,
      "learning_rate": 1.9308497944889178e-05,
      "loss": 1.9314,
      "step": 203947
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.3875031471252441,
      "learning_rate": 1.9307054165377e-05,
      "loss": 2.1191,
      "step": 203948
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.4137428998947144,
      "learning_rate": 1.930561043805088e-05,
      "loss": 2.0674,
      "step": 203949
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.3698817491531372,
      "learning_rate": 1.930416676291121e-05,
      "loss": 2.0785,
      "step": 203950
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.4723998308181763,
      "learning_rate": 1.930272313995813e-05,
      "loss": 1.9927,
      "step": 203951
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.6047555208206177,
      "learning_rate": 1.9301279569192073e-05,
      "loss": 1.9878,
      "step": 203952
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.5337024927139282,
      "learning_rate": 1.929983605061314e-05,
      "loss": 1.8992,
      "step": 203953
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.5091102123260498,
      "learning_rate": 1.929839258422169e-05,
      "loss": 1.9972,
      "step": 203954
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.4571539163589478,
      "learning_rate": 1.9296949170017895e-05,
      "loss": 1.8858,
      "step": 203955
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.5619711875915527,
      "learning_rate": 1.9295505808002186e-05,
      "loss": 1.9502,
      "step": 203956
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.7097394466400146,
      "learning_rate": 1.9294062498174667e-05,
      "loss": 2.161,
      "step": 203957
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.5054800510406494,
      "learning_rate": 1.92926192405358e-05,
      "loss": 1.8918,
      "step": 203958
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.5904589891433716,
      "learning_rate": 1.929117603508562e-05,
      "loss": 1.7086,
      "step": 203959
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.5461899042129517,
      "learning_rate": 1.928973288182456e-05,
      "loss": 2.0871,
      "step": 203960
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.4536912441253662,
      "learning_rate": 1.9288289780752785e-05,
      "loss": 2.0067,
      "step": 203961
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.34750235080719,
      "learning_rate": 1.9286846731870665e-05,
      "loss": 1.7774,
      "step": 203962
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.4274578094482422,
      "learning_rate": 1.9285403735178362e-05,
      "loss": 1.9842,
      "step": 203963
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.6482689380645752,
      "learning_rate": 1.9283960790676312e-05,
      "loss": 2.2388,
      "step": 203964
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.6028120517730713,
      "learning_rate": 1.928251789836458e-05,
      "loss": 2.1367,
      "step": 203965
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.431740164756775,
      "learning_rate": 1.9281075058243567e-05,
      "loss": 1.9243,
      "step": 203966
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.5137332677841187,
      "learning_rate": 1.927963227031347e-05,
      "loss": 1.7687,
      "step": 203967
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.4145046472549438,
      "learning_rate": 1.9278189534574627e-05,
      "loss": 2.078,
      "step": 203968
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.4533460140228271,
      "learning_rate": 1.9276746851027203e-05,
      "loss": 2.0629,
      "step": 203969
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.6166008710861206,
      "learning_rate": 1.9275304219671628e-05,
      "loss": 2.0041,
      "step": 203970
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.5375173091888428,
      "learning_rate": 1.9273861640508004e-05,
      "loss": 2.0018,
      "step": 203971
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.4928432703018188,
      "learning_rate": 1.9272419113536697e-05,
      "loss": 2.0292,
      "step": 203972
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.4105517864227295,
      "learning_rate": 1.9270976638757906e-05,
      "loss": 2.0956,
      "step": 203973
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.4222468137741089,
      "learning_rate": 1.926953421617197e-05,
      "loss": 1.8701,
      "step": 203974
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.4028310775756836,
      "learning_rate": 1.926809184577911e-05,
      "loss": 1.9839,
      "step": 203975
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.3468483686447144,
      "learning_rate": 1.926664952757967e-05,
      "loss": 2.1454,
      "step": 203976
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.4748659133911133,
      "learning_rate": 1.9265207261573778e-05,
      "loss": 2.159,
      "step": 203977
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.6210545301437378,
      "learning_rate": 1.9263765047761838e-05,
      "loss": 2.2038,
      "step": 203978
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.411545991897583,
      "learning_rate": 1.926232288614401e-05,
      "loss": 1.9621,
      "step": 203979
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.4543617963790894,
      "learning_rate": 1.9260880776720667e-05,
      "loss": 1.8969,
      "step": 203980
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.4676700830459595,
      "learning_rate": 1.9259438719491972e-05,
      "loss": 1.9438,
      "step": 203981
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.6010465621948242,
      "learning_rate": 1.9257996714458325e-05,
      "loss": 1.993,
      "step": 203982
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.3300049304962158,
      "learning_rate": 1.9256554761619892e-05,
      "loss": 2.1375,
      "step": 203983
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.4900184869766235,
      "learning_rate": 1.925511286097694e-05,
      "loss": 2.0788,
      "step": 203984
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.4479148387908936,
      "learning_rate": 1.9253671012529737e-05,
      "loss": 1.9829,
      "step": 203985
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.4523156881332397,
      "learning_rate": 1.9252229216278646e-05,
      "loss": 2.0045,
      "step": 203986
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.413778305053711,
      "learning_rate": 1.925078747222377e-05,
      "loss": 1.9279,
      "step": 203987
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.488219976425171,
      "learning_rate": 1.9249345780365544e-05,
      "loss": 1.8631,
      "step": 203988
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.4001352787017822,
      "learning_rate": 1.9247904140704163e-05,
      "loss": 2.1021,
      "step": 203989
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.3982253074645996,
      "learning_rate": 1.9246462553239927e-05,
      "loss": 1.8742,
      "step": 203990
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.4602129459381104,
      "learning_rate": 1.9245021017972972e-05,
      "loss": 1.93,
      "step": 203991
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.438506007194519,
      "learning_rate": 1.9243579534903765e-05,
      "loss": 2.0409,
      "step": 203992
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.4079393148422241,
      "learning_rate": 1.92421381040324e-05,
      "loss": 1.9161,
      "step": 203993
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.3084399700164795,
      "learning_rate": 1.924069672535925e-05,
      "loss": 1.8096,
      "step": 203994
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.6323282718658447,
      "learning_rate": 1.923925539888458e-05,
      "loss": 1.9028,
      "step": 203995
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.3708443641662598,
      "learning_rate": 1.9237814124608652e-05,
      "loss": 1.9061,
      "step": 203996
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.497195839881897,
      "learning_rate": 1.9236372902531638e-05,
      "loss": 1.929,
      "step": 203997
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.328137993812561,
      "learning_rate": 1.9234931732653934e-05,
      "loss": 2.1697,
      "step": 203998
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.3927404880523682,
      "learning_rate": 1.9233490614975712e-05,
      "loss": 1.8739,
      "step": 203999
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.5880742073059082,
      "learning_rate": 1.923204954949733e-05,
      "loss": 1.8699,
      "step": 204000
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.4549331665039062,
      "learning_rate": 1.9230608536218995e-05,
      "loss": 1.8844,
      "step": 204001
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.4272245168685913,
      "learning_rate": 1.922916757514097e-05,
      "loss": 2.1008,
      "step": 204002
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.3868037462234497,
      "learning_rate": 1.9227726666263553e-05,
      "loss": 1.951,
      "step": 204003
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.5708872079849243,
      "learning_rate": 1.922628580958705e-05,
      "loss": 1.7667,
      "step": 204004
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.5150874853134155,
      "learning_rate": 1.922484500511159e-05,
      "loss": 2.0508,
      "step": 204005
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.437172532081604,
      "learning_rate": 1.9223404252837604e-05,
      "loss": 2.0981,
      "step": 204006
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.442863941192627,
      "learning_rate": 1.9221963552765298e-05,
      "loss": 2.0592,
      "step": 204007
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.6919598579406738,
      "learning_rate": 1.9220522904894862e-05,
      "loss": 2.0865,
      "step": 204008
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.388306975364685,
      "learning_rate": 1.9219082309226675e-05,
      "loss": 1.9447,
      "step": 204009
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.3587738275527954,
      "learning_rate": 1.9217641765760928e-05,
      "loss": 2.1031,
      "step": 204010
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.457672119140625,
      "learning_rate": 1.9216201274497954e-05,
      "loss": 1.9345,
      "step": 204011
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.3986479043960571,
      "learning_rate": 1.9214760835437993e-05,
      "loss": 1.9775,
      "step": 204012
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.4250315427780151,
      "learning_rate": 1.9213320448581303e-05,
      "loss": 2.1778,
      "step": 204013
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.5578826665878296,
      "learning_rate": 1.9211880113928124e-05,
      "loss": 2.0231,
      "step": 204014
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.370334267616272,
      "learning_rate": 1.9210439831478785e-05,
      "loss": 1.8627,
      "step": 204015
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.4823040962219238,
      "learning_rate": 1.920899960123349e-05,
      "loss": 1.8869,
      "step": 204016
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.4291564226150513,
      "learning_rate": 1.92075594231926e-05,
      "loss": 1.9022,
      "step": 204017
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.54728102684021,
      "learning_rate": 1.9206119297356314e-05,
      "loss": 2.0079,
      "step": 204018
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.389427661895752,
      "learning_rate": 1.9204679223724907e-05,
      "loss": 1.9691,
      "step": 204019
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.5940059423446655,
      "learning_rate": 1.9203239202298603e-05,
      "loss": 2.203,
      "step": 204020
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.3982346057891846,
      "learning_rate": 1.9201799233077775e-05,
      "loss": 1.804,
      "step": 204021
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.3774468898773193,
      "learning_rate": 1.9200359316062553e-05,
      "loss": 1.8402,
      "step": 204022
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.3821418285369873,
      "learning_rate": 1.919891945125337e-05,
      "loss": 2.059,
      "step": 204023
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.3407312631607056,
      "learning_rate": 1.919747963865036e-05,
      "loss": 2.0703,
      "step": 204024
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.4025993347167969,
      "learning_rate": 1.919603987825392e-05,
      "loss": 1.8512,
      "step": 204025
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.530382752418518,
      "learning_rate": 1.919460017006412e-05,
      "loss": 2.2662,
      "step": 204026
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.505850076675415,
      "learning_rate": 1.9193160514081428e-05,
      "loss": 1.9915,
      "step": 204027
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.4213025569915771,
      "learning_rate": 1.9191720910305976e-05,
      "loss": 1.8717,
      "step": 204028
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.5716731548309326,
      "learning_rate": 1.9190281358738124e-05,
      "loss": 2.0745,
      "step": 204029
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.705487608909607,
      "learning_rate": 1.9188841859378046e-05,
      "loss": 2.2781,
      "step": 204030
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.5602045059204102,
      "learning_rate": 1.9187402412226172e-05,
      "loss": 1.7292,
      "step": 204031
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.6265366077423096,
      "learning_rate": 1.918596301728257e-05,
      "loss": 1.9936,
      "step": 204032
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.461551308631897,
      "learning_rate": 1.9184523674547636e-05,
      "loss": 2.1226,
      "step": 204033
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.561289668083191,
      "learning_rate": 1.9183084384021542e-05,
      "loss": 2.0531,
      "step": 204034
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.405914545059204,
      "learning_rate": 1.9181645145704682e-05,
      "loss": 1.8511,
      "step": 204035
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.5528467893600464,
      "learning_rate": 1.9180205959597195e-05,
      "loss": 2.0736,
      "step": 204036
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.4168058633804321,
      "learning_rate": 1.9178766825699478e-05,
      "loss": 1.9298,
      "step": 204037
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.5285176038742065,
      "learning_rate": 1.917732774401166e-05,
      "loss": 1.9853,
      "step": 204038
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.5268207788467407,
      "learning_rate": 1.917588871453415e-05,
      "loss": 2.0679,
      "step": 204039
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.602118968963623,
      "learning_rate": 1.917444973726704e-05,
      "loss": 2.2152,
      "step": 204040
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.4367806911468506,
      "learning_rate": 1.9173010812210798e-05,
      "loss": 2.1119,
      "step": 204041
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.3308767080307007,
      "learning_rate": 1.9171571939365527e-05,
      "loss": 1.9399,
      "step": 204042
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.4827741384506226,
      "learning_rate": 1.9170133118731656e-05,
      "loss": 1.8914,
      "step": 204043
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.42267644405365,
      "learning_rate": 1.9168694350309254e-05,
      "loss": 1.9605,
      "step": 204044
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.265692949295044,
      "learning_rate": 1.9167255634098756e-05,
      "loss": 1.924,
      "step": 204045
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.347531795501709,
      "learning_rate": 1.916581697010032e-05,
      "loss": 1.9012,
      "step": 204046
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.4526679515838623,
      "learning_rate": 1.9164378358314292e-05,
      "loss": 1.9623,
      "step": 204047
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.3641600608825684,
      "learning_rate": 1.916293979874086e-05,
      "loss": 1.9482,
      "step": 204048
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.4524915218353271,
      "learning_rate": 1.91615012913804e-05,
      "loss": 2.0029,
      "step": 204049
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.5282306671142578,
      "learning_rate": 1.9160062836233137e-05,
      "loss": 1.9439,
      "step": 204050
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.421466588973999,
      "learning_rate": 1.9158624433299276e-05,
      "loss": 1.996,
      "step": 204051
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.5853607654571533,
      "learning_rate": 1.915718608257911e-05,
      "loss": 1.9823,
      "step": 204052
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.3719604015350342,
      "learning_rate": 1.915574778407295e-05,
      "loss": 1.9272,
      "step": 204053
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.334096908569336,
      "learning_rate": 1.915430953778102e-05,
      "loss": 1.7902,
      "step": 204054
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.3846936225891113,
      "learning_rate": 1.9152871343703623e-05,
      "loss": 1.6932,
      "step": 204055
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.5935876369476318,
      "learning_rate": 1.9151433201841027e-05,
      "loss": 2.0066,
      "step": 204056
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.3698267936706543,
      "learning_rate": 1.9149995112193462e-05,
      "loss": 2.028,
      "step": 204057
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.326877236366272,
      "learning_rate": 1.9148557074761195e-05,
      "loss": 2.1055,
      "step": 204058
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.4468947649002075,
      "learning_rate": 1.914711908954456e-05,
      "loss": 1.7896,
      "step": 204059
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.624890685081482,
      "learning_rate": 1.914568115654369e-05,
      "loss": 2.0408,
      "step": 204060
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.5730445384979248,
      "learning_rate": 1.914424327575902e-05,
      "loss": 1.772,
      "step": 204061
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.6177493333816528,
      "learning_rate": 1.9142805447190712e-05,
      "loss": 1.9413,
      "step": 204062
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.3601073026657104,
      "learning_rate": 1.9141367670839102e-05,
      "loss": 1.9536,
      "step": 204063
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.3653745651245117,
      "learning_rate": 1.9139929946704326e-05,
      "loss": 2.0152,
      "step": 204064
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.250132441520691,
      "learning_rate": 1.913849227478681e-05,
      "loss": 2.091,
      "step": 204065
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.5149778127670288,
      "learning_rate": 1.9137054655086693e-05,
      "loss": 1.9914,
      "step": 204066
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.4470343589782715,
      "learning_rate": 1.913561708760434e-05,
      "loss": 1.9179,
      "step": 204067
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.5092263221740723,
      "learning_rate": 1.913417957233998e-05,
      "loss": 2.0743,
      "step": 204068
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.57680082321167,
      "learning_rate": 1.9132742109293887e-05,
      "loss": 1.8695,
      "step": 204069
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.6452314853668213,
      "learning_rate": 1.9131304698466254e-05,
      "loss": 2.0952,
      "step": 204070
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.3507834672927856,
      "learning_rate": 1.9129867339857487e-05,
      "loss": 1.8324,
      "step": 204071
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.5057045221328735,
      "learning_rate": 1.9128430033467712e-05,
      "loss": 1.9444,
      "step": 204072
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.4029241800308228,
      "learning_rate": 1.91269927792973e-05,
      "loss": 1.932,
      "step": 204073
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.3376003503799438,
      "learning_rate": 1.9125555577346485e-05,
      "loss": 1.9317,
      "step": 204074
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.5817242860794067,
      "learning_rate": 1.9124118427615566e-05,
      "loss": 2.1043,
      "step": 204075
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.395187258720398,
      "learning_rate": 1.9122681330104705e-05,
      "loss": 1.9222,
      "step": 204076
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.4579623937606812,
      "learning_rate": 1.9121244284814276e-05,
      "loss": 1.7862,
      "step": 204077
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.4831870794296265,
      "learning_rate": 1.911980729174447e-05,
      "loss": 1.9392,
      "step": 204078
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.3813263177871704,
      "learning_rate": 1.911837035089566e-05,
      "loss": 2.0103,
      "step": 204079
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.4029492139816284,
      "learning_rate": 1.9116933462268014e-05,
      "loss": 2.1286,
      "step": 204080
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.5951884984970093,
      "learning_rate": 1.911549662586186e-05,
      "loss": 1.8827,
      "step": 204081
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.4126920700073242,
      "learning_rate": 1.9114059841677366e-05,
      "loss": 1.7861,
      "step": 204082
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.5534979104995728,
      "learning_rate": 1.911262310971493e-05,
      "loss": 2.0085,
      "step": 204083
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.3751145601272583,
      "learning_rate": 1.9111186429974723e-05,
      "loss": 1.75,
      "step": 204084
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.4412200450897217,
      "learning_rate": 1.910974980245711e-05,
      "loss": 2.0017,
      "step": 204085
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.4442540407180786,
      "learning_rate": 1.9108313227162253e-05,
      "loss": 1.9809,
      "step": 204086
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.4952583312988281,
      "learning_rate": 1.9106876704090424e-05,
      "loss": 1.7034,
      "step": 204087
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.5703821182250977,
      "learning_rate": 1.910544023324202e-05,
      "loss": 2.1243,
      "step": 204088
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.4769335985183716,
      "learning_rate": 1.9104003814617176e-05,
      "loss": 1.9291,
      "step": 204089
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.5273927450180054,
      "learning_rate": 1.9102567448216156e-05,
      "loss": 1.7601,
      "step": 204090
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.5062493085861206,
      "learning_rate": 1.9101131134039326e-05,
      "loss": 2.0681,
      "step": 204091
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.5079835653305054,
      "learning_rate": 1.909969487208689e-05,
      "loss": 1.9894,
      "step": 204092
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.4137654304504395,
      "learning_rate": 1.909825866235908e-05,
      "loss": 2.0605,
      "step": 204093
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.5222280025482178,
      "learning_rate": 1.9096822504856258e-05,
      "loss": 2.0851,
      "step": 204094
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.590930461883545,
      "learning_rate": 1.9095386399578626e-05,
      "loss": 2.0415,
      "step": 204095
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.5456047058105469,
      "learning_rate": 1.9093950346526454e-05,
      "loss": 2.0462,
      "step": 204096
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.4422069787979126,
      "learning_rate": 1.909251434570004e-05,
      "loss": 1.9836,
      "step": 204097
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.585572361946106,
      "learning_rate": 1.9091078397099614e-05,
      "loss": 2.1765,
      "step": 204098
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.3734214305877686,
      "learning_rate": 1.9089642500725444e-05,
      "loss": 2.3222,
      "step": 204099
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.4655665159225464,
      "learning_rate": 1.908820665657783e-05,
      "loss": 2.2079,
      "step": 204100
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.4857348203659058,
      "learning_rate": 1.9086770864656975e-05,
      "loss": 2.0708,
      "step": 204101
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.422875165939331,
      "learning_rate": 1.9085335124963276e-05,
      "loss": 1.9973,
      "step": 204102
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.471845269203186,
      "learning_rate": 1.9083899437496864e-05,
      "loss": 1.9895,
      "step": 204103
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.4851528406143188,
      "learning_rate": 1.908246380225811e-05,
      "loss": 2.0357,
      "step": 204104
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.428231954574585,
      "learning_rate": 1.9081028219247142e-05,
      "loss": 2.1151,
      "step": 204105
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.4990708827972412,
      "learning_rate": 1.9079592688464396e-05,
      "loss": 2.0852,
      "step": 204106
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.5646164417266846,
      "learning_rate": 1.9078157209909972e-05,
      "loss": 1.9781,
      "step": 204107
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.4236279726028442,
      "learning_rate": 1.907672178358427e-05,
      "loss": 1.8878,
      "step": 204108
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.5609102249145508,
      "learning_rate": 1.907528640948749e-05,
      "loss": 2.1295,
      "step": 204109
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.3675017356872559,
      "learning_rate": 1.9073851087619995e-05,
      "loss": 2.1044,
      "step": 204110
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.3574641942977905,
      "learning_rate": 1.907241581798189e-05,
      "loss": 1.7182,
      "step": 204111
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.2969056367874146,
      "learning_rate": 1.9070980600573573e-05,
      "loss": 1.9497,
      "step": 204112
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.373180627822876,
      "learning_rate": 1.9069545435395207e-05,
      "loss": 1.7518,
      "step": 204113
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.4194347858428955,
      "learning_rate": 1.9068110322447163e-05,
      "loss": 2.1329,
      "step": 204114
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.3649349212646484,
      "learning_rate": 1.9066675261729603e-05,
      "loss": 2.0195,
      "step": 204115
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.420076608657837,
      "learning_rate": 1.906524025324293e-05,
      "loss": 2.1384,
      "step": 204116
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.458977460861206,
      "learning_rate": 1.906380529698731e-05,
      "loss": 2.0143,
      "step": 204117
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.414025902748108,
      "learning_rate": 1.906237039296301e-05,
      "loss": 1.9718,
      "step": 204118
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.3502894639968872,
      "learning_rate": 1.9060935541170297e-05,
      "loss": 2.0252,
      "step": 204119
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.4336791038513184,
      "learning_rate": 1.90595007416095e-05,
      "loss": 2.0171,
      "step": 204120
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.3561863899230957,
      "learning_rate": 1.9058065994280823e-05,
      "loss": 1.9782,
      "step": 204121
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.4176199436187744,
      "learning_rate": 1.9056631299184568e-05,
      "loss": 1.8895,
      "step": 204122
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.589056134223938,
      "learning_rate": 1.905519665632099e-05,
      "loss": 2.1515,
      "step": 204123
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.512428879737854,
      "learning_rate": 1.905376206569037e-05,
      "loss": 2.07,
      "step": 204124
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.5438870191574097,
      "learning_rate": 1.9052327527292932e-05,
      "loss": 1.8955,
      "step": 204125
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.4267573356628418,
      "learning_rate": 1.9050893041128977e-05,
      "loss": 2.0938,
      "step": 204126
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.5239399671554565,
      "learning_rate": 1.9049458607198743e-05,
      "loss": 2.0369,
      "step": 204127
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.4029957056045532,
      "learning_rate": 1.9048024225502556e-05,
      "loss": 2.0035,
      "step": 204128
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.3928215503692627,
      "learning_rate": 1.904658989604062e-05,
      "loss": 2.0618,
      "step": 204129
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.4174346923828125,
      "learning_rate": 1.9045155618813268e-05,
      "loss": 1.9094,
      "step": 204130
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.6451040506362915,
      "learning_rate": 1.9043721393820667e-05,
      "loss": 1.789,
      "step": 204131
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.4538356065750122,
      "learning_rate": 1.904228722106318e-05,
      "loss": 1.8933,
      "step": 204132
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.3391342163085938,
      "learning_rate": 1.9040853100540975e-05,
      "loss": 1.9872,
      "step": 204133
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.4375230073928833,
      "learning_rate": 1.9039419032254423e-05,
      "loss": 1.8732,
      "step": 204134
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.535197138786316,
      "learning_rate": 1.9037985016203783e-05,
      "loss": 1.978,
      "step": 204135
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.628945231437683,
      "learning_rate": 1.9036551052389258e-05,
      "loss": 2.1375,
      "step": 204136
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.4039890766143799,
      "learning_rate": 1.903511714081112e-05,
      "loss": 1.9777,
      "step": 204137
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.4471739530563354,
      "learning_rate": 1.903368328146966e-05,
      "loss": 1.9417,
      "step": 204138
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.5520687103271484,
      "learning_rate": 1.9032249474365147e-05,
      "loss": 2.0735,
      "step": 204139
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.389330267906189,
      "learning_rate": 1.9030815719497848e-05,
      "loss": 1.8882,
      "step": 204140
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.4082636833190918,
      "learning_rate": 1.9029382016868033e-05,
      "loss": 1.7871,
      "step": 204141
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.3881977796554565,
      "learning_rate": 1.9027948366475966e-05,
      "loss": 2.0126,
      "step": 204142
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.3794705867767334,
      "learning_rate": 1.9026514768321875e-05,
      "loss": 2.2469,
      "step": 204143
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.5775082111358643,
      "learning_rate": 1.90250812224061e-05,
      "loss": 1.9662,
      "step": 204144
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.5337505340576172,
      "learning_rate": 1.9023647728728808e-05,
      "loss": 2.1039,
      "step": 204145
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.5160011053085327,
      "learning_rate": 1.902221428729036e-05,
      "loss": 1.7897,
      "step": 204146
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.363582730293274,
      "learning_rate": 1.9020780898090993e-05,
      "loss": 1.8452,
      "step": 204147
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.275231957435608,
      "learning_rate": 1.901934756113097e-05,
      "loss": 1.6418,
      "step": 204148
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.4520149230957031,
      "learning_rate": 1.9017914276410528e-05,
      "loss": 1.9039,
      "step": 204149
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.4564743041992188,
      "learning_rate": 1.901648104393e-05,
      "loss": 2.1513,
      "step": 204150
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.3780888319015503,
      "learning_rate": 1.9015047863689547e-05,
      "loss": 2.181,
      "step": 204151
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.5120611190795898,
      "learning_rate": 1.9013614735689543e-05,
      "loss": 1.8615,
      "step": 204152
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.4515149593353271,
      "learning_rate": 1.901218165993025e-05,
      "loss": 2.0269,
      "step": 204153
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.5768160820007324,
      "learning_rate": 1.9010748636411864e-05,
      "loss": 1.7444,
      "step": 204154
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.3698879480361938,
      "learning_rate": 1.9009315665134626e-05,
      "loss": 1.9567,
      "step": 204155
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.5559757947921753,
      "learning_rate": 1.9007882746098935e-05,
      "loss": 2.1192,
      "step": 204156
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.3823421001434326,
      "learning_rate": 1.900644987930492e-05,
      "loss": 2.138,
      "step": 204157
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.5516387224197388,
      "learning_rate": 1.900501706475298e-05,
      "loss": 2.0366,
      "step": 204158
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.404937505722046,
      "learning_rate": 1.9003584302443287e-05,
      "loss": 2.0752,
      "step": 204159
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.4750920534133911,
      "learning_rate": 1.9002151592376136e-05,
      "loss": 2.1682,
      "step": 204160
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.4784563779830933,
      "learning_rate": 1.9000718934551727e-05,
      "loss": 2.2245,
      "step": 204161
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.3527811765670776,
      "learning_rate": 1.8999286328970464e-05,
      "loss": 1.8656,
      "step": 204162
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.4156516790390015,
      "learning_rate": 1.8997853775632476e-05,
      "loss": 2.0233,
      "step": 204163
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.5266058444976807,
      "learning_rate": 1.8996421274538165e-05,
      "loss": 1.9852,
      "step": 204164
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.4759465456008911,
      "learning_rate": 1.8994988825687697e-05,
      "loss": 2.3469,
      "step": 204165
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.3361775875091553,
      "learning_rate": 1.899355642908137e-05,
      "loss": 2.1302,
      "step": 204166
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.3700129985809326,
      "learning_rate": 1.8992124084719385e-05,
      "loss": 1.8498,
      "step": 204167
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.4530580043792725,
      "learning_rate": 1.8990691792602142e-05,
      "loss": 2.1827,
      "step": 204168
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.5355600118637085,
      "learning_rate": 1.8989259552729773e-05,
      "loss": 1.9606,
      "step": 204169
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.3614619970321655,
      "learning_rate": 1.8987827365102647e-05,
      "loss": 2.0086,
      "step": 204170
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.5546847581863403,
      "learning_rate": 1.8986395229720995e-05,
      "loss": 2.0577,
      "step": 204171
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.3758405447006226,
      "learning_rate": 1.898496314658502e-05,
      "loss": 2.0757,
      "step": 204172
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.4040311574935913,
      "learning_rate": 1.8983531115695116e-05,
      "loss": 2.0551,
      "step": 204173
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.3036004304885864,
      "learning_rate": 1.8982099137051455e-05,
      "loss": 1.945,
      "step": 204174
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.5009628534317017,
      "learning_rate": 1.8980667210654298e-05,
      "loss": 2.0733,
      "step": 204175
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.477515459060669,
      "learning_rate": 1.8979235336503983e-05,
      "loss": 2.0821,
      "step": 204176
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.4883939027786255,
      "learning_rate": 1.897780351460071e-05,
      "loss": 2.0699,
      "step": 204177
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.588377594947815,
      "learning_rate": 1.897637174494474e-05,
      "loss": 1.9071,
      "step": 204178
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.5150668621063232,
      "learning_rate": 1.897494002753641e-05,
      "loss": 1.8181,
      "step": 204179
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.4049768447875977,
      "learning_rate": 1.8973508362375954e-05,
      "loss": 1.9424,
      "step": 204180
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.3738443851470947,
      "learning_rate": 1.8972076749463573e-05,
      "loss": 1.9466,
      "step": 204181
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.4586403369903564,
      "learning_rate": 1.8970645188799626e-05,
      "loss": 1.9024,
      "step": 204182
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.4250941276550293,
      "learning_rate": 1.8969213680384287e-05,
      "loss": 1.8903,
      "step": 204183
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.5464746952056885,
      "learning_rate": 1.8967782224217955e-05,
      "loss": 1.9408,
      "step": 204184
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.3753694295883179,
      "learning_rate": 1.896635082030079e-05,
      "loss": 2.071,
      "step": 204185
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.3806043863296509,
      "learning_rate": 1.8964919468633034e-05,
      "loss": 1.83,
      "step": 204186
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.4426301717758179,
      "learning_rate": 1.8963488169215078e-05,
      "loss": 2.0588,
      "step": 204187
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.267273187637329,
      "learning_rate": 1.89620569220471e-05,
      "loss": 1.8894,
      "step": 204188
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.5132085084915161,
      "learning_rate": 1.8960625727129353e-05,
      "loss": 2.055,
      "step": 204189
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.3525011539459229,
      "learning_rate": 1.8959194584462146e-05,
      "loss": 2.0326,
      "step": 204190
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.417605996131897,
      "learning_rate": 1.895776349404574e-05,
      "loss": 1.9527,
      "step": 204191
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.426113486289978,
      "learning_rate": 1.895633245588034e-05,
      "loss": 2.0321,
      "step": 204192
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.5616785287857056,
      "learning_rate": 1.895490146996631e-05,
      "loss": 1.9835,
      "step": 204193
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.5310750007629395,
      "learning_rate": 1.895347053630385e-05,
      "loss": 2.0921,
      "step": 204194
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.4933756589889526,
      "learning_rate": 1.8952039654893258e-05,
      "loss": 1.717,
      "step": 204195
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.509851336479187,
      "learning_rate": 1.89506088257348e-05,
      "loss": 1.9482,
      "step": 204196
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.4454456567764282,
      "learning_rate": 1.894917804882875e-05,
      "loss": 2.0086,
      "step": 204197
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.348544955253601,
      "learning_rate": 1.8947747324175267e-05,
      "loss": 1.808,
      "step": 204198
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.441529631614685,
      "learning_rate": 1.8946316651774786e-05,
      "loss": 2.0122,
      "step": 204199
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.3754377365112305,
      "learning_rate": 1.8944886031627404e-05,
      "loss": 1.9555,
      "step": 204200
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.4777313470840454,
      "learning_rate": 1.894345546373356e-05,
      "loss": 1.8084,
      "step": 204201
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.5250296592712402,
      "learning_rate": 1.8942024948093415e-05,
      "loss": 1.9187,
      "step": 204202
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.5690566301345825,
      "learning_rate": 1.8940594484707238e-05,
      "loss": 2.0819,
      "step": 204203
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.4884319305419922,
      "learning_rate": 1.8939164073575296e-05,
      "loss": 1.9516,
      "step": 204204
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.5088039636611938,
      "learning_rate": 1.893773371469789e-05,
      "loss": 1.8808,
      "step": 204205
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.35346257686615,
      "learning_rate": 1.8936303408075214e-05,
      "loss": 2.0038,
      "step": 204206
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.3855538368225098,
      "learning_rate": 1.893487315370764e-05,
      "loss": 1.8697,
      "step": 204207
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.613805890083313,
      "learning_rate": 1.8933442951595403e-05,
      "loss": 1.9597,
      "step": 204208
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.4064475297927856,
      "learning_rate": 1.8932012801738694e-05,
      "loss": 1.8776,
      "step": 204209
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.5343066453933716,
      "learning_rate": 1.893058270413782e-05,
      "loss": 1.8664,
      "step": 204210
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.5602391958236694,
      "learning_rate": 1.8929152658793112e-05,
      "loss": 1.8264,
      "step": 204211
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.5585339069366455,
      "learning_rate": 1.8927722665704703e-05,
      "loss": 2.0211,
      "step": 204212
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.4379355907440186,
      "learning_rate": 1.8926292724872992e-05,
      "loss": 2.024,
      "step": 204213
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.3410552740097046,
      "learning_rate": 1.8924862836298182e-05,
      "loss": 2.0073,
      "step": 204214
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.4357812404632568,
      "learning_rate": 1.892343299998057e-05,
      "loss": 1.9019,
      "step": 204215
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.4597879648208618,
      "learning_rate": 1.8922003215920324e-05,
      "loss": 1.9623,
      "step": 204216
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.3903131484985352,
      "learning_rate": 1.892057348411784e-05,
      "loss": 2.0945,
      "step": 204217
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.4982757568359375,
      "learning_rate": 1.891914380457329e-05,
      "loss": 2.0604,
      "step": 204218
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.4958595037460327,
      "learning_rate": 1.8917714177287037e-05,
      "loss": 2.1053,
      "step": 204219
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.6619999408721924,
      "learning_rate": 1.8916284602259247e-05,
      "loss": 1.9633,
      "step": 204220
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.3944228887557983,
      "learning_rate": 1.8914855079490255e-05,
      "loss": 2.2365,
      "step": 204221
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.3770091533660889,
      "learning_rate": 1.8913425608980225e-05,
      "loss": 2.1755,
      "step": 204222
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.3775752782821655,
      "learning_rate": 1.891199619072956e-05,
      "loss": 1.8565,
      "step": 204223
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.5226351022720337,
      "learning_rate": 1.8910566824738425e-05,
      "loss": 2.1179,
      "step": 204224
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.437927007675171,
      "learning_rate": 1.8909137511007157e-05,
      "loss": 1.7756,
      "step": 204225
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.3828437328338623,
      "learning_rate": 1.890770824953598e-05,
      "loss": 1.8303,
      "step": 204226
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.5110868215560913,
      "learning_rate": 1.8906279040325167e-05,
      "loss": 2.0988,
      "step": 204227
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.4784367084503174,
      "learning_rate": 1.890484988337495e-05,
      "loss": 2.0295,
      "step": 204228
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.632878065109253,
      "learning_rate": 1.8903420778685663e-05,
      "loss": 2.0974,
      "step": 204229
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.4781825542449951,
      "learning_rate": 1.8901991726257472e-05,
      "loss": 1.9651,
      "step": 204230
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.7640316486358643,
      "learning_rate": 1.8900562726090774e-05,
      "loss": 2.0303,
      "step": 204231
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.589263916015625,
      "learning_rate": 1.889913377818577e-05,
      "loss": 2.1163,
      "step": 204232
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.4298027753829956,
      "learning_rate": 1.889770488254273e-05,
      "loss": 2.0401,
      "step": 204233
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.4217344522476196,
      "learning_rate": 1.8896276039161852e-05,
      "loss": 1.8934,
      "step": 204234
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.4743452072143555,
      "learning_rate": 1.8894847248043498e-05,
      "loss": 2.1139,
      "step": 204235
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.4694461822509766,
      "learning_rate": 1.8893418509187873e-05,
      "loss": 1.7627,
      "step": 204236
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.7372519969940186,
      "learning_rate": 1.889198982259531e-05,
      "loss": 1.8435,
      "step": 204237
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.376352310180664,
      "learning_rate": 1.889056118826604e-05,
      "loss": 1.7677,
      "step": 204238
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.5877999067306519,
      "learning_rate": 1.8889132606200297e-05,
      "loss": 1.9219,
      "step": 204239
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.4311907291412354,
      "learning_rate": 1.8887704076398346e-05,
      "loss": 1.9974,
      "step": 204240
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.4353163242340088,
      "learning_rate": 1.888627559886049e-05,
      "loss": 2.0201,
      "step": 204241
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.5254011154174805,
      "learning_rate": 1.888484717358696e-05,
      "loss": 1.9221,
      "step": 204242
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.5380690097808838,
      "learning_rate": 1.8883418800578122e-05,
      "loss": 2.0849,
      "step": 204243
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.4228383302688599,
      "learning_rate": 1.8881990479834108e-05,
      "loss": 1.9146,
      "step": 204244
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.5551741123199463,
      "learning_rate": 1.8880562211355255e-05,
      "loss": 2.0301,
      "step": 204245
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.4626556634902954,
      "learning_rate": 1.887913399514179e-05,
      "loss": 1.9158,
      "step": 204246
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.5776281356811523,
      "learning_rate": 1.887770583119402e-05,
      "loss": 1.8818,
      "step": 204247
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.4269112348556519,
      "learning_rate": 1.8876277719512145e-05,
      "loss": 2.1485,
      "step": 204248
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.5589406490325928,
      "learning_rate": 1.887484966009656e-05,
      "loss": 1.8335,
      "step": 204249
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.43856942653656,
      "learning_rate": 1.8873421652947364e-05,
      "loss": 2.0084,
      "step": 204250
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.502752661705017,
      "learning_rate": 1.887199369806499e-05,
      "loss": 2.2343,
      "step": 204251
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.4706982374191284,
      "learning_rate": 1.8870565795449545e-05,
      "loss": 2.0177,
      "step": 204252
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.5423617362976074,
      "learning_rate": 1.886913794510142e-05,
      "loss": 1.9631,
      "step": 204253
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.486637830734253,
      "learning_rate": 1.8867710147020786e-05,
      "loss": 1.8297,
      "step": 204254
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.4722909927368164,
      "learning_rate": 1.8866282401207977e-05,
      "loss": 2.0754,
      "step": 204255
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.6545536518096924,
      "learning_rate": 1.886485470766319e-05,
      "loss": 2.1107,
      "step": 204256
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.5641703605651855,
      "learning_rate": 1.8863427066386827e-05,
      "loss": 1.9478,
      "step": 204257
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.4866480827331543,
      "learning_rate": 1.8861999477378988e-05,
      "loss": 1.9081,
      "step": 204258
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.4565553665161133,
      "learning_rate": 1.8860571940640033e-05,
      "loss": 2.1677,
      "step": 204259
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.5046871900558472,
      "learning_rate": 1.885914445617017e-05,
      "loss": 1.9218,
      "step": 204260
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.5445152521133423,
      "learning_rate": 1.8857717023969766e-05,
      "loss": 1.864,
      "step": 204261
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.5019721984863281,
      "learning_rate": 1.8856289644038948e-05,
      "loss": 1.9898,
      "step": 204262
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.426141619682312,
      "learning_rate": 1.8854862316378082e-05,
      "loss": 2.1041,
      "step": 204263
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.5961185693740845,
      "learning_rate": 1.885343504098744e-05,
      "loss": 2.0195,
      "step": 204264
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.6010600328445435,
      "learning_rate": 1.8852007817867222e-05,
      "loss": 1.9888,
      "step": 204265
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.6969860792160034,
      "learning_rate": 1.8850580647017688e-05,
      "loss": 2.0069,
      "step": 204266
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.4913520812988281,
      "learning_rate": 1.8849153528439177e-05,
      "loss": 2.0532,
      "step": 204267
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.416422963142395,
      "learning_rate": 1.8847726462131886e-05,
      "loss": 1.8581,
      "step": 204268
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.371549129486084,
      "learning_rate": 1.8846299448096147e-05,
      "loss": 1.9463,
      "step": 204269
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.464593529701233,
      "learning_rate": 1.8844872486332196e-05,
      "loss": 1.9392,
      "step": 204270
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.4335148334503174,
      "learning_rate": 1.884344557684023e-05,
      "loss": 1.8348,
      "step": 204271
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.524141788482666,
      "learning_rate": 1.884201871962062e-05,
      "loss": 1.9885,
      "step": 204272
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.500958800315857,
      "learning_rate": 1.8840591914673596e-05,
      "loss": 2.0468,
      "step": 204273
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.466756820678711,
      "learning_rate": 1.883916516199936e-05,
      "loss": 1.9347,
      "step": 204274
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.3600490093231201,
      "learning_rate": 1.8837738461598273e-05,
      "loss": 1.8506,
      "step": 204275
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.3527021408081055,
      "learning_rate": 1.8836311813470572e-05,
      "loss": 1.8694,
      "step": 204276
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.2931349277496338,
      "learning_rate": 1.8834885217616456e-05,
      "loss": 2.1017,
      "step": 204277
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.4122002124786377,
      "learning_rate": 1.8833458674036294e-05,
      "loss": 2.0755,
      "step": 204278
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.2837941646575928,
      "learning_rate": 1.8832032182730285e-05,
      "loss": 1.8832,
      "step": 204279
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.5482375621795654,
      "learning_rate": 1.883060574369869e-05,
      "loss": 1.9667,
      "step": 204280
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.3533000946044922,
      "learning_rate": 1.8829179356941814e-05,
      "loss": 2.0653,
      "step": 204281
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.5007635354995728,
      "learning_rate": 1.8827753022459923e-05,
      "loss": 1.991,
      "step": 204282
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.3643878698349,
      "learning_rate": 1.8826326740253183e-05,
      "loss": 1.948,
      "step": 204283
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.3989405632019043,
      "learning_rate": 1.8824900510321996e-05,
      "loss": 1.9431,
      "step": 204284
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.3486604690551758,
      "learning_rate": 1.882347433266652e-05,
      "loss": 1.7456,
      "step": 204285
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.3655314445495605,
      "learning_rate": 1.882204820728713e-05,
      "loss": 2.0694,
      "step": 204286
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.4324078559875488,
      "learning_rate": 1.8820622134184026e-05,
      "loss": 2.005,
      "step": 204287
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.3821901082992554,
      "learning_rate": 1.8819196113357437e-05,
      "loss": 2.0636,
      "step": 204288
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.5222222805023193,
      "learning_rate": 1.8817770144807664e-05,
      "loss": 2.018,
      "step": 204289
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.8355305194854736,
      "learning_rate": 1.8816344228535007e-05,
      "loss": 2.0972,
      "step": 204290
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.3455314636230469,
      "learning_rate": 1.8814918364539668e-05,
      "loss": 2.1074,
      "step": 204291
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.4442771673202515,
      "learning_rate": 1.8813492552821972e-05,
      "loss": 2.093,
      "step": 204292
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.689213514328003,
      "learning_rate": 1.8812066793382164e-05,
      "loss": 2.0129,
      "step": 204293
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.4375413656234741,
      "learning_rate": 1.8810641086220468e-05,
      "loss": 2.0864,
      "step": 204294
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.4468433856964111,
      "learning_rate": 1.8809215431337187e-05,
      "loss": 1.8618,
      "step": 204295
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.4004971981048584,
      "learning_rate": 1.880778982873259e-05,
      "loss": 1.9811,
      "step": 204296
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.5745172500610352,
      "learning_rate": 1.8806364278406904e-05,
      "loss": 2.029,
      "step": 204297
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.488262414932251,
      "learning_rate": 1.8804938780360467e-05,
      "loss": 2.1269,
      "step": 204298
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.429249882698059,
      "learning_rate": 1.880351333459351e-05,
      "loss": 1.9877,
      "step": 204299
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.4460524320602417,
      "learning_rate": 1.8802087941106268e-05,
      "loss": 1.7152,
      "step": 204300
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.4056988954544067,
      "learning_rate": 1.8800662599898976e-05,
      "loss": 1.9437,
      "step": 204301
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.4729896783828735,
      "learning_rate": 1.8799237310972025e-05,
      "loss": 2.1964,
      "step": 204302
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.5349297523498535,
      "learning_rate": 1.8797812074325526e-05,
      "loss": 2.0605,
      "step": 204303
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.4151629209518433,
      "learning_rate": 1.879638688995987e-05,
      "loss": 1.9671,
      "step": 204304
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.3671499490737915,
      "learning_rate": 1.8794961757875295e-05,
      "loss": 1.9304,
      "step": 204305
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.3320443630218506,
      "learning_rate": 1.879353667807203e-05,
      "loss": 1.957,
      "step": 204306
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.3138548135757446,
      "learning_rate": 1.8792111650550312e-05,
      "loss": 1.973,
      "step": 204307
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.476629614830017,
      "learning_rate": 1.8790686675310507e-05,
      "loss": 2.0824,
      "step": 204308
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.3115415573120117,
      "learning_rate": 1.8789261752352746e-05,
      "loss": 1.935,
      "step": 204309
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.4819490909576416,
      "learning_rate": 1.878783688167743e-05,
      "loss": 1.8888,
      "step": 204310
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.4836630821228027,
      "learning_rate": 1.8786412063284796e-05,
      "loss": 2.134,
      "step": 204311
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.3651008605957031,
      "learning_rate": 1.8784987297175037e-05,
      "loss": 2.1312,
      "step": 204312
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.5779426097869873,
      "learning_rate": 1.8783562583348423e-05,
      "loss": 2.1507,
      "step": 204313
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.4770723581314087,
      "learning_rate": 1.8782137921805284e-05,
      "loss": 2.0157,
      "step": 204314
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.4566856622695923,
      "learning_rate": 1.8780713312545827e-05,
      "loss": 2.0122,
      "step": 204315
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.4006237983703613,
      "learning_rate": 1.8779288755570376e-05,
      "loss": 2.0751,
      "step": 204316
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.5710065364837646,
      "learning_rate": 1.8777864250879138e-05,
      "loss": 1.9359,
      "step": 204317
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.4994804859161377,
      "learning_rate": 1.8776439798472476e-05,
      "loss": 2.0251,
      "step": 204318
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.4975394010543823,
      "learning_rate": 1.8775015398350525e-05,
      "loss": 1.8761,
      "step": 204319
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.5359302759170532,
      "learning_rate": 1.8773591050513614e-05,
      "loss": 2.0427,
      "step": 204320
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.692700982093811,
      "learning_rate": 1.8772166754961946e-05,
      "loss": 2.1353,
      "step": 204321
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.5929450988769531,
      "learning_rate": 1.8770742511695925e-05,
      "loss": 2.0865,
      "step": 204322
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.69710373878479,
      "learning_rate": 1.8769318320715675e-05,
      "loss": 2.1377,
      "step": 204323
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.36210298538208,
      "learning_rate": 1.8767894182021604e-05,
      "loss": 1.7859,
      "step": 204324
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.5282137393951416,
      "learning_rate": 1.8766470095613806e-05,
      "loss": 1.9394,
      "step": 204325
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.5523101091384888,
      "learning_rate": 1.876504606149265e-05,
      "loss": 2.0191,
      "step": 204326
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.6534756422042847,
      "learning_rate": 1.8763622079658368e-05,
      "loss": 1.8174,
      "step": 204327
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.5830392837524414,
      "learning_rate": 1.8762198150111263e-05,
      "loss": 1.8783,
      "step": 204328
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.5156729221343994,
      "learning_rate": 1.8760774272851532e-05,
      "loss": 1.8332,
      "step": 204329
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.9258464574813843,
      "learning_rate": 1.8759350447879572e-05,
      "loss": 1.8983,
      "step": 204330
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.6274845600128174,
      "learning_rate": 1.8757926675195454e-05,
      "loss": 2.1762,
      "step": 204331
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.5533829927444458,
      "learning_rate": 1.8756502954799612e-05,
      "loss": 1.9055,
      "step": 204332
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.536717414855957,
      "learning_rate": 1.8755079286692208e-05,
      "loss": 1.9688,
      "step": 204333
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.473336100578308,
      "learning_rate": 1.8753655670873547e-05,
      "loss": 1.8371,
      "step": 204334
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.3687454462051392,
      "learning_rate": 1.875223210734389e-05,
      "loss": 1.9957,
      "step": 204335
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.6030036211013794,
      "learning_rate": 1.8750808596103572e-05,
      "loss": 1.9471,
      "step": 204336
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.3426748514175415,
      "learning_rate": 1.8749385137152696e-05,
      "loss": 1.8071,
      "step": 204337
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.7382258176803589,
      "learning_rate": 1.8747961730491657e-05,
      "loss": 2.023,
      "step": 204338
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.419745683670044,
      "learning_rate": 1.8746538376120655e-05,
      "loss": 1.9684,
      "step": 204339
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.45057213306427,
      "learning_rate": 1.874511507404003e-05,
      "loss": 1.8727,
      "step": 204340
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.2823737859725952,
      "learning_rate": 1.874369182424994e-05,
      "loss": 1.9051,
      "step": 204341
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.4653609991073608,
      "learning_rate": 1.874226862675079e-05,
      "loss": 1.921,
      "step": 204342
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.3413468599319458,
      "learning_rate": 1.8740845481542677e-05,
      "loss": 1.7316,
      "step": 204343
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.4583748579025269,
      "learning_rate": 1.8739422388626003e-05,
      "loss": 1.9936,
      "step": 204344
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.5715043544769287,
      "learning_rate": 1.873799934800093e-05,
      "loss": 1.9222,
      "step": 204345
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.5726044178009033,
      "learning_rate": 1.8736576359667798e-05,
      "loss": 1.9149,
      "step": 204346
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.4024568796157837,
      "learning_rate": 1.8735153423626836e-05,
      "loss": 1.7812,
      "step": 204347
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.419482946395874,
      "learning_rate": 1.8733730539878345e-05,
      "loss": 2.0482,
      "step": 204348
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.4308513402938843,
      "learning_rate": 1.8732307708422557e-05,
      "loss": 2.0255,
      "step": 204349
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.4451788663864136,
      "learning_rate": 1.8730884929259738e-05,
      "loss": 2.0079,
      "step": 204350
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.3827409744262695,
      "learning_rate": 1.872946220239012e-05,
      "loss": 1.8364,
      "step": 204351
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.5686285495758057,
      "learning_rate": 1.8728039527814044e-05,
      "loss": 1.8846,
      "step": 204352
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.4632689952850342,
      "learning_rate": 1.87266169055317e-05,
      "loss": 2.0242,
      "step": 204353
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.50791335105896,
      "learning_rate": 1.8725194335543427e-05,
      "loss": 2.0202,
      "step": 204354
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.592455506324768,
      "learning_rate": 1.8723771817849453e-05,
      "loss": 1.8638,
      "step": 204355
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.319779872894287,
      "learning_rate": 1.8722349352450016e-05,
      "loss": 2.0661,
      "step": 204356
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.5598634481430054,
      "learning_rate": 1.8720926939345417e-05,
      "loss": 1.9765,
      "step": 204357
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.4577819108963013,
      "learning_rate": 1.8719504578535914e-05,
      "loss": 1.8657,
      "step": 204358
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.741380214691162,
      "learning_rate": 1.8718082270021717e-05,
      "loss": 1.9911,
      "step": 204359
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.6009362936019897,
      "learning_rate": 1.8716660013803185e-05,
      "loss": 1.8729,
      "step": 204360
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.346665382385254,
      "learning_rate": 1.8715237809880555e-05,
      "loss": 1.9279,
      "step": 204361
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.5719947814941406,
      "learning_rate": 1.8713815658253994e-05,
      "loss": 1.8491,
      "step": 204362
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.3113553524017334,
      "learning_rate": 1.87123935589239e-05,
      "loss": 1.977,
      "step": 204363
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.418209195137024,
      "learning_rate": 1.8710971511890506e-05,
      "loss": 1.8356,
      "step": 204364
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.4794739484786987,
      "learning_rate": 1.8709549517153975e-05,
      "loss": 1.8458,
      "step": 204365
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.378064751625061,
      "learning_rate": 1.8708127574714716e-05,
      "loss": 2.0414,
      "step": 204366
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.560133695602417,
      "learning_rate": 1.870670568457292e-05,
      "loss": 2.2192,
      "step": 204367
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.5215016603469849,
      "learning_rate": 1.870528384672879e-05,
      "loss": 2.1362,
      "step": 204368
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.5497992038726807,
      "learning_rate": 1.870386206118273e-05,
      "loss": 1.8857,
      "step": 204369
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.45452082157135,
      "learning_rate": 1.87024403279349e-05,
      "loss": 1.8791,
      "step": 204370
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.657191514968872,
      "learning_rate": 1.87010186469856e-05,
      "loss": 2.1571,
      "step": 204371
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.384959101676941,
      "learning_rate": 1.86995970183351e-05,
      "loss": 1.8333,
      "step": 204372
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.381121039390564,
      "learning_rate": 1.8698175441983664e-05,
      "loss": 2.1695,
      "step": 204373
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.5608267784118652,
      "learning_rate": 1.8696753917931528e-05,
      "loss": 2.0419,
      "step": 204374
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.6443301439285278,
      "learning_rate": 1.869533244617899e-05,
      "loss": 1.9484,
      "step": 204375
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.477116584777832,
      "learning_rate": 1.869391102672625e-05,
      "loss": 2.0659,
      "step": 204376
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.472310185432434,
      "learning_rate": 1.8692489659573674e-05,
      "loss": 2.0245,
      "step": 204377
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.8481780290603638,
      "learning_rate": 1.8691068344721495e-05,
      "loss": 2.0512,
      "step": 204378
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.4361090660095215,
      "learning_rate": 1.8689647082169945e-05,
      "loss": 1.9469,
      "step": 204379
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.4089750051498413,
      "learning_rate": 1.8688225871919226e-05,
      "loss": 1.8984,
      "step": 204380
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.588308334350586,
      "learning_rate": 1.8686804713969773e-05,
      "loss": 2.0424,
      "step": 204381
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.3294297456741333,
      "learning_rate": 1.868538360832168e-05,
      "loss": 1.9266,
      "step": 204382
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.5177417993545532,
      "learning_rate": 1.8683962554975318e-05,
      "loss": 1.9374,
      "step": 204383
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.417287826538086,
      "learning_rate": 1.8682541553930886e-05,
      "loss": 1.9394,
      "step": 204384
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.608890175819397,
      "learning_rate": 1.868112060518875e-05,
      "loss": 2.0751,
      "step": 204385
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.4720419645309448,
      "learning_rate": 1.867969970874904e-05,
      "loss": 2.147,
      "step": 204386
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.3908636569976807,
      "learning_rate": 1.8678278864612127e-05,
      "loss": 1.9856,
      "step": 204387
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.532523274421692,
      "learning_rate": 1.8676858072778178e-05,
      "loss": 2.1354,
      "step": 204388
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.4883440732955933,
      "learning_rate": 1.8675437333247554e-05,
      "loss": 2.1063,
      "step": 204389
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.480419635772705,
      "learning_rate": 1.867401664602043e-05,
      "loss": 2.1312,
      "step": 204390
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.4878238439559937,
      "learning_rate": 1.867259601109723e-05,
      "loss": 2.0654,
      "step": 204391
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.4388293027877808,
      "learning_rate": 1.8671175428477993e-05,
      "loss": 1.9895,
      "step": 204392
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.5837645530700684,
      "learning_rate": 1.866975489816315e-05,
      "loss": 1.9886,
      "step": 204393
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.6378939151763916,
      "learning_rate": 1.8668334420152864e-05,
      "loss": 2.1214,
      "step": 204394
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.3793480396270752,
      "learning_rate": 1.8666913994447473e-05,
      "loss": 2.1041,
      "step": 204395
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.4544713497161865,
      "learning_rate": 1.866549362104718e-05,
      "loss": 1.9098,
      "step": 204396
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.441880226135254,
      "learning_rate": 1.8664073299952377e-05,
      "loss": 2.0629,
      "step": 204397
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.379870057106018,
      "learning_rate": 1.8662653031163133e-05,
      "loss": 1.8921,
      "step": 204398
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.4116297960281372,
      "learning_rate": 1.8661232814679885e-05,
      "loss": 1.9866,
      "step": 204399
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.4948064088821411,
      "learning_rate": 1.865981265050276e-05,
      "loss": 2.0743,
      "step": 204400
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.4296483993530273,
      "learning_rate": 1.8658392538632126e-05,
      "loss": 2.2483,
      "step": 204401
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.6199654340744019,
      "learning_rate": 1.8656972479068155e-05,
      "loss": 1.7653,
      "step": 204402
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.4077379703521729,
      "learning_rate": 1.8655552471811274e-05,
      "loss": 1.9851,
      "step": 204403
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.563977837562561,
      "learning_rate": 1.8654132516861553e-05,
      "loss": 2.2983,
      "step": 204404
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.4157061576843262,
      "learning_rate": 1.8652712614219357e-05,
      "loss": 2.3126,
      "step": 204405
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.3806308507919312,
      "learning_rate": 1.8651292763884917e-05,
      "loss": 1.931,
      "step": 204406
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.573386549949646,
      "learning_rate": 1.8649872965858537e-05,
      "loss": 1.8071,
      "step": 204407
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.486673355102539,
      "learning_rate": 1.8648453220140415e-05,
      "loss": 1.892,
      "step": 204408
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.4340777397155762,
      "learning_rate": 1.8647033526730948e-05,
      "loss": 2.1367,
      "step": 204409
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.4166196584701538,
      "learning_rate": 1.8645613885630208e-05,
      "loss": 1.983,
      "step": 204410
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.4891951084136963,
      "learning_rate": 1.864419429683862e-05,
      "loss": 1.9171,
      "step": 204411
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.247127890586853,
      "learning_rate": 1.8642774760356326e-05,
      "loss": 2.0368,
      "step": 204412
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.483591079711914,
      "learning_rate": 1.8641355276183723e-05,
      "loss": 2.0454,
      "step": 204413
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.332107663154602,
      "learning_rate": 1.863993584432094e-05,
      "loss": 2.1929,
      "step": 204414
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.3631412982940674,
      "learning_rate": 1.8638516464768382e-05,
      "loss": 1.8834,
      "step": 204415
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.5427331924438477,
      "learning_rate": 1.863709713752618e-05,
      "loss": 2.1742,
      "step": 204416
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.4448156356811523,
      "learning_rate": 1.8635677862594666e-05,
      "loss": 1.7465,
      "step": 204417
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.4861449003219604,
      "learning_rate": 1.8634258639974043e-05,
      "loss": 1.9909,
      "step": 204418
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.6169123649597168,
      "learning_rate": 1.8632839469664708e-05,
      "loss": 2.2506,
      "step": 204419
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.5216431617736816,
      "learning_rate": 1.863142035166676e-05,
      "loss": 2.0474,
      "step": 204420
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.417296051979065,
      "learning_rate": 1.8630001285980633e-05,
      "loss": 1.8857,
      "step": 204421
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.3973288536071777,
      "learning_rate": 1.8628582272606428e-05,
      "loss": 2.087,
      "step": 204422
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.40035879611969,
      "learning_rate": 1.8627163311544514e-05,
      "loss": 2.1423,
      "step": 204423
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.4110430479049683,
      "learning_rate": 1.862574440279505e-05,
      "loss": 1.9021,
      "step": 204424
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.3760520219802856,
      "learning_rate": 1.8624325546358442e-05,
      "loss": 1.8128,
      "step": 204425
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.464178442955017,
      "learning_rate": 1.8622906742234857e-05,
      "loss": 2.1554,
      "step": 204426
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.3492103815078735,
      "learning_rate": 1.8621487990424653e-05,
      "loss": 1.9323,
      "step": 204427
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.4388941526412964,
      "learning_rate": 1.862006929092794e-05,
      "loss": 2.0635,
      "step": 204428
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.4161344766616821,
      "learning_rate": 1.8618650643745113e-05,
      "loss": 2.0543,
      "step": 204429
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.4111900329589844,
      "learning_rate": 1.861723204887634e-05,
      "loss": 1.9617,
      "step": 204430
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.4422268867492676,
      "learning_rate": 1.8615813506321985e-05,
      "loss": 1.9167,
      "step": 204431
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.4611015319824219,
      "learning_rate": 1.8614395016082217e-05,
      "loss": 1.7399,
      "step": 204432
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.5566976070404053,
      "learning_rate": 1.8612976578157402e-05,
      "loss": 1.842,
      "step": 204433
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.4244153499603271,
      "learning_rate": 1.8611558192547738e-05,
      "loss": 1.9004,
      "step": 204434
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.39336097240448,
      "learning_rate": 1.861013985925349e-05,
      "loss": 1.9572,
      "step": 204435
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.5685994625091553,
      "learning_rate": 1.86087215782749e-05,
      "loss": 2.046,
      "step": 204436
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.465942144393921,
      "learning_rate": 1.860730334961229e-05,
      "loss": 2.0539,
      "step": 204437
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.4148833751678467,
      "learning_rate": 1.8605885173265834e-05,
      "loss": 2.0631,
      "step": 204438
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.4079029560089111,
      "learning_rate": 1.8604467049235925e-05,
      "loss": 1.8438,
      "step": 204439
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.5593287944793701,
      "learning_rate": 1.860304897752274e-05,
      "loss": 2.0849,
      "step": 204440
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.4723803997039795,
      "learning_rate": 1.8601630958126567e-05,
      "loss": 2.1159,
      "step": 204441
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.5031288862228394,
      "learning_rate": 1.860021299104765e-05,
      "loss": 1.9381,
      "step": 204442
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.3836441040039062,
      "learning_rate": 1.8598795076286276e-05,
      "loss": 1.7963,
      "step": 204443
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.5532113313674927,
      "learning_rate": 1.8597377213842656e-05,
      "loss": 2.1738,
      "step": 204444
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.549869418144226,
      "learning_rate": 1.859595940371712e-05,
      "loss": 2.0203,
      "step": 204445
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.4944292306900024,
      "learning_rate": 1.8594541645909933e-05,
      "loss": 1.8883,
      "step": 204446
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.365860939025879,
      "learning_rate": 1.8593123940421294e-05,
      "loss": 1.8881,
      "step": 204447
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.676840901374817,
      "learning_rate": 1.859170628725154e-05,
      "loss": 1.9849,
      "step": 204448
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.561218500137329,
      "learning_rate": 1.8590288686400902e-05,
      "loss": 1.8385,
      "step": 204449
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.5731494426727295,
      "learning_rate": 1.8588871137869576e-05,
      "loss": 2.0789,
      "step": 204450
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.4255191087722778,
      "learning_rate": 1.8587453641657967e-05,
      "loss": 2.0375,
      "step": 204451
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.3118951320648193,
      "learning_rate": 1.858603619776624e-05,
      "loss": 1.9313,
      "step": 204452
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.398729920387268,
      "learning_rate": 1.858461880619466e-05,
      "loss": 1.975,
      "step": 204453
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.5593969821929932,
      "learning_rate": 1.8583201466943532e-05,
      "loss": 2.0941,
      "step": 204454
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.4150997400283813,
      "learning_rate": 1.858178418001308e-05,
      "loss": 2.1053,
      "step": 204455
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.4929425716400146,
      "learning_rate": 1.8580366945403613e-05,
      "loss": 2.1666,
      "step": 204456
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.4919755458831787,
      "learning_rate": 1.8578949763115357e-05,
      "loss": 2.0666,
      "step": 204457
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.460872769355774,
      "learning_rate": 1.8577532633148618e-05,
      "loss": 1.8652,
      "step": 204458
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.4606587886810303,
      "learning_rate": 1.8576115555503557e-05,
      "loss": 2.0277,
      "step": 204459
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.5885025262832642,
      "learning_rate": 1.8574698530180542e-05,
      "loss": 2.054,
      "step": 204460
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.466503620147705,
      "learning_rate": 1.8573281557179776e-05,
      "loss": 1.7913,
      "step": 204461
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.4736465215682983,
      "learning_rate": 1.857186463650162e-05,
      "loss": 2.1418,
      "step": 204462
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.448587417602539,
      "learning_rate": 1.8570447768146248e-05,
      "loss": 1.9256,
      "step": 204463
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.4083986282348633,
      "learning_rate": 1.8569030952113917e-05,
      "loss": 1.9751,
      "step": 204464
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.5264214277267456,
      "learning_rate": 1.85676141884049e-05,
      "loss": 2.0731,
      "step": 204465
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.410022497177124,
      "learning_rate": 1.8566197477019496e-05,
      "loss": 2.0395,
      "step": 204466
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.4482530355453491,
      "learning_rate": 1.8564780817957936e-05,
      "loss": 2.1228,
      "step": 204467
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.3342361450195312,
      "learning_rate": 1.856336421122052e-05,
      "loss": 1.938,
      "step": 204468
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.4894683361053467,
      "learning_rate": 1.856194765680745e-05,
      "loss": 1.9298,
      "step": 204469
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.3919875621795654,
      "learning_rate": 1.8560531154719093e-05,
      "loss": 1.7242,
      "step": 204470
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.410939335823059,
      "learning_rate": 1.855911470495558e-05,
      "loss": 1.975,
      "step": 204471
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.5273568630218506,
      "learning_rate": 1.8557698307517278e-05,
      "loss": 1.9161,
      "step": 204472
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.4357393980026245,
      "learning_rate": 1.8556281962404383e-05,
      "loss": 1.916,
      "step": 204473
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.4401240348815918,
      "learning_rate": 1.8554865669617202e-05,
      "loss": 1.7743,
      "step": 204474
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.4942690134048462,
      "learning_rate": 1.8553449429155963e-05,
      "loss": 1.8715,
      "step": 204475
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.3862265348434448,
      "learning_rate": 1.8552033241021035e-05,
      "loss": 1.9944,
      "step": 204476
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.3652667999267578,
      "learning_rate": 1.855061710521252e-05,
      "loss": 1.8068,
      "step": 204477
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.4340096712112427,
      "learning_rate": 1.8549201021730774e-05,
      "loss": 1.8517,
      "step": 204478
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.4964228868484497,
      "learning_rate": 1.8547784990576007e-05,
      "loss": 1.8413,
      "step": 204479
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.4320462942123413,
      "learning_rate": 1.8546369011748585e-05,
      "loss": 1.9932,
      "step": 204480
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.4489496946334839,
      "learning_rate": 1.8544953085248636e-05,
      "loss": 1.8502,
      "step": 204481
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.5670546293258667,
      "learning_rate": 1.8543537211076596e-05,
      "loss": 1.8824,
      "step": 204482
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.3645009994506836,
      "learning_rate": 1.8542121389232535e-05,
      "loss": 2.0889,
      "step": 204483
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.36500883102417,
      "learning_rate": 1.8540705619716844e-05,
      "loss": 2.3105,
      "step": 204484
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.5186563730239868,
      "learning_rate": 1.853928990252973e-05,
      "loss": 2.1447,
      "step": 204485
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.4609562158584595,
      "learning_rate": 1.8537874237671492e-05,
      "loss": 2.1511,
      "step": 204486
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.4667333364486694,
      "learning_rate": 1.8536458625142326e-05,
      "loss": 1.9317,
      "step": 204487
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.3376634120941162,
      "learning_rate": 1.853504306494267e-05,
      "loss": 2.113,
      "step": 204488
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.5115634202957153,
      "learning_rate": 1.853362755707255e-05,
      "loss": 1.9654,
      "step": 204489
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.5225781202316284,
      "learning_rate": 1.8532212101532374e-05,
      "loss": 2.1222,
      "step": 204490
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.5586154460906982,
      "learning_rate": 1.8530796698322335e-05,
      "loss": 2.1002,
      "step": 204491
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.5221381187438965,
      "learning_rate": 1.8529381347442805e-05,
      "loss": 1.9154,
      "step": 204492
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.3601688146591187,
      "learning_rate": 1.8527966048893916e-05,
      "loss": 2.0471,
      "step": 204493
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.5090516805648804,
      "learning_rate": 1.8526550802676064e-05,
      "loss": 2.1839,
      "step": 204494
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.4531415700912476,
      "learning_rate": 1.8525135608789355e-05,
      "loss": 1.9544,
      "step": 204495
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.3886724710464478,
      "learning_rate": 1.8523720467234183e-05,
      "loss": 2.0645,
      "step": 204496
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.4045454263687134,
      "learning_rate": 1.8522305378010747e-05,
      "loss": 1.8972,
      "step": 204497
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.4199458360671997,
      "learning_rate": 1.8520890341119355e-05,
      "loss": 1.9738,
      "step": 204498
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.4549697637557983,
      "learning_rate": 1.85194753565602e-05,
      "loss": 1.9726,
      "step": 204499
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.5816110372543335,
      "learning_rate": 1.8518060424333646e-05,
      "loss": 2.0034,
      "step": 204500
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.4367797374725342,
      "learning_rate": 1.8516645544439834e-05,
      "loss": 2.2195,
      "step": 204501
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.4771320819854736,
      "learning_rate": 1.851523071687916e-05,
      "loss": 2.0423,
      "step": 204502
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.537601351737976,
      "learning_rate": 1.851381594165172e-05,
      "loss": 2.1148,
      "step": 204503
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.468551516532898,
      "learning_rate": 1.8512401218757956e-05,
      "loss": 1.7428,
      "step": 204504
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.4727073907852173,
      "learning_rate": 1.8510986548197993e-05,
      "loss": 1.8293,
      "step": 204505
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.430790901184082,
      "learning_rate": 1.8509571929972266e-05,
      "loss": 2.1372,
      "step": 204506
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.4725332260131836,
      "learning_rate": 1.850815736408081e-05,
      "loss": 1.9609,
      "step": 204507
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.3484792709350586,
      "learning_rate": 1.850674285052406e-05,
      "loss": 1.9081,
      "step": 204508
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.3878649473190308,
      "learning_rate": 1.8505328389302143e-05,
      "loss": 2.0727,
      "step": 204509
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.5721830129623413,
      "learning_rate": 1.8503913980415462e-05,
      "loss": 1.9826,
      "step": 204510
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.4023919105529785,
      "learning_rate": 1.8502499623864186e-05,
      "loss": 1.9405,
      "step": 204511
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.313848614692688,
      "learning_rate": 1.8501085319648713e-05,
      "loss": 1.9645,
      "step": 204512
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.4629311561584473,
      "learning_rate": 1.8499671067769072e-05,
      "loss": 1.7224,
      "step": 204513
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.676558017730713,
      "learning_rate": 1.8498256868225703e-05,
      "loss": 2.076,
      "step": 204514
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.3866339921951294,
      "learning_rate": 1.8496842721018802e-05,
      "loss": 2.0609,
      "step": 204515
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.4830849170684814,
      "learning_rate": 1.8495428626148702e-05,
      "loss": 1.8779,
      "step": 204516
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.3522368669509888,
      "learning_rate": 1.8494014583615534e-05,
      "loss": 1.8022,
      "step": 204517
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.429752230644226,
      "learning_rate": 1.8492600593419705e-05,
      "loss": 1.9001,
      "step": 204518
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.4838602542877197,
      "learning_rate": 1.849118665556144e-05,
      "loss": 1.8913,
      "step": 204519
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.5241998434066772,
      "learning_rate": 1.8489772770040946e-05,
      "loss": 1.9183,
      "step": 204520
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.5040249824523926,
      "learning_rate": 1.8488358936858483e-05,
      "loss": 2.081,
      "step": 204521
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.6654822826385498,
      "learning_rate": 1.848694515601439e-05,
      "loss": 2.1865,
      "step": 204522
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.3595304489135742,
      "learning_rate": 1.8485531427508827e-05,
      "loss": 2.1815,
      "step": 204523
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.7225497961044312,
      "learning_rate": 1.8484117751342197e-05,
      "loss": 2.1533,
      "step": 204524
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.4670058488845825,
      "learning_rate": 1.848270412751467e-05,
      "loss": 2.0118,
      "step": 204525
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.5272935628890991,
      "learning_rate": 1.8481290556026506e-05,
      "loss": 1.9243,
      "step": 204526
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.6584597826004028,
      "learning_rate": 1.8479877036877944e-05,
      "loss": 2.1077,
      "step": 204527
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.611403465270996,
      "learning_rate": 1.8478463570069347e-05,
      "loss": 2.0757,
      "step": 204528
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.611497402191162,
      "learning_rate": 1.847705015560088e-05,
      "loss": 1.7798,
      "step": 204529
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.4435486793518066,
      "learning_rate": 1.8475636793472847e-05,
      "loss": 2.0002,
      "step": 204530
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.4143675565719604,
      "learning_rate": 1.8474223483685547e-05,
      "loss": 2.0203,
      "step": 204531
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.475003957748413,
      "learning_rate": 1.8472810226239144e-05,
      "loss": 1.9709,
      "step": 204532
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.6292243003845215,
      "learning_rate": 1.8471397021134003e-05,
      "loss": 1.8781,
      "step": 204533
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.432307481765747,
      "learning_rate": 1.8469983868370362e-05,
      "loss": 2.0091,
      "step": 204534
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.517240285873413,
      "learning_rate": 1.8468570767948387e-05,
      "loss": 2.0212,
      "step": 204535
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.4481955766677856,
      "learning_rate": 1.8467157719868508e-05,
      "loss": 1.8361,
      "step": 204536
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.4245333671569824,
      "learning_rate": 1.846574472413086e-05,
      "loss": 2.0967,
      "step": 204537
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.3833985328674316,
      "learning_rate": 1.846433178073571e-05,
      "loss": 2.1947,
      "step": 204538
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.4486546516418457,
      "learning_rate": 1.846291888968342e-05,
      "loss": 2.0349,
      "step": 204539
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.3508915901184082,
      "learning_rate": 1.846150605097416e-05,
      "loss": 1.9454,
      "step": 204540
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.462800145149231,
      "learning_rate": 1.8460093264608167e-05,
      "loss": 1.7814,
      "step": 204541
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.6013693809509277,
      "learning_rate": 1.8458680530585835e-05,
      "loss": 1.9961,
      "step": 204542
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.4851038455963135,
      "learning_rate": 1.845726784890733e-05,
      "loss": 2.1562,
      "step": 204543
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.63111412525177,
      "learning_rate": 1.845585521957289e-05,
      "loss": 1.9993,
      "step": 204544
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.4534409046173096,
      "learning_rate": 1.845444264258288e-05,
      "loss": 2.0246,
      "step": 204545
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.4778096675872803,
      "learning_rate": 1.8453030117937462e-05,
      "loss": 1.9208,
      "step": 204546
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.4766978025436401,
      "learning_rate": 1.845161764563694e-05,
      "loss": 2.0483,
      "step": 204547
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.4522216320037842,
      "learning_rate": 1.8450205225681614e-05,
      "loss": 1.7209,
      "step": 204548
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.4737604856491089,
      "learning_rate": 1.8448792858071714e-05,
      "loss": 2.0836,
      "step": 204549
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.478542685508728,
      "learning_rate": 1.8447380542807445e-05,
      "loss": 1.9074,
      "step": 204550
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.349092960357666,
      "learning_rate": 1.8445968279889168e-05,
      "loss": 1.776,
      "step": 204551
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.4719064235687256,
      "learning_rate": 1.8444556069317052e-05,
      "loss": 2.0915,
      "step": 204552
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.4055649042129517,
      "learning_rate": 1.844314391109146e-05,
      "loss": 2.1505,
      "step": 204553
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.6008168458938599,
      "learning_rate": 1.8441731805212533e-05,
      "loss": 1.7955,
      "step": 204554
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.4333109855651855,
      "learning_rate": 1.8440319751680733e-05,
      "loss": 2.2071,
      "step": 204555
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.4937466382980347,
      "learning_rate": 1.8438907750496057e-05,
      "loss": 2.086,
      "step": 204556
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.420491337776184,
      "learning_rate": 1.8437495801658975e-05,
      "loss": 1.9127,
      "step": 204557
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.3851643800735474,
      "learning_rate": 1.8436083905169617e-05,
      "loss": 2.1415,
      "step": 204558
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.4282631874084473,
      "learning_rate": 1.8434672061028387e-05,
      "loss": 2.0741,
      "step": 204559
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.578981876373291,
      "learning_rate": 1.8433260269235384e-05,
      "loss": 1.9569,
      "step": 204560
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.5534727573394775,
      "learning_rate": 1.843184852979107e-05,
      "loss": 2.1271,
      "step": 204561
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.4680864810943604,
      "learning_rate": 1.8430436842695485e-05,
      "loss": 1.9629,
      "step": 204562
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.4489322900772095,
      "learning_rate": 1.8429025207949022e-05,
      "loss": 2.1698,
      "step": 204563
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.3697540760040283,
      "learning_rate": 1.842761362555192e-05,
      "loss": 1.808,
      "step": 204564
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.5147731304168701,
      "learning_rate": 1.8426202095504437e-05,
      "loss": 2.3526,
      "step": 204565
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.532751202583313,
      "learning_rate": 1.8424790617806817e-05,
      "loss": 1.8942,
      "step": 204566
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.3912739753723145,
      "learning_rate": 1.8423379192459455e-05,
      "loss": 2.0533,
      "step": 204567
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.411332368850708,
      "learning_rate": 1.842196781946238e-05,
      "loss": 1.7744,
      "step": 204568
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.5318129062652588,
      "learning_rate": 1.842055649881603e-05,
      "loss": 2.0993,
      "step": 204569
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.3482236862182617,
      "learning_rate": 1.841914523052057e-05,
      "loss": 1.9698,
      "step": 204570
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.4402977228164673,
      "learning_rate": 1.8417734014576334e-05,
      "loss": 2.072,
      "step": 204571
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.4102565050125122,
      "learning_rate": 1.841632285098352e-05,
      "loss": 2.075,
      "step": 204572
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.4372704029083252,
      "learning_rate": 1.8414911739742532e-05,
      "loss": 2.0363,
      "step": 204573
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.7569270133972168,
      "learning_rate": 1.8413500680853434e-05,
      "loss": 1.9008,
      "step": 204574
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.6351637840270996,
      "learning_rate": 1.841208967431662e-05,
      "loss": 2.0492,
      "step": 204575
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.5289287567138672,
      "learning_rate": 1.8410678720132266e-05,
      "loss": 1.6843,
      "step": 204576
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.4417619705200195,
      "learning_rate": 1.8409267818300732e-05,
      "loss": 1.8995,
      "step": 204577
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.4914085865020752,
      "learning_rate": 1.8407856968822187e-05,
      "loss": 1.8637,
      "step": 204578
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.4576079845428467,
      "learning_rate": 1.840644617169703e-05,
      "loss": 2.2132,
      "step": 204579
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.47231125831604,
      "learning_rate": 1.840503542692533e-05,
      "loss": 2.073,
      "step": 204580
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.5722861289978027,
      "learning_rate": 1.8403624734507517e-05,
      "loss": 1.9572,
      "step": 204581
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.5074846744537354,
      "learning_rate": 1.8402214094443723e-05,
      "loss": 2.2061,
      "step": 204582
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.4034255743026733,
      "learning_rate": 1.840080350673432e-05,
      "loss": 2.0116,
      "step": 204583
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.7539589405059814,
      "learning_rate": 1.8399392971379468e-05,
      "loss": 2.1003,
      "step": 204584
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.5275206565856934,
      "learning_rate": 1.8397982488379536e-05,
      "loss": 2.1297,
      "step": 204585
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.6847354173660278,
      "learning_rate": 1.839657205773476e-05,
      "loss": 1.9475,
      "step": 204586
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.5627632141113281,
      "learning_rate": 1.8395161679445337e-05,
      "loss": 1.94,
      "step": 204587
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.5135670900344849,
      "learning_rate": 1.839375135351153e-05,
      "loss": 1.9226,
      "step": 204588
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.3816404342651367,
      "learning_rate": 1.8392341079933715e-05,
      "loss": 2.2275,
      "step": 204589
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.4442291259765625,
      "learning_rate": 1.8390930858712016e-05,
      "loss": 2.1597,
      "step": 204590
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.3709028959274292,
      "learning_rate": 1.8389520689846802e-05,
      "loss": 1.8516,
      "step": 204591
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.431577205657959,
      "learning_rate": 1.8388110573338277e-05,
      "loss": 2.2418,
      "step": 204592
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.442944884300232,
      "learning_rate": 1.8386700509186736e-05,
      "loss": 1.7962,
      "step": 204593
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.4146497249603271,
      "learning_rate": 1.838529049739238e-05,
      "loss": 1.9366,
      "step": 204594
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.4705294370651245,
      "learning_rate": 1.8383880537955575e-05,
      "loss": 1.9493,
      "step": 204595
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.4176969528198242,
      "learning_rate": 1.8382470630876455e-05,
      "loss": 2.0896,
      "step": 204596
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.7182809114456177,
      "learning_rate": 1.8381060776155388e-05,
      "loss": 1.8563,
      "step": 204597
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.363256573677063,
      "learning_rate": 1.8379650973792603e-05,
      "loss": 1.8294,
      "step": 204598
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.4457228183746338,
      "learning_rate": 1.837824122378837e-05,
      "loss": 1.9786,
      "step": 204599
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.4258501529693604,
      "learning_rate": 1.837683152614289e-05,
      "loss": 1.9799,
      "step": 204600
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.579969882965088,
      "learning_rate": 1.8375421880856522e-05,
      "loss": 2.179,
      "step": 204601
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.4578559398651123,
      "learning_rate": 1.8374012287929406e-05,
      "loss": 2.111,
      "step": 204602
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.3480228185653687,
      "learning_rate": 1.8372602747361943e-05,
      "loss": 2.0102,
      "step": 204603
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.4111610651016235,
      "learning_rate": 1.8371193259154327e-05,
      "loss": 2.1178,
      "step": 204604
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.455046534538269,
      "learning_rate": 1.8369783823306828e-05,
      "loss": 1.8891,
      "step": 204605
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.4835799932479858,
      "learning_rate": 1.8368374439819677e-05,
      "loss": 2.1642,
      "step": 204606
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.489612102508545,
      "learning_rate": 1.836696510869318e-05,
      "loss": 2.2076,
      "step": 204607
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.3611849546432495,
      "learning_rate": 1.8365555829927526e-05,
      "loss": 1.9844,
      "step": 204608
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.4060667753219604,
      "learning_rate": 1.836414660352309e-05,
      "loss": 1.9293,
      "step": 204609
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.5125460624694824,
      "learning_rate": 1.8362737429480067e-05,
      "loss": 1.9484,
      "step": 204610
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.4494253396987915,
      "learning_rate": 1.836132830779873e-05,
      "loss": 1.78,
      "step": 204611
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.4038878679275513,
      "learning_rate": 1.8359919238479303e-05,
      "loss": 1.7343,
      "step": 204612
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.5883541107177734,
      "learning_rate": 1.8358510221522126e-05,
      "loss": 2.1221,
      "step": 204613
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.3969331979751587,
      "learning_rate": 1.8357101256927363e-05,
      "loss": 1.8231,
      "step": 204614
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.3711516857147217,
      "learning_rate": 1.835569234469538e-05,
      "loss": 2.0205,
      "step": 204615
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.369168758392334,
      "learning_rate": 1.8354283484826414e-05,
      "loss": 2.086,
      "step": 204616
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.3931188583374023,
      "learning_rate": 1.8352874677320627e-05,
      "loss": 2.0672,
      "step": 204617
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.4388036727905273,
      "learning_rate": 1.835146592217842e-05,
      "loss": 2.1309,
      "step": 204618
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.4407671689987183,
      "learning_rate": 1.8350057219399993e-05,
      "loss": 1.907,
      "step": 204619
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.5384372472763062,
      "learning_rate": 1.8348648568985547e-05,
      "loss": 2.0714,
      "step": 204620
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.3998992443084717,
      "learning_rate": 1.8347239970935446e-05,
      "loss": 2.1183,
      "step": 204621
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.3242651224136353,
      "learning_rate": 1.8345831425249924e-05,
      "loss": 1.9362,
      "step": 204622
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.4818750619888306,
      "learning_rate": 1.8344422931929183e-05,
      "loss": 1.812,
      "step": 204623
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.374157190322876,
      "learning_rate": 1.834301449097355e-05,
      "loss": 1.952,
      "step": 204624
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.4740864038467407,
      "learning_rate": 1.83416061023833e-05,
      "loss": 2.1445,
      "step": 204625
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.4140574932098389,
      "learning_rate": 1.8340197766158627e-05,
      "loss": 1.9317,
      "step": 204626
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.4237377643585205,
      "learning_rate": 1.8338789482299833e-05,
      "loss": 2.1393,
      "step": 204627
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.381326675415039,
      "learning_rate": 1.833738125080718e-05,
      "loss": 2.0166,
      "step": 204628
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.4857631921768188,
      "learning_rate": 1.833597307168091e-05,
      "loss": 1.7627,
      "step": 204629
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.4291003942489624,
      "learning_rate": 1.8334564944921317e-05,
      "loss": 1.8877,
      "step": 204630
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.3758511543273926,
      "learning_rate": 1.833315687052863e-05,
      "loss": 1.8165,
      "step": 204631
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.5324443578720093,
      "learning_rate": 1.833174884850316e-05,
      "loss": 2.0289,
      "step": 204632
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.471757411956787,
      "learning_rate": 1.8330340878845095e-05,
      "loss": 1.9292,
      "step": 204633
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.5174903869628906,
      "learning_rate": 1.8328932961554777e-05,
      "loss": 2.0536,
      "step": 204634
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.5662606954574585,
      "learning_rate": 1.8327525096632367e-05,
      "loss": 1.8998,
      "step": 204635
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.4448586702346802,
      "learning_rate": 1.832611728407827e-05,
      "loss": 2.0546,
      "step": 204636
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.7735209465026855,
      "learning_rate": 1.8324709523892577e-05,
      "loss": 2.0516,
      "step": 204637
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.5074446201324463,
      "learning_rate": 1.8323301816075696e-05,
      "loss": 1.9318,
      "step": 204638
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.5663968324661255,
      "learning_rate": 1.832189416062776e-05,
      "loss": 2.0183,
      "step": 204639
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.437880039215088,
      "learning_rate": 1.832048655754923e-05,
      "loss": 2.0303,
      "step": 204640
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.3062101602554321,
      "learning_rate": 1.831907900684014e-05,
      "loss": 2.0044,
      "step": 204641
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.2896453142166138,
      "learning_rate": 1.8317671508500865e-05,
      "loss": 1.9408,
      "step": 204642
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.3654087781906128,
      "learning_rate": 1.831626406253166e-05,
      "loss": 1.9296,
      "step": 204643
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.296668291091919,
      "learning_rate": 1.8314856668932763e-05,
      "loss": 1.7667,
      "step": 204644
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.4653455018997192,
      "learning_rate": 1.8313449327704444e-05,
      "loss": 1.9244,
      "step": 204645
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.5793402194976807,
      "learning_rate": 1.8312042038847064e-05,
      "loss": 2.0445,
      "step": 204646
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.4662846326828003,
      "learning_rate": 1.831063480236069e-05,
      "loss": 2.0526,
      "step": 204647
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.5398921966552734,
      "learning_rate": 1.8309227618245724e-05,
      "loss": 2.0693,
      "step": 204648
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.48782479763031,
      "learning_rate": 1.8307820486502368e-05,
      "loss": 1.8595,
      "step": 204649
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.4533694982528687,
      "learning_rate": 1.830641340713095e-05,
      "loss": 1.959,
      "step": 204650
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.6394320726394653,
      "learning_rate": 1.8305006380131637e-05,
      "loss": 1.8613,
      "step": 204651
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.4750362634658813,
      "learning_rate": 1.8303599405504765e-05,
      "loss": 1.8234,
      "step": 204652
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.3853181600570679,
      "learning_rate": 1.83021924832506e-05,
      "loss": 1.9359,
      "step": 204653
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.345535159111023,
      "learning_rate": 1.830078561336934e-05,
      "loss": 1.9759,
      "step": 204654
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.4667069911956787,
      "learning_rate": 1.8299378795861254e-05,
      "loss": 2.0865,
      "step": 204655
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.3286991119384766,
      "learning_rate": 1.8297972030726703e-05,
      "loss": 1.6776,
      "step": 204656
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.4703396558761597,
      "learning_rate": 1.8296565317965795e-05,
      "loss": 1.9253,
      "step": 204657
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.7445825338363647,
      "learning_rate": 1.8295158657578923e-05,
      "loss": 2.1189,
      "step": 204658
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.5364961624145508,
      "learning_rate": 1.8293752049566325e-05,
      "loss": 2.0172,
      "step": 204659
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.3375922441482544,
      "learning_rate": 1.8292345493928196e-05,
      "loss": 1.9491,
      "step": 204660
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.5110293626785278,
      "learning_rate": 1.8290938990664838e-05,
      "loss": 2.0915,
      "step": 204661
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.6142735481262207,
      "learning_rate": 1.828953253977652e-05,
      "loss": 2.2362,
      "step": 204662
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.291063904762268,
      "learning_rate": 1.828812614126347e-05,
      "loss": 1.8696,
      "step": 204663
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.4930880069732666,
      "learning_rate": 1.8286719795125993e-05,
      "loss": 1.9024,
      "step": 204664
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.5653928518295288,
      "learning_rate": 1.8285313501364353e-05,
      "loss": 1.9745,
      "step": 204665
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.4553335905075073,
      "learning_rate": 1.8283907259978815e-05,
      "loss": 2.0267,
      "step": 204666
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.5376670360565186,
      "learning_rate": 1.8282501070969546e-05,
      "loss": 2.1331,
      "step": 204667
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.412595272064209,
      "learning_rate": 1.8281094934336915e-05,
      "loss": 2.0649,
      "step": 204668
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.3899543285369873,
      "learning_rate": 1.8279688850081122e-05,
      "loss": 2.0236,
      "step": 204669
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.3962681293487549,
      "learning_rate": 1.8278282818202495e-05,
      "loss": 1.8883,
      "step": 204670
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.527714729309082,
      "learning_rate": 1.827687683870124e-05,
      "loss": 2.0742,
      "step": 204671
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.412219762802124,
      "learning_rate": 1.8275470911577616e-05,
      "loss": 2.159,
      "step": 204672
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.4855068922042847,
      "learning_rate": 1.82740650368319e-05,
      "loss": 1.821,
      "step": 204673
    },
    {
      "epoch": 2.66,
      "grad_norm": 1.5264748334884644,
      "learning_rate": 1.8272659214464348e-05,
      "loss": 1.9413,
      "step": 204674
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.314157485961914,
      "learning_rate": 1.827125344447523e-05,
      "loss": 1.8854,
      "step": 204675
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.3561967611312866,
      "learning_rate": 1.8269847726864818e-05,
      "loss": 1.8297,
      "step": 204676
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.4833877086639404,
      "learning_rate": 1.826844206163337e-05,
      "loss": 2.0829,
      "step": 204677
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.4541445970535278,
      "learning_rate": 1.8267036448781123e-05,
      "loss": 2.2792,
      "step": 204678
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.3539316654205322,
      "learning_rate": 1.8265630888308313e-05,
      "loss": 1.8717,
      "step": 204679
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.3079545497894287,
      "learning_rate": 1.82642253802153e-05,
      "loss": 1.9661,
      "step": 204680
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.6182376146316528,
      "learning_rate": 1.8262819924502224e-05,
      "loss": 1.9737,
      "step": 204681
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.336007833480835,
      "learning_rate": 1.8261414521169482e-05,
      "loss": 2.1283,
      "step": 204682
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.359631896018982,
      "learning_rate": 1.8260009170217237e-05,
      "loss": 2.0566,
      "step": 204683
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.3906396627426147,
      "learning_rate": 1.825860387164576e-05,
      "loss": 2.0317,
      "step": 204684
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.4857484102249146,
      "learning_rate": 1.8257198625455315e-05,
      "loss": 2.1695,
      "step": 204685
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.462506890296936,
      "learning_rate": 1.8255793431646204e-05,
      "loss": 1.9271,
      "step": 204686
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.6280102729797363,
      "learning_rate": 1.825438829021859e-05,
      "loss": 1.8231,
      "step": 204687
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.920654058456421,
      "learning_rate": 1.8252983201172876e-05,
      "loss": 1.8624,
      "step": 204688
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.5449787378311157,
      "learning_rate": 1.825157816450926e-05,
      "loss": 1.9155,
      "step": 204689
    },
    {
      "epoch": 2.67,
      "grad_norm": 2.1197662353515625,
      "learning_rate": 1.8250173180227977e-05,
      "loss": 2.2227,
      "step": 204690
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.515285611152649,
      "learning_rate": 1.824876824832926e-05,
      "loss": 2.0832,
      "step": 204691
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.4641358852386475,
      "learning_rate": 1.8247363368813473e-05,
      "loss": 2.1366,
      "step": 204692
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.5229578018188477,
      "learning_rate": 1.824595854168078e-05,
      "loss": 2.172,
      "step": 204693
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.391637921333313,
      "learning_rate": 1.824455376693149e-05,
      "loss": 2.0316,
      "step": 204694
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.5546164512634277,
      "learning_rate": 1.8243149044565895e-05,
      "loss": 1.8932,
      "step": 204695
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.339269995689392,
      "learning_rate": 1.82417443745842e-05,
      "loss": 1.9955,
      "step": 204696
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.359721064567566,
      "learning_rate": 1.8240339756986632e-05,
      "loss": 1.8678,
      "step": 204697
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.4494141340255737,
      "learning_rate": 1.8238935191773563e-05,
      "loss": 2.05,
      "step": 204698
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.349707841873169,
      "learning_rate": 1.8237530678945156e-05,
      "loss": 1.957,
      "step": 204699
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.3961551189422607,
      "learning_rate": 1.8236126218501745e-05,
      "loss": 1.8363,
      "step": 204700
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.2421287298202515,
      "learning_rate": 1.823472181044353e-05,
      "loss": 2.1303,
      "step": 204701
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.3894329071044922,
      "learning_rate": 1.8233317454770844e-05,
      "loss": 1.9908,
      "step": 204702
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.4305224418640137,
      "learning_rate": 1.8231913151483823e-05,
      "loss": 1.9212,
      "step": 204703
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.3677831888198853,
      "learning_rate": 1.8230508900582895e-05,
      "loss": 1.838,
      "step": 204704
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.414934515953064,
      "learning_rate": 1.8229104702068163e-05,
      "loss": 1.9224,
      "step": 204705
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.6098793745040894,
      "learning_rate": 1.8227700555939994e-05,
      "loss": 2.1531,
      "step": 204706
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.6325129270553589,
      "learning_rate": 1.822629646219862e-05,
      "loss": 2.154,
      "step": 204707
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.4339314699172974,
      "learning_rate": 1.822489242084424e-05,
      "loss": 1.881,
      "step": 204708
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.4377708435058594,
      "learning_rate": 1.8223488431877254e-05,
      "loss": 1.8441,
      "step": 204709
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.4761327505111694,
      "learning_rate": 1.822208449529783e-05,
      "loss": 2.0636,
      "step": 204710
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.4684158563613892,
      "learning_rate": 1.8220680611106165e-05,
      "loss": 1.8935,
      "step": 204711
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.552504062652588,
      "learning_rate": 1.821927677930266e-05,
      "loss": 2.0642,
      "step": 204712
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.426893711090088,
      "learning_rate": 1.8217872999887517e-05,
      "loss": 2.0784,
      "step": 204713
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.3459707498550415,
      "learning_rate": 1.8216469272860933e-05,
      "loss": 1.7688,
      "step": 204714
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.5773377418518066,
      "learning_rate": 1.821506559822328e-05,
      "loss": 2.1337,
      "step": 204715
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.457411289215088,
      "learning_rate": 1.8213661975974714e-05,
      "loss": 1.8077,
      "step": 204716
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.318630337715149,
      "learning_rate": 1.8212258406115578e-05,
      "loss": 1.8331,
      "step": 204717
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.4851222038269043,
      "learning_rate": 1.821085488864613e-05,
      "loss": 2.2094,
      "step": 204718
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.356355905532837,
      "learning_rate": 1.8209451423566545e-05,
      "loss": 2.0377,
      "step": 204719
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.4498720169067383,
      "learning_rate": 1.8208048010877184e-05,
      "loss": 2.1298,
      "step": 204720
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.570510745048523,
      "learning_rate": 1.820664465057825e-05,
      "loss": 2.0772,
      "step": 204721
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.5078703165054321,
      "learning_rate": 1.8205241342670006e-05,
      "loss": 1.9937,
      "step": 204722
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.4229635000228882,
      "learning_rate": 1.8203838087152755e-05,
      "loss": 2.0818,
      "step": 204723
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.5596308708190918,
      "learning_rate": 1.8202434884026762e-05,
      "loss": 1.8606,
      "step": 204724
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.6367390155792236,
      "learning_rate": 1.820103173329216e-05,
      "loss": 2.1366,
      "step": 204725
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.4186848402023315,
      "learning_rate": 1.8199628634949383e-05,
      "loss": 1.8481,
      "step": 204726
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.510111927986145,
      "learning_rate": 1.8198225588998593e-05,
      "loss": 2.1433,
      "step": 204727
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.4771065711975098,
      "learning_rate": 1.8196822595440064e-05,
      "loss": 2.1055,
      "step": 204728
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.5796140432357788,
      "learning_rate": 1.819541965427409e-05,
      "loss": 2.3264,
      "step": 204729
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.4279849529266357,
      "learning_rate": 1.819401676550084e-05,
      "loss": 1.8336,
      "step": 204730
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.5247626304626465,
      "learning_rate": 1.8192613929120714e-05,
      "loss": 2.1216,
      "step": 204731
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.4350470304489136,
      "learning_rate": 1.819121114513391e-05,
      "loss": 2.1053,
      "step": 204732
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.5190380811691284,
      "learning_rate": 1.8189808413540663e-05,
      "loss": 2.0823,
      "step": 204733
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.4825767278671265,
      "learning_rate": 1.818840573434117e-05,
      "loss": 1.969,
      "step": 204734
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.3965678215026855,
      "learning_rate": 1.818700310753587e-05,
      "loss": 1.907,
      "step": 204735
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.5111192464828491,
      "learning_rate": 1.8185600533124855e-05,
      "loss": 1.9408,
      "step": 204736
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.3995925188064575,
      "learning_rate": 1.8184198011108497e-05,
      "loss": 1.7924,
      "step": 204737
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.412536859512329,
      "learning_rate": 1.818279554148706e-05,
      "loss": 2.0179,
      "step": 204738
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.5258959531784058,
      "learning_rate": 1.8181393124260712e-05,
      "loss": 1.8962,
      "step": 204739
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.360622525215149,
      "learning_rate": 1.8179990759429718e-05,
      "loss": 1.8154,
      "step": 204740
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.5789662599563599,
      "learning_rate": 1.8178588446994448e-05,
      "loss": 2.0641,
      "step": 204741
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.5235517024993896,
      "learning_rate": 1.817718618695506e-05,
      "loss": 1.9578,
      "step": 204742
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.4899275302886963,
      "learning_rate": 1.8175783979311897e-05,
      "loss": 1.9591,
      "step": 204743
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.515113115310669,
      "learning_rate": 1.8174381824065156e-05,
      "loss": 2.1823,
      "step": 204744
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.4276314973831177,
      "learning_rate": 1.817297972121513e-05,
      "loss": 2.0241,
      "step": 204745
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.3848075866699219,
      "learning_rate": 1.8171577670762028e-05,
      "loss": 2.1587,
      "step": 204746
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.3243365287780762,
      "learning_rate": 1.8170175672706176e-05,
      "loss": 1.9116,
      "step": 204747
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.5075411796569824,
      "learning_rate": 1.816877372704778e-05,
      "loss": 1.9524,
      "step": 204748
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.5284755229949951,
      "learning_rate": 1.816737183378717e-05,
      "loss": 1.9239,
      "step": 204749
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.531847357749939,
      "learning_rate": 1.8165969992924578e-05,
      "loss": 1.8913,
      "step": 204750
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.417035698890686,
      "learning_rate": 1.8164568204460206e-05,
      "loss": 2.1176,
      "step": 204751
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.4523773193359375,
      "learning_rate": 1.8163166468394352e-05,
      "loss": 2.0679,
      "step": 204752
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.4455903768539429,
      "learning_rate": 1.816176478472735e-05,
      "loss": 2.0615,
      "step": 204753
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.5012593269348145,
      "learning_rate": 1.8160363153459302e-05,
      "loss": 1.8874,
      "step": 204754
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.521285057067871,
      "learning_rate": 1.8158961574590634e-05,
      "loss": 1.9498,
      "step": 204755
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.3426560163497925,
      "learning_rate": 1.8157560048121556e-05,
      "loss": 1.9793,
      "step": 204756
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.6267471313476562,
      "learning_rate": 1.815615857405226e-05,
      "loss": 1.9302,
      "step": 204757
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.4408193826675415,
      "learning_rate": 1.8154757152383047e-05,
      "loss": 1.8093,
      "step": 204758
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.4104440212249756,
      "learning_rate": 1.815335578311422e-05,
      "loss": 1.9266,
      "step": 204759
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.4324084520339966,
      "learning_rate": 1.8151954466245944e-05,
      "loss": 1.8164,
      "step": 204760
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.5997906923294067,
      "learning_rate": 1.8150553201778617e-05,
      "loss": 1.9912,
      "step": 204761
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.3840577602386475,
      "learning_rate": 1.8149151989712406e-05,
      "loss": 1.8965,
      "step": 204762
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.482314109802246,
      "learning_rate": 1.814775083004758e-05,
      "loss": 1.9391,
      "step": 204763
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.4984756708145142,
      "learning_rate": 1.814634972278437e-05,
      "loss": 2.1029,
      "step": 204764
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.3849316835403442,
      "learning_rate": 1.814494866792311e-05,
      "loss": 1.9991,
      "step": 204765
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.4750787019729614,
      "learning_rate": 1.8143547665463997e-05,
      "loss": 1.8207,
      "step": 204766
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.3549643754959106,
      "learning_rate": 1.8142146715407334e-05,
      "loss": 1.8275,
      "step": 204767
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.4752360582351685,
      "learning_rate": 1.8140745817753387e-05,
      "loss": 2.1703,
      "step": 204768
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.3836784362792969,
      "learning_rate": 1.813934497250239e-05,
      "loss": 1.9593,
      "step": 204769
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.4690487384796143,
      "learning_rate": 1.813794417965457e-05,
      "loss": 2.2447,
      "step": 204770
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.4058995246887207,
      "learning_rate": 1.8136543439210272e-05,
      "loss": 1.847,
      "step": 204771
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.391158103942871,
      "learning_rate": 1.8135142751169685e-05,
      "loss": 1.9716,
      "step": 204772
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.4690293073654175,
      "learning_rate": 1.8133742115533112e-05,
      "loss": 2.1234,
      "step": 204773
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.5075258016586304,
      "learning_rate": 1.813234153230082e-05,
      "loss": 2.2039,
      "step": 204774
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.4569785594940186,
      "learning_rate": 1.813094100147301e-05,
      "loss": 1.8953,
      "step": 204775
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.3893083333969116,
      "learning_rate": 1.8129540523049946e-05,
      "loss": 2.2375,
      "step": 204776
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.5134656429290771,
      "learning_rate": 1.8128140097031994e-05,
      "loss": 2.112,
      "step": 204777
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.470754861831665,
      "learning_rate": 1.812673972341926e-05,
      "loss": 2.1386,
      "step": 204778
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.6513705253601074,
      "learning_rate": 1.812533940221217e-05,
      "loss": 2.0445,
      "step": 204779
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.4768877029418945,
      "learning_rate": 1.8123939133410858e-05,
      "loss": 1.8903,
      "step": 204780
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.3234323263168335,
      "learning_rate": 1.8122538917015662e-05,
      "loss": 1.7361,
      "step": 204781
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.6273841857910156,
      "learning_rate": 1.8121138753026743e-05,
      "loss": 1.8643,
      "step": 204782
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.3220109939575195,
      "learning_rate": 1.811973864144447e-05,
      "loss": 1.7726,
      "step": 204783
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.3811315298080444,
      "learning_rate": 1.811833858226904e-05,
      "loss": 1.9091,
      "step": 204784
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.5090059041976929,
      "learning_rate": 1.8116938575500727e-05,
      "loss": 1.8297,
      "step": 204785
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.4464547634124756,
      "learning_rate": 1.811553862113979e-05,
      "loss": 2.1407,
      "step": 204786
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.409195899963379,
      "learning_rate": 1.8114138719186565e-05,
      "loss": 1.9129,
      "step": 204787
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.6661722660064697,
      "learning_rate": 1.8112738869641186e-05,
      "loss": 1.7773,
      "step": 204788
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.3423150777816772,
      "learning_rate": 1.8111339072503985e-05,
      "loss": 1.8308,
      "step": 204789
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.3643839359283447,
      "learning_rate": 1.8109939327775158e-05,
      "loss": 1.9309,
      "step": 204790
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.5311154127120972,
      "learning_rate": 1.8108539635455075e-05,
      "loss": 1.9101,
      "step": 204791
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.3299528360366821,
      "learning_rate": 1.8107139995543872e-05,
      "loss": 1.7904,
      "step": 204792
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.397401213645935,
      "learning_rate": 1.8105740408041945e-05,
      "loss": 1.9981,
      "step": 204793
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.523600459098816,
      "learning_rate": 1.8104340872949463e-05,
      "loss": 1.8633,
      "step": 204794
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.3680745363235474,
      "learning_rate": 1.8102941390266722e-05,
      "loss": 2.0484,
      "step": 204795
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.5014023780822754,
      "learning_rate": 1.8101541959993893e-05,
      "loss": 2.0698,
      "step": 204796
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.5320438146591187,
      "learning_rate": 1.8100142582131373e-05,
      "loss": 1.9832,
      "step": 204797
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.4070135354995728,
      "learning_rate": 1.8098743256679326e-05,
      "loss": 1.8458,
      "step": 204798
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.5236390829086304,
      "learning_rate": 1.809734398363806e-05,
      "loss": 1.838,
      "step": 204799
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.4152696132659912,
      "learning_rate": 1.809594476300783e-05,
      "loss": 1.7875,
      "step": 204800
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.4557793140411377,
      "learning_rate": 1.8094545594788844e-05,
      "loss": 1.9547,
      "step": 204801
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.4190131425857544,
      "learning_rate": 1.8093146478981435e-05,
      "loss": 1.9895,
      "step": 204802
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.430565357208252,
      "learning_rate": 1.8091747415585867e-05,
      "loss": 1.9922,
      "step": 204803
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.4225152730941772,
      "learning_rate": 1.8090348404602272e-05,
      "loss": 1.8793,
      "step": 204804
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.5371530055999756,
      "learning_rate": 1.8088949446031087e-05,
      "loss": 2.0811,
      "step": 204805
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.5158557891845703,
      "learning_rate": 1.8087550539872474e-05,
      "loss": 1.9983,
      "step": 204806
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.4004263877868652,
      "learning_rate": 1.8086151686126637e-05,
      "loss": 1.9497,
      "step": 204807
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.452467679977417,
      "learning_rate": 1.8084752884793973e-05,
      "loss": 2.0882,
      "step": 204808
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.5263139009475708,
      "learning_rate": 1.8083354135874684e-05,
      "loss": 2.0791,
      "step": 204809
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.4421961307525635,
      "learning_rate": 1.8081955439368967e-05,
      "loss": 1.9215,
      "step": 204810
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.4402951002120972,
      "learning_rate": 1.8080556795277192e-05,
      "loss": 2.07,
      "step": 204811
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.5754860639572144,
      "learning_rate": 1.8079158203599554e-05,
      "loss": 2.0512,
      "step": 204812
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.423672080039978,
      "learning_rate": 1.8077759664336254e-05,
      "loss": 2.1204,
      "step": 204813
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.4163990020751953,
      "learning_rate": 1.8076361177487696e-05,
      "loss": 1.8051,
      "step": 204814
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.5660048723220825,
      "learning_rate": 1.807496274305401e-05,
      "loss": 1.9084,
      "step": 204815
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.5170094966888428,
      "learning_rate": 1.8073564361035563e-05,
      "loss": 1.7275,
      "step": 204816
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.5238721370697021,
      "learning_rate": 1.8072166031432556e-05,
      "loss": 1.816,
      "step": 204817
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.3794499635696411,
      "learning_rate": 1.807076775424525e-05,
      "loss": 1.9682,
      "step": 204818
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.401600956916809,
      "learning_rate": 1.8069369529473853e-05,
      "loss": 2.1295,
      "step": 204819
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.5399806499481201,
      "learning_rate": 1.806797135711876e-05,
      "loss": 2.0093,
      "step": 204820
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.4260940551757812,
      "learning_rate": 1.806657323718007e-05,
      "loss": 1.9976,
      "step": 204821
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.463149905204773,
      "learning_rate": 1.8065175169658185e-05,
      "loss": 2.1513,
      "step": 204822
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.5515800714492798,
      "learning_rate": 1.8063777154553337e-05,
      "loss": 2.1199,
      "step": 204823
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.54771888256073,
      "learning_rate": 1.8062379191865727e-05,
      "loss": 1.9319,
      "step": 204824
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.5586645603179932,
      "learning_rate": 1.806098128159559e-05,
      "loss": 1.8819,
      "step": 204825
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.4341825246810913,
      "learning_rate": 1.8059583423743285e-05,
      "loss": 1.8894,
      "step": 204826
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.517309308052063,
      "learning_rate": 1.8058185618309018e-05,
      "loss": 1.7275,
      "step": 204827
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.396886944770813,
      "learning_rate": 1.8056787865293054e-05,
      "loss": 1.8437,
      "step": 204828
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.4969608783721924,
      "learning_rate": 1.8055390164695693e-05,
      "loss": 2.1703,
      "step": 204829
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.4205546379089355,
      "learning_rate": 1.8053992516517135e-05,
      "loss": 2.1025,
      "step": 204830
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.4597543478012085,
      "learning_rate": 1.805259492075761e-05,
      "loss": 1.9711,
      "step": 204831
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.4710891246795654,
      "learning_rate": 1.805119737741749e-05,
      "loss": 1.9528,
      "step": 204832
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.4733657836914062,
      "learning_rate": 1.804979988649691e-05,
      "loss": 1.8819,
      "step": 204833
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.4913047552108765,
      "learning_rate": 1.8048402447996258e-05,
      "loss": 1.9686,
      "step": 204834
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.4087457656860352,
      "learning_rate": 1.8047005061915742e-05,
      "loss": 2.1751,
      "step": 204835
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.5126807689666748,
      "learning_rate": 1.8045607728255595e-05,
      "loss": 2.1026,
      "step": 204836
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.447445034980774,
      "learning_rate": 1.8044210447016017e-05,
      "loss": 2.0561,
      "step": 204837
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.419995903968811,
      "learning_rate": 1.8042813218197437e-05,
      "loss": 2.0186,
      "step": 204838
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.4803224802017212,
      "learning_rate": 1.804141604179996e-05,
      "loss": 1.8889,
      "step": 204839
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.4448848962783813,
      "learning_rate": 1.8040018917823952e-05,
      "loss": 2.071,
      "step": 204840
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.4881681203842163,
      "learning_rate": 1.803862184626961e-05,
      "loss": 1.9209,
      "step": 204841
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.5654608011245728,
      "learning_rate": 1.803722482713723e-05,
      "loss": 1.9341,
      "step": 204842
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.6060038805007935,
      "learning_rate": 1.803582786042699e-05,
      "loss": 2.0186,
      "step": 204843
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.3731448650360107,
      "learning_rate": 1.803443094613928e-05,
      "loss": 2.0085,
      "step": 204844
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.4651294946670532,
      "learning_rate": 1.80330340842742e-05,
      "loss": 2.0574,
      "step": 204845
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.4294109344482422,
      "learning_rate": 1.803163727483219e-05,
      "loss": 1.9184,
      "step": 204846
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.7216886281967163,
      "learning_rate": 1.803024051781341e-05,
      "loss": 1.8702,
      "step": 204847
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.3313345909118652,
      "learning_rate": 1.802884381321813e-05,
      "loss": 1.9031,
      "step": 204848
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.376883864402771,
      "learning_rate": 1.8027447161046582e-05,
      "loss": 1.8973,
      "step": 204849
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.4835472106933594,
      "learning_rate": 1.8026050561299064e-05,
      "loss": 2.052,
      "step": 204850
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.4563603401184082,
      "learning_rate": 1.8024654013975814e-05,
      "loss": 2.0821,
      "step": 204851
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.3603382110595703,
      "learning_rate": 1.8023257519077128e-05,
      "loss": 1.9362,
      "step": 204852
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.369860291481018,
      "learning_rate": 1.8021861076603204e-05,
      "loss": 1.9048,
      "step": 204853
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.2862056493759155,
      "learning_rate": 1.8020464686554447e-05,
      "loss": 1.8917,
      "step": 204854
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.4202979803085327,
      "learning_rate": 1.8019068348930887e-05,
      "loss": 1.9889,
      "step": 204855
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.4299613237380981,
      "learning_rate": 1.8017672063732957e-05,
      "loss": 1.859,
      "step": 204856
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.3409401178359985,
      "learning_rate": 1.8016275830960826e-05,
      "loss": 1.9147,
      "step": 204857
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.579056739807129,
      "learning_rate": 1.8014879650614855e-05,
      "loss": 2.2098,
      "step": 204858
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.4793757200241089,
      "learning_rate": 1.801348352269518e-05,
      "loss": 1.9341,
      "step": 204859
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.4113742113113403,
      "learning_rate": 1.8012087447202206e-05,
      "loss": 1.8582,
      "step": 204860
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.3982234001159668,
      "learning_rate": 1.8010691424136026e-05,
      "loss": 1.9456,
      "step": 204861
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.43220055103302,
      "learning_rate": 1.800929545349701e-05,
      "loss": 1.65,
      "step": 204862
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.4045398235321045,
      "learning_rate": 1.8007899535285352e-05,
      "loss": 1.9091,
      "step": 204863
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.5140557289123535,
      "learning_rate": 1.8006503669501393e-05,
      "loss": 1.9688,
      "step": 204864
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.5189253091812134,
      "learning_rate": 1.8005107856145294e-05,
      "loss": 1.8272,
      "step": 204865
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.6443935632705688,
      "learning_rate": 1.8003712095217492e-05,
      "loss": 2.1187,
      "step": 204866
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.5061718225479126,
      "learning_rate": 1.8002316386718017e-05,
      "loss": 2.1801,
      "step": 204867
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.458126187324524,
      "learning_rate": 1.8000920730647273e-05,
      "loss": 2.0886,
      "step": 204868
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.543350100517273,
      "learning_rate": 1.7999525127005422e-05,
      "loss": 2.1131,
      "step": 204869
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.44368577003479,
      "learning_rate": 1.7998129575792864e-05,
      "loss": 1.7862,
      "step": 204870
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.4966565370559692,
      "learning_rate": 1.79967340770097e-05,
      "loss": 1.844,
      "step": 204871
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.3911783695220947,
      "learning_rate": 1.7995338630656362e-05,
      "loss": 2.1149,
      "step": 204872
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.3229405879974365,
      "learning_rate": 1.799394323673292e-05,
      "loss": 2.0417,
      "step": 204873
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.4583395719528198,
      "learning_rate": 1.7992547895239808e-05,
      "loss": 2.0754,
      "step": 204874
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.3998970985412598,
      "learning_rate": 1.7991152606177116e-05,
      "loss": 1.833,
      "step": 204875
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.4524129629135132,
      "learning_rate": 1.7989757369545256e-05,
      "loss": 2.0101,
      "step": 204876
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.4812331199645996,
      "learning_rate": 1.7988362185344383e-05,
      "loss": 1.8016,
      "step": 204877
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.6081990003585815,
      "learning_rate": 1.798696705357484e-05,
      "loss": 2.0647,
      "step": 204878
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.5235487222671509,
      "learning_rate": 1.798557197423682e-05,
      "loss": 1.9039,
      "step": 204879
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.4691712856292725,
      "learning_rate": 1.798417694733063e-05,
      "loss": 2.1388,
      "step": 204880
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.3465627431869507,
      "learning_rate": 1.7982781972856464e-05,
      "loss": 1.7871,
      "step": 204881
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.5369205474853516,
      "learning_rate": 1.7981387050814657e-05,
      "loss": 2.1999,
      "step": 204882
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.7598800659179688,
      "learning_rate": 1.797999218120537e-05,
      "loss": 1.8844,
      "step": 204883
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.6672719717025757,
      "learning_rate": 1.7978597364029014e-05,
      "loss": 2.076,
      "step": 204884
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.374647617340088,
      "learning_rate": 1.797720259928571e-05,
      "loss": 1.9971,
      "step": 204885
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.291303277015686,
      "learning_rate": 1.7975807886975803e-05,
      "loss": 1.7351,
      "step": 204886
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.315527319908142,
      "learning_rate": 1.797441322709948e-05,
      "loss": 1.8689,
      "step": 204887
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.3611077070236206,
      "learning_rate": 1.7973018619657054e-05,
      "loss": 1.9818,
      "step": 204888
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.446292757987976,
      "learning_rate": 1.7971624064648748e-05,
      "loss": 1.8397,
      "step": 204889
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.4431859254837036,
      "learning_rate": 1.7970229562074867e-05,
      "loss": 2.0495,
      "step": 204890
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.4778553247451782,
      "learning_rate": 1.7968835111935643e-05,
      "loss": 1.9204,
      "step": 204891
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.6302821636199951,
      "learning_rate": 1.7967440714231306e-05,
      "loss": 2.0246,
      "step": 204892
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.4614601135253906,
      "learning_rate": 1.796604636896216e-05,
      "loss": 1.8437,
      "step": 204893
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.4425760507583618,
      "learning_rate": 1.796465207612847e-05,
      "loss": 1.6588,
      "step": 204894
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.441734790802002,
      "learning_rate": 1.7963257835730438e-05,
      "loss": 1.9015,
      "step": 204895
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.5039390325546265,
      "learning_rate": 1.7961863647768393e-05,
      "loss": 1.9444,
      "step": 204896
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.5039952993392944,
      "learning_rate": 1.796046951224257e-05,
      "loss": 1.8224,
      "step": 204897
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.724564790725708,
      "learning_rate": 1.795907542915317e-05,
      "loss": 2.0464,
      "step": 204898
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.4394210577011108,
      "learning_rate": 1.7957681398500555e-05,
      "loss": 1.9361,
      "step": 204899
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.3927780389785767,
      "learning_rate": 1.7956287420284864e-05,
      "loss": 2.1492,
      "step": 204900
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.4801002740859985,
      "learning_rate": 1.7954893494506495e-05,
      "loss": 2.0146,
      "step": 204901
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.6157445907592773,
      "learning_rate": 1.7953499621165612e-05,
      "loss": 2.0448,
      "step": 204902
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.719642162322998,
      "learning_rate": 1.7952105800262516e-05,
      "loss": 1.9547,
      "step": 204903
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.3819772005081177,
      "learning_rate": 1.795071203179741e-05,
      "loss": 1.9609,
      "step": 204904
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.3933558464050293,
      "learning_rate": 1.7949318315770624e-05,
      "loss": 1.8225,
      "step": 204905
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.3820338249206543,
      "learning_rate": 1.7947924652182322e-05,
      "loss": 2.0456,
      "step": 204906
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.4997445344924927,
      "learning_rate": 1.7946531041032874e-05,
      "loss": 2.1294,
      "step": 204907
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.5509690046310425,
      "learning_rate": 1.7945137482322515e-05,
      "loss": 2.1819,
      "step": 204908
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.5793166160583496,
      "learning_rate": 1.794374397605147e-05,
      "loss": 1.8604,
      "step": 204909
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.4233052730560303,
      "learning_rate": 1.7942350522219952e-05,
      "loss": 1.9864,
      "step": 204910
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.5853909254074097,
      "learning_rate": 1.7940957120828314e-05,
      "loss": 1.9241,
      "step": 204911
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.4141467809677124,
      "learning_rate": 1.7939563771876763e-05,
      "loss": 1.9089,
      "step": 204912
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.5024949312210083,
      "learning_rate": 1.7938170475365598e-05,
      "loss": 1.7286,
      "step": 204913
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.587283730506897,
      "learning_rate": 1.7936777231294985e-05,
      "loss": 2.0914,
      "step": 204914
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.4853562116622925,
      "learning_rate": 1.7935384039665358e-05,
      "loss": 1.8318,
      "step": 204915
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.3510756492614746,
      "learning_rate": 1.793399090047678e-05,
      "loss": 1.9211,
      "step": 204916
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.4459964036941528,
      "learning_rate": 1.7932597813729655e-05,
      "loss": 2.1194,
      "step": 204917
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.832401990890503,
      "learning_rate": 1.7931204779424112e-05,
      "loss": 2.089,
      "step": 204918
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.3350492715835571,
      "learning_rate": 1.7929811797560557e-05,
      "loss": 1.8247,
      "step": 204919
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.5235471725463867,
      "learning_rate": 1.7928418868139116e-05,
      "loss": 1.9724,
      "step": 204920
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.3132120370864868,
      "learning_rate": 1.7927025991160192e-05,
      "loss": 2.0056,
      "step": 204921
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.3811492919921875,
      "learning_rate": 1.7925633166623853e-05,
      "loss": 2.0009,
      "step": 204922
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.4617693424224854,
      "learning_rate": 1.792424039453053e-05,
      "loss": 1.8766,
      "step": 204923
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.5514150857925415,
      "learning_rate": 1.7922847674880357e-05,
      "loss": 2.1506,
      "step": 204924
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.4098459482192993,
      "learning_rate": 1.7921455007673703e-05,
      "loss": 2.037,
      "step": 204925
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.5660380125045776,
      "learning_rate": 1.7920062392910728e-05,
      "loss": 2.0942,
      "step": 204926
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.7196786403656006,
      "learning_rate": 1.7918669830591835e-05,
      "loss": 2.095,
      "step": 204927
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.7654004096984863,
      "learning_rate": 1.7917277320717095e-05,
      "loss": 2.129,
      "step": 204928
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.7278006076812744,
      "learning_rate": 1.79158848632869e-05,
      "loss": 1.939,
      "step": 204929
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.4675017595291138,
      "learning_rate": 1.791449245830142e-05,
      "loss": 1.9907,
      "step": 204930
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.5005908012390137,
      "learning_rate": 1.7913100105761025e-05,
      "loss": 2.0608,
      "step": 204931
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.5159553289413452,
      "learning_rate": 1.7911707805665843e-05,
      "loss": 2.1035,
      "step": 204932
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.5024662017822266,
      "learning_rate": 1.7910315558016308e-05,
      "loss": 1.9413,
      "step": 204933
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.4108295440673828,
      "learning_rate": 1.7908923362812455e-05,
      "loss": 1.9787,
      "step": 204934
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.4489315748214722,
      "learning_rate": 1.7907531220054717e-05,
      "loss": 2.1131,
      "step": 204935
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.4370288848876953,
      "learning_rate": 1.7906139129743257e-05,
      "loss": 2.0312,
      "step": 204936
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.4280771017074585,
      "learning_rate": 1.790474709187838e-05,
      "loss": 2.0657,
      "step": 204937
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.6219313144683838,
      "learning_rate": 1.7903355106460314e-05,
      "loss": 2.2052,
      "step": 204938
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.3095173835754395,
      "learning_rate": 1.7901963173489432e-05,
      "loss": 2.0252,
      "step": 204939
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.4145206212997437,
      "learning_rate": 1.7900571292965827e-05,
      "loss": 2.0065,
      "step": 204940
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.3691492080688477,
      "learning_rate": 1.7899179464889868e-05,
      "loss": 2.0569,
      "step": 204941
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.4373525381088257,
      "learning_rate": 1.7897787689261722e-05,
      "loss": 1.8529,
      "step": 204942
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.3986716270446777,
      "learning_rate": 1.7896395966081724e-05,
      "loss": 1.8729,
      "step": 204943
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.445852279663086,
      "learning_rate": 1.78950042953501e-05,
      "loss": 1.9269,
      "step": 204944
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.4754359722137451,
      "learning_rate": 1.7893612677067195e-05,
      "loss": 1.8655,
      "step": 204945
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.4973890781402588,
      "learning_rate": 1.7892221111233097e-05,
      "loss": 1.8908,
      "step": 204946
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.4040595293045044,
      "learning_rate": 1.789082959784821e-05,
      "loss": 1.8579,
      "step": 204947
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.5108778476715088,
      "learning_rate": 1.7889438136912704e-05,
      "loss": 2.0299,
      "step": 204948
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.405579924583435,
      "learning_rate": 1.788804672842694e-05,
      "loss": 1.9747,
      "step": 204949
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.477776050567627,
      "learning_rate": 1.7886655372391025e-05,
      "loss": 1.9483,
      "step": 204950
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.5434845685958862,
      "learning_rate": 1.788526406880545e-05,
      "loss": 2.2133,
      "step": 204951
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.5692716836929321,
      "learning_rate": 1.7883872817670188e-05,
      "loss": 2.2306,
      "step": 204952
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.4036487340927124,
      "learning_rate": 1.78824816189857e-05,
      "loss": 2.0939,
      "step": 204953
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.339483618736267,
      "learning_rate": 1.7881090472752158e-05,
      "loss": 2.1491,
      "step": 204954
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.5726158618927002,
      "learning_rate": 1.787969937896989e-05,
      "loss": 1.9156,
      "step": 204955
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.4931825399398804,
      "learning_rate": 1.7878308337639035e-05,
      "loss": 1.9862,
      "step": 204956
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.3244110345840454,
      "learning_rate": 1.7876917348760054e-05,
      "loss": 2.0283,
      "step": 204957
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.5142252445220947,
      "learning_rate": 1.787552641233295e-05,
      "loss": 1.8519,
      "step": 204958
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.3416426181793213,
      "learning_rate": 1.787413552835819e-05,
      "loss": 1.9238,
      "step": 204959
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.5348050594329834,
      "learning_rate": 1.7872744696835906e-05,
      "loss": 1.9564,
      "step": 204960
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.4336085319519043,
      "learning_rate": 1.787135391776643e-05,
      "loss": 2.0271,
      "step": 204961
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.3910819292068481,
      "learning_rate": 1.786996319114996e-05,
      "loss": 2.0478,
      "step": 204962
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.430886149406433,
      "learning_rate": 1.7868572516986834e-05,
      "loss": 2.1103,
      "step": 204963
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.3468774557113647,
      "learning_rate": 1.7867181895277283e-05,
      "loss": 1.8376,
      "step": 204964
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.387908935546875,
      "learning_rate": 1.7865791326021538e-05,
      "loss": 2.22,
      "step": 204965
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.4797505140304565,
      "learning_rate": 1.7864400809219802e-05,
      "loss": 1.9015,
      "step": 204966
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.3863005638122559,
      "learning_rate": 1.7863010344872475e-05,
      "loss": 2.0597,
      "step": 204967
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.412848949432373,
      "learning_rate": 1.7861619932979687e-05,
      "loss": 1.9095,
      "step": 204968
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.5197945833206177,
      "learning_rate": 1.7860229573541772e-05,
      "loss": 2.1043,
      "step": 204969
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.4498603343963623,
      "learning_rate": 1.7858839266559e-05,
      "loss": 2.194,
      "step": 204970
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.4556492567062378,
      "learning_rate": 1.7857449012031565e-05,
      "loss": 2.1467,
      "step": 204971
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.5452226400375366,
      "learning_rate": 1.785605880995974e-05,
      "loss": 2.0968,
      "step": 204972
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.5290818214416504,
      "learning_rate": 1.785466866034382e-05,
      "loss": 2.0435,
      "step": 204973
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.559449553489685,
      "learning_rate": 1.7853278563184004e-05,
      "loss": 1.9598,
      "step": 204974
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.4524297714233398,
      "learning_rate": 1.785188851848063e-05,
      "loss": 1.912,
      "step": 204975
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.439087986946106,
      "learning_rate": 1.7850498526233925e-05,
      "loss": 2.1011,
      "step": 204976
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.3966258764266968,
      "learning_rate": 1.7849108586444095e-05,
      "loss": 2.0125,
      "step": 204977
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.4493037462234497,
      "learning_rate": 1.7847718699111468e-05,
      "loss": 1.9719,
      "step": 204978
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.4753140211105347,
      "learning_rate": 1.7846328864236314e-05,
      "loss": 1.975,
      "step": 204979
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.3834797143936157,
      "learning_rate": 1.7844939081818764e-05,
      "loss": 1.9015,
      "step": 204980
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.4602551460266113,
      "learning_rate": 1.7843549351859255e-05,
      "loss": 1.9071,
      "step": 204981
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.31002938747406,
      "learning_rate": 1.7842159674357913e-05,
      "loss": 1.9215,
      "step": 204982
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.3676784038543701,
      "learning_rate": 1.7840770049315046e-05,
      "loss": 2.0579,
      "step": 204983
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.675723671913147,
      "learning_rate": 1.7839380476730915e-05,
      "loss": 2.0146,
      "step": 204984
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.3710755109786987,
      "learning_rate": 1.7837990956605753e-05,
      "loss": 1.9664,
      "step": 204985
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.457661747932434,
      "learning_rate": 1.783660148893983e-05,
      "loss": 2.1126,
      "step": 204986
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.5803916454315186,
      "learning_rate": 1.783521207373344e-05,
      "loss": 2.1097,
      "step": 204987
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.4134985208511353,
      "learning_rate": 1.783382271098679e-05,
      "loss": 1.879,
      "step": 204988
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.6217198371887207,
      "learning_rate": 1.7832433400700143e-05,
      "loss": 2.086,
      "step": 204989
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.4729785919189453,
      "learning_rate": 1.7831044142873797e-05,
      "loss": 2.0618,
      "step": 204990
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.394482135772705,
      "learning_rate": 1.7829654937507952e-05,
      "loss": 1.9685,
      "step": 204991
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.4356528520584106,
      "learning_rate": 1.7828265784602947e-05,
      "loss": 2.013,
      "step": 204992
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.4232022762298584,
      "learning_rate": 1.7826876684158975e-05,
      "loss": 1.9394,
      "step": 204993
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.5334811210632324,
      "learning_rate": 1.7825487636176338e-05,
      "loss": 1.8591,
      "step": 204994
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.4862134456634521,
      "learning_rate": 1.7824098640655205e-05,
      "loss": 2.2895,
      "step": 204995
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.4557158946990967,
      "learning_rate": 1.7822709697595972e-05,
      "loss": 1.9687,
      "step": 204996
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.4214380979537964,
      "learning_rate": 1.782132080699874e-05,
      "loss": 2.0036,
      "step": 204997
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.4473894834518433,
      "learning_rate": 1.781993196886391e-05,
      "loss": 1.9135,
      "step": 204998
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.4376935958862305,
      "learning_rate": 1.7818543183191647e-05,
      "loss": 2.0345,
      "step": 204999
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.4998557567596436,
      "learning_rate": 1.781715444998235e-05,
      "loss": 2.1409,
      "step": 205000
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.3163373470306396,
      "learning_rate": 1.781576576923606e-05,
      "loss": 1.9173,
      "step": 205001
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.385715126991272,
      "learning_rate": 1.781437714095316e-05,
      "loss": 1.8936,
      "step": 205002
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.4216060638427734,
      "learning_rate": 1.7812988565133902e-05,
      "loss": 2.314,
      "step": 205003
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.3949968814849854,
      "learning_rate": 1.781160004177854e-05,
      "loss": 1.9475,
      "step": 205004
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.4058377742767334,
      "learning_rate": 1.7810211570887312e-05,
      "loss": 1.8359,
      "step": 205005
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.3298848867416382,
      "learning_rate": 1.780882315246058e-05,
      "loss": 1.9201,
      "step": 205006
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.489343523979187,
      "learning_rate": 1.7807434786498387e-05,
      "loss": 2.0226,
      "step": 205007
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.435335636138916,
      "learning_rate": 1.7806046473001224e-05,
      "loss": 1.8624,
      "step": 205008
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.4260656833648682,
      "learning_rate": 1.780465821196916e-05,
      "loss": 1.9026,
      "step": 205009
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.5408178567886353,
      "learning_rate": 1.7803270003402593e-05,
      "loss": 1.8786,
      "step": 205010
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.425781011581421,
      "learning_rate": 1.7801881847301658e-05,
      "loss": 1.846,
      "step": 205011
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.4883698225021362,
      "learning_rate": 1.7800493743666787e-05,
      "loss": 1.9916,
      "step": 205012
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.4606115818023682,
      "learning_rate": 1.779910569249805e-05,
      "loss": 1.8287,
      "step": 205013
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.4705997705459595,
      "learning_rate": 1.779771769379581e-05,
      "loss": 1.8774,
      "step": 205014
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.4027292728424072,
      "learning_rate": 1.7796329747560268e-05,
      "loss": 1.9615,
      "step": 205015
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.3474187850952148,
      "learning_rate": 1.779494185379179e-05,
      "loss": 1.7317,
      "step": 205016
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.4628945589065552,
      "learning_rate": 1.7793554012490473e-05,
      "loss": 1.8729,
      "step": 205017
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.4190635681152344,
      "learning_rate": 1.7792166223656755e-05,
      "loss": 1.8037,
      "step": 205018
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.3726717233657837,
      "learning_rate": 1.7790778487290735e-05,
      "loss": 1.9688,
      "step": 205019
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.47239351272583,
      "learning_rate": 1.7789390803392745e-05,
      "loss": 1.8741,
      "step": 205020
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.4385416507720947,
      "learning_rate": 1.7788003171963018e-05,
      "loss": 2.0192,
      "step": 205021
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.5715835094451904,
      "learning_rate": 1.7786615593001886e-05,
      "loss": 1.9731,
      "step": 205022
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.4845417737960815,
      "learning_rate": 1.778522806650948e-05,
      "loss": 1.8865,
      "step": 205023
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.4311342239379883,
      "learning_rate": 1.7783840592486208e-05,
      "loss": 1.9516,
      "step": 205024
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.5486290454864502,
      "learning_rate": 1.7782453170932166e-05,
      "loss": 1.8958,
      "step": 205025
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.4299774169921875,
      "learning_rate": 1.778106580184775e-05,
      "loss": 2.0242,
      "step": 205026
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.5699055194854736,
      "learning_rate": 1.777967848523313e-05,
      "loss": 2.0444,
      "step": 205027
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.4824934005737305,
      "learning_rate": 1.7778291221088604e-05,
      "loss": 1.732,
      "step": 205028
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.7095956802368164,
      "learning_rate": 1.7776904009414405e-05,
      "loss": 1.8926,
      "step": 205029
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.7060580253601074,
      "learning_rate": 1.777551685021087e-05,
      "loss": 1.7763,
      "step": 205030
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.4834543466567993,
      "learning_rate": 1.7774129743478127e-05,
      "loss": 2.0947,
      "step": 205031
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.513609766960144,
      "learning_rate": 1.777274268921651e-05,
      "loss": 1.8188,
      "step": 205032
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.4970754384994507,
      "learning_rate": 1.777135568742626e-05,
      "loss": 1.8901,
      "step": 205033
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.4321037530899048,
      "learning_rate": 1.7769968738107665e-05,
      "loss": 1.9293,
      "step": 205034
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.3468644618988037,
      "learning_rate": 1.7768581841260932e-05,
      "loss": 1.7736,
      "step": 205035
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.4270732402801514,
      "learning_rate": 1.7767194996886457e-05,
      "loss": 2.017,
      "step": 205036
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.6566522121429443,
      "learning_rate": 1.7765808204984244e-05,
      "loss": 2.0299,
      "step": 205037
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.73714017868042,
      "learning_rate": 1.776442146555479e-05,
      "loss": 1.9444,
      "step": 205038
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.4997678995132446,
      "learning_rate": 1.7763034778598196e-05,
      "loss": 1.9432,
      "step": 205039
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.5523431301116943,
      "learning_rate": 1.7761648144114793e-05,
      "loss": 1.9546,
      "step": 205040
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.512325406074524,
      "learning_rate": 1.7760261562104816e-05,
      "loss": 1.8054,
      "step": 205041
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.3193817138671875,
      "learning_rate": 1.7758875032568632e-05,
      "loss": 2.1328,
      "step": 205042
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.4393450021743774,
      "learning_rate": 1.7757488555506305e-05,
      "loss": 1.9816,
      "step": 205043
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.4678231477737427,
      "learning_rate": 1.7756102130918205e-05,
      "loss": 1.9791,
      "step": 205044
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.5100198984146118,
      "learning_rate": 1.775471575880456e-05,
      "loss": 1.9403,
      "step": 205045
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.393850564956665,
      "learning_rate": 1.7753329439165676e-05,
      "loss": 1.9912,
      "step": 205046
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.5172137022018433,
      "learning_rate": 1.775194317200175e-05,
      "loss": 2.0261,
      "step": 205047
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.447495460510254,
      "learning_rate": 1.7750556957313078e-05,
      "loss": 1.89,
      "step": 205048
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.4904890060424805,
      "learning_rate": 1.7749170795099898e-05,
      "loss": 2.0389,
      "step": 205049
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.4018335342407227,
      "learning_rate": 1.774778468536251e-05,
      "loss": 1.9438,
      "step": 205050
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.4045062065124512,
      "learning_rate": 1.7746398628101078e-05,
      "loss": 1.8926,
      "step": 205051
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.5495353937149048,
      "learning_rate": 1.7745012623315967e-05,
      "loss": 2.0864,
      "step": 205052
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.330029010772705,
      "learning_rate": 1.774362667100735e-05,
      "loss": 1.9512,
      "step": 205053
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.3795380592346191,
      "learning_rate": 1.7742240771175553e-05,
      "loss": 1.7997,
      "step": 205054
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.5016738176345825,
      "learning_rate": 1.774085492382078e-05,
      "loss": 1.9411,
      "step": 205055
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.7281920909881592,
      "learning_rate": 1.7739469128943328e-05,
      "loss": 2.0349,
      "step": 205056
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.5275838375091553,
      "learning_rate": 1.77380833865434e-05,
      "loss": 1.756,
      "step": 205057
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.6236072778701782,
      "learning_rate": 1.7736697696621327e-05,
      "loss": 2.0739,
      "step": 205058
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.4347326755523682,
      "learning_rate": 1.773531205917731e-05,
      "loss": 2.1297,
      "step": 205059
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.4977145195007324,
      "learning_rate": 1.7733926474211647e-05,
      "loss": 2.1133,
      "step": 205060
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.4541043043136597,
      "learning_rate": 1.7732540941724573e-05,
      "loss": 1.8746,
      "step": 205061
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.4375227689743042,
      "learning_rate": 1.773115546171632e-05,
      "loss": 2.1045,
      "step": 205062
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.576338768005371,
      "learning_rate": 1.7729770034187184e-05,
      "loss": 2.0592,
      "step": 205063
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.3893229961395264,
      "learning_rate": 1.772838465913744e-05,
      "loss": 1.9392,
      "step": 205064
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.439191222190857,
      "learning_rate": 1.7726999336567283e-05,
      "loss": 2.0935,
      "step": 205065
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.4111920595169067,
      "learning_rate": 1.7725614066477013e-05,
      "loss": 1.9549,
      "step": 205066
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.4301801919937134,
      "learning_rate": 1.7724228848866894e-05,
      "loss": 2.0667,
      "step": 205067
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.9358116388320923,
      "learning_rate": 1.772284368373713e-05,
      "loss": 1.9325,
      "step": 205068
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.6283315420150757,
      "learning_rate": 1.7721458571088087e-05,
      "loss": 1.9639,
      "step": 205069
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.400718331336975,
      "learning_rate": 1.772007351091993e-05,
      "loss": 2.0726,
      "step": 205070
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.441457986831665,
      "learning_rate": 1.7718688503232893e-05,
      "loss": 1.9843,
      "step": 205071
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.4063339233398438,
      "learning_rate": 1.771730354802734e-05,
      "loss": 2.0285,
      "step": 205072
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.5654916763305664,
      "learning_rate": 1.771591864530344e-05,
      "loss": 1.6303,
      "step": 205073
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.3432339429855347,
      "learning_rate": 1.771453379506146e-05,
      "loss": 1.815,
      "step": 205074
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.4111799001693726,
      "learning_rate": 1.7713148997301698e-05,
      "loss": 1.9802,
      "step": 205075
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.5675290822982788,
      "learning_rate": 1.771176425202435e-05,
      "loss": 1.9762,
      "step": 205076
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.4097588062286377,
      "learning_rate": 1.771037955922979e-05,
      "loss": 1.9191,
      "step": 205077
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.357677698135376,
      "learning_rate": 1.7708994918918183e-05,
      "loss": 2.0436,
      "step": 205078
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.3732110261917114,
      "learning_rate": 1.7707610331089794e-05,
      "loss": 1.9817,
      "step": 205079
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.3688344955444336,
      "learning_rate": 1.7706225795744823e-05,
      "loss": 2.0424,
      "step": 205080
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.58016836643219,
      "learning_rate": 1.7704841312883667e-05,
      "loss": 2.0324,
      "step": 205081
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.4713160991668701,
      "learning_rate": 1.770345688250646e-05,
      "loss": 1.8608,
      "step": 205082
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.48173189163208,
      "learning_rate": 1.7702072504613574e-05,
      "loss": 2.1024,
      "step": 205083
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.5164920091629028,
      "learning_rate": 1.7700688179205136e-05,
      "loss": 2.1067,
      "step": 205084
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.4757543802261353,
      "learning_rate": 1.7699303906281547e-05,
      "loss": 1.9551,
      "step": 205085
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.3947193622589111,
      "learning_rate": 1.769791968584291e-05,
      "loss": 1.9822,
      "step": 205086
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.4492313861846924,
      "learning_rate": 1.7696535517889586e-05,
      "loss": 1.9912,
      "step": 205087
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.3264505863189697,
      "learning_rate": 1.769515140242178e-05,
      "loss": 1.8282,
      "step": 205088
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.3587793111801147,
      "learning_rate": 1.7693767339439824e-05,
      "loss": 1.9885,
      "step": 205089
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.5643057823181152,
      "learning_rate": 1.7692383328943883e-05,
      "loss": 2.0343,
      "step": 205090
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.3897278308868408,
      "learning_rate": 1.7690999370934323e-05,
      "loss": 1.8543,
      "step": 205091
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.3872625827789307,
      "learning_rate": 1.7689615465411243e-05,
      "loss": 2.0021,
      "step": 205092
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.499558687210083,
      "learning_rate": 1.7688231612375083e-05,
      "loss": 1.92,
      "step": 205093
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.4003899097442627,
      "learning_rate": 1.768684781182593e-05,
      "loss": 1.809,
      "step": 205094
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.4421249628067017,
      "learning_rate": 1.7685464063764166e-05,
      "loss": 2.0137,
      "step": 205095
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.480865240097046,
      "learning_rate": 1.7684080368189946e-05,
      "loss": 2.0174,
      "step": 205096
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.2603230476379395,
      "learning_rate": 1.7682696725103708e-05,
      "loss": 1.9754,
      "step": 205097
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.4962685108184814,
      "learning_rate": 1.7681313134505483e-05,
      "loss": 1.8171,
      "step": 205098
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.3744131326675415,
      "learning_rate": 1.767992959639567e-05,
      "loss": 2.0371,
      "step": 205099
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.4666961431503296,
      "learning_rate": 1.767854611077444e-05,
      "loss": 1.9477,
      "step": 205100
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.4485357999801636,
      "learning_rate": 1.7677162677642153e-05,
      "loss": 1.9919,
      "step": 205101
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.517633318901062,
      "learning_rate": 1.7675779296998947e-05,
      "loss": 1.9404,
      "step": 205102
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.4796578884124756,
      "learning_rate": 1.7674395968845258e-05,
      "loss": 1.986,
      "step": 205103
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.4122846126556396,
      "learning_rate": 1.767301269318111e-05,
      "loss": 2.0797,
      "step": 205104
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.3371614217758179,
      "learning_rate": 1.7671629470006942e-05,
      "loss": 2.0182,
      "step": 205105
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.564893364906311,
      "learning_rate": 1.7670246299322888e-05,
      "loss": 1.9438,
      "step": 205106
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.4861156940460205,
      "learning_rate": 1.7668863181129344e-05,
      "loss": 2.1321,
      "step": 205107
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.5721309185028076,
      "learning_rate": 1.7667480115426415e-05,
      "loss": 2.1668,
      "step": 205108
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.4149264097213745,
      "learning_rate": 1.7666097102214527e-05,
      "loss": 1.8647,
      "step": 205109
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.3385765552520752,
      "learning_rate": 1.7664714141493752e-05,
      "loss": 1.8938,
      "step": 205110
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.6317819356918335,
      "learning_rate": 1.7663331233264456e-05,
      "loss": 1.9003,
      "step": 205111
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.5202866792678833,
      "learning_rate": 1.766194837752687e-05,
      "loss": 1.7369,
      "step": 205112
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.4720239639282227,
      "learning_rate": 1.7660565574281294e-05,
      "loss": 1.8058,
      "step": 205113
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.5049543380737305,
      "learning_rate": 1.7659182823527894e-05,
      "loss": 2.0003,
      "step": 205114
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.411749005317688,
      "learning_rate": 1.7657800125267007e-05,
      "loss": 2.0125,
      "step": 205115
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.635216236114502,
      "learning_rate": 1.7656417479498896e-05,
      "loss": 2.0826,
      "step": 205116
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.8122590780258179,
      "learning_rate": 1.7655034886223795e-05,
      "loss": 1.9084,
      "step": 205117
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.4637075662612915,
      "learning_rate": 1.765365234544187e-05,
      "loss": 1.9916,
      "step": 205118
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.4362244606018066,
      "learning_rate": 1.7652269857153523e-05,
      "loss": 2.1631,
      "step": 205119
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.459936261177063,
      "learning_rate": 1.7650887421358917e-05,
      "loss": 1.828,
      "step": 205120
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.496206283569336,
      "learning_rate": 1.7649505038058354e-05,
      "loss": 1.9335,
      "step": 205121
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.4874780178070068,
      "learning_rate": 1.76481227072521e-05,
      "loss": 1.989,
      "step": 205122
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.5055389404296875,
      "learning_rate": 1.7646740428940386e-05,
      "loss": 2.0967,
      "step": 205123
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.4064416885375977,
      "learning_rate": 1.7645358203123448e-05,
      "loss": 1.7455,
      "step": 205124
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.3885198831558228,
      "learning_rate": 1.7643976029801587e-05,
      "loss": 1.793,
      "step": 205125
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.39485764503479,
      "learning_rate": 1.7642593908975e-05,
      "loss": 1.8323,
      "step": 205126
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.3254210948944092,
      "learning_rate": 1.7641211840644055e-05,
      "loss": 1.8499,
      "step": 205127
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.5278247594833374,
      "learning_rate": 1.7639829824808916e-05,
      "loss": 2.1208,
      "step": 205128
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.508914828300476,
      "learning_rate": 1.763844786146985e-05,
      "loss": 2.1226,
      "step": 205129
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.4666398763656616,
      "learning_rate": 1.7637065950627095e-05,
      "loss": 1.9783,
      "step": 205130
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.4432142972946167,
      "learning_rate": 1.7635684092280978e-05,
      "loss": 2.072,
      "step": 205131
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.4558603763580322,
      "learning_rate": 1.763430228643167e-05,
      "loss": 2.0079,
      "step": 205132
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.4261960983276367,
      "learning_rate": 1.763292053307953e-05,
      "loss": 1.9004,
      "step": 205133
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.5839077234268188,
      "learning_rate": 1.7631538832224768e-05,
      "loss": 1.9521,
      "step": 205134
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.3810319900512695,
      "learning_rate": 1.7630157183867643e-05,
      "loss": 1.9267,
      "step": 205135
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.486981749534607,
      "learning_rate": 1.7628775588008325e-05,
      "loss": 1.9823,
      "step": 205136
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.4031100273132324,
      "learning_rate": 1.7627394044647215e-05,
      "loss": 2.0291,
      "step": 205137
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.3374338150024414,
      "learning_rate": 1.7626012553784442e-05,
      "loss": 1.9308,
      "step": 205138
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.360142469406128,
      "learning_rate": 1.7624631115420374e-05,
      "loss": 1.9304,
      "step": 205139
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.4237124919891357,
      "learning_rate": 1.7623249729555212e-05,
      "loss": 2.0,
      "step": 205140
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.4550659656524658,
      "learning_rate": 1.762186839618922e-05,
      "loss": 1.9969,
      "step": 205141
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.424286127090454,
      "learning_rate": 1.7620487115322634e-05,
      "loss": 2.0514,
      "step": 205142
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.4262665510177612,
      "learning_rate": 1.7619105886955754e-05,
      "loss": 2.1182,
      "step": 205143
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.5595182180404663,
      "learning_rate": 1.7617724711088745e-05,
      "loss": 1.9007,
      "step": 205144
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.459926962852478,
      "learning_rate": 1.7616343587722003e-05,
      "loss": 1.9676,
      "step": 205145
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.4820517301559448,
      "learning_rate": 1.7614962516855702e-05,
      "loss": 2.0034,
      "step": 205146
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.5034466981887817,
      "learning_rate": 1.7613581498490136e-05,
      "loss": 2.2202,
      "step": 205147
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.6291425228118896,
      "learning_rate": 1.761220053262544e-05,
      "loss": 2.0061,
      "step": 205148
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.4892219305038452,
      "learning_rate": 1.761081961926205e-05,
      "loss": 2.04,
      "step": 205149
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.5635530948638916,
      "learning_rate": 1.7609438758400096e-05,
      "loss": 2.0349,
      "step": 205150
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.4606832265853882,
      "learning_rate": 1.760805795003991e-05,
      "loss": 1.7896,
      "step": 205151
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.4949798583984375,
      "learning_rate": 1.760667719418173e-05,
      "loss": 2.0706,
      "step": 205152
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.375435709953308,
      "learning_rate": 1.7605296490825748e-05,
      "loss": 1.9204,
      "step": 205153
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.577736258506775,
      "learning_rate": 1.7603915839972305e-05,
      "loss": 2.0648,
      "step": 205154
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.5243104696273804,
      "learning_rate": 1.760253524162163e-05,
      "loss": 2.1104,
      "step": 205155
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.5686402320861816,
      "learning_rate": 1.760115469577392e-05,
      "loss": 1.9234,
      "step": 205156
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.5150214433670044,
      "learning_rate": 1.759977420242955e-05,
      "loss": 2.0679,
      "step": 205157
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.3837535381317139,
      "learning_rate": 1.7598393761588714e-05,
      "loss": 1.952,
      "step": 205158
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.3948543071746826,
      "learning_rate": 1.759701337325161e-05,
      "loss": 2.1045,
      "step": 205159
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.3522974252700806,
      "learning_rate": 1.759563303741861e-05,
      "loss": 1.7376,
      "step": 205160
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.521309733390808,
      "learning_rate": 1.7594252754089842e-05,
      "loss": 2.0925,
      "step": 205161
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.3487401008605957,
      "learning_rate": 1.7592872523265712e-05,
      "loss": 2.088,
      "step": 205162
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.3516087532043457,
      "learning_rate": 1.759149234494638e-05,
      "loss": 1.974,
      "step": 205163
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.3144129514694214,
      "learning_rate": 1.759011221913211e-05,
      "loss": 2.0942,
      "step": 205164
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.4702237844467163,
      "learning_rate": 1.7588732145823146e-05,
      "loss": 2.1508,
      "step": 205165
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.4782922267913818,
      "learning_rate": 1.7587352125019782e-05,
      "loss": 1.9856,
      "step": 205166
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.5194951295852661,
      "learning_rate": 1.758597215672225e-05,
      "loss": 2.082,
      "step": 205167
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.5807604789733887,
      "learning_rate": 1.758459224093085e-05,
      "loss": 1.9205,
      "step": 205168
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.578910231590271,
      "learning_rate": 1.7583212377645785e-05,
      "loss": 1.9617,
      "step": 205169
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.5258325338363647,
      "learning_rate": 1.758183256686735e-05,
      "loss": 1.9174,
      "step": 205170
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.4365390539169312,
      "learning_rate": 1.7580452808595745e-05,
      "loss": 1.9283,
      "step": 205171
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.4663527011871338,
      "learning_rate": 1.757907310283131e-05,
      "loss": 2.1215,
      "step": 205172
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.3183377981185913,
      "learning_rate": 1.7577693449574202e-05,
      "loss": 1.9863,
      "step": 205173
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.3379243612289429,
      "learning_rate": 1.7576313848824763e-05,
      "loss": 1.7151,
      "step": 205174
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.5727646350860596,
      "learning_rate": 1.7574934300583187e-05,
      "loss": 2.0585,
      "step": 205175
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.566674828529358,
      "learning_rate": 1.7573554804849878e-05,
      "loss": 1.8328,
      "step": 205176
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.394487977027893,
      "learning_rate": 1.7572175361624863e-05,
      "loss": 1.9392,
      "step": 205177
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.4889098405838013,
      "learning_rate": 1.7570795970908545e-05,
      "loss": 2.2192,
      "step": 205178
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.364123821258545,
      "learning_rate": 1.7569416632701128e-05,
      "loss": 1.8055,
      "step": 205179
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.4995683431625366,
      "learning_rate": 1.7568037347002906e-05,
      "loss": 1.7722,
      "step": 205180
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.4615687131881714,
      "learning_rate": 1.756665811381408e-05,
      "loss": 2.0312,
      "step": 205181
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.4241355657577515,
      "learning_rate": 1.7565278933134984e-05,
      "loss": 1.9517,
      "step": 205182
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.4695366621017456,
      "learning_rate": 1.7563899804965852e-05,
      "loss": 2.2741,
      "step": 205183
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.3858543634414673,
      "learning_rate": 1.7562520729306916e-05,
      "loss": 2.011,
      "step": 205184
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.8007835149765015,
      "learning_rate": 1.7561141706158377e-05,
      "loss": 1.7576,
      "step": 205185
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.4411630630493164,
      "learning_rate": 1.7559762735520633e-05,
      "loss": 1.9401,
      "step": 205186
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.4770711660385132,
      "learning_rate": 1.7558383817393785e-05,
      "loss": 2.01,
      "step": 205187
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.3735182285308838,
      "learning_rate": 1.755700495177823e-05,
      "loss": 1.9891,
      "step": 205188
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.362229347229004,
      "learning_rate": 1.755562613867414e-05,
      "loss": 2.0401,
      "step": 205189
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.5033825635910034,
      "learning_rate": 1.755424737808181e-05,
      "loss": 2.0693,
      "step": 205190
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.3888087272644043,
      "learning_rate": 1.7552868670001406e-05,
      "loss": 1.9403,
      "step": 205191
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.461986780166626,
      "learning_rate": 1.7551490014433335e-05,
      "loss": 1.9294,
      "step": 205192
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.540958046913147,
      "learning_rate": 1.755011141137769e-05,
      "loss": 2.115,
      "step": 205193
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.5355502367019653,
      "learning_rate": 1.754873286083487e-05,
      "loss": 2.1622,
      "step": 205194
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.408550500869751,
      "learning_rate": 1.7547354362805077e-05,
      "loss": 1.9306,
      "step": 205195
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.4768294095993042,
      "learning_rate": 1.754597591728858e-05,
      "loss": 1.9358,
      "step": 205196
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.4911998510360718,
      "learning_rate": 1.754459752428554e-05,
      "loss": 1.9753,
      "step": 205197
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.464254379272461,
      "learning_rate": 1.7543219183796365e-05,
      "loss": 2.0768,
      "step": 205198
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.5153108835220337,
      "learning_rate": 1.754184089582118e-05,
      "loss": 1.7432,
      "step": 205199
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.4620864391326904,
      "learning_rate": 1.7540462660360355e-05,
      "loss": 2.3129,
      "step": 205200
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.4950546026229858,
      "learning_rate": 1.7539084477414057e-05,
      "loss": 1.826,
      "step": 205201
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.448175311088562,
      "learning_rate": 1.7537706346982615e-05,
      "loss": 2.0467,
      "step": 205202
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.4715518951416016,
      "learning_rate": 1.753632826906617e-05,
      "loss": 1.9097,
      "step": 205203
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.5193625688552856,
      "learning_rate": 1.7534950243665115e-05,
      "loss": 2.3573,
      "step": 205204
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.43291175365448,
      "learning_rate": 1.7533572270779582e-05,
      "loss": 2.0428,
      "step": 205205
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.4867655038833618,
      "learning_rate": 1.7532194350409946e-05,
      "loss": 1.9214,
      "step": 205206
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.433206558227539,
      "learning_rate": 1.753081648255643e-05,
      "loss": 1.9089,
      "step": 205207
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.5922300815582275,
      "learning_rate": 1.7529438667219243e-05,
      "loss": 2.093,
      "step": 205208
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.4368356466293335,
      "learning_rate": 1.7528060904398612e-05,
      "loss": 2.0343,
      "step": 205209
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.3512684106826782,
      "learning_rate": 1.7526683194094902e-05,
      "loss": 1.9784,
      "step": 205210
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.3650988340377808,
      "learning_rate": 1.7525305536308286e-05,
      "loss": 1.6604,
      "step": 205211
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.4663878679275513,
      "learning_rate": 1.752392793103906e-05,
      "loss": 1.8794,
      "step": 205212
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.4964200258255005,
      "learning_rate": 1.7522550378287493e-05,
      "loss": 1.9771,
      "step": 205213
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.4836177825927734,
      "learning_rate": 1.7521172878053812e-05,
      "loss": 1.9603,
      "step": 205214
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.4410014152526855,
      "learning_rate": 1.7519795430338224e-05,
      "loss": 2.0995,
      "step": 205215
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.39000403881073,
      "learning_rate": 1.751841803514109e-05,
      "loss": 1.9206,
      "step": 205216
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.5787301063537598,
      "learning_rate": 1.7517040692462548e-05,
      "loss": 1.7851,
      "step": 205217
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.4730862379074097,
      "learning_rate": 1.7515663402302992e-05,
      "loss": 1.9199,
      "step": 205218
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.3852075338363647,
      "learning_rate": 1.7514286164662593e-05,
      "loss": 2.0582,
      "step": 205219
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.385895848274231,
      "learning_rate": 1.7512908979541618e-05,
      "loss": 1.7317,
      "step": 205220
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.4925307035446167,
      "learning_rate": 1.7511531846940297e-05,
      "loss": 2.1729,
      "step": 205221
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.448531150817871,
      "learning_rate": 1.7510154766858963e-05,
      "loss": 1.8233,
      "step": 205222
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.5144225358963013,
      "learning_rate": 1.7508777739297754e-05,
      "loss": 1.6973,
      "step": 205223
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.6079342365264893,
      "learning_rate": 1.7507400764257064e-05,
      "loss": 2.0661,
      "step": 205224
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.4347671270370483,
      "learning_rate": 1.750602384173706e-05,
      "loss": 2.0472,
      "step": 205225
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.5194357633590698,
      "learning_rate": 1.7504646971738045e-05,
      "loss": 1.9911,
      "step": 205226
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.392529845237732,
      "learning_rate": 1.7503270154260184e-05,
      "loss": 2.0074,
      "step": 205227
    },
    {
      "epoch": 2.67,
      "grad_norm": 4.860893249511719,
      "learning_rate": 1.7501893389303844e-05,
      "loss": 1.9061,
      "step": 205228
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.3575222492218018,
      "learning_rate": 1.750051667686919e-05,
      "loss": 1.8999,
      "step": 205229
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.5335583686828613,
      "learning_rate": 1.749914001695656e-05,
      "loss": 1.9501,
      "step": 205230
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.6087114810943604,
      "learning_rate": 1.749776340956618e-05,
      "loss": 1.9046,
      "step": 205231
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.4467931985855103,
      "learning_rate": 1.7496386854698286e-05,
      "loss": 2.0199,
      "step": 205232
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.623130440711975,
      "learning_rate": 1.7495010352353145e-05,
      "loss": 2.1258,
      "step": 205233
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.6398437023162842,
      "learning_rate": 1.7493633902531022e-05,
      "loss": 1.9359,
      "step": 205234
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.5462629795074463,
      "learning_rate": 1.749225750523212e-05,
      "loss": 1.8417,
      "step": 205235
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.435019850730896,
      "learning_rate": 1.7490881160456806e-05,
      "loss": 1.955,
      "step": 205236
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.506068468093872,
      "learning_rate": 1.748950486820524e-05,
      "loss": 2.0131,
      "step": 205237
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.710545539855957,
      "learning_rate": 1.7488128628477694e-05,
      "loss": 2.0527,
      "step": 205238
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.4721466302871704,
      "learning_rate": 1.7486752441274464e-05,
      "loss": 1.8613,
      "step": 205239
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.7252901792526245,
      "learning_rate": 1.7485376306595787e-05,
      "loss": 2.0746,
      "step": 205240
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.5394374132156372,
      "learning_rate": 1.7484000224441865e-05,
      "loss": 2.0169,
      "step": 205241
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.477006196975708,
      "learning_rate": 1.748262419481302e-05,
      "loss": 1.9812,
      "step": 205242
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.485155463218689,
      "learning_rate": 1.748124821770953e-05,
      "loss": 2.2022,
      "step": 205243
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.6078931093215942,
      "learning_rate": 1.7479872293131525e-05,
      "loss": 1.9975,
      "step": 205244
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.4263432025909424,
      "learning_rate": 1.7478496421079436e-05,
      "loss": 1.8273,
      "step": 205245
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.647987961769104,
      "learning_rate": 1.747712060155333e-05,
      "loss": 2.0593,
      "step": 205246
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.5038446187973022,
      "learning_rate": 1.747574483455364e-05,
      "loss": 1.9485,
      "step": 205247
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.3581210374832153,
      "learning_rate": 1.7474369120080532e-05,
      "loss": 1.9051,
      "step": 205248
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.3299574851989746,
      "learning_rate": 1.7472993458134244e-05,
      "loss": 1.8858,
      "step": 205249
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.412704348564148,
      "learning_rate": 1.747161784871507e-05,
      "loss": 2.0286,
      "step": 205250
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.6364712715148926,
      "learning_rate": 1.7470242291823276e-05,
      "loss": 1.8368,
      "step": 205251
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.5350399017333984,
      "learning_rate": 1.746886678745907e-05,
      "loss": 1.9466,
      "step": 205252
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.3447484970092773,
      "learning_rate": 1.746749133562274e-05,
      "loss": 1.922,
      "step": 205253
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.4987024068832397,
      "learning_rate": 1.7466115936314563e-05,
      "loss": 1.9367,
      "step": 205254
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.3225035667419434,
      "learning_rate": 1.746474058953473e-05,
      "loss": 1.9492,
      "step": 205255
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.2985459566116333,
      "learning_rate": 1.746336529528355e-05,
      "loss": 1.8642,
      "step": 205256
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.4865355491638184,
      "learning_rate": 1.746199005356128e-05,
      "loss": 1.9247,
      "step": 205257
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.649773359298706,
      "learning_rate": 1.746061486436813e-05,
      "loss": 2.0833,
      "step": 205258
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.5470311641693115,
      "learning_rate": 1.745923972770442e-05,
      "loss": 2.0135,
      "step": 205259
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.5320965051651,
      "learning_rate": 1.7457864643570296e-05,
      "loss": 1.9698,
      "step": 205260
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.550710678100586,
      "learning_rate": 1.7456489611966184e-05,
      "loss": 1.906,
      "step": 205261
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.6121196746826172,
      "learning_rate": 1.745511463289222e-05,
      "loss": 2.0501,
      "step": 205262
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.47726309299469,
      "learning_rate": 1.7453739706348668e-05,
      "loss": 1.9749,
      "step": 205263
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.4796921014785767,
      "learning_rate": 1.7452364832335764e-05,
      "loss": 1.8663,
      "step": 205264
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.4232535362243652,
      "learning_rate": 1.7450990010853872e-05,
      "loss": 1.9911,
      "step": 205265
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.4673315286636353,
      "learning_rate": 1.7449615241903093e-05,
      "loss": 2.1319,
      "step": 205266
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.5861611366271973,
      "learning_rate": 1.7448240525483824e-05,
      "loss": 2.0222,
      "step": 205267
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.3337205648422241,
      "learning_rate": 1.744686586159627e-05,
      "loss": 1.7959,
      "step": 205268
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.5416655540466309,
      "learning_rate": 1.744549125024066e-05,
      "loss": 2.1115,
      "step": 205269
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.4975154399871826,
      "learning_rate": 1.744411669141723e-05,
      "loss": 1.7629,
      "step": 205270
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.507855772972107,
      "learning_rate": 1.744274218512631e-05,
      "loss": 1.928,
      "step": 205271
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.382726788520813,
      "learning_rate": 1.7441367731368073e-05,
      "loss": 1.7827,
      "step": 205272
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.4351725578308105,
      "learning_rate": 1.7439993330142875e-05,
      "loss": 1.8915,
      "step": 205273
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.340142011642456,
      "learning_rate": 1.743861898145089e-05,
      "loss": 1.9458,
      "step": 205274
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.4370335340499878,
      "learning_rate": 1.7437244685292416e-05,
      "loss": 1.5846,
      "step": 205275
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.6860134601593018,
      "learning_rate": 1.7435870441667653e-05,
      "loss": 1.9106,
      "step": 205276
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.6468634605407715,
      "learning_rate": 1.7434496250576967e-05,
      "loss": 1.8786,
      "step": 205277
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.5406464338302612,
      "learning_rate": 1.7433122112020457e-05,
      "loss": 1.9778,
      "step": 205278
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.4431753158569336,
      "learning_rate": 1.7431748025998527e-05,
      "loss": 1.8858,
      "step": 205279
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.5930951833724976,
      "learning_rate": 1.7430373992511338e-05,
      "loss": 1.9175,
      "step": 205280
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.4950567483901978,
      "learning_rate": 1.7429000011559223e-05,
      "loss": 1.95,
      "step": 205281
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.4863574504852295,
      "learning_rate": 1.742762608314232e-05,
      "loss": 1.903,
      "step": 205282
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.4715744256973267,
      "learning_rate": 1.7426252207260995e-05,
      "loss": 1.8984,
      "step": 205283
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.4779443740844727,
      "learning_rate": 1.742487838391544e-05,
      "loss": 1.9919,
      "step": 205284
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.4944546222686768,
      "learning_rate": 1.7423504613105965e-05,
      "loss": 2.0692,
      "step": 205285
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.5971578359603882,
      "learning_rate": 1.7422130894832798e-05,
      "loss": 2.0349,
      "step": 205286
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.4450008869171143,
      "learning_rate": 1.742075722909617e-05,
      "loss": 2.2553,
      "step": 205287
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.501980185508728,
      "learning_rate": 1.7419383615896353e-05,
      "loss": 2.0235,
      "step": 205288
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.517707347869873,
      "learning_rate": 1.741801005523361e-05,
      "loss": 1.8644,
      "step": 205289
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.5111360549926758,
      "learning_rate": 1.7416636547108176e-05,
      "loss": 2.0053,
      "step": 205290
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.5671031475067139,
      "learning_rate": 1.741526309152035e-05,
      "loss": 1.8941,
      "step": 205291
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.4691088199615479,
      "learning_rate": 1.741388968847036e-05,
      "loss": 1.8563,
      "step": 205292
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.3648250102996826,
      "learning_rate": 1.7412516337958482e-05,
      "loss": 1.9833,
      "step": 205293
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.4416719675064087,
      "learning_rate": 1.741114303998491e-05,
      "loss": 2.0337,
      "step": 205294
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.4553457498550415,
      "learning_rate": 1.740976979454998e-05,
      "loss": 2.1333,
      "step": 205295
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.7438417673110962,
      "learning_rate": 1.7408396601653852e-05,
      "loss": 2.1412,
      "step": 205296
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.3771233558654785,
      "learning_rate": 1.7407023461296865e-05,
      "loss": 1.9462,
      "step": 205297
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.4291640520095825,
      "learning_rate": 1.7405650373479285e-05,
      "loss": 2.0365,
      "step": 205298
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.3311861753463745,
      "learning_rate": 1.7404277338201312e-05,
      "loss": 1.8334,
      "step": 205299
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.5244396924972534,
      "learning_rate": 1.7402904355463177e-05,
      "loss": 1.8679,
      "step": 205300
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.3800666332244873,
      "learning_rate": 1.7401531425265215e-05,
      "loss": 1.8783,
      "step": 205301
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.3695894479751587,
      "learning_rate": 1.740015854760759e-05,
      "loss": 1.8844,
      "step": 205302
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.4932652711868286,
      "learning_rate": 1.739878572249067e-05,
      "loss": 1.9968,
      "step": 205303
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.516014575958252,
      "learning_rate": 1.7397412949914657e-05,
      "loss": 1.8936,
      "step": 205304
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.4707903861999512,
      "learning_rate": 1.7396040229879782e-05,
      "loss": 1.8474,
      "step": 205305
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.488172173500061,
      "learning_rate": 1.7394667562386276e-05,
      "loss": 2.1142,
      "step": 205306
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.4193295240402222,
      "learning_rate": 1.739329494743448e-05,
      "loss": 2.1787,
      "step": 205307
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.4263174533843994,
      "learning_rate": 1.739192238502458e-05,
      "loss": 2.0632,
      "step": 205308
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.3516041040420532,
      "learning_rate": 1.739054987515689e-05,
      "loss": 2.0329,
      "step": 205309
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.4817860126495361,
      "learning_rate": 1.7389177417831602e-05,
      "loss": 2.0123,
      "step": 205310
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.4814702272415161,
      "learning_rate": 1.738780501304905e-05,
      "loss": 1.8496,
      "step": 205311
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.5582501888275146,
      "learning_rate": 1.738643266080937e-05,
      "loss": 1.8994,
      "step": 205312
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.3470096588134766,
      "learning_rate": 1.7385060361112925e-05,
      "loss": 1.8259,
      "step": 205313
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.5281413793563843,
      "learning_rate": 1.7383688113959882e-05,
      "loss": 2.0359,
      "step": 205314
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.4553906917572021,
      "learning_rate": 1.738231591935061e-05,
      "loss": 2.0406,
      "step": 205315
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.5074164867401123,
      "learning_rate": 1.7380943777285238e-05,
      "loss": 1.9902,
      "step": 205316
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.3946393728256226,
      "learning_rate": 1.7379571687764203e-05,
      "loss": 2.0245,
      "step": 205317
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.604463815689087,
      "learning_rate": 1.737819965078754e-05,
      "loss": 2.1655,
      "step": 205318
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.3631280660629272,
      "learning_rate": 1.7376827666355642e-05,
      "loss": 1.9577,
      "step": 205319
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.4680432081222534,
      "learning_rate": 1.737545573446868e-05,
      "loss": 2.2358,
      "step": 205320
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.3349928855895996,
      "learning_rate": 1.7374083855126984e-05,
      "loss": 2.0378,
      "step": 205321
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.3098989725112915,
      "learning_rate": 1.737271202833076e-05,
      "loss": 1.9424,
      "step": 205322
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.4319441318511963,
      "learning_rate": 1.7371340254080333e-05,
      "loss": 1.9119,
      "step": 205323
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.4382684230804443,
      "learning_rate": 1.7369968532375878e-05,
      "loss": 2.0684,
      "step": 205324
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.721299171447754,
      "learning_rate": 1.736859686321772e-05,
      "loss": 2.0735,
      "step": 205325
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.4329930543899536,
      "learning_rate": 1.7367225246606e-05,
      "loss": 1.9174,
      "step": 205326
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.4222038984298706,
      "learning_rate": 1.7365853682541108e-05,
      "loss": 1.885,
      "step": 205327
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.4003738164901733,
      "learning_rate": 1.7364482171023186e-05,
      "loss": 1.8851,
      "step": 205328
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.5424354076385498,
      "learning_rate": 1.7363110712052596e-05,
      "loss": 1.925,
      "step": 205329
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.3162685632705688,
      "learning_rate": 1.7361739305629542e-05,
      "loss": 1.9338,
      "step": 205330
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.4357678890228271,
      "learning_rate": 1.736036795175425e-05,
      "loss": 1.9697,
      "step": 205331
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.4584391117095947,
      "learning_rate": 1.735899665042699e-05,
      "loss": 2.0194,
      "step": 205332
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.4274312257766724,
      "learning_rate": 1.7357625401648034e-05,
      "loss": 2.0083,
      "step": 205333
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.4737271070480347,
      "learning_rate": 1.735625420541761e-05,
      "loss": 1.9867,
      "step": 205334
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.5231705904006958,
      "learning_rate": 1.7354883061736046e-05,
      "loss": 2.1428,
      "step": 205335
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.4378706216812134,
      "learning_rate": 1.735351197060355e-05,
      "loss": 2.0171,
      "step": 205336
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.4375728368759155,
      "learning_rate": 1.7352140932020285e-05,
      "loss": 2.0156,
      "step": 205337
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.3480700254440308,
      "learning_rate": 1.7350769945986685e-05,
      "loss": 2.0419,
      "step": 205338
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.397655725479126,
      "learning_rate": 1.7349399012502885e-05,
      "loss": 1.9698,
      "step": 205339
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.3815799951553345,
      "learning_rate": 1.7348028131569113e-05,
      "loss": 2.2645,
      "step": 205340
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.495728611946106,
      "learning_rate": 1.734665730318574e-05,
      "loss": 2.0031,
      "step": 205341
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.500527262687683,
      "learning_rate": 1.7345286527352965e-05,
      "loss": 1.9368,
      "step": 205342
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.4670906066894531,
      "learning_rate": 1.7343915804070985e-05,
      "loss": 2.0305,
      "step": 205343
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.9093141555786133,
      "learning_rate": 1.734254513334017e-05,
      "loss": 2.2543,
      "step": 205344
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.511325716972351,
      "learning_rate": 1.734117451516065e-05,
      "loss": 1.9531,
      "step": 205345
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.434542179107666,
      "learning_rate": 1.733980394953276e-05,
      "loss": 1.8924,
      "step": 205346
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.3528074026107788,
      "learning_rate": 1.7338433436456766e-05,
      "loss": 1.769,
      "step": 205347
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.3673022985458374,
      "learning_rate": 1.7337062975932903e-05,
      "loss": 1.8527,
      "step": 205348
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.4418816566467285,
      "learning_rate": 1.7335692567961333e-05,
      "loss": 2.014,
      "step": 205349
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.5634608268737793,
      "learning_rate": 1.7334322212542462e-05,
      "loss": 1.9745,
      "step": 205350
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.7500431537628174,
      "learning_rate": 1.733295190967642e-05,
      "loss": 1.8815,
      "step": 205351
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.3184585571289062,
      "learning_rate": 1.733158165936357e-05,
      "loss": 2.1153,
      "step": 205352
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.3137004375457764,
      "learning_rate": 1.7330211461604115e-05,
      "loss": 1.9526,
      "step": 205353
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.4269657135009766,
      "learning_rate": 1.7328841316398324e-05,
      "loss": 1.939,
      "step": 205354
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.5221383571624756,
      "learning_rate": 1.7327471223746393e-05,
      "loss": 1.9905,
      "step": 205355
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.3537697792053223,
      "learning_rate": 1.7326101183648655e-05,
      "loss": 1.7904,
      "step": 205356
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.560792088508606,
      "learning_rate": 1.732473119610528e-05,
      "loss": 1.9781,
      "step": 205357
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.470255732536316,
      "learning_rate": 1.7323361261116596e-05,
      "loss": 2.0511,
      "step": 205358
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.3818118572235107,
      "learning_rate": 1.7321991378682877e-05,
      "loss": 2.0979,
      "step": 205359
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.4528969526290894,
      "learning_rate": 1.7320621548804314e-05,
      "loss": 1.9517,
      "step": 205360
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.624948501586914,
      "learning_rate": 1.7319251771481147e-05,
      "loss": 1.979,
      "step": 205361
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.4644349813461304,
      "learning_rate": 1.7317882046713704e-05,
      "loss": 1.8789,
      "step": 205362
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.4600144624710083,
      "learning_rate": 1.7316512374502155e-05,
      "loss": 2.0573,
      "step": 205363
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.5672658681869507,
      "learning_rate": 1.7315142754846866e-05,
      "loss": 1.9739,
      "step": 205364
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.5241647958755493,
      "learning_rate": 1.7313773187748003e-05,
      "loss": 2.2693,
      "step": 205365
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.5671601295471191,
      "learning_rate": 1.7312403673205864e-05,
      "loss": 1.8965,
      "step": 205366
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.4794085025787354,
      "learning_rate": 1.7311034211220653e-05,
      "loss": 1.8321,
      "step": 205367
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.36777663230896,
      "learning_rate": 1.7309664801792666e-05,
      "loss": 1.8108,
      "step": 205368
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.385335087776184,
      "learning_rate": 1.730829544492214e-05,
      "loss": 1.899,
      "step": 205369
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.5993123054504395,
      "learning_rate": 1.7306926140609366e-05,
      "loss": 2.1047,
      "step": 205370
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.525149941444397,
      "learning_rate": 1.7305556888854554e-05,
      "loss": 2.227,
      "step": 205371
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.4407261610031128,
      "learning_rate": 1.7304187689658e-05,
      "loss": 2.1436,
      "step": 205372
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.5134739875793457,
      "learning_rate": 1.730281854301987e-05,
      "loss": 1.9254,
      "step": 205373
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.5743992328643799,
      "learning_rate": 1.7301449448940565e-05,
      "loss": 1.9039,
      "step": 205374
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.5125652551651,
      "learning_rate": 1.730008040742018e-05,
      "loss": 2.0959,
      "step": 205375
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.3707542419433594,
      "learning_rate": 1.7298711418459088e-05,
      "loss": 1.9162,
      "step": 205376
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.456349492073059,
      "learning_rate": 1.729734248205752e-05,
      "loss": 2.1038,
      "step": 205377
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.442289113998413,
      "learning_rate": 1.7295973598215672e-05,
      "loss": 2.1214,
      "step": 205378
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.495937466621399,
      "learning_rate": 1.7294604766933852e-05,
      "loss": 1.9756,
      "step": 205379
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.3238016366958618,
      "learning_rate": 1.7293235988212318e-05,
      "loss": 1.9693,
      "step": 205380
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.4478445053100586,
      "learning_rate": 1.729186726205124e-05,
      "loss": 1.9774,
      "step": 205381
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.514689326286316,
      "learning_rate": 1.7290498588451018e-05,
      "loss": 1.9722,
      "step": 205382
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.308292031288147,
      "learning_rate": 1.7289129967411785e-05,
      "loss": 1.7674,
      "step": 205383
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.547128677368164,
      "learning_rate": 1.7287761398933908e-05,
      "loss": 1.9394,
      "step": 205384
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.4034538269042969,
      "learning_rate": 1.7286392883017484e-05,
      "loss": 1.9795,
      "step": 205385
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.525622844696045,
      "learning_rate": 1.7285024419662918e-05,
      "loss": 1.8501,
      "step": 205386
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.3900948762893677,
      "learning_rate": 1.728365600887037e-05,
      "loss": 1.8566,
      "step": 205387
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.4034850597381592,
      "learning_rate": 1.728228765064015e-05,
      "loss": 1.8559,
      "step": 205388
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.394779920578003,
      "learning_rate": 1.7280919344972445e-05,
      "loss": 2.0436,
      "step": 205389
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.453938364982605,
      "learning_rate": 1.7279551091867627e-05,
      "loss": 2.0174,
      "step": 205390
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.3609188795089722,
      "learning_rate": 1.7278182891325832e-05,
      "loss": 2.0397,
      "step": 205391
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.6239311695098877,
      "learning_rate": 1.7276814743347355e-05,
      "loss": 1.968,
      "step": 205392
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.3717340230941772,
      "learning_rate": 1.7275446647932435e-05,
      "loss": 1.8388,
      "step": 205393
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.4274924993515015,
      "learning_rate": 1.72740786050814e-05,
      "loss": 1.8594,
      "step": 205394
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.5771524906158447,
      "learning_rate": 1.7272710614794384e-05,
      "loss": 2.0099,
      "step": 205395
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.4754496812820435,
      "learning_rate": 1.727134267707182e-05,
      "loss": 2.1475,
      "step": 205396
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.521453619003296,
      "learning_rate": 1.7269974791913743e-05,
      "loss": 1.7796,
      "step": 205397
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.44688880443573,
      "learning_rate": 1.7268606959320587e-05,
      "loss": 2.0396,
      "step": 205398
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.5271214246749878,
      "learning_rate": 1.7267239179292446e-05,
      "loss": 2.1124,
      "step": 205399
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.4618488550186157,
      "learning_rate": 1.7265871451829726e-05,
      "loss": 1.8177,
      "step": 205400
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.5488801002502441,
      "learning_rate": 1.726450377693259e-05,
      "loss": 1.8422,
      "step": 205401
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.687502384185791,
      "learning_rate": 1.7263136154601408e-05,
      "loss": 1.9921,
      "step": 205402
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.5864006280899048,
      "learning_rate": 1.7261768584836243e-05,
      "loss": 2.0069,
      "step": 205403
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.4635560512542725,
      "learning_rate": 1.7260401067637496e-05,
      "loss": 1.9253,
      "step": 205404
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.5715701580047607,
      "learning_rate": 1.7259033603005336e-05,
      "loss": 1.998,
      "step": 205405
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.4209164381027222,
      "learning_rate": 1.725766619094009e-05,
      "loss": 1.8176,
      "step": 205406
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.574407935142517,
      "learning_rate": 1.725629883144196e-05,
      "loss": 2.1065,
      "step": 205407
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.6850898265838623,
      "learning_rate": 1.7254931524511252e-05,
      "loss": 1.9784,
      "step": 205408
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.3654584884643555,
      "learning_rate": 1.725356427014819e-05,
      "loss": 1.9356,
      "step": 205409
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.3902819156646729,
      "learning_rate": 1.7252197068353013e-05,
      "loss": 1.9938,
      "step": 205410
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.6717196702957153,
      "learning_rate": 1.7250829919125985e-05,
      "loss": 1.8834,
      "step": 205411
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.406867504119873,
      "learning_rate": 1.7249462822467373e-05,
      "loss": 1.9832,
      "step": 205412
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.4253056049346924,
      "learning_rate": 1.7248095778377415e-05,
      "loss": 1.9595,
      "step": 205413
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.4490516185760498,
      "learning_rate": 1.7246728786856368e-05,
      "loss": 1.9593,
      "step": 205414
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.5160284042358398,
      "learning_rate": 1.7245361847904537e-05,
      "loss": 2.1274,
      "step": 205415
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.483984351158142,
      "learning_rate": 1.724399496152209e-05,
      "loss": 1.9907,
      "step": 205416
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.340059757232666,
      "learning_rate": 1.7242628127709322e-05,
      "loss": 1.8962,
      "step": 205417
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.4119269847869873,
      "learning_rate": 1.7241261346466505e-05,
      "loss": 1.9624,
      "step": 205418
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.4512712955474854,
      "learning_rate": 1.7239894617793835e-05,
      "loss": 1.8458,
      "step": 205419
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.4584054946899414,
      "learning_rate": 1.723852794169165e-05,
      "loss": 1.7616,
      "step": 205420
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.4083160161972046,
      "learning_rate": 1.7237161318160142e-05,
      "loss": 2.023,
      "step": 205421
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.4085389375686646,
      "learning_rate": 1.7235794747199583e-05,
      "loss": 1.9219,
      "step": 205422
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.4376277923583984,
      "learning_rate": 1.723442822881024e-05,
      "loss": 2.0008,
      "step": 205423
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.3345078229904175,
      "learning_rate": 1.7233061762992374e-05,
      "loss": 2.016,
      "step": 205424
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.340509057044983,
      "learning_rate": 1.7231695349746154e-05,
      "loss": 1.7795,
      "step": 205425
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.6578410863876343,
      "learning_rate": 1.723032898907195e-05,
      "loss": 2.1582,
      "step": 205426
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.4658809900283813,
      "learning_rate": 1.722896268096996e-05,
      "loss": 1.7186,
      "step": 205427
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.485469102859497,
      "learning_rate": 1.7227596425440414e-05,
      "loss": 1.9915,
      "step": 205428
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.6444664001464844,
      "learning_rate": 1.7226230222483615e-05,
      "loss": 1.8885,
      "step": 205429
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.4464654922485352,
      "learning_rate": 1.722486407209983e-05,
      "loss": 1.9267,
      "step": 205430
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.5384278297424316,
      "learning_rate": 1.7223497974289223e-05,
      "loss": 1.8844,
      "step": 205431
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.4927282333374023,
      "learning_rate": 1.722213192905213e-05,
      "loss": 2.3121,
      "step": 205432
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.4992396831512451,
      "learning_rate": 1.7220765936388815e-05,
      "loss": 1.6562,
      "step": 205433
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.4455740451812744,
      "learning_rate": 1.7219399996299445e-05,
      "loss": 2.0527,
      "step": 205434
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.575756549835205,
      "learning_rate": 1.721803410878435e-05,
      "loss": 1.9807,
      "step": 205435
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.4720981121063232,
      "learning_rate": 1.7216668273843736e-05,
      "loss": 1.7866,
      "step": 205436
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.4005353450775146,
      "learning_rate": 1.72153024914779e-05,
      "loss": 1.9296,
      "step": 205437
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.4810028076171875,
      "learning_rate": 1.7213936761687107e-05,
      "loss": 2.1249,
      "step": 205438
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.4443538188934326,
      "learning_rate": 1.721257108447156e-05,
      "loss": 2.0528,
      "step": 205439
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.4981296062469482,
      "learning_rate": 1.721120545983149e-05,
      "loss": 2.0287,
      "step": 205440
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.4707781076431274,
      "learning_rate": 1.7209839887767262e-05,
      "loss": 2.1706,
      "step": 205441
    },
    {
      "epoch": 2.67,
      "grad_norm": 1.3663312196731567,
      "learning_rate": 1.7208474368279014e-05,
      "loss": 2.2469,
      "step": 205442
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.402349591255188,
      "learning_rate": 1.7207108901367074e-05,
      "loss": 2.0609,
      "step": 205443
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.4290341138839722,
      "learning_rate": 1.7205743487031674e-05,
      "loss": 2.0239,
      "step": 205444
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.550498127937317,
      "learning_rate": 1.7204378125273087e-05,
      "loss": 2.0839,
      "step": 205445
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.6048691272735596,
      "learning_rate": 1.7203012816091478e-05,
      "loss": 2.0398,
      "step": 205446
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.397939682006836,
      "learning_rate": 1.720164755948721e-05,
      "loss": 1.9167,
      "step": 205447
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.3261100053787231,
      "learning_rate": 1.720028235546045e-05,
      "loss": 2.0835,
      "step": 205448
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.4006706476211548,
      "learning_rate": 1.7198917204011564e-05,
      "loss": 2.0522,
      "step": 205449
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.4488388299942017,
      "learning_rate": 1.7197552105140656e-05,
      "loss": 1.8933,
      "step": 205450
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.3911182880401611,
      "learning_rate": 1.7196187058848188e-05,
      "loss": 1.8512,
      "step": 205451
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.4179731607437134,
      "learning_rate": 1.7194822065134195e-05,
      "loss": 1.8346,
      "step": 205452
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.5591490268707275,
      "learning_rate": 1.7193457123999043e-05,
      "loss": 1.9873,
      "step": 205453
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.4502547979354858,
      "learning_rate": 1.7192092235442934e-05,
      "loss": 1.9448,
      "step": 205454
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.3731740713119507,
      "learning_rate": 1.7190727399466198e-05,
      "loss": 2.112,
      "step": 205455
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.408454179763794,
      "learning_rate": 1.7189362616069002e-05,
      "loss": 1.9549,
      "step": 205456
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.4500153064727783,
      "learning_rate": 1.7187997885251715e-05,
      "loss": 1.7525,
      "step": 205457
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.4132745265960693,
      "learning_rate": 1.718663320701443e-05,
      "loss": 2.0055,
      "step": 205458
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.4517312049865723,
      "learning_rate": 1.718526858135756e-05,
      "loss": 2.0426,
      "step": 205459
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.4399607181549072,
      "learning_rate": 1.7183904008281225e-05,
      "loss": 1.7839,
      "step": 205460
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.5373470783233643,
      "learning_rate": 1.7182539487785763e-05,
      "loss": 2.083,
      "step": 205461
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.488612174987793,
      "learning_rate": 1.7181175019871406e-05,
      "loss": 1.9637,
      "step": 205462
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.5859571695327759,
      "learning_rate": 1.7179810604538458e-05,
      "loss": 1.994,
      "step": 205463
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.350709319114685,
      "learning_rate": 1.7178446241787046e-05,
      "loss": 1.7397,
      "step": 205464
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.38321852684021,
      "learning_rate": 1.7177081931617542e-05,
      "loss": 2.1771,
      "step": 205465
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.5791419744491577,
      "learning_rate": 1.7175717674030143e-05,
      "loss": 2.2895,
      "step": 205466
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.542662262916565,
      "learning_rate": 1.7174353469025115e-05,
      "loss": 1.6081,
      "step": 205467
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.4305442571640015,
      "learning_rate": 1.717298931660269e-05,
      "loss": 1.9307,
      "step": 205468
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.4829833507537842,
      "learning_rate": 1.717162521676324e-05,
      "loss": 1.9427,
      "step": 205469
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.4651517868041992,
      "learning_rate": 1.7170261169506828e-05,
      "loss": 2.0737,
      "step": 205470
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.4452309608459473,
      "learning_rate": 1.716889717483385e-05,
      "loss": 2.0179,
      "step": 205471
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.3402594327926636,
      "learning_rate": 1.716753323274448e-05,
      "loss": 1.7833,
      "step": 205472
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.3384896516799927,
      "learning_rate": 1.716616934323901e-05,
      "loss": 2.064,
      "step": 205473
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.3655943870544434,
      "learning_rate": 1.716480550631768e-05,
      "loss": 2.0013,
      "step": 205474
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.4492852687835693,
      "learning_rate": 1.716344172198082e-05,
      "loss": 2.1786,
      "step": 205475
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.441065788269043,
      "learning_rate": 1.716207799022853e-05,
      "loss": 1.9767,
      "step": 205476
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.464861512184143,
      "learning_rate": 1.7160714311061206e-05,
      "loss": 2.1468,
      "step": 205477
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.4184085130691528,
      "learning_rate": 1.715935068447899e-05,
      "loss": 1.839,
      "step": 205478
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.455940842628479,
      "learning_rate": 1.715798711048224e-05,
      "loss": 1.9515,
      "step": 205479
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.4835325479507446,
      "learning_rate": 1.7156623589071127e-05,
      "loss": 1.8647,
      "step": 205480
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.4376935958862305,
      "learning_rate": 1.7155260120245985e-05,
      "loss": 1.9957,
      "step": 205481
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.4591768980026245,
      "learning_rate": 1.7153896704006976e-05,
      "loss": 2.1176,
      "step": 205482
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.4532740116119385,
      "learning_rate": 1.7152533340354402e-05,
      "loss": 1.933,
      "step": 205483
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.3591853380203247,
      "learning_rate": 1.7151170029288496e-05,
      "loss": 1.733,
      "step": 205484
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.4234336614608765,
      "learning_rate": 1.714980677080956e-05,
      "loss": 1.8717,
      "step": 205485
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.5593645572662354,
      "learning_rate": 1.7148443564917756e-05,
      "loss": 1.8385,
      "step": 205486
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.6109007596969604,
      "learning_rate": 1.7147080411613522e-05,
      "loss": 1.7562,
      "step": 205487
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.561221957206726,
      "learning_rate": 1.7145717310896857e-05,
      "loss": 2.0996,
      "step": 205488
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.3667664527893066,
      "learning_rate": 1.7144354262768223e-05,
      "loss": 2.0564,
      "step": 205489
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.3496354818344116,
      "learning_rate": 1.7142991267227725e-05,
      "loss": 1.9206,
      "step": 205490
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.507156252861023,
      "learning_rate": 1.7141628324275724e-05,
      "loss": 2.0649,
      "step": 205491
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.567671298980713,
      "learning_rate": 1.7140265433912394e-05,
      "loss": 1.9491,
      "step": 205492
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.3937901258468628,
      "learning_rate": 1.7138902596138127e-05,
      "loss": 1.7744,
      "step": 205493
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.340968370437622,
      "learning_rate": 1.7137539810952962e-05,
      "loss": 1.8939,
      "step": 205494
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.370351791381836,
      "learning_rate": 1.7136177078357362e-05,
      "loss": 1.7854,
      "step": 205495
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.4756547212600708,
      "learning_rate": 1.713481439835139e-05,
      "loss": 1.8448,
      "step": 205496
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.4180104732513428,
      "learning_rate": 1.7133451770935457e-05,
      "loss": 1.9954,
      "step": 205497
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.4608476161956787,
      "learning_rate": 1.713208919610972e-05,
      "loss": 1.9337,
      "step": 205498
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.6181809902191162,
      "learning_rate": 1.7130726673874516e-05,
      "loss": 1.8415,
      "step": 205499
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.4628281593322754,
      "learning_rate": 1.7129364204230046e-05,
      "loss": 2.0404,
      "step": 205500
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.5114691257476807,
      "learning_rate": 1.7128001787176536e-05,
      "loss": 2.0809,
      "step": 205501
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.4240435361862183,
      "learning_rate": 1.7126639422714262e-05,
      "loss": 1.9934,
      "step": 205502
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.3863186836242676,
      "learning_rate": 1.7125277110843515e-05,
      "loss": 1.9213,
      "step": 205503
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.4161633253097534,
      "learning_rate": 1.7123914851564467e-05,
      "loss": 2.0599,
      "step": 205504
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.4366947412490845,
      "learning_rate": 1.7122552644877485e-05,
      "loss": 2.0195,
      "step": 205505
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.4705877304077148,
      "learning_rate": 1.7121190490782766e-05,
      "loss": 1.947,
      "step": 205506
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.3992283344268799,
      "learning_rate": 1.711982838928051e-05,
      "loss": 1.9852,
      "step": 205507
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.3570265769958496,
      "learning_rate": 1.711846634037105e-05,
      "loss": 1.9784,
      "step": 205508
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.6123102903366089,
      "learning_rate": 1.711710434405459e-05,
      "loss": 1.9471,
      "step": 205509
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.454283595085144,
      "learning_rate": 1.711574240033139e-05,
      "loss": 1.9047,
      "step": 205510
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.4143719673156738,
      "learning_rate": 1.711438050920172e-05,
      "loss": 2.0295,
      "step": 205511
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.4554024934768677,
      "learning_rate": 1.7113018670665878e-05,
      "loss": 2.0379,
      "step": 205512
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.3778674602508545,
      "learning_rate": 1.711165688472397e-05,
      "loss": 2.0115,
      "step": 205513
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.4639102220535278,
      "learning_rate": 1.711029515137642e-05,
      "loss": 2.0069,
      "step": 205514
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.369065523147583,
      "learning_rate": 1.71089334706234e-05,
      "loss": 1.9087,
      "step": 205515
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.5512416362762451,
      "learning_rate": 1.7107571842465108e-05,
      "loss": 2.02,
      "step": 205516
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.5118660926818848,
      "learning_rate": 1.710621026690194e-05,
      "loss": 2.2676,
      "step": 205517
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.3750574588775635,
      "learning_rate": 1.7104848743934042e-05,
      "loss": 1.8311,
      "step": 205518
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.744699239730835,
      "learning_rate": 1.7103487273561634e-05,
      "loss": 2.0726,
      "step": 205519
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.5205739736557007,
      "learning_rate": 1.7102125855785086e-05,
      "loss": 2.0093,
      "step": 205520
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.4939035177230835,
      "learning_rate": 1.710076449060457e-05,
      "loss": 1.8645,
      "step": 205521
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.4545001983642578,
      "learning_rate": 1.7099403178020375e-05,
      "loss": 2.1086,
      "step": 205522
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.5275179147720337,
      "learning_rate": 1.7098041918032747e-05,
      "loss": 1.8765,
      "step": 205523
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.2991290092468262,
      "learning_rate": 1.709668071064194e-05,
      "loss": 2.1137,
      "step": 205524
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.2836021184921265,
      "learning_rate": 1.7095319555848164e-05,
      "loss": 2.1306,
      "step": 205525
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.5514088869094849,
      "learning_rate": 1.7093958453651745e-05,
      "loss": 2.0481,
      "step": 205526
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.708167552947998,
      "learning_rate": 1.7092597404052854e-05,
      "loss": 1.9558,
      "step": 205527
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.3693228960037231,
      "learning_rate": 1.7091236407051856e-05,
      "loss": 1.8974,
      "step": 205528
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.6126896142959595,
      "learning_rate": 1.708987546264885e-05,
      "loss": 1.8894,
      "step": 205529
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.4343056678771973,
      "learning_rate": 1.70885145708443e-05,
      "loss": 2.1386,
      "step": 205530
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.3804818391799927,
      "learning_rate": 1.7087153731638216e-05,
      "loss": 1.8994,
      "step": 205531
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.4545427560806274,
      "learning_rate": 1.708579294503105e-05,
      "loss": 1.7756,
      "step": 205532
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.3973575830459595,
      "learning_rate": 1.7084432211022913e-05,
      "loss": 2.0586,
      "step": 205533
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.3590147495269775,
      "learning_rate": 1.7083071529614167e-05,
      "loss": 1.833,
      "step": 205534
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.4340697526931763,
      "learning_rate": 1.708171090080498e-05,
      "loss": 1.9984,
      "step": 205535
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.3947489261627197,
      "learning_rate": 1.7080350324595714e-05,
      "loss": 1.8423,
      "step": 205536
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.475238561630249,
      "learning_rate": 1.7078989800986475e-05,
      "loss": 2.1219,
      "step": 205537
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.5261110067367554,
      "learning_rate": 1.7077629329977626e-05,
      "loss": 2.1783,
      "step": 205538
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.408001184463501,
      "learning_rate": 1.7076268911569367e-05,
      "loss": 2.0908,
      "step": 205539
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.4431791305541992,
      "learning_rate": 1.7074908545762e-05,
      "loss": 2.0015,
      "step": 205540
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.3158597946166992,
      "learning_rate": 1.707354823255569e-05,
      "loss": 1.9576,
      "step": 205541
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.5248627662658691,
      "learning_rate": 1.707218797195087e-05,
      "loss": 1.9175,
      "step": 205542
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.3777776956558228,
      "learning_rate": 1.7070827763947536e-05,
      "loss": 2.0639,
      "step": 205543
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.4665076732635498,
      "learning_rate": 1.7069467608546162e-05,
      "loss": 1.9266,
      "step": 205544
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.5590147972106934,
      "learning_rate": 1.706810750574684e-05,
      "loss": 1.8417,
      "step": 205545
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.4952105283737183,
      "learning_rate": 1.7066747455549978e-05,
      "loss": 1.9203,
      "step": 205546
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.3744025230407715,
      "learning_rate": 1.706538745795567e-05,
      "loss": 1.8749,
      "step": 205547
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.3205039501190186,
      "learning_rate": 1.706402751296435e-05,
      "loss": 1.8886,
      "step": 205548
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.3421144485473633,
      "learning_rate": 1.7062667620576088e-05,
      "loss": 1.9689,
      "step": 205549
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.480212926864624,
      "learning_rate": 1.7061307780791245e-05,
      "loss": 1.9957,
      "step": 205550
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.433362603187561,
      "learning_rate": 1.705994799360999e-05,
      "loss": 2.0559,
      "step": 205551
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.4666873216629028,
      "learning_rate": 1.7058588259032725e-05,
      "loss": 1.7366,
      "step": 205552
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.4330002069473267,
      "learning_rate": 1.705722857705951e-05,
      "loss": 1.7889,
      "step": 205553
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.5522431135177612,
      "learning_rate": 1.7055868947690822e-05,
      "loss": 1.9997,
      "step": 205554
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.4213553667068481,
      "learning_rate": 1.7054509370926684e-05,
      "loss": 1.845,
      "step": 205555
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.431448221206665,
      "learning_rate": 1.70531498467675e-05,
      "loss": 1.8694,
      "step": 205556
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.318794846534729,
      "learning_rate": 1.7051790375213438e-05,
      "loss": 1.9064,
      "step": 205557
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.30620276927948,
      "learning_rate": 1.7050430956264826e-05,
      "loss": 1.667,
      "step": 205558
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.4763531684875488,
      "learning_rate": 1.7049071589921835e-05,
      "loss": 1.9073,
      "step": 205559
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.5103822946548462,
      "learning_rate": 1.7047712276184866e-05,
      "loss": 1.857,
      "step": 205560
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.7714430093765259,
      "learning_rate": 1.704635301505395e-05,
      "loss": 1.8745,
      "step": 205561
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.7395764589309692,
      "learning_rate": 1.7044993806529517e-05,
      "loss": 2.1074,
      "step": 205562
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.4654148817062378,
      "learning_rate": 1.7043634650611736e-05,
      "loss": 2.0418,
      "step": 205563
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.5504779815673828,
      "learning_rate": 1.704227554730091e-05,
      "loss": 1.8394,
      "step": 205564
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.390236258506775,
      "learning_rate": 1.7040916496597235e-05,
      "loss": 1.9194,
      "step": 205565
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.4444226026535034,
      "learning_rate": 1.7039557498501076e-05,
      "loss": 1.804,
      "step": 205566
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.4075603485107422,
      "learning_rate": 1.7038198553012504e-05,
      "loss": 1.8442,
      "step": 205567
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.5106192827224731,
      "learning_rate": 1.7036839660131918e-05,
      "loss": 1.9862,
      "step": 205568
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.3685892820358276,
      "learning_rate": 1.7035480819859515e-05,
      "loss": 2.0291,
      "step": 205569
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.5032007694244385,
      "learning_rate": 1.7034122032195563e-05,
      "loss": 1.9613,
      "step": 205570
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.5288047790527344,
      "learning_rate": 1.7032763297140262e-05,
      "loss": 2.1816,
      "step": 205571
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.442864179611206,
      "learning_rate": 1.7031404614694045e-05,
      "loss": 2.0319,
      "step": 205572
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.4913787841796875,
      "learning_rate": 1.703004598485691e-05,
      "loss": 2.1176,
      "step": 205573
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.309478521347046,
      "learning_rate": 1.702868740762926e-05,
      "loss": 1.9322,
      "step": 205574
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.4615720510482788,
      "learning_rate": 1.7027328883011294e-05,
      "loss": 1.8348,
      "step": 205575
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.3886069059371948,
      "learning_rate": 1.702597041100334e-05,
      "loss": 1.7449,
      "step": 205576
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.4222103357315063,
      "learning_rate": 1.702461199160554e-05,
      "loss": 1.8973,
      "step": 205577
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.5077269077301025,
      "learning_rate": 1.7023253624818324e-05,
      "loss": 1.9227,
      "step": 205578
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.4969911575317383,
      "learning_rate": 1.702189531064172e-05,
      "loss": 2.0212,
      "step": 205579
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.4333311319351196,
      "learning_rate": 1.7020537049076098e-05,
      "loss": 2.0809,
      "step": 205580
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.497466802597046,
      "learning_rate": 1.7019178840121695e-05,
      "loss": 1.9487,
      "step": 205581
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.3752561807632446,
      "learning_rate": 1.7017820683778805e-05,
      "loss": 2.025,
      "step": 205582
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.41253662109375,
      "learning_rate": 1.7016462580047595e-05,
      "loss": 2.0053,
      "step": 205583
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.461057424545288,
      "learning_rate": 1.7015104528928436e-05,
      "loss": 2.0139,
      "step": 205584
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.5373941659927368,
      "learning_rate": 1.701374653042149e-05,
      "loss": 2.0853,
      "step": 205585
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.5068379640579224,
      "learning_rate": 1.7012388584527026e-05,
      "loss": 1.8865,
      "step": 205586
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.4677326679229736,
      "learning_rate": 1.701103069124524e-05,
      "loss": 1.9866,
      "step": 205587
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.5445401668548584,
      "learning_rate": 1.7009672850576507e-05,
      "loss": 1.9626,
      "step": 205588
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.4184805154800415,
      "learning_rate": 1.7008315062520985e-05,
      "loss": 1.9773,
      "step": 205589
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.5676695108413696,
      "learning_rate": 1.700695732707901e-05,
      "loss": 1.812,
      "step": 205590
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.4197880029678345,
      "learning_rate": 1.7005599644250745e-05,
      "loss": 2.0124,
      "step": 205591
    },
    {
      "epoch": 2.68,
      "grad_norm": 2.019225597381592,
      "learning_rate": 1.7004242014036495e-05,
      "loss": 2.0785,
      "step": 205592
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.4098607301712036,
      "learning_rate": 1.7002884436436493e-05,
      "loss": 1.8739,
      "step": 205593
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.601116418838501,
      "learning_rate": 1.7001526911451003e-05,
      "loss": 1.9677,
      "step": 205594
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.3490705490112305,
      "learning_rate": 1.7000169439080225e-05,
      "loss": 1.6981,
      "step": 205595
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.542175054550171,
      "learning_rate": 1.6998812019324525e-05,
      "loss": 2.0415,
      "step": 205596
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.4293421506881714,
      "learning_rate": 1.6997454652184072e-05,
      "loss": 1.8987,
      "step": 205597
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.46195650100708,
      "learning_rate": 1.6996097337659096e-05,
      "loss": 1.8964,
      "step": 205598
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.573039174079895,
      "learning_rate": 1.6994740075749904e-05,
      "loss": 2.0925,
      "step": 205599
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.554611086845398,
      "learning_rate": 1.699338286645675e-05,
      "loss": 1.8489,
      "step": 205600
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.3977786302566528,
      "learning_rate": 1.6992025709779845e-05,
      "loss": 1.9523,
      "step": 205601
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.410672664642334,
      "learning_rate": 1.6990668605719483e-05,
      "loss": 1.8565,
      "step": 205602
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.6819978952407837,
      "learning_rate": 1.698931155427593e-05,
      "loss": 2.0127,
      "step": 205603
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.6622964143753052,
      "learning_rate": 1.6987954555449324e-05,
      "loss": 1.8773,
      "step": 205604
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.429327368736267,
      "learning_rate": 1.698659760924006e-05,
      "loss": 2.1313,
      "step": 205605
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.424180030822754,
      "learning_rate": 1.698524071564831e-05,
      "loss": 2.0811,
      "step": 205606
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.5888488292694092,
      "learning_rate": 1.6983883874674365e-05,
      "loss": 2.1658,
      "step": 205607
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.5149264335632324,
      "learning_rate": 1.698252708631843e-05,
      "loss": 2.0414,
      "step": 205608
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.4567354917526245,
      "learning_rate": 1.6981170350580842e-05,
      "loss": 2.1363,
      "step": 205609
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.545689344406128,
      "learning_rate": 1.6979813667461727e-05,
      "loss": 2.1654,
      "step": 205610
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.3430986404418945,
      "learning_rate": 1.6978457036961424e-05,
      "loss": 2.1316,
      "step": 205611
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.6127656698226929,
      "learning_rate": 1.6977100459080162e-05,
      "loss": 2.0972,
      "step": 205612
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.62126886844635,
      "learning_rate": 1.697574393381824e-05,
      "loss": 1.9711,
      "step": 205613
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.442232608795166,
      "learning_rate": 1.6974387461175866e-05,
      "loss": 2.1367,
      "step": 205614
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.5998400449752808,
      "learning_rate": 1.6973031041153295e-05,
      "loss": 2.0162,
      "step": 205615
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.321522831916809,
      "learning_rate": 1.6971674673750702e-05,
      "loss": 2.0255,
      "step": 205616
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.4585726261138916,
      "learning_rate": 1.6970318358968516e-05,
      "loss": 1.973,
      "step": 205617
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.5291334390640259,
      "learning_rate": 1.6968962096806805e-05,
      "loss": 1.8588,
      "step": 205618
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.5360159873962402,
      "learning_rate": 1.6967605887265966e-05,
      "loss": 1.9983,
      "step": 205619
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.4712790250778198,
      "learning_rate": 1.696624973034617e-05,
      "loss": 1.9753,
      "step": 205620
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.5352586507797241,
      "learning_rate": 1.696489362604775e-05,
      "loss": 1.8613,
      "step": 205621
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.4348608255386353,
      "learning_rate": 1.6963537574370835e-05,
      "loss": 2.1552,
      "step": 205622
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.4158308506011963,
      "learning_rate": 1.696218157531576e-05,
      "loss": 2.1585,
      "step": 205623
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.525430679321289,
      "learning_rate": 1.6960825628882722e-05,
      "loss": 1.9537,
      "step": 205624
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.4762483835220337,
      "learning_rate": 1.6959469735072062e-05,
      "loss": 2.174,
      "step": 205625
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.6731435060501099,
      "learning_rate": 1.6958113893883907e-05,
      "loss": 2.0708,
      "step": 205626
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.4634618759155273,
      "learning_rate": 1.695675810531869e-05,
      "loss": 1.9027,
      "step": 205627
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.5630923509597778,
      "learning_rate": 1.695540236937648e-05,
      "loss": 2.0088,
      "step": 205628
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.4279264211654663,
      "learning_rate": 1.6954046686057642e-05,
      "loss": 1.8655,
      "step": 205629
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.5227762460708618,
      "learning_rate": 1.695269105536231e-05,
      "loss": 2.0635,
      "step": 205630
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.4163159132003784,
      "learning_rate": 1.6951335477290916e-05,
      "loss": 1.9944,
      "step": 205631
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.4181345701217651,
      "learning_rate": 1.6949979951843528e-05,
      "loss": 1.9515,
      "step": 205632
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.4724414348602295,
      "learning_rate": 1.694862447902058e-05,
      "loss": 1.9905,
      "step": 205633
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.5824172496795654,
      "learning_rate": 1.6947269058822133e-05,
      "loss": 2.1209,
      "step": 205634
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.347062110900879,
      "learning_rate": 1.694591369124859e-05,
      "loss": 1.9734,
      "step": 205635
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.5408878326416016,
      "learning_rate": 1.6944558376300088e-05,
      "loss": 2.0143,
      "step": 205636
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.4037768840789795,
      "learning_rate": 1.694320311397699e-05,
      "loss": 2.0056,
      "step": 205637
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.3441630601882935,
      "learning_rate": 1.6941847904279428e-05,
      "loss": 2.1885,
      "step": 205638
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.3537887334823608,
      "learning_rate": 1.6940492747207834e-05,
      "loss": 2.0506,
      "step": 205639
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.4137184619903564,
      "learning_rate": 1.6939137642762248e-05,
      "loss": 1.7916,
      "step": 205640
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.466200351715088,
      "learning_rate": 1.6937782590943062e-05,
      "loss": 1.8294,
      "step": 205641
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.5028918981552124,
      "learning_rate": 1.6936427591750444e-05,
      "loss": 2.0465,
      "step": 205642
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.4360926151275635,
      "learning_rate": 1.693507264518473e-05,
      "loss": 2.0772,
      "step": 205643
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.549519658088684,
      "learning_rate": 1.693371775124609e-05,
      "loss": 1.8845,
      "step": 205644
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.3200234174728394,
      "learning_rate": 1.693236290993488e-05,
      "loss": 1.9069,
      "step": 205645
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.3726797103881836,
      "learning_rate": 1.6931008121251245e-05,
      "loss": 1.9246,
      "step": 205646
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.460239291191101,
      "learning_rate": 1.6929653385195473e-05,
      "loss": 1.9646,
      "step": 205647
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.4181294441223145,
      "learning_rate": 1.6928298701767806e-05,
      "loss": 1.8043,
      "step": 205648
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.4949239492416382,
      "learning_rate": 1.692694407096854e-05,
      "loss": 1.8562,
      "step": 205649
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.3711647987365723,
      "learning_rate": 1.6925589492797876e-05,
      "loss": 1.7312,
      "step": 205650
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.5402551889419556,
      "learning_rate": 1.692423496725611e-05,
      "loss": 1.8323,
      "step": 205651
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.4791072607040405,
      "learning_rate": 1.6922880494343484e-05,
      "loss": 1.8989,
      "step": 205652
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.424556016921997,
      "learning_rate": 1.6921526074060253e-05,
      "loss": 2.0297,
      "step": 205653
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.3837809562683105,
      "learning_rate": 1.6920171706406594e-05,
      "loss": 2.0649,
      "step": 205654
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.4391452074050903,
      "learning_rate": 1.6918817391382867e-05,
      "loss": 2.2076,
      "step": 205655
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.3474246263504028,
      "learning_rate": 1.6917463128989207e-05,
      "loss": 2.1803,
      "step": 205656
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.3637617826461792,
      "learning_rate": 1.6916108919226013e-05,
      "loss": 1.8382,
      "step": 205657
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.4794964790344238,
      "learning_rate": 1.691475476209342e-05,
      "loss": 1.9782,
      "step": 205658
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.7164238691329956,
      "learning_rate": 1.6913400657591758e-05,
      "loss": 2.1226,
      "step": 205659
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.573704481124878,
      "learning_rate": 1.6912046605721164e-05,
      "loss": 2.0514,
      "step": 205660
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.6605327129364014,
      "learning_rate": 1.6910692606482033e-05,
      "loss": 1.9378,
      "step": 205661
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.4802980422973633,
      "learning_rate": 1.6909338659874473e-05,
      "loss": 2.0708,
      "step": 205662
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.4650336503982544,
      "learning_rate": 1.6907984765898874e-05,
      "loss": 2.0434,
      "step": 205663
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.4860929250717163,
      "learning_rate": 1.6906630924555443e-05,
      "loss": 1.7944,
      "step": 205664
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.6676549911499023,
      "learning_rate": 1.6905277135844375e-05,
      "loss": 2.1492,
      "step": 205665
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.4801459312438965,
      "learning_rate": 1.690392339976594e-05,
      "loss": 2.2124,
      "step": 205666
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.44806706905365,
      "learning_rate": 1.6902569716320468e-05,
      "loss": 1.801,
      "step": 205667
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.5376616716384888,
      "learning_rate": 1.690121608550806e-05,
      "loss": 1.8855,
      "step": 205668
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.6299757957458496,
      "learning_rate": 1.6899862507329154e-05,
      "loss": 2.0339,
      "step": 205669
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.5853151082992554,
      "learning_rate": 1.6898508981783874e-05,
      "loss": 2.2111,
      "step": 205670
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.6129827499389648,
      "learning_rate": 1.6897155508872528e-05,
      "loss": 2.1179,
      "step": 205671
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.5217857360839844,
      "learning_rate": 1.689580208859528e-05,
      "loss": 1.8893,
      "step": 205672
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.559792160987854,
      "learning_rate": 1.6894448720952493e-05,
      "loss": 1.96,
      "step": 205673
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.3599776029586792,
      "learning_rate": 1.6893095405944334e-05,
      "loss": 1.9346,
      "step": 205674
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.3460273742675781,
      "learning_rate": 1.689174214357114e-05,
      "loss": 1.8272,
      "step": 205675
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.4398024082183838,
      "learning_rate": 1.689038893383311e-05,
      "loss": 1.946,
      "step": 205676
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.5353164672851562,
      "learning_rate": 1.688903577673051e-05,
      "loss": 1.9538,
      "step": 205677
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.4081991910934448,
      "learning_rate": 1.6887682672263536e-05,
      "loss": 2.1058,
      "step": 205678
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.5649290084838867,
      "learning_rate": 1.6886329620432525e-05,
      "loss": 1.7826,
      "step": 205679
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.3928149938583374,
      "learning_rate": 1.6884976621237646e-05,
      "loss": 1.8429,
      "step": 205680
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.5303136110305786,
      "learning_rate": 1.6883623674679257e-05,
      "loss": 1.9829,
      "step": 205681
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.5517103672027588,
      "learning_rate": 1.6882270780757502e-05,
      "loss": 2.0354,
      "step": 205682
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.3721840381622314,
      "learning_rate": 1.688091793947267e-05,
      "loss": 1.8339,
      "step": 205683
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.4612680673599243,
      "learning_rate": 1.687956515082507e-05,
      "loss": 2.048,
      "step": 205684
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.4042744636535645,
      "learning_rate": 1.6878212414814863e-05,
      "loss": 2.0234,
      "step": 205685
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.3930306434631348,
      "learning_rate": 1.6876859731442317e-05,
      "loss": 1.9323,
      "step": 205686
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.468869686126709,
      "learning_rate": 1.6875507100707762e-05,
      "loss": 2.0317,
      "step": 205687
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.5550987720489502,
      "learning_rate": 1.6874154522611406e-05,
      "loss": 2.1392,
      "step": 205688
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.365640640258789,
      "learning_rate": 1.6872801997153408e-05,
      "loss": 2.0741,
      "step": 205689
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.445217251777649,
      "learning_rate": 1.687144952433417e-05,
      "loss": 2.0335,
      "step": 205690
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.4973253011703491,
      "learning_rate": 1.6870097104153822e-05,
      "loss": 2.1219,
      "step": 205691
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.3444783687591553,
      "learning_rate": 1.6868744736612706e-05,
      "loss": 1.9393,
      "step": 205692
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.4491353034973145,
      "learning_rate": 1.6867392421711045e-05,
      "loss": 2.1629,
      "step": 205693
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.5384232997894287,
      "learning_rate": 1.686604015944908e-05,
      "loss": 2.2152,
      "step": 205694
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.381691336631775,
      "learning_rate": 1.6864687949827005e-05,
      "loss": 2.2662,
      "step": 205695
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.5708932876586914,
      "learning_rate": 1.686333579284519e-05,
      "loss": 2.1285,
      "step": 205696
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.3933086395263672,
      "learning_rate": 1.6861983688503768e-05,
      "loss": 1.8546,
      "step": 205697
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.4836125373840332,
      "learning_rate": 1.68606316368031e-05,
      "loss": 1.9392,
      "step": 205698
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.466200351715088,
      "learning_rate": 1.6859279637743394e-05,
      "loss": 1.9147,
      "step": 205699
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.5517231225967407,
      "learning_rate": 1.685792769132488e-05,
      "loss": 1.7729,
      "step": 205700
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.4836136102676392,
      "learning_rate": 1.6856575797547757e-05,
      "loss": 1.6615,
      "step": 205701
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.522850513458252,
      "learning_rate": 1.6855223956412423e-05,
      "loss": 1.9498,
      "step": 205702
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.474106788635254,
      "learning_rate": 1.685387216791898e-05,
      "loss": 1.9071,
      "step": 205703
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.5067963600158691,
      "learning_rate": 1.6852520432067796e-05,
      "loss": 1.9469,
      "step": 205704
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.4409308433532715,
      "learning_rate": 1.6851168748859035e-05,
      "loss": 1.9324,
      "step": 205705
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.5388213396072388,
      "learning_rate": 1.6849817118293063e-05,
      "loss": 2.1873,
      "step": 205706
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.415518879890442,
      "learning_rate": 1.6848465540369982e-05,
      "loss": 2.1871,
      "step": 205707
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.5640590190887451,
      "learning_rate": 1.6847114015090158e-05,
      "loss": 1.9486,
      "step": 205708
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.5228439569473267,
      "learning_rate": 1.6845762542453757e-05,
      "loss": 1.9423,
      "step": 205709
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.507106900215149,
      "learning_rate": 1.684441112246111e-05,
      "loss": 2.0081,
      "step": 205710
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.3486852645874023,
      "learning_rate": 1.6843059755112386e-05,
      "loss": 1.7276,
      "step": 205711
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.6131356954574585,
      "learning_rate": 1.6841708440407985e-05,
      "loss": 2.1412,
      "step": 205712
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.6026626825332642,
      "learning_rate": 1.6840357178347976e-05,
      "loss": 2.0085,
      "step": 205713
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.5857460498809814,
      "learning_rate": 1.6839005968932684e-05,
      "loss": 2.0396,
      "step": 205714
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.6005828380584717,
      "learning_rate": 1.6837654812162384e-05,
      "loss": 2.0868,
      "step": 205715
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.6172688007354736,
      "learning_rate": 1.6836303708037302e-05,
      "loss": 2.4434,
      "step": 205716
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.4608173370361328,
      "learning_rate": 1.6834952656557676e-05,
      "loss": 1.8985,
      "step": 205717
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.390264868736267,
      "learning_rate": 1.683360165772384e-05,
      "loss": 1.8967,
      "step": 205718
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.5511529445648193,
      "learning_rate": 1.6832250711535954e-05,
      "loss": 1.9847,
      "step": 205719
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.4332603216171265,
      "learning_rate": 1.6830899817994292e-05,
      "loss": 2.0095,
      "step": 205720
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.4394710063934326,
      "learning_rate": 1.6829548977099085e-05,
      "loss": 1.9681,
      "step": 205721
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.4743587970733643,
      "learning_rate": 1.682819818885066e-05,
      "loss": 2.1518,
      "step": 205722
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.4370052814483643,
      "learning_rate": 1.6826847453249158e-05,
      "loss": 1.9215,
      "step": 205723
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.360433578491211,
      "learning_rate": 1.6825496770294943e-05,
      "loss": 1.9941,
      "step": 205724
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.5256541967391968,
      "learning_rate": 1.6824146139988217e-05,
      "loss": 2.1017,
      "step": 205725
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.5544443130493164,
      "learning_rate": 1.6822795562329206e-05,
      "loss": 2.0858,
      "step": 205726
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.7038105726242065,
      "learning_rate": 1.682144503731815e-05,
      "loss": 1.837,
      "step": 205727
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.3632128238677979,
      "learning_rate": 1.682009456495538e-05,
      "loss": 1.8579,
      "step": 205728
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.3993794918060303,
      "learning_rate": 1.6818744145241057e-05,
      "loss": 2.0139,
      "step": 205729
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.3613083362579346,
      "learning_rate": 1.6817393778175525e-05,
      "loss": 2.0038,
      "step": 205730
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.296961784362793,
      "learning_rate": 1.6816043463758976e-05,
      "loss": 2.0448,
      "step": 205731
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.468732476234436,
      "learning_rate": 1.681469320199168e-05,
      "loss": 2.1353,
      "step": 205732
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.5138503313064575,
      "learning_rate": 1.68133429928738e-05,
      "loss": 1.7577,
      "step": 205733
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.4276524782180786,
      "learning_rate": 1.6811992836405742e-05,
      "loss": 1.8617,
      "step": 205734
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.4885178804397583,
      "learning_rate": 1.681064273258763e-05,
      "loss": 1.9132,
      "step": 205735
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.4142869710922241,
      "learning_rate": 1.6809292681419805e-05,
      "loss": 2.0407,
      "step": 205736
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.4829496145248413,
      "learning_rate": 1.68079426829025e-05,
      "loss": 2.1011,
      "step": 205737
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.507081389427185,
      "learning_rate": 1.6806592737035908e-05,
      "loss": 2.2288,
      "step": 205738
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.475991129875183,
      "learning_rate": 1.6805242843820298e-05,
      "loss": 2.1305,
      "step": 205739
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.3607227802276611,
      "learning_rate": 1.6803893003255975e-05,
      "loss": 1.8134,
      "step": 205740
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.3712431192398071,
      "learning_rate": 1.6802543215343102e-05,
      "loss": 2.0635,
      "step": 205741
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.4165593385696411,
      "learning_rate": 1.6801193480082044e-05,
      "loss": 1.9322,
      "step": 205742
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.46260404586792,
      "learning_rate": 1.6799843797472967e-05,
      "loss": 1.9635,
      "step": 205743
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.4265525341033936,
      "learning_rate": 1.6798494167516173e-05,
      "loss": 2.0718,
      "step": 205744
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.4573211669921875,
      "learning_rate": 1.6797144590211832e-05,
      "loss": 2.1569,
      "step": 205745
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.39682936668396,
      "learning_rate": 1.6795795065560268e-05,
      "loss": 1.8487,
      "step": 205746
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.5685920715332031,
      "learning_rate": 1.6794445593561688e-05,
      "loss": 1.8381,
      "step": 205747
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.3984626531600952,
      "learning_rate": 1.679309617421639e-05,
      "loss": 1.9039,
      "step": 205748
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.4555020332336426,
      "learning_rate": 1.6791746807524643e-05,
      "loss": 2.0336,
      "step": 205749
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.505638837814331,
      "learning_rate": 1.679039749348664e-05,
      "loss": 1.9547,
      "step": 205750
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.520142912864685,
      "learning_rate": 1.6789048232102585e-05,
      "loss": 2.1268,
      "step": 205751
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.5325950384140015,
      "learning_rate": 1.6787699023372847e-05,
      "loss": 2.0379,
      "step": 205752
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.3979542255401611,
      "learning_rate": 1.6786349867297587e-05,
      "loss": 2.0702,
      "step": 205753
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.512282371520996,
      "learning_rate": 1.678500076387711e-05,
      "loss": 1.9772,
      "step": 205754
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.5649974346160889,
      "learning_rate": 1.6783651713111677e-05,
      "loss": 1.9679,
      "step": 205755
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.5833100080490112,
      "learning_rate": 1.6782302715001496e-05,
      "loss": 2.0662,
      "step": 205756
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.4437270164489746,
      "learning_rate": 1.6780953769546755e-05,
      "loss": 1.8344,
      "step": 205757
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.4590473175048828,
      "learning_rate": 1.6779604876747865e-05,
      "loss": 1.9789,
      "step": 205758
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.4017112255096436,
      "learning_rate": 1.6778256036604954e-05,
      "loss": 2.253,
      "step": 205759
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.3287070989608765,
      "learning_rate": 1.677690724911832e-05,
      "loss": 1.7823,
      "step": 205760
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.5155832767486572,
      "learning_rate": 1.6775558514288235e-05,
      "loss": 1.9537,
      "step": 205761
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.4091796875,
      "learning_rate": 1.6774209832114926e-05,
      "loss": 1.8488,
      "step": 205762
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.3614393472671509,
      "learning_rate": 1.6772861202598565e-05,
      "loss": 2.014,
      "step": 205763
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.6681632995605469,
      "learning_rate": 1.677151262573955e-05,
      "loss": 2.2727,
      "step": 205764
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.638597011566162,
      "learning_rate": 1.6770164101537975e-05,
      "loss": 1.9688,
      "step": 205765
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.5122320652008057,
      "learning_rate": 1.6768815629994248e-05,
      "loss": 2.0499,
      "step": 205766
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.4751536846160889,
      "learning_rate": 1.6767467211108533e-05,
      "loss": 1.7931,
      "step": 205767
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.3716343641281128,
      "learning_rate": 1.676611884488106e-05,
      "loss": 2.1258,
      "step": 205768
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.555981993675232,
      "learning_rate": 1.6764770531312133e-05,
      "loss": 1.9269,
      "step": 205769
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.5772053003311157,
      "learning_rate": 1.6763422270401983e-05,
      "loss": 1.9405,
      "step": 205770
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.490270733833313,
      "learning_rate": 1.676207406215084e-05,
      "loss": 1.8006,
      "step": 205771
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.4733277559280396,
      "learning_rate": 1.6760725906559013e-05,
      "loss": 1.8229,
      "step": 205772
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.5185577869415283,
      "learning_rate": 1.6759377803626694e-05,
      "loss": 1.939,
      "step": 205773
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.5569677352905273,
      "learning_rate": 1.675802975335412e-05,
      "loss": 1.8082,
      "step": 205774
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.4216188192367554,
      "learning_rate": 1.675668175574162e-05,
      "loss": 1.843,
      "step": 205775
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.3495867252349854,
      "learning_rate": 1.6755333810789396e-05,
      "loss": 2.0307,
      "step": 205776
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.60318922996521,
      "learning_rate": 1.675398591849768e-05,
      "loss": 1.9492,
      "step": 205777
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.4954768419265747,
      "learning_rate": 1.6752638078866777e-05,
      "loss": 1.9115,
      "step": 205778
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.355728268623352,
      "learning_rate": 1.675129029189688e-05,
      "loss": 2.0289,
      "step": 205779
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.4371528625488281,
      "learning_rate": 1.674994255758826e-05,
      "loss": 1.9712,
      "step": 205780
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.5351707935333252,
      "learning_rate": 1.674859487594118e-05,
      "loss": 1.9234,
      "step": 205781
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.546675682067871,
      "learning_rate": 1.674724724695584e-05,
      "loss": 1.9741,
      "step": 205782
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.6561448574066162,
      "learning_rate": 1.6745899670632612e-05,
      "loss": 1.9925,
      "step": 205783
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.3528972864151,
      "learning_rate": 1.6744552146971656e-05,
      "loss": 1.7725,
      "step": 205784
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.5659328699111938,
      "learning_rate": 1.6743204675973177e-05,
      "loss": 1.8415,
      "step": 205785
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.4517287015914917,
      "learning_rate": 1.6741857257637538e-05,
      "loss": 1.9389,
      "step": 205786
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.7026057243347168,
      "learning_rate": 1.6740509891964902e-05,
      "loss": 2.0549,
      "step": 205787
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.3531157970428467,
      "learning_rate": 1.6739162578955544e-05,
      "loss": 1.7251,
      "step": 205788
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.439204454421997,
      "learning_rate": 1.673781531860976e-05,
      "loss": 1.9731,
      "step": 205789
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.4356237649917603,
      "learning_rate": 1.6736468110927713e-05,
      "loss": 1.7928,
      "step": 205790
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.5251754522323608,
      "learning_rate": 1.6735120955909776e-05,
      "loss": 2.0476,
      "step": 205791
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.4200814962387085,
      "learning_rate": 1.6733773853556075e-05,
      "loss": 1.8324,
      "step": 205792
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.7211167812347412,
      "learning_rate": 1.673242680386695e-05,
      "loss": 1.9906,
      "step": 205793
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.4971565008163452,
      "learning_rate": 1.673107980684256e-05,
      "loss": 1.7638,
      "step": 205794
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.557337999343872,
      "learning_rate": 1.6729732862483246e-05,
      "loss": 1.7933,
      "step": 205795
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.4498915672302246,
      "learning_rate": 1.6728385970789204e-05,
      "loss": 1.8949,
      "step": 205796
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.4864037036895752,
      "learning_rate": 1.6727039131760733e-05,
      "loss": 2.0935,
      "step": 205797
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.4908620119094849,
      "learning_rate": 1.6725692345398033e-05,
      "loss": 2.0274,
      "step": 205798
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.4379549026489258,
      "learning_rate": 1.6724345611701406e-05,
      "loss": 1.9151,
      "step": 205799
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.4968719482421875,
      "learning_rate": 1.6722998930670983e-05,
      "loss": 2.0661,
      "step": 205800
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.399977207183838,
      "learning_rate": 1.67216523023072e-05,
      "loss": 1.8573,
      "step": 205801
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.615265965461731,
      "learning_rate": 1.6720305726610117e-05,
      "loss": 1.913,
      "step": 205802
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.4406352043151855,
      "learning_rate": 1.671895920358014e-05,
      "loss": 1.9609,
      "step": 205803
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.4932844638824463,
      "learning_rate": 1.6717612733217466e-05,
      "loss": 2.1376,
      "step": 205804
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.4701526165008545,
      "learning_rate": 1.6716266315522332e-05,
      "loss": 1.9092,
      "step": 205805
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.6273783445358276,
      "learning_rate": 1.6714919950494932e-05,
      "loss": 2.0458,
      "step": 205806
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.498218059539795,
      "learning_rate": 1.6713573638135637e-05,
      "loss": 2.1258,
      "step": 205807
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.4825488328933716,
      "learning_rate": 1.6712227378444576e-05,
      "loss": 1.9153,
      "step": 205808
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.4499307870864868,
      "learning_rate": 1.671088117142212e-05,
      "loss": 1.9776,
      "step": 205809
    },
    {
      "epoch": 2.68,
      "grad_norm": 2.358848810195923,
      "learning_rate": 1.670953501706843e-05,
      "loss": 1.932,
      "step": 205810
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.4463765621185303,
      "learning_rate": 1.6708188915383814e-05,
      "loss": 2.0503,
      "step": 205811
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.3821824789047241,
      "learning_rate": 1.670684286636843e-05,
      "loss": 1.7729,
      "step": 205812
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.4822012186050415,
      "learning_rate": 1.6705496870022648e-05,
      "loss": 1.8662,
      "step": 205813
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.3626396656036377,
      "learning_rate": 1.6704150926346604e-05,
      "loss": 1.9253,
      "step": 205814
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.4611144065856934,
      "learning_rate": 1.670280503534066e-05,
      "loss": 1.9098,
      "step": 205815
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.4620474576950073,
      "learning_rate": 1.6701459197005017e-05,
      "loss": 1.9705,
      "step": 205816
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.4632866382598877,
      "learning_rate": 1.6700113411339943e-05,
      "loss": 1.9699,
      "step": 205817
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.4630248546600342,
      "learning_rate": 1.669876767834557e-05,
      "loss": 2.0635,
      "step": 205818
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.437412142753601,
      "learning_rate": 1.669742199802233e-05,
      "loss": 2.0173,
      "step": 205819
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.5486640930175781,
      "learning_rate": 1.6696076370370325e-05,
      "loss": 2.1204,
      "step": 205820
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.5376241207122803,
      "learning_rate": 1.669473079538992e-05,
      "loss": 1.9124,
      "step": 205821
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.4392120838165283,
      "learning_rate": 1.6693385273081285e-05,
      "loss": 1.9194,
      "step": 205822
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.5279854536056519,
      "learning_rate": 1.6692039803444712e-05,
      "loss": 1.9061,
      "step": 205823
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.3405523300170898,
      "learning_rate": 1.6690694386480408e-05,
      "loss": 1.9321,
      "step": 205824
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.4549087285995483,
      "learning_rate": 1.6689349022188703e-05,
      "loss": 2.009,
      "step": 205825
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.6491072177886963,
      "learning_rate": 1.668800371056973e-05,
      "loss": 2.1218,
      "step": 205826
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.5651570558547974,
      "learning_rate": 1.668665845162386e-05,
      "loss": 1.9938,
      "step": 205827
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.5894376039505005,
      "learning_rate": 1.6685313245351286e-05,
      "loss": 1.9806,
      "step": 205828
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.6781835556030273,
      "learning_rate": 1.6683968091752243e-05,
      "loss": 2.0892,
      "step": 205829
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.5779163837432861,
      "learning_rate": 1.6682622990826967e-05,
      "loss": 2.0938,
      "step": 205830
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.4714844226837158,
      "learning_rate": 1.6681277942575786e-05,
      "loss": 1.9387,
      "step": 205831
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.382333755493164,
      "learning_rate": 1.667993294699884e-05,
      "loss": 2.0091,
      "step": 205832
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.4968079328536987,
      "learning_rate": 1.6678588004096526e-05,
      "loss": 1.804,
      "step": 205833
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.3377132415771484,
      "learning_rate": 1.6677243113868978e-05,
      "loss": 1.9231,
      "step": 205834
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.4780497550964355,
      "learning_rate": 1.667589827631649e-05,
      "loss": 2.0602,
      "step": 205835
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.3969718217849731,
      "learning_rate": 1.667455349143927e-05,
      "loss": 1.7507,
      "step": 205836
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.5863949060440063,
      "learning_rate": 1.667320875923761e-05,
      "loss": 1.884,
      "step": 205837
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.474064826965332,
      "learning_rate": 1.6671864079711716e-05,
      "loss": 2.1997,
      "step": 205838
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.4369597434997559,
      "learning_rate": 1.6670519452861918e-05,
      "loss": 1.9291,
      "step": 205839
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.4887596368789673,
      "learning_rate": 1.666917487868845e-05,
      "loss": 1.8039,
      "step": 205840
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.489076852798462,
      "learning_rate": 1.6667830357191482e-05,
      "loss": 1.9929,
      "step": 205841
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.3432711362838745,
      "learning_rate": 1.6666485888371305e-05,
      "loss": 2.0457,
      "step": 205842
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.6388466358184814,
      "learning_rate": 1.6665141472228193e-05,
      "loss": 1.8845,
      "step": 205843
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.4595259428024292,
      "learning_rate": 1.6663797108762344e-05,
      "loss": 1.9826,
      "step": 205844
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.3736084699630737,
      "learning_rate": 1.666245279797409e-05,
      "loss": 1.8846,
      "step": 205845
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.611296534538269,
      "learning_rate": 1.6661108539863632e-05,
      "loss": 2.2481,
      "step": 205846
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.321035623550415,
      "learning_rate": 1.6659764334431234e-05,
      "loss": 1.9391,
      "step": 205847
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.6726247072219849,
      "learning_rate": 1.6658420181677103e-05,
      "loss": 1.9759,
      "step": 205848
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.5573303699493408,
      "learning_rate": 1.665707608160153e-05,
      "loss": 1.9609,
      "step": 205849
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.4756211042404175,
      "learning_rate": 1.6655732034204716e-05,
      "loss": 2.0099,
      "step": 205850
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.3484199047088623,
      "learning_rate": 1.6654388039487e-05,
      "loss": 1.832,
      "step": 205851
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.6917604207992554,
      "learning_rate": 1.6653044097448544e-05,
      "loss": 1.8847,
      "step": 205852
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.5720235109329224,
      "learning_rate": 1.665170020808968e-05,
      "loss": 2.0866,
      "step": 205853
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.5086833238601685,
      "learning_rate": 1.665035637141061e-05,
      "loss": 2.1849,
      "step": 205854
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.5519351959228516,
      "learning_rate": 1.66490125874116e-05,
      "loss": 1.9336,
      "step": 205855
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.5512945652008057,
      "learning_rate": 1.664766885609282e-05,
      "loss": 2.0924,
      "step": 205856
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.628064513206482,
      "learning_rate": 1.6646325177454634e-05,
      "loss": 2.0228,
      "step": 205857
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.5071382522583008,
      "learning_rate": 1.6644981551497205e-05,
      "loss": 1.9439,
      "step": 205858
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.3838284015655518,
      "learning_rate": 1.664363797822087e-05,
      "loss": 1.9179,
      "step": 205859
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.3549801111221313,
      "learning_rate": 1.6642294457625828e-05,
      "loss": 1.82,
      "step": 205860
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.4459209442138672,
      "learning_rate": 1.6640950989712342e-05,
      "loss": 1.9858,
      "step": 205861
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.4316127300262451,
      "learning_rate": 1.6639607574480616e-05,
      "loss": 2.0699,
      "step": 205862
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.7106115818023682,
      "learning_rate": 1.663826421193095e-05,
      "loss": 2.0822,
      "step": 205863
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.4094403982162476,
      "learning_rate": 1.6636920902063578e-05,
      "loss": 1.9229,
      "step": 205864
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.4880050420761108,
      "learning_rate": 1.663557764487876e-05,
      "loss": 1.9975,
      "step": 205865
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.5414670705795288,
      "learning_rate": 1.6634234440376736e-05,
      "loss": 2.0033,
      "step": 205866
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.4412962198257446,
      "learning_rate": 1.663289128855774e-05,
      "loss": 1.9525,
      "step": 205867
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.5501993894577026,
      "learning_rate": 1.6631548189422062e-05,
      "loss": 1.9399,
      "step": 205868
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.4895358085632324,
      "learning_rate": 1.663020514296991e-05,
      "loss": 2.1942,
      "step": 205869
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.606654167175293,
      "learning_rate": 1.662886214920155e-05,
      "loss": 2.0189,
      "step": 205870
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.5597258806228638,
      "learning_rate": 1.6627519208117247e-05,
      "loss": 2.2415,
      "step": 205871
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.4609267711639404,
      "learning_rate": 1.6626176319717232e-05,
      "loss": 2.0213,
      "step": 205872
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.4940674304962158,
      "learning_rate": 1.6624833484001742e-05,
      "loss": 2.0566,
      "step": 205873
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.4866604804992676,
      "learning_rate": 1.6623490700971077e-05,
      "loss": 1.8098,
      "step": 205874
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.3579710721969604,
      "learning_rate": 1.6622147970625466e-05,
      "loss": 1.9834,
      "step": 205875
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.556069016456604,
      "learning_rate": 1.6620805292965077e-05,
      "loss": 1.9978,
      "step": 205876
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.6787503957748413,
      "learning_rate": 1.661946266799028e-05,
      "loss": 2.1667,
      "step": 205877
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.50586998462677,
      "learning_rate": 1.6618120095701303e-05,
      "loss": 1.983,
      "step": 205878
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.4689182043075562,
      "learning_rate": 1.6616777576098282e-05,
      "loss": 2.042,
      "step": 205879
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.5088931322097778,
      "learning_rate": 1.6615435109181618e-05,
      "loss": 1.8705,
      "step": 205880
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.6353075504302979,
      "learning_rate": 1.6614092694951443e-05,
      "loss": 1.9123,
      "step": 205881
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.4187461137771606,
      "learning_rate": 1.661275033340812e-05,
      "loss": 2.1376,
      "step": 205882
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.4863779544830322,
      "learning_rate": 1.661140802455182e-05,
      "loss": 1.9336,
      "step": 205883
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.4932880401611328,
      "learning_rate": 1.6610065768382807e-05,
      "loss": 1.8846,
      "step": 205884
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.4605345726013184,
      "learning_rate": 1.6608723564901284e-05,
      "loss": 1.937,
      "step": 205885
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.3449805974960327,
      "learning_rate": 1.6607381414107612e-05,
      "loss": 1.8883,
      "step": 205886
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.7300326824188232,
      "learning_rate": 1.6606039316001897e-05,
      "loss": 2.0537,
      "step": 205887
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.462533712387085,
      "learning_rate": 1.6604697270584565e-05,
      "loss": 1.9502,
      "step": 205888
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.2880085706710815,
      "learning_rate": 1.6603355277855722e-05,
      "loss": 2.0414,
      "step": 205889
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.5683742761611938,
      "learning_rate": 1.66020133378157e-05,
      "loss": 1.9446,
      "step": 205890
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.6385748386383057,
      "learning_rate": 1.660067145046463e-05,
      "loss": 2.0935,
      "step": 205891
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.574676275253296,
      "learning_rate": 1.6599329615802915e-05,
      "loss": 1.8358,
      "step": 205892
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.4020825624465942,
      "learning_rate": 1.6597987833830682e-05,
      "loss": 2.0282,
      "step": 205893
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.5907727479934692,
      "learning_rate": 1.6596646104548306e-05,
      "loss": 1.9014,
      "step": 205894
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.4950460195541382,
      "learning_rate": 1.6595304427955913e-05,
      "loss": 2.0356,
      "step": 205895
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.4242171049118042,
      "learning_rate": 1.6593962804053838e-05,
      "loss": 1.9839,
      "step": 205896
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.4411437511444092,
      "learning_rate": 1.659262123284225e-05,
      "loss": 1.9625,
      "step": 205897
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.3506146669387817,
      "learning_rate": 1.6591279714321483e-05,
      "loss": 2.0707,
      "step": 205898
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.5673779249191284,
      "learning_rate": 1.6589938248491697e-05,
      "loss": 2.1543,
      "step": 205899
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.328870415687561,
      "learning_rate": 1.658859683535323e-05,
      "loss": 1.8606,
      "step": 205900
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.425436019897461,
      "learning_rate": 1.6587255474906315e-05,
      "loss": 1.9647,
      "step": 205901
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.8435994386672974,
      "learning_rate": 1.658591416715115e-05,
      "loss": 1.9545,
      "step": 205902
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.4913711547851562,
      "learning_rate": 1.6584572912088002e-05,
      "loss": 2.1509,
      "step": 205903
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.6101211309432983,
      "learning_rate": 1.658323170971714e-05,
      "loss": 2.0437,
      "step": 205904
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.527203917503357,
      "learning_rate": 1.658189056003879e-05,
      "loss": 2.0582,
      "step": 205905
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.4158899784088135,
      "learning_rate": 1.6580549463053226e-05,
      "loss": 1.9729,
      "step": 205906
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.5332132577896118,
      "learning_rate": 1.6579208418760713e-05,
      "loss": 2.1114,
      "step": 205907
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.4786192178726196,
      "learning_rate": 1.6577867427161486e-05,
      "loss": 1.9387,
      "step": 205908
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.503943681716919,
      "learning_rate": 1.6576526488255704e-05,
      "loss": 1.8525,
      "step": 205909
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.4570034742355347,
      "learning_rate": 1.657518560204377e-05,
      "loss": 2.0099,
      "step": 205910
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.3742364645004272,
      "learning_rate": 1.6573844768525825e-05,
      "loss": 1.8082,
      "step": 205911
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.4194411039352417,
      "learning_rate": 1.6572503987702158e-05,
      "loss": 1.9722,
      "step": 205912
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.8031572103500366,
      "learning_rate": 1.657116325957304e-05,
      "loss": 2.0392,
      "step": 205913
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.3718620538711548,
      "learning_rate": 1.6569822584138703e-05,
      "loss": 1.8415,
      "step": 205914
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.3309216499328613,
      "learning_rate": 1.6568481961399316e-05,
      "loss": 1.9946,
      "step": 205915
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.5123982429504395,
      "learning_rate": 1.6567141391355244e-05,
      "loss": 1.9586,
      "step": 205916
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.4237529039382935,
      "learning_rate": 1.656580087400665e-05,
      "loss": 1.7963,
      "step": 205917
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.3698042631149292,
      "learning_rate": 1.6564460409353874e-05,
      "loss": 1.7135,
      "step": 205918
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.4321043491363525,
      "learning_rate": 1.656311999739708e-05,
      "loss": 2.2008,
      "step": 205919
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.4323025941848755,
      "learning_rate": 1.656177963813663e-05,
      "loss": 1.8956,
      "step": 205920
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.402154564857483,
      "learning_rate": 1.6560439331572594e-05,
      "loss": 1.9305,
      "step": 205921
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.5206376314163208,
      "learning_rate": 1.6559099077705372e-05,
      "loss": 2.0161,
      "step": 205922
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.5140060186386108,
      "learning_rate": 1.6557758876535133e-05,
      "loss": 2.2219,
      "step": 205923
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.3582853078842163,
      "learning_rate": 1.6556418728062205e-05,
      "loss": 2.0584,
      "step": 205924
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.4887776374816895,
      "learning_rate": 1.6555078632286724e-05,
      "loss": 2.0253,
      "step": 205925
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.4036684036254883,
      "learning_rate": 1.655373858920912e-05,
      "loss": 2.1799,
      "step": 205926
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.3668267726898193,
      "learning_rate": 1.65523985988294e-05,
      "loss": 1.8781,
      "step": 205927
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.5722498893737793,
      "learning_rate": 1.655105866114802e-05,
      "loss": 1.9179,
      "step": 205928
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.4330296516418457,
      "learning_rate": 1.654971877616509e-05,
      "loss": 2.0611,
      "step": 205929
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.5220803022384644,
      "learning_rate": 1.654837894388097e-05,
      "loss": 1.9624,
      "step": 205930
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.4326187372207642,
      "learning_rate": 1.6547039164295795e-05,
      "loss": 2.1172,
      "step": 205931
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.3276809453964233,
      "learning_rate": 1.6545699437409966e-05,
      "loss": 1.7911,
      "step": 205932
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.4955768585205078,
      "learning_rate": 1.654435976322358e-05,
      "loss": 1.9184,
      "step": 205933
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.5891083478927612,
      "learning_rate": 1.6543020141736973e-05,
      "loss": 1.876,
      "step": 205934
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.4611581563949585,
      "learning_rate": 1.654168057295031e-05,
      "loss": 1.8951,
      "step": 205935
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.3915331363677979,
      "learning_rate": 1.6540341056863958e-05,
      "loss": 1.8675,
      "step": 205936
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.5754015445709229,
      "learning_rate": 1.6539001593478085e-05,
      "loss": 2.0945,
      "step": 205937
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.6329082250595093,
      "learning_rate": 1.6537662182793022e-05,
      "loss": 2.1377,
      "step": 205938
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.4830687046051025,
      "learning_rate": 1.6536322824808866e-05,
      "loss": 1.9278,
      "step": 205939
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.6367614269256592,
      "learning_rate": 1.6534983519526025e-05,
      "loss": 1.8632,
      "step": 205940
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.448082447052002,
      "learning_rate": 1.6533644266944623e-05,
      "loss": 1.9106,
      "step": 205941
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.456201195716858,
      "learning_rate": 1.653230506706503e-05,
      "loss": 2.0346,
      "step": 205942
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.4525610208511353,
      "learning_rate": 1.653096591988735e-05,
      "loss": 1.9745,
      "step": 205943
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.5489717721939087,
      "learning_rate": 1.652962682541198e-05,
      "loss": 2.1188,
      "step": 205944
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.5207103490829468,
      "learning_rate": 1.6528287783639114e-05,
      "loss": 2.1743,
      "step": 205945
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.4098433256149292,
      "learning_rate": 1.6526948794568994e-05,
      "loss": 1.8213,
      "step": 205946
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.552359938621521,
      "learning_rate": 1.6525609858201816e-05,
      "loss": 1.9559,
      "step": 205947
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.402420997619629,
      "learning_rate": 1.6524270974537913e-05,
      "loss": 1.9434,
      "step": 205948
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.4471790790557861,
      "learning_rate": 1.652293214357745e-05,
      "loss": 2.1174,
      "step": 205949
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.4287625551223755,
      "learning_rate": 1.6521593365320762e-05,
      "loss": 1.8287,
      "step": 205950
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.4298397302627563,
      "learning_rate": 1.6520254639768084e-05,
      "loss": 1.9217,
      "step": 205951
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.5986487865447998,
      "learning_rate": 1.6518915966919578e-05,
      "loss": 2.1161,
      "step": 205952
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.3922456502914429,
      "learning_rate": 1.6517577346775613e-05,
      "loss": 1.9561,
      "step": 205953
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.4273810386657715,
      "learning_rate": 1.6516238779336387e-05,
      "loss": 2.0094,
      "step": 205954
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.4671704769134521,
      "learning_rate": 1.6514900264602105e-05,
      "loss": 1.8902,
      "step": 205955
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.3025996685028076,
      "learning_rate": 1.6513561802573063e-05,
      "loss": 1.7579,
      "step": 205956
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.4412627220153809,
      "learning_rate": 1.6512223393249524e-05,
      "loss": 1.9548,
      "step": 205957
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.5325263738632202,
      "learning_rate": 1.6510885036631695e-05,
      "loss": 2.2143,
      "step": 205958
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.4437694549560547,
      "learning_rate": 1.6509546732719835e-05,
      "loss": 1.9565,
      "step": 205959
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.6079814434051514,
      "learning_rate": 1.650820848151425e-05,
      "loss": 1.859,
      "step": 205960
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.4647982120513916,
      "learning_rate": 1.6506870283015074e-05,
      "loss": 1.7839,
      "step": 205961
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.4116930961608887,
      "learning_rate": 1.6505532137222665e-05,
      "loss": 1.9087,
      "step": 205962
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.452288269996643,
      "learning_rate": 1.650419404413723e-05,
      "loss": 2.0742,
      "step": 205963
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.3502988815307617,
      "learning_rate": 1.650285600375897e-05,
      "loss": 1.8561,
      "step": 205964
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.4568647146224976,
      "learning_rate": 1.6501518016088243e-05,
      "loss": 1.9398,
      "step": 205965
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.4754717350006104,
      "learning_rate": 1.6500180081125157e-05,
      "loss": 1.9828,
      "step": 205966
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.5475406646728516,
      "learning_rate": 1.649884219887011e-05,
      "loss": 2.0415,
      "step": 205967
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.406465768814087,
      "learning_rate": 1.6497504369323265e-05,
      "loss": 1.8361,
      "step": 205968
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.6188247203826904,
      "learning_rate": 1.6496166592484893e-05,
      "loss": 1.8065,
      "step": 205969
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.6528990268707275,
      "learning_rate": 1.6494828868355192e-05,
      "loss": 1.9285,
      "step": 205970
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.4223662614822388,
      "learning_rate": 1.6493491196934495e-05,
      "loss": 2.0257,
      "step": 205971
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.408495545387268,
      "learning_rate": 1.6492153578222965e-05,
      "loss": 1.8638,
      "step": 205972
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.5608478784561157,
      "learning_rate": 1.6490816012220942e-05,
      "loss": 1.9867,
      "step": 205973
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.4138356447219849,
      "learning_rate": 1.6489478498928656e-05,
      "loss": 1.8612,
      "step": 205974
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.4917548894882202,
      "learning_rate": 1.6488141038346304e-05,
      "loss": 1.8961,
      "step": 205975
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.4191662073135376,
      "learning_rate": 1.6486803630474088e-05,
      "loss": 2.0183,
      "step": 205976
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.5317310094833374,
      "learning_rate": 1.648546627531241e-05,
      "loss": 2.1088,
      "step": 205977
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.4935461282730103,
      "learning_rate": 1.6484128972861366e-05,
      "loss": 2.0678,
      "step": 205978
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.407395839691162,
      "learning_rate": 1.6482791723121324e-05,
      "loss": 2.0324,
      "step": 205979
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.5909974575042725,
      "learning_rate": 1.6481454526092484e-05,
      "loss": 1.9906,
      "step": 205980
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.5593461990356445,
      "learning_rate": 1.648011738177508e-05,
      "loss": 2.0027,
      "step": 205981
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.420245885848999,
      "learning_rate": 1.6478780290169348e-05,
      "loss": 2.0735,
      "step": 205982
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.5585271120071411,
      "learning_rate": 1.6477443251275614e-05,
      "loss": 1.8272,
      "step": 205983
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.6723827123641968,
      "learning_rate": 1.6476106265094014e-05,
      "loss": 1.9231,
      "step": 205984
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.6209772825241089,
      "learning_rate": 1.6474769331624914e-05,
      "loss": 2.1168,
      "step": 205985
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.3743767738342285,
      "learning_rate": 1.647343245086845e-05,
      "loss": 2.0127,
      "step": 205986
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.3254246711730957,
      "learning_rate": 1.647209562282502e-05,
      "loss": 1.777,
      "step": 205987
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.4291386604309082,
      "learning_rate": 1.6470758847494692e-05,
      "loss": 2.1452,
      "step": 205988
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.4861282110214233,
      "learning_rate": 1.6469422124877863e-05,
      "loss": 2.1932,
      "step": 205989
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.3475556373596191,
      "learning_rate": 1.6468085454974665e-05,
      "loss": 2.1338,
      "step": 205990
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.3381162881851196,
      "learning_rate": 1.646674883778544e-05,
      "loss": 2.0457,
      "step": 205991
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.423346996307373,
      "learning_rate": 1.6465412273310342e-05,
      "loss": 1.9325,
      "step": 205992
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.4730123281478882,
      "learning_rate": 1.6464075761549778e-05,
      "loss": 2.1145,
      "step": 205993
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.4184805154800415,
      "learning_rate": 1.6462739302503815e-05,
      "loss": 1.8742,
      "step": 205994
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.415184497833252,
      "learning_rate": 1.6461402896172815e-05,
      "loss": 1.9789,
      "step": 205995
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.4967200756072998,
      "learning_rate": 1.6460066542556948e-05,
      "loss": 1.9356,
      "step": 205996
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.4402167797088623,
      "learning_rate": 1.645873024165655e-05,
      "loss": 1.8886,
      "step": 205997
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.4792379140853882,
      "learning_rate": 1.6457393993471778e-05,
      "loss": 1.9094,
      "step": 205998
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.547456979751587,
      "learning_rate": 1.645605779800304e-05,
      "loss": 2.0322,
      "step": 205999
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.5750823020935059,
      "learning_rate": 1.645472165525037e-05,
      "loss": 1.9996,
      "step": 206000
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.337380290031433,
      "learning_rate": 1.6453385565214162e-05,
      "loss": 1.7728,
      "step": 206001
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.6218109130859375,
      "learning_rate": 1.6452049527894584e-05,
      "loss": 2.0059,
      "step": 206002
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.5046418905258179,
      "learning_rate": 1.6450713543291937e-05,
      "loss": 2.0108,
      "step": 206003
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.4105887413024902,
      "learning_rate": 1.6449377611406456e-05,
      "loss": 2.0779,
      "step": 206004
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.6020091772079468,
      "learning_rate": 1.6448041732238436e-05,
      "loss": 2.0875,
      "step": 206005
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.3013052940368652,
      "learning_rate": 1.6446705905788014e-05,
      "loss": 1.9222,
      "step": 206006
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.492398738861084,
      "learning_rate": 1.644537013205556e-05,
      "loss": 2.0284,
      "step": 206007
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.6174119710922241,
      "learning_rate": 1.6444034411041196e-05,
      "loss": 1.9517,
      "step": 206008
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.670470118522644,
      "learning_rate": 1.6442698742745264e-05,
      "loss": 2.1434,
      "step": 206009
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.4841684103012085,
      "learning_rate": 1.6441363127167994e-05,
      "loss": 1.9329,
      "step": 206010
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.5767078399658203,
      "learning_rate": 1.644002756430969e-05,
      "loss": 2.0622,
      "step": 206011
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.6367334127426147,
      "learning_rate": 1.6438692054170444e-05,
      "loss": 1.9375,
      "step": 206012
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.3860424757003784,
      "learning_rate": 1.6437356596750628e-05,
      "loss": 1.8941,
      "step": 206013
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.5078420639038086,
      "learning_rate": 1.643602119205044e-05,
      "loss": 2.0288,
      "step": 206014
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.497615098953247,
      "learning_rate": 1.6434685840070183e-05,
      "loss": 2.1739,
      "step": 206015
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.5429868698120117,
      "learning_rate": 1.643335054081002e-05,
      "loss": 2.0097,
      "step": 206016
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.3423246145248413,
      "learning_rate": 1.643201529427035e-05,
      "loss": 1.8719,
      "step": 206017
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.4070377349853516,
      "learning_rate": 1.6430680100451244e-05,
      "loss": 2.0128,
      "step": 206018
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.4101910591125488,
      "learning_rate": 1.642934495935303e-05,
      "loss": 2.104,
      "step": 206019
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.608027696609497,
      "learning_rate": 1.6428009870975943e-05,
      "loss": 2.0687,
      "step": 206020
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.7633492946624756,
      "learning_rate": 1.6426674835320286e-05,
      "loss": 2.0451,
      "step": 206021
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.8367009162902832,
      "learning_rate": 1.6425339852386223e-05,
      "loss": 1.9159,
      "step": 206022
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.3585585355758667,
      "learning_rate": 1.6424004922174116e-05,
      "loss": 2.0258,
      "step": 206023
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.3679190874099731,
      "learning_rate": 1.642267004468404e-05,
      "loss": 1.8185,
      "step": 206024
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.315811276435852,
      "learning_rate": 1.642133521991642e-05,
      "loss": 1.7036,
      "step": 206025
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.585236668586731,
      "learning_rate": 1.6420000447871363e-05,
      "loss": 2.0002,
      "step": 206026
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.42495858669281,
      "learning_rate": 1.6418665728549196e-05,
      "loss": 1.8593,
      "step": 206027
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.4633058309555054,
      "learning_rate": 1.6417331061950158e-05,
      "loss": 1.8848,
      "step": 206028
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.5210994482040405,
      "learning_rate": 1.641599644807451e-05,
      "loss": 2.0169,
      "step": 206029
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.4258145093917847,
      "learning_rate": 1.6414661886922487e-05,
      "loss": 2.0571,
      "step": 206030
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.4819532632827759,
      "learning_rate": 1.6413327378494323e-05,
      "loss": 2.0671,
      "step": 206031
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.4857584238052368,
      "learning_rate": 1.6411992922790252e-05,
      "loss": 1.7561,
      "step": 206032
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.5406687259674072,
      "learning_rate": 1.641065851981057e-05,
      "loss": 1.9789,
      "step": 206033
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.4104386568069458,
      "learning_rate": 1.6409324169555448e-05,
      "loss": 1.8245,
      "step": 206034
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.4308799505233765,
      "learning_rate": 1.640798987202525e-05,
      "loss": 1.934,
      "step": 206035
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.365129828453064,
      "learning_rate": 1.6406655627220146e-05,
      "loss": 1.8991,
      "step": 206036
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.507403016090393,
      "learning_rate": 1.640532143514043e-05,
      "loss": 1.7704,
      "step": 206037
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.6670578718185425,
      "learning_rate": 1.6403987295786235e-05,
      "loss": 1.9968,
      "step": 206038
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.3308695554733276,
      "learning_rate": 1.6402653209157933e-05,
      "loss": 2.0271,
      "step": 206039
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.48919677734375,
      "learning_rate": 1.640131917525572e-05,
      "loss": 2.0239,
      "step": 206040
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.4493789672851562,
      "learning_rate": 1.6399985194079868e-05,
      "loss": 1.8962,
      "step": 206041
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.4317377805709839,
      "learning_rate": 1.6398651265630635e-05,
      "loss": 1.9871,
      "step": 206042
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.5393840074539185,
      "learning_rate": 1.6397317389908195e-05,
      "loss": 2.0469,
      "step": 206043
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.4758213758468628,
      "learning_rate": 1.6395983566912908e-05,
      "loss": 2.129,
      "step": 206044
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.4757813215255737,
      "learning_rate": 1.6394649796644944e-05,
      "loss": 1.9926,
      "step": 206045
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.4940879344940186,
      "learning_rate": 1.63933160791045e-05,
      "loss": 2.2061,
      "step": 206046
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.4569742679595947,
      "learning_rate": 1.6391982414291982e-05,
      "loss": 2.0182,
      "step": 206047
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.7074902057647705,
      "learning_rate": 1.6390648802207516e-05,
      "loss": 2.1071,
      "step": 206048
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.534488320350647,
      "learning_rate": 1.6389315242851343e-05,
      "loss": 2.0007,
      "step": 206049
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.547206163406372,
      "learning_rate": 1.6387981736223786e-05,
      "loss": 1.808,
      "step": 206050
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.4163062572479248,
      "learning_rate": 1.638664828232502e-05,
      "loss": 2.1047,
      "step": 206051
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.4708062410354614,
      "learning_rate": 1.6385314881155377e-05,
      "loss": 1.7232,
      "step": 206052
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.4396930932998657,
      "learning_rate": 1.6383981532715086e-05,
      "loss": 2.2206,
      "step": 206053
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.5332443714141846,
      "learning_rate": 1.6382648237004314e-05,
      "loss": 2.0233,
      "step": 206054
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.381085991859436,
      "learning_rate": 1.6381314994023366e-05,
      "loss": 2.0599,
      "step": 206055
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.429290771484375,
      "learning_rate": 1.6379981803772502e-05,
      "loss": 2.0763,
      "step": 206056
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.38908851146698,
      "learning_rate": 1.6378648666251892e-05,
      "loss": 1.8951,
      "step": 206057
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.493226170539856,
      "learning_rate": 1.6377315581461936e-05,
      "loss": 1.9433,
      "step": 206058
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.4600498676300049,
      "learning_rate": 1.6375982549402766e-05,
      "loss": 1.9636,
      "step": 206059
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.601750135421753,
      "learning_rate": 1.637464957007465e-05,
      "loss": 1.92,
      "step": 206060
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.573769211769104,
      "learning_rate": 1.6373316643477786e-05,
      "loss": 1.8941,
      "step": 206061
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.6626895666122437,
      "learning_rate": 1.637198376961254e-05,
      "loss": 2.2405,
      "step": 206062
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.4539564847946167,
      "learning_rate": 1.6370650948479046e-05,
      "loss": 2.0178,
      "step": 206063
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.422785758972168,
      "learning_rate": 1.636931818007764e-05,
      "loss": 1.9925,
      "step": 206064
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.7232378721237183,
      "learning_rate": 1.6367985464408516e-05,
      "loss": 1.9245,
      "step": 206065
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.4135913848876953,
      "learning_rate": 1.636665280147198e-05,
      "loss": 2.0324,
      "step": 206066
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.3983943462371826,
      "learning_rate": 1.6365320191268193e-05,
      "loss": 1.7929,
      "step": 206067
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.5086268186569214,
      "learning_rate": 1.636398763379746e-05,
      "loss": 1.8233,
      "step": 206068
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.4718432426452637,
      "learning_rate": 1.6362655129059976e-05,
      "loss": 2.031,
      "step": 206069
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.4454149007797241,
      "learning_rate": 1.6361322677056076e-05,
      "loss": 1.9268,
      "step": 206070
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.5238738059997559,
      "learning_rate": 1.6359990277785927e-05,
      "loss": 1.9911,
      "step": 206071
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.6094892024993896,
      "learning_rate": 1.6358657931249898e-05,
      "loss": 1.9313,
      "step": 206072
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.6815909147262573,
      "learning_rate": 1.635732563744805e-05,
      "loss": 2.0303,
      "step": 206073
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.3312650918960571,
      "learning_rate": 1.6355993396380785e-05,
      "loss": 1.8344,
      "step": 206074
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.4189752340316772,
      "learning_rate": 1.6354661208048237e-05,
      "loss": 1.8985,
      "step": 206075
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.9623568058013916,
      "learning_rate": 1.635332907245077e-05,
      "loss": 2.1396,
      "step": 206076
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.410264015197754,
      "learning_rate": 1.6351996989588522e-05,
      "loss": 1.936,
      "step": 206077
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.5085564851760864,
      "learning_rate": 1.6350664959461857e-05,
      "loss": 1.9765,
      "step": 206078
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.4530342817306519,
      "learning_rate": 1.6349332982070908e-05,
      "loss": 2.2124,
      "step": 206079
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.586532711982727,
      "learning_rate": 1.6348001057416006e-05,
      "loss": 1.8907,
      "step": 206080
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.54526686668396,
      "learning_rate": 1.6346669185497284e-05,
      "loss": 2.0547,
      "step": 206081
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.4651767015457153,
      "learning_rate": 1.6345337366315148e-05,
      "loss": 1.908,
      "step": 206082
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.4638935327529907,
      "learning_rate": 1.6344005599869724e-05,
      "loss": 1.8118,
      "step": 206083
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.4068694114685059,
      "learning_rate": 1.6342673886161385e-05,
      "loss": 1.8879,
      "step": 206084
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.6639641523361206,
      "learning_rate": 1.6341342225190225e-05,
      "loss": 1.949,
      "step": 206085
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.4691144227981567,
      "learning_rate": 1.634001061695661e-05,
      "loss": 1.8328,
      "step": 206086
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.5621235370635986,
      "learning_rate": 1.633867906146068e-05,
      "loss": 1.9376,
      "step": 206087
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.5385894775390625,
      "learning_rate": 1.6337347558702795e-05,
      "loss": 2.0293,
      "step": 206088
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.4569454193115234,
      "learning_rate": 1.6336016108683093e-05,
      "loss": 1.939,
      "step": 206089
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.3337217569351196,
      "learning_rate": 1.6334684711401967e-05,
      "loss": 1.8134,
      "step": 206090
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.39742112159729,
      "learning_rate": 1.6333353366859524e-05,
      "loss": 2.0207,
      "step": 206091
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.434581995010376,
      "learning_rate": 1.6332022075056063e-05,
      "loss": 1.9416,
      "step": 206092
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.3864156007766724,
      "learning_rate": 1.6330690835991844e-05,
      "loss": 2.082,
      "step": 206093
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.513232707977295,
      "learning_rate": 1.6329359649667106e-05,
      "loss": 1.8385,
      "step": 206094
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.477985143661499,
      "learning_rate": 1.6328028516082047e-05,
      "loss": 1.9014,
      "step": 206095
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.3220884799957275,
      "learning_rate": 1.6326697435237068e-05,
      "loss": 1.8765,
      "step": 206096
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.4474016427993774,
      "learning_rate": 1.6325366407132236e-05,
      "loss": 1.63,
      "step": 206097
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.5841645002365112,
      "learning_rate": 1.6324035431767914e-05,
      "loss": 2.1832,
      "step": 206098
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.3342424631118774,
      "learning_rate": 1.6322704509144237e-05,
      "loss": 1.9957,
      "step": 206099
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.4689083099365234,
      "learning_rate": 1.632137363926157e-05,
      "loss": 2.0922,
      "step": 206100
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.5514308214187622,
      "learning_rate": 1.6320042822120117e-05,
      "loss": 1.9611,
      "step": 206101
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.6094588041305542,
      "learning_rate": 1.6318712057720172e-05,
      "loss": 2.1442,
      "step": 206102
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.4061288833618164,
      "learning_rate": 1.631738134606184e-05,
      "loss": 1.913,
      "step": 206103
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.4433153867721558,
      "learning_rate": 1.6316050687145522e-05,
      "loss": 1.8848,
      "step": 206104
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.4941376447677612,
      "learning_rate": 1.6314720080971345e-05,
      "loss": 2.0488,
      "step": 206105
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.5937366485595703,
      "learning_rate": 1.6313389527539677e-05,
      "loss": 2.0252,
      "step": 206106
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.4449018239974976,
      "learning_rate": 1.6312059026850653e-05,
      "loss": 2.0489,
      "step": 206107
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.532723307609558,
      "learning_rate": 1.6310728578904674e-05,
      "loss": 2.373,
      "step": 206108
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.5570372343063354,
      "learning_rate": 1.6309398183701773e-05,
      "loss": 2.0834,
      "step": 206109
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.4590967893600464,
      "learning_rate": 1.630806784124238e-05,
      "loss": 1.8104,
      "step": 206110
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.3630999326705933,
      "learning_rate": 1.6306737551526627e-05,
      "loss": 2.0455,
      "step": 206111
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.5141844749450684,
      "learning_rate": 1.630540731455482e-05,
      "loss": 1.89,
      "step": 206112
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.6198093891143799,
      "learning_rate": 1.6304077130327185e-05,
      "loss": 2.0138,
      "step": 206113
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.4320178031921387,
      "learning_rate": 1.6302746998843997e-05,
      "loss": 2.0114,
      "step": 206114
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.5633190870285034,
      "learning_rate": 1.630141692010548e-05,
      "loss": 1.9942,
      "step": 206115
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.5617289543151855,
      "learning_rate": 1.6300086894111908e-05,
      "loss": 1.879,
      "step": 206116
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.4047276973724365,
      "learning_rate": 1.6298756920863443e-05,
      "loss": 2.2272,
      "step": 206117
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.5174299478530884,
      "learning_rate": 1.629742700036045e-05,
      "loss": 2.0298,
      "step": 206118
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.3583263158798218,
      "learning_rate": 1.629609713260307e-05,
      "loss": 1.8729,
      "step": 206119
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.4228614568710327,
      "learning_rate": 1.629476731759166e-05,
      "loss": 2.1489,
      "step": 206120
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.4830023050308228,
      "learning_rate": 1.6293437555326395e-05,
      "loss": 1.9066,
      "step": 206121
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.4692721366882324,
      "learning_rate": 1.6292107845807533e-05,
      "loss": 1.9907,
      "step": 206122
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.2652640342712402,
      "learning_rate": 1.629077818903528e-05,
      "loss": 1.9177,
      "step": 206123
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.4937113523483276,
      "learning_rate": 1.6289448585009968e-05,
      "loss": 1.9544,
      "step": 206124
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.4186649322509766,
      "learning_rate": 1.6288119033731762e-05,
      "loss": 1.9472,
      "step": 206125
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.6158192157745361,
      "learning_rate": 1.6286789535200994e-05,
      "loss": 2.0865,
      "step": 206126
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.427632212638855,
      "learning_rate": 1.628546008941787e-05,
      "loss": 1.9824,
      "step": 206127
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.4025394916534424,
      "learning_rate": 1.628413069638258e-05,
      "loss": 2.0121,
      "step": 206128
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.5340238809585571,
      "learning_rate": 1.6282801356095465e-05,
      "loss": 2.128,
      "step": 206129
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.509909987449646,
      "learning_rate": 1.6281472068556756e-05,
      "loss": 1.8254,
      "step": 206130
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.5036423206329346,
      "learning_rate": 1.6280142833766617e-05,
      "loss": 1.889,
      "step": 206131
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.4630653858184814,
      "learning_rate": 1.627881365172542e-05,
      "loss": 1.9681,
      "step": 206132
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.4652808904647827,
      "learning_rate": 1.6277484522433328e-05,
      "loss": 1.9464,
      "step": 206133
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.3430529832839966,
      "learning_rate": 1.627615544589057e-05,
      "loss": 2.0159,
      "step": 206134
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.5450483560562134,
      "learning_rate": 1.6274826422097487e-05,
      "loss": 1.9722,
      "step": 206135
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.4341727495193481,
      "learning_rate": 1.6273497451054207e-05,
      "loss": 1.9323,
      "step": 206136
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.4197989702224731,
      "learning_rate": 1.6272168532761066e-05,
      "loss": 2.0395,
      "step": 206137
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.5266956090927124,
      "learning_rate": 1.627083966721833e-05,
      "loss": 1.9527,
      "step": 206138
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.3297423124313354,
      "learning_rate": 1.626951085442616e-05,
      "loss": 1.8447,
      "step": 206139
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.5008395910263062,
      "learning_rate": 1.626818209438483e-05,
      "loss": 1.9206,
      "step": 206140
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.3767826557159424,
      "learning_rate": 1.6266853387094636e-05,
      "loss": 1.8851,
      "step": 206141
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.5069221258163452,
      "learning_rate": 1.6265524732555714e-05,
      "loss": 2.1166,
      "step": 206142
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.474029779434204,
      "learning_rate": 1.6264196130768458e-05,
      "loss": 1.7018,
      "step": 206143
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.4971742630004883,
      "learning_rate": 1.6262867581733042e-05,
      "loss": 2.0224,
      "step": 206144
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.4584629535675049,
      "learning_rate": 1.6261539085449727e-05,
      "loss": 1.8062,
      "step": 206145
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.4422401189804077,
      "learning_rate": 1.626021064191868e-05,
      "loss": 1.8407,
      "step": 206146
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.4745951890945435,
      "learning_rate": 1.6258882251140304e-05,
      "loss": 1.9736,
      "step": 206147
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.5902026891708374,
      "learning_rate": 1.6257553913114663e-05,
      "loss": 1.817,
      "step": 206148
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.4293622970581055,
      "learning_rate": 1.6256225627842156e-05,
      "loss": 1.8878,
      "step": 206149
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.344387173652649,
      "learning_rate": 1.6254897395322952e-05,
      "loss": 2.0992,
      "step": 206150
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.3505890369415283,
      "learning_rate": 1.625356921555738e-05,
      "loss": 1.645,
      "step": 206151
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.5336054563522339,
      "learning_rate": 1.6252241088545547e-05,
      "loss": 2.1251,
      "step": 206152
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.3768948316574097,
      "learning_rate": 1.6250913014287846e-05,
      "loss": 2.16,
      "step": 206153
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.4792628288269043,
      "learning_rate": 1.624958499278438e-05,
      "loss": 2.0579,
      "step": 206154
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.444055438041687,
      "learning_rate": 1.6248257024035516e-05,
      "loss": 1.9599,
      "step": 206155
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.4971535205841064,
      "learning_rate": 1.624692910804142e-05,
      "loss": 1.9883,
      "step": 206156
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.5824108123779297,
      "learning_rate": 1.6245601244802485e-05,
      "loss": 2.0493,
      "step": 206157
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.4800533056259155,
      "learning_rate": 1.624427343431872e-05,
      "loss": 2.0438,
      "step": 206158
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.4357869625091553,
      "learning_rate": 1.6242945676590592e-05,
      "loss": 2.0076,
      "step": 206159
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.444942593574524,
      "learning_rate": 1.6241617971618192e-05,
      "loss": 1.7999,
      "step": 206160
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.496522068977356,
      "learning_rate": 1.624029031940186e-05,
      "loss": 2.1377,
      "step": 206161
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.466487169265747,
      "learning_rate": 1.6238962719941794e-05,
      "loss": 2.0547,
      "step": 206162
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.4261789321899414,
      "learning_rate": 1.6237635173238327e-05,
      "loss": 1.8823,
      "step": 206163
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.4751230478286743,
      "learning_rate": 1.6236307679291562e-05,
      "loss": 2.1035,
      "step": 206164
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.408098578453064,
      "learning_rate": 1.6234980238101857e-05,
      "loss": 2.0119,
      "step": 206165
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.3416156768798828,
      "learning_rate": 1.623365284966942e-05,
      "loss": 1.9354,
      "step": 206166
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.5118167400360107,
      "learning_rate": 1.6232325513994516e-05,
      "loss": 2.2531,
      "step": 206167
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.4572960138320923,
      "learning_rate": 1.6230998231077343e-05,
      "loss": 2.1341,
      "step": 206168
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.382337212562561,
      "learning_rate": 1.6229671000918264e-05,
      "loss": 1.8551,
      "step": 206169
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.6000711917877197,
      "learning_rate": 1.6228343823517354e-05,
      "loss": 2.1937,
      "step": 206170
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.4914839267730713,
      "learning_rate": 1.6227016698875005e-05,
      "loss": 2.3028,
      "step": 206171
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.4374953508377075,
      "learning_rate": 1.622568962699139e-05,
      "loss": 2.0506,
      "step": 206172
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.5835130214691162,
      "learning_rate": 1.622436260786677e-05,
      "loss": 1.9853,
      "step": 206173
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.5285288095474243,
      "learning_rate": 1.622303564150138e-05,
      "loss": 2.177,
      "step": 206174
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.480218768119812,
      "learning_rate": 1.6221708727895587e-05,
      "loss": 1.8386,
      "step": 206175
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.4592114686965942,
      "learning_rate": 1.6220381867049426e-05,
      "loss": 1.9259,
      "step": 206176
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.503825306892395,
      "learning_rate": 1.6219055058963294e-05,
      "loss": 1.9802,
      "step": 206177
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.4438029527664185,
      "learning_rate": 1.621772830363739e-05,
      "loss": 1.9194,
      "step": 206178
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.4882365465164185,
      "learning_rate": 1.6216401601071983e-05,
      "loss": 1.9663,
      "step": 206179
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.7072808742523193,
      "learning_rate": 1.621507495126724e-05,
      "loss": 1.9625,
      "step": 206180
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.3172345161437988,
      "learning_rate": 1.6213748354223586e-05,
      "loss": 2.1033,
      "step": 206181
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.5760835409164429,
      "learning_rate": 1.6212421809941068e-05,
      "loss": 1.9993,
      "step": 206182
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.5538849830627441,
      "learning_rate": 1.6211095318420042e-05,
      "loss": 1.8625,
      "step": 206183
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.5139836072921753,
      "learning_rate": 1.6209768879660678e-05,
      "loss": 1.8406,
      "step": 206184
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.4186058044433594,
      "learning_rate": 1.620844249366334e-05,
      "loss": 2.0396,
      "step": 206185
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.292649507522583,
      "learning_rate": 1.620711616042817e-05,
      "loss": 1.7789,
      "step": 206186
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.3075335025787354,
      "learning_rate": 1.6205789879955488e-05,
      "loss": 1.7514,
      "step": 206187
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.364085078239441,
      "learning_rate": 1.6204463652245503e-05,
      "loss": 2.0242,
      "step": 206188
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.4875785112380981,
      "learning_rate": 1.6203137477298477e-05,
      "loss": 1.9595,
      "step": 206189
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.3513401746749878,
      "learning_rate": 1.6201811355114614e-05,
      "loss": 2.1148,
      "step": 206190
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.3047840595245361,
      "learning_rate": 1.620048528569421e-05,
      "loss": 1.7925,
      "step": 206191
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.4476453065872192,
      "learning_rate": 1.6199159269037464e-05,
      "loss": 2.1252,
      "step": 206192
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.3604989051818848,
      "learning_rate": 1.619783330514468e-05,
      "loss": 1.9261,
      "step": 206193
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.318933129310608,
      "learning_rate": 1.619650739401609e-05,
      "loss": 2.056,
      "step": 206194
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.4615365266799927,
      "learning_rate": 1.619518153565189e-05,
      "loss": 1.8331,
      "step": 206195
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.3316800594329834,
      "learning_rate": 1.6193855730052352e-05,
      "loss": 2.0571,
      "step": 206196
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.575002670288086,
      "learning_rate": 1.619252997721777e-05,
      "loss": 1.9803,
      "step": 206197
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.4739534854888916,
      "learning_rate": 1.619120427714832e-05,
      "loss": 1.9584,
      "step": 206198
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.3906102180480957,
      "learning_rate": 1.6189878629844323e-05,
      "loss": 2.0763,
      "step": 206199
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.4573601484298706,
      "learning_rate": 1.6188553035305984e-05,
      "loss": 1.9138,
      "step": 206200
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.649057149887085,
      "learning_rate": 1.618722749353354e-05,
      "loss": 2.1034,
      "step": 206201
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.3046603202819824,
      "learning_rate": 1.6185902004527217e-05,
      "loss": 1.8005,
      "step": 206202
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.3954764604568481,
      "learning_rate": 1.6184576568287322e-05,
      "loss": 1.8383,
      "step": 206203
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.4400430917739868,
      "learning_rate": 1.618325118481405e-05,
      "loss": 1.9424,
      "step": 206204
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.3441413640975952,
      "learning_rate": 1.6181925854107702e-05,
      "loss": 2.0629,
      "step": 206205
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.471709132194519,
      "learning_rate": 1.618060057616848e-05,
      "loss": 2.1599,
      "step": 206206
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.470585584640503,
      "learning_rate": 1.6179275350996644e-05,
      "loss": 1.7977,
      "step": 206207
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.5984936952590942,
      "learning_rate": 1.6177950178592435e-05,
      "loss": 2.0928,
      "step": 206208
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.778346300125122,
      "learning_rate": 1.6176625058956084e-05,
      "loss": 1.8482,
      "step": 206209
    },
    {
      "epoch": 2.68,
      "grad_norm": 1.5563448667526245,
      "learning_rate": 1.6175299992087852e-05,
      "loss": 2.1984,
      "step": 206210
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.660936713218689,
      "learning_rate": 1.6173974977988012e-05,
      "loss": 2.0291,
      "step": 206211
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.4169920682907104,
      "learning_rate": 1.6172650016656796e-05,
      "loss": 2.2321,
      "step": 206212
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.3695226907730103,
      "learning_rate": 1.61713251080944e-05,
      "loss": 1.9993,
      "step": 206213
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.4917367696762085,
      "learning_rate": 1.6170000252301163e-05,
      "loss": 1.961,
      "step": 206214
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.4019691944122314,
      "learning_rate": 1.6168675449277278e-05,
      "loss": 1.9197,
      "step": 206215
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.5777482986450195,
      "learning_rate": 1.6167350699022952e-05,
      "loss": 2.3726,
      "step": 206216
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.6747252941131592,
      "learning_rate": 1.6166026001538512e-05,
      "loss": 1.9124,
      "step": 206217
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.6489568948745728,
      "learning_rate": 1.6164701356824163e-05,
      "loss": 2.1062,
      "step": 206218
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.3380135297775269,
      "learning_rate": 1.6163376764880133e-05,
      "loss": 1.7801,
      "step": 206219
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.3403701782226562,
      "learning_rate": 1.616205222570669e-05,
      "loss": 2.0683,
      "step": 206220
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.4176490306854248,
      "learning_rate": 1.6160727739304102e-05,
      "loss": 1.9491,
      "step": 206221
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.353442907333374,
      "learning_rate": 1.6159403305672535e-05,
      "loss": 1.8229,
      "step": 206222
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.5573726892471313,
      "learning_rate": 1.6158078924812358e-05,
      "loss": 1.8265,
      "step": 206223
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.4711114168167114,
      "learning_rate": 1.6156754596723766e-05,
      "loss": 1.8992,
      "step": 206224
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.316201090812683,
      "learning_rate": 1.6155430321406926e-05,
      "loss": 2.0115,
      "step": 206225
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.4627420902252197,
      "learning_rate": 1.6154106098862174e-05,
      "loss": 1.8934,
      "step": 206226
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.388006329536438,
      "learning_rate": 1.6152781929089708e-05,
      "loss": 1.8448,
      "step": 206227
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.4361032247543335,
      "learning_rate": 1.615145781208983e-05,
      "loss": 1.9587,
      "step": 206228
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.3879748582839966,
      "learning_rate": 1.615013374786277e-05,
      "loss": 1.9181,
      "step": 206229
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.4195324182510376,
      "learning_rate": 1.6148809736408762e-05,
      "loss": 2.0947,
      "step": 206230
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.4049327373504639,
      "learning_rate": 1.6147485777727976e-05,
      "loss": 2.0501,
      "step": 206231
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.5646852254867554,
      "learning_rate": 1.6146161871820805e-05,
      "loss": 2.0789,
      "step": 206232
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.4685688018798828,
      "learning_rate": 1.6144838018687355e-05,
      "loss": 2.0465,
      "step": 206233
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.4904060363769531,
      "learning_rate": 1.614351421832799e-05,
      "loss": 1.9923,
      "step": 206234
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.4255698919296265,
      "learning_rate": 1.614219047074288e-05,
      "loss": 2.0187,
      "step": 206235
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.3875503540039062,
      "learning_rate": 1.6140866775932348e-05,
      "loss": 1.9119,
      "step": 206236
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.628682017326355,
      "learning_rate": 1.6139543133896537e-05,
      "loss": 2.0122,
      "step": 206237
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.3366293907165527,
      "learning_rate": 1.6138219544635744e-05,
      "loss": 2.0326,
      "step": 206238
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.4257659912109375,
      "learning_rate": 1.6136896008150202e-05,
      "loss": 2.0982,
      "step": 206239
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.3705546855926514,
      "learning_rate": 1.613557252444021e-05,
      "loss": 2.0593,
      "step": 206240
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.4851982593536377,
      "learning_rate": 1.6134249093505935e-05,
      "loss": 2.0819,
      "step": 206241
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.5028384923934937,
      "learning_rate": 1.613292571534771e-05,
      "loss": 2.127,
      "step": 206242
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.5590914487838745,
      "learning_rate": 1.6131602389965704e-05,
      "loss": 1.8454,
      "step": 206243
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.3871901035308838,
      "learning_rate": 1.613027911736018e-05,
      "loss": 1.9324,
      "step": 206244
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.4763695001602173,
      "learning_rate": 1.612895589753137e-05,
      "loss": 1.8197,
      "step": 206245
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.5430325269699097,
      "learning_rate": 1.6127632730479613e-05,
      "loss": 2.0284,
      "step": 206246
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.4739238023757935,
      "learning_rate": 1.6126309616205034e-05,
      "loss": 1.7797,
      "step": 206247
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.464516043663025,
      "learning_rate": 1.6124986554708007e-05,
      "loss": 1.7791,
      "step": 206248
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.6258702278137207,
      "learning_rate": 1.6123663545988596e-05,
      "loss": 2.007,
      "step": 206249
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.4434114694595337,
      "learning_rate": 1.6122340590047233e-05,
      "loss": 1.8956,
      "step": 206250
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.4735832214355469,
      "learning_rate": 1.6121017686884052e-05,
      "loss": 1.8955,
      "step": 206251
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.452126145362854,
      "learning_rate": 1.611969483649935e-05,
      "loss": 1.8889,
      "step": 206252
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.4203968048095703,
      "learning_rate": 1.61183720388933e-05,
      "loss": 2.1083,
      "step": 206253
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.5131045579910278,
      "learning_rate": 1.6117049294066264e-05,
      "loss": 2.2424,
      "step": 206254
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.3816015720367432,
      "learning_rate": 1.6115726602018442e-05,
      "loss": 1.8727,
      "step": 206255
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.4404114484786987,
      "learning_rate": 1.6114403962750032e-05,
      "loss": 2.0691,
      "step": 206256
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.455617070198059,
      "learning_rate": 1.6113081376261306e-05,
      "loss": 2.087,
      "step": 206257
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.4890376329421997,
      "learning_rate": 1.6111758842552525e-05,
      "loss": 1.9885,
      "step": 206258
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.3982672691345215,
      "learning_rate": 1.6110436361623924e-05,
      "loss": 1.9798,
      "step": 206259
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.3333454132080078,
      "learning_rate": 1.610911393347577e-05,
      "loss": 1.7901,
      "step": 206260
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.435020089149475,
      "learning_rate": 1.6107791558108295e-05,
      "loss": 1.8166,
      "step": 206261
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.606581687927246,
      "learning_rate": 1.6106469235521734e-05,
      "loss": 1.9,
      "step": 206262
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.5218459367752075,
      "learning_rate": 1.6105146965716285e-05,
      "loss": 2.0179,
      "step": 206263
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.5827642679214478,
      "learning_rate": 1.610382474869232e-05,
      "loss": 2.2245,
      "step": 206264
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.383347749710083,
      "learning_rate": 1.610250258444996e-05,
      "loss": 1.763,
      "step": 206265
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.4084886312484741,
      "learning_rate": 1.610118047298955e-05,
      "loss": 1.9468,
      "step": 206266
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.382155179977417,
      "learning_rate": 1.6099858414311285e-05,
      "loss": 1.8978,
      "step": 206267
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.3486095666885376,
      "learning_rate": 1.6098536408415398e-05,
      "loss": 1.8064,
      "step": 206268
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.4555299282073975,
      "learning_rate": 1.609721445530212e-05,
      "loss": 2.1699,
      "step": 206269
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.4766610860824585,
      "learning_rate": 1.609589255497179e-05,
      "loss": 1.6619,
      "step": 206270
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.502908706665039,
      "learning_rate": 1.6094570707424538e-05,
      "loss": 1.8221,
      "step": 206271
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.6055458784103394,
      "learning_rate": 1.6093248912660696e-05,
      "loss": 2.0447,
      "step": 206272
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.6024575233459473,
      "learning_rate": 1.6091927170680498e-05,
      "loss": 1.8351,
      "step": 206273
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.4676729440689087,
      "learning_rate": 1.609060548148414e-05,
      "loss": 2.0861,
      "step": 206274
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.4736440181732178,
      "learning_rate": 1.60892838450719e-05,
      "loss": 2.0051,
      "step": 206275
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.609560489654541,
      "learning_rate": 1.6087962261444033e-05,
      "loss": 2.065,
      "step": 206276
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.426339864730835,
      "learning_rate": 1.608664073060071e-05,
      "loss": 2.0363,
      "step": 206277
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.4598612785339355,
      "learning_rate": 1.6085319252542326e-05,
      "loss": 1.894,
      "step": 206278
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.4879018068313599,
      "learning_rate": 1.6083997827269025e-05,
      "loss": 1.9846,
      "step": 206279
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.5416630506515503,
      "learning_rate": 1.608267645478106e-05,
      "loss": 1.9587,
      "step": 206280
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.3946478366851807,
      "learning_rate": 1.6081355135078643e-05,
      "loss": 1.894,
      "step": 206281
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.3169819116592407,
      "learning_rate": 1.60800338681621e-05,
      "loss": 1.8824,
      "step": 206282
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.3422061204910278,
      "learning_rate": 1.6078712654031634e-05,
      "loss": 2.0559,
      "step": 206283
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.4343022108078003,
      "learning_rate": 1.6077391492687507e-05,
      "loss": 1.9475,
      "step": 206284
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.3480618000030518,
      "learning_rate": 1.6076070384129924e-05,
      "loss": 2.0575,
      "step": 206285
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.3748838901519775,
      "learning_rate": 1.6074749328359215e-05,
      "loss": 2.0652,
      "step": 206286
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.4052726030349731,
      "learning_rate": 1.607342832537548e-05,
      "loss": 1.9029,
      "step": 206287
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.5026606321334839,
      "learning_rate": 1.6072107375179123e-05,
      "loss": 1.8897,
      "step": 206288
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.453442931175232,
      "learning_rate": 1.607078647777027e-05,
      "loss": 1.8967,
      "step": 206289
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.5025105476379395,
      "learning_rate": 1.6069465633149258e-05,
      "loss": 2.023,
      "step": 206290
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.4112355709075928,
      "learning_rate": 1.6068144841316322e-05,
      "loss": 1.8555,
      "step": 206291
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.410141944885254,
      "learning_rate": 1.6066824102271624e-05,
      "loss": 2.0107,
      "step": 206292
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.4396871328353882,
      "learning_rate": 1.6065503416015467e-05,
      "loss": 2.0814,
      "step": 206293
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.4612858295440674,
      "learning_rate": 1.6064182782548117e-05,
      "loss": 1.8304,
      "step": 206294
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.3797613382339478,
      "learning_rate": 1.606286220186974e-05,
      "loss": 2.0173,
      "step": 206295
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.4724643230438232,
      "learning_rate": 1.6061541673980705e-05,
      "loss": 1.9752,
      "step": 206296
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.4828505516052246,
      "learning_rate": 1.6060221198881174e-05,
      "loss": 1.7913,
      "step": 206297
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.6555255651474,
      "learning_rate": 1.605890077657135e-05,
      "loss": 2.3596,
      "step": 206298
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.4796926975250244,
      "learning_rate": 1.6057580407051595e-05,
      "loss": 2.0146,
      "step": 206299
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.378944754600525,
      "learning_rate": 1.6056260090322116e-05,
      "loss": 2.1276,
      "step": 206300
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.4707815647125244,
      "learning_rate": 1.6054939826383075e-05,
      "loss": 1.8394,
      "step": 206301
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.5529186725616455,
      "learning_rate": 1.6053619615234802e-05,
      "loss": 1.9382,
      "step": 206302
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.6807258129119873,
      "learning_rate": 1.6052299456877536e-05,
      "loss": 2.0815,
      "step": 206303
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.5948892831802368,
      "learning_rate": 1.6050979351311477e-05,
      "loss": 1.9112,
      "step": 206304
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.5770875215530396,
      "learning_rate": 1.6049659298536953e-05,
      "loss": 2.0805,
      "step": 206305
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.4786702394485474,
      "learning_rate": 1.6048339298554137e-05,
      "loss": 1.7383,
      "step": 206306
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.5325891971588135,
      "learning_rate": 1.6047019351363255e-05,
      "loss": 1.8689,
      "step": 206307
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.4818902015686035,
      "learning_rate": 1.6045699456964644e-05,
      "loss": 2.0165,
      "step": 206308
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.3957921266555786,
      "learning_rate": 1.604437961535847e-05,
      "loss": 1.964,
      "step": 206309
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.4375559091567993,
      "learning_rate": 1.6043059826545003e-05,
      "loss": 2.0297,
      "step": 206310
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.4100031852722168,
      "learning_rate": 1.6041740090524504e-05,
      "loss": 2.0713,
      "step": 206311
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.3973267078399658,
      "learning_rate": 1.6040420407297173e-05,
      "loss": 1.9297,
      "step": 206312
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.4480586051940918,
      "learning_rate": 1.6039100776863346e-05,
      "loss": 2.2031,
      "step": 206313
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.3835729360580444,
      "learning_rate": 1.6037781199223188e-05,
      "loss": 2.1572,
      "step": 206314
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.4146316051483154,
      "learning_rate": 1.6036461674376965e-05,
      "loss": 1.9321,
      "step": 206315
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.4038567543029785,
      "learning_rate": 1.6035142202324912e-05,
      "loss": 2.0496,
      "step": 206316
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.4278459548950195,
      "learning_rate": 1.6033822783067295e-05,
      "loss": 2.1926,
      "step": 206317
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.449604868888855,
      "learning_rate": 1.6032503416604313e-05,
      "loss": 1.8718,
      "step": 206318
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.4731262922286987,
      "learning_rate": 1.6031184102936302e-05,
      "loss": 2.0715,
      "step": 206319
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.4199529886245728,
      "learning_rate": 1.6029864842063454e-05,
      "loss": 1.8666,
      "step": 206320
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.5266104936599731,
      "learning_rate": 1.6028545633985977e-05,
      "loss": 1.9948,
      "step": 206321
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.661224365234375,
      "learning_rate": 1.602722647870417e-05,
      "loss": 1.7807,
      "step": 206322
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.4956902265548706,
      "learning_rate": 1.6025907376218294e-05,
      "loss": 1.9294,
      "step": 206323
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.4867684841156006,
      "learning_rate": 1.602458832652852e-05,
      "loss": 1.8872,
      "step": 206324
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.35878324508667,
      "learning_rate": 1.6023269329635148e-05,
      "loss": 1.7984,
      "step": 206325
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.4053876399993896,
      "learning_rate": 1.6021950385538372e-05,
      "loss": 2.1564,
      "step": 206326
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.5693758726119995,
      "learning_rate": 1.6020631494238533e-05,
      "loss": 2.1642,
      "step": 206327
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.3389190435409546,
      "learning_rate": 1.601931265573583e-05,
      "loss": 1.9362,
      "step": 206328
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.4015191793441772,
      "learning_rate": 1.6017993870030487e-05,
      "loss": 2.1004,
      "step": 206329
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.593466877937317,
      "learning_rate": 1.6016675137122715e-05,
      "loss": 1.8245,
      "step": 206330
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.4600995779037476,
      "learning_rate": 1.601535645701284e-05,
      "loss": 2.0069,
      "step": 206331
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.5932543277740479,
      "learning_rate": 1.601403782970103e-05,
      "loss": 1.9086,
      "step": 206332
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.4652371406555176,
      "learning_rate": 1.6012719255187624e-05,
      "loss": 1.7349,
      "step": 206333
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.4465367794036865,
      "learning_rate": 1.6011400733472813e-05,
      "loss": 1.9556,
      "step": 206334
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.4225084781646729,
      "learning_rate": 1.6010082264556835e-05,
      "loss": 2.0869,
      "step": 206335
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.4814181327819824,
      "learning_rate": 1.6008763848439886e-05,
      "loss": 2.0394,
      "step": 206336
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.4089733362197876,
      "learning_rate": 1.600744548512234e-05,
      "loss": 1.9333,
      "step": 206337
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.4124329090118408,
      "learning_rate": 1.6006127174604322e-05,
      "loss": 2.1527,
      "step": 206338
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.41191828250885,
      "learning_rate": 1.600480891688617e-05,
      "loss": 2.0458,
      "step": 206339
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.4165841341018677,
      "learning_rate": 1.600349071196808e-05,
      "loss": 1.9514,
      "step": 206340
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.4846818447113037,
      "learning_rate": 1.600217255985029e-05,
      "loss": 1.9573,
      "step": 206341
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.5693669319152832,
      "learning_rate": 1.6000854460533032e-05,
      "loss": 2.0346,
      "step": 206342
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.4004806280136108,
      "learning_rate": 1.59995364140166e-05,
      "loss": 1.8254,
      "step": 206343
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.4101699590682983,
      "learning_rate": 1.5998218420301202e-05,
      "loss": 2.0457,
      "step": 206344
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.7124663591384888,
      "learning_rate": 1.5996900479387098e-05,
      "loss": 1.9978,
      "step": 206345
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.3876465559005737,
      "learning_rate": 1.5995582591274558e-05,
      "loss": 2.0339,
      "step": 206346
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.4897936582565308,
      "learning_rate": 1.5994264755963816e-05,
      "loss": 1.9748,
      "step": 206347
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.4482266902923584,
      "learning_rate": 1.5992946973455034e-05,
      "loss": 1.8883,
      "step": 206348
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.415305733680725,
      "learning_rate": 1.599162924374855e-05,
      "loss": 2.0705,
      "step": 206349
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.5057989358901978,
      "learning_rate": 1.599031156684456e-05,
      "loss": 2.09,
      "step": 206350
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.513244390487671,
      "learning_rate": 1.59889939427434e-05,
      "loss": 1.823,
      "step": 206351
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.4884140491485596,
      "learning_rate": 1.59876763714452e-05,
      "loss": 1.8984,
      "step": 206352
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.458187222480774,
      "learning_rate": 1.5986358852950297e-05,
      "loss": 1.8613,
      "step": 206353
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.408488392829895,
      "learning_rate": 1.5985041387258824e-05,
      "loss": 2.0513,
      "step": 206354
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.5260639190673828,
      "learning_rate": 1.5983723974371144e-05,
      "loss": 1.8241,
      "step": 206355
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.5139789581298828,
      "learning_rate": 1.598240661428739e-05,
      "loss": 1.9267,
      "step": 206356
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.4642595052719116,
      "learning_rate": 1.5981089307007934e-05,
      "loss": 2.013,
      "step": 206357
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.6667454242706299,
      "learning_rate": 1.5979772052532935e-05,
      "loss": 2.1115,
      "step": 206358
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.371193766593933,
      "learning_rate": 1.5978454850862665e-05,
      "loss": 2.0074,
      "step": 206359
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.5023921728134155,
      "learning_rate": 1.5977137701997323e-05,
      "loss": 2.0692,
      "step": 206360
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.4681353569030762,
      "learning_rate": 1.5975820605937238e-05,
      "loss": 2.075,
      "step": 206361
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.5088751316070557,
      "learning_rate": 1.597450356268255e-05,
      "loss": 1.9142,
      "step": 206362
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.499997615814209,
      "learning_rate": 1.597318657223362e-05,
      "loss": 2.0276,
      "step": 206363
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.4869412183761597,
      "learning_rate": 1.597186963459065e-05,
      "loss": 2.0713,
      "step": 206364
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.5410393476486206,
      "learning_rate": 1.597055274975384e-05,
      "loss": 2.1238,
      "step": 206365
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.5194591283798218,
      "learning_rate": 1.5969235917723456e-05,
      "loss": 2.1341,
      "step": 206366
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.4693199396133423,
      "learning_rate": 1.5967919138499762e-05,
      "loss": 1.9828,
      "step": 206367
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.3796777725219727,
      "learning_rate": 1.596660241208296e-05,
      "loss": 1.8633,
      "step": 206368
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.4935177564620972,
      "learning_rate": 1.5965285738473386e-05,
      "loss": 1.9834,
      "step": 206369
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.4423456192016602,
      "learning_rate": 1.5963969117671206e-05,
      "loss": 1.9439,
      "step": 206370
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.478297472000122,
      "learning_rate": 1.5962652549676713e-05,
      "loss": 1.959,
      "step": 206371
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.5597375631332397,
      "learning_rate": 1.5961336034490048e-05,
      "loss": 2.0111,
      "step": 206372
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.405686616897583,
      "learning_rate": 1.5960019572111604e-05,
      "loss": 1.9884,
      "step": 206373
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.4378318786621094,
      "learning_rate": 1.5958703162541487e-05,
      "loss": 1.892,
      "step": 206374
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.6244832277297974,
      "learning_rate": 1.5957386805780094e-05,
      "loss": 2.0637,
      "step": 206375
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.582682490348816,
      "learning_rate": 1.5956070501827555e-05,
      "loss": 2.0642,
      "step": 206376
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.3700168132781982,
      "learning_rate": 1.5954754250684143e-05,
      "loss": 1.8279,
      "step": 206377
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.389641523361206,
      "learning_rate": 1.5953438052350054e-05,
      "loss": 1.8871,
      "step": 206378
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.5082403421401978,
      "learning_rate": 1.595212190682562e-05,
      "loss": 2.0254,
      "step": 206379
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.3383063077926636,
      "learning_rate": 1.5950805814111013e-05,
      "loss": 1.9097,
      "step": 206380
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.6142842769622803,
      "learning_rate": 1.5949489774206593e-05,
      "loss": 1.9341,
      "step": 206381
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.4342188835144043,
      "learning_rate": 1.5948173787112427e-05,
      "loss": 1.9162,
      "step": 206382
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.4979325532913208,
      "learning_rate": 1.5946857852828987e-05,
      "loss": 2.0551,
      "step": 206383
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.5032800436019897,
      "learning_rate": 1.594554197135627e-05,
      "loss": 1.879,
      "step": 206384
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.527898907661438,
      "learning_rate": 1.5944226142694704e-05,
      "loss": 1.8512,
      "step": 206385
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.4503734111785889,
      "learning_rate": 1.5942910366844396e-05,
      "loss": 2.1317,
      "step": 206386
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.5553044080734253,
      "learning_rate": 1.5941594643805745e-05,
      "loss": 2.1773,
      "step": 206387
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.4496480226516724,
      "learning_rate": 1.5940278973578847e-05,
      "loss": 2.0671,
      "step": 206388
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.4010587930679321,
      "learning_rate": 1.593896335616407e-05,
      "loss": 1.9216,
      "step": 206389
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.9000471830368042,
      "learning_rate": 1.593764779156158e-05,
      "loss": 1.8799,
      "step": 206390
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.4937584400177002,
      "learning_rate": 1.5936332279771648e-05,
      "loss": 2.0805,
      "step": 206391
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.501080870628357,
      "learning_rate": 1.59350168207945e-05,
      "loss": 2.1258,
      "step": 206392
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.4444843530654907,
      "learning_rate": 1.5933701414630406e-05,
      "loss": 1.9505,
      "step": 206393
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.540509819984436,
      "learning_rate": 1.5932386061279566e-05,
      "loss": 1.9657,
      "step": 206394
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.5597972869873047,
      "learning_rate": 1.593107076074228e-05,
      "loss": 1.6906,
      "step": 206395
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.4883580207824707,
      "learning_rate": 1.5929755513018815e-05,
      "loss": 2.0674,
      "step": 206396
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.4330979585647583,
      "learning_rate": 1.5928440318109303e-05,
      "loss": 2.1272,
      "step": 206397
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.4182353019714355,
      "learning_rate": 1.5927125176014077e-05,
      "loss": 1.9791,
      "step": 206398
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.4110561609268188,
      "learning_rate": 1.5925810086733402e-05,
      "loss": 1.9553,
      "step": 206399
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.401334524154663,
      "learning_rate": 1.5924495050267417e-05,
      "loss": 1.9776,
      "step": 206400
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.3334124088287354,
      "learning_rate": 1.592318006661648e-05,
      "loss": 1.8517,
      "step": 206401
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.5379446744918823,
      "learning_rate": 1.5921865135780766e-05,
      "loss": 2.0154,
      "step": 206402
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.391188144683838,
      "learning_rate": 1.5920550257760533e-05,
      "loss": 1.957,
      "step": 206403
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.4106292724609375,
      "learning_rate": 1.5919235432556054e-05,
      "loss": 2.16,
      "step": 206404
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.4379796981811523,
      "learning_rate": 1.591792066016756e-05,
      "loss": 1.8729,
      "step": 206405
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.4562996625900269,
      "learning_rate": 1.5916605940595217e-05,
      "loss": 1.9774,
      "step": 206406
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.4000049829483032,
      "learning_rate": 1.5915291273839425e-05,
      "loss": 1.7421,
      "step": 206407
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.444251298904419,
      "learning_rate": 1.5913976659900318e-05,
      "loss": 1.8936,
      "step": 206408
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.5687828063964844,
      "learning_rate": 1.5912662098778096e-05,
      "loss": 2.0801,
      "step": 206409
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.4109551906585693,
      "learning_rate": 1.5911347590473156e-05,
      "loss": 1.9965,
      "step": 206410
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.4463281631469727,
      "learning_rate": 1.5910033134985602e-05,
      "loss": 2.0329,
      "step": 206411
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.394777774810791,
      "learning_rate": 1.5908718732315794e-05,
      "loss": 1.7953,
      "step": 206412
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.4883496761322021,
      "learning_rate": 1.5907404382463908e-05,
      "loss": 2.038,
      "step": 206413
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.4920471906661987,
      "learning_rate": 1.5906090085430167e-05,
      "loss": 1.9569,
      "step": 206414
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.5764530897140503,
      "learning_rate": 1.5904775841214846e-05,
      "loss": 1.7902,
      "step": 206415
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.4129518270492554,
      "learning_rate": 1.5903461649818206e-05,
      "loss": 1.9361,
      "step": 206416
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.3726458549499512,
      "learning_rate": 1.5902147511240447e-05,
      "loss": 2.1259,
      "step": 206417
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.499128818511963,
      "learning_rate": 1.5900833425481907e-05,
      "loss": 1.9691,
      "step": 206418
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.388643503189087,
      "learning_rate": 1.5899519392542716e-05,
      "loss": 1.9613,
      "step": 206419
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.456066608428955,
      "learning_rate": 1.5898205412423205e-05,
      "loss": 1.9646,
      "step": 206420
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.4627060890197754,
      "learning_rate": 1.589689148512351e-05,
      "loss": 1.8123,
      "step": 206421
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.558947205543518,
      "learning_rate": 1.5895577610644027e-05,
      "loss": 2.2269,
      "step": 206422
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.4771641492843628,
      "learning_rate": 1.5894263788984827e-05,
      "loss": 2.1453,
      "step": 206423
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.3895941972732544,
      "learning_rate": 1.589295002014631e-05,
      "loss": 1.9688,
      "step": 206424
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.5582611560821533,
      "learning_rate": 1.5891636304128674e-05,
      "loss": 1.8792,
      "step": 206425
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.5268386602401733,
      "learning_rate": 1.5890322640932118e-05,
      "loss": 1.9839,
      "step": 206426
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.610863208770752,
      "learning_rate": 1.588900903055691e-05,
      "loss": 1.9927,
      "step": 206427
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.4101201295852661,
      "learning_rate": 1.5887695473003313e-05,
      "loss": 2.0007,
      "step": 206428
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.392232894897461,
      "learning_rate": 1.58863819682715e-05,
      "loss": 1.8652,
      "step": 206429
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.4758350849151611,
      "learning_rate": 1.5885068516361832e-05,
      "loss": 2.2196,
      "step": 206430
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.5482245683670044,
      "learning_rate": 1.588375511727451e-05,
      "loss": 2.0029,
      "step": 206431
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.4347844123840332,
      "learning_rate": 1.5882441771009736e-05,
      "loss": 1.8884,
      "step": 206432
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.5063929557800293,
      "learning_rate": 1.588112847756774e-05,
      "loss": 1.9493,
      "step": 206433
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.4870221614837646,
      "learning_rate": 1.5879815236948855e-05,
      "loss": 1.9972,
      "step": 206434
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.5781537294387817,
      "learning_rate": 1.587850204915322e-05,
      "loss": 2.3141,
      "step": 206435
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.4349154233932495,
      "learning_rate": 1.5877188914181194e-05,
      "loss": 2.2136,
      "step": 206436
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.5472793579101562,
      "learning_rate": 1.5875875832032946e-05,
      "loss": 1.9508,
      "step": 206437
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.4493530988693237,
      "learning_rate": 1.5874562802708745e-05,
      "loss": 2.1539,
      "step": 206438
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.3764533996582031,
      "learning_rate": 1.5873249826208788e-05,
      "loss": 1.9014,
      "step": 206439
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.5198875665664673,
      "learning_rate": 1.5871936902533376e-05,
      "loss": 1.9041,
      "step": 206440
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.4331787824630737,
      "learning_rate": 1.587062403168271e-05,
      "loss": 1.7795,
      "step": 206441
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.493823766708374,
      "learning_rate": 1.586931121365712e-05,
      "loss": 1.8115,
      "step": 206442
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.5123686790466309,
      "learning_rate": 1.5867998448456775e-05,
      "loss": 2.0699,
      "step": 206443
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.3601350784301758,
      "learning_rate": 1.5866685736081908e-05,
      "loss": 1.7367,
      "step": 206444
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.5552339553833008,
      "learning_rate": 1.5865373076532783e-05,
      "loss": 1.9687,
      "step": 206445
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.521729588508606,
      "learning_rate": 1.586406046980967e-05,
      "loss": 1.6947,
      "step": 206446
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.4940446615219116,
      "learning_rate": 1.5862747915912733e-05,
      "loss": 1.9612,
      "step": 206447
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.5330299139022827,
      "learning_rate": 1.5861435414842337e-05,
      "loss": 2.0349,
      "step": 206448
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.4136571884155273,
      "learning_rate": 1.586012296659862e-05,
      "loss": 2.1783,
      "step": 206449
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.5842530727386475,
      "learning_rate": 1.585881057118198e-05,
      "loss": 1.9712,
      "step": 206450
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.5110950469970703,
      "learning_rate": 1.5857498228592413e-05,
      "loss": 1.8146,
      "step": 206451
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.51166570186615,
      "learning_rate": 1.585618593883039e-05,
      "loss": 2.1938,
      "step": 206452
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.449172019958496,
      "learning_rate": 1.585487370189601e-05,
      "loss": 2.0265,
      "step": 206453
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.4198981523513794,
      "learning_rate": 1.5853561517789604e-05,
      "loss": 2.035,
      "step": 206454
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.430340051651001,
      "learning_rate": 1.585224938651134e-05,
      "loss": 1.9119,
      "step": 206455
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.3669389486312866,
      "learning_rate": 1.5850937308061586e-05,
      "loss": 2.0638,
      "step": 206456
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.4145112037658691,
      "learning_rate": 1.5849625282440436e-05,
      "loss": 1.9013,
      "step": 206457
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.674998164176941,
      "learning_rate": 1.5848313309648263e-05,
      "loss": 1.8614,
      "step": 206458
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.4229007959365845,
      "learning_rate": 1.5847001389685167e-05,
      "loss": 1.9899,
      "step": 206459
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.9092607498168945,
      "learning_rate": 1.5845689522551542e-05,
      "loss": 2.0369,
      "step": 206460
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.539670705795288,
      "learning_rate": 1.584437770824756e-05,
      "loss": 1.8917,
      "step": 206461
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.4925727844238281,
      "learning_rate": 1.5843065946773515e-05,
      "loss": 1.9532,
      "step": 206462
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.54832124710083,
      "learning_rate": 1.5841754238129512e-05,
      "loss": 1.9889,
      "step": 206463
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.3929343223571777,
      "learning_rate": 1.5840442582315947e-05,
      "loss": 2.0813,
      "step": 206464
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.3224475383758545,
      "learning_rate": 1.5839130979332993e-05,
      "loss": 1.9487,
      "step": 206465
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.6178789138793945,
      "learning_rate": 1.583781942918091e-05,
      "loss": 1.9749,
      "step": 206466
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.6463685035705566,
      "learning_rate": 1.583650793185993e-05,
      "loss": 1.9136,
      "step": 206467
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.6240533590316772,
      "learning_rate": 1.5835196487370394e-05,
      "loss": 2.1421,
      "step": 206468
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.407016634941101,
      "learning_rate": 1.583388509571236e-05,
      "loss": 2.0741,
      "step": 206469
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.3903894424438477,
      "learning_rate": 1.5832573756886203e-05,
      "loss": 1.8985,
      "step": 206470
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.6631686687469482,
      "learning_rate": 1.583126247089208e-05,
      "loss": 2.2674,
      "step": 206471
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.4237849712371826,
      "learning_rate": 1.5829951237730366e-05,
      "loss": 1.9097,
      "step": 206472
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.2868409156799316,
      "learning_rate": 1.582864005740119e-05,
      "loss": 2.105,
      "step": 206473
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.402677297592163,
      "learning_rate": 1.582732892990485e-05,
      "loss": 2.1868,
      "step": 206474
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.5049691200256348,
      "learning_rate": 1.5826017855241578e-05,
      "loss": 2.0805,
      "step": 206475
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.5183032751083374,
      "learning_rate": 1.5824706833411615e-05,
      "loss": 2.1214,
      "step": 206476
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.3380532264709473,
      "learning_rate": 1.5823395864415156e-05,
      "loss": 1.8258,
      "step": 206477
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.422538161277771,
      "learning_rate": 1.5822084948252566e-05,
      "loss": 2.0159,
      "step": 206478
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.445051670074463,
      "learning_rate": 1.582077408492395e-05,
      "loss": 1.9656,
      "step": 206479
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.3758682012557983,
      "learning_rate": 1.5819463274429634e-05,
      "loss": 1.9577,
      "step": 206480
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.612879991531372,
      "learning_rate": 1.5818152516769887e-05,
      "loss": 1.9603,
      "step": 206481
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.5094255208969116,
      "learning_rate": 1.581684181194488e-05,
      "loss": 2.2017,
      "step": 206482
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.430774211883545,
      "learning_rate": 1.5815531159954873e-05,
      "loss": 1.9227,
      "step": 206483
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.379468560218811,
      "learning_rate": 1.5814220560800138e-05,
      "loss": 2.0859,
      "step": 206484
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.4324511289596558,
      "learning_rate": 1.581291001448087e-05,
      "loss": 1.9602,
      "step": 206485
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.384219765663147,
      "learning_rate": 1.5811599520997376e-05,
      "loss": 1.827,
      "step": 206486
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.4566357135772705,
      "learning_rate": 1.5810289080349915e-05,
      "loss": 1.8834,
      "step": 206487
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.4195712804794312,
      "learning_rate": 1.5808978692538587e-05,
      "loss": 1.9034,
      "step": 206488
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.4436994791030884,
      "learning_rate": 1.5807668357563795e-05,
      "loss": 1.9349,
      "step": 206489
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.500725507736206,
      "learning_rate": 1.580635807542574e-05,
      "loss": 1.7872,
      "step": 206490
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.44072425365448,
      "learning_rate": 1.5805047846124586e-05,
      "loss": 1.9808,
      "step": 206491
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.4260847568511963,
      "learning_rate": 1.5803737669660698e-05,
      "loss": 1.8289,
      "step": 206492
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.5686256885528564,
      "learning_rate": 1.5802427546034247e-05,
      "loss": 1.9208,
      "step": 206493
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.5291156768798828,
      "learning_rate": 1.580111747524546e-05,
      "loss": 1.9511,
      "step": 206494
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.5662966966629028,
      "learning_rate": 1.5799807457294643e-05,
      "loss": 2.0407,
      "step": 206495
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.538759469985962,
      "learning_rate": 1.579849749218196e-05,
      "loss": 2.3805,
      "step": 206496
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.5754001140594482,
      "learning_rate": 1.5797187579907744e-05,
      "loss": 2.0507,
      "step": 206497
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.4078850746154785,
      "learning_rate": 1.5795877720472194e-05,
      "loss": 2.1252,
      "step": 206498
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.4497108459472656,
      "learning_rate": 1.579456791387558e-05,
      "loss": 1.9931,
      "step": 206499
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.4537724256515503,
      "learning_rate": 1.5793258160118062e-05,
      "loss": 1.9745,
      "step": 206500
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.6516238451004028,
      "learning_rate": 1.5791948459199978e-05,
      "loss": 1.8074,
      "step": 206501
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.4616572856903076,
      "learning_rate": 1.5790638811121493e-05,
      "loss": 2.0208,
      "step": 206502
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.3970681428909302,
      "learning_rate": 1.578932921588294e-05,
      "loss": 2.0576,
      "step": 206503
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.4150471687316895,
      "learning_rate": 1.578801967348452e-05,
      "loss": 1.8389,
      "step": 206504
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.5958075523376465,
      "learning_rate": 1.5786710183926465e-05,
      "loss": 2.0453,
      "step": 206505
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.569004774093628,
      "learning_rate": 1.5785400747208976e-05,
      "loss": 2.0471,
      "step": 206506
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.38050377368927,
      "learning_rate": 1.5784091363332386e-05,
      "loss": 1.9664,
      "step": 206507
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.4586046934127808,
      "learning_rate": 1.5782782032296893e-05,
      "loss": 2.1662,
      "step": 206508
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.4509081840515137,
      "learning_rate": 1.578147275410273e-05,
      "loss": 1.8739,
      "step": 206509
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.6694693565368652,
      "learning_rate": 1.57801635287502e-05,
      "loss": 2.02,
      "step": 206510
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.3407742977142334,
      "learning_rate": 1.57788543562395e-05,
      "loss": 2.0351,
      "step": 206511
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.425727128982544,
      "learning_rate": 1.5777545236570833e-05,
      "loss": 2.0323,
      "step": 206512
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.4619959592819214,
      "learning_rate": 1.5776236169744493e-05,
      "loss": 2.2183,
      "step": 206513
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.299617886543274,
      "learning_rate": 1.577492715576072e-05,
      "loss": 2.0237,
      "step": 206514
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.4445911645889282,
      "learning_rate": 1.5773618194619775e-05,
      "loss": 1.9666,
      "step": 206515
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.5443062782287598,
      "learning_rate": 1.5772309286321828e-05,
      "loss": 1.7882,
      "step": 206516
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.4275271892547607,
      "learning_rate": 1.5771000430867275e-05,
      "loss": 1.9263,
      "step": 206517
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.5204718112945557,
      "learning_rate": 1.5769691628256153e-05,
      "loss": 1.8707,
      "step": 206518
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.5069183111190796,
      "learning_rate": 1.5768382878488895e-05,
      "loss": 1.8876,
      "step": 206519
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.5464305877685547,
      "learning_rate": 1.5767074181565564e-05,
      "loss": 2.0194,
      "step": 206520
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.5362542867660522,
      "learning_rate": 1.5765765537486564e-05,
      "loss": 1.8028,
      "step": 206521
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.5410934686660767,
      "learning_rate": 1.5764456946252057e-05,
      "loss": 2.0534,
      "step": 206522
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.4511659145355225,
      "learning_rate": 1.5763148407862345e-05,
      "loss": 2.082,
      "step": 206523
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.4608484506607056,
      "learning_rate": 1.5761839922317566e-05,
      "loss": 1.8356,
      "step": 206524
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.550212025642395,
      "learning_rate": 1.576053148961808e-05,
      "loss": 2.1151,
      "step": 206525
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.6247104406356812,
      "learning_rate": 1.5759223109764017e-05,
      "loss": 2.1027,
      "step": 206526
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.6590019464492798,
      "learning_rate": 1.5757914782755722e-05,
      "loss": 2.0731,
      "step": 206527
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.4799487590789795,
      "learning_rate": 1.575660650859335e-05,
      "loss": 1.92,
      "step": 206528
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.3571001291275024,
      "learning_rate": 1.575529828727731e-05,
      "loss": 1.9585,
      "step": 206529
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.364194631576538,
      "learning_rate": 1.575399011880759e-05,
      "loss": 1.7877,
      "step": 206530
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.5179262161254883,
      "learning_rate": 1.5752682003184637e-05,
      "loss": 2.1078,
      "step": 206531
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.5753649473190308,
      "learning_rate": 1.5751373940408574e-05,
      "loss": 1.9427,
      "step": 206532
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.3768033981323242,
      "learning_rate": 1.575006593047977e-05,
      "loss": 2.0044,
      "step": 206533
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.429504156112671,
      "learning_rate": 1.574875797339833e-05,
      "loss": 1.8623,
      "step": 206534
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.5799309015274048,
      "learning_rate": 1.5747450069164646e-05,
      "loss": 1.9908,
      "step": 206535
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.416487216949463,
      "learning_rate": 1.574614221777879e-05,
      "loss": 1.776,
      "step": 206536
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.6953704357147217,
      "learning_rate": 1.5744834419241127e-05,
      "loss": 1.915,
      "step": 206537
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.4072926044464111,
      "learning_rate": 1.574352667355182e-05,
      "loss": 1.9052,
      "step": 206538
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.4106096029281616,
      "learning_rate": 1.5742218980711208e-05,
      "loss": 1.9575,
      "step": 206539
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.43927800655365,
      "learning_rate": 1.5740911340719454e-05,
      "loss": 1.9406,
      "step": 206540
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.4040194749832153,
      "learning_rate": 1.573960375357689e-05,
      "loss": 1.7782,
      "step": 206541
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.4910156726837158,
      "learning_rate": 1.5738296219283654e-05,
      "loss": 2.1346,
      "step": 206542
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.4791204929351807,
      "learning_rate": 1.573698873784004e-05,
      "loss": 1.9994,
      "step": 206543
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.5103943347930908,
      "learning_rate": 1.573568130924625e-05,
      "loss": 2.0234,
      "step": 206544
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.4277492761611938,
      "learning_rate": 1.573437393350262e-05,
      "loss": 2.116,
      "step": 206545
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.5161921977996826,
      "learning_rate": 1.573306661060928e-05,
      "loss": 1.9912,
      "step": 206546
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.7437041997909546,
      "learning_rate": 1.573175934056663e-05,
      "loss": 2.0745,
      "step": 206547
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.6752219200134277,
      "learning_rate": 1.5730452123374702e-05,
      "loss": 2.0863,
      "step": 206548
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.526093602180481,
      "learning_rate": 1.57291449590339e-05,
      "loss": 1.9535,
      "step": 206549
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.445359706878662,
      "learning_rate": 1.572783784754439e-05,
      "loss": 2.335,
      "step": 206550
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.3693372011184692,
      "learning_rate": 1.5726530788906465e-05,
      "loss": 1.8183,
      "step": 206551
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.439843773841858,
      "learning_rate": 1.57252237831203e-05,
      "loss": 1.8516,
      "step": 206552
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.4359867572784424,
      "learning_rate": 1.572391683018629e-05,
      "loss": 2.1022,
      "step": 206553
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.6142653226852417,
      "learning_rate": 1.572260993010447e-05,
      "loss": 2.1775,
      "step": 206554
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.5483262538909912,
      "learning_rate": 1.5721303082875204e-05,
      "loss": 2.0661,
      "step": 206555
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.374009609222412,
      "learning_rate": 1.5719996288498694e-05,
      "loss": 1.9905,
      "step": 206556
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.639181137084961,
      "learning_rate": 1.5718689546975238e-05,
      "loss": 2.0899,
      "step": 206557
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.66010582447052,
      "learning_rate": 1.5717382858305005e-05,
      "loss": 1.8784,
      "step": 206558
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.4130158424377441,
      "learning_rate": 1.5716076222488326e-05,
      "loss": 1.8329,
      "step": 206559
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.4899600744247437,
      "learning_rate": 1.5714769639525406e-05,
      "loss": 1.9604,
      "step": 206560
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.2935762405395508,
      "learning_rate": 1.5713463109416436e-05,
      "loss": 1.8705,
      "step": 206561
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.5490697622299194,
      "learning_rate": 1.5712156632161686e-05,
      "loss": 2.1077,
      "step": 206562
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.457130789756775,
      "learning_rate": 1.571085020776146e-05,
      "loss": 2.0597,
      "step": 206563
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.61592435836792,
      "learning_rate": 1.5709543836215886e-05,
      "loss": 2.0207,
      "step": 206564
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.3993021249771118,
      "learning_rate": 1.5708237517525334e-05,
      "loss": 2.0708,
      "step": 206565
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.5256701707839966,
      "learning_rate": 1.570693125169e-05,
      "loss": 2.1959,
      "step": 206566
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.4465781450271606,
      "learning_rate": 1.570562503871009e-05,
      "loss": 1.9389,
      "step": 206567
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.3677791357040405,
      "learning_rate": 1.570431887858583e-05,
      "loss": 1.9905,
      "step": 206568
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.5367640256881714,
      "learning_rate": 1.5703012771317558e-05,
      "loss": 1.9631,
      "step": 206569
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.5581355094909668,
      "learning_rate": 1.570170671690544e-05,
      "loss": 1.7043,
      "step": 206570
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.3840430974960327,
      "learning_rate": 1.5700400715349735e-05,
      "loss": 1.783,
      "step": 206571
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.327365756034851,
      "learning_rate": 1.569909476665072e-05,
      "loss": 2.276,
      "step": 206572
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.5567505359649658,
      "learning_rate": 1.569778887080856e-05,
      "loss": 2.1383,
      "step": 206573
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.478415846824646,
      "learning_rate": 1.5696483027823614e-05,
      "loss": 2.2158,
      "step": 206574
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.5081167221069336,
      "learning_rate": 1.569517723769602e-05,
      "loss": 1.9929,
      "step": 206575
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.5817633867263794,
      "learning_rate": 1.569387150042605e-05,
      "loss": 1.9612,
      "step": 206576
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.4781014919281006,
      "learning_rate": 1.569256581601399e-05,
      "loss": 1.9493,
      "step": 206577
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.4094213247299194,
      "learning_rate": 1.569126018446002e-05,
      "loss": 2.046,
      "step": 206578
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.405762791633606,
      "learning_rate": 1.56899546057644e-05,
      "loss": 2.0727,
      "step": 206579
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.3737163543701172,
      "learning_rate": 1.568864907992743e-05,
      "loss": 1.8505,
      "step": 206580
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.481719970703125,
      "learning_rate": 1.5687343606949243e-05,
      "loss": 1.7589,
      "step": 206581
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.3739031553268433,
      "learning_rate": 1.568603818683021e-05,
      "loss": 1.9142,
      "step": 206582
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.4070779085159302,
      "learning_rate": 1.568473281957049e-05,
      "loss": 1.9006,
      "step": 206583
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.3095180988311768,
      "learning_rate": 1.5683427505170355e-05,
      "loss": 1.8715,
      "step": 206584
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.535855770111084,
      "learning_rate": 1.568212224362997e-05,
      "loss": 1.8571,
      "step": 206585
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.3098523616790771,
      "learning_rate": 1.5680817034949732e-05,
      "loss": 1.932,
      "step": 206586
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.5171210765838623,
      "learning_rate": 1.5679511879129714e-05,
      "loss": 2.1242,
      "step": 206587
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.5299092531204224,
      "learning_rate": 1.567820677617031e-05,
      "loss": 1.8185,
      "step": 206588
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.34526526927948,
      "learning_rate": 1.5676901726071687e-05,
      "loss": 1.8407,
      "step": 206589
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.3636804819107056,
      "learning_rate": 1.5675596728834082e-05,
      "loss": 2.1251,
      "step": 206590
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.6160407066345215,
      "learning_rate": 1.5674291784457726e-05,
      "loss": 2.0354,
      "step": 206591
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.4454745054244995,
      "learning_rate": 1.5672986892942918e-05,
      "loss": 2.0436,
      "step": 206592
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.5167659521102905,
      "learning_rate": 1.5671682054289857e-05,
      "loss": 1.8939,
      "step": 206593
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.740560531616211,
      "learning_rate": 1.5670377268498814e-05,
      "loss": 1.9629,
      "step": 206594
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.4185981750488281,
      "learning_rate": 1.566907253556995e-05,
      "loss": 2.26,
      "step": 206595
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.461587905883789,
      "learning_rate": 1.56677678555037e-05,
      "loss": 2.0564,
      "step": 206596
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.4846221208572388,
      "learning_rate": 1.5666463228300064e-05,
      "loss": 2.0799,
      "step": 206597
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.9015157222747803,
      "learning_rate": 1.5665158653959477e-05,
      "loss": 1.9325,
      "step": 206598
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.5173099040985107,
      "learning_rate": 1.5663854132482033e-05,
      "loss": 2.0723,
      "step": 206599
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.43060302734375,
      "learning_rate": 1.5662549663868075e-05,
      "loss": 1.8088,
      "step": 206600
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.6110061407089233,
      "learning_rate": 1.566124524811779e-05,
      "loss": 2.0917,
      "step": 206601
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.4470447301864624,
      "learning_rate": 1.565994088523156e-05,
      "loss": 1.8194,
      "step": 206602
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.4143973588943481,
      "learning_rate": 1.5658636575209373e-05,
      "loss": 1.9331,
      "step": 206603
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.4157443046569824,
      "learning_rate": 1.5657332318051698e-05,
      "loss": 1.9417,
      "step": 206604
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.410176396369934,
      "learning_rate": 1.5656028113758633e-05,
      "loss": 1.7309,
      "step": 206605
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.4620037078857422,
      "learning_rate": 1.5654723962330517e-05,
      "loss": 1.9723,
      "step": 206606
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.428894281387329,
      "learning_rate": 1.5653419863767547e-05,
      "loss": 2.016,
      "step": 206607
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.3746188879013062,
      "learning_rate": 1.565211581807002e-05,
      "loss": 1.9026,
      "step": 206608
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.4073293209075928,
      "learning_rate": 1.5650811825238074e-05,
      "loss": 2.0037,
      "step": 206609
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.5841305255889893,
      "learning_rate": 1.564950788527204e-05,
      "loss": 1.933,
      "step": 206610
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.3746720552444458,
      "learning_rate": 1.564820399817208e-05,
      "loss": 1.8749,
      "step": 206611
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.4039489030838013,
      "learning_rate": 1.5646900163938535e-05,
      "loss": 1.9793,
      "step": 206612
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.4875788688659668,
      "learning_rate": 1.564559638257157e-05,
      "loss": 1.888,
      "step": 206613
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.4464497566223145,
      "learning_rate": 1.564429265407151e-05,
      "loss": 2.0613,
      "step": 206614
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.4295463562011719,
      "learning_rate": 1.5642988978438498e-05,
      "loss": 2.2211,
      "step": 206615
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.4532251358032227,
      "learning_rate": 1.564168535567283e-05,
      "loss": 1.9253,
      "step": 206616
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.5331310033798218,
      "learning_rate": 1.5640381785774703e-05,
      "loss": 1.9772,
      "step": 206617
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.440891146659851,
      "learning_rate": 1.5639078268744452e-05,
      "loss": 1.8064,
      "step": 206618
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.5256638526916504,
      "learning_rate": 1.5637774804582214e-05,
      "loss": 2.0993,
      "step": 206619
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.5125806331634521,
      "learning_rate": 1.5636471393288384e-05,
      "loss": 1.9889,
      "step": 206620
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.3918321132659912,
      "learning_rate": 1.5635168034863e-05,
      "loss": 1.8018,
      "step": 206621
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.3338977098464966,
      "learning_rate": 1.5633864729306455e-05,
      "loss": 2.0677,
      "step": 206622
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.5933815240859985,
      "learning_rate": 1.5632561476618887e-05,
      "loss": 2.1571,
      "step": 206623
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.4393892288208008,
      "learning_rate": 1.563125827680063e-05,
      "loss": 2.0055,
      "step": 206624
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.446531891822815,
      "learning_rate": 1.562995512985188e-05,
      "loss": 2.1355,
      "step": 206625
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.5035523176193237,
      "learning_rate": 1.5628652035772936e-05,
      "loss": 2.0899,
      "step": 206626
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.4253708124160767,
      "learning_rate": 1.562734899456394e-05,
      "loss": 2.1167,
      "step": 206627
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.4393677711486816,
      "learning_rate": 1.5626046006225213e-05,
      "loss": 1.7887,
      "step": 206628
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.4191138744354248,
      "learning_rate": 1.5624743070756962e-05,
      "loss": 1.8683,
      "step": 206629
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.5056242942810059,
      "learning_rate": 1.562344018815945e-05,
      "loss": 1.917,
      "step": 206630
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.4787801504135132,
      "learning_rate": 1.5622137358432853e-05,
      "loss": 2.0514,
      "step": 206631
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.3973503112792969,
      "learning_rate": 1.562083458157759e-05,
      "loss": 2.0602,
      "step": 206632
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.4390896558761597,
      "learning_rate": 1.561953185759367e-05,
      "loss": 2.1648,
      "step": 206633
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.4713175296783447,
      "learning_rate": 1.561822918648149e-05,
      "loss": 1.9238,
      "step": 206634
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.5342061519622803,
      "learning_rate": 1.5616926568241218e-05,
      "loss": 2.275,
      "step": 206635
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.5177888870239258,
      "learning_rate": 1.5615624002873182e-05,
      "loss": 1.9248,
      "step": 206636
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.5198036432266235,
      "learning_rate": 1.5614321490377522e-05,
      "loss": 2.0376,
      "step": 206637
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.4413135051727295,
      "learning_rate": 1.5613019030754604e-05,
      "loss": 2.1438,
      "step": 206638
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.4453866481781006,
      "learning_rate": 1.561171662400449e-05,
      "loss": 1.8292,
      "step": 206639
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.682047724723816,
      "learning_rate": 1.561041427012758e-05,
      "loss": 1.9974,
      "step": 206640
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.4337228536605835,
      "learning_rate": 1.5609111969124045e-05,
      "loss": 1.7715,
      "step": 206641
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.4759341478347778,
      "learning_rate": 1.560780972099418e-05,
      "loss": 1.779,
      "step": 206642
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.3408496379852295,
      "learning_rate": 1.560650752573812e-05,
      "loss": 1.9551,
      "step": 206643
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.429315447807312,
      "learning_rate": 1.5605205383356302e-05,
      "loss": 2.0215,
      "step": 206644
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.437424659729004,
      "learning_rate": 1.560390329384872e-05,
      "loss": 1.7515,
      "step": 206645
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.4025577306747437,
      "learning_rate": 1.5602601257215806e-05,
      "loss": 2.2359,
      "step": 206646
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.4306235313415527,
      "learning_rate": 1.5601299273457702e-05,
      "loss": 1.9789,
      "step": 206647
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.3598272800445557,
      "learning_rate": 1.5599997342574698e-05,
      "loss": 2.0302,
      "step": 206648
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.3868921995162964,
      "learning_rate": 1.5598695464567e-05,
      "loss": 2.0886,
      "step": 206649
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.4092754125595093,
      "learning_rate": 1.5597393639434908e-05,
      "loss": 1.7885,
      "step": 206650
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.4987683296203613,
      "learning_rate": 1.559609186717865e-05,
      "loss": 2.0182,
      "step": 206651
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.5243760347366333,
      "learning_rate": 1.559479014779843e-05,
      "loss": 1.7376,
      "step": 206652
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.3929715156555176,
      "learning_rate": 1.559348848129448e-05,
      "loss": 2.1658,
      "step": 206653
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.3843684196472168,
      "learning_rate": 1.55921868676671e-05,
      "loss": 1.781,
      "step": 206654
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.4261051416397095,
      "learning_rate": 1.5590885306916455e-05,
      "loss": 1.7847,
      "step": 206655
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.5946331024169922,
      "learning_rate": 1.558958379904288e-05,
      "loss": 2.0327,
      "step": 206656
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.4401763677597046,
      "learning_rate": 1.5588282344046543e-05,
      "loss": 2.0654,
      "step": 206657
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.548454761505127,
      "learning_rate": 1.5586980941927705e-05,
      "loss": 1.9836,
      "step": 206658
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.4710626602172852,
      "learning_rate": 1.558567959268664e-05,
      "loss": 1.972,
      "step": 206659
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.4513816833496094,
      "learning_rate": 1.5584378296323573e-05,
      "loss": 1.952,
      "step": 206660
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.6099216938018799,
      "learning_rate": 1.5583077052838712e-05,
      "loss": 1.9283,
      "step": 206661
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.524295687675476,
      "learning_rate": 1.558177586223235e-05,
      "loss": 2.0686,
      "step": 206662
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.437710165977478,
      "learning_rate": 1.5580474724504722e-05,
      "loss": 1.7891,
      "step": 206663
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.3884960412979126,
      "learning_rate": 1.5579173639655996e-05,
      "loss": 1.8806,
      "step": 206664
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.6622638702392578,
      "learning_rate": 1.5577872607686504e-05,
      "loss": 1.9697,
      "step": 206665
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.4340206384658813,
      "learning_rate": 1.557657162859648e-05,
      "loss": 1.9989,
      "step": 206666
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.4186137914657593,
      "learning_rate": 1.5575270702386057e-05,
      "loss": 1.8745,
      "step": 206667
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.5104786157608032,
      "learning_rate": 1.5573969829055634e-05,
      "loss": 1.9521,
      "step": 206668
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.8306448459625244,
      "learning_rate": 1.5572669008605376e-05,
      "loss": 1.9966,
      "step": 206669
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.3550058603286743,
      "learning_rate": 1.5571368241035487e-05,
      "loss": 1.9229,
      "step": 206670
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.4641399383544922,
      "learning_rate": 1.5570067526346295e-05,
      "loss": 1.9528,
      "step": 206671
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.5707415342330933,
      "learning_rate": 1.556876686453794e-05,
      "loss": 1.8568,
      "step": 206672
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.5974047183990479,
      "learning_rate": 1.556746625561075e-05,
      "loss": 1.9691,
      "step": 206673
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.5154095888137817,
      "learning_rate": 1.556616569956496e-05,
      "loss": 1.8202,
      "step": 206674
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.4505723714828491,
      "learning_rate": 1.55648651964008e-05,
      "loss": 1.9895,
      "step": 206675
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.6170459985733032,
      "learning_rate": 1.556356474611844e-05,
      "loss": 2.0195,
      "step": 206676
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.4606561660766602,
      "learning_rate": 1.5562264348718213e-05,
      "loss": 1.9868,
      "step": 206677
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.4028122425079346,
      "learning_rate": 1.5560964004200315e-05,
      "loss": 1.9886,
      "step": 206678
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.386538028717041,
      "learning_rate": 1.555966371256502e-05,
      "loss": 2.0757,
      "step": 206679
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.4538013935089111,
      "learning_rate": 1.555836347381255e-05,
      "loss": 2.1598,
      "step": 206680
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.5855116844177246,
      "learning_rate": 1.5557063287943185e-05,
      "loss": 1.7699,
      "step": 206681
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.6183338165283203,
      "learning_rate": 1.5555763154957078e-05,
      "loss": 1.8566,
      "step": 206682
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.511518955230713,
      "learning_rate": 1.5554463074854572e-05,
      "loss": 2.1948,
      "step": 206683
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.6484007835388184,
      "learning_rate": 1.5553163047635796e-05,
      "loss": 2.2405,
      "step": 206684
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.5394976139068604,
      "learning_rate": 1.5551863073301084e-05,
      "loss": 1.9083,
      "step": 206685
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.341672658920288,
      "learning_rate": 1.5550563151850638e-05,
      "loss": 2.0865,
      "step": 206686
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.3409276008605957,
      "learning_rate": 1.5549263283284785e-05,
      "loss": 1.8568,
      "step": 206687
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.5814754962921143,
      "learning_rate": 1.554796346760363e-05,
      "loss": 1.6571,
      "step": 206688
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.6052155494689941,
      "learning_rate": 1.5546663704807506e-05,
      "loss": 1.9554,
      "step": 206689
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.4763824939727783,
      "learning_rate": 1.5545363994896576e-05,
      "loss": 1.9217,
      "step": 206690
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.606728196144104,
      "learning_rate": 1.5544064337871177e-05,
      "loss": 1.9128,
      "step": 206691
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.6062244176864624,
      "learning_rate": 1.5542764733731473e-05,
      "loss": 2.1794,
      "step": 206692
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.4104222059249878,
      "learning_rate": 1.5541465182477796e-05,
      "loss": 2.0863,
      "step": 206693
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.4703221321105957,
      "learning_rate": 1.5540165684110285e-05,
      "loss": 1.984,
      "step": 206694
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.510087251663208,
      "learning_rate": 1.5538866238629234e-05,
      "loss": 2.2169,
      "step": 206695
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.5001012086868286,
      "learning_rate": 1.5537566846034877e-05,
      "loss": 2.2797,
      "step": 206696
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.462613582611084,
      "learning_rate": 1.5536267506327448e-05,
      "loss": 1.8563,
      "step": 206697
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.4301886558532715,
      "learning_rate": 1.553496821950718e-05,
      "loss": 1.8924,
      "step": 206698
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.489210605621338,
      "learning_rate": 1.553366898557441e-05,
      "loss": 1.964,
      "step": 206699
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.544712781906128,
      "learning_rate": 1.5532369804529233e-05,
      "loss": 1.7987,
      "step": 206700
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.4853512048721313,
      "learning_rate": 1.5531070676371982e-05,
      "loss": 1.8613,
      "step": 206701
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.444892168045044,
      "learning_rate": 1.552977160110286e-05,
      "loss": 1.9676,
      "step": 206702
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.5376418828964233,
      "learning_rate": 1.5528472578722128e-05,
      "loss": 2.0586,
      "step": 206703
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.4504493474960327,
      "learning_rate": 1.552717360922999e-05,
      "loss": 2.0865,
      "step": 206704
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.4633234739303589,
      "learning_rate": 1.5525874692626815e-05,
      "loss": 2.0427,
      "step": 206705
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.4720498323440552,
      "learning_rate": 1.5524575828912665e-05,
      "loss": 1.8704,
      "step": 206706
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.4431118965148926,
      "learning_rate": 1.552327701808791e-05,
      "loss": 2.1195,
      "step": 206707
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.5365221500396729,
      "learning_rate": 1.5521978260152713e-05,
      "loss": 2.0501,
      "step": 206708
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.5228686332702637,
      "learning_rate": 1.5520679555107373e-05,
      "loss": 2.1208,
      "step": 206709
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.462853193283081,
      "learning_rate": 1.551938090295206e-05,
      "loss": 2.1832,
      "step": 206710
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.6486488580703735,
      "learning_rate": 1.5518082303687174e-05,
      "loss": 2.0243,
      "step": 206711
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.6014373302459717,
      "learning_rate": 1.5516783757312745e-05,
      "loss": 2.0709,
      "step": 206712
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.3709663152694702,
      "learning_rate": 1.5515485263829174e-05,
      "loss": 2.2362,
      "step": 206713
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.6742641925811768,
      "learning_rate": 1.5514186823236563e-05,
      "loss": 1.9041,
      "step": 206714
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.401666283607483,
      "learning_rate": 1.5512888435535308e-05,
      "loss": 1.8185,
      "step": 206715
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.409030556678772,
      "learning_rate": 1.5511590100725545e-05,
      "loss": 2.0687,
      "step": 206716
    },
    {
      "epoch": 2.69,
      "grad_norm": 2.4812169075012207,
      "learning_rate": 1.5510291818807575e-05,
      "loss": 1.8324,
      "step": 206717
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.450416088104248,
      "learning_rate": 1.5508993589781625e-05,
      "loss": 2.1175,
      "step": 206718
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.491505742073059,
      "learning_rate": 1.5507695413647903e-05,
      "loss": 2.1945,
      "step": 206719
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.3818697929382324,
      "learning_rate": 1.5506397290406635e-05,
      "loss": 2.0013,
      "step": 206720
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.6047273874282837,
      "learning_rate": 1.550509922005816e-05,
      "loss": 1.8846,
      "step": 206721
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.5881445407867432,
      "learning_rate": 1.550380120260257e-05,
      "loss": 2.1246,
      "step": 206722
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.4457051753997803,
      "learning_rate": 1.5502503238040274e-05,
      "loss": 1.8582,
      "step": 206723
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.556484580039978,
      "learning_rate": 1.5501205326371434e-05,
      "loss": 2.0224,
      "step": 206724
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.5479912757873535,
      "learning_rate": 1.5499907467596285e-05,
      "loss": 1.8637,
      "step": 206725
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.6543047428131104,
      "learning_rate": 1.5498609661715022e-05,
      "loss": 2.0409,
      "step": 206726
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.5141083002090454,
      "learning_rate": 1.5497311908727983e-05,
      "loss": 1.9932,
      "step": 206727
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.4402434825897217,
      "learning_rate": 1.5496014208635298e-05,
      "loss": 1.8878,
      "step": 206728
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.361971139907837,
      "learning_rate": 1.549471656143737e-05,
      "loss": 1.8031,
      "step": 206729
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.4133001565933228,
      "learning_rate": 1.5493418967134297e-05,
      "loss": 1.7423,
      "step": 206730
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.4894747734069824,
      "learning_rate": 1.5492121425726377e-05,
      "loss": 2.0989,
      "step": 206731
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.4460991621017456,
      "learning_rate": 1.5490823937213815e-05,
      "loss": 1.9379,
      "step": 206732
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.4798059463500977,
      "learning_rate": 1.548952650159694e-05,
      "loss": 1.7883,
      "step": 206733
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.6158623695373535,
      "learning_rate": 1.548822911887585e-05,
      "loss": 1.9599,
      "step": 206734
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.5237796306610107,
      "learning_rate": 1.5486931789050917e-05,
      "loss": 2.0487,
      "step": 206735
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.397607445716858,
      "learning_rate": 1.548563451212237e-05,
      "loss": 1.842,
      "step": 206736
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.4845550060272217,
      "learning_rate": 1.5484337288090377e-05,
      "loss": 2.1188,
      "step": 206737
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.4292724132537842,
      "learning_rate": 1.548304011695517e-05,
      "loss": 1.8993,
      "step": 206738
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.4290474653244019,
      "learning_rate": 1.5481742998717117e-05,
      "loss": 1.8772,
      "step": 206739
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.4213899374008179,
      "learning_rate": 1.5480445933376286e-05,
      "loss": 2.1108,
      "step": 206740
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.3783133029937744,
      "learning_rate": 1.5479148920933103e-05,
      "loss": 1.9189,
      "step": 206741
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.3704689741134644,
      "learning_rate": 1.5477851961387676e-05,
      "loss": 1.6828,
      "step": 206742
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.511448621749878,
      "learning_rate": 1.547655505474027e-05,
      "loss": 1.8848,
      "step": 206743
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.5321509838104248,
      "learning_rate": 1.547525820099118e-05,
      "loss": 2.0621,
      "step": 206744
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.427217721939087,
      "learning_rate": 1.5473961400140612e-05,
      "loss": 2.0574,
      "step": 206745
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.7047370672225952,
      "learning_rate": 1.547266465218876e-05,
      "loss": 1.9156,
      "step": 206746
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.3832590579986572,
      "learning_rate": 1.547136795713596e-05,
      "loss": 2.2572,
      "step": 206747
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.5255635976791382,
      "learning_rate": 1.5470071314982378e-05,
      "loss": 1.9215,
      "step": 206748
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.4313831329345703,
      "learning_rate": 1.5468774725728284e-05,
      "loss": 1.9157,
      "step": 206749
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.4391485452651978,
      "learning_rate": 1.5467478189373904e-05,
      "loss": 2.096,
      "step": 206750
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.5650315284729004,
      "learning_rate": 1.5466181705919544e-05,
      "loss": 2.066,
      "step": 206751
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.5684583187103271,
      "learning_rate": 1.5464885275365302e-05,
      "loss": 2.0905,
      "step": 206752
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.3742344379425049,
      "learning_rate": 1.5463588897711575e-05,
      "loss": 2.0336,
      "step": 206753
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.5089083909988403,
      "learning_rate": 1.5462292572958535e-05,
      "loss": 1.935,
      "step": 206754
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.7038527727127075,
      "learning_rate": 1.5460996301106378e-05,
      "loss": 2.0465,
      "step": 206755
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.5632102489471436,
      "learning_rate": 1.5459700082155467e-05,
      "loss": 1.9752,
      "step": 206756
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.5049476623535156,
      "learning_rate": 1.5458403916105878e-05,
      "loss": 2.0632,
      "step": 206757
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.4795969724655151,
      "learning_rate": 1.5457107802958002e-05,
      "loss": 2.0412,
      "step": 206758
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.3577940464019775,
      "learning_rate": 1.5455811742712043e-05,
      "loss": 2.0619,
      "step": 206759
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.5007600784301758,
      "learning_rate": 1.54545157353682e-05,
      "loss": 1.9935,
      "step": 206760
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.566137433052063,
      "learning_rate": 1.5453219780926706e-05,
      "loss": 2.0929,
      "step": 206761
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.3786892890930176,
      "learning_rate": 1.545192387938786e-05,
      "loss": 1.8666,
      "step": 206762
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.498897671699524,
      "learning_rate": 1.5450628030751833e-05,
      "loss": 1.9045,
      "step": 206763
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.4349149465560913,
      "learning_rate": 1.5449332235018953e-05,
      "loss": 1.8586,
      "step": 206764
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.3891679048538208,
      "learning_rate": 1.5448036492189386e-05,
      "loss": 2.1692,
      "step": 206765
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.361130952835083,
      "learning_rate": 1.5446740802263436e-05,
      "loss": 1.9186,
      "step": 206766
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.3992855548858643,
      "learning_rate": 1.5445445165241233e-05,
      "loss": 2.1208,
      "step": 206767
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.3804234266281128,
      "learning_rate": 1.5444149581123177e-05,
      "loss": 2.0122,
      "step": 206768
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.454064130783081,
      "learning_rate": 1.5442854049909338e-05,
      "loss": 1.9918,
      "step": 206769
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.3382831811904907,
      "learning_rate": 1.544155857160011e-05,
      "loss": 1.9106,
      "step": 206770
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.4698537588119507,
      "learning_rate": 1.544026314619563e-05,
      "loss": 2.0103,
      "step": 206771
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.3336299657821655,
      "learning_rate": 1.5438967773696264e-05,
      "loss": 2.0315,
      "step": 206772
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.5248786211013794,
      "learning_rate": 1.5437672454102047e-05,
      "loss": 1.9387,
      "step": 206773
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.4297864437103271,
      "learning_rate": 1.5436377187413406e-05,
      "loss": 2.1839,
      "step": 206774
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.5232617855072021,
      "learning_rate": 1.543508197363048e-05,
      "loss": 1.9048,
      "step": 206775
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.463905930519104,
      "learning_rate": 1.5433786812753567e-05,
      "loss": 1.9957,
      "step": 206776
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.5119267702102661,
      "learning_rate": 1.5432491704782868e-05,
      "loss": 2.2356,
      "step": 206777
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.344760775566101,
      "learning_rate": 1.543119664971868e-05,
      "loss": 2.0943,
      "step": 206778
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.5067123174667358,
      "learning_rate": 1.5429901647561173e-05,
      "loss": 1.8476,
      "step": 206779
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.5861598253250122,
      "learning_rate": 1.542860669831064e-05,
      "loss": 2.1663,
      "step": 206780
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.5212178230285645,
      "learning_rate": 1.5427311801967258e-05,
      "loss": 1.9642,
      "step": 206781
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.5434750318527222,
      "learning_rate": 1.5426016958531318e-05,
      "loss": 1.9961,
      "step": 206782
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.4470899105072021,
      "learning_rate": 1.5424722168003058e-05,
      "loss": 1.9037,
      "step": 206783
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.2940175533294678,
      "learning_rate": 1.542342743038274e-05,
      "loss": 1.8873,
      "step": 206784
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.429877519607544,
      "learning_rate": 1.5422132745670568e-05,
      "loss": 1.9235,
      "step": 206785
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.532698392868042,
      "learning_rate": 1.5420838113866806e-05,
      "loss": 1.9864,
      "step": 206786
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.5669879913330078,
      "learning_rate": 1.5419543534971655e-05,
      "loss": 1.9335,
      "step": 206787
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.5049550533294678,
      "learning_rate": 1.541824900898542e-05,
      "loss": 1.7652,
      "step": 206788
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.511175513267517,
      "learning_rate": 1.541695453590822e-05,
      "loss": 2.1549,
      "step": 206789
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.444245457649231,
      "learning_rate": 1.541566011574047e-05,
      "loss": 1.9784,
      "step": 206790
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.409294843673706,
      "learning_rate": 1.5414365748482294e-05,
      "loss": 1.9646,
      "step": 206791
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.3842107057571411,
      "learning_rate": 1.541307143413396e-05,
      "loss": 1.9859,
      "step": 206792
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.3730329275131226,
      "learning_rate": 1.5411777172695673e-05,
      "loss": 1.9931,
      "step": 206793
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.5318801403045654,
      "learning_rate": 1.541048296416776e-05,
      "loss": 1.8936,
      "step": 206794
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.41151762008667,
      "learning_rate": 1.5409188808550355e-05,
      "loss": 1.9563,
      "step": 206795
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.448476791381836,
      "learning_rate": 1.5407894705843826e-05,
      "loss": 1.8162,
      "step": 206796
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.3862152099609375,
      "learning_rate": 1.5406600656048308e-05,
      "loss": 2.1369,
      "step": 206797
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.4160900115966797,
      "learning_rate": 1.5405306659164095e-05,
      "loss": 1.8549,
      "step": 206798
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.614593267440796,
      "learning_rate": 1.5404012715191327e-05,
      "loss": 1.9406,
      "step": 206799
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.4394721984863281,
      "learning_rate": 1.54027188241304e-05,
      "loss": 1.7112,
      "step": 206800
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.4440933465957642,
      "learning_rate": 1.540142498598145e-05,
      "loss": 1.8999,
      "step": 206801
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.65753972530365,
      "learning_rate": 1.540013120074477e-05,
      "loss": 2.0741,
      "step": 206802
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.4398502111434937,
      "learning_rate": 1.5398837468420567e-05,
      "loss": 1.8836,
      "step": 206803
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.4102755784988403,
      "learning_rate": 1.5397543789009104e-05,
      "loss": 1.97,
      "step": 206804
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.4904159307479858,
      "learning_rate": 1.5396250162510583e-05,
      "loss": 1.9756,
      "step": 206805
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.459538221359253,
      "learning_rate": 1.5394956588925302e-05,
      "loss": 2.0151,
      "step": 206806
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.5537492036819458,
      "learning_rate": 1.5393663068253427e-05,
      "loss": 2.1409,
      "step": 206807
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.5226839780807495,
      "learning_rate": 1.5392369600495257e-05,
      "loss": 1.9093,
      "step": 206808
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.4537925720214844,
      "learning_rate": 1.5391076185651062e-05,
      "loss": 2.0759,
      "step": 206809
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.353576421737671,
      "learning_rate": 1.538978282372101e-05,
      "loss": 1.9087,
      "step": 206810
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.4218149185180664,
      "learning_rate": 1.538848951470536e-05,
      "loss": 1.8601,
      "step": 206811
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.48341703414917,
      "learning_rate": 1.5387196258604383e-05,
      "loss": 1.9977,
      "step": 206812
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.7645477056503296,
      "learning_rate": 1.5385903055418248e-05,
      "loss": 2.1402,
      "step": 206813
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.6072890758514404,
      "learning_rate": 1.538460990514728e-05,
      "loss": 1.7911,
      "step": 206814
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.4531846046447754,
      "learning_rate": 1.5383316807791723e-05,
      "loss": 1.9397,
      "step": 206815
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.536854863166809,
      "learning_rate": 1.5382023763351736e-05,
      "loss": 1.9979,
      "step": 206816
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.4996439218521118,
      "learning_rate": 1.5380730771827588e-05,
      "loss": 2.1167,
      "step": 206817
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.4714059829711914,
      "learning_rate": 1.5379437833219576e-05,
      "loss": 1.8856,
      "step": 206818
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.5966780185699463,
      "learning_rate": 1.537814494752787e-05,
      "loss": 2.2269,
      "step": 206819
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.3788933753967285,
      "learning_rate": 1.5376852114752768e-05,
      "loss": 1.9571,
      "step": 206820
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.3884776830673218,
      "learning_rate": 1.5375559334894473e-05,
      "loss": 2.1942,
      "step": 206821
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.534796953201294,
      "learning_rate": 1.5374266607953246e-05,
      "loss": 1.8422,
      "step": 206822
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.4892774820327759,
      "learning_rate": 1.5372973933929257e-05,
      "loss": 1.7646,
      "step": 206823
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.5237592458724976,
      "learning_rate": 1.537168131282287e-05,
      "loss": 1.8329,
      "step": 206824
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.6455830335617065,
      "learning_rate": 1.5370388744634187e-05,
      "loss": 2.0169,
      "step": 206825
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.5031652450561523,
      "learning_rate": 1.5369096229363607e-05,
      "loss": 1.8399,
      "step": 206826
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.4534128904342651,
      "learning_rate": 1.5367803767011265e-05,
      "loss": 1.7926,
      "step": 206827
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.4256932735443115,
      "learning_rate": 1.5366511357577394e-05,
      "loss": 1.8982,
      "step": 206828
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.5462795495986938,
      "learning_rate": 1.5365219001062256e-05,
      "loss": 1.8291,
      "step": 206829
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.368722677230835,
      "learning_rate": 1.536392669746612e-05,
      "loss": 1.9383,
      "step": 206830
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.4360920190811157,
      "learning_rate": 1.536263444678916e-05,
      "loss": 2.0925,
      "step": 206831
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.4661145210266113,
      "learning_rate": 1.5361342249031728e-05,
      "loss": 2.1683,
      "step": 206832
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.4554388523101807,
      "learning_rate": 1.5360050104193965e-05,
      "loss": 1.9436,
      "step": 206833
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.4997689723968506,
      "learning_rate": 1.5358758012276107e-05,
      "loss": 2.0132,
      "step": 206834
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.628941297531128,
      "learning_rate": 1.5357465973278482e-05,
      "loss": 1.9394,
      "step": 206835
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.4214227199554443,
      "learning_rate": 1.535617398720126e-05,
      "loss": 2.048,
      "step": 206836
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.4267277717590332,
      "learning_rate": 1.535488205404467e-05,
      "loss": 1.8334,
      "step": 206837
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.555044174194336,
      "learning_rate": 1.5353590173809017e-05,
      "loss": 2.0155,
      "step": 206838
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.522302269935608,
      "learning_rate": 1.5352298346494497e-05,
      "loss": 1.9787,
      "step": 206839
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.4179801940917969,
      "learning_rate": 1.5351006572101342e-05,
      "loss": 2.03,
      "step": 206840
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.5085091590881348,
      "learning_rate": 1.5349714850629823e-05,
      "loss": 2.105,
      "step": 206841
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.5390918254852295,
      "learning_rate": 1.5348423182080137e-05,
      "loss": 1.9345,
      "step": 206842
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.4850900173187256,
      "learning_rate": 1.5347131566452588e-05,
      "loss": 2.0973,
      "step": 206843
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.4826596975326538,
      "learning_rate": 1.5345840003747366e-05,
      "loss": 1.9222,
      "step": 206844
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.4853907823562622,
      "learning_rate": 1.534454849396475e-05,
      "loss": 2.0275,
      "step": 206845
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.309817910194397,
      "learning_rate": 1.5343257037104893e-05,
      "loss": 1.9587,
      "step": 206846
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.400518774986267,
      "learning_rate": 1.5341965633168174e-05,
      "loss": 1.9161,
      "step": 206847
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.5298402309417725,
      "learning_rate": 1.5340674282154687e-05,
      "loss": 2.036,
      "step": 206848
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.3901525735855103,
      "learning_rate": 1.5339382984064797e-05,
      "loss": 2.2,
      "step": 206849
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.3747367858886719,
      "learning_rate": 1.533809173889867e-05,
      "loss": 1.7569,
      "step": 206850
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.3744560480117798,
      "learning_rate": 1.5336800546656547e-05,
      "loss": 1.7838,
      "step": 206851
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.400037407875061,
      "learning_rate": 1.533550940733872e-05,
      "loss": 2.0608,
      "step": 206852
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.5175549983978271,
      "learning_rate": 1.533421832094539e-05,
      "loss": 2.2143,
      "step": 206853
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.4522126913070679,
      "learning_rate": 1.533292728747676e-05,
      "loss": 2.0893,
      "step": 206854
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.3845107555389404,
      "learning_rate": 1.533163630693316e-05,
      "loss": 2.0079,
      "step": 206855
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.374731183052063,
      "learning_rate": 1.5330345379314724e-05,
      "loss": 1.9365,
      "step": 206856
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.3447991609573364,
      "learning_rate": 1.532905450462182e-05,
      "loss": 2.0877,
      "step": 206857
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.4257093667984009,
      "learning_rate": 1.532776368285458e-05,
      "loss": 2.0051,
      "step": 206858
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.3652842044830322,
      "learning_rate": 1.532647291401333e-05,
      "loss": 1.8984,
      "step": 206859
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.3583643436431885,
      "learning_rate": 1.5325182198098185e-05,
      "loss": 1.9767,
      "step": 206860
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.4143245220184326,
      "learning_rate": 1.53238915351095e-05,
      "loss": 2.0093,
      "step": 206861
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.4565404653549194,
      "learning_rate": 1.5322600925047478e-05,
      "loss": 2.1052,
      "step": 206862
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.4368364810943604,
      "learning_rate": 1.5321310367912355e-05,
      "loss": 1.935,
      "step": 206863
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.5862219333648682,
      "learning_rate": 1.5320019863704393e-05,
      "loss": 1.8136,
      "step": 206864
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.4678691625595093,
      "learning_rate": 1.5318729412423824e-05,
      "loss": 1.7595,
      "step": 206865
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.3595870733261108,
      "learning_rate": 1.531743901407082e-05,
      "loss": 2.0525,
      "step": 206866
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.446922779083252,
      "learning_rate": 1.531614866864571e-05,
      "loss": 2.0534,
      "step": 206867
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.3632301092147827,
      "learning_rate": 1.5314858376148697e-05,
      "loss": 1.9259,
      "step": 206868
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.7491304874420166,
      "learning_rate": 1.5313568136580045e-05,
      "loss": 1.9899,
      "step": 206869
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.3923890590667725,
      "learning_rate": 1.5312277949939955e-05,
      "loss": 1.8997,
      "step": 206870
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.6587542295455933,
      "learning_rate": 1.5310987816228694e-05,
      "loss": 1.93,
      "step": 206871
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.4875422716140747,
      "learning_rate": 1.5309697735446492e-05,
      "loss": 2.0272,
      "step": 206872
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.4232995510101318,
      "learning_rate": 1.5308407707593584e-05,
      "loss": 1.8815,
      "step": 206873
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.5012036561965942,
      "learning_rate": 1.5307117732670205e-05,
      "loss": 1.7727,
      "step": 206874
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.4167665243148804,
      "learning_rate": 1.530582781067662e-05,
      "loss": 1.8729,
      "step": 206875
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.51682448387146,
      "learning_rate": 1.5304537941613094e-05,
      "loss": 2.041,
      "step": 206876
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.4841288328170776,
      "learning_rate": 1.5303248125479794e-05,
      "loss": 2.0642,
      "step": 206877
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.5090150833129883,
      "learning_rate": 1.530195836227699e-05,
      "loss": 1.92,
      "step": 206878
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.4302103519439697,
      "learning_rate": 1.5300668652004943e-05,
      "loss": 1.8505,
      "step": 206879
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.4480271339416504,
      "learning_rate": 1.5299378994663825e-05,
      "loss": 2.0269,
      "step": 206880
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.5658104419708252,
      "learning_rate": 1.5298089390253966e-05,
      "loss": 2.0602,
      "step": 206881
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.5349584817886353,
      "learning_rate": 1.52967998387756e-05,
      "loss": 2.0538,
      "step": 206882
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.6108158826828003,
      "learning_rate": 1.529551034022889e-05,
      "loss": 1.9766,
      "step": 206883
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.6040277481079102,
      "learning_rate": 1.5294220894614106e-05,
      "loss": 1.8871,
      "step": 206884
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.3617562055587769,
      "learning_rate": 1.529293150193155e-05,
      "loss": 2.0537,
      "step": 206885
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.7129435539245605,
      "learning_rate": 1.529164216218135e-05,
      "loss": 2.1144,
      "step": 206886
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.3606045246124268,
      "learning_rate": 1.5290352875363842e-05,
      "loss": 1.7349,
      "step": 206887
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.4414889812469482,
      "learning_rate": 1.528906364147926e-05,
      "loss": 1.7486,
      "step": 206888
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.5313293933868408,
      "learning_rate": 1.52877744605278e-05,
      "loss": 1.9254,
      "step": 206889
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.5530918836593628,
      "learning_rate": 1.5286485332509667e-05,
      "loss": 1.8728,
      "step": 206890
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.5224277973175049,
      "learning_rate": 1.5285196257425225e-05,
      "loss": 1.9019,
      "step": 206891
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.5779175758361816,
      "learning_rate": 1.5283907235274572e-05,
      "loss": 1.9166,
      "step": 206892
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.6728298664093018,
      "learning_rate": 1.5282618266058044e-05,
      "loss": 2.0147,
      "step": 206893
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.4921892881393433,
      "learning_rate": 1.5281329349775874e-05,
      "loss": 1.978,
      "step": 206894
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.5128167867660522,
      "learning_rate": 1.528004048642829e-05,
      "loss": 1.9203,
      "step": 206895
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.489057183265686,
      "learning_rate": 1.5278751676015467e-05,
      "loss": 2.0082,
      "step": 206896
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.5201205015182495,
      "learning_rate": 1.527746291853773e-05,
      "loss": 2.0019,
      "step": 206897
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.638547658920288,
      "learning_rate": 1.527617421399525e-05,
      "loss": 1.9093,
      "step": 206898
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.3774014711380005,
      "learning_rate": 1.527488556238833e-05,
      "loss": 2.1358,
      "step": 206899
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.4687806367874146,
      "learning_rate": 1.5273596963717195e-05,
      "loss": 2.0799,
      "step": 206900
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.4496029615402222,
      "learning_rate": 1.5272308417982084e-05,
      "loss": 1.8445,
      "step": 206901
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.528799295425415,
      "learning_rate": 1.527101992518319e-05,
      "loss": 2.2023,
      "step": 206902
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.5882512331008911,
      "learning_rate": 1.5269731485320824e-05,
      "loss": 2.1,
      "step": 206903
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.554224967956543,
      "learning_rate": 1.5268443098395144e-05,
      "loss": 2.1141,
      "step": 206904
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.420032262802124,
      "learning_rate": 1.526715476440645e-05,
      "loss": 2.0282,
      "step": 206905
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.8173604011535645,
      "learning_rate": 1.5265866483355017e-05,
      "loss": 1.9783,
      "step": 206906
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.3934937715530396,
      "learning_rate": 1.5264578255241e-05,
      "loss": 1.949,
      "step": 206907
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.3374396562576294,
      "learning_rate": 1.526329008006464e-05,
      "loss": 1.7793,
      "step": 206908
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.5271648168563843,
      "learning_rate": 1.5262001957826232e-05,
      "loss": 1.9772,
      "step": 206909
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.4413800239562988,
      "learning_rate": 1.5260713888525978e-05,
      "loss": 1.9824,
      "step": 206910
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.4257012605667114,
      "learning_rate": 1.5259425872164178e-05,
      "loss": 1.9521,
      "step": 206911
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.4257144927978516,
      "learning_rate": 1.5258137908740997e-05,
      "loss": 1.8643,
      "step": 206912
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.338958501815796,
      "learning_rate": 1.5256849998256738e-05,
      "loss": 1.8456,
      "step": 206913
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.4435861110687256,
      "learning_rate": 1.525556214071153e-05,
      "loss": 1.9618,
      "step": 206914
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.3226569890975952,
      "learning_rate": 1.5254274336105743e-05,
      "loss": 1.7434,
      "step": 206915
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.491266131401062,
      "learning_rate": 1.5252986584439508e-05,
      "loss": 2.0805,
      "step": 206916
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.391653299331665,
      "learning_rate": 1.5251698885713193e-05,
      "loss": 1.9607,
      "step": 206917
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.3977956771850586,
      "learning_rate": 1.5250411239926896e-05,
      "loss": 1.9457,
      "step": 206918
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.443030595779419,
      "learning_rate": 1.5249123647080951e-05,
      "loss": 1.9802,
      "step": 206919
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.5178782939910889,
      "learning_rate": 1.5247836107175592e-05,
      "loss": 1.848,
      "step": 206920
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.4553800821304321,
      "learning_rate": 1.5246548620211053e-05,
      "loss": 1.8864,
      "step": 206921
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.4376270771026611,
      "learning_rate": 1.5245261186187464e-05,
      "loss": 2.1725,
      "step": 206922
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.7800703048706055,
      "learning_rate": 1.5243973805105226e-05,
      "loss": 1.9229,
      "step": 206923
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.4313790798187256,
      "learning_rate": 1.5242686476964472e-05,
      "loss": 2.151,
      "step": 206924
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.2602356672286987,
      "learning_rate": 1.5241399201765503e-05,
      "loss": 1.7442,
      "step": 206925
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.443541169166565,
      "learning_rate": 1.5240111979508551e-05,
      "loss": 1.9254,
      "step": 206926
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.3648918867111206,
      "learning_rate": 1.5238824810193817e-05,
      "loss": 1.8681,
      "step": 206927
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.643437385559082,
      "learning_rate": 1.5237537693821533e-05,
      "loss": 1.9494,
      "step": 206928
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.5758464336395264,
      "learning_rate": 1.5236250630392033e-05,
      "loss": 2.1308,
      "step": 206929
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.4766513109207153,
      "learning_rate": 1.5234963619905416e-05,
      "loss": 2.1595,
      "step": 206930
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.390498399734497,
      "learning_rate": 1.5233676662362015e-05,
      "loss": 2.029,
      "step": 206931
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.4363632202148438,
      "learning_rate": 1.5232389757762098e-05,
      "loss": 1.9862,
      "step": 206932
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.5608656406402588,
      "learning_rate": 1.5231102906105797e-05,
      "loss": 2.1225,
      "step": 206933
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.601205825805664,
      "learning_rate": 1.5229816107393444e-05,
      "loss": 2.25,
      "step": 206934
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.4474081993103027,
      "learning_rate": 1.522852936162524e-05,
      "loss": 2.0119,
      "step": 206935
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.521493911743164,
      "learning_rate": 1.5227242668801388e-05,
      "loss": 1.9973,
      "step": 206936
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.375192403793335,
      "learning_rate": 1.5225956028922215e-05,
      "loss": 1.8925,
      "step": 206937
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.5125927925109863,
      "learning_rate": 1.522466944198789e-05,
      "loss": 1.974,
      "step": 206938
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.4345734119415283,
      "learning_rate": 1.522338290799865e-05,
      "loss": 1.9705,
      "step": 206939
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.4015014171600342,
      "learning_rate": 1.5222096426954821e-05,
      "loss": 1.823,
      "step": 206940
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.4717592000961304,
      "learning_rate": 1.522080999885651e-05,
      "loss": 2.0392,
      "step": 206941
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.5916764736175537,
      "learning_rate": 1.5219523623704078e-05,
      "loss": 1.9981,
      "step": 206942
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.408622145652771,
      "learning_rate": 1.5218237301497694e-05,
      "loss": 1.8348,
      "step": 206943
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.487552523612976,
      "learning_rate": 1.5216951032237657e-05,
      "loss": 2.2453,
      "step": 206944
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.522518277168274,
      "learning_rate": 1.5215664815924101e-05,
      "loss": 2.0483,
      "step": 206945
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.4270306825637817,
      "learning_rate": 1.5214378652557357e-05,
      "loss": 1.9219,
      "step": 206946
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.6312942504882812,
      "learning_rate": 1.5213092542137595e-05,
      "loss": 1.9702,
      "step": 206947
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.4798240661621094,
      "learning_rate": 1.5211806484665146e-05,
      "loss": 2.003,
      "step": 206948
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.5790365934371948,
      "learning_rate": 1.5210520480140209e-05,
      "loss": 2.0879,
      "step": 206949
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.5180284976959229,
      "learning_rate": 1.5209234528562985e-05,
      "loss": 1.9988,
      "step": 206950
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.4599136114120483,
      "learning_rate": 1.5207948629933709e-05,
      "loss": 1.9134,
      "step": 206951
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.4641664028167725,
      "learning_rate": 1.520666278425271e-05,
      "loss": 2.1484,
      "step": 206952
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.5242537260055542,
      "learning_rate": 1.5205376991520123e-05,
      "loss": 1.9787,
      "step": 206953
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.5166386365890503,
      "learning_rate": 1.520409125173625e-05,
      "loss": 1.8101,
      "step": 206954
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.4489696025848389,
      "learning_rate": 1.520280556490132e-05,
      "loss": 1.9787,
      "step": 206955
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.5202094316482544,
      "learning_rate": 1.5201519931015604e-05,
      "loss": 1.7889,
      "step": 206956
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.5573264360427856,
      "learning_rate": 1.5200234350079198e-05,
      "loss": 2.1824,
      "step": 206957
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.4967952966690063,
      "learning_rate": 1.5198948822092539e-05,
      "loss": 2.0219,
      "step": 206958
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.4985525608062744,
      "learning_rate": 1.519766334705569e-05,
      "loss": 1.9805,
      "step": 206959
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.4084830284118652,
      "learning_rate": 1.519637792496905e-05,
      "loss": 1.9136,
      "step": 206960
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.298424243927002,
      "learning_rate": 1.5195092555832755e-05,
      "loss": 2.1327,
      "step": 206961
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.3971819877624512,
      "learning_rate": 1.5193807239647072e-05,
      "loss": 2.018,
      "step": 206962
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.4413378238677979,
      "learning_rate": 1.5192521976412198e-05,
      "loss": 1.8935,
      "step": 206963
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.457726001739502,
      "learning_rate": 1.5191236766128434e-05,
      "loss": 1.9554,
      "step": 206964
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.3923417329788208,
      "learning_rate": 1.518995160879598e-05,
      "loss": 1.9608,
      "step": 206965
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.5394647121429443,
      "learning_rate": 1.5188666504415104e-05,
      "loss": 1.9318,
      "step": 206966
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.4728835821151733,
      "learning_rate": 1.5187381452986069e-05,
      "loss": 1.8015,
      "step": 206967
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.5406663417816162,
      "learning_rate": 1.5186096454509044e-05,
      "loss": 2.0803,
      "step": 206968
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.3191882371902466,
      "learning_rate": 1.5184811508984262e-05,
      "loss": 1.9527,
      "step": 206969
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.4777424335479736,
      "learning_rate": 1.5183526616412057e-05,
      "loss": 2.0863,
      "step": 206970
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.5127097368240356,
      "learning_rate": 1.518224177679256e-05,
      "loss": 2.0961,
      "step": 206971
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.4779009819030762,
      "learning_rate": 1.5180956990126103e-05,
      "loss": 1.9923,
      "step": 206972
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.363334059715271,
      "learning_rate": 1.5179672256412889e-05,
      "loss": 1.9875,
      "step": 206973
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.5966360569000244,
      "learning_rate": 1.517838757565315e-05,
      "loss": 1.9742,
      "step": 206974
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.5382708311080933,
      "learning_rate": 1.5177102947847086e-05,
      "loss": 1.9462,
      "step": 206975
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.402450680732727,
      "learning_rate": 1.5175818372994998e-05,
      "loss": 1.8652,
      "step": 206976
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.5168589353561401,
      "learning_rate": 1.5174533851097081e-05,
      "loss": 2.224,
      "step": 206977
    },
    {
      "epoch": 2.69,
      "grad_norm": 1.3533602952957153,
      "learning_rate": 1.5173249382153606e-05,
      "loss": 1.8696,
      "step": 206978
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.311178207397461,
      "learning_rate": 1.517196496616484e-05,
      "loss": 1.8933,
      "step": 206979
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.537992000579834,
      "learning_rate": 1.5170680603130947e-05,
      "loss": 1.9904,
      "step": 206980
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.417559266090393,
      "learning_rate": 1.5169396293052193e-05,
      "loss": 1.9559,
      "step": 206981
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.539398431777954,
      "learning_rate": 1.5168112035928848e-05,
      "loss": 1.8868,
      "step": 206982
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.484070062637329,
      "learning_rate": 1.5166827831761074e-05,
      "loss": 1.7736,
      "step": 206983
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.5661518573760986,
      "learning_rate": 1.5165543680549208e-05,
      "loss": 2.1771,
      "step": 206984
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.4378507137298584,
      "learning_rate": 1.5164259582293414e-05,
      "loss": 2.037,
      "step": 206985
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.4202699661254883,
      "learning_rate": 1.5162975536994059e-05,
      "loss": 1.9558,
      "step": 206986
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.3723024129867554,
      "learning_rate": 1.5161691544651178e-05,
      "loss": 2.0198,
      "step": 206987
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.6046797037124634,
      "learning_rate": 1.5160407605265167e-05,
      "loss": 2.0016,
      "step": 206988
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.5302278995513916,
      "learning_rate": 1.5159123718836164e-05,
      "loss": 2.0104,
      "step": 206989
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.488294243812561,
      "learning_rate": 1.5157839885364497e-05,
      "loss": 1.7477,
      "step": 206990
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.5369740724563599,
      "learning_rate": 1.5156556104850337e-05,
      "loss": 2.0199,
      "step": 206991
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.4517757892608643,
      "learning_rate": 1.5155272377294015e-05,
      "loss": 2.1284,
      "step": 206992
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.5345110893249512,
      "learning_rate": 1.515398870269563e-05,
      "loss": 2.0106,
      "step": 206993
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.4659788608551025,
      "learning_rate": 1.5152705081055517e-05,
      "loss": 1.9904,
      "step": 206994
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.5576750040054321,
      "learning_rate": 1.5151421512373873e-05,
      "loss": 1.8799,
      "step": 206995
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.4341033697128296,
      "learning_rate": 1.5150137996651001e-05,
      "loss": 1.9632,
      "step": 206996
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.3930180072784424,
      "learning_rate": 1.5148854533887066e-05,
      "loss": 2.0023,
      "step": 206997
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.3294425010681152,
      "learning_rate": 1.5147571124082403e-05,
      "loss": 1.845,
      "step": 206998
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.5527760982513428,
      "learning_rate": 1.5146287767237076e-05,
      "loss": 1.8508,
      "step": 206999
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.4738081693649292,
      "learning_rate": 1.5145004463351517e-05,
      "loss": 1.8386,
      "step": 207000
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.5948634147644043,
      "learning_rate": 1.5143721212425797e-05,
      "loss": 1.9192,
      "step": 207001
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.5075091123580933,
      "learning_rate": 1.5142438014460313e-05,
      "loss": 2.1179,
      "step": 207002
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.4878427982330322,
      "learning_rate": 1.5141154869455164e-05,
      "loss": 1.8804,
      "step": 207003
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.3339946269989014,
      "learning_rate": 1.5139871777410684e-05,
      "loss": 2.0557,
      "step": 207004
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.530632495880127,
      "learning_rate": 1.5138588738327107e-05,
      "loss": 2.0989,
      "step": 207005
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.2831425666809082,
      "learning_rate": 1.5137305752204631e-05,
      "loss": 1.9725,
      "step": 207006
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.4352673292160034,
      "learning_rate": 1.5136022819043459e-05,
      "loss": 1.9234,
      "step": 207007
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.4581201076507568,
      "learning_rate": 1.513473993884392e-05,
      "loss": 1.9923,
      "step": 207008
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.4341946840286255,
      "learning_rate": 1.5133457111606185e-05,
      "loss": 1.9489,
      "step": 207009
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.5439496040344238,
      "learning_rate": 1.5132174337330549e-05,
      "loss": 1.9661,
      "step": 207010
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.6227059364318848,
      "learning_rate": 1.5130891616017216e-05,
      "loss": 1.9576,
      "step": 207011
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.574495553970337,
      "learning_rate": 1.5129608947666416e-05,
      "loss": 2.0905,
      "step": 207012
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.493759036064148,
      "learning_rate": 1.5128326332278384e-05,
      "loss": 1.9312,
      "step": 207013
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.7888661623001099,
      "learning_rate": 1.5127043769853386e-05,
      "loss": 2.0095,
      "step": 207014
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.672207236289978,
      "learning_rate": 1.5125761260391623e-05,
      "loss": 2.135,
      "step": 207015
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.5312384366989136,
      "learning_rate": 1.5124478803893426e-05,
      "loss": 1.9401,
      "step": 207016
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.427565574645996,
      "learning_rate": 1.5123196400358928e-05,
      "loss": 2.0052,
      "step": 207017
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.6759499311447144,
      "learning_rate": 1.5121914049788365e-05,
      "loss": 1.8848,
      "step": 207018
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.4394441843032837,
      "learning_rate": 1.5120631752182066e-05,
      "loss": 1.928,
      "step": 207019
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.554162621498108,
      "learning_rate": 1.5119349507540235e-05,
      "loss": 1.9209,
      "step": 207020
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.565757155418396,
      "learning_rate": 1.5118067315863036e-05,
      "loss": 1.9606,
      "step": 207021
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.3465136289596558,
      "learning_rate": 1.5116785177150803e-05,
      "loss": 2.2252,
      "step": 207022
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.4091951847076416,
      "learning_rate": 1.5115503091403736e-05,
      "loss": 1.7979,
      "step": 207023
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.528313159942627,
      "learning_rate": 1.5114221058622033e-05,
      "loss": 1.9876,
      "step": 207024
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.3731731176376343,
      "learning_rate": 1.511293907880603e-05,
      "loss": 2.0475,
      "step": 207025
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.417459487915039,
      "learning_rate": 1.5111657151955858e-05,
      "loss": 2.2111,
      "step": 207026
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.4872381687164307,
      "learning_rate": 1.5110375278071852e-05,
      "loss": 2.1245,
      "step": 207027
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.4131662845611572,
      "learning_rate": 1.5109093457154208e-05,
      "loss": 2.1152,
      "step": 207028
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.5625797510147095,
      "learning_rate": 1.5107811689203165e-05,
      "loss": 1.9155,
      "step": 207029
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.4285825490951538,
      "learning_rate": 1.5106529974218918e-05,
      "loss": 2.0449,
      "step": 207030
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.488750696182251,
      "learning_rate": 1.5105248312201767e-05,
      "loss": 2.0172,
      "step": 207031
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.4973946809768677,
      "learning_rate": 1.5103966703151882e-05,
      "loss": 1.8353,
      "step": 207032
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.505662202835083,
      "learning_rate": 1.5102685147069593e-05,
      "loss": 1.9607,
      "step": 207033
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.4253556728363037,
      "learning_rate": 1.5101403643955135e-05,
      "loss": 1.9581,
      "step": 207034
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.5145021677017212,
      "learning_rate": 1.510012219380864e-05,
      "loss": 2.1915,
      "step": 207035
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.4089632034301758,
      "learning_rate": 1.5098840796630407e-05,
      "loss": 1.9176,
      "step": 207036
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.3332618474960327,
      "learning_rate": 1.5097559452420704e-05,
      "loss": 1.9371,
      "step": 207037
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.4225496053695679,
      "learning_rate": 1.509627816117973e-05,
      "loss": 1.7472,
      "step": 207038
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.4498368501663208,
      "learning_rate": 1.5094996922907754e-05,
      "loss": 2.0276,
      "step": 207039
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.44891357421875,
      "learning_rate": 1.5093715737605005e-05,
      "loss": 2.1473,
      "step": 207040
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.4267501831054688,
      "learning_rate": 1.5092434605271686e-05,
      "loss": 1.8013,
      "step": 207041
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.3623477220535278,
      "learning_rate": 1.5091153525908061e-05,
      "loss": 2.001,
      "step": 207042
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.4461731910705566,
      "learning_rate": 1.5089872499514366e-05,
      "loss": 2.0513,
      "step": 207043
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.3221129179000854,
      "learning_rate": 1.5088591526090832e-05,
      "loss": 1.9223,
      "step": 207044
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.4800602197647095,
      "learning_rate": 1.5087310605637759e-05,
      "loss": 2.0403,
      "step": 207045
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.616774082183838,
      "learning_rate": 1.5086029738155314e-05,
      "loss": 1.9284,
      "step": 207046
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.4908488988876343,
      "learning_rate": 1.5084748923643731e-05,
      "loss": 1.9972,
      "step": 207047
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.5274320840835571,
      "learning_rate": 1.5083468162103274e-05,
      "loss": 1.9204,
      "step": 207048
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.5203571319580078,
      "learning_rate": 1.5082187453534178e-05,
      "loss": 1.9123,
      "step": 207049
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.4892657995224,
      "learning_rate": 1.5080906797936676e-05,
      "loss": 1.7478,
      "step": 207050
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.4643117189407349,
      "learning_rate": 1.5079626195311034e-05,
      "loss": 1.9385,
      "step": 207051
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.3960700035095215,
      "learning_rate": 1.5078345645657419e-05,
      "loss": 2.1288,
      "step": 207052
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.538015604019165,
      "learning_rate": 1.507706514897623e-05,
      "loss": 1.9487,
      "step": 207053
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.4370096921920776,
      "learning_rate": 1.5075784705267469e-05,
      "loss": 1.9571,
      "step": 207054
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.4418530464172363,
      "learning_rate": 1.5074504314531565e-05,
      "loss": 2.1854,
      "step": 207055
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.3496407270431519,
      "learning_rate": 1.5073223976768655e-05,
      "loss": 2.0512,
      "step": 207056
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.4620375633239746,
      "learning_rate": 1.5071943691979038e-05,
      "loss": 2.0986,
      "step": 207057
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.4843403100967407,
      "learning_rate": 1.507066346016288e-05,
      "loss": 1.8556,
      "step": 207058
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.4141820669174194,
      "learning_rate": 1.5069383281320579e-05,
      "loss": 2.2427,
      "step": 207059
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.604134202003479,
      "learning_rate": 1.5068103155452139e-05,
      "loss": 1.9495,
      "step": 207060
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.4606037139892578,
      "learning_rate": 1.5066823082557988e-05,
      "loss": 1.8262,
      "step": 207061
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.5158978700637817,
      "learning_rate": 1.506554306263823e-05,
      "loss": 2.0071,
      "step": 207062
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.3528748750686646,
      "learning_rate": 1.5064263095693196e-05,
      "loss": 2.0328,
      "step": 207063
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.5032877922058105,
      "learning_rate": 1.5062983181723087e-05,
      "loss": 1.578,
      "step": 207064
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.4182699918746948,
      "learning_rate": 1.5061703320728202e-05,
      "loss": 1.8537,
      "step": 207065
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.3967249393463135,
      "learning_rate": 1.5060423512708642e-05,
      "loss": 2.1235,
      "step": 207066
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.4436124563217163,
      "learning_rate": 1.5059143757664804e-05,
      "loss": 1.8069,
      "step": 207067
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.4993878602981567,
      "learning_rate": 1.5057864055596791e-05,
      "loss": 2.2792,
      "step": 207068
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.5187844038009644,
      "learning_rate": 1.5056584406504935e-05,
      "loss": 2.1246,
      "step": 207069
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.4956717491149902,
      "learning_rate": 1.5055304810389367e-05,
      "loss": 1.9838,
      "step": 207070
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.8902356624603271,
      "learning_rate": 1.5054025267250525e-05,
      "loss": 2.0629,
      "step": 207071
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.4724122285842896,
      "learning_rate": 1.5052745777088404e-05,
      "loss": 2.0412,
      "step": 207072
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.3645515441894531,
      "learning_rate": 1.505146633990344e-05,
      "loss": 2.0723,
      "step": 207073
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.4082045555114746,
      "learning_rate": 1.5050186955695698e-05,
      "loss": 2.0179,
      "step": 207074
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.4768023490905762,
      "learning_rate": 1.5048907624465579e-05,
      "loss": 1.9854,
      "step": 207075
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.361147403717041,
      "learning_rate": 1.504762834621318e-05,
      "loss": 1.8451,
      "step": 207076
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.4550621509552002,
      "learning_rate": 1.5046349120938906e-05,
      "loss": 2.0631,
      "step": 207077
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.5709108114242554,
      "learning_rate": 1.5045069948642818e-05,
      "loss": 1.9343,
      "step": 207078
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.4929823875427246,
      "learning_rate": 1.5043790829325253e-05,
      "loss": 1.9884,
      "step": 207079
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.4525409936904907,
      "learning_rate": 1.5042511762986408e-05,
      "loss": 1.9257,
      "step": 207080
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.5019985437393188,
      "learning_rate": 1.5041232749626586e-05,
      "loss": 2.0593,
      "step": 207081
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.5131548643112183,
      "learning_rate": 1.5039953789245918e-05,
      "loss": 2.1499,
      "step": 207082
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.54965341091156,
      "learning_rate": 1.503867488184477e-05,
      "loss": 2.0115,
      "step": 207083
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.5127207040786743,
      "learning_rate": 1.5037396027423242e-05,
      "loss": 2.1494,
      "step": 207084
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.520790934562683,
      "learning_rate": 1.503611722598167e-05,
      "loss": 2.2256,
      "step": 207085
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.5080912113189697,
      "learning_rate": 1.503483847752025e-05,
      "loss": 2.0464,
      "step": 207086
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.3707144260406494,
      "learning_rate": 1.5033559782039251e-05,
      "loss": 1.8445,
      "step": 207087
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.4446355104446411,
      "learning_rate": 1.5032281139538871e-05,
      "loss": 2.0039,
      "step": 207088
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.5594415664672852,
      "learning_rate": 1.5031002550019444e-05,
      "loss": 2.1045,
      "step": 207089
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.5404640436172485,
      "learning_rate": 1.5029724013481038e-05,
      "loss": 1.9759,
      "step": 207090
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.352782130241394,
      "learning_rate": 1.5028445529924049e-05,
      "loss": 1.7486,
      "step": 207091
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.6354401111602783,
      "learning_rate": 1.5027167099348581e-05,
      "loss": 2.0232,
      "step": 207092
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.590031385421753,
      "learning_rate": 1.5025888721754997e-05,
      "loss": 2.0976,
      "step": 207093
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.4597516059875488,
      "learning_rate": 1.5024610397143433e-05,
      "loss": 1.9371,
      "step": 207094
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.360579490661621,
      "learning_rate": 1.502333212551422e-05,
      "loss": 1.9004,
      "step": 207095
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.4750652313232422,
      "learning_rate": 1.5022053906867559e-05,
      "loss": 2.1704,
      "step": 207096
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.4496716260910034,
      "learning_rate": 1.5020775741203684e-05,
      "loss": 2.2995,
      "step": 207097
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.4564012289047241,
      "learning_rate": 1.501949762852276e-05,
      "loss": 2.073,
      "step": 207098
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.4822657108306885,
      "learning_rate": 1.5018219568825118e-05,
      "loss": 1.9718,
      "step": 207099
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.527603268623352,
      "learning_rate": 1.5016941562110963e-05,
      "loss": 2.2454,
      "step": 207100
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.3894089460372925,
      "learning_rate": 1.5015663608380557e-05,
      "loss": 2.0785,
      "step": 207101
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.4479036331176758,
      "learning_rate": 1.5014385707634135e-05,
      "loss": 1.8467,
      "step": 207102
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.4119716882705688,
      "learning_rate": 1.5013107859871864e-05,
      "loss": 1.8672,
      "step": 207103
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.5302631855010986,
      "learning_rate": 1.501183006509411e-05,
      "loss": 2.1768,
      "step": 207104
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.6924957036972046,
      "learning_rate": 1.5010552323301006e-05,
      "loss": 1.9711,
      "step": 207105
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.4775350093841553,
      "learning_rate": 1.5009274634492785e-05,
      "loss": 2.1859,
      "step": 207106
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.534555435180664,
      "learning_rate": 1.5007996998669746e-05,
      "loss": 2.1086,
      "step": 207107
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.4536075592041016,
      "learning_rate": 1.5006719415832125e-05,
      "loss": 1.8651,
      "step": 207108
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.3659780025482178,
      "learning_rate": 1.5005441885980085e-05,
      "loss": 1.9932,
      "step": 207109
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.3701550960540771,
      "learning_rate": 1.5004164409113928e-05,
      "loss": 1.9281,
      "step": 207110
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.4436908960342407,
      "learning_rate": 1.500288698523392e-05,
      "loss": 1.8745,
      "step": 207111
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.4814590215682983,
      "learning_rate": 1.5001609614340193e-05,
      "loss": 1.9654,
      "step": 207112
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.3789578676223755,
      "learning_rate": 1.5000332296433082e-05,
      "loss": 1.8539,
      "step": 207113
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.553499460220337,
      "learning_rate": 1.4999055031512785e-05,
      "loss": 1.9434,
      "step": 207114
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.6092782020568848,
      "learning_rate": 1.4997777819579504e-05,
      "loss": 1.8814,
      "step": 207115
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.4234099388122559,
      "learning_rate": 1.499650066063357e-05,
      "loss": 1.837,
      "step": 207116
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.4698396921157837,
      "learning_rate": 1.4995223554675117e-05,
      "loss": 1.8837,
      "step": 207117
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.3483035564422607,
      "learning_rate": 1.4993946501704479e-05,
      "loss": 2.1238,
      "step": 207118
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.4054224491119385,
      "learning_rate": 1.4992669501721855e-05,
      "loss": 2.0444,
      "step": 207119
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.496691107749939,
      "learning_rate": 1.4991392554727444e-05,
      "loss": 2.0936,
      "step": 207120
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.4674746990203857,
      "learning_rate": 1.499011566072148e-05,
      "loss": 1.9292,
      "step": 207121
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.4945390224456787,
      "learning_rate": 1.4988838819704297e-05,
      "loss": 1.8836,
      "step": 207122
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.3632208108901978,
      "learning_rate": 1.4987562031676027e-05,
      "loss": 2.1605,
      "step": 207123
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.4984465837478638,
      "learning_rate": 1.4986285296636969e-05,
      "loss": 1.984,
      "step": 207124
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.4608937501907349,
      "learning_rate": 1.4985008614587291e-05,
      "loss": 1.7926,
      "step": 207125
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.5674846172332764,
      "learning_rate": 1.4983731985527391e-05,
      "loss": 1.8452,
      "step": 207126
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.3719507455825806,
      "learning_rate": 1.4982455409457305e-05,
      "loss": 1.8848,
      "step": 207127
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.4704800844192505,
      "learning_rate": 1.4981178886377399e-05,
      "loss": 1.8049,
      "step": 207128
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.6145765781402588,
      "learning_rate": 1.4979902416287836e-05,
      "loss": 1.9255,
      "step": 207129
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.392983317375183,
      "learning_rate": 1.497862599918892e-05,
      "loss": 2.0876,
      "step": 207130
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.455243468284607,
      "learning_rate": 1.4977349635080815e-05,
      "loss": 1.8106,
      "step": 207131
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.4712821245193481,
      "learning_rate": 1.4976073323963888e-05,
      "loss": 2.2063,
      "step": 207132
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.5327391624450684,
      "learning_rate": 1.4974797065838207e-05,
      "loss": 1.9961,
      "step": 207133
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.4291971921920776,
      "learning_rate": 1.4973520860704137e-05,
      "loss": 2.0585,
      "step": 207134
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.5538443326950073,
      "learning_rate": 1.4972244708561843e-05,
      "loss": 2.1181,
      "step": 207135
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.5455608367919922,
      "learning_rate": 1.4970968609411593e-05,
      "loss": 2.1224,
      "step": 207136
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.3872108459472656,
      "learning_rate": 1.4969692563253621e-05,
      "loss": 1.8269,
      "step": 207137
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.5691633224487305,
      "learning_rate": 1.4968416570088226e-05,
      "loss": 2.1454,
      "step": 207138
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.557584524154663,
      "learning_rate": 1.4967140629915474e-05,
      "loss": 2.136,
      "step": 207139
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.4367496967315674,
      "learning_rate": 1.49658647427358e-05,
      "loss": 1.9526,
      "step": 207140
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.4612151384353638,
      "learning_rate": 1.4964588908549269e-05,
      "loss": 2.0011,
      "step": 207141
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.4233918190002441,
      "learning_rate": 1.496331312735628e-05,
      "loss": 2.0805,
      "step": 207142
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.4929248094558716,
      "learning_rate": 1.4962037399156934e-05,
      "loss": 2.0361,
      "step": 207143
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.4019906520843506,
      "learning_rate": 1.4960761723951597e-05,
      "loss": 1.9471,
      "step": 207144
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.449825406074524,
      "learning_rate": 1.495948610174037e-05,
      "loss": 2.0608,
      "step": 207145
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.5959988832473755,
      "learning_rate": 1.4958210532523584e-05,
      "loss": 1.956,
      "step": 207146
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.3902337551116943,
      "learning_rate": 1.4956935016301408e-05,
      "loss": 1.9448,
      "step": 207147
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.5168769359588623,
      "learning_rate": 1.495565955307414e-05,
      "loss": 2.1183,
      "step": 207148
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.4725935459136963,
      "learning_rate": 1.495438414284198e-05,
      "loss": 1.8625,
      "step": 207149
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.4778823852539062,
      "learning_rate": 1.4953108785605262e-05,
      "loss": 1.8333,
      "step": 207150
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.5020437240600586,
      "learning_rate": 1.495183348136405e-05,
      "loss": 1.955,
      "step": 207151
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.480629324913025,
      "learning_rate": 1.4950558230118715e-05,
      "loss": 2.0735,
      "step": 207152
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.407288670539856,
      "learning_rate": 1.494928303186942e-05,
      "loss": 1.9784,
      "step": 207153
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.484062910079956,
      "learning_rate": 1.4948007886616498e-05,
      "loss": 2.3533,
      "step": 207154
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.467635989189148,
      "learning_rate": 1.494673279436005e-05,
      "loss": 2.093,
      "step": 207155
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.5160962343215942,
      "learning_rate": 1.4945457755100443e-05,
      "loss": 2.0887,
      "step": 207156
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.5065524578094482,
      "learning_rate": 1.4944182768837809e-05,
      "loss": 2.1671,
      "step": 207157
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.4391921758651733,
      "learning_rate": 1.4942907835572481e-05,
      "loss": 2.0608,
      "step": 207158
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.479274034500122,
      "learning_rate": 1.4941632955304595e-05,
      "loss": 1.8072,
      "step": 207159
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.3698831796646118,
      "learning_rate": 1.494035812803448e-05,
      "loss": 1.9844,
      "step": 207160
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.4605116844177246,
      "learning_rate": 1.4939083353762271e-05,
      "loss": 2.0184,
      "step": 207161
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.553832769393921,
      "learning_rate": 1.4937808632488368e-05,
      "loss": 2.031,
      "step": 207162
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.4918311834335327,
      "learning_rate": 1.4936533964212837e-05,
      "loss": 2.0685,
      "step": 207163
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.5064892768859863,
      "learning_rate": 1.4935259348936012e-05,
      "loss": 1.9397,
      "step": 207164
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.5272883176803589,
      "learning_rate": 1.493398478665806e-05,
      "loss": 1.8173,
      "step": 207165
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.3823492527008057,
      "learning_rate": 1.4932710277379312e-05,
      "loss": 2.0953,
      "step": 207166
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.5649110078811646,
      "learning_rate": 1.4931435821099902e-05,
      "loss": 1.9346,
      "step": 207167
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.5404804944992065,
      "learning_rate": 1.493016141782023e-05,
      "loss": 1.9797,
      "step": 207168
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.4897165298461914,
      "learning_rate": 1.4928887067540296e-05,
      "loss": 2.1679,
      "step": 207169
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.519135594367981,
      "learning_rate": 1.4927612770260532e-05,
      "loss": 2.0238,
      "step": 207170
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.427883267402649,
      "learning_rate": 1.4926338525981074e-05,
      "loss": 1.8864,
      "step": 207171
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.3949153423309326,
      "learning_rate": 1.492506433470222e-05,
      "loss": 1.7751,
      "step": 207172
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.4719140529632568,
      "learning_rate": 1.4923790196424135e-05,
      "loss": 2.1005,
      "step": 207173
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.455743670463562,
      "learning_rate": 1.4922516111147154e-05,
      "loss": 2.082,
      "step": 207174
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.487442970275879,
      "learning_rate": 1.4921242078871409e-05,
      "loss": 2.1003,
      "step": 207175
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.4075806140899658,
      "learning_rate": 1.4919968099597234e-05,
      "loss": 1.9655,
      "step": 207176
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.4633842706680298,
      "learning_rate": 1.4918694173324763e-05,
      "loss": 1.9725,
      "step": 207177
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.3840597867965698,
      "learning_rate": 1.4917420300054329e-05,
      "loss": 2.0501,
      "step": 207178
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.4357085227966309,
      "learning_rate": 1.4916146479786096e-05,
      "loss": 2.0069,
      "step": 207179
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.4081379175186157,
      "learning_rate": 1.4914872712520365e-05,
      "loss": 2.1791,
      "step": 207180
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.4681068658828735,
      "learning_rate": 1.4913598998257336e-05,
      "loss": 1.8493,
      "step": 207181
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.6581394672393799,
      "learning_rate": 1.4912325336997243e-05,
      "loss": 2.0257,
      "step": 207182
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.350895881652832,
      "learning_rate": 1.491105172874032e-05,
      "loss": 1.8273,
      "step": 207183
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.4601022005081177,
      "learning_rate": 1.490977817348683e-05,
      "loss": 2.1954,
      "step": 207184
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.3345612287521362,
      "learning_rate": 1.4908504671236943e-05,
      "loss": 1.9114,
      "step": 207185
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.4751430749893188,
      "learning_rate": 1.4907231221991023e-05,
      "loss": 2.0686,
      "step": 207186
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.476078987121582,
      "learning_rate": 1.4905957825749205e-05,
      "loss": 1.8737,
      "step": 207187
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.5685268640518188,
      "learning_rate": 1.490468448251172e-05,
      "loss": 1.8647,
      "step": 207188
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.650628685951233,
      "learning_rate": 1.490341119227887e-05,
      "loss": 2.0819,
      "step": 207189
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.3568408489227295,
      "learning_rate": 1.4902137955050853e-05,
      "loss": 2.0809,
      "step": 207190
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.4398179054260254,
      "learning_rate": 1.490086477082787e-05,
      "loss": 1.9191,
      "step": 207191
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.491234540939331,
      "learning_rate": 1.4899591639610253e-05,
      "loss": 2.0769,
      "step": 207192
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.4646207094192505,
      "learning_rate": 1.489831856139817e-05,
      "loss": 2.0166,
      "step": 207193
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.3501322269439697,
      "learning_rate": 1.4897045536191854e-05,
      "loss": 1.8522,
      "step": 207194
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.5349401235580444,
      "learning_rate": 1.4895772563991571e-05,
      "loss": 1.9354,
      "step": 207195
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.3761570453643799,
      "learning_rate": 1.4894499644797553e-05,
      "loss": 1.8847,
      "step": 207196
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.463603138923645,
      "learning_rate": 1.4893226778610001e-05,
      "loss": 1.9947,
      "step": 207197
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.531738519668579,
      "learning_rate": 1.4891953965429216e-05,
      "loss": 1.8905,
      "step": 207198
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.5198270082473755,
      "learning_rate": 1.4890681205255395e-05,
      "loss": 2.1252,
      "step": 207199
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.331116795539856,
      "learning_rate": 1.488940849808874e-05,
      "loss": 1.9959,
      "step": 207200
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.4430252313613892,
      "learning_rate": 1.4888135843929549e-05,
      "loss": 1.9224,
      "step": 207201
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.6839057207107544,
      "learning_rate": 1.4886863242778025e-05,
      "loss": 1.8939,
      "step": 207202
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.599955677986145,
      "learning_rate": 1.488559069463443e-05,
      "loss": 1.8608,
      "step": 207203
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.4420253038406372,
      "learning_rate": 1.4884318199499002e-05,
      "loss": 1.8317,
      "step": 207204
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.3941997289657593,
      "learning_rate": 1.488304575737197e-05,
      "loss": 1.9813,
      "step": 207205
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.3176833391189575,
      "learning_rate": 1.4881773368253503e-05,
      "loss": 1.9234,
      "step": 207206
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.3355885744094849,
      "learning_rate": 1.4880501032143932e-05,
      "loss": 2.1182,
      "step": 207207
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.3467634916305542,
      "learning_rate": 1.4879228749043425e-05,
      "loss": 2.0111,
      "step": 207208
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.5208379030227661,
      "learning_rate": 1.4877956518952282e-05,
      "loss": 2.0543,
      "step": 207209
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.4939498901367188,
      "learning_rate": 1.4876684341870736e-05,
      "loss": 2.1137,
      "step": 207210
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.4794139862060547,
      "learning_rate": 1.4875412217798955e-05,
      "loss": 2.2324,
      "step": 207211
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.4786497354507446,
      "learning_rate": 1.4874140146737201e-05,
      "loss": 1.9174,
      "step": 207212
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.6790847778320312,
      "learning_rate": 1.4872868128685777e-05,
      "loss": 2.1293,
      "step": 207213
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.349387526512146,
      "learning_rate": 1.4871596163644816e-05,
      "loss": 1.8251,
      "step": 207214
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.3728258609771729,
      "learning_rate": 1.4870324251614652e-05,
      "loss": 2.0761,
      "step": 207215
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.4199544191360474,
      "learning_rate": 1.4869052392595448e-05,
      "loss": 1.9275,
      "step": 207216
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.5235819816589355,
      "learning_rate": 1.4867780586587542e-05,
      "loss": 1.7969,
      "step": 207217
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.4068523645401,
      "learning_rate": 1.4866508833590996e-05,
      "loss": 2.1726,
      "step": 207218
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.475881814956665,
      "learning_rate": 1.486523713360621e-05,
      "loss": 1.9323,
      "step": 207219
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.3847088813781738,
      "learning_rate": 1.4863965486633323e-05,
      "loss": 1.7874,
      "step": 207220
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.668988823890686,
      "learning_rate": 1.4862693892672627e-05,
      "loss": 2.1199,
      "step": 207221
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.347875714302063,
      "learning_rate": 1.4861422351724328e-05,
      "loss": 2.057,
      "step": 207222
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.5986547470092773,
      "learning_rate": 1.4860150863788723e-05,
      "loss": 2.1659,
      "step": 207223
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.4650943279266357,
      "learning_rate": 1.4858879428865944e-05,
      "loss": 1.8807,
      "step": 207224
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.3780144453048706,
      "learning_rate": 1.4857608046956293e-05,
      "loss": 1.7983,
      "step": 207225
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.548582673072815,
      "learning_rate": 1.4856336718059969e-05,
      "loss": 1.9411,
      "step": 207226
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.43638277053833,
      "learning_rate": 1.4855065442177272e-05,
      "loss": 1.9262,
      "step": 207227
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.4433234930038452,
      "learning_rate": 1.4853794219308369e-05,
      "loss": 2.0658,
      "step": 207228
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.4864224195480347,
      "learning_rate": 1.4852523049453624e-05,
      "loss": 1.7699,
      "step": 207229
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.4342350959777832,
      "learning_rate": 1.4851251932613072e-05,
      "loss": 1.8946,
      "step": 207230
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.4165561199188232,
      "learning_rate": 1.4849980868787081e-05,
      "loss": 1.9065,
      "step": 207231
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.507703185081482,
      "learning_rate": 1.484870985797585e-05,
      "loss": 2.1441,
      "step": 207232
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.5859527587890625,
      "learning_rate": 1.4847438900179642e-05,
      "loss": 1.94,
      "step": 207233
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.366786241531372,
      "learning_rate": 1.484616799539866e-05,
      "loss": 2.0783,
      "step": 207234
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.514817476272583,
      "learning_rate": 1.4844897143633238e-05,
      "loss": 2.1805,
      "step": 207235
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.618361234664917,
      "learning_rate": 1.4843626344883441e-05,
      "loss": 1.926,
      "step": 207236
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.406677484512329,
      "learning_rate": 1.4842355599149635e-05,
      "loss": 1.9416,
      "step": 207237
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.5220431089401245,
      "learning_rate": 1.4841084906431989e-05,
      "loss": 2.0096,
      "step": 207238
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.604341745376587,
      "learning_rate": 1.4839814266730798e-05,
      "loss": 1.9608,
      "step": 207239
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.318652629852295,
      "learning_rate": 1.4838543680046234e-05,
      "loss": 1.8797,
      "step": 207240
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.5209239721298218,
      "learning_rate": 1.483727314637866e-05,
      "loss": 1.8035,
      "step": 207241
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.4497559070587158,
      "learning_rate": 1.4836002665728108e-05,
      "loss": 1.8159,
      "step": 207242
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.4542170763015747,
      "learning_rate": 1.4834732238094981e-05,
      "loss": 2.0449,
      "step": 207243
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.4425501823425293,
      "learning_rate": 1.4833461863479445e-05,
      "loss": 2.2009,
      "step": 207244
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.5640203952789307,
      "learning_rate": 1.4832191541881766e-05,
      "loss": 1.8547,
      "step": 207245
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.514662265777588,
      "learning_rate": 1.4830921273302143e-05,
      "loss": 1.8502,
      "step": 207246
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.5202245712280273,
      "learning_rate": 1.4829651057740876e-05,
      "loss": 2.166,
      "step": 207247
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.5215332508087158,
      "learning_rate": 1.4828380895198099e-05,
      "loss": 2.0435,
      "step": 207248
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.562281608581543,
      "learning_rate": 1.4827110785674178e-05,
      "loss": 1.9987,
      "step": 207249
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.5031193494796753,
      "learning_rate": 1.4825840729169215e-05,
      "loss": 1.9318,
      "step": 207250
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.4535350799560547,
      "learning_rate": 1.4824570725683538e-05,
      "loss": 1.9762,
      "step": 207251
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.4860137701034546,
      "learning_rate": 1.4823300775217351e-05,
      "loss": 2.1463,
      "step": 207252
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.4120718240737915,
      "learning_rate": 1.482203087777092e-05,
      "loss": 2.0886,
      "step": 207253
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.4913909435272217,
      "learning_rate": 1.4820761033344442e-05,
      "loss": 1.9449,
      "step": 207254
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.6170623302459717,
      "learning_rate": 1.4819491241938186e-05,
      "loss": 2.0856,
      "step": 207255
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.4846383333206177,
      "learning_rate": 1.481822150355232e-05,
      "loss": 1.999,
      "step": 207256
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.4126101732254028,
      "learning_rate": 1.4816951818187172e-05,
      "loss": 2.093,
      "step": 207257
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.4968860149383545,
      "learning_rate": 1.4815682185842881e-05,
      "loss": 1.9259,
      "step": 207258
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.536194086074829,
      "learning_rate": 1.481441260651981e-05,
      "loss": 1.9463,
      "step": 207259
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.4102306365966797,
      "learning_rate": 1.4813143080218092e-05,
      "loss": 2.0625,
      "step": 207260
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.4553494453430176,
      "learning_rate": 1.481187360693803e-05,
      "loss": 1.9938,
      "step": 207261
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.5218359231948853,
      "learning_rate": 1.4810604186679754e-05,
      "loss": 1.9129,
      "step": 207262
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.4990283250808716,
      "learning_rate": 1.480933481944363e-05,
      "loss": 2.02,
      "step": 207263
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.4910361766815186,
      "learning_rate": 1.480806550522976e-05,
      "loss": 2.3537,
      "step": 207264
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.474937915802002,
      "learning_rate": 1.4806796244038511e-05,
      "loss": 2.1215,
      "step": 207265
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.505483627319336,
      "learning_rate": 1.480552703587008e-05,
      "loss": 2.0433,
      "step": 207266
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.3065576553344727,
      "learning_rate": 1.4804257880724668e-05,
      "loss": 2.1315,
      "step": 207267
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.403890609741211,
      "learning_rate": 1.480298877860251e-05,
      "loss": 1.85,
      "step": 207268
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.3210397958755493,
      "learning_rate": 1.480171972950387e-05,
      "loss": 1.9084,
      "step": 207269
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.380611538887024,
      "learning_rate": 1.4800450733428948e-05,
      "loss": 2.0471,
      "step": 207270
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.3347289562225342,
      "learning_rate": 1.4799181790378046e-05,
      "loss": 1.7461,
      "step": 207271
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.2789466381072998,
      "learning_rate": 1.4797912900351361e-05,
      "loss": 2.1613,
      "step": 207272
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.4265562295913696,
      "learning_rate": 1.4796644063349129e-05,
      "loss": 2.118,
      "step": 207273
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.3504995107650757,
      "learning_rate": 1.4795375279371513e-05,
      "loss": 1.8924,
      "step": 207274
    },
    {
      "epoch": 2.7,
      "grad_norm": 2.12540864944458,
      "learning_rate": 1.4794106548418916e-05,
      "loss": 1.6734,
      "step": 207275
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.5412724018096924,
      "learning_rate": 1.4792837870491402e-05,
      "loss": 2.1453,
      "step": 207276
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.4344626665115356,
      "learning_rate": 1.4791569245589308e-05,
      "loss": 1.9424,
      "step": 207277
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.4294402599334717,
      "learning_rate": 1.4790300673712896e-05,
      "loss": 2.0181,
      "step": 207278
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.429877758026123,
      "learning_rate": 1.4789032154862267e-05,
      "loss": 1.7684,
      "step": 207279
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.6017093658447266,
      "learning_rate": 1.478776368903779e-05,
      "loss": 2.1339,
      "step": 207280
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.5759358406066895,
      "learning_rate": 1.4786495276239663e-05,
      "loss": 1.7486,
      "step": 207281
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.4210466146469116,
      "learning_rate": 1.4785226916468051e-05,
      "loss": 1.9938,
      "step": 207282
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.4771113395690918,
      "learning_rate": 1.4783958609723322e-05,
      "loss": 1.9984,
      "step": 207283
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.431320309638977,
      "learning_rate": 1.4782690356005611e-05,
      "loss": 1.9673,
      "step": 207284
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.8077914714813232,
      "learning_rate": 1.4781422155315148e-05,
      "loss": 2.0128,
      "step": 207285
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.447370171546936,
      "learning_rate": 1.4780154007652267e-05,
      "loss": 2.0663,
      "step": 207286
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.453869342803955,
      "learning_rate": 1.4778885913017069e-05,
      "loss": 2.188,
      "step": 207287
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.4001693725585938,
      "learning_rate": 1.4777617871409919e-05,
      "loss": 1.8905,
      "step": 207288
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.5220953226089478,
      "learning_rate": 1.4776349882830984e-05,
      "loss": 1.8537,
      "step": 207289
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.3661653995513916,
      "learning_rate": 1.4775081947280498e-05,
      "loss": 1.9802,
      "step": 207290
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.3293066024780273,
      "learning_rate": 1.477381406475866e-05,
      "loss": 2.0326,
      "step": 207291
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.559146523475647,
      "learning_rate": 1.4772546235265836e-05,
      "loss": 1.9871,
      "step": 207292
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.2922841310501099,
      "learning_rate": 1.4771278458802094e-05,
      "loss": 1.8137,
      "step": 207293
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.4725940227508545,
      "learning_rate": 1.4770010735367831e-05,
      "loss": 1.9432,
      "step": 207294
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.3711962699890137,
      "learning_rate": 1.4768743064963184e-05,
      "loss": 1.9153,
      "step": 207295
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.3311963081359863,
      "learning_rate": 1.476747544758845e-05,
      "loss": 2.2522,
      "step": 207296
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.413702368736267,
      "learning_rate": 1.476620788324373e-05,
      "loss": 1.9318,
      "step": 207297
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.42093026638031,
      "learning_rate": 1.4764940371929458e-05,
      "loss": 1.9433,
      "step": 207298
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.4052002429962158,
      "learning_rate": 1.4763672913645664e-05,
      "loss": 2.1131,
      "step": 207299
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.471231460571289,
      "learning_rate": 1.4762405508392783e-05,
      "loss": 1.8471,
      "step": 207300
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.5425227880477905,
      "learning_rate": 1.4761138156170882e-05,
      "loss": 1.878,
      "step": 207301
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.5177502632141113,
      "learning_rate": 1.475987085698036e-05,
      "loss": 2.0071,
      "step": 207302
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.4634567499160767,
      "learning_rate": 1.4758603610821284e-05,
      "loss": 1.7948,
      "step": 207303
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.5362478494644165,
      "learning_rate": 1.4757336417693988e-05,
      "loss": 2.1145,
      "step": 207304
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.5946711301803589,
      "learning_rate": 1.475606927759867e-05,
      "loss": 2.1276,
      "step": 207305
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.399745225906372,
      "learning_rate": 1.475480219053563e-05,
      "loss": 1.8708,
      "step": 207306
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.3899762630462646,
      "learning_rate": 1.4753535156505003e-05,
      "loss": 1.9731,
      "step": 207307
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.355613112449646,
      "learning_rate": 1.4752268175507153e-05,
      "loss": 1.8122,
      "step": 207308
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.5406650304794312,
      "learning_rate": 1.475100124754215e-05,
      "loss": 2.0509,
      "step": 207309
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.4020709991455078,
      "learning_rate": 1.4749734372610388e-05,
      "loss": 1.8677,
      "step": 207310
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.6204196214675903,
      "learning_rate": 1.4748467550711973e-05,
      "loss": 2.2226,
      "step": 207311
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.4079645872116089,
      "learning_rate": 1.4747200781847269e-05,
      "loss": 1.9401,
      "step": 207312
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.5223162174224854,
      "learning_rate": 1.4745934066016374e-05,
      "loss": 1.8471,
      "step": 207313
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.46535062789917,
      "learning_rate": 1.4744667403219724e-05,
      "loss": 2.0075,
      "step": 207314
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.3034429550170898,
      "learning_rate": 1.4743400793457283e-05,
      "loss": 2.0081,
      "step": 207315
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.4128074645996094,
      "learning_rate": 1.4742134236729519e-05,
      "loss": 2.0664,
      "step": 207316
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.4764779806137085,
      "learning_rate": 1.4740867733036498e-05,
      "loss": 1.9857,
      "step": 207317
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.4596227407455444,
      "learning_rate": 1.4739601282378589e-05,
      "loss": 1.8771,
      "step": 207318
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.533255934715271,
      "learning_rate": 1.4738334884755953e-05,
      "loss": 1.8531,
      "step": 207319
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.4860858917236328,
      "learning_rate": 1.473706854016886e-05,
      "loss": 2.0813,
      "step": 207320
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.4426429271697998,
      "learning_rate": 1.4735802248617546e-05,
      "loss": 1.8571,
      "step": 207321
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.5662803649902344,
      "learning_rate": 1.4734536010102204e-05,
      "loss": 1.9073,
      "step": 207322
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.3944650888442993,
      "learning_rate": 1.4733269824623073e-05,
      "loss": 1.8638,
      "step": 207323
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.5085878372192383,
      "learning_rate": 1.473200369218045e-05,
      "loss": 1.8433,
      "step": 207324
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.4501948356628418,
      "learning_rate": 1.47307376127745e-05,
      "loss": 1.8491,
      "step": 207325
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.3535152673721313,
      "learning_rate": 1.4729471586405528e-05,
      "loss": 1.7948,
      "step": 207326
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.3549771308898926,
      "learning_rate": 1.4728205613073729e-05,
      "loss": 2.0188,
      "step": 207327
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.4759310483932495,
      "learning_rate": 1.4726939692779338e-05,
      "loss": 2.1122,
      "step": 207328
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.406779170036316,
      "learning_rate": 1.4725673825522588e-05,
      "loss": 1.7958,
      "step": 207329
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.5717625617980957,
      "learning_rate": 1.4724408011303712e-05,
      "loss": 2.1978,
      "step": 207330
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.6537625789642334,
      "learning_rate": 1.4723142250122944e-05,
      "loss": 1.9794,
      "step": 207331
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.6539932489395142,
      "learning_rate": 1.4721876541980548e-05,
      "loss": 1.9368,
      "step": 207332
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.4000623226165771,
      "learning_rate": 1.472061088687676e-05,
      "loss": 2.1456,
      "step": 207333
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.4539633989334106,
      "learning_rate": 1.4719345284811779e-05,
      "loss": 1.8496,
      "step": 207334
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.4262861013412476,
      "learning_rate": 1.4718079735785837e-05,
      "loss": 1.9852,
      "step": 207335
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.2908965349197388,
      "learning_rate": 1.4716814239799201e-05,
      "loss": 2.0728,
      "step": 207336
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.550950527191162,
      "learning_rate": 1.4715548796852072e-05,
      "loss": 2.0912,
      "step": 207337
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.4863312244415283,
      "learning_rate": 1.471428340694475e-05,
      "loss": 2.0099,
      "step": 207338
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.5330352783203125,
      "learning_rate": 1.4713018070077432e-05,
      "loss": 1.6959,
      "step": 207339
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.4931907653808594,
      "learning_rate": 1.471175278625032e-05,
      "loss": 1.9573,
      "step": 207340
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.4377987384796143,
      "learning_rate": 1.4710487555463678e-05,
      "loss": 1.9866,
      "step": 207341
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.4535213708877563,
      "learning_rate": 1.4709222377717744e-05,
      "loss": 1.8453,
      "step": 207342
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.6426141262054443,
      "learning_rate": 1.4707957253012747e-05,
      "loss": 2.1107,
      "step": 207343
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.415877103805542,
      "learning_rate": 1.4706692181348923e-05,
      "loss": 2.0205,
      "step": 207344
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.6032333374023438,
      "learning_rate": 1.4705427162726536e-05,
      "loss": 1.844,
      "step": 207345
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.4198849201202393,
      "learning_rate": 1.4704162197145819e-05,
      "loss": 2.3113,
      "step": 207346
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.5112342834472656,
      "learning_rate": 1.4702897284606907e-05,
      "loss": 1.9048,
      "step": 207347
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.5020641088485718,
      "learning_rate": 1.4701632425110165e-05,
      "loss": 1.9957,
      "step": 207348
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.4197968244552612,
      "learning_rate": 1.4700367618655728e-05,
      "loss": 2.2106,
      "step": 207349
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.5345710515975952,
      "learning_rate": 1.4699102865243929e-05,
      "loss": 2.0312,
      "step": 207350
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.5086708068847656,
      "learning_rate": 1.4697838164874964e-05,
      "loss": 1.9918,
      "step": 207351
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.3710527420043945,
      "learning_rate": 1.4696573517549037e-05,
      "loss": 1.7472,
      "step": 207352
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.5121610164642334,
      "learning_rate": 1.4695308923266346e-05,
      "loss": 2.0345,
      "step": 207353
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.4490593671798706,
      "learning_rate": 1.4694044382027259e-05,
      "loss": 2.0054,
      "step": 207354
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.5507147312164307,
      "learning_rate": 1.4692779893831874e-05,
      "loss": 1.7499,
      "step": 207355
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.593643307685852,
      "learning_rate": 1.4691515458680526e-05,
      "loss": 2.1616,
      "step": 207356
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.4315863847732544,
      "learning_rate": 1.4690251076573411e-05,
      "loss": 1.8477,
      "step": 207357
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.4149889945983887,
      "learning_rate": 1.4688986747510767e-05,
      "loss": 1.8704,
      "step": 207358
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.4855226278305054,
      "learning_rate": 1.4687722471492791e-05,
      "loss": 1.9053,
      "step": 207359
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.412920594215393,
      "learning_rate": 1.4686458248519783e-05,
      "loss": 1.7726,
      "step": 207360
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.4030681848526,
      "learning_rate": 1.4685194078591911e-05,
      "loss": 1.9455,
      "step": 207361
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.3457012176513672,
      "learning_rate": 1.4683929961709507e-05,
      "loss": 1.9773,
      "step": 207362
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.5282378196716309,
      "learning_rate": 1.4682665897872736e-05,
      "loss": 1.9164,
      "step": 207363
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.595786452293396,
      "learning_rate": 1.4681401887081767e-05,
      "loss": 1.9607,
      "step": 207364
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.5269697904586792,
      "learning_rate": 1.4680137929336999e-05,
      "loss": 1.956,
      "step": 207365
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.5317325592041016,
      "learning_rate": 1.4678874024638565e-05,
      "loss": 2.1723,
      "step": 207366
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.543540358543396,
      "learning_rate": 1.4677610172986665e-05,
      "loss": 1.9632,
      "step": 207367
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.434353232383728,
      "learning_rate": 1.4676346374381632e-05,
      "loss": 1.9199,
      "step": 207368
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.3702861070632935,
      "learning_rate": 1.4675082628823632e-05,
      "loss": 1.9546,
      "step": 207369
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.5266700983047485,
      "learning_rate": 1.4673818936312897e-05,
      "loss": 2.2243,
      "step": 207370
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.4432886838912964,
      "learning_rate": 1.4672555296849731e-05,
      "loss": 2.1458,
      "step": 207371
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.3833768367767334,
      "learning_rate": 1.4671291710434329e-05,
      "loss": 1.9322,
      "step": 207372
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.4400745630264282,
      "learning_rate": 1.467002817706686e-05,
      "loss": 1.8341,
      "step": 207373
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.54571533203125,
      "learning_rate": 1.466876469674766e-05,
      "loss": 2.0322,
      "step": 207374
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.4005311727523804,
      "learning_rate": 1.4667501269476955e-05,
      "loss": 2.0397,
      "step": 207375
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.3399059772491455,
      "learning_rate": 1.4666237895254851e-05,
      "loss": 1.8119,
      "step": 207376
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.519378423690796,
      "learning_rate": 1.4664974574081777e-05,
      "loss": 1.7641,
      "step": 207377
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.5243366956710815,
      "learning_rate": 1.4663711305957804e-05,
      "loss": 2.0494,
      "step": 207378
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.4455028772354126,
      "learning_rate": 1.466244809088326e-05,
      "loss": 1.9872,
      "step": 207379
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.4624301195144653,
      "learning_rate": 1.4661184928858382e-05,
      "loss": 2.1162,
      "step": 207380
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.6111830472946167,
      "learning_rate": 1.4659921819883336e-05,
      "loss": 1.9001,
      "step": 207381
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.4553025960922241,
      "learning_rate": 1.4658658763958386e-05,
      "loss": 1.7819,
      "step": 207382
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.5411943197250366,
      "learning_rate": 1.4657395761083802e-05,
      "loss": 1.8924,
      "step": 207383
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.5041921138763428,
      "learning_rate": 1.4656132811259779e-05,
      "loss": 2.148,
      "step": 207384
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.304672122001648,
      "learning_rate": 1.4654869914486588e-05,
      "loss": 1.984,
      "step": 207385
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.4724446535110474,
      "learning_rate": 1.4653607070764394e-05,
      "loss": 1.9402,
      "step": 207386
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.641465663909912,
      "learning_rate": 1.4652344280093564e-05,
      "loss": 2.0711,
      "step": 207387
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.5536983013153076,
      "learning_rate": 1.4651081542474197e-05,
      "loss": 1.9229,
      "step": 207388
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.437143325805664,
      "learning_rate": 1.4649818857906625e-05,
      "loss": 1.9156,
      "step": 207389
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.4151502847671509,
      "learning_rate": 1.464855622639095e-05,
      "loss": 1.9875,
      "step": 207390
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.5700421333312988,
      "learning_rate": 1.4647293647927572e-05,
      "loss": 1.9604,
      "step": 207391
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.4851677417755127,
      "learning_rate": 1.4646031122516589e-05,
      "loss": 1.8991,
      "step": 207392
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.512932538986206,
      "learning_rate": 1.4644768650158367e-05,
      "loss": 1.9072,
      "step": 207393
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.444419503211975,
      "learning_rate": 1.4643506230853042e-05,
      "loss": 2.0527,
      "step": 207394
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.5202287435531616,
      "learning_rate": 1.4642243864600878e-05,
      "loss": 2.1737,
      "step": 207395
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.387563705444336,
      "learning_rate": 1.4640981551402076e-05,
      "loss": 2.0612,
      "step": 207396
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.5373457670211792,
      "learning_rate": 1.4639719291256934e-05,
      "loss": 1.8779,
      "step": 207397
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.423334002494812,
      "learning_rate": 1.463845708416559e-05,
      "loss": 2.0783,
      "step": 207398
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.4317317008972168,
      "learning_rate": 1.4637194930128438e-05,
      "loss": 2.046,
      "step": 207399
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.488745093345642,
      "learning_rate": 1.463593282914558e-05,
      "loss": 1.7757,
      "step": 207400
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.4290047883987427,
      "learning_rate": 1.4634670781217283e-05,
      "loss": 2.0013,
      "step": 207401
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.3957663774490356,
      "learning_rate": 1.4633408786343782e-05,
      "loss": 1.9622,
      "step": 207402
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.5219476222991943,
      "learning_rate": 1.463214684452534e-05,
      "loss": 2.0078,
      "step": 207403
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.5208656787872314,
      "learning_rate": 1.4630884955762124e-05,
      "loss": 1.9626,
      "step": 207404
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.3467106819152832,
      "learning_rate": 1.4629623120054434e-05,
      "loss": 1.926,
      "step": 207405
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.5134738683700562,
      "learning_rate": 1.4628361337402505e-05,
      "loss": 2.1928,
      "step": 207406
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.3604938983917236,
      "learning_rate": 1.462709960780657e-05,
      "loss": 1.9474,
      "step": 207407
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.8604518175125122,
      "learning_rate": 1.4625837931266794e-05,
      "loss": 2.0996,
      "step": 207408
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.4145238399505615,
      "learning_rate": 1.4624576307783475e-05,
      "loss": 1.9278,
      "step": 207409
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.4493321180343628,
      "learning_rate": 1.4623314737356817e-05,
      "loss": 2.2957,
      "step": 207410
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.5280821323394775,
      "learning_rate": 1.4622053219987118e-05,
      "loss": 2.1301,
      "step": 207411
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.4734046459197998,
      "learning_rate": 1.4620791755674543e-05,
      "loss": 1.9899,
      "step": 207412
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.4744747877120972,
      "learning_rate": 1.4619530344419361e-05,
      "loss": 1.997,
      "step": 207413
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.4590883255004883,
      "learning_rate": 1.4618268986221737e-05,
      "loss": 1.8363,
      "step": 207414
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.5051193237304688,
      "learning_rate": 1.4617007681082039e-05,
      "loss": 2.0361,
      "step": 207415
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.4632753133773804,
      "learning_rate": 1.4615746429000364e-05,
      "loss": 2.0532,
      "step": 207416
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.4911937713623047,
      "learning_rate": 1.4614485229977046e-05,
      "loss": 2.0825,
      "step": 207417
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.4307520389556885,
      "learning_rate": 1.4613224084012287e-05,
      "loss": 1.8864,
      "step": 207418
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.4528311491012573,
      "learning_rate": 1.4611962991106351e-05,
      "loss": 1.8965,
      "step": 207419
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.522825837135315,
      "learning_rate": 1.461070195125934e-05,
      "loss": 1.9348,
      "step": 207420
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.6334385871887207,
      "learning_rate": 1.4609440964471652e-05,
      "loss": 1.9303,
      "step": 207421
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.4150854349136353,
      "learning_rate": 1.4608180030743422e-05,
      "loss": 2.1298,
      "step": 207422
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.4193272590637207,
      "learning_rate": 1.4606919150074981e-05,
      "loss": 2.0988,
      "step": 207423
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.348766803741455,
      "learning_rate": 1.4605658322466463e-05,
      "loss": 1.8479,
      "step": 207424
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.5265024900436401,
      "learning_rate": 1.4604397547918135e-05,
      "loss": 1.9063,
      "step": 207425
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.3705084323883057,
      "learning_rate": 1.4603136826430229e-05,
      "loss": 2.0103,
      "step": 207426
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.6188520193099976,
      "learning_rate": 1.4601876158003011e-05,
      "loss": 1.9552,
      "step": 207427
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.4676852226257324,
      "learning_rate": 1.4600615542636651e-05,
      "loss": 1.8272,
      "step": 207428
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.5219190120697021,
      "learning_rate": 1.459935498033148e-05,
      "loss": 1.9996,
      "step": 207429
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.3870065212249756,
      "learning_rate": 1.4598094471087662e-05,
      "loss": 1.9373,
      "step": 207430
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.655126929283142,
      "learning_rate": 1.4596834014905434e-05,
      "loss": 2.0614,
      "step": 207431
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.7502429485321045,
      "learning_rate": 1.4595573611785027e-05,
      "loss": 1.8979,
      "step": 207432
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.5491396188735962,
      "learning_rate": 1.4594313261726708e-05,
      "loss": 2.0744,
      "step": 207433
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.5533714294433594,
      "learning_rate": 1.4593052964730645e-05,
      "loss": 1.9507,
      "step": 207434
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.407261848449707,
      "learning_rate": 1.4591792720797202e-05,
      "loss": 2.0839,
      "step": 207435
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.3455642461776733,
      "learning_rate": 1.459053252992648e-05,
      "loss": 2.1699,
      "step": 207436
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.5167689323425293,
      "learning_rate": 1.4589272392118812e-05,
      "loss": 2.1353,
      "step": 207437
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.3193374872207642,
      "learning_rate": 1.458801230737433e-05,
      "loss": 2.1669,
      "step": 207438
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.4669262170791626,
      "learning_rate": 1.4586752275693369e-05,
      "loss": 1.9435,
      "step": 207439
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.4643149375915527,
      "learning_rate": 1.4585492297076062e-05,
      "loss": 2.1245,
      "step": 207440
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.3524284362792969,
      "learning_rate": 1.4584232371522741e-05,
      "loss": 1.9604,
      "step": 207441
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.3814321756362915,
      "learning_rate": 1.4582972499033607e-05,
      "loss": 1.9416,
      "step": 207442
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.4942079782485962,
      "learning_rate": 1.4581712679608892e-05,
      "loss": 1.9113,
      "step": 207443
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.420162558555603,
      "learning_rate": 1.4580452913248763e-05,
      "loss": 2.0955,
      "step": 207444
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.4292523860931396,
      "learning_rate": 1.4579193199953588e-05,
      "loss": 1.9892,
      "step": 207445
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.5622477531433105,
      "learning_rate": 1.4577933539723463e-05,
      "loss": 1.883,
      "step": 207446
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.4101988077163696,
      "learning_rate": 1.4576673932558725e-05,
      "loss": 1.8927,
      "step": 207447
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.383131504058838,
      "learning_rate": 1.4575414378459604e-05,
      "loss": 1.9623,
      "step": 207448
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.361557126045227,
      "learning_rate": 1.4574154877426237e-05,
      "loss": 1.8655,
      "step": 207449
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.5327610969543457,
      "learning_rate": 1.4572895429458952e-05,
      "loss": 1.9943,
      "step": 207450
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.5100702047348022,
      "learning_rate": 1.4571636034557987e-05,
      "loss": 2.2596,
      "step": 207451
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.44923996925354,
      "learning_rate": 1.4570376692723473e-05,
      "loss": 1.8096,
      "step": 207452
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.5506302118301392,
      "learning_rate": 1.4569117403955776e-05,
      "loss": 1.8454,
      "step": 207453
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.601279377937317,
      "learning_rate": 1.4567858168254997e-05,
      "loss": 2.257,
      "step": 207454
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.4117803573608398,
      "learning_rate": 1.4566598985621503e-05,
      "loss": 1.9881,
      "step": 207455
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.3565618991851807,
      "learning_rate": 1.4565339856055458e-05,
      "loss": 2.0549,
      "step": 207456
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.3988239765167236,
      "learning_rate": 1.4564080779557097e-05,
      "loss": 1.9868,
      "step": 207457
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.5143626928329468,
      "learning_rate": 1.4562821756126652e-05,
      "loss": 2.0113,
      "step": 207458
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.518519639968872,
      "learning_rate": 1.4561562785764392e-05,
      "loss": 2.0915,
      "step": 207459
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.4483453035354614,
      "learning_rate": 1.456030386847048e-05,
      "loss": 2.0744,
      "step": 207460
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.5126676559448242,
      "learning_rate": 1.4559045004245218e-05,
      "loss": 1.7811,
      "step": 207461
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.538554072380066,
      "learning_rate": 1.4557786193088838e-05,
      "loss": 2.0747,
      "step": 207462
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.4259933233261108,
      "learning_rate": 1.4556527435001508e-05,
      "loss": 1.8893,
      "step": 207463
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.6222128868103027,
      "learning_rate": 1.455526872998356e-05,
      "loss": 2.0873,
      "step": 207464
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.4956001043319702,
      "learning_rate": 1.455401007803516e-05,
      "loss": 1.9826,
      "step": 207465
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.4628793001174927,
      "learning_rate": 1.4552751479156543e-05,
      "loss": 2.1066,
      "step": 207466
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.5587983131408691,
      "learning_rate": 1.4551492933347974e-05,
      "loss": 2.0367,
      "step": 207467
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.442169189453125,
      "learning_rate": 1.4550234440609686e-05,
      "loss": 1.9779,
      "step": 207468
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.6406482458114624,
      "learning_rate": 1.4548976000941848e-05,
      "loss": 2.0197,
      "step": 207469
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.454012393951416,
      "learning_rate": 1.454771761434479e-05,
      "loss": 2.1348,
      "step": 207470
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.5172287225723267,
      "learning_rate": 1.4546459280818646e-05,
      "loss": 2.0279,
      "step": 207471
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.4573826789855957,
      "learning_rate": 1.4545201000363748e-05,
      "loss": 2.0532,
      "step": 207472
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.3937793970108032,
      "learning_rate": 1.4543942772980299e-05,
      "loss": 2.0896,
      "step": 207473
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.4032893180847168,
      "learning_rate": 1.4542684598668498e-05,
      "loss": 2.191,
      "step": 207474
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.5035861730575562,
      "learning_rate": 1.4541426477428575e-05,
      "loss": 1.9403,
      "step": 207475
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.3794480562210083,
      "learning_rate": 1.4540168409260832e-05,
      "loss": 2.2201,
      "step": 207476
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.4343223571777344,
      "learning_rate": 1.4538910394165404e-05,
      "loss": 2.0445,
      "step": 207477
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.4752665758132935,
      "learning_rate": 1.4537652432142622e-05,
      "loss": 1.8913,
      "step": 207478
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.3644040822982788,
      "learning_rate": 1.4536394523192686e-05,
      "loss": 2.1254,
      "step": 207479
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.5628525018692017,
      "learning_rate": 1.4535136667315828e-05,
      "loss": 1.9521,
      "step": 207480
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.3857762813568115,
      "learning_rate": 1.453387886451225e-05,
      "loss": 1.8978,
      "step": 207481
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.4544684886932373,
      "learning_rate": 1.4532621114782217e-05,
      "loss": 2.0687,
      "step": 207482
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.5100241899490356,
      "learning_rate": 1.4531363418125963e-05,
      "loss": 2.0332,
      "step": 207483
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.5673859119415283,
      "learning_rate": 1.453010577454372e-05,
      "loss": 1.9581,
      "step": 207484
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.4418615102767944,
      "learning_rate": 1.4528848184035724e-05,
      "loss": 2.0291,
      "step": 207485
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.5439000129699707,
      "learning_rate": 1.4527590646602205e-05,
      "loss": 2.1662,
      "step": 207486
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.6574347019195557,
      "learning_rate": 1.4526333162243364e-05,
      "loss": 1.8937,
      "step": 207487
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.609856367111206,
      "learning_rate": 1.4525075730959502e-05,
      "loss": 1.908,
      "step": 207488
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.3911702632904053,
      "learning_rate": 1.4523818352750783e-05,
      "loss": 2.0193,
      "step": 207489
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.445622205734253,
      "learning_rate": 1.4522561027617508e-05,
      "loss": 2.0937,
      "step": 207490
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.4021445512771606,
      "learning_rate": 1.4521303755559876e-05,
      "loss": 1.9331,
      "step": 207491
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.347530722618103,
      "learning_rate": 1.4520046536578122e-05,
      "loss": 2.0337,
      "step": 207492
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.5248183012008667,
      "learning_rate": 1.4518789370672445e-05,
      "loss": 1.9282,
      "step": 207493
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.5022262334823608,
      "learning_rate": 1.4517532257843179e-05,
      "loss": 2.0602,
      "step": 207494
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.6105692386627197,
      "learning_rate": 1.4516275198090422e-05,
      "loss": 2.0023,
      "step": 207495
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.468550443649292,
      "learning_rate": 1.4515018191414506e-05,
      "loss": 1.9107,
      "step": 207496
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.4268450736999512,
      "learning_rate": 1.4513761237815669e-05,
      "loss": 1.8457,
      "step": 207497
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.4068366289138794,
      "learning_rate": 1.4512504337294106e-05,
      "loss": 2.219,
      "step": 207498
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.5019762516021729,
      "learning_rate": 1.4511247489849986e-05,
      "loss": 2.0241,
      "step": 207499
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.3366469144821167,
      "learning_rate": 1.4509990695483675e-05,
      "loss": 1.8594,
      "step": 207500
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.5393840074539185,
      "learning_rate": 1.450873395419534e-05,
      "loss": 1.9417,
      "step": 207501
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.452377438545227,
      "learning_rate": 1.4507477265985212e-05,
      "loss": 1.8498,
      "step": 207502
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.4722625017166138,
      "learning_rate": 1.450622063085356e-05,
      "loss": 2.2063,
      "step": 207503
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.600642204284668,
      "learning_rate": 1.4504964048800583e-05,
      "loss": 2.1252,
      "step": 207504
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.5993999242782593,
      "learning_rate": 1.450370751982648e-05,
      "loss": 1.8425,
      "step": 207505
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.4854246377944946,
      "learning_rate": 1.4502451043931552e-05,
      "loss": 2.0641,
      "step": 207506
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.4230154752731323,
      "learning_rate": 1.4501194621115997e-05,
      "loss": 1.9825,
      "step": 207507
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.4439516067504883,
      "learning_rate": 1.4499938251380084e-05,
      "loss": 2.119,
      "step": 207508
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.4656386375427246,
      "learning_rate": 1.449868193472401e-05,
      "loss": 1.9922,
      "step": 207509
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.4809026718139648,
      "learning_rate": 1.4497425671148044e-05,
      "loss": 2.0655,
      "step": 207510
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.5004862546920776,
      "learning_rate": 1.4496169460652352e-05,
      "loss": 1.9537,
      "step": 207511
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.4051538705825806,
      "learning_rate": 1.4494913303237232e-05,
      "loss": 1.9895,
      "step": 207512
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.4628241062164307,
      "learning_rate": 1.4493657198902886e-05,
      "loss": 1.9451,
      "step": 207513
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.6259167194366455,
      "learning_rate": 1.4492401147649579e-05,
      "loss": 1.9388,
      "step": 207514
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.43738853931427,
      "learning_rate": 1.4491145149477513e-05,
      "loss": 1.9698,
      "step": 207515
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.5446099042892456,
      "learning_rate": 1.4489889204386951e-05,
      "loss": 2.046,
      "step": 207516
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.3628087043762207,
      "learning_rate": 1.4488633312378062e-05,
      "loss": 2.2038,
      "step": 207517
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.616105318069458,
      "learning_rate": 1.4487377473451178e-05,
      "loss": 1.9361,
      "step": 207518
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.3419040441513062,
      "learning_rate": 1.44861216876064e-05,
      "loss": 2.0081,
      "step": 207519
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.4861950874328613,
      "learning_rate": 1.4484865954844128e-05,
      "loss": 1.9789,
      "step": 207520
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.486966848373413,
      "learning_rate": 1.4483610275164459e-05,
      "loss": 1.9463,
      "step": 207521
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.4738643169403076,
      "learning_rate": 1.4482354648567728e-05,
      "loss": 1.7594,
      "step": 207522
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.4284011125564575,
      "learning_rate": 1.448109907505407e-05,
      "loss": 1.8461,
      "step": 207523
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.3767635822296143,
      "learning_rate": 1.4479843554623783e-05,
      "loss": 1.9413,
      "step": 207524
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.4608607292175293,
      "learning_rate": 1.4478588087277065e-05,
      "loss": 2.2171,
      "step": 207525
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.558280110359192,
      "learning_rate": 1.4477332673014186e-05,
      "loss": 2.1498,
      "step": 207526
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.5190333127975464,
      "learning_rate": 1.4476077311835343e-05,
      "loss": 2.1045,
      "step": 207527
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.4416455030441284,
      "learning_rate": 1.4474822003740837e-05,
      "loss": 2.0265,
      "step": 207528
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.4217075109481812,
      "learning_rate": 1.4473566748730803e-05,
      "loss": 1.961,
      "step": 207529
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.4425526857376099,
      "learning_rate": 1.4472311546805536e-05,
      "loss": 1.8739,
      "step": 207530
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.5616027116775513,
      "learning_rate": 1.4471056397965209e-05,
      "loss": 1.8817,
      "step": 207531
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.386780858039856,
      "learning_rate": 1.4469801302210183e-05,
      "loss": 1.8646,
      "step": 207532
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.40955650806427,
      "learning_rate": 1.4468546259540525e-05,
      "loss": 1.9437,
      "step": 207533
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.3541088104248047,
      "learning_rate": 1.4467291269956671e-05,
      "loss": 1.8507,
      "step": 207534
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.4928381443023682,
      "learning_rate": 1.4466036333458652e-05,
      "loss": 2.2114,
      "step": 207535
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.5230122804641724,
      "learning_rate": 1.4464781450046803e-05,
      "loss": 1.9924,
      "step": 207536
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.4461711645126343,
      "learning_rate": 1.4463526619721321e-05,
      "loss": 1.7585,
      "step": 207537
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.4626750946044922,
      "learning_rate": 1.446227184248251e-05,
      "loss": 1.946,
      "step": 207538
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.3265161514282227,
      "learning_rate": 1.4461017118330498e-05,
      "loss": 1.9144,
      "step": 207539
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.3618003129959106,
      "learning_rate": 1.4459762447265588e-05,
      "loss": 2.1749,
      "step": 207540
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.3812769651412964,
      "learning_rate": 1.4458507829288045e-05,
      "loss": 1.9238,
      "step": 207541
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.4542156457901,
      "learning_rate": 1.4457253264398005e-05,
      "loss": 1.9608,
      "step": 207542
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.330824375152588,
      "learning_rate": 1.4455998752595765e-05,
      "loss": 1.78,
      "step": 207543
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.4452580213546753,
      "learning_rate": 1.445474429388156e-05,
      "loss": 1.8213,
      "step": 207544
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.4291993379592896,
      "learning_rate": 1.4453489888255554e-05,
      "loss": 1.9791,
      "step": 207545
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.7820243835449219,
      "learning_rate": 1.4452235535718082e-05,
      "loss": 1.9715,
      "step": 207546
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.4355757236480713,
      "learning_rate": 1.4450981236269343e-05,
      "loss": 2.0534,
      "step": 207547
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.462388038635254,
      "learning_rate": 1.4449726989909505e-05,
      "loss": 2.1794,
      "step": 207548
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.5381619930267334,
      "learning_rate": 1.4448472796638899e-05,
      "loss": 2.1193,
      "step": 207549
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.4241029024124146,
      "learning_rate": 1.4447218656457726e-05,
      "loss": 1.9929,
      "step": 207550
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.5270332098007202,
      "learning_rate": 1.444596456936612e-05,
      "loss": 2.2028,
      "step": 207551
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.3778446912765503,
      "learning_rate": 1.4444710535364479e-05,
      "loss": 1.9631,
      "step": 207552
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.5192930698394775,
      "learning_rate": 1.4443456554452937e-05,
      "loss": 1.8408,
      "step": 207553
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.5048872232437134,
      "learning_rate": 1.4442202626631728e-05,
      "loss": 1.8975,
      "step": 207554
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.5165764093399048,
      "learning_rate": 1.4440948751901116e-05,
      "loss": 2.0749,
      "step": 207555
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.5686973333358765,
      "learning_rate": 1.443969493026137e-05,
      "loss": 2.019,
      "step": 207556
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.2670717239379883,
      "learning_rate": 1.4438441161712589e-05,
      "loss": 1.9168,
      "step": 207557
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.56417715549469,
      "learning_rate": 1.4437187446255138e-05,
      "loss": 2.1665,
      "step": 207558
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.6018236875534058,
      "learning_rate": 1.443593378388922e-05,
      "loss": 1.8554,
      "step": 207559
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.557942271232605,
      "learning_rate": 1.4434680174614998e-05,
      "loss": 2.368,
      "step": 207560
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.4629406929016113,
      "learning_rate": 1.4433426618432808e-05,
      "loss": 1.6756,
      "step": 207561
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.395188331604004,
      "learning_rate": 1.4432173115342782e-05,
      "loss": 2.0558,
      "step": 207562
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.3616877794265747,
      "learning_rate": 1.4430919665345254e-05,
      "loss": 1.8828,
      "step": 207563
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.4552663564682007,
      "learning_rate": 1.442966626844042e-05,
      "loss": 2.0599,
      "step": 207564
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.7797722816467285,
      "learning_rate": 1.4428412924628485e-05,
      "loss": 2.0957,
      "step": 207565
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.4166067838668823,
      "learning_rate": 1.4427159633909679e-05,
      "loss": 1.9198,
      "step": 207566
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.4051457643508911,
      "learning_rate": 1.442590639628427e-05,
      "loss": 1.8026,
      "step": 207567
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.4319087266921997,
      "learning_rate": 1.4424653211752457e-05,
      "loss": 2.0714,
      "step": 207568
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.3655316829681396,
      "learning_rate": 1.4423400080314507e-05,
      "loss": 1.9976,
      "step": 207569
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.5517795085906982,
      "learning_rate": 1.4422147001970652e-05,
      "loss": 1.9475,
      "step": 207570
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.4127655029296875,
      "learning_rate": 1.4420893976721127e-05,
      "loss": 1.8878,
      "step": 207571
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.560444951057434,
      "learning_rate": 1.4419641004566096e-05,
      "loss": 1.8375,
      "step": 207572
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.5687261819839478,
      "learning_rate": 1.4418388085505861e-05,
      "loss": 2.0466,
      "step": 207573
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.4217702150344849,
      "learning_rate": 1.4417135219540622e-05,
      "loss": 1.9217,
      "step": 207574
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.4084484577178955,
      "learning_rate": 1.4415882406670675e-05,
      "loss": 2.181,
      "step": 207575
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.5052810907363892,
      "learning_rate": 1.4414629646896192e-05,
      "loss": 1.9561,
      "step": 207576
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.3439216613769531,
      "learning_rate": 1.4413376940217436e-05,
      "loss": 1.8862,
      "step": 207577
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.4423307180404663,
      "learning_rate": 1.4412124286634575e-05,
      "loss": 2.1029,
      "step": 207578
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.4341307878494263,
      "learning_rate": 1.4410871686147908e-05,
      "loss": 2.0555,
      "step": 207579
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.4175890684127808,
      "learning_rate": 1.4409619138757633e-05,
      "loss": 1.8593,
      "step": 207580
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.3798831701278687,
      "learning_rate": 1.4408366644464053e-05,
      "loss": 2.038,
      "step": 207581
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.4573181867599487,
      "learning_rate": 1.4407114203267334e-05,
      "loss": 1.9327,
      "step": 207582
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.4288005828857422,
      "learning_rate": 1.4405861815167707e-05,
      "loss": 2.0947,
      "step": 207583
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.521908164024353,
      "learning_rate": 1.4404609480165408e-05,
      "loss": 1.9926,
      "step": 207584
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.5382612943649292,
      "learning_rate": 1.4403357198260701e-05,
      "loss": 1.9627,
      "step": 207585
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.4930351972579956,
      "learning_rate": 1.4402104969453787e-05,
      "loss": 1.9586,
      "step": 207586
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.4387156963348389,
      "learning_rate": 1.4400852793744932e-05,
      "loss": 1.7978,
      "step": 207587
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.4497219324111938,
      "learning_rate": 1.4399600671134337e-05,
      "loss": 1.7977,
      "step": 207588
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.523491621017456,
      "learning_rate": 1.43983486016223e-05,
      "loss": 1.994,
      "step": 207589
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.4526513814926147,
      "learning_rate": 1.4397096585208923e-05,
      "loss": 2.1472,
      "step": 207590
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.6012028455734253,
      "learning_rate": 1.439584462189457e-05,
      "loss": 1.9205,
      "step": 207591
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.298041820526123,
      "learning_rate": 1.4394592711679375e-05,
      "loss": 1.7415,
      "step": 207592
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.406943917274475,
      "learning_rate": 1.439334085456364e-05,
      "loss": 2.051,
      "step": 207593
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.459920883178711,
      "learning_rate": 1.4392089050547562e-05,
      "loss": 2.2036,
      "step": 207594
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.5581954717636108,
      "learning_rate": 1.4390837299631441e-05,
      "loss": 1.9401,
      "step": 207595
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.4913893938064575,
      "learning_rate": 1.4389585601815378e-05,
      "loss": 1.8962,
      "step": 207596
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.4689552783966064,
      "learning_rate": 1.438833395709974e-05,
      "loss": 1.9397,
      "step": 207597
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.4832453727722168,
      "learning_rate": 1.438708236548466e-05,
      "loss": 2.001,
      "step": 207598
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.5054773092269897,
      "learning_rate": 1.4385830826970434e-05,
      "loss": 2.2127,
      "step": 207599
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.4567056894302368,
      "learning_rate": 1.4384579341557234e-05,
      "loss": 2.1018,
      "step": 207600
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.3194019794464111,
      "learning_rate": 1.4383327909245424e-05,
      "loss": 2.1059,
      "step": 207601
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.5463647842407227,
      "learning_rate": 1.438207653003507e-05,
      "loss": 2.045,
      "step": 207602
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.4499869346618652,
      "learning_rate": 1.4380825203926505e-05,
      "loss": 1.9617,
      "step": 207603
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.62282133102417,
      "learning_rate": 1.4379573930919929e-05,
      "loss": 1.886,
      "step": 207604
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.6020947694778442,
      "learning_rate": 1.4378322711015578e-05,
      "loss": 1.9789,
      "step": 207605
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.5239702463150024,
      "learning_rate": 1.437707154421368e-05,
      "loss": 1.9059,
      "step": 207606
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.5112022161483765,
      "learning_rate": 1.437582043051454e-05,
      "loss": 2.1886,
      "step": 207607
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.536582589149475,
      "learning_rate": 1.4374569369918255e-05,
      "loss": 2.1169,
      "step": 207608
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.393553614616394,
      "learning_rate": 1.4373318362425157e-05,
      "loss": 1.9699,
      "step": 207609
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.5889769792556763,
      "learning_rate": 1.4372067408035414e-05,
      "loss": 2.1294,
      "step": 207610
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.3851174116134644,
      "learning_rate": 1.4370816506749328e-05,
      "loss": 1.987,
      "step": 207611
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.3758180141448975,
      "learning_rate": 1.436956565856706e-05,
      "loss": 2.0398,
      "step": 207612
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.7506014108657837,
      "learning_rate": 1.4368314863488984e-05,
      "loss": 2.0385,
      "step": 207613
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.48375403881073,
      "learning_rate": 1.4367064121515126e-05,
      "loss": 2.0311,
      "step": 207614
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.4321329593658447,
      "learning_rate": 1.4365813432645889e-05,
      "loss": 2.177,
      "step": 207615
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.3507882356643677,
      "learning_rate": 1.4364562796881374e-05,
      "loss": 1.9485,
      "step": 207616
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.4580495357513428,
      "learning_rate": 1.4363312214221945e-05,
      "loss": 2.0272,
      "step": 207617
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.4082266092300415,
      "learning_rate": 1.4362061684667703e-05,
      "loss": 1.8531,
      "step": 207618
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.4961787462234497,
      "learning_rate": 1.4360811208219047e-05,
      "loss": 1.8246,
      "step": 207619
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.3660119771957397,
      "learning_rate": 1.4359560784876012e-05,
      "loss": 2.0398,
      "step": 207620
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.431627869606018,
      "learning_rate": 1.4358310414638962e-05,
      "loss": 1.8715,
      "step": 207621
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.409739375114441,
      "learning_rate": 1.4357060097508067e-05,
      "loss": 1.8899,
      "step": 207622
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.457102656364441,
      "learning_rate": 1.4355809833483622e-05,
      "loss": 1.9389,
      "step": 207623
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.9815534353256226,
      "learning_rate": 1.4354559622565765e-05,
      "loss": 1.9295,
      "step": 207624
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.3496770858764648,
      "learning_rate": 1.4353309464754859e-05,
      "loss": 2.1964,
      "step": 207625
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.3273284435272217,
      "learning_rate": 1.4352059360051038e-05,
      "loss": 1.9567,
      "step": 207626
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.3039391040802002,
      "learning_rate": 1.4350809308454569e-05,
      "loss": 2.0343,
      "step": 207627
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.668948769569397,
      "learning_rate": 1.4349559309965653e-05,
      "loss": 2.1235,
      "step": 207628
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.3983097076416016,
      "learning_rate": 1.4348309364584587e-05,
      "loss": 2.0222,
      "step": 207629
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.5411626100540161,
      "learning_rate": 1.4347059472311505e-05,
      "loss": 1.8345,
      "step": 207630
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.4190062284469604,
      "learning_rate": 1.4345809633146744e-05,
      "loss": 1.9848,
      "step": 207631
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.4503402709960938,
      "learning_rate": 1.4344559847090464e-05,
      "loss": 2.2265,
      "step": 207632
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.4106414318084717,
      "learning_rate": 1.4343310114142903e-05,
      "loss": 1.9537,
      "step": 207633
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.3827061653137207,
      "learning_rate": 1.4342060434304358e-05,
      "loss": 1.9571,
      "step": 207634
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.4903860092163086,
      "learning_rate": 1.4340810807575031e-05,
      "loss": 1.8633,
      "step": 207635
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.5671898126602173,
      "learning_rate": 1.4339561233955088e-05,
      "loss": 2.0235,
      "step": 207636
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.4522541761398315,
      "learning_rate": 1.4338311713444828e-05,
      "loss": 1.9291,
      "step": 207637
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.3245325088500977,
      "learning_rate": 1.4337062246044484e-05,
      "loss": 1.9868,
      "step": 207638
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.3362170457839966,
      "learning_rate": 1.4335812831754223e-05,
      "loss": 1.9749,
      "step": 207639
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.4807478189468384,
      "learning_rate": 1.4334563470574411e-05,
      "loss": 1.8402,
      "step": 207640
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.4596580266952515,
      "learning_rate": 1.433331416250515e-05,
      "loss": 1.7654,
      "step": 207641
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.423825740814209,
      "learning_rate": 1.4332064907546703e-05,
      "loss": 1.9356,
      "step": 207642
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.5497772693634033,
      "learning_rate": 1.4330815705699338e-05,
      "loss": 2.1528,
      "step": 207643
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.6094716787338257,
      "learning_rate": 1.4329566556963256e-05,
      "loss": 1.8904,
      "step": 207644
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.4225002527236938,
      "learning_rate": 1.432831746133869e-05,
      "loss": 2.076,
      "step": 207645
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.4380440711975098,
      "learning_rate": 1.4327068418825905e-05,
      "loss": 1.9284,
      "step": 207646
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.4606575965881348,
      "learning_rate": 1.43258194294251e-05,
      "loss": 1.8334,
      "step": 207647
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.4961580038070679,
      "learning_rate": 1.4324570493136512e-05,
      "loss": 2.0543,
      "step": 207648
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.4493870735168457,
      "learning_rate": 1.4323321609960404e-05,
      "loss": 2.1269,
      "step": 207649
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.3678162097930908,
      "learning_rate": 1.4322072779896977e-05,
      "loss": 2.0246,
      "step": 207650
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.4690173864364624,
      "learning_rate": 1.4320824002946429e-05,
      "loss": 1.9856,
      "step": 207651
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.5472462177276611,
      "learning_rate": 1.4319575279109097e-05,
      "loss": 2.0492,
      "step": 207652
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.7289655208587646,
      "learning_rate": 1.4318326608385078e-05,
      "loss": 2.1052,
      "step": 207653
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.4460170269012451,
      "learning_rate": 1.431707799077474e-05,
      "loss": 1.7929,
      "step": 207654
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.4576911926269531,
      "learning_rate": 1.4315829426278213e-05,
      "loss": 2.1496,
      "step": 207655
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.4642730951309204,
      "learning_rate": 1.4314580914895801e-05,
      "loss": 2.1976,
      "step": 207656
    },
    {
      "epoch": 2.7,
      "grad_norm": 2.0307421684265137,
      "learning_rate": 1.4313332456627635e-05,
      "loss": 1.8793,
      "step": 207657
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.4146150350570679,
      "learning_rate": 1.4312084051474081e-05,
      "loss": 2.0434,
      "step": 207658
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.5755428075790405,
      "learning_rate": 1.4310835699435241e-05,
      "loss": 2.132,
      "step": 207659
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.4488451480865479,
      "learning_rate": 1.4309587400511447e-05,
      "loss": 2.0236,
      "step": 207660
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.4397763013839722,
      "learning_rate": 1.4308339154702863e-05,
      "loss": 2.0238,
      "step": 207661
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.459267497062683,
      "learning_rate": 1.430709096200986e-05,
      "loss": 2.1295,
      "step": 207662
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.5045245885849,
      "learning_rate": 1.4305842822432433e-05,
      "loss": 2.0256,
      "step": 207663
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.5157417058944702,
      "learning_rate": 1.430459473597102e-05,
      "loss": 1.8469,
      "step": 207664
    },
    {
      "epoch": 2.7,
      "grad_norm": 2.3745527267456055,
      "learning_rate": 1.4303346702625717e-05,
      "loss": 2.009,
      "step": 207665
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.59043550491333,
      "learning_rate": 1.430209872239686e-05,
      "loss": 2.0683,
      "step": 207666
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.3711673021316528,
      "learning_rate": 1.430085079528458e-05,
      "loss": 1.9065,
      "step": 207667
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.5215927362442017,
      "learning_rate": 1.4299602921289244e-05,
      "loss": 1.8804,
      "step": 207668
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.4706238508224487,
      "learning_rate": 1.4298355100410952e-05,
      "loss": 2.0939,
      "step": 207669
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.459729552268982,
      "learning_rate": 1.4297107332650004e-05,
      "loss": 1.9761,
      "step": 207670
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.451881766319275,
      "learning_rate": 1.42958596180066e-05,
      "loss": 1.8514,
      "step": 207671
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.530359148979187,
      "learning_rate": 1.4294611956481006e-05,
      "loss": 1.8022,
      "step": 207672
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.393505334854126,
      "learning_rate": 1.4293364348073422e-05,
      "loss": 1.9925,
      "step": 207673
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.4018268585205078,
      "learning_rate": 1.4292116792784147e-05,
      "loss": 2.0497,
      "step": 207674
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.3915283679962158,
      "learning_rate": 1.4290869290613315e-05,
      "loss": 2.0101,
      "step": 207675
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.5235285758972168,
      "learning_rate": 1.4289621841561227e-05,
      "loss": 1.8882,
      "step": 207676
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.3999131917953491,
      "learning_rate": 1.4288374445628048e-05,
      "loss": 2.0043,
      "step": 207677
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.4271519184112549,
      "learning_rate": 1.4287127102814079e-05,
      "loss": 1.7374,
      "step": 207678
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.3611321449279785,
      "learning_rate": 1.4285879813119484e-05,
      "loss": 1.8938,
      "step": 207679
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.3505527973175049,
      "learning_rate": 1.428463257654463e-05,
      "loss": 1.9379,
      "step": 207680
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.3787660598754883,
      "learning_rate": 1.4283385393089587e-05,
      "loss": 1.9677,
      "step": 207681
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.481459140777588,
      "learning_rate": 1.4282138262754683e-05,
      "loss": 1.9601,
      "step": 207682
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.3289213180541992,
      "learning_rate": 1.4280891185540088e-05,
      "loss": 1.9425,
      "step": 207683
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.487221360206604,
      "learning_rate": 1.4279644161446135e-05,
      "loss": 1.9434,
      "step": 207684
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.3881398439407349,
      "learning_rate": 1.4278397190472923e-05,
      "loss": 1.9105,
      "step": 207685
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.4440068006515503,
      "learning_rate": 1.4277150272620818e-05,
      "loss": 1.8296,
      "step": 207686
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.44364333152771,
      "learning_rate": 1.4275903407889922e-05,
      "loss": 1.9039,
      "step": 207687
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.3466408252716064,
      "learning_rate": 1.4274656596280565e-05,
      "loss": 1.7785,
      "step": 207688
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.4337965250015259,
      "learning_rate": 1.4273409837792883e-05,
      "loss": 1.863,
      "step": 207689
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.4255231618881226,
      "learning_rate": 1.427216313242724e-05,
      "loss": 2.0767,
      "step": 207690
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.3334274291992188,
      "learning_rate": 1.4270916480183736e-05,
      "loss": 1.9848,
      "step": 207691
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.591228723526001,
      "learning_rate": 1.426966988106274e-05,
      "loss": 1.8251,
      "step": 207692
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.4487611055374146,
      "learning_rate": 1.426842333506435e-05,
      "loss": 2.164,
      "step": 207693
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.4540430307388306,
      "learning_rate": 1.4267176842188865e-05,
      "loss": 1.9541,
      "step": 207694
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.4433412551879883,
      "learning_rate": 1.4265930402436487e-05,
      "loss": 1.8603,
      "step": 207695
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.5704935789108276,
      "learning_rate": 1.4264684015807482e-05,
      "loss": 1.9937,
      "step": 207696
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.45261549949646,
      "learning_rate": 1.4263437682302048e-05,
      "loss": 1.8332,
      "step": 207697
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.4661905765533447,
      "learning_rate": 1.4262191401920519e-05,
      "loss": 1.9803,
      "step": 207698
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.6647511720657349,
      "learning_rate": 1.426094517466293e-05,
      "loss": 2.1374,
      "step": 207699
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.4758975505828857,
      "learning_rate": 1.4259699000529711e-05,
      "loss": 1.9902,
      "step": 207700
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.529917597770691,
      "learning_rate": 1.4258452879520933e-05,
      "loss": 1.8772,
      "step": 207701
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.466786503791809,
      "learning_rate": 1.4257206811636956e-05,
      "loss": 2.1323,
      "step": 207702
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.3588923215866089,
      "learning_rate": 1.4255960796877918e-05,
      "loss": 2.0297,
      "step": 207703
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.4677187204360962,
      "learning_rate": 1.4254714835244152e-05,
      "loss": 2.028,
      "step": 207704
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.433676838874817,
      "learning_rate": 1.4253468926735757e-05,
      "loss": 2.0791,
      "step": 207705
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.551173210144043,
      "learning_rate": 1.4252223071353097e-05,
      "loss": 1.8421,
      "step": 207706
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.5790584087371826,
      "learning_rate": 1.4250977269096275e-05,
      "loss": 1.921,
      "step": 207707
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.4359972476959229,
      "learning_rate": 1.4249731519965657e-05,
      "loss": 2.0008,
      "step": 207708
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.342713475227356,
      "learning_rate": 1.4248485823961342e-05,
      "loss": 2.1438,
      "step": 207709
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.4822758436203003,
      "learning_rate": 1.4247240181083697e-05,
      "loss": 1.8602,
      "step": 207710
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.346238374710083,
      "learning_rate": 1.4245994591332854e-05,
      "loss": 1.8837,
      "step": 207711
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.5542657375335693,
      "learning_rate": 1.4244749054709082e-05,
      "loss": 2.2289,
      "step": 207712
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.453739047050476,
      "learning_rate": 1.4243503571212579e-05,
      "loss": 2.0939,
      "step": 207713
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.9807775020599365,
      "learning_rate": 1.4242258140843643e-05,
      "loss": 1.9104,
      "step": 207714
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.4670734405517578,
      "learning_rate": 1.424101276360241e-05,
      "loss": 2.0237,
      "step": 207715
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.4644711017608643,
      "learning_rate": 1.4239767439489213e-05,
      "loss": 2.1806,
      "step": 207716
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.4658035039901733,
      "learning_rate": 1.4238522168504218e-05,
      "loss": 2.0174,
      "step": 207717
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.3614871501922607,
      "learning_rate": 1.4237276950647724e-05,
      "loss": 1.8454,
      "step": 207718
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.5293231010437012,
      "learning_rate": 1.4236031785919832e-05,
      "loss": 1.8662,
      "step": 207719
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.5240280628204346,
      "learning_rate": 1.4234786674320908e-05,
      "loss": 2.009,
      "step": 207720
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.5407936573028564,
      "learning_rate": 1.4233541615851085e-05,
      "loss": 2.1012,
      "step": 207721
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.4859164953231812,
      "learning_rate": 1.4232296610510696e-05,
      "loss": 1.9163,
      "step": 207722
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.402548909187317,
      "learning_rate": 1.4231051658299908e-05,
      "loss": 1.976,
      "step": 207723
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.451259970664978,
      "learning_rate": 1.4229806759218921e-05,
      "loss": 2.0986,
      "step": 207724
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.4599545001983643,
      "learning_rate": 1.4228561913268065e-05,
      "loss": 1.9271,
      "step": 207725
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.4234830141067505,
      "learning_rate": 1.4227317120447479e-05,
      "loss": 2.0186,
      "step": 207726
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.4142072200775146,
      "learning_rate": 1.4226072380757425e-05,
      "loss": 1.9712,
      "step": 207727
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.3364259004592896,
      "learning_rate": 1.4224827694198171e-05,
      "loss": 2.0744,
      "step": 207728
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.536716103553772,
      "learning_rate": 1.4223583060769883e-05,
      "loss": 2.3581,
      "step": 207729
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.583024501800537,
      "learning_rate": 1.4222338480472827e-05,
      "loss": 2.0508,
      "step": 207730
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.3587079048156738,
      "learning_rate": 1.4221093953307238e-05,
      "loss": 1.9939,
      "step": 207731
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.5877721309661865,
      "learning_rate": 1.4219849479273349e-05,
      "loss": 1.8876,
      "step": 207732
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.411673665046692,
      "learning_rate": 1.421860505837139e-05,
      "loss": 1.8137,
      "step": 207733
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.4982259273529053,
      "learning_rate": 1.4217360690601597e-05,
      "loss": 1.7706,
      "step": 207734
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.621658205986023,
      "learning_rate": 1.421611637596417e-05,
      "loss": 1.7313,
      "step": 207735
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.5309021472930908,
      "learning_rate": 1.4214872114459342e-05,
      "loss": 2.1572,
      "step": 207736
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.6061009168624878,
      "learning_rate": 1.421362790608741e-05,
      "loss": 2.1465,
      "step": 207737
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.334418773651123,
      "learning_rate": 1.421238375084851e-05,
      "loss": 1.9249,
      "step": 207738
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.467835783958435,
      "learning_rate": 1.4211139648742941e-05,
      "loss": 1.8212,
      "step": 207739
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.4675332307815552,
      "learning_rate": 1.4209895599770937e-05,
      "loss": 1.8773,
      "step": 207740
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.3562438488006592,
      "learning_rate": 1.420865160393273e-05,
      "loss": 1.8519,
      "step": 207741
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.7454502582550049,
      "learning_rate": 1.4207407661228454e-05,
      "loss": 2.2126,
      "step": 207742
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.4620736837387085,
      "learning_rate": 1.4206163771658474e-05,
      "loss": 1.8339,
      "step": 207743
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.4008756875991821,
      "learning_rate": 1.4204919935222925e-05,
      "loss": 2.058,
      "step": 207744
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.3399336338043213,
      "learning_rate": 1.4203676151922105e-05,
      "loss": 1.9938,
      "step": 207745
    },
    {
      "epoch": 2.7,
      "grad_norm": 1.4256073236465454,
      "learning_rate": 1.4202432421756182e-05,
      "loss": 1.9999,
      "step": 207746
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.3784551620483398,
      "learning_rate": 1.4201188744725488e-05,
      "loss": 1.8869,
      "step": 207747
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.4418245553970337,
      "learning_rate": 1.4199945120830125e-05,
      "loss": 1.8523,
      "step": 207748
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.549185037612915,
      "learning_rate": 1.4198701550070424e-05,
      "loss": 2.0357,
      "step": 207749
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.6700994968414307,
      "learning_rate": 1.4197458032446518e-05,
      "loss": 2.1177,
      "step": 207750
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.3582098484039307,
      "learning_rate": 1.4196214567958774e-05,
      "loss": 2.0536,
      "step": 207751
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.6012729406356812,
      "learning_rate": 1.4194971156607293e-05,
      "loss": 1.866,
      "step": 207752
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.4685499668121338,
      "learning_rate": 1.4193727798392441e-05,
      "loss": 1.9851,
      "step": 207753
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.4215093851089478,
      "learning_rate": 1.4192484493314282e-05,
      "loss": 1.9997,
      "step": 207754
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.4557743072509766,
      "learning_rate": 1.4191241241373186e-05,
      "loss": 1.9219,
      "step": 207755
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.6427229642868042,
      "learning_rate": 1.4189998042569283e-05,
      "loss": 2.0228,
      "step": 207756
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.4534807205200195,
      "learning_rate": 1.418875489690291e-05,
      "loss": 2.0967,
      "step": 207757
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.3792879581451416,
      "learning_rate": 1.4187511804374196e-05,
      "loss": 1.9426,
      "step": 207758
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.4030681848526,
      "learning_rate": 1.4186268764983511e-05,
      "loss": 2.2431,
      "step": 207759
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.5459760427474976,
      "learning_rate": 1.4185025778730886e-05,
      "loss": 1.9688,
      "step": 207760
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.4407870769500732,
      "learning_rate": 1.418378284561672e-05,
      "loss": 1.8511,
      "step": 207761
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.443263292312622,
      "learning_rate": 1.4182539965641148e-05,
      "loss": 2.1339,
      "step": 207762
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.3667386770248413,
      "learning_rate": 1.4181297138804471e-05,
      "loss": 1.8527,
      "step": 207763
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.5120768547058105,
      "learning_rate": 1.4180054365106819e-05,
      "loss": 1.9781,
      "step": 207764
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.4295072555541992,
      "learning_rate": 1.4178811644548594e-05,
      "loss": 1.973,
      "step": 207765
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.417898178100586,
      "learning_rate": 1.4177568977129861e-05,
      "loss": 1.9477,
      "step": 207766
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.5669562816619873,
      "learning_rate": 1.417632636285092e-05,
      "loss": 2.2308,
      "step": 207767
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.594581127166748,
      "learning_rate": 1.4175083801711973e-05,
      "loss": 1.9495,
      "step": 207768
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.4125045537948608,
      "learning_rate": 1.4173841293713316e-05,
      "loss": 1.9325,
      "step": 207769
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.3590439558029175,
      "learning_rate": 1.4172598838855086e-05,
      "loss": 1.9978,
      "step": 207770
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.4856257438659668,
      "learning_rate": 1.4171356437137682e-05,
      "loss": 2.0827,
      "step": 207771
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.585888385772705,
      "learning_rate": 1.4170114088561101e-05,
      "loss": 2.3162,
      "step": 207772
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.3184584379196167,
      "learning_rate": 1.4168871793125714e-05,
      "loss": 2.1578,
      "step": 207773
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.3559584617614746,
      "learning_rate": 1.4167629550831716e-05,
      "loss": 1.9748,
      "step": 207774
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.4322134256362915,
      "learning_rate": 1.416638736167941e-05,
      "loss": 1.9493,
      "step": 207775
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.4897383451461792,
      "learning_rate": 1.4165145225668895e-05,
      "loss": 2.0433,
      "step": 207776
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.3543438911437988,
      "learning_rate": 1.416390314280057e-05,
      "loss": 1.9009,
      "step": 207777
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.5193440914154053,
      "learning_rate": 1.4162661113074503e-05,
      "loss": 2.0208,
      "step": 207778
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.5297706127166748,
      "learning_rate": 1.4161419136490992e-05,
      "loss": 2.0578,
      "step": 207779
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.5003024339675903,
      "learning_rate": 1.4160177213050273e-05,
      "loss": 2.0277,
      "step": 207780
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.5795369148254395,
      "learning_rate": 1.4158935342752575e-05,
      "loss": 1.9971,
      "step": 207781
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.5215421915054321,
      "learning_rate": 1.4157693525598102e-05,
      "loss": 2.0631,
      "step": 207782
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.408779263496399,
      "learning_rate": 1.4156451761587218e-05,
      "loss": 2.0672,
      "step": 207783
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.3975132703781128,
      "learning_rate": 1.4155210050719923e-05,
      "loss": 1.8873,
      "step": 207784
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.375204086303711,
      "learning_rate": 1.4153968392996651e-05,
      "loss": 1.9344,
      "step": 207785
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.6074975728988647,
      "learning_rate": 1.4152726788417468e-05,
      "loss": 1.9247,
      "step": 207786
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.5996843576431274,
      "learning_rate": 1.415148523698274e-05,
      "loss": 2.0596,
      "step": 207787
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.49055016040802,
      "learning_rate": 1.4150243738692635e-05,
      "loss": 1.8744,
      "step": 207788
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.5213415622711182,
      "learning_rate": 1.4149002293547418e-05,
      "loss": 1.9637,
      "step": 207789
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.5842548608779907,
      "learning_rate": 1.4147760901547289e-05,
      "loss": 2.3148,
      "step": 207790
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.369878888130188,
      "learning_rate": 1.4146519562692482e-05,
      "loss": 1.9558,
      "step": 207791
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.4622939825057983,
      "learning_rate": 1.414527827698323e-05,
      "loss": 2.1113,
      "step": 207792
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.4417344331741333,
      "learning_rate": 1.4144037044419765e-05,
      "loss": 1.903,
      "step": 207793
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.4675449132919312,
      "learning_rate": 1.4142795865002322e-05,
      "loss": 1.9586,
      "step": 207794
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.5174834728240967,
      "learning_rate": 1.4141554738731131e-05,
      "loss": 1.8105,
      "step": 207795
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.5760177373886108,
      "learning_rate": 1.4140313665606428e-05,
      "loss": 2.1038,
      "step": 207796
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.5791178941726685,
      "learning_rate": 1.4139072645628446e-05,
      "loss": 2.0266,
      "step": 207797
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.358934760093689,
      "learning_rate": 1.413783167879735e-05,
      "loss": 1.9255,
      "step": 207798
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.3668984174728394,
      "learning_rate": 1.4136590765113476e-05,
      "loss": 1.8132,
      "step": 207799
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.5490212440490723,
      "learning_rate": 1.4135349904576987e-05,
      "loss": 2.0088,
      "step": 207800
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.4569249153137207,
      "learning_rate": 1.413410909718815e-05,
      "loss": 2.0032,
      "step": 207801
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.38539457321167,
      "learning_rate": 1.4132868342947202e-05,
      "loss": 1.9996,
      "step": 207802
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.4866915941238403,
      "learning_rate": 1.4131627641854304e-05,
      "loss": 1.7852,
      "step": 207803
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.6411387920379639,
      "learning_rate": 1.4130386993909725e-05,
      "loss": 1.8431,
      "step": 207804
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.4766846895217896,
      "learning_rate": 1.4129146399113733e-05,
      "loss": 1.982,
      "step": 207805
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.4270395040512085,
      "learning_rate": 1.4127905857466493e-05,
      "loss": 1.9173,
      "step": 207806
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.3755767345428467,
      "learning_rate": 1.4126665368968337e-05,
      "loss": 2.1876,
      "step": 207807
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.4614039659500122,
      "learning_rate": 1.41254249336194e-05,
      "loss": 1.9671,
      "step": 207808
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.4002797603607178,
      "learning_rate": 1.4124184551419915e-05,
      "loss": 1.8885,
      "step": 207809
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.4308171272277832,
      "learning_rate": 1.4122944222370181e-05,
      "loss": 1.9245,
      "step": 207810
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.4334912300109863,
      "learning_rate": 1.4121703946470364e-05,
      "loss": 1.7931,
      "step": 207811
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.5637736320495605,
      "learning_rate": 1.4120463723720698e-05,
      "loss": 2.132,
      "step": 207812
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.675222635269165,
      "learning_rate": 1.4119223554121483e-05,
      "loss": 1.858,
      "step": 207813
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.4300333261489868,
      "learning_rate": 1.4117983437672886e-05,
      "loss": 1.9702,
      "step": 207814
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.463594675064087,
      "learning_rate": 1.4116743374375106e-05,
      "loss": 1.8929,
      "step": 207815
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.5214109420776367,
      "learning_rate": 1.4115503364228476e-05,
      "loss": 2.0131,
      "step": 207816
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.3350728750228882,
      "learning_rate": 1.4114263407233162e-05,
      "loss": 1.949,
      "step": 207817
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.585423231124878,
      "learning_rate": 1.4113023503389365e-05,
      "loss": 1.9521,
      "step": 207818
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.4838032722473145,
      "learning_rate": 1.4111783652697383e-05,
      "loss": 1.9494,
      "step": 207819
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.377130150794983,
      "learning_rate": 1.4110543855157453e-05,
      "loss": 1.9945,
      "step": 207820
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.4904289245605469,
      "learning_rate": 1.4109304110769704e-05,
      "loss": 1.9541,
      "step": 207821
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.56279456615448,
      "learning_rate": 1.4108064419534471e-05,
      "loss": 2.0899,
      "step": 207822
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.3549000024795532,
      "learning_rate": 1.4106824781451886e-05,
      "loss": 1.9976,
      "step": 207823
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.441495656967163,
      "learning_rate": 1.4105585196522318e-05,
      "loss": 1.8624,
      "step": 207824
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.4169939756393433,
      "learning_rate": 1.4104345664745897e-05,
      "loss": 1.7216,
      "step": 207825
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.422360897064209,
      "learning_rate": 1.410310618612286e-05,
      "loss": 2.0549,
      "step": 207826
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.5745807886123657,
      "learning_rate": 1.4101866760653435e-05,
      "loss": 2.1261,
      "step": 207827
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.432892084121704,
      "learning_rate": 1.4100627388337925e-05,
      "loss": 2.014,
      "step": 207828
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.4762241840362549,
      "learning_rate": 1.4099388069176432e-05,
      "loss": 2.1205,
      "step": 207829
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.3768826723098755,
      "learning_rate": 1.4098148803169318e-05,
      "loss": 2.0101,
      "step": 207830
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.4043123722076416,
      "learning_rate": 1.4096909590316719e-05,
      "loss": 1.9676,
      "step": 207831
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.5076130628585815,
      "learning_rate": 1.4095670430618966e-05,
      "loss": 2.0429,
      "step": 207832
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.498818278312683,
      "learning_rate": 1.4094431324076127e-05,
      "loss": 2.0299,
      "step": 207833
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.5464900732040405,
      "learning_rate": 1.4093192270688603e-05,
      "loss": 1.9205,
      "step": 207834
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.4884166717529297,
      "learning_rate": 1.4091953270456524e-05,
      "loss": 1.9208,
      "step": 207835
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.3828321695327759,
      "learning_rate": 1.409071432338016e-05,
      "loss": 1.7822,
      "step": 207836
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.3970328569412231,
      "learning_rate": 1.4089475429459673e-05,
      "loss": 1.8458,
      "step": 207837
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.653451681137085,
      "learning_rate": 1.4088236588695468e-05,
      "loss": 1.8883,
      "step": 207838
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.5131869316101074,
      "learning_rate": 1.4086997801087574e-05,
      "loss": 2.2356,
      "step": 207839
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.4353913068771362,
      "learning_rate": 1.4085759066636327e-05,
      "loss": 1.975,
      "step": 207840
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.4141284227371216,
      "learning_rate": 1.4084520385341891e-05,
      "loss": 1.9896,
      "step": 207841
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.3690205812454224,
      "learning_rate": 1.4083281757204567e-05,
      "loss": 1.7661,
      "step": 207842
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.4400241374969482,
      "learning_rate": 1.4082043182224555e-05,
      "loss": 1.9996,
      "step": 207843
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.5573277473449707,
      "learning_rate": 1.4080804660402156e-05,
      "loss": 2.042,
      "step": 207844
    },
    {
      "epoch": 2.71,
      "grad_norm": 2.8476061820983887,
      "learning_rate": 1.4079566191737434e-05,
      "loss": 2.0314,
      "step": 207845
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.5496702194213867,
      "learning_rate": 1.407832777623079e-05,
      "loss": 1.9335,
      "step": 207846
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.56051504611969,
      "learning_rate": 1.4077089413882325e-05,
      "loss": 2.1623,
      "step": 207847
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.5335006713867188,
      "learning_rate": 1.4075851104692337e-05,
      "loss": 1.9662,
      "step": 207848
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.5565872192382812,
      "learning_rate": 1.407461284866106e-05,
      "loss": 1.89,
      "step": 207849
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.5178269147872925,
      "learning_rate": 1.4073374645788726e-05,
      "loss": 1.8745,
      "step": 207850
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.6286855936050415,
      "learning_rate": 1.4072136496075537e-05,
      "loss": 1.9166,
      "step": 207851
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.6428323984146118,
      "learning_rate": 1.4070898399521723e-05,
      "loss": 1.8289,
      "step": 207852
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.4397821426391602,
      "learning_rate": 1.406966035612752e-05,
      "loss": 2.166,
      "step": 207853
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.5225112438201904,
      "learning_rate": 1.4068422365893194e-05,
      "loss": 2.1077,
      "step": 207854
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.5014066696166992,
      "learning_rate": 1.4067184428818911e-05,
      "loss": 1.7308,
      "step": 207855
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.5241276025772095,
      "learning_rate": 1.4065946544904971e-05,
      "loss": 1.8195,
      "step": 207856
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.496429681777954,
      "learning_rate": 1.4064708714151573e-05,
      "loss": 2.1255,
      "step": 207857
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.3713139295578003,
      "learning_rate": 1.4063470936558919e-05,
      "loss": 1.8577,
      "step": 207858
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.5500725507736206,
      "learning_rate": 1.406223321212724e-05,
      "loss": 1.9454,
      "step": 207859
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.386349081993103,
      "learning_rate": 1.4060995540856835e-05,
      "loss": 1.9092,
      "step": 207860
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.5298900604248047,
      "learning_rate": 1.4059757922747838e-05,
      "loss": 1.9403,
      "step": 207861
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.549615502357483,
      "learning_rate": 1.4058520357800585e-05,
      "loss": 2.1465,
      "step": 207862
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.3116730451583862,
      "learning_rate": 1.4057282846015238e-05,
      "loss": 1.8784,
      "step": 207863
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.610987901687622,
      "learning_rate": 1.4056045387392035e-05,
      "loss": 2.1704,
      "step": 207864
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.467930555343628,
      "learning_rate": 1.4054807981931171e-05,
      "loss": 1.9243,
      "step": 207865
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.5327035188674927,
      "learning_rate": 1.4053570629632981e-05,
      "loss": 2.1267,
      "step": 207866
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.6707695722579956,
      "learning_rate": 1.4052333330497566e-05,
      "loss": 2.0943,
      "step": 207867
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.3974910974502563,
      "learning_rate": 1.405109608452526e-05,
      "loss": 1.9858,
      "step": 207868
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.5954065322875977,
      "learning_rate": 1.4049858891716259e-05,
      "loss": 1.927,
      "step": 207869
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.5243401527404785,
      "learning_rate": 1.4048621752070798e-05,
      "loss": 1.9729,
      "step": 207870
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.3039768934249878,
      "learning_rate": 1.4047384665589045e-05,
      "loss": 1.8211,
      "step": 207871
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.5428309440612793,
      "learning_rate": 1.4046147632271299e-05,
      "loss": 2.1572,
      "step": 207872
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.5105974674224854,
      "learning_rate": 1.4044910652117758e-05,
      "loss": 1.8722,
      "step": 207873
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.392598032951355,
      "learning_rate": 1.4043673725128723e-05,
      "loss": 1.9619,
      "step": 207874
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.3359438180923462,
      "learning_rate": 1.404243685130433e-05,
      "loss": 1.9223,
      "step": 207875
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.4052982330322266,
      "learning_rate": 1.4041200030644873e-05,
      "loss": 1.8297,
      "step": 207876
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.4389420747756958,
      "learning_rate": 1.4039963263150489e-05,
      "loss": 1.776,
      "step": 207877
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.3709226846694946,
      "learning_rate": 1.4038726548821543e-05,
      "loss": 1.9,
      "step": 207878
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.408552646636963,
      "learning_rate": 1.4037489887658137e-05,
      "loss": 2.0439,
      "step": 207879
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.4137063026428223,
      "learning_rate": 1.4036253279660604e-05,
      "loss": 1.9479,
      "step": 207880
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.387197494506836,
      "learning_rate": 1.4035016724829108e-05,
      "loss": 1.7119,
      "step": 207881
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.5465419292449951,
      "learning_rate": 1.4033780223163948e-05,
      "loss": 2.0246,
      "step": 207882
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.4285074472427368,
      "learning_rate": 1.4032543774665228e-05,
      "loss": 2.0014,
      "step": 207883
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.4045840501785278,
      "learning_rate": 1.4031307379333312e-05,
      "loss": 2.0741,
      "step": 207884
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.437381386756897,
      "learning_rate": 1.4030071037168333e-05,
      "loss": 1.7544,
      "step": 207885
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.3244560956954956,
      "learning_rate": 1.4028834748170592e-05,
      "loss": 1.8467,
      "step": 207886
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.2944507598876953,
      "learning_rate": 1.4027598512340288e-05,
      "loss": 1.8594,
      "step": 207887
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.3482775688171387,
      "learning_rate": 1.4026362329677654e-05,
      "loss": 1.9494,
      "step": 207888
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.6367571353912354,
      "learning_rate": 1.402512620018289e-05,
      "loss": 1.9897,
      "step": 207889
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.580161213874817,
      "learning_rate": 1.4023890123856296e-05,
      "loss": 2.1529,
      "step": 207890
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.451296091079712,
      "learning_rate": 1.4022654100698006e-05,
      "loss": 2.1326,
      "step": 207891
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.3458226919174194,
      "learning_rate": 1.4021418130708351e-05,
      "loss": 2.0339,
      "step": 207892
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.5009938478469849,
      "learning_rate": 1.4020182213887499e-05,
      "loss": 2.1225,
      "step": 207893
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.3385530710220337,
      "learning_rate": 1.4018946350235648e-05,
      "loss": 1.9626,
      "step": 207894
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.3674507141113281,
      "learning_rate": 1.4017710539753135e-05,
      "loss": 1.9421,
      "step": 207895
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.4837007522583008,
      "learning_rate": 1.4016474782440123e-05,
      "loss": 1.9976,
      "step": 207896
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.4317511320114136,
      "learning_rate": 1.4015239078296814e-05,
      "loss": 1.6663,
      "step": 207897
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.4904662370681763,
      "learning_rate": 1.4014003427323473e-05,
      "loss": 2.0504,
      "step": 207898
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.473522424697876,
      "learning_rate": 1.4012767829520365e-05,
      "loss": 1.8933,
      "step": 207899
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.676060676574707,
      "learning_rate": 1.4011532284887628e-05,
      "loss": 2.1243,
      "step": 207900
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.3590656518936157,
      "learning_rate": 1.401029679342559e-05,
      "loss": 1.8177,
      "step": 207901
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.4476674795150757,
      "learning_rate": 1.400906135513442e-05,
      "loss": 2.0473,
      "step": 207902
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.413343906402588,
      "learning_rate": 1.4007825970014352e-05,
      "loss": 2.0004,
      "step": 207903
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.4694796800613403,
      "learning_rate": 1.4006590638065652e-05,
      "loss": 1.896,
      "step": 207904
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.332388997077942,
      "learning_rate": 1.4005355359288517e-05,
      "loss": 2.0637,
      "step": 207905
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.4365705251693726,
      "learning_rate": 1.4004120133683183e-05,
      "loss": 1.9706,
      "step": 207906
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.547875165939331,
      "learning_rate": 1.4002884961249883e-05,
      "loss": 1.9621,
      "step": 207907
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.425544023513794,
      "learning_rate": 1.4001649841988816e-05,
      "loss": 2.1095,
      "step": 207908
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.6182876825332642,
      "learning_rate": 1.4000414775900281e-05,
      "loss": 1.975,
      "step": 207909
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.339355707168579,
      "learning_rate": 1.399917976298448e-05,
      "loss": 1.9421,
      "step": 207910
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.64814031124115,
      "learning_rate": 1.3997944803241612e-05,
      "loss": 1.9703,
      "step": 207911
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.604947566986084,
      "learning_rate": 1.3996709896671876e-05,
      "loss": 1.9893,
      "step": 207912
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.3753747940063477,
      "learning_rate": 1.3995475043275606e-05,
      "loss": 2.0398,
      "step": 207913
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.5419100522994995,
      "learning_rate": 1.3994240243052935e-05,
      "loss": 1.8949,
      "step": 207914
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.5895881652832031,
      "learning_rate": 1.3993005496004195e-05,
      "loss": 2.1227,
      "step": 207915
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.567635178565979,
      "learning_rate": 1.3991770802129488e-05,
      "loss": 2.1992,
      "step": 207916
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.4407626390457153,
      "learning_rate": 1.3990536161429144e-05,
      "loss": 1.8299,
      "step": 207917
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.5453722476959229,
      "learning_rate": 1.3989301573903366e-05,
      "loss": 1.9956,
      "step": 207918
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.4535959959030151,
      "learning_rate": 1.3988067039552387e-05,
      "loss": 1.8997,
      "step": 207919
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.3362542390823364,
      "learning_rate": 1.398683255837637e-05,
      "loss": 2.0817,
      "step": 207920
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.4566783905029297,
      "learning_rate": 1.3985598130375653e-05,
      "loss": 1.9531,
      "step": 207921
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.43743896484375,
      "learning_rate": 1.3984363755550365e-05,
      "loss": 1.6458,
      "step": 207922
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.424056053161621,
      "learning_rate": 1.3983129433900809e-05,
      "loss": 1.6944,
      "step": 207923
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.4752929210662842,
      "learning_rate": 1.3981895165427216e-05,
      "loss": 1.8604,
      "step": 207924
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.395003318786621,
      "learning_rate": 1.3980660950129785e-05,
      "loss": 1.883,
      "step": 207925
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.5632275342941284,
      "learning_rate": 1.3979426788008686e-05,
      "loss": 2.0859,
      "step": 207926
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.4588439464569092,
      "learning_rate": 1.3978192679064248e-05,
      "loss": 1.9169,
      "step": 207927
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.3140968084335327,
      "learning_rate": 1.3976958623296642e-05,
      "loss": 2.012,
      "step": 207928
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.4742236137390137,
      "learning_rate": 1.3975724620706164e-05,
      "loss": 2.1549,
      "step": 207929
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.304163932800293,
      "learning_rate": 1.3974490671292981e-05,
      "loss": 1.7599,
      "step": 207930
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.4331110715866089,
      "learning_rate": 1.3973256775057363e-05,
      "loss": 1.8928,
      "step": 207931
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.542405366897583,
      "learning_rate": 1.3972022931999438e-05,
      "loss": 1.9844,
      "step": 207932
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.3377543687820435,
      "learning_rate": 1.3970789142119576e-05,
      "loss": 2.0372,
      "step": 207933
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.5355958938598633,
      "learning_rate": 1.396955540541791e-05,
      "loss": 1.9662,
      "step": 207934
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.6463364362716675,
      "learning_rate": 1.3968321721894738e-05,
      "loss": 1.9052,
      "step": 207935
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.4488189220428467,
      "learning_rate": 1.3967088091550227e-05,
      "loss": 1.9777,
      "step": 207936
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.4908074140548706,
      "learning_rate": 1.3965854514384678e-05,
      "loss": 1.9372,
      "step": 207937
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.4980591535568237,
      "learning_rate": 1.396462099039819e-05,
      "loss": 1.973,
      "step": 207938
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.4345524311065674,
      "learning_rate": 1.3963387519591162e-05,
      "loss": 1.739,
      "step": 207939
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.404894471168518,
      "learning_rate": 1.3962154101963664e-05,
      "loss": 2.1108,
      "step": 207940
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.5137825012207031,
      "learning_rate": 1.3960920737516056e-05,
      "loss": 1.9316,
      "step": 207941
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.4265116453170776,
      "learning_rate": 1.3959687426248511e-05,
      "loss": 1.9425,
      "step": 207942
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.4471466541290283,
      "learning_rate": 1.3958454168161293e-05,
      "loss": 2.0667,
      "step": 207943
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.5102307796478271,
      "learning_rate": 1.3957220963254501e-05,
      "loss": 2.1215,
      "step": 207944
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.553009033203125,
      "learning_rate": 1.3955987811528536e-05,
      "loss": 1.8433,
      "step": 207945
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.4552074670791626,
      "learning_rate": 1.3954754712983497e-05,
      "loss": 1.8822,
      "step": 207946
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.4490647315979004,
      "learning_rate": 1.3953521667619716e-05,
      "loss": 1.9482,
      "step": 207947
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.5445762872695923,
      "learning_rate": 1.3952288675437395e-05,
      "loss": 2.0964,
      "step": 207948
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.4792630672454834,
      "learning_rate": 1.39510557364367e-05,
      "loss": 1.932,
      "step": 207949
    },
    {
      "epoch": 2.71,
      "grad_norm": 2.6478142738342285,
      "learning_rate": 1.3949822850617897e-05,
      "loss": 2.2073,
      "step": 207950
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.3861545324325562,
      "learning_rate": 1.3948590017981254e-05,
      "loss": 1.9106,
      "step": 207951
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.488331913948059,
      "learning_rate": 1.3947357238526935e-05,
      "loss": 1.9712,
      "step": 207952
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.3458412885665894,
      "learning_rate": 1.3946124512255209e-05,
      "loss": 2.0532,
      "step": 207953
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.5712305307388306,
      "learning_rate": 1.3944891839166338e-05,
      "loss": 2.1045,
      "step": 207954
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.5166065692901611,
      "learning_rate": 1.3943659219260495e-05,
      "loss": 2.2528,
      "step": 207955
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.381147027015686,
      "learning_rate": 1.3942426652537875e-05,
      "loss": 2.1009,
      "step": 207956
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.416001319885254,
      "learning_rate": 1.3941194138998813e-05,
      "loss": 1.9421,
      "step": 207957
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.5163313150405884,
      "learning_rate": 1.3939961678643441e-05,
      "loss": 2.0907,
      "step": 207958
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.4352315664291382,
      "learning_rate": 1.393872927147206e-05,
      "loss": 1.8833,
      "step": 207959
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.3788151741027832,
      "learning_rate": 1.393749691748487e-05,
      "loss": 1.8605,
      "step": 207960
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.4371753931045532,
      "learning_rate": 1.3936264616682102e-05,
      "loss": 2.0216,
      "step": 207961
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.4340612888336182,
      "learning_rate": 1.3935032369063926e-05,
      "loss": 1.752,
      "step": 207962
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.3981817960739136,
      "learning_rate": 1.3933800174630704e-05,
      "loss": 1.7815,
      "step": 207963
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.4850232601165771,
      "learning_rate": 1.3932568033382541e-05,
      "loss": 1.9584,
      "step": 207964
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.355370283126831,
      "learning_rate": 1.3931335945319732e-05,
      "loss": 1.7452,
      "step": 207965
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.367994785308838,
      "learning_rate": 1.3930103910442481e-05,
      "loss": 1.925,
      "step": 207966
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.4808388948440552,
      "learning_rate": 1.3928871928751051e-05,
      "loss": 1.9738,
      "step": 207967
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.4860787391662598,
      "learning_rate": 1.3927640000245577e-05,
      "loss": 1.883,
      "step": 207968
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.5249195098876953,
      "learning_rate": 1.3926408124926424e-05,
      "loss": 2.0098,
      "step": 207969
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.3903188705444336,
      "learning_rate": 1.3925176302793695e-05,
      "loss": 1.7752,
      "step": 207970
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.4003015756607056,
      "learning_rate": 1.392394453384772e-05,
      "loss": 1.9836,
      "step": 207971
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.4142078161239624,
      "learning_rate": 1.3922712818088699e-05,
      "loss": 1.9049,
      "step": 207972
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.4324284791946411,
      "learning_rate": 1.3921481155516833e-05,
      "loss": 1.9584,
      "step": 207973
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.48955237865448,
      "learning_rate": 1.3920249546132323e-05,
      "loss": 2.0049,
      "step": 207974
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.655644178390503,
      "learning_rate": 1.3919017989935466e-05,
      "loss": 1.9772,
      "step": 207975
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.623995065689087,
      "learning_rate": 1.391778648692643e-05,
      "loss": 1.9371,
      "step": 207976
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.3644838333129883,
      "learning_rate": 1.3916555037105515e-05,
      "loss": 2.0559,
      "step": 207977
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.4130897521972656,
      "learning_rate": 1.391532364047292e-05,
      "loss": 2.0616,
      "step": 207978
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.4472140073776245,
      "learning_rate": 1.3914092297028879e-05,
      "loss": 1.8805,
      "step": 207979
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.457281470298767,
      "learning_rate": 1.3912861006773557e-05,
      "loss": 2.0143,
      "step": 207980
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.4399513006210327,
      "learning_rate": 1.3911629769707255e-05,
      "loss": 2.0956,
      "step": 207981
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.4199512004852295,
      "learning_rate": 1.3910398585830173e-05,
      "loss": 2.0875,
      "step": 207982
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.489159345626831,
      "learning_rate": 1.3909167455142578e-05,
      "loss": 1.9382,
      "step": 207983
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.5110279321670532,
      "learning_rate": 1.3907936377644635e-05,
      "loss": 2.0654,
      "step": 207984
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.5855706930160522,
      "learning_rate": 1.3906705353336611e-05,
      "loss": 2.0612,
      "step": 207985
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.633894920349121,
      "learning_rate": 1.3905474382218774e-05,
      "loss": 1.989,
      "step": 207986
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.580088496208191,
      "learning_rate": 1.3904243464291287e-05,
      "loss": 1.9057,
      "step": 207987
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.5262279510498047,
      "learning_rate": 1.3903012599554386e-05,
      "loss": 2.2184,
      "step": 207988
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.5737651586532593,
      "learning_rate": 1.3901781788008337e-05,
      "loss": 2.1201,
      "step": 207989
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.373397946357727,
      "learning_rate": 1.3900551029653307e-05,
      "loss": 1.7783,
      "step": 207990
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.436890721321106,
      "learning_rate": 1.3899320324489593e-05,
      "loss": 1.9871,
      "step": 207991
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.4596648216247559,
      "learning_rate": 1.3898089672517399e-05,
      "loss": 1.9503,
      "step": 207992
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.7056806087493896,
      "learning_rate": 1.3896859073736922e-05,
      "loss": 2.0149,
      "step": 207993
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.319506287574768,
      "learning_rate": 1.3895628528148461e-05,
      "loss": 1.868,
      "step": 207994
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.4768381118774414,
      "learning_rate": 1.3894398035752219e-05,
      "loss": 2.0521,
      "step": 207995
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.547297716140747,
      "learning_rate": 1.3893167596548328e-05,
      "loss": 1.9211,
      "step": 207996
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.397658109664917,
      "learning_rate": 1.3891937210537186e-05,
      "loss": 1.9646,
      "step": 207997
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.4436615705490112,
      "learning_rate": 1.3890706877718893e-05,
      "loss": 1.9079,
      "step": 207998
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.4651843309402466,
      "learning_rate": 1.3889476598093685e-05,
      "loss": 1.8014,
      "step": 207999
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.402570366859436,
      "learning_rate": 1.3888246371661894e-05,
      "loss": 1.9566,
      "step": 208000
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.49288809299469,
      "learning_rate": 1.3887016198423651e-05,
      "loss": 1.9439,
      "step": 208001
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.5906215906143188,
      "learning_rate": 1.3885786078379191e-05,
      "loss": 1.9496,
      "step": 208002
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.6552734375,
      "learning_rate": 1.388455601152878e-05,
      "loss": 2.1057,
      "step": 208003
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.432990312576294,
      "learning_rate": 1.3883325997872652e-05,
      "loss": 1.999,
      "step": 208004
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.5390781164169312,
      "learning_rate": 1.388209603741094e-05,
      "loss": 2.1271,
      "step": 208005
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.5967897176742554,
      "learning_rate": 1.3880866130144042e-05,
      "loss": 2.1061,
      "step": 208006
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.5342520475387573,
      "learning_rate": 1.3879636276071992e-05,
      "loss": 2.2402,
      "step": 208007
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.511272668838501,
      "learning_rate": 1.3878406475195191e-05,
      "loss": 2.1563,
      "step": 208008
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.5286298990249634,
      "learning_rate": 1.3877176727513806e-05,
      "loss": 1.8111,
      "step": 208009
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.4785773754119873,
      "learning_rate": 1.3875947033028034e-05,
      "loss": 1.9074,
      "step": 208010
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.475691795349121,
      "learning_rate": 1.3874717391738111e-05,
      "loss": 2.0637,
      "step": 208011
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.512984037399292,
      "learning_rate": 1.38734878036443e-05,
      "loss": 1.9968,
      "step": 208012
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.5317727327346802,
      "learning_rate": 1.3872258268746772e-05,
      "loss": 2.0737,
      "step": 208013
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.4612075090408325,
      "learning_rate": 1.3871028787045823e-05,
      "loss": 2.0911,
      "step": 208014
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.4356811046600342,
      "learning_rate": 1.3869799358541655e-05,
      "loss": 2.1158,
      "step": 208015
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.6218599081039429,
      "learning_rate": 1.3868569983234502e-05,
      "loss": 1.9149,
      "step": 208016
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.3750489950180054,
      "learning_rate": 1.3867340661124526e-05,
      "loss": 1.812,
      "step": 208017
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.4237161874771118,
      "learning_rate": 1.3866111392212065e-05,
      "loss": 1.9114,
      "step": 208018
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.3901492357254028,
      "learning_rate": 1.3864882176497249e-05,
      "loss": 2.0831,
      "step": 208019
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.5860813856124878,
      "learning_rate": 1.3863653013980413e-05,
      "loss": 2.0289,
      "step": 208020
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.6162166595458984,
      "learning_rate": 1.386242390466169e-05,
      "loss": 1.8931,
      "step": 208021
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.4578582048416138,
      "learning_rate": 1.386119484854138e-05,
      "loss": 1.9109,
      "step": 208022
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.4903357028961182,
      "learning_rate": 1.3859965845619614e-05,
      "loss": 1.9681,
      "step": 208023
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.3369704484939575,
      "learning_rate": 1.3858736895896726e-05,
      "loss": 1.8975,
      "step": 208024
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.4133714437484741,
      "learning_rate": 1.3857507999372853e-05,
      "loss": 1.9587,
      "step": 208025
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.5607588291168213,
      "learning_rate": 1.3856279156048322e-05,
      "loss": 1.9463,
      "step": 208026
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.4275873899459839,
      "learning_rate": 1.3855050365923304e-05,
      "loss": 2.0294,
      "step": 208027
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.4668464660644531,
      "learning_rate": 1.385382162899803e-05,
      "loss": 1.8653,
      "step": 208028
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.4740463495254517,
      "learning_rate": 1.3852592945272701e-05,
      "loss": 1.9596,
      "step": 208029
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.512830376625061,
      "learning_rate": 1.3851364314747615e-05,
      "loss": 1.9254,
      "step": 208030
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.4113273620605469,
      "learning_rate": 1.3850135737422907e-05,
      "loss": 2.1739,
      "step": 208031
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.4524296522140503,
      "learning_rate": 1.384890721329891e-05,
      "loss": 1.968,
      "step": 208032
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.5055406093597412,
      "learning_rate": 1.384767874237579e-05,
      "loss": 1.9194,
      "step": 208033
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.3621127605438232,
      "learning_rate": 1.3846450324653812e-05,
      "loss": 1.9864,
      "step": 208034
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.4475902318954468,
      "learning_rate": 1.3845221960133145e-05,
      "loss": 1.9284,
      "step": 208035
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.3922489881515503,
      "learning_rate": 1.3843993648814055e-05,
      "loss": 2.1013,
      "step": 208036
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.5453197956085205,
      "learning_rate": 1.3842765390696774e-05,
      "loss": 1.8997,
      "step": 208037
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.4463666677474976,
      "learning_rate": 1.3841537185781538e-05,
      "loss": 1.9314,
      "step": 208038
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.4532718658447266,
      "learning_rate": 1.3840309034068575e-05,
      "loss": 1.8626,
      "step": 208039
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.4971166849136353,
      "learning_rate": 1.3839080935558088e-05,
      "loss": 1.9792,
      "step": 208040
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.3852591514587402,
      "learning_rate": 1.3837852890250278e-05,
      "loss": 2.0827,
      "step": 208041
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.4895673990249634,
      "learning_rate": 1.3836624898145442e-05,
      "loss": 2.1663,
      "step": 208042
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.5356011390686035,
      "learning_rate": 1.383539695924375e-05,
      "loss": 2.2569,
      "step": 208043
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.5347845554351807,
      "learning_rate": 1.3834169073545498e-05,
      "loss": 1.95,
      "step": 208044
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.6640740633010864,
      "learning_rate": 1.3832941241050887e-05,
      "loss": 1.8736,
      "step": 208045
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.5272886753082275,
      "learning_rate": 1.3831713461760152e-05,
      "loss": 2.0815,
      "step": 208046
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.4886754751205444,
      "learning_rate": 1.3830485735673424e-05,
      "loss": 1.936,
      "step": 208047
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.4876786470413208,
      "learning_rate": 1.382925806279107e-05,
      "loss": 1.9214,
      "step": 208048
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.4406318664550781,
      "learning_rate": 1.3828030443113192e-05,
      "loss": 1.9239,
      "step": 208049
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.4772902727127075,
      "learning_rate": 1.3826802876640152e-05,
      "loss": 1.9905,
      "step": 208050
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.6366329193115234,
      "learning_rate": 1.3825575363372088e-05,
      "loss": 1.8902,
      "step": 208051
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.4253886938095093,
      "learning_rate": 1.3824347903309296e-05,
      "loss": 2.0235,
      "step": 208052
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.5379517078399658,
      "learning_rate": 1.3823120496451912e-05,
      "loss": 1.9192,
      "step": 208053
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.4419723749160767,
      "learning_rate": 1.3821893142800234e-05,
      "loss": 2.0577,
      "step": 208054
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.4733861684799194,
      "learning_rate": 1.3820665842354427e-05,
      "loss": 1.9186,
      "step": 208055
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.4565660953521729,
      "learning_rate": 1.3819438595114796e-05,
      "loss": 1.8682,
      "step": 208056
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.4283294677734375,
      "learning_rate": 1.3818211401081503e-05,
      "loss": 1.8789,
      "step": 208057
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.4553862810134888,
      "learning_rate": 1.3816984260254882e-05,
      "loss": 1.9396,
      "step": 208058
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.4112569093704224,
      "learning_rate": 1.3815757172635e-05,
      "loss": 1.7228,
      "step": 208059
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.4065864086151123,
      "learning_rate": 1.3814530138222224e-05,
      "loss": 2.0046,
      "step": 208060
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.3449769020080566,
      "learning_rate": 1.3813303157016653e-05,
      "loss": 1.9083,
      "step": 208061
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.7535990476608276,
      "learning_rate": 1.3812076229018687e-05,
      "loss": 2.0703,
      "step": 208062
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.502506971359253,
      "learning_rate": 1.3810849354228393e-05,
      "loss": 2.1249,
      "step": 208063
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.444717526435852,
      "learning_rate": 1.3809622532646136e-05,
      "loss": 1.85,
      "step": 208064
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.4791102409362793,
      "learning_rate": 1.3808395764271985e-05,
      "loss": 1.7566,
      "step": 208065
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.484258770942688,
      "learning_rate": 1.3807169049106304e-05,
      "loss": 1.9146,
      "step": 208066
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.439125895500183,
      "learning_rate": 1.380594238714926e-05,
      "loss": 2.1142,
      "step": 208067
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.5942410230636597,
      "learning_rate": 1.3804715778401087e-05,
      "loss": 2.0377,
      "step": 208068
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.4452491998672485,
      "learning_rate": 1.3803489222862018e-05,
      "loss": 1.94,
      "step": 208069
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.5919663906097412,
      "learning_rate": 1.380226272053232e-05,
      "loss": 1.9854,
      "step": 208070
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.4160454273223877,
      "learning_rate": 1.3801036271412157e-05,
      "loss": 2.016,
      "step": 208071
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.50005042552948,
      "learning_rate": 1.37998098755018e-05,
      "loss": 2.2394,
      "step": 208072
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.5622531175613403,
      "learning_rate": 1.379858353280141e-05,
      "loss": 1.9271,
      "step": 208073
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.6432565450668335,
      "learning_rate": 1.3797357243311324e-05,
      "loss": 1.9257,
      "step": 208074
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.4807491302490234,
      "learning_rate": 1.3796131007031674e-05,
      "loss": 1.8847,
      "step": 208075
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.4095572233200073,
      "learning_rate": 1.379490482396276e-05,
      "loss": 1.9932,
      "step": 208076
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.4205985069274902,
      "learning_rate": 1.3793678694104748e-05,
      "loss": 1.8693,
      "step": 208077
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.6218810081481934,
      "learning_rate": 1.3792452617457905e-05,
      "loss": 1.9921,
      "step": 208078
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.4829002618789673,
      "learning_rate": 1.3791226594022465e-05,
      "loss": 1.9194,
      "step": 208079
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.4112091064453125,
      "learning_rate": 1.3790000623798625e-05,
      "loss": 2.0739,
      "step": 208080
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.4119033813476562,
      "learning_rate": 1.3788774706786587e-05,
      "loss": 2.0667,
      "step": 208081
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.4979535341262817,
      "learning_rate": 1.3787548842986652e-05,
      "loss": 1.9848,
      "step": 208082
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.4523677825927734,
      "learning_rate": 1.378632303239905e-05,
      "loss": 2.1455,
      "step": 208083
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.4179502725601196,
      "learning_rate": 1.3785097275023915e-05,
      "loss": 1.9435,
      "step": 208084
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.388922929763794,
      "learning_rate": 1.3783871570861549e-05,
      "loss": 2.1623,
      "step": 208085
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.4130247831344604,
      "learning_rate": 1.3782645919912183e-05,
      "loss": 2.0786,
      "step": 208086
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.448071837425232,
      "learning_rate": 1.3781420322175984e-05,
      "loss": 2.0735,
      "step": 208087
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.4380744695663452,
      "learning_rate": 1.3780194777653286e-05,
      "loss": 2.0647,
      "step": 208088
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.5011229515075684,
      "learning_rate": 1.3778969286344221e-05,
      "loss": 1.8674,
      "step": 208089
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.3510944843292236,
      "learning_rate": 1.3777743848249024e-05,
      "loss": 1.8209,
      "step": 208090
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.4361495971679688,
      "learning_rate": 1.3776518463367992e-05,
      "loss": 1.9632,
      "step": 208091
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.4369667768478394,
      "learning_rate": 1.377529313170126e-05,
      "loss": 1.9119,
      "step": 208092
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.4954990148544312,
      "learning_rate": 1.3774067853249126e-05,
      "loss": 2.14,
      "step": 208093
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.5674843788146973,
      "learning_rate": 1.3772842628011826e-05,
      "loss": 2.1453,
      "step": 208094
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.4738788604736328,
      "learning_rate": 1.3771617455989525e-05,
      "loss": 2.0638,
      "step": 208095
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.5456403493881226,
      "learning_rate": 1.3770392337182489e-05,
      "loss": 2.1593,
      "step": 208096
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.4446998834609985,
      "learning_rate": 1.3769167271590953e-05,
      "loss": 2.0499,
      "step": 208097
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.31379234790802,
      "learning_rate": 1.3767942259215081e-05,
      "loss": 1.9663,
      "step": 208098
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.4417732954025269,
      "learning_rate": 1.3766717300055207e-05,
      "loss": 2.0073,
      "step": 208099
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.5373728275299072,
      "learning_rate": 1.37654923941115e-05,
      "loss": 1.9432,
      "step": 208100
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.6324824094772339,
      "learning_rate": 1.3764267541384189e-05,
      "loss": 2.0096,
      "step": 208101
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.4513047933578491,
      "learning_rate": 1.3763042741873475e-05,
      "loss": 1.8813,
      "step": 208102
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.6640554666519165,
      "learning_rate": 1.3761817995579627e-05,
      "loss": 2.0637,
      "step": 208103
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.4315619468688965,
      "learning_rate": 1.3760593302502842e-05,
      "loss": 2.0657,
      "step": 208104
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.4895985126495361,
      "learning_rate": 1.3759368662643422e-05,
      "loss": 1.8489,
      "step": 208105
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.3578885793685913,
      "learning_rate": 1.3758144076001498e-05,
      "loss": 1.9774,
      "step": 208106
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.4764893054962158,
      "learning_rate": 1.3756919542577372e-05,
      "loss": 2.0408,
      "step": 208107
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.3344601392745972,
      "learning_rate": 1.3755695062371175e-05,
      "loss": 1.9221,
      "step": 208108
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.3742976188659668,
      "learning_rate": 1.375447063538324e-05,
      "loss": 1.9314,
      "step": 208109
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.5265415906906128,
      "learning_rate": 1.3753246261613737e-05,
      "loss": 1.9685,
      "step": 208110
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.38724684715271,
      "learning_rate": 1.3752021941062929e-05,
      "loss": 2.1807,
      "step": 208111
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.4302387237548828,
      "learning_rate": 1.3750797673731018e-05,
      "loss": 2.0319,
      "step": 208112
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.469914436340332,
      "learning_rate": 1.3749573459618235e-05,
      "loss": 2.2152,
      "step": 208113
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.4618186950683594,
      "learning_rate": 1.374834929872478e-05,
      "loss": 1.834,
      "step": 208114
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.5044329166412354,
      "learning_rate": 1.3747125191050955e-05,
      "loss": 2.1202,
      "step": 208115
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.421452522277832,
      "learning_rate": 1.3745901136596893e-05,
      "loss": 1.7699,
      "step": 208116
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.4933809041976929,
      "learning_rate": 1.3744677135362925e-05,
      "loss": 2.0596,
      "step": 208117
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.353271245956421,
      "learning_rate": 1.3743453187349185e-05,
      "loss": 2.2217,
      "step": 208118
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.6627413034439087,
      "learning_rate": 1.3742229292556006e-05,
      "loss": 1.8309,
      "step": 208119
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.564365029335022,
      "learning_rate": 1.3741005450983455e-05,
      "loss": 2.0561,
      "step": 208120
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.3843392133712769,
      "learning_rate": 1.3739781662631933e-05,
      "loss": 1.8611,
      "step": 208121
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.4565318822860718,
      "learning_rate": 1.3738557927501536e-05,
      "loss": 1.9174,
      "step": 208122
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.4109903573989868,
      "learning_rate": 1.3737334245592568e-05,
      "loss": 1.986,
      "step": 208123
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.42258882522583,
      "learning_rate": 1.3736110616905227e-05,
      "loss": 1.9143,
      "step": 208124
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.401935338973999,
      "learning_rate": 1.373488704143978e-05,
      "loss": 1.9007,
      "step": 208125
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.3994085788726807,
      "learning_rate": 1.3733663519196391e-05,
      "loss": 1.8866,
      "step": 208126
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.4782060384750366,
      "learning_rate": 1.373244005017533e-05,
      "loss": 1.8952,
      "step": 208127
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.4850455522537231,
      "learning_rate": 1.3731216634376763e-05,
      "loss": 1.9643,
      "step": 208128
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.6978923082351685,
      "learning_rate": 1.3729993271801021e-05,
      "loss": 2.3106,
      "step": 208129
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.517998218536377,
      "learning_rate": 1.3728769962448239e-05,
      "loss": 2.0529,
      "step": 208130
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.4780871868133545,
      "learning_rate": 1.3727546706318748e-05,
      "loss": 1.9928,
      "step": 208131
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.5429834127426147,
      "learning_rate": 1.372632350341265e-05,
      "loss": 2.0584,
      "step": 208132
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.372499942779541,
      "learning_rate": 1.3725100353730246e-05,
      "loss": 1.8427,
      "step": 208133
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.5089631080627441,
      "learning_rate": 1.372387725727173e-05,
      "loss": 1.991,
      "step": 208134
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.3196817636489868,
      "learning_rate": 1.3722654214037376e-05,
      "loss": 2.1766,
      "step": 208135
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.4410359859466553,
      "learning_rate": 1.3721431224027379e-05,
      "loss": 1.9761,
      "step": 208136
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.5111109018325806,
      "learning_rate": 1.3720208287242008e-05,
      "loss": 1.9354,
      "step": 208137
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.4567023515701294,
      "learning_rate": 1.3718985403681392e-05,
      "loss": 2.0772,
      "step": 208138
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.4228657484054565,
      "learning_rate": 1.3717762573345869e-05,
      "loss": 2.0123,
      "step": 208139
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.4165188074111938,
      "learning_rate": 1.371653979623557e-05,
      "loss": 1.9837,
      "step": 208140
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.432740330696106,
      "learning_rate": 1.3715317072350795e-05,
      "loss": 1.8992,
      "step": 208141
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.3379307985305786,
      "learning_rate": 1.3714094401691745e-05,
      "loss": 2.0102,
      "step": 208142
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.4684096574783325,
      "learning_rate": 1.3712871784258683e-05,
      "loss": 1.995,
      "step": 208143
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.414254903793335,
      "learning_rate": 1.3711649220051746e-05,
      "loss": 2.1202,
      "step": 208144
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.430060863494873,
      "learning_rate": 1.3710426709071265e-05,
      "loss": 2.159,
      "step": 208145
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.4491273164749146,
      "learning_rate": 1.3709204251317374e-05,
      "loss": 2.0643,
      "step": 208146
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.5305917263031006,
      "learning_rate": 1.3707981846790372e-05,
      "loss": 1.8593,
      "step": 208147
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.4021543264389038,
      "learning_rate": 1.370675949549046e-05,
      "loss": 2.0057,
      "step": 208148
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.3808073997497559,
      "learning_rate": 1.3705537197417904e-05,
      "loss": 2.1702,
      "step": 208149
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.394113302230835,
      "learning_rate": 1.3704314952572838e-05,
      "loss": 1.7917,
      "step": 208150
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.4515612125396729,
      "learning_rate": 1.370309276095556e-05,
      "loss": 1.9472,
      "step": 208151
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.5546047687530518,
      "learning_rate": 1.3701870622566268e-05,
      "loss": 2.0519,
      "step": 208152
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.5474895238876343,
      "learning_rate": 1.3700648537405202e-05,
      "loss": 2.0114,
      "step": 208153
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.6294162273406982,
      "learning_rate": 1.3699426505472589e-05,
      "loss": 1.8518,
      "step": 208154
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.5271673202514648,
      "learning_rate": 1.3698204526768697e-05,
      "loss": 1.8869,
      "step": 208155
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.4366283416748047,
      "learning_rate": 1.3696982601293694e-05,
      "loss": 1.9045,
      "step": 208156
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.640110731124878,
      "learning_rate": 1.3695760729047811e-05,
      "loss": 2.1045,
      "step": 208157
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.5934573411941528,
      "learning_rate": 1.3694538910031283e-05,
      "loss": 1.7183,
      "step": 208158
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.5009713172912598,
      "learning_rate": 1.3693317144244376e-05,
      "loss": 1.8911,
      "step": 208159
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.5032745599746704,
      "learning_rate": 1.3692095431687255e-05,
      "loss": 2.0375,
      "step": 208160
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.630672574043274,
      "learning_rate": 1.3690873772360189e-05,
      "loss": 1.9803,
      "step": 208161
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.3412630558013916,
      "learning_rate": 1.368965216626341e-05,
      "loss": 1.9192,
      "step": 208162
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.4554134607315063,
      "learning_rate": 1.368843061339715e-05,
      "loss": 2.0775,
      "step": 208163
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.3887654542922974,
      "learning_rate": 1.3687209113761544e-05,
      "loss": 1.879,
      "step": 208164
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.5053104162216187,
      "learning_rate": 1.3685987667356958e-05,
      "loss": 2.0406,
      "step": 208165
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.547871470451355,
      "learning_rate": 1.368476627418349e-05,
      "loss": 2.1774,
      "step": 208166
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.6924763917922974,
      "learning_rate": 1.3683544934241475e-05,
      "loss": 1.8777,
      "step": 208167
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.5436606407165527,
      "learning_rate": 1.368232364753108e-05,
      "loss": 1.896,
      "step": 208168
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.4644951820373535,
      "learning_rate": 1.3681102414052536e-05,
      "loss": 1.9072,
      "step": 208169
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.4841471910476685,
      "learning_rate": 1.3679881233806078e-05,
      "loss": 2.0284,
      "step": 208170
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.3934860229492188,
      "learning_rate": 1.3678660106791971e-05,
      "loss": 1.9062,
      "step": 208171
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.5336971282958984,
      "learning_rate": 1.3677439033010352e-05,
      "loss": 1.8259,
      "step": 208172
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.43137526512146,
      "learning_rate": 1.3676218012461548e-05,
      "loss": 1.9142,
      "step": 208173
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.4111738204956055,
      "learning_rate": 1.367499704514573e-05,
      "loss": 1.99,
      "step": 208174
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.5323035717010498,
      "learning_rate": 1.3673776131063097e-05,
      "loss": 1.9741,
      "step": 208175
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.3846323490142822,
      "learning_rate": 1.367255527021398e-05,
      "loss": 2.1689,
      "step": 208176
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.3994754552841187,
      "learning_rate": 1.3671334462598449e-05,
      "loss": 2.047,
      "step": 208177
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.3072738647460938,
      "learning_rate": 1.36701137082169e-05,
      "loss": 1.9236,
      "step": 208178
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.5835216045379639,
      "learning_rate": 1.366889300706947e-05,
      "loss": 1.8353,
      "step": 208179
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.4260814189910889,
      "learning_rate": 1.366767235915639e-05,
      "loss": 1.8437,
      "step": 208180
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.4407963752746582,
      "learning_rate": 1.3666451764477892e-05,
      "loss": 1.9013,
      "step": 208181
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.4968228340148926,
      "learning_rate": 1.3665231223034213e-05,
      "loss": 1.9661,
      "step": 208182
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.5997215509414673,
      "learning_rate": 1.3664010734825514e-05,
      "loss": 2.0755,
      "step": 208183
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.4068310260772705,
      "learning_rate": 1.3662790299852167e-05,
      "loss": 2.0041,
      "step": 208184
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.5038418769836426,
      "learning_rate": 1.3661569918114268e-05,
      "loss": 1.969,
      "step": 208185
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.568365454673767,
      "learning_rate": 1.3660349589612119e-05,
      "loss": 2.0573,
      "step": 208186
    },
    {
      "epoch": 2.71,
      "grad_norm": 2.109916925430298,
      "learning_rate": 1.3659129314345884e-05,
      "loss": 2.0625,
      "step": 208187
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.4464138746261597,
      "learning_rate": 1.3657909092315834e-05,
      "loss": 1.8369,
      "step": 208188
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.515231966972351,
      "learning_rate": 1.3656688923522163e-05,
      "loss": 2.0602,
      "step": 208189
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.5393128395080566,
      "learning_rate": 1.3655468807965142e-05,
      "loss": 1.9654,
      "step": 208190
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.5252927541732788,
      "learning_rate": 1.3654248745644935e-05,
      "loss": 2.1151,
      "step": 208191
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.335618495941162,
      "learning_rate": 1.365302873656191e-05,
      "loss": 1.9462,
      "step": 208192
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.4457885026931763,
      "learning_rate": 1.36518087807161e-05,
      "loss": 1.7511,
      "step": 208193
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.5566335916519165,
      "learning_rate": 1.3650588878107838e-05,
      "loss": 1.9027,
      "step": 208194
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.5672398805618286,
      "learning_rate": 1.3649369028737323e-05,
      "loss": 1.86,
      "step": 208195
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.5288907289505005,
      "learning_rate": 1.3648149232604854e-05,
      "loss": 2.0233,
      "step": 208196
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.4664121866226196,
      "learning_rate": 1.3646929489710534e-05,
      "loss": 1.9172,
      "step": 208197
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.529560923576355,
      "learning_rate": 1.3645709800054726e-05,
      "loss": 1.9133,
      "step": 208198
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.5147289037704468,
      "learning_rate": 1.3644490163637533e-05,
      "loss": 1.9602,
      "step": 208199
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.392828106880188,
      "learning_rate": 1.3643270580459253e-05,
      "loss": 1.8418,
      "step": 208200
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.4957488775253296,
      "learning_rate": 1.3642051050520052e-05,
      "loss": 1.9025,
      "step": 208201
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.4076054096221924,
      "learning_rate": 1.3640831573820266e-05,
      "loss": 1.9656,
      "step": 208202
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.416093111038208,
      "learning_rate": 1.3639612150359992e-05,
      "loss": 1.9307,
      "step": 208203
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.4271725416183472,
      "learning_rate": 1.3638392780139595e-05,
      "loss": 1.867,
      "step": 208204
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.496476173400879,
      "learning_rate": 1.3637173463159179e-05,
      "loss": 1.8892,
      "step": 208205
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.73390531539917,
      "learning_rate": 1.363595419941904e-05,
      "loss": 1.802,
      "step": 208206
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.301479697227478,
      "learning_rate": 1.3634734988919316e-05,
      "loss": 1.7663,
      "step": 208207
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.4056192636489868,
      "learning_rate": 1.3633515831660368e-05,
      "loss": 1.6079,
      "step": 208208
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.5243637561798096,
      "learning_rate": 1.36322967276423e-05,
      "loss": 1.8583,
      "step": 208209
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.4955437183380127,
      "learning_rate": 1.363107767686551e-05,
      "loss": 2.1021,
      "step": 208210
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.4432618618011475,
      "learning_rate": 1.3629858679329997e-05,
      "loss": 2.0219,
      "step": 208211
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.5101699829101562,
      "learning_rate": 1.3628639735036129e-05,
      "loss": 1.9611,
      "step": 208212
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.4003064632415771,
      "learning_rate": 1.3627420843984072e-05,
      "loss": 2.2318,
      "step": 208213
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.459899663925171,
      "learning_rate": 1.3626202006174158e-05,
      "loss": 2.1499,
      "step": 208214
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.4894534349441528,
      "learning_rate": 1.3624983221606456e-05,
      "loss": 2.1946,
      "step": 208215
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.4462741613388062,
      "learning_rate": 1.3623764490281364e-05,
      "loss": 1.9435,
      "step": 208216
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.4042421579360962,
      "learning_rate": 1.3622545812198948e-05,
      "loss": 2.0316,
      "step": 208217
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.7038427591323853,
      "learning_rate": 1.3621327187359543e-05,
      "loss": 2.0085,
      "step": 208218
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.4594461917877197,
      "learning_rate": 1.3620108615763314e-05,
      "loss": 2.1263,
      "step": 208219
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.4612963199615479,
      "learning_rate": 1.3618890097410529e-05,
      "loss": 2.0706,
      "step": 208220
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.442238450050354,
      "learning_rate": 1.3617671632301353e-05,
      "loss": 1.983,
      "step": 208221
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.4652668237686157,
      "learning_rate": 1.3616453220436153e-05,
      "loss": 2.2117,
      "step": 208222
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.6035112142562866,
      "learning_rate": 1.3615234861814995e-05,
      "loss": 1.9317,
      "step": 208223
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.4209516048431396,
      "learning_rate": 1.3614016556438179e-05,
      "loss": 2.0912,
      "step": 208224
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.4635932445526123,
      "learning_rate": 1.3612798304305906e-05,
      "loss": 1.9661,
      "step": 208225
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.4453786611557007,
      "learning_rate": 1.3611580105418474e-05,
      "loss": 1.9754,
      "step": 208226
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.728811264038086,
      "learning_rate": 1.3610361959775985e-05,
      "loss": 2.1839,
      "step": 208227
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.4579437971115112,
      "learning_rate": 1.3609143867378835e-05,
      "loss": 1.8258,
      "step": 208228
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.3997876644134521,
      "learning_rate": 1.3607925828227061e-05,
      "loss": 1.9841,
      "step": 208229
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.4071563482284546,
      "learning_rate": 1.3606707842320997e-05,
      "loss": 1.9556,
      "step": 208230
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.458536982536316,
      "learning_rate": 1.3605489909660837e-05,
      "loss": 1.9321,
      "step": 208231
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.4309664964675903,
      "learning_rate": 1.3604272030246855e-05,
      "loss": 2.0146,
      "step": 208232
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.4481900930404663,
      "learning_rate": 1.3603054204079212e-05,
      "loss": 2.0158,
      "step": 208233
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.3901385068893433,
      "learning_rate": 1.3601836431158241e-05,
      "loss": 2.1247,
      "step": 208234
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.3886237144470215,
      "learning_rate": 1.3600618711484046e-05,
      "loss": 1.9906,
      "step": 208235
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.6043529510498047,
      "learning_rate": 1.3599401045056923e-05,
      "loss": 2.108,
      "step": 208236
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.616262674331665,
      "learning_rate": 1.3598183431877007e-05,
      "loss": 2.0798,
      "step": 208237
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.6800522804260254,
      "learning_rate": 1.3596965871944665e-05,
      "loss": 2.1052,
      "step": 208238
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.4219595193862915,
      "learning_rate": 1.3595748365259996e-05,
      "loss": 2.1552,
      "step": 208239
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.4382059574127197,
      "learning_rate": 1.3594530911823331e-05,
      "loss": 1.9916,
      "step": 208240
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.5667749643325806,
      "learning_rate": 1.3593313511634873e-05,
      "loss": 2.071,
      "step": 208241
    },
    {
      "epoch": 2.71,
      "grad_norm": 2.432832717895508,
      "learning_rate": 1.3592096164694788e-05,
      "loss": 2.0209,
      "step": 208242
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.3208831548690796,
      "learning_rate": 1.3590878871003341e-05,
      "loss": 1.907,
      "step": 208243
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.4420099258422852,
      "learning_rate": 1.3589661630560767e-05,
      "loss": 2.0751,
      "step": 208244
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.499208927154541,
      "learning_rate": 1.358844444336723e-05,
      "loss": 1.8997,
      "step": 208245
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.4692926406860352,
      "learning_rate": 1.3587227309423065e-05,
      "loss": 2.0861,
      "step": 208246
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.3100978136062622,
      "learning_rate": 1.358601022872844e-05,
      "loss": 1.8077,
      "step": 208247
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.5364248752593994,
      "learning_rate": 1.3584793201283583e-05,
      "loss": 1.8918,
      "step": 208248
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.4354883432388306,
      "learning_rate": 1.3583576227088667e-05,
      "loss": 1.8854,
      "step": 208249
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.5110676288604736,
      "learning_rate": 1.358235930614402e-05,
      "loss": 2.1124,
      "step": 208250
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.5164601802825928,
      "learning_rate": 1.3581142438449777e-05,
      "loss": 1.9529,
      "step": 208251
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.4785103797912598,
      "learning_rate": 1.3579925624006238e-05,
      "loss": 1.9545,
      "step": 208252
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.6414124965667725,
      "learning_rate": 1.3578708862813636e-05,
      "loss": 2.0615,
      "step": 208253
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.3823236227035522,
      "learning_rate": 1.3577492154872071e-05,
      "loss": 1.8173,
      "step": 208254
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.4640554189682007,
      "learning_rate": 1.3576275500181911e-05,
      "loss": 1.9313,
      "step": 208255
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.295901894569397,
      "learning_rate": 1.3575058898743351e-05,
      "loss": 1.8721,
      "step": 208256
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.4031068086624146,
      "learning_rate": 1.357384235055653e-05,
      "loss": 1.8611,
      "step": 208257
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.4598681926727295,
      "learning_rate": 1.3572625855621778e-05,
      "loss": 1.8605,
      "step": 208258
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.5723079442977905,
      "learning_rate": 1.3571409413939294e-05,
      "loss": 2.2152,
      "step": 208259
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.4933634996414185,
      "learning_rate": 1.3570193025509246e-05,
      "loss": 2.0093,
      "step": 208260
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.5931552648544312,
      "learning_rate": 1.3568976690331934e-05,
      "loss": 2.0732,
      "step": 208261
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.4098222255706787,
      "learning_rate": 1.3567760408407591e-05,
      "loss": 1.9687,
      "step": 208262
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.5252933502197266,
      "learning_rate": 1.356654417973635e-05,
      "loss": 2.1013,
      "step": 208263
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.4361923933029175,
      "learning_rate": 1.356532800431851e-05,
      "loss": 1.9251,
      "step": 208264
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.293201208114624,
      "learning_rate": 1.3564111882154306e-05,
      "loss": 1.9285,
      "step": 208265
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.5510478019714355,
      "learning_rate": 1.3562895813243902e-05,
      "loss": 2.0475,
      "step": 208266
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.5056841373443604,
      "learning_rate": 1.3561679797587599e-05,
      "loss": 2.0038,
      "step": 208267
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.3475149869918823,
      "learning_rate": 1.3560463835185532e-05,
      "loss": 1.8673,
      "step": 208268
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.4470412731170654,
      "learning_rate": 1.355924792603803e-05,
      "loss": 1.995,
      "step": 208269
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.4100593328475952,
      "learning_rate": 1.3558032070145298e-05,
      "loss": 2.0499,
      "step": 208270
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.5970646142959595,
      "learning_rate": 1.3556816267507498e-05,
      "loss": 2.0936,
      "step": 208271
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.5501415729522705,
      "learning_rate": 1.3555600518124864e-05,
      "loss": 1.9278,
      "step": 208272
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.4331414699554443,
      "learning_rate": 1.3554384821997699e-05,
      "loss": 1.9135,
      "step": 208273
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.5792936086654663,
      "learning_rate": 1.3553169179126133e-05,
      "loss": 1.9972,
      "step": 208274
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.3855170011520386,
      "learning_rate": 1.3551953589510501e-05,
      "loss": 1.8189,
      "step": 208275
    },
    {
      "epoch": 2.71,
      "grad_norm": 2.4199376106262207,
      "learning_rate": 1.35507380531509e-05,
      "loss": 1.8182,
      "step": 208276
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.2888774871826172,
      "learning_rate": 1.3549522570047732e-05,
      "loss": 1.7222,
      "step": 208277
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.3704352378845215,
      "learning_rate": 1.3548307140200998e-05,
      "loss": 1.8849,
      "step": 208278
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.4556437730789185,
      "learning_rate": 1.3547091763611095e-05,
      "loss": 2.0429,
      "step": 208279
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.5330018997192383,
      "learning_rate": 1.3545876440278159e-05,
      "loss": 2.0397,
      "step": 208280
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.3744007349014282,
      "learning_rate": 1.354466117020252e-05,
      "loss": 1.9308,
      "step": 208281
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.5443583726882935,
      "learning_rate": 1.3543445953384247e-05,
      "loss": 1.8865,
      "step": 208282
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.7865538597106934,
      "learning_rate": 1.3542230789823772e-05,
      "loss": 1.9742,
      "step": 208283
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.517215371131897,
      "learning_rate": 1.3541015679521095e-05,
      "loss": 1.9696,
      "step": 208284
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.6666120290756226,
      "learning_rate": 1.3539800622476616e-05,
      "loss": 2.0043,
      "step": 208285
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.5197288990020752,
      "learning_rate": 1.3538585618690434e-05,
      "loss": 2.0293,
      "step": 208286
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.4056740999221802,
      "learning_rate": 1.3537370668162884e-05,
      "loss": 1.8008,
      "step": 208287
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.4778932332992554,
      "learning_rate": 1.3536155770894097e-05,
      "loss": 1.9601,
      "step": 208288
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.43825101852417,
      "learning_rate": 1.3534940926884442e-05,
      "loss": 1.9251,
      "step": 208289
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.4604204893112183,
      "learning_rate": 1.3533726136133982e-05,
      "loss": 1.965,
      "step": 208290
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.3550859689712524,
      "learning_rate": 1.353251139864302e-05,
      "loss": 1.7788,
      "step": 208291
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.3904005289077759,
      "learning_rate": 1.353129671441172e-05,
      "loss": 1.94,
      "step": 208292
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.6384247541427612,
      "learning_rate": 1.3530082083440452e-05,
      "loss": 2.0017,
      "step": 208293
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.385202407836914,
      "learning_rate": 1.3528867505729246e-05,
      "loss": 2.1331,
      "step": 208294
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.3739855289459229,
      "learning_rate": 1.352765298127857e-05,
      "loss": 1.9448,
      "step": 208295
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.508278489112854,
      "learning_rate": 1.352643851008839e-05,
      "loss": 2.1319,
      "step": 208296
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.6493357419967651,
      "learning_rate": 1.3525224092159104e-05,
      "loss": 2.0746,
      "step": 208297
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.4737364053726196,
      "learning_rate": 1.3524009727490848e-05,
      "loss": 1.9049,
      "step": 208298
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.5387150049209595,
      "learning_rate": 1.352279541608392e-05,
      "loss": 2.063,
      "step": 208299
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.60419499874115,
      "learning_rate": 1.3521581157938488e-05,
      "loss": 1.9685,
      "step": 208300
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.455759048461914,
      "learning_rate": 1.3520366953054884e-05,
      "loss": 2.1058,
      "step": 208301
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.6079291105270386,
      "learning_rate": 1.3519152801433142e-05,
      "loss": 1.9525,
      "step": 208302
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.5143533945083618,
      "learning_rate": 1.351793870307366e-05,
      "loss": 1.8051,
      "step": 208303
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.516402006149292,
      "learning_rate": 1.3516724657976574e-05,
      "loss": 1.8611,
      "step": 208304
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.3989335298538208,
      "learning_rate": 1.3515510666142148e-05,
      "loss": 2.0866,
      "step": 208305
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.3800008296966553,
      "learning_rate": 1.3514296727570584e-05,
      "loss": 1.8611,
      "step": 208306
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.3187172412872314,
      "learning_rate": 1.351308284226218e-05,
      "loss": 2.0337,
      "step": 208307
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.6289491653442383,
      "learning_rate": 1.3511869010217002e-05,
      "loss": 1.9356,
      "step": 208308
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.4582663774490356,
      "learning_rate": 1.3510655231435452e-05,
      "loss": 1.9649,
      "step": 208309
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.5569941997528076,
      "learning_rate": 1.3509441505917628e-05,
      "loss": 1.8134,
      "step": 208310
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.5709189176559448,
      "learning_rate": 1.3508227833663831e-05,
      "loss": 2.1041,
      "step": 208311
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.3401316404342651,
      "learning_rate": 1.350701421467426e-05,
      "loss": 1.9852,
      "step": 208312
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.5209105014801025,
      "learning_rate": 1.3505800648949184e-05,
      "loss": 2.0682,
      "step": 208313
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.371332049369812,
      "learning_rate": 1.3504587136488731e-05,
      "loss": 1.9356,
      "step": 208314
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.4308373928070068,
      "learning_rate": 1.3503373677293206e-05,
      "loss": 1.8678,
      "step": 208315
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.5072821378707886,
      "learning_rate": 1.3502160271362771e-05,
      "loss": 2.0701,
      "step": 208316
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.4587490558624268,
      "learning_rate": 1.3500946918697764e-05,
      "loss": 1.9615,
      "step": 208317
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.4216243028640747,
      "learning_rate": 1.349973361929828e-05,
      "loss": 2.0234,
      "step": 208318
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.6331901550292969,
      "learning_rate": 1.3498520373164623e-05,
      "loss": 2.0056,
      "step": 208319
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.3934215307235718,
      "learning_rate": 1.3497307180297024e-05,
      "loss": 1.8673,
      "step": 208320
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.4404712915420532,
      "learning_rate": 1.3496094040695682e-05,
      "loss": 1.8858,
      "step": 208321
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.3822455406188965,
      "learning_rate": 1.3494880954360765e-05,
      "loss": 2.203,
      "step": 208322
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.3513951301574707,
      "learning_rate": 1.3493667921292606e-05,
      "loss": 1.8246,
      "step": 208323
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.4679667949676514,
      "learning_rate": 1.349245494149137e-05,
      "loss": 2.056,
      "step": 208324
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.4598321914672852,
      "learning_rate": 1.3491242014957293e-05,
      "loss": 1.9331,
      "step": 208325
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.4337769746780396,
      "learning_rate": 1.3490029141690639e-05,
      "loss": 2.0371,
      "step": 208326
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.5030986070632935,
      "learning_rate": 1.3488816321691576e-05,
      "loss": 1.9431,
      "step": 208327
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.589932918548584,
      "learning_rate": 1.3487603554960303e-05,
      "loss": 2.0155,
      "step": 208328
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.5864492654800415,
      "learning_rate": 1.3486390841497153e-05,
      "loss": 1.8607,
      "step": 208329
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.5050238370895386,
      "learning_rate": 1.3485178181302258e-05,
      "loss": 1.865,
      "step": 208330
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.328144907951355,
      "learning_rate": 1.3483965574375888e-05,
      "loss": 1.8452,
      "step": 208331
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.3040308952331543,
      "learning_rate": 1.3482753020718274e-05,
      "loss": 1.9051,
      "step": 208332
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.3756015300750732,
      "learning_rate": 1.3481540520329614e-05,
      "loss": 1.7658,
      "step": 208333
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.3562188148498535,
      "learning_rate": 1.3480328073210112e-05,
      "loss": 2.1795,
      "step": 208334
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.4081315994262695,
      "learning_rate": 1.3479115679360064e-05,
      "loss": 2.143,
      "step": 208335
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.4231677055358887,
      "learning_rate": 1.3477903338779638e-05,
      "loss": 1.8248,
      "step": 208336
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.5454338788986206,
      "learning_rate": 1.3476691051469101e-05,
      "loss": 1.9874,
      "step": 208337
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.5026036500930786,
      "learning_rate": 1.3475478817428653e-05,
      "loss": 1.9931,
      "step": 208338
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.5662779808044434,
      "learning_rate": 1.347426663665846e-05,
      "loss": 2.003,
      "step": 208339
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.503745198249817,
      "learning_rate": 1.3473054509158886e-05,
      "loss": 1.9451,
      "step": 208340
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.515005111694336,
      "learning_rate": 1.3471842434930068e-05,
      "loss": 1.8796,
      "step": 208341
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.506822109222412,
      "learning_rate": 1.3470630413972206e-05,
      "loss": 1.9915,
      "step": 208342
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.4674164056777954,
      "learning_rate": 1.3469418446285596e-05,
      "loss": 2.1529,
      "step": 208343
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.4538981914520264,
      "learning_rate": 1.3468206531870406e-05,
      "loss": 2.0322,
      "step": 208344
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.4234459400177002,
      "learning_rate": 1.3466994670726871e-05,
      "loss": 1.8692,
      "step": 208345
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.4425723552703857,
      "learning_rate": 1.346578286285529e-05,
      "loss": 2.1279,
      "step": 208346
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.5316252708435059,
      "learning_rate": 1.3464571108255795e-05,
      "loss": 1.9612,
      "step": 208347
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.4236226081848145,
      "learning_rate": 1.346335940692862e-05,
      "loss": 1.9931,
      "step": 208348
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.5932281017303467,
      "learning_rate": 1.3462147758874064e-05,
      "loss": 1.9872,
      "step": 208349
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.4495552778244019,
      "learning_rate": 1.3460936164092295e-05,
      "loss": 2.201,
      "step": 208350
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.37815260887146,
      "learning_rate": 1.3459724622583512e-05,
      "loss": 1.9446,
      "step": 208351
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.3443019390106201,
      "learning_rate": 1.3458513134347981e-05,
      "loss": 1.9759,
      "step": 208352
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.4029983282089233,
      "learning_rate": 1.3457301699385936e-05,
      "loss": 1.9306,
      "step": 208353
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.3840665817260742,
      "learning_rate": 1.3456090317697577e-05,
      "loss": 1.8096,
      "step": 208354
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.41990327835083,
      "learning_rate": 1.345487898928317e-05,
      "loss": 1.9613,
      "step": 208355
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.5070714950561523,
      "learning_rate": 1.3453667714142913e-05,
      "loss": 1.8109,
      "step": 208356
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.3864681720733643,
      "learning_rate": 1.3452456492276975e-05,
      "loss": 1.9125,
      "step": 208357
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.5450706481933594,
      "learning_rate": 1.3451245323685656e-05,
      "loss": 1.9337,
      "step": 208358
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.727879524230957,
      "learning_rate": 1.3450034208369154e-05,
      "loss": 2.1375,
      "step": 208359
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.5699540376663208,
      "learning_rate": 1.3448823146327735e-05,
      "loss": 1.953,
      "step": 208360
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.456350564956665,
      "learning_rate": 1.3447612137561536e-05,
      "loss": 2.0181,
      "step": 208361
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.4301445484161377,
      "learning_rate": 1.3446401182070921e-05,
      "loss": 1.7137,
      "step": 208362
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.4276522397994995,
      "learning_rate": 1.3445190279855922e-05,
      "loss": 1.9585,
      "step": 208363
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.4863587617874146,
      "learning_rate": 1.344397943091694e-05,
      "loss": 2.0187,
      "step": 208364
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.431822657585144,
      "learning_rate": 1.3442768635254109e-05,
      "loss": 1.9955,
      "step": 208365
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.416503667831421,
      "learning_rate": 1.3441557892867693e-05,
      "loss": 1.9132,
      "step": 208366
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.5727300643920898,
      "learning_rate": 1.3440347203757863e-05,
      "loss": 1.939,
      "step": 208367
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.3499665260314941,
      "learning_rate": 1.3439136567924946e-05,
      "loss": 1.9159,
      "step": 208368
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.3190761804580688,
      "learning_rate": 1.3437925985369047e-05,
      "loss": 2.0184,
      "step": 208369
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.3975577354431152,
      "learning_rate": 1.3436715456090463e-05,
      "loss": 1.8107,
      "step": 208370
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.4321231842041016,
      "learning_rate": 1.3435504980089395e-05,
      "loss": 1.8778,
      "step": 208371
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.573455810546875,
      "learning_rate": 1.343429455736611e-05,
      "loss": 1.9856,
      "step": 208372
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.404899001121521,
      "learning_rate": 1.3433084187920739e-05,
      "loss": 1.859,
      "step": 208373
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.4152289628982544,
      "learning_rate": 1.3431873871753652e-05,
      "loss": 1.9387,
      "step": 208374
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.3981823921203613,
      "learning_rate": 1.3430663608864911e-05,
      "loss": 2.1285,
      "step": 208375
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.6092532873153687,
      "learning_rate": 1.3429453399254852e-05,
      "loss": 2.0659,
      "step": 208376
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.2687040567398071,
      "learning_rate": 1.3428243242923642e-05,
      "loss": 1.837,
      "step": 208377
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.369489312171936,
      "learning_rate": 1.342703313987158e-05,
      "loss": 2.0253,
      "step": 208378
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.4718968868255615,
      "learning_rate": 1.3425823090098763e-05,
      "loss": 1.8635,
      "step": 208379
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.4998445510864258,
      "learning_rate": 1.3424613093605597e-05,
      "loss": 1.9054,
      "step": 208380
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.513689637184143,
      "learning_rate": 1.3423403150392142e-05,
      "loss": 2.2426,
      "step": 208381
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.5180796384811401,
      "learning_rate": 1.3422193260458702e-05,
      "loss": 1.8184,
      "step": 208382
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.4706984758377075,
      "learning_rate": 1.3420983423805475e-05,
      "loss": 2.0011,
      "step": 208383
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.427146315574646,
      "learning_rate": 1.3419773640432697e-05,
      "loss": 1.9847,
      "step": 208384
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.521302342414856,
      "learning_rate": 1.3418563910340596e-05,
      "loss": 2.0624,
      "step": 208385
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.4796168804168701,
      "learning_rate": 1.341735423352941e-05,
      "loss": 2.0111,
      "step": 208386
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.4951286315917969,
      "learning_rate": 1.3416144609999336e-05,
      "loss": 1.9949,
      "step": 208387
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.5538725852966309,
      "learning_rate": 1.3414935039750607e-05,
      "loss": 1.9923,
      "step": 208388
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.3862369060516357,
      "learning_rate": 1.3413725522783424e-05,
      "loss": 1.7555,
      "step": 208389
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.5759823322296143,
      "learning_rate": 1.3412516059098088e-05,
      "loss": 2.1738,
      "step": 208390
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.3926217555999756,
      "learning_rate": 1.341130664869473e-05,
      "loss": 2.0056,
      "step": 208391
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.3610957860946655,
      "learning_rate": 1.341009729157365e-05,
      "loss": 1.7091,
      "step": 208392
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.4458595514297485,
      "learning_rate": 1.3408887987735051e-05,
      "loss": 1.9802,
      "step": 208393
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.37015962600708,
      "learning_rate": 1.340767873717916e-05,
      "loss": 2.0333,
      "step": 208394
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.386931300163269,
      "learning_rate": 1.340646953990615e-05,
      "loss": 2.0237,
      "step": 208395
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.473805546760559,
      "learning_rate": 1.3405260395916318e-05,
      "loss": 2.0979,
      "step": 208396
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.2852922677993774,
      "learning_rate": 1.3404051305209829e-05,
      "loss": 1.9453,
      "step": 208397
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.4254319667816162,
      "learning_rate": 1.340284226778695e-05,
      "loss": 2.0099,
      "step": 208398
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.5264346599578857,
      "learning_rate": 1.3401633283647916e-05,
      "loss": 2.1051,
      "step": 208399
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.3639662265777588,
      "learning_rate": 1.3400424352792926e-05,
      "loss": 1.8843,
      "step": 208400
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.3374438285827637,
      "learning_rate": 1.3399215475222148e-05,
      "loss": 1.882,
      "step": 208401
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.546599268913269,
      "learning_rate": 1.339800665093591e-05,
      "loss": 1.9148,
      "step": 208402
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.4785765409469604,
      "learning_rate": 1.3396797879934384e-05,
      "loss": 2.0642,
      "step": 208403
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.5548824071884155,
      "learning_rate": 1.3395589162217801e-05,
      "loss": 1.9232,
      "step": 208404
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.5089625120162964,
      "learning_rate": 1.3394380497786428e-05,
      "loss": 2.0736,
      "step": 208405
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.3989053964614868,
      "learning_rate": 1.339317188664043e-05,
      "loss": 1.9122,
      "step": 208406
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.38127863407135,
      "learning_rate": 1.339196332878001e-05,
      "loss": 2.0188,
      "step": 208407
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.4523879289627075,
      "learning_rate": 1.3390754824205463e-05,
      "loss": 1.871,
      "step": 208408
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.3657547235488892,
      "learning_rate": 1.338954637291696e-05,
      "loss": 1.699,
      "step": 208409
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.328320860862732,
      "learning_rate": 1.3388337974914798e-05,
      "loss": 1.8414,
      "step": 208410
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.8389564752578735,
      "learning_rate": 1.3387129630199145e-05,
      "loss": 2.1081,
      "step": 208411
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.4445298910140991,
      "learning_rate": 1.3385921338770233e-05,
      "loss": 1.7836,
      "step": 208412
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.6243855953216553,
      "learning_rate": 1.338471310062823e-05,
      "loss": 2.0941,
      "step": 208413
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.447264313697815,
      "learning_rate": 1.3383504915773502e-05,
      "loss": 2.0641,
      "step": 208414
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.411220669746399,
      "learning_rate": 1.3382296784206114e-05,
      "loss": 1.9027,
      "step": 208415
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.44839346408844,
      "learning_rate": 1.3381088705926435e-05,
      "loss": 1.8341,
      "step": 208416
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.330819845199585,
      "learning_rate": 1.3379880680934596e-05,
      "loss": 1.8692,
      "step": 208417
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.5610276460647583,
      "learning_rate": 1.3378672709230864e-05,
      "loss": 2.1461,
      "step": 208418
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.4122811555862427,
      "learning_rate": 1.3377464790815406e-05,
      "loss": 2.0533,
      "step": 208419
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.3847719430923462,
      "learning_rate": 1.3376256925688522e-05,
      "loss": 2.1351,
      "step": 208420
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.616713047027588,
      "learning_rate": 1.3375049113850344e-05,
      "loss": 1.8099,
      "step": 208421
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.4331560134887695,
      "learning_rate": 1.337384135530124e-05,
      "loss": 1.9249,
      "step": 208422
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.402236819267273,
      "learning_rate": 1.3372633650041308e-05,
      "loss": 1.9068,
      "step": 208423
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.3754572868347168,
      "learning_rate": 1.337142599807085e-05,
      "loss": 2.0712,
      "step": 208424
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.4707379341125488,
      "learning_rate": 1.3370218399389998e-05,
      "loss": 1.9003,
      "step": 208425
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.5070643424987793,
      "learning_rate": 1.3369010853999052e-05,
      "loss": 2.0936,
      "step": 208426
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.5767282247543335,
      "learning_rate": 1.336780336189821e-05,
      "loss": 1.9587,
      "step": 208427
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.5669646263122559,
      "learning_rate": 1.3366595923087708e-05,
      "loss": 2.1487,
      "step": 208428
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.3520902395248413,
      "learning_rate": 1.3365388537567811e-05,
      "loss": 1.8832,
      "step": 208429
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.3886662721633911,
      "learning_rate": 1.336418120533862e-05,
      "loss": 1.8479,
      "step": 208430
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.4543280601501465,
      "learning_rate": 1.33629739264005e-05,
      "loss": 2.1189,
      "step": 208431
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.4761394262313843,
      "learning_rate": 1.3361766700753618e-05,
      "loss": 2.0301,
      "step": 208432
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.5410406589508057,
      "learning_rate": 1.336055952839814e-05,
      "loss": 1.8948,
      "step": 208433
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.34268319606781,
      "learning_rate": 1.3359352409334401e-05,
      "loss": 1.978,
      "step": 208434
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.3577078580856323,
      "learning_rate": 1.3358145343562531e-05,
      "loss": 1.9199,
      "step": 208435
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.429514765739441,
      "learning_rate": 1.33569383310828e-05,
      "loss": 2.0017,
      "step": 208436
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.3544031381607056,
      "learning_rate": 1.335573137189544e-05,
      "loss": 2.0576,
      "step": 208437
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.3559722900390625,
      "learning_rate": 1.3354524466000615e-05,
      "loss": 1.8483,
      "step": 208438
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.4474315643310547,
      "learning_rate": 1.3353317613398661e-05,
      "loss": 1.7452,
      "step": 208439
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.580551266670227,
      "learning_rate": 1.3352110814089711e-05,
      "loss": 1.9805,
      "step": 208440
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.4922549724578857,
      "learning_rate": 1.3350904068074031e-05,
      "loss": 1.9327,
      "step": 208441
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.5626308917999268,
      "learning_rate": 1.3349697375351753e-05,
      "loss": 2.0682,
      "step": 208442
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.4002033472061157,
      "learning_rate": 1.3348490735923245e-05,
      "loss": 1.9237,
      "step": 208443
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.5509557723999023,
      "learning_rate": 1.334728414978864e-05,
      "loss": 2.0077,
      "step": 208444
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.415175437927246,
      "learning_rate": 1.3346077616948204e-05,
      "loss": 1.7401,
      "step": 208445
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.5668038129806519,
      "learning_rate": 1.3344871137402136e-05,
      "loss": 1.9244,
      "step": 208446
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.3762869834899902,
      "learning_rate": 1.3343664711150703e-05,
      "loss": 1.9022,
      "step": 208447
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.4987674951553345,
      "learning_rate": 1.3342458338194006e-05,
      "loss": 2.0235,
      "step": 208448
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.4465947151184082,
      "learning_rate": 1.3341252018532445e-05,
      "loss": 2.0339,
      "step": 208449
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.4304752349853516,
      "learning_rate": 1.3340045752166085e-05,
      "loss": 1.9794,
      "step": 208450
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.544015884399414,
      "learning_rate": 1.3338839539095258e-05,
      "loss": 1.8777,
      "step": 208451
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.4410744905471802,
      "learning_rate": 1.3337633379320134e-05,
      "loss": 1.9154,
      "step": 208452
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.489180564880371,
      "learning_rate": 1.3336427272840976e-05,
      "loss": 1.8514,
      "step": 208453
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.4486229419708252,
      "learning_rate": 1.3335221219658022e-05,
      "loss": 1.7551,
      "step": 208454
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.5478746891021729,
      "learning_rate": 1.3334015219771432e-05,
      "loss": 1.7781,
      "step": 208455
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.4172004461288452,
      "learning_rate": 1.333280927318141e-05,
      "loss": 1.8952,
      "step": 208456
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.4946179389953613,
      "learning_rate": 1.333160337988829e-05,
      "loss": 1.9165,
      "step": 208457
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.5253247022628784,
      "learning_rate": 1.3330397539892169e-05,
      "loss": 2.0127,
      "step": 208458
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.3631328344345093,
      "learning_rate": 1.3329191753193413e-05,
      "loss": 1.8803,
      "step": 208459
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.4164639711380005,
      "learning_rate": 1.3327986019792158e-05,
      "loss": 1.8498,
      "step": 208460
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.4015141725540161,
      "learning_rate": 1.3326780339688636e-05,
      "loss": 2.0542,
      "step": 208461
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.3329654932022095,
      "learning_rate": 1.3325574712883048e-05,
      "loss": 2.088,
      "step": 208462
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.7429349422454834,
      "learning_rate": 1.332436913937569e-05,
      "loss": 2.0723,
      "step": 208463
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.4052402973175049,
      "learning_rate": 1.3323163619166699e-05,
      "loss": 1.8136,
      "step": 208464
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.4636825323104858,
      "learning_rate": 1.3321958152256373e-05,
      "loss": 1.9356,
      "step": 208465
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.4196465015411377,
      "learning_rate": 1.3320752738644914e-05,
      "loss": 1.9084,
      "step": 208466
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.5126527547836304,
      "learning_rate": 1.3319547378332518e-05,
      "loss": 1.9038,
      "step": 208467
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.5748029947280884,
      "learning_rate": 1.3318342071319421e-05,
      "loss": 1.9951,
      "step": 208468
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.492310643196106,
      "learning_rate": 1.331713681760589e-05,
      "loss": 1.8451,
      "step": 208469
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.3655238151550293,
      "learning_rate": 1.3315931617192055e-05,
      "loss": 1.8733,
      "step": 208470
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.5029960870742798,
      "learning_rate": 1.3314726470078252e-05,
      "loss": 2.0066,
      "step": 208471
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.466650366783142,
      "learning_rate": 1.3313521376264647e-05,
      "loss": 2.1027,
      "step": 208472
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.4111133813858032,
      "learning_rate": 1.3312316335751472e-05,
      "loss": 2.0023,
      "step": 208473
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.5551422834396362,
      "learning_rate": 1.3311111348538927e-05,
      "loss": 1.8568,
      "step": 208474
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.3934202194213867,
      "learning_rate": 1.330990641462728e-05,
      "loss": 1.9476,
      "step": 208475
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.707314372062683,
      "learning_rate": 1.3308701534016697e-05,
      "loss": 2.1133,
      "step": 208476
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.6573460102081299,
      "learning_rate": 1.3307496706707477e-05,
      "loss": 2.0194,
      "step": 208477
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.4535722732543945,
      "learning_rate": 1.330629193269982e-05,
      "loss": 2.2654,
      "step": 208478
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.6483949422836304,
      "learning_rate": 1.3305087211993926e-05,
      "loss": 2.0813,
      "step": 208479
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.5209059715270996,
      "learning_rate": 1.3303882544589995e-05,
      "loss": 1.9921,
      "step": 208480
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.4621027708053589,
      "learning_rate": 1.3302677930488292e-05,
      "loss": 2.0322,
      "step": 208481
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.5923961400985718,
      "learning_rate": 1.330147336968902e-05,
      "loss": 2.0304,
      "step": 208482
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.4462088346481323,
      "learning_rate": 1.3300268862192443e-05,
      "loss": 2.0983,
      "step": 208483
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.5035933256149292,
      "learning_rate": 1.3299064407998762e-05,
      "loss": 1.8572,
      "step": 208484
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.420199990272522,
      "learning_rate": 1.3297860007108208e-05,
      "loss": 2.1238,
      "step": 208485
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.492622971534729,
      "learning_rate": 1.329665565952095e-05,
      "loss": 1.8134,
      "step": 208486
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.6268543004989624,
      "learning_rate": 1.3295451365237286e-05,
      "loss": 2.065,
      "step": 208487
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.6248910427093506,
      "learning_rate": 1.3294247124257384e-05,
      "loss": 1.9536,
      "step": 208488
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.580504298210144,
      "learning_rate": 1.329304293658151e-05,
      "loss": 2.0756,
      "step": 208489
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.3307616710662842,
      "learning_rate": 1.3291838802209898e-05,
      "loss": 1.9716,
      "step": 208490
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.5745443105697632,
      "learning_rate": 1.3290634721142746e-05,
      "loss": 2.0636,
      "step": 208491
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.5043612718582153,
      "learning_rate": 1.328943069338022e-05,
      "loss": 1.8904,
      "step": 208492
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.489088773727417,
      "learning_rate": 1.3288226718922623e-05,
      "loss": 2.108,
      "step": 208493
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.4437808990478516,
      "learning_rate": 1.3287022797770153e-05,
      "loss": 2.1607,
      "step": 208494
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.3174103498458862,
      "learning_rate": 1.3285818929923075e-05,
      "loss": 1.8546,
      "step": 208495
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.4543265104293823,
      "learning_rate": 1.3284615115381558e-05,
      "loss": 1.8903,
      "step": 208496
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.3998315334320068,
      "learning_rate": 1.3283411354145835e-05,
      "loss": 1.983,
      "step": 208497
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.3974109888076782,
      "learning_rate": 1.3282207646216103e-05,
      "loss": 2.0256,
      "step": 208498
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.4980137348175049,
      "learning_rate": 1.3281003991592698e-05,
      "loss": 1.96,
      "step": 208499
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.4265671968460083,
      "learning_rate": 1.3279800390275685e-05,
      "loss": 2.0319,
      "step": 208500
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.4714947938919067,
      "learning_rate": 1.3278596842265432e-05,
      "loss": 2.1429,
      "step": 208501
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.4271167516708374,
      "learning_rate": 1.327739334756207e-05,
      "loss": 1.7623,
      "step": 208502
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.393096923828125,
      "learning_rate": 1.3276189906165901e-05,
      "loss": 1.8248,
      "step": 208503
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.5208055973052979,
      "learning_rate": 1.3274986518077024e-05,
      "loss": 1.9779,
      "step": 208504
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.5359140634536743,
      "learning_rate": 1.3273783183295772e-05,
      "loss": 1.9414,
      "step": 208505
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.6067200899124146,
      "learning_rate": 1.3272579901822311e-05,
      "loss": 1.9321,
      "step": 208506
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.458788275718689,
      "learning_rate": 1.3271376673656942e-05,
      "loss": 2.1441,
      "step": 208507
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.449946403503418,
      "learning_rate": 1.327017349879983e-05,
      "loss": 1.8428,
      "step": 208508
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.55143141746521,
      "learning_rate": 1.326897037725121e-05,
      "loss": 1.8389,
      "step": 208509
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.4027104377746582,
      "learning_rate": 1.3267767309011246e-05,
      "loss": 1.9145,
      "step": 208510
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.6789803504943848,
      "learning_rate": 1.3266564294080274e-05,
      "loss": 2.0008,
      "step": 208511
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.5615897178649902,
      "learning_rate": 1.326536133245839e-05,
      "loss": 1.9299,
      "step": 208512
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.481427788734436,
      "learning_rate": 1.3264158424145966e-05,
      "loss": 2.1278,
      "step": 208513
    },
    {
      "epoch": 2.71,
      "grad_norm": 1.4452606439590454,
      "learning_rate": 1.326295556914313e-05,
      "loss": 2.1055,
      "step": 208514
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.482964038848877,
      "learning_rate": 1.3261752767450085e-05,
      "loss": 2.0233,
      "step": 208515
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.4714990854263306,
      "learning_rate": 1.3260550019067128e-05,
      "loss": 1.9116,
      "step": 208516
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.5541930198669434,
      "learning_rate": 1.325934732399443e-05,
      "loss": 1.8167,
      "step": 208517
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.3453315496444702,
      "learning_rate": 1.325814468223222e-05,
      "loss": 1.7187,
      "step": 208518
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.4041398763656616,
      "learning_rate": 1.3256942093780764e-05,
      "loss": 1.931,
      "step": 208519
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.4101089239120483,
      "learning_rate": 1.32557395586402e-05,
      "loss": 1.7037,
      "step": 208520
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.4874141216278076,
      "learning_rate": 1.325453707681089e-05,
      "loss": 1.8688,
      "step": 208521
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.525681734085083,
      "learning_rate": 1.3253334648292935e-05,
      "loss": 1.9234,
      "step": 208522
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.540249228477478,
      "learning_rate": 1.3252132273086569e-05,
      "loss": 2.065,
      "step": 208523
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.558894157409668,
      "learning_rate": 1.325092995119209e-05,
      "loss": 1.9271,
      "step": 208524
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.5411651134490967,
      "learning_rate": 1.3249727682609668e-05,
      "loss": 1.8892,
      "step": 208525
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.4410001039505005,
      "learning_rate": 1.3248525467339499e-05,
      "loss": 1.993,
      "step": 208526
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.3910043239593506,
      "learning_rate": 1.3247323305381885e-05,
      "loss": 1.9075,
      "step": 208527
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.4965755939483643,
      "learning_rate": 1.3246121196736991e-05,
      "loss": 2.0164,
      "step": 208528
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.5115323066711426,
      "learning_rate": 1.324491914140502e-05,
      "loss": 1.8894,
      "step": 208529
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.47706139087677,
      "learning_rate": 1.3243717139386268e-05,
      "loss": 1.8707,
      "step": 208530
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.4268462657928467,
      "learning_rate": 1.3242515190680935e-05,
      "loss": 1.9387,
      "step": 208531
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.4053105115890503,
      "learning_rate": 1.324131329528919e-05,
      "loss": 1.8402,
      "step": 208532
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.4558446407318115,
      "learning_rate": 1.3240111453211333e-05,
      "loss": 2.0724,
      "step": 208533
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.491105556488037,
      "learning_rate": 1.3238909664447561e-05,
      "loss": 1.8897,
      "step": 208534
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.4955965280532837,
      "learning_rate": 1.3237707928998043e-05,
      "loss": 1.8687,
      "step": 208535
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.3113982677459717,
      "learning_rate": 1.323650624686311e-05,
      "loss": 1.8866,
      "step": 208536
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.383802890777588,
      "learning_rate": 1.323530461804283e-05,
      "loss": 2.0922,
      "step": 208537
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.7591397762298584,
      "learning_rate": 1.3234103042537602e-05,
      "loss": 2.1413,
      "step": 208538
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.4525372982025146,
      "learning_rate": 1.323290152034756e-05,
      "loss": 2.078,
      "step": 208539
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.3857238292694092,
      "learning_rate": 1.3231700051472938e-05,
      "loss": 1.9858,
      "step": 208540
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.5485800504684448,
      "learning_rate": 1.32304986359139e-05,
      "loss": 1.9237,
      "step": 208541
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.503283143043518,
      "learning_rate": 1.3229297273670781e-05,
      "loss": 1.8824,
      "step": 208542
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.5674545764923096,
      "learning_rate": 1.3228095964743713e-05,
      "loss": 2.0688,
      "step": 208543
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.5098718404769897,
      "learning_rate": 1.3226894709132962e-05,
      "loss": 1.6701,
      "step": 208544
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.5252736806869507,
      "learning_rate": 1.3225693506838797e-05,
      "loss": 2.1193,
      "step": 208545
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.4908411502838135,
      "learning_rate": 1.3224492357861349e-05,
      "loss": 1.9479,
      "step": 208546
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.361736536026001,
      "learning_rate": 1.3223291262200853e-05,
      "loss": 1.9739,
      "step": 208547
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.4636770486831665,
      "learning_rate": 1.3222090219857606e-05,
      "loss": 1.8568,
      "step": 208548
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.455520749092102,
      "learning_rate": 1.3220889230831744e-05,
      "loss": 2.1813,
      "step": 208549
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.6878260374069214,
      "learning_rate": 1.3219688295123565e-05,
      "loss": 2.1691,
      "step": 208550
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.444380760192871,
      "learning_rate": 1.3218487412733236e-05,
      "loss": 1.8454,
      "step": 208551
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.4725874662399292,
      "learning_rate": 1.3217286583661057e-05,
      "loss": 1.8577,
      "step": 208552
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.4106433391571045,
      "learning_rate": 1.321608580790713e-05,
      "loss": 1.9215,
      "step": 208553
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.3032751083374023,
      "learning_rate": 1.3214885085471782e-05,
      "loss": 2.0123,
      "step": 208554
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.7232098579406738,
      "learning_rate": 1.3213684416355152e-05,
      "loss": 1.8935,
      "step": 208555
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.5021830797195435,
      "learning_rate": 1.3212483800557572e-05,
      "loss": 2.1217,
      "step": 208556
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.4831297397613525,
      "learning_rate": 1.3211283238079206e-05,
      "loss": 2.0169,
      "step": 208557
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.5244696140289307,
      "learning_rate": 1.3210082728920257e-05,
      "loss": 2.0241,
      "step": 208558
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.4024683237075806,
      "learning_rate": 1.3208882273080922e-05,
      "loss": 1.9841,
      "step": 208559
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.4333891868591309,
      "learning_rate": 1.3207681870561537e-05,
      "loss": 2.0545,
      "step": 208560
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.6274645328521729,
      "learning_rate": 1.32064815213622e-05,
      "loss": 2.2077,
      "step": 208561
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.4746534824371338,
      "learning_rate": 1.3205281225483244e-05,
      "loss": 2.0361,
      "step": 208562
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.5212340354919434,
      "learning_rate": 1.3204080982924836e-05,
      "loss": 2.1732,
      "step": 208563
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.3541390895843506,
      "learning_rate": 1.320288079368721e-05,
      "loss": 2.1534,
      "step": 208564
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.2946171760559082,
      "learning_rate": 1.320168065777053e-05,
      "loss": 1.9546,
      "step": 208565
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.3604096174240112,
      "learning_rate": 1.3200480575175099e-05,
      "loss": 1.9688,
      "step": 208566
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.3754935264587402,
      "learning_rate": 1.319928054590108e-05,
      "loss": 1.9761,
      "step": 208567
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.5926603078842163,
      "learning_rate": 1.3198080569948777e-05,
      "loss": 2.3616,
      "step": 208568
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.5893750190734863,
      "learning_rate": 1.3196880647318387e-05,
      "loss": 2.1699,
      "step": 208569
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.4204655885696411,
      "learning_rate": 1.3195680778010076e-05,
      "loss": 2.0053,
      "step": 208570
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.4662165641784668,
      "learning_rate": 1.319448096202408e-05,
      "loss": 1.8865,
      "step": 208571
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.4986618757247925,
      "learning_rate": 1.3193281199360662e-05,
      "loss": 2.0707,
      "step": 208572
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.4040225744247437,
      "learning_rate": 1.3192081490020024e-05,
      "loss": 2.0703,
      "step": 208573
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.5249080657958984,
      "learning_rate": 1.3190881834002398e-05,
      "loss": 2.1507,
      "step": 208574
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.4606444835662842,
      "learning_rate": 1.3189682231308019e-05,
      "loss": 2.0524,
      "step": 208575
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.4259059429168701,
      "learning_rate": 1.3188482681937085e-05,
      "loss": 1.999,
      "step": 208576
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.3928053379058838,
      "learning_rate": 1.3187283185889763e-05,
      "loss": 1.9442,
      "step": 208577
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.4135234355926514,
      "learning_rate": 1.318608374316642e-05,
      "loss": 2.0383,
      "step": 208578
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.4361165761947632,
      "learning_rate": 1.3184884353767123e-05,
      "loss": 1.7553,
      "step": 208579
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.48606276512146,
      "learning_rate": 1.3183685017692236e-05,
      "loss": 1.8703,
      "step": 208580
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.4257018566131592,
      "learning_rate": 1.3182485734941894e-05,
      "loss": 2.1388,
      "step": 208581
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.3518915176391602,
      "learning_rate": 1.3181286505516365e-05,
      "loss": 1.8537,
      "step": 208582
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.5071344375610352,
      "learning_rate": 1.3180087329415811e-05,
      "loss": 2.1718,
      "step": 208583
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.360762596130371,
      "learning_rate": 1.3178888206640503e-05,
      "loss": 1.8266,
      "step": 208584
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.4123352766036987,
      "learning_rate": 1.3177689137190639e-05,
      "loss": 2.0765,
      "step": 208585
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.4826632738113403,
      "learning_rate": 1.3176490121066484e-05,
      "loss": 2.0668,
      "step": 208586
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.4505281448364258,
      "learning_rate": 1.3175291158268209e-05,
      "loss": 1.8762,
      "step": 208587
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.3995791673660278,
      "learning_rate": 1.3174092248796143e-05,
      "loss": 1.7194,
      "step": 208588
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.5700784921646118,
      "learning_rate": 1.3172893392650318e-05,
      "loss": 2.0223,
      "step": 208589
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.4144513607025146,
      "learning_rate": 1.3171694589831139e-05,
      "loss": 1.9407,
      "step": 208590
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.499565839767456,
      "learning_rate": 1.3170495840338702e-05,
      "loss": 2.1694,
      "step": 208591
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.3838862180709839,
      "learning_rate": 1.316929714417334e-05,
      "loss": 2.0218,
      "step": 208592
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.3655294179916382,
      "learning_rate": 1.3168098501335155e-05,
      "loss": 1.7882,
      "step": 208593
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.406995177268982,
      "learning_rate": 1.3166899911824513e-05,
      "loss": 1.8117,
      "step": 208594
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.618129849433899,
      "learning_rate": 1.3165701375641514e-05,
      "loss": 1.9836,
      "step": 208595
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.3862031698226929,
      "learning_rate": 1.316450289278642e-05,
      "loss": 2.0705,
      "step": 208596
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.4662972688674927,
      "learning_rate": 1.3163304463259438e-05,
      "loss": 1.744,
      "step": 208597
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.5142881870269775,
      "learning_rate": 1.3162106087060864e-05,
      "loss": 1.8723,
      "step": 208598
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.3801946640014648,
      "learning_rate": 1.3160907764190832e-05,
      "loss": 2.0398,
      "step": 208599
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.482671856880188,
      "learning_rate": 1.3159709494649606e-05,
      "loss": 2.169,
      "step": 208600
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.5894708633422852,
      "learning_rate": 1.3158511278437456e-05,
      "loss": 2.1043,
      "step": 208601
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.314319133758545,
      "learning_rate": 1.3157313115554512e-05,
      "loss": 1.9598,
      "step": 208602
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.384926199913025,
      "learning_rate": 1.3156115006001012e-05,
      "loss": 1.948,
      "step": 208603
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.3747518062591553,
      "learning_rate": 1.315491694977725e-05,
      "loss": 2.1822,
      "step": 208604
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.4473932981491089,
      "learning_rate": 1.3153718946883363e-05,
      "loss": 1.9965,
      "step": 208605
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.5847095251083374,
      "learning_rate": 1.315252099731965e-05,
      "loss": 1.987,
      "step": 208606
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.4941585063934326,
      "learning_rate": 1.3151323101086276e-05,
      "loss": 2.1329,
      "step": 208607
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.3960278034210205,
      "learning_rate": 1.3150125258183508e-05,
      "loss": 1.6614,
      "step": 208608
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.4840366840362549,
      "learning_rate": 1.3148927468611514e-05,
      "loss": 1.9873,
      "step": 208609
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.4313210248947144,
      "learning_rate": 1.3147729732370594e-05,
      "loss": 2.0257,
      "step": 208610
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.5125477313995361,
      "learning_rate": 1.3146532049460879e-05,
      "loss": 2.1518,
      "step": 208611
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.5489085912704468,
      "learning_rate": 1.3145334419882669e-05,
      "loss": 2.0918,
      "step": 208612
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.538346290588379,
      "learning_rate": 1.3144136843636132e-05,
      "loss": 1.9692,
      "step": 208613
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.5362945795059204,
      "learning_rate": 1.3142939320721501e-05,
      "loss": 1.8447,
      "step": 208614
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.4792001247406006,
      "learning_rate": 1.3141741851139075e-05,
      "loss": 2.2381,
      "step": 208615
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.4549533128738403,
      "learning_rate": 1.3140544434888988e-05,
      "loss": 1.9868,
      "step": 208616
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.4461219310760498,
      "learning_rate": 1.313934707197144e-05,
      "loss": 2.0059,
      "step": 208617
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.5494030714035034,
      "learning_rate": 1.3138149762386762e-05,
      "loss": 1.8741,
      "step": 208618
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.5625096559524536,
      "learning_rate": 1.3136952506135124e-05,
      "loss": 1.8938,
      "step": 208619
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.6644158363342285,
      "learning_rate": 1.3135755303216688e-05,
      "loss": 2.0329,
      "step": 208620
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.3572245836257935,
      "learning_rate": 1.3134558153631758e-05,
      "loss": 1.9808,
      "step": 208621
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.488882064819336,
      "learning_rate": 1.3133361057380497e-05,
      "loss": 2.1134,
      "step": 208622
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.445717692375183,
      "learning_rate": 1.3132164014463209e-05,
      "loss": 1.9886,
      "step": 208623
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.3600609302520752,
      "learning_rate": 1.3130967024880056e-05,
      "loss": 2.0276,
      "step": 208624
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.3959895372390747,
      "learning_rate": 1.3129770088631275e-05,
      "loss": 2.0283,
      "step": 208625
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.5412718057632446,
      "learning_rate": 1.3128573205717063e-05,
      "loss": 2.1886,
      "step": 208626
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.5620919466018677,
      "learning_rate": 1.3127376376137687e-05,
      "loss": 2.0265,
      "step": 208627
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.5990720987319946,
      "learning_rate": 1.3126179599893283e-05,
      "loss": 1.8043,
      "step": 208628
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.4353859424591064,
      "learning_rate": 1.3124982876984213e-05,
      "loss": 1.7822,
      "step": 208629
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.4175670146942139,
      "learning_rate": 1.3123786207410615e-05,
      "loss": 2.0976,
      "step": 208630
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.3400121927261353,
      "learning_rate": 1.3122589591172717e-05,
      "loss": 1.8901,
      "step": 208631
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.5570429563522339,
      "learning_rate": 1.3121393028270721e-05,
      "loss": 2.1022,
      "step": 208632
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.3626867532730103,
      "learning_rate": 1.3120196518704895e-05,
      "loss": 1.9204,
      "step": 208633
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.4619972705841064,
      "learning_rate": 1.3119000062475405e-05,
      "loss": 1.7868,
      "step": 208634
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.4319348335266113,
      "learning_rate": 1.3117803659582548e-05,
      "loss": 1.9538,
      "step": 208635
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.5018351078033447,
      "learning_rate": 1.3116607310026495e-05,
      "loss": 2.0081,
      "step": 208636
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.3418506383895874,
      "learning_rate": 1.3115411013807476e-05,
      "loss": 1.9536,
      "step": 208637
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.4628984928131104,
      "learning_rate": 1.311421477092569e-05,
      "loss": 2.0337,
      "step": 208638
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.3849183320999146,
      "learning_rate": 1.311301858138144e-05,
      "loss": 2.059,
      "step": 208639
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.6391464471817017,
      "learning_rate": 1.3111822445174824e-05,
      "loss": 2.0058,
      "step": 208640
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.6807000637054443,
      "learning_rate": 1.311062636230621e-05,
      "loss": 1.6804,
      "step": 208641
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.4412744045257568,
      "learning_rate": 1.3109430332775729e-05,
      "loss": 1.8903,
      "step": 208642
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.4106062650680542,
      "learning_rate": 1.3108234356583614e-05,
      "loss": 1.8746,
      "step": 208643
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.5190157890319824,
      "learning_rate": 1.3107038433730065e-05,
      "loss": 2.3587,
      "step": 208644
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.3374842405319214,
      "learning_rate": 1.3105842564215386e-05,
      "loss": 1.6865,
      "step": 208645
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.4603774547576904,
      "learning_rate": 1.310464674803967e-05,
      "loss": 2.2289,
      "step": 208646
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.3242080211639404,
      "learning_rate": 1.310345098520329e-05,
      "loss": 2.0001,
      "step": 208647
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.4536596536636353,
      "learning_rate": 1.3102255275706375e-05,
      "loss": 1.8864,
      "step": 208648
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.5656177997589111,
      "learning_rate": 1.310105961954916e-05,
      "loss": 1.9066,
      "step": 208649
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.3462363481521606,
      "learning_rate": 1.3099864016731843e-05,
      "loss": 2.1678,
      "step": 208650
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.4593141078948975,
      "learning_rate": 1.3098668467254725e-05,
      "loss": 2.0826,
      "step": 208651
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.513809084892273,
      "learning_rate": 1.3097472971117939e-05,
      "loss": 1.958,
      "step": 208652
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.531915307044983,
      "learning_rate": 1.3096277528321785e-05,
      "loss": 2.191,
      "step": 208653
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.4455652236938477,
      "learning_rate": 1.3095082138866397e-05,
      "loss": 2.0342,
      "step": 208654
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.633955955505371,
      "learning_rate": 1.309388680275214e-05,
      "loss": 1.9014,
      "step": 208655
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.3092014789581299,
      "learning_rate": 1.3092691519979082e-05,
      "loss": 1.7414,
      "step": 208656
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.5246937274932861,
      "learning_rate": 1.3091496290547521e-05,
      "loss": 1.9785,
      "step": 208657
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.4475996494293213,
      "learning_rate": 1.3090301114457624e-05,
      "loss": 2.1068,
      "step": 208658
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.3467564582824707,
      "learning_rate": 1.3089105991709726e-05,
      "loss": 2.1057,
      "step": 208659
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.53596031665802,
      "learning_rate": 1.3087910922303923e-05,
      "loss": 1.9177,
      "step": 208660
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.4038994312286377,
      "learning_rate": 1.3086715906240553e-05,
      "loss": 2.2387,
      "step": 208661
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.7649685144424438,
      "learning_rate": 1.3085520943519744e-05,
      "loss": 1.801,
      "step": 208662
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.577734112739563,
      "learning_rate": 1.3084326034141768e-05,
      "loss": 2.2262,
      "step": 208663
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.5233383178710938,
      "learning_rate": 1.3083131178106787e-05,
      "loss": 1.9961,
      "step": 208664
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.5072500705718994,
      "learning_rate": 1.3081936375415103e-05,
      "loss": 2.2258,
      "step": 208665
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.332454800605774,
      "learning_rate": 1.3080741626066882e-05,
      "loss": 1.8371,
      "step": 208666
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.4905000925064087,
      "learning_rate": 1.3079546930062423e-05,
      "loss": 2.1562,
      "step": 208667
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.36825430393219,
      "learning_rate": 1.3078352287401828e-05,
      "loss": 1.9734,
      "step": 208668
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.6595340967178345,
      "learning_rate": 1.3077157698085427e-05,
      "loss": 1.9472,
      "step": 208669
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.4186065196990967,
      "learning_rate": 1.3075963162113357e-05,
      "loss": 2.0376,
      "step": 208670
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.5026097297668457,
      "learning_rate": 1.3074768679485913e-05,
      "loss": 2.0387,
      "step": 208671
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.5301461219787598,
      "learning_rate": 1.3073574250203233e-05,
      "loss": 2.1018,
      "step": 208672
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.3986377716064453,
      "learning_rate": 1.3072379874265681e-05,
      "loss": 1.9484,
      "step": 208673
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.4626609086990356,
      "learning_rate": 1.3071185551673324e-05,
      "loss": 2.0892,
      "step": 208674
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.4496208429336548,
      "learning_rate": 1.3069991282426494e-05,
      "loss": 1.908,
      "step": 208675
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.579406976699829,
      "learning_rate": 1.3068797066525295e-05,
      "loss": 2.0819,
      "step": 208676
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.5890018939971924,
      "learning_rate": 1.3067602903970087e-05,
      "loss": 2.1596,
      "step": 208677
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.3663794994354248,
      "learning_rate": 1.3066408794761007e-05,
      "loss": 2.1632,
      "step": 208678
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.5221409797668457,
      "learning_rate": 1.3065214738898354e-05,
      "loss": 1.9857,
      "step": 208679
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.602367639541626,
      "learning_rate": 1.3064020736382196e-05,
      "loss": 2.0599,
      "step": 208680
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.6246412992477417,
      "learning_rate": 1.3062826787212932e-05,
      "loss": 1.8663,
      "step": 208681
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.381165862083435,
      "learning_rate": 1.3061632891390627e-05,
      "loss": 2.0422,
      "step": 208682
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.5783957242965698,
      "learning_rate": 1.3060439048915649e-05,
      "loss": 2.0616,
      "step": 208683
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.58869469165802,
      "learning_rate": 1.3059245259788098e-05,
      "loss": 1.9649,
      "step": 208684
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.3552430868148804,
      "learning_rate": 1.3058051524008305e-05,
      "loss": 1.8022,
      "step": 208685
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.342504620552063,
      "learning_rate": 1.305685784157644e-05,
      "loss": 2.0326,
      "step": 208686
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.3666831254959106,
      "learning_rate": 1.30556642124927e-05,
      "loss": 2.0519,
      "step": 208687
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.5261571407318115,
      "learning_rate": 1.3054470636757285e-05,
      "loss": 1.68,
      "step": 208688
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.503398060798645,
      "learning_rate": 1.305327711437053e-05,
      "loss": 2.1468,
      "step": 208689
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.6185200214385986,
      "learning_rate": 1.30520836453325e-05,
      "loss": 2.0522,
      "step": 208690
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.862622857093811,
      "learning_rate": 1.3050890229643595e-05,
      "loss": 2.0668,
      "step": 208691
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.4364495277404785,
      "learning_rate": 1.3049696867303917e-05,
      "loss": 1.7249,
      "step": 208692
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.3902785778045654,
      "learning_rate": 1.3048503558313728e-05,
      "loss": 2.0571,
      "step": 208693
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.5017411708831787,
      "learning_rate": 1.3047310302673197e-05,
      "loss": 2.0136,
      "step": 208694
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.3799388408660889,
      "learning_rate": 1.3046117100382625e-05,
      "loss": 1.8548,
      "step": 208695
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.4385086297988892,
      "learning_rate": 1.3044923951442144e-05,
      "loss": 2.1363,
      "step": 208696
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.324723720550537,
      "learning_rate": 1.3043730855852085e-05,
      "loss": 2.0432,
      "step": 208697
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.62345290184021,
      "learning_rate": 1.3042537813612619e-05,
      "loss": 1.9046,
      "step": 208698
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.460586428642273,
      "learning_rate": 1.3041344824723909e-05,
      "loss": 2.099,
      "step": 208699
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.5781161785125732,
      "learning_rate": 1.304015188918629e-05,
      "loss": 1.8832,
      "step": 208700
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.4368102550506592,
      "learning_rate": 1.3038959006999893e-05,
      "loss": 2.1045,
      "step": 208701
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.4478352069854736,
      "learning_rate": 1.3037766178164955e-05,
      "loss": 1.9699,
      "step": 208702
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.3763076066970825,
      "learning_rate": 1.3036573402681738e-05,
      "loss": 1.8203,
      "step": 208703
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.4713982343673706,
      "learning_rate": 1.3035380680550445e-05,
      "loss": 1.9877,
      "step": 208704
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.5476824045181274,
      "learning_rate": 1.303418801177124e-05,
      "loss": 2.0735,
      "step": 208705
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.5047322511672974,
      "learning_rate": 1.303299539634446e-05,
      "loss": 2.0435,
      "step": 208706
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.5417271852493286,
      "learning_rate": 1.3031802834270233e-05,
      "loss": 1.8455,
      "step": 208707
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.3707157373428345,
      "learning_rate": 1.3030610325548796e-05,
      "loss": 2.0276,
      "step": 208708
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.4384492635726929,
      "learning_rate": 1.3029417870180414e-05,
      "loss": 1.9905,
      "step": 208709
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.3890494108200073,
      "learning_rate": 1.3028225468165288e-05,
      "loss": 2.1333,
      "step": 208710
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.5081076622009277,
      "learning_rate": 1.3027033119503583e-05,
      "loss": 1.7825,
      "step": 208711
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.2988475561141968,
      "learning_rate": 1.30258408241956e-05,
      "loss": 1.9352,
      "step": 208712
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.4509564638137817,
      "learning_rate": 1.3024648582241504e-05,
      "loss": 1.9408,
      "step": 208713
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.5370115041732788,
      "learning_rate": 1.3023456393641596e-05,
      "loss": 2.0101,
      "step": 208714
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.4909335374832153,
      "learning_rate": 1.3022264258396043e-05,
      "loss": 2.0037,
      "step": 208715
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.633191704750061,
      "learning_rate": 1.3021072176505042e-05,
      "loss": 2.0192,
      "step": 208716
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.561256766319275,
      "learning_rate": 1.301988014796883e-05,
      "loss": 1.944,
      "step": 208717
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.4432873725891113,
      "learning_rate": 1.3018688172787638e-05,
      "loss": 1.8652,
      "step": 208718
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.600206971168518,
      "learning_rate": 1.3017496250961701e-05,
      "loss": 2.1963,
      "step": 208719
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.5094693899154663,
      "learning_rate": 1.3016304382491216e-05,
      "loss": 1.8669,
      "step": 208720
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.4175792932510376,
      "learning_rate": 1.3015112567376419e-05,
      "loss": 1.8843,
      "step": 208721
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.4269095659255981,
      "learning_rate": 1.3013920805617606e-05,
      "loss": 1.8859,
      "step": 208722
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.5136606693267822,
      "learning_rate": 1.3012729097214814e-05,
      "loss": 2.0506,
      "step": 208723
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.446743369102478,
      "learning_rate": 1.3011537442168441e-05,
      "loss": 1.965,
      "step": 208724
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.44478440284729,
      "learning_rate": 1.3010345840478587e-05,
      "loss": 1.9275,
      "step": 208725
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.4708720445632935,
      "learning_rate": 1.3009154292145585e-05,
      "loss": 1.9623,
      "step": 208726
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.4421439170837402,
      "learning_rate": 1.3007962797169535e-05,
      "loss": 1.9505,
      "step": 208727
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.4189708232879639,
      "learning_rate": 1.3006771355550805e-05,
      "loss": 2.0737,
      "step": 208728
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.4087713956832886,
      "learning_rate": 1.300557996728946e-05,
      "loss": 1.9835,
      "step": 208729
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.507520079612732,
      "learning_rate": 1.3004388632385831e-05,
      "loss": 2.1577,
      "step": 208730
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.5032678842544556,
      "learning_rate": 1.300319735084009e-05,
      "loss": 1.9207,
      "step": 208731
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.430783987045288,
      "learning_rate": 1.3002006122652465e-05,
      "loss": 1.9542,
      "step": 208732
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.425485372543335,
      "learning_rate": 1.3000814947823191e-05,
      "loss": 2.0302,
      "step": 208733
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.4500796794891357,
      "learning_rate": 1.2999623826352534e-05,
      "loss": 2.082,
      "step": 208734
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.4080724716186523,
      "learning_rate": 1.2998432758240596e-05,
      "loss": 1.8183,
      "step": 208735
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.4869043827056885,
      "learning_rate": 1.2997241743487708e-05,
      "loss": 2.0017,
      "step": 208736
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.420715570449829,
      "learning_rate": 1.2996050782094003e-05,
      "loss": 1.9141,
      "step": 208737
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.3933160305023193,
      "learning_rate": 1.2994859874059783e-05,
      "loss": 1.8589,
      "step": 208738
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.408088207244873,
      "learning_rate": 1.299366901938521e-05,
      "loss": 1.8447,
      "step": 208739
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.3701952695846558,
      "learning_rate": 1.299247821807059e-05,
      "loss": 2.0858,
      "step": 208740
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.4809869527816772,
      "learning_rate": 1.2991287470116018e-05,
      "loss": 2.1476,
      "step": 208741
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.5219544172286987,
      "learning_rate": 1.299009677552183e-05,
      "loss": 2.0298,
      "step": 208742
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.4109400510787964,
      "learning_rate": 1.2988906134288158e-05,
      "loss": 1.7654,
      "step": 208743
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.4358415603637695,
      "learning_rate": 1.29877155464153e-05,
      "loss": 1.9426,
      "step": 208744
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.409595251083374,
      "learning_rate": 1.2986525011903425e-05,
      "loss": 1.9867,
      "step": 208745
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.5347179174423218,
      "learning_rate": 1.2985334530752834e-05,
      "loss": 2.2497,
      "step": 208746
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.3105090856552124,
      "learning_rate": 1.2984144102963623e-05,
      "loss": 1.8925,
      "step": 208747
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.5383379459381104,
      "learning_rate": 1.2982953728536095e-05,
      "loss": 2.0468,
      "step": 208748
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.5757381916046143,
      "learning_rate": 1.2981763407470414e-05,
      "loss": 1.999,
      "step": 208749
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.363938808441162,
      "learning_rate": 1.2980573139766915e-05,
      "loss": 2.0453,
      "step": 208750
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.3646979331970215,
      "learning_rate": 1.2979382925425662e-05,
      "loss": 1.9915,
      "step": 208751
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.44242262840271,
      "learning_rate": 1.2978192764447059e-05,
      "loss": 1.8532,
      "step": 208752
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.4950892925262451,
      "learning_rate": 1.297700265683117e-05,
      "loss": 1.8545,
      "step": 208753
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.4755998849868774,
      "learning_rate": 1.2975812602578295e-05,
      "loss": 2.154,
      "step": 208754
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.4028685092926025,
      "learning_rate": 1.2974622601688566e-05,
      "loss": 2.1218,
      "step": 208755
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.3864644765853882,
      "learning_rate": 1.2973432654162352e-05,
      "loss": 1.7725,
      "step": 208756
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.387271761894226,
      "learning_rate": 1.2972242759999752e-05,
      "loss": 1.9683,
      "step": 208757
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.5184262990951538,
      "learning_rate": 1.2971052919201097e-05,
      "loss": 1.977,
      "step": 208758
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.4538519382476807,
      "learning_rate": 1.2969863131766456e-05,
      "loss": 1.8476,
      "step": 208759
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.4130511283874512,
      "learning_rate": 1.2968673397696195e-05,
      "loss": 1.9026,
      "step": 208760
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.3866732120513916,
      "learning_rate": 1.2967483716990412e-05,
      "loss": 2.0175,
      "step": 208761
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.4459447860717773,
      "learning_rate": 1.2966294089649442e-05,
      "loss": 2.0286,
      "step": 208762
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.472545862197876,
      "learning_rate": 1.296510451567342e-05,
      "loss": 2.1586,
      "step": 208763
    },
    {
      "epoch": 2.72,
      "grad_norm": 2.2397496700286865,
      "learning_rate": 1.2963914995062674e-05,
      "loss": 2.0118,
      "step": 208764
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.3627409934997559,
      "learning_rate": 1.2962725527817308e-05,
      "loss": 2.2586,
      "step": 208765
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.477918267250061,
      "learning_rate": 1.2961536113937588e-05,
      "loss": 2.1354,
      "step": 208766
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.4657385349273682,
      "learning_rate": 1.2960346753423712e-05,
      "loss": 2.0457,
      "step": 208767
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.5132826566696167,
      "learning_rate": 1.295915744627598e-05,
      "loss": 1.8971,
      "step": 208768
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.4969550371170044,
      "learning_rate": 1.2957968192494495e-05,
      "loss": 1.7684,
      "step": 208769
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.3730159997940063,
      "learning_rate": 1.2956778992079619e-05,
      "loss": 1.9492,
      "step": 208770
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.3944783210754395,
      "learning_rate": 1.2955589845031422e-05,
      "loss": 1.9011,
      "step": 208771
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.384415864944458,
      "learning_rate": 1.2954400751350268e-05,
      "loss": 1.6551,
      "step": 208772
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.45332932472229,
      "learning_rate": 1.295321171103626e-05,
      "loss": 1.9113,
      "step": 208773
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.5161761045455933,
      "learning_rate": 1.2952022724089695e-05,
      "loss": 2.096,
      "step": 208774
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.3850371837615967,
      "learning_rate": 1.2950833790510706e-05,
      "loss": 1.891,
      "step": 208775
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.4676989316940308,
      "learning_rate": 1.2949644910299662e-05,
      "loss": 1.9826,
      "step": 208776
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.417043685913086,
      "learning_rate": 1.294845608345666e-05,
      "loss": 2.0113,
      "step": 208777
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.4839437007904053,
      "learning_rate": 1.2947267309981968e-05,
      "loss": 2.1719,
      "step": 208778
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.414340615272522,
      "learning_rate": 1.2946078589875753e-05,
      "loss": 2.1701,
      "step": 208779
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.4372453689575195,
      "learning_rate": 1.2944889923138346e-05,
      "loss": 2.0307,
      "step": 208780
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.3484346866607666,
      "learning_rate": 1.294370130976985e-05,
      "loss": 2.0045,
      "step": 208781
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.3550817966461182,
      "learning_rate": 1.294251274977056e-05,
      "loss": 1.9981,
      "step": 208782
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.6397489309310913,
      "learning_rate": 1.2941324243140682e-05,
      "loss": 1.973,
      "step": 208783
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.467901587486267,
      "learning_rate": 1.2940135789880413e-05,
      "loss": 2.0779,
      "step": 208784
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.4264320135116577,
      "learning_rate": 1.2938947389990018e-05,
      "loss": 2.0367,
      "step": 208785
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.337491512298584,
      "learning_rate": 1.2937759043469698e-05,
      "loss": 1.9584,
      "step": 208786
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.4172722101211548,
      "learning_rate": 1.2936570750319618e-05,
      "loss": 2.0294,
      "step": 208787
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.6213830709457397,
      "learning_rate": 1.2935382510540082e-05,
      "loss": 1.9375,
      "step": 208788
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.4526927471160889,
      "learning_rate": 1.2934194324131287e-05,
      "loss": 1.9103,
      "step": 208789
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.5555413961410522,
      "learning_rate": 1.2933006191093399e-05,
      "loss": 1.9294,
      "step": 208790
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.5108449459075928,
      "learning_rate": 1.2931818111426718e-05,
      "loss": 2.0325,
      "step": 208791
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.407112956047058,
      "learning_rate": 1.2930630085131444e-05,
      "loss": 1.8424,
      "step": 208792
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.6670348644256592,
      "learning_rate": 1.2929442112207744e-05,
      "loss": 1.7394,
      "step": 208793
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.4348453283309937,
      "learning_rate": 1.2928254192655918e-05,
      "loss": 2.266,
      "step": 208794
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.4307100772857666,
      "learning_rate": 1.2927066326476133e-05,
      "loss": 2.1022,
      "step": 208795
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.4658938646316528,
      "learning_rate": 1.2925878513668619e-05,
      "loss": 2.024,
      "step": 208796
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.47336745262146,
      "learning_rate": 1.2924690754233613e-05,
      "loss": 1.9909,
      "step": 208797
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.4346815347671509,
      "learning_rate": 1.2923503048171313e-05,
      "loss": 2.1638,
      "step": 208798
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.3938177824020386,
      "learning_rate": 1.2922315395481952e-05,
      "loss": 1.8719,
      "step": 208799
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.3949955701828003,
      "learning_rate": 1.2921127796165798e-05,
      "loss": 1.9874,
      "step": 208800
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.495633840560913,
      "learning_rate": 1.2919940250222981e-05,
      "loss": 1.7635,
      "step": 208801
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.4501951932907104,
      "learning_rate": 1.2918752757653739e-05,
      "loss": 1.9619,
      "step": 208802
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.4543936252593994,
      "learning_rate": 1.2917565318458366e-05,
      "loss": 1.7218,
      "step": 208803
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.4520241022109985,
      "learning_rate": 1.2916377932637001e-05,
      "loss": 2.0071,
      "step": 208804
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.53268563747406,
      "learning_rate": 1.2915190600189939e-05,
      "loss": 1.9322,
      "step": 208805
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.3776592016220093,
      "learning_rate": 1.291400332111735e-05,
      "loss": 1.9831,
      "step": 208806
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.3856757879257202,
      "learning_rate": 1.2912816095419498e-05,
      "loss": 2.0162,
      "step": 208807
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.4975249767303467,
      "learning_rate": 1.2911628923096518e-05,
      "loss": 1.9952,
      "step": 208808
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.4278194904327393,
      "learning_rate": 1.2910441804148707e-05,
      "loss": 2.0147,
      "step": 208809
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.42984139919281,
      "learning_rate": 1.290925473857627e-05,
      "loss": 2.1738,
      "step": 208810
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.3744337558746338,
      "learning_rate": 1.2908067726379401e-05,
      "loss": 1.849,
      "step": 208811
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.7170498371124268,
      "learning_rate": 1.2906880767558336e-05,
      "loss": 2.0838,
      "step": 208812
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.3701372146606445,
      "learning_rate": 1.2905693862113376e-05,
      "loss": 1.9663,
      "step": 208813
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.5125328302383423,
      "learning_rate": 1.2904507010044585e-05,
      "loss": 2.0178,
      "step": 208814
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.4596209526062012,
      "learning_rate": 1.290332021135233e-05,
      "loss": 2.0607,
      "step": 208815
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.5143753290176392,
      "learning_rate": 1.2902133466036679e-05,
      "loss": 1.9376,
      "step": 208816
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.4764587879180908,
      "learning_rate": 1.2900946774098031e-05,
      "loss": 2.0136,
      "step": 208817
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.428430438041687,
      "learning_rate": 1.2899760135536453e-05,
      "loss": 1.8287,
      "step": 208818
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.542271375656128,
      "learning_rate": 1.2898573550352276e-05,
      "loss": 2.2235,
      "step": 208819
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.4726768732070923,
      "learning_rate": 1.2897387018545635e-05,
      "loss": 1.8595,
      "step": 208820
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.4400231838226318,
      "learning_rate": 1.289620054011683e-05,
      "loss": 1.9188,
      "step": 208821
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.3928108215332031,
      "learning_rate": 1.2895014115065993e-05,
      "loss": 1.8199,
      "step": 208822
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.4089778661727905,
      "learning_rate": 1.2893827743393425e-05,
      "loss": 1.8352,
      "step": 208823
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.3533852100372314,
      "learning_rate": 1.2892641425099293e-05,
      "loss": 1.9263,
      "step": 208824
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.410870909690857,
      "learning_rate": 1.2891455160183895e-05,
      "loss": 2.1976,
      "step": 208825
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.5234556198120117,
      "learning_rate": 1.2890268948647331e-05,
      "loss": 2.2048,
      "step": 208826
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.4501824378967285,
      "learning_rate": 1.2889082790489902e-05,
      "loss": 1.8958,
      "step": 208827
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.4314661026000977,
      "learning_rate": 1.2887896685711807e-05,
      "loss": 2.0358,
      "step": 208828
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.3905848264694214,
      "learning_rate": 1.288671063431328e-05,
      "loss": 1.9615,
      "step": 208829
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.498494267463684,
      "learning_rate": 1.2885524636294519e-05,
      "loss": 2.0144,
      "step": 208830
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.8411604166030884,
      "learning_rate": 1.2884338691655794e-05,
      "loss": 1.8946,
      "step": 208831
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.574665904045105,
      "learning_rate": 1.2883152800397267e-05,
      "loss": 2.0668,
      "step": 208832
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.4987661838531494,
      "learning_rate": 1.2881966962519174e-05,
      "loss": 1.703,
      "step": 208833
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.4460952281951904,
      "learning_rate": 1.2880781178021748e-05,
      "loss": 1.956,
      "step": 208834
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.4151053428649902,
      "learning_rate": 1.2879595446905223e-05,
      "loss": 1.9137,
      "step": 208835
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.5378639698028564,
      "learning_rate": 1.2878409769169762e-05,
      "loss": 2.0166,
      "step": 208836
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.465082049369812,
      "learning_rate": 1.2877224144815702e-05,
      "loss": 2.0521,
      "step": 208837
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.7042096853256226,
      "learning_rate": 1.2876038573843106e-05,
      "loss": 1.9568,
      "step": 208838
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.5593079328536987,
      "learning_rate": 1.287485305625231e-05,
      "loss": 2.1366,
      "step": 208839
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.4662185907363892,
      "learning_rate": 1.287366759204348e-05,
      "loss": 1.9671,
      "step": 208840
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.4851638078689575,
      "learning_rate": 1.2872482181216848e-05,
      "loss": 2.1999,
      "step": 208841
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.4818768501281738,
      "learning_rate": 1.2871296823772647e-05,
      "loss": 1.9555,
      "step": 208842
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.5673604011535645,
      "learning_rate": 1.2870111519711147e-05,
      "loss": 1.864,
      "step": 208843
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.577205777168274,
      "learning_rate": 1.2868926269032442e-05,
      "loss": 1.8311,
      "step": 208844
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.373204231262207,
      "learning_rate": 1.286774107173687e-05,
      "loss": 1.7343,
      "step": 208845
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.4397457838058472,
      "learning_rate": 1.2866555927824563e-05,
      "loss": 2.0298,
      "step": 208846
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.5944323539733887,
      "learning_rate": 1.2865370837295819e-05,
      "loss": 1.9846,
      "step": 208847
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.575894832611084,
      "learning_rate": 1.2864185800150805e-05,
      "loss": 1.8997,
      "step": 208848
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.3834470510482788,
      "learning_rate": 1.2863000816389791e-05,
      "loss": 1.9139,
      "step": 208849
    },
    {
      "epoch": 2.72,
      "grad_norm": 2.372215509414673,
      "learning_rate": 1.2861815886012938e-05,
      "loss": 1.9545,
      "step": 208850
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.4001489877700806,
      "learning_rate": 1.2860631009020484e-05,
      "loss": 1.8102,
      "step": 208851
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.3614606857299805,
      "learning_rate": 1.2859446185412658e-05,
      "loss": 1.8532,
      "step": 208852
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.4213502407073975,
      "learning_rate": 1.2858261415189696e-05,
      "loss": 1.8369,
      "step": 208853
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.5558453798294067,
      "learning_rate": 1.2857076698351797e-05,
      "loss": 1.9062,
      "step": 208854
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.4380269050598145,
      "learning_rate": 1.2855892034899195e-05,
      "loss": 1.9384,
      "step": 208855
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.5580743551254272,
      "learning_rate": 1.2854707424832122e-05,
      "loss": 1.9548,
      "step": 208856
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.572040319442749,
      "learning_rate": 1.2853522868150746e-05,
      "loss": 1.9521,
      "step": 208857
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.4492390155792236,
      "learning_rate": 1.285233836485533e-05,
      "loss": 2.0027,
      "step": 208858
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.330706238746643,
      "learning_rate": 1.2851153914946077e-05,
      "loss": 2.0406,
      "step": 208859
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.4994865655899048,
      "learning_rate": 1.284996951842322e-05,
      "loss": 2.134,
      "step": 208860
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.394304633140564,
      "learning_rate": 1.284878517528699e-05,
      "loss": 1.9621,
      "step": 208861
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.4746204614639282,
      "learning_rate": 1.284760088553759e-05,
      "loss": 2.0416,
      "step": 208862
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.5188186168670654,
      "learning_rate": 1.2846416649175217e-05,
      "loss": 1.9217,
      "step": 208863
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.4840450286865234,
      "learning_rate": 1.2845232466200107e-05,
      "loss": 2.0284,
      "step": 208864
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.5570579767227173,
      "learning_rate": 1.2844048336612522e-05,
      "loss": 1.9423,
      "step": 208865
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.3708053827285767,
      "learning_rate": 1.2842864260412633e-05,
      "loss": 1.9126,
      "step": 208866
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.457290768623352,
      "learning_rate": 1.2841680237600671e-05,
      "loss": 2.0355,
      "step": 208867
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.4758552312850952,
      "learning_rate": 1.2840496268176903e-05,
      "loss": 1.9423,
      "step": 208868
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.5404579639434814,
      "learning_rate": 1.2839312352141461e-05,
      "loss": 2.0868,
      "step": 208869
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.4494045972824097,
      "learning_rate": 1.2838128489494614e-05,
      "loss": 1.9419,
      "step": 208870
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.4307217597961426,
      "learning_rate": 1.2836944680236594e-05,
      "loss": 2.029,
      "step": 208871
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.4594758749008179,
      "learning_rate": 1.28357609243676e-05,
      "loss": 1.9296,
      "step": 208872
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.391298770904541,
      "learning_rate": 1.2834577221887864e-05,
      "loss": 1.8201,
      "step": 208873
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.3400568962097168,
      "learning_rate": 1.2833393572797624e-05,
      "loss": 1.8587,
      "step": 208874
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.638004183769226,
      "learning_rate": 1.2832209977097008e-05,
      "loss": 1.891,
      "step": 208875
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.6389856338500977,
      "learning_rate": 1.2831026434786351e-05,
      "loss": 1.9197,
      "step": 208876
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.5142778158187866,
      "learning_rate": 1.2829842945865853e-05,
      "loss": 2.085,
      "step": 208877
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.4349761009216309,
      "learning_rate": 1.2828659510335647e-05,
      "loss": 2.2156,
      "step": 208878
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.595387578010559,
      "learning_rate": 1.2827476128196068e-05,
      "loss": 2.0398,
      "step": 208879
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.432305932044983,
      "learning_rate": 1.282629279944728e-05,
      "loss": 1.9816,
      "step": 208880
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.5260127782821655,
      "learning_rate": 1.2825109524089449e-05,
      "loss": 1.8955,
      "step": 208881
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.4183363914489746,
      "learning_rate": 1.282392630212291e-05,
      "loss": 1.942,
      "step": 208882
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.3970353603363037,
      "learning_rate": 1.2822743133547764e-05,
      "loss": 1.9158,
      "step": 208883
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.42417573928833,
      "learning_rate": 1.282156001836434e-05,
      "loss": 2.2432,
      "step": 208884
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.5190855264663696,
      "learning_rate": 1.282037695657281e-05,
      "loss": 2.1165,
      "step": 208885
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.7362066507339478,
      "learning_rate": 1.2819193948173401e-05,
      "loss": 1.7156,
      "step": 208886
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.5890491008758545,
      "learning_rate": 1.2818010993166284e-05,
      "loss": 2.2255,
      "step": 208887
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.4833873510360718,
      "learning_rate": 1.2816828091551723e-05,
      "loss": 1.993,
      "step": 208888
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.4445173740386963,
      "learning_rate": 1.2815645243329953e-05,
      "loss": 1.8254,
      "step": 208889
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.415648341178894,
      "learning_rate": 1.2814462448501172e-05,
      "loss": 2.0213,
      "step": 208890
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.3756263256072998,
      "learning_rate": 1.2813279707065615e-05,
      "loss": 1.6829,
      "step": 208891
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.5901663303375244,
      "learning_rate": 1.2812097019023481e-05,
      "loss": 1.921,
      "step": 208892
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.4158000946044922,
      "learning_rate": 1.2810914384375004e-05,
      "loss": 2.3804,
      "step": 208893
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.3264213800430298,
      "learning_rate": 1.2809731803120383e-05,
      "loss": 1.9165,
      "step": 208894
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.3219690322875977,
      "learning_rate": 1.280854927525985e-05,
      "loss": 1.8464,
      "step": 208895
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.432462453842163,
      "learning_rate": 1.280736680079364e-05,
      "loss": 1.8915,
      "step": 208896
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.3797894716262817,
      "learning_rate": 1.280618437972195e-05,
      "loss": 1.9247,
      "step": 208897
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.4009228944778442,
      "learning_rate": 1.2805002012045085e-05,
      "loss": 1.8075,
      "step": 208898
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.4794204235076904,
      "learning_rate": 1.2803819697763107e-05,
      "loss": 2.0126,
      "step": 208899
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.5394459962844849,
      "learning_rate": 1.2802637436876351e-05,
      "loss": 1.7199,
      "step": 208900
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.4853426218032837,
      "learning_rate": 1.2801455229384982e-05,
      "loss": 2.0114,
      "step": 208901
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.3548762798309326,
      "learning_rate": 1.2800273075289268e-05,
      "loss": 1.9166,
      "step": 208902
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.4577842950820923,
      "learning_rate": 1.2799090974589377e-05,
      "loss": 2.0404,
      "step": 208903
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.4835549592971802,
      "learning_rate": 1.2797908927285638e-05,
      "loss": 1.944,
      "step": 208904
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.415143370628357,
      "learning_rate": 1.2796726933378087e-05,
      "loss": 1.9131,
      "step": 208905
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.408659815788269,
      "learning_rate": 1.2795544992867123e-05,
      "loss": 2.1012,
      "step": 208906
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.406313180923462,
      "learning_rate": 1.2794363105752813e-05,
      "loss": 1.9206,
      "step": 208907
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.4468084573745728,
      "learning_rate": 1.2793181272035525e-05,
      "loss": 2.0871,
      "step": 208908
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.435431957244873,
      "learning_rate": 1.2791999491715355e-05,
      "loss": 1.9641,
      "step": 208909
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.6431328058242798,
      "learning_rate": 1.2790817764792637e-05,
      "loss": 1.9679,
      "step": 208910
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.5283823013305664,
      "learning_rate": 1.2789636091267441e-05,
      "loss": 2.0352,
      "step": 208911
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.4862765073776245,
      "learning_rate": 1.278845447114013e-05,
      "loss": 1.6414,
      "step": 208912
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.3190866708755493,
      "learning_rate": 1.278727290441084e-05,
      "loss": 2.0163,
      "step": 208913
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.3855788707733154,
      "learning_rate": 1.2786091391079833e-05,
      "loss": 2.1052,
      "step": 208914
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.520391821861267,
      "learning_rate": 1.278490993114728e-05,
      "loss": 2.0884,
      "step": 208915
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.4319055080413818,
      "learning_rate": 1.2783728524613513e-05,
      "loss": 1.6903,
      "step": 208916
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.5199744701385498,
      "learning_rate": 1.278254717147863e-05,
      "loss": 1.9904,
      "step": 208917
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.3279131650924683,
      "learning_rate": 1.27813658717429e-05,
      "loss": 2.0551,
      "step": 208918
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.4918010234832764,
      "learning_rate": 1.2780184625406487e-05,
      "loss": 2.0616,
      "step": 208919
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.5545958280563354,
      "learning_rate": 1.2779003432469692e-05,
      "loss": 1.8274,
      "step": 208920
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.6915711164474487,
      "learning_rate": 1.2777822292932682e-05,
      "loss": 1.9129,
      "step": 208921
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.5550562143325806,
      "learning_rate": 1.2776641206795757e-05,
      "loss": 2.0508,
      "step": 208922
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.4443154335021973,
      "learning_rate": 1.2775460174059049e-05,
      "loss": 2.027,
      "step": 208923
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.5423436164855957,
      "learning_rate": 1.2774279194722791e-05,
      "loss": 2.0045,
      "step": 208924
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.488789439201355,
      "learning_rate": 1.2773098268787186e-05,
      "loss": 1.9064,
      "step": 208925
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.512715220451355,
      "learning_rate": 1.277191739625253e-05,
      "loss": 1.9365,
      "step": 208926
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.520204782485962,
      "learning_rate": 1.2770736577118956e-05,
      "loss": 2.0648,
      "step": 208927
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.4444645643234253,
      "learning_rate": 1.2769555811386768e-05,
      "loss": 1.8237,
      "step": 208928
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.536607265472412,
      "learning_rate": 1.276837509905616e-05,
      "loss": 1.8304,
      "step": 208929
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.5078243017196655,
      "learning_rate": 1.2767194440127304e-05,
      "loss": 1.9975,
      "step": 208930
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.4247525930404663,
      "learning_rate": 1.276601383460043e-05,
      "loss": 2.2123,
      "step": 208931
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.6356340646743774,
      "learning_rate": 1.2764833282475773e-05,
      "loss": 1.9238,
      "step": 208932
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.4596631526947021,
      "learning_rate": 1.2763652783753564e-05,
      "loss": 2.0425,
      "step": 208933
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.494493007659912,
      "learning_rate": 1.2762472338434037e-05,
      "loss": 1.8886,
      "step": 208934
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.428218126296997,
      "learning_rate": 1.2761291946517393e-05,
      "loss": 1.7141,
      "step": 208935
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.5172030925750732,
      "learning_rate": 1.2760111608003831e-05,
      "loss": 1.9197,
      "step": 208936
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.4501038789749146,
      "learning_rate": 1.2758931322893551e-05,
      "loss": 2.222,
      "step": 208937
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.5444914102554321,
      "learning_rate": 1.2757751091186852e-05,
      "loss": 1.8823,
      "step": 208938
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.383481740951538,
      "learning_rate": 1.2756570912883867e-05,
      "loss": 1.9605,
      "step": 208939
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.5826290845870972,
      "learning_rate": 1.2755390787984899e-05,
      "loss": 1.9098,
      "step": 208940
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.5445098876953125,
      "learning_rate": 1.2754210716490143e-05,
      "loss": 1.9662,
      "step": 208941
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.4424169063568115,
      "learning_rate": 1.2753030698399768e-05,
      "loss": 2.0346,
      "step": 208942
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.4934766292572021,
      "learning_rate": 1.2751850733714009e-05,
      "loss": 1.931,
      "step": 208943
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.4181066751480103,
      "learning_rate": 1.275067082243313e-05,
      "loss": 1.8847,
      "step": 208944
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.3298243284225464,
      "learning_rate": 1.2749490964557296e-05,
      "loss": 1.9545,
      "step": 208945
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.5005669593811035,
      "learning_rate": 1.2748311160086811e-05,
      "loss": 2.1953,
      "step": 208946
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.5887784957885742,
      "learning_rate": 1.2747131409021805e-05,
      "loss": 2.0086,
      "step": 208947
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.3779650926589966,
      "learning_rate": 1.2745951711362546e-05,
      "loss": 2.0131,
      "step": 208948
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.4523996114730835,
      "learning_rate": 1.27447720671092e-05,
      "loss": 2.1533,
      "step": 208949
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.4772510528564453,
      "learning_rate": 1.2743592476262033e-05,
      "loss": 1.9152,
      "step": 208950
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.5117470026016235,
      "learning_rate": 1.2742412938821245e-05,
      "loss": 1.9679,
      "step": 208951
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.6819251775741577,
      "learning_rate": 1.2741233454787103e-05,
      "loss": 1.9465,
      "step": 208952
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.32704496383667,
      "learning_rate": 1.2740054024159774e-05,
      "loss": 1.9473,
      "step": 208953
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.5092644691467285,
      "learning_rate": 1.2738874646939489e-05,
      "loss": 2.0247,
      "step": 208954
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.3922163248062134,
      "learning_rate": 1.2737695323126452e-05,
      "loss": 2.0016,
      "step": 208955
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.4053261280059814,
      "learning_rate": 1.2736516052720924e-05,
      "loss": 2.0759,
      "step": 208956
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.4559158086776733,
      "learning_rate": 1.2735336835723075e-05,
      "loss": 1.8116,
      "step": 208957
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.5836530923843384,
      "learning_rate": 1.2734157672133171e-05,
      "loss": 2.0297,
      "step": 208958
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.5071271657943726,
      "learning_rate": 1.2732978561951413e-05,
      "loss": 1.9413,
      "step": 208959
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.1987192630767822,
      "learning_rate": 1.2731799505177998e-05,
      "loss": 1.9104,
      "step": 208960
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.4194821119308472,
      "learning_rate": 1.273062050181316e-05,
      "loss": 1.8715,
      "step": 208961
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.4412953853607178,
      "learning_rate": 1.2729441551857167e-05,
      "loss": 2.0123,
      "step": 208962
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.4938056468963623,
      "learning_rate": 1.2728262655310117e-05,
      "loss": 2.0919,
      "step": 208963
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.4062745571136475,
      "learning_rate": 1.2727083812172379e-05,
      "loss": 1.9274,
      "step": 208964
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.5531457662582397,
      "learning_rate": 1.2725905022444083e-05,
      "loss": 1.8025,
      "step": 208965
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.4871164560317993,
      "learning_rate": 1.272472628612543e-05,
      "loss": 2.1617,
      "step": 208966
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.505927562713623,
      "learning_rate": 1.2723547603216688e-05,
      "loss": 1.8612,
      "step": 208967
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.5858590602874756,
      "learning_rate": 1.2722368973718056e-05,
      "loss": 1.8382,
      "step": 208968
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.455160140991211,
      "learning_rate": 1.2721190397629799e-05,
      "loss": 2.1613,
      "step": 208969
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.342298984527588,
      "learning_rate": 1.2720011874952051e-05,
      "loss": 1.7817,
      "step": 208970
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.4067373275756836,
      "learning_rate": 1.2718833405685113e-05,
      "loss": 1.9476,
      "step": 208971
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.3995181322097778,
      "learning_rate": 1.2717654989829119e-05,
      "loss": 2.0642,
      "step": 208972
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.3006033897399902,
      "learning_rate": 1.2716476627384365e-05,
      "loss": 2.027,
      "step": 208973
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.4529224634170532,
      "learning_rate": 1.2715298318351019e-05,
      "loss": 2.0256,
      "step": 208974
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.3553433418273926,
      "learning_rate": 1.271412006272935e-05,
      "loss": 2.0886,
      "step": 208975
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.4361928701400757,
      "learning_rate": 1.2712941860519554e-05,
      "loss": 1.8622,
      "step": 208976
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.4776637554168701,
      "learning_rate": 1.2711763711721868e-05,
      "loss": 1.933,
      "step": 208977
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.4383141994476318,
      "learning_rate": 1.2710585616336421e-05,
      "loss": 1.9579,
      "step": 208978
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.606195092201233,
      "learning_rate": 1.2709407574363551e-05,
      "loss": 1.7534,
      "step": 208979
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.4910489320755005,
      "learning_rate": 1.2708229585803354e-05,
      "loss": 2.0897,
      "step": 208980
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.5822422504425049,
      "learning_rate": 1.2707051650656197e-05,
      "loss": 2.1099,
      "step": 208981
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.3470793962478638,
      "learning_rate": 1.2705873768922148e-05,
      "loss": 2.0329,
      "step": 208982
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.5174150466918945,
      "learning_rate": 1.2704695940601606e-05,
      "loss": 2.016,
      "step": 208983
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.5203701257705688,
      "learning_rate": 1.2703518165694604e-05,
      "loss": 1.9486,
      "step": 208984
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.4052698612213135,
      "learning_rate": 1.2702340444201475e-05,
      "loss": 1.9092,
      "step": 208985
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.527003526687622,
      "learning_rate": 1.2701162776122353e-05,
      "loss": 2.0008,
      "step": 208986
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.436586618423462,
      "learning_rate": 1.269998516145757e-05,
      "loss": 2.0049,
      "step": 208987
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.4356762170791626,
      "learning_rate": 1.2698807600207228e-05,
      "loss": 2.0612,
      "step": 208988
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.5141748189926147,
      "learning_rate": 1.2697630092371657e-05,
      "loss": 1.8669,
      "step": 208989
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.4362481832504272,
      "learning_rate": 1.2696452637950994e-05,
      "loss": 1.8727,
      "step": 208990
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.5078595876693726,
      "learning_rate": 1.2695275236945502e-05,
      "loss": 2.1191,
      "step": 208991
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.569901943206787,
      "learning_rate": 1.2694097889355348e-05,
      "loss": 1.8035,
      "step": 208992
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.4567556381225586,
      "learning_rate": 1.26929205951808e-05,
      "loss": 2.0066,
      "step": 208993
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.4140737056732178,
      "learning_rate": 1.2691743354422057e-05,
      "loss": 1.9374,
      "step": 208994
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.3722648620605469,
      "learning_rate": 1.2690566167079352e-05,
      "loss": 2.1334,
      "step": 208995
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.3226388692855835,
      "learning_rate": 1.2689389033152886e-05,
      "loss": 1.9024,
      "step": 208996
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.4300918579101562,
      "learning_rate": 1.2688211952642924e-05,
      "loss": 1.9045,
      "step": 208997
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.3291730880737305,
      "learning_rate": 1.2687034925549566e-05,
      "loss": 1.8862,
      "step": 208998
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.7278565168380737,
      "learning_rate": 1.2685857951873179e-05,
      "loss": 2.0173,
      "step": 208999
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.3388646841049194,
      "learning_rate": 1.2684681031613864e-05,
      "loss": 2.0482,
      "step": 209000
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.5408341884613037,
      "learning_rate": 1.2683504164771951e-05,
      "loss": 1.8,
      "step": 209001
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.5901468992233276,
      "learning_rate": 1.2682327351347577e-05,
      "loss": 2.1622,
      "step": 209002
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.6404472589492798,
      "learning_rate": 1.2681150591340971e-05,
      "loss": 2.0808,
      "step": 209003
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.6793646812438965,
      "learning_rate": 1.2679973884752337e-05,
      "loss": 1.7714,
      "step": 209004
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.3938803672790527,
      "learning_rate": 1.2678797231581972e-05,
      "loss": 2.0021,
      "step": 209005
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.5880082845687866,
      "learning_rate": 1.2677620631829977e-05,
      "loss": 2.2604,
      "step": 209006
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.3501636981964111,
      "learning_rate": 1.2676444085496685e-05,
      "loss": 1.8156,
      "step": 209007
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.3720262050628662,
      "learning_rate": 1.2675267592582295e-05,
      "loss": 1.9276,
      "step": 209008
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.4874125719070435,
      "learning_rate": 1.2674091153086973e-05,
      "loss": 1.917,
      "step": 209009
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.3844741582870483,
      "learning_rate": 1.2672914767010889e-05,
      "loss": 2.126,
      "step": 209010
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.458743691444397,
      "learning_rate": 1.2671738434354406e-05,
      "loss": 1.8149,
      "step": 209011
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.4973870515823364,
      "learning_rate": 1.2670562155117624e-05,
      "loss": 1.9987,
      "step": 209012
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.4907385110855103,
      "learning_rate": 1.266938592930088e-05,
      "loss": 1.8664,
      "step": 209013
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.415814757347107,
      "learning_rate": 1.2668209756904269e-05,
      "loss": 1.8828,
      "step": 209014
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.5871219635009766,
      "learning_rate": 1.2667033637928093e-05,
      "loss": 2.0089,
      "step": 209015
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.5356428623199463,
      "learning_rate": 1.2665857572372484e-05,
      "loss": 2.0828,
      "step": 209016
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.3946659564971924,
      "learning_rate": 1.2664681560237777e-05,
      "loss": 1.779,
      "step": 209017
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.6565601825714111,
      "learning_rate": 1.2663505601524071e-05,
      "loss": 1.8465,
      "step": 209018
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.3383457660675049,
      "learning_rate": 1.2662329696231699e-05,
      "loss": 1.9243,
      "step": 209019
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.3481000661849976,
      "learning_rate": 1.2661153844360827e-05,
      "loss": 1.7745,
      "step": 209020
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.3867220878601074,
      "learning_rate": 1.2659978045911656e-05,
      "loss": 2.1767,
      "step": 209021
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.530974268913269,
      "learning_rate": 1.2658802300884352e-05,
      "loss": 1.9065,
      "step": 209022
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.525782585144043,
      "learning_rate": 1.2657626609279281e-05,
      "loss": 2.1143,
      "step": 209023
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.3147624731063843,
      "learning_rate": 1.2656450971096543e-05,
      "loss": 2.1855,
      "step": 209024
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.5659043788909912,
      "learning_rate": 1.2655275386336405e-05,
      "loss": 1.9844,
      "step": 209025
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.4578367471694946,
      "learning_rate": 1.2654099854999101e-05,
      "loss": 2.0329,
      "step": 209026
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.4831351041793823,
      "learning_rate": 1.2652924377084795e-05,
      "loss": 1.9283,
      "step": 209027
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.5353949069976807,
      "learning_rate": 1.2651748952593721e-05,
      "loss": 2.089,
      "step": 209028
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.4594465494155884,
      "learning_rate": 1.2650573581526147e-05,
      "loss": 2.1431,
      "step": 209029
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.5579283237457275,
      "learning_rate": 1.2649398263882204e-05,
      "loss": 2.0515,
      "step": 209030
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.2815022468566895,
      "learning_rate": 1.2648222999662227e-05,
      "loss": 1.9466,
      "step": 209031
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.6012401580810547,
      "learning_rate": 1.2647047788866349e-05,
      "loss": 1.806,
      "step": 209032
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.4482276439666748,
      "learning_rate": 1.2645872631494803e-05,
      "loss": 1.9385,
      "step": 209033
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.445961594581604,
      "learning_rate": 1.2644697527547787e-05,
      "loss": 2.0289,
      "step": 209034
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.4693008661270142,
      "learning_rate": 1.2643522477025604e-05,
      "loss": 1.8759,
      "step": 209035
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.4881634712219238,
      "learning_rate": 1.2642347479928349e-05,
      "loss": 2.2129,
      "step": 209036
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.3920085430145264,
      "learning_rate": 1.2641172536256327e-05,
      "loss": 1.9777,
      "step": 209037
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.5507367849349976,
      "learning_rate": 1.263999764600977e-05,
      "loss": 1.983,
      "step": 209038
    },
    {
      "epoch": 2.72,
      "grad_norm": 2.4322497844696045,
      "learning_rate": 1.2638822809188842e-05,
      "loss": 1.9151,
      "step": 209039
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.60653555393219,
      "learning_rate": 1.2637648025793744e-05,
      "loss": 1.9704,
      "step": 209040
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.4241752624511719,
      "learning_rate": 1.2636473295824779e-05,
      "loss": 1.8232,
      "step": 209041
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.465516448020935,
      "learning_rate": 1.2635298619282075e-05,
      "loss": 1.9764,
      "step": 209042
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.5090479850769043,
      "learning_rate": 1.2634123996165935e-05,
      "loss": 2.1191,
      "step": 209043
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.3792450428009033,
      "learning_rate": 1.2632949426476557e-05,
      "loss": 1.9485,
      "step": 209044
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.477124571800232,
      "learning_rate": 1.2631774910214077e-05,
      "loss": 2.1451,
      "step": 209045
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.4781111478805542,
      "learning_rate": 1.2630600447378792e-05,
      "loss": 1.7291,
      "step": 209046
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.5545213222503662,
      "learning_rate": 1.2629426037970936e-05,
      "loss": 1.9251,
      "step": 209047
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.4365354776382446,
      "learning_rate": 1.2628251681990643e-05,
      "loss": 2.0966,
      "step": 209048
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.5170960426330566,
      "learning_rate": 1.262707737943821e-05,
      "loss": 1.8782,
      "step": 209049
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.4342873096466064,
      "learning_rate": 1.262590313031384e-05,
      "loss": 2.0983,
      "step": 209050
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.2726980447769165,
      "learning_rate": 1.26247289346177e-05,
      "loss": 1.9893,
      "step": 209051
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.5825973749160767,
      "learning_rate": 1.2623554792350087e-05,
      "loss": 1.8704,
      "step": 209052
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.503793716430664,
      "learning_rate": 1.262238070351117e-05,
      "loss": 1.9761,
      "step": 209053
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.5361602306365967,
      "learning_rate": 1.2621206668101147e-05,
      "loss": 1.9381,
      "step": 209054
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.4763028621673584,
      "learning_rate": 1.2620032686120318e-05,
      "loss": 2.2283,
      "step": 209055
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.3633602857589722,
      "learning_rate": 1.2618858757568818e-05,
      "loss": 1.8697,
      "step": 209056
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.483675241470337,
      "learning_rate": 1.261768488244691e-05,
      "loss": 1.7443,
      "step": 209057
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.4814618825912476,
      "learning_rate": 1.2616511060754797e-05,
      "loss": 1.7458,
      "step": 209058
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.4276882410049438,
      "learning_rate": 1.2615337292492678e-05,
      "loss": 2.0429,
      "step": 209059
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.4369027614593506,
      "learning_rate": 1.261416357766082e-05,
      "loss": 2.1254,
      "step": 209060
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.7295986413955688,
      "learning_rate": 1.261298991625942e-05,
      "loss": 1.9378,
      "step": 209061
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.3174738883972168,
      "learning_rate": 1.2611816308288646e-05,
      "loss": 2.1794,
      "step": 209062
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.5114110708236694,
      "learning_rate": 1.2610642753748801e-05,
      "loss": 2.0176,
      "step": 209063
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.4205381870269775,
      "learning_rate": 1.2609469252640047e-05,
      "loss": 1.9372,
      "step": 209064
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.3422913551330566,
      "learning_rate": 1.2608295804962587e-05,
      "loss": 2.0023,
      "step": 209065
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.6986098289489746,
      "learning_rate": 1.2607122410716718e-05,
      "loss": 2.0893,
      "step": 209066
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.5273542404174805,
      "learning_rate": 1.2605949069902576e-05,
      "loss": 1.9098,
      "step": 209067
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.4716416597366333,
      "learning_rate": 1.2604775782520426e-05,
      "loss": 2.2188,
      "step": 209068
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.4377810955047607,
      "learning_rate": 1.2603602548570502e-05,
      "loss": 2.0002,
      "step": 209069
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.3901089429855347,
      "learning_rate": 1.2602429368053003e-05,
      "loss": 1.961,
      "step": 209070
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.400663137435913,
      "learning_rate": 1.260125624096806e-05,
      "loss": 1.8914,
      "step": 209071
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.8714977502822876,
      "learning_rate": 1.2600083167316044e-05,
      "loss": 2.0873,
      "step": 209072
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.498528003692627,
      "learning_rate": 1.2598910147097019e-05,
      "loss": 2.1453,
      "step": 209073
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.5220693349838257,
      "learning_rate": 1.2597737180311352e-05,
      "loss": 2.0783,
      "step": 209074
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.3862706422805786,
      "learning_rate": 1.2596564266959175e-05,
      "loss": 1.8387,
      "step": 209075
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.514595866203308,
      "learning_rate": 1.2595391407040722e-05,
      "loss": 1.9919,
      "step": 209076
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.4104655981063843,
      "learning_rate": 1.2594218600556193e-05,
      "loss": 1.8376,
      "step": 209077
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.5258426666259766,
      "learning_rate": 1.2593045847505855e-05,
      "loss": 2.0354,
      "step": 209078
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.465680480003357,
      "learning_rate": 1.2591873147889842e-05,
      "loss": 2.0967,
      "step": 209079
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.478403925895691,
      "learning_rate": 1.2590700501708451e-05,
      "loss": 1.8509,
      "step": 209080
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.4837064743041992,
      "learning_rate": 1.2589527908961916e-05,
      "loss": 2.0833,
      "step": 209081
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.5181275606155396,
      "learning_rate": 1.2588355369650372e-05,
      "loss": 1.9724,
      "step": 209082
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.4819461107254028,
      "learning_rate": 1.258718288377405e-05,
      "loss": 1.9169,
      "step": 209083
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.5983433723449707,
      "learning_rate": 1.2586010451333251e-05,
      "loss": 2.045,
      "step": 209084
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.5377256870269775,
      "learning_rate": 1.2584838072328107e-05,
      "loss": 1.7829,
      "step": 209085
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.3492939472198486,
      "learning_rate": 1.2583665746758887e-05,
      "loss": 1.9373,
      "step": 209086
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.4772013425827026,
      "learning_rate": 1.2582493474625787e-05,
      "loss": 2.2159,
      "step": 209087
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.4412662982940674,
      "learning_rate": 1.2581321255929045e-05,
      "loss": 1.9906,
      "step": 209088
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.4707928895950317,
      "learning_rate": 1.258014909066879e-05,
      "loss": 2.1409,
      "step": 209089
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.5772373676300049,
      "learning_rate": 1.2578976978845357e-05,
      "loss": 1.8009,
      "step": 209090
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.5527573823928833,
      "learning_rate": 1.257780492045891e-05,
      "loss": 1.926,
      "step": 209091
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.5681880712509155,
      "learning_rate": 1.2576632915509688e-05,
      "loss": 1.9627,
      "step": 209092
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.4467628002166748,
      "learning_rate": 1.2575460963997886e-05,
      "loss": 1.8443,
      "step": 209093
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.595639705657959,
      "learning_rate": 1.257428906592377e-05,
      "loss": 2.0116,
      "step": 209094
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.370897650718689,
      "learning_rate": 1.2573117221287444e-05,
      "loss": 2.007,
      "step": 209095
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.3474618196487427,
      "learning_rate": 1.257194543008927e-05,
      "loss": 1.7602,
      "step": 209096
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.583353042602539,
      "learning_rate": 1.2570773692329316e-05,
      "loss": 1.9934,
      "step": 209097
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.36629056930542,
      "learning_rate": 1.256960200800795e-05,
      "loss": 1.8775,
      "step": 209098
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.4878954887390137,
      "learning_rate": 1.2568430377125304e-05,
      "loss": 1.9502,
      "step": 209099
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.5334619283676147,
      "learning_rate": 1.2567258799681645e-05,
      "loss": 2.0743,
      "step": 209100
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.4592206478118896,
      "learning_rate": 1.2566087275677106e-05,
      "loss": 1.9885,
      "step": 209101
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.4940029382705688,
      "learning_rate": 1.2564915805111986e-05,
      "loss": 1.8851,
      "step": 209102
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.3816417455673218,
      "learning_rate": 1.2563744387986418e-05,
      "loss": 2.0569,
      "step": 209103
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.4315409660339355,
      "learning_rate": 1.2562573024300737e-05,
      "loss": 2.0,
      "step": 209104
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.6019232273101807,
      "learning_rate": 1.2561401714055108e-05,
      "loss": 2.0813,
      "step": 209105
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.4194482564926147,
      "learning_rate": 1.2560230457249731e-05,
      "loss": 1.901,
      "step": 209106
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.516169548034668,
      "learning_rate": 1.2559059253884774e-05,
      "loss": 2.1232,
      "step": 209107
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.502418875694275,
      "learning_rate": 1.2557888103960568e-05,
      "loss": 1.9026,
      "step": 209108
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.403472900390625,
      "learning_rate": 1.2556717007477245e-05,
      "loss": 2.0265,
      "step": 209109
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.418022871017456,
      "learning_rate": 1.2555545964435077e-05,
      "loss": 1.9977,
      "step": 209110
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.397233486175537,
      "learning_rate": 1.2554374974834292e-05,
      "loss": 2.0274,
      "step": 209111
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.369473934173584,
      "learning_rate": 1.2553204038675058e-05,
      "loss": 1.794,
      "step": 209112
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.5367668867111206,
      "learning_rate": 1.2552033155957542e-05,
      "loss": 1.925,
      "step": 209113
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.4977186918258667,
      "learning_rate": 1.255086232668211e-05,
      "loss": 1.9474,
      "step": 209114
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.4949144124984741,
      "learning_rate": 1.2549691550848829e-05,
      "loss": 2.1331,
      "step": 209115
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.406129002571106,
      "learning_rate": 1.2548520828458031e-05,
      "loss": 2.1189,
      "step": 209116
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.3423404693603516,
      "learning_rate": 1.2547350159509917e-05,
      "loss": 2.0021,
      "step": 209117
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.6338446140289307,
      "learning_rate": 1.2546179544004653e-05,
      "loss": 1.8421,
      "step": 209118
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.3981457948684692,
      "learning_rate": 1.2545008981942439e-05,
      "loss": 1.9131,
      "step": 209119
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.474825143814087,
      "learning_rate": 1.2543838473323575e-05,
      "loss": 1.9971,
      "step": 209120
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.6600711345672607,
      "learning_rate": 1.2542668018148194e-05,
      "loss": 2.0914,
      "step": 209121
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.5213991403579712,
      "learning_rate": 1.2541497616416596e-05,
      "loss": 1.8152,
      "step": 209122
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.5800156593322754,
      "learning_rate": 1.2540327268128947e-05,
      "loss": 1.9274,
      "step": 209123
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.409760594367981,
      "learning_rate": 1.2539156973285513e-05,
      "loss": 2.0172,
      "step": 209124
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.6007729768753052,
      "learning_rate": 1.2537986731886429e-05,
      "loss": 2.0056,
      "step": 209125
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.4791990518569946,
      "learning_rate": 1.2536816543931994e-05,
      "loss": 2.0582,
      "step": 209126
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.5616337060928345,
      "learning_rate": 1.253564640942234e-05,
      "loss": 2.1644,
      "step": 209127
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.3380733728408813,
      "learning_rate": 1.2534476328357802e-05,
      "loss": 1.9753,
      "step": 209128
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.3290983438491821,
      "learning_rate": 1.2533306300738444e-05,
      "loss": 1.9572,
      "step": 209129
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.4978142976760864,
      "learning_rate": 1.2532136326564667e-05,
      "loss": 2.1138,
      "step": 209130
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.5893303155899048,
      "learning_rate": 1.253096640583654e-05,
      "loss": 2.0643,
      "step": 209131
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.4375197887420654,
      "learning_rate": 1.252979653855436e-05,
      "loss": 1.8844,
      "step": 209132
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.5176734924316406,
      "learning_rate": 1.2528626724718293e-05,
      "loss": 2.1698,
      "step": 209133
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.5995538234710693,
      "learning_rate": 1.2527456964328575e-05,
      "loss": 2.2964,
      "step": 209134
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.503509759902954,
      "learning_rate": 1.2526287257385437e-05,
      "loss": 1.9528,
      "step": 209135
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.4881798028945923,
      "learning_rate": 1.2525117603889112e-05,
      "loss": 2.0258,
      "step": 209136
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.605393409729004,
      "learning_rate": 1.2523948003839768e-05,
      "loss": 2.3242,
      "step": 209137
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.6076158285140991,
      "learning_rate": 1.252277845723767e-05,
      "loss": 2.1992,
      "step": 209138
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.4463199377059937,
      "learning_rate": 1.2521608964082986e-05,
      "loss": 1.8995,
      "step": 209139
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.4095064401626587,
      "learning_rate": 1.252043952437598e-05,
      "loss": 2.0686,
      "step": 209140
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.4957517385482788,
      "learning_rate": 1.2519270138116822e-05,
      "loss": 2.1036,
      "step": 209141
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.5027424097061157,
      "learning_rate": 1.2518100805305808e-05,
      "loss": 2.1103,
      "step": 209142
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.5554155111312866,
      "learning_rate": 1.2516931525943108e-05,
      "loss": 2.031,
      "step": 209143
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.3279938697814941,
      "learning_rate": 1.2515762300028886e-05,
      "loss": 2.1825,
      "step": 209144
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.574224591255188,
      "learning_rate": 1.2514593127563444e-05,
      "loss": 1.9629,
      "step": 209145
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.4803169965744019,
      "learning_rate": 1.251342400854698e-05,
      "loss": 2.0539,
      "step": 209146
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.5044915676116943,
      "learning_rate": 1.2512254942979626e-05,
      "loss": 2.2128,
      "step": 209147
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.4981446266174316,
      "learning_rate": 1.2511085930861753e-05,
      "loss": 1.8135,
      "step": 209148
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.451200246810913,
      "learning_rate": 1.2509916972193457e-05,
      "loss": 2.028,
      "step": 209149
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.3020761013031006,
      "learning_rate": 1.2508748066974972e-05,
      "loss": 2.1509,
      "step": 209150
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.474277377128601,
      "learning_rate": 1.2507579215206597e-05,
      "loss": 1.83,
      "step": 209151
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.509440541267395,
      "learning_rate": 1.2506410416888468e-05,
      "loss": 2.2008,
      "step": 209152
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.4980463981628418,
      "learning_rate": 1.2505241672020782e-05,
      "loss": 2.2099,
      "step": 209153
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.4666273593902588,
      "learning_rate": 1.250407298060384e-05,
      "loss": 1.8528,
      "step": 209154
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.5087450742721558,
      "learning_rate": 1.250290434263781e-05,
      "loss": 2.0245,
      "step": 209155
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.6109094619750977,
      "learning_rate": 1.2501735758122889e-05,
      "loss": 2.0818,
      "step": 209156
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.5546293258666992,
      "learning_rate": 1.2500567227059377e-05,
      "loss": 1.964,
      "step": 209157
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.336645245552063,
      "learning_rate": 1.2499398749447376e-05,
      "loss": 2.0796,
      "step": 209158
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.5112898349761963,
      "learning_rate": 1.2498230325287217e-05,
      "loss": 2.0165,
      "step": 209159
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.41908860206604,
      "learning_rate": 1.2497061954579036e-05,
      "loss": 2.0051,
      "step": 209160
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.454581618309021,
      "learning_rate": 1.2495893637323096e-05,
      "loss": 2.0937,
      "step": 209161
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.3504612445831299,
      "learning_rate": 1.2494725373519565e-05,
      "loss": 1.8799,
      "step": 209162
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.3573707342147827,
      "learning_rate": 1.2493557163168743e-05,
      "loss": 1.7467,
      "step": 209163
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.4066611528396606,
      "learning_rate": 1.2492389006270731e-05,
      "loss": 1.7627,
      "step": 209164
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.4330648183822632,
      "learning_rate": 1.249122090282586e-05,
      "loss": 1.9669,
      "step": 209165
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.4810506105422974,
      "learning_rate": 1.2490052852834298e-05,
      "loss": 2.0786,
      "step": 209166
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.5835645198822021,
      "learning_rate": 1.2488884856296245e-05,
      "loss": 2.1131,
      "step": 209167
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.363781213760376,
      "learning_rate": 1.248771691321193e-05,
      "loss": 2.0255,
      "step": 209168
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.4964828491210938,
      "learning_rate": 1.2486549023581594e-05,
      "loss": 1.7648,
      "step": 209169
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.3528136014938354,
      "learning_rate": 1.2485381187405396e-05,
      "loss": 2.0471,
      "step": 209170
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.5585267543792725,
      "learning_rate": 1.248421340468364e-05,
      "loss": 2.0259,
      "step": 209171
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.3255188465118408,
      "learning_rate": 1.2483045675416492e-05,
      "loss": 2.0692,
      "step": 209172
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.5064687728881836,
      "learning_rate": 1.248187799960415e-05,
      "loss": 2.0315,
      "step": 209173
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.7303071022033691,
      "learning_rate": 1.248071037724685e-05,
      "loss": 2.2583,
      "step": 209174
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.3696041107177734,
      "learning_rate": 1.2479542808344822e-05,
      "loss": 1.9681,
      "step": 209175
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.3353369235992432,
      "learning_rate": 1.2478375292898268e-05,
      "loss": 1.9213,
      "step": 209176
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.4633444547653198,
      "learning_rate": 1.2477207830907421e-05,
      "loss": 1.7712,
      "step": 209177
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.488609790802002,
      "learning_rate": 1.2476040422372513e-05,
      "loss": 2.0776,
      "step": 209178
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.4920494556427002,
      "learning_rate": 1.2474873067293745e-05,
      "loss": 2.0392,
      "step": 209179
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.5849177837371826,
      "learning_rate": 1.247370576567125e-05,
      "loss": 1.9522,
      "step": 209180
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.4422321319580078,
      "learning_rate": 1.2472538517505393e-05,
      "loss": 2.028,
      "step": 209181
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.5413950681686401,
      "learning_rate": 1.2471371322796241e-05,
      "loss": 2.174,
      "step": 209182
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.4699878692626953,
      "learning_rate": 1.2470204181544164e-05,
      "loss": 1.899,
      "step": 209183
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.50877046585083,
      "learning_rate": 1.246903709374929e-05,
      "loss": 2.2147,
      "step": 209184
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.5941696166992188,
      "learning_rate": 1.2467870059411856e-05,
      "loss": 2.0072,
      "step": 209185
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.604356050491333,
      "learning_rate": 1.2466703078532025e-05,
      "loss": 1.7838,
      "step": 209186
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.5024728775024414,
      "learning_rate": 1.24655361511101e-05,
      "loss": 2.0499,
      "step": 209187
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.3765523433685303,
      "learning_rate": 1.2464369277146213e-05,
      "loss": 2.0605,
      "step": 209188
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.5181447267532349,
      "learning_rate": 1.2463202456640698e-05,
      "loss": 2.0124,
      "step": 209189
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.556073784828186,
      "learning_rate": 1.246203568959362e-05,
      "loss": 1.938,
      "step": 209190
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.374613642692566,
      "learning_rate": 1.2460868976005377e-05,
      "loss": 1.8688,
      "step": 209191
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.3677843809127808,
      "learning_rate": 1.2459702315876009e-05,
      "loss": 1.8981,
      "step": 209192
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.498378872871399,
      "learning_rate": 1.2458535709205842e-05,
      "loss": 1.9934,
      "step": 209193
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.5006712675094604,
      "learning_rate": 1.2457369155995045e-05,
      "loss": 1.8551,
      "step": 209194
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.4516303539276123,
      "learning_rate": 1.2456202656243853e-05,
      "loss": 1.8109,
      "step": 209195
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.5669524669647217,
      "learning_rate": 1.2455036209952463e-05,
      "loss": 1.9256,
      "step": 209196
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.921141266822815,
      "learning_rate": 1.245386981712121e-05,
      "loss": 1.9304,
      "step": 209197
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.433236837387085,
      "learning_rate": 1.2452703477750093e-05,
      "loss": 1.8023,
      "step": 209198
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.4195525646209717,
      "learning_rate": 1.245153719183951e-05,
      "loss": 1.965,
      "step": 209199
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.6242810487747192,
      "learning_rate": 1.2450370959389567e-05,
      "loss": 1.9694,
      "step": 209200
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.4830782413482666,
      "learning_rate": 1.2449204780400557e-05,
      "loss": 1.9568,
      "step": 209201
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.4570441246032715,
      "learning_rate": 1.244803865487265e-05,
      "loss": 2.0043,
      "step": 209202
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.494543194770813,
      "learning_rate": 1.2446872582806145e-05,
      "loss": 1.93,
      "step": 209203
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.6458675861358643,
      "learning_rate": 1.2445706564201107e-05,
      "loss": 2.0571,
      "step": 209204
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.4858595132827759,
      "learning_rate": 1.2444540599057872e-05,
      "loss": 1.7195,
      "step": 209205
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.4326363801956177,
      "learning_rate": 1.2443374687376606e-05,
      "loss": 1.8733,
      "step": 209206
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.5609490871429443,
      "learning_rate": 1.2442208829157607e-05,
      "loss": 2.0777,
      "step": 209207
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.332666277885437,
      "learning_rate": 1.2441043024400943e-05,
      "loss": 2.0159,
      "step": 209208
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.52384614944458,
      "learning_rate": 1.2439877273107013e-05,
      "loss": 1.7528,
      "step": 209209
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.4315333366394043,
      "learning_rate": 1.2438711575275884e-05,
      "loss": 1.9614,
      "step": 209210
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.5080640316009521,
      "learning_rate": 1.2437545930907822e-05,
      "loss": 1.9408,
      "step": 209211
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.5006277561187744,
      "learning_rate": 1.243638034000306e-05,
      "loss": 1.9699,
      "step": 209212
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.3340256214141846,
      "learning_rate": 1.2435214802561799e-05,
      "loss": 1.883,
      "step": 209213
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.6450817584991455,
      "learning_rate": 1.2434049318584239e-05,
      "loss": 1.8092,
      "step": 209214
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.5902419090270996,
      "learning_rate": 1.243288388807071e-05,
      "loss": 1.8039,
      "step": 209215
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.3644156455993652,
      "learning_rate": 1.2431718511021215e-05,
      "loss": 1.9222,
      "step": 209216
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.4215543270111084,
      "learning_rate": 1.2430553187436154e-05,
      "loss": 1.8557,
      "step": 209217
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.574249029159546,
      "learning_rate": 1.2429387917315658e-05,
      "loss": 2.0061,
      "step": 209218
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.4718259572982788,
      "learning_rate": 1.2428222700659995e-05,
      "loss": 2.1604,
      "step": 209219
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.7409828901290894,
      "learning_rate": 1.2427057537469331e-05,
      "loss": 1.9677,
      "step": 209220
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.5441086292266846,
      "learning_rate": 1.24258924277439e-05,
      "loss": 2.0036,
      "step": 209221
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.3800878524780273,
      "learning_rate": 1.2424727371483967e-05,
      "loss": 1.9264,
      "step": 209222
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.6289141178131104,
      "learning_rate": 1.24235623686897e-05,
      "loss": 1.9334,
      "step": 209223
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.528362512588501,
      "learning_rate": 1.2422397419361264e-05,
      "loss": 2.2036,
      "step": 209224
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.4417495727539062,
      "learning_rate": 1.2421232523498958e-05,
      "loss": 2.046,
      "step": 209225
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.614722728729248,
      "learning_rate": 1.2420067681102952e-05,
      "loss": 2.0504,
      "step": 209226
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.4109888076782227,
      "learning_rate": 1.2418902892173543e-05,
      "loss": 1.9367,
      "step": 209227
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.3829059600830078,
      "learning_rate": 1.2417738156710865e-05,
      "loss": 1.8333,
      "step": 209228
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.38727605342865,
      "learning_rate": 1.241657347471512e-05,
      "loss": 2.2264,
      "step": 209229
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.4249606132507324,
      "learning_rate": 1.2415408846186604e-05,
      "loss": 1.9402,
      "step": 209230
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.292322039604187,
      "learning_rate": 1.2414244271125518e-05,
      "loss": 1.7428,
      "step": 209231
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.4826713800430298,
      "learning_rate": 1.2413079749531995e-05,
      "loss": 1.9853,
      "step": 209232
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.479358434677124,
      "learning_rate": 1.2411915281406337e-05,
      "loss": 1.7728,
      "step": 209233
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.467986822128296,
      "learning_rate": 1.2410750866748742e-05,
      "loss": 2.0698,
      "step": 209234
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.421343207359314,
      "learning_rate": 1.2409586505559377e-05,
      "loss": 1.9198,
      "step": 209235
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.4472051858901978,
      "learning_rate": 1.2408422197838541e-05,
      "loss": 2.1421,
      "step": 209236
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.3982006311416626,
      "learning_rate": 1.2407257943586402e-05,
      "loss": 1.9583,
      "step": 209237
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.58203125,
      "learning_rate": 1.2406093742803125e-05,
      "loss": 1.894,
      "step": 209238
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.3512485027313232,
      "learning_rate": 1.2404929595489044e-05,
      "loss": 2.1367,
      "step": 209239
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.4463653564453125,
      "learning_rate": 1.2403765501644325e-05,
      "loss": 1.8935,
      "step": 209240
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.3944318294525146,
      "learning_rate": 1.2402601461269135e-05,
      "loss": 1.884,
      "step": 209241
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.3023464679718018,
      "learning_rate": 1.240143747436374e-05,
      "loss": 2.0858,
      "step": 209242
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.423524022102356,
      "learning_rate": 1.240027354092834e-05,
      "loss": 1.6538,
      "step": 209243
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.5965065956115723,
      "learning_rate": 1.2399109660963169e-05,
      "loss": 1.9498,
      "step": 209244
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.4824670553207397,
      "learning_rate": 1.2397945834468459e-05,
      "loss": 1.8961,
      "step": 209245
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.6391156911849976,
      "learning_rate": 1.2396782061444377e-05,
      "loss": 2.0125,
      "step": 209246
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.4410648345947266,
      "learning_rate": 1.2395618341891156e-05,
      "loss": 2.0204,
      "step": 209247
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.4218735694885254,
      "learning_rate": 1.2394454675809029e-05,
      "loss": 1.9549,
      "step": 209248
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.4272866249084473,
      "learning_rate": 1.2393291063198162e-05,
      "loss": 1.9666,
      "step": 209249
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.3598097562789917,
      "learning_rate": 1.2392127504058858e-05,
      "loss": 1.8352,
      "step": 209250
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.6396232843399048,
      "learning_rate": 1.239096399839128e-05,
      "loss": 2.0047,
      "step": 209251
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.5407226085662842,
      "learning_rate": 1.238980054619566e-05,
      "loss": 2.1892,
      "step": 209252
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.5844699144363403,
      "learning_rate": 1.238863714747217e-05,
      "loss": 2.1115,
      "step": 209253
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.4860252141952515,
      "learning_rate": 1.2387473802221104e-05,
      "loss": 1.852,
      "step": 209254
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.3588156700134277,
      "learning_rate": 1.2386310510442565e-05,
      "loss": 1.9726,
      "step": 209255
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.3903473615646362,
      "learning_rate": 1.238514727213692e-05,
      "loss": 2.026,
      "step": 209256
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.5659812688827515,
      "learning_rate": 1.2383984087304233e-05,
      "loss": 2.0334,
      "step": 209257
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.4910918474197388,
      "learning_rate": 1.2382820955944871e-05,
      "loss": 2.032,
      "step": 209258
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.497933268547058,
      "learning_rate": 1.2381657878058937e-05,
      "loss": 2.0977,
      "step": 209259
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.5602809190750122,
      "learning_rate": 1.2380494853646662e-05,
      "loss": 1.8828,
      "step": 209260
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.6791131496429443,
      "learning_rate": 1.2379331882708276e-05,
      "loss": 2.1989,
      "step": 209261
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.7570337057113647,
      "learning_rate": 1.237816896524405e-05,
      "loss": 1.8949,
      "step": 209262
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.4804209470748901,
      "learning_rate": 1.2377006101254083e-05,
      "loss": 1.8644,
      "step": 209263
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.466949701309204,
      "learning_rate": 1.2375843290738774e-05,
      "loss": 1.8491,
      "step": 209264
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.472550868988037,
      "learning_rate": 1.2374680533698122e-05,
      "loss": 1.821,
      "step": 209265
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.5367841720581055,
      "learning_rate": 1.2373517830132496e-05,
      "loss": 1.9666,
      "step": 209266
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.5682129859924316,
      "learning_rate": 1.2372355180042025e-05,
      "loss": 1.9715,
      "step": 209267
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.4910814762115479,
      "learning_rate": 1.2371192583426981e-05,
      "loss": 2.227,
      "step": 209268
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.6320284605026245,
      "learning_rate": 1.2370030040287527e-05,
      "loss": 2.1299,
      "step": 209269
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.5482560396194458,
      "learning_rate": 1.2368867550623995e-05,
      "loss": 1.9409,
      "step": 209270
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.4934766292572021,
      "learning_rate": 1.2367705114436454e-05,
      "loss": 1.825,
      "step": 209271
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.446521520614624,
      "learning_rate": 1.2366542731725238e-05,
      "loss": 2.0198,
      "step": 209272
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.461517572402954,
      "learning_rate": 1.2365380402490443e-05,
      "loss": 2.0993,
      "step": 209273
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.4156033992767334,
      "learning_rate": 1.2364218126732406e-05,
      "loss": 1.8891,
      "step": 209274
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.3644320964813232,
      "learning_rate": 1.2363055904451258e-05,
      "loss": 1.8187,
      "step": 209275
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.780621886253357,
      "learning_rate": 1.2361893735647298e-05,
      "loss": 1.8455,
      "step": 209276
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.3997260332107544,
      "learning_rate": 1.236073162032063e-05,
      "loss": 1.9467,
      "step": 209277
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.3947681188583374,
      "learning_rate": 1.2359569558471583e-05,
      "loss": 1.7478,
      "step": 209278
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.5766334533691406,
      "learning_rate": 1.2358407550100258e-05,
      "loss": 1.9383,
      "step": 209279
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.4598371982574463,
      "learning_rate": 1.2357245595206988e-05,
      "loss": 2.0774,
      "step": 209280
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.6138545274734497,
      "learning_rate": 1.2356083693791907e-05,
      "loss": 1.8063,
      "step": 209281
    },
    {
      "epoch": 2.72,
      "grad_norm": 1.5062251091003418,
      "learning_rate": 1.2354921845855314e-05,
      "loss": 1.7742,
      "step": 209282
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.4815541505813599,
      "learning_rate": 1.235376005139731e-05,
      "loss": 1.6985,
      "step": 209283
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.4555678367614746,
      "learning_rate": 1.2352598310418228e-05,
      "loss": 1.9604,
      "step": 209284
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.4898710250854492,
      "learning_rate": 1.2351436622918165e-05,
      "loss": 1.8957,
      "step": 209285
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.479551076889038,
      "learning_rate": 1.2350274988897423e-05,
      "loss": 2.1132,
      "step": 209286
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.4714454412460327,
      "learning_rate": 1.234911340835617e-05,
      "loss": 1.829,
      "step": 209287
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.4696954488754272,
      "learning_rate": 1.2347951881294737e-05,
      "loss": 1.7675,
      "step": 209288
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.57319974899292,
      "learning_rate": 1.2346790407713159e-05,
      "loss": 1.8672,
      "step": 209289
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.5211865901947021,
      "learning_rate": 1.23456289876118e-05,
      "loss": 1.9212,
      "step": 209290
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.483534574508667,
      "learning_rate": 1.2344467620990762e-05,
      "loss": 2.0677,
      "step": 209291
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.3792804479599,
      "learning_rate": 1.2343306307850342e-05,
      "loss": 1.9173,
      "step": 209292
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.413020133972168,
      "learning_rate": 1.2342145048190711e-05,
      "loss": 1.9857,
      "step": 209293
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.3523221015930176,
      "learning_rate": 1.2340983842012198e-05,
      "loss": 1.7499,
      "step": 209294
    },
    {
      "epoch": 2.73,
      "grad_norm": 2.6169888973236084,
      "learning_rate": 1.2339822689314804e-05,
      "loss": 2.0189,
      "step": 209295
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.3204939365386963,
      "learning_rate": 1.233866159009893e-05,
      "loss": 1.9522,
      "step": 209296
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.3530988693237305,
      "learning_rate": 1.2337500544364709e-05,
      "loss": 1.9849,
      "step": 209297
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.3080377578735352,
      "learning_rate": 1.2336339552112407e-05,
      "loss": 2.039,
      "step": 209298
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.5850965976715088,
      "learning_rate": 1.2335178613342156e-05,
      "loss": 2.0457,
      "step": 209299
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.583878755569458,
      "learning_rate": 1.2334017728054324e-05,
      "loss": 1.9695,
      "step": 209300
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.4527361392974854,
      "learning_rate": 1.233285689624891e-05,
      "loss": 1.8267,
      "step": 209301
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.5440561771392822,
      "learning_rate": 1.2331696117926315e-05,
      "loss": 2.0675,
      "step": 209302
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.4319157600402832,
      "learning_rate": 1.2330535393086672e-05,
      "loss": 2.0094,
      "step": 209303
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.4961323738098145,
      "learning_rate": 1.2329374721730211e-05,
      "loss": 1.8502,
      "step": 209304
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.6064406633377075,
      "learning_rate": 1.2328214103857137e-05,
      "loss": 1.9051,
      "step": 209305
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.48051917552948,
      "learning_rate": 1.2327053539467712e-05,
      "loss": 2.0908,
      "step": 209306
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.5410594940185547,
      "learning_rate": 1.2325893028562106e-05,
      "loss": 1.9152,
      "step": 209307
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.5082733631134033,
      "learning_rate": 1.232473257114055e-05,
      "loss": 2.0146,
      "step": 209308
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.323879361152649,
      "learning_rate": 1.2323572167203211e-05,
      "loss": 1.9238,
      "step": 209309
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.5419743061065674,
      "learning_rate": 1.232241181675042e-05,
      "loss": 2.014,
      "step": 209310
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.4758049249649048,
      "learning_rate": 1.2321251519782249e-05,
      "loss": 1.8793,
      "step": 209311
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.615046501159668,
      "learning_rate": 1.232009127629906e-05,
      "loss": 1.6718,
      "step": 209312
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.5325154066085815,
      "learning_rate": 1.2318931086300954e-05,
      "loss": 2.0764,
      "step": 209313
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.5529242753982544,
      "learning_rate": 1.231777094978823e-05,
      "loss": 1.8103,
      "step": 209314
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.456122636795044,
      "learning_rate": 1.2316610866760989e-05,
      "loss": 2.0056,
      "step": 209315
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.431904673576355,
      "learning_rate": 1.2315450837219599e-05,
      "loss": 1.9471,
      "step": 209316
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.4119117259979248,
      "learning_rate": 1.2314290861164123e-05,
      "loss": 1.8981,
      "step": 209317
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.3696565628051758,
      "learning_rate": 1.2313130938594896e-05,
      "loss": 1.7959,
      "step": 209318
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.561398983001709,
      "learning_rate": 1.2311971069512117e-05,
      "loss": 1.9955,
      "step": 209319
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.5850319862365723,
      "learning_rate": 1.231081125391592e-05,
      "loss": 1.9878,
      "step": 209320
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.5480215549468994,
      "learning_rate": 1.2309651491806605e-05,
      "loss": 2.1407,
      "step": 209321
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.301160454750061,
      "learning_rate": 1.2308491783184338e-05,
      "loss": 1.8805,
      "step": 209322
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.4343159198760986,
      "learning_rate": 1.2307332128049353e-05,
      "loss": 1.8155,
      "step": 209323
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.516579031944275,
      "learning_rate": 1.230617252640188e-05,
      "loss": 1.8613,
      "step": 209324
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.4895139932632446,
      "learning_rate": 1.2305012978242123e-05,
      "loss": 2.3081,
      "step": 209325
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.3898935317993164,
      "learning_rate": 1.2303853483570248e-05,
      "loss": 1.8215,
      "step": 209326
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.425518274307251,
      "learning_rate": 1.2302694042386552e-05,
      "loss": 1.8347,
      "step": 209327
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.4716119766235352,
      "learning_rate": 1.2301534654691203e-05,
      "loss": 1.8255,
      "step": 209328
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.4861916303634644,
      "learning_rate": 1.2300375320484468e-05,
      "loss": 2.1799,
      "step": 209329
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.541469693183899,
      "learning_rate": 1.2299216039766512e-05,
      "loss": 1.8319,
      "step": 209330
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.48166024684906,
      "learning_rate": 1.2298056812537539e-05,
      "loss": 2.2074,
      "step": 209331
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.4518333673477173,
      "learning_rate": 1.2296897638797776e-05,
      "loss": 2.1765,
      "step": 209332
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.411188006401062,
      "learning_rate": 1.2295738518547492e-05,
      "loss": 1.958,
      "step": 209333
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.4370203018188477,
      "learning_rate": 1.2294579451786823e-05,
      "loss": 2.2147,
      "step": 209334
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.4425276517868042,
      "learning_rate": 1.2293420438516033e-05,
      "loss": 1.942,
      "step": 209335
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.430373191833496,
      "learning_rate": 1.2292261478735355e-05,
      "loss": 2.0508,
      "step": 209336
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.5162928104400635,
      "learning_rate": 1.2291102572444988e-05,
      "loss": 1.9175,
      "step": 209337
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.5220006704330444,
      "learning_rate": 1.2289943719645069e-05,
      "loss": 1.9695,
      "step": 209338
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.5534547567367554,
      "learning_rate": 1.2288784920335925e-05,
      "loss": 2.0522,
      "step": 209339
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.4217392206192017,
      "learning_rate": 1.2287626174517695e-05,
      "loss": 1.9711,
      "step": 209340
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.6869210004806519,
      "learning_rate": 1.2286467482190676e-05,
      "loss": 1.9348,
      "step": 209341
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.3707952499389648,
      "learning_rate": 1.2285308843354969e-05,
      "loss": 1.7141,
      "step": 209342
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.4020816087722778,
      "learning_rate": 1.2284150258010938e-05,
      "loss": 1.8619,
      "step": 209343
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.3548579216003418,
      "learning_rate": 1.2282991726158653e-05,
      "loss": 1.9583,
      "step": 209344
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.515410304069519,
      "learning_rate": 1.2281833247798411e-05,
      "loss": 1.8464,
      "step": 209345
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.4964441061019897,
      "learning_rate": 1.228067482293038e-05,
      "loss": 1.8543,
      "step": 209346
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.4055464267730713,
      "learning_rate": 1.227951645155486e-05,
      "loss": 1.8268,
      "step": 209347
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.503482699394226,
      "learning_rate": 1.2278358133671951e-05,
      "loss": 1.9902,
      "step": 209348
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.4833825826644897,
      "learning_rate": 1.2277199869281984e-05,
      "loss": 1.9703,
      "step": 209349
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.4383991956710815,
      "learning_rate": 1.227604165838506e-05,
      "loss": 1.9755,
      "step": 209350
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.572300672531128,
      "learning_rate": 1.227488350098148e-05,
      "loss": 2.0743,
      "step": 209351
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.4260274171829224,
      "learning_rate": 1.227372539707141e-05,
      "loss": 1.761,
      "step": 209352
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.4517648220062256,
      "learning_rate": 1.2272567346655115e-05,
      "loss": 2.1047,
      "step": 209353
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.3884336948394775,
      "learning_rate": 1.227140934973273e-05,
      "loss": 1.8623,
      "step": 209354
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.4386579990386963,
      "learning_rate": 1.2270251406304621e-05,
      "loss": 2.0976,
      "step": 209355
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.3227945566177368,
      "learning_rate": 1.226909351637082e-05,
      "loss": 1.9882,
      "step": 209356
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.5688529014587402,
      "learning_rate": 1.2267935679931628e-05,
      "loss": 1.9764,
      "step": 209357
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.5523045063018799,
      "learning_rate": 1.2266777896987279e-05,
      "loss": 2.1157,
      "step": 209358
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.3817005157470703,
      "learning_rate": 1.226562016753797e-05,
      "loss": 1.9342,
      "step": 209359
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.4081599712371826,
      "learning_rate": 1.226446249158387e-05,
      "loss": 2.0432,
      "step": 209360
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.4074137210845947,
      "learning_rate": 1.2263304869125346e-05,
      "loss": 1.9214,
      "step": 209361
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.403181791305542,
      "learning_rate": 1.2262147300162396e-05,
      "loss": 1.9424,
      "step": 209362
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.5354565382003784,
      "learning_rate": 1.2260989784695385e-05,
      "loss": 2.0053,
      "step": 209363
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.4989831447601318,
      "learning_rate": 1.2259832322724483e-05,
      "loss": 2.1305,
      "step": 209364
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.4116891622543335,
      "learning_rate": 1.2258674914249922e-05,
      "loss": 1.9117,
      "step": 209365
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.4084590673446655,
      "learning_rate": 1.2257517559271868e-05,
      "loss": 2.0156,
      "step": 209366
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.3745529651641846,
      "learning_rate": 1.2256360257790655e-05,
      "loss": 2.0159,
      "step": 209367
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.557140827178955,
      "learning_rate": 1.2255203009806313e-05,
      "loss": 2.1298,
      "step": 209368
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.458938479423523,
      "learning_rate": 1.2254045815319247e-05,
      "loss": 2.1666,
      "step": 209369
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.360550045967102,
      "learning_rate": 1.2252888674329519e-05,
      "loss": 1.8276,
      "step": 209370
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.4206410646438599,
      "learning_rate": 1.2251731586837433e-05,
      "loss": 1.8151,
      "step": 209371
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.5271403789520264,
      "learning_rate": 1.2250574552843185e-05,
      "loss": 1.9787,
      "step": 209372
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.4449734687805176,
      "learning_rate": 1.2249417572347042e-05,
      "loss": 1.9156,
      "step": 209373
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.8060510158538818,
      "learning_rate": 1.2248260645349073e-05,
      "loss": 2.0956,
      "step": 209374
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.5359418392181396,
      "learning_rate": 1.2247103771849642e-05,
      "loss": 2.1068,
      "step": 209375
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.4738221168518066,
      "learning_rate": 1.2245946951848851e-05,
      "loss": 2.0972,
      "step": 209376
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.5003929138183594,
      "learning_rate": 1.224479018534703e-05,
      "loss": 2.0365,
      "step": 209377
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.4528766870498657,
      "learning_rate": 1.2243633472344284e-05,
      "loss": 1.9871,
      "step": 209378
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.5663875341415405,
      "learning_rate": 1.224247681284094e-05,
      "loss": 1.9737,
      "step": 209379
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.3930939435958862,
      "learning_rate": 1.2241320206837101e-05,
      "loss": 1.7963,
      "step": 209380
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.4386183023452759,
      "learning_rate": 1.2240163654333068e-05,
      "loss": 1.8294,
      "step": 209381
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.6343135833740234,
      "learning_rate": 1.2239007155328973e-05,
      "loss": 1.9942,
      "step": 209382
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.4666341543197632,
      "learning_rate": 1.2237850709825114e-05,
      "loss": 1.9673,
      "step": 209383
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.382986068725586,
      "learning_rate": 1.2236694317821627e-05,
      "loss": 2.1355,
      "step": 209384
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.45418119430542,
      "learning_rate": 1.223553797931881e-05,
      "loss": 1.912,
      "step": 209385
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.5181313753128052,
      "learning_rate": 1.2234381694316863e-05,
      "loss": 2.0733,
      "step": 209386
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.4906178712844849,
      "learning_rate": 1.2233225462815955e-05,
      "loss": 1.942,
      "step": 209387
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.4261258840560913,
      "learning_rate": 1.2232069284816282e-05,
      "loss": 1.9127,
      "step": 209388
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.5958977937698364,
      "learning_rate": 1.2230913160318145e-05,
      "loss": 2.1182,
      "step": 209389
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.3680419921875,
      "learning_rate": 1.222975708932168e-05,
      "loss": 1.8679,
      "step": 209390
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.435495376586914,
      "learning_rate": 1.2228601071827182e-05,
      "loss": 1.9804,
      "step": 209391
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.5525795221328735,
      "learning_rate": 1.2227445107834788e-05,
      "loss": 1.9219,
      "step": 209392
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.4065498113632202,
      "learning_rate": 1.2226289197344763e-05,
      "loss": 1.836,
      "step": 209393
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.706015706062317,
      "learning_rate": 1.2225133340357274e-05,
      "loss": 2.0905,
      "step": 209394
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.3503389358520508,
      "learning_rate": 1.2223977536872586e-05,
      "loss": 1.9474,
      "step": 209395
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.8591254949569702,
      "learning_rate": 1.2222821786890869e-05,
      "loss": 1.976,
      "step": 209396
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.4223498106002808,
      "learning_rate": 1.2221666090412385e-05,
      "loss": 1.8604,
      "step": 209397
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.4409582614898682,
      "learning_rate": 1.2220510447437338e-05,
      "loss": 2.0239,
      "step": 209398
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.4955171346664429,
      "learning_rate": 1.2219354857965924e-05,
      "loss": 1.9511,
      "step": 209399
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.3961135149002075,
      "learning_rate": 1.2218199321998345e-05,
      "loss": 2.0923,
      "step": 209400
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.4493728876113892,
      "learning_rate": 1.2217043839534835e-05,
      "loss": 1.9737,
      "step": 209401
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.4057260751724243,
      "learning_rate": 1.2215888410575592e-05,
      "loss": 2.0988,
      "step": 209402
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.4961845874786377,
      "learning_rate": 1.2214733035120883e-05,
      "loss": 2.0674,
      "step": 209403
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.2841676473617554,
      "learning_rate": 1.2213577713170908e-05,
      "loss": 1.8679,
      "step": 209404
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.3447015285491943,
      "learning_rate": 1.22124224447258e-05,
      "loss": 2.1148,
      "step": 209405
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.4198882579803467,
      "learning_rate": 1.221126722978586e-05,
      "loss": 1.9918,
      "step": 209406
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.4671324491500854,
      "learning_rate": 1.2210112068351318e-05,
      "loss": 1.9503,
      "step": 209407
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.3918752670288086,
      "learning_rate": 1.2208956960422277e-05,
      "loss": 1.9255,
      "step": 209408
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.4011305570602417,
      "learning_rate": 1.2207801905999071e-05,
      "loss": 1.9374,
      "step": 209409
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.6457360982894897,
      "learning_rate": 1.2206646905081864e-05,
      "loss": 2.0882,
      "step": 209410
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.3682223558425903,
      "learning_rate": 1.2205491957670855e-05,
      "loss": 2.061,
      "step": 209411
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.5750000476837158,
      "learning_rate": 1.2204337063766279e-05,
      "loss": 2.2094,
      "step": 209412
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.3958165645599365,
      "learning_rate": 1.2203182223368336e-05,
      "loss": 2.0134,
      "step": 209413
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.5468448400497437,
      "learning_rate": 1.220202743647729e-05,
      "loss": 1.6975,
      "step": 209414
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.4983243942260742,
      "learning_rate": 1.2200872703093312e-05,
      "loss": 1.9209,
      "step": 209415
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.416867971420288,
      "learning_rate": 1.2199718023216631e-05,
      "loss": 1.9908,
      "step": 209416
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.4564582109451294,
      "learning_rate": 1.2198563396847417e-05,
      "loss": 2.1943,
      "step": 209417
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.4841821193695068,
      "learning_rate": 1.2197408823985966e-05,
      "loss": 1.8606,
      "step": 209418
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.478546142578125,
      "learning_rate": 1.2196254304632414e-05,
      "loss": 2.127,
      "step": 209419
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.4004508256912231,
      "learning_rate": 1.219509983878706e-05,
      "loss": 2.0901,
      "step": 209420
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.7959978580474854,
      "learning_rate": 1.2193945426450035e-05,
      "loss": 1.9369,
      "step": 209421
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.4888299703598022,
      "learning_rate": 1.219279106762161e-05,
      "loss": 2.0544,
      "step": 209422
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.3858474493026733,
      "learning_rate": 1.2191636762301914e-05,
      "loss": 2.0304,
      "step": 209423
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.6711889505386353,
      "learning_rate": 1.2190482510491284e-05,
      "loss": 1.7443,
      "step": 209424
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.5275931358337402,
      "learning_rate": 1.218932831218985e-05,
      "loss": 1.9544,
      "step": 209425
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.510486364364624,
      "learning_rate": 1.218817416739788e-05,
      "loss": 2.0091,
      "step": 209426
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.3994423151016235,
      "learning_rate": 1.218702007611554e-05,
      "loss": 1.8737,
      "step": 209427
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.4806467294692993,
      "learning_rate": 1.2185866038343094e-05,
      "loss": 1.9313,
      "step": 209428
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.3830088376998901,
      "learning_rate": 1.2184712054080714e-05,
      "loss": 1.8176,
      "step": 209429
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.5149321556091309,
      "learning_rate": 1.2183558123328629e-05,
      "loss": 2.0804,
      "step": 209430
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.459259271621704,
      "learning_rate": 1.2182404246087008e-05,
      "loss": 2.0102,
      "step": 209431
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.3763655424118042,
      "learning_rate": 1.2181250422356182e-05,
      "loss": 1.9048,
      "step": 209432
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.4966065883636475,
      "learning_rate": 1.2180096652136217e-05,
      "loss": 1.9612,
      "step": 209433
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.4562205076217651,
      "learning_rate": 1.2178942935427516e-05,
      "loss": 1.8304,
      "step": 209434
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.4418104887008667,
      "learning_rate": 1.2177789272230076e-05,
      "loss": 1.8427,
      "step": 209435
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.5496679544448853,
      "learning_rate": 1.2176635662544266e-05,
      "loss": 2.0343,
      "step": 209436
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.4367127418518066,
      "learning_rate": 1.217548210637025e-05,
      "loss": 1.8849,
      "step": 209437
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.8106228113174438,
      "learning_rate": 1.2174328603708228e-05,
      "loss": 1.9765,
      "step": 209438
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.5713571310043335,
      "learning_rate": 1.2173175154558434e-05,
      "loss": 2.0035,
      "step": 209439
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.3832892179489136,
      "learning_rate": 1.2172021758921135e-05,
      "loss": 1.9646,
      "step": 209440
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.4228609800338745,
      "learning_rate": 1.217086841679643e-05,
      "loss": 1.8507,
      "step": 209441
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.4072978496551514,
      "learning_rate": 1.216971512818462e-05,
      "loss": 1.9177,
      "step": 209442
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.5569077730178833,
      "learning_rate": 1.2168561893085838e-05,
      "loss": 2.1856,
      "step": 209443
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.5501338243484497,
      "learning_rate": 1.2167408711500415e-05,
      "loss": 1.8529,
      "step": 209444
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.5199443101882935,
      "learning_rate": 1.2166255583428485e-05,
      "loss": 2.1238,
      "step": 209445
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.3563377857208252,
      "learning_rate": 1.2165102508870317e-05,
      "loss": 2.1525,
      "step": 209446
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.3379136323928833,
      "learning_rate": 1.2163949487826042e-05,
      "loss": 1.9129,
      "step": 209447
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.3887149095535278,
      "learning_rate": 1.2162796520295926e-05,
      "loss": 1.9841,
      "step": 209448
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.4871100187301636,
      "learning_rate": 1.216164360628017e-05,
      "loss": 1.8871,
      "step": 209449
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.481443166732788,
      "learning_rate": 1.216049074577904e-05,
      "loss": 1.9625,
      "step": 209450
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.6142524480819702,
      "learning_rate": 1.2159337938792669e-05,
      "loss": 1.9837,
      "step": 209451
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.5768479108810425,
      "learning_rate": 1.2158185185321323e-05,
      "loss": 1.9897,
      "step": 209452
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.366694450378418,
      "learning_rate": 1.2157032485365237e-05,
      "loss": 2.0609,
      "step": 209453
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.4990406036376953,
      "learning_rate": 1.2155879838924576e-05,
      "loss": 1.9067,
      "step": 209454
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.6590734720230103,
      "learning_rate": 1.215472724599954e-05,
      "loss": 1.8342,
      "step": 209455
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.6061831712722778,
      "learning_rate": 1.2153574706590396e-05,
      "loss": 2.2171,
      "step": 209456
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.4395326375961304,
      "learning_rate": 1.215242222069731e-05,
      "loss": 1.9112,
      "step": 209457
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.4431580305099487,
      "learning_rate": 1.215126978832055e-05,
      "loss": 2.1913,
      "step": 209458
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.4883859157562256,
      "learning_rate": 1.2150117409460313e-05,
      "loss": 2.0726,
      "step": 209459
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.4575508832931519,
      "learning_rate": 1.2148965084116802e-05,
      "loss": 1.7481,
      "step": 209460
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.4509949684143066,
      "learning_rate": 1.214781281229018e-05,
      "loss": 2.0164,
      "step": 209461
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.591926097869873,
      "learning_rate": 1.214666059398075e-05,
      "loss": 1.6863,
      "step": 209462
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.6281993389129639,
      "learning_rate": 1.2145508429188677e-05,
      "loss": 1.9031,
      "step": 209463
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.7606950998306274,
      "learning_rate": 1.2144356317914194e-05,
      "loss": 2.0064,
      "step": 209464
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.5214142799377441,
      "learning_rate": 1.2143204260157535e-05,
      "loss": 2.3046,
      "step": 209465
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.6077324151992798,
      "learning_rate": 1.21420522559189e-05,
      "loss": 1.9324,
      "step": 209466
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.3989828824996948,
      "learning_rate": 1.214090030519842e-05,
      "loss": 1.9208,
      "step": 209467
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.4067294597625732,
      "learning_rate": 1.2139748407996431e-05,
      "loss": 1.9261,
      "step": 209468
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.511866569519043,
      "learning_rate": 1.2138596564313063e-05,
      "loss": 1.986,
      "step": 209469
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.514485478401184,
      "learning_rate": 1.2137444774148587e-05,
      "loss": 1.9446,
      "step": 209470
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.4000827074050903,
      "learning_rate": 1.2136293037503197e-05,
      "loss": 1.8645,
      "step": 209471
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.4156752824783325,
      "learning_rate": 1.2135141354377132e-05,
      "loss": 1.9975,
      "step": 209472
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.3783921003341675,
      "learning_rate": 1.213398972477052e-05,
      "loss": 1.981,
      "step": 209473
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.4576603174209595,
      "learning_rate": 1.2132838148683665e-05,
      "loss": 2.0069,
      "step": 209474
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.435800313949585,
      "learning_rate": 1.2131686626116732e-05,
      "loss": 2.187,
      "step": 209475
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.5853650569915771,
      "learning_rate": 1.2130535157069954e-05,
      "loss": 1.8917,
      "step": 209476
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.3898625373840332,
      "learning_rate": 1.2129383741543564e-05,
      "loss": 2.1253,
      "step": 209477
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.474181890487671,
      "learning_rate": 1.2128232379537762e-05,
      "loss": 1.7731,
      "step": 209478
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.4916642904281616,
      "learning_rate": 1.2127081071052713e-05,
      "loss": 2.1137,
      "step": 209479
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.497091293334961,
      "learning_rate": 1.2125929816088719e-05,
      "loss": 2.1139,
      "step": 209480
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.3722425699234009,
      "learning_rate": 1.212477861464588e-05,
      "loss": 1.8879,
      "step": 209481
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.491144061088562,
      "learning_rate": 1.2123627466724562e-05,
      "loss": 1.8479,
      "step": 209482
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.3822176456451416,
      "learning_rate": 1.2122476372324862e-05,
      "loss": 2.1214,
      "step": 209483
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.473427414894104,
      "learning_rate": 1.2121325331447052e-05,
      "loss": 2.1329,
      "step": 209484
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.351805329322815,
      "learning_rate": 1.212017434409126e-05,
      "loss": 1.9577,
      "step": 209485
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.4796643257141113,
      "learning_rate": 1.2119023410257823e-05,
      "loss": 2.0667,
      "step": 209486
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.4352715015411377,
      "learning_rate": 1.2117872529946837e-05,
      "loss": 1.8219,
      "step": 209487
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.5178674459457397,
      "learning_rate": 1.2116721703158605e-05,
      "loss": 1.9836,
      "step": 209488
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.3630921840667725,
      "learning_rate": 1.2115570929893326e-05,
      "loss": 2.0621,
      "step": 209489
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.4620171785354614,
      "learning_rate": 1.2114420210151166e-05,
      "loss": 1.8074,
      "step": 209490
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.3391523361206055,
      "learning_rate": 1.2113269543932392e-05,
      "loss": 1.9267,
      "step": 209491
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.4557018280029297,
      "learning_rate": 1.211211893123717e-05,
      "loss": 2.1557,
      "step": 209492
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.3733186721801758,
      "learning_rate": 1.2110968372065733e-05,
      "loss": 1.9372,
      "step": 209493
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.4717631340026855,
      "learning_rate": 1.2109817866418347e-05,
      "loss": 2.0471,
      "step": 209494
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.5474165678024292,
      "learning_rate": 1.2108667414295181e-05,
      "loss": 2.1337,
      "step": 209495
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.4660850763320923,
      "learning_rate": 1.21075170156964e-05,
      "loss": 2.0741,
      "step": 209496
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.450426459312439,
      "learning_rate": 1.2106366670622304e-05,
      "loss": 1.9194,
      "step": 209497
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.5938762426376343,
      "learning_rate": 1.2105216379073057e-05,
      "loss": 1.9182,
      "step": 209498
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.3917078971862793,
      "learning_rate": 1.2104066141048863e-05,
      "loss": 1.9368,
      "step": 209499
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.739976406097412,
      "learning_rate": 1.2102915956549986e-05,
      "loss": 2.3806,
      "step": 209500
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.749397873878479,
      "learning_rate": 1.2101765825576592e-05,
      "loss": 2.0262,
      "step": 209501
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.5347994565963745,
      "learning_rate": 1.2100615748128916e-05,
      "loss": 1.8891,
      "step": 209502
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.490401268005371,
      "learning_rate": 1.209946572420719e-05,
      "loss": 1.9886,
      "step": 209503
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.4232022762298584,
      "learning_rate": 1.2098315753811549e-05,
      "loss": 2.0985,
      "step": 209504
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.444159746170044,
      "learning_rate": 1.2097165836942358e-05,
      "loss": 2.0654,
      "step": 209505
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.4776018857955933,
      "learning_rate": 1.2096015973599682e-05,
      "loss": 2.2144,
      "step": 209506
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.4563566446304321,
      "learning_rate": 1.2094866163783822e-05,
      "loss": 2.0095,
      "step": 209507
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.642508625984192,
      "learning_rate": 1.2093716407494913e-05,
      "loss": 1.8608,
      "step": 209508
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.564165472984314,
      "learning_rate": 1.2092566704733252e-05,
      "loss": 1.9387,
      "step": 209509
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.7052665948867798,
      "learning_rate": 1.2091417055499009e-05,
      "loss": 2.1142,
      "step": 209510
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.5817787647247314,
      "learning_rate": 1.2090267459792413e-05,
      "loss": 2.1042,
      "step": 209511
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.3741474151611328,
      "learning_rate": 1.2089117917613634e-05,
      "loss": 1.8936,
      "step": 209512
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.47011137008667,
      "learning_rate": 1.2087968428963001e-05,
      "loss": 1.9353,
      "step": 209513
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.5201021432876587,
      "learning_rate": 1.2086818993840586e-05,
      "loss": 2.0882,
      "step": 209514
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.3916882276535034,
      "learning_rate": 1.2085669612246685e-05,
      "loss": 2.0206,
      "step": 209515
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.5078377723693848,
      "learning_rate": 1.2084520284181465e-05,
      "loss": 2.1352,
      "step": 209516
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.4050731658935547,
      "learning_rate": 1.2083371009645226e-05,
      "loss": 1.8688,
      "step": 209517
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.4381884336471558,
      "learning_rate": 1.2082221788638035e-05,
      "loss": 2.0735,
      "step": 209518
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.476354956626892,
      "learning_rate": 1.2081072621160292e-05,
      "loss": 1.9123,
      "step": 209519
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.3493895530700684,
      "learning_rate": 1.2079923507212064e-05,
      "loss": 1.9487,
      "step": 209520
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.5475226640701294,
      "learning_rate": 1.2078774446793648e-05,
      "loss": 2.0162,
      "step": 209521
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.4177600145339966,
      "learning_rate": 1.2077625439905148e-05,
      "loss": 1.8198,
      "step": 209522
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.387195110321045,
      "learning_rate": 1.2076476486546926e-05,
      "loss": 2.1214,
      "step": 209523
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.3399633169174194,
      "learning_rate": 1.2075327586719053e-05,
      "loss": 1.85,
      "step": 209524
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.452268123626709,
      "learning_rate": 1.2074178740421859e-05,
      "loss": 1.9452,
      "step": 209525
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.4440118074417114,
      "learning_rate": 1.2073029947655544e-05,
      "loss": 1.7756,
      "step": 209526
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.4257469177246094,
      "learning_rate": 1.2071881208420243e-05,
      "loss": 1.9115,
      "step": 209527
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.3626195192337036,
      "learning_rate": 1.2070732522716187e-05,
      "loss": 1.8502,
      "step": 209528
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.54325532913208,
      "learning_rate": 1.2069583890543677e-05,
      "loss": 1.9714,
      "step": 209529
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.682616949081421,
      "learning_rate": 1.2068435311902814e-05,
      "loss": 2.4269,
      "step": 209530
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.3473871946334839,
      "learning_rate": 1.2067286786793894e-05,
      "loss": 1.9152,
      "step": 209531
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.5195764303207397,
      "learning_rate": 1.2066138315217121e-05,
      "loss": 2.1158,
      "step": 209532
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.4601725339889526,
      "learning_rate": 1.2064989897172662e-05,
      "loss": 1.8747,
      "step": 209533
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.574605941772461,
      "learning_rate": 1.2063841532660745e-05,
      "loss": 1.9799,
      "step": 209534
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.5138298273086548,
      "learning_rate": 1.2062693221681607e-05,
      "loss": 2.1013,
      "step": 209535
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.5804400444030762,
      "learning_rate": 1.2061544964235414e-05,
      "loss": 1.8887,
      "step": 209536
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.4157841205596924,
      "learning_rate": 1.2060396760322465e-05,
      "loss": 2.0322,
      "step": 209537
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.381277084350586,
      "learning_rate": 1.2059248609942929e-05,
      "loss": 1.9987,
      "step": 209538
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.5071995258331299,
      "learning_rate": 1.2058100513097002e-05,
      "loss": 1.9369,
      "step": 209539
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.605008602142334,
      "learning_rate": 1.2056952469784886e-05,
      "loss": 1.9411,
      "step": 209540
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.5507525205612183,
      "learning_rate": 1.2055804480006847e-05,
      "loss": 2.1914,
      "step": 209541
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.4315201044082642,
      "learning_rate": 1.2054656543763019e-05,
      "loss": 1.8881,
      "step": 209542
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.5024263858795166,
      "learning_rate": 1.2053508661053734e-05,
      "loss": 1.8841,
      "step": 209543
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.6001344919204712,
      "learning_rate": 1.2052360831879093e-05,
      "loss": 1.9592,
      "step": 209544
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.4413594007492065,
      "learning_rate": 1.2051213056239394e-05,
      "loss": 2.0222,
      "step": 209545
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.5547924041748047,
      "learning_rate": 1.2050065334134739e-05,
      "loss": 2.1748,
      "step": 209546
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.6004524230957031,
      "learning_rate": 1.2048917665565494e-05,
      "loss": 2.0051,
      "step": 209547
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.4847925901412964,
      "learning_rate": 1.2047770050531724e-05,
      "loss": 1.8236,
      "step": 209548
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.5747836828231812,
      "learning_rate": 1.2046622489033731e-05,
      "loss": 2.0268,
      "step": 209549
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.366923451423645,
      "learning_rate": 1.2045474981071746e-05,
      "loss": 1.8888,
      "step": 209550
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.363865852355957,
      "learning_rate": 1.2044327526645903e-05,
      "loss": 1.9757,
      "step": 209551
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.4399268627166748,
      "learning_rate": 1.2043180125756436e-05,
      "loss": 1.8929,
      "step": 209552
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.4060190916061401,
      "learning_rate": 1.204203277840361e-05,
      "loss": 2.1156,
      "step": 209553
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.7873811721801758,
      "learning_rate": 1.2040885484587592e-05,
      "loss": 2.0466,
      "step": 209554
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.5693148374557495,
      "learning_rate": 1.2039738244308617e-05,
      "loss": 2.154,
      "step": 209555
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.511759638786316,
      "learning_rate": 1.2038591057566916e-05,
      "loss": 1.9383,
      "step": 209556
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.5315868854522705,
      "learning_rate": 1.2037443924362654e-05,
      "loss": 1.9796,
      "step": 209557
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.3200335502624512,
      "learning_rate": 1.2036296844696037e-05,
      "loss": 2.0889,
      "step": 209558
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.3293429613113403,
      "learning_rate": 1.2035149818567358e-05,
      "loss": 2.121,
      "step": 209559
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.378849744796753,
      "learning_rate": 1.203400284597672e-05,
      "loss": 1.9082,
      "step": 209560
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.6107226610183716,
      "learning_rate": 1.2032855926924422e-05,
      "loss": 1.9747,
      "step": 209561
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.7469737529754639,
      "learning_rate": 1.2031709061410698e-05,
      "loss": 1.9061,
      "step": 209562
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.4155597686767578,
      "learning_rate": 1.2030562249435683e-05,
      "loss": 1.9093,
      "step": 209563
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.3892062902450562,
      "learning_rate": 1.2029415490999605e-05,
      "loss": 2.0887,
      "step": 209564
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.4957433938980103,
      "learning_rate": 1.2028268786102702e-05,
      "loss": 1.8282,
      "step": 209565
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.440524935722351,
      "learning_rate": 1.202712213474517e-05,
      "loss": 1.9369,
      "step": 209566
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.4738023281097412,
      "learning_rate": 1.2025975536927246e-05,
      "loss": 2.0845,
      "step": 209567
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.5511527061462402,
      "learning_rate": 1.202482899264916e-05,
      "loss": 1.8617,
      "step": 209568
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.5038620233535767,
      "learning_rate": 1.2023682501911081e-05,
      "loss": 2.1022,
      "step": 209569
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.5885064601898193,
      "learning_rate": 1.2022536064713206e-05,
      "loss": 1.9007,
      "step": 209570
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.4694255590438843,
      "learning_rate": 1.2021389681055804e-05,
      "loss": 1.8591,
      "step": 209571
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.4864990711212158,
      "learning_rate": 1.2020243350939008e-05,
      "loss": 1.9266,
      "step": 209572
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.5039753913879395,
      "learning_rate": 1.2019097074363148e-05,
      "loss": 1.9084,
      "step": 209573
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.4039698839187622,
      "learning_rate": 1.2017950851328362e-05,
      "loss": 2.0904,
      "step": 209574
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.5494567155838013,
      "learning_rate": 1.2016804681834847e-05,
      "loss": 2.0679,
      "step": 209575
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.5511093139648438,
      "learning_rate": 1.2015658565882869e-05,
      "loss": 2.0752,
      "step": 209576
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.5052672624588013,
      "learning_rate": 1.2014512503472595e-05,
      "loss": 2.1227,
      "step": 209577
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.4178781509399414,
      "learning_rate": 1.201336649460426e-05,
      "loss": 1.9314,
      "step": 209578
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.5273100137710571,
      "learning_rate": 1.2012220539278094e-05,
      "loss": 2.0631,
      "step": 209579
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.4108260869979858,
      "learning_rate": 1.20110746374943e-05,
      "loss": 1.8975,
      "step": 209580
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.3876113891601562,
      "learning_rate": 1.2009928789253042e-05,
      "loss": 1.9757,
      "step": 209581
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.6318392753601074,
      "learning_rate": 1.2008782994554623e-05,
      "loss": 2.1545,
      "step": 209582
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.4241806268692017,
      "learning_rate": 1.2007637253399204e-05,
      "loss": 1.945,
      "step": 209583
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.5255935192108154,
      "learning_rate": 1.2006491565786958e-05,
      "loss": 1.9458,
      "step": 209584
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.4608713388442993,
      "learning_rate": 1.200534593171818e-05,
      "loss": 2.1341,
      "step": 209585
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.6536985635757446,
      "learning_rate": 1.2004200351193005e-05,
      "loss": 2.1108,
      "step": 209586
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.4788105487823486,
      "learning_rate": 1.2003054824211733e-05,
      "loss": 2.0619,
      "step": 209587
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.6152921915054321,
      "learning_rate": 1.200190935077453e-05,
      "loss": 1.9428,
      "step": 209588
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.4400570392608643,
      "learning_rate": 1.2000763930881563e-05,
      "loss": 1.9989,
      "step": 209589
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.4074681997299194,
      "learning_rate": 1.1999618564533131e-05,
      "loss": 1.8775,
      "step": 209590
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.490565538406372,
      "learning_rate": 1.1998473251729401e-05,
      "loss": 2.2239,
      "step": 209591
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.444340467453003,
      "learning_rate": 1.1997327992470573e-05,
      "loss": 2.1295,
      "step": 209592
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.4449776411056519,
      "learning_rate": 1.1996182786756881e-05,
      "loss": 2.122,
      "step": 209593
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.4792765378952026,
      "learning_rate": 1.1995037634588555e-05,
      "loss": 2.1755,
      "step": 209594
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.3954479694366455,
      "learning_rate": 1.1993892535965766e-05,
      "loss": 1.8302,
      "step": 209595
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.4589335918426514,
      "learning_rate": 1.1992747490888776e-05,
      "loss": 2.0183,
      "step": 209596
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.5839849710464478,
      "learning_rate": 1.1991602499357755e-05,
      "loss": 1.9413,
      "step": 209597
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.4388799667358398,
      "learning_rate": 1.19904575613729e-05,
      "loss": 1.9717,
      "step": 209598
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.4501652717590332,
      "learning_rate": 1.1989312676934481e-05,
      "loss": 1.8809,
      "step": 209599
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.5566095113754272,
      "learning_rate": 1.1988167846042728e-05,
      "loss": 2.1072,
      "step": 209600
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.4656546115875244,
      "learning_rate": 1.1987023068697743e-05,
      "loss": 1.9726,
      "step": 209601
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.5873689651489258,
      "learning_rate": 1.1985878344899824e-05,
      "loss": 2.2032,
      "step": 209602
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.3736110925674438,
      "learning_rate": 1.1984733674649172e-05,
      "loss": 1.7657,
      "step": 209603
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.3786256313323975,
      "learning_rate": 1.1983589057946018e-05,
      "loss": 1.9442,
      "step": 209604
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.4071218967437744,
      "learning_rate": 1.1982444494790533e-05,
      "loss": 2.0035,
      "step": 209605
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.4079433679580688,
      "learning_rate": 1.198129998518298e-05,
      "loss": 1.9177,
      "step": 209606
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.4871739149093628,
      "learning_rate": 1.1980155529123492e-05,
      "loss": 1.8311,
      "step": 209607
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.4683034420013428,
      "learning_rate": 1.1979011126612337e-05,
      "loss": 2.0442,
      "step": 209608
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.4080519676208496,
      "learning_rate": 1.1977866777649713e-05,
      "loss": 1.9117,
      "step": 209609
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.67625892162323,
      "learning_rate": 1.197672248223589e-05,
      "loss": 1.8499,
      "step": 209610
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.3722728490829468,
      "learning_rate": 1.1975578240370997e-05,
      "loss": 1.8853,
      "step": 209611
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.593920350074768,
      "learning_rate": 1.1974434052055304e-05,
      "loss": 2.1074,
      "step": 209612
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.4212199449539185,
      "learning_rate": 1.1973289917288942e-05,
      "loss": 2.0405,
      "step": 209613
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.4340776205062866,
      "learning_rate": 1.1972145836072244e-05,
      "loss": 1.8822,
      "step": 209614
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.4453120231628418,
      "learning_rate": 1.1971001808405311e-05,
      "loss": 1.9006,
      "step": 209615
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.4385223388671875,
      "learning_rate": 1.1969857834288443e-05,
      "loss": 1.8716,
      "step": 209616
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.4316985607147217,
      "learning_rate": 1.1968713913721838e-05,
      "loss": 1.8698,
      "step": 209617
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.485970377922058,
      "learning_rate": 1.1967570046705666e-05,
      "loss": 1.9961,
      "step": 209618
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.4236114025115967,
      "learning_rate": 1.1966426233240123e-05,
      "loss": 1.7958,
      "step": 209619
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.4962332248687744,
      "learning_rate": 1.1965282473325477e-05,
      "loss": 2.0487,
      "step": 209620
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.5929491519927979,
      "learning_rate": 1.1964138766961895e-05,
      "loss": 1.9191,
      "step": 209621
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.5483988523483276,
      "learning_rate": 1.1962995114149675e-05,
      "loss": 1.7425,
      "step": 209622
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.6119661331176758,
      "learning_rate": 1.1961851514888954e-05,
      "loss": 2.0591,
      "step": 209623
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.522762417793274,
      "learning_rate": 1.1960707969179962e-05,
      "loss": 2.0141,
      "step": 209624
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.407367467880249,
      "learning_rate": 1.1959564477022897e-05,
      "loss": 2.0067,
      "step": 209625
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.619725227355957,
      "learning_rate": 1.1958421038417997e-05,
      "loss": 1.9245,
      "step": 209626
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.374187707901001,
      "learning_rate": 1.1957277653365427e-05,
      "loss": 1.8875,
      "step": 209627
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.3770474195480347,
      "learning_rate": 1.1956134321865485e-05,
      "loss": 1.8131,
      "step": 209628
    },
    {
      "epoch": 2.73,
      "grad_norm": 2.4511258602142334,
      "learning_rate": 1.1954991043918338e-05,
      "loss": 1.945,
      "step": 209629
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.3679701089859009,
      "learning_rate": 1.1953847819524188e-05,
      "loss": 1.8496,
      "step": 209630
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.341807246208191,
      "learning_rate": 1.1952704648683231e-05,
      "loss": 1.9362,
      "step": 209631
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.4125523567199707,
      "learning_rate": 1.1951561531395704e-05,
      "loss": 2.1571,
      "step": 209632
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.4816890954971313,
      "learning_rate": 1.1950418467661804e-05,
      "loss": 2.1083,
      "step": 209633
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.4990121126174927,
      "learning_rate": 1.19492754574818e-05,
      "loss": 1.9368,
      "step": 209634
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.4024195671081543,
      "learning_rate": 1.1948132500855856e-05,
      "loss": 2.0259,
      "step": 209635
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.435044765472412,
      "learning_rate": 1.1946989597784206e-05,
      "loss": 2.1088,
      "step": 209636
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.488925576210022,
      "learning_rate": 1.1945846748267018e-05,
      "loss": 1.7172,
      "step": 209637
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.4568487405776978,
      "learning_rate": 1.1944703952304558e-05,
      "loss": 1.7431,
      "step": 209638
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.5983372926712036,
      "learning_rate": 1.1943561209896957e-05,
      "loss": 1.9041,
      "step": 209639
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.3638073205947876,
      "learning_rate": 1.194241852104455e-05,
      "loss": 1.8163,
      "step": 209640
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.346308708190918,
      "learning_rate": 1.194127588574747e-05,
      "loss": 1.8527,
      "step": 209641
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.5272185802459717,
      "learning_rate": 1.194013330400595e-05,
      "loss": 2.0556,
      "step": 209642
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.5318681001663208,
      "learning_rate": 1.1938990775820156e-05,
      "loss": 2.1741,
      "step": 209643
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.5423052310943604,
      "learning_rate": 1.1937848301190389e-05,
      "loss": 2.0409,
      "step": 209644
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.4279714822769165,
      "learning_rate": 1.1936705880116749e-05,
      "loss": 2.0281,
      "step": 209645
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.517676830291748,
      "learning_rate": 1.1935563512599566e-05,
      "loss": 1.9753,
      "step": 209646
    },
    {
      "epoch": 2.73,
      "grad_norm": 2.4650065898895264,
      "learning_rate": 1.193442119863901e-05,
      "loss": 1.9326,
      "step": 209647
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.4456709623336792,
      "learning_rate": 1.193327893823528e-05,
      "loss": 2.0167,
      "step": 209648
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.84835946559906,
      "learning_rate": 1.1932136731388542e-05,
      "loss": 2.0345,
      "step": 209649
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.4121085405349731,
      "learning_rate": 1.193099457809913e-05,
      "loss": 2.0636,
      "step": 209650
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.397354006767273,
      "learning_rate": 1.1929852478367108e-05,
      "loss": 1.9956,
      "step": 209651
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.668714165687561,
      "learning_rate": 1.1928710432192812e-05,
      "loss": 2.0255,
      "step": 209652
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.317645788192749,
      "learning_rate": 1.1927568439576374e-05,
      "loss": 2.036,
      "step": 209653
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.5278449058532715,
      "learning_rate": 1.1926426500518093e-05,
      "loss": 2.0394,
      "step": 209654
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.3382298946380615,
      "learning_rate": 1.1925284615018072e-05,
      "loss": 1.843,
      "step": 209655
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.3783795833587646,
      "learning_rate": 1.1924142783076606e-05,
      "loss": 1.9674,
      "step": 209656
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.4563406705856323,
      "learning_rate": 1.1923001004693866e-05,
      "loss": 1.9914,
      "step": 209657
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.586310863494873,
      "learning_rate": 1.1921859279870084e-05,
      "loss": 1.8388,
      "step": 209658
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.377303123474121,
      "learning_rate": 1.1920717608605457e-05,
      "loss": 1.9253,
      "step": 209659
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.5134981870651245,
      "learning_rate": 1.1919575990900253e-05,
      "loss": 1.9169,
      "step": 209660
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.4490435123443604,
      "learning_rate": 1.1918434426754608e-05,
      "loss": 1.9188,
      "step": 209661
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.5383023023605347,
      "learning_rate": 1.1917292916168752e-05,
      "loss": 1.8168,
      "step": 209662
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.4815382957458496,
      "learning_rate": 1.1916151459142887e-05,
      "loss": 1.8861,
      "step": 209663
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.3899084329605103,
      "learning_rate": 1.191501005567731e-05,
      "loss": 1.8999,
      "step": 209664
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.3343995809555054,
      "learning_rate": 1.1913868705772089e-05,
      "loss": 2.0816,
      "step": 209665
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.4824106693267822,
      "learning_rate": 1.191272740942759e-05,
      "loss": 2.1108,
      "step": 209666
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.4097892045974731,
      "learning_rate": 1.1911586166643949e-05,
      "loss": 2.0825,
      "step": 209667
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.377281904220581,
      "learning_rate": 1.1910444977421363e-05,
      "loss": 1.9457,
      "step": 209668
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.4919496774673462,
      "learning_rate": 1.1909303841760031e-05,
      "loss": 1.9632,
      "step": 209669
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.5736013650894165,
      "learning_rate": 1.1908162759660255e-05,
      "loss": 2.1665,
      "step": 209670
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.4005061388015747,
      "learning_rate": 1.1907021731122135e-05,
      "loss": 1.9068,
      "step": 209671
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.5225037336349487,
      "learning_rate": 1.1905880756145969e-05,
      "loss": 2.0505,
      "step": 209672
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.4143975973129272,
      "learning_rate": 1.1904739834731958e-05,
      "loss": 1.8655,
      "step": 209673
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.5915108919143677,
      "learning_rate": 1.1903598966880234e-05,
      "loss": 1.7777,
      "step": 209674
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.4091333150863647,
      "learning_rate": 1.1902458152591132e-05,
      "loss": 1.8054,
      "step": 209675
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.4649858474731445,
      "learning_rate": 1.1901317391864784e-05,
      "loss": 2.1054,
      "step": 209676
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.3851269483566284,
      "learning_rate": 1.190017668470139e-05,
      "loss": 2.0644,
      "step": 209677
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.4443016052246094,
      "learning_rate": 1.1899036031101216e-05,
      "loss": 2.1599,
      "step": 209678
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.5379315614700317,
      "learning_rate": 1.1897895431064464e-05,
      "loss": 1.9464,
      "step": 209679
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.5372952222824097,
      "learning_rate": 1.1896754884591298e-05,
      "loss": 2.0584,
      "step": 209680
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.5026215314865112,
      "learning_rate": 1.1895614391681984e-05,
      "loss": 1.8991,
      "step": 209681
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.3515675067901611,
      "learning_rate": 1.1894473952336724e-05,
      "loss": 1.7761,
      "step": 209682
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.5143227577209473,
      "learning_rate": 1.1893333566555685e-05,
      "loss": 2.0702,
      "step": 209683
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.5123651027679443,
      "learning_rate": 1.1892193234339131e-05,
      "loss": 2.0761,
      "step": 209684
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.3962957859039307,
      "learning_rate": 1.1891052955687297e-05,
      "loss": 2.1037,
      "step": 209685
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.5004609823226929,
      "learning_rate": 1.1889912730600282e-05,
      "loss": 1.8895,
      "step": 209686
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.4010257720947266,
      "learning_rate": 1.1888772559078419e-05,
      "loss": 1.8374,
      "step": 209687
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.564608097076416,
      "learning_rate": 1.1887632441121842e-05,
      "loss": 1.9976,
      "step": 209688
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.4370653629302979,
      "learning_rate": 1.1886492376730816e-05,
      "loss": 2.0004,
      "step": 209689
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.460510492324829,
      "learning_rate": 1.1885352365905543e-05,
      "loss": 1.9432,
      "step": 209690
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.6716597080230713,
      "learning_rate": 1.1884212408646187e-05,
      "loss": 1.9324,
      "step": 209691
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.3940294981002808,
      "learning_rate": 1.1883072504952984e-05,
      "loss": 1.8958,
      "step": 209692
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.5042784214019775,
      "learning_rate": 1.1881932654826199e-05,
      "loss": 2.0566,
      "step": 209693
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.4738218784332275,
      "learning_rate": 1.1880792858265964e-05,
      "loss": 2.0748,
      "step": 209694
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.3345263004302979,
      "learning_rate": 1.1879653115272547e-05,
      "loss": 1.9923,
      "step": 209695
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.4008933305740356,
      "learning_rate": 1.187851342584618e-05,
      "loss": 1.7756,
      "step": 209696
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.3856863975524902,
      "learning_rate": 1.1877373789987e-05,
      "loss": 1.9778,
      "step": 209697
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.5328152179718018,
      "learning_rate": 1.1876234207695234e-05,
      "loss": 1.9493,
      "step": 209698
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.4212745428085327,
      "learning_rate": 1.1875094678971152e-05,
      "loss": 2.0423,
      "step": 209699
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.5147053003311157,
      "learning_rate": 1.187395520381489e-05,
      "loss": 1.7997,
      "step": 209700
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.5293426513671875,
      "learning_rate": 1.1872815782226707e-05,
      "loss": 1.9218,
      "step": 209701
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.492638111114502,
      "learning_rate": 1.1871676414206842e-05,
      "loss": 2.0709,
      "step": 209702
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.435019850730896,
      "learning_rate": 1.1870537099755462e-05,
      "loss": 1.9261,
      "step": 209703
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.5959701538085938,
      "learning_rate": 1.1869397838872763e-05,
      "loss": 1.7769,
      "step": 209704
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.484932541847229,
      "learning_rate": 1.186825863155898e-05,
      "loss": 1.8703,
      "step": 209705
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.4527513980865479,
      "learning_rate": 1.1867119477814313e-05,
      "loss": 1.9781,
      "step": 209706
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.3929097652435303,
      "learning_rate": 1.1865980377639028e-05,
      "loss": 1.9942,
      "step": 209707
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.5713342428207397,
      "learning_rate": 1.1864841331033292e-05,
      "loss": 1.9933,
      "step": 209708
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.4706059694290161,
      "learning_rate": 1.1863702337997339e-05,
      "loss": 2.1488,
      "step": 209709
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.5976141691207886,
      "learning_rate": 1.18625633985313e-05,
      "loss": 1.9267,
      "step": 209710
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.4453136920928955,
      "learning_rate": 1.1861424512635509e-05,
      "loss": 1.7765,
      "step": 209711
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.4990850687026978,
      "learning_rate": 1.1860285680310066e-05,
      "loss": 2.0614,
      "step": 209712
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.5127323865890503,
      "learning_rate": 1.1859146901555273e-05,
      "loss": 2.0677,
      "step": 209713
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.435115098953247,
      "learning_rate": 1.1858008176371292e-05,
      "loss": 1.9321,
      "step": 209714
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.4408605098724365,
      "learning_rate": 1.185686950475836e-05,
      "loss": 2.2311,
      "step": 209715
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.4942171573638916,
      "learning_rate": 1.1855730886716642e-05,
      "loss": 2.1556,
      "step": 209716
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.3783360719680786,
      "learning_rate": 1.1854592322246437e-05,
      "loss": 2.0572,
      "step": 209717
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.4690163135528564,
      "learning_rate": 1.1853453811347812e-05,
      "loss": 1.9858,
      "step": 209718
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.4578158855438232,
      "learning_rate": 1.1852315354021169e-05,
      "loss": 2.0415,
      "step": 209719
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.3990598917007446,
      "learning_rate": 1.1851176950266538e-05,
      "loss": 2.1267,
      "step": 209720
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.4027897119522095,
      "learning_rate": 1.1850038600084289e-05,
      "loss": 2.0164,
      "step": 209721
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.4828237295150757,
      "learning_rate": 1.1848900303474518e-05,
      "loss": 1.9103,
      "step": 209722
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.6230432987213135,
      "learning_rate": 1.1847762060437493e-05,
      "loss": 1.9772,
      "step": 209723
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.4322704076766968,
      "learning_rate": 1.1846623870973348e-05,
      "loss": 2.124,
      "step": 209724
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.5389018058776855,
      "learning_rate": 1.1845485735082416e-05,
      "loss": 1.8733,
      "step": 209725
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.5105621814727783,
      "learning_rate": 1.1844347652764797e-05,
      "loss": 2.1016,
      "step": 209726
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.5632578134536743,
      "learning_rate": 1.1843209624020855e-05,
      "loss": 2.0508,
      "step": 209727
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.3374028205871582,
      "learning_rate": 1.1842071648850592e-05,
      "loss": 2.0658,
      "step": 209728
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.8197556734085083,
      "learning_rate": 1.1840933727254376e-05,
      "loss": 1.9331,
      "step": 209729
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.705985188484192,
      "learning_rate": 1.1839795859232337e-05,
      "loss": 1.8458,
      "step": 209730
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.3863095045089722,
      "learning_rate": 1.1838658044784777e-05,
      "loss": 1.8414,
      "step": 209731
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.3245844841003418,
      "learning_rate": 1.183752028391176e-05,
      "loss": 2.0338,
      "step": 209732
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.4710792303085327,
      "learning_rate": 1.1836382576613723e-05,
      "loss": 1.9891,
      "step": 209733
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.43726646900177,
      "learning_rate": 1.183524492289063e-05,
      "loss": 1.8845,
      "step": 209734
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.408076286315918,
      "learning_rate": 1.1834107322742848e-05,
      "loss": 2.1676,
      "step": 209735
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.405638337135315,
      "learning_rate": 1.1832969776170508e-05,
      "loss": 1.8754,
      "step": 209736
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.4591422080993652,
      "learning_rate": 1.1831832283173882e-05,
      "loss": 2.2963,
      "step": 209737
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.486207365989685,
      "learning_rate": 1.1830694843753131e-05,
      "loss": 1.7269,
      "step": 209738
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.5496248006820679,
      "learning_rate": 1.182955745790859e-05,
      "loss": 1.9776,
      "step": 209739
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.5197933912277222,
      "learning_rate": 1.1828420125640292e-05,
      "loss": 2.1364,
      "step": 209740
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.3680191040039062,
      "learning_rate": 1.1827282846948539e-05,
      "loss": 1.9402,
      "step": 209741
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.5566681623458862,
      "learning_rate": 1.1826145621833527e-05,
      "loss": 2.1738,
      "step": 209742
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.5763500928878784,
      "learning_rate": 1.1825008450295493e-05,
      "loss": 2.1755,
      "step": 209743
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.5060771703720093,
      "learning_rate": 1.18238713323346e-05,
      "loss": 2.0202,
      "step": 209744
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.579085111618042,
      "learning_rate": 1.182273426795115e-05,
      "loss": 2.123,
      "step": 209745
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.4398748874664307,
      "learning_rate": 1.1821597257145243e-05,
      "loss": 1.9459,
      "step": 209746
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.5341873168945312,
      "learning_rate": 1.1820460299917179e-05,
      "loss": 1.8681,
      "step": 209747
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.5544824600219727,
      "learning_rate": 1.1819323396267054e-05,
      "loss": 1.8593,
      "step": 209748
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.3847992420196533,
      "learning_rate": 1.181818654619524e-05,
      "loss": 2.0232,
      "step": 209749
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.4030097723007202,
      "learning_rate": 1.1817049749701802e-05,
      "loss": 1.9742,
      "step": 209750
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.3314560651779175,
      "learning_rate": 1.181591300678707e-05,
      "loss": 1.9724,
      "step": 209751
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.499926209449768,
      "learning_rate": 1.181477631745118e-05,
      "loss": 1.9239,
      "step": 209752
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.3997927904129028,
      "learning_rate": 1.1813639681694364e-05,
      "loss": 2.073,
      "step": 209753
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.5544813871383667,
      "learning_rate": 1.1812503099516824e-05,
      "loss": 2.0535,
      "step": 209754
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.3557876348495483,
      "learning_rate": 1.1811366570918789e-05,
      "loss": 2.0323,
      "step": 209755
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.425826072692871,
      "learning_rate": 1.181023009590043e-05,
      "loss": 1.9124,
      "step": 209756
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.3358497619628906,
      "learning_rate": 1.1809093674462011e-05,
      "loss": 1.9245,
      "step": 209757
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.3688634634017944,
      "learning_rate": 1.1807957306603733e-05,
      "loss": 1.8924,
      "step": 209758
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.4258854389190674,
      "learning_rate": 1.1806820992325827e-05,
      "loss": 1.8433,
      "step": 209759
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.461005687713623,
      "learning_rate": 1.1805684731628395e-05,
      "loss": 1.9265,
      "step": 209760
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.3631789684295654,
      "learning_rate": 1.180454852451177e-05,
      "loss": 2.0502,
      "step": 209761
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.6811853647232056,
      "learning_rate": 1.1803412370976084e-05,
      "loss": 1.9329,
      "step": 209762
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.4080662727355957,
      "learning_rate": 1.1802276271021637e-05,
      "loss": 2.0665,
      "step": 209763
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.4721633195877075,
      "learning_rate": 1.1801140224648564e-05,
      "loss": 2.0589,
      "step": 209764
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.5604816675186157,
      "learning_rate": 1.1800004231857097e-05,
      "loss": 2.1313,
      "step": 209765
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.3285777568817139,
      "learning_rate": 1.1798868292647468e-05,
      "loss": 1.973,
      "step": 209766
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.4982068538665771,
      "learning_rate": 1.1797732407019844e-05,
      "loss": 2.0674,
      "step": 209767
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.5663115978240967,
      "learning_rate": 1.179659657497446e-05,
      "loss": 1.859,
      "step": 209768
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.6318929195404053,
      "learning_rate": 1.1795460796511547e-05,
      "loss": 1.9303,
      "step": 209769
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.3443940877914429,
      "learning_rate": 1.1794325071631305e-05,
      "loss": 1.9099,
      "step": 209770
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.5638306140899658,
      "learning_rate": 1.1793189400333903e-05,
      "loss": 2.0386,
      "step": 209771
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.3938125371932983,
      "learning_rate": 1.1792053782619603e-05,
      "loss": 1.9182,
      "step": 209772
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.40900719165802,
      "learning_rate": 1.179091821848861e-05,
      "loss": 2.0682,
      "step": 209773
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.3493564128875732,
      "learning_rate": 1.178978270794112e-05,
      "loss": 2.0548,
      "step": 209774
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.3611119985580444,
      "learning_rate": 1.1788647250977368e-05,
      "loss": 2.0153,
      "step": 209775
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.4902726411819458,
      "learning_rate": 1.1787511847597553e-05,
      "loss": 2.0187,
      "step": 209776
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.387096643447876,
      "learning_rate": 1.1786376497801807e-05,
      "loss": 1.9485,
      "step": 209777
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.4642466306686401,
      "learning_rate": 1.1785241201590502e-05,
      "loss": 2.1755,
      "step": 209778
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.4582202434539795,
      "learning_rate": 1.1784105958963696e-05,
      "loss": 1.8782,
      "step": 209779
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.5909721851348877,
      "learning_rate": 1.1782970769921695e-05,
      "loss": 2.0105,
      "step": 209780
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.4585307836532593,
      "learning_rate": 1.1781835634464698e-05,
      "loss": 1.7955,
      "step": 209781
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.612935185432434,
      "learning_rate": 1.178070055259287e-05,
      "loss": 1.9213,
      "step": 209782
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.4056310653686523,
      "learning_rate": 1.1779565524306444e-05,
      "loss": 1.8597,
      "step": 209783
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.353679895401001,
      "learning_rate": 1.1778430549605655e-05,
      "loss": 2.0109,
      "step": 209784
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.5546635389328003,
      "learning_rate": 1.1777295628490669e-05,
      "loss": 2.062,
      "step": 209785
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.4875311851501465,
      "learning_rate": 1.1776160760961784e-05,
      "loss": 1.9891,
      "step": 209786
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.4203298091888428,
      "learning_rate": 1.1775025947019068e-05,
      "loss": 2.1584,
      "step": 209787
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.4041666984558105,
      "learning_rate": 1.177389118666292e-05,
      "loss": 1.9363,
      "step": 209788
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.463979721069336,
      "learning_rate": 1.1772756479893375e-05,
      "loss": 2.2416,
      "step": 209789
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.3859503269195557,
      "learning_rate": 1.1771621826710731e-05,
      "loss": 1.9788,
      "step": 209790
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.364941120147705,
      "learning_rate": 1.1770487227115156e-05,
      "loss": 1.9159,
      "step": 209791
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.8719671964645386,
      "learning_rate": 1.1769352681106946e-05,
      "loss": 1.9863,
      "step": 209792
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.3727909326553345,
      "learning_rate": 1.1768218188686174e-05,
      "loss": 2.0204,
      "step": 209793
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.533153772354126,
      "learning_rate": 1.1767083749853235e-05,
      "loss": 1.9042,
      "step": 209794
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.6517707109451294,
      "learning_rate": 1.1765949364608162e-05,
      "loss": 1.7037,
      "step": 209795
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.4520963430404663,
      "learning_rate": 1.1764815032951258e-05,
      "loss": 1.9588,
      "step": 209796
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.462444543838501,
      "learning_rate": 1.1763680754882721e-05,
      "loss": 1.9993,
      "step": 209797
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.5735650062561035,
      "learning_rate": 1.176254653040275e-05,
      "loss": 2.0986,
      "step": 209798
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.4404206275939941,
      "learning_rate": 1.1761412359511545e-05,
      "loss": 2.0913,
      "step": 209799
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.5512986183166504,
      "learning_rate": 1.1760278242209408e-05,
      "loss": 1.772,
      "step": 209800
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.6051167249679565,
      "learning_rate": 1.1759144178496404e-05,
      "loss": 1.8149,
      "step": 209801
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.5717226266860962,
      "learning_rate": 1.1758010168372834e-05,
      "loss": 1.9369,
      "step": 209802
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.5180432796478271,
      "learning_rate": 1.1756876211838861e-05,
      "loss": 2.027,
      "step": 209803
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.3722587823867798,
      "learning_rate": 1.1755742308894755e-05,
      "loss": 1.8848,
      "step": 209804
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.3891562223434448,
      "learning_rate": 1.1754608459540683e-05,
      "loss": 1.8573,
      "step": 209805
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.6326441764831543,
      "learning_rate": 1.1753474663776941e-05,
      "loss": 2.1803,
      "step": 209806
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.5448272228240967,
      "learning_rate": 1.1752340921603597e-05,
      "loss": 2.1145,
      "step": 209807
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.3929836750030518,
      "learning_rate": 1.1751207233020954e-05,
      "loss": 2.0702,
      "step": 209808
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.530895471572876,
      "learning_rate": 1.1750073598029175e-05,
      "loss": 1.9919,
      "step": 209809
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.556439995765686,
      "learning_rate": 1.1748940016628527e-05,
      "loss": 1.7723,
      "step": 209810
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.62276291847229,
      "learning_rate": 1.1747806488819178e-05,
      "loss": 2.0855,
      "step": 209811
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.5932945013046265,
      "learning_rate": 1.1746673014601393e-05,
      "loss": 2.1806,
      "step": 209812
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.4648191928863525,
      "learning_rate": 1.1745539593975306e-05,
      "loss": 1.8417,
      "step": 209813
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.58595609664917,
      "learning_rate": 1.1744406226941182e-05,
      "loss": 1.8749,
      "step": 209814
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.6039291620254517,
      "learning_rate": 1.174327291349919e-05,
      "loss": 1.8988,
      "step": 209815
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.457112431526184,
      "learning_rate": 1.1742139653649563e-05,
      "loss": 2.252,
      "step": 209816
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.4440057277679443,
      "learning_rate": 1.1741006447392531e-05,
      "loss": 1.8759,
      "step": 209817
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.5584253072738647,
      "learning_rate": 1.173987329472833e-05,
      "loss": 1.7385,
      "step": 209818
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.8122249841690063,
      "learning_rate": 1.173874019565706e-05,
      "loss": 2.1321,
      "step": 209819
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.4074064493179321,
      "learning_rate": 1.1737607150179018e-05,
      "loss": 1.8841,
      "step": 209820
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.6446412801742554,
      "learning_rate": 1.1736474158294373e-05,
      "loss": 1.8771,
      "step": 209821
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.599571943283081,
      "learning_rate": 1.173534122000339e-05,
      "loss": 2.164,
      "step": 209822
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.441801905632019,
      "learning_rate": 1.173420833530624e-05,
      "loss": 1.8836,
      "step": 209823
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.4114856719970703,
      "learning_rate": 1.1733075504203182e-05,
      "loss": 1.7398,
      "step": 209824
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.532766342163086,
      "learning_rate": 1.1731942726694321e-05,
      "loss": 1.8961,
      "step": 209825
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.5552343130111694,
      "learning_rate": 1.173081000277999e-05,
      "loss": 2.0568,
      "step": 209826
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.4500021934509277,
      "learning_rate": 1.1729677332460285e-05,
      "loss": 1.8991,
      "step": 209827
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.4482018947601318,
      "learning_rate": 1.1728544715735544e-05,
      "loss": 1.8259,
      "step": 209828
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.7638226747512817,
      "learning_rate": 1.172741215260583e-05,
      "loss": 1.9957,
      "step": 209829
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.597454309463501,
      "learning_rate": 1.1726279643071512e-05,
      "loss": 2.1616,
      "step": 209830
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.509351372718811,
      "learning_rate": 1.1725147187132655e-05,
      "loss": 2.0962,
      "step": 209831
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.6424915790557861,
      "learning_rate": 1.1724014784789593e-05,
      "loss": 2.0259,
      "step": 209832
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.6523693799972534,
      "learning_rate": 1.1722882436042424e-05,
      "loss": 1.967,
      "step": 209833
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.450317621231079,
      "learning_rate": 1.1721750140891417e-05,
      "loss": 2.0373,
      "step": 209834
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.481679081916809,
      "learning_rate": 1.172061789933677e-05,
      "loss": 1.7934,
      "step": 209835
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.7202117443084717,
      "learning_rate": 1.1719485711378751e-05,
      "loss": 1.7885,
      "step": 209836
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.5219229459762573,
      "learning_rate": 1.1718353577017525e-05,
      "loss": 2.0231,
      "step": 209837
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.6489306688308716,
      "learning_rate": 1.1717221496253259e-05,
      "loss": 1.7117,
      "step": 209838
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.6085902452468872,
      "learning_rate": 1.171608946908622e-05,
      "loss": 2.0677,
      "step": 209839
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.576173186302185,
      "learning_rate": 1.1714957495516608e-05,
      "loss": 2.1108,
      "step": 209840
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.4526948928833008,
      "learning_rate": 1.171382557554459e-05,
      "loss": 1.8756,
      "step": 209841
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.4511871337890625,
      "learning_rate": 1.1712693709170429e-05,
      "loss": 2.0994,
      "step": 209842
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.385088324546814,
      "learning_rate": 1.1711561896394361e-05,
      "loss": 2.0667,
      "step": 209843
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.4742498397827148,
      "learning_rate": 1.171043013721652e-05,
      "loss": 1.9178,
      "step": 209844
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.52402663230896,
      "learning_rate": 1.1709298431637137e-05,
      "loss": 1.8821,
      "step": 209845
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.4672530889511108,
      "learning_rate": 1.170816677965648e-05,
      "loss": 2.0271,
      "step": 209846
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.7168737649917603,
      "learning_rate": 1.1707035181274682e-05,
      "loss": 1.9302,
      "step": 209847
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.5022591352462769,
      "learning_rate": 1.170590363649201e-05,
      "loss": 2.0581,
      "step": 209848
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.4250015020370483,
      "learning_rate": 1.170477214530866e-05,
      "loss": 1.957,
      "step": 209849
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.3980642557144165,
      "learning_rate": 1.1703640707724771e-05,
      "loss": 1.9991,
      "step": 209850
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.3781661987304688,
      "learning_rate": 1.1702509323740671e-05,
      "loss": 1.8916,
      "step": 209851
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.5099233388900757,
      "learning_rate": 1.1701377993356532e-05,
      "loss": 1.8661,
      "step": 209852
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.3545827865600586,
      "learning_rate": 1.1700246716572514e-05,
      "loss": 1.93,
      "step": 209853
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.624406099319458,
      "learning_rate": 1.169911549338889e-05,
      "loss": 2.0419,
      "step": 209854
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.4818809032440186,
      "learning_rate": 1.169798432380582e-05,
      "loss": 2.006,
      "step": 209855
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.519964575767517,
      "learning_rate": 1.1696853207823543e-05,
      "loss": 1.9615,
      "step": 209856
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.6074432134628296,
      "learning_rate": 1.1695722145442254e-05,
      "loss": 1.8181,
      "step": 209857
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.406245470046997,
      "learning_rate": 1.1694591136662158e-05,
      "loss": 1.9664,
      "step": 209858
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.336835265159607,
      "learning_rate": 1.1693460181483515e-05,
      "loss": 1.9745,
      "step": 209859
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.4320344924926758,
      "learning_rate": 1.1692329279906531e-05,
      "loss": 1.8804,
      "step": 209860
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.4183526039123535,
      "learning_rate": 1.1691198431931336e-05,
      "loss": 1.8713,
      "step": 209861
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.5629689693450928,
      "learning_rate": 1.1690067637558165e-05,
      "loss": 1.8798,
      "step": 209862
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.4895175695419312,
      "learning_rate": 1.1688936896787315e-05,
      "loss": 1.9434,
      "step": 209863
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.518359899520874,
      "learning_rate": 1.1687806209618889e-05,
      "loss": 1.9846,
      "step": 209864
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.5200390815734863,
      "learning_rate": 1.1686675576053151e-05,
      "loss": 2.0003,
      "step": 209865
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.5659103393554688,
      "learning_rate": 1.1685544996090335e-05,
      "loss": 1.9448,
      "step": 209866
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.3166733980178833,
      "learning_rate": 1.1684414469730608e-05,
      "loss": 1.9477,
      "step": 209867
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.4590719938278198,
      "learning_rate": 1.1683283996974136e-05,
      "loss": 2.0499,
      "step": 209868
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.5244866609573364,
      "learning_rate": 1.1682153577821251e-05,
      "loss": 1.8918,
      "step": 209869
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.5297454595565796,
      "learning_rate": 1.1681023212272023e-05,
      "loss": 2.1517,
      "step": 209870
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.485751986503601,
      "learning_rate": 1.1679892900326815e-05,
      "loss": 2.2728,
      "step": 209871
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.3954060077667236,
      "learning_rate": 1.1678762641985695e-05,
      "loss": 2.1035,
      "step": 209872
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.4898452758789062,
      "learning_rate": 1.1677632437248996e-05,
      "loss": 2.102,
      "step": 209873
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.5392091274261475,
      "learning_rate": 1.1676502286116818e-05,
      "loss": 1.9269,
      "step": 209874
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.6154251098632812,
      "learning_rate": 1.167537218858946e-05,
      "loss": 1.9838,
      "step": 209875
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.5289721488952637,
      "learning_rate": 1.1674242144667023e-05,
      "loss": 2.1459,
      "step": 209876
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.5212215185165405,
      "learning_rate": 1.167311215434984e-05,
      "loss": 2.1846,
      "step": 209877
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.4758038520812988,
      "learning_rate": 1.1671982217638043e-05,
      "loss": 2.0867,
      "step": 209878
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.3655264377593994,
      "learning_rate": 1.1670852334531933e-05,
      "loss": 1.9822,
      "step": 209879
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.5629407167434692,
      "learning_rate": 1.1669722505031575e-05,
      "loss": 2.2211,
      "step": 209880
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.5047907829284668,
      "learning_rate": 1.1668592729137305e-05,
      "loss": 2.0624,
      "step": 209881
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.5303210020065308,
      "learning_rate": 1.1667463006849254e-05,
      "loss": 1.8865,
      "step": 209882
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.392880916595459,
      "learning_rate": 1.1666333338167688e-05,
      "loss": 1.9625,
      "step": 209883
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.384628176689148,
      "learning_rate": 1.1665203723092775e-05,
      "loss": 1.8393,
      "step": 209884
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.5245805978775024,
      "learning_rate": 1.1664074161624814e-05,
      "loss": 1.947,
      "step": 209885
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.5979995727539062,
      "learning_rate": 1.1662944653763873e-05,
      "loss": 1.9256,
      "step": 209886
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.4719514846801758,
      "learning_rate": 1.1661815199510249e-05,
      "loss": 2.2011,
      "step": 209887
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.3639726638793945,
      "learning_rate": 1.166068579886411e-05,
      "loss": 2.001,
      "step": 209888
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.4193938970565796,
      "learning_rate": 1.1659556451825725e-05,
      "loss": 1.9117,
      "step": 209889
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.3341859579086304,
      "learning_rate": 1.1658427158395223e-05,
      "loss": 1.8063,
      "step": 209890
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.362815260887146,
      "learning_rate": 1.1657297918572972e-05,
      "loss": 2.1967,
      "step": 209891
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.6040323972702026,
      "learning_rate": 1.1656168732358973e-05,
      "loss": 1.9397,
      "step": 209892
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.3481099605560303,
      "learning_rate": 1.1655039599753556e-05,
      "loss": 1.7874,
      "step": 209893
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.4148229360580444,
      "learning_rate": 1.1653910520756926e-05,
      "loss": 1.9247,
      "step": 209894
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.6257210969924927,
      "learning_rate": 1.1652781495369278e-05,
      "loss": 2.0043,
      "step": 209895
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.3808729648590088,
      "learning_rate": 1.1651652523590781e-05,
      "loss": 2.0912,
      "step": 209896
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.3849915266036987,
      "learning_rate": 1.1650523605421769e-05,
      "loss": 2.0891,
      "step": 209897
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.4338319301605225,
      "learning_rate": 1.1649394740862273e-05,
      "loss": 2.0393,
      "step": 209898
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.4279217720031738,
      "learning_rate": 1.1648265929912659e-05,
      "loss": 1.8235,
      "step": 209899
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.4547102451324463,
      "learning_rate": 1.164713717257303e-05,
      "loss": 2.0378,
      "step": 209900
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.4836803674697876,
      "learning_rate": 1.1646008468843682e-05,
      "loss": 1.9632,
      "step": 209901
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.5178688764572144,
      "learning_rate": 1.164487981872475e-05,
      "loss": 1.8898,
      "step": 209902
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.4536738395690918,
      "learning_rate": 1.1643751222216535e-05,
      "loss": 1.7387,
      "step": 209903
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.5845892429351807,
      "learning_rate": 1.1642622679319136e-05,
      "loss": 1.7708,
      "step": 209904
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.4055863618850708,
      "learning_rate": 1.1641494190032852e-05,
      "loss": 1.9766,
      "step": 209905
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.4797500371932983,
      "learning_rate": 1.1640365754357816e-05,
      "loss": 1.9199,
      "step": 209906
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.4241498708724976,
      "learning_rate": 1.1639237372294297e-05,
      "loss": 1.7863,
      "step": 209907
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.5270684957504272,
      "learning_rate": 1.1638109043842458e-05,
      "loss": 1.8586,
      "step": 209908
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.5854383707046509,
      "learning_rate": 1.1636980769002635e-05,
      "loss": 1.9258,
      "step": 209909
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.4621573686599731,
      "learning_rate": 1.163585254777486e-05,
      "loss": 2.1471,
      "step": 209910
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.452219009399414,
      "learning_rate": 1.1634724380159433e-05,
      "loss": 2.0671,
      "step": 209911
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.4754905700683594,
      "learning_rate": 1.1633596266156553e-05,
      "loss": 2.0848,
      "step": 209912
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.5273587703704834,
      "learning_rate": 1.1632468205766454e-05,
      "loss": 1.9137,
      "step": 209913
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.4165323972702026,
      "learning_rate": 1.163134019898927e-05,
      "loss": 1.8916,
      "step": 209914
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.554551362991333,
      "learning_rate": 1.1630212245825365e-05,
      "loss": 2.0823,
      "step": 209915
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.4762531518936157,
      "learning_rate": 1.1629084346274741e-05,
      "loss": 1.9095,
      "step": 209916
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.5372825860977173,
      "learning_rate": 1.1627956500337765e-05,
      "loss": 1.9742,
      "step": 209917
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.484810471534729,
      "learning_rate": 1.1626828708014568e-05,
      "loss": 1.8613,
      "step": 209918
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.3679801225662231,
      "learning_rate": 1.1625700969305419e-05,
      "loss": 1.9278,
      "step": 209919
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.5447882413864136,
      "learning_rate": 1.1624573284210448e-05,
      "loss": 1.8885,
      "step": 209920
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.4050378799438477,
      "learning_rate": 1.1623445652729956e-05,
      "loss": 1.8571,
      "step": 209921
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.6202641725540161,
      "learning_rate": 1.1622318074864112e-05,
      "loss": 1.9466,
      "step": 209922
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.7443475723266602,
      "learning_rate": 1.1621190550613113e-05,
      "loss": 1.9979,
      "step": 209923
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.4272388219833374,
      "learning_rate": 1.162006307997716e-05,
      "loss": 2.1355,
      "step": 209924
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.4438230991363525,
      "learning_rate": 1.1618935662956519e-05,
      "loss": 2.1462,
      "step": 209925
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.4084117412567139,
      "learning_rate": 1.161780829955129e-05,
      "loss": 1.9502,
      "step": 209926
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.673880696296692,
      "learning_rate": 1.161668098976184e-05,
      "loss": 2.306,
      "step": 209927
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.4403363466262817,
      "learning_rate": 1.1615553733588267e-05,
      "loss": 2.0485,
      "step": 209928
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.3560261726379395,
      "learning_rate": 1.1614426531030773e-05,
      "loss": 1.9342,
      "step": 209929
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.5666561126708984,
      "learning_rate": 1.1613299382089625e-05,
      "loss": 2.033,
      "step": 209930
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.4481539726257324,
      "learning_rate": 1.161217228676502e-05,
      "loss": 1.8669,
      "step": 209931
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.5006954669952393,
      "learning_rate": 1.1611045245057093e-05,
      "loss": 2.1958,
      "step": 209932
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.3693034648895264,
      "learning_rate": 1.1609918256966177e-05,
      "loss": 1.7992,
      "step": 209933
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.301257610321045,
      "learning_rate": 1.160879132249244e-05,
      "loss": 1.9376,
      "step": 209934
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.4595389366149902,
      "learning_rate": 1.1607664441636012e-05,
      "loss": 1.7462,
      "step": 209935
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.323150634765625,
      "learning_rate": 1.1606537614397194e-05,
      "loss": 2.0777,
      "step": 209936
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.482199788093567,
      "learning_rate": 1.1605410840776186e-05,
      "loss": 2.0546,
      "step": 209937
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.468357801437378,
      "learning_rate": 1.1604284120773122e-05,
      "loss": 2.0416,
      "step": 209938
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.4455270767211914,
      "learning_rate": 1.16031574543883e-05,
      "loss": 1.9462,
      "step": 209939
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.4231871366500854,
      "learning_rate": 1.160203084162189e-05,
      "loss": 1.9993,
      "step": 209940
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.461739182472229,
      "learning_rate": 1.1600904282474089e-05,
      "loss": 2.0538,
      "step": 209941
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.4662461280822754,
      "learning_rate": 1.1599777776945163e-05,
      "loss": 2.1395,
      "step": 209942
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.374370813369751,
      "learning_rate": 1.159865132503528e-05,
      "loss": 1.733,
      "step": 209943
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.4377012252807617,
      "learning_rate": 1.1597524926744605e-05,
      "loss": 1.7072,
      "step": 209944
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.4209671020507812,
      "learning_rate": 1.1596398582073441e-05,
      "loss": 1.9797,
      "step": 209945
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.4801188707351685,
      "learning_rate": 1.1595272291021917e-05,
      "loss": 1.9431,
      "step": 209946
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.4796639680862427,
      "learning_rate": 1.1594146053590269e-05,
      "loss": 2.002,
      "step": 209947
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.475485920906067,
      "learning_rate": 1.1593019869778763e-05,
      "loss": 1.8825,
      "step": 209948
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.3791295289993286,
      "learning_rate": 1.15918937395875e-05,
      "loss": 1.9597,
      "step": 209949
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.5765918493270874,
      "learning_rate": 1.1590767663016809e-05,
      "loss": 2.2468,
      "step": 209950
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.5113015174865723,
      "learning_rate": 1.1589641640066793e-05,
      "loss": 1.8581,
      "step": 209951
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.5574212074279785,
      "learning_rate": 1.1588515670737753e-05,
      "loss": 2.0292,
      "step": 209952
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.4377986192703247,
      "learning_rate": 1.1587389755029786e-05,
      "loss": 1.8815,
      "step": 209953
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.3818000555038452,
      "learning_rate": 1.1586263892943226e-05,
      "loss": 1.8863,
      "step": 209954
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.4932126998901367,
      "learning_rate": 1.1585138084478175e-05,
      "loss": 2.1175,
      "step": 209955
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.5288481712341309,
      "learning_rate": 1.1584012329634929e-05,
      "loss": 2.2899,
      "step": 209956
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.3932830095291138,
      "learning_rate": 1.1582886628413624e-05,
      "loss": 1.7233,
      "step": 209957
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.4298641681671143,
      "learning_rate": 1.1581760980814558e-05,
      "loss": 2.0791,
      "step": 209958
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.52975332736969,
      "learning_rate": 1.1580635386837832e-05,
      "loss": 2.1851,
      "step": 209959
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.429129719734192,
      "learning_rate": 1.1579509846483747e-05,
      "loss": 1.9398,
      "step": 209960
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.517188310623169,
      "learning_rate": 1.1578384359752435e-05,
      "loss": 1.9768,
      "step": 209961
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.4583245515823364,
      "learning_rate": 1.1577258926644195e-05,
      "loss": 1.9632,
      "step": 209962
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.4082509279251099,
      "learning_rate": 1.1576133547159128e-05,
      "loss": 1.7626,
      "step": 209963
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.5589131116867065,
      "learning_rate": 1.1575008221297598e-05,
      "loss": 2.1467,
      "step": 209964
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.5612436532974243,
      "learning_rate": 1.1573882949059643e-05,
      "loss": 1.9499,
      "step": 209965
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.3870999813079834,
      "learning_rate": 1.1572757730445558e-05,
      "loss": 1.8206,
      "step": 209966
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.5842969417572021,
      "learning_rate": 1.1571632565455547e-05,
      "loss": 2.0615,
      "step": 209967
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.5298118591308594,
      "learning_rate": 1.1570507454089806e-05,
      "loss": 1.9938,
      "step": 209968
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.5728307962417603,
      "learning_rate": 1.1569382396348536e-05,
      "loss": 2.088,
      "step": 209969
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.413912296295166,
      "learning_rate": 1.1568257392232038e-05,
      "loss": 2.1604,
      "step": 209970
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.3948798179626465,
      "learning_rate": 1.1567132441740378e-05,
      "loss": 1.7672,
      "step": 209971
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.3453491926193237,
      "learning_rate": 1.1566007544873857e-05,
      "loss": 1.9999,
      "step": 209972
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.4823603630065918,
      "learning_rate": 1.1564882701632638e-05,
      "loss": 1.998,
      "step": 209973
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.5499732494354248,
      "learning_rate": 1.1563757912016991e-05,
      "loss": 1.9123,
      "step": 209974
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.494023323059082,
      "learning_rate": 1.1562633176027015e-05,
      "loss": 1.9988,
      "step": 209975
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.4327572584152222,
      "learning_rate": 1.1561508493663107e-05,
      "loss": 1.8673,
      "step": 209976
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.5198668241500854,
      "learning_rate": 1.1560383864925272e-05,
      "loss": 2.0534,
      "step": 209977
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.5384212732315063,
      "learning_rate": 1.1559259289813838e-05,
      "loss": 2.1224,
      "step": 209978
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.4475466012954712,
      "learning_rate": 1.1558134768328941e-05,
      "loss": 2.1124,
      "step": 209979
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.406461477279663,
      "learning_rate": 1.1557010300470849e-05,
      "loss": 2.0556,
      "step": 209980
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.3664630651474,
      "learning_rate": 1.1555885886239757e-05,
      "loss": 1.9083,
      "step": 209981
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.4775558710098267,
      "learning_rate": 1.1554761525635902e-05,
      "loss": 2.0531,
      "step": 209982
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.4640730619430542,
      "learning_rate": 1.1553637218659418e-05,
      "loss": 1.9795,
      "step": 209983
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.512906789779663,
      "learning_rate": 1.1552512965310601e-05,
      "loss": 1.9028,
      "step": 209984
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.388558030128479,
      "learning_rate": 1.1551388765589553e-05,
      "loss": 2.0529,
      "step": 209985
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.3558497428894043,
      "learning_rate": 1.1550264619496608e-05,
      "loss": 1.853,
      "step": 209986
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.3143056631088257,
      "learning_rate": 1.1549140527031864e-05,
      "loss": 1.7968,
      "step": 209987
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.3906145095825195,
      "learning_rate": 1.1548016488195622e-05,
      "loss": 1.973,
      "step": 209988
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.4229686260223389,
      "learning_rate": 1.1546892502988048e-05,
      "loss": 1.8784,
      "step": 209989
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.4128766059875488,
      "learning_rate": 1.1545768571409342e-05,
      "loss": 2.0535,
      "step": 209990
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.4946717023849487,
      "learning_rate": 1.1544644693459704e-05,
      "loss": 2.0188,
      "step": 209991
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.4269919395446777,
      "learning_rate": 1.15435208691394e-05,
      "loss": 1.9665,
      "step": 209992
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.5877894163131714,
      "learning_rate": 1.154239709844853e-05,
      "loss": 2.1052,
      "step": 209993
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.6229966878890991,
      "learning_rate": 1.1541273381387428e-05,
      "loss": 2.0859,
      "step": 209994
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.4386374950408936,
      "learning_rate": 1.154014971795626e-05,
      "loss": 1.9982,
      "step": 209995
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.3375167846679688,
      "learning_rate": 1.1539026108155225e-05,
      "loss": 2.0856,
      "step": 209996
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.4417781829833984,
      "learning_rate": 1.1537902551984491e-05,
      "loss": 1.9793,
      "step": 209997
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.5724608898162842,
      "learning_rate": 1.1536779049444322e-05,
      "loss": 2.0113,
      "step": 209998
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.4749414920806885,
      "learning_rate": 1.1535655600534888e-05,
      "loss": 1.845,
      "step": 209999
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.5465837717056274,
      "learning_rate": 1.1534532205256453e-05,
      "loss": 1.87,
      "step": 210000
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.7889569997787476,
      "learning_rate": 1.1533408863609217e-05,
      "loss": 1.8932,
      "step": 210001
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.5692366361618042,
      "learning_rate": 1.1532285575593347e-05,
      "loss": 2.0712,
      "step": 210002
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.4575644731521606,
      "learning_rate": 1.153116234120901e-05,
      "loss": 2.091,
      "step": 210003
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.5114942789077759,
      "learning_rate": 1.153003916045654e-05,
      "loss": 1.818,
      "step": 210004
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.428923487663269,
      "learning_rate": 1.1528916033336033e-05,
      "loss": 2.1021,
      "step": 210005
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.3848340511322021,
      "learning_rate": 1.1527792959847793e-05,
      "loss": 1.9234,
      "step": 210006
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.4173139333724976,
      "learning_rate": 1.1526669939991984e-05,
      "loss": 2.2385,
      "step": 210007
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.7514023780822754,
      "learning_rate": 1.1525546973768806e-05,
      "loss": 1.8401,
      "step": 210008
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.5982227325439453,
      "learning_rate": 1.1524424061178428e-05,
      "loss": 2.0064,
      "step": 210009
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.380059003829956,
      "learning_rate": 1.1523301202221147e-05,
      "loss": 2.163,
      "step": 210010
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.385250210762024,
      "learning_rate": 1.1522178396897097e-05,
      "loss": 1.9157,
      "step": 210011
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.351839303970337,
      "learning_rate": 1.1521055645206579e-05,
      "loss": 2.0761,
      "step": 210012
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.379975438117981,
      "learning_rate": 1.1519932947149723e-05,
      "loss": 2.0804,
      "step": 210013
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.3760379552841187,
      "learning_rate": 1.1518810302726732e-05,
      "loss": 1.9843,
      "step": 210014
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.540460228919983,
      "learning_rate": 1.1517687711937839e-05,
      "loss": 2.042,
      "step": 210015
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.3903344869613647,
      "learning_rate": 1.1516565174783277e-05,
      "loss": 2.1137,
      "step": 210016
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.5081154108047485,
      "learning_rate": 1.1515442691263176e-05,
      "loss": 1.9009,
      "step": 210017
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.6117140054702759,
      "learning_rate": 1.1514320261377875e-05,
      "loss": 2.188,
      "step": 210018
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.5279351472854614,
      "learning_rate": 1.1513197885127468e-05,
      "loss": 2.1419,
      "step": 210019
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.3914411067962646,
      "learning_rate": 1.1512075562512191e-05,
      "loss": 1.8398,
      "step": 210020
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.5269567966461182,
      "learning_rate": 1.151095329353231e-05,
      "loss": 1.9505,
      "step": 210021
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.3840543031692505,
      "learning_rate": 1.1509831078187959e-05,
      "loss": 2.052,
      "step": 210022
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.499436378479004,
      "learning_rate": 1.150870891647937e-05,
      "loss": 2.1917,
      "step": 210023
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.435204267501831,
      "learning_rate": 1.1507586808406778e-05,
      "loss": 1.827,
      "step": 210024
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.5181689262390137,
      "learning_rate": 1.1506464753970346e-05,
      "loss": 2.0825,
      "step": 210025
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.7503314018249512,
      "learning_rate": 1.150534275317031e-05,
      "loss": 2.1517,
      "step": 210026
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.48711097240448,
      "learning_rate": 1.1504220806006904e-05,
      "loss": 2.0085,
      "step": 210027
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.3556904792785645,
      "learning_rate": 1.1503098912480291e-05,
      "loss": 1.8325,
      "step": 210028
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.3610700368881226,
      "learning_rate": 1.1501977072590674e-05,
      "loss": 2.0664,
      "step": 210029
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.4459649324417114,
      "learning_rate": 1.1500855286338318e-05,
      "loss": 2.1926,
      "step": 210030
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.450875163078308,
      "learning_rate": 1.1499733553723423e-05,
      "loss": 1.9443,
      "step": 210031
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.4536998271942139,
      "learning_rate": 1.1498611874746123e-05,
      "loss": 2.0321,
      "step": 210032
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.4669275283813477,
      "learning_rate": 1.1497490249406682e-05,
      "loss": 2.1167,
      "step": 210033
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.4189081192016602,
      "learning_rate": 1.1496368677705303e-05,
      "loss": 2.1245,
      "step": 210034
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.5718841552734375,
      "learning_rate": 1.1495247159642217e-05,
      "loss": 1.9555,
      "step": 210035
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.439129114151001,
      "learning_rate": 1.1494125695217627e-05,
      "loss": 1.9823,
      "step": 210036
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.5537590980529785,
      "learning_rate": 1.1493004284431694e-05,
      "loss": 2.2306,
      "step": 210037
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.3390847444534302,
      "learning_rate": 1.1491882927284657e-05,
      "loss": 1.8331,
      "step": 210038
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.5499968528747559,
      "learning_rate": 1.1490761623776745e-05,
      "loss": 2.1473,
      "step": 210039
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.5149431228637695,
      "learning_rate": 1.1489640373908094e-05,
      "loss": 2.0504,
      "step": 210040
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.5029447078704834,
      "learning_rate": 1.1488519177679034e-05,
      "loss": 2.0715,
      "step": 210041
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.3741308450698853,
      "learning_rate": 1.1487398035089667e-05,
      "loss": 2.0424,
      "step": 210042
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.5581175088882446,
      "learning_rate": 1.148627694614026e-05,
      "loss": 1.8111,
      "step": 210043
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.4113447666168213,
      "learning_rate": 1.148515591083098e-05,
      "loss": 1.8953,
      "step": 210044
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.5269193649291992,
      "learning_rate": 1.1484034929162089e-05,
      "loss": 2.0993,
      "step": 210045
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.4380134344100952,
      "learning_rate": 1.1482914001133692e-05,
      "loss": 1.895,
      "step": 210046
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.4258660078048706,
      "learning_rate": 1.1481793126746119e-05,
      "loss": 2.008,
      "step": 210047
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.528529167175293,
      "learning_rate": 1.1480672305999505e-05,
      "loss": 2.0156,
      "step": 210048
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.5440056324005127,
      "learning_rate": 1.1479551538894149e-05,
      "loss": 1.8104,
      "step": 210049
    },
    {
      "epoch": 2.73,
      "grad_norm": 1.4696720838546753,
      "learning_rate": 1.1478430825430118e-05,
      "loss": 1.9269,
      "step": 210050
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.4113532304763794,
      "learning_rate": 1.1477310165607711e-05,
      "loss": 2.1185,
      "step": 210051
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.4035634994506836,
      "learning_rate": 1.1476189559427096e-05,
      "loss": 2.2533,
      "step": 210052
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.509966492652893,
      "learning_rate": 1.1475069006888538e-05,
      "loss": 2.0962,
      "step": 210053
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.7702341079711914,
      "learning_rate": 1.147394850799217e-05,
      "loss": 1.7998,
      "step": 210054
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.4116771221160889,
      "learning_rate": 1.1472828062738293e-05,
      "loss": 2.0136,
      "step": 210055
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.4006892442703247,
      "learning_rate": 1.147170767112704e-05,
      "loss": 1.9811,
      "step": 210056
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.4951765537261963,
      "learning_rate": 1.1470587333158676e-05,
      "loss": 2.0992,
      "step": 210057
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.5195443630218506,
      "learning_rate": 1.1469467048833303e-05,
      "loss": 2.0942,
      "step": 210058
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.5082485675811768,
      "learning_rate": 1.1468346818151252e-05,
      "loss": 1.8064,
      "step": 210059
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.4586944580078125,
      "learning_rate": 1.1467226641112658e-05,
      "loss": 1.7451,
      "step": 210060
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.4689204692840576,
      "learning_rate": 1.1466106517717788e-05,
      "loss": 1.9491,
      "step": 210061
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.6389563083648682,
      "learning_rate": 1.1464986447966806e-05,
      "loss": 2.1925,
      "step": 210062
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.3804150819778442,
      "learning_rate": 1.1463866431859913e-05,
      "loss": 1.8232,
      "step": 210063
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.5722784996032715,
      "learning_rate": 1.146274646939731e-05,
      "loss": 2.1195,
      "step": 210064
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.4797388315200806,
      "learning_rate": 1.1461626560579296e-05,
      "loss": 2.0745,
      "step": 210065
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.495651125907898,
      "learning_rate": 1.1460506705405936e-05,
      "loss": 1.7057,
      "step": 210066
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.401281476020813,
      "learning_rate": 1.1459386903877565e-05,
      "loss": 2.1448,
      "step": 210067
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.4304416179656982,
      "learning_rate": 1.145826715599435e-05,
      "loss": 1.9832,
      "step": 210068
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.382590413093567,
      "learning_rate": 1.1457147461756489e-05,
      "loss": 2.1191,
      "step": 210069
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.4365595579147339,
      "learning_rate": 1.1456027821164148e-05,
      "loss": 2.1063,
      "step": 210070
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.4640452861785889,
      "learning_rate": 1.1454908234217596e-05,
      "loss": 2.0952,
      "step": 210071
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.3562343120574951,
      "learning_rate": 1.1453788700917e-05,
      "loss": 2.0237,
      "step": 210072
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.4431661367416382,
      "learning_rate": 1.1452669221262656e-05,
      "loss": 1.9685,
      "step": 210073
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.5663552284240723,
      "learning_rate": 1.1451549795254666e-05,
      "loss": 1.9401,
      "step": 210074
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.454468011856079,
      "learning_rate": 1.1450430422893298e-05,
      "loss": 2.0313,
      "step": 210075
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.6452313661575317,
      "learning_rate": 1.1449311104178716e-05,
      "loss": 1.9813,
      "step": 210076
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.4177473783493042,
      "learning_rate": 1.1448191839111153e-05,
      "loss": 1.9602,
      "step": 210077
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.4446228742599487,
      "learning_rate": 1.1447072627690812e-05,
      "loss": 2.1755,
      "step": 210078
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.424892783164978,
      "learning_rate": 1.1445953469917957e-05,
      "loss": 2.019,
      "step": 210079
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.5602647066116333,
      "learning_rate": 1.144483436579272e-05,
      "loss": 2.076,
      "step": 210080
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.6104305982589722,
      "learning_rate": 1.1443715315315338e-05,
      "loss": 2.0973,
      "step": 210081
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.4500013589859009,
      "learning_rate": 1.1442596318486008e-05,
      "loss": 1.9208,
      "step": 210082
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.442366123199463,
      "learning_rate": 1.1441477375304931e-05,
      "loss": 1.9752,
      "step": 210083
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.460404634475708,
      "learning_rate": 1.1440358485772338e-05,
      "loss": 1.9275,
      "step": 210084
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.3844246864318848,
      "learning_rate": 1.1439239649888432e-05,
      "loss": 1.8115,
      "step": 210085
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.3489937782287598,
      "learning_rate": 1.1438120867653444e-05,
      "loss": 1.9573,
      "step": 210086
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.5532296895980835,
      "learning_rate": 1.1437002139067575e-05,
      "loss": 1.998,
      "step": 210087
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.5618959665298462,
      "learning_rate": 1.1435883464130924e-05,
      "loss": 2.1195,
      "step": 210088
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.3315695524215698,
      "learning_rate": 1.1434764842843857e-05,
      "loss": 1.7259,
      "step": 210089
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.4568023681640625,
      "learning_rate": 1.1433646275206476e-05,
      "loss": 2.0339,
      "step": 210090
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.5901243686676025,
      "learning_rate": 1.1432527761219079e-05,
      "loss": 1.9144,
      "step": 210091
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.5489068031311035,
      "learning_rate": 1.14314093008818e-05,
      "loss": 1.9674,
      "step": 210092
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.4597257375717163,
      "learning_rate": 1.1430290894194872e-05,
      "loss": 2.0039,
      "step": 210093
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.5497655868530273,
      "learning_rate": 1.142917254115846e-05,
      "loss": 1.8726,
      "step": 210094
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.4572674036026,
      "learning_rate": 1.1428054241772865e-05,
      "loss": 1.8141,
      "step": 210095
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.477081298828125,
      "learning_rate": 1.1426935996038189e-05,
      "loss": 2.0097,
      "step": 210096
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.2948262691497803,
      "learning_rate": 1.1425817803954728e-05,
      "loss": 2.0495,
      "step": 210097
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.5095378160476685,
      "learning_rate": 1.1424699665522685e-05,
      "loss": 2.0283,
      "step": 210098
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.3862751722335815,
      "learning_rate": 1.1423581580742192e-05,
      "loss": 1.7381,
      "step": 210099
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.6621261835098267,
      "learning_rate": 1.1422463549613515e-05,
      "loss": 2.2471,
      "step": 210100
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.538521409034729,
      "learning_rate": 1.1421345572136853e-05,
      "loss": 1.9803,
      "step": 210101
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.3638778924942017,
      "learning_rate": 1.1420227648312374e-05,
      "loss": 2.0567,
      "step": 210102
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.69193434715271,
      "learning_rate": 1.1419109778140346e-05,
      "loss": 2.0384,
      "step": 210103
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.3801465034484863,
      "learning_rate": 1.1417991961620998e-05,
      "loss": 2.2113,
      "step": 210104
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.7125736474990845,
      "learning_rate": 1.1416874198754466e-05,
      "loss": 1.7004,
      "step": 210105
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.4746358394622803,
      "learning_rate": 1.141575648954095e-05,
      "loss": 2.1798,
      "step": 210106
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.3851810693740845,
      "learning_rate": 1.1414638833980716e-05,
      "loss": 1.8052,
      "step": 210107
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.5373988151550293,
      "learning_rate": 1.141352123207393e-05,
      "loss": 1.901,
      "step": 210108
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.4609049558639526,
      "learning_rate": 1.1412403683820826e-05,
      "loss": 2.1891,
      "step": 210109
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.5151941776275635,
      "learning_rate": 1.1411286189221636e-05,
      "loss": 2.085,
      "step": 210110
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.4572877883911133,
      "learning_rate": 1.1410168748276494e-05,
      "loss": 2.0567,
      "step": 210111
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.5765414237976074,
      "learning_rate": 1.1409051360985665e-05,
      "loss": 1.9374,
      "step": 210112
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.3632903099060059,
      "learning_rate": 1.1407934027349353e-05,
      "loss": 1.953,
      "step": 210113
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.4766937494277954,
      "learning_rate": 1.140681674736772e-05,
      "loss": 2.1256,
      "step": 210114
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.4790505170822144,
      "learning_rate": 1.1405699521041068e-05,
      "loss": 2.2167,
      "step": 210115
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.4829727411270142,
      "learning_rate": 1.1404582348369496e-05,
      "loss": 1.9049,
      "step": 210116
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.4474047422409058,
      "learning_rate": 1.140346522935327e-05,
      "loss": 2.2507,
      "step": 210117
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.541281819343567,
      "learning_rate": 1.1402348163992593e-05,
      "loss": 2.1347,
      "step": 210118
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.5369662046432495,
      "learning_rate": 1.1401231152287627e-05,
      "loss": 2.0078,
      "step": 210119
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.576305627822876,
      "learning_rate": 1.1400114194238674e-05,
      "loss": 2.1298,
      "step": 210120
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.6690620183944702,
      "learning_rate": 1.1398997289845902e-05,
      "loss": 2.0085,
      "step": 210121
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.3888503313064575,
      "learning_rate": 1.1397880439109442e-05,
      "loss": 1.9928,
      "step": 210122
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.4108296632766724,
      "learning_rate": 1.1396763642029594e-05,
      "loss": 1.7758,
      "step": 210123
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.3004567623138428,
      "learning_rate": 1.1395646898606558e-05,
      "loss": 2.0525,
      "step": 210124
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.6115834712982178,
      "learning_rate": 1.1394530208840469e-05,
      "loss": 1.7828,
      "step": 210125
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.4212738275527954,
      "learning_rate": 1.1393413572731624e-05,
      "loss": 2.0231,
      "step": 210126
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.5813565254211426,
      "learning_rate": 1.1392296990280191e-05,
      "loss": 1.8619,
      "step": 210127
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.398345947265625,
      "learning_rate": 1.1391180461486337e-05,
      "loss": 1.7265,
      "step": 210128
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.5102437734603882,
      "learning_rate": 1.1390063986350362e-05,
      "loss": 2.0371,
      "step": 210129
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.5319031476974487,
      "learning_rate": 1.1388947564872397e-05,
      "loss": 1.8173,
      "step": 210130
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.3957552909851074,
      "learning_rate": 1.1387831197052643e-05,
      "loss": 1.9514,
      "step": 210131
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.4058045148849487,
      "learning_rate": 1.1386714882891402e-05,
      "loss": 1.7914,
      "step": 210132
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.314469575881958,
      "learning_rate": 1.1385598622388769e-05,
      "loss": 1.781,
      "step": 210133
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.5319304466247559,
      "learning_rate": 1.1384482415545016e-05,
      "loss": 1.6835,
      "step": 210134
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.4739770889282227,
      "learning_rate": 1.1383366262360372e-05,
      "loss": 1.9748,
      "step": 210135
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.3306469917297363,
      "learning_rate": 1.1382250162834972e-05,
      "loss": 1.8086,
      "step": 210136
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.4859145879745483,
      "learning_rate": 1.1381134116969048e-05,
      "loss": 1.9359,
      "step": 210137
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.4728288650512695,
      "learning_rate": 1.1380018124762868e-05,
      "loss": 2.0695,
      "step": 210138
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.3614256381988525,
      "learning_rate": 1.1378902186216532e-05,
      "loss": 1.8802,
      "step": 210139
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.4331153631210327,
      "learning_rate": 1.1377786301330338e-05,
      "loss": 2.0962,
      "step": 210140
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.4231065511703491,
      "learning_rate": 1.1376670470104455e-05,
      "loss": 1.7564,
      "step": 210141
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.4591752290725708,
      "learning_rate": 1.1375554692539112e-05,
      "loss": 1.9282,
      "step": 210142
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.552185297012329,
      "learning_rate": 1.137443896863448e-05,
      "loss": 1.9865,
      "step": 210143
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.476367712020874,
      "learning_rate": 1.1373323298390825e-05,
      "loss": 2.0969,
      "step": 210144
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.5269856452941895,
      "learning_rate": 1.1372207681808243e-05,
      "loss": 1.9724,
      "step": 210145
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.378899335861206,
      "learning_rate": 1.1371092118887071e-05,
      "loss": 2.0029,
      "step": 210146
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.4836015701293945,
      "learning_rate": 1.1369976609627473e-05,
      "loss": 1.9016,
      "step": 210147
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.387276530265808,
      "learning_rate": 1.1368861154029652e-05,
      "loss": 1.92,
      "step": 210148
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.4159042835235596,
      "learning_rate": 1.1367745752093738e-05,
      "loss": 2.0205,
      "step": 210149
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.4522854089736938,
      "learning_rate": 1.1366630403820065e-05,
      "loss": 1.9854,
      "step": 210150
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.373876690864563,
      "learning_rate": 1.1365515109208733e-05,
      "loss": 2.156,
      "step": 210151
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.378873586654663,
      "learning_rate": 1.1364399868260077e-05,
      "loss": 1.7589,
      "step": 210152
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.4167051315307617,
      "learning_rate": 1.1363284680974194e-05,
      "loss": 2.1221,
      "step": 210153
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.3682169914245605,
      "learning_rate": 1.1362169547351318e-05,
      "loss": 1.8461,
      "step": 210154
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.3322486877441406,
      "learning_rate": 1.136105446739165e-05,
      "loss": 2.258,
      "step": 210155
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.542258858680725,
      "learning_rate": 1.1359939441095422e-05,
      "loss": 1.8854,
      "step": 210156
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.417284369468689,
      "learning_rate": 1.1358824468462835e-05,
      "loss": 2.1169,
      "step": 210157
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.4037009477615356,
      "learning_rate": 1.1357709549494087e-05,
      "loss": 2.0411,
      "step": 210158
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.4818893671035767,
      "learning_rate": 1.1356594684189412e-05,
      "loss": 2.1348,
      "step": 210159
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.5012011528015137,
      "learning_rate": 1.1355479872548979e-05,
      "loss": 2.0486,
      "step": 210160
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.4126880168914795,
      "learning_rate": 1.1354365114572983e-05,
      "loss": 1.9737,
      "step": 210161
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.4599478244781494,
      "learning_rate": 1.1353250410261694e-05,
      "loss": 2.0771,
      "step": 210162
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.3970434665679932,
      "learning_rate": 1.1352135759615244e-05,
      "loss": 1.8591,
      "step": 210163
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.4576473236083984,
      "learning_rate": 1.13510211626339e-05,
      "loss": 2.0929,
      "step": 210164
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.4594752788543701,
      "learning_rate": 1.1349906619317895e-05,
      "loss": 1.9457,
      "step": 210165
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.5011649131774902,
      "learning_rate": 1.1348792129667361e-05,
      "loss": 1.9959,
      "step": 210166
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.3617184162139893,
      "learning_rate": 1.13476776936825e-05,
      "loss": 1.9424,
      "step": 210167
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.5167667865753174,
      "learning_rate": 1.1346563311363577e-05,
      "loss": 2.357,
      "step": 210168
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.5058655738830566,
      "learning_rate": 1.1345448982710759e-05,
      "loss": 1.9177,
      "step": 210169
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.392511248588562,
      "learning_rate": 1.1344334707724312e-05,
      "loss": 1.8891,
      "step": 210170
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.6986616849899292,
      "learning_rate": 1.1343220486404402e-05,
      "loss": 2.2037,
      "step": 210171
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.4555665254592896,
      "learning_rate": 1.1342106318751231e-05,
      "loss": 2.0331,
      "step": 210172
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.4086054563522339,
      "learning_rate": 1.1340992204764965e-05,
      "loss": 2.0277,
      "step": 210173
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.3912041187286377,
      "learning_rate": 1.1339878144445902e-05,
      "loss": 1.7827,
      "step": 210174
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.455651879310608,
      "learning_rate": 1.1338764137794177e-05,
      "loss": 2.0796,
      "step": 210175
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.5456477403640747,
      "learning_rate": 1.1337650184810054e-05,
      "loss": 1.8452,
      "step": 210176
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.5340733528137207,
      "learning_rate": 1.1336536285493702e-05,
      "loss": 1.8253,
      "step": 210177
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.3487762212753296,
      "learning_rate": 1.1335422439845355e-05,
      "loss": 2.0725,
      "step": 210178
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.561516284942627,
      "learning_rate": 1.1334308647865143e-05,
      "loss": 1.9339,
      "step": 210179
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.4270514249801636,
      "learning_rate": 1.13331949095534e-05,
      "loss": 1.9943,
      "step": 210180
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.5136874914169312,
      "learning_rate": 1.1332081224910194e-05,
      "loss": 1.8965,
      "step": 210181
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.409616231918335,
      "learning_rate": 1.1330967593935857e-05,
      "loss": 1.9603,
      "step": 210182
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.4842579364776611,
      "learning_rate": 1.1329854016630557e-05,
      "loss": 2.0412,
      "step": 210183
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.5604389905929565,
      "learning_rate": 1.132874049299446e-05,
      "loss": 2.2041,
      "step": 210184
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.4557535648345947,
      "learning_rate": 1.1327627023027763e-05,
      "loss": 1.8836,
      "step": 210185
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.435564398765564,
      "learning_rate": 1.1326513606730769e-05,
      "loss": 1.9496,
      "step": 210186
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.6195577383041382,
      "learning_rate": 1.1325400244103577e-05,
      "loss": 2.121,
      "step": 210187
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.3801653385162354,
      "learning_rate": 1.1324286935146454e-05,
      "loss": 2.0724,
      "step": 210188
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.4641340970993042,
      "learning_rate": 1.1323173679859599e-05,
      "loss": 2.008,
      "step": 210189
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.4161403179168701,
      "learning_rate": 1.1322060478243278e-05,
      "loss": 2.044,
      "step": 210190
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.4061399698257446,
      "learning_rate": 1.1320947330297525e-05,
      "loss": 1.9561,
      "step": 210191
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.5922468900680542,
      "learning_rate": 1.131983423602274e-05,
      "loss": 1.9026,
      "step": 210192
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.4662408828735352,
      "learning_rate": 1.131872119541899e-05,
      "loss": 1.9998,
      "step": 210193
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.4406157732009888,
      "learning_rate": 1.1317608208486572e-05,
      "loss": 1.8956,
      "step": 210194
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.5787062644958496,
      "learning_rate": 1.1316495275225623e-05,
      "loss": 1.9047,
      "step": 210195
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.2965185642242432,
      "learning_rate": 1.1315382395636441e-05,
      "loss": 1.8277,
      "step": 210196
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.368450403213501,
      "learning_rate": 1.131426956971916e-05,
      "loss": 1.904,
      "step": 210197
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.6740155220031738,
      "learning_rate": 1.1313156797474044e-05,
      "loss": 1.8855,
      "step": 210198
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.4307461977005005,
      "learning_rate": 1.1312044078901195e-05,
      "loss": 1.822,
      "step": 210199
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.3942418098449707,
      "learning_rate": 1.1310931414000912e-05,
      "loss": 2.0021,
      "step": 210200
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.4378094673156738,
      "learning_rate": 1.130981880277333e-05,
      "loss": 1.9218,
      "step": 210201
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.4372878074645996,
      "learning_rate": 1.130870624521878e-05,
      "loss": 1.8631,
      "step": 210202
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.6208205223083496,
      "learning_rate": 1.1307593741337361e-05,
      "loss": 1.9256,
      "step": 210203
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.50504469871521,
      "learning_rate": 1.130648129112931e-05,
      "loss": 1.8577,
      "step": 210204
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.4477143287658691,
      "learning_rate": 1.1305368894594824e-05,
      "loss": 2.1329,
      "step": 210205
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.4932717084884644,
      "learning_rate": 1.1304256551734137e-05,
      "loss": 1.8993,
      "step": 210206
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.4975194931030273,
      "learning_rate": 1.1303144262547414e-05,
      "loss": 2.2038,
      "step": 210207
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.4642207622528076,
      "learning_rate": 1.130203202703489e-05,
      "loss": 1.6718,
      "step": 210208
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.50706946849823,
      "learning_rate": 1.13009198451968e-05,
      "loss": 2.172,
      "step": 210209
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.2743840217590332,
      "learning_rate": 1.1299807717033272e-05,
      "loss": 1.8225,
      "step": 210210
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.5769522190093994,
      "learning_rate": 1.1298695642544608e-05,
      "loss": 1.9424,
      "step": 210211
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.4327120780944824,
      "learning_rate": 1.1297583621730944e-05,
      "loss": 2.115,
      "step": 210212
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.492453932762146,
      "learning_rate": 1.1296471654592476e-05,
      "loss": 2.037,
      "step": 210213
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.4277576208114624,
      "learning_rate": 1.1295359741129473e-05,
      "loss": 1.939,
      "step": 210214
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.4160102605819702,
      "learning_rate": 1.1294247881342133e-05,
      "loss": 1.9956,
      "step": 210215
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.5834518671035767,
      "learning_rate": 1.129313607523059e-05,
      "loss": 1.9357,
      "step": 210216
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.6253482103347778,
      "learning_rate": 1.129202432279518e-05,
      "loss": 2.0441,
      "step": 210217
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.5402092933654785,
      "learning_rate": 1.129091262403593e-05,
      "loss": 1.9306,
      "step": 210218
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.5301302671432495,
      "learning_rate": 1.1289800978953245e-05,
      "loss": 2.1211,
      "step": 210219
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.3661967515945435,
      "learning_rate": 1.128868938754719e-05,
      "loss": 1.9072,
      "step": 210220
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.5308828353881836,
      "learning_rate": 1.1287577849818063e-05,
      "loss": 2.0498,
      "step": 210221
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.3627477884292603,
      "learning_rate": 1.1286466365765933e-05,
      "loss": 1.8806,
      "step": 210222
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.5185859203338623,
      "learning_rate": 1.1285354935391167e-05,
      "loss": 2.039,
      "step": 210223
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.4065611362457275,
      "learning_rate": 1.1284243558693895e-05,
      "loss": 2.0044,
      "step": 210224
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.5017123222351074,
      "learning_rate": 1.1283132235674318e-05,
      "loss": 1.9994,
      "step": 210225
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.7369011640548706,
      "learning_rate": 1.1282020966332672e-05,
      "loss": 1.9221,
      "step": 210226
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.5636000633239746,
      "learning_rate": 1.1280909750669153e-05,
      "loss": 2.0233,
      "step": 210227
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.6340970993041992,
      "learning_rate": 1.127979858868393e-05,
      "loss": 2.1402,
      "step": 210228
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.4493520259857178,
      "learning_rate": 1.1278687480377269e-05,
      "loss": 1.9372,
      "step": 210229
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.5324734449386597,
      "learning_rate": 1.1277576425749335e-05,
      "loss": 2.1348,
      "step": 210230
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.4275929927825928,
      "learning_rate": 1.1276465424800363e-05,
      "loss": 1.8854,
      "step": 210231
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.3941363096237183,
      "learning_rate": 1.1275354477530551e-05,
      "loss": 1.9764,
      "step": 210232
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.5703620910644531,
      "learning_rate": 1.1274243583940101e-05,
      "loss": 2.1793,
      "step": 210233
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.4762921333312988,
      "learning_rate": 1.1273132744029212e-05,
      "loss": 2.0824,
      "step": 210234
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.41912043094635,
      "learning_rate": 1.1272021957798083e-05,
      "loss": 2.0369,
      "step": 210235
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.4183560609817505,
      "learning_rate": 1.1270911225246948e-05,
      "loss": 1.8012,
      "step": 210236
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.4692821502685547,
      "learning_rate": 1.1269800546376007e-05,
      "loss": 2.0711,
      "step": 210237
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.6427278518676758,
      "learning_rate": 1.1268689921185458e-05,
      "loss": 2.2609,
      "step": 210238
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.4737735986709595,
      "learning_rate": 1.1267579349675537e-05,
      "loss": 2.0399,
      "step": 210239
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.4534416198730469,
      "learning_rate": 1.1266468831846343e-05,
      "loss": 1.9173,
      "step": 210240
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.4894828796386719,
      "learning_rate": 1.1265358367698241e-05,
      "loss": 2.1801,
      "step": 210241
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.4883652925491333,
      "learning_rate": 1.1264247957231298e-05,
      "loss": 1.8566,
      "step": 210242
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.2889448404312134,
      "learning_rate": 1.1263137600445849e-05,
      "loss": 1.8149,
      "step": 210243
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.5056884288787842,
      "learning_rate": 1.126202729734199e-05,
      "loss": 1.8523,
      "step": 210244
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.371852993965149,
      "learning_rate": 1.1260917047919994e-05,
      "loss": 2.0017,
      "step": 210245
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.296831727027893,
      "learning_rate": 1.125980685218002e-05,
      "loss": 1.9298,
      "step": 210246
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.4012607336044312,
      "learning_rate": 1.1258696710122306e-05,
      "loss": 1.8914,
      "step": 210247
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.3834739923477173,
      "learning_rate": 1.1257586621747017e-05,
      "loss": 1.8826,
      "step": 210248
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.498928189277649,
      "learning_rate": 1.1256476587054452e-05,
      "loss": 1.761,
      "step": 210249
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.5631632804870605,
      "learning_rate": 1.1255366606044746e-05,
      "loss": 2.0937,
      "step": 210250
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.4094855785369873,
      "learning_rate": 1.1254256678718098e-05,
      "loss": 2.1209,
      "step": 210251
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.3792511224746704,
      "learning_rate": 1.1253146805074741e-05,
      "loss": 1.8102,
      "step": 210252
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.4339032173156738,
      "learning_rate": 1.1252036985114876e-05,
      "loss": 1.9003,
      "step": 210253
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.4608951807022095,
      "learning_rate": 1.1250927218838702e-05,
      "loss": 1.8546,
      "step": 210254
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.5742396116256714,
      "learning_rate": 1.124981750624645e-05,
      "loss": 2.1037,
      "step": 210255
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.3508450984954834,
      "learning_rate": 1.1248707847338256e-05,
      "loss": 1.9254,
      "step": 210256
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.5201659202575684,
      "learning_rate": 1.1247598242114453e-05,
      "loss": 1.9319,
      "step": 210257
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.3968071937561035,
      "learning_rate": 1.1246488690575107e-05,
      "loss": 1.8589,
      "step": 210258
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.3993583917617798,
      "learning_rate": 1.124537919272055e-05,
      "loss": 2.0661,
      "step": 210259
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.4445643424987793,
      "learning_rate": 1.1244269748550883e-05,
      "loss": 1.9731,
      "step": 210260
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.4167876243591309,
      "learning_rate": 1.1243160358066372e-05,
      "loss": 1.8309,
      "step": 210261
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.3831067085266113,
      "learning_rate": 1.1242051021267184e-05,
      "loss": 1.9344,
      "step": 210262
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.6331098079681396,
      "learning_rate": 1.124094173815362e-05,
      "loss": 2.032,
      "step": 210263
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.5380887985229492,
      "learning_rate": 1.1239832508725743e-05,
      "loss": 1.8362,
      "step": 210264
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.4026439189910889,
      "learning_rate": 1.1238723332983857e-05,
      "loss": 1.847,
      "step": 210265
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.5464178323745728,
      "learning_rate": 1.1237614210928125e-05,
      "loss": 2.064,
      "step": 210266
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.461281180381775,
      "learning_rate": 1.1236505142558783e-05,
      "loss": 2.1309,
      "step": 210267
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.349316954612732,
      "learning_rate": 1.1235396127876029e-05,
      "loss": 1.905,
      "step": 210268
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.461877465248108,
      "learning_rate": 1.1234287166880096e-05,
      "loss": 1.8321,
      "step": 210269
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.4539053440093994,
      "learning_rate": 1.1233178259571118e-05,
      "loss": 1.8979,
      "step": 210270
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.4692819118499756,
      "learning_rate": 1.1232069405949362e-05,
      "loss": 2.0437,
      "step": 210271
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.4611015319824219,
      "learning_rate": 1.1230960606014994e-05,
      "loss": 2.0172,
      "step": 210272
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.5359498262405396,
      "learning_rate": 1.1229851859768247e-05,
      "loss": 1.975,
      "step": 210273
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.520410180091858,
      "learning_rate": 1.1228743167209319e-05,
      "loss": 2.1233,
      "step": 210274
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.309341549873352,
      "learning_rate": 1.122763452833848e-05,
      "loss": 1.7453,
      "step": 210275
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.4830788373947144,
      "learning_rate": 1.1226525943155795e-05,
      "loss": 1.8661,
      "step": 210276
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.5273863077163696,
      "learning_rate": 1.1225417411661596e-05,
      "loss": 1.7707,
      "step": 210277
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.3454874753952026,
      "learning_rate": 1.1224308933856019e-05,
      "loss": 2.0229,
      "step": 210278
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.5475066900253296,
      "learning_rate": 1.1223200509739294e-05,
      "loss": 1.9403,
      "step": 210279
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.4174344539642334,
      "learning_rate": 1.1222092139311623e-05,
      "loss": 2.1215,
      "step": 210280
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.4658606052398682,
      "learning_rate": 1.1220983822573238e-05,
      "loss": 2.0211,
      "step": 210281
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.4497840404510498,
      "learning_rate": 1.1219875559524338e-05,
      "loss": 1.801,
      "step": 210282
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.5189236402511597,
      "learning_rate": 1.1218767350165092e-05,
      "loss": 2.0621,
      "step": 210283
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.4837515354156494,
      "learning_rate": 1.1217659194495699e-05,
      "loss": 1.923,
      "step": 210284
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.5399020910263062,
      "learning_rate": 1.1216551092516424e-05,
      "loss": 1.9603,
      "step": 210285
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.4860279560089111,
      "learning_rate": 1.1215443044227434e-05,
      "loss": 2.0585,
      "step": 210286
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.4197438955307007,
      "learning_rate": 1.1214335049628964e-05,
      "loss": 1.9149,
      "step": 210287
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.4464470148086548,
      "learning_rate": 1.1213227108721178e-05,
      "loss": 1.997,
      "step": 210288
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.5426559448242188,
      "learning_rate": 1.1212119221504346e-05,
      "loss": 2.0327,
      "step": 210289
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.3870959281921387,
      "learning_rate": 1.1211011387978564e-05,
      "loss": 2.0095,
      "step": 210290
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.5786497592926025,
      "learning_rate": 1.1209903608144166e-05,
      "loss": 2.0236,
      "step": 210291
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.4128539562225342,
      "learning_rate": 1.1208795882001254e-05,
      "loss": 1.962,
      "step": 210292
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.5239534378051758,
      "learning_rate": 1.1207688209550091e-05,
      "loss": 1.9117,
      "step": 210293
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.411360502243042,
      "learning_rate": 1.1206580590790914e-05,
      "loss": 1.8768,
      "step": 210294
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.4231408834457397,
      "learning_rate": 1.120547302572382e-05,
      "loss": 1.9743,
      "step": 210295
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.3126453161239624,
      "learning_rate": 1.1204365514349112e-05,
      "loss": 1.7935,
      "step": 210296
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.4973220825195312,
      "learning_rate": 1.1203258056666985e-05,
      "loss": 1.822,
      "step": 210297
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.6293635368347168,
      "learning_rate": 1.1202150652677577e-05,
      "loss": 2.0405,
      "step": 210298
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.395879864692688,
      "learning_rate": 1.1201043302381153e-05,
      "loss": 2.1141,
      "step": 210299
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.3648200035095215,
      "learning_rate": 1.1199936005777943e-05,
      "loss": 1.892,
      "step": 210300
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.7400718927383423,
      "learning_rate": 1.1198828762868084e-05,
      "loss": 2.0985,
      "step": 210301
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.4079307317733765,
      "learning_rate": 1.1197721573651809e-05,
      "loss": 1.8807,
      "step": 210302
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.420512080192566,
      "learning_rate": 1.1196614438129314e-05,
      "loss": 1.7562,
      "step": 210303
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.4007563591003418,
      "learning_rate": 1.1195507356300871e-05,
      "loss": 1.9752,
      "step": 210304
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.6706547737121582,
      "learning_rate": 1.1194400328166609e-05,
      "loss": 2.0848,
      "step": 210305
    },
    {
      "epoch": 2.74,
      "grad_norm": 2.7040069103240967,
      "learning_rate": 1.1193293353726796e-05,
      "loss": 1.9794,
      "step": 210306
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.5505781173706055,
      "learning_rate": 1.1192186432981531e-05,
      "loss": 1.7276,
      "step": 210307
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.5010545253753662,
      "learning_rate": 1.119107956593115e-05,
      "loss": 2.0731,
      "step": 210308
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.4811795949935913,
      "learning_rate": 1.118997275257575e-05,
      "loss": 1.9524,
      "step": 210309
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.3323217630386353,
      "learning_rate": 1.1188865992915629e-05,
      "loss": 1.9113,
      "step": 210310
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.6076425313949585,
      "learning_rate": 1.1187759286950926e-05,
      "loss": 2.0228,
      "step": 210311
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.5479609966278076,
      "learning_rate": 1.1186652634681904e-05,
      "loss": 2.1805,
      "step": 210312
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.4192934036254883,
      "learning_rate": 1.1185546036108661e-05,
      "loss": 1.9521,
      "step": 210313
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.48899245262146,
      "learning_rate": 1.1184439491231534e-05,
      "loss": 1.932,
      "step": 210314
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.4965488910675049,
      "learning_rate": 1.1183333000050653e-05,
      "loss": 2.101,
      "step": 210315
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.4579380750656128,
      "learning_rate": 1.1182226562566255e-05,
      "loss": 1.8152,
      "step": 210316
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.5742911100387573,
      "learning_rate": 1.1181120178778535e-05,
      "loss": 1.7958,
      "step": 210317
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.5308599472045898,
      "learning_rate": 1.1180013848687696e-05,
      "loss": 1.8028,
      "step": 210318
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.418499231338501,
      "learning_rate": 1.1178907572293938e-05,
      "loss": 1.7823,
      "step": 210319
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.456390380859375,
      "learning_rate": 1.117780134959746e-05,
      "loss": 2.0096,
      "step": 210320
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.4771537780761719,
      "learning_rate": 1.1176695180598494e-05,
      "loss": 2.126,
      "step": 210321
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.6080650091171265,
      "learning_rate": 1.1175589065297241e-05,
      "loss": 1.7602,
      "step": 210322
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.4326664209365845,
      "learning_rate": 1.1174483003693834e-05,
      "loss": 1.8076,
      "step": 210323
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.4519054889678955,
      "learning_rate": 1.1173376995788675e-05,
      "loss": 1.9129,
      "step": 210324
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.4779651165008545,
      "learning_rate": 1.1172271041581727e-05,
      "loss": 2.0049,
      "step": 210325
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.5804344415664673,
      "learning_rate": 1.1171165141073358e-05,
      "loss": 2.1328,
      "step": 210326
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.5187163352966309,
      "learning_rate": 1.1170059294263667e-05,
      "loss": 1.9401,
      "step": 210327
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.5448857545852661,
      "learning_rate": 1.1168953501152955e-05,
      "loss": 1.9836,
      "step": 210328
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.398202657699585,
      "learning_rate": 1.1167847761741355e-05,
      "loss": 1.9569,
      "step": 210329
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.3915239572525024,
      "learning_rate": 1.1166742076029167e-05,
      "loss": 1.994,
      "step": 210330
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.3696054220199585,
      "learning_rate": 1.116563644401649e-05,
      "loss": 1.9996,
      "step": 210331
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.3978915214538574,
      "learning_rate": 1.1164530865703591e-05,
      "loss": 2.0223,
      "step": 210332
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.431752324104309,
      "learning_rate": 1.1163425341090604e-05,
      "loss": 1.9597,
      "step": 210333
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.6325920820236206,
      "learning_rate": 1.116231987017786e-05,
      "loss": 2.0424,
      "step": 210334
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.3963276147842407,
      "learning_rate": 1.1161214452965428e-05,
      "loss": 1.9015,
      "step": 210335
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.5310057401657104,
      "learning_rate": 1.1160109089453672e-05,
      "loss": 1.8732,
      "step": 210336
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.540156602859497,
      "learning_rate": 1.1159003779642627e-05,
      "loss": 1.8721,
      "step": 210337
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.3515307903289795,
      "learning_rate": 1.1157898523532594e-05,
      "loss": 1.8959,
      "step": 210338
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.505889654159546,
      "learning_rate": 1.1156793321123737e-05,
      "loss": 1.9587,
      "step": 210339
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.5911810398101807,
      "learning_rate": 1.1155688172416322e-05,
      "loss": 2.1924,
      "step": 210340
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.6894007921218872,
      "learning_rate": 1.1154583077410483e-05,
      "loss": 1.9785,
      "step": 210341
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.4044593572616577,
      "learning_rate": 1.1153478036106522e-05,
      "loss": 2.1157,
      "step": 210342
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.4271517992019653,
      "learning_rate": 1.1152373048504537e-05,
      "loss": 2.042,
      "step": 210343
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.5027767419815063,
      "learning_rate": 1.1151268114604795e-05,
      "loss": 1.99,
      "step": 210344
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.536258578300476,
      "learning_rate": 1.1150163234407428e-05,
      "loss": 2.2213,
      "step": 210345
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.4571633338928223,
      "learning_rate": 1.114905840791277e-05,
      "loss": 1.9287,
      "step": 210346
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.4877341985702515,
      "learning_rate": 1.1147953635120888e-05,
      "loss": 2.1587,
      "step": 210347
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.4753859043121338,
      "learning_rate": 1.1146848916032147e-05,
      "loss": 2.0778,
      "step": 210348
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.3197002410888672,
      "learning_rate": 1.1145744250646583e-05,
      "loss": 1.7347,
      "step": 210349
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.4213095903396606,
      "learning_rate": 1.1144639638964493e-05,
      "loss": 2.1088,
      "step": 210350
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.3628036975860596,
      "learning_rate": 1.1143535080986043e-05,
      "loss": 1.8239,
      "step": 210351
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.443284273147583,
      "learning_rate": 1.1142430576711504e-05,
      "loss": 2.0717,
      "step": 210352
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.5219677686691284,
      "learning_rate": 1.1141326126141004e-05,
      "loss": 1.9457,
      "step": 210353
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.4122748374938965,
      "learning_rate": 1.114022172927488e-05,
      "loss": 2.042,
      "step": 210354
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.4477717876434326,
      "learning_rate": 1.1139117386113128e-05,
      "loss": 2.0605,
      "step": 210355
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.4430437088012695,
      "learning_rate": 1.1138013096656117e-05,
      "loss": 2.0578,
      "step": 210356
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.457439661026001,
      "learning_rate": 1.1136908860903981e-05,
      "loss": 2.0513,
      "step": 210357
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.527016043663025,
      "learning_rate": 1.1135804678856953e-05,
      "loss": 1.9009,
      "step": 210358
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.5407527685165405,
      "learning_rate": 1.113470055051523e-05,
      "loss": 1.8616,
      "step": 210359
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.6353890895843506,
      "learning_rate": 1.1133596475879048e-05,
      "loss": 2.0384,
      "step": 210360
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.3916717767715454,
      "learning_rate": 1.1132492454948538e-05,
      "loss": 1.9222,
      "step": 210361
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.43644380569458,
      "learning_rate": 1.1131388487724002e-05,
      "loss": 1.9048,
      "step": 210362
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.5641366243362427,
      "learning_rate": 1.113028457420554e-05,
      "loss": 1.9097,
      "step": 210363
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.7093263864517212,
      "learning_rate": 1.1129180714393448e-05,
      "loss": 2.0423,
      "step": 210364
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.5981831550598145,
      "learning_rate": 1.1128076908287863e-05,
      "loss": 1.9794,
      "step": 210365
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.4873992204666138,
      "learning_rate": 1.1126973155889085e-05,
      "loss": 1.9932,
      "step": 210366
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.4488056898117065,
      "learning_rate": 1.1125869457197178e-05,
      "loss": 1.9849,
      "step": 210367
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.42807137966156,
      "learning_rate": 1.1124765812212476e-05,
      "loss": 1.9351,
      "step": 210368
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.423495888710022,
      "learning_rate": 1.112366222093508e-05,
      "loss": 1.9495,
      "step": 210369
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.4262218475341797,
      "learning_rate": 1.112255868336529e-05,
      "loss": 1.9476,
      "step": 210370
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.5556174516677856,
      "learning_rate": 1.1121455199503237e-05,
      "loss": 1.9306,
      "step": 210371
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.3719977140426636,
      "learning_rate": 1.1120351769349223e-05,
      "loss": 1.947,
      "step": 210372
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.46334707736969,
      "learning_rate": 1.1119248392903346e-05,
      "loss": 2.0251,
      "step": 210373
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.4941694736480713,
      "learning_rate": 1.1118145070165874e-05,
      "loss": 2.0127,
      "step": 210374
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.4080257415771484,
      "learning_rate": 1.1117041801136939e-05,
      "loss": 2.0547,
      "step": 210375
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.4438166618347168,
      "learning_rate": 1.1115938585816842e-05,
      "loss": 1.9716,
      "step": 210376
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.4518814086914062,
      "learning_rate": 1.1114835424205714e-05,
      "loss": 1.9799,
      "step": 210377
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.5974736213684082,
      "learning_rate": 1.1113732316303825e-05,
      "loss": 2.0018,
      "step": 210378
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.4362525939941406,
      "learning_rate": 1.1112629262111373e-05,
      "loss": 2.2032,
      "step": 210379
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.540675163269043,
      "learning_rate": 1.1111526261628457e-05,
      "loss": 1.8587,
      "step": 210380
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.6065700054168701,
      "learning_rate": 1.111042331485541e-05,
      "loss": 1.7654,
      "step": 210381
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.4691110849380493,
      "learning_rate": 1.1109320421792401e-05,
      "loss": 2.0107,
      "step": 210382
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.3678652048110962,
      "learning_rate": 1.1108217582439593e-05,
      "loss": 1.9281,
      "step": 210383
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.4544966220855713,
      "learning_rate": 1.1107114796797257e-05,
      "loss": 1.8794,
      "step": 210384
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.4763590097427368,
      "learning_rate": 1.1106012064865522e-05,
      "loss": 1.8375,
      "step": 210385
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.4812296628952026,
      "learning_rate": 1.1104909386644622e-05,
      "loss": 2.0165,
      "step": 210386
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.841662883758545,
      "learning_rate": 1.1103806762134827e-05,
      "loss": 2.0962,
      "step": 210387
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.5223556756973267,
      "learning_rate": 1.1102704191336265e-05,
      "loss": 1.9102,
      "step": 210388
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.524982213973999,
      "learning_rate": 1.1101601674249139e-05,
      "loss": 2.2461,
      "step": 210389
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.3690056800842285,
      "learning_rate": 1.1100499210873715e-05,
      "loss": 1.8344,
      "step": 210390
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.6091958284378052,
      "learning_rate": 1.1099396801210159e-05,
      "loss": 2.0388,
      "step": 210391
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.4985144138336182,
      "learning_rate": 1.1098294445258637e-05,
      "loss": 1.9244,
      "step": 210392
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.695846438407898,
      "learning_rate": 1.1097192143019418e-05,
      "loss": 1.9356,
      "step": 210393
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.4309911727905273,
      "learning_rate": 1.1096089894492666e-05,
      "loss": 2.0862,
      "step": 210394
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.6359570026397705,
      "learning_rate": 1.1094987699678648e-05,
      "loss": 2.1042,
      "step": 210395
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.5272403955459595,
      "learning_rate": 1.1093885558577498e-05,
      "loss": 2.1505,
      "step": 210396
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.482055425643921,
      "learning_rate": 1.1092783471189482e-05,
      "loss": 2.103,
      "step": 210397
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.5465543270111084,
      "learning_rate": 1.1091681437514698e-05,
      "loss": 1.9207,
      "step": 210398
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.369309663772583,
      "learning_rate": 1.1090579457553484e-05,
      "loss": 2.0668,
      "step": 210399
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.5296847820281982,
      "learning_rate": 1.1089477531305968e-05,
      "loss": 2.1509,
      "step": 210400
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.3565068244934082,
      "learning_rate": 1.1088375658772386e-05,
      "loss": 2.0362,
      "step": 210401
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.434612512588501,
      "learning_rate": 1.108727383995287e-05,
      "loss": 1.8901,
      "step": 210402
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.541675329208374,
      "learning_rate": 1.1086172074847788e-05,
      "loss": 2.0601,
      "step": 210403
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.4854016304016113,
      "learning_rate": 1.1085070363457172e-05,
      "loss": 1.8938,
      "step": 210404
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.3807034492492676,
      "learning_rate": 1.1083968705781321e-05,
      "loss": 1.8727,
      "step": 210405
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.6585935354232788,
      "learning_rate": 1.108286710182037e-05,
      "loss": 2.2031,
      "step": 210406
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.5275752544403076,
      "learning_rate": 1.1081765551574584e-05,
      "loss": 2.0332,
      "step": 210407
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.4840750694274902,
      "learning_rate": 1.1080664055044164e-05,
      "loss": 1.9789,
      "step": 210408
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.4099828004837036,
      "learning_rate": 1.1079562612229342e-05,
      "loss": 1.9512,
      "step": 210409
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.475597620010376,
      "learning_rate": 1.1078461223130219e-05,
      "loss": 1.8836,
      "step": 210410
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.4886997938156128,
      "learning_rate": 1.1077359887747095e-05,
      "loss": 1.966,
      "step": 210411
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.4323176145553589,
      "learning_rate": 1.1076258606080101e-05,
      "loss": 1.8347,
      "step": 210412
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.5425676107406616,
      "learning_rate": 1.107515737812954e-05,
      "loss": 2.0196,
      "step": 210413
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.583588719367981,
      "learning_rate": 1.1074056203895509e-05,
      "loss": 2.0055,
      "step": 210414
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.5110843181610107,
      "learning_rate": 1.1072955083378343e-05,
      "loss": 1.8475,
      "step": 210415
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.424142599105835,
      "learning_rate": 1.1071854016578107e-05,
      "loss": 2.0353,
      "step": 210416
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.4540258646011353,
      "learning_rate": 1.1070753003495102e-05,
      "loss": 1.9952,
      "step": 210417
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.6032073497772217,
      "learning_rate": 1.1069652044129462e-05,
      "loss": 1.9263,
      "step": 210418
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.6418356895446777,
      "learning_rate": 1.1068551138481452e-05,
      "loss": 2.106,
      "step": 210419
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.540611743927002,
      "learning_rate": 1.1067450286551206e-05,
      "loss": 1.9668,
      "step": 210420
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.4711532592773438,
      "learning_rate": 1.1066349488339055e-05,
      "loss": 1.88,
      "step": 210421
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.4489490985870361,
      "learning_rate": 1.1065248743845068e-05,
      "loss": 2.094,
      "step": 210422
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.3869837522506714,
      "learning_rate": 1.1064148053069544e-05,
      "loss": 2.0643,
      "step": 210423
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.4932799339294434,
      "learning_rate": 1.1063047416012583e-05,
      "loss": 2.0451,
      "step": 210424
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.7179515361785889,
      "learning_rate": 1.1061946832674518e-05,
      "loss": 1.85,
      "step": 210425
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.518692970275879,
      "learning_rate": 1.1060846303055448e-05,
      "loss": 1.9976,
      "step": 210426
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.4896018505096436,
      "learning_rate": 1.1059745827155674e-05,
      "loss": 1.9624,
      "step": 210427
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.3620014190673828,
      "learning_rate": 1.1058645404975298e-05,
      "loss": 1.7781,
      "step": 210428
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.5169564485549927,
      "learning_rate": 1.1057545036514615e-05,
      "loss": 1.9825,
      "step": 210429
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.4842555522918701,
      "learning_rate": 1.1056444721773727e-05,
      "loss": 1.9282,
      "step": 210430
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.5519487857818604,
      "learning_rate": 1.1055344460752968e-05,
      "loss": 1.8493,
      "step": 210431
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.442225456237793,
      "learning_rate": 1.1054244253452404e-05,
      "loss": 1.8759,
      "step": 210432
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.476777195930481,
      "learning_rate": 1.1053144099872403e-05,
      "loss": 1.9079,
      "step": 210433
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.3482905626296997,
      "learning_rate": 1.1052044000013028e-05,
      "loss": 2.0134,
      "step": 210434
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.5670545101165771,
      "learning_rate": 1.1050943953874514e-05,
      "loss": 2.1132,
      "step": 210435
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.4188354015350342,
      "learning_rate": 1.1049843961457095e-05,
      "loss": 2.181,
      "step": 210436
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.6489893198013306,
      "learning_rate": 1.104874402276097e-05,
      "loss": 1.8266,
      "step": 210437
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.4361048936843872,
      "learning_rate": 1.1047644137786305e-05,
      "loss": 1.7726,
      "step": 210438
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.4507848024368286,
      "learning_rate": 1.1046544306533433e-05,
      "loss": 1.9206,
      "step": 210439
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.5352694988250732,
      "learning_rate": 1.1045444529002389e-05,
      "loss": 1.8289,
      "step": 210440
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.4834822416305542,
      "learning_rate": 1.1044344805193473e-05,
      "loss": 2.0202,
      "step": 210441
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.4884421825408936,
      "learning_rate": 1.1043245135106815e-05,
      "loss": 1.9327,
      "step": 210442
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.4648935794830322,
      "learning_rate": 1.1042145518742718e-05,
      "loss": 1.9636,
      "step": 210443
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.4449031352996826,
      "learning_rate": 1.1041045956101314e-05,
      "loss": 1.8199,
      "step": 210444
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.6400991678237915,
      "learning_rate": 1.1039946447182901e-05,
      "loss": 2.028,
      "step": 210445
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.4194351434707642,
      "learning_rate": 1.1038846991987548e-05,
      "loss": 2.002,
      "step": 210446
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.4915671348571777,
      "learning_rate": 1.1037747590515588e-05,
      "loss": 1.9855,
      "step": 210447
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.4657100439071655,
      "learning_rate": 1.1036648242767088e-05,
      "loss": 1.9987,
      "step": 210448
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.4684925079345703,
      "learning_rate": 1.103554894874238e-05,
      "loss": 1.9788,
      "step": 210449
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.499981164932251,
      "learning_rate": 1.1034449708441596e-05,
      "loss": 2.0351,
      "step": 210450
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.4782836437225342,
      "learning_rate": 1.1033350521865037e-05,
      "loss": 1.931,
      "step": 210451
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.5159673690795898,
      "learning_rate": 1.1032251389012736e-05,
      "loss": 2.0218,
      "step": 210452
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.4931166172027588,
      "learning_rate": 1.1031152309885028e-05,
      "loss": 2.1534,
      "step": 210453
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.576785922050476,
      "learning_rate": 1.1030053284482076e-05,
      "loss": 1.9254,
      "step": 210454
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.3789111375808716,
      "learning_rate": 1.1028954312804117e-05,
      "loss": 2.1082,
      "step": 210455
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.5227174758911133,
      "learning_rate": 1.1027855394851281e-05,
      "loss": 1.9067,
      "step": 210456
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.497720718383789,
      "learning_rate": 1.102675653062387e-05,
      "loss": 2.216,
      "step": 210457
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.5060901641845703,
      "learning_rate": 1.1025657720122017e-05,
      "loss": 1.9195,
      "step": 210458
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.4231020212173462,
      "learning_rate": 1.1024558963345986e-05,
      "loss": 1.8889,
      "step": 210459
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.6442463397979736,
      "learning_rate": 1.102346026029588e-05,
      "loss": 2.0279,
      "step": 210460
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.4658482074737549,
      "learning_rate": 1.102236161097203e-05,
      "loss": 2.0463,
      "step": 210461
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.4757579565048218,
      "learning_rate": 1.1021263015374505e-05,
      "loss": 1.9649,
      "step": 210462
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.2789263725280762,
      "learning_rate": 1.1020164473503667e-05,
      "loss": 1.8872,
      "step": 210463
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.3369003534317017,
      "learning_rate": 1.1019065985359587e-05,
      "loss": 1.7938,
      "step": 210464
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.5865541696548462,
      "learning_rate": 1.1017967550942497e-05,
      "loss": 1.8752,
      "step": 210465
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.512213945388794,
      "learning_rate": 1.1016869170252662e-05,
      "loss": 2.0436,
      "step": 210466
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.4794244766235352,
      "learning_rate": 1.1015770843290251e-05,
      "loss": 1.9597,
      "step": 210467
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.5972607135772705,
      "learning_rate": 1.1014672570055426e-05,
      "loss": 2.0518,
      "step": 210468
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.4757113456726074,
      "learning_rate": 1.1013574350548459e-05,
      "loss": 1.9093,
      "step": 210469
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.5196423530578613,
      "learning_rate": 1.1012476184769547e-05,
      "loss": 2.0287,
      "step": 210470
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.4864373207092285,
      "learning_rate": 1.101137807271879e-05,
      "loss": 1.9685,
      "step": 210471
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.5724477767944336,
      "learning_rate": 1.1010280014396555e-05,
      "loss": 1.8699,
      "step": 210472
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.4839247465133667,
      "learning_rate": 1.1009182009802941e-05,
      "loss": 1.9578,
      "step": 210473
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.4172892570495605,
      "learning_rate": 1.1008084058938116e-05,
      "loss": 1.9508,
      "step": 210474
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.4115513563156128,
      "learning_rate": 1.1006986161802412e-05,
      "loss": 1.9948,
      "step": 210475
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.3820422887802124,
      "learning_rate": 1.1005888318395961e-05,
      "loss": 2.0162,
      "step": 210476
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.526003360748291,
      "learning_rate": 1.1004790528718932e-05,
      "loss": 1.992,
      "step": 210477
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.4777897596359253,
      "learning_rate": 1.100369279277159e-05,
      "loss": 1.9149,
      "step": 210478
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.4122456312179565,
      "learning_rate": 1.1002595110554102e-05,
      "loss": 1.7509,
      "step": 210479
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.5715994834899902,
      "learning_rate": 1.1001497482066701e-05,
      "loss": 2.2586,
      "step": 210480
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.558706521987915,
      "learning_rate": 1.1000399907309587e-05,
      "loss": 1.975,
      "step": 210481
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.449009895324707,
      "learning_rate": 1.099930238628296e-05,
      "loss": 1.8707,
      "step": 210482
    },
    {
      "epoch": 2.74,
      "grad_norm": 2.0131468772888184,
      "learning_rate": 1.0998204918986985e-05,
      "loss": 1.7269,
      "step": 210483
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.401291847229004,
      "learning_rate": 1.0997107505421931e-05,
      "loss": 1.9986,
      "step": 210484
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.2920079231262207,
      "learning_rate": 1.099601014558793e-05,
      "loss": 1.8464,
      "step": 210485
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.5031176805496216,
      "learning_rate": 1.0994912839485248e-05,
      "loss": 2.019,
      "step": 210486
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.602297306060791,
      "learning_rate": 1.0993815587114085e-05,
      "loss": 2.2223,
      "step": 210487
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.5695867538452148,
      "learning_rate": 1.0992718388474609e-05,
      "loss": 1.9035,
      "step": 210488
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.426424264907837,
      "learning_rate": 1.099162124356705e-05,
      "loss": 1.9095,
      "step": 210489
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.5038501024246216,
      "learning_rate": 1.0990524152391612e-05,
      "loss": 1.9871,
      "step": 210490
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.3031326532363892,
      "learning_rate": 1.0989427114948423e-05,
      "loss": 1.5265,
      "step": 210491
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.4342483282089233,
      "learning_rate": 1.0988330131237821e-05,
      "loss": 1.9163,
      "step": 210492
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.6047310829162598,
      "learning_rate": 1.0987233201259937e-05,
      "loss": 2.0021,
      "step": 210493
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.685939073562622,
      "learning_rate": 1.0986136325015005e-05,
      "loss": 2.0349,
      "step": 210494
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.4017541408538818,
      "learning_rate": 1.0985039502503157e-05,
      "loss": 1.7634,
      "step": 210495
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.4509528875350952,
      "learning_rate": 1.0983942733724692e-05,
      "loss": 1.932,
      "step": 210496
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.4930952787399292,
      "learning_rate": 1.0982846018679714e-05,
      "loss": 1.9654,
      "step": 210497
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.321341872215271,
      "learning_rate": 1.0981749357368519e-05,
      "loss": 1.8669,
      "step": 210498
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.4619237184524536,
      "learning_rate": 1.098065274979124e-05,
      "loss": 1.8649,
      "step": 210499
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.4042105674743652,
      "learning_rate": 1.0979556195948181e-05,
      "loss": 1.9217,
      "step": 210500
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.3360000848770142,
      "learning_rate": 1.0978459695839404e-05,
      "loss": 1.8631,
      "step": 210501
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.3503133058547974,
      "learning_rate": 1.0977363249465243e-05,
      "loss": 2.0527,
      "step": 210502
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.420333981513977,
      "learning_rate": 1.0976266856825799e-05,
      "loss": 1.8218,
      "step": 210503
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.4762364625930786,
      "learning_rate": 1.0975170517921339e-05,
      "loss": 1.9799,
      "step": 210504
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.6994072198867798,
      "learning_rate": 1.0974074232752027e-05,
      "loss": 2.0538,
      "step": 210505
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.5572787523269653,
      "learning_rate": 1.0972978001318168e-05,
      "loss": 2.0537,
      "step": 210506
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.4585785865783691,
      "learning_rate": 1.0971881823619822e-05,
      "loss": 1.9886,
      "step": 210507
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.4742064476013184,
      "learning_rate": 1.0970785699657292e-05,
      "loss": 1.8622,
      "step": 210508
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.375901699066162,
      "learning_rate": 1.0969689629430711e-05,
      "loss": 1.9684,
      "step": 210509
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.5302528142929077,
      "learning_rate": 1.0968593612940347e-05,
      "loss": 2.0206,
      "step": 210510
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.3542249202728271,
      "learning_rate": 1.096749765018633e-05,
      "loss": 1.9117,
      "step": 210511
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.4270052909851074,
      "learning_rate": 1.0966401741168996e-05,
      "loss": 1.8473,
      "step": 210512
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.6054431200027466,
      "learning_rate": 1.0965305885888409e-05,
      "loss": 2.2501,
      "step": 210513
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.5106512308120728,
      "learning_rate": 1.0964210084344837e-05,
      "loss": 2.0181,
      "step": 210514
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.5614335536956787,
      "learning_rate": 1.096311433653848e-05,
      "loss": 2.0395,
      "step": 210515
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.6310698986053467,
      "learning_rate": 1.0962018642469539e-05,
      "loss": 2.0158,
      "step": 210516
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.543618083000183,
      "learning_rate": 1.0960923002138177e-05,
      "loss": 1.9233,
      "step": 210517
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.5237058401107788,
      "learning_rate": 1.095982741554473e-05,
      "loss": 1.9945,
      "step": 210518
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.3829326629638672,
      "learning_rate": 1.0958731882689198e-05,
      "loss": 1.9779,
      "step": 210519
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.477156400680542,
      "learning_rate": 1.0957636403571946e-05,
      "loss": 2.1249,
      "step": 210520
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.4833173751831055,
      "learning_rate": 1.0956540978193108e-05,
      "loss": 2.0386,
      "step": 210521
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.4139726161956787,
      "learning_rate": 1.0955445606552916e-05,
      "loss": 1.9871,
      "step": 210522
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.477211594581604,
      "learning_rate": 1.095435028865157e-05,
      "loss": 1.9694,
      "step": 210523
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.3555433750152588,
      "learning_rate": 1.0953255024489272e-05,
      "loss": 2.1977,
      "step": 210524
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.4985097646713257,
      "learning_rate": 1.095215981406622e-05,
      "loss": 2.0657,
      "step": 210525
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.6586673259735107,
      "learning_rate": 1.0951064657382613e-05,
      "loss": 1.9446,
      "step": 210526
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.578041434288025,
      "learning_rate": 1.094996955443862e-05,
      "loss": 1.9499,
      "step": 210527
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.5655500888824463,
      "learning_rate": 1.094887450523454e-05,
      "loss": 1.9079,
      "step": 210528
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.4811066389083862,
      "learning_rate": 1.0947779509770472e-05,
      "loss": 1.9498,
      "step": 210529
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.4236598014831543,
      "learning_rate": 1.0946684568046682e-05,
      "loss": 1.8242,
      "step": 210530
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.4381628036499023,
      "learning_rate": 1.0945589680063372e-05,
      "loss": 1.9767,
      "step": 210531
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.5858453512191772,
      "learning_rate": 1.0944494845820772e-05,
      "loss": 2.0871,
      "step": 210532
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.4088023900985718,
      "learning_rate": 1.0943400065318952e-05,
      "loss": 1.9446,
      "step": 210533
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.5019439458847046,
      "learning_rate": 1.0942305338558277e-05,
      "loss": 2.0138,
      "step": 210534
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.3528960943222046,
      "learning_rate": 1.0941210665538847e-05,
      "loss": 1.9574,
      "step": 210535
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.4107730388641357,
      "learning_rate": 1.0940116046260961e-05,
      "loss": 2.1325,
      "step": 210536
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.489035725593567,
      "learning_rate": 1.0939021480724719e-05,
      "loss": 2.1557,
      "step": 210537
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.6076301336288452,
      "learning_rate": 1.0937926968930389e-05,
      "loss": 2.0853,
      "step": 210538
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.6712948083877563,
      "learning_rate": 1.0936832510878135e-05,
      "loss": 2.069,
      "step": 210539
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.414085865020752,
      "learning_rate": 1.0935738106568193e-05,
      "loss": 1.863,
      "step": 210540
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.7042425870895386,
      "learning_rate": 1.0934643756000727e-05,
      "loss": 1.9842,
      "step": 210541
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.6269056797027588,
      "learning_rate": 1.0933549459176005e-05,
      "loss": 1.8402,
      "step": 210542
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.5844879150390625,
      "learning_rate": 1.0932455216094194e-05,
      "loss": 2.0986,
      "step": 210543
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.4314793348312378,
      "learning_rate": 1.0931361026755492e-05,
      "loss": 2.0,
      "step": 210544
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.359621286392212,
      "learning_rate": 1.0930266891160067e-05,
      "loss": 2.1866,
      "step": 210545
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.4068193435668945,
      "learning_rate": 1.0929172809308218e-05,
      "loss": 2.025,
      "step": 210546
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.5229073762893677,
      "learning_rate": 1.0928078781200045e-05,
      "loss": 2.2195,
      "step": 210547
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.5118554830551147,
      "learning_rate": 1.0926984806835815e-05,
      "loss": 1.8415,
      "step": 210548
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.4175058603286743,
      "learning_rate": 1.092589088621576e-05,
      "loss": 2.1777,
      "step": 210549
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.4253085851669312,
      "learning_rate": 1.092479701933998e-05,
      "loss": 1.9289,
      "step": 210550
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.609389066696167,
      "learning_rate": 1.0923703206208745e-05,
      "loss": 2.175,
      "step": 210551
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.3732706308364868,
      "learning_rate": 1.0922609446822284e-05,
      "loss": 1.9223,
      "step": 210552
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.5019651651382446,
      "learning_rate": 1.0921515741180698e-05,
      "loss": 1.7829,
      "step": 210553
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.411237359046936,
      "learning_rate": 1.092042208928432e-05,
      "loss": 2.0958,
      "step": 210554
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.4818354845046997,
      "learning_rate": 1.0919328491133283e-05,
      "loss": 1.8946,
      "step": 210555
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.4856109619140625,
      "learning_rate": 1.0918234946727788e-05,
      "loss": 2.0595,
      "step": 210556
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.3713070154190063,
      "learning_rate": 1.0917141456068068e-05,
      "loss": 1.8018,
      "step": 210557
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.5769283771514893,
      "learning_rate": 1.0916048019154288e-05,
      "loss": 2.0219,
      "step": 210558
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.516784906387329,
      "learning_rate": 1.091495463598665e-05,
      "loss": 1.9878,
      "step": 210559
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.4166414737701416,
      "learning_rate": 1.0913861306565419e-05,
      "loss": 2.0491,
      "step": 210560
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.5303672552108765,
      "learning_rate": 1.0912768030890762e-05,
      "loss": 2.0388,
      "step": 210561
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.3861156702041626,
      "learning_rate": 1.0911674808962845e-05,
      "loss": 1.9863,
      "step": 210562
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.5752370357513428,
      "learning_rate": 1.0910581640781934e-05,
      "loss": 1.8925,
      "step": 210563
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.3673518896102905,
      "learning_rate": 1.0909488526348164e-05,
      "loss": 1.9166,
      "step": 210564
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.664503574371338,
      "learning_rate": 1.0908395465661835e-05,
      "loss": 1.9053,
      "step": 210565
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.4877089262008667,
      "learning_rate": 1.0907302458723078e-05,
      "loss": 1.8832,
      "step": 210566
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.6073713302612305,
      "learning_rate": 1.0906209505532092e-05,
      "loss": 2.007,
      "step": 210567
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.449047565460205,
      "learning_rate": 1.0905116606089081e-05,
      "loss": 1.9484,
      "step": 210568
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.5491544008255005,
      "learning_rate": 1.090402376039431e-05,
      "loss": 1.9794,
      "step": 210569
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.438740849494934,
      "learning_rate": 1.0902930968447875e-05,
      "loss": 1.9681,
      "step": 210570
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.3625383377075195,
      "learning_rate": 1.0901838230250081e-05,
      "loss": 2.0441,
      "step": 210571
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.5572019815444946,
      "learning_rate": 1.0900745545801126e-05,
      "loss": 1.978,
      "step": 210572
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.556054949760437,
      "learning_rate": 1.0899652915101142e-05,
      "loss": 2.0535,
      "step": 210573
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.3997678756713867,
      "learning_rate": 1.0898560338150364e-05,
      "loss": 1.7533,
      "step": 210574
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.4798524379730225,
      "learning_rate": 1.0897467814949023e-05,
      "loss": 1.9971,
      "step": 210575
    },
    {
      "epoch": 2.74,
      "grad_norm": 2.613816022872925,
      "learning_rate": 1.0896375345497254e-05,
      "loss": 1.9679,
      "step": 210576
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.604590892791748,
      "learning_rate": 1.0895282929795357e-05,
      "loss": 2.0827,
      "step": 210577
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.9735352993011475,
      "learning_rate": 1.089419056784343e-05,
      "loss": 1.7573,
      "step": 210578
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.4597285985946655,
      "learning_rate": 1.0893098259641841e-05,
      "loss": 2.0379,
      "step": 210579
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.440869688987732,
      "learning_rate": 1.0892006005190557e-05,
      "loss": 1.9765,
      "step": 210580
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.2883144617080688,
      "learning_rate": 1.0890913804489975e-05,
      "loss": 1.7925,
      "step": 210581
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.607218623161316,
      "learning_rate": 1.0889821657540199e-05,
      "loss": 2.0246,
      "step": 210582
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.4774287939071655,
      "learning_rate": 1.0888729564341492e-05,
      "loss": 2.0422,
      "step": 210583
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.4492396116256714,
      "learning_rate": 1.0887637524893988e-05,
      "loss": 1.773,
      "step": 210584
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.4468801021575928,
      "learning_rate": 1.0886545539197989e-05,
      "loss": 1.9665,
      "step": 210585
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.5596108436584473,
      "learning_rate": 1.088545360725359e-05,
      "loss": 1.8807,
      "step": 210586
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.606329083442688,
      "learning_rate": 1.0884361729061063e-05,
      "loss": 2.0682,
      "step": 210587
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.3766322135925293,
      "learning_rate": 1.0883269904620572e-05,
      "loss": 2.0101,
      "step": 210588
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.4786759614944458,
      "learning_rate": 1.088217813393235e-05,
      "loss": 2.1245,
      "step": 210589
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.4641554355621338,
      "learning_rate": 1.0881086416996564e-05,
      "loss": 2.2613,
      "step": 210590
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.567818522453308,
      "learning_rate": 1.0879994753813481e-05,
      "loss": 2.0243,
      "step": 210591
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.4261616468429565,
      "learning_rate": 1.0878903144383266e-05,
      "loss": 2.0437,
      "step": 210592
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.529449701309204,
      "learning_rate": 1.0877811588706121e-05,
      "loss": 2.193,
      "step": 210593
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.513283371925354,
      "learning_rate": 1.0876720086782208e-05,
      "loss": 1.9198,
      "step": 210594
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.4211851358413696,
      "learning_rate": 1.08756286386118e-05,
      "loss": 2.0272,
      "step": 210595
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.6351920366287231,
      "learning_rate": 1.0874537244195025e-05,
      "loss": 2.0618,
      "step": 210596
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.5188918113708496,
      "learning_rate": 1.0873445903532186e-05,
      "loss": 1.9513,
      "step": 210597
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.4267743825912476,
      "learning_rate": 1.0872354616623447e-05,
      "loss": 1.92,
      "step": 210598
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.3953081369400024,
      "learning_rate": 1.0871263383468975e-05,
      "loss": 2.1437,
      "step": 210599
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.4943230152130127,
      "learning_rate": 1.087017220406894e-05,
      "loss": 1.9589,
      "step": 210600
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.5916913747787476,
      "learning_rate": 1.0869081078423669e-05,
      "loss": 2.047,
      "step": 210601
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.4241560697555542,
      "learning_rate": 1.0867990006533234e-05,
      "loss": 1.9499,
      "step": 210602
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.4698892831802368,
      "learning_rate": 1.0866898988397932e-05,
      "loss": 1.8716,
      "step": 210603
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.427724838256836,
      "learning_rate": 1.086580802401793e-05,
      "loss": 1.6582,
      "step": 210604
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.4628984928131104,
      "learning_rate": 1.0864717113393462e-05,
      "loss": 1.9254,
      "step": 210605
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.517929196357727,
      "learning_rate": 1.0863626256524628e-05,
      "loss": 2.1038,
      "step": 210606
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.6424986124038696,
      "learning_rate": 1.0862535453411758e-05,
      "loss": 2.0375,
      "step": 210607
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.478137493133545,
      "learning_rate": 1.0861444704054956e-05,
      "loss": 2.13,
      "step": 210608
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.46943199634552,
      "learning_rate": 1.086035400845452e-05,
      "loss": 2.1177,
      "step": 210609
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.462814211845398,
      "learning_rate": 1.0859263366610583e-05,
      "loss": 1.7899,
      "step": 210610
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.4034146070480347,
      "learning_rate": 1.0858172778523378e-05,
      "loss": 1.7036,
      "step": 210611
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.501936912536621,
      "learning_rate": 1.0857082244193072e-05,
      "loss": 1.6853,
      "step": 210612
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.755935788154602,
      "learning_rate": 1.0855991763619898e-05,
      "loss": 2.2323,
      "step": 210613
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.6996906995773315,
      "learning_rate": 1.0854901336804057e-05,
      "loss": 2.0831,
      "step": 210614
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.3448201417922974,
      "learning_rate": 1.0853810963745746e-05,
      "loss": 2.082,
      "step": 210615
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.416481375694275,
      "learning_rate": 1.08527206444452e-05,
      "loss": 1.8782,
      "step": 210616
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.4906741380691528,
      "learning_rate": 1.0851630378902554e-05,
      "loss": 2.1526,
      "step": 210617
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.5040980577468872,
      "learning_rate": 1.0850540167118039e-05,
      "loss": 1.8898,
      "step": 210618
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.5477344989776611,
      "learning_rate": 1.084945000909192e-05,
      "loss": 2.0413,
      "step": 210619
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.4998445510864258,
      "learning_rate": 1.0848359904824266e-05,
      "loss": 1.975,
      "step": 210620
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.3609349727630615,
      "learning_rate": 1.0847269854315444e-05,
      "loss": 1.9541,
      "step": 210621
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.4460291862487793,
      "learning_rate": 1.0846179857565552e-05,
      "loss": 1.8214,
      "step": 210622
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.7139207124710083,
      "learning_rate": 1.0845089914574789e-05,
      "loss": 2.0779,
      "step": 210623
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.4050897359848022,
      "learning_rate": 1.0844000025343358e-05,
      "loss": 1.7512,
      "step": 210624
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.5632693767547607,
      "learning_rate": 1.0842910189871556e-05,
      "loss": 2.007,
      "step": 210625
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.466837763786316,
      "learning_rate": 1.0841820408159451e-05,
      "loss": 1.8087,
      "step": 210626
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.5163553953170776,
      "learning_rate": 1.0840730680207343e-05,
      "loss": 1.9156,
      "step": 210627
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.392175316810608,
      "learning_rate": 1.0839641006015398e-05,
      "loss": 1.7776,
      "step": 210628
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.5468688011169434,
      "learning_rate": 1.0838551385583849e-05,
      "loss": 2.0579,
      "step": 210629
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.682560920715332,
      "learning_rate": 1.0837461818912796e-05,
      "loss": 2.0764,
      "step": 210630
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.446243405342102,
      "learning_rate": 1.0836372306002572e-05,
      "loss": 1.9782,
      "step": 210631
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.4651440382003784,
      "learning_rate": 1.083528284685331e-05,
      "loss": 1.998,
      "step": 210632
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.695820927619934,
      "learning_rate": 1.0834193441465244e-05,
      "loss": 2.0389,
      "step": 210633
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.5519845485687256,
      "learning_rate": 1.0833104089838541e-05,
      "loss": 1.9692,
      "step": 210634
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.5004266500473022,
      "learning_rate": 1.0832014791973465e-05,
      "loss": 1.8408,
      "step": 210635
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.4305756092071533,
      "learning_rate": 1.0830925547870118e-05,
      "loss": 1.8677,
      "step": 210636
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.4311739206314087,
      "learning_rate": 1.0829836357528798e-05,
      "loss": 1.9418,
      "step": 210637
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.436823844909668,
      "learning_rate": 1.0828747220949607e-05,
      "loss": 1.9997,
      "step": 210638
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.4909268617630005,
      "learning_rate": 1.0827658138132878e-05,
      "loss": 1.9536,
      "step": 210639
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.4486029148101807,
      "learning_rate": 1.0826569109078742e-05,
      "loss": 1.9868,
      "step": 210640
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.304681420326233,
      "learning_rate": 1.0825480133787367e-05,
      "loss": 1.6782,
      "step": 210641
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.421710729598999,
      "learning_rate": 1.0824391212259021e-05,
      "loss": 2.2782,
      "step": 210642
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.6282116174697876,
      "learning_rate": 1.0823302344493901e-05,
      "loss": 2.0338,
      "step": 210643
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.4872397184371948,
      "learning_rate": 1.082221353049214e-05,
      "loss": 1.8831,
      "step": 210644
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.5822844505310059,
      "learning_rate": 1.0821124770254008e-05,
      "loss": 1.9579,
      "step": 210645
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.5510329008102417,
      "learning_rate": 1.0820036063779702e-05,
      "loss": 1.9641,
      "step": 210646
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.4386909008026123,
      "learning_rate": 1.081894741106939e-05,
      "loss": 1.8668,
      "step": 210647
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.4955129623413086,
      "learning_rate": 1.0817858812123303e-05,
      "loss": 1.9364,
      "step": 210648
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.4747638702392578,
      "learning_rate": 1.0816770266941643e-05,
      "loss": 1.9225,
      "step": 210649
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.5086814165115356,
      "learning_rate": 1.0815681775524576e-05,
      "loss": 2.1997,
      "step": 210650
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.4349310398101807,
      "learning_rate": 1.0814593337872368e-05,
      "loss": 1.9318,
      "step": 210651
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.8511990308761597,
      "learning_rate": 1.0813504953985153e-05,
      "loss": 2.1619,
      "step": 210652
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.43544602394104,
      "learning_rate": 1.0812416623863163e-05,
      "loss": 1.7725,
      "step": 210653
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.4992355108261108,
      "learning_rate": 1.0811328347506631e-05,
      "loss": 1.768,
      "step": 210654
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.4435546398162842,
      "learning_rate": 1.0810240124915726e-05,
      "loss": 1.8706,
      "step": 210655
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.4967190027236938,
      "learning_rate": 1.0809151956090644e-05,
      "loss": 2.0646,
      "step": 210656
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.451738953590393,
      "learning_rate": 1.0808063841031622e-05,
      "loss": 1.8602,
      "step": 210657
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.713260531425476,
      "learning_rate": 1.0806975779738824e-05,
      "loss": 1.8921,
      "step": 210658
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.4965778589248657,
      "learning_rate": 1.0805887772212484e-05,
      "loss": 2.0118,
      "step": 210659
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.4916703701019287,
      "learning_rate": 1.0804799818452769e-05,
      "loss": 1.9653,
      "step": 210660
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.4658546447753906,
      "learning_rate": 1.080371191845988e-05,
      "loss": 1.9581,
      "step": 210661
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.4245812892913818,
      "learning_rate": 1.0802624072234078e-05,
      "loss": 2.017,
      "step": 210662
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.3447785377502441,
      "learning_rate": 1.0801536279775503e-05,
      "loss": 2.0131,
      "step": 210663
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.5850558280944824,
      "learning_rate": 1.0800448541084417e-05,
      "loss": 2.0949,
      "step": 210664
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.3365737199783325,
      "learning_rate": 1.0799360856160989e-05,
      "loss": 1.8209,
      "step": 210665
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.486201524734497,
      "learning_rate": 1.0798273225005382e-05,
      "loss": 2.0155,
      "step": 210666
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.5033437013626099,
      "learning_rate": 1.0797185647617835e-05,
      "loss": 2.086,
      "step": 210667
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.5102659463882446,
      "learning_rate": 1.0796098123998576e-05,
      "loss": 2.0404,
      "step": 210668
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.3316975831985474,
      "learning_rate": 1.079501065414774e-05,
      "loss": 1.8795,
      "step": 210669
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.511675238609314,
      "learning_rate": 1.0793923238065626e-05,
      "loss": 1.9293,
      "step": 210670
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.5233771800994873,
      "learning_rate": 1.079283587575237e-05,
      "loss": 1.9526,
      "step": 210671
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.4609599113464355,
      "learning_rate": 1.079174856720817e-05,
      "loss": 1.8873,
      "step": 210672
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.7564023733139038,
      "learning_rate": 1.0790661312433224e-05,
      "loss": 2.2846,
      "step": 210673
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.4539587497711182,
      "learning_rate": 1.0789574111427768e-05,
      "loss": 2.0774,
      "step": 210674
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.3784070014953613,
      "learning_rate": 1.0788486964191966e-05,
      "loss": 1.9574,
      "step": 210675
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.3936580419540405,
      "learning_rate": 1.0787399870726088e-05,
      "loss": 1.789,
      "step": 210676
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.4054594039916992,
      "learning_rate": 1.0786312831030298e-05,
      "loss": 2.0721,
      "step": 210677
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.4725403785705566,
      "learning_rate": 1.0785225845104761e-05,
      "loss": 1.8919,
      "step": 210678
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.5012531280517578,
      "learning_rate": 1.0784138912949713e-05,
      "loss": 2.0103,
      "step": 210679
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.620257019996643,
      "learning_rate": 1.0783052034565353e-05,
      "loss": 1.8633,
      "step": 210680
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.5732616186141968,
      "learning_rate": 1.0781965209951882e-05,
      "loss": 1.8784,
      "step": 210681
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.423635482788086,
      "learning_rate": 1.0780878439109497e-05,
      "loss": 2.223,
      "step": 210682
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.450152039527893,
      "learning_rate": 1.0779791722038433e-05,
      "loss": 1.8585,
      "step": 210683
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.4805006980895996,
      "learning_rate": 1.0778705058738856e-05,
      "loss": 2.1394,
      "step": 210684
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.367087960243225,
      "learning_rate": 1.0777618449210968e-05,
      "loss": 1.9126,
      "step": 210685
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.4253934621810913,
      "learning_rate": 1.0776531893454999e-05,
      "loss": 1.9692,
      "step": 210686
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.4058806896209717,
      "learning_rate": 1.0775445391471083e-05,
      "loss": 1.9776,
      "step": 210687
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.4270657300949097,
      "learning_rate": 1.077435894325952e-05,
      "loss": 2.0726,
      "step": 210688
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.443184494972229,
      "learning_rate": 1.0773272548820444e-05,
      "loss": 2.0453,
      "step": 210689
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.4367365837097168,
      "learning_rate": 1.0772186208154122e-05,
      "loss": 1.9777,
      "step": 210690
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.4366207122802734,
      "learning_rate": 1.0771099921260618e-05,
      "loss": 1.8306,
      "step": 210691
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.4401214122772217,
      "learning_rate": 1.0770013688140299e-05,
      "loss": 1.8942,
      "step": 210692
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.3952544927597046,
      "learning_rate": 1.0768927508793235e-05,
      "loss": 1.9555,
      "step": 210693
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.4480754137039185,
      "learning_rate": 1.0767841383219755e-05,
      "loss": 1.8536,
      "step": 210694
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.4459072351455688,
      "learning_rate": 1.076675531141996e-05,
      "loss": 1.8702,
      "step": 210695
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.4517457485198975,
      "learning_rate": 1.076566929339412e-05,
      "loss": 1.7434,
      "step": 210696
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.41023588180542,
      "learning_rate": 1.076458332914233e-05,
      "loss": 1.9349,
      "step": 210697
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.5134730339050293,
      "learning_rate": 1.0763497418664924e-05,
      "loss": 2.1505,
      "step": 210698
    },
    {
      "epoch": 2.74,
      "grad_norm": 2.358712911605835,
      "learning_rate": 1.076241156196197e-05,
      "loss": 1.8569,
      "step": 210699
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.5734539031982422,
      "learning_rate": 1.0761325759033834e-05,
      "loss": 1.9926,
      "step": 210700
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.4697405099868774,
      "learning_rate": 1.0760240009880584e-05,
      "loss": 1.8554,
      "step": 210701
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.6085288524627686,
      "learning_rate": 1.0759154314502482e-05,
      "loss": 2.0455,
      "step": 210702
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.5063738822937012,
      "learning_rate": 1.0758068672899667e-05,
      "loss": 2.1143,
      "step": 210703
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.5227789878845215,
      "learning_rate": 1.0756983085072434e-05,
      "loss": 2.0172,
      "step": 210704
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.4772478342056274,
      "learning_rate": 1.075589755102092e-05,
      "loss": 1.8123,
      "step": 210705
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.4740269184112549,
      "learning_rate": 1.0754812070745356e-05,
      "loss": 1.9929,
      "step": 210706
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.380482792854309,
      "learning_rate": 1.075372664424594e-05,
      "loss": 1.9574,
      "step": 210707
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.4913686513900757,
      "learning_rate": 1.0752641271522843e-05,
      "loss": 2.1354,
      "step": 210708
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.4725308418273926,
      "learning_rate": 1.0751555952576296e-05,
      "loss": 1.9708,
      "step": 210709
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.4574867486953735,
      "learning_rate": 1.0750470687406498e-05,
      "loss": 1.9591,
      "step": 210710
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.4430603981018066,
      "learning_rate": 1.0749385476013617e-05,
      "loss": 1.9665,
      "step": 210711
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.5216851234436035,
      "learning_rate": 1.0748300318397918e-05,
      "loss": 1.9868,
      "step": 210712
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.6799155473709106,
      "learning_rate": 1.07472152145596e-05,
      "loss": 2.0073,
      "step": 210713
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.451999545097351,
      "learning_rate": 1.07461301644988e-05,
      "loss": 1.8182,
      "step": 210714
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.414594292640686,
      "learning_rate": 1.0745045168215749e-05,
      "loss": 1.978,
      "step": 210715
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.6889137029647827,
      "learning_rate": 1.074396022571068e-05,
      "loss": 2.0791,
      "step": 210716
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.6601440906524658,
      "learning_rate": 1.0742875336983725e-05,
      "loss": 1.9504,
      "step": 210717
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.5766451358795166,
      "learning_rate": 1.0741790502035153e-05,
      "loss": 1.9623,
      "step": 210718
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.4949811697006226,
      "learning_rate": 1.0740705720865129e-05,
      "loss": 1.9427,
      "step": 210719
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.4597904682159424,
      "learning_rate": 1.0739620993473918e-05,
      "loss": 2.1104,
      "step": 210720
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.443922758102417,
      "learning_rate": 1.0738536319861624e-05,
      "loss": 1.8964,
      "step": 210721
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.4332025051116943,
      "learning_rate": 1.073745170002851e-05,
      "loss": 1.8253,
      "step": 210722
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.4216840267181396,
      "learning_rate": 1.0736367133974743e-05,
      "loss": 2.0813,
      "step": 210723
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.995308518409729,
      "learning_rate": 1.0735282621700592e-05,
      "loss": 2.3302,
      "step": 210724
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.6155266761779785,
      "learning_rate": 1.0734198163206153e-05,
      "loss": 1.9634,
      "step": 210725
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.4837766885757446,
      "learning_rate": 1.0733113758491729e-05,
      "loss": 1.9414,
      "step": 210726
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.4237309694290161,
      "learning_rate": 1.0732029407557486e-05,
      "loss": 1.936,
      "step": 210727
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.417508602142334,
      "learning_rate": 1.0730945110403622e-05,
      "loss": 1.9137,
      "step": 210728
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.4909770488739014,
      "learning_rate": 1.0729860867030271e-05,
      "loss": 1.9704,
      "step": 210729
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.4918245077133179,
      "learning_rate": 1.0728776677437767e-05,
      "loss": 2.142,
      "step": 210730
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.6314184665679932,
      "learning_rate": 1.072769254162621e-05,
      "loss": 1.8693,
      "step": 210731
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.3402179479599,
      "learning_rate": 1.0726608459595832e-05,
      "loss": 2.0496,
      "step": 210732
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.592902660369873,
      "learning_rate": 1.0725524431346866e-05,
      "loss": 1.7697,
      "step": 210733
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.3636314868927002,
      "learning_rate": 1.072444045687948e-05,
      "loss": 1.9679,
      "step": 210734
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.4668549299240112,
      "learning_rate": 1.0723356536193873e-05,
      "loss": 2.046,
      "step": 210735
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.5440571308135986,
      "learning_rate": 1.0722272669290277e-05,
      "loss": 2.0345,
      "step": 210736
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.3873109817504883,
      "learning_rate": 1.0721188856168794e-05,
      "loss": 1.9924,
      "step": 210737
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.481977939605713,
      "learning_rate": 1.0720105096829789e-05,
      "loss": 1.8337,
      "step": 210738
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.4592047929763794,
      "learning_rate": 1.0719021391273364e-05,
      "loss": 1.9109,
      "step": 210739
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.617091178894043,
      "learning_rate": 1.0717937739499683e-05,
      "loss": 2.1963,
      "step": 210740
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.4565690755844116,
      "learning_rate": 1.0716854141509046e-05,
      "loss": 1.9799,
      "step": 210741
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.4339008331298828,
      "learning_rate": 1.0715770597301588e-05,
      "loss": 2.0147,
      "step": 210742
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.422788143157959,
      "learning_rate": 1.0714687106877506e-05,
      "loss": 1.924,
      "step": 210743
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.3995288610458374,
      "learning_rate": 1.071360367023707e-05,
      "loss": 1.8139,
      "step": 210744
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.390990138053894,
      "learning_rate": 1.0712520287380443e-05,
      "loss": 1.7955,
      "step": 210745
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.4165152311325073,
      "learning_rate": 1.0711436958307763e-05,
      "loss": 1.8686,
      "step": 210746
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.551211953163147,
      "learning_rate": 1.0710353683019323e-05,
      "loss": 1.9161,
      "step": 210747
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.6768505573272705,
      "learning_rate": 1.0709270461515263e-05,
      "loss": 2.146,
      "step": 210748
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.4566668272018433,
      "learning_rate": 1.0708187293795844e-05,
      "loss": 1.9971,
      "step": 210749
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.4484238624572754,
      "learning_rate": 1.0707104179861236e-05,
      "loss": 1.8195,
      "step": 210750
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.5952115058898926,
      "learning_rate": 1.0706021119711639e-05,
      "loss": 1.9601,
      "step": 210751
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.345650315284729,
      "learning_rate": 1.0704938113347217e-05,
      "loss": 2.1172,
      "step": 210752
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.428443431854248,
      "learning_rate": 1.0703855160768237e-05,
      "loss": 2.0397,
      "step": 210753
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.5063111782073975,
      "learning_rate": 1.0702772261974834e-05,
      "loss": 2.1194,
      "step": 210754
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.421751618385315,
      "learning_rate": 1.0701689416967274e-05,
      "loss": 1.8777,
      "step": 210755
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.346848726272583,
      "learning_rate": 1.0700606625745756e-05,
      "loss": 1.9971,
      "step": 210756
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.5771533250808716,
      "learning_rate": 1.0699523888310446e-05,
      "loss": 1.9953,
      "step": 210757
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.4265443086624146,
      "learning_rate": 1.0698441204661511e-05,
      "loss": 1.8888,
      "step": 210758
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.7872886657714844,
      "learning_rate": 1.069735857479922e-05,
      "loss": 2.244,
      "step": 210759
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.6574084758758545,
      "learning_rate": 1.0696275998723735e-05,
      "loss": 2.2569,
      "step": 210760
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.5559717416763306,
      "learning_rate": 1.0695193476435326e-05,
      "loss": 2.0422,
      "step": 210761
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.3782483339309692,
      "learning_rate": 1.0694111007934125e-05,
      "loss": 1.9434,
      "step": 210762
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.6081300973892212,
      "learning_rate": 1.0693028593220332e-05,
      "loss": 2.1217,
      "step": 210763
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.3625342845916748,
      "learning_rate": 1.0691946232294147e-05,
      "loss": 1.9303,
      "step": 210764
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.388032078742981,
      "learning_rate": 1.0690863925155834e-05,
      "loss": 1.9437,
      "step": 210765
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.4396623373031616,
      "learning_rate": 1.0689781671805497e-05,
      "loss": 2.1187,
      "step": 210766
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.483204960823059,
      "learning_rate": 1.0688699472243433e-05,
      "loss": 1.8237,
      "step": 210767
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.4526320695877075,
      "learning_rate": 1.068761732646981e-05,
      "loss": 1.9225,
      "step": 210768
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.526379942893982,
      "learning_rate": 1.0686535234484828e-05,
      "loss": 2.0989,
      "step": 210769
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.3863860368728638,
      "learning_rate": 1.0685453196288618e-05,
      "loss": 1.8633,
      "step": 210770
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.4880414009094238,
      "learning_rate": 1.0684371211881481e-05,
      "loss": 1.9406,
      "step": 210771
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.3700056076049805,
      "learning_rate": 1.0683289281263585e-05,
      "loss": 1.9545,
      "step": 210772
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.4401884078979492,
      "learning_rate": 1.0682207404435129e-05,
      "loss": 1.9633,
      "step": 210773
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.4602391719818115,
      "learning_rate": 1.068112558139631e-05,
      "loss": 1.8357,
      "step": 210774
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.5428463220596313,
      "learning_rate": 1.0680043812147332e-05,
      "loss": 2.0766,
      "step": 210775
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.52555251121521,
      "learning_rate": 1.0678962096688359e-05,
      "loss": 1.8866,
      "step": 210776
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.9145610332489014,
      "learning_rate": 1.0677880435019693e-05,
      "loss": 1.9862,
      "step": 210777
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.4555306434631348,
      "learning_rate": 1.0676798827141397e-05,
      "loss": 2.0424,
      "step": 210778
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.4606767892837524,
      "learning_rate": 1.0675717273053808e-05,
      "loss": 2.0575,
      "step": 210779
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.5196152925491333,
      "learning_rate": 1.0674635772757057e-05,
      "loss": 1.8862,
      "step": 210780
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.5925042629241943,
      "learning_rate": 1.0673554326251343e-05,
      "loss": 2.1722,
      "step": 210781
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.5063021183013916,
      "learning_rate": 1.0672472933536836e-05,
      "loss": 2.0635,
      "step": 210782
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.4555351734161377,
      "learning_rate": 1.0671391594613832e-05,
      "loss": 1.9442,
      "step": 210783
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.6497905254364014,
      "learning_rate": 1.0670310309482433e-05,
      "loss": 2.0901,
      "step": 210784
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.3688809871673584,
      "learning_rate": 1.0669229078142905e-05,
      "loss": 1.8456,
      "step": 210785
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.4669767618179321,
      "learning_rate": 1.0668147900595413e-05,
      "loss": 2.1779,
      "step": 210786
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.4259566068649292,
      "learning_rate": 1.0667066776840227e-05,
      "loss": 1.9902,
      "step": 210787
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.5533132553100586,
      "learning_rate": 1.0665985706877444e-05,
      "loss": 1.9522,
      "step": 210788
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.4581784009933472,
      "learning_rate": 1.0664904690707365e-05,
      "loss": 2.0112,
      "step": 210789
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.494112491607666,
      "learning_rate": 1.066382372833009e-05,
      "loss": 2.0202,
      "step": 210790
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.442984938621521,
      "learning_rate": 1.0662742819745885e-05,
      "loss": 1.8279,
      "step": 210791
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.3921222686767578,
      "learning_rate": 1.0661661964954916e-05,
      "loss": 1.8457,
      "step": 210792
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.3934056758880615,
      "learning_rate": 1.0660581163957483e-05,
      "loss": 1.9635,
      "step": 210793
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.4118890762329102,
      "learning_rate": 1.0659500416753653e-05,
      "loss": 1.8412,
      "step": 210794
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.3986597061157227,
      "learning_rate": 1.0658419723343692e-05,
      "loss": 2.031,
      "step": 210795
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.8524208068847656,
      "learning_rate": 1.0657339083727767e-05,
      "loss": 2.1005,
      "step": 210796
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.452504277229309,
      "learning_rate": 1.0656258497906144e-05,
      "loss": 1.9342,
      "step": 210797
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.4610241651535034,
      "learning_rate": 1.0655177965878959e-05,
      "loss": 2.1529,
      "step": 210798
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.3708316087722778,
      "learning_rate": 1.0654097487646507e-05,
      "loss": 1.9638,
      "step": 210799
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.5273994207382202,
      "learning_rate": 1.0653017063208823e-05,
      "loss": 1.9894,
      "step": 210800
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.3744345903396606,
      "learning_rate": 1.0651936692566243e-05,
      "loss": 1.9686,
      "step": 210801
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.4428691864013672,
      "learning_rate": 1.0650856375718931e-05,
      "loss": 2.0923,
      "step": 210802
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.395828366279602,
      "learning_rate": 1.0649776112667085e-05,
      "loss": 1.7899,
      "step": 210803
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.445921540260315,
      "learning_rate": 1.064869590341091e-05,
      "loss": 1.834,
      "step": 210804
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.3996690511703491,
      "learning_rate": 1.0647615747950634e-05,
      "loss": 1.9666,
      "step": 210805
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.5097358226776123,
      "learning_rate": 1.0646535646286359e-05,
      "loss": 2.1113,
      "step": 210806
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.3732165098190308,
      "learning_rate": 1.0645455598418418e-05,
      "loss": 1.8783,
      "step": 210807
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.4929957389831543,
      "learning_rate": 1.0644375604346911e-05,
      "loss": 1.9123,
      "step": 210808
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.4865223169326782,
      "learning_rate": 1.0643295664072104e-05,
      "loss": 2.1507,
      "step": 210809
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.456400990486145,
      "learning_rate": 1.0642215777594132e-05,
      "loss": 2.1484,
      "step": 210810
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.6016367673873901,
      "learning_rate": 1.0641135944913293e-05,
      "loss": 2.1505,
      "step": 210811
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.6309720277786255,
      "learning_rate": 1.0640056166029687e-05,
      "loss": 2.1226,
      "step": 210812
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.550836205482483,
      "learning_rate": 1.063897644094358e-05,
      "loss": 2.0446,
      "step": 210813
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.426513433456421,
      "learning_rate": 1.0637896769655108e-05,
      "loss": 1.9651,
      "step": 210814
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.47140371799469,
      "learning_rate": 1.0636817152164534e-05,
      "loss": 1.8649,
      "step": 210815
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.3749874830245972,
      "learning_rate": 1.0635737588472027e-05,
      "loss": 1.8044,
      "step": 210816
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.4871686697006226,
      "learning_rate": 1.0634658078577819e-05,
      "loss": 1.8961,
      "step": 210817
    },
    {
      "epoch": 2.74,
      "grad_norm": 1.4091928005218506,
      "learning_rate": 1.0633578622482108e-05,
      "loss": 2.0443,
      "step": 210818
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.4242359399795532,
      "learning_rate": 1.0632499220185065e-05,
      "loss": 2.0413,
      "step": 210819
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.555936574935913,
      "learning_rate": 1.0631419871686852e-05,
      "loss": 2.0335,
      "step": 210820
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.458373785018921,
      "learning_rate": 1.0630340576987773e-05,
      "loss": 2.0091,
      "step": 210821
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.46638822555542,
      "learning_rate": 1.0629261336087957e-05,
      "loss": 1.7066,
      "step": 210822
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.4249895811080933,
      "learning_rate": 1.062818214898764e-05,
      "loss": 1.9975,
      "step": 210823
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.5613878965377808,
      "learning_rate": 1.0627103015687022e-05,
      "loss": 1.8272,
      "step": 210824
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.5351543426513672,
      "learning_rate": 1.0626023936186234e-05,
      "loss": 1.938,
      "step": 210825
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.4434984922409058,
      "learning_rate": 1.0624944910485577e-05,
      "loss": 1.8447,
      "step": 210826
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.386325478553772,
      "learning_rate": 1.0623865938585185e-05,
      "loss": 1.8895,
      "step": 210827
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.5707050561904907,
      "learning_rate": 1.0622787020485257e-05,
      "loss": 2.1931,
      "step": 210828
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.5275605916976929,
      "learning_rate": 1.0621708156186059e-05,
      "loss": 1.8215,
      "step": 210829
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.5022772550582886,
      "learning_rate": 1.0620629345687726e-05,
      "loss": 1.8833,
      "step": 210830
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.3703030347824097,
      "learning_rate": 1.0619550588990456e-05,
      "loss": 1.9434,
      "step": 210831
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.591564655303955,
      "learning_rate": 1.0618471886094514e-05,
      "loss": 1.9476,
      "step": 210832
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.5329989194869995,
      "learning_rate": 1.0617393237000071e-05,
      "loss": 2.0993,
      "step": 210833
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.497495174407959,
      "learning_rate": 1.0616314641707257e-05,
      "loss": 1.8769,
      "step": 210834
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.4102503061294556,
      "learning_rate": 1.0615236100216373e-05,
      "loss": 1.9288,
      "step": 210835
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.4046506881713867,
      "learning_rate": 1.0614157612527585e-05,
      "loss": 1.8056,
      "step": 210836
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.4229950904846191,
      "learning_rate": 1.0613079178641059e-05,
      "loss": 1.9198,
      "step": 210837
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.484593391418457,
      "learning_rate": 1.0612000798557063e-05,
      "loss": 1.8279,
      "step": 210838
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.4310742616653442,
      "learning_rate": 1.061092247227573e-05,
      "loss": 2.2197,
      "step": 210839
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.4945582151412964,
      "learning_rate": 1.0609844199797291e-05,
      "loss": 2.009,
      "step": 210840
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.4626954793930054,
      "learning_rate": 1.0608765981121981e-05,
      "loss": 1.8977,
      "step": 210841
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.5331631898880005,
      "learning_rate": 1.0607687816249933e-05,
      "loss": 1.8987,
      "step": 210842
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.5153312683105469,
      "learning_rate": 1.0606609705181347e-05,
      "loss": 2.0354,
      "step": 210843
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.4193435907363892,
      "learning_rate": 1.0605531647916521e-05,
      "loss": 1.9811,
      "step": 210844
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.596319317817688,
      "learning_rate": 1.0604453644455524e-05,
      "loss": 1.9734,
      "step": 210845
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.4041730165481567,
      "learning_rate": 1.0603375694798656e-05,
      "loss": 1.8843,
      "step": 210846
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.491047978401184,
      "learning_rate": 1.0602297798946113e-05,
      "loss": 2.0411,
      "step": 210847
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.382567286491394,
      "learning_rate": 1.0601219956898033e-05,
      "loss": 2.1455,
      "step": 210848
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.5670356750488281,
      "learning_rate": 1.0600142168654612e-05,
      "loss": 2.1419,
      "step": 210849
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.6435645818710327,
      "learning_rate": 1.0599064434216153e-05,
      "loss": 1.9451,
      "step": 210850
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.4553016424179077,
      "learning_rate": 1.0597986753582754e-05,
      "loss": 2.2254,
      "step": 210851
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.3789875507354736,
      "learning_rate": 1.0596909126754648e-05,
      "loss": 2.0613,
      "step": 210852
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.4217631816864014,
      "learning_rate": 1.0595831553732037e-05,
      "loss": 1.9517,
      "step": 210853
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.6299612522125244,
      "learning_rate": 1.0594754034515185e-05,
      "loss": 1.9491,
      "step": 210854
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.5329395532608032,
      "learning_rate": 1.0593676569104159e-05,
      "loss": 1.8758,
      "step": 210855
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.4962376356124878,
      "learning_rate": 1.0592599157499293e-05,
      "loss": 1.9903,
      "step": 210856
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.5145576000213623,
      "learning_rate": 1.0591521799700652e-05,
      "loss": 1.9648,
      "step": 210857
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.530617356300354,
      "learning_rate": 1.0590444495708571e-05,
      "loss": 1.9955,
      "step": 210858
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.4949840307235718,
      "learning_rate": 1.058936724552315e-05,
      "loss": 1.8457,
      "step": 210859
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.3915008306503296,
      "learning_rate": 1.0588290049144688e-05,
      "loss": 1.9556,
      "step": 210860
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.381269931793213,
      "learning_rate": 1.058721290657325e-05,
      "loss": 1.9596,
      "step": 210861
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.37269127368927,
      "learning_rate": 1.0586135817809171e-05,
      "loss": 2.0202,
      "step": 210862
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.3827558755874634,
      "learning_rate": 1.0585058782852519e-05,
      "loss": 2.1969,
      "step": 210863
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.4048337936401367,
      "learning_rate": 1.0583981801703656e-05,
      "loss": 1.9102,
      "step": 210864
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.416207194328308,
      "learning_rate": 1.0582904874362619e-05,
      "loss": 1.8589,
      "step": 210865
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.5170226097106934,
      "learning_rate": 1.0581828000829773e-05,
      "loss": 1.9283,
      "step": 210866
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.41932213306427,
      "learning_rate": 1.0580751181105151e-05,
      "loss": 1.898,
      "step": 210867
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.6697531938552856,
      "learning_rate": 1.0579674415189054e-05,
      "loss": 1.9417,
      "step": 210868
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.5783424377441406,
      "learning_rate": 1.0578597703081649e-05,
      "loss": 2.0881,
      "step": 210869
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.46805739402771,
      "learning_rate": 1.0577521044783166e-05,
      "loss": 1.842,
      "step": 210870
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.427049994468689,
      "learning_rate": 1.0576444440293741e-05,
      "loss": 2.0476,
      "step": 210871
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.3783907890319824,
      "learning_rate": 1.0575367889613707e-05,
      "loss": 1.9944,
      "step": 210872
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.3894659280776978,
      "learning_rate": 1.0574291392743094e-05,
      "loss": 2.1275,
      "step": 210873
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.4571453332901,
      "learning_rate": 1.0573214949682208e-05,
      "loss": 2.2901,
      "step": 210874
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.4236855506896973,
      "learning_rate": 1.0572138560431208e-05,
      "loss": 1.9112,
      "step": 210875
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.463173747062683,
      "learning_rate": 1.0571062224990334e-05,
      "loss": 1.9639,
      "step": 210876
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.536589503288269,
      "learning_rate": 1.056998594335975e-05,
      "loss": 2.0341,
      "step": 210877
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.6155027151107788,
      "learning_rate": 1.056890971553972e-05,
      "loss": 2.0942,
      "step": 210878
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.536675214767456,
      "learning_rate": 1.0567833541530312e-05,
      "loss": 2.0548,
      "step": 210879
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.4399433135986328,
      "learning_rate": 1.0566757421331861e-05,
      "loss": 1.8568,
      "step": 210880
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.5622586011886597,
      "learning_rate": 1.05656813549445e-05,
      "loss": 2.0227,
      "step": 210881
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.5017813444137573,
      "learning_rate": 1.0564605342368426e-05,
      "loss": 1.9436,
      "step": 210882
    },
    {
      "epoch": 2.75,
      "grad_norm": 2.3142056465148926,
      "learning_rate": 1.0563529383603841e-05,
      "loss": 1.9973,
      "step": 210883
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.553558111190796,
      "learning_rate": 1.0562453478651045e-05,
      "loss": 1.7719,
      "step": 210884
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.3971526622772217,
      "learning_rate": 1.056137762751007e-05,
      "loss": 1.8271,
      "step": 210885
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.4522273540496826,
      "learning_rate": 1.0560301830181251e-05,
      "loss": 2.0663,
      "step": 210886
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.5107475519180298,
      "learning_rate": 1.0559226086664684e-05,
      "loss": 2.0408,
      "step": 210887
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.5256245136260986,
      "learning_rate": 1.055815039696064e-05,
      "loss": 2.0809,
      "step": 210888
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.3984942436218262,
      "learning_rate": 1.0557074761069284e-05,
      "loss": 1.9162,
      "step": 210889
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.3611091375350952,
      "learning_rate": 1.0555999178990915e-05,
      "loss": 2.0441,
      "step": 210890
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.5540181398391724,
      "learning_rate": 1.0554923650725566e-05,
      "loss": 1.901,
      "step": 210891
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.4147472381591797,
      "learning_rate": 1.0553848176273538e-05,
      "loss": 1.9963,
      "step": 210892
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.5912021398544312,
      "learning_rate": 1.0552772755634997e-05,
      "loss": 1.8714,
      "step": 210893
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.5480421781539917,
      "learning_rate": 1.0551697388810176e-05,
      "loss": 1.9053,
      "step": 210894
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.4583115577697754,
      "learning_rate": 1.0550622075799242e-05,
      "loss": 1.9222,
      "step": 210895
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.4350805282592773,
      "learning_rate": 1.0549546816602494e-05,
      "loss": 1.7845,
      "step": 210896
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.461622953414917,
      "learning_rate": 1.0548471611219933e-05,
      "loss": 2.1017,
      "step": 210897
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.522720217704773,
      "learning_rate": 1.0547396459651957e-05,
      "loss": 1.9753,
      "step": 210898
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.4756864309310913,
      "learning_rate": 1.0546321361898635e-05,
      "loss": 1.9666,
      "step": 210899
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.564307689666748,
      "learning_rate": 1.0545246317960232e-05,
      "loss": 2.0168,
      "step": 210900
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.5320178270339966,
      "learning_rate": 1.0544171327836915e-05,
      "loss": 1.9842,
      "step": 210901
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.4447355270385742,
      "learning_rate": 1.0543096391528916e-05,
      "loss": 2.102,
      "step": 210902
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.387086033821106,
      "learning_rate": 1.0542021509036436e-05,
      "loss": 2.0086,
      "step": 210903
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.4490959644317627,
      "learning_rate": 1.0540946680359674e-05,
      "loss": 1.8292,
      "step": 210904
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.354470133781433,
      "learning_rate": 1.0539871905498764e-05,
      "loss": 1.9417,
      "step": 210905
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.4295395612716675,
      "learning_rate": 1.0538797184453974e-05,
      "loss": 2.0362,
      "step": 210906
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.3413833379745483,
      "learning_rate": 1.0537722517225466e-05,
      "loss": 2.0091,
      "step": 210907
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.3638678789138794,
      "learning_rate": 1.0536647903813511e-05,
      "loss": 1.8885,
      "step": 210908
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.4286267757415771,
      "learning_rate": 1.0535573344218239e-05,
      "loss": 1.9695,
      "step": 210909
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.3850147724151611,
      "learning_rate": 1.053449883843982e-05,
      "loss": 2.234,
      "step": 210910
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.4621586799621582,
      "learning_rate": 1.053342438647855e-05,
      "loss": 1.9002,
      "step": 210911
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.5376338958740234,
      "learning_rate": 1.0532349988334599e-05,
      "loss": 1.8324,
      "step": 210912
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.5552420616149902,
      "learning_rate": 1.0531275644008064e-05,
      "loss": 1.7812,
      "step": 210913
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.5645334720611572,
      "learning_rate": 1.0530201353499313e-05,
      "loss": 2.0196,
      "step": 210914
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.406733751296997,
      "learning_rate": 1.0529127116808445e-05,
      "loss": 1.8804,
      "step": 210915
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.4485843181610107,
      "learning_rate": 1.0528052933935626e-05,
      "loss": 1.9539,
      "step": 210916
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.5409998893737793,
      "learning_rate": 1.0526978804881158e-05,
      "loss": 2.1808,
      "step": 210917
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.4114660024642944,
      "learning_rate": 1.0525904729645207e-05,
      "loss": 2.1946,
      "step": 210918
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.4485609531402588,
      "learning_rate": 1.0524830708227871e-05,
      "loss": 2.0092,
      "step": 210919
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.5219173431396484,
      "learning_rate": 1.052375674062952e-05,
      "loss": 1.9555,
      "step": 210920
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.3097848892211914,
      "learning_rate": 1.0522682826850215e-05,
      "loss": 1.7455,
      "step": 210921
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.4582362174987793,
      "learning_rate": 1.0521608966890226e-05,
      "loss": 1.9718,
      "step": 210922
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.4887197017669678,
      "learning_rate": 1.052053516074972e-05,
      "loss": 1.967,
      "step": 210923
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.66200852394104,
      "learning_rate": 1.0519461408428897e-05,
      "loss": 1.9722,
      "step": 210924
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.4535905122756958,
      "learning_rate": 1.0518387709928022e-05,
      "loss": 1.89,
      "step": 210925
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.5153018236160278,
      "learning_rate": 1.051731406524723e-05,
      "loss": 2.0151,
      "step": 210926
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.4568536281585693,
      "learning_rate": 1.051624047438675e-05,
      "loss": 1.9474,
      "step": 210927
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.5245610475540161,
      "learning_rate": 1.0515166937346686e-05,
      "loss": 1.9548,
      "step": 210928
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.4443082809448242,
      "learning_rate": 1.0514093454127403e-05,
      "loss": 1.9608,
      "step": 210929
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.4012813568115234,
      "learning_rate": 1.0513020024728936e-05,
      "loss": 1.9678,
      "step": 210930
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.382552146911621,
      "learning_rate": 1.0511946649151615e-05,
      "loss": 1.8398,
      "step": 210931
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.5500342845916748,
      "learning_rate": 1.0510873327395608e-05,
      "loss": 1.858,
      "step": 210932
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.3606061935424805,
      "learning_rate": 1.0509800059461082e-05,
      "loss": 1.8974,
      "step": 210933
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.4163310527801514,
      "learning_rate": 1.0508726845348203e-05,
      "loss": 1.9915,
      "step": 210934
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.487398624420166,
      "learning_rate": 1.0507653685057238e-05,
      "loss": 1.9792,
      "step": 210935
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.461668610572815,
      "learning_rate": 1.0506580578588352e-05,
      "loss": 2.3796,
      "step": 210936
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.4734997749328613,
      "learning_rate": 1.050550752594178e-05,
      "loss": 2.0641,
      "step": 210937
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.3884871006011963,
      "learning_rate": 1.0504434527117689e-05,
      "loss": 1.9211,
      "step": 210938
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.4586137533187866,
      "learning_rate": 1.0503361582116342e-05,
      "loss": 2.1069,
      "step": 210939
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.5501724481582642,
      "learning_rate": 1.050228869093781e-05,
      "loss": 2.15,
      "step": 210940
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.5077325105667114,
      "learning_rate": 1.050121585358239e-05,
      "loss": 1.9649,
      "step": 210941
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.4965486526489258,
      "learning_rate": 1.0500143070050249e-05,
      "loss": 2.1308,
      "step": 210942
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.3511183261871338,
      "learning_rate": 1.049907034034162e-05,
      "loss": 2.0199,
      "step": 210943
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.3060379028320312,
      "learning_rate": 1.049799766445667e-05,
      "loss": 1.825,
      "step": 210944
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.3840457201004028,
      "learning_rate": 1.0496925042395632e-05,
      "loss": 1.7434,
      "step": 210945
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.4922419786453247,
      "learning_rate": 1.049585247415864e-05,
      "loss": 1.9164,
      "step": 210946
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.4410676956176758,
      "learning_rate": 1.0494779959745959e-05,
      "loss": 2.1277,
      "step": 210947
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.3786399364471436,
      "learning_rate": 1.0493707499157722e-05,
      "loss": 1.987,
      "step": 210948
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.492371678352356,
      "learning_rate": 1.0492635092394231e-05,
      "loss": 2.0612,
      "step": 210949
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.414749264717102,
      "learning_rate": 1.049156273945555e-05,
      "loss": 2.0989,
      "step": 210950
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.4434473514556885,
      "learning_rate": 1.0490490440342048e-05,
      "loss": 2.0895,
      "step": 210951
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.4728186130523682,
      "learning_rate": 1.0489418195053755e-05,
      "loss": 1.9649,
      "step": 210952
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.3092443943023682,
      "learning_rate": 1.0488346003590974e-05,
      "loss": 2.2277,
      "step": 210953
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.5360223054885864,
      "learning_rate": 1.0487273865953839e-05,
      "loss": 1.9121,
      "step": 210954
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.5832188129425049,
      "learning_rate": 1.0486201782142645e-05,
      "loss": 2.0226,
      "step": 210955
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.4071329832077026,
      "learning_rate": 1.0485129752157461e-05,
      "loss": 1.9193,
      "step": 210956
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.5530805587768555,
      "learning_rate": 1.0484057775998622e-05,
      "loss": 2.1392,
      "step": 210957
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.3931300640106201,
      "learning_rate": 1.0482985853666227e-05,
      "loss": 2.0751,
      "step": 210958
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.4412826299667358,
      "learning_rate": 1.0481913985160539e-05,
      "loss": 1.9409,
      "step": 210959
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.4699056148529053,
      "learning_rate": 1.0480842170481662e-05,
      "loss": 1.8022,
      "step": 210960
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.5495144128799438,
      "learning_rate": 1.0479770409629928e-05,
      "loss": 1.7083,
      "step": 210961
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.5805461406707764,
      "learning_rate": 1.0478698702605403e-05,
      "loss": 2.0974,
      "step": 210962
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.5258642435073853,
      "learning_rate": 1.0477627049408455e-05,
      "loss": 2.0908,
      "step": 210963
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.4355412721633911,
      "learning_rate": 1.0476555450039114e-05,
      "loss": 2.0649,
      "step": 210964
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.411895513534546,
      "learning_rate": 1.047548390449765e-05,
      "loss": 1.7654,
      "step": 210965
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.446512222290039,
      "learning_rate": 1.0474412412784262e-05,
      "loss": 1.8431,
      "step": 210966
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.7467868328094482,
      "learning_rate": 1.0473340974899147e-05,
      "loss": 1.911,
      "step": 210967
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.349510669708252,
      "learning_rate": 1.0472269590842476e-05,
      "loss": 2.1132,
      "step": 210968
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.462178349494934,
      "learning_rate": 1.0471198260614543e-05,
      "loss": 2.0639,
      "step": 210969
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.5500890016555786,
      "learning_rate": 1.0470126984215421e-05,
      "loss": 1.8328,
      "step": 210970
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.3911008834838867,
      "learning_rate": 1.0469055761645405e-05,
      "loss": 1.9776,
      "step": 210971
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.3961795568466187,
      "learning_rate": 1.04679845929046e-05,
      "loss": 2.1633,
      "step": 210972
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.4998971223831177,
      "learning_rate": 1.0466913477993334e-05,
      "loss": 2.1889,
      "step": 210973
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.6281201839447021,
      "learning_rate": 1.0465842416911674e-05,
      "loss": 2.0766,
      "step": 210974
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.4298897981643677,
      "learning_rate": 1.0464771409659956e-05,
      "loss": 1.9683,
      "step": 210975
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.2939257621765137,
      "learning_rate": 1.0463700456238211e-05,
      "loss": 1.9593,
      "step": 210976
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.5120011568069458,
      "learning_rate": 1.0462629556646807e-05,
      "loss": 2.1233,
      "step": 210977
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.5300501585006714,
      "learning_rate": 1.0461558710885776e-05,
      "loss": 1.9669,
      "step": 210978
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.8817311525344849,
      "learning_rate": 1.0460487918955484e-05,
      "loss": 2.1206,
      "step": 210979
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.351113200187683,
      "learning_rate": 1.0459417180856e-05,
      "loss": 1.9644,
      "step": 210980
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.5181413888931274,
      "learning_rate": 1.0458346496587655e-05,
      "loss": 2.0952,
      "step": 210981
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.4436569213867188,
      "learning_rate": 1.0457275866150483e-05,
      "loss": 1.9145,
      "step": 210982
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.3489177227020264,
      "learning_rate": 1.0456205289544817e-05,
      "loss": 2.0788,
      "step": 210983
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.3661633729934692,
      "learning_rate": 1.045513476677079e-05,
      "loss": 1.8937,
      "step": 210984
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.3666198253631592,
      "learning_rate": 1.0454064297828635e-05,
      "loss": 2.0063,
      "step": 210985
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.5439865589141846,
      "learning_rate": 1.0452993882718485e-05,
      "loss": 1.8439,
      "step": 210986
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.5003087520599365,
      "learning_rate": 1.0451923521440642e-05,
      "loss": 2.0117,
      "step": 210987
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.4297126531600952,
      "learning_rate": 1.045085321399527e-05,
      "loss": 1.9946,
      "step": 210988
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.5774246454238892,
      "learning_rate": 1.0449782960382536e-05,
      "loss": 2.1178,
      "step": 210989
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.3762025833129883,
      "learning_rate": 1.0448712760602606e-05,
      "loss": 2.1037,
      "step": 210990
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.4292975664138794,
      "learning_rate": 1.044764261465575e-05,
      "loss": 1.8812,
      "step": 210991
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.5317753553390503,
      "learning_rate": 1.044657252254213e-05,
      "loss": 1.9567,
      "step": 210992
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.4444621801376343,
      "learning_rate": 1.044550248426198e-05,
      "loss": 1.7895,
      "step": 210993
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.4538251161575317,
      "learning_rate": 1.0444432499815503e-05,
      "loss": 1.9065,
      "step": 210994
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.4655293226242065,
      "learning_rate": 1.0443362569202863e-05,
      "loss": 2.1445,
      "step": 210995
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.3034299612045288,
      "learning_rate": 1.0442292692424226e-05,
      "loss": 1.8263,
      "step": 210996
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.400143027305603,
      "learning_rate": 1.044122286947986e-05,
      "loss": 1.8108,
      "step": 210997
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.5015619993209839,
      "learning_rate": 1.0440153100369896e-05,
      "loss": 2.0447,
      "step": 210998
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.4729682207107544,
      "learning_rate": 1.0439083385094637e-05,
      "loss": 1.8815,
      "step": 210999
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.3687424659729004,
      "learning_rate": 1.0438013723654182e-05,
      "loss": 2.222,
      "step": 211000
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.3709744215011597,
      "learning_rate": 1.0436944116048762e-05,
      "loss": 1.9059,
      "step": 211001
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.449451208114624,
      "learning_rate": 1.0435874562278578e-05,
      "loss": 2.0698,
      "step": 211002
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.5062185525894165,
      "learning_rate": 1.0434805062343865e-05,
      "loss": 2.1231,
      "step": 211003
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.3025139570236206,
      "learning_rate": 1.0433735616244754e-05,
      "loss": 1.7515,
      "step": 211004
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.4482654333114624,
      "learning_rate": 1.0432666223981511e-05,
      "loss": 2.163,
      "step": 211005
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.3896011114120483,
      "learning_rate": 1.0431596885554272e-05,
      "loss": 1.979,
      "step": 211006
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.491777777671814,
      "learning_rate": 1.0430527600963268e-05,
      "loss": 2.0674,
      "step": 211007
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.5059115886688232,
      "learning_rate": 1.0429458370208698e-05,
      "loss": 1.8177,
      "step": 211008
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.5175362825393677,
      "learning_rate": 1.0428389193290731e-05,
      "loss": 1.9194,
      "step": 211009
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.3110766410827637,
      "learning_rate": 1.0427320070209633e-05,
      "loss": 1.9401,
      "step": 211010
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.5853580236434937,
      "learning_rate": 1.0426251000965568e-05,
      "loss": 1.9961,
      "step": 211011
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.7257611751556396,
      "learning_rate": 1.042518198555874e-05,
      "loss": 2.0255,
      "step": 211012
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.4335829019546509,
      "learning_rate": 1.0424113023989277e-05,
      "loss": 1.8971,
      "step": 211013
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.4394488334655762,
      "learning_rate": 1.0423044116257485e-05,
      "loss": 2.006,
      "step": 211014
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.4532911777496338,
      "learning_rate": 1.0421975262363457e-05,
      "loss": 2.0057,
      "step": 211015
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.4787653684616089,
      "learning_rate": 1.0420906462307532e-05,
      "loss": 1.9578,
      "step": 211016
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.6039519309997559,
      "learning_rate": 1.0419837716089774e-05,
      "loss": 2.0118,
      "step": 211017
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.495460033416748,
      "learning_rate": 1.0418769023710483e-05,
      "loss": 2.0072,
      "step": 211018
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.567270040512085,
      "learning_rate": 1.0417700385169758e-05,
      "loss": 1.9628,
      "step": 211019
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.403990626335144,
      "learning_rate": 1.0416631800467868e-05,
      "loss": 1.9827,
      "step": 211020
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.4376739263534546,
      "learning_rate": 1.0415563269604944e-05,
      "loss": 1.9534,
      "step": 211021
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.3084063529968262,
      "learning_rate": 1.0414494792581318e-05,
      "loss": 1.8469,
      "step": 211022
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.3997303247451782,
      "learning_rate": 1.0413426369397027e-05,
      "loss": 1.9639,
      "step": 211023
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.384681224822998,
      "learning_rate": 1.0412358000052434e-05,
      "loss": 2.1106,
      "step": 211024
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.360995888710022,
      "learning_rate": 1.0411289684547574e-05,
      "loss": 2.1089,
      "step": 211025
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.6390548944473267,
      "learning_rate": 1.0410221422882747e-05,
      "loss": 1.9433,
      "step": 211026
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.5030053853988647,
      "learning_rate": 1.040915321505812e-05,
      "loss": 1.8782,
      "step": 211027
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.5648341178894043,
      "learning_rate": 1.0408085061073923e-05,
      "loss": 1.896,
      "step": 211028
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.470657467842102,
      "learning_rate": 1.0407016960930292e-05,
      "loss": 1.8035,
      "step": 211029
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.4435462951660156,
      "learning_rate": 1.0405948914627526e-05,
      "loss": 1.9134,
      "step": 211030
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.459818959236145,
      "learning_rate": 1.0404880922165692e-05,
      "loss": 1.8628,
      "step": 211031
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.3894637823104858,
      "learning_rate": 1.040381298354509e-05,
      "loss": 2.0143,
      "step": 211032
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.4848744869232178,
      "learning_rate": 1.0402745098765852e-05,
      "loss": 1.9814,
      "step": 211033
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.3525104522705078,
      "learning_rate": 1.0401677267828246e-05,
      "loss": 1.8314,
      "step": 211034
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.5183234214782715,
      "learning_rate": 1.0400609490732403e-05,
      "loss": 2.2127,
      "step": 211035
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.4554665088653564,
      "learning_rate": 1.0399541767478625e-05,
      "loss": 1.9061,
      "step": 211036
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.6177231073379517,
      "learning_rate": 1.0398474098066945e-05,
      "loss": 2.1995,
      "step": 211037
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.3812178373336792,
      "learning_rate": 1.0397406482497727e-05,
      "loss": 2.0675,
      "step": 211038
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.5649824142456055,
      "learning_rate": 1.0396338920771041e-05,
      "loss": 1.9384,
      "step": 211039
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.533095121383667,
      "learning_rate": 1.0395271412887185e-05,
      "loss": 1.9878,
      "step": 211040
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.419439673423767,
      "learning_rate": 1.0394203958846291e-05,
      "loss": 1.9679,
      "step": 211041
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.4152709245681763,
      "learning_rate": 1.039313655864863e-05,
      "loss": 2.1282,
      "step": 211042
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.4791711568832397,
      "learning_rate": 1.0392069212294295e-05,
      "loss": 1.9049,
      "step": 211043
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.4441492557525635,
      "learning_rate": 1.0391001919783559e-05,
      "loss": 2.0404,
      "step": 211044
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.4154974222183228,
      "learning_rate": 1.0389934681116586e-05,
      "loss": 2.0224,
      "step": 211045
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.8267114162445068,
      "learning_rate": 1.0388867496293607e-05,
      "loss": 1.8738,
      "step": 211046
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.4689154624938965,
      "learning_rate": 1.038780036531479e-05,
      "loss": 2.0477,
      "step": 211047
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.529536247253418,
      "learning_rate": 1.0386733288180404e-05,
      "loss": 2.1677,
      "step": 211048
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.3996944427490234,
      "learning_rate": 1.0385666264890513e-05,
      "loss": 1.8633,
      "step": 211049
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.3979123830795288,
      "learning_rate": 1.038459929544545e-05,
      "loss": 2.1493,
      "step": 211050
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.4102225303649902,
      "learning_rate": 1.0383532379845316e-05,
      "loss": 2.1441,
      "step": 211051
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.6220091581344604,
      "learning_rate": 1.0382465518090377e-05,
      "loss": 2.1218,
      "step": 211052
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.485935091972351,
      "learning_rate": 1.03813987101808e-05,
      "loss": 1.9571,
      "step": 211053
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.5584137439727783,
      "learning_rate": 1.0380331956116784e-05,
      "loss": 1.8893,
      "step": 211054
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.4255298376083374,
      "learning_rate": 1.0379265255898562e-05,
      "loss": 1.9399,
      "step": 211055
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.4246060848236084,
      "learning_rate": 1.0378198609526267e-05,
      "loss": 2.0887,
      "step": 211056
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.5031142234802246,
      "learning_rate": 1.0377132017000134e-05,
      "loss": 2.366,
      "step": 211057
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.467940330505371,
      "learning_rate": 1.037606547832036e-05,
      "loss": 1.9678,
      "step": 211058
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.4164472818374634,
      "learning_rate": 1.0374998993487117e-05,
      "loss": 1.9286,
      "step": 211059
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.3874261379241943,
      "learning_rate": 1.0373932562500663e-05,
      "loss": 1.8128,
      "step": 211060
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.4010281562805176,
      "learning_rate": 1.0372866185361172e-05,
      "loss": 1.9709,
      "step": 211061
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.6259011030197144,
      "learning_rate": 1.0371799862068841e-05,
      "loss": 1.9252,
      "step": 211062
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.3672828674316406,
      "learning_rate": 1.0370733592623804e-05,
      "loss": 2.0934,
      "step": 211063
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.3405057191848755,
      "learning_rate": 1.0369667377026359e-05,
      "loss": 1.8605,
      "step": 211064
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.4711427688598633,
      "learning_rate": 1.036860121527664e-05,
      "loss": 1.9813,
      "step": 211065
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.411604642868042,
      "learning_rate": 1.036753510737488e-05,
      "loss": 1.9054,
      "step": 211066
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.495936632156372,
      "learning_rate": 1.036646905332128e-05,
      "loss": 2.0498,
      "step": 211067
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.459261178970337,
      "learning_rate": 1.0365403053116005e-05,
      "loss": 1.9169,
      "step": 211068
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.5049855709075928,
      "learning_rate": 1.0364337106759224e-05,
      "loss": 1.9943,
      "step": 211069
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.47975492477417,
      "learning_rate": 1.0363271214251235e-05,
      "loss": 1.957,
      "step": 211070
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.6779510974884033,
      "learning_rate": 1.0362205375592136e-05,
      "loss": 2.0948,
      "step": 211071
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.6148548126220703,
      "learning_rate": 1.0361139590782197e-05,
      "loss": 2.0162,
      "step": 211072
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.4449419975280762,
      "learning_rate": 1.0360073859821615e-05,
      "loss": 2.1464,
      "step": 211073
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.3921200037002563,
      "learning_rate": 1.0359008182710526e-05,
      "loss": 2.0795,
      "step": 211074
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.4958477020263672,
      "learning_rate": 1.0357942559449162e-05,
      "loss": 2.0943,
      "step": 211075
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.5445808172225952,
      "learning_rate": 1.0356876990037755e-05,
      "loss": 2.024,
      "step": 211076
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.4743309020996094,
      "learning_rate": 1.0355811474476439e-05,
      "loss": 1.9953,
      "step": 211077
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.493887186050415,
      "learning_rate": 1.035474601276548e-05,
      "loss": 2.0743,
      "step": 211078
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.4166748523712158,
      "learning_rate": 1.0353680604905012e-05,
      "loss": 1.9701,
      "step": 211079
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.423333764076233,
      "learning_rate": 1.0352615250895302e-05,
      "loss": 1.927,
      "step": 211080
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.3420034646987915,
      "learning_rate": 1.0351549950736448e-05,
      "loss": 2.1245,
      "step": 211081
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.6839094161987305,
      "learning_rate": 1.0350484704428752e-05,
      "loss": 2.0049,
      "step": 211082
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.3723902702331543,
      "learning_rate": 1.0349419511972312e-05,
      "loss": 2.0118,
      "step": 211083
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.4839903116226196,
      "learning_rate": 1.0348354373367463e-05,
      "loss": 2.0297,
      "step": 211084
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.5875605344772339,
      "learning_rate": 1.0347289288614301e-05,
      "loss": 2.1647,
      "step": 211085
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.3768610954284668,
      "learning_rate": 1.0346224257712998e-05,
      "loss": 1.9348,
      "step": 211086
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.5742979049682617,
      "learning_rate": 1.034515928066385e-05,
      "loss": 2.2818,
      "step": 211087
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.366066575050354,
      "learning_rate": 1.0344094357466993e-05,
      "loss": 1.9298,
      "step": 211088
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.451332926750183,
      "learning_rate": 1.0343029488122589e-05,
      "loss": 1.9613,
      "step": 211089
    },
    {
      "epoch": 2.75,
      "grad_norm": 2.957716941833496,
      "learning_rate": 1.0341964672630942e-05,
      "loss": 2.0158,
      "step": 211090
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.5242927074432373,
      "learning_rate": 1.0340899910992185e-05,
      "loss": 1.993,
      "step": 211091
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.5765876770019531,
      "learning_rate": 1.0339835203206482e-05,
      "loss": 1.9701,
      "step": 211092
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.5131516456604004,
      "learning_rate": 1.0338770549274134e-05,
      "loss": 2.008,
      "step": 211093
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.6495158672332764,
      "learning_rate": 1.033770594919524e-05,
      "loss": 1.8982,
      "step": 211094
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.5199886560440063,
      "learning_rate": 1.0336641402970003e-05,
      "loss": 2.0082,
      "step": 211095
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.4513851404190063,
      "learning_rate": 1.033557691059872e-05,
      "loss": 1.9528,
      "step": 211096
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.4793968200683594,
      "learning_rate": 1.033451247208149e-05,
      "loss": 1.7312,
      "step": 211097
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.4449567794799805,
      "learning_rate": 1.0333448087418516e-05,
      "loss": 1.9731,
      "step": 211098
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.3837389945983887,
      "learning_rate": 1.0332383756610064e-05,
      "loss": 1.993,
      "step": 211099
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.4703351259231567,
      "learning_rate": 1.0331319479656231e-05,
      "loss": 2.1074,
      "step": 211100
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.494370937347412,
      "learning_rate": 1.0330255256557318e-05,
      "loss": 2.028,
      "step": 211101
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.4716269969940186,
      "learning_rate": 1.0329191087313493e-05,
      "loss": 2.0765,
      "step": 211102
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.5307341814041138,
      "learning_rate": 1.0328126971924955e-05,
      "loss": 1.8593,
      "step": 211103
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.4357023239135742,
      "learning_rate": 1.0327062910391804e-05,
      "loss": 1.7803,
      "step": 211104
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.3959283828735352,
      "learning_rate": 1.0325998902714405e-05,
      "loss": 1.9366,
      "step": 211105
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.601982593536377,
      "learning_rate": 1.0324934948892794e-05,
      "loss": 2.0918,
      "step": 211106
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.457518458366394,
      "learning_rate": 1.0323871048927302e-05,
      "loss": 1.8279,
      "step": 211107
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.4921668767929077,
      "learning_rate": 1.0322807202818029e-05,
      "loss": 1.8697,
      "step": 211108
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.518839716911316,
      "learning_rate": 1.0321743410565275e-05,
      "loss": 1.971,
      "step": 211109
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.4232804775238037,
      "learning_rate": 1.0320679672169142e-05,
      "loss": 1.8041,
      "step": 211110
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.4523077011108398,
      "learning_rate": 1.0319615987629859e-05,
      "loss": 1.9918,
      "step": 211111
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.5296112298965454,
      "learning_rate": 1.0318552356947595e-05,
      "loss": 1.6871,
      "step": 211112
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.4160025119781494,
      "learning_rate": 1.0317488780122651e-05,
      "loss": 1.8804,
      "step": 211113
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.4040828943252563,
      "learning_rate": 1.0316425257155091e-05,
      "loss": 1.882,
      "step": 211114
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.62965726852417,
      "learning_rate": 1.031536178804525e-05,
      "loss": 1.7866,
      "step": 211115
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.495147943496704,
      "learning_rate": 1.0314298372793161e-05,
      "loss": 2.0606,
      "step": 211116
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.4963783025741577,
      "learning_rate": 1.031323501139919e-05,
      "loss": 2.1645,
      "step": 211117
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.4641627073287964,
      "learning_rate": 1.031217170386337e-05,
      "loss": 2.1232,
      "step": 211118
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.649076223373413,
      "learning_rate": 1.0311108450186067e-05,
      "loss": 2.0183,
      "step": 211119
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.8123294115066528,
      "learning_rate": 1.031004525036735e-05,
      "loss": 2.0782,
      "step": 211120
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.411292314529419,
      "learning_rate": 1.0308982104407481e-05,
      "loss": 2.0358,
      "step": 211121
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.5181468725204468,
      "learning_rate": 1.0307919012306664e-05,
      "loss": 1.8371,
      "step": 211122
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.5314942598342896,
      "learning_rate": 1.0306855974065065e-05,
      "loss": 1.9936,
      "step": 211123
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.5310806035995483,
      "learning_rate": 1.0305792989682849e-05,
      "loss": 1.972,
      "step": 211124
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.4514551162719727,
      "learning_rate": 1.0304730059160283e-05,
      "loss": 2.0065,
      "step": 211125
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.4168269634246826,
      "learning_rate": 1.0303667182497532e-05,
      "loss": 1.7771,
      "step": 211126
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.4607360363006592,
      "learning_rate": 1.0302604359694799e-05,
      "loss": 1.7994,
      "step": 211127
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.4972994327545166,
      "learning_rate": 1.0301541590752282e-05,
      "loss": 2.0628,
      "step": 211128
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.438232421875,
      "learning_rate": 1.0300478875670181e-05,
      "loss": 2.0822,
      "step": 211129
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.4890234470367432,
      "learning_rate": 1.0299416214448663e-05,
      "loss": 2.0346,
      "step": 211130
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.3770664930343628,
      "learning_rate": 1.0298353607087994e-05,
      "loss": 2.1608,
      "step": 211131
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.6201199293136597,
      "learning_rate": 1.0297291053588308e-05,
      "loss": 2.1074,
      "step": 211132
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.5167827606201172,
      "learning_rate": 1.0296228553949837e-05,
      "loss": 2.1077,
      "step": 211133
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.3246127367019653,
      "learning_rate": 1.0295166108172748e-05,
      "loss": 2.0698,
      "step": 211134
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.7023793458938599,
      "learning_rate": 1.0294103716257274e-05,
      "loss": 2.2169,
      "step": 211135
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.7185806035995483,
      "learning_rate": 1.0293041378203548e-05,
      "loss": 2.0153,
      "step": 211136
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.6084415912628174,
      "learning_rate": 1.029197909401187e-05,
      "loss": 1.9915,
      "step": 211137
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.3866337537765503,
      "learning_rate": 1.029091686368234e-05,
      "loss": 2.0695,
      "step": 211138
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.4593216180801392,
      "learning_rate": 1.0289854687215227e-05,
      "loss": 1.9509,
      "step": 211139
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.4054858684539795,
      "learning_rate": 1.0288792564610726e-05,
      "loss": 1.9356,
      "step": 211140
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.339861273765564,
      "learning_rate": 1.0287730495868973e-05,
      "loss": 1.8432,
      "step": 211141
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.475123643875122,
      "learning_rate": 1.0286668480990168e-05,
      "loss": 1.9251,
      "step": 211142
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.4016393423080444,
      "learning_rate": 1.0285606519974576e-05,
      "loss": 2.1101,
      "step": 211143
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.4845950603485107,
      "learning_rate": 1.0284544612822332e-05,
      "loss": 2.0327,
      "step": 211144
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.5407261848449707,
      "learning_rate": 1.0283482759533702e-05,
      "loss": 2.1325,
      "step": 211145
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.3532317876815796,
      "learning_rate": 1.0282420960108816e-05,
      "loss": 1.9243,
      "step": 211146
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.458383321762085,
      "learning_rate": 1.0281359214547913e-05,
      "loss": 2.0054,
      "step": 211147
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.8619061708450317,
      "learning_rate": 1.0280297522851155e-05,
      "loss": 2.1297,
      "step": 211148
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.4556410312652588,
      "learning_rate": 1.0279235885018777e-05,
      "loss": 1.898,
      "step": 211149
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.3943290710449219,
      "learning_rate": 1.0278174301050912e-05,
      "loss": 1.9466,
      "step": 211150
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.425722360610962,
      "learning_rate": 1.027711277094786e-05,
      "loss": 1.7864,
      "step": 211151
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.4920555353164673,
      "learning_rate": 1.0276051294709753e-05,
      "loss": 1.8601,
      "step": 211152
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.4581865072250366,
      "learning_rate": 1.0274989872336792e-05,
      "loss": 2.166,
      "step": 211153
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.3818751573562622,
      "learning_rate": 1.0273928503829143e-05,
      "loss": 2.1748,
      "step": 211154
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.4685322046279907,
      "learning_rate": 1.0272867189187073e-05,
      "loss": 2.0754,
      "step": 211155
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.4935429096221924,
      "learning_rate": 1.0271805928410715e-05,
      "loss": 1.9814,
      "step": 211156
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.4479076862335205,
      "learning_rate": 1.0270744721500335e-05,
      "loss": 1.9381,
      "step": 211157
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.5419334173202515,
      "learning_rate": 1.0269683568456099e-05,
      "loss": 1.9736,
      "step": 211158
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.4694770574569702,
      "learning_rate": 1.0268622469278176e-05,
      "loss": 2.0389,
      "step": 211159
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.4067697525024414,
      "learning_rate": 1.0267561423966797e-05,
      "loss": 1.902,
      "step": 211160
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.529671311378479,
      "learning_rate": 1.0266500432522129e-05,
      "loss": 2.1003,
      "step": 211161
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.4008314609527588,
      "learning_rate": 1.0265439494944372e-05,
      "loss": 1.8837,
      "step": 211162
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.5092452764511108,
      "learning_rate": 1.0264378611233792e-05,
      "loss": 1.778,
      "step": 211163
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.5085606575012207,
      "learning_rate": 1.0263317781390523e-05,
      "loss": 1.9661,
      "step": 211164
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.6502571105957031,
      "learning_rate": 1.0262257005414764e-05,
      "loss": 2.0149,
      "step": 211165
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.5551577806472778,
      "learning_rate": 1.0261196283306683e-05,
      "loss": 2.0715,
      "step": 211166
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.5408949851989746,
      "learning_rate": 1.0260135615066545e-05,
      "loss": 1.9637,
      "step": 211167
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.4076931476593018,
      "learning_rate": 1.0259075000694516e-05,
      "loss": 1.6043,
      "step": 211168
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.784924864768982,
      "learning_rate": 1.0258014440190799e-05,
      "loss": 1.8441,
      "step": 211169
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.4285328388214111,
      "learning_rate": 1.0256953933555589e-05,
      "loss": 1.9816,
      "step": 211170
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.5684058666229248,
      "learning_rate": 1.0255893480789057e-05,
      "loss": 1.8618,
      "step": 211171
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.6048798561096191,
      "learning_rate": 1.0254833081891434e-05,
      "loss": 1.9433,
      "step": 211172
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.5265843868255615,
      "learning_rate": 1.025377273686292e-05,
      "loss": 2.0585,
      "step": 211173
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.5026466846466064,
      "learning_rate": 1.025271244570368e-05,
      "loss": 2.3148,
      "step": 211174
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.5757259130477905,
      "learning_rate": 1.0251652208413952e-05,
      "loss": 2.0076,
      "step": 211175
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.509699821472168,
      "learning_rate": 1.0250592024993897e-05,
      "loss": 1.9075,
      "step": 211176
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.4566603899002075,
      "learning_rate": 1.0249531895443718e-05,
      "loss": 1.9662,
      "step": 211177
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.3424336910247803,
      "learning_rate": 1.0248471819763649e-05,
      "loss": 1.8009,
      "step": 211178
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.3549855947494507,
      "learning_rate": 1.0247411797953852e-05,
      "loss": 1.8948,
      "step": 211179
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.3291417360305786,
      "learning_rate": 1.0246351830014498e-05,
      "loss": 2.17,
      "step": 211180
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.3231918811798096,
      "learning_rate": 1.0245291915945852e-05,
      "loss": 2.0364,
      "step": 211181
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.4957431554794312,
      "learning_rate": 1.024423205574808e-05,
      "loss": 2.0934,
      "step": 211182
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.4974240064620972,
      "learning_rate": 1.0243172249421349e-05,
      "loss": 2.0613,
      "step": 211183
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.3840709924697876,
      "learning_rate": 1.0242112496965893e-05,
      "loss": 1.8607,
      "step": 211184
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.5253323316574097,
      "learning_rate": 1.0241052798381877e-05,
      "loss": 1.959,
      "step": 211185
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.4117120504379272,
      "learning_rate": 1.0239993153669535e-05,
      "loss": 1.9305,
      "step": 211186
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.466415524482727,
      "learning_rate": 1.0238933562829066e-05,
      "loss": 2.0938,
      "step": 211187
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.4777978658676147,
      "learning_rate": 1.0237874025860603e-05,
      "loss": 1.9212,
      "step": 211188
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.5010342597961426,
      "learning_rate": 1.0236814542764449e-05,
      "loss": 1.8886,
      "step": 211189
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.5052974224090576,
      "learning_rate": 1.0235755113540733e-05,
      "loss": 2.0513,
      "step": 211190
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.501768946647644,
      "learning_rate": 1.0234695738189591e-05,
      "loss": 1.8111,
      "step": 211191
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.4768906831741333,
      "learning_rate": 1.0233636416711355e-05,
      "loss": 1.9623,
      "step": 211192
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.4078998565673828,
      "learning_rate": 1.0232577149106125e-05,
      "loss": 2.1377,
      "step": 211193
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.460750937461853,
      "learning_rate": 1.0231517935374133e-05,
      "loss": 1.8604,
      "step": 211194
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.5334055423736572,
      "learning_rate": 1.0230458775515616e-05,
      "loss": 2.2477,
      "step": 211195
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.607334852218628,
      "learning_rate": 1.0229399669530703e-05,
      "loss": 1.9514,
      "step": 211196
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.5230578184127808,
      "learning_rate": 1.0228340617419562e-05,
      "loss": 1.9794,
      "step": 211197
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.576489806175232,
      "learning_rate": 1.0227281619182493e-05,
      "loss": 1.8629,
      "step": 211198
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.4932887554168701,
      "learning_rate": 1.0226222674819628e-05,
      "loss": 2.2327,
      "step": 211199
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.4872604608535767,
      "learning_rate": 1.022516378433117e-05,
      "loss": 1.8551,
      "step": 211200
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.475726842880249,
      "learning_rate": 1.0224104947717348e-05,
      "loss": 1.9482,
      "step": 211201
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.6752660274505615,
      "learning_rate": 1.0223046164978333e-05,
      "loss": 2.1035,
      "step": 211202
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.4619038105010986,
      "learning_rate": 1.0221987436114287e-05,
      "loss": 1.9771,
      "step": 211203
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.4208455085754395,
      "learning_rate": 1.0220928761125479e-05,
      "loss": 1.8436,
      "step": 211204
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.499139666557312,
      "learning_rate": 1.0219870140012044e-05,
      "loss": 1.9823,
      "step": 211205
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.4323810338974,
      "learning_rate": 1.0218811572774244e-05,
      "loss": 1.8566,
      "step": 211206
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.7472690343856812,
      "learning_rate": 1.0217753059412214e-05,
      "loss": 2.0681,
      "step": 211207
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.4464906454086304,
      "learning_rate": 1.021669459992619e-05,
      "loss": 2.0069,
      "step": 211208
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.3546671867370605,
      "learning_rate": 1.0215636194316335e-05,
      "loss": 1.8145,
      "step": 211209
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.325910210609436,
      "learning_rate": 1.0214577842582849e-05,
      "loss": 2.1884,
      "step": 211210
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.4449576139450073,
      "learning_rate": 1.0213519544725935e-05,
      "loss": 1.9505,
      "step": 211211
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.3930895328521729,
      "learning_rate": 1.0212461300745856e-05,
      "loss": 1.9162,
      "step": 211212
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.454276442527771,
      "learning_rate": 1.0211403110642713e-05,
      "loss": 2.1095,
      "step": 211213
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.441941499710083,
      "learning_rate": 1.0210344974416773e-05,
      "loss": 1.9346,
      "step": 211214
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.417523741722107,
      "learning_rate": 1.0209286892068136e-05,
      "loss": 2.0123,
      "step": 211215
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.3422943353652954,
      "learning_rate": 1.0208228863597102e-05,
      "loss": 1.9503,
      "step": 211216
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.4052135944366455,
      "learning_rate": 1.0207170889003802e-05,
      "loss": 1.9322,
      "step": 211217
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.4985157251358032,
      "learning_rate": 1.0206112968288505e-05,
      "loss": 2.0467,
      "step": 211218
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.5375924110412598,
      "learning_rate": 1.0205055101451342e-05,
      "loss": 2.2581,
      "step": 211219
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.4556463956832886,
      "learning_rate": 1.0203997288492548e-05,
      "loss": 2.1326,
      "step": 211220
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.5249385833740234,
      "learning_rate": 1.0202939529412257e-05,
      "loss": 2.0969,
      "step": 211221
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.5813047885894775,
      "learning_rate": 1.0201881824210733e-05,
      "loss": 2.0489,
      "step": 211222
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.374614953994751,
      "learning_rate": 1.0200824172888145e-05,
      "loss": 1.9884,
      "step": 211223
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.3386338949203491,
      "learning_rate": 1.019976657544469e-05,
      "loss": 2.045,
      "step": 211224
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.402998447418213,
      "learning_rate": 1.0198709031880603e-05,
      "loss": 1.9913,
      "step": 211225
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.423355221748352,
      "learning_rate": 1.0197651542196017e-05,
      "loss": 2.1041,
      "step": 211226
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.4076619148254395,
      "learning_rate": 1.0196594106391131e-05,
      "loss": 1.9781,
      "step": 211227
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.4157589673995972,
      "learning_rate": 1.0195536724466214e-05,
      "loss": 2.09,
      "step": 211228
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.446665644645691,
      "learning_rate": 1.0194479396421395e-05,
      "loss": 1.847,
      "step": 211229
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.473087191581726,
      "learning_rate": 1.0193422122256912e-05,
      "loss": 1.7723,
      "step": 211230
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.3907346725463867,
      "learning_rate": 1.0192364901972926e-05,
      "loss": 1.9545,
      "step": 211231
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.4522459506988525,
      "learning_rate": 1.0191307735569675e-05,
      "loss": 1.9469,
      "step": 211232
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.6595001220703125,
      "learning_rate": 1.0190250623047291e-05,
      "loss": 1.9998,
      "step": 211233
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.5727022886276245,
      "learning_rate": 1.018919356440604e-05,
      "loss": 1.8255,
      "step": 211234
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.518542766571045,
      "learning_rate": 1.0188136559646054e-05,
      "loss": 2.0874,
      "step": 211235
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.29511296749115,
      "learning_rate": 1.01870796087676e-05,
      "loss": 1.7518,
      "step": 211236
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.5424383878707886,
      "learning_rate": 1.0186022711770848e-05,
      "loss": 2.0475,
      "step": 211237
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.6680400371551514,
      "learning_rate": 1.018496586865596e-05,
      "loss": 2.1316,
      "step": 211238
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.4058095216751099,
      "learning_rate": 1.0183909079423136e-05,
      "loss": 1.9749,
      "step": 211239
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.6141610145568848,
      "learning_rate": 1.0182852344072644e-05,
      "loss": 1.7824,
      "step": 211240
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.4566435813903809,
      "learning_rate": 1.0181795662604585e-05,
      "loss": 1.7978,
      "step": 211241
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.7171399593353271,
      "learning_rate": 1.0180739035019226e-05,
      "loss": 1.9562,
      "step": 211242
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.3859091997146606,
      "learning_rate": 1.017968246131673e-05,
      "loss": 1.7789,
      "step": 211243
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.4244623184204102,
      "learning_rate": 1.0178625941497332e-05,
      "loss": 1.946,
      "step": 211244
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.5810590982437134,
      "learning_rate": 1.0177569475561131e-05,
      "loss": 2.0153,
      "step": 211245
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.409970760345459,
      "learning_rate": 1.0176513063508462e-05,
      "loss": 1.7521,
      "step": 211246
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.4182155132293701,
      "learning_rate": 1.0175456705339391e-05,
      "loss": 1.9986,
      "step": 211247
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.645189881324768,
      "learning_rate": 1.0174400401054217e-05,
      "loss": 2.1427,
      "step": 211248
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.362396001815796,
      "learning_rate": 1.0173344150653073e-05,
      "loss": 1.9645,
      "step": 211249
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.502476453781128,
      "learning_rate": 1.0172287954136193e-05,
      "loss": 2.0592,
      "step": 211250
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.4662748575210571,
      "learning_rate": 1.017123181150371e-05,
      "loss": 1.9827,
      "step": 211251
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.4720237255096436,
      "learning_rate": 1.0170175722755924e-05,
      "loss": 1.9259,
      "step": 211252
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.3672876358032227,
      "learning_rate": 1.0169119687892935e-05,
      "loss": 1.8665,
      "step": 211253
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.6171108484268188,
      "learning_rate": 1.0168063706914976e-05,
      "loss": 1.9838,
      "step": 211254
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.432934045791626,
      "learning_rate": 1.0167007779822245e-05,
      "loss": 1.9461,
      "step": 211255
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.4846093654632568,
      "learning_rate": 1.0165951906615012e-05,
      "loss": 1.8701,
      "step": 211256
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.422501802444458,
      "learning_rate": 1.0164896087293305e-05,
      "loss": 1.7698,
      "step": 211257
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.3715007305145264,
      "learning_rate": 1.0163840321857463e-05,
      "loss": 2.0499,
      "step": 211258
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.5074831247329712,
      "learning_rate": 1.0162784610307584e-05,
      "loss": 1.8628,
      "step": 211259
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.4089747667312622,
      "learning_rate": 1.0161728952643966e-05,
      "loss": 2.095,
      "step": 211260
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.477005958557129,
      "learning_rate": 1.0160673348866744e-05,
      "loss": 2.1077,
      "step": 211261
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.3561992645263672,
      "learning_rate": 1.0159617798976116e-05,
      "loss": 2.0183,
      "step": 211262
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.4567945003509521,
      "learning_rate": 1.0158562302972285e-05,
      "loss": 2.0193,
      "step": 211263
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.5061355829238892,
      "learning_rate": 1.015750686085548e-05,
      "loss": 1.9864,
      "step": 211264
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.4326701164245605,
      "learning_rate": 1.0156451472625803e-05,
      "loss": 1.7006,
      "step": 211265
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.4674098491668701,
      "learning_rate": 1.0155396138283557e-05,
      "loss": 2.1007,
      "step": 211266
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.4395155906677246,
      "learning_rate": 1.0154340857828868e-05,
      "loss": 2.0233,
      "step": 211267
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.4317981004714966,
      "learning_rate": 1.015328563126201e-05,
      "loss": 1.9038,
      "step": 211268
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.500476598739624,
      "learning_rate": 1.0152230458583076e-05,
      "loss": 2.1744,
      "step": 211269
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.4286025762557983,
      "learning_rate": 1.0151175339792305e-05,
      "loss": 1.88,
      "step": 211270
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.599783182144165,
      "learning_rate": 1.0150120274889962e-05,
      "loss": 2.2645,
      "step": 211271
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.6111187934875488,
      "learning_rate": 1.0149065263876144e-05,
      "loss": 2.0638,
      "step": 211272
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.3839259147644043,
      "learning_rate": 1.0148010306751087e-05,
      "loss": 2.05,
      "step": 211273
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.5107427835464478,
      "learning_rate": 1.014695540351499e-05,
      "loss": 2.0935,
      "step": 211274
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.4348554611206055,
      "learning_rate": 1.0145900554168052e-05,
      "loss": 2.0539,
      "step": 211275
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.4764041900634766,
      "learning_rate": 1.014484575871044e-05,
      "loss": 1.9714,
      "step": 211276
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.5493934154510498,
      "learning_rate": 1.0143791017142388e-05,
      "loss": 2.0672,
      "step": 211277
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.4302877187728882,
      "learning_rate": 1.0142736329464096e-05,
      "loss": 2.0859,
      "step": 211278
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.4497522115707397,
      "learning_rate": 1.0141681695675696e-05,
      "loss": 1.8278,
      "step": 211279
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.4980241060256958,
      "learning_rate": 1.014062711577749e-05,
      "loss": 2.0612,
      "step": 211280
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.5443295240402222,
      "learning_rate": 1.0139572589769573e-05,
      "loss": 1.7593,
      "step": 211281
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.4725861549377441,
      "learning_rate": 1.013851811765215e-05,
      "loss": 1.7925,
      "step": 211282
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.444993495941162,
      "learning_rate": 1.013746369942552e-05,
      "loss": 2.118,
      "step": 211283
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.4312186241149902,
      "learning_rate": 1.0136409335089745e-05,
      "loss": 1.8652,
      "step": 211284
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.4537235498428345,
      "learning_rate": 1.0135355024645098e-05,
      "loss": 1.9771,
      "step": 211285
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.4335122108459473,
      "learning_rate": 1.0134300768091808e-05,
      "loss": 2.0801,
      "step": 211286
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.4520941972732544,
      "learning_rate": 1.0133246565429975e-05,
      "loss": 1.8862,
      "step": 211287
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.393807053565979,
      "learning_rate": 1.0132192416659834e-05,
      "loss": 1.8518,
      "step": 211288
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.4194947481155396,
      "learning_rate": 1.0131138321781617e-05,
      "loss": 1.8445,
      "step": 211289
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.5819401741027832,
      "learning_rate": 1.0130084280795492e-05,
      "loss": 2.1009,
      "step": 211290
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.4753835201263428,
      "learning_rate": 1.0129030293701657e-05,
      "loss": 1.7248,
      "step": 211291
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.4646650552749634,
      "learning_rate": 1.0127976360500312e-05,
      "loss": 2.018,
      "step": 211292
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.3302150964736938,
      "learning_rate": 1.0126922481191624e-05,
      "loss": 2.1008,
      "step": 211293
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.4552576541900635,
      "learning_rate": 1.0125868655775827e-05,
      "loss": 2.1774,
      "step": 211294
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.411806344985962,
      "learning_rate": 1.0124814884253084e-05,
      "loss": 1.8681,
      "step": 211295
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.4510290622711182,
      "learning_rate": 1.0123761166623633e-05,
      "loss": 2.0276,
      "step": 211296
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.3951891660690308,
      "learning_rate": 1.012270750288764e-05,
      "loss": 2.0342,
      "step": 211297
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.541879415512085,
      "learning_rate": 1.0121653893045334e-05,
      "loss": 2.1067,
      "step": 211298
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.4627206325531006,
      "learning_rate": 1.012060033709685e-05,
      "loss": 1.8248,
      "step": 211299
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.3458006381988525,
      "learning_rate": 1.0119546835042425e-05,
      "loss": 1.8811,
      "step": 211300
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.5893959999084473,
      "learning_rate": 1.0118493386882254e-05,
      "loss": 2.1618,
      "step": 211301
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.4761128425598145,
      "learning_rate": 1.0117439992616505e-05,
      "loss": 2.1294,
      "step": 211302
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.6902889013290405,
      "learning_rate": 1.0116386652245412e-05,
      "loss": 1.8857,
      "step": 211303
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.4053374528884888,
      "learning_rate": 1.0115333365769174e-05,
      "loss": 2.1155,
      "step": 211304
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.4264695644378662,
      "learning_rate": 1.0114280133187958e-05,
      "loss": 2.0119,
      "step": 211305
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.4012377262115479,
      "learning_rate": 1.0113226954501929e-05,
      "loss": 1.9949,
      "step": 211306
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.6056294441223145,
      "learning_rate": 1.011217382971139e-05,
      "loss": 2.1496,
      "step": 211307
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.5272921323776245,
      "learning_rate": 1.0111120758816404e-05,
      "loss": 2.0015,
      "step": 211308
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.5018017292022705,
      "learning_rate": 1.0110067741817274e-05,
      "loss": 1.9284,
      "step": 211309
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.6587467193603516,
      "learning_rate": 1.0109014778714164e-05,
      "loss": 2.0004,
      "step": 211310
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.5383765697479248,
      "learning_rate": 1.0107961869507242e-05,
      "loss": 1.9908,
      "step": 211311
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.744972825050354,
      "learning_rate": 1.0106909014196707e-05,
      "loss": 1.8699,
      "step": 211312
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.4059014320373535,
      "learning_rate": 1.0105856212782793e-05,
      "loss": 2.0216,
      "step": 211313
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.5606987476348877,
      "learning_rate": 1.0104803465265632e-05,
      "loss": 2.0019,
      "step": 211314
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.6973923444747925,
      "learning_rate": 1.0103750771645492e-05,
      "loss": 2.1017,
      "step": 211315
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.4489529132843018,
      "learning_rate": 1.0102698131922571e-05,
      "loss": 1.8556,
      "step": 211316
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.4846807718276978,
      "learning_rate": 1.0101645546097003e-05,
      "loss": 1.7201,
      "step": 211317
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.372862458229065,
      "learning_rate": 1.010059301416899e-05,
      "loss": 1.9462,
      "step": 211318
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.5434457063674927,
      "learning_rate": 1.0099540536138762e-05,
      "loss": 2.0732,
      "step": 211319
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.4350773096084595,
      "learning_rate": 1.0098488112006486e-05,
      "loss": 1.9441,
      "step": 211320
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.3376713991165161,
      "learning_rate": 1.0097435741772397e-05,
      "loss": 1.7318,
      "step": 211321
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.5768994092941284,
      "learning_rate": 1.0096383425436627e-05,
      "loss": 2.0107,
      "step": 211322
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.5038188695907593,
      "learning_rate": 1.009533116299951e-05,
      "loss": 1.9556,
      "step": 211323
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.535750150680542,
      "learning_rate": 1.0094278954461044e-05,
      "loss": 1.9311,
      "step": 211324
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.39802885055542,
      "learning_rate": 1.0093226799821563e-05,
      "loss": 1.8524,
      "step": 211325
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.409074306488037,
      "learning_rate": 1.0092174699081201e-05,
      "loss": 2.0395,
      "step": 211326
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.4466948509216309,
      "learning_rate": 1.0091122652240224e-05,
      "loss": 2.0909,
      "step": 211327
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.4799587726593018,
      "learning_rate": 1.0090070659298699e-05,
      "loss": 2.045,
      "step": 211328
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.4352266788482666,
      "learning_rate": 1.0089018720257025e-05,
      "loss": 1.9738,
      "step": 211329
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.4734833240509033,
      "learning_rate": 1.008796683511517e-05,
      "loss": 2.1286,
      "step": 211330
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.3867794275283813,
      "learning_rate": 1.0086915003873464e-05,
      "loss": 1.8705,
      "step": 211331
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.6716419458389282,
      "learning_rate": 1.0085863226532075e-05,
      "loss": 2.0409,
      "step": 211332
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.67042875289917,
      "learning_rate": 1.0084811503091206e-05,
      "loss": 2.1351,
      "step": 211333
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.3911131620407104,
      "learning_rate": 1.008375983355102e-05,
      "loss": 2.0788,
      "step": 211334
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.4219400882720947,
      "learning_rate": 1.0082708217911783e-05,
      "loss": 1.9793,
      "step": 211335
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.3976774215698242,
      "learning_rate": 1.0081656656173598e-05,
      "loss": 1.8035,
      "step": 211336
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.493944764137268,
      "learning_rate": 1.008060514833673e-05,
      "loss": 2.104,
      "step": 211337
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.3997981548309326,
      "learning_rate": 1.007955369440131e-05,
      "loss": 2.2162,
      "step": 211338
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.3539445400238037,
      "learning_rate": 1.0078502294367607e-05,
      "loss": 1.8144,
      "step": 211339
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.3726006746292114,
      "learning_rate": 1.0077450948235754e-05,
      "loss": 1.9524,
      "step": 211340
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.5615549087524414,
      "learning_rate": 1.0076399656006051e-05,
      "loss": 1.9502,
      "step": 211341
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.5780314207077026,
      "learning_rate": 1.0075348417678565e-05,
      "loss": 2.1106,
      "step": 211342
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.4124492406845093,
      "learning_rate": 1.0074297233253525e-05,
      "loss": 1.7594,
      "step": 211343
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.4736826419830322,
      "learning_rate": 1.0073246102731169e-05,
      "loss": 2.0303,
      "step": 211344
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.4167155027389526,
      "learning_rate": 1.0072195026111662e-05,
      "loss": 1.9349,
      "step": 211345
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.4616597890853882,
      "learning_rate": 1.0071144003395204e-05,
      "loss": 1.9162,
      "step": 211346
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.5260505676269531,
      "learning_rate": 1.0070093034581995e-05,
      "loss": 1.9159,
      "step": 211347
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.4423420429229736,
      "learning_rate": 1.0069042119672265e-05,
      "loss": 1.8362,
      "step": 211348
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.4574341773986816,
      "learning_rate": 1.0067991258666152e-05,
      "loss": 2.0848,
      "step": 211349
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.4123557806015015,
      "learning_rate": 1.0066940451563853e-05,
      "loss": 2.0213,
      "step": 211350
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.497009038925171,
      "learning_rate": 1.0065889698365603e-05,
      "loss": 2.0946,
      "step": 211351
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.539842963218689,
      "learning_rate": 1.0064838999071534e-05,
      "loss": 1.9416,
      "step": 211352
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.4589124917984009,
      "learning_rate": 1.0063788353681945e-05,
      "loss": 2.0996,
      "step": 211353
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.6072555780410767,
      "learning_rate": 1.0062737762196937e-05,
      "loss": 2.056,
      "step": 211354
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.5706987380981445,
      "learning_rate": 1.006168722461671e-05,
      "loss": 1.8617,
      "step": 211355
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.5244585275650024,
      "learning_rate": 1.0060636740941564e-05,
      "loss": 1.8268,
      "step": 211356
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.271752953529358,
      "learning_rate": 1.0059586311171564e-05,
      "loss": 1.8297,
      "step": 211357
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.5559132099151611,
      "learning_rate": 1.0058535935306977e-05,
      "loss": 1.8062,
      "step": 211358
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.3877923488616943,
      "learning_rate": 1.005748561334797e-05,
      "loss": 1.8229,
      "step": 211359
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.4479148387908936,
      "learning_rate": 1.0056435345294778e-05,
      "loss": 1.8929,
      "step": 211360
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.4728087186813354,
      "learning_rate": 1.005538513114753e-05,
      "loss": 2.1794,
      "step": 211361
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.5194989442825317,
      "learning_rate": 1.0054334970906497e-05,
      "loss": 2.1907,
      "step": 211362
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.4412821531295776,
      "learning_rate": 1.0053284864571808e-05,
      "loss": 1.9047,
      "step": 211363
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.3797264099121094,
      "learning_rate": 1.0052234812143667e-05,
      "loss": 2.0503,
      "step": 211364
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.4051613807678223,
      "learning_rate": 1.0051184813622337e-05,
      "loss": 1.9007,
      "step": 211365
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.4123085737228394,
      "learning_rate": 1.0050134869007954e-05,
      "loss": 1.8736,
      "step": 211366
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.6139531135559082,
      "learning_rate": 1.0049084978300714e-05,
      "loss": 2.0797,
      "step": 211367
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.4837993383407593,
      "learning_rate": 1.0048035141500821e-05,
      "loss": 1.924,
      "step": 211368
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.5953787565231323,
      "learning_rate": 1.004698535860844e-05,
      "loss": 1.9997,
      "step": 211369
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.3101253509521484,
      "learning_rate": 1.0045935629623836e-05,
      "loss": 1.921,
      "step": 211370
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.546194076538086,
      "learning_rate": 1.0044885954547176e-05,
      "loss": 1.9245,
      "step": 211371
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.435172200202942,
      "learning_rate": 1.0043836333378663e-05,
      "loss": 1.7824,
      "step": 211372
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.5064576864242554,
      "learning_rate": 1.0042786766118393e-05,
      "loss": 1.9789,
      "step": 211373
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.7394013404846191,
      "learning_rate": 1.0041737252766734e-05,
      "loss": 1.9709,
      "step": 211374
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.5253831148147583,
      "learning_rate": 1.0040687793323721e-05,
      "loss": 1.9168,
      "step": 211375
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.5062614679336548,
      "learning_rate": 1.003963838778965e-05,
      "loss": 2.0256,
      "step": 211376
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.3592334985733032,
      "learning_rate": 1.003858903616469e-05,
      "loss": 1.9421,
      "step": 211377
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.4022237062454224,
      "learning_rate": 1.0037539738449007e-05,
      "loss": 2.1019,
      "step": 211378
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.4639755487442017,
      "learning_rate": 1.0036490494642801e-05,
      "loss": 1.9731,
      "step": 211379
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.6510652303695679,
      "learning_rate": 1.0035441304746338e-05,
      "loss": 2.1,
      "step": 211380
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.5084353685379028,
      "learning_rate": 1.003439216875972e-05,
      "loss": 2.0164,
      "step": 211381
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.5930272340774536,
      "learning_rate": 1.0033343086683209e-05,
      "loss": 2.0314,
      "step": 211382
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.4031710624694824,
      "learning_rate": 1.0032294058516977e-05,
      "loss": 1.984,
      "step": 211383
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.4282112121582031,
      "learning_rate": 1.0031245084261185e-05,
      "loss": 1.9834,
      "step": 211384
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.3883589506149292,
      "learning_rate": 1.0030196163916071e-05,
      "loss": 1.9939,
      "step": 211385
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.542302131652832,
      "learning_rate": 1.0029147297481798e-05,
      "loss": 1.8799,
      "step": 211386
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.6125212907791138,
      "learning_rate": 1.0028098484958601e-05,
      "loss": 1.9292,
      "step": 211387
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.3688980340957642,
      "learning_rate": 1.0027049726346647e-05,
      "loss": 1.8404,
      "step": 211388
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.5310441255569458,
      "learning_rate": 1.0026001021646135e-05,
      "loss": 2.084,
      "step": 211389
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.5131826400756836,
      "learning_rate": 1.0024952370857298e-05,
      "loss": 1.9275,
      "step": 211390
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.4620754718780518,
      "learning_rate": 1.0023903773980268e-05,
      "loss": 1.8865,
      "step": 211391
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.513620376586914,
      "learning_rate": 1.002285523101528e-05,
      "loss": 2.0009,
      "step": 211392
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.4856595993041992,
      "learning_rate": 1.0021806741962468e-05,
      "loss": 2.0117,
      "step": 211393
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.5133613348007202,
      "learning_rate": 1.002075830682213e-05,
      "loss": 1.924,
      "step": 211394
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.5512419939041138,
      "learning_rate": 1.0019709925594365e-05,
      "loss": 1.8834,
      "step": 211395
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.4543565511703491,
      "learning_rate": 1.001866159827951e-05,
      "loss": 1.7178,
      "step": 211396
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.566698670387268,
      "learning_rate": 1.001761332487756e-05,
      "loss": 1.9522,
      "step": 211397
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.5193853378295898,
      "learning_rate": 1.001656510538882e-05,
      "loss": 2.1723,
      "step": 211398
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.4108766317367554,
      "learning_rate": 1.0015516939813483e-05,
      "loss": 1.9797,
      "step": 211399
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.426072359085083,
      "learning_rate": 1.0014468828151756e-05,
      "loss": 1.9783,
      "step": 211400
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.3675482273101807,
      "learning_rate": 1.0013420770403768e-05,
      "loss": 1.8939,
      "step": 211401
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.5901182889938354,
      "learning_rate": 1.0012372766569855e-05,
      "loss": 2.1174,
      "step": 211402
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.4590742588043213,
      "learning_rate": 1.0011324816650012e-05,
      "loss": 1.9818,
      "step": 211403
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.5583354234695435,
      "learning_rate": 1.0010276920644578e-05,
      "loss": 1.8252,
      "step": 211404
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.5092524290084839,
      "learning_rate": 1.0009229078553681e-05,
      "loss": 1.8751,
      "step": 211405
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.3513826131820679,
      "learning_rate": 1.0008181290377592e-05,
      "loss": 1.9356,
      "step": 211406
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.3476351499557495,
      "learning_rate": 1.0007133556116408e-05,
      "loss": 1.8304,
      "step": 211407
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.4031035900115967,
      "learning_rate": 1.000608587577043e-05,
      "loss": 2.0676,
      "step": 211408
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.6944001913070679,
      "learning_rate": 1.0005038249339758e-05,
      "loss": 1.8985,
      "step": 211409
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.481370449066162,
      "learning_rate": 1.0003990676824625e-05,
      "loss": 1.9622,
      "step": 211410
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.5420805215835571,
      "learning_rate": 1.0002943158225197e-05,
      "loss": 2.1448,
      "step": 211411
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.400490164756775,
      "learning_rate": 1.0001895693541739e-05,
      "loss": 1.902,
      "step": 211412
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.4591927528381348,
      "learning_rate": 1.0000848282774388e-05,
      "loss": 1.9738,
      "step": 211413
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.8069112300872803,
      "learning_rate": 9.999800925923374e-06,
      "loss": 1.9453,
      "step": 211414
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.4475452899932861,
      "learning_rate": 9.998753622988831e-06,
      "loss": 2.0272,
      "step": 211415
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.4883228540420532,
      "learning_rate": 9.997706373971027e-06,
      "loss": 2.0842,
      "step": 211416
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.4371145963668823,
      "learning_rate": 9.996659178870092e-06,
      "loss": 1.911,
      "step": 211417
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.5493887662887573,
      "learning_rate": 9.995612037686296e-06,
      "loss": 1.9377,
      "step": 211418
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.6009620428085327,
      "learning_rate": 9.994564950419736e-06,
      "loss": 2.0527,
      "step": 211419
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.4995903968811035,
      "learning_rate": 9.993517917070715e-06,
      "loss": 2.1239,
      "step": 211420
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.4787455797195435,
      "learning_rate": 9.992470937639329e-06,
      "loss": 1.9983,
      "step": 211421
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.4071600437164307,
      "learning_rate": 9.991424012125849e-06,
      "loss": 2.0089,
      "step": 211422
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.4247257709503174,
      "learning_rate": 9.990377140530403e-06,
      "loss": 1.9813,
      "step": 211423
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.4284476041793823,
      "learning_rate": 9.989330322853262e-06,
      "loss": 1.8956,
      "step": 211424
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.5394326448440552,
      "learning_rate": 9.988283559094523e-06,
      "loss": 1.9019,
      "step": 211425
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.413425326347351,
      "learning_rate": 9.987236849254522e-06,
      "loss": 1.9788,
      "step": 211426
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.352096676826477,
      "learning_rate": 9.986190193333289e-06,
      "loss": 2.1491,
      "step": 211427
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.3863427639007568,
      "learning_rate": 9.985143591331124e-06,
      "loss": 1.8954,
      "step": 211428
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.6227937936782837,
      "learning_rate": 9.984097043248196e-06,
      "loss": 1.9504,
      "step": 211429
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.408092737197876,
      "learning_rate": 9.983050549084703e-06,
      "loss": 1.9756,
      "step": 211430
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.3971152305603027,
      "learning_rate": 9.982004108840813e-06,
      "loss": 1.8252,
      "step": 211431
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.678839921951294,
      "learning_rate": 9.980957722516758e-06,
      "loss": 1.929,
      "step": 211432
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.3162974119186401,
      "learning_rate": 9.979911390112738e-06,
      "loss": 1.8574,
      "step": 211433
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.508682370185852,
      "learning_rate": 9.97886511162892e-06,
      "loss": 2.0406,
      "step": 211434
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.488689661026001,
      "learning_rate": 9.977818887065504e-06,
      "loss": 2.1176,
      "step": 211435
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.3229089975357056,
      "learning_rate": 9.976772716422687e-06,
      "loss": 1.9365,
      "step": 211436
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.5387893915176392,
      "learning_rate": 9.975726599700607e-06,
      "loss": 2.0864,
      "step": 211437
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.3897627592086792,
      "learning_rate": 9.974680536899592e-06,
      "loss": 1.9972,
      "step": 211438
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.428497314453125,
      "learning_rate": 9.973634528019746e-06,
      "loss": 1.9251,
      "step": 211439
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.7293891906738281,
      "learning_rate": 9.97258857306127e-06,
      "loss": 1.9802,
      "step": 211440
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.3936887979507446,
      "learning_rate": 9.971542672024323e-06,
      "loss": 1.9374,
      "step": 211441
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.373259425163269,
      "learning_rate": 9.97049682490918e-06,
      "loss": 1.9678,
      "step": 211442
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.3162658214569092,
      "learning_rate": 9.969451031715969e-06,
      "loss": 1.9532,
      "step": 211443
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.4444897174835205,
      "learning_rate": 9.968405292444925e-06,
      "loss": 1.9914,
      "step": 211444
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.5691211223602295,
      "learning_rate": 9.967359607096249e-06,
      "loss": 2.0241,
      "step": 211445
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.4051088094711304,
      "learning_rate": 9.966313975670071e-06,
      "loss": 1.9934,
      "step": 211446
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.4237033128738403,
      "learning_rate": 9.965268398166692e-06,
      "loss": 2.0147,
      "step": 211447
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.4818246364593506,
      "learning_rate": 9.964222874586215e-06,
      "loss": 1.9796,
      "step": 211448
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.4139074087142944,
      "learning_rate": 9.963177404928836e-06,
      "loss": 2.2076,
      "step": 211449
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.4411669969558716,
      "learning_rate": 9.96213198919482e-06,
      "loss": 1.9046,
      "step": 211450
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.578514575958252,
      "learning_rate": 9.961086627384307e-06,
      "loss": 2.1916,
      "step": 211451
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.634013295173645,
      "learning_rate": 9.960041319497458e-06,
      "loss": 1.8673,
      "step": 211452
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.5161981582641602,
      "learning_rate": 9.958996065534575e-06,
      "loss": 1.9439,
      "step": 211453
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.3593640327453613,
      "learning_rate": 9.957950865495723e-06,
      "loss": 1.8751,
      "step": 211454
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.481851577758789,
      "learning_rate": 9.956905719381236e-06,
      "loss": 2.1776,
      "step": 211455
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.4184354543685913,
      "learning_rate": 9.955860627191214e-06,
      "loss": 2.0524,
      "step": 211456
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.3918931484222412,
      "learning_rate": 9.954815588925858e-06,
      "loss": 2.1612,
      "step": 211457
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.5611858367919922,
      "learning_rate": 9.953770604585364e-06,
      "loss": 2.1534,
      "step": 211458
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.3987653255462646,
      "learning_rate": 9.952725674169971e-06,
      "loss": 1.9749,
      "step": 211459
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.3795170783996582,
      "learning_rate": 9.951680797679807e-06,
      "loss": 1.7793,
      "step": 211460
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.4767539501190186,
      "learning_rate": 9.95063597511514e-06,
      "loss": 1.849,
      "step": 211461
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.4704771041870117,
      "learning_rate": 9.949591206476105e-06,
      "loss": 1.743,
      "step": 211462
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.4921929836273193,
      "learning_rate": 9.948546491762932e-06,
      "loss": 2.1571,
      "step": 211463
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.6154800653457642,
      "learning_rate": 9.947501830975758e-06,
      "loss": 1.9763,
      "step": 211464
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.531026840209961,
      "learning_rate": 9.946457224114846e-06,
      "loss": 1.9236,
      "step": 211465
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.496871829032898,
      "learning_rate": 9.945412671180363e-06,
      "loss": 1.8878,
      "step": 211466
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.620131492614746,
      "learning_rate": 9.944368172172513e-06,
      "loss": 2.156,
      "step": 211467
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.4003385305404663,
      "learning_rate": 9.943323727091423e-06,
      "loss": 1.8505,
      "step": 211468
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.742799162864685,
      "learning_rate": 9.942279335937431e-06,
      "loss": 2.1209,
      "step": 211469
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.4927482604980469,
      "learning_rate": 9.9412349987106e-06,
      "loss": 1.7714,
      "step": 211470
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.5645778179168701,
      "learning_rate": 9.940190715411167e-06,
      "loss": 1.8582,
      "step": 211471
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.5008360147476196,
      "learning_rate": 9.939146486039328e-06,
      "loss": 2.1606,
      "step": 211472
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.3243794441223145,
      "learning_rate": 9.938102310595285e-06,
      "loss": 1.9397,
      "step": 211473
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.4011752605438232,
      "learning_rate": 9.937058189079206e-06,
      "loss": 1.8542,
      "step": 211474
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.555504560470581,
      "learning_rate": 9.936014121491353e-06,
      "loss": 1.8734,
      "step": 211475
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.4990034103393555,
      "learning_rate": 9.93497010783183e-06,
      "loss": 1.8705,
      "step": 211476
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.6878032684326172,
      "learning_rate": 9.933926148100868e-06,
      "loss": 2.1345,
      "step": 211477
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.4479014873504639,
      "learning_rate": 9.932882242298667e-06,
      "loss": 2.0803,
      "step": 211478
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.4758411645889282,
      "learning_rate": 9.931838390425428e-06,
      "loss": 2.07,
      "step": 211479
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.4529845714569092,
      "learning_rate": 9.930794592481317e-06,
      "loss": 1.9054,
      "step": 211480
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.4553929567337036,
      "learning_rate": 9.9297508484666e-06,
      "loss": 2.1679,
      "step": 211481
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.6126700639724731,
      "learning_rate": 9.928707158381344e-06,
      "loss": 2.1325,
      "step": 211482
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.5281552076339722,
      "learning_rate": 9.927663522225882e-06,
      "loss": 2.077,
      "step": 211483
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.4649438858032227,
      "learning_rate": 9.92661994000028e-06,
      "loss": 1.9982,
      "step": 211484
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.442731261253357,
      "learning_rate": 9.92557641170484e-06,
      "loss": 2.1121,
      "step": 211485
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.5056854486465454,
      "learning_rate": 9.92453293733969e-06,
      "loss": 1.9751,
      "step": 211486
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.4278782606124878,
      "learning_rate": 9.923489516905103e-06,
      "loss": 1.9043,
      "step": 211487
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.5494920015335083,
      "learning_rate": 9.922446150401143e-06,
      "loss": 2.2578,
      "step": 211488
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.4269771575927734,
      "learning_rate": 9.921402837828108e-06,
      "loss": 2.0097,
      "step": 211489
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.3459022045135498,
      "learning_rate": 9.920359579186132e-06,
      "loss": 1.9883,
      "step": 211490
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.3792387247085571,
      "learning_rate": 9.91931637447545e-06,
      "loss": 1.9473,
      "step": 211491
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.5471924543380737,
      "learning_rate": 9.918273223696227e-06,
      "loss": 1.7684,
      "step": 211492
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.4033340215682983,
      "learning_rate": 9.917230126848763e-06,
      "loss": 1.8914,
      "step": 211493
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.422648310661316,
      "learning_rate": 9.916187083933058e-06,
      "loss": 2.0507,
      "step": 211494
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.4849275350570679,
      "learning_rate": 9.915144094949445e-06,
      "loss": 2.1441,
      "step": 211495
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.5000022649765015,
      "learning_rate": 9.91410115989809e-06,
      "loss": 2.0114,
      "step": 211496
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.4735441207885742,
      "learning_rate": 9.913058278779162e-06,
      "loss": 1.7911,
      "step": 211497
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.568130373954773,
      "learning_rate": 9.912015451592858e-06,
      "loss": 1.8753,
      "step": 211498
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.4256654977798462,
      "learning_rate": 9.910972678339479e-06,
      "loss": 2.0187,
      "step": 211499
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.417720079421997,
      "learning_rate": 9.909929959019025e-06,
      "loss": 2.0381,
      "step": 211500
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.3855884075164795,
      "learning_rate": 9.908887293631829e-06,
      "loss": 2.1276,
      "step": 211501
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.3968318700790405,
      "learning_rate": 9.907844682178024e-06,
      "loss": 2.0367,
      "step": 211502
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.5331904888153076,
      "learning_rate": 9.906802124657842e-06,
      "loss": 2.0475,
      "step": 211503
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.4876413345336914,
      "learning_rate": 9.905759621071452e-06,
      "loss": 1.999,
      "step": 211504
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.3532367944717407,
      "learning_rate": 9.90471717141912e-06,
      "loss": 2.031,
      "step": 211505
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.487295389175415,
      "learning_rate": 9.90367477570091e-06,
      "loss": 1.9936,
      "step": 211506
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.4868518114089966,
      "learning_rate": 9.902632433917091e-06,
      "loss": 2.1354,
      "step": 211507
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.4030966758728027,
      "learning_rate": 9.901590146067828e-06,
      "loss": 2.1146,
      "step": 211508
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.4923512935638428,
      "learning_rate": 9.900547912153389e-06,
      "loss": 1.8154,
      "step": 211509
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.3521926403045654,
      "learning_rate": 9.899505732173873e-06,
      "loss": 1.9278,
      "step": 211510
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.5362999439239502,
      "learning_rate": 9.898463606129581e-06,
      "loss": 2.0947,
      "step": 211511
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.5703281164169312,
      "learning_rate": 9.897421534020545e-06,
      "loss": 1.9391,
      "step": 211512
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.5968469381332397,
      "learning_rate": 9.89637951584713e-06,
      "loss": 2.0079,
      "step": 211513
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.6159205436706543,
      "learning_rate": 9.895337551609373e-06,
      "loss": 2.2318,
      "step": 211514
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.4585047960281372,
      "learning_rate": 9.894295641307637e-06,
      "loss": 2.1356,
      "step": 211515
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.484872817993164,
      "learning_rate": 9.893253784941956e-06,
      "loss": 2.1015,
      "step": 211516
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.5181336402893066,
      "learning_rate": 9.892211982512632e-06,
      "loss": 2.2488,
      "step": 211517
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.4426945447921753,
      "learning_rate": 9.89117023401983e-06,
      "loss": 2.0521,
      "step": 211518
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.377961277961731,
      "learning_rate": 9.890128539463748e-06,
      "loss": 2.0521,
      "step": 211519
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.5504659414291382,
      "learning_rate": 9.889086898844523e-06,
      "loss": 1.9079,
      "step": 211520
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.4035699367523193,
      "learning_rate": 9.888045312162418e-06,
      "loss": 2.1543,
      "step": 211521
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.5527052879333496,
      "learning_rate": 9.887003779417568e-06,
      "loss": 2.0267,
      "step": 211522
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.3887050151824951,
      "learning_rate": 9.885962300610273e-06,
      "loss": 1.9995,
      "step": 211523
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.353675127029419,
      "learning_rate": 9.884920875740599e-06,
      "loss": 2.1789,
      "step": 211524
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.4429272413253784,
      "learning_rate": 9.883879504808812e-06,
      "loss": 1.9349,
      "step": 211525
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.4750704765319824,
      "learning_rate": 9.882838187815079e-06,
      "loss": 1.9454,
      "step": 211526
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.4457509517669678,
      "learning_rate": 9.881796924759599e-06,
      "loss": 1.907,
      "step": 211527
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.457202434539795,
      "learning_rate": 9.880755715642574e-06,
      "loss": 1.9063,
      "step": 211528
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.7232524156570435,
      "learning_rate": 9.879714560464203e-06,
      "loss": 2.102,
      "step": 211529
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.4143773317337036,
      "learning_rate": 9.878673459224684e-06,
      "loss": 2.1443,
      "step": 211530
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.381617546081543,
      "learning_rate": 9.877632411924153e-06,
      "loss": 1.9056,
      "step": 211531
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.3804024457931519,
      "learning_rate": 9.876591418562874e-06,
      "loss": 1.9633,
      "step": 211532
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.6067311763763428,
      "learning_rate": 9.875550479141014e-06,
      "loss": 1.9753,
      "step": 211533
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.5086761713027954,
      "learning_rate": 9.874509593658741e-06,
      "loss": 2.3177,
      "step": 211534
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.3330596685409546,
      "learning_rate": 9.873468762116287e-06,
      "loss": 1.7168,
      "step": 211535
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.5289171934127808,
      "learning_rate": 9.872427984513853e-06,
      "loss": 2.0357,
      "step": 211536
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.5069938898086548,
      "learning_rate": 9.871387260851571e-06,
      "loss": 2.0381,
      "step": 211537
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.3192414045333862,
      "learning_rate": 9.870346591129708e-06,
      "loss": 1.8943,
      "step": 211538
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.5376259088516235,
      "learning_rate": 9.86930597534843e-06,
      "loss": 2.1863,
      "step": 211539
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.7669376134872437,
      "learning_rate": 9.868265413507903e-06,
      "loss": 1.9758,
      "step": 211540
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.5451596975326538,
      "learning_rate": 9.867224905608362e-06,
      "loss": 2.142,
      "step": 211541
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.4310184717178345,
      "learning_rate": 9.866184451649972e-06,
      "loss": 1.9157,
      "step": 211542
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.4296942949295044,
      "learning_rate": 9.865144051632901e-06,
      "loss": 1.8016,
      "step": 211543
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.3995230197906494,
      "learning_rate": 9.864103705557414e-06,
      "loss": 1.969,
      "step": 211544
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.4889785051345825,
      "learning_rate": 9.863063413423643e-06,
      "loss": 1.9626,
      "step": 211545
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.4647176265716553,
      "learning_rate": 9.862023175231826e-06,
      "loss": 2.1958,
      "step": 211546
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.38957941532135,
      "learning_rate": 9.860982990982158e-06,
      "loss": 2.1089,
      "step": 211547
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.4224987030029297,
      "learning_rate": 9.859942860674774e-06,
      "loss": 1.8856,
      "step": 211548
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.442083477973938,
      "learning_rate": 9.858902784309907e-06,
      "loss": 2.0291,
      "step": 211549
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.5287851095199585,
      "learning_rate": 9.857862761887757e-06,
      "loss": 2.0578,
      "step": 211550
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.572494626045227,
      "learning_rate": 9.856822793408459e-06,
      "loss": 1.9144,
      "step": 211551
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.431277871131897,
      "learning_rate": 9.85578287887231e-06,
      "loss": 1.8202,
      "step": 211552
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.843973994255066,
      "learning_rate": 9.85474301827941e-06,
      "loss": 1.9877,
      "step": 211553
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.4197834730148315,
      "learning_rate": 9.853703211630059e-06,
      "loss": 1.8618,
      "step": 211554
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.5428073406219482,
      "learning_rate": 9.852663458924326e-06,
      "loss": 1.7095,
      "step": 211555
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.5833250284194946,
      "learning_rate": 9.851623760162475e-06,
      "loss": 1.8113,
      "step": 211556
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.472326636314392,
      "learning_rate": 9.850584115344672e-06,
      "loss": 2.0322,
      "step": 211557
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.4875476360321045,
      "learning_rate": 9.849544524471154e-06,
      "loss": 1.9262,
      "step": 211558
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.4836771488189697,
      "learning_rate": 9.848504987542017e-06,
      "loss": 2.0725,
      "step": 211559
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.5082461833953857,
      "learning_rate": 9.847465504557628e-06,
      "loss": 1.9329,
      "step": 211560
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.3409476280212402,
      "learning_rate": 9.846426075517989e-06,
      "loss": 2.0338,
      "step": 211561
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.5520601272583008,
      "learning_rate": 9.845386700423396e-06,
      "loss": 1.9601,
      "step": 211562
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.6636377573013306,
      "learning_rate": 9.844347379273986e-06,
      "loss": 2.1387,
      "step": 211563
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.8478368520736694,
      "learning_rate": 9.843308112070026e-06,
      "loss": 1.9063,
      "step": 211564
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.4046295881271362,
      "learning_rate": 9.842268898811678e-06,
      "loss": 2.1456,
      "step": 211565
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.4461088180541992,
      "learning_rate": 9.841229739499146e-06,
      "loss": 1.8606,
      "step": 211566
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.4553576707839966,
      "learning_rate": 9.840190634132561e-06,
      "loss": 1.9261,
      "step": 211567
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.4150203466415405,
      "learning_rate": 9.839151582712191e-06,
      "loss": 1.9035,
      "step": 211568
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.645240068435669,
      "learning_rate": 9.838112585238166e-06,
      "loss": 1.7831,
      "step": 211569
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.7612204551696777,
      "learning_rate": 9.837073641710757e-06,
      "loss": 1.9845,
      "step": 211570
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.4203617572784424,
      "learning_rate": 9.836034752130061e-06,
      "loss": 2.1051,
      "step": 211571
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.6059622764587402,
      "learning_rate": 9.834995916496413e-06,
      "loss": 1.9731,
      "step": 211572
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.586606740951538,
      "learning_rate": 9.833957134809844e-06,
      "loss": 1.8802,
      "step": 211573
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.4090466499328613,
      "learning_rate": 9.832918407070656e-06,
      "loss": 2.1216,
      "step": 211574
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.3600932359695435,
      "learning_rate": 9.831879733278947e-06,
      "loss": 1.9406,
      "step": 211575
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.4530503749847412,
      "learning_rate": 9.830841113435018e-06,
      "loss": 1.8574,
      "step": 211576
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.5669114589691162,
      "learning_rate": 9.829802547539001e-06,
      "loss": 1.9399,
      "step": 211577
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.4072884321212769,
      "learning_rate": 9.82876403559113e-06,
      "loss": 1.7611,
      "step": 211578
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.6281993389129639,
      "learning_rate": 9.827725577591539e-06,
      "loss": 1.8678,
      "step": 211579
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.3999223709106445,
      "learning_rate": 9.826687173540459e-06,
      "loss": 2.0055,
      "step": 211580
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.4469478130340576,
      "learning_rate": 9.82564882343806e-06,
      "loss": 1.7524,
      "step": 211581
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.4581667184829712,
      "learning_rate": 9.82461052728457e-06,
      "loss": 2.0457,
      "step": 211582
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.4471858739852905,
      "learning_rate": 9.823572285080128e-06,
      "loss": 2.0728,
      "step": 211583
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.4032304286956787,
      "learning_rate": 9.82253409682503e-06,
      "loss": 1.9631,
      "step": 211584
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.4992045164108276,
      "learning_rate": 9.821495962519344e-06,
      "loss": 1.9377,
      "step": 211585
    },
    {
      "epoch": 2.75,
      "grad_norm": 1.4815053939819336,
      "learning_rate": 9.820457882163335e-06,
      "loss": 2.1032,
      "step": 211586
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.4258108139038086,
      "learning_rate": 9.819419855757137e-06,
      "loss": 2.0524,
      "step": 211587
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.517871379852295,
      "learning_rate": 9.818381883301052e-06,
      "loss": 2.2505,
      "step": 211588
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.4108421802520752,
      "learning_rate": 9.817343964795177e-06,
      "loss": 1.785,
      "step": 211589
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.533555507659912,
      "learning_rate": 9.816306100239745e-06,
      "loss": 2.0653,
      "step": 211590
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.4648610353469849,
      "learning_rate": 9.815268289634925e-06,
      "loss": 2.0598,
      "step": 211591
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.4090189933776855,
      "learning_rate": 9.814230532980949e-06,
      "loss": 1.9243,
      "step": 211592
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.4972139596939087,
      "learning_rate": 9.813192830277949e-06,
      "loss": 2.1765,
      "step": 211593
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.5657516717910767,
      "learning_rate": 9.812155181526194e-06,
      "loss": 1.9921,
      "step": 211594
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.4937909841537476,
      "learning_rate": 9.811117586725814e-06,
      "loss": 1.907,
      "step": 211595
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.4533891677856445,
      "learning_rate": 9.810080045877044e-06,
      "loss": 1.965,
      "step": 211596
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.361162543296814,
      "learning_rate": 9.809042558980051e-06,
      "loss": 1.5803,
      "step": 211597
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.5206609964370728,
      "learning_rate": 9.808005126035035e-06,
      "loss": 2.2368,
      "step": 211598
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.4127198457717896,
      "learning_rate": 9.80696774704216e-06,
      "loss": 2.2141,
      "step": 211599
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.718320608139038,
      "learning_rate": 9.805930422001695e-06,
      "loss": 1.9907,
      "step": 211600
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.5711681842803955,
      "learning_rate": 9.80489315091374e-06,
      "loss": 2.0832,
      "step": 211601
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.5770293474197388,
      "learning_rate": 9.80385593377856e-06,
      "loss": 1.9307,
      "step": 211602
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.4733587503433228,
      "learning_rate": 9.802818770596321e-06,
      "loss": 2.158,
      "step": 211603
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.5522640943527222,
      "learning_rate": 9.801781661367225e-06,
      "loss": 1.9386,
      "step": 211604
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.3876832723617554,
      "learning_rate": 9.800744606091404e-06,
      "loss": 2.0411,
      "step": 211605
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.4475138187408447,
      "learning_rate": 9.799707604769159e-06,
      "loss": 2.0464,
      "step": 211606
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.4292265176773071,
      "learning_rate": 9.798670657400586e-06,
      "loss": 1.7879,
      "step": 211607
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.4706625938415527,
      "learning_rate": 9.797633763985923e-06,
      "loss": 2.1536,
      "step": 211608
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.718380331993103,
      "learning_rate": 9.796596924525402e-06,
      "loss": 2.0201,
      "step": 211609
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.4702836275100708,
      "learning_rate": 9.795560139019154e-06,
      "loss": 1.9916,
      "step": 211610
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.5237988233566284,
      "learning_rate": 9.794523407467348e-06,
      "loss": 2.0456,
      "step": 211611
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.541333794593811,
      "learning_rate": 9.793486729870281e-06,
      "loss": 1.9323,
      "step": 211612
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.5981031656265259,
      "learning_rate": 9.792450106228023e-06,
      "loss": 1.7579,
      "step": 211613
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.5246896743774414,
      "learning_rate": 9.791413536540838e-06,
      "loss": 2.0509,
      "step": 211614
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.548403263092041,
      "learning_rate": 9.79037702080896e-06,
      "loss": 1.944,
      "step": 211615
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.5241252183914185,
      "learning_rate": 9.789340559032454e-06,
      "loss": 1.9957,
      "step": 211616
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.4373345375061035,
      "learning_rate": 9.788304151211656e-06,
      "loss": 2.0336,
      "step": 211617
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.586266040802002,
      "learning_rate": 9.787267797346665e-06,
      "loss": 1.7656,
      "step": 211618
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.3447976112365723,
      "learning_rate": 9.78623149743768e-06,
      "loss": 1.9145,
      "step": 211619
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.3655301332473755,
      "learning_rate": 9.785195251484935e-06,
      "loss": 1.9872,
      "step": 211620
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.6775630712509155,
      "learning_rate": 9.784159059488594e-06,
      "loss": 1.8995,
      "step": 211621
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.4861187934875488,
      "learning_rate": 9.783122921448827e-06,
      "loss": 1.8582,
      "step": 211622
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.4133427143096924,
      "learning_rate": 9.782086837365898e-06,
      "loss": 1.966,
      "step": 211623
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.5656408071517944,
      "learning_rate": 9.781050807239943e-06,
      "loss": 2.1297,
      "step": 211624
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.5264619588851929,
      "learning_rate": 9.78001483107116e-06,
      "loss": 1.8472,
      "step": 211625
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.4662119150161743,
      "learning_rate": 9.77897890885978e-06,
      "loss": 1.8642,
      "step": 211626
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.353973627090454,
      "learning_rate": 9.77794304060594e-06,
      "loss": 2.0191,
      "step": 211627
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.5114396810531616,
      "learning_rate": 9.776907226309838e-06,
      "loss": 2.0619,
      "step": 211628
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.411681890487671,
      "learning_rate": 9.77587146597174e-06,
      "loss": 2.0829,
      "step": 211629
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.5287151336669922,
      "learning_rate": 9.774835759591714e-06,
      "loss": 2.04,
      "step": 211630
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.412896752357483,
      "learning_rate": 9.773800107170093e-06,
      "loss": 1.911,
      "step": 211631
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.5187982320785522,
      "learning_rate": 9.772764508706975e-06,
      "loss": 2.033,
      "step": 211632
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.682705283164978,
      "learning_rate": 9.771728964202596e-06,
      "loss": 2.201,
      "step": 211633
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.4257097244262695,
      "learning_rate": 9.770693473657121e-06,
      "loss": 2.0219,
      "step": 211634
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.412174940109253,
      "learning_rate": 9.769658037070748e-06,
      "loss": 1.828,
      "step": 211635
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.417958378791809,
      "learning_rate": 9.768622654443647e-06,
      "loss": 1.9724,
      "step": 211636
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.3267407417297363,
      "learning_rate": 9.767587325776082e-06,
      "loss": 1.9093,
      "step": 211637
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.60623300075531,
      "learning_rate": 9.766552051068154e-06,
      "loss": 2.2893,
      "step": 211638
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.414642095565796,
      "learning_rate": 9.765516830320164e-06,
      "loss": 2.0596,
      "step": 211639
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.5126277208328247,
      "learning_rate": 9.764481663532208e-06,
      "loss": 1.8656,
      "step": 211640
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.5007067918777466,
      "learning_rate": 9.763446550704523e-06,
      "loss": 1.9858,
      "step": 211641
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.537406086921692,
      "learning_rate": 9.762411491837274e-06,
      "loss": 1.947,
      "step": 211642
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.555400013923645,
      "learning_rate": 9.761376486930695e-06,
      "loss": 2.0789,
      "step": 211643
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.4512486457824707,
      "learning_rate": 9.760341535984916e-06,
      "loss": 1.9008,
      "step": 211644
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.3790394067764282,
      "learning_rate": 9.759306639000242e-06,
      "loss": 2.162,
      "step": 211645
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.5354098081588745,
      "learning_rate": 9.758271795976702e-06,
      "loss": 1.9117,
      "step": 211646
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.5018956661224365,
      "learning_rate": 9.757237006914631e-06,
      "loss": 1.9973,
      "step": 211647
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.4486939907073975,
      "learning_rate": 9.756202271814162e-06,
      "loss": 1.9127,
      "step": 211648
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.4986546039581299,
      "learning_rate": 9.755167590675494e-06,
      "loss": 1.8933,
      "step": 211649
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.4892692565917969,
      "learning_rate": 9.754132963498795e-06,
      "loss": 1.8665,
      "step": 211650
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.404531478881836,
      "learning_rate": 9.753098390284331e-06,
      "loss": 1.9978,
      "step": 211651
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.372923731803894,
      "learning_rate": 9.752063871032201e-06,
      "loss": 1.8669,
      "step": 211652
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.7252391576766968,
      "learning_rate": 9.751029405742673e-06,
      "loss": 2.1301,
      "step": 211653
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.451993465423584,
      "learning_rate": 9.749994994415843e-06,
      "loss": 1.9789,
      "step": 211654
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.6021678447723389,
      "learning_rate": 9.74896063705205e-06,
      "loss": 1.9294,
      "step": 211655
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.4951488971710205,
      "learning_rate": 9.747926333651324e-06,
      "loss": 1.9059,
      "step": 211656
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.3675185441970825,
      "learning_rate": 9.746892084213997e-06,
      "loss": 1.8998,
      "step": 211657
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.379172921180725,
      "learning_rate": 9.745857888740206e-06,
      "loss": 1.9954,
      "step": 211658
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.4464031457901,
      "learning_rate": 9.744823747230147e-06,
      "loss": 2.0382,
      "step": 211659
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.5072702169418335,
      "learning_rate": 9.743789659683953e-06,
      "loss": 1.9034,
      "step": 211660
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.4562020301818848,
      "learning_rate": 9.742755626101895e-06,
      "loss": 2.1085,
      "step": 211661
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.3315353393554688,
      "learning_rate": 9.741721646484102e-06,
      "loss": 1.8604,
      "step": 211662
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.4411898851394653,
      "learning_rate": 9.740687720830843e-06,
      "loss": 2.0437,
      "step": 211663
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.5049761533737183,
      "learning_rate": 9.739653849142282e-06,
      "loss": 1.8884,
      "step": 211664
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.4221575260162354,
      "learning_rate": 9.738620031418586e-06,
      "loss": 2.0539,
      "step": 211665
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.5103039741516113,
      "learning_rate": 9.737586267659924e-06,
      "loss": 1.9477,
      "step": 211666
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.4291235208511353,
      "learning_rate": 9.736552557866561e-06,
      "loss": 2.143,
      "step": 211667
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.5340989828109741,
      "learning_rate": 9.735518902038631e-06,
      "loss": 1.8811,
      "step": 211668
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.5490492582321167,
      "learning_rate": 9.734485300176364e-06,
      "loss": 1.7246,
      "step": 211669
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.3568283319473267,
      "learning_rate": 9.73345175227993e-06,
      "loss": 1.9112,
      "step": 211670
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.437334418296814,
      "learning_rate": 9.732418258349561e-06,
      "loss": 1.6788,
      "step": 211671
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.4552539587020874,
      "learning_rate": 9.731384818385357e-06,
      "loss": 1.9197,
      "step": 211672
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.3471094369888306,
      "learning_rate": 9.730351432387584e-06,
      "loss": 1.9583,
      "step": 211673
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.4602644443511963,
      "learning_rate": 9.729318100356409e-06,
      "loss": 2.095,
      "step": 211674
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.4083479642868042,
      "learning_rate": 9.728284822292064e-06,
      "loss": 2.0891,
      "step": 211675
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.413685917854309,
      "learning_rate": 9.727251598194685e-06,
      "loss": 1.6812,
      "step": 211676
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.570587158203125,
      "learning_rate": 9.726218428064537e-06,
      "loss": 2.2056,
      "step": 211677
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.5193819999694824,
      "learning_rate": 9.725185311901684e-06,
      "loss": 2.1289,
      "step": 211678
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.3492814302444458,
      "learning_rate": 9.724152249706463e-06,
      "loss": 1.8681,
      "step": 211679
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.656055212020874,
      "learning_rate": 9.723119241478938e-06,
      "loss": 2.1014,
      "step": 211680
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.480046272277832,
      "learning_rate": 9.722086287219445e-06,
      "loss": 1.7806,
      "step": 211681
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.3837497234344482,
      "learning_rate": 9.721053386928046e-06,
      "loss": 1.6418,
      "step": 211682
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.506884217262268,
      "learning_rate": 9.72002054060501e-06,
      "loss": 2.0344,
      "step": 211683
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.4799175262451172,
      "learning_rate": 9.718987748250473e-06,
      "loss": 1.9199,
      "step": 211684
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.4928818941116333,
      "learning_rate": 9.717955009864664e-06,
      "loss": 2.039,
      "step": 211685
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.5933140516281128,
      "learning_rate": 9.716922325447751e-06,
      "loss": 1.9136,
      "step": 211686
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.4385427236557007,
      "learning_rate": 9.715889694999967e-06,
      "loss": 1.9865,
      "step": 211687
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.7697348594665527,
      "learning_rate": 9.71485711852148e-06,
      "loss": 1.7665,
      "step": 211688
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.41786789894104,
      "learning_rate": 9.713824596012488e-06,
      "loss": 1.976,
      "step": 211689
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.3907182216644287,
      "learning_rate": 9.712792127473123e-06,
      "loss": 1.8574,
      "step": 211690
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.3976945877075195,
      "learning_rate": 9.711759712903688e-06,
      "loss": 1.8714,
      "step": 211691
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.3978132009506226,
      "learning_rate": 9.710727352304282e-06,
      "loss": 1.918,
      "step": 211692
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.4013022184371948,
      "learning_rate": 9.709695045675136e-06,
      "loss": 1.9287,
      "step": 211693
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.4608699083328247,
      "learning_rate": 9.708662793016486e-06,
      "loss": 1.9794,
      "step": 211694
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.312135100364685,
      "learning_rate": 9.70763059432843e-06,
      "loss": 1.9667,
      "step": 211695
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.655821681022644,
      "learning_rate": 9.706598449611203e-06,
      "loss": 2.1865,
      "step": 211696
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.6569536924362183,
      "learning_rate": 9.705566358865003e-06,
      "loss": 1.9682,
      "step": 211697
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.4879512786865234,
      "learning_rate": 9.704534322089997e-06,
      "loss": 2.0087,
      "step": 211698
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.5612400770187378,
      "learning_rate": 9.70350233928645e-06,
      "loss": 1.8947,
      "step": 211699
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.434442400932312,
      "learning_rate": 9.702470410454465e-06,
      "loss": 2.1149,
      "step": 211700
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.43670654296875,
      "learning_rate": 9.701438535594309e-06,
      "loss": 1.8947,
      "step": 211701
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.4108858108520508,
      "learning_rate": 9.700406714706076e-06,
      "loss": 2.0353,
      "step": 211702
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.5392776727676392,
      "learning_rate": 9.69937494779004e-06,
      "loss": 2.1523,
      "step": 211703
    },
    {
      "epoch": 2.76,
      "grad_norm": 2.567784547805786,
      "learning_rate": 9.698343234846362e-06,
      "loss": 2.2316,
      "step": 211704
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.5906345844268799,
      "learning_rate": 9.697311575875278e-06,
      "loss": 1.8346,
      "step": 211705
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.628182053565979,
      "learning_rate": 9.69627997087692e-06,
      "loss": 1.9668,
      "step": 211706
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.6185427904129028,
      "learning_rate": 9.695248419851487e-06,
      "loss": 1.9885,
      "step": 211707
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.7489166259765625,
      "learning_rate": 9.694216922799214e-06,
      "loss": 2.0051,
      "step": 211708
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.4053410291671753,
      "learning_rate": 9.693185479720267e-06,
      "loss": 1.9585,
      "step": 211709
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.6733558177947998,
      "learning_rate": 9.692154090614811e-06,
      "loss": 2.0086,
      "step": 211710
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.5612874031066895,
      "learning_rate": 9.691122755483083e-06,
      "loss": 1.9127,
      "step": 211711
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.5363061428070068,
      "learning_rate": 9.690091474325278e-06,
      "loss": 2.0338,
      "step": 211712
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.4983631372451782,
      "learning_rate": 9.689060247141533e-06,
      "loss": 1.9653,
      "step": 211713
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.5122512578964233,
      "learning_rate": 9.68802907393208e-06,
      "loss": 2.0866,
      "step": 211714
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.4689481258392334,
      "learning_rate": 9.686997954697085e-06,
      "loss": 1.9972,
      "step": 211715
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.462752342224121,
      "learning_rate": 9.68596688943678e-06,
      "loss": 1.9398,
      "step": 211716
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.3668700456619263,
      "learning_rate": 9.684935878151334e-06,
      "loss": 1.7398,
      "step": 211717
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.4650212526321411,
      "learning_rate": 9.683904920840945e-06,
      "loss": 1.886,
      "step": 211718
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.4159290790557861,
      "learning_rate": 9.682874017505749e-06,
      "loss": 2.0833,
      "step": 211719
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.4298912286758423,
      "learning_rate": 9.681843168146042e-06,
      "loss": 2.0481,
      "step": 211720
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.47847580909729,
      "learning_rate": 9.680812372761926e-06,
      "loss": 2.0769,
      "step": 211721
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.456008791923523,
      "learning_rate": 9.679781631353634e-06,
      "loss": 1.7668,
      "step": 211722
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.4360713958740234,
      "learning_rate": 9.678750943921398e-06,
      "loss": 1.9035,
      "step": 211723
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.5222474336624146,
      "learning_rate": 9.677720310465287e-06,
      "loss": 1.7209,
      "step": 211724
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.5086251497268677,
      "learning_rate": 9.676689730985632e-06,
      "loss": 2.2678,
      "step": 211725
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.5050565004348755,
      "learning_rate": 9.675659205482533e-06,
      "loss": 1.9274,
      "step": 211726
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.352979302406311,
      "learning_rate": 9.67462873395619e-06,
      "loss": 2.0558,
      "step": 211727
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.416863203048706,
      "learning_rate": 9.67359831640684e-06,
      "loss": 1.9791,
      "step": 211728
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.5506964921951294,
      "learning_rate": 9.672567952834609e-06,
      "loss": 1.8841,
      "step": 211729
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.5052404403686523,
      "learning_rate": 9.671537643239768e-06,
      "loss": 2.102,
      "step": 211730
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.4321186542510986,
      "learning_rate": 9.670507387622484e-06,
      "loss": 1.8011,
      "step": 211731
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.479343295097351,
      "learning_rate": 9.66947718598292e-06,
      "loss": 2.0556,
      "step": 211732
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.4794484376907349,
      "learning_rate": 9.668447038321247e-06,
      "loss": 1.9262,
      "step": 211733
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.446592926979065,
      "learning_rate": 9.66741694463773e-06,
      "loss": 2.2267,
      "step": 211734
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.3636753559112549,
      "learning_rate": 9.666386904932467e-06,
      "loss": 1.8176,
      "step": 211735
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.403368592262268,
      "learning_rate": 9.665356919205759e-06,
      "loss": 1.5615,
      "step": 211736
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.4269787073135376,
      "learning_rate": 9.664326987457738e-06,
      "loss": 2.0602,
      "step": 211737
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.3845244646072388,
      "learning_rate": 9.663297109688606e-06,
      "loss": 1.8835,
      "step": 211738
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.359479546546936,
      "learning_rate": 9.662267285898496e-06,
      "loss": 1.8413,
      "step": 211739
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.473272442817688,
      "learning_rate": 9.661237516087706e-06,
      "loss": 2.1001,
      "step": 211740
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.4413844347000122,
      "learning_rate": 9.660207800256303e-06,
      "loss": 1.9144,
      "step": 211741
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.532317876815796,
      "learning_rate": 9.659178138404622e-06,
      "loss": 1.9405,
      "step": 211742
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.5214873552322388,
      "learning_rate": 9.65814853053276e-06,
      "loss": 2.0306,
      "step": 211743
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.4502204656600952,
      "learning_rate": 9.657118976640922e-06,
      "loss": 2.0661,
      "step": 211744
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.41716468334198,
      "learning_rate": 9.656089476729301e-06,
      "loss": 1.7567,
      "step": 211745
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.4802048206329346,
      "learning_rate": 9.6550600307981e-06,
      "loss": 2.0917,
      "step": 211746
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.4380204677581787,
      "learning_rate": 9.654030638847487e-06,
      "loss": 2.0196,
      "step": 211747
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.3437618017196655,
      "learning_rate": 9.653001300877694e-06,
      "loss": 1.8109,
      "step": 211748
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.5365277528762817,
      "learning_rate": 9.651972016888887e-06,
      "loss": 2.0345,
      "step": 211749
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.3112279176712036,
      "learning_rate": 9.650942786881266e-06,
      "loss": 1.8652,
      "step": 211750
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.4111781120300293,
      "learning_rate": 9.649913610854965e-06,
      "loss": 1.6587,
      "step": 211751
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.419748306274414,
      "learning_rate": 9.648884488810282e-06,
      "loss": 2.0218,
      "step": 211752
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.5035892724990845,
      "learning_rate": 9.647855420747286e-06,
      "loss": 2.117,
      "step": 211753
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.4507629871368408,
      "learning_rate": 9.646826406666308e-06,
      "loss": 2.0111,
      "step": 211754
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.5977115631103516,
      "learning_rate": 9.645797446567449e-06,
      "loss": 2.0502,
      "step": 211755
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.493068814277649,
      "learning_rate": 9.644768540450909e-06,
      "loss": 2.1401,
      "step": 211756
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.5242202281951904,
      "learning_rate": 9.643739688316854e-06,
      "loss": 2.0363,
      "step": 211757
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.4620089530944824,
      "learning_rate": 9.64271089016555e-06,
      "loss": 2.0369,
      "step": 211758
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.3939149379730225,
      "learning_rate": 9.641682145997132e-06,
      "loss": 2.0361,
      "step": 211759
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.5199559926986694,
      "learning_rate": 9.640653455811831e-06,
      "loss": 1.919,
      "step": 211760
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.4835700988769531,
      "learning_rate": 9.639624819609782e-06,
      "loss": 1.8378,
      "step": 211761
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.484719157218933,
      "learning_rate": 9.63859623739125e-06,
      "loss": 2.067,
      "step": 211762
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.4783639907836914,
      "learning_rate": 9.637567709156335e-06,
      "loss": 1.8906,
      "step": 211763
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.565697431564331,
      "learning_rate": 9.636539234905304e-06,
      "loss": 1.9443,
      "step": 211764
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.4969232082366943,
      "learning_rate": 9.635510814638293e-06,
      "loss": 2.1302,
      "step": 211765
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.4543572664260864,
      "learning_rate": 9.63448244835553e-06,
      "loss": 1.7567,
      "step": 211766
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.5621427297592163,
      "learning_rate": 9.633454136057218e-06,
      "loss": 1.9657,
      "step": 211767
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.59317946434021,
      "learning_rate": 9.632425877743522e-06,
      "loss": 2.036,
      "step": 211768
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.5243643522262573,
      "learning_rate": 9.631397673414643e-06,
      "loss": 2.1237,
      "step": 211769
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.4986628293991089,
      "learning_rate": 9.630369523070747e-06,
      "loss": 2.121,
      "step": 211770
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.4383962154388428,
      "learning_rate": 9.629341426712034e-06,
      "loss": 2.1272,
      "step": 211771
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.5178368091583252,
      "learning_rate": 9.62831338433877e-06,
      "loss": 1.946,
      "step": 211772
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.3874154090881348,
      "learning_rate": 9.627285395951056e-06,
      "loss": 2.4247,
      "step": 211773
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.4460992813110352,
      "learning_rate": 9.626257461549092e-06,
      "loss": 1.8991,
      "step": 211774
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.534775733947754,
      "learning_rate": 9.625229581133076e-06,
      "loss": 1.897,
      "step": 211775
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.3529274463653564,
      "learning_rate": 9.624201754703276e-06,
      "loss": 1.8636,
      "step": 211776
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.5126781463623047,
      "learning_rate": 9.623173982259723e-06,
      "loss": 1.9301,
      "step": 211777
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.4295839071273804,
      "learning_rate": 9.622146263802789e-06,
      "loss": 1.7939,
      "step": 211778
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.4834222793579102,
      "learning_rate": 9.621118599332534e-06,
      "loss": 1.8866,
      "step": 211779
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.5637763738632202,
      "learning_rate": 9.620090988849228e-06,
      "loss": 1.98,
      "step": 211780
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.4867818355560303,
      "learning_rate": 9.619063432353002e-06,
      "loss": 2.0414,
      "step": 211781
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.3027820587158203,
      "learning_rate": 9.618035929844091e-06,
      "loss": 1.9476,
      "step": 211782
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.5025122165679932,
      "learning_rate": 9.61700848132263e-06,
      "loss": 2.0768,
      "step": 211783
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.4498323202133179,
      "learning_rate": 9.615981086788882e-06,
      "loss": 1.9385,
      "step": 211784
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.4856889247894287,
      "learning_rate": 9.614953746243015e-06,
      "loss": 1.9393,
      "step": 211785
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.325229525566101,
      "learning_rate": 9.613926459685195e-06,
      "loss": 1.8285,
      "step": 211786
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.400633692741394,
      "learning_rate": 9.61289922711559e-06,
      "loss": 1.9875,
      "step": 211787
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.4590827226638794,
      "learning_rate": 9.611872048534497e-06,
      "loss": 1.8447,
      "step": 211788
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.4027196168899536,
      "learning_rate": 9.610844923941951e-06,
      "loss": 1.989,
      "step": 211789
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.4655901193618774,
      "learning_rate": 9.60981785333832e-06,
      "loss": 1.8496,
      "step": 211790
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.7004203796386719,
      "learning_rate": 9.608790836723634e-06,
      "loss": 2.0849,
      "step": 211791
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.4158775806427002,
      "learning_rate": 9.607763874098196e-06,
      "loss": 2.0264,
      "step": 211792
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.3665008544921875,
      "learning_rate": 9.606736965462169e-06,
      "loss": 1.9825,
      "step": 211793
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.4475834369659424,
      "learning_rate": 9.605710110815724e-06,
      "loss": 2.2694,
      "step": 211794
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.480238437652588,
      "learning_rate": 9.604683310159023e-06,
      "loss": 2.0162,
      "step": 211795
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.4130290746688843,
      "learning_rate": 9.603656563492335e-06,
      "loss": 2.0551,
      "step": 211796
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.5422130823135376,
      "learning_rate": 9.602629870815758e-06,
      "loss": 1.8326,
      "step": 211797
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.3266522884368896,
      "learning_rate": 9.601603232129596e-06,
      "loss": 1.8359,
      "step": 211798
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.3231889009475708,
      "learning_rate": 9.600576647433945e-06,
      "loss": 2.0353,
      "step": 211799
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.4754705429077148,
      "learning_rate": 9.599550116729004e-06,
      "loss": 2.0553,
      "step": 211800
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.541045069694519,
      "learning_rate": 9.598523640015043e-06,
      "loss": 2.0717,
      "step": 211801
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.6141902208328247,
      "learning_rate": 9.597497217292194e-06,
      "loss": 2.01,
      "step": 211802
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.5010699033737183,
      "learning_rate": 9.59647084856059e-06,
      "loss": 1.8039,
      "step": 211803
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.5731289386749268,
      "learning_rate": 9.59544453382053e-06,
      "loss": 1.9926,
      "step": 211804
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.3791619539260864,
      "learning_rate": 9.59441827307218e-06,
      "loss": 2.1024,
      "step": 211805
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.417324423789978,
      "learning_rate": 9.593392066315675e-06,
      "loss": 2.0922,
      "step": 211806
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.5451619625091553,
      "learning_rate": 9.592365913551248e-06,
      "loss": 1.9408,
      "step": 211807
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.426445484161377,
      "learning_rate": 9.591339814779097e-06,
      "loss": 2.1067,
      "step": 211808
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.3769831657409668,
      "learning_rate": 9.590313769999359e-06,
      "loss": 2.0219,
      "step": 211809
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.5215073823928833,
      "learning_rate": 9.589287779212296e-06,
      "loss": 1.8352,
      "step": 211810
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.5939754247665405,
      "learning_rate": 9.588261842418077e-06,
      "loss": 1.9945,
      "step": 211811
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.4630483388900757,
      "learning_rate": 9.587235959616834e-06,
      "loss": 1.9324,
      "step": 211812
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.3579199314117432,
      "learning_rate": 9.586210130808835e-06,
      "loss": 1.8221,
      "step": 211813
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.4532219171524048,
      "learning_rate": 9.585184355994213e-06,
      "loss": 2.0579,
      "step": 211814
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.5685704946517944,
      "learning_rate": 9.584158635173233e-06,
      "loss": 1.8082,
      "step": 211815
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.4555487632751465,
      "learning_rate": 9.58313296834603e-06,
      "loss": 1.9521,
      "step": 211816
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.4242650270462036,
      "learning_rate": 9.582107355512803e-06,
      "loss": 1.9322,
      "step": 211817
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.4669859409332275,
      "learning_rate": 9.581081796673718e-06,
      "loss": 1.8558,
      "step": 211818
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.4862761497497559,
      "learning_rate": 9.580056291829042e-06,
      "loss": 1.9942,
      "step": 211819
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.407395601272583,
      "learning_rate": 9.579030840978842e-06,
      "loss": 1.8566,
      "step": 211820
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.5497512817382812,
      "learning_rate": 9.57800544412345e-06,
      "loss": 1.9758,
      "step": 211821
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.402023434638977,
      "learning_rate": 9.576980101262965e-06,
      "loss": 1.639,
      "step": 211822
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.601235270500183,
      "learning_rate": 9.575954812397623e-06,
      "loss": 2.2153,
      "step": 211823
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.5108892917633057,
      "learning_rate": 9.574929577527556e-06,
      "loss": 1.9565,
      "step": 211824
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.4100264310836792,
      "learning_rate": 9.573904396652999e-06,
      "loss": 2.0276,
      "step": 211825
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.3398422002792358,
      "learning_rate": 9.572879269774148e-06,
      "loss": 2.2392,
      "step": 211826
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.4632494449615479,
      "learning_rate": 9.57185419689117e-06,
      "loss": 2.018,
      "step": 211827
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.3644391298294067,
      "learning_rate": 9.570829178004302e-06,
      "loss": 1.8759,
      "step": 211828
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.6212836503982544,
      "learning_rate": 9.569804213113675e-06,
      "loss": 1.9584,
      "step": 211829
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.5877419710159302,
      "learning_rate": 9.568779302219453e-06,
      "loss": 2.0717,
      "step": 211830
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.4782432317733765,
      "learning_rate": 9.56775444532194e-06,
      "loss": 2.1028,
      "step": 211831
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.6582162380218506,
      "learning_rate": 9.566729642421233e-06,
      "loss": 2.0635,
      "step": 211832
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.4982928037643433,
      "learning_rate": 9.565704893517568e-06,
      "loss": 1.8572,
      "step": 211833
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.572106122970581,
      "learning_rate": 9.56468019861114e-06,
      "loss": 1.8686,
      "step": 211834
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.431999683380127,
      "learning_rate": 9.563655557702122e-06,
      "loss": 1.8243,
      "step": 211835
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.3416811227798462,
      "learning_rate": 9.562630970790642e-06,
      "loss": 1.968,
      "step": 211836
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.4871729612350464,
      "learning_rate": 9.561606437877e-06,
      "loss": 2.0978,
      "step": 211837
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.4524325132369995,
      "learning_rate": 9.560581958961334e-06,
      "loss": 2.0589,
      "step": 211838
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.4671173095703125,
      "learning_rate": 9.559557534043838e-06,
      "loss": 1.8069,
      "step": 211839
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.3228719234466553,
      "learning_rate": 9.558533163124716e-06,
      "loss": 1.7726,
      "step": 211840
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.6369084119796753,
      "learning_rate": 9.557508846204132e-06,
      "loss": 1.8646,
      "step": 211841
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.481380581855774,
      "learning_rate": 9.556484583282253e-06,
      "loss": 2.1295,
      "step": 211842
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.518804669380188,
      "learning_rate": 9.555460374359348e-06,
      "loss": 2.0263,
      "step": 211843
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.66189706325531,
      "learning_rate": 9.554436219435547e-06,
      "loss": 2.0088,
      "step": 211844
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.4870665073394775,
      "learning_rate": 9.55341211851105e-06,
      "loss": 1.7405,
      "step": 211845
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.705122709274292,
      "learning_rate": 9.552388071586092e-06,
      "loss": 1.805,
      "step": 211846
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.455334186553955,
      "learning_rate": 9.551364078660839e-06,
      "loss": 1.9306,
      "step": 211847
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.4355915784835815,
      "learning_rate": 9.550340139735391e-06,
      "loss": 2.0605,
      "step": 211848
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.4727994203567505,
      "learning_rate": 9.54931625481008e-06,
      "loss": 1.8507,
      "step": 211849
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.554909110069275,
      "learning_rate": 9.548292423885007e-06,
      "loss": 2.1067,
      "step": 211850
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.4149055480957031,
      "learning_rate": 9.547268646960404e-06,
      "loss": 2.1376,
      "step": 211851
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.5867893695831299,
      "learning_rate": 9.546244924036472e-06,
      "loss": 1.857,
      "step": 211852
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.4889628887176514,
      "learning_rate": 9.545221255113344e-06,
      "loss": 1.8996,
      "step": 211853
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.659824013710022,
      "learning_rate": 9.54419764019122e-06,
      "loss": 1.7906,
      "step": 211854
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.4373961687088013,
      "learning_rate": 9.543174079270365e-06,
      "loss": 1.8261,
      "step": 211855
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.6929895877838135,
      "learning_rate": 9.542150572350882e-06,
      "loss": 1.8441,
      "step": 211856
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.436341404914856,
      "learning_rate": 9.541127119433e-06,
      "loss": 1.7972,
      "step": 211857
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.5421112775802612,
      "learning_rate": 9.54010372051689e-06,
      "loss": 1.7779,
      "step": 211858
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.501611351966858,
      "learning_rate": 9.539080375602815e-06,
      "loss": 2.1112,
      "step": 211859
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.6177561283111572,
      "learning_rate": 9.538057084690842e-06,
      "loss": 2.1019,
      "step": 211860
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.522391438484192,
      "learning_rate": 9.537033847781273e-06,
      "loss": 1.8264,
      "step": 211861
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.582336664199829,
      "learning_rate": 9.536010664874238e-06,
      "loss": 2.0256,
      "step": 211862
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.562466025352478,
      "learning_rate": 9.53498753596994e-06,
      "loss": 2.1307,
      "step": 211863
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.552438497543335,
      "learning_rate": 9.533964461068545e-06,
      "loss": 2.0826,
      "step": 211864
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.3154056072235107,
      "learning_rate": 9.53294144017035e-06,
      "loss": 2.0254,
      "step": 211865
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.6275136470794678,
      "learning_rate": 9.531918473275357e-06,
      "loss": 1.9847,
      "step": 211866
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.6512126922607422,
      "learning_rate": 9.530895560383933e-06,
      "loss": 1.9379,
      "step": 211867
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.5269205570220947,
      "learning_rate": 9.529872701496177e-06,
      "loss": 2.1126,
      "step": 211868
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.5220826864242554,
      "learning_rate": 9.52884989661229e-06,
      "loss": 2.1545,
      "step": 211869
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.5872281789779663,
      "learning_rate": 9.52782714573247e-06,
      "loss": 2.2155,
      "step": 211870
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.5135430097579956,
      "learning_rate": 9.526804448856983e-06,
      "loss": 1.9604,
      "step": 211871
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.5123616456985474,
      "learning_rate": 9.525781805985865e-06,
      "loss": 2.0042,
      "step": 211872
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.521262288093567,
      "learning_rate": 9.524759217119415e-06,
      "loss": 2.0101,
      "step": 211873
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.570357322692871,
      "learning_rate": 9.523736682257765e-06,
      "loss": 1.7666,
      "step": 211874
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.30524742603302,
      "learning_rate": 9.522714201401183e-06,
      "loss": 2.2098,
      "step": 211875
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.676399827003479,
      "learning_rate": 9.521691774549767e-06,
      "loss": 2.1069,
      "step": 211876
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.5169668197631836,
      "learning_rate": 9.520669401703785e-06,
      "loss": 1.9372,
      "step": 211877
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.6312512159347534,
      "learning_rate": 9.519647082863402e-06,
      "loss": 2.2208,
      "step": 211878
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.579329252243042,
      "learning_rate": 9.518624818028785e-06,
      "loss": 1.916,
      "step": 211879
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.4221616983413696,
      "learning_rate": 9.517602607200104e-06,
      "loss": 2.1439,
      "step": 211880
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.4585987329483032,
      "learning_rate": 9.516580450377654e-06,
      "loss": 2.0111,
      "step": 211881
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.5360203981399536,
      "learning_rate": 9.51555834756147e-06,
      "loss": 2.044,
      "step": 211882
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.487778902053833,
      "learning_rate": 9.514536298751885e-06,
      "loss": 1.8806,
      "step": 211883
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.511576771736145,
      "learning_rate": 9.513514303949033e-06,
      "loss": 1.9816,
      "step": 211884
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.5304831266403198,
      "learning_rate": 9.512492363153112e-06,
      "loss": 2.0135,
      "step": 211885
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.3828141689300537,
      "learning_rate": 9.511470476364258e-06,
      "loss": 1.9714,
      "step": 211886
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.3348066806793213,
      "learning_rate": 9.510448643582736e-06,
      "loss": 1.7899,
      "step": 211887
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.490397334098816,
      "learning_rate": 9.509426864808678e-06,
      "loss": 1.9433,
      "step": 211888
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.4015196561813354,
      "learning_rate": 9.50840514004232e-06,
      "loss": 2.012,
      "step": 211889
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.3823496103286743,
      "learning_rate": 9.507383469283858e-06,
      "loss": 1.8268,
      "step": 211890
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.4765303134918213,
      "learning_rate": 9.506361852533396e-06,
      "loss": 2.1449,
      "step": 211891
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.5217410326004028,
      "learning_rate": 9.50534028979123e-06,
      "loss": 2.0246,
      "step": 211892
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.4758601188659668,
      "learning_rate": 9.504318781057496e-06,
      "loss": 1.9934,
      "step": 211893
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.4418153762817383,
      "learning_rate": 9.50329732633236e-06,
      "loss": 1.7513,
      "step": 211894
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.5432707071304321,
      "learning_rate": 9.502275925616087e-06,
      "loss": 2.1032,
      "step": 211895
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.5717171430587769,
      "learning_rate": 9.501254578908846e-06,
      "loss": 1.9721,
      "step": 211896
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.3712403774261475,
      "learning_rate": 9.500233286210735e-06,
      "loss": 2.0037,
      "step": 211897
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.4733213186264038,
      "learning_rate": 9.499212047522053e-06,
      "loss": 2.0453,
      "step": 211898
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.381022572517395,
      "learning_rate": 9.498190862842937e-06,
      "loss": 1.7951,
      "step": 211899
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.473237156867981,
      "learning_rate": 9.497169732173616e-06,
      "loss": 1.9853,
      "step": 211900
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.6248152256011963,
      "learning_rate": 9.496148655514258e-06,
      "loss": 2.0831,
      "step": 211901
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.4531023502349854,
      "learning_rate": 9.49512763286503e-06,
      "loss": 2.1048,
      "step": 211902
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.471107840538025,
      "learning_rate": 9.494106664226131e-06,
      "loss": 2.0794,
      "step": 211903
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.3961787223815918,
      "learning_rate": 9.493085749597795e-06,
      "loss": 2.1758,
      "step": 211904
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.595720887184143,
      "learning_rate": 9.492064888980123e-06,
      "loss": 2.1137,
      "step": 211905
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.3940099477767944,
      "learning_rate": 9.491044082373412e-06,
      "loss": 1.9072,
      "step": 211906
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.4138460159301758,
      "learning_rate": 9.490023329777796e-06,
      "loss": 2.1484,
      "step": 211907
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.4897078275680542,
      "learning_rate": 9.489002631193476e-06,
      "loss": 2.0422,
      "step": 211908
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.3415666818618774,
      "learning_rate": 9.487981986620585e-06,
      "loss": 1.9737,
      "step": 211909
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.3873074054718018,
      "learning_rate": 9.48696139605939e-06,
      "loss": 1.7863,
      "step": 211910
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.4632903337478638,
      "learning_rate": 9.485940859510055e-06,
      "loss": 2.0834,
      "step": 211911
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.4388976097106934,
      "learning_rate": 9.48492037697275e-06,
      "loss": 2.1481,
      "step": 211912
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.3840281963348389,
      "learning_rate": 9.483899948447738e-06,
      "loss": 1.9,
      "step": 211913
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.4880971908569336,
      "learning_rate": 9.48287957393512e-06,
      "loss": 1.8064,
      "step": 211914
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.409035563468933,
      "learning_rate": 9.481859253435064e-06,
      "loss": 1.9945,
      "step": 211915
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.4993586540222168,
      "learning_rate": 9.480838986947902e-06,
      "loss": 1.9834,
      "step": 211916
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.5185567140579224,
      "learning_rate": 9.479818774473668e-06,
      "loss": 2.0647,
      "step": 211917
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.4981380701065063,
      "learning_rate": 9.478798616012662e-06,
      "loss": 2.0268,
      "step": 211918
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.440253496170044,
      "learning_rate": 9.477778511565016e-06,
      "loss": 1.7365,
      "step": 211919
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.3878751993179321,
      "learning_rate": 9.476758461130963e-06,
      "loss": 2.318,
      "step": 211920
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.4212968349456787,
      "learning_rate": 9.475738464710603e-06,
      "loss": 1.91,
      "step": 211921
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.4356943368911743,
      "learning_rate": 9.474718522304236e-06,
      "loss": 1.8969,
      "step": 211922
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.5321438312530518,
      "learning_rate": 9.473698633911963e-06,
      "loss": 1.9615,
      "step": 211923
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.5604256391525269,
      "learning_rate": 9.47267879953405e-06,
      "loss": 1.8551,
      "step": 211924
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.4136213064193726,
      "learning_rate": 9.47165901917063e-06,
      "loss": 1.9334,
      "step": 211925
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.542693853378296,
      "learning_rate": 9.470639292821935e-06,
      "loss": 1.9855,
      "step": 211926
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.4571789503097534,
      "learning_rate": 9.4696196204881e-06,
      "loss": 1.9932,
      "step": 211927
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.633571743965149,
      "learning_rate": 9.468600002169391e-06,
      "loss": 2.0879,
      "step": 211928
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.508245587348938,
      "learning_rate": 9.467580437865907e-06,
      "loss": 2.0221,
      "step": 211929
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.6213120222091675,
      "learning_rate": 9.466560927577916e-06,
      "loss": 2.0196,
      "step": 211930
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.5410236120224,
      "learning_rate": 9.465541471305549e-06,
      "loss": 2.0929,
      "step": 211931
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.5341527462005615,
      "learning_rate": 9.464522069049075e-06,
      "loss": 2.2816,
      "step": 211932
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.4405851364135742,
      "learning_rate": 9.463502720808591e-06,
      "loss": 1.9136,
      "step": 211933
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.3756011724472046,
      "learning_rate": 9.462483426584334e-06,
      "loss": 1.9534,
      "step": 211934
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.4128525257110596,
      "learning_rate": 9.461464186376467e-06,
      "loss": 2.0019,
      "step": 211935
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.3578153848648071,
      "learning_rate": 9.46044500018519e-06,
      "loss": 2.0952,
      "step": 211936
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.4635781049728394,
      "learning_rate": 9.459425868010705e-06,
      "loss": 2.1334,
      "step": 211937
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.3821852207183838,
      "learning_rate": 9.458406789853279e-06,
      "loss": 1.9545,
      "step": 211938
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.562915325164795,
      "learning_rate": 9.457387765712942e-06,
      "loss": 1.8203,
      "step": 211939
    },
    {
      "epoch": 2.76,
      "grad_norm": 2.408623218536377,
      "learning_rate": 9.456368795589964e-06,
      "loss": 2.0621,
      "step": 211940
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.3437302112579346,
      "learning_rate": 9.455349879484541e-06,
      "loss": 1.6639,
      "step": 211941
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.388171911239624,
      "learning_rate": 9.454331017396844e-06,
      "loss": 1.8366,
      "step": 211942
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.6176073551177979,
      "learning_rate": 9.453312209327069e-06,
      "loss": 1.85,
      "step": 211943
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.5288844108581543,
      "learning_rate": 9.45229345527545e-06,
      "loss": 1.9216,
      "step": 211944
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.7509419918060303,
      "learning_rate": 9.451274755242088e-06,
      "loss": 2.2178,
      "step": 211945
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.5987966060638428,
      "learning_rate": 9.45025610922725e-06,
      "loss": 1.9327,
      "step": 211946
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.5023528337478638,
      "learning_rate": 9.449237517231068e-06,
      "loss": 1.9735,
      "step": 211947
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.5654948949813843,
      "learning_rate": 9.448218979253774e-06,
      "loss": 1.9769,
      "step": 211948
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.4292091131210327,
      "learning_rate": 9.447200495295504e-06,
      "loss": 2.1219,
      "step": 211949
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.4994890689849854,
      "learning_rate": 9.446182065356556e-06,
      "loss": 2.0862,
      "step": 211950
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.5666279792785645,
      "learning_rate": 9.445163689436997e-06,
      "loss": 2.0773,
      "step": 211951
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.483569622039795,
      "learning_rate": 9.444145367537092e-06,
      "loss": 1.991,
      "step": 211952
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.4935752153396606,
      "learning_rate": 9.443127099656945e-06,
      "loss": 1.9789,
      "step": 211953
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.3683196306228638,
      "learning_rate": 9.442108885796884e-06,
      "loss": 1.8809,
      "step": 211954
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.3740344047546387,
      "learning_rate": 9.441090725956946e-06,
      "loss": 2.0189,
      "step": 211955
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.5691323280334473,
      "learning_rate": 9.440072620137496e-06,
      "loss": 1.9349,
      "step": 211956
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.4843162298202515,
      "learning_rate": 9.439054568338533e-06,
      "loss": 1.8416,
      "step": 211957
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.6266255378723145,
      "learning_rate": 9.438036570560358e-06,
      "loss": 1.9804,
      "step": 211958
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.5355656147003174,
      "learning_rate": 9.437018626803106e-06,
      "loss": 1.837,
      "step": 211959
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.4803316593170166,
      "learning_rate": 9.43600073706704e-06,
      "loss": 2.3264,
      "step": 211960
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.5175727605819702,
      "learning_rate": 9.434982901352295e-06,
      "loss": 2.0158,
      "step": 211961
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.4263449907302856,
      "learning_rate": 9.433965119659071e-06,
      "loss": 2.0614,
      "step": 211962
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.4835578203201294,
      "learning_rate": 9.432947391987567e-06,
      "loss": 2.0475,
      "step": 211963
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.404377818107605,
      "learning_rate": 9.431929718337983e-06,
      "loss": 2.0465,
      "step": 211964
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.4704580307006836,
      "learning_rate": 9.430912098710419e-06,
      "loss": 1.9948,
      "step": 211965
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.5288565158843994,
      "learning_rate": 9.429894533105176e-06,
      "loss": 2.1201,
      "step": 211966
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.5011965036392212,
      "learning_rate": 9.428877021522385e-06,
      "loss": 1.9231,
      "step": 211967
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.4072933197021484,
      "learning_rate": 9.427859563962281e-06,
      "loss": 1.9753,
      "step": 211968
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.4876302480697632,
      "learning_rate": 9.42684216042503e-06,
      "loss": 1.8027,
      "step": 211969
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.4339808225631714,
      "learning_rate": 9.425824810910798e-06,
      "loss": 2.1595,
      "step": 211970
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.6092162132263184,
      "learning_rate": 9.424807515419752e-06,
      "loss": 1.8507,
      "step": 211971
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.4400417804718018,
      "learning_rate": 9.423790273952158e-06,
      "loss": 1.8857,
      "step": 211972
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.389236330986023,
      "learning_rate": 9.422773086508151e-06,
      "loss": 1.8602,
      "step": 211973
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.4334547519683838,
      "learning_rate": 9.421755953087929e-06,
      "loss": 1.883,
      "step": 211974
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.5291985273361206,
      "learning_rate": 9.420738873691724e-06,
      "loss": 2.0867,
      "step": 211975
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.478793978691101,
      "learning_rate": 9.41972184831964e-06,
      "loss": 1.8088,
      "step": 211976
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.4614468812942505,
      "learning_rate": 9.418704876971972e-06,
      "loss": 1.9288,
      "step": 211977
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.5342501401901245,
      "learning_rate": 9.417687959648823e-06,
      "loss": 1.9157,
      "step": 211978
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.4004194736480713,
      "learning_rate": 9.416671096350392e-06,
      "loss": 1.9622,
      "step": 211979
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.7767068147659302,
      "learning_rate": 9.415654287076912e-06,
      "loss": 2.0126,
      "step": 211980
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.6242496967315674,
      "learning_rate": 9.414637531828551e-06,
      "loss": 2.0422,
      "step": 211981
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.5295246839523315,
      "learning_rate": 9.413620830605472e-06,
      "loss": 2.2273,
      "step": 211982
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.4047728776931763,
      "learning_rate": 9.41260418340788e-06,
      "loss": 1.9245,
      "step": 211983
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.72905695438385,
      "learning_rate": 9.411587590236003e-06,
      "loss": 2.2121,
      "step": 211984
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.4895776510238647,
      "learning_rate": 9.410571051089976e-06,
      "loss": 2.014,
      "step": 211985
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.4114152193069458,
      "learning_rate": 9.409554565970002e-06,
      "loss": 1.8831,
      "step": 211986
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.4978151321411133,
      "learning_rate": 9.40853813487631e-06,
      "loss": 2.0725,
      "step": 211987
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.4112036228179932,
      "learning_rate": 9.407521757809e-06,
      "loss": 1.962,
      "step": 211988
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.4706885814666748,
      "learning_rate": 9.406505434768374e-06,
      "loss": 2.0981,
      "step": 211989
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.5973843336105347,
      "learning_rate": 9.405489165754532e-06,
      "loss": 1.8695,
      "step": 211990
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.35215425491333,
      "learning_rate": 9.404472950767705e-06,
      "loss": 1.856,
      "step": 211991
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.5341358184814453,
      "learning_rate": 9.403456789808095e-06,
      "loss": 2.0767,
      "step": 211992
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.4978207349777222,
      "learning_rate": 9.402440682875834e-06,
      "loss": 2.1551,
      "step": 211993
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.5278034210205078,
      "learning_rate": 9.401424629971154e-06,
      "loss": 1.8533,
      "step": 211994
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.433258056640625,
      "learning_rate": 9.400408631094258e-06,
      "loss": 1.9951,
      "step": 211995
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.4482030868530273,
      "learning_rate": 9.399392686245243e-06,
      "loss": 1.9292,
      "step": 211996
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.4498389959335327,
      "learning_rate": 9.398376795424445e-06,
      "loss": 1.82,
      "step": 211997
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.5622508525848389,
      "learning_rate": 9.397360958631894e-06,
      "loss": 1.9613,
      "step": 211998
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.3724430799484253,
      "learning_rate": 9.396345175867958e-06,
      "loss": 2.1103,
      "step": 211999
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.3955961465835571,
      "learning_rate": 9.395329447132671e-06,
      "loss": 1.9231,
      "step": 212000
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.5817208290100098,
      "learning_rate": 9.394313772426299e-06,
      "loss": 1.9886,
      "step": 212001
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.373536229133606,
      "learning_rate": 9.393298151748973e-06,
      "loss": 1.9914,
      "step": 212002
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.463249683380127,
      "learning_rate": 9.39228258510093e-06,
      "loss": 1.8119,
      "step": 212003
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.4065488576889038,
      "learning_rate": 9.391267072482333e-06,
      "loss": 2.0117,
      "step": 212004
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.439098596572876,
      "learning_rate": 9.39025161389345e-06,
      "loss": 1.936,
      "step": 212005
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.5480254888534546,
      "learning_rate": 9.38923620933435e-06,
      "loss": 2.0925,
      "step": 212006
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.3929086923599243,
      "learning_rate": 9.388220858805295e-06,
      "loss": 1.8404,
      "step": 212007
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.4129623174667358,
      "learning_rate": 9.387205562306455e-06,
      "loss": 1.8568,
      "step": 212008
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.3271746635437012,
      "learning_rate": 9.386190319838027e-06,
      "loss": 2.1671,
      "step": 212009
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.5190470218658447,
      "learning_rate": 9.385175131400146e-06,
      "loss": 1.956,
      "step": 212010
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.6357312202453613,
      "learning_rate": 9.384159996993112e-06,
      "loss": 1.9769,
      "step": 212011
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.3975664377212524,
      "learning_rate": 9.383144916616992e-06,
      "loss": 2.0955,
      "step": 212012
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.5785070657730103,
      "learning_rate": 9.382129890272082e-06,
      "loss": 1.9263,
      "step": 212013
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.4647611379623413,
      "learning_rate": 9.381114917958488e-06,
      "loss": 1.975,
      "step": 212014
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.4681775569915771,
      "learning_rate": 9.380099999676439e-06,
      "loss": 2.0274,
      "step": 212015
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.4992485046386719,
      "learning_rate": 9.379085135426101e-06,
      "loss": 2.0558,
      "step": 212016
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.522299885749817,
      "learning_rate": 9.37807032520771e-06,
      "loss": 1.9479,
      "step": 212017
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.4825962781906128,
      "learning_rate": 9.377055569021397e-06,
      "loss": 1.8366,
      "step": 212018
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.4211922883987427,
      "learning_rate": 9.376040866867395e-06,
      "loss": 1.741,
      "step": 212019
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.4058284759521484,
      "learning_rate": 9.375026218745807e-06,
      "loss": 1.7851,
      "step": 212020
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.5652130842208862,
      "learning_rate": 9.374011624656963e-06,
      "loss": 2.1708,
      "step": 212021
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.5370749235153198,
      "learning_rate": 9.37299708460093e-06,
      "loss": 2.0932,
      "step": 212022
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.6949502229690552,
      "learning_rate": 9.37198259857801e-06,
      "loss": 2.0879,
      "step": 212023
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.466160535812378,
      "learning_rate": 9.370968166588267e-06,
      "loss": 2.0502,
      "step": 212024
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.4007447957992554,
      "learning_rate": 9.369953788631967e-06,
      "loss": 2.066,
      "step": 212025
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.4100333452224731,
      "learning_rate": 9.368939464709247e-06,
      "loss": 1.9577,
      "step": 212026
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.560123085975647,
      "learning_rate": 9.367925194820336e-06,
      "loss": 2.012,
      "step": 212027
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.374913215637207,
      "learning_rate": 9.366910978965436e-06,
      "loss": 1.8302,
      "step": 212028
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.3705558776855469,
      "learning_rate": 9.365896817144746e-06,
      "loss": 1.8028,
      "step": 212029
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.5549718141555786,
      "learning_rate": 9.364882709358335e-06,
      "loss": 2.0317,
      "step": 212030
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.4888452291488647,
      "learning_rate": 9.363868655606565e-06,
      "loss": 2.1231,
      "step": 212031
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.4648982286453247,
      "learning_rate": 9.362854655889474e-06,
      "loss": 2.0729,
      "step": 212032
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.4656330347061157,
      "learning_rate": 9.361840710207324e-06,
      "loss": 2.235,
      "step": 212033
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.4652533531188965,
      "learning_rate": 9.360826818560285e-06,
      "loss": 1.8438,
      "step": 212034
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.3306255340576172,
      "learning_rate": 9.359812980948622e-06,
      "loss": 1.8294,
      "step": 212035
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.5588672161102295,
      "learning_rate": 9.358799197372402e-06,
      "loss": 1.787,
      "step": 212036
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.4828135967254639,
      "learning_rate": 9.357785467831892e-06,
      "loss": 2.128,
      "step": 212037
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.5587743520736694,
      "learning_rate": 9.356771792327222e-06,
      "loss": 2.1211,
      "step": 212038
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.5491724014282227,
      "learning_rate": 9.355758170858629e-06,
      "loss": 2.0317,
      "step": 212039
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.6708751916885376,
      "learning_rate": 9.35474460342628e-06,
      "loss": 2.1677,
      "step": 212040
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.3601969480514526,
      "learning_rate": 9.353731090030437e-06,
      "loss": 1.9891,
      "step": 212041
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.485153317451477,
      "learning_rate": 9.352717630671136e-06,
      "loss": 1.8657,
      "step": 212042
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.491195797920227,
      "learning_rate": 9.351704225348711e-06,
      "loss": 2.1004,
      "step": 212043
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.5328078269958496,
      "learning_rate": 9.350690874063227e-06,
      "loss": 2.1298,
      "step": 212044
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.368500828742981,
      "learning_rate": 9.349677576814985e-06,
      "loss": 1.79,
      "step": 212045
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.3300033807754517,
      "learning_rate": 9.348664333604083e-06,
      "loss": 1.9952,
      "step": 212046
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.4966591596603394,
      "learning_rate": 9.347651144430823e-06,
      "loss": 1.9584,
      "step": 212047
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.385070562362671,
      "learning_rate": 9.346638009295238e-06,
      "loss": 2.0368,
      "step": 212048
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.500929594039917,
      "learning_rate": 9.34562492819766e-06,
      "loss": 1.9108,
      "step": 212049
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.3911021947860718,
      "learning_rate": 9.344611901138188e-06,
      "loss": 2.2339,
      "step": 212050
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.448309063911438,
      "learning_rate": 9.343598928117058e-06,
      "loss": 1.9015,
      "step": 212051
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.4487885236740112,
      "learning_rate": 9.342586009134401e-06,
      "loss": 1.9602,
      "step": 212052
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.3431612253189087,
      "learning_rate": 9.341573144190485e-06,
      "loss": 1.7129,
      "step": 212053
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.4804279804229736,
      "learning_rate": 9.34056033328544e-06,
      "loss": 1.9553,
      "step": 212054
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.3443634510040283,
      "learning_rate": 9.339547576419471e-06,
      "loss": 1.6975,
      "step": 212055
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.5270217657089233,
      "learning_rate": 9.338534873592773e-06,
      "loss": 1.9697,
      "step": 212056
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.5677080154418945,
      "learning_rate": 9.337522224805516e-06,
      "loss": 2.0459,
      "step": 212057
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.5317262411117554,
      "learning_rate": 9.336509630057898e-06,
      "loss": 2.0538,
      "step": 212058
    },
    {
      "epoch": 2.76,
      "grad_norm": 2.3528385162353516,
      "learning_rate": 9.335497089350118e-06,
      "loss": 2.0036,
      "step": 212059
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.5909615755081177,
      "learning_rate": 9.33448460268238e-06,
      "loss": 2.1704,
      "step": 212060
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.347731113433838,
      "learning_rate": 9.333472170054812e-06,
      "loss": 1.9062,
      "step": 212061
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.4458452463150024,
      "learning_rate": 9.33245979146765e-06,
      "loss": 1.878,
      "step": 212062
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.2961674928665161,
      "learning_rate": 9.331447466921093e-06,
      "loss": 1.8576,
      "step": 212063
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.61583411693573,
      "learning_rate": 9.330435196415276e-06,
      "loss": 2.1208,
      "step": 212064
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.537279486656189,
      "learning_rate": 9.329422979950429e-06,
      "loss": 1.9601,
      "step": 212065
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.499465823173523,
      "learning_rate": 9.328410817526722e-06,
      "loss": 1.8727,
      "step": 212066
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.5338377952575684,
      "learning_rate": 9.327398709144351e-06,
      "loss": 1.899,
      "step": 212067
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.4037973880767822,
      "learning_rate": 9.326386654803519e-06,
      "loss": 2.1053,
      "step": 212068
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.5206387042999268,
      "learning_rate": 9.325374654504392e-06,
      "loss": 1.9672,
      "step": 212069
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.4341943264007568,
      "learning_rate": 9.324362708247135e-06,
      "loss": 1.9789,
      "step": 212070
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.6820740699768066,
      "learning_rate": 9.323350816032015e-06,
      "loss": 1.9799,
      "step": 212071
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.4526017904281616,
      "learning_rate": 9.322338977859167e-06,
      "loss": 1.9938,
      "step": 212072
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.5255346298217773,
      "learning_rate": 9.321327193728723e-06,
      "loss": 2.176,
      "step": 212073
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.37913978099823,
      "learning_rate": 9.320315463640982e-06,
      "loss": 1.8517,
      "step": 212074
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.5717530250549316,
      "learning_rate": 9.319303787596077e-06,
      "loss": 2.0704,
      "step": 212075
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.4590826034545898,
      "learning_rate": 9.31829216559421e-06,
      "loss": 1.9709,
      "step": 212076
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.4568328857421875,
      "learning_rate": 9.317280597635546e-06,
      "loss": 2.1371,
      "step": 212077
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.4238406419754028,
      "learning_rate": 9.316269083720285e-06,
      "loss": 1.8902,
      "step": 212078
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.3927974700927734,
      "learning_rate": 9.315257623848593e-06,
      "loss": 2.007,
      "step": 212079
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.4686191082000732,
      "learning_rate": 9.314246218020738e-06,
      "loss": 2.0873,
      "step": 212080
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.3792262077331543,
      "learning_rate": 9.313234866236786e-06,
      "loss": 1.8322,
      "step": 212081
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.4554882049560547,
      "learning_rate": 9.312223568497035e-06,
      "loss": 1.9281,
      "step": 212082
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.5075345039367676,
      "learning_rate": 9.311212324801586e-06,
      "loss": 1.7654,
      "step": 212083
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.943568229675293,
      "learning_rate": 9.310201135150775e-06,
      "loss": 1.7749,
      "step": 212084
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.481242299079895,
      "learning_rate": 9.309189999544564e-06,
      "loss": 2.062,
      "step": 212085
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.3894233703613281,
      "learning_rate": 9.308178917983322e-06,
      "loss": 1.9697,
      "step": 212086
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.4387857913970947,
      "learning_rate": 9.30716789046715e-06,
      "loss": 2.0153,
      "step": 212087
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.435104489326477,
      "learning_rate": 9.306156916996277e-06,
      "loss": 1.8184,
      "step": 212088
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.6215972900390625,
      "learning_rate": 9.305145997570873e-06,
      "loss": 1.9946,
      "step": 212089
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.5370020866394043,
      "learning_rate": 9.30413513219117e-06,
      "loss": 2.1614,
      "step": 212090
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.4571980237960815,
      "learning_rate": 9.30312432085727e-06,
      "loss": 1.8834,
      "step": 212091
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.365774154663086,
      "learning_rate": 9.302113563569435e-06,
      "loss": 1.815,
      "step": 212092
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.3924809694290161,
      "learning_rate": 9.30110286032777e-06,
      "loss": 1.9516,
      "step": 212093
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.4530987739562988,
      "learning_rate": 9.30009221113257e-06,
      "loss": 1.9284,
      "step": 212094
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.628011703491211,
      "learning_rate": 9.299081615983938e-06,
      "loss": 2.0003,
      "step": 212095
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.5474474430084229,
      "learning_rate": 9.298071074882141e-06,
      "loss": 1.8714,
      "step": 212096
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.4806981086730957,
      "learning_rate": 9.297060587827276e-06,
      "loss": 1.9313,
      "step": 212097
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.5279449224472046,
      "learning_rate": 9.296050154819578e-06,
      "loss": 1.7633,
      "step": 212098
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.390100359916687,
      "learning_rate": 9.295039775859215e-06,
      "loss": 1.9632,
      "step": 212099
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.5557692050933838,
      "learning_rate": 9.29402945094645e-06,
      "loss": 2.0963,
      "step": 212100
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.3413517475128174,
      "learning_rate": 9.293019180081352e-06,
      "loss": 2.0058,
      "step": 212101
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.5561175346374512,
      "learning_rate": 9.292008963264252e-06,
      "loss": 2.0019,
      "step": 212102
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.4258286952972412,
      "learning_rate": 9.290998800495154e-06,
      "loss": 1.8859,
      "step": 212103
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.4786654710769653,
      "learning_rate": 9.28998869177442e-06,
      "loss": 1.9526,
      "step": 212104
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.3930788040161133,
      "learning_rate": 9.288978637102118e-06,
      "loss": 1.853,
      "step": 212105
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.6081992387771606,
      "learning_rate": 9.287968636478516e-06,
      "loss": 2.0122,
      "step": 212106
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.6418964862823486,
      "learning_rate": 9.286958689903712e-06,
      "loss": 2.1742,
      "step": 212107
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.411656379699707,
      "learning_rate": 9.28594879737804e-06,
      "loss": 2.0686,
      "step": 212108
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.3989615440368652,
      "learning_rate": 9.284938958901533e-06,
      "loss": 1.7068,
      "step": 212109
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.5770225524902344,
      "learning_rate": 9.283929174474457e-06,
      "loss": 1.971,
      "step": 212110
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.4395672082901,
      "learning_rate": 9.282919444096981e-06,
      "loss": 1.9952,
      "step": 212111
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.399611473083496,
      "learning_rate": 9.281909767769335e-06,
      "loss": 1.791,
      "step": 212112
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.4402283430099487,
      "learning_rate": 9.280900145491622e-06,
      "loss": 1.8377,
      "step": 212113
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.3713936805725098,
      "learning_rate": 9.279890577264138e-06,
      "loss": 2.1422,
      "step": 212114
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.4480302333831787,
      "learning_rate": 9.278881063086919e-06,
      "loss": 2.0187,
      "step": 212115
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.5059868097305298,
      "learning_rate": 9.27787160296033e-06,
      "loss": 2.0726,
      "step": 212116
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.5205210447311401,
      "learning_rate": 9.276862196884405e-06,
      "loss": 2.1508,
      "step": 212117
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.4839030504226685,
      "learning_rate": 9.275852844859445e-06,
      "loss": 2.0606,
      "step": 212118
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.406730055809021,
      "learning_rate": 9.274843546885546e-06,
      "loss": 1.8956,
      "step": 212119
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.5286544561386108,
      "learning_rate": 9.273834302963013e-06,
      "loss": 2.1209,
      "step": 212120
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.619289755821228,
      "learning_rate": 9.272825113091908e-06,
      "loss": 1.9909,
      "step": 212121
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.7427208423614502,
      "learning_rate": 9.2718159772725e-06,
      "loss": 1.8719,
      "step": 212122
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.471727728843689,
      "learning_rate": 9.270806895504923e-06,
      "loss": 2.0571,
      "step": 212123
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.367655873298645,
      "learning_rate": 9.269797867789407e-06,
      "loss": 1.9583,
      "step": 212124
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.445204734802246,
      "learning_rate": 9.26878889412609e-06,
      "loss": 1.9512,
      "step": 212125
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.440760850906372,
      "learning_rate": 9.267779974515233e-06,
      "loss": 1.8481,
      "step": 212126
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.4677162170410156,
      "learning_rate": 9.266771108957006e-06,
      "loss": 1.9572,
      "step": 212127
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.381908893585205,
      "learning_rate": 9.26576229745154e-06,
      "loss": 1.8967,
      "step": 212128
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.596261739730835,
      "learning_rate": 9.264753539999036e-06,
      "loss": 1.9853,
      "step": 212129
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.4937760829925537,
      "learning_rate": 9.263744836599728e-06,
      "loss": 2.2441,
      "step": 212130
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.3952038288116455,
      "learning_rate": 9.26273618725375e-06,
      "loss": 2.1377,
      "step": 212131
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.3796440362930298,
      "learning_rate": 9.261727591961366e-06,
      "loss": 1.8017,
      "step": 212132
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.4917662143707275,
      "learning_rate": 9.26071905072271e-06,
      "loss": 1.9833,
      "step": 212133
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.5956530570983887,
      "learning_rate": 9.259710563537948e-06,
      "loss": 1.96,
      "step": 212134
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.3954004049301147,
      "learning_rate": 9.25870213040728e-06,
      "loss": 1.9664,
      "step": 212135
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.4177263975143433,
      "learning_rate": 9.257693751330942e-06,
      "loss": 2.1104,
      "step": 212136
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.4091825485229492,
      "learning_rate": 9.256685426309063e-06,
      "loss": 1.949,
      "step": 212137
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.3692405223846436,
      "learning_rate": 9.255677155341879e-06,
      "loss": 2.0593,
      "step": 212138
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.6437925100326538,
      "learning_rate": 9.254668938429554e-06,
      "loss": 2.2072,
      "step": 212139
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.4646812677383423,
      "learning_rate": 9.253660775572259e-06,
      "loss": 1.9771,
      "step": 212140
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.4417225122451782,
      "learning_rate": 9.252652666770188e-06,
      "loss": 1.9582,
      "step": 212141
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.529981017112732,
      "learning_rate": 9.251644612023546e-06,
      "loss": 1.9507,
      "step": 212142
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.6421622037887573,
      "learning_rate": 9.250636611332495e-06,
      "loss": 1.9266,
      "step": 212143
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.5001128911972046,
      "learning_rate": 9.249628664697273e-06,
      "loss": 1.8681,
      "step": 212144
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.5375245809555054,
      "learning_rate": 9.248620772118043e-06,
      "loss": 1.7478,
      "step": 212145
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.4239531755447388,
      "learning_rate": 9.24761293359494e-06,
      "loss": 1.9062,
      "step": 212146
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.4408743381500244,
      "learning_rate": 9.246605149128195e-06,
      "loss": 1.7932,
      "step": 212147
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.4326907396316528,
      "learning_rate": 9.245597418718042e-06,
      "loss": 2.1043,
      "step": 212148
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.5268465280532837,
      "learning_rate": 9.24458974236455e-06,
      "loss": 2.1919,
      "step": 212149
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.3962265253067017,
      "learning_rate": 9.24358212006805e-06,
      "loss": 1.9484,
      "step": 212150
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.4465335607528687,
      "learning_rate": 9.242574551828608e-06,
      "loss": 1.8161,
      "step": 212151
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.440199375152588,
      "learning_rate": 9.241567037646458e-06,
      "loss": 2.0541,
      "step": 212152
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.5409510135650635,
      "learning_rate": 9.240559577521833e-06,
      "loss": 2.1089,
      "step": 212153
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.4473150968551636,
      "learning_rate": 9.239552171454835e-06,
      "loss": 1.8332,
      "step": 212154
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.6134966611862183,
      "learning_rate": 9.238544819445692e-06,
      "loss": 2.2146,
      "step": 212155
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.5005279779434204,
      "learning_rate": 9.23753752149461e-06,
      "loss": 2.0593,
      "step": 212156
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.4794858694076538,
      "learning_rate": 9.236530277601783e-06,
      "loss": 1.9807,
      "step": 212157
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.4107497930526733,
      "learning_rate": 9.235523087767316e-06,
      "loss": 1.8228,
      "step": 212158
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.616119384765625,
      "learning_rate": 9.234515951991506e-06,
      "loss": 2.1081,
      "step": 212159
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.4136106967926025,
      "learning_rate": 9.233508870274453e-06,
      "loss": 2.142,
      "step": 212160
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.507016897201538,
      "learning_rate": 9.232501842616392e-06,
      "loss": 2.0951,
      "step": 212161
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.4000684022903442,
      "learning_rate": 9.23149486901752e-06,
      "loss": 1.7848,
      "step": 212162
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.4091850519180298,
      "learning_rate": 9.23048794947797e-06,
      "loss": 2.014,
      "step": 212163
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.4627095460891724,
      "learning_rate": 9.229481083997947e-06,
      "loss": 2.0447,
      "step": 212164
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.5348634719848633,
      "learning_rate": 9.228474272577713e-06,
      "loss": 1.9099,
      "step": 212165
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.5864086151123047,
      "learning_rate": 9.227467515217335e-06,
      "loss": 2.0321,
      "step": 212166
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.4624254703521729,
      "learning_rate": 9.22646081191708e-06,
      "loss": 1.7261,
      "step": 212167
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.4748355150222778,
      "learning_rate": 9.225454162677115e-06,
      "loss": 2.0583,
      "step": 212168
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.6344528198242188,
      "learning_rate": 9.224447567497639e-06,
      "loss": 1.9193,
      "step": 212169
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.4761557579040527,
      "learning_rate": 9.22344102637882e-06,
      "loss": 1.8814,
      "step": 212170
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.6147522926330566,
      "learning_rate": 9.222434539320856e-06,
      "loss": 1.7678,
      "step": 212171
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.3582361936569214,
      "learning_rate": 9.22142810632388e-06,
      "loss": 1.874,
      "step": 212172
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.3661247491836548,
      "learning_rate": 9.220421727388195e-06,
      "loss": 1.7803,
      "step": 212173
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.3547405004501343,
      "learning_rate": 9.219415402513897e-06,
      "loss": 2.1647,
      "step": 212174
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.4521833658218384,
      "learning_rate": 9.21840913170122e-06,
      "loss": 2.0757,
      "step": 212175
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.3890398740768433,
      "learning_rate": 9.2174029149503e-06,
      "loss": 1.9287,
      "step": 212176
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.493862509727478,
      "learning_rate": 9.216396752261368e-06,
      "loss": 1.8711,
      "step": 212177
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.510735034942627,
      "learning_rate": 9.215390643634591e-06,
      "loss": 2.0687,
      "step": 212178
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.537483811378479,
      "learning_rate": 9.214384589070167e-06,
      "loss": 2.0661,
      "step": 212179
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.440637469291687,
      "learning_rate": 9.213378588568266e-06,
      "loss": 1.885,
      "step": 212180
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.5371862649917603,
      "learning_rate": 9.212372642129118e-06,
      "loss": 2.1654,
      "step": 212181
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.3959263563156128,
      "learning_rate": 9.211366749752858e-06,
      "loss": 1.9402,
      "step": 212182
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.5780471563339233,
      "learning_rate": 9.210360911439686e-06,
      "loss": 1.8026,
      "step": 212183
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.5182807445526123,
      "learning_rate": 9.209355127189799e-06,
      "loss": 2.0908,
      "step": 212184
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.5240477323532104,
      "learning_rate": 9.208349397003367e-06,
      "loss": 2.0101,
      "step": 212185
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.4850754737854004,
      "learning_rate": 9.207343720880622e-06,
      "loss": 1.9342,
      "step": 212186
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.3784619569778442,
      "learning_rate": 9.206338098821697e-06,
      "loss": 1.7816,
      "step": 212187
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.3950207233428955,
      "learning_rate": 9.205332530826826e-06,
      "loss": 1.7224,
      "step": 212188
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.4789233207702637,
      "learning_rate": 9.204327016896174e-06,
      "loss": 1.8912,
      "step": 212189
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.6302765607833862,
      "learning_rate": 9.203321557029908e-06,
      "loss": 2.118,
      "step": 212190
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.4107643365859985,
      "learning_rate": 9.202316151228262e-06,
      "loss": 2.0314,
      "step": 212191
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.4388145208358765,
      "learning_rate": 9.201310799491334e-06,
      "loss": 2.0074,
      "step": 212192
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.4756813049316406,
      "learning_rate": 9.200305501819427e-06,
      "loss": 2.0161,
      "step": 212193
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.3933346271514893,
      "learning_rate": 9.199300258212671e-06,
      "loss": 1.7636,
      "step": 212194
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.4888795614242554,
      "learning_rate": 9.19829506867127e-06,
      "loss": 1.9034,
      "step": 212195
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.5565769672393799,
      "learning_rate": 9.19728993319535e-06,
      "loss": 1.8887,
      "step": 212196
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.6259515285491943,
      "learning_rate": 9.196284851785151e-06,
      "loss": 2.1979,
      "step": 212197
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.6204612255096436,
      "learning_rate": 9.195279824440838e-06,
      "loss": 2.0592,
      "step": 212198
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.546971082687378,
      "learning_rate": 9.194274851162641e-06,
      "loss": 1.9794,
      "step": 212199
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.420926570892334,
      "learning_rate": 9.19326993195073e-06,
      "loss": 1.9139,
      "step": 212200
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.410267949104309,
      "learning_rate": 9.192265066805271e-06,
      "loss": 1.8324,
      "step": 212201
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.5024783611297607,
      "learning_rate": 9.19126025572643e-06,
      "loss": 1.991,
      "step": 212202
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.3788681030273438,
      "learning_rate": 9.190255498714472e-06,
      "loss": 1.9389,
      "step": 212203
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.5482194423675537,
      "learning_rate": 9.189250795769465e-06,
      "loss": 1.9052,
      "step": 212204
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.3936123847961426,
      "learning_rate": 9.188246146891709e-06,
      "loss": 1.8402,
      "step": 212205
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.4789135456085205,
      "learning_rate": 9.18724155208137e-06,
      "loss": 1.9984,
      "step": 212206
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.5838749408721924,
      "learning_rate": 9.186237011338583e-06,
      "loss": 1.9741,
      "step": 212207
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.6327728033065796,
      "learning_rate": 9.185232524663543e-06,
      "loss": 1.9833,
      "step": 212208
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.4834774732589722,
      "learning_rate": 9.184228092056522e-06,
      "loss": 1.9141,
      "step": 212209
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.466269612312317,
      "learning_rate": 9.183223713517585e-06,
      "loss": 1.89,
      "step": 212210
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.5431835651397705,
      "learning_rate": 9.182219389046997e-06,
      "loss": 2.1579,
      "step": 212211
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.464585542678833,
      "learning_rate": 9.181215118644925e-06,
      "loss": 1.6612,
      "step": 212212
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.5454415082931519,
      "learning_rate": 9.180210902311569e-06,
      "loss": 1.8907,
      "step": 212213
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.4236153364181519,
      "learning_rate": 9.179206740047063e-06,
      "loss": 2.0376,
      "step": 212214
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.4494974613189697,
      "learning_rate": 9.178202631851673e-06,
      "loss": 1.9447,
      "step": 212215
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.4871430397033691,
      "learning_rate": 9.177198577725497e-06,
      "loss": 2.0273,
      "step": 212216
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.5510764122009277,
      "learning_rate": 9.176194577668805e-06,
      "loss": 2.041,
      "step": 212217
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.4677444696426392,
      "learning_rate": 9.175190631681762e-06,
      "loss": 1.7896,
      "step": 212218
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.4190298318862915,
      "learning_rate": 9.1741867397645e-06,
      "loss": 1.8017,
      "step": 212219
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.5051865577697754,
      "learning_rate": 9.173182901917253e-06,
      "loss": 1.9207,
      "step": 212220
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.6494455337524414,
      "learning_rate": 9.172179118140221e-06,
      "loss": 1.8737,
      "step": 212221
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.5130115747451782,
      "learning_rate": 9.171175388433537e-06,
      "loss": 1.8842,
      "step": 212222
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.5345284938812256,
      "learning_rate": 9.17017171279747e-06,
      "loss": 2.035,
      "step": 212223
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.641749620437622,
      "learning_rate": 9.169168091232148e-06,
      "loss": 2.1288,
      "step": 212224
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.3434559106826782,
      "learning_rate": 9.168164523737776e-06,
      "loss": 1.7575,
      "step": 212225
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.369606852531433,
      "learning_rate": 9.167161010314484e-06,
      "loss": 2.1619,
      "step": 212226
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.4158185720443726,
      "learning_rate": 9.166157550962538e-06,
      "loss": 2.0713,
      "step": 212227
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.576528549194336,
      "learning_rate": 9.165154145682075e-06,
      "loss": 1.6441,
      "step": 212228
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.2838656902313232,
      "learning_rate": 9.164150794473324e-06,
      "loss": 1.9083,
      "step": 212229
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.5152559280395508,
      "learning_rate": 9.163147497336454e-06,
      "loss": 1.876,
      "step": 212230
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.3570873737335205,
      "learning_rate": 9.16214425427163e-06,
      "loss": 2.1582,
      "step": 212231
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.373909831047058,
      "learning_rate": 9.16114106527902e-06,
      "loss": 1.7398,
      "step": 212232
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.5315228700637817,
      "learning_rate": 9.16013793035889e-06,
      "loss": 2.0745,
      "step": 212233
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.3984915018081665,
      "learning_rate": 9.15913484951134e-06,
      "loss": 1.9693,
      "step": 212234
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.441624641418457,
      "learning_rate": 9.158131822736637e-06,
      "loss": 1.8097,
      "step": 212235
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.4787976741790771,
      "learning_rate": 9.157128850034945e-06,
      "loss": 1.9979,
      "step": 212236
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.4814435243606567,
      "learning_rate": 9.156125931406365e-06,
      "loss": 1.9676,
      "step": 212237
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.3961502313613892,
      "learning_rate": 9.1551230668512e-06,
      "loss": 2.1189,
      "step": 212238
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.5383728742599487,
      "learning_rate": 9.154120256369579e-06,
      "loss": 2.04,
      "step": 212239
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.5294815301895142,
      "learning_rate": 9.15311749996167e-06,
      "loss": 2.0746,
      "step": 212240
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.5768203735351562,
      "learning_rate": 9.152114797627741e-06,
      "loss": 1.9491,
      "step": 212241
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.3445252180099487,
      "learning_rate": 9.15111214936789e-06,
      "loss": 1.8722,
      "step": 212242
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.4646633863449097,
      "learning_rate": 9.150109555182316e-06,
      "loss": 2.0319,
      "step": 212243
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.6564644575119019,
      "learning_rate": 9.149107015071255e-06,
      "loss": 1.8887,
      "step": 212244
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.5789825916290283,
      "learning_rate": 9.148104529034872e-06,
      "loss": 1.9321,
      "step": 212245
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.5658175945281982,
      "learning_rate": 9.147102097073333e-06,
      "loss": 2.3059,
      "step": 212246
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.4487662315368652,
      "learning_rate": 9.146099719186838e-06,
      "loss": 1.7167,
      "step": 212247
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.6890860795974731,
      "learning_rate": 9.14509739537559e-06,
      "loss": 2.0918,
      "step": 212248
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.6361851692199707,
      "learning_rate": 9.14409512563975e-06,
      "loss": 1.8886,
      "step": 212249
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.695855975151062,
      "learning_rate": 9.143092909979488e-06,
      "loss": 1.9084,
      "step": 212250
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.71226966381073,
      "learning_rate": 9.142090748395037e-06,
      "loss": 2.1301,
      "step": 212251
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.409319519996643,
      "learning_rate": 9.141088640886563e-06,
      "loss": 2.1333,
      "step": 212252
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.522952914237976,
      "learning_rate": 9.140086587454265e-06,
      "loss": 2.1983,
      "step": 212253
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.4359105825424194,
      "learning_rate": 9.139084588098277e-06,
      "loss": 1.9375,
      "step": 212254
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.3287526369094849,
      "learning_rate": 9.138082642818868e-06,
      "loss": 1.9866,
      "step": 212255
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.5634808540344238,
      "learning_rate": 9.137080751616166e-06,
      "loss": 2.2167,
      "step": 212256
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.5616339445114136,
      "learning_rate": 9.136078914490341e-06,
      "loss": 1.8987,
      "step": 212257
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.3976231813430786,
      "learning_rate": 9.135077131441626e-06,
      "loss": 1.7509,
      "step": 212258
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.4048935174942017,
      "learning_rate": 9.134075402470186e-06,
      "loss": 1.6038,
      "step": 212259
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.5042593479156494,
      "learning_rate": 9.133073727576223e-06,
      "loss": 1.8447,
      "step": 212260
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.3502922058105469,
      "learning_rate": 9.132072106759903e-06,
      "loss": 2.1024,
      "step": 212261
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.566474199295044,
      "learning_rate": 9.131070540021457e-06,
      "loss": 2.1064,
      "step": 212262
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.4421026706695557,
      "learning_rate": 9.130069027360986e-06,
      "loss": 1.9263,
      "step": 212263
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.7039461135864258,
      "learning_rate": 9.129067568778758e-06,
      "loss": 2.1768,
      "step": 212264
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.3331269025802612,
      "learning_rate": 9.128066164274905e-06,
      "loss": 2.1446,
      "step": 212265
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.5493170022964478,
      "learning_rate": 9.127064813849628e-06,
      "loss": 2.0942,
      "step": 212266
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.351635217666626,
      "learning_rate": 9.126063517503157e-06,
      "loss": 2.0877,
      "step": 212267
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.5238361358642578,
      "learning_rate": 9.125062275235629e-06,
      "loss": 2.0358,
      "step": 212268
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.4304684400558472,
      "learning_rate": 9.124061087047241e-06,
      "loss": 2.1105,
      "step": 212269
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.5083372592926025,
      "learning_rate": 9.123059952938161e-06,
      "loss": 1.9995,
      "step": 212270
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.4795151948928833,
      "learning_rate": 9.122058872908588e-06,
      "loss": 1.9914,
      "step": 212271
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.469298243522644,
      "learning_rate": 9.121057846958756e-06,
      "loss": 2.0256,
      "step": 212272
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.5106364488601685,
      "learning_rate": 9.1200568750888e-06,
      "loss": 2.1737,
      "step": 212273
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.4023208618164062,
      "learning_rate": 9.119055957298914e-06,
      "loss": 1.9447,
      "step": 212274
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.5222511291503906,
      "learning_rate": 9.118055093589272e-06,
      "loss": 2.0334,
      "step": 212275
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.4906178712844849,
      "learning_rate": 9.117054283960101e-06,
      "loss": 2.0551,
      "step": 212276
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.5165361166000366,
      "learning_rate": 9.116053528411505e-06,
      "loss": 1.965,
      "step": 212277
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.53202486038208,
      "learning_rate": 9.115052826943781e-06,
      "loss": 1.9749,
      "step": 212278
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.5958722829818726,
      "learning_rate": 9.114052179557063e-06,
      "loss": 2.0746,
      "step": 212279
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.381686806678772,
      "learning_rate": 9.113051586251518e-06,
      "loss": 1.8357,
      "step": 212280
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.559781789779663,
      "learning_rate": 9.112051047027313e-06,
      "loss": 1.9463,
      "step": 212281
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.5043152570724487,
      "learning_rate": 9.111050561884714e-06,
      "loss": 2.0549,
      "step": 212282
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.4373974800109863,
      "learning_rate": 9.110050130823821e-06,
      "loss": 2.0543,
      "step": 212283
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.468125581741333,
      "learning_rate": 9.1090497538449e-06,
      "loss": 1.9496,
      "step": 212284
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.565643072128296,
      "learning_rate": 9.108049430948084e-06,
      "loss": 1.9075,
      "step": 212285
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.3968701362609863,
      "learning_rate": 9.107049162133573e-06,
      "loss": 1.8882,
      "step": 212286
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.4325639009475708,
      "learning_rate": 9.106048947401534e-06,
      "loss": 2.0798,
      "step": 212287
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.3602229356765747,
      "learning_rate": 9.1050487867522e-06,
      "loss": 2.1756,
      "step": 212288
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.544495940208435,
      "learning_rate": 9.104048680185704e-06,
      "loss": 1.9525,
      "step": 212289
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.620579481124878,
      "learning_rate": 9.103048627702248e-06,
      "loss": 2.0945,
      "step": 212290
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.5018922090530396,
      "learning_rate": 9.10204862930206e-06,
      "loss": 2.0809,
      "step": 212291
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.3877501487731934,
      "learning_rate": 9.101048684985313e-06,
      "loss": 1.8673,
      "step": 212292
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.4187839031219482,
      "learning_rate": 9.100048794752102e-06,
      "loss": 1.8696,
      "step": 212293
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.4237949848175049,
      "learning_rate": 9.099048958602728e-06,
      "loss": 1.9764,
      "step": 212294
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.445801019668579,
      "learning_rate": 9.098049176537291e-06,
      "loss": 1.8835,
      "step": 212295
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.5717476606369019,
      "learning_rate": 9.097049448556092e-06,
      "loss": 1.9906,
      "step": 212296
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.4859004020690918,
      "learning_rate": 9.096049774659198e-06,
      "loss": 1.9579,
      "step": 212297
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.3803266286849976,
      "learning_rate": 9.095050154846839e-06,
      "loss": 1.9034,
      "step": 212298
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.4482966661453247,
      "learning_rate": 9.094050589119184e-06,
      "loss": 1.9214,
      "step": 212299
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.356696367263794,
      "learning_rate": 9.093051077476466e-06,
      "loss": 1.7911,
      "step": 212300
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.468715786933899,
      "learning_rate": 9.092051619918816e-06,
      "loss": 1.9433,
      "step": 212301
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.4214484691619873,
      "learning_rate": 9.09105221644647e-06,
      "loss": 2.0573,
      "step": 212302
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.401440143585205,
      "learning_rate": 9.090052867059595e-06,
      "loss": 1.9052,
      "step": 212303
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.4669430255889893,
      "learning_rate": 9.089053571758387e-06,
      "loss": 1.9708,
      "step": 212304
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.4074047803878784,
      "learning_rate": 9.088054330542949e-06,
      "loss": 2.0987,
      "step": 212305
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.47732675075531,
      "learning_rate": 9.08705514341358e-06,
      "loss": 1.9056,
      "step": 212306
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.5393438339233398,
      "learning_rate": 9.086056010370379e-06,
      "loss": 1.875,
      "step": 212307
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.4000571966171265,
      "learning_rate": 9.085056931413614e-06,
      "loss": 2.0052,
      "step": 212308
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.4597657918930054,
      "learning_rate": 9.084057906543418e-06,
      "loss": 2.0522,
      "step": 212309
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.4543853998184204,
      "learning_rate": 9.08305893575999e-06,
      "loss": 2.0696,
      "step": 212310
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.4469034671783447,
      "learning_rate": 9.082060019063498e-06,
      "loss": 1.9288,
      "step": 212311
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.4947948455810547,
      "learning_rate": 9.08106115645414e-06,
      "loss": 2.1315,
      "step": 212312
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.5207929611206055,
      "learning_rate": 9.080062347932116e-06,
      "loss": 2.2167,
      "step": 212313
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.5564550161361694,
      "learning_rate": 9.079063593497593e-06,
      "loss": 1.791,
      "step": 212314
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.454294204711914,
      "learning_rate": 9.078064893150771e-06,
      "loss": 2.1035,
      "step": 212315
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.4698106050491333,
      "learning_rate": 9.07706624689185e-06,
      "loss": 2.0065,
      "step": 212316
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.5443106889724731,
      "learning_rate": 9.076067654720965e-06,
      "loss": 1.8648,
      "step": 212317
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.3503754138946533,
      "learning_rate": 9.075069116638345e-06,
      "loss": 2.1612,
      "step": 212318
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.503068208694458,
      "learning_rate": 9.074070632644126e-06,
      "loss": 1.7816,
      "step": 212319
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.5716716051101685,
      "learning_rate": 9.073072202738574e-06,
      "loss": 1.9895,
      "step": 212320
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.4030554294586182,
      "learning_rate": 9.072073826921788e-06,
      "loss": 2.069,
      "step": 212321
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.5192527770996094,
      "learning_rate": 9.071075505194003e-06,
      "loss": 1.8713,
      "step": 212322
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.3488845825195312,
      "learning_rate": 9.070077237555418e-06,
      "loss": 1.9725,
      "step": 212323
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.5050956010818481,
      "learning_rate": 9.069079024006232e-06,
      "loss": 2.0033,
      "step": 212324
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.5884268283843994,
      "learning_rate": 9.068080864546511e-06,
      "loss": 1.9268,
      "step": 212325
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.3688762187957764,
      "learning_rate": 9.067082759176592e-06,
      "loss": 1.784,
      "step": 212326
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.4345602989196777,
      "learning_rate": 9.066084707896538e-06,
      "loss": 2.0106,
      "step": 212327
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.4507505893707275,
      "learning_rate": 9.065086710706648e-06,
      "loss": 1.6943,
      "step": 212328
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.5995981693267822,
      "learning_rate": 9.064088767607024e-06,
      "loss": 1.9886,
      "step": 212329
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.309996485710144,
      "learning_rate": 9.0630908785979e-06,
      "loss": 1.9272,
      "step": 212330
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.6153342723846436,
      "learning_rate": 9.062093043679409e-06,
      "loss": 1.9932,
      "step": 212331
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.4438631534576416,
      "learning_rate": 9.061095262851781e-06,
      "loss": 1.7119,
      "step": 212332
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.6227518320083618,
      "learning_rate": 9.060097536115152e-06,
      "loss": 2.0325,
      "step": 212333
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.4468659162521362,
      "learning_rate": 9.059099863469788e-06,
      "loss": 1.9686,
      "step": 212334
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.656567931175232,
      "learning_rate": 9.058102244915822e-06,
      "loss": 1.9161,
      "step": 212335
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.4843626022338867,
      "learning_rate": 9.05710468045342e-06,
      "loss": 2.0814,
      "step": 212336
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.3729521036148071,
      "learning_rate": 9.056107170082815e-06,
      "loss": 1.9921,
      "step": 212337
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.4346410036087036,
      "learning_rate": 9.05510971380421e-06,
      "loss": 2.0188,
      "step": 212338
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.4095240831375122,
      "learning_rate": 9.054112311617668e-06,
      "loss": 1.9027,
      "step": 212339
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.477631688117981,
      "learning_rate": 9.053114963523523e-06,
      "loss": 1.9898,
      "step": 212340
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.4802207946777344,
      "learning_rate": 9.052117669521875e-06,
      "loss": 1.8966,
      "step": 212341
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.5242211818695068,
      "learning_rate": 9.051120429612924e-06,
      "loss": 1.9491,
      "step": 212342
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.530179738998413,
      "learning_rate": 9.05012324379687e-06,
      "loss": 1.9512,
      "step": 212343
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.8994975090026855,
      "learning_rate": 9.04912611207388e-06,
      "loss": 2.0881,
      "step": 212344
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.4343035221099854,
      "learning_rate": 9.048129034444151e-06,
      "loss": 2.0262,
      "step": 212345
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.5035996437072754,
      "learning_rate": 9.047132010907886e-06,
      "loss": 2.2115,
      "step": 212346
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.623962640762329,
      "learning_rate": 9.046135041465252e-06,
      "loss": 1.978,
      "step": 212347
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.5283421277999878,
      "learning_rate": 9.045138126116414e-06,
      "loss": 1.7248,
      "step": 212348
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.38337242603302,
      "learning_rate": 9.044141264861605e-06,
      "loss": 2.036,
      "step": 212349
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.4333232641220093,
      "learning_rate": 9.043144457700925e-06,
      "loss": 1.7847,
      "step": 212350
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.5733834505081177,
      "learning_rate": 9.042147704634672e-06,
      "loss": 1.9362,
      "step": 212351
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.3489705324172974,
      "learning_rate": 9.041151005662983e-06,
      "loss": 2.0048,
      "step": 212352
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.3510879278182983,
      "learning_rate": 9.040154360786023e-06,
      "loss": 2.1027,
      "step": 212353
    },
    {
      "epoch": 2.76,
      "grad_norm": 1.6052733659744263,
      "learning_rate": 9.039157770003958e-06,
      "loss": 1.9997,
      "step": 212354
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.4148492813110352,
      "learning_rate": 9.03816123331702e-06,
      "loss": 2.0092,
      "step": 212355
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.5664863586425781,
      "learning_rate": 9.03716475072538e-06,
      "loss": 1.9442,
      "step": 212356
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.340376853942871,
      "learning_rate": 9.036168322229232e-06,
      "loss": 2.062,
      "step": 212357
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.428558111190796,
      "learning_rate": 9.035171947828779e-06,
      "loss": 2.1224,
      "step": 212358
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.4126942157745361,
      "learning_rate": 9.034175627524155e-06,
      "loss": 2.0951,
      "step": 212359
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.4813224077224731,
      "learning_rate": 9.033179361315557e-06,
      "loss": 2.0623,
      "step": 212360
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.5757761001586914,
      "learning_rate": 9.032183149203187e-06,
      "loss": 1.9457,
      "step": 212361
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.3906524181365967,
      "learning_rate": 9.031186991187212e-06,
      "loss": 2.1858,
      "step": 212362
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.3423829078674316,
      "learning_rate": 9.030190887267862e-06,
      "loss": 1.9501,
      "step": 212363
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.5236777067184448,
      "learning_rate": 9.029194837445309e-06,
      "loss": 2.1408,
      "step": 212364
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.4258581399917603,
      "learning_rate": 9.028198841719681e-06,
      "loss": 1.9758,
      "step": 212365
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.3800910711288452,
      "learning_rate": 9.027202900091212e-06,
      "loss": 1.9855,
      "step": 212366
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.621380090713501,
      "learning_rate": 9.026207012560105e-06,
      "loss": 1.824,
      "step": 212367
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.4628658294677734,
      "learning_rate": 9.025211179126457e-06,
      "loss": 2.0387,
      "step": 212368
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.5244067907333374,
      "learning_rate": 9.02421539979057e-06,
      "loss": 1.8973,
      "step": 212369
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.451586127281189,
      "learning_rate": 9.023219674552572e-06,
      "loss": 1.9347,
      "step": 212370
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.5404149293899536,
      "learning_rate": 9.02222400341267e-06,
      "loss": 1.8634,
      "step": 212371
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.473178505897522,
      "learning_rate": 9.02122838637096e-06,
      "loss": 1.9533,
      "step": 212372
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.4107985496520996,
      "learning_rate": 9.020232823427743e-06,
      "loss": 1.8919,
      "step": 212373
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.8033066987991333,
      "learning_rate": 9.019237314583117e-06,
      "loss": 2.0085,
      "step": 212374
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.491661787033081,
      "learning_rate": 9.01824185983735e-06,
      "loss": 1.8608,
      "step": 212375
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.5856266021728516,
      "learning_rate": 9.017246459190608e-06,
      "loss": 1.9751,
      "step": 212376
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.4701615571975708,
      "learning_rate": 9.016251112643025e-06,
      "loss": 1.7884,
      "step": 212377
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.6364752054214478,
      "learning_rate": 9.015255820194766e-06,
      "loss": 2.1218,
      "step": 212378
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.5909876823425293,
      "learning_rate": 9.014260581846134e-06,
      "loss": 2.0813,
      "step": 212379
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.4125412702560425,
      "learning_rate": 9.01326539759719e-06,
      "loss": 2.2166,
      "step": 212380
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.4588944911956787,
      "learning_rate": 9.012270267448207e-06,
      "loss": 1.9398,
      "step": 212381
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.4408694505691528,
      "learning_rate": 9.011275191399315e-06,
      "loss": 1.9596,
      "step": 212382
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.4676196575164795,
      "learning_rate": 9.010280169450745e-06,
      "loss": 1.8796,
      "step": 212383
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.4525831937789917,
      "learning_rate": 9.009285201602635e-06,
      "loss": 2.0679,
      "step": 212384
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.3680360317230225,
      "learning_rate": 9.008290287855213e-06,
      "loss": 1.889,
      "step": 212385
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.5487642288208008,
      "learning_rate": 9.007295428208583e-06,
      "loss": 2.0538,
      "step": 212386
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.549550175666809,
      "learning_rate": 9.006300622663043e-06,
      "loss": 2.1181,
      "step": 212387
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.534838318824768,
      "learning_rate": 9.005305871218693e-06,
      "loss": 1.9234,
      "step": 212388
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.4582202434539795,
      "learning_rate": 9.004311173875833e-06,
      "loss": 2.0516,
      "step": 212389
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.4500503540039062,
      "learning_rate": 9.003316530634497e-06,
      "loss": 2.1279,
      "step": 212390
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.3675614595413208,
      "learning_rate": 9.00232194149495e-06,
      "loss": 1.8321,
      "step": 212391
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.4275310039520264,
      "learning_rate": 9.001327406457325e-06,
      "loss": 2.0733,
      "step": 212392
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.5033667087554932,
      "learning_rate": 9.000332925521892e-06,
      "loss": 1.8241,
      "step": 212393
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.7141613960266113,
      "learning_rate": 8.999338498688746e-06,
      "loss": 1.9164,
      "step": 212394
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.5587449073791504,
      "learning_rate": 8.99834412595819e-06,
      "loss": 1.9014,
      "step": 212395
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.4327250719070435,
      "learning_rate": 8.997349807330289e-06,
      "loss": 1.9691,
      "step": 212396
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.3545761108398438,
      "learning_rate": 8.996355542805278e-06,
      "loss": 2.083,
      "step": 212397
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.4110888242721558,
      "learning_rate": 8.995361332383322e-06,
      "loss": 2.1956,
      "step": 212398
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.451887607574463,
      "learning_rate": 8.994367176064654e-06,
      "loss": 1.9946,
      "step": 212399
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.4771015644073486,
      "learning_rate": 8.993373073849374e-06,
      "loss": 1.6812,
      "step": 212400
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.3730003833770752,
      "learning_rate": 8.992379025737817e-06,
      "loss": 1.9473,
      "step": 212401
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.5621834993362427,
      "learning_rate": 8.99138503172998e-06,
      "loss": 2.0855,
      "step": 212402
    },
    {
      "epoch": 2.77,
      "grad_norm": 2.509842872619629,
      "learning_rate": 8.990391091826166e-06,
      "loss": 1.8706,
      "step": 212403
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.465489149093628,
      "learning_rate": 8.989397206026538e-06,
      "loss": 2.0142,
      "step": 212404
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.6825858354568481,
      "learning_rate": 8.988403374331266e-06,
      "loss": 2.0361,
      "step": 212405
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.4994794130325317,
      "learning_rate": 8.987409596740513e-06,
      "loss": 2.1252,
      "step": 212406
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.4604802131652832,
      "learning_rate": 8.986415873254548e-06,
      "loss": 1.9144,
      "step": 212407
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.44533109664917,
      "learning_rate": 8.985422203873472e-06,
      "loss": 1.8233,
      "step": 212408
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.5059478282928467,
      "learning_rate": 8.984428588597515e-06,
      "loss": 2.157,
      "step": 212409
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.5303157567977905,
      "learning_rate": 8.983435027426844e-06,
      "loss": 1.957,
      "step": 212410
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.5375232696533203,
      "learning_rate": 8.982441520361628e-06,
      "loss": 2.2384,
      "step": 212411
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.6042627096176147,
      "learning_rate": 8.981448067402064e-06,
      "loss": 2.1421,
      "step": 212412
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.6524662971496582,
      "learning_rate": 8.980454668548386e-06,
      "loss": 1.95,
      "step": 212413
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.3935894966125488,
      "learning_rate": 8.979461323800697e-06,
      "loss": 2.0032,
      "step": 212414
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.4455329179763794,
      "learning_rate": 8.97846803315926e-06,
      "loss": 1.9853,
      "step": 212415
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.3363189697265625,
      "learning_rate": 8.977474796624174e-06,
      "loss": 1.9491,
      "step": 212416
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.44705331325531,
      "learning_rate": 8.976481614195675e-06,
      "loss": 1.8598,
      "step": 212417
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.5529717206954956,
      "learning_rate": 8.975488485873927e-06,
      "loss": 1.903,
      "step": 212418
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.4036625623703003,
      "learning_rate": 8.974495411659166e-06,
      "loss": 1.8151,
      "step": 212419
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.4590595960617065,
      "learning_rate": 8.973502391551556e-06,
      "loss": 2.0281,
      "step": 212420
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.3531057834625244,
      "learning_rate": 8.9725094255512e-06,
      "loss": 1.9916,
      "step": 212421
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.4534028768539429,
      "learning_rate": 8.971516513658395e-06,
      "loss": 2.173,
      "step": 212422
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.398995041847229,
      "learning_rate": 8.970523655873274e-06,
      "loss": 2.1396,
      "step": 212423
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.5112577676773071,
      "learning_rate": 8.969530852196006e-06,
      "loss": 2.0348,
      "step": 212424
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.4074773788452148,
      "learning_rate": 8.968538102626821e-06,
      "loss": 1.9663,
      "step": 212425
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.4288761615753174,
      "learning_rate": 8.967545407165888e-06,
      "loss": 1.9053,
      "step": 212426
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.516263484954834,
      "learning_rate": 8.96655276581334e-06,
      "loss": 2.115,
      "step": 212427
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.3186417818069458,
      "learning_rate": 8.965560178569441e-06,
      "loss": 1.9295,
      "step": 212428
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.4550182819366455,
      "learning_rate": 8.964567645434329e-06,
      "loss": 1.829,
      "step": 212429
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.4279507398605347,
      "learning_rate": 8.963575166408166e-06,
      "loss": 1.937,
      "step": 212430
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.5905399322509766,
      "learning_rate": 8.962582741491187e-06,
      "loss": 2.0523,
      "step": 212431
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.6104795932769775,
      "learning_rate": 8.961590370683591e-06,
      "loss": 1.9854,
      "step": 212432
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.4296599626541138,
      "learning_rate": 8.960598053985479e-06,
      "loss": 1.9055,
      "step": 212433
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.4481631517410278,
      "learning_rate": 8.959605791397118e-06,
      "loss": 1.7692,
      "step": 212434
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.4611762762069702,
      "learning_rate": 8.958613582918638e-06,
      "loss": 2.0168,
      "step": 212435
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.6284946203231812,
      "learning_rate": 8.957621428550244e-06,
      "loss": 1.7761,
      "step": 212436
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.571960210800171,
      "learning_rate": 8.956629328292165e-06,
      "loss": 1.9708,
      "step": 212437
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.6854711771011353,
      "learning_rate": 8.955637282144534e-06,
      "loss": 1.8893,
      "step": 212438
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.5044875144958496,
      "learning_rate": 8.954645290107487e-06,
      "loss": 1.8185,
      "step": 212439
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.4398893117904663,
      "learning_rate": 8.953653352181322e-06,
      "loss": 1.9904,
      "step": 212440
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.505273461341858,
      "learning_rate": 8.952661468366107e-06,
      "loss": 2.1364,
      "step": 212441
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.5298575162887573,
      "learning_rate": 8.95166963866214e-06,
      "loss": 2.0135,
      "step": 212442
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.444236159324646,
      "learning_rate": 8.950677863069555e-06,
      "loss": 1.8469,
      "step": 212443
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.508442759513855,
      "learning_rate": 8.94968614158852e-06,
      "loss": 1.8418,
      "step": 212444
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.4677777290344238,
      "learning_rate": 8.948694474219198e-06,
      "loss": 2.0126,
      "step": 212445
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.467017412185669,
      "learning_rate": 8.947702860961858e-06,
      "loss": 1.9487,
      "step": 212446
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.4643362760543823,
      "learning_rate": 8.9467113018166e-06,
      "loss": 1.9751,
      "step": 212447
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.7295597791671753,
      "learning_rate": 8.945719796783656e-06,
      "loss": 2.0184,
      "step": 212448
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.6148066520690918,
      "learning_rate": 8.944728345863228e-06,
      "loss": 2.0486,
      "step": 212449
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.4049068689346313,
      "learning_rate": 8.943736949055447e-06,
      "loss": 1.9051,
      "step": 212450
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.4410978555679321,
      "learning_rate": 8.94274560636048e-06,
      "loss": 1.9877,
      "step": 212451
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.5146822929382324,
      "learning_rate": 8.941754317778593e-06,
      "loss": 1.8251,
      "step": 212452
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.6756871938705444,
      "learning_rate": 8.940763083309888e-06,
      "loss": 1.9167,
      "step": 212453
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.3873143196105957,
      "learning_rate": 8.93977190295463e-06,
      "loss": 1.6768,
      "step": 212454
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.5712050199508667,
      "learning_rate": 8.938780776712918e-06,
      "loss": 2.2718,
      "step": 212455
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.4652042388916016,
      "learning_rate": 8.937789704585085e-06,
      "loss": 1.9946,
      "step": 212456
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.4533947706222534,
      "learning_rate": 8.9367986865711e-06,
      "loss": 1.8443,
      "step": 212457
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.382869005203247,
      "learning_rate": 8.935807722671295e-06,
      "loss": 2.0286,
      "step": 212458
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.5862650871276855,
      "learning_rate": 8.934816812885803e-06,
      "loss": 1.9233,
      "step": 212459
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.3838036060333252,
      "learning_rate": 8.933825957214857e-06,
      "loss": 2.2428,
      "step": 212460
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.4591354131698608,
      "learning_rate": 8.932835155658558e-06,
      "loss": 2.2407,
      "step": 212461
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.5109957456588745,
      "learning_rate": 8.931844408217238e-06,
      "loss": 2.1731,
      "step": 212462
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.3722889423370361,
      "learning_rate": 8.930853714890896e-06,
      "loss": 1.6833,
      "step": 212463
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.3129957914352417,
      "learning_rate": 8.929863075679834e-06,
      "loss": 1.9346,
      "step": 212464
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.4193907976150513,
      "learning_rate": 8.92887249058415e-06,
      "loss": 2.01,
      "step": 212465
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.5581306219100952,
      "learning_rate": 8.927881959604144e-06,
      "loss": 1.9727,
      "step": 212466
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.4467557668685913,
      "learning_rate": 8.926891482739884e-06,
      "loss": 2.0413,
      "step": 212467
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.5909141302108765,
      "learning_rate": 8.925901059991703e-06,
      "loss": 2.153,
      "step": 212468
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.5437556505203247,
      "learning_rate": 8.924910691359632e-06,
      "loss": 1.9885,
      "step": 212469
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.5541731119155884,
      "learning_rate": 8.923920376843907e-06,
      "loss": 1.9153,
      "step": 212470
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.4219534397125244,
      "learning_rate": 8.922930116444727e-06,
      "loss": 1.9017,
      "step": 212471
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.4151440858840942,
      "learning_rate": 8.921939910162257e-06,
      "loss": 1.9996,
      "step": 212472
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.4377025365829468,
      "learning_rate": 8.920949757996698e-06,
      "loss": 2.1826,
      "step": 212473
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.502215027809143,
      "learning_rate": 8.919959659948284e-06,
      "loss": 1.9403,
      "step": 212474
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.4719396829605103,
      "learning_rate": 8.91896961601708e-06,
      "loss": 2.1189,
      "step": 212475
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.4615689516067505,
      "learning_rate": 8.917979626203353e-06,
      "loss": 2.0749,
      "step": 212476
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.4862929582595825,
      "learning_rate": 8.91698969050727e-06,
      "loss": 1.9173,
      "step": 212477
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.3699259757995605,
      "learning_rate": 8.915999808928997e-06,
      "loss": 1.9809,
      "step": 212478
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.4861129522323608,
      "learning_rate": 8.915009981468734e-06,
      "loss": 1.99,
      "step": 212479
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.5476374626159668,
      "learning_rate": 8.914020208126715e-06,
      "loss": 1.7744,
      "step": 212480
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.5072367191314697,
      "learning_rate": 8.91303048890304e-06,
      "loss": 2.1016,
      "step": 212481
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.5053365230560303,
      "learning_rate": 8.912040823797939e-06,
      "loss": 2.0597,
      "step": 212482
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.4949976205825806,
      "learning_rate": 8.911051212811549e-06,
      "loss": 1.8536,
      "step": 212483
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.3855400085449219,
      "learning_rate": 8.910061655944101e-06,
      "loss": 2.1486,
      "step": 212484
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.476728916168213,
      "learning_rate": 8.909072153195762e-06,
      "loss": 1.8996,
      "step": 212485
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.4066567420959473,
      "learning_rate": 8.908082704566799e-06,
      "loss": 1.9428,
      "step": 212486
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.384658694267273,
      "learning_rate": 8.907093310057212e-06,
      "loss": 1.9482,
      "step": 212487
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.6190568208694458,
      "learning_rate": 8.906103969667366e-06,
      "loss": 2.0631,
      "step": 212488
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.3451443910598755,
      "learning_rate": 8.905114683397297e-06,
      "loss": 2.0227,
      "step": 212489
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.4913184642791748,
      "learning_rate": 8.904125451247335e-06,
      "loss": 1.7556,
      "step": 212490
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.5150879621505737,
      "learning_rate": 8.903136273217548e-06,
      "loss": 1.9935,
      "step": 212491
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.6324431896209717,
      "learning_rate": 8.902147149308203e-06,
      "loss": 1.9848,
      "step": 212492
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.394803524017334,
      "learning_rate": 8.9011580795194e-06,
      "loss": 1.9281,
      "step": 212493
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.4235506057739258,
      "learning_rate": 8.900169063851404e-06,
      "loss": 1.9359,
      "step": 212494
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.5575069189071655,
      "learning_rate": 8.899180102304316e-06,
      "loss": 1.8835,
      "step": 212495
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.3091951608657837,
      "learning_rate": 8.898191194878401e-06,
      "loss": 1.8804,
      "step": 212496
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.4030662775039673,
      "learning_rate": 8.897202341573762e-06,
      "loss": 1.8057,
      "step": 212497
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.4004403352737427,
      "learning_rate": 8.896213542390662e-06,
      "loss": 2.0301,
      "step": 212498
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.43739652633667,
      "learning_rate": 8.89522479732927e-06,
      "loss": 1.7661,
      "step": 212499
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.6396294832229614,
      "learning_rate": 8.894236106389751e-06,
      "loss": 2.0185,
      "step": 212500
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.564252257347107,
      "learning_rate": 8.89324746957224e-06,
      "loss": 1.9097,
      "step": 212501
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.4813566207885742,
      "learning_rate": 8.892258886877002e-06,
      "loss": 2.0576,
      "step": 212502
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.617397427558899,
      "learning_rate": 8.89127035830417e-06,
      "loss": 2.202,
      "step": 212503
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.3597996234893799,
      "learning_rate": 8.890281883853978e-06,
      "loss": 2.1083,
      "step": 212504
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.5384271144866943,
      "learning_rate": 8.889293463526558e-06,
      "loss": 2.0394,
      "step": 212505
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.4299675226211548,
      "learning_rate": 8.888305097322113e-06,
      "loss": 2.0157,
      "step": 212506
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.4559251070022583,
      "learning_rate": 8.88731678524084e-06,
      "loss": 2.0639,
      "step": 212507
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.4094539880752563,
      "learning_rate": 8.886328527282904e-06,
      "loss": 1.9102,
      "step": 212508
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.5588452816009521,
      "learning_rate": 8.885340323448476e-06,
      "loss": 1.9387,
      "step": 212509
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.3929457664489746,
      "learning_rate": 8.884352173737785e-06,
      "loss": 1.8538,
      "step": 212510
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.453254222869873,
      "learning_rate": 8.883364078151e-06,
      "loss": 1.9633,
      "step": 212511
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.5074152946472168,
      "learning_rate": 8.882376036688222e-06,
      "loss": 2.1179,
      "step": 212512
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.5799829959869385,
      "learning_rate": 8.881388049349781e-06,
      "loss": 1.9995,
      "step": 212513
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.5024652481079102,
      "learning_rate": 8.880400116135778e-06,
      "loss": 1.9169,
      "step": 212514
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.509896993637085,
      "learning_rate": 8.879412237046346e-06,
      "loss": 1.915,
      "step": 212515
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.3509938716888428,
      "learning_rate": 8.878424412081787e-06,
      "loss": 2.0165,
      "step": 212516
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.5141491889953613,
      "learning_rate": 8.877436641242197e-06,
      "loss": 2.0974,
      "step": 212517
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.5001254081726074,
      "learning_rate": 8.87644892452778e-06,
      "loss": 1.7172,
      "step": 212518
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.3088093996047974,
      "learning_rate": 8.875461261938732e-06,
      "loss": 1.9882,
      "step": 212519
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.6498783826828003,
      "learning_rate": 8.874473653475223e-06,
      "loss": 1.9831,
      "step": 212520
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.427479863166809,
      "learning_rate": 8.873486099137484e-06,
      "loss": 1.7966,
      "step": 212521
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.4629234075546265,
      "learning_rate": 8.872498598925648e-06,
      "loss": 2.2989,
      "step": 212522
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.514197587966919,
      "learning_rate": 8.871511152839883e-06,
      "loss": 1.9934,
      "step": 212523
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.4078441858291626,
      "learning_rate": 8.870523760880388e-06,
      "loss": 2.1093,
      "step": 212524
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.4377362728118896,
      "learning_rate": 8.869536423047396e-06,
      "loss": 1.8316,
      "step": 212525
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.366899013519287,
      "learning_rate": 8.868549139341042e-06,
      "loss": 1.7999,
      "step": 212526
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.4136377573013306,
      "learning_rate": 8.867561909761522e-06,
      "loss": 1.7918,
      "step": 212527
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.4423891305923462,
      "learning_rate": 8.866574734308974e-06,
      "loss": 1.93,
      "step": 212528
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.4675034284591675,
      "learning_rate": 8.865587612983727e-06,
      "loss": 2.1233,
      "step": 212529
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.4089895486831665,
      "learning_rate": 8.864600545785783e-06,
      "loss": 1.979,
      "step": 212530
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.4762012958526611,
      "learning_rate": 8.863613532715441e-06,
      "loss": 1.596,
      "step": 212531
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.5509727001190186,
      "learning_rate": 8.862626573772802e-06,
      "loss": 2.0704,
      "step": 212532
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.404853105545044,
      "learning_rate": 8.861639668958131e-06,
      "loss": 2.0734,
      "step": 212533
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.4231033325195312,
      "learning_rate": 8.860652818271562e-06,
      "loss": 1.7728,
      "step": 212534
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.4842990636825562,
      "learning_rate": 8.85966602171333e-06,
      "loss": 2.0296,
      "step": 212535
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.3518216609954834,
      "learning_rate": 8.858679279283532e-06,
      "loss": 1.9551,
      "step": 212536
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.4024919271469116,
      "learning_rate": 8.857692590982434e-06,
      "loss": 1.8437,
      "step": 212537
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.3908984661102295,
      "learning_rate": 8.85670595681014e-06,
      "loss": 1.854,
      "step": 212538
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.431024432182312,
      "learning_rate": 8.855719376766946e-06,
      "loss": 1.936,
      "step": 212539
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.4852821826934814,
      "learning_rate": 8.85473285085292e-06,
      "loss": 1.7752,
      "step": 212540
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.514374852180481,
      "learning_rate": 8.853746379068361e-06,
      "loss": 1.9371,
      "step": 212541
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.5657695531845093,
      "learning_rate": 8.852759961413302e-06,
      "loss": 1.837,
      "step": 212542
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.5298802852630615,
      "learning_rate": 8.851773597888079e-06,
      "loss": 1.9052,
      "step": 212543
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.4646296501159668,
      "learning_rate": 8.850787288492755e-06,
      "loss": 2.0342,
      "step": 212544
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.4720326662063599,
      "learning_rate": 8.849801033227599e-06,
      "loss": 1.9552,
      "step": 212545
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.366157054901123,
      "learning_rate": 8.84881483209271e-06,
      "loss": 2.1328,
      "step": 212546
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.5362507104873657,
      "learning_rate": 8.84782868508842e-06,
      "loss": 2.0208,
      "step": 212547
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.493078351020813,
      "learning_rate": 8.846842592214731e-06,
      "loss": 1.9109,
      "step": 212548
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.4562103748321533,
      "learning_rate": 8.845856553471942e-06,
      "loss": 2.1657,
      "step": 212549
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.3816637992858887,
      "learning_rate": 8.844870568860185e-06,
      "loss": 2.0637,
      "step": 212550
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.4875469207763672,
      "learning_rate": 8.843884638379662e-06,
      "loss": 2.1701,
      "step": 212551
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.3891113996505737,
      "learning_rate": 8.84289876203057e-06,
      "loss": 1.8519,
      "step": 212552
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.4635883569717407,
      "learning_rate": 8.841912939813111e-06,
      "loss": 1.9081,
      "step": 212553
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.532205581665039,
      "learning_rate": 8.840927171727386e-06,
      "loss": 2.0589,
      "step": 212554
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.4993304014205933,
      "learning_rate": 8.839941457773658e-06,
      "loss": 1.9498,
      "step": 212555
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.4486204385757446,
      "learning_rate": 8.838955797952063e-06,
      "loss": 2.1138,
      "step": 212556
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.5421102046966553,
      "learning_rate": 8.8379701922628e-06,
      "loss": 2.0055,
      "step": 212557
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.3183513879776,
      "learning_rate": 8.836984640706035e-06,
      "loss": 1.8153,
      "step": 212558
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.4996172189712524,
      "learning_rate": 8.83599914328207e-06,
      "loss": 2.1666,
      "step": 212559
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.4501928091049194,
      "learning_rate": 8.835013699990867e-06,
      "loss": 1.8697,
      "step": 212560
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.5448206663131714,
      "learning_rate": 8.834028310832798e-06,
      "loss": 1.8178,
      "step": 212561
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.4527454376220703,
      "learning_rate": 8.833042975807957e-06,
      "loss": 1.905,
      "step": 212562
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.4636610746383667,
      "learning_rate": 8.83205769491655e-06,
      "loss": 1.9137,
      "step": 212563
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.3959956169128418,
      "learning_rate": 8.83107246815874e-06,
      "loss": 2.146,
      "step": 212564
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.4311069250106812,
      "learning_rate": 8.830087295534826e-06,
      "loss": 1.8217,
      "step": 212565
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.5016182661056519,
      "learning_rate": 8.829102177044777e-06,
      "loss": 1.9606,
      "step": 212566
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.6023262739181519,
      "learning_rate": 8.82811711268896e-06,
      "loss": 2.1823,
      "step": 212567
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.3596668243408203,
      "learning_rate": 8.827132102467471e-06,
      "loss": 1.7523,
      "step": 212568
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.413758397102356,
      "learning_rate": 8.826147146380514e-06,
      "loss": 1.9256,
      "step": 212569
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.3661682605743408,
      "learning_rate": 8.825162244428253e-06,
      "loss": 1.8545,
      "step": 212570
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.4354679584503174,
      "learning_rate": 8.824177396610955e-06,
      "loss": 1.9752,
      "step": 212571
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.511947751045227,
      "learning_rate": 8.823192602928686e-06,
      "loss": 2.0447,
      "step": 212572
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.4634119272232056,
      "learning_rate": 8.822207863381714e-06,
      "loss": 2.0377,
      "step": 212573
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.4085434675216675,
      "learning_rate": 8.821223177970137e-06,
      "loss": 2.0971,
      "step": 212574
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.4161114692687988,
      "learning_rate": 8.820238546694225e-06,
      "loss": 1.9116,
      "step": 212575
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.527006983757019,
      "learning_rate": 8.819253969554108e-06,
      "loss": 2.2376,
      "step": 212576
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.5055259466171265,
      "learning_rate": 8.818269446550052e-06,
      "loss": 1.8782,
      "step": 212577
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.3363252878189087,
      "learning_rate": 8.817284977682127e-06,
      "loss": 1.9766,
      "step": 212578
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.520473837852478,
      "learning_rate": 8.816300562950562e-06,
      "loss": 1.8111,
      "step": 212579
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.4642698764801025,
      "learning_rate": 8.815316202355526e-06,
      "loss": 2.0215,
      "step": 212580
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.557725429534912,
      "learning_rate": 8.814331895897252e-06,
      "loss": 1.8527,
      "step": 212581
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.3878806829452515,
      "learning_rate": 8.813347643575875e-06,
      "loss": 1.9176,
      "step": 212582
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.5038610696792603,
      "learning_rate": 8.81236344539159e-06,
      "loss": 2.2339,
      "step": 212583
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.3576608896255493,
      "learning_rate": 8.811379301344601e-06,
      "loss": 1.8576,
      "step": 212584
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.5323190689086914,
      "learning_rate": 8.810395211435073e-06,
      "loss": 1.8293,
      "step": 212585
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.3814284801483154,
      "learning_rate": 8.809411175663172e-06,
      "loss": 2.0238,
      "step": 212586
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.5569713115692139,
      "learning_rate": 8.808427194029133e-06,
      "loss": 1.9452,
      "step": 212587
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.4923665523529053,
      "learning_rate": 8.807443266533055e-06,
      "loss": 1.8434,
      "step": 212588
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.3104230165481567,
      "learning_rate": 8.806459393175202e-06,
      "loss": 1.6668,
      "step": 212589
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.311226487159729,
      "learning_rate": 8.80547557395571e-06,
      "loss": 1.9795,
      "step": 212590
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.4793647527694702,
      "learning_rate": 8.804491808874813e-06,
      "loss": 2.0841,
      "step": 212591
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.4742789268493652,
      "learning_rate": 8.803508097932577e-06,
      "loss": 1.8707,
      "step": 212592
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.61426842212677,
      "learning_rate": 8.802524441129333e-06,
      "loss": 2.1294,
      "step": 212593
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.4511380195617676,
      "learning_rate": 8.801540838465182e-06,
      "loss": 1.9566,
      "step": 212594
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.3841832876205444,
      "learning_rate": 8.800557289940324e-06,
      "loss": 2.0134,
      "step": 212595
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.5568082332611084,
      "learning_rate": 8.799573795554925e-06,
      "loss": 1.9402,
      "step": 212596
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.4626195430755615,
      "learning_rate": 8.798590355309188e-06,
      "loss": 2.1464,
      "step": 212597
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.4471677541732788,
      "learning_rate": 8.797606969203308e-06,
      "loss": 1.8949,
      "step": 212598
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.4774225950241089,
      "learning_rate": 8.79662363723742e-06,
      "loss": 2.0599,
      "step": 212599
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.498901605606079,
      "learning_rate": 8.795640359411726e-06,
      "loss": 1.8773,
      "step": 212600
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.5836107730865479,
      "learning_rate": 8.794657135726457e-06,
      "loss": 2.1671,
      "step": 212601
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.4749003648757935,
      "learning_rate": 8.79367396618178e-06,
      "loss": 1.8703,
      "step": 212602
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.5245132446289062,
      "learning_rate": 8.792690850777794e-06,
      "loss": 1.9431,
      "step": 212603
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.4429882764816284,
      "learning_rate": 8.791707789514768e-06,
      "loss": 2.0544,
      "step": 212604
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.6815212965011597,
      "learning_rate": 8.790724782392866e-06,
      "loss": 1.9816,
      "step": 212605
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.4880964756011963,
      "learning_rate": 8.789741829412255e-06,
      "loss": 2.1014,
      "step": 212606
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.4385324716567993,
      "learning_rate": 8.78875893057317e-06,
      "loss": 2.0863,
      "step": 212607
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.5405489206314087,
      "learning_rate": 8.787776085875709e-06,
      "loss": 1.816,
      "step": 212608
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.463781714439392,
      "learning_rate": 8.786793295320105e-06,
      "loss": 2.0506,
      "step": 212609
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.554888129234314,
      "learning_rate": 8.785810558906525e-06,
      "loss": 1.9,
      "step": 212610
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.5565294027328491,
      "learning_rate": 8.78482787663517e-06,
      "loss": 2.0834,
      "step": 212611
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.366524577140808,
      "learning_rate": 8.783845248506204e-06,
      "loss": 2.019,
      "step": 212612
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.4393012523651123,
      "learning_rate": 8.78286267451983e-06,
      "loss": 2.0903,
      "step": 212613
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.4922481775283813,
      "learning_rate": 8.781880154676246e-06,
      "loss": 1.9113,
      "step": 212614
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.5412331819534302,
      "learning_rate": 8.780897688975552e-06,
      "loss": 1.8398,
      "step": 212615
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.718321681022644,
      "learning_rate": 8.779915277418015e-06,
      "loss": 1.9494,
      "step": 212616
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.5351520776748657,
      "learning_rate": 8.778932920003768e-06,
      "loss": 2.0139,
      "step": 212617
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.4086589813232422,
      "learning_rate": 8.777950616733042e-06,
      "loss": 1.78,
      "step": 212618
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.4688665866851807,
      "learning_rate": 8.776968367605975e-06,
      "loss": 1.8285,
      "step": 212619
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.5828100442886353,
      "learning_rate": 8.775986172622796e-06,
      "loss": 2.2428,
      "step": 212620
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.602494478225708,
      "learning_rate": 8.775004031783606e-06,
      "loss": 2.065,
      "step": 212621
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.518440842628479,
      "learning_rate": 8.774021945088705e-06,
      "loss": 1.8803,
      "step": 212622
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.5075191259384155,
      "learning_rate": 8.77303991253816e-06,
      "loss": 2.0647,
      "step": 212623
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.4041473865509033,
      "learning_rate": 8.772057934132204e-06,
      "loss": 1.7842,
      "step": 212624
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.4189544916152954,
      "learning_rate": 8.771076009871036e-06,
      "loss": 2.0132,
      "step": 212625
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.4776759147644043,
      "learning_rate": 8.770094139754857e-06,
      "loss": 1.9885,
      "step": 212626
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.5941081047058105,
      "learning_rate": 8.769112323783767e-06,
      "loss": 2.1366,
      "step": 212627
    },
    {
      "epoch": 2.77,
      "grad_norm": 2.2397122383117676,
      "learning_rate": 8.768130561957997e-06,
      "loss": 1.8519,
      "step": 212628
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.5791572332382202,
      "learning_rate": 8.76714885427775e-06,
      "loss": 1.982,
      "step": 212629
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.7517427206039429,
      "learning_rate": 8.76616720074319e-06,
      "loss": 2.1001,
      "step": 212630
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.3941221237182617,
      "learning_rate": 8.765185601354452e-06,
      "loss": 1.8655,
      "step": 212631
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.5506688356399536,
      "learning_rate": 8.764204056111834e-06,
      "loss": 2.0359,
      "step": 212632
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.389977216720581,
      "learning_rate": 8.763222565015404e-06,
      "loss": 1.7965,
      "step": 212633
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.539406180381775,
      "learning_rate": 8.762241128065394e-06,
      "loss": 2.1275,
      "step": 212634
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.5676757097244263,
      "learning_rate": 8.761259745261972e-06,
      "loss": 2.0691,
      "step": 212635
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.5508556365966797,
      "learning_rate": 8.760278416605338e-06,
      "loss": 2.1452,
      "step": 212636
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.3742564916610718,
      "learning_rate": 8.759297142095656e-06,
      "loss": 2.0944,
      "step": 212637
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.42641019821167,
      "learning_rate": 8.75831592173316e-06,
      "loss": 2.2854,
      "step": 212638
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.546163558959961,
      "learning_rate": 8.757334755517952e-06,
      "loss": 1.9352,
      "step": 212639
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.332180142402649,
      "learning_rate": 8.756353643450265e-06,
      "loss": 1.9171,
      "step": 212640
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.595714807510376,
      "learning_rate": 8.755372585530262e-06,
      "loss": 2.0799,
      "step": 212641
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.374918818473816,
      "learning_rate": 8.754391581758146e-06,
      "loss": 1.9353,
      "step": 212642
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.3619354963302612,
      "learning_rate": 8.75341063213405e-06,
      "loss": 1.8434,
      "step": 212643
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.3953158855438232,
      "learning_rate": 8.752429736658273e-06,
      "loss": 1.951,
      "step": 212644
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.4264088869094849,
      "learning_rate": 8.751448895330847e-06,
      "loss": 2.0181,
      "step": 212645
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.6061697006225586,
      "learning_rate": 8.750468108152043e-06,
      "loss": 1.9238,
      "step": 212646
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.4229059219360352,
      "learning_rate": 8.749487375121989e-06,
      "loss": 2.1587,
      "step": 212647
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.5782147645950317,
      "learning_rate": 8.748506696240953e-06,
      "loss": 1.9485,
      "step": 212648
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.4975478649139404,
      "learning_rate": 8.747526071509037e-06,
      "loss": 2.0379,
      "step": 212649
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.5059454441070557,
      "learning_rate": 8.746545500926504e-06,
      "loss": 2.1783,
      "step": 212650
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.4324783086776733,
      "learning_rate": 8.745564984493424e-06,
      "loss": 1.8549,
      "step": 212651
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.8741165399551392,
      "learning_rate": 8.744584522210097e-06,
      "loss": 2.1351,
      "step": 212652
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.4378540515899658,
      "learning_rate": 8.743604114076586e-06,
      "loss": 1.9541,
      "step": 212653
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.3916329145431519,
      "learning_rate": 8.742623760093192e-06,
      "loss": 1.9706,
      "step": 212654
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.4810818433761597,
      "learning_rate": 8.741643460260018e-06,
      "loss": 2.1362,
      "step": 212655
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.50315523147583,
      "learning_rate": 8.740663214577259e-06,
      "loss": 2.0345,
      "step": 212656
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.453446626663208,
      "learning_rate": 8.739683023045151e-06,
      "loss": 1.976,
      "step": 212657
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.728023886680603,
      "learning_rate": 8.738702885663796e-06,
      "loss": 2.016,
      "step": 212658
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.84532630443573,
      "learning_rate": 8.737722802433423e-06,
      "loss": 1.9486,
      "step": 212659
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.6059634685516357,
      "learning_rate": 8.7367427733542e-06,
      "loss": 1.847,
      "step": 212660
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.4214235544204712,
      "learning_rate": 8.735762798426327e-06,
      "loss": 1.9324,
      "step": 212661
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.3417835235595703,
      "learning_rate": 8.73478287764997e-06,
      "loss": 2.1304,
      "step": 212662
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.4796640872955322,
      "learning_rate": 8.73380301102533e-06,
      "loss": 1.9968,
      "step": 212663
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.463111162185669,
      "learning_rate": 8.732823198552575e-06,
      "loss": 1.9083,
      "step": 212664
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.3426575660705566,
      "learning_rate": 8.731843440231835e-06,
      "loss": 2.0229,
      "step": 212665
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.5078707933425903,
      "learning_rate": 8.730863736063376e-06,
      "loss": 2.1488,
      "step": 212666
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.4106031656265259,
      "learning_rate": 8.729884086047333e-06,
      "loss": 1.9427,
      "step": 212667
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.4214106798171997,
      "learning_rate": 8.728904490183941e-06,
      "loss": 2.1486,
      "step": 212668
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.531411051750183,
      "learning_rate": 8.72792494847333e-06,
      "loss": 2.0118,
      "step": 212669
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.323495864868164,
      "learning_rate": 8.726945460915701e-06,
      "loss": 1.9004,
      "step": 212670
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.3558963537216187,
      "learning_rate": 8.725966027511189e-06,
      "loss": 1.843,
      "step": 212671
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.4351190328598022,
      "learning_rate": 8.724986648260058e-06,
      "loss": 1.9662,
      "step": 212672
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.4607182741165161,
      "learning_rate": 8.724007323162408e-06,
      "loss": 2.1527,
      "step": 212673
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.7332531213760376,
      "learning_rate": 8.723028052218506e-06,
      "loss": 2.023,
      "step": 212674
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.4121346473693848,
      "learning_rate": 8.722048835428486e-06,
      "loss": 1.8372,
      "step": 212675
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.4699680805206299,
      "learning_rate": 8.721069672792546e-06,
      "loss": 1.9376,
      "step": 212676
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.505919098854065,
      "learning_rate": 8.720090564310823e-06,
      "loss": 1.9931,
      "step": 212677
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.392143726348877,
      "learning_rate": 8.719111509983578e-06,
      "loss": 2.291,
      "step": 212678
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.5603528022766113,
      "learning_rate": 8.718132509810882e-06,
      "loss": 2.0654,
      "step": 212679
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.2845579385757446,
      "learning_rate": 8.717153563793034e-06,
      "loss": 1.8953,
      "step": 212680
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.4766993522644043,
      "learning_rate": 8.716174671930165e-06,
      "loss": 2.2035,
      "step": 212681
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.4459319114685059,
      "learning_rate": 8.715195834222443e-06,
      "loss": 2.0572,
      "step": 212682
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.5011463165283203,
      "learning_rate": 8.714217050670068e-06,
      "loss": 1.9207,
      "step": 212683
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.4404962062835693,
      "learning_rate": 8.713238321273241e-06,
      "loss": 1.9677,
      "step": 212684
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.564905047416687,
      "learning_rate": 8.71225964603206e-06,
      "loss": 2.0584,
      "step": 212685
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.6596174240112305,
      "learning_rate": 8.711281024946825e-06,
      "loss": 1.9507,
      "step": 212686
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.3701980113983154,
      "learning_rate": 8.710302458017637e-06,
      "loss": 1.8485,
      "step": 212687
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.4772881269454956,
      "learning_rate": 8.709323945244696e-06,
      "loss": 1.9148,
      "step": 212688
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.3846865892410278,
      "learning_rate": 8.708345486628232e-06,
      "loss": 1.9949,
      "step": 212689
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.4839391708374023,
      "learning_rate": 8.707367082168315e-06,
      "loss": 1.9698,
      "step": 212690
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.4026857614517212,
      "learning_rate": 8.706388731865244e-06,
      "loss": 2.084,
      "step": 212691
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.5751392841339111,
      "learning_rate": 8.705410435719185e-06,
      "loss": 2.0074,
      "step": 212692
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.3169665336608887,
      "learning_rate": 8.704432193730237e-06,
      "loss": 1.9086,
      "step": 212693
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.524726152420044,
      "learning_rate": 8.703454005898636e-06,
      "loss": 1.9595,
      "step": 212694
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.4804694652557373,
      "learning_rate": 8.70247587222458e-06,
      "loss": 1.9566,
      "step": 212695
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.4609849452972412,
      "learning_rate": 8.701497792708235e-06,
      "loss": 1.8732,
      "step": 212696
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.3444323539733887,
      "learning_rate": 8.700519767349767e-06,
      "loss": 2.0416,
      "step": 212697
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.425707459449768,
      "learning_rate": 8.699541796149378e-06,
      "loss": 2.0702,
      "step": 212698
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.4708683490753174,
      "learning_rate": 8.698563879107268e-06,
      "loss": 2.1004,
      "step": 212699
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.4231303930282593,
      "learning_rate": 8.697586016223535e-06,
      "loss": 2.0009,
      "step": 212700
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.4994653463363647,
      "learning_rate": 8.696608207498479e-06,
      "loss": 1.9709,
      "step": 212701
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.4166128635406494,
      "learning_rate": 8.695630452932167e-06,
      "loss": 2.1061,
      "step": 212702
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.4899882078170776,
      "learning_rate": 8.694652752524867e-06,
      "loss": 1.8013,
      "step": 212703
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.655438780784607,
      "learning_rate": 8.69367510627671e-06,
      "loss": 1.9332,
      "step": 212704
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.5546051263809204,
      "learning_rate": 8.692697514187963e-06,
      "loss": 1.9137,
      "step": 212705
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.3632023334503174,
      "learning_rate": 8.691719976258659e-06,
      "loss": 1.951,
      "step": 212706
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.5360873937606812,
      "learning_rate": 8.690742492489133e-06,
      "loss": 2.0927,
      "step": 212707
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.4204472303390503,
      "learning_rate": 8.689765062879417e-06,
      "loss": 1.8623,
      "step": 212708
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.4855704307556152,
      "learning_rate": 8.688787687429843e-06,
      "loss": 2.3437,
      "step": 212709
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.4627766609191895,
      "learning_rate": 8.68781036614048e-06,
      "loss": 2.0617,
      "step": 212710
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.606533169746399,
      "learning_rate": 8.686833099011592e-06,
      "loss": 1.8855,
      "step": 212711
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.5355827808380127,
      "learning_rate": 8.68585588604328e-06,
      "loss": 1.8995,
      "step": 212712
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.4536373615264893,
      "learning_rate": 8.68487872723581e-06,
      "loss": 2.0964,
      "step": 212713
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.865318775177002,
      "learning_rate": 8.683901622589251e-06,
      "loss": 1.828,
      "step": 212714
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.4037926197052002,
      "learning_rate": 8.682924572103934e-06,
      "loss": 2.0757,
      "step": 212715
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.4594937562942505,
      "learning_rate": 8.681947575779891e-06,
      "loss": 2.1225,
      "step": 212716
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.3701058626174927,
      "learning_rate": 8.680970633617457e-06,
      "loss": 1.7995,
      "step": 212717
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.4885376691818237,
      "learning_rate": 8.679993745616631e-06,
      "loss": 2.0152,
      "step": 212718
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.5107420682907104,
      "learning_rate": 8.679016911777747e-06,
      "loss": 2.032,
      "step": 212719
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.6208018064498901,
      "learning_rate": 8.678040132100905e-06,
      "loss": 1.8784,
      "step": 212720
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.518051028251648,
      "learning_rate": 8.677063406586371e-06,
      "loss": 2.1844,
      "step": 212721
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.5109200477600098,
      "learning_rate": 8.676086735234212e-06,
      "loss": 2.0375,
      "step": 212722
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.4791959524154663,
      "learning_rate": 8.675110118044692e-06,
      "loss": 2.0529,
      "step": 212723
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.461352825164795,
      "learning_rate": 8.674133555017982e-06,
      "loss": 1.7261,
      "step": 212724
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.4936699867248535,
      "learning_rate": 8.673157046154243e-06,
      "loss": 1.8432,
      "step": 212725
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.5634199380874634,
      "learning_rate": 8.672180591453615e-06,
      "loss": 2.0194,
      "step": 212726
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.3950012922286987,
      "learning_rate": 8.671204190916392e-06,
      "loss": 2.0564,
      "step": 212727
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.5248944759368896,
      "learning_rate": 8.670227844542643e-06,
      "loss": 1.7182,
      "step": 212728
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.645491123199463,
      "learning_rate": 8.669251552332634e-06,
      "loss": 2.1319,
      "step": 212729
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.5174373388290405,
      "learning_rate": 8.6682753142865e-06,
      "loss": 2.0808,
      "step": 212730
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.480809211730957,
      "learning_rate": 8.667299130404437e-06,
      "loss": 2.0292,
      "step": 212731
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.5248051881790161,
      "learning_rate": 8.666323000686615e-06,
      "loss": 1.7746,
      "step": 212732
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.416124939918518,
      "learning_rate": 8.665346925133231e-06,
      "loss": 1.8647,
      "step": 212733
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.4455275535583496,
      "learning_rate": 8.664370903744421e-06,
      "loss": 2.049,
      "step": 212734
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.4950878620147705,
      "learning_rate": 8.663394936520451e-06,
      "loss": 1.9581,
      "step": 212735
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.3498061895370483,
      "learning_rate": 8.662419023461453e-06,
      "loss": 1.916,
      "step": 212736
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.5286085605621338,
      "learning_rate": 8.661443164567594e-06,
      "loss": 2.0829,
      "step": 212737
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.4512147903442383,
      "learning_rate": 8.660467359839074e-06,
      "loss": 2.0078,
      "step": 212738
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.3892383575439453,
      "learning_rate": 8.659491609276058e-06,
      "loss": 2.0197,
      "step": 212739
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.5531779527664185,
      "learning_rate": 8.65851591287875e-06,
      "loss": 1.8173,
      "step": 212740
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.4647029638290405,
      "learning_rate": 8.657540270647344e-06,
      "loss": 1.9049,
      "step": 212741
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.4125317335128784,
      "learning_rate": 8.65656468258198e-06,
      "loss": 2.0775,
      "step": 212742
    },
    {
      "epoch": 2.77,
      "grad_norm": 2.712045907974243,
      "learning_rate": 8.655589148682884e-06,
      "loss": 1.9011,
      "step": 212743
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.340112328529358,
      "learning_rate": 8.654613668950195e-06,
      "loss": 2.045,
      "step": 212744
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.5494954586029053,
      "learning_rate": 8.65363824338411e-06,
      "loss": 2.0018,
      "step": 212745
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.5349830389022827,
      "learning_rate": 8.652662871984795e-06,
      "loss": 1.9697,
      "step": 212746
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.4244569540023804,
      "learning_rate": 8.651687554752484e-06,
      "loss": 2.1179,
      "step": 212747
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.4187501668930054,
      "learning_rate": 8.650712291687345e-06,
      "loss": 1.9735,
      "step": 212748
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.399835467338562,
      "learning_rate": 8.64973708278951e-06,
      "loss": 2.1083,
      "step": 212749
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.5069035291671753,
      "learning_rate": 8.648761928059144e-06,
      "loss": 2.0544,
      "step": 212750
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.4212238788604736,
      "learning_rate": 8.64778682749655e-06,
      "loss": 2.0162,
      "step": 212751
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.3964356184005737,
      "learning_rate": 8.64681178110176e-06,
      "loss": 1.9765,
      "step": 212752
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.4397352933883667,
      "learning_rate": 8.645836788875072e-06,
      "loss": 2.0598,
      "step": 212753
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.4736639261245728,
      "learning_rate": 8.64486185081662e-06,
      "loss": 2.0995,
      "step": 212754
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.5941424369812012,
      "learning_rate": 8.643886966926606e-06,
      "loss": 2.0553,
      "step": 212755
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.518113374710083,
      "learning_rate": 8.642912137205127e-06,
      "loss": 1.8164,
      "step": 212756
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.4887135028839111,
      "learning_rate": 8.641937361652484e-06,
      "loss": 1.9218,
      "step": 212757
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.4951770305633545,
      "learning_rate": 8.640962640268778e-06,
      "loss": 2.0434,
      "step": 212758
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.4179301261901855,
      "learning_rate": 8.639987973054241e-06,
      "loss": 2.1172,
      "step": 212759
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.4879885911941528,
      "learning_rate": 8.639013360009039e-06,
      "loss": 2.0161,
      "step": 212760
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.5137197971343994,
      "learning_rate": 8.63803880113334e-06,
      "loss": 1.9977,
      "step": 212761
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.5268936157226562,
      "learning_rate": 8.637064296427276e-06,
      "loss": 1.825,
      "step": 212762
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.4670333862304688,
      "learning_rate": 8.636089845891147e-06,
      "loss": 1.9765,
      "step": 212763
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.3064459562301636,
      "learning_rate": 8.63511544952502e-06,
      "loss": 1.9697,
      "step": 212764
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.4806931018829346,
      "learning_rate": 8.634141107329162e-06,
      "loss": 1.8418,
      "step": 212765
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.4325861930847168,
      "learning_rate": 8.633166819303705e-06,
      "loss": 1.9016,
      "step": 212766
    },
    {
      "epoch": 2.77,
      "grad_norm": 2.2415266036987305,
      "learning_rate": 8.632192585448815e-06,
      "loss": 1.834,
      "step": 212767
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.6129367351531982,
      "learning_rate": 8.631218405764729e-06,
      "loss": 2.0819,
      "step": 212768
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.6550471782684326,
      "learning_rate": 8.630244280251607e-06,
      "loss": 1.8457,
      "step": 212769
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.4002914428710938,
      "learning_rate": 8.629270208909589e-06,
      "loss": 1.8363,
      "step": 212770
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.4258556365966797,
      "learning_rate": 8.628296191738937e-06,
      "loss": 2.023,
      "step": 212771
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.7229071855545044,
      "learning_rate": 8.627322228739753e-06,
      "loss": 2.0175,
      "step": 212772
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.4162025451660156,
      "learning_rate": 8.626348319912235e-06,
      "loss": 1.8826,
      "step": 212773
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.3715647459030151,
      "learning_rate": 8.625374465256618e-06,
      "loss": 2.1113,
      "step": 212774
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.522844910621643,
      "learning_rate": 8.624400664773034e-06,
      "loss": 2.0852,
      "step": 212775
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.4380987882614136,
      "learning_rate": 8.62342691846165e-06,
      "loss": 1.8431,
      "step": 212776
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.5504209995269775,
      "learning_rate": 8.622453226322701e-06,
      "loss": 1.96,
      "step": 212777
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.5757768154144287,
      "learning_rate": 8.62147958835635e-06,
      "loss": 2.1777,
      "step": 212778
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.363568902015686,
      "learning_rate": 8.6205060045627e-06,
      "loss": 1.9338,
      "step": 212779
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.5358840227127075,
      "learning_rate": 8.619532474942048e-06,
      "loss": 1.9437,
      "step": 212780
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.461933970451355,
      "learning_rate": 8.618558999494496e-06,
      "loss": 2.0616,
      "step": 212781
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.440475583076477,
      "learning_rate": 8.61758557822031e-06,
      "loss": 1.7824,
      "step": 212782
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.4812995195388794,
      "learning_rate": 8.616612211119589e-06,
      "loss": 1.8988,
      "step": 212783
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.5101577043533325,
      "learning_rate": 8.615638898192534e-06,
      "loss": 1.7441,
      "step": 212784
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.520027995109558,
      "learning_rate": 8.61466563943931e-06,
      "loss": 2.0632,
      "step": 212785
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.9899792671203613,
      "learning_rate": 8.613692434860154e-06,
      "loss": 2.0772,
      "step": 212786
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.4034830331802368,
      "learning_rate": 8.612719284455161e-06,
      "loss": 1.8184,
      "step": 212787
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.4341075420379639,
      "learning_rate": 8.611746188224633e-06,
      "loss": 1.9856,
      "step": 212788
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.4088090658187866,
      "learning_rate": 8.610773146168603e-06,
      "loss": 1.9524,
      "step": 212789
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.511169195175171,
      "learning_rate": 8.609800158287405e-06,
      "loss": 1.9593,
      "step": 212790
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.6446692943572998,
      "learning_rate": 8.608827224581138e-06,
      "loss": 1.8871,
      "step": 212791
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.4873507022857666,
      "learning_rate": 8.607854345049969e-06,
      "loss": 2.1578,
      "step": 212792
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.5138823986053467,
      "learning_rate": 8.606881519694065e-06,
      "loss": 1.7643,
      "step": 212793
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.4103018045425415,
      "learning_rate": 8.60590874851369e-06,
      "loss": 1.8354,
      "step": 212794
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.5454638004302979,
      "learning_rate": 8.604936031508946e-06,
      "loss": 2.1018,
      "step": 212795
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.4261081218719482,
      "learning_rate": 8.6039633686801e-06,
      "loss": 1.8033,
      "step": 212796
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.5487333536148071,
      "learning_rate": 8.60299076002725e-06,
      "loss": 1.8581,
      "step": 212797
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.478980541229248,
      "learning_rate": 8.602018205550598e-06,
      "loss": 1.9042,
      "step": 212798
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.383644700050354,
      "learning_rate": 8.601045705250309e-06,
      "loss": 1.9019,
      "step": 212799
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.4890027046203613,
      "learning_rate": 8.600073259126617e-06,
      "loss": 2.1009,
      "step": 212800
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.4575985670089722,
      "learning_rate": 8.599100867179653e-06,
      "loss": 1.7877,
      "step": 212801
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.375625491142273,
      "learning_rate": 8.59812852940962e-06,
      "loss": 1.9769,
      "step": 212802
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.5894241333007812,
      "learning_rate": 8.597156245816716e-06,
      "loss": 1.9991,
      "step": 212803
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.447299599647522,
      "learning_rate": 8.59618401640111e-06,
      "loss": 2.012,
      "step": 212804
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.584537148475647,
      "learning_rate": 8.59521184116293e-06,
      "loss": 2.0709,
      "step": 212805
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.3525185585021973,
      "learning_rate": 8.594239720102414e-06,
      "loss": 1.9598,
      "step": 212806
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.3994181156158447,
      "learning_rate": 8.593267653219726e-06,
      "loss": 2.0084,
      "step": 212807
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.3402372598648071,
      "learning_rate": 8.592295640515068e-06,
      "loss": 1.8599,
      "step": 212808
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.6320881843566895,
      "learning_rate": 8.591323681988604e-06,
      "loss": 2.0088,
      "step": 212809
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.4452427625656128,
      "learning_rate": 8.590351777640536e-06,
      "loss": 1.9037,
      "step": 212810
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.4273767471313477,
      "learning_rate": 8.589379927470962e-06,
      "loss": 2.0979,
      "step": 212811
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.4519288539886475,
      "learning_rate": 8.588408131480151e-06,
      "loss": 1.9886,
      "step": 212812
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.490187644958496,
      "learning_rate": 8.587436389668234e-06,
      "loss": 2.0088,
      "step": 212813
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.5804625749588013,
      "learning_rate": 8.586464702035479e-06,
      "loss": 1.8386,
      "step": 212814
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.544023036956787,
      "learning_rate": 8.58549306858195e-06,
      "loss": 1.8836,
      "step": 212815
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.5054736137390137,
      "learning_rate": 8.584521489307916e-06,
      "loss": 1.8687,
      "step": 212816
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.5419402122497559,
      "learning_rate": 8.583549964213476e-06,
      "loss": 1.9659,
      "step": 212817
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.668905258178711,
      "learning_rate": 8.582578493298864e-06,
      "loss": 1.83,
      "step": 212818
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.6519808769226074,
      "learning_rate": 8.581607076564245e-06,
      "loss": 1.9105,
      "step": 212819
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.36968195438385,
      "learning_rate": 8.58063571400982e-06,
      "loss": 1.8041,
      "step": 212820
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.3901870250701904,
      "learning_rate": 8.57966440563579e-06,
      "loss": 1.9672,
      "step": 212821
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.6132335662841797,
      "learning_rate": 8.578693151442251e-06,
      "loss": 2.2631,
      "step": 212822
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.4426649808883667,
      "learning_rate": 8.57772195142944e-06,
      "loss": 2.1788,
      "step": 212823
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.3433102369308472,
      "learning_rate": 8.576750805597555e-06,
      "loss": 1.9357,
      "step": 212824
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.3959412574768066,
      "learning_rate": 8.575779713946728e-06,
      "loss": 2.2722,
      "step": 212825
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.4888333082199097,
      "learning_rate": 8.574808676477197e-06,
      "loss": 1.8256,
      "step": 212826
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.3583015203475952,
      "learning_rate": 8.57383769318909e-06,
      "loss": 1.9297,
      "step": 212827
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.4233620166778564,
      "learning_rate": 8.572866764082608e-06,
      "loss": 2.0338,
      "step": 212828
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.4630255699157715,
      "learning_rate": 8.57189588915792e-06,
      "loss": 1.9715,
      "step": 212829
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.520374059677124,
      "learning_rate": 8.570925068415223e-06,
      "loss": 1.8101,
      "step": 212830
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.4698516130447388,
      "learning_rate": 8.569954301854687e-06,
      "loss": 2.0937,
      "step": 212831
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.4241909980773926,
      "learning_rate": 8.568983589476509e-06,
      "loss": 1.8583,
      "step": 212832
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.467939853668213,
      "learning_rate": 8.568012931280888e-06,
      "loss": 2.0654,
      "step": 212833
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.399390697479248,
      "learning_rate": 8.567042327267958e-06,
      "loss": 1.8313,
      "step": 212834
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.4407285451889038,
      "learning_rate": 8.566071777437888e-06,
      "loss": 2.0251,
      "step": 212835
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.4636129140853882,
      "learning_rate": 8.565101281790909e-06,
      "loss": 1.9515,
      "step": 212836
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.4319896697998047,
      "learning_rate": 8.564130840327122e-06,
      "loss": 2.0688,
      "step": 212837
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.7604272365570068,
      "learning_rate": 8.563160453046824e-06,
      "loss": 2.0018,
      "step": 212838
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.5536696910858154,
      "learning_rate": 8.562190119950152e-06,
      "loss": 2.0188,
      "step": 212839
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.6217840909957886,
      "learning_rate": 8.561219841037237e-06,
      "loss": 2.0113,
      "step": 212840
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.3937371969223022,
      "learning_rate": 8.56024961630828e-06,
      "loss": 1.9505,
      "step": 212841
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.3372130393981934,
      "learning_rate": 8.559279445763511e-06,
      "loss": 1.7532,
      "step": 212842
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.5115996599197388,
      "learning_rate": 8.558309329403035e-06,
      "loss": 1.8921,
      "step": 212843
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.4168027639389038,
      "learning_rate": 8.557339267227115e-06,
      "loss": 1.8433,
      "step": 212844
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.6489697694778442,
      "learning_rate": 8.556369259235852e-06,
      "loss": 1.7949,
      "step": 212845
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.4232022762298584,
      "learning_rate": 8.555399305429512e-06,
      "loss": 1.8308,
      "step": 212846
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.5279572010040283,
      "learning_rate": 8.55442940580816e-06,
      "loss": 1.8872,
      "step": 212847
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.4723767042160034,
      "learning_rate": 8.553459560372066e-06,
      "loss": 1.9059,
      "step": 212848
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.5112332105636597,
      "learning_rate": 8.552489769121362e-06,
      "loss": 2.0304,
      "step": 212849
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.401291847229004,
      "learning_rate": 8.551520032056281e-06,
      "loss": 1.8199,
      "step": 212850
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.4012541770935059,
      "learning_rate": 8.550550349176989e-06,
      "loss": 2.0538,
      "step": 212851
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.4529842138290405,
      "learning_rate": 8.549580720483618e-06,
      "loss": 1.9737,
      "step": 212852
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.3560079336166382,
      "learning_rate": 8.548611145976403e-06,
      "loss": 1.9584,
      "step": 212853
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.4094511270523071,
      "learning_rate": 8.547641625655477e-06,
      "loss": 2.0673,
      "step": 212854
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.4390140771865845,
      "learning_rate": 8.54667215952104e-06,
      "loss": 2.1063,
      "step": 212855
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.5193043947219849,
      "learning_rate": 8.545702747573325e-06,
      "loss": 1.8816,
      "step": 212856
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.4064579010009766,
      "learning_rate": 8.544733389812431e-06,
      "loss": 1.9358,
      "step": 212857
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.3683476448059082,
      "learning_rate": 8.543764086238592e-06,
      "loss": 1.7929,
      "step": 212858
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.4813389778137207,
      "learning_rate": 8.542794836851975e-06,
      "loss": 1.9711,
      "step": 212859
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.41325843334198,
      "learning_rate": 8.541825641652745e-06,
      "loss": 1.8894,
      "step": 212860
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.4363495111465454,
      "learning_rate": 8.54085650064107e-06,
      "loss": 2.022,
      "step": 212861
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.363295555114746,
      "learning_rate": 8.539887413817147e-06,
      "loss": 2.0436,
      "step": 212862
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.5611697435379028,
      "learning_rate": 8.53891838118118e-06,
      "loss": 2.2625,
      "step": 212863
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.4289036989212036,
      "learning_rate": 8.537949402733335e-06,
      "loss": 2.0874,
      "step": 212864
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.495267391204834,
      "learning_rate": 8.53698047847381e-06,
      "loss": 1.8405,
      "step": 212865
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.4833166599273682,
      "learning_rate": 8.536011608402705e-06,
      "loss": 2.3475,
      "step": 212866
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.4768941402435303,
      "learning_rate": 8.535042792520286e-06,
      "loss": 2.1369,
      "step": 212867
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.3669897317886353,
      "learning_rate": 8.534074030826721e-06,
      "loss": 1.92,
      "step": 212868
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.3271394968032837,
      "learning_rate": 8.533105323322142e-06,
      "loss": 2.1464,
      "step": 212869
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.3237954378128052,
      "learning_rate": 8.532136670006784e-06,
      "loss": 1.8621,
      "step": 212870
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.5340009927749634,
      "learning_rate": 8.531168070880811e-06,
      "loss": 2.0076,
      "step": 212871
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.4572793245315552,
      "learning_rate": 8.530199525944359e-06,
      "loss": 1.8599,
      "step": 212872
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.3836157321929932,
      "learning_rate": 8.529231035197659e-06,
      "loss": 1.8774,
      "step": 212873
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.542707920074463,
      "learning_rate": 8.528262598640912e-06,
      "loss": 1.9462,
      "step": 212874
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.4765706062316895,
      "learning_rate": 8.527294216274216e-06,
      "loss": 1.9859,
      "step": 212875
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.4001163244247437,
      "learning_rate": 8.526325888097808e-06,
      "loss": 1.9678,
      "step": 212876
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.4026843309402466,
      "learning_rate": 8.525357614111882e-06,
      "loss": 1.8881,
      "step": 212877
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.3821709156036377,
      "learning_rate": 8.524389394316577e-06,
      "loss": 1.9024,
      "step": 212878
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.404952883720398,
      "learning_rate": 8.52342122871209e-06,
      "loss": 2.0365,
      "step": 212879
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.4726712703704834,
      "learning_rate": 8.52245311729859e-06,
      "loss": 2.0512,
      "step": 212880
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.4925955533981323,
      "learning_rate": 8.521485060076271e-06,
      "loss": 1.9452,
      "step": 212881
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.522477388381958,
      "learning_rate": 8.52051705704534e-06,
      "loss": 1.8138,
      "step": 212882
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.4787347316741943,
      "learning_rate": 8.519549108205926e-06,
      "loss": 2.0067,
      "step": 212883
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.4363621473312378,
      "learning_rate": 8.518581213558229e-06,
      "loss": 2.0074,
      "step": 212884
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.4922069311141968,
      "learning_rate": 8.51761337310245e-06,
      "loss": 1.9484,
      "step": 212885
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.4432642459869385,
      "learning_rate": 8.516645586838689e-06,
      "loss": 1.881,
      "step": 212886
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.5304192304611206,
      "learning_rate": 8.515677854767245e-06,
      "loss": 2.0813,
      "step": 212887
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.3896872997283936,
      "learning_rate": 8.514710176888217e-06,
      "loss": 1.9842,
      "step": 212888
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.473638892173767,
      "learning_rate": 8.51374255320184e-06,
      "loss": 1.9394,
      "step": 212889
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.6475377082824707,
      "learning_rate": 8.512774983708215e-06,
      "loss": 1.9631,
      "step": 212890
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.4253208637237549,
      "learning_rate": 8.511807468407573e-06,
      "loss": 1.8701,
      "step": 212891
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.4127202033996582,
      "learning_rate": 8.51084000730008e-06,
      "loss": 2.024,
      "step": 212892
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.5313386917114258,
      "learning_rate": 8.50987260038597e-06,
      "loss": 2.0139,
      "step": 212893
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.5710374116897583,
      "learning_rate": 8.508905247665344e-06,
      "loss": 1.9935,
      "step": 212894
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.4724222421646118,
      "learning_rate": 8.507937949138432e-06,
      "loss": 1.8913,
      "step": 212895
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.4009565114974976,
      "learning_rate": 8.506970704805371e-06,
      "loss": 1.9646,
      "step": 212896
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.642260193824768,
      "learning_rate": 8.506003514666393e-06,
      "loss": 2.057,
      "step": 212897
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.5053781270980835,
      "learning_rate": 8.50503637872163e-06,
      "loss": 2.038,
      "step": 212898
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.628771185874939,
      "learning_rate": 8.504069296971283e-06,
      "loss": 1.9046,
      "step": 212899
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.560482144355774,
      "learning_rate": 8.503102269415551e-06,
      "loss": 1.9963,
      "step": 212900
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.5662600994110107,
      "learning_rate": 8.502135296054602e-06,
      "loss": 2.1908,
      "step": 212901
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.5060737133026123,
      "learning_rate": 8.501168376888567e-06,
      "loss": 2.0931,
      "step": 212902
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.4774819612503052,
      "learning_rate": 8.500201511917714e-06,
      "loss": 1.8777,
      "step": 212903
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.4715999364852905,
      "learning_rate": 8.499234701142143e-06,
      "loss": 1.9572,
      "step": 212904
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.366672396659851,
      "learning_rate": 8.498267944562087e-06,
      "loss": 2.1272,
      "step": 212905
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.4882806539535522,
      "learning_rate": 8.497301242177713e-06,
      "loss": 1.9104,
      "step": 212906
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.5251208543777466,
      "learning_rate": 8.496334593989184e-06,
      "loss": 1.9489,
      "step": 212907
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.4234745502471924,
      "learning_rate": 8.495367999996638e-06,
      "loss": 2.2331,
      "step": 212908
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.757419228553772,
      "learning_rate": 8.494401460200373e-06,
      "loss": 2.218,
      "step": 212909
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.5744017362594604,
      "learning_rate": 8.493434974600455e-06,
      "loss": 1.971,
      "step": 212910
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.4909003973007202,
      "learning_rate": 8.49246854319715e-06,
      "loss": 1.8793,
      "step": 212911
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.3351765871047974,
      "learning_rate": 8.491502165990594e-06,
      "loss": 2.005,
      "step": 212912
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.3993611335754395,
      "learning_rate": 8.49053584298095e-06,
      "loss": 2.0463,
      "step": 212913
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.5510603189468384,
      "learning_rate": 8.48956957416842e-06,
      "loss": 1.99,
      "step": 212914
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.3805992603302002,
      "learning_rate": 8.488603359553203e-06,
      "loss": 2.099,
      "step": 212915
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.544966459274292,
      "learning_rate": 8.4876371991354e-06,
      "loss": 1.8874,
      "step": 212916
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.4338096380233765,
      "learning_rate": 8.48667109291531e-06,
      "loss": 2.0523,
      "step": 212917
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.5029497146606445,
      "learning_rate": 8.485705040893032e-06,
      "loss": 1.9478,
      "step": 212918
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.4951796531677246,
      "learning_rate": 8.484739043068766e-06,
      "loss": 2.0324,
      "step": 212919
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.6876133680343628,
      "learning_rate": 8.48377309944268e-06,
      "loss": 1.8655,
      "step": 212920
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.3902883529663086,
      "learning_rate": 8.482807210014974e-06,
      "loss": 1.8563,
      "step": 212921
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.4702880382537842,
      "learning_rate": 8.48184137478578e-06,
      "loss": 2.0363,
      "step": 212922
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.4593418836593628,
      "learning_rate": 8.480875593755365e-06,
      "loss": 1.9256,
      "step": 212923
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.3005825281143188,
      "learning_rate": 8.479909866923828e-06,
      "loss": 1.7604,
      "step": 212924
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.5010404586791992,
      "learning_rate": 8.478944194291436e-06,
      "loss": 1.982,
      "step": 212925
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.3780553340911865,
      "learning_rate": 8.477978575858225e-06,
      "loss": 1.6913,
      "step": 212926
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.6382941007614136,
      "learning_rate": 8.477013011624524e-06,
      "loss": 1.9342,
      "step": 212927
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.4804681539535522,
      "learning_rate": 8.476047501590399e-06,
      "loss": 1.8077,
      "step": 212928
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.3887678384780884,
      "learning_rate": 8.47508204575612e-06,
      "loss": 1.9969,
      "step": 212929
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.7049295902252197,
      "learning_rate": 8.474116644121786e-06,
      "loss": 2.1048,
      "step": 212930
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.540981411933899,
      "learning_rate": 8.473151296687698e-06,
      "loss": 1.8964,
      "step": 212931
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.5238573551177979,
      "learning_rate": 8.472186003453851e-06,
      "loss": 2.1439,
      "step": 212932
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.3979548215866089,
      "learning_rate": 8.471220764420617e-06,
      "loss": 2.0429,
      "step": 212933
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.554918646812439,
      "learning_rate": 8.470255579588026e-06,
      "loss": 1.9895,
      "step": 212934
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.4714174270629883,
      "learning_rate": 8.469290448956346e-06,
      "loss": 2.0075,
      "step": 212935
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.5702588558197021,
      "learning_rate": 8.468325372525675e-06,
      "loss": 2.242,
      "step": 212936
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.422413945198059,
      "learning_rate": 8.467360350296349e-06,
      "loss": 2.0116,
      "step": 212937
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.438066840171814,
      "learning_rate": 8.466395382268365e-06,
      "loss": 1.9641,
      "step": 212938
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.5855772495269775,
      "learning_rate": 8.465430468441992e-06,
      "loss": 1.855,
      "step": 212939
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.5244120359420776,
      "learning_rate": 8.464465608817395e-06,
      "loss": 1.861,
      "step": 212940
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.4882051944732666,
      "learning_rate": 8.463500803394773e-06,
      "loss": 2.0364,
      "step": 212941
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.5279988050460815,
      "learning_rate": 8.462536052174262e-06,
      "loss": 2.0823,
      "step": 212942
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.5036581754684448,
      "learning_rate": 8.461571355156095e-06,
      "loss": 2.1217,
      "step": 212943
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.3963267803192139,
      "learning_rate": 8.460606712340434e-06,
      "loss": 1.9275,
      "step": 212944
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.489454984664917,
      "learning_rate": 8.45964212372745e-06,
      "loss": 1.9983,
      "step": 212945
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.4243890047073364,
      "learning_rate": 8.45867758931731e-06,
      "loss": 1.9622,
      "step": 212946
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.5800447463989258,
      "learning_rate": 8.45771310911021e-06,
      "loss": 1.9161,
      "step": 212947
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.5171573162078857,
      "learning_rate": 8.456748683106284e-06,
      "loss": 1.8694,
      "step": 212948
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.3023078441619873,
      "learning_rate": 8.455784311305802e-06,
      "loss": 1.8617,
      "step": 212949
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.5234863758087158,
      "learning_rate": 8.454819993708895e-06,
      "loss": 1.9216,
      "step": 212950
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.567889928817749,
      "learning_rate": 8.453855730315696e-06,
      "loss": 2.1872,
      "step": 212951
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.8217514753341675,
      "learning_rate": 8.45289152112647e-06,
      "loss": 2.0258,
      "step": 212952
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.5110297203063965,
      "learning_rate": 8.451927366141354e-06,
      "loss": 2.0597,
      "step": 212953
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.447470784187317,
      "learning_rate": 8.450963265360477e-06,
      "loss": 2.1439,
      "step": 212954
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.3070242404937744,
      "learning_rate": 8.449999218784142e-06,
      "loss": 1.9825,
      "step": 212955
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.4644296169281006,
      "learning_rate": 8.449035226412416e-06,
      "loss": 2.1242,
      "step": 212956
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.5161385536193848,
      "learning_rate": 8.448071288245495e-06,
      "loss": 2.0118,
      "step": 212957
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.4923334121704102,
      "learning_rate": 8.447107404283615e-06,
      "loss": 2.0197,
      "step": 212958
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.48637056350708,
      "learning_rate": 8.446143574526942e-06,
      "loss": 1.8181,
      "step": 212959
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.4454411268234253,
      "learning_rate": 8.445179798975577e-06,
      "loss": 2.0717,
      "step": 212960
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.4414722919464111,
      "learning_rate": 8.444216077629817e-06,
      "loss": 1.9447,
      "step": 212961
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.4674947261810303,
      "learning_rate": 8.443252410489732e-06,
      "loss": 1.8045,
      "step": 212962
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.4426041841506958,
      "learning_rate": 8.442288797555552e-06,
      "loss": 2.0423,
      "step": 212963
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.5191065073013306,
      "learning_rate": 8.441325238827479e-06,
      "loss": 1.7942,
      "step": 212964
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.5181097984313965,
      "learning_rate": 8.440361734305645e-06,
      "loss": 1.8616,
      "step": 212965
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.473430871963501,
      "learning_rate": 8.439398283990285e-06,
      "loss": 1.8838,
      "step": 212966
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.394890308380127,
      "learning_rate": 8.43843488788153e-06,
      "loss": 1.9303,
      "step": 212967
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.627983570098877,
      "learning_rate": 8.43747154597958e-06,
      "loss": 2.1794,
      "step": 212968
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.4767967462539673,
      "learning_rate": 8.436508258284569e-06,
      "loss": 1.9903,
      "step": 212969
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.4927306175231934,
      "learning_rate": 8.435545024796763e-06,
      "loss": 1.8959,
      "step": 212970
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.5613197088241577,
      "learning_rate": 8.43458184551623e-06,
      "loss": 2.0673,
      "step": 212971
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.4317858219146729,
      "learning_rate": 8.43361872044327e-06,
      "loss": 1.9903,
      "step": 212972
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.5600590705871582,
      "learning_rate": 8.432655649578014e-06,
      "loss": 1.9946,
      "step": 212973
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.5578867197036743,
      "learning_rate": 8.431692632920595e-06,
      "loss": 1.9534,
      "step": 212974
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.484131932258606,
      "learning_rate": 8.430729670471214e-06,
      "loss": 1.77,
      "step": 212975
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.383985161781311,
      "learning_rate": 8.429766762230106e-06,
      "loss": 2.1343,
      "step": 212976
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.3857556581497192,
      "learning_rate": 8.428803908197369e-06,
      "loss": 2.098,
      "step": 212977
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.4770171642303467,
      "learning_rate": 8.427841108373268e-06,
      "loss": 1.9524,
      "step": 212978
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.4417918920516968,
      "learning_rate": 8.426878362757905e-06,
      "loss": 1.7384,
      "step": 212979
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.4235094785690308,
      "learning_rate": 8.425915671351513e-06,
      "loss": 1.9167,
      "step": 212980
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.6332200765609741,
      "learning_rate": 8.42495303415419e-06,
      "loss": 1.9739,
      "step": 212981
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.5217067003250122,
      "learning_rate": 8.42399045116624e-06,
      "loss": 1.9254,
      "step": 212982
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.5171984434127808,
      "learning_rate": 8.423027922387727e-06,
      "loss": 2.1378,
      "step": 212983
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.3883601427078247,
      "learning_rate": 8.422065447818882e-06,
      "loss": 1.7576,
      "step": 212984
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.5748502016067505,
      "learning_rate": 8.421103027459908e-06,
      "loss": 1.9602,
      "step": 212985
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.530290961265564,
      "learning_rate": 8.420140661310937e-06,
      "loss": 1.8993,
      "step": 212986
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.4400758743286133,
      "learning_rate": 8.419178349372168e-06,
      "loss": 2.105,
      "step": 212987
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.487228274345398,
      "learning_rate": 8.41821609164377e-06,
      "loss": 2.0684,
      "step": 212988
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.4106379747390747,
      "learning_rate": 8.417253888125908e-06,
      "loss": 1.9034,
      "step": 212989
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.4675601720809937,
      "learning_rate": 8.416291738818848e-06,
      "loss": 1.9529,
      "step": 212990
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.5184165239334106,
      "learning_rate": 8.415329643722624e-06,
      "loss": 2.0383,
      "step": 212991
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.4966645240783691,
      "learning_rate": 8.414367602837602e-06,
      "loss": 2.0784,
      "step": 212992
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.5248278379440308,
      "learning_rate": 8.413405616163748e-06,
      "loss": 1.9094,
      "step": 212993
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.558988332748413,
      "learning_rate": 8.412443683701397e-06,
      "loss": 1.9851,
      "step": 212994
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.4602043628692627,
      "learning_rate": 8.411481805450648e-06,
      "loss": 1.946,
      "step": 212995
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.536440372467041,
      "learning_rate": 8.410519981411735e-06,
      "loss": 1.7408,
      "step": 212996
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.4001129865646362,
      "learning_rate": 8.409558211584788e-06,
      "loss": 2.1295,
      "step": 212997
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.5581625699996948,
      "learning_rate": 8.408596495970044e-06,
      "loss": 2.2159,
      "step": 212998
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.4972373247146606,
      "learning_rate": 8.4076348345676e-06,
      "loss": 2.0742,
      "step": 212999
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.4378340244293213,
      "learning_rate": 8.406673227377724e-06,
      "loss": 1.9774,
      "step": 213000
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.443227767944336,
      "learning_rate": 8.405711674400517e-06,
      "loss": 2.2378,
      "step": 213001
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.5530633926391602,
      "learning_rate": 8.40475017563621e-06,
      "loss": 1.8115,
      "step": 213002
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.5798969268798828,
      "learning_rate": 8.403788731084938e-06,
      "loss": 1.7934,
      "step": 213003
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.553035855293274,
      "learning_rate": 8.402827340746998e-06,
      "loss": 2.0984,
      "step": 213004
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.4048428535461426,
      "learning_rate": 8.401866004622393e-06,
      "loss": 2.0327,
      "step": 213005
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.4751371145248413,
      "learning_rate": 8.400904722711388e-06,
      "loss": 2.1873,
      "step": 213006
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.5302603244781494,
      "learning_rate": 8.39994349501415e-06,
      "loss": 1.9548,
      "step": 213007
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.3905280828475952,
      "learning_rate": 8.398982321530911e-06,
      "loss": 1.8203,
      "step": 213008
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.4316422939300537,
      "learning_rate": 8.398021202261773e-06,
      "loss": 1.8526,
      "step": 213009
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.4237573146820068,
      "learning_rate": 8.397060137207001e-06,
      "loss": 1.8769,
      "step": 213010
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.6522107124328613,
      "learning_rate": 8.39609912636663e-06,
      "loss": 1.9319,
      "step": 213011
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.3608125448226929,
      "learning_rate": 8.395138169740989e-06,
      "loss": 2.0255,
      "step": 213012
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.4797749519348145,
      "learning_rate": 8.394177267330182e-06,
      "loss": 1.9164,
      "step": 213013
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.4116109609603882,
      "learning_rate": 8.393216419134408e-06,
      "loss": 1.7954,
      "step": 213014
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.4101810455322266,
      "learning_rate": 8.392255625153798e-06,
      "loss": 1.7901,
      "step": 213015
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.3435653448104858,
      "learning_rate": 8.391294885388655e-06,
      "loss": 1.7404,
      "step": 213016
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.4629815816879272,
      "learning_rate": 8.39033419983901e-06,
      "loss": 1.8182,
      "step": 213017
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.5987372398376465,
      "learning_rate": 8.389373568505132e-06,
      "loss": 2.2157,
      "step": 213018
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.3645187616348267,
      "learning_rate": 8.38841299138715e-06,
      "loss": 1.8244,
      "step": 213019
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.8601940870285034,
      "learning_rate": 8.387452468485301e-06,
      "loss": 2.0574,
      "step": 213020
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.4882103204727173,
      "learning_rate": 8.386491999799683e-06,
      "loss": 1.9918,
      "step": 213021
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.4203085899353027,
      "learning_rate": 8.385531585330596e-06,
      "loss": 1.8987,
      "step": 213022
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.5101929903030396,
      "learning_rate": 8.384571225078074e-06,
      "loss": 2.1116,
      "step": 213023
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.6706093549728394,
      "learning_rate": 8.383610919042416e-06,
      "loss": 2.016,
      "step": 213024
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.538062334060669,
      "learning_rate": 8.382650667223689e-06,
      "loss": 2.0504,
      "step": 213025
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.4152777194976807,
      "learning_rate": 8.38169046962216e-06,
      "loss": 1.7921,
      "step": 213026
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.4664814472198486,
      "learning_rate": 8.380730326237994e-06,
      "loss": 1.9147,
      "step": 213027
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.406922698020935,
      "learning_rate": 8.379770237071325e-06,
      "loss": 1.7291,
      "step": 213028
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.4097421169281006,
      "learning_rate": 8.378810202122421e-06,
      "loss": 1.7932,
      "step": 213029
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.486208438873291,
      "learning_rate": 8.377850221391347e-06,
      "loss": 2.0143,
      "step": 213030
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.5281386375427246,
      "learning_rate": 8.376890294878336e-06,
      "loss": 1.95,
      "step": 213031
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.5160582065582275,
      "learning_rate": 8.375930422583588e-06,
      "loss": 2.2182,
      "step": 213032
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.6221061944961548,
      "learning_rate": 8.374970604507236e-06,
      "loss": 1.8349,
      "step": 213033
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.506841778755188,
      "learning_rate": 8.374010840649514e-06,
      "loss": 2.0106,
      "step": 213034
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.4966233968734741,
      "learning_rate": 8.373051131010556e-06,
      "loss": 2.0458,
      "step": 213035
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.4948720932006836,
      "learning_rate": 8.37209147559056e-06,
      "loss": 2.1756,
      "step": 213036
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.335869550704956,
      "learning_rate": 8.371131874389658e-06,
      "loss": 2.1576,
      "step": 213037
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.3391084671020508,
      "learning_rate": 8.37017232740812e-06,
      "loss": 2.1062,
      "step": 213038
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.5646640062332153,
      "learning_rate": 8.369212834646011e-06,
      "loss": 1.8622,
      "step": 213039
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.3869434595108032,
      "learning_rate": 8.36825339610363e-06,
      "loss": 1.9814,
      "step": 213040
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.483842134475708,
      "learning_rate": 8.367294011781079e-06,
      "loss": 1.7786,
      "step": 213041
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.3439732789993286,
      "learning_rate": 8.366334681678522e-06,
      "loss": 1.7916,
      "step": 213042
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.4898302555084229,
      "learning_rate": 8.365375405796193e-06,
      "loss": 2.0563,
      "step": 213043
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.594583511352539,
      "learning_rate": 8.364416184134259e-06,
      "loss": 2.2462,
      "step": 213044
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.4838069677352905,
      "learning_rate": 8.363457016692854e-06,
      "loss": 1.978,
      "step": 213045
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.470729112625122,
      "learning_rate": 8.36249790347221e-06,
      "loss": 2.1692,
      "step": 213046
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.4531736373901367,
      "learning_rate": 8.361538844472492e-06,
      "loss": 1.9362,
      "step": 213047
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.5151609182357788,
      "learning_rate": 8.360579839693837e-06,
      "loss": 2.1454,
      "step": 213048
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.407509207725525,
      "learning_rate": 8.359620889136475e-06,
      "loss": 2.2012,
      "step": 213049
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.4488786458969116,
      "learning_rate": 8.358661992800541e-06,
      "loss": 1.9191,
      "step": 213050
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.4216934442520142,
      "learning_rate": 8.357703150686268e-06,
      "loss": 2.0558,
      "step": 213051
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.523729681968689,
      "learning_rate": 8.356744362793787e-06,
      "loss": 1.8754,
      "step": 213052
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.5219131708145142,
      "learning_rate": 8.355785629123335e-06,
      "loss": 2.1711,
      "step": 213053
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.4583402872085571,
      "learning_rate": 8.354826949674975e-06,
      "loss": 1.7036,
      "step": 213054
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.3726006746292114,
      "learning_rate": 8.353868324449008e-06,
      "loss": 1.8221,
      "step": 213055
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.3505148887634277,
      "learning_rate": 8.352909753445536e-06,
      "loss": 2.0385,
      "step": 213056
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.5189578533172607,
      "learning_rate": 8.351951236664756e-06,
      "loss": 1.847,
      "step": 213057
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.3478368520736694,
      "learning_rate": 8.350992774106902e-06,
      "loss": 2.0208,
      "step": 213058
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.501517653465271,
      "learning_rate": 8.350034365772074e-06,
      "loss": 1.9738,
      "step": 213059
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.459072232246399,
      "learning_rate": 8.349076011660471e-06,
      "loss": 2.1313,
      "step": 213060
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.5189257860183716,
      "learning_rate": 8.348117711772296e-06,
      "loss": 1.8128,
      "step": 213061
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.4189296960830688,
      "learning_rate": 8.347159466107679e-06,
      "loss": 2.1048,
      "step": 213062
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.4940698146820068,
      "learning_rate": 8.346201274666887e-06,
      "loss": 1.8841,
      "step": 213063
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.5970268249511719,
      "learning_rate": 8.345243137449987e-06,
      "loss": 1.9131,
      "step": 213064
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.5226584672927856,
      "learning_rate": 8.344285054457278e-06,
      "loss": 2.0908,
      "step": 213065
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.4347947835922241,
      "learning_rate": 8.343327025688796e-06,
      "loss": 2.0484,
      "step": 213066
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.4714219570159912,
      "learning_rate": 8.342369051144837e-06,
      "loss": 2.1808,
      "step": 213067
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.476291537284851,
      "learning_rate": 8.341411130825504e-06,
      "loss": 2.0091,
      "step": 213068
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.4510201215744019,
      "learning_rate": 8.340453264731029e-06,
      "loss": 2.0369,
      "step": 213069
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.4357116222381592,
      "learning_rate": 8.339495452861578e-06,
      "loss": 1.9334,
      "step": 213070
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.5220810174942017,
      "learning_rate": 8.338537695217351e-06,
      "loss": 1.9988,
      "step": 213071
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.4186351299285889,
      "learning_rate": 8.33757999179845e-06,
      "loss": 1.9865,
      "step": 213072
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.4401243925094604,
      "learning_rate": 8.336622342605104e-06,
      "loss": 1.89,
      "step": 213073
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.5079090595245361,
      "learning_rate": 8.33566474763745e-06,
      "loss": 1.9324,
      "step": 213074
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.4423081874847412,
      "learning_rate": 8.334707206895752e-06,
      "loss": 2.0935,
      "step": 213075
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.4624695777893066,
      "learning_rate": 8.333749720380112e-06,
      "loss": 2.2189,
      "step": 213076
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.3557060956954956,
      "learning_rate": 8.332792288090795e-06,
      "loss": 1.9412,
      "step": 213077
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.5098129510879517,
      "learning_rate": 8.331834910027868e-06,
      "loss": 2.2012,
      "step": 213078
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.6007024049758911,
      "learning_rate": 8.330877586191564e-06,
      "loss": 2.1595,
      "step": 213079
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.4299404621124268,
      "learning_rate": 8.329920316582018e-06,
      "loss": 1.7886,
      "step": 213080
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.3939018249511719,
      "learning_rate": 8.328963101199493e-06,
      "loss": 1.909,
      "step": 213081
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.629620909690857,
      "learning_rate": 8.328005940044091e-06,
      "loss": 1.9379,
      "step": 213082
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.4822462797164917,
      "learning_rate": 8.327048833116079e-06,
      "loss": 1.9641,
      "step": 213083
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.446665644645691,
      "learning_rate": 8.32609178041549e-06,
      "loss": 1.7986,
      "step": 213084
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.5608019828796387,
      "learning_rate": 8.325134781942655e-06,
      "loss": 2.1278,
      "step": 213085
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.4535017013549805,
      "learning_rate": 8.32417783769761e-06,
      "loss": 2.0109,
      "step": 213086
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.5220917463302612,
      "learning_rate": 8.323220947680687e-06,
      "loss": 1.9746,
      "step": 213087
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.4798988103866577,
      "learning_rate": 8.322264111891919e-06,
      "loss": 1.9233,
      "step": 213088
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.5792267322540283,
      "learning_rate": 8.321307330331606e-06,
      "loss": 1.8644,
      "step": 213089
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.4463672637939453,
      "learning_rate": 8.320350602999815e-06,
      "loss": 1.9339,
      "step": 213090
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.5659829378128052,
      "learning_rate": 8.319393929896811e-06,
      "loss": 2.0582,
      "step": 213091
    },
    {
      "epoch": 2.77,
      "grad_norm": 2.377927780151367,
      "learning_rate": 8.318437311022696e-06,
      "loss": 1.989,
      "step": 213092
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.3973037004470825,
      "learning_rate": 8.317480746377737e-06,
      "loss": 1.7989,
      "step": 213093
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.4391170740127563,
      "learning_rate": 8.31652423596203e-06,
      "loss": 1.898,
      "step": 213094
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.5123684406280518,
      "learning_rate": 8.315567779775845e-06,
      "loss": 1.8145,
      "step": 213095
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.5610156059265137,
      "learning_rate": 8.314611377819246e-06,
      "loss": 2.2338,
      "step": 213096
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.5181902647018433,
      "learning_rate": 8.31365503009247e-06,
      "loss": 2.0914,
      "step": 213097
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.358786702156067,
      "learning_rate": 8.31269873659568e-06,
      "loss": 2.0853,
      "step": 213098
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.4207149744033813,
      "learning_rate": 8.31174249732911e-06,
      "loss": 1.9688,
      "step": 213099
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.6974518299102783,
      "learning_rate": 8.31078631229286e-06,
      "loss": 1.9887,
      "step": 213100
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.354761004447937,
      "learning_rate": 8.309830181487199e-06,
      "loss": 1.8778,
      "step": 213101
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.4478366374969482,
      "learning_rate": 8.308874104912156e-06,
      "loss": 1.8169,
      "step": 213102
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.476790189743042,
      "learning_rate": 8.307918082568065e-06,
      "loss": 1.8255,
      "step": 213103
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.4680390357971191,
      "learning_rate": 8.306962114454996e-06,
      "loss": 1.8155,
      "step": 213104
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.450125813484192,
      "learning_rate": 8.306006200573178e-06,
      "loss": 1.9894,
      "step": 213105
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.5326493978500366,
      "learning_rate": 8.305050340922782e-06,
      "loss": 1.8888,
      "step": 213106
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.44695246219635,
      "learning_rate": 8.304094535504035e-06,
      "loss": 1.8576,
      "step": 213107
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.3205746412277222,
      "learning_rate": 8.303138784317009e-06,
      "loss": 2.0977,
      "step": 213108
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.4233675003051758,
      "learning_rate": 8.302183087361935e-06,
      "loss": 1.6558,
      "step": 213109
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.4016352891921997,
      "learning_rate": 8.30122744463898e-06,
      "loss": 1.7939,
      "step": 213110
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.3516310453414917,
      "learning_rate": 8.300271856148377e-06,
      "loss": 2.1204,
      "step": 213111
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.53499174118042,
      "learning_rate": 8.299316321890225e-06,
      "loss": 2.1189,
      "step": 213112
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.4338220357894897,
      "learning_rate": 8.29836084186476e-06,
      "loss": 2.0373,
      "step": 213113
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.5448092222213745,
      "learning_rate": 8.297405416072111e-06,
      "loss": 1.9407,
      "step": 213114
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.5714768171310425,
      "learning_rate": 8.296450044512516e-06,
      "loss": 1.9737,
      "step": 213115
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.5391123294830322,
      "learning_rate": 8.29549472718607e-06,
      "loss": 1.9546,
      "step": 213116
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.593062162399292,
      "learning_rate": 8.294539464093041e-06,
      "loss": 2.2875,
      "step": 213117
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.4336050748825073,
      "learning_rate": 8.293584255233532e-06,
      "loss": 1.8532,
      "step": 213118
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.4492981433868408,
      "learning_rate": 8.292629100607773e-06,
      "loss": 1.9503,
      "step": 213119
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.4187756776809692,
      "learning_rate": 8.29167400021593e-06,
      "loss": 1.9949,
      "step": 213120
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.527205228805542,
      "learning_rate": 8.290718954058173e-06,
      "loss": 1.9849,
      "step": 213121
    },
    {
      "epoch": 2.77,
      "grad_norm": 1.3959180116653442,
      "learning_rate": 8.289763962134632e-06,
      "loss": 1.7808,
      "step": 213122
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.4053236246109009,
      "learning_rate": 8.288809024445575e-06,
      "loss": 2.0419,
      "step": 213123
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.439252495765686,
      "learning_rate": 8.287854140991102e-06,
      "loss": 1.9387,
      "step": 213124
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.516571044921875,
      "learning_rate": 8.286899311771445e-06,
      "loss": 2.078,
      "step": 213125
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.6001710891723633,
      "learning_rate": 8.285944536786771e-06,
      "loss": 2.2019,
      "step": 213126
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.5908101797103882,
      "learning_rate": 8.284989816037212e-06,
      "loss": 2.1089,
      "step": 213127
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.6699150800704956,
      "learning_rate": 8.284035149523006e-06,
      "loss": 1.9288,
      "step": 213128
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.3683360815048218,
      "learning_rate": 8.283080537244313e-06,
      "loss": 2.0099,
      "step": 213129
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.466671109199524,
      "learning_rate": 8.282125979201237e-06,
      "loss": 1.9138,
      "step": 213130
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.4447396993637085,
      "learning_rate": 8.281171475394077e-06,
      "loss": 2.1632,
      "step": 213131
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.4150524139404297,
      "learning_rate": 8.280217025822966e-06,
      "loss": 1.8259,
      "step": 213132
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.559999704360962,
      "learning_rate": 8.279262630488037e-06,
      "loss": 1.9973,
      "step": 213133
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.5068186521530151,
      "learning_rate": 8.278308289389523e-06,
      "loss": 2.0055,
      "step": 213134
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.4826428890228271,
      "learning_rate": 8.277354002527559e-06,
      "loss": 2.0577,
      "step": 213135
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.4726009368896484,
      "learning_rate": 8.276399769902342e-06,
      "loss": 2.173,
      "step": 213136
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.6148982048034668,
      "learning_rate": 8.275445591514073e-06,
      "loss": 2.0562,
      "step": 213137
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.4221606254577637,
      "learning_rate": 8.274491467362887e-06,
      "loss": 2.0517,
      "step": 213138
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.3833239078521729,
      "learning_rate": 8.273537397448981e-06,
      "loss": 1.9272,
      "step": 213139
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.295698642730713,
      "learning_rate": 8.272583381772525e-06,
      "loss": 1.8035,
      "step": 213140
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.3934566974639893,
      "learning_rate": 8.271629420333681e-06,
      "loss": 2.089,
      "step": 213141
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.4076426029205322,
      "learning_rate": 8.270675513132685e-06,
      "loss": 1.9397,
      "step": 213142
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.4789646863937378,
      "learning_rate": 8.269721660169704e-06,
      "loss": 1.9283,
      "step": 213143
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.574648141860962,
      "learning_rate": 8.26876786144487e-06,
      "loss": 1.9213,
      "step": 213144
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.5148011445999146,
      "learning_rate": 8.267814116958315e-06,
      "loss": 1.9461,
      "step": 213145
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.4693185091018677,
      "learning_rate": 8.266860426710343e-06,
      "loss": 2.1468,
      "step": 213146
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.5812567472457886,
      "learning_rate": 8.26590679070105e-06,
      "loss": 2.1897,
      "step": 213147
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.3126411437988281,
      "learning_rate": 8.264953208930635e-06,
      "loss": 1.9063,
      "step": 213148
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.43712317943573,
      "learning_rate": 8.263999681399235e-06,
      "loss": 1.8806,
      "step": 213149
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.6260936260223389,
      "learning_rate": 8.26304620810715e-06,
      "loss": 2.0315,
      "step": 213150
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.3576971292495728,
      "learning_rate": 8.26209278905441e-06,
      "loss": 2.0768,
      "step": 213151
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.455346941947937,
      "learning_rate": 8.261139424241281e-06,
      "loss": 2.1516,
      "step": 213152
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.6576623916625977,
      "learning_rate": 8.260186113667865e-06,
      "loss": 1.9668,
      "step": 213153
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.459429383277893,
      "learning_rate": 8.25923285733443e-06,
      "loss": 1.9959,
      "step": 213154
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.3005192279815674,
      "learning_rate": 8.258279655241106e-06,
      "loss": 1.8354,
      "step": 213155
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.394063949584961,
      "learning_rate": 8.257326507388096e-06,
      "loss": 1.8984,
      "step": 213156
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.5130935907363892,
      "learning_rate": 8.25637341377553e-06,
      "loss": 2.3307,
      "step": 213157
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.3227218389511108,
      "learning_rate": 8.25542037440361e-06,
      "loss": 1.919,
      "step": 213158
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.4356473684310913,
      "learning_rate": 8.2544673892725e-06,
      "loss": 1.9795,
      "step": 213159
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.43259859085083,
      "learning_rate": 8.253514458382404e-06,
      "loss": 1.9277,
      "step": 213160
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.6712404489517212,
      "learning_rate": 8.252561581733486e-06,
      "loss": 1.9505,
      "step": 213161
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.5156821012496948,
      "learning_rate": 8.251608759325978e-06,
      "loss": 2.1169,
      "step": 213162
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.55740225315094,
      "learning_rate": 8.250655991159915e-06,
      "loss": 1.8069,
      "step": 213163
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.4386589527130127,
      "learning_rate": 8.24970327723563e-06,
      "loss": 2.0122,
      "step": 213164
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.3897804021835327,
      "learning_rate": 8.248750617553158e-06,
      "loss": 1.8428,
      "step": 213165
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.483650803565979,
      "learning_rate": 8.247798012112827e-06,
      "loss": 1.6861,
      "step": 213166
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.4960898160934448,
      "learning_rate": 8.246845460914676e-06,
      "loss": 2.1112,
      "step": 213167
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.4731073379516602,
      "learning_rate": 8.245892963959034e-06,
      "loss": 2.0815,
      "step": 213168
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.5188813209533691,
      "learning_rate": 8.244940521245903e-06,
      "loss": 2.2617,
      "step": 213169
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.3975367546081543,
      "learning_rate": 8.24398813277558e-06,
      "loss": 2.1939,
      "step": 213170
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.426903486251831,
      "learning_rate": 8.24303579854817e-06,
      "loss": 1.912,
      "step": 213171
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.4910554885864258,
      "learning_rate": 8.242083518563935e-06,
      "loss": 2.1334,
      "step": 213172
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.5543246269226074,
      "learning_rate": 8.241131292822945e-06,
      "loss": 2.0323,
      "step": 213173
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.4404901266098022,
      "learning_rate": 8.24017912132553e-06,
      "loss": 2.0941,
      "step": 213174
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.558366060256958,
      "learning_rate": 8.239227004071692e-06,
      "loss": 1.9831,
      "step": 213175
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.3856483697891235,
      "learning_rate": 8.238274941061728e-06,
      "loss": 1.9906,
      "step": 213176
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.567061185836792,
      "learning_rate": 8.237322932295743e-06,
      "loss": 1.9229,
      "step": 213177
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.4786429405212402,
      "learning_rate": 8.236370977773966e-06,
      "loss": 2.1345,
      "step": 213178
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.4847689867019653,
      "learning_rate": 8.23541907749653e-06,
      "loss": 2.1767,
      "step": 213179
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.6363704204559326,
      "learning_rate": 8.234467231463705e-06,
      "loss": 1.9891,
      "step": 213180
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.3434584140777588,
      "learning_rate": 8.233515439675553e-06,
      "loss": 1.9975,
      "step": 213181
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.4902796745300293,
      "learning_rate": 8.23256370213231e-06,
      "loss": 2.3195,
      "step": 213182
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.476536512374878,
      "learning_rate": 8.23161201883411e-06,
      "loss": 1.9341,
      "step": 213183
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.844164252281189,
      "learning_rate": 8.230660389781185e-06,
      "loss": 1.9928,
      "step": 213184
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.636118769645691,
      "learning_rate": 8.229708814973667e-06,
      "loss": 2.2359,
      "step": 213185
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.5504366159439087,
      "learning_rate": 8.228757294411826e-06,
      "loss": 1.7859,
      "step": 213186
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.481890082359314,
      "learning_rate": 8.22780582809569e-06,
      "loss": 1.861,
      "step": 213187
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.4593926668167114,
      "learning_rate": 8.226854416025563e-06,
      "loss": 1.8731,
      "step": 213188
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.5144898891448975,
      "learning_rate": 8.225903058201511e-06,
      "loss": 2.0644,
      "step": 213189
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.373158574104309,
      "learning_rate": 8.224951754623832e-06,
      "loss": 1.7812,
      "step": 213190
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.3639816045761108,
      "learning_rate": 8.224000505292593e-06,
      "loss": 1.8827,
      "step": 213191
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.5489238500595093,
      "learning_rate": 8.223049310208029e-06,
      "loss": 1.945,
      "step": 213192
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.9119936227798462,
      "learning_rate": 8.22209816937034e-06,
      "loss": 1.9935,
      "step": 213193
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.5055714845657349,
      "learning_rate": 8.221147082779656e-06,
      "loss": 2.0078,
      "step": 213194
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.4432376623153687,
      "learning_rate": 8.220196050436145e-06,
      "loss": 2.2063,
      "step": 213195
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.4428684711456299,
      "learning_rate": 8.219245072340041e-06,
      "loss": 2.0197,
      "step": 213196
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.4719090461730957,
      "learning_rate": 8.218294148491478e-06,
      "loss": 1.9183,
      "step": 213197
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.9510349035263062,
      "learning_rate": 8.21734327889062e-06,
      "loss": 1.9821,
      "step": 213198
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.526007056236267,
      "learning_rate": 8.216392463537703e-06,
      "loss": 1.9524,
      "step": 213199
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.5947891473770142,
      "learning_rate": 8.215441702432856e-06,
      "loss": 2.224,
      "step": 213200
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.5007801055908203,
      "learning_rate": 8.214490995576251e-06,
      "loss": 2.0817,
      "step": 213201
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.3679900169372559,
      "learning_rate": 8.213540342968084e-06,
      "loss": 2.123,
      "step": 213202
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.6907199621200562,
      "learning_rate": 8.212589744608522e-06,
      "loss": 2.1765,
      "step": 213203
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.4219752550125122,
      "learning_rate": 8.211639200497767e-06,
      "loss": 2.2159,
      "step": 213204
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.4398791790008545,
      "learning_rate": 8.210688710635948e-06,
      "loss": 2.1401,
      "step": 213205
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.5431170463562012,
      "learning_rate": 8.209738275023303e-06,
      "loss": 1.742,
      "step": 213206
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.440007209777832,
      "learning_rate": 8.208787893659962e-06,
      "loss": 1.7541,
      "step": 213207
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.3989914655685425,
      "learning_rate": 8.207837566546094e-06,
      "loss": 2.0776,
      "step": 213208
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.6754605770111084,
      "learning_rate": 8.206887293681897e-06,
      "loss": 2.0285,
      "step": 213209
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.5400793552398682,
      "learning_rate": 8.205937075067604e-06,
      "loss": 2.0019,
      "step": 213210
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.399837851524353,
      "learning_rate": 8.204986910703283e-06,
      "loss": 2.0531,
      "step": 213211
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.4834429025650024,
      "learning_rate": 8.204036800589165e-06,
      "loss": 2.1407,
      "step": 213212
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.6055619716644287,
      "learning_rate": 8.203086744725451e-06,
      "loss": 1.8849,
      "step": 213213
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.5084279775619507,
      "learning_rate": 8.202136743112275e-06,
      "loss": 1.9997,
      "step": 213214
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.5256942510604858,
      "learning_rate": 8.201186795749836e-06,
      "loss": 1.9458,
      "step": 213215
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.563222050666809,
      "learning_rate": 8.200236902638302e-06,
      "loss": 2.0477,
      "step": 213216
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.5077579021453857,
      "learning_rate": 8.19928706377787e-06,
      "loss": 1.9438,
      "step": 213217
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.4101706743240356,
      "learning_rate": 8.198337279168643e-06,
      "loss": 2.105,
      "step": 213218
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.8411240577697754,
      "learning_rate": 8.197387548810919e-06,
      "loss": 1.7623,
      "step": 213219
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.4436309337615967,
      "learning_rate": 8.196437872704764e-06,
      "loss": 1.8649,
      "step": 213220
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.3967472314834595,
      "learning_rate": 8.195488250850414e-06,
      "loss": 1.7673,
      "step": 213221
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.5022764205932617,
      "learning_rate": 8.19453868324803e-06,
      "loss": 1.9116,
      "step": 213222
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.5080375671386719,
      "learning_rate": 8.193589169897818e-06,
      "loss": 1.9517,
      "step": 213223
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.3524787425994873,
      "learning_rate": 8.192639710799875e-06,
      "loss": 2.0468,
      "step": 213224
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.5208463668823242,
      "learning_rate": 8.191690305954468e-06,
      "loss": 1.7703,
      "step": 213225
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.378578782081604,
      "learning_rate": 8.190740955361697e-06,
      "loss": 2.0226,
      "step": 213226
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.45297110080719,
      "learning_rate": 8.189791659021794e-06,
      "loss": 2.184,
      "step": 213227
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.3771544694900513,
      "learning_rate": 8.188842416934926e-06,
      "loss": 2.0085,
      "step": 213228
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.4916975498199463,
      "learning_rate": 8.187893229101261e-06,
      "loss": 1.887,
      "step": 213229
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.4842299222946167,
      "learning_rate": 8.186944095520964e-06,
      "loss": 2.272,
      "step": 213230
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.4135020971298218,
      "learning_rate": 8.185995016194235e-06,
      "loss": 2.0272,
      "step": 213231
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.7807515859603882,
      "learning_rate": 8.185045991121209e-06,
      "loss": 1.7989,
      "step": 213232
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.4625452756881714,
      "learning_rate": 8.184097020302116e-06,
      "loss": 2.0601,
      "step": 213233
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.6371244192123413,
      "learning_rate": 8.183148103737092e-06,
      "loss": 1.7588,
      "step": 213234
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.419948935508728,
      "learning_rate": 8.182199241426401e-06,
      "loss": 2.1248,
      "step": 213235
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.4742028713226318,
      "learning_rate": 8.181250433370046e-06,
      "loss": 2.003,
      "step": 213236
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.538539171218872,
      "learning_rate": 8.180301679568357e-06,
      "loss": 1.8669,
      "step": 213237
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.4812891483306885,
      "learning_rate": 8.179352980021437e-06,
      "loss": 2.0889,
      "step": 213238
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.4033794403076172,
      "learning_rate": 8.178404334729517e-06,
      "loss": 1.9447,
      "step": 213239
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.419919490814209,
      "learning_rate": 8.177455743692696e-06,
      "loss": 2.1426,
      "step": 213240
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.3542989492416382,
      "learning_rate": 8.176507206911243e-06,
      "loss": 2.1027,
      "step": 213241
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.4748685359954834,
      "learning_rate": 8.175558724385257e-06,
      "loss": 2.0738,
      "step": 213242
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.3664807081222534,
      "learning_rate": 8.17461029611497e-06,
      "loss": 1.7859,
      "step": 213243
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.3398544788360596,
      "learning_rate": 8.173661922100483e-06,
      "loss": 2.1239,
      "step": 213244
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.3828281164169312,
      "learning_rate": 8.172713602342062e-06,
      "loss": 1.9331,
      "step": 213245
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.6060961484909058,
      "learning_rate": 8.171765336839775e-06,
      "loss": 1.8811,
      "step": 213246
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.4333316087722778,
      "learning_rate": 8.170817125593988e-06,
      "loss": 2.032,
      "step": 213247
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.4238344430923462,
      "learning_rate": 8.169868968604664e-06,
      "loss": 1.8242,
      "step": 213248
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.4104042053222656,
      "learning_rate": 8.168920865872075e-06,
      "loss": 1.8016,
      "step": 213249
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.4746356010437012,
      "learning_rate": 8.167972817396385e-06,
      "loss": 1.995,
      "step": 213250
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.3617404699325562,
      "learning_rate": 8.167024823177825e-06,
      "loss": 1.9206,
      "step": 213251
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.6187331676483154,
      "learning_rate": 8.166076883216466e-06,
      "loss": 1.9835,
      "step": 213252
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.3738597631454468,
      "learning_rate": 8.165128997512637e-06,
      "loss": 1.8391,
      "step": 213253
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.470276951789856,
      "learning_rate": 8.16418116606634e-06,
      "loss": 1.8888,
      "step": 213254
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.432313084602356,
      "learning_rate": 8.163233388877844e-06,
      "loss": 2.0392,
      "step": 213255
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.5421044826507568,
      "learning_rate": 8.16228566594731e-06,
      "loss": 2.2837,
      "step": 213256
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.573258876800537,
      "learning_rate": 8.161337997274909e-06,
      "loss": 1.6818,
      "step": 213257
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.6990153789520264,
      "learning_rate": 8.160390382860837e-06,
      "loss": 2.0137,
      "step": 213258
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.402125597000122,
      "learning_rate": 8.159442822705265e-06,
      "loss": 2.1507,
      "step": 213259
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.4419465065002441,
      "learning_rate": 8.158495316808355e-06,
      "loss": 1.9679,
      "step": 213260
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.5400795936584473,
      "learning_rate": 8.157547865170311e-06,
      "loss": 2.0658,
      "step": 213261
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.4545735120773315,
      "learning_rate": 8.15660046779123e-06,
      "loss": 1.8485,
      "step": 213262
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.6475489139556885,
      "learning_rate": 8.155653124671414e-06,
      "loss": 2.2697,
      "step": 213263
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.3883962631225586,
      "learning_rate": 8.154705835810893e-06,
      "loss": 1.8683,
      "step": 213264
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.6207118034362793,
      "learning_rate": 8.153758601209969e-06,
      "loss": 1.9285,
      "step": 213265
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.3748267889022827,
      "learning_rate": 8.15281142086881e-06,
      "loss": 1.8336,
      "step": 213266
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.5052112340927124,
      "learning_rate": 8.151864294787514e-06,
      "loss": 2.0374,
      "step": 213267
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.4901292324066162,
      "learning_rate": 8.150917222966246e-06,
      "loss": 2.1109,
      "step": 213268
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.4749079942703247,
      "learning_rate": 8.14997020540531e-06,
      "loss": 1.9575,
      "step": 213269
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.5170036554336548,
      "learning_rate": 8.149023242104734e-06,
      "loss": 1.9375,
      "step": 213270
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.3349213600158691,
      "learning_rate": 8.148076333064824e-06,
      "loss": 1.8919,
      "step": 213271
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.5150196552276611,
      "learning_rate": 8.147129478285674e-06,
      "loss": 1.9428,
      "step": 213272
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.6042795181274414,
      "learning_rate": 8.146182677767488e-06,
      "loss": 1.8122,
      "step": 213273
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.454857587814331,
      "learning_rate": 8.14523593151043e-06,
      "loss": 2.1178,
      "step": 213274
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.4847372770309448,
      "learning_rate": 8.144289239514667e-06,
      "loss": 1.9005,
      "step": 213275
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.523040533065796,
      "learning_rate": 8.1433426017804e-06,
      "loss": 2.2811,
      "step": 213276
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.396064281463623,
      "learning_rate": 8.142396018307795e-06,
      "loss": 1.9527,
      "step": 213277
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.427430272102356,
      "learning_rate": 8.141449489097051e-06,
      "loss": 2.1337,
      "step": 213278
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.3314549922943115,
      "learning_rate": 8.140503014148302e-06,
      "loss": 2.1058,
      "step": 213279
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.5875391960144043,
      "learning_rate": 8.139556593461715e-06,
      "loss": 1.9862,
      "step": 213280
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.4797759056091309,
      "learning_rate": 8.138610227037523e-06,
      "loss": 1.8124,
      "step": 213281
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.4254134893417358,
      "learning_rate": 8.137663914875859e-06,
      "loss": 2.1534,
      "step": 213282
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.5316132307052612,
      "learning_rate": 8.13671765697692e-06,
      "loss": 1.9498,
      "step": 213283
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.520138144493103,
      "learning_rate": 8.135771453340912e-06,
      "loss": 1.8869,
      "step": 213284
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.4999706745147705,
      "learning_rate": 8.13482530396793e-06,
      "loss": 1.5197,
      "step": 213285
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.4123315811157227,
      "learning_rate": 8.133879208858207e-06,
      "loss": 1.905,
      "step": 213286
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.6057239770889282,
      "learning_rate": 8.132933168011912e-06,
      "loss": 1.9491,
      "step": 213287
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.3940645456314087,
      "learning_rate": 8.131987181429177e-06,
      "loss": 1.9558,
      "step": 213288
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.502828598022461,
      "learning_rate": 8.131041249110238e-06,
      "loss": 2.0829,
      "step": 213289
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.407372236251831,
      "learning_rate": 8.130095371055256e-06,
      "loss": 1.963,
      "step": 213290
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.5240514278411865,
      "learning_rate": 8.129149547264401e-06,
      "loss": 2.1314,
      "step": 213291
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.5551108121871948,
      "learning_rate": 8.12820377773784e-06,
      "loss": 2.136,
      "step": 213292
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.5465338230133057,
      "learning_rate": 8.127258062475738e-06,
      "loss": 1.9885,
      "step": 213293
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.5002098083496094,
      "learning_rate": 8.126312401478296e-06,
      "loss": 1.8911,
      "step": 213294
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.3900386095046997,
      "learning_rate": 8.125366794745714e-06,
      "loss": 1.8553,
      "step": 213295
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.3260687589645386,
      "learning_rate": 8.124421242278124e-06,
      "loss": 1.7666,
      "step": 213296
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.5022977590560913,
      "learning_rate": 8.123475744075658e-06,
      "loss": 1.9479,
      "step": 213297
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.5745656490325928,
      "learning_rate": 8.12253030013862e-06,
      "loss": 2.0445,
      "step": 213298
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.2949299812316895,
      "learning_rate": 8.121584910467106e-06,
      "loss": 2.2517,
      "step": 213299
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.4491851329803467,
      "learning_rate": 8.120639575061249e-06,
      "loss": 1.9192,
      "step": 213300
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.469566822052002,
      "learning_rate": 8.11969429392132e-06,
      "loss": 1.8635,
      "step": 213301
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.3002514839172363,
      "learning_rate": 8.118749067047414e-06,
      "loss": 1.7378,
      "step": 213302
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.5223246812820435,
      "learning_rate": 8.117803894439767e-06,
      "loss": 2.2975,
      "step": 213303
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.5140799283981323,
      "learning_rate": 8.116858776098512e-06,
      "loss": 2.0407,
      "step": 213304
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.5444209575653076,
      "learning_rate": 8.11591371202388e-06,
      "loss": 2.0246,
      "step": 213305
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.4489754438400269,
      "learning_rate": 8.114968702215973e-06,
      "loss": 1.9465,
      "step": 213306
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.6668646335601807,
      "learning_rate": 8.114023746675025e-06,
      "loss": 1.8003,
      "step": 213307
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.4519442319869995,
      "learning_rate": 8.113078845401167e-06,
      "loss": 1.9803,
      "step": 213308
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.3745121955871582,
      "learning_rate": 8.1121339983946e-06,
      "loss": 1.7975,
      "step": 213309
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.4564305543899536,
      "learning_rate": 8.111189205655521e-06,
      "loss": 2.0076,
      "step": 213310
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.6932165622711182,
      "learning_rate": 8.110244467184035e-06,
      "loss": 1.9737,
      "step": 213311
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.4309343099594116,
      "learning_rate": 8.109299782980405e-06,
      "loss": 1.6996,
      "step": 213312
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.5948539972305298,
      "learning_rate": 8.108355153044766e-06,
      "loss": 1.8926,
      "step": 213313
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.3207721710205078,
      "learning_rate": 8.107410577377315e-06,
      "loss": 1.8432,
      "step": 213314
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.5171622037887573,
      "learning_rate": 8.106466055978156e-06,
      "loss": 2.0697,
      "step": 213315
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.4662290811538696,
      "learning_rate": 8.10552158884752e-06,
      "loss": 1.8507,
      "step": 213316
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.4581482410430908,
      "learning_rate": 8.104577175985571e-06,
      "loss": 1.9281,
      "step": 213317
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.3788142204284668,
      "learning_rate": 8.103632817392546e-06,
      "loss": 2.0255,
      "step": 213318
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.5769628286361694,
      "learning_rate": 8.102688513068544e-06,
      "loss": 2.038,
      "step": 213319
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.6975631713867188,
      "learning_rate": 8.101744263013765e-06,
      "loss": 1.9687,
      "step": 213320
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.45357346534729,
      "learning_rate": 8.100800067228341e-06,
      "loss": 1.8872,
      "step": 213321
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.410465955734253,
      "learning_rate": 8.099855925712507e-06,
      "loss": 1.9614,
      "step": 213322
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.6149150133132935,
      "learning_rate": 8.098911838466427e-06,
      "loss": 1.8698,
      "step": 213323
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.6953569650650024,
      "learning_rate": 8.097967805490268e-06,
      "loss": 2.0775,
      "step": 213324
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.5226995944976807,
      "learning_rate": 8.0970238267842e-06,
      "loss": 2.1225,
      "step": 213325
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.4784259796142578,
      "learning_rate": 8.096079902348418e-06,
      "loss": 2.0368,
      "step": 213326
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.4120755195617676,
      "learning_rate": 8.095136032183091e-06,
      "loss": 2.0997,
      "step": 213327
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.3617329597473145,
      "learning_rate": 8.094192216288387e-06,
      "loss": 1.9086,
      "step": 213328
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.6907442808151245,
      "learning_rate": 8.093248454664469e-06,
      "loss": 1.8883,
      "step": 213329
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.4609088897705078,
      "learning_rate": 8.09230474731154e-06,
      "loss": 2.1046,
      "step": 213330
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.4456714391708374,
      "learning_rate": 8.091361094229732e-06,
      "loss": 1.9966,
      "step": 213331
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.4402700662612915,
      "learning_rate": 8.090417495419277e-06,
      "loss": 2.0305,
      "step": 213332
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.6875826120376587,
      "learning_rate": 8.08947395088031e-06,
      "loss": 1.9097,
      "step": 213333
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.5137141942977905,
      "learning_rate": 8.088530460613063e-06,
      "loss": 2.044,
      "step": 213334
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.4057648181915283,
      "learning_rate": 8.087587024617603e-06,
      "loss": 2.0304,
      "step": 213335
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.4498927593231201,
      "learning_rate": 8.086643642894198e-06,
      "loss": 1.8158,
      "step": 213336
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.6097378730773926,
      "learning_rate": 8.085700315442977e-06,
      "loss": 1.8714,
      "step": 213337
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.486284613609314,
      "learning_rate": 8.084757042264178e-06,
      "loss": 2.1155,
      "step": 213338
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.5974750518798828,
      "learning_rate": 8.083813823357932e-06,
      "loss": 2.0313,
      "step": 213339
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.6894843578338623,
      "learning_rate": 8.082870658724406e-06,
      "loss": 1.7594,
      "step": 213340
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.4095631837844849,
      "learning_rate": 8.081927548363765e-06,
      "loss": 2.0034,
      "step": 213341
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.3953403234481812,
      "learning_rate": 8.08098449227621e-06,
      "loss": 1.9552,
      "step": 213342
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.6360383033752441,
      "learning_rate": 8.080041490461908e-06,
      "loss": 1.9294,
      "step": 213343
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.5029139518737793,
      "learning_rate": 8.079098542921026e-06,
      "loss": 1.8155,
      "step": 213344
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.4025511741638184,
      "learning_rate": 8.078155649653795e-06,
      "loss": 2.1263,
      "step": 213345
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.6581780910491943,
      "learning_rate": 8.077212810660316e-06,
      "loss": 1.8881,
      "step": 213346
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.4193905591964722,
      "learning_rate": 8.07627002594079e-06,
      "loss": 2.1876,
      "step": 213347
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.4949344396591187,
      "learning_rate": 8.075327295495415e-06,
      "loss": 2.041,
      "step": 213348
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.4602208137512207,
      "learning_rate": 8.074384619324292e-06,
      "loss": 1.9088,
      "step": 213349
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.4138222932815552,
      "learning_rate": 8.073441997427721e-06,
      "loss": 2.063,
      "step": 213350
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.562167763710022,
      "learning_rate": 8.072499429805768e-06,
      "loss": 2.021,
      "step": 213351
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.4850050210952759,
      "learning_rate": 8.071556916458667e-06,
      "loss": 2.0293,
      "step": 213352
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.4549391269683838,
      "learning_rate": 8.070614457386548e-06,
      "loss": 1.9933,
      "step": 213353
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.6696913242340088,
      "learning_rate": 8.069672052589648e-06,
      "loss": 1.9421,
      "step": 213354
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.6622076034545898,
      "learning_rate": 8.068729702068067e-06,
      "loss": 1.9419,
      "step": 213355
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.3738175630569458,
      "learning_rate": 8.067787405822035e-06,
      "loss": 2.0323,
      "step": 213356
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.554012656211853,
      "learning_rate": 8.066845163851754e-06,
      "loss": 2.055,
      "step": 213357
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.4809850454330444,
      "learning_rate": 8.065902976157324e-06,
      "loss": 1.9537,
      "step": 213358
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.4548417329788208,
      "learning_rate": 8.064960842738943e-06,
      "loss": 1.7771,
      "step": 213359
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.4550199508666992,
      "learning_rate": 8.064018763596814e-06,
      "loss": 2.1193,
      "step": 213360
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.5577664375305176,
      "learning_rate": 8.063076738731067e-06,
      "loss": 1.9285,
      "step": 213361
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.5535401105880737,
      "learning_rate": 8.062134768141903e-06,
      "loss": 2.1075,
      "step": 213362
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.3599532842636108,
      "learning_rate": 8.061192851829556e-06,
      "loss": 1.8706,
      "step": 213363
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.3445683717727661,
      "learning_rate": 8.060250989794092e-06,
      "loss": 1.8955,
      "step": 213364
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.5709445476531982,
      "learning_rate": 8.059309182035744e-06,
      "loss": 1.9341,
      "step": 213365
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.4592394828796387,
      "learning_rate": 8.058367428554712e-06,
      "loss": 2.0541,
      "step": 213366
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.4876469373703003,
      "learning_rate": 8.057425729351097e-06,
      "loss": 1.836,
      "step": 213367
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.3896327018737793,
      "learning_rate": 8.05648408442513e-06,
      "loss": 1.9447,
      "step": 213368
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.6521131992340088,
      "learning_rate": 8.055542493777012e-06,
      "loss": 2.0526,
      "step": 213369
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.7040660381317139,
      "learning_rate": 8.054600957406843e-06,
      "loss": 1.9399,
      "step": 213370
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.417428731918335,
      "learning_rate": 8.053659475314822e-06,
      "loss": 1.9552,
      "step": 213371
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.5034289360046387,
      "learning_rate": 8.052718047501183e-06,
      "loss": 1.9484,
      "step": 213372
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.413648247718811,
      "learning_rate": 8.051776673965992e-06,
      "loss": 1.9642,
      "step": 213373
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.4631479978561401,
      "learning_rate": 8.050835354709551e-06,
      "loss": 2.0373,
      "step": 213374
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.4863851070404053,
      "learning_rate": 8.049894089731924e-06,
      "loss": 1.932,
      "step": 213375
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.4265304803848267,
      "learning_rate": 8.048952879033377e-06,
      "loss": 2.0711,
      "step": 213376
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.502941370010376,
      "learning_rate": 8.048011722613978e-06,
      "loss": 1.7657,
      "step": 213377
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.4927870035171509,
      "learning_rate": 8.047070620474027e-06,
      "loss": 2.0344,
      "step": 213378
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.6023072004318237,
      "learning_rate": 8.046129572613592e-06,
      "loss": 1.9081,
      "step": 213379
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.3564594984054565,
      "learning_rate": 8.045188579032901e-06,
      "loss": 2.0479,
      "step": 213380
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.424025297164917,
      "learning_rate": 8.04424763973216e-06,
      "loss": 2.0809,
      "step": 213381
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.3948794603347778,
      "learning_rate": 8.043306754711465e-06,
      "loss": 1.9826,
      "step": 213382
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.4417780637741089,
      "learning_rate": 8.042365923971051e-06,
      "loss": 2.1698,
      "step": 213383
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.5844718217849731,
      "learning_rate": 8.04142514751105e-06,
      "loss": 1.9568,
      "step": 213384
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.4379035234451294,
      "learning_rate": 8.040484425331662e-06,
      "loss": 1.913,
      "step": 213385
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.4913798570632935,
      "learning_rate": 8.039543757433086e-06,
      "loss": 2.0372,
      "step": 213386
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.4072668552398682,
      "learning_rate": 8.038603143815459e-06,
      "loss": 2.0356,
      "step": 213387
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.478451132774353,
      "learning_rate": 8.037662584478943e-06,
      "loss": 1.9787,
      "step": 213388
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.4096059799194336,
      "learning_rate": 8.03672207942374e-06,
      "loss": 1.8779,
      "step": 213389
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.4148765802383423,
      "learning_rate": 8.03578162865005e-06,
      "loss": 2.0572,
      "step": 213390
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.536435842514038,
      "learning_rate": 8.034841232157974e-06,
      "loss": 1.8995,
      "step": 213391
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.8271267414093018,
      "learning_rate": 8.033900889947776e-06,
      "loss": 1.7407,
      "step": 213392
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.428507685661316,
      "learning_rate": 8.032960602019556e-06,
      "loss": 2.0625,
      "step": 213393
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.4608222246170044,
      "learning_rate": 8.032020368373548e-06,
      "loss": 1.9226,
      "step": 213394
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.516085147857666,
      "learning_rate": 8.031080189009919e-06,
      "loss": 1.9536,
      "step": 213395
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.5438377857208252,
      "learning_rate": 8.030140063928769e-06,
      "loss": 2.2263,
      "step": 213396
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.4029948711395264,
      "learning_rate": 8.029199993130364e-06,
      "loss": 1.9359,
      "step": 213397
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.4613748788833618,
      "learning_rate": 8.028259976614836e-06,
      "loss": 2.0232,
      "step": 213398
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.4749784469604492,
      "learning_rate": 8.027320014382321e-06,
      "loss": 1.9905,
      "step": 213399
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.4863286018371582,
      "learning_rate": 8.026380106433117e-06,
      "loss": 1.8799,
      "step": 213400
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.4550540447235107,
      "learning_rate": 8.02544025276729e-06,
      "loss": 1.9627,
      "step": 213401
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.4286057949066162,
      "learning_rate": 8.024500453385008e-06,
      "loss": 1.9508,
      "step": 213402
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.458253264427185,
      "learning_rate": 8.023560708286502e-06,
      "loss": 1.9447,
      "step": 213403
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.468451738357544,
      "learning_rate": 8.022621017471943e-06,
      "loss": 1.9375,
      "step": 213404
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.5571534633636475,
      "learning_rate": 8.021681380941492e-06,
      "loss": 1.973,
      "step": 213405
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.3826936483383179,
      "learning_rate": 8.020741798695318e-06,
      "loss": 1.9541,
      "step": 213406
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.4957627058029175,
      "learning_rate": 8.019802270733588e-06,
      "loss": 2.1966,
      "step": 213407
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.4162143468856812,
      "learning_rate": 8.018862797056468e-06,
      "loss": 1.8758,
      "step": 213408
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.395430088043213,
      "learning_rate": 8.017923377664193e-06,
      "loss": 2.1512,
      "step": 213409
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.4908255338668823,
      "learning_rate": 8.016984012556893e-06,
      "loss": 2.1327,
      "step": 213410
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.4334968328475952,
      "learning_rate": 8.016044701734736e-06,
      "loss": 1.9026,
      "step": 213411
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.3823379278182983,
      "learning_rate": 8.015105445197923e-06,
      "loss": 1.8834,
      "step": 213412
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.4399853944778442,
      "learning_rate": 8.014166242946617e-06,
      "loss": 1.8742,
      "step": 213413
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.5223138332366943,
      "learning_rate": 8.013227094980956e-06,
      "loss": 2.0256,
      "step": 213414
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.517194390296936,
      "learning_rate": 8.01228800130117e-06,
      "loss": 1.8667,
      "step": 213415
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.4764522314071655,
      "learning_rate": 8.011348961907393e-06,
      "loss": 2.0602,
      "step": 213416
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.4955207109451294,
      "learning_rate": 8.010409976799825e-06,
      "loss": 1.9488,
      "step": 213417
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.4917776584625244,
      "learning_rate": 8.009471045978665e-06,
      "loss": 1.8547,
      "step": 213418
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.471901535987854,
      "learning_rate": 8.008532169444015e-06,
      "loss": 1.953,
      "step": 213419
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.3524444103240967,
      "learning_rate": 8.007593347196107e-06,
      "loss": 2.0022,
      "step": 213420
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.4788613319396973,
      "learning_rate": 8.006654579235105e-06,
      "loss": 1.8031,
      "step": 213421
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.4055649042129517,
      "learning_rate": 8.005715865561147e-06,
      "loss": 1.9711,
      "step": 213422
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.5032132863998413,
      "learning_rate": 8.004777206174462e-06,
      "loss": 2.1083,
      "step": 213423
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.3693000078201294,
      "learning_rate": 8.003838601075218e-06,
      "loss": 1.8001,
      "step": 213424
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.4199742078781128,
      "learning_rate": 8.002900050263583e-06,
      "loss": 1.9723,
      "step": 213425
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.4022619724273682,
      "learning_rate": 8.001961553739655e-06,
      "loss": 2.0968,
      "step": 213426
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.5227913856506348,
      "learning_rate": 8.001023111503735e-06,
      "loss": 2.247,
      "step": 213427
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.414301872253418,
      "learning_rate": 8.000084723555889e-06,
      "loss": 2.1351,
      "step": 213428
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.4157620668411255,
      "learning_rate": 7.999146389896383e-06,
      "loss": 1.7969,
      "step": 213429
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.506542682647705,
      "learning_rate": 7.998208110525351e-06,
      "loss": 1.8395,
      "step": 213430
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.5202159881591797,
      "learning_rate": 7.997269885442926e-06,
      "loss": 2.0428,
      "step": 213431
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.357994556427002,
      "learning_rate": 7.99633171464934e-06,
      "loss": 2.1387,
      "step": 213432
    },
    {
      "epoch": 2.78,
      "grad_norm": 2.650425434112549,
      "learning_rate": 7.995393598144729e-06,
      "loss": 2.1427,
      "step": 213433
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.4813437461853027,
      "learning_rate": 7.994455535929289e-06,
      "loss": 1.8957,
      "step": 213434
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.423904299736023,
      "learning_rate": 7.993517528003224e-06,
      "loss": 1.9283,
      "step": 213435
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.513920545578003,
      "learning_rate": 7.992579574366665e-06,
      "loss": 2.0852,
      "step": 213436
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.4703434705734253,
      "learning_rate": 7.99164167501981e-06,
      "loss": 1.7723,
      "step": 213437
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.4915064573287964,
      "learning_rate": 7.990703829962763e-06,
      "loss": 2.1325,
      "step": 213438
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.3498367071151733,
      "learning_rate": 7.989766039195822e-06,
      "loss": 1.8932,
      "step": 213439
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.4322478771209717,
      "learning_rate": 7.988828302719052e-06,
      "loss": 1.8827,
      "step": 213440
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.4454859495162964,
      "learning_rate": 7.98789062053269e-06,
      "loss": 2.0894,
      "step": 213441
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.4345122575759888,
      "learning_rate": 7.986952992636896e-06,
      "loss": 1.8361,
      "step": 213442
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.5906572341918945,
      "learning_rate": 7.986015419031843e-06,
      "loss": 2.0009,
      "step": 213443
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.4413660764694214,
      "learning_rate": 7.985077899717663e-06,
      "loss": 2.1478,
      "step": 213444
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.5209031105041504,
      "learning_rate": 7.98414043469462e-06,
      "loss": 1.9748,
      "step": 213445
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.5813640356063843,
      "learning_rate": 7.98320302396278e-06,
      "loss": 2.027,
      "step": 213446
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.3959873914718628,
      "learning_rate": 7.982265667522414e-06,
      "loss": 2.0842,
      "step": 213447
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.4520541429519653,
      "learning_rate": 7.981328365373685e-06,
      "loss": 1.9047,
      "step": 213448
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.5589399337768555,
      "learning_rate": 7.980391117516693e-06,
      "loss": 2.0985,
      "step": 213449
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.4408634901046753,
      "learning_rate": 7.979453923951672e-06,
      "loss": 2.091,
      "step": 213450
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.4708902835845947,
      "learning_rate": 7.97851678467879e-06,
      "loss": 2.1731,
      "step": 213451
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.4063457250595093,
      "learning_rate": 7.977579699698179e-06,
      "loss": 2.1786,
      "step": 213452
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.4239094257354736,
      "learning_rate": 7.97664266901007e-06,
      "loss": 1.9479,
      "step": 213453
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.3193447589874268,
      "learning_rate": 7.97570569261463e-06,
      "loss": 1.847,
      "step": 213454
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.4745906591415405,
      "learning_rate": 7.97476877051203e-06,
      "loss": 2.1088,
      "step": 213455
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.5608683824539185,
      "learning_rate": 7.973831902702398e-06,
      "loss": 2.1792,
      "step": 213456
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.4124667644500732,
      "learning_rate": 7.972895089185972e-06,
      "loss": 1.9634,
      "step": 213457
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.700631856918335,
      "learning_rate": 7.971958329962846e-06,
      "loss": 2.0376,
      "step": 213458
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.4836384057998657,
      "learning_rate": 7.971021625033292e-06,
      "loss": 1.9942,
      "step": 213459
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.3399182558059692,
      "learning_rate": 7.970084974397439e-06,
      "loss": 2.0396,
      "step": 213460
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.365200161933899,
      "learning_rate": 7.96914837805549e-06,
      "loss": 1.8563,
      "step": 213461
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.438227891921997,
      "learning_rate": 7.96821183600751e-06,
      "loss": 1.9845,
      "step": 213462
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.543330430984497,
      "learning_rate": 7.967275348253799e-06,
      "loss": 1.8988,
      "step": 213463
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.540634036064148,
      "learning_rate": 7.966338914794457e-06,
      "loss": 2.3431,
      "step": 213464
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.4795020818710327,
      "learning_rate": 7.965402535629716e-06,
      "loss": 1.948,
      "step": 213465
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.4183330535888672,
      "learning_rate": 7.96446621075968e-06,
      "loss": 1.8052,
      "step": 213466
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.50458562374115,
      "learning_rate": 7.963529940184644e-06,
      "loss": 2.1848,
      "step": 213467
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.487474799156189,
      "learning_rate": 7.96259372390461e-06,
      "loss": 1.812,
      "step": 213468
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.5766013860702515,
      "learning_rate": 7.961657561919876e-06,
      "loss": 1.8183,
      "step": 213469
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.510228157043457,
      "learning_rate": 7.960721454230578e-06,
      "loss": 1.8612,
      "step": 213470
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.4430058002471924,
      "learning_rate": 7.959785400836916e-06,
      "loss": 2.0981,
      "step": 213471
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.3376497030258179,
      "learning_rate": 7.958849401739021e-06,
      "loss": 1.9,
      "step": 213472
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.5534512996673584,
      "learning_rate": 7.957913456937093e-06,
      "loss": 1.9517,
      "step": 213473
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.452637791633606,
      "learning_rate": 7.9569775664313e-06,
      "loss": 2.0757,
      "step": 213474
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.3722553253173828,
      "learning_rate": 7.95604173022184e-06,
      "loss": 2.0391,
      "step": 213475
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.4141923189163208,
      "learning_rate": 7.955105948308816e-06,
      "loss": 1.8227,
      "step": 213476
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.8164788484573364,
      "learning_rate": 7.95417022069249e-06,
      "loss": 1.9023,
      "step": 213477
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.4196619987487793,
      "learning_rate": 7.953234547372967e-06,
      "loss": 1.9119,
      "step": 213478
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.4460545778274536,
      "learning_rate": 7.952298928350476e-06,
      "loss": 1.9505,
      "step": 213479
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.4909870624542236,
      "learning_rate": 7.951363363625185e-06,
      "loss": 1.9179,
      "step": 213480
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.5160939693450928,
      "learning_rate": 7.950427853197228e-06,
      "loss": 2.1062,
      "step": 213481
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.4884809255599976,
      "learning_rate": 7.949492397066771e-06,
      "loss": 1.9043,
      "step": 213482
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.4150140285491943,
      "learning_rate": 7.948556995234045e-06,
      "loss": 2.1182,
      "step": 213483
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.3286588191986084,
      "learning_rate": 7.947621647699187e-06,
      "loss": 1.9907,
      "step": 213484
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.5469081401824951,
      "learning_rate": 7.946686354462395e-06,
      "loss": 1.9654,
      "step": 213485
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.4483771324157715,
      "learning_rate": 7.945751115523802e-06,
      "loss": 2.0802,
      "step": 213486
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.4016143083572388,
      "learning_rate": 7.944815930883608e-06,
      "loss": 1.841,
      "step": 213487
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.491090178489685,
      "learning_rate": 7.943880800542013e-06,
      "loss": 2.1391,
      "step": 213488
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.4295954704284668,
      "learning_rate": 7.94294572449915e-06,
      "loss": 2.0717,
      "step": 213489
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.4761537313461304,
      "learning_rate": 7.942010702755186e-06,
      "loss": 1.9573,
      "step": 213490
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.5566792488098145,
      "learning_rate": 7.941075735310354e-06,
      "loss": 1.9791,
      "step": 213491
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.4800081253051758,
      "learning_rate": 7.940140822164753e-06,
      "loss": 2.0354,
      "step": 213492
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.570099949836731,
      "learning_rate": 7.939205963318618e-06,
      "loss": 1.702,
      "step": 213493
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.5131233930587769,
      "learning_rate": 7.938271158772081e-06,
      "loss": 2.1495,
      "step": 213494
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.4768036603927612,
      "learning_rate": 7.937336408525308e-06,
      "loss": 2.0546,
      "step": 213495
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.4763672351837158,
      "learning_rate": 7.936401712578532e-06,
      "loss": 1.7402,
      "step": 213496
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.357007384300232,
      "learning_rate": 7.935467070931922e-06,
      "loss": 2.1824,
      "step": 213497
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.4619423151016235,
      "learning_rate": 7.934532483585576e-06,
      "loss": 1.9025,
      "step": 213498
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.5987414121627808,
      "learning_rate": 7.933597950539728e-06,
      "loss": 1.9466,
      "step": 213499
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.3632961511611938,
      "learning_rate": 7.932663471794543e-06,
      "loss": 1.8288,
      "step": 213500
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.6249250173568726,
      "learning_rate": 7.931729047350155e-06,
      "loss": 2.0785,
      "step": 213501
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.5079267024993896,
      "learning_rate": 7.930794677206798e-06,
      "loss": 1.9121,
      "step": 213502
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.5906789302825928,
      "learning_rate": 7.929860361364637e-06,
      "loss": 2.1494,
      "step": 213503
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.459800124168396,
      "learning_rate": 7.928926099823808e-06,
      "loss": 1.8866,
      "step": 213504
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.4826687574386597,
      "learning_rate": 7.927991892584507e-06,
      "loss": 2.0191,
      "step": 213505
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.4850856065750122,
      "learning_rate": 7.927057739646903e-06,
      "loss": 1.9372,
      "step": 213506
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.4920698404312134,
      "learning_rate": 7.926123641011161e-06,
      "loss": 1.9845,
      "step": 213507
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.440462589263916,
      "learning_rate": 7.925189596677484e-06,
      "loss": 1.9422,
      "step": 213508
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.4601171016693115,
      "learning_rate": 7.924255606646035e-06,
      "loss": 2.2734,
      "step": 213509
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.3552658557891846,
      "learning_rate": 7.923321670916983e-06,
      "loss": 1.8698,
      "step": 213510
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.3759498596191406,
      "learning_rate": 7.922387789490459e-06,
      "loss": 1.918,
      "step": 213511
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.445921540260315,
      "learning_rate": 7.92145396236673e-06,
      "loss": 2.2142,
      "step": 213512
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.514907717704773,
      "learning_rate": 7.920520189545865e-06,
      "loss": 1.7613,
      "step": 213513
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.477583885192871,
      "learning_rate": 7.919586471028128e-06,
      "loss": 1.9221,
      "step": 213514
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.434110164642334,
      "learning_rate": 7.918652806813619e-06,
      "loss": 2.1766,
      "step": 213515
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.583624005317688,
      "learning_rate": 7.917719196902572e-06,
      "loss": 1.8566,
      "step": 213516
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.379347324371338,
      "learning_rate": 7.91678564129512e-06,
      "loss": 1.761,
      "step": 213517
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.4987789392471313,
      "learning_rate": 7.915852139991463e-06,
      "loss": 1.9023,
      "step": 213518
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.449296236038208,
      "learning_rate": 7.914918692991734e-06,
      "loss": 1.9445,
      "step": 213519
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.8118281364440918,
      "learning_rate": 7.913985300296166e-06,
      "loss": 1.7298,
      "step": 213520
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.8424885272979736,
      "learning_rate": 7.913051961904926e-06,
      "loss": 2.1973,
      "step": 213521
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.423946499824524,
      "learning_rate": 7.912118677818147e-06,
      "loss": 1.8627,
      "step": 213522
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.51422917842865,
      "learning_rate": 7.911185448035962e-06,
      "loss": 2.0515,
      "step": 213523
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.3704737424850464,
      "learning_rate": 7.91025227255867e-06,
      "loss": 1.7756,
      "step": 213524
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.5348342657089233,
      "learning_rate": 7.909319151386307e-06,
      "loss": 1.8978,
      "step": 213525
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.4493935108184814,
      "learning_rate": 7.908386084519169e-06,
      "loss": 2.1512,
      "step": 213526
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.416014313697815,
      "learning_rate": 7.90745307195736e-06,
      "loss": 2.0326,
      "step": 213527
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.5623829364776611,
      "learning_rate": 7.90652011370111e-06,
      "loss": 2.2281,
      "step": 213528
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.5018373727798462,
      "learning_rate": 7.905587209750486e-06,
      "loss": 2.153,
      "step": 213529
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.5087839365005493,
      "learning_rate": 7.904654360105756e-06,
      "loss": 1.8081,
      "step": 213530
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.5952908992767334,
      "learning_rate": 7.903721564767052e-06,
      "loss": 1.9814,
      "step": 213531
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.4551023244857788,
      "learning_rate": 7.902788823734573e-06,
      "loss": 1.902,
      "step": 213532
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.6820979118347168,
      "learning_rate": 7.901856137008456e-06,
      "loss": 1.8781,
      "step": 213533
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.4216175079345703,
      "learning_rate": 7.900923504588964e-06,
      "loss": 1.9699,
      "step": 213534
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.442923665046692,
      "learning_rate": 7.89999092647613e-06,
      "loss": 2.0429,
      "step": 213535
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.5026386976242065,
      "learning_rate": 7.899058402670255e-06,
      "loss": 1.9016,
      "step": 213536
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.5082710981369019,
      "learning_rate": 7.898125933171407e-06,
      "loss": 2.2801,
      "step": 213537
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.6565895080566406,
      "learning_rate": 7.89719351797985e-06,
      "loss": 2.1146,
      "step": 213538
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.5341591835021973,
      "learning_rate": 7.896261157095685e-06,
      "loss": 1.9609,
      "step": 213539
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.4065380096435547,
      "learning_rate": 7.895328850519178e-06,
      "loss": 1.9609,
      "step": 213540
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.4421850442886353,
      "learning_rate": 7.894396598250396e-06,
      "loss": 1.9658,
      "step": 213541
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.583389163017273,
      "learning_rate": 7.893464400289573e-06,
      "loss": 1.9949,
      "step": 213542
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.5700122117996216,
      "learning_rate": 7.892532256636841e-06,
      "loss": 2.0552,
      "step": 213543
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.4440977573394775,
      "learning_rate": 7.891600167292433e-06,
      "loss": 2.1241,
      "step": 213544
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.477402687072754,
      "learning_rate": 7.890668132256483e-06,
      "loss": 2.0626,
      "step": 213545
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.4726492166519165,
      "learning_rate": 7.889736151529224e-06,
      "loss": 2.1676,
      "step": 213546
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.6584582328796387,
      "learning_rate": 7.88880422511069e-06,
      "loss": 1.8111,
      "step": 213547
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.348034381866455,
      "learning_rate": 7.887872353001179e-06,
      "loss": 1.8841,
      "step": 213548
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.3575788736343384,
      "learning_rate": 7.886940535200826e-06,
      "loss": 1.955,
      "step": 213549
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.3391685485839844,
      "learning_rate": 7.886008771709796e-06,
      "loss": 1.9469,
      "step": 213550
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.4463062286376953,
      "learning_rate": 7.885077062528256e-06,
      "loss": 2.2642,
      "step": 213551
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.4378607273101807,
      "learning_rate": 7.884145407656472e-06,
      "loss": 2.0586,
      "step": 213552
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.410892128944397,
      "learning_rate": 7.883213807094446e-06,
      "loss": 2.0301,
      "step": 213553
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.5503615140914917,
      "learning_rate": 7.88228226084251e-06,
      "loss": 1.9383,
      "step": 213554
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.6407352685928345,
      "learning_rate": 7.88135076890073e-06,
      "loss": 1.9119,
      "step": 213555
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.3783471584320068,
      "learning_rate": 7.88041933126934e-06,
      "loss": 2.1911,
      "step": 213556
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.3604317903518677,
      "learning_rate": 7.87948794794847e-06,
      "loss": 2.0123,
      "step": 213557
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.4005357027053833,
      "learning_rate": 7.878556618938326e-06,
      "loss": 1.9313,
      "step": 213558
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.5051478147506714,
      "learning_rate": 7.877625344239103e-06,
      "loss": 1.8318,
      "step": 213559
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.5315533876419067,
      "learning_rate": 7.876694123850936e-06,
      "loss": 1.9637,
      "step": 213560
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.4211753606796265,
      "learning_rate": 7.87576295777399e-06,
      "loss": 1.8768,
      "step": 213561
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.3893643617630005,
      "learning_rate": 7.874831846008467e-06,
      "loss": 1.9476,
      "step": 213562
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.4492584466934204,
      "learning_rate": 7.8739007885545e-06,
      "loss": 2.1649,
      "step": 213563
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.4616929292678833,
      "learning_rate": 7.87296978541232e-06,
      "loss": 2.1403,
      "step": 213564
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.451999306678772,
      "learning_rate": 7.872038836582062e-06,
      "loss": 1.9797,
      "step": 213565
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.3951970338821411,
      "learning_rate": 7.871107942063892e-06,
      "loss": 1.9979,
      "step": 213566
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.3544751405715942,
      "learning_rate": 7.87017710185801e-06,
      "loss": 1.8957,
      "step": 213567
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.4558899402618408,
      "learning_rate": 7.869246315964584e-06,
      "loss": 1.9742,
      "step": 213568
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.6182540655136108,
      "learning_rate": 7.868315584383744e-06,
      "loss": 2.0862,
      "step": 213569
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.5534855127334595,
      "learning_rate": 7.867384907115727e-06,
      "loss": 2.0727,
      "step": 213570
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.4420169591903687,
      "learning_rate": 7.866454284160695e-06,
      "loss": 1.7488,
      "step": 213571
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.3802807331085205,
      "learning_rate": 7.865523715518752e-06,
      "loss": 1.8921,
      "step": 213572
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.4491785764694214,
      "learning_rate": 7.864593201190161e-06,
      "loss": 1.912,
      "step": 213573
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.5335729122161865,
      "learning_rate": 7.863662741175059e-06,
      "loss": 1.9689,
      "step": 213574
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.4537445306777954,
      "learning_rate": 7.862732335473576e-06,
      "loss": 2.0569,
      "step": 213575
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.5587517023086548,
      "learning_rate": 7.861801984085981e-06,
      "loss": 2.0793,
      "step": 213576
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.6582595109939575,
      "learning_rate": 7.86087168701237e-06,
      "loss": 2.0705,
      "step": 213577
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.5058379173278809,
      "learning_rate": 7.859941444252915e-06,
      "loss": 1.7996,
      "step": 213578
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.548911690711975,
      "learning_rate": 7.859011255807845e-06,
      "loss": 1.8142,
      "step": 213579
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.467239260673523,
      "learning_rate": 7.858081121677262e-06,
      "loss": 1.8848,
      "step": 213580
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.5206811428070068,
      "learning_rate": 7.857151041861397e-06,
      "loss": 2.0454,
      "step": 213581
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.4485849142074585,
      "learning_rate": 7.85622101636042e-06,
      "loss": 1.8417,
      "step": 213582
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.488375186920166,
      "learning_rate": 7.855291045174494e-06,
      "loss": 1.7807,
      "step": 213583
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.4339733123779297,
      "learning_rate": 7.854361128303722e-06,
      "loss": 2.1101,
      "step": 213584
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.4995256662368774,
      "learning_rate": 7.8534312657484e-06,
      "loss": 2.1661,
      "step": 213585
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.4789987802505493,
      "learning_rate": 7.852501457508598e-06,
      "loss": 2.111,
      "step": 213586
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.5024415254592896,
      "learning_rate": 7.851571703584547e-06,
      "loss": 2.0783,
      "step": 213587
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.5599462985992432,
      "learning_rate": 7.850642003976415e-06,
      "loss": 2.0264,
      "step": 213588
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.4223313331604004,
      "learning_rate": 7.849712358684368e-06,
      "loss": 1.9438,
      "step": 213589
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.4255245923995972,
      "learning_rate": 7.84878276770854e-06,
      "loss": 1.7733,
      "step": 213590
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.5291961431503296,
      "learning_rate": 7.847853231049161e-06,
      "loss": 1.9216,
      "step": 213591
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.5749417543411255,
      "learning_rate": 7.846923748706368e-06,
      "loss": 1.9654,
      "step": 213592
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.5132120847702026,
      "learning_rate": 7.84599432068036e-06,
      "loss": 1.9723,
      "step": 213593
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.4469490051269531,
      "learning_rate": 7.845064946971268e-06,
      "loss": 1.7932,
      "step": 213594
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.3694928884506226,
      "learning_rate": 7.844135627579363e-06,
      "loss": 1.9385,
      "step": 213595
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.3395652770996094,
      "learning_rate": 7.843206362504705e-06,
      "loss": 1.929,
      "step": 213596
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.542493224143982,
      "learning_rate": 7.8422771517475e-06,
      "loss": 2.0253,
      "step": 213597
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.463104009628296,
      "learning_rate": 7.841347995307944e-06,
      "loss": 2.025,
      "step": 213598
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.4190163612365723,
      "learning_rate": 7.840418893186173e-06,
      "loss": 1.9064,
      "step": 213599
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.4086374044418335,
      "learning_rate": 7.839489845382385e-06,
      "loss": 1.8565,
      "step": 213600
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.598646640777588,
      "learning_rate": 7.838560851896814e-06,
      "loss": 1.8196,
      "step": 213601
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.4240602254867554,
      "learning_rate": 7.837631912729525e-06,
      "loss": 1.7819,
      "step": 213602
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.7710152864456177,
      "learning_rate": 7.83670302788072e-06,
      "loss": 2.0371,
      "step": 213603
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.5560319423675537,
      "learning_rate": 7.835774197350597e-06,
      "loss": 2.0433,
      "step": 213604
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.4494811296463013,
      "learning_rate": 7.834845421139324e-06,
      "loss": 2.0795,
      "step": 213605
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.5379654169082642,
      "learning_rate": 7.833916699247033e-06,
      "loss": 2.2538,
      "step": 213606
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.4940553903579712,
      "learning_rate": 7.832988031674026e-06,
      "loss": 1.7689,
      "step": 213607
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.372385859489441,
      "learning_rate": 7.8320594184203e-06,
      "loss": 1.9769,
      "step": 213608
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.3935457468032837,
      "learning_rate": 7.831130859486123e-06,
      "loss": 1.8259,
      "step": 213609
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.4904814958572388,
      "learning_rate": 7.830202354871662e-06,
      "loss": 2.1097,
      "step": 213610
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.533684253692627,
      "learning_rate": 7.829273904577082e-06,
      "loss": 1.9801,
      "step": 213611
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.4785045385360718,
      "learning_rate": 7.828345508602518e-06,
      "loss": 2.0935,
      "step": 213612
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.470738172531128,
      "learning_rate": 7.827417166948268e-06,
      "loss": 2.0286,
      "step": 213613
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.4387449026107788,
      "learning_rate": 7.826488879614335e-06,
      "loss": 2.0364,
      "step": 213614
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.5350658893585205,
      "learning_rate": 7.825560646601014e-06,
      "loss": 2.194,
      "step": 213615
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.500860571861267,
      "learning_rate": 7.824632467908409e-06,
      "loss": 2.2726,
      "step": 213616
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.4724633693695068,
      "learning_rate": 7.823704343536752e-06,
      "loss": 2.0938,
      "step": 213617
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.729316234588623,
      "learning_rate": 7.822776273486142e-06,
      "loss": 2.0922,
      "step": 213618
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.541481852531433,
      "learning_rate": 7.821848257756847e-06,
      "loss": 2.1873,
      "step": 213619
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.628484845161438,
      "learning_rate": 7.820920296348966e-06,
      "loss": 2.0784,
      "step": 213620
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.475940465927124,
      "learning_rate": 7.819992389262664e-06,
      "loss": 1.9321,
      "step": 213621
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.4708181619644165,
      "learning_rate": 7.819064536498144e-06,
      "loss": 1.812,
      "step": 213622
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.6746103763580322,
      "learning_rate": 7.818136738055603e-06,
      "loss": 2.1302,
      "step": 213623
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.4445704221725464,
      "learning_rate": 7.817208993935142e-06,
      "loss": 1.7034,
      "step": 213624
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.5949122905731201,
      "learning_rate": 7.816281304137063e-06,
      "loss": 2.2383,
      "step": 213625
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.4188904762268066,
      "learning_rate": 7.815353668661394e-06,
      "loss": 2.0598,
      "step": 213626
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.3532449007034302,
      "learning_rate": 7.814426087508373e-06,
      "loss": 1.7763,
      "step": 213627
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.3973255157470703,
      "learning_rate": 7.813498560678133e-06,
      "loss": 1.9868,
      "step": 213628
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.3476049900054932,
      "learning_rate": 7.812571088170905e-06,
      "loss": 2.0206,
      "step": 213629
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.3580251932144165,
      "learning_rate": 7.811643669986823e-06,
      "loss": 1.9602,
      "step": 213630
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.4195457696914673,
      "learning_rate": 7.810716306126152e-06,
      "loss": 1.8165,
      "step": 213631
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.525741457939148,
      "learning_rate": 7.809788996588894e-06,
      "loss": 2.017,
      "step": 213632
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.4406204223632812,
      "learning_rate": 7.808861741375349e-06,
      "loss": 1.923,
      "step": 213633
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.4754632711410522,
      "learning_rate": 7.807934540485616e-06,
      "loss": 1.9986,
      "step": 213634
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.4826675653457642,
      "learning_rate": 7.807007393919929e-06,
      "loss": 2.0772,
      "step": 213635
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.427198052406311,
      "learning_rate": 7.80608030167842e-06,
      "loss": 2.1427,
      "step": 213636
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.4351847171783447,
      "learning_rate": 7.80515326376132e-06,
      "loss": 1.7333,
      "step": 213637
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.5367158651351929,
      "learning_rate": 7.804226280168734e-06,
      "loss": 1.9324,
      "step": 213638
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.48955237865448,
      "learning_rate": 7.80329935090086e-06,
      "loss": 2.0004,
      "step": 213639
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.5521210432052612,
      "learning_rate": 7.80237247595783e-06,
      "loss": 1.8619,
      "step": 213640
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.5018967390060425,
      "learning_rate": 7.80144565533991e-06,
      "loss": 2.0192,
      "step": 213641
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.4539787769317627,
      "learning_rate": 7.800518889047169e-06,
      "loss": 2.1056,
      "step": 213642
    },
    {
      "epoch": 2.78,
      "grad_norm": 2.537055253982544,
      "learning_rate": 7.799592177079872e-06,
      "loss": 1.8861,
      "step": 213643
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.3346853256225586,
      "learning_rate": 7.798665519438118e-06,
      "loss": 1.9541,
      "step": 213644
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.4948073625564575,
      "learning_rate": 7.79773891612211e-06,
      "loss": 1.8374,
      "step": 213645
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.54346764087677,
      "learning_rate": 7.79681236713201e-06,
      "loss": 2.0879,
      "step": 213646
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.5037634372711182,
      "learning_rate": 7.795885872468022e-06,
      "loss": 2.0568,
      "step": 213647
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.5147693157196045,
      "learning_rate": 7.794959432130277e-06,
      "loss": 1.9206,
      "step": 213648
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.2571282386779785,
      "learning_rate": 7.794033046118974e-06,
      "loss": 1.9595,
      "step": 213649
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.4305157661437988,
      "learning_rate": 7.793106714434283e-06,
      "loss": 2.0658,
      "step": 213650
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.6714377403259277,
      "learning_rate": 7.792180437076367e-06,
      "loss": 2.1096,
      "step": 213651
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.5895756483078003,
      "learning_rate": 7.791254214045362e-06,
      "loss": 1.8803,
      "step": 213652
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.4382184743881226,
      "learning_rate": 7.790328045341532e-06,
      "loss": 1.8262,
      "step": 213653
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.6458791494369507,
      "learning_rate": 7.789401930964945e-06,
      "loss": 2.0755,
      "step": 213654
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.3424164056777954,
      "learning_rate": 7.788475870915866e-06,
      "loss": 2.0753,
      "step": 213655
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.4961544275283813,
      "learning_rate": 7.787549865194432e-06,
      "loss": 1.8731,
      "step": 213656
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.5089532136917114,
      "learning_rate": 7.78662391380077e-06,
      "loss": 2.0157,
      "step": 213657
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.451485514640808,
      "learning_rate": 7.78569801673512e-06,
      "loss": 2.0521,
      "step": 213658
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.4534372091293335,
      "learning_rate": 7.78477217399761e-06,
      "loss": 2.2625,
      "step": 213659
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.4310405254364014,
      "learning_rate": 7.783846385588411e-06,
      "loss": 1.9793,
      "step": 213660
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.558911681175232,
      "learning_rate": 7.78292065150775e-06,
      "loss": 2.1435,
      "step": 213661
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.4774055480957031,
      "learning_rate": 7.781994971755767e-06,
      "loss": 2.0073,
      "step": 213662
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.3678830862045288,
      "learning_rate": 7.781069346332557e-06,
      "loss": 1.825,
      "step": 213663
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.6615363359451294,
      "learning_rate": 7.780143775238424e-06,
      "loss": 1.8628,
      "step": 213664
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.4463236331939697,
      "learning_rate": 7.779218258473463e-06,
      "loss": 1.8104,
      "step": 213665
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.5959973335266113,
      "learning_rate": 7.778292796037843e-06,
      "loss": 1.9484,
      "step": 213666
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.5609567165374756,
      "learning_rate": 7.777367387931799e-06,
      "loss": 2.1848,
      "step": 213667
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.5082253217697144,
      "learning_rate": 7.776442034155428e-06,
      "loss": 1.976,
      "step": 213668
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.5557903051376343,
      "learning_rate": 7.77551673470893e-06,
      "loss": 1.8302,
      "step": 213669
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.540067434310913,
      "learning_rate": 7.774591489592474e-06,
      "loss": 1.7942,
      "step": 213670
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.3799649477005005,
      "learning_rate": 7.773666298806226e-06,
      "loss": 1.9794,
      "step": 213671
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.490615963935852,
      "learning_rate": 7.772741162350415e-06,
      "loss": 2.1791,
      "step": 213672
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.418458104133606,
      "learning_rate": 7.771816080225146e-06,
      "loss": 1.984,
      "step": 213673
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.5142916440963745,
      "learning_rate": 7.770891052430617e-06,
      "loss": 2.1338,
      "step": 213674
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.4766111373901367,
      "learning_rate": 7.769966078966961e-06,
      "loss": 2.1366,
      "step": 213675
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.4586173295974731,
      "learning_rate": 7.769041159834443e-06,
      "loss": 1.82,
      "step": 213676
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.4021180868148804,
      "learning_rate": 7.768116295033134e-06,
      "loss": 1.8133,
      "step": 213677
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.4506527185440063,
      "learning_rate": 7.767191484563262e-06,
      "loss": 2.0465,
      "step": 213678
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.3783011436462402,
      "learning_rate": 7.766266728424964e-06,
      "loss": 1.6286,
      "step": 213679
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.3934577703475952,
      "learning_rate": 7.765342026618503e-06,
      "loss": 1.8741,
      "step": 213680
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.5274622440338135,
      "learning_rate": 7.764417379143917e-06,
      "loss": 1.9372,
      "step": 213681
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.52691650390625,
      "learning_rate": 7.763492786001436e-06,
      "loss": 2.0287,
      "step": 213682
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.5110950469970703,
      "learning_rate": 7.762568247191258e-06,
      "loss": 1.8279,
      "step": 213683
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.5823898315429688,
      "learning_rate": 7.761643762713555e-06,
      "loss": 2.0309,
      "step": 213684
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.4620193243026733,
      "learning_rate": 7.760719332568422e-06,
      "loss": 1.9653,
      "step": 213685
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.5087380409240723,
      "learning_rate": 7.75979495675616e-06,
      "loss": 2.1654,
      "step": 213686
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.4044593572616577,
      "learning_rate": 7.758870635276837e-06,
      "loss": 2.1902,
      "step": 213687
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.7430877685546875,
      "learning_rate": 7.757946368130653e-06,
      "loss": 1.9812,
      "step": 213688
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.3041486740112305,
      "learning_rate": 7.75702215531777e-06,
      "loss": 1.9122,
      "step": 213689
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.5527422428131104,
      "learning_rate": 7.756097996838395e-06,
      "loss": 1.9791,
      "step": 213690
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.3572653532028198,
      "learning_rate": 7.755173892692656e-06,
      "loss": 1.9024,
      "step": 213691
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.620172142982483,
      "learning_rate": 7.754249842880788e-06,
      "loss": 1.9115,
      "step": 213692
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.5773682594299316,
      "learning_rate": 7.753325847402891e-06,
      "loss": 1.8998,
      "step": 213693
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.4742376804351807,
      "learning_rate": 7.752401906259165e-06,
      "loss": 2.0716,
      "step": 213694
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.464128851890564,
      "learning_rate": 7.751478019449776e-06,
      "loss": 1.9407,
      "step": 213695
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.5005282163619995,
      "learning_rate": 7.750554186974923e-06,
      "loss": 1.9122,
      "step": 213696
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.39057195186615,
      "learning_rate": 7.74963040883474e-06,
      "loss": 2.0417,
      "step": 213697
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.3241218328475952,
      "learning_rate": 7.74870668502946e-06,
      "loss": 2.074,
      "step": 213698
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.4689960479736328,
      "learning_rate": 7.747783015559183e-06,
      "loss": 1.8507,
      "step": 213699
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.3885141611099243,
      "learning_rate": 7.74685940042411e-06,
      "loss": 1.8972,
      "step": 213700
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.425706148147583,
      "learning_rate": 7.745935839624406e-06,
      "loss": 1.9103,
      "step": 213701
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.5372976064682007,
      "learning_rate": 7.74501233316024e-06,
      "loss": 2.1393,
      "step": 213702
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.494776964187622,
      "learning_rate": 7.744088881031808e-06,
      "loss": 2.1001,
      "step": 213703
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.359839916229248,
      "learning_rate": 7.743165483239311e-06,
      "loss": 1.8961,
      "step": 213704
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.4890023469924927,
      "learning_rate": 7.742242139782784e-06,
      "loss": 1.9608,
      "step": 213705
    },
    {
      "epoch": 2.78,
      "grad_norm": 2.35552716255188,
      "learning_rate": 7.741318850662559e-06,
      "loss": 1.9308,
      "step": 213706
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.4987586736679077,
      "learning_rate": 7.740395615878702e-06,
      "loss": 2.0132,
      "step": 213707
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.4589827060699463,
      "learning_rate": 7.739472435431448e-06,
      "loss": 1.8964,
      "step": 213708
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.3635982275009155,
      "learning_rate": 7.73854930932093e-06,
      "loss": 1.9689,
      "step": 213709
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.4564173221588135,
      "learning_rate": 7.737626237547378e-06,
      "loss": 2.1073,
      "step": 213710
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.5006555318832397,
      "learning_rate": 7.736703220110862e-06,
      "loss": 1.7731,
      "step": 213711
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.4147443771362305,
      "learning_rate": 7.735780257011614e-06,
      "loss": 1.9016,
      "step": 213712
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.4544669389724731,
      "learning_rate": 7.7348573482498e-06,
      "loss": 1.9376,
      "step": 213713
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.4156299829483032,
      "learning_rate": 7.733934493825623e-06,
      "loss": 1.8852,
      "step": 213714
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.4538984298706055,
      "learning_rate": 7.733011693739178e-06,
      "loss": 1.9768,
      "step": 213715
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.576330542564392,
      "learning_rate": 7.732088947990766e-06,
      "loss": 2.0277,
      "step": 213716
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.467380404472351,
      "learning_rate": 7.73116625658039e-06,
      "loss": 1.9337,
      "step": 213717
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.4850257635116577,
      "learning_rate": 7.730243619508348e-06,
      "loss": 2.1857,
      "step": 213718
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.4233306646347046,
      "learning_rate": 7.729321036774738e-06,
      "loss": 2.0273,
      "step": 213719
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.4461852312088013,
      "learning_rate": 7.728398508379797e-06,
      "loss": 1.8286,
      "step": 213720
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.5132032632827759,
      "learning_rate": 7.727476034323655e-06,
      "loss": 1.8546,
      "step": 213721
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.4366055727005005,
      "learning_rate": 7.72655361460648e-06,
      "loss": 1.8778,
      "step": 213722
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.4465020895004272,
      "learning_rate": 7.725631249228504e-06,
      "loss": 1.9669,
      "step": 213723
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.5414806604385376,
      "learning_rate": 7.724708938189828e-06,
      "loss": 1.8869,
      "step": 213724
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.6103698015213013,
      "learning_rate": 7.723786681490585e-06,
      "loss": 1.9895,
      "step": 213725
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.4132291078567505,
      "learning_rate": 7.722864479131075e-06,
      "loss": 2.203,
      "step": 213726
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.577209234237671,
      "learning_rate": 7.721942331111364e-06,
      "loss": 2.0583,
      "step": 213727
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.3891047239303589,
      "learning_rate": 7.721020237431686e-06,
      "loss": 1.8778,
      "step": 213728
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.4505956172943115,
      "learning_rate": 7.720098198092172e-06,
      "loss": 1.8729,
      "step": 213729
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.3932987451553345,
      "learning_rate": 7.719176213093059e-06,
      "loss": 2.1309,
      "step": 213730
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.5127424001693726,
      "learning_rate": 7.718254282434377e-06,
      "loss": 1.8774,
      "step": 213731
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.491943359375,
      "learning_rate": 7.717332406116461e-06,
      "loss": 2.0792,
      "step": 213732
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.4625563621520996,
      "learning_rate": 7.716410584139376e-06,
      "loss": 1.8371,
      "step": 213733
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.4016132354736328,
      "learning_rate": 7.715488816503357e-06,
      "loss": 1.8834,
      "step": 213734
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.3258941173553467,
      "learning_rate": 7.714567103208536e-06,
      "loss": 1.9859,
      "step": 213735
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.4506837129592896,
      "learning_rate": 7.713645444255079e-06,
      "loss": 2.0806,
      "step": 213736
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.6674443483352661,
      "learning_rate": 7.712723839643153e-06,
      "loss": 1.9079,
      "step": 213737
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.6000754833221436,
      "learning_rate": 7.711802289372992e-06,
      "loss": 1.9653,
      "step": 213738
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.5103107690811157,
      "learning_rate": 7.710880793444696e-06,
      "loss": 1.9144,
      "step": 213739
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.4644310474395752,
      "learning_rate": 7.709959351858464e-06,
      "loss": 1.8716,
      "step": 213740
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.414533257484436,
      "learning_rate": 7.709037964614495e-06,
      "loss": 2.1546,
      "step": 213741
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.3683454990386963,
      "learning_rate": 7.708116631712924e-06,
      "loss": 1.8888,
      "step": 213742
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.8230443000793457,
      "learning_rate": 7.707195353153917e-06,
      "loss": 2.0768,
      "step": 213743
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.5689202547073364,
      "learning_rate": 7.706274128937673e-06,
      "loss": 1.8991,
      "step": 213744
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.5353972911834717,
      "learning_rate": 7.705352959064326e-06,
      "loss": 1.8769,
      "step": 213745
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.4971755743026733,
      "learning_rate": 7.704431843534108e-06,
      "loss": 2.0829,
      "step": 213746
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.5704976320266724,
      "learning_rate": 7.703510782347156e-06,
      "loss": 1.8885,
      "step": 213747
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.6002025604248047,
      "learning_rate": 7.702589775503598e-06,
      "loss": 1.7752,
      "step": 213748
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.6228721141815186,
      "learning_rate": 7.70166882300367e-06,
      "loss": 1.8207,
      "step": 213749
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.5125856399536133,
      "learning_rate": 7.700747924847539e-06,
      "loss": 1.8625,
      "step": 213750
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.4468834400177002,
      "learning_rate": 7.699827081035304e-06,
      "loss": 1.9876,
      "step": 213751
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.3244653940200806,
      "learning_rate": 7.698906291567231e-06,
      "loss": 1.8411,
      "step": 213752
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.3995476961135864,
      "learning_rate": 7.697985556443453e-06,
      "loss": 1.9242,
      "step": 213753
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.5325453281402588,
      "learning_rate": 7.697064875664105e-06,
      "loss": 1.977,
      "step": 213754
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.6283148527145386,
      "learning_rate": 7.696144249229419e-06,
      "loss": 1.8469,
      "step": 213755
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.499839425086975,
      "learning_rate": 7.695223677139495e-06,
      "loss": 1.9287,
      "step": 213756
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.313619613647461,
      "learning_rate": 7.694303159394599e-06,
      "loss": 1.7772,
      "step": 213757
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.4590364694595337,
      "learning_rate": 7.693382695994832e-06,
      "loss": 2.0852,
      "step": 213758
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.5596017837524414,
      "learning_rate": 7.692462286940392e-06,
      "loss": 1.7966,
      "step": 213759
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.543365478515625,
      "learning_rate": 7.691541932231415e-06,
      "loss": 1.8721,
      "step": 213760
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.4025068283081055,
      "learning_rate": 7.6906216318681e-06,
      "loss": 2.0002,
      "step": 213761
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.3865543603897095,
      "learning_rate": 7.68970138585061e-06,
      "loss": 1.9541,
      "step": 213762
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.5956610441207886,
      "learning_rate": 7.68878119417915e-06,
      "loss": 1.9252,
      "step": 213763
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.439420223236084,
      "learning_rate": 7.68786105685385e-06,
      "loss": 1.9973,
      "step": 213764
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.4024637937545776,
      "learning_rate": 7.686940973874912e-06,
      "loss": 1.8157,
      "step": 213765
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.631168007850647,
      "learning_rate": 7.686020945242466e-06,
      "loss": 2.2045,
      "step": 213766
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.4542936086654663,
      "learning_rate": 7.685100970956714e-06,
      "loss": 1.8402,
      "step": 213767
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.4567630290985107,
      "learning_rate": 7.684181051017824e-06,
      "loss": 2.0381,
      "step": 213768
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.4095507860183716,
      "learning_rate": 7.68326118542596e-06,
      "loss": 1.8157,
      "step": 213769
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.742215871810913,
      "learning_rate": 7.682341374181256e-06,
      "loss": 2.2491,
      "step": 213770
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.7150492668151855,
      "learning_rate": 7.681421617284012e-06,
      "loss": 1.9378,
      "step": 213771
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.527745008468628,
      "learning_rate": 7.680501914734227e-06,
      "loss": 2.0329,
      "step": 213772
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.472562551498413,
      "learning_rate": 7.679582266532203e-06,
      "loss": 2.0193,
      "step": 213773
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.4904005527496338,
      "learning_rate": 7.678662672678038e-06,
      "loss": 1.7768,
      "step": 213774
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.5391509532928467,
      "learning_rate": 7.677743133171931e-06,
      "loss": 1.9863,
      "step": 213775
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.4694658517837524,
      "learning_rate": 7.676823648014052e-06,
      "loss": 1.8015,
      "step": 213776
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.5322445631027222,
      "learning_rate": 7.675904217204597e-06,
      "loss": 1.8028,
      "step": 213777
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.2840626239776611,
      "learning_rate": 7.67498484074367e-06,
      "loss": 1.7979,
      "step": 213778
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.4007664918899536,
      "learning_rate": 7.6740655186315e-06,
      "loss": 1.7979,
      "step": 213779
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.3487564325332642,
      "learning_rate": 7.673146250868223e-06,
      "loss": 1.6741,
      "step": 213780
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.468464970588684,
      "learning_rate": 7.67222703745407e-06,
      "loss": 1.9638,
      "step": 213781
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.3933643102645874,
      "learning_rate": 7.67130787838911e-06,
      "loss": 1.9319,
      "step": 213782
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.5029041767120361,
      "learning_rate": 7.670388773673642e-06,
      "loss": 2.2198,
      "step": 213783
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.3850229978561401,
      "learning_rate": 7.669469723307697e-06,
      "loss": 1.8911,
      "step": 213784
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.3034063577651978,
      "learning_rate": 7.668550727291545e-06,
      "loss": 2.1083,
      "step": 213785
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.530078649520874,
      "learning_rate": 7.667631785625317e-06,
      "loss": 2.2196,
      "step": 213786
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.3971585035324097,
      "learning_rate": 7.666712898309213e-06,
      "loss": 1.8051,
      "step": 213787
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.4316823482513428,
      "learning_rate": 7.665794065343334e-06,
      "loss": 2.0334,
      "step": 213788
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.4997650384902954,
      "learning_rate": 7.664875286727978e-06,
      "loss": 2.0186,
      "step": 213789
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.6456327438354492,
      "learning_rate": 7.663956562463214e-06,
      "loss": 2.0082,
      "step": 213790
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.3905094861984253,
      "learning_rate": 7.663037892549239e-06,
      "loss": 1.801,
      "step": 213791
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.5410569906234741,
      "learning_rate": 7.66211927698619e-06,
      "loss": 1.8863,
      "step": 213792
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.5012153387069702,
      "learning_rate": 7.661200715774296e-06,
      "loss": 2.0215,
      "step": 213793
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.427079439163208,
      "learning_rate": 7.660282208913693e-06,
      "loss": 1.9656,
      "step": 213794
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.4220287799835205,
      "learning_rate": 7.65936375640458e-06,
      "loss": 1.8325,
      "step": 213795
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.3810726404190063,
      "learning_rate": 7.65844535824709e-06,
      "loss": 1.7383,
      "step": 213796
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.5456421375274658,
      "learning_rate": 7.657527014441456e-06,
      "loss": 2.2082,
      "step": 213797
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.4618098735809326,
      "learning_rate": 7.656608724987745e-06,
      "loss": 2.0256,
      "step": 213798
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.4750282764434814,
      "learning_rate": 7.655690489886224e-06,
      "loss": 1.8359,
      "step": 213799
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.3247325420379639,
      "learning_rate": 7.654772309136992e-06,
      "loss": 2.0015,
      "step": 213800
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.3695299625396729,
      "learning_rate": 7.653854182740315e-06,
      "loss": 1.8562,
      "step": 213801
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.4153467416763306,
      "learning_rate": 7.652936110696296e-06,
      "loss": 2.0019,
      "step": 213802
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.4250879287719727,
      "learning_rate": 7.652018093005097e-06,
      "loss": 1.9774,
      "step": 213803
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.6038373708724976,
      "learning_rate": 7.651100129666887e-06,
      "loss": 2.1739,
      "step": 213804
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.6191587448120117,
      "learning_rate": 7.650182220681867e-06,
      "loss": 2.1498,
      "step": 213805
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.6114476919174194,
      "learning_rate": 7.6492643660502e-06,
      "loss": 1.7741,
      "step": 213806
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.4263797998428345,
      "learning_rate": 7.648346565772057e-06,
      "loss": 1.96,
      "step": 213807
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.342348337173462,
      "learning_rate": 7.647428819847634e-06,
      "loss": 1.882,
      "step": 213808
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.4011579751968384,
      "learning_rate": 7.646511128277033e-06,
      "loss": 2.1689,
      "step": 213809
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.4379011392593384,
      "learning_rate": 7.645593491060454e-06,
      "loss": 2.0267,
      "step": 213810
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.5186465978622437,
      "learning_rate": 7.644675908198128e-06,
      "loss": 1.9893,
      "step": 213811
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.5970269441604614,
      "learning_rate": 7.643758379690123e-06,
      "loss": 1.9475,
      "step": 213812
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.571751356124878,
      "learning_rate": 7.642840905536673e-06,
      "loss": 1.9319,
      "step": 213813
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.4074676036834717,
      "learning_rate": 7.641923485737977e-06,
      "loss": 1.9091,
      "step": 213814
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.4700649976730347,
      "learning_rate": 7.641006120294135e-06,
      "loss": 1.9202,
      "step": 213815
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.3934855461120605,
      "learning_rate": 7.640088809205347e-06,
      "loss": 2.0566,
      "step": 213816
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.393120288848877,
      "learning_rate": 7.639171552471778e-06,
      "loss": 1.9158,
      "step": 213817
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.435165524482727,
      "learning_rate": 7.638254350093632e-06,
      "loss": 1.919,
      "step": 213818
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.4496715068817139,
      "learning_rate": 7.637337202071036e-06,
      "loss": 2.0364,
      "step": 213819
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.4653347730636597,
      "learning_rate": 7.636420108404195e-06,
      "loss": 1.9882,
      "step": 213820
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.3796515464782715,
      "learning_rate": 7.635503069093241e-06,
      "loss": 1.7421,
      "step": 213821
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.5254786014556885,
      "learning_rate": 7.634586084138373e-06,
      "loss": 1.8339,
      "step": 213822
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.3823833465576172,
      "learning_rate": 7.633669153539756e-06,
      "loss": 2.1583,
      "step": 213823
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.5546183586120605,
      "learning_rate": 7.632752277297528e-06,
      "loss": 2.1074,
      "step": 213824
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.500970482826233,
      "learning_rate": 7.63183545541195e-06,
      "loss": 1.8906,
      "step": 213825
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.4295625686645508,
      "learning_rate": 7.630918687883093e-06,
      "loss": 1.9608,
      "step": 213826
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.4289231300354004,
      "learning_rate": 7.630001974711186e-06,
      "loss": 1.9175,
      "step": 213827
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.4919993877410889,
      "learning_rate": 7.629085315896366e-06,
      "loss": 1.9634,
      "step": 213828
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.4404339790344238,
      "learning_rate": 7.6281687114388314e-06,
      "loss": 1.9989,
      "step": 213829
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.478764533996582,
      "learning_rate": 7.627252161338715e-06,
      "loss": 1.7644,
      "step": 213830
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.3993264436721802,
      "learning_rate": 7.626335665596217e-06,
      "loss": 1.7209,
      "step": 213831
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.421238660812378,
      "learning_rate": 7.625419224211537e-06,
      "loss": 2.1989,
      "step": 213832
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.4086025953292847,
      "learning_rate": 7.624502837184743e-06,
      "loss": 2.0224,
      "step": 213833
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.4611684083938599,
      "learning_rate": 7.623586504516132e-06,
      "loss": 2.0317,
      "step": 213834
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.467047929763794,
      "learning_rate": 7.622670226205807e-06,
      "loss": 2.1991,
      "step": 213835
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.5740364789962769,
      "learning_rate": 7.621754002253932e-06,
      "loss": 1.9482,
      "step": 213836
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.4747860431671143,
      "learning_rate": 7.6208378326607085e-06,
      "loss": 1.9126,
      "step": 213837
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.4208943843841553,
      "learning_rate": 7.619921717426303e-06,
      "loss": 1.9504,
      "step": 213838
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.512127161026001,
      "learning_rate": 7.619005656550814e-06,
      "loss": 1.9426,
      "step": 213839
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.4866876602172852,
      "learning_rate": 7.618089650034509e-06,
      "loss": 2.1759,
      "step": 213840
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.5854188203811646,
      "learning_rate": 7.617173697877521e-06,
      "loss": 1.9382,
      "step": 213841
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.6424219608306885,
      "learning_rate": 7.616257800080017e-06,
      "loss": 1.8533,
      "step": 213842
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.3599148988723755,
      "learning_rate": 7.615341956642196e-06,
      "loss": 1.928,
      "step": 213843
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.481601357460022,
      "learning_rate": 7.614426167564158e-06,
      "loss": 1.859,
      "step": 213844
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.4293670654296875,
      "learning_rate": 7.613510432846137e-06,
      "loss": 2.1299,
      "step": 213845
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.5665109157562256,
      "learning_rate": 7.6125947524882994e-06,
      "loss": 1.8482,
      "step": 213846
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.5827475786209106,
      "learning_rate": 7.611679126490744e-06,
      "loss": 1.9923,
      "step": 213847
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.497504472732544,
      "learning_rate": 7.610763554853738e-06,
      "loss": 1.9216,
      "step": 213848
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.7735373973846436,
      "learning_rate": 7.609848037577415e-06,
      "loss": 2.1193,
      "step": 213849
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.4411633014678955,
      "learning_rate": 7.608932574661941e-06,
      "loss": 1.9338,
      "step": 213850
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.55479097366333,
      "learning_rate": 7.608017166107483e-06,
      "loss": 1.9263,
      "step": 213851
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.4675849676132202,
      "learning_rate": 7.607101811914207e-06,
      "loss": 2.0223,
      "step": 213852
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.472887635231018,
      "learning_rate": 7.606186512082279e-06,
      "loss": 2.0921,
      "step": 213853
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.4047796726226807,
      "learning_rate": 7.605271266611901e-06,
      "loss": 1.8357,
      "step": 213854
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.4548879861831665,
      "learning_rate": 7.604356075503204e-06,
      "loss": 2.1711,
      "step": 213855
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.4917525053024292,
      "learning_rate": 7.6034409387563885e-06,
      "loss": 2.1499,
      "step": 213856
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.4719874858856201,
      "learning_rate": 7.6025258563716216e-06,
      "loss": 1.7283,
      "step": 213857
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.543315052986145,
      "learning_rate": 7.601610828349069e-06,
      "loss": 1.8757,
      "step": 213858
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.551832914352417,
      "learning_rate": 7.600695854688865e-06,
      "loss": 1.9111,
      "step": 213859
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.6154712438583374,
      "learning_rate": 7.599780935391242e-06,
      "loss": 1.9259,
      "step": 213860
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.4057236909866333,
      "learning_rate": 7.5988660704563e-06,
      "loss": 1.9091,
      "step": 213861
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.5063778162002563,
      "learning_rate": 7.597951259884272e-06,
      "loss": 2.2155,
      "step": 213862
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.3810367584228516,
      "learning_rate": 7.597036503675324e-06,
      "loss": 1.9079,
      "step": 213863
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.349938154220581,
      "learning_rate": 7.5961218018295915e-06,
      "loss": 1.84,
      "step": 213864
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.457423448562622,
      "learning_rate": 7.595207154347238e-06,
      "loss": 1.9533,
      "step": 213865
    },
    {
      "epoch": 2.78,
      "grad_norm": 2.3873610496520996,
      "learning_rate": 7.594292561228466e-06,
      "loss": 2.2209,
      "step": 213866
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.3519282341003418,
      "learning_rate": 7.59337802247344e-06,
      "loss": 1.8829,
      "step": 213867
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.4684994220733643,
      "learning_rate": 7.592463538082328e-06,
      "loss": 1.7993,
      "step": 213868
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.3793145418167114,
      "learning_rate": 7.591549108055328e-06,
      "loss": 1.9569,
      "step": 213869
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.3775354623794556,
      "learning_rate": 7.590634732392542e-06,
      "loss": 2.0441,
      "step": 213870
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.4117072820663452,
      "learning_rate": 7.589720411094169e-06,
      "loss": 1.9778,
      "step": 213871
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.562490463256836,
      "learning_rate": 7.588806144160409e-06,
      "loss": 2.1467,
      "step": 213872
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.486684799194336,
      "learning_rate": 7.587891931591361e-06,
      "loss": 2.0149,
      "step": 213873
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.3655725717544556,
      "learning_rate": 7.586977773387292e-06,
      "loss": 1.9082,
      "step": 213874
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.5302566289901733,
      "learning_rate": 7.5860636695483364e-06,
      "loss": 2.1039,
      "step": 213875
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.621315836906433,
      "learning_rate": 7.585149620074627e-06,
      "loss": 1.8603,
      "step": 213876
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.6329315900802612,
      "learning_rate": 7.584235624966328e-06,
      "loss": 2.0146,
      "step": 213877
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.4708482027053833,
      "learning_rate": 7.583321684223708e-06,
      "loss": 1.8437,
      "step": 213878
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.620996117591858,
      "learning_rate": 7.5824077978468015e-06,
      "loss": 2.1469,
      "step": 213879
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.370802640914917,
      "learning_rate": 7.581493965835872e-06,
      "loss": 1.9158,
      "step": 213880
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.3257180452346802,
      "learning_rate": 7.580580188191088e-06,
      "loss": 1.8131,
      "step": 213881
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.784319281578064,
      "learning_rate": 7.579666464912581e-06,
      "loss": 1.9885,
      "step": 213882
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.5786296129226685,
      "learning_rate": 7.578752796000487e-06,
      "loss": 2.117,
      "step": 213883
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.3529616594314575,
      "learning_rate": 7.5778391814550695e-06,
      "loss": 1.8708,
      "step": 213884
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.306552767753601,
      "learning_rate": 7.576925621276431e-06,
      "loss": 1.9171,
      "step": 213885
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.5259159803390503,
      "learning_rate": 7.576012115464769e-06,
      "loss": 1.933,
      "step": 213886
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.4982720613479614,
      "learning_rate": 7.575098664020285e-06,
      "loss": 2.053,
      "step": 213887
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.4504643678665161,
      "learning_rate": 7.574185266943078e-06,
      "loss": 2.027,
      "step": 213888
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.4865615367889404,
      "learning_rate": 7.5732719242333154e-06,
      "loss": 1.9916,
      "step": 213889
    },
    {
      "epoch": 2.78,
      "grad_norm": 1.4096423387527466,
      "learning_rate": 7.572358635891262e-06,
      "loss": 2.0046,
      "step": 213890
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.4810664653778076,
      "learning_rate": 7.571445401916954e-06,
      "loss": 2.2208,
      "step": 213891
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.4237760305404663,
      "learning_rate": 7.570532222310688e-06,
      "loss": 1.8538,
      "step": 213892
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.5130735635757446,
      "learning_rate": 7.569619097072599e-06,
      "loss": 1.8739,
      "step": 213893
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.5003268718719482,
      "learning_rate": 7.56870602620282e-06,
      "loss": 2.0803,
      "step": 213894
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.6127451658248901,
      "learning_rate": 7.567793009701517e-06,
      "loss": 2.0332,
      "step": 213895
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.5436731576919556,
      "learning_rate": 7.566880047568891e-06,
      "loss": 1.8734,
      "step": 213896
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.5334900617599487,
      "learning_rate": 7.565967139805074e-06,
      "loss": 2.0423,
      "step": 213897
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.646040439605713,
      "learning_rate": 7.5650542864102996e-06,
      "loss": 2.2097,
      "step": 213898
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.4038795232772827,
      "learning_rate": 7.564141487384701e-06,
      "loss": 2.0099,
      "step": 213899
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.5432764291763306,
      "learning_rate": 7.563228742728477e-06,
      "loss": 2.089,
      "step": 213900
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.4572035074234009,
      "learning_rate": 7.562316052441697e-06,
      "loss": 1.7641,
      "step": 213901
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.3671729564666748,
      "learning_rate": 7.561403416524659e-06,
      "loss": 2.0206,
      "step": 213902
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.5148452520370483,
      "learning_rate": 7.560490834977429e-06,
      "loss": 2.2288,
      "step": 213903
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.4471372365951538,
      "learning_rate": 7.559578307800274e-06,
      "loss": 2.0375,
      "step": 213904
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.5653871297836304,
      "learning_rate": 7.558665834993294e-06,
      "loss": 2.1417,
      "step": 213905
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.5902049541473389,
      "learning_rate": 7.557753416556689e-06,
      "loss": 1.8721,
      "step": 213906
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.5174158811569214,
      "learning_rate": 7.556841052490592e-06,
      "loss": 2.1328,
      "step": 213907
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.603409767150879,
      "learning_rate": 7.555928742795236e-06,
      "loss": 1.9592,
      "step": 213908
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.5544897317886353,
      "learning_rate": 7.5550164874706884e-06,
      "loss": 1.9991,
      "step": 213909
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.4444628953933716,
      "learning_rate": 7.554104286517248e-06,
      "loss": 2.1976,
      "step": 213910
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.4048893451690674,
      "learning_rate": 7.553192139935016e-06,
      "loss": 1.904,
      "step": 213911
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.4516232013702393,
      "learning_rate": 7.552280047724157e-06,
      "loss": 2.078,
      "step": 213912
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.5166083574295044,
      "learning_rate": 7.551368009884806e-06,
      "loss": 1.9461,
      "step": 213913
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.486088514328003,
      "learning_rate": 7.5504560264172284e-06,
      "loss": 1.9654,
      "step": 213914
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.4278721809387207,
      "learning_rate": 7.549544097321492e-06,
      "loss": 2.2388,
      "step": 213915
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.4688472747802734,
      "learning_rate": 7.5486322225978615e-06,
      "loss": 1.9063,
      "step": 213916
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.3936846256256104,
      "learning_rate": 7.547720402246437e-06,
      "loss": 1.8965,
      "step": 213917
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.464491605758667,
      "learning_rate": 7.546808636267421e-06,
      "loss": 1.9874,
      "step": 213918
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.748287558555603,
      "learning_rate": 7.5458969246609765e-06,
      "loss": 2.0683,
      "step": 213919
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.4943989515304565,
      "learning_rate": 7.544985267427273e-06,
      "loss": 1.9221,
      "step": 213920
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.5629324913024902,
      "learning_rate": 7.544073664566441e-06,
      "loss": 2.1286,
      "step": 213921
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.5948753356933594,
      "learning_rate": 7.543162116078749e-06,
      "loss": 2.1773,
      "step": 213922
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.6869219541549683,
      "learning_rate": 7.542250621964229e-06,
      "loss": 1.9547,
      "step": 213923
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.5339953899383545,
      "learning_rate": 7.541339182223183e-06,
      "loss": 2.0292,
      "step": 213924
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.6691499948501587,
      "learning_rate": 7.540427796855708e-06,
      "loss": 1.9644,
      "step": 213925
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.421622633934021,
      "learning_rate": 7.539516465862005e-06,
      "loss": 1.9207,
      "step": 213926
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.4585514068603516,
      "learning_rate": 7.538605189242175e-06,
      "loss": 2.0451,
      "step": 213927
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.587886095046997,
      "learning_rate": 7.537693966996483e-06,
      "loss": 1.8073,
      "step": 213928
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.9944583177566528,
      "learning_rate": 7.536782799125029e-06,
      "loss": 2.1024,
      "step": 213929
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.3201655149459839,
      "learning_rate": 7.535871685628048e-06,
      "loss": 1.8316,
      "step": 213930
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.5535523891448975,
      "learning_rate": 7.534960626505637e-06,
      "loss": 2.1122,
      "step": 213931
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.6551157236099243,
      "learning_rate": 7.5340496217579985e-06,
      "loss": 1.9524,
      "step": 213932
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.5395780801773071,
      "learning_rate": 7.533138671385297e-06,
      "loss": 2.1115,
      "step": 213933
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.6845598220825195,
      "learning_rate": 7.532227775387734e-06,
      "loss": 1.8235,
      "step": 213934
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.657212495803833,
      "learning_rate": 7.5313169337654075e-06,
      "loss": 2.0639,
      "step": 213935
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.4852861166000366,
      "learning_rate": 7.530406146518586e-06,
      "loss": 2.0731,
      "step": 213936
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.4901272058486938,
      "learning_rate": 7.5294954136473355e-06,
      "loss": 2.0749,
      "step": 213937
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.3989473581314087,
      "learning_rate": 7.528584735151888e-06,
      "loss": 2.1307,
      "step": 213938
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.4344924688339233,
      "learning_rate": 7.5276741110324114e-06,
      "loss": 2.0067,
      "step": 213939
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.4072494506835938,
      "learning_rate": 7.5267635412890045e-06,
      "loss": 1.9904,
      "step": 213940
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.5232279300689697,
      "learning_rate": 7.525853025921969e-06,
      "loss": 2.0792,
      "step": 213941
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.5104095935821533,
      "learning_rate": 7.524942564931369e-06,
      "loss": 2.0931,
      "step": 213942
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.5107336044311523,
      "learning_rate": 7.5240321583174055e-06,
      "loss": 2.1403,
      "step": 213943
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.5576660633087158,
      "learning_rate": 7.523121806080212e-06,
      "loss": 1.9254,
      "step": 213944
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.4456909894943237,
      "learning_rate": 7.5222115082200214e-06,
      "loss": 1.8819,
      "step": 213945
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.4339382648468018,
      "learning_rate": 7.521301264736968e-06,
      "loss": 2.1736,
      "step": 213946
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.3804125785827637,
      "learning_rate": 7.520391075631216e-06,
      "loss": 1.8054,
      "step": 213947
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.476333498954773,
      "learning_rate": 7.519480940902966e-06,
      "loss": 2.0999,
      "step": 213948
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.3623231649398804,
      "learning_rate": 7.518570860552387e-06,
      "loss": 2.0275,
      "step": 213949
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.5557900667190552,
      "learning_rate": 7.5176608345795756e-06,
      "loss": 1.8216,
      "step": 213950
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.4906904697418213,
      "learning_rate": 7.516750862984766e-06,
      "loss": 1.9736,
      "step": 213951
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.5035815238952637,
      "learning_rate": 7.515840945768093e-06,
      "loss": 1.9943,
      "step": 213952
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.5435513257980347,
      "learning_rate": 7.514931082929787e-06,
      "loss": 2.0704,
      "step": 213953
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.4908758401870728,
      "learning_rate": 7.514021274469983e-06,
      "loss": 2.016,
      "step": 213954
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.375205159187317,
      "learning_rate": 7.513111520388815e-06,
      "loss": 2.0009,
      "step": 213955
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.5499258041381836,
      "learning_rate": 7.512201820686481e-06,
      "loss": 2.0625,
      "step": 213956
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.5209065675735474,
      "learning_rate": 7.511292175363148e-06,
      "loss": 2.0165,
      "step": 213957
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.5515484809875488,
      "learning_rate": 7.510382584418984e-06,
      "loss": 1.8788,
      "step": 213958
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.5590882301330566,
      "learning_rate": 7.509473047854186e-06,
      "loss": 1.971,
      "step": 213959
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.5562185049057007,
      "learning_rate": 7.50856356566889e-06,
      "loss": 1.9207,
      "step": 213960
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.4713157415390015,
      "learning_rate": 7.507654137863261e-06,
      "loss": 1.9685,
      "step": 213961
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.3431243896484375,
      "learning_rate": 7.506744764437467e-06,
      "loss": 1.7857,
      "step": 213962
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.5223420858383179,
      "learning_rate": 7.505835445391739e-06,
      "loss": 1.9463,
      "step": 213963
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.6841908693313599,
      "learning_rate": 7.504926180726145e-06,
      "loss": 1.8932,
      "step": 213964
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.7479504346847534,
      "learning_rate": 7.504016970440952e-06,
      "loss": 1.9229,
      "step": 213965
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.4217791557312012,
      "learning_rate": 7.503107814536257e-06,
      "loss": 1.9892,
      "step": 213966
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.5254148244857788,
      "learning_rate": 7.502198713012264e-06,
      "loss": 2.0691,
      "step": 213967
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.5097664594650269,
      "learning_rate": 7.501289665869104e-06,
      "loss": 1.9808,
      "step": 213968
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.3905766010284424,
      "learning_rate": 7.500380673107009e-06,
      "loss": 2.0001,
      "step": 213969
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.6207691431045532,
      "learning_rate": 7.499471734726081e-06,
      "loss": 2.0836,
      "step": 213970
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.4446736574172974,
      "learning_rate": 7.498562850726586e-06,
      "loss": 1.9644,
      "step": 213971
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.4911518096923828,
      "learning_rate": 7.49765402110859e-06,
      "loss": 1.9175,
      "step": 213972
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.3780181407928467,
      "learning_rate": 7.4967452458723265e-06,
      "loss": 1.9176,
      "step": 213973
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.4712142944335938,
      "learning_rate": 7.495836525017895e-06,
      "loss": 2.2986,
      "step": 213974
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.409312129020691,
      "learning_rate": 7.49492785854553e-06,
      "loss": 1.8732,
      "step": 213975
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.3627333641052246,
      "learning_rate": 7.494019246455363e-06,
      "loss": 1.7018,
      "step": 213976
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.473044753074646,
      "learning_rate": 7.493110688747595e-06,
      "loss": 1.8461,
      "step": 213977
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.4200488328933716,
      "learning_rate": 7.492202185422391e-06,
      "loss": 2.0793,
      "step": 213978
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.4698721170425415,
      "learning_rate": 7.4912937364799194e-06,
      "loss": 1.9431,
      "step": 213979
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.4560033082962036,
      "learning_rate": 7.49038534192028e-06,
      "loss": 1.914,
      "step": 213980
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.4325495958328247,
      "learning_rate": 7.489477001743738e-06,
      "loss": 2.0173,
      "step": 213981
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.4650495052337646,
      "learning_rate": 7.488568715950427e-06,
      "loss": 2.0254,
      "step": 213982
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.3998523950576782,
      "learning_rate": 7.487660484540514e-06,
      "loss": 1.9623,
      "step": 213983
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.4891374111175537,
      "learning_rate": 7.486752307514166e-06,
      "loss": 1.9068,
      "step": 213984
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.3732123374938965,
      "learning_rate": 7.4858441848715475e-06,
      "loss": 1.8649,
      "step": 213985
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.4485318660736084,
      "learning_rate": 7.484936116612828e-06,
      "loss": 1.9857,
      "step": 213986
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.4965503215789795,
      "learning_rate": 7.484028102738171e-06,
      "loss": 2.0271,
      "step": 213987
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.3897007703781128,
      "learning_rate": 7.483120143247745e-06,
      "loss": 2.093,
      "step": 213988
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.5419244766235352,
      "learning_rate": 7.482212238141783e-06,
      "loss": 2.0077,
      "step": 213989
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.401174783706665,
      "learning_rate": 7.481304387420317e-06,
      "loss": 2.0315,
      "step": 213990
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.472703218460083,
      "learning_rate": 7.480396591083715e-06,
      "loss": 2.2574,
      "step": 213991
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.5479660034179688,
      "learning_rate": 7.479488849131943e-06,
      "loss": 1.9231,
      "step": 213992
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.441428303718567,
      "learning_rate": 7.478581161565267e-06,
      "loss": 1.7893,
      "step": 213993
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.309192419052124,
      "learning_rate": 7.47767352838382e-06,
      "loss": 1.8429,
      "step": 213994
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.6000194549560547,
      "learning_rate": 7.476765949587871e-06,
      "loss": 2.0265,
      "step": 213995
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.4037458896636963,
      "learning_rate": 7.475858425177417e-06,
      "loss": 2.0666,
      "step": 213996
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.40436851978302,
      "learning_rate": 7.4749509551528256e-06,
      "loss": 1.862,
      "step": 213997
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.4958690404891968,
      "learning_rate": 7.474043539514096e-06,
      "loss": 2.1716,
      "step": 213998
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.5708434581756592,
      "learning_rate": 7.473136178261496e-06,
      "loss": 2.0948,
      "step": 213999
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.4551911354064941,
      "learning_rate": 7.472228871395125e-06,
      "loss": 2.0161,
      "step": 214000
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.423177719116211,
      "learning_rate": 7.471321618915216e-06,
      "loss": 2.0748,
      "step": 214001
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.4507405757904053,
      "learning_rate": 7.470414420821868e-06,
      "loss": 1.9295,
      "step": 214002
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.5175367593765259,
      "learning_rate": 7.46950727711535e-06,
      "loss": 2.0343,
      "step": 214003
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.5632416009902954,
      "learning_rate": 7.468600187795759e-06,
      "loss": 1.9726,
      "step": 214004
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.5924488306045532,
      "learning_rate": 7.467693152863296e-06,
      "loss": 1.9377,
      "step": 214005
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.5688848495483398,
      "learning_rate": 7.466786172318062e-06,
      "loss": 1.8653,
      "step": 214006
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.3660047054290771,
      "learning_rate": 7.465879246160289e-06,
      "loss": 1.9555,
      "step": 214007
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.3944286108016968,
      "learning_rate": 7.46497237439011e-06,
      "loss": 2.0977,
      "step": 214008
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.4652878046035767,
      "learning_rate": 7.464065557007759e-06,
      "loss": 1.921,
      "step": 214009
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.389290452003479,
      "learning_rate": 7.463158794013369e-06,
      "loss": 1.8558,
      "step": 214010
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.458472490310669,
      "learning_rate": 7.4622520854070724e-06,
      "loss": 2.1031,
      "step": 214011
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.5984201431274414,
      "learning_rate": 7.461345431189037e-06,
      "loss": 1.9535,
      "step": 214012
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.549346923828125,
      "learning_rate": 7.460438831359494e-06,
      "loss": 2.0486,
      "step": 214013
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.3954322338104248,
      "learning_rate": 7.459532285918579e-06,
      "loss": 1.7446,
      "step": 214014
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.4036279916763306,
      "learning_rate": 7.458625794866457e-06,
      "loss": 1.983,
      "step": 214015
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.4767206907272339,
      "learning_rate": 7.4577193582032954e-06,
      "loss": 1.9221,
      "step": 214016
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.4711475372314453,
      "learning_rate": 7.456812975929261e-06,
      "loss": 1.8746,
      "step": 214017
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.5600003004074097,
      "learning_rate": 7.455906648044519e-06,
      "loss": 2.0669,
      "step": 214018
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.4544473886489868,
      "learning_rate": 7.455000374549269e-06,
      "loss": 2.0197,
      "step": 214019
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.401776671409607,
      "learning_rate": 7.454094155443646e-06,
      "loss": 1.8398,
      "step": 214020
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.4892181158065796,
      "learning_rate": 7.4531879907278155e-06,
      "loss": 2.0357,
      "step": 214021
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.4567629098892212,
      "learning_rate": 7.452281880402011e-06,
      "loss": 1.8234,
      "step": 214022
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.379862904548645,
      "learning_rate": 7.451375824466266e-06,
      "loss": 1.9413,
      "step": 214023
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.4194928407669067,
      "learning_rate": 7.450469822920913e-06,
      "loss": 2.0722,
      "step": 214024
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.4366627931594849,
      "learning_rate": 7.449563875765985e-06,
      "loss": 2.0764,
      "step": 214025
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.4148237705230713,
      "learning_rate": 7.448657983001715e-06,
      "loss": 1.8623,
      "step": 214026
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.4154698848724365,
      "learning_rate": 7.4477521446283054e-06,
      "loss": 1.857,
      "step": 214027
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.5693389177322388,
      "learning_rate": 7.4468463606458525e-06,
      "loss": 2.0737,
      "step": 214028
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.4130398035049438,
      "learning_rate": 7.445940631054526e-06,
      "loss": 2.0389,
      "step": 214029
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.506281852722168,
      "learning_rate": 7.445034955854556e-06,
      "loss": 2.0386,
      "step": 214030
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.462342619895935,
      "learning_rate": 7.444129335046045e-06,
      "loss": 1.8106,
      "step": 214031
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.5236923694610596,
      "learning_rate": 7.443223768629225e-06,
      "loss": 2.0598,
      "step": 214032
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.5184375047683716,
      "learning_rate": 7.442318256604229e-06,
      "loss": 1.9689,
      "step": 214033
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.4231401681900024,
      "learning_rate": 7.4414127989712246e-06,
      "loss": 1.8857,
      "step": 214034
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.715682029724121,
      "learning_rate": 7.440507395730344e-06,
      "loss": 1.9867,
      "step": 214035
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.349862813949585,
      "learning_rate": 7.439602046881854e-06,
      "loss": 1.5808,
      "step": 214036
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.4134292602539062,
      "learning_rate": 7.438696752425821e-06,
      "loss": 1.9664,
      "step": 214037
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.5513941049575806,
      "learning_rate": 7.437791512362478e-06,
      "loss": 2.0681,
      "step": 214038
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.4916685819625854,
      "learning_rate": 7.436886326691993e-06,
      "loss": 1.9647,
      "step": 214039
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.5675078630447388,
      "learning_rate": 7.435981195414497e-06,
      "loss": 2.0582,
      "step": 214040
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.47881019115448,
      "learning_rate": 7.435076118530159e-06,
      "loss": 2.0039,
      "step": 214041
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.4271610975265503,
      "learning_rate": 7.434171096039176e-06,
      "loss": 1.906,
      "step": 214042
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.4641081094741821,
      "learning_rate": 7.433266127941684e-06,
      "loss": 1.9196,
      "step": 214043
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.4557826519012451,
      "learning_rate": 7.432361214237881e-06,
      "loss": 1.9301,
      "step": 214044
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.4035451412200928,
      "learning_rate": 7.431456354927934e-06,
      "loss": 1.875,
      "step": 214045
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.7555519342422485,
      "learning_rate": 7.4305515500120095e-06,
      "loss": 1.996,
      "step": 214046
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.4023269414901733,
      "learning_rate": 7.429646799490241e-06,
      "loss": 1.9264,
      "step": 214047
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.4184696674346924,
      "learning_rate": 7.428742103362861e-06,
      "loss": 1.8937,
      "step": 214048
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.5430303812026978,
      "learning_rate": 7.427837461629971e-06,
      "loss": 2.0154,
      "step": 214049
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.6510111093521118,
      "learning_rate": 7.4269328742917686e-06,
      "loss": 2.1319,
      "step": 214050
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.4723293781280518,
      "learning_rate": 7.426028341348456e-06,
      "loss": 1.9415,
      "step": 214051
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.4376749992370605,
      "learning_rate": 7.425123862800131e-06,
      "loss": 1.8446,
      "step": 214052
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.7660259008407593,
      "learning_rate": 7.424219438646995e-06,
      "loss": 1.6967,
      "step": 214053
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.5065525770187378,
      "learning_rate": 7.423315068889246e-06,
      "loss": 2.1016,
      "step": 214054
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.4154292345046997,
      "learning_rate": 7.4224107535269864e-06,
      "loss": 2.0594,
      "step": 214055
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.414347529411316,
      "learning_rate": 7.42150649256048e-06,
      "loss": 1.8996,
      "step": 214056
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.742691159248352,
      "learning_rate": 7.4206022859897625e-06,
      "loss": 2.1871,
      "step": 214057
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.4403196573257446,
      "learning_rate": 7.419698133815166e-06,
      "loss": 1.9387,
      "step": 214058
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.405566692352295,
      "learning_rate": 7.418794036036724e-06,
      "loss": 1.8726,
      "step": 214059
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.3996868133544922,
      "learning_rate": 7.417889992654668e-06,
      "loss": 1.9829,
      "step": 214060
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.4458738565444946,
      "learning_rate": 7.416986003669101e-06,
      "loss": 1.961,
      "step": 214061
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.4789401292800903,
      "learning_rate": 7.416082069080287e-06,
      "loss": 1.7425,
      "step": 214062
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.649816632270813,
      "learning_rate": 7.4151781888883265e-06,
      "loss": 2.1475,
      "step": 214063
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.4290802478790283,
      "learning_rate": 7.414274363093453e-06,
      "loss": 1.9909,
      "step": 214064
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.418745994567871,
      "learning_rate": 7.413370591695733e-06,
      "loss": 2.0614,
      "step": 214065
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.4337712526321411,
      "learning_rate": 7.412466874695399e-06,
      "loss": 1.9754,
      "step": 214066
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.495247483253479,
      "learning_rate": 7.411563212092619e-06,
      "loss": 1.9727,
      "step": 214067
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.4344483613967896,
      "learning_rate": 7.4106596038875585e-06,
      "loss": 1.9464,
      "step": 214068
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.5067843198776245,
      "learning_rate": 7.409756050080351e-06,
      "loss": 1.998,
      "step": 214069
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.389398455619812,
      "learning_rate": 7.408852550671263e-06,
      "loss": 1.7322,
      "step": 214070
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.5468006134033203,
      "learning_rate": 7.407949105660327e-06,
      "loss": 2.0402,
      "step": 214071
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.431624412536621,
      "learning_rate": 7.40704571504781e-06,
      "loss": 1.8294,
      "step": 214072
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.3346805572509766,
      "learning_rate": 7.406142378833813e-06,
      "loss": 1.9024,
      "step": 214073
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.4650274515151978,
      "learning_rate": 7.4052390970185675e-06,
      "loss": 1.8952,
      "step": 214074
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.6407357454299927,
      "learning_rate": 7.4043358696021735e-06,
      "loss": 2.139,
      "step": 214075
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.4777653217315674,
      "learning_rate": 7.403432696584899e-06,
      "loss": 1.9992,
      "step": 214076
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.5451048612594604,
      "learning_rate": 7.402529577966776e-06,
      "loss": 1.8123,
      "step": 214077
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.60899019241333,
      "learning_rate": 7.401626513748105e-06,
      "loss": 1.8683,
      "step": 214078
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.4138481616973877,
      "learning_rate": 7.4007235039289516e-06,
      "loss": 1.9254,
      "step": 214079
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.4331297874450684,
      "learning_rate": 7.39982054850955e-06,
      "loss": 1.9888,
      "step": 214080
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.5745346546173096,
      "learning_rate": 7.398917647490033e-06,
      "loss": 2.0254,
      "step": 214081
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.3891242742538452,
      "learning_rate": 7.3980148008706335e-06,
      "loss": 1.9297,
      "step": 214082
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.5677019357681274,
      "learning_rate": 7.397112008651385e-06,
      "loss": 1.9002,
      "step": 214083
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.778625249862671,
      "learning_rate": 7.396209270832587e-06,
      "loss": 1.8923,
      "step": 214084
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.5168766975402832,
      "learning_rate": 7.39530658741434e-06,
      "loss": 2.0397,
      "step": 214085
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.4127469062805176,
      "learning_rate": 7.394403958396844e-06,
      "loss": 1.9448,
      "step": 214086
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.4972840547561646,
      "learning_rate": 7.393501383780232e-06,
      "loss": 1.8886,
      "step": 214087
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.560067892074585,
      "learning_rate": 7.392598863564736e-06,
      "loss": 1.8691,
      "step": 214088
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.5823496580123901,
      "learning_rate": 7.391696397750424e-06,
      "loss": 1.9257,
      "step": 214089
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.454493761062622,
      "learning_rate": 7.390793986337529e-06,
      "loss": 1.9668,
      "step": 214090
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.4763317108154297,
      "learning_rate": 7.389891629326216e-06,
      "loss": 1.9553,
      "step": 214091
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.3978244066238403,
      "learning_rate": 7.388989326716655e-06,
      "loss": 1.8812,
      "step": 214092
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.7083326578140259,
      "learning_rate": 7.388087078508975e-06,
      "loss": 1.9166,
      "step": 214093
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.3228880167007446,
      "learning_rate": 7.387184884703412e-06,
      "loss": 2.0752,
      "step": 214094
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.4140278100967407,
      "learning_rate": 7.386282745300098e-06,
      "loss": 2.0485,
      "step": 214095
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.384356141090393,
      "learning_rate": 7.385380660299167e-06,
      "loss": 1.9829,
      "step": 214096
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.560156226158142,
      "learning_rate": 7.384478629700819e-06,
      "loss": 2.1217,
      "step": 214097
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.4611104726791382,
      "learning_rate": 7.3835766535052524e-06,
      "loss": 2.1068,
      "step": 214098
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.4356024265289307,
      "learning_rate": 7.382674731712535e-06,
      "loss": 2.041,
      "step": 214099
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.4247121810913086,
      "learning_rate": 7.381772864322966e-06,
      "loss": 1.9726,
      "step": 214100
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.4047411680221558,
      "learning_rate": 7.380871051336646e-06,
      "loss": 1.9807,
      "step": 214101
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.4806569814682007,
      "learning_rate": 7.3799692927537065e-06,
      "loss": 2.0146,
      "step": 214102
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.441361904144287,
      "learning_rate": 7.379067588574383e-06,
      "loss": 1.6959,
      "step": 214103
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.5252267122268677,
      "learning_rate": 7.378165938798808e-06,
      "loss": 1.9673,
      "step": 214104
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.3226374387741089,
      "learning_rate": 7.377264343427147e-06,
      "loss": 2.1663,
      "step": 214105
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.3182365894317627,
      "learning_rate": 7.376362802459601e-06,
      "loss": 2.102,
      "step": 214106
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.4071677923202515,
      "learning_rate": 7.375461315896303e-06,
      "loss": 1.7986,
      "step": 214107
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.4849900007247925,
      "learning_rate": 7.374559883737418e-06,
      "loss": 1.9057,
      "step": 214108
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.286877989768982,
      "learning_rate": 7.3736585059831155e-06,
      "loss": 2.09,
      "step": 214109
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.5956536531448364,
      "learning_rate": 7.372757182633626e-06,
      "loss": 1.9444,
      "step": 214110
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.6107966899871826,
      "learning_rate": 7.371855913688984e-06,
      "loss": 2.0346,
      "step": 214111
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.450058102607727,
      "learning_rate": 7.370954699149523e-06,
      "loss": 1.6719,
      "step": 214112
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.5850507020950317,
      "learning_rate": 7.370053539015275e-06,
      "loss": 1.8152,
      "step": 214113
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.5224425792694092,
      "learning_rate": 7.3691524332864405e-06,
      "loss": 2.1166,
      "step": 214114
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.608736515045166,
      "learning_rate": 7.368251381963253e-06,
      "loss": 1.8546,
      "step": 214115
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.3021992444992065,
      "learning_rate": 7.367350385045779e-06,
      "loss": 2.1145,
      "step": 214116
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.5849156379699707,
      "learning_rate": 7.366449442534284e-06,
      "loss": 2.0575,
      "step": 214117
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.545032262802124,
      "learning_rate": 7.365548554428902e-06,
      "loss": 2.1056,
      "step": 214118
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.4171671867370605,
      "learning_rate": 7.364647720729766e-06,
      "loss": 1.9257,
      "step": 214119
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.5704929828643799,
      "learning_rate": 7.363746941437043e-06,
      "loss": 2.0523,
      "step": 214120
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.2953119277954102,
      "learning_rate": 7.362846216550966e-06,
      "loss": 2.0297,
      "step": 214121
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.4385894536972046,
      "learning_rate": 7.361945546071602e-06,
      "loss": 1.8221,
      "step": 214122
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.5596845149993896,
      "learning_rate": 7.361044929999216e-06,
      "loss": 1.9868,
      "step": 214123
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.5324476957321167,
      "learning_rate": 7.360144368333909e-06,
      "loss": 2.0527,
      "step": 214124
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.7888563871383667,
      "learning_rate": 7.359243861075947e-06,
      "loss": 1.9609,
      "step": 214125
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.3722091913223267,
      "learning_rate": 7.358343408225364e-06,
      "loss": 2.0031,
      "step": 214126
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.6620951890945435,
      "learning_rate": 7.357443009782393e-06,
      "loss": 1.9525,
      "step": 214127
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.4739521741867065,
      "learning_rate": 7.356542665747201e-06,
      "loss": 1.8765,
      "step": 214128
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.403380274772644,
      "learning_rate": 7.3556423761199525e-06,
      "loss": 2.1113,
      "step": 214129
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.5056867599487305,
      "learning_rate": 7.354742140900815e-06,
      "loss": 2.0988,
      "step": 214130
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.459865927696228,
      "learning_rate": 7.35384196008999e-06,
      "loss": 1.812,
      "step": 214131
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.4446330070495605,
      "learning_rate": 7.3529418336875425e-06,
      "loss": 2.2175,
      "step": 214132
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.716952919960022,
      "learning_rate": 7.352041761693772e-06,
      "loss": 1.8613,
      "step": 214133
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.425305724143982,
      "learning_rate": 7.351141744108713e-06,
      "loss": 2.0424,
      "step": 214134
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.4234843254089355,
      "learning_rate": 7.3502417809326645e-06,
      "loss": 1.8471,
      "step": 214135
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.5480055809020996,
      "learning_rate": 7.349341872165693e-06,
      "loss": 2.1701,
      "step": 214136
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.484386682510376,
      "learning_rate": 7.348442017808031e-06,
      "loss": 1.9964,
      "step": 214137
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.4004557132720947,
      "learning_rate": 7.347542217859781e-06,
      "loss": 2.2133,
      "step": 214138
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.5020158290863037,
      "learning_rate": 7.346642472321207e-06,
      "loss": 1.9743,
      "step": 214139
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.4692134857177734,
      "learning_rate": 7.345742781192343e-06,
      "loss": 1.8784,
      "step": 214140
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.4715346097946167,
      "learning_rate": 7.344843144473489e-06,
      "loss": 2.0423,
      "step": 214141
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.6827318668365479,
      "learning_rate": 7.343943562164711e-06,
      "loss": 2.0571,
      "step": 214142
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.508188009262085,
      "learning_rate": 7.343044034266244e-06,
      "loss": 2.0362,
      "step": 214143
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.5165081024169922,
      "learning_rate": 7.342144560778218e-06,
      "loss": 1.9268,
      "step": 214144
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.624316692352295,
      "learning_rate": 7.341245141700835e-06,
      "loss": 2.0625,
      "step": 214145
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.4488272666931152,
      "learning_rate": 7.340345777034196e-06,
      "loss": 1.9406,
      "step": 214146
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.3716285228729248,
      "learning_rate": 7.339446466778531e-06,
      "loss": 1.8621,
      "step": 214147
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.4909662008285522,
      "learning_rate": 7.338547210933976e-06,
      "loss": 2.0589,
      "step": 214148
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.4566326141357422,
      "learning_rate": 7.337648009500763e-06,
      "loss": 1.9642,
      "step": 214149
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.4341241121292114,
      "learning_rate": 7.336748862478958e-06,
      "loss": 1.8753,
      "step": 214150
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.4703863859176636,
      "learning_rate": 7.3358497698687625e-06,
      "loss": 1.7981,
      "step": 214151
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.3592915534973145,
      "learning_rate": 7.334950731670375e-06,
      "loss": 2.0309,
      "step": 214152
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.6460121870040894,
      "learning_rate": 7.334051747883962e-06,
      "loss": 2.1369,
      "step": 214153
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.378706455230713,
      "learning_rate": 7.333152818509624e-06,
      "loss": 1.9685,
      "step": 214154
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.5063214302062988,
      "learning_rate": 7.332253943547661e-06,
      "loss": 1.9518,
      "step": 214155
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.454962968826294,
      "learning_rate": 7.331355122998106e-06,
      "loss": 2.0785,
      "step": 214156
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.5087603330612183,
      "learning_rate": 7.330456356861158e-06,
      "loss": 1.9464,
      "step": 214157
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.4617811441421509,
      "learning_rate": 7.329557645137018e-06,
      "loss": 1.9785,
      "step": 214158
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.5734938383102417,
      "learning_rate": 7.328658987825853e-06,
      "loss": 2.0727,
      "step": 214159
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.6001660823822021,
      "learning_rate": 7.327760384927794e-06,
      "loss": 2.1702,
      "step": 214160
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.5145769119262695,
      "learning_rate": 7.326861836443076e-06,
      "loss": 1.9379,
      "step": 214161
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.4587023258209229,
      "learning_rate": 7.325963342371765e-06,
      "loss": 1.9296,
      "step": 214162
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.5250569581985474,
      "learning_rate": 7.325064902714095e-06,
      "loss": 1.9522,
      "step": 214163
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.466591715812683,
      "learning_rate": 7.324166517470232e-06,
      "loss": 1.9833,
      "step": 214164
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.642316222190857,
      "learning_rate": 7.323268186640307e-06,
      "loss": 1.9846,
      "step": 214165
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.4000952243804932,
      "learning_rate": 7.322369910224524e-06,
      "loss": 1.8721,
      "step": 214166
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.530242919921875,
      "learning_rate": 7.32147168822308e-06,
      "loss": 1.9531,
      "step": 214167
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.4510642290115356,
      "learning_rate": 7.320573520636042e-06,
      "loss": 1.9064,
      "step": 214168
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.3207916021347046,
      "learning_rate": 7.319675407463644e-06,
      "loss": 1.9199,
      "step": 214169
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.4739717245101929,
      "learning_rate": 7.318777348706051e-06,
      "loss": 2.0126,
      "step": 214170
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.4073070287704468,
      "learning_rate": 7.317879344363431e-06,
      "loss": 2.0161,
      "step": 214171
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.515164852142334,
      "learning_rate": 7.316981394435917e-06,
      "loss": 2.0508,
      "step": 214172
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.4956380128860474,
      "learning_rate": 7.316083498923741e-06,
      "loss": 1.9554,
      "step": 214173
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.4194613695144653,
      "learning_rate": 7.3151856578270055e-06,
      "loss": 2.0545,
      "step": 214174
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.5491966009140015,
      "learning_rate": 7.3142878711459075e-06,
      "loss": 1.8767,
      "step": 214175
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.4011168479919434,
      "learning_rate": 7.313390138880582e-06,
      "loss": 1.9918,
      "step": 214176
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.5225087404251099,
      "learning_rate": 7.312492461031261e-06,
      "loss": 1.9975,
      "step": 214177
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.551733374595642,
      "learning_rate": 7.3115948375980445e-06,
      "loss": 2.036,
      "step": 214178
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.7216264009475708,
      "learning_rate": 7.310697268581134e-06,
      "loss": 1.9277,
      "step": 214179
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.7024883031845093,
      "learning_rate": 7.309799753980694e-06,
      "loss": 2.0408,
      "step": 214180
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.4239360094070435,
      "learning_rate": 7.308902293796925e-06,
      "loss": 1.7832,
      "step": 214181
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.490404725074768,
      "learning_rate": 7.3080048880298944e-06,
      "loss": 1.9017,
      "step": 214182
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.4351637363433838,
      "learning_rate": 7.307107536679868e-06,
      "loss": 2.0562,
      "step": 214183
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.4442386627197266,
      "learning_rate": 7.306210239746946e-06,
      "loss": 1.8769,
      "step": 214184
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.6454527378082275,
      "learning_rate": 7.305312997231327e-06,
      "loss": 2.0193,
      "step": 214185
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.4597562551498413,
      "learning_rate": 7.304415809133213e-06,
      "loss": 1.897,
      "step": 214186
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.4832074642181396,
      "learning_rate": 7.3035186754526685e-06,
      "loss": 2.0793,
      "step": 214187
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.597022294998169,
      "learning_rate": 7.302621596189995e-06,
      "loss": 1.8831,
      "step": 214188
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.4531241655349731,
      "learning_rate": 7.301724571345258e-06,
      "loss": 1.9603,
      "step": 214189
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.5550516843795776,
      "learning_rate": 7.300827600918657e-06,
      "loss": 2.2907,
      "step": 214190
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.4074254035949707,
      "learning_rate": 7.29993068491036e-06,
      "loss": 1.7658,
      "step": 214191
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.4261040687561035,
      "learning_rate": 7.299033823320533e-06,
      "loss": 2.1695,
      "step": 214192
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.3547694683074951,
      "learning_rate": 7.298137016149308e-06,
      "loss": 1.8127,
      "step": 214193
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.4443060159683228,
      "learning_rate": 7.29724026339692e-06,
      "loss": 2.0394,
      "step": 214194
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.6101880073547363,
      "learning_rate": 7.296343565063534e-06,
      "loss": 1.984,
      "step": 214195
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.4851223230361938,
      "learning_rate": 7.2954469211492175e-06,
      "loss": 1.8864,
      "step": 214196
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.56874680519104,
      "learning_rate": 7.294550331654236e-06,
      "loss": 1.8461,
      "step": 214197
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.6468383073806763,
      "learning_rate": 7.293653796578725e-06,
      "loss": 2.0899,
      "step": 214198
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.5555099248886108,
      "learning_rate": 7.292757315922815e-06,
      "loss": 1.9129,
      "step": 214199
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.382818579673767,
      "learning_rate": 7.29186088968674e-06,
      "loss": 2.1018,
      "step": 214200
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.4976078271865845,
      "learning_rate": 7.290964517870601e-06,
      "loss": 1.9431,
      "step": 214201
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.5152912139892578,
      "learning_rate": 7.29006820047463e-06,
      "loss": 1.8586,
      "step": 214202
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.4892486333847046,
      "learning_rate": 7.289171937498961e-06,
      "loss": 1.9833,
      "step": 214203
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.4693703651428223,
      "learning_rate": 7.28827572894376e-06,
      "loss": 1.8447,
      "step": 214204
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.435056209564209,
      "learning_rate": 7.28737957480916e-06,
      "loss": 2.0158,
      "step": 214205
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.4982365369796753,
      "learning_rate": 7.2864834750953945e-06,
      "loss": 2.1802,
      "step": 214206
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.4921540021896362,
      "learning_rate": 7.285587429802564e-06,
      "loss": 2.0491,
      "step": 214207
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.400116205215454,
      "learning_rate": 7.284691438930901e-06,
      "loss": 1.95,
      "step": 214208
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.4171091318130493,
      "learning_rate": 7.283795502480538e-06,
      "loss": 1.9388,
      "step": 214209
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.4428006410598755,
      "learning_rate": 7.282899620451643e-06,
      "loss": 2.0864,
      "step": 214210
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.5515109300613403,
      "learning_rate": 7.282003792844349e-06,
      "loss": 2.1663,
      "step": 214211
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.4039522409439087,
      "learning_rate": 7.281108019658921e-06,
      "loss": 2.0199,
      "step": 214212
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.4798604249954224,
      "learning_rate": 7.280212300895394e-06,
      "loss": 1.9552,
      "step": 214213
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.5138053894042969,
      "learning_rate": 7.279316636554034e-06,
      "loss": 1.8584,
      "step": 214214
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.4521907567977905,
      "learning_rate": 7.27842102663494e-06,
      "loss": 1.9714,
      "step": 214215
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.4709861278533936,
      "learning_rate": 7.277525471138379e-06,
      "loss": 1.87,
      "step": 214216
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.4121640920639038,
      "learning_rate": 7.276629970064385e-06,
      "loss": 1.9334,
      "step": 214217
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.377188801765442,
      "learning_rate": 7.275734523413257e-06,
      "loss": 1.9072,
      "step": 214218
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.4105778932571411,
      "learning_rate": 7.274839131185028e-06,
      "loss": 1.8706,
      "step": 214219
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.4837464094161987,
      "learning_rate": 7.273943793379966e-06,
      "loss": 1.9885,
      "step": 214220
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.6059813499450684,
      "learning_rate": 7.273048509998169e-06,
      "loss": 2.1326,
      "step": 214221
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.5069799423217773,
      "learning_rate": 7.272153281039938e-06,
      "loss": 1.9212,
      "step": 214222
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.6416062116622925,
      "learning_rate": 7.27125810650524e-06,
      "loss": 2.1559,
      "step": 214223
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.606502652168274,
      "learning_rate": 7.2703629863943735e-06,
      "loss": 2.0813,
      "step": 214224
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.4608182907104492,
      "learning_rate": 7.2694679207074725e-06,
      "loss": 1.8792,
      "step": 214225
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.5047106742858887,
      "learning_rate": 7.268572909444703e-06,
      "loss": 1.833,
      "step": 214226
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.3719687461853027,
      "learning_rate": 7.267677952606199e-06,
      "loss": 2.073,
      "step": 214227
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.411911964416504,
      "learning_rate": 7.266783050192226e-06,
      "loss": 2.0211,
      "step": 214228
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.4747986793518066,
      "learning_rate": 7.265888202202819e-06,
      "loss": 1.8177,
      "step": 214229
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.4465930461883545,
      "learning_rate": 7.2649934086382755e-06,
      "loss": 1.8735,
      "step": 214230
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.4800766706466675,
      "learning_rate": 7.26409866949863e-06,
      "loss": 2.0277,
      "step": 214231
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.4186829328536987,
      "learning_rate": 7.2632039847841495e-06,
      "loss": 1.9246,
      "step": 214232
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.586658239364624,
      "learning_rate": 7.262309354494933e-06,
      "loss": 1.9396,
      "step": 214233
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.438595175743103,
      "learning_rate": 7.261414778631247e-06,
      "loss": 1.8736,
      "step": 214234
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.5797219276428223,
      "learning_rate": 7.2605202571931255e-06,
      "loss": 1.9545,
      "step": 214235
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.4445329904556274,
      "learning_rate": 7.259625790180834e-06,
      "loss": 1.8465,
      "step": 214236
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.5124270915985107,
      "learning_rate": 7.258731377594473e-06,
      "loss": 2.0674,
      "step": 214237
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.569604516029358,
      "learning_rate": 7.257837019434242e-06,
      "loss": 2.1131,
      "step": 214238
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.6723849773406982,
      "learning_rate": 7.256942715700309e-06,
      "loss": 2.0009,
      "step": 214239
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.3954235315322876,
      "learning_rate": 7.2560484663929055e-06,
      "loss": 1.8105,
      "step": 214240
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.4711743593215942,
      "learning_rate": 7.255154271512032e-06,
      "loss": 2.0181,
      "step": 214241
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.4302765130996704,
      "learning_rate": 7.2542601310580205e-06,
      "loss": 1.8799,
      "step": 214242
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.7470656633377075,
      "learning_rate": 7.253366045030906e-06,
      "loss": 1.945,
      "step": 214243
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.3337303400039673,
      "learning_rate": 7.252472013430955e-06,
      "loss": 1.7283,
      "step": 214244
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.4409661293029785,
      "learning_rate": 7.251578036258265e-06,
      "loss": 1.9241,
      "step": 214245
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.486426591873169,
      "learning_rate": 7.250684113513106e-06,
      "loss": 1.882,
      "step": 214246
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.3947319984436035,
      "learning_rate": 7.2497902451955084e-06,
      "loss": 2.1869,
      "step": 214247
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.53694748878479,
      "learning_rate": 7.2488964313057065e-06,
      "loss": 2.0632,
      "step": 214248
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.403428077697754,
      "learning_rate": 7.248002671843867e-06,
      "loss": 1.9772,
      "step": 214249
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.3967839479446411,
      "learning_rate": 7.247108966810156e-06,
      "loss": 1.9982,
      "step": 214250
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.353041648864746,
      "learning_rate": 7.246215316204707e-06,
      "loss": 1.8541,
      "step": 214251
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.4808987379074097,
      "learning_rate": 7.245321720027752e-06,
      "loss": 1.9891,
      "step": 214252
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.4613374471664429,
      "learning_rate": 7.244428178279393e-06,
      "loss": 2.1315,
      "step": 214253
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.3636873960494995,
      "learning_rate": 7.243534690959829e-06,
      "loss": 1.9435,
      "step": 214254
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.4168667793273926,
      "learning_rate": 7.242641258069193e-06,
      "loss": 1.9625,
      "step": 214255
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.524841070175171,
      "learning_rate": 7.2417478796077175e-06,
      "loss": 2.003,
      "step": 214256
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.495786428451538,
      "learning_rate": 7.240854555575504e-06,
      "loss": 1.9743,
      "step": 214257
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.4117544889450073,
      "learning_rate": 7.239961285972751e-06,
      "loss": 1.833,
      "step": 214258
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.444765329360962,
      "learning_rate": 7.239068070799625e-06,
      "loss": 1.916,
      "step": 214259
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.4854334592819214,
      "learning_rate": 7.23817491005626e-06,
      "loss": 1.8463,
      "step": 214260
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.4294073581695557,
      "learning_rate": 7.237281803742856e-06,
      "loss": 2.0324,
      "step": 214261
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.4175540208816528,
      "learning_rate": 7.236388751859579e-06,
      "loss": 1.8951,
      "step": 214262
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.4249380826950073,
      "learning_rate": 7.2354957544065284e-06,
      "loss": 1.8031,
      "step": 214263
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.3826619386672974,
      "learning_rate": 7.234602811384005e-06,
      "loss": 1.923,
      "step": 214264
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.7448354959487915,
      "learning_rate": 7.233709922792041e-06,
      "loss": 2.004,
      "step": 214265
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.4725044965744019,
      "learning_rate": 7.232817088630905e-06,
      "loss": 2.1863,
      "step": 214266
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.5266634225845337,
      "learning_rate": 7.231924308900661e-06,
      "loss": 1.8696,
      "step": 214267
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.5867475271224976,
      "learning_rate": 7.231031583601543e-06,
      "loss": 2.0565,
      "step": 214268
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.515875220298767,
      "learning_rate": 7.230138912733719e-06,
      "loss": 1.981,
      "step": 214269
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.4914692640304565,
      "learning_rate": 7.22924629629732e-06,
      "loss": 2.1642,
      "step": 214270
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.5297930240631104,
      "learning_rate": 7.228353734292547e-06,
      "loss": 2.137,
      "step": 214271
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.4246015548706055,
      "learning_rate": 7.227461226719567e-06,
      "loss": 1.8834,
      "step": 214272
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.4506869316101074,
      "learning_rate": 7.226568773578478e-06,
      "loss": 1.8957,
      "step": 214273
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.4324774742126465,
      "learning_rate": 7.225676374869549e-06,
      "loss": 1.9876,
      "step": 214274
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.4548375606536865,
      "learning_rate": 7.224784030592845e-06,
      "loss": 1.8591,
      "step": 214275
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.4794327020645142,
      "learning_rate": 7.2238917407486e-06,
      "loss": 1.9578,
      "step": 214276
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.4762208461761475,
      "learning_rate": 7.22299950533698e-06,
      "loss": 2.0358,
      "step": 214277
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.459180235862732,
      "learning_rate": 7.222107324358084e-06,
      "loss": 2.0728,
      "step": 214278
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.3302773237228394,
      "learning_rate": 7.2212151978121805e-06,
      "loss": 1.8951,
      "step": 214279
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.376915454864502,
      "learning_rate": 7.220323125699368e-06,
      "loss": 1.8884,
      "step": 214280
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.4370110034942627,
      "learning_rate": 7.2194311080197795e-06,
      "loss": 1.8779,
      "step": 214281
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.448425531387329,
      "learning_rate": 7.2185391447736826e-06,
      "loss": 2.055,
      "step": 214282
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.5101449489593506,
      "learning_rate": 7.217647235961144e-06,
      "loss": 1.873,
      "step": 214283
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.5112282037734985,
      "learning_rate": 7.216755381582395e-06,
      "loss": 1.877,
      "step": 214284
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.4027818441390991,
      "learning_rate": 7.215863581637571e-06,
      "loss": 1.8789,
      "step": 214285
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.4772145748138428,
      "learning_rate": 7.214971836126804e-06,
      "loss": 1.7684,
      "step": 214286
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.4137179851531982,
      "learning_rate": 7.214080145050361e-06,
      "loss": 1.8359,
      "step": 214287
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.4494221210479736,
      "learning_rate": 7.213188508408341e-06,
      "loss": 2.2603,
      "step": 214288
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.4024795293807983,
      "learning_rate": 7.212296926200911e-06,
      "loss": 1.8786,
      "step": 214289
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.4202241897583008,
      "learning_rate": 7.211405398428205e-06,
      "loss": 2.0174,
      "step": 214290
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.440938115119934,
      "learning_rate": 7.210513925090456e-06,
      "loss": 1.8553,
      "step": 214291
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.547595500946045,
      "learning_rate": 7.209622506187795e-06,
      "loss": 1.8938,
      "step": 214292
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.6147738695144653,
      "learning_rate": 7.208731141720392e-06,
      "loss": 1.9531,
      "step": 214293
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.72853684425354,
      "learning_rate": 7.207839831688445e-06,
      "loss": 2.0409,
      "step": 214294
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.7389419078826904,
      "learning_rate": 7.2069485760920534e-06,
      "loss": 2.0087,
      "step": 214295
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.4564785957336426,
      "learning_rate": 7.206057374931384e-06,
      "loss": 1.9934,
      "step": 214296
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.516639232635498,
      "learning_rate": 7.205166228206705e-06,
      "loss": 2.1363,
      "step": 214297
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.4915200471878052,
      "learning_rate": 7.204275135918047e-06,
      "loss": 2.0816,
      "step": 214298
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.418864369392395,
      "learning_rate": 7.203384098065679e-06,
      "loss": 1.8884,
      "step": 214299
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.3193854093551636,
      "learning_rate": 7.202493114649699e-06,
      "loss": 2.039,
      "step": 214300
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.4905261993408203,
      "learning_rate": 7.2016021856703745e-06,
      "loss": 1.9757,
      "step": 214301
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.758617877960205,
      "learning_rate": 7.200711311127738e-06,
      "loss": 2.0932,
      "step": 214302
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.5419853925704956,
      "learning_rate": 7.199820491022024e-06,
      "loss": 1.9623,
      "step": 214303
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.5972853899002075,
      "learning_rate": 7.198929725353397e-06,
      "loss": 1.9805,
      "step": 214304
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.5757598876953125,
      "learning_rate": 7.198039014122026e-06,
      "loss": 2.0838,
      "step": 214305
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.4790878295898438,
      "learning_rate": 7.197148357328042e-06,
      "loss": 1.9599,
      "step": 214306
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.676436424255371,
      "learning_rate": 7.19625775497168e-06,
      "loss": 1.8594,
      "step": 214307
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.4639760255813599,
      "learning_rate": 7.195367207053005e-06,
      "loss": 2.1017,
      "step": 214308
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.3234851360321045,
      "learning_rate": 7.1944767135722834e-06,
      "loss": 2.1092,
      "step": 214309
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.4394781589508057,
      "learning_rate": 7.193586274529584e-06,
      "loss": 1.8996,
      "step": 214310
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.4163079261779785,
      "learning_rate": 7.1926958899251705e-06,
      "loss": 2.139,
      "step": 214311
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.536838412284851,
      "learning_rate": 7.191805559759112e-06,
      "loss": 1.9093,
      "step": 214312
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.331260085105896,
      "learning_rate": 7.190915284031706e-06,
      "loss": 1.7679,
      "step": 214313
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.4257161617279053,
      "learning_rate": 7.190025062742955e-06,
      "loss": 1.9888,
      "step": 214314
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.3699593544006348,
      "learning_rate": 7.189134895893156e-06,
      "loss": 1.885,
      "step": 214315
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.4778070449829102,
      "learning_rate": 7.188244783482377e-06,
      "loss": 1.9218,
      "step": 214316
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.5762271881103516,
      "learning_rate": 7.187354725510885e-06,
      "loss": 1.8798,
      "step": 214317
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.468500018119812,
      "learning_rate": 7.1864647219787465e-06,
      "loss": 2.2764,
      "step": 214318
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.5807833671569824,
      "learning_rate": 7.185574772886227e-06,
      "loss": 2.0272,
      "step": 214319
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.6191670894622803,
      "learning_rate": 7.1846848782333936e-06,
      "loss": 2.0859,
      "step": 214320
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.4557722806930542,
      "learning_rate": 7.183795038020479e-06,
      "loss": 1.9639,
      "step": 214321
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.3969019651412964,
      "learning_rate": 7.182905252247584e-06,
      "loss": 1.9568,
      "step": 214322
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.4084364175796509,
      "learning_rate": 7.182015520914941e-06,
      "loss": 2.049,
      "step": 214323
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.499105453491211,
      "learning_rate": 7.18112584402265e-06,
      "loss": 2.2241,
      "step": 214324
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.765433430671692,
      "learning_rate": 7.1802362215709784e-06,
      "loss": 2.1093,
      "step": 214325
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.4763638973236084,
      "learning_rate": 7.179346653559992e-06,
      "loss": 1.7884,
      "step": 214326
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.4722731113433838,
      "learning_rate": 7.178457139989924e-06,
      "loss": 1.9739,
      "step": 214327
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.774121642112732,
      "learning_rate": 7.177567680860874e-06,
      "loss": 2.0119,
      "step": 214328
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.514617919921875,
      "learning_rate": 7.176678276173043e-06,
      "loss": 2.2726,
      "step": 214329
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.5686681270599365,
      "learning_rate": 7.175788925926595e-06,
      "loss": 1.897,
      "step": 214330
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.418191909790039,
      "learning_rate": 7.174899630121733e-06,
      "loss": 1.9095,
      "step": 214331
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.537593126296997,
      "learning_rate": 7.1740103887585555e-06,
      "loss": 1.8802,
      "step": 214332
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.5678315162658691,
      "learning_rate": 7.173121201837262e-06,
      "loss": 1.9245,
      "step": 214333
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.4354703426361084,
      "learning_rate": 7.172232069357986e-06,
      "loss": 1.9892,
      "step": 214334
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.4397214651107788,
      "learning_rate": 7.171342991320961e-06,
      "loss": 2.1506,
      "step": 214335
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.654128074645996,
      "learning_rate": 7.1704539677262865e-06,
      "loss": 1.9724,
      "step": 214336
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.4097867012023926,
      "learning_rate": 7.169564998574162e-06,
      "loss": 1.8795,
      "step": 214337
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.477609634399414,
      "learning_rate": 7.168676083864755e-06,
      "loss": 1.9534,
      "step": 214338
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.5067270994186401,
      "learning_rate": 7.1677872235981985e-06,
      "loss": 2.0868,
      "step": 214339
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.4852532148361206,
      "learning_rate": 7.166898417774658e-06,
      "loss": 2.0028,
      "step": 214340
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.4632915258407593,
      "learning_rate": 7.166009666394368e-06,
      "loss": 1.8717,
      "step": 214341
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.3802695274353027,
      "learning_rate": 7.165120969457394e-06,
      "loss": 1.8616,
      "step": 214342
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.4349844455718994,
      "learning_rate": 7.16423232696397e-06,
      "loss": 1.7937,
      "step": 214343
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.6585057973861694,
      "learning_rate": 7.1633437389142625e-06,
      "loss": 2.0793,
      "step": 214344
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.4620810747146606,
      "learning_rate": 7.162455205308437e-06,
      "loss": 1.9807,
      "step": 214345
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.5294642448425293,
      "learning_rate": 7.161566726146561e-06,
      "loss": 1.8125,
      "step": 214346
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.6341004371643066,
      "learning_rate": 7.160678301428935e-06,
      "loss": 2.0441,
      "step": 214347
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.3607938289642334,
      "learning_rate": 7.159789931155657e-06,
      "loss": 1.9708,
      "step": 214348
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.433493971824646,
      "learning_rate": 7.158901615326895e-06,
      "loss": 2.0779,
      "step": 214349
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.4297198057174683,
      "learning_rate": 7.158013353942849e-06,
      "loss": 1.8111,
      "step": 214350
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.4203320741653442,
      "learning_rate": 7.15712514700365e-06,
      "loss": 2.2983,
      "step": 214351
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.452823519706726,
      "learning_rate": 7.156236994509435e-06,
      "loss": 2.1062,
      "step": 214352
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.4041504859924316,
      "learning_rate": 7.1553488964604e-06,
      "loss": 1.9686,
      "step": 214353
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.4893221855163574,
      "learning_rate": 7.154460852856713e-06,
      "loss": 2.1891,
      "step": 214354
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.5085697174072266,
      "learning_rate": 7.153572863698576e-06,
      "loss": 2.0485,
      "step": 214355
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.5587822198867798,
      "learning_rate": 7.152684928986119e-06,
      "loss": 1.9833,
      "step": 214356
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.499769687652588,
      "learning_rate": 7.15179704871951e-06,
      "loss": 1.7563,
      "step": 214357
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.4518959522247314,
      "learning_rate": 7.150909222898849e-06,
      "loss": 2.0282,
      "step": 214358
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.3792468309402466,
      "learning_rate": 7.150021451524435e-06,
      "loss": 2.0381,
      "step": 214359
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.5588841438293457,
      "learning_rate": 7.1491337345963015e-06,
      "loss": 2.11,
      "step": 214360
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.4723526239395142,
      "learning_rate": 7.148246072114716e-06,
      "loss": 1.9651,
      "step": 214361
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.4888296127319336,
      "learning_rate": 7.1473584640797774e-06,
      "loss": 1.711,
      "step": 214362
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.46920645236969,
      "learning_rate": 7.1464709104916525e-06,
      "loss": 2.0926,
      "step": 214363
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.4073503017425537,
      "learning_rate": 7.145583411350575e-06,
      "loss": 1.8886,
      "step": 214364
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.4305933713912964,
      "learning_rate": 7.144695966656644e-06,
      "loss": 2.0108,
      "step": 214365
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.4168264865875244,
      "learning_rate": 7.143808576409993e-06,
      "loss": 1.9859,
      "step": 214366
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.4789018630981445,
      "learning_rate": 7.1429212406108884e-06,
      "loss": 1.9495,
      "step": 214367
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.4456368684768677,
      "learning_rate": 7.142033959259464e-06,
      "loss": 1.8173,
      "step": 214368
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.4264209270477295,
      "learning_rate": 7.141146732355818e-06,
      "loss": 2.0593,
      "step": 214369
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.530693531036377,
      "learning_rate": 7.1402595599001855e-06,
      "loss": 1.8006,
      "step": 214370
    },
    {
      "epoch": 2.79,
      "grad_norm": 2.3789992332458496,
      "learning_rate": 7.139372441892699e-06,
      "loss": 2.1074,
      "step": 214371
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.4167972803115845,
      "learning_rate": 7.138485378333492e-06,
      "loss": 1.9301,
      "step": 214372
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.5167663097381592,
      "learning_rate": 7.13759836922283e-06,
      "loss": 2.0307,
      "step": 214373
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.4852595329284668,
      "learning_rate": 7.136711414560814e-06,
      "loss": 1.6851,
      "step": 214374
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.3630790710449219,
      "learning_rate": 7.135824514347544e-06,
      "loss": 2.1529,
      "step": 214375
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.4752981662750244,
      "learning_rate": 7.1349376685833185e-06,
      "loss": 1.9744,
      "step": 214376
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.3723533153533936,
      "learning_rate": 7.134050877268205e-06,
      "loss": 2.021,
      "step": 214377
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.501896619796753,
      "learning_rate": 7.133164140402403e-06,
      "loss": 2.1842,
      "step": 214378
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.3135923147201538,
      "learning_rate": 7.132277457986113e-06,
      "loss": 1.8058,
      "step": 214379
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.5147180557250977,
      "learning_rate": 7.131390830019434e-06,
      "loss": 1.9298,
      "step": 214380
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.6064958572387695,
      "learning_rate": 7.1305042565025336e-06,
      "loss": 2.1086,
      "step": 214381
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.4804072380065918,
      "learning_rate": 7.129617737435644e-06,
      "loss": 2.0677,
      "step": 214382
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.5568856000900269,
      "learning_rate": 7.128731272818833e-06,
      "loss": 2.0722,
      "step": 214383
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.4996356964111328,
      "learning_rate": 7.127844862652365e-06,
      "loss": 2.2167,
      "step": 214384
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.446563959121704,
      "learning_rate": 7.126958506936309e-06,
      "loss": 1.8286,
      "step": 214385
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.4930788278579712,
      "learning_rate": 7.126072205670963e-06,
      "loss": 1.9969,
      "step": 214386
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.6767569780349731,
      "learning_rate": 7.1251859588563275e-06,
      "loss": 1.8245,
      "step": 214387
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.5055387020111084,
      "learning_rate": 7.124299766492703e-06,
      "loss": 1.9662,
      "step": 214388
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.4436675310134888,
      "learning_rate": 7.123413628580155e-06,
      "loss": 2.0485,
      "step": 214389
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.3534793853759766,
      "learning_rate": 7.12252754511895e-06,
      "loss": 1.7923,
      "step": 214390
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.4764010906219482,
      "learning_rate": 7.121641516109122e-06,
      "loss": 1.9027,
      "step": 214391
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.6571279764175415,
      "learning_rate": 7.120755541550971e-06,
      "loss": 2.3536,
      "step": 214392
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.3542383909225464,
      "learning_rate": 7.119869621444596e-06,
      "loss": 1.7761,
      "step": 214393
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.6660199165344238,
      "learning_rate": 7.1189837557901305e-06,
      "loss": 2.1161,
      "step": 214394
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.4606328010559082,
      "learning_rate": 7.1180979445877755e-06,
      "loss": 1.9439,
      "step": 214395
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.6332647800445557,
      "learning_rate": 7.117212187837729e-06,
      "loss": 1.9324,
      "step": 214396
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.3782750368118286,
      "learning_rate": 7.116326485540091e-06,
      "loss": 2.0942,
      "step": 214397
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.5198115110397339,
      "learning_rate": 7.115440837695064e-06,
      "loss": 1.9911,
      "step": 214398
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.493080496788025,
      "learning_rate": 7.114555244302844e-06,
      "loss": 1.9322,
      "step": 214399
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.406372308731079,
      "learning_rate": 7.113669705363501e-06,
      "loss": 1.9058,
      "step": 214400
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.6318820714950562,
      "learning_rate": 7.112784220877266e-06,
      "loss": 1.9622,
      "step": 214401
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.480955958366394,
      "learning_rate": 7.111898790844306e-06,
      "loss": 2.0564,
      "step": 214402
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.4237409830093384,
      "learning_rate": 7.111013415264755e-06,
      "loss": 2.056,
      "step": 214403
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.5412317514419556,
      "learning_rate": 7.110128094138845e-06,
      "loss": 1.8893,
      "step": 214404
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.4281750917434692,
      "learning_rate": 7.109242827466643e-06,
      "loss": 2.0541,
      "step": 214405
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.544830560684204,
      "learning_rate": 7.108357615248384e-06,
      "loss": 1.8724,
      "step": 214406
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.5293394327163696,
      "learning_rate": 7.1074724574841976e-06,
      "loss": 1.8412,
      "step": 214407
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.4320815801620483,
      "learning_rate": 7.106587354174287e-06,
      "loss": 2.0932,
      "step": 214408
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.45262610912323,
      "learning_rate": 7.10570230531875e-06,
      "loss": 1.8034,
      "step": 214409
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.5379678010940552,
      "learning_rate": 7.104817310917821e-06,
      "loss": 1.8728,
      "step": 214410
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.3968273401260376,
      "learning_rate": 7.1039323709716325e-06,
      "loss": 1.9472,
      "step": 214411
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.4456603527069092,
      "learning_rate": 7.103047485480351e-06,
      "loss": 1.9078,
      "step": 214412
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.4162729978561401,
      "learning_rate": 7.10216265444411e-06,
      "loss": 2.1894,
      "step": 214413
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.4732013940811157,
      "learning_rate": 7.101277877863143e-06,
      "loss": 2.109,
      "step": 214414
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.437694787979126,
      "learning_rate": 7.100393155737549e-06,
      "loss": 1.903,
      "step": 214415
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.4297661781311035,
      "learning_rate": 7.099508488067529e-06,
      "loss": 1.6698,
      "step": 214416
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.5804307460784912,
      "learning_rate": 7.098623874853282e-06,
      "loss": 1.7965,
      "step": 214417
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.4484153985977173,
      "learning_rate": 7.097739316094875e-06,
      "loss": 1.8842,
      "step": 214418
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.465390920639038,
      "learning_rate": 7.09685481179254e-06,
      "loss": 1.7441,
      "step": 214419
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.5761806964874268,
      "learning_rate": 7.095970361946446e-06,
      "loss": 2.0604,
      "step": 214420
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.5579348802566528,
      "learning_rate": 7.095085966556724e-06,
      "loss": 2.0235,
      "step": 214421
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.521377444267273,
      "learning_rate": 7.094201625623575e-06,
      "loss": 1.9655,
      "step": 214422
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.57733154296875,
      "learning_rate": 7.093317339147131e-06,
      "loss": 1.8336,
      "step": 214423
    },
    {
      "epoch": 2.79,
      "grad_norm": 2.383542776107788,
      "learning_rate": 7.09243310712756e-06,
      "loss": 1.9614,
      "step": 214424
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.5475196838378906,
      "learning_rate": 7.0915489295650274e-06,
      "loss": 2.0145,
      "step": 214425
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.515794277191162,
      "learning_rate": 7.090664806459734e-06,
      "loss": 2.0676,
      "step": 214426
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.6505333185195923,
      "learning_rate": 7.089780737811779e-06,
      "loss": 1.9461,
      "step": 214427
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.5639283657073975,
      "learning_rate": 7.088896723621395e-06,
      "loss": 2.1178,
      "step": 214428
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.4802404642105103,
      "learning_rate": 7.0880127638887175e-06,
      "loss": 1.8341,
      "step": 214429
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.5521290302276611,
      "learning_rate": 7.0871288586138775e-06,
      "loss": 2.1541,
      "step": 214430
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.513539433479309,
      "learning_rate": 7.086245007797075e-06,
      "loss": 2.1787,
      "step": 214431
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.4417552947998047,
      "learning_rate": 7.0853612114384784e-06,
      "loss": 2.0971,
      "step": 214432
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.4667776823043823,
      "learning_rate": 7.084477469538219e-06,
      "loss": 2.1791,
      "step": 214433
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.4834840297698975,
      "learning_rate": 7.08359378209653e-06,
      "loss": 2.0568,
      "step": 214434
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.5918469429016113,
      "learning_rate": 7.082710149113513e-06,
      "loss": 1.963,
      "step": 214435
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.5317412614822388,
      "learning_rate": 7.0818265705893665e-06,
      "loss": 2.0755,
      "step": 214436
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.5462827682495117,
      "learning_rate": 7.0809430465241905e-06,
      "loss": 1.7081,
      "step": 214437
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.3267226219177246,
      "learning_rate": 7.080059576918218e-06,
      "loss": 1.987,
      "step": 214438
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.5792980194091797,
      "learning_rate": 7.079176161771583e-06,
      "loss": 2.158,
      "step": 214439
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.5555223226547241,
      "learning_rate": 7.078292801084484e-06,
      "loss": 2.138,
      "step": 214440
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.3626741170883179,
      "learning_rate": 7.077409494857056e-06,
      "loss": 1.8757,
      "step": 214441
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.5629750490188599,
      "learning_rate": 7.076526243089464e-06,
      "loss": 2.0714,
      "step": 214442
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.4220190048217773,
      "learning_rate": 7.075643045781843e-06,
      "loss": 2.0776,
      "step": 214443
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.3982319831848145,
      "learning_rate": 7.074759902934424e-06,
      "loss": 2.0979,
      "step": 214444
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.4512813091278076,
      "learning_rate": 7.073876814547308e-06,
      "loss": 2.1322,
      "step": 214445
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.5223766565322876,
      "learning_rate": 7.072993780620728e-06,
      "loss": 2.0736,
      "step": 214446
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.4680677652359009,
      "learning_rate": 7.072110801154784e-06,
      "loss": 2.0477,
      "step": 214447
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.4246052503585815,
      "learning_rate": 7.071227876149644e-06,
      "loss": 1.9187,
      "step": 214448
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.4203733205795288,
      "learning_rate": 7.0703450056055375e-06,
      "loss": 2.198,
      "step": 214449
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.4888654947280884,
      "learning_rate": 7.0694621895225674e-06,
      "loss": 1.9841,
      "step": 214450
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.4923821687698364,
      "learning_rate": 7.0685794279009e-06,
      "loss": 1.886,
      "step": 214451
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.4576071500778198,
      "learning_rate": 7.067696720740735e-06,
      "loss": 1.7792,
      "step": 214452
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.5529727935791016,
      "learning_rate": 7.066814068042204e-06,
      "loss": 1.917,
      "step": 214453
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.4165120124816895,
      "learning_rate": 7.065931469805441e-06,
      "loss": 1.8781,
      "step": 214454
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.4778540134429932,
      "learning_rate": 7.065048926030714e-06,
      "loss": 1.9683,
      "step": 214455
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.365938663482666,
      "learning_rate": 7.064166436718088e-06,
      "loss": 1.8561,
      "step": 214456
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.3769989013671875,
      "learning_rate": 7.063284001867764e-06,
      "loss": 2.0218,
      "step": 214457
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.451459527015686,
      "learning_rate": 7.062401621479907e-06,
      "loss": 2.1096,
      "step": 214458
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.34267258644104,
      "learning_rate": 7.061519295554685e-06,
      "loss": 1.8997,
      "step": 214459
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.4592857360839844,
      "learning_rate": 7.060637024092264e-06,
      "loss": 1.8103,
      "step": 214460
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.3878309726715088,
      "learning_rate": 7.05975480709281e-06,
      "loss": 2.0937,
      "step": 214461
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.4086452722549438,
      "learning_rate": 7.058872644556423e-06,
      "loss": 1.8965,
      "step": 214462
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.397015929222107,
      "learning_rate": 7.057990536483371e-06,
      "loss": 2.0847,
      "step": 214463
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.5815056562423706,
      "learning_rate": 7.0571084828737854e-06,
      "loss": 1.9458,
      "step": 214464
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.4335170984268188,
      "learning_rate": 7.056226483727767e-06,
      "loss": 2.1027,
      "step": 214465
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.5091692209243774,
      "learning_rate": 7.055344539045549e-06,
      "loss": 1.951,
      "step": 214466
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.4366544485092163,
      "learning_rate": 7.054462648827264e-06,
      "loss": 2.1006,
      "step": 214467
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.4064737558364868,
      "learning_rate": 7.053580813073079e-06,
      "loss": 1.7677,
      "step": 214468
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.5822144746780396,
      "learning_rate": 7.05269903178316e-06,
      "loss": 1.9797,
      "step": 214469
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.4379981756210327,
      "learning_rate": 7.051817304957674e-06,
      "loss": 1.9055,
      "step": 214470
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.5553220510482788,
      "learning_rate": 7.050935632596788e-06,
      "loss": 1.9241,
      "step": 214471
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.5089099407196045,
      "learning_rate": 7.050054014700701e-06,
      "loss": 2.1166,
      "step": 214472
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.5463554859161377,
      "learning_rate": 7.0491724512695135e-06,
      "loss": 1.8273,
      "step": 214473
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.3246594667434692,
      "learning_rate": 7.048290942303391e-06,
      "loss": 2.0202,
      "step": 214474
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.4694575071334839,
      "learning_rate": 7.047409487802536e-06,
      "loss": 2.1441,
      "step": 214475
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.4156594276428223,
      "learning_rate": 7.046528087767078e-06,
      "loss": 1.9845,
      "step": 214476
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.516058325767517,
      "learning_rate": 7.045646742197253e-06,
      "loss": 1.9315,
      "step": 214477
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.587615728378296,
      "learning_rate": 7.044765451093159e-06,
      "loss": 1.8789,
      "step": 214478
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.3836079835891724,
      "learning_rate": 7.043884214454964e-06,
      "loss": 2.0352,
      "step": 214479
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.4841244220733643,
      "learning_rate": 7.043003032282802e-06,
      "loss": 2.1911,
      "step": 214480
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.5279786586761475,
      "learning_rate": 7.042121904576903e-06,
      "loss": 2.0093,
      "step": 214481
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.4914551973342896,
      "learning_rate": 7.041240831337403e-06,
      "loss": 2.1205,
      "step": 214482
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.3621764183044434,
      "learning_rate": 7.040359812564467e-06,
      "loss": 2.0245,
      "step": 214483
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.7010146379470825,
      "learning_rate": 7.0394788482582955e-06,
      "loss": 1.9498,
      "step": 214484
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.4161605834960938,
      "learning_rate": 7.038597938418988e-06,
      "loss": 2.1433,
      "step": 214485
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.46235191822052,
      "learning_rate": 7.037717083046712e-06,
      "loss": 1.8297,
      "step": 214486
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.5404776334762573,
      "learning_rate": 7.0368362821416665e-06,
      "loss": 2.1196,
      "step": 214487
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.4846364259719849,
      "learning_rate": 7.035955535703985e-06,
      "loss": 2.0514,
      "step": 214488
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.5325261354446411,
      "learning_rate": 7.035074843733901e-06,
      "loss": 1.7898,
      "step": 214489
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.5270277261734009,
      "learning_rate": 7.03419420623148e-06,
      "loss": 1.9953,
      "step": 214490
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.5435656309127808,
      "learning_rate": 7.03331362319699e-06,
      "loss": 1.7774,
      "step": 214491
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.4149583578109741,
      "learning_rate": 7.032433094630463e-06,
      "loss": 1.9531,
      "step": 214492
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.5895426273345947,
      "learning_rate": 7.031552620532166e-06,
      "loss": 1.7731,
      "step": 214493
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.4585411548614502,
      "learning_rate": 7.030672200902232e-06,
      "loss": 2.0157,
      "step": 214494
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.456632137298584,
      "learning_rate": 7.029791835740861e-06,
      "loss": 2.1155,
      "step": 214495
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.5424280166625977,
      "learning_rate": 7.028911525048153e-06,
      "loss": 2.0321,
      "step": 214496
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.4367927312850952,
      "learning_rate": 7.028031268824308e-06,
      "loss": 1.9455,
      "step": 214497
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.3915724754333496,
      "learning_rate": 7.027151067069458e-06,
      "loss": 1.822,
      "step": 214498
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.405348539352417,
      "learning_rate": 7.026270919783805e-06,
      "loss": 1.9621,
      "step": 214499
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.486679196357727,
      "learning_rate": 7.025390826967481e-06,
      "loss": 2.0764,
      "step": 214500
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.3985695838928223,
      "learning_rate": 7.0245107886207184e-06,
      "loss": 1.9168,
      "step": 214501
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.4072929620742798,
      "learning_rate": 7.023630804743618e-06,
      "loss": 1.8933,
      "step": 214502
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.4820234775543213,
      "learning_rate": 7.022750875336347e-06,
      "loss": 1.9939,
      "step": 214503
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.4703447818756104,
      "learning_rate": 7.0218710003990375e-06,
      "loss": 1.943,
      "step": 214504
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.4318960905075073,
      "learning_rate": 7.020991179931923e-06,
      "loss": 1.7226,
      "step": 214505
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.4272937774658203,
      "learning_rate": 7.020111413935137e-06,
      "loss": 1.8585,
      "step": 214506
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.4809544086456299,
      "learning_rate": 7.0192317024088455e-06,
      "loss": 1.8835,
      "step": 214507
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.4602254629135132,
      "learning_rate": 7.018352045353215e-06,
      "loss": 2.3414,
      "step": 214508
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.5517473220825195,
      "learning_rate": 7.017472442768379e-06,
      "loss": 2.0801,
      "step": 214509
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.7113454341888428,
      "learning_rate": 7.016592894654538e-06,
      "loss": 2.0782,
      "step": 214510
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.5545483827590942,
      "learning_rate": 7.015713401011858e-06,
      "loss": 1.8565,
      "step": 214511
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.3946460485458374,
      "learning_rate": 7.014833961840438e-06,
      "loss": 1.957,
      "step": 214512
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.542284369468689,
      "learning_rate": 7.013954577140546e-06,
      "loss": 1.9299,
      "step": 214513
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.3935397863388062,
      "learning_rate": 7.013075246912281e-06,
      "loss": 1.9035,
      "step": 214514
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.3780218362808228,
      "learning_rate": 7.0121959711558094e-06,
      "loss": 1.6799,
      "step": 214515
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.522189974784851,
      "learning_rate": 7.0113167498712985e-06,
      "loss": 1.9248,
      "step": 214516
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.9140182733535767,
      "learning_rate": 7.010437583058914e-06,
      "loss": 2.063,
      "step": 214517
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.369657278060913,
      "learning_rate": 7.0095584707187906e-06,
      "loss": 1.9808,
      "step": 214518
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.3976120948791504,
      "learning_rate": 7.0086794128511595e-06,
      "loss": 1.9756,
      "step": 214519
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.6781418323516846,
      "learning_rate": 7.0078004094561545e-06,
      "loss": 1.9328,
      "step": 214520
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.4377107620239258,
      "learning_rate": 7.0069214605339096e-06,
      "loss": 1.8734,
      "step": 214521
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.3695151805877686,
      "learning_rate": 7.006042566084591e-06,
      "loss": 1.8383,
      "step": 214522
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.3716901540756226,
      "learning_rate": 7.0051637261084314e-06,
      "loss": 1.9613,
      "step": 214523
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.4074070453643799,
      "learning_rate": 7.004284940605464e-06,
      "loss": 2.1065,
      "step": 214524
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.5433809757232666,
      "learning_rate": 7.0034062095759906e-06,
      "loss": 2.0834,
      "step": 214525
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.5705794095993042,
      "learning_rate": 7.002527533020108e-06,
      "loss": 2.0017,
      "step": 214526
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.5099906921386719,
      "learning_rate": 7.0016489109380185e-06,
      "loss": 1.9739,
      "step": 214527
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.6436258554458618,
      "learning_rate": 7.000770343329787e-06,
      "loss": 1.9173,
      "step": 214528
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.8366109132766724,
      "learning_rate": 6.999891830195681e-06,
      "loss": 2.1177,
      "step": 214529
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.6905157566070557,
      "learning_rate": 6.9990133715358e-06,
      "loss": 2.1131,
      "step": 214530
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.5373889207839966,
      "learning_rate": 6.998134967350344e-06,
      "loss": 1.9288,
      "step": 214531
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.4423916339874268,
      "learning_rate": 6.997256617639446e-06,
      "loss": 2.0486,
      "step": 214532
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.4158307313919067,
      "learning_rate": 6.996378322403373e-06,
      "loss": 1.8652,
      "step": 214533
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.4623029232025146,
      "learning_rate": 6.995500081642125e-06,
      "loss": 1.8372,
      "step": 214534
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.5523688793182373,
      "learning_rate": 6.9946218953559675e-06,
      "loss": 2.1594,
      "step": 214535
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.4299036264419556,
      "learning_rate": 6.993743763545001e-06,
      "loss": 1.886,
      "step": 214536
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.6584349870681763,
      "learning_rate": 6.992865686209492e-06,
      "loss": 2.065,
      "step": 214537
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.5549449920654297,
      "learning_rate": 6.991987663349474e-06,
      "loss": 2.1362,
      "step": 214538
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.6056506633758545,
      "learning_rate": 6.991109694965247e-06,
      "loss": 2.2452,
      "step": 214539
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.7990808486938477,
      "learning_rate": 6.990231781056876e-06,
      "loss": 1.9399,
      "step": 214540
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.5197947025299072,
      "learning_rate": 6.989353921624563e-06,
      "loss": 2.104,
      "step": 214541
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.4451029300689697,
      "learning_rate": 6.988476116668407e-06,
      "loss": 2.0147,
      "step": 214542
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.5008333921432495,
      "learning_rate": 6.987598366188673e-06,
      "loss": 2.2362,
      "step": 214543
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.5452815294265747,
      "learning_rate": 6.986720670185464e-06,
      "loss": 2.0293,
      "step": 214544
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.366922378540039,
      "learning_rate": 6.985843028658944e-06,
      "loss": 1.9439,
      "step": 214545
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.3824093341827393,
      "learning_rate": 6.9849654416093135e-06,
      "loss": 1.7821,
      "step": 214546
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.5139867067337036,
      "learning_rate": 6.984087909036673e-06,
      "loss": 2.0238,
      "step": 214547
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.6328016519546509,
      "learning_rate": 6.9832104309412555e-06,
      "loss": 2.0687,
      "step": 214548
    },
    {
      "epoch": 2.79,
      "grad_norm": 2.1079745292663574,
      "learning_rate": 6.982333007323193e-06,
      "loss": 1.7032,
      "step": 214549
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.5519793033599854,
      "learning_rate": 6.981455638182587e-06,
      "loss": 2.1394,
      "step": 214550
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.5302220582962036,
      "learning_rate": 6.980578323519737e-06,
      "loss": 2.011,
      "step": 214551
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.5134319067001343,
      "learning_rate": 6.979701063334708e-06,
      "loss": 2.068,
      "step": 214552
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.4400906562805176,
      "learning_rate": 6.978823857627636e-06,
      "loss": 1.9245,
      "step": 214553
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.4874790906906128,
      "learning_rate": 6.977946706398785e-06,
      "loss": 2.1083,
      "step": 214554
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.5739270448684692,
      "learning_rate": 6.977069609648256e-06,
      "loss": 1.8872,
      "step": 214555
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.5350329875946045,
      "learning_rate": 6.976192567376182e-06,
      "loss": 2.0854,
      "step": 214556
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.5507607460021973,
      "learning_rate": 6.975315579582797e-06,
      "loss": 1.8908,
      "step": 214557
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.5100451707839966,
      "learning_rate": 6.974438646268232e-06,
      "loss": 1.9727,
      "step": 214558
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.4295713901519775,
      "learning_rate": 6.9735617674326224e-06,
      "loss": 1.7905,
      "step": 214559
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.6343631744384766,
      "learning_rate": 6.972684943076201e-06,
      "loss": 1.9648,
      "step": 214560
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.351548194885254,
      "learning_rate": 6.971808173199067e-06,
      "loss": 1.8908,
      "step": 214561
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.5349355936050415,
      "learning_rate": 6.970931457801421e-06,
      "loss": 1.9393,
      "step": 214562
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.4003193378448486,
      "learning_rate": 6.9700547968833955e-06,
      "loss": 1.9961,
      "step": 214563
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.3326423168182373,
      "learning_rate": 6.96917819044519e-06,
      "loss": 2.0705,
      "step": 214564
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.513564944267273,
      "learning_rate": 6.968301638486906e-06,
      "loss": 2.1889,
      "step": 214565
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.5152349472045898,
      "learning_rate": 6.967425141008775e-06,
      "loss": 1.8902,
      "step": 214566
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.3977783918380737,
      "learning_rate": 6.966548698010898e-06,
      "loss": 1.9484,
      "step": 214567
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.4627047777175903,
      "learning_rate": 6.965672309493509e-06,
      "loss": 1.9396,
      "step": 214568
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.4464983940124512,
      "learning_rate": 6.964795975456738e-06,
      "loss": 2.0683,
      "step": 214569
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.5535153150558472,
      "learning_rate": 6.963919695900722e-06,
      "loss": 2.0012,
      "step": 214570
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.454706072807312,
      "learning_rate": 6.963043470825625e-06,
      "loss": 2.1056,
      "step": 214571
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.3939330577850342,
      "learning_rate": 6.96216730023168e-06,
      "loss": 1.9825,
      "step": 214572
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.7367141246795654,
      "learning_rate": 6.9612911841189556e-06,
      "loss": 2.2906,
      "step": 214573
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.5470975637435913,
      "learning_rate": 6.960415122487683e-06,
      "loss": 2.0838,
      "step": 214574
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.4360523223876953,
      "learning_rate": 6.959539115337997e-06,
      "loss": 1.9524,
      "step": 214575
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.4495958089828491,
      "learning_rate": 6.958663162670064e-06,
      "loss": 2.0664,
      "step": 214576
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.4770784378051758,
      "learning_rate": 6.957787264484016e-06,
      "loss": 2.0422,
      "step": 214577
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.426095962524414,
      "learning_rate": 6.956911420780087e-06,
      "loss": 1.8276,
      "step": 214578
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.451176404953003,
      "learning_rate": 6.956035631558376e-06,
      "loss": 1.901,
      "step": 214579
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.5141830444335938,
      "learning_rate": 6.955159896819084e-06,
      "loss": 2.0262,
      "step": 214580
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.4158427715301514,
      "learning_rate": 6.954284216562378e-06,
      "loss": 1.8353,
      "step": 214581
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.443953037261963,
      "learning_rate": 6.95340859078839e-06,
      "loss": 1.7544,
      "step": 214582
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.5098742246627808,
      "learning_rate": 6.952533019497253e-06,
      "loss": 1.9936,
      "step": 214583
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.5405670404434204,
      "learning_rate": 6.951657502689234e-06,
      "loss": 1.9691,
      "step": 214584
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.457499623298645,
      "learning_rate": 6.950782040364367e-06,
      "loss": 1.7796,
      "step": 214585
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.368399977684021,
      "learning_rate": 6.949906632522917e-06,
      "loss": 1.8484,
      "step": 214586
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.519499659538269,
      "learning_rate": 6.9490312791650515e-06,
      "loss": 1.7952,
      "step": 214587
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.5129058361053467,
      "learning_rate": 6.948155980290837e-06,
      "loss": 1.7903,
      "step": 214588
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.470729112625122,
      "learning_rate": 6.947280735900507e-06,
      "loss": 1.9464,
      "step": 214589
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.296482801437378,
      "learning_rate": 6.946405545994227e-06,
      "loss": 2.0553,
      "step": 214590
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.5958017110824585,
      "learning_rate": 6.945530410572098e-06,
      "loss": 2.1411,
      "step": 214591
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.4956624507904053,
      "learning_rate": 6.944655329634352e-06,
      "loss": 2.0768,
      "step": 214592
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.5253440141677856,
      "learning_rate": 6.943780303181123e-06,
      "loss": 1.845,
      "step": 214593
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.459852933883667,
      "learning_rate": 6.942905331212611e-06,
      "loss": 1.7218,
      "step": 214594
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.4939169883728027,
      "learning_rate": 6.942030413728917e-06,
      "loss": 2.0914,
      "step": 214595
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.4777005910873413,
      "learning_rate": 6.941155550730238e-06,
      "loss": 1.9856,
      "step": 214596
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.54292893409729,
      "learning_rate": 6.940280742216708e-06,
      "loss": 1.913,
      "step": 214597
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.3631850481033325,
      "learning_rate": 6.9394059881885285e-06,
      "loss": 1.8443,
      "step": 214598
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.410971760749817,
      "learning_rate": 6.938531288645832e-06,
      "loss": 2.1551,
      "step": 214599
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.582139253616333,
      "learning_rate": 6.937656643588851e-06,
      "loss": 2.0903,
      "step": 214600
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.3535932302474976,
      "learning_rate": 6.93678205301762e-06,
      "loss": 1.8801,
      "step": 214601
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.4268455505371094,
      "learning_rate": 6.935907516932404e-06,
      "loss": 2.0017,
      "step": 214602
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.4884637594223022,
      "learning_rate": 6.935033035333304e-06,
      "loss": 2.0666,
      "step": 214603
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.4844791889190674,
      "learning_rate": 6.934158608220552e-06,
      "loss": 1.7146,
      "step": 214604
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.5187511444091797,
      "learning_rate": 6.933284235594249e-06,
      "loss": 2.0067,
      "step": 214605
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.373358964920044,
      "learning_rate": 6.932409917454629e-06,
      "loss": 1.9743,
      "step": 214606
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.3871123790740967,
      "learning_rate": 6.931535653801757e-06,
      "loss": 1.976,
      "step": 214607
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.3569464683532715,
      "learning_rate": 6.930661444635866e-06,
      "loss": 2.2477,
      "step": 214608
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.3912078142166138,
      "learning_rate": 6.929787289957056e-06,
      "loss": 1.6902,
      "step": 214609
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.252077579498291,
      "learning_rate": 6.928913189765595e-06,
      "loss": 2.0892,
      "step": 214610
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.5620774030685425,
      "learning_rate": 6.928039144061515e-06,
      "loss": 2.2035,
      "step": 214611
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.4920382499694824,
      "learning_rate": 6.927165152845115e-06,
      "loss": 2.1095,
      "step": 214612
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.6262186765670776,
      "learning_rate": 6.926291216116431e-06,
      "loss": 1.9881,
      "step": 214613
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.5712003707885742,
      "learning_rate": 6.925417333875727e-06,
      "loss": 2.032,
      "step": 214614
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.4659854173660278,
      "learning_rate": 6.92454350612307e-06,
      "loss": 2.08,
      "step": 214615
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.478488564491272,
      "learning_rate": 6.923669732858694e-06,
      "loss": 1.8848,
      "step": 214616
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.3332254886627197,
      "learning_rate": 6.9227960140827324e-06,
      "loss": 2.0182,
      "step": 214617
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.4836143255233765,
      "learning_rate": 6.921922349795417e-06,
      "loss": 2.1605,
      "step": 214618
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.4220014810562134,
      "learning_rate": 6.921048739996749e-06,
      "loss": 2.0413,
      "step": 214619
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.3490787744522095,
      "learning_rate": 6.92017518468706e-06,
      "loss": 1.9043,
      "step": 214620
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.580459475517273,
      "learning_rate": 6.919301683866418e-06,
      "loss": 1.858,
      "step": 214621
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.5702823400497437,
      "learning_rate": 6.918428237535023e-06,
      "loss": 1.9132,
      "step": 214622
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.4860538244247437,
      "learning_rate": 6.917554845692974e-06,
      "loss": 2.0409,
      "step": 214623
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.5103422403335571,
      "learning_rate": 6.916681508340538e-06,
      "loss": 1.9758,
      "step": 214624
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.6109613180160522,
      "learning_rate": 6.915808225477814e-06,
      "loss": 2.0368,
      "step": 214625
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.4604545831680298,
      "learning_rate": 6.9149349971050036e-06,
      "loss": 2.1077,
      "step": 214626
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.5018912553787231,
      "learning_rate": 6.914061823222172e-06,
      "loss": 2.0205,
      "step": 214627
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.4645367860794067,
      "learning_rate": 6.913188703829586e-06,
      "loss": 2.0148,
      "step": 214628
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.4217735528945923,
      "learning_rate": 6.912315638927346e-06,
      "loss": 2.0593,
      "step": 214629
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.4044116735458374,
      "learning_rate": 6.911442628515684e-06,
      "loss": 2.0401,
      "step": 214630
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.4998509883880615,
      "learning_rate": 6.9105696725947015e-06,
      "loss": 1.9021,
      "step": 214631
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.5435473918914795,
      "learning_rate": 6.909696771164564e-06,
      "loss": 1.8739,
      "step": 214632
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.5526036024093628,
      "learning_rate": 6.908823924225438e-06,
      "loss": 2.02,
      "step": 214633
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.4500848054885864,
      "learning_rate": 6.9079511317775225e-06,
      "loss": 1.9345,
      "step": 214634
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.6101433038711548,
      "learning_rate": 6.90707839382092e-06,
      "loss": 1.7482,
      "step": 214635
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.651026964187622,
      "learning_rate": 6.906205710355861e-06,
      "loss": 2.1253,
      "step": 214636
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.3552063703536987,
      "learning_rate": 6.905333081382447e-06,
      "loss": 2.0774,
      "step": 214637
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.4577958583831787,
      "learning_rate": 6.904460506900844e-06,
      "loss": 1.9751,
      "step": 214638
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.5561213493347168,
      "learning_rate": 6.903587986911285e-06,
      "loss": 2.114,
      "step": 214639
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.4918560981750488,
      "learning_rate": 6.902715521413837e-06,
      "loss": 2.1001,
      "step": 214640
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.469618558883667,
      "learning_rate": 6.901843110408734e-06,
      "loss": 2.0896,
      "step": 214641
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.3809877634048462,
      "learning_rate": 6.900970753896106e-06,
      "loss": 1.8146,
      "step": 214642
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.392296552658081,
      "learning_rate": 6.900098451876124e-06,
      "loss": 1.8329,
      "step": 214643
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.6025928258895874,
      "learning_rate": 6.899226204348918e-06,
      "loss": 2.1386,
      "step": 214644
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.417709231376648,
      "learning_rate": 6.898354011314689e-06,
      "loss": 2.1316,
      "step": 214645
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.479901909828186,
      "learning_rate": 6.897481872773603e-06,
      "loss": 2.1252,
      "step": 214646
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.565303087234497,
      "learning_rate": 6.896609788725793e-06,
      "loss": 2.2636,
      "step": 214647
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.3652139902114868,
      "learning_rate": 6.89573775917146e-06,
      "loss": 2.059,
      "step": 214648
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.3614022731781006,
      "learning_rate": 6.894865784110737e-06,
      "loss": 2.0139,
      "step": 214649
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.5250599384307861,
      "learning_rate": 6.893993863543756e-06,
      "loss": 1.9431,
      "step": 214650
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.448961853981018,
      "learning_rate": 6.893121997470752e-06,
      "loss": 1.9426,
      "step": 214651
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.3516138792037964,
      "learning_rate": 6.892250185891823e-06,
      "loss": 2.1969,
      "step": 214652
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.5360002517700195,
      "learning_rate": 6.891378428807171e-06,
      "loss": 2.0331,
      "step": 214653
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.4754502773284912,
      "learning_rate": 6.89050672621696e-06,
      "loss": 2.0876,
      "step": 214654
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.4010944366455078,
      "learning_rate": 6.889635078121325e-06,
      "loss": 1.8057,
      "step": 214655
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.5002132654190063,
      "learning_rate": 6.888763484520432e-06,
      "loss": 1.8975,
      "step": 214656
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.4544025659561157,
      "learning_rate": 6.887891945414481e-06,
      "loss": 2.2825,
      "step": 214657
    },
    {
      "epoch": 2.79,
      "grad_norm": 1.3760563135147095,
      "learning_rate": 6.887020460803538e-06,
      "loss": 1.9381,
      "step": 214658
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.3584295511245728,
      "learning_rate": 6.8861490306879045e-06,
      "loss": 1.8917,
      "step": 214659
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.627960443496704,
      "learning_rate": 6.885277655067612e-06,
      "loss": 2.0794,
      "step": 214660
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.483018398284912,
      "learning_rate": 6.884406333942927e-06,
      "loss": 1.8455,
      "step": 214661
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.5508673191070557,
      "learning_rate": 6.883535067313917e-06,
      "loss": 1.9856,
      "step": 214662
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.4830018281936646,
      "learning_rate": 6.8826638551808475e-06,
      "loss": 1.8523,
      "step": 214663
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.4808093309402466,
      "learning_rate": 6.881792697543753e-06,
      "loss": 1.8188,
      "step": 214664
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.6879414319992065,
      "learning_rate": 6.8809215944029325e-06,
      "loss": 1.8674,
      "step": 214665
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.531035304069519,
      "learning_rate": 6.88005054575842e-06,
      "loss": 2.1616,
      "step": 214666
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.386567234992981,
      "learning_rate": 6.879179551610514e-06,
      "loss": 1.8056,
      "step": 214667
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.7403217554092407,
      "learning_rate": 6.878308611959249e-06,
      "loss": 1.86,
      "step": 214668
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.4159659147262573,
      "learning_rate": 6.877437726804858e-06,
      "loss": 1.8177,
      "step": 214669
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.5831960439682007,
      "learning_rate": 6.876566896147439e-06,
      "loss": 2.1045,
      "step": 214670
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.3981012105941772,
      "learning_rate": 6.875696119987262e-06,
      "loss": 1.9967,
      "step": 214671
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.4053188562393188,
      "learning_rate": 6.8748253983243906e-06,
      "loss": 1.9814,
      "step": 214672
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.4660298824310303,
      "learning_rate": 6.873954731159059e-06,
      "loss": 2.0179,
      "step": 214673
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.462460994720459,
      "learning_rate": 6.873084118491335e-06,
      "loss": 2.0681,
      "step": 214674
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.6042662858963013,
      "learning_rate": 6.872213560321482e-06,
      "loss": 2.0443,
      "step": 214675
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.4786425828933716,
      "learning_rate": 6.871343056649603e-06,
      "loss": 1.8072,
      "step": 214676
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.5464926958084106,
      "learning_rate": 6.870472607475863e-06,
      "loss": 1.9179,
      "step": 214677
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.5716660022735596,
      "learning_rate": 6.8696022128004294e-06,
      "loss": 2.0548,
      "step": 214678
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.6636929512023926,
      "learning_rate": 6.868731872623534e-06,
      "loss": 2.1286,
      "step": 214679
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.4689407348632812,
      "learning_rate": 6.867861586945211e-06,
      "loss": 2.0034,
      "step": 214680
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.7269946336746216,
      "learning_rate": 6.866991355765728e-06,
      "loss": 1.9867,
      "step": 214681
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.3564116954803467,
      "learning_rate": 6.866121179085149e-06,
      "loss": 1.8556,
      "step": 214682
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.4685211181640625,
      "learning_rate": 6.865251056903742e-06,
      "loss": 1.9671,
      "step": 214683
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.674235224723816,
      "learning_rate": 6.864380989221574e-06,
      "loss": 2.0186,
      "step": 214684
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.394809603691101,
      "learning_rate": 6.863510976038911e-06,
      "loss": 2.0149,
      "step": 214685
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.3897120952606201,
      "learning_rate": 6.862641017355819e-06,
      "loss": 2.027,
      "step": 214686
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.4554054737091064,
      "learning_rate": 6.861771113172498e-06,
      "loss": 1.9484,
      "step": 214687
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.531039834022522,
      "learning_rate": 6.860901263489083e-06,
      "loss": 2.138,
      "step": 214688
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.5043272972106934,
      "learning_rate": 6.860031468305804e-06,
      "loss": 2.092,
      "step": 214689
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.4403456449508667,
      "learning_rate": 6.85916172762273e-06,
      "loss": 1.8695,
      "step": 214690
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.5808064937591553,
      "learning_rate": 6.858292041440161e-06,
      "loss": 2.0127,
      "step": 214691
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.7745826244354248,
      "learning_rate": 6.857422409758062e-06,
      "loss": 2.0901,
      "step": 214692
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.5156967639923096,
      "learning_rate": 6.856552832576767e-06,
      "loss": 1.8697,
      "step": 214693
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.3933436870574951,
      "learning_rate": 6.855683309896343e-06,
      "loss": 2.0399,
      "step": 214694
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.511145830154419,
      "learning_rate": 6.8548138417169885e-06,
      "loss": 1.9873,
      "step": 214695
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.5258984565734863,
      "learning_rate": 6.853944428038838e-06,
      "loss": 1.9847,
      "step": 214696
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.4566926956176758,
      "learning_rate": 6.853075068862124e-06,
      "loss": 1.8403,
      "step": 214697
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.4590678215026855,
      "learning_rate": 6.8522057641869134e-06,
      "loss": 1.7956,
      "step": 214698
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.5355783700942993,
      "learning_rate": 6.851336514013439e-06,
      "loss": 2.0144,
      "step": 214699
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.628074049949646,
      "learning_rate": 6.850467318341802e-06,
      "loss": 1.9375,
      "step": 214700
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.3783557415008545,
      "learning_rate": 6.849598177172233e-06,
      "loss": 1.9669,
      "step": 214701
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.9081496000289917,
      "learning_rate": 6.848729090504801e-06,
      "loss": 2.1714,
      "step": 214702
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.4432995319366455,
      "learning_rate": 6.847860058339805e-06,
      "loss": 1.8512,
      "step": 214703
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.4744914770126343,
      "learning_rate": 6.846991080677278e-06,
      "loss": 1.8193,
      "step": 214704
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.477850079536438,
      "learning_rate": 6.846122157517453e-06,
      "loss": 1.9543,
      "step": 214705
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.5913697481155396,
      "learning_rate": 6.845253288860431e-06,
      "loss": 1.8129,
      "step": 214706
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.5125683546066284,
      "learning_rate": 6.8443844747064105e-06,
      "loss": 1.8957,
      "step": 214707
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.4990135431289673,
      "learning_rate": 6.843515715055559e-06,
      "loss": 1.7194,
      "step": 214708
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.4486192464828491,
      "learning_rate": 6.842647009908075e-06,
      "loss": 1.8027,
      "step": 214709
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.611463189125061,
      "learning_rate": 6.841778359264027e-06,
      "loss": 1.8823,
      "step": 214710
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.6785783767700195,
      "learning_rate": 6.840909763123681e-06,
      "loss": 1.881,
      "step": 214711
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.5291318893432617,
      "learning_rate": 6.8400412214870694e-06,
      "loss": 1.9013,
      "step": 214712
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.602259635925293,
      "learning_rate": 6.839172734354492e-06,
      "loss": 2.0842,
      "step": 214713
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.5176056623458862,
      "learning_rate": 6.838304301725983e-06,
      "loss": 2.0537,
      "step": 214714
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.4132076501846313,
      "learning_rate": 6.837435923601808e-06,
      "loss": 1.9813,
      "step": 214715
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.5723984241485596,
      "learning_rate": 6.8365675999821015e-06,
      "loss": 1.9878,
      "step": 214716
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.7201697826385498,
      "learning_rate": 6.835699330866995e-06,
      "loss": 2.0802,
      "step": 214717
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.4638038873672485,
      "learning_rate": 6.834831116256656e-06,
      "loss": 2.0025,
      "step": 214718
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.612894058227539,
      "learning_rate": 6.833962956151284e-06,
      "loss": 2.1877,
      "step": 214719
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.68910813331604,
      "learning_rate": 6.833094850550946e-06,
      "loss": 1.9406,
      "step": 214720
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.4515317678451538,
      "learning_rate": 6.8322267994559425e-06,
      "loss": 2.0116,
      "step": 214721
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.5638134479522705,
      "learning_rate": 6.831358802866338e-06,
      "loss": 1.9114,
      "step": 214722
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.3988592624664307,
      "learning_rate": 6.8304908607823005e-06,
      "loss": 1.7373,
      "step": 214723
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.442270278930664,
      "learning_rate": 6.82962297320403e-06,
      "loss": 2.0597,
      "step": 214724
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.4068129062652588,
      "learning_rate": 6.828755140131659e-06,
      "loss": 2.0583,
      "step": 214725
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.438997745513916,
      "learning_rate": 6.827887361565354e-06,
      "loss": 1.9276,
      "step": 214726
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.61941397190094,
      "learning_rate": 6.827019637505282e-06,
      "loss": 2.1426,
      "step": 214727
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.459999680519104,
      "learning_rate": 6.82615196795161e-06,
      "loss": 2.168,
      "step": 214728
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.4907665252685547,
      "learning_rate": 6.82528435290447e-06,
      "loss": 2.1887,
      "step": 214729
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.5541925430297852,
      "learning_rate": 6.8244167923640626e-06,
      "loss": 1.9622,
      "step": 214730
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.4266475439071655,
      "learning_rate": 6.823549286330487e-06,
      "loss": 1.7956,
      "step": 214731
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.4073375463485718,
      "learning_rate": 6.822681834804011e-06,
      "loss": 1.9335,
      "step": 214732
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.4747120141983032,
      "learning_rate": 6.821814437784701e-06,
      "loss": 2.0585,
      "step": 214733
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.4641468524932861,
      "learning_rate": 6.820947095272755e-06,
      "loss": 2.0877,
      "step": 214734
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.443837285041809,
      "learning_rate": 6.820079807268308e-06,
      "loss": 1.8233,
      "step": 214735
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.348445177078247,
      "learning_rate": 6.819212573771593e-06,
      "loss": 1.7634,
      "step": 214736
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.4592162370681763,
      "learning_rate": 6.818345394782676e-06,
      "loss": 2.0937,
      "step": 214737
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.3025685548782349,
      "learning_rate": 6.817478270301757e-06,
      "loss": 1.7368,
      "step": 214738
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.61174738407135,
      "learning_rate": 6.816611200329036e-06,
      "loss": 1.8396,
      "step": 214739
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.4470491409301758,
      "learning_rate": 6.815744184864646e-06,
      "loss": 2.1183,
      "step": 214740
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.556458830833435,
      "learning_rate": 6.814877223908721e-06,
      "loss": 2.0841,
      "step": 214741
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.5630508661270142,
      "learning_rate": 6.8140103174614605e-06,
      "loss": 2.1276,
      "step": 214742
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.3933467864990234,
      "learning_rate": 6.813143465522964e-06,
      "loss": 1.8235,
      "step": 214743
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.4068081378936768,
      "learning_rate": 6.812276668093497e-06,
      "loss": 1.8882,
      "step": 214744
    },
    {
      "epoch": 2.8,
      "grad_norm": 2.217958927154541,
      "learning_rate": 6.811409925173095e-06,
      "loss": 1.8525,
      "step": 214745
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.5869754552841187,
      "learning_rate": 6.810543236762089e-06,
      "loss": 2.0595,
      "step": 214746
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.3806709051132202,
      "learning_rate": 6.809676602860448e-06,
      "loss": 2.0461,
      "step": 214747
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.5201393365859985,
      "learning_rate": 6.80881002346847e-06,
      "loss": 1.9121,
      "step": 214748
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.484336256980896,
      "learning_rate": 6.807943498586221e-06,
      "loss": 2.0932,
      "step": 214749
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.5228368043899536,
      "learning_rate": 6.8070770282139365e-06,
      "loss": 1.8608,
      "step": 214750
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.3902778625488281,
      "learning_rate": 6.806210612351748e-06,
      "loss": 2.0787,
      "step": 214751
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.4994300603866577,
      "learning_rate": 6.8053442509998556e-06,
      "loss": 1.9168,
      "step": 214752
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.4609578847885132,
      "learning_rate": 6.804477944158326e-06,
      "loss": 1.805,
      "step": 214753
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.4183087348937988,
      "learning_rate": 6.803611691827426e-06,
      "loss": 1.7568,
      "step": 214754
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.6782102584838867,
      "learning_rate": 6.802745494007222e-06,
      "loss": 2.0752,
      "step": 214755
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.3566457033157349,
      "learning_rate": 6.8018793506979804e-06,
      "loss": 1.9563,
      "step": 214756
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.708090901374817,
      "learning_rate": 6.801013261899735e-06,
      "loss": 1.8356,
      "step": 214757
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.357100009918213,
      "learning_rate": 6.800147227612818e-06,
      "loss": 2.0938,
      "step": 214758
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.4869612455368042,
      "learning_rate": 6.799281247837196e-06,
      "loss": 2.015,
      "step": 214759
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.4565911293029785,
      "learning_rate": 6.79841532257317e-06,
      "loss": 1.9654,
      "step": 214760
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.6324867010116577,
      "learning_rate": 6.797549451820805e-06,
      "loss": 2.0002,
      "step": 214761
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.4601943492889404,
      "learning_rate": 6.796683635580336e-06,
      "loss": 2.1237,
      "step": 214762
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.3263111114501953,
      "learning_rate": 6.795817873851894e-06,
      "loss": 2.0001,
      "step": 214763
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.5151454210281372,
      "learning_rate": 6.794952166635681e-06,
      "loss": 2.1291,
      "step": 214764
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.3626093864440918,
      "learning_rate": 6.794086513931762e-06,
      "loss": 1.8981,
      "step": 214765
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.514027714729309,
      "learning_rate": 6.793220915740405e-06,
      "loss": 1.9389,
      "step": 214766
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.4179162979125977,
      "learning_rate": 6.792355372061675e-06,
      "loss": 1.9816,
      "step": 214767
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.4318585395812988,
      "learning_rate": 6.7914898828958395e-06,
      "loss": 2.1168,
      "step": 214768
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.6473770141601562,
      "learning_rate": 6.790624448242932e-06,
      "loss": 1.9213,
      "step": 214769
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.6832712888717651,
      "learning_rate": 6.789759068103285e-06,
      "loss": 1.999,
      "step": 214770
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.4746804237365723,
      "learning_rate": 6.788893742476864e-06,
      "loss": 1.7864,
      "step": 214771
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.4922207593917847,
      "learning_rate": 6.7880284713639376e-06,
      "loss": 1.9856,
      "step": 214772
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.570706844329834,
      "learning_rate": 6.787163254764672e-06,
      "loss": 2.0027,
      "step": 214773
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.429098129272461,
      "learning_rate": 6.7862980926791985e-06,
      "loss": 1.9074,
      "step": 214774
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.5678786039352417,
      "learning_rate": 6.785432985107686e-06,
      "loss": 2.1141,
      "step": 214775
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.5247961282730103,
      "learning_rate": 6.784567932050334e-06,
      "loss": 1.8427,
      "step": 214776
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.7621945142745972,
      "learning_rate": 6.783702933507207e-06,
      "loss": 1.8103,
      "step": 214777
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.480953574180603,
      "learning_rate": 6.78283798947854e-06,
      "loss": 1.9035,
      "step": 214778
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.4376254081726074,
      "learning_rate": 6.781973099964466e-06,
      "loss": 1.8607,
      "step": 214779
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.3830411434173584,
      "learning_rate": 6.781108264965218e-06,
      "loss": 2.1002,
      "step": 214780
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.5193758010864258,
      "learning_rate": 6.78024348448083e-06,
      "loss": 2.1145,
      "step": 214781
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.5906649827957153,
      "learning_rate": 6.7793787585116e-06,
      "loss": 2.0084,
      "step": 214782
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.4811800718307495,
      "learning_rate": 6.778514087057563e-06,
      "loss": 1.9518,
      "step": 214783
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.6096051931381226,
      "learning_rate": 6.777649470118951e-06,
      "loss": 1.9017,
      "step": 214784
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.5200837850570679,
      "learning_rate": 6.776784907695864e-06,
      "loss": 2.0526,
      "step": 214785
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.2883706092834473,
      "learning_rate": 6.7759203997885705e-06,
      "loss": 1.7025,
      "step": 214786
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.5768324136734009,
      "learning_rate": 6.775055946397135e-06,
      "loss": 1.9673,
      "step": 214787
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.4248329401016235,
      "learning_rate": 6.77419154752179e-06,
      "loss": 2.1053,
      "step": 214788
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.329554796218872,
      "learning_rate": 6.773327203162638e-06,
      "loss": 1.9569,
      "step": 214789
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.4673666954040527,
      "learning_rate": 6.772462913319843e-06,
      "loss": 2.0348,
      "step": 214790
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.4907262325286865,
      "learning_rate": 6.771598677993573e-06,
      "loss": 1.9841,
      "step": 214791
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.477571964263916,
      "learning_rate": 6.7707344971840276e-06,
      "loss": 2.0192,
      "step": 214792
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.4795676469802856,
      "learning_rate": 6.769870370891306e-06,
      "loss": 2.0619,
      "step": 214793
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.434036135673523,
      "learning_rate": 6.769006299115642e-06,
      "loss": 1.9368,
      "step": 214794
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.5053300857543945,
      "learning_rate": 6.768142281857136e-06,
      "loss": 2.0659,
      "step": 214795
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.5229356288909912,
      "learning_rate": 6.767278319115987e-06,
      "loss": 2.2437,
      "step": 214796
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.6189119815826416,
      "learning_rate": 6.7664144108922955e-06,
      "loss": 2.0231,
      "step": 214797
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.5329018831253052,
      "learning_rate": 6.765550557186294e-06,
      "loss": 2.1684,
      "step": 214798
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.4466384649276733,
      "learning_rate": 6.764686757998083e-06,
      "loss": 1.8699,
      "step": 214799
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.4978619813919067,
      "learning_rate": 6.763823013327896e-06,
      "loss": 1.8925,
      "step": 214800
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.5094786882400513,
      "learning_rate": 6.7629593231758316e-06,
      "loss": 1.9754,
      "step": 214801
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.556449294090271,
      "learning_rate": 6.762095687542057e-06,
      "loss": 1.8949,
      "step": 214802
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.4634137153625488,
      "learning_rate": 6.761232106426739e-06,
      "loss": 2.0614,
      "step": 214803
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.388491153717041,
      "learning_rate": 6.760368579830044e-06,
      "loss": 1.9947,
      "step": 214804
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.4433457851409912,
      "learning_rate": 6.759505107752139e-06,
      "loss": 1.8983,
      "step": 214805
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.4004007577896118,
      "learning_rate": 6.75864169019319e-06,
      "loss": 1.9625,
      "step": 214806
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.3822497129440308,
      "learning_rate": 6.75777832715333e-06,
      "loss": 1.8959,
      "step": 214807
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.4848146438598633,
      "learning_rate": 6.756915018632725e-06,
      "loss": 2.0116,
      "step": 214808
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.390730381011963,
      "learning_rate": 6.756051764631576e-06,
      "loss": 1.754,
      "step": 214809
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.5568666458129883,
      "learning_rate": 6.7551885651499825e-06,
      "loss": 1.9322,
      "step": 214810
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.3336564302444458,
      "learning_rate": 6.754325420188145e-06,
      "loss": 1.9244,
      "step": 214811
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.3342461585998535,
      "learning_rate": 6.753462329746229e-06,
      "loss": 1.8964,
      "step": 214812
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.6492890119552612,
      "learning_rate": 6.7525992938243674e-06,
      "loss": 2.228,
      "step": 214813
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.6389247179031372,
      "learning_rate": 6.751736312422695e-06,
      "loss": 2.2892,
      "step": 214814
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.462502360343933,
      "learning_rate": 6.750873385541478e-06,
      "loss": 1.8016,
      "step": 214815
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.4321188926696777,
      "learning_rate": 6.7500105131807815e-06,
      "loss": 2.0072,
      "step": 214816
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.3666908740997314,
      "learning_rate": 6.7491476953407734e-06,
      "loss": 1.9489,
      "step": 214817
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.4469618797302246,
      "learning_rate": 6.748284932021653e-06,
      "loss": 1.9565,
      "step": 214818
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.4178714752197266,
      "learning_rate": 6.747422223223586e-06,
      "loss": 2.0207,
      "step": 214819
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.4425960779190063,
      "learning_rate": 6.746559568946641e-06,
      "loss": 1.9414,
      "step": 214820
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.4069117307662964,
      "learning_rate": 6.745696969191117e-06,
      "loss": 1.7513,
      "step": 214821
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.2881057262420654,
      "learning_rate": 6.744834423957046e-06,
      "loss": 1.6894,
      "step": 214822
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.4949933290481567,
      "learning_rate": 6.743971933244696e-06,
      "loss": 1.8634,
      "step": 214823
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.446645736694336,
      "learning_rate": 6.743109497054167e-06,
      "loss": 1.816,
      "step": 214824
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.524875521659851,
      "learning_rate": 6.742247115385624e-06,
      "loss": 2.1564,
      "step": 214825
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.5804874897003174,
      "learning_rate": 6.741384788239201e-06,
      "loss": 1.8797,
      "step": 214826
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.4954090118408203,
      "learning_rate": 6.740522515615132e-06,
      "loss": 1.8995,
      "step": 214827
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.5105249881744385,
      "learning_rate": 6.739660297513483e-06,
      "loss": 2.0362,
      "step": 214828
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.5435550212860107,
      "learning_rate": 6.73879813393452e-06,
      "loss": 1.9507,
      "step": 214829
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.4100239276885986,
      "learning_rate": 6.73793602487831e-06,
      "loss": 1.9459,
      "step": 214830
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.494222640991211,
      "learning_rate": 6.737073970345119e-06,
      "loss": 2.037,
      "step": 214831
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.6432565450668335,
      "learning_rate": 6.736211970334981e-06,
      "loss": 2.0116,
      "step": 214832
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.4531888961791992,
      "learning_rate": 6.735350024848163e-06,
      "loss": 1.8767,
      "step": 214833
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.3751249313354492,
      "learning_rate": 6.73448813388473e-06,
      "loss": 1.9446,
      "step": 214834
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.314787745475769,
      "learning_rate": 6.733626297444916e-06,
      "loss": 1.8275,
      "step": 214835
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.4802422523498535,
      "learning_rate": 6.732764515528822e-06,
      "loss": 2.1437,
      "step": 214836
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.6174592971801758,
      "learning_rate": 6.731902788136711e-06,
      "loss": 1.8511,
      "step": 214837
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.5046266317367554,
      "learning_rate": 6.73104111526862e-06,
      "loss": 2.0536,
      "step": 214838
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.5129295587539673,
      "learning_rate": 6.730179496924815e-06,
      "loss": 2.0161,
      "step": 214839
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.3756110668182373,
      "learning_rate": 6.7293179331053606e-06,
      "loss": 1.9811,
      "step": 214840
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.5648423433303833,
      "learning_rate": 6.728456423810458e-06,
      "loss": 2.0528,
      "step": 214841
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.3702681064605713,
      "learning_rate": 6.727594969040273e-06,
      "loss": 2.1361,
      "step": 214842
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.3675893545150757,
      "learning_rate": 6.72673356879504e-06,
      "loss": 1.9492,
      "step": 214843
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.4187841415405273,
      "learning_rate": 6.725872223074757e-06,
      "loss": 1.9545,
      "step": 214844
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.5131502151489258,
      "learning_rate": 6.725010931879693e-06,
      "loss": 1.9915,
      "step": 214845
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.381750226020813,
      "learning_rate": 6.7241496952099796e-06,
      "loss": 1.9729,
      "step": 214846
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.484403371810913,
      "learning_rate": 6.723288513065817e-06,
      "loss": 2.0055,
      "step": 214847
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.5058939456939697,
      "learning_rate": 6.722427385447271e-06,
      "loss": 2.0085,
      "step": 214848
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.505225419998169,
      "learning_rate": 6.7215663123546425e-06,
      "loss": 1.9645,
      "step": 214849
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.4978705644607544,
      "learning_rate": 6.7207052937879316e-06,
      "loss": 2.0888,
      "step": 214850
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.4356868267059326,
      "learning_rate": 6.719844329747437e-06,
      "loss": 1.8823,
      "step": 214851
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.4544782638549805,
      "learning_rate": 6.718983420233192e-06,
      "loss": 2.0171,
      "step": 214852
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.4316096305847168,
      "learning_rate": 6.718122565245499e-06,
      "loss": 1.8567,
      "step": 214853
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.6422638893127441,
      "learning_rate": 6.717261764784387e-06,
      "loss": 2.0096,
      "step": 214854
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.461460828781128,
      "learning_rate": 6.716401018850126e-06,
      "loss": 2.0062,
      "step": 214855
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.4070671796798706,
      "learning_rate": 6.715540327442781e-06,
      "loss": 1.8401,
      "step": 214856
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.5857301950454712,
      "learning_rate": 6.714679690562586e-06,
      "loss": 2.0201,
      "step": 214857
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.4440717697143555,
      "learning_rate": 6.713819108209606e-06,
      "loss": 2.0511,
      "step": 214858
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.3337583541870117,
      "learning_rate": 6.712958580384109e-06,
      "loss": 2.0404,
      "step": 214859
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.4612385034561157,
      "learning_rate": 6.712098107086195e-06,
      "loss": 2.1367,
      "step": 214860
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.6236121654510498,
      "learning_rate": 6.7112376883160624e-06,
      "loss": 2.0974,
      "step": 214861
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.3460445404052734,
      "learning_rate": 6.710377324073846e-06,
      "loss": 1.9313,
      "step": 214862
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.5757837295532227,
      "learning_rate": 6.709517014359678e-06,
      "loss": 1.8983,
      "step": 214863
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.3487722873687744,
      "learning_rate": 6.708656759173758e-06,
      "loss": 2.0075,
      "step": 214864
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.4947893619537354,
      "learning_rate": 6.707796558516254e-06,
      "loss": 1.9509,
      "step": 214865
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.4660588502883911,
      "learning_rate": 6.706936412387265e-06,
      "loss": 1.9935,
      "step": 214866
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.354455828666687,
      "learning_rate": 6.706076320787024e-06,
      "loss": 1.8745,
      "step": 214867
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.5199791193008423,
      "learning_rate": 6.7052162837156646e-06,
      "loss": 1.9069,
      "step": 214868
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.6448256969451904,
      "learning_rate": 6.704356301173352e-06,
      "loss": 1.8284,
      "step": 214869
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.486023187637329,
      "learning_rate": 6.703496373160189e-06,
      "loss": 1.9139,
      "step": 214870
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.4458239078521729,
      "learning_rate": 6.702636499676406e-06,
      "loss": 1.784,
      "step": 214871
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.381363868713379,
      "learning_rate": 6.701776680722137e-06,
      "loss": 1.9621,
      "step": 214872
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.3860284090042114,
      "learning_rate": 6.70091691629755e-06,
      "loss": 2.0606,
      "step": 214873
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.6176613569259644,
      "learning_rate": 6.7000572064028095e-06,
      "loss": 1.9509,
      "step": 214874
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.4102513790130615,
      "learning_rate": 6.699197551038082e-06,
      "loss": 1.8723,
      "step": 214875
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.5647335052490234,
      "learning_rate": 6.698337950203436e-06,
      "loss": 1.9342,
      "step": 214876
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.3466600179672241,
      "learning_rate": 6.69747840389917e-06,
      "loss": 2.0256,
      "step": 214877
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.5858958959579468,
      "learning_rate": 6.696618912125351e-06,
      "loss": 2.0114,
      "step": 214878
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.4655954837799072,
      "learning_rate": 6.6957594748821785e-06,
      "loss": 1.9578,
      "step": 214879
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.5190393924713135,
      "learning_rate": 6.694900092169819e-06,
      "loss": 1.9252,
      "step": 214880
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.5653741359710693,
      "learning_rate": 6.6940407639884065e-06,
      "loss": 2.2069,
      "step": 214881
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.3947219848632812,
      "learning_rate": 6.693181490338073e-06,
      "loss": 1.8476,
      "step": 214882
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.5485203266143799,
      "learning_rate": 6.692322271219052e-06,
      "loss": 1.9752,
      "step": 214883
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.422154426574707,
      "learning_rate": 6.691463106631411e-06,
      "loss": 2.105,
      "step": 214884
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.4975634813308716,
      "learning_rate": 6.690603996575416e-06,
      "loss": 2.149,
      "step": 214885
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.5179036855697632,
      "learning_rate": 6.6897449410511675e-06,
      "loss": 1.9703,
      "step": 214886
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.5525814294815063,
      "learning_rate": 6.6888859400588305e-06,
      "loss": 2.0771,
      "step": 214887
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.3849074840545654,
      "learning_rate": 6.688026993598572e-06,
      "loss": 2.0193,
      "step": 214888
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.4275215864181519,
      "learning_rate": 6.687168101670526e-06,
      "loss": 2.0719,
      "step": 214889
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.4807366132736206,
      "learning_rate": 6.686309264274858e-06,
      "loss": 1.9982,
      "step": 214890
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.5927149057388306,
      "learning_rate": 6.685450481411769e-06,
      "loss": 1.9572,
      "step": 214891
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.6210054159164429,
      "learning_rate": 6.684591753081426e-06,
      "loss": 1.8553,
      "step": 214892
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.4155175685882568,
      "learning_rate": 6.683733079283893e-06,
      "loss": 1.9053,
      "step": 214893
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.352541446685791,
      "learning_rate": 6.682874460019405e-06,
      "loss": 2.0573,
      "step": 214894
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.5199521780014038,
      "learning_rate": 6.682015895288129e-06,
      "loss": 2.0221,
      "step": 214895
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.53487229347229,
      "learning_rate": 6.681157385090164e-06,
      "loss": 2.0187,
      "step": 214896
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.466876745223999,
      "learning_rate": 6.6802989294257425e-06,
      "loss": 1.8305,
      "step": 214897
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.4832229614257812,
      "learning_rate": 6.679440528294999e-06,
      "loss": 1.8546,
      "step": 214898
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.3914893865585327,
      "learning_rate": 6.6785821816980665e-06,
      "loss": 2.0316,
      "step": 214899
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.4753762483596802,
      "learning_rate": 6.677723889635111e-06,
      "loss": 1.966,
      "step": 214900
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.4633723497390747,
      "learning_rate": 6.676865652106334e-06,
      "loss": 1.9198,
      "step": 214901
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.608986496925354,
      "learning_rate": 6.676007469111833e-06,
      "loss": 2.0979,
      "step": 214902
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.7701131105422974,
      "learning_rate": 6.6751493406518085e-06,
      "loss": 2.0397,
      "step": 214903
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.3823161125183105,
      "learning_rate": 6.674291266726428e-06,
      "loss": 2.0701,
      "step": 214904
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.4309940338134766,
      "learning_rate": 6.673433247335824e-06,
      "loss": 1.8329,
      "step": 214905
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.334536075592041,
      "learning_rate": 6.672575282480164e-06,
      "loss": 1.8363,
      "step": 214906
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.51473867893219,
      "learning_rate": 6.671717372159579e-06,
      "loss": 2.0482,
      "step": 214907
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.4627962112426758,
      "learning_rate": 6.670859516374305e-06,
      "loss": 2.0573,
      "step": 214908
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.4650675058364868,
      "learning_rate": 6.67000171512444e-06,
      "loss": 2.06,
      "step": 214909
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.4677051305770874,
      "learning_rate": 6.669143968410185e-06,
      "loss": 1.8933,
      "step": 214910
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.6145083904266357,
      "learning_rate": 6.668286276231605e-06,
      "loss": 2.2092,
      "step": 214911
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.5611705780029297,
      "learning_rate": 6.667428638589001e-06,
      "loss": 1.987,
      "step": 214912
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.5517501831054688,
      "learning_rate": 6.666571055482406e-06,
      "loss": 1.9839,
      "step": 214913
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.4295705556869507,
      "learning_rate": 6.6657135269120534e-06,
      "loss": 1.913,
      "step": 214914
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.606621265411377,
      "learning_rate": 6.6648560528780426e-06,
      "loss": 1.863,
      "step": 214915
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.5696414709091187,
      "learning_rate": 6.663998633380674e-06,
      "loss": 2.0608,
      "step": 214916
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.471005916595459,
      "learning_rate": 6.663141268419914e-06,
      "loss": 1.8441,
      "step": 214917
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.3618669509887695,
      "learning_rate": 6.6622839579960615e-06,
      "loss": 2.0292,
      "step": 214918
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.548534870147705,
      "learning_rate": 6.661426702109185e-06,
      "loss": 1.8848,
      "step": 214919
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.426344871520996,
      "learning_rate": 6.660569500759516e-06,
      "loss": 1.8783,
      "step": 214920
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.4562677145004272,
      "learning_rate": 6.659712353947155e-06,
      "loss": 1.9202,
      "step": 214921
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.400195837020874,
      "learning_rate": 6.658855261672336e-06,
      "loss": 1.762,
      "step": 214922
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.6464818716049194,
      "learning_rate": 6.657998223935124e-06,
      "loss": 1.9288,
      "step": 214923
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.3995338678359985,
      "learning_rate": 6.657141240735786e-06,
      "loss": 2.2558,
      "step": 214924
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.5945483446121216,
      "learning_rate": 6.656284312074356e-06,
      "loss": 2.2143,
      "step": 214925
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.5442118644714355,
      "learning_rate": 6.6554274379510996e-06,
      "loss": 2.2261,
      "step": 214926
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.4161055088043213,
      "learning_rate": 6.654570618366117e-06,
      "loss": 1.7278,
      "step": 214927
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.446925163269043,
      "learning_rate": 6.653713853319609e-06,
      "loss": 1.7892,
      "step": 214928
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.4600071907043457,
      "learning_rate": 6.6528571428117075e-06,
      "loss": 1.9815,
      "step": 214929
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.5261435508728027,
      "learning_rate": 6.6520004868425795e-06,
      "loss": 2.0274,
      "step": 214930
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.4576444625854492,
      "learning_rate": 6.651143885412358e-06,
      "loss": 1.9335,
      "step": 214931
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.4622889757156372,
      "learning_rate": 6.650287338521276e-06,
      "loss": 1.7895,
      "step": 214932
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.503494143486023,
      "learning_rate": 6.649430846169368e-06,
      "loss": 1.8568,
      "step": 214933
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.7235016822814941,
      "learning_rate": 6.6485744083569326e-06,
      "loss": 2.2093,
      "step": 214934
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.4391732215881348,
      "learning_rate": 6.64771802508407e-06,
      "loss": 1.9186,
      "step": 214935
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.5081894397735596,
      "learning_rate": 6.646861696350914e-06,
      "loss": 1.9005,
      "step": 214936
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.3097282648086548,
      "learning_rate": 6.646005422157596e-06,
      "loss": 2.2652,
      "step": 214937
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.405720829963684,
      "learning_rate": 6.645149202504385e-06,
      "loss": 1.8534,
      "step": 214938
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.490409016609192,
      "learning_rate": 6.644293037391346e-06,
      "loss": 1.8182,
      "step": 214939
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.5085852146148682,
      "learning_rate": 6.6434369268186795e-06,
      "loss": 2.0032,
      "step": 214940
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.4009253978729248,
      "learning_rate": 6.642580870786551e-06,
      "loss": 1.9263,
      "step": 214941
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.4803588390350342,
      "learning_rate": 6.641724869295095e-06,
      "loss": 2.0763,
      "step": 214942
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.414007306098938,
      "learning_rate": 6.640868922344478e-06,
      "loss": 2.0223,
      "step": 214943
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.3874421119689941,
      "learning_rate": 6.640013029934865e-06,
      "loss": 1.9605,
      "step": 214944
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.3794069290161133,
      "learning_rate": 6.639157192066391e-06,
      "loss": 2.0056,
      "step": 214945
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.562183141708374,
      "learning_rate": 6.638301408739255e-06,
      "loss": 2.0343,
      "step": 214946
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.5511445999145508,
      "learning_rate": 6.63744567995359e-06,
      "loss": 2.2539,
      "step": 214947
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.4448827505111694,
      "learning_rate": 6.636590005709597e-06,
      "loss": 1.9437,
      "step": 214948
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.496370553970337,
      "learning_rate": 6.635734386007341e-06,
      "loss": 1.911,
      "step": 214949
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.50798499584198,
      "learning_rate": 6.634878820847056e-06,
      "loss": 2.1859,
      "step": 214950
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.4584121704101562,
      "learning_rate": 6.6340233102288755e-06,
      "loss": 1.8413,
      "step": 214951
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.5531044006347656,
      "learning_rate": 6.633167854152999e-06,
      "loss": 1.9929,
      "step": 214952
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.5594178438186646,
      "learning_rate": 6.6323124526195596e-06,
      "loss": 2.0941,
      "step": 214953
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.5678560733795166,
      "learning_rate": 6.63145710562869e-06,
      "loss": 1.9644,
      "step": 214954
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.4768433570861816,
      "learning_rate": 6.630601813180558e-06,
      "loss": 1.9952,
      "step": 214955
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.4225287437438965,
      "learning_rate": 6.629746575275363e-06,
      "loss": 1.8307,
      "step": 214956
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.3608639240264893,
      "learning_rate": 6.628891391913205e-06,
      "loss": 1.9959,
      "step": 214957
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.5681098699569702,
      "learning_rate": 6.628036263094316e-06,
      "loss": 2.0468,
      "step": 214958
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.4838831424713135,
      "learning_rate": 6.627181188818798e-06,
      "loss": 1.9443,
      "step": 214959
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.3264989852905273,
      "learning_rate": 6.626326169086815e-06,
      "loss": 1.8393,
      "step": 214960
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.4897172451019287,
      "learning_rate": 6.625471203898502e-06,
      "loss": 1.9132,
      "step": 214961
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.4887410402297974,
      "learning_rate": 6.624616293254093e-06,
      "loss": 1.934,
      "step": 214962
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.538651466369629,
      "learning_rate": 6.6237614371536854e-06,
      "loss": 1.9324,
      "step": 214963
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.3973078727722168,
      "learning_rate": 6.62290663559748e-06,
      "loss": 2.0924,
      "step": 214964
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.5059583187103271,
      "learning_rate": 6.622051888585611e-06,
      "loss": 2.1532,
      "step": 214965
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.3970365524291992,
      "learning_rate": 6.621197196118244e-06,
      "loss": 1.9282,
      "step": 214966
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.5317562818527222,
      "learning_rate": 6.6203425581955125e-06,
      "loss": 1.9703,
      "step": 214967
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.4973896741867065,
      "learning_rate": 6.619487974817617e-06,
      "loss": 2.0207,
      "step": 214968
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.6427208185195923,
      "learning_rate": 6.618633445984689e-06,
      "loss": 2.1964,
      "step": 214969
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.561349868774414,
      "learning_rate": 6.617778971696897e-06,
      "loss": 1.9349,
      "step": 214970
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.448427438735962,
      "learning_rate": 6.616924551954405e-06,
      "loss": 2.236,
      "step": 214971
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.4651294946670532,
      "learning_rate": 6.616070186757383e-06,
      "loss": 1.9553,
      "step": 214972
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.4511544704437256,
      "learning_rate": 6.6152158761058946e-06,
      "loss": 1.9372,
      "step": 214973
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.3784949779510498,
      "learning_rate": 6.614361620000241e-06,
      "loss": 1.9678,
      "step": 214974
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.4737955331802368,
      "learning_rate": 6.613507418440489e-06,
      "loss": 1.7941,
      "step": 214975
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.4369022846221924,
      "learning_rate": 6.6126532714268375e-06,
      "loss": 2.0671,
      "step": 214976
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.3836458921432495,
      "learning_rate": 6.611799178959454e-06,
      "loss": 2.2647,
      "step": 214977
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.4582573175430298,
      "learning_rate": 6.61094514103847e-06,
      "loss": 1.8563,
      "step": 214978
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.4343911409378052,
      "learning_rate": 6.610091157663988e-06,
      "loss": 2.0198,
      "step": 214979
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.4176043272018433,
      "learning_rate": 6.609237228836306e-06,
      "loss": 2.1267,
      "step": 214980
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.4408780336380005,
      "learning_rate": 6.608383354555425e-06,
      "loss": 1.9683,
      "step": 214981
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.5244190692901611,
      "learning_rate": 6.607529534821643e-06,
      "loss": 2.114,
      "step": 214982
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.3916068077087402,
      "learning_rate": 6.6066757696350616e-06,
      "loss": 1.7783,
      "step": 214983
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.6967219114303589,
      "learning_rate": 6.605822058995813e-06,
      "loss": 1.9274,
      "step": 214984
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.4601147174835205,
      "learning_rate": 6.604968402904098e-06,
      "loss": 2.0207,
      "step": 214985
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.5865460634231567,
      "learning_rate": 6.60411480136005e-06,
      "loss": 1.9642,
      "step": 214986
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.389064908027649,
      "learning_rate": 6.603261254363801e-06,
      "loss": 1.7904,
      "step": 214987
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.5266727209091187,
      "learning_rate": 6.602407761915584e-06,
      "loss": 1.9798,
      "step": 214988
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.5182043313980103,
      "learning_rate": 6.601554324015534e-06,
      "loss": 1.9949,
      "step": 214989
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.4489495754241943,
      "learning_rate": 6.60070094066375e-06,
      "loss": 1.921,
      "step": 214990
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.4373341798782349,
      "learning_rate": 6.599847611860465e-06,
      "loss": 1.9613,
      "step": 214991
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.3897091150283813,
      "learning_rate": 6.5989943376057784e-06,
      "loss": 1.9164,
      "step": 214992
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.38934326171875,
      "learning_rate": 6.598141117899891e-06,
      "loss": 2.1748,
      "step": 214993
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.3836439847946167,
      "learning_rate": 6.597287952742969e-06,
      "loss": 2.0356,
      "step": 214994
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.5794485807418823,
      "learning_rate": 6.596434842135112e-06,
      "loss": 2.0318,
      "step": 214995
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.503643274307251,
      "learning_rate": 6.595581786076554e-06,
      "loss": 1.8581,
      "step": 214996
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.4653083086013794,
      "learning_rate": 6.5947287845673934e-06,
      "loss": 1.9942,
      "step": 214997
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.401448369026184,
      "learning_rate": 6.593875837607799e-06,
      "loss": 1.9978,
      "step": 214998
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.443596363067627,
      "learning_rate": 6.593022945197968e-06,
      "loss": 1.9902,
      "step": 214999
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.3455824851989746,
      "learning_rate": 6.592170107338035e-06,
      "loss": 1.8478,
      "step": 215000
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.4321924448013306,
      "learning_rate": 6.591317324028134e-06,
      "loss": 1.9365,
      "step": 215001
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.3446102142333984,
      "learning_rate": 6.590464595268463e-06,
      "loss": 1.9129,
      "step": 215002
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.5490981340408325,
      "learning_rate": 6.589611921059157e-06,
      "loss": 1.9792,
      "step": 215003
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.4510599374771118,
      "learning_rate": 6.588759301400348e-06,
      "loss": 2.1393,
      "step": 215004
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.501444697380066,
      "learning_rate": 6.58790673629227e-06,
      "loss": 1.8174,
      "step": 215005
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.5765573978424072,
      "learning_rate": 6.5870542257350224e-06,
      "loss": 1.965,
      "step": 215006
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.4118565320968628,
      "learning_rate": 6.586201769728805e-06,
      "loss": 2.0899,
      "step": 215007
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.466956615447998,
      "learning_rate": 6.585349368273718e-06,
      "loss": 1.7398,
      "step": 215008
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.4640380144119263,
      "learning_rate": 6.5844970213699945e-06,
      "loss": 1.8456,
      "step": 215009
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.4442046880722046,
      "learning_rate": 6.583644729017701e-06,
      "loss": 1.9291,
      "step": 215010
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.3504316806793213,
      "learning_rate": 6.582792491217071e-06,
      "loss": 2.0851,
      "step": 215011
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.7424757480621338,
      "learning_rate": 6.581940307968203e-06,
      "loss": 2.0869,
      "step": 215012
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.4374470710754395,
      "learning_rate": 6.581088179271332e-06,
      "loss": 2.0519,
      "step": 215013
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.5360788106918335,
      "learning_rate": 6.580236105126591e-06,
      "loss": 1.8431,
      "step": 215014
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.5036735534667969,
      "learning_rate": 6.579384085534112e-06,
      "loss": 2.0077,
      "step": 215015
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.571291446685791,
      "learning_rate": 6.578532120494029e-06,
      "loss": 2.149,
      "step": 215016
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.4695470333099365,
      "learning_rate": 6.577680210006575e-06,
      "loss": 2.0095,
      "step": 215017
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.5340073108673096,
      "learning_rate": 6.576828354071817e-06,
      "loss": 2.0918,
      "step": 215018
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.4684303998947144,
      "learning_rate": 6.575976552690021e-06,
      "loss": 1.9699,
      "step": 215019
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.5499870777130127,
      "learning_rate": 6.575124805861287e-06,
      "loss": 1.847,
      "step": 215020
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.5081117153167725,
      "learning_rate": 6.574273113585749e-06,
      "loss": 2.0272,
      "step": 215021
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.560899019241333,
      "learning_rate": 6.573421475863605e-06,
      "loss": 2.0077,
      "step": 215022
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.5051895380020142,
      "learning_rate": 6.57256989269499e-06,
      "loss": 1.8877,
      "step": 215023
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.8007771968841553,
      "learning_rate": 6.571718364080069e-06,
      "loss": 2.0556,
      "step": 215024
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.5691543817520142,
      "learning_rate": 6.570866890019011e-06,
      "loss": 2.2303,
      "step": 215025
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.3822510242462158,
      "learning_rate": 6.57001547051198e-06,
      "loss": 2.3063,
      "step": 215026
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.5696877241134644,
      "learning_rate": 6.569164105559144e-06,
      "loss": 1.7716,
      "step": 215027
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.348973274230957,
      "learning_rate": 6.568312795160569e-06,
      "loss": 1.8311,
      "step": 215028
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.4891515970230103,
      "learning_rate": 6.5674615393165546e-06,
      "loss": 2.1889,
      "step": 215029
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.5714046955108643,
      "learning_rate": 6.566610338027134e-06,
      "loss": 2.1346,
      "step": 215030
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.4557548761367798,
      "learning_rate": 6.565759191292541e-06,
      "loss": 1.8755,
      "step": 215031
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.5497865676879883,
      "learning_rate": 6.564908099112942e-06,
      "loss": 2.0072,
      "step": 215032
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.4944394826889038,
      "learning_rate": 6.564057061488437e-06,
      "loss": 2.1656,
      "step": 215033
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.46474027633667,
      "learning_rate": 6.563206078419192e-06,
      "loss": 1.8854,
      "step": 215034
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.4781297445297241,
      "learning_rate": 6.562355149905441e-06,
      "loss": 1.9277,
      "step": 215035
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.5072938203811646,
      "learning_rate": 6.561504275947216e-06,
      "loss": 1.9504,
      "step": 215036
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.478095293045044,
      "learning_rate": 6.560653456544818e-06,
      "loss": 1.8673,
      "step": 215037
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.4932608604431152,
      "learning_rate": 6.5598026916982795e-06,
      "loss": 1.8158,
      "step": 215038
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.4472887516021729,
      "learning_rate": 6.558951981407867e-06,
      "loss": 2.2587,
      "step": 215039
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.5668874979019165,
      "learning_rate": 6.558101325673614e-06,
      "loss": 2.0546,
      "step": 215040
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.5077911615371704,
      "learning_rate": 6.557250724495822e-06,
      "loss": 2.0298,
      "step": 215041
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.540321707725525,
      "learning_rate": 6.556400177874521e-06,
      "loss": 1.9297,
      "step": 215042
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.3728581666946411,
      "learning_rate": 6.555549685809946e-06,
      "loss": 1.9964,
      "step": 215043
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.4408926963806152,
      "learning_rate": 6.554699248302231e-06,
      "loss": 1.9681,
      "step": 215044
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.3691492080688477,
      "learning_rate": 6.553848865351541e-06,
      "loss": 1.8349,
      "step": 215045
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.6440807580947876,
      "learning_rate": 6.55299853695801e-06,
      "loss": 2.0249,
      "step": 215046
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.6496769189834595,
      "learning_rate": 6.552148263121837e-06,
      "loss": 1.9406,
      "step": 215047
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.425515055656433,
      "learning_rate": 6.551298043843156e-06,
      "loss": 1.9518,
      "step": 215048
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.3172950744628906,
      "learning_rate": 6.550447879122133e-06,
      "loss": 2.1093,
      "step": 215049
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.4895135164260864,
      "learning_rate": 6.5495977689589365e-06,
      "loss": 2.0513,
      "step": 215050
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.3142884969711304,
      "learning_rate": 6.548747713353664e-06,
      "loss": 2.13,
      "step": 215051
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.3516870737075806,
      "learning_rate": 6.547897712306549e-06,
      "loss": 1.6397,
      "step": 215052
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.4052602052688599,
      "learning_rate": 6.547047765817725e-06,
      "loss": 2.0132,
      "step": 215053
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.5958387851715088,
      "learning_rate": 6.546197873887294e-06,
      "loss": 1.9178,
      "step": 215054
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.458101511001587,
      "learning_rate": 6.545348036515519e-06,
      "loss": 2.0493,
      "step": 215055
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.4557276964187622,
      "learning_rate": 6.544498253702501e-06,
      "loss": 1.8394,
      "step": 215056
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.4421133995056152,
      "learning_rate": 6.543648525448375e-06,
      "loss": 1.7852,
      "step": 215057
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.6436980962753296,
      "learning_rate": 6.542798851753306e-06,
      "loss": 1.9115,
      "step": 215058
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.4646528959274292,
      "learning_rate": 6.5419492326175275e-06,
      "loss": 2.0686,
      "step": 215059
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.4539178609848022,
      "learning_rate": 6.541099668041072e-06,
      "loss": 1.8259,
      "step": 215060
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.556363582611084,
      "learning_rate": 6.540250158024241e-06,
      "loss": 2.1543,
      "step": 215061
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.4752802848815918,
      "learning_rate": 6.5394007025670325e-06,
      "loss": 2.0058,
      "step": 215062
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.3397254943847656,
      "learning_rate": 6.538551301669781e-06,
      "loss": 1.9589,
      "step": 215063
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.401418924331665,
      "learning_rate": 6.537701955332486e-06,
      "loss": 1.9194,
      "step": 215064
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.5928748846054077,
      "learning_rate": 6.536852663555414e-06,
      "loss": 1.9472,
      "step": 215065
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.4763617515563965,
      "learning_rate": 6.536003426338632e-06,
      "loss": 2.0418,
      "step": 215066
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.7844830751419067,
      "learning_rate": 6.535154243682406e-06,
      "loss": 2.1269,
      "step": 215067
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.47365403175354,
      "learning_rate": 6.534305115586769e-06,
      "loss": 2.1394,
      "step": 215068
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.5801687240600586,
      "learning_rate": 6.533456042051987e-06,
      "loss": 2.0067,
      "step": 215069
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.5092562437057495,
      "learning_rate": 6.532607023078162e-06,
      "loss": 1.8429,
      "step": 215070
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.465518832206726,
      "learning_rate": 6.531758058665493e-06,
      "loss": 1.6921,
      "step": 215071
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.4932291507720947,
      "learning_rate": 6.5309091488140784e-06,
      "loss": 2.0132,
      "step": 215072
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.3592275381088257,
      "learning_rate": 6.53006029352412e-06,
      "loss": 2.1493,
      "step": 215073
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.417826533317566,
      "learning_rate": 6.529211492795716e-06,
      "loss": 1.9232,
      "step": 215074
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.4239816665649414,
      "learning_rate": 6.528362746629135e-06,
      "loss": 2.046,
      "step": 215075
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.355754017829895,
      "learning_rate": 6.527514055024475e-06,
      "loss": 1.8763,
      "step": 215076
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.4652838706970215,
      "learning_rate": 6.526665417981836e-06,
      "loss": 2.0361,
      "step": 215077
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.4804468154907227,
      "learning_rate": 6.525816835501452e-06,
      "loss": 1.8823,
      "step": 215078
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.4990105628967285,
      "learning_rate": 6.524968307583489e-06,
      "loss": 2.0128,
      "step": 215079
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.4758338928222656,
      "learning_rate": 6.524119834228014e-06,
      "loss": 1.9851,
      "step": 215080
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.5704480409622192,
      "learning_rate": 6.523271415435294e-06,
      "loss": 1.8242,
      "step": 215081
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.476688027381897,
      "learning_rate": 6.522423051205428e-06,
      "loss": 1.8304,
      "step": 215082
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.4049988985061646,
      "learning_rate": 6.521574741538582e-06,
      "loss": 1.8146,
      "step": 215083
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.5590267181396484,
      "learning_rate": 6.520726486434924e-06,
      "loss": 1.9707,
      "step": 215084
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.535742998123169,
      "learning_rate": 6.519878285894586e-06,
      "loss": 1.9839,
      "step": 215085
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.5600147247314453,
      "learning_rate": 6.5190301399177356e-06,
      "loss": 1.9863,
      "step": 215086
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.5415706634521484,
      "learning_rate": 6.518182048504572e-06,
      "loss": 2.0497,
      "step": 215087
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.482778549194336,
      "learning_rate": 6.517334011655195e-06,
      "loss": 2.0371,
      "step": 215088
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.3992341756820679,
      "learning_rate": 6.516486029369772e-06,
      "loss": 2.189,
      "step": 215089
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.461756706237793,
      "learning_rate": 6.515638101648502e-06,
      "loss": 2.0767,
      "step": 215090
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.7557777166366577,
      "learning_rate": 6.5147902284914845e-06,
      "loss": 1.9455,
      "step": 215091
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.3966683149337769,
      "learning_rate": 6.51394240989892e-06,
      "loss": 1.9959,
      "step": 215092
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.5095247030258179,
      "learning_rate": 6.5130946458709755e-06,
      "loss": 1.9889,
      "step": 215093
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.4986871480941772,
      "learning_rate": 6.51224693640775e-06,
      "loss": 2.0907,
      "step": 215094
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.4612741470336914,
      "learning_rate": 6.511399281509444e-06,
      "loss": 1.6404,
      "step": 215095
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.343936800956726,
      "learning_rate": 6.510551681176257e-06,
      "loss": 2.0313,
      "step": 215096
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.590944528579712,
      "learning_rate": 6.509704135408222e-06,
      "loss": 2.0895,
      "step": 215097
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.4101389646530151,
      "learning_rate": 6.50885664420564e-06,
      "loss": 2.044,
      "step": 215098
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.7073924541473389,
      "learning_rate": 6.508009207568576e-06,
      "loss": 2.0739,
      "step": 215099
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.5272903442382812,
      "learning_rate": 6.507161825497231e-06,
      "loss": 2.0871,
      "step": 215100
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.5254075527191162,
      "learning_rate": 6.506314497991705e-06,
      "loss": 1.8999,
      "step": 215101
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.504141926765442,
      "learning_rate": 6.5054672250522296e-06,
      "loss": 2.135,
      "step": 215102
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.4618871212005615,
      "learning_rate": 6.5046200066788735e-06,
      "loss": 1.7853,
      "step": 215103
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.637450098991394,
      "learning_rate": 6.503772842871902e-06,
      "loss": 1.8837,
      "step": 215104
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.461032509803772,
      "learning_rate": 6.502925733631414e-06,
      "loss": 1.9239,
      "step": 215105
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.5833197832107544,
      "learning_rate": 6.502078678957578e-06,
      "loss": 2.1321,
      "step": 215106
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.446358561515808,
      "learning_rate": 6.5012316788505266e-06,
      "loss": 2.0378,
      "step": 215107
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.7401748895645142,
      "learning_rate": 6.50038473331046e-06,
      "loss": 2.0731,
      "step": 215108
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.4560825824737549,
      "learning_rate": 6.499537842337477e-06,
      "loss": 2.0127,
      "step": 215109
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.49825918674469,
      "learning_rate": 6.498691005931777e-06,
      "loss": 2.0358,
      "step": 215110
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.535446047782898,
      "learning_rate": 6.497844224093563e-06,
      "loss": 2.172,
      "step": 215111
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.489074468612671,
      "learning_rate": 6.496997496822898e-06,
      "loss": 2.1242,
      "step": 215112
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.5924407243728638,
      "learning_rate": 6.49615082411995e-06,
      "loss": 2.0691,
      "step": 215113
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.5596787929534912,
      "learning_rate": 6.495304205984952e-06,
      "loss": 2.0505,
      "step": 215114
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.4435551166534424,
      "learning_rate": 6.494457642418005e-06,
      "loss": 2.03,
      "step": 215115
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.4590364694595337,
      "learning_rate": 6.493611133419274e-06,
      "loss": 1.8912,
      "step": 215116
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.5121291875839233,
      "learning_rate": 6.49276467898896e-06,
      "loss": 2.1695,
      "step": 215117
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.4986543655395508,
      "learning_rate": 6.491918279127128e-06,
      "loss": 1.989,
      "step": 215118
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.4405226707458496,
      "learning_rate": 6.491071933834013e-06,
      "loss": 2.1973,
      "step": 215119
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.5162978172302246,
      "learning_rate": 6.490225643109748e-06,
      "loss": 2.0444,
      "step": 215120
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.6134140491485596,
      "learning_rate": 6.489379406954465e-06,
      "loss": 2.2646,
      "step": 215121
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.5473036766052246,
      "learning_rate": 6.488533225368364e-06,
      "loss": 2.079,
      "step": 215122
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.5164340734481812,
      "learning_rate": 6.48768709835158e-06,
      "loss": 2.0005,
      "step": 215123
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.3790602684020996,
      "learning_rate": 6.4868410259043436e-06,
      "loss": 1.9851,
      "step": 215124
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.444040060043335,
      "learning_rate": 6.485995008026657e-06,
      "loss": 1.9788,
      "step": 215125
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.398546576499939,
      "learning_rate": 6.485149044718785e-06,
      "loss": 2.1185,
      "step": 215126
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.3781025409698486,
      "learning_rate": 6.4843031359808616e-06,
      "loss": 1.7644,
      "step": 215127
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.491977572441101,
      "learning_rate": 6.483457281813054e-06,
      "loss": 1.7553,
      "step": 215128
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.4560885429382324,
      "learning_rate": 6.482611482215527e-06,
      "loss": 1.9313,
      "step": 215129
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.4218934774398804,
      "learning_rate": 6.481765737188449e-06,
      "loss": 1.9359,
      "step": 215130
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.4014068841934204,
      "learning_rate": 6.480920046731886e-06,
      "loss": 1.9397,
      "step": 215131
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.538608431816101,
      "learning_rate": 6.480074410846103e-06,
      "loss": 2.0147,
      "step": 215132
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.388594388961792,
      "learning_rate": 6.479228829531169e-06,
      "loss": 1.8437,
      "step": 215133
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.4076920747756958,
      "learning_rate": 6.478383302787316e-06,
      "loss": 1.9976,
      "step": 215134
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.430809497833252,
      "learning_rate": 6.477537830614676e-06,
      "loss": 1.848,
      "step": 215135
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.4001595973968506,
      "learning_rate": 6.4766924130134175e-06,
      "loss": 1.9364,
      "step": 215136
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.4334561824798584,
      "learning_rate": 6.4758470499836735e-06,
      "loss": 1.9482,
      "step": 215137
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.2717351913452148,
      "learning_rate": 6.4750017415256095e-06,
      "loss": 1.9168,
      "step": 215138
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.4728550910949707,
      "learning_rate": 6.474156487639359e-06,
      "loss": 1.8843,
      "step": 215139
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.5801472663879395,
      "learning_rate": 6.473311288325122e-06,
      "loss": 2.1535,
      "step": 215140
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.5536366701126099,
      "learning_rate": 6.472466143583033e-06,
      "loss": 1.9492,
      "step": 215141
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.4825109243392944,
      "learning_rate": 6.4716210534132894e-06,
      "loss": 1.9655,
      "step": 215142
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.4101303815841675,
      "learning_rate": 6.470776017815959e-06,
      "loss": 2.0868,
      "step": 215143
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.3656976222991943,
      "learning_rate": 6.469931036791276e-06,
      "loss": 1.9417,
      "step": 215144
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.5161542892456055,
      "learning_rate": 6.469086110339339e-06,
      "loss": 2.1296,
      "step": 215145
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.5161328315734863,
      "learning_rate": 6.468241238460381e-06,
      "loss": 1.969,
      "step": 215146
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.4085079431533813,
      "learning_rate": 6.467396421154503e-06,
      "loss": 1.9243,
      "step": 215147
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.728255271911621,
      "learning_rate": 6.4665516584219035e-06,
      "loss": 1.9473,
      "step": 215148
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.4570623636245728,
      "learning_rate": 6.465706950262683e-06,
      "loss": 2.0603,
      "step": 215149
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.3582844734191895,
      "learning_rate": 6.464862296677042e-06,
      "loss": 1.8163,
      "step": 215150
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.4746850728988647,
      "learning_rate": 6.4640176976651136e-06,
      "loss": 1.8911,
      "step": 215151
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.579629898071289,
      "learning_rate": 6.463173153227064e-06,
      "loss": 1.7779,
      "step": 215152
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.4653102159500122,
      "learning_rate": 6.462328663363059e-06,
      "loss": 2.1762,
      "step": 215153
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.491466760635376,
      "learning_rate": 6.461484228073233e-06,
      "loss": 1.9831,
      "step": 215154
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.4344755411148071,
      "learning_rate": 6.460639847357785e-06,
      "loss": 1.818,
      "step": 215155
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.6188180446624756,
      "learning_rate": 6.459795521216849e-06,
      "loss": 2.0642,
      "step": 215156
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.3756129741668701,
      "learning_rate": 6.458951249650557e-06,
      "loss": 1.9624,
      "step": 215157
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.456977367401123,
      "learning_rate": 6.4581070326590765e-06,
      "loss": 1.971,
      "step": 215158
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.3948662281036377,
      "learning_rate": 6.457262870242574e-06,
      "loss": 1.82,
      "step": 215159
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.4856830835342407,
      "learning_rate": 6.4564187624012495e-06,
      "loss": 1.9967,
      "step": 215160
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.3387641906738281,
      "learning_rate": 6.455574709135203e-06,
      "loss": 1.8666,
      "step": 215161
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.3869714736938477,
      "learning_rate": 6.4547307104446e-06,
      "loss": 2.1412,
      "step": 215162
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.4813272953033447,
      "learning_rate": 6.453886766329574e-06,
      "loss": 1.9567,
      "step": 215163
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.931750774383545,
      "learning_rate": 6.453042876790327e-06,
      "loss": 2.0868,
      "step": 215164
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.5010873079299927,
      "learning_rate": 6.452199041826989e-06,
      "loss": 1.9542,
      "step": 215165
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.5451061725616455,
      "learning_rate": 6.451355261439761e-06,
      "loss": 1.9994,
      "step": 215166
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.3675442934036255,
      "learning_rate": 6.4505115356287775e-06,
      "loss": 2.1757,
      "step": 215167
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.4515516757965088,
      "learning_rate": 6.449667864394137e-06,
      "loss": 1.9927,
      "step": 215168
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.4286894798278809,
      "learning_rate": 6.448824247736073e-06,
      "loss": 1.9344,
      "step": 215169
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.4352872371673584,
      "learning_rate": 6.447980685654719e-06,
      "loss": 2.0747,
      "step": 215170
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.5329684019088745,
      "learning_rate": 6.447137178150208e-06,
      "loss": 1.8832,
      "step": 215171
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.395914912223816,
      "learning_rate": 6.44629372522274e-06,
      "loss": 1.7338,
      "step": 215172
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.4673093557357788,
      "learning_rate": 6.445450326872448e-06,
      "loss": 1.9491,
      "step": 215173
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.3426278829574585,
      "learning_rate": 6.444606983099432e-06,
      "loss": 1.8338,
      "step": 215174
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.7825653553009033,
      "learning_rate": 6.443763693903958e-06,
      "loss": 2.1079,
      "step": 215175
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.5791802406311035,
      "learning_rate": 6.442920459286094e-06,
      "loss": 2.1692,
      "step": 215176
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.4669713973999023,
      "learning_rate": 6.4420772792460715e-06,
      "loss": 2.1272,
      "step": 215177
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.5138988494873047,
      "learning_rate": 6.441234153783992e-06,
      "loss": 1.9566,
      "step": 215178
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.6134659051895142,
      "learning_rate": 6.440391082900054e-06,
      "loss": 2.1669,
      "step": 215179
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.350131869316101,
      "learning_rate": 6.4395480665943575e-06,
      "loss": 1.967,
      "step": 215180
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.4976210594177246,
      "learning_rate": 6.438705104867103e-06,
      "loss": 2.0191,
      "step": 215181
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.5859081745147705,
      "learning_rate": 6.4378621977183905e-06,
      "loss": 2.1445,
      "step": 215182
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.5072561502456665,
      "learning_rate": 6.437019345148486e-06,
      "loss": 2.1089,
      "step": 215183
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.314383625984192,
      "learning_rate": 6.436176547157457e-06,
      "loss": 1.9325,
      "step": 215184
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.6821479797363281,
      "learning_rate": 6.4353338037454686e-06,
      "loss": 1.9645,
      "step": 215185
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.493696689605713,
      "learning_rate": 6.434491114912688e-06,
      "loss": 1.8525,
      "step": 215186
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.5165740251541138,
      "learning_rate": 6.4336484806593146e-06,
      "loss": 2.0043,
      "step": 215187
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.4524188041687012,
      "learning_rate": 6.432805900985416e-06,
      "loss": 1.8686,
      "step": 215188
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.4692187309265137,
      "learning_rate": 6.4319633758912585e-06,
      "loss": 1.8465,
      "step": 215189
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.406536340713501,
      "learning_rate": 6.431120905376874e-06,
      "loss": 1.9971,
      "step": 215190
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.5106828212738037,
      "learning_rate": 6.430278489442564e-06,
      "loss": 2.104,
      "step": 215191
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.3872425556182861,
      "learning_rate": 6.429436128088361e-06,
      "loss": 1.6739,
      "step": 215192
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.3868340253829956,
      "learning_rate": 6.428593821314498e-06,
      "loss": 1.921,
      "step": 215193
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.4231712818145752,
      "learning_rate": 6.427751569121042e-06,
      "loss": 1.7608,
      "step": 215194
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.4556909799575806,
      "learning_rate": 6.42690937150826e-06,
      "loss": 1.8987,
      "step": 215195
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.5453665256500244,
      "learning_rate": 6.4260672284762175e-06,
      "loss": 1.7942,
      "step": 215196
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.4727287292480469,
      "learning_rate": 6.425225140025181e-06,
      "loss": 1.8765,
      "step": 215197
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.4951276779174805,
      "learning_rate": 6.424383106155151e-06,
      "loss": 1.9191,
      "step": 215198
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.5354180335998535,
      "learning_rate": 6.423541126866427e-06,
      "loss": 1.9841,
      "step": 215199
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.5636744499206543,
      "learning_rate": 6.422699202159076e-06,
      "loss": 1.8912,
      "step": 215200
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.4975510835647583,
      "learning_rate": 6.421857332033298e-06,
      "loss": 2.0627,
      "step": 215201
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.4307010173797607,
      "learning_rate": 6.421015516489225e-06,
      "loss": 1.895,
      "step": 215202
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.6233733892440796,
      "learning_rate": 6.420173755527091e-06,
      "loss": 2.0395,
      "step": 215203
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.5709840059280396,
      "learning_rate": 6.41933204914693e-06,
      "loss": 2.0555,
      "step": 215204
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.470755934715271,
      "learning_rate": 6.4184903973489744e-06,
      "loss": 1.844,
      "step": 215205
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.3777650594711304,
      "learning_rate": 6.417648800133324e-06,
      "loss": 2.0164,
      "step": 215206
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.5816097259521484,
      "learning_rate": 6.416807257500212e-06,
      "loss": 2.0803,
      "step": 215207
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.4337756633758545,
      "learning_rate": 6.415965769449738e-06,
      "loss": 2.0457,
      "step": 215208
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.5477784872055054,
      "learning_rate": 6.415124335982136e-06,
      "loss": 2.0329,
      "step": 215209
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.5247942209243774,
      "learning_rate": 6.414282957097439e-06,
      "loss": 1.8988,
      "step": 215210
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.4062058925628662,
      "learning_rate": 6.4134416327959125e-06,
      "loss": 2.0315,
      "step": 215211
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.6513779163360596,
      "learning_rate": 6.412600363077625e-06,
      "loss": 1.8667,
      "step": 215212
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.59932541847229,
      "learning_rate": 6.411759147942807e-06,
      "loss": 1.9945,
      "step": 215213
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.5433220863342285,
      "learning_rate": 6.410917987391561e-06,
      "loss": 2.2237,
      "step": 215214
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.483351707458496,
      "learning_rate": 6.41007688142412e-06,
      "loss": 1.9254,
      "step": 215215
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.5456366539001465,
      "learning_rate": 6.409235830040515e-06,
      "loss": 1.9603,
      "step": 215216
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.3702564239501953,
      "learning_rate": 6.408394833241015e-06,
      "loss": 1.8839,
      "step": 215217
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.5199501514434814,
      "learning_rate": 6.4075538910257185e-06,
      "loss": 1.9867,
      "step": 215218
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.5375280380249023,
      "learning_rate": 6.406713003394826e-06,
      "loss": 1.9516,
      "step": 215219
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.4990389347076416,
      "learning_rate": 6.405872170348436e-06,
      "loss": 2.0994,
      "step": 215220
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.4395216703414917,
      "learning_rate": 6.405031391886783e-06,
      "loss": 1.8317,
      "step": 215221
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.3721400499343872,
      "learning_rate": 6.404190668009934e-06,
      "loss": 2.0125,
      "step": 215222
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.5647109746932983,
      "learning_rate": 6.403349998718122e-06,
      "loss": 2.0073,
      "step": 215223
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.536306381225586,
      "learning_rate": 6.4025093840114115e-06,
      "loss": 2.0492,
      "step": 215224
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.5022140741348267,
      "learning_rate": 6.401668823890072e-06,
      "loss": 1.8392,
      "step": 215225
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.4373486042022705,
      "learning_rate": 6.4008283183541676e-06,
      "loss": 1.8974,
      "step": 215226
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.423431634902954,
      "learning_rate": 6.399987867403966e-06,
      "loss": 2.041,
      "step": 215227
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.4399806261062622,
      "learning_rate": 6.399147471039468e-06,
      "loss": 1.7557,
      "step": 215228
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.5742086172103882,
      "learning_rate": 6.398307129260938e-06,
      "loss": 1.9727,
      "step": 215229
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.603218674659729,
      "learning_rate": 6.3974668420684775e-06,
      "loss": 1.9754,
      "step": 215230
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.426016092300415,
      "learning_rate": 6.396626609462318e-06,
      "loss": 1.9775,
      "step": 215231
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.4389344453811646,
      "learning_rate": 6.395786431442529e-06,
      "loss": 1.9543,
      "step": 215232
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.484820008277893,
      "learning_rate": 6.394946308009341e-06,
      "loss": 1.8827,
      "step": 215233
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.6042808294296265,
      "learning_rate": 6.394106239162855e-06,
      "loss": 1.8102,
      "step": 215234
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.4331026077270508,
      "learning_rate": 6.3932662249032375e-06,
      "loss": 2.2312,
      "step": 215235
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.4900922775268555,
      "learning_rate": 6.3924262652306546e-06,
      "loss": 1.9852,
      "step": 215236
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.7361313104629517,
      "learning_rate": 6.391586360145273e-06,
      "loss": 2.091,
      "step": 215237
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.4806655645370483,
      "learning_rate": 6.390746509647227e-06,
      "loss": 2.1017,
      "step": 215238
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.5141090154647827,
      "learning_rate": 6.389906713736714e-06,
      "loss": 1.9102,
      "step": 215239
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.4140692949295044,
      "learning_rate": 6.389066972413837e-06,
      "loss": 1.9576,
      "step": 215240
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.4988890886306763,
      "learning_rate": 6.3882272856787934e-06,
      "loss": 2.0678,
      "step": 215241
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.506577968597412,
      "learning_rate": 6.3873876535316835e-06,
      "loss": 1.7992,
      "step": 215242
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.5985394716262817,
      "learning_rate": 6.386548075972742e-06,
      "loss": 2.2215,
      "step": 215243
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.8062870502471924,
      "learning_rate": 6.3857085530020334e-06,
      "loss": 2.0113,
      "step": 215244
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.3477774858474731,
      "learning_rate": 6.384869084619792e-06,
      "loss": 1.9007,
      "step": 215245
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.4338935613632202,
      "learning_rate": 6.3840296708261515e-06,
      "loss": 2.017,
      "step": 215246
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.4222133159637451,
      "learning_rate": 6.383190311621278e-06,
      "loss": 1.8462,
      "step": 215247
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.376007080078125,
      "learning_rate": 6.382351007005271e-06,
      "loss": 2.0858,
      "step": 215248
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.7099982500076294,
      "learning_rate": 6.381511756978363e-06,
      "loss": 2.0862,
      "step": 215249
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.4139851331710815,
      "learning_rate": 6.380672561540623e-06,
      "loss": 2.095,
      "step": 215250
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.5268535614013672,
      "learning_rate": 6.379833420692316e-06,
      "loss": 1.9993,
      "step": 215251
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.4299787282943726,
      "learning_rate": 6.3789943344335405e-06,
      "loss": 2.0663,
      "step": 215252
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.4663424491882324,
      "learning_rate": 6.378155302764398e-06,
      "loss": 2.214,
      "step": 215253
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.512352466583252,
      "learning_rate": 6.377316325685155e-06,
      "loss": 1.7975,
      "step": 215254
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.445493221282959,
      "learning_rate": 6.376477403195879e-06,
      "loss": 1.9986,
      "step": 215255
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.4771171808242798,
      "learning_rate": 6.3756385352967674e-06,
      "loss": 2.0075,
      "step": 215256
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.511497974395752,
      "learning_rate": 6.374799721987955e-06,
      "loss": 2.0181,
      "step": 215257
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.4874157905578613,
      "learning_rate": 6.373960963269642e-06,
      "loss": 1.8219,
      "step": 215258
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.4699077606201172,
      "learning_rate": 6.3731222591419275e-06,
      "loss": 1.886,
      "step": 215259
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.4334768056869507,
      "learning_rate": 6.372283609605011e-06,
      "loss": 2.0478,
      "step": 215260
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.4368640184402466,
      "learning_rate": 6.371445014659027e-06,
      "loss": 1.826,
      "step": 215261
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.48720121383667,
      "learning_rate": 6.370606474304108e-06,
      "loss": 1.9699,
      "step": 215262
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.472686767578125,
      "learning_rate": 6.369767988540453e-06,
      "loss": 2.1826,
      "step": 215263
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.5044207572937012,
      "learning_rate": 6.368929557368196e-06,
      "loss": 1.9244,
      "step": 215264
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.5778640508651733,
      "learning_rate": 6.368091180787471e-06,
      "loss": 1.9112,
      "step": 215265
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.4781270027160645,
      "learning_rate": 6.3672528587985106e-06,
      "loss": 2.2176,
      "step": 215266
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.7871607542037964,
      "learning_rate": 6.366414591401381e-06,
      "loss": 1.9267,
      "step": 215267
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.411700963973999,
      "learning_rate": 6.365576378596316e-06,
      "loss": 2.1119,
      "step": 215268
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.3491681814193726,
      "learning_rate": 6.364738220383414e-06,
      "loss": 1.933,
      "step": 215269
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.5397619009017944,
      "learning_rate": 6.363900116762877e-06,
      "loss": 1.8286,
      "step": 215270
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.460322380065918,
      "learning_rate": 6.36306206773477e-06,
      "loss": 1.9815,
      "step": 215271
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.4924719333648682,
      "learning_rate": 6.36222407329936e-06,
      "loss": 1.918,
      "step": 215272
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.5205951929092407,
      "learning_rate": 6.3613861334567475e-06,
      "loss": 1.8903,
      "step": 215273
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.4879759550094604,
      "learning_rate": 6.360548248207098e-06,
      "loss": 2.0646,
      "step": 215274
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.4264352321624756,
      "learning_rate": 6.3597104175505454e-06,
      "loss": 2.0112,
      "step": 215275
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.5284005403518677,
      "learning_rate": 6.358872641487289e-06,
      "loss": 2.1755,
      "step": 215276
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.4544410705566406,
      "learning_rate": 6.358034920017463e-06,
      "loss": 1.9358,
      "step": 215277
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.6372220516204834,
      "learning_rate": 6.3571972531411995e-06,
      "loss": 1.8979,
      "step": 215278
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.4113484621047974,
      "learning_rate": 6.356359640858666e-06,
      "loss": 1.7699,
      "step": 215279
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.4082392454147339,
      "learning_rate": 6.3555220831700615e-06,
      "loss": 2.149,
      "step": 215280
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.3825209140777588,
      "learning_rate": 6.354684580075487e-06,
      "loss": 1.7899,
      "step": 215281
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.4883249998092651,
      "learning_rate": 6.3538471315751735e-06,
      "loss": 1.7536,
      "step": 215282
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.54893159866333,
      "learning_rate": 6.353009737669157e-06,
      "loss": 1.7859,
      "step": 215283
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.4640777111053467,
      "learning_rate": 6.352172398357702e-06,
      "loss": 2.1239,
      "step": 215284
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.479738473892212,
      "learning_rate": 6.3513351136408764e-06,
      "loss": 1.6772,
      "step": 215285
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.5885543823242188,
      "learning_rate": 6.350497883518912e-06,
      "loss": 1.987,
      "step": 215286
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.4830228090286255,
      "learning_rate": 6.3496607079919106e-06,
      "loss": 2.1912,
      "step": 215287
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.342326283454895,
      "learning_rate": 6.348823587060103e-06,
      "loss": 1.9325,
      "step": 215288
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.406032681465149,
      "learning_rate": 6.347986520723525e-06,
      "loss": 1.9193,
      "step": 215289
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.6456125974655151,
      "learning_rate": 6.347149508982441e-06,
      "loss": 2.1141,
      "step": 215290
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.4876742362976074,
      "learning_rate": 6.346312551836952e-06,
      "loss": 2.0529,
      "step": 215291
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.5119528770446777,
      "learning_rate": 6.3454756492872235e-06,
      "loss": 2.0814,
      "step": 215292
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.4996685981750488,
      "learning_rate": 6.34463880133339e-06,
      "loss": 2.0266,
      "step": 215293
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.3863005638122559,
      "learning_rate": 6.343802007975718e-06,
      "loss": 2.0765,
      "step": 215294
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.6122568845748901,
      "learning_rate": 6.342965269214173e-06,
      "loss": 1.9986,
      "step": 215295
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.6489251852035522,
      "learning_rate": 6.342128585049089e-06,
      "loss": 1.9586,
      "step": 215296
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.8602830171585083,
      "learning_rate": 6.341291955480498e-06,
      "loss": 1.778,
      "step": 215297
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.256186842918396,
      "learning_rate": 6.340455380508636e-06,
      "loss": 2.2809,
      "step": 215298
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.368830919265747,
      "learning_rate": 6.3396188601336e-06,
      "loss": 1.9932,
      "step": 215299
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.498220443725586,
      "learning_rate": 6.338782394355591e-06,
      "loss": 1.8956,
      "step": 215300
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.496803641319275,
      "learning_rate": 6.337945983174742e-06,
      "loss": 1.9827,
      "step": 215301
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.6346454620361328,
      "learning_rate": 6.3371096265912195e-06,
      "loss": 1.975,
      "step": 215302
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.399973750114441,
      "learning_rate": 6.336273324605124e-06,
      "loss": 1.9643,
      "step": 215303
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.3651231527328491,
      "learning_rate": 6.335437077216721e-06,
      "loss": 1.7299,
      "step": 215304
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.3815144300460815,
      "learning_rate": 6.3346008844260446e-06,
      "loss": 1.9927,
      "step": 215305
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.55085027217865,
      "learning_rate": 6.333764746233394e-06,
      "loss": 2.0319,
      "step": 215306
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.491644263267517,
      "learning_rate": 6.3329286626387695e-06,
      "loss": 1.9295,
      "step": 215307
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.4395266771316528,
      "learning_rate": 6.332092633642438e-06,
      "loss": 2.059,
      "step": 215308
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.4755576848983765,
      "learning_rate": 6.331256659244466e-06,
      "loss": 2.1311,
      "step": 215309
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.6399120092391968,
      "learning_rate": 6.330420739445086e-06,
      "loss": 1.9489,
      "step": 215310
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.6951549053192139,
      "learning_rate": 6.329584874244431e-06,
      "loss": 1.9194,
      "step": 215311
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.3717175722122192,
      "learning_rate": 6.328749063642668e-06,
      "loss": 1.7925,
      "step": 215312
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.4411041736602783,
      "learning_rate": 6.327913307639898e-06,
      "loss": 1.9324,
      "step": 215313
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.4715312719345093,
      "learning_rate": 6.327077606236319e-06,
      "loss": 2.0514,
      "step": 215314
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.6274069547653198,
      "learning_rate": 6.3262419594320655e-06,
      "loss": 2.0432,
      "step": 215315
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.4045077562332153,
      "learning_rate": 6.325406367227337e-06,
      "loss": 1.9155,
      "step": 215316
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.456217885017395,
      "learning_rate": 6.324570829622233e-06,
      "loss": 1.8699,
      "step": 215317
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.5628217458724976,
      "learning_rate": 6.323735346616954e-06,
      "loss": 2.2474,
      "step": 215318
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.4981998205184937,
      "learning_rate": 6.322899918211633e-06,
      "loss": 1.9107,
      "step": 215319
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.3962594270706177,
      "learning_rate": 6.322064544406435e-06,
      "loss": 1.9661,
      "step": 215320
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.4948433637619019,
      "learning_rate": 6.321229225201463e-06,
      "loss": 1.8935,
      "step": 215321
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.620582103729248,
      "learning_rate": 6.320393960596981e-06,
      "loss": 2.0492,
      "step": 215322
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.48003351688385,
      "learning_rate": 6.319558750593023e-06,
      "loss": 1.9522,
      "step": 215323
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.5663286447525024,
      "learning_rate": 6.318723595189856e-06,
      "loss": 1.9236,
      "step": 215324
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.444619059562683,
      "learning_rate": 6.3178884943875795e-06,
      "loss": 1.8746,
      "step": 215325
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.4651899337768555,
      "learning_rate": 6.317053448186327e-06,
      "loss": 2.0325,
      "step": 215326
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.4132510423660278,
      "learning_rate": 6.3162184565862635e-06,
      "loss": 1.9019,
      "step": 215327
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.4312247037887573,
      "learning_rate": 6.315383519587592e-06,
      "loss": 1.8442,
      "step": 215328
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.3544795513153076,
      "learning_rate": 6.314548637190408e-06,
      "loss": 2.0389,
      "step": 215329
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.6577529907226562,
      "learning_rate": 6.313713809394915e-06,
      "loss": 1.9969,
      "step": 215330
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.4123815298080444,
      "learning_rate": 6.312879036201246e-06,
      "loss": 1.7814,
      "step": 215331
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.5244662761688232,
      "learning_rate": 6.312044317609566e-06,
      "loss": 1.9129,
      "step": 215332
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.4575711488723755,
      "learning_rate": 6.311209653619976e-06,
      "loss": 1.7947,
      "step": 215333
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.5869379043579102,
      "learning_rate": 6.310375044232707e-06,
      "loss": 1.8903,
      "step": 215334
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.5246959924697876,
      "learning_rate": 6.309540489447862e-06,
      "loss": 2.1833,
      "step": 215335
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.4536701440811157,
      "learning_rate": 6.308705989265639e-06,
      "loss": 1.9823,
      "step": 215336
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.4801414012908936,
      "learning_rate": 6.307871543686171e-06,
      "loss": 1.8614,
      "step": 215337
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.3894121646881104,
      "learning_rate": 6.307037152709593e-06,
      "loss": 1.8644,
      "step": 215338
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.4178794622421265,
      "learning_rate": 6.3062028163361034e-06,
      "loss": 2.1808,
      "step": 215339
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.4217002391815186,
      "learning_rate": 6.305368534565835e-06,
      "loss": 1.7884,
      "step": 215340
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.4685956239700317,
      "learning_rate": 6.304534307398923e-06,
      "loss": 2.0102,
      "step": 215341
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.4812356233596802,
      "learning_rate": 6.303700134835565e-06,
      "loss": 2.0639,
      "step": 215342
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.3586676120758057,
      "learning_rate": 6.302866016875896e-06,
      "loss": 1.9814,
      "step": 215343
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.4345117807388306,
      "learning_rate": 6.302031953520015e-06,
      "loss": 2.0303,
      "step": 215344
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.3649708032608032,
      "learning_rate": 6.301197944768188e-06,
      "loss": 2.0177,
      "step": 215345
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.5465047359466553,
      "learning_rate": 6.300363990620516e-06,
      "loss": 1.9352,
      "step": 215346
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.3776812553405762,
      "learning_rate": 6.299530091077098e-06,
      "loss": 2.0222,
      "step": 215347
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.408262014389038,
      "learning_rate": 6.2986962461381685e-06,
      "loss": 2.1013,
      "step": 215348
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.4177788496017456,
      "learning_rate": 6.297862455803893e-06,
      "loss": 1.9428,
      "step": 215349
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.370131254196167,
      "learning_rate": 6.297028720074337e-06,
      "loss": 2.05,
      "step": 215350
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.380420207977295,
      "learning_rate": 6.2961950389497365e-06,
      "loss": 1.9714,
      "step": 215351
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.4071331024169922,
      "learning_rate": 6.295361412430189e-06,
      "loss": 1.9758,
      "step": 215352
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.5315728187561035,
      "learning_rate": 6.294527840515895e-06,
      "loss": 1.9011,
      "step": 215353
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.450286865234375,
      "learning_rate": 6.293694323206988e-06,
      "loss": 1.865,
      "step": 215354
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.5600875616073608,
      "learning_rate": 6.292860860503634e-06,
      "loss": 1.8979,
      "step": 215355
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.2848891019821167,
      "learning_rate": 6.292027452405968e-06,
      "loss": 1.9255,
      "step": 215356
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.5735803842544556,
      "learning_rate": 6.291194098914187e-06,
      "loss": 2.1391,
      "step": 215357
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.4195696115493774,
      "learning_rate": 6.2903608000283934e-06,
      "loss": 2.0244,
      "step": 215358
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.3968558311462402,
      "learning_rate": 6.289527555748752e-06,
      "loss": 2.0095,
      "step": 215359
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.5044671297073364,
      "learning_rate": 6.288694366075464e-06,
      "loss": 2.1163,
      "step": 215360
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.6005276441574097,
      "learning_rate": 6.287861231008662e-06,
      "loss": 1.9103,
      "step": 215361
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.4890650510787964,
      "learning_rate": 6.287028150548445e-06,
      "loss": 2.0999,
      "step": 215362
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.4073963165283203,
      "learning_rate": 6.286195124695048e-06,
      "loss": 1.9154,
      "step": 215363
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.3344370126724243,
      "learning_rate": 6.28536215344857e-06,
      "loss": 1.8535,
      "step": 215364
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.4002596139907837,
      "learning_rate": 6.284529236809211e-06,
      "loss": 1.9551,
      "step": 215365
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.530747652053833,
      "learning_rate": 6.28369637477707e-06,
      "loss": 1.9313,
      "step": 215366
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.28705894947052,
      "learning_rate": 6.282863567352381e-06,
      "loss": 1.8392,
      "step": 215367
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.4756656885147095,
      "learning_rate": 6.282030814535211e-06,
      "loss": 1.9095,
      "step": 215368
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.4211506843566895,
      "learning_rate": 6.281198116325758e-06,
      "loss": 1.8,
      "step": 215369
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.475921630859375,
      "learning_rate": 6.280365472724191e-06,
      "loss": 1.8677,
      "step": 215370
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.5427284240722656,
      "learning_rate": 6.279532883730643e-06,
      "loss": 1.8118,
      "step": 215371
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.63318932056427,
      "learning_rate": 6.278700349345245e-06,
      "loss": 2.0979,
      "step": 215372
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.4444423913955688,
      "learning_rate": 6.277867869568265e-06,
      "loss": 1.9128,
      "step": 215373
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.4702621698379517,
      "learning_rate": 6.277035444399703e-06,
      "loss": 2.0294,
      "step": 215374
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.3677018880844116,
      "learning_rate": 6.276203073839791e-06,
      "loss": 1.8869,
      "step": 215375
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.4869076013565063,
      "learning_rate": 6.275370757888698e-06,
      "loss": 1.9727,
      "step": 215376
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.4656176567077637,
      "learning_rate": 6.274538496546555e-06,
      "loss": 2.024,
      "step": 215377
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.413191795349121,
      "learning_rate": 6.273706289813496e-06,
      "loss": 1.9989,
      "step": 215378
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.6118327379226685,
      "learning_rate": 6.272874137689754e-06,
      "loss": 1.9752,
      "step": 215379
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.584700107574463,
      "learning_rate": 6.272042040175362e-06,
      "loss": 1.9345,
      "step": 215380
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.551914930343628,
      "learning_rate": 6.271209997270588e-06,
      "loss": 2.109,
      "step": 215381
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.4465605020523071,
      "learning_rate": 6.2703780089754965e-06,
      "loss": 1.8567,
      "step": 215382
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.4811009168624878,
      "learning_rate": 6.269546075290322e-06,
      "loss": 2.1185,
      "step": 215383
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.2852511405944824,
      "learning_rate": 6.268714196215163e-06,
      "loss": 2.0415,
      "step": 215384
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.448961615562439,
      "learning_rate": 6.267882371750222e-06,
      "loss": 2.0755,
      "step": 215385
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.5560482740402222,
      "learning_rate": 6.267050601895595e-06,
      "loss": 2.0051,
      "step": 215386
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.5124719142913818,
      "learning_rate": 6.2662188866514864e-06,
      "loss": 1.8486,
      "step": 215387
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.4520974159240723,
      "learning_rate": 6.265387226018026e-06,
      "loss": 1.9932,
      "step": 215388
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.4013103246688843,
      "learning_rate": 6.264555619995382e-06,
      "loss": 1.926,
      "step": 215389
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.4672869443893433,
      "learning_rate": 6.2637240685836865e-06,
      "loss": 1.9896,
      "step": 215390
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.4486037492752075,
      "learning_rate": 6.2628925717831405e-06,
      "loss": 1.7509,
      "step": 215391
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.4009562730789185,
      "learning_rate": 6.262061129593843e-06,
      "loss": 1.7843,
      "step": 215392
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.5924930572509766,
      "learning_rate": 6.261229742015994e-06,
      "loss": 2.0141,
      "step": 215393
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.4474447965621948,
      "learning_rate": 6.260398409049694e-06,
      "loss": 1.9749,
      "step": 215394
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.4313124418258667,
      "learning_rate": 6.259567130695142e-06,
      "loss": 2.0022,
      "step": 215395
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.4865821599960327,
      "learning_rate": 6.258735906952472e-06,
      "loss": 1.8076,
      "step": 215396
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.4558465480804443,
      "learning_rate": 6.257904737821884e-06,
      "loss": 2.1516,
      "step": 215397
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.5008825063705444,
      "learning_rate": 6.257073623303477e-06,
      "loss": 1.9465,
      "step": 215398
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.42532217502594,
      "learning_rate": 6.256242563397451e-06,
      "loss": 1.9831,
      "step": 215399
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.5482213497161865,
      "learning_rate": 6.255411558103873e-06,
      "loss": 1.9958,
      "step": 215400
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.9518885612487793,
      "learning_rate": 6.25458060742301e-06,
      "loss": 1.9786,
      "step": 215401
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.4018319845199585,
      "learning_rate": 6.2537497113549275e-06,
      "loss": 2.1971,
      "step": 215402
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.4831348657608032,
      "learning_rate": 6.252918869899826e-06,
      "loss": 2.0471,
      "step": 215403
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.404723882675171,
      "learning_rate": 6.252088083057905e-06,
      "loss": 1.9146,
      "step": 215404
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.4318886995315552,
      "learning_rate": 6.251257350829231e-06,
      "loss": 1.9117,
      "step": 215405
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.6120178699493408,
      "learning_rate": 6.250426673213971e-06,
      "loss": 1.9788,
      "step": 215406
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.6412649154663086,
      "learning_rate": 6.249596050212324e-06,
      "loss": 1.905,
      "step": 215407
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.396742820739746,
      "learning_rate": 6.248765481824392e-06,
      "loss": 2.097,
      "step": 215408
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.4319182634353638,
      "learning_rate": 6.247934968050372e-06,
      "loss": 1.9091,
      "step": 215409
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.493803858757019,
      "learning_rate": 6.247104508890433e-06,
      "loss": 1.9616,
      "step": 215410
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.3873112201690674,
      "learning_rate": 6.246274104344706e-06,
      "loss": 2.0977,
      "step": 215411
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.6716439723968506,
      "learning_rate": 6.245443754413293e-06,
      "loss": 1.861,
      "step": 215412
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.3792104721069336,
      "learning_rate": 6.2446134590964255e-06,
      "loss": 2.069,
      "step": 215413
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.4524801969528198,
      "learning_rate": 6.243783218394238e-06,
      "loss": 1.8336,
      "step": 215414
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.3921793699264526,
      "learning_rate": 6.242953032306863e-06,
      "loss": 1.7748,
      "step": 215415
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.3718141317367554,
      "learning_rate": 6.242122900834501e-06,
      "loss": 1.9785,
      "step": 215416
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.504550814628601,
      "learning_rate": 6.2412928239772505e-06,
      "loss": 1.9219,
      "step": 215417
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.418554425239563,
      "learning_rate": 6.2404628017352465e-06,
      "loss": 1.9411,
      "step": 215418
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.502017855644226,
      "learning_rate": 6.239632834108755e-06,
      "loss": 2.0722,
      "step": 215419
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.46412193775177,
      "learning_rate": 6.238802921097807e-06,
      "loss": 1.8169,
      "step": 215420
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.4297841787338257,
      "learning_rate": 6.237973062702639e-06,
      "loss": 2.1558,
      "step": 215421
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.4778048992156982,
      "learning_rate": 6.237143258923383e-06,
      "loss": 2.0277,
      "step": 215422
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.4598497152328491,
      "learning_rate": 6.236313509760171e-06,
      "loss": 1.9682,
      "step": 215423
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.4895249605178833,
      "learning_rate": 6.235483815213171e-06,
      "loss": 2.1036,
      "step": 215424
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.4000283479690552,
      "learning_rate": 6.23465417528255e-06,
      "loss": 1.9794,
      "step": 215425
    },
    {
      "epoch": 2.8,
      "grad_norm": 1.4068052768707275,
      "learning_rate": 6.2338245899684395e-06,
      "loss": 1.8008,
      "step": 215426
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.49920654296875,
      "learning_rate": 6.232995059271007e-06,
      "loss": 1.998,
      "step": 215427
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.356744408607483,
      "learning_rate": 6.2321655831904185e-06,
      "loss": 1.8122,
      "step": 215428
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.4750864505767822,
      "learning_rate": 6.231336161726808e-06,
      "loss": 1.9089,
      "step": 215429
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.4719499349594116,
      "learning_rate": 6.2305067948803414e-06,
      "loss": 2.0447,
      "step": 215430
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.3159208297729492,
      "learning_rate": 6.2296774826511855e-06,
      "loss": 2.0859,
      "step": 215431
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.5384093523025513,
      "learning_rate": 6.22884822503944e-06,
      "loss": 2.0091,
      "step": 215432
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.3235057592391968,
      "learning_rate": 6.228019022045338e-06,
      "loss": 2.072,
      "step": 215433
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.5575041770935059,
      "learning_rate": 6.227189873668981e-06,
      "loss": 1.9931,
      "step": 215434
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.4300719499588013,
      "learning_rate": 6.2263607799105e-06,
      "loss": 1.9807,
      "step": 215435
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.617314100265503,
      "learning_rate": 6.2255317407700956e-06,
      "loss": 1.6842,
      "step": 215436
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.6079531908035278,
      "learning_rate": 6.224702756247934e-06,
      "loss": 1.9237,
      "step": 215437
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.3670539855957031,
      "learning_rate": 6.22387382634415e-06,
      "loss": 1.9998,
      "step": 215438
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.5048248767852783,
      "learning_rate": 6.223044951058875e-06,
      "loss": 1.9757,
      "step": 215439
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.445677399635315,
      "learning_rate": 6.222216130392277e-06,
      "loss": 1.7385,
      "step": 215440
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.669714093208313,
      "learning_rate": 6.221387364344521e-06,
      "loss": 2.1164,
      "step": 215441
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.49501371383667,
      "learning_rate": 6.2205586529157745e-06,
      "loss": 1.95,
      "step": 215442
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.311975359916687,
      "learning_rate": 6.219729996106137e-06,
      "loss": 2.0754,
      "step": 215443
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.4112271070480347,
      "learning_rate": 6.21890139391581e-06,
      "loss": 1.8039,
      "step": 215444
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.4530980587005615,
      "learning_rate": 6.218072846344957e-06,
      "loss": 2.1382,
      "step": 215445
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.5117064714431763,
      "learning_rate": 6.217244353393713e-06,
      "loss": 2.1161,
      "step": 215446
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.4941173791885376,
      "learning_rate": 6.216415915062179e-06,
      "loss": 2.1399,
      "step": 215447
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.6391656398773193,
      "learning_rate": 6.215587531350585e-06,
      "loss": 1.9159,
      "step": 215448
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.494147539138794,
      "learning_rate": 6.214759202259034e-06,
      "loss": 1.9167,
      "step": 215449
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.423126459121704,
      "learning_rate": 6.213930927787758e-06,
      "loss": 1.9003,
      "step": 215450
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.4791665077209473,
      "learning_rate": 6.213102707936823e-06,
      "loss": 1.7903,
      "step": 215451
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.3720343112945557,
      "learning_rate": 6.212274542706464e-06,
      "loss": 1.9808,
      "step": 215452
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.4241281747817993,
      "learning_rate": 6.211446432096712e-06,
      "loss": 1.8176,
      "step": 215453
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.456661581993103,
      "learning_rate": 6.210618376107868e-06,
      "loss": 1.8907,
      "step": 215454
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.4438228607177734,
      "learning_rate": 6.209790374739965e-06,
      "loss": 1.8645,
      "step": 215455
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.524154782295227,
      "learning_rate": 6.208962427993236e-06,
      "loss": 1.8942,
      "step": 215456
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.5738049745559692,
      "learning_rate": 6.208134535867781e-06,
      "loss": 1.6742,
      "step": 215457
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.5085710287094116,
      "learning_rate": 6.207306698363801e-06,
      "loss": 1.9251,
      "step": 215458
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.5175113677978516,
      "learning_rate": 6.206478915481428e-06,
      "loss": 1.7675,
      "step": 215459
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.3527148962020874,
      "learning_rate": 6.205651187220828e-06,
      "loss": 1.8901,
      "step": 215460
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.3847215175628662,
      "learning_rate": 6.2048235135821026e-06,
      "loss": 1.8868,
      "step": 215461
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.5187352895736694,
      "learning_rate": 6.203995894565483e-06,
      "loss": 2.139,
      "step": 215462
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.2764167785644531,
      "learning_rate": 6.203168330171071e-06,
      "loss": 1.9232,
      "step": 215463
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.4581341743469238,
      "learning_rate": 6.202340820399032e-06,
      "loss": 1.8318,
      "step": 215464
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.484257459640503,
      "learning_rate": 6.201513365249566e-06,
      "loss": 1.9128,
      "step": 215465
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.4968987703323364,
      "learning_rate": 6.20068596472274e-06,
      "loss": 2.0651,
      "step": 215466
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.6174203157424927,
      "learning_rate": 6.199858618818754e-06,
      "loss": 2.0321,
      "step": 215467
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.5403516292572021,
      "learning_rate": 6.199031327537774e-06,
      "loss": 2.0106,
      "step": 215468
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.3905515670776367,
      "learning_rate": 6.198204090879932e-06,
      "loss": 1.9682,
      "step": 215469
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.5100634098052979,
      "learning_rate": 6.197376908845397e-06,
      "loss": 1.8084,
      "step": 215470
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.5944435596466064,
      "learning_rate": 6.196549781434301e-06,
      "loss": 2.206,
      "step": 215471
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.3510470390319824,
      "learning_rate": 6.195722708646844e-06,
      "loss": 2.2182,
      "step": 215472
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.4929759502410889,
      "learning_rate": 6.194895690483126e-06,
      "loss": 2.0567,
      "step": 215473
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.409980297088623,
      "learning_rate": 6.1940687269433135e-06,
      "loss": 1.9985,
      "step": 215474
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.367412805557251,
      "learning_rate": 6.193241818027572e-06,
      "loss": 2.1201,
      "step": 215475
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.6793286800384521,
      "learning_rate": 6.1924149637360695e-06,
      "loss": 2.2967,
      "step": 215476
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.4746127128601074,
      "learning_rate": 6.191588164068939e-06,
      "loss": 1.8863,
      "step": 215477
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.5819928646087646,
      "learning_rate": 6.190761419026347e-06,
      "loss": 1.9207,
      "step": 215478
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.4348634481430054,
      "learning_rate": 6.189934728608392e-06,
      "loss": 2.2499,
      "step": 215479
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.5065069198608398,
      "learning_rate": 6.189108092815343e-06,
      "loss": 1.9795,
      "step": 215480
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.3373733758926392,
      "learning_rate": 6.188281511647231e-06,
      "loss": 2.0466,
      "step": 215481
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.6490213871002197,
      "learning_rate": 6.18745498510429e-06,
      "loss": 2.0173,
      "step": 215482
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.5518858432769775,
      "learning_rate": 6.18662851318662e-06,
      "loss": 1.8519,
      "step": 215483
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.4554797410964966,
      "learning_rate": 6.185802095894454e-06,
      "loss": 1.936,
      "step": 215484
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.441899299621582,
      "learning_rate": 6.184975733227826e-06,
      "loss": 2.0039,
      "step": 215485
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.6199990510940552,
      "learning_rate": 6.184149425187002e-06,
      "loss": 1.998,
      "step": 215486
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.611627459526062,
      "learning_rate": 6.183323171772081e-06,
      "loss": 2.2383,
      "step": 215487
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.5183786153793335,
      "learning_rate": 6.182496972983231e-06,
      "loss": 1.8931,
      "step": 215488
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.3205912113189697,
      "learning_rate": 6.1816708288206175e-06,
      "loss": 2.0217,
      "step": 215489
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.5055047273635864,
      "learning_rate": 6.1808447392843404e-06,
      "loss": 1.9829,
      "step": 215490
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.5306068658828735,
      "learning_rate": 6.1800187043746005e-06,
      "loss": 1.9852,
      "step": 215491
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.4257978200912476,
      "learning_rate": 6.179192724091564e-06,
      "loss": 1.7059,
      "step": 215492
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.508673906326294,
      "learning_rate": 6.17836679843533e-06,
      "loss": 2.0138,
      "step": 215493
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.5910090208053589,
      "learning_rate": 6.177540927406099e-06,
      "loss": 1.8933,
      "step": 215494
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.5841479301452637,
      "learning_rate": 6.176715111004005e-06,
      "loss": 1.9411,
      "step": 215495
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.325192928314209,
      "learning_rate": 6.175889349229246e-06,
      "loss": 1.9266,
      "step": 215496
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.4159082174301147,
      "learning_rate": 6.175063642081857e-06,
      "loss": 1.8985,
      "step": 215497
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.4321022033691406,
      "learning_rate": 6.1742379895621365e-06,
      "loss": 1.8523,
      "step": 215498
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.4454911947250366,
      "learning_rate": 6.173412391670118e-06,
      "loss": 1.8117,
      "step": 215499
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.3908051252365112,
      "learning_rate": 6.172586848406069e-06,
      "loss": 2.0358,
      "step": 215500
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.405532956123352,
      "learning_rate": 6.1717613597700555e-06,
      "loss": 1.897,
      "step": 215501
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.4751414060592651,
      "learning_rate": 6.1709359257622435e-06,
      "loss": 1.9344,
      "step": 215502
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.3289726972579956,
      "learning_rate": 6.170110546382801e-06,
      "loss": 1.8584,
      "step": 215503
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.4266234636306763,
      "learning_rate": 6.1692852216319256e-06,
      "loss": 2.0668,
      "step": 215504
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.4836080074310303,
      "learning_rate": 6.1684599515096525e-06,
      "loss": 1.9744,
      "step": 215505
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.742742657661438,
      "learning_rate": 6.167634736016281e-06,
      "loss": 1.9036,
      "step": 215506
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.5822449922561646,
      "learning_rate": 6.166809575151876e-06,
      "loss": 2.1231,
      "step": 215507
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.4713807106018066,
      "learning_rate": 6.165984468916573e-06,
      "loss": 2.0072,
      "step": 215508
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.5126560926437378,
      "learning_rate": 6.165159417310573e-06,
      "loss": 2.2244,
      "step": 215509
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.486793875694275,
      "learning_rate": 6.164334420334038e-06,
      "loss": 1.8665,
      "step": 215510
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.5284520387649536,
      "learning_rate": 6.163509477987072e-06,
      "loss": 2.3235,
      "step": 215511
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.4249900579452515,
      "learning_rate": 6.162684590269873e-06,
      "loss": 1.8776,
      "step": 215512
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.5726314783096313,
      "learning_rate": 6.161859757182574e-06,
      "loss": 1.8252,
      "step": 215513
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.3627609014511108,
      "learning_rate": 6.1610349787253435e-06,
      "loss": 2.1201,
      "step": 215514
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.8348451852798462,
      "learning_rate": 6.160210254898312e-06,
      "loss": 1.9144,
      "step": 215515
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.360032558441162,
      "learning_rate": 6.159385585701649e-06,
      "loss": 1.9376,
      "step": 215516
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.4852880239486694,
      "learning_rate": 6.158560971135484e-06,
      "loss": 2.0447,
      "step": 215517
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.3911149501800537,
      "learning_rate": 6.1577364112000205e-06,
      "loss": 1.752,
      "step": 215518
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.4295921325683594,
      "learning_rate": 6.156911905895357e-06,
      "loss": 1.9717,
      "step": 215519
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.5081247091293335,
      "learning_rate": 6.1560874552216585e-06,
      "loss": 1.9939,
      "step": 215520
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.5079184770584106,
      "learning_rate": 6.155263059179127e-06,
      "loss": 2.0183,
      "step": 215521
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.4059510231018066,
      "learning_rate": 6.154438717767829e-06,
      "loss": 1.9552,
      "step": 215522
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.4337538480758667,
      "learning_rate": 6.15361443098803e-06,
      "loss": 1.9434,
      "step": 215523
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.5512728691101074,
      "learning_rate": 6.152790198839797e-06,
      "loss": 1.9239,
      "step": 215524
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.4324452877044678,
      "learning_rate": 6.1519660213232625e-06,
      "loss": 2.1407,
      "step": 215525
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.364540934562683,
      "learning_rate": 6.151141898438694e-06,
      "loss": 1.9285,
      "step": 215526
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.624065637588501,
      "learning_rate": 6.150317830186124e-06,
      "loss": 2.1191,
      "step": 215527
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.5999947786331177,
      "learning_rate": 6.149493816565787e-06,
      "loss": 1.9208,
      "step": 215528
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.3867518901824951,
      "learning_rate": 6.148669857577782e-06,
      "loss": 1.9478,
      "step": 215529
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.3878581523895264,
      "learning_rate": 6.147845953222308e-06,
      "loss": 1.9528,
      "step": 215530
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.2847546339035034,
      "learning_rate": 6.147022103499466e-06,
      "loss": 1.7308,
      "step": 215531
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.55186128616333,
      "learning_rate": 6.146198308409489e-06,
      "loss": 2.1173,
      "step": 215532
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.4556993246078491,
      "learning_rate": 6.145374567952444e-06,
      "loss": 2.0752,
      "step": 215533
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.6441545486450195,
      "learning_rate": 6.1445508821285295e-06,
      "loss": 1.9468,
      "step": 215534
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.3878793716430664,
      "learning_rate": 6.1437272509378795e-06,
      "loss": 1.8365,
      "step": 215535
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.6154310703277588,
      "learning_rate": 6.142903674380661e-06,
      "loss": 1.7909,
      "step": 215536
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.4341148138046265,
      "learning_rate": 6.14208015245704e-06,
      "loss": 2.2564,
      "step": 215537
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.4706475734710693,
      "learning_rate": 6.14125668516715e-06,
      "loss": 1.9313,
      "step": 215538
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.4494973421096802,
      "learning_rate": 6.140433272511158e-06,
      "loss": 1.9056,
      "step": 215539
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.5243953466415405,
      "learning_rate": 6.139609914489163e-06,
      "loss": 1.9022,
      "step": 215540
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.4240672588348389,
      "learning_rate": 6.1387866111013985e-06,
      "loss": 1.8743,
      "step": 215541
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.4683918952941895,
      "learning_rate": 6.137963362347964e-06,
      "loss": 2.0441,
      "step": 215542
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.5004067420959473,
      "learning_rate": 6.137140168229027e-06,
      "loss": 1.7915,
      "step": 215543
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.4722718000411987,
      "learning_rate": 6.136317028744752e-06,
      "loss": 1.8676,
      "step": 215544
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.5542291402816772,
      "learning_rate": 6.1354939438953084e-06,
      "loss": 2.0354,
      "step": 215545
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.6592226028442383,
      "learning_rate": 6.134670913680762e-06,
      "loss": 2.1711,
      "step": 215546
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.3592995405197144,
      "learning_rate": 6.133847938101377e-06,
      "loss": 1.9172,
      "step": 215547
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.509526014328003,
      "learning_rate": 6.133025017157223e-06,
      "loss": 2.2412,
      "step": 215548
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.4030269384384155,
      "learning_rate": 6.132202150848531e-06,
      "loss": 1.9067,
      "step": 215549
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.4667840003967285,
      "learning_rate": 6.131379339175402e-06,
      "loss": 1.8705,
      "step": 215550
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.4919623136520386,
      "learning_rate": 6.130556582137969e-06,
      "loss": 1.9095,
      "step": 215551
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.3217628002166748,
      "learning_rate": 6.129733879736398e-06,
      "loss": 1.8838,
      "step": 215552
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.2668365240097046,
      "learning_rate": 6.12891123197089e-06,
      "loss": 1.6751,
      "step": 215553
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.4989134073257446,
      "learning_rate": 6.128088638841544e-06,
      "loss": 1.9834,
      "step": 215554
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.4921941757202148,
      "learning_rate": 6.12726610034856e-06,
      "loss": 2.1017,
      "step": 215555
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.5372352600097656,
      "learning_rate": 6.126443616492071e-06,
      "loss": 2.0824,
      "step": 215556
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.3907222747802734,
      "learning_rate": 6.125621187272212e-06,
      "loss": 1.9701,
      "step": 215557
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.4085419178009033,
      "learning_rate": 6.1247988126891135e-06,
      "loss": 2.1197,
      "step": 215558
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.4312044382095337,
      "learning_rate": 6.1239764927430105e-06,
      "loss": 1.9256,
      "step": 215559
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.541890025138855,
      "learning_rate": 6.123154227433968e-06,
      "loss": 2.0442,
      "step": 215560
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.4461122751235962,
      "learning_rate": 6.122332016762188e-06,
      "loss": 1.8509,
      "step": 215561
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.5650722980499268,
      "learning_rate": 6.1215098607278354e-06,
      "loss": 2.0218,
      "step": 215562
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.4335980415344238,
      "learning_rate": 6.120687759331044e-06,
      "loss": 2.028,
      "step": 215563
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.475342869758606,
      "learning_rate": 6.119865712571914e-06,
      "loss": 2.0061,
      "step": 215564
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.4079118967056274,
      "learning_rate": 6.119043720450678e-06,
      "loss": 2.1729,
      "step": 215565
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.5855181217193604,
      "learning_rate": 6.11822178296747e-06,
      "loss": 1.9027,
      "step": 215566
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.3616315126419067,
      "learning_rate": 6.117399900122422e-06,
      "loss": 1.8613,
      "step": 215567
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.5382542610168457,
      "learning_rate": 6.116578071915701e-06,
      "loss": 1.9931,
      "step": 215568
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.4765616655349731,
      "learning_rate": 6.115756298347474e-06,
      "loss": 2.1541,
      "step": 215569
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.4462186098098755,
      "learning_rate": 6.11493457941784e-06,
      "loss": 1.9323,
      "step": 215570
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.5938247442245483,
      "learning_rate": 6.114112915127001e-06,
      "loss": 1.9972,
      "step": 215571
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.4801719188690186,
      "learning_rate": 6.113291305475088e-06,
      "loss": 2.0774,
      "step": 215572
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.6546921730041504,
      "learning_rate": 6.1124697504622675e-06,
      "loss": 1.9129,
      "step": 215573
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.6275458335876465,
      "learning_rate": 6.111648250088708e-06,
      "loss": 2.0107,
      "step": 215574
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.4299850463867188,
      "learning_rate": 6.11082680435454e-06,
      "loss": 1.9387,
      "step": 215575
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.3546278476715088,
      "learning_rate": 6.110005413259866e-06,
      "loss": 2.0456,
      "step": 215576
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.6585373878479004,
      "learning_rate": 6.109184076804952e-06,
      "loss": 2.0167,
      "step": 215577
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.4840904474258423,
      "learning_rate": 6.108362794989829e-06,
      "loss": 1.9857,
      "step": 215578
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.551416277885437,
      "learning_rate": 6.107541567814766e-06,
      "loss": 2.0718,
      "step": 215579
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.4735326766967773,
      "learning_rate": 6.106720395279829e-06,
      "loss": 1.9704,
      "step": 215580
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.566942811012268,
      "learning_rate": 6.105899277385218e-06,
      "loss": 1.9459,
      "step": 215581
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.366579532623291,
      "learning_rate": 6.105078214131032e-06,
      "loss": 1.9493,
      "step": 215582
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.48616361618042,
      "learning_rate": 6.104257205517471e-06,
      "loss": 2.2131,
      "step": 215583
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.45393967628479,
      "learning_rate": 6.103436251544669e-06,
      "loss": 1.8959,
      "step": 215584
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.4083834886550903,
      "learning_rate": 6.102615352212825e-06,
      "loss": 1.8625,
      "step": 215585
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.3975518941879272,
      "learning_rate": 6.10179450752204e-06,
      "loss": 2.0128,
      "step": 215586
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.5482351779937744,
      "learning_rate": 6.100973717472479e-06,
      "loss": 2.0805,
      "step": 215587
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.5829975605010986,
      "learning_rate": 6.100152982064277e-06,
      "loss": 1.9374,
      "step": 215588
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.5728377103805542,
      "learning_rate": 6.0993323012976325e-06,
      "loss": 2.1494,
      "step": 215589
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.3809212446212769,
      "learning_rate": 6.098511675172613e-06,
      "loss": 1.8886,
      "step": 215590
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.5099766254425049,
      "learning_rate": 6.097691103689484e-06,
      "loss": 1.8005,
      "step": 215591
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.3712852001190186,
      "learning_rate": 6.096870586848313e-06,
      "loss": 2.2135,
      "step": 215592
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.5657044649124146,
      "learning_rate": 6.096050124649332e-06,
      "loss": 2.02,
      "step": 215593
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.371820330619812,
      "learning_rate": 6.095229717092575e-06,
      "loss": 2.0389,
      "step": 215594
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.4073479175567627,
      "learning_rate": 6.09440936417831e-06,
      "loss": 1.9731,
      "step": 215595
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.549121618270874,
      "learning_rate": 6.093589065906601e-06,
      "loss": 2.0685,
      "step": 215596
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.4989045858383179,
      "learning_rate": 6.092768822277649e-06,
      "loss": 1.9389,
      "step": 215597
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.5111804008483887,
      "learning_rate": 6.091948633291621e-06,
      "loss": 1.9991,
      "step": 215598
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.536920189857483,
      "learning_rate": 6.09112849894865e-06,
      "loss": 1.9251,
      "step": 215599
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.5029162168502808,
      "learning_rate": 6.0903084192488685e-06,
      "loss": 1.9535,
      "step": 215600
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.4418927431106567,
      "learning_rate": 6.089488394192476e-06,
      "loss": 2.1918,
      "step": 215601
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.487317442893982,
      "learning_rate": 6.0886684237795415e-06,
      "loss": 1.8897,
      "step": 215602
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.439277172088623,
      "learning_rate": 6.087848508010329e-06,
      "loss": 2.1191,
      "step": 215603
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.557551622390747,
      "learning_rate": 6.087028646884873e-06,
      "loss": 1.8825,
      "step": 215604
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.4953415393829346,
      "learning_rate": 6.086208840403439e-06,
      "loss": 2.0433,
      "step": 215605
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.4965884685516357,
      "learning_rate": 6.0853890885661285e-06,
      "loss": 2.1115,
      "step": 215606
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.349290370941162,
      "learning_rate": 6.084569391373073e-06,
      "loss": 2.1307,
      "step": 215607
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.4757534265518188,
      "learning_rate": 6.08374974882444e-06,
      "loss": 1.8376,
      "step": 215608
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.4343141317367554,
      "learning_rate": 6.082930160920396e-06,
      "loss": 2.1286,
      "step": 215609
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.6832680702209473,
      "learning_rate": 6.082110627661074e-06,
      "loss": 2.2538,
      "step": 215610
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.7194193601608276,
      "learning_rate": 6.081291149046641e-06,
      "loss": 1.9805,
      "step": 215611
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.4400492906570435,
      "learning_rate": 6.080471725077262e-06,
      "loss": 1.9811,
      "step": 215612
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.6643799543380737,
      "learning_rate": 6.07965235575304e-06,
      "loss": 1.9923,
      "step": 215613
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.3865269422531128,
      "learning_rate": 6.078833041074171e-06,
      "loss": 1.9645,
      "step": 215614
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.4003727436065674,
      "learning_rate": 6.078013781040791e-06,
      "loss": 1.9614,
      "step": 215615
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.4741631746292114,
      "learning_rate": 6.077194575653066e-06,
      "loss": 1.8344,
      "step": 215616
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.4342870712280273,
      "learning_rate": 6.0763754249111285e-06,
      "loss": 1.9019,
      "step": 215617
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.5519704818725586,
      "learning_rate": 6.075556328815145e-06,
      "loss": 2.0757,
      "step": 215618
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.4128684997558594,
      "learning_rate": 6.07473728736525e-06,
      "loss": 1.7779,
      "step": 215619
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.4461115598678589,
      "learning_rate": 6.0739183005616424e-06,
      "loss": 2.0848,
      "step": 215620
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.387850284576416,
      "learning_rate": 6.0730993684044216e-06,
      "loss": 2.0049,
      "step": 215621
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.4053317308425903,
      "learning_rate": 6.072280490893755e-06,
      "loss": 1.9853,
      "step": 215622
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.4425290822982788,
      "learning_rate": 6.071461668029809e-06,
      "loss": 2.0376,
      "step": 215623
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.4520567655563354,
      "learning_rate": 6.07064289981275e-06,
      "loss": 1.8203,
      "step": 215624
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.3814475536346436,
      "learning_rate": 6.069824186242678e-06,
      "loss": 1.8243,
      "step": 215625
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.3770016431808472,
      "learning_rate": 6.069005527319759e-06,
      "loss": 1.9812,
      "step": 215626
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.5840612649917603,
      "learning_rate": 6.068186923044161e-06,
      "loss": 2.1266,
      "step": 215627
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.440264344215393,
      "learning_rate": 6.067368373416081e-06,
      "loss": 1.9359,
      "step": 215628
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.4384814500808716,
      "learning_rate": 6.066549878435589e-06,
      "loss": 1.957,
      "step": 215629
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.5603032112121582,
      "learning_rate": 6.065731438102916e-06,
      "loss": 1.9787,
      "step": 215630
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.5828073024749756,
      "learning_rate": 6.064913052418097e-06,
      "loss": 1.9373,
      "step": 215631
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.3001548051834106,
      "learning_rate": 6.06409472138143e-06,
      "loss": 1.8708,
      "step": 215632
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.3817307949066162,
      "learning_rate": 6.0632764449929485e-06,
      "loss": 1.9931,
      "step": 215633
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.4569122791290283,
      "learning_rate": 6.0624582232528865e-06,
      "loss": 1.8307,
      "step": 215634
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.3677352666854858,
      "learning_rate": 6.061640056161343e-06,
      "loss": 1.9598,
      "step": 215635
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.5258692502975464,
      "learning_rate": 6.060821943718519e-06,
      "loss": 2.2593,
      "step": 215636
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.526808738708496,
      "learning_rate": 6.0600038859244806e-06,
      "loss": 2.0869,
      "step": 215637
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.5241285562515259,
      "learning_rate": 6.059185882779494e-06,
      "loss": 1.8578,
      "step": 215638
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.3548431396484375,
      "learning_rate": 6.058367934283592e-06,
      "loss": 1.8815,
      "step": 215639
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.512747883796692,
      "learning_rate": 6.057550040437042e-06,
      "loss": 1.9476,
      "step": 215640
    },
    {
      "epoch": 2.81,
      "grad_norm": 2.4409019947052,
      "learning_rate": 6.056732201239911e-06,
      "loss": 2.0338,
      "step": 215641
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.5648906230926514,
      "learning_rate": 6.0559144166923974e-06,
      "loss": 2.1638,
      "step": 215642
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.5414749383926392,
      "learning_rate": 6.055096686794636e-06,
      "loss": 2.0827,
      "step": 215643
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.4202905893325806,
      "learning_rate": 6.054279011546792e-06,
      "loss": 2.0813,
      "step": 215644
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.4340920448303223,
      "learning_rate": 6.053461390948966e-06,
      "loss": 2.0437,
      "step": 215645
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.438785195350647,
      "learning_rate": 6.052643825001391e-06,
      "loss": 1.8323,
      "step": 215646
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.4586560726165771,
      "learning_rate": 6.051826313704167e-06,
      "loss": 2.1791,
      "step": 215647
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.4063369035720825,
      "learning_rate": 6.05100885705746e-06,
      "loss": 2.0613,
      "step": 215648
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.4540047645568848,
      "learning_rate": 6.050191455061404e-06,
      "loss": 2.1171,
      "step": 215649
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.3856353759765625,
      "learning_rate": 6.049374107716165e-06,
      "loss": 1.9832,
      "step": 215650
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.5325895547866821,
      "learning_rate": 6.04855681502191e-06,
      "loss": 1.9465,
      "step": 215651
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.4706295728683472,
      "learning_rate": 6.047739576978772e-06,
      "loss": 1.8308,
      "step": 215652
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.49037766456604,
      "learning_rate": 6.0469223935869175e-06,
      "loss": 1.9946,
      "step": 215653
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.4395437240600586,
      "learning_rate": 6.046105264846479e-06,
      "loss": 2.2161,
      "step": 215654
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.3538076877593994,
      "learning_rate": 6.0452881907576245e-06,
      "loss": 1.9533,
      "step": 215655
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.5031710863113403,
      "learning_rate": 6.044471171320486e-06,
      "loss": 1.8486,
      "step": 215656
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.4151787757873535,
      "learning_rate": 6.043654206535231e-06,
      "loss": 1.9697,
      "step": 215657
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.566085696220398,
      "learning_rate": 6.042837296402025e-06,
      "loss": 1.9675,
      "step": 215658
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.3787473440170288,
      "learning_rate": 6.042020440920969e-06,
      "loss": 1.9592,
      "step": 215659
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.4619396924972534,
      "learning_rate": 6.0412036400923295e-06,
      "loss": 1.8926,
      "step": 215660
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.529376745223999,
      "learning_rate": 6.040386893916105e-06,
      "loss": 2.0162,
      "step": 215661
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.4747081995010376,
      "learning_rate": 6.039570202392563e-06,
      "loss": 2.0993,
      "step": 215662
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.48151433467865,
      "learning_rate": 6.03875356552177e-06,
      "loss": 1.8708,
      "step": 215663
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.4900696277618408,
      "learning_rate": 6.037936983303959e-06,
      "loss": 1.8253,
      "step": 215664
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.5309659242630005,
      "learning_rate": 6.0371204557391976e-06,
      "loss": 1.9882,
      "step": 215665
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.4342247247695923,
      "learning_rate": 6.036303982827751e-06,
      "loss": 1.7401,
      "step": 215666
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.4068015813827515,
      "learning_rate": 6.035487564569652e-06,
      "loss": 1.8356,
      "step": 215667
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.4616241455078125,
      "learning_rate": 6.034671200965135e-06,
      "loss": 1.8177,
      "step": 215668
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.381406545639038,
      "learning_rate": 6.033854892014301e-06,
      "loss": 1.7037,
      "step": 215669
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.3758805990219116,
      "learning_rate": 6.033038637717347e-06,
      "loss": 1.835,
      "step": 215670
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.5148301124572754,
      "learning_rate": 6.0322224380743745e-06,
      "loss": 1.9591,
      "step": 215671
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.5433034896850586,
      "learning_rate": 6.031406293085617e-06,
      "loss": 1.8304,
      "step": 215672
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.6369632482528687,
      "learning_rate": 6.030590202751107e-06,
      "loss": 1.9409,
      "step": 215673
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.3955353498458862,
      "learning_rate": 6.029774167071077e-06,
      "loss": 2.1077,
      "step": 215674
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.603231430053711,
      "learning_rate": 6.02895818604563e-06,
      "loss": 2.2091,
      "step": 215675
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.5178639888763428,
      "learning_rate": 6.0281422596749955e-06,
      "loss": 2.0945,
      "step": 215676
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.6657238006591797,
      "learning_rate": 6.027326387959241e-06,
      "loss": 2.0888,
      "step": 215677
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.4616042375564575,
      "learning_rate": 6.026510570898635e-06,
      "loss": 1.9104,
      "step": 215678
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.3645620346069336,
      "learning_rate": 6.025694808493175e-06,
      "loss": 1.9007,
      "step": 215679
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.3422707319259644,
      "learning_rate": 6.024879100743097e-06,
      "loss": 2.0397,
      "step": 215680
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.3821184635162354,
      "learning_rate": 6.0240634476485305e-06,
      "loss": 1.9951,
      "step": 215681
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.500914216041565,
      "learning_rate": 6.023247849209678e-06,
      "loss": 1.8605,
      "step": 215682
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.4726711511611938,
      "learning_rate": 6.0224323054266385e-06,
      "loss": 1.8654,
      "step": 215683
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.5167737007141113,
      "learning_rate": 6.0216168162995796e-06,
      "loss": 2.0972,
      "step": 215684
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.4062836170196533,
      "learning_rate": 6.020801381828666e-06,
      "loss": 1.9526,
      "step": 215685
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.6058313846588135,
      "learning_rate": 6.019986002013999e-06,
      "loss": 2.0693,
      "step": 215686
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.4192538261413574,
      "learning_rate": 6.0191706768557775e-06,
      "loss": 1.9258,
      "step": 215687
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.3463722467422485,
      "learning_rate": 6.018355406354169e-06,
      "loss": 1.9011,
      "step": 215688
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.4775408506393433,
      "learning_rate": 6.01754019050924e-06,
      "loss": 2.1415,
      "step": 215689
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.521467924118042,
      "learning_rate": 6.016725029321256e-06,
      "loss": 2.0227,
      "step": 215690
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.4748358726501465,
      "learning_rate": 6.015909922790285e-06,
      "loss": 1.9166,
      "step": 215691
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.6391984224319458,
      "learning_rate": 6.015094870916526e-06,
      "loss": 2.1087,
      "step": 215692
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.491327166557312,
      "learning_rate": 6.014279873700079e-06,
      "loss": 1.9389,
      "step": 215693
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.4415937662124634,
      "learning_rate": 6.013464931141177e-06,
      "loss": 1.873,
      "step": 215694
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.4996201992034912,
      "learning_rate": 6.012650043239853e-06,
      "loss": 2.1006,
      "step": 215695
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.4293426275253296,
      "learning_rate": 6.011835209996374e-06,
      "loss": 1.8824,
      "step": 215696
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.3595702648162842,
      "learning_rate": 6.0110204314108404e-06,
      "loss": 1.8797,
      "step": 215697
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.414264440536499,
      "learning_rate": 6.010205707483384e-06,
      "loss": 2.0572,
      "step": 215698
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.445900559425354,
      "learning_rate": 6.0093910382142065e-06,
      "loss": 1.9201,
      "step": 215699
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.520458698272705,
      "learning_rate": 6.00857642360344e-06,
      "loss": 2.2047,
      "step": 215700
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.4898144006729126,
      "learning_rate": 6.007761863651183e-06,
      "loss": 1.8574,
      "step": 215701
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.4094301462173462,
      "learning_rate": 6.0069473583576725e-06,
      "loss": 1.9591,
      "step": 215702
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.4323647022247314,
      "learning_rate": 6.0061329077230045e-06,
      "loss": 1.9819,
      "step": 215703
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.5105520486831665,
      "learning_rate": 6.005318511747348e-06,
      "loss": 1.9485,
      "step": 215704
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.6680375337600708,
      "learning_rate": 6.004504170430868e-06,
      "loss": 2.044,
      "step": 215705
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.5376101732254028,
      "learning_rate": 6.003689883773699e-06,
      "loss": 1.9683,
      "step": 215706
    },
    {
      "epoch": 2.81,
      "grad_norm": 2.9555273056030273,
      "learning_rate": 6.00287565177594e-06,
      "loss": 1.8744,
      "step": 215707
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.4913097620010376,
      "learning_rate": 6.002061474437858e-06,
      "loss": 2.0241,
      "step": 215708
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.369655966758728,
      "learning_rate": 6.00124735175952e-06,
      "loss": 1.8397,
      "step": 215709
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.4604941606521606,
      "learning_rate": 6.000433283741091e-06,
      "loss": 2.1703,
      "step": 215710
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.4038722515106201,
      "learning_rate": 5.999619270382772e-06,
      "loss": 2.1319,
      "step": 215711
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.406237006187439,
      "learning_rate": 5.998805311684596e-06,
      "loss": 1.8372,
      "step": 215712
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.4157832860946655,
      "learning_rate": 5.997991407646863e-06,
      "loss": 2.0328,
      "step": 215713
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.4635975360870361,
      "learning_rate": 5.9971775582696394e-06,
      "loss": 2.3148,
      "step": 215714
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.4347484111785889,
      "learning_rate": 5.996363763553091e-06,
      "loss": 1.9458,
      "step": 215715
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.4261928796768188,
      "learning_rate": 5.99555002349732e-06,
      "loss": 1.8359,
      "step": 215716
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.459396481513977,
      "learning_rate": 5.99473633810259e-06,
      "loss": 1.9503,
      "step": 215717
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.4561984539031982,
      "learning_rate": 5.993922707368937e-06,
      "loss": 2.0271,
      "step": 215718
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.491292953491211,
      "learning_rate": 5.993109131296592e-06,
      "loss": 1.9878,
      "step": 215719
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.3775992393493652,
      "learning_rate": 5.992295609885689e-06,
      "loss": 1.8809,
      "step": 215720
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.655910849571228,
      "learning_rate": 5.9914821431363616e-06,
      "loss": 2.2451,
      "step": 215721
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.5616788864135742,
      "learning_rate": 5.990668731048742e-06,
      "loss": 2.0366,
      "step": 215722
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.4612125158309937,
      "learning_rate": 5.9898553736230315e-06,
      "loss": 2.0865,
      "step": 215723
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.377814531326294,
      "learning_rate": 5.989042070859329e-06,
      "loss": 1.88,
      "step": 215724
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.4578040838241577,
      "learning_rate": 5.988228822757835e-06,
      "loss": 2.0464,
      "step": 215725
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.387702226638794,
      "learning_rate": 5.987415629318681e-06,
      "loss": 2.0401,
      "step": 215726
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.5116113424301147,
      "learning_rate": 5.986602490542037e-06,
      "loss": 2.0131,
      "step": 215727
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.4760398864746094,
      "learning_rate": 5.985789406427998e-06,
      "loss": 1.7585,
      "step": 215728
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.433489441871643,
      "learning_rate": 5.984976376976769e-06,
      "loss": 1.9275,
      "step": 215729
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.515056848526001,
      "learning_rate": 5.9841634021884464e-06,
      "loss": 1.7713,
      "step": 215730
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.4052064418792725,
      "learning_rate": 5.983350482063265e-06,
      "loss": 1.9642,
      "step": 215731
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.5232760906219482,
      "learning_rate": 5.982537616601291e-06,
      "loss": 1.9627,
      "step": 215732
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.5051844120025635,
      "learning_rate": 5.981724805802757e-06,
      "loss": 1.8946,
      "step": 215733
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.4614965915679932,
      "learning_rate": 5.98091204966773e-06,
      "loss": 2.0047,
      "step": 215734
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.3982449769973755,
      "learning_rate": 5.980099348196443e-06,
      "loss": 2.0407,
      "step": 215735
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.4942522048950195,
      "learning_rate": 5.979286701388963e-06,
      "loss": 2.0118,
      "step": 215736
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.501657247543335,
      "learning_rate": 5.978474109245523e-06,
      "loss": 2.0639,
      "step": 215737
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.4227986335754395,
      "learning_rate": 5.9776615717661904e-06,
      "loss": 1.9921,
      "step": 215738
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.3956992626190186,
      "learning_rate": 5.976849088951196e-06,
      "loss": 2.0372,
      "step": 215739
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.567559003829956,
      "learning_rate": 5.976036660800643e-06,
      "loss": 1.8594,
      "step": 215740
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.6330734491348267,
      "learning_rate": 5.975224287314695e-06,
      "loss": 2.1295,
      "step": 215741
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.4139453172683716,
      "learning_rate": 5.974411968493487e-06,
      "loss": 2.0442,
      "step": 215742
    },
    {
      "epoch": 2.81,
      "grad_norm": 2.040776491165161,
      "learning_rate": 5.973599704337218e-06,
      "loss": 1.7221,
      "step": 215743
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.4390119314193726,
      "learning_rate": 5.972787494845954e-06,
      "loss": 1.8367,
      "step": 215744
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.652919054031372,
      "learning_rate": 5.971975340019996e-06,
      "loss": 2.0751,
      "step": 215745
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.510120153427124,
      "learning_rate": 5.971163239859311e-06,
      "loss": 1.8238,
      "step": 215746
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.9007267951965332,
      "learning_rate": 5.970351194364164e-06,
      "loss": 1.9221,
      "step": 215747
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.5064915418624878,
      "learning_rate": 5.969539203534657e-06,
      "loss": 1.9022,
      "step": 215748
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.4362190961837769,
      "learning_rate": 5.968727267371021e-06,
      "loss": 1.9069,
      "step": 215749
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.4093769788742065,
      "learning_rate": 5.9679153858732896e-06,
      "loss": 1.9308,
      "step": 215750
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.4732449054718018,
      "learning_rate": 5.967103559041731e-06,
      "loss": 1.9865,
      "step": 215751
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.5236488580703735,
      "learning_rate": 5.966291786876376e-06,
      "loss": 2.0726,
      "step": 215752
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.4680945873260498,
      "learning_rate": 5.965480069377459e-06,
      "loss": 1.9773,
      "step": 215753
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.3859058618545532,
      "learning_rate": 5.964668406545115e-06,
      "loss": 1.9808,
      "step": 215754
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.4588375091552734,
      "learning_rate": 5.963856798379507e-06,
      "loss": 2.1795,
      "step": 215755
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.4008204936981201,
      "learning_rate": 5.963045244880738e-06,
      "loss": 1.8201,
      "step": 215756
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.4565707445144653,
      "learning_rate": 5.962233746049006e-06,
      "loss": 1.8543,
      "step": 215757
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.4596458673477173,
      "learning_rate": 5.961422301884444e-06,
      "loss": 2.0849,
      "step": 215758
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.4721179008483887,
      "learning_rate": 5.96061091238722e-06,
      "loss": 2.1031,
      "step": 215759
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.4467341899871826,
      "learning_rate": 5.959799577557434e-06,
      "loss": 2.0005,
      "step": 215760
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.4251667261123657,
      "learning_rate": 5.958988297395284e-06,
      "loss": 1.9049,
      "step": 215761
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.6741172075271606,
      "learning_rate": 5.958177071900905e-06,
      "loss": 1.9819,
      "step": 215762
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.4770524501800537,
      "learning_rate": 5.9573659010744955e-06,
      "loss": 1.9822,
      "step": 215763
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.5599747896194458,
      "learning_rate": 5.95655478491609e-06,
      "loss": 2.0636,
      "step": 215764
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.4637107849121094,
      "learning_rate": 5.955743723425954e-06,
      "loss": 1.9168,
      "step": 215765
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.5270429849624634,
      "learning_rate": 5.954932716604189e-06,
      "loss": 2.1573,
      "step": 215766
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.3714512586593628,
      "learning_rate": 5.954121764450959e-06,
      "loss": 2.0344,
      "step": 215767
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.5008553266525269,
      "learning_rate": 5.953310866966365e-06,
      "loss": 1.8222,
      "step": 215768
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.39460027217865,
      "learning_rate": 5.952500024150675e-06,
      "loss": 2.1618,
      "step": 215769
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.399288535118103,
      "learning_rate": 5.9516892360039205e-06,
      "loss": 2.0434,
      "step": 215770
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.5137128829956055,
      "learning_rate": 5.950878502526302e-06,
      "loss": 2.0365,
      "step": 215771
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.5160410404205322,
      "learning_rate": 5.950067823717953e-06,
      "loss": 1.9393,
      "step": 215772
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.6652084589004517,
      "learning_rate": 5.949257199579072e-06,
      "loss": 2.1211,
      "step": 215773
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.5078175067901611,
      "learning_rate": 5.948446630109727e-06,
      "loss": 2.0499,
      "step": 215774
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.5117567777633667,
      "learning_rate": 5.947636115310151e-06,
      "loss": 2.0473,
      "step": 215775
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.4570034742355347,
      "learning_rate": 5.9468256551804764e-06,
      "loss": 1.8322,
      "step": 215776
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.4262856245040894,
      "learning_rate": 5.946015249720837e-06,
      "loss": 1.9844,
      "step": 215777
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.3969013690948486,
      "learning_rate": 5.945204898931333e-06,
      "loss": 2.1501,
      "step": 215778
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.5391498804092407,
      "learning_rate": 5.944394602812197e-06,
      "loss": 1.8764,
      "step": 215779
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.4770610332489014,
      "learning_rate": 5.943584361363529e-06,
      "loss": 2.1064,
      "step": 215780
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.433308720588684,
      "learning_rate": 5.942774174585529e-06,
      "loss": 1.8927,
      "step": 215781
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.430653691291809,
      "learning_rate": 5.941964042478331e-06,
      "loss": 1.9579,
      "step": 215782
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.6023377180099487,
      "learning_rate": 5.941153965042034e-06,
      "loss": 2.0765,
      "step": 215783
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.3479305505752563,
      "learning_rate": 5.9403439422768375e-06,
      "loss": 1.8012,
      "step": 215784
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.496313452720642,
      "learning_rate": 5.9395339741829086e-06,
      "loss": 2.0056,
      "step": 215785
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.4977229833602905,
      "learning_rate": 5.938724060760314e-06,
      "loss": 1.985,
      "step": 215786
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.3724861145019531,
      "learning_rate": 5.93791420200932e-06,
      "loss": 1.8177,
      "step": 215787
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.586208701133728,
      "learning_rate": 5.937104397929992e-06,
      "loss": 2.0825,
      "step": 215788
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.497584581375122,
      "learning_rate": 5.936294648522499e-06,
      "loss": 2.1008,
      "step": 215789
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.3483043909072876,
      "learning_rate": 5.935484953787039e-06,
      "loss": 1.9663,
      "step": 215790
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.4819989204406738,
      "learning_rate": 5.93467531372368e-06,
      "loss": 1.6522,
      "step": 215791
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.375666618347168,
      "learning_rate": 5.933865728332621e-06,
      "loss": 1.9616,
      "step": 215792
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.5035924911499023,
      "learning_rate": 5.933056197614028e-06,
      "loss": 2.0287,
      "step": 215793
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.4326117038726807,
      "learning_rate": 5.932246721568035e-06,
      "loss": 1.8251,
      "step": 215794
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.4175153970718384,
      "learning_rate": 5.931437300194741e-06,
      "loss": 1.8324,
      "step": 215795
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.4379626512527466,
      "learning_rate": 5.930627933494381e-06,
      "loss": 1.956,
      "step": 215796
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.5193400382995605,
      "learning_rate": 5.929818621467053e-06,
      "loss": 2.1304,
      "step": 215797
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.4779688119888306,
      "learning_rate": 5.929009364112958e-06,
      "loss": 2.0147,
      "step": 215798
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.4379514455795288,
      "learning_rate": 5.928200161432195e-06,
      "loss": 1.9396,
      "step": 215799
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.4985145330429077,
      "learning_rate": 5.9273910134249316e-06,
      "loss": 1.9522,
      "step": 215800
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.5080710649490356,
      "learning_rate": 5.926581920091267e-06,
      "loss": 2.3546,
      "step": 215801
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.5938305854797363,
      "learning_rate": 5.925772881431468e-06,
      "loss": 1.83,
      "step": 215802
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.6812893152236938,
      "learning_rate": 5.924963897445567e-06,
      "loss": 2.255,
      "step": 215803
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.5119469165802002,
      "learning_rate": 5.924154968133798e-06,
      "loss": 1.8169,
      "step": 215804
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.4740532636642456,
      "learning_rate": 5.923346093496262e-06,
      "loss": 1.9575,
      "step": 215805
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.3821632862091064,
      "learning_rate": 5.92253727353319e-06,
      "loss": 1.9906,
      "step": 215806
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.476235270500183,
      "learning_rate": 5.921728508244583e-06,
      "loss": 2.0114,
      "step": 215807
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.6660072803497314,
      "learning_rate": 5.920919797630741e-06,
      "loss": 1.9242,
      "step": 215808
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.5220805406570435,
      "learning_rate": 5.920111141691697e-06,
      "loss": 2.1449,
      "step": 215809
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.4922982454299927,
      "learning_rate": 5.919302540427684e-06,
      "loss": 1.8432,
      "step": 215810
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.4356313943862915,
      "learning_rate": 5.918493993838802e-06,
      "loss": 1.8086,
      "step": 215811
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.4976187944412231,
      "learning_rate": 5.917685501925284e-06,
      "loss": 1.775,
      "step": 215812
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.36505126953125,
      "learning_rate": 5.916877064687165e-06,
      "loss": 2.0711,
      "step": 215813
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.4615545272827148,
      "learning_rate": 5.916068682124675e-06,
      "loss": 1.8768,
      "step": 215814
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.449167013168335,
      "learning_rate": 5.915260354237916e-06,
      "loss": 1.9299,
      "step": 215815
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.5169380903244019,
      "learning_rate": 5.914452081027088e-06,
      "loss": 1.8687,
      "step": 215816
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.4382593631744385,
      "learning_rate": 5.913643862492256e-06,
      "loss": 2.0817,
      "step": 215817
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.5311336517333984,
      "learning_rate": 5.912835698633722e-06,
      "loss": 1.8874,
      "step": 215818
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.4732707738876343,
      "learning_rate": 5.912027589451485e-06,
      "loss": 2.0726,
      "step": 215819
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.6266127824783325,
      "learning_rate": 5.911219534945777e-06,
      "loss": 2.0759,
      "step": 215820
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.5802448987960815,
      "learning_rate": 5.9104115351166655e-06,
      "loss": 2.1646,
      "step": 215821
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.5512306690216064,
      "learning_rate": 5.909603589964418e-06,
      "loss": 1.9315,
      "step": 215822
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.3794957399368286,
      "learning_rate": 5.9087956994890995e-06,
      "loss": 1.8596,
      "step": 215823
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.6511468887329102,
      "learning_rate": 5.907987863690944e-06,
      "loss": 1.9793,
      "step": 215824
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.5865343809127808,
      "learning_rate": 5.907180082569984e-06,
      "loss": 1.9215,
      "step": 215825
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.4572417736053467,
      "learning_rate": 5.906372356126454e-06,
      "loss": 2.1752,
      "step": 215826
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.4484832286834717,
      "learning_rate": 5.905564684360453e-06,
      "loss": 1.9238,
      "step": 215827
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.4046173095703125,
      "learning_rate": 5.904757067272181e-06,
      "loss": 1.8188,
      "step": 215828
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.4664288759231567,
      "learning_rate": 5.903949504861772e-06,
      "loss": 2.1123,
      "step": 215829
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.374666452407837,
      "learning_rate": 5.90314199712939e-06,
      "loss": 1.9123,
      "step": 215830
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.40611732006073,
      "learning_rate": 5.902334544075138e-06,
      "loss": 1.8986,
      "step": 215831
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.6583502292633057,
      "learning_rate": 5.901527145699181e-06,
      "loss": 2.2123,
      "step": 215832
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.488917589187622,
      "learning_rate": 5.900719802001685e-06,
      "loss": 1.8287,
      "step": 215833
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.4032255411148071,
      "learning_rate": 5.899912512982819e-06,
      "loss": 2.0565,
      "step": 215834
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.8150416612625122,
      "learning_rate": 5.899105278642679e-06,
      "loss": 1.9201,
      "step": 215835
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.4056991338729858,
      "learning_rate": 5.898298098981502e-06,
      "loss": 1.8749,
      "step": 215836
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.6008291244506836,
      "learning_rate": 5.897490973999352e-06,
      "loss": 1.9644,
      "step": 215837
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.4878084659576416,
      "learning_rate": 5.896683903696398e-06,
      "loss": 2.0693,
      "step": 215838
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.5024734735488892,
      "learning_rate": 5.895876888072804e-06,
      "loss": 1.9829,
      "step": 215839
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.3289666175842285,
      "learning_rate": 5.8950699271287375e-06,
      "loss": 1.9941,
      "step": 215840
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.4315276145935059,
      "learning_rate": 5.894263020864299e-06,
      "loss": 1.68,
      "step": 215841
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.4137665033340454,
      "learning_rate": 5.89345616927972e-06,
      "loss": 2.0593,
      "step": 215842
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.4799062013626099,
      "learning_rate": 5.892649372375036e-06,
      "loss": 1.8329,
      "step": 215843
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.3877191543579102,
      "learning_rate": 5.891842630150512e-06,
      "loss": 1.7772,
      "step": 215844
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.6103382110595703,
      "learning_rate": 5.891035942606182e-06,
      "loss": 2.0184,
      "step": 215845
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.6916395425796509,
      "learning_rate": 5.890229309742311e-06,
      "loss": 2.0644,
      "step": 215846
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.400497317314148,
      "learning_rate": 5.889422731559001e-06,
      "loss": 2.0242,
      "step": 215847
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.4381442070007324,
      "learning_rate": 5.8886162080564185e-06,
      "loss": 1.8876,
      "step": 215848
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.6003849506378174,
      "learning_rate": 5.887809739234628e-06,
      "loss": 2.0121,
      "step": 215849
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.5221319198608398,
      "learning_rate": 5.887003325093898e-06,
      "loss": 2.099,
      "step": 215850
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.4301141500473022,
      "learning_rate": 5.886196965634293e-06,
      "loss": 2.1006,
      "step": 215851
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.3311171531677246,
      "learning_rate": 5.885390660856015e-06,
      "loss": 1.7981,
      "step": 215852
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.5723979473114014,
      "learning_rate": 5.884584410759163e-06,
      "loss": 1.9239,
      "step": 215853
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.5665048360824585,
      "learning_rate": 5.8837782153440035e-06,
      "loss": 1.9553,
      "step": 215854
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.4720877408981323,
      "learning_rate": 5.882972074610504e-06,
      "loss": 2.0101,
      "step": 215855
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.4834520816802979,
      "learning_rate": 5.882165988558962e-06,
      "loss": 2.1328,
      "step": 215856
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.5291283130645752,
      "learning_rate": 5.881359957189446e-06,
      "loss": 2.0075,
      "step": 215857
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.5412757396697998,
      "learning_rate": 5.880553980502189e-06,
      "loss": 2.0609,
      "step": 215858
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.3572741746902466,
      "learning_rate": 5.879748058497225e-06,
      "loss": 1.696,
      "step": 215859
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.380876064300537,
      "learning_rate": 5.878942191174818e-06,
      "loss": 2.0277,
      "step": 215860
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.3596864938735962,
      "learning_rate": 5.87813637853507e-06,
      "loss": 1.911,
      "step": 215861
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.488584041595459,
      "learning_rate": 5.877330620578114e-06,
      "loss": 1.8877,
      "step": 215862
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.4594453573226929,
      "learning_rate": 5.8765249173040815e-06,
      "loss": 1.9051,
      "step": 215863
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.5288242101669312,
      "learning_rate": 5.875719268713208e-06,
      "loss": 2.1817,
      "step": 215864
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.4757919311523438,
      "learning_rate": 5.8749136748055255e-06,
      "loss": 1.8149,
      "step": 215865
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.5557525157928467,
      "learning_rate": 5.8741081355813015e-06,
      "loss": 2.0766,
      "step": 215866
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.4791128635406494,
      "learning_rate": 5.8733026510406344e-06,
      "loss": 1.8205,
      "step": 215867
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.446703553199768,
      "learning_rate": 5.872497221183659e-06,
      "loss": 2.0533,
      "step": 215868
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.4644020795822144,
      "learning_rate": 5.87169184601054e-06,
      "loss": 2.0448,
      "step": 215869
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.4159282445907593,
      "learning_rate": 5.8708865255214124e-06,
      "loss": 2.0144,
      "step": 215870
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.4303985834121704,
      "learning_rate": 5.8700812597164415e-06,
      "loss": 1.9775,
      "step": 215871
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.4615321159362793,
      "learning_rate": 5.869276048595795e-06,
      "loss": 2.2561,
      "step": 215872
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.3910751342773438,
      "learning_rate": 5.868470892159604e-06,
      "loss": 2.0713,
      "step": 215873
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.6602312326431274,
      "learning_rate": 5.867665790407971e-06,
      "loss": 1.8995,
      "step": 215874
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.3458701372146606,
      "learning_rate": 5.866860743341128e-06,
      "loss": 1.8729,
      "step": 215875
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.740488052368164,
      "learning_rate": 5.866055750959209e-06,
      "loss": 2.046,
      "step": 215876
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.4627071619033813,
      "learning_rate": 5.865250813262279e-06,
      "loss": 1.8803,
      "step": 215877
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.3447588682174683,
      "learning_rate": 5.864445930250605e-06,
      "loss": 1.9746,
      "step": 215878
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.395559549331665,
      "learning_rate": 5.863641101924288e-06,
      "loss": 1.9378,
      "step": 215879
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.4894376993179321,
      "learning_rate": 5.862836328283427e-06,
      "loss": 2.0143,
      "step": 215880
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.4685155153274536,
      "learning_rate": 5.862031609328255e-06,
      "loss": 2.0661,
      "step": 215881
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.54157292842865,
      "learning_rate": 5.861226945058839e-06,
      "loss": 1.9506,
      "step": 215882
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.5520137548446655,
      "learning_rate": 5.860422335475412e-06,
      "loss": 2.0178,
      "step": 215883
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.6855309009552002,
      "learning_rate": 5.859617780578074e-06,
      "loss": 2.0475,
      "step": 215884
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.5106781721115112,
      "learning_rate": 5.858813280366992e-06,
      "loss": 1.9607,
      "step": 215885
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.5368906259536743,
      "learning_rate": 5.858008834842298e-06,
      "loss": 2.149,
      "step": 215886
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.5844672918319702,
      "learning_rate": 5.85720444400416e-06,
      "loss": 2.0946,
      "step": 215887
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.4716769456863403,
      "learning_rate": 5.85640010785271e-06,
      "loss": 1.889,
      "step": 215888
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.3593531847000122,
      "learning_rate": 5.855595826388115e-06,
      "loss": 1.8517,
      "step": 215889
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.502274513244629,
      "learning_rate": 5.854791599610542e-06,
      "loss": 2.0373,
      "step": 215890
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.5762826204299927,
      "learning_rate": 5.8539874275200905e-06,
      "loss": 1.9709,
      "step": 215891
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.539319634437561,
      "learning_rate": 5.853183310116927e-06,
      "loss": 2.0346,
      "step": 215892
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.561462640762329,
      "learning_rate": 5.852379247401217e-06,
      "loss": 1.9804,
      "step": 215893
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.7180918455123901,
      "learning_rate": 5.8515752393730965e-06,
      "loss": 1.9188,
      "step": 215894
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.4669723510742188,
      "learning_rate": 5.8507712860327295e-06,
      "loss": 1.8677,
      "step": 215895
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.5697234869003296,
      "learning_rate": 5.84996738738025e-06,
      "loss": 2.0561,
      "step": 215896
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.520877718925476,
      "learning_rate": 5.849163543415858e-06,
      "loss": 1.9468,
      "step": 215897
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.531846523284912,
      "learning_rate": 5.848359754139587e-06,
      "loss": 2.1178,
      "step": 215898
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.6397840976715088,
      "learning_rate": 5.8475560195517026e-06,
      "loss": 1.9638,
      "step": 215899
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.4204708337783813,
      "learning_rate": 5.846752339652272e-06,
      "loss": 1.8935,
      "step": 215900
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.4100314378738403,
      "learning_rate": 5.845948714441528e-06,
      "loss": 1.9008,
      "step": 215901
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.589542269706726,
      "learning_rate": 5.845145143919538e-06,
      "loss": 1.7204,
      "step": 215902
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.3844027519226074,
      "learning_rate": 5.844341628086535e-06,
      "loss": 1.9366,
      "step": 215903
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.415135145187378,
      "learning_rate": 5.843538166942585e-06,
      "loss": 1.8004,
      "step": 215904
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.5192488431930542,
      "learning_rate": 5.842734760487888e-06,
      "loss": 1.9355,
      "step": 215905
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.364234447479248,
      "learning_rate": 5.841931408722578e-06,
      "loss": 1.8139,
      "step": 215906
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.432086706161499,
      "learning_rate": 5.841128111646787e-06,
      "loss": 1.9606,
      "step": 215907
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.5273469686508179,
      "learning_rate": 5.840324869260682e-06,
      "loss": 2.124,
      "step": 215908
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.4097203016281128,
      "learning_rate": 5.839521681564463e-06,
      "loss": 1.8646,
      "step": 215909
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.5071319341659546,
      "learning_rate": 5.838718548558197e-06,
      "loss": 1.9589,
      "step": 215910
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.4325670003890991,
      "learning_rate": 5.837915470242049e-06,
      "loss": 1.921,
      "step": 215911
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.4545537233352661,
      "learning_rate": 5.837112446616188e-06,
      "loss": 1.7163,
      "step": 215912
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.4959105253219604,
      "learning_rate": 5.836309477680778e-06,
      "loss": 1.9821,
      "step": 215913
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.5892388820648193,
      "learning_rate": 5.835506563435921e-06,
      "loss": 1.9134,
      "step": 215914
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.396772027015686,
      "learning_rate": 5.834703703881882e-06,
      "loss": 2.0987,
      "step": 215915
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.441257119178772,
      "learning_rate": 5.833900899018629e-06,
      "loss": 1.9652,
      "step": 215916
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.4274801015853882,
      "learning_rate": 5.83309814884646e-06,
      "loss": 1.4766,
      "step": 215917
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.5676321983337402,
      "learning_rate": 5.8322954533654434e-06,
      "loss": 1.8918,
      "step": 215918
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.3312932252883911,
      "learning_rate": 5.831492812575744e-06,
      "loss": 1.967,
      "step": 215919
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.443684697151184,
      "learning_rate": 5.83069022647753e-06,
      "loss": 2.034,
      "step": 215920
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.4731862545013428,
      "learning_rate": 5.829887695071001e-06,
      "loss": 1.7918,
      "step": 215921
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.5714927911758423,
      "learning_rate": 5.8290852183561885e-06,
      "loss": 1.9892,
      "step": 215922
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.5272293090820312,
      "learning_rate": 5.828282796333328e-06,
      "loss": 2.21,
      "step": 215923
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.489447832107544,
      "learning_rate": 5.827480429002518e-06,
      "loss": 2.0269,
      "step": 215924
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.5627166032791138,
      "learning_rate": 5.8266781163639585e-06,
      "loss": 1.9336,
      "step": 215925
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.5708098411560059,
      "learning_rate": 5.82587585841775e-06,
      "loss": 2.0336,
      "step": 215926
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.3938151597976685,
      "learning_rate": 5.825073655164059e-06,
      "loss": 1.7665,
      "step": 215927
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.5828875303268433,
      "learning_rate": 5.824271506603084e-06,
      "loss": 2.1274,
      "step": 215928
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.437940001487732,
      "learning_rate": 5.823469412734893e-06,
      "loss": 1.8684,
      "step": 215929
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.455317735671997,
      "learning_rate": 5.822667373559653e-06,
      "loss": 2.0103,
      "step": 215930
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.6510634422302246,
      "learning_rate": 5.821865389077596e-06,
      "loss": 1.9377,
      "step": 215931
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.5664100646972656,
      "learning_rate": 5.821063459288755e-06,
      "loss": 1.7926,
      "step": 215932
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.5547910928726196,
      "learning_rate": 5.820261584193331e-06,
      "loss": 1.6324,
      "step": 215933
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.5839065313339233,
      "learning_rate": 5.819459763791523e-06,
      "loss": 1.8517,
      "step": 215934
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.4562623500823975,
      "learning_rate": 5.818657998083398e-06,
      "loss": 1.892,
      "step": 215935
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.4925205707550049,
      "learning_rate": 5.817856287069122e-06,
      "loss": 2.0238,
      "step": 215936
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.397260069847107,
      "learning_rate": 5.817054630748863e-06,
      "loss": 1.9938,
      "step": 215937
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.6516987085342407,
      "learning_rate": 5.816253029122786e-06,
      "loss": 2.1459,
      "step": 215938
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.524402141571045,
      "learning_rate": 5.815451482191025e-06,
      "loss": 1.808,
      "step": 215939
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.4326807260513306,
      "learning_rate": 5.814649989953713e-06,
      "loss": 1.8889,
      "step": 215940
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.8946411609649658,
      "learning_rate": 5.813848552411016e-06,
      "loss": 2.0779,
      "step": 215941
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.4947261810302734,
      "learning_rate": 5.813047169563067e-06,
      "loss": 1.8527,
      "step": 215942
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.4935675859451294,
      "learning_rate": 5.812245841410035e-06,
      "loss": 2.003,
      "step": 215943
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.313617467880249,
      "learning_rate": 5.811444567952017e-06,
      "loss": 1.9392,
      "step": 215944
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.5787324905395508,
      "learning_rate": 5.810643349189281e-06,
      "loss": 2.1391,
      "step": 215945
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.6281481981277466,
      "learning_rate": 5.80984218512186e-06,
      "loss": 2.1541,
      "step": 215946
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.4356200695037842,
      "learning_rate": 5.809041075749954e-06,
      "loss": 1.8645,
      "step": 215947
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.725268006324768,
      "learning_rate": 5.808240021073662e-06,
      "loss": 1.9316,
      "step": 215948
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.4992547035217285,
      "learning_rate": 5.807439021093219e-06,
      "loss": 2.0324,
      "step": 215949
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.4104738235473633,
      "learning_rate": 5.806638075808689e-06,
      "loss": 2.0568,
      "step": 215950
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.5481879711151123,
      "learning_rate": 5.805837185220274e-06,
      "loss": 2.0701,
      "step": 215951
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.5121495723724365,
      "learning_rate": 5.8050363493281405e-06,
      "loss": 1.9601,
      "step": 215952
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.359313726425171,
      "learning_rate": 5.804235568132387e-06,
      "loss": 1.7714,
      "step": 215953
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.5152291059494019,
      "learning_rate": 5.803434841633148e-06,
      "loss": 1.9658,
      "step": 215954
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.4198524951934814,
      "learning_rate": 5.8026341698306224e-06,
      "loss": 2.0359,
      "step": 215955
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.4311892986297607,
      "learning_rate": 5.80183355272491e-06,
      "loss": 1.9144,
      "step": 215956
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.5264531373977661,
      "learning_rate": 5.801032990316245e-06,
      "loss": 1.9804,
      "step": 215957
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.6049352884292603,
      "learning_rate": 5.800232482604694e-06,
      "loss": 2.0434,
      "step": 215958
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.4348795413970947,
      "learning_rate": 5.7994320295904226e-06,
      "loss": 1.7747,
      "step": 215959
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.4323432445526123,
      "learning_rate": 5.798631631273598e-06,
      "loss": 2.1053,
      "step": 215960
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.5080982446670532,
      "learning_rate": 5.797831287654387e-06,
      "loss": 2.0289,
      "step": 215961
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.5174214839935303,
      "learning_rate": 5.797030998732888e-06,
      "loss": 1.9228,
      "step": 215962
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.4003524780273438,
      "learning_rate": 5.796230764509269e-06,
      "loss": 1.9344,
      "step": 215963
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.519149661064148,
      "learning_rate": 5.795430584983729e-06,
      "loss": 1.9914,
      "step": 215964
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.4774364233016968,
      "learning_rate": 5.794630460156302e-06,
      "loss": 2.0802,
      "step": 215965
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.4184269905090332,
      "learning_rate": 5.793830390027254e-06,
      "loss": 1.7687,
      "step": 215966
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.486991286277771,
      "learning_rate": 5.793030374596652e-06,
      "loss": 2.033,
      "step": 215967
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.560907006263733,
      "learning_rate": 5.7922304138647295e-06,
      "loss": 2.0342,
      "step": 215968
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.5763427019119263,
      "learning_rate": 5.791430507831552e-06,
      "loss": 2.0205,
      "step": 215969
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.4823986291885376,
      "learning_rate": 5.79063065649732e-06,
      "loss": 2.0071,
      "step": 215970
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.4126033782958984,
      "learning_rate": 5.7898308598621324e-06,
      "loss": 1.8134,
      "step": 215971
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.5933709144592285,
      "learning_rate": 5.789031117926191e-06,
      "loss": 2.0485,
      "step": 215972
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.4254509210586548,
      "learning_rate": 5.788231430689594e-06,
      "loss": 1.8849,
      "step": 215973
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.4348399639129639,
      "learning_rate": 5.787431798152575e-06,
      "loss": 1.7064,
      "step": 215974
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.6326788663864136,
      "learning_rate": 5.7866322203152015e-06,
      "loss": 2.0803,
      "step": 215975
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.4551206827163696,
      "learning_rate": 5.785832697177639e-06,
      "loss": 2.1123,
      "step": 215976
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.5535088777542114,
      "learning_rate": 5.785033228740022e-06,
      "loss": 1.9085,
      "step": 215977
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.6300830841064453,
      "learning_rate": 5.784233815002548e-06,
      "loss": 2.0185,
      "step": 215978
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.4482641220092773,
      "learning_rate": 5.783434455965319e-06,
      "loss": 2.0508,
      "step": 215979
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.4709638357162476,
      "learning_rate": 5.7826351516285345e-06,
      "loss": 2.023,
      "step": 215980
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.4438210725784302,
      "learning_rate": 5.78183590199226e-06,
      "loss": 2.1002,
      "step": 215981
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.3490487337112427,
      "learning_rate": 5.7810367070567636e-06,
      "loss": 1.7738,
      "step": 215982
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.4843205213546753,
      "learning_rate": 5.780237566822077e-06,
      "loss": 1.921,
      "step": 215983
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.410562515258789,
      "learning_rate": 5.7794384812884344e-06,
      "loss": 1.8172,
      "step": 215984
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.5535207986831665,
      "learning_rate": 5.778639450455902e-06,
      "loss": 1.7561,
      "step": 215985
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.5241930484771729,
      "learning_rate": 5.777840474324713e-06,
      "loss": 2.0501,
      "step": 215986
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.6409919261932373,
      "learning_rate": 5.777041552894967e-06,
      "loss": 2.1164,
      "step": 215987
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.508981466293335,
      "learning_rate": 5.776242686166832e-06,
      "loss": 1.77,
      "step": 215988
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.576804280281067,
      "learning_rate": 5.775443874140439e-06,
      "loss": 1.8456,
      "step": 215989
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.5634706020355225,
      "learning_rate": 5.7746451168159565e-06,
      "loss": 1.9335,
      "step": 215990
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.538018822669983,
      "learning_rate": 5.773846414193484e-06,
      "loss": 1.7508,
      "step": 215991
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.350821614265442,
      "learning_rate": 5.773047766273253e-06,
      "loss": 1.9789,
      "step": 215992
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.4324198961257935,
      "learning_rate": 5.7722491730552985e-06,
      "loss": 1.9122,
      "step": 215993
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.4777730703353882,
      "learning_rate": 5.77145063453992e-06,
      "loss": 2.0949,
      "step": 215994
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.393157958984375,
      "learning_rate": 5.770652150727151e-06,
      "loss": 1.8562,
      "step": 215995
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.3300879001617432,
      "learning_rate": 5.769853721617157e-06,
      "loss": 1.9308,
      "step": 215996
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.576696515083313,
      "learning_rate": 5.769055347210105e-06,
      "loss": 1.9632,
      "step": 215997
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.4611760377883911,
      "learning_rate": 5.768257027506163e-06,
      "loss": 1.8056,
      "step": 215998
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.5284467935562134,
      "learning_rate": 5.7674587625053944e-06,
      "loss": 2.2338,
      "step": 215999
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.6090712547302246,
      "learning_rate": 5.766660552208069e-06,
      "loss": 2.1687,
      "step": 216000
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.4872326850891113,
      "learning_rate": 5.765862396614285e-06,
      "loss": 2.1855,
      "step": 216001
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.4930790662765503,
      "learning_rate": 5.765064295724142e-06,
      "loss": 2.1015,
      "step": 216002
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.6256250143051147,
      "learning_rate": 5.764266249537841e-06,
      "loss": 1.9713,
      "step": 216003
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.5684469938278198,
      "learning_rate": 5.763468258055514e-06,
      "loss": 1.8765,
      "step": 216004
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.4439414739608765,
      "learning_rate": 5.762670321277296e-06,
      "loss": 1.8839,
      "step": 216005
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.438879370689392,
      "learning_rate": 5.761872439203386e-06,
      "loss": 1.9046,
      "step": 216006
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.5913665294647217,
      "learning_rate": 5.761074611833882e-06,
      "loss": 2.0313,
      "step": 216007
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.4375735521316528,
      "learning_rate": 5.7602768391689865e-06,
      "loss": 2.0021,
      "step": 216008
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.4989086389541626,
      "learning_rate": 5.759479121208732e-06,
      "loss": 1.96,
      "step": 216009
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.4604177474975586,
      "learning_rate": 5.758681457953418e-06,
      "loss": 2.0236,
      "step": 216010
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.3524364233016968,
      "learning_rate": 5.757883849403044e-06,
      "loss": 1.8133,
      "step": 216011
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.5486040115356445,
      "learning_rate": 5.757086295557911e-06,
      "loss": 2.012,
      "step": 216012
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.6539099216461182,
      "learning_rate": 5.756288796418051e-06,
      "loss": 2.0235,
      "step": 216013
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.443250298500061,
      "learning_rate": 5.755491351983665e-06,
      "loss": 1.7815,
      "step": 216014
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.3576364517211914,
      "learning_rate": 5.754693962254853e-06,
      "loss": 2.0726,
      "step": 216015
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.5544910430908203,
      "learning_rate": 5.753896627231846e-06,
      "loss": 2.0905,
      "step": 216016
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.422580599784851,
      "learning_rate": 5.753099346914713e-06,
      "loss": 2.2278,
      "step": 216017
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.392380952835083,
      "learning_rate": 5.75230212130362e-06,
      "loss": 2.0168,
      "step": 216018
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.3747284412384033,
      "learning_rate": 5.7515049503987665e-06,
      "loss": 1.9259,
      "step": 216019
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.4780387878417969,
      "learning_rate": 5.750707834200252e-06,
      "loss": 2.0978,
      "step": 216020
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.500210165977478,
      "learning_rate": 5.7499107727082105e-06,
      "loss": 1.789,
      "step": 216021
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.4528487920761108,
      "learning_rate": 5.749113765922841e-06,
      "loss": 1.7764,
      "step": 216022
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.4861114025115967,
      "learning_rate": 5.748316813844245e-06,
      "loss": 1.8841,
      "step": 216023
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.4555907249450684,
      "learning_rate": 5.74751991647262e-06,
      "loss": 1.9148,
      "step": 216024
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.618857741355896,
      "learning_rate": 5.746723073808068e-06,
      "loss": 1.9678,
      "step": 216025
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.5554934740066528,
      "learning_rate": 5.7459262858507886e-06,
      "loss": 1.8601,
      "step": 216026
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.37851881980896,
      "learning_rate": 5.745129552600847e-06,
      "loss": 1.8342,
      "step": 216027
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.59494948387146,
      "learning_rate": 5.744332874058443e-06,
      "loss": 2.1223,
      "step": 216028
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.5052006244659424,
      "learning_rate": 5.743536250223746e-06,
      "loss": 1.9404,
      "step": 216029
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.4550607204437256,
      "learning_rate": 5.742739681096853e-06,
      "loss": 1.8388,
      "step": 216030
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.4097750186920166,
      "learning_rate": 5.7419431666779645e-06,
      "loss": 1.9239,
      "step": 216031
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.3031834363937378,
      "learning_rate": 5.741146706967215e-06,
      "loss": 1.8877,
      "step": 216032
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.5705456733703613,
      "learning_rate": 5.740350301964702e-06,
      "loss": 1.922,
      "step": 216033
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.4118387699127197,
      "learning_rate": 5.739553951670628e-06,
      "loss": 1.9239,
      "step": 216034
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.4005600214004517,
      "learning_rate": 5.738757656085125e-06,
      "loss": 1.987,
      "step": 216035
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.4183436632156372,
      "learning_rate": 5.7379614152083585e-06,
      "loss": 2.1781,
      "step": 216036
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.3623197078704834,
      "learning_rate": 5.7371652290404635e-06,
      "loss": 1.9534,
      "step": 216037
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.4202849864959717,
      "learning_rate": 5.736369097581573e-06,
      "loss": 2.0256,
      "step": 216038
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.6157879829406738,
      "learning_rate": 5.735573020831818e-06,
      "loss": 2.0145,
      "step": 216039
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.4389723539352417,
      "learning_rate": 5.734776998791402e-06,
      "loss": 1.8811,
      "step": 216040
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.449094295501709,
      "learning_rate": 5.733981031460422e-06,
      "loss": 2.0584,
      "step": 216041
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.4736132621765137,
      "learning_rate": 5.7331851188390785e-06,
      "loss": 2.102,
      "step": 216042
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.4868520498275757,
      "learning_rate": 5.732389260927506e-06,
      "loss": 1.8665,
      "step": 216043
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.3619855642318726,
      "learning_rate": 5.73159345772577e-06,
      "loss": 1.928,
      "step": 216044
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.4674949645996094,
      "learning_rate": 5.730797709234136e-06,
      "loss": 1.8314,
      "step": 216045
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.3986294269561768,
      "learning_rate": 5.730002015452706e-06,
      "loss": 2.101,
      "step": 216046
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.5311291217803955,
      "learning_rate": 5.7292063763816125e-06,
      "loss": 2.008,
      "step": 216047
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.560320496559143,
      "learning_rate": 5.728410792021021e-06,
      "loss": 2.1137,
      "step": 216048
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.4007141590118408,
      "learning_rate": 5.727615262371066e-06,
      "loss": 1.7805,
      "step": 216049
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.3887213468551636,
      "learning_rate": 5.72681978743188e-06,
      "loss": 1.9885,
      "step": 216050
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.5105468034744263,
      "learning_rate": 5.726024367203663e-06,
      "loss": 2.0027,
      "step": 216051
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.479479432106018,
      "learning_rate": 5.7252290016865154e-06,
      "loss": 2.0896,
      "step": 216052
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.5439119338989258,
      "learning_rate": 5.724433690880603e-06,
      "loss": 2.1367,
      "step": 216053
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.4583994150161743,
      "learning_rate": 5.723638434786093e-06,
      "loss": 1.8502,
      "step": 216054
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.4219906330108643,
      "learning_rate": 5.722843233403118e-06,
      "loss": 1.8182,
      "step": 216055
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.4530233144760132,
      "learning_rate": 5.722048086731779e-06,
      "loss": 2.0457,
      "step": 216056
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.4904402494430542,
      "learning_rate": 5.721252994772307e-06,
      "loss": 2.089,
      "step": 216057
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.4333263635635376,
      "learning_rate": 5.720457957524771e-06,
      "loss": 2.0094,
      "step": 216058
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.5271764993667603,
      "learning_rate": 5.719662974989403e-06,
      "loss": 2.023,
      "step": 216059
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.4550480842590332,
      "learning_rate": 5.718868047166269e-06,
      "loss": 1.9165,
      "step": 216060
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.7791014909744263,
      "learning_rate": 5.718073174055537e-06,
      "loss": 2.047,
      "step": 216061
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.430009126663208,
      "learning_rate": 5.71727835565744e-06,
      "loss": 2.0092,
      "step": 216062
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.538466215133667,
      "learning_rate": 5.71648359197201e-06,
      "loss": 2.0564,
      "step": 216063
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.448620319366455,
      "learning_rate": 5.7156888829994145e-06,
      "loss": 2.063,
      "step": 216064
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.646422266960144,
      "learning_rate": 5.714894228739886e-06,
      "loss": 2.083,
      "step": 216065
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.5006614923477173,
      "learning_rate": 5.714099629193458e-06,
      "loss": 2.1522,
      "step": 216066
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.4841495752334595,
      "learning_rate": 5.7133050843603645e-06,
      "loss": 2.0637,
      "step": 216067
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.3712170124053955,
      "learning_rate": 5.712510594240738e-06,
      "loss": 2.0549,
      "step": 216068
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.3928935527801514,
      "learning_rate": 5.711716158834678e-06,
      "loss": 1.9525,
      "step": 216069
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.5691308975219727,
      "learning_rate": 5.710921778142385e-06,
      "loss": 1.8963,
      "step": 216070
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.535358190536499,
      "learning_rate": 5.710127452163993e-06,
      "loss": 1.937,
      "step": 216071
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.3467857837677002,
      "learning_rate": 5.7093331808996e-06,
      "loss": 2.1388,
      "step": 216072
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.393924593925476,
      "learning_rate": 5.708538964349441e-06,
      "loss": 1.9865,
      "step": 216073
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.5757653713226318,
      "learning_rate": 5.707744802513614e-06,
      "loss": 2.0414,
      "step": 216074
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.5114115476608276,
      "learning_rate": 5.706950695392287e-06,
      "loss": 1.9138,
      "step": 216075
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.5496015548706055,
      "learning_rate": 5.70615664298556e-06,
      "loss": 2.0462,
      "step": 216076
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.4051858186721802,
      "learning_rate": 5.705362645293632e-06,
      "loss": 2.0924,
      "step": 216077
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.5942193269729614,
      "learning_rate": 5.704568702316603e-06,
      "loss": 2.111,
      "step": 216078
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.414622187614441,
      "learning_rate": 5.703774814054673e-06,
      "loss": 2.0498,
      "step": 216079
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.5771719217300415,
      "learning_rate": 5.702980980507976e-06,
      "loss": 1.939,
      "step": 216080
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.539180040359497,
      "learning_rate": 5.7021872016766445e-06,
      "loss": 1.9602,
      "step": 216081
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.4035121202468872,
      "learning_rate": 5.701393477560811e-06,
      "loss": 1.8859,
      "step": 216082
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.4248676300048828,
      "learning_rate": 5.700599808160677e-06,
      "loss": 2.1008,
      "step": 216083
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.3693056106567383,
      "learning_rate": 5.699806193476309e-06,
      "loss": 1.9454,
      "step": 216084
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.3758805990219116,
      "learning_rate": 5.699012633507938e-06,
      "loss": 2.0022,
      "step": 216085
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.3092479705810547,
      "learning_rate": 5.698219128255699e-06,
      "loss": 2.2365,
      "step": 216086
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.5579088926315308,
      "learning_rate": 5.6974256777196914e-06,
      "loss": 1.8678,
      "step": 216087
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.451802372932434,
      "learning_rate": 5.696632281900082e-06,
      "loss": 1.8117,
      "step": 216088
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.3291687965393066,
      "learning_rate": 5.695838940797038e-06,
      "loss": 1.9031,
      "step": 216089
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.4513933658599854,
      "learning_rate": 5.695045654410657e-06,
      "loss": 1.9124,
      "step": 216090
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.4017099142074585,
      "learning_rate": 5.694252422741141e-06,
      "loss": 2.0927,
      "step": 216091
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.5687823295593262,
      "learning_rate": 5.693459245788656e-06,
      "loss": 2.107,
      "step": 216092
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.6651118993759155,
      "learning_rate": 5.692666123553269e-06,
      "loss": 2.0294,
      "step": 216093
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.516167163848877,
      "learning_rate": 5.691873056035179e-06,
      "loss": 1.7974,
      "step": 216094
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.4682461023330688,
      "learning_rate": 5.691080043234519e-06,
      "loss": 2.0648,
      "step": 216095
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.4135881662368774,
      "learning_rate": 5.690287085151457e-06,
      "loss": 2.0896,
      "step": 216096
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.549709439277649,
      "learning_rate": 5.689494181786125e-06,
      "loss": 1.9999,
      "step": 216097
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.5853347778320312,
      "learning_rate": 5.68870133313869e-06,
      "loss": 1.84,
      "step": 216098
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.474186658859253,
      "learning_rate": 5.687908539209252e-06,
      "loss": 1.9873,
      "step": 216099
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.6947306394577026,
      "learning_rate": 5.687115799997976e-06,
      "loss": 1.882,
      "step": 216100
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.4971197843551636,
      "learning_rate": 5.686323115505065e-06,
      "loss": 1.8716,
      "step": 216101
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.583540439605713,
      "learning_rate": 5.685530485730582e-06,
      "loss": 1.9328,
      "step": 216102
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.606713891029358,
      "learning_rate": 5.68473791067473e-06,
      "loss": 1.9713,
      "step": 216103
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.498637080192566,
      "learning_rate": 5.6839453903376405e-06,
      "loss": 2.1805,
      "step": 216104
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.4873652458190918,
      "learning_rate": 5.683152924719481e-06,
      "loss": 2.1707,
      "step": 216105
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.6061161756515503,
      "learning_rate": 5.68236051382035e-06,
      "loss": 2.1491,
      "step": 216106
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.4259823560714722,
      "learning_rate": 5.681568157640448e-06,
      "loss": 1.8098,
      "step": 216107
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.4854403734207153,
      "learning_rate": 5.6807758561798424e-06,
      "loss": 2.286,
      "step": 216108
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.4858962297439575,
      "learning_rate": 5.679983609438799e-06,
      "loss": 2.0569,
      "step": 216109
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.4077212810516357,
      "learning_rate": 5.679191417417383e-06,
      "loss": 2.1323,
      "step": 216110
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.7076588869094849,
      "learning_rate": 5.6783992801157645e-06,
      "loss": 1.845,
      "step": 216111
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.4431043863296509,
      "learning_rate": 5.677607197534073e-06,
      "loss": 1.9317,
      "step": 216112
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.6364054679870605,
      "learning_rate": 5.676815169672477e-06,
      "loss": 2.2999,
      "step": 216113
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.3158481121063232,
      "learning_rate": 5.67602319653111e-06,
      "loss": 1.9058,
      "step": 216114
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.4229528903961182,
      "learning_rate": 5.675231278110138e-06,
      "loss": 1.9888,
      "step": 216115
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.407912015914917,
      "learning_rate": 5.674439414409726e-06,
      "loss": 1.8358,
      "step": 216116
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.4545797109603882,
      "learning_rate": 5.673647605429943e-06,
      "loss": 2.0973,
      "step": 216117
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.4081398248672485,
      "learning_rate": 5.6728558511709876e-06,
      "loss": 1.859,
      "step": 216118
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.4535248279571533,
      "learning_rate": 5.672064151633027e-06,
      "loss": 2.06,
      "step": 216119
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.532770037651062,
      "learning_rate": 5.6712725068161605e-06,
      "loss": 2.0306,
      "step": 216120
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.471908450126648,
      "learning_rate": 5.670480916720588e-06,
      "loss": 1.9267,
      "step": 216121
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.5660548210144043,
      "learning_rate": 5.669689381346443e-06,
      "loss": 1.8703,
      "step": 216122
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.3649232387542725,
      "learning_rate": 5.6688979006938255e-06,
      "loss": 2.0319,
      "step": 216123
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.56902015209198,
      "learning_rate": 5.668106474762901e-06,
      "loss": 2.0416,
      "step": 216124
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.4572618007659912,
      "learning_rate": 5.667315103553871e-06,
      "loss": 1.9792,
      "step": 216125
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.449964165687561,
      "learning_rate": 5.666523787066801e-06,
      "loss": 2.1048,
      "step": 216126
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.2738734483718872,
      "learning_rate": 5.665732525301925e-06,
      "loss": 1.6565,
      "step": 216127
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.3643614053726196,
      "learning_rate": 5.664941318259308e-06,
      "loss": 2.0732,
      "step": 216128
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.3917211294174194,
      "learning_rate": 5.664150165939152e-06,
      "loss": 1.9876,
      "step": 216129
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.3564684391021729,
      "learning_rate": 5.663359068341589e-06,
      "loss": 2.069,
      "step": 216130
    },
    {
      "epoch": 2.81,
      "grad_norm": 2.4425008296966553,
      "learning_rate": 5.662568025466785e-06,
      "loss": 2.0177,
      "step": 216131
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.4446674585342407,
      "learning_rate": 5.661777037314808e-06,
      "loss": 2.0918,
      "step": 216132
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.5104703903198242,
      "learning_rate": 5.6609861038859225e-06,
      "loss": 1.9925,
      "step": 216133
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.4758561849594116,
      "learning_rate": 5.660195225180164e-06,
      "loss": 2.3158,
      "step": 216134
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.7349655628204346,
      "learning_rate": 5.659404401197765e-06,
      "loss": 2.0977,
      "step": 216135
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.4737751483917236,
      "learning_rate": 5.6586136319388246e-06,
      "loss": 2.0029,
      "step": 216136
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.5187780857086182,
      "learning_rate": 5.65782291740351e-06,
      "loss": 2.1149,
      "step": 216137
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.4515366554260254,
      "learning_rate": 5.657032257591954e-06,
      "loss": 2.0404,
      "step": 216138
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.4539556503295898,
      "learning_rate": 5.656241652504323e-06,
      "loss": 1.9121,
      "step": 216139
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.4587522745132446,
      "learning_rate": 5.655451102140751e-06,
      "loss": 2.0148,
      "step": 216140
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.5436152219772339,
      "learning_rate": 5.654660606501371e-06,
      "loss": 2.2713,
      "step": 216141
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.5754104852676392,
      "learning_rate": 5.6538701655863826e-06,
      "loss": 1.8329,
      "step": 216142
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.565746784210205,
      "learning_rate": 5.653079779395853e-06,
      "loss": 1.9813,
      "step": 216143
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.4514062404632568,
      "learning_rate": 5.652289447930014e-06,
      "loss": 2.2641,
      "step": 216144
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.6131540536880493,
      "learning_rate": 5.651499171188934e-06,
      "loss": 2.0676,
      "step": 216145
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.4314504861831665,
      "learning_rate": 5.650708949172811e-06,
      "loss": 1.9927,
      "step": 216146
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.4888055324554443,
      "learning_rate": 5.649918781881779e-06,
      "loss": 1.9699,
      "step": 216147
    },
    {
      "epoch": 2.81,
      "grad_norm": 3.398447275161743,
      "learning_rate": 5.6491286693160055e-06,
      "loss": 1.8295,
      "step": 216148
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.5246307849884033,
      "learning_rate": 5.648338611475556e-06,
      "loss": 1.8955,
      "step": 216149
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.4542081356048584,
      "learning_rate": 5.647548608360697e-06,
      "loss": 1.8353,
      "step": 216150
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.384346604347229,
      "learning_rate": 5.646758659971495e-06,
      "loss": 1.9047,
      "step": 216151
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.5196330547332764,
      "learning_rate": 5.645968766308085e-06,
      "loss": 1.8943,
      "step": 216152
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.325174331665039,
      "learning_rate": 5.645178927370697e-06,
      "loss": 1.9557,
      "step": 216153
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.3434021472930908,
      "learning_rate": 5.6443891431594e-06,
      "loss": 1.7493,
      "step": 216154
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.3936741352081299,
      "learning_rate": 5.64359941367436e-06,
      "loss": 2.0351,
      "step": 216155
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.5991833209991455,
      "learning_rate": 5.642809738915743e-06,
      "loss": 1.9889,
      "step": 216156
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.646882176399231,
      "learning_rate": 5.642020118883683e-06,
      "loss": 1.9056,
      "step": 216157
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.4358422756195068,
      "learning_rate": 5.641230553578313e-06,
      "loss": 1.9593,
      "step": 216158
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.6141583919525146,
      "learning_rate": 5.6404410429998324e-06,
      "loss": 2.1362,
      "step": 216159
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.4285249710083008,
      "learning_rate": 5.639651587148342e-06,
      "loss": 1.9615,
      "step": 216160
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.5047473907470703,
      "learning_rate": 5.638862186023974e-06,
      "loss": 2.0739,
      "step": 216161
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.3807809352874756,
      "learning_rate": 5.638072839626895e-06,
      "loss": 1.8927,
      "step": 216162
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.5145018100738525,
      "learning_rate": 5.637283547957272e-06,
      "loss": 2.0612,
      "step": 216163
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.5659443140029907,
      "learning_rate": 5.636494311015205e-06,
      "loss": 2.1727,
      "step": 216164
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.5364761352539062,
      "learning_rate": 5.635705128800927e-06,
      "loss": 2.1026,
      "step": 216165
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.4905163049697876,
      "learning_rate": 5.634916001314504e-06,
      "loss": 2.0732,
      "step": 216166
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.3868228197097778,
      "learning_rate": 5.634126928556071e-06,
      "loss": 1.7755,
      "step": 216167
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.4652745723724365,
      "learning_rate": 5.633337910525859e-06,
      "loss": 1.9077,
      "step": 216168
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.48085618019104,
      "learning_rate": 5.632548947223903e-06,
      "loss": 2.1524,
      "step": 216169
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.7075893878936768,
      "learning_rate": 5.631760038650468e-06,
      "loss": 1.8213,
      "step": 216170
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.7023528814315796,
      "learning_rate": 5.630971184805655e-06,
      "loss": 2.0154,
      "step": 216171
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.426324486732483,
      "learning_rate": 5.6301823856895634e-06,
      "loss": 1.8878,
      "step": 216172
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.6035711765289307,
      "learning_rate": 5.629393641302393e-06,
      "loss": 1.9447,
      "step": 216173
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.4193310737609863,
      "learning_rate": 5.6286049516442775e-06,
      "loss": 2.0624,
      "step": 216174
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.4334444999694824,
      "learning_rate": 5.62781631671535e-06,
      "loss": 2.1757,
      "step": 216175
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.470827579498291,
      "learning_rate": 5.6270277365157765e-06,
      "loss": 2.2042,
      "step": 216176
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.469054937362671,
      "learning_rate": 5.626239211045691e-06,
      "loss": 2.0359,
      "step": 216177
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.705080270767212,
      "learning_rate": 5.62545074030526e-06,
      "loss": 1.9831,
      "step": 216178
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.456982135772705,
      "learning_rate": 5.624662324294582e-06,
      "loss": 2.2007,
      "step": 216179
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.502314567565918,
      "learning_rate": 5.6238739630138585e-06,
      "loss": 2.0684,
      "step": 216180
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.5270578861236572,
      "learning_rate": 5.623085656463222e-06,
      "loss": 2.0577,
      "step": 216181
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.4447219371795654,
      "learning_rate": 5.622297404642806e-06,
      "loss": 1.9868,
      "step": 216182
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.35236656665802,
      "learning_rate": 5.6215092075527436e-06,
      "loss": 1.8859,
      "step": 216183
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.5639383792877197,
      "learning_rate": 5.620721065193234e-06,
      "loss": 1.8988,
      "step": 216184
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.5951955318450928,
      "learning_rate": 5.619932977564346e-06,
      "loss": 2.0069,
      "step": 216185
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.4316084384918213,
      "learning_rate": 5.61914494466631e-06,
      "loss": 2.0891,
      "step": 216186
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.8138699531555176,
      "learning_rate": 5.618356966499193e-06,
      "loss": 2.0359,
      "step": 216187
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.3554712533950806,
      "learning_rate": 5.617569043063197e-06,
      "loss": 2.0573,
      "step": 216188
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.4515308141708374,
      "learning_rate": 5.616781174358453e-06,
      "loss": 2.0472,
      "step": 216189
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.5325103998184204,
      "learning_rate": 5.6159933603851295e-06,
      "loss": 1.9897,
      "step": 216190
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.5345011949539185,
      "learning_rate": 5.615205601143291e-06,
      "loss": 2.146,
      "step": 216191
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.537192940711975,
      "learning_rate": 5.614417896633205e-06,
      "loss": 1.9154,
      "step": 216192
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.459871768951416,
      "learning_rate": 5.613630246854906e-06,
      "loss": 1.8756,
      "step": 216193
    },
    {
      "epoch": 2.81,
      "grad_norm": 1.4031620025634766,
      "learning_rate": 5.612842651808625e-06,
      "loss": 1.8459,
      "step": 216194
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.4609891176223755,
      "learning_rate": 5.612055111494429e-06,
      "loss": 1.8769,
      "step": 216195
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.4543585777282715,
      "learning_rate": 5.611267625912585e-06,
      "loss": 2.1594,
      "step": 216196
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.4128086566925049,
      "learning_rate": 5.610480195063094e-06,
      "loss": 1.9336,
      "step": 216197
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.4397274255752563,
      "learning_rate": 5.609692818946221e-06,
      "loss": 2.1206,
      "step": 216198
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.5037914514541626,
      "learning_rate": 5.608905497562033e-06,
      "loss": 1.8388,
      "step": 216199
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.4670460224151611,
      "learning_rate": 5.60811823091073e-06,
      "loss": 2.0346,
      "step": 216200
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.4579938650131226,
      "learning_rate": 5.607331018992378e-06,
      "loss": 1.983,
      "step": 216201
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.3973474502563477,
      "learning_rate": 5.606543861807278e-06,
      "loss": 1.9534,
      "step": 216202
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.3436013460159302,
      "learning_rate": 5.605756759355396e-06,
      "loss": 2.0175,
      "step": 216203
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.5987133979797363,
      "learning_rate": 5.604969711636997e-06,
      "loss": 1.9113,
      "step": 216204
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.3925875425338745,
      "learning_rate": 5.60418271865215e-06,
      "loss": 1.806,
      "step": 216205
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.5385689735412598,
      "learning_rate": 5.603395780401088e-06,
      "loss": 1.6034,
      "step": 216206
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.3379138708114624,
      "learning_rate": 5.602608896883876e-06,
      "loss": 1.8262,
      "step": 216207
    },
    {
      "epoch": 2.82,
      "grad_norm": 2.2565841674804688,
      "learning_rate": 5.601822068100747e-06,
      "loss": 1.9777,
      "step": 216208
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.489100694656372,
      "learning_rate": 5.60103529405177e-06,
      "loss": 1.9909,
      "step": 216209
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.48039710521698,
      "learning_rate": 5.60024857473711e-06,
      "loss": 2.0498,
      "step": 216210
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.5074589252471924,
      "learning_rate": 5.5994619101569e-06,
      "loss": 2.1438,
      "step": 216211
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.3828895092010498,
      "learning_rate": 5.59867530031134e-06,
      "loss": 2.116,
      "step": 216212
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.4281506538391113,
      "learning_rate": 5.5978887452005295e-06,
      "loss": 2.0175,
      "step": 216213
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.356374979019165,
      "learning_rate": 5.597102244824669e-06,
      "loss": 1.8297,
      "step": 216214
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.3433314561843872,
      "learning_rate": 5.596315799183826e-06,
      "loss": 2.1753,
      "step": 216215
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.4654875993728638,
      "learning_rate": 5.595529408278199e-06,
      "loss": 1.8811,
      "step": 216216
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.4916236400604248,
      "learning_rate": 5.594743072107888e-06,
      "loss": 1.8853,
      "step": 216217
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.5497413873672485,
      "learning_rate": 5.593956790673093e-06,
      "loss": 2.1836,
      "step": 216218
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.3947105407714844,
      "learning_rate": 5.593170563973947e-06,
      "loss": 1.9701,
      "step": 216219
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.411885142326355,
      "learning_rate": 5.592384392010585e-06,
      "loss": 1.7465,
      "step": 216220
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.5856190919876099,
      "learning_rate": 5.591598274783171e-06,
      "loss": 1.9781,
      "step": 216221
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.4122732877731323,
      "learning_rate": 5.590812212291807e-06,
      "loss": 1.8347,
      "step": 216222
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.4203239679336548,
      "learning_rate": 5.590026204536658e-06,
      "loss": 2.1129,
      "step": 216223
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.5690206289291382,
      "learning_rate": 5.589240251517924e-06,
      "loss": 2.0336,
      "step": 216224
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.5290755033493042,
      "learning_rate": 5.5884543532356715e-06,
      "loss": 1.8675,
      "step": 216225
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.5175540447235107,
      "learning_rate": 5.587668509690102e-06,
      "loss": 2.136,
      "step": 216226
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.4604451656341553,
      "learning_rate": 5.586882720881347e-06,
      "loss": 2.0693,
      "step": 216227
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.689639925956726,
      "learning_rate": 5.586096986809507e-06,
      "loss": 2.1628,
      "step": 216228
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.60352623462677,
      "learning_rate": 5.5853113074748155e-06,
      "loss": 2.0295,
      "step": 216229
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.504608392715454,
      "learning_rate": 5.584525682877372e-06,
      "loss": 1.8684,
      "step": 216230
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.4732779264450073,
      "learning_rate": 5.58374011301731e-06,
      "loss": 1.8767,
      "step": 216231
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.55734121799469,
      "learning_rate": 5.5829545978947955e-06,
      "loss": 1.914,
      "step": 216232
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.4640014171600342,
      "learning_rate": 5.582169137509962e-06,
      "loss": 2.0049,
      "step": 216233
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.727415919303894,
      "learning_rate": 5.581383731862943e-06,
      "loss": 1.8966,
      "step": 216234
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.4078248739242554,
      "learning_rate": 5.580598380953905e-06,
      "loss": 1.9492,
      "step": 216235
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.387229084968567,
      "learning_rate": 5.579813084783013e-06,
      "loss": 1.9412,
      "step": 216236
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.4671326875686646,
      "learning_rate": 5.57902784335037e-06,
      "loss": 2.0717,
      "step": 216237
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.5727063417434692,
      "learning_rate": 5.5782426566561735e-06,
      "loss": 2.0472,
      "step": 216238
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.4534547328948975,
      "learning_rate": 5.577457524700523e-06,
      "loss": 1.7303,
      "step": 216239
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.5526869297027588,
      "learning_rate": 5.576672447483554e-06,
      "loss": 1.9891,
      "step": 216240
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.3928272724151611,
      "learning_rate": 5.575887425005465e-06,
      "loss": 2.0172,
      "step": 216241
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.450931191444397,
      "learning_rate": 5.575102457266356e-06,
      "loss": 1.7042,
      "step": 216242
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.414430856704712,
      "learning_rate": 5.574317544266427e-06,
      "loss": 1.9376,
      "step": 216243
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.4553813934326172,
      "learning_rate": 5.573532686005777e-06,
      "loss": 2.1726,
      "step": 216244
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.475394368171692,
      "learning_rate": 5.572747882484574e-06,
      "loss": 1.9787,
      "step": 216245
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.5697568655014038,
      "learning_rate": 5.57196313370295e-06,
      "loss": 1.7657,
      "step": 216246
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.4802942276000977,
      "learning_rate": 5.571178439661039e-06,
      "loss": 2.1598,
      "step": 216247
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.7467515468597412,
      "learning_rate": 5.570393800359008e-06,
      "loss": 2.1079,
      "step": 216248
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.450863242149353,
      "learning_rate": 5.569609215797022e-06,
      "loss": 1.8796,
      "step": 216249
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.382285714149475,
      "learning_rate": 5.568824685975182e-06,
      "loss": 1.8239,
      "step": 216250
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.4633259773254395,
      "learning_rate": 5.568040210893654e-06,
      "loss": 2.0168,
      "step": 216251
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.448771357536316,
      "learning_rate": 5.567255790552605e-06,
      "loss": 2.0035,
      "step": 216252
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.809931755065918,
      "learning_rate": 5.566471424952135e-06,
      "loss": 1.9116,
      "step": 216253
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.5342628955841064,
      "learning_rate": 5.56568711409241e-06,
      "loss": 1.9221,
      "step": 216254
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.5516470670700073,
      "learning_rate": 5.564902857973597e-06,
      "loss": 2.0472,
      "step": 216255
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.5232495069503784,
      "learning_rate": 5.564118656595862e-06,
      "loss": 1.9313,
      "step": 216256
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.5352530479431152,
      "learning_rate": 5.563334509959272e-06,
      "loss": 2.1876,
      "step": 216257
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.4592769145965576,
      "learning_rate": 5.562550418064027e-06,
      "loss": 1.8324,
      "step": 216258
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.3676962852478027,
      "learning_rate": 5.56176638091026e-06,
      "loss": 1.9962,
      "step": 216259
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.461011290550232,
      "learning_rate": 5.560982398498104e-06,
      "loss": 1.9641,
      "step": 216260
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.4451379776000977,
      "learning_rate": 5.560198470827759e-06,
      "loss": 1.7807,
      "step": 216261
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.442552089691162,
      "learning_rate": 5.5594145978992584e-06,
      "loss": 1.7868,
      "step": 216262
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.4795552492141724,
      "learning_rate": 5.558630779712902e-06,
      "loss": 1.7617,
      "step": 216263
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.3747437000274658,
      "learning_rate": 5.557847016268724e-06,
      "loss": 1.7964,
      "step": 216264
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.4577953815460205,
      "learning_rate": 5.557063307566889e-06,
      "loss": 2.1867,
      "step": 216265
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.48332679271698,
      "learning_rate": 5.556279653607565e-06,
      "loss": 2.0309,
      "step": 216266
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.6662981510162354,
      "learning_rate": 5.555496054390884e-06,
      "loss": 1.9195,
      "step": 216267
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.507237195968628,
      "learning_rate": 5.554712509916981e-06,
      "loss": 1.8328,
      "step": 216268
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.4117282629013062,
      "learning_rate": 5.553929020186054e-06,
      "loss": 2.1362,
      "step": 216269
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.4846339225769043,
      "learning_rate": 5.553145585198171e-06,
      "loss": 1.9928,
      "step": 216270
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.4410746097564697,
      "learning_rate": 5.552362204953531e-06,
      "loss": 1.9198,
      "step": 216271
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.4313738346099854,
      "learning_rate": 5.551578879452268e-06,
      "loss": 1.7959,
      "step": 216272
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.4007657766342163,
      "learning_rate": 5.550795608694514e-06,
      "loss": 1.8859,
      "step": 216273
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.4188061952590942,
      "learning_rate": 5.550012392680436e-06,
      "loss": 1.9372,
      "step": 216274
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.444047451019287,
      "learning_rate": 5.549229231410168e-06,
      "loss": 1.924,
      "step": 216275
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.662304401397705,
      "learning_rate": 5.548446124883843e-06,
      "loss": 1.9748,
      "step": 216276
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.5341227054595947,
      "learning_rate": 5.5476630731016605e-06,
      "loss": 2.0743,
      "step": 216277
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.4590328931808472,
      "learning_rate": 5.546880076063653e-06,
      "loss": 1.7022,
      "step": 216278
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.4861568212509155,
      "learning_rate": 5.546097133770089e-06,
      "loss": 2.0223,
      "step": 216279
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.5570662021636963,
      "learning_rate": 5.545314246221067e-06,
      "loss": 1.8846,
      "step": 216280
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.3654478788375854,
      "learning_rate": 5.544531413416719e-06,
      "loss": 1.859,
      "step": 216281
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.3969531059265137,
      "learning_rate": 5.5437486353572144e-06,
      "loss": 1.958,
      "step": 216282
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.3792134523391724,
      "learning_rate": 5.542965912042652e-06,
      "loss": 2.0328,
      "step": 216283
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.460540771484375,
      "learning_rate": 5.542183243473231e-06,
      "loss": 2.0173,
      "step": 216284
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.4563597440719604,
      "learning_rate": 5.541400629649084e-06,
      "loss": 2.0871,
      "step": 216285
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.4963399171829224,
      "learning_rate": 5.540618070570313e-06,
      "loss": 1.7305,
      "step": 216286
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.4849406480789185,
      "learning_rate": 5.539835566237149e-06,
      "loss": 1.9424,
      "step": 216287
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.465203881263733,
      "learning_rate": 5.539053116649661e-06,
      "loss": 1.9262,
      "step": 216288
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.5097726583480835,
      "learning_rate": 5.5382707218080465e-06,
      "loss": 2.2152,
      "step": 216289
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.525975227355957,
      "learning_rate": 5.537488381712374e-06,
      "loss": 1.9634,
      "step": 216290
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.3374823331832886,
      "learning_rate": 5.536706096362875e-06,
      "loss": 1.935,
      "step": 216291
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.4424083232879639,
      "learning_rate": 5.535923865759651e-06,
      "loss": 1.8505,
      "step": 216292
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.5220565795898438,
      "learning_rate": 5.5351416899029e-06,
      "loss": 2.0863,
      "step": 216293
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.3345141410827637,
      "learning_rate": 5.534359568792657e-06,
      "loss": 2.1212,
      "step": 216294
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.447332501411438,
      "learning_rate": 5.533577502429187e-06,
      "loss": 1.7975,
      "step": 216295
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.4951355457305908,
      "learning_rate": 5.532795490812525e-06,
      "loss": 2.3298,
      "step": 216296
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.3544352054595947,
      "learning_rate": 5.532013533942936e-06,
      "loss": 1.9,
      "step": 216297
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.392425298690796,
      "learning_rate": 5.5312316318204545e-06,
      "loss": 1.7028,
      "step": 216298
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.5579240322113037,
      "learning_rate": 5.530449784445345e-06,
      "loss": 1.9448,
      "step": 216299
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.4032014608383179,
      "learning_rate": 5.52966799181761e-06,
      "loss": 2.1988,
      "step": 216300
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.3856959342956543,
      "learning_rate": 5.528886253937514e-06,
      "loss": 1.871,
      "step": 216301
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.467149019241333,
      "learning_rate": 5.528104570805125e-06,
      "loss": 1.8346,
      "step": 216302
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.5824923515319824,
      "learning_rate": 5.5273229424206425e-06,
      "loss": 1.9514,
      "step": 216303
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.4615436792373657,
      "learning_rate": 5.526541368784165e-06,
      "loss": 2.1211,
      "step": 216304
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.6692707538604736,
      "learning_rate": 5.525759849895894e-06,
      "loss": 2.0244,
      "step": 216305
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.4926072359085083,
      "learning_rate": 5.524978385755929e-06,
      "loss": 2.1224,
      "step": 216306
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.4250836372375488,
      "learning_rate": 5.524196976364436e-06,
      "loss": 1.8419,
      "step": 216307
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.5155643224716187,
      "learning_rate": 5.523415621721549e-06,
      "loss": 1.8171,
      "step": 216308
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.3771729469299316,
      "learning_rate": 5.522634321827402e-06,
      "loss": 2.1518,
      "step": 216309
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.6229368448257446,
      "learning_rate": 5.521853076682159e-06,
      "loss": 2.1536,
      "step": 216310
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.3675013780593872,
      "learning_rate": 5.521071886285988e-06,
      "loss": 2.0148,
      "step": 216311
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.5883060693740845,
      "learning_rate": 5.520290750638989e-06,
      "loss": 1.8836,
      "step": 216312
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.3967115879058838,
      "learning_rate": 5.519509669741362e-06,
      "loss": 1.961,
      "step": 216313
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.5263804197311401,
      "learning_rate": 5.518728643593173e-06,
      "loss": 1.8781,
      "step": 216314
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.4739205837249756,
      "learning_rate": 5.517947672194622e-06,
      "loss": 1.9109,
      "step": 216315
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.5989644527435303,
      "learning_rate": 5.5171667555458095e-06,
      "loss": 2.1637,
      "step": 216316
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.434885025024414,
      "learning_rate": 5.516385893646969e-06,
      "loss": 2.0196,
      "step": 216317
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.4759023189544678,
      "learning_rate": 5.515605086498198e-06,
      "loss": 2.0305,
      "step": 216318
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.6650054454803467,
      "learning_rate": 5.514824334099566e-06,
      "loss": 1.8935,
      "step": 216319
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.4264534711837769,
      "learning_rate": 5.514043636451338e-06,
      "loss": 2.103,
      "step": 216320
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.4377049207687378,
      "learning_rate": 5.513262993553613e-06,
      "loss": 1.8663,
      "step": 216321
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.467561960220337,
      "learning_rate": 5.512482405406493e-06,
      "loss": 1.9087,
      "step": 216322
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.6411548852920532,
      "learning_rate": 5.51170187201021e-06,
      "loss": 2.0321,
      "step": 216323
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.5043679475784302,
      "learning_rate": 5.510921393364831e-06,
      "loss": 2.142,
      "step": 216324
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.3470346927642822,
      "learning_rate": 5.510140969470522e-06,
      "loss": 2.1963,
      "step": 216325
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.34187912940979,
      "learning_rate": 5.509360600327484e-06,
      "loss": 1.9112,
      "step": 216326
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.3973420858383179,
      "learning_rate": 5.508580285935748e-06,
      "loss": 1.7482,
      "step": 216327
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.3693292140960693,
      "learning_rate": 5.507800026295583e-06,
      "loss": 2.0947,
      "step": 216328
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.3821085691452026,
      "learning_rate": 5.507019821407055e-06,
      "loss": 2.11,
      "step": 216329
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.4711138010025024,
      "learning_rate": 5.506239671270329e-06,
      "loss": 2.1029,
      "step": 216330
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.4472898244857788,
      "learning_rate": 5.50545957588554e-06,
      "loss": 2.0024,
      "step": 216331
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.5008608102798462,
      "learning_rate": 5.504679535252854e-06,
      "loss": 2.165,
      "step": 216332
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.4424335956573486,
      "learning_rate": 5.503899549372404e-06,
      "loss": 1.8787,
      "step": 216333
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.4073150157928467,
      "learning_rate": 5.503119618244356e-06,
      "loss": 1.9132,
      "step": 216334
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.5658985376358032,
      "learning_rate": 5.502339741868844e-06,
      "loss": 1.8592,
      "step": 216335
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.4980918169021606,
      "learning_rate": 5.5015599202460015e-06,
      "loss": 1.7322,
      "step": 216336
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.4366371631622314,
      "learning_rate": 5.500780153375928e-06,
      "loss": 2.2798,
      "step": 216337
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.456920862197876,
      "learning_rate": 5.50000044125889e-06,
      "loss": 1.9072,
      "step": 216338
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.575650930404663,
      "learning_rate": 5.499220783894886e-06,
      "loss": 2.0091,
      "step": 216339
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.4740980863571167,
      "learning_rate": 5.498441181284185e-06,
      "loss": 2.0374,
      "step": 216340
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.5548338890075684,
      "learning_rate": 5.497661633426853e-06,
      "loss": 2.066,
      "step": 216341
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.4344537258148193,
      "learning_rate": 5.496882140323123e-06,
      "loss": 1.9461,
      "step": 216342
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.4894155263900757,
      "learning_rate": 5.496102701973027e-06,
      "loss": 1.8922,
      "step": 216343
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.434277057647705,
      "learning_rate": 5.495323318376799e-06,
      "loss": 1.8359,
      "step": 216344
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.374548077583313,
      "learning_rate": 5.494543989534506e-06,
      "loss": 1.9196,
      "step": 216345
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.4503268003463745,
      "learning_rate": 5.49376471544638e-06,
      "loss": 2.0287,
      "step": 216346
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.5118153095245361,
      "learning_rate": 5.49298549611249e-06,
      "loss": 2.205,
      "step": 216347
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.496464490890503,
      "learning_rate": 5.4922063315330665e-06,
      "loss": 2.1805,
      "step": 216348
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.617758870124817,
      "learning_rate": 5.491427221708144e-06,
      "loss": 1.9459,
      "step": 216349
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.3724117279052734,
      "learning_rate": 5.4906481666379564e-06,
      "loss": 1.8369,
      "step": 216350
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.401511788368225,
      "learning_rate": 5.4898691663226026e-06,
      "loss": 1.8199,
      "step": 216351
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.3746531009674072,
      "learning_rate": 5.4890902207622485e-06,
      "loss": 1.9902,
      "step": 216352
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.4517968893051147,
      "learning_rate": 5.488311329956996e-06,
      "loss": 1.8567,
      "step": 216353
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.5003855228424072,
      "learning_rate": 5.48753249390711e-06,
      "loss": 2.0886,
      "step": 216354
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.5644724369049072,
      "learning_rate": 5.486753712612557e-06,
      "loss": 1.9852,
      "step": 216355
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.461709976196289,
      "learning_rate": 5.485974986073638e-06,
      "loss": 2.2013,
      "step": 216356
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.5127685070037842,
      "learning_rate": 5.485196314290385e-06,
      "loss": 2.0887,
      "step": 216357
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.5229320526123047,
      "learning_rate": 5.484417697263033e-06,
      "loss": 1.9712,
      "step": 216358
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.464800238609314,
      "learning_rate": 5.48363913499168e-06,
      "loss": 2.0578,
      "step": 216359
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.6156768798828125,
      "learning_rate": 5.482860627476493e-06,
      "loss": 1.9392,
      "step": 216360
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.369978904724121,
      "learning_rate": 5.482082174717573e-06,
      "loss": 1.9748,
      "step": 216361
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.5470892190933228,
      "learning_rate": 5.481303776715118e-06,
      "loss": 1.9155,
      "step": 216362
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.5152151584625244,
      "learning_rate": 5.48052543346923e-06,
      "loss": 1.976,
      "step": 216363
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.4446943998336792,
      "learning_rate": 5.479747144980073e-06,
      "loss": 1.8588,
      "step": 216364
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.5101357698440552,
      "learning_rate": 5.478968911247783e-06,
      "loss": 2.0505,
      "step": 216365
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.5349029302597046,
      "learning_rate": 5.4781907322725584e-06,
      "loss": 1.9196,
      "step": 216366
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.5458446741104126,
      "learning_rate": 5.477412608054432e-06,
      "loss": 2.05,
      "step": 216367
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.487179160118103,
      "learning_rate": 5.476634538593639e-06,
      "loss": 1.8089,
      "step": 216368
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.470807671546936,
      "learning_rate": 5.47585652389031e-06,
      "loss": 1.9877,
      "step": 216369
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.3345674276351929,
      "learning_rate": 5.4750785639445795e-06,
      "loss": 1.9516,
      "step": 216370
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.446755051612854,
      "learning_rate": 5.474300658756548e-06,
      "loss": 2.0185,
      "step": 216371
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.4650474786758423,
      "learning_rate": 5.47352280832648e-06,
      "loss": 1.8586,
      "step": 216372
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.3572545051574707,
      "learning_rate": 5.472745012654378e-06,
      "loss": 1.903,
      "step": 216373
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.4764310121536255,
      "learning_rate": 5.471967271740507e-06,
      "loss": 2.1155,
      "step": 216374
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.507224678993225,
      "learning_rate": 5.471189585584901e-06,
      "loss": 2.1519,
      "step": 216375
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.5958582162857056,
      "learning_rate": 5.470411954187792e-06,
      "loss": 2.0854,
      "step": 216376
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.4527082443237305,
      "learning_rate": 5.469634377549281e-06,
      "loss": 1.8696,
      "step": 216377
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.4428637027740479,
      "learning_rate": 5.468856855669568e-06,
      "loss": 1.7736,
      "step": 216378
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.372254490852356,
      "learning_rate": 5.4680793885487185e-06,
      "loss": 1.927,
      "step": 216379
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.5208019018173218,
      "learning_rate": 5.4673019761869e-06,
      "loss": 1.9868,
      "step": 216380
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.453052043914795,
      "learning_rate": 5.4665246185842785e-06,
      "loss": 1.9321,
      "step": 216381
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.619455337524414,
      "learning_rate": 5.465747315740987e-06,
      "loss": 2.0283,
      "step": 216382
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.5359563827514648,
      "learning_rate": 5.46497006765716e-06,
      "loss": 1.9677,
      "step": 216383
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.399855613708496,
      "learning_rate": 5.464192874333029e-06,
      "loss": 2.0021,
      "step": 216384
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.4502283334732056,
      "learning_rate": 5.463415735768595e-06,
      "loss": 1.9219,
      "step": 216385
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.7177834510803223,
      "learning_rate": 5.462638651964091e-06,
      "loss": 2.0521,
      "step": 216386
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.3772586584091187,
      "learning_rate": 5.461861622919617e-06,
      "loss": 1.9413,
      "step": 216387
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.6836349964141846,
      "learning_rate": 5.4610846486353724e-06,
      "loss": 2.1954,
      "step": 216388
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.4688823223114014,
      "learning_rate": 5.460307729111457e-06,
      "loss": 1.7903,
      "step": 216389
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.4704734086990356,
      "learning_rate": 5.459530864348038e-06,
      "loss": 2.2295,
      "step": 216390
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.5060948133468628,
      "learning_rate": 5.458754054345249e-06,
      "loss": 1.9893,
      "step": 216391
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.4229973554611206,
      "learning_rate": 5.457977299103256e-06,
      "loss": 1.893,
      "step": 216392
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.398063063621521,
      "learning_rate": 5.457200598622158e-06,
      "loss": 1.9754,
      "step": 216393
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.5141124725341797,
      "learning_rate": 5.456423952902156e-06,
      "loss": 1.8902,
      "step": 216394
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.4898195266723633,
      "learning_rate": 5.455647361943316e-06,
      "loss": 1.9824,
      "step": 216395
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.4515422582626343,
      "learning_rate": 5.454870825745871e-06,
      "loss": 1.9303,
      "step": 216396
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.4651103019714355,
      "learning_rate": 5.454094344309922e-06,
      "loss": 1.8954,
      "step": 216397
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.4108411073684692,
      "learning_rate": 5.453317917635636e-06,
      "loss": 2.1028,
      "step": 216398
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.4845478534698486,
      "learning_rate": 5.452541545723077e-06,
      "loss": 1.9462,
      "step": 216399
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.4461393356323242,
      "learning_rate": 5.4517652285725135e-06,
      "loss": 2.1333,
      "step": 216400
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.5957221984863281,
      "learning_rate": 5.450988966183978e-06,
      "loss": 2.0425,
      "step": 216401
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.5299681425094604,
      "learning_rate": 5.45021275855767e-06,
      "loss": 2.0178,
      "step": 216402
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.5526528358459473,
      "learning_rate": 5.449436605693758e-06,
      "loss": 2.0283,
      "step": 216403
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.4455201625823975,
      "learning_rate": 5.4486605075923395e-06,
      "loss": 2.0014,
      "step": 216404
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.5603959560394287,
      "learning_rate": 5.4478844642535825e-06,
      "loss": 1.9567,
      "step": 216405
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.3382501602172852,
      "learning_rate": 5.44710847567762e-06,
      "loss": 1.9334,
      "step": 216406
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.406471610069275,
      "learning_rate": 5.446332541864584e-06,
      "loss": 1.9305,
      "step": 216407
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.4018614292144775,
      "learning_rate": 5.445556662814643e-06,
      "loss": 2.0964,
      "step": 216408
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.4796561002731323,
      "learning_rate": 5.4447808385279625e-06,
      "loss": 2.1214,
      "step": 216409
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.3899372816085815,
      "learning_rate": 5.444005069004609e-06,
      "loss": 2.1437,
      "step": 216410
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.600755214691162,
      "learning_rate": 5.443229354244816e-06,
      "loss": 1.8383,
      "step": 216411
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.5262559652328491,
      "learning_rate": 5.44245369424865e-06,
      "loss": 2.1688,
      "step": 216412
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.4642912149429321,
      "learning_rate": 5.44167808901631e-06,
      "loss": 1.8013,
      "step": 216413
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.4799013137817383,
      "learning_rate": 5.440902538547931e-06,
      "loss": 2.0075,
      "step": 216414
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.4720553159713745,
      "learning_rate": 5.440127042843645e-06,
      "loss": 2.1007,
      "step": 216415
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.4287388324737549,
      "learning_rate": 5.439351601903585e-06,
      "loss": 1.9109,
      "step": 216416
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.4088410139083862,
      "learning_rate": 5.4385762157279525e-06,
      "loss": 2.1531,
      "step": 216417
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.4329673051834106,
      "learning_rate": 5.4378008843167785e-06,
      "loss": 1.9489,
      "step": 216418
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.463674783706665,
      "learning_rate": 5.437025607670331e-06,
      "loss": 1.9254,
      "step": 216419
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.4945709705352783,
      "learning_rate": 5.4362503857887095e-06,
      "loss": 2.0128,
      "step": 216420
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.274360179901123,
      "learning_rate": 5.435475218672014e-06,
      "loss": 1.8216,
      "step": 216421
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.5051078796386719,
      "learning_rate": 5.434700106320444e-06,
      "loss": 1.9726,
      "step": 216422
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.4227635860443115,
      "learning_rate": 5.4339250487341006e-06,
      "loss": 2.0353,
      "step": 216423
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.539747714996338,
      "learning_rate": 5.433150045913148e-06,
      "loss": 1.8673,
      "step": 216424
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.49928617477417,
      "learning_rate": 5.432375097857788e-06,
      "loss": 2.0573,
      "step": 216425
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.5026088953018188,
      "learning_rate": 5.431600204568054e-06,
      "loss": 2.2206,
      "step": 216426
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.4583123922348022,
      "learning_rate": 5.430825366044178e-06,
      "loss": 1.9143,
      "step": 216427
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.3872798681259155,
      "learning_rate": 5.43005058228626e-06,
      "loss": 2.1504,
      "step": 216428
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.4737396240234375,
      "learning_rate": 5.429275853294468e-06,
      "loss": 2.0448,
      "step": 216429
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.5113389492034912,
      "learning_rate": 5.4285011790689e-06,
      "loss": 1.9438,
      "step": 216430
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.5002025365829468,
      "learning_rate": 5.427726559609757e-06,
      "loss": 2.2801,
      "step": 216431
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.589194893836975,
      "learning_rate": 5.426951994917139e-06,
      "loss": 2.0794,
      "step": 216432
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.4985171556472778,
      "learning_rate": 5.426177484991278e-06,
      "loss": 1.9584,
      "step": 216433
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.5265079736709595,
      "learning_rate": 5.425403029832176e-06,
      "loss": 2.0636,
      "step": 216434
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.3598477840423584,
      "learning_rate": 5.424628629440097e-06,
      "loss": 1.9681,
      "step": 216435
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.4889580011367798,
      "learning_rate": 5.42385428381511e-06,
      "loss": 1.995,
      "step": 216436
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.539028525352478,
      "learning_rate": 5.423079992957413e-06,
      "loss": 2.0197,
      "step": 216437
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.5584717988967896,
      "learning_rate": 5.422305756867107e-06,
      "loss": 1.9825,
      "step": 216438
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.5629172325134277,
      "learning_rate": 5.421531575544391e-06,
      "loss": 1.9612,
      "step": 216439
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.5842244625091553,
      "learning_rate": 5.4207574489893324e-06,
      "loss": 1.9848,
      "step": 216440
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.5756772756576538,
      "learning_rate": 5.419983377202131e-06,
      "loss": 2.116,
      "step": 216441
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.541599154472351,
      "learning_rate": 5.419209360182886e-06,
      "loss": 1.9382,
      "step": 216442
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.5424171686172485,
      "learning_rate": 5.418435397931797e-06,
      "loss": 1.9707,
      "step": 216443
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.4832499027252197,
      "learning_rate": 5.417661490448966e-06,
      "loss": 1.8657,
      "step": 216444
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.5592727661132812,
      "learning_rate": 5.416887637734624e-06,
      "loss": 2.0729,
      "step": 216445
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.5638372898101807,
      "learning_rate": 5.416113839788772e-06,
      "loss": 2.0311,
      "step": 216446
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.4756724834442139,
      "learning_rate": 5.415340096611642e-06,
      "loss": 1.9127,
      "step": 216447
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.4815895557403564,
      "learning_rate": 5.414566408203336e-06,
      "loss": 1.9429,
      "step": 216448
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.5038261413574219,
      "learning_rate": 5.413792774564052e-06,
      "loss": 1.9952,
      "step": 216449
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.2835105657577515,
      "learning_rate": 5.413019195693891e-06,
      "loss": 1.77,
      "step": 216450
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.4576141834259033,
      "learning_rate": 5.412245671593052e-06,
      "loss": 1.8914,
      "step": 216451
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.4647064208984375,
      "learning_rate": 5.411472202261602e-06,
      "loss": 1.8375,
      "step": 216452
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.4242262840270996,
      "learning_rate": 5.410698787699707e-06,
      "loss": 1.9289,
      "step": 216453
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.6225919723510742,
      "learning_rate": 5.409925427907535e-06,
      "loss": 1.8538,
      "step": 216454
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.4045250415802002,
      "learning_rate": 5.4091521228852185e-06,
      "loss": 1.9799,
      "step": 216455
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.4533575773239136,
      "learning_rate": 5.4083788726328905e-06,
      "loss": 2.0177,
      "step": 216456
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.4241764545440674,
      "learning_rate": 5.4076056771507505e-06,
      "loss": 2.0205,
      "step": 216457
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.498650312423706,
      "learning_rate": 5.4068325364388655e-06,
      "loss": 1.8914,
      "step": 216458
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.3857777118682861,
      "learning_rate": 5.406059450497402e-06,
      "loss": 1.9089,
      "step": 216459
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.4824579954147339,
      "learning_rate": 5.405286419326527e-06,
      "loss": 1.9964,
      "step": 216460
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.6158239841461182,
      "learning_rate": 5.404513442926373e-06,
      "loss": 2.0106,
      "step": 216461
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.4003853797912598,
      "learning_rate": 5.403740521297073e-06,
      "loss": 1.9344,
      "step": 216462
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.421979546546936,
      "learning_rate": 5.402967654438761e-06,
      "loss": 2.0545,
      "step": 216463
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.6330360174179077,
      "learning_rate": 5.402194842351637e-06,
      "loss": 1.9574,
      "step": 216464
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.563907504081726,
      "learning_rate": 5.401422085035801e-06,
      "loss": 2.0018,
      "step": 216465
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.5969001054763794,
      "learning_rate": 5.400649382491384e-06,
      "loss": 2.0894,
      "step": 216466
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.3412773609161377,
      "learning_rate": 5.399876734718555e-06,
      "loss": 2.2169,
      "step": 216467
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.5190414190292358,
      "learning_rate": 5.399104141717414e-06,
      "loss": 1.9728,
      "step": 216468
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.348418116569519,
      "learning_rate": 5.3983316034881925e-06,
      "loss": 1.7675,
      "step": 216469
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.4905775785446167,
      "learning_rate": 5.397559120030959e-06,
      "loss": 1.8822,
      "step": 216470
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.5151522159576416,
      "learning_rate": 5.396786691345911e-06,
      "loss": 1.9724,
      "step": 216471
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.5008093118667603,
      "learning_rate": 5.396014317433117e-06,
      "loss": 1.8248,
      "step": 216472
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.480379343032837,
      "learning_rate": 5.395241998292777e-06,
      "loss": 1.8046,
      "step": 216473
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.581073522567749,
      "learning_rate": 5.394469733925022e-06,
      "loss": 2.0618,
      "step": 216474
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.461829423904419,
      "learning_rate": 5.3936975243299876e-06,
      "loss": 2.1947,
      "step": 216475
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.413150429725647,
      "learning_rate": 5.392925369507872e-06,
      "loss": 1.9842,
      "step": 216476
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.4301739931106567,
      "learning_rate": 5.392153269458743e-06,
      "loss": 2.0737,
      "step": 216477
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.4315205812454224,
      "learning_rate": 5.391381224182767e-06,
      "loss": 1.9048,
      "step": 216478
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.3790502548217773,
      "learning_rate": 5.3906092336801095e-06,
      "loss": 1.9341,
      "step": 216479
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.481552004814148,
      "learning_rate": 5.389837297950872e-06,
      "loss": 1.9161,
      "step": 216480
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.6526938676834106,
      "learning_rate": 5.389065416995219e-06,
      "loss": 2.1468,
      "step": 216481
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.4819082021713257,
      "learning_rate": 5.388293590813353e-06,
      "loss": 2.0456,
      "step": 216482
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.327568769454956,
      "learning_rate": 5.3875218194053384e-06,
      "loss": 1.8859,
      "step": 216483
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.649437665939331,
      "learning_rate": 5.3867501027713085e-06,
      "loss": 1.9569,
      "step": 216484
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.6757545471191406,
      "learning_rate": 5.385978440911498e-06,
      "loss": 1.9515,
      "step": 216485
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.457983136177063,
      "learning_rate": 5.385206833825939e-06,
      "loss": 2.0053,
      "step": 216486
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.465889573097229,
      "learning_rate": 5.384435281514899e-06,
      "loss": 2.2348,
      "step": 216487
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.4312435388565063,
      "learning_rate": 5.383663783978409e-06,
      "loss": 1.9227,
      "step": 216488
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.4245760440826416,
      "learning_rate": 5.382892341216672e-06,
      "loss": 2.1286,
      "step": 216489
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.3699601888656616,
      "learning_rate": 5.3821209532298184e-06,
      "loss": 2.0977,
      "step": 216490
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.444748044013977,
      "learning_rate": 5.381349620017983e-06,
      "loss": 1.8286,
      "step": 216491
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.401465654373169,
      "learning_rate": 5.380578341581298e-06,
      "loss": 2.0637,
      "step": 216492
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.4726855754852295,
      "learning_rate": 5.379807117919965e-06,
      "loss": 2.0178,
      "step": 216493
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.5204341411590576,
      "learning_rate": 5.379035949034049e-06,
      "loss": 1.8658,
      "step": 216494
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.4923934936523438,
      "learning_rate": 5.378264834923751e-06,
      "loss": 1.7933,
      "step": 216495
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.4960517883300781,
      "learning_rate": 5.3774937755892035e-06,
      "loss": 1.861,
      "step": 216496
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.503173589706421,
      "learning_rate": 5.376722771030539e-06,
      "loss": 2.1054,
      "step": 216497
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.4507064819335938,
      "learning_rate": 5.375951821247859e-06,
      "loss": 1.9406,
      "step": 216498
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.6749383211135864,
      "learning_rate": 5.375180926241396e-06,
      "loss": 1.8354,
      "step": 216499
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.4281306266784668,
      "learning_rate": 5.3744100860112496e-06,
      "loss": 1.9671,
      "step": 216500
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.58608078956604,
      "learning_rate": 5.373639300557553e-06,
      "loss": 1.9012,
      "step": 216501
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.526760458946228,
      "learning_rate": 5.3728685698804396e-06,
      "loss": 2.0125,
      "step": 216502
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.4989323616027832,
      "learning_rate": 5.372097893980076e-06,
      "loss": 2.0285,
      "step": 216503
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.3240256309509277,
      "learning_rate": 5.371327272856629e-06,
      "loss": 1.9384,
      "step": 216504
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.66148042678833,
      "learning_rate": 5.370556706510199e-06,
      "loss": 1.9993,
      "step": 216505
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.5599842071533203,
      "learning_rate": 5.369786194940951e-06,
      "loss": 1.9906,
      "step": 216506
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.3907690048217773,
      "learning_rate": 5.369015738149018e-06,
      "loss": 1.8635,
      "step": 216507
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.366129755973816,
      "learning_rate": 5.368245336134536e-06,
      "loss": 2.0433,
      "step": 216508
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.5015820264816284,
      "learning_rate": 5.367474988897636e-06,
      "loss": 2.0277,
      "step": 216509
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.4015893936157227,
      "learning_rate": 5.36670469643855e-06,
      "loss": 2.122,
      "step": 216510
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.7456494569778442,
      "learning_rate": 5.365934458757281e-06,
      "loss": 1.9416,
      "step": 216511
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.5463967323303223,
      "learning_rate": 5.365164275854128e-06,
      "loss": 1.7764,
      "step": 216512
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.6128627061843872,
      "learning_rate": 5.364394147729089e-06,
      "loss": 2.2266,
      "step": 216513
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.3898757696151733,
      "learning_rate": 5.3636240743824e-06,
      "loss": 1.9495,
      "step": 216514
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.3413554430007935,
      "learning_rate": 5.3628540558141586e-06,
      "loss": 1.8924,
      "step": 216515
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.568541169166565,
      "learning_rate": 5.362084092024532e-06,
      "loss": 1.9942,
      "step": 216516
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.6390780210494995,
      "learning_rate": 5.36131418301362e-06,
      "loss": 2.1779,
      "step": 216517
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.579418420791626,
      "learning_rate": 5.360544328781691e-06,
      "loss": 1.8072,
      "step": 216518
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.4456291198730469,
      "learning_rate": 5.359774529328708e-06,
      "loss": 1.959,
      "step": 216519
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.4221971035003662,
      "learning_rate": 5.359004784654941e-06,
      "loss": 1.8373,
      "step": 216520
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.4426242113113403,
      "learning_rate": 5.358235094760488e-06,
      "loss": 1.8681,
      "step": 216521
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.6479711532592773,
      "learning_rate": 5.3574654596455155e-06,
      "loss": 2.1996,
      "step": 216522
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.510920524597168,
      "learning_rate": 5.356695879310124e-06,
      "loss": 1.9061,
      "step": 216523
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.4357044696807861,
      "learning_rate": 5.355926353754547e-06,
      "loss": 1.9955,
      "step": 216524
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.4225537776947021,
      "learning_rate": 5.355156882978817e-06,
      "loss": 2.0484,
      "step": 216525
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.5922726392745972,
      "learning_rate": 5.354387466983134e-06,
      "loss": 1.8117,
      "step": 216526
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.4257879257202148,
      "learning_rate": 5.353618105767599e-06,
      "loss": 1.927,
      "step": 216527
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.48539137840271,
      "learning_rate": 5.352848799332443e-06,
      "loss": 1.8449,
      "step": 216528
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.4465968608856201,
      "learning_rate": 5.352079547677701e-06,
      "loss": 2.1991,
      "step": 216529
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.6964985132217407,
      "learning_rate": 5.351310350803606e-06,
      "loss": 1.8608,
      "step": 216530
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.4307934045791626,
      "learning_rate": 5.350541208710291e-06,
      "loss": 1.8564,
      "step": 216531
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.4583183526992798,
      "learning_rate": 5.349772121397822e-06,
      "loss": 1.9221,
      "step": 216532
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.3502343893051147,
      "learning_rate": 5.3490030888664e-06,
      "loss": 1.7974,
      "step": 216533
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.465410828590393,
      "learning_rate": 5.34823411111619e-06,
      "loss": 1.8292,
      "step": 216534
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.4758687019348145,
      "learning_rate": 5.34746518814726e-06,
      "loss": 1.8957,
      "step": 216535
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.367534875869751,
      "learning_rate": 5.346696319959842e-06,
      "loss": 1.7794,
      "step": 216536
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.5848121643066406,
      "learning_rate": 5.345927506554004e-06,
      "loss": 2.1515,
      "step": 216537
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.5578420162200928,
      "learning_rate": 5.345158747929945e-06,
      "loss": 2.0021,
      "step": 216538
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.5040216445922852,
      "learning_rate": 5.344390044087765e-06,
      "loss": 2.0501,
      "step": 216539
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.3515701293945312,
      "learning_rate": 5.343621395027631e-06,
      "loss": 1.9419,
      "step": 216540
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.372088074684143,
      "learning_rate": 5.342852800749675e-06,
      "loss": 2.2569,
      "step": 216541
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.448933720588684,
      "learning_rate": 5.342084261254065e-06,
      "loss": 1.9318,
      "step": 216542
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.319331407546997,
      "learning_rate": 5.3413157765409e-06,
      "loss": 1.7551,
      "step": 216543
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.419204831123352,
      "learning_rate": 5.340547346610347e-06,
      "loss": 2.0346,
      "step": 216544
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.4485045671463013,
      "learning_rate": 5.339778971462539e-06,
      "loss": 1.7376,
      "step": 216545
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.5773649215698242,
      "learning_rate": 5.339010651097675e-06,
      "loss": 2.0086,
      "step": 216546
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.4894883632659912,
      "learning_rate": 5.33824238551579e-06,
      "loss": 2.2238,
      "step": 216547
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.4464606046676636,
      "learning_rate": 5.337474174717149e-06,
      "loss": 1.9505,
      "step": 216548
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.4788873195648193,
      "learning_rate": 5.336706018701786e-06,
      "loss": 2.0958,
      "step": 216549
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.407627820968628,
      "learning_rate": 5.335937917469935e-06,
      "loss": 2.0267,
      "step": 216550
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.4136385917663574,
      "learning_rate": 5.335169871021661e-06,
      "loss": 1.8555,
      "step": 216551
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.4722667932510376,
      "learning_rate": 5.334401879357131e-06,
      "loss": 1.9024,
      "step": 216552
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.4348353147506714,
      "learning_rate": 5.333633942476512e-06,
      "loss": 1.986,
      "step": 216553
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.4079504013061523,
      "learning_rate": 5.332866060379937e-06,
      "loss": 1.8136,
      "step": 216554
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.3644939661026,
      "learning_rate": 5.332098233067572e-06,
      "loss": 2.0828,
      "step": 216555
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.3788585662841797,
      "learning_rate": 5.331330460539485e-06,
      "loss": 1.7713,
      "step": 216556
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.5030544996261597,
      "learning_rate": 5.330562742795874e-06,
      "loss": 1.9904,
      "step": 216557
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.4152028560638428,
      "learning_rate": 5.329795079836907e-06,
      "loss": 1.8694,
      "step": 216558
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.4065033197402954,
      "learning_rate": 5.32902747166265e-06,
      "loss": 1.7544,
      "step": 216559
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.5225132703781128,
      "learning_rate": 5.328259918273337e-06,
      "loss": 2.1753,
      "step": 216560
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.3786730766296387,
      "learning_rate": 5.327492419669032e-06,
      "loss": 2.1767,
      "step": 216561
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.5471608638763428,
      "learning_rate": 5.326724975849905e-06,
      "loss": 1.9278,
      "step": 216562
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.4604448080062866,
      "learning_rate": 5.325957586816121e-06,
      "loss": 1.9278,
      "step": 216563
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.6248955726623535,
      "learning_rate": 5.325190252567779e-06,
      "loss": 1.9604,
      "step": 216564
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.574256420135498,
      "learning_rate": 5.324422973105047e-06,
      "loss": 2.1023,
      "step": 216565
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.4742369651794434,
      "learning_rate": 5.323655748428091e-06,
      "loss": 1.892,
      "step": 216566
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.3280653953552246,
      "learning_rate": 5.322888578537044e-06,
      "loss": 1.8,
      "step": 216567
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.4144362211227417,
      "learning_rate": 5.322121463432005e-06,
      "loss": 1.9076,
      "step": 216568
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.4084314107894897,
      "learning_rate": 5.321354403113143e-06,
      "loss": 1.9268,
      "step": 216569
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.612328290939331,
      "learning_rate": 5.320587397580623e-06,
      "loss": 1.991,
      "step": 216570
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.4964003562927246,
      "learning_rate": 5.319820446834544e-06,
      "loss": 1.9262,
      "step": 216571
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.4753869771957397,
      "learning_rate": 5.319053550875074e-06,
      "loss": 1.7619,
      "step": 216572
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.516921043395996,
      "learning_rate": 5.3182867097023796e-06,
      "loss": 2.0054,
      "step": 216573
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.5009346008300781,
      "learning_rate": 5.31751992331656e-06,
      "loss": 1.976,
      "step": 216574
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.5660368204116821,
      "learning_rate": 5.316753191717782e-06,
      "loss": 1.9811,
      "step": 216575
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.7826634645462036,
      "learning_rate": 5.315986514906178e-06,
      "loss": 1.8566,
      "step": 216576
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.446190357208252,
      "learning_rate": 5.315219892881883e-06,
      "loss": 2.071,
      "step": 216577
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.673611044883728,
      "learning_rate": 5.314453325645096e-06,
      "loss": 1.8082,
      "step": 216578
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.5638635158538818,
      "learning_rate": 5.313686813195883e-06,
      "loss": 2.0908,
      "step": 216579
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.3920871019363403,
      "learning_rate": 5.3129203555344115e-06,
      "loss": 2.1232,
      "step": 216580
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.5064902305603027,
      "learning_rate": 5.312153952660847e-06,
      "loss": 1.8761,
      "step": 216581
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.4591174125671387,
      "learning_rate": 5.3113876045753234e-06,
      "loss": 2.103,
      "step": 216582
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.3366432189941406,
      "learning_rate": 5.310621311277941e-06,
      "loss": 1.88,
      "step": 216583
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.5422601699829102,
      "learning_rate": 5.309855072768898e-06,
      "loss": 1.8811,
      "step": 216584
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.6787625551223755,
      "learning_rate": 5.3090888890483295e-06,
      "loss": 2.1801,
      "step": 216585
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.3361175060272217,
      "learning_rate": 5.308322760116335e-06,
      "loss": 1.9081,
      "step": 216586
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.5174074172973633,
      "learning_rate": 5.307556685973113e-06,
      "loss": 2.0635,
      "step": 216587
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.4510523080825806,
      "learning_rate": 5.306790666618732e-06,
      "loss": 1.8878,
      "step": 216588
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.442696213722229,
      "learning_rate": 5.306024702053424e-06,
      "loss": 1.7962,
      "step": 216589
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.506646990776062,
      "learning_rate": 5.305258792277289e-06,
      "loss": 2.0571,
      "step": 216590
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.4459140300750732,
      "learning_rate": 5.304492937290461e-06,
      "loss": 1.6867,
      "step": 216591
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.60163152217865,
      "learning_rate": 5.303727137093072e-06,
      "loss": 2.2272,
      "step": 216592
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.4657249450683594,
      "learning_rate": 5.3029613916853235e-06,
      "loss": 1.8334,
      "step": 216593
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.7296899557113647,
      "learning_rate": 5.30219570106728e-06,
      "loss": 2.0694,
      "step": 216594
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.4140644073486328,
      "learning_rate": 5.301430065239143e-06,
      "loss": 1.8673,
      "step": 216595
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.4711637496948242,
      "learning_rate": 5.300664484201045e-06,
      "loss": 1.9203,
      "step": 216596
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.386815071105957,
      "learning_rate": 5.299898957953053e-06,
      "loss": 1.9184,
      "step": 216597
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.5370186567306519,
      "learning_rate": 5.2991334864954325e-06,
      "loss": 2.079,
      "step": 216598
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.4529629945755005,
      "learning_rate": 5.298368069828252e-06,
      "loss": 2.0755,
      "step": 216599
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.5050064325332642,
      "learning_rate": 5.297602707951676e-06,
      "loss": 1.9691,
      "step": 216600
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.4349455833435059,
      "learning_rate": 5.296837400865806e-06,
      "loss": 1.8202,
      "step": 216601
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.4563806056976318,
      "learning_rate": 5.29607214857084e-06,
      "loss": 2.0535,
      "step": 216602
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.4263432025909424,
      "learning_rate": 5.295306951066913e-06,
      "loss": 2.1369,
      "step": 216603
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.4211944341659546,
      "learning_rate": 5.294541808354125e-06,
      "loss": 2.1378,
      "step": 216604
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.4993367195129395,
      "learning_rate": 5.293776720432674e-06,
      "loss": 2.0447,
      "step": 216605
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.4509413242340088,
      "learning_rate": 5.2930116873026285e-06,
      "loss": 1.8503,
      "step": 216606
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.4924798011779785,
      "learning_rate": 5.29224670896422e-06,
      "loss": 1.8567,
      "step": 216607
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.4047858715057373,
      "learning_rate": 5.291481785417517e-06,
      "loss": 2.0146,
      "step": 216608
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.433590292930603,
      "learning_rate": 5.290716916662685e-06,
      "loss": 1.9755,
      "step": 216609
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.3899415731430054,
      "learning_rate": 5.28995210269989e-06,
      "loss": 2.0406,
      "step": 216610
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.4528480768203735,
      "learning_rate": 5.2891873435292665e-06,
      "loss": 1.8705,
      "step": 216611
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.4016296863555908,
      "learning_rate": 5.288422639150913e-06,
      "loss": 1.7409,
      "step": 216612
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.4631112813949585,
      "learning_rate": 5.2876579895650304e-06,
      "loss": 1.909,
      "step": 216613
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.4057774543762207,
      "learning_rate": 5.286893394771719e-06,
      "loss": 1.9955,
      "step": 216614
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.389296054840088,
      "learning_rate": 5.286128854771177e-06,
      "loss": 1.9266,
      "step": 216615
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.5901024341583252,
      "learning_rate": 5.285364369563472e-06,
      "loss": 1.8924,
      "step": 216616
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.4022854566574097,
      "learning_rate": 5.2845999391488034e-06,
      "loss": 1.9261,
      "step": 216617
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.6099233627319336,
      "learning_rate": 5.283835563527239e-06,
      "loss": 2.0406,
      "step": 216618
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.4550738334655762,
      "learning_rate": 5.2830712426990105e-06,
      "loss": 1.8348,
      "step": 216619
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.5718002319335938,
      "learning_rate": 5.282306976664219e-06,
      "loss": 2.0067,
      "step": 216620
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.4588465690612793,
      "learning_rate": 5.2815427654230305e-06,
      "loss": 2.0135,
      "step": 216621
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.5508075952529907,
      "learning_rate": 5.280778608975545e-06,
      "loss": 1.9697,
      "step": 216622
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.4206334352493286,
      "learning_rate": 5.280014507321928e-06,
      "loss": 1.9737,
      "step": 216623
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.3766531944274902,
      "learning_rate": 5.279250460462314e-06,
      "loss": 1.7521,
      "step": 216624
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.2761188745498657,
      "learning_rate": 5.278486468396837e-06,
      "loss": 2.1422,
      "step": 216625
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.4793812036514282,
      "learning_rate": 5.277722531125661e-06,
      "loss": 1.906,
      "step": 216626
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.4220387935638428,
      "learning_rate": 5.276958648648955e-06,
      "loss": 1.8684,
      "step": 216627
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.6677086353302002,
      "learning_rate": 5.276194820966784e-06,
      "loss": 1.9435,
      "step": 216628
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.4396040439605713,
      "learning_rate": 5.275431048079382e-06,
      "loss": 2.0174,
      "step": 216629
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.3811005353927612,
      "learning_rate": 5.2746673299867816e-06,
      "loss": 2.0062,
      "step": 216630
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.5272283554077148,
      "learning_rate": 5.273903666689216e-06,
      "loss": 2.0664,
      "step": 216631
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.4610824584960938,
      "learning_rate": 5.273140058186754e-06,
      "loss": 1.7017,
      "step": 216632
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.663282871246338,
      "learning_rate": 5.272376504479625e-06,
      "loss": 1.8588,
      "step": 216633
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.5455836057662964,
      "learning_rate": 5.2716130055679315e-06,
      "loss": 1.9755,
      "step": 216634
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.433363676071167,
      "learning_rate": 5.270849561451807e-06,
      "loss": 1.9018,
      "step": 216635
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.4721667766571045,
      "learning_rate": 5.270086172131349e-06,
      "loss": 2.18,
      "step": 216636
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.4562312364578247,
      "learning_rate": 5.2693228376067935e-06,
      "loss": 2.0378,
      "step": 216637
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.5204927921295166,
      "learning_rate": 5.268559557878205e-06,
      "loss": 1.9752,
      "step": 216638
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.3971190452575684,
      "learning_rate": 5.267796332945751e-06,
      "loss": 1.8855,
      "step": 216639
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.5690561532974243,
      "learning_rate": 5.267033162809597e-06,
      "loss": 2.1277,
      "step": 216640
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.3736600875854492,
      "learning_rate": 5.266270047469878e-06,
      "loss": 1.857,
      "step": 216641
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.3742384910583496,
      "learning_rate": 5.265506986926693e-06,
      "loss": 1.9035,
      "step": 216642
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.3816814422607422,
      "learning_rate": 5.264743981180209e-06,
      "loss": 1.8126,
      "step": 216643
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.488564133644104,
      "learning_rate": 5.263981030230591e-06,
      "loss": 2.0305,
      "step": 216644
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.5215036869049072,
      "learning_rate": 5.263218134077973e-06,
      "loss": 1.8729,
      "step": 216645
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.3877770900726318,
      "learning_rate": 5.2624552927224565e-06,
      "loss": 2.0956,
      "step": 216646
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.3097050189971924,
      "learning_rate": 5.261692506164239e-06,
      "loss": 1.8303,
      "step": 216647
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.4523952007293701,
      "learning_rate": 5.260929774403422e-06,
      "loss": 1.7901,
      "step": 216648
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.4672120809555054,
      "learning_rate": 5.260167097440171e-06,
      "loss": 1.8788,
      "step": 216649
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.438062310218811,
      "learning_rate": 5.259404475274587e-06,
      "loss": 2.0673,
      "step": 216650
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.576570749282837,
      "learning_rate": 5.258641907906902e-06,
      "loss": 1.927,
      "step": 216651
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.5063879489898682,
      "learning_rate": 5.2578793953371835e-06,
      "loss": 1.8349,
      "step": 216652
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.4710087776184082,
      "learning_rate": 5.257116937565564e-06,
      "loss": 1.9397,
      "step": 216653
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.6192195415496826,
      "learning_rate": 5.2563545345922105e-06,
      "loss": 1.9957,
      "step": 216654
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.396283507347107,
      "learning_rate": 5.25559218641729e-06,
      "loss": 1.9103,
      "step": 216655
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.3465059995651245,
      "learning_rate": 5.2548298930408685e-06,
      "loss": 1.9694,
      "step": 216656
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.6436907052993774,
      "learning_rate": 5.254067654463179e-06,
      "loss": 2.0674,
      "step": 216657
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.4187827110290527,
      "learning_rate": 5.253305470684288e-06,
      "loss": 2.1234,
      "step": 216658
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.4491647481918335,
      "learning_rate": 5.2525433417044295e-06,
      "loss": 2.0064,
      "step": 216659
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.4317011833190918,
      "learning_rate": 5.251781267523636e-06,
      "loss": 1.9663,
      "step": 216660
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.4095109701156616,
      "learning_rate": 5.251019248142141e-06,
      "loss": 1.8909,
      "step": 216661
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.34373939037323,
      "learning_rate": 5.250257283560011e-06,
      "loss": 1.8249,
      "step": 216662
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.5571757555007935,
      "learning_rate": 5.249495373777446e-06,
      "loss": 1.9526,
      "step": 216663
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.410293459892273,
      "learning_rate": 5.248733518794546e-06,
      "loss": 2.0587,
      "step": 216664
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.4010149240493774,
      "learning_rate": 5.247971718611477e-06,
      "loss": 1.8195,
      "step": 216665
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.4277923107147217,
      "learning_rate": 5.247209973228372e-06,
      "loss": 1.8606,
      "step": 216666
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.5006680488586426,
      "learning_rate": 5.246448282645399e-06,
      "loss": 1.8663,
      "step": 216667
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.5044605731964111,
      "learning_rate": 5.245686646862624e-06,
      "loss": 2.0851,
      "step": 216668
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.495847225189209,
      "learning_rate": 5.244925065880279e-06,
      "loss": 2.0337,
      "step": 216669
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.3496085405349731,
      "learning_rate": 5.244163539698432e-06,
      "loss": 1.98,
      "step": 216670
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.4118801355361938,
      "learning_rate": 5.2434020683172815e-06,
      "loss": 1.8396,
      "step": 216671
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.6147769689559937,
      "learning_rate": 5.2426406517369625e-06,
      "loss": 1.871,
      "step": 216672
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.6643192768096924,
      "learning_rate": 5.241879289957574e-06,
      "loss": 1.9659,
      "step": 216673
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.4605714082717896,
      "learning_rate": 5.241117982979315e-06,
      "loss": 1.9946,
      "step": 216674
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.4669655561447144,
      "learning_rate": 5.240356730802286e-06,
      "loss": 2.0274,
      "step": 216675
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.565812587738037,
      "learning_rate": 5.239595533426588e-06,
      "loss": 2.1256,
      "step": 216676
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.3926289081573486,
      "learning_rate": 5.238834390852486e-06,
      "loss": 2.0461,
      "step": 216677
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.672799825668335,
      "learning_rate": 5.238073303080015e-06,
      "loss": 1.9423,
      "step": 216678
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.5316917896270752,
      "learning_rate": 5.23731227010934e-06,
      "loss": 2.0822,
      "step": 216679
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.350465178489685,
      "learning_rate": 5.236551291940627e-06,
      "loss": 1.955,
      "step": 216680
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.4860255718231201,
      "learning_rate": 5.235790368574011e-06,
      "loss": 1.9712,
      "step": 216681
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.4237200021743774,
      "learning_rate": 5.235029500009591e-06,
      "loss": 1.9972,
      "step": 216682
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.510164737701416,
      "learning_rate": 5.2342686862476006e-06,
      "loss": 1.8986,
      "step": 216683
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.5945887565612793,
      "learning_rate": 5.233507927288072e-06,
      "loss": 1.8488,
      "step": 216684
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.4105863571166992,
      "learning_rate": 5.232747223131206e-06,
      "loss": 1.9904,
      "step": 216685
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.5088838338851929,
      "learning_rate": 5.231986573777169e-06,
      "loss": 1.9867,
      "step": 216686
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.4731115102767944,
      "learning_rate": 5.231225979226028e-06,
      "loss": 1.916,
      "step": 216687
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.5356136560440063,
      "learning_rate": 5.230465439477982e-06,
      "loss": 2.1002,
      "step": 216688
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.634658694267273,
      "learning_rate": 5.229704954533198e-06,
      "loss": 1.8633,
      "step": 216689
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.4237546920776367,
      "learning_rate": 5.228944524391743e-06,
      "loss": 2.1987,
      "step": 216690
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.5955644845962524,
      "learning_rate": 5.2281841490537825e-06,
      "loss": 1.9617,
      "step": 216691
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.367045283317566,
      "learning_rate": 5.22742382851945e-06,
      "loss": 1.8545,
      "step": 216692
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.4873243570327759,
      "learning_rate": 5.226663562788913e-06,
      "loss": 2.028,
      "step": 216693
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.4825692176818848,
      "learning_rate": 5.225903351862337e-06,
      "loss": 1.9548,
      "step": 216694
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.5404235124588013,
      "learning_rate": 5.2251431957398226e-06,
      "loss": 2.0046,
      "step": 216695
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.5404579639434814,
      "learning_rate": 5.2243830944215024e-06,
      "loss": 1.9137,
      "step": 216696
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.4931801557540894,
      "learning_rate": 5.223623047907544e-06,
      "loss": 2.0458,
      "step": 216697
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.6594069004058838,
      "learning_rate": 5.222863056198079e-06,
      "loss": 2.0411,
      "step": 216698
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.5952744483947754,
      "learning_rate": 5.2221031192932084e-06,
      "loss": 2.1001,
      "step": 216699
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.37885320186615,
      "learning_rate": 5.221343237193165e-06,
      "loss": 1.7164,
      "step": 216700
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.452818751335144,
      "learning_rate": 5.220583409898016e-06,
      "loss": 1.7999,
      "step": 216701
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.4580883979797363,
      "learning_rate": 5.219823637407927e-06,
      "loss": 2.1565,
      "step": 216702
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.3829575777053833,
      "learning_rate": 5.219063919723032e-06,
      "loss": 1.8683,
      "step": 216703
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.452359914779663,
      "learning_rate": 5.218304256843497e-06,
      "loss": 2.027,
      "step": 216704
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.6529120206832886,
      "learning_rate": 5.2175446487694225e-06,
      "loss": 2.0776,
      "step": 216705
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.471526861190796,
      "learning_rate": 5.216785095500975e-06,
      "loss": 1.9527,
      "step": 216706
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.5174344778060913,
      "learning_rate": 5.216025597038287e-06,
      "loss": 1.903,
      "step": 216707
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.434949278831482,
      "learning_rate": 5.215266153381525e-06,
      "loss": 1.9807,
      "step": 216708
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.5532593727111816,
      "learning_rate": 5.21450676453079e-06,
      "loss": 1.9639,
      "step": 216709
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.7195000648498535,
      "learning_rate": 5.213747430486248e-06,
      "loss": 2.0651,
      "step": 216710
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.5641392469406128,
      "learning_rate": 5.212988151248032e-06,
      "loss": 1.8519,
      "step": 216711
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.44472336769104,
      "learning_rate": 5.2122289268163085e-06,
      "loss": 1.7953,
      "step": 216712
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.361263632774353,
      "learning_rate": 5.211469757191178e-06,
      "loss": 2.1208,
      "step": 216713
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.6009037494659424,
      "learning_rate": 5.2107106423728065e-06,
      "loss": 1.9139,
      "step": 216714
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.4406875371932983,
      "learning_rate": 5.2099515823612934e-06,
      "loss": 1.9891,
      "step": 216715
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.4354108572006226,
      "learning_rate": 5.2091925771568736e-06,
      "loss": 2.0558,
      "step": 216716
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.3965164422988892,
      "learning_rate": 5.208433626759578e-06,
      "loss": 2.0518,
      "step": 216717
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.4167060852050781,
      "learning_rate": 5.207674731169609e-06,
      "loss": 2.0541,
      "step": 216718
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.5024508237838745,
      "learning_rate": 5.206915890387131e-06,
      "loss": 1.878,
      "step": 216719
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.4522531032562256,
      "learning_rate": 5.2061571044122455e-06,
      "loss": 2.1739,
      "step": 216720
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.407469391822815,
      "learning_rate": 5.205398373245051e-06,
      "loss": 1.8109,
      "step": 216721
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.4277554750442505,
      "learning_rate": 5.2046396968857814e-06,
      "loss": 1.965,
      "step": 216722
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.4758468866348267,
      "learning_rate": 5.203881075334504e-06,
      "loss": 2.1024,
      "step": 216723
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.4159828424453735,
      "learning_rate": 5.203122508591417e-06,
      "loss": 2.0602,
      "step": 216724
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.5282676219940186,
      "learning_rate": 5.202363996656622e-06,
      "loss": 1.7489,
      "step": 216725
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.6045286655426025,
      "learning_rate": 5.201605539530318e-06,
      "loss": 1.7462,
      "step": 216726
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.3799065351486206,
      "learning_rate": 5.200847137212538e-06,
      "loss": 1.6571,
      "step": 216727
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.411739468574524,
      "learning_rate": 5.200088789703516e-06,
      "loss": 1.9526,
      "step": 216728
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.4462250471115112,
      "learning_rate": 5.199330497003351e-06,
      "loss": 2.0046,
      "step": 216729
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.5916495323181152,
      "learning_rate": 5.198572259112177e-06,
      "loss": 1.887,
      "step": 216730
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.5091512203216553,
      "learning_rate": 5.19781407603016e-06,
      "loss": 2.0126,
      "step": 216731
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.6889475584030151,
      "learning_rate": 5.1970559477575e-06,
      "loss": 2.0551,
      "step": 216732
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.4518553018569946,
      "learning_rate": 5.196297874294198e-06,
      "loss": 1.9222,
      "step": 216733
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.5095363855361938,
      "learning_rate": 5.195539855640485e-06,
      "loss": 2.011,
      "step": 216734
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.5500601530075073,
      "learning_rate": 5.194781891796496e-06,
      "loss": 2.2189,
      "step": 216735
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.5784337520599365,
      "learning_rate": 5.194023982762363e-06,
      "loss": 1.9173,
      "step": 216736
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.4976972341537476,
      "learning_rate": 5.193266128538187e-06,
      "loss": 2.0,
      "step": 216737
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.5568214654922485,
      "learning_rate": 5.192508329124234e-06,
      "loss": 1.8622,
      "step": 216738
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.5210096836090088,
      "learning_rate": 5.191750584520471e-06,
      "loss": 1.9761,
      "step": 216739
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.4844238758087158,
      "learning_rate": 5.190992894727197e-06,
      "loss": 1.862,
      "step": 216740
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.534841537475586,
      "learning_rate": 5.190235259744413e-06,
      "loss": 2.0973,
      "step": 216741
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.5106579065322876,
      "learning_rate": 5.189477679572385e-06,
      "loss": 2.1441,
      "step": 216742
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.4766520261764526,
      "learning_rate": 5.188720154211146e-06,
      "loss": 1.8585,
      "step": 216743
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.3084272146224976,
      "learning_rate": 5.187962683660962e-06,
      "loss": 2.0459,
      "step": 216744
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.3322423696517944,
      "learning_rate": 5.187205267921834e-06,
      "loss": 1.9156,
      "step": 216745
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.6002917289733887,
      "learning_rate": 5.186447906993996e-06,
      "loss": 1.8775,
      "step": 216746
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.6085224151611328,
      "learning_rate": 5.1856906008775456e-06,
      "loss": 2.0455,
      "step": 216747
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.458945631980896,
      "learning_rate": 5.184933349572684e-06,
      "loss": 2.0654,
      "step": 216748
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.5561859607696533,
      "learning_rate": 5.1841761530794445e-06,
      "loss": 1.9822,
      "step": 216749
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.4643142223358154,
      "learning_rate": 5.1834190113980934e-06,
      "loss": 2.1027,
      "step": 216750
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.3485556840896606,
      "learning_rate": 5.182661924528664e-06,
      "loss": 2.0581,
      "step": 216751
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.6449121236801147,
      "learning_rate": 5.181904892471389e-06,
      "loss": 1.9196,
      "step": 216752
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.5826507806777954,
      "learning_rate": 5.181147915226302e-06,
      "loss": 1.9717,
      "step": 216753
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.4106417894363403,
      "learning_rate": 5.180390992793671e-06,
      "loss": 1.8826,
      "step": 216754
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.565434217453003,
      "learning_rate": 5.179634125173493e-06,
      "loss": 2.1196,
      "step": 216755
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.5255553722381592,
      "learning_rate": 5.17887731236607e-06,
      "loss": 1.7511,
      "step": 216756
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.5047365427017212,
      "learning_rate": 5.178120554371401e-06,
      "loss": 1.7492,
      "step": 216757
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.4638532400131226,
      "learning_rate": 5.17736385118972e-06,
      "loss": 2.0586,
      "step": 216758
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.5285942554473877,
      "learning_rate": 5.176607202821093e-06,
      "loss": 1.9695,
      "step": 216759
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.441318392753601,
      "learning_rate": 5.175850609265753e-06,
      "loss": 1.9801,
      "step": 216760
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.5044188499450684,
      "learning_rate": 5.175094070523733e-06,
      "loss": 1.8985,
      "step": 216761
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.6731913089752197,
      "learning_rate": 5.174337586595267e-06,
      "loss": 2.0524,
      "step": 216762
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.655840516090393,
      "learning_rate": 5.173581157480455e-06,
      "loss": 1.96,
      "step": 216763
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.5088402032852173,
      "learning_rate": 5.172824783179397e-06,
      "loss": 2.119,
      "step": 216764
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.3617850542068481,
      "learning_rate": 5.1720684636923246e-06,
      "loss": 1.6431,
      "step": 216765
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.5173171758651733,
      "learning_rate": 5.171312199019306e-06,
      "loss": 2.0019,
      "step": 216766
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.4802109003067017,
      "learning_rate": 5.170555989160507e-06,
      "loss": 1.7005,
      "step": 216767
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.5590389966964722,
      "learning_rate": 5.169799834116062e-06,
      "loss": 1.9845,
      "step": 216768
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.3903851509094238,
      "learning_rate": 5.1690437338861355e-06,
      "loss": 1.8955,
      "step": 216769
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.5590428113937378,
      "learning_rate": 5.168287688470829e-06,
      "loss": 2.1817,
      "step": 216770
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.4588978290557861,
      "learning_rate": 5.167531697870309e-06,
      "loss": 1.9756,
      "step": 216771
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.4768121242523193,
      "learning_rate": 5.166775762084707e-06,
      "loss": 1.9706,
      "step": 216772
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.5605837106704712,
      "learning_rate": 5.166019881114192e-06,
      "loss": 1.9896,
      "step": 216773
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.6903969049453735,
      "learning_rate": 5.165264054958862e-06,
      "loss": 1.922,
      "step": 216774
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.5500844717025757,
      "learning_rate": 5.1645082836188846e-06,
      "loss": 1.9184,
      "step": 216775
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.605966567993164,
      "learning_rate": 5.1637525670943595e-06,
      "loss": 2.08,
      "step": 216776
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.6693083047866821,
      "learning_rate": 5.162996905385486e-06,
      "loss": 1.9816,
      "step": 216777
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.4500892162322998,
      "learning_rate": 5.162241298492364e-06,
      "loss": 2.0307,
      "step": 216778
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.511595368385315,
      "learning_rate": 5.1614857464151616e-06,
      "loss": 1.9537,
      "step": 216779
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.5159380435943604,
      "learning_rate": 5.16073024915401e-06,
      "loss": 1.7003,
      "step": 216780
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.3129098415374756,
      "learning_rate": 5.159974806709044e-06,
      "loss": 1.9667,
      "step": 216781
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.4485445022583008,
      "learning_rate": 5.159219419080363e-06,
      "loss": 2.0208,
      "step": 216782
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.381095290184021,
      "learning_rate": 5.158464086268199e-06,
      "loss": 1.9788,
      "step": 216783
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.411273717880249,
      "learning_rate": 5.1577088082725874e-06,
      "loss": 1.9556,
      "step": 216784
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.4450249671936035,
      "learning_rate": 5.156953585093759e-06,
      "loss": 1.883,
      "step": 216785
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.5594887733459473,
      "learning_rate": 5.156198416731816e-06,
      "loss": 2.0134,
      "step": 216786
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.4898333549499512,
      "learning_rate": 5.155443303186924e-06,
      "loss": 1.7167,
      "step": 216787
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.4886211156845093,
      "learning_rate": 5.154688244459149e-06,
      "loss": 1.9607,
      "step": 216788
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.528484582901001,
      "learning_rate": 5.153933240548724e-06,
      "loss": 1.8934,
      "step": 216789
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.4810223579406738,
      "learning_rate": 5.1531782914557175e-06,
      "loss": 2.0271,
      "step": 216790
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.4459102153778076,
      "learning_rate": 5.1524233971803274e-06,
      "loss": 1.9366,
      "step": 216791
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.5788438320159912,
      "learning_rate": 5.151668557722655e-06,
      "loss": 2.0589,
      "step": 216792
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.5658166408538818,
      "learning_rate": 5.150913773082898e-06,
      "loss": 2.0186,
      "step": 216793
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.4983640909194946,
      "learning_rate": 5.150159043261093e-06,
      "loss": 2.0932,
      "step": 216794
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.4476126432418823,
      "learning_rate": 5.14940436825747e-06,
      "loss": 2.127,
      "step": 216795
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.4210435152053833,
      "learning_rate": 5.1486497480720975e-06,
      "loss": 2.1411,
      "step": 216796
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.418197512626648,
      "learning_rate": 5.147895182705175e-06,
      "loss": 1.8474,
      "step": 216797
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.5711480379104614,
      "learning_rate": 5.147140672156835e-06,
      "loss": 2.087,
      "step": 216798
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.4596518278121948,
      "learning_rate": 5.1463862164272455e-06,
      "loss": 1.8201,
      "step": 216799
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.5700619220733643,
      "learning_rate": 5.145631815516438e-06,
      "loss": 2.0041,
      "step": 216800
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.40339195728302,
      "learning_rate": 5.1448774694246794e-06,
      "loss": 2.0544,
      "step": 216801
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.514934778213501,
      "learning_rate": 5.144123178152004e-06,
      "loss": 2.0936,
      "step": 216802
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.6084895133972168,
      "learning_rate": 5.143368941698644e-06,
      "loss": 2.1528,
      "step": 216803
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.4144790172576904,
      "learning_rate": 5.1426147600646674e-06,
      "loss": 1.9277,
      "step": 216804
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.724094033241272,
      "learning_rate": 5.1418606332502725e-06,
      "loss": 2.0458,
      "step": 216805
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.3902850151062012,
      "learning_rate": 5.141106561255559e-06,
      "loss": 2.0387,
      "step": 216806
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.4107165336608887,
      "learning_rate": 5.140352544080662e-06,
      "loss": 1.7031,
      "step": 216807
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.523659586906433,
      "learning_rate": 5.139598581725746e-06,
      "loss": 2.0234,
      "step": 216808
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.6382520198822021,
      "learning_rate": 5.138844674190979e-06,
      "loss": 2.1499,
      "step": 216809
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.5494263172149658,
      "learning_rate": 5.138090821476426e-06,
      "loss": 2.139,
      "step": 216810
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.4342221021652222,
      "learning_rate": 5.137337023582322e-06,
      "loss": 2.1327,
      "step": 216811
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.359684705734253,
      "learning_rate": 5.136583280508666e-06,
      "loss": 1.972,
      "step": 216812
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.4685369729995728,
      "learning_rate": 5.135829592255758e-06,
      "loss": 2.0793,
      "step": 216813
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.405107021331787,
      "learning_rate": 5.1350759588236315e-06,
      "loss": 1.9756,
      "step": 216814
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.4679343700408936,
      "learning_rate": 5.134322380212485e-06,
      "loss": 1.8614,
      "step": 216815
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.656599760055542,
      "learning_rate": 5.133568856422387e-06,
      "loss": 2.1066,
      "step": 216816
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.4079151153564453,
      "learning_rate": 5.132815387453604e-06,
      "loss": 2.0684,
      "step": 216817
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.423746943473816,
      "learning_rate": 5.132061973306134e-06,
      "loss": 1.8983,
      "step": 216818
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.4080328941345215,
      "learning_rate": 5.131308613980178e-06,
      "loss": 1.8127,
      "step": 216819
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.3874180316925049,
      "learning_rate": 5.130555309475903e-06,
      "loss": 1.8602,
      "step": 216820
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.4102193117141724,
      "learning_rate": 5.1298020597934085e-06,
      "loss": 1.9277,
      "step": 216821
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.4731836318969727,
      "learning_rate": 5.129048864932828e-06,
      "loss": 2.0604,
      "step": 216822
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.5283174514770508,
      "learning_rate": 5.128295724894394e-06,
      "loss": 2.0168,
      "step": 216823
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.526617407798767,
      "learning_rate": 5.1275426396781064e-06,
      "loss": 1.9984,
      "step": 216824
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.525991678237915,
      "learning_rate": 5.126789609284232e-06,
      "loss": 1.9199,
      "step": 216825
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.5322294235229492,
      "learning_rate": 5.126036633712804e-06,
      "loss": 1.869,
      "step": 216826
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.534767985343933,
      "learning_rate": 5.1252837129640235e-06,
      "loss": 1.903,
      "step": 216827
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.4424208402633667,
      "learning_rate": 5.124530847038022e-06,
      "loss": 2.1056,
      "step": 216828
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.4006177186965942,
      "learning_rate": 5.123778035934967e-06,
      "loss": 1.9212,
      "step": 216829
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.4017086029052734,
      "learning_rate": 5.123025279654924e-06,
      "loss": 1.764,
      "step": 216830
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.4148774147033691,
      "learning_rate": 5.122272578198128e-06,
      "loss": 1.9792,
      "step": 216831
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.3884668350219727,
      "learning_rate": 5.1215199315646106e-06,
      "loss": 2.1137,
      "step": 216832
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.4542824029922485,
      "learning_rate": 5.120767339754606e-06,
      "loss": 2.0135,
      "step": 216833
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.4825501441955566,
      "learning_rate": 5.1200148027681795e-06,
      "loss": 1.9745,
      "step": 216834
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.387608289718628,
      "learning_rate": 5.119262320605533e-06,
      "loss": 1.9903,
      "step": 216835
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.6248209476470947,
      "learning_rate": 5.118509893266798e-06,
      "loss": 1.9686,
      "step": 216836
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.9509305953979492,
      "learning_rate": 5.117757520752108e-06,
      "loss": 1.8978,
      "step": 216837
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.4926044940948486,
      "learning_rate": 5.117005203061531e-06,
      "loss": 2.0088,
      "step": 216838
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.5684536695480347,
      "learning_rate": 5.116252940195331e-06,
      "loss": 1.9426,
      "step": 216839
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.6064794063568115,
      "learning_rate": 5.1155007321535435e-06,
      "loss": 1.989,
      "step": 216840
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.4395856857299805,
      "learning_rate": 5.114748578936367e-06,
      "loss": 2.1046,
      "step": 216841
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.435165524482727,
      "learning_rate": 5.113996480543936e-06,
      "loss": 2.0978,
      "step": 216842
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.4390642642974854,
      "learning_rate": 5.113244436976383e-06,
      "loss": 1.9533,
      "step": 216843
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.3846405744552612,
      "learning_rate": 5.112492448233807e-06,
      "loss": 2.1309,
      "step": 216844
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.4335826635360718,
      "learning_rate": 5.111740514316442e-06,
      "loss": 1.9123,
      "step": 216845
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.4830814599990845,
      "learning_rate": 5.1109886352243225e-06,
      "loss": 2.0249,
      "step": 216846
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.5649504661560059,
      "learning_rate": 5.110236810957646e-06,
      "loss": 2.0304,
      "step": 216847
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.4243730306625366,
      "learning_rate": 5.109485041516548e-06,
      "loss": 1.9538,
      "step": 216848
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.3884235620498657,
      "learning_rate": 5.10873332690116e-06,
      "loss": 2.2386,
      "step": 216849
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.447313666343689,
      "learning_rate": 5.107981667111649e-06,
      "loss": 2.0204,
      "step": 216850
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.68966805934906,
      "learning_rate": 5.1072300621481156e-06,
      "loss": 1.9624,
      "step": 216851
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.3870086669921875,
      "learning_rate": 5.106478512010692e-06,
      "loss": 2.0951,
      "step": 216852
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.3984923362731934,
      "learning_rate": 5.105727016699546e-06,
      "loss": 1.6468,
      "step": 216853
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.617287278175354,
      "learning_rate": 5.104975576214843e-06,
      "loss": 1.9481,
      "step": 216854
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.586242437362671,
      "learning_rate": 5.10422419055665e-06,
      "loss": 1.9507,
      "step": 216855
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.3973045349121094,
      "learning_rate": 5.103472859725199e-06,
      "loss": 1.9717,
      "step": 216856
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.5158530473709106,
      "learning_rate": 5.102721583720526e-06,
      "loss": 2.013,
      "step": 216857
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.473795771598816,
      "learning_rate": 5.101970362542862e-06,
      "loss": 2.1393,
      "step": 216858
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.3442599773406982,
      "learning_rate": 5.101219196192308e-06,
      "loss": 2.0641,
      "step": 216859
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.415751576423645,
      "learning_rate": 5.100468084668996e-06,
      "loss": 2.0208,
      "step": 216860
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.5306979417800903,
      "learning_rate": 5.09971702797306e-06,
      "loss": 1.9584,
      "step": 216861
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.3408046960830688,
      "learning_rate": 5.098966026104667e-06,
      "loss": 1.9352,
      "step": 216862
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.4199579954147339,
      "learning_rate": 5.098215079063916e-06,
      "loss": 2.0162,
      "step": 216863
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.4757442474365234,
      "learning_rate": 5.097464186851008e-06,
      "loss": 1.8721,
      "step": 216864
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.6052643060684204,
      "learning_rate": 5.096713349466042e-06,
      "loss": 1.9695,
      "step": 216865
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.5127750635147095,
      "learning_rate": 5.095962566909184e-06,
      "loss": 2.2957,
      "step": 216866
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.5189226865768433,
      "learning_rate": 5.095211839180502e-06,
      "loss": 2.0652,
      "step": 216867
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.4403846263885498,
      "learning_rate": 5.094461166280228e-06,
      "loss": 1.8627,
      "step": 216868
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.3128966093063354,
      "learning_rate": 5.093710548208429e-06,
      "loss": 1.9634,
      "step": 216869
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.5729550123214722,
      "learning_rate": 5.092959984965273e-06,
      "loss": 1.8462,
      "step": 216870
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.4624675512313843,
      "learning_rate": 5.092209476550924e-06,
      "loss": 1.9421,
      "step": 216871
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.4330763816833496,
      "learning_rate": 5.091459022965516e-06,
      "loss": 2.0387,
      "step": 216872
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.437301754951477,
      "learning_rate": 5.0907086242091496e-06,
      "loss": 2.0004,
      "step": 216873
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.4010552167892456,
      "learning_rate": 5.0899582802819915e-06,
      "loss": 2.1369,
      "step": 216874
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.51382577419281,
      "learning_rate": 5.089207991184174e-06,
      "loss": 1.9583,
      "step": 216875
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.476564645767212,
      "learning_rate": 5.088457756915832e-06,
      "loss": 2.0323,
      "step": 216876
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.416616678237915,
      "learning_rate": 5.087707577477129e-06,
      "loss": 1.972,
      "step": 216877
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.6877236366271973,
      "learning_rate": 5.086957452868201e-06,
      "loss": 2.1283,
      "step": 216878
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.3938299417495728,
      "learning_rate": 5.086207383089147e-06,
      "loss": 1.834,
      "step": 216879
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.4144105911254883,
      "learning_rate": 5.085457368140133e-06,
      "loss": 2.0858,
      "step": 216880
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.6359080076217651,
      "learning_rate": 5.084707408021293e-06,
      "loss": 2.0941,
      "step": 216881
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.4970825910568237,
      "learning_rate": 5.083957502732827e-06,
      "loss": 1.9465,
      "step": 216882
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.4412156343460083,
      "learning_rate": 5.0832076522747674e-06,
      "loss": 1.8902,
      "step": 216883
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.436165452003479,
      "learning_rate": 5.082457856647349e-06,
      "loss": 1.9313,
      "step": 216884
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.5895060300827026,
      "learning_rate": 5.081708115850602e-06,
      "loss": 2.0874,
      "step": 216885
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.2846996784210205,
      "learning_rate": 5.080958429884796e-06,
      "loss": 1.8856,
      "step": 216886
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.5518183708190918,
      "learning_rate": 5.080208798749963e-06,
      "loss": 1.8675,
      "step": 216887
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.5457395315170288,
      "learning_rate": 5.079459222446302e-06,
      "loss": 2.1127,
      "step": 216888
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.468469262123108,
      "learning_rate": 5.078709700973948e-06,
      "loss": 1.8597,
      "step": 216889
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.564551830291748,
      "learning_rate": 5.077960234333034e-06,
      "loss": 1.8268,
      "step": 216890
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.4124563932418823,
      "learning_rate": 5.0772108225236586e-06,
      "loss": 2.2072,
      "step": 216891
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.5875788927078247,
      "learning_rate": 5.076461465546056e-06,
      "loss": 1.838,
      "step": 216892
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.3947081565856934,
      "learning_rate": 5.075712163400225e-06,
      "loss": 1.8796,
      "step": 216893
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.441819190979004,
      "learning_rate": 5.074962916086467e-06,
      "loss": 1.9105,
      "step": 216894
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.6251896619796753,
      "learning_rate": 5.074213723604781e-06,
      "loss": 2.1075,
      "step": 216895
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.4398796558380127,
      "learning_rate": 5.073464585955433e-06,
      "loss": 1.9512,
      "step": 216896
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.4877532720565796,
      "learning_rate": 5.072715503138424e-06,
      "loss": 2.1659,
      "step": 216897
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.3013815879821777,
      "learning_rate": 5.0719664751539876e-06,
      "loss": 1.9939,
      "step": 216898
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.5389049053192139,
      "learning_rate": 5.071217502002223e-06,
      "loss": 1.8374,
      "step": 216899
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.4767858982086182,
      "learning_rate": 5.0704685836833295e-06,
      "loss": 1.7408,
      "step": 216900
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.4405754804611206,
      "learning_rate": 5.069719720197374e-06,
      "loss": 2.076,
      "step": 216901
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.455246925354004,
      "learning_rate": 5.068970911544557e-06,
      "loss": 1.8953,
      "step": 216902
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.563592553138733,
      "learning_rate": 5.068222157724944e-06,
      "loss": 2.0516,
      "step": 216903
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.4419875144958496,
      "learning_rate": 5.067473458738736e-06,
      "loss": 1.9559,
      "step": 216904
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.357623815536499,
      "learning_rate": 5.066724814586031e-06,
      "loss": 2.0856,
      "step": 216905
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.416164517402649,
      "learning_rate": 5.065976225267032e-06,
      "loss": 1.98,
      "step": 216906
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.4296631813049316,
      "learning_rate": 5.06522769078177e-06,
      "loss": 1.809,
      "step": 216907
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.446470022201538,
      "learning_rate": 5.064479211130545e-06,
      "loss": 1.8199,
      "step": 216908
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.5196810960769653,
      "learning_rate": 5.063730786313325e-06,
      "loss": 1.7895,
      "step": 216909
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.4692535400390625,
      "learning_rate": 5.062982416330341e-06,
      "loss": 1.8656,
      "step": 216910
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.4808319807052612,
      "learning_rate": 5.062234101181695e-06,
      "loss": 2.0469,
      "step": 216911
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.380707025527954,
      "learning_rate": 5.061485840867585e-06,
      "loss": 1.9437,
      "step": 216912
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.3712811470031738,
      "learning_rate": 5.060737635388079e-06,
      "loss": 1.7187,
      "step": 216913
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.6467832326889038,
      "learning_rate": 5.05998948474341e-06,
      "loss": 2.045,
      "step": 216914
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.3296608924865723,
      "learning_rate": 5.05924138893361e-06,
      "loss": 2.095,
      "step": 216915
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.4274364709854126,
      "learning_rate": 5.058493347958881e-06,
      "loss": 2.0212,
      "step": 216916
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.5587011575698853,
      "learning_rate": 5.057745361819321e-06,
      "loss": 2.0067,
      "step": 216917
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.465887188911438,
      "learning_rate": 5.056997430515097e-06,
      "loss": 1.9696,
      "step": 216918
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.517090082168579,
      "learning_rate": 5.056249554046344e-06,
      "loss": 1.9946,
      "step": 216919
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.3539177179336548,
      "learning_rate": 5.055501732413225e-06,
      "loss": 1.9725,
      "step": 216920
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.4266705513000488,
      "learning_rate": 5.054753965615843e-06,
      "loss": 1.9368,
      "step": 216921
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.4309369325637817,
      "learning_rate": 5.054006253654363e-06,
      "loss": 1.8266,
      "step": 216922
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.657273530960083,
      "learning_rate": 5.053258596528886e-06,
      "loss": 2.0522,
      "step": 216923
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.4995282888412476,
      "learning_rate": 5.052510994239611e-06,
      "loss": 2.0704,
      "step": 216924
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.4450401067733765,
      "learning_rate": 5.051763446786572e-06,
      "loss": 2.0719,
      "step": 216925
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.4751231670379639,
      "learning_rate": 5.051015954170035e-06,
      "loss": 2.0228,
      "step": 216926
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.4898370504379272,
      "learning_rate": 5.050268516390099e-06,
      "loss": 1.8578,
      "step": 216927
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.46755051612854,
      "learning_rate": 5.049521133446866e-06,
      "loss": 1.7523,
      "step": 216928
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.7186042070388794,
      "learning_rate": 5.048773805340467e-06,
      "loss": 1.8498,
      "step": 216929
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.5385981798171997,
      "learning_rate": 5.048026532071103e-06,
      "loss": 1.9396,
      "step": 216930
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.4828670024871826,
      "learning_rate": 5.047279313638841e-06,
      "loss": 1.8667,
      "step": 216931
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.4583302736282349,
      "learning_rate": 5.04653215004388e-06,
      "loss": 1.8733,
      "step": 216932
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.438775658607483,
      "learning_rate": 5.045785041286354e-06,
      "loss": 2.0211,
      "step": 216933
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.5757321119308472,
      "learning_rate": 5.04503798736633e-06,
      "loss": 1.9021,
      "step": 216934
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.4398196935653687,
      "learning_rate": 5.044290988284072e-06,
      "loss": 2.0023,
      "step": 216935
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.556551218032837,
      "learning_rate": 5.043544044039616e-06,
      "loss": 1.9659,
      "step": 216936
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.3612128496170044,
      "learning_rate": 5.0427971546330935e-06,
      "loss": 1.7154,
      "step": 216937
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.5034185647964478,
      "learning_rate": 5.0420503200647385e-06,
      "loss": 2.2614,
      "step": 216938
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.492946982383728,
      "learning_rate": 5.041303540334618e-06,
      "loss": 1.9661,
      "step": 216939
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.3665677309036255,
      "learning_rate": 5.040556815442865e-06,
      "loss": 1.7999,
      "step": 216940
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.4876270294189453,
      "learning_rate": 5.039810145389678e-06,
      "loss": 2.022,
      "step": 216941
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.4223661422729492,
      "learning_rate": 5.039063530175158e-06,
      "loss": 1.9136,
      "step": 216942
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.425013780593872,
      "learning_rate": 5.038316969799405e-06,
      "loss": 2.1176,
      "step": 216943
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.5486488342285156,
      "learning_rate": 5.0375704642626195e-06,
      "loss": 1.9563,
      "step": 216944
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.6110594272613525,
      "learning_rate": 5.036824013564933e-06,
      "loss": 1.9459,
      "step": 216945
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.4509880542755127,
      "learning_rate": 5.036077617706446e-06,
      "loss": 1.9094,
      "step": 216946
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.5702053308486938,
      "learning_rate": 5.035331276687327e-06,
      "loss": 1.9518,
      "step": 216947
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.6167927980422974,
      "learning_rate": 5.034584990507673e-06,
      "loss": 2.0461,
      "step": 216948
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.6768767833709717,
      "learning_rate": 5.033838759167719e-06,
      "loss": 1.8383,
      "step": 216949
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.404403567314148,
      "learning_rate": 5.033092582667497e-06,
      "loss": 1.8516,
      "step": 216950
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.41716468334198,
      "learning_rate": 5.032346461007242e-06,
      "loss": 2.0174,
      "step": 216951
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.3708117008209229,
      "learning_rate": 5.031600394186986e-06,
      "loss": 1.7126,
      "step": 216952
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.6721739768981934,
      "learning_rate": 5.030854382206961e-06,
      "loss": 2.1189,
      "step": 216953
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.562381625175476,
      "learning_rate": 5.0301084250672365e-06,
      "loss": 1.9405,
      "step": 216954
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.427976369857788,
      "learning_rate": 5.029362522767977e-06,
      "loss": 1.8535,
      "step": 216955
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.4998847246170044,
      "learning_rate": 5.028616675309349e-06,
      "loss": 1.8928,
      "step": 216956
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.488295555114746,
      "learning_rate": 5.027870882691487e-06,
      "loss": 2.0671,
      "step": 216957
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.7179903984069824,
      "learning_rate": 5.02712514491449e-06,
      "loss": 2.0957,
      "step": 216958
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.3314374685287476,
      "learning_rate": 5.0263794619785244e-06,
      "loss": 1.8111,
      "step": 216959
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.4975215196609497,
      "learning_rate": 5.025633833883724e-06,
      "loss": 1.8103,
      "step": 216960
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.3448491096496582,
      "learning_rate": 5.024888260630222e-06,
      "loss": 2.1999,
      "step": 216961
    },
    {
      "epoch": 2.82,
      "grad_norm": 1.516342282295227,
      "learning_rate": 5.024142742218151e-06,
      "loss": 1.9831,
      "step": 216962
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.4192724227905273,
      "learning_rate": 5.023397278647678e-06,
      "loss": 1.8423,
      "step": 216963
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.467949390411377,
      "learning_rate": 5.022651869918904e-06,
      "loss": 2.0091,
      "step": 216964
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.5173805952072144,
      "learning_rate": 5.021906516032026e-06,
      "loss": 1.8606,
      "step": 216965
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.4608765840530396,
      "learning_rate": 5.02116121698708e-06,
      "loss": 1.9136,
      "step": 216966
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.4665547609329224,
      "learning_rate": 5.020415972784331e-06,
      "loss": 2.0081,
      "step": 216967
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.5965173244476318,
      "learning_rate": 5.019670783423813e-06,
      "loss": 2.0052,
      "step": 216968
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.5206644535064697,
      "learning_rate": 5.018925648905758e-06,
      "loss": 2.0375,
      "step": 216969
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.4326448440551758,
      "learning_rate": 5.018180569230201e-06,
      "loss": 2.1229,
      "step": 216970
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.51836097240448,
      "learning_rate": 5.017435544397341e-06,
      "loss": 1.8477,
      "step": 216971
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.5715614557266235,
      "learning_rate": 5.0166905744073115e-06,
      "loss": 1.8561,
      "step": 216972
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.5064811706542969,
      "learning_rate": 5.015945659260279e-06,
      "loss": 1.8127,
      "step": 216973
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.3961302042007446,
      "learning_rate": 5.015200798956276e-06,
      "loss": 2.0681,
      "step": 216974
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.4041157960891724,
      "learning_rate": 5.014455993495603e-06,
      "loss": 1.9271,
      "step": 216975
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.5521408319473267,
      "learning_rate": 5.01371124287826e-06,
      "loss": 1.9245,
      "step": 216976
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.462923526763916,
      "learning_rate": 5.012966547104447e-06,
      "loss": 1.8956,
      "step": 216977
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.3875586986541748,
      "learning_rate": 5.012221906174296e-06,
      "loss": 1.983,
      "step": 216978
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.473729133605957,
      "learning_rate": 5.0114773200879425e-06,
      "loss": 2.0022,
      "step": 216979
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.5266876220703125,
      "learning_rate": 5.010732788845484e-06,
      "loss": 2.0588,
      "step": 216980
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.3975850343704224,
      "learning_rate": 5.009988312447189e-06,
      "loss": 1.9869,
      "step": 216981
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.3368260860443115,
      "learning_rate": 5.009243890893022e-06,
      "loss": 1.9664,
      "step": 216982
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.450654149055481,
      "learning_rate": 5.008499524183251e-06,
      "loss": 1.865,
      "step": 216983
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.3851432800292969,
      "learning_rate": 5.0077552123179434e-06,
      "loss": 1.9952,
      "step": 216984
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.5028672218322754,
      "learning_rate": 5.007010955297263e-06,
      "loss": 2.1048,
      "step": 216985
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.4999727010726929,
      "learning_rate": 5.006266753121347e-06,
      "loss": 2.0206,
      "step": 216986
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.4327456951141357,
      "learning_rate": 5.0055226057903575e-06,
      "loss": 1.8451,
      "step": 216987
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.4021025896072388,
      "learning_rate": 5.004778513304397e-06,
      "loss": 1.9283,
      "step": 216988
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.490187406539917,
      "learning_rate": 5.004034475663598e-06,
      "loss": 2.0784,
      "step": 216989
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.4918845891952515,
      "learning_rate": 5.003290492868128e-06,
      "loss": 2.1474,
      "step": 216990
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.4767206907272339,
      "learning_rate": 5.002546564918119e-06,
      "loss": 2.2774,
      "step": 216991
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.5619374513626099,
      "learning_rate": 5.001802691813705e-06,
      "loss": 2.1524,
      "step": 216992
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.465903878211975,
      "learning_rate": 5.00105887355502e-06,
      "loss": 1.8661,
      "step": 216993
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.4440072774887085,
      "learning_rate": 5.000315110142228e-06,
      "loss": 1.8509,
      "step": 216994
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.4816067218780518,
      "learning_rate": 4.9995714015754304e-06,
      "loss": 1.9046,
      "step": 216995
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.607937216758728,
      "learning_rate": 4.9988277478547616e-06,
      "loss": 2.1006,
      "step": 216996
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.4554047584533691,
      "learning_rate": 4.998084148980419e-06,
      "loss": 2.0182,
      "step": 216997
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.473935604095459,
      "learning_rate": 4.9973406049524715e-06,
      "loss": 1.7631,
      "step": 216998
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.4669132232666016,
      "learning_rate": 4.996597115771084e-06,
      "loss": 2.0265,
      "step": 216999
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.4138849973678589,
      "learning_rate": 4.9958536814364235e-06,
      "loss": 1.9964,
      "step": 217000
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.456408143043518,
      "learning_rate": 4.99511030194859e-06,
      "loss": 2.186,
      "step": 217001
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.6129193305969238,
      "learning_rate": 4.994366977307718e-06,
      "loss": 2.0947,
      "step": 217002
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.4345015287399292,
      "learning_rate": 4.9936237075140055e-06,
      "loss": 1.8981,
      "step": 217003
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.409531593322754,
      "learning_rate": 4.992880492567486e-06,
      "loss": 1.9755,
      "step": 217004
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.557326316833496,
      "learning_rate": 4.9921373324683935e-06,
      "loss": 2.068,
      "step": 217005
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.4893097877502441,
      "learning_rate": 4.9913942272168604e-06,
      "loss": 2.0984,
      "step": 217006
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.4977596998214722,
      "learning_rate": 4.9906511768129545e-06,
      "loss": 1.8614,
      "step": 217007
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.2162808179855347,
      "learning_rate": 4.989908181256841e-06,
      "loss": 1.6906,
      "step": 217008
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.6264867782592773,
      "learning_rate": 4.98916524054872e-06,
      "loss": 2.0406,
      "step": 217009
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.4034748077392578,
      "learning_rate": 4.988422354688626e-06,
      "loss": 1.8769,
      "step": 217010
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.4262512922286987,
      "learning_rate": 4.987679523676824e-06,
      "loss": 1.9356,
      "step": 217011
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.4056366682052612,
      "learning_rate": 4.986936747513348e-06,
      "loss": 1.8463,
      "step": 217012
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.4675201177597046,
      "learning_rate": 4.986194026198365e-06,
      "loss": 1.9825,
      "step": 217013
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.5832879543304443,
      "learning_rate": 4.985451359732007e-06,
      "loss": 1.8608,
      "step": 217014
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.4685944318771362,
      "learning_rate": 4.984708748114441e-06,
      "loss": 2.1536,
      "step": 217015
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.4805399179458618,
      "learning_rate": 4.983966191345734e-06,
      "loss": 2.1237,
      "step": 217016
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.4660074710845947,
      "learning_rate": 4.983223689426152e-06,
      "loss": 1.9244,
      "step": 217017
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.4298577308654785,
      "learning_rate": 4.982481242355729e-06,
      "loss": 2.1269,
      "step": 217018
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.5113517045974731,
      "learning_rate": 4.981738850134598e-06,
      "loss": 1.9981,
      "step": 217019
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.3881690502166748,
      "learning_rate": 4.9809965127629585e-06,
      "loss": 2.1155,
      "step": 217020
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.5644904375076294,
      "learning_rate": 4.980254230240943e-06,
      "loss": 1.848,
      "step": 217021
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.592483639717102,
      "learning_rate": 4.97951200256862e-06,
      "loss": 1.9066,
      "step": 217022
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.5177546739578247,
      "learning_rate": 4.978769829746221e-06,
      "loss": 2.0957,
      "step": 217023
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.5291059017181396,
      "learning_rate": 4.978027711773814e-06,
      "loss": 1.9563,
      "step": 217024
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.4741146564483643,
      "learning_rate": 4.977285648651563e-06,
      "loss": 2.0478,
      "step": 217025
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.4147018194198608,
      "learning_rate": 4.976543640379604e-06,
      "loss": 1.9932,
      "step": 217026
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.5863759517669678,
      "learning_rate": 4.975801686958103e-06,
      "loss": 2.2254,
      "step": 217027
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.553471326828003,
      "learning_rate": 4.975059788387126e-06,
      "loss": 1.9634,
      "step": 217028
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.4204350709915161,
      "learning_rate": 4.974317944666872e-06,
      "loss": 2.0212,
      "step": 217029
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.6918286085128784,
      "learning_rate": 4.9735761557974766e-06,
      "loss": 2.101,
      "step": 217030
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.4067128896713257,
      "learning_rate": 4.972834421779004e-06,
      "loss": 1.9716,
      "step": 217031
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.428175449371338,
      "learning_rate": 4.972092742611722e-06,
      "loss": 2.0407,
      "step": 217032
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.6644991636276245,
      "learning_rate": 4.97135111829563e-06,
      "loss": 2.0148,
      "step": 217033
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.4354281425476074,
      "learning_rate": 4.970609548830995e-06,
      "loss": 1.9599,
      "step": 217034
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.4574363231658936,
      "learning_rate": 4.9698680342178835e-06,
      "loss": 1.9474,
      "step": 217035
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.5307072401046753,
      "learning_rate": 4.9691265744564276e-06,
      "loss": 2.1714,
      "step": 217036
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.459906816482544,
      "learning_rate": 4.968385169546763e-06,
      "loss": 1.8968,
      "step": 217037
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.3850921392440796,
      "learning_rate": 4.9676438194890534e-06,
      "loss": 2.0788,
      "step": 217038
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.5633668899536133,
      "learning_rate": 4.966902524283433e-06,
      "loss": 1.9921,
      "step": 217039
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.5014768838882446,
      "learning_rate": 4.9661612839300365e-06,
      "loss": 1.8237,
      "step": 217040
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.4667288064956665,
      "learning_rate": 4.96542009842903e-06,
      "loss": 1.9004,
      "step": 217041
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.4359880685806274,
      "learning_rate": 4.964678967780478e-06,
      "loss": 2.086,
      "step": 217042
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.4807593822479248,
      "learning_rate": 4.963937891984582e-06,
      "loss": 1.8812,
      "step": 217043
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.4425311088562012,
      "learning_rate": 4.9631968710414415e-06,
      "loss": 2.1872,
      "step": 217044
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.3997809886932373,
      "learning_rate": 4.9624559049512236e-06,
      "loss": 1.8759,
      "step": 217045
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.3075965642929077,
      "learning_rate": 4.961714993714061e-06,
      "loss": 1.9645,
      "step": 217046
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.3135664463043213,
      "learning_rate": 4.960974137330054e-06,
      "loss": 2.0957,
      "step": 217047
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.4359830617904663,
      "learning_rate": 4.960233335799435e-06,
      "loss": 1.9741,
      "step": 217048
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.359338402748108,
      "learning_rate": 4.959492589122205e-06,
      "loss": 1.7518,
      "step": 217049
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.5468724966049194,
      "learning_rate": 4.958751897298629e-06,
      "loss": 1.8182,
      "step": 217050
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.4458087682724,
      "learning_rate": 4.958011260328742e-06,
      "loss": 2.068,
      "step": 217051
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.4491342306137085,
      "learning_rate": 4.957270678212777e-06,
      "loss": 1.9467,
      "step": 217052
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.4416121244430542,
      "learning_rate": 4.956530150950799e-06,
      "loss": 2.0289,
      "step": 217053
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.3296059370040894,
      "learning_rate": 4.955789678543009e-06,
      "loss": 2.1464,
      "step": 217054
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.3538999557495117,
      "learning_rate": 4.955049260989441e-06,
      "loss": 1.9387,
      "step": 217055
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.4648993015289307,
      "learning_rate": 4.95430889829036e-06,
      "loss": 1.8096,
      "step": 217056
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.5081754922866821,
      "learning_rate": 4.9535685904458e-06,
      "loss": 2.0927,
      "step": 217057
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.4756121635437012,
      "learning_rate": 4.95282833745596e-06,
      "loss": 2.0687,
      "step": 217058
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.4603146314620972,
      "learning_rate": 4.952088139320942e-06,
      "loss": 1.7563,
      "step": 217059
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.4770114421844482,
      "learning_rate": 4.951347996040911e-06,
      "loss": 1.5669,
      "step": 217060
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.4345430135726929,
      "learning_rate": 4.950607907616e-06,
      "loss": 2.0317,
      "step": 217061
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.5353803634643555,
      "learning_rate": 4.949867874046343e-06,
      "loss": 2.1278,
      "step": 217062
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.4651203155517578,
      "learning_rate": 4.949127895332072e-06,
      "loss": 2.0439,
      "step": 217063
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.5313289165496826,
      "learning_rate": 4.9483879714733224e-06,
      "loss": 1.7383,
      "step": 217064
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.5873711109161377,
      "learning_rate": 4.947648102470225e-06,
      "loss": 2.3038,
      "step": 217065
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.4525073766708374,
      "learning_rate": 4.946908288322949e-06,
      "loss": 1.9879,
      "step": 217066
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.4349507093429565,
      "learning_rate": 4.946168529031625e-06,
      "loss": 2.0469,
      "step": 217067
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.4342875480651855,
      "learning_rate": 4.945428824596353e-06,
      "loss": 2.1497,
      "step": 217068
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.4010674953460693,
      "learning_rate": 4.944689175017302e-06,
      "loss": 2.097,
      "step": 217069
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.4224932193756104,
      "learning_rate": 4.943949580294604e-06,
      "loss": 1.9533,
      "step": 217070
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.483957052230835,
      "learning_rate": 4.9432100404283585e-06,
      "loss": 1.9219,
      "step": 217071
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.6948806047439575,
      "learning_rate": 4.942470555418798e-06,
      "loss": 1.745,
      "step": 217072
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.5001745223999023,
      "learning_rate": 4.941731125265991e-06,
      "loss": 1.9524,
      "step": 217073
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.6283379793167114,
      "learning_rate": 4.940991749970069e-06,
      "loss": 2.0837,
      "step": 217074
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.4231213331222534,
      "learning_rate": 4.9402524295312e-06,
      "loss": 1.9226,
      "step": 217075
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.5324392318725586,
      "learning_rate": 4.939513163949482e-06,
      "loss": 2.1227,
      "step": 217076
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.4133156538009644,
      "learning_rate": 4.938773953225084e-06,
      "loss": 1.9432,
      "step": 217077
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.492253303527832,
      "learning_rate": 4.93803479735817e-06,
      "loss": 1.8264,
      "step": 217078
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.3713387250900269,
      "learning_rate": 4.937295696348809e-06,
      "loss": 1.8131,
      "step": 217079
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.4506970643997192,
      "learning_rate": 4.936556650197199e-06,
      "loss": 1.8289,
      "step": 217080
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.4955517053604126,
      "learning_rate": 4.93581765890344e-06,
      "loss": 1.9669,
      "step": 217081
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.4312433004379272,
      "learning_rate": 4.9350787224677e-06,
      "loss": 1.794,
      "step": 217082
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.7227625846862793,
      "learning_rate": 4.934339840890078e-06,
      "loss": 1.8865,
      "step": 217083
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.4830769300460815,
      "learning_rate": 4.933601014170741e-06,
      "loss": 1.974,
      "step": 217084
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.4595636129379272,
      "learning_rate": 4.932862242309821e-06,
      "loss": 2.0503,
      "step": 217085
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.548857569694519,
      "learning_rate": 4.932123525307452e-06,
      "loss": 2.0011,
      "step": 217086
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.2787566184997559,
      "learning_rate": 4.931384863163735e-06,
      "loss": 1.7772,
      "step": 217087
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.399845838546753,
      "learning_rate": 4.9306462558789005e-06,
      "loss": 1.9114,
      "step": 217088
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.4768171310424805,
      "learning_rate": 4.9299077034529844e-06,
      "loss": 1.9323,
      "step": 217089
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.536516547203064,
      "learning_rate": 4.9291692058861855e-06,
      "loss": 2.1094,
      "step": 217090
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.4729266166687012,
      "learning_rate": 4.928430763178637e-06,
      "loss": 2.0726,
      "step": 217091
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.5692821741104126,
      "learning_rate": 4.927692375330439e-06,
      "loss": 1.9609,
      "step": 217092
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.565153956413269,
      "learning_rate": 4.926954042341758e-06,
      "loss": 2.2901,
      "step": 217093
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.7460802793502808,
      "learning_rate": 4.926215764212727e-06,
      "loss": 1.9957,
      "step": 217094
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.4788686037063599,
      "learning_rate": 4.925477540943479e-06,
      "loss": 1.9842,
      "step": 217095
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.5408519506454468,
      "learning_rate": 4.924739372534181e-06,
      "loss": 2.0963,
      "step": 217096
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.512096881866455,
      "learning_rate": 4.924001258984933e-06,
      "loss": 1.8847,
      "step": 217097
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.5899676084518433,
      "learning_rate": 4.923263200295868e-06,
      "loss": 1.8047,
      "step": 217098
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.4624496698379517,
      "learning_rate": 4.922525196467153e-06,
      "loss": 2.0372,
      "step": 217099
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.466213345527649,
      "learning_rate": 4.92178724749892e-06,
      "loss": 1.8676,
      "step": 217100
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.4612421989440918,
      "learning_rate": 4.9210493533912375e-06,
      "loss": 2.003,
      "step": 217101
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.407241702079773,
      "learning_rate": 4.92031151414437e-06,
      "loss": 2.0375,
      "step": 217102
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.6257153749465942,
      "learning_rate": 4.919573729758353e-06,
      "loss": 2.1906,
      "step": 217103
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.4230526685714722,
      "learning_rate": 4.918836000233383e-06,
      "loss": 2.1252,
      "step": 217104
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.4004485607147217,
      "learning_rate": 4.918098325569531e-06,
      "loss": 1.6465,
      "step": 217105
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.4497863054275513,
      "learning_rate": 4.917360705767026e-06,
      "loss": 1.8816,
      "step": 217106
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.6336160898208618,
      "learning_rate": 4.916623140825904e-06,
      "loss": 1.9621,
      "step": 217107
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.6631240844726562,
      "learning_rate": 4.9158856307463965e-06,
      "loss": 2.1392,
      "step": 217108
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.4717117547988892,
      "learning_rate": 4.915148175528571e-06,
      "loss": 2.3364,
      "step": 217109
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.3842191696166992,
      "learning_rate": 4.9144107751725615e-06,
      "loss": 1.8378,
      "step": 217110
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.4644170999526978,
      "learning_rate": 4.9136734296786e-06,
      "loss": 1.8867,
      "step": 217111
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.4941353797912598,
      "learning_rate": 4.912936139046719e-06,
      "loss": 1.9892,
      "step": 217112
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.5937222242355347,
      "learning_rate": 4.912198903277087e-06,
      "loss": 2.0749,
      "step": 217113
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.5281503200531006,
      "learning_rate": 4.911461722369836e-06,
      "loss": 2.3143,
      "step": 217114
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.3815934658050537,
      "learning_rate": 4.910724596325166e-06,
      "loss": 1.9472,
      "step": 217115
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.3338791131973267,
      "learning_rate": 4.909987525143111e-06,
      "loss": 1.9299,
      "step": 217116
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.4756402969360352,
      "learning_rate": 4.90925050882387e-06,
      "loss": 1.9724,
      "step": 217117
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.3029202222824097,
      "learning_rate": 4.9085135473675765e-06,
      "loss": 1.6836,
      "step": 217118
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.4393980503082275,
      "learning_rate": 4.907776640774363e-06,
      "loss": 2.0296,
      "step": 217119
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.3626269102096558,
      "learning_rate": 4.907039789044365e-06,
      "loss": 2.1235,
      "step": 217120
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.5682876110076904,
      "learning_rate": 4.906302992177746e-06,
      "loss": 1.8113,
      "step": 217121
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.4869118928909302,
      "learning_rate": 4.9055662501745755e-06,
      "loss": 1.8626,
      "step": 217122
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.4555470943450928,
      "learning_rate": 4.904829563035051e-06,
      "loss": 2.078,
      "step": 217123
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.382950782775879,
      "learning_rate": 4.90409293075924e-06,
      "loss": 2.1793,
      "step": 217124
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.4738976955413818,
      "learning_rate": 4.9033563533473766e-06,
      "loss": 1.9571,
      "step": 217125
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.4848201274871826,
      "learning_rate": 4.902619830799559e-06,
      "loss": 2.0648,
      "step": 217126
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.6092628240585327,
      "learning_rate": 4.901883363115888e-06,
      "loss": 2.1576,
      "step": 217127
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.4598822593688965,
      "learning_rate": 4.901146950296531e-06,
      "loss": 1.7873,
      "step": 217128
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.3591828346252441,
      "learning_rate": 4.9004105923416524e-06,
      "loss": 1.9736,
      "step": 217129
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.4905563592910767,
      "learning_rate": 4.899674289251321e-06,
      "loss": 2.0175,
      "step": 217130
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.6200371980667114,
      "learning_rate": 4.8989380410257016e-06,
      "loss": 2.1111,
      "step": 217131
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.48259699344635,
      "learning_rate": 4.898201847664962e-06,
      "loss": 1.9148,
      "step": 217132
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.631319522857666,
      "learning_rate": 4.897465709169235e-06,
      "loss": 2.138,
      "step": 217133
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.4260659217834473,
      "learning_rate": 4.89672962553862e-06,
      "loss": 2.0054,
      "step": 217134
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.5023703575134277,
      "learning_rate": 4.895993596773285e-06,
      "loss": 1.9045,
      "step": 217135
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.5300034284591675,
      "learning_rate": 4.8952576228733276e-06,
      "loss": 2.0367,
      "step": 217136
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.5093557834625244,
      "learning_rate": 4.894521703838916e-06,
      "loss": 1.7393,
      "step": 217137
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.4537159204483032,
      "learning_rate": 4.893785839670183e-06,
      "loss": 2.0424,
      "step": 217138
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.6549724340438843,
      "learning_rate": 4.8930500303672955e-06,
      "loss": 2.0502,
      "step": 217139
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.4984996318817139,
      "learning_rate": 4.892314275930354e-06,
      "loss": 2.0248,
      "step": 217140
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.551476001739502,
      "learning_rate": 4.891578576359489e-06,
      "loss": 2.0978,
      "step": 217141
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.5108469724655151,
      "learning_rate": 4.890842931654837e-06,
      "loss": 2.0663,
      "step": 217142
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.4477040767669678,
      "learning_rate": 4.890107341816596e-06,
      "loss": 2.0728,
      "step": 217143
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.5427563190460205,
      "learning_rate": 4.8893718068448e-06,
      "loss": 1.914,
      "step": 217144
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.3815505504608154,
      "learning_rate": 4.888636326739681e-06,
      "loss": 1.9756,
      "step": 217145
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.3907164335250854,
      "learning_rate": 4.887900901501307e-06,
      "loss": 2.1015,
      "step": 217146
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.5532158613204956,
      "learning_rate": 4.887165531129877e-06,
      "loss": 1.9891,
      "step": 217147
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.5323609113693237,
      "learning_rate": 4.886430215625459e-06,
      "loss": 2.0671,
      "step": 217148
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.5144647359848022,
      "learning_rate": 4.885694954988217e-06,
      "loss": 1.9616,
      "step": 217149
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.3739575147628784,
      "learning_rate": 4.8849597492183204e-06,
      "loss": 2.0392,
      "step": 217150
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.482689619064331,
      "learning_rate": 4.8842245983158665e-06,
      "loss": 2.1203,
      "step": 217151
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.6571874618530273,
      "learning_rate": 4.883489502281024e-06,
      "loss": 2.0835,
      "step": 217152
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.4272491931915283,
      "learning_rate": 4.882754461113891e-06,
      "loss": 2.0466,
      "step": 217153
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.416722059249878,
      "learning_rate": 4.882019474814636e-06,
      "loss": 2.037,
      "step": 217154
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.5110505819320679,
      "learning_rate": 4.8812845433833905e-06,
      "loss": 2.0981,
      "step": 217155
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.5461935997009277,
      "learning_rate": 4.8805496668202545e-06,
      "loss": 2.1,
      "step": 217156
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.5048869848251343,
      "learning_rate": 4.879814845125429e-06,
      "loss": 2.1687,
      "step": 217157
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.3693301677703857,
      "learning_rate": 4.879080078299014e-06,
      "loss": 1.7671,
      "step": 217158
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.4660063982009888,
      "learning_rate": 4.878345366341141e-06,
      "loss": 1.8628,
      "step": 217159
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.468591332435608,
      "learning_rate": 4.877610709251944e-06,
      "loss": 2.0557,
      "step": 217160
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.5427180528640747,
      "learning_rate": 4.876876107031591e-06,
      "loss": 2.0577,
      "step": 217161
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.5571115016937256,
      "learning_rate": 4.8761415596801805e-06,
      "loss": 1.9205,
      "step": 217162
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.4776487350463867,
      "learning_rate": 4.875407067197878e-06,
      "loss": 1.8863,
      "step": 217163
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.396225929260254,
      "learning_rate": 4.874672629584786e-06,
      "loss": 2.1869,
      "step": 217164
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.533328652381897,
      "learning_rate": 4.873938246841103e-06,
      "loss": 2.0055,
      "step": 217165
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.4395880699157715,
      "learning_rate": 4.873203918966895e-06,
      "loss": 1.9558,
      "step": 217166
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.3690053224563599,
      "learning_rate": 4.872469645962329e-06,
      "loss": 1.9895,
      "step": 217167
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.497497320175171,
      "learning_rate": 4.871735427827539e-06,
      "loss": 2.0724,
      "step": 217168
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.380391001701355,
      "learning_rate": 4.871001264562691e-06,
      "loss": 1.8901,
      "step": 217169
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.3738361597061157,
      "learning_rate": 4.870267156167884e-06,
      "loss": 1.948,
      "step": 217170
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.465328574180603,
      "learning_rate": 4.869533102643253e-06,
      "loss": 2.0169,
      "step": 217171
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.3817166090011597,
      "learning_rate": 4.868799103988963e-06,
      "loss": 1.984,
      "step": 217172
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.4776206016540527,
      "learning_rate": 4.868065160205115e-06,
      "loss": 1.9382,
      "step": 217173
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.3825923204421997,
      "learning_rate": 4.8673312712918744e-06,
      "loss": 1.7573,
      "step": 217174
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.6648645401000977,
      "learning_rate": 4.866597437249376e-06,
      "loss": 1.8389,
      "step": 217175
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.566420078277588,
      "learning_rate": 4.865863658077751e-06,
      "loss": 2.0445,
      "step": 217176
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.5283814668655396,
      "learning_rate": 4.865129933777101e-06,
      "loss": 2.0805,
      "step": 217177
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.5267879962921143,
      "learning_rate": 4.864396264347625e-06,
      "loss": 2.1714,
      "step": 217178
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.5690213441848755,
      "learning_rate": 4.863662649789424e-06,
      "loss": 1.9887,
      "step": 217179
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.6849257946014404,
      "learning_rate": 4.86292909010263e-06,
      "loss": 1.982,
      "step": 217180
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.4271974563598633,
      "learning_rate": 4.8621955852873764e-06,
      "loss": 1.7963,
      "step": 217181
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.413761854171753,
      "learning_rate": 4.8614621353438635e-06,
      "loss": 1.6904,
      "step": 217182
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.4390162229537964,
      "learning_rate": 4.860728740272124e-06,
      "loss": 1.9938,
      "step": 217183
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.5809932947158813,
      "learning_rate": 4.859995400072358e-06,
      "loss": 1.9398,
      "step": 217184
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.381545066833496,
      "learning_rate": 4.8592621147447e-06,
      "loss": 1.8445,
      "step": 217185
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.4660426378250122,
      "learning_rate": 4.858528884289248e-06,
      "loss": 1.9581,
      "step": 217186
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.4593348503112793,
      "learning_rate": 4.857795708706202e-06,
      "loss": 1.9189,
      "step": 217187
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.5294865369796753,
      "learning_rate": 4.857062587995664e-06,
      "loss": 2.242,
      "step": 217188
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.4638062715530396,
      "learning_rate": 4.856329522157765e-06,
      "loss": 1.9893,
      "step": 217189
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.4292551279067993,
      "learning_rate": 4.8555965111926385e-06,
      "loss": 1.8193,
      "step": 217190
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.382449984550476,
      "learning_rate": 4.854863555100419e-06,
      "loss": 1.9114,
      "step": 217191
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.5149965286254883,
      "learning_rate": 4.854130653881239e-06,
      "loss": 1.8525,
      "step": 217192
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.5896815061569214,
      "learning_rate": 4.853397807535298e-06,
      "loss": 1.9513,
      "step": 217193
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.4180326461791992,
      "learning_rate": 4.852665016062629e-06,
      "loss": 1.7828,
      "step": 217194
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.4731171131134033,
      "learning_rate": 4.851932279463433e-06,
      "loss": 2.0685,
      "step": 217195
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.386033535003662,
      "learning_rate": 4.851199597737876e-06,
      "loss": 2.1356,
      "step": 217196
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.4290955066680908,
      "learning_rate": 4.850466970886024e-06,
      "loss": 1.9077,
      "step": 217197
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.514191746711731,
      "learning_rate": 4.849734398908012e-06,
      "loss": 2.1997,
      "step": 217198
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.4666211605072021,
      "learning_rate": 4.8490018818040376e-06,
      "loss": 2.0119,
      "step": 217199
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.4632341861724854,
      "learning_rate": 4.848269419574202e-06,
      "loss": 1.9218,
      "step": 217200
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.4388384819030762,
      "learning_rate": 4.8475370122186384e-06,
      "loss": 2.0175,
      "step": 217201
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.3768783807754517,
      "learning_rate": 4.846804659737513e-06,
      "loss": 1.855,
      "step": 217202
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.3738795518875122,
      "learning_rate": 4.846072362130926e-06,
      "loss": 2.196,
      "step": 217203
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.355912446975708,
      "learning_rate": 4.8453401193990105e-06,
      "loss": 2.0052,
      "step": 217204
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.4724421501159668,
      "learning_rate": 4.844607931541933e-06,
      "loss": 1.932,
      "step": 217205
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.3893811702728271,
      "learning_rate": 4.843875798559793e-06,
      "loss": 2.1089,
      "step": 217206
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.4172072410583496,
      "learning_rate": 4.843143720452758e-06,
      "loss": 2.0414,
      "step": 217207
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.5229270458221436,
      "learning_rate": 4.84241169722096e-06,
      "loss": 1.8582,
      "step": 217208
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.4570523500442505,
      "learning_rate": 4.8416797288645e-06,
      "loss": 2.0869,
      "step": 217209
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.434145450592041,
      "learning_rate": 4.840947815383611e-06,
      "loss": 2.2247,
      "step": 217210
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.5127328634262085,
      "learning_rate": 4.840215956778326e-06,
      "loss": 2.0568,
      "step": 217211
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.6543419361114502,
      "learning_rate": 4.8394841530487785e-06,
      "loss": 1.8551,
      "step": 217212
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.5498263835906982,
      "learning_rate": 4.838752404195201e-06,
      "loss": 2.0809,
      "step": 217213
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.4484614133834839,
      "learning_rate": 4.838020710217627e-06,
      "loss": 1.7317,
      "step": 217214
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.3653548955917358,
      "learning_rate": 4.837289071116257e-06,
      "loss": 1.8968,
      "step": 217215
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.4108768701553345,
      "learning_rate": 4.836557486891224e-06,
      "loss": 2.2124,
      "step": 217216
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.510683536529541,
      "learning_rate": 4.835825957542594e-06,
      "loss": 2.0509,
      "step": 217217
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.3854745626449585,
      "learning_rate": 4.8350944830706e-06,
      "loss": 1.9801,
      "step": 217218
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.5238022804260254,
      "learning_rate": 4.834363063475344e-06,
      "loss": 1.7138,
      "step": 217219
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.4267003536224365,
      "learning_rate": 4.833631698756923e-06,
      "loss": 1.9418,
      "step": 217220
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.6506019830703735,
      "learning_rate": 4.832900388915506e-06,
      "loss": 1.9829,
      "step": 217221
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.6597802639007568,
      "learning_rate": 4.832169133951225e-06,
      "loss": 1.8738,
      "step": 217222
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.5149333477020264,
      "learning_rate": 4.831437933864213e-06,
      "loss": 1.9909,
      "step": 217223
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.4267526865005493,
      "learning_rate": 4.830706788654604e-06,
      "loss": 1.9828,
      "step": 217224
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.4528706073760986,
      "learning_rate": 4.829975698322564e-06,
      "loss": 1.8749,
      "step": 217225
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.464902639389038,
      "learning_rate": 4.8292446628681926e-06,
      "loss": 2.0293,
      "step": 217226
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.6577863693237305,
      "learning_rate": 4.828513682291624e-06,
      "loss": 2.0212,
      "step": 217227
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.5284194946289062,
      "learning_rate": 4.82778275659299e-06,
      "loss": 1.9253,
      "step": 217228
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.5678516626358032,
      "learning_rate": 4.827051885772459e-06,
      "loss": 2.1772,
      "step": 217229
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.455922245979309,
      "learning_rate": 4.826321069830163e-06,
      "loss": 2.0453,
      "step": 217230
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.3605937957763672,
      "learning_rate": 4.825590308766236e-06,
      "loss": 1.8787,
      "step": 217231
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.458875060081482,
      "learning_rate": 4.8248596025808104e-06,
      "loss": 1.9585,
      "step": 217232
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.5970467329025269,
      "learning_rate": 4.824128951273953e-06,
      "loss": 2.0973,
      "step": 217233
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.3427677154541016,
      "learning_rate": 4.823398354845931e-06,
      "loss": 1.9481,
      "step": 217234
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.4481257200241089,
      "learning_rate": 4.822667813296743e-06,
      "loss": 1.8358,
      "step": 217235
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.4653301239013672,
      "learning_rate": 4.821937326626657e-06,
      "loss": 2.0594,
      "step": 217236
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.435376763343811,
      "learning_rate": 4.821206894835738e-06,
      "loss": 2.0035,
      "step": 217237
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.4450838565826416,
      "learning_rate": 4.820476517924088e-06,
      "loss": 2.0013,
      "step": 217238
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.477184772491455,
      "learning_rate": 4.819746195891905e-06,
      "loss": 1.9227,
      "step": 217239
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.4120856523513794,
      "learning_rate": 4.819015928739323e-06,
      "loss": 2.0667,
      "step": 217240
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.4282211065292358,
      "learning_rate": 4.818285716466408e-06,
      "loss": 1.9071,
      "step": 217241
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.4735733270645142,
      "learning_rate": 4.817555559073394e-06,
      "loss": 2.0148,
      "step": 217242
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.4026297330856323,
      "learning_rate": 4.816825456560347e-06,
      "loss": 1.673,
      "step": 217243
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.438368797302246,
      "learning_rate": 4.816095408927434e-06,
      "loss": 1.9353,
      "step": 217244
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.5163695812225342,
      "learning_rate": 4.815365416174754e-06,
      "loss": 1.8628,
      "step": 217245
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.4715420007705688,
      "learning_rate": 4.814635478302509e-06,
      "loss": 2.0721,
      "step": 217246
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.5157864093780518,
      "learning_rate": 4.813905595310763e-06,
      "loss": 2.1239,
      "step": 217247
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.413407564163208,
      "learning_rate": 4.8131757671997175e-06,
      "loss": 1.9904,
      "step": 217248
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.424546480178833,
      "learning_rate": 4.812445993969438e-06,
      "loss": 2.0212,
      "step": 217249
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.5904680490493774,
      "learning_rate": 4.811716275620125e-06,
      "loss": 2.0109,
      "step": 217250
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.6142768859863281,
      "learning_rate": 4.810986612151879e-06,
      "loss": 2.1304,
      "step": 217251
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.6176353693008423,
      "learning_rate": 4.810257003564832e-06,
      "loss": 1.9082,
      "step": 217252
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.5308736562728882,
      "learning_rate": 4.809527449859119e-06,
      "loss": 2.0818,
      "step": 217253
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.5757216215133667,
      "learning_rate": 4.808797951034904e-06,
      "loss": 2.0123,
      "step": 217254
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.4809752702713013,
      "learning_rate": 4.808068507092322e-06,
      "loss": 1.9881,
      "step": 217255
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.5819936990737915,
      "learning_rate": 4.807339118031506e-06,
      "loss": 2.0317,
      "step": 217256
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.4533700942993164,
      "learning_rate": 4.806609783852522e-06,
      "loss": 1.9847,
      "step": 217257
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.4149607419967651,
      "learning_rate": 4.805880504555604e-06,
      "loss": 1.8293,
      "step": 217258
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.5034561157226562,
      "learning_rate": 4.805151280140818e-06,
      "loss": 2.0135,
      "step": 217259
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.4565119743347168,
      "learning_rate": 4.804422110608331e-06,
      "loss": 2.1012,
      "step": 217260
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.3983619213104248,
      "learning_rate": 4.803692995958275e-06,
      "loss": 1.8263,
      "step": 217261
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.4462000131607056,
      "learning_rate": 4.802963936190818e-06,
      "loss": 1.7389,
      "step": 217262
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.4456273317337036,
      "learning_rate": 4.802234931306026e-06,
      "loss": 2.0555,
      "step": 217263
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.5808515548706055,
      "learning_rate": 4.8015059813040654e-06,
      "loss": 2.0556,
      "step": 217264
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.4476439952850342,
      "learning_rate": 4.80077708618507e-06,
      "loss": 1.8073,
      "step": 217265
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.8863776922225952,
      "learning_rate": 4.8000482459492395e-06,
      "loss": 1.8579,
      "step": 217266
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.5397614240646362,
      "learning_rate": 4.799319460596574e-06,
      "loss": 2.1252,
      "step": 217267
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.5152939558029175,
      "learning_rate": 4.798590730127372e-06,
      "loss": 1.88,
      "step": 217268
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.4487935304641724,
      "learning_rate": 4.797862054541634e-06,
      "loss": 2.0363,
      "step": 217269
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.4861053228378296,
      "learning_rate": 4.797133433839562e-06,
      "loss": 1.6863,
      "step": 217270
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.498741865158081,
      "learning_rate": 4.796404868021253e-06,
      "loss": 1.8114,
      "step": 217271
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.500791072845459,
      "learning_rate": 4.795676357086875e-06,
      "loss": 2.0766,
      "step": 217272
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.4633954763412476,
      "learning_rate": 4.794947901036561e-06,
      "loss": 1.8455,
      "step": 217273
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.4481114149093628,
      "learning_rate": 4.794219499870444e-06,
      "loss": 1.9925,
      "step": 217274
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.4503728151321411,
      "learning_rate": 4.793491153588658e-06,
      "loss": 1.9663,
      "step": 217275
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.471841812133789,
      "learning_rate": 4.792762862191302e-06,
      "loss": 2.3123,
      "step": 217276
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.5425286293029785,
      "learning_rate": 4.792034625678576e-06,
      "loss": 2.0643,
      "step": 217277
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.3830517530441284,
      "learning_rate": 4.791306444050547e-06,
      "loss": 2.16,
      "step": 217278
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.414150595664978,
      "learning_rate": 4.790578317307414e-06,
      "loss": 1.8457,
      "step": 217279
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.5018970966339111,
      "learning_rate": 4.7898502454492785e-06,
      "loss": 1.786,
      "step": 217280
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.4574335813522339,
      "learning_rate": 4.7891222284763054e-06,
      "loss": 2.1464,
      "step": 217281
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.5196329355239868,
      "learning_rate": 4.788394266388596e-06,
      "loss": 2.0348,
      "step": 217282
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.5695968866348267,
      "learning_rate": 4.7876663591862485e-06,
      "loss": 1.827,
      "step": 217283
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.627712607383728,
      "learning_rate": 4.7869385068694975e-06,
      "loss": 1.9789,
      "step": 217284
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.5251519680023193,
      "learning_rate": 4.78621070943841e-06,
      "loss": 1.9104,
      "step": 217285
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.4366117715835571,
      "learning_rate": 4.78548296689315e-06,
      "loss": 1.9669,
      "step": 217286
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.3957600593566895,
      "learning_rate": 4.7847552792338535e-06,
      "loss": 1.9566,
      "step": 217287
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.4298455715179443,
      "learning_rate": 4.78402764646062e-06,
      "loss": 1.8913,
      "step": 217288
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.3491594791412354,
      "learning_rate": 4.783300068573614e-06,
      "loss": 2.0915,
      "step": 217289
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.5360690355300903,
      "learning_rate": 4.782572545572972e-06,
      "loss": 1.9597,
      "step": 217290
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.499118685722351,
      "learning_rate": 4.78184507745879e-06,
      "loss": 1.9134,
      "step": 217291
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.419052004814148,
      "learning_rate": 4.781117664231271e-06,
      "loss": 1.7812,
      "step": 217292
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.432370662689209,
      "learning_rate": 4.7803903058905136e-06,
      "loss": 1.9873,
      "step": 217293
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.4719794988632202,
      "learning_rate": 4.779663002436618e-06,
      "loss": 1.8308,
      "step": 217294
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.5169472694396973,
      "learning_rate": 4.778935753869784e-06,
      "loss": 2.1282,
      "step": 217295
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.5626951456069946,
      "learning_rate": 4.7782085601901445e-06,
      "loss": 2.2889,
      "step": 217296
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.5254275798797607,
      "learning_rate": 4.7774814213977665e-06,
      "loss": 1.9965,
      "step": 217297
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.4815013408660889,
      "learning_rate": 4.77675433749285e-06,
      "loss": 1.8359,
      "step": 217298
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.4732754230499268,
      "learning_rate": 4.776027308475494e-06,
      "loss": 1.9431,
      "step": 217299
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.445300817489624,
      "learning_rate": 4.775300334345833e-06,
      "loss": 2.0793,
      "step": 217300
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.4153343439102173,
      "learning_rate": 4.774573415104066e-06,
      "loss": 2.0527,
      "step": 217301
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.533612847328186,
      "learning_rate": 4.773846550750227e-06,
      "loss": 2.1047,
      "step": 217302
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.313082218170166,
      "learning_rate": 4.773119741284548e-06,
      "loss": 1.9807,
      "step": 217303
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.395519733428955,
      "learning_rate": 4.772392986707096e-06,
      "loss": 1.7811,
      "step": 217304
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.4410876035690308,
      "learning_rate": 4.771666287018039e-06,
      "loss": 2.1249,
      "step": 217305
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.6167376041412354,
      "learning_rate": 4.770939642217475e-06,
      "loss": 1.8459,
      "step": 217306
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.3689548969268799,
      "learning_rate": 4.770213052305604e-06,
      "loss": 1.9402,
      "step": 217307
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.5302623510360718,
      "learning_rate": 4.769486517282462e-06,
      "loss": 1.9206,
      "step": 217308
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.5154619216918945,
      "learning_rate": 4.768760037148312e-06,
      "loss": 1.9256,
      "step": 217309
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.4814916849136353,
      "learning_rate": 4.768033611903188e-06,
      "loss": 2.1481,
      "step": 217310
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.5327999591827393,
      "learning_rate": 4.767307241547291e-06,
      "loss": 1.7699,
      "step": 217311
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.4104504585266113,
      "learning_rate": 4.766580926080687e-06,
      "loss": 2.1153,
      "step": 217312
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.3864891529083252,
      "learning_rate": 4.765854665503577e-06,
      "loss": 1.9974,
      "step": 217313
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.4977976083755493,
      "learning_rate": 4.765128459816026e-06,
      "loss": 1.9638,
      "step": 217314
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.4878395795822144,
      "learning_rate": 4.7644023090182335e-06,
      "loss": 2.0479,
      "step": 217315
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.6082159280776978,
      "learning_rate": 4.763676213110334e-06,
      "loss": 2.1553,
      "step": 217316
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.424440860748291,
      "learning_rate": 4.762950172092428e-06,
      "loss": 2.0044,
      "step": 217317
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.4393633604049683,
      "learning_rate": 4.762224185964647e-06,
      "loss": 2.01,
      "step": 217318
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.4907026290893555,
      "learning_rate": 4.761498254727158e-06,
      "loss": 2.0161,
      "step": 217319
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.423909306526184,
      "learning_rate": 4.760772378380063e-06,
      "loss": 1.7973,
      "step": 217320
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.4785268306732178,
      "learning_rate": 4.760046556923524e-06,
      "loss": 1.8838,
      "step": 217321
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.4851244688034058,
      "learning_rate": 4.759320790357679e-06,
      "loss": 2.2882,
      "step": 217322
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.4432703256607056,
      "learning_rate": 4.758595078682658e-06,
      "loss": 1.8191,
      "step": 217323
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.4246630668640137,
      "learning_rate": 4.757869421898531e-06,
      "loss": 1.6127,
      "step": 217324
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.531858205795288,
      "learning_rate": 4.757143820005527e-06,
      "loss": 1.9735,
      "step": 217325
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.5641744136810303,
      "learning_rate": 4.756418273003748e-06,
      "loss": 2.0449,
      "step": 217326
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.4439167976379395,
      "learning_rate": 4.755692780893294e-06,
      "loss": 1.8241,
      "step": 217327
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.490996241569519,
      "learning_rate": 4.754967343674365e-06,
      "loss": 2.1753,
      "step": 217328
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.6519625186920166,
      "learning_rate": 4.754241961347061e-06,
      "loss": 1.9749,
      "step": 217329
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.357371211051941,
      "learning_rate": 4.75351663391148e-06,
      "loss": 1.9663,
      "step": 217330
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.4418567419052124,
      "learning_rate": 4.752791361367825e-06,
      "loss": 2.058,
      "step": 217331
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.4739930629730225,
      "learning_rate": 4.752066143716193e-06,
      "loss": 2.0706,
      "step": 217332
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.4349048137664795,
      "learning_rate": 4.751340980956719e-06,
      "loss": 2.1486,
      "step": 217333
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.431082844734192,
      "learning_rate": 4.750615873089536e-06,
      "loss": 1.9664,
      "step": 217334
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.6458921432495117,
      "learning_rate": 4.7498908201148435e-06,
      "loss": 1.959,
      "step": 217335
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.3936007022857666,
      "learning_rate": 4.749165822032641e-06,
      "loss": 1.954,
      "step": 217336
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.4324870109558105,
      "learning_rate": 4.748440878843196e-06,
      "loss": 2.1234,
      "step": 217337
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.3556441068649292,
      "learning_rate": 4.747715990546541e-06,
      "loss": 1.9702,
      "step": 217338
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.5133399963378906,
      "learning_rate": 4.74699115714291e-06,
      "loss": 2.1527,
      "step": 217339
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.3458172082901,
      "learning_rate": 4.746266378632335e-06,
      "loss": 2.1977,
      "step": 217340
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.4486346244812012,
      "learning_rate": 4.745541655015051e-06,
      "loss": 1.8936,
      "step": 217341
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.327415943145752,
      "learning_rate": 4.744816986291122e-06,
      "loss": 1.7383,
      "step": 217342
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.5265001058578491,
      "learning_rate": 4.744092372460717e-06,
      "loss": 2.1921,
      "step": 217343
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.3971244096755981,
      "learning_rate": 4.743367813523935e-06,
      "loss": 1.8378,
      "step": 217344
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.394492268562317,
      "learning_rate": 4.742643309480942e-06,
      "loss": 1.8759,
      "step": 217345
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.4887213706970215,
      "learning_rate": 4.741918860331872e-06,
      "loss": 2.1369,
      "step": 217346
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.5041896104812622,
      "learning_rate": 4.7411944660768586e-06,
      "loss": 2.0562,
      "step": 217347
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.3863952159881592,
      "learning_rate": 4.740470126716034e-06,
      "loss": 1.8546,
      "step": 217348
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.5303813219070435,
      "learning_rate": 4.739745842249498e-06,
      "loss": 1.7571,
      "step": 217349
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.565925121307373,
      "learning_rate": 4.739021612677452e-06,
      "loss": 1.8,
      "step": 217350
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.4033616781234741,
      "learning_rate": 4.7382974379999604e-06,
      "loss": 2.1824,
      "step": 217351
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.4346390962600708,
      "learning_rate": 4.737573318217225e-06,
      "loss": 2.0301,
      "step": 217352
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.6084102392196655,
      "learning_rate": 4.736849253329344e-06,
      "loss": 2.1478,
      "step": 217353
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.4205437898635864,
      "learning_rate": 4.736125243336453e-06,
      "loss": 2.2195,
      "step": 217354
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.5623342990875244,
      "learning_rate": 4.735401288238683e-06,
      "loss": 2.0483,
      "step": 217355
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.5523133277893066,
      "learning_rate": 4.734677388036168e-06,
      "loss": 1.9541,
      "step": 217356
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.395387887954712,
      "learning_rate": 4.7339535427290744e-06,
      "loss": 2.0225,
      "step": 217357
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.3472501039505005,
      "learning_rate": 4.733229752317469e-06,
      "loss": 2.0111,
      "step": 217358
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.5442006587982178,
      "learning_rate": 4.732506016801585e-06,
      "loss": 1.9105,
      "step": 217359
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.3793156147003174,
      "learning_rate": 4.731782336181455e-06,
      "loss": 1.8801,
      "step": 217360
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.6121934652328491,
      "learning_rate": 4.731058710457314e-06,
      "loss": 2.0621,
      "step": 217361
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.502734661102295,
      "learning_rate": 4.730335139629193e-06,
      "loss": 2.0632,
      "step": 217362
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.7763952016830444,
      "learning_rate": 4.7296116236972935e-06,
      "loss": 2.0844,
      "step": 217363
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.4226475954055786,
      "learning_rate": 4.728888162661715e-06,
      "loss": 2.0503,
      "step": 217364
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.4584904909133911,
      "learning_rate": 4.728164756522623e-06,
      "loss": 1.8088,
      "step": 217365
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.3491541147232056,
      "learning_rate": 4.727441405280153e-06,
      "loss": 1.7671,
      "step": 217366
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.4542733430862427,
      "learning_rate": 4.726718108934435e-06,
      "loss": 1.8867,
      "step": 217367
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.4795156717300415,
      "learning_rate": 4.725994867485538e-06,
      "loss": 1.9798,
      "step": 217368
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.4343820810317993,
      "learning_rate": 4.7252716809336955e-06,
      "loss": 1.9708,
      "step": 217369
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.469079613685608,
      "learning_rate": 4.724548549279006e-06,
      "loss": 2.0826,
      "step": 217370
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.5808557271957397,
      "learning_rate": 4.7238254725215695e-06,
      "loss": 1.8932,
      "step": 217371
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.3374638557434082,
      "learning_rate": 4.723102450661587e-06,
      "loss": 1.9418,
      "step": 217372
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.5036141872406006,
      "learning_rate": 4.722379483699123e-06,
      "loss": 2.0657,
      "step": 217373
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.4863572120666504,
      "learning_rate": 4.7216565716343135e-06,
      "loss": 1.9511,
      "step": 217374
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.4758200645446777,
      "learning_rate": 4.7209337144673895e-06,
      "loss": 2.1403,
      "step": 217375
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.4393203258514404,
      "learning_rate": 4.720210912198352e-06,
      "loss": 1.9796,
      "step": 217376
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.442720651626587,
      "learning_rate": 4.7194881648274344e-06,
      "loss": 1.9496,
      "step": 217377
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.4875049591064453,
      "learning_rate": 4.7187654723547685e-06,
      "loss": 2.0319,
      "step": 217378
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.4208133220672607,
      "learning_rate": 4.718042834780389e-06,
      "loss": 1.942,
      "step": 217379
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.4365689754486084,
      "learning_rate": 4.717320252104562e-06,
      "loss": 1.6983,
      "step": 217380
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.427259087562561,
      "learning_rate": 4.71659772432732e-06,
      "loss": 2.0015,
      "step": 217381
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.492294192314148,
      "learning_rate": 4.715875251448831e-06,
      "loss": 2.2186,
      "step": 217382
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.5216447114944458,
      "learning_rate": 4.71515283346926e-06,
      "loss": 2.1336,
      "step": 217383
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.320189356803894,
      "learning_rate": 4.7144304703887414e-06,
      "loss": 1.9316,
      "step": 217384
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.5622392892837524,
      "learning_rate": 4.713708162207308e-06,
      "loss": 1.9706,
      "step": 217385
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.423393964767456,
      "learning_rate": 4.7129859089252265e-06,
      "loss": 2.0199,
      "step": 217386
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.3945059776306152,
      "learning_rate": 4.712263710542563e-06,
      "loss": 1.9207,
      "step": 217387
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.5772775411605835,
      "learning_rate": 4.711541567059452e-06,
      "loss": 2.0637,
      "step": 217388
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.4519236087799072,
      "learning_rate": 4.7108194784760575e-06,
      "loss": 2.2359,
      "step": 217389
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.460477590560913,
      "learning_rate": 4.710097444792516e-06,
      "loss": 2.0279,
      "step": 217390
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.421154499053955,
      "learning_rate": 4.709375466008891e-06,
      "loss": 1.9879,
      "step": 217391
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.463670015335083,
      "learning_rate": 4.708653542125384e-06,
      "loss": 1.8511,
      "step": 217392
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.3219832181930542,
      "learning_rate": 4.707931673142096e-06,
      "loss": 2.001,
      "step": 217393
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.4293683767318726,
      "learning_rate": 4.7072098590592244e-06,
      "loss": 2.0331,
      "step": 217394
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.4592314958572388,
      "learning_rate": 4.706488099876837e-06,
      "loss": 1.9134,
      "step": 217395
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.335269808769226,
      "learning_rate": 4.705766395595101e-06,
      "loss": 1.8733,
      "step": 217396
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.511579990386963,
      "learning_rate": 4.705044746214082e-06,
      "loss": 1.9596,
      "step": 217397
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.3936153650283813,
      "learning_rate": 4.704323151734013e-06,
      "loss": 1.9091,
      "step": 217398
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.592598795890808,
      "learning_rate": 4.703601612154961e-06,
      "loss": 1.8827,
      "step": 217399
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.5382624864578247,
      "learning_rate": 4.702880127477127e-06,
      "loss": 2.0861,
      "step": 217400
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.5679625272750854,
      "learning_rate": 4.702158697700542e-06,
      "loss": 1.9615,
      "step": 217401
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.4004493951797485,
      "learning_rate": 4.701437322825474e-06,
      "loss": 1.8979,
      "step": 217402
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.4440019130706787,
      "learning_rate": 4.700716002851923e-06,
      "loss": 2.1697,
      "step": 217403
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.3585764169692993,
      "learning_rate": 4.699994737780121e-06,
      "loss": 1.7705,
      "step": 217404
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.4999414682388306,
      "learning_rate": 4.699273527610104e-06,
      "loss": 1.9612,
      "step": 217405
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.4205107688903809,
      "learning_rate": 4.698552372342135e-06,
      "loss": 1.9522,
      "step": 217406
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.4707300662994385,
      "learning_rate": 4.697831271976215e-06,
      "loss": 1.999,
      "step": 217407
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.448078989982605,
      "learning_rate": 4.697110226512612e-06,
      "loss": 2.0025,
      "step": 217408
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.306309700012207,
      "learning_rate": 4.696389235951359e-06,
      "loss": 1.9649,
      "step": 217409
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.460034966468811,
      "learning_rate": 4.695668300292621e-06,
      "loss": 1.904,
      "step": 217410
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.5138171911239624,
      "learning_rate": 4.694947419536533e-06,
      "loss": 2.0326,
      "step": 217411
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.8084046840667725,
      "learning_rate": 4.694226593683225e-06,
      "loss": 2.1478,
      "step": 217412
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.4434001445770264,
      "learning_rate": 4.6935058227328014e-06,
      "loss": 2.1638,
      "step": 217413
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.4510008096694946,
      "learning_rate": 4.692785106685526e-06,
      "loss": 1.8567,
      "step": 217414
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.6617738008499146,
      "learning_rate": 4.692064445541333e-06,
      "loss": 2.1762,
      "step": 217415
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.5810701847076416,
      "learning_rate": 4.691343839300521e-06,
      "loss": 1.766,
      "step": 217416
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.4934377670288086,
      "learning_rate": 4.690623287963125e-06,
      "loss": 1.9193,
      "step": 217417
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.5474005937576294,
      "learning_rate": 4.689902791529343e-06,
      "loss": 1.8707,
      "step": 217418
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.3231021165847778,
      "learning_rate": 4.689182349999276e-06,
      "loss": 1.6838,
      "step": 217419
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.6760274171829224,
      "learning_rate": 4.688461963373091e-06,
      "loss": 2.2313,
      "step": 217420
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.4540575742721558,
      "learning_rate": 4.687741631650854e-06,
      "loss": 2.0137,
      "step": 217421
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.5061702728271484,
      "learning_rate": 4.687021354832732e-06,
      "loss": 2.2085,
      "step": 217422
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.4578603506088257,
      "learning_rate": 4.686301132918891e-06,
      "loss": 2.0133,
      "step": 217423
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.4963175058364868,
      "learning_rate": 4.6855809659094304e-06,
      "loss": 2.0337,
      "step": 217424
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.4427047967910767,
      "learning_rate": 4.684860853804484e-06,
      "loss": 1.8883,
      "step": 217425
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.4447637796401978,
      "learning_rate": 4.684140796604252e-06,
      "loss": 2.1322,
      "step": 217426
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.5240778923034668,
      "learning_rate": 4.683420794308768e-06,
      "loss": 1.8823,
      "step": 217427
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.5351622104644775,
      "learning_rate": 4.68270084691823e-06,
      "loss": 2.0313,
      "step": 217428
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.6194779872894287,
      "learning_rate": 4.681980954432707e-06,
      "loss": 2.162,
      "step": 217429
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.4565222263336182,
      "learning_rate": 4.681261116852431e-06,
      "loss": 1.9384,
      "step": 217430
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.5144761800765991,
      "learning_rate": 4.6805413341774675e-06,
      "loss": 2.0705,
      "step": 217431
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.4592441320419312,
      "learning_rate": 4.679821606407986e-06,
      "loss": 2.0583,
      "step": 217432
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.3344509601593018,
      "learning_rate": 4.679101933544049e-06,
      "loss": 1.9976,
      "step": 217433
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.5947645902633667,
      "learning_rate": 4.678382315585894e-06,
      "loss": 1.9759,
      "step": 217434
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.640587329864502,
      "learning_rate": 4.677662752533551e-06,
      "loss": 1.9265,
      "step": 217435
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.5506120920181274,
      "learning_rate": 4.676943244387255e-06,
      "loss": 1.9718,
      "step": 217436
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.6986546516418457,
      "learning_rate": 4.676223791147038e-06,
      "loss": 1.9065,
      "step": 217437
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.5302656888961792,
      "learning_rate": 4.675504392813134e-06,
      "loss": 2.152,
      "step": 217438
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.3572932481765747,
      "learning_rate": 4.67478504938561e-06,
      "loss": 1.8753,
      "step": 217439
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.3437907695770264,
      "learning_rate": 4.674065760864631e-06,
      "loss": 1.9957,
      "step": 217440
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.4391454458236694,
      "learning_rate": 4.673346527250266e-06,
      "loss": 1.9429,
      "step": 217441
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.4327387809753418,
      "learning_rate": 4.672627348542745e-06,
      "loss": 2.0813,
      "step": 217442
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.517972707748413,
      "learning_rate": 4.671908224742138e-06,
      "loss": 2.1656,
      "step": 217443
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.436259150505066,
      "learning_rate": 4.671189155848642e-06,
      "loss": 2.1204,
      "step": 217444
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.4678311347961426,
      "learning_rate": 4.670470141862326e-06,
      "loss": 1.8187,
      "step": 217445
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.592504858970642,
      "learning_rate": 4.6697511827833216e-06,
      "loss": 1.9293,
      "step": 217446
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.4393856525421143,
      "learning_rate": 4.669032278611795e-06,
      "loss": 1.8595,
      "step": 217447
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.4858603477478027,
      "learning_rate": 4.668313429347881e-06,
      "loss": 2.1547,
      "step": 217448
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.5365869998931885,
      "learning_rate": 4.667594634991678e-06,
      "loss": 1.9178,
      "step": 217449
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.4883403778076172,
      "learning_rate": 4.666875895543387e-06,
      "loss": 1.914,
      "step": 217450
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.472893238067627,
      "learning_rate": 4.666157211003041e-06,
      "loss": 2.0333,
      "step": 217451
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.4930380582809448,
      "learning_rate": 4.665438581370873e-06,
      "loss": 1.9771,
      "step": 217452
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.512385368347168,
      "learning_rate": 4.664720006646949e-06,
      "loss": 2.1036,
      "step": 217453
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.5483589172363281,
      "learning_rate": 4.66400148683147e-06,
      "loss": 2.1216,
      "step": 217454
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.378340482711792,
      "learning_rate": 4.663283021924469e-06,
      "loss": 1.9744,
      "step": 217455
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.470119595527649,
      "learning_rate": 4.6625646119261785e-06,
      "loss": 2.0196,
      "step": 217456
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.4501363039016724,
      "learning_rate": 4.661846256836698e-06,
      "loss": 1.9064,
      "step": 217457
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.482235312461853,
      "learning_rate": 4.661127956656163e-06,
      "loss": 1.8727,
      "step": 217458
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.4488807916641235,
      "learning_rate": 4.660409711384638e-06,
      "loss": 1.987,
      "step": 217459
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.4613100290298462,
      "learning_rate": 4.659691521022391e-06,
      "loss": 1.9983,
      "step": 217460
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.5094279050827026,
      "learning_rate": 4.65897338556942e-06,
      "loss": 1.8447,
      "step": 217461
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.5110175609588623,
      "learning_rate": 4.65825530502596e-06,
      "loss": 1.9985,
      "step": 217462
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.5809335708618164,
      "learning_rate": 4.6575372793921094e-06,
      "loss": 1.9902,
      "step": 217463
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.4368146657943726,
      "learning_rate": 4.6568193086679695e-06,
      "loss": 1.9266,
      "step": 217464
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.4636075496673584,
      "learning_rate": 4.656101392853706e-06,
      "loss": 1.8792,
      "step": 217465
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.6532459259033203,
      "learning_rate": 4.655383531949486e-06,
      "loss": 1.931,
      "step": 217466
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.4910609722137451,
      "learning_rate": 4.654665725955342e-06,
      "loss": 1.8437,
      "step": 217467
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.364981770515442,
      "learning_rate": 4.653947974871508e-06,
      "loss": 2.1791,
      "step": 217468
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.4580395221710205,
      "learning_rate": 4.653230278698117e-06,
      "loss": 2.0733,
      "step": 217469
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.4767714738845825,
      "learning_rate": 4.652512637435202e-06,
      "loss": 1.7339,
      "step": 217470
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.4236319065093994,
      "learning_rate": 4.651795051082996e-06,
      "loss": 1.8485,
      "step": 217471
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.5332562923431396,
      "learning_rate": 4.651077519641599e-06,
      "loss": 2.0041,
      "step": 217472
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.4724903106689453,
      "learning_rate": 4.6503600431111125e-06,
      "loss": 1.8724,
      "step": 217473
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.5307786464691162,
      "learning_rate": 4.649642621491701e-06,
      "loss": 1.7953,
      "step": 217474
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.4162864685058594,
      "learning_rate": 4.648925254783531e-06,
      "loss": 2.0379,
      "step": 217475
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.3922746181488037,
      "learning_rate": 4.648207942986671e-06,
      "loss": 1.9812,
      "step": 217476
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.4754226207733154,
      "learning_rate": 4.647490686101285e-06,
      "loss": 1.8446,
      "step": 217477
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.4439280033111572,
      "learning_rate": 4.646773484127508e-06,
      "loss": 1.7398,
      "step": 217478
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.3908902406692505,
      "learning_rate": 4.646056337065474e-06,
      "loss": 2.048,
      "step": 217479
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.5040740966796875,
      "learning_rate": 4.645339244915347e-06,
      "loss": 2.0207,
      "step": 217480
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.566371202468872,
      "learning_rate": 4.644622207677195e-06,
      "loss": 2.2629,
      "step": 217481
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.5656918287277222,
      "learning_rate": 4.643905225351152e-06,
      "loss": 1.8359,
      "step": 217482
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.5478878021240234,
      "learning_rate": 4.64318829793745e-06,
      "loss": 1.9771,
      "step": 217483
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.2640339136123657,
      "learning_rate": 4.64247142543609e-06,
      "loss": 2.0908,
      "step": 217484
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.3772088289260864,
      "learning_rate": 4.6417546078473035e-06,
      "loss": 2.0528,
      "step": 217485
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.3768409490585327,
      "learning_rate": 4.641037845171191e-06,
      "loss": 1.9144,
      "step": 217486
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.4985857009887695,
      "learning_rate": 4.640321137407887e-06,
      "loss": 2.0126,
      "step": 217487
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.4916638135910034,
      "learning_rate": 4.639604484557491e-06,
      "loss": 1.9199,
      "step": 217488
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.3421521186828613,
      "learning_rate": 4.638887886620202e-06,
      "loss": 1.8932,
      "step": 217489
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.3324055671691895,
      "learning_rate": 4.63817134359612e-06,
      "loss": 1.8665,
      "step": 217490
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.5999670028686523,
      "learning_rate": 4.637454855485345e-06,
      "loss": 1.8357,
      "step": 217491
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.5560945272445679,
      "learning_rate": 4.6367384222880774e-06,
      "loss": 1.9684,
      "step": 217492
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.6381560564041138,
      "learning_rate": 4.636022044004417e-06,
      "loss": 1.994,
      "step": 217493
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.4786772727966309,
      "learning_rate": 4.635305720634463e-06,
      "loss": 1.821,
      "step": 217494
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.4309046268463135,
      "learning_rate": 4.634589452178416e-06,
      "loss": 1.8141,
      "step": 217495
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.423045039176941,
      "learning_rate": 4.633873238636343e-06,
      "loss": 1.8186,
      "step": 217496
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.5161539316177368,
      "learning_rate": 4.6331570800084095e-06,
      "loss": 1.9112,
      "step": 217497
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.441293478012085,
      "learning_rate": 4.632440976294749e-06,
      "loss": 2.0018,
      "step": 217498
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.7707890272140503,
      "learning_rate": 4.631724927495561e-06,
      "loss": 1.912,
      "step": 217499
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.5290776491165161,
      "learning_rate": 4.6310089336108136e-06,
      "loss": 1.8633,
      "step": 217500
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.522951602935791,
      "learning_rate": 4.630292994640805e-06,
      "loss": 2.0388,
      "step": 217501
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.5097506046295166,
      "learning_rate": 4.629577110585536e-06,
      "loss": 2.0043,
      "step": 217502
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.5054413080215454,
      "learning_rate": 4.628861281445273e-06,
      "loss": 2.0158,
      "step": 217503
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.5157814025878906,
      "learning_rate": 4.628145507220016e-06,
      "loss": 1.8138,
      "step": 217504
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.5474481582641602,
      "learning_rate": 4.627429787910031e-06,
      "loss": 1.9745,
      "step": 217505
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.548538088798523,
      "learning_rate": 4.6267141235153516e-06,
      "loss": 2.1128,
      "step": 217506
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.5043178796768188,
      "learning_rate": 4.625998514036144e-06,
      "loss": 1.9321,
      "step": 217507
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.5051172971725464,
      "learning_rate": 4.625282959472509e-06,
      "loss": 2.0289,
      "step": 217508
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.4155213832855225,
      "learning_rate": 4.624567459824646e-06,
      "loss": 1.9768,
      "step": 217509
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.352506399154663,
      "learning_rate": 4.623852015092621e-06,
      "loss": 2.0041,
      "step": 217510
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.4738551378250122,
      "learning_rate": 4.623136625276636e-06,
      "loss": 1.9071,
      "step": 217511
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.420034646987915,
      "learning_rate": 4.622421290376754e-06,
      "loss": 1.9318,
      "step": 217512
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.498827338218689,
      "learning_rate": 4.621706010393145e-06,
      "loss": 1.962,
      "step": 217513
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.4177378416061401,
      "learning_rate": 4.62099078532594e-06,
      "loss": 1.9889,
      "step": 217514
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.6065382957458496,
      "learning_rate": 4.620275615175273e-06,
      "loss": 2.0688,
      "step": 217515
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.5470468997955322,
      "learning_rate": 4.619560499941244e-06,
      "loss": 2.215,
      "step": 217516
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.5240141153335571,
      "learning_rate": 4.618845439624086e-06,
      "loss": 2.0137,
      "step": 217517
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.6286803483963013,
      "learning_rate": 4.618130434223799e-06,
      "loss": 1.9653,
      "step": 217518
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.5659524202346802,
      "learning_rate": 4.617415483740583e-06,
      "loss": 1.9186,
      "step": 217519
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.3718620538711548,
      "learning_rate": 4.616700588174571e-06,
      "loss": 2.1202,
      "step": 217520
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.5217205286026,
      "learning_rate": 4.615985747525897e-06,
      "loss": 1.767,
      "step": 217521
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.4296081066131592,
      "learning_rate": 4.615270961794659e-06,
      "loss": 1.8795,
      "step": 217522
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.7339272499084473,
      "learning_rate": 4.6145562309810925e-06,
      "loss": 2.0555,
      "step": 217523
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.5048530101776123,
      "learning_rate": 4.613841555085162e-06,
      "loss": 1.9846,
      "step": 217524
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.4433324337005615,
      "learning_rate": 4.61312693410717e-06,
      "loss": 2.0089,
      "step": 217525
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.5434869527816772,
      "learning_rate": 4.612412368047114e-06,
      "loss": 1.9694,
      "step": 217526
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.5669575929641724,
      "learning_rate": 4.611697856905228e-06,
      "loss": 2.0733,
      "step": 217527
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.4554611444473267,
      "learning_rate": 4.610983400681578e-06,
      "loss": 2.1793,
      "step": 217528
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.5440069437026978,
      "learning_rate": 4.610268999376332e-06,
      "loss": 1.9786,
      "step": 217529
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.4996349811553955,
      "learning_rate": 4.609554652989622e-06,
      "loss": 2.1224,
      "step": 217530
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.5605733394622803,
      "learning_rate": 4.608840361521582e-06,
      "loss": 2.0481,
      "step": 217531
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.4823871850967407,
      "learning_rate": 4.608126124972311e-06,
      "loss": 2.0459,
      "step": 217532
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.4591556787490845,
      "learning_rate": 4.607411943341977e-06,
      "loss": 1.9465,
      "step": 217533
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.3886927366256714,
      "learning_rate": 4.606697816630678e-06,
      "loss": 2.0923,
      "step": 217534
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.5545639991760254,
      "learning_rate": 4.605983744838582e-06,
      "loss": 1.9057,
      "step": 217535
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.4345676898956299,
      "learning_rate": 4.6052697279658214e-06,
      "loss": 1.9791,
      "step": 217536
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.3965317010879517,
      "learning_rate": 4.60455576601253e-06,
      "loss": 1.8357,
      "step": 217537
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.3913350105285645,
      "learning_rate": 4.603841858978774e-06,
      "loss": 1.9533,
      "step": 217538
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.4685378074645996,
      "learning_rate": 4.603128006864787e-06,
      "loss": 2.0648,
      "step": 217539
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.3935284614562988,
      "learning_rate": 4.602414209670635e-06,
      "loss": 1.7291,
      "step": 217540
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.4111336469650269,
      "learning_rate": 4.6017004673964856e-06,
      "loss": 1.8055,
      "step": 217541
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.551084280014038,
      "learning_rate": 4.600986780042437e-06,
      "loss": 2.1439,
      "step": 217542
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.4597134590148926,
      "learning_rate": 4.6002731476086575e-06,
      "loss": 2.0925,
      "step": 217543
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.5022841691970825,
      "learning_rate": 4.599559570095246e-06,
      "loss": 1.9963,
      "step": 217544
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.4554463624954224,
      "learning_rate": 4.598846047502369e-06,
      "loss": 1.8115,
      "step": 217545
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.4749622344970703,
      "learning_rate": 4.598132579830127e-06,
      "loss": 2.0153,
      "step": 217546
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.4272829294204712,
      "learning_rate": 4.597419167078687e-06,
      "loss": 2.0838,
      "step": 217547
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.370944857597351,
      "learning_rate": 4.59670580924818e-06,
      "loss": 2.037,
      "step": 217548
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.4464093446731567,
      "learning_rate": 4.5959925063386745e-06,
      "loss": 2.0264,
      "step": 217549
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.4046837091445923,
      "learning_rate": 4.595279258350371e-06,
      "loss": 1.8251,
      "step": 217550
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.375478982925415,
      "learning_rate": 4.5945660652834e-06,
      "loss": 2.094,
      "step": 217551
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.4997631311416626,
      "learning_rate": 4.593852927137831e-06,
      "loss": 2.1161,
      "step": 217552
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.5808719396591187,
      "learning_rate": 4.593139843913862e-06,
      "loss": 1.9455,
      "step": 217553
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.4561922550201416,
      "learning_rate": 4.592426815611627e-06,
      "loss": 2.1389,
      "step": 217554
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.6228177547454834,
      "learning_rate": 4.591713842231226e-06,
      "loss": 1.9437,
      "step": 217555
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.490666389465332,
      "learning_rate": 4.591000923772792e-06,
      "loss": 2.2533,
      "step": 217556
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.485602617263794,
      "learning_rate": 4.590288060236458e-06,
      "loss": 1.8527,
      "step": 217557
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.2983918190002441,
      "learning_rate": 4.589575251622391e-06,
      "loss": 1.7663,
      "step": 217558
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.3862545490264893,
      "learning_rate": 4.58886249793069e-06,
      "loss": 2.1158,
      "step": 217559
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.6952261924743652,
      "learning_rate": 4.5881497991614895e-06,
      "loss": 2.231,
      "step": 217560
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.4313966035842896,
      "learning_rate": 4.5874371553149215e-06,
      "loss": 2.1802,
      "step": 217561
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.4002187252044678,
      "learning_rate": 4.58672456639112e-06,
      "loss": 2.0093,
      "step": 217562
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.3717050552368164,
      "learning_rate": 4.586012032390218e-06,
      "loss": 1.9376,
      "step": 217563
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.9447927474975586,
      "learning_rate": 4.585299553312383e-06,
      "loss": 1.8629,
      "step": 217564
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.5937168598175049,
      "learning_rate": 4.584587129157713e-06,
      "loss": 2.0565,
      "step": 217565
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.4984022378921509,
      "learning_rate": 4.583874759926342e-06,
      "loss": 1.9717,
      "step": 217566
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.388291358947754,
      "learning_rate": 4.583162445618405e-06,
      "loss": 2.0769,
      "step": 217567
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.4326391220092773,
      "learning_rate": 4.582450186234033e-06,
      "loss": 1.8019,
      "step": 217568
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.495418667793274,
      "learning_rate": 4.581737981773326e-06,
      "loss": 1.906,
      "step": 217569
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.6892281770706177,
      "learning_rate": 4.581025832236484e-06,
      "loss": 1.8893,
      "step": 217570
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.3683799505233765,
      "learning_rate": 4.580313737623609e-06,
      "loss": 1.8115,
      "step": 217571
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.3449451923370361,
      "learning_rate": 4.579601697934831e-06,
      "loss": 2.0176,
      "step": 217572
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.3692402839660645,
      "learning_rate": 4.578889713170286e-06,
      "loss": 2.0865,
      "step": 217573
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.5207983255386353,
      "learning_rate": 4.578177783330073e-06,
      "loss": 1.8819,
      "step": 217574
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.5257225036621094,
      "learning_rate": 4.577465908414357e-06,
      "loss": 2.2259,
      "step": 217575
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.4142324924468994,
      "learning_rate": 4.576754088423307e-06,
      "loss": 1.8551,
      "step": 217576
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.4075038433074951,
      "learning_rate": 4.576042323356954e-06,
      "loss": 2.0583,
      "step": 217577
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.4304953813552856,
      "learning_rate": 4.5753306132155665e-06,
      "loss": 1.9403,
      "step": 217578
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.3602031469345093,
      "learning_rate": 4.574618957999143e-06,
      "loss": 2.0305,
      "step": 217579
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.5369665622711182,
      "learning_rate": 4.573907357707885e-06,
      "loss": 1.8665,
      "step": 217580
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.5147196054458618,
      "learning_rate": 4.573195812341924e-06,
      "loss": 1.9133,
      "step": 217581
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.5548266172409058,
      "learning_rate": 4.572484321901393e-06,
      "loss": 1.8349,
      "step": 217582
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.4025897979736328,
      "learning_rate": 4.57177288638636e-06,
      "loss": 1.9664,
      "step": 217583
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.5697252750396729,
      "learning_rate": 4.571061505797091e-06,
      "loss": 1.9899,
      "step": 217584
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.4318046569824219,
      "learning_rate": 4.570350180133586e-06,
      "loss": 2.0174,
      "step": 217585
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.623632550239563,
      "learning_rate": 4.569638909396045e-06,
      "loss": 1.9513,
      "step": 217586
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.5149527788162231,
      "learning_rate": 4.568927693584568e-06,
      "loss": 1.8657,
      "step": 217587
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.5686875581741333,
      "learning_rate": 4.5682165326993206e-06,
      "loss": 1.8742,
      "step": 217588
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.3725980520248413,
      "learning_rate": 4.567505426740403e-06,
      "loss": 2.1127,
      "step": 217589
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.454922080039978,
      "learning_rate": 4.566794375708016e-06,
      "loss": 2.0396,
      "step": 217590
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.42752206325531,
      "learning_rate": 4.566083379602192e-06,
      "loss": 2.1551,
      "step": 217591
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.4398236274719238,
      "learning_rate": 4.565372438423099e-06,
      "loss": 2.056,
      "step": 217592
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.4870887994766235,
      "learning_rate": 4.564661552170901e-06,
      "loss": 1.7945,
      "step": 217593
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.5126560926437378,
      "learning_rate": 4.5639507208456995e-06,
      "loss": 1.9234,
      "step": 217594
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.474911093711853,
      "learning_rate": 4.563239944447628e-06,
      "loss": 1.9889,
      "step": 217595
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.5120927095413208,
      "learning_rate": 4.562529222976852e-06,
      "loss": 1.9334,
      "step": 217596
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.5096354484558105,
      "learning_rate": 4.561818556433472e-06,
      "loss": 1.939,
      "step": 217597
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.4952887296676636,
      "learning_rate": 4.561107944817622e-06,
      "loss": 1.9517,
      "step": 217598
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.6058685779571533,
      "learning_rate": 4.560397388129433e-06,
      "loss": 2.2909,
      "step": 217599
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.5750418901443481,
      "learning_rate": 4.559686886369074e-06,
      "loss": 1.8479,
      "step": 217600
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.5869568586349487,
      "learning_rate": 4.558976439536577e-06,
      "loss": 1.9064,
      "step": 217601
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.7109986543655396,
      "learning_rate": 4.558266047632209e-06,
      "loss": 2.0025,
      "step": 217602
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.4484517574310303,
      "learning_rate": 4.557555710656035e-06,
      "loss": 2.0054,
      "step": 217603
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.4458043575286865,
      "learning_rate": 4.556845428608158e-06,
      "loss": 1.9973,
      "step": 217604
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.5832756757736206,
      "learning_rate": 4.556135201488742e-06,
      "loss": 2.1192,
      "step": 217605
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.4909687042236328,
      "learning_rate": 4.5554250292979215e-06,
      "loss": 1.7911,
      "step": 217606
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.6513409614562988,
      "learning_rate": 4.554714912035828e-06,
      "loss": 1.8096,
      "step": 217607
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.4916115999221802,
      "learning_rate": 4.554004849702597e-06,
      "loss": 1.8504,
      "step": 217608
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.502786636352539,
      "learning_rate": 4.55329484229836e-06,
      "loss": 1.7746,
      "step": 217609
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.4173903465270996,
      "learning_rate": 4.552584889823219e-06,
      "loss": 1.927,
      "step": 217610
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.7687064409255981,
      "learning_rate": 4.5518749922773376e-06,
      "loss": 1.983,
      "step": 217611
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.4177939891815186,
      "learning_rate": 4.551165149660851e-06,
      "loss": 1.9165,
      "step": 217612
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.3891587257385254,
      "learning_rate": 4.550455361973859e-06,
      "loss": 2.019,
      "step": 217613
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.566016435623169,
      "learning_rate": 4.549745629216528e-06,
      "loss": 1.9074,
      "step": 217614
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.3897861242294312,
      "learning_rate": 4.549035951388991e-06,
      "loss": 1.9478,
      "step": 217615
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.442864179611206,
      "learning_rate": 4.548326328491348e-06,
      "loss": 1.9595,
      "step": 217616
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.4130847454071045,
      "learning_rate": 4.547616760523732e-06,
      "loss": 2.378,
      "step": 217617
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.3388075828552246,
      "learning_rate": 4.546907247486309e-06,
      "loss": 1.907,
      "step": 217618
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.5379748344421387,
      "learning_rate": 4.54619778937918e-06,
      "loss": 1.8561,
      "step": 217619
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.4678744077682495,
      "learning_rate": 4.545488386202512e-06,
      "loss": 1.8712,
      "step": 217620
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.4309998750686646,
      "learning_rate": 4.544779037956403e-06,
      "loss": 1.8291,
      "step": 217621
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.4832428693771362,
      "learning_rate": 4.544069744641022e-06,
      "loss": 2.0757,
      "step": 217622
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.7524842023849487,
      "learning_rate": 4.5433605062564325e-06,
      "loss": 1.7821,
      "step": 217623
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.4664160013198853,
      "learning_rate": 4.542651322802837e-06,
      "loss": 1.9065,
      "step": 217624
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.3828060626983643,
      "learning_rate": 4.541942194280335e-06,
      "loss": 2.0445,
      "step": 217625
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.5164510011672974,
      "learning_rate": 4.5412331206890584e-06,
      "loss": 1.8159,
      "step": 217626
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.527189016342163,
      "learning_rate": 4.540524102029142e-06,
      "loss": 2.1881,
      "step": 217627
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.3894239664077759,
      "learning_rate": 4.539815138300751e-06,
      "loss": 2.0647,
      "step": 217628
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.4526368379592896,
      "learning_rate": 4.539106229503919e-06,
      "loss": 1.8814,
      "step": 217629
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.5535703897476196,
      "learning_rate": 4.5383973756389135e-06,
      "loss": 2.132,
      "step": 217630
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.335378885269165,
      "learning_rate": 4.537688576705733e-06,
      "loss": 1.9108,
      "step": 217631
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.563956379890442,
      "learning_rate": 4.5369798327046124e-06,
      "loss": 2.1319,
      "step": 217632
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.437146544456482,
      "learning_rate": 4.53627114363565e-06,
      "loss": 1.887,
      "step": 217633
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.463255763053894,
      "learning_rate": 4.53556250949898e-06,
      "loss": 1.9094,
      "step": 217634
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.3099446296691895,
      "learning_rate": 4.534853930294702e-06,
      "loss": 1.9257,
      "step": 217635
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.4265576601028442,
      "learning_rate": 4.534145406022982e-06,
      "loss": 1.9999,
      "step": 217636
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.371705174446106,
      "learning_rate": 4.5334369366839205e-06,
      "loss": 1.7977,
      "step": 217637
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.554532527923584,
      "learning_rate": 4.532728522277684e-06,
      "loss": 1.9138,
      "step": 217638
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.5029770135879517,
      "learning_rate": 4.532020162804406e-06,
      "loss": 1.9763,
      "step": 217639
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.3851163387298584,
      "learning_rate": 4.531311858264186e-06,
      "loss": 1.8799,
      "step": 217640
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.4479066133499146,
      "learning_rate": 4.530603608657158e-06,
      "loss": 1.8915,
      "step": 217641
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.5416009426116943,
      "learning_rate": 4.529895413983519e-06,
      "loss": 1.9487,
      "step": 217642
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.6321516036987305,
      "learning_rate": 4.529187274243307e-06,
      "loss": 1.9551,
      "step": 217643
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.36264967918396,
      "learning_rate": 4.528479189436718e-06,
      "loss": 1.8302,
      "step": 217644
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.495671272277832,
      "learning_rate": 4.527771159563853e-06,
      "loss": 1.9729,
      "step": 217645
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.3740425109863281,
      "learning_rate": 4.527063184624813e-06,
      "loss": 1.9076,
      "step": 217646
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.5130105018615723,
      "learning_rate": 4.52635526461983e-06,
      "loss": 1.8943,
      "step": 217647
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.4293779134750366,
      "learning_rate": 4.5256473995489705e-06,
      "loss": 2.0287,
      "step": 217648
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.51710045337677,
      "learning_rate": 4.524939589412336e-06,
      "loss": 1.8493,
      "step": 217649
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.6194603443145752,
      "learning_rate": 4.524231834210123e-06,
      "loss": 1.9204,
      "step": 217650
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.4970500469207764,
      "learning_rate": 4.523524133942435e-06,
      "loss": 1.9133,
      "step": 217651
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.3989301919937134,
      "learning_rate": 4.5228164886093706e-06,
      "loss": 1.8477,
      "step": 217652
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.3841650485992432,
      "learning_rate": 4.522108898211097e-06,
      "loss": 1.8516,
      "step": 217653
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.4993503093719482,
      "learning_rate": 4.5214013627477455e-06,
      "loss": 1.9621,
      "step": 217654
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.5965973138809204,
      "learning_rate": 4.52069388221945e-06,
      "loss": 2.0227,
      "step": 217655
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.4597773551940918,
      "learning_rate": 4.519986456626345e-06,
      "loss": 2.0757,
      "step": 217656
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.535678744316101,
      "learning_rate": 4.519279085968564e-06,
      "loss": 2.1406,
      "step": 217657
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.5246397256851196,
      "learning_rate": 4.518571770246171e-06,
      "loss": 2.1262,
      "step": 217658
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.3822189569473267,
      "learning_rate": 4.517864509459401e-06,
      "loss": 2.1287,
      "step": 217659
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.4502595663070679,
      "learning_rate": 4.517157303608321e-06,
      "loss": 2.0486,
      "step": 217660
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.4912368059158325,
      "learning_rate": 4.516450152693063e-06,
      "loss": 1.8722,
      "step": 217661
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.3683223724365234,
      "learning_rate": 4.515743056713794e-06,
      "loss": 1.9968,
      "step": 217662
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.4619308710098267,
      "learning_rate": 4.5150360156706145e-06,
      "loss": 1.9506,
      "step": 217663
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.4906301498413086,
      "learning_rate": 4.5143290295636906e-06,
      "loss": 1.8447,
      "step": 217664
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.7001714706420898,
      "learning_rate": 4.513622098393121e-06,
      "loss": 2.0191,
      "step": 217665
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.6198416948318481,
      "learning_rate": 4.512915222159041e-06,
      "loss": 2.0024,
      "step": 217666
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.5869003534317017,
      "learning_rate": 4.512208400861617e-06,
      "loss": 2.2175,
      "step": 217667
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.605051875114441,
      "learning_rate": 4.51150163450088e-06,
      "loss": 1.9629,
      "step": 217668
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.5112611055374146,
      "learning_rate": 4.510794923077099e-06,
      "loss": 2.078,
      "step": 217669
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.4430195093154907,
      "learning_rate": 4.5100882665903394e-06,
      "loss": 1.923,
      "step": 217670
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.3896021842956543,
      "learning_rate": 4.509381665040701e-06,
      "loss": 1.7895,
      "step": 217671
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.71622633934021,
      "learning_rate": 4.508675118428351e-06,
      "loss": 1.9048,
      "step": 217672
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.4775277376174927,
      "learning_rate": 4.5079686267534225e-06,
      "loss": 2.0564,
      "step": 217673
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.445208191871643,
      "learning_rate": 4.507262190016014e-06,
      "loss": 1.8931,
      "step": 217674
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.4651610851287842,
      "learning_rate": 4.506555808216328e-06,
      "loss": 2.0946,
      "step": 217675
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.424125075340271,
      "learning_rate": 4.505849481354429e-06,
      "loss": 1.9968,
      "step": 217676
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.714106559753418,
      "learning_rate": 4.505143209430484e-06,
      "loss": 1.8919,
      "step": 217677
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.4935529232025146,
      "learning_rate": 4.504436992444593e-06,
      "loss": 2.1972,
      "step": 217678
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.3343409299850464,
      "learning_rate": 4.503730830396923e-06,
      "loss": 1.8334,
      "step": 217679
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.60441255569458,
      "learning_rate": 4.5030247232875405e-06,
      "loss": 2.2715,
      "step": 217680
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.397712230682373,
      "learning_rate": 4.5023186711166785e-06,
      "loss": 1.9363,
      "step": 217681
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.4799684286117554,
      "learning_rate": 4.501612673884403e-06,
      "loss": 1.7678,
      "step": 217682
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.4613158702850342,
      "learning_rate": 4.500906731590847e-06,
      "loss": 2.0999,
      "step": 217683
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.3770110607147217,
      "learning_rate": 4.500200844236146e-06,
      "loss": 1.9337,
      "step": 217684
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.411372423171997,
      "learning_rate": 4.4994950118204646e-06,
      "loss": 2.0709,
      "step": 217685
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.6068236827850342,
      "learning_rate": 4.498789234343836e-06,
      "loss": 2.1152,
      "step": 217686
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.4409492015838623,
      "learning_rate": 4.498083511806527e-06,
      "loss": 1.9876,
      "step": 217687
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.302240014076233,
      "learning_rate": 4.497377844208572e-06,
      "loss": 1.9752,
      "step": 217688
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.560502052307129,
      "learning_rate": 4.496672231550169e-06,
      "loss": 2.1168,
      "step": 217689
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.5348281860351562,
      "learning_rate": 4.495966673831353e-06,
      "loss": 1.9754,
      "step": 217690
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.4091750383377075,
      "learning_rate": 4.495261171052356e-06,
      "loss": 2.0218,
      "step": 217691
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.3325436115264893,
      "learning_rate": 4.4945557232132445e-06,
      "loss": 2.1905,
      "step": 217692
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.5467581748962402,
      "learning_rate": 4.493850330314186e-06,
      "loss": 2.0327,
      "step": 217693
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.640751600265503,
      "learning_rate": 4.493144992355313e-06,
      "loss": 1.8659,
      "step": 217694
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.5035479068756104,
      "learning_rate": 4.4924397093367595e-06,
      "loss": 2.1488,
      "step": 217695
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.579460620880127,
      "learning_rate": 4.491734481258591e-06,
      "loss": 1.9489,
      "step": 217696
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.3687623739242554,
      "learning_rate": 4.491029308121008e-06,
      "loss": 2.1532,
      "step": 217697
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.6280956268310547,
      "learning_rate": 4.490324189924111e-06,
      "loss": 1.8304,
      "step": 217698
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.307265281677246,
      "learning_rate": 4.4896191266680315e-06,
      "loss": 1.9218,
      "step": 217699
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.6161600351333618,
      "learning_rate": 4.4889141183529375e-06,
      "loss": 1.9653,
      "step": 217700
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.4380258321762085,
      "learning_rate": 4.488209164978929e-06,
      "loss": 1.8887,
      "step": 217701
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.4930464029312134,
      "learning_rate": 4.487504266546138e-06,
      "loss": 2.135,
      "step": 217702
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.3710122108459473,
      "learning_rate": 4.486799423054699e-06,
      "loss": 2.0152,
      "step": 217703
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.6441404819488525,
      "learning_rate": 4.486094634504711e-06,
      "loss": 1.8773,
      "step": 217704
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.6621966361999512,
      "learning_rate": 4.485389900896375e-06,
      "loss": 2.0853,
      "step": 217705
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.5627418756484985,
      "learning_rate": 4.4846852222297894e-06,
      "loss": 1.9763,
      "step": 217706
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.5142697095870972,
      "learning_rate": 4.483980598505055e-06,
      "loss": 2.0958,
      "step": 217707
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.59809148311615,
      "learning_rate": 4.483276029722305e-06,
      "loss": 2.0948,
      "step": 217708
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.396366000175476,
      "learning_rate": 4.482571515881739e-06,
      "loss": 1.7966,
      "step": 217709
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.3995048999786377,
      "learning_rate": 4.481867056983424e-06,
      "loss": 1.8485,
      "step": 217710
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.3287851810455322,
      "learning_rate": 4.4811626530274925e-06,
      "loss": 2.0285,
      "step": 217711
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.4488133192062378,
      "learning_rate": 4.480458304014112e-06,
      "loss": 1.9231,
      "step": 217712
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.6357861757278442,
      "learning_rate": 4.4797540099434145e-06,
      "loss": 1.9273,
      "step": 217713
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.6280564069747925,
      "learning_rate": 4.479049770815435e-06,
      "loss": 1.9938,
      "step": 217714
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.324180245399475,
      "learning_rate": 4.478345586630438e-06,
      "loss": 1.7539,
      "step": 217715
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.4607985019683838,
      "learning_rate": 4.477641457388459e-06,
      "loss": 2.0045,
      "step": 217716
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.5233687162399292,
      "learning_rate": 4.476937383089696e-06,
      "loss": 2.0666,
      "step": 217717
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.4760390520095825,
      "learning_rate": 4.476233363734249e-06,
      "loss": 2.1168,
      "step": 217718
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.5132501125335693,
      "learning_rate": 4.475529399322253e-06,
      "loss": 1.9847,
      "step": 217719
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.4126145839691162,
      "learning_rate": 4.474825489853806e-06,
      "loss": 1.887,
      "step": 217720
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.5012174844741821,
      "learning_rate": 4.474121635329109e-06,
      "loss": 2.2856,
      "step": 217721
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.4381769895553589,
      "learning_rate": 4.473417835748194e-06,
      "loss": 1.9488,
      "step": 217722
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.4572880268096924,
      "learning_rate": 4.472714091111296e-06,
      "loss": 1.9332,
      "step": 217723
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.5342556238174438,
      "learning_rate": 4.47201040141848e-06,
      "loss": 1.7907,
      "step": 217724
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.4137457609176636,
      "learning_rate": 4.4713067666698794e-06,
      "loss": 1.8551,
      "step": 217725
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.4595263004302979,
      "learning_rate": 4.470603186865662e-06,
      "loss": 2.1418,
      "step": 217726
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.5499374866485596,
      "learning_rate": 4.469899662005961e-06,
      "loss": 2.0854,
      "step": 217727
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.4703558683395386,
      "learning_rate": 4.46919619209084e-06,
      "loss": 1.8457,
      "step": 217728
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.5943330526351929,
      "learning_rate": 4.468492777120502e-06,
      "loss": 1.7804,
      "step": 217729
    },
    {
      "epoch": 2.83,
      "grad_norm": 1.2825171947479248,
      "learning_rate": 4.467789417095013e-06,
      "loss": 1.7596,
      "step": 217730
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.4607229232788086,
      "learning_rate": 4.467086112014573e-06,
      "loss": 1.8426,
      "step": 217731
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.5356353521347046,
      "learning_rate": 4.466382861879281e-06,
      "loss": 1.9958,
      "step": 217732
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.4755709171295166,
      "learning_rate": 4.4656796666892705e-06,
      "loss": 1.8457,
      "step": 217733
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.3953825235366821,
      "learning_rate": 4.464976526444642e-06,
      "loss": 2.0106,
      "step": 217734
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.6071583032608032,
      "learning_rate": 4.464273441145593e-06,
      "loss": 2.2219,
      "step": 217735
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.494747281074524,
      "learning_rate": 4.463570410792161e-06,
      "loss": 1.9645,
      "step": 217736
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.4438897371292114,
      "learning_rate": 4.4628674353845765e-06,
      "loss": 2.1106,
      "step": 217737
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.6004269123077393,
      "learning_rate": 4.462164514922939e-06,
      "loss": 1.8859,
      "step": 217738
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.5137146711349487,
      "learning_rate": 4.461461649407317e-06,
      "loss": 1.9222,
      "step": 217739
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.5367071628570557,
      "learning_rate": 4.4607588388379075e-06,
      "loss": 2.0559,
      "step": 217740
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.4434776306152344,
      "learning_rate": 4.460056083214813e-06,
      "loss": 1.9559,
      "step": 217741
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.5530729293823242,
      "learning_rate": 4.459353382538166e-06,
      "loss": 1.8787,
      "step": 217742
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.5424224138259888,
      "learning_rate": 4.458650736808134e-06,
      "loss": 1.9891,
      "step": 217743
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.407020092010498,
      "learning_rate": 4.457948146024815e-06,
      "loss": 1.857,
      "step": 217744
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.5559028387069702,
      "learning_rate": 4.457245610188309e-06,
      "loss": 2.1015,
      "step": 217745
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.51512610912323,
      "learning_rate": 4.456543129298784e-06,
      "loss": 1.8752,
      "step": 217746
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.4673126935958862,
      "learning_rate": 4.455840703356373e-06,
      "loss": 1.92,
      "step": 217747
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.4749720096588135,
      "learning_rate": 4.4551383323612075e-06,
      "loss": 1.9057,
      "step": 217748
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.4866411685943604,
      "learning_rate": 4.454436016313423e-06,
      "loss": 1.8123,
      "step": 217749
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.4908143281936646,
      "learning_rate": 4.4537337552131516e-06,
      "loss": 2.0381,
      "step": 217750
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.4569183588027954,
      "learning_rate": 4.45303154906046e-06,
      "loss": 1.743,
      "step": 217751
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.4364559650421143,
      "learning_rate": 4.452329397855548e-06,
      "loss": 2.0357,
      "step": 217752
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.4769046306610107,
      "learning_rate": 4.451627301598515e-06,
      "loss": 1.9658,
      "step": 217753
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.494539499282837,
      "learning_rate": 4.450925260289528e-06,
      "loss": 1.8119,
      "step": 217754
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.3724662065505981,
      "learning_rate": 4.450223273928688e-06,
      "loss": 2.0221,
      "step": 217755
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.4429737329483032,
      "learning_rate": 4.44952134251616e-06,
      "loss": 1.9888,
      "step": 217756
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.5353425741195679,
      "learning_rate": 4.448819466051978e-06,
      "loss": 1.8883,
      "step": 217757
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.471787452697754,
      "learning_rate": 4.448117644536375e-06,
      "loss": 2.0188,
      "step": 217758
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.4702750444412231,
      "learning_rate": 4.447415877969451e-06,
      "loss": 1.9909,
      "step": 217759
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.4533028602600098,
      "learning_rate": 4.446714166351306e-06,
      "loss": 2.1181,
      "step": 217760
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.527258038520813,
      "learning_rate": 4.44601250968214e-06,
      "loss": 1.8873,
      "step": 217761
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.4455722570419312,
      "learning_rate": 4.445310907962019e-06,
      "loss": 1.9778,
      "step": 217762
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.4206064939498901,
      "learning_rate": 4.444609361191076e-06,
      "loss": 2.017,
      "step": 217763
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.3654696941375732,
      "learning_rate": 4.443907869369479e-06,
      "loss": 1.8538,
      "step": 217764
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.4780207872390747,
      "learning_rate": 4.443206432497326e-06,
      "loss": 2.0945,
      "step": 217765
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.5414397716522217,
      "learning_rate": 4.442505050574751e-06,
      "loss": 2.245,
      "step": 217766
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.4889296293258667,
      "learning_rate": 4.441803723601922e-06,
      "loss": 2.1529,
      "step": 217767
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.4119336605072021,
      "learning_rate": 4.441102451578937e-06,
      "loss": 2.0444,
      "step": 217768
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.5494648218154907,
      "learning_rate": 4.440401234505897e-06,
      "loss": 1.9258,
      "step": 217769
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.5440144538879395,
      "learning_rate": 4.4397000723830015e-06,
      "loss": 2.0613,
      "step": 217770
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.375635027885437,
      "learning_rate": 4.438998965210316e-06,
      "loss": 1.9966,
      "step": 217771
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.530993103981018,
      "learning_rate": 4.438297912988009e-06,
      "loss": 2.1009,
      "step": 217772
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.4707658290863037,
      "learning_rate": 4.4375969157162125e-06,
      "loss": 1.9361,
      "step": 217773
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.5809741020202637,
      "learning_rate": 4.436895973395061e-06,
      "loss": 1.9208,
      "step": 217774
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.3815606832504272,
      "learning_rate": 4.436195086024652e-06,
      "loss": 2.2118,
      "step": 217775
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.3611490726470947,
      "learning_rate": 4.435494253605121e-06,
      "loss": 2.043,
      "step": 217776
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.580635905265808,
      "learning_rate": 4.4347934761366e-06,
      "loss": 2.0269,
      "step": 217777
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.5050586462020874,
      "learning_rate": 4.434092753619256e-06,
      "loss": 1.8943,
      "step": 217778
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.3826496601104736,
      "learning_rate": 4.4333920860532225e-06,
      "loss": 2.1348,
      "step": 217779
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.3754903078079224,
      "learning_rate": 4.4326914734385655e-06,
      "loss": 1.9629,
      "step": 217780
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.507752537727356,
      "learning_rate": 4.431990915775418e-06,
      "loss": 2.0027,
      "step": 217781
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.6016757488250732,
      "learning_rate": 4.431290413064015e-06,
      "loss": 2.0247,
      "step": 217782
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.5212807655334473,
      "learning_rate": 4.4305899653043545e-06,
      "loss": 2.0618,
      "step": 217783
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.5020434856414795,
      "learning_rate": 4.429889572496636e-06,
      "loss": 1.9034,
      "step": 217784
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.5275777578353882,
      "learning_rate": 4.429189234640995e-06,
      "loss": 2.0175,
      "step": 217785
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.5768568515777588,
      "learning_rate": 4.4284889517375636e-06,
      "loss": 1.9366,
      "step": 217786
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.433743953704834,
      "learning_rate": 4.4277887237864075e-06,
      "loss": 1.9748,
      "step": 217787
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.4972435235977173,
      "learning_rate": 4.427088550787761e-06,
      "loss": 1.8318,
      "step": 217788
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.535313367843628,
      "learning_rate": 4.426388432741656e-06,
      "loss": 2.1159,
      "step": 217789
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.543206810951233,
      "learning_rate": 4.425688369648295e-06,
      "loss": 2.1712,
      "step": 217790
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.381750464439392,
      "learning_rate": 4.424988361507742e-06,
      "loss": 1.8175,
      "step": 217791
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.3829935789108276,
      "learning_rate": 4.4242884083201975e-06,
      "loss": 1.7823,
      "step": 217792
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.4846705198287964,
      "learning_rate": 4.423588510085729e-06,
      "loss": 1.9118,
      "step": 217793
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.5856715440750122,
      "learning_rate": 4.422888666804536e-06,
      "loss": 2.0502,
      "step": 217794
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.5758060216903687,
      "learning_rate": 4.422188878476651e-06,
      "loss": 2.0922,
      "step": 217795
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.4615473747253418,
      "learning_rate": 4.421489145102308e-06,
      "loss": 2.0215,
      "step": 217796
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.3955435752868652,
      "learning_rate": 4.42078946668154e-06,
      "loss": 1.8478,
      "step": 217797
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.4494400024414062,
      "learning_rate": 4.4200898432146135e-06,
      "loss": 2.032,
      "step": 217798
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.4588059186935425,
      "learning_rate": 4.419390274701495e-06,
      "loss": 2.0256,
      "step": 217799
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.3869469165802002,
      "learning_rate": 4.418690761142418e-06,
      "loss": 1.8431,
      "step": 217800
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.545902967453003,
      "learning_rate": 4.417991302537449e-06,
      "loss": 1.7907,
      "step": 217801
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.5876660346984863,
      "learning_rate": 4.417291898886821e-06,
      "loss": 2.0406,
      "step": 217802
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.4025009870529175,
      "learning_rate": 4.416592550190534e-06,
      "loss": 1.9502,
      "step": 217803
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.5144081115722656,
      "learning_rate": 4.415893256448855e-06,
      "loss": 2.044,
      "step": 217804
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.5964895486831665,
      "learning_rate": 4.415194017661783e-06,
      "loss": 1.9991,
      "step": 217805
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.4433624744415283,
      "learning_rate": 4.414494833829518e-06,
      "loss": 2.0644,
      "step": 217806
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.4678494930267334,
      "learning_rate": 4.413795704952161e-06,
      "loss": 1.9528,
      "step": 217807
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.6181058883666992,
      "learning_rate": 4.413096631029878e-06,
      "loss": 2.0568,
      "step": 217808
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.4528263807296753,
      "learning_rate": 4.412397612062768e-06,
      "loss": 2.0349,
      "step": 217809
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.4813235998153687,
      "learning_rate": 4.411698648050999e-06,
      "loss": 2.0122,
      "step": 217810
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.3471964597702026,
      "learning_rate": 4.410999738994669e-06,
      "loss": 1.9639,
      "step": 217811
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.3219587802886963,
      "learning_rate": 4.410300884893914e-06,
      "loss": 2.0527,
      "step": 217812
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.4099303483963013,
      "learning_rate": 4.409602085748831e-06,
      "loss": 1.9789,
      "step": 217813
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.4934616088867188,
      "learning_rate": 4.408903341559622e-06,
      "loss": 2.0427,
      "step": 217814
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.4126629829406738,
      "learning_rate": 4.408204652326352e-06,
      "loss": 2.1742,
      "step": 217815
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.377160668373108,
      "learning_rate": 4.407506018049189e-06,
      "loss": 2.0458,
      "step": 217816
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.5058389902114868,
      "learning_rate": 4.406807438728266e-06,
      "loss": 1.956,
      "step": 217817
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.4564433097839355,
      "learning_rate": 4.406108914363682e-06,
      "loss": 1.8968,
      "step": 217818
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.4766377210617065,
      "learning_rate": 4.4054104449555705e-06,
      "loss": 1.7474,
      "step": 217819
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.501116156578064,
      "learning_rate": 4.404712030504098e-06,
      "loss": 2.1306,
      "step": 217820
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.423203706741333,
      "learning_rate": 4.4040136710093324e-06,
      "loss": 2.1199,
      "step": 217821
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.4529350996017456,
      "learning_rate": 4.403315366471471e-06,
      "loss": 1.9555,
      "step": 217822
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.3567442893981934,
      "learning_rate": 4.402617116890617e-06,
      "loss": 2.0539,
      "step": 217823
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.5308401584625244,
      "learning_rate": 4.401918922266867e-06,
      "loss": 2.1009,
      "step": 217824
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.4856630563735962,
      "learning_rate": 4.40122078260039e-06,
      "loss": 1.9958,
      "step": 217825
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.4485844373703003,
      "learning_rate": 4.400522697891318e-06,
      "loss": 2.1172,
      "step": 217826
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.3534537553787231,
      "learning_rate": 4.399824668139751e-06,
      "loss": 1.8282,
      "step": 217827
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.424854040145874,
      "learning_rate": 4.3991266933458555e-06,
      "loss": 2.1674,
      "step": 217828
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.604251742362976,
      "learning_rate": 4.398428773509766e-06,
      "loss": 2.2418,
      "step": 217829
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.5557844638824463,
      "learning_rate": 4.397730908631547e-06,
      "loss": 1.9401,
      "step": 217830
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.3974127769470215,
      "learning_rate": 4.397033098711367e-06,
      "loss": 1.8844,
      "step": 217831
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.5628938674926758,
      "learning_rate": 4.396335343749391e-06,
      "loss": 1.9805,
      "step": 217832
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.43572199344635,
      "learning_rate": 4.3956376437456865e-06,
      "loss": 1.7809,
      "step": 217833
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.5464047193527222,
      "learning_rate": 4.39493999870042e-06,
      "loss": 1.9873,
      "step": 217834
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.4683985710144043,
      "learning_rate": 4.394242408613724e-06,
      "loss": 2.0576,
      "step": 217835
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.5388751029968262,
      "learning_rate": 4.3935448734857e-06,
      "loss": 2.0038,
      "step": 217836
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.4571479558944702,
      "learning_rate": 4.392847393316512e-06,
      "loss": 2.0074,
      "step": 217837
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.4314954280853271,
      "learning_rate": 4.392149968106262e-06,
      "loss": 1.8985,
      "step": 217838
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.4196230173110962,
      "learning_rate": 4.391452597855116e-06,
      "loss": 1.9629,
      "step": 217839
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.4889805316925049,
      "learning_rate": 4.390755282563174e-06,
      "loss": 2.0875,
      "step": 217840
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.4351001977920532,
      "learning_rate": 4.390058022230569e-06,
      "loss": 1.7358,
      "step": 217841
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.4287155866622925,
      "learning_rate": 4.3893608168574015e-06,
      "loss": 1.9822,
      "step": 217842
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.4816182851791382,
      "learning_rate": 4.38866366644387e-06,
      "loss": 1.7858,
      "step": 217843
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.4518038034439087,
      "learning_rate": 4.387966570990009e-06,
      "loss": 2.1628,
      "step": 217844
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.371163249015808,
      "learning_rate": 4.387269530496085e-06,
      "loss": 1.9812,
      "step": 217845
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.8317790031433105,
      "learning_rate": 4.386572544962097e-06,
      "loss": 2.2196,
      "step": 217846
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.5159038305282593,
      "learning_rate": 4.385875614388246e-06,
      "loss": 1.8103,
      "step": 217847
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.4839175939559937,
      "learning_rate": 4.385178738774631e-06,
      "loss": 2.0163,
      "step": 217848
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.7390542030334473,
      "learning_rate": 4.384481918121385e-06,
      "loss": 1.9016,
      "step": 217849
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.5653873682022095,
      "learning_rate": 4.3837851524286425e-06,
      "loss": 2.1021,
      "step": 217850
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.518356204032898,
      "learning_rate": 4.383088441696536e-06,
      "loss": 1.9201,
      "step": 217851
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.541633129119873,
      "learning_rate": 4.382391785925199e-06,
      "loss": 2.0594,
      "step": 217852
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.3649961948394775,
      "learning_rate": 4.381695185114764e-06,
      "loss": 2.0489,
      "step": 217853
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.3639525175094604,
      "learning_rate": 4.380998639265332e-06,
      "loss": 1.7038,
      "step": 217854
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.3632638454437256,
      "learning_rate": 4.380302148377102e-06,
      "loss": 1.7426,
      "step": 217855
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.3580317497253418,
      "learning_rate": 4.379605712450074e-06,
      "loss": 1.8995,
      "step": 217856
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.6090983152389526,
      "learning_rate": 4.3789093314845145e-06,
      "loss": 1.8365,
      "step": 217857
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.4741270542144775,
      "learning_rate": 4.378213005480491e-06,
      "loss": 1.9861,
      "step": 217858
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.4839808940887451,
      "learning_rate": 4.377516734438135e-06,
      "loss": 1.9573,
      "step": 217859
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.5270302295684814,
      "learning_rate": 4.376820518357582e-06,
      "loss": 1.95,
      "step": 217860
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.459202527999878,
      "learning_rate": 4.376124357238931e-06,
      "loss": 2.1602,
      "step": 217861
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.4332424402236938,
      "learning_rate": 4.375428251082347e-06,
      "loss": 1.993,
      "step": 217862
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.4411870241165161,
      "learning_rate": 4.374732199887964e-06,
      "loss": 2.0776,
      "step": 217863
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.4712018966674805,
      "learning_rate": 4.374036203655884e-06,
      "loss": 1.8643,
      "step": 217864
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.483985424041748,
      "learning_rate": 4.3733402623863045e-06,
      "loss": 2.0633,
      "step": 217865
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.4095149040222168,
      "learning_rate": 4.37264437607926e-06,
      "loss": 1.7637,
      "step": 217866
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.4778075218200684,
      "learning_rate": 4.371948544734916e-06,
      "loss": 1.9936,
      "step": 217867
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.501219630241394,
      "learning_rate": 4.371252768353406e-06,
      "loss": 1.9847,
      "step": 217868
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.5187392234802246,
      "learning_rate": 4.370557046934864e-06,
      "loss": 1.8204,
      "step": 217869
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.484782099723816,
      "learning_rate": 4.369861380479423e-06,
      "loss": 1.8342,
      "step": 217870
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.424005389213562,
      "learning_rate": 4.369165768987215e-06,
      "loss": 1.8329,
      "step": 217871
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.4744651317596436,
      "learning_rate": 4.368470212458341e-06,
      "loss": 1.751,
      "step": 217872
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.458147406578064,
      "learning_rate": 4.367774710892968e-06,
      "loss": 1.9662,
      "step": 217873
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.3935796022415161,
      "learning_rate": 4.367079264291162e-06,
      "loss": 1.7345,
      "step": 217874
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.4099897146224976,
      "learning_rate": 4.366383872653123e-06,
      "loss": 1.9844,
      "step": 217875
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.3766448497772217,
      "learning_rate": 4.365688535978951e-06,
      "loss": 1.9505,
      "step": 217876
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.5064973831176758,
      "learning_rate": 4.364993254268812e-06,
      "loss": 1.9692,
      "step": 217877
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.3442765474319458,
      "learning_rate": 4.364298027522772e-06,
      "loss": 1.9117,
      "step": 217878
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.4195122718811035,
      "learning_rate": 4.3636028557409996e-06,
      "loss": 1.8863,
      "step": 217879
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.4141733646392822,
      "learning_rate": 4.362907738923593e-06,
      "loss": 2.0189,
      "step": 217880
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.4918378591537476,
      "learning_rate": 4.36221267707072e-06,
      "loss": 1.7968,
      "step": 217881
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.4749549627304077,
      "learning_rate": 4.361517670182479e-06,
      "loss": 2.0019,
      "step": 217882
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.4885435104370117,
      "learning_rate": 4.360822718259038e-06,
      "loss": 2.0377,
      "step": 217883
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.4653764963150024,
      "learning_rate": 4.360127821300463e-06,
      "loss": 2.116,
      "step": 217884
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.4126338958740234,
      "learning_rate": 4.359432979306954e-06,
      "loss": 2.0695,
      "step": 217885
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.464259147644043,
      "learning_rate": 4.358738192278577e-06,
      "loss": 2.0696,
      "step": 217886
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.5280512571334839,
      "learning_rate": 4.358043460215532e-06,
      "loss": 2.1234,
      "step": 217887
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.4632987976074219,
      "learning_rate": 4.357348783117853e-06,
      "loss": 1.9888,
      "step": 217888
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.6595325469970703,
      "learning_rate": 4.3566541609858065e-06,
      "loss": 2.0831,
      "step": 217889
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.4141532182693481,
      "learning_rate": 4.3559595938193584e-06,
      "loss": 2.0781,
      "step": 217890
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.399234652519226,
      "learning_rate": 4.355265081618742e-06,
      "loss": 2.1071,
      "step": 217891
    },
    {
      "epoch": 2.84,
      "grad_norm": 2.103860378265381,
      "learning_rate": 4.354570624384057e-06,
      "loss": 1.912,
      "step": 217892
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.5907756090164185,
      "learning_rate": 4.353876222115471e-06,
      "loss": 1.9725,
      "step": 217893
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.5518933534622192,
      "learning_rate": 4.35318187481305e-06,
      "loss": 2.1004,
      "step": 217894
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.40714430809021,
      "learning_rate": 4.35248758247696e-06,
      "loss": 2.1131,
      "step": 217895
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.5618125200271606,
      "learning_rate": 4.351793345107335e-06,
      "loss": 2.0799,
      "step": 217896
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.4707199335098267,
      "learning_rate": 4.351099162704275e-06,
      "loss": 1.8097,
      "step": 217897
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.492763876914978,
      "learning_rate": 4.350405035267912e-06,
      "loss": 2.0687,
      "step": 217898
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.7133054733276367,
      "learning_rate": 4.349710962798414e-06,
      "loss": 2.0002,
      "step": 217899
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.3125174045562744,
      "learning_rate": 4.34901694529588e-06,
      "loss": 2.1364,
      "step": 217900
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.4062089920043945,
      "learning_rate": 4.348322982760443e-06,
      "loss": 1.8096,
      "step": 217901
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.4609979391098022,
      "learning_rate": 4.347629075192272e-06,
      "loss": 1.883,
      "step": 217902
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.447501540184021,
      "learning_rate": 4.34693522259143e-06,
      "loss": 2.0303,
      "step": 217903
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.4098832607269287,
      "learning_rate": 4.346241424958052e-06,
      "loss": 2.1273,
      "step": 217904
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.4666426181793213,
      "learning_rate": 4.345547682292305e-06,
      "loss": 2.0422,
      "step": 217905
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.3986024856567383,
      "learning_rate": 4.344853994594289e-06,
      "loss": 1.9043,
      "step": 217906
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.5073400735855103,
      "learning_rate": 4.344160361864135e-06,
      "loss": 1.9334,
      "step": 217907
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.4476808309555054,
      "learning_rate": 4.343466784102012e-06,
      "loss": 1.8699,
      "step": 217908
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.5782620906829834,
      "learning_rate": 4.342773261308019e-06,
      "loss": 1.9049,
      "step": 217909
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.4647154808044434,
      "learning_rate": 4.342079793482289e-06,
      "loss": 1.9055,
      "step": 217910
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.4654040336608887,
      "learning_rate": 4.341386380624923e-06,
      "loss": 2.1067,
      "step": 217911
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.601539969444275,
      "learning_rate": 4.340693022736086e-06,
      "loss": 2.0374,
      "step": 217912
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.69656503200531,
      "learning_rate": 4.3399997198159116e-06,
      "loss": 2.0891,
      "step": 217913
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.583014726638794,
      "learning_rate": 4.339306471864501e-06,
      "loss": 2.009,
      "step": 217914
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.4447771310806274,
      "learning_rate": 4.338613278881953e-06,
      "loss": 2.1049,
      "step": 217915
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.4733531475067139,
      "learning_rate": 4.3379201408685e-06,
      "loss": 2.1476,
      "step": 217916
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.353210210800171,
      "learning_rate": 4.337227057824177e-06,
      "loss": 1.9493,
      "step": 217917
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.443625569343567,
      "learning_rate": 4.33653402974915e-06,
      "loss": 2.0943,
      "step": 217918
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.4186575412750244,
      "learning_rate": 4.335841056643552e-06,
      "loss": 1.9385,
      "step": 217919
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.620476484298706,
      "learning_rate": 4.335148138507516e-06,
      "loss": 2.0558,
      "step": 217920
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.4048610925674438,
      "learning_rate": 4.334455275341109e-06,
      "loss": 2.0251,
      "step": 217921
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.4659498929977417,
      "learning_rate": 4.33376246714453e-06,
      "loss": 2.1138,
      "step": 217922
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.332216739654541,
      "learning_rate": 4.33306971391788e-06,
      "loss": 2.0052,
      "step": 217923
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.5525341033935547,
      "learning_rate": 4.332377015661326e-06,
      "loss": 1.8925,
      "step": 217924
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.3926568031311035,
      "learning_rate": 4.331684372374933e-06,
      "loss": 1.926,
      "step": 217925
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.6344093084335327,
      "learning_rate": 4.330991784058868e-06,
      "loss": 2.0597,
      "step": 217926
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.459213376045227,
      "learning_rate": 4.330299250713265e-06,
      "loss": 1.8387,
      "step": 217927
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.3346593379974365,
      "learning_rate": 4.329606772338223e-06,
      "loss": 2.0433,
      "step": 217928
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.3898921012878418,
      "learning_rate": 4.328914348933876e-06,
      "loss": 1.8471,
      "step": 217929
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.3978873491287231,
      "learning_rate": 4.328221980500391e-06,
      "loss": 2.025,
      "step": 217930
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.581311583518982,
      "learning_rate": 4.3275296670378656e-06,
      "loss": 1.9535,
      "step": 217931
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.427775502204895,
      "learning_rate": 4.326837408546468e-06,
      "loss": 2.1723,
      "step": 217932
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.5462367534637451,
      "learning_rate": 4.326145205026232e-06,
      "loss": 1.9585,
      "step": 217933
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.405250072479248,
      "learning_rate": 4.325453056477357e-06,
      "loss": 2.184,
      "step": 217934
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.5233573913574219,
      "learning_rate": 4.324760962899976e-06,
      "loss": 2.0725,
      "step": 217935
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.4518460035324097,
      "learning_rate": 4.324068924294221e-06,
      "loss": 1.7459,
      "step": 217936
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.5160579681396484,
      "learning_rate": 4.323376940660161e-06,
      "loss": 2.0054,
      "step": 217937
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.3974919319152832,
      "learning_rate": 4.3226850119980285e-06,
      "loss": 1.9924,
      "step": 217938
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.3935447931289673,
      "learning_rate": 4.3219931383078225e-06,
      "loss": 1.9207,
      "step": 217939
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.4507110118865967,
      "learning_rate": 4.321301319589776e-06,
      "loss": 2.0608,
      "step": 217940
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.5603784322738647,
      "learning_rate": 4.320609555843957e-06,
      "loss": 2.1257,
      "step": 217941
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.4637070894241333,
      "learning_rate": 4.319917847070565e-06,
      "loss": 2.1477,
      "step": 217942
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.402089238166809,
      "learning_rate": 4.3192261932696315e-06,
      "loss": 1.81,
      "step": 217943
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.6112852096557617,
      "learning_rate": 4.3185345944413916e-06,
      "loss": 1.8125,
      "step": 217944
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.6069831848144531,
      "learning_rate": 4.317843050585845e-06,
      "loss": 2.0974,
      "step": 217945
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.5177640914916992,
      "learning_rate": 4.317151561703258e-06,
      "loss": 2.066,
      "step": 217946
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.424385905265808,
      "learning_rate": 4.316460127793631e-06,
      "loss": 1.8734,
      "step": 217947
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.4272854328155518,
      "learning_rate": 4.315768748857196e-06,
      "loss": 1.9041,
      "step": 217948
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.497079610824585,
      "learning_rate": 4.31507742489402e-06,
      "loss": 1.8299,
      "step": 217949
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.4845857620239258,
      "learning_rate": 4.314386155904304e-06,
      "loss": 1.9992,
      "step": 217950
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.6753898859024048,
      "learning_rate": 4.313694941888079e-06,
      "loss": 1.9219,
      "step": 217951
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.5270546674728394,
      "learning_rate": 4.313003782845548e-06,
      "loss": 1.7908,
      "step": 217952
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.459718108177185,
      "learning_rate": 4.312312678776775e-06,
      "loss": 2.0396,
      "step": 217953
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.4303597211837769,
      "learning_rate": 4.3116216296819606e-06,
      "loss": 1.8497,
      "step": 217954
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.4920272827148438,
      "learning_rate": 4.310930635561138e-06,
      "loss": 2.093,
      "step": 217955
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.5091354846954346,
      "learning_rate": 4.310239696414575e-06,
      "loss": 1.8195,
      "step": 217956
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.5636004209518433,
      "learning_rate": 4.30954881224227e-06,
      "loss": 1.9639,
      "step": 217957
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.3906974792480469,
      "learning_rate": 4.308857983044423e-06,
      "loss": 2.0245,
      "step": 217958
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.4638956785202026,
      "learning_rate": 4.3081672088211e-06,
      "loss": 1.8754,
      "step": 217959
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.3814541101455688,
      "learning_rate": 4.307476489572503e-06,
      "loss": 1.7957,
      "step": 217960
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.5265954732894897,
      "learning_rate": 4.3067858252987305e-06,
      "loss": 1.9641,
      "step": 217961
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.440737009048462,
      "learning_rate": 4.306095215999916e-06,
      "loss": 2.0421,
      "step": 217962
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.4253461360931396,
      "learning_rate": 4.3054046616761595e-06,
      "loss": 1.8901,
      "step": 217963
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.3688938617706299,
      "learning_rate": 4.304714162327627e-06,
      "loss": 2.1199,
      "step": 217964
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.5592304468154907,
      "learning_rate": 4.304023717954419e-06,
      "loss": 1.9195,
      "step": 217965
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.7376751899719238,
      "learning_rate": 4.303333328556668e-06,
      "loss": 2.1193,
      "step": 217966
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.4689921140670776,
      "learning_rate": 4.302642994134509e-06,
      "loss": 2.0066,
      "step": 217967
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.47118079662323,
      "learning_rate": 4.301952714688106e-06,
      "loss": 2.1212,
      "step": 217968
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.404342532157898,
      "learning_rate": 4.301262490217494e-06,
      "loss": 1.8764,
      "step": 217969
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.4294558763504028,
      "learning_rate": 4.300572320722906e-06,
      "loss": 2.0364,
      "step": 217970
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.5483685731887817,
      "learning_rate": 4.299882206204408e-06,
      "loss": 2.0263,
      "step": 217971
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.3321788311004639,
      "learning_rate": 4.299192146662134e-06,
      "loss": 1.8281,
      "step": 217972
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.42889404296875,
      "learning_rate": 4.298502142096216e-06,
      "loss": 1.9885,
      "step": 217973
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.5547175407409668,
      "learning_rate": 4.297812192506855e-06,
      "loss": 1.8069,
      "step": 217974
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.4797905683517456,
      "learning_rate": 4.297122297894018e-06,
      "loss": 1.9908,
      "step": 217975
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.388669490814209,
      "learning_rate": 4.2964324582580036e-06,
      "loss": 2.018,
      "step": 217976
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.382326602935791,
      "learning_rate": 4.295742673598812e-06,
      "loss": 2.1758,
      "step": 217977
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.4884763956069946,
      "learning_rate": 4.295052943916644e-06,
      "loss": 2.0263,
      "step": 217978
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.3592078685760498,
      "learning_rate": 4.2943632692115985e-06,
      "loss": 1.8699,
      "step": 217979
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.643179178237915,
      "learning_rate": 4.293673649483842e-06,
      "loss": 2.0187,
      "step": 217980
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.4617087841033936,
      "learning_rate": 4.292984084733442e-06,
      "loss": 2.1051,
      "step": 217981
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.4445583820343018,
      "learning_rate": 4.2922945749605646e-06,
      "loss": 2.1015,
      "step": 217982
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.4087698459625244,
      "learning_rate": 4.291605120165309e-06,
      "loss": 1.8014,
      "step": 217983
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.3995789289474487,
      "learning_rate": 4.290915720347876e-06,
      "loss": 2.1302,
      "step": 217984
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.4591730833053589,
      "learning_rate": 4.290226375508299e-06,
      "loss": 1.9666,
      "step": 217985
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.3562686443328857,
      "learning_rate": 4.289537085646777e-06,
      "loss": 2.1697,
      "step": 217986
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.6136634349822998,
      "learning_rate": 4.2888478507634105e-06,
      "loss": 2.0616,
      "step": 217987
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.5229917764663696,
      "learning_rate": 4.2881586708583325e-06,
      "loss": 1.9705,
      "step": 217988
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.3071246147155762,
      "learning_rate": 4.287469545931643e-06,
      "loss": 2.0433,
      "step": 217989
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.332204818725586,
      "learning_rate": 4.286780475983509e-06,
      "loss": 1.9228,
      "step": 217990
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.2836129665374756,
      "learning_rate": 4.286091461014063e-06,
      "loss": 2.0917,
      "step": 217991
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.4218472242355347,
      "learning_rate": 4.285402501023405e-06,
      "loss": 1.8445,
      "step": 217992
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.5960875749588013,
      "learning_rate": 4.284713596011668e-06,
      "loss": 1.8184,
      "step": 217993
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.3456412553787231,
      "learning_rate": 4.2840247459789866e-06,
      "loss": 2.0516,
      "step": 217994
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.378390908241272,
      "learning_rate": 4.283335950925459e-06,
      "loss": 1.9509,
      "step": 217995
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.4250653982162476,
      "learning_rate": 4.282647210851287e-06,
      "loss": 2.0686,
      "step": 217996
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.4995023012161255,
      "learning_rate": 4.281958525756501e-06,
      "loss": 2.0269,
      "step": 217997
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.404067873954773,
      "learning_rate": 4.2812698956413375e-06,
      "loss": 2.0704,
      "step": 217998
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.3422284126281738,
      "learning_rate": 4.280581320505827e-06,
      "loss": 1.9379,
      "step": 217999
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.465977430343628,
      "learning_rate": 4.279892800350171e-06,
      "loss": 1.8146,
      "step": 218000
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.5360199213027954,
      "learning_rate": 4.279204335174435e-06,
      "loss": 1.9538,
      "step": 218001
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.5949090719223022,
      "learning_rate": 4.278515924978821e-06,
      "loss": 2.1159,
      "step": 218002
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.5544511079788208,
      "learning_rate": 4.277827569763359e-06,
      "loss": 2.1115,
      "step": 218003
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.496140718460083,
      "learning_rate": 4.277139269528251e-06,
      "loss": 2.011,
      "step": 218004
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.5804826021194458,
      "learning_rate": 4.2764510242736305e-06,
      "loss": 2.0624,
      "step": 218005
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.4833917617797852,
      "learning_rate": 4.275762833999563e-06,
      "loss": 1.8871,
      "step": 218006
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.5406527519226074,
      "learning_rate": 4.275074698706216e-06,
      "loss": 1.9912,
      "step": 218007
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.4930086135864258,
      "learning_rate": 4.274386618393722e-06,
      "loss": 2.0143,
      "step": 218008
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.3792721033096313,
      "learning_rate": 4.2736985930622134e-06,
      "loss": 1.969,
      "step": 218009
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.3594366312026978,
      "learning_rate": 4.2730106227118255e-06,
      "loss": 2.0514,
      "step": 218010
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.4664943218231201,
      "learning_rate": 4.272322707342624e-06,
      "loss": 1.9331,
      "step": 218011
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.5843302011489868,
      "learning_rate": 4.271634846954808e-06,
      "loss": 1.9826,
      "step": 218012
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.3653265237808228,
      "learning_rate": 4.2709470415484444e-06,
      "loss": 2.1243,
      "step": 218013
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.528551697731018,
      "learning_rate": 4.270259291123734e-06,
      "loss": 1.8084,
      "step": 218014
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.338621735572815,
      "learning_rate": 4.269571595680743e-06,
      "loss": 1.9743,
      "step": 218015
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.5421323776245117,
      "learning_rate": 4.268883955219638e-06,
      "loss": 2.2529,
      "step": 218016
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.5282299518585205,
      "learning_rate": 4.268196369740518e-06,
      "loss": 1.7859,
      "step": 218017
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.3391177654266357,
      "learning_rate": 4.267508839243517e-06,
      "loss": 1.8033,
      "step": 218018
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.6470104455947876,
      "learning_rate": 4.266821363728768e-06,
      "loss": 1.8429,
      "step": 218019
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.512884259223938,
      "learning_rate": 4.266133943196404e-06,
      "loss": 2.1586,
      "step": 218020
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.4582946300506592,
      "learning_rate": 4.265446577646559e-06,
      "loss": 1.8997,
      "step": 218021
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.5211740732192993,
      "learning_rate": 4.264759267079298e-06,
      "loss": 1.9461,
      "step": 218022
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.4678183794021606,
      "learning_rate": 4.26407201149489e-06,
      "loss": 1.9266,
      "step": 218023
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.4820785522460938,
      "learning_rate": 4.2633848108933e-06,
      "loss": 2.2177,
      "step": 218024
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.5079307556152344,
      "learning_rate": 4.2626976652747616e-06,
      "loss": 1.8832,
      "step": 218025
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.4788801670074463,
      "learning_rate": 4.262010574639374e-06,
      "loss": 2.0347,
      "step": 218026
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.3871572017669678,
      "learning_rate": 4.261323538987238e-06,
      "loss": 2.2243,
      "step": 218027
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.359750747680664,
      "learning_rate": 4.26063655831852e-06,
      "loss": 2.1427,
      "step": 218028
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.443280577659607,
      "learning_rate": 4.259949632633353e-06,
      "loss": 2.0454,
      "step": 218029
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.3301559686660767,
      "learning_rate": 4.259262761931803e-06,
      "loss": 2.0404,
      "step": 218030
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.5393937826156616,
      "learning_rate": 4.2585759462140714e-06,
      "loss": 2.1522,
      "step": 218031
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.4217619895935059,
      "learning_rate": 4.257889185480223e-06,
      "loss": 1.9553,
      "step": 218032
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.4114246368408203,
      "learning_rate": 4.257202479730426e-06,
      "loss": 1.8457,
      "step": 218033
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.8672508001327515,
      "learning_rate": 4.256515828964779e-06,
      "loss": 2.0147,
      "step": 218034
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.544931173324585,
      "learning_rate": 4.255829233183483e-06,
      "loss": 1.8643,
      "step": 218035
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.5513042211532593,
      "learning_rate": 4.25514269238657e-06,
      "loss": 2.1506,
      "step": 218036
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.5700262784957886,
      "learning_rate": 4.254456206574241e-06,
      "loss": 2.0136,
      "step": 218037
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.568095088005066,
      "learning_rate": 4.253769775746529e-06,
      "loss": 2.0788,
      "step": 218038
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.5700627565383911,
      "learning_rate": 4.253083399903701e-06,
      "loss": 1.9518,
      "step": 218039
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.463718295097351,
      "learning_rate": 4.252397079045721e-06,
      "loss": 1.9267,
      "step": 218040
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.4751520156860352,
      "learning_rate": 4.251710813172893e-06,
      "loss": 1.8605,
      "step": 218041
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.4265676736831665,
      "learning_rate": 4.2510246022852135e-06,
      "loss": 2.0363,
      "step": 218042
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.5606614351272583,
      "learning_rate": 4.25033844638285e-06,
      "loss": 1.9646,
      "step": 218043
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.3662501573562622,
      "learning_rate": 4.249652345465904e-06,
      "loss": 1.905,
      "step": 218044
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.5002555847167969,
      "learning_rate": 4.248966299534573e-06,
      "loss": 1.8635,
      "step": 218045
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.4228876829147339,
      "learning_rate": 4.248280308588925e-06,
      "loss": 2.1825,
      "step": 218046
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.5236421823501587,
      "learning_rate": 4.247594372629126e-06,
      "loss": 1.9942,
      "step": 218047
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.4087201356887817,
      "learning_rate": 4.246908491655243e-06,
      "loss": 1.9606,
      "step": 218048
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.5105938911437988,
      "learning_rate": 4.246222665667476e-06,
      "loss": 2.0823,
      "step": 218049
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.378798007965088,
      "learning_rate": 4.24553689466589e-06,
      "loss": 1.7529,
      "step": 218050
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.5479090213775635,
      "learning_rate": 4.244851178650655e-06,
      "loss": 2.2211,
      "step": 218051
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.5808169841766357,
      "learning_rate": 4.2441655176219e-06,
      "loss": 2.0313,
      "step": 218052
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.4317376613616943,
      "learning_rate": 4.243479911579728e-06,
      "loss": 1.9334,
      "step": 218053
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.446977972984314,
      "learning_rate": 4.242794360524271e-06,
      "loss": 1.8579,
      "step": 218054
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.44971764087677,
      "learning_rate": 4.242108864455663e-06,
      "loss": 2.0828,
      "step": 218055
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.5751149654388428,
      "learning_rate": 4.241423423374035e-06,
      "loss": 1.8505,
      "step": 218056
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.467738389968872,
      "learning_rate": 4.24073803727949e-06,
      "loss": 2.1617,
      "step": 218057
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.3620718717575073,
      "learning_rate": 4.240052706172192e-06,
      "loss": 2.024,
      "step": 218058
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.5545605421066284,
      "learning_rate": 4.239367430052243e-06,
      "loss": 2.0501,
      "step": 218059
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.425683617591858,
      "learning_rate": 4.238682208919775e-06,
      "loss": 2.2114,
      "step": 218060
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.425621509552002,
      "learning_rate": 4.237997042774955e-06,
      "loss": 2.0811,
      "step": 218061
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.6127334833145142,
      "learning_rate": 4.237311931617815e-06,
      "loss": 1.9471,
      "step": 218062
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.441752552986145,
      "learning_rate": 4.23662687544859e-06,
      "loss": 1.9313,
      "step": 218063
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.5776115655899048,
      "learning_rate": 4.235941874267312e-06,
      "loss": 2.0655,
      "step": 218064
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.6135339736938477,
      "learning_rate": 4.235256928074216e-06,
      "loss": 1.7819,
      "step": 218065
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.4558533430099487,
      "learning_rate": 4.234572036869333e-06,
      "loss": 1.9089,
      "step": 218066
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.3717461824417114,
      "learning_rate": 4.23388720065283e-06,
      "loss": 1.7417,
      "step": 218067
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.5877678394317627,
      "learning_rate": 4.233202419424841e-06,
      "loss": 1.9576,
      "step": 218068
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.6026755571365356,
      "learning_rate": 4.232517693185467e-06,
      "loss": 1.7659,
      "step": 218069
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.432444453239441,
      "learning_rate": 4.231833021934838e-06,
      "loss": 2.0154,
      "step": 218070
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.6471774578094482,
      "learning_rate": 4.231148405673124e-06,
      "loss": 1.9006,
      "step": 218071
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.4290848970413208,
      "learning_rate": 4.230463844400422e-06,
      "loss": 2.0746,
      "step": 218072
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.5897142887115479,
      "learning_rate": 4.229779338116868e-06,
      "loss": 1.9293,
      "step": 218073
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.3478305339813232,
      "learning_rate": 4.229094886822526e-06,
      "loss": 2.0393,
      "step": 218074
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.4844735860824585,
      "learning_rate": 4.228410490517631e-06,
      "loss": 1.9675,
      "step": 218075
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.4547322988510132,
      "learning_rate": 4.227726149202215e-06,
      "loss": 1.8841,
      "step": 218076
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.334505319595337,
      "learning_rate": 4.227041862876512e-06,
      "loss": 1.8672,
      "step": 218077
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.5136219263076782,
      "learning_rate": 4.226357631540555e-06,
      "loss": 1.9827,
      "step": 218078
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.462049126625061,
      "learning_rate": 4.225673455194478e-06,
      "loss": 1.9806,
      "step": 218079
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.5189173221588135,
      "learning_rate": 4.224989333838447e-06,
      "loss": 1.8536,
      "step": 218080
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.509558081626892,
      "learning_rate": 4.2243052674725945e-06,
      "loss": 1.8832,
      "step": 218081
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.510275959968567,
      "learning_rate": 4.2236212560969875e-06,
      "loss": 2.1656,
      "step": 218082
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.4502804279327393,
      "learning_rate": 4.222937299711793e-06,
      "loss": 1.9867,
      "step": 218083
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.6458691358566284,
      "learning_rate": 4.2222533983171765e-06,
      "loss": 1.8223,
      "step": 218084
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.3683921098709106,
      "learning_rate": 4.221569551913206e-06,
      "loss": 1.7437,
      "step": 218085
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.433459758758545,
      "learning_rate": 4.220885760500014e-06,
      "loss": 2.053,
      "step": 218086
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.4814716577529907,
      "learning_rate": 4.220202024077768e-06,
      "loss": 1.9823,
      "step": 218087
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.4713503122329712,
      "learning_rate": 4.219518342646566e-06,
      "loss": 1.8119,
      "step": 218088
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.3446640968322754,
      "learning_rate": 4.2188347162065095e-06,
      "loss": 1.9776,
      "step": 218089
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.4811400175094604,
      "learning_rate": 4.2181511447577975e-06,
      "loss": 1.8445,
      "step": 218090
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.5153640508651733,
      "learning_rate": 4.217467628300497e-06,
      "loss": 2.1456,
      "step": 218091
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.3870245218276978,
      "learning_rate": 4.216784166834741e-06,
      "loss": 1.9921,
      "step": 218092
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.4942171573638916,
      "learning_rate": 4.216100760360697e-06,
      "loss": 2.0123,
      "step": 218093
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.54888916015625,
      "learning_rate": 4.21541740887843e-06,
      "loss": 1.9381,
      "step": 218094
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.533774495124817,
      "learning_rate": 4.214734112388108e-06,
      "loss": 2.1306,
      "step": 218095
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.5098414421081543,
      "learning_rate": 4.214050870889862e-06,
      "loss": 1.8339,
      "step": 218096
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.457222580909729,
      "learning_rate": 4.213367684383795e-06,
      "loss": 1.8966,
      "step": 218097
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.561341404914856,
      "learning_rate": 4.212684552870072e-06,
      "loss": 1.9322,
      "step": 218098
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.6122850179672241,
      "learning_rate": 4.2120014763487585e-06,
      "loss": 1.8481,
      "step": 218099
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.4903782606124878,
      "learning_rate": 4.211318454820023e-06,
      "loss": 2.1384,
      "step": 218100
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.5453592538833618,
      "learning_rate": 4.210635488283997e-06,
      "loss": 1.9701,
      "step": 218101
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.4256170988082886,
      "learning_rate": 4.209952576740816e-06,
      "loss": 2.0267,
      "step": 218102
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.476769208908081,
      "learning_rate": 4.209269720190544e-06,
      "loss": 1.9129,
      "step": 218103
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.3722813129425049,
      "learning_rate": 4.208586918633416e-06,
      "loss": 1.8029,
      "step": 218104
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.6486812829971313,
      "learning_rate": 4.207904172069432e-06,
      "loss": 2.0594,
      "step": 218105
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.5569161176681519,
      "learning_rate": 4.207221480498824e-06,
      "loss": 2.0809,
      "step": 218106
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.6140315532684326,
      "learning_rate": 4.206538843921625e-06,
      "loss": 1.8756,
      "step": 218107
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.4452952146530151,
      "learning_rate": 4.205856262338103e-06,
      "loss": 1.8983,
      "step": 218108
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.6208235025405884,
      "learning_rate": 4.205173735748224e-06,
      "loss": 2.0501,
      "step": 218109
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.460713267326355,
      "learning_rate": 4.204491264152221e-06,
      "loss": 1.9688,
      "step": 218110
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.5649869441986084,
      "learning_rate": 4.2038088475501275e-06,
      "loss": 1.8415,
      "step": 218111
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.4565329551696777,
      "learning_rate": 4.20312648594221e-06,
      "loss": 1.941,
      "step": 218112
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.4037829637527466,
      "learning_rate": 4.2024441793284345e-06,
      "loss": 1.9601,
      "step": 218113
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.4660712480545044,
      "learning_rate": 4.201761927709102e-06,
      "loss": 1.7895,
      "step": 218114
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.691126823425293,
      "learning_rate": 4.201079731084178e-06,
      "loss": 2.0619,
      "step": 218115
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.574589490890503,
      "learning_rate": 4.200397589453863e-06,
      "loss": 2.0566,
      "step": 218116
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.4277130365371704,
      "learning_rate": 4.199715502818291e-06,
      "loss": 1.9563,
      "step": 218117
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.37118661403656,
      "learning_rate": 4.199033471177593e-06,
      "loss": 1.949,
      "step": 218118
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.5257089138031006,
      "learning_rate": 4.1983514945318375e-06,
      "loss": 1.86,
      "step": 218119
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.4737080335617065,
      "learning_rate": 4.197669572881224e-06,
      "loss": 1.9438,
      "step": 218120
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.562238097190857,
      "learning_rate": 4.196987706225852e-06,
      "loss": 2.0687,
      "step": 218121
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.4526766538619995,
      "learning_rate": 4.196305894565821e-06,
      "loss": 1.8853,
      "step": 218122
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.4919068813323975,
      "learning_rate": 4.195624137901266e-06,
      "loss": 2.0356,
      "step": 218123
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.4223707914352417,
      "learning_rate": 4.194942436232352e-06,
      "loss": 2.0258,
      "step": 218124
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.4056040048599243,
      "learning_rate": 4.194260789559179e-06,
      "loss": 1.8486,
      "step": 218125
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.4402729272842407,
      "learning_rate": 4.19357919788188e-06,
      "loss": 1.8869,
      "step": 218126
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.6223493814468384,
      "learning_rate": 4.192897661200589e-06,
      "loss": 2.2578,
      "step": 218127
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.4568886756896973,
      "learning_rate": 4.192216179515407e-06,
      "loss": 1.9662,
      "step": 218128
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.3933634757995605,
      "learning_rate": 4.191534752826464e-06,
      "loss": 1.8353,
      "step": 218129
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.3503280878067017,
      "learning_rate": 4.19085338113393e-06,
      "loss": 2.1674,
      "step": 218130
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.5391634702682495,
      "learning_rate": 4.190172064437869e-06,
      "loss": 2.0079,
      "step": 218131
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.4316575527191162,
      "learning_rate": 4.189490802738449e-06,
      "loss": 1.8795,
      "step": 218132
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.5324827432632446,
      "learning_rate": 4.1888095960358025e-06,
      "loss": 2.0448,
      "step": 218133
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.4720702171325684,
      "learning_rate": 4.18812844433003e-06,
      "loss": 1.985,
      "step": 218134
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.4999974966049194,
      "learning_rate": 4.187447347621265e-06,
      "loss": 2.014,
      "step": 218135
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.5031887292861938,
      "learning_rate": 4.186766305909639e-06,
      "loss": 2.0413,
      "step": 218136
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.73129141330719,
      "learning_rate": 4.186085319195287e-06,
      "loss": 1.8261,
      "step": 218137
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.5408263206481934,
      "learning_rate": 4.185404387478308e-06,
      "loss": 2.2135,
      "step": 218138
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.3905048370361328,
      "learning_rate": 4.184723510758869e-06,
      "loss": 1.8468,
      "step": 218139
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.3901816606521606,
      "learning_rate": 4.184042689037071e-06,
      "loss": 1.8873,
      "step": 218140
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.3449691534042358,
      "learning_rate": 4.183361922313011e-06,
      "loss": 2.1462,
      "step": 218141
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.3857743740081787,
      "learning_rate": 4.1826812105868575e-06,
      "loss": 2.2385,
      "step": 218142
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.5182803869247437,
      "learning_rate": 4.182000553858745e-06,
      "loss": 2.1979,
      "step": 218143
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.3907995223999023,
      "learning_rate": 4.18131995212877e-06,
      "loss": 2.1089,
      "step": 218144
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.667014479637146,
      "learning_rate": 4.180639405397102e-06,
      "loss": 2.0293,
      "step": 218145
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.5231740474700928,
      "learning_rate": 4.1799589136638055e-06,
      "loss": 2.0001,
      "step": 218146
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.6903764009475708,
      "learning_rate": 4.179278476929049e-06,
      "loss": 1.9078,
      "step": 218147
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.5116424560546875,
      "learning_rate": 4.178598095192931e-06,
      "loss": 2.2587,
      "step": 218148
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.5654888153076172,
      "learning_rate": 4.1779177684556185e-06,
      "loss": 2.0021,
      "step": 218149
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.4299867153167725,
      "learning_rate": 4.177237496717212e-06,
      "loss": 2.2288,
      "step": 218150
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.3982423543930054,
      "learning_rate": 4.176557279977843e-06,
      "loss": 1.9453,
      "step": 218151
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.4510365724563599,
      "learning_rate": 4.175877118237614e-06,
      "loss": 1.6764,
      "step": 218152
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.461469292640686,
      "learning_rate": 4.17519701149669e-06,
      "loss": 1.992,
      "step": 218153
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.456086277961731,
      "learning_rate": 4.1745169597551695e-06,
      "loss": 2.0812,
      "step": 218154
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.4431713819503784,
      "learning_rate": 4.173836963013188e-06,
      "loss": 1.7411,
      "step": 218155
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.5396500825881958,
      "learning_rate": 4.173157021270912e-06,
      "loss": 1.8173,
      "step": 218156
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.3968571424484253,
      "learning_rate": 4.172477134528407e-06,
      "loss": 1.9791,
      "step": 218157
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.4054456949234009,
      "learning_rate": 4.17179730278584e-06,
      "loss": 1.8839,
      "step": 218158
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.5307658910751343,
      "learning_rate": 4.1711175260432775e-06,
      "loss": 2.0112,
      "step": 218159
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.4654715061187744,
      "learning_rate": 4.170437804300919e-06,
      "loss": 2.0102,
      "step": 218160
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.4527887105941772,
      "learning_rate": 4.169758137558832e-06,
      "loss": 1.8693,
      "step": 218161
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.3425581455230713,
      "learning_rate": 4.169078525817215e-06,
      "loss": 2.0651,
      "step": 218162
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.5416717529296875,
      "learning_rate": 4.168398969076137e-06,
      "loss": 2.1557,
      "step": 218163
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.6547762155532837,
      "learning_rate": 4.167719467335729e-06,
      "loss": 1.8614,
      "step": 218164
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.5674837827682495,
      "learning_rate": 4.167040020596091e-06,
      "loss": 1.9656,
      "step": 218165
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.4262067079544067,
      "learning_rate": 4.166360628857457e-06,
      "loss": 1.8689,
      "step": 218166
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.3421434164047241,
      "learning_rate": 4.1656812921198265e-06,
      "loss": 1.9742,
      "step": 218167
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.5135220289230347,
      "learning_rate": 4.1650020103834e-06,
      "loss": 2.0151,
      "step": 218168
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.4924486875534058,
      "learning_rate": 4.164322783648277e-06,
      "loss": 1.9133,
      "step": 218169
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.4335124492645264,
      "learning_rate": 4.163643611914591e-06,
      "loss": 1.9509,
      "step": 218170
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.4279884099960327,
      "learning_rate": 4.162964495182475e-06,
      "loss": 1.9227,
      "step": 218171
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.4015979766845703,
      "learning_rate": 4.162285433452061e-06,
      "loss": 1.9023,
      "step": 218172
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.5065003633499146,
      "learning_rate": 4.161606426723418e-06,
      "loss": 2.1515,
      "step": 218173
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.357905387878418,
      "learning_rate": 4.160927474996745e-06,
      "loss": 1.9016,
      "step": 218174
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.4237854480743408,
      "learning_rate": 4.160248578272141e-06,
      "loss": 2.0016,
      "step": 218175
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.4728775024414062,
      "learning_rate": 4.159569736549739e-06,
      "loss": 1.9497,
      "step": 218176
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.4561432600021362,
      "learning_rate": 4.158890949829641e-06,
      "loss": 1.9109,
      "step": 218177
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.6110785007476807,
      "learning_rate": 4.158212218112011e-06,
      "loss": 1.8534,
      "step": 218178
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.4973769187927246,
      "learning_rate": 4.157533541396918e-06,
      "loss": 1.7551,
      "step": 218179
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.5332820415496826,
      "learning_rate": 4.156854919684561e-06,
      "loss": 2.0133,
      "step": 218180
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.4633852243423462,
      "learning_rate": 4.156176352975038e-06,
      "loss": 1.9251,
      "step": 218181
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.4455276727676392,
      "learning_rate": 4.155497841268418e-06,
      "loss": 1.7897,
      "step": 218182
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.5054521560668945,
      "learning_rate": 4.1548193845649e-06,
      "loss": 2.07,
      "step": 218183
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.492103934288025,
      "learning_rate": 4.154140982864585e-06,
      "loss": 1.9412,
      "step": 218184
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.4077038764953613,
      "learning_rate": 4.153462636167604e-06,
      "loss": 1.9744,
      "step": 218185
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.613311529159546,
      "learning_rate": 4.152784344474091e-06,
      "loss": 1.9513,
      "step": 218186
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.4899003505706787,
      "learning_rate": 4.152106107784147e-06,
      "loss": 1.9554,
      "step": 218187
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.49020254611969,
      "learning_rate": 4.151427926097873e-06,
      "loss": 2.005,
      "step": 218188
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.54713773727417,
      "learning_rate": 4.150749799415498e-06,
      "loss": 1.8496,
      "step": 218189
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.5262815952301025,
      "learning_rate": 4.150071727737025e-06,
      "loss": 1.9991,
      "step": 218190
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.4674845933914185,
      "learning_rate": 4.149393711062687e-06,
      "loss": 1.7867,
      "step": 218191
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.6492120027542114,
      "learning_rate": 4.148715749392517e-06,
      "loss": 2.1077,
      "step": 218192
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.624746322631836,
      "learning_rate": 4.148037842726715e-06,
      "loss": 2.0146,
      "step": 218193
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.3758580684661865,
      "learning_rate": 4.14735999106538e-06,
      "loss": 2.0313,
      "step": 218194
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.4291738271713257,
      "learning_rate": 4.146682194408646e-06,
      "loss": 1.8266,
      "step": 218195
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.535475254058838,
      "learning_rate": 4.146004452756579e-06,
      "loss": 1.9954,
      "step": 218196
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.3947690725326538,
      "learning_rate": 4.145326766109414e-06,
      "loss": 1.9529,
      "step": 218197
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.4346855878829956,
      "learning_rate": 4.144649134467149e-06,
      "loss": 2.0234,
      "step": 218198
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.6223948001861572,
      "learning_rate": 4.1439715578300505e-06,
      "loss": 1.9538,
      "step": 218199
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.5265933275222778,
      "learning_rate": 4.143294036198119e-06,
      "loss": 2.1027,
      "step": 218200
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.4136073589324951,
      "learning_rate": 4.142616569571588e-06,
      "loss": 1.9574,
      "step": 218201
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.5048937797546387,
      "learning_rate": 4.141939157950458e-06,
      "loss": 1.9021,
      "step": 218202
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.490066647529602,
      "learning_rate": 4.141261801334994e-06,
      "loss": 1.8624,
      "step": 218203
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.542091965675354,
      "learning_rate": 4.140584499725197e-06,
      "loss": 1.8633,
      "step": 218204
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.3634898662567139,
      "learning_rate": 4.139907253121299e-06,
      "loss": 2.0012,
      "step": 218205
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.4365673065185547,
      "learning_rate": 4.139230061523369e-06,
      "loss": 1.9504,
      "step": 218206
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.5164257287979126,
      "learning_rate": 4.1385529249315046e-06,
      "loss": 2.1134,
      "step": 218207
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.596946358680725,
      "learning_rate": 4.137875843345906e-06,
      "loss": 2.1283,
      "step": 218208
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.5490224361419678,
      "learning_rate": 4.13719881676664e-06,
      "loss": 1.9466,
      "step": 218209
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.3893707990646362,
      "learning_rate": 4.136521845193841e-06,
      "loss": 1.7518,
      "step": 218210
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.3523753881454468,
      "learning_rate": 4.135844928627674e-06,
      "loss": 1.9947,
      "step": 218211
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.326242446899414,
      "learning_rate": 4.13516806706824e-06,
      "loss": 1.899,
      "step": 218212
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.328669548034668,
      "learning_rate": 4.134491260515671e-06,
      "loss": 2.0332,
      "step": 218213
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.4120458364486694,
      "learning_rate": 4.133814508970035e-06,
      "loss": 2.1358,
      "step": 218214
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.3610795736312866,
      "learning_rate": 4.133137812431564e-06,
      "loss": 2.195,
      "step": 218215
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.5427345037460327,
      "learning_rate": 4.132461170900292e-06,
      "loss": 2.0087,
      "step": 218216
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.4733816385269165,
      "learning_rate": 4.131784584376385e-06,
      "loss": 1.8716,
      "step": 218217
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.3739162683486938,
      "learning_rate": 4.1311080528599774e-06,
      "loss": 2.066,
      "step": 218218
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.5723408460617065,
      "learning_rate": 4.130431576351201e-06,
      "loss": 1.7812,
      "step": 218219
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.4872854948043823,
      "learning_rate": 4.1297551548501225e-06,
      "loss": 1.7621,
      "step": 218220
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.6400126218795776,
      "learning_rate": 4.12907878835691e-06,
      "loss": 2.0364,
      "step": 218221
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.4689682722091675,
      "learning_rate": 4.128402476871695e-06,
      "loss": 1.9687,
      "step": 218222
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.453020691871643,
      "learning_rate": 4.127726220394612e-06,
      "loss": 2.0235,
      "step": 218223
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.5876466035842896,
      "learning_rate": 4.127050018925759e-06,
      "loss": 1.9658,
      "step": 218224
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.437454104423523,
      "learning_rate": 4.126373872465272e-06,
      "loss": 1.7582,
      "step": 218225
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.4977425336837769,
      "learning_rate": 4.125697781013282e-06,
      "loss": 1.937,
      "step": 218226
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.4656978845596313,
      "learning_rate": 4.125021744569889e-06,
      "loss": 1.8827,
      "step": 218227
    },
    {
      "epoch": 2.84,
      "grad_norm": 2.0028927326202393,
      "learning_rate": 4.1243457631352616e-06,
      "loss": 2.1057,
      "step": 218228
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.4248850345611572,
      "learning_rate": 4.123669836709498e-06,
      "loss": 1.9016,
      "step": 218229
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.5299335718154907,
      "learning_rate": 4.122993965292731e-06,
      "loss": 2.0457,
      "step": 218230
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.4172309637069702,
      "learning_rate": 4.122318148885096e-06,
      "loss": 1.9694,
      "step": 218231
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.4523706436157227,
      "learning_rate": 4.121642387486657e-06,
      "loss": 1.9364,
      "step": 218232
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.5014724731445312,
      "learning_rate": 4.120966681097648e-06,
      "loss": 2.0397,
      "step": 218233
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.5711948871612549,
      "learning_rate": 4.120291029718103e-06,
      "loss": 2.1903,
      "step": 218234
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.471308946609497,
      "learning_rate": 4.119615433348189e-06,
      "loss": 1.8286,
      "step": 218235
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.4668837785720825,
      "learning_rate": 4.118939891988038e-06,
      "loss": 1.7433,
      "step": 218236
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.4775385856628418,
      "learning_rate": 4.11826440563775e-06,
      "loss": 2.1058,
      "step": 218237
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.4576985836029053,
      "learning_rate": 4.117588974297426e-06,
      "loss": 1.8352,
      "step": 218238
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.484720230102539,
      "learning_rate": 4.116913597967298e-06,
      "loss": 2.1497,
      "step": 218239
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.3362020254135132,
      "learning_rate": 4.116238276647365e-06,
      "loss": 1.8484,
      "step": 218240
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.536093831062317,
      "learning_rate": 4.11556301033783e-06,
      "loss": 2.0615,
      "step": 218241
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.5216991901397705,
      "learning_rate": 4.11488779903879e-06,
      "loss": 2.0684,
      "step": 218242
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.3803101778030396,
      "learning_rate": 4.114212642750414e-06,
      "loss": 1.9837,
      "step": 218243
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.406467318534851,
      "learning_rate": 4.1135375414727335e-06,
      "loss": 1.8674,
      "step": 218244
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.4416776895523071,
      "learning_rate": 4.1128624952059815e-06,
      "loss": 1.8081,
      "step": 218245
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.6451550722122192,
      "learning_rate": 4.1121875039501925e-06,
      "loss": 2.0413,
      "step": 218246
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.398371696472168,
      "learning_rate": 4.111512567705533e-06,
      "loss": 1.9571,
      "step": 218247
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.400183916091919,
      "learning_rate": 4.1108376864721686e-06,
      "loss": 2.3443,
      "step": 218248
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.6141176223754883,
      "learning_rate": 4.110162860250166e-06,
      "loss": 1.9812,
      "step": 218249
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.5080333948135376,
      "learning_rate": 4.109488089039659e-06,
      "loss": 2.2169,
      "step": 218250
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.4290962219238281,
      "learning_rate": 4.108813372840813e-06,
      "loss": 1.9356,
      "step": 218251
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.6003632545471191,
      "learning_rate": 4.108138711653664e-06,
      "loss": 2.1565,
      "step": 218252
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.4351998567581177,
      "learning_rate": 4.107464105478442e-06,
      "loss": 1.9461,
      "step": 218253
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.46342134475708,
      "learning_rate": 4.106789554315248e-06,
      "loss": 2.1254,
      "step": 218254
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.4750573635101318,
      "learning_rate": 4.10611505816415e-06,
      "loss": 1.8424,
      "step": 218255
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.4799408912658691,
      "learning_rate": 4.105440617025313e-06,
      "loss": 1.9434,
      "step": 218256
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.427127718925476,
      "learning_rate": 4.10476623089887e-06,
      "loss": 1.8227,
      "step": 218257
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.5051220655441284,
      "learning_rate": 4.104091899784922e-06,
      "loss": 2.0014,
      "step": 218258
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.4511945247650146,
      "learning_rate": 4.103417623683636e-06,
      "loss": 2.0238,
      "step": 218259
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.5021864175796509,
      "learning_rate": 4.1027434025950764e-06,
      "loss": 1.9466,
      "step": 218260
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.4569478034973145,
      "learning_rate": 4.102069236519445e-06,
      "loss": 1.9874,
      "step": 218261
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.468855857849121,
      "learning_rate": 4.101395125456808e-06,
      "loss": 1.9663,
      "step": 218262
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.4850001335144043,
      "learning_rate": 4.100721069407298e-06,
      "loss": 1.9126,
      "step": 218263
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.4634718894958496,
      "learning_rate": 4.1000470683710485e-06,
      "loss": 2.1225,
      "step": 218264
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.514991044998169,
      "learning_rate": 4.099373122348193e-06,
      "loss": 1.8088,
      "step": 218265
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.3039709329605103,
      "learning_rate": 4.0986992313388315e-06,
      "loss": 1.8668,
      "step": 218266
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.409419298171997,
      "learning_rate": 4.09802539534313e-06,
      "loss": 1.9612,
      "step": 218267
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.4228652715682983,
      "learning_rate": 4.097351614361188e-06,
      "loss": 1.7811,
      "step": 218268
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.5314041376113892,
      "learning_rate": 4.0966778883931075e-06,
      "loss": 1.8729,
      "step": 218269
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.3265305757522583,
      "learning_rate": 4.096004217439086e-06,
      "loss": 1.8308,
      "step": 218270
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.525089979171753,
      "learning_rate": 4.095330601499158e-06,
      "loss": 2.2587,
      "step": 218271
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.5672399997711182,
      "learning_rate": 4.0946570405735234e-06,
      "loss": 2.026,
      "step": 218272
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.4289968013763428,
      "learning_rate": 4.093983534662248e-06,
      "loss": 1.9091,
      "step": 218273
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.3418850898742676,
      "learning_rate": 4.093310083765533e-06,
      "loss": 1.9985,
      "step": 218274
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.5741245746612549,
      "learning_rate": 4.092636687883411e-06,
      "loss": 2.0365,
      "step": 218275
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.3999711275100708,
      "learning_rate": 4.091963347016081e-06,
      "loss": 1.9197,
      "step": 218276
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.5737032890319824,
      "learning_rate": 4.091290061163644e-06,
      "loss": 1.8755,
      "step": 218277
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.418472409248352,
      "learning_rate": 4.090616830326199e-06,
      "loss": 2.06,
      "step": 218278
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.4238985776901245,
      "learning_rate": 4.089943654503913e-06,
      "loss": 1.84,
      "step": 218279
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.5196139812469482,
      "learning_rate": 4.089270533696887e-06,
      "loss": 1.9233,
      "step": 218280
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.4572004079818726,
      "learning_rate": 4.088597467905252e-06,
      "loss": 1.9143,
      "step": 218281
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.4765762090682983,
      "learning_rate": 4.08792445712911e-06,
      "loss": 2.2349,
      "step": 218282
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.4800982475280762,
      "learning_rate": 4.087251501368627e-06,
      "loss": 2.0282,
      "step": 218283
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.4836410284042358,
      "learning_rate": 4.086578600623936e-06,
      "loss": 2.0503,
      "step": 218284
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.4196600914001465,
      "learning_rate": 4.0859057548951025e-06,
      "loss": 1.9169,
      "step": 218285
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.5425710678100586,
      "learning_rate": 4.085232964182294e-06,
      "loss": 1.9772,
      "step": 218286
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.5334407091140747,
      "learning_rate": 4.084560228485612e-06,
      "loss": 1.8998,
      "step": 218287
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.4642399549484253,
      "learning_rate": 4.08388754780522e-06,
      "loss": 1.9462,
      "step": 218288
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.4676178693771362,
      "learning_rate": 4.0832149221412206e-06,
      "loss": 1.8455,
      "step": 218289
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.5609698295593262,
      "learning_rate": 4.082542351493712e-06,
      "loss": 1.8278,
      "step": 218290
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.3936601877212524,
      "learning_rate": 4.081869835862861e-06,
      "loss": 1.9618,
      "step": 218291
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.4404923915863037,
      "learning_rate": 4.081197375248801e-06,
      "loss": 1.9373,
      "step": 218292
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.478129267692566,
      "learning_rate": 4.0805249696516e-06,
      "loss": 1.9441,
      "step": 218293
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.5737680196762085,
      "learning_rate": 4.079852619071422e-06,
      "loss": 1.928,
      "step": 218294
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.6031465530395508,
      "learning_rate": 4.0791803235083686e-06,
      "loss": 1.9278,
      "step": 218295
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.5267105102539062,
      "learning_rate": 4.078508082962606e-06,
      "loss": 1.9064,
      "step": 218296
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.5243362188339233,
      "learning_rate": 4.077835897434234e-06,
      "loss": 2.2047,
      "step": 218297
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.5377625226974487,
      "learning_rate": 4.077163766923386e-06,
      "loss": 2.187,
      "step": 218298
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.3723368644714355,
      "learning_rate": 4.076491691430128e-06,
      "loss": 1.9765,
      "step": 218299
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.349843144416809,
      "learning_rate": 4.075819670954694e-06,
      "loss": 1.7925,
      "step": 218300
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.4978986978530884,
      "learning_rate": 4.075147705497117e-06,
      "loss": 1.8875,
      "step": 218301
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.5170578956604004,
      "learning_rate": 4.074475795057564e-06,
      "loss": 2.1027,
      "step": 218302
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.4778082370758057,
      "learning_rate": 4.0738039396361665e-06,
      "loss": 2.0726,
      "step": 218303
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.4537054300308228,
      "learning_rate": 4.073132139233026e-06,
      "loss": 2.0589,
      "step": 218304
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.6487473249435425,
      "learning_rate": 4.072460393848276e-06,
      "loss": 1.8621,
      "step": 218305
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.4521898031234741,
      "learning_rate": 4.071788703482048e-06,
      "loss": 2.0481,
      "step": 218306
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.6540708541870117,
      "learning_rate": 4.071117068134411e-06,
      "loss": 1.9371,
      "step": 218307
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.320638656616211,
      "learning_rate": 4.070445487805596e-06,
      "loss": 1.9993,
      "step": 218308
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.547163963317871,
      "learning_rate": 4.069773962495671e-06,
      "loss": 1.8658,
      "step": 218309
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.34174382686615,
      "learning_rate": 4.0691024922047675e-06,
      "loss": 1.9059,
      "step": 218310
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.3810726404190063,
      "learning_rate": 4.068431076932955e-06,
      "loss": 1.8676,
      "step": 218311
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.3877044916152954,
      "learning_rate": 4.06775971668043e-06,
      "loss": 1.7544,
      "step": 218312
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.606548547744751,
      "learning_rate": 4.067088411447294e-06,
      "loss": 2.1881,
      "step": 218313
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.4600105285644531,
      "learning_rate": 4.066417161233682e-06,
      "loss": 1.9595,
      "step": 218314
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.5245413780212402,
      "learning_rate": 4.065745966039724e-06,
      "loss": 2.1438,
      "step": 218315
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.8108632564544678,
      "learning_rate": 4.065074825865522e-06,
      "loss": 1.8556,
      "step": 218316
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.5664509534835815,
      "learning_rate": 4.064403740711175e-06,
      "loss": 2.0703,
      "step": 218317
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.7690162658691406,
      "learning_rate": 4.063732710576884e-06,
      "loss": 1.9446,
      "step": 218318
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.6314631700515747,
      "learning_rate": 4.063061735462681e-06,
      "loss": 2.004,
      "step": 218319
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.4233458042144775,
      "learning_rate": 4.062390815368799e-06,
      "loss": 1.7931,
      "step": 218320
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.470058798789978,
      "learning_rate": 4.0617199502952725e-06,
      "loss": 1.7507,
      "step": 218321
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.5118049383163452,
      "learning_rate": 4.061049140242267e-06,
      "loss": 2.0332,
      "step": 218322
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.42523992061615,
      "learning_rate": 4.060378385209884e-06,
      "loss": 2.059,
      "step": 218323
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.458648920059204,
      "learning_rate": 4.059707685198288e-06,
      "loss": 1.9789,
      "step": 218324
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.5126186609268188,
      "learning_rate": 4.059037040207547e-06,
      "loss": 1.9555,
      "step": 218325
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.4487829208374023,
      "learning_rate": 4.05836645023786e-06,
      "loss": 1.9812,
      "step": 218326
    },
    {
      "epoch": 2.84,
      "grad_norm": 2.871375799179077,
      "learning_rate": 4.057695915289261e-06,
      "loss": 2.1271,
      "step": 218327
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.3910280466079712,
      "learning_rate": 4.057025435361982e-06,
      "loss": 1.8387,
      "step": 218328
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.4004229307174683,
      "learning_rate": 4.056355010456025e-06,
      "loss": 1.9658,
      "step": 218329
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.4953417778015137,
      "learning_rate": 4.055684640571621e-06,
      "loss": 1.7916,
      "step": 218330
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.5250922441482544,
      "learning_rate": 4.055014325708805e-06,
      "loss": 2.1156,
      "step": 218331
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.4294655323028564,
      "learning_rate": 4.054344065867809e-06,
      "loss": 1.9298,
      "step": 218332
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.5025901794433594,
      "learning_rate": 4.053673861048634e-06,
      "loss": 1.957,
      "step": 218333
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.542547583580017,
      "learning_rate": 4.0530037112515455e-06,
      "loss": 1.9195,
      "step": 218334
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.596789002418518,
      "learning_rate": 4.052333616476544e-06,
      "loss": 1.9472,
      "step": 218335
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.4655641317367554,
      "learning_rate": 4.051663576723796e-06,
      "loss": 2.0972,
      "step": 218336
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.4471029043197632,
      "learning_rate": 4.050993591993434e-06,
      "loss": 2.0959,
      "step": 218337
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.3857303857803345,
      "learning_rate": 4.05032366228556e-06,
      "loss": 1.9079,
      "step": 218338
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.396060585975647,
      "learning_rate": 4.049653787600338e-06,
      "loss": 2.0147,
      "step": 218339
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.4273165464401245,
      "learning_rate": 4.048983967937902e-06,
      "loss": 1.9631,
      "step": 218340
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.4123481512069702,
      "learning_rate": 4.048314203298286e-06,
      "loss": 1.7891,
      "step": 218341
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.3797506093978882,
      "learning_rate": 4.047644493681723e-06,
      "loss": 1.9244,
      "step": 218342
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.4747834205627441,
      "learning_rate": 4.04697483908828e-06,
      "loss": 1.9073,
      "step": 218343
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.625802755355835,
      "learning_rate": 4.046305239518088e-06,
      "loss": 2.1984,
      "step": 218344
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.5440630912780762,
      "learning_rate": 4.0456356949712494e-06,
      "loss": 1.9787,
      "step": 218345
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.4701789617538452,
      "learning_rate": 4.04496620544793e-06,
      "loss": 2.0443,
      "step": 218346
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.4868412017822266,
      "learning_rate": 4.044296770948263e-06,
      "loss": 2.2239,
      "step": 218347
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.3991467952728271,
      "learning_rate": 4.043627391472315e-06,
      "loss": 1.9294,
      "step": 218348
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.5292307138442993,
      "learning_rate": 4.042958067020252e-06,
      "loss": 1.9715,
      "step": 218349
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.462647557258606,
      "learning_rate": 4.042288797592208e-06,
      "loss": 1.9049,
      "step": 218350
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.476786494255066,
      "learning_rate": 4.041619583188249e-06,
      "loss": 2.054,
      "step": 218351
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.537305474281311,
      "learning_rate": 4.040950423808575e-06,
      "loss": 2.1397,
      "step": 218352
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.4713895320892334,
      "learning_rate": 4.040281319453254e-06,
      "loss": 1.9772,
      "step": 218353
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.4559763669967651,
      "learning_rate": 4.03961227012245e-06,
      "loss": 2.185,
      "step": 218354
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.3972517251968384,
      "learning_rate": 4.0389432758162644e-06,
      "loss": 1.9989,
      "step": 218355
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.6183465719223022,
      "learning_rate": 4.03827433653483e-06,
      "loss": 1.9286,
      "step": 218356
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.64126455783844,
      "learning_rate": 4.037605452278248e-06,
      "loss": 1.8816,
      "step": 218357
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.5994669198989868,
      "learning_rate": 4.036936623046649e-06,
      "loss": 2.0878,
      "step": 218358
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.3911125659942627,
      "learning_rate": 4.0362678488402025e-06,
      "loss": 1.8916,
      "step": 218359
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.4576128721237183,
      "learning_rate": 4.0355991296590065e-06,
      "loss": 1.9519,
      "step": 218360
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.4389280080795288,
      "learning_rate": 4.034930465503161e-06,
      "loss": 2.0167,
      "step": 218361
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.485306978225708,
      "learning_rate": 4.0342618563728006e-06,
      "loss": 1.9749,
      "step": 218362
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.6432631015777588,
      "learning_rate": 4.0335933022680566e-06,
      "loss": 2.01,
      "step": 218363
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.4667088985443115,
      "learning_rate": 4.032924803189097e-06,
      "loss": 1.9955,
      "step": 218364
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.4093191623687744,
      "learning_rate": 4.032256359135988e-06,
      "loss": 1.8227,
      "step": 218365
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.4649031162261963,
      "learning_rate": 4.031587970108829e-06,
      "loss": 2.0564,
      "step": 218366
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.5153892040252686,
      "learning_rate": 4.030919636107821e-06,
      "loss": 1.9199,
      "step": 218367
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.54794442653656,
      "learning_rate": 4.03025135713303e-06,
      "loss": 2.0854,
      "step": 218368
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.2947824001312256,
      "learning_rate": 4.029583133184622e-06,
      "loss": 1.8135,
      "step": 218369
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.3948640823364258,
      "learning_rate": 4.028914964262697e-06,
      "loss": 1.9899,
      "step": 218370
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.4347490072250366,
      "learning_rate": 4.02824685036739e-06,
      "loss": 1.872,
      "step": 218371
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.355404019355774,
      "learning_rate": 4.027578791498797e-06,
      "loss": 1.7573,
      "step": 218372
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.633352279663086,
      "learning_rate": 4.026910787657056e-06,
      "loss": 2.2098,
      "step": 218373
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.5077837705612183,
      "learning_rate": 4.0262428388423305e-06,
      "loss": 2.1309,
      "step": 218374
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.458126425743103,
      "learning_rate": 4.025574945054688e-06,
      "loss": 2.0268,
      "step": 218375
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.4999003410339355,
      "learning_rate": 4.024907106294295e-06,
      "loss": 1.9371,
      "step": 218376
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.5230941772460938,
      "learning_rate": 4.024239322561284e-06,
      "loss": 2.0416,
      "step": 218377
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.5129259824752808,
      "learning_rate": 4.02357159385569e-06,
      "loss": 1.8802,
      "step": 218378
    },
    {
      "epoch": 2.84,
      "grad_norm": 2.9648256301879883,
      "learning_rate": 4.022903920177745e-06,
      "loss": 2.0865,
      "step": 218379
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.6402177810668945,
      "learning_rate": 4.022236301527515e-06,
      "loss": 1.9808,
      "step": 218380
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.5859218835830688,
      "learning_rate": 4.0215687379051675e-06,
      "loss": 1.9719,
      "step": 218381
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.5134828090667725,
      "learning_rate": 4.020901229310769e-06,
      "loss": 1.8158,
      "step": 218382
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.4586704969406128,
      "learning_rate": 4.0202337757444855e-06,
      "loss": 2.025,
      "step": 218383
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.3859368562698364,
      "learning_rate": 4.019566377206385e-06,
      "loss": 1.9765,
      "step": 218384
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.6222118139266968,
      "learning_rate": 4.018899033696699e-06,
      "loss": 2.0988,
      "step": 218385
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.5459058284759521,
      "learning_rate": 4.018231745215428e-06,
      "loss": 1.9323,
      "step": 218386
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.2969900369644165,
      "learning_rate": 4.017564511762805e-06,
      "loss": 1.9593,
      "step": 218387
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.5925294160842896,
      "learning_rate": 4.016897333338864e-06,
      "loss": 1.8385,
      "step": 218388
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.457415223121643,
      "learning_rate": 4.016230209943805e-06,
      "loss": 2.0665,
      "step": 218389
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.5821083784103394,
      "learning_rate": 4.0155631415776936e-06,
      "loss": 2.085,
      "step": 218390
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.5946712493896484,
      "learning_rate": 4.014896128240697e-06,
      "loss": 1.8875,
      "step": 218391
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.4397542476654053,
      "learning_rate": 4.014229169932881e-06,
      "loss": 1.801,
      "step": 218392
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.4235215187072754,
      "learning_rate": 4.0135622666544135e-06,
      "loss": 1.6992,
      "step": 218393
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.3178690671920776,
      "learning_rate": 4.012895418405427e-06,
      "loss": 1.9743,
      "step": 218394
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.5105280876159668,
      "learning_rate": 4.012228625186054e-06,
      "loss": 1.8326,
      "step": 218395
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.5473721027374268,
      "learning_rate": 4.011561886996361e-06,
      "loss": 1.9013,
      "step": 218396
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.4663841724395752,
      "learning_rate": 4.0108952038365175e-06,
      "loss": 1.8481,
      "step": 218397
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.5910859107971191,
      "learning_rate": 4.01022857570662e-06,
      "loss": 2.0487,
      "step": 218398
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.433114767074585,
      "learning_rate": 4.009562002606836e-06,
      "loss": 2.0525,
      "step": 218399
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.4282604455947876,
      "learning_rate": 4.008895484537233e-06,
      "loss": 1.8879,
      "step": 218400
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.4830771684646606,
      "learning_rate": 4.008229021498011e-06,
      "loss": 1.9681,
      "step": 218401
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.6745474338531494,
      "learning_rate": 4.0075626134891685e-06,
      "loss": 1.8482,
      "step": 218402
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.644364833831787,
      "learning_rate": 4.006896260510972e-06,
      "loss": 1.9258,
      "step": 218403
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.4383565187454224,
      "learning_rate": 4.006229962563456e-06,
      "loss": 1.9198,
      "step": 218404
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.6319767236709595,
      "learning_rate": 4.005563719646787e-06,
      "loss": 2.0914,
      "step": 218405
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.4679679870605469,
      "learning_rate": 4.004897531761031e-06,
      "loss": 2.0766,
      "step": 218406
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.510823130607605,
      "learning_rate": 4.004231398906421e-06,
      "loss": 2.0034,
      "step": 218407
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.4065498113632202,
      "learning_rate": 4.0035653210829245e-06,
      "loss": 1.7918,
      "step": 218408
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.4871941804885864,
      "learning_rate": 4.002899298290807e-06,
      "loss": 2.0852,
      "step": 218409
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.4213513135910034,
      "learning_rate": 4.002233330530136e-06,
      "loss": 1.8646,
      "step": 218410
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.492596983909607,
      "learning_rate": 4.001567417801044e-06,
      "loss": 1.9574,
      "step": 218411
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.4229198694229126,
      "learning_rate": 4.000901560103598e-06,
      "loss": 1.9624,
      "step": 218412
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.467427372932434,
      "learning_rate": 4.000235757438031e-06,
      "loss": 1.8188,
      "step": 218413
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.439178228378296,
      "learning_rate": 3.999570009804376e-06,
      "loss": 1.6999,
      "step": 218414
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.605508804321289,
      "learning_rate": 3.9989043172028e-06,
      "loss": 1.8338,
      "step": 218415
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.533412218093872,
      "learning_rate": 3.998238679633403e-06,
      "loss": 1.8295,
      "step": 218416
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.4006555080413818,
      "learning_rate": 3.9975730970963515e-06,
      "loss": 2.0261,
      "step": 218417
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.6287611722946167,
      "learning_rate": 3.996907569591678e-06,
      "loss": 2.076,
      "step": 218418
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.3708778619766235,
      "learning_rate": 3.996242097119651e-06,
      "loss": 1.7727,
      "step": 218419
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.4816802740097046,
      "learning_rate": 3.9955766796802345e-06,
      "loss": 1.9042,
      "step": 218420
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.6691378355026245,
      "learning_rate": 3.9949113172736636e-06,
      "loss": 2.0665,
      "step": 218421
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.399271845817566,
      "learning_rate": 3.994246009900037e-06,
      "loss": 1.8318,
      "step": 218422
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.5086568593978882,
      "learning_rate": 3.993580757559456e-06,
      "loss": 1.8981,
      "step": 218423
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.4124667644500732,
      "learning_rate": 3.9929155602520194e-06,
      "loss": 1.9602,
      "step": 218424
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.4388517141342163,
      "learning_rate": 3.9922504179779604e-06,
      "loss": 2.0664,
      "step": 218425
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.5858514308929443,
      "learning_rate": 3.9915853307372795e-06,
      "loss": 1.9512,
      "step": 218426
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.4856865406036377,
      "learning_rate": 3.990920298530143e-06,
      "loss": 1.9343,
      "step": 218427
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.5677964687347412,
      "learning_rate": 3.990255321356684e-06,
      "loss": 1.8935,
      "step": 218428
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.4422662258148193,
      "learning_rate": 3.989590399217069e-06,
      "loss": 1.9051,
      "step": 218429
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.3652108907699585,
      "learning_rate": 3.988925532111331e-06,
      "loss": 1.8128,
      "step": 218430
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.44955575466156,
      "learning_rate": 3.988260720039638e-06,
      "loss": 2.0178,
      "step": 218431
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.5218204259872437,
      "learning_rate": 3.9875959630021544e-06,
      "loss": 2.092,
      "step": 218432
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.4558613300323486,
      "learning_rate": 3.986931260998949e-06,
      "loss": 1.8684,
      "step": 218433
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.7041157484054565,
      "learning_rate": 3.98626661403012e-06,
      "loss": 1.9687,
      "step": 218434
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.8304214477539062,
      "learning_rate": 3.9856020220959014e-06,
      "loss": 2.0161,
      "step": 218435
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.421454906463623,
      "learning_rate": 3.984937485196293e-06,
      "loss": 2.0146,
      "step": 218436
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.359510064125061,
      "learning_rate": 3.984273003331495e-06,
      "loss": 2.0092,
      "step": 218437
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.4291434288024902,
      "learning_rate": 3.983608576501607e-06,
      "loss": 2.0542,
      "step": 218438
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.5129445791244507,
      "learning_rate": 3.982944204706762e-06,
      "loss": 2.0612,
      "step": 218439
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.6475685834884644,
      "learning_rate": 3.982279887947026e-06,
      "loss": 2.0067,
      "step": 218440
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.6954845190048218,
      "learning_rate": 3.981615626222634e-06,
      "loss": 2.0945,
      "step": 218441
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.4699550867080688,
      "learning_rate": 3.980951419533618e-06,
      "loss": 1.8811,
      "step": 218442
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.4437904357910156,
      "learning_rate": 3.980287267880145e-06,
      "loss": 1.9667,
      "step": 218443
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.401694655418396,
      "learning_rate": 3.979623171262314e-06,
      "loss": 1.9425,
      "step": 218444
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.4266626834869385,
      "learning_rate": 3.97895912968026e-06,
      "loss": 1.967,
      "step": 218445
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.5606131553649902,
      "learning_rate": 3.978295143134081e-06,
      "loss": 2.049,
      "step": 218446
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.4673454761505127,
      "learning_rate": 3.9776312116239775e-06,
      "loss": 1.9881,
      "step": 218447
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.3740346431732178,
      "learning_rate": 3.976967335149983e-06,
      "loss": 1.9437,
      "step": 218448
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.4647040367126465,
      "learning_rate": 3.976303513712265e-06,
      "loss": 2.1804,
      "step": 218449
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.4925510883331299,
      "learning_rate": 3.975639747310955e-06,
      "loss": 1.9835,
      "step": 218450
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.7133694887161255,
      "learning_rate": 3.97497603594612e-06,
      "loss": 1.903,
      "step": 218451
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.471882700920105,
      "learning_rate": 3.974312379617961e-06,
      "loss": 1.9191,
      "step": 218452
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.5308793783187866,
      "learning_rate": 3.973648778326544e-06,
      "loss": 1.9021,
      "step": 218453
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.4566733837127686,
      "learning_rate": 3.972985232072035e-06,
      "loss": 2.0103,
      "step": 218454
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.4598239660263062,
      "learning_rate": 3.972321740854534e-06,
      "loss": 1.9098,
      "step": 218455
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.738123893737793,
      "learning_rate": 3.971658304674175e-06,
      "loss": 1.8435,
      "step": 218456
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.5059199333190918,
      "learning_rate": 3.970994923531057e-06,
      "loss": 1.997,
      "step": 218457
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.6043142080307007,
      "learning_rate": 3.970331597425314e-06,
      "loss": 2.0863,
      "step": 218458
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.4567604064941406,
      "learning_rate": 3.9696683263570786e-06,
      "loss": 2.0719,
      "step": 218459
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.4467580318450928,
      "learning_rate": 3.969005110326484e-06,
      "loss": 1.91,
      "step": 218460
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.4293352365493774,
      "learning_rate": 3.968341949333631e-06,
      "loss": 1.9782,
      "step": 218461
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.637017011642456,
      "learning_rate": 3.967678843378652e-06,
      "loss": 2.0366,
      "step": 218462
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.5412248373031616,
      "learning_rate": 3.967015792461647e-06,
      "loss": 1.9736,
      "step": 218463
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.4327446222305298,
      "learning_rate": 3.966352796582783e-06,
      "loss": 1.9189,
      "step": 218464
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.5238651037216187,
      "learning_rate": 3.96568985574216e-06,
      "loss": 1.8925,
      "step": 218465
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.581973671913147,
      "learning_rate": 3.965026969939911e-06,
      "loss": 1.9411,
      "step": 218466
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.3689333200454712,
      "learning_rate": 3.964364139176135e-06,
      "loss": 1.909,
      "step": 218467
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.334126353263855,
      "learning_rate": 3.963701363450999e-06,
      "loss": 1.8894,
      "step": 218468
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.5212347507476807,
      "learning_rate": 3.963038642764571e-06,
      "loss": 1.8666,
      "step": 218469
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.4261760711669922,
      "learning_rate": 3.962375977117016e-06,
      "loss": 1.9995,
      "step": 218470
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.443345546722412,
      "learning_rate": 3.961713366508401e-06,
      "loss": 2.0466,
      "step": 218471
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.4648921489715576,
      "learning_rate": 3.961050810938959e-06,
      "loss": 1.7116,
      "step": 218472
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.4732375144958496,
      "learning_rate": 3.960388310408691e-06,
      "loss": 1.9489,
      "step": 218473
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.516795039176941,
      "learning_rate": 3.959725864917829e-06,
      "loss": 1.9972,
      "step": 218474
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.361112117767334,
      "learning_rate": 3.959063474466406e-06,
      "loss": 1.8671,
      "step": 218475
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.8527190685272217,
      "learning_rate": 3.95840113905459e-06,
      "loss": 2.043,
      "step": 218476
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.5486066341400146,
      "learning_rate": 3.95773885868248e-06,
      "loss": 1.9725,
      "step": 218477
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.407088279724121,
      "learning_rate": 3.957076633350242e-06,
      "loss": 1.8641,
      "step": 218478
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.3543498516082764,
      "learning_rate": 3.956414463057944e-06,
      "loss": 2.0218,
      "step": 218479
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.4261082410812378,
      "learning_rate": 3.955752347805785e-06,
      "loss": 1.8088,
      "step": 218480
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.3734338283538818,
      "learning_rate": 3.9550902875937985e-06,
      "loss": 1.9845,
      "step": 218481
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.3393672704696655,
      "learning_rate": 3.954428282422151e-06,
      "loss": 1.8858,
      "step": 218482
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.3865936994552612,
      "learning_rate": 3.953766332290942e-06,
      "loss": 2.0132,
      "step": 218483
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.5114375352859497,
      "learning_rate": 3.953104437200372e-06,
      "loss": 1.9537,
      "step": 218484
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.335656762123108,
      "learning_rate": 3.9524425971504406e-06,
      "loss": 2.0536,
      "step": 218485
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.4042006731033325,
      "learning_rate": 3.951780812141414e-06,
      "loss": 2.0476,
      "step": 218486
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.4713178873062134,
      "learning_rate": 3.951119082173293e-06,
      "loss": 1.8966,
      "step": 218487
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.3801411390304565,
      "learning_rate": 3.950457407246277e-06,
      "loss": 1.7376,
      "step": 218488
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.3865537643432617,
      "learning_rate": 3.949795787360433e-06,
      "loss": 1.8827,
      "step": 218489
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.538857340812683,
      "learning_rate": 3.949134222515893e-06,
      "loss": 2.0207,
      "step": 218490
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.5040677785873413,
      "learning_rate": 3.9484727127128245e-06,
      "loss": 2.1495,
      "step": 218491
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.4224802255630493,
      "learning_rate": 3.947811257951361e-06,
      "loss": 1.886,
      "step": 218492
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.448868989944458,
      "learning_rate": 3.947149858231536e-06,
      "loss": 1.8089,
      "step": 218493
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.5490583181381226,
      "learning_rate": 3.946488513553514e-06,
      "loss": 1.973,
      "step": 218494
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.5632072687149048,
      "learning_rate": 3.9458272239174304e-06,
      "loss": 2.0321,
      "step": 218495
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.4413282871246338,
      "learning_rate": 3.945165989323451e-06,
      "loss": 1.9404,
      "step": 218496
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.679703712463379,
      "learning_rate": 3.944504809771609e-06,
      "loss": 1.8425,
      "step": 218497
    },
    {
      "epoch": 2.84,
      "grad_norm": 1.409896969795227,
      "learning_rate": 3.943843685262138e-06,
      "loss": 1.8496,
      "step": 218498
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.3877451419830322,
      "learning_rate": 3.943182615795004e-06,
      "loss": 1.8454,
      "step": 218499
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.5403507947921753,
      "learning_rate": 3.942521601370474e-06,
      "loss": 1.912,
      "step": 218500
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.4380736351013184,
      "learning_rate": 3.941860641988615e-06,
      "loss": 1.9004,
      "step": 218501
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.4423565864562988,
      "learning_rate": 3.941199737649559e-06,
      "loss": 2.07,
      "step": 218502
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.588605523109436,
      "learning_rate": 3.940538888353373e-06,
      "loss": 1.9243,
      "step": 218503
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.3860869407653809,
      "learning_rate": 3.939878094100291e-06,
      "loss": 1.8941,
      "step": 218504
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.4551126956939697,
      "learning_rate": 3.939217354890345e-06,
      "loss": 1.9656,
      "step": 218505
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.4707226753234863,
      "learning_rate": 3.938556670723703e-06,
      "loss": 1.8576,
      "step": 218506
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.4162040948867798,
      "learning_rate": 3.93789604160043e-06,
      "loss": 2.0114,
      "step": 218507
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.5594719648361206,
      "learning_rate": 3.937235467520727e-06,
      "loss": 2.215,
      "step": 218508
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.4998828172683716,
      "learning_rate": 3.936574948484661e-06,
      "loss": 2.0656,
      "step": 218509
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.5080015659332275,
      "learning_rate": 3.9359144844924305e-06,
      "loss": 2.2709,
      "step": 218510
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.4271600246429443,
      "learning_rate": 3.935254075544036e-06,
      "loss": 1.9854,
      "step": 218511
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.4885964393615723,
      "learning_rate": 3.9345937216397114e-06,
      "loss": 1.9603,
      "step": 218512
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.5122863054275513,
      "learning_rate": 3.93393342277949e-06,
      "loss": 1.9705,
      "step": 218513
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.4900187253952026,
      "learning_rate": 3.9332731789635694e-06,
      "loss": 2.0257,
      "step": 218514
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.5910913944244385,
      "learning_rate": 3.932612990192019e-06,
      "loss": 1.9186,
      "step": 218515
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.403781533241272,
      "learning_rate": 3.931952856465004e-06,
      "loss": 1.9716,
      "step": 218516
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.5259445905685425,
      "learning_rate": 3.931292777782624e-06,
      "loss": 1.8949,
      "step": 218517
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.5186843872070312,
      "learning_rate": 3.930632754145013e-06,
      "loss": 1.9978,
      "step": 218518
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.5305520296096802,
      "learning_rate": 3.929972785552271e-06,
      "loss": 2.061,
      "step": 218519
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.4226881265640259,
      "learning_rate": 3.929312872004564e-06,
      "loss": 1.971,
      "step": 218520
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.4745949506759644,
      "learning_rate": 3.928653013501959e-06,
      "loss": 1.951,
      "step": 218521
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.4294633865356445,
      "learning_rate": 3.927993210044622e-06,
      "loss": 1.8845,
      "step": 218522
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.4951744079589844,
      "learning_rate": 3.927333461632654e-06,
      "loss": 2.0327,
      "step": 218523
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.4284230470657349,
      "learning_rate": 3.926673768266187e-06,
      "loss": 1.8684,
      "step": 218524
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.5883393287658691,
      "learning_rate": 3.926014129945321e-06,
      "loss": 1.9049,
      "step": 218525
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.3279954195022583,
      "learning_rate": 3.92535454667019e-06,
      "loss": 1.888,
      "step": 218526
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.6095870733261108,
      "learning_rate": 3.92469501844096e-06,
      "loss": 1.9061,
      "step": 218527
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.5754485130310059,
      "learning_rate": 3.924035545257698e-06,
      "loss": 2.1603,
      "step": 218528
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.3953698873519897,
      "learning_rate": 3.92337612712057e-06,
      "loss": 1.8707,
      "step": 218529
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.6065467596054077,
      "learning_rate": 3.922716764029643e-06,
      "loss": 1.9257,
      "step": 218530
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.553227424621582,
      "learning_rate": 3.922057455985084e-06,
      "loss": 1.9121,
      "step": 218531
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.5567104816436768,
      "learning_rate": 3.921398202986992e-06,
      "loss": 1.9091,
      "step": 218532
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.4430867433547974,
      "learning_rate": 3.9207390050355e-06,
      "loss": 1.9672,
      "step": 218533
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.3337454795837402,
      "learning_rate": 3.920079862130776e-06,
      "loss": 2.0279,
      "step": 218534
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.3483127355575562,
      "learning_rate": 3.9194207742728525e-06,
      "loss": 1.9185,
      "step": 218535
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.4504390954971313,
      "learning_rate": 3.918761741461929e-06,
      "loss": 1.9235,
      "step": 218536
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.4486430883407593,
      "learning_rate": 3.918102763698072e-06,
      "loss": 2.0241,
      "step": 218537
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.614743947982788,
      "learning_rate": 3.917443840981449e-06,
      "loss": 1.9535,
      "step": 218538
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.5168342590332031,
      "learning_rate": 3.916784973312126e-06,
      "loss": 1.9305,
      "step": 218539
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.4662883281707764,
      "learning_rate": 3.916126160690303e-06,
      "loss": 2.054,
      "step": 218540
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.4309793710708618,
      "learning_rate": 3.915467403116046e-06,
      "loss": 2.1094,
      "step": 218541
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.443053960800171,
      "learning_rate": 3.914808700589489e-06,
      "loss": 1.9179,
      "step": 218542
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.4508883953094482,
      "learning_rate": 3.914150053110765e-06,
      "loss": 1.7854,
      "step": 218543
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.595535397529602,
      "learning_rate": 3.913491460679974e-06,
      "loss": 2.2444,
      "step": 218544
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.4807449579238892,
      "learning_rate": 3.9128329232972486e-06,
      "loss": 1.9567,
      "step": 218545
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.4859567880630493,
      "learning_rate": 3.912174440962756e-06,
      "loss": 2.0428,
      "step": 218546
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.4674973487854004,
      "learning_rate": 3.9115160136765634e-06,
      "loss": 1.8152,
      "step": 218547
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.5434101819992065,
      "learning_rate": 3.910857641438769e-06,
      "loss": 1.8127,
      "step": 218548
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.5198915004730225,
      "learning_rate": 3.910199324249574e-06,
      "loss": 2.0691,
      "step": 218549
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.4456104040145874,
      "learning_rate": 3.909541062109045e-06,
      "loss": 2.2001,
      "step": 218550
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.690420150756836,
      "learning_rate": 3.908882855017348e-06,
      "loss": 1.9185,
      "step": 218551
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.4497572183609009,
      "learning_rate": 3.90822470297455e-06,
      "loss": 2.081,
      "step": 218552
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.529585599899292,
      "learning_rate": 3.907566605980849e-06,
      "loss": 1.9051,
      "step": 218553
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.5134046077728271,
      "learning_rate": 3.906908564036249e-06,
      "loss": 2.0126,
      "step": 218554
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.5012617111206055,
      "learning_rate": 3.906250577141012e-06,
      "loss": 2.1446,
      "step": 218555
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.4329081773757935,
      "learning_rate": 3.905592645295142e-06,
      "loss": 1.8952,
      "step": 218556
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.3539347648620605,
      "learning_rate": 3.904934768498835e-06,
      "loss": 1.9302,
      "step": 218557
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.442521333694458,
      "learning_rate": 3.90427694675216e-06,
      "loss": 2.076,
      "step": 218558
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.4845314025878906,
      "learning_rate": 3.903619180055351e-06,
      "loss": 1.9457,
      "step": 218559
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.484654426574707,
      "learning_rate": 3.9029614684083725e-06,
      "loss": 1.8882,
      "step": 218560
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.5543696880340576,
      "learning_rate": 3.902303811811458e-06,
      "loss": 2.0866,
      "step": 218561
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.5175913572311401,
      "learning_rate": 3.901646210264642e-06,
      "loss": 2.1636,
      "step": 218562
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.4335967302322388,
      "learning_rate": 3.900988663768156e-06,
      "loss": 1.9198,
      "step": 218563
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.539111852645874,
      "learning_rate": 3.900331172322036e-06,
      "loss": 1.9982,
      "step": 218564
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.3772306442260742,
      "learning_rate": 3.899673735926478e-06,
      "loss": 2.0237,
      "step": 218565
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.4952812194824219,
      "learning_rate": 3.899016354581486e-06,
      "loss": 1.8614,
      "step": 218566
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.5559910535812378,
      "learning_rate": 3.898359028287324e-06,
      "loss": 2.0628,
      "step": 218567
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.3918150663375854,
      "learning_rate": 3.897701757043992e-06,
      "loss": 1.9822,
      "step": 218568
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.3296923637390137,
      "learning_rate": 3.89704454085169e-06,
      "loss": 1.8143,
      "step": 218569
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.4662822484970093,
      "learning_rate": 3.89638737971052e-06,
      "loss": 2.0233,
      "step": 218570
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.4167596101760864,
      "learning_rate": 3.8957302736206454e-06,
      "loss": 1.9099,
      "step": 218571
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.521812081336975,
      "learning_rate": 3.895073222582068e-06,
      "loss": 1.8455,
      "step": 218572
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.5035847425460815,
      "learning_rate": 3.894416226595021e-06,
      "loss": 2.0954,
      "step": 218573
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.4159183502197266,
      "learning_rate": 3.8937592856595705e-06,
      "loss": 1.7797,
      "step": 218574
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.5655860900878906,
      "learning_rate": 3.893102399775882e-06,
      "loss": 2.2594,
      "step": 218575
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.3834515810012817,
      "learning_rate": 3.892445568944058e-06,
      "loss": 2.1006,
      "step": 218576
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.4609075784683228,
      "learning_rate": 3.89178879316423e-06,
      "loss": 2.0,
      "step": 218577
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.3742597103118896,
      "learning_rate": 3.891132072436464e-06,
      "loss": 1.8451,
      "step": 218578
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.4650976657867432,
      "learning_rate": 3.890475406760962e-06,
      "loss": 2.0738,
      "step": 218579
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.5392684936523438,
      "learning_rate": 3.889818796137789e-06,
      "loss": 1.9695,
      "step": 218580
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.484471082687378,
      "learning_rate": 3.889162240567112e-06,
      "loss": 1.7782,
      "step": 218581
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.5048668384552002,
      "learning_rate": 3.888505740048997e-06,
      "loss": 1.939,
      "step": 218582
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.4161361455917358,
      "learning_rate": 3.887849294583644e-06,
      "loss": 2.0462,
      "step": 218583
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.6026819944381714,
      "learning_rate": 3.887192904171088e-06,
      "loss": 1.8657,
      "step": 218584
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.3990541696548462,
      "learning_rate": 3.886536568811494e-06,
      "loss": 2.1389,
      "step": 218585
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.378250002861023,
      "learning_rate": 3.885880288504994e-06,
      "loss": 1.8085,
      "step": 218586
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.4118636846542358,
      "learning_rate": 3.885224063251691e-06,
      "loss": 1.8576,
      "step": 218587
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.4640140533447266,
      "learning_rate": 3.884567893051715e-06,
      "loss": 1.9989,
      "step": 218588
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.375380277633667,
      "learning_rate": 3.883911777905235e-06,
      "loss": 1.9861,
      "step": 218589
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.5566998720169067,
      "learning_rate": 3.88325571781225e-06,
      "loss": 2.0251,
      "step": 218590
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.5545905828475952,
      "learning_rate": 3.8825997127730265e-06,
      "loss": 2.0033,
      "step": 218591
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.5606896877288818,
      "learning_rate": 3.881943762787565e-06,
      "loss": 1.9832,
      "step": 218592
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.4921650886535645,
      "learning_rate": 3.881287867856065e-06,
      "loss": 1.7334,
      "step": 218593
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.5264275074005127,
      "learning_rate": 3.8806320279785915e-06,
      "loss": 1.8281,
      "step": 218594
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.4260486364364624,
      "learning_rate": 3.879976243155347e-06,
      "loss": 1.8936,
      "step": 218595
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.4516618251800537,
      "learning_rate": 3.8793205133863635e-06,
      "loss": 2.1062,
      "step": 218596
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.393608570098877,
      "learning_rate": 3.8786648386718415e-06,
      "loss": 1.8506,
      "step": 218597
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.485851526260376,
      "learning_rate": 3.878009219011813e-06,
      "loss": 1.9652,
      "step": 218598
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.5334086418151855,
      "learning_rate": 3.8773536544065124e-06,
      "loss": 1.8951,
      "step": 218599
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.4378844499588013,
      "learning_rate": 3.876698144855938e-06,
      "loss": 1.8408,
      "step": 218600
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.494490146636963,
      "learning_rate": 3.876042690360326e-06,
      "loss": 1.8972,
      "step": 218601
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.5432652235031128,
      "learning_rate": 3.87538729091974e-06,
      "loss": 2.1526,
      "step": 218602
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.5598162412643433,
      "learning_rate": 3.874731946534315e-06,
      "loss": 1.9149,
      "step": 218603
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.3773326873779297,
      "learning_rate": 3.87407665720415e-06,
      "loss": 2.1006,
      "step": 218604
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.5513437986373901,
      "learning_rate": 3.873421422929379e-06,
      "loss": 2.0079,
      "step": 218605
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.7082825899124146,
      "learning_rate": 3.872766243710135e-06,
      "loss": 2.114,
      "step": 218606
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.5274631977081299,
      "learning_rate": 3.8721111195465175e-06,
      "loss": 1.9671,
      "step": 218607
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.4691401720046997,
      "learning_rate": 3.8714560504386925e-06,
      "loss": 2.0247,
      "step": 218608
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.6464756727218628,
      "learning_rate": 3.870801036386761e-06,
      "loss": 1.9071,
      "step": 218609
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.3994252681732178,
      "learning_rate": 3.870146077390823e-06,
      "loss": 1.9092,
      "step": 218610
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.350401759147644,
      "learning_rate": 3.869491173451011e-06,
      "loss": 1.952,
      "step": 218611
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.4835184812545776,
      "learning_rate": 3.868836324567426e-06,
      "loss": 1.8433,
      "step": 218612
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.4345591068267822,
      "learning_rate": 3.868181530740266e-06,
      "loss": 1.9103,
      "step": 218613
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.458184838294983,
      "learning_rate": 3.8675267919695665e-06,
      "loss": 2.0299,
      "step": 218614
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.5165280103683472,
      "learning_rate": 3.8668721082554925e-06,
      "loss": 1.9725,
      "step": 218615
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.4268864393234253,
      "learning_rate": 3.866217479598144e-06,
      "loss": 1.8308,
      "step": 218616
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.4197415113449097,
      "learning_rate": 3.865562905997687e-06,
      "loss": 2.076,
      "step": 218617
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.7290115356445312,
      "learning_rate": 3.864908387454157e-06,
      "loss": 2.1123,
      "step": 218618
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.3074100017547607,
      "learning_rate": 3.864253923967786e-06,
      "loss": 1.9877,
      "step": 218619
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.4836164712905884,
      "learning_rate": 3.86359951553864e-06,
      "loss": 2.0614,
      "step": 218620
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.3703683614730835,
      "learning_rate": 3.862945162166786e-06,
      "loss": 1.9912,
      "step": 218621
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.4077410697937012,
      "learning_rate": 3.862290863852457e-06,
      "loss": 2.2034,
      "step": 218622
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.4701741933822632,
      "learning_rate": 3.861636620595688e-06,
      "loss": 2.0171,
      "step": 218623
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.4821887016296387,
      "learning_rate": 3.860982432396642e-06,
      "loss": 1.973,
      "step": 218624
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.6467483043670654,
      "learning_rate": 3.8603282992554215e-06,
      "loss": 2.1402,
      "step": 218625
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.5682151317596436,
      "learning_rate": 3.859674221172193e-06,
      "loss": 1.7956,
      "step": 218626
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.4843751192092896,
      "learning_rate": 3.8590201981469895e-06,
      "loss": 1.9746,
      "step": 218627
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.3788460493087769,
      "learning_rate": 3.85836623018001e-06,
      "loss": 1.882,
      "step": 218628
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.4963271617889404,
      "learning_rate": 3.857712317271322e-06,
      "loss": 1.9898,
      "step": 218629
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.3760695457458496,
      "learning_rate": 3.857058459421125e-06,
      "loss": 1.8633,
      "step": 218630
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.425066351890564,
      "learning_rate": 3.856404656629453e-06,
      "loss": 1.7869,
      "step": 218631
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.3173068761825562,
      "learning_rate": 3.8557509088964715e-06,
      "loss": 1.7738,
      "step": 218632
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.3698984384536743,
      "learning_rate": 3.855097216222314e-06,
      "loss": 1.7715,
      "step": 218633
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.42626953125,
      "learning_rate": 3.854443578607081e-06,
      "loss": 2.0623,
      "step": 218634
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.3880901336669922,
      "learning_rate": 3.853789996050871e-06,
      "loss": 2.007,
      "step": 218635
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.4663491249084473,
      "learning_rate": 3.853136468553852e-06,
      "loss": 2.1704,
      "step": 218636
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.489371418952942,
      "learning_rate": 3.852482996116091e-06,
      "loss": 2.0463,
      "step": 218637
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.5040256977081299,
      "learning_rate": 3.8518295787378195e-06,
      "loss": 1.9752,
      "step": 218638
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.4895597696304321,
      "learning_rate": 3.851176216419005e-06,
      "loss": 2.0084,
      "step": 218639
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.591597080230713,
      "learning_rate": 3.850522909159881e-06,
      "loss": 2.0457,
      "step": 218640
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.5071041584014893,
      "learning_rate": 3.849869656960547e-06,
      "loss": 1.8937,
      "step": 218641
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.4089077711105347,
      "learning_rate": 3.849216459821103e-06,
      "loss": 2.0429,
      "step": 218642
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.4166676998138428,
      "learning_rate": 3.848563317741649e-06,
      "loss": 1.9288,
      "step": 218643
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.474835753440857,
      "learning_rate": 3.8479102307223845e-06,
      "loss": 1.8515,
      "step": 218644
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.431931734085083,
      "learning_rate": 3.847257198763376e-06,
      "loss": 2.2014,
      "step": 218645
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.3645395040512085,
      "learning_rate": 3.846604221864724e-06,
      "loss": 1.9217,
      "step": 218646
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.6549553871154785,
      "learning_rate": 3.845951300026595e-06,
      "loss": 1.9238,
      "step": 218647
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.5116522312164307,
      "learning_rate": 3.845298433249089e-06,
      "loss": 2.0154,
      "step": 218648
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.4889047145843506,
      "learning_rate": 3.844645621532339e-06,
      "loss": 1.8762,
      "step": 218649
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.335316777229309,
      "learning_rate": 3.843992864876477e-06,
      "loss": 1.8912,
      "step": 218650
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.5634928941726685,
      "learning_rate": 3.843340163281572e-06,
      "loss": 1.9231,
      "step": 218651
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.535621166229248,
      "learning_rate": 3.842687516747822e-06,
      "loss": 1.907,
      "step": 218652
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.3555828332901,
      "learning_rate": 3.842034925275261e-06,
      "loss": 1.9254,
      "step": 218653
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.5288740396499634,
      "learning_rate": 3.841382388864089e-06,
      "loss": 1.9591,
      "step": 218654
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.442085862159729,
      "learning_rate": 3.840729907514339e-06,
      "loss": 2.0811,
      "step": 218655
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.2756012678146362,
      "learning_rate": 3.840077481226278e-06,
      "loss": 1.9458,
      "step": 218656
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.5299232006072998,
      "learning_rate": 3.839425109999872e-06,
      "loss": 2.0201,
      "step": 218657
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.4420078992843628,
      "learning_rate": 3.8387727938353205e-06,
      "loss": 2.0078,
      "step": 218658
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.3876138925552368,
      "learning_rate": 3.838120532732725e-06,
      "loss": 1.8343,
      "step": 218659
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.5460984706878662,
      "learning_rate": 3.83746832669225e-06,
      "loss": 1.8314,
      "step": 218660
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.4953118562698364,
      "learning_rate": 3.83681617571393e-06,
      "loss": 2.0624,
      "step": 218661
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.5229331254959106,
      "learning_rate": 3.836164079797966e-06,
      "loss": 1.8789,
      "step": 218662
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.5269402265548706,
      "learning_rate": 3.835512038944455e-06,
      "loss": 2.2869,
      "step": 218663
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.7118889093399048,
      "learning_rate": 3.834860053153499e-06,
      "loss": 2.1848,
      "step": 218664
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.5440274477005005,
      "learning_rate": 3.834208122425231e-06,
      "loss": 1.9679,
      "step": 218665
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.5824000835418701,
      "learning_rate": 3.833556246759783e-06,
      "loss": 1.8145,
      "step": 218666
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.4867559671401978,
      "learning_rate": 3.8329044261572575e-06,
      "loss": 1.8466,
      "step": 218667
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.5279791355133057,
      "learning_rate": 3.8322526606178186e-06,
      "loss": 2.1141,
      "step": 218668
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.4745721817016602,
      "learning_rate": 3.831600950141533e-06,
      "loss": 2.1866,
      "step": 218669
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.4748568534851074,
      "learning_rate": 3.830949294728536e-06,
      "loss": 2.0682,
      "step": 218670
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.717921495437622,
      "learning_rate": 3.830297694378925e-06,
      "loss": 1.7776,
      "step": 218671
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.4844863414764404,
      "learning_rate": 3.829646149092902e-06,
      "loss": 2.0948,
      "step": 218672
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.3536858558654785,
      "learning_rate": 3.828994658870532e-06,
      "loss": 1.984,
      "step": 218673
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.6445072889328003,
      "learning_rate": 3.828343223711916e-06,
      "loss": 1.9949,
      "step": 218674
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.4030293226242065,
      "learning_rate": 3.82769184361722e-06,
      "loss": 1.8975,
      "step": 218675
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.395458698272705,
      "learning_rate": 3.827040518586577e-06,
      "loss": 2.1197,
      "step": 218676
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.3968133926391602,
      "learning_rate": 3.826389248620021e-06,
      "loss": 1.9195,
      "step": 218677
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.3225445747375488,
      "learning_rate": 3.825738033717751e-06,
      "loss": 1.9792,
      "step": 218678
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.5278297662734985,
      "learning_rate": 3.8250868738798675e-06,
      "loss": 1.8913,
      "step": 218679
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.6477166414260864,
      "learning_rate": 3.824435769106504e-06,
      "loss": 1.9123,
      "step": 218680
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.4717581272125244,
      "learning_rate": 3.82378471939776e-06,
      "loss": 2.203,
      "step": 218681
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.5129506587982178,
      "learning_rate": 3.8231337247537685e-06,
      "loss": 2.1251,
      "step": 218682
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.4935886859893799,
      "learning_rate": 3.82248278517463e-06,
      "loss": 2.0127,
      "step": 218683
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.385195255279541,
      "learning_rate": 3.82183190066051e-06,
      "loss": 2.0104,
      "step": 218684
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.36278235912323,
      "learning_rate": 3.821181071211477e-06,
      "loss": 1.7914,
      "step": 218685
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.4652128219604492,
      "learning_rate": 3.820530296827695e-06,
      "loss": 2.0019,
      "step": 218686
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.586488962173462,
      "learning_rate": 3.819879577509266e-06,
      "loss": 2.0661,
      "step": 218687
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.364546298980713,
      "learning_rate": 3.819228913256322e-06,
      "loss": 2.1698,
      "step": 218688
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.5773118734359741,
      "learning_rate": 3.818578304068964e-06,
      "loss": 2.0333,
      "step": 218689
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.5968931913375854,
      "learning_rate": 3.817927749947325e-06,
      "loss": 2.1454,
      "step": 218690
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.397316575050354,
      "learning_rate": 3.817277250891504e-06,
      "loss": 1.7592,
      "step": 218691
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.3466359376907349,
      "learning_rate": 3.8166268069016344e-06,
      "loss": 2.0299,
      "step": 218692
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.506715178489685,
      "learning_rate": 3.815976417977884e-06,
      "loss": 1.9174,
      "step": 218693
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.4015253782272339,
      "learning_rate": 3.815326084120318e-06,
      "loss": 2.0364,
      "step": 218694
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.4617016315460205,
      "learning_rate": 3.8146758053290705e-06,
      "loss": 1.8456,
      "step": 218695
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.4857145547866821,
      "learning_rate": 3.8140255816042743e-06,
      "loss": 2.0669,
      "step": 218696
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.4538167715072632,
      "learning_rate": 3.813375412946029e-06,
      "loss": 1.93,
      "step": 218697
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.483967900276184,
      "learning_rate": 3.812725299354469e-06,
      "loss": 2.0426,
      "step": 218698
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.6146900653839111,
      "learning_rate": 3.8120752408297262e-06,
      "loss": 1.9303,
      "step": 218699
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.4090147018432617,
      "learning_rate": 3.8114252373719013e-06,
      "loss": 1.9768,
      "step": 218700
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.4418835639953613,
      "learning_rate": 3.8107752889811273e-06,
      "loss": 1.9641,
      "step": 218701
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.4080435037612915,
      "learning_rate": 3.8101253956575373e-06,
      "loss": 1.7067,
      "step": 218702
    },
    {
      "epoch": 2.85,
      "grad_norm": 2.2791903018951416,
      "learning_rate": 3.8094755574011983e-06,
      "loss": 2.0263,
      "step": 218703
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.3406003713607788,
      "learning_rate": 3.808825774212276e-06,
      "loss": 1.9642,
      "step": 218704
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.5092976093292236,
      "learning_rate": 3.808176046090905e-06,
      "loss": 1.963,
      "step": 218705
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.532631754875183,
      "learning_rate": 3.807526373037184e-06,
      "loss": 2.0139,
      "step": 218706
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.4989687204360962,
      "learning_rate": 3.8068767550512137e-06,
      "loss": 2.0158,
      "step": 218707
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.4860625267028809,
      "learning_rate": 3.80622719213316e-06,
      "loss": 1.9771,
      "step": 218708
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.5024932622909546,
      "learning_rate": 3.80557768428309e-06,
      "loss": 1.9414,
      "step": 218709
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.414188027381897,
      "learning_rate": 3.804928231501203e-06,
      "loss": 1.9043,
      "step": 218710
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.6173673868179321,
      "learning_rate": 3.804278833787533e-06,
      "loss": 2.2229,
      "step": 218711
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.5054677724838257,
      "learning_rate": 3.803629491142246e-06,
      "loss": 2.1882,
      "step": 218712
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.5033804178237915,
      "learning_rate": 3.802980203565442e-06,
      "loss": 2.0198,
      "step": 218713
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.431761384010315,
      "learning_rate": 3.8023309710572546e-06,
      "loss": 1.9362,
      "step": 218714
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.3596391677856445,
      "learning_rate": 3.8016817936178503e-06,
      "loss": 1.9191,
      "step": 218715
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.490565538406372,
      "learning_rate": 3.8010326712472618e-06,
      "loss": 2.1233,
      "step": 218716
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.4263806343078613,
      "learning_rate": 3.8003836039456894e-06,
      "loss": 2.042,
      "step": 218717
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.4003933668136597,
      "learning_rate": 3.799734591713166e-06,
      "loss": 2.0179,
      "step": 218718
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.5807578563690186,
      "learning_rate": 3.799085634549892e-06,
      "loss": 2.0708,
      "step": 218719
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.4145253896713257,
      "learning_rate": 3.7984367324559673e-06,
      "loss": 2.05,
      "step": 218720
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.4994865655899048,
      "learning_rate": 3.7977878854314916e-06,
      "loss": 1.9774,
      "step": 218721
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.3476365804672241,
      "learning_rate": 3.7971390934765978e-06,
      "loss": 2.0087,
      "step": 218722
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.5042484998703003,
      "learning_rate": 3.7964903565914197e-06,
      "loss": 2.0395,
      "step": 218723
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.5728232860565186,
      "learning_rate": 3.7958416747760566e-06,
      "loss": 1.9648,
      "step": 218724
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.5016520023345947,
      "learning_rate": 3.7951930480306426e-06,
      "loss": 1.6333,
      "step": 218725
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.4242140054702759,
      "learning_rate": 3.7945444763552768e-06,
      "loss": 1.7873,
      "step": 218726
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.4745509624481201,
      "learning_rate": 3.793895959750093e-06,
      "loss": 2.1062,
      "step": 218727
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.5082987546920776,
      "learning_rate": 3.793247498215224e-06,
      "loss": 2.0362,
      "step": 218728
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.500266671180725,
      "learning_rate": 3.7925990917508033e-06,
      "loss": 2.0154,
      "step": 218729
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.6540229320526123,
      "learning_rate": 3.7919507403569303e-06,
      "loss": 2.0641,
      "step": 218730
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.4111310243606567,
      "learning_rate": 3.791302444033706e-06,
      "loss": 2.1753,
      "step": 218731
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.4438835382461548,
      "learning_rate": 3.7906542027812626e-06,
      "loss": 2.1163,
      "step": 218732
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.4050395488739014,
      "learning_rate": 3.7900060165997336e-06,
      "loss": 1.9336,
      "step": 218733
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.5635026693344116,
      "learning_rate": 3.7893578854892192e-06,
      "loss": 2.0276,
      "step": 218734
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.3682937622070312,
      "learning_rate": 3.7887098094498856e-06,
      "loss": 1.9853,
      "step": 218735
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.6498112678527832,
      "learning_rate": 3.788061788481833e-06,
      "loss": 2.0272,
      "step": 218736
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.489032506942749,
      "learning_rate": 3.787413822585161e-06,
      "loss": 1.9178,
      "step": 218737
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.4189380407333374,
      "learning_rate": 3.7867659117599703e-06,
      "loss": 1.9877,
      "step": 218738
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.3922886848449707,
      "learning_rate": 3.7861180560064263e-06,
      "loss": 2.1083,
      "step": 218739
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.4316860437393188,
      "learning_rate": 3.78547025532463e-06,
      "loss": 2.0134,
      "step": 218740
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.729964017868042,
      "learning_rate": 3.7848225097147132e-06,
      "loss": 2.0088,
      "step": 218741
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.629330039024353,
      "learning_rate": 3.7841748191768104e-06,
      "loss": 1.9567,
      "step": 218742
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.547197699546814,
      "learning_rate": 3.7835271837109883e-06,
      "loss": 2.0415,
      "step": 218743
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.4045343399047852,
      "learning_rate": 3.7828796033174125e-06,
      "loss": 1.9499,
      "step": 218744
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.4751875400543213,
      "learning_rate": 3.7822320779961834e-06,
      "loss": 1.9508,
      "step": 218745
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.5207717418670654,
      "learning_rate": 3.7815846077474343e-06,
      "loss": 1.9547,
      "step": 218746
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.438435435295105,
      "learning_rate": 3.780937192571298e-06,
      "loss": 2.0003,
      "step": 218747
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.538248062133789,
      "learning_rate": 3.780289832467842e-06,
      "loss": 1.9604,
      "step": 218748
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.4460861682891846,
      "learning_rate": 3.7796425274372655e-06,
      "loss": 1.9435,
      "step": 218749
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.4788001775741577,
      "learning_rate": 3.7789952774796018e-06,
      "loss": 1.8227,
      "step": 218750
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.5332425832748413,
      "learning_rate": 3.778348082595051e-06,
      "loss": 1.9775,
      "step": 218751
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.3676767349243164,
      "learning_rate": 3.7777009427836457e-06,
      "loss": 1.9221,
      "step": 218752
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.3138195276260376,
      "learning_rate": 3.777053858045587e-06,
      "loss": 1.8951,
      "step": 218753
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.487512469291687,
      "learning_rate": 3.7764068283809736e-06,
      "loss": 1.7974,
      "step": 218754
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.460096001625061,
      "learning_rate": 3.775759853789939e-06,
      "loss": 1.9074,
      "step": 218755
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.4110099077224731,
      "learning_rate": 3.7751129342725504e-06,
      "loss": 1.8526,
      "step": 218756
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.9425581693649292,
      "learning_rate": 3.7744660698289743e-06,
      "loss": 2.2731,
      "step": 218757
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.5520082712173462,
      "learning_rate": 3.7738192604592767e-06,
      "loss": 2.0806,
      "step": 218758
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.5113650560379028,
      "learning_rate": 3.7731725061636575e-06,
      "loss": 2.0425,
      "step": 218759
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.4022446870803833,
      "learning_rate": 3.772525806942217e-06,
      "loss": 1.921,
      "step": 218760
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.4530481100082397,
      "learning_rate": 3.7718791627950216e-06,
      "loss": 1.9726,
      "step": 218761
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.4806203842163086,
      "learning_rate": 3.7712325737222047e-06,
      "loss": 1.9964,
      "step": 218762
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.5029594898223877,
      "learning_rate": 3.770586039723966e-06,
      "loss": 2.0882,
      "step": 218763
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.57498300075531,
      "learning_rate": 3.769939560800306e-06,
      "loss": 2.0321,
      "step": 218764
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.5055795907974243,
      "learning_rate": 3.7692931369514566e-06,
      "loss": 1.9262,
      "step": 218765
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.5264848470687866,
      "learning_rate": 3.768646768177452e-06,
      "loss": 2.1027,
      "step": 218766
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.550945520401001,
      "learning_rate": 3.7680004544784926e-06,
      "loss": 1.7649,
      "step": 218767
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.5138499736785889,
      "learning_rate": 3.7673541958546106e-06,
      "loss": 1.8834,
      "step": 218768
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.4755167961120605,
      "learning_rate": 3.766707992305973e-06,
      "loss": 1.7706,
      "step": 218769
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.5729717016220093,
      "learning_rate": 3.7660618438326796e-06,
      "loss": 1.8421,
      "step": 218770
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.4344940185546875,
      "learning_rate": 3.765415750434897e-06,
      "loss": 1.7646,
      "step": 218771
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.6735966205596924,
      "learning_rate": 3.764769712112725e-06,
      "loss": 1.7473,
      "step": 218772
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.5723915100097656,
      "learning_rate": 3.764123728866264e-06,
      "loss": 1.9552,
      "step": 218773
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.450649619102478,
      "learning_rate": 3.7634778006956465e-06,
      "loss": 1.959,
      "step": 218774
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.4194997549057007,
      "learning_rate": 3.7628319276009734e-06,
      "loss": 2.0122,
      "step": 218775
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.4102429151535034,
      "learning_rate": 3.7621861095823767e-06,
      "loss": 1.9374,
      "step": 218776
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.5472254753112793,
      "learning_rate": 3.761540346640024e-06,
      "loss": 2.0705,
      "step": 218777
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.451343297958374,
      "learning_rate": 3.7608946387739815e-06,
      "loss": 2.1678,
      "step": 218778
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.4195493459701538,
      "learning_rate": 3.7602489859843486e-06,
      "loss": 2.0079,
      "step": 218779
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.6680335998535156,
      "learning_rate": 3.7596033882712928e-06,
      "loss": 2.0266,
      "step": 218780
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.5324602127075195,
      "learning_rate": 3.7589578456349132e-06,
      "loss": 2.0286,
      "step": 218781
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.4053401947021484,
      "learning_rate": 3.758312358075344e-06,
      "loss": 1.9596,
      "step": 218782
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.4355767965316772,
      "learning_rate": 3.757666925592717e-06,
      "loss": 2.032,
      "step": 218783
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.5759841203689575,
      "learning_rate": 3.7570215481871003e-06,
      "loss": 1.9236,
      "step": 218784
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.7752587795257568,
      "learning_rate": 3.7563762258586595e-06,
      "loss": 1.9582,
      "step": 218785
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.5174243450164795,
      "learning_rate": 3.7557309586074947e-06,
      "loss": 2.0766,
      "step": 218786
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.5709969997406006,
      "learning_rate": 3.7550857464337393e-06,
      "loss": 2.032,
      "step": 218787
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.3688629865646362,
      "learning_rate": 3.75444058933746e-06,
      "loss": 1.9835,
      "step": 218788
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.3366400003433228,
      "learning_rate": 3.753795487318889e-06,
      "loss": 1.8993,
      "step": 218789
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.6719591617584229,
      "learning_rate": 3.753150440378061e-06,
      "loss": 2.3168,
      "step": 218790
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.617932677268982,
      "learning_rate": 3.752505448515075e-06,
      "loss": 2.1184,
      "step": 218791
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.5174012184143066,
      "learning_rate": 3.7518605117301314e-06,
      "loss": 1.9242,
      "step": 218792
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.534820318222046,
      "learning_rate": 3.7512156300232967e-06,
      "loss": 2.1627,
      "step": 218793
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.4670276641845703,
      "learning_rate": 3.7505708033947036e-06,
      "loss": 1.9027,
      "step": 218794
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.51729154586792,
      "learning_rate": 3.749926031844486e-06,
      "loss": 2.0078,
      "step": 218795
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.4887462854385376,
      "learning_rate": 3.7492813153727097e-06,
      "loss": 2.0387,
      "step": 218796
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.5029082298278809,
      "learning_rate": 3.7486366539795753e-06,
      "loss": 2.0924,
      "step": 218797
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.482621669769287,
      "learning_rate": 3.747992047665149e-06,
      "loss": 2.0284,
      "step": 218798
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.492247223854065,
      "learning_rate": 3.7473474964295647e-06,
      "loss": 1.9462,
      "step": 218799
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.4144318103790283,
      "learning_rate": 3.7467030002729216e-06,
      "loss": 2.0361,
      "step": 218800
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.346071720123291,
      "learning_rate": 3.746058559195386e-06,
      "loss": 1.8763,
      "step": 218801
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.8199490308761597,
      "learning_rate": 3.7454141731970256e-06,
      "loss": 1.9355,
      "step": 218802
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.4529132843017578,
      "learning_rate": 3.7447698422780057e-06,
      "loss": 1.8602,
      "step": 218803
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.485321044921875,
      "learning_rate": 3.7441255664384273e-06,
      "loss": 1.918,
      "step": 218804
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.501050591468811,
      "learning_rate": 3.7434813456783898e-06,
      "loss": 1.8879,
      "step": 218805
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.5021406412124634,
      "learning_rate": 3.742837179998026e-06,
      "loss": 2.0696,
      "step": 218806
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.3603603839874268,
      "learning_rate": 3.7421930693975033e-06,
      "loss": 1.8644,
      "step": 218807
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.584181785583496,
      "learning_rate": 3.7415490138768543e-06,
      "loss": 2.0947,
      "step": 218808
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.5781618356704712,
      "learning_rate": 3.740905013436246e-06,
      "loss": 1.9806,
      "step": 218809
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.5358655452728271,
      "learning_rate": 3.740261068075812e-06,
      "loss": 1.9838,
      "step": 218810
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.43954336643219,
      "learning_rate": 3.7396171777956508e-06,
      "loss": 1.7062,
      "step": 218811
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.4506175518035889,
      "learning_rate": 3.7389733425958968e-06,
      "loss": 1.9921,
      "step": 218812
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.5374513864517212,
      "learning_rate": 3.7383295624766497e-06,
      "loss": 1.8675,
      "step": 218813
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.6173324584960938,
      "learning_rate": 3.7376858374380427e-06,
      "loss": 2.0453,
      "step": 218814
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.4590487480163574,
      "learning_rate": 3.7370421674802085e-06,
      "loss": 1.8871,
      "step": 218815
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.5152732133865356,
      "learning_rate": 3.7363985526032147e-06,
      "loss": 2.0752,
      "step": 218816
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.4354965686798096,
      "learning_rate": 3.735754992807227e-06,
      "loss": 1.9963,
      "step": 218817
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.498416543006897,
      "learning_rate": 3.7351114880923793e-06,
      "loss": 1.8248,
      "step": 218818
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.4510570764541626,
      "learning_rate": 3.7344680384587377e-06,
      "loss": 2.1156,
      "step": 218819
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.3851555585861206,
      "learning_rate": 3.733824643906469e-06,
      "loss": 1.9505,
      "step": 218820
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.3086735010147095,
      "learning_rate": 3.7331813044356727e-06,
      "loss": 1.8764,
      "step": 218821
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.375044345855713,
      "learning_rate": 3.7325380200464827e-06,
      "loss": 2.0703,
      "step": 218822
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.5919158458709717,
      "learning_rate": 3.7318947907389983e-06,
      "loss": 2.0969,
      "step": 218823
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.4615776538848877,
      "learning_rate": 3.7312516165133197e-06,
      "loss": 2.0156,
      "step": 218824
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.4899539947509766,
      "learning_rate": 3.730608497369614e-06,
      "loss": 2.1428,
      "step": 218825
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.5214630365371704,
      "learning_rate": 3.7299654333079796e-06,
      "loss": 2.0729,
      "step": 218826
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.4377787113189697,
      "learning_rate": 3.7293224243285514e-06,
      "loss": 2.0661,
      "step": 218827
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.5017720460891724,
      "learning_rate": 3.7286794704314284e-06,
      "loss": 1.8456,
      "step": 218828
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.6126068830490112,
      "learning_rate": 3.7280365716167106e-06,
      "loss": 2.1107,
      "step": 218829
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.4206645488739014,
      "learning_rate": 3.727393727884598e-06,
      "loss": 1.8826,
      "step": 218830
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.4478427171707153,
      "learning_rate": 3.7267509392350902e-06,
      "loss": 2.1442,
      "step": 218831
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.4232102632522583,
      "learning_rate": 3.726108205668421e-06,
      "loss": 1.9312,
      "step": 218832
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.4808367490768433,
      "learning_rate": 3.7254655271846567e-06,
      "loss": 1.9036,
      "step": 218833
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.3758125305175781,
      "learning_rate": 3.7248229037838973e-06,
      "loss": 1.8628,
      "step": 218834
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.5681140422821045,
      "learning_rate": 3.7241803354662757e-06,
      "loss": 2.131,
      "step": 218835
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.4034799337387085,
      "learning_rate": 3.723537822231959e-06,
      "loss": 1.9495,
      "step": 218836
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.537068247795105,
      "learning_rate": 3.72289536408098e-06,
      "loss": 2.1572,
      "step": 218837
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.3669755458831787,
      "learning_rate": 3.722252961013572e-06,
      "loss": 1.9444,
      "step": 218838
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.4341121912002563,
      "learning_rate": 3.7216106130297685e-06,
      "loss": 1.9255,
      "step": 218839
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.5151256322860718,
      "learning_rate": 3.7209683201297027e-06,
      "loss": 1.9817,
      "step": 218840
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.377875804901123,
      "learning_rate": 3.720326082313474e-06,
      "loss": 1.8326,
      "step": 218841
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.4904940128326416,
      "learning_rate": 3.7196838995812495e-06,
      "loss": 1.8993,
      "step": 218842
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.4503815174102783,
      "learning_rate": 3.719041771933129e-06,
      "loss": 1.8759,
      "step": 218843
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.5013765096664429,
      "learning_rate": 3.7183996993692455e-06,
      "loss": 1.8977,
      "step": 218844
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.3952304124832153,
      "learning_rate": 3.717757681889699e-06,
      "loss": 1.9456,
      "step": 218845
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.4716414213180542,
      "learning_rate": 3.717115719494623e-06,
      "loss": 1.7557,
      "step": 218846
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.6218323707580566,
      "learning_rate": 3.716473812184084e-06,
      "loss": 2.2471,
      "step": 218847
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.653386116027832,
      "learning_rate": 3.715831959958282e-06,
      "loss": 2.1808,
      "step": 218848
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.5259627103805542,
      "learning_rate": 3.715190162817283e-06,
      "loss": 2.052,
      "step": 218849
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.3770781755447388,
      "learning_rate": 3.714548420761254e-06,
      "loss": 1.9562,
      "step": 218850
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.4331775903701782,
      "learning_rate": 3.7139067337902617e-06,
      "loss": 2.0178,
      "step": 218851
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.5781248807907104,
      "learning_rate": 3.713265101904439e-06,
      "loss": 2.0864,
      "step": 218852
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.4777718782424927,
      "learning_rate": 3.7126235251039194e-06,
      "loss": 2.1051,
      "step": 218853
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.3499345779418945,
      "learning_rate": 3.7119820033888028e-06,
      "loss": 1.8514,
      "step": 218854
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.372063159942627,
      "learning_rate": 3.711340536759222e-06,
      "loss": 1.9266,
      "step": 218855
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.3618037700653076,
      "learning_rate": 3.710699125215344e-06,
      "loss": 1.8817,
      "step": 218856
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.3728727102279663,
      "learning_rate": 3.710057768757202e-06,
      "loss": 1.8055,
      "step": 218857
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.4269626140594482,
      "learning_rate": 3.7094164673849626e-06,
      "loss": 1.826,
      "step": 218858
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.4722691774368286,
      "learning_rate": 3.7087752210987253e-06,
      "loss": 2.0767,
      "step": 218859
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.5739845037460327,
      "learning_rate": 3.708134029898624e-06,
      "loss": 2.0286,
      "step": 218860
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.3381237983703613,
      "learning_rate": 3.7074928937847583e-06,
      "loss": 1.7117,
      "step": 218861
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.6083557605743408,
      "learning_rate": 3.706851812757294e-06,
      "loss": 2.1244,
      "step": 218862
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.439171314239502,
      "learning_rate": 3.7062107868162993e-06,
      "loss": 2.2731,
      "step": 218863
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.4240925312042236,
      "learning_rate": 3.7055698159619398e-06,
      "loss": 1.9044,
      "step": 218864
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.4405713081359863,
      "learning_rate": 3.7049289001942485e-06,
      "loss": 1.8463,
      "step": 218865
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.6501590013504028,
      "learning_rate": 3.704288039513459e-06,
      "loss": 1.9235,
      "step": 218866
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.3845739364624023,
      "learning_rate": 3.7036472339196043e-06,
      "loss": 1.8508,
      "step": 218867
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.4631085395812988,
      "learning_rate": 3.7030064834128513e-06,
      "loss": 2.0563,
      "step": 218868
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.4718588590621948,
      "learning_rate": 3.7023657879933e-06,
      "loss": 1.9196,
      "step": 218869
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.4599206447601318,
      "learning_rate": 3.7017251476610833e-06,
      "loss": 2.0389,
      "step": 218870
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.621109962463379,
      "learning_rate": 3.701084562416301e-06,
      "loss": 2.2693,
      "step": 218871
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.5241819620132446,
      "learning_rate": 3.7004440322590534e-06,
      "loss": 1.8489,
      "step": 218872
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.4581198692321777,
      "learning_rate": 3.699803557189507e-06,
      "loss": 1.8638,
      "step": 218873
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.5631405115127563,
      "learning_rate": 3.699163137207761e-06,
      "loss": 1.8698,
      "step": 218874
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.4774255752563477,
      "learning_rate": 3.6985227723139166e-06,
      "loss": 1.7621,
      "step": 218875
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.3108519315719604,
      "learning_rate": 3.697882462508139e-06,
      "loss": 1.918,
      "step": 218876
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.6117737293243408,
      "learning_rate": 3.6972422077905293e-06,
      "loss": 2.0797,
      "step": 218877
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.5548648834228516,
      "learning_rate": 3.6966020081611535e-06,
      "loss": 1.8451,
      "step": 218878
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.4418443441390991,
      "learning_rate": 3.695961863620178e-06,
      "loss": 1.977,
      "step": 218879
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.4951454401016235,
      "learning_rate": 3.6953217741677697e-06,
      "loss": 1.9045,
      "step": 218880
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.4059619903564453,
      "learning_rate": 3.6946817398039286e-06,
      "loss": 1.9227,
      "step": 218881
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.4471997022628784,
      "learning_rate": 3.6940417605288543e-06,
      "loss": 1.9172,
      "step": 218882
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.3452316522598267,
      "learning_rate": 3.69340183634268e-06,
      "loss": 1.8065,
      "step": 218883
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.5181885957717896,
      "learning_rate": 3.692761967245472e-06,
      "loss": 1.8667,
      "step": 218884
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.5555953979492188,
      "learning_rate": 3.6921221532373644e-06,
      "loss": 2.0382,
      "step": 218885
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.522158145904541,
      "learning_rate": 3.6914823943185234e-06,
      "loss": 1.9691,
      "step": 218886
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.367272138595581,
      "learning_rate": 3.690842690488982e-06,
      "loss": 1.8904,
      "step": 218887
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.4127557277679443,
      "learning_rate": 3.6902030417489404e-06,
      "loss": 2.0709,
      "step": 218888
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.656631588935852,
      "learning_rate": 3.689563448098465e-06,
      "loss": 2.0027,
      "step": 218889
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.475032091140747,
      "learning_rate": 3.6889239095376886e-06,
      "loss": 1.7889,
      "step": 218890
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.418514609336853,
      "learning_rate": 3.6882844260667788e-06,
      "loss": 1.7904,
      "step": 218891
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.618729829788208,
      "learning_rate": 3.687644997685768e-06,
      "loss": 2.0279,
      "step": 218892
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.378132939338684,
      "learning_rate": 3.6870056243948233e-06,
      "loss": 1.8369,
      "step": 218893
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.5328923463821411,
      "learning_rate": 3.686366306194044e-06,
      "loss": 2.0982,
      "step": 218894
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.5301905870437622,
      "learning_rate": 3.685727043083597e-06,
      "loss": 1.8365,
      "step": 218895
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.5869226455688477,
      "learning_rate": 3.685087835063549e-06,
      "loss": 1.8436,
      "step": 218896
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.6871691942214966,
      "learning_rate": 3.6844486821340337e-06,
      "loss": 2.1165,
      "step": 218897
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.3378918170928955,
      "learning_rate": 3.6838095842951497e-06,
      "loss": 1.8928,
      "step": 218898
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.4082247018814087,
      "learning_rate": 3.6831705415470646e-06,
      "loss": 1.6835,
      "step": 218899
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.4885659217834473,
      "learning_rate": 3.682531553889878e-06,
      "loss": 1.9301,
      "step": 218900
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.4666097164154053,
      "learning_rate": 3.6818926213237234e-06,
      "loss": 2.0417,
      "step": 218901
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.302315354347229,
      "learning_rate": 3.6812537438486332e-06,
      "loss": 1.9613,
      "step": 218902
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.4054635763168335,
      "learning_rate": 3.6806149214648417e-06,
      "loss": 2.028,
      "step": 218903
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.5830150842666626,
      "learning_rate": 3.679976154172415e-06,
      "loss": 2.0029,
      "step": 218904
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.4201104640960693,
      "learning_rate": 3.6793374419714526e-06,
      "loss": 1.9182,
      "step": 218905
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.556881070137024,
      "learning_rate": 3.678698784862122e-06,
      "loss": 2.131,
      "step": 218906
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.4852349758148193,
      "learning_rate": 3.678060182844489e-06,
      "loss": 1.9381,
      "step": 218907
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.5367687940597534,
      "learning_rate": 3.6774216359187203e-06,
      "loss": 1.9797,
      "step": 218908
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.4623910188674927,
      "learning_rate": 3.676783144084883e-06,
      "loss": 1.9302,
      "step": 218909
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.515974760055542,
      "learning_rate": 3.6761447073431427e-06,
      "loss": 1.7546,
      "step": 218910
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.4823567867279053,
      "learning_rate": 3.6755063256936003e-06,
      "loss": 2.0345,
      "step": 218911
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.451524019241333,
      "learning_rate": 3.674867999136388e-06,
      "loss": 2.0384,
      "step": 218912
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.4383666515350342,
      "learning_rate": 3.674229727671607e-06,
      "loss": 1.8673,
      "step": 218913
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.6190351247787476,
      "learning_rate": 3.6735915112993564e-06,
      "loss": 2.0265,
      "step": 218914
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.5916835069656372,
      "learning_rate": 3.6729533500198026e-06,
      "loss": 1.8997,
      "step": 218915
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.5209001302719116,
      "learning_rate": 3.6723152438330128e-06,
      "loss": 2.1476,
      "step": 218916
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.6345981359481812,
      "learning_rate": 3.671677192739153e-06,
      "loss": 1.9149,
      "step": 218917
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.636574387550354,
      "learning_rate": 3.6710391967383234e-06,
      "loss": 2.0528,
      "step": 218918
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.739967703819275,
      "learning_rate": 3.6704012558306573e-06,
      "loss": 1.9034,
      "step": 218919
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.472499132156372,
      "learning_rate": 3.669763370016221e-06,
      "loss": 1.9306,
      "step": 218920
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.441321849822998,
      "learning_rate": 3.6691255392951814e-06,
      "loss": 1.802,
      "step": 218921
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.4897942543029785,
      "learning_rate": 3.668487763667638e-06,
      "loss": 2.2809,
      "step": 218922
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.5075246095657349,
      "learning_rate": 3.6678500431337577e-06,
      "loss": 2.0096,
      "step": 218923
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.6286262273788452,
      "learning_rate": 3.6672123776935736e-06,
      "loss": 1.905,
      "step": 218924
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.2764972448349,
      "learning_rate": 3.6665747673472854e-06,
      "loss": 1.945,
      "step": 218925
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.4136847257614136,
      "learning_rate": 3.665937212094927e-06,
      "loss": 1.9657,
      "step": 218926
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.3507777452468872,
      "learning_rate": 3.665299711936698e-06,
      "loss": 1.9392,
      "step": 218927
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.432600498199463,
      "learning_rate": 3.6646622668726645e-06,
      "loss": 2.0104,
      "step": 218928
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.4285979270935059,
      "learning_rate": 3.6640248769029935e-06,
      "loss": 2.0466,
      "step": 218929
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.3703484535217285,
      "learning_rate": 3.663387542027718e-06,
      "loss": 2.1241,
      "step": 218930
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.4784833192825317,
      "learning_rate": 3.662750262247105e-06,
      "loss": 1.992,
      "step": 218931
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.515305995941162,
      "learning_rate": 3.6621130375610876e-06,
      "loss": 1.9157,
      "step": 218932
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.421926498413086,
      "learning_rate": 3.6614758679699317e-06,
      "loss": 1.8989,
      "step": 218933
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.4435755014419556,
      "learning_rate": 3.6608387534736714e-06,
      "loss": 1.7813,
      "step": 218934
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.5511174201965332,
      "learning_rate": 3.660201694072473e-06,
      "loss": 1.8196,
      "step": 218935
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.4297242164611816,
      "learning_rate": 3.6595646897664367e-06,
      "loss": 2.1205,
      "step": 218936
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.4730664491653442,
      "learning_rate": 3.6589277405556948e-06,
      "loss": 1.9872,
      "step": 218937
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.5277823209762573,
      "learning_rate": 3.658290846440315e-06,
      "loss": 1.7297,
      "step": 218938
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.4231268167495728,
      "learning_rate": 3.6576540074204962e-06,
      "loss": 1.9071,
      "step": 218939
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.4206037521362305,
      "learning_rate": 3.6570172234962726e-06,
      "loss": 2.0337,
      "step": 218940
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.4078115224838257,
      "learning_rate": 3.65638049466781e-06,
      "loss": 2.029,
      "step": 218941
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.4667454957962036,
      "learning_rate": 3.655743820935242e-06,
      "loss": 1.9084,
      "step": 218942
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.4608460664749146,
      "learning_rate": 3.655107202298668e-06,
      "loss": 2.0802,
      "step": 218943
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.5268298387527466,
      "learning_rate": 3.654470638758189e-06,
      "loss": 1.9194,
      "step": 218944
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.464486002922058,
      "learning_rate": 3.6538341303139375e-06,
      "loss": 1.9644,
      "step": 218945
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.3537262678146362,
      "learning_rate": 3.653197676966013e-06,
      "loss": 1.8557,
      "step": 218946
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.5147705078125,
      "learning_rate": 3.652561278714583e-06,
      "loss": 2.1952,
      "step": 218947
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.4653611183166504,
      "learning_rate": 3.6519249355597135e-06,
      "loss": 2.0414,
      "step": 218948
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.5172390937805176,
      "learning_rate": 3.651288647501571e-06,
      "loss": 1.8945,
      "step": 218949
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.372994303703308,
      "learning_rate": 3.650652414540223e-06,
      "loss": 2.0511,
      "step": 218950
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.3593965768814087,
      "learning_rate": 3.6500162366758347e-06,
      "loss": 2.0118,
      "step": 218951
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.363210678100586,
      "learning_rate": 3.6493801139084732e-06,
      "loss": 1.8077,
      "step": 218952
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.5236539840698242,
      "learning_rate": 3.6487440462383388e-06,
      "loss": 1.8958,
      "step": 218953
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.8813105821609497,
      "learning_rate": 3.6481080336654313e-06,
      "loss": 1.8695,
      "step": 218954
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.4910916090011597,
      "learning_rate": 3.6474720761899833e-06,
      "loss": 1.9895,
      "step": 218955
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.2941149473190308,
      "learning_rate": 3.646836173812062e-06,
      "loss": 1.9312,
      "step": 218956
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.4945652484893799,
      "learning_rate": 3.6462003265317673e-06,
      "loss": 1.8999,
      "step": 218957
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.4544758796691895,
      "learning_rate": 3.6455645343492323e-06,
      "loss": 2.1851,
      "step": 218958
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.4851831197738647,
      "learning_rate": 3.64492879726459e-06,
      "loss": 2.0205,
      "step": 218959
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.4972362518310547,
      "learning_rate": 3.644293115277941e-06,
      "loss": 1.8894,
      "step": 218960
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.3943374156951904,
      "learning_rate": 3.6436574883894178e-06,
      "loss": 2.0453,
      "step": 218961
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.475141167640686,
      "learning_rate": 3.643021916599154e-06,
      "loss": 1.9539,
      "step": 218962
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.4353920221328735,
      "learning_rate": 3.642386399907249e-06,
      "loss": 1.9652,
      "step": 218963
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.5187519788742065,
      "learning_rate": 3.64175093831377e-06,
      "loss": 2.1175,
      "step": 218964
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.5377508401870728,
      "learning_rate": 3.64111553181895e-06,
      "loss": 1.9857,
      "step": 218965
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.3769798278808594,
      "learning_rate": 3.640480180422789e-06,
      "loss": 1.8099,
      "step": 218966
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.4570720195770264,
      "learning_rate": 3.6398448841254537e-06,
      "loss": 2.0715,
      "step": 218967
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.375919222831726,
      "learning_rate": 3.63920964292711e-06,
      "loss": 1.8453,
      "step": 218968
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.4978245496749878,
      "learning_rate": 3.638574456827792e-06,
      "loss": 2.009,
      "step": 218969
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.4775810241699219,
      "learning_rate": 3.6379393258276657e-06,
      "loss": 1.9046,
      "step": 218970
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.6018118858337402,
      "learning_rate": 3.6373042499268645e-06,
      "loss": 2.1889,
      "step": 218971
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.6013188362121582,
      "learning_rate": 3.6366692291254552e-06,
      "loss": 2.052,
      "step": 218972
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.6108015775680542,
      "learning_rate": 3.636034263423604e-06,
      "loss": 2.0218,
      "step": 218973
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.354927897453308,
      "learning_rate": 3.6353993528213777e-06,
      "loss": 1.9553,
      "step": 218974
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.4673559665679932,
      "learning_rate": 3.634764497318943e-06,
      "loss": 1.858,
      "step": 218975
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.3992053270339966,
      "learning_rate": 3.6341296969163992e-06,
      "loss": 1.8356,
      "step": 218976
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.6182650327682495,
      "learning_rate": 3.63349495161388e-06,
      "loss": 2.0707,
      "step": 218977
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.5446697473526,
      "learning_rate": 3.6328602614114523e-06,
      "loss": 1.9002,
      "step": 218978
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.4559316635131836,
      "learning_rate": 3.6322256263093154e-06,
      "loss": 1.9393,
      "step": 218979
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.4430276155471802,
      "learning_rate": 3.631591046307536e-06,
      "loss": 2.0181,
      "step": 218980
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.4720834493637085,
      "learning_rate": 3.6309565214062143e-06,
      "loss": 2.069,
      "step": 218981
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.3945891857147217,
      "learning_rate": 3.630322051605483e-06,
      "loss": 1.9435,
      "step": 218982
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.459925651550293,
      "learning_rate": 3.6296876369055093e-06,
      "loss": 1.9787,
      "step": 218983
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.4175037145614624,
      "learning_rate": 3.6290532773063264e-06,
      "loss": 2.1875,
      "step": 218984
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.4759749174118042,
      "learning_rate": 3.6284189728081334e-06,
      "loss": 2.1526,
      "step": 218985
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.4404805898666382,
      "learning_rate": 3.627784723410998e-06,
      "loss": 1.9621,
      "step": 218986
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.4868017435073853,
      "learning_rate": 3.6271505291150527e-06,
      "loss": 1.9703,
      "step": 218987
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.610695719718933,
      "learning_rate": 3.6265163899203974e-06,
      "loss": 2.193,
      "step": 218988
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.479384422302246,
      "learning_rate": 3.6258823058271657e-06,
      "loss": 1.7854,
      "step": 218989
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.5818599462509155,
      "learning_rate": 3.625248276835524e-06,
      "loss": 2.0417,
      "step": 218990
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.325145959854126,
      "learning_rate": 3.6246143029455056e-06,
      "loss": 2.1312,
      "step": 218991
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.3351315259933472,
      "learning_rate": 3.6239803841572766e-06,
      "loss": 1.8037,
      "step": 218992
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.475112795829773,
      "learning_rate": 3.623346520470938e-06,
      "loss": 2.0241,
      "step": 218993
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.5726051330566406,
      "learning_rate": 3.6227127118866217e-06,
      "loss": 1.9857,
      "step": 218994
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.4454538822174072,
      "learning_rate": 3.6220789584044287e-06,
      "loss": 2.0575,
      "step": 218995
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.4126827716827393,
      "learning_rate": 3.6214452600244916e-06,
      "loss": 2.1252,
      "step": 218996
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.4709632396697998,
      "learning_rate": 3.62081161674691e-06,
      "loss": 1.8418,
      "step": 218997
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.654427170753479,
      "learning_rate": 3.620178028571852e-06,
      "loss": 1.9341,
      "step": 218998
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.4223442077636719,
      "learning_rate": 3.619544495499349e-06,
      "loss": 2.1223,
      "step": 218999
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.3994834423065186,
      "learning_rate": 3.618911017529602e-06,
      "loss": 1.9074,
      "step": 219000
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.4993352890014648,
      "learning_rate": 3.6182775946626775e-06,
      "loss": 1.9596,
      "step": 219001
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.3897942304611206,
      "learning_rate": 3.6176442268987083e-06,
      "loss": 1.7168,
      "step": 219002
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.640190601348877,
      "learning_rate": 3.6170109142377946e-06,
      "loss": 1.8136,
      "step": 219003
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.4898263216018677,
      "learning_rate": 3.616377656680136e-06,
      "loss": 1.947,
      "step": 219004
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.416032075881958,
      "learning_rate": 3.615744454225733e-06,
      "loss": 1.9631,
      "step": 219005
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.415236473083496,
      "learning_rate": 3.615111306874785e-06,
      "loss": 1.8882,
      "step": 219006
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.3699160814285278,
      "learning_rate": 3.6144782146273586e-06,
      "loss": 1.8371,
      "step": 219007
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.36683189868927,
      "learning_rate": 3.613845177483621e-06,
      "loss": 2.1289,
      "step": 219008
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.5608724355697632,
      "learning_rate": 3.6132121954436376e-06,
      "loss": 2.1151,
      "step": 219009
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.4555392265319824,
      "learning_rate": 3.612579268507576e-06,
      "loss": 2.0434,
      "step": 219010
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.460451602935791,
      "learning_rate": 3.6119463966755023e-06,
      "loss": 1.8843,
      "step": 219011
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.513667345046997,
      "learning_rate": 3.6113135799475832e-06,
      "loss": 1.8899,
      "step": 219012
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.4441155195236206,
      "learning_rate": 3.610680818323919e-06,
      "loss": 1.7215,
      "step": 219013
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.455064058303833,
      "learning_rate": 3.6100481118046088e-06,
      "loss": 1.85,
      "step": 219014
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.5147079229354858,
      "learning_rate": 3.6094154603897862e-06,
      "loss": 2.0059,
      "step": 219015
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.602699637413025,
      "learning_rate": 3.608782864079618e-06,
      "loss": 2.1119,
      "step": 219016
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.4988222122192383,
      "learning_rate": 3.6081503228741037e-06,
      "loss": 2.0974,
      "step": 219017
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.3184621334075928,
      "learning_rate": 3.607517836773477e-06,
      "loss": 1.9913,
      "step": 219018
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.4333137273788452,
      "learning_rate": 3.6068854057777707e-06,
      "loss": 1.9603,
      "step": 219019
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.363938570022583,
      "learning_rate": 3.6062530298871517e-06,
      "loss": 1.9637,
      "step": 219020
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.400355339050293,
      "learning_rate": 3.6056207091017197e-06,
      "loss": 1.7731,
      "step": 219021
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.4137718677520752,
      "learning_rate": 3.6049884434216415e-06,
      "loss": 1.8472,
      "step": 219022
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.6565097570419312,
      "learning_rate": 3.60435623284695e-06,
      "loss": 1.9521,
      "step": 219023
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.564832091331482,
      "learning_rate": 3.603724077377812e-06,
      "loss": 2.1189,
      "step": 219024
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.3438560962677002,
      "learning_rate": 3.6030919770143274e-06,
      "loss": 1.8571,
      "step": 219025
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.5517021417617798,
      "learning_rate": 3.6024599317566294e-06,
      "loss": 1.7854,
      "step": 219026
    },
    {
      "epoch": 2.85,
      "grad_norm": 3.2669947147369385,
      "learning_rate": 3.601827941604818e-06,
      "loss": 1.9019,
      "step": 219027
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.6558910608291626,
      "learning_rate": 3.6011960065590596e-06,
      "loss": 2.0607,
      "step": 219028
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.4752572774887085,
      "learning_rate": 3.6005641266193873e-06,
      "loss": 2.0414,
      "step": 219029
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.526036262512207,
      "learning_rate": 3.5999323017860015e-06,
      "loss": 1.6967,
      "step": 219030
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.4819589853286743,
      "learning_rate": 3.599300532058935e-06,
      "loss": 2.2494,
      "step": 219031
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.3392242193222046,
      "learning_rate": 3.598668817438388e-06,
      "loss": 1.7903,
      "step": 219032
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.4758943319320679,
      "learning_rate": 3.5980371579244603e-06,
      "loss": 1.9895,
      "step": 219033
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.4223986864089966,
      "learning_rate": 3.597405553517252e-06,
      "loss": 1.685,
      "step": 219034
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.5181362628936768,
      "learning_rate": 3.596774004216829e-06,
      "loss": 1.8135,
      "step": 219035
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.4704726934432983,
      "learning_rate": 3.5961425100234256e-06,
      "loss": 2.0255,
      "step": 219036
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.4783557653427124,
      "learning_rate": 3.5955110709370406e-06,
      "loss": 1.8181,
      "step": 219037
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.6103932857513428,
      "learning_rate": 3.5948796869578743e-06,
      "loss": 1.9403,
      "step": 219038
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.4560142755508423,
      "learning_rate": 3.594248358085994e-06,
      "loss": 1.9826,
      "step": 219039
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.467202067375183,
      "learning_rate": 3.593617084321565e-06,
      "loss": 2.1253,
      "step": 219040
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.4385735988616943,
      "learning_rate": 3.592985865664655e-06,
      "loss": 2.0776,
      "step": 219041
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.4626842737197876,
      "learning_rate": 3.5923547021154297e-06,
      "loss": 1.847,
      "step": 219042
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.5361827611923218,
      "learning_rate": 3.5917235936739564e-06,
      "loss": 1.8616,
      "step": 219043
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.5981900691986084,
      "learning_rate": 3.5910925403404012e-06,
      "loss": 2.0901,
      "step": 219044
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.5027499198913574,
      "learning_rate": 3.5904615421147973e-06,
      "loss": 2.0105,
      "step": 219045
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.5160449743270874,
      "learning_rate": 3.589830598997412e-06,
      "loss": 1.9886,
      "step": 219046
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.5355385541915894,
      "learning_rate": 3.5891997109882108e-06,
      "loss": 2.2003,
      "step": 219047
    },
    {
      "epoch": 2.85,
      "grad_norm": 2.3517539501190186,
      "learning_rate": 3.5885688780873945e-06,
      "loss": 1.8877,
      "step": 219048
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.46694815158844,
      "learning_rate": 3.5879381002950293e-06,
      "loss": 2.0385,
      "step": 219049
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.647873044013977,
      "learning_rate": 3.5873073776113146e-06,
      "loss": 2.0452,
      "step": 219050
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.4205256700515747,
      "learning_rate": 3.5866767100362516e-06,
      "loss": 1.9141,
      "step": 219051
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.5916104316711426,
      "learning_rate": 3.586046097570072e-06,
      "loss": 2.1506,
      "step": 219052
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.4835141897201538,
      "learning_rate": 3.5854155402128103e-06,
      "loss": 1.9741,
      "step": 219053
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.451845407485962,
      "learning_rate": 3.5847850379646327e-06,
      "loss": 1.7697,
      "step": 219054
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.3867641687393188,
      "learning_rate": 3.5841545908256387e-06,
      "loss": 2.0257,
      "step": 219055
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.661637783050537,
      "learning_rate": 3.5835241987959285e-06,
      "loss": 1.9601,
      "step": 219056
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.456107258796692,
      "learning_rate": 3.5828938618756352e-06,
      "loss": 1.6877,
      "step": 219057
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.489775538444519,
      "learning_rate": 3.582263580064926e-06,
      "loss": 1.8619,
      "step": 219058
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.4183868169784546,
      "learning_rate": 3.581633353363833e-06,
      "loss": 1.8631,
      "step": 219059
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.4796593189239502,
      "learning_rate": 3.5810031817724903e-06,
      "loss": 2.0863,
      "step": 219060
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.5676883459091187,
      "learning_rate": 3.5803730652910643e-06,
      "loss": 1.78,
      "step": 219061
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.8101072311401367,
      "learning_rate": 3.579743003919655e-06,
      "loss": 2.0514,
      "step": 219062
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.516440510749817,
      "learning_rate": 3.5791129976583286e-06,
      "loss": 1.8019,
      "step": 219063
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.6121801137924194,
      "learning_rate": 3.5784830465072523e-06,
      "loss": 2.2051,
      "step": 219064
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.3990046977996826,
      "learning_rate": 3.577853150466525e-06,
      "loss": 1.9786,
      "step": 219065
    },
    {
      "epoch": 2.85,
      "grad_norm": 3.3056273460388184,
      "learning_rate": 3.577223309536281e-06,
      "loss": 2.1629,
      "step": 219066
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.4675432443618774,
      "learning_rate": 3.57659352371662e-06,
      "loss": 1.8383,
      "step": 219067
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.4569059610366821,
      "learning_rate": 3.575963793007708e-06,
      "loss": 2.0125,
      "step": 219068
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.637663722038269,
      "learning_rate": 3.575334117409545e-06,
      "loss": 1.8547,
      "step": 219069
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.4531238079071045,
      "learning_rate": 3.5747044969223647e-06,
      "loss": 1.9406,
      "step": 219070
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.4719291925430298,
      "learning_rate": 3.574074931546267e-06,
      "loss": 1.7159,
      "step": 219071
    },
    {
      "epoch": 2.85,
      "grad_norm": 2.0440454483032227,
      "learning_rate": 3.5734454212812847e-06,
      "loss": 1.8958,
      "step": 219072
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.5462093353271484,
      "learning_rate": 3.5728159661276513e-06,
      "loss": 2.0409,
      "step": 219073
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.2964190244674683,
      "learning_rate": 3.5721865660854e-06,
      "loss": 1.7331,
      "step": 219074
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.479355812072754,
      "learning_rate": 3.571557221154664e-06,
      "loss": 1.884,
      "step": 219075
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.397673487663269,
      "learning_rate": 3.57092793133561e-06,
      "loss": 2.1913,
      "step": 219076
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.4611924886703491,
      "learning_rate": 3.570298696628271e-06,
      "loss": 1.982,
      "step": 219077
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.575209140777588,
      "learning_rate": 3.569669517032814e-06,
      "loss": 1.935,
      "step": 219078
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.4855886697769165,
      "learning_rate": 3.5690403925493716e-06,
      "loss": 2.0452,
      "step": 219079
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.4340403079986572,
      "learning_rate": 3.5684113231780443e-06,
      "loss": 1.795,
      "step": 219080
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.5735410451889038,
      "learning_rate": 3.567782308918932e-06,
      "loss": 1.9037,
      "step": 219081
    },
    {
      "epoch": 2.85,
      "grad_norm": 2.3301823139190674,
      "learning_rate": 3.567153349772134e-06,
      "loss": 2.0962,
      "step": 219082
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.402830958366394,
      "learning_rate": 3.566524445737884e-06,
      "loss": 1.9412,
      "step": 219083
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.5571033954620361,
      "learning_rate": 3.5658955968161486e-06,
      "loss": 2.0662,
      "step": 219084
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.4805577993392944,
      "learning_rate": 3.5652668030070943e-06,
      "loss": 1.926,
      "step": 219085
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.52242910861969,
      "learning_rate": 3.5646380643108875e-06,
      "loss": 2.0289,
      "step": 219086
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.5433074235916138,
      "learning_rate": 3.5640093807275948e-06,
      "loss": 1.989,
      "step": 219087
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.5042929649353027,
      "learning_rate": 3.56338075225735e-06,
      "loss": 1.8847,
      "step": 219088
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.4932361841201782,
      "learning_rate": 3.5627521789002854e-06,
      "loss": 2.0086,
      "step": 219089
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.467943549156189,
      "learning_rate": 3.5621236606564683e-06,
      "loss": 1.9511,
      "step": 219090
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.5685086250305176,
      "learning_rate": 3.5614951975260985e-06,
      "loss": 1.9612,
      "step": 219091
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.4855941534042358,
      "learning_rate": 3.560866789509176e-06,
      "loss": 2.0757,
      "step": 219092
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.3995895385742188,
      "learning_rate": 3.5602384366059334e-06,
      "loss": 1.8712,
      "step": 219093
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.5900187492370605,
      "learning_rate": 3.5596101388164046e-06,
      "loss": 2.0512,
      "step": 219094
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.416906714439392,
      "learning_rate": 3.558981896140789e-06,
      "loss": 1.9227,
      "step": 219095
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.5989890098571777,
      "learning_rate": 3.55835370857912e-06,
      "loss": 1.9013,
      "step": 219096
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.3122001886367798,
      "learning_rate": 3.557725576131565e-06,
      "loss": 2.0364,
      "step": 219097
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.5650105476379395,
      "learning_rate": 3.5570974987981895e-06,
      "loss": 1.9838,
      "step": 219098
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.4537516832351685,
      "learning_rate": 3.5564694765791935e-06,
      "loss": 1.9554,
      "step": 219099
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.4549750089645386,
      "learning_rate": 3.555841509474611e-06,
      "loss": 2.0508,
      "step": 219100
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.401135802268982,
      "learning_rate": 3.5552135974846074e-06,
      "loss": 1.7369,
      "step": 219101
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.4093561172485352,
      "learning_rate": 3.5545857406092837e-06,
      "loss": 1.857,
      "step": 219102
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.4977192878723145,
      "learning_rate": 3.5539579388487726e-06,
      "loss": 2.108,
      "step": 219103
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.4600505828857422,
      "learning_rate": 3.553330192203141e-06,
      "loss": 2.0029,
      "step": 219104
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.5785467624664307,
      "learning_rate": 3.5527025006725552e-06,
      "loss": 2.0594,
      "step": 219105
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.5180920362472534,
      "learning_rate": 3.552074864257115e-06,
      "loss": 1.7708,
      "step": 219106
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.360573649406433,
      "learning_rate": 3.5514472829569873e-06,
      "loss": 2.0783,
      "step": 219107
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.471065878868103,
      "learning_rate": 3.550819756772172e-06,
      "loss": 2.1826,
      "step": 219108
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.4721176624298096,
      "learning_rate": 3.550192285702902e-06,
      "loss": 2.0048,
      "step": 219109
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.4154115915298462,
      "learning_rate": 3.5495648697492106e-06,
      "loss": 1.9103,
      "step": 219110
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.3924685716629028,
      "learning_rate": 3.548937508911265e-06,
      "loss": 1.9533,
      "step": 219111
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.5996365547180176,
      "learning_rate": 3.5483102031891643e-06,
      "loss": 2.1096,
      "step": 219112
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.8704967498779297,
      "learning_rate": 3.5476829525830754e-06,
      "loss": 2.0225,
      "step": 219113
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.451812982559204,
      "learning_rate": 3.5470557570930315e-06,
      "loss": 1.9526,
      "step": 219114
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.5598416328430176,
      "learning_rate": 3.546428616719166e-06,
      "loss": 2.3269,
      "step": 219115
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.3001439571380615,
      "learning_rate": 3.545801531461645e-06,
      "loss": 1.9069,
      "step": 219116
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.45280122756958,
      "learning_rate": 3.5451745013205356e-06,
      "loss": 1.9913,
      "step": 219117
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.4036967754364014,
      "learning_rate": 3.544547526295971e-06,
      "loss": 2.1706,
      "step": 219118
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.2685012817382812,
      "learning_rate": 3.5439206063881178e-06,
      "loss": 2.0665,
      "step": 219119
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.435171365737915,
      "learning_rate": 3.543293741597009e-06,
      "loss": 1.9094,
      "step": 219120
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.5418705940246582,
      "learning_rate": 3.5426669319228106e-06,
      "loss": 1.8531,
      "step": 219121
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.43007493019104,
      "learning_rate": 3.5420401773655905e-06,
      "loss": 1.7239,
      "step": 219122
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.5426082611083984,
      "learning_rate": 3.5414134779255476e-06,
      "loss": 1.9036,
      "step": 219123
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.4678655862808228,
      "learning_rate": 3.5407868336027156e-06,
      "loss": 2.0269,
      "step": 219124
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.700195074081421,
      "learning_rate": 3.5401602443972942e-06,
      "loss": 1.9756,
      "step": 219125
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.5374958515167236,
      "learning_rate": 3.5395337103093168e-06,
      "loss": 1.9416,
      "step": 219126
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.4025405645370483,
      "learning_rate": 3.5389072313389497e-06,
      "loss": 1.9565,
      "step": 219127
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.4590232372283936,
      "learning_rate": 3.53828080748626e-06,
      "loss": 1.8384,
      "step": 219128
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.3520011901855469,
      "learning_rate": 3.537654438751447e-06,
      "loss": 2.1484,
      "step": 219129
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.4808545112609863,
      "learning_rate": 3.5370281251345444e-06,
      "loss": 1.959,
      "step": 219130
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.40223228931427,
      "learning_rate": 3.5364018666357185e-06,
      "loss": 2.0288,
      "step": 219131
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.458640456199646,
      "learning_rate": 3.535775663255069e-06,
      "loss": 1.8814,
      "step": 219132
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.425304651260376,
      "learning_rate": 3.53514951499273e-06,
      "loss": 2.0459,
      "step": 219133
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.4522382020950317,
      "learning_rate": 3.5345234218488006e-06,
      "loss": 2.0608,
      "step": 219134
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.4046889543533325,
      "learning_rate": 3.533897383823381e-06,
      "loss": 2.1164,
      "step": 219135
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.6741076707839966,
      "learning_rate": 3.533271400916604e-06,
      "loss": 1.9466,
      "step": 219136
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.5620352029800415,
      "learning_rate": 3.5326454731286367e-06,
      "loss": 1.9131,
      "step": 219137
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.4907053709030151,
      "learning_rate": 3.5320196004595126e-06,
      "loss": 2.0114,
      "step": 219138
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.4583311080932617,
      "learning_rate": 3.5313937829093973e-06,
      "loss": 1.9063,
      "step": 219139
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.5873509645462036,
      "learning_rate": 3.5307680204783582e-06,
      "loss": 1.9124,
      "step": 219140
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.336006760597229,
      "learning_rate": 3.5301423131665618e-06,
      "loss": 1.868,
      "step": 219141
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.4600250720977783,
      "learning_rate": 3.5295166609741075e-06,
      "loss": 2.0777,
      "step": 219142
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.3942534923553467,
      "learning_rate": 3.5288910639010958e-06,
      "loss": 1.8622,
      "step": 219143
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.3880363702774048,
      "learning_rate": 3.528265521947693e-06,
      "loss": 1.8303,
      "step": 219144
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.477781891822815,
      "learning_rate": 3.5276400351139657e-06,
      "loss": 2.1067,
      "step": 219145
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.4371840953826904,
      "learning_rate": 3.527014603400047e-06,
      "loss": 1.8534,
      "step": 219146
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.511131763458252,
      "learning_rate": 3.52638922680607e-06,
      "loss": 1.9315,
      "step": 219147
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.505811333656311,
      "learning_rate": 3.5257639053321017e-06,
      "loss": 2.0505,
      "step": 219148
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.5026823282241821,
      "learning_rate": 3.5251386389783088e-06,
      "loss": 1.816,
      "step": 219149
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.456197738647461,
      "learning_rate": 3.5245134277447906e-06,
      "loss": 2.0333,
      "step": 219150
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.53395676612854,
      "learning_rate": 3.5238882716316475e-06,
      "loss": 2.0174,
      "step": 219151
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.4261432886123657,
      "learning_rate": 3.523263170639046e-06,
      "loss": 2.044,
      "step": 219152
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.4063555002212524,
      "learning_rate": 3.522638124767052e-06,
      "loss": 2.0163,
      "step": 219153
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.4617069959640503,
      "learning_rate": 3.5220131340157666e-06,
      "loss": 2.1122,
      "step": 219154
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.478679895401001,
      "learning_rate": 3.5213881983853556e-06,
      "loss": 1.903,
      "step": 219155
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.4756239652633667,
      "learning_rate": 3.5207633178759185e-06,
      "loss": 1.876,
      "step": 219156
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.5788061618804932,
      "learning_rate": 3.5201384924875562e-06,
      "loss": 2.0307,
      "step": 219157
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.350712537765503,
      "learning_rate": 3.5195137222204017e-06,
      "loss": 1.8986,
      "step": 219158
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.3544222116470337,
      "learning_rate": 3.5188890070745545e-06,
      "loss": 2.1875,
      "step": 219159
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.39106285572052,
      "learning_rate": 3.518264347050148e-06,
      "loss": 2.0268,
      "step": 219160
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.4214487075805664,
      "learning_rate": 3.5176397421473155e-06,
      "loss": 1.8166,
      "step": 219161
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.5167292356491089,
      "learning_rate": 3.517015192366157e-06,
      "loss": 2.134,
      "step": 219162
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.5158910751342773,
      "learning_rate": 3.5163906977067392e-06,
      "loss": 1.8173,
      "step": 219163
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.434195876121521,
      "learning_rate": 3.5157662581692614e-06,
      "loss": 2.1644,
      "step": 219164
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.2945489883422852,
      "learning_rate": 3.5151418737537574e-06,
      "loss": 1.8498,
      "step": 219165
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.3086704015731812,
      "learning_rate": 3.5145175444603934e-06,
      "loss": 2.1407,
      "step": 219166
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.492693305015564,
      "learning_rate": 3.513893270289303e-06,
      "loss": 1.7504,
      "step": 219167
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.3280726671218872,
      "learning_rate": 3.5132690512405857e-06,
      "loss": 2.2393,
      "step": 219168
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.5371218919754028,
      "learning_rate": 3.512644887314309e-06,
      "loss": 1.7685,
      "step": 219169
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.578782320022583,
      "learning_rate": 3.512020778510638e-06,
      "loss": 2.212,
      "step": 219170
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.552404522895813,
      "learning_rate": 3.5113967248296738e-06,
      "loss": 1.9784,
      "step": 219171
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.5559018850326538,
      "learning_rate": 3.5107727262715823e-06,
      "loss": 2.0778,
      "step": 219172
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.7528120279312134,
      "learning_rate": 3.5101487828363974e-06,
      "loss": 2.0423,
      "step": 219173
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.571946144104004,
      "learning_rate": 3.509524894524285e-06,
      "loss": 1.9352,
      "step": 219174
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.4539387226104736,
      "learning_rate": 3.5089010613353453e-06,
      "loss": 1.9193,
      "step": 219175
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.459856390953064,
      "learning_rate": 3.508277283269678e-06,
      "loss": 2.0613,
      "step": 219176
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.3429523706436157,
      "learning_rate": 3.5076535603274502e-06,
      "loss": 1.9609,
      "step": 219177
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.476293683052063,
      "learning_rate": 3.507029892508728e-06,
      "loss": 1.8961,
      "step": 219178
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.5524038076400757,
      "learning_rate": 3.506406279813645e-06,
      "loss": 1.9435,
      "step": 219179
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.4316297769546509,
      "learning_rate": 3.505782722242334e-06,
      "loss": 1.9891,
      "step": 219180
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.6042879819869995,
      "learning_rate": 3.505159219794862e-06,
      "loss": 2.0748,
      "step": 219181
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.621902346611023,
      "learning_rate": 3.5045357724714283e-06,
      "loss": 1.8997,
      "step": 219182
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.4588508605957031,
      "learning_rate": 3.5039123802720336e-06,
      "loss": 1.7932,
      "step": 219183
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.4494072198867798,
      "learning_rate": 3.5032890431969107e-06,
      "loss": 2.0716,
      "step": 219184
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.4227073192596436,
      "learning_rate": 3.502665761246093e-06,
      "loss": 2.0512,
      "step": 219185
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.5303906202316284,
      "learning_rate": 3.5020425344197468e-06,
      "loss": 1.9675,
      "step": 219186
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.4458402395248413,
      "learning_rate": 3.501419362717972e-06,
      "loss": 2.0849,
      "step": 219187
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.3603489398956299,
      "learning_rate": 3.500796246140869e-06,
      "loss": 2.03,
      "step": 219188
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.3710681200027466,
      "learning_rate": 3.500173184688537e-06,
      "loss": 1.9837,
      "step": 219189
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.3959869146347046,
      "learning_rate": 3.4995501783611768e-06,
      "loss": 1.7354,
      "step": 219190
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.456217646598816,
      "learning_rate": 3.4989272271587876e-06,
      "loss": 2.0979,
      "step": 219191
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.5118343830108643,
      "learning_rate": 3.498304331081603e-06,
      "loss": 1.9993,
      "step": 219192
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.51889967918396,
      "learning_rate": 3.4976814901296223e-06,
      "loss": 1.8617,
      "step": 219193
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.5404090881347656,
      "learning_rate": 3.4970587043030795e-06,
      "loss": 1.995,
      "step": 219194
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.4698630571365356,
      "learning_rate": 3.496435973601974e-06,
      "loss": 1.8198,
      "step": 219195
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.5293638706207275,
      "learning_rate": 3.4958132980265064e-06,
      "loss": 1.8532,
      "step": 219196
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.5984615087509155,
      "learning_rate": 3.4951906775767426e-06,
      "loss": 1.845,
      "step": 219197
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.5677833557128906,
      "learning_rate": 3.494568112252849e-06,
      "loss": 1.759,
      "step": 219198
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.4018745422363281,
      "learning_rate": 3.493945602054926e-06,
      "loss": 1.896,
      "step": 219199
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.3492062091827393,
      "learning_rate": 3.49332314698304e-06,
      "loss": 1.785,
      "step": 219200
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.4233630895614624,
      "learning_rate": 3.492700747037358e-06,
      "loss": 1.8784,
      "step": 219201
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.4407389163970947,
      "learning_rate": 3.4920784022179795e-06,
      "loss": 1.8492,
      "step": 219202
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.5006999969482422,
      "learning_rate": 3.491456112525004e-06,
      "loss": 2.059,
      "step": 219203
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.4048681259155273,
      "learning_rate": 3.4908338779585652e-06,
      "loss": 1.8925,
      "step": 219204
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.5252591371536255,
      "learning_rate": 3.4902116985187966e-06,
      "loss": 1.9773,
      "step": 219205
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.427291989326477,
      "learning_rate": 3.4895895742057978e-06,
      "loss": 1.8619,
      "step": 219206
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.5971935987472534,
      "learning_rate": 3.4889675050196687e-06,
      "loss": 1.9099,
      "step": 219207
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.4654481410980225,
      "learning_rate": 3.4883454909605423e-06,
      "loss": 1.9473,
      "step": 219208
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.438773512840271,
      "learning_rate": 3.487723532028519e-06,
      "loss": 1.9049,
      "step": 219209
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.5530543327331543,
      "learning_rate": 3.487101628223732e-06,
      "loss": 1.9769,
      "step": 219210
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.4106169939041138,
      "learning_rate": 3.486479779546314e-06,
      "loss": 1.9651,
      "step": 219211
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.4908548593521118,
      "learning_rate": 3.485857985996332e-06,
      "loss": 2.0038,
      "step": 219212
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.450322151184082,
      "learning_rate": 3.4852362475739194e-06,
      "loss": 1.6926,
      "step": 219213
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.4569960832595825,
      "learning_rate": 3.484614564279209e-06,
      "loss": 1.8766,
      "step": 219214
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.5581433773040771,
      "learning_rate": 3.4839929361122677e-06,
      "loss": 1.9019,
      "step": 219215
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.389716386795044,
      "learning_rate": 3.483371363073295e-06,
      "loss": 2.0919,
      "step": 219216
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.5539672374725342,
      "learning_rate": 3.482749845162358e-06,
      "loss": 1.9385,
      "step": 219217
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.3934050798416138,
      "learning_rate": 3.4821283823795897e-06,
      "loss": 1.8237,
      "step": 219218
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.5001436471939087,
      "learning_rate": 3.481506974725057e-06,
      "loss": 1.8667,
      "step": 219219
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.5021030902862549,
      "learning_rate": 3.480885622198926e-06,
      "loss": 1.9669,
      "step": 219220
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.3177512884140015,
      "learning_rate": 3.480264324801263e-06,
      "loss": 1.7587,
      "step": 219221
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.4395229816436768,
      "learning_rate": 3.4796430825322686e-06,
      "loss": 1.8967,
      "step": 219222
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.4170286655426025,
      "learning_rate": 3.479021895391976e-06,
      "loss": 2.1592,
      "step": 219223
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.478602647781372,
      "learning_rate": 3.478400763380551e-06,
      "loss": 2.0067,
      "step": 219224
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.4364813566207886,
      "learning_rate": 3.477779686498061e-06,
      "loss": 2.0106,
      "step": 219225
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.4339916706085205,
      "learning_rate": 3.4771586647446727e-06,
      "loss": 1.9234,
      "step": 219226
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.423134207725525,
      "learning_rate": 3.4765376981204517e-06,
      "loss": 1.8962,
      "step": 219227
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.5841208696365356,
      "learning_rate": 3.4759167866255653e-06,
      "loss": 2.1241,
      "step": 219228
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.4764097929000854,
      "learning_rate": 3.47529593026008e-06,
      "loss": 1.9395,
      "step": 219229
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.3614451885223389,
      "learning_rate": 3.4746751290241627e-06,
      "loss": 2.139,
      "step": 219230
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.4422059059143066,
      "learning_rate": 3.474054382917879e-06,
      "loss": 1.7989,
      "step": 219231
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.6894129514694214,
      "learning_rate": 3.473433691941363e-06,
      "loss": 2.0849,
      "step": 219232
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.4587496519088745,
      "learning_rate": 3.4728130560947478e-06,
      "loss": 1.932,
      "step": 219233
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.4500073194503784,
      "learning_rate": 3.4721924753781328e-06,
      "loss": 2.1103,
      "step": 219234
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.470578670501709,
      "learning_rate": 3.4715719497916185e-06,
      "loss": 1.9436,
      "step": 219235
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.4226185083389282,
      "learning_rate": 3.4709514793353375e-06,
      "loss": 2.1236,
      "step": 219236
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.478884220123291,
      "learning_rate": 3.470331064009424e-06,
      "loss": 2.0336,
      "step": 219237
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.4433817863464355,
      "learning_rate": 3.469710703813977e-06,
      "loss": 2.0389,
      "step": 219238
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.480438470840454,
      "learning_rate": 3.4690903987490635e-06,
      "loss": 1.8237,
      "step": 219239
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.3879190683364868,
      "learning_rate": 3.4684701488148834e-06,
      "loss": 2.0844,
      "step": 219240
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.5810307264328003,
      "learning_rate": 3.467849954011537e-06,
      "loss": 2.0558,
      "step": 219241
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.4570025205612183,
      "learning_rate": 3.4672298143390562e-06,
      "loss": 2.0295,
      "step": 219242
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.7935614585876465,
      "learning_rate": 3.4666097297976425e-06,
      "loss": 2.0049,
      "step": 219243
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.3518457412719727,
      "learning_rate": 3.4659897003873616e-06,
      "loss": 1.7115,
      "step": 219244
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.4277456998825073,
      "learning_rate": 3.46536972610838e-06,
      "loss": 2.154,
      "step": 219245
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.477866768836975,
      "learning_rate": 3.464749806960798e-06,
      "loss": 2.1973,
      "step": 219246
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.984337568283081,
      "learning_rate": 3.4641299429446823e-06,
      "loss": 2.1182,
      "step": 219247
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.6443899869918823,
      "learning_rate": 3.463510134060199e-06,
      "loss": 2.0108,
      "step": 219248
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.4555697441101074,
      "learning_rate": 3.4628903803074147e-06,
      "loss": 1.9464,
      "step": 219249
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.5400017499923706,
      "learning_rate": 3.4622706816864966e-06,
      "loss": 2.033,
      "step": 219250
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.5526994466781616,
      "learning_rate": 3.461651038197577e-06,
      "loss": 1.9875,
      "step": 219251
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.3572161197662354,
      "learning_rate": 3.46103144984069e-06,
      "loss": 1.7243,
      "step": 219252
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.4729639291763306,
      "learning_rate": 3.4604119166160017e-06,
      "loss": 2.0646,
      "step": 219253
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.4664697647094727,
      "learning_rate": 3.4597924385236453e-06,
      "loss": 1.8382,
      "step": 219254
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.4717503786087036,
      "learning_rate": 3.4591730155636876e-06,
      "loss": 2.0649,
      "step": 219255
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.5599669218063354,
      "learning_rate": 3.4585536477362617e-06,
      "loss": 1.8432,
      "step": 219256
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.4866359233856201,
      "learning_rate": 3.4579343350415013e-06,
      "loss": 1.8696,
      "step": 219257
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.4640998840332031,
      "learning_rate": 3.457315077479472e-06,
      "loss": 1.8694,
      "step": 219258
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.4062336683273315,
      "learning_rate": 3.456695875050408e-06,
      "loss": 1.9072,
      "step": 219259
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.6283973455429077,
      "learning_rate": 3.4560767277542754e-06,
      "loss": 1.8829,
      "step": 219260
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.4657150506973267,
      "learning_rate": 3.4554576355912743e-06,
      "loss": 1.8566,
      "step": 219261
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.47353196144104,
      "learning_rate": 3.4548385985615045e-06,
      "loss": 1.9738,
      "step": 219262
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.6741561889648438,
      "learning_rate": 3.454219616665066e-06,
      "loss": 1.8102,
      "step": 219263
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.4222652912139893,
      "learning_rate": 3.453600689902092e-06,
      "loss": 1.9187,
      "step": 219264
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.3602510690689087,
      "learning_rate": 3.452981818272682e-06,
      "loss": 1.9569,
      "step": 219265
    },
    {
      "epoch": 2.85,
      "grad_norm": 1.407490849494934,
      "learning_rate": 3.4523630017770033e-06,
      "loss": 2.1759,
      "step": 219266
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.5115548372268677,
      "learning_rate": 3.4517442404150884e-06,
      "loss": 2.0456,
      "step": 219267
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.4867075681686401,
      "learning_rate": 3.4511255341870714e-06,
      "loss": 2.2487,
      "step": 219268
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.5615330934524536,
      "learning_rate": 3.4505068830931515e-06,
      "loss": 1.9898,
      "step": 219269
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.466811180114746,
      "learning_rate": 3.4498882871333287e-06,
      "loss": 1.9021,
      "step": 219270
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.42011296749115,
      "learning_rate": 3.44926974630777e-06,
      "loss": 1.8782,
      "step": 219271
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.4071604013442993,
      "learning_rate": 3.4486512606166416e-06,
      "loss": 1.9849,
      "step": 219272
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.3782745599746704,
      "learning_rate": 3.448032830059977e-06,
      "loss": 2.0257,
      "step": 219273
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.5237343311309814,
      "learning_rate": 3.447414454637876e-06,
      "loss": 1.8838,
      "step": 219274
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.4927912950515747,
      "learning_rate": 3.446796134350571e-06,
      "loss": 2.1754,
      "step": 219275
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.344668984413147,
      "learning_rate": 3.44617786919803e-06,
      "loss": 1.9364,
      "step": 219276
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.531944990158081,
      "learning_rate": 3.4455596591805192e-06,
      "loss": 1.9411,
      "step": 219277
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.3707560300827026,
      "learning_rate": 3.444941504298038e-06,
      "loss": 1.9636,
      "step": 219278
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.597007393836975,
      "learning_rate": 3.4443234045507527e-06,
      "loss": 1.9011,
      "step": 219279
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.4086166620254517,
      "learning_rate": 3.443705359938731e-06,
      "loss": 1.7755,
      "step": 219280
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.5381461381912231,
      "learning_rate": 3.443087370462172e-06,
      "loss": 2.0863,
      "step": 219281
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.5292913913726807,
      "learning_rate": 3.442469436121076e-06,
      "loss": 1.9114,
      "step": 219282
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.452858567237854,
      "learning_rate": 3.4418515569156757e-06,
      "loss": 1.8115,
      "step": 219283
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.4256761074066162,
      "learning_rate": 3.4412337328460383e-06,
      "loss": 2.0145,
      "step": 219284
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.5082136392593384,
      "learning_rate": 3.440615963912263e-06,
      "loss": 2.0012,
      "step": 219285
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.4412251710891724,
      "learning_rate": 3.4399982501144506e-06,
      "loss": 1.9033,
      "step": 219286
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.5695475339889526,
      "learning_rate": 3.439380591452767e-06,
      "loss": 1.8399,
      "step": 219287
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.5542234182357788,
      "learning_rate": 3.4387629879272793e-06,
      "loss": 1.9677,
      "step": 219288
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.347244381904602,
      "learning_rate": 3.4381454395381534e-06,
      "loss": 1.9074,
      "step": 219289
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.3883787393569946,
      "learning_rate": 3.4375279462854565e-06,
      "loss": 2.1106,
      "step": 219290
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.446069598197937,
      "learning_rate": 3.4369105081693215e-06,
      "loss": 2.2167,
      "step": 219291
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.4961988925933838,
      "learning_rate": 3.4362931251898484e-06,
      "loss": 2.1675,
      "step": 219292
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.4677560329437256,
      "learning_rate": 3.4356757973472036e-06,
      "loss": 2.0518,
      "step": 219293
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.5664411783218384,
      "learning_rate": 3.435058524641421e-06,
      "loss": 2.1882,
      "step": 219294
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.4490727186203003,
      "learning_rate": 3.4344413070726662e-06,
      "loss": 1.9943,
      "step": 219295
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.437551498413086,
      "learning_rate": 3.433824144641073e-06,
      "loss": 2.0097,
      "step": 219296
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.4627214670181274,
      "learning_rate": 3.4332070373467414e-06,
      "loss": 1.9557,
      "step": 219297
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.3707201480865479,
      "learning_rate": 3.4325899851897375e-06,
      "loss": 1.8798,
      "step": 219298
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.4095187187194824,
      "learning_rate": 3.4319729881702284e-06,
      "loss": 1.8701,
      "step": 219299
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.3836119174957275,
      "learning_rate": 3.431356046288314e-06,
      "loss": 1.9385,
      "step": 219300
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.5888581275939941,
      "learning_rate": 3.4307391595440936e-06,
      "loss": 1.979,
      "step": 219301
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.5721603631973267,
      "learning_rate": 3.4301223279377343e-06,
      "loss": 2.0974,
      "step": 219302
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.4622652530670166,
      "learning_rate": 3.4295055514692692e-06,
      "loss": 2.0877,
      "step": 219303
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.643851637840271,
      "learning_rate": 3.428888830138865e-06,
      "loss": 2.0,
      "step": 219304
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.4710402488708496,
      "learning_rate": 3.428272163946655e-06,
      "loss": 1.9546,
      "step": 219305
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.5346025228500366,
      "learning_rate": 3.4276555528927053e-06,
      "loss": 2.1346,
      "step": 219306
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.4640767574310303,
      "learning_rate": 3.427038996977183e-06,
      "loss": 1.9936,
      "step": 219307
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.496277093887329,
      "learning_rate": 3.426422496200154e-06,
      "loss": 2.0999,
      "step": 219308
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.5036122798919678,
      "learning_rate": 3.4258060505617523e-06,
      "loss": 2.0554,
      "step": 219309
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.4060556888580322,
      "learning_rate": 3.425189660062078e-06,
      "loss": 2.2932,
      "step": 219310
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.4836909770965576,
      "learning_rate": 3.424573324701263e-06,
      "loss": 2.0283,
      "step": 219311
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.4810072183609009,
      "learning_rate": 3.4239570444794086e-06,
      "loss": 1.9265,
      "step": 219312
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.5242761373519897,
      "learning_rate": 3.423340819396647e-06,
      "loss": 1.9014,
      "step": 219313
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.5444234609603882,
      "learning_rate": 3.4227246494531125e-06,
      "loss": 2.0883,
      "step": 219314
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.5864356756210327,
      "learning_rate": 3.422108534648871e-06,
      "loss": 2.1265,
      "step": 219315
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.5668771266937256,
      "learning_rate": 3.4214924749840556e-06,
      "loss": 2.0106,
      "step": 219316
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.4043608903884888,
      "learning_rate": 3.4208764704587998e-06,
      "loss": 1.8789,
      "step": 219317
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.5826822519302368,
      "learning_rate": 3.4202605210731705e-06,
      "loss": 2.0963,
      "step": 219318
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.4314310550689697,
      "learning_rate": 3.4196446268273336e-06,
      "loss": 1.947,
      "step": 219319
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.4348562955856323,
      "learning_rate": 3.4190287877213897e-06,
      "loss": 2.0978,
      "step": 219320
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.4366265535354614,
      "learning_rate": 3.4184130037554047e-06,
      "loss": 1.7714,
      "step": 219321
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.5690261125564575,
      "learning_rate": 3.417797274929579e-06,
      "loss": 1.9565,
      "step": 219322
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.5672528743743896,
      "learning_rate": 3.417181601243979e-06,
      "loss": 2.1377,
      "step": 219323
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.3015050888061523,
      "learning_rate": 3.4165659826987044e-06,
      "loss": 1.6967,
      "step": 219324
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.5819330215454102,
      "learning_rate": 3.4159504192938893e-06,
      "loss": 2.113,
      "step": 219325
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.5994597673416138,
      "learning_rate": 3.415334911029666e-06,
      "loss": 2.0955,
      "step": 219326
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.635899543762207,
      "learning_rate": 3.4147194579061345e-06,
      "loss": 1.9547,
      "step": 219327
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.4943090677261353,
      "learning_rate": 3.4141040599233947e-06,
      "loss": 2.0852,
      "step": 219328
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.4117629528045654,
      "learning_rate": 3.4134887170815808e-06,
      "loss": 1.7785,
      "step": 219329
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.4312084913253784,
      "learning_rate": 3.412873429380758e-06,
      "loss": 2.0911,
      "step": 219330
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.4165433645248413,
      "learning_rate": 3.412258196821127e-06,
      "loss": 2.0017,
      "step": 219331
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.4931696653366089,
      "learning_rate": 3.411643019402721e-06,
      "loss": 2.2211,
      "step": 219332
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.4039894342422485,
      "learning_rate": 3.4110278971257064e-06,
      "loss": 2.0182,
      "step": 219333
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.5938751697540283,
      "learning_rate": 3.4104128299901835e-06,
      "loss": 1.9472,
      "step": 219334
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.4767100811004639,
      "learning_rate": 3.4097978179962513e-06,
      "loss": 2.1103,
      "step": 219335
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.57442307472229,
      "learning_rate": 3.409182861144044e-06,
      "loss": 1.7715,
      "step": 219336
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.5374614000320435,
      "learning_rate": 3.408567959433661e-06,
      "loss": 2.068,
      "step": 219337
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.5647777318954468,
      "learning_rate": 3.407953112865236e-06,
      "loss": 1.8965,
      "step": 219338
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.5629703998565674,
      "learning_rate": 3.4073383214388685e-06,
      "loss": 1.9759,
      "step": 219339
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.5145536661148071,
      "learning_rate": 3.4067235851546913e-06,
      "loss": 1.9938,
      "step": 219340
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.449921727180481,
      "learning_rate": 3.406108904012772e-06,
      "loss": 2.0637,
      "step": 219341
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.4100825786590576,
      "learning_rate": 3.4054942780132764e-06,
      "loss": 1.982,
      "step": 219342
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.370648741722107,
      "learning_rate": 3.4048797071562716e-06,
      "loss": 1.987,
      "step": 219343
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.5267189741134644,
      "learning_rate": 3.404265191441924e-06,
      "loss": 2.0689,
      "step": 219344
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.5965526103973389,
      "learning_rate": 3.4036507308703e-06,
      "loss": 1.9676,
      "step": 219345
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.571352481842041,
      "learning_rate": 3.403036325441566e-06,
      "loss": 1.9583,
      "step": 219346
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.448274850845337,
      "learning_rate": 3.4024219751557893e-06,
      "loss": 1.9778,
      "step": 219347
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.320539116859436,
      "learning_rate": 3.4018076800131024e-06,
      "loss": 1.8674,
      "step": 219348
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.512656331062317,
      "learning_rate": 3.4011934400136055e-06,
      "loss": 1.8219,
      "step": 219349
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.3632100820541382,
      "learning_rate": 3.400579255157465e-06,
      "loss": 1.9065,
      "step": 219350
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.4659000635147095,
      "learning_rate": 3.399965125444715e-06,
      "loss": 1.9426,
      "step": 219351
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.4044723510742188,
      "learning_rate": 3.3993510508755205e-06,
      "loss": 1.9324,
      "step": 219352
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.4859577417373657,
      "learning_rate": 3.398737031449983e-06,
      "loss": 1.9404,
      "step": 219353
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.4796031713485718,
      "learning_rate": 3.3981230671682348e-06,
      "loss": 2.0479,
      "step": 219354
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.4838969707489014,
      "learning_rate": 3.3975091580303425e-06,
      "loss": 1.9042,
      "step": 219355
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.4745510816574097,
      "learning_rate": 3.3968953040365066e-06,
      "loss": 2.0157,
      "step": 219356
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.9863879680633545,
      "learning_rate": 3.39628150518676e-06,
      "loss": 2.1668,
      "step": 219357
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.4380629062652588,
      "learning_rate": 3.3956677614812356e-06,
      "loss": 1.9019,
      "step": 219358
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.5381938219070435,
      "learning_rate": 3.395054072920034e-06,
      "loss": 2.1614,
      "step": 219359
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.4065364599227905,
      "learning_rate": 3.394440439503321e-06,
      "loss": 2.0003,
      "step": 219360
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.4780733585357666,
      "learning_rate": 3.3938268612311637e-06,
      "loss": 1.9005,
      "step": 219361
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.3905372619628906,
      "learning_rate": 3.3932133381037284e-06,
      "loss": 1.9539,
      "step": 219362
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.6529338359832764,
      "learning_rate": 3.392599870121082e-06,
      "loss": 1.9934,
      "step": 219363
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.3868299722671509,
      "learning_rate": 3.3919864572833242e-06,
      "loss": 1.9253,
      "step": 219364
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.443166971206665,
      "learning_rate": 3.3913730995905884e-06,
      "loss": 1.9359,
      "step": 219365
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.5237386226654053,
      "learning_rate": 3.390759797043041e-06,
      "loss": 2.0613,
      "step": 219366
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.578447699546814,
      "learning_rate": 3.3901465496407153e-06,
      "loss": 2.1647,
      "step": 219367
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.4787945747375488,
      "learning_rate": 3.3895333573837446e-06,
      "loss": 1.9204,
      "step": 219368
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.650417447090149,
      "learning_rate": 3.388920220272295e-06,
      "loss": 2.1027,
      "step": 219369
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.444695234298706,
      "learning_rate": 3.388307138306434e-06,
      "loss": 2.1113,
      "step": 219370
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.5461362600326538,
      "learning_rate": 3.3876941114862944e-06,
      "loss": 2.0411,
      "step": 219371
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.4513442516326904,
      "learning_rate": 3.3870811398119755e-06,
      "loss": 1.9116,
      "step": 219372
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.4794403314590454,
      "learning_rate": 3.3864682232835783e-06,
      "loss": 1.9548,
      "step": 219373
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.478546380996704,
      "learning_rate": 3.3858553619012352e-06,
      "loss": 1.8551,
      "step": 219374
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.4140154123306274,
      "learning_rate": 3.3852425556651132e-06,
      "loss": 1.9005,
      "step": 219375
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.410367488861084,
      "learning_rate": 3.384629804575212e-06,
      "loss": 2.0577,
      "step": 219376
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.5050042867660522,
      "learning_rate": 3.3840171086317315e-06,
      "loss": 2.0063,
      "step": 219377
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.5731873512268066,
      "learning_rate": 3.3834044678347716e-06,
      "loss": 2.0741,
      "step": 219378
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.389699935913086,
      "learning_rate": 3.382791882184399e-06,
      "loss": 1.9173,
      "step": 219379
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.3602975606918335,
      "learning_rate": 3.382179351680814e-06,
      "loss": 1.9493,
      "step": 219380
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.4995704889297485,
      "learning_rate": 3.381566876324049e-06,
      "loss": 2.1042,
      "step": 219381
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.4836450815200806,
      "learning_rate": 3.380954456114271e-06,
      "loss": 1.8899,
      "step": 219382
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.5144896507263184,
      "learning_rate": 3.3803420910515466e-06,
      "loss": 1.9686,
      "step": 219383
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.5524473190307617,
      "learning_rate": 3.3797297811360423e-06,
      "loss": 1.9497,
      "step": 219384
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.371724009513855,
      "learning_rate": 3.379117526367825e-06,
      "loss": 2.2115,
      "step": 219385
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.4875197410583496,
      "learning_rate": 3.3785053267470276e-06,
      "loss": 2.0632,
      "step": 219386
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.4876248836517334,
      "learning_rate": 3.377893182273783e-06,
      "loss": 2.0716,
      "step": 219387
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.49868905544281,
      "learning_rate": 3.377281092948192e-06,
      "loss": 1.9967,
      "step": 219388
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.4009126424789429,
      "learning_rate": 3.3766690587703206e-06,
      "loss": 1.8827,
      "step": 219389
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.4519717693328857,
      "learning_rate": 3.3760570797403686e-06,
      "loss": 1.804,
      "step": 219390
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.5751441717147827,
      "learning_rate": 3.3754451558583695e-06,
      "loss": 2.0082,
      "step": 219391
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.553939700126648,
      "learning_rate": 3.374833287124523e-06,
      "loss": 1.9881,
      "step": 219392
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.4716908931732178,
      "learning_rate": 3.374221473538863e-06,
      "loss": 1.9692,
      "step": 219393
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.5057320594787598,
      "learning_rate": 3.373609715101555e-06,
      "loss": 2.0334,
      "step": 219394
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.627942681312561,
      "learning_rate": 3.3729980118126665e-06,
      "loss": 2.1925,
      "step": 219395
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.6824289560317993,
      "learning_rate": 3.37238636367233e-06,
      "loss": 2.0954,
      "step": 219396
    },
    {
      "epoch": 2.86,
      "grad_norm": 2.177048683166504,
      "learning_rate": 3.3717747706806797e-06,
      "loss": 1.9548,
      "step": 219397
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.505786418914795,
      "learning_rate": 3.3711632328378478e-06,
      "loss": 2.0422,
      "step": 219398
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.4358140230178833,
      "learning_rate": 3.370551750143868e-06,
      "loss": 1.9366,
      "step": 219399
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.5582939386367798,
      "learning_rate": 3.369940322598941e-06,
      "loss": 2.0948,
      "step": 219400
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.5877436399459839,
      "learning_rate": 3.3693289502030984e-06,
      "loss": 2.1011,
      "step": 219401
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.5446981191635132,
      "learning_rate": 3.3687176329565412e-06,
      "loss": 2.1011,
      "step": 219402
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.4365365505218506,
      "learning_rate": 3.3681063708593027e-06,
      "loss": 1.9899,
      "step": 219403
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.426443338394165,
      "learning_rate": 3.3674951639115488e-06,
      "loss": 2.0028,
      "step": 219404
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.4847615957260132,
      "learning_rate": 3.36688401211338e-06,
      "loss": 1.9419,
      "step": 219405
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.4381335973739624,
      "learning_rate": 3.366272915464896e-06,
      "loss": 1.9796,
      "step": 219406
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.4523687362670898,
      "learning_rate": 3.36566187396623e-06,
      "loss": 1.8198,
      "step": 219407
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.4587604999542236,
      "learning_rate": 3.3650508876175154e-06,
      "loss": 1.7979,
      "step": 219408
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.538314938545227,
      "learning_rate": 3.364439956418785e-06,
      "loss": 2.0727,
      "step": 219409
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.4236854314804077,
      "learning_rate": 3.3638290803702393e-06,
      "loss": 2.1462,
      "step": 219410
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.3796988725662231,
      "learning_rate": 3.3632182594719114e-06,
      "loss": 1.831,
      "step": 219411
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.6335384845733643,
      "learning_rate": 3.362607493724001e-06,
      "loss": 1.8932,
      "step": 219412
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.4253839254379272,
      "learning_rate": 3.3619967831265747e-06,
      "loss": 2.0487,
      "step": 219413
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.6277196407318115,
      "learning_rate": 3.361386127679766e-06,
      "loss": 1.9132,
      "step": 219414
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.5183436870574951,
      "learning_rate": 3.360775527383641e-06,
      "loss": 1.8534,
      "step": 219415
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.5375421047210693,
      "learning_rate": 3.3601649822384002e-06,
      "loss": 2.0381,
      "step": 219416
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.3854374885559082,
      "learning_rate": 3.359554492244043e-06,
      "loss": 2.1596,
      "step": 219417
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.5442904233932495,
      "learning_rate": 3.3589440574007696e-06,
      "loss": 2.0513,
      "step": 219418
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.473368763923645,
      "learning_rate": 3.35833367770868e-06,
      "loss": 1.9392,
      "step": 219419
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.3705369234085083,
      "learning_rate": 3.357723353167874e-06,
      "loss": 1.8984,
      "step": 219420
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.426142930984497,
      "learning_rate": 3.357113083778451e-06,
      "loss": 1.9059,
      "step": 219421
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.574526071548462,
      "learning_rate": 3.3565028695405782e-06,
      "loss": 2.0975,
      "step": 219422
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.4269973039627075,
      "learning_rate": 3.3558927104542887e-06,
      "loss": 2.0546,
      "step": 219423
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.5387433767318726,
      "learning_rate": 3.3552826065197493e-06,
      "loss": 1.8972,
      "step": 219424
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.5020604133605957,
      "learning_rate": 3.3546725577370594e-06,
      "loss": 1.9497,
      "step": 219425
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.575963020324707,
      "learning_rate": 3.354062564106319e-06,
      "loss": 2.04,
      "step": 219426
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.4998893737792969,
      "learning_rate": 3.353452625627695e-06,
      "loss": 1.8371,
      "step": 219427
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.3571423292160034,
      "learning_rate": 3.3528427423012538e-06,
      "loss": 1.9652,
      "step": 219428
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.3826428651809692,
      "learning_rate": 3.352232914127095e-06,
      "loss": 2.032,
      "step": 219429
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.4160691499710083,
      "learning_rate": 3.3516231411053864e-06,
      "loss": 2.0751,
      "step": 219430
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.3861724138259888,
      "learning_rate": 3.3510134232362263e-06,
      "loss": 2.0384,
      "step": 219431
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.5293105840682983,
      "learning_rate": 3.3504037605196486e-06,
      "loss": 1.9995,
      "step": 219432
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.507323980331421,
      "learning_rate": 3.349794152955887e-06,
      "loss": 2.0041,
      "step": 219433
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.5572481155395508,
      "learning_rate": 3.349184600544974e-06,
      "loss": 1.9132,
      "step": 219434
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.3793851137161255,
      "learning_rate": 3.3485751032870434e-06,
      "loss": 1.8328,
      "step": 219435
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.4593636989593506,
      "learning_rate": 3.347965661182228e-06,
      "loss": 2.0975,
      "step": 219436
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.5298539400100708,
      "learning_rate": 3.347356274230628e-06,
      "loss": 2.0195,
      "step": 219437
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.5698200464248657,
      "learning_rate": 3.3467469424323433e-06,
      "loss": 1.9512,
      "step": 219438
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.4436875581741333,
      "learning_rate": 3.346137665787507e-06,
      "loss": 1.9044,
      "step": 219439
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.4290332794189453,
      "learning_rate": 3.345528444296186e-06,
      "loss": 1.8767,
      "step": 219440
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.3773491382598877,
      "learning_rate": 3.34491927795858e-06,
      "loss": 1.9016,
      "step": 219441
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.6939343214035034,
      "learning_rate": 3.344310166774722e-06,
      "loss": 2.3438,
      "step": 219442
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.4594603776931763,
      "learning_rate": 3.3437011107447786e-06,
      "loss": 1.7689,
      "step": 219443
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.5770570039749146,
      "learning_rate": 3.343092109868817e-06,
      "loss": 1.856,
      "step": 219444
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.4065730571746826,
      "learning_rate": 3.342483164147003e-06,
      "loss": 1.947,
      "step": 219445
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.5388702154159546,
      "learning_rate": 3.3418742735794035e-06,
      "loss": 2.0662,
      "step": 219446
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.4713172912597656,
      "learning_rate": 3.3412654381661517e-06,
      "loss": 2.0038,
      "step": 219447
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.3164269924163818,
      "learning_rate": 3.340656657907348e-06,
      "loss": 1.6394,
      "step": 219448
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.4904710054397583,
      "learning_rate": 3.3400479328031584e-06,
      "loss": 1.8702,
      "step": 219449
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.5575692653656006,
      "learning_rate": 3.339439262853616e-06,
      "loss": 2.0488,
      "step": 219450
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.5200774669647217,
      "learning_rate": 3.3388306480588543e-06,
      "loss": 2.0672,
      "step": 219451
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.4046950340270996,
      "learning_rate": 3.33822208841904e-06,
      "loss": 1.7638,
      "step": 219452
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.3682881593704224,
      "learning_rate": 3.33761358393424e-06,
      "loss": 2.027,
      "step": 219453
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.366147756576538,
      "learning_rate": 3.337005134604587e-06,
      "loss": 2.0395,
      "step": 219454
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.4721804857254028,
      "learning_rate": 3.3363967404301805e-06,
      "loss": 1.9602,
      "step": 219455
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.5349041223526,
      "learning_rate": 3.3357884014111215e-06,
      "loss": 2.0689,
      "step": 219456
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.4918087720870972,
      "learning_rate": 3.3351801175475425e-06,
      "loss": 2.093,
      "step": 219457
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.4706673622131348,
      "learning_rate": 3.3345718888395434e-06,
      "loss": 1.7905,
      "step": 219458
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.5899507999420166,
      "learning_rate": 3.3339637152872576e-06,
      "loss": 1.8837,
      "step": 219459
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.5056607723236084,
      "learning_rate": 3.3333555968908186e-06,
      "loss": 2.047,
      "step": 219460
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.3609594106674194,
      "learning_rate": 3.332747533650293e-06,
      "loss": 1.813,
      "step": 219461
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.4745750427246094,
      "learning_rate": 3.3321395255657803e-06,
      "loss": 2.2015,
      "step": 219462
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.3668642044067383,
      "learning_rate": 3.331531572637447e-06,
      "loss": 1.9573,
      "step": 219463
    },
    {
      "epoch": 2.86,
      "grad_norm": 2.504214286804199,
      "learning_rate": 3.33092367486536e-06,
      "loss": 2.0056,
      "step": 219464
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.5272579193115234,
      "learning_rate": 3.3303158322496858e-06,
      "loss": 1.9203,
      "step": 219465
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.5779900550842285,
      "learning_rate": 3.329708044790491e-06,
      "loss": 1.8197,
      "step": 219466
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.7384247779846191,
      "learning_rate": 3.329100312487942e-06,
      "loss": 1.8666,
      "step": 219467
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.389744758605957,
      "learning_rate": 3.328492635342039e-06,
      "loss": 1.9229,
      "step": 219468
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.4243558645248413,
      "learning_rate": 3.3278850133530155e-06,
      "loss": 1.7317,
      "step": 219469
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.3806990385055542,
      "learning_rate": 3.327277446520937e-06,
      "loss": 2.1846,
      "step": 219470
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.3905450105667114,
      "learning_rate": 3.3266699348459045e-06,
      "loss": 1.7729,
      "step": 219471
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.324177622795105,
      "learning_rate": 3.326062478328051e-06,
      "loss": 1.9147,
      "step": 219472
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.5066436529159546,
      "learning_rate": 3.3254550769675093e-06,
      "loss": 1.9563,
      "step": 219473
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.4738843441009521,
      "learning_rate": 3.324847730764313e-06,
      "loss": 2.1587,
      "step": 219474
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.6073914766311646,
      "learning_rate": 3.324240439718662e-06,
      "loss": 2.0193,
      "step": 219475
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.6389623880386353,
      "learning_rate": 3.323633203830622e-06,
      "loss": 1.9427,
      "step": 219476
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.4634182453155518,
      "learning_rate": 3.323026023100328e-06,
      "loss": 1.9461,
      "step": 219477
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.433858871459961,
      "learning_rate": 3.3224188975278455e-06,
      "loss": 2.1602,
      "step": 219478
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.482102632522583,
      "learning_rate": 3.3218118271134075e-06,
      "loss": 1.8535,
      "step": 219479
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.559429407119751,
      "learning_rate": 3.321204811856981e-06,
      "loss": 1.9633,
      "step": 219480
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.4000630378723145,
      "learning_rate": 3.320597851758766e-06,
      "loss": 1.8523,
      "step": 219481
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.472971796989441,
      "learning_rate": 3.3199909468188292e-06,
      "loss": 1.8497,
      "step": 219482
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.487671971321106,
      "learning_rate": 3.319384097037303e-06,
      "loss": 2.0395,
      "step": 219483
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.3165000677108765,
      "learning_rate": 3.3187773024143214e-06,
      "loss": 1.81,
      "step": 219484
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.4313852787017822,
      "learning_rate": 3.3181705629500177e-06,
      "loss": 2.0877,
      "step": 219485
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.3985216617584229,
      "learning_rate": 3.3175638786443916e-06,
      "loss": 1.752,
      "step": 219486
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.3934153318405151,
      "learning_rate": 3.316957249497676e-06,
      "loss": 1.8875,
      "step": 219487
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.5280580520629883,
      "learning_rate": 3.316350675509938e-06,
      "loss": 2.0309,
      "step": 219488
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.4531292915344238,
      "learning_rate": 3.315744156681277e-06,
      "loss": 1.9756,
      "step": 219489
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.5079247951507568,
      "learning_rate": 3.3151376930118268e-06,
      "loss": 1.8695,
      "step": 219490
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.3892909288406372,
      "learning_rate": 3.3145312845017202e-06,
      "loss": 2.1646,
      "step": 219491
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.3485513925552368,
      "learning_rate": 3.313924931150991e-06,
      "loss": 1.9258,
      "step": 219492
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.451424479484558,
      "learning_rate": 3.313318632959838e-06,
      "loss": 2.0911,
      "step": 219493
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.5855159759521484,
      "learning_rate": 3.312712389928296e-06,
      "loss": 1.7969,
      "step": 219494
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.4802297353744507,
      "learning_rate": 3.3121062020565634e-06,
      "loss": 2.0291,
      "step": 219495
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.4753739833831787,
      "learning_rate": 3.3115000693447082e-06,
      "loss": 2.1351,
      "step": 219496
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.394331932067871,
      "learning_rate": 3.310893991792829e-06,
      "loss": 2.0786,
      "step": 219497
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.4413052797317505,
      "learning_rate": 3.3102879694010597e-06,
      "loss": 1.9755,
      "step": 219498
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.530475378036499,
      "learning_rate": 3.3096820021695336e-06,
      "loss": 1.9269,
      "step": 219499
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.482053518295288,
      "learning_rate": 3.309076090098284e-06,
      "loss": 2.0478,
      "step": 219500
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.3763374090194702,
      "learning_rate": 3.3084702331875433e-06,
      "loss": 1.8857,
      "step": 219501
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.6520174741744995,
      "learning_rate": 3.3078644314373125e-06,
      "loss": 2.0394,
      "step": 219502
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.519454002380371,
      "learning_rate": 3.3072586848477577e-06,
      "loss": 1.8612,
      "step": 219503
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.4815868139266968,
      "learning_rate": 3.306652993418979e-06,
      "loss": 1.8933,
      "step": 219504
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.4663574695587158,
      "learning_rate": 3.3060473571511093e-06,
      "loss": 1.9031,
      "step": 219505
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.4457389116287231,
      "learning_rate": 3.3054417760442487e-06,
      "loss": 2.0697,
      "step": 219506
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.5770862102508545,
      "learning_rate": 3.304836250098497e-06,
      "loss": 1.9295,
      "step": 219507
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.4502238035202026,
      "learning_rate": 3.304230779313988e-06,
      "loss": 1.7599,
      "step": 219508
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.589547872543335,
      "learning_rate": 3.3036253636908204e-06,
      "loss": 1.7809,
      "step": 219509
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.5509097576141357,
      "learning_rate": 3.3030200032290955e-06,
      "loss": 1.7462,
      "step": 219510
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.4673794507980347,
      "learning_rate": 3.3024146979289455e-06,
      "loss": 1.638,
      "step": 219511
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.4314106702804565,
      "learning_rate": 3.301809447790471e-06,
      "loss": 1.722,
      "step": 219512
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.3986587524414062,
      "learning_rate": 3.301204252813805e-06,
      "loss": 1.8526,
      "step": 219513
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.4486198425292969,
      "learning_rate": 3.300599112999014e-06,
      "loss": 2.0286,
      "step": 219514
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.5316280126571655,
      "learning_rate": 3.2999940283462977e-06,
      "loss": 2.1031,
      "step": 219515
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.3392794132232666,
      "learning_rate": 3.29938899885569e-06,
      "loss": 2.053,
      "step": 219516
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.5045746564865112,
      "learning_rate": 3.29878402452729e-06,
      "loss": 1.9544,
      "step": 219517
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.6084132194519043,
      "learning_rate": 3.2981791053612985e-06,
      "loss": 2.0861,
      "step": 219518
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.4488790035247803,
      "learning_rate": 3.297574241357748e-06,
      "loss": 1.9307,
      "step": 219519
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.454966425895691,
      "learning_rate": 3.296969432516772e-06,
      "loss": 2.1651,
      "step": 219520
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.5214895009994507,
      "learning_rate": 3.296364678838503e-06,
      "loss": 2.1751,
      "step": 219521
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.4640501737594604,
      "learning_rate": 3.295759980323076e-06,
      "loss": 1.8086,
      "step": 219522
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.5601526498794556,
      "learning_rate": 3.295155336970523e-06,
      "loss": 1.9417,
      "step": 219523
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.5297648906707764,
      "learning_rate": 3.294550748781011e-06,
      "loss": 1.8845,
      "step": 219524
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.367246150970459,
      "learning_rate": 3.293946215754606e-06,
      "loss": 2.0601,
      "step": 219525
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.4916061162948608,
      "learning_rate": 3.2933417378915085e-06,
      "loss": 2.0633,
      "step": 219526
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.5284385681152344,
      "learning_rate": 3.2927373151917515e-06,
      "loss": 2.1681,
      "step": 219527
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.4596368074417114,
      "learning_rate": 3.292132947655535e-06,
      "loss": 1.9762,
      "step": 219528
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.4478670358657837,
      "learning_rate": 3.2915286352828586e-06,
      "loss": 1.6576,
      "step": 219529
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.398988127708435,
      "learning_rate": 3.290924378073889e-06,
      "loss": 1.9275,
      "step": 219530
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.3766119480133057,
      "learning_rate": 3.2903201760287267e-06,
      "loss": 2.0241,
      "step": 219531
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.5627192258834839,
      "learning_rate": 3.2897160291475045e-06,
      "loss": 2.1205,
      "step": 219532
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.4424123764038086,
      "learning_rate": 3.289111937430322e-06,
      "loss": 2.0392,
      "step": 219533
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.4277621507644653,
      "learning_rate": 3.2885079008773464e-06,
      "loss": 1.8812,
      "step": 219534
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.4597200155258179,
      "learning_rate": 3.287903919488577e-06,
      "loss": 2.156,
      "step": 219535
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.3864375352859497,
      "learning_rate": 3.2872999932642143e-06,
      "loss": 1.6393,
      "step": 219536
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.4964450597763062,
      "learning_rate": 3.286696122204324e-06,
      "loss": 1.8978,
      "step": 219537
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.426705002784729,
      "learning_rate": 3.2860923063090404e-06,
      "loss": 1.9468,
      "step": 219538
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.550240397453308,
      "learning_rate": 3.2854885455784964e-06,
      "loss": 2.3174,
      "step": 219539
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.5091651678085327,
      "learning_rate": 3.284884840012791e-06,
      "loss": 1.904,
      "step": 219540
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.5279872417449951,
      "learning_rate": 3.2842811896119926e-06,
      "loss": 2.1141,
      "step": 219541
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.6248018741607666,
      "learning_rate": 3.283677594376233e-06,
      "loss": 1.9411,
      "step": 219542
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.4092756509780884,
      "learning_rate": 3.283074054305679e-06,
      "loss": 1.9366,
      "step": 219543
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.5323717594146729,
      "learning_rate": 3.282470569400364e-06,
      "loss": 1.9669,
      "step": 219544
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.48237943649292,
      "learning_rate": 3.2818671396604547e-06,
      "loss": 2.0227,
      "step": 219545
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.5313122272491455,
      "learning_rate": 3.281263765086084e-06,
      "loss": 1.8838,
      "step": 219546
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.4551066160202026,
      "learning_rate": 3.2806604456772856e-06,
      "loss": 1.776,
      "step": 219547
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.517156720161438,
      "learning_rate": 3.2800571814342257e-06,
      "loss": 1.9813,
      "step": 219548
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.5842413902282715,
      "learning_rate": 3.2794539723570045e-06,
      "loss": 2.0695,
      "step": 219549
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.4911705255508423,
      "learning_rate": 3.278850818445722e-06,
      "loss": 2.3325,
      "step": 219550
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.555374026298523,
      "learning_rate": 3.278247719700511e-06,
      "loss": 1.8769,
      "step": 219551
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.460747480392456,
      "learning_rate": 3.277644676121505e-06,
      "loss": 1.922,
      "step": 219552
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.5758121013641357,
      "learning_rate": 3.277041687708737e-06,
      "loss": 1.9119,
      "step": 219553
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.4389649629592896,
      "learning_rate": 3.276438754462407e-06,
      "loss": 1.9002,
      "step": 219554
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.338841199874878,
      "learning_rate": 3.2758358763825486e-06,
      "loss": 1.8579,
      "step": 219555
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.4976364374160767,
      "learning_rate": 3.275233053469328e-06,
      "loss": 1.9671,
      "step": 219556
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.393344521522522,
      "learning_rate": 3.2746302857228457e-06,
      "loss": 1.9681,
      "step": 219557
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.5336108207702637,
      "learning_rate": 3.274027573143234e-06,
      "loss": 2.0607,
      "step": 219558
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.6934027671813965,
      "learning_rate": 3.27342491573056e-06,
      "loss": 2.2157,
      "step": 219559
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.464226484298706,
      "learning_rate": 3.2728223134849575e-06,
      "loss": 1.9045,
      "step": 219560
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.5502697229385376,
      "learning_rate": 3.272219766406525e-06,
      "loss": 1.8263,
      "step": 219561
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.4236688613891602,
      "learning_rate": 3.2716172744953973e-06,
      "loss": 1.9723,
      "step": 219562
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.4993817806243896,
      "learning_rate": 3.271014837751673e-06,
      "loss": 1.8785,
      "step": 219563
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.50437593460083,
      "learning_rate": 3.2704124561755196e-06,
      "loss": 1.8609,
      "step": 219564
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.4346429109573364,
      "learning_rate": 3.2698101297669365e-06,
      "loss": 2.0941,
      "step": 219565
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.6003291606903076,
      "learning_rate": 3.269207858526124e-06,
      "loss": 2.1195,
      "step": 219566
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.4279816150665283,
      "learning_rate": 3.268605642453148e-06,
      "loss": 2.1169,
      "step": 219567
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.5084739923477173,
      "learning_rate": 3.2680034815481427e-06,
      "loss": 2.0538,
      "step": 219568
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.540951132774353,
      "learning_rate": 3.2674013758112406e-06,
      "loss": 2.0466,
      "step": 219569
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.6443108320236206,
      "learning_rate": 3.2667993252425417e-06,
      "loss": 2.0357,
      "step": 219570
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.4979557991027832,
      "learning_rate": 3.2661973298420796e-06,
      "loss": 1.8954,
      "step": 219571
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.387346625328064,
      "learning_rate": 3.265595389610087e-06,
      "loss": 1.9271,
      "step": 219572
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.5632439851760864,
      "learning_rate": 3.2649935045465982e-06,
      "loss": 1.9494,
      "step": 219573
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.5118484497070312,
      "learning_rate": 3.2643916746517784e-06,
      "loss": 2.0935,
      "step": 219574
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.4775899648666382,
      "learning_rate": 3.2637898999256616e-06,
      "loss": 1.9457,
      "step": 219575
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.3802924156188965,
      "learning_rate": 3.2631881803684477e-06,
      "loss": 1.9638,
      "step": 219576
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.6018222570419312,
      "learning_rate": 3.2625865159802033e-06,
      "loss": 2.1,
      "step": 219577
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.4188313484191895,
      "learning_rate": 3.261984906761028e-06,
      "loss": 2.1574,
      "step": 219578
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.4152183532714844,
      "learning_rate": 3.2613833527110553e-06,
      "loss": 1.8408,
      "step": 219579
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.4574484825134277,
      "learning_rate": 3.2607818538303853e-06,
      "loss": 2.0042,
      "step": 219580
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.438628911972046,
      "learning_rate": 3.260180410119151e-06,
      "loss": 2.0453,
      "step": 219581
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.6610603332519531,
      "learning_rate": 3.259579021577452e-06,
      "loss": 1.9859,
      "step": 219582
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.643977403640747,
      "learning_rate": 3.2589776882053886e-06,
      "loss": 2.0298,
      "step": 219583
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.5558314323425293,
      "learning_rate": 3.258376410003094e-06,
      "loss": 1.9527,
      "step": 219584
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.5422563552856445,
      "learning_rate": 3.257775186970668e-06,
      "loss": 1.8862,
      "step": 219585
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.5629597902297974,
      "learning_rate": 3.257174019108211e-06,
      "loss": 1.9815,
      "step": 219586
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.4768428802490234,
      "learning_rate": 3.256572906415855e-06,
      "loss": 1.787,
      "step": 219587
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.4398038387298584,
      "learning_rate": 3.2559718488937013e-06,
      "loss": 1.9421,
      "step": 219588
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.6255985498428345,
      "learning_rate": 3.2553708465418827e-06,
      "loss": 2.0424,
      "step": 219589
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.3809645175933838,
      "learning_rate": 3.2547698993604654e-06,
      "loss": 2.1243,
      "step": 219590
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.6021368503570557,
      "learning_rate": 3.2541690073496163e-06,
      "loss": 2.0888,
      "step": 219591
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.4127180576324463,
      "learning_rate": 3.253568170509402e-06,
      "loss": 2.094,
      "step": 219592
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.4284234046936035,
      "learning_rate": 3.252967388839922e-06,
      "loss": 1.9578,
      "step": 219593
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.5935615301132202,
      "learning_rate": 3.2523666623413437e-06,
      "loss": 1.9052,
      "step": 219594
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.5068713426589966,
      "learning_rate": 3.2517659910137663e-06,
      "loss": 1.9218,
      "step": 219595
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.8684070110321045,
      "learning_rate": 3.2511653748572565e-06,
      "loss": 1.8727,
      "step": 219596
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.4941129684448242,
      "learning_rate": 3.250564813871981e-06,
      "loss": 1.9747,
      "step": 219597
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.4542996883392334,
      "learning_rate": 3.2499643080580395e-06,
      "loss": 1.9399,
      "step": 219598
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.4486407041549683,
      "learning_rate": 3.249363857415499e-06,
      "loss": 2.0176,
      "step": 219599
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.434525489807129,
      "learning_rate": 3.2487634619445256e-06,
      "loss": 1.9652,
      "step": 219600
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.4436345100402832,
      "learning_rate": 3.2481631216451863e-06,
      "loss": 1.7834,
      "step": 219601
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.7012025117874146,
      "learning_rate": 3.247562836517614e-06,
      "loss": 1.8264,
      "step": 219602
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.4490255117416382,
      "learning_rate": 3.2469626065619425e-06,
      "loss": 1.944,
      "step": 219603
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.6864960193634033,
      "learning_rate": 3.246362431778238e-06,
      "loss": 1.8858,
      "step": 219604
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.4171127080917358,
      "learning_rate": 3.2457623121666664e-06,
      "loss": 2.029,
      "step": 219605
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.3995507955551147,
      "learning_rate": 3.2451622477272953e-06,
      "loss": 1.8934,
      "step": 219606
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.3812823295593262,
      "learning_rate": 3.2445622384602576e-06,
      "loss": 2.0146,
      "step": 219607
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.490915298461914,
      "learning_rate": 3.2439622843656533e-06,
      "loss": 1.8196,
      "step": 219608
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.453078269958496,
      "learning_rate": 3.2433623854436153e-06,
      "loss": 2.1002,
      "step": 219609
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.457865595817566,
      "learning_rate": 3.242762541694177e-06,
      "loss": 1.8378,
      "step": 219610
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.547712802886963,
      "learning_rate": 3.242162753117572e-06,
      "loss": 2.0251,
      "step": 219611
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.4408849477767944,
      "learning_rate": 3.241563019713833e-06,
      "loss": 1.8635,
      "step": 219612
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.3218927383422852,
      "learning_rate": 3.2409633414830937e-06,
      "loss": 1.8385,
      "step": 219613
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.3984805345535278,
      "learning_rate": 3.240363718425454e-06,
      "loss": 1.9435,
      "step": 219614
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.490105390548706,
      "learning_rate": 3.2397641505410466e-06,
      "loss": 1.9169,
      "step": 219615
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.5357948541641235,
      "learning_rate": 3.2391646378299383e-06,
      "loss": 1.9841,
      "step": 219616
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.4488639831542969,
      "learning_rate": 3.2385651802922964e-06,
      "loss": 2.073,
      "step": 219617
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.4551727771759033,
      "learning_rate": 3.2379657779281863e-06,
      "loss": 1.9515,
      "step": 219618
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.5328340530395508,
      "learning_rate": 3.2373664307377755e-06,
      "loss": 1.9303,
      "step": 219619
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.4594707489013672,
      "learning_rate": 3.236767138721097e-06,
      "loss": 2.1564,
      "step": 219620
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.4635765552520752,
      "learning_rate": 3.2361679018783172e-06,
      "loss": 2.1595,
      "step": 219621
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.6005311012268066,
      "learning_rate": 3.235568720209536e-06,
      "loss": 1.8519,
      "step": 219622
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.6501668691635132,
      "learning_rate": 3.2349695937148866e-06,
      "loss": 2.1352,
      "step": 219623
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.458763599395752,
      "learning_rate": 3.234370522394436e-06,
      "loss": 2.0003,
      "step": 219624
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.455967903137207,
      "learning_rate": 3.2337715062483505e-06,
      "loss": 1.9093,
      "step": 219625
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.5061631202697754,
      "learning_rate": 3.233172545276663e-06,
      "loss": 1.7923,
      "step": 219626
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.572241187095642,
      "learning_rate": 3.232573639479541e-06,
      "loss": 1.9144,
      "step": 219627
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.5846989154815674,
      "learning_rate": 3.2319747888570836e-06,
      "loss": 2.0834,
      "step": 219628
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.534032940864563,
      "learning_rate": 3.231375993409424e-06,
      "loss": 2.0129,
      "step": 219629
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.4676443338394165,
      "learning_rate": 3.230777253136663e-06,
      "loss": 1.9701,
      "step": 219630
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.4267491102218628,
      "learning_rate": 3.2301785680388993e-06,
      "loss": 1.8734,
      "step": 219631
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.4286019802093506,
      "learning_rate": 3.2295799381162335e-06,
      "loss": 1.9223,
      "step": 219632
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.5093144178390503,
      "learning_rate": 3.2289813633687657e-06,
      "loss": 1.8934,
      "step": 219633
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.3637547492980957,
      "learning_rate": 3.228382843796662e-06,
      "loss": 1.9606,
      "step": 219634
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.3459287881851196,
      "learning_rate": 3.2277843793999893e-06,
      "loss": 2.034,
      "step": 219635
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.4679679870605469,
      "learning_rate": 3.2271859701788806e-06,
      "loss": 1.9916,
      "step": 219636
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.5156077146530151,
      "learning_rate": 3.2265876161334692e-06,
      "loss": 2.0367,
      "step": 219637
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.4371755123138428,
      "learning_rate": 3.2259893172637883e-06,
      "loss": 2.0333,
      "step": 219638
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.4121427536010742,
      "learning_rate": 3.2253910735700385e-06,
      "loss": 1.8562,
      "step": 219639
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.3833221197128296,
      "learning_rate": 3.224792885052252e-06,
      "loss": 1.7951,
      "step": 219640
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.5092066526412964,
      "learning_rate": 3.2241947517106294e-06,
      "loss": 1.8096,
      "step": 219641
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.8526819944381714,
      "learning_rate": 3.2235966735451704e-06,
      "loss": 2.0428,
      "step": 219642
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.4877015352249146,
      "learning_rate": 3.222998650556108e-06,
      "loss": 1.9734,
      "step": 219643
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.548121452331543,
      "learning_rate": 3.2224006827434425e-06,
      "loss": 2.0394,
      "step": 219644
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.7758922576904297,
      "learning_rate": 3.2218027701073736e-06,
      "loss": 2.0818,
      "step": 219645
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.552548885345459,
      "learning_rate": 3.2212049126479342e-06,
      "loss": 1.8134,
      "step": 219646
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.5002498626708984,
      "learning_rate": 3.220607110365292e-06,
      "loss": 2.2904,
      "step": 219647
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.5336565971374512,
      "learning_rate": 3.220009363259546e-06,
      "loss": 1.928,
      "step": 219648
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.537192940711975,
      "learning_rate": 3.219411671330829e-06,
      "loss": 2.018,
      "step": 219649
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.589657187461853,
      "learning_rate": 3.2188140345791756e-06,
      "loss": 2.086,
      "step": 219650
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.4690749645233154,
      "learning_rate": 3.218216453004785e-06,
      "loss": 2.0703,
      "step": 219651
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.4999706745147705,
      "learning_rate": 3.21761892660769e-06,
      "loss": 2.2131,
      "step": 219652
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.4403342008590698,
      "learning_rate": 3.217021455388058e-06,
      "loss": 2.1299,
      "step": 219653
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.5957934856414795,
      "learning_rate": 3.2164240393459883e-06,
      "loss": 1.9771,
      "step": 219654
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.4759234189987183,
      "learning_rate": 3.2158266784816147e-06,
      "loss": 2.0697,
      "step": 219655
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.6362143754959106,
      "learning_rate": 3.215229372795003e-06,
      "loss": 1.8257,
      "step": 219656
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.3143209218978882,
      "learning_rate": 3.2146321222862535e-06,
      "loss": 1.9408,
      "step": 219657
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.4065238237380981,
      "learning_rate": 3.214034926955533e-06,
      "loss": 2.1886,
      "step": 219658
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.3322514295578003,
      "learning_rate": 3.213437786802908e-06,
      "loss": 1.977,
      "step": 219659
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.3938325643539429,
      "learning_rate": 3.212840701828512e-06,
      "loss": 1.9799,
      "step": 219660
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.4466756582260132,
      "learning_rate": 3.2122436720324773e-06,
      "loss": 1.8514,
      "step": 219661
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.5456335544586182,
      "learning_rate": 3.2116466974148714e-06,
      "loss": 1.9936,
      "step": 219662
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.5475846529006958,
      "learning_rate": 3.211049777975827e-06,
      "loss": 1.9842,
      "step": 219663
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.4591156244277954,
      "learning_rate": 3.2104529137154445e-06,
      "loss": 2.1404,
      "step": 219664
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.5700232982635498,
      "learning_rate": 3.2098561046338568e-06,
      "loss": 2.0333,
      "step": 219665
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.3633986711502075,
      "learning_rate": 3.2092593507311305e-06,
      "loss": 2.1399,
      "step": 219666
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.3903858661651611,
      "learning_rate": 3.208662652007432e-06,
      "loss": 1.9725,
      "step": 219667
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.44410240650177,
      "learning_rate": 3.2080660084628283e-06,
      "loss": 1.8131,
      "step": 219668
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.4979482889175415,
      "learning_rate": 3.2074694200974527e-06,
      "loss": 2.1947,
      "step": 219669
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.4480997323989868,
      "learning_rate": 3.2068728869114048e-06,
      "loss": 2.1428,
      "step": 219670
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.4797459840774536,
      "learning_rate": 3.2062764089048176e-06,
      "loss": 1.8744,
      "step": 219671
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.53928804397583,
      "learning_rate": 3.205679986077758e-06,
      "loss": 2.0844,
      "step": 219672
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.521354079246521,
      "learning_rate": 3.205083618430393e-06,
      "loss": 2.009,
      "step": 219673
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.4867421388626099,
      "learning_rate": 3.2044873059627885e-06,
      "loss": 1.8523,
      "step": 219674
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.4513565301895142,
      "learning_rate": 3.203891048675111e-06,
      "loss": 1.6698,
      "step": 219675
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.475252389907837,
      "learning_rate": 3.2032948465673613e-06,
      "loss": 1.8463,
      "step": 219676
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.429095983505249,
      "learning_rate": 3.202698699639772e-06,
      "loss": 2.07,
      "step": 219677
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.3912603855133057,
      "learning_rate": 3.2021026078923764e-06,
      "loss": 1.9177,
      "step": 219678
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.5395771265029907,
      "learning_rate": 3.2015065713253407e-06,
      "loss": 1.9093,
      "step": 219679
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.4665066003799438,
      "learning_rate": 3.2009105899387657e-06,
      "loss": 2.0598,
      "step": 219680
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.4466667175292969,
      "learning_rate": 3.200314663732684e-06,
      "loss": 2.0984,
      "step": 219681
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.6464768648147583,
      "learning_rate": 3.1997187927072954e-06,
      "loss": 2.128,
      "step": 219682
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.3809809684753418,
      "learning_rate": 3.1991229768627004e-06,
      "loss": 1.8948,
      "step": 219683
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.4306050539016724,
      "learning_rate": 3.198527216198965e-06,
      "loss": 1.9366,
      "step": 219684
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.5678913593292236,
      "learning_rate": 3.1979315107162228e-06,
      "loss": 2.0128,
      "step": 219685
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.4438145160675049,
      "learning_rate": 3.1973358604146073e-06,
      "loss": 2.0473,
      "step": 219686
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.5145246982574463,
      "learning_rate": 3.1967402652941844e-06,
      "loss": 1.8335,
      "step": 219687
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.531083106994629,
      "learning_rate": 3.196144725355121e-06,
      "loss": 1.992,
      "step": 219688
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.4208321571350098,
      "learning_rate": 3.1955492405974835e-06,
      "loss": 1.7825,
      "step": 219689
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.6022106409072876,
      "learning_rate": 3.1949538110214055e-06,
      "loss": 1.9929,
      "step": 219690
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.405417561531067,
      "learning_rate": 3.1943584366269537e-06,
      "loss": 1.7392,
      "step": 219691
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.5532033443450928,
      "learning_rate": 3.1937631174143276e-06,
      "loss": 2.1118,
      "step": 219692
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.564867377281189,
      "learning_rate": 3.193167853383527e-06,
      "loss": 2.0905,
      "step": 219693
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.5387169122695923,
      "learning_rate": 3.1925726445347522e-06,
      "loss": 1.8166,
      "step": 219694
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.5389056205749512,
      "learning_rate": 3.1919774908680694e-06,
      "loss": 2.0266,
      "step": 219695
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.6825177669525146,
      "learning_rate": 3.1913823923836125e-06,
      "loss": 2.0042,
      "step": 219696
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.5632343292236328,
      "learning_rate": 3.1907873490814805e-06,
      "loss": 1.8846,
      "step": 219697
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.4452537298202515,
      "learning_rate": 3.190192360961774e-06,
      "loss": 2.0852,
      "step": 219698
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.4694339036941528,
      "learning_rate": 3.1895974280245927e-06,
      "loss": 1.9489,
      "step": 219699
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.5276554822921753,
      "learning_rate": 3.189002550270103e-06,
      "loss": 1.8579,
      "step": 219700
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.4801595211029053,
      "learning_rate": 3.1884077276983723e-06,
      "loss": 1.786,
      "step": 219701
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.4916749000549316,
      "learning_rate": 3.1878129603095327e-06,
      "loss": 1.7933,
      "step": 219702
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.4021761417388916,
      "learning_rate": 3.1872182481036513e-06,
      "loss": 1.863,
      "step": 219703
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.4618711471557617,
      "learning_rate": 3.1866235910808944e-06,
      "loss": 1.9031,
      "step": 219704
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.5379918813705444,
      "learning_rate": 3.1860289892413295e-06,
      "loss": 1.9299,
      "step": 219705
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.7047196626663208,
      "learning_rate": 3.1854344425850887e-06,
      "loss": 1.9928,
      "step": 219706
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.4258570671081543,
      "learning_rate": 3.1848399511122723e-06,
      "loss": 1.9555,
      "step": 219707
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.4371719360351562,
      "learning_rate": 3.184245514823014e-06,
      "loss": 1.9836,
      "step": 219708
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.448433756828308,
      "learning_rate": 3.18365113371738e-06,
      "loss": 2.0956,
      "step": 219709
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.4578962326049805,
      "learning_rate": 3.1830568077955365e-06,
      "loss": 1.9148,
      "step": 219710
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.4318819046020508,
      "learning_rate": 3.1824625370575508e-06,
      "loss": 1.7818,
      "step": 219711
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.5823028087615967,
      "learning_rate": 3.1818683215035557e-06,
      "loss": 1.982,
      "step": 219712
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.3543083667755127,
      "learning_rate": 3.1812741611336513e-06,
      "loss": 1.8491,
      "step": 219713
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.5346654653549194,
      "learning_rate": 3.180680055947937e-06,
      "loss": 1.9614,
      "step": 219714
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.4464370012283325,
      "learning_rate": 3.1800860059465473e-06,
      "loss": 1.9897,
      "step": 219715
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.4092729091644287,
      "learning_rate": 3.179492011129614e-06,
      "loss": 1.8641,
      "step": 219716
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.5727691650390625,
      "learning_rate": 3.178898071497171e-06,
      "loss": 1.742,
      "step": 219717
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.5050324201583862,
      "learning_rate": 3.178304187049385e-06,
      "loss": 2.2602,
      "step": 219718
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.4189499616622925,
      "learning_rate": 3.177710357786356e-06,
      "loss": 1.8544,
      "step": 219719
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.7088702917099,
      "learning_rate": 3.1771165837082167e-06,
      "loss": 2.0893,
      "step": 219720
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.4415279626846313,
      "learning_rate": 3.176522864815001e-06,
      "loss": 1.9223,
      "step": 219721
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.400671124458313,
      "learning_rate": 3.1759292011069416e-06,
      "loss": 2.0814,
      "step": 219722
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.5059157609939575,
      "learning_rate": 3.1753355925840384e-06,
      "loss": 1.663,
      "step": 219723
    },
    {
      "epoch": 2.86,
      "grad_norm": 2.37969708442688,
      "learning_rate": 3.1747420392464586e-06,
      "loss": 2.1976,
      "step": 219724
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.420716404914856,
      "learning_rate": 3.174148541094268e-06,
      "loss": 2.1016,
      "step": 219725
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.653501272201538,
      "learning_rate": 3.173555098127634e-06,
      "loss": 2.1527,
      "step": 219726
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.4529293775558472,
      "learning_rate": 3.172961710346622e-06,
      "loss": 2.0911,
      "step": 219727
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.474112629890442,
      "learning_rate": 3.1723683777514e-06,
      "loss": 2.0486,
      "step": 219728
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.8209447860717773,
      "learning_rate": 3.171775100342e-06,
      "loss": 1.8456,
      "step": 219729
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.6504510641098022,
      "learning_rate": 3.171181878118589e-06,
      "loss": 1.9387,
      "step": 219730
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.4764337539672852,
      "learning_rate": 3.1705887110812343e-06,
      "loss": 1.9084,
      "step": 219731
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.5200424194335938,
      "learning_rate": 3.1699955992300685e-06,
      "loss": 1.9526,
      "step": 219732
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.429924488067627,
      "learning_rate": 3.169402542565225e-06,
      "loss": 2.0462,
      "step": 219733
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.446547269821167,
      "learning_rate": 3.168809541086803e-06,
      "loss": 2.1103,
      "step": 219734
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.4335650205612183,
      "learning_rate": 3.16821659479487e-06,
      "loss": 1.9314,
      "step": 219735
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.649845838546753,
      "learning_rate": 3.1676237036895925e-06,
      "loss": 1.9497,
      "step": 219736
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.6275397539138794,
      "learning_rate": 3.1670308677710365e-06,
      "loss": 2.1609,
      "step": 219737
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.5358644723892212,
      "learning_rate": 3.166438087039369e-06,
      "loss": 1.7829,
      "step": 219738
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.4245942831039429,
      "learning_rate": 3.165845361494623e-06,
      "loss": 1.7664,
      "step": 219739
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.4870166778564453,
      "learning_rate": 3.1652526911369657e-06,
      "loss": 2.0275,
      "step": 219740
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.7483195066452026,
      "learning_rate": 3.1646600759664964e-06,
      "loss": 2.1009,
      "step": 219741
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.4263699054718018,
      "learning_rate": 3.164067515983315e-06,
      "loss": 2.1259,
      "step": 219742
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.7806850671768188,
      "learning_rate": 3.1634750111875216e-06,
      "loss": 2.1515,
      "step": 219743
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.4123542308807373,
      "learning_rate": 3.162882561579283e-06,
      "loss": 2.0535,
      "step": 219744
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.7355303764343262,
      "learning_rate": 3.162290167158632e-06,
      "loss": 1.8565,
      "step": 219745
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.460092306137085,
      "learning_rate": 3.161697827925702e-06,
      "loss": 1.8528,
      "step": 219746
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.4464815855026245,
      "learning_rate": 3.1611055438806597e-06,
      "loss": 1.9502,
      "step": 219747
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.4303114414215088,
      "learning_rate": 3.160513315023572e-06,
      "loss": 2.0033,
      "step": 219748
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.5138689279556274,
      "learning_rate": 3.1599211413545044e-06,
      "loss": 1.9453,
      "step": 219749
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.6464208364486694,
      "learning_rate": 3.1593290228736247e-06,
      "loss": 1.9517,
      "step": 219750
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.3528048992156982,
      "learning_rate": 3.158736959581032e-06,
      "loss": 1.9474,
      "step": 219751
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.5238980054855347,
      "learning_rate": 3.1581449514768266e-06,
      "loss": 2.0519,
      "step": 219752
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.483952522277832,
      "learning_rate": 3.1575529985611416e-06,
      "loss": 1.851,
      "step": 219753
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.4377509355545044,
      "learning_rate": 3.1569611008340767e-06,
      "loss": 2.0199,
      "step": 219754
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.4889578819274902,
      "learning_rate": 3.1563692582957323e-06,
      "loss": 2.1668,
      "step": 219755
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.54147469997406,
      "learning_rate": 3.155777470946208e-06,
      "loss": 2.0441,
      "step": 219756
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.483491063117981,
      "learning_rate": 3.155185738785637e-06,
      "loss": 1.8726,
      "step": 219757
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.467069387435913,
      "learning_rate": 3.1545940618141196e-06,
      "loss": 1.8856,
      "step": 219758
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.3907837867736816,
      "learning_rate": 3.1540024400317554e-06,
      "loss": 1.9031,
      "step": 219759
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.4487054347991943,
      "learning_rate": 3.1534108734386776e-06,
      "loss": 1.9617,
      "step": 219760
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.4529575109481812,
      "learning_rate": 3.152819362034986e-06,
      "loss": 1.9705,
      "step": 219761
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.5564789772033691,
      "learning_rate": 3.1522279058207813e-06,
      "loss": 1.8116,
      "step": 219762
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.4264063835144043,
      "learning_rate": 3.151636504796162e-06,
      "loss": 1.9418,
      "step": 219763
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.476146936416626,
      "learning_rate": 3.151045158961296e-06,
      "loss": 2.1093,
      "step": 219764
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.538718342781067,
      "learning_rate": 3.1504538683162496e-06,
      "loss": 2.1613,
      "step": 219765
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.5937410593032837,
      "learning_rate": 3.149862632861122e-06,
      "loss": 1.995,
      "step": 219766
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.5490158796310425,
      "learning_rate": 3.149271452596047e-06,
      "loss": 2.2044,
      "step": 219767
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.5877108573913574,
      "learning_rate": 3.1486803275211247e-06,
      "loss": 1.8684,
      "step": 219768
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.5753660202026367,
      "learning_rate": 3.1480892576364215e-06,
      "loss": 2.1262,
      "step": 219769
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.4906352758407593,
      "learning_rate": 3.14749824294217e-06,
      "loss": 1.8385,
      "step": 219770
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.4824821949005127,
      "learning_rate": 3.1469072834383715e-06,
      "loss": 2.126,
      "step": 219771
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.3312591314315796,
      "learning_rate": 3.1463163791251246e-06,
      "loss": 1.9055,
      "step": 219772
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.4447715282440186,
      "learning_rate": 3.1457255300026295e-06,
      "loss": 1.9348,
      "step": 219773
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.608975887298584,
      "learning_rate": 3.1451347360709534e-06,
      "loss": 2.0894,
      "step": 219774
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.4626106023788452,
      "learning_rate": 3.1445439973301624e-06,
      "loss": 1.9171,
      "step": 219775
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.5192887783050537,
      "learning_rate": 3.143953313780423e-06,
      "loss": 2.1258,
      "step": 219776
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.4449223279953003,
      "learning_rate": 3.1433626854218354e-06,
      "loss": 2.012,
      "step": 219777
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.457190990447998,
      "learning_rate": 3.142772112254466e-06,
      "loss": 1.9367,
      "step": 219778
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.4118354320526123,
      "learning_rate": 3.1421815942784812e-06,
      "loss": 2.1011,
      "step": 219779
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.4788315296173096,
      "learning_rate": 3.1415911314939814e-06,
      "loss": 1.791,
      "step": 219780
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.5069148540496826,
      "learning_rate": 3.141000723901066e-06,
      "loss": 1.9836,
      "step": 219781
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.637871265411377,
      "learning_rate": 3.140410371499835e-06,
      "loss": 1.8944,
      "step": 219782
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.4606915712356567,
      "learning_rate": 3.139820074290389e-06,
      "loss": 1.8728,
      "step": 219783
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.4761508703231812,
      "learning_rate": 3.1392298322728604e-06,
      "loss": 1.9001,
      "step": 219784
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.3309853076934814,
      "learning_rate": 3.1386396454473495e-06,
      "loss": 2.1037,
      "step": 219785
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.3927943706512451,
      "learning_rate": 3.138049513813956e-06,
      "loss": 1.9432,
      "step": 219786
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.4820455312728882,
      "learning_rate": 3.1374594373728467e-06,
      "loss": 1.8634,
      "step": 219787
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.4626661539077759,
      "learning_rate": 3.1368694161240548e-06,
      "loss": 1.9784,
      "step": 219788
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.4003551006317139,
      "learning_rate": 3.1362794500677467e-06,
      "loss": 1.7795,
      "step": 219789
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.4682382345199585,
      "learning_rate": 3.1356895392039894e-06,
      "loss": 1.9991,
      "step": 219790
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.558257818222046,
      "learning_rate": 3.1350996835329155e-06,
      "loss": 1.9599,
      "step": 219791
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.3879809379577637,
      "learning_rate": 3.1345098830546258e-06,
      "loss": 1.784,
      "step": 219792
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.4900070428848267,
      "learning_rate": 3.1339201377692193e-06,
      "loss": 1.9964,
      "step": 219793
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.4103504419326782,
      "learning_rate": 3.1333304476768295e-06,
      "loss": 1.9945,
      "step": 219794
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.4126083850860596,
      "learning_rate": 3.13274081277759e-06,
      "loss": 1.9949,
      "step": 219795
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.3806513547897339,
      "learning_rate": 3.132151233071567e-06,
      "loss": 1.786,
      "step": 219796
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.3169114589691162,
      "learning_rate": 3.1315617085588607e-06,
      "loss": 1.7433,
      "step": 219797
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.5459926128387451,
      "learning_rate": 3.130972239239604e-06,
      "loss": 2.0622,
      "step": 219798
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.611440896987915,
      "learning_rate": 3.1303828251139306e-06,
      "loss": 1.9154,
      "step": 219799
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.3759126663208008,
      "learning_rate": 3.1297934661818735e-06,
      "loss": 1.8899,
      "step": 219800
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.342136263847351,
      "learning_rate": 3.1292041624436328e-06,
      "loss": 1.9656,
      "step": 219801
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.6280542612075806,
      "learning_rate": 3.1286149138992743e-06,
      "loss": 2.2314,
      "step": 219802
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.2463231086730957,
      "learning_rate": 3.128025720548899e-06,
      "loss": 1.6616,
      "step": 219803
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.5062642097473145,
      "learning_rate": 3.1274365823926064e-06,
      "loss": 1.9499,
      "step": 219804
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.366612434387207,
      "learning_rate": 3.1268474994305624e-06,
      "loss": 1.7798,
      "step": 219805
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.4219999313354492,
      "learning_rate": 3.1262584716628014e-06,
      "loss": 2.0692,
      "step": 219806
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.49848210811615,
      "learning_rate": 3.1256694990895225e-06,
      "loss": 2.0225,
      "step": 219807
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.5327876806259155,
      "learning_rate": 3.1250805817107594e-06,
      "loss": 2.08,
      "step": 219808
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.4461333751678467,
      "learning_rate": 3.124491719526645e-06,
      "loss": 1.9675,
      "step": 219809
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.5696357488632202,
      "learning_rate": 3.1239029125373126e-06,
      "loss": 2.0516,
      "step": 219810
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.398573637008667,
      "learning_rate": 3.123314160742829e-06,
      "loss": 1.7967,
      "step": 219811
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.4565229415893555,
      "learning_rate": 3.1227254641433275e-06,
      "loss": 1.754,
      "step": 219812
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.40045964717865,
      "learning_rate": 3.1221368227389076e-06,
      "loss": 2.0757,
      "step": 219813
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.4072741270065308,
      "learning_rate": 3.121548236529703e-06,
      "loss": 1.8955,
      "step": 219814
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.3971813917160034,
      "learning_rate": 3.120959705515813e-06,
      "loss": 1.7283,
      "step": 219815
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.398612141609192,
      "learning_rate": 3.120371229697338e-06,
      "loss": 2.0412,
      "step": 219816
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.5038087368011475,
      "learning_rate": 3.1197828090743784e-06,
      "loss": 1.9856,
      "step": 219817
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.4408018589019775,
      "learning_rate": 3.119194443647033e-06,
      "loss": 1.8843,
      "step": 219818
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.4983958005905151,
      "learning_rate": 3.118606133415469e-06,
      "loss": 2.1011,
      "step": 219819
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.3730361461639404,
      "learning_rate": 3.1180178783797527e-06,
      "loss": 1.8471,
      "step": 219820
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.6041868925094604,
      "learning_rate": 3.1174296785400177e-06,
      "loss": 1.9169,
      "step": 219821
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.6170891523361206,
      "learning_rate": 3.1168415338963304e-06,
      "loss": 1.9026,
      "step": 219822
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.3535953760147095,
      "learning_rate": 3.116253444448824e-06,
      "loss": 1.9135,
      "step": 219823
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.4736770391464233,
      "learning_rate": 3.1156654101975986e-06,
      "loss": 1.8659,
      "step": 219824
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.5962718725204468,
      "learning_rate": 3.1150774311428208e-06,
      "loss": 1.8249,
      "step": 219825
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.4326063394546509,
      "learning_rate": 3.114489507284523e-06,
      "loss": 2.2157,
      "step": 219826
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.4778528213500977,
      "learning_rate": 3.113901638622873e-06,
      "loss": 2.0799,
      "step": 219827
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.373415470123291,
      "learning_rate": 3.113313825157904e-06,
      "loss": 1.7549,
      "step": 219828
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.4839041233062744,
      "learning_rate": 3.1127260668898146e-06,
      "loss": 2.0818,
      "step": 219829
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.6198821067810059,
      "learning_rate": 3.1121383638186393e-06,
      "loss": 1.9401,
      "step": 219830
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.4891548156738281,
      "learning_rate": 3.1115507159445443e-06,
      "loss": 2.1007,
      "step": 219831
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.456680417060852,
      "learning_rate": 3.110963123267629e-06,
      "loss": 2.1025,
      "step": 219832
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.4320991039276123,
      "learning_rate": 3.110375585787994e-06,
      "loss": 1.959,
      "step": 219833
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.393001675605774,
      "learning_rate": 3.1097881035057058e-06,
      "loss": 1.867,
      "step": 219834
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.4608203172683716,
      "learning_rate": 3.109200676420931e-06,
      "loss": 1.7341,
      "step": 219835
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.4535763263702393,
      "learning_rate": 3.108613304533769e-06,
      "loss": 1.9503,
      "step": 219836
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.3394732475280762,
      "learning_rate": 3.1080259878443204e-06,
      "loss": 1.764,
      "step": 219837
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.5222370624542236,
      "learning_rate": 3.1074387263526844e-06,
      "loss": 1.8906,
      "step": 219838
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.376129150390625,
      "learning_rate": 3.106851520058995e-06,
      "loss": 1.9183,
      "step": 219839
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.5922825336456299,
      "learning_rate": 3.106264368963318e-06,
      "loss": 1.9355,
      "step": 219840
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.4530529975891113,
      "learning_rate": 3.1056772730658206e-06,
      "loss": 2.0577,
      "step": 219841
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.4485517740249634,
      "learning_rate": 3.105090232366536e-06,
      "loss": 2.1677,
      "step": 219842
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.524841070175171,
      "learning_rate": 3.104503246865664e-06,
      "loss": 1.9082,
      "step": 219843
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.5636913776397705,
      "learning_rate": 3.103916316563271e-06,
      "loss": 1.9848,
      "step": 219844
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.571110486984253,
      "learning_rate": 3.103329441459457e-06,
      "loss": 2.0434,
      "step": 219845
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.6021698713302612,
      "learning_rate": 3.1027426215543215e-06,
      "loss": 1.9136,
      "step": 219846
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.387757420539856,
      "learning_rate": 3.1021558568479987e-06,
      "loss": 1.8345,
      "step": 219847
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.461663842201233,
      "learning_rate": 3.101569147340588e-06,
      "loss": 2.1642,
      "step": 219848
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.4113234281539917,
      "learning_rate": 3.1009824930321893e-06,
      "loss": 1.7281,
      "step": 219849
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.4871090650558472,
      "learning_rate": 3.100395893922936e-06,
      "loss": 2.0062,
      "step": 219850
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.5902364253997803,
      "learning_rate": 3.099809350012927e-06,
      "loss": 1.8894,
      "step": 219851
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.4503614902496338,
      "learning_rate": 3.099222861302264e-06,
      "loss": 1.9356,
      "step": 219852
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.4200531244277954,
      "learning_rate": 3.098636427791079e-06,
      "loss": 2.1127,
      "step": 219853
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.505039095878601,
      "learning_rate": 3.098050049479406e-06,
      "loss": 1.9553,
      "step": 219854
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.4763954877853394,
      "learning_rate": 3.097463726367477e-06,
      "loss": 2.1455,
      "step": 219855
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.472007155418396,
      "learning_rate": 3.0968774584552935e-06,
      "loss": 2.2806,
      "step": 219856
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.3684008121490479,
      "learning_rate": 3.096291245743021e-06,
      "loss": 2.0602,
      "step": 219857
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.4931350946426392,
      "learning_rate": 3.095705088230727e-06,
      "loss": 1.9239,
      "step": 219858
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.543912649154663,
      "learning_rate": 3.0951189859185767e-06,
      "loss": 2.2029,
      "step": 219859
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.4879616498947144,
      "learning_rate": 3.0945329388066375e-06,
      "loss": 1.902,
      "step": 219860
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.4858355522155762,
      "learning_rate": 3.0939469468950427e-06,
      "loss": 2.0009,
      "step": 219861
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.6359411478042603,
      "learning_rate": 3.0933610101838592e-06,
      "loss": 2.0331,
      "step": 219862
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.4137836694717407,
      "learning_rate": 3.0927751286732525e-06,
      "loss": 1.8452,
      "step": 219863
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.4852173328399658,
      "learning_rate": 3.0921893023632905e-06,
      "loss": 2.1312,
      "step": 219864
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.4879679679870605,
      "learning_rate": 3.091603531254072e-06,
      "loss": 1.9965,
      "step": 219865
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.4945266246795654,
      "learning_rate": 3.0910178153457644e-06,
      "loss": 2.1267,
      "step": 219866
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.3960047960281372,
      "learning_rate": 3.0904321546384338e-06,
      "loss": 2.0431,
      "step": 219867
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.364964246749878,
      "learning_rate": 3.08984654913218e-06,
      "loss": 1.981,
      "step": 219868
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.4517502784729004,
      "learning_rate": 3.0892609988271366e-06,
      "loss": 1.9752,
      "step": 219869
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.450954794883728,
      "learning_rate": 3.0886755037234033e-06,
      "loss": 1.9535,
      "step": 219870
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.416684865951538,
      "learning_rate": 3.0880900638210805e-06,
      "loss": 1.9084,
      "step": 219871
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.4833312034606934,
      "learning_rate": 3.087504679120334e-06,
      "loss": 1.8158,
      "step": 219872
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.4006385803222656,
      "learning_rate": 3.0869193496211973e-06,
      "loss": 1.7907,
      "step": 219873
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.5923525094985962,
      "learning_rate": 3.0863340753237708e-06,
      "loss": 2.0734,
      "step": 219874
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.3989061117172241,
      "learning_rate": 3.0857488562282205e-06,
      "loss": 1.955,
      "step": 219875
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.3950446844100952,
      "learning_rate": 3.0851636923346802e-06,
      "loss": 2.0394,
      "step": 219876
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.4874358177185059,
      "learning_rate": 3.084578583643149e-06,
      "loss": 1.9122,
      "step": 219877
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.4553526639938354,
      "learning_rate": 3.0839935301538276e-06,
      "loss": 2.1509,
      "step": 219878
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.6714457273483276,
      "learning_rate": 3.083408531866782e-06,
      "loss": 1.9984,
      "step": 219879
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.5937554836273193,
      "learning_rate": 3.0828235887821464e-06,
      "loss": 2.0114,
      "step": 219880
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.5084258317947388,
      "learning_rate": 3.0822387009000526e-06,
      "loss": 1.9564,
      "step": 219881
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.4628266096115112,
      "learning_rate": 3.0816538682205348e-06,
      "loss": 1.8632,
      "step": 219882
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.4950611591339111,
      "learning_rate": 3.081069090743726e-06,
      "loss": 1.8853,
      "step": 219883
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.3911869525909424,
      "learning_rate": 3.0804843684697933e-06,
      "loss": 1.859,
      "step": 219884
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.4027448892593384,
      "learning_rate": 3.079899701398769e-06,
      "loss": 2.1983,
      "step": 219885
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.459743618965149,
      "learning_rate": 3.0793150895308206e-06,
      "loss": 2.08,
      "step": 219886
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.4340890645980835,
      "learning_rate": 3.078730532866047e-06,
      "loss": 1.8342,
      "step": 219887
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.5728988647460938,
      "learning_rate": 3.078146031404516e-06,
      "loss": 2.0017,
      "step": 219888
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.4748269319534302,
      "learning_rate": 3.07756158514636e-06,
      "loss": 2.1503,
      "step": 219889
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.4700418710708618,
      "learning_rate": 3.0769771940916786e-06,
      "loss": 1.9842,
      "step": 219890
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.488895058631897,
      "learning_rate": 3.076392858240606e-06,
      "loss": 1.7878,
      "step": 219891
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.4261767864227295,
      "learning_rate": 3.075808577593242e-06,
      "loss": 1.9558,
      "step": 219892
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.6279569864273071,
      "learning_rate": 3.075224352149719e-06,
      "loss": 2.3477,
      "step": 219893
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.4606928825378418,
      "learning_rate": 3.074640181910071e-06,
      "loss": 2.0889,
      "step": 219894
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.5878345966339111,
      "learning_rate": 3.074056066874464e-06,
      "loss": 2.0526,
      "step": 219895
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.593321442604065,
      "learning_rate": 3.0734720070429984e-06,
      "loss": 1.899,
      "step": 219896
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.4921503067016602,
      "learning_rate": 3.0728880024157743e-06,
      "loss": 1.9945,
      "step": 219897
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.5978134870529175,
      "learning_rate": 3.0723040529929243e-06,
      "loss": 2.0046,
      "step": 219898
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.3335756063461304,
      "learning_rate": 3.0717201587745153e-06,
      "loss": 1.7882,
      "step": 219899
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.3785632848739624,
      "learning_rate": 3.071136319760714e-06,
      "loss": 1.8293,
      "step": 219900
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.431636929512024,
      "learning_rate": 3.0705525359515536e-06,
      "loss": 2.0767,
      "step": 219901
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.5188064575195312,
      "learning_rate": 3.0699688073472008e-06,
      "loss": 2.017,
      "step": 219902
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.458386778831482,
      "learning_rate": 3.0693851339477214e-06,
      "loss": 1.9635,
      "step": 219903
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.5116866827011108,
      "learning_rate": 3.0688015157532497e-06,
      "loss": 1.9217,
      "step": 219904
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.4040806293487549,
      "learning_rate": 3.0682179527639183e-06,
      "loss": 2.0894,
      "step": 219905
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.4489675760269165,
      "learning_rate": 3.0676344449798275e-06,
      "loss": 1.8713,
      "step": 219906
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.5633587837219238,
      "learning_rate": 3.0670509924010433e-06,
      "loss": 1.9478,
      "step": 219907
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.5653538703918457,
      "learning_rate": 3.0664675950276997e-06,
      "loss": 1.717,
      "step": 219908
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.445378303527832,
      "learning_rate": 3.0658842528598958e-06,
      "loss": 1.8021,
      "step": 219909
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.4218555688858032,
      "learning_rate": 3.0653009658977657e-06,
      "loss": 1.9512,
      "step": 219910
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.4872835874557495,
      "learning_rate": 3.0647177341413755e-06,
      "loss": 1.5192,
      "step": 219911
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.464375376701355,
      "learning_rate": 3.0641345575908914e-06,
      "loss": 1.8762,
      "step": 219912
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.4615339040756226,
      "learning_rate": 3.0635514362463475e-06,
      "loss": 1.8346,
      "step": 219913
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.4897112846374512,
      "learning_rate": 3.0629683701079433e-06,
      "loss": 1.6674,
      "step": 219914
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.5514521598815918,
      "learning_rate": 3.062385359175712e-06,
      "loss": 1.9531,
      "step": 219915
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.5663819313049316,
      "learning_rate": 3.0618024034497868e-06,
      "loss": 2.2079,
      "step": 219916
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.4396693706512451,
      "learning_rate": 3.061219502930301e-06,
      "loss": 1.8891,
      "step": 219917
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.3571208715438843,
      "learning_rate": 3.0606366576173213e-06,
      "loss": 2.0916,
      "step": 219918
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.4159226417541504,
      "learning_rate": 3.060053867510981e-06,
      "loss": 1.938,
      "step": 219919
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.3970718383789062,
      "learning_rate": 3.0594711326113795e-06,
      "loss": 1.9831,
      "step": 219920
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.3804126977920532,
      "learning_rate": 3.0588884529186175e-06,
      "loss": 1.8779,
      "step": 219921
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.4539823532104492,
      "learning_rate": 3.058305828432828e-06,
      "loss": 1.7745,
      "step": 219922
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.6304737329483032,
      "learning_rate": 3.05772325915411e-06,
      "loss": 1.9892,
      "step": 219923
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.5052969455718994,
      "learning_rate": 3.0571407450825313e-06,
      "loss": 2.007,
      "step": 219924
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.540328860282898,
      "learning_rate": 3.056558286218258e-06,
      "loss": 2.121,
      "step": 219925
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.5093846321105957,
      "learning_rate": 3.05597588256139e-06,
      "loss": 2.1855,
      "step": 219926
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.425248384475708,
      "learning_rate": 3.055393534111994e-06,
      "loss": 2.2085,
      "step": 219927
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.3901976346969604,
      "learning_rate": 3.0548112408702362e-06,
      "loss": 2.1209,
      "step": 219928
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.522425889968872,
      "learning_rate": 3.0542290028361506e-06,
      "loss": 2.3397,
      "step": 219929
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.4239484071731567,
      "learning_rate": 3.0536468200099364e-06,
      "loss": 2.0799,
      "step": 219930
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.4155832529067993,
      "learning_rate": 3.053064692391627e-06,
      "loss": 2.0024,
      "step": 219931
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.4664170742034912,
      "learning_rate": 3.0524826199813556e-06,
      "loss": 1.9339,
      "step": 219932
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.5184917449951172,
      "learning_rate": 3.0519006027792557e-06,
      "loss": 2.1671,
      "step": 219933
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.5945627689361572,
      "learning_rate": 3.051318640785394e-06,
      "loss": 1.8101,
      "step": 219934
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.4700335264205933,
      "learning_rate": 3.0507367339999034e-06,
      "loss": 1.8209,
      "step": 219935
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.4294015169143677,
      "learning_rate": 3.050154882422884e-06,
      "loss": 1.8749,
      "step": 219936
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.3363126516342163,
      "learning_rate": 3.0495730860544354e-06,
      "loss": 1.9233,
      "step": 219937
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.502945065498352,
      "learning_rate": 3.0489913448947246e-06,
      "loss": 1.9216,
      "step": 219938
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.6177407503128052,
      "learning_rate": 3.0484096589437514e-06,
      "loss": 2.1816,
      "step": 219939
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.4716625213623047,
      "learning_rate": 3.0478280282017153e-06,
      "loss": 2.0041,
      "step": 219940
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.3037346601486206,
      "learning_rate": 3.0472464526686836e-06,
      "loss": 1.7936,
      "step": 219941
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.5012072324752808,
      "learning_rate": 3.046664932344789e-06,
      "loss": 1.9724,
      "step": 219942
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.4637300968170166,
      "learning_rate": 3.0460834672300984e-06,
      "loss": 1.7987,
      "step": 219943
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.7570993900299072,
      "learning_rate": 3.045502057324778e-06,
      "loss": 1.9982,
      "step": 219944
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.6049124002456665,
      "learning_rate": 3.044920702628861e-06,
      "loss": 1.9939,
      "step": 219945
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.4437999725341797,
      "learning_rate": 3.044339403142548e-06,
      "loss": 2.0211,
      "step": 219946
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.4877723455429077,
      "learning_rate": 3.0437581588658723e-06,
      "loss": 2.074,
      "step": 219947
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.4115862846374512,
      "learning_rate": 3.043176969798966e-06,
      "loss": 1.9922,
      "step": 219948
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.5567281246185303,
      "learning_rate": 3.0425958359419635e-06,
      "loss": 1.8396,
      "step": 219949
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.5754761695861816,
      "learning_rate": 3.0420147572948974e-06,
      "loss": 2.2708,
      "step": 219950
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.3762925863265991,
      "learning_rate": 3.0414337338579674e-06,
      "loss": 1.8601,
      "step": 219951
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.4149963855743408,
      "learning_rate": 3.0408527656312408e-06,
      "loss": 1.8171,
      "step": 219952
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.4473252296447754,
      "learning_rate": 3.040271852614784e-06,
      "loss": 1.9096,
      "step": 219953
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.5310019254684448,
      "learning_rate": 3.0396909948087966e-06,
      "loss": 1.8914,
      "step": 219954
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.4752542972564697,
      "learning_rate": 3.039110192213312e-06,
      "loss": 1.8813,
      "step": 219955
    },
    {
      "epoch": 2.86,
      "grad_norm": 2.430192232131958,
      "learning_rate": 3.0385294448284637e-06,
      "loss": 2.1081,
      "step": 219956
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.4772148132324219,
      "learning_rate": 3.037948752654351e-06,
      "loss": 2.1716,
      "step": 219957
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.671749472618103,
      "learning_rate": 3.0373681156911077e-06,
      "loss": 2.211,
      "step": 219958
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.5350075960159302,
      "learning_rate": 3.0367875339388004e-06,
      "loss": 2.1628,
      "step": 219959
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.4007012844085693,
      "learning_rate": 3.036207007397595e-06,
      "loss": 1.9455,
      "step": 219960
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.392507791519165,
      "learning_rate": 3.0356265360675593e-06,
      "loss": 1.871,
      "step": 219961
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.5156344175338745,
      "learning_rate": 3.0350461199487585e-06,
      "loss": 1.9216,
      "step": 219962
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.4129208326339722,
      "learning_rate": 3.0344657590413936e-06,
      "loss": 1.7455,
      "step": 219963
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.6321940422058105,
      "learning_rate": 3.033885453345497e-06,
      "loss": 1.9399,
      "step": 219964
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.6117937564849854,
      "learning_rate": 3.033305202861236e-06,
      "loss": 2.0512,
      "step": 219965
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.4511889219284058,
      "learning_rate": 3.032725007588677e-06,
      "loss": 2.0129,
      "step": 219966
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.608866810798645,
      "learning_rate": 3.0321448675279526e-06,
      "loss": 1.9513,
      "step": 219967
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.4782878160476685,
      "learning_rate": 3.03156478267913e-06,
      "loss": 1.8592,
      "step": 219968
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.5678319931030273,
      "learning_rate": 3.0309847530423757e-06,
      "loss": 1.8332,
      "step": 219969
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.3728581666946411,
      "learning_rate": 3.0304047786177233e-06,
      "loss": 1.8672,
      "step": 219970
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.4604471921920776,
      "learning_rate": 3.029824859405372e-06,
      "loss": 2.0281,
      "step": 219971
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.3993008136749268,
      "learning_rate": 3.0292449954053557e-06,
      "loss": 2.0727,
      "step": 219972
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.6010817289352417,
      "learning_rate": 3.0286651866178403e-06,
      "loss": 1.8891,
      "step": 219973
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.4766356945037842,
      "learning_rate": 3.0280854330428593e-06,
      "loss": 1.9482,
      "step": 219974
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.4773831367492676,
      "learning_rate": 3.0275057346805796e-06,
      "loss": 1.9731,
      "step": 219975
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.534182071685791,
      "learning_rate": 3.0269260915310677e-06,
      "loss": 1.9451,
      "step": 219976
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.401764154434204,
      "learning_rate": 3.02634650359449e-06,
      "loss": 1.9013,
      "step": 219977
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.4740616083145142,
      "learning_rate": 3.0257669708709465e-06,
      "loss": 1.7831,
      "step": 219978
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.4172486066818237,
      "learning_rate": 3.0251874933604702e-06,
      "loss": 1.7159,
      "step": 219979
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.4025911092758179,
      "learning_rate": 3.024608071063228e-06,
      "loss": 2.047,
      "step": 219980
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.4461116790771484,
      "learning_rate": 3.02402870397932e-06,
      "loss": 1.9272,
      "step": 219981
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.5112642049789429,
      "learning_rate": 3.0234493921088456e-06,
      "loss": 2.2475,
      "step": 219982
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.507380723953247,
      "learning_rate": 3.022870135451938e-06,
      "loss": 2.0548,
      "step": 219983
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.3341017961502075,
      "learning_rate": 3.022290934008664e-06,
      "loss": 1.8751,
      "step": 219984
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.4933243989944458,
      "learning_rate": 3.0217117877791575e-06,
      "loss": 1.9675,
      "step": 219985
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.409265398979187,
      "learning_rate": 3.0211326967635174e-06,
      "loss": 2.1033,
      "step": 219986
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.5133678913116455,
      "learning_rate": 3.0205536609618775e-06,
      "loss": 2.0252,
      "step": 219987
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.4866498708724976,
      "learning_rate": 3.0199746803742708e-06,
      "loss": 1.9462,
      "step": 219988
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.456408143043518,
      "learning_rate": 3.0193957550008972e-06,
      "loss": 2.1951,
      "step": 219989
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.462764859199524,
      "learning_rate": 3.0188168848418237e-06,
      "loss": 2.031,
      "step": 219990
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.5575882196426392,
      "learning_rate": 3.01823806989715e-06,
      "loss": 1.8691,
      "step": 219991
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.4488962888717651,
      "learning_rate": 3.017659310167009e-06,
      "loss": 1.9417,
      "step": 219992
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.5055245161056519,
      "learning_rate": 3.017080605651467e-06,
      "loss": 1.844,
      "step": 219993
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.4359906911849976,
      "learning_rate": 3.0165019563506587e-06,
      "loss": 1.8657,
      "step": 219994
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.5133624076843262,
      "learning_rate": 3.0159233622646826e-06,
      "loss": 2.1793,
      "step": 219995
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.4554208517074585,
      "learning_rate": 3.0153448233936728e-06,
      "loss": 2.1488,
      "step": 219996
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.5340702533721924,
      "learning_rate": 3.0147663397377287e-06,
      "loss": 1.8842,
      "step": 219997
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.4644579887390137,
      "learning_rate": 3.014187911296917e-06,
      "loss": 2.1191,
      "step": 219998
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.5424185991287231,
      "learning_rate": 3.013609538071371e-06,
      "loss": 1.8978,
      "step": 219999
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.3394783735275269,
      "learning_rate": 3.0130312200611905e-06,
      "loss": 1.7911,
      "step": 220000
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.4561687707901,
      "learning_rate": 3.0124529572665425e-06,
      "loss": 1.851,
      "step": 220001
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.4747916460037231,
      "learning_rate": 3.0118747496874262e-06,
      "loss": 1.8344,
      "step": 220002
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.5383857488632202,
      "learning_rate": 3.0112965973240754e-06,
      "loss": 1.9719,
      "step": 220003
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.464137315750122,
      "learning_rate": 3.01071850017649e-06,
      "loss": 1.7868,
      "step": 220004
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.4197649955749512,
      "learning_rate": 3.010140458244803e-06,
      "loss": 2.0354,
      "step": 220005
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.5239808559417725,
      "learning_rate": 3.0095624715291477e-06,
      "loss": 2.1865,
      "step": 220006
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.5698734521865845,
      "learning_rate": 3.0089845400296573e-06,
      "loss": 2.0446,
      "step": 220007
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.6084578037261963,
      "learning_rate": 3.0084066637463656e-06,
      "loss": 1.9398,
      "step": 220008
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.3745540380477905,
      "learning_rate": 3.007828842679438e-06,
      "loss": 1.7488,
      "step": 220009
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.5344388484954834,
      "learning_rate": 3.0072510768289424e-06,
      "loss": 1.8022,
      "step": 220010
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.4392601251602173,
      "learning_rate": 3.0066733661950117e-06,
      "loss": 1.9466,
      "step": 220011
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.4969675540924072,
      "learning_rate": 3.006095710777745e-06,
      "loss": 2.0198,
      "step": 220012
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.3400425910949707,
      "learning_rate": 3.005518110577243e-06,
      "loss": 1.8862,
      "step": 220013
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.431028962135315,
      "learning_rate": 3.0049405655936387e-06,
      "loss": 1.8119,
      "step": 220014
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.3765400648117065,
      "learning_rate": 3.004363075827032e-06,
      "loss": 1.8088,
      "step": 220015
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.658519983291626,
      "learning_rate": 3.0037856412774895e-06,
      "loss": 1.9357,
      "step": 220016
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.4332847595214844,
      "learning_rate": 3.003208261945178e-06,
      "loss": 1.9049,
      "step": 220017
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.5111244916915894,
      "learning_rate": 3.002630937830164e-06,
      "loss": 1.8427,
      "step": 220018
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.5040442943572998,
      "learning_rate": 3.002053668932547e-06,
      "loss": 2.073,
      "step": 220019
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.3964465856552124,
      "learning_rate": 3.0014764552524605e-06,
      "loss": 1.8945,
      "step": 220020
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.4184930324554443,
      "learning_rate": 3.000899296790038e-06,
      "loss": 2.0473,
      "step": 220021
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.6039875745773315,
      "learning_rate": 3.000322193545346e-06,
      "loss": 2.1328,
      "step": 220022
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.5133711099624634,
      "learning_rate": 2.999745145518484e-06,
      "loss": 1.9909,
      "step": 220023
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.5327715873718262,
      "learning_rate": 2.999168152709586e-06,
      "loss": 2.1661,
      "step": 220024
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.760930061340332,
      "learning_rate": 2.9985912151187506e-06,
      "loss": 1.9691,
      "step": 220025
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.6062520742416382,
      "learning_rate": 2.998014332746046e-06,
      "loss": 2.0144,
      "step": 220026
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.6085882186889648,
      "learning_rate": 2.997437505591671e-06,
      "loss": 2.0161,
      "step": 220027
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.4741780757904053,
      "learning_rate": 2.9968607336556592e-06,
      "loss": 1.8227,
      "step": 220028
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.3769543170928955,
      "learning_rate": 2.996284016938144e-06,
      "loss": 1.9955,
      "step": 220029
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.576310157775879,
      "learning_rate": 2.9957073554391917e-06,
      "loss": 2.1392,
      "step": 220030
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.4990166425704956,
      "learning_rate": 2.9951307491589694e-06,
      "loss": 2.0694,
      "step": 220031
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.4024646282196045,
      "learning_rate": 2.994554198097543e-06,
      "loss": 1.9871,
      "step": 220032
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.4227896928787231,
      "learning_rate": 2.993977702255046e-06,
      "loss": 1.8534,
      "step": 220033
    },
    {
      "epoch": 2.86,
      "grad_norm": 1.307488203048706,
      "learning_rate": 2.9934012616316115e-06,
      "loss": 1.9437,
      "step": 220034
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.383528470993042,
      "learning_rate": 2.99282487622724e-06,
      "loss": 1.9626,
      "step": 220035
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.5498156547546387,
      "learning_rate": 2.992248546042164e-06,
      "loss": 2.106,
      "step": 220036
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.2700045108795166,
      "learning_rate": 2.991672271076417e-06,
      "loss": 1.801,
      "step": 220037
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.6315913200378418,
      "learning_rate": 2.991096051330133e-06,
      "loss": 2.0955,
      "step": 220038
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.4481291770935059,
      "learning_rate": 2.9905198868034107e-06,
      "loss": 1.8528,
      "step": 220039
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.704997181892395,
      "learning_rate": 2.989943777496351e-06,
      "loss": 1.8648,
      "step": 220040
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.4496830701828003,
      "learning_rate": 2.989367723409053e-06,
      "loss": 1.9769,
      "step": 220041
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.6889129877090454,
      "learning_rate": 2.9887917245416506e-06,
      "loss": 2.0137,
      "step": 220042
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.4345780611038208,
      "learning_rate": 2.9882157808942433e-06,
      "loss": 2.1156,
      "step": 220043
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.5435298681259155,
      "learning_rate": 2.9876398924668975e-06,
      "loss": 1.9575,
      "step": 220044
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.4916688203811646,
      "learning_rate": 2.9870640592598137e-06,
      "loss": 1.9751,
      "step": 220045
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.4700016975402832,
      "learning_rate": 2.9864882812730253e-06,
      "loss": 2.062,
      "step": 220046
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.6571341753005981,
      "learning_rate": 2.9859125585066313e-06,
      "loss": 1.9784,
      "step": 220047
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.4751698970794678,
      "learning_rate": 2.9853368909607656e-06,
      "loss": 1.9269,
      "step": 220048
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.5918352603912354,
      "learning_rate": 2.984761278635528e-06,
      "loss": 1.7752,
      "step": 220049
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.3751270771026611,
      "learning_rate": 2.9841857215310515e-06,
      "loss": 1.8798,
      "step": 220050
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.6070752143859863,
      "learning_rate": 2.9836102196474364e-06,
      "loss": 1.8637,
      "step": 220051
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.2548750638961792,
      "learning_rate": 2.983034772984749e-06,
      "loss": 1.7563,
      "step": 220052
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.5621514320373535,
      "learning_rate": 2.982459381543123e-06,
      "loss": 1.852,
      "step": 220053
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.4347978830337524,
      "learning_rate": 2.981884045322658e-06,
      "loss": 1.9044,
      "step": 220054
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.3780056238174438,
      "learning_rate": 2.9813087643234867e-06,
      "loss": 1.9376,
      "step": 220055
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.4189212322235107,
      "learning_rate": 2.9807335385456764e-06,
      "loss": 1.9609,
      "step": 220056
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.4442973136901855,
      "learning_rate": 2.9801583679893605e-06,
      "loss": 1.7958,
      "step": 220057
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.4459816217422485,
      "learning_rate": 2.9795832526546714e-06,
      "loss": 2.0189,
      "step": 220058
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.4551982879638672,
      "learning_rate": 2.979008192541643e-06,
      "loss": 1.7028,
      "step": 220059
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.4447616338729858,
      "learning_rate": 2.978433187650442e-06,
      "loss": 2.0748,
      "step": 220060
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.4181097745895386,
      "learning_rate": 2.9778582379811343e-06,
      "loss": 2.0076,
      "step": 220061
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.4355366230010986,
      "learning_rate": 2.977283343533887e-06,
      "loss": 2.0447,
      "step": 220062
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.5688129663467407,
      "learning_rate": 2.9767085043087335e-06,
      "loss": 1.945,
      "step": 220063
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.4436793327331543,
      "learning_rate": 2.97613372030584e-06,
      "loss": 1.968,
      "step": 220064
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.3102182149887085,
      "learning_rate": 2.975558991525273e-06,
      "loss": 1.8423,
      "step": 220065
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.429519772529602,
      "learning_rate": 2.974984317967166e-06,
      "loss": 1.815,
      "step": 220066
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.4403117895126343,
      "learning_rate": 2.9744096996316193e-06,
      "loss": 1.8243,
      "step": 220067
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.4548245668411255,
      "learning_rate": 2.9738351365187316e-06,
      "loss": 2.0117,
      "step": 220068
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.3832472562789917,
      "learning_rate": 2.973260628628604e-06,
      "loss": 1.7983,
      "step": 220069
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.5759299993515015,
      "learning_rate": 2.972686175961403e-06,
      "loss": 2.1191,
      "step": 220070
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.4674961566925049,
      "learning_rate": 2.9721117785171276e-06,
      "loss": 1.7255,
      "step": 220071
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.5189704895019531,
      "learning_rate": 2.971537436295979e-06,
      "loss": 1.7846,
      "step": 220072
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.4800673723220825,
      "learning_rate": 2.9709631492980223e-06,
      "loss": 2.0931,
      "step": 220073
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.4843995571136475,
      "learning_rate": 2.9703889175233587e-06,
      "loss": 2.1397,
      "step": 220074
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.5633777379989624,
      "learning_rate": 2.9698147409720872e-06,
      "loss": 1.9798,
      "step": 220075
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.3651193380355835,
      "learning_rate": 2.9692406196444086e-06,
      "loss": 1.7304,
      "step": 220076
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.5094671249389648,
      "learning_rate": 2.9686665535402886e-06,
      "loss": 2.1098,
      "step": 220077
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.4809157848358154,
      "learning_rate": 2.968092542659928e-06,
      "loss": 1.9019,
      "step": 220078
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.4405843019485474,
      "learning_rate": 2.967518587003359e-06,
      "loss": 1.9285,
      "step": 220079
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.4061603546142578,
      "learning_rate": 2.9669446865707825e-06,
      "loss": 1.7717,
      "step": 220080
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.4275782108306885,
      "learning_rate": 2.966370841362231e-06,
      "loss": 1.8415,
      "step": 220081
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.4970510005950928,
      "learning_rate": 2.965797051377905e-06,
      "loss": 2.0329,
      "step": 220082
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.6105105876922607,
      "learning_rate": 2.9652233166177707e-06,
      "loss": 2.0896,
      "step": 220083
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.3256691694259644,
      "learning_rate": 2.9646496370820283e-06,
      "loss": 2.1402,
      "step": 220084
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.5452566146850586,
      "learning_rate": 2.9640760127707775e-06,
      "loss": 2.2672,
      "step": 220085
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.4064741134643555,
      "learning_rate": 2.9635024436840847e-06,
      "loss": 1.9396,
      "step": 220086
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.4539241790771484,
      "learning_rate": 2.962928929822084e-06,
      "loss": 2.0282,
      "step": 220087
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.4216890335083008,
      "learning_rate": 2.9623554711849405e-06,
      "loss": 1.9301,
      "step": 220088
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.3876765966415405,
      "learning_rate": 2.9617820677726554e-06,
      "loss": 1.9038,
      "step": 220089
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.5100810527801514,
      "learning_rate": 2.961208719585395e-06,
      "loss": 1.9968,
      "step": 220090
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.444750428199768,
      "learning_rate": 2.9606354266232257e-06,
      "loss": 1.8716,
      "step": 220091
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.5384598970413208,
      "learning_rate": 2.960062188886314e-06,
      "loss": 2.0309,
      "step": 220092
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.4032814502716064,
      "learning_rate": 2.9594890063746934e-06,
      "loss": 1.8815,
      "step": 220093
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.4604190587997437,
      "learning_rate": 2.9589158790885635e-06,
      "loss": 2.2255,
      "step": 220094
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.4533213376998901,
      "learning_rate": 2.9583428070279246e-06,
      "loss": 2.1066,
      "step": 220095
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.5024518966674805,
      "learning_rate": 2.9577697901929763e-06,
      "loss": 2.0197,
      "step": 220096
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.4300131797790527,
      "learning_rate": 2.9571968285837854e-06,
      "loss": 1.8682,
      "step": 220097
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.4766839742660522,
      "learning_rate": 2.9566239222004516e-06,
      "loss": 1.949,
      "step": 220098
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.3846700191497803,
      "learning_rate": 2.956051071043075e-06,
      "loss": 1.9224,
      "step": 220099
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.3391051292419434,
      "learning_rate": 2.955478275111789e-06,
      "loss": 2.1199,
      "step": 220100
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.4877922534942627,
      "learning_rate": 2.9549055344066595e-06,
      "loss": 2.0095,
      "step": 220101
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.4573702812194824,
      "learning_rate": 2.954332848927854e-06,
      "loss": 1.8879,
      "step": 220102
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.5648812055587769,
      "learning_rate": 2.9537602186754384e-06,
      "loss": 1.9488,
      "step": 220103
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.4254357814788818,
      "learning_rate": 2.9531876436495126e-06,
      "loss": 2.071,
      "step": 220104
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.431922197341919,
      "learning_rate": 2.9526151238502104e-06,
      "loss": 2.0566,
      "step": 220105
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.452768325805664,
      "learning_rate": 2.9520426592776313e-06,
      "loss": 1.8465,
      "step": 220106
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.358830451965332,
      "learning_rate": 2.9514702499318756e-06,
      "loss": 1.993,
      "step": 220107
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.5940123796463013,
      "learning_rate": 2.9508978958130424e-06,
      "loss": 1.9032,
      "step": 220108
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.5343364477157593,
      "learning_rate": 2.950325596921199e-06,
      "loss": 2.041,
      "step": 220109
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.7150530815124512,
      "learning_rate": 2.9497533532565453e-06,
      "loss": 1.9359,
      "step": 220110
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.3724558353424072,
      "learning_rate": 2.9491811648191142e-06,
      "loss": 1.9362,
      "step": 220111
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.4880414009094238,
      "learning_rate": 2.948609031609073e-06,
      "loss": 2.0436,
      "step": 220112
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.3782848119735718,
      "learning_rate": 2.948036953626487e-06,
      "loss": 1.9426,
      "step": 220113
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.670328140258789,
      "learning_rate": 2.947464930871457e-06,
      "loss": 2.1824,
      "step": 220114
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.375013828277588,
      "learning_rate": 2.946892963344083e-06,
      "loss": 1.9854,
      "step": 220115
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.481411337852478,
      "learning_rate": 2.9463210510445314e-06,
      "loss": 1.832,
      "step": 220116
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.4701076745986938,
      "learning_rate": 2.9457491939728017e-06,
      "loss": 1.8289,
      "step": 220117
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.5085868835449219,
      "learning_rate": 2.9451773921291276e-06,
      "loss": 2.067,
      "step": 220118
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.3816226720809937,
      "learning_rate": 2.9446056455135093e-06,
      "loss": 1.9601,
      "step": 220119
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.4043374061584473,
      "learning_rate": 2.944033954126146e-06,
      "loss": 1.9117,
      "step": 220120
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.4651069641113281,
      "learning_rate": 2.943462317967038e-06,
      "loss": 1.9872,
      "step": 220121
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.423642635345459,
      "learning_rate": 2.942890737036385e-06,
      "loss": 1.7732,
      "step": 220122
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.402651309967041,
      "learning_rate": 2.942319211334221e-06,
      "loss": 1.8427,
      "step": 220123
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.4925271272659302,
      "learning_rate": 2.9417477408607114e-06,
      "loss": 2.0842,
      "step": 220124
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.5435125827789307,
      "learning_rate": 2.941176325615957e-06,
      "loss": 2.0298,
      "step": 220125
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.6633838415145874,
      "learning_rate": 2.940604965600024e-06,
      "loss": 1.8381,
      "step": 220126
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.515487790107727,
      "learning_rate": 2.940033660813046e-06,
      "loss": 2.2126,
      "step": 220127
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.4796035289764404,
      "learning_rate": 2.939462411255122e-06,
      "loss": 2.2195,
      "step": 220128
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.7224974632263184,
      "learning_rate": 2.93889121692632e-06,
      "loss": 2.0087,
      "step": 220129
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.6411871910095215,
      "learning_rate": 2.938320077826839e-06,
      "loss": 2.0188,
      "step": 220130
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.5110063552856445,
      "learning_rate": 2.9377489939567123e-06,
      "loss": 2.072,
      "step": 220131
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.5564302206039429,
      "learning_rate": 2.9371779653160733e-06,
      "loss": 2.1592,
      "step": 220132
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.5275577306747437,
      "learning_rate": 2.9366069919050217e-06,
      "loss": 1.8641,
      "step": 220133
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.668525218963623,
      "learning_rate": 2.9360360737236243e-06,
      "loss": 2.0926,
      "step": 220134
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.5424766540527344,
      "learning_rate": 2.935465210772048e-06,
      "loss": 1.9871,
      "step": 220135
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.4733506441116333,
      "learning_rate": 2.934894403050392e-06,
      "loss": 2.2146,
      "step": 220136
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.3740651607513428,
      "learning_rate": 2.9343236505587563e-06,
      "loss": 2.1824,
      "step": 220137
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.5085984468460083,
      "learning_rate": 2.9337529532972083e-06,
      "loss": 1.872,
      "step": 220138
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.6607766151428223,
      "learning_rate": 2.9331823112659137e-06,
      "loss": 2.0055,
      "step": 220139
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.8172693252563477,
      "learning_rate": 2.932611724464906e-06,
      "loss": 1.9567,
      "step": 220140
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.4389973878860474,
      "learning_rate": 2.932041192894352e-06,
      "loss": 1.9625,
      "step": 220141
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.4593793153762817,
      "learning_rate": 2.9314707165543516e-06,
      "loss": 1.8702,
      "step": 220142
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.490173101425171,
      "learning_rate": 2.9309002954450045e-06,
      "loss": 2.0974,
      "step": 220143
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.5540286302566528,
      "learning_rate": 2.9303299295663772e-06,
      "loss": 2.2321,
      "step": 220144
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.4649574756622314,
      "learning_rate": 2.9297596189186366e-06,
      "loss": 2.0471,
      "step": 220145
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.3991987705230713,
      "learning_rate": 2.929189363501849e-06,
      "loss": 2.0896,
      "step": 220146
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.4728201627731323,
      "learning_rate": 2.9286191633161482e-06,
      "loss": 1.8428,
      "step": 220147
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.435143232345581,
      "learning_rate": 2.9280490183616e-06,
      "loss": 2.3473,
      "step": 220148
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.7070893049240112,
      "learning_rate": 2.9274789286383716e-06,
      "loss": 2.0144,
      "step": 220149
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.476996660232544,
      "learning_rate": 2.9269088941464958e-06,
      "loss": 1.9995,
      "step": 220150
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.4303977489471436,
      "learning_rate": 2.9263389148861394e-06,
      "loss": 1.9742,
      "step": 220151
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.4399100542068481,
      "learning_rate": 2.925768990857336e-06,
      "loss": 2.0015,
      "step": 220152
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.4794124364852905,
      "learning_rate": 2.9251991220602846e-06,
      "loss": 1.9119,
      "step": 220153
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.46384596824646,
      "learning_rate": 2.924629308495052e-06,
      "loss": 2.1551,
      "step": 220154
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.5367571115493774,
      "learning_rate": 2.9240595501617394e-06,
      "loss": 1.7399,
      "step": 220155
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.5618042945861816,
      "learning_rate": 2.923489847060412e-06,
      "loss": 2.1206,
      "step": 220156
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.468858242034912,
      "learning_rate": 2.92292019919127e-06,
      "loss": 2.0728,
      "step": 220157
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.4220916032791138,
      "learning_rate": 2.9223506065543135e-06,
      "loss": 1.7594,
      "step": 220158
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.4561363458633423,
      "learning_rate": 2.9217810691497423e-06,
      "loss": 2.0147,
      "step": 220159
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.5674283504486084,
      "learning_rate": 2.92121158697759e-06,
      "loss": 2.0581,
      "step": 220160
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.5512709617614746,
      "learning_rate": 2.9206421600380225e-06,
      "loss": 1.9199,
      "step": 220161
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.5175344944000244,
      "learning_rate": 2.9200727883310736e-06,
      "loss": 1.9823,
      "step": 220162
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.440269947052002,
      "learning_rate": 2.91950347185691e-06,
      "loss": 1.9405,
      "step": 220163
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.6203452348709106,
      "learning_rate": 2.918934210615631e-06,
      "loss": 1.8839,
      "step": 220164
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.4105517864227295,
      "learning_rate": 2.9183650046073035e-06,
      "loss": 1.9059,
      "step": 220165
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.5957626104354858,
      "learning_rate": 2.9177958538320946e-06,
      "loss": 2.173,
      "step": 220166
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.4491677284240723,
      "learning_rate": 2.91722675829007e-06,
      "loss": 1.9259,
      "step": 220167
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.6271719932556152,
      "learning_rate": 2.91665771798133e-06,
      "loss": 1.9107,
      "step": 220168
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.465183138847351,
      "learning_rate": 2.9160887329059747e-06,
      "loss": 2.0277,
      "step": 220169
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.4207748174667358,
      "learning_rate": 2.9155198030641367e-06,
      "loss": 2.1293,
      "step": 220170
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.5197545289993286,
      "learning_rate": 2.914950928455917e-06,
      "loss": 2.017,
      "step": 220171
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.380793809890747,
      "learning_rate": 2.9143821090814145e-06,
      "loss": 2.1572,
      "step": 220172
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.4472676515579224,
      "learning_rate": 2.913813344940763e-06,
      "loss": 1.8054,
      "step": 220173
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.5333220958709717,
      "learning_rate": 2.9132446360339957e-06,
      "loss": 2.2553,
      "step": 220174
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.4317384958267212,
      "learning_rate": 2.912675982361279e-06,
      "loss": 1.8406,
      "step": 220175
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.5920265913009644,
      "learning_rate": 2.9121073839227126e-06,
      "loss": 1.7343,
      "step": 220176
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.5163780450820923,
      "learning_rate": 2.911538840718397e-06,
      "loss": 2.0689,
      "step": 220177
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.5213428735733032,
      "learning_rate": 2.9109703527483987e-06,
      "loss": 1.8744,
      "step": 220178
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.4728727340698242,
      "learning_rate": 2.910401920012917e-06,
      "loss": 2.0735,
      "step": 220179
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.5410863161087036,
      "learning_rate": 2.909833542511952e-06,
      "loss": 1.7966,
      "step": 220180
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.4103007316589355,
      "learning_rate": 2.9092652202457046e-06,
      "loss": 1.9499,
      "step": 220181
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.4846789836883545,
      "learning_rate": 2.9086969532141736e-06,
      "loss": 1.9262,
      "step": 220182
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.495375394821167,
      "learning_rate": 2.908128741417559e-06,
      "loss": 1.7956,
      "step": 220183
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.539901614189148,
      "learning_rate": 2.907560584855928e-06,
      "loss": 2.0404,
      "step": 220184
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.4627162218093872,
      "learning_rate": 2.906992483529413e-06,
      "loss": 1.9446,
      "step": 220185
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.532218337059021,
      "learning_rate": 2.9064244374380484e-06,
      "loss": 1.7189,
      "step": 220186
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.5429878234863281,
      "learning_rate": 2.905856446582e-06,
      "loss": 2.0025,
      "step": 220187
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.4962618350982666,
      "learning_rate": 2.9052885109613676e-06,
      "loss": 2.1254,
      "step": 220188
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.3664450645446777,
      "learning_rate": 2.9047206305762517e-06,
      "loss": 2.0193,
      "step": 220189
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.5448706150054932,
      "learning_rate": 2.9041528054267516e-06,
      "loss": 2.0762,
      "step": 220190
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.4526710510253906,
      "learning_rate": 2.903585035513001e-06,
      "loss": 2.0044,
      "step": 220191
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.4671201705932617,
      "learning_rate": 2.903017320835066e-06,
      "loss": 1.8167,
      "step": 220192
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.486696720123291,
      "learning_rate": 2.9024496613930805e-06,
      "loss": 2.0251,
      "step": 220193
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.396876573562622,
      "learning_rate": 2.9018820571871106e-06,
      "loss": 2.0933,
      "step": 220194
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.6331104040145874,
      "learning_rate": 2.9013145082172895e-06,
      "loss": 1.92,
      "step": 220195
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.441288948059082,
      "learning_rate": 2.9007470144837173e-06,
      "loss": 1.9949,
      "step": 220196
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.440675973892212,
      "learning_rate": 2.9001795759865277e-06,
      "loss": 1.8923,
      "step": 220197
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.507224678993225,
      "learning_rate": 2.89961219272582e-06,
      "loss": 2.0573,
      "step": 220198
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.5086811780929565,
      "learning_rate": 2.8990448647016607e-06,
      "loss": 2.0942,
      "step": 220199
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.6593552827835083,
      "learning_rate": 2.898477591914183e-06,
      "loss": 2.0851,
      "step": 220200
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.3982305526733398,
      "learning_rate": 2.897910374363488e-06,
      "loss": 2.0144,
      "step": 220201
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.3727234601974487,
      "learning_rate": 2.897343212049641e-06,
      "loss": 1.9335,
      "step": 220202
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.5490463972091675,
      "learning_rate": 2.896776104972842e-06,
      "loss": 2.1349,
      "step": 220203
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.4624392986297607,
      "learning_rate": 2.8962090531330918e-06,
      "loss": 1.706,
      "step": 220204
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.4884796142578125,
      "learning_rate": 2.8956420565305894e-06,
      "loss": 2.1871,
      "step": 220205
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.5134316682815552,
      "learning_rate": 2.8950751151653684e-06,
      "loss": 2.0518,
      "step": 220206
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.598261833190918,
      "learning_rate": 2.894508229037562e-06,
      "loss": 2.0576,
      "step": 220207
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.5320476293563843,
      "learning_rate": 2.89394139814727e-06,
      "loss": 1.7131,
      "step": 220208
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.4510583877563477,
      "learning_rate": 2.8933746224946263e-06,
      "loss": 2.0468,
      "step": 220209
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.4695931673049927,
      "learning_rate": 2.8928079020796966e-06,
      "loss": 1.8797,
      "step": 220210
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.6455022096633911,
      "learning_rate": 2.8922412369025815e-06,
      "loss": 2.2929,
      "step": 220211
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.4731613397598267,
      "learning_rate": 2.8916746269634473e-06,
      "loss": 1.8923,
      "step": 220212
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.5398304462432861,
      "learning_rate": 2.89110807226236e-06,
      "loss": 2.2129,
      "step": 220213
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.606765627861023,
      "learning_rate": 2.8905415727993875e-06,
      "loss": 2.0175,
      "step": 220214
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.4723001718521118,
      "learning_rate": 2.889975128574695e-06,
      "loss": 1.8007,
      "step": 220215
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.3924534320831299,
      "learning_rate": 2.8894087395883502e-06,
      "loss": 1.9561,
      "step": 220216
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.462518334388733,
      "learning_rate": 2.8888424058404856e-06,
      "loss": 1.9393,
      "step": 220217
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.525046706199646,
      "learning_rate": 2.8882761273312016e-06,
      "loss": 2.0196,
      "step": 220218
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.4908536672592163,
      "learning_rate": 2.887709904060598e-06,
      "loss": 1.8538,
      "step": 220219
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.4221707582473755,
      "learning_rate": 2.8871437360287408e-06,
      "loss": 1.8464,
      "step": 220220
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.5429710149765015,
      "learning_rate": 2.8865776232357973e-06,
      "loss": 2.2201,
      "step": 220221
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.4282991886138916,
      "learning_rate": 2.8860115656818674e-06,
      "loss": 1.9543,
      "step": 220222
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.3349684476852417,
      "learning_rate": 2.8854455633669835e-06,
      "loss": 2.1231,
      "step": 220223
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.4478271007537842,
      "learning_rate": 2.8848796162913466e-06,
      "loss": 1.9659,
      "step": 220224
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.475878357887268,
      "learning_rate": 2.8843137244549896e-06,
      "loss": 2.0701,
      "step": 220225
    },
    {
      "epoch": 2.87,
      "grad_norm": 2.7211267948150635,
      "learning_rate": 2.883747887858079e-06,
      "loss": 2.1614,
      "step": 220226
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.6071686744689941,
      "learning_rate": 2.883182106500681e-06,
      "loss": 2.0254,
      "step": 220227
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.6923578977584839,
      "learning_rate": 2.882616380382896e-06,
      "loss": 1.9191,
      "step": 220228
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.4809587001800537,
      "learning_rate": 2.882050709504824e-06,
      "loss": 1.9395,
      "step": 220229
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.6275739669799805,
      "learning_rate": 2.881485093866631e-06,
      "loss": 1.9392,
      "step": 220230
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.5046508312225342,
      "learning_rate": 2.880919533468351e-06,
      "loss": 1.9177,
      "step": 220231
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.471648097038269,
      "learning_rate": 2.8803540283101166e-06,
      "loss": 2.1645,
      "step": 220232
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.5349854230880737,
      "learning_rate": 2.8797885783920282e-06,
      "loss": 1.9422,
      "step": 220233
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.4630341529846191,
      "learning_rate": 2.879223183714219e-06,
      "loss": 1.9161,
      "step": 220234
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.5750370025634766,
      "learning_rate": 2.878657844276755e-06,
      "loss": 1.9905,
      "step": 220235
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.4312225580215454,
      "learning_rate": 2.87809256007977e-06,
      "loss": 1.9362,
      "step": 220236
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.5740927457809448,
      "learning_rate": 2.8775273311233302e-06,
      "loss": 1.9963,
      "step": 220237
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.485521912574768,
      "learning_rate": 2.876962157407603e-06,
      "loss": 2.0303,
      "step": 220238
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.4768868684768677,
      "learning_rate": 2.8763970389326207e-06,
      "loss": 2.0519,
      "step": 220239
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.401193380355835,
      "learning_rate": 2.8758319756985503e-06,
      "loss": 2.1126,
      "step": 220240
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.4168187379837036,
      "learning_rate": 2.8752669677054585e-06,
      "loss": 2.0454,
      "step": 220241
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.4025657176971436,
      "learning_rate": 2.8747020149534784e-06,
      "loss": 1.9818,
      "step": 220242
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.6844784021377563,
      "learning_rate": 2.8741371174426764e-06,
      "loss": 2.0495,
      "step": 220243
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.5082110166549683,
      "learning_rate": 2.873572275173186e-06,
      "loss": 1.9498,
      "step": 220244
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.4764435291290283,
      "learning_rate": 2.8730074881451072e-06,
      "loss": 1.8185,
      "step": 220245
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.5386152267456055,
      "learning_rate": 2.8724427563585727e-06,
      "loss": 1.9556,
      "step": 220246
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.5185346603393555,
      "learning_rate": 2.8718780798136164e-06,
      "loss": 2.1374,
      "step": 220247
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.601674199104309,
      "learning_rate": 2.871313458510438e-06,
      "loss": 1.963,
      "step": 220248
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.654344916343689,
      "learning_rate": 2.8707488924490374e-06,
      "loss": 1.9427,
      "step": 220249
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.7891333103179932,
      "learning_rate": 2.8701843816296145e-06,
      "loss": 2.0237,
      "step": 220250
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.6086053848266602,
      "learning_rate": 2.8696199260522023e-06,
      "loss": 1.8951,
      "step": 220251
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.5473579168319702,
      "learning_rate": 2.8690555257169678e-06,
      "loss": 2.0087,
      "step": 220252
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.6373275518417358,
      "learning_rate": 2.8684911806239443e-06,
      "loss": 2.152,
      "step": 220253
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.5178704261779785,
      "learning_rate": 2.867926890773331e-06,
      "loss": 2.0096,
      "step": 220254
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.5456550121307373,
      "learning_rate": 2.8673626561651286e-06,
      "loss": 2.2027,
      "step": 220255
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.4630534648895264,
      "learning_rate": 2.866798476799503e-06,
      "loss": 2.0494,
      "step": 220256
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.4513427019119263,
      "learning_rate": 2.866234352676555e-06,
      "loss": 1.9615,
      "step": 220257
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.4909677505493164,
      "learning_rate": 2.8656702837964175e-06,
      "loss": 1.9389,
      "step": 220258
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.6272521018981934,
      "learning_rate": 2.865106270159123e-06,
      "loss": 2.1521,
      "step": 220259
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.4751944541931152,
      "learning_rate": 2.8645423117648057e-06,
      "loss": 2.2123,
      "step": 220260
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.429582953453064,
      "learning_rate": 2.8639784086135988e-06,
      "loss": 2.0882,
      "step": 220261
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.6140316724777222,
      "learning_rate": 2.863414560705568e-06,
      "loss": 1.9328,
      "step": 220262
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.4714632034301758,
      "learning_rate": 2.8628507680408476e-06,
      "loss": 1.8885,
      "step": 220263
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.581560492515564,
      "learning_rate": 2.8622870306195367e-06,
      "loss": 2.0665,
      "step": 220264
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.474387526512146,
      "learning_rate": 2.8617233484417358e-06,
      "loss": 1.9263,
      "step": 220265
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.4374250173568726,
      "learning_rate": 2.861159721507578e-06,
      "loss": 2.0274,
      "step": 220266
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.5239434242248535,
      "learning_rate": 2.860596149817096e-06,
      "loss": 2.0125,
      "step": 220267
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.3772145509719849,
      "learning_rate": 2.860032633370457e-06,
      "loss": 2.1447,
      "step": 220268
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.3716925382614136,
      "learning_rate": 2.8594691721677276e-06,
      "loss": 1.879,
      "step": 220269
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.4631584882736206,
      "learning_rate": 2.8589057662090408e-06,
      "loss": 2.0646,
      "step": 220270
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.532933235168457,
      "learning_rate": 2.8583424154944966e-06,
      "loss": 2.0428,
      "step": 220271
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.6839566230773926,
      "learning_rate": 2.857779120024195e-06,
      "loss": 2.0771,
      "step": 220272
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.407455563545227,
      "learning_rate": 2.857215879798236e-06,
      "loss": 2.0938,
      "step": 220273
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.4960274696350098,
      "learning_rate": 2.856652694816752e-06,
      "loss": 2.0965,
      "step": 220274
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.592777967453003,
      "learning_rate": 2.8560895650797776e-06,
      "loss": 1.9458,
      "step": 220275
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.4124776124954224,
      "learning_rate": 2.855526490587512e-06,
      "loss": 1.9768,
      "step": 220276
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.3100576400756836,
      "learning_rate": 2.8549634713400214e-06,
      "loss": 1.8565,
      "step": 220277
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.3703616857528687,
      "learning_rate": 2.854400507337373e-06,
      "loss": 1.9815,
      "step": 220278
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.3706142902374268,
      "learning_rate": 2.8538375985797004e-06,
      "loss": 1.7326,
      "step": 220279
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.4784066677093506,
      "learning_rate": 2.8532747450671024e-06,
      "loss": 2.1542,
      "step": 220280
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.5462851524353027,
      "learning_rate": 2.8527119467997128e-06,
      "loss": 2.0344,
      "step": 220281
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.9972455501556396,
      "learning_rate": 2.8521492037775983e-06,
      "loss": 1.6778,
      "step": 220282
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.713320255279541,
      "learning_rate": 2.851586516000892e-06,
      "loss": 1.9017,
      "step": 220283
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.5410264730453491,
      "learning_rate": 2.851023883469661e-06,
      "loss": 1.8973,
      "step": 220284
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.593822717666626,
      "learning_rate": 2.8504613061840375e-06,
      "loss": 1.9788,
      "step": 220285
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.4881188869476318,
      "learning_rate": 2.8498987841441555e-06,
      "loss": 2.0344,
      "step": 220286
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.3965785503387451,
      "learning_rate": 2.849336317350048e-06,
      "loss": 1.9078,
      "step": 220287
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.6275936365127563,
      "learning_rate": 2.848773905801882e-06,
      "loss": 2.0709,
      "step": 220288
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.3649494647979736,
      "learning_rate": 2.8482115494997236e-06,
      "loss": 1.9681,
      "step": 220289
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.5259743928909302,
      "learning_rate": 2.8476492484437063e-06,
      "loss": 1.7562,
      "step": 220290
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.4607632160186768,
      "learning_rate": 2.8470870026338965e-06,
      "loss": 2.044,
      "step": 220291
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.5343655347824097,
      "learning_rate": 2.846524812070461e-06,
      "loss": 1.9989,
      "step": 220292
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.5665841102600098,
      "learning_rate": 2.8459626767534325e-06,
      "loss": 1.9951,
      "step": 220293
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.5861190557479858,
      "learning_rate": 2.8454005966829452e-06,
      "loss": 1.9124,
      "step": 220294
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.343358039855957,
      "learning_rate": 2.8448385718591317e-06,
      "loss": 2.014,
      "step": 220295
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.4865490198135376,
      "learning_rate": 2.8442766022820586e-06,
      "loss": 2.0508,
      "step": 220296
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.7796523571014404,
      "learning_rate": 2.8437146879518257e-06,
      "loss": 1.9323,
      "step": 220297
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.4440494775772095,
      "learning_rate": 2.8431528288686002e-06,
      "loss": 1.9027,
      "step": 220298
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.7081546783447266,
      "learning_rate": 2.8425910250324147e-06,
      "loss": 1.7443,
      "step": 220299
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.523606538772583,
      "learning_rate": 2.8420292764434026e-06,
      "loss": 2.1147,
      "step": 220300
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.536220908164978,
      "learning_rate": 2.8414675831016643e-06,
      "loss": 1.818,
      "step": 220301
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.5080907344818115,
      "learning_rate": 2.840905945007299e-06,
      "loss": 1.8905,
      "step": 220302
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.4636765718460083,
      "learning_rate": 2.8403443621604407e-06,
      "loss": 2.0175,
      "step": 220303
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.4629106521606445,
      "learning_rate": 2.839782834561155e-06,
      "loss": 1.8316,
      "step": 220304
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.6239492893218994,
      "learning_rate": 2.839221362209576e-06,
      "loss": 1.9467,
      "step": 220305
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.3103654384613037,
      "learning_rate": 2.83865994510577e-06,
      "loss": 1.9525,
      "step": 220306
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.4915953874588013,
      "learning_rate": 2.8380985832499035e-06,
      "loss": 1.8823,
      "step": 220307
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.6237627267837524,
      "learning_rate": 2.83753727664201e-06,
      "loss": 1.954,
      "step": 220308
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.482043743133545,
      "learning_rate": 2.836976025282256e-06,
      "loss": 1.9882,
      "step": 220309
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.5602542161941528,
      "learning_rate": 2.8364148291706746e-06,
      "loss": 1.8262,
      "step": 220310
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.4338598251342773,
      "learning_rate": 2.8358536883074657e-06,
      "loss": 2.0244,
      "step": 220311
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.5775948762893677,
      "learning_rate": 2.8352926026926627e-06,
      "loss": 1.9775,
      "step": 220312
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.3029612302780151,
      "learning_rate": 2.834731572326365e-06,
      "loss": 1.8079,
      "step": 220313
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.5001693964004517,
      "learning_rate": 2.834170597208707e-06,
      "loss": 1.9401,
      "step": 220314
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.5717041492462158,
      "learning_rate": 2.833609677339821e-06,
      "loss": 1.906,
      "step": 220315
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.6109050512313843,
      "learning_rate": 2.8330488127197403e-06,
      "loss": 2.0646,
      "step": 220316
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.4444698095321655,
      "learning_rate": 2.8324880033486318e-06,
      "loss": 1.6685,
      "step": 220317
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.3949434757232666,
      "learning_rate": 2.8319272492265623e-06,
      "loss": 1.9411,
      "step": 220318
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.4890044927597046,
      "learning_rate": 2.831366550353631e-06,
      "loss": 2.0757,
      "step": 220319
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.4457566738128662,
      "learning_rate": 2.8308059067299714e-06,
      "loss": 1.954,
      "step": 220320
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.489344835281372,
      "learning_rate": 2.830245318355684e-06,
      "loss": 2.1679,
      "step": 220321
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.5879850387573242,
      "learning_rate": 2.8296847852308346e-06,
      "loss": 1.9497,
      "step": 220322
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.3980706930160522,
      "learning_rate": 2.8291243073555903e-06,
      "loss": 2.1309,
      "step": 220323
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.4327418804168701,
      "learning_rate": 2.828563884729984e-06,
      "loss": 1.9811,
      "step": 220324
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.3819409608840942,
      "learning_rate": 2.828003517354216e-06,
      "loss": 1.9102,
      "step": 220325
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.3928534984588623,
      "learning_rate": 2.8274432052282527e-06,
      "loss": 1.7699,
      "step": 220326
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.4668316841125488,
      "learning_rate": 2.826882948352327e-06,
      "loss": 1.9718,
      "step": 220327
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.4454280138015747,
      "learning_rate": 2.8263227467264727e-06,
      "loss": 1.9896,
      "step": 220328
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.501880407333374,
      "learning_rate": 2.825762600350856e-06,
      "loss": 1.8511,
      "step": 220329
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.4661555290222168,
      "learning_rate": 2.825202509225477e-06,
      "loss": 2.1535,
      "step": 220330
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.55380380153656,
      "learning_rate": 2.824642473350569e-06,
      "loss": 2.1963,
      "step": 220331
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.5676733255386353,
      "learning_rate": 2.824082492726132e-06,
      "loss": 2.2722,
      "step": 220332
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.4588154554367065,
      "learning_rate": 2.823522567352332e-06,
      "loss": 2.0308,
      "step": 220333
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.6447880268096924,
      "learning_rate": 2.822962697229203e-06,
      "loss": 1.954,
      "step": 220334
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.4380441904067993,
      "learning_rate": 2.8224028823569443e-06,
      "loss": 1.9889,
      "step": 220335
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.5489650964736938,
      "learning_rate": 2.8218431227355564e-06,
      "loss": 2.0208,
      "step": 220336
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.4159830808639526,
      "learning_rate": 2.8212834183652387e-06,
      "loss": 2.1259,
      "step": 220337
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.4487500190734863,
      "learning_rate": 2.820723769246058e-06,
      "loss": 1.79,
      "step": 220338
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.4885388612747192,
      "learning_rate": 2.8201641753781145e-06,
      "loss": 1.9175,
      "step": 220339
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.6170392036437988,
      "learning_rate": 2.8196046367614745e-06,
      "loss": 1.862,
      "step": 220340
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.3375558853149414,
      "learning_rate": 2.8190451533963045e-06,
      "loss": 1.8984,
      "step": 220341
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.5245516300201416,
      "learning_rate": 2.818485725282671e-06,
      "loss": 2.1725,
      "step": 220342
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.5528841018676758,
      "learning_rate": 2.8179263524207073e-06,
      "loss": 2.1569,
      "step": 220343
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.5736284255981445,
      "learning_rate": 2.817367034810514e-06,
      "loss": 1.8658,
      "step": 220344
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.5540763139724731,
      "learning_rate": 2.8168077724521565e-06,
      "loss": 1.9271,
      "step": 220345
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.512615442276001,
      "learning_rate": 2.816248565345769e-06,
      "loss": 1.9816,
      "step": 220346
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.4419560432434082,
      "learning_rate": 2.815689413491451e-06,
      "loss": 1.9928,
      "step": 220347
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.4114408493041992,
      "learning_rate": 2.8151303168893027e-06,
      "loss": 1.9387,
      "step": 220348
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.3886233568191528,
      "learning_rate": 2.814571275539457e-06,
      "loss": 2.0087,
      "step": 220349
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.3846465349197388,
      "learning_rate": 2.81401228944198e-06,
      "loss": 2.1483,
      "step": 220350
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.4929440021514893,
      "learning_rate": 2.813453358596973e-06,
      "loss": 1.927,
      "step": 220351
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.4624823331832886,
      "learning_rate": 2.8128944830045352e-06,
      "loss": 1.9466,
      "step": 220352
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.6419466733932495,
      "learning_rate": 2.812335662664833e-06,
      "loss": 2.0242,
      "step": 220353
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.445642352104187,
      "learning_rate": 2.8117768975779e-06,
      "loss": 2.0119,
      "step": 220354
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.3594779968261719,
      "learning_rate": 2.8112181877439023e-06,
      "loss": 1.8127,
      "step": 220355
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.548875331878662,
      "learning_rate": 2.8106595331628735e-06,
      "loss": 1.9436,
      "step": 220356
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.4975674152374268,
      "learning_rate": 2.81010093383498e-06,
      "loss": 2.0291,
      "step": 220357
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.4890121221542358,
      "learning_rate": 2.8095423897602556e-06,
      "loss": 2.0171,
      "step": 220358
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.4294883012771606,
      "learning_rate": 2.8089839009388993e-06,
      "loss": 1.9485,
      "step": 220359
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.3493239879608154,
      "learning_rate": 2.8084254673709117e-06,
      "loss": 1.9575,
      "step": 220360
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.5406906604766846,
      "learning_rate": 2.8078670890564926e-06,
      "loss": 1.9617,
      "step": 220361
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.3966326713562012,
      "learning_rate": 2.807308765995675e-06,
      "loss": 2.0016,
      "step": 220362
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.545926809310913,
      "learning_rate": 2.806750498188626e-06,
      "loss": 2.2224,
      "step": 220363
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.6445900201797485,
      "learning_rate": 2.806192285635378e-06,
      "loss": 1.9062,
      "step": 220364
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.5668526887893677,
      "learning_rate": 2.805634128336065e-06,
      "loss": 1.9088,
      "step": 220365
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.4934589862823486,
      "learning_rate": 2.805076026290787e-06,
      "loss": 1.9803,
      "step": 220366
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.390247106552124,
      "learning_rate": 2.8045179794996763e-06,
      "loss": 2.0128,
      "step": 220367
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.4923933744430542,
      "learning_rate": 2.8039599879628338e-06,
      "loss": 2.1372,
      "step": 220368
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.5118799209594727,
      "learning_rate": 2.803402051680326e-06,
      "loss": 1.8709,
      "step": 220369
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.4462759494781494,
      "learning_rate": 2.802844170652252e-06,
      "loss": 1.9234,
      "step": 220370
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.3754743337631226,
      "learning_rate": 2.802286344878779e-06,
      "loss": 2.0511,
      "step": 220371
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.4888699054718018,
      "learning_rate": 2.8017285743599406e-06,
      "loss": 1.9482,
      "step": 220372
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.4753938913345337,
      "learning_rate": 2.801170859095869e-06,
      "loss": 1.9567,
      "step": 220373
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.63520085811615,
      "learning_rate": 2.8006131990866985e-06,
      "loss": 2.1848,
      "step": 220374
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.420375943183899,
      "learning_rate": 2.8000555943324953e-06,
      "loss": 1.9892,
      "step": 220375
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.4962941408157349,
      "learning_rate": 2.7994980448333594e-06,
      "loss": 2.0131,
      "step": 220376
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.4947584867477417,
      "learning_rate": 2.7989405505894235e-06,
      "loss": 2.0647,
      "step": 220377
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.6042729616165161,
      "learning_rate": 2.7983831116007547e-06,
      "loss": 1.9809,
      "step": 220378
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.3839603662490845,
      "learning_rate": 2.7978257278674866e-06,
      "loss": 1.8265,
      "step": 220379
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.4574449062347412,
      "learning_rate": 2.7972683993897183e-06,
      "loss": 2.028,
      "step": 220380
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.448612093925476,
      "learning_rate": 2.79671112616755e-06,
      "loss": 1.9542,
      "step": 220381
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.661794662475586,
      "learning_rate": 2.796153908201049e-06,
      "loss": 2.1612,
      "step": 220382
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.5173529386520386,
      "learning_rate": 2.7955967454903803e-06,
      "loss": 2.0368,
      "step": 220383
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.3931976556777954,
      "learning_rate": 2.795039638035612e-06,
      "loss": 1.8397,
      "step": 220384
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.4164588451385498,
      "learning_rate": 2.794482585836877e-06,
      "loss": 1.759,
      "step": 220385
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.6113077402114868,
      "learning_rate": 2.793925588894241e-06,
      "loss": 1.9682,
      "step": 220386
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.5284702777862549,
      "learning_rate": 2.7933686472078052e-06,
      "loss": 2.0769,
      "step": 220387
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.3937945365905762,
      "learning_rate": 2.7928117607777356e-06,
      "loss": 2.1576,
      "step": 220388
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.502634882926941,
      "learning_rate": 2.792254929604065e-06,
      "loss": 1.828,
      "step": 220389
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.541332483291626,
      "learning_rate": 2.7916981536869275e-06,
      "loss": 1.8004,
      "step": 220390
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.5290471315383911,
      "learning_rate": 2.7911414330264225e-06,
      "loss": 2.171,
      "step": 220391
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.540280818939209,
      "learning_rate": 2.7905847676226835e-06,
      "loss": 1.9338,
      "step": 220392
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.7714635133743286,
      "learning_rate": 2.790028157475743e-06,
      "loss": 1.9066,
      "step": 220393
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.404893398284912,
      "learning_rate": 2.789471602585769e-06,
      "loss": 1.8902,
      "step": 220394
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.5075775384902954,
      "learning_rate": 2.7889151029528266e-06,
      "loss": 1.8643,
      "step": 220395
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.3830461502075195,
      "learning_rate": 2.78835865857705e-06,
      "loss": 1.8987,
      "step": 220396
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.5662493705749512,
      "learning_rate": 2.787802269458539e-06,
      "loss": 2.0585,
      "step": 220397
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.4828604459762573,
      "learning_rate": 2.78724593559736e-06,
      "loss": 1.7746,
      "step": 220398
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.6531951427459717,
      "learning_rate": 2.786689656993646e-06,
      "loss": 1.8315,
      "step": 220399
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.4124938249588013,
      "learning_rate": 2.7861334336475304e-06,
      "loss": 2.1231,
      "step": 220400
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.4373655319213867,
      "learning_rate": 2.785577265559047e-06,
      "loss": 2.039,
      "step": 220401
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.3321090936660767,
      "learning_rate": 2.7850211527283286e-06,
      "loss": 1.9827,
      "step": 220402
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.577087640762329,
      "learning_rate": 2.784465095155508e-06,
      "loss": 1.7515,
      "step": 220403
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.5179580450057983,
      "learning_rate": 2.7839090928406528e-06,
      "loss": 2.0991,
      "step": 220404
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.4049502611160278,
      "learning_rate": 2.783353145783895e-06,
      "loss": 2.2221,
      "step": 220405
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.440233588218689,
      "learning_rate": 2.7827972539853026e-06,
      "loss": 2.0208,
      "step": 220406
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.4627360105514526,
      "learning_rate": 2.7822414174450083e-06,
      "loss": 1.9089,
      "step": 220407
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.5053813457489014,
      "learning_rate": 2.7816856361631114e-06,
      "loss": 1.9467,
      "step": 220408
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.4541455507278442,
      "learning_rate": 2.7811299101397124e-06,
      "loss": 1.9334,
      "step": 220409
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.4063332080841064,
      "learning_rate": 2.7805742393748777e-06,
      "loss": 2.0348,
      "step": 220410
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.4694535732269287,
      "learning_rate": 2.780018623868807e-06,
      "loss": 2.1,
      "step": 220411
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.5904289484024048,
      "learning_rate": 2.779463063621501e-06,
      "loss": 1.9864,
      "step": 220412
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.4284663200378418,
      "learning_rate": 2.778907558633092e-06,
      "loss": 1.9849,
      "step": 220413
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.512837290763855,
      "learning_rate": 2.778352108903714e-06,
      "loss": 1.9709,
      "step": 220414
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.4032742977142334,
      "learning_rate": 2.7777967144334336e-06,
      "loss": 1.7826,
      "step": 220415
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.4147840738296509,
      "learning_rate": 2.777241375222383e-06,
      "loss": 1.8822,
      "step": 220416
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.504606008529663,
      "learning_rate": 2.7766860912706635e-06,
      "loss": 1.9648,
      "step": 220417
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.5404112339019775,
      "learning_rate": 2.776130862578374e-06,
      "loss": 2.1376,
      "step": 220418
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.521308183670044,
      "learning_rate": 2.7755756891455816e-06,
      "loss": 1.975,
      "step": 220419
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.652247428894043,
      "learning_rate": 2.7750205709724527e-06,
      "loss": 1.9377,
      "step": 220420
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.3807252645492554,
      "learning_rate": 2.7744655080590205e-06,
      "loss": 1.8755,
      "step": 220421
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.5294804573059082,
      "learning_rate": 2.773910500405452e-06,
      "loss": 2.0356,
      "step": 220422
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.582830786705017,
      "learning_rate": 2.7733555480118463e-06,
      "loss": 1.8699,
      "step": 220423
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.4376639127731323,
      "learning_rate": 2.7728006508782376e-06,
      "loss": 1.7557,
      "step": 220424
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.5540196895599365,
      "learning_rate": 2.7722458090047916e-06,
      "loss": 1.9815,
      "step": 220425
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.6248410940170288,
      "learning_rate": 2.771691022391609e-06,
      "loss": 2.1132,
      "step": 220426
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.4616221189498901,
      "learning_rate": 2.7711362910387556e-06,
      "loss": 1.8336,
      "step": 220427
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.4924843311309814,
      "learning_rate": 2.770581614946399e-06,
      "loss": 2.1267,
      "step": 220428
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.5802350044250488,
      "learning_rate": 2.7700269941145714e-06,
      "loss": 2.0323,
      "step": 220429
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.6436518430709839,
      "learning_rate": 2.7694724285434066e-06,
      "loss": 2.0133,
      "step": 220430
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.379224419593811,
      "learning_rate": 2.7689179182330046e-06,
      "loss": 1.9726,
      "step": 220431
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.3808261156082153,
      "learning_rate": 2.7683634631834984e-06,
      "loss": 1.8437,
      "step": 220432
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.4991692304611206,
      "learning_rate": 2.7678090633949214e-06,
      "loss": 2.0118,
      "step": 220433
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.4646178483963013,
      "learning_rate": 2.7672547188674397e-06,
      "loss": 2.1827,
      "step": 220434
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.4778294563293457,
      "learning_rate": 2.766700429601154e-06,
      "loss": 2.0128,
      "step": 220435
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.5018311738967896,
      "learning_rate": 2.7661461955961304e-06,
      "loss": 2.0964,
      "step": 220436
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.5683200359344482,
      "learning_rate": 2.7655920168525023e-06,
      "loss": 2.0545,
      "step": 220437
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.7182255983352661,
      "learning_rate": 2.7650378933703697e-06,
      "loss": 1.8975,
      "step": 220438
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.4327447414398193,
      "learning_rate": 2.764483825149799e-06,
      "loss": 2.2103,
      "step": 220439
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.3510563373565674,
      "learning_rate": 2.763929812190924e-06,
      "loss": 1.7158,
      "step": 220440
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.4735842943191528,
      "learning_rate": 2.7633758544938765e-06,
      "loss": 2.1045,
      "step": 220441
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.8090285062789917,
      "learning_rate": 2.7628219520586914e-06,
      "loss": 2.0013,
      "step": 220442
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.498452067375183,
      "learning_rate": 2.7622681048855345e-06,
      "loss": 1.9999,
      "step": 220443
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.7256518602371216,
      "learning_rate": 2.7617143129744724e-06,
      "loss": 1.8805,
      "step": 220444
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.641129732131958,
      "learning_rate": 2.761160576325605e-06,
      "loss": 2.0428,
      "step": 220445
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.4931633472442627,
      "learning_rate": 2.7606068949390657e-06,
      "loss": 1.7763,
      "step": 220446
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.5395526885986328,
      "learning_rate": 2.7600532688149545e-06,
      "loss": 2.0456,
      "step": 220447
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.6230324506759644,
      "learning_rate": 2.7594996979533378e-06,
      "loss": 1.886,
      "step": 220448
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.446612000465393,
      "learning_rate": 2.758946182354349e-06,
      "loss": 1.9807,
      "step": 220449
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.5692757368087769,
      "learning_rate": 2.7583927220180877e-06,
      "loss": 2.0093,
      "step": 220450
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.384530782699585,
      "learning_rate": 2.7578393169446213e-06,
      "loss": 2.1193,
      "step": 220451
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.5845997333526611,
      "learning_rate": 2.7572859671341153e-06,
      "loss": 1.8647,
      "step": 220452
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.3514463901519775,
      "learning_rate": 2.756732672586637e-06,
      "loss": 2.0296,
      "step": 220453
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.4026408195495605,
      "learning_rate": 2.756179433302286e-06,
      "loss": 1.9376,
      "step": 220454
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.5118401050567627,
      "learning_rate": 2.755626249281162e-06,
      "loss": 1.811,
      "step": 220455
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.5853345394134521,
      "learning_rate": 2.755073120523399e-06,
      "loss": 2.0662,
      "step": 220456
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.6274607181549072,
      "learning_rate": 2.754520047029063e-06,
      "loss": 2.1238,
      "step": 220457
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.5461547374725342,
      "learning_rate": 2.7539670287982875e-06,
      "loss": 2.0879,
      "step": 220458
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.4072259664535522,
      "learning_rate": 2.753414065831172e-06,
      "loss": 1.9632,
      "step": 220459
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.5453827381134033,
      "learning_rate": 2.7528611581277837e-06,
      "loss": 1.9547,
      "step": 220460
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.3693541288375854,
      "learning_rate": 2.752308305688222e-06,
      "loss": 1.9049,
      "step": 220461
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.5128787755966187,
      "learning_rate": 2.7517555085126875e-06,
      "loss": 2.2745,
      "step": 220462
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.5315266847610474,
      "learning_rate": 2.751202766601146e-06,
      "loss": 2.194,
      "step": 220463
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.407677412033081,
      "learning_rate": 2.7506500799537977e-06,
      "loss": 1.8349,
      "step": 220464
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.4982647895812988,
      "learning_rate": 2.7500974485707094e-06,
      "loss": 1.991,
      "step": 220465
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.430259346961975,
      "learning_rate": 2.7495448724520143e-06,
      "loss": 1.7022,
      "step": 220466
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.4832431077957153,
      "learning_rate": 2.7489923515977454e-06,
      "loss": 2.0224,
      "step": 220467
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.419381022453308,
      "learning_rate": 2.7484398860081026e-06,
      "loss": 1.9237,
      "step": 220468
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.4772446155548096,
      "learning_rate": 2.747887475683086e-06,
      "loss": 2.0557,
      "step": 220469
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.432426929473877,
      "learning_rate": 2.747335120622862e-06,
      "loss": 2.1323,
      "step": 220470
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.3779361248016357,
      "learning_rate": 2.746782820827531e-06,
      "loss": 1.9723,
      "step": 220471
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.3798696994781494,
      "learning_rate": 2.746230576297159e-06,
      "loss": 1.734,
      "step": 220472
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.7408406734466553,
      "learning_rate": 2.745678387031913e-06,
      "loss": 2.1524,
      "step": 220473
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.4937838315963745,
      "learning_rate": 2.7451262530318264e-06,
      "loss": 1.9279,
      "step": 220474
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.5224522352218628,
      "learning_rate": 2.7445741742970317e-06,
      "loss": 1.8101,
      "step": 220475
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.4598714113235474,
      "learning_rate": 2.7440221508276626e-06,
      "loss": 1.8191,
      "step": 220476
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.452324390411377,
      "learning_rate": 2.7434701826237526e-06,
      "loss": 1.9605,
      "step": 220477
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.407867193222046,
      "learning_rate": 2.742918269685468e-06,
      "loss": 2.0169,
      "step": 220478
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.4618303775787354,
      "learning_rate": 2.7423664120128752e-06,
      "loss": 1.8791,
      "step": 220479
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.3980026245117188,
      "learning_rate": 2.7418146096060746e-06,
      "loss": 1.7317,
      "step": 220480
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.4033055305480957,
      "learning_rate": 2.741262862465232e-06,
      "loss": 2.0914,
      "step": 220481
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.6836631298065186,
      "learning_rate": 2.7407111705903484e-06,
      "loss": 1.9975,
      "step": 220482
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.5655794143676758,
      "learning_rate": 2.7401595339815897e-06,
      "loss": 2.0444,
      "step": 220483
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.52909255027771,
      "learning_rate": 2.739607952639056e-06,
      "loss": 2.137,
      "step": 220484
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.38142991065979,
      "learning_rate": 2.739056426562847e-06,
      "loss": 1.8432,
      "step": 220485
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.5202587842941284,
      "learning_rate": 2.7385049557530292e-06,
      "loss": 2.0374,
      "step": 220486
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.5056041479110718,
      "learning_rate": 2.7379535402097697e-06,
      "loss": 2.1445,
      "step": 220487
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.4493718147277832,
      "learning_rate": 2.737402179933135e-06,
      "loss": 2.0803,
      "step": 220488
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.5454415082931519,
      "learning_rate": 2.736850874923191e-06,
      "loss": 1.925,
      "step": 220489
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.3529691696166992,
      "learning_rate": 2.7362996251801052e-06,
      "loss": 1.8826,
      "step": 220490
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.4026479721069336,
      "learning_rate": 2.7357484307039433e-06,
      "loss": 1.9359,
      "step": 220491
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.6705704927444458,
      "learning_rate": 2.7351972914948393e-06,
      "loss": 1.9357,
      "step": 220492
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.567899465560913,
      "learning_rate": 2.7346462075528597e-06,
      "loss": 2.2304,
      "step": 220493
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.4631860256195068,
      "learning_rate": 2.734095178878104e-06,
      "loss": 1.9095,
      "step": 220494
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.4330471754074097,
      "learning_rate": 2.733544205470706e-06,
      "loss": 1.8987,
      "step": 220495
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.372317910194397,
      "learning_rate": 2.732993287330765e-06,
      "loss": 1.6865,
      "step": 220496
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.6129109859466553,
      "learning_rate": 2.7324424244583478e-06,
      "loss": 2.0106,
      "step": 220497
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.4835933446884155,
      "learning_rate": 2.7318916168535874e-06,
      "loss": 1.9556,
      "step": 220498
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.4660812616348267,
      "learning_rate": 2.7313408645165847e-06,
      "loss": 1.9663,
      "step": 220499
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.3660627603530884,
      "learning_rate": 2.730790167447405e-06,
      "loss": 2.1728,
      "step": 220500
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.4011567831039429,
      "learning_rate": 2.7302395256462496e-06,
      "loss": 1.95,
      "step": 220501
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.352587342262268,
      "learning_rate": 2.729688939113117e-06,
      "loss": 1.9132,
      "step": 220502
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.5446140766143799,
      "learning_rate": 2.7291384078481416e-06,
      "loss": 2.0122,
      "step": 220503
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.4183845520019531,
      "learning_rate": 2.7285879318514227e-06,
      "loss": 1.8704,
      "step": 220504
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.4659998416900635,
      "learning_rate": 2.7280375111230936e-06,
      "loss": 1.9502,
      "step": 220505
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.4907971620559692,
      "learning_rate": 2.727487145663221e-06,
      "loss": 1.9487,
      "step": 220506
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.5209715366363525,
      "learning_rate": 2.726936835471938e-06,
      "loss": 1.8623,
      "step": 220507
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.6127225160598755,
      "learning_rate": 2.7263865805493113e-06,
      "loss": 2.0436,
      "step": 220508
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.4837424755096436,
      "learning_rate": 2.725836380895474e-06,
      "loss": 2.3105,
      "step": 220509
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.5748556852340698,
      "learning_rate": 2.7252862365104932e-06,
      "loss": 2.0683,
      "step": 220510
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.4348961114883423,
      "learning_rate": 2.7247361473945017e-06,
      "loss": 1.6172,
      "step": 220511
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.5142889022827148,
      "learning_rate": 2.7241861135475996e-06,
      "loss": 1.7509,
      "step": 220512
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.3748688697814941,
      "learning_rate": 2.7236361349698867e-06,
      "loss": 2.0124,
      "step": 220513
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.4261858463287354,
      "learning_rate": 2.7230862116614627e-06,
      "loss": 2.1109,
      "step": 220514
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.4057942628860474,
      "learning_rate": 2.7225363436224278e-06,
      "loss": 2.0966,
      "step": 220515
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.4818525314331055,
      "learning_rate": 2.7219865308528487e-06,
      "loss": 2.0612,
      "step": 220516
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.4775364398956299,
      "learning_rate": 2.7214367733529254e-06,
      "loss": 1.9614,
      "step": 220517
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.5000137090682983,
      "learning_rate": 2.720887071122624e-06,
      "loss": 1.8494,
      "step": 220518
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.3329025506973267,
      "learning_rate": 2.720337424162178e-06,
      "loss": 2.1175,
      "step": 220519
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.5546753406524658,
      "learning_rate": 2.7197878324716204e-06,
      "loss": 2.2579,
      "step": 220520
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.435149073600769,
      "learning_rate": 2.719238296051085e-06,
      "loss": 2.1481,
      "step": 220521
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.4621129035949707,
      "learning_rate": 2.7186888149006046e-06,
      "loss": 2.1084,
      "step": 220522
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.4987632036209106,
      "learning_rate": 2.7181393890203795e-06,
      "loss": 2.0211,
      "step": 220523
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.550087571144104,
      "learning_rate": 2.7175900184104426e-06,
      "loss": 1.8907,
      "step": 220524
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.379952311515808,
      "learning_rate": 2.7170407030709275e-06,
      "loss": 1.8972,
      "step": 220525
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.314767837524414,
      "learning_rate": 2.7164914430019e-06,
      "loss": 1.8465,
      "step": 220526
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.4576300382614136,
      "learning_rate": 2.715942238203528e-06,
      "loss": 2.0383,
      "step": 220527
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.481635570526123,
      "learning_rate": 2.7153930886758436e-06,
      "loss": 2.1624,
      "step": 220528
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.6324204206466675,
      "learning_rate": 2.7148439944189806e-06,
      "loss": 1.9072,
      "step": 220529
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.3077136278152466,
      "learning_rate": 2.7142949554330384e-06,
      "loss": 1.8685,
      "step": 220530
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.3497412204742432,
      "learning_rate": 2.713745971718151e-06,
      "loss": 1.897,
      "step": 220531
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.5144890546798706,
      "learning_rate": 2.713197043274351e-06,
      "loss": 1.9484,
      "step": 220532
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.4125027656555176,
      "learning_rate": 2.712648170101839e-06,
      "loss": 2.0723,
      "step": 220533
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.6740899085998535,
      "learning_rate": 2.712099352200581e-06,
      "loss": 1.897,
      "step": 220534
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.5971214771270752,
      "learning_rate": 2.7115505895708102e-06,
      "loss": 2.171,
      "step": 220535
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.7760690450668335,
      "learning_rate": 2.7110018822125267e-06,
      "loss": 2.0239,
      "step": 220536
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.461254358291626,
      "learning_rate": 2.7104532301259306e-06,
      "loss": 1.8063,
      "step": 220537
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.4936530590057373,
      "learning_rate": 2.709904633311022e-06,
      "loss": 2.2469,
      "step": 220538
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.5839052200317383,
      "learning_rate": 2.709356091768e-06,
      "loss": 2.0795,
      "step": 220539
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.5549590587615967,
      "learning_rate": 2.708807605496899e-06,
      "loss": 2.0143,
      "step": 220540
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.4816116094589233,
      "learning_rate": 2.708259174497851e-06,
      "loss": 2.2267,
      "step": 220541
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.3978397846221924,
      "learning_rate": 2.7077107987709236e-06,
      "loss": 1.8065,
      "step": 220542
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.5699867010116577,
      "learning_rate": 2.7071624783162827e-06,
      "loss": 1.9957,
      "step": 220543
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.4831253290176392,
      "learning_rate": 2.706614213133962e-06,
      "loss": 2.03,
      "step": 220544
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.4919015169143677,
      "learning_rate": 2.7060660032241277e-06,
      "loss": 2.0128,
      "step": 220545
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.560078740119934,
      "learning_rate": 2.7055178485868135e-06,
      "loss": 1.8823,
      "step": 220546
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.6011312007904053,
      "learning_rate": 2.7049697492221854e-06,
      "loss": 1.9474,
      "step": 220547
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.3698281049728394,
      "learning_rate": 2.7044217051302776e-06,
      "loss": 1.8951,
      "step": 220548
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.4969656467437744,
      "learning_rate": 2.7038737163112554e-06,
      "loss": 1.9344,
      "step": 220549
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.4336845874786377,
      "learning_rate": 2.7033257827651864e-06,
      "loss": 1.9537,
      "step": 220550
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.353576898574829,
      "learning_rate": 2.7027779044922037e-06,
      "loss": 1.9438,
      "step": 220551
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.4624305963516235,
      "learning_rate": 2.7022300814923403e-06,
      "loss": 1.984,
      "step": 220552
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.3979952335357666,
      "learning_rate": 2.701682313765796e-06,
      "loss": 1.9609,
      "step": 220553
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.4234979152679443,
      "learning_rate": 2.701134601312571e-06,
      "loss": 1.9301,
      "step": 220554
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.3960340023040771,
      "learning_rate": 2.7005869441328653e-06,
      "loss": 2.2479,
      "step": 220555
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.5629935264587402,
      "learning_rate": 2.7000393422266785e-06,
      "loss": 2.093,
      "step": 220556
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.4410961866378784,
      "learning_rate": 2.6994917955942107e-06,
      "loss": 1.8888,
      "step": 220557
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.4245171546936035,
      "learning_rate": 2.6989443042354954e-06,
      "loss": 1.951,
      "step": 220558
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.4632751941680908,
      "learning_rate": 2.6983968681506985e-06,
      "loss": 1.8295,
      "step": 220559
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.5206412076950073,
      "learning_rate": 2.6978494873398206e-06,
      "loss": 2.1982,
      "step": 220560
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.5559661388397217,
      "learning_rate": 2.697302161803061e-06,
      "loss": 1.8018,
      "step": 220561
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.4548211097717285,
      "learning_rate": 2.6967548915404868e-06,
      "loss": 1.9416,
      "step": 220562
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.4231650829315186,
      "learning_rate": 2.6962076765521977e-06,
      "loss": 1.916,
      "step": 220563
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.5041651725769043,
      "learning_rate": 2.695660516838294e-06,
      "loss": 2.059,
      "step": 220564
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.5042831897735596,
      "learning_rate": 2.695113412398875e-06,
      "loss": 1.8789,
      "step": 220565
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.4418586492538452,
      "learning_rate": 2.6945663632340408e-06,
      "loss": 2.0425,
      "step": 220566
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.3707903623580933,
      "learning_rate": 2.694019369343892e-06,
      "loss": 2.1517,
      "step": 220567
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.2935214042663574,
      "learning_rate": 2.693472430728527e-06,
      "loss": 2.0277,
      "step": 220568
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.509636640548706,
      "learning_rate": 2.6929255473880805e-06,
      "loss": 1.9694,
      "step": 220569
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.384502649307251,
      "learning_rate": 2.692378719322652e-06,
      "loss": 2.0883,
      "step": 220570
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.4143025875091553,
      "learning_rate": 2.6918319465322745e-06,
      "loss": 1.8595,
      "step": 220571
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.4807615280151367,
      "learning_rate": 2.6912852290171482e-06,
      "loss": 2.0157,
      "step": 220572
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.4732352495193481,
      "learning_rate": 2.6907385667773063e-06,
      "loss": 2.2074,
      "step": 220573
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.5198256969451904,
      "learning_rate": 2.6901919598128486e-06,
      "loss": 1.9308,
      "step": 220574
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.5320526361465454,
      "learning_rate": 2.6896454081239082e-06,
      "loss": 2.1025,
      "step": 220575
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.4552931785583496,
      "learning_rate": 2.6890989117105854e-06,
      "loss": 1.9449,
      "step": 220576
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.4704755544662476,
      "learning_rate": 2.6885524705729468e-06,
      "loss": 2.2098,
      "step": 220577
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.4504282474517822,
      "learning_rate": 2.6880060847111252e-06,
      "loss": 1.9069,
      "step": 220578
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.5500187873840332,
      "learning_rate": 2.6874597541252207e-06,
      "loss": 2.0016,
      "step": 220579
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.351735234260559,
      "learning_rate": 2.6869134788153336e-06,
      "loss": 2.1092,
      "step": 220580
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.5993050336837769,
      "learning_rate": 2.686367258781563e-06,
      "loss": 2.038,
      "step": 220581
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.5341792106628418,
      "learning_rate": 2.68582109402401e-06,
      "loss": 2.0986,
      "step": 220582
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.4130444526672363,
      "learning_rate": 2.6852749845427735e-06,
      "loss": 2.2209,
      "step": 220583
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.4138389825820923,
      "learning_rate": 2.684728930337954e-06,
      "loss": 1.9743,
      "step": 220584
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.5200313329696655,
      "learning_rate": 2.6841829314096506e-06,
      "loss": 2.0163,
      "step": 220585
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.5559409856796265,
      "learning_rate": 2.6836369877579645e-06,
      "loss": 1.9408,
      "step": 220586
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.6063761711120605,
      "learning_rate": 2.683091099383028e-06,
      "loss": 1.7948,
      "step": 220587
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.4426639080047607,
      "learning_rate": 2.682545266284908e-06,
      "loss": 2.0911,
      "step": 220588
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.6667389869689941,
      "learning_rate": 2.681999488463704e-06,
      "loss": 2.0709,
      "step": 220589
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.450080394744873,
      "learning_rate": 2.681453765919517e-06,
      "loss": 1.8425,
      "step": 220590
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.6268322467803955,
      "learning_rate": 2.6809080986524788e-06,
      "loss": 2.1125,
      "step": 220591
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.4274431467056274,
      "learning_rate": 2.680362486662657e-06,
      "loss": 2.1551,
      "step": 220592
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.559786319732666,
      "learning_rate": 2.679816929950218e-06,
      "loss": 2.1089,
      "step": 220593
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.4630683660507202,
      "learning_rate": 2.6792714285151617e-06,
      "loss": 1.7769,
      "step": 220594
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.6454147100448608,
      "learning_rate": 2.6787259823576545e-06,
      "loss": 1.9564,
      "step": 220595
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.5841869115829468,
      "learning_rate": 2.6781805914777965e-06,
      "loss": 2.0395,
      "step": 220596
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.5266145467758179,
      "learning_rate": 2.6776352558756543e-06,
      "loss": 2.1305,
      "step": 220597
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.4639816284179688,
      "learning_rate": 2.6770899755513608e-06,
      "loss": 2.0451,
      "step": 220598
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.3780055046081543,
      "learning_rate": 2.6765447505050163e-06,
      "loss": 1.9105,
      "step": 220599
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.3686473369598389,
      "learning_rate": 2.675999580736721e-06,
      "loss": 1.892,
      "step": 220600
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.382645606994629,
      "learning_rate": 2.6754544662465406e-06,
      "loss": 1.8327,
      "step": 220601
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.5317962169647217,
      "learning_rate": 2.6749094070346423e-06,
      "loss": 2.0718,
      "step": 220602
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.4679760932922363,
      "learning_rate": 2.674364403101059e-06,
      "loss": 2.0706,
      "step": 220603
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.503412127494812,
      "learning_rate": 2.673819454445958e-06,
      "loss": 2.0311,
      "step": 220604
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.5467194318771362,
      "learning_rate": 2.673274561069372e-06,
      "loss": 1.8944,
      "step": 220605
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.5590147972106934,
      "learning_rate": 2.6727297229714675e-06,
      "loss": 2.0391,
      "step": 220606
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.498833179473877,
      "learning_rate": 2.6721849401522775e-06,
      "loss": 2.0771,
      "step": 220607
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.5520111322402954,
      "learning_rate": 2.6716402126119696e-06,
      "loss": 2.0925,
      "step": 220608
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.5101354122161865,
      "learning_rate": 2.6710955403506094e-06,
      "loss": 1.8612,
      "step": 220609
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.4835751056671143,
      "learning_rate": 2.6705509233682974e-06,
      "loss": 1.8774,
      "step": 220610
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.3923215866088867,
      "learning_rate": 2.6700063616651334e-06,
      "loss": 2.044,
      "step": 220611
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.5531072616577148,
      "learning_rate": 2.669461855241284e-06,
      "loss": 1.7872,
      "step": 220612
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.4405800104141235,
      "learning_rate": 2.6689174040967155e-06,
      "loss": 1.8118,
      "step": 220613
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.608237624168396,
      "learning_rate": 2.668373008231661e-06,
      "loss": 2.1704,
      "step": 220614
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.4878201484680176,
      "learning_rate": 2.667828667646155e-06,
      "loss": 1.7923,
      "step": 220615
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.5426396131515503,
      "learning_rate": 2.6672843823402957e-06,
      "loss": 1.8755,
      "step": 220616
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.4015018939971924,
      "learning_rate": 2.6667401523142172e-06,
      "loss": 1.7291,
      "step": 220617
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.4136130809783936,
      "learning_rate": 2.6661959775680196e-06,
      "loss": 1.8578,
      "step": 220618
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.3799409866333008,
      "learning_rate": 2.6656518581017695e-06,
      "loss": 1.9411,
      "step": 220619
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.5169665813446045,
      "learning_rate": 2.6651077939155997e-06,
      "loss": 2.1354,
      "step": 220620
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.507283091545105,
      "learning_rate": 2.664563785009577e-06,
      "loss": 1.9119,
      "step": 220621
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.573908805847168,
      "learning_rate": 2.6640198313838345e-06,
      "loss": 1.7791,
      "step": 220622
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.6289317607879639,
      "learning_rate": 2.6634759330384725e-06,
      "loss": 2.0335,
      "step": 220623
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.465194821357727,
      "learning_rate": 2.6629320899735905e-06,
      "loss": 2.1754,
      "step": 220624
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.5573283433914185,
      "learning_rate": 2.662388302189256e-06,
      "loss": 2.036,
      "step": 220625
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.4861990213394165,
      "learning_rate": 2.6618445696856006e-06,
      "loss": 2.2175,
      "step": 220626
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.4102673530578613,
      "learning_rate": 2.6613008924627253e-06,
      "loss": 1.8436,
      "step": 220627
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.3892848491668701,
      "learning_rate": 2.6607572705207302e-06,
      "loss": 1.8924,
      "step": 220628
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.5224124193191528,
      "learning_rate": 2.660213703859715e-06,
      "loss": 1.9951,
      "step": 220629
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.591492772102356,
      "learning_rate": 2.6596701924797794e-06,
      "loss": 1.929,
      "step": 220630
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.4619144201278687,
      "learning_rate": 2.659126736381023e-06,
      "loss": 1.967,
      "step": 220631
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.3913218975067139,
      "learning_rate": 2.6585833355635467e-06,
      "loss": 2.1112,
      "step": 220632
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.522369384765625,
      "learning_rate": 2.6580399900274496e-06,
      "loss": 1.9848,
      "step": 220633
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.627070426940918,
      "learning_rate": 2.657496699772832e-06,
      "loss": 2.269,
      "step": 220634
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.541193962097168,
      "learning_rate": 2.656953464799794e-06,
      "loss": 2.0479,
      "step": 220635
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.5440313816070557,
      "learning_rate": 2.6564102851084677e-06,
      "loss": 2.0319,
      "step": 220636
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.6079745292663574,
      "learning_rate": 2.6558671606989213e-06,
      "loss": 2.1187,
      "step": 220637
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.380147933959961,
      "learning_rate": 2.655324091571254e-06,
      "loss": 1.9192,
      "step": 220638
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.4664660692214966,
      "learning_rate": 2.6547810777255653e-06,
      "loss": 2.0677,
      "step": 220639
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.4588066339492798,
      "learning_rate": 2.6542381191619555e-06,
      "loss": 2.2021,
      "step": 220640
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.4909510612487793,
      "learning_rate": 2.653695215880558e-06,
      "loss": 2.0814,
      "step": 220641
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.5038868188858032,
      "learning_rate": 2.6531523678814394e-06,
      "loss": 2.1993,
      "step": 220642
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.3888304233551025,
      "learning_rate": 2.6526095751647326e-06,
      "loss": 1.9827,
      "step": 220643
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.4549590349197388,
      "learning_rate": 2.6520668377305044e-06,
      "loss": 1.9349,
      "step": 220644
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.4537384510040283,
      "learning_rate": 2.6515241555788547e-06,
      "loss": 2.0932,
      "step": 220645
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.4939225912094116,
      "learning_rate": 2.65098152870995e-06,
      "loss": 2.1987,
      "step": 220646
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.6154165267944336,
      "learning_rate": 2.6504389571237903e-06,
      "loss": 1.9441,
      "step": 220647
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.5509042739868164,
      "learning_rate": 2.6498964408205425e-06,
      "loss": 2.004,
      "step": 220648
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.3673464059829712,
      "learning_rate": 2.649353979800306e-06,
      "loss": 1.8183,
      "step": 220649
    },
    {
      "epoch": 2.87,
      "grad_norm": 2.113016128540039,
      "learning_rate": 2.648811574063148e-06,
      "loss": 1.7938,
      "step": 220650
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.590303659439087,
      "learning_rate": 2.648269223609201e-06,
      "loss": 2.036,
      "step": 220651
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.5237137079238892,
      "learning_rate": 2.6477269284385318e-06,
      "loss": 1.8022,
      "step": 220652
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.4141216278076172,
      "learning_rate": 2.647184688551274e-06,
      "loss": 1.863,
      "step": 220653
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.605065107345581,
      "learning_rate": 2.646642503947527e-06,
      "loss": 1.9716,
      "step": 220654
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.4713820219039917,
      "learning_rate": 2.646100374627391e-06,
      "loss": 1.9431,
      "step": 220655
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.4458039999008179,
      "learning_rate": 2.645558300590933e-06,
      "loss": 1.9574,
      "step": 220656
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.4912694692611694,
      "learning_rate": 2.645016281838286e-06,
      "loss": 2.2366,
      "step": 220657
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.4739530086517334,
      "learning_rate": 2.6444743183695495e-06,
      "loss": 1.9409,
      "step": 220658
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.4589344263076782,
      "learning_rate": 2.6439324101848237e-06,
      "loss": 2.0874,
      "step": 220659
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.3456737995147705,
      "learning_rate": 2.6433905572842083e-06,
      "loss": 1.9306,
      "step": 220660
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.423707127571106,
      "learning_rate": 2.6428487596677706e-06,
      "loss": 2.0955,
      "step": 220661
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.4322583675384521,
      "learning_rate": 2.642307017335643e-06,
      "loss": 1.8312,
      "step": 220662
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.4204516410827637,
      "learning_rate": 2.6417653302879595e-06,
      "loss": 2.0161,
      "step": 220663
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.5274925231933594,
      "learning_rate": 2.6412236985247526e-06,
      "loss": 1.9189,
      "step": 220664
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.3542568683624268,
      "learning_rate": 2.6406821220461562e-06,
      "loss": 1.7892,
      "step": 220665
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.3646225929260254,
      "learning_rate": 2.640140600852303e-06,
      "loss": 1.9353,
      "step": 220666
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.4882820844650269,
      "learning_rate": 2.639599134943227e-06,
      "loss": 2.0761,
      "step": 220667
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.508486270904541,
      "learning_rate": 2.63905772431906e-06,
      "loss": 1.9851,
      "step": 220668
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.621323585510254,
      "learning_rate": 2.638516368979937e-06,
      "loss": 2.0014,
      "step": 220669
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.6748533248901367,
      "learning_rate": 2.6379750689258906e-06,
      "loss": 2.0633,
      "step": 220670
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.3890236616134644,
      "learning_rate": 2.6374338241570867e-06,
      "loss": 2.0941,
      "step": 220671
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.3677433729171753,
      "learning_rate": 2.636892634673593e-06,
      "loss": 1.9495,
      "step": 220672
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.434518575668335,
      "learning_rate": 2.6363515004755087e-06,
      "loss": 2.0699,
      "step": 220673
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.321840524673462,
      "learning_rate": 2.6358104215629337e-06,
      "loss": 1.9298,
      "step": 220674
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.5928150415420532,
      "learning_rate": 2.6352693979360017e-06,
      "loss": 1.9973,
      "step": 220675
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.5590741634368896,
      "learning_rate": 2.634728429594746e-06,
      "loss": 2.3791,
      "step": 220676
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.531445860862732,
      "learning_rate": 2.6341875165393323e-06,
      "loss": 2.1413,
      "step": 220677
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.4054213762283325,
      "learning_rate": 2.6336466587697947e-06,
      "loss": 2.0826,
      "step": 220678
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.4147834777832031,
      "learning_rate": 2.633105856286366e-06,
      "loss": 1.8073,
      "step": 220679
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.5143616199493408,
      "learning_rate": 2.63256510908898e-06,
      "loss": 1.8911,
      "step": 220680
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.5217912197113037,
      "learning_rate": 2.632024417177836e-06,
      "loss": 1.965,
      "step": 220681
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.4459693431854248,
      "learning_rate": 2.631483780553001e-06,
      "loss": 2.1028,
      "step": 220682
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.492812991142273,
      "learning_rate": 2.630943199214608e-06,
      "loss": 2.3643,
      "step": 220683
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.517846941947937,
      "learning_rate": 2.6304026731626906e-06,
      "loss": 1.8898,
      "step": 220684
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.488747239112854,
      "learning_rate": 2.629862202397448e-06,
      "loss": 2.1829,
      "step": 220685
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.44072687625885,
      "learning_rate": 2.6293217869189143e-06,
      "loss": 1.8879,
      "step": 220686
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.5203361511230469,
      "learning_rate": 2.628781426727189e-06,
      "loss": 2.0537,
      "step": 220687
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.4292311668395996,
      "learning_rate": 2.628241121822372e-06,
      "loss": 2.0638,
      "step": 220688
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.5055878162384033,
      "learning_rate": 2.62770087220463e-06,
      "loss": 2.0142,
      "step": 220689
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.5867455005645752,
      "learning_rate": 2.627160677873963e-06,
      "loss": 1.9909,
      "step": 220690
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.5142861604690552,
      "learning_rate": 2.6266205388305706e-06,
      "loss": 1.8809,
      "step": 220691
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.4421491622924805,
      "learning_rate": 2.626080455074453e-06,
      "loss": 1.8892,
      "step": 220692
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.3823648691177368,
      "learning_rate": 2.625540426605777e-06,
      "loss": 1.9652,
      "step": 220693
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.462349772453308,
      "learning_rate": 2.6250004534246084e-06,
      "loss": 1.8191,
      "step": 220694
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.3490409851074219,
      "learning_rate": 2.6244605355311143e-06,
      "loss": 2.1431,
      "step": 220695
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.598409652709961,
      "learning_rate": 2.623920672925295e-06,
      "loss": 1.9562,
      "step": 220696
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.596282958984375,
      "learning_rate": 2.62338086560735e-06,
      "loss": 2.244,
      "step": 220697
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.5366692543029785,
      "learning_rate": 2.622841113577279e-06,
      "loss": 1.7237,
      "step": 220698
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.5030463933944702,
      "learning_rate": 2.622301416835282e-06,
      "loss": 1.8147,
      "step": 220699
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.5171067714691162,
      "learning_rate": 2.621761775381359e-06,
      "loss": 2.185,
      "step": 220700
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.516894817352295,
      "learning_rate": 2.6212221892157104e-06,
      "loss": 1.9407,
      "step": 220701
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.628404974937439,
      "learning_rate": 2.620682658338369e-06,
      "loss": 1.9382,
      "step": 220702
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.3626357316970825,
      "learning_rate": 2.620143182749501e-06,
      "loss": 1.9443,
      "step": 220703
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.4283087253570557,
      "learning_rate": 2.619603762449107e-06,
      "loss": 1.7671,
      "step": 220704
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.467718482017517,
      "learning_rate": 2.6190643974373536e-06,
      "loss": 2.1061,
      "step": 220705
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.4588521718978882,
      "learning_rate": 2.6185250877143406e-06,
      "loss": 1.8964,
      "step": 220706
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.4465627670288086,
      "learning_rate": 2.6179858332801675e-06,
      "loss": 2.137,
      "step": 220707
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.4994786977767944,
      "learning_rate": 2.617446634134901e-06,
      "loss": 2.0406,
      "step": 220708
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.4246649742126465,
      "learning_rate": 2.616907490278675e-06,
      "loss": 2.0802,
      "step": 220709
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.5805445909500122,
      "learning_rate": 2.6163684017115882e-06,
      "loss": 2.0425,
      "step": 220710
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.4327837228775024,
      "learning_rate": 2.6158293684337082e-06,
      "loss": 1.9323,
      "step": 220711
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.4306777715682983,
      "learning_rate": 2.615290390445168e-06,
      "loss": 2.1188,
      "step": 220712
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.5574480295181274,
      "learning_rate": 2.614751467746068e-06,
      "loss": 1.8591,
      "step": 220713
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.3751715421676636,
      "learning_rate": 2.614212600336474e-06,
      "loss": 1.6629,
      "step": 220714
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.6016243696212769,
      "learning_rate": 2.6136737882165527e-06,
      "loss": 2.0765,
      "step": 220715
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.5522619485855103,
      "learning_rate": 2.6131350313863374e-06,
      "loss": 1.9476,
      "step": 220716
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.47532057762146,
      "learning_rate": 2.6125963298459618e-06,
      "loss": 1.8196,
      "step": 220717
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.4854135513305664,
      "learning_rate": 2.6120576835955254e-06,
      "loss": 1.8554,
      "step": 220718
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.5027681589126587,
      "learning_rate": 2.611519092635095e-06,
      "loss": 1.8065,
      "step": 220719
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.5763219594955444,
      "learning_rate": 2.6109805569648034e-06,
      "loss": 1.8021,
      "step": 220720
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.3113102912902832,
      "learning_rate": 2.6104420765847845e-06,
      "loss": 1.8617,
      "step": 220721
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.4112058877944946,
      "learning_rate": 2.6099036514950713e-06,
      "loss": 1.8854,
      "step": 220722
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.4467374086380005,
      "learning_rate": 2.6093652816957966e-06,
      "loss": 1.9807,
      "step": 220723
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.5444921255111694,
      "learning_rate": 2.608826967187028e-06,
      "loss": 1.9406,
      "step": 220724
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.7466821670532227,
      "learning_rate": 2.6082887079688976e-06,
      "loss": 2.0496,
      "step": 220725
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.4184876680374146,
      "learning_rate": 2.6077505040415057e-06,
      "loss": 1.714,
      "step": 220726
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.4132630825042725,
      "learning_rate": 2.607212355404986e-06,
      "loss": 1.8128,
      "step": 220727
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.3662489652633667,
      "learning_rate": 2.6066742620593385e-06,
      "loss": 1.9601,
      "step": 220728
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.4640105962753296,
      "learning_rate": 2.6061362240047625e-06,
      "loss": 1.95,
      "step": 220729
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.532704472541809,
      "learning_rate": 2.6055982412412913e-06,
      "loss": 2.1064,
      "step": 220730
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.4796924591064453,
      "learning_rate": 2.6050603137690917e-06,
      "loss": 1.8451,
      "step": 220731
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.5192722082138062,
      "learning_rate": 2.6045224415881972e-06,
      "loss": 1.8951,
      "step": 220732
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.681316614151001,
      "learning_rate": 2.6039846246987404e-06,
      "loss": 1.8595,
      "step": 220733
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.3960444927215576,
      "learning_rate": 2.603446863100822e-06,
      "loss": 1.8099,
      "step": 220734
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.3853386640548706,
      "learning_rate": 2.6029091567945413e-06,
      "loss": 2.0026,
      "step": 220735
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.63074791431427,
      "learning_rate": 2.6023715057799654e-06,
      "loss": 2.0234,
      "step": 220736
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.703467845916748,
      "learning_rate": 2.6018339100572605e-06,
      "loss": 1.9743,
      "step": 220737
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.3777263164520264,
      "learning_rate": 2.60129636962646e-06,
      "loss": 1.9008,
      "step": 220738
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.47270667552948,
      "learning_rate": 2.600758884487697e-06,
      "loss": 1.9826,
      "step": 220739
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.4760011434555054,
      "learning_rate": 2.6002214546410716e-06,
      "loss": 2.1956,
      "step": 220740
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.4922791719436646,
      "learning_rate": 2.599684080086684e-06,
      "loss": 1.9654,
      "step": 220741
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.4979045391082764,
      "learning_rate": 2.599146760824633e-06,
      "loss": 2.0655,
      "step": 220742
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.5018852949142456,
      "learning_rate": 2.5986094968549863e-06,
      "loss": 1.8498,
      "step": 220743
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.437653660774231,
      "learning_rate": 2.5980722881779104e-06,
      "loss": 2.0751,
      "step": 220744
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.4662179946899414,
      "learning_rate": 2.597535134793438e-06,
      "loss": 1.9284,
      "step": 220745
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.3804062604904175,
      "learning_rate": 2.596998036701703e-06,
      "loss": 2.1664,
      "step": 220746
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.5171667337417603,
      "learning_rate": 2.596460993902805e-06,
      "loss": 1.9612,
      "step": 220747
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.4329839944839478,
      "learning_rate": 2.5959240063968435e-06,
      "loss": 1.8986,
      "step": 220748
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.537075400352478,
      "learning_rate": 2.595387074183919e-06,
      "loss": 1.9899,
      "step": 220749
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.502496361732483,
      "learning_rate": 2.5948501972640977e-06,
      "loss": 1.9471,
      "step": 220750
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.401986002922058,
      "learning_rate": 2.5943133756375467e-06,
      "loss": 1.9012,
      "step": 220751
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.6145952939987183,
      "learning_rate": 2.593776609304299e-06,
      "loss": 2.1166,
      "step": 220752
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.5128867626190186,
      "learning_rate": 2.5932398982644874e-06,
      "loss": 1.9958,
      "step": 220753
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.395047903060913,
      "learning_rate": 2.5927032425182125e-06,
      "loss": 1.9767,
      "step": 220754
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.3873811960220337,
      "learning_rate": 2.5921666420655407e-06,
      "loss": 2.1332,
      "step": 220755
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.4464545249938965,
      "learning_rate": 2.5916300969066384e-06,
      "loss": 2.0174,
      "step": 220756
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.5934650897979736,
      "learning_rate": 2.5910936070415388e-06,
      "loss": 1.8133,
      "step": 220757
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.532600998878479,
      "learning_rate": 2.590557172470409e-06,
      "loss": 2.0692,
      "step": 220758
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.536764144897461,
      "learning_rate": 2.590020793193248e-06,
      "loss": 1.9723,
      "step": 220759
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.46066153049469,
      "learning_rate": 2.5894844692102567e-06,
      "loss": 1.9704,
      "step": 220760
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.4617515802383423,
      "learning_rate": 2.588948200521468e-06,
      "loss": 1.6224,
      "step": 220761
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.52568519115448,
      "learning_rate": 2.588411987127048e-06,
      "loss": 1.9215,
      "step": 220762
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.3581331968307495,
      "learning_rate": 2.5878758290270306e-06,
      "loss": 1.9505,
      "step": 220763
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.515043020248413,
      "learning_rate": 2.5873397262215488e-06,
      "loss": 1.8764,
      "step": 220764
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.5447016954421997,
      "learning_rate": 2.5868036787107025e-06,
      "loss": 2.0063,
      "step": 220765
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.402682900428772,
      "learning_rate": 2.586267686494559e-06,
      "loss": 2.134,
      "step": 220766
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.5071572065353394,
      "learning_rate": 2.58573174957325e-06,
      "loss": 1.9339,
      "step": 220767
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.5190472602844238,
      "learning_rate": 2.5851958679468763e-06,
      "loss": 1.9973,
      "step": 220768
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.4621659517288208,
      "learning_rate": 2.584660041615505e-06,
      "loss": 2.0089,
      "step": 220769
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.3279629945755005,
      "learning_rate": 2.584124270579302e-06,
      "loss": 1.7791,
      "step": 220770
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.582070231437683,
      "learning_rate": 2.583588554838301e-06,
      "loss": 1.8542,
      "step": 220771
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.5509049892425537,
      "learning_rate": 2.5830528943926344e-06,
      "loss": 1.891,
      "step": 220772
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.4103366136550903,
      "learning_rate": 2.5825172892423694e-06,
      "loss": 1.8981,
      "step": 220773
    },
    {
      "epoch": 2.87,
      "grad_norm": 2.059314250946045,
      "learning_rate": 2.5819817393876395e-06,
      "loss": 1.8198,
      "step": 220774
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.3955695629119873,
      "learning_rate": 2.5814462448285443e-06,
      "loss": 1.8088,
      "step": 220775
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.463123083114624,
      "learning_rate": 2.580910805565184e-06,
      "loss": 1.9783,
      "step": 220776
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.6881515979766846,
      "learning_rate": 2.5803754215975915e-06,
      "loss": 1.9558,
      "step": 220777
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.349256157875061,
      "learning_rate": 2.5798400929259667e-06,
      "loss": 2.1217,
      "step": 220778
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.7068023681640625,
      "learning_rate": 2.579304819550343e-06,
      "loss": 2.012,
      "step": 220779
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.4786899089813232,
      "learning_rate": 2.578769601470887e-06,
      "loss": 2.0958,
      "step": 220780
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.4664021730422974,
      "learning_rate": 2.5782344386875984e-06,
      "loss": 1.8966,
      "step": 220781
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.4552624225616455,
      "learning_rate": 2.577699331200678e-06,
      "loss": 1.9721,
      "step": 220782
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.5449448823928833,
      "learning_rate": 2.5771642790101244e-06,
      "loss": 1.8964,
      "step": 220783
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.560958743095398,
      "learning_rate": 2.576629282116138e-06,
      "loss": 2.1187,
      "step": 220784
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.3266074657440186,
      "learning_rate": 2.5760943405187196e-06,
      "loss": 1.826,
      "step": 220785
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.412617564201355,
      "learning_rate": 2.575559454218068e-06,
      "loss": 2.0658,
      "step": 220786
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.550150752067566,
      "learning_rate": 2.5750246232141837e-06,
      "loss": 2.0675,
      "step": 220787
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.337181568145752,
      "learning_rate": 2.5744898475072662e-06,
      "loss": 1.9999,
      "step": 220788
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.3471887111663818,
      "learning_rate": 2.573955127097349e-06,
      "loss": 1.7735,
      "step": 220789
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.4906374216079712,
      "learning_rate": 2.573420461984532e-06,
      "loss": 2.1997,
      "step": 220790
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.4148226976394653,
      "learning_rate": 2.5728858521689487e-06,
      "loss": 1.8315,
      "step": 220791
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.429059624671936,
      "learning_rate": 2.5723512976506657e-06,
      "loss": 1.8731,
      "step": 220792
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.5445609092712402,
      "learning_rate": 2.5718167984298153e-06,
      "loss": 1.9715,
      "step": 220793
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.4599164724349976,
      "learning_rate": 2.5712823545064656e-06,
      "loss": 1.8153,
      "step": 220794
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.3963299989700317,
      "learning_rate": 2.5707479658807483e-06,
      "loss": 1.9849,
      "step": 220795
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.477182149887085,
      "learning_rate": 2.5702136325527312e-06,
      "loss": 2.0361,
      "step": 220796
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.5126957893371582,
      "learning_rate": 2.5696793545225135e-06,
      "loss": 1.9033,
      "step": 220797
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.4265515804290771,
      "learning_rate": 2.5691451317901958e-06,
      "loss": 1.8819,
      "step": 220798
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.4518382549285889,
      "learning_rate": 2.5686109643559104e-06,
      "loss": 1.9654,
      "step": 220799
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.5309679508209229,
      "learning_rate": 2.568076852219725e-06,
      "loss": 1.9099,
      "step": 220800
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.4029877185821533,
      "learning_rate": 2.5675427953817716e-06,
      "loss": 2.0714,
      "step": 220801
    },
    {
      "epoch": 2.87,
      "grad_norm": 1.439015507698059,
      "learning_rate": 2.567008793842118e-06,
      "loss": 2.2022,
      "step": 220802
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.4585191011428833,
      "learning_rate": 2.5664748476008634e-06,
      "loss": 2.0784,
      "step": 220803
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.4530659914016724,
      "learning_rate": 2.565940956658108e-06,
      "loss": 1.9913,
      "step": 220804
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.4770292043685913,
      "learning_rate": 2.5654071210139514e-06,
      "loss": 2.1152,
      "step": 220805
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.5503954887390137,
      "learning_rate": 2.564873340668527e-06,
      "loss": 2.0682,
      "step": 220806
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.4048064947128296,
      "learning_rate": 2.564339615621902e-06,
      "loss": 2.1122,
      "step": 220807
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.4045339822769165,
      "learning_rate": 2.5638059458741755e-06,
      "loss": 2.0655,
      "step": 220808
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.397072196006775,
      "learning_rate": 2.563272331425448e-06,
      "loss": 2.0142,
      "step": 220809
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.4494054317474365,
      "learning_rate": 2.562738772275852e-06,
      "loss": 1.9549,
      "step": 220810
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.4357627630233765,
      "learning_rate": 2.5622052684254213e-06,
      "loss": 2.0459,
      "step": 220811
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.956390380859375,
      "learning_rate": 2.561671819874289e-06,
      "loss": 1.966,
      "step": 220812
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.5635161399841309,
      "learning_rate": 2.561138426622589e-06,
      "loss": 1.7869,
      "step": 220813
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.4678986072540283,
      "learning_rate": 2.5606050886703532e-06,
      "loss": 1.9161,
      "step": 220814
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.5218582153320312,
      "learning_rate": 2.5600718060177494e-06,
      "loss": 1.9908,
      "step": 220815
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.5008070468902588,
      "learning_rate": 2.5595385786648105e-06,
      "loss": 1.8997,
      "step": 220816
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.4780550003051758,
      "learning_rate": 2.5590054066116695e-06,
      "loss": 1.8932,
      "step": 220817
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.4761759042739868,
      "learning_rate": 2.5584722898584597e-06,
      "loss": 1.9564,
      "step": 220818
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.4124600887298584,
      "learning_rate": 2.5579392284052147e-06,
      "loss": 1.9132,
      "step": 220819
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.3562467098236084,
      "learning_rate": 2.5574062222520675e-06,
      "loss": 1.8881,
      "step": 220820
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.4803094863891602,
      "learning_rate": 2.556873271399118e-06,
      "loss": 1.8937,
      "step": 220821
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.5678794384002686,
      "learning_rate": 2.5563403758464663e-06,
      "loss": 2.1206,
      "step": 220822
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.54132080078125,
      "learning_rate": 2.555807535594179e-06,
      "loss": 1.6997,
      "step": 220823
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.5227822065353394,
      "learning_rate": 2.5552747506423885e-06,
      "loss": 1.9885,
      "step": 220824
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.4629368782043457,
      "learning_rate": 2.5547420209912295e-06,
      "loss": 1.9676,
      "step": 220825
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.4710397720336914,
      "learning_rate": 2.554209346640734e-06,
      "loss": 2.0127,
      "step": 220826
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.4583628177642822,
      "learning_rate": 2.553676727591003e-06,
      "loss": 1.7022,
      "step": 220827
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.4866783618927002,
      "learning_rate": 2.553144163842169e-06,
      "loss": 1.9622,
      "step": 220828
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.4972862005233765,
      "learning_rate": 2.552611655394332e-06,
      "loss": 2.1562,
      "step": 220829
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.4757318496704102,
      "learning_rate": 2.5520792022475588e-06,
      "loss": 1.942,
      "step": 220830
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.3581382036209106,
      "learning_rate": 2.551546804402016e-06,
      "loss": 1.9338,
      "step": 220831
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.6014745235443115,
      "learning_rate": 2.5510144618577035e-06,
      "loss": 2.0276,
      "step": 220832
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.484094500541687,
      "learning_rate": 2.5504821746147874e-06,
      "loss": 1.9074,
      "step": 220833
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.616808295249939,
      "learning_rate": 2.5499499426733684e-06,
      "loss": 1.8723,
      "step": 220834
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.5759608745574951,
      "learning_rate": 2.5494177660335124e-06,
      "loss": 1.8429,
      "step": 220835
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.4259599447250366,
      "learning_rate": 2.54888564469532e-06,
      "loss": 1.8046,
      "step": 220836
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.5132579803466797,
      "learning_rate": 2.548353578658924e-06,
      "loss": 1.9361,
      "step": 220837
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.4082835912704468,
      "learning_rate": 2.5478215679243907e-06,
      "loss": 1.8445,
      "step": 220838
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.3528462648391724,
      "learning_rate": 2.547289612491854e-06,
      "loss": 2.0078,
      "step": 220839
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.4923248291015625,
      "learning_rate": 2.54675771236138e-06,
      "loss": 2.0021,
      "step": 220840
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.3881454467773438,
      "learning_rate": 2.5462258675330694e-06,
      "loss": 1.9236,
      "step": 220841
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.5161988735198975,
      "learning_rate": 2.5456940780070545e-06,
      "loss": 1.9891,
      "step": 220842
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.5533196926116943,
      "learning_rate": 2.5451623437834022e-06,
      "loss": 2.0373,
      "step": 220843
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.514216423034668,
      "learning_rate": 2.5446306648622126e-06,
      "loss": 1.9001,
      "step": 220844
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.4432971477508545,
      "learning_rate": 2.5440990412435856e-06,
      "loss": 1.9981,
      "step": 220845
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.4465546607971191,
      "learning_rate": 2.543567472927621e-06,
      "loss": 1.8918,
      "step": 220846
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.5477920770645142,
      "learning_rate": 2.5430359599144523e-06,
      "loss": 2.0945,
      "step": 220847
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.3758926391601562,
      "learning_rate": 2.5425045022041123e-06,
      "loss": 1.9483,
      "step": 220848
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.6412898302078247,
      "learning_rate": 2.541973099796768e-06,
      "loss": 1.9596,
      "step": 220849
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.540925145149231,
      "learning_rate": 2.5414417526924526e-06,
      "loss": 2.1407,
      "step": 220850
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.5887809991836548,
      "learning_rate": 2.5409104608913323e-06,
      "loss": 2.033,
      "step": 220851
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.4729148149490356,
      "learning_rate": 2.5403792243934407e-06,
      "loss": 1.945,
      "step": 220852
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.5361050367355347,
      "learning_rate": 2.539848043198911e-06,
      "loss": 2.1101,
      "step": 220853
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.3402973413467407,
      "learning_rate": 2.539316917307843e-06,
      "loss": 1.7848,
      "step": 220854
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.4951252937316895,
      "learning_rate": 2.5387858467203705e-06,
      "loss": 2.0306,
      "step": 220855
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.544782042503357,
      "learning_rate": 2.5382548314364927e-06,
      "loss": 1.936,
      "step": 220856
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.5119860172271729,
      "learning_rate": 2.5377238714564095e-06,
      "loss": 1.8037,
      "step": 220857
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.4276052713394165,
      "learning_rate": 2.5371929667801548e-06,
      "loss": 2.0639,
      "step": 220858
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.592248797416687,
      "learning_rate": 2.5366621174078615e-06,
      "loss": 1.8711,
      "step": 220859
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.4179540872573853,
      "learning_rate": 2.5361313233395962e-06,
      "loss": 2.0286,
      "step": 220860
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.4220408201217651,
      "learning_rate": 2.5356005845755258e-06,
      "loss": 2.0243,
      "step": 220861
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.5213541984558105,
      "learning_rate": 2.5350699011156827e-06,
      "loss": 1.987,
      "step": 220862
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.4057427644729614,
      "learning_rate": 2.5345392729601675e-06,
      "loss": 1.9008,
      "step": 220863
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.4521504640579224,
      "learning_rate": 2.5340087001091135e-06,
      "loss": 2.0068,
      "step": 220864
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.4557695388793945,
      "learning_rate": 2.5334781825625872e-06,
      "loss": 1.8829,
      "step": 220865
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.4569592475891113,
      "learning_rate": 2.532947720320688e-06,
      "loss": 1.8982,
      "step": 220866
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.3809995651245117,
      "learning_rate": 2.5324173133835836e-06,
      "loss": 2.0215,
      "step": 220867
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.576230525970459,
      "learning_rate": 2.5318869617512728e-06,
      "loss": 2.0696,
      "step": 220868
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.588002324104309,
      "learning_rate": 2.5313566654239225e-06,
      "loss": 2.0451,
      "step": 220869
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.6788065433502197,
      "learning_rate": 2.5308264244015996e-06,
      "loss": 2.0596,
      "step": 220870
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.4895658493041992,
      "learning_rate": 2.5302962386844037e-06,
      "loss": 2.1026,
      "step": 220871
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.3259057998657227,
      "learning_rate": 2.5297661082724352e-06,
      "loss": 1.8534,
      "step": 220872
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.3659098148345947,
      "learning_rate": 2.5292360331658266e-06,
      "loss": 1.8781,
      "step": 220873
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.390804648399353,
      "learning_rate": 2.5287060133646453e-06,
      "loss": 2.1617,
      "step": 220874
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.4142705202102661,
      "learning_rate": 2.5281760488689573e-06,
      "loss": 2.0097,
      "step": 220875
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.4804837703704834,
      "learning_rate": 2.5276461396789294e-06,
      "loss": 2.0924,
      "step": 220876
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.3698692321777344,
      "learning_rate": 2.5271162857946283e-06,
      "loss": 1.904,
      "step": 220877
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.4813790321350098,
      "learning_rate": 2.52658648721612e-06,
      "loss": 2.0264,
      "step": 220878
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.4488128423690796,
      "learning_rate": 2.5260567439435385e-06,
      "loss": 1.7692,
      "step": 220879
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.3584667444229126,
      "learning_rate": 2.5255270559770167e-06,
      "loss": 2.039,
      "step": 220880
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.5533703565597534,
      "learning_rate": 2.5249974233165883e-06,
      "loss": 2.0955,
      "step": 220881
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.4332678318023682,
      "learning_rate": 2.5244678459623856e-06,
      "loss": 2.0709,
      "step": 220882
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.4719237089157104,
      "learning_rate": 2.5239383239145096e-06,
      "loss": 2.02,
      "step": 220883
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.3968385457992554,
      "learning_rate": 2.523408857173026e-06,
      "loss": 2.184,
      "step": 220884
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.4298979043960571,
      "learning_rate": 2.5228794457380684e-06,
      "loss": 1.9888,
      "step": 220885
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.6041558980941772,
      "learning_rate": 2.522350089609704e-06,
      "loss": 2.0371,
      "step": 220886
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.435746192932129,
      "learning_rate": 2.521820788788065e-06,
      "loss": 2.0052,
      "step": 220887
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.4291399717330933,
      "learning_rate": 2.521291543273252e-06,
      "loss": 1.9403,
      "step": 220888
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.435463547706604,
      "learning_rate": 2.520762353065331e-06,
      "loss": 1.88,
      "step": 220889
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.5763438940048218,
      "learning_rate": 2.5202332181644024e-06,
      "loss": 2.1971,
      "step": 220890
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.6243096590042114,
      "learning_rate": 2.5197041385705997e-06,
      "loss": 2.0006,
      "step": 220891
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.436016321182251,
      "learning_rate": 2.519175114283989e-06,
      "loss": 2.0595,
      "step": 220892
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.533239722251892,
      "learning_rate": 2.51864614530467e-06,
      "loss": 2.0053,
      "step": 220893
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.630142331123352,
      "learning_rate": 2.518117231632744e-06,
      "loss": 2.1301,
      "step": 220894
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.6000113487243652,
      "learning_rate": 2.517588373268342e-06,
      "loss": 1.9765,
      "step": 220895
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.4868782758712769,
      "learning_rate": 2.5170595702115326e-06,
      "loss": 2.1314,
      "step": 220896
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.3501386642456055,
      "learning_rate": 2.516530822462415e-06,
      "loss": 1.8866,
      "step": 220897
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.471538782119751,
      "learning_rate": 2.5160021300210886e-06,
      "loss": 1.8482,
      "step": 220898
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.5193395614624023,
      "learning_rate": 2.515473492887654e-06,
      "loss": 1.9548,
      "step": 220899
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.4065722227096558,
      "learning_rate": 2.514944911062178e-06,
      "loss": 1.9096,
      "step": 220900
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.5142736434936523,
      "learning_rate": 2.514416384544826e-06,
      "loss": 2.0085,
      "step": 220901
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.5077157020568848,
      "learning_rate": 2.513887913335633e-06,
      "loss": 1.8466,
      "step": 220902
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.5847078561782837,
      "learning_rate": 2.5133594974347637e-06,
      "loss": 1.8865,
      "step": 220903
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.4526056051254272,
      "learning_rate": 2.5128311368422525e-06,
      "loss": 2.0373,
      "step": 220904
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.4513518810272217,
      "learning_rate": 2.5123028315581997e-06,
      "loss": 2.0023,
      "step": 220905
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.4504451751708984,
      "learning_rate": 2.5117745815827373e-06,
      "loss": 1.9161,
      "step": 220906
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.441141128540039,
      "learning_rate": 2.511246386915966e-06,
      "loss": 1.9978,
      "step": 220907
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.3954389095306396,
      "learning_rate": 2.5107182475579522e-06,
      "loss": 1.9552,
      "step": 220908
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.4786494970321655,
      "learning_rate": 2.5101901635088293e-06,
      "loss": 1.8899,
      "step": 220909
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.4515233039855957,
      "learning_rate": 2.509662134768697e-06,
      "loss": 1.7872,
      "step": 220910
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.3684022426605225,
      "learning_rate": 2.509134161337589e-06,
      "loss": 2.0965,
      "step": 220911
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.5464720726013184,
      "learning_rate": 2.508606243215672e-06,
      "loss": 1.8561,
      "step": 220912
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.280604362487793,
      "learning_rate": 2.508078380403011e-06,
      "loss": 2.1348,
      "step": 220913
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.524908185005188,
      "learning_rate": 2.507550572899708e-06,
      "loss": 1.8791,
      "step": 220914
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.4415680170059204,
      "learning_rate": 2.5070228207058953e-06,
      "loss": 2.0067,
      "step": 220915
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.445766806602478,
      "learning_rate": 2.506495123821639e-06,
      "loss": 1.9867,
      "step": 220916
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.433029055595398,
      "learning_rate": 2.5059674822470065e-06,
      "loss": 2.1186,
      "step": 220917
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.451920747756958,
      "learning_rate": 2.505439895982164e-06,
      "loss": 1.9637,
      "step": 220918
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.5088144540786743,
      "learning_rate": 2.504912365027145e-06,
      "loss": 1.7804,
      "step": 220919
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.4467658996582031,
      "learning_rate": 2.504384889382083e-06,
      "loss": 1.8877,
      "step": 220920
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.5108752250671387,
      "learning_rate": 2.5038574690471103e-06,
      "loss": 1.9671,
      "step": 220921
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.6384326219558716,
      "learning_rate": 2.503330104022261e-06,
      "loss": 2.0371,
      "step": 220922
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.5381903648376465,
      "learning_rate": 2.5028027943076345e-06,
      "loss": 2.0761,
      "step": 220923
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.4896223545074463,
      "learning_rate": 2.502275539903398e-06,
      "loss": 1.9293,
      "step": 220924
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.5482076406478882,
      "learning_rate": 2.501748340809584e-06,
      "loss": 1.8903,
      "step": 220925
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.6296616792678833,
      "learning_rate": 2.501221197026293e-06,
      "loss": 1.9486,
      "step": 220926
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.4461740255355835,
      "learning_rate": 2.500694108553658e-06,
      "loss": 2.1398,
      "step": 220927
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.6405187845230103,
      "learning_rate": 2.500167075391746e-06,
      "loss": 1.8965,
      "step": 220928
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.506805658340454,
      "learning_rate": 2.4996400975406893e-06,
      "loss": 2.0534,
      "step": 220929
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.5070922374725342,
      "learning_rate": 2.4991131750005556e-06,
      "loss": 2.0843,
      "step": 220930
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.499489188194275,
      "learning_rate": 2.4985863077714442e-06,
      "loss": 1.8879,
      "step": 220931
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.3373112678527832,
      "learning_rate": 2.498059495853488e-06,
      "loss": 2.0396,
      "step": 220932
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.401698112487793,
      "learning_rate": 2.4975327392467548e-06,
      "loss": 1.9993,
      "step": 220933
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.4914700984954834,
      "learning_rate": 2.49700603795131e-06,
      "loss": 1.8633,
      "step": 220934
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.5120735168457031,
      "learning_rate": 2.496479391967321e-06,
      "loss": 2.0605,
      "step": 220935
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.569899082183838,
      "learning_rate": 2.4959528012948535e-06,
      "loss": 2.0291,
      "step": 220936
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.4794219732284546,
      "learning_rate": 2.495426265933975e-06,
      "loss": 1.8448,
      "step": 220937
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.5527156591415405,
      "learning_rate": 2.4948997858848517e-06,
      "loss": 1.9802,
      "step": 220938
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.421586275100708,
      "learning_rate": 2.4943733611474837e-06,
      "loss": 2.0232,
      "step": 220939
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.4964758157730103,
      "learning_rate": 2.4938469917220707e-06,
      "loss": 1.7056,
      "step": 220940
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.384278416633606,
      "learning_rate": 2.493320677608679e-06,
      "loss": 1.8852,
      "step": 220941
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.5486913919448853,
      "learning_rate": 2.4927944188073757e-06,
      "loss": 2.0919,
      "step": 220942
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.6630791425704956,
      "learning_rate": 2.4922682153182606e-06,
      "loss": 2.1011,
      "step": 220943
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.536220908164978,
      "learning_rate": 2.4917420671415e-06,
      "loss": 1.9708,
      "step": 220944
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.618391513824463,
      "learning_rate": 2.4912159742770942e-06,
      "loss": 2.0284,
      "step": 220945
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.512509822845459,
      "learning_rate": 2.4906899367252098e-06,
      "loss": 2.0569,
      "step": 220946
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.4615685939788818,
      "learning_rate": 2.490163954485913e-06,
      "loss": 2.028,
      "step": 220947
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.5619014501571655,
      "learning_rate": 2.489638027559304e-06,
      "loss": 2.2372,
      "step": 220948
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.4896143674850464,
      "learning_rate": 2.489112155945483e-06,
      "loss": 2.1387,
      "step": 220949
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.362790822982788,
      "learning_rate": 2.4885863396445826e-06,
      "loss": 2.013,
      "step": 220950
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.4406168460845947,
      "learning_rate": 2.4880605786566368e-06,
      "loss": 2.0182,
      "step": 220951
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.6747394800186157,
      "learning_rate": 2.487534872981778e-06,
      "loss": 2.0589,
      "step": 220952
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.4512301683425903,
      "learning_rate": 2.4870092226201398e-06,
      "loss": 2.0624,
      "step": 220953
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.4045538902282715,
      "learning_rate": 2.486483627571756e-06,
      "loss": 1.7886,
      "step": 220954
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.6320981979370117,
      "learning_rate": 2.485958087836726e-06,
      "loss": 2.1333,
      "step": 220955
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.4536069631576538,
      "learning_rate": 2.485432603415216e-06,
      "loss": 1.8576,
      "step": 220956
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.4499484300613403,
      "learning_rate": 2.48490717430726e-06,
      "loss": 1.9966,
      "step": 220957
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.541496992111206,
      "learning_rate": 2.484381800512958e-06,
      "loss": 2.1375,
      "step": 220958
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.3812756538391113,
      "learning_rate": 2.4838564820324756e-06,
      "loss": 1.9378,
      "step": 220959
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.4920647144317627,
      "learning_rate": 2.4833312188658137e-06,
      "loss": 1.9453,
      "step": 220960
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.4278632402420044,
      "learning_rate": 2.482806011013139e-06,
      "loss": 1.8354,
      "step": 220961
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.4638041257858276,
      "learning_rate": 2.482280858474517e-06,
      "loss": 1.8833,
      "step": 220962
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.5206741094589233,
      "learning_rate": 2.4817557612500484e-06,
      "loss": 2.0961,
      "step": 220963
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.435786247253418,
      "learning_rate": 2.4812307193398663e-06,
      "loss": 1.898,
      "step": 220964
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.3983879089355469,
      "learning_rate": 2.480705732744037e-06,
      "loss": 2.1854,
      "step": 220965
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.499455213546753,
      "learning_rate": 2.480180801462628e-06,
      "loss": 2.1743,
      "step": 220966
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.5119520425796509,
      "learning_rate": 2.4796559254957716e-06,
      "loss": 2.0658,
      "step": 220967
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.6263197660446167,
      "learning_rate": 2.4791311048436015e-06,
      "loss": 1.8765,
      "step": 220968
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.4645572900772095,
      "learning_rate": 2.4786063395061506e-06,
      "loss": 2.0574,
      "step": 220969
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.469834804534912,
      "learning_rate": 2.4780816294835527e-06,
      "loss": 1.8302,
      "step": 220970
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.6802968978881836,
      "learning_rate": 2.477556974775874e-06,
      "loss": 2.0868,
      "step": 220971
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.3278682231903076,
      "learning_rate": 2.4770323753832477e-06,
      "loss": 1.9959,
      "step": 220972
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.5881757736206055,
      "learning_rate": 2.476507831305774e-06,
      "loss": 1.9261,
      "step": 220973
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.5244327783584595,
      "learning_rate": 2.4759833425435195e-06,
      "loss": 1.8475,
      "step": 220974
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.3740103244781494,
      "learning_rate": 2.4754589090965835e-06,
      "loss": 1.9421,
      "step": 220975
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.5282639265060425,
      "learning_rate": 2.474934530965067e-06,
      "loss": 1.9639,
      "step": 220976
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.4595210552215576,
      "learning_rate": 2.4744102081491025e-06,
      "loss": 1.921,
      "step": 220977
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.3984962701797485,
      "learning_rate": 2.47388594064879e-06,
      "loss": 1.956,
      "step": 220978
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.5019899606704712,
      "learning_rate": 2.4733617284641296e-06,
      "loss": 2.028,
      "step": 220979
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.3551454544067383,
      "learning_rate": 2.472837571595321e-06,
      "loss": 1.8866,
      "step": 220980
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.6364331245422363,
      "learning_rate": 2.472313470042431e-06,
      "loss": 2.0649,
      "step": 220981
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.5325419902801514,
      "learning_rate": 2.4717894238055593e-06,
      "loss": 1.9554,
      "step": 220982
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.4186782836914062,
      "learning_rate": 2.471265432884806e-06,
      "loss": 2.0564,
      "step": 220983
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.472785472869873,
      "learning_rate": 2.4707414972802374e-06,
      "loss": 1.9696,
      "step": 220984
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.6081750392913818,
      "learning_rate": 2.4702176169919874e-06,
      "loss": 2.2259,
      "step": 220985
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.6831756830215454,
      "learning_rate": 2.469693792020122e-06,
      "loss": 1.8481,
      "step": 220986
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.4400897026062012,
      "learning_rate": 2.4691700223647746e-06,
      "loss": 2.0418,
      "step": 220987
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.496989130973816,
      "learning_rate": 2.468646308026012e-06,
      "loss": 1.9403,
      "step": 220988
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.3358491659164429,
      "learning_rate": 2.4681226490039675e-06,
      "loss": 1.7861,
      "step": 220989
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.3985354900360107,
      "learning_rate": 2.467599045298707e-06,
      "loss": 1.9724,
      "step": 220990
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.5279452800750732,
      "learning_rate": 2.467075496910298e-06,
      "loss": 2.0688,
      "step": 220991
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.443466067314148,
      "learning_rate": 2.46655200383894e-06,
      "loss": 1.9214,
      "step": 220992
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.360338807106018,
      "learning_rate": 2.466028566084599e-06,
      "loss": 1.7658,
      "step": 220993
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.5504509210586548,
      "learning_rate": 2.465505183647476e-06,
      "loss": 1.9897,
      "step": 220994
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.4660271406173706,
      "learning_rate": 2.464981856527637e-06,
      "loss": 2.1405,
      "step": 220995
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.4076123237609863,
      "learning_rate": 2.4644585847251817e-06,
      "loss": 1.8397,
      "step": 220996
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.4064127206802368,
      "learning_rate": 2.4639353682401773e-06,
      "loss": 2.0165,
      "step": 220997
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.4364287853240967,
      "learning_rate": 2.463412207072757e-06,
      "loss": 1.8694,
      "step": 220998
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.4558641910552979,
      "learning_rate": 2.462889101222987e-06,
      "loss": 1.9162,
      "step": 220999
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.4304165840148926,
      "learning_rate": 2.4623660506910004e-06,
      "loss": 2.0405,
      "step": 221000
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.5135924816131592,
      "learning_rate": 2.461843055476864e-06,
      "loss": 1.9496,
      "step": 221001
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.5553905963897705,
      "learning_rate": 2.4613201155807117e-06,
      "loss": 1.9908,
      "step": 221002
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.595520257949829,
      "learning_rate": 2.4607972310026093e-06,
      "loss": 1.8926,
      "step": 221003
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.4588146209716797,
      "learning_rate": 2.4602744017426566e-06,
      "loss": 2.0537,
      "step": 221004
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.47859525680542,
      "learning_rate": 2.459751627800921e-06,
      "loss": 1.8446,
      "step": 221005
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.5039142370224,
      "learning_rate": 2.4592289091775684e-06,
      "loss": 1.901,
      "step": 221006
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.5440788269042969,
      "learning_rate": 2.458706245872666e-06,
      "loss": 1.7726,
      "step": 221007
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.6412200927734375,
      "learning_rate": 2.4581836378862797e-06,
      "loss": 1.741,
      "step": 221008
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.4875255823135376,
      "learning_rate": 2.4576610852185764e-06,
      "loss": 1.8666,
      "step": 221009
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.4238225221633911,
      "learning_rate": 2.4571385878695895e-06,
      "loss": 1.866,
      "step": 221010
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.4948736429214478,
      "learning_rate": 2.4566161458394185e-06,
      "loss": 2.0817,
      "step": 221011
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.764762282371521,
      "learning_rate": 2.456093759128197e-06,
      "loss": 1.8633,
      "step": 221012
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.4211363792419434,
      "learning_rate": 2.4555714277359916e-06,
      "loss": 2.0562,
      "step": 221013
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.54770827293396,
      "learning_rate": 2.4550491516629357e-06,
      "loss": 2.07,
      "step": 221014
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.3687492609024048,
      "learning_rate": 2.4545269309090955e-06,
      "loss": 2.0186,
      "step": 221015
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.5841097831726074,
      "learning_rate": 2.4540047654745375e-06,
      "loss": 1.954,
      "step": 221016
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.401570200920105,
      "learning_rate": 2.453482655359429e-06,
      "loss": 1.8588,
      "step": 221017
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.4636629819869995,
      "learning_rate": 2.452960600563836e-06,
      "loss": 2.0389,
      "step": 221018
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.4009510278701782,
      "learning_rate": 2.4524386010878582e-06,
      "loss": 1.9942,
      "step": 221019
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.5006719827651978,
      "learning_rate": 2.451916656931596e-06,
      "loss": 2.0796,
      "step": 221020
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.514122724533081,
      "learning_rate": 2.4513947680951163e-06,
      "loss": 2.1193,
      "step": 221021
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.6263840198516846,
      "learning_rate": 2.4508729345785514e-06,
      "loss": 2.0355,
      "step": 221022
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.463697075843811,
      "learning_rate": 2.450351156381969e-06,
      "loss": 2.1595,
      "step": 221023
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.5102851390838623,
      "learning_rate": 2.4498294335055014e-06,
      "loss": 2.1036,
      "step": 221024
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.4681129455566406,
      "learning_rate": 2.4493077659492157e-06,
      "loss": 1.9194,
      "step": 221025
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.5628834962844849,
      "learning_rate": 2.4487861537132447e-06,
      "loss": 1.8174,
      "step": 221026
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.5355288982391357,
      "learning_rate": 2.4482645967976554e-06,
      "loss": 2.052,
      "step": 221027
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.4909526109695435,
      "learning_rate": 2.4477430952025146e-06,
      "loss": 2.0058,
      "step": 221028
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.567730188369751,
      "learning_rate": 2.4472216489279883e-06,
      "loss": 2.1281,
      "step": 221029
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.4209754467010498,
      "learning_rate": 2.44670025797411e-06,
      "loss": 1.9337,
      "step": 221030
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.3590940237045288,
      "learning_rate": 2.4461789223410468e-06,
      "loss": 1.8636,
      "step": 221031
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.3481072187423706,
      "learning_rate": 2.445657642028831e-06,
      "loss": 1.9196,
      "step": 221032
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.3947865962982178,
      "learning_rate": 2.4451364170375964e-06,
      "loss": 1.9439,
      "step": 221033
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.4626295566558838,
      "learning_rate": 2.4446152473674096e-06,
      "loss": 1.9563,
      "step": 221034
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.5486875772476196,
      "learning_rate": 2.4440941330184037e-06,
      "loss": 1.8323,
      "step": 221035
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.4155670404434204,
      "learning_rate": 2.4435730739906455e-06,
      "loss": 2.0385,
      "step": 221036
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.4296363592147827,
      "learning_rate": 2.443052070284235e-06,
      "loss": 2.0732,
      "step": 221037
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.5198029279708862,
      "learning_rate": 2.442531121899305e-06,
      "loss": 2.1001,
      "step": 221038
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.4381030797958374,
      "learning_rate": 2.4420102288359223e-06,
      "loss": 1.8431,
      "step": 221039
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.423807144165039,
      "learning_rate": 2.4414893910941536e-06,
      "loss": 1.9464,
      "step": 221040
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.315200924873352,
      "learning_rate": 2.4409686086741654e-06,
      "loss": 1.8641,
      "step": 221041
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.4476293325424194,
      "learning_rate": 2.4404478815759908e-06,
      "loss": 2.0076,
      "step": 221042
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.5170023441314697,
      "learning_rate": 2.439927209799797e-06,
      "loss": 1.9486,
      "step": 221043
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.4124455451965332,
      "learning_rate": 2.439406593345583e-06,
      "loss": 1.9435,
      "step": 221044
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.5313084125518799,
      "learning_rate": 2.4388860322135495e-06,
      "loss": 2.0859,
      "step": 221045
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.3958410024642944,
      "learning_rate": 2.4383655264036962e-06,
      "loss": 2.1158,
      "step": 221046
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.4074662923812866,
      "learning_rate": 2.4378450759162227e-06,
      "loss": 1.7984,
      "step": 221047
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.4264450073242188,
      "learning_rate": 2.437324680751096e-06,
      "loss": 1.9233,
      "step": 221048
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.424087643623352,
      "learning_rate": 2.4368043409085493e-06,
      "loss": 1.9732,
      "step": 221049
    },
    {
      "epoch": 2.88,
      "grad_norm": 2.495713233947754,
      "learning_rate": 2.4362840563886154e-06,
      "loss": 1.9108,
      "step": 221050
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.5874544382095337,
      "learning_rate": 2.4357638271913615e-06,
      "loss": 2.119,
      "step": 221051
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.5118131637573242,
      "learning_rate": 2.4352436533169206e-06,
      "loss": 2.0372,
      "step": 221052
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.4624404907226562,
      "learning_rate": 2.4347235347654258e-06,
      "loss": 2.0207,
      "step": 221053
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.5697991847991943,
      "learning_rate": 2.4342034715368773e-06,
      "loss": 1.9561,
      "step": 221054
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.4331461191177368,
      "learning_rate": 2.433683463631475e-06,
      "loss": 2.0919,
      "step": 221055
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.5043097734451294,
      "learning_rate": 2.433163511049252e-06,
      "loss": 1.9373,
      "step": 221056
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.3862704038619995,
      "learning_rate": 2.4326436137903083e-06,
      "loss": 1.92,
      "step": 221057
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.5047457218170166,
      "learning_rate": 2.432123771854744e-06,
      "loss": 2.0054,
      "step": 221058
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.53383469581604,
      "learning_rate": 2.4316039852426916e-06,
      "loss": 2.028,
      "step": 221059
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.4402551651000977,
      "learning_rate": 2.4310842539541854e-06,
      "loss": 1.7894,
      "step": 221060
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.4659780263900757,
      "learning_rate": 2.430564577989391e-06,
      "loss": 2.0987,
      "step": 221061
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.3042621612548828,
      "learning_rate": 2.430044957348376e-06,
      "loss": 1.535,
      "step": 221062
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.641581654548645,
      "learning_rate": 2.42952539203124e-06,
      "loss": 1.9685,
      "step": 221063
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.4656574726104736,
      "learning_rate": 2.4290058820380156e-06,
      "loss": 2.0087,
      "step": 221064
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.4343960285186768,
      "learning_rate": 2.428486427368903e-06,
      "loss": 1.9333,
      "step": 221065
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.546891689300537,
      "learning_rate": 2.427967028023936e-06,
      "loss": 2.0343,
      "step": 221066
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.356734275817871,
      "learning_rate": 2.4274476840032475e-06,
      "loss": 1.8484,
      "step": 221067
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.4168041944503784,
      "learning_rate": 2.426928395306871e-06,
      "loss": 1.9128,
      "step": 221068
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.4103646278381348,
      "learning_rate": 2.4264091619350056e-06,
      "loss": 1.8483,
      "step": 221069
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.5068163871765137,
      "learning_rate": 2.425889983887652e-06,
      "loss": 1.9363,
      "step": 221070
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.5444204807281494,
      "learning_rate": 2.4253708611649438e-06,
      "loss": 1.8783,
      "step": 221071
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.5725483894348145,
      "learning_rate": 2.4248517937669797e-06,
      "loss": 2.0156,
      "step": 221072
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.5516966581344604,
      "learning_rate": 2.4243327816938606e-06,
      "loss": 1.9565,
      "step": 221073
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.3780686855316162,
      "learning_rate": 2.4238138249456527e-06,
      "loss": 1.832,
      "step": 221074
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.5079063177108765,
      "learning_rate": 2.42329492352249e-06,
      "loss": 2.0208,
      "step": 221075
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.4583083391189575,
      "learning_rate": 2.422776077424471e-06,
      "loss": 1.8703,
      "step": 221076
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.3883411884307861,
      "learning_rate": 2.422257286651663e-06,
      "loss": 1.9578,
      "step": 221077
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.6067376136779785,
      "learning_rate": 2.421738551204133e-06,
      "loss": 2.0811,
      "step": 221078
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.5253326892852783,
      "learning_rate": 2.421219871082081e-06,
      "loss": 2.087,
      "step": 221079
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.5062751770019531,
      "learning_rate": 2.4207012462855056e-06,
      "loss": 2.1162,
      "step": 221080
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.3577547073364258,
      "learning_rate": 2.4201826768145416e-06,
      "loss": 2.1811,
      "step": 221081
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.4679099321365356,
      "learning_rate": 2.4196641626692883e-06,
      "loss": 1.953,
      "step": 221082
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.5496894121170044,
      "learning_rate": 2.4191457038498452e-06,
      "loss": 1.9782,
      "step": 221083
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.6168028116226196,
      "learning_rate": 2.4186273003562794e-06,
      "loss": 2.0859,
      "step": 221084
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.5275697708129883,
      "learning_rate": 2.418108952188724e-06,
      "loss": 1.9609,
      "step": 221085
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.6761934757232666,
      "learning_rate": 2.417590659347246e-06,
      "loss": 2.0018,
      "step": 221086
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.5751575231552124,
      "learning_rate": 2.417072421831978e-06,
      "loss": 1.8387,
      "step": 221087
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.5546997785568237,
      "learning_rate": 2.416554239642987e-06,
      "loss": 1.9582,
      "step": 221088
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.3824670314788818,
      "learning_rate": 2.416036112780373e-06,
      "loss": 1.9259,
      "step": 221089
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.407427191734314,
      "learning_rate": 2.4155180412442353e-06,
      "loss": 2.0491,
      "step": 221090
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.4178003072738647,
      "learning_rate": 2.4150000250346747e-06,
      "loss": 2.093,
      "step": 221091
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.3425568342208862,
      "learning_rate": 2.414482064151757e-06,
      "loss": 1.9101,
      "step": 221092
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.44136643409729,
      "learning_rate": 2.413964158595616e-06,
      "loss": 2.011,
      "step": 221093
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.5047167539596558,
      "learning_rate": 2.4134463083663514e-06,
      "loss": 2.0775,
      "step": 221094
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.48377525806427,
      "learning_rate": 2.412928513464063e-06,
      "loss": 2.1448,
      "step": 221095
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.558021068572998,
      "learning_rate": 2.4124107738887845e-06,
      "loss": 1.9742,
      "step": 221096
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.4373112916946411,
      "learning_rate": 2.4118930896406817e-06,
      "loss": 1.8657,
      "step": 221097
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.4944462776184082,
      "learning_rate": 2.411375460719822e-06,
      "loss": 1.9942,
      "step": 221098
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.5160362720489502,
      "learning_rate": 2.410857887126305e-06,
      "loss": 2.08,
      "step": 221099
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.557965636253357,
      "learning_rate": 2.410340368860231e-06,
      "loss": 1.9959,
      "step": 221100
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.5386443138122559,
      "learning_rate": 2.4098229059216656e-06,
      "loss": 2.0956,
      "step": 221101
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.443535566329956,
      "learning_rate": 2.409305498310743e-06,
      "loss": 1.93,
      "step": 221102
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.5532819032669067,
      "learning_rate": 2.4087881460275626e-06,
      "loss": 1.9374,
      "step": 221103
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.5918256044387817,
      "learning_rate": 2.4082708490721913e-06,
      "loss": 1.999,
      "step": 221104
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.4351160526275635,
      "learning_rate": 2.407753607444762e-06,
      "loss": 2.0961,
      "step": 221105
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.4816185235977173,
      "learning_rate": 2.4072364211453423e-06,
      "loss": 1.8731,
      "step": 221106
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.4921690225601196,
      "learning_rate": 2.406719290173997e-06,
      "loss": 1.7542,
      "step": 221107
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.5746077299118042,
      "learning_rate": 2.4062022145308945e-06,
      "loss": 1.8626,
      "step": 221108
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.5005491971969604,
      "learning_rate": 2.4056851942161005e-06,
      "loss": 2.185,
      "step": 221109
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.465859293937683,
      "learning_rate": 2.4051682292296816e-06,
      "loss": 1.7316,
      "step": 221110
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.5114325284957886,
      "learning_rate": 2.404651319571804e-06,
      "loss": 2.2153,
      "step": 221111
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.4834680557250977,
      "learning_rate": 2.4041344652424687e-06,
      "loss": 1.9386,
      "step": 221112
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.4271000623703003,
      "learning_rate": 2.4036176662418415e-06,
      "loss": 1.9337,
      "step": 221113
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.6876074075698853,
      "learning_rate": 2.4031009225700228e-06,
      "loss": 2.0004,
      "step": 221114
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.3669127225875854,
      "learning_rate": 2.4025842342270453e-06,
      "loss": 1.8377,
      "step": 221115
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.6965663433074951,
      "learning_rate": 2.4020676012130757e-06,
      "loss": 2.0876,
      "step": 221116
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.589958667755127,
      "learning_rate": 2.401551023528181e-06,
      "loss": 1.9576,
      "step": 221117
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.7187010049819946,
      "learning_rate": 2.4010345011724276e-06,
      "loss": 1.9009,
      "step": 221118
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.362522006034851,
      "learning_rate": 2.4005180341459486e-06,
      "loss": 2.0606,
      "step": 221119
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.3895621299743652,
      "learning_rate": 2.4000016224488437e-06,
      "loss": 1.989,
      "step": 221120
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.3431100845336914,
      "learning_rate": 2.3994852660811803e-06,
      "loss": 1.7057,
      "step": 221121
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.596983790397644,
      "learning_rate": 2.398968965043091e-06,
      "loss": 2.0307,
      "step": 221122
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.5260627269744873,
      "learning_rate": 2.3984527193346425e-06,
      "loss": 1.9075,
      "step": 221123
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.4884438514709473,
      "learning_rate": 2.397936528955935e-06,
      "loss": 2.032,
      "step": 221124
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.4347374439239502,
      "learning_rate": 2.397420393907068e-06,
      "loss": 1.947,
      "step": 221125
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.3270182609558105,
      "learning_rate": 2.3969043141881417e-06,
      "loss": 1.7034,
      "step": 221126
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.526262640953064,
      "learning_rate": 2.396388289799256e-06,
      "loss": 1.9987,
      "step": 221127
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.4365839958190918,
      "learning_rate": 2.3958723207404772e-06,
      "loss": 1.8487,
      "step": 221128
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.5254559516906738,
      "learning_rate": 2.3953564070119393e-06,
      "loss": 1.9033,
      "step": 221129
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.491517186164856,
      "learning_rate": 2.394840548613741e-06,
      "loss": 1.9979,
      "step": 221130
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.539281964302063,
      "learning_rate": 2.394324745545917e-06,
      "loss": 1.9363,
      "step": 221131
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.4659572839736938,
      "learning_rate": 2.393808997808633e-06,
      "loss": 1.8969,
      "step": 221132
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.48573637008667,
      "learning_rate": 2.3932933054019554e-06,
      "loss": 2.078,
      "step": 221133
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.884451985359192,
      "learning_rate": 2.3927776683259845e-06,
      "loss": 2.1539,
      "step": 221134
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.4506736993789673,
      "learning_rate": 2.3922620865807875e-06,
      "loss": 2.0901,
      "step": 221135
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.436395525932312,
      "learning_rate": 2.39174656016653e-06,
      "loss": 1.9529,
      "step": 221136
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.5964375734329224,
      "learning_rate": 2.3912310890832452e-06,
      "loss": 2.0671,
      "step": 221137
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.4616267681121826,
      "learning_rate": 2.390715673331034e-06,
      "loss": 1.8547,
      "step": 221138
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.5002450942993164,
      "learning_rate": 2.3902003129100288e-06,
      "loss": 2.0795,
      "step": 221139
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.4941264390945435,
      "learning_rate": 2.389685007820297e-06,
      "loss": 1.7503,
      "step": 221140
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.4042333364486694,
      "learning_rate": 2.389169758061904e-06,
      "loss": 1.8445,
      "step": 221141
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.4725733995437622,
      "learning_rate": 2.388654563635051e-06,
      "loss": 2.0879,
      "step": 221142
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.488507628440857,
      "learning_rate": 2.3881394245397035e-06,
      "loss": 1.9934,
      "step": 221143
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.6060163974761963,
      "learning_rate": 2.387624340776062e-06,
      "loss": 2.1744,
      "step": 221144
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.4699655771255493,
      "learning_rate": 2.38710931234416e-06,
      "loss": 1.9541,
      "step": 221145
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.408050775527954,
      "learning_rate": 2.3865943392440967e-06,
      "loss": 2.046,
      "step": 221146
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.7026854753494263,
      "learning_rate": 2.386079421476006e-06,
      "loss": 1.9296,
      "step": 221147
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.5415170192718506,
      "learning_rate": 2.385564559039954e-06,
      "loss": 2.0091,
      "step": 221148
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.4429616928100586,
      "learning_rate": 2.385049751936041e-06,
      "loss": 2.114,
      "step": 221149
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.3620578050613403,
      "learning_rate": 2.384535000164367e-06,
      "loss": 1.9789,
      "step": 221150
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.4423794746398926,
      "learning_rate": 2.3840203037250314e-06,
      "loss": 2.1063,
      "step": 221151
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.5477153062820435,
      "learning_rate": 2.3835056626181347e-06,
      "loss": 1.9242,
      "step": 221152
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.4954264163970947,
      "learning_rate": 2.3829910768437097e-06,
      "loss": 1.8306,
      "step": 221153
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.5308743715286255,
      "learning_rate": 2.3824765464019903e-06,
      "loss": 2.1414,
      "step": 221154
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.4351392984390259,
      "learning_rate": 2.381962071292909e-06,
      "loss": 1.9792,
      "step": 221155
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.4220762252807617,
      "learning_rate": 2.381447651516666e-06,
      "loss": 2.2147,
      "step": 221156
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.4604134559631348,
      "learning_rate": 2.3809332870733277e-06,
      "loss": 2.2302,
      "step": 221157
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.426432728767395,
      "learning_rate": 2.3804189779629945e-06,
      "loss": 1.8959,
      "step": 221158
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.482089877128601,
      "learning_rate": 2.3799047241857662e-06,
      "loss": 2.0079,
      "step": 221159
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.5317296981811523,
      "learning_rate": 2.3793905257417424e-06,
      "loss": 1.9482,
      "step": 221160
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.390146255493164,
      "learning_rate": 2.37887638263099e-06,
      "loss": 1.8477,
      "step": 221161
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.511199712753296,
      "learning_rate": 2.3783622948536417e-06,
      "loss": 1.8645,
      "step": 221162
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.4051045179367065,
      "learning_rate": 2.3778482624097317e-06,
      "loss": 1.7573,
      "step": 221163
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.3625439405441284,
      "learning_rate": 2.377334285299426e-06,
      "loss": 1.922,
      "step": 221164
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.6190546751022339,
      "learning_rate": 2.376820363522791e-06,
      "loss": 2.1429,
      "step": 221165
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.4046237468719482,
      "learning_rate": 2.376306497079927e-06,
      "loss": 1.9468,
      "step": 221166
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.4145770072937012,
      "learning_rate": 2.375792685970934e-06,
      "loss": 1.9217,
      "step": 221167
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.5356618165969849,
      "learning_rate": 2.3752789301958784e-06,
      "loss": 1.9489,
      "step": 221168
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.3652337789535522,
      "learning_rate": 2.37476522975486e-06,
      "loss": 1.9483,
      "step": 221169
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.3974411487579346,
      "learning_rate": 2.374251584648046e-06,
      "loss": 1.8902,
      "step": 221170
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.6025125980377197,
      "learning_rate": 2.373737994875402e-06,
      "loss": 2.0122,
      "step": 221171
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.4461783170700073,
      "learning_rate": 2.3732244604371954e-06,
      "loss": 1.6994,
      "step": 221172
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.3246499300003052,
      "learning_rate": 2.372710981333359e-06,
      "loss": 1.9714,
      "step": 221173
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.5003682374954224,
      "learning_rate": 2.3721975575640594e-06,
      "loss": 2.1617,
      "step": 221174
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.563705563545227,
      "learning_rate": 2.3716841891293638e-06,
      "loss": 1.9386,
      "step": 221175
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.3808361291885376,
      "learning_rate": 2.371170876029438e-06,
      "loss": 1.9274,
      "step": 221176
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.4772882461547852,
      "learning_rate": 2.3706576182643156e-06,
      "loss": 1.7246,
      "step": 221177
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.3879894018173218,
      "learning_rate": 2.3701444158340965e-06,
      "loss": 1.9615,
      "step": 221178
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.4841631650924683,
      "learning_rate": 2.369631268738881e-06,
      "loss": 2.0841,
      "step": 221179
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.4959665536880493,
      "learning_rate": 2.3691181769788014e-06,
      "loss": 2.0234,
      "step": 221180
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.3830424547195435,
      "learning_rate": 2.368605140553892e-06,
      "loss": 1.911,
      "step": 221181
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.5810978412628174,
      "learning_rate": 2.3680921594642853e-06,
      "loss": 1.9687,
      "step": 221182
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.6677906513214111,
      "learning_rate": 2.3675792337100486e-06,
      "loss": 1.8628,
      "step": 221183
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.5185832977294922,
      "learning_rate": 2.3670663632913143e-06,
      "loss": 1.9845,
      "step": 221184
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.5356942415237427,
      "learning_rate": 2.3665535482081833e-06,
      "loss": 2.1777,
      "step": 221185
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.6466078758239746,
      "learning_rate": 2.366040788460688e-06,
      "loss": 2.1569,
      "step": 221186
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.4706274271011353,
      "learning_rate": 2.365528084048995e-06,
      "loss": 1.8698,
      "step": 221187
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.6113405227661133,
      "learning_rate": 2.365015434973139e-06,
      "loss": 2.2591,
      "step": 221188
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.438367486000061,
      "learning_rate": 2.3645028412332513e-06,
      "loss": 1.9599,
      "step": 221189
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.4889401197433472,
      "learning_rate": 2.3639903028294328e-06,
      "loss": 1.8769,
      "step": 221190
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.4447964429855347,
      "learning_rate": 2.3634778197617833e-06,
      "loss": 1.9302,
      "step": 221191
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.3873733282089233,
      "learning_rate": 2.3629653920303694e-06,
      "loss": 2.007,
      "step": 221192
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.615639090538025,
      "learning_rate": 2.362453019635291e-06,
      "loss": 2.0493,
      "step": 221193
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.4251537322998047,
      "learning_rate": 2.361940702576648e-06,
      "loss": 1.8092,
      "step": 221194
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.564805269241333,
      "learning_rate": 2.3614284408545404e-06,
      "loss": 1.9352,
      "step": 221195
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.5648932456970215,
      "learning_rate": 2.360916234469068e-06,
      "loss": 2.0742,
      "step": 221196
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.4526021480560303,
      "learning_rate": 2.3604040834203308e-06,
      "loss": 1.9541,
      "step": 221197
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.4663763046264648,
      "learning_rate": 2.359891987708362e-06,
      "loss": 2.117,
      "step": 221198
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.3910495042800903,
      "learning_rate": 2.3593799473333617e-06,
      "loss": 2.0136,
      "step": 221199
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.850337266921997,
      "learning_rate": 2.35886796229533e-06,
      "loss": 1.9632,
      "step": 221200
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.619266390800476,
      "learning_rate": 2.358356032594433e-06,
      "loss": 1.9985,
      "step": 221201
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.507981538772583,
      "learning_rate": 2.3578441582307373e-06,
      "loss": 1.6853,
      "step": 221202
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.3852758407592773,
      "learning_rate": 2.357332339204343e-06,
      "loss": 1.9918,
      "step": 221203
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.460257887840271,
      "learning_rate": 2.356820575515317e-06,
      "loss": 2.0857,
      "step": 221204
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.4221466779708862,
      "learning_rate": 2.356308867163792e-06,
      "loss": 2.1261,
      "step": 221205
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.4716531038284302,
      "learning_rate": 2.355797214149835e-06,
      "loss": 1.9027,
      "step": 221206
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.4968928098678589,
      "learning_rate": 2.3552856164735457e-06,
      "loss": 1.982,
      "step": 221207
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.4836513996124268,
      "learning_rate": 2.3547740741350574e-06,
      "loss": 2.1632,
      "step": 221208
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.4211759567260742,
      "learning_rate": 2.354262587134437e-06,
      "loss": 2.1095,
      "step": 221209
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.4064640998840332,
      "learning_rate": 2.353751155471717e-06,
      "loss": 1.9406,
      "step": 221210
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.4431171417236328,
      "learning_rate": 2.3532397791471315e-06,
      "loss": 1.9847,
      "step": 221211
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.4344472885131836,
      "learning_rate": 2.3527284581606466e-06,
      "loss": 1.7546,
      "step": 221212
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.5611472129821777,
      "learning_rate": 2.352217192512429e-06,
      "loss": 2.1202,
      "step": 221213
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.5212517976760864,
      "learning_rate": 2.3517059822025118e-06,
      "loss": 2.022,
      "step": 221214
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.5371161699295044,
      "learning_rate": 2.3511948272310956e-06,
      "loss": 1.8646,
      "step": 221215
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.3506996631622314,
      "learning_rate": 2.3506837275981795e-06,
      "loss": 2.0625,
      "step": 221216
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.460229516029358,
      "learning_rate": 2.350172683303897e-06,
      "loss": 2.055,
      "step": 221217
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.6366013288497925,
      "learning_rate": 2.3496616943483147e-06,
      "loss": 2.1267,
      "step": 221218
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.422956943511963,
      "learning_rate": 2.3491507607315663e-06,
      "loss": 2.0166,
      "step": 221219
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.4944202899932861,
      "learning_rate": 2.3486398824537177e-06,
      "loss": 1.9386,
      "step": 221220
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.5116615295410156,
      "learning_rate": 2.3481290595149025e-06,
      "loss": 1.8657,
      "step": 221221
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.3993597030639648,
      "learning_rate": 2.3476182919151545e-06,
      "loss": 2.0206,
      "step": 221222
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.448097586631775,
      "learning_rate": 2.347107579654639e-06,
      "loss": 1.887,
      "step": 221223
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.490095853805542,
      "learning_rate": 2.34659692273339e-06,
      "loss": 2.0563,
      "step": 221224
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.5101076364517212,
      "learning_rate": 2.346086321151508e-06,
      "loss": 2.1372,
      "step": 221225
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.417049765586853,
      "learning_rate": 2.345575774909125e-06,
      "loss": 1.8845,
      "step": 221226
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.4908201694488525,
      "learning_rate": 2.3450652840063422e-06,
      "loss": 1.894,
      "step": 221227
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.3578506708145142,
      "learning_rate": 2.3445548484432253e-06,
      "loss": 1.986,
      "step": 221228
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.3732823133468628,
      "learning_rate": 2.3440444682198744e-06,
      "loss": 2.0956,
      "step": 221229
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.3927520513534546,
      "learning_rate": 2.3435341433363563e-06,
      "loss": 1.886,
      "step": 221230
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.4557291269302368,
      "learning_rate": 2.343023873792804e-06,
      "loss": 2.0242,
      "step": 221231
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.416553020477295,
      "learning_rate": 2.3425136595892848e-06,
      "loss": 1.9424,
      "step": 221232
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.4724396467208862,
      "learning_rate": 2.342003500725964e-06,
      "loss": 2.0317,
      "step": 221233
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.5550612211227417,
      "learning_rate": 2.341493397202843e-06,
      "loss": 1.9715,
      "step": 221234
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.543745517730713,
      "learning_rate": 2.3409833490200867e-06,
      "loss": 1.8343,
      "step": 221235
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.6174861192703247,
      "learning_rate": 2.34047335617773e-06,
      "loss": 1.9247,
      "step": 221236
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.4485726356506348,
      "learning_rate": 2.3399634186759053e-06,
      "loss": 1.9814,
      "step": 221237
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.5085339546203613,
      "learning_rate": 2.3394535365147125e-06,
      "loss": 2.3809,
      "step": 221238
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.5439122915267944,
      "learning_rate": 2.3389437096942187e-06,
      "loss": 2.0209,
      "step": 221239
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.6120949983596802,
      "learning_rate": 2.338433938214557e-06,
      "loss": 1.9154,
      "step": 221240
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.5257488489151,
      "learning_rate": 2.33792422207576e-06,
      "loss": 2.0209,
      "step": 221241
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.582824945449829,
      "learning_rate": 2.337414561277995e-06,
      "loss": 2.0205,
      "step": 221242
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.45164155960083,
      "learning_rate": 2.3369049558213284e-06,
      "loss": 1.9097,
      "step": 221243
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.5219199657440186,
      "learning_rate": 2.336395405705793e-06,
      "loss": 1.9708,
      "step": 221244
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.64847731590271,
      "learning_rate": 2.335885910931623e-06,
      "loss": 2.0012,
      "step": 221245
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.345929503440857,
      "learning_rate": 2.3353764714987842e-06,
      "loss": 1.9348,
      "step": 221246
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.495219111442566,
      "learning_rate": 2.3348670874074106e-06,
      "loss": 2.0158,
      "step": 221247
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.5741595029830933,
      "learning_rate": 2.3343577586576346e-06,
      "loss": 2.1025,
      "step": 221248
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.4139686822891235,
      "learning_rate": 2.3338484852494898e-06,
      "loss": 1.9566,
      "step": 221249
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.3520106077194214,
      "learning_rate": 2.33333926718311e-06,
      "loss": 2.0279,
      "step": 221250
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.4231504201889038,
      "learning_rate": 2.332830104458594e-06,
      "loss": 1.97,
      "step": 221251
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.5961098670959473,
      "learning_rate": 2.3323209970760093e-06,
      "loss": 2.0059,
      "step": 221252
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.4352266788482666,
      "learning_rate": 2.331811945035489e-06,
      "loss": 2.0475,
      "step": 221253
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.2978911399841309,
      "learning_rate": 2.331302948337066e-06,
      "loss": 1.9577,
      "step": 221254
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.4889813661575317,
      "learning_rate": 2.330794006980907e-06,
      "loss": 2.0665,
      "step": 221255
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.6202630996704102,
      "learning_rate": 2.330285120967046e-06,
      "loss": 1.7667,
      "step": 221256
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.3736628293991089,
      "learning_rate": 2.3297762902956487e-06,
      "loss": 1.9112,
      "step": 221257
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.585710048675537,
      "learning_rate": 2.329267514966715e-06,
      "loss": 2.0597,
      "step": 221258
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.450253963470459,
      "learning_rate": 2.328758794980412e-06,
      "loss": 2.0482,
      "step": 221259
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.3393443822860718,
      "learning_rate": 2.328250130336806e-06,
      "loss": 1.8639,
      "step": 221260
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.489423155784607,
      "learning_rate": 2.327741521035997e-06,
      "loss": 1.839,
      "step": 221261
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.448818564414978,
      "learning_rate": 2.327232967078052e-06,
      "loss": 1.8699,
      "step": 221262
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.5202745199203491,
      "learning_rate": 2.3267244684631368e-06,
      "loss": 1.9502,
      "step": 221263
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.511781930923462,
      "learning_rate": 2.3262160251912853e-06,
      "loss": 2.0614,
      "step": 221264
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.6045050621032715,
      "learning_rate": 2.3257076372626303e-06,
      "loss": 2.0454,
      "step": 221265
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.4070512056350708,
      "learning_rate": 2.3251993046772056e-06,
      "loss": 1.9303,
      "step": 221266
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.4497802257537842,
      "learning_rate": 2.3246910274351773e-06,
      "loss": 1.8961,
      "step": 221267
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.4574750661849976,
      "learning_rate": 2.3241828055365783e-06,
      "loss": 1.9532,
      "step": 221268
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.5706509351730347,
      "learning_rate": 2.3236746389815765e-06,
      "loss": 1.968,
      "step": 221269
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.5037791728973389,
      "learning_rate": 2.323166527770204e-06,
      "loss": 2.087,
      "step": 221270
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.4721318483352661,
      "learning_rate": 2.322658471902561e-06,
      "loss": 1.9714,
      "step": 221271
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.442447304725647,
      "learning_rate": 2.3221504713787474e-06,
      "loss": 1.8533,
      "step": 221272
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.5539087057113647,
      "learning_rate": 2.3216425261988967e-06,
      "loss": 1.9312,
      "step": 221273
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.6473288536071777,
      "learning_rate": 2.3211346363630425e-06,
      "loss": 1.9183,
      "step": 221274
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.4988245964050293,
      "learning_rate": 2.320626801871317e-06,
      "loss": 2.2198,
      "step": 221275
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.535798192024231,
      "learning_rate": 2.3201190227238208e-06,
      "loss": 2.0232,
      "step": 221276
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.4706989526748657,
      "learning_rate": 2.3196112989206205e-06,
      "loss": 2.2196,
      "step": 221277
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.6450996398925781,
      "learning_rate": 2.319103630461816e-06,
      "loss": 2.238,
      "step": 221278
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.4852681159973145,
      "learning_rate": 2.3185960173475406e-06,
      "loss": 1.812,
      "step": 221279
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.489919662475586,
      "learning_rate": 2.318088459577827e-06,
      "loss": 1.985,
      "step": 221280
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.483825445175171,
      "learning_rate": 2.3175809571528093e-06,
      "loss": 2.0593,
      "step": 221281
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.4964473247528076,
      "learning_rate": 2.317073510072587e-06,
      "loss": 2.1534,
      "step": 221282
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.385035514831543,
      "learning_rate": 2.316566118337193e-06,
      "loss": 1.9905,
      "step": 221283
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.446602702140808,
      "learning_rate": 2.316058781946828e-06,
      "loss": 2.0395,
      "step": 221284
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.4986529350280762,
      "learning_rate": 2.3155515009014914e-06,
      "loss": 2.0877,
      "step": 221285
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.5160332918167114,
      "learning_rate": 2.3150442752013165e-06,
      "loss": 1.8806,
      "step": 221286
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.4760881662368774,
      "learning_rate": 2.3145371048464036e-06,
      "loss": 2.1734,
      "step": 221287
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.5551843643188477,
      "learning_rate": 2.3140299898368186e-06,
      "loss": 1.9107,
      "step": 221288
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.4335978031158447,
      "learning_rate": 2.3135229301726955e-06,
      "loss": 1.9415,
      "step": 221289
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.4836275577545166,
      "learning_rate": 2.3130159258541004e-06,
      "loss": 1.9336,
      "step": 221290
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.4699078798294067,
      "learning_rate": 2.3125089768811335e-06,
      "loss": 1.9611,
      "step": 221291
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.4694844484329224,
      "learning_rate": 2.3120020832538944e-06,
      "loss": 1.9194,
      "step": 221292
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.5402954816818237,
      "learning_rate": 2.311495244972483e-06,
      "loss": 2.1154,
      "step": 221293
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.5950913429260254,
      "learning_rate": 2.310988462036967e-06,
      "loss": 2.085,
      "step": 221294
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.4268028736114502,
      "learning_rate": 2.3104817344474445e-06,
      "loss": 1.7921,
      "step": 221295
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.4277902841567993,
      "learning_rate": 2.30997506220405e-06,
      "loss": 1.9714,
      "step": 221296
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.5153299570083618,
      "learning_rate": 2.3094684453068502e-06,
      "loss": 2.1289,
      "step": 221297
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.4601359367370605,
      "learning_rate": 2.308961883755911e-06,
      "loss": 1.834,
      "step": 221298
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.4614646434783936,
      "learning_rate": 2.3084553775513657e-06,
      "loss": 2.2086,
      "step": 221299
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.6730289459228516,
      "learning_rate": 2.3079489266933483e-06,
      "loss": 1.9849,
      "step": 221300
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.5376828908920288,
      "learning_rate": 2.307442531181858e-06,
      "loss": 2.189,
      "step": 221301
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.5223674774169922,
      "learning_rate": 2.306936191017028e-06,
      "loss": 1.8216,
      "step": 221302
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.497562050819397,
      "learning_rate": 2.3064299061989586e-06,
      "loss": 1.9832,
      "step": 221303
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.5069822072982788,
      "learning_rate": 2.30592367672775e-06,
      "loss": 1.9862,
      "step": 221304
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.4115570783615112,
      "learning_rate": 2.3054175026035014e-06,
      "loss": 1.9259,
      "step": 221305
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.3775439262390137,
      "learning_rate": 2.30491138382628e-06,
      "loss": 1.8545,
      "step": 221306
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.4980032444000244,
      "learning_rate": 2.3044053203961854e-06,
      "loss": 1.8809,
      "step": 221307
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.4261424541473389,
      "learning_rate": 2.303899312313351e-06,
      "loss": 2.0077,
      "step": 221308
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.5074313879013062,
      "learning_rate": 2.30339335957781e-06,
      "loss": 1.9901,
      "step": 221309
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.5645157098770142,
      "learning_rate": 2.3028874621897287e-06,
      "loss": 1.8206,
      "step": 221310
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.5177911520004272,
      "learning_rate": 2.3023816201491074e-06,
      "loss": 1.9314,
      "step": 221311
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.4006991386413574,
      "learning_rate": 2.301875833456146e-06,
      "loss": 2.0138,
      "step": 221312
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.4254710674285889,
      "learning_rate": 2.3013701021108443e-06,
      "loss": 2.1099,
      "step": 221313
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.3902807235717773,
      "learning_rate": 2.300864426113369e-06,
      "loss": 1.9612,
      "step": 221314
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.4290976524353027,
      "learning_rate": 2.3003588054637535e-06,
      "loss": 2.0773,
      "step": 221315
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.52449631690979,
      "learning_rate": 2.2998532401621306e-06,
      "loss": 1.7678,
      "step": 221316
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.5228136777877808,
      "learning_rate": 2.2993477302086004e-06,
      "loss": 1.8987,
      "step": 221317
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.4513062238693237,
      "learning_rate": 2.29884227560323e-06,
      "loss": 1.7906,
      "step": 221318
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.5708675384521484,
      "learning_rate": 2.2983368763461184e-06,
      "loss": 1.7986,
      "step": 221319
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.4117431640625,
      "learning_rate": 2.2978315324373666e-06,
      "loss": 1.8081,
      "step": 221320
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.4282337427139282,
      "learning_rate": 2.2973262438770734e-06,
      "loss": 2.0234,
      "step": 221321
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.4481087923049927,
      "learning_rate": 2.2968210106653397e-06,
      "loss": 2.0182,
      "step": 221322
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.4864442348480225,
      "learning_rate": 2.296315832802198e-06,
      "loss": 1.92,
      "step": 221323
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.5301264524459839,
      "learning_rate": 2.2958107102878487e-06,
      "loss": 1.9471,
      "step": 221324
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.4174652099609375,
      "learning_rate": 2.2953056431222918e-06,
      "loss": 1.8698,
      "step": 221325
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.4571564197540283,
      "learning_rate": 2.2948006313056932e-06,
      "loss": 1.938,
      "step": 221326
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.535910964012146,
      "learning_rate": 2.2942956748380535e-06,
      "loss": 2.1869,
      "step": 221327
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.4046106338500977,
      "learning_rate": 2.2937907737195725e-06,
      "loss": 1.674,
      "step": 221328
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.474661946296692,
      "learning_rate": 2.2932859279502503e-06,
      "loss": 2.0369,
      "step": 221329
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.2839915752410889,
      "learning_rate": 2.292781137530286e-06,
      "loss": 2.1016,
      "step": 221330
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.6103018522262573,
      "learning_rate": 2.2922764024596808e-06,
      "loss": 1.9944,
      "step": 221331
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.558847188949585,
      "learning_rate": 2.291771722738567e-06,
      "loss": 1.9308,
      "step": 221332
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.4265214204788208,
      "learning_rate": 2.291267098367011e-06,
      "loss": 2.0645,
      "step": 221333
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.411886215209961,
      "learning_rate": 2.290762529345147e-06,
      "loss": 1.9481,
      "step": 221334
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.6031394004821777,
      "learning_rate": 2.290258015673041e-06,
      "loss": 2.028,
      "step": 221335
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.445949912071228,
      "learning_rate": 2.2897535573508263e-06,
      "loss": 2.1168,
      "step": 221336
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.482802152633667,
      "learning_rate": 2.2892491543785363e-06,
      "loss": 1.7064,
      "step": 221337
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.453563928604126,
      "learning_rate": 2.288744806756304e-06,
      "loss": 1.784,
      "step": 221338
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.5633964538574219,
      "learning_rate": 2.2882405144841964e-06,
      "loss": 2.0088,
      "step": 221339
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.3952912092208862,
      "learning_rate": 2.2877362775623464e-06,
      "loss": 1.9997,
      "step": 221340
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.5788053274154663,
      "learning_rate": 2.287232095990821e-06,
      "loss": 1.9738,
      "step": 221341
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.4405878782272339,
      "learning_rate": 2.2867279697697194e-06,
      "loss": 2.1635,
      "step": 221342
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.5117467641830444,
      "learning_rate": 2.2862238988991423e-06,
      "loss": 1.8125,
      "step": 221343
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.4854973554611206,
      "learning_rate": 2.285719883379156e-06,
      "loss": 2.0316,
      "step": 221344
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.4762645959854126,
      "learning_rate": 2.28521592320986e-06,
      "loss": 1.8663,
      "step": 221345
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.4841309785842896,
      "learning_rate": 2.284712018391388e-06,
      "loss": 1.8382,
      "step": 221346
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.588395357131958,
      "learning_rate": 2.284208168923807e-06,
      "loss": 1.8227,
      "step": 221347
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.4291759729385376,
      "learning_rate": 2.283704374807216e-06,
      "loss": 1.7837,
      "step": 221348
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.5850903987884521,
      "learning_rate": 2.2832006360417153e-06,
      "loss": 1.9715,
      "step": 221349
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.4344005584716797,
      "learning_rate": 2.282696952627372e-06,
      "loss": 1.9868,
      "step": 221350
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.4078853130340576,
      "learning_rate": 2.2821933245642856e-06,
      "loss": 1.8842,
      "step": 221351
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.4040607213974,
      "learning_rate": 2.2816897518525555e-06,
      "loss": 1.9465,
      "step": 221352
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.4797509908676147,
      "learning_rate": 2.2811862344922825e-06,
      "loss": 1.9801,
      "step": 221353
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.421421766281128,
      "learning_rate": 2.2806827724835662e-06,
      "loss": 1.8771,
      "step": 221354
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.3895667791366577,
      "learning_rate": 2.2801793658265066e-06,
      "loss": 1.8209,
      "step": 221355
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.480858564376831,
      "learning_rate": 2.27967601452117e-06,
      "loss": 2.1114,
      "step": 221356
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.4637845754623413,
      "learning_rate": 2.2791727185676566e-06,
      "loss": 2.0173,
      "step": 221357
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.4712591171264648,
      "learning_rate": 2.2786694779660665e-06,
      "loss": 1.8511,
      "step": 221358
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.4927184581756592,
      "learning_rate": 2.2781662927164655e-06,
      "loss": 1.908,
      "step": 221359
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.5041241645812988,
      "learning_rate": 2.2776631628190214e-06,
      "loss": 1.9415,
      "step": 221360
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.4139987230300903,
      "learning_rate": 2.277160088273766e-06,
      "loss": 1.9799,
      "step": 221361
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.487589716911316,
      "learning_rate": 2.2766570690807673e-06,
      "loss": 1.9285,
      "step": 221362
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.438520073890686,
      "learning_rate": 2.276154105240191e-06,
      "loss": 2.1463,
      "step": 221363
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.45639967918396,
      "learning_rate": 2.275651196752104e-06,
      "loss": 1.7933,
      "step": 221364
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.515600562095642,
      "learning_rate": 2.275148343616573e-06,
      "loss": 2.1094,
      "step": 221365
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.4774209260940552,
      "learning_rate": 2.274645545833731e-06,
      "loss": 1.7771,
      "step": 221366
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.4340190887451172,
      "learning_rate": 2.2741428034036446e-06,
      "loss": 1.9628,
      "step": 221367
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.491317629814148,
      "learning_rate": 2.273640116326414e-06,
      "loss": 1.924,
      "step": 221368
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.5593173503875732,
      "learning_rate": 2.2731374846021387e-06,
      "loss": 1.8134,
      "step": 221369
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.4399446249008179,
      "learning_rate": 2.272634908230886e-06,
      "loss": 1.8889,
      "step": 221370
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.5433216094970703,
      "learning_rate": 2.2721323872127883e-06,
      "loss": 2.2712,
      "step": 221371
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.518136978149414,
      "learning_rate": 2.271629921547913e-06,
      "loss": 1.9589,
      "step": 221372
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.4854614734649658,
      "learning_rate": 2.2711275112363926e-06,
      "loss": 1.9813,
      "step": 221373
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.6218538284301758,
      "learning_rate": 2.2706251562782606e-06,
      "loss": 2.213,
      "step": 221374
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.4401518106460571,
      "learning_rate": 2.270122856673651e-06,
      "loss": 1.8273,
      "step": 221375
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.443083643913269,
      "learning_rate": 2.269620612422629e-06,
      "loss": 2.0369,
      "step": 221376
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.5080313682556152,
      "learning_rate": 2.2691184235252956e-06,
      "loss": 1.9616,
      "step": 221377
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.5839263200759888,
      "learning_rate": 2.268616289981784e-06,
      "loss": 1.9761,
      "step": 221378
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.5344351530075073,
      "learning_rate": 2.2681142117921604e-06,
      "loss": 2.0502,
      "step": 221379
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.5932097434997559,
      "learning_rate": 2.2676121889564915e-06,
      "loss": 2.0547,
      "step": 221380
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.4896093606948853,
      "learning_rate": 2.26711022147491e-06,
      "loss": 1.9591,
      "step": 221381
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.4155550003051758,
      "learning_rate": 2.266608309347451e-06,
      "loss": 1.7922,
      "step": 221382
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.4687747955322266,
      "learning_rate": 2.2661064525743124e-06,
      "loss": 2.0034,
      "step": 221383
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.5555006265640259,
      "learning_rate": 2.2656046511554615e-06,
      "loss": 1.9394,
      "step": 221384
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.4257904291152954,
      "learning_rate": 2.265102905091132e-06,
      "loss": 1.9454,
      "step": 221385
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.6395173072814941,
      "learning_rate": 2.2646012143812896e-06,
      "loss": 2.0127,
      "step": 221386
    },
    {
      "epoch": 2.88,
      "grad_norm": 2.130579710006714,
      "learning_rate": 2.2640995790260686e-06,
      "loss": 2.0912,
      "step": 221387
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.5941020250320435,
      "learning_rate": 2.263597999025568e-06,
      "loss": 1.9881,
      "step": 221388
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.565393328666687,
      "learning_rate": 2.263096474379922e-06,
      "loss": 1.9471,
      "step": 221389
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.3710720539093018,
      "learning_rate": 2.2625950050891626e-06,
      "loss": 1.765,
      "step": 221390
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.4692200422286987,
      "learning_rate": 2.2620935911534244e-06,
      "loss": 1.8136,
      "step": 221391
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.5906587839126587,
      "learning_rate": 2.2615922325727733e-06,
      "loss": 2.084,
      "step": 221392
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.4470202922821045,
      "learning_rate": 2.2610909293473088e-06,
      "loss": 1.6922,
      "step": 221393
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.487174391746521,
      "learning_rate": 2.2605896814771317e-06,
      "loss": 1.8622,
      "step": 221394
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.586266279220581,
      "learning_rate": 2.2600884889623416e-06,
      "loss": 2.1268,
      "step": 221395
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.5623266696929932,
      "learning_rate": 2.259587351803005e-06,
      "loss": 1.9821,
      "step": 221396
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.5947389602661133,
      "learning_rate": 2.2590862699992216e-06,
      "loss": 1.9742,
      "step": 221397
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.52897310256958,
      "learning_rate": 2.258585243551125e-06,
      "loss": 2.0001,
      "step": 221398
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.444463849067688,
      "learning_rate": 2.258084272458749e-06,
      "loss": 2.0429,
      "step": 221399
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.460644006729126,
      "learning_rate": 2.2575833567222256e-06,
      "loss": 2.0581,
      "step": 221400
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.4606571197509766,
      "learning_rate": 2.257082496341656e-06,
      "loss": 1.9536,
      "step": 221401
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.4102504253387451,
      "learning_rate": 2.256581691317072e-06,
      "loss": 1.9193,
      "step": 221402
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.5123693943023682,
      "learning_rate": 2.2560809416486414e-06,
      "loss": 2.0245,
      "step": 221403
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.4771206378936768,
      "learning_rate": 2.2555802473364306e-06,
      "loss": 1.7736,
      "step": 221404
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.436173915863037,
      "learning_rate": 2.255079608380539e-06,
      "loss": 2.0598,
      "step": 221405
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.771330714225769,
      "learning_rate": 2.254579024781e-06,
      "loss": 2.1339,
      "step": 221406
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.41374933719635,
      "learning_rate": 2.2540784965380142e-06,
      "loss": 1.9793,
      "step": 221407
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.392377495765686,
      "learning_rate": 2.253578023651548e-06,
      "loss": 1.8111,
      "step": 221408
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.4397242069244385,
      "learning_rate": 2.2530776061218338e-06,
      "loss": 2.0282,
      "step": 221409
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.411012053489685,
      "learning_rate": 2.2525772439488387e-06,
      "loss": 2.099,
      "step": 221410
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.4264905452728271,
      "learning_rate": 2.2520769371327297e-06,
      "loss": 2.1264,
      "step": 221411
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.4534693956375122,
      "learning_rate": 2.2515766856735727e-06,
      "loss": 1.8317,
      "step": 221412
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.5051536560058594,
      "learning_rate": 2.2510764895715017e-06,
      "loss": 2.1944,
      "step": 221413
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.346916675567627,
      "learning_rate": 2.250576348826516e-06,
      "loss": 1.8975,
      "step": 221414
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.4672232866287231,
      "learning_rate": 2.250076263438816e-06,
      "loss": 2.0167,
      "step": 221415
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.4543248414993286,
      "learning_rate": 2.2495762334084343e-06,
      "loss": 2.1075,
      "step": 221416
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.4383693933486938,
      "learning_rate": 2.249076258735505e-06,
      "loss": 2.1364,
      "step": 221417
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.5056508779525757,
      "learning_rate": 2.2485763394200273e-06,
      "loss": 1.9785,
      "step": 221418
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.6211038827896118,
      "learning_rate": 2.2480764754622018e-06,
      "loss": 2.0237,
      "step": 221419
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.363718032836914,
      "learning_rate": 2.247576666862061e-06,
      "loss": 1.8526,
      "step": 221420
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.3663617372512817,
      "learning_rate": 2.2470769136197385e-06,
      "loss": 1.7572,
      "step": 221421
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.5242362022399902,
      "learning_rate": 2.246577215735301e-06,
      "loss": 2.0609,
      "step": 221422
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.4177275896072388,
      "learning_rate": 2.246077573208849e-06,
      "loss": 2.0398,
      "step": 221423
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.5092008113861084,
      "learning_rate": 2.2455779860404476e-06,
      "loss": 2.1401,
      "step": 221424
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.4078307151794434,
      "learning_rate": 2.2450784542302315e-06,
      "loss": 1.8059,
      "step": 221425
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.4619637727737427,
      "learning_rate": 2.2445789777782665e-06,
      "loss": 2.0221,
      "step": 221426
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.587282419204712,
      "learning_rate": 2.2440795566846857e-06,
      "loss": 2.2036,
      "step": 221427
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.4032471179962158,
      "learning_rate": 2.2435801909495234e-06,
      "loss": 1.8141,
      "step": 221428
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.4980312585830688,
      "learning_rate": 2.2430808805729118e-06,
      "loss": 1.8254,
      "step": 221429
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.7120652198791504,
      "learning_rate": 2.2425816255549176e-06,
      "loss": 2.1127,
      "step": 221430
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.4919092655181885,
      "learning_rate": 2.2420824258956414e-06,
      "loss": 1.957,
      "step": 221431
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.583340048789978,
      "learning_rate": 2.241583281595183e-06,
      "loss": 2.0232,
      "step": 221432
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.4286916255950928,
      "learning_rate": 2.241084192653675e-06,
      "loss": 2.0881,
      "step": 221433
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.45168936252594,
      "learning_rate": 2.2405851590711176e-06,
      "loss": 1.8603,
      "step": 221434
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.637696623802185,
      "learning_rate": 2.2400861808477112e-06,
      "loss": 2.043,
      "step": 221435
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.4206832647323608,
      "learning_rate": 2.239587257983455e-06,
      "loss": 2.0216,
      "step": 221436
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.405457854270935,
      "learning_rate": 2.239088390478483e-06,
      "loss": 2.08,
      "step": 221437
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.416263461112976,
      "learning_rate": 2.2385895783328943e-06,
      "loss": 1.9348,
      "step": 221438
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.4332596063613892,
      "learning_rate": 2.2380908215467562e-06,
      "loss": 1.7915,
      "step": 221439
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.3915355205535889,
      "learning_rate": 2.2375921201202018e-06,
      "loss": 1.7425,
      "step": 221440
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.5250506401062012,
      "learning_rate": 2.2370934740532974e-06,
      "loss": 2.1082,
      "step": 221441
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.539686918258667,
      "learning_rate": 2.23659488334611e-06,
      "loss": 1.8976,
      "step": 221442
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.449272871017456,
      "learning_rate": 2.2360963479987726e-06,
      "loss": 1.8153,
      "step": 221443
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.5880694389343262,
      "learning_rate": 2.235597868011385e-06,
      "loss": 1.8125,
      "step": 221444
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.3929284811019897,
      "learning_rate": 2.235099443384014e-06,
      "loss": 1.8259,
      "step": 221445
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.5658079385757446,
      "learning_rate": 2.2346010741167597e-06,
      "loss": 1.9338,
      "step": 221446
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.6317946910858154,
      "learning_rate": 2.2341027602096883e-06,
      "loss": 1.9549,
      "step": 221447
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.5197687149047852,
      "learning_rate": 2.2336045016629335e-06,
      "loss": 1.9823,
      "step": 221448
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.5390266180038452,
      "learning_rate": 2.2331062984765946e-06,
      "loss": 2.0341,
      "step": 221449
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.3805487155914307,
      "learning_rate": 2.2326081506507053e-06,
      "loss": 1.7933,
      "step": 221450
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.5483635663986206,
      "learning_rate": 2.232110058185399e-06,
      "loss": 2.135,
      "step": 221451
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.4507415294647217,
      "learning_rate": 2.2316120210808085e-06,
      "loss": 1.7958,
      "step": 221452
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.504366397857666,
      "learning_rate": 2.2311140393369344e-06,
      "loss": 1.9255,
      "step": 221453
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.629782795906067,
      "learning_rate": 2.2306161129539424e-06,
      "loss": 2.1703,
      "step": 221454
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.4511704444885254,
      "learning_rate": 2.2301182419318663e-06,
      "loss": 1.7888,
      "step": 221455
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.404405117034912,
      "learning_rate": 2.2296204262708727e-06,
      "loss": 1.9057,
      "step": 221456
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.5060317516326904,
      "learning_rate": 2.229122665970995e-06,
      "loss": 1.8821,
      "step": 221457
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.4574834108352661,
      "learning_rate": 2.2286249610323325e-06,
      "loss": 1.8153,
      "step": 221458
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.449906349182129,
      "learning_rate": 2.2281273114550193e-06,
      "loss": 1.9347,
      "step": 221459
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.4830588102340698,
      "learning_rate": 2.227629717239088e-06,
      "loss": 1.9468,
      "step": 221460
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.481707215309143,
      "learning_rate": 2.227132178384672e-06,
      "loss": 1.7417,
      "step": 221461
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.4549211263656616,
      "learning_rate": 2.2266346948918713e-06,
      "loss": 1.888,
      "step": 221462
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.5463312864303589,
      "learning_rate": 2.2261372667607525e-06,
      "loss": 1.8106,
      "step": 221463
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.366735577583313,
      "learning_rate": 2.2256398939913823e-06,
      "loss": 1.8553,
      "step": 221464
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.409515619277954,
      "learning_rate": 2.225142576583927e-06,
      "loss": 2.0416,
      "step": 221465
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.5515837669372559,
      "learning_rate": 2.2246453145384204e-06,
      "loss": 1.9434,
      "step": 221466
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.4888097047805786,
      "learning_rate": 2.2241481078549617e-06,
      "loss": 1.9218,
      "step": 221467
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.589964509010315,
      "learning_rate": 2.2236509565336845e-06,
      "loss": 1.8746,
      "step": 221468
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.5473562479019165,
      "learning_rate": 2.2231538605746226e-06,
      "loss": 2.13,
      "step": 221469
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.5550475120544434,
      "learning_rate": 2.2226568199779415e-06,
      "loss": 2.2983,
      "step": 221470
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.339147925376892,
      "learning_rate": 2.222159834743642e-06,
      "loss": 1.9975,
      "step": 221471
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.61685049533844,
      "learning_rate": 2.2216629048718903e-06,
      "loss": 2.078,
      "step": 221472
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.4939709901809692,
      "learning_rate": 2.221166030362753e-06,
      "loss": 2.0823,
      "step": 221473
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.6187019348144531,
      "learning_rate": 2.220669211216297e-06,
      "loss": 1.9105,
      "step": 221474
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.4377613067626953,
      "learning_rate": 2.2201724474326556e-06,
      "loss": 1.9769,
      "step": 221475
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.425920844078064,
      "learning_rate": 2.2196757390118948e-06,
      "loss": 2.0797,
      "step": 221476
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.5800306797027588,
      "learning_rate": 2.219179085954148e-06,
      "loss": 1.8848,
      "step": 221477
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.4885492324829102,
      "learning_rate": 2.2186824882594488e-06,
      "loss": 2.0598,
      "step": 221478
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.4559462070465088,
      "learning_rate": 2.2181859459279305e-06,
      "loss": 2.2732,
      "step": 221479
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.4708482027053833,
      "learning_rate": 2.2176894589596596e-06,
      "loss": 2.1007,
      "step": 221480
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.5600112676620483,
      "learning_rate": 2.2171930273547356e-06,
      "loss": 2.1213,
      "step": 221481
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.485593557357788,
      "learning_rate": 2.216696651113292e-06,
      "loss": 2.0489,
      "step": 221482
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.5237504243850708,
      "learning_rate": 2.2162003302353627e-06,
      "loss": 1.9191,
      "step": 221483
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.6331212520599365,
      "learning_rate": 2.21570406472108e-06,
      "loss": 2.0902,
      "step": 221484
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.4396637678146362,
      "learning_rate": 2.215207854570478e-06,
      "loss": 1.963,
      "step": 221485
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.3420554399490356,
      "learning_rate": 2.2147116997837222e-06,
      "loss": 2.1549,
      "step": 221486
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.4732111692428589,
      "learning_rate": 2.214215600360847e-06,
      "loss": 2.0602,
      "step": 221487
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.7217265367507935,
      "learning_rate": 2.2137195563020184e-06,
      "loss": 2.2412,
      "step": 221488
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.3590776920318604,
      "learning_rate": 2.213223567607236e-06,
      "loss": 2.0082,
      "step": 221489
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.4784877300262451,
      "learning_rate": 2.2127276342766677e-06,
      "loss": 2.073,
      "step": 221490
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.4687319993972778,
      "learning_rate": 2.2122317563103454e-06,
      "loss": 2.0694,
      "step": 221491
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.582973599433899,
      "learning_rate": 2.211735933708403e-06,
      "loss": 1.8671,
      "step": 221492
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.4248071908950806,
      "learning_rate": 2.2112401664709066e-06,
      "loss": 1.8775,
      "step": 221493
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.467282772064209,
      "learning_rate": 2.2107444545979904e-06,
      "loss": 2.0067,
      "step": 221494
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.388793706893921,
      "learning_rate": 2.21024879808972e-06,
      "loss": 1.7452,
      "step": 221495
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.4862022399902344,
      "learning_rate": 2.2097531969461625e-06,
      "loss": 1.8863,
      "step": 221496
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.590871810913086,
      "learning_rate": 2.2092576511674177e-06,
      "loss": 2.0779,
      "step": 221497
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.446067214012146,
      "learning_rate": 2.2087621607536186e-06,
      "loss": 1.9954,
      "step": 221498
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.4309170246124268,
      "learning_rate": 2.2082667257048326e-06,
      "loss": 2.1553,
      "step": 221499
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.5204582214355469,
      "learning_rate": 2.2077713460211588e-06,
      "loss": 2.0944,
      "step": 221500
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.4316797256469727,
      "learning_rate": 2.207276021702664e-06,
      "loss": 1.7914,
      "step": 221501
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.3505040407180786,
      "learning_rate": 2.2067807527494817e-06,
      "loss": 1.9306,
      "step": 221502
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.4875513315200806,
      "learning_rate": 2.2062855391616453e-06,
      "loss": 1.8892,
      "step": 221503
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.6699820756912231,
      "learning_rate": 2.2057903809392875e-06,
      "loss": 2.0268,
      "step": 221504
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.3089898824691772,
      "learning_rate": 2.2052952780825085e-06,
      "loss": 1.9736,
      "step": 221505
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.52591073513031,
      "learning_rate": 2.204800230591375e-06,
      "loss": 1.7428,
      "step": 221506
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.5797009468078613,
      "learning_rate": 2.2043052384660197e-06,
      "loss": 1.9441,
      "step": 221507
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.4536616802215576,
      "learning_rate": 2.2038103017064767e-06,
      "loss": 1.9948,
      "step": 221508
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.5471434593200684,
      "learning_rate": 2.203315420312879e-06,
      "loss": 2.0258,
      "step": 221509
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.4327867031097412,
      "learning_rate": 2.202820594285293e-06,
      "loss": 1.9527,
      "step": 221510
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.5389918088912964,
      "learning_rate": 2.2023258236238184e-06,
      "loss": 2.0219,
      "step": 221511
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.464562177658081,
      "learning_rate": 2.201831108328589e-06,
      "loss": 1.8452,
      "step": 221512
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.5236951112747192,
      "learning_rate": 2.2013364483996374e-06,
      "loss": 2.1102,
      "step": 221513
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.425286054611206,
      "learning_rate": 2.2008418438370977e-06,
      "loss": 1.8027,
      "step": 221514
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.5559217929840088,
      "learning_rate": 2.200347294641003e-06,
      "loss": 1.7967,
      "step": 221515
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.4873936176300049,
      "learning_rate": 2.199852800811519e-06,
      "loss": 1.8705,
      "step": 221516
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.5421391725540161,
      "learning_rate": 2.19935836234868e-06,
      "loss": 2.1048,
      "step": 221517
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.7503806352615356,
      "learning_rate": 2.1988639792526184e-06,
      "loss": 1.8637,
      "step": 221518
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.529343843460083,
      "learning_rate": 2.1983696515234017e-06,
      "loss": 1.844,
      "step": 221519
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.4790655374526978,
      "learning_rate": 2.1978753791611293e-06,
      "loss": 1.9721,
      "step": 221520
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.6055063009262085,
      "learning_rate": 2.197381162165901e-06,
      "loss": 2.0602,
      "step": 221521
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.4452265501022339,
      "learning_rate": 2.1968870005377837e-06,
      "loss": 1.8683,
      "step": 221522
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.3763651847839355,
      "learning_rate": 2.1963928942769104e-06,
      "loss": 2.041,
      "step": 221523
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.4046343564987183,
      "learning_rate": 2.195898843383348e-06,
      "loss": 1.9924,
      "step": 221524
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.5186457633972168,
      "learning_rate": 2.1954048478571627e-06,
      "loss": 1.8635,
      "step": 221525
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.4643187522888184,
      "learning_rate": 2.194910907698488e-06,
      "loss": 1.9173,
      "step": 221526
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.5262417793273926,
      "learning_rate": 2.194417022907391e-06,
      "loss": 2.2203,
      "step": 221527
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.4941916465759277,
      "learning_rate": 2.193923193484004e-06,
      "loss": 1.8531,
      "step": 221528
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.60465669631958,
      "learning_rate": 2.193429419428361e-06,
      "loss": 2.1941,
      "step": 221529
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.4261088371276855,
      "learning_rate": 2.1929357007405945e-06,
      "loss": 2.0704,
      "step": 221530
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.5322942733764648,
      "learning_rate": 2.192442037420772e-06,
      "loss": 1.961,
      "step": 221531
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.5261176824569702,
      "learning_rate": 2.1919484294689928e-06,
      "loss": 1.8638,
      "step": 221532
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.3980499505996704,
      "learning_rate": 2.1914548768853903e-06,
      "loss": 1.9763,
      "step": 221533
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.7620079517364502,
      "learning_rate": 2.190961379669964e-06,
      "loss": 2.0593,
      "step": 221534
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.430204153060913,
      "learning_rate": 2.1904679378228816e-06,
      "loss": 1.8054,
      "step": 221535
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.3881745338439941,
      "learning_rate": 2.1899745513442426e-06,
      "loss": 1.8474,
      "step": 221536
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.5423730611801147,
      "learning_rate": 2.1894812202340462e-06,
      "loss": 2.1141,
      "step": 221537
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.6357215642929077,
      "learning_rate": 2.188987944492493e-06,
      "loss": 1.919,
      "step": 221538
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.3297191858291626,
      "learning_rate": 2.188494724119616e-06,
      "loss": 2.0004,
      "step": 221539
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.40103280544281,
      "learning_rate": 2.1880015591155487e-06,
      "loss": 1.8483,
      "step": 221540
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.8234120607376099,
      "learning_rate": 2.187508449480291e-06,
      "loss": 2.0819,
      "step": 221541
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.4635018110275269,
      "learning_rate": 2.1870153952140423e-06,
      "loss": 2.1801,
      "step": 221542
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.5087774991989136,
      "learning_rate": 2.1865223963168366e-06,
      "loss": 2.0055,
      "step": 221543
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.6315418481826782,
      "learning_rate": 2.186029452788773e-06,
      "loss": 2.1131,
      "step": 221544
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.4583183526992798,
      "learning_rate": 2.185536564629986e-06,
      "loss": 1.9213,
      "step": 221545
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.3476930856704712,
      "learning_rate": 2.1850437318404745e-06,
      "loss": 2.0362,
      "step": 221546
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.5559635162353516,
      "learning_rate": 2.1845509544204053e-06,
      "loss": 1.8891,
      "step": 221547
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.5441334247589111,
      "learning_rate": 2.1840582323698786e-06,
      "loss": 2.0715,
      "step": 221548
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.430067539215088,
      "learning_rate": 2.183565565688927e-06,
      "loss": 2.1995,
      "step": 221549
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.4525790214538574,
      "learning_rate": 2.1830729543776846e-06,
      "loss": 2.0611,
      "step": 221550
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.4400650262832642,
      "learning_rate": 2.1825803984362176e-06,
      "loss": 1.8361,
      "step": 221551
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.5117701292037964,
      "learning_rate": 2.182087897864626e-06,
      "loss": 2.0356,
      "step": 221552
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.4551784992218018,
      "learning_rate": 2.1815954526630096e-06,
      "loss": 1.915,
      "step": 221553
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.4296690225601196,
      "learning_rate": 2.1811030628314684e-06,
      "loss": 2.1404,
      "step": 221554
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.5281211137771606,
      "learning_rate": 2.180610728370069e-06,
      "loss": 2.0813,
      "step": 221555
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.4794682264328003,
      "learning_rate": 2.180118449278945e-06,
      "loss": 1.9919,
      "step": 221556
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.3685023784637451,
      "learning_rate": 2.1796262255581287e-06,
      "loss": 2.1181,
      "step": 221557
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.4933949708938599,
      "learning_rate": 2.179134057207754e-06,
      "loss": 1.9281,
      "step": 221558
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.5186890363693237,
      "learning_rate": 2.178641944227888e-06,
      "loss": 2.0021,
      "step": 221559
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.8501815795898438,
      "learning_rate": 2.17814988661863e-06,
      "loss": 1.9821,
      "step": 221560
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.4442349672317505,
      "learning_rate": 2.177657884380113e-06,
      "loss": 1.8258,
      "step": 221561
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.4491647481918335,
      "learning_rate": 2.1771659375123707e-06,
      "loss": 2.1463,
      "step": 221562
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.4682345390319824,
      "learning_rate": 2.176674046015503e-06,
      "loss": 2.0181,
      "step": 221563
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.6697834730148315,
      "learning_rate": 2.1761822098896097e-06,
      "loss": 2.012,
      "step": 221564
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.3989051580429077,
      "learning_rate": 2.1756904291348244e-06,
      "loss": 1.9467,
      "step": 221565
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.4737519025802612,
      "learning_rate": 2.1751987037511463e-06,
      "loss": 1.8959,
      "step": 221566
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.57809579372406,
      "learning_rate": 2.1747070337387763e-06,
      "loss": 1.9811,
      "step": 221567
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.6588445901870728,
      "learning_rate": 2.1742154190977133e-06,
      "loss": 1.9783,
      "step": 221568
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.672961711883545,
      "learning_rate": 2.1737238598280915e-06,
      "loss": 1.898,
      "step": 221569
    },
    {
      "epoch": 2.88,
      "grad_norm": 1.4491339921951294,
      "learning_rate": 2.1732323559300104e-06,
      "loss": 1.9635,
      "step": 221570
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.6311243772506714,
      "learning_rate": 2.172740907403536e-06,
      "loss": 2.1707,
      "step": 221571
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.555959701538086,
      "learning_rate": 2.1722495142487363e-06,
      "loss": 2.1266,
      "step": 221572
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.4746711254119873,
      "learning_rate": 2.1717581764658097e-06,
      "loss": 2.0623,
      "step": 221573
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.4324970245361328,
      "learning_rate": 2.171266894054724e-06,
      "loss": 1.7869,
      "step": 221574
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.332715392112732,
      "learning_rate": 2.170775667015645e-06,
      "loss": 2.0669,
      "step": 221575
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.7381377220153809,
      "learning_rate": 2.1702844953486065e-06,
      "loss": 1.8615,
      "step": 221576
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.5090440511703491,
      "learning_rate": 2.1697933790537745e-06,
      "loss": 1.8604,
      "step": 221577
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.474419116973877,
      "learning_rate": 2.1693023181311832e-06,
      "loss": 1.858,
      "step": 221578
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.3696668148040771,
      "learning_rate": 2.1688113125809315e-06,
      "loss": 2.0076,
      "step": 221579
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.6120840311050415,
      "learning_rate": 2.168320362403153e-06,
      "loss": 1.8731,
      "step": 221580
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.4335110187530518,
      "learning_rate": 2.1678294675978813e-06,
      "loss": 1.9347,
      "step": 221581
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.5169192552566528,
      "learning_rate": 2.1673386281652162e-06,
      "loss": 2.0145,
      "step": 221582
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.6021186113357544,
      "learning_rate": 2.1668478441052905e-06,
      "loss": 1.8422,
      "step": 221583
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.47049081325531,
      "learning_rate": 2.166357115418138e-06,
      "loss": 1.9763,
      "step": 221584
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.4510151147842407,
      "learning_rate": 2.165866442103925e-06,
      "loss": 2.0801,
      "step": 221585
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.3784815073013306,
      "learning_rate": 2.165375824162685e-06,
      "loss": 2.0419,
      "step": 221586
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.3651360273361206,
      "learning_rate": 2.1648852615945176e-06,
      "loss": 1.7557,
      "step": 221587
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.4925553798675537,
      "learning_rate": 2.1643947543995232e-06,
      "loss": 2.0835,
      "step": 221588
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.490987777709961,
      "learning_rate": 2.1639043025778013e-06,
      "loss": 1.9393,
      "step": 221589
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.5427556037902832,
      "learning_rate": 2.1634139061293853e-06,
      "loss": 2.1594,
      "step": 221590
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.4298959970474243,
      "learning_rate": 2.162923565054475e-06,
      "loss": 1.8561,
      "step": 221591
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.5366969108581543,
      "learning_rate": 2.162433279353071e-06,
      "loss": 1.9812,
      "step": 221592
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.4830915927886963,
      "learning_rate": 2.161943049025272e-06,
      "loss": 1.8387,
      "step": 221593
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.485170841217041,
      "learning_rate": 2.161452874071212e-06,
      "loss": 1.9996,
      "step": 221594
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.5717616081237793,
      "learning_rate": 2.160962754490958e-06,
      "loss": 1.9903,
      "step": 221595
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.4272946119308472,
      "learning_rate": 2.160472690284609e-06,
      "loss": 2.0908,
      "step": 221596
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.518546223640442,
      "learning_rate": 2.1599826814522324e-06,
      "loss": 2.0849,
      "step": 221597
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.4254505634307861,
      "learning_rate": 2.159492727993961e-06,
      "loss": 2.0969,
      "step": 221598
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.520228624343872,
      "learning_rate": 2.1590028299098616e-06,
      "loss": 2.1015,
      "step": 221599
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.39633309841156,
      "learning_rate": 2.1585129872000005e-06,
      "loss": 1.9533,
      "step": 221600
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.3258190155029297,
      "learning_rate": 2.1580231998645114e-06,
      "loss": 1.7179,
      "step": 221601
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.5090962648391724,
      "learning_rate": 2.1575334679034606e-06,
      "loss": 1.8955,
      "step": 221602
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.3591077327728271,
      "learning_rate": 2.157043791316948e-06,
      "loss": 1.9319,
      "step": 221603
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.437765121459961,
      "learning_rate": 2.1565541701050404e-06,
      "loss": 1.907,
      "step": 221604
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.374592900276184,
      "learning_rate": 2.156064604267904e-06,
      "loss": 2.1162,
      "step": 221605
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.539069414138794,
      "learning_rate": 2.1555750938055393e-06,
      "loss": 1.9012,
      "step": 221606
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.58738112449646,
      "learning_rate": 2.1550856387181125e-06,
      "loss": 1.8174,
      "step": 221607
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.6213454008102417,
      "learning_rate": 2.1545962390056236e-06,
      "loss": 2.1514,
      "step": 221608
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.438982605934143,
      "learning_rate": 2.154106894668239e-06,
      "loss": 2.0241,
      "step": 221609
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.487324595451355,
      "learning_rate": 2.1536176057060263e-06,
      "loss": 1.8978,
      "step": 221610
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.450458288192749,
      "learning_rate": 2.153128372119117e-06,
      "loss": 2.0918,
      "step": 221611
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.5766946077346802,
      "learning_rate": 2.152639193907513e-06,
      "loss": 2.0879,
      "step": 221612
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.4793956279754639,
      "learning_rate": 2.152150071071379e-06,
      "loss": 1.9981,
      "step": 221613
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.4322509765625,
      "learning_rate": 2.151661003610783e-06,
      "loss": 1.8682,
      "step": 221614
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.3734055757522583,
      "learning_rate": 2.151171991525824e-06,
      "loss": 2.1244,
      "step": 221615
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.4516289234161377,
      "learning_rate": 2.1506830348165695e-06,
      "loss": 1.8636,
      "step": 221616
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.4635205268859863,
      "learning_rate": 2.1501941334831183e-06,
      "loss": 1.9871,
      "step": 221617
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.2967132329940796,
      "learning_rate": 2.1497052875255714e-06,
      "loss": 2.014,
      "step": 221618
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.483113169670105,
      "learning_rate": 2.1492164969440286e-06,
      "loss": 2.0395,
      "step": 221619
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.496809959411621,
      "learning_rate": 2.1487277617385556e-06,
      "loss": 2.0873,
      "step": 221620
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.5017045736312866,
      "learning_rate": 2.1482390819092867e-06,
      "loss": 2.1424,
      "step": 221621
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.5972658395767212,
      "learning_rate": 2.1477504574562545e-06,
      "loss": 2.0017,
      "step": 221622
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.3705867528915405,
      "learning_rate": 2.147261888379559e-06,
      "loss": 2.1685,
      "step": 221623
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.6122634410858154,
      "learning_rate": 2.146773374679367e-06,
      "loss": 2.2506,
      "step": 221624
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.4832841157913208,
      "learning_rate": 2.1462849163556785e-06,
      "loss": 2.0034,
      "step": 221625
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.4473252296447754,
      "learning_rate": 2.1457965134085932e-06,
      "loss": 1.9261,
      "step": 221626
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.3372066020965576,
      "learning_rate": 2.1453081658382775e-06,
      "loss": 1.813,
      "step": 221627
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.515656590461731,
      "learning_rate": 2.144819873644732e-06,
      "loss": 2.1076,
      "step": 221628
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.4610552787780762,
      "learning_rate": 2.1443316368280894e-06,
      "loss": 2.0515,
      "step": 221629
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.4047101736068726,
      "learning_rate": 2.143843455388483e-06,
      "loss": 2.1664,
      "step": 221630
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.4937145709991455,
      "learning_rate": 2.143355329325913e-06,
      "loss": 2.0045,
      "step": 221631
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.574979305267334,
      "learning_rate": 2.1428672586405126e-06,
      "loss": 2.29,
      "step": 221632
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.4996140003204346,
      "learning_rate": 2.142379243332415e-06,
      "loss": 2.1272,
      "step": 221633
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.5417110919952393,
      "learning_rate": 2.14189128340162e-06,
      "loss": 1.9785,
      "step": 221634
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.5495069026947021,
      "learning_rate": 2.1414033788483278e-06,
      "loss": 2.0404,
      "step": 221635
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.4480199813842773,
      "learning_rate": 2.140915529672538e-06,
      "loss": 2.0188,
      "step": 221636
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.5844879150390625,
      "learning_rate": 2.1404277358743504e-06,
      "loss": 2.1848,
      "step": 221637
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.457173228263855,
      "learning_rate": 2.1399399974539323e-06,
      "loss": 1.9822,
      "step": 221638
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.6848680973052979,
      "learning_rate": 2.139452314411283e-06,
      "loss": 2.0848,
      "step": 221639
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.43099844455719,
      "learning_rate": 2.138964686746536e-06,
      "loss": 1.7844,
      "step": 221640
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.5522515773773193,
      "learning_rate": 2.1384771144597912e-06,
      "loss": 1.9077,
      "step": 221641
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.5252872705459595,
      "learning_rate": 2.1379895975511486e-06,
      "loss": 2.1413,
      "step": 221642
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.333936333656311,
      "learning_rate": 2.1375021360206413e-06,
      "loss": 2.0256,
      "step": 221643
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.5898644924163818,
      "learning_rate": 2.1370147298684027e-06,
      "loss": 1.9383,
      "step": 221644
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.5768193006515503,
      "learning_rate": 2.1365273790944993e-06,
      "loss": 1.9562,
      "step": 221645
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.5172423124313354,
      "learning_rate": 2.136040083699064e-06,
      "loss": 2.117,
      "step": 221646
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.3761101961135864,
      "learning_rate": 2.1355528436821644e-06,
      "loss": 1.9181,
      "step": 221647
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.419049859046936,
      "learning_rate": 2.1350656590438663e-06,
      "loss": 1.9436,
      "step": 221648
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.502181887626648,
      "learning_rate": 2.134578529784303e-06,
      "loss": 2.0501,
      "step": 221649
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.668699026107788,
      "learning_rate": 2.1340914559035415e-06,
      "loss": 1.9927,
      "step": 221650
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.5695146322250366,
      "learning_rate": 2.1336044374016483e-06,
      "loss": 1.9943,
      "step": 221651
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.370621919631958,
      "learning_rate": 2.133117474278756e-06,
      "loss": 2.0717,
      "step": 221652
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.4550861120224,
      "learning_rate": 2.132630566534965e-06,
      "loss": 2.0528,
      "step": 221653
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.7756356000900269,
      "learning_rate": 2.132143714170309e-06,
      "loss": 2.0545,
      "step": 221654
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.4923505783081055,
      "learning_rate": 2.1316569171849206e-06,
      "loss": 2.1041,
      "step": 221655
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.5082672834396362,
      "learning_rate": 2.1311701755789e-06,
      "loss": 1.9704,
      "step": 221656
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.4796565771102905,
      "learning_rate": 2.1306834893522805e-06,
      "loss": 2.0142,
      "step": 221657
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.5912688970565796,
      "learning_rate": 2.1301968585051952e-06,
      "loss": 2.0607,
      "step": 221658
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.5609183311462402,
      "learning_rate": 2.1297102830377445e-06,
      "loss": 1.8659,
      "step": 221659
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.467760443687439,
      "learning_rate": 2.1292237629500274e-06,
      "loss": 2.2215,
      "step": 221660
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.5548547506332397,
      "learning_rate": 2.1287372982420782e-06,
      "loss": 2.0255,
      "step": 221661
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.545425295829773,
      "learning_rate": 2.128250888914029e-06,
      "loss": 1.9627,
      "step": 221662
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.4070981740951538,
      "learning_rate": 2.127764534965948e-06,
      "loss": 1.7392,
      "step": 221663
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.624412178993225,
      "learning_rate": 2.1272782363979333e-06,
      "loss": 1.8497,
      "step": 221664
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.583591103553772,
      "learning_rate": 2.1267919932100864e-06,
      "loss": 1.8628,
      "step": 221665
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.5255539417266846,
      "learning_rate": 2.1263058054025063e-06,
      "loss": 1.9521,
      "step": 221666
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.4505761861801147,
      "learning_rate": 2.1258196729752596e-06,
      "loss": 1.873,
      "step": 221667
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.5715047121047974,
      "learning_rate": 2.1253335959284466e-06,
      "loss": 1.9301,
      "step": 221668
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.4163621664047241,
      "learning_rate": 2.1248475742621672e-06,
      "loss": 1.9969,
      "step": 221669
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.575819730758667,
      "learning_rate": 2.124361607976488e-06,
      "loss": 1.9805,
      "step": 221670
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.3982592821121216,
      "learning_rate": 2.1238756970715086e-06,
      "loss": 1.9929,
      "step": 221671
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.516974925994873,
      "learning_rate": 2.1233898415473626e-06,
      "loss": 1.9933,
      "step": 221672
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.5350877046585083,
      "learning_rate": 2.1229040414040497e-06,
      "loss": 1.9196,
      "step": 221673
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.4415656328201294,
      "learning_rate": 2.122418296641737e-06,
      "loss": 2.0197,
      "step": 221674
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.4689358472824097,
      "learning_rate": 2.121932607260457e-06,
      "loss": 2.1166,
      "step": 221675
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.4609800577163696,
      "learning_rate": 2.121446973260377e-06,
      "loss": 1.8973,
      "step": 221676
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.6280851364135742,
      "learning_rate": 2.1209613946415292e-06,
      "loss": 2.0494,
      "step": 221677
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.6022229194641113,
      "learning_rate": 2.120475871404015e-06,
      "loss": 1.989,
      "step": 221678
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.4713881015777588,
      "learning_rate": 2.119990403547933e-06,
      "loss": 2.1142,
      "step": 221679
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.5693219900131226,
      "learning_rate": 2.1195049910733507e-06,
      "loss": 1.9642,
      "step": 221680
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.62236750125885,
      "learning_rate": 2.119019633980401e-06,
      "loss": 1.9742,
      "step": 221681
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.6762092113494873,
      "learning_rate": 2.1185343322691173e-06,
      "loss": 1.6816,
      "step": 221682
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.5169191360473633,
      "learning_rate": 2.1180490859396327e-06,
      "loss": 2.0487,
      "step": 221683
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.4478124380111694,
      "learning_rate": 2.117563894992047e-06,
      "loss": 2.1391,
      "step": 221684
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.4593884944915771,
      "learning_rate": 2.117078759426394e-06,
      "loss": 1.9812,
      "step": 221685
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.4972935914993286,
      "learning_rate": 2.1165936792428395e-06,
      "loss": 1.891,
      "step": 221686
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.5390782356262207,
      "learning_rate": 2.116108654441384e-06,
      "loss": 2.0253,
      "step": 221687
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.4601173400878906,
      "learning_rate": 2.1156236850222274e-06,
      "loss": 2.115,
      "step": 221688
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.4561383724212646,
      "learning_rate": 2.115138770985336e-06,
      "loss": 1.929,
      "step": 221689
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.4397943019866943,
      "learning_rate": 2.1146539123309103e-06,
      "loss": 2.163,
      "step": 221690
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.4885520935058594,
      "learning_rate": 2.114169109058983e-06,
      "loss": 2.1835,
      "step": 221691
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.4773468971252441,
      "learning_rate": 2.113684361169654e-06,
      "loss": 2.1259,
      "step": 221692
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.4306647777557373,
      "learning_rate": 2.1131996686629905e-06,
      "loss": 2.1319,
      "step": 221693
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.5221099853515625,
      "learning_rate": 2.112715031539125e-06,
      "loss": 2.1889,
      "step": 221694
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.4251163005828857,
      "learning_rate": 2.1122304497981245e-06,
      "loss": 1.9292,
      "step": 221695
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.5243818759918213,
      "learning_rate": 2.1117459234401224e-06,
      "loss": 2.0753,
      "step": 221696
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.3688404560089111,
      "learning_rate": 2.1112614524651184e-06,
      "loss": 2.0659,
      "step": 221697
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.450571894645691,
      "learning_rate": 2.110777036873279e-06,
      "loss": 1.8933,
      "step": 221698
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.3600019216537476,
      "learning_rate": 2.1102926766646376e-06,
      "loss": 1.6586,
      "step": 221699
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.378841757774353,
      "learning_rate": 2.109808371839361e-06,
      "loss": 1.9645,
      "step": 221700
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.3244749307632446,
      "learning_rate": 2.109324122397449e-06,
      "loss": 2.0771,
      "step": 221701
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.495216727256775,
      "learning_rate": 2.1088399283390677e-06,
      "loss": 1.9812,
      "step": 221702
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.3216735124588013,
      "learning_rate": 2.1083557896642513e-06,
      "loss": 1.9037,
      "step": 221703
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.4309630393981934,
      "learning_rate": 2.1078717063731653e-06,
      "loss": 1.9914,
      "step": 221704
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.464937686920166,
      "learning_rate": 2.1073876784657774e-06,
      "loss": 1.982,
      "step": 221705
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.440167784690857,
      "learning_rate": 2.10690370594232e-06,
      "loss": 1.9287,
      "step": 221706
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.5409399271011353,
      "learning_rate": 2.10641978880276e-06,
      "loss": 1.8584,
      "step": 221707
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.4131029844284058,
      "learning_rate": 2.105935927047264e-06,
      "loss": 1.9613,
      "step": 221708
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.361624836921692,
      "learning_rate": 2.105452120675899e-06,
      "loss": 1.8368,
      "step": 221709
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.6349817514419556,
      "learning_rate": 2.104968369688764e-06,
      "loss": 1.9446,
      "step": 221710
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.4787768125534058,
      "learning_rate": 2.1044846740858934e-06,
      "loss": 1.88,
      "step": 221711
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.5833638906478882,
      "learning_rate": 2.1040010338674527e-06,
      "loss": 2.0945,
      "step": 221712
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.3987847566604614,
      "learning_rate": 2.1035174490335094e-06,
      "loss": 1.905,
      "step": 221713
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.4599663019180298,
      "learning_rate": 2.1030339195841295e-06,
      "loss": 2.1717,
      "step": 221714
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.4728529453277588,
      "learning_rate": 2.102550445519413e-06,
      "loss": 1.9858,
      "step": 221715
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.4416368007659912,
      "learning_rate": 2.1020670268394933e-06,
      "loss": 2.1944,
      "step": 221716
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.497840166091919,
      "learning_rate": 2.1015836635443708e-06,
      "loss": 2.0593,
      "step": 221717
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.3190785646438599,
      "learning_rate": 2.101100355634244e-06,
      "loss": 1.9659,
      "step": 221718
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.4432882070541382,
      "learning_rate": 2.100617103109081e-06,
      "loss": 1.8642,
      "step": 221719
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.3898035287857056,
      "learning_rate": 2.1001339059690816e-06,
      "loss": 2.0249,
      "step": 221720
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.4147846698760986,
      "learning_rate": 2.099650764214278e-06,
      "loss": 2.0147,
      "step": 221721
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.3965290784835815,
      "learning_rate": 2.0991676778447706e-06,
      "loss": 1.7814,
      "step": 221722
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.4984489679336548,
      "learning_rate": 2.09868464686066e-06,
      "loss": 1.8263,
      "step": 221723
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.4798496961593628,
      "learning_rate": 2.0982016712620452e-06,
      "loss": 1.9194,
      "step": 221724
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.6001375913619995,
      "learning_rate": 2.09771875104896e-06,
      "loss": 2.1563,
      "step": 221725
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.5749510526657104,
      "learning_rate": 2.0972358862215376e-06,
      "loss": 2.179,
      "step": 221726
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.4320112466812134,
      "learning_rate": 2.096753076779878e-06,
      "loss": 1.9383,
      "step": 221727
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.5008608102798462,
      "learning_rate": 2.0962703227240474e-06,
      "loss": 2.0251,
      "step": 221728
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.551232099533081,
      "learning_rate": 2.095787624054146e-06,
      "loss": 2.1261,
      "step": 221729
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.4791568517684937,
      "learning_rate": 2.0953049807702405e-06,
      "loss": 1.8392,
      "step": 221730
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.4447373151779175,
      "learning_rate": 2.094822392872497e-06,
      "loss": 2.0149,
      "step": 221731
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.3663424253463745,
      "learning_rate": 2.0943398603609163e-06,
      "loss": 1.9714,
      "step": 221732
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.4886170625686646,
      "learning_rate": 2.0938573832355976e-06,
      "loss": 2.1753,
      "step": 221733
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.33640456199646,
      "learning_rate": 2.0933749614966746e-06,
      "loss": 2.0301,
      "step": 221734
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.3436402082443237,
      "learning_rate": 2.092892595144213e-06,
      "loss": 2.1132,
      "step": 221735
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.471296787261963,
      "learning_rate": 2.092410284178314e-06,
      "loss": 2.0641,
      "step": 221736
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.6237789392471313,
      "learning_rate": 2.091928028599077e-06,
      "loss": 2.0392,
      "step": 221737
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.4612491130828857,
      "learning_rate": 2.091445828406535e-06,
      "loss": 1.7959,
      "step": 221738
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.3917105197906494,
      "learning_rate": 2.0909636836008547e-06,
      "loss": 1.8608,
      "step": 221739
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.4444148540496826,
      "learning_rate": 2.0904815941820364e-06,
      "loss": 1.8859,
      "step": 221740
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.3974910974502563,
      "learning_rate": 2.0899995601502793e-06,
      "loss": 1.974,
      "step": 221741
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.4792640209197998,
      "learning_rate": 2.0895175815055844e-06,
      "loss": 1.9201,
      "step": 221742
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.324726939201355,
      "learning_rate": 2.0890356582480504e-06,
      "loss": 1.9319,
      "step": 221743
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.4836761951446533,
      "learning_rate": 2.0885537903778115e-06,
      "loss": 1.9637,
      "step": 221744
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.4080709218978882,
      "learning_rate": 2.0880719778949672e-06,
      "loss": 1.8676,
      "step": 221745
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.3975532054901123,
      "learning_rate": 2.0875902207995176e-06,
      "loss": 2.2018,
      "step": 221746
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.4031336307525635,
      "learning_rate": 2.0871085190916625e-06,
      "loss": 2.0013,
      "step": 221747
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.4691259860992432,
      "learning_rate": 2.0866268727713685e-06,
      "loss": 2.0692,
      "step": 221748
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.4999170303344727,
      "learning_rate": 2.086145281838869e-06,
      "loss": 1.7679,
      "step": 221749
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.4413992166519165,
      "learning_rate": 2.0856637462941307e-06,
      "loss": 1.9924,
      "step": 221750
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.395740032196045,
      "learning_rate": 2.0851822661373195e-06,
      "loss": 1.6848,
      "step": 221751
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.4668724536895752,
      "learning_rate": 2.0847008413684696e-06,
      "loss": 1.7951,
      "step": 221752
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.500614881515503,
      "learning_rate": 2.0842194719877136e-06,
      "loss": 1.9836,
      "step": 221753
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.4788196086883545,
      "learning_rate": 2.0837381579950853e-06,
      "loss": 1.9337,
      "step": 221754
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.5687910318374634,
      "learning_rate": 2.083256899390784e-06,
      "loss": 1.8399,
      "step": 221755
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.5516480207443237,
      "learning_rate": 2.0827756961747765e-06,
      "loss": 2.1644,
      "step": 221756
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.4687775373458862,
      "learning_rate": 2.08229454834723e-06,
      "loss": 2.0598,
      "step": 221757
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.4587150812149048,
      "learning_rate": 2.0818134559081768e-06,
      "loss": 2.0897,
      "step": 221758
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.507318139076233,
      "learning_rate": 2.081332418857784e-06,
      "loss": 1.9971,
      "step": 221759
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.4232546091079712,
      "learning_rate": 2.0808514371960515e-06,
      "loss": 2.1229,
      "step": 221760
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.421543002128601,
      "learning_rate": 2.080370510923146e-06,
      "loss": 2.0756,
      "step": 221761
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.5326882600784302,
      "learning_rate": 2.0798896400391006e-06,
      "loss": 1.9959,
      "step": 221762
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.5726795196533203,
      "learning_rate": 2.0794088245440486e-06,
      "loss": 1.7552,
      "step": 221763
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.3428549766540527,
      "learning_rate": 2.078928064438057e-06,
      "loss": 2.024,
      "step": 221764
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.4305118322372437,
      "learning_rate": 2.0784473597211916e-06,
      "loss": 1.8259,
      "step": 221765
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.5472278594970703,
      "learning_rate": 2.077966710393586e-06,
      "loss": 1.9624,
      "step": 221766
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.3936766386032104,
      "learning_rate": 2.077486116455307e-06,
      "loss": 2.0547,
      "step": 221767
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.6142975091934204,
      "learning_rate": 2.077005577906421e-06,
      "loss": 1.8655,
      "step": 221768
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.4609270095825195,
      "learning_rate": 2.0765250947470946e-06,
      "loss": 1.8452,
      "step": 221769
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.4498820304870605,
      "learning_rate": 2.076044666977328e-06,
      "loss": 2.0989,
      "step": 221770
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.468051791191101,
      "learning_rate": 2.075564294597254e-06,
      "loss": 1.8588,
      "step": 221771
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.433915138244629,
      "learning_rate": 2.075083977606973e-06,
      "loss": 1.6899,
      "step": 221772
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.2645541429519653,
      "learning_rate": 2.074603716006551e-06,
      "loss": 1.9363,
      "step": 221773
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.5653884410858154,
      "learning_rate": 2.074123509796055e-06,
      "loss": 1.8055,
      "step": 221774
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.3902376890182495,
      "learning_rate": 2.073643358975652e-06,
      "loss": 1.8747,
      "step": 221775
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.547451376914978,
      "learning_rate": 2.073163263545341e-06,
      "loss": 2.1085,
      "step": 221776
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.5749199390411377,
      "learning_rate": 2.0726832235052893e-06,
      "loss": 1.8365,
      "step": 221777
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.412337064743042,
      "learning_rate": 2.0722032388554964e-06,
      "loss": 1.8809,
      "step": 221778
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.410064697265625,
      "learning_rate": 2.0717233095961627e-06,
      "loss": 2.0122,
      "step": 221779
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.4945446252822876,
      "learning_rate": 2.0712434357272878e-06,
      "loss": 1.9371,
      "step": 221780
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.5430123805999756,
      "learning_rate": 2.070763617249005e-06,
      "loss": 2.1404,
      "step": 221781
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.4015583992004395,
      "learning_rate": 2.070283854161381e-06,
      "loss": 1.8978,
      "step": 221782
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.4466544389724731,
      "learning_rate": 2.0698041464645155e-06,
      "loss": 2.1244,
      "step": 221783
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.3921804428100586,
      "learning_rate": 2.0693244941585086e-06,
      "loss": 1.8115,
      "step": 221784
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.496643304824829,
      "learning_rate": 2.068844897243427e-06,
      "loss": 2.0467,
      "step": 221785
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.5251842737197876,
      "learning_rate": 2.0683653557193704e-06,
      "loss": 1.9574,
      "step": 221786
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.568225383758545,
      "learning_rate": 2.0678858695864386e-06,
      "loss": 1.9399,
      "step": 221787
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.8434507846832275,
      "learning_rate": 2.0674064388446987e-06,
      "loss": 2.0699,
      "step": 221788
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.4597936868667603,
      "learning_rate": 2.0669270634942836e-06,
      "loss": 1.9168,
      "step": 221789
    },
    {
      "epoch": 2.89,
      "grad_norm": 2.11592960357666,
      "learning_rate": 2.0664477435351934e-06,
      "loss": 2.098,
      "step": 221790
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.4325494766235352,
      "learning_rate": 2.065968478967628e-06,
      "loss": 1.9364,
      "step": 221791
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.6199582815170288,
      "learning_rate": 2.065489269791587e-06,
      "loss": 2.2916,
      "step": 221792
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.5961685180664062,
      "learning_rate": 2.065010116007204e-06,
      "loss": 2.0235,
      "step": 221793
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.5064622163772583,
      "learning_rate": 2.0645310176145792e-06,
      "loss": 2.0608,
      "step": 221794
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.4719488620758057,
      "learning_rate": 2.0640519746137786e-06,
      "loss": 2.0878,
      "step": 221795
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.5168880224227905,
      "learning_rate": 2.063572987004869e-06,
      "loss": 2.0218,
      "step": 221796
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.4998811483383179,
      "learning_rate": 2.0630940547879838e-06,
      "loss": 2.2361,
      "step": 221797
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.3797540664672852,
      "learning_rate": 2.0626151779631894e-06,
      "loss": 1.8478,
      "step": 221798
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.426507592201233,
      "learning_rate": 2.062136356530586e-06,
      "loss": 2.2015,
      "step": 221799
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.9842756986618042,
      "learning_rate": 2.06165759049024e-06,
      "loss": 2.2101,
      "step": 221800
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.5763567686080933,
      "learning_rate": 2.0611788798422845e-06,
      "loss": 2.176,
      "step": 221801
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.5078976154327393,
      "learning_rate": 2.0607002245867533e-06,
      "loss": 2.0319,
      "step": 221802
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.4537043571472168,
      "learning_rate": 2.060221624723779e-06,
      "loss": 2.0115,
      "step": 221803
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.592908263206482,
      "learning_rate": 2.0597430802533954e-06,
      "loss": 1.8017,
      "step": 221804
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.406553864479065,
      "learning_rate": 2.059264591175769e-06,
      "loss": 2.0723,
      "step": 221805
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.5277096033096313,
      "learning_rate": 2.0587861574909324e-06,
      "loss": 2.0509,
      "step": 221806
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.5380967855453491,
      "learning_rate": 2.0583077791989865e-06,
      "loss": 2.139,
      "step": 221807
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.371727466583252,
      "learning_rate": 2.057829456300064e-06,
      "loss": 1.9318,
      "step": 221808
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.5503478050231934,
      "learning_rate": 2.0573511887941984e-06,
      "loss": 2.075,
      "step": 221809
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.4753437042236328,
      "learning_rate": 2.056872976681456e-06,
      "loss": 2.0713,
      "step": 221810
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.407847285270691,
      "learning_rate": 2.056394819962004e-06,
      "loss": 1.8241,
      "step": 221811
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.4806013107299805,
      "learning_rate": 2.055916718635875e-06,
      "loss": 2.255,
      "step": 221812
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.4369558095932007,
      "learning_rate": 2.0554386727032025e-06,
      "loss": 2.0639,
      "step": 221813
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.6299855709075928,
      "learning_rate": 2.0549606821640196e-06,
      "loss": 2.149,
      "step": 221814
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.556472897529602,
      "learning_rate": 2.0544827470184265e-06,
      "loss": 2.1679,
      "step": 221815
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.5176751613616943,
      "learning_rate": 2.0540048672665567e-06,
      "loss": 2.1998,
      "step": 221816
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.4752404689788818,
      "learning_rate": 2.053527042908476e-06,
      "loss": 1.9421,
      "step": 221817
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.414942979812622,
      "learning_rate": 2.053049273944285e-06,
      "loss": 1.9572,
      "step": 221818
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.4637809991836548,
      "learning_rate": 2.0525715603740165e-06,
      "loss": 1.7808,
      "step": 221819
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.7482599020004272,
      "learning_rate": 2.0520939021978376e-06,
      "loss": 2.0027,
      "step": 221820
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.441806435585022,
      "learning_rate": 2.0516162994157482e-06,
      "loss": 1.9865,
      "step": 221821
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.320534110069275,
      "learning_rate": 2.0511387520279145e-06,
      "loss": 1.9402,
      "step": 221822
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.4739875793457031,
      "learning_rate": 2.050661260034403e-06,
      "loss": 1.9326,
      "step": 221823
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.4063475131988525,
      "learning_rate": 2.050183823435314e-06,
      "loss": 2.0454,
      "step": 221824
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.508985996246338,
      "learning_rate": 2.049706442230681e-06,
      "loss": 2.0121,
      "step": 221825
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.4179348945617676,
      "learning_rate": 2.04922911642067e-06,
      "loss": 1.7957,
      "step": 221826
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.5819321870803833,
      "learning_rate": 2.048751846005314e-06,
      "loss": 2.3679,
      "step": 221827
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.714497685432434,
      "learning_rate": 2.0482746309847144e-06,
      "loss": 2.0665,
      "step": 221828
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.505772590637207,
      "learning_rate": 2.0477974713589695e-06,
      "loss": 2.0765,
      "step": 221829
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.422191858291626,
      "learning_rate": 2.04732036712818e-06,
      "loss": 2.0842,
      "step": 221830
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.4417692422866821,
      "learning_rate": 2.0468433182923795e-06,
      "loss": 2.0337,
      "step": 221831
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.5319408178329468,
      "learning_rate": 2.046366324851734e-06,
      "loss": 1.877,
      "step": 221832
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.4267518520355225,
      "learning_rate": 2.0458893868062432e-06,
      "loss": 1.9259,
      "step": 221833
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.450151801109314,
      "learning_rate": 2.0454125041561077e-06,
      "loss": 2.0082,
      "step": 221834
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.4528237581253052,
      "learning_rate": 2.044935676901294e-06,
      "loss": 1.9861,
      "step": 221835
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.4936503171920776,
      "learning_rate": 2.0444589050420014e-06,
      "loss": 2.0987,
      "step": 221836
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.547243356704712,
      "learning_rate": 2.0439821885782635e-06,
      "loss": 1.8057,
      "step": 221837
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.4125396013259888,
      "learning_rate": 2.0435055275101474e-06,
      "loss": 1.8708,
      "step": 221838
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.4957842826843262,
      "learning_rate": 2.043028921837786e-06,
      "loss": 2.1709,
      "step": 221839
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.3515934944152832,
      "learning_rate": 2.0425523715612456e-06,
      "loss": 2.0268,
      "step": 221840
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.5677834749221802,
      "learning_rate": 2.0420758766805936e-06,
      "loss": 2.1372,
      "step": 221841
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.351919412612915,
      "learning_rate": 2.0415994371959954e-06,
      "loss": 2.1557,
      "step": 221842
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.7583345174789429,
      "learning_rate": 2.041123053107452e-06,
      "loss": 2.1366,
      "step": 221843
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.5681626796722412,
      "learning_rate": 2.0406467244150958e-06,
      "loss": 1.8525,
      "step": 221844
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.6117759943008423,
      "learning_rate": 2.040170451118994e-06,
      "loss": 2.1258,
      "step": 221845
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.5383914709091187,
      "learning_rate": 2.03969423321928e-06,
      "loss": 1.8211,
      "step": 221846
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.5712443590164185,
      "learning_rate": 2.039218070715987e-06,
      "loss": 1.9241,
      "step": 221847
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.5996021032333374,
      "learning_rate": 2.0387419636092474e-06,
      "loss": 1.9662,
      "step": 221848
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.5507752895355225,
      "learning_rate": 2.038265911899095e-06,
      "loss": 1.8906,
      "step": 221849
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.3692364692687988,
      "learning_rate": 2.0377899155856968e-06,
      "loss": 2.0303,
      "step": 221850
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.542393445968628,
      "learning_rate": 2.0373139746690858e-06,
      "loss": 2.036,
      "step": 221851
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.3863706588745117,
      "learning_rate": 2.0368380891493286e-06,
      "loss": 1.9357,
      "step": 221852
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.5416327714920044,
      "learning_rate": 2.0363622590265914e-06,
      "loss": 2.1726,
      "step": 221853
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.4848213195800781,
      "learning_rate": 2.035886484300908e-06,
      "loss": 1.992,
      "step": 221854
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.5453357696533203,
      "learning_rate": 2.0354107649723783e-06,
      "loss": 1.8714,
      "step": 221855
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.4098385572433472,
      "learning_rate": 2.0349351010410688e-06,
      "loss": 1.8983,
      "step": 221856
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.4214202165603638,
      "learning_rate": 2.0344594925071125e-06,
      "loss": 1.8353,
      "step": 221857
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.48136568069458,
      "learning_rate": 2.0339839393705424e-06,
      "loss": 2.2833,
      "step": 221858
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.5181902647018433,
      "learning_rate": 2.033508441631493e-06,
      "loss": 2.1671,
      "step": 221859
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.3658009767532349,
      "learning_rate": 2.0330329992900963e-06,
      "loss": 2.0118,
      "step": 221860
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.3618797063827515,
      "learning_rate": 2.0325576123463194e-06,
      "loss": 1.8688,
      "step": 221861
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.4835604429244995,
      "learning_rate": 2.032082280800329e-06,
      "loss": 2.1264,
      "step": 221862
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.430952548980713,
      "learning_rate": 2.0316070046521914e-06,
      "loss": 2.1346,
      "step": 221863
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.4080092906951904,
      "learning_rate": 2.0311317839020068e-06,
      "loss": 2.0124,
      "step": 221864
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.4612737894058228,
      "learning_rate": 2.030656618549875e-06,
      "loss": 1.9786,
      "step": 221865
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.3142716884613037,
      "learning_rate": 2.030181508595863e-06,
      "loss": 1.707,
      "step": 221866
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.4673489332199097,
      "learning_rate": 2.02970645404007e-06,
      "loss": 1.9655,
      "step": 221867
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.4809236526489258,
      "learning_rate": 2.029231454882563e-06,
      "loss": 1.8852,
      "step": 221868
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.5251877307891846,
      "learning_rate": 2.0287565111234418e-06,
      "loss": 2.1921,
      "step": 221869
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.526429533958435,
      "learning_rate": 2.0282816227628063e-06,
      "loss": 2.0208,
      "step": 221870
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.4291527271270752,
      "learning_rate": 2.0278067898007564e-06,
      "loss": 2.0763,
      "step": 221871
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.5667202472686768,
      "learning_rate": 2.027332012237359e-06,
      "loss": 1.7794,
      "step": 221872
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.5796138048171997,
      "learning_rate": 2.026857290072681e-06,
      "loss": 2.0018,
      "step": 221873
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.6035497188568115,
      "learning_rate": 2.0263826233068546e-06,
      "loss": 1.8607,
      "step": 221874
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.5235140323638916,
      "learning_rate": 2.0259080119399473e-06,
      "loss": 2.1481,
      "step": 221875
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.9207355976104736,
      "learning_rate": 2.0254334559720254e-06,
      "loss": 2.044,
      "step": 221876
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.4528875350952148,
      "learning_rate": 2.024958955403222e-06,
      "loss": 1.9618,
      "step": 221877
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.4300236701965332,
      "learning_rate": 2.024484510233604e-06,
      "loss": 1.9746,
      "step": 221878
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.528951644897461,
      "learning_rate": 2.024010120463271e-06,
      "loss": 2.1727,
      "step": 221879
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.373868703842163,
      "learning_rate": 2.0235357860922565e-06,
      "loss": 1.8667,
      "step": 221880
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.508876919746399,
      "learning_rate": 2.0230615071207267e-06,
      "loss": 2.0511,
      "step": 221881
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.4291346073150635,
      "learning_rate": 2.0225872835487156e-06,
      "loss": 1.9922,
      "step": 221882
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.620274543762207,
      "learning_rate": 2.0221131153763557e-06,
      "loss": 1.978,
      "step": 221883
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.399882197380066,
      "learning_rate": 2.021639002603681e-06,
      "loss": 2.0584,
      "step": 221884
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.6381425857543945,
      "learning_rate": 2.021164945230824e-06,
      "loss": 1.685,
      "step": 221885
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.4541643857955933,
      "learning_rate": 2.0206909432578523e-06,
      "loss": 1.8707,
      "step": 221886
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.5114825963974,
      "learning_rate": 2.0202169966848647e-06,
      "loss": 2.0648,
      "step": 221887
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.5544997453689575,
      "learning_rate": 2.019743105511962e-06,
      "loss": 2.0158,
      "step": 221888
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.392343521118164,
      "learning_rate": 2.0192692697391765e-06,
      "loss": 1.9862,
      "step": 221889
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.49893057346344,
      "learning_rate": 2.0187954893666425e-06,
      "loss": 1.8639,
      "step": 221890
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.408843755722046,
      "learning_rate": 2.0183217643944594e-06,
      "loss": 2.0224,
      "step": 221891
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.5045057535171509,
      "learning_rate": 2.0178480948226606e-06,
      "loss": 2.04,
      "step": 221892
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.5089353322982788,
      "learning_rate": 2.0173744806514125e-06,
      "loss": 1.7622,
      "step": 221893
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.4921793937683105,
      "learning_rate": 2.016900921880715e-06,
      "loss": 1.9052,
      "step": 221894
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.40333890914917,
      "learning_rate": 2.0164274185107355e-06,
      "loss": 2.1485,
      "step": 221895
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.3194794654846191,
      "learning_rate": 2.0159539705415064e-06,
      "loss": 1.9234,
      "step": 221896
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.444919228553772,
      "learning_rate": 2.015480577973161e-06,
      "loss": 1.9443,
      "step": 221897
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.4863767623901367,
      "learning_rate": 2.0150072408057327e-06,
      "loss": 1.9005,
      "step": 221898
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.327521800994873,
      "learning_rate": 2.014533959039355e-06,
      "loss": 2.0934,
      "step": 221899
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.4868730306625366,
      "learning_rate": 2.0140607326740944e-06,
      "loss": 1.9808,
      "step": 221900
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.4631842374801636,
      "learning_rate": 2.0135875617100505e-06,
      "loss": 1.8643,
      "step": 221901
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.4775177240371704,
      "learning_rate": 2.0131144461472905e-06,
      "loss": 2.079,
      "step": 221902
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.5483444929122925,
      "learning_rate": 2.012641385985947e-06,
      "loss": 1.9688,
      "step": 221903
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.4798834323883057,
      "learning_rate": 2.0121683812260536e-06,
      "loss": 1.877,
      "step": 221904
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.3573428392410278,
      "learning_rate": 2.0116954318677435e-06,
      "loss": 1.8827,
      "step": 221905
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.480543851852417,
      "learning_rate": 2.0112225379110502e-06,
      "loss": 1.7649,
      "step": 221906
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.4878429174423218,
      "learning_rate": 2.0107496993561067e-06,
      "loss": 1.9192,
      "step": 221907
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.498842477798462,
      "learning_rate": 2.0102769162030132e-06,
      "loss": 2.0612,
      "step": 221908
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.411233901977539,
      "learning_rate": 2.0098041884518355e-06,
      "loss": 2.0655,
      "step": 221909
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.4363998174667358,
      "learning_rate": 2.0093315161026413e-06,
      "loss": 1.7743,
      "step": 221910
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.5221840143203735,
      "learning_rate": 2.0088588991555297e-06,
      "loss": 2.0846,
      "step": 221911
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.5941182374954224,
      "learning_rate": 2.0083863376106015e-06,
      "loss": 2.2432,
      "step": 221912
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.9567577838897705,
      "learning_rate": 2.0079138314679554e-06,
      "loss": 2.1009,
      "step": 221913
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.442839503288269,
      "learning_rate": 2.007441380727659e-06,
      "loss": 2.1442,
      "step": 221914
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.499336838722229,
      "learning_rate": 2.006968985389812e-06,
      "loss": 1.8029,
      "step": 221915
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.4586269855499268,
      "learning_rate": 2.0064966454544473e-06,
      "loss": 1.9753,
      "step": 221916
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.5310167074203491,
      "learning_rate": 2.006024360921765e-06,
      "loss": 1.9221,
      "step": 221917
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.3895565271377563,
      "learning_rate": 2.0055521317917323e-06,
      "loss": 1.9478,
      "step": 221918
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.4968684911727905,
      "learning_rate": 2.005079958064515e-06,
      "loss": 1.8422,
      "step": 221919
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.5937607288360596,
      "learning_rate": 2.0046078397401797e-06,
      "loss": 1.8736,
      "step": 221920
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.4618035554885864,
      "learning_rate": 2.004135776818827e-06,
      "loss": 2.0101,
      "step": 221921
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.6302624940872192,
      "learning_rate": 2.003663769300523e-06,
      "loss": 1.9173,
      "step": 221922
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.5002106428146362,
      "learning_rate": 2.0031918171853677e-06,
      "loss": 2.0405,
      "step": 221923
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.433712363243103,
      "learning_rate": 2.002719920473428e-06,
      "loss": 1.789,
      "step": 221924
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.491260051727295,
      "learning_rate": 2.002248079164803e-06,
      "loss": 1.818,
      "step": 221925
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.4685779809951782,
      "learning_rate": 2.001776293259594e-06,
      "loss": 2.0484,
      "step": 221926
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.424323320388794,
      "learning_rate": 2.0013045627578993e-06,
      "loss": 1.9236,
      "step": 221927
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.4898154735565186,
      "learning_rate": 2.0008328876597535e-06,
      "loss": 1.8982,
      "step": 221928
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.5825117826461792,
      "learning_rate": 2.000361267965289e-06,
      "loss": 1.9482,
      "step": 221929
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.5775816440582275,
      "learning_rate": 1.9998897036746064e-06,
      "loss": 1.9918,
      "step": 221930
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.4223569631576538,
      "learning_rate": 1.9994181947877385e-06,
      "loss": 1.8939,
      "step": 221931
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.3394335508346558,
      "learning_rate": 1.998946741304819e-06,
      "loss": 1.9498,
      "step": 221932
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.518533706665039,
      "learning_rate": 1.998475343225914e-06,
      "loss": 2.1658,
      "step": 221933
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.563628911972046,
      "learning_rate": 1.998004000551123e-06,
      "loss": 2.1604,
      "step": 221934
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.3585255146026611,
      "learning_rate": 1.9975327132805474e-06,
      "loss": 1.9748,
      "step": 221935
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.4221466779708862,
      "learning_rate": 1.9970614814142193e-06,
      "loss": 1.9968,
      "step": 221936
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.4725641012191772,
      "learning_rate": 1.9965903049522726e-06,
      "loss": 2.0945,
      "step": 221937
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.5211677551269531,
      "learning_rate": 1.996119183894773e-06,
      "loss": 1.8924,
      "step": 221938
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.34566330909729,
      "learning_rate": 1.995648118241855e-06,
      "loss": 2.0265,
      "step": 221939
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.4922212362289429,
      "learning_rate": 1.9951771079935506e-06,
      "loss": 2.0337,
      "step": 221940
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.4225075244903564,
      "learning_rate": 1.994706153149994e-06,
      "loss": 1.9226,
      "step": 221941
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.5691709518432617,
      "learning_rate": 1.994235253711218e-06,
      "loss": 1.9821,
      "step": 221942
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.3732621669769287,
      "learning_rate": 1.9937644096773564e-06,
      "loss": 1.844,
      "step": 221943
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.5726470947265625,
      "learning_rate": 1.993293621048442e-06,
      "loss": 1.7431,
      "step": 221944
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.4458781480789185,
      "learning_rate": 1.992822887824641e-06,
      "loss": 1.9213,
      "step": 221945
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.5171759128570557,
      "learning_rate": 1.9923522100059875e-06,
      "loss": 2.0023,
      "step": 221946
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.435258150100708,
      "learning_rate": 1.991881587592581e-06,
      "loss": 1.9533,
      "step": 221947
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.5468287467956543,
      "learning_rate": 1.991411020584488e-06,
      "loss": 2.0079,
      "step": 221948
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.4315840005874634,
      "learning_rate": 1.990940508981842e-06,
      "loss": 1.7897,
      "step": 221949
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.4605337381362915,
      "learning_rate": 1.9904700527847096e-06,
      "loss": 1.9287,
      "step": 221950
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.3598991632461548,
      "learning_rate": 1.9899996519931573e-06,
      "loss": 2.0337,
      "step": 221951
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.5080655813217163,
      "learning_rate": 1.989529306607285e-06,
      "loss": 2.0709,
      "step": 221952
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.5117870569229126,
      "learning_rate": 1.989059016627226e-06,
      "loss": 1.8915,
      "step": 221953
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.596295714378357,
      "learning_rate": 1.98858878205298e-06,
      "loss": 1.9566,
      "step": 221954
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.4825267791748047,
      "learning_rate": 1.9881186028846806e-06,
      "loss": 2.1154,
      "step": 221955
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.4905638694763184,
      "learning_rate": 1.987648479122428e-06,
      "loss": 1.875,
      "step": 221956
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.5421395301818848,
      "learning_rate": 1.9871784107663213e-06,
      "loss": 2.0868,
      "step": 221957
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.450006127357483,
      "learning_rate": 1.9867083978163943e-06,
      "loss": 2.0391,
      "step": 221958
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.4068466424942017,
      "learning_rate": 1.9862384402727804e-06,
      "loss": 1.7471,
      "step": 221959
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.5177925825119019,
      "learning_rate": 1.9857685381355457e-06,
      "loss": 2.0243,
      "step": 221960
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.5261168479919434,
      "learning_rate": 1.9852986914047575e-06,
      "loss": 1.8182,
      "step": 221961
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.3392059803009033,
      "learning_rate": 1.9848289000805484e-06,
      "loss": 2.1678,
      "step": 221962
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.4575449228286743,
      "learning_rate": 1.9843591641629853e-06,
      "loss": 1.8024,
      "step": 221963
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.4467620849609375,
      "learning_rate": 1.9838894836521678e-06,
      "loss": 1.9721,
      "step": 221964
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.5679820775985718,
      "learning_rate": 1.983419858548163e-06,
      "loss": 1.977,
      "step": 221965
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.61506986618042,
      "learning_rate": 1.9829502888510373e-06,
      "loss": 1.9781,
      "step": 221966
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.5521628856658936,
      "learning_rate": 1.9824807745609573e-06,
      "loss": 2.0448,
      "step": 221967
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.4666861295700073,
      "learning_rate": 1.9820113156779227e-06,
      "loss": 2.1006,
      "step": 221968
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.4146686792373657,
      "learning_rate": 1.981541912202067e-06,
      "loss": 2.074,
      "step": 221969
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.6621897220611572,
      "learning_rate": 1.981072564133457e-06,
      "loss": 2.0133,
      "step": 221970
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.5866689682006836,
      "learning_rate": 1.9806032714722254e-06,
      "loss": 2.1033,
      "step": 221971
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.4151101112365723,
      "learning_rate": 1.9801340342183726e-06,
      "loss": 1.9009,
      "step": 221972
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.4631249904632568,
      "learning_rate": 1.979664852372098e-06,
      "loss": 1.9633,
      "step": 221973
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.4135104417800903,
      "learning_rate": 1.979195725933369e-06,
      "loss": 1.9946,
      "step": 221974
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.5955414772033691,
      "learning_rate": 1.978726654902385e-06,
      "loss": 1.9765,
      "step": 221975
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.4809314012527466,
      "learning_rate": 1.978257639279146e-06,
      "loss": 2.0222,
      "step": 221976
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.4222029447555542,
      "learning_rate": 1.9777886790637854e-06,
      "loss": 1.8277,
      "step": 221977
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.5723079442977905,
      "learning_rate": 1.977319774256403e-06,
      "loss": 2.1211,
      "step": 221978
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.67762291431427,
      "learning_rate": 1.9768509248570317e-06,
      "loss": 1.9704,
      "step": 221979
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.581248164176941,
      "learning_rate": 1.9763821308658058e-06,
      "loss": 1.9912,
      "step": 221980
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.4283509254455566,
      "learning_rate": 1.9759133922827907e-06,
      "loss": 1.9829,
      "step": 221981
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.3413788080215454,
      "learning_rate": 1.975444709108087e-06,
      "loss": 1.9983,
      "step": 221982
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.6614086627960205,
      "learning_rate": 1.9749760813417283e-06,
      "loss": 2.1511,
      "step": 221983
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.6097639799118042,
      "learning_rate": 1.9745075089839135e-06,
      "loss": 2.0915,
      "step": 221984
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.5642681121826172,
      "learning_rate": 1.9740389920346433e-06,
      "loss": 2.1986,
      "step": 221985
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.5442836284637451,
      "learning_rate": 1.973570530493984e-06,
      "loss": 2.1706,
      "step": 221986
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.4411386251449585,
      "learning_rate": 1.9731021243621026e-06,
      "loss": 2.0491,
      "step": 221987
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.768459439277649,
      "learning_rate": 1.972633773639065e-06,
      "loss": 1.9594,
      "step": 221988
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.472550392150879,
      "learning_rate": 1.972165478324905e-06,
      "loss": 2.2089,
      "step": 221989
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.5379356145858765,
      "learning_rate": 1.971697238419756e-06,
      "loss": 1.9814,
      "step": 221990
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.354201316833496,
      "learning_rate": 1.971229053923684e-06,
      "loss": 2.1458,
      "step": 221991
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.4853315353393555,
      "learning_rate": 1.9707609248367893e-06,
      "loss": 1.8723,
      "step": 221992
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.43095862865448,
      "learning_rate": 1.970292851159172e-06,
      "loss": 1.9925,
      "step": 221993
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.5126605033874512,
      "learning_rate": 1.9698248328908983e-06,
      "loss": 1.9605,
      "step": 221994
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.4042279720306396,
      "learning_rate": 1.969356870032035e-06,
      "loss": 1.761,
      "step": 221995
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.6640011072158813,
      "learning_rate": 1.9688889625827154e-06,
      "loss": 1.8773,
      "step": 221996
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.635229229927063,
      "learning_rate": 1.968421110543006e-06,
      "loss": 2.0389,
      "step": 221997
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.5384681224822998,
      "learning_rate": 1.9679533139130066e-06,
      "loss": 2.1476,
      "step": 221998
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.4427400827407837,
      "learning_rate": 1.967485572692784e-06,
      "loss": 2.0913,
      "step": 221999
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.5568702220916748,
      "learning_rate": 1.967017886882405e-06,
      "loss": 1.9724,
      "step": 222000
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.4139498472213745,
      "learning_rate": 1.9665502564820023e-06,
      "loss": 1.9894,
      "step": 222001
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.4997609853744507,
      "learning_rate": 1.966082681491643e-06,
      "loss": 1.9566,
      "step": 222002
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.5427073240280151,
      "learning_rate": 1.9656151619113936e-06,
      "loss": 2.0492,
      "step": 222003
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.5139238834381104,
      "learning_rate": 1.9651476977413873e-06,
      "loss": 1.9844,
      "step": 222004
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.4376639127731323,
      "learning_rate": 1.9646802889816905e-06,
      "loss": 2.0014,
      "step": 222005
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.4839119911193848,
      "learning_rate": 1.96421293563237e-06,
      "loss": 2.0689,
      "step": 222006
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.4063724279403687,
      "learning_rate": 1.963745637693526e-06,
      "loss": 1.825,
      "step": 222007
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.506203532218933,
      "learning_rate": 1.963278395165291e-06,
      "loss": 2.1149,
      "step": 222008
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.3356927633285522,
      "learning_rate": 1.962811208047632e-06,
      "loss": 1.9307,
      "step": 222009
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.4334391355514526,
      "learning_rate": 1.9623440763407826e-06,
      "loss": 2.001,
      "step": 222010
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.4636825323104858,
      "learning_rate": 1.961877000044709e-06,
      "loss": 2.0709,
      "step": 222011
    },
    {
      "epoch": 2.89,
      "grad_norm": 2.049771785736084,
      "learning_rate": 1.9614099791595783e-06,
      "loss": 2.0989,
      "step": 222012
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.6010798215866089,
      "learning_rate": 1.9609430136854565e-06,
      "loss": 2.0202,
      "step": 222013
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.530369758605957,
      "learning_rate": 1.960476103622377e-06,
      "loss": 2.0275,
      "step": 222014
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.678592324256897,
      "learning_rate": 1.9600092489705066e-06,
      "loss": 2.2202,
      "step": 222015
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.391357660293579,
      "learning_rate": 1.9595424497298783e-06,
      "loss": 2.0349,
      "step": 222016
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.5885531902313232,
      "learning_rate": 1.9590757059005924e-06,
      "loss": 2.0791,
      "step": 222017
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.3661235570907593,
      "learning_rate": 1.958609017482782e-06,
      "loss": 1.9266,
      "step": 222018
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.4753366708755493,
      "learning_rate": 1.9581423844764467e-06,
      "loss": 1.9794,
      "step": 222019
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.4853565692901611,
      "learning_rate": 1.9576758068817533e-06,
      "loss": 1.8804,
      "step": 222020
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.4703733921051025,
      "learning_rate": 1.957209284698735e-06,
      "loss": 1.9426,
      "step": 222021
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.4083331823349,
      "learning_rate": 1.9567428179274926e-06,
      "loss": 1.8024,
      "step": 222022
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.5387647151947021,
      "learning_rate": 1.956276406568125e-06,
      "loss": 2.1053,
      "step": 222023
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.359961748123169,
      "learning_rate": 1.9558100506206988e-06,
      "loss": 2.0468,
      "step": 222024
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.477745532989502,
      "learning_rate": 1.9553437500853473e-06,
      "loss": 2.1336,
      "step": 222025
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.39461350440979,
      "learning_rate": 1.9548775049621047e-06,
      "loss": 2.1319,
      "step": 222026
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.516785740852356,
      "learning_rate": 1.95441131525107e-06,
      "loss": 1.9888,
      "step": 222027
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.6408872604370117,
      "learning_rate": 1.9539451809523433e-06,
      "loss": 2.0802,
      "step": 222028
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.388452172279358,
      "learning_rate": 1.9534791020659913e-06,
      "loss": 2.2188,
      "step": 222029
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.4945169687271118,
      "learning_rate": 1.9530130785921137e-06,
      "loss": 2.1432,
      "step": 222030
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.3172415494918823,
      "learning_rate": 1.952547110530811e-06,
      "loss": 2.0161,
      "step": 222031
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.4667013883590698,
      "learning_rate": 1.9520811978821494e-06,
      "loss": 1.8361,
      "step": 222032
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.3688443899154663,
      "learning_rate": 1.951615340646229e-06,
      "loss": 1.8261,
      "step": 222033
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.4595836400985718,
      "learning_rate": 1.951149538823116e-06,
      "loss": 1.9159,
      "step": 222034
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.3841023445129395,
      "learning_rate": 1.9506837924129102e-06,
      "loss": 2.0551,
      "step": 222035
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.4774210453033447,
      "learning_rate": 1.9502181014157126e-06,
      "loss": 1.972,
      "step": 222036
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.6391609907150269,
      "learning_rate": 1.9497524658315888e-06,
      "loss": 2.2276,
      "step": 222037
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.4322408437728882,
      "learning_rate": 1.949286885660639e-06,
      "loss": 2.1166,
      "step": 222038
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.4723875522613525,
      "learning_rate": 1.94882136090293e-06,
      "loss": 1.9761,
      "step": 222039
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.4556970596313477,
      "learning_rate": 1.948355891558562e-06,
      "loss": 2.1193,
      "step": 222040
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.454017996788025,
      "learning_rate": 1.9478904776276337e-06,
      "loss": 2.0382,
      "step": 222041
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.5643376111984253,
      "learning_rate": 1.947425119110213e-06,
      "loss": 1.9125,
      "step": 222042
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.6381194591522217,
      "learning_rate": 1.9469598160063994e-06,
      "loss": 2.0495,
      "step": 222043
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.4592845439910889,
      "learning_rate": 1.9464945683162593e-06,
      "loss": 2.2154,
      "step": 222044
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.4285019636154175,
      "learning_rate": 1.9460293760398926e-06,
      "loss": 1.9527,
      "step": 222045
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.520595908164978,
      "learning_rate": 1.9455642391773997e-06,
      "loss": 1.9384,
      "step": 222046
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.5100252628326416,
      "learning_rate": 1.9450991577288134e-06,
      "loss": 1.8392,
      "step": 222047
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.4597924947738647,
      "learning_rate": 1.9446341316943004e-06,
      "loss": 1.892,
      "step": 222048
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.2957172393798828,
      "learning_rate": 1.944169161073894e-06,
      "loss": 1.843,
      "step": 222049
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.6642100811004639,
      "learning_rate": 1.9437042458676944e-06,
      "loss": 1.8662,
      "step": 222050
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.4954442977905273,
      "learning_rate": 1.9432393860758012e-06,
      "loss": 2.0657,
      "step": 222051
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.5496288537979126,
      "learning_rate": 1.9427745816982477e-06,
      "loss": 1.9793,
      "step": 222052
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.4488584995269775,
      "learning_rate": 1.9423098327352006e-06,
      "loss": 2.0123,
      "step": 222053
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.5399727821350098,
      "learning_rate": 1.9418451391866595e-06,
      "loss": 1.8047,
      "step": 222054
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.3595200777053833,
      "learning_rate": 1.941380501052792e-06,
      "loss": 2.1076,
      "step": 222055
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.4271326065063477,
      "learning_rate": 1.940915918333663e-06,
      "loss": 1.8866,
      "step": 222056
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.7281538248062134,
      "learning_rate": 1.940451391029307e-06,
      "loss": 2.0509,
      "step": 222057
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.550460934638977,
      "learning_rate": 1.9399869191398576e-06,
      "loss": 2.1675,
      "step": 222058
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.4921766519546509,
      "learning_rate": 1.9395225026653803e-06,
      "loss": 1.9866,
      "step": 222059
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.3552476167678833,
      "learning_rate": 1.9390581416059757e-06,
      "loss": 1.7461,
      "step": 222060
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.526136040687561,
      "learning_rate": 1.938593835961744e-06,
      "loss": 1.8555,
      "step": 222061
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.6215894222259521,
      "learning_rate": 1.9381295857327506e-06,
      "loss": 1.8388,
      "step": 222062
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.4642647504806519,
      "learning_rate": 1.937665390919063e-06,
      "loss": 1.8683,
      "step": 222063
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.476647138595581,
      "learning_rate": 1.9372012515208146e-06,
      "loss": 2.0129,
      "step": 222064
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.5251227617263794,
      "learning_rate": 1.936737167538038e-06,
      "loss": 2.0205,
      "step": 222065
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.5775036811828613,
      "learning_rate": 1.9362731389708676e-06,
      "loss": 1.9888,
      "step": 222066
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.5484362840652466,
      "learning_rate": 1.9358091658193687e-06,
      "loss": 1.8238,
      "step": 222067
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.4273126125335693,
      "learning_rate": 1.9353452480836417e-06,
      "loss": 2.0229,
      "step": 222068
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.5196013450622559,
      "learning_rate": 1.9348813857637535e-06,
      "loss": 2.2192,
      "step": 222069
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.5149178504943848,
      "learning_rate": 1.9344175788598038e-06,
      "loss": 1.9611,
      "step": 222070
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.496421217918396,
      "learning_rate": 1.9339538273718593e-06,
      "loss": 2.0066,
      "step": 222071
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.5116465091705322,
      "learning_rate": 1.9334901313000196e-06,
      "loss": 1.8957,
      "step": 222072
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.3957017660140991,
      "learning_rate": 1.9330264906443516e-06,
      "loss": 2.0183,
      "step": 222073
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.4192821979522705,
      "learning_rate": 1.9325629054049884e-06,
      "loss": 1.8822,
      "step": 222074
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.4967223405838013,
      "learning_rate": 1.932099375581997e-06,
      "loss": 1.8759,
      "step": 222075
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.378372073173523,
      "learning_rate": 1.9316359011754434e-06,
      "loss": 1.8885,
      "step": 222076
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.599977970123291,
      "learning_rate": 1.931172482185428e-06,
      "loss": 1.8941,
      "step": 222077
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.5500701665878296,
      "learning_rate": 1.930709118612017e-06,
      "loss": 1.9975,
      "step": 222078
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.600224256515503,
      "learning_rate": 1.930245810455311e-06,
      "loss": 2.0339,
      "step": 222079
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.5215407609939575,
      "learning_rate": 1.929782557715409e-06,
      "loss": 2.0233,
      "step": 222080
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.456897258758545,
      "learning_rate": 1.9293193603924115e-06,
      "loss": 1.8802,
      "step": 222081
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.3820074796676636,
      "learning_rate": 1.928856218486352e-06,
      "loss": 1.8818,
      "step": 222082
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.5349074602127075,
      "learning_rate": 1.9283931319973633e-06,
      "loss": 1.6733,
      "step": 222083
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.4265044927597046,
      "learning_rate": 1.927930100925479e-06,
      "loss": 1.9349,
      "step": 222084
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.4818346500396729,
      "learning_rate": 1.927467125270832e-06,
      "loss": 2.0362,
      "step": 222085
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.5175416469573975,
      "learning_rate": 1.927004205033522e-06,
      "loss": 2.0904,
      "step": 222086
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.4822566509246826,
      "learning_rate": 1.9265413402135833e-06,
      "loss": 2.0483,
      "step": 222087
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.4800171852111816,
      "learning_rate": 1.9260785308111483e-06,
      "loss": 2.0462,
      "step": 222088
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.6147352457046509,
      "learning_rate": 1.9256157768262504e-06,
      "loss": 2.0932,
      "step": 222089
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.569927453994751,
      "learning_rate": 1.925153078259023e-06,
      "loss": 2.2703,
      "step": 222090
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.4482207298278809,
      "learning_rate": 1.924690435109533e-06,
      "loss": 2.0092,
      "step": 222091
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.54067862033844,
      "learning_rate": 1.9242278473778792e-06,
      "loss": 2.15,
      "step": 222092
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.4732472896575928,
      "learning_rate": 1.923765315064163e-06,
      "loss": 1.8893,
      "step": 222093
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.4059480428695679,
      "learning_rate": 1.9233028381683836e-06,
      "loss": 1.9481,
      "step": 222094
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.4894285202026367,
      "learning_rate": 1.9228404166907407e-06,
      "loss": 2.1719,
      "step": 222095
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.507066011428833,
      "learning_rate": 1.9223780506312346e-06,
      "loss": 1.8304,
      "step": 222096
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.720231533050537,
      "learning_rate": 1.9219157399899988e-06,
      "loss": 1.8882,
      "step": 222097
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.6351789236068726,
      "learning_rate": 1.9214534847670993e-06,
      "loss": 1.7837,
      "step": 222098
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.381365418434143,
      "learning_rate": 1.9209912849626695e-06,
      "loss": 1.9564,
      "step": 222099
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.4138990640640259,
      "learning_rate": 1.92052914057671e-06,
      "loss": 2.0727,
      "step": 222100
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.35492742061615,
      "learning_rate": 1.920067051609353e-06,
      "loss": 1.6711,
      "step": 222101
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.4667441844940186,
      "learning_rate": 1.9196050180606658e-06,
      "loss": 1.9881,
      "step": 222102
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.5016534328460693,
      "learning_rate": 1.9191430399307816e-06,
      "loss": 2.0188,
      "step": 222103
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.5239790678024292,
      "learning_rate": 1.918681117219767e-06,
      "loss": 1.9074,
      "step": 222104
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.4346966743469238,
      "learning_rate": 1.9182192499276882e-06,
      "loss": 1.686,
      "step": 222105
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.4260412454605103,
      "learning_rate": 1.9177574380546125e-06,
      "loss": 1.8739,
      "step": 222106
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.4540382623672485,
      "learning_rate": 1.9172956816006723e-06,
      "loss": 2.0054,
      "step": 222107
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.347041130065918,
      "learning_rate": 1.916833980565935e-06,
      "loss": 1.9205,
      "step": 222108
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.4506815671920776,
      "learning_rate": 1.916372334950467e-06,
      "loss": 2.0511,
      "step": 222109
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.5397801399230957,
      "learning_rate": 1.9159107447544007e-06,
      "loss": 1.9236,
      "step": 222110
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.4700210094451904,
      "learning_rate": 1.915449209977771e-06,
      "loss": 1.9618,
      "step": 222111
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.460534930229187,
      "learning_rate": 1.914987730620676e-06,
      "loss": 1.8933,
      "step": 222112
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.4532229900360107,
      "learning_rate": 1.9145263066832506e-06,
      "loss": 1.819,
      "step": 222113
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.3531627655029297,
      "learning_rate": 1.914064938165494e-06,
      "loss": 1.8692,
      "step": 222114
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.4555999040603638,
      "learning_rate": 1.913603625067572e-06,
      "loss": 1.8745,
      "step": 222115
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.464212417602539,
      "learning_rate": 1.913142367389553e-06,
      "loss": 2.0019,
      "step": 222116
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.41657555103302,
      "learning_rate": 1.912681165131469e-06,
      "loss": 2.0014,
      "step": 222117
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.2784409523010254,
      "learning_rate": 1.9122200182934534e-06,
      "loss": 1.8108,
      "step": 222118
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.5688749551773071,
      "learning_rate": 1.9117589268756063e-06,
      "loss": 1.9363,
      "step": 222119
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.4261868000030518,
      "learning_rate": 1.9112978908779607e-06,
      "loss": 1.9528,
      "step": 222120
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.557139277458191,
      "learning_rate": 1.9108369103006504e-06,
      "loss": 1.8453,
      "step": 222121
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.486096739768982,
      "learning_rate": 1.910375985143742e-06,
      "loss": 1.885,
      "step": 222122
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.4500058889389038,
      "learning_rate": 1.9099151154073343e-06,
      "loss": 1.8731,
      "step": 222123
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.5884019136428833,
      "learning_rate": 1.909454301091462e-06,
      "loss": 2.1212,
      "step": 222124
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.4994510412216187,
      "learning_rate": 1.908993542196291e-06,
      "loss": 2.2717,
      "step": 222125
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.395035982131958,
      "learning_rate": 1.908532838721821e-06,
      "loss": 2.0251,
      "step": 222126
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.4030004739761353,
      "learning_rate": 1.9080721906682193e-06,
      "loss": 2.0013,
      "step": 222127
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.426121473312378,
      "learning_rate": 1.9076115980355187e-06,
      "loss": 1.9083,
      "step": 222128
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.4749703407287598,
      "learning_rate": 1.9071510608238194e-06,
      "loss": 2.0109,
      "step": 222129
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.344990611076355,
      "learning_rate": 1.9066905790331877e-06,
      "loss": 1.8557,
      "step": 222130
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.4899226427078247,
      "learning_rate": 1.9062301526637568e-06,
      "loss": 1.7621,
      "step": 222131
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.473917841911316,
      "learning_rate": 1.9057697817155603e-06,
      "loss": 1.9089,
      "step": 222132
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.4317666292190552,
      "learning_rate": 1.9053094661887315e-06,
      "loss": 1.9566,
      "step": 222133
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.42841374874115,
      "learning_rate": 1.9048492060833365e-06,
      "loss": 2.0413,
      "step": 222134
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.4804067611694336,
      "learning_rate": 1.9043890013994755e-06,
      "loss": 2.1298,
      "step": 222135
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.3142744302749634,
      "learning_rate": 1.9039288521371488e-06,
      "loss": 2.012,
      "step": 222136
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.462948203086853,
      "learning_rate": 1.9034687582965558e-06,
      "loss": 2.1216,
      "step": 222137
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.4643285274505615,
      "learning_rate": 1.90300871987773e-06,
      "loss": 2.0966,
      "step": 222138
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.5529283285140991,
      "learning_rate": 1.9025487368807713e-06,
      "loss": 2.089,
      "step": 222139
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.5977239608764648,
      "learning_rate": 1.902088809305713e-06,
      "loss": 2.3046,
      "step": 222140
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.4130016565322876,
      "learning_rate": 1.901628937152755e-06,
      "loss": 1.8373,
      "step": 222141
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.4796634912490845,
      "learning_rate": 1.9011691204218638e-06,
      "loss": 2.1488,
      "step": 222142
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.5459295511245728,
      "learning_rate": 1.9007093591131727e-06,
      "loss": 2.145,
      "step": 222143
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.4024215936660767,
      "learning_rate": 1.9002496532267487e-06,
      "loss": 1.9028,
      "step": 222144
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.5512197017669678,
      "learning_rate": 1.8997900027627243e-06,
      "loss": 1.8729,
      "step": 222145
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.394317388534546,
      "learning_rate": 1.8993304077211336e-06,
      "loss": 1.7784,
      "step": 222146
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.5414395332336426,
      "learning_rate": 1.8988708681021426e-06,
      "loss": 1.916,
      "step": 222147
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.3903971910476685,
      "learning_rate": 1.898411383905718e-06,
      "loss": 1.9092,
      "step": 222148
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.4568779468536377,
      "learning_rate": 1.8979519551320266e-06,
      "loss": 2.004,
      "step": 222149
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.5712716579437256,
      "learning_rate": 1.8974925817811348e-06,
      "loss": 2.1581,
      "step": 222150
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.4925405979156494,
      "learning_rate": 1.8970332638531094e-06,
      "loss": 2.1318,
      "step": 222151
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.7995766401290894,
      "learning_rate": 1.8965740013480835e-06,
      "loss": 2.013,
      "step": 222152
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.407392978668213,
      "learning_rate": 1.8961147942660903e-06,
      "loss": 1.9542,
      "step": 222153
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.3901349306106567,
      "learning_rate": 1.89565564260723e-06,
      "loss": 2.035,
      "step": 222154
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.4347174167633057,
      "learning_rate": 1.8951965463716358e-06,
      "loss": 1.8093,
      "step": 222155
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.4560918807983398,
      "learning_rate": 1.8947375055593073e-06,
      "loss": 2.0014,
      "step": 222156
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.530452013015747,
      "learning_rate": 1.8942785201703782e-06,
      "loss": 2.1626,
      "step": 222157
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.3834669589996338,
      "learning_rate": 1.893819590204948e-06,
      "loss": 1.8993,
      "step": 222158
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.5491496324539185,
      "learning_rate": 1.8933607156630837e-06,
      "loss": 2.1053,
      "step": 222159
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.5714539289474487,
      "learning_rate": 1.8929018965448516e-06,
      "loss": 2.1362,
      "step": 222160
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.451644778251648,
      "learning_rate": 1.8924431328503853e-06,
      "loss": 1.9307,
      "step": 222161
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.6516203880310059,
      "learning_rate": 1.8919844245797177e-06,
      "loss": 2.197,
      "step": 222162
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.3758236169815063,
      "learning_rate": 1.891525771732949e-06,
      "loss": 2.0796,
      "step": 222163
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.5654473304748535,
      "learning_rate": 1.8910671743101791e-06,
      "loss": 2.0898,
      "step": 222164
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.36180579662323,
      "learning_rate": 1.8906086323115077e-06,
      "loss": 1.9501,
      "step": 222165
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.259477138519287,
      "learning_rate": 1.8901501457369682e-06,
      "loss": 1.9851,
      "step": 222166
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.6837408542633057,
      "learning_rate": 1.889691714586694e-06,
      "loss": 1.9664,
      "step": 222167
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.4940944910049438,
      "learning_rate": 1.8892333388607518e-06,
      "loss": 1.9776,
      "step": 222168
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.3648964166641235,
      "learning_rate": 1.888775018559241e-06,
      "loss": 1.9482,
      "step": 222169
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.4456206560134888,
      "learning_rate": 1.8883167536822286e-06,
      "loss": 1.9862,
      "step": 222170
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.6011134386062622,
      "learning_rate": 1.8878585442298144e-06,
      "loss": 2.0009,
      "step": 222171
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.4521710872650146,
      "learning_rate": 1.887400390202065e-06,
      "loss": 2.0039,
      "step": 222172
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.4600712060928345,
      "learning_rate": 1.8869422915990474e-06,
      "loss": 1.89,
      "step": 222173
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.6049193143844604,
      "learning_rate": 1.8864842484208942e-06,
      "loss": 2.0006,
      "step": 222174
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.470292091369629,
      "learning_rate": 1.8860262606676724e-06,
      "loss": 2.1286,
      "step": 222175
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.40728759765625,
      "learning_rate": 1.8855683283394818e-06,
      "loss": 1.9948,
      "step": 222176
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.5581904649734497,
      "learning_rate": 1.8851104514363891e-06,
      "loss": 2.0634,
      "step": 222177
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.391676425933838,
      "learning_rate": 1.8846526299584607e-06,
      "loss": 2.0487,
      "step": 222178
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.5994726419448853,
      "learning_rate": 1.8841948639058302e-06,
      "loss": 1.8664,
      "step": 222179
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.45488703250885,
      "learning_rate": 1.8837371532785305e-06,
      "loss": 2.2557,
      "step": 222180
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.5924663543701172,
      "learning_rate": 1.883279498076662e-06,
      "loss": 1.8491,
      "step": 222181
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.7289618253707886,
      "learning_rate": 1.8828218983003573e-06,
      "loss": 1.9917,
      "step": 222182
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.523662805557251,
      "learning_rate": 1.8823643539496503e-06,
      "loss": 1.932,
      "step": 222183
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.4220376014709473,
      "learning_rate": 1.8819068650246071e-06,
      "loss": 1.7044,
      "step": 222184
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.4351762533187866,
      "learning_rate": 1.8814494315253613e-06,
      "loss": 1.9649,
      "step": 222185
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.4653302431106567,
      "learning_rate": 1.880992053452013e-06,
      "loss": 2.1144,
      "step": 222186
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.555503010749817,
      "learning_rate": 1.8805347308045615e-06,
      "loss": 2.0544,
      "step": 222187
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.4541409015655518,
      "learning_rate": 1.8800774635832073e-06,
      "loss": 2.0916,
      "step": 222188
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.9853862524032593,
      "learning_rate": 1.87962025178795e-06,
      "loss": 1.9293,
      "step": 222189
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.4178781509399414,
      "learning_rate": 1.8791630954188896e-06,
      "loss": 1.8878,
      "step": 222190
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.6684651374816895,
      "learning_rate": 1.878705994476093e-06,
      "loss": 1.9663,
      "step": 222191
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.4552196264266968,
      "learning_rate": 1.8782489489597264e-06,
      "loss": 1.8954,
      "step": 222192
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.4896917343139648,
      "learning_rate": 1.8777919588697899e-06,
      "loss": 1.9981,
      "step": 222193
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.5864758491516113,
      "learning_rate": 1.8773350242064166e-06,
      "loss": 1.8772,
      "step": 222194
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.5501937866210938,
      "learning_rate": 1.8768781449696402e-06,
      "loss": 2.03,
      "step": 222195
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.4977695941925049,
      "learning_rate": 1.8764213211596268e-06,
      "loss": 2.0374,
      "step": 222196
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.4762561321258545,
      "learning_rate": 1.8759645527763767e-06,
      "loss": 2.0152,
      "step": 222197
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.3750792741775513,
      "learning_rate": 1.875507839820023e-06,
      "loss": 1.7722,
      "step": 222198
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.4837745428085327,
      "learning_rate": 1.8750511822906322e-06,
      "loss": 2.0386,
      "step": 222199
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.3529467582702637,
      "learning_rate": 1.8745945801883378e-06,
      "loss": 1.7189,
      "step": 222200
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.3791507482528687,
      "learning_rate": 1.8741380335131395e-06,
      "loss": 1.7663,
      "step": 222201
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.406235694885254,
      "learning_rate": 1.873681542265204e-06,
      "loss": 2.1113,
      "step": 222202
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.3723318576812744,
      "learning_rate": 1.8732251064445313e-06,
      "loss": 1.9859,
      "step": 222203
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.657956838607788,
      "learning_rate": 1.872768726051288e-06,
      "loss": 1.9348,
      "step": 222204
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.621112585067749,
      "learning_rate": 1.8723124010855074e-06,
      "loss": 1.9565,
      "step": 222205
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.6165744066238403,
      "learning_rate": 1.871856131547289e-06,
      "loss": 2.013,
      "step": 222206
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.5756499767303467,
      "learning_rate": 1.8713999174367334e-06,
      "loss": 2.0723,
      "step": 222207
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.595542073249817,
      "learning_rate": 1.8709437587539066e-06,
      "loss": 2.2152,
      "step": 222208
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.4889708757400513,
      "learning_rate": 1.870487655498909e-06,
      "loss": 2.0849,
      "step": 222209
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.4544556140899658,
      "learning_rate": 1.870031607671807e-06,
      "loss": 2.0433,
      "step": 222210
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.6001023054122925,
      "learning_rate": 1.8695756152727004e-06,
      "loss": 1.9525,
      "step": 222211
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.3923134803771973,
      "learning_rate": 1.869119678301656e-06,
      "loss": 2.1344,
      "step": 222212
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.6245002746582031,
      "learning_rate": 1.8686637967587736e-06,
      "loss": 1.9917,
      "step": 222213
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.5518077611923218,
      "learning_rate": 1.8682079706441534e-06,
      "loss": 1.9542,
      "step": 222214
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.5893499851226807,
      "learning_rate": 1.8677521999578283e-06,
      "loss": 1.9314,
      "step": 222215
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.741233468055725,
      "learning_rate": 1.8672964846999318e-06,
      "loss": 1.9527,
      "step": 222216
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.474610686302185,
      "learning_rate": 1.8668408248705302e-06,
      "loss": 1.8537,
      "step": 222217
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.360290288925171,
      "learning_rate": 1.8663852204696906e-06,
      "loss": 1.861,
      "step": 222218
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.4167742729187012,
      "learning_rate": 1.8659296714975458e-06,
      "loss": 1.9948,
      "step": 222219
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.4942413568496704,
      "learning_rate": 1.8654741779541627e-06,
      "loss": 1.9432,
      "step": 222220
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.4780566692352295,
      "learning_rate": 1.8650187398395744e-06,
      "loss": 2.0555,
      "step": 222221
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.4364612102508545,
      "learning_rate": 1.8645633571539475e-06,
      "loss": 2.2844,
      "step": 222222
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.4065614938735962,
      "learning_rate": 1.8641080298973154e-06,
      "loss": 1.8605,
      "step": 222223
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.4323121309280396,
      "learning_rate": 1.8636527580697447e-06,
      "loss": 1.9175,
      "step": 222224
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.3988432884216309,
      "learning_rate": 1.8631975416713686e-06,
      "loss": 2.2159,
      "step": 222225
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.5050642490386963,
      "learning_rate": 1.8627423807022868e-06,
      "loss": 2.0548,
      "step": 222226
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.425525426864624,
      "learning_rate": 1.8622872751624995e-06,
      "loss": 1.9554,
      "step": 222227
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.3074493408203125,
      "learning_rate": 1.8618322250521734e-06,
      "loss": 1.8566,
      "step": 222228
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.490007758140564,
      "learning_rate": 1.8613772303713414e-06,
      "loss": 1.939,
      "step": 222229
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.4184725284576416,
      "learning_rate": 1.8609222911201039e-06,
      "loss": 2.0385,
      "step": 222230
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.5166566371917725,
      "learning_rate": 1.860467407298527e-06,
      "loss": 1.9208,
      "step": 222231
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.3966174125671387,
      "learning_rate": 1.8600125789067776e-06,
      "loss": 1.9501,
      "step": 222232
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.4842008352279663,
      "learning_rate": 1.8595578059448556e-06,
      "loss": 1.9287,
      "step": 222233
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.5344595909118652,
      "learning_rate": 1.8591030884128611e-06,
      "loss": 2.1373,
      "step": 222234
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.455271601676941,
      "learning_rate": 1.8586484263108602e-06,
      "loss": 1.9745,
      "step": 222235
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.3805925846099854,
      "learning_rate": 1.85819381963902e-06,
      "loss": 2.0465,
      "step": 222236
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.538807988166809,
      "learning_rate": 1.857739268397307e-06,
      "loss": 1.9517,
      "step": 222237
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.6550949811935425,
      "learning_rate": 1.8572847725859209e-06,
      "loss": 2.0526,
      "step": 222238
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.6109931468963623,
      "learning_rate": 1.8568303322048616e-06,
      "loss": 2.2193,
      "step": 222239
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.5140193700790405,
      "learning_rate": 1.856375947254296e-06,
      "loss": 1.8776,
      "step": 222240
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.5352526903152466,
      "learning_rate": 1.8559216177341906e-06,
      "loss": 1.8166,
      "step": 222241
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.4888876676559448,
      "learning_rate": 1.8554673436447455e-06,
      "loss": 2.1901,
      "step": 222242
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.5726109743118286,
      "learning_rate": 1.85501312498596e-06,
      "loss": 2.2732,
      "step": 222243
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.399448037147522,
      "learning_rate": 1.8545589617579682e-06,
      "loss": 1.9756,
      "step": 222244
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.4576226472854614,
      "learning_rate": 1.8541048539608694e-06,
      "loss": 1.8383,
      "step": 222245
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.5503588914871216,
      "learning_rate": 1.853650801594697e-06,
      "loss": 1.8949,
      "step": 222246
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.5521446466445923,
      "learning_rate": 1.8531968046595513e-06,
      "loss": 1.9167,
      "step": 222247
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.5444930791854858,
      "learning_rate": 1.8527428631555318e-06,
      "loss": 1.9444,
      "step": 222248
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.3948627710342407,
      "learning_rate": 1.8522889770827053e-06,
      "loss": 2.0697,
      "step": 222249
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.5164061784744263,
      "learning_rate": 1.851835146441205e-06,
      "loss": 2.1004,
      "step": 222250
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.3304767608642578,
      "learning_rate": 1.851381371231031e-06,
      "loss": 2.0225,
      "step": 222251
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.486387014389038,
      "learning_rate": 1.8509276514523497e-06,
      "loss": 2.114,
      "step": 222252
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.3495484590530396,
      "learning_rate": 1.8504739871051944e-06,
      "loss": 2.1636,
      "step": 222253
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.5202622413635254,
      "learning_rate": 1.850020378189665e-06,
      "loss": 1.9172,
      "step": 222254
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.5368865728378296,
      "learning_rate": 1.849566824705828e-06,
      "loss": 1.8028,
      "step": 222255
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.3984938859939575,
      "learning_rate": 1.849113326653784e-06,
      "loss": 1.9103,
      "step": 222256
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.6295267343521118,
      "learning_rate": 1.8486598840336653e-06,
      "loss": 1.9158,
      "step": 222257
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.4674255847930908,
      "learning_rate": 1.848206496845439e-06,
      "loss": 1.8561,
      "step": 222258
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.3562185764312744,
      "learning_rate": 1.8477531650893052e-06,
      "loss": 1.7918,
      "step": 222259
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.570732831954956,
      "learning_rate": 1.8472998887652967e-06,
      "loss": 1.9121,
      "step": 222260
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.3887720108032227,
      "learning_rate": 1.8468466678734805e-06,
      "loss": 1.9634,
      "step": 222261
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.4343767166137695,
      "learning_rate": 1.8463935024139898e-06,
      "loss": 1.8518,
      "step": 222262
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.4695172309875488,
      "learning_rate": 1.845940392386891e-06,
      "loss": 2.1131,
      "step": 222263
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.5180834531784058,
      "learning_rate": 1.8454873377922507e-06,
      "loss": 2.0408,
      "step": 222264
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.717972993850708,
      "learning_rate": 1.845034338630136e-06,
      "loss": 2.1285,
      "step": 222265
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.6247597932815552,
      "learning_rate": 1.8445813949006793e-06,
      "loss": 1.9214,
      "step": 222266
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.5425540208816528,
      "learning_rate": 1.844128506603948e-06,
      "loss": 1.8486,
      "step": 222267
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.6826037168502808,
      "learning_rate": 1.8436756737400082e-06,
      "loss": 2.0386,
      "step": 222268
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.5552068948745728,
      "learning_rate": 1.8432228963089602e-06,
      "loss": 2.0141,
      "step": 222269
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.5893237590789795,
      "learning_rate": 1.8427701743109036e-06,
      "loss": 1.9911,
      "step": 222270
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.3793830871582031,
      "learning_rate": 1.8423175077458718e-06,
      "loss": 2.0543,
      "step": 222271
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.5595245361328125,
      "learning_rate": 1.8418648966139981e-06,
      "loss": 1.8982,
      "step": 222272
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.5706276893615723,
      "learning_rate": 1.8414123409153493e-06,
      "loss": 1.992,
      "step": 222273
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.4378993511199951,
      "learning_rate": 1.8409598406499915e-06,
      "loss": 2.0895,
      "step": 222274
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.5147253274917603,
      "learning_rate": 1.8405073958180916e-06,
      "loss": 2.1178,
      "step": 222275
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.499254822731018,
      "learning_rate": 1.8400550064196163e-06,
      "loss": 2.0812,
      "step": 222276
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.4545778036117554,
      "learning_rate": 1.8396026724546986e-06,
      "loss": 1.9635,
      "step": 222277
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.4827302694320679,
      "learning_rate": 1.8391503939234388e-06,
      "loss": 1.8999,
      "step": 222278
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.5017844438552856,
      "learning_rate": 1.838698170825903e-06,
      "loss": 1.9614,
      "step": 222279
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.4512317180633545,
      "learning_rate": 1.8382460031621915e-06,
      "loss": 1.8286,
      "step": 222280
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.5185216665267944,
      "learning_rate": 1.8377938909323707e-06,
      "loss": 2.1859,
      "step": 222281
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.5421183109283447,
      "learning_rate": 1.837341834136541e-06,
      "loss": 1.9786,
      "step": 222282
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.675925850868225,
      "learning_rate": 1.8368898327747682e-06,
      "loss": 1.971,
      "step": 222283
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.430299162864685,
      "learning_rate": 1.836437886847153e-06,
      "loss": 2.0042,
      "step": 222284
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.5989747047424316,
      "learning_rate": 1.8359859963537615e-06,
      "loss": 1.9316,
      "step": 222285
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.549907922744751,
      "learning_rate": 1.835534161294694e-06,
      "loss": 1.8403,
      "step": 222286
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.4581334590911865,
      "learning_rate": 1.83508238167005e-06,
      "loss": 1.9764,
      "step": 222287
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.4821280241012573,
      "learning_rate": 1.8346306574798631e-06,
      "loss": 1.8188,
      "step": 222288
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.4591208696365356,
      "learning_rate": 1.8341789887242663e-06,
      "loss": 2.0,
      "step": 222289
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.3957006931304932,
      "learning_rate": 1.8337273754033266e-06,
      "loss": 1.91,
      "step": 222290
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.4648430347442627,
      "learning_rate": 1.8332758175171104e-06,
      "loss": 1.9071,
      "step": 222291
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.5941352844238281,
      "learning_rate": 1.8328243150657173e-06,
      "loss": 2.1014,
      "step": 222292
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.4221601486206055,
      "learning_rate": 1.8323728680492478e-06,
      "loss": 1.8617,
      "step": 222293
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.3706387281417847,
      "learning_rate": 1.8319214764677347e-06,
      "loss": 2.0698,
      "step": 222294
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.6540391445159912,
      "learning_rate": 1.8314701403213116e-06,
      "loss": 2.1366,
      "step": 222295
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.4500175714492798,
      "learning_rate": 1.831018859610045e-06,
      "loss": 2.0303,
      "step": 222296
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.4176890850067139,
      "learning_rate": 1.8305676343340346e-06,
      "loss": 1.8775,
      "step": 222297
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.5502417087554932,
      "learning_rate": 1.8301164644933474e-06,
      "loss": 2.1886,
      "step": 222298
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.4118419885635376,
      "learning_rate": 1.829665350088083e-06,
      "loss": 1.9367,
      "step": 222299
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.4521546363830566,
      "learning_rate": 1.829214291118275e-06,
      "loss": 1.9131,
      "step": 222300
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.574286699295044,
      "learning_rate": 1.8287632875840564e-06,
      "loss": 1.8367,
      "step": 222301
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.5002919435501099,
      "learning_rate": 1.8283123394854938e-06,
      "loss": 2.1215,
      "step": 222302
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.5423258543014526,
      "learning_rate": 1.8278614468227204e-06,
      "loss": 1.8638,
      "step": 222303
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.4192885160446167,
      "learning_rate": 1.8274106095957364e-06,
      "loss": 1.9965,
      "step": 222304
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.3225306272506714,
      "learning_rate": 1.8269598278046748e-06,
      "loss": 1.8649,
      "step": 222305
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.5153409242630005,
      "learning_rate": 1.8265091014496024e-06,
      "loss": 2.2354,
      "step": 222306
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.4419208765029907,
      "learning_rate": 1.826058430530619e-06,
      "loss": 2.0066,
      "step": 222307
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.6480751037597656,
      "learning_rate": 1.8256078150477915e-06,
      "loss": 2.0684,
      "step": 222308
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.5839277505874634,
      "learning_rate": 1.8251572550012195e-06,
      "loss": 2.0261,
      "step": 222309
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.5210902690887451,
      "learning_rate": 1.8247067503909697e-06,
      "loss": 1.9865,
      "step": 222310
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.477005958557129,
      "learning_rate": 1.8242563012171418e-06,
      "loss": 2.1227,
      "step": 222311
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.5412958860397339,
      "learning_rate": 1.823805907479836e-06,
      "loss": 2.0399,
      "step": 222312
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.3927977085113525,
      "learning_rate": 1.8233555691790857e-06,
      "loss": 1.7741,
      "step": 222313
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.3955519199371338,
      "learning_rate": 1.8229052863149907e-06,
      "loss": 1.9238,
      "step": 222314
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.536299705505371,
      "learning_rate": 1.8224550588876841e-06,
      "loss": 1.8894,
      "step": 222315
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.5097694396972656,
      "learning_rate": 1.8220048868971659e-06,
      "loss": 1.9243,
      "step": 222316
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.4799834489822388,
      "learning_rate": 1.8215547703436028e-06,
      "loss": 1.9569,
      "step": 222317
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.4780274629592896,
      "learning_rate": 1.8211047092270282e-06,
      "loss": 2.0138,
      "step": 222318
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.5484358072280884,
      "learning_rate": 1.8206547035475417e-06,
      "loss": 1.9904,
      "step": 222319
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.419731616973877,
      "learning_rate": 1.82020475330521e-06,
      "loss": 1.9243,
      "step": 222320
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.416697382926941,
      "learning_rate": 1.8197548585001332e-06,
      "loss": 1.8669,
      "step": 222321
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.5412945747375488,
      "learning_rate": 1.819305019132411e-06,
      "loss": 2.0422,
      "step": 222322
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.601948618888855,
      "learning_rate": 1.81885523520211e-06,
      "loss": 1.966,
      "step": 222323
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.6084420680999756,
      "learning_rate": 1.8184055067092972e-06,
      "loss": 2.1808,
      "step": 222324
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.3968793153762817,
      "learning_rate": 1.8179558336540723e-06,
      "loss": 1.8022,
      "step": 222325
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.4087692499160767,
      "learning_rate": 1.8175062160365017e-06,
      "loss": 2.0171,
      "step": 222326
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.4578981399536133,
      "learning_rate": 1.817056653856719e-06,
      "loss": 2.0388,
      "step": 222327
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.4893397092819214,
      "learning_rate": 1.816607147114757e-06,
      "loss": 2.1204,
      "step": 222328
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.4743075370788574,
      "learning_rate": 1.816157695810716e-06,
      "loss": 1.8026,
      "step": 222329
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.4533213376998901,
      "learning_rate": 1.8157082999446958e-06,
      "loss": 2.0842,
      "step": 222330
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.5026087760925293,
      "learning_rate": 1.8152589595167633e-06,
      "loss": 1.8653,
      "step": 222331
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.5487117767333984,
      "learning_rate": 1.8148096745269846e-06,
      "loss": 2.1163,
      "step": 222332
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.5084152221679688,
      "learning_rate": 1.8143604449754601e-06,
      "loss": 1.9967,
      "step": 222333
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.7409708499908447,
      "learning_rate": 1.8139112708622894e-06,
      "loss": 1.9805,
      "step": 222334
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.3991459608078003,
      "learning_rate": 1.8134621521875392e-06,
      "loss": 2.0035,
      "step": 222335
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.5365556478500366,
      "learning_rate": 1.8130130889512761e-06,
      "loss": 1.9337,
      "step": 222336
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.4009214639663696,
      "learning_rate": 1.8125640811536002e-06,
      "loss": 1.9829,
      "step": 222337
    },
    {
      "epoch": 2.89,
      "grad_norm": 1.5348966121673584,
      "learning_rate": 1.8121151287946444e-06,
      "loss": 2.1622,
      "step": 222338
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.460982322692871,
      "learning_rate": 1.811666231874409e-06,
      "loss": 1.8749,
      "step": 222339
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.4643850326538086,
      "learning_rate": 1.811217390392994e-06,
      "loss": 1.9012,
      "step": 222340
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.4848440885543823,
      "learning_rate": 1.8107686043505655e-06,
      "loss": 2.1248,
      "step": 222341
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.5525143146514893,
      "learning_rate": 1.8103198737470903e-06,
      "loss": 1.8524,
      "step": 222342
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.3294070959091187,
      "learning_rate": 1.8098711985827019e-06,
      "loss": 1.7859,
      "step": 222343
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.3658543825149536,
      "learning_rate": 1.8094225788575333e-06,
      "loss": 2.0073,
      "step": 222344
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.514491319656372,
      "learning_rate": 1.8089740145715847e-06,
      "loss": 2.079,
      "step": 222345
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.3690850734710693,
      "learning_rate": 1.8085255057249893e-06,
      "loss": 1.8869,
      "step": 222346
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.3742843866348267,
      "learning_rate": 1.8080770523178134e-06,
      "loss": 1.8805,
      "step": 222347
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.512423038482666,
      "learning_rate": 1.8076286543501573e-06,
      "loss": 1.9693,
      "step": 222348
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.4735773801803589,
      "learning_rate": 1.807180311822054e-06,
      "loss": 1.8385,
      "step": 222349
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.6423704624176025,
      "learning_rate": 1.8067320247336702e-06,
      "loss": 2.1294,
      "step": 222350
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.486849069595337,
      "learning_rate": 1.806283793085006e-06,
      "loss": 2.0966,
      "step": 222351
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.4961234331130981,
      "learning_rate": 1.8058356168762278e-06,
      "loss": 2.0389,
      "step": 222352
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.505116581916809,
      "learning_rate": 1.8053874961073357e-06,
      "loss": 2.2632,
      "step": 222353
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.6056593656539917,
      "learning_rate": 1.8049394307784626e-06,
      "loss": 2.194,
      "step": 222354
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.5930441617965698,
      "learning_rate": 1.8044914208896754e-06,
      "loss": 1.9431,
      "step": 222355
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.5291566848754883,
      "learning_rate": 1.8040434664410741e-06,
      "loss": 2.0219,
      "step": 222356
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.4489890336990356,
      "learning_rate": 1.803595567432725e-06,
      "loss": 1.9696,
      "step": 222357
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.4122231006622314,
      "learning_rate": 1.8031477238647285e-06,
      "loss": 1.9573,
      "step": 222358
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.4741162061691284,
      "learning_rate": 1.8026999357371175e-06,
      "loss": 1.9003,
      "step": 222359
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.5279674530029297,
      "learning_rate": 1.8022522030500586e-06,
      "loss": 2.0151,
      "step": 222360
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.4211000204086304,
      "learning_rate": 1.8018045258035518e-06,
      "loss": 1.8961,
      "step": 222361
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.4943922758102417,
      "learning_rate": 1.8013569039977638e-06,
      "loss": 2.0014,
      "step": 222362
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.4546308517456055,
      "learning_rate": 1.8009093376326943e-06,
      "loss": 1.8125,
      "step": 222363
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.5484669208526611,
      "learning_rate": 1.800461826708477e-06,
      "loss": 1.9368,
      "step": 222364
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.4966906309127808,
      "learning_rate": 1.800014371225178e-06,
      "loss": 1.9363,
      "step": 222365
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.4111040830612183,
      "learning_rate": 1.7995669711828975e-06,
      "loss": 1.8646,
      "step": 222366
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.3021489381790161,
      "learning_rate": 1.7991196265816689e-06,
      "loss": 1.775,
      "step": 222367
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.5023993253707886,
      "learning_rate": 1.7986723374216582e-06,
      "loss": 2.1189,
      "step": 222368
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.4598166942596436,
      "learning_rate": 1.7982251037028661e-06,
      "loss": 1.9866,
      "step": 222369
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.4982447624206543,
      "learning_rate": 1.7977779254254588e-06,
      "loss": 1.9872,
      "step": 222370
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.4652117490768433,
      "learning_rate": 1.797330802589436e-06,
      "loss": 1.8698,
      "step": 222371
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.4324671030044556,
      "learning_rate": 1.7968837351949316e-06,
      "loss": 1.8589,
      "step": 222372
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.4670237302780151,
      "learning_rate": 1.796436723242012e-06,
      "loss": 1.939,
      "step": 222373
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.5272152423858643,
      "learning_rate": 1.7959897667307766e-06,
      "loss": 1.7825,
      "step": 222374
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.4453519582748413,
      "learning_rate": 1.7955428656612926e-06,
      "loss": 2.1365,
      "step": 222375
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.3784983158111572,
      "learning_rate": 1.795096020033626e-06,
      "loss": 1.8532,
      "step": 222376
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.5013827085494995,
      "learning_rate": 1.7946492298478775e-06,
      "loss": 2.0217,
      "step": 222377
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.6081268787384033,
      "learning_rate": 1.79420249510418e-06,
      "loss": 1.8507,
      "step": 222378
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.4617632627487183,
      "learning_rate": 1.7937558158025334e-06,
      "loss": 1.8982,
      "step": 222379
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.538944125175476,
      "learning_rate": 1.7933091919430709e-06,
      "loss": 1.9279,
      "step": 222380
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.4097453355789185,
      "learning_rate": 1.7928626235258258e-06,
      "loss": 1.9793,
      "step": 222381
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.5240325927734375,
      "learning_rate": 1.7924161105509648e-06,
      "loss": 1.8825,
      "step": 222382
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.4848216772079468,
      "learning_rate": 1.7919696530184879e-06,
      "loss": 1.8934,
      "step": 222383
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.5129454135894775,
      "learning_rate": 1.7915232509285283e-06,
      "loss": 1.9446,
      "step": 222384
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.5599132776260376,
      "learning_rate": 1.7910769042811524e-06,
      "loss": 1.9639,
      "step": 222385
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.580592393875122,
      "learning_rate": 1.790630613076427e-06,
      "loss": 2.0689,
      "step": 222386
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.358638882637024,
      "learning_rate": 1.7901843773144853e-06,
      "loss": 1.8989,
      "step": 222387
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.4261564016342163,
      "learning_rate": 1.7897381969953605e-06,
      "loss": 1.9385,
      "step": 222388
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.5100712776184082,
      "learning_rate": 1.7892920721191195e-06,
      "loss": 1.8157,
      "step": 222389
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.4700427055358887,
      "learning_rate": 1.7888460026859287e-06,
      "loss": 1.7018,
      "step": 222390
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.5091781616210938,
      "learning_rate": 1.7883999886958211e-06,
      "loss": 2.046,
      "step": 222391
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.4405802488327026,
      "learning_rate": 1.7879540301488637e-06,
      "loss": 1.9875,
      "step": 222392
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.4571255445480347,
      "learning_rate": 1.7875081270451231e-06,
      "loss": 1.6921,
      "step": 222393
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.4961903095245361,
      "learning_rate": 1.7870622793847656e-06,
      "loss": 1.8576,
      "step": 222394
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.6210905313491821,
      "learning_rate": 1.7866164871677913e-06,
      "loss": 1.9114,
      "step": 222395
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.4851475954055786,
      "learning_rate": 1.7861707503943001e-06,
      "loss": 1.9437,
      "step": 222396
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.5437012910842896,
      "learning_rate": 1.7857250690644254e-06,
      "loss": 2.098,
      "step": 222397
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.5960774421691895,
      "learning_rate": 1.7852794431782002e-06,
      "loss": 2.0049,
      "step": 222398
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.456364631652832,
      "learning_rate": 1.7848338727357247e-06,
      "loss": 1.9605,
      "step": 222399
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.553932547569275,
      "learning_rate": 1.7843883577370654e-06,
      "loss": 1.9347,
      "step": 222400
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.4471814632415771,
      "learning_rate": 1.7839428981823223e-06,
      "loss": 2.1045,
      "step": 222401
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.5361168384552002,
      "learning_rate": 1.7834974940715952e-06,
      "loss": 1.9654,
      "step": 222402
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.4169468879699707,
      "learning_rate": 1.7830521454049174e-06,
      "loss": 1.8521,
      "step": 222403
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.464892029762268,
      "learning_rate": 1.782606852182422e-06,
      "loss": 2.0319,
      "step": 222404
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.4589533805847168,
      "learning_rate": 1.7821616144041429e-06,
      "loss": 1.9585,
      "step": 222405
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.4012541770935059,
      "learning_rate": 1.7817164320702126e-06,
      "loss": 1.9529,
      "step": 222406
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.4337695837020874,
      "learning_rate": 1.781271305180665e-06,
      "loss": 1.9712,
      "step": 222407
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.535476565361023,
      "learning_rate": 1.7808262337356328e-06,
      "loss": 1.9842,
      "step": 222408
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.3151201009750366,
      "learning_rate": 1.7803812177351828e-06,
      "loss": 2.102,
      "step": 222409
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.468331217765808,
      "learning_rate": 1.7799362571793484e-06,
      "loss": 1.9375,
      "step": 222410
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.4936647415161133,
      "learning_rate": 1.7794913520682963e-06,
      "loss": 2.0391,
      "step": 222411
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.4586122035980225,
      "learning_rate": 1.7790465024020261e-06,
      "loss": 2.2466,
      "step": 222412
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.4708037376403809,
      "learning_rate": 1.7786017081806713e-06,
      "loss": 1.8642,
      "step": 222413
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.7200791835784912,
      "learning_rate": 1.7781569694043319e-06,
      "loss": 2.2356,
      "step": 222414
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.5102617740631104,
      "learning_rate": 1.7777122860730407e-06,
      "loss": 1.9157,
      "step": 222415
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.4309682846069336,
      "learning_rate": 1.7772676581868984e-06,
      "loss": 1.7998,
      "step": 222416
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.482619047164917,
      "learning_rate": 1.7768230857460042e-06,
      "loss": 2.0518,
      "step": 222417
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.473280906677246,
      "learning_rate": 1.776378568750425e-06,
      "loss": 2.0041,
      "step": 222418
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.4720426797866821,
      "learning_rate": 1.7759341072002275e-06,
      "loss": 1.8367,
      "step": 222419
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.5051239728927612,
      "learning_rate": 1.7754897010955449e-06,
      "loss": 2.0882,
      "step": 222420
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.4204976558685303,
      "learning_rate": 1.7750453504364103e-06,
      "loss": 1.9403,
      "step": 222421
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.5201507806777954,
      "learning_rate": 1.7746010552229239e-06,
      "loss": 1.8839,
      "step": 222422
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.4964863061904907,
      "learning_rate": 1.7741568154551856e-06,
      "loss": 2.0003,
      "step": 222423
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.447732925415039,
      "learning_rate": 1.7737126311332284e-06,
      "loss": 2.3336,
      "step": 222424
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.5204918384552002,
      "learning_rate": 1.7732685022571858e-06,
      "loss": 2.0397,
      "step": 222425
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.5365090370178223,
      "learning_rate": 1.7728244288271244e-06,
      "loss": 2.0472,
      "step": 222426
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.4455174207687378,
      "learning_rate": 1.7723804108431438e-06,
      "loss": 1.85,
      "step": 222427
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.6115295886993408,
      "learning_rate": 1.7719364483052779e-06,
      "loss": 1.9503,
      "step": 222428
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.5674270391464233,
      "learning_rate": 1.771492541213626e-06,
      "loss": 1.9829,
      "step": 222429
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.3510304689407349,
      "learning_rate": 1.7710486895683218e-06,
      "loss": 1.8368,
      "step": 222430
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.4287736415863037,
      "learning_rate": 1.7706048933693651e-06,
      "loss": 1.9283,
      "step": 222431
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.5978927612304688,
      "learning_rate": 1.7701611526169224e-06,
      "loss": 2.165,
      "step": 222432
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.4081311225891113,
      "learning_rate": 1.769717467311027e-06,
      "loss": 2.1399,
      "step": 222433
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.463792085647583,
      "learning_rate": 1.7692738374517458e-06,
      "loss": 1.7938,
      "step": 222434
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.4308993816375732,
      "learning_rate": 1.7688302630392116e-06,
      "loss": 2.1447,
      "step": 222435
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.4584481716156006,
      "learning_rate": 1.7683867440734912e-06,
      "loss": 1.9391,
      "step": 222436
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.4632717370986938,
      "learning_rate": 1.7679432805546178e-06,
      "loss": 1.9435,
      "step": 222437
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.4075798988342285,
      "learning_rate": 1.7674998724827584e-06,
      "loss": 2.1887,
      "step": 222438
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.4928957223892212,
      "learning_rate": 1.7670565198579456e-06,
      "loss": 1.8255,
      "step": 222439
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.5288166999816895,
      "learning_rate": 1.7666132226802133e-06,
      "loss": 2.0403,
      "step": 222440
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.4840563535690308,
      "learning_rate": 1.766169980949761e-06,
      "loss": 2.2524,
      "step": 222441
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.485783338546753,
      "learning_rate": 1.7657267946665555e-06,
      "loss": 1.9015,
      "step": 222442
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.651485800743103,
      "learning_rate": 1.7652836638307633e-06,
      "loss": 2.1876,
      "step": 222443
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.353655457496643,
      "learning_rate": 1.764840588442451e-06,
      "loss": 1.9729,
      "step": 222444
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.4405837059020996,
      "learning_rate": 1.764397568501652e-06,
      "loss": 2.1326,
      "step": 222445
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.5741910934448242,
      "learning_rate": 1.7639546040084661e-06,
      "loss": 2.1162,
      "step": 222446
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.4527043104171753,
      "learning_rate": 1.7635116949630267e-06,
      "loss": 1.8996,
      "step": 222447
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.5905861854553223,
      "learning_rate": 1.763068841365367e-06,
      "loss": 1.9696,
      "step": 222448
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.4887664318084717,
      "learning_rate": 1.762626043215587e-06,
      "loss": 2.2165,
      "step": 222449
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.4352065324783325,
      "learning_rate": 1.762183300513753e-06,
      "loss": 1.8101,
      "step": 222450
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.4798353910446167,
      "learning_rate": 1.7617406132599655e-06,
      "loss": 2.1092,
      "step": 222451
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.352729320526123,
      "learning_rate": 1.7612979814542904e-06,
      "loss": 1.8055,
      "step": 222452
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.572310447692871,
      "learning_rate": 1.7608554050968615e-06,
      "loss": 1.9215,
      "step": 222453
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.4826198816299438,
      "learning_rate": 1.7604128841876786e-06,
      "loss": 2.1432,
      "step": 222454
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.399045705795288,
      "learning_rate": 1.759970418726875e-06,
      "loss": 2.0671,
      "step": 222455
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.4575600624084473,
      "learning_rate": 1.759528008714517e-06,
      "loss": 2.0093,
      "step": 222456
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.5282180309295654,
      "learning_rate": 1.7590856541507048e-06,
      "loss": 1.9385,
      "step": 222457
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.7471040487289429,
      "learning_rate": 1.758643355035505e-06,
      "loss": 2.0526,
      "step": 222458
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.3607139587402344,
      "learning_rate": 1.7582011113689841e-06,
      "loss": 2.0421,
      "step": 222459
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.3690208196640015,
      "learning_rate": 1.7577589231512423e-06,
      "loss": 1.8801,
      "step": 222460
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.6204205751419067,
      "learning_rate": 1.7573167903823793e-06,
      "loss": 1.934,
      "step": 222461
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.4231929779052734,
      "learning_rate": 1.7568747130624616e-06,
      "loss": 1.9909,
      "step": 222462
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.8012717962265015,
      "learning_rate": 1.7564326911915895e-06,
      "loss": 2.0731,
      "step": 222463
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.445067048072815,
      "learning_rate": 1.7559907247697958e-06,
      "loss": 1.9342,
      "step": 222464
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.6417227983474731,
      "learning_rate": 1.755548813797214e-06,
      "loss": 2.1281,
      "step": 222465
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.4537550210952759,
      "learning_rate": 1.7551069582738775e-06,
      "loss": 1.9843,
      "step": 222466
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.4044597148895264,
      "learning_rate": 1.7546651581999194e-06,
      "loss": 1.9782,
      "step": 222467
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.4480314254760742,
      "learning_rate": 1.754223413575373e-06,
      "loss": 1.952,
      "step": 222468
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.4354848861694336,
      "learning_rate": 1.7537817244003715e-06,
      "loss": 2.0577,
      "step": 222469
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.5957956314086914,
      "learning_rate": 1.7533400906749817e-06,
      "loss": 2.0134,
      "step": 222470
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.7340340614318848,
      "learning_rate": 1.75289851239927e-06,
      "loss": 1.7378,
      "step": 222471
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.5699735879898071,
      "learning_rate": 1.7524569895733031e-06,
      "loss": 2.0458,
      "step": 222472
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.3935396671295166,
      "learning_rate": 1.7520155221972144e-06,
      "loss": 1.8696,
      "step": 222473
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.5183669328689575,
      "learning_rate": 1.7515741102710035e-06,
      "loss": 1.799,
      "step": 222474
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.5115063190460205,
      "learning_rate": 1.7511327537948705e-06,
      "loss": 1.9455,
      "step": 222475
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.4061009883880615,
      "learning_rate": 1.750691452768782e-06,
      "loss": 2.0461,
      "step": 222476
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.5913842916488647,
      "learning_rate": 1.7502502071929047e-06,
      "loss": 1.8035,
      "step": 222477
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.5352659225463867,
      "learning_rate": 1.7498090170672717e-06,
      "loss": 1.8916,
      "step": 222478
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.5296335220336914,
      "learning_rate": 1.7493678823919832e-06,
      "loss": 2.2185,
      "step": 222479
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.471056342124939,
      "learning_rate": 1.7489268031671054e-06,
      "loss": 1.8064,
      "step": 222480
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.4445959329605103,
      "learning_rate": 1.7484857793927387e-06,
      "loss": 2.042,
      "step": 222481
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.6223145723342896,
      "learning_rate": 1.7480448110689494e-06,
      "loss": 1.6789,
      "step": 222482
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.5094032287597656,
      "learning_rate": 1.7476038981958374e-06,
      "loss": 1.9304,
      "step": 222483
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.462931752204895,
      "learning_rate": 1.7471630407734694e-06,
      "loss": 2.1611,
      "step": 222484
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.4749807119369507,
      "learning_rate": 1.7467222388019453e-06,
      "loss": 1.9824,
      "step": 222485
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.4548288583755493,
      "learning_rate": 1.7462814922813317e-06,
      "loss": 1.9083,
      "step": 222486
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.397792935371399,
      "learning_rate": 1.7458408012116953e-06,
      "loss": 1.7814,
      "step": 222487
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.6256909370422363,
      "learning_rate": 1.7454001655931692e-06,
      "loss": 1.7974,
      "step": 222488
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.7300562858581543,
      "learning_rate": 1.744959585425787e-06,
      "loss": 2.0665,
      "step": 222489
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.451482892036438,
      "learning_rate": 1.744519060709648e-06,
      "loss": 2.0371,
      "step": 222490
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.4265366792678833,
      "learning_rate": 1.7440785914448197e-06,
      "loss": 1.979,
      "step": 222491
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.354681134223938,
      "learning_rate": 1.7436381776314012e-06,
      "loss": 1.6933,
      "step": 222492
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.3836593627929688,
      "learning_rate": 1.7431978192694928e-06,
      "loss": 1.7927,
      "step": 222493
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.450730562210083,
      "learning_rate": 1.7427575163591611e-06,
      "loss": 1.9142,
      "step": 222494
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.354826807975769,
      "learning_rate": 1.7423172689004394e-06,
      "loss": 1.9525,
      "step": 222495
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.7332589626312256,
      "learning_rate": 1.7418770768934608e-06,
      "loss": 2.2898,
      "step": 222496
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.563997745513916,
      "learning_rate": 1.7414369403383254e-06,
      "loss": 1.8455,
      "step": 222497
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.431808352470398,
      "learning_rate": 1.7409968592350665e-06,
      "loss": 1.9478,
      "step": 222498
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.5006595849990845,
      "learning_rate": 1.740556833583784e-06,
      "loss": 2.1424,
      "step": 222499
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.439252257347107,
      "learning_rate": 1.7401168633845775e-06,
      "loss": 1.9445,
      "step": 222500
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.5537514686584473,
      "learning_rate": 1.739676948637514e-06,
      "loss": 2.1331,
      "step": 222501
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.3997373580932617,
      "learning_rate": 1.73923708934266e-06,
      "loss": 1.8913,
      "step": 222502
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.4785583019256592,
      "learning_rate": 1.7387972855001153e-06,
      "loss": 2.0683,
      "step": 222503
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.5853997468948364,
      "learning_rate": 1.7383575371099468e-06,
      "loss": 1.9713,
      "step": 222504
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.474189043045044,
      "learning_rate": 1.7379178441722542e-06,
      "loss": 1.9313,
      "step": 222505
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.4665244817733765,
      "learning_rate": 1.7374782066871373e-06,
      "loss": 2.2063,
      "step": 222506
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.5167032480239868,
      "learning_rate": 1.7370386246546298e-06,
      "loss": 2.2037,
      "step": 222507
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.4127097129821777,
      "learning_rate": 1.7365990980748313e-06,
      "loss": 1.9451,
      "step": 222508
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.4321078062057495,
      "learning_rate": 1.7361596269478418e-06,
      "loss": 1.9696,
      "step": 222509
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.4459481239318848,
      "learning_rate": 1.735720211273728e-06,
      "loss": 2.0859,
      "step": 222510
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.4994158744812012,
      "learning_rate": 1.7352808510525896e-06,
      "loss": 2.2806,
      "step": 222511
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.545242428779602,
      "learning_rate": 1.73484154628446e-06,
      "loss": 1.9174,
      "step": 222512
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.5149873495101929,
      "learning_rate": 1.7344022969694725e-06,
      "loss": 2.0523,
      "step": 222513
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.4961656332015991,
      "learning_rate": 1.733963103107694e-06,
      "loss": 2.1297,
      "step": 222514
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.5164350271224976,
      "learning_rate": 1.7335239646991904e-06,
      "loss": 2.2008,
      "step": 222515
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.5196295976638794,
      "learning_rate": 1.7330848817440624e-06,
      "loss": 1.8428,
      "step": 222516
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.5242384672164917,
      "learning_rate": 1.732645854242376e-06,
      "loss": 2.1367,
      "step": 222517
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.4320799112319946,
      "learning_rate": 1.7322068821942315e-06,
      "loss": 1.9962,
      "step": 222518
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.4527766704559326,
      "learning_rate": 1.7317679655996951e-06,
      "loss": 2.0774,
      "step": 222519
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.3475761413574219,
      "learning_rate": 1.7313291044588674e-06,
      "loss": 1.9929,
      "step": 222520
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.4956880807876587,
      "learning_rate": 1.730890298771781e-06,
      "loss": 2.0497,
      "step": 222521
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.4733864068984985,
      "learning_rate": 1.7304515485385696e-06,
      "loss": 2.1268,
      "step": 222522
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.5291121006011963,
      "learning_rate": 1.7300128537593327e-06,
      "loss": 2.2484,
      "step": 222523
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.4640321731567383,
      "learning_rate": 1.7295742144340707e-06,
      "loss": 1.8114,
      "step": 222524
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.6706913709640503,
      "learning_rate": 1.7291356305629167e-06,
      "loss": 1.8137,
      "step": 222525
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.4743667840957642,
      "learning_rate": 1.7286971021459706e-06,
      "loss": 2.1744,
      "step": 222526
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.6657485961914062,
      "learning_rate": 1.7282586291832655e-06,
      "loss": 2.0002,
      "step": 222527
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.6275285482406616,
      "learning_rate": 1.727820211674935e-06,
      "loss": 2.1109,
      "step": 222528
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.7394288778305054,
      "learning_rate": 1.7273818496210122e-06,
      "loss": 1.8365,
      "step": 222529
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.5459094047546387,
      "learning_rate": 1.7269435430215972e-06,
      "loss": 1.8129,
      "step": 222530
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.443773865699768,
      "learning_rate": 1.7265052918767563e-06,
      "loss": 1.9164,
      "step": 222531
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.4224438667297363,
      "learning_rate": 1.726067096186623e-06,
      "loss": 1.8911,
      "step": 222532
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.5154112577438354,
      "learning_rate": 1.7256289559512304e-06,
      "loss": 2.15,
      "step": 222533
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.4705140590667725,
      "learning_rate": 1.7251908711706787e-06,
      "loss": 1.8967,
      "step": 222534
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.5562676191329956,
      "learning_rate": 1.724752841845034e-06,
      "loss": 1.9419,
      "step": 222535
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.5986151695251465,
      "learning_rate": 1.7243148679743968e-06,
      "loss": 1.9253,
      "step": 222536
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.4745591878890991,
      "learning_rate": 1.7238769495588667e-06,
      "loss": 1.8255,
      "step": 222537
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.5120593309402466,
      "learning_rate": 1.7234390865984438e-06,
      "loss": 2.1088,
      "step": 222538
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.5767639875411987,
      "learning_rate": 1.7230012790932945e-06,
      "loss": 2.0075,
      "step": 222539
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.4245303869247437,
      "learning_rate": 1.7225635270434856e-06,
      "loss": 1.9706,
      "step": 222540
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.382433533668518,
      "learning_rate": 1.7221258304490504e-06,
      "loss": 2.0525,
      "step": 222541
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.3312169313430786,
      "learning_rate": 1.721688189310122e-06,
      "loss": 1.9816,
      "step": 222542
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.5081731081008911,
      "learning_rate": 1.7212506036267337e-06,
      "loss": 1.8249,
      "step": 222543
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.4441438913345337,
      "learning_rate": 1.7208130733990188e-06,
      "loss": 1.9979,
      "step": 222544
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.5182729959487915,
      "learning_rate": 1.720375598627044e-06,
      "loss": 2.0351,
      "step": 222545
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.4673808813095093,
      "learning_rate": 1.7199381793108756e-06,
      "loss": 1.9307,
      "step": 222546
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.4781768321990967,
      "learning_rate": 1.7195008154505808e-06,
      "loss": 2.044,
      "step": 222547
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.4695510864257812,
      "learning_rate": 1.719063507046292e-06,
      "loss": 2.0987,
      "step": 222548
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.4582217931747437,
      "learning_rate": 1.7186262540980433e-06,
      "loss": 2.2641,
      "step": 222549
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.3395800590515137,
      "learning_rate": 1.7181890566059341e-06,
      "loss": 2.0435,
      "step": 222550
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.4769384860992432,
      "learning_rate": 1.7177519145700313e-06,
      "loss": 1.9719,
      "step": 222551
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.4230962991714478,
      "learning_rate": 1.717314827990468e-06,
      "loss": 1.9395,
      "step": 222552
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.5946314334869385,
      "learning_rate": 1.7168777968672442e-06,
      "loss": 1.9721,
      "step": 222553
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.5723217725753784,
      "learning_rate": 1.7164408212004933e-06,
      "loss": 2.1803,
      "step": 222554
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.400290846824646,
      "learning_rate": 1.7160039009903148e-06,
      "loss": 1.807,
      "step": 222555
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.5362178087234497,
      "learning_rate": 1.7155670362367423e-06,
      "loss": 2.0515,
      "step": 222556
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.6100645065307617,
      "learning_rate": 1.7151302269398426e-06,
      "loss": 1.9288,
      "step": 222557
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.4219691753387451,
      "learning_rate": 1.714693473099782e-06,
      "loss": 1.9348,
      "step": 222558
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.379380226135254,
      "learning_rate": 1.7142567747165604e-06,
      "loss": 1.9993,
      "step": 222559
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.5177679061889648,
      "learning_rate": 1.713820131790311e-06,
      "loss": 2.2475,
      "step": 222560
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.5716227293014526,
      "learning_rate": 1.7133835443210674e-06,
      "loss": 1.8805,
      "step": 222561
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.5649609565734863,
      "learning_rate": 1.712947012308963e-06,
      "loss": 1.9868,
      "step": 222562
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.4908071756362915,
      "learning_rate": 1.7125105357540303e-06,
      "loss": 1.9422,
      "step": 222563
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.3189557790756226,
      "learning_rate": 1.7120741146563699e-06,
      "loss": 1.9481,
      "step": 222564
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.369718313217163,
      "learning_rate": 1.7116377490160814e-06,
      "loss": 1.9551,
      "step": 222565
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.4458987712860107,
      "learning_rate": 1.7112014388332318e-06,
      "loss": 1.9769,
      "step": 222566
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.5952802896499634,
      "learning_rate": 1.7107651841078873e-06,
      "loss": 1.917,
      "step": 222567
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.4543973207473755,
      "learning_rate": 1.7103289848401481e-06,
      "loss": 2.0211,
      "step": 222568
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.4714864492416382,
      "learning_rate": 1.7098928410300806e-06,
      "loss": 1.9901,
      "step": 222569
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.4283143281936646,
      "learning_rate": 1.7094567526777847e-06,
      "loss": 2.0042,
      "step": 222570
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.608784794807434,
      "learning_rate": 1.7090207197833271e-06,
      "loss": 2.1669,
      "step": 222571
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.6460391283035278,
      "learning_rate": 1.708584742346808e-06,
      "loss": 2.0598,
      "step": 222572
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.4459052085876465,
      "learning_rate": 1.7081488203682936e-06,
      "loss": 2.0162,
      "step": 222573
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.5967137813568115,
      "learning_rate": 1.7077129538478506e-06,
      "loss": 2.115,
      "step": 222574
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.4893988370895386,
      "learning_rate": 1.707277142785579e-06,
      "loss": 1.9893,
      "step": 222575
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.3853349685668945,
      "learning_rate": 1.7068413871815456e-06,
      "loss": 2.1822,
      "step": 222576
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.408531904220581,
      "learning_rate": 1.70640568703585e-06,
      "loss": 1.9182,
      "step": 222577
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.5144306421279907,
      "learning_rate": 1.705970042348559e-06,
      "loss": 1.8605,
      "step": 222578
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.4242007732391357,
      "learning_rate": 1.7055344531197723e-06,
      "loss": 1.8784,
      "step": 222579
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.53203284740448,
      "learning_rate": 1.7050989193495567e-06,
      "loss": 1.9385,
      "step": 222580
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.4712648391723633,
      "learning_rate": 1.7046634410379788e-06,
      "loss": 2.1238,
      "step": 222581
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.4572834968566895,
      "learning_rate": 1.7042280181851387e-06,
      "loss": 2.0233,
      "step": 222582
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.3674806356430054,
      "learning_rate": 1.7037926507911359e-06,
      "loss": 1.8447,
      "step": 222583
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.5231820344924927,
      "learning_rate": 1.703357338856004e-06,
      "loss": 2.011,
      "step": 222584
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.355059266090393,
      "learning_rate": 1.702922082379843e-06,
      "loss": 2.0748,
      "step": 222585
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.517939805984497,
      "learning_rate": 1.7024868813627856e-06,
      "loss": 1.9236,
      "step": 222586
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.4782339334487915,
      "learning_rate": 1.7020517358047992e-06,
      "loss": 2.0399,
      "step": 222587
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.4650235176086426,
      "learning_rate": 1.7016166457060831e-06,
      "loss": 2.0342,
      "step": 222588
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.5566327571868896,
      "learning_rate": 1.7011816110666376e-06,
      "loss": 2.0073,
      "step": 222589
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.4104070663452148,
      "learning_rate": 1.7007466318865959e-06,
      "loss": 1.8499,
      "step": 222590
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.5209230184555054,
      "learning_rate": 1.700311708165991e-06,
      "loss": 1.9542,
      "step": 222591
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.4066091775894165,
      "learning_rate": 1.6998768399049566e-06,
      "loss": 1.9712,
      "step": 222592
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.4009413719177246,
      "learning_rate": 1.699442027103526e-06,
      "loss": 1.9145,
      "step": 222593
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.4599167108535767,
      "learning_rate": 1.6990072697618318e-06,
      "loss": 1.9397,
      "step": 222594
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.4812206029891968,
      "learning_rate": 1.6985725678798746e-06,
      "loss": 2.0916,
      "step": 222595
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.620732307434082,
      "learning_rate": 1.6981379214578206e-06,
      "loss": 1.9953,
      "step": 222596
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.554060459136963,
      "learning_rate": 1.69770333049567e-06,
      "loss": 2.0131,
      "step": 222597
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.496936559677124,
      "learning_rate": 1.6972687949935893e-06,
      "loss": 1.9254,
      "step": 222598
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.5150461196899414,
      "learning_rate": 1.6968343149516116e-06,
      "loss": 1.9373,
      "step": 222599
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.4239946603775024,
      "learning_rate": 1.6963998903698372e-06,
      "loss": 1.9104,
      "step": 222600
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.3132957220077515,
      "learning_rate": 1.6959655212482991e-06,
      "loss": 2.0374,
      "step": 222601
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.467572808265686,
      "learning_rate": 1.6955312075871308e-06,
      "loss": 2.0428,
      "step": 222602
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.3596583604812622,
      "learning_rate": 1.6950969493863653e-06,
      "loss": 1.8193,
      "step": 222603
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.305293321609497,
      "learning_rate": 1.694662746646136e-06,
      "loss": 1.8904,
      "step": 222604
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.498429536819458,
      "learning_rate": 1.6942285993665094e-06,
      "loss": 2.0001,
      "step": 222605
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.5320667028427124,
      "learning_rate": 1.6937945075475522e-06,
      "loss": 2.1599,
      "step": 222606
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.43717360496521,
      "learning_rate": 1.6933604711893312e-06,
      "loss": 1.9897,
      "step": 222607
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.4539421796798706,
      "learning_rate": 1.6929264902919793e-06,
      "loss": 2.0182,
      "step": 222608
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.6085517406463623,
      "learning_rate": 1.6924925648555298e-06,
      "loss": 1.9677,
      "step": 222609
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.3224760293960571,
      "learning_rate": 1.6920586948800496e-06,
      "loss": 1.8783,
      "step": 222610
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.4775484800338745,
      "learning_rate": 1.6916248803656718e-06,
      "loss": 2.0338,
      "step": 222611
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.3742904663085938,
      "learning_rate": 1.691191121312463e-06,
      "loss": 1.683,
      "step": 222612
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.4402763843536377,
      "learning_rate": 1.6907574177204563e-06,
      "loss": 2.0939,
      "step": 222613
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.4071298837661743,
      "learning_rate": 1.6903237695898186e-06,
      "loss": 1.9675,
      "step": 222614
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.5541720390319824,
      "learning_rate": 1.6898901769205165e-06,
      "loss": 2.0748,
      "step": 222615
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.4813077449798584,
      "learning_rate": 1.6894566397127497e-06,
      "loss": 2.1228,
      "step": 222616
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.5707467794418335,
      "learning_rate": 1.689023157966518e-06,
      "loss": 2.1222,
      "step": 222617
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.446783423423767,
      "learning_rate": 1.6885897316819218e-06,
      "loss": 1.8961,
      "step": 222618
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.5154472589492798,
      "learning_rate": 1.6881563608590941e-06,
      "loss": 1.9543,
      "step": 222619
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.4392544031143188,
      "learning_rate": 1.6877230454980018e-06,
      "loss": 1.8781,
      "step": 222620
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.4296051263809204,
      "learning_rate": 1.6872897855988443e-06,
      "loss": 1.9595,
      "step": 222621
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.4763121604919434,
      "learning_rate": 1.686856581161622e-06,
      "loss": 1.9003,
      "step": 222622
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.5020999908447266,
      "learning_rate": 1.6864234321864679e-06,
      "loss": 2.1141,
      "step": 222623
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.5200111865997314,
      "learning_rate": 1.6859903386734153e-06,
      "loss": 1.9094,
      "step": 222624
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.4139963388442993,
      "learning_rate": 1.6855573006225975e-06,
      "loss": 1.9041,
      "step": 222625
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.4110149145126343,
      "learning_rate": 1.6851243180340146e-06,
      "loss": 1.9695,
      "step": 222626
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.4999059438705444,
      "learning_rate": 1.6846913909078663e-06,
      "loss": 2.1522,
      "step": 222627
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.421398401260376,
      "learning_rate": 1.6842585192441194e-06,
      "loss": 1.9794,
      "step": 222628
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.6327223777770996,
      "learning_rate": 1.683825703042907e-06,
      "loss": 1.983,
      "step": 222629
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.5482426881790161,
      "learning_rate": 1.6833929423042958e-06,
      "loss": 1.8359,
      "step": 222630
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.4142332077026367,
      "learning_rate": 1.6829602370283857e-06,
      "loss": 1.9758,
      "step": 222631
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.435584545135498,
      "learning_rate": 1.6825275872152434e-06,
      "loss": 1.9389,
      "step": 222632
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.529676914215088,
      "learning_rate": 1.6820949928649684e-06,
      "loss": 2.2055,
      "step": 222633
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.3960411548614502,
      "learning_rate": 1.6816624539775946e-06,
      "loss": 1.9093,
      "step": 222634
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.412488579750061,
      "learning_rate": 1.681229970553255e-06,
      "loss": 1.8522,
      "step": 222635
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.3539823293685913,
      "learning_rate": 1.680797542591983e-06,
      "loss": 2.0843,
      "step": 222636
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.3819161653518677,
      "learning_rate": 1.680365170093878e-06,
      "loss": 1.9575,
      "step": 222637
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.5146803855895996,
      "learning_rate": 1.6799328530590406e-06,
      "loss": 1.9176,
      "step": 222638
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.5649900436401367,
      "learning_rate": 1.679500591487537e-06,
      "loss": 1.8676,
      "step": 222639
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.5120139122009277,
      "learning_rate": 1.679068385379434e-06,
      "loss": 2.0259,
      "step": 222640
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.4399042129516602,
      "learning_rate": 1.6786362347348314e-06,
      "loss": 2.0605,
      "step": 222641
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.4250500202178955,
      "learning_rate": 1.678204139553796e-06,
      "loss": 2.0231,
      "step": 222642
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.3780100345611572,
      "learning_rate": 1.6777720998364274e-06,
      "loss": 1.9746,
      "step": 222643
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.3875246047973633,
      "learning_rate": 1.6773401155827593e-06,
      "loss": 1.8887,
      "step": 222644
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.5294381380081177,
      "learning_rate": 1.6769081867929245e-06,
      "loss": 2.207,
      "step": 222645
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.3963924646377563,
      "learning_rate": 1.6764763134669901e-06,
      "loss": 1.9034,
      "step": 222646
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.4278743267059326,
      "learning_rate": 1.6760444956050224e-06,
      "loss": 1.8656,
      "step": 222647
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.559464693069458,
      "learning_rate": 1.6756127332071212e-06,
      "loss": 1.9352,
      "step": 222648
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.4663163423538208,
      "learning_rate": 1.6751810262733535e-06,
      "loss": 1.7448,
      "step": 222649
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.410325050354004,
      "learning_rate": 1.6747493748037856e-06,
      "loss": 1.8329,
      "step": 222650
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.5067449808120728,
      "learning_rate": 1.6743177787985173e-06,
      "loss": 1.9824,
      "step": 222651
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.6163018941879272,
      "learning_rate": 1.673886238257649e-06,
      "loss": 1.7988,
      "step": 222652
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.4490100145339966,
      "learning_rate": 1.6734547531812136e-06,
      "loss": 2.1679,
      "step": 222653
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.880836009979248,
      "learning_rate": 1.6730233235693113e-06,
      "loss": 2.0199,
      "step": 222654
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.491767406463623,
      "learning_rate": 1.6725919494220085e-06,
      "loss": 1.9323,
      "step": 222655
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.4862233400344849,
      "learning_rate": 1.6721606307394386e-06,
      "loss": 2.1614,
      "step": 222656
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.569453477859497,
      "learning_rate": 1.6717293675216348e-06,
      "loss": 2.018,
      "step": 222657
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.4684484004974365,
      "learning_rate": 1.671298159768697e-06,
      "loss": 2.1134,
      "step": 222658
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.5475859642028809,
      "learning_rate": 1.6708670074806919e-06,
      "loss": 2.0225,
      "step": 222659
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.3911798000335693,
      "learning_rate": 1.6704359106576859e-06,
      "loss": 2.0013,
      "step": 222660
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.4589704275131226,
      "learning_rate": 1.6700048692997792e-06,
      "loss": 2.1122,
      "step": 222661
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.4597464799880981,
      "learning_rate": 1.6695738834070382e-06,
      "loss": 2.1408,
      "step": 222662
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.4160263538360596,
      "learning_rate": 1.6691429529795963e-06,
      "loss": 1.7712,
      "step": 222663
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.5417416095733643,
      "learning_rate": 1.6687120780174535e-06,
      "loss": 1.8109,
      "step": 222664
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.4602477550506592,
      "learning_rate": 1.6682812585207761e-06,
      "loss": 2.1108,
      "step": 222665
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.5486582517623901,
      "learning_rate": 1.667850494489531e-06,
      "loss": 2.0828,
      "step": 222666
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.4624065160751343,
      "learning_rate": 1.667419785923918e-06,
      "loss": 1.9298,
      "step": 222667
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.4370360374450684,
      "learning_rate": 1.6669891328239372e-06,
      "loss": 1.688,
      "step": 222668
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.4877164363861084,
      "learning_rate": 1.6665585351897215e-06,
      "loss": 1.8947,
      "step": 222669
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.4866362810134888,
      "learning_rate": 1.6661279930212711e-06,
      "loss": 1.9757,
      "step": 222670
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.4172933101654053,
      "learning_rate": 1.665697506318786e-06,
      "loss": 1.9988,
      "step": 222671
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.5787172317504883,
      "learning_rate": 1.6652670750822328e-06,
      "loss": 2.06,
      "step": 222672
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.4499740600585938,
      "learning_rate": 1.6648366993117445e-06,
      "loss": 2.0366,
      "step": 222673
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.3861169815063477,
      "learning_rate": 1.664406379007388e-06,
      "loss": 1.9921,
      "step": 222674
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.4560589790344238,
      "learning_rate": 1.663976114169263e-06,
      "loss": 1.8758,
      "step": 222675
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.481921672821045,
      "learning_rate": 1.6635459047974365e-06,
      "loss": 2.0289,
      "step": 222676
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.484224557876587,
      "learning_rate": 1.6631157508920078e-06,
      "loss": 1.9961,
      "step": 222677
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.3497320413589478,
      "learning_rate": 1.6626856524530108e-06,
      "loss": 1.9044,
      "step": 222678
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.44287109375,
      "learning_rate": 1.6622556094805783e-06,
      "loss": 1.8884,
      "step": 222679
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.42808198928833,
      "learning_rate": 1.6618256219747107e-06,
      "loss": 1.7966,
      "step": 222680
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.6232109069824219,
      "learning_rate": 1.6613956899356074e-06,
      "loss": 2.0963,
      "step": 222681
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.4438284635543823,
      "learning_rate": 1.6609658133632353e-06,
      "loss": 1.9487,
      "step": 222682
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.4895180463790894,
      "learning_rate": 1.6605359922577277e-06,
      "loss": 1.9318,
      "step": 222683
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.5546538829803467,
      "learning_rate": 1.6601062266191844e-06,
      "loss": 1.9961,
      "step": 222684
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.500104308128357,
      "learning_rate": 1.6596765164476388e-06,
      "loss": 2.0872,
      "step": 222685
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.7129729986190796,
      "learning_rate": 1.659246861743191e-06,
      "loss": 2.1213,
      "step": 222686
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.5875027179718018,
      "learning_rate": 1.658817262505907e-06,
      "loss": 2.1116,
      "step": 222687
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.6557198762893677,
      "learning_rate": 1.6583877187358874e-06,
      "loss": 1.9714,
      "step": 222688
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.501239538192749,
      "learning_rate": 1.6579582304332317e-06,
      "loss": 1.9582,
      "step": 222689
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.4627864360809326,
      "learning_rate": 1.6575287975979735e-06,
      "loss": 1.859,
      "step": 222690
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.4170526266098022,
      "learning_rate": 1.6570994202302123e-06,
      "loss": 1.9261,
      "step": 222691
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.5358636379241943,
      "learning_rate": 1.6566700983300152e-06,
      "loss": 2.111,
      "step": 222692
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.34334135055542,
      "learning_rate": 1.6562408318975152e-06,
      "loss": 1.9452,
      "step": 222693
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.6495314836502075,
      "learning_rate": 1.6558116209327121e-06,
      "loss": 1.8514,
      "step": 222694
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.5737566947937012,
      "learning_rate": 1.6553824654357395e-06,
      "loss": 1.9608,
      "step": 222695
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.4578653573989868,
      "learning_rate": 1.6549533654066639e-06,
      "loss": 1.8096,
      "step": 222696
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.5059943199157715,
      "learning_rate": 1.6545243208455516e-06,
      "loss": 1.9536,
      "step": 222697
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.4764363765716553,
      "learning_rate": 1.6540953317525363e-06,
      "loss": 1.8888,
      "step": 222698
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.4568355083465576,
      "learning_rate": 1.6536663981276176e-06,
      "loss": 1.9315,
      "step": 222699
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.3810958862304688,
      "learning_rate": 1.6532375199708958e-06,
      "loss": 1.8412,
      "step": 222700
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.5568913221359253,
      "learning_rate": 1.6528086972825039e-06,
      "loss": 2.1317,
      "step": 222701
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.3945380449295044,
      "learning_rate": 1.6523799300625084e-06,
      "loss": 1.8465,
      "step": 222702
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.5346360206604004,
      "learning_rate": 1.6519512183109097e-06,
      "loss": 1.899,
      "step": 222703
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.467368245124817,
      "learning_rate": 1.6515225620278738e-06,
      "loss": 2.0336,
      "step": 222704
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.621667742729187,
      "learning_rate": 1.6510939612134677e-06,
      "loss": 2.0386,
      "step": 222705
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.4561623334884644,
      "learning_rate": 1.6506654158677246e-06,
      "loss": 1.9359,
      "step": 222706
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.5270826816558838,
      "learning_rate": 1.6502369259907777e-06,
      "loss": 1.9633,
      "step": 222707
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.4514110088348389,
      "learning_rate": 1.6498084915826605e-06,
      "loss": 1.8746,
      "step": 222708
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.603618860244751,
      "learning_rate": 1.6493801126435059e-06,
      "loss": 1.8947,
      "step": 222709
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.365048885345459,
      "learning_rate": 1.6489517891733473e-06,
      "loss": 1.7517,
      "step": 222710
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.4128050804138184,
      "learning_rate": 1.6485235211722514e-06,
      "loss": 2.0833,
      "step": 222711
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.3808388710021973,
      "learning_rate": 1.6480953086403847e-06,
      "loss": 1.8108,
      "step": 222712
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.4873522520065308,
      "learning_rate": 1.647667151577714e-06,
      "loss": 2.0554,
      "step": 222713
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.4635940790176392,
      "learning_rate": 1.6472390499844058e-06,
      "loss": 1.8428,
      "step": 222714
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.6191905736923218,
      "learning_rate": 1.6468110038604932e-06,
      "loss": 1.9558,
      "step": 222715
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.5464152097702026,
      "learning_rate": 1.6463830132061095e-06,
      "loss": 2.2238,
      "step": 222716
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.531646966934204,
      "learning_rate": 1.645955078021255e-06,
      "loss": 2.0405,
      "step": 222717
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.4309176206588745,
      "learning_rate": 1.6455271983060626e-06,
      "loss": 2.0643,
      "step": 222718
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.5206743478775024,
      "learning_rate": 1.6450993740605988e-06,
      "loss": 2.0238,
      "step": 222719
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.5362589359283447,
      "learning_rate": 1.6446716052849308e-06,
      "loss": 1.7618,
      "step": 222720
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.5889830589294434,
      "learning_rate": 1.6442438919791577e-06,
      "loss": 2.0518,
      "step": 222721
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.5564595460891724,
      "learning_rate": 1.643816234143347e-06,
      "loss": 1.9355,
      "step": 222722
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.4543797969818115,
      "learning_rate": 1.6433886317775646e-06,
      "loss": 1.974,
      "step": 222723
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.5383312702178955,
      "learning_rate": 1.642961084881944e-06,
      "loss": 2.0367,
      "step": 222724
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.5679062604904175,
      "learning_rate": 1.6425335934565187e-06,
      "loss": 2.0912,
      "step": 222725
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.4515876770019531,
      "learning_rate": 1.6421061575013884e-06,
      "loss": 1.9741,
      "step": 222726
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.4816950559616089,
      "learning_rate": 1.6416787770165863e-06,
      "loss": 2.1652,
      "step": 222727
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.4767382144927979,
      "learning_rate": 1.6412514520022456e-06,
      "loss": 2.0028,
      "step": 222728
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.4984036684036255,
      "learning_rate": 1.6408241824584334e-06,
      "loss": 1.977,
      "step": 222729
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.4118175506591797,
      "learning_rate": 1.6403969683852492e-06,
      "loss": 2.0837,
      "step": 222730
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.5079779624938965,
      "learning_rate": 1.639969809782693e-06,
      "loss": 1.9924,
      "step": 222731
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.4719316959381104,
      "learning_rate": 1.6395427066509648e-06,
      "loss": 2.0319,
      "step": 222732
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.419061541557312,
      "learning_rate": 1.6391156589900311e-06,
      "loss": 1.8962,
      "step": 222733
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.4488916397094727,
      "learning_rate": 1.6386886668000254e-06,
      "loss": 1.9065,
      "step": 222734
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.5775336027145386,
      "learning_rate": 1.6382617300810142e-06,
      "loss": 2.1106,
      "step": 222735
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.3324297666549683,
      "learning_rate": 1.6378348488330971e-06,
      "loss": 1.9623,
      "step": 222736
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.4829086065292358,
      "learning_rate": 1.637408023056308e-06,
      "loss": 2.0055,
      "step": 222737
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.5309765338897705,
      "learning_rate": 1.636981252750813e-06,
      "loss": 1.8894,
      "step": 222738
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.5733568668365479,
      "learning_rate": 1.636554537916579e-06,
      "loss": 2.0583,
      "step": 222739
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.6153384447097778,
      "learning_rate": 1.6361278785537724e-06,
      "loss": 2.1446,
      "step": 222740
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.5476552248001099,
      "learning_rate": 1.6357012746624266e-06,
      "loss": 1.8222,
      "step": 222741
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.5039304494857788,
      "learning_rate": 1.6352747262426413e-06,
      "loss": 1.9103,
      "step": 222742
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.6122028827667236,
      "learning_rate": 1.6348482332944834e-06,
      "loss": 2.0878,
      "step": 222743
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.5274670124053955,
      "learning_rate": 1.6344217958180527e-06,
      "loss": 2.1123,
      "step": 222744
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.4028822183609009,
      "learning_rate": 1.6339954138134158e-06,
      "loss": 2.2822,
      "step": 222745
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.4660032987594604,
      "learning_rate": 1.6335690872806395e-06,
      "loss": 1.827,
      "step": 222746
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.491148591041565,
      "learning_rate": 1.6331428162198234e-06,
      "loss": 1.8832,
      "step": 222747
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.5596883296966553,
      "learning_rate": 1.6327166006310344e-06,
      "loss": 1.9159,
      "step": 222748
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.3826109170913696,
      "learning_rate": 1.6322904405143388e-06,
      "loss": 2.0723,
      "step": 222749
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.3638334274291992,
      "learning_rate": 1.6318643358698701e-06,
      "loss": 1.9603,
      "step": 222750
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.4046200513839722,
      "learning_rate": 1.6314382866976616e-06,
      "loss": 1.8836,
      "step": 222751
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.442480206489563,
      "learning_rate": 1.6310122929977799e-06,
      "loss": 1.9635,
      "step": 222752
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.439814805984497,
      "learning_rate": 1.6305863547703245e-06,
      "loss": 1.5581,
      "step": 222753
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.5139127969741821,
      "learning_rate": 1.630160472015396e-06,
      "loss": 1.9564,
      "step": 222754
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.3683003187179565,
      "learning_rate": 1.6297346447330273e-06,
      "loss": 2.0324,
      "step": 222755
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.458795428276062,
      "learning_rate": 1.6293088729233517e-06,
      "loss": 2.0585,
      "step": 222756
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.4736106395721436,
      "learning_rate": 1.6288831565864025e-06,
      "loss": 1.7284,
      "step": 222757
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.3762309551239014,
      "learning_rate": 1.6284574957222796e-06,
      "loss": 1.9143,
      "step": 222758
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.4293882846832275,
      "learning_rate": 1.6280318903310497e-06,
      "loss": 1.9773,
      "step": 222759
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.4175593852996826,
      "learning_rate": 1.6276063404128125e-06,
      "loss": 1.9085,
      "step": 222760
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.5098217725753784,
      "learning_rate": 1.627180845967635e-06,
      "loss": 2.0659,
      "step": 222761
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.5650075674057007,
      "learning_rate": 1.6267554069956168e-06,
      "loss": 1.8279,
      "step": 222762
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.4669945240020752,
      "learning_rate": 1.6263300234967913e-06,
      "loss": 1.9087,
      "step": 222763
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.4428884983062744,
      "learning_rate": 1.6259046954712584e-06,
      "loss": 2.0776,
      "step": 222764
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.705883264541626,
      "learning_rate": 1.6254794229191182e-06,
      "loss": 1.814,
      "step": 222765
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.5678646564483643,
      "learning_rate": 1.6250542058404038e-06,
      "loss": 1.9848,
      "step": 222766
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.4310977458953857,
      "learning_rate": 1.6246290442352483e-06,
      "loss": 2.01,
      "step": 222767
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.6396406888961792,
      "learning_rate": 1.6242039381037187e-06,
      "loss": 2.0334,
      "step": 222768
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.5621840953826904,
      "learning_rate": 1.6237788874458813e-06,
      "loss": 1.8126,
      "step": 222769
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.4208685159683228,
      "learning_rate": 1.6233538922618029e-06,
      "loss": 1.7836,
      "step": 222770
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.3899646997451782,
      "learning_rate": 1.62292895255155e-06,
      "loss": 1.9047,
      "step": 222771
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.4088271856307983,
      "learning_rate": 1.6225040683152558e-06,
      "loss": 1.967,
      "step": 222772
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.518362045288086,
      "learning_rate": 1.6220792395529536e-06,
      "loss": 1.8857,
      "step": 222773
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.4004640579223633,
      "learning_rate": 1.6216544662647769e-06,
      "loss": 2.0406,
      "step": 222774
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.3410977125167847,
      "learning_rate": 1.6212297484507252e-06,
      "loss": 2.0409,
      "step": 222775
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.5953832864761353,
      "learning_rate": 1.6208050861109656e-06,
      "loss": 1.8624,
      "step": 222776
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.5330771207809448,
      "learning_rate": 1.6203804792454644e-06,
      "loss": 1.8405,
      "step": 222777
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.382859468460083,
      "learning_rate": 1.6199559278544217e-06,
      "loss": 2.0007,
      "step": 222778
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.4812732934951782,
      "learning_rate": 1.619531431937837e-06,
      "loss": 2.3314,
      "step": 222779
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.5638203620910645,
      "learning_rate": 1.619106991495811e-06,
      "loss": 1.9007,
      "step": 222780
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.6102060079574585,
      "learning_rate": 1.618682606528443e-06,
      "loss": 2.1681,
      "step": 222781
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.4107311964035034,
      "learning_rate": 1.6182582770357998e-06,
      "loss": 1.7567,
      "step": 222782
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.5547837018966675,
      "learning_rate": 1.617834003017915e-06,
      "loss": 1.9213,
      "step": 222783
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.5473746061325073,
      "learning_rate": 1.6174097844749545e-06,
      "loss": 2.0827,
      "step": 222784
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.4473875761032104,
      "learning_rate": 1.6169856214069188e-06,
      "loss": 2.0736,
      "step": 222785
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.4819992780685425,
      "learning_rate": 1.616561513813941e-06,
      "loss": 2.041,
      "step": 222786
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.3178932666778564,
      "learning_rate": 1.6161374616960543e-06,
      "loss": 1.987,
      "step": 222787
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.4749640226364136,
      "learning_rate": 1.6157134650533587e-06,
      "loss": 1.9846,
      "step": 222788
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.3992806673049927,
      "learning_rate": 1.6152895238859541e-06,
      "loss": 2.0684,
      "step": 222789
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.5457504987716675,
      "learning_rate": 1.6148656381939073e-06,
      "loss": 2.0301,
      "step": 222790
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.2975550889968872,
      "learning_rate": 1.6144418079772514e-06,
      "loss": 1.8274,
      "step": 222791
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.3562207221984863,
      "learning_rate": 1.6140180332361198e-06,
      "loss": 1.8282,
      "step": 222792
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.4617706537246704,
      "learning_rate": 1.6135943139705788e-06,
      "loss": 1.9051,
      "step": 222793
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.6977269649505615,
      "learning_rate": 1.6131706501806951e-06,
      "loss": 2.1405,
      "step": 222794
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.4372837543487549,
      "learning_rate": 1.6127470418665688e-06,
      "loss": 1.7049,
      "step": 222795
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.424013376235962,
      "learning_rate": 1.6123234890282665e-06,
      "loss": 2.0399,
      "step": 222796
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.4312303066253662,
      "learning_rate": 1.6118999916658548e-06,
      "loss": 1.835,
      "step": 222797
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.527490496635437,
      "learning_rate": 1.6114765497794335e-06,
      "loss": 1.8878,
      "step": 222798
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.8216465711593628,
      "learning_rate": 1.6110531633690694e-06,
      "loss": 1.9038,
      "step": 222799
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.493707299232483,
      "learning_rate": 1.6106298324348287e-06,
      "loss": 1.9792,
      "step": 222800
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.4986512660980225,
      "learning_rate": 1.6102065569768119e-06,
      "loss": 1.9938,
      "step": 222801
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.7054400444030762,
      "learning_rate": 1.6097833369950851e-06,
      "loss": 2.127,
      "step": 222802
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.4486126899719238,
      "learning_rate": 1.6093601724897487e-06,
      "loss": 1.9398,
      "step": 222803
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.6345115900039673,
      "learning_rate": 1.608937063460869e-06,
      "loss": 2.0887,
      "step": 222804
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.4918735027313232,
      "learning_rate": 1.6085140099085125e-06,
      "loss": 1.7796,
      "step": 222805
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.4984859228134155,
      "learning_rate": 1.608091011832746e-06,
      "loss": 1.713,
      "step": 222806
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.4545930624008179,
      "learning_rate": 1.607668069233703e-06,
      "loss": 2.1654,
      "step": 222807
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.5964770317077637,
      "learning_rate": 1.607245182111383e-06,
      "loss": 1.9989,
      "step": 222808
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.5137075185775757,
      "learning_rate": 1.6068223504659528e-06,
      "loss": 1.9907,
      "step": 222809
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.3397600650787354,
      "learning_rate": 1.6063995742974122e-06,
      "loss": 1.9918,
      "step": 222810
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.3686439990997314,
      "learning_rate": 1.6059768536058947e-06,
      "loss": 1.987,
      "step": 222811
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.436553955078125,
      "learning_rate": 1.6055541883914668e-06,
      "loss": 1.9013,
      "step": 222812
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.3760706186294556,
      "learning_rate": 1.6051315786541952e-06,
      "loss": 1.9284,
      "step": 222813
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.6470880508422852,
      "learning_rate": 1.6047090243941462e-06,
      "loss": 1.9632,
      "step": 222814
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.4212170839309692,
      "learning_rate": 1.6042865256114201e-06,
      "loss": 2.2118,
      "step": 222815
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.4940844774246216,
      "learning_rate": 1.6038640823060834e-06,
      "loss": 2.0657,
      "step": 222816
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.6017428636550903,
      "learning_rate": 1.6034416944782359e-06,
      "loss": 1.8835,
      "step": 222817
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.499904751777649,
      "learning_rate": 1.6030193621279442e-06,
      "loss": 1.917,
      "step": 222818
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.5043320655822754,
      "learning_rate": 1.602597085255275e-06,
      "loss": 1.9355,
      "step": 222819
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.4737080335617065,
      "learning_rate": 1.602174863860295e-06,
      "loss": 2.0723,
      "step": 222820
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.5202128887176514,
      "learning_rate": 1.6017526979431372e-06,
      "loss": 2.2656,
      "step": 222821
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.381075143814087,
      "learning_rate": 1.6013305875038018e-06,
      "loss": 2.0926,
      "step": 222822
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.4573770761489868,
      "learning_rate": 1.6009085325424554e-06,
      "loss": 1.9233,
      "step": 222823
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.5215568542480469,
      "learning_rate": 1.600486533059131e-06,
      "loss": 2.1696,
      "step": 222824
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.3585882186889648,
      "learning_rate": 1.6000645890538953e-06,
      "loss": 1.8725,
      "step": 222825
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.4945671558380127,
      "learning_rate": 1.5996427005268486e-06,
      "loss": 2.0952,
      "step": 222826
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.5050936937332153,
      "learning_rate": 1.5992208674780572e-06,
      "loss": 1.9959,
      "step": 222827
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.3457026481628418,
      "learning_rate": 1.5987990899075876e-06,
      "loss": 1.8839,
      "step": 222828
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.5476901531219482,
      "learning_rate": 1.5983773678155398e-06,
      "loss": 2.0581,
      "step": 222829
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.5812783241271973,
      "learning_rate": 1.5979557012020139e-06,
      "loss": 1.8948,
      "step": 222830
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.3999000787734985,
      "learning_rate": 1.597534090067043e-06,
      "loss": 2.0385,
      "step": 222831
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.4577194452285767,
      "learning_rate": 1.597112534410694e-06,
      "loss": 2.116,
      "step": 222832
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.4740148782730103,
      "learning_rate": 1.5966910342330996e-06,
      "loss": 2.0806,
      "step": 222833
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.5370954275131226,
      "learning_rate": 1.596269589534327e-06,
      "loss": 1.9046,
      "step": 222834
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.611714482307434,
      "learning_rate": 1.5958482003144423e-06,
      "loss": 1.8537,
      "step": 222835
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.4041965007781982,
      "learning_rate": 1.5954268665735125e-06,
      "loss": 1.7153,
      "step": 222836
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.4074783325195312,
      "learning_rate": 1.5950055883116043e-06,
      "loss": 1.7661,
      "step": 222837
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.4927172660827637,
      "learning_rate": 1.5945843655288172e-06,
      "loss": 1.8726,
      "step": 222838
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.4235018491744995,
      "learning_rate": 1.5941631982252845e-06,
      "loss": 1.9154,
      "step": 222839
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.6606141328811646,
      "learning_rate": 1.5937420864009732e-06,
      "loss": 1.9613,
      "step": 222840
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.450171709060669,
      "learning_rate": 1.5933210300560496e-06,
      "loss": 1.9384,
      "step": 222841
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.6321195363998413,
      "learning_rate": 1.5929000291905469e-06,
      "loss": 2.1752,
      "step": 222842
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.3654602766036987,
      "learning_rate": 1.5924790838045653e-06,
      "loss": 1.841,
      "step": 222843
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.5162193775177002,
      "learning_rate": 1.592058193898138e-06,
      "loss": 2.0344,
      "step": 222844
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.4775869846343994,
      "learning_rate": 1.5916373594714315e-06,
      "loss": 1.937,
      "step": 222845
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.398665428161621,
      "learning_rate": 1.5912165805244459e-06,
      "loss": 1.9252,
      "step": 222846
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.4449862241744995,
      "learning_rate": 1.5907958570573143e-06,
      "loss": 1.996,
      "step": 222847
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.5814539194107056,
      "learning_rate": 1.5903751890700367e-06,
      "loss": 1.9555,
      "step": 222848
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.433162808418274,
      "learning_rate": 1.5899545765627797e-06,
      "loss": 1.7303,
      "step": 222849
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.4212571382522583,
      "learning_rate": 1.5895340195355432e-06,
      "loss": 2.0164,
      "step": 222850
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.3194347620010376,
      "learning_rate": 1.589113517988494e-06,
      "loss": 2.0999,
      "step": 222851
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.473458170890808,
      "learning_rate": 1.5886930719216317e-06,
      "loss": 1.7882,
      "step": 222852
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.713720440864563,
      "learning_rate": 1.5882726813350898e-06,
      "loss": 1.9537,
      "step": 222853
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.5367106199264526,
      "learning_rate": 1.5878523462289018e-06,
      "loss": 1.9767,
      "step": 222854
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.438899278640747,
      "learning_rate": 1.587432066603167e-06,
      "loss": 2.107,
      "step": 222855
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.3376929759979248,
      "learning_rate": 1.5870118424579526e-06,
      "loss": 2.0705,
      "step": 222856
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.712280511856079,
      "learning_rate": 1.5865916737933581e-06,
      "loss": 1.9864,
      "step": 222857
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.3942830562591553,
      "learning_rate": 1.5861715606094505e-06,
      "loss": 2.2918,
      "step": 222858
    },
    {
      "epoch": 2.9,
      "grad_norm": 2.473532199859619,
      "learning_rate": 1.5857515029062962e-06,
      "loss": 2.0604,
      "step": 222859
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.394235372543335,
      "learning_rate": 1.585331500683995e-06,
      "loss": 2.0259,
      "step": 222860
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.5841538906097412,
      "learning_rate": 1.5849115539426138e-06,
      "loss": 1.9941,
      "step": 222861
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.4345977306365967,
      "learning_rate": 1.584491662682219e-06,
      "loss": 1.877,
      "step": 222862
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.5940167903900146,
      "learning_rate": 1.5840718269029107e-06,
      "loss": 1.9343,
      "step": 222863
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.496751308441162,
      "learning_rate": 1.5836520466047554e-06,
      "loss": 1.8764,
      "step": 222864
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.548693299293518,
      "learning_rate": 1.5832323217878196e-06,
      "loss": 1.9721,
      "step": 222865
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.4828819036483765,
      "learning_rate": 1.5828126524522034e-06,
      "loss": 1.9833,
      "step": 222866
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.5501000881195068,
      "learning_rate": 1.5823930385979733e-06,
      "loss": 2.0686,
      "step": 222867
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.7298808097839355,
      "learning_rate": 1.5819734802252292e-06,
      "loss": 2.0344,
      "step": 222868
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.4559913873672485,
      "learning_rate": 1.5815539773340046e-06,
      "loss": 2.1918,
      "step": 222869
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.4985377788543701,
      "learning_rate": 1.5811345299243994e-06,
      "loss": 1.7539,
      "step": 222870
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.3788678646087646,
      "learning_rate": 1.5807151379965132e-06,
      "loss": 2.0163,
      "step": 222871
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.3642922639846802,
      "learning_rate": 1.580295801550413e-06,
      "loss": 1.787,
      "step": 222872
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.5334223508834839,
      "learning_rate": 1.5798765205861318e-06,
      "loss": 2.057,
      "step": 222873
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.5021387338638306,
      "learning_rate": 1.5794572951038031e-06,
      "loss": 1.8502,
      "step": 222874
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.6080552339553833,
      "learning_rate": 1.5790381251034934e-06,
      "loss": 1.7957,
      "step": 222875
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.412214994430542,
      "learning_rate": 1.5786190105852692e-06,
      "loss": 1.8194,
      "step": 222876
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.5568758249282837,
      "learning_rate": 1.5781999515492305e-06,
      "loss": 2.0026,
      "step": 222877
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.7559139728546143,
      "learning_rate": 1.5777809479954107e-06,
      "loss": 1.8306,
      "step": 222878
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.5357046127319336,
      "learning_rate": 1.5773619999239095e-06,
      "loss": 1.9784,
      "step": 222879
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.4371083974838257,
      "learning_rate": 1.5769431073348271e-06,
      "loss": 1.8587,
      "step": 222880
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.4747810363769531,
      "learning_rate": 1.5765242702282299e-06,
      "loss": 2.0229,
      "step": 222881
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.6921627521514893,
      "learning_rate": 1.5761054886041846e-06,
      "loss": 1.9217,
      "step": 222882
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.421932578086853,
      "learning_rate": 1.5756867624627579e-06,
      "loss": 1.9131,
      "step": 222883
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.5700522661209106,
      "learning_rate": 1.5752680918040828e-06,
      "loss": 2.0937,
      "step": 222884
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.6353412866592407,
      "learning_rate": 1.5748494766281595e-06,
      "loss": 1.9756,
      "step": 222885
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.4862570762634277,
      "learning_rate": 1.574430916935121e-06,
      "loss": 1.9975,
      "step": 222886
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.3996528387069702,
      "learning_rate": 1.574012412725001e-06,
      "loss": 2.0432,
      "step": 222887
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.599125623703003,
      "learning_rate": 1.5735939639979322e-06,
      "loss": 2.1456,
      "step": 222888
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.729786992073059,
      "learning_rate": 1.5731755707539816e-06,
      "loss": 1.9949,
      "step": 222889
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.6282504796981812,
      "learning_rate": 1.5727572329931826e-06,
      "loss": 2.0845,
      "step": 222890
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.484747052192688,
      "learning_rate": 1.572338950715668e-06,
      "loss": 2.0238,
      "step": 222891
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.5240416526794434,
      "learning_rate": 1.5719207239214715e-06,
      "loss": 2.0734,
      "step": 222892
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.5269650220870972,
      "learning_rate": 1.5715025526106928e-06,
      "loss": 2.3672,
      "step": 222893
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.442731499671936,
      "learning_rate": 1.5710844367833985e-06,
      "loss": 1.9545,
      "step": 222894
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.4220268726348877,
      "learning_rate": 1.5706663764396555e-06,
      "loss": 1.9932,
      "step": 222895
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.4434133768081665,
      "learning_rate": 1.5702483715795965e-06,
      "loss": 1.8291,
      "step": 222896
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.4201709032058716,
      "learning_rate": 1.5698304222032553e-06,
      "loss": 1.9187,
      "step": 222897
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.6767578125,
      "learning_rate": 1.5694125283106983e-06,
      "loss": 2.1198,
      "step": 222898
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.442716121673584,
      "learning_rate": 1.5689946899020255e-06,
      "loss": 1.9982,
      "step": 222899
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.5147466659545898,
      "learning_rate": 1.5685769069773036e-06,
      "loss": 2.1603,
      "step": 222900
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.4303642511367798,
      "learning_rate": 1.5681591795366321e-06,
      "loss": 1.795,
      "step": 222901
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.3686625957489014,
      "learning_rate": 1.5677415075800782e-06,
      "loss": 1.9626,
      "step": 222902
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.432911992073059,
      "learning_rate": 1.5673238911076746e-06,
      "loss": 1.7699,
      "step": 222903
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.3481134176254272,
      "learning_rate": 1.5669063301195884e-06,
      "loss": 2.1102,
      "step": 222904
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.4437808990478516,
      "learning_rate": 1.566488824615819e-06,
      "loss": 1.9444,
      "step": 222905
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.4850640296936035,
      "learning_rate": 1.566071374596467e-06,
      "loss": 1.9556,
      "step": 222906
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.5648133754730225,
      "learning_rate": 1.5656539800616318e-06,
      "loss": 1.9455,
      "step": 222907
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.4352688789367676,
      "learning_rate": 1.5652366410113804e-06,
      "loss": 1.9049,
      "step": 222908
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.3663932085037231,
      "learning_rate": 1.564819357445779e-06,
      "loss": 1.9662,
      "step": 222909
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.3650325536727905,
      "learning_rate": 1.5644021293649278e-06,
      "loss": 1.9092,
      "step": 222910
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.4726608991622925,
      "learning_rate": 1.56398495676886e-06,
      "loss": 1.8019,
      "step": 222911
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.4340205192565918,
      "learning_rate": 1.5635678396576756e-06,
      "loss": 1.9372,
      "step": 222912
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.3853747844696045,
      "learning_rate": 1.5631507780314746e-06,
      "loss": 2.1079,
      "step": 222913
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.4107197523117065,
      "learning_rate": 1.5627337718903232e-06,
      "loss": 1.9371,
      "step": 222914
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.6295568943023682,
      "learning_rate": 1.5623168212342885e-06,
      "loss": 2.0131,
      "step": 222915
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.5628212690353394,
      "learning_rate": 1.5618999260634368e-06,
      "loss": 1.7965,
      "step": 222916
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.5561224222183228,
      "learning_rate": 1.5614830863778682e-06,
      "loss": 2.0103,
      "step": 222917
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.4984451532363892,
      "learning_rate": 1.5610663021776492e-06,
      "loss": 2.0848,
      "step": 222918
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.5092084407806396,
      "learning_rate": 1.5606495734628799e-06,
      "loss": 2.0816,
      "step": 222919
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.512279987335205,
      "learning_rate": 1.5602329002336267e-06,
      "loss": 2.1031,
      "step": 222920
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.4581408500671387,
      "learning_rate": 1.5598162824899229e-06,
      "loss": 2.0243,
      "step": 222921
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.4227370023727417,
      "learning_rate": 1.559399720231902e-06,
      "loss": 1.6644,
      "step": 222922
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.433760166168213,
      "learning_rate": 1.5589832134596304e-06,
      "loss": 1.8768,
      "step": 222923
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.300336480140686,
      "learning_rate": 1.5585667621731745e-06,
      "loss": 1.9273,
      "step": 222924
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.5891975164413452,
      "learning_rate": 1.558150366372568e-06,
      "loss": 2.072,
      "step": 222925
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.3698228597640991,
      "learning_rate": 1.5577340260580106e-06,
      "loss": 1.7276,
      "step": 222926
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.439751148223877,
      "learning_rate": 1.557317741229469e-06,
      "loss": 1.8969,
      "step": 222927
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.6373109817504883,
      "learning_rate": 1.5569015118870431e-06,
      "loss": 2.2036,
      "step": 222928
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.4970030784606934,
      "learning_rate": 1.5564853380308329e-06,
      "loss": 1.7957,
      "step": 222929
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.4765286445617676,
      "learning_rate": 1.5560692196609047e-06,
      "loss": 1.9634,
      "step": 222930
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.4564276933670044,
      "learning_rate": 1.5556531567773256e-06,
      "loss": 2.1092,
      "step": 222931
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.4208402633666992,
      "learning_rate": 1.5552371493802286e-06,
      "loss": 2.2547,
      "step": 222932
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.4508814811706543,
      "learning_rate": 1.5548211974696133e-06,
      "loss": 1.9888,
      "step": 222933
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.3982303142547607,
      "learning_rate": 1.5544053010455803e-06,
      "loss": 1.8225,
      "step": 222934
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.594902515411377,
      "learning_rate": 1.5539894601082292e-06,
      "loss": 1.9226,
      "step": 222935
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.4739131927490234,
      "learning_rate": 1.5535736746576267e-06,
      "loss": 2.069,
      "step": 222936
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.6223392486572266,
      "learning_rate": 1.5531579446938391e-06,
      "loss": 2.1027,
      "step": 222937
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.4088222980499268,
      "learning_rate": 1.5527422702169666e-06,
      "loss": 2.0296,
      "step": 222938
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.4204723834991455,
      "learning_rate": 1.552326651227076e-06,
      "loss": 2.011,
      "step": 222939
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.4393800497055054,
      "learning_rate": 1.5519110877242335e-06,
      "loss": 2.0125,
      "step": 222940
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.4323163032531738,
      "learning_rate": 1.5514955797085393e-06,
      "loss": 2.0331,
      "step": 222941
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.470504879951477,
      "learning_rate": 1.5510801271800266e-06,
      "loss": 1.9911,
      "step": 222942
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.4185551404953003,
      "learning_rate": 1.5506647301388287e-06,
      "loss": 2.0737,
      "step": 222943
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.4488216638565063,
      "learning_rate": 1.5502493885849787e-06,
      "loss": 2.1757,
      "step": 222944
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.5020464658737183,
      "learning_rate": 1.5498341025185767e-06,
      "loss": 1.7402,
      "step": 222945
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.6058580875396729,
      "learning_rate": 1.5494188719396893e-06,
      "loss": 1.9534,
      "step": 222946
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.3812432289123535,
      "learning_rate": 1.5490036968484165e-06,
      "loss": 1.7619,
      "step": 222947
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.4260668754577637,
      "learning_rate": 1.5485885772447915e-06,
      "loss": 2.1615,
      "step": 222948
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.5583109855651855,
      "learning_rate": 1.5481735131289142e-06,
      "loss": 1.8731,
      "step": 222949
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.5513291358947754,
      "learning_rate": 1.5477585045008845e-06,
      "loss": 1.9784,
      "step": 222950
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.518221378326416,
      "learning_rate": 1.547343551360769e-06,
      "loss": 2.0696,
      "step": 222951
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.5925333499908447,
      "learning_rate": 1.5469286537086346e-06,
      "loss": 1.8404,
      "step": 222952
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.4018633365631104,
      "learning_rate": 1.5465138115445475e-06,
      "loss": 1.7397,
      "step": 222953
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.3760164976119995,
      "learning_rate": 1.5460990248685744e-06,
      "loss": 1.9381,
      "step": 222954
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.459741473197937,
      "learning_rate": 1.5456842936808489e-06,
      "loss": 1.9321,
      "step": 222955
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.4306880235671997,
      "learning_rate": 1.545269617981404e-06,
      "loss": 1.7984,
      "step": 222956
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.4611607789993286,
      "learning_rate": 1.5448549977703396e-06,
      "loss": 1.9841,
      "step": 222957
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.4929600954055786,
      "learning_rate": 1.544440433047689e-06,
      "loss": 1.8437,
      "step": 222958
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.4711475372314453,
      "learning_rate": 1.5440259238135854e-06,
      "loss": 1.8584,
      "step": 222959
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.376542329788208,
      "learning_rate": 1.5436114700680958e-06,
      "loss": 1.9948,
      "step": 222960
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.3495022058486938,
      "learning_rate": 1.543197071811253e-06,
      "loss": 1.8293,
      "step": 222961
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.5793911218643188,
      "learning_rate": 1.5427827290431572e-06,
      "loss": 1.9487,
      "step": 222962
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.6061227321624756,
      "learning_rate": 1.5423684417639416e-06,
      "loss": 1.9792,
      "step": 222963
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.626875400543213,
      "learning_rate": 1.5419542099735727e-06,
      "loss": 2.0935,
      "step": 222964
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.5415273904800415,
      "learning_rate": 1.5415400336722172e-06,
      "loss": 2.1091,
      "step": 222965
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.4808094501495361,
      "learning_rate": 1.5411259128599084e-06,
      "loss": 1.771,
      "step": 222966
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.328233003616333,
      "learning_rate": 1.5407118475367796e-06,
      "loss": 1.9872,
      "step": 222967
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.5588102340698242,
      "learning_rate": 1.5402978377028309e-06,
      "loss": 2.0486,
      "step": 222968
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.5190763473510742,
      "learning_rate": 1.539883883358195e-06,
      "loss": 1.9108,
      "step": 222969
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.5213004350662231,
      "learning_rate": 1.5394699845029057e-06,
      "loss": 1.9284,
      "step": 222970
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.7357547283172607,
      "learning_rate": 1.5390561411370627e-06,
      "loss": 2.1006,
      "step": 222971
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.5470017194747925,
      "learning_rate": 1.538642353260766e-06,
      "loss": 2.1317,
      "step": 222972
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.420621395111084,
      "learning_rate": 1.538228620874049e-06,
      "loss": 2.0592,
      "step": 222973
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.5640202760696411,
      "learning_rate": 1.5378149439770116e-06,
      "loss": 2.0416,
      "step": 222974
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.4461405277252197,
      "learning_rate": 1.5374013225697536e-06,
      "loss": 2.0186,
      "step": 222975
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.4052953720092773,
      "learning_rate": 1.536987756652308e-06,
      "loss": 1.9118,
      "step": 222976
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.5879603624343872,
      "learning_rate": 1.5365742462247754e-06,
      "loss": 2.1145,
      "step": 222977
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.5557559728622437,
      "learning_rate": 1.5361607912871888e-06,
      "loss": 2.0163,
      "step": 222978
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.524822473526001,
      "learning_rate": 1.5357473918397146e-06,
      "loss": 1.924,
      "step": 222979
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.3078203201293945,
      "learning_rate": 1.5353340478823528e-06,
      "loss": 2.0593,
      "step": 222980
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.3530415296554565,
      "learning_rate": 1.5349207594152369e-06,
      "loss": 1.8099,
      "step": 222981
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.451527714729309,
      "learning_rate": 1.5345075264383665e-06,
      "loss": 1.9945,
      "step": 222982
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.4422818422317505,
      "learning_rate": 1.5340943489519086e-06,
      "loss": 2.0586,
      "step": 222983
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.618185043334961,
      "learning_rate": 1.533681226955863e-06,
      "loss": 2.0536,
      "step": 222984
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.6169402599334717,
      "learning_rate": 1.5332681604503627e-06,
      "loss": 1.9485,
      "step": 222985
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.3845975399017334,
      "learning_rate": 1.5328551494354412e-06,
      "loss": 2.0952,
      "step": 222986
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.4818766117095947,
      "learning_rate": 1.532442193911232e-06,
      "loss": 1.9341,
      "step": 222987
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.45683753490448,
      "learning_rate": 1.5320292938777345e-06,
      "loss": 2.0431,
      "step": 222988
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.5466899871826172,
      "learning_rate": 1.5316164493350824e-06,
      "loss": 1.8991,
      "step": 222989
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.495956301689148,
      "learning_rate": 1.5312036602833423e-06,
      "loss": 1.8701,
      "step": 222990
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.4723289012908936,
      "learning_rate": 1.5307909267225804e-06,
      "loss": 2.0775,
      "step": 222991
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.4960013628005981,
      "learning_rate": 1.5303782486528636e-06,
      "loss": 1.9445,
      "step": 222992
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.5316052436828613,
      "learning_rate": 1.5299656260743254e-06,
      "loss": 2.0039,
      "step": 222993
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.5066397190093994,
      "learning_rate": 1.5295530589869652e-06,
      "loss": 2.0883,
      "step": 222994
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.544475793838501,
      "learning_rate": 1.5291405473909168e-06,
      "loss": 1.9844,
      "step": 222995
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.4469743967056274,
      "learning_rate": 1.5287280912862133e-06,
      "loss": 1.9493,
      "step": 222996
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.4366589784622192,
      "learning_rate": 1.5283156906729543e-06,
      "loss": 1.9433,
      "step": 222997
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.4109628200531006,
      "learning_rate": 1.5279033455512401e-06,
      "loss": 2.0368,
      "step": 222998
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.4196083545684814,
      "learning_rate": 1.527491055921104e-06,
      "loss": 2.0275,
      "step": 222999
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.3812975883483887,
      "learning_rate": 1.5270788217826456e-06,
      "loss": 1.9295,
      "step": 223000
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.5416975021362305,
      "learning_rate": 1.526666643135932e-06,
      "loss": 2.3763,
      "step": 223001
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.532102346420288,
      "learning_rate": 1.5262545199810628e-06,
      "loss": 2.0151,
      "step": 223002
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.4272675514221191,
      "learning_rate": 1.5258424523180712e-06,
      "loss": 1.6925,
      "step": 223003
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.5098775625228882,
      "learning_rate": 1.5254304401470573e-06,
      "loss": 2.0442,
      "step": 223004
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.5499062538146973,
      "learning_rate": 1.5250184834681213e-06,
      "loss": 1.8677,
      "step": 223005
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.5217725038528442,
      "learning_rate": 1.5246065822813292e-06,
      "loss": 2.1458,
      "step": 223006
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.5527056455612183,
      "learning_rate": 1.524194736586748e-06,
      "loss": 1.8883,
      "step": 223007
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.408341646194458,
      "learning_rate": 1.523782946384411e-06,
      "loss": 2.0966,
      "step": 223008
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.372896671295166,
      "learning_rate": 1.5233712116744512e-06,
      "loss": 1.8609,
      "step": 223009
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.40507173538208,
      "learning_rate": 1.5229595324569355e-06,
      "loss": 2.0014,
      "step": 223010
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.462817907333374,
      "learning_rate": 1.5225479087319637e-06,
      "loss": 2.0539,
      "step": 223011
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.5036487579345703,
      "learning_rate": 1.522136340499569e-06,
      "loss": 1.8092,
      "step": 223012
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.530017614364624,
      "learning_rate": 1.5217248277598181e-06,
      "loss": 1.8898,
      "step": 223013
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.3890215158462524,
      "learning_rate": 1.5213133705128443e-06,
      "loss": 2.0534,
      "step": 223014
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.4833943843841553,
      "learning_rate": 1.520901968758681e-06,
      "loss": 1.697,
      "step": 223015
    },
    {
      "epoch": 2.9,
      "grad_norm": 2.151110887527466,
      "learning_rate": 1.5204906224973944e-06,
      "loss": 2.1005,
      "step": 223016
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.4925340414047241,
      "learning_rate": 1.5200793317291183e-06,
      "loss": 1.9666,
      "step": 223017
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.6619391441345215,
      "learning_rate": 1.5196680964538855e-06,
      "loss": 2.0746,
      "step": 223018
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.4939029216766357,
      "learning_rate": 1.5192569166717628e-06,
      "loss": 2.0646,
      "step": 223019
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.4941473007202148,
      "learning_rate": 1.51884579238285e-06,
      "loss": 2.0456,
      "step": 223020
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.5794684886932373,
      "learning_rate": 1.5184347235872473e-06,
      "loss": 1.9094,
      "step": 223021
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.400107979774475,
      "learning_rate": 1.5180237102849545e-06,
      "loss": 1.8714,
      "step": 223022
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.6211837530136108,
      "learning_rate": 1.5176127524761384e-06,
      "loss": 2.1132,
      "step": 223023
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.4550158977508545,
      "learning_rate": 1.5172018501607986e-06,
      "loss": 2.0258,
      "step": 223024
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.5753412246704102,
      "learning_rate": 1.5167910033390685e-06,
      "loss": 1.9866,
      "step": 223025
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.4905296564102173,
      "learning_rate": 1.5163802120109814e-06,
      "loss": 1.9592,
      "step": 223026
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.4438552856445312,
      "learning_rate": 1.5159694761766373e-06,
      "loss": 2.0367,
      "step": 223027
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.4831395149230957,
      "learning_rate": 1.5155587958361027e-06,
      "loss": 1.953,
      "step": 223028
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.4583659172058105,
      "learning_rate": 1.5151481709894776e-06,
      "loss": 1.7727,
      "step": 223029
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.5231833457946777,
      "learning_rate": 1.5147376016368284e-06,
      "loss": 1.9696,
      "step": 223030
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.5183082818984985,
      "learning_rate": 1.514327087778222e-06,
      "loss": 1.7748,
      "step": 223031
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.5144739151000977,
      "learning_rate": 1.513916629413725e-06,
      "loss": 2.0269,
      "step": 223032
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.5137046575546265,
      "learning_rate": 1.5135062265434372e-06,
      "loss": 1.9823,
      "step": 223033
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.386646032333374,
      "learning_rate": 1.5130958791673919e-06,
      "loss": 1.9073,
      "step": 223034
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.516761064529419,
      "learning_rate": 1.5126855872857225e-06,
      "loss": 1.9936,
      "step": 223035
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.5413622856140137,
      "learning_rate": 1.5122753508984619e-06,
      "loss": 1.8031,
      "step": 223036
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.442260980606079,
      "learning_rate": 1.5118651700057439e-06,
      "loss": 2.0851,
      "step": 223037
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.365143060684204,
      "learning_rate": 1.5114550446075679e-06,
      "loss": 1.7533,
      "step": 223038
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.5390254259109497,
      "learning_rate": 1.5110449747040342e-06,
      "loss": 1.9753,
      "step": 223039
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.3935508728027344,
      "learning_rate": 1.5106349602952428e-06,
      "loss": 1.7668,
      "step": 223040
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.5406205654144287,
      "learning_rate": 1.5102250013812933e-06,
      "loss": 1.9019,
      "step": 223041
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.595945954322815,
      "learning_rate": 1.5098150979622192e-06,
      "loss": 1.8321,
      "step": 223042
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.4768012762069702,
      "learning_rate": 1.509405250038054e-06,
      "loss": 2.0607,
      "step": 223043
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.4372864961624146,
      "learning_rate": 1.5089954576089635e-06,
      "loss": 2.0195,
      "step": 223044
    },
    {
      "epoch": 2.9,
      "grad_norm": 2.1774609088897705,
      "learning_rate": 1.5085857206749818e-06,
      "loss": 1.8483,
      "step": 223045
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.6580524444580078,
      "learning_rate": 1.5081760392361753e-06,
      "loss": 1.8802,
      "step": 223046
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.4118125438690186,
      "learning_rate": 1.507766413292677e-06,
      "loss": 1.9548,
      "step": 223047
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.5025897026062012,
      "learning_rate": 1.507356842844487e-06,
      "loss": 1.9661,
      "step": 223048
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.5053677558898926,
      "learning_rate": 1.5069473278917054e-06,
      "loss": 1.9833,
      "step": 223049
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.4483585357666016,
      "learning_rate": 1.506537868434432e-06,
      "loss": 1.8757,
      "step": 223050
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.666299819946289,
      "learning_rate": 1.5061284644727333e-06,
      "loss": 2.2282,
      "step": 223051
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.4439278841018677,
      "learning_rate": 1.5057191160066428e-06,
      "loss": 2.0189,
      "step": 223052
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.4896619319915771,
      "learning_rate": 1.5053098230362936e-06,
      "loss": 2.0903,
      "step": 223053
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.5793976783752441,
      "learning_rate": 1.5049005855617523e-06,
      "loss": 2.0395,
      "step": 223054
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.4948307275772095,
      "learning_rate": 1.5044914035830858e-06,
      "loss": 1.939,
      "step": 223055
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.6704010963439941,
      "learning_rate": 1.5040822771003602e-06,
      "loss": 1.8435,
      "step": 223056
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.51227867603302,
      "learning_rate": 1.5036732061136425e-06,
      "loss": 2.0247,
      "step": 223057
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.5078942775726318,
      "learning_rate": 1.5032641906230658e-06,
      "loss": 1.9108,
      "step": 223058
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.4443002939224243,
      "learning_rate": 1.5028552306286635e-06,
      "loss": 1.8578,
      "step": 223059
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.3865731954574585,
      "learning_rate": 1.502446326130502e-06,
      "loss": 2.0674,
      "step": 223060
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.5524390935897827,
      "learning_rate": 1.502037477128648e-06,
      "loss": 1.877,
      "step": 223061
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.519956111907959,
      "learning_rate": 1.501628683623235e-06,
      "loss": 1.9693,
      "step": 223062
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.46638822555542,
      "learning_rate": 1.5012199456142625e-06,
      "loss": 1.9794,
      "step": 223063
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.5546329021453857,
      "learning_rate": 1.5008112631018976e-06,
      "loss": 2.1454,
      "step": 223064
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.3687782287597656,
      "learning_rate": 1.5004026360861065e-06,
      "loss": 1.9403,
      "step": 223065
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.4672753810882568,
      "learning_rate": 1.4999940645670893e-06,
      "loss": 1.9171,
      "step": 223066
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.5517939329147339,
      "learning_rate": 1.4995855485448127e-06,
      "loss": 1.9542,
      "step": 223067
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.439422845840454,
      "learning_rate": 1.499177088019443e-06,
      "loss": 1.8631,
      "step": 223068
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.600865125656128,
      "learning_rate": 1.4987686829909473e-06,
      "loss": 1.9312,
      "step": 223069
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.5453135967254639,
      "learning_rate": 1.498360333459525e-06,
      "loss": 1.7827,
      "step": 223070
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.5145387649536133,
      "learning_rate": 1.4979520394251433e-06,
      "loss": 2.0204,
      "step": 223071
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.3978713750839233,
      "learning_rate": 1.497543800887968e-06,
      "loss": 2.1486,
      "step": 223072
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.3335978984832764,
      "learning_rate": 1.4971356178480332e-06,
      "loss": 1.8691,
      "step": 223073
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.6353071928024292,
      "learning_rate": 1.496727490305405e-06,
      "loss": 2.097,
      "step": 223074
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.4655295610427856,
      "learning_rate": 1.4963194182601502e-06,
      "loss": 1.9922,
      "step": 223075
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.7131426334381104,
      "learning_rate": 1.4959114017123686e-06,
      "loss": 2.0348,
      "step": 223076
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.455183506011963,
      "learning_rate": 1.4955034406621268e-06,
      "loss": 1.7057,
      "step": 223077
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.4831281900405884,
      "learning_rate": 1.495095535109525e-06,
      "loss": 1.6757,
      "step": 223078
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.388609766960144,
      "learning_rate": 1.4946876850546296e-06,
      "loss": 2.0758,
      "step": 223079
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.4651687145233154,
      "learning_rate": 1.4942798904975073e-06,
      "loss": 2.1453,
      "step": 223080
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.5243011713027954,
      "learning_rate": 1.4938721514381912e-06,
      "loss": 2.0274,
      "step": 223081
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.39776611328125,
      "learning_rate": 1.4934644678768482e-06,
      "loss": 1.9174,
      "step": 223082
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.6678943634033203,
      "learning_rate": 1.493056839813478e-06,
      "loss": 2.0643,
      "step": 223083
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.5095525979995728,
      "learning_rate": 1.4926492672481804e-06,
      "loss": 1.8054,
      "step": 223084
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.3731621503829956,
      "learning_rate": 1.4922417501810557e-06,
      "loss": 1.8598,
      "step": 223085
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.495821237564087,
      "learning_rate": 1.491834288612137e-06,
      "loss": 1.8255,
      "step": 223086
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.3977056741714478,
      "learning_rate": 1.4914268825415244e-06,
      "loss": 2.1483,
      "step": 223087
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.5280005931854248,
      "learning_rate": 1.4910195319692842e-06,
      "loss": 2.0875,
      "step": 223088
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.5199300050735474,
      "learning_rate": 1.4906122368955164e-06,
      "loss": 1.9497,
      "step": 223089
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.417724847793579,
      "learning_rate": 1.4902049973202546e-06,
      "loss": 1.7612,
      "step": 223090
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.4099891185760498,
      "learning_rate": 1.4897978132436316e-06,
      "loss": 1.8687,
      "step": 223091
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.5055376291275024,
      "learning_rate": 1.489390684665681e-06,
      "loss": 2.0788,
      "step": 223092
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.420729160308838,
      "learning_rate": 1.4889836115864695e-06,
      "loss": 1.8835,
      "step": 223093
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.4877523183822632,
      "learning_rate": 1.4885765940060968e-06,
      "loss": 1.9575,
      "step": 223094
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.5568190813064575,
      "learning_rate": 1.4881696319246295e-06,
      "loss": 1.9659,
      "step": 223095
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.4823349714279175,
      "learning_rate": 1.4877627253421676e-06,
      "loss": 2.1039,
      "step": 223096
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.406553864479065,
      "learning_rate": 1.4873558742587444e-06,
      "loss": 2.0352,
      "step": 223097
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.727626919746399,
      "learning_rate": 1.4869490786744598e-06,
      "loss": 1.8576,
      "step": 223098
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.3566863536834717,
      "learning_rate": 1.4865423385893805e-06,
      "loss": 2.1034,
      "step": 223099
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.5239520072937012,
      "learning_rate": 1.4861356540036062e-06,
      "loss": 1.8687,
      "step": 223100
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.3886059522628784,
      "learning_rate": 1.4857290249171705e-06,
      "loss": 2.034,
      "step": 223101
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.4956169128417969,
      "learning_rate": 1.4853224513302065e-06,
      "loss": 1.8392,
      "step": 223102
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.5929043292999268,
      "learning_rate": 1.4849159332427474e-06,
      "loss": 2.0197,
      "step": 223103
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.6269035339355469,
      "learning_rate": 1.48450947065486e-06,
      "loss": 1.9359,
      "step": 223104
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.4758180379867554,
      "learning_rate": 1.484103063566644e-06,
      "loss": 1.9112,
      "step": 223105
    },
    {
      "epoch": 2.9,
      "grad_norm": 1.5881407260894775,
      "learning_rate": 1.4836967119781663e-06,
      "loss": 2.1555,
      "step": 223106
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.5166691541671753,
      "learning_rate": 1.4832904158894931e-06,
      "loss": 1.782,
      "step": 223107
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.332828402519226,
      "learning_rate": 1.4828841753007248e-06,
      "loss": 2.0104,
      "step": 223108
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.6397221088409424,
      "learning_rate": 1.4824779902119276e-06,
      "loss": 1.8976,
      "step": 223109
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.418289303779602,
      "learning_rate": 1.4820718606231684e-06,
      "loss": 2.3228,
      "step": 223110
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.4473360776901245,
      "learning_rate": 1.481665786534514e-06,
      "loss": 1.9501,
      "step": 223111
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.4937294721603394,
      "learning_rate": 1.4812597679460636e-06,
      "loss": 1.9507,
      "step": 223112
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.5783759355545044,
      "learning_rate": 1.4808538048578845e-06,
      "loss": 1.8387,
      "step": 223113
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.4153265953063965,
      "learning_rate": 1.4804478972700428e-06,
      "loss": 1.9013,
      "step": 223114
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.4418355226516724,
      "learning_rate": 1.480042045182639e-06,
      "loss": 1.8417,
      "step": 223115
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.6231105327606201,
      "learning_rate": 1.4796362485957058e-06,
      "loss": 1.9856,
      "step": 223116
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.4810503721237183,
      "learning_rate": 1.4792305075093435e-06,
      "loss": 1.8396,
      "step": 223117
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.4535632133483887,
      "learning_rate": 1.478824821923652e-06,
      "loss": 2.0072,
      "step": 223118
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.409566879272461,
      "learning_rate": 1.4784191918386646e-06,
      "loss": 1.9202,
      "step": 223119
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.42316734790802,
      "learning_rate": 1.478013617254481e-06,
      "loss": 1.9063,
      "step": 223120
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.6888911724090576,
      "learning_rate": 1.477608098171168e-06,
      "loss": 2.2674,
      "step": 223121
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.4928580522537231,
      "learning_rate": 1.4772026345888256e-06,
      "loss": 2.2421,
      "step": 223122
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.4280799627304077,
      "learning_rate": 1.4767972265074535e-06,
      "loss": 2.102,
      "step": 223123
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.5469145774841309,
      "learning_rate": 1.4763918739272184e-06,
      "loss": 1.9829,
      "step": 223124
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.4562143087387085,
      "learning_rate": 1.4759865768481537e-06,
      "loss": 2.0107,
      "step": 223125
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.5654550790786743,
      "learning_rate": 1.475581335270326e-06,
      "loss": 1.8676,
      "step": 223126
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.3899095058441162,
      "learning_rate": 1.4751761491938353e-06,
      "loss": 2.0229,
      "step": 223127
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.3675005435943604,
      "learning_rate": 1.4747710186187145e-06,
      "loss": 1.8748,
      "step": 223128
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.58896803855896,
      "learning_rate": 1.474365943545097e-06,
      "loss": 1.9322,
      "step": 223129
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.5707954168319702,
      "learning_rate": 1.4739609239730499e-06,
      "loss": 1.8395,
      "step": 223130
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.517035961151123,
      "learning_rate": 1.4735559599025727e-06,
      "loss": 2.0025,
      "step": 223131
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.5211557149887085,
      "learning_rate": 1.473151051333832e-06,
      "loss": 1.9346,
      "step": 223132
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.427014946937561,
      "learning_rate": 1.4727461982668609e-06,
      "loss": 1.9331,
      "step": 223133
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.4835522174835205,
      "learning_rate": 1.4723414007017597e-06,
      "loss": 1.8503,
      "step": 223134
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.4357073307037354,
      "learning_rate": 1.4719366586385617e-06,
      "loss": 2.0024,
      "step": 223135
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.4749009609222412,
      "learning_rate": 1.4715319720774e-06,
      "loss": 1.7902,
      "step": 223136
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.4772017002105713,
      "learning_rate": 1.4711273410182745e-06,
      "loss": 1.9147,
      "step": 223137
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.4071662425994873,
      "learning_rate": 1.4707227654613185e-06,
      "loss": 1.882,
      "step": 223138
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.5320465564727783,
      "learning_rate": 1.4703182454065654e-06,
      "loss": 2.1019,
      "step": 223139
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.385870337486267,
      "learning_rate": 1.4699137808541484e-06,
      "loss": 2.0527,
      "step": 223140
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.4594544172286987,
      "learning_rate": 1.4695093718041007e-06,
      "loss": 1.8725,
      "step": 223141
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.687147617340088,
      "learning_rate": 1.469105018256489e-06,
      "loss": 1.8554,
      "step": 223142
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.4604276418685913,
      "learning_rate": 1.4687007202114464e-06,
      "loss": 1.9049,
      "step": 223143
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.3890875577926636,
      "learning_rate": 1.468296477668973e-06,
      "loss": 1.7401,
      "step": 223144
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.4083499908447266,
      "learning_rate": 1.467892290629169e-06,
      "loss": 2.0056,
      "step": 223145
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.7605735063552856,
      "learning_rate": 1.4674881590921672e-06,
      "loss": 1.8785,
      "step": 223146
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.5366954803466797,
      "learning_rate": 1.4670840830579678e-06,
      "loss": 2.0242,
      "step": 223147
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.4141541719436646,
      "learning_rate": 1.4666800625266706e-06,
      "loss": 2.1097,
      "step": 223148
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.414149284362793,
      "learning_rate": 1.4662760974983424e-06,
      "loss": 2.035,
      "step": 223149
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.4800333976745605,
      "learning_rate": 1.465872187973083e-06,
      "loss": 1.9007,
      "step": 223150
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.4910880327224731,
      "learning_rate": 1.4654683339509588e-06,
      "loss": 2.0999,
      "step": 223151
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.5389964580535889,
      "learning_rate": 1.4650645354320035e-06,
      "loss": 2.0842,
      "step": 223152
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.3660234212875366,
      "learning_rate": 1.4646607924163834e-06,
      "loss": 2.0222,
      "step": 223153
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.459814429283142,
      "learning_rate": 1.4642571049040653e-06,
      "loss": 1.7911,
      "step": 223154
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.4082998037338257,
      "learning_rate": 1.4638534728952156e-06,
      "loss": 2.0543,
      "step": 223155
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.421994924545288,
      "learning_rate": 1.4634498963898344e-06,
      "loss": 2.1239,
      "step": 223156
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.3656456470489502,
      "learning_rate": 1.463046375388055e-06,
      "loss": 1.9057,
      "step": 223157
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.5011305809020996,
      "learning_rate": 1.462642909889944e-06,
      "loss": 1.7547,
      "step": 223158
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.5115731954574585,
      "learning_rate": 1.4622394998955677e-06,
      "loss": 2.0338,
      "step": 223159
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.4705570936203003,
      "learning_rate": 1.46183614540496e-06,
      "loss": 2.1666,
      "step": 223160
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.3835294246673584,
      "learning_rate": 1.4614328464182535e-06,
      "loss": 2.1229,
      "step": 223161
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.4277262687683105,
      "learning_rate": 1.4610296029354817e-06,
      "loss": 1.9659,
      "step": 223162
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.4597970247268677,
      "learning_rate": 1.460626414956778e-06,
      "loss": 2.0268,
      "step": 223163
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.5035314559936523,
      "learning_rate": 1.4602232824821758e-06,
      "loss": 2.0648,
      "step": 223164
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.3727095127105713,
      "learning_rate": 1.4598202055117414e-06,
      "loss": 2.0595,
      "step": 223165
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.3854283094406128,
      "learning_rate": 1.4594171840455416e-06,
      "loss": 1.7011,
      "step": 223166
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.3944863080978394,
      "learning_rate": 1.459014218083676e-06,
      "loss": 1.8807,
      "step": 223167
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.4458391666412354,
      "learning_rate": 1.4586113076262118e-06,
      "loss": 2.1436,
      "step": 223168
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.5087699890136719,
      "learning_rate": 1.4582084526732485e-06,
      "loss": 2.2705,
      "step": 223169
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.5843408107757568,
      "learning_rate": 1.4578056532248528e-06,
      "loss": 1.9962,
      "step": 223170
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.4847686290740967,
      "learning_rate": 1.457402909281058e-06,
      "loss": 2.1764,
      "step": 223171
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.4795920848846436,
      "learning_rate": 1.457000220841964e-06,
      "loss": 1.9649,
      "step": 223172
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.4803564548492432,
      "learning_rate": 1.4565975879076708e-06,
      "loss": 2.0653,
      "step": 223173
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.6151552200317383,
      "learning_rate": 1.4561950104782115e-06,
      "loss": 1.9603,
      "step": 223174
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.4296737909317017,
      "learning_rate": 1.4557924885536866e-06,
      "loss": 2.0378,
      "step": 223175
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.3779622316360474,
      "learning_rate": 1.455390022134195e-06,
      "loss": 2.1697,
      "step": 223176
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.4418679475784302,
      "learning_rate": 1.4549876112197378e-06,
      "loss": 2.0598,
      "step": 223177
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.5165575742721558,
      "learning_rate": 1.4545852558104477e-06,
      "loss": 1.9705,
      "step": 223178
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.47267484664917,
      "learning_rate": 1.4541829559063912e-06,
      "loss": 1.9646,
      "step": 223179
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.5817160606384277,
      "learning_rate": 1.453780711507635e-06,
      "loss": 2.0146,
      "step": 223180
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.5397005081176758,
      "learning_rate": 1.453378522614279e-06,
      "loss": 2.0202,
      "step": 223181
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.3487213850021362,
      "learning_rate": 1.4529763892263568e-06,
      "loss": 1.952,
      "step": 223182
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.635371446609497,
      "learning_rate": 1.452574311343968e-06,
      "loss": 2.0149,
      "step": 223183
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.5316169261932373,
      "learning_rate": 1.4521722889671461e-06,
      "loss": 1.917,
      "step": 223184
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.468868613243103,
      "learning_rate": 1.4517703220960242e-06,
      "loss": 2.0562,
      "step": 223185
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.465682029724121,
      "learning_rate": 1.451368410730669e-06,
      "loss": 1.9802,
      "step": 223186
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.4776232242584229,
      "learning_rate": 1.4509665548711136e-06,
      "loss": 2.192,
      "step": 223187
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.6931239366531372,
      "learning_rate": 1.4505647545174914e-06,
      "loss": 2.1795,
      "step": 223188
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.3959367275238037,
      "learning_rate": 1.4501630096698358e-06,
      "loss": 1.991,
      "step": 223189
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.4545917510986328,
      "learning_rate": 1.4497613203282131e-06,
      "loss": 1.9726,
      "step": 223190
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.503750205039978,
      "learning_rate": 1.4493596864927236e-06,
      "loss": 2.2065,
      "step": 223191
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.4548146724700928,
      "learning_rate": 1.4489581081634338e-06,
      "loss": 2.0412,
      "step": 223192
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.4735616445541382,
      "learning_rate": 1.4485565853404102e-06,
      "loss": 1.8398,
      "step": 223193
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.5169038772583008,
      "learning_rate": 1.4481551180237527e-06,
      "loss": 1.7942,
      "step": 223194
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.5239530801773071,
      "learning_rate": 1.447753706213528e-06,
      "loss": 2.1507,
      "step": 223195
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.2742538452148438,
      "learning_rate": 1.4473523499097694e-06,
      "loss": 2.1573,
      "step": 223196
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.5168038606643677,
      "learning_rate": 1.4469510491125769e-06,
      "loss": 1.8085,
      "step": 223197
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.5997334718704224,
      "learning_rate": 1.4465498038220503e-06,
      "loss": 1.7723,
      "step": 223198
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.5375152826309204,
      "learning_rate": 1.4461486140382562e-06,
      "loss": 2.0985,
      "step": 223199
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.460146427154541,
      "learning_rate": 1.445747479761261e-06,
      "loss": 2.0947,
      "step": 223200
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.5557199716567993,
      "learning_rate": 1.445346400991132e-06,
      "loss": 1.708,
      "step": 223201
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.50562584400177,
      "learning_rate": 1.4449453777279351e-06,
      "loss": 1.8244,
      "step": 223202
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.5213559865951538,
      "learning_rate": 1.4445444099717708e-06,
      "loss": 1.9856,
      "step": 223203
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.3675204515457153,
      "learning_rate": 1.4441434977226718e-06,
      "loss": 1.8962,
      "step": 223204
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.5675442218780518,
      "learning_rate": 1.4437426409807718e-06,
      "loss": 2.111,
      "step": 223205
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.470779538154602,
      "learning_rate": 1.443341839746104e-06,
      "loss": 1.9508,
      "step": 223206
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.5363749265670776,
      "learning_rate": 1.4429410940188014e-06,
      "loss": 2.105,
      "step": 223207
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.3897299766540527,
      "learning_rate": 1.442540403798831e-06,
      "loss": 2.1858,
      "step": 223208
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.434399962425232,
      "learning_rate": 1.442139769086359e-06,
      "loss": 2.0728,
      "step": 223209
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.4630274772644043,
      "learning_rate": 1.441739189881419e-06,
      "loss": 1.9799,
      "step": 223210
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.437066674232483,
      "learning_rate": 1.4413386661840776e-06,
      "loss": 1.7677,
      "step": 223211
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.3617371320724487,
      "learning_rate": 1.4409381979944345e-06,
      "loss": 1.7632,
      "step": 223212
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.4625869989395142,
      "learning_rate": 1.4405377853125899e-06,
      "loss": 2.0278,
      "step": 223213
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.4408714771270752,
      "learning_rate": 1.440137428138577e-06,
      "loss": 1.9258,
      "step": 223214
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.4753228425979614,
      "learning_rate": 1.4397371264724622e-06,
      "loss": 1.9547,
      "step": 223215
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.5056366920471191,
      "learning_rate": 1.4393368803143124e-06,
      "loss": 1.9204,
      "step": 223216
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.4943156242370605,
      "learning_rate": 1.4389366896642607e-06,
      "loss": 2.0062,
      "step": 223217
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.3583760261535645,
      "learning_rate": 1.4385365545223403e-06,
      "loss": 1.74,
      "step": 223218
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.36569082736969,
      "learning_rate": 1.4381364748886182e-06,
      "loss": 1.9495,
      "step": 223219
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.4125169515609741,
      "learning_rate": 1.437736450763194e-06,
      "loss": 1.8647,
      "step": 223220
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.7279630899429321,
      "learning_rate": 1.4373364821461341e-06,
      "loss": 1.7871,
      "step": 223221
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.5107289552688599,
      "learning_rate": 1.4369365690375057e-06,
      "loss": 1.8948,
      "step": 223222
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.5558007955551147,
      "learning_rate": 1.436536711437375e-06,
      "loss": 2.064,
      "step": 223223
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.5394625663757324,
      "learning_rate": 1.436136909345842e-06,
      "loss": 1.9918,
      "step": 223224
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.3837813138961792,
      "learning_rate": 1.4357371627629732e-06,
      "loss": 1.8768,
      "step": 223225
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.4426318407058716,
      "learning_rate": 1.4353374716888355e-06,
      "loss": 1.8499,
      "step": 223226
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.3347599506378174,
      "learning_rate": 1.434937836123462e-06,
      "loss": 2.078,
      "step": 223227
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.5337717533111572,
      "learning_rate": 1.4345382560670194e-06,
      "loss": 1.9891,
      "step": 223228
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.5993139743804932,
      "learning_rate": 1.4341387315195075e-06,
      "loss": 2.0329,
      "step": 223229
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.4324960708618164,
      "learning_rate": 1.4337392624810262e-06,
      "loss": 1.9485,
      "step": 223230
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.3568695783615112,
      "learning_rate": 1.4333398489516756e-06,
      "loss": 1.8188,
      "step": 223231
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.7340210676193237,
      "learning_rate": 1.432940490931489e-06,
      "loss": 1.9925,
      "step": 223232
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.5024161338806152,
      "learning_rate": 1.432541188420533e-06,
      "loss": 2.0325,
      "step": 223233
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.7927861213684082,
      "learning_rate": 1.4321419414189074e-06,
      "loss": 2.1201,
      "step": 223234
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.612666368484497,
      "learning_rate": 1.431742749926712e-06,
      "loss": 2.096,
      "step": 223235
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.5570189952850342,
      "learning_rate": 1.4313436139439804e-06,
      "loss": 1.9866,
      "step": 223236
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.4954659938812256,
      "learning_rate": 1.4309445334707792e-06,
      "loss": 2.0174,
      "step": 223237
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.7315399646759033,
      "learning_rate": 1.4305455085072082e-06,
      "loss": 2.0315,
      "step": 223238
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.4398126602172852,
      "learning_rate": 1.430146539053334e-06,
      "loss": 1.9384,
      "step": 223239
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.6398075819015503,
      "learning_rate": 1.4297476251092566e-06,
      "loss": 2.1067,
      "step": 223240
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.5477194786071777,
      "learning_rate": 1.4293487666749759e-06,
      "loss": 2.0405,
      "step": 223241
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.4710631370544434,
      "learning_rate": 1.4289499637506585e-06,
      "loss": 2.0363,
      "step": 223242
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.418454647064209,
      "learning_rate": 1.4285512163363378e-06,
      "loss": 2.1483,
      "step": 223243
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.4236910343170166,
      "learning_rate": 1.42815252443208e-06,
      "loss": 2.0394,
      "step": 223244
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.5174086093902588,
      "learning_rate": 1.427753888037919e-06,
      "loss": 1.8183,
      "step": 223245
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.416001796722412,
      "learning_rate": 1.4273553071540211e-06,
      "loss": 1.9942,
      "step": 223246
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.4953160285949707,
      "learning_rate": 1.4269567817803862e-06,
      "loss": 2.0858,
      "step": 223247
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.377483606338501,
      "learning_rate": 1.4265583119171475e-06,
      "loss": 1.9969,
      "step": 223248
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.4311853647232056,
      "learning_rate": 1.4261598975643386e-06,
      "loss": 2.0483,
      "step": 223249
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.4050648212432861,
      "learning_rate": 1.4257615387220255e-06,
      "loss": 1.923,
      "step": 223250
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.5217084884643555,
      "learning_rate": 1.4253632353903088e-06,
      "loss": 1.8976,
      "step": 223251
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.5058903694152832,
      "learning_rate": 1.424964987569255e-06,
      "loss": 2.0672,
      "step": 223252
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.4733368158340454,
      "learning_rate": 1.4245667952589302e-06,
      "loss": 2.0388,
      "step": 223253
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.660132646560669,
      "learning_rate": 1.4241686584594348e-06,
      "loss": 2.2372,
      "step": 223254
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.4589985609054565,
      "learning_rate": 1.423770577170802e-06,
      "loss": 2.0692,
      "step": 223255
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.5909175872802734,
      "learning_rate": 1.4233725513931316e-06,
      "loss": 2.1396,
      "step": 223256
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.6296931505203247,
      "learning_rate": 1.4229745811264904e-06,
      "loss": 1.8381,
      "step": 223257
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.462033748626709,
      "learning_rate": 1.4225766663709782e-06,
      "loss": 2.0534,
      "step": 223258
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.5907264947891235,
      "learning_rate": 1.422178807126595e-06,
      "loss": 2.2375,
      "step": 223259
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.4429994821548462,
      "learning_rate": 1.4217810033935073e-06,
      "loss": 1.7098,
      "step": 223260
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.5741125345230103,
      "learning_rate": 1.4213832551717152e-06,
      "loss": 2.0725,
      "step": 223261
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.504791498184204,
      "learning_rate": 1.420985562461352e-06,
      "loss": 2.0001,
      "step": 223262
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.4356439113616943,
      "learning_rate": 1.4205879252624508e-06,
      "loss": 2.274,
      "step": 223263
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.5884816646575928,
      "learning_rate": 1.4201903435751116e-06,
      "loss": 1.9082,
      "step": 223264
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.5953410863876343,
      "learning_rate": 1.4197928173993678e-06,
      "loss": 2.1583,
      "step": 223265
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.462599754333496,
      "learning_rate": 1.4193953467353525e-06,
      "loss": 2.0696,
      "step": 223266
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.38021981716156,
      "learning_rate": 1.4189979315830658e-06,
      "loss": 1.9555,
      "step": 223267
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.5344457626342773,
      "learning_rate": 1.418600571942674e-06,
      "loss": 2.1202,
      "step": 223268
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.4001481533050537,
      "learning_rate": 1.4182032678141775e-06,
      "loss": 1.8469,
      "step": 223269
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.6517406702041626,
      "learning_rate": 1.4178060191976758e-06,
      "loss": 1.9128,
      "step": 223270
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.4583487510681152,
      "learning_rate": 1.417408826093236e-06,
      "loss": 1.9189,
      "step": 223271
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.4109686613082886,
      "learning_rate": 1.4170116885009243e-06,
      "loss": 1.894,
      "step": 223272
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.5554182529449463,
      "learning_rate": 1.4166146064208407e-06,
      "loss": 1.8698,
      "step": 223273
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.5138031244277954,
      "learning_rate": 1.4162175798530518e-06,
      "loss": 1.9387,
      "step": 223274
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.4919568300247192,
      "learning_rate": 1.4158206087976243e-06,
      "loss": 2.0785,
      "step": 223275
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.6623486280441284,
      "learning_rate": 1.4154236932546248e-06,
      "loss": 1.8406,
      "step": 223276
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.3851174116134644,
      "learning_rate": 1.4150268332241199e-06,
      "loss": 2.0739,
      "step": 223277
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.3523019552230835,
      "learning_rate": 1.4146300287062095e-06,
      "loss": 2.1844,
      "step": 223278
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.3809704780578613,
      "learning_rate": 1.414233279700927e-06,
      "loss": 1.8001,
      "step": 223279
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.5665524005889893,
      "learning_rate": 1.4138365862084388e-06,
      "loss": 2.0256,
      "step": 223280
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.527487874031067,
      "learning_rate": 1.4134399482287118e-06,
      "loss": 1.9763,
      "step": 223281
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.4390718936920166,
      "learning_rate": 1.413043365761879e-06,
      "loss": 1.984,
      "step": 223282
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.4478123188018799,
      "learning_rate": 1.4126468388079736e-06,
      "loss": 1.8361,
      "step": 223283
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.393821358680725,
      "learning_rate": 1.4122503673670959e-06,
      "loss": 1.9184,
      "step": 223284
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.3403472900390625,
      "learning_rate": 1.4118539514393457e-06,
      "loss": 1.9306,
      "step": 223285
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.5847868919372559,
      "learning_rate": 1.4114575910247562e-06,
      "loss": 2.036,
      "step": 223286
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.5376782417297363,
      "learning_rate": 1.411061286123394e-06,
      "loss": 2.1301,
      "step": 223287
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.4986631870269775,
      "learning_rate": 1.4106650367353922e-06,
      "loss": 1.9272,
      "step": 223288
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.3384109735488892,
      "learning_rate": 1.4102688428607512e-06,
      "loss": 1.8929,
      "step": 223289
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.342885971069336,
      "learning_rate": 1.4098727044995706e-06,
      "loss": 1.9343,
      "step": 223290
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.4116047620773315,
      "learning_rate": 1.4094766216519505e-06,
      "loss": 2.0974,
      "step": 223291
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.431393027305603,
      "learning_rate": 1.4090805943179572e-06,
      "loss": 1.8597,
      "step": 223292
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.3224520683288574,
      "learning_rate": 1.4086846224976245e-06,
      "loss": 1.9798,
      "step": 223293
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.4641509056091309,
      "learning_rate": 1.4082887061910852e-06,
      "loss": 1.8169,
      "step": 223294
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.6038531064987183,
      "learning_rate": 1.4078928453983396e-06,
      "loss": 1.8355,
      "step": 223295
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.461714744567871,
      "learning_rate": 1.4074970401195207e-06,
      "loss": 1.9416,
      "step": 223296
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.5147615671157837,
      "learning_rate": 1.4071012903546951e-06,
      "loss": 1.8801,
      "step": 223297
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.3775144815444946,
      "learning_rate": 1.4067055961039297e-06,
      "loss": 1.7782,
      "step": 223298
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.4773597717285156,
      "learning_rate": 1.4063099573672576e-06,
      "loss": 2.2427,
      "step": 223299
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.345859408378601,
      "learning_rate": 1.4059143741448454e-06,
      "loss": 1.8101,
      "step": 223300
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.463714838027954,
      "learning_rate": 1.4055188464366596e-06,
      "loss": 2.0677,
      "step": 223301
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.4789766073226929,
      "learning_rate": 1.4051233742428335e-06,
      "loss": 1.8824,
      "step": 223302
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.574111819267273,
      "learning_rate": 1.404727957563434e-06,
      "loss": 1.8463,
      "step": 223303
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.4987387657165527,
      "learning_rate": 1.4043325963985607e-06,
      "loss": 1.9759,
      "step": 223304
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.4333000183105469,
      "learning_rate": 1.403937290748247e-06,
      "loss": 1.9825,
      "step": 223305
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.315718650817871,
      "learning_rate": 1.4035420406125596e-06,
      "loss": 2.0332,
      "step": 223306
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.3926901817321777,
      "learning_rate": 1.4031468459915983e-06,
      "loss": 1.9561,
      "step": 223307
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.533765196800232,
      "learning_rate": 1.4027517068854299e-06,
      "loss": 2.209,
      "step": 223308
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.5065550804138184,
      "learning_rate": 1.4023566232941208e-06,
      "loss": 1.9264,
      "step": 223309
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.5837359428405762,
      "learning_rate": 1.4019615952177377e-06,
      "loss": 2.005,
      "step": 223310
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.4084752798080444,
      "learning_rate": 1.4015666226564138e-06,
      "loss": 1.8145,
      "step": 223311
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.4080077409744263,
      "learning_rate": 1.4011717056101157e-06,
      "loss": 1.6992,
      "step": 223312
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.3387192487716675,
      "learning_rate": 1.4007768440790102e-06,
      "loss": 1.7945,
      "step": 223313
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.4599640369415283,
      "learning_rate": 1.4003820380631302e-06,
      "loss": 1.9051,
      "step": 223314
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.650386095046997,
      "learning_rate": 1.399987287562576e-06,
      "loss": 2.0465,
      "step": 223315
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.4335002899169922,
      "learning_rate": 1.399592592577381e-06,
      "loss": 1.771,
      "step": 223316
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.4765573740005493,
      "learning_rate": 1.3991979531076447e-06,
      "loss": 1.9214,
      "step": 223317
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.580714225769043,
      "learning_rate": 1.3988033691534007e-06,
      "loss": 1.8238,
      "step": 223318
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.5207451581954956,
      "learning_rate": 1.398408840714782e-06,
      "loss": 1.9233,
      "step": 223319
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.3579658269882202,
      "learning_rate": 1.3980143677918554e-06,
      "loss": 1.7548,
      "step": 223320
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.552600622177124,
      "learning_rate": 1.397619950384654e-06,
      "loss": 2.087,
      "step": 223321
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.4213902950286865,
      "learning_rate": 1.3972255884932781e-06,
      "loss": 1.8751,
      "step": 223322
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.6096906661987305,
      "learning_rate": 1.396831282117794e-06,
      "loss": 1.8764,
      "step": 223323
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.4261553287506104,
      "learning_rate": 1.3964370312582685e-06,
      "loss": 2.1059,
      "step": 223324
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.7276171445846558,
      "learning_rate": 1.396042835914768e-06,
      "loss": 1.924,
      "step": 223325
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.507224440574646,
      "learning_rate": 1.3956486960873925e-06,
      "loss": 1.9851,
      "step": 223326
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.4728963375091553,
      "learning_rate": 1.3952546117762087e-06,
      "loss": 1.9237,
      "step": 223327
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.4305288791656494,
      "learning_rate": 1.3948605829812831e-06,
      "loss": 2.2279,
      "step": 223328
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.3765991926193237,
      "learning_rate": 1.3944666097027157e-06,
      "loss": 1.8592,
      "step": 223329
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.536073923110962,
      "learning_rate": 1.3940726919405065e-06,
      "loss": 1.9082,
      "step": 223330
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.3978320360183716,
      "learning_rate": 1.3936788296947888e-06,
      "loss": 2.0521,
      "step": 223331
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.3737937211990356,
      "learning_rate": 1.393285022965629e-06,
      "loss": 2.0352,
      "step": 223332
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.3281879425048828,
      "learning_rate": 1.3928912717530938e-06,
      "loss": 2.0698,
      "step": 223333
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.4807605743408203,
      "learning_rate": 1.3924975760572833e-06,
      "loss": 1.9865,
      "step": 223334
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.497700810432434,
      "learning_rate": 1.3921039358781972e-06,
      "loss": 2.1252,
      "step": 223335
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.4083855152130127,
      "learning_rate": 1.391710351215969e-06,
      "loss": 2.1293,
      "step": 223336
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.4843682050704956,
      "learning_rate": 1.3913168220706982e-06,
      "loss": 1.9649,
      "step": 223337
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.6470673084259033,
      "learning_rate": 1.390923348442352e-06,
      "loss": 1.969,
      "step": 223338
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.4699658155441284,
      "learning_rate": 1.3905299303311302e-06,
      "loss": 2.0804,
      "step": 223339
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.6066309213638306,
      "learning_rate": 1.3901365677369992e-06,
      "loss": 2.1124,
      "step": 223340
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.3798433542251587,
      "learning_rate": 1.3897432606601256e-06,
      "loss": 1.9469,
      "step": 223341
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.641230583190918,
      "learning_rate": 1.3893500091004761e-06,
      "loss": 2.2196,
      "step": 223342
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.5416878461837769,
      "learning_rate": 1.3889568130582508e-06,
      "loss": 1.9012,
      "step": 223343
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.4871748685836792,
      "learning_rate": 1.3885636725333826e-06,
      "loss": 1.8315,
      "step": 223344
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.445444107055664,
      "learning_rate": 1.3881705875260719e-06,
      "loss": 1.9519,
      "step": 223345
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.404478669166565,
      "learning_rate": 1.3877775580363182e-06,
      "loss": 2.0289,
      "step": 223346
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.4592559337615967,
      "learning_rate": 1.3873845840642216e-06,
      "loss": 2.1029,
      "step": 223347
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.5240025520324707,
      "learning_rate": 1.386991665609849e-06,
      "loss": 2.0467,
      "step": 223348
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.4690327644348145,
      "learning_rate": 1.3865988026732666e-06,
      "loss": 2.0824,
      "step": 223349
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.374895453453064,
      "learning_rate": 1.3862059952545412e-06,
      "loss": 1.8931,
      "step": 223350
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.521793007850647,
      "learning_rate": 1.3858132433537728e-06,
      "loss": 1.9518,
      "step": 223351
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.5226807594299316,
      "learning_rate": 1.3854205469709944e-06,
      "loss": 2.1538,
      "step": 223352
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.3622052669525146,
      "learning_rate": 1.3850279061063396e-06,
      "loss": 1.7396,
      "step": 223353
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.5136339664459229,
      "learning_rate": 1.3846353207598083e-06,
      "loss": 1.8862,
      "step": 223354
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.4207870960235596,
      "learning_rate": 1.3842427909315334e-06,
      "loss": 2.0164,
      "step": 223355
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.43716561794281,
      "learning_rate": 1.3838503166215487e-06,
      "loss": 1.8736,
      "step": 223356
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.3829652070999146,
      "learning_rate": 1.3834578978299538e-06,
      "loss": 2.1275,
      "step": 223357
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.3451402187347412,
      "learning_rate": 1.3830655345567821e-06,
      "loss": 2.0863,
      "step": 223358
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.5215829610824585,
      "learning_rate": 1.3826732268022001e-06,
      "loss": 1.9712,
      "step": 223359
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.4947357177734375,
      "learning_rate": 1.3822809745661413e-06,
      "loss": 2.1242,
      "step": 223360
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.4034920930862427,
      "learning_rate": 1.3818887778488053e-06,
      "loss": 1.874,
      "step": 223361
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.4880915880203247,
      "learning_rate": 1.3814966366501923e-06,
      "loss": 1.9968,
      "step": 223362
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.52948796749115,
      "learning_rate": 1.3811045509704022e-06,
      "loss": 1.9092,
      "step": 223363
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.4179885387420654,
      "learning_rate": 1.3807125208095016e-06,
      "loss": 2.0879,
      "step": 223364
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.5829524993896484,
      "learning_rate": 1.380320546167557e-06,
      "loss": 2.1007,
      "step": 223365
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.5567165613174438,
      "learning_rate": 1.3799286270446686e-06,
      "loss": 2.1302,
      "step": 223366
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.413202166557312,
      "learning_rate": 1.3795367634408694e-06,
      "loss": 1.8573,
      "step": 223367
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.4625041484832764,
      "learning_rate": 1.3791449553562595e-06,
      "loss": 2.0007,
      "step": 223368
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.6946238279342651,
      "learning_rate": 1.378753202790872e-06,
      "loss": 2.0352,
      "step": 223369
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.4165993928909302,
      "learning_rate": 1.3783615057448406e-06,
      "loss": 1.8731,
      "step": 223370
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.565605878829956,
      "learning_rate": 1.3779698642182314e-06,
      "loss": 1.9335,
      "step": 223371
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.6845773458480835,
      "learning_rate": 1.3775782782110445e-06,
      "loss": 2.0327,
      "step": 223372
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.5191353559494019,
      "learning_rate": 1.3771867477234466e-06,
      "loss": 2.0238,
      "step": 223373
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.5589385032653809,
      "learning_rate": 1.3767952727554377e-06,
      "loss": 1.9402,
      "step": 223374
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.6325485706329346,
      "learning_rate": 1.3764038533071176e-06,
      "loss": 2.0259,
      "step": 223375
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.4085278511047363,
      "learning_rate": 1.3760124893785862e-06,
      "loss": 1.972,
      "step": 223376
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.3804200887680054,
      "learning_rate": 1.375621180969877e-06,
      "loss": 1.9453,
      "step": 223377
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.543846607208252,
      "learning_rate": 1.3752299280810898e-06,
      "loss": 1.9363,
      "step": 223378
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.4970284700393677,
      "learning_rate": 1.3748387307122578e-06,
      "loss": 1.8953,
      "step": 223379
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.475156545639038,
      "learning_rate": 1.374447588863481e-06,
      "loss": 1.8468,
      "step": 223380
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.422361969947815,
      "learning_rate": 1.3740565025348592e-06,
      "loss": 1.8887,
      "step": 223381
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.5136860609054565,
      "learning_rate": 1.3736654717264261e-06,
      "loss": 2.0711,
      "step": 223382
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.5291451215744019,
      "learning_rate": 1.3732744964382481e-06,
      "loss": 1.9848,
      "step": 223383
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.4835695028305054,
      "learning_rate": 1.3728835766704248e-06,
      "loss": 1.9104,
      "step": 223384
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.5283119678497314,
      "learning_rate": 1.3724927124230233e-06,
      "loss": 1.8715,
      "step": 223385
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.5540707111358643,
      "learning_rate": 1.37210190369611e-06,
      "loss": 2.0059,
      "step": 223386
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.4663199186325073,
      "learning_rate": 1.3717111504897849e-06,
      "loss": 1.888,
      "step": 223387
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.406897783279419,
      "learning_rate": 1.3713204528040477e-06,
      "loss": 1.8591,
      "step": 223388
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.4987318515777588,
      "learning_rate": 1.3709298106390654e-06,
      "loss": 2.0402,
      "step": 223389
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.49063241481781,
      "learning_rate": 1.3705392239948376e-06,
      "loss": 1.9852,
      "step": 223390
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.5385332107543945,
      "learning_rate": 1.3701486928714644e-06,
      "loss": 1.8897,
      "step": 223391
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.5737727880477905,
      "learning_rate": 1.3697582172690458e-06,
      "loss": 2.0652,
      "step": 223392
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.4234248399734497,
      "learning_rate": 1.3693677971875815e-06,
      "loss": 1.7524,
      "step": 223393
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.3456019163131714,
      "learning_rate": 1.3689774326272053e-06,
      "loss": 1.9822,
      "step": 223394
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.4530606269836426,
      "learning_rate": 1.368587123587983e-06,
      "loss": 1.963,
      "step": 223395
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.526123285293579,
      "learning_rate": 1.3681968700699818e-06,
      "loss": 1.7646,
      "step": 223396
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.47400963306427,
      "learning_rate": 1.3678066720732683e-06,
      "loss": 1.8749,
      "step": 223397
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.5938830375671387,
      "learning_rate": 1.3674165295979089e-06,
      "loss": 2.0092,
      "step": 223398
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.3142833709716797,
      "learning_rate": 1.3670264426439703e-06,
      "loss": 1.9813,
      "step": 223399
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.3728317022323608,
      "learning_rate": 1.3666364112115524e-06,
      "loss": 2.0826,
      "step": 223400
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.431979775428772,
      "learning_rate": 1.3662464353007218e-06,
      "loss": 1.8997,
      "step": 223401
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.4444528818130493,
      "learning_rate": 1.3658565149115452e-06,
      "loss": 1.989,
      "step": 223402
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.4200607538223267,
      "learning_rate": 1.3654666500440892e-06,
      "loss": 1.972,
      "step": 223403
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.4744218587875366,
      "learning_rate": 1.3650768406984203e-06,
      "loss": 1.949,
      "step": 223404
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.4325718879699707,
      "learning_rate": 1.3646870868746384e-06,
      "loss": 2.1134,
      "step": 223405
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.3211674690246582,
      "learning_rate": 1.364297388572777e-06,
      "loss": 1.7974,
      "step": 223406
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.4795303344726562,
      "learning_rate": 1.363907745792936e-06,
      "loss": 1.899,
      "step": 223407
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.4580678939819336,
      "learning_rate": 1.3635181585351818e-06,
      "loss": 2.2976,
      "step": 223408
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.5464868545532227,
      "learning_rate": 1.363128626799581e-06,
      "loss": 2.1192,
      "step": 223409
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.5213212966918945,
      "learning_rate": 1.3627391505862339e-06,
      "loss": 2.0114,
      "step": 223410
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.5183489322662354,
      "learning_rate": 1.3623497298951403e-06,
      "loss": 2.2476,
      "step": 223411
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.3516582250595093,
      "learning_rate": 1.3619603647264665e-06,
      "loss": 1.8235,
      "step": 223412
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.4500620365142822,
      "learning_rate": 1.3615710550802462e-06,
      "loss": 1.9606,
      "step": 223413
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.564612865447998,
      "learning_rate": 1.3611818009565457e-06,
      "loss": 2.05,
      "step": 223414
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.7833443880081177,
      "learning_rate": 1.3607926023554316e-06,
      "loss": 1.9452,
      "step": 223415
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.3987935781478882,
      "learning_rate": 1.3604034592769708e-06,
      "loss": 1.7104,
      "step": 223416
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.3497225046157837,
      "learning_rate": 1.360014371721263e-06,
      "loss": 2.0015,
      "step": 223417
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.58428955078125,
      "learning_rate": 1.3596253396883417e-06,
      "loss": 1.8994,
      "step": 223418
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.4776335954666138,
      "learning_rate": 1.3592363631783397e-06,
      "loss": 2.2019,
      "step": 223419
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.6415687799453735,
      "learning_rate": 1.358847442191291e-06,
      "loss": 2.1696,
      "step": 223420
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.6211812496185303,
      "learning_rate": 1.3584585767272283e-06,
      "loss": 2.0546,
      "step": 223421
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.5503292083740234,
      "learning_rate": 1.358069766786285e-06,
      "loss": 2.1245,
      "step": 223422
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.4688094854354858,
      "learning_rate": 1.3576810123685278e-06,
      "loss": 2.0303,
      "step": 223423
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.3861922025680542,
      "learning_rate": 1.3572923134740232e-06,
      "loss": 2.0804,
      "step": 223424
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.4210115671157837,
      "learning_rate": 1.3569036701028047e-06,
      "loss": 1.8486,
      "step": 223425
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.3608994483947754,
      "learning_rate": 1.3565150822550053e-06,
      "loss": 2.058,
      "step": 223426
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.630527138710022,
      "learning_rate": 1.356126549930625e-06,
      "loss": 2.0834,
      "step": 223427
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.5744903087615967,
      "learning_rate": 1.3557380731298306e-06,
      "loss": 1.9765,
      "step": 223428
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.4480215311050415,
      "learning_rate": 1.3553496518525886e-06,
      "loss": 1.9978,
      "step": 223429
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.3571704626083374,
      "learning_rate": 1.3549612860990655e-06,
      "loss": 1.7025,
      "step": 223430
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.4402930736541748,
      "learning_rate": 1.3545729758692615e-06,
      "loss": 2.0341,
      "step": 223431
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.4683483839035034,
      "learning_rate": 1.3541847211633095e-06,
      "loss": 2.1402,
      "step": 223432
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.6068791151046753,
      "learning_rate": 1.353796521981243e-06,
      "loss": 2.1042,
      "step": 223433
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.5501655340194702,
      "learning_rate": 1.3534083783231286e-06,
      "loss": 1.8489,
      "step": 223434
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.3625528812408447,
      "learning_rate": 1.3530202901890664e-06,
      "loss": 1.8773,
      "step": 223435
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.5212596654891968,
      "learning_rate": 1.3526322575791226e-06,
      "loss": 1.7565,
      "step": 223436
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.527702808380127,
      "learning_rate": 1.3522442804933309e-06,
      "loss": 1.9878,
      "step": 223437
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.5983737707138062,
      "learning_rate": 1.3518563589318576e-06,
      "loss": 2.0815,
      "step": 223438
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.4504690170288086,
      "learning_rate": 1.3514684928946361e-06,
      "loss": 2.1231,
      "step": 223439
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.441112756729126,
      "learning_rate": 1.3510806823818665e-06,
      "loss": 1.9428,
      "step": 223440
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.5274121761322021,
      "learning_rate": 1.3506929273935486e-06,
      "loss": 1.8506,
      "step": 223441
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.5072336196899414,
      "learning_rate": 1.3503052279297822e-06,
      "loss": 2.1983,
      "step": 223442
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.504454493522644,
      "learning_rate": 1.3499175839906007e-06,
      "loss": 1.9788,
      "step": 223443
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.5130970478057861,
      "learning_rate": 1.3495299955761374e-06,
      "loss": 1.9723,
      "step": 223444
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.549914836883545,
      "learning_rate": 1.3491424626864256e-06,
      "loss": 1.9833,
      "step": 223445
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.392927885055542,
      "learning_rate": 1.348754985321565e-06,
      "loss": 1.789,
      "step": 223446
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.410714864730835,
      "learning_rate": 1.348367563481556e-06,
      "loss": 2.0715,
      "step": 223447
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.4155068397521973,
      "learning_rate": 1.347980197166565e-06,
      "loss": 2.047,
      "step": 223448
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.349344253540039,
      "learning_rate": 1.347592886376625e-06,
      "loss": 1.9332,
      "step": 223449
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.4674125909805298,
      "learning_rate": 1.3472056311118028e-06,
      "loss": 1.7576,
      "step": 223450
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.5737346410751343,
      "learning_rate": 1.3468184313721653e-06,
      "loss": 2.1007,
      "step": 223451
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.3603672981262207,
      "learning_rate": 1.3464312871577788e-06,
      "loss": 1.9913,
      "step": 223452
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.4655787944793701,
      "learning_rate": 1.346044198468743e-06,
      "loss": 2.1432,
      "step": 223453
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.4689476490020752,
      "learning_rate": 1.3456571653051252e-06,
      "loss": 2.1526,
      "step": 223454
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.354934573173523,
      "learning_rate": 1.345270187666958e-06,
      "loss": 2.0413,
      "step": 223455
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.5948610305786133,
      "learning_rate": 1.344883265554375e-06,
      "loss": 1.8057,
      "step": 223456
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.5472209453582764,
      "learning_rate": 1.3444963989673763e-06,
      "loss": 1.967,
      "step": 223457
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.4861762523651123,
      "learning_rate": 1.344109587906128e-06,
      "loss": 1.9319,
      "step": 223458
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.4348398447036743,
      "learning_rate": 1.3437228323705972e-06,
      "loss": 1.7884,
      "step": 223459
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.417773723602295,
      "learning_rate": 1.343336132360917e-06,
      "loss": 1.9011,
      "step": 223460
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.4103301763534546,
      "learning_rate": 1.342949487877154e-06,
      "loss": 2.1437,
      "step": 223461
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.4875390529632568,
      "learning_rate": 1.3425628989194083e-06,
      "loss": 2.053,
      "step": 223462
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.4371755123138428,
      "learning_rate": 1.3421763654876794e-06,
      "loss": 2.1268,
      "step": 223463
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.5286990404129028,
      "learning_rate": 1.3417898875820676e-06,
      "loss": 2.0347,
      "step": 223464
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.5982530117034912,
      "learning_rate": 1.341403465202673e-06,
      "loss": 1.9397,
      "step": 223465
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.4381917715072632,
      "learning_rate": 1.3410170983495615e-06,
      "loss": 1.8603,
      "step": 223466
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.543782114982605,
      "learning_rate": 1.3406307870227673e-06,
      "loss": 1.8444,
      "step": 223467
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.4932020902633667,
      "learning_rate": 1.3402445312223897e-06,
      "loss": 1.9804,
      "step": 223468
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.493151068687439,
      "learning_rate": 1.3398583309484955e-06,
      "loss": 2.2012,
      "step": 223469
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.3975989818572998,
      "learning_rate": 1.3394721862011849e-06,
      "loss": 1.8103,
      "step": 223470
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.5403978824615479,
      "learning_rate": 1.3390860969804906e-06,
      "loss": 2.0686,
      "step": 223471
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.3586488962173462,
      "learning_rate": 1.338700063286513e-06,
      "loss": 1.8366,
      "step": 223472
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.5622754096984863,
      "learning_rate": 1.338314085119252e-06,
      "loss": 2.0945,
      "step": 223473
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.4679888486862183,
      "learning_rate": 1.3379281624789074e-06,
      "loss": 1.8971,
      "step": 223474
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.5222183465957642,
      "learning_rate": 1.3375422953654457e-06,
      "loss": 1.9145,
      "step": 223475
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.4738585948944092,
      "learning_rate": 1.3371564837789672e-06,
      "loss": 1.9611,
      "step": 223476
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.5415112972259521,
      "learning_rate": 1.3367707277195717e-06,
      "loss": 1.9536,
      "step": 223477
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.5189584493637085,
      "learning_rate": 1.3363850271872921e-06,
      "loss": 1.8119,
      "step": 223478
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.5614304542541504,
      "learning_rate": 1.3359993821821958e-06,
      "loss": 1.9161,
      "step": 223479
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.349058747291565,
      "learning_rate": 1.3356137927044153e-06,
      "loss": 2.0375,
      "step": 223480
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.5324933528900146,
      "learning_rate": 1.335228258753951e-06,
      "loss": 1.7549,
      "step": 223481
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.4820729494094849,
      "learning_rate": 1.3348427803309359e-06,
      "loss": 2.0076,
      "step": 223482
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.4077922105789185,
      "learning_rate": 1.3344573574353702e-06,
      "loss": 2.0972,
      "step": 223483
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.675440788269043,
      "learning_rate": 1.3340719900674201e-06,
      "loss": 1.9995,
      "step": 223484
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.5071587562561035,
      "learning_rate": 1.3336866782270527e-06,
      "loss": 2.2252,
      "step": 223485
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.3268405199050903,
      "learning_rate": 1.3333014219144345e-06,
      "loss": 2.0968,
      "step": 223486
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.5455130338668823,
      "learning_rate": 1.3329162211295652e-06,
      "loss": 2.0216,
      "step": 223487
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.443708062171936,
      "learning_rate": 1.3325310758725449e-06,
      "loss": 2.0672,
      "step": 223488
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.39176344871521,
      "learning_rate": 1.3321459861434402e-06,
      "loss": 2.0033,
      "step": 223489
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.5730971097946167,
      "learning_rate": 1.331760951942351e-06,
      "loss": 1.9821,
      "step": 223490
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.4346325397491455,
      "learning_rate": 1.3313759732693108e-06,
      "loss": 1.8465,
      "step": 223491
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.4966284036636353,
      "learning_rate": 1.330991050124386e-06,
      "loss": 2.0241,
      "step": 223492
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.3851149082183838,
      "learning_rate": 1.3306061825077096e-06,
      "loss": 1.7926,
      "step": 223493
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.5809415578842163,
      "learning_rate": 1.3302213704192822e-06,
      "loss": 1.8696,
      "step": 223494
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.6090210676193237,
      "learning_rate": 1.3298366138592032e-06,
      "loss": 1.863,
      "step": 223495
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.4815917015075684,
      "learning_rate": 1.3294519128275728e-06,
      "loss": 2.2573,
      "step": 223496
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.4528625011444092,
      "learning_rate": 1.3290672673243908e-06,
      "loss": 1.9569,
      "step": 223497
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.4476879835128784,
      "learning_rate": 1.3286826773498238e-06,
      "loss": 1.982,
      "step": 223498
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.4940147399902344,
      "learning_rate": 1.328298142903872e-06,
      "loss": 1.9652,
      "step": 223499
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.4596771001815796,
      "learning_rate": 1.3279136639866017e-06,
      "loss": 1.9306,
      "step": 223500
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.4386143684387207,
      "learning_rate": 1.3275292405981463e-06,
      "loss": 1.9208,
      "step": 223501
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.353515863418579,
      "learning_rate": 1.327144872738506e-06,
      "loss": 2.0194,
      "step": 223502
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.5604431629180908,
      "learning_rate": 1.3267605604078136e-06,
      "loss": 1.8825,
      "step": 223503
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.3069204092025757,
      "learning_rate": 1.3263763036061025e-06,
      "loss": 2.0318,
      "step": 223504
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.436111330986023,
      "learning_rate": 1.3259921023334729e-06,
      "loss": 1.7665,
      "step": 223505
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.4957115650177002,
      "learning_rate": 1.325607956589958e-06,
      "loss": 1.8295,
      "step": 223506
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.4873061180114746,
      "learning_rate": 1.3252238663756576e-06,
      "loss": 1.8344,
      "step": 223507
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.5207419395446777,
      "learning_rate": 1.3248398316906384e-06,
      "loss": 1.932,
      "step": 223508
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.4634711742401123,
      "learning_rate": 1.3244558525349669e-06,
      "loss": 1.8988,
      "step": 223509
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.483150601387024,
      "learning_rate": 1.324071928908743e-06,
      "loss": 2.0256,
      "step": 223510
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.4241924285888672,
      "learning_rate": 1.3236880608120003e-06,
      "loss": 1.9667,
      "step": 223511
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.4589464664459229,
      "learning_rate": 1.3233042482448053e-06,
      "loss": 2.1334,
      "step": 223512
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.4045889377593994,
      "learning_rate": 1.3229204912072577e-06,
      "loss": 1.911,
      "step": 223513
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.4707683324813843,
      "learning_rate": 1.322536789699391e-06,
      "loss": 2.2101,
      "step": 223514
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.498935341835022,
      "learning_rate": 1.3221531437213383e-06,
      "loss": 2.1554,
      "step": 223515
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.5905978679656982,
      "learning_rate": 1.3217695532731e-06,
      "loss": 1.8161,
      "step": 223516
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.4018783569335938,
      "learning_rate": 1.3213860183548086e-06,
      "loss": 1.9387,
      "step": 223517
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.3552900552749634,
      "learning_rate": 1.321002538966498e-06,
      "loss": 1.9921,
      "step": 223518
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.4315025806427002,
      "learning_rate": 1.3206191151082679e-06,
      "loss": 2.1068,
      "step": 223519
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.3468356132507324,
      "learning_rate": 1.3202357467801516e-06,
      "loss": 1.8322,
      "step": 223520
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.4449464082717896,
      "learning_rate": 1.3198524339822493e-06,
      "loss": 1.897,
      "step": 223521
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.3557580709457397,
      "learning_rate": 1.3194691767146271e-06,
      "loss": 1.9692,
      "step": 223522
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.3959739208221436,
      "learning_rate": 1.3190859749773519e-06,
      "loss": 2.0507,
      "step": 223523
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.4156289100646973,
      "learning_rate": 1.3187028287704905e-06,
      "loss": 2.0839,
      "step": 223524
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.5605978965759277,
      "learning_rate": 1.3183197380941424e-06,
      "loss": 1.8072,
      "step": 223525
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.3936747312545776,
      "learning_rate": 1.3179367029483411e-06,
      "loss": 1.7694,
      "step": 223526
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.3871687650680542,
      "learning_rate": 1.3175537233331868e-06,
      "loss": 1.874,
      "step": 223527
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.4523377418518066,
      "learning_rate": 1.3171707992487123e-06,
      "loss": 1.9864,
      "step": 223528
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.4574788808822632,
      "learning_rate": 1.316787930695018e-06,
      "loss": 2.0852,
      "step": 223529
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.590351939201355,
      "learning_rate": 1.3164051176721702e-06,
      "loss": 2.0939,
      "step": 223530
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.5994571447372437,
      "learning_rate": 1.316022360180269e-06,
      "loss": 1.9586,
      "step": 223531
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.3472849130630493,
      "learning_rate": 1.315639658219314e-06,
      "loss": 1.7665,
      "step": 223532
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.5220967531204224,
      "learning_rate": 1.3152570117894723e-06,
      "loss": 2.0323,
      "step": 223533
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.515657663345337,
      "learning_rate": 1.3148744208907103e-06,
      "loss": 1.9869,
      "step": 223534
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.3794035911560059,
      "learning_rate": 1.3144918855231944e-06,
      "loss": 1.8939,
      "step": 223535
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.4602922201156616,
      "learning_rate": 1.314109405686925e-06,
      "loss": 1.8544,
      "step": 223536
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.5178134441375732,
      "learning_rate": 1.3137269813820016e-06,
      "loss": 1.9212,
      "step": 223537
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.5134340524673462,
      "learning_rate": 1.3133446126085246e-06,
      "loss": 1.992,
      "step": 223538
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.352219820022583,
      "learning_rate": 1.3129622993664935e-06,
      "loss": 1.7932,
      "step": 223539
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.4358277320861816,
      "learning_rate": 1.3125800416560417e-06,
      "loss": 1.9293,
      "step": 223540
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.5073639154434204,
      "learning_rate": 1.3121978394772025e-06,
      "loss": 2.0414,
      "step": 223541
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.6138088703155518,
      "learning_rate": 1.3118156928301093e-06,
      "loss": 1.9419,
      "step": 223542
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.4852203130722046,
      "learning_rate": 1.311433601714762e-06,
      "loss": 1.9446,
      "step": 223543
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.4859893321990967,
      "learning_rate": 1.311051566131227e-06,
      "loss": 1.7765,
      "step": 223544
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.4248729944229126,
      "learning_rate": 1.3106695860796378e-06,
      "loss": 1.9962,
      "step": 223545
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.4207968711853027,
      "learning_rate": 1.3102876615600277e-06,
      "loss": 1.8324,
      "step": 223546
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.578497052192688,
      "learning_rate": 1.3099057925724632e-06,
      "loss": 2.033,
      "step": 223547
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.4464422464370728,
      "learning_rate": 1.3095239791170443e-06,
      "loss": 1.9318,
      "step": 223548
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.4683765172958374,
      "learning_rate": 1.3091422211938041e-06,
      "loss": 2.0877,
      "step": 223549
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.4470305442810059,
      "learning_rate": 1.3087605188028428e-06,
      "loss": 1.9948,
      "step": 223550
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.535900354385376,
      "learning_rate": 1.3083788719442267e-06,
      "loss": 1.7201,
      "step": 223551
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.3910598754882812,
      "learning_rate": 1.3079972806180229e-06,
      "loss": 2.0101,
      "step": 223552
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.5502994060516357,
      "learning_rate": 1.3076157448242642e-06,
      "loss": 1.9023,
      "step": 223553
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.3700989484786987,
      "learning_rate": 1.3072342645631174e-06,
      "loss": 1.9296,
      "step": 223554
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.4105546474456787,
      "learning_rate": 1.306852839834549e-06,
      "loss": 2.0809,
      "step": 223555
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.5226001739501953,
      "learning_rate": 1.3064714706386924e-06,
      "loss": 1.8984,
      "step": 223556
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.4492131471633911,
      "learning_rate": 1.3060901569755812e-06,
      "loss": 2.0055,
      "step": 223557
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.4704805612564087,
      "learning_rate": 1.3057088988453147e-06,
      "loss": 2.0123,
      "step": 223558
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.5694489479064941,
      "learning_rate": 1.3053276962479932e-06,
      "loss": 2.0178,
      "step": 223559
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.5452810525894165,
      "learning_rate": 1.3049465491836164e-06,
      "loss": 2.1457,
      "step": 223560
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.4453110694885254,
      "learning_rate": 1.3045654576522845e-06,
      "loss": 2.0944,
      "step": 223561
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.4886572360992432,
      "learning_rate": 1.3041844216540643e-06,
      "loss": 2.0644,
      "step": 223562
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.4987035989761353,
      "learning_rate": 1.3038034411890552e-06,
      "loss": 1.9424,
      "step": 223563
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.5011671781539917,
      "learning_rate": 1.303422516257291e-06,
      "loss": 2.1635,
      "step": 223564
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.4118351936340332,
      "learning_rate": 1.3030416468588712e-06,
      "loss": 1.8258,
      "step": 223565
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.3576871156692505,
      "learning_rate": 1.3026608329938626e-06,
      "loss": 1.9562,
      "step": 223566
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.3275046348571777,
      "learning_rate": 1.3022800746623318e-06,
      "loss": 1.8926,
      "step": 223567
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.561066746711731,
      "learning_rate": 1.3018993718643123e-06,
      "loss": 2.0117,
      "step": 223568
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.7227729558944702,
      "learning_rate": 1.301518724599937e-06,
      "loss": 2.0251,
      "step": 223569
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.425013780593872,
      "learning_rate": 1.3011381328692394e-06,
      "loss": 2.0001,
      "step": 223570
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.4865541458129883,
      "learning_rate": 1.300757596672286e-06,
      "loss": 2.0407,
      "step": 223571
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.5871293544769287,
      "learning_rate": 1.3003771160091768e-06,
      "loss": 2.1034,
      "step": 223572
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.4142987728118896,
      "learning_rate": 1.2999966908799454e-06,
      "loss": 2.0663,
      "step": 223573
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.4868111610412598,
      "learning_rate": 1.2996163212847244e-06,
      "loss": 2.0137,
      "step": 223574
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.4766862392425537,
      "learning_rate": 1.2992360072235143e-06,
      "loss": 2.0139,
      "step": 223575
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.4775590896606445,
      "learning_rate": 1.2988557486964146e-06,
      "loss": 1.9632,
      "step": 223576
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.4634244441986084,
      "learning_rate": 1.2984755457034923e-06,
      "loss": 2.088,
      "step": 223577
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.4629484415054321,
      "learning_rate": 1.2980953982448138e-06,
      "loss": 1.9271,
      "step": 223578
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.5950227975845337,
      "learning_rate": 1.2977153063204793e-06,
      "loss": 1.9411,
      "step": 223579
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.402343988418579,
      "learning_rate": 1.2973352699305218e-06,
      "loss": 1.9167,
      "step": 223580
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.5188730955123901,
      "learning_rate": 1.2969552890750411e-06,
      "loss": 1.968,
      "step": 223581
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.3711735010147095,
      "learning_rate": 1.2965753637541042e-06,
      "loss": 1.9682,
      "step": 223582
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.4821972846984863,
      "learning_rate": 1.2961954939677443e-06,
      "loss": 1.965,
      "step": 223583
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.4885271787643433,
      "learning_rate": 1.2958156797160945e-06,
      "loss": 2.1546,
      "step": 223584
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.453208327293396,
      "learning_rate": 1.295435920999155e-06,
      "loss": 1.8353,
      "step": 223585
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.6123040914535522,
      "learning_rate": 1.2950562178170588e-06,
      "loss": 2.1672,
      "step": 223586
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.389804720878601,
      "learning_rate": 1.294676570169806e-06,
      "loss": 1.9895,
      "step": 223587
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.5606549978256226,
      "learning_rate": 1.294296978057563e-06,
      "loss": 2.0233,
      "step": 223588
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.723008394241333,
      "learning_rate": 1.29391744148033e-06,
      "loss": 1.8915,
      "step": 223589
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.6753548383712769,
      "learning_rate": 1.2935379604382068e-06,
      "loss": 2.0019,
      "step": 223590
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.5123275518417358,
      "learning_rate": 1.2931585349312269e-06,
      "loss": 1.8873,
      "step": 223591
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.5544713735580444,
      "learning_rate": 1.292779164959523e-06,
      "loss": 2.0681,
      "step": 223592
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.476356863975525,
      "learning_rate": 1.292399850523096e-06,
      "loss": 2.0642,
      "step": 223593
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.7256656885147095,
      "learning_rate": 1.2920205916220783e-06,
      "loss": 2.1193,
      "step": 223594
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.411147117614746,
      "learning_rate": 1.2916413882565034e-06,
      "loss": 1.9315,
      "step": 223595
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.450934648513794,
      "learning_rate": 1.2912622404264384e-06,
      "loss": 1.8591,
      "step": 223596
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.5115909576416016,
      "learning_rate": 1.2908831481319826e-06,
      "loss": 1.623,
      "step": 223597
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.359897494316101,
      "learning_rate": 1.2905041113731696e-06,
      "loss": 1.8658,
      "step": 223598
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.5326287746429443,
      "learning_rate": 1.2901251301500992e-06,
      "loss": 2.0822,
      "step": 223599
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.4120129346847534,
      "learning_rate": 1.289746204462838e-06,
      "loss": 1.9303,
      "step": 223600
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.5204811096191406,
      "learning_rate": 1.289367334311453e-06,
      "loss": 1.8295,
      "step": 223601
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.4713964462280273,
      "learning_rate": 1.2889885196960104e-06,
      "loss": 1.9856,
      "step": 223602
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.3709999322891235,
      "learning_rate": 1.2886097606165767e-06,
      "loss": 1.7138,
      "step": 223603
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.3790225982666016,
      "learning_rate": 1.2882310570732524e-06,
      "loss": 1.9864,
      "step": 223604
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.5225881338119507,
      "learning_rate": 1.2878524090660703e-06,
      "loss": 1.8619,
      "step": 223605
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.4516003131866455,
      "learning_rate": 1.287473816595097e-06,
      "loss": 2.0171,
      "step": 223606
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.393114447593689,
      "learning_rate": 1.2870952796604327e-06,
      "loss": 1.88,
      "step": 223607
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.507922649383545,
      "learning_rate": 1.286716798262144e-06,
      "loss": 1.9329,
      "step": 223608
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.3949846029281616,
      "learning_rate": 1.2863383724002973e-06,
      "loss": 1.9823,
      "step": 223609
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.382961630821228,
      "learning_rate": 1.2859600020749595e-06,
      "loss": 2.1473,
      "step": 223610
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.351607322692871,
      "learning_rate": 1.2855816872861967e-06,
      "loss": 2.0469,
      "step": 223611
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.4144203662872314,
      "learning_rate": 1.285203428034076e-06,
      "loss": 2.1258,
      "step": 223612
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.3555723428726196,
      "learning_rate": 1.2848252243186642e-06,
      "loss": 1.7739,
      "step": 223613
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.6577188968658447,
      "learning_rate": 1.2844470761400605e-06,
      "loss": 1.7689,
      "step": 223614
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.4876418113708496,
      "learning_rate": 1.2840689834982986e-06,
      "loss": 2.0187,
      "step": 223615
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.48054838180542,
      "learning_rate": 1.2836909463935119e-06,
      "loss": 2.1034,
      "step": 223616
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.4295027256011963,
      "learning_rate": 1.2833129648256668e-06,
      "loss": 1.9183,
      "step": 223617
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.539685606956482,
      "learning_rate": 1.28293503879493e-06,
      "loss": 2.0186,
      "step": 223618
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.4716845750808716,
      "learning_rate": 1.2825571683013013e-06,
      "loss": 2.0585,
      "step": 223619
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.5213725566864014,
      "learning_rate": 1.2821793533449143e-06,
      "loss": 2.0555,
      "step": 223620
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.5719523429870605,
      "learning_rate": 1.281801593925802e-06,
      "loss": 1.9385,
      "step": 223621
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.5313148498535156,
      "learning_rate": 1.2814238900440643e-06,
      "loss": 1.8982,
      "step": 223622
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.493414282798767,
      "learning_rate": 1.2810462416997013e-06,
      "loss": 1.8599,
      "step": 223623
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.416947364807129,
      "learning_rate": 1.2806686488928797e-06,
      "loss": 2.0738,
      "step": 223624
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.5838807821273804,
      "learning_rate": 1.2802911116235658e-06,
      "loss": 1.9898,
      "step": 223625
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.3955714702606201,
      "learning_rate": 1.2799136298919266e-06,
      "loss": 1.947,
      "step": 223626
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.421338438987732,
      "learning_rate": 1.279536203697995e-06,
      "loss": 1.819,
      "step": 223627
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.5117323398590088,
      "learning_rate": 1.2791588330418045e-06,
      "loss": 2.0093,
      "step": 223628
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.4512027502059937,
      "learning_rate": 1.2787815179234551e-06,
      "loss": 1.8878,
      "step": 223629
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.3820240497589111,
      "learning_rate": 1.2784042583430465e-06,
      "loss": 1.7819,
      "step": 223630
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.4234211444854736,
      "learning_rate": 1.2780270543006122e-06,
      "loss": 1.8775,
      "step": 223631
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.3604203462600708,
      "learning_rate": 1.2776499057962185e-06,
      "loss": 2.0616,
      "step": 223632
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.5701426267623901,
      "learning_rate": 1.2772728128299657e-06,
      "loss": 1.972,
      "step": 223633
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.5522844791412354,
      "learning_rate": 1.2768957754019206e-06,
      "loss": 2.1407,
      "step": 223634
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.4215561151504517,
      "learning_rate": 1.2765187935121158e-06,
      "loss": 2.0273,
      "step": 223635
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.5451431274414062,
      "learning_rate": 1.2761418671606515e-06,
      "loss": 1.8785,
      "step": 223636
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.407899022102356,
      "learning_rate": 1.2757649963475613e-06,
      "loss": 1.9086,
      "step": 223637
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.5260965824127197,
      "learning_rate": 1.2753881810729781e-06,
      "loss": 2.1623,
      "step": 223638
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.4960709810256958,
      "learning_rate": 1.2750114213369356e-06,
      "loss": 1.9337,
      "step": 223639
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.505537748336792,
      "learning_rate": 1.2746347171395333e-06,
      "loss": 2.0813,
      "step": 223640
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.4804221391677856,
      "learning_rate": 1.2742580684807712e-06,
      "loss": 2.155,
      "step": 223641
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.4087390899658203,
      "learning_rate": 1.273881475360783e-06,
      "loss": 1.9313,
      "step": 223642
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.3077598810195923,
      "learning_rate": 1.2735049377796015e-06,
      "loss": 1.9091,
      "step": 223643
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.542995810508728,
      "learning_rate": 1.2731284557373266e-06,
      "loss": 2.0963,
      "step": 223644
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.679466724395752,
      "learning_rate": 1.2727520292340253e-06,
      "loss": 2.3104,
      "step": 223645
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.4414353370666504,
      "learning_rate": 1.2723756582697642e-06,
      "loss": 1.8808,
      "step": 223646
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.4809767007827759,
      "learning_rate": 1.2719993428445763e-06,
      "loss": 1.9081,
      "step": 223647
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.7041404247283936,
      "learning_rate": 1.271623082958595e-06,
      "loss": 2.0945,
      "step": 223648
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.4672939777374268,
      "learning_rate": 1.2712468786118202e-06,
      "loss": 1.8435,
      "step": 223649
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.4696786403656006,
      "learning_rate": 1.2708707298043853e-06,
      "loss": 2.0036,
      "step": 223650
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.3457157611846924,
      "learning_rate": 1.2704946365363567e-06,
      "loss": 1.9804,
      "step": 223651
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.3005551099777222,
      "learning_rate": 1.2701185988077677e-06,
      "loss": 1.8152,
      "step": 223652
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.4180816411972046,
      "learning_rate": 1.269742616618652e-06,
      "loss": 2.1533,
      "step": 223653
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.5148307085037231,
      "learning_rate": 1.2693666899691756e-06,
      "loss": 2.1167,
      "step": 223654
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.486992597579956,
      "learning_rate": 1.2689908188593723e-06,
      "loss": 2.0623,
      "step": 223655
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.4980111122131348,
      "learning_rate": 1.268615003289275e-06,
      "loss": 2.0654,
      "step": 223656
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.5381730794906616,
      "learning_rate": 1.268239243259017e-06,
      "loss": 1.9198,
      "step": 223657
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.3436912298202515,
      "learning_rate": 1.2678635387685986e-06,
      "loss": 1.9553,
      "step": 223658
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.546754240989685,
      "learning_rate": 1.2674878898181196e-06,
      "loss": 2.0439,
      "step": 223659
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.4389411211013794,
      "learning_rate": 1.2671122964076795e-06,
      "loss": 1.9378,
      "step": 223660
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.4977308511734009,
      "learning_rate": 1.2667367585373122e-06,
      "loss": 1.9356,
      "step": 223661
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.5000427961349487,
      "learning_rate": 1.266361276207084e-06,
      "loss": 2.0118,
      "step": 223662
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.445288062095642,
      "learning_rate": 1.265985849417095e-06,
      "loss": 1.9061,
      "step": 223663
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.5411961078643799,
      "learning_rate": 1.265610478167378e-06,
      "loss": 2.1099,
      "step": 223664
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.5939983129501343,
      "learning_rate": 1.265235162458067e-06,
      "loss": 2.0423,
      "step": 223665
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.3957184553146362,
      "learning_rate": 1.2648599022891614e-06,
      "loss": 1.9098,
      "step": 223666
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.6283950805664062,
      "learning_rate": 1.264484697660728e-06,
      "loss": 1.9706,
      "step": 223667
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.4863483905792236,
      "learning_rate": 1.2641095485729003e-06,
      "loss": 1.8459,
      "step": 223668
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.497885823249817,
      "learning_rate": 1.263734455025711e-06,
      "loss": 1.8528,
      "step": 223669
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.5469859838485718,
      "learning_rate": 1.2633594170192273e-06,
      "loss": 1.9075,
      "step": 223670
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.5073997974395752,
      "learning_rate": 1.2629844345535488e-06,
      "loss": 1.7866,
      "step": 223671
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.4612857103347778,
      "learning_rate": 1.2626095076286758e-06,
      "loss": 1.9199,
      "step": 223672
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.5952937602996826,
      "learning_rate": 1.262234636244741e-06,
      "loss": 1.9676,
      "step": 223673
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.5000070333480835,
      "learning_rate": 1.2618598204018116e-06,
      "loss": 2.1011,
      "step": 223674
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.6322338581085205,
      "learning_rate": 1.2614850600999204e-06,
      "loss": 2.1016,
      "step": 223675
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.4294408559799194,
      "learning_rate": 1.2611103553391677e-06,
      "loss": 1.9714,
      "step": 223676
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.5498639345169067,
      "learning_rate": 1.2607357061196199e-06,
      "loss": 2.0156,
      "step": 223677
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.391046404838562,
      "learning_rate": 1.2603611124413105e-06,
      "loss": 1.99,
      "step": 223678
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.4256616830825806,
      "learning_rate": 1.2599865743043724e-06,
      "loss": 1.8065,
      "step": 223679
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.4311656951904297,
      "learning_rate": 1.2596120917088392e-06,
      "loss": 1.9835,
      "step": 223680
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.6601866483688354,
      "learning_rate": 1.2592376646547774e-06,
      "loss": 2.1229,
      "step": 223681
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.4042465686798096,
      "learning_rate": 1.2588632931422537e-06,
      "loss": 1.8081,
      "step": 223682
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.3870867490768433,
      "learning_rate": 1.258488977171368e-06,
      "loss": 1.9355,
      "step": 223683
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.496679425239563,
      "learning_rate": 1.2581147167421534e-06,
      "loss": 2.043,
      "step": 223684
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.385850191116333,
      "learning_rate": 1.257740511854677e-06,
      "loss": 2.0926,
      "step": 223685
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.4232456684112549,
      "learning_rate": 1.257366362509038e-06,
      "loss": 2.141,
      "step": 223686
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.6054179668426514,
      "learning_rate": 1.2569922687053035e-06,
      "loss": 2.0085,
      "step": 223687
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.6712785959243774,
      "learning_rate": 1.2566182304435401e-06,
      "loss": 2.0086,
      "step": 223688
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.4554227590560913,
      "learning_rate": 1.2562442477238145e-06,
      "loss": 1.8417,
      "step": 223689
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.7776364088058472,
      "learning_rate": 1.2558703205461595e-06,
      "loss": 1.8435,
      "step": 223690
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.586856484413147,
      "learning_rate": 1.255496448910709e-06,
      "loss": 1.9003,
      "step": 223691
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.413663387298584,
      "learning_rate": 1.2551226328174624e-06,
      "loss": 2.0244,
      "step": 223692
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.7589988708496094,
      "learning_rate": 1.2547488722665533e-06,
      "loss": 2.1266,
      "step": 223693
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.5151445865631104,
      "learning_rate": 1.2543751672580482e-06,
      "loss": 1.7815,
      "step": 223694
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.629706859588623,
      "learning_rate": 1.2540015177919803e-06,
      "loss": 1.7079,
      "step": 223695
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.6146447658538818,
      "learning_rate": 1.2536279238684166e-06,
      "loss": 1.9725,
      "step": 223696
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.6196722984313965,
      "learning_rate": 1.2532543854874234e-06,
      "loss": 2.2768,
      "step": 223697
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.6159316301345825,
      "learning_rate": 1.2528809026491337e-06,
      "loss": 1.9279,
      "step": 223698
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.458601951599121,
      "learning_rate": 1.252507475353548e-06,
      "loss": 1.9852,
      "step": 223699
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.6090259552001953,
      "learning_rate": 1.252134103600766e-06,
      "loss": 2.2002,
      "step": 223700
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.435925841331482,
      "learning_rate": 1.2517607873908542e-06,
      "loss": 2.0852,
      "step": 223701
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.4173086881637573,
      "learning_rate": 1.2513875267238793e-06,
      "loss": 1.8743,
      "step": 223702
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.589465618133545,
      "learning_rate": 1.251014321599908e-06,
      "loss": 1.8363,
      "step": 223703
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.3506168127059937,
      "learning_rate": 1.2506411720190068e-06,
      "loss": 1.8796,
      "step": 223704
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.4400285482406616,
      "learning_rate": 1.2502680779812756e-06,
      "loss": 1.9341,
      "step": 223705
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.510209321975708,
      "learning_rate": 1.249895039486748e-06,
      "loss": 1.9743,
      "step": 223706
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.5505574941635132,
      "learning_rate": 1.2495220565354902e-06,
      "loss": 1.773,
      "step": 223707
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.4077562093734741,
      "learning_rate": 1.249149129127569e-06,
      "loss": 2.0629,
      "step": 223708
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.475054383277893,
      "learning_rate": 1.2487762572631176e-06,
      "loss": 1.8684,
      "step": 223709
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.390586495399475,
      "learning_rate": 1.248403440942136e-06,
      "loss": 2.0255,
      "step": 223710
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.4678374528884888,
      "learning_rate": 1.248030680164691e-06,
      "loss": 1.912,
      "step": 223711
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.3032697439193726,
      "learning_rate": 1.2476579749309156e-06,
      "loss": 1.7007,
      "step": 223712
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.528019666671753,
      "learning_rate": 1.2472853252408432e-06,
      "loss": 1.8072,
      "step": 223713
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.4642438888549805,
      "learning_rate": 1.246912731094507e-06,
      "loss": 2.2609,
      "step": 223714
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.5174602270126343,
      "learning_rate": 1.246540192492007e-06,
      "loss": 2.087,
      "step": 223715
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.4467419385910034,
      "learning_rate": 1.2461677094334432e-06,
      "loss": 1.8639,
      "step": 223716
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.6338855028152466,
      "learning_rate": 1.2457952819188488e-06,
      "loss": 1.9474,
      "step": 223717
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.4982178211212158,
      "learning_rate": 1.2454229099482904e-06,
      "loss": 2.1791,
      "step": 223718
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.483781337738037,
      "learning_rate": 1.245050593521868e-06,
      "loss": 1.9313,
      "step": 223719
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.5148603916168213,
      "learning_rate": 1.2446783326395815e-06,
      "loss": 2.2789,
      "step": 223720
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.4814951419830322,
      "learning_rate": 1.2443061273015975e-06,
      "loss": 1.9109,
      "step": 223721
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.432752013206482,
      "learning_rate": 1.243933977507916e-06,
      "loss": 2.0357,
      "step": 223722
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.250869631767273,
      "learning_rate": 1.243561883258637e-06,
      "loss": 1.6863,
      "step": 223723
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.545621633529663,
      "learning_rate": 1.2431898445538268e-06,
      "loss": 2.2661,
      "step": 223724
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.4805325269699097,
      "learning_rate": 1.242817861393519e-06,
      "loss": 1.9833,
      "step": 223725
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.4617600440979004,
      "learning_rate": 1.2424459337778137e-06,
      "loss": 2.0693,
      "step": 223726
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.507927656173706,
      "learning_rate": 1.2420740617068104e-06,
      "loss": 2.2176,
      "step": 223727
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.4048537015914917,
      "learning_rate": 1.2417022451805092e-06,
      "loss": 2.0787,
      "step": 223728
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.3788787126541138,
      "learning_rate": 1.2413304841990436e-06,
      "loss": 2.052,
      "step": 223729
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.5583404302597046,
      "learning_rate": 1.2409587787624464e-06,
      "loss": 2.1051,
      "step": 223730
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.486177682876587,
      "learning_rate": 1.2405871288707847e-06,
      "loss": 1.928,
      "step": 223731
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.452126145362854,
      "learning_rate": 1.240215534524125e-06,
      "loss": 1.9564,
      "step": 223732
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.367679238319397,
      "learning_rate": 1.239843995722567e-06,
      "loss": 1.7778,
      "step": 223733
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.4888743162155151,
      "learning_rate": 1.2394725124661442e-06,
      "loss": 1.9512,
      "step": 223734
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.4995752573013306,
      "learning_rate": 1.2391010847549565e-06,
      "loss": 1.9056,
      "step": 223735
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.409677267074585,
      "learning_rate": 1.2387297125890704e-06,
      "loss": 2.0898,
      "step": 223736
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.634920358657837,
      "learning_rate": 1.2383583959685528e-06,
      "loss": 1.9384,
      "step": 223737
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.5517666339874268,
      "learning_rate": 1.2379871348934368e-06,
      "loss": 2.1172,
      "step": 223738
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.3932126760482788,
      "learning_rate": 1.2376159293638221e-06,
      "loss": 1.8966,
      "step": 223739
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.4227221012115479,
      "learning_rate": 1.2372447793797758e-06,
      "loss": 2.0559,
      "step": 223740
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.4031953811645508,
      "learning_rate": 1.236873684941364e-06,
      "loss": 2.0585,
      "step": 223741
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.4562677145004272,
      "learning_rate": 1.2365026460486538e-06,
      "loss": 1.9129,
      "step": 223742
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.5715739727020264,
      "learning_rate": 1.2361316627017448e-06,
      "loss": 2.0228,
      "step": 223743
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.5239145755767822,
      "learning_rate": 1.2357607349006704e-06,
      "loss": 2.1368,
      "step": 223744
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.508231520652771,
      "learning_rate": 1.2353898626454973e-06,
      "loss": 1.9529,
      "step": 223745
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.358078122138977,
      "learning_rate": 1.2350190459362918e-06,
      "loss": 2.0113,
      "step": 223746
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.492218017578125,
      "learning_rate": 1.2346482847731543e-06,
      "loss": 2.1029,
      "step": 223747
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.4757788181304932,
      "learning_rate": 1.2342775791561178e-06,
      "loss": 2.1259,
      "step": 223748
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.3968342542648315,
      "learning_rate": 1.233906929085282e-06,
      "loss": 2.0942,
      "step": 223749
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.5199811458587646,
      "learning_rate": 1.233536334560714e-06,
      "loss": 1.9509,
      "step": 223750
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.4419313669204712,
      "learning_rate": 1.233165795582447e-06,
      "loss": 1.8896,
      "step": 223751
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.4464631080627441,
      "learning_rate": 1.2327953121505807e-06,
      "loss": 2.1288,
      "step": 223752
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.4892936944961548,
      "learning_rate": 1.2324248842651818e-06,
      "loss": 1.6962,
      "step": 223753
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.3680697679519653,
      "learning_rate": 1.232054511926317e-06,
      "loss": 2.0693,
      "step": 223754
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.4478280544281006,
      "learning_rate": 1.2316841951340529e-06,
      "loss": 1.9691,
      "step": 223755
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.401228904724121,
      "learning_rate": 1.2313139338884558e-06,
      "loss": 2.2006,
      "step": 223756
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.4912328720092773,
      "learning_rate": 1.2309437281895928e-06,
      "loss": 2.0285,
      "step": 223757
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.6257892847061157,
      "learning_rate": 1.2305735780375303e-06,
      "loss": 1.8622,
      "step": 223758
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.4047662019729614,
      "learning_rate": 1.230203483432368e-06,
      "loss": 2.1579,
      "step": 223759
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.4464954137802124,
      "learning_rate": 1.2298334443741398e-06,
      "loss": 1.8299,
      "step": 223760
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.5548657178878784,
      "learning_rate": 1.2294634608629117e-06,
      "loss": 2.416,
      "step": 223761
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.515031337738037,
      "learning_rate": 1.229093532898784e-06,
      "loss": 2.0124,
      "step": 223762
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.4880104064941406,
      "learning_rate": 1.2287236604817896e-06,
      "loss": 1.9665,
      "step": 223763
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.5002567768096924,
      "learning_rate": 1.2283538436120289e-06,
      "loss": 1.8965,
      "step": 223764
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.444159984588623,
      "learning_rate": 1.2279840822895682e-06,
      "loss": 2.0072,
      "step": 223765
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.4785196781158447,
      "learning_rate": 1.2276143765144408e-06,
      "loss": 1.8158,
      "step": 223766
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.4507627487182617,
      "learning_rate": 1.2272447262867468e-06,
      "loss": 1.9397,
      "step": 223767
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.5979551076889038,
      "learning_rate": 1.226875131606586e-06,
      "loss": 1.9519,
      "step": 223768
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.438524603843689,
      "learning_rate": 1.2265055924739253e-06,
      "loss": 2.1102,
      "step": 223769
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.9233875274658203,
      "learning_rate": 1.226136108888931e-06,
      "loss": 2.0667,
      "step": 223770
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.433758020401001,
      "learning_rate": 1.2257666808516364e-06,
      "loss": 1.8748,
      "step": 223771
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.390533447265625,
      "learning_rate": 1.2253973083621082e-06,
      "loss": 1.8436,
      "step": 223772
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.5018879175186157,
      "learning_rate": 1.225027991420413e-06,
      "loss": 1.9695,
      "step": 223773
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.704704999923706,
      "learning_rate": 1.2246587300266507e-06,
      "loss": 2.086,
      "step": 223774
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.4125314950942993,
      "learning_rate": 1.2242895241808214e-06,
      "loss": 1.7494,
      "step": 223775
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.4674090147018433,
      "learning_rate": 1.2239203738830583e-06,
      "loss": 1.9815,
      "step": 223776
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.3029544353485107,
      "learning_rate": 1.2235512791333946e-06,
      "loss": 1.8825,
      "step": 223777
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.3868434429168701,
      "learning_rate": 1.2231822399319635e-06,
      "loss": 2.0735,
      "step": 223778
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.467773199081421,
      "learning_rate": 1.2228132562787318e-06,
      "loss": 1.9965,
      "step": 223779
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.5090430974960327,
      "learning_rate": 1.2224443281738328e-06,
      "loss": 1.9829,
      "step": 223780
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.3206530809402466,
      "learning_rate": 1.222075455617333e-06,
      "loss": 1.9054,
      "step": 223781
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.4862990379333496,
      "learning_rate": 1.2217066386092655e-06,
      "loss": 1.8282,
      "step": 223782
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.5722825527191162,
      "learning_rate": 1.2213378771497306e-06,
      "loss": 2.0607,
      "step": 223783
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.5301361083984375,
      "learning_rate": 1.2209691712387948e-06,
      "loss": 2.0544,
      "step": 223784
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.429221749305725,
      "learning_rate": 1.2206005208765245e-06,
      "loss": 2.0353,
      "step": 223785
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.5536096096038818,
      "learning_rate": 1.2202319260629866e-06,
      "loss": 1.8082,
      "step": 223786
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.4279507398605347,
      "learning_rate": 1.2198633867982143e-06,
      "loss": 1.9583,
      "step": 223787
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.4323841333389282,
      "learning_rate": 1.2194949030823409e-06,
      "loss": 1.8502,
      "step": 223788
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.5649455785751343,
      "learning_rate": 1.2191264749153995e-06,
      "loss": 1.9369,
      "step": 223789
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.3934046030044556,
      "learning_rate": 1.218758102297457e-06,
      "loss": 1.7419,
      "step": 223790
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.5205695629119873,
      "learning_rate": 1.2183897852286129e-06,
      "loss": 2.1991,
      "step": 223791
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.485655426979065,
      "learning_rate": 1.2180215237089008e-06,
      "loss": 2.0332,
      "step": 223792
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.5195986032485962,
      "learning_rate": 1.2176533177383541e-06,
      "loss": 2.0185,
      "step": 223793
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.5209085941314697,
      "learning_rate": 1.2172851673171391e-06,
      "loss": 2.0764,
      "step": 223794
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.463179588317871,
      "learning_rate": 1.2169170724452226e-06,
      "loss": 2.0348,
      "step": 223795
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.4370321035385132,
      "learning_rate": 1.216549033122771e-06,
      "loss": 1.985,
      "step": 223796
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.3766471147537231,
      "learning_rate": 1.2161810493497849e-06,
      "loss": 1.8568,
      "step": 223797
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.4831422567367554,
      "learning_rate": 1.21581312112633e-06,
      "loss": 1.9156,
      "step": 223798
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.4466184377670288,
      "learning_rate": 1.2154452484525069e-06,
      "loss": 1.8525,
      "step": 223799
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.4448844194412231,
      "learning_rate": 1.2150774313283818e-06,
      "loss": 2.1179,
      "step": 223800
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.4598726034164429,
      "learning_rate": 1.2147096697539883e-06,
      "loss": 1.9721,
      "step": 223801
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.4743082523345947,
      "learning_rate": 1.2143419637294593e-06,
      "loss": 1.8376,
      "step": 223802
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.477781891822815,
      "learning_rate": 1.2139743132547952e-06,
      "loss": 1.9453,
      "step": 223803
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.4309242963790894,
      "learning_rate": 1.2136067183301291e-06,
      "loss": 1.9749,
      "step": 223804
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.5303661823272705,
      "learning_rate": 1.2132391789554607e-06,
      "loss": 1.887,
      "step": 223805
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.4284998178482056,
      "learning_rate": 1.2128716951308903e-06,
      "loss": 1.9899,
      "step": 223806
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.4700493812561035,
      "learning_rate": 1.2125042668564844e-06,
      "loss": 1.9193,
      "step": 223807
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.4816725254058838,
      "learning_rate": 1.2121368941323428e-06,
      "loss": 1.9266,
      "step": 223808
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.4229768514633179,
      "learning_rate": 1.2117695769584656e-06,
      "loss": 2.143,
      "step": 223809
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.3791766166687012,
      "learning_rate": 1.211402315334986e-06,
      "loss": 2.0342,
      "step": 223810
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.4868412017822266,
      "learning_rate": 1.2110351092619374e-06,
      "loss": 2.2634,
      "step": 223811
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.4703065156936646,
      "learning_rate": 1.2106679587394197e-06,
      "loss": 1.8455,
      "step": 223812
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.4979914426803589,
      "learning_rate": 1.2103008637674327e-06,
      "loss": 1.8773,
      "step": 223813
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.468617558479309,
      "learning_rate": 1.2099338243461431e-06,
      "loss": 1.8608,
      "step": 223814
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.428689956665039,
      "learning_rate": 1.2095668404755178e-06,
      "loss": 2.0798,
      "step": 223815
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.5113613605499268,
      "learning_rate": 1.2091999121557227e-06,
      "loss": 2.1742,
      "step": 223816
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.4568419456481934,
      "learning_rate": 1.2088330393867585e-06,
      "loss": 1.7979,
      "step": 223817
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.4674021005630493,
      "learning_rate": 1.2084662221686913e-06,
      "loss": 2.1128,
      "step": 223818
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.442522406578064,
      "learning_rate": 1.2080994605016215e-06,
      "loss": 1.9927,
      "step": 223819
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.434339165687561,
      "learning_rate": 1.2077327543856153e-06,
      "loss": 1.8737,
      "step": 223820
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.4238322973251343,
      "learning_rate": 1.2073661038207394e-06,
      "loss": 1.9407,
      "step": 223821
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.3825187683105469,
      "learning_rate": 1.2069995088070606e-06,
      "loss": 1.8315,
      "step": 223822
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.432295799255371,
      "learning_rate": 1.206632969344612e-06,
      "loss": 1.8771,
      "step": 223823
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.7158560752868652,
      "learning_rate": 1.2062664854334935e-06,
      "loss": 1.835,
      "step": 223824
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.4394426345825195,
      "learning_rate": 1.205900057073772e-06,
      "loss": 1.8638,
      "step": 223825
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.4787051677703857,
      "learning_rate": 1.2055336842655139e-06,
      "loss": 2.0308,
      "step": 223826
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.4079121351242065,
      "learning_rate": 1.2051673670087859e-06,
      "loss": 1.8153,
      "step": 223827
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.825587272644043,
      "learning_rate": 1.2048011053036878e-06,
      "loss": 1.8766,
      "step": 223828
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.4090349674224854,
      "learning_rate": 1.2044348991502195e-06,
      "loss": 1.8919,
      "step": 223829
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.6004642248153687,
      "learning_rate": 1.2040687485485146e-06,
      "loss": 2.2259,
      "step": 223830
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.3693102598190308,
      "learning_rate": 1.2037026534986062e-06,
      "loss": 2.0887,
      "step": 223831
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.628799557685852,
      "learning_rate": 1.2033366140005608e-06,
      "loss": 2.0897,
      "step": 223832
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.5437016487121582,
      "learning_rate": 1.202970630054445e-06,
      "loss": 2.0649,
      "step": 223833
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.5107992887496948,
      "learning_rate": 1.202604701660359e-06,
      "loss": 2.2095,
      "step": 223834
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.4342447519302368,
      "learning_rate": 1.202238828818336e-06,
      "loss": 1.9571,
      "step": 223835
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.5218487977981567,
      "learning_rate": 1.2018730115284758e-06,
      "loss": 1.7848,
      "step": 223836
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.9154711961746216,
      "learning_rate": 1.2015072497908118e-06,
      "loss": 1.8929,
      "step": 223837
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.433029055595398,
      "learning_rate": 1.201141543605444e-06,
      "loss": 1.7853,
      "step": 223838
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.4949872493743896,
      "learning_rate": 1.2007758929724055e-06,
      "loss": 1.8525,
      "step": 223839
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.5481255054473877,
      "learning_rate": 1.2004102978917963e-06,
      "loss": 1.909,
      "step": 223840
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.4972364902496338,
      "learning_rate": 1.2000447583636498e-06,
      "loss": 2.006,
      "step": 223841
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.4209837913513184,
      "learning_rate": 1.1996792743880657e-06,
      "loss": 1.849,
      "step": 223842
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.532458782196045,
      "learning_rate": 1.1993138459651109e-06,
      "loss": 1.9978,
      "step": 223843
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.7074264287948608,
      "learning_rate": 1.1989484730948519e-06,
      "loss": 2.1086,
      "step": 223844
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.3845930099487305,
      "learning_rate": 1.1985831557773218e-06,
      "loss": 2.3834,
      "step": 223845
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.451780915260315,
      "learning_rate": 1.1982178940126207e-06,
      "loss": 1.677,
      "step": 223846
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.4785230159759521,
      "learning_rate": 1.1978526878008154e-06,
      "loss": 1.9075,
      "step": 223847
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.5144344568252563,
      "learning_rate": 1.1974875371419722e-06,
      "loss": 1.8757,
      "step": 223848
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.661681890487671,
      "learning_rate": 1.197122442036158e-06,
      "loss": 1.8935,
      "step": 223849
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.5003963708877563,
      "learning_rate": 1.1967574024834392e-06,
      "loss": 1.7349,
      "step": 223850
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.4746780395507812,
      "learning_rate": 1.1963924184838825e-06,
      "loss": 2.0229,
      "step": 223851
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.567812442779541,
      "learning_rate": 1.1960274900375543e-06,
      "loss": 2.0324,
      "step": 223852
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.4659961462020874,
      "learning_rate": 1.1956626171445217e-06,
      "loss": 2.0078,
      "step": 223853
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.9185409545898438,
      "learning_rate": 1.1952977998048841e-06,
      "loss": 2.0878,
      "step": 223854
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.538860559463501,
      "learning_rate": 1.1949330380186417e-06,
      "loss": 2.0883,
      "step": 223855
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.4817042350769043,
      "learning_rate": 1.1945683317859277e-06,
      "loss": 2.2839,
      "step": 223856
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.3999056816101074,
      "learning_rate": 1.1942036811067758e-06,
      "loss": 2.1324,
      "step": 223857
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.542693018913269,
      "learning_rate": 1.1938390859812518e-06,
      "loss": 2.0253,
      "step": 223858
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.5494353771209717,
      "learning_rate": 1.1934745464094564e-06,
      "loss": 1.9405,
      "step": 223859
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.4675133228302002,
      "learning_rate": 1.193110062391389e-06,
      "loss": 2.0068,
      "step": 223860
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.7836061716079712,
      "learning_rate": 1.1927456339272168e-06,
      "loss": 2.1166,
      "step": 223861
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.4961116313934326,
      "learning_rate": 1.1923812610169058e-06,
      "loss": 2.1017,
      "step": 223862
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.5206891298294067,
      "learning_rate": 1.1920169436606232e-06,
      "loss": 1.924,
      "step": 223863
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.4352275133132935,
      "learning_rate": 1.1916526818583683e-06,
      "loss": 1.6837,
      "step": 223864
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.5651955604553223,
      "learning_rate": 1.1912884756102082e-06,
      "loss": 1.8623,
      "step": 223865
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.3824784755706787,
      "learning_rate": 1.1909243249162093e-06,
      "loss": 2.0207,
      "step": 223866
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.5563846826553345,
      "learning_rate": 1.1905602297765049e-06,
      "loss": 1.8718,
      "step": 223867
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.5510789155960083,
      "learning_rate": 1.190196190191095e-06,
      "loss": 1.9927,
      "step": 223868
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.4568376541137695,
      "learning_rate": 1.1898322061600462e-06,
      "loss": 1.9563,
      "step": 223869
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.5203627347946167,
      "learning_rate": 1.1894682776834919e-06,
      "loss": 1.9748,
      "step": 223870
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.458383560180664,
      "learning_rate": 1.1891044047614319e-06,
      "loss": 1.8892,
      "step": 223871
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.5960372686386108,
      "learning_rate": 1.188740587393966e-06,
      "loss": 1.818,
      "step": 223872
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.3979847431182861,
      "learning_rate": 1.1883768255811277e-06,
      "loss": 1.9321,
      "step": 223873
    },
    {
      "epoch": 2.91,
      "grad_norm": 1.5520992279052734,
      "learning_rate": 1.1880131193230503e-06,
      "loss": 2.0563,
      "step": 223874
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.4526430368423462,
      "learning_rate": 1.1876494686197335e-06,
      "loss": 2.0181,
      "step": 223875
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.4194743633270264,
      "learning_rate": 1.1872858734712776e-06,
      "loss": 1.8346,
      "step": 223876
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.4255223274230957,
      "learning_rate": 1.186922333877749e-06,
      "loss": 2.0992,
      "step": 223877
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.469190001487732,
      "learning_rate": 1.1865588498392143e-06,
      "loss": 1.9513,
      "step": 223878
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.3217599391937256,
      "learning_rate": 1.18619542135574e-06,
      "loss": 1.7807,
      "step": 223879
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.4786031246185303,
      "learning_rate": 1.185832048427393e-06,
      "loss": 1.8503,
      "step": 223880
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.5690388679504395,
      "learning_rate": 1.1854687310542399e-06,
      "loss": 1.9881,
      "step": 223881
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.4004560708999634,
      "learning_rate": 1.185105469236347e-06,
      "loss": 1.9407,
      "step": 223882
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.5323336124420166,
      "learning_rate": 1.1847422629737813e-06,
      "loss": 2.0684,
      "step": 223883
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.6251001358032227,
      "learning_rate": 1.184379112266609e-06,
      "loss": 2.097,
      "step": 223884
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.473846673965454,
      "learning_rate": 1.1840160171149305e-06,
      "loss": 2.061,
      "step": 223885
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.491248369216919,
      "learning_rate": 1.1836529775187454e-06,
      "loss": 1.8509,
      "step": 223886
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.580855369567871,
      "learning_rate": 1.183289993478187e-06,
      "loss": 2.1668,
      "step": 223887
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.526428461074829,
      "learning_rate": 1.1829270649932555e-06,
      "loss": 2.0597,
      "step": 223888
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.3524967432022095,
      "learning_rate": 1.182564192064117e-06,
      "loss": 1.8679,
      "step": 223889
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.4405908584594727,
      "learning_rate": 1.1822013746907388e-06,
      "loss": 2.0046,
      "step": 223890
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.3729013204574585,
      "learning_rate": 1.1818386128732204e-06,
      "loss": 1.9368,
      "step": 223891
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.5999817848205566,
      "learning_rate": 1.1814759066116619e-06,
      "loss": 2.0038,
      "step": 223892
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.5305991172790527,
      "learning_rate": 1.1811132559061297e-06,
      "loss": 2.1217,
      "step": 223893
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.3532718420028687,
      "learning_rate": 1.1807506607566242e-06,
      "loss": 1.8791,
      "step": 223894
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.5171092748641968,
      "learning_rate": 1.1803881211632783e-06,
      "loss": 2.0634,
      "step": 223895
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.4652585983276367,
      "learning_rate": 1.1800256371261584e-06,
      "loss": 1.7792,
      "step": 223896
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.6203621625900269,
      "learning_rate": 1.179663208645265e-06,
      "loss": 1.9553,
      "step": 223897
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.4922288656234741,
      "learning_rate": 1.1793008357207645e-06,
      "loss": 1.6987,
      "step": 223898
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.363729476928711,
      "learning_rate": 1.1789385183526234e-06,
      "loss": 1.871,
      "step": 223899
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.7326024770736694,
      "learning_rate": 1.178576256540975e-06,
      "loss": 1.9406,
      "step": 223900
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.7589515447616577,
      "learning_rate": 1.1782140502859195e-06,
      "loss": 1.8419,
      "step": 223901
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.5006718635559082,
      "learning_rate": 1.1778518995874232e-06,
      "loss": 2.1243,
      "step": 223902
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.4250930547714233,
      "learning_rate": 1.1774898044456194e-06,
      "loss": 2.2273,
      "step": 223903
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.6225106716156006,
      "learning_rate": 1.1771277648605415e-06,
      "loss": 2.0386,
      "step": 223904
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.448364019393921,
      "learning_rate": 1.1767657808322893e-06,
      "loss": 1.9768,
      "step": 223905
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.4084042310714722,
      "learning_rate": 1.17640385236093e-06,
      "loss": 1.8609,
      "step": 223906
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.4443347454071045,
      "learning_rate": 1.176041979446496e-06,
      "loss": 2.1096,
      "step": 223907
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.5777266025543213,
      "learning_rate": 1.1756801620890876e-06,
      "loss": 2.1778,
      "step": 223908
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.3493270874023438,
      "learning_rate": 1.1753184002887716e-06,
      "loss": 1.9212,
      "step": 223909
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.5809499025344849,
      "learning_rate": 1.1749566940455812e-06,
      "loss": 2.0961,
      "step": 223910
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.4380812644958496,
      "learning_rate": 1.1745950433596164e-06,
      "loss": 1.7666,
      "step": 223911
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.3492504358291626,
      "learning_rate": 1.1742334482309434e-06,
      "loss": 1.8372,
      "step": 223912
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.3365432024002075,
      "learning_rate": 1.1738719086596293e-06,
      "loss": 1.9655,
      "step": 223913
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.4301108121871948,
      "learning_rate": 1.1735104246457072e-06,
      "loss": 2.0437,
      "step": 223914
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.4893090724945068,
      "learning_rate": 1.1731489961893103e-06,
      "loss": 1.8914,
      "step": 223915
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.396807074546814,
      "learning_rate": 1.1727876232904054e-06,
      "loss": 2.1626,
      "step": 223916
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.7304226160049438,
      "learning_rate": 1.172426305949159e-06,
      "loss": 1.9741,
      "step": 223917
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.4706156253814697,
      "learning_rate": 1.1720650441656043e-06,
      "loss": 1.9531,
      "step": 223918
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.4853075742721558,
      "learning_rate": 1.171703837939808e-06,
      "loss": 2.0152,
      "step": 223919
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.3711285591125488,
      "learning_rate": 1.1713426872718034e-06,
      "loss": 1.8329,
      "step": 223920
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.5028598308563232,
      "learning_rate": 1.1709815921617238e-06,
      "loss": 2.0566,
      "step": 223921
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.4608993530273438,
      "learning_rate": 1.170620552609569e-06,
      "loss": 2.189,
      "step": 223922
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.4280540943145752,
      "learning_rate": 1.1702595686154392e-06,
      "loss": 2.0869,
      "step": 223923
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.4846299886703491,
      "learning_rate": 1.1698986401794009e-06,
      "loss": 1.7757,
      "step": 223924
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.4164180755615234,
      "learning_rate": 1.1695377673015205e-06,
      "loss": 2.1676,
      "step": 223925
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.5217781066894531,
      "learning_rate": 1.1691769499818648e-06,
      "loss": 1.8415,
      "step": 223926
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.407968521118164,
      "learning_rate": 1.1688161882205005e-06,
      "loss": 1.8899,
      "step": 223927
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.5250627994537354,
      "learning_rate": 1.168455482017494e-06,
      "loss": 2.0312,
      "step": 223928
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.6510130167007446,
      "learning_rate": 1.168094831372912e-06,
      "loss": 2.0562,
      "step": 223929
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.3793697357177734,
      "learning_rate": 1.1677342362868214e-06,
      "loss": 1.8315,
      "step": 223930
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.4523183107376099,
      "learning_rate": 1.1673736967593217e-06,
      "loss": 1.9487,
      "step": 223931
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.7302613258361816,
      "learning_rate": 1.1670132127904131e-06,
      "loss": 2.2367,
      "step": 223932
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.5396076440811157,
      "learning_rate": 1.1666527843801954e-06,
      "loss": 2.0707,
      "step": 223933
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.4792486429214478,
      "learning_rate": 1.1662924115287354e-06,
      "loss": 1.9926,
      "step": 223934
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.4817540645599365,
      "learning_rate": 1.1659320942361328e-06,
      "loss": 1.9558,
      "step": 223935
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.5247833728790283,
      "learning_rate": 1.1655718325023877e-06,
      "loss": 2.0756,
      "step": 223936
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.45534086227417,
      "learning_rate": 1.1652116263276335e-06,
      "loss": 1.7652,
      "step": 223937
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.4941520690917969,
      "learning_rate": 1.1648514757119031e-06,
      "loss": 2.1493,
      "step": 223938
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.4077001810073853,
      "learning_rate": 1.1644913806552303e-06,
      "loss": 1.8056,
      "step": 223939
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.5544915199279785,
      "learning_rate": 1.1641313411577814e-06,
      "loss": 2.0498,
      "step": 223940
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.5200488567352295,
      "learning_rate": 1.163771357219523e-06,
      "loss": 1.8771,
      "step": 223941
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.4147919416427612,
      "learning_rate": 1.163411428840555e-06,
      "loss": 1.864,
      "step": 223942
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.5296326875686646,
      "learning_rate": 1.1630515560209774e-06,
      "loss": 2.0538,
      "step": 223943
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.6620713472366333,
      "learning_rate": 1.1626917387607904e-06,
      "loss": 2.092,
      "step": 223944
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.5180485248565674,
      "learning_rate": 1.162331977060127e-06,
      "loss": 1.9341,
      "step": 223945
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.3667505979537964,
      "learning_rate": 1.1619722709190204e-06,
      "loss": 2.019,
      "step": 223946
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.4171488285064697,
      "learning_rate": 1.1616126203375376e-06,
      "loss": 2.0634,
      "step": 223947
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.3907732963562012,
      "learning_rate": 1.161253025315778e-06,
      "loss": 1.9616,
      "step": 223948
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.571955919265747,
      "learning_rate": 1.1608934858537755e-06,
      "loss": 1.8681,
      "step": 223949
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.4645509719848633,
      "learning_rate": 1.1605340019515963e-06,
      "loss": 1.973,
      "step": 223950
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.6062415838241577,
      "learning_rate": 1.1601745736093071e-06,
      "loss": 1.9253,
      "step": 223951
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.5607638359069824,
      "learning_rate": 1.1598152008269745e-06,
      "loss": 2.2263,
      "step": 223952
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.4189883470535278,
      "learning_rate": 1.1594558836046985e-06,
      "loss": 1.8836,
      "step": 223953
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.5753556489944458,
      "learning_rate": 1.1590966219425124e-06,
      "loss": 2.0246,
      "step": 223954
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.4352720975875854,
      "learning_rate": 1.1587374158404827e-06,
      "loss": 1.8287,
      "step": 223955
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.5338300466537476,
      "learning_rate": 1.1583782652987094e-06,
      "loss": 1.933,
      "step": 223956
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.4929404258728027,
      "learning_rate": 1.1580191703171925e-06,
      "loss": 1.7617,
      "step": 223957
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.4294674396514893,
      "learning_rate": 1.1576601308960653e-06,
      "loss": 1.8677,
      "step": 223958
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.4647703170776367,
      "learning_rate": 1.157301147035361e-06,
      "loss": 1.9362,
      "step": 223959
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.422611117362976,
      "learning_rate": 1.1569422187351796e-06,
      "loss": 1.8314,
      "step": 223960
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.434985876083374,
      "learning_rate": 1.156583345995521e-06,
      "loss": 1.8365,
      "step": 223961
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.4848319292068481,
      "learning_rate": 1.1562245288165517e-06,
      "loss": 2.0438,
      "step": 223962
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.4811490774154663,
      "learning_rate": 1.1558657671982386e-06,
      "loss": 2.0115,
      "step": 223963
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.4754278659820557,
      "learning_rate": 1.1555070611407147e-06,
      "loss": 2.1338,
      "step": 223964
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.4100956916809082,
      "learning_rate": 1.1551484106439801e-06,
      "loss": 2.0392,
      "step": 223965
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.4595266580581665,
      "learning_rate": 1.1547898157082014e-06,
      "loss": 2.0871,
      "step": 223966
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.4250397682189941,
      "learning_rate": 1.1544312763333452e-06,
      "loss": 1.7717,
      "step": 223967
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.648991346359253,
      "learning_rate": 1.1540727925195447e-06,
      "loss": 2.1623,
      "step": 223968
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.7139428853988647,
      "learning_rate": 1.1537143642668335e-06,
      "loss": 2.1163,
      "step": 223969
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.5224038362503052,
      "learning_rate": 1.153355991575311e-06,
      "loss": 2.0885,
      "step": 223970
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.5830258131027222,
      "learning_rate": 1.1529976744449776e-06,
      "loss": 2.192,
      "step": 223971
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.477292537689209,
      "learning_rate": 1.1526394128759663e-06,
      "loss": 2.164,
      "step": 223972
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.633496642112732,
      "learning_rate": 1.152281206868344e-06,
      "loss": 2.047,
      "step": 223973
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.511866807937622,
      "learning_rate": 1.1519230564221438e-06,
      "loss": 2.0824,
      "step": 223974
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.3984037637710571,
      "learning_rate": 1.151564961537399e-06,
      "loss": 1.8081,
      "step": 223975
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.4662251472473145,
      "learning_rate": 1.151206922214276e-06,
      "loss": 2.0038,
      "step": 223976
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.662514328956604,
      "learning_rate": 1.150848938452742e-06,
      "loss": 1.8905,
      "step": 223977
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.5903040170669556,
      "learning_rate": 1.1504910102529296e-06,
      "loss": 1.956,
      "step": 223978
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.42029869556427,
      "learning_rate": 1.1501331376148725e-06,
      "loss": 1.9592,
      "step": 223979
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.4966024160385132,
      "learning_rate": 1.1497753205386706e-06,
      "loss": 1.8716,
      "step": 223980
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.3580899238586426,
      "learning_rate": 1.1494175590243238e-06,
      "loss": 1.8901,
      "step": 223981
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.4098466634750366,
      "learning_rate": 1.1490598530719653e-06,
      "loss": 2.0267,
      "step": 223982
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.5406802892684937,
      "learning_rate": 1.1487022026816284e-06,
      "loss": 1.799,
      "step": 223983
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.6958794593811035,
      "learning_rate": 1.148344607853413e-06,
      "loss": 2.024,
      "step": 223984
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.2905874252319336,
      "learning_rate": 1.1479870685873195e-06,
      "loss": 1.9109,
      "step": 223985
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.5138301849365234,
      "learning_rate": 1.1476295848835138e-06,
      "loss": 1.8483,
      "step": 223986
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.479164481163025,
      "learning_rate": 1.147272156741963e-06,
      "loss": 1.9277,
      "step": 223987
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.3922271728515625,
      "learning_rate": 1.1469147841628002e-06,
      "loss": 2.1227,
      "step": 223988
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.3998119831085205,
      "learning_rate": 1.1465574671460255e-06,
      "loss": 2.0505,
      "step": 223989
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.6028783321380615,
      "learning_rate": 1.1462002056918052e-06,
      "loss": 2.2193,
      "step": 223990
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.498064398765564,
      "learning_rate": 1.1458429998001062e-06,
      "loss": 1.9587,
      "step": 223991
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.481436014175415,
      "learning_rate": 1.1454858494710616e-06,
      "loss": 1.9356,
      "step": 223992
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.415685772895813,
      "learning_rate": 1.1451287547046717e-06,
      "loss": 1.9518,
      "step": 223993
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.3132874965667725,
      "learning_rate": 1.1447717155011026e-06,
      "loss": 1.7686,
      "step": 223994
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.4791319370269775,
      "learning_rate": 1.1444147318603214e-06,
      "loss": 1.8546,
      "step": 223995
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.4060722589492798,
      "learning_rate": 1.1440578037824277e-06,
      "loss": 1.7195,
      "step": 223996
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.620404839515686,
      "learning_rate": 1.1437009312675216e-06,
      "loss": 2.0189,
      "step": 223997
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.435189962387085,
      "learning_rate": 1.1433441143156363e-06,
      "loss": 1.9609,
      "step": 223998
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.4967691898345947,
      "learning_rate": 1.1429873529268385e-06,
      "loss": 1.976,
      "step": 223999
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.4115225076675415,
      "learning_rate": 1.1426306471011948e-06,
      "loss": 1.8171,
      "step": 224000
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.5618573427200317,
      "learning_rate": 1.1422739968387718e-06,
      "loss": 2.0938,
      "step": 224001
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.3098517656326294,
      "learning_rate": 1.1419174021396692e-06,
      "loss": 2.0566,
      "step": 224002
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.3642886877059937,
      "learning_rate": 1.1415608630039207e-06,
      "loss": 1.8199,
      "step": 224003
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.4388809204101562,
      "learning_rate": 1.1412043794315596e-06,
      "loss": 2.1708,
      "step": 224004
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.4625684022903442,
      "learning_rate": 1.140847951422752e-06,
      "loss": 2.1848,
      "step": 224005
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.5030194520950317,
      "learning_rate": 1.1404915789774649e-06,
      "loss": 2.0032,
      "step": 224006
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.4313422441482544,
      "learning_rate": 1.1401352620957983e-06,
      "loss": 2.1098,
      "step": 224007
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.3879098892211914,
      "learning_rate": 1.1397790007778184e-06,
      "loss": 1.8495,
      "step": 224008
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.5147111415863037,
      "learning_rate": 1.1394227950235922e-06,
      "loss": 1.9427,
      "step": 224009
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.46588134765625,
      "learning_rate": 1.1390666448332197e-06,
      "loss": 1.8572,
      "step": 224010
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.4964230060577393,
      "learning_rate": 1.138710550206734e-06,
      "loss": 1.7457,
      "step": 224011
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.4772902727127075,
      "learning_rate": 1.1383545111441682e-06,
      "loss": 2.0037,
      "step": 224012
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.557798147201538,
      "learning_rate": 1.137998527645656e-06,
      "loss": 2.0542,
      "step": 224013
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.423290491104126,
      "learning_rate": 1.1376425997112305e-06,
      "loss": 1.8479,
      "step": 224014
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.6615434885025024,
      "learning_rate": 1.1372867273409248e-06,
      "loss": 2.0649,
      "step": 224015
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.4685670137405396,
      "learning_rate": 1.1369309105348723e-06,
      "loss": 1.9119,
      "step": 224016
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.3983384370803833,
      "learning_rate": 1.1365751492931062e-06,
      "loss": 1.8831,
      "step": 224017
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.576698899269104,
      "learning_rate": 1.1362194436156601e-06,
      "loss": 2.0423,
      "step": 224018
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.4912936687469482,
      "learning_rate": 1.135863793502667e-06,
      "loss": 2.0384,
      "step": 224019
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.5354609489440918,
      "learning_rate": 1.135508198954127e-06,
      "loss": 2.0292,
      "step": 224020
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.7522515058517456,
      "learning_rate": 1.1351526599701732e-06,
      "loss": 1.9533,
      "step": 224021
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.5593767166137695,
      "learning_rate": 1.1347971765508057e-06,
      "loss": 1.9391,
      "step": 224022
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.4396426677703857,
      "learning_rate": 1.1344417486961244e-06,
      "loss": 1.9855,
      "step": 224023
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.3945080041885376,
      "learning_rate": 1.134086376406196e-06,
      "loss": 2.2597,
      "step": 224024
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.4076844453811646,
      "learning_rate": 1.1337310596810867e-06,
      "loss": 1.8761,
      "step": 224025
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.4975337982177734,
      "learning_rate": 1.1333757985208635e-06,
      "loss": 1.8509,
      "step": 224026
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.373083472251892,
      "learning_rate": 1.133020592925593e-06,
      "loss": 1.9699,
      "step": 224027
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.5079987049102783,
      "learning_rate": 1.132665442895342e-06,
      "loss": 1.8581,
      "step": 224028
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.4572809934616089,
      "learning_rate": 1.1323103484301432e-06,
      "loss": 2.096,
      "step": 224029
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.6664544343948364,
      "learning_rate": 1.131955309530097e-06,
      "loss": 2.0367,
      "step": 224030
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.5709731578826904,
      "learning_rate": 1.13160032619527e-06,
      "loss": 2.1305,
      "step": 224031
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.4521654844284058,
      "learning_rate": 1.1312453984256953e-06,
      "loss": 1.8722,
      "step": 224032
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.4016664028167725,
      "learning_rate": 1.1308905262214728e-06,
      "loss": 2.0178,
      "step": 224033
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.5850547552108765,
      "learning_rate": 1.1305357095827028e-06,
      "loss": 1.9506,
      "step": 224034
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.4567903280258179,
      "learning_rate": 1.1301809485093515e-06,
      "loss": 2.2824,
      "step": 224035
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.4792898893356323,
      "learning_rate": 1.1298262430015526e-06,
      "loss": 2.0384,
      "step": 224036
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.5300840139389038,
      "learning_rate": 1.1294715930594055e-06,
      "loss": 1.8932,
      "step": 224037
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.7990477085113525,
      "learning_rate": 1.1291169986828775e-06,
      "loss": 1.945,
      "step": 224038
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.4115182161331177,
      "learning_rate": 1.1287624598721013e-06,
      "loss": 1.7352,
      "step": 224039
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.5804765224456787,
      "learning_rate": 1.1284079766271436e-06,
      "loss": 1.9697,
      "step": 224040
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.5558925867080688,
      "learning_rate": 1.1280535489480714e-06,
      "loss": 1.8532,
      "step": 224041
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.4846612215042114,
      "learning_rate": 1.1276991768348842e-06,
      "loss": 2.1312,
      "step": 224042
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.5111265182495117,
      "learning_rate": 1.1273448602877488e-06,
      "loss": 1.9729,
      "step": 224043
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.7793753147125244,
      "learning_rate": 1.1269905993066652e-06,
      "loss": 1.9329,
      "step": 224044
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.5229268074035645,
      "learning_rate": 1.1266363938917e-06,
      "loss": 1.9302,
      "step": 224045
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.6765519380569458,
      "learning_rate": 1.1262822440429531e-06,
      "loss": 2.0295,
      "step": 224046
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.4790380001068115,
      "learning_rate": 1.1259281497604577e-06,
      "loss": 1.8499,
      "step": 224047
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.5352407693862915,
      "learning_rate": 1.1255741110443139e-06,
      "loss": 1.8543,
      "step": 224048
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.5079452991485596,
      "learning_rate": 1.125220127894555e-06,
      "loss": 2.1183,
      "step": 224049
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.516026496887207,
      "learning_rate": 1.1248662003112474e-06,
      "loss": 1.8779,
      "step": 224050
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.4991627931594849,
      "learning_rate": 1.1245123282944912e-06,
      "loss": 1.9169,
      "step": 224051
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.6073964834213257,
      "learning_rate": 1.1241585118443196e-06,
      "loss": 1.8004,
      "step": 224052
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.518276572227478,
      "learning_rate": 1.1238047509608327e-06,
      "loss": 2.1642,
      "step": 224053
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.5618137121200562,
      "learning_rate": 1.1234510456440304e-06,
      "loss": 2.0306,
      "step": 224054
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.386033535003662,
      "learning_rate": 1.1230973958940459e-06,
      "loss": 1.9636,
      "step": 224055
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.4141629934310913,
      "learning_rate": 1.1227438017109125e-06,
      "loss": 1.931,
      "step": 224056
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.6033344268798828,
      "learning_rate": 1.1223902630946968e-06,
      "loss": 2.0842,
      "step": 224057
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.4097228050231934,
      "learning_rate": 1.1220367800454656e-06,
      "loss": 1.8656,
      "step": 224058
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.5830304622650146,
      "learning_rate": 1.1216833525633184e-06,
      "loss": 2.0084,
      "step": 224059
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.5005598068237305,
      "learning_rate": 1.1213299806482556e-06,
      "loss": 2.0841,
      "step": 224060
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.478681206703186,
      "learning_rate": 1.1209766643004104e-06,
      "loss": 2.0272,
      "step": 224061
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.4528383016586304,
      "learning_rate": 1.1206234035197826e-06,
      "loss": 1.7212,
      "step": 224062
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.4854501485824585,
      "learning_rate": 1.1202701983065054e-06,
      "loss": 2.0768,
      "step": 224063
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.4351832866668701,
      "learning_rate": 1.1199170486606123e-06,
      "loss": 2.0505,
      "step": 224064
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.4366902112960815,
      "learning_rate": 1.1195639545821367e-06,
      "loss": 1.8741,
      "step": 224065
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.6826379299163818,
      "learning_rate": 1.1192109160712116e-06,
      "loss": 2.1328,
      "step": 224066
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.38633131980896,
      "learning_rate": 1.1188579331278702e-06,
      "loss": 2.0688,
      "step": 224067
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.574988603591919,
      "learning_rate": 1.118505005752146e-06,
      "loss": 2.1621,
      "step": 224068
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.4021432399749756,
      "learning_rate": 1.118152133944139e-06,
      "loss": 2.0811,
      "step": 224069
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.48057222366333,
      "learning_rate": 1.1177993177039158e-06,
      "loss": 1.8617,
      "step": 224070
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.402540922164917,
      "learning_rate": 1.1174465570315428e-06,
      "loss": 1.8714,
      "step": 224071
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.4818593263626099,
      "learning_rate": 1.1170938519270533e-06,
      "loss": 1.84,
      "step": 224072
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.3570594787597656,
      "learning_rate": 1.1167412023905809e-06,
      "loss": 1.9072,
      "step": 224073
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.6730223894119263,
      "learning_rate": 1.116388608422092e-06,
      "loss": 1.9891,
      "step": 224074
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.4874745607376099,
      "learning_rate": 1.116036070021753e-06,
      "loss": 2.0801,
      "step": 224075
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.6312446594238281,
      "learning_rate": 1.115683587189564e-06,
      "loss": 2.0003,
      "step": 224076
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.4152823686599731,
      "learning_rate": 1.1153311599256253e-06,
      "loss": 2.1245,
      "step": 224077
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.5193254947662354,
      "learning_rate": 1.11497878822997e-06,
      "loss": 2.2406,
      "step": 224078
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.4986587762832642,
      "learning_rate": 1.1146264721026976e-06,
      "loss": 1.8035,
      "step": 224079
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.3916805982589722,
      "learning_rate": 1.114274211543842e-06,
      "loss": 2.0933,
      "step": 224080
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.45755934715271,
      "learning_rate": 1.1139220065535026e-06,
      "loss": 1.8993,
      "step": 224081
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.4607266187667847,
      "learning_rate": 1.113569857131713e-06,
      "loss": 2.2415,
      "step": 224082
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.5677944421768188,
      "learning_rate": 1.1132177632785732e-06,
      "loss": 2.0616,
      "step": 224083
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.4953267574310303,
      "learning_rate": 1.1128657249941163e-06,
      "loss": 2.0741,
      "step": 224084
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.4102468490600586,
      "learning_rate": 1.1125137422784092e-06,
      "loss": 1.965,
      "step": 224085
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.405779242515564,
      "learning_rate": 1.1121618151315514e-06,
      "loss": 1.9458,
      "step": 224086
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.4902738332748413,
      "learning_rate": 1.1118099435535765e-06,
      "loss": 1.9059,
      "step": 224087
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.573768973350525,
      "learning_rate": 1.1114581275445178e-06,
      "loss": 1.8637,
      "step": 224088
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.5424696207046509,
      "learning_rate": 1.1111063671045416e-06,
      "loss": 1.9768,
      "step": 224089
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.4659816026687622,
      "learning_rate": 1.110754662233615e-06,
      "loss": 2.1824,
      "step": 224090
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.5374141931533813,
      "learning_rate": 1.1104030129318708e-06,
      "loss": 2.1486,
      "step": 224091
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.4663244485855103,
      "learning_rate": 1.1100514191993092e-06,
      "loss": 2.001,
      "step": 224092
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.6317763328552246,
      "learning_rate": 1.1096998810360303e-06,
      "loss": 2.1207,
      "step": 224093
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.4587088823318481,
      "learning_rate": 1.1093483984421335e-06,
      "loss": 2.0206,
      "step": 224094
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.4014822244644165,
      "learning_rate": 1.1089969714176194e-06,
      "loss": 1.8623,
      "step": 224095
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.3512368202209473,
      "learning_rate": 1.108645599962621e-06,
      "loss": 1.9145,
      "step": 224096
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.3478654623031616,
      "learning_rate": 1.108294284077138e-06,
      "loss": 1.8763,
      "step": 224097
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.4565508365631104,
      "learning_rate": 1.1079430237612708e-06,
      "loss": 1.9934,
      "step": 224098
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.4676363468170166,
      "learning_rate": 1.1075918190150857e-06,
      "loss": 2.0014,
      "step": 224099
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.4104273319244385,
      "learning_rate": 1.107240669838616e-06,
      "loss": 2.0772,
      "step": 224100
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.458627462387085,
      "learning_rate": 1.1068895762319952e-06,
      "loss": 2.1284,
      "step": 224101
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.5674535036087036,
      "learning_rate": 1.106538538195223e-06,
      "loss": 2.1512,
      "step": 224102
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.550010323524475,
      "learning_rate": 1.1061875557283994e-06,
      "loss": 2.0131,
      "step": 224103
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.539530873298645,
      "learning_rate": 1.1058366288315578e-06,
      "loss": 1.9145,
      "step": 224104
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.3595917224884033,
      "learning_rate": 1.1054857575047981e-06,
      "loss": 1.7746,
      "step": 224105
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.5442357063293457,
      "learning_rate": 1.1051349417481537e-06,
      "loss": 2.0397,
      "step": 224106
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.4006526470184326,
      "learning_rate": 1.1047841815617242e-06,
      "loss": 1.8639,
      "step": 224107
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.4566965103149414,
      "learning_rate": 1.1044334769455432e-06,
      "loss": 2.157,
      "step": 224108
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.3516433238983154,
      "learning_rate": 1.1040828278997106e-06,
      "loss": 2.0249,
      "step": 224109
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.46702241897583,
      "learning_rate": 1.1037322344242595e-06,
      "loss": 1.99,
      "step": 224110
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.483282446861267,
      "learning_rate": 1.1033816965192566e-06,
      "loss": 1.9579,
      "step": 224111
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.5825704336166382,
      "learning_rate": 1.1030312141847686e-06,
      "loss": 1.9227,
      "step": 224112
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.470819354057312,
      "learning_rate": 1.1026807874208955e-06,
      "loss": 2.1299,
      "step": 224113
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.5501247644424438,
      "learning_rate": 1.1023304162276702e-06,
      "loss": 1.8181,
      "step": 224114
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.510284185409546,
      "learning_rate": 1.1019801006051597e-06,
      "loss": 2.0618,
      "step": 224115
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.5210846662521362,
      "learning_rate": 1.1016298405534308e-06,
      "loss": 1.8366,
      "step": 224116
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.5793436765670776,
      "learning_rate": 1.1012796360725496e-06,
      "loss": 2.0101,
      "step": 224117
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.351728081703186,
      "learning_rate": 1.100929487162583e-06,
      "loss": 2.049,
      "step": 224118
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.3962641954421997,
      "learning_rate": 1.1005793938236307e-06,
      "loss": 1.8588,
      "step": 224119
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.4552315473556519,
      "learning_rate": 1.1002293560556929e-06,
      "loss": 2.0729,
      "step": 224120
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.4204001426696777,
      "learning_rate": 1.0998793738588362e-06,
      "loss": 1.966,
      "step": 224121
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.3859243392944336,
      "learning_rate": 1.0995294472331939e-06,
      "loss": 1.9215,
      "step": 224122
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.4321415424346924,
      "learning_rate": 1.099179576178766e-06,
      "loss": 2.0494,
      "step": 224123
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.5572822093963623,
      "learning_rate": 1.0988297606956852e-06,
      "loss": 2.0464,
      "step": 224124
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.492579698562622,
      "learning_rate": 1.0984800007839189e-06,
      "loss": 1.8394,
      "step": 224125
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.3679248094558716,
      "learning_rate": 1.0981302964436334e-06,
      "loss": 1.9301,
      "step": 224126
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.4871981143951416,
      "learning_rate": 1.0977806476748285e-06,
      "loss": 2.0273,
      "step": 224127
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.6269363164901733,
      "learning_rate": 1.097431054477571e-06,
      "loss": 1.9017,
      "step": 224128
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.5999860763549805,
      "learning_rate": 1.097081516851961e-06,
      "loss": 2.0485,
      "step": 224129
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.4256330728530884,
      "learning_rate": 1.0967320347980647e-06,
      "loss": 1.9373,
      "step": 224130
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.6130681037902832,
      "learning_rate": 1.096382608315882e-06,
      "loss": 1.9191,
      "step": 224131
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.5680378675460815,
      "learning_rate": 1.096033237405547e-06,
      "loss": 2.0804,
      "step": 224132
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.5374211072921753,
      "learning_rate": 1.0956839220671253e-06,
      "loss": 1.8301,
      "step": 224133
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.7051026821136475,
      "learning_rate": 1.0953346623006508e-06,
      "loss": 2.0811,
      "step": 224134
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.339659333229065,
      "learning_rate": 1.0949854581061568e-06,
      "loss": 1.8264,
      "step": 224135
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.605014443397522,
      "learning_rate": 1.0946363094837763e-06,
      "loss": 2.0017,
      "step": 224136
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.6939995288848877,
      "learning_rate": 1.0942872164335426e-06,
      "loss": 1.9381,
      "step": 224137
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.5288221836090088,
      "learning_rate": 1.0939381789555224e-06,
      "loss": 1.9461,
      "step": 224138
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.5205109119415283,
      "learning_rate": 1.0935891970497823e-06,
      "loss": 1.8968,
      "step": 224139
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.6012920141220093,
      "learning_rate": 1.093240270716389e-06,
      "loss": 2.21,
      "step": 224140
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.400989055633545,
      "learning_rate": 1.0928913999554089e-06,
      "loss": 1.8683,
      "step": 224141
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.560103416442871,
      "learning_rate": 1.092542584766909e-06,
      "loss": 2.0177,
      "step": 224142
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.4795444011688232,
      "learning_rate": 1.092193825150922e-06,
      "loss": 1.9529,
      "step": 224143
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.3172434568405151,
      "learning_rate": 1.0918451211075486e-06,
      "loss": 1.8754,
      "step": 224144
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.4201326370239258,
      "learning_rate": 1.0914964726368546e-06,
      "loss": 1.7653,
      "step": 224145
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.493651032447815,
      "learning_rate": 1.091147879738874e-06,
      "loss": 1.9236,
      "step": 224146
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.5886194705963135,
      "learning_rate": 1.0907993424137063e-06,
      "loss": 2.0607,
      "step": 224147
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.434477686882019,
      "learning_rate": 1.090450860661385e-06,
      "loss": 1.8851,
      "step": 224148
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.5799366235733032,
      "learning_rate": 1.0901024344820097e-06,
      "loss": 2.0546,
      "step": 224149
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.4105889797210693,
      "learning_rate": 1.0897540638756142e-06,
      "loss": 2.0074,
      "step": 224150
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.399206280708313,
      "learning_rate": 1.089405748842298e-06,
      "loss": 1.9034,
      "step": 224151
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.516516923904419,
      "learning_rate": 1.0890574893820948e-06,
      "loss": 1.796,
      "step": 224152
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.5698645114898682,
      "learning_rate": 1.0887092854950708e-06,
      "loss": 1.8463,
      "step": 224153
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.5254615545272827,
      "learning_rate": 1.0883611371812928e-06,
      "loss": 1.9253,
      "step": 224154
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.5573126077651978,
      "learning_rate": 1.0880130444408274e-06,
      "loss": 2.0695,
      "step": 224155
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.4519160985946655,
      "learning_rate": 1.0876650072737413e-06,
      "loss": 1.8809,
      "step": 224156
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.510657548904419,
      "learning_rate": 1.0873170256801344e-06,
      "loss": 1.9438,
      "step": 224157
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.4564687013626099,
      "learning_rate": 1.0869690996600068e-06,
      "loss": 1.9328,
      "step": 224158
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.496337890625,
      "learning_rate": 1.086621229213458e-06,
      "loss": 2.0447,
      "step": 224159
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.393656849861145,
      "learning_rate": 1.0862734143405549e-06,
      "loss": 1.7985,
      "step": 224160
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.3969676494598389,
      "learning_rate": 1.0859256550413308e-06,
      "loss": 2.1203,
      "step": 224161
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.5682741403579712,
      "learning_rate": 1.0855779513159191e-06,
      "loss": 1.861,
      "step": 224162
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.5021612644195557,
      "learning_rate": 1.0852303031643194e-06,
      "loss": 2.2024,
      "step": 224163
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.4941877126693726,
      "learning_rate": 1.084882710586632e-06,
      "loss": 2.1171,
      "step": 224164
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.4538486003875732,
      "learning_rate": 1.08453517358289e-06,
      "loss": 1.954,
      "step": 224165
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.5061992406845093,
      "learning_rate": 1.08418769215316e-06,
      "loss": 2.1911,
      "step": 224166
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.4994603395462036,
      "learning_rate": 1.083840266297542e-06,
      "loss": 1.937,
      "step": 224167
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.446077823638916,
      "learning_rate": 1.0834928960161027e-06,
      "loss": 1.9764,
      "step": 224168
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.751878023147583,
      "learning_rate": 1.0831455813088419e-06,
      "loss": 2.0113,
      "step": 224169
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.4172312021255493,
      "learning_rate": 1.0827983221758928e-06,
      "loss": 1.9196,
      "step": 224170
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.6454800367355347,
      "learning_rate": 1.082451118617289e-06,
      "loss": 1.9332,
      "step": 224171
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.3805911540985107,
      "learning_rate": 1.082103970633097e-06,
      "loss": 2.1881,
      "step": 224172
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.4845073223114014,
      "learning_rate": 1.0817568782233832e-06,
      "loss": 1.997,
      "step": 224173
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.590960144996643,
      "learning_rate": 1.0814098413882144e-06,
      "loss": 1.9791,
      "step": 224174
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.5927512645721436,
      "learning_rate": 1.0810628601276573e-06,
      "loss": 1.8549,
      "step": 224175
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.47742760181427,
      "learning_rate": 1.0807159344417782e-06,
      "loss": 2.0948,
      "step": 224176
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.4007059335708618,
      "learning_rate": 1.0803690643306106e-06,
      "loss": 2.0606,
      "step": 224177
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.3438714742660522,
      "learning_rate": 1.080022249794288e-06,
      "loss": 1.8417,
      "step": 224178
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.3493465185165405,
      "learning_rate": 1.0796754908328098e-06,
      "loss": 1.8668,
      "step": 224179
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.5348199605941772,
      "learning_rate": 1.0793287874462431e-06,
      "loss": 2.0397,
      "step": 224180
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.4185736179351807,
      "learning_rate": 1.0789821396346876e-06,
      "loss": 2.0505,
      "step": 224181
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.5714796781539917,
      "learning_rate": 1.07863554739821e-06,
      "loss": 2.2975,
      "step": 224182
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.3378443717956543,
      "learning_rate": 1.0782890107368436e-06,
      "loss": 1.8326,
      "step": 224183
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.4854803085327148,
      "learning_rate": 1.077942529650655e-06,
      "loss": 1.8409,
      "step": 224184
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.5171512365341187,
      "learning_rate": 1.0775961041397108e-06,
      "loss": 1.9254,
      "step": 224185
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.3820486068725586,
      "learning_rate": 1.0772497342041108e-06,
      "loss": 1.8385,
      "step": 224186
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.4476275444030762,
      "learning_rate": 1.0769034198438886e-06,
      "loss": 1.8836,
      "step": 224187
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.5017955303192139,
      "learning_rate": 1.0765571610591105e-06,
      "loss": 1.8494,
      "step": 224188
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.5476369857788086,
      "learning_rate": 1.0762109578498434e-06,
      "loss": 2.1393,
      "step": 224189
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.4913830757141113,
      "learning_rate": 1.0758648102161537e-06,
      "loss": 1.9121,
      "step": 224190
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.4978584051132202,
      "learning_rate": 1.0755187181580748e-06,
      "loss": 1.8869,
      "step": 224191
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.3821065425872803,
      "learning_rate": 1.07517268167574e-06,
      "loss": 1.9769,
      "step": 224192
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.4333068132400513,
      "learning_rate": 1.074826700769149e-06,
      "loss": 1.902,
      "step": 224193
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.5182976722717285,
      "learning_rate": 1.074480775438402e-06,
      "loss": 2.0236,
      "step": 224194
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.4273573160171509,
      "learning_rate": 1.0741349056835324e-06,
      "loss": 2.0316,
      "step": 224195
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.5231647491455078,
      "learning_rate": 1.07378909150464e-06,
      "loss": 1.7771,
      "step": 224196
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.489825963973999,
      "learning_rate": 1.0734433329017579e-06,
      "loss": 2.0182,
      "step": 224197
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.4903900623321533,
      "learning_rate": 1.073097629874986e-06,
      "loss": 2.0376,
      "step": 224198
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.4106792211532593,
      "learning_rate": 1.072751982424358e-06,
      "loss": 1.7063,
      "step": 224199
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.3791236877441406,
      "learning_rate": 1.0724063905499404e-06,
      "loss": 2.0069,
      "step": 224200
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.4796196222305298,
      "learning_rate": 1.0720608542518328e-06,
      "loss": 1.9606,
      "step": 224201
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.4129958152770996,
      "learning_rate": 1.0717153735300355e-06,
      "loss": 1.9191,
      "step": 224202
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.496250033378601,
      "learning_rate": 1.0713699483846816e-06,
      "loss": 2.2072,
      "step": 224203
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.511523723602295,
      "learning_rate": 1.0710245788157713e-06,
      "loss": 2.3277,
      "step": 224204
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.5147082805633545,
      "learning_rate": 1.0706792648234375e-06,
      "loss": 1.9866,
      "step": 224205
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.407378911972046,
      "learning_rate": 1.0703340064076804e-06,
      "loss": 1.8679,
      "step": 224206
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.566855788230896,
      "learning_rate": 1.0699888035686e-06,
      "loss": 2.2267,
      "step": 224207
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.435176134109497,
      "learning_rate": 1.0696436563062294e-06,
      "loss": 1.9532,
      "step": 224208
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.4827290773391724,
      "learning_rate": 1.0692985646207019e-06,
      "loss": 2.0076,
      "step": 224209
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.4948049783706665,
      "learning_rate": 1.0689535285119844e-06,
      "loss": 2.1664,
      "step": 224210
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.5091328620910645,
      "learning_rate": 1.0686085479802098e-06,
      "loss": 2.2015,
      "step": 224211
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.5709317922592163,
      "learning_rate": 1.068263623025445e-06,
      "loss": 1.9842,
      "step": 224212
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.5975115299224854,
      "learning_rate": 1.067918753647723e-06,
      "loss": 1.9439,
      "step": 224213
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.4171135425567627,
      "learning_rate": 1.0675739398470774e-06,
      "loss": 1.9789,
      "step": 224214
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.502091407775879,
      "learning_rate": 1.0672291816236744e-06,
      "loss": 1.998,
      "step": 224215
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.3648507595062256,
      "learning_rate": 1.066884478977481e-06,
      "loss": 1.9716,
      "step": 224216
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.5366036891937256,
      "learning_rate": 1.0665398319085972e-06,
      "loss": 1.967,
      "step": 224217
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.547309160232544,
      "learning_rate": 1.0661952404170893e-06,
      "loss": 2.0363,
      "step": 224218
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.4254798889160156,
      "learning_rate": 1.065850704503024e-06,
      "loss": 2.0022,
      "step": 224219
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.4394747018814087,
      "learning_rate": 1.0655062241664347e-06,
      "loss": 1.8236,
      "step": 224220
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.336729884147644,
      "learning_rate": 1.0651617994074546e-06,
      "loss": 2.0094,
      "step": 224221
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.490867257118225,
      "learning_rate": 1.0648174302260838e-06,
      "loss": 2.0343,
      "step": 224222
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.5436025857925415,
      "learning_rate": 1.0644731166223886e-06,
      "loss": 2.1248,
      "step": 224223
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.4360933303833008,
      "learning_rate": 1.0641288585964691e-06,
      "loss": 1.9669,
      "step": 224224
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.4401320219039917,
      "learning_rate": 1.063784656148392e-06,
      "loss": 1.9671,
      "step": 224225
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.3894022703170776,
      "learning_rate": 1.0634405092781573e-06,
      "loss": 1.8222,
      "step": 224226
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.3561121225357056,
      "learning_rate": 1.0630964179858981e-06,
      "loss": 1.999,
      "step": 224227
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.4827083349227905,
      "learning_rate": 1.0627523822716478e-06,
      "loss": 1.9494,
      "step": 224228
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.50884211063385,
      "learning_rate": 1.062408402135473e-06,
      "loss": 1.8793,
      "step": 224229
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.4481264352798462,
      "learning_rate": 1.06206447757744e-06,
      "loss": 1.9158,
      "step": 224230
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.3959728479385376,
      "learning_rate": 1.061720608597616e-06,
      "loss": 1.9791,
      "step": 224231
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.4683470726013184,
      "learning_rate": 1.0613767951960338e-06,
      "loss": 1.7913,
      "step": 224232
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.4505985975265503,
      "learning_rate": 1.0610330373727938e-06,
      "loss": 1.9322,
      "step": 224233
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.9071604013442993,
      "learning_rate": 1.0606893351279621e-06,
      "loss": 1.8903,
      "step": 224234
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.4794586896896362,
      "learning_rate": 1.0603456884616058e-06,
      "loss": 1.9583,
      "step": 224235
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.6238137483596802,
      "learning_rate": 1.0600020973737578e-06,
      "loss": 2.0262,
      "step": 224236
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.3615132570266724,
      "learning_rate": 1.0596585618645182e-06,
      "loss": 2.191,
      "step": 224237
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.422302484512329,
      "learning_rate": 1.059315081933887e-06,
      "loss": 2.0314,
      "step": 224238
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.5069997310638428,
      "learning_rate": 1.0589716575819973e-06,
      "loss": 2.1784,
      "step": 224239
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.5484867095947266,
      "learning_rate": 1.0586282888088826e-06,
      "loss": 2.2406,
      "step": 224240
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.5293397903442383,
      "learning_rate": 1.0582849756146428e-06,
      "loss": 2.0248,
      "step": 224241
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.448468565940857,
      "learning_rate": 1.0579417179992777e-06,
      "loss": 2.1093,
      "step": 224242
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.3926873207092285,
      "learning_rate": 1.0575985159628874e-06,
      "loss": 2.1423,
      "step": 224243
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.4818389415740967,
      "learning_rate": 1.0572553695055385e-06,
      "loss": 2.0082,
      "step": 224244
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.3962732553482056,
      "learning_rate": 1.0569122786272976e-06,
      "loss": 1.9686,
      "step": 224245
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.5183722972869873,
      "learning_rate": 1.056569243328198e-06,
      "loss": 1.9385,
      "step": 224246
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.6078455448150635,
      "learning_rate": 1.0562262636083397e-06,
      "loss": 2.1094,
      "step": 224247
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.470706582069397,
      "learning_rate": 1.0558833394677891e-06,
      "loss": 2.1023,
      "step": 224248
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.372873067855835,
      "learning_rate": 1.0555404709065795e-06,
      "loss": 2.0399,
      "step": 224249
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.4869558811187744,
      "learning_rate": 1.0551976579247778e-06,
      "loss": 1.9828,
      "step": 224250
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.344499111175537,
      "learning_rate": 1.0548549005224505e-06,
      "loss": 1.9134,
      "step": 224251
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.6186100244522095,
      "learning_rate": 1.054512198699664e-06,
      "loss": 1.8981,
      "step": 224252
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.5080933570861816,
      "learning_rate": 1.0541695524565187e-06,
      "loss": 1.9995,
      "step": 224253
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.532454013824463,
      "learning_rate": 1.0538269617930474e-06,
      "loss": 2.1662,
      "step": 224254
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.5087034702301025,
      "learning_rate": 1.0534844267092834e-06,
      "loss": 1.9461,
      "step": 224255
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.4439716339111328,
      "learning_rate": 1.053141947205327e-06,
      "loss": 2.0217,
      "step": 224256
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.468274474143982,
      "learning_rate": 1.0527995232812448e-06,
      "loss": 1.9589,
      "step": 224257
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.5087742805480957,
      "learning_rate": 1.0524571549370697e-06,
      "loss": 2.0452,
      "step": 224258
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.523181676864624,
      "learning_rate": 1.052114842172902e-06,
      "loss": 2.2338,
      "step": 224259
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.494016408920288,
      "learning_rate": 1.0517725849888082e-06,
      "loss": 2.1891,
      "step": 224260
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.566301941871643,
      "learning_rate": 1.0514303833848215e-06,
      "loss": 1.9689,
      "step": 224261
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.443678617477417,
      "learning_rate": 1.0510882373609754e-06,
      "loss": 1.9982,
      "step": 224262
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.5610963106155396,
      "learning_rate": 1.050746146917436e-06,
      "loss": 1.8787,
      "step": 224263
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.3807183504104614,
      "learning_rate": 1.0504041120541705e-06,
      "loss": 2.1157,
      "step": 224264
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.453979253768921,
      "learning_rate": 1.0500621327712787e-06,
      "loss": 1.8902,
      "step": 224265
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.4845322370529175,
      "learning_rate": 1.0497202090688272e-06,
      "loss": 2.0788,
      "step": 224266
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.4412753582000732,
      "learning_rate": 1.0493783409468825e-06,
      "loss": 2.0478,
      "step": 224267
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.531798243522644,
      "learning_rate": 1.0490365284055113e-06,
      "loss": 1.8483,
      "step": 224268
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.5355736017227173,
      "learning_rate": 1.048694771444747e-06,
      "loss": 1.7784,
      "step": 224269
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.532610297203064,
      "learning_rate": 1.0483530700646892e-06,
      "loss": 2.0922,
      "step": 224270
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.527880072593689,
      "learning_rate": 1.0480114242653715e-06,
      "loss": 1.8605,
      "step": 224271
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.7865939140319824,
      "learning_rate": 1.0476698340468604e-06,
      "loss": 1.8944,
      "step": 224272
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.5183823108673096,
      "learning_rate": 1.0473282994092558e-06,
      "loss": 2.0293,
      "step": 224273
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.3886182308197021,
      "learning_rate": 1.0469868203525912e-06,
      "loss": 1.8817,
      "step": 224274
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.511098861694336,
      "learning_rate": 1.0466453968769328e-06,
      "loss": 1.9651,
      "step": 224275
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.4888981580734253,
      "learning_rate": 1.0463040289823476e-06,
      "loss": 1.8448,
      "step": 224276
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.4296422004699707,
      "learning_rate": 1.0459627166689022e-06,
      "loss": 1.9575,
      "step": 224277
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.5087424516677856,
      "learning_rate": 1.0456214599366298e-06,
      "loss": 1.803,
      "step": 224278
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.631415843963623,
      "learning_rate": 1.0452802587856302e-06,
      "loss": 1.9139,
      "step": 224279
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.581102967262268,
      "learning_rate": 1.0449391132159702e-06,
      "loss": 2.1126,
      "step": 224280
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.509770393371582,
      "learning_rate": 1.0445980232277162e-06,
      "loss": 2.0359,
      "step": 224281
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.4736326932907104,
      "learning_rate": 1.0442569888208685e-06,
      "loss": 2.1645,
      "step": 224282
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.5165448188781738,
      "learning_rate": 1.0439160099955602e-06,
      "loss": 2.0872,
      "step": 224283
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.4800788164138794,
      "learning_rate": 1.0435750867518244e-06,
      "loss": 1.8055,
      "step": 224284
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.3898580074310303,
      "learning_rate": 1.0432342190897613e-06,
      "loss": 2.1462,
      "step": 224285
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.547075867652893,
      "learning_rate": 1.0428934070093708e-06,
      "loss": 1.9734,
      "step": 224286
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.4985597133636475,
      "learning_rate": 1.0425526505107862e-06,
      "loss": 1.9518,
      "step": 224287
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.536442518234253,
      "learning_rate": 1.042211949593974e-06,
      "loss": 1.9988,
      "step": 224288
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.3679075241088867,
      "learning_rate": 1.0418713042591008e-06,
      "loss": 2.0632,
      "step": 224289
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.3496549129486084,
      "learning_rate": 1.041530714506167e-06,
      "loss": 1.7763,
      "step": 224290
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.3891408443450928,
      "learning_rate": 1.0411901803352718e-06,
      "loss": 1.8425,
      "step": 224291
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.727044939994812,
      "learning_rate": 1.0408497017464823e-06,
      "loss": 2.164,
      "step": 224292
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.6696854829788208,
      "learning_rate": 1.0405092787397984e-06,
      "loss": 2.1027,
      "step": 224293
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.4099758863449097,
      "learning_rate": 1.0401689113153533e-06,
      "loss": 2.1428,
      "step": 224294
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.42653226852417,
      "learning_rate": 1.0398285994731804e-06,
      "loss": 2.0988,
      "step": 224295
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.5429303646087646,
      "learning_rate": 1.0394883432133127e-06,
      "loss": 1.956,
      "step": 224296
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.45054030418396,
      "learning_rate": 1.039148142535884e-06,
      "loss": 1.8616,
      "step": 224297
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.5017566680908203,
      "learning_rate": 1.0388079974409268e-06,
      "loss": 2.0449,
      "step": 224298
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.473854899406433,
      "learning_rate": 1.0384679079284752e-06,
      "loss": 2.0614,
      "step": 224299
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.3529253005981445,
      "learning_rate": 1.0381278739986288e-06,
      "loss": 1.8879,
      "step": 224300
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.4394879341125488,
      "learning_rate": 1.037787895651454e-06,
      "loss": 1.9697,
      "step": 224301
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.533374309539795,
      "learning_rate": 1.0374479728869845e-06,
      "loss": 1.9261,
      "step": 224302
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.4992996454238892,
      "learning_rate": 1.0371081057052865e-06,
      "loss": 1.9193,
      "step": 224303
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.3309516906738281,
      "learning_rate": 1.036768294106427e-06,
      "loss": 1.8502,
      "step": 224304
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.4489127397537231,
      "learning_rate": 1.0364285380905058e-06,
      "loss": 2.169,
      "step": 224305
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.5188277959823608,
      "learning_rate": 1.0360888376575226e-06,
      "loss": 2.1054,
      "step": 224306
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.5244019031524658,
      "learning_rate": 1.0357491928075778e-06,
      "loss": 1.888,
      "step": 224307
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.5046590566635132,
      "learning_rate": 1.0354096035407379e-06,
      "loss": 1.9094,
      "step": 224308
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.3930590152740479,
      "learning_rate": 1.035070069857069e-06,
      "loss": 1.9225,
      "step": 224309
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.636709451675415,
      "learning_rate": 1.0347305917566384e-06,
      "loss": 2.0455,
      "step": 224310
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.4462331533432007,
      "learning_rate": 1.0343911692394457e-06,
      "loss": 1.7789,
      "step": 224311
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.5752768516540527,
      "learning_rate": 1.0340518023056576e-06,
      "loss": 2.078,
      "step": 224312
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.493159532546997,
      "learning_rate": 1.0337124909552407e-06,
      "loss": 1.8701,
      "step": 224313
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.5362555980682373,
      "learning_rate": 1.0333732351883284e-06,
      "loss": 2.0058,
      "step": 224314
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.460070252418518,
      "learning_rate": 1.0330340350049537e-06,
      "loss": 2.3217,
      "step": 224315
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.5188418626785278,
      "learning_rate": 1.0326948904051502e-06,
      "loss": 1.8744,
      "step": 224316
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.427444577217102,
      "learning_rate": 1.0323558013890508e-06,
      "loss": 1.9306,
      "step": 224317
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.5158605575561523,
      "learning_rate": 1.032016767956656e-06,
      "loss": 1.9418,
      "step": 224318
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.5352035760879517,
      "learning_rate": 1.031677790108032e-06,
      "loss": 2.013,
      "step": 224319
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.3624979257583618,
      "learning_rate": 1.0313388678433122e-06,
      "loss": 1.8815,
      "step": 224320
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.4312996864318848,
      "learning_rate": 1.0310000011624963e-06,
      "loss": 1.8794,
      "step": 224321
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.472481608390808,
      "learning_rate": 1.0306611900656514e-06,
      "loss": 1.9399,
      "step": 224322
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.6892873048782349,
      "learning_rate": 1.0303224345528438e-06,
      "loss": 1.9068,
      "step": 224323
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.4241821765899658,
      "learning_rate": 1.0299837346241402e-06,
      "loss": 1.9873,
      "step": 224324
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.541233777999878,
      "learning_rate": 1.0296450902796072e-06,
      "loss": 2.112,
      "step": 224325
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.5219569206237793,
      "learning_rate": 1.0293065015193115e-06,
      "loss": 2.0073,
      "step": 224326
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.4347649812698364,
      "learning_rate": 1.0289679683433195e-06,
      "loss": 2.1879,
      "step": 224327
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.5146503448486328,
      "learning_rate": 1.028629490751698e-06,
      "loss": 1.8506,
      "step": 224328
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.3940848112106323,
      "learning_rate": 1.02829106874448e-06,
      "loss": 1.9714,
      "step": 224329
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.428985834121704,
      "learning_rate": 1.0279527023217327e-06,
      "loss": 1.9751,
      "step": 224330
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.4040145874023438,
      "learning_rate": 1.0276143914835556e-06,
      "loss": 1.959,
      "step": 224331
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.518607258796692,
      "learning_rate": 1.0272761362299819e-06,
      "loss": 2.0899,
      "step": 224332
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.6085538864135742,
      "learning_rate": 1.0269379365610787e-06,
      "loss": 1.7751,
      "step": 224333
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.4668567180633545,
      "learning_rate": 1.026599792476912e-06,
      "loss": 2.1232,
      "step": 224334
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.3995064496994019,
      "learning_rate": 1.026261703977549e-06,
      "loss": 2.1514,
      "step": 224335
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.5089799165725708,
      "learning_rate": 1.0259236710630226e-06,
      "loss": 1.8563,
      "step": 224336
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.5965557098388672,
      "learning_rate": 1.0255856937334328e-06,
      "loss": 2.2226,
      "step": 224337
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.3531229496002197,
      "learning_rate": 1.0252477719888463e-06,
      "loss": 1.7255,
      "step": 224338
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.554261326789856,
      "learning_rate": 1.0249099058292965e-06,
      "loss": 2.2539,
      "step": 224339
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.3525229692459106,
      "learning_rate": 1.024572095254883e-06,
      "loss": 1.8936,
      "step": 224340
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.6011950969696045,
      "learning_rate": 1.0242343402656062e-06,
      "loss": 2.0579,
      "step": 224341
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.6222223043441772,
      "learning_rate": 1.0238966408615989e-06,
      "loss": 1.9324,
      "step": 224342
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.3618090152740479,
      "learning_rate": 1.0235589970428614e-06,
      "loss": 1.9479,
      "step": 224343
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.4729994535446167,
      "learning_rate": 1.023221408809527e-06,
      "loss": 1.9668,
      "step": 224344
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.5222326517105103,
      "learning_rate": 1.0228838761615954e-06,
      "loss": 1.9995,
      "step": 224345
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.4156441688537598,
      "learning_rate": 1.0225463990991666e-06,
      "loss": 2.2331,
      "step": 224346
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.5564583539962769,
      "learning_rate": 1.0222089776222742e-06,
      "loss": 2.1347,
      "step": 224347
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.4704296588897705,
      "learning_rate": 1.0218716117310178e-06,
      "loss": 2.0235,
      "step": 224348
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.6111723184585571,
      "learning_rate": 1.0215343014253974e-06,
      "loss": 2.171,
      "step": 224349
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.4568415880203247,
      "learning_rate": 1.0211970467055464e-06,
      "loss": 1.8776,
      "step": 224350
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.4496461153030396,
      "learning_rate": 1.020859847571498e-06,
      "loss": 2.1718,
      "step": 224351
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.5236114263534546,
      "learning_rate": 1.0205227040233188e-06,
      "loss": 1.9249,
      "step": 224352
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.4522626399993896,
      "learning_rate": 1.0201856160610754e-06,
      "loss": 1.7138,
      "step": 224353
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.4274719953536987,
      "learning_rate": 1.0198485836848015e-06,
      "loss": 2.0071,
      "step": 224354
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.4776344299316406,
      "learning_rate": 1.0195116068945963e-06,
      "loss": 1.9682,
      "step": 224355
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.608013391494751,
      "learning_rate": 1.0191746856904937e-06,
      "loss": 2.0349,
      "step": 224356
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.4740993976593018,
      "learning_rate": 1.0188378200725933e-06,
      "loss": 2.1482,
      "step": 224357
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.4925026893615723,
      "learning_rate": 1.0185010100409286e-06,
      "loss": 1.9869,
      "step": 224358
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.5342075824737549,
      "learning_rate": 1.0181642555955661e-06,
      "loss": 2.1022,
      "step": 224359
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.3914070129394531,
      "learning_rate": 1.017827556736539e-06,
      "loss": 1.8837,
      "step": 224360
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.4835964441299438,
      "learning_rate": 1.017490913463981e-06,
      "loss": 1.9079,
      "step": 224361
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.3771787881851196,
      "learning_rate": 1.0171543257778913e-06,
      "loss": 1.7754,
      "step": 224362
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.4651316404342651,
      "learning_rate": 1.0168177936783705e-06,
      "loss": 2.0638,
      "step": 224363
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.4986904859542847,
      "learning_rate": 1.0164813171654518e-06,
      "loss": 1.9962,
      "step": 224364
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.4003759622573853,
      "learning_rate": 1.0161448962392348e-06,
      "loss": 2.1639,
      "step": 224365
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.356902837753296,
      "learning_rate": 1.0158085308997532e-06,
      "loss": 2.0612,
      "step": 224366
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.4686226844787598,
      "learning_rate": 1.0154722211470733e-06,
      "loss": 1.791,
      "step": 224367
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.5547195672988892,
      "learning_rate": 1.0151359669812619e-06,
      "loss": 2.2345,
      "step": 224368
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.4083908796310425,
      "learning_rate": 1.0147997684023856e-06,
      "loss": 1.6688,
      "step": 224369
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.6130367517471313,
      "learning_rate": 1.0144636254104777e-06,
      "loss": 1.8594,
      "step": 224370
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.3207626342773438,
      "learning_rate": 1.0141275380056712e-06,
      "loss": 2.0074,
      "step": 224371
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.4282524585723877,
      "learning_rate": 1.0137915061879331e-06,
      "loss": 1.9701,
      "step": 224372
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.4004511833190918,
      "learning_rate": 1.0134555299573965e-06,
      "loss": 2.3014,
      "step": 224373
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.5535482168197632,
      "learning_rate": 1.013119609314128e-06,
      "loss": 2.1786,
      "step": 224374
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.4845818281173706,
      "learning_rate": 1.0127837442581278e-06,
      "loss": 1.8828,
      "step": 224375
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.454504370689392,
      "learning_rate": 1.0124479347894953e-06,
      "loss": 2.0042,
      "step": 224376
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.5177456140518188,
      "learning_rate": 1.0121121809082977e-06,
      "loss": 1.8834,
      "step": 224377
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.540328025817871,
      "learning_rate": 1.0117764826146014e-06,
      "loss": 2.0291,
      "step": 224378
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.448439121246338,
      "learning_rate": 1.0114408399084394e-06,
      "loss": 1.8559,
      "step": 224379
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.5226572751998901,
      "learning_rate": 1.0111052527899454e-06,
      "loss": 2.1847,
      "step": 224380
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.5210970640182495,
      "learning_rate": 1.0107697212590859e-06,
      "loss": 2.0671,
      "step": 224381
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.4751760959625244,
      "learning_rate": 1.0104342453159608e-06,
      "loss": 1.789,
      "step": 224382
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.4261549711227417,
      "learning_rate": 1.01009882496067e-06,
      "loss": 1.9989,
      "step": 224383
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.501261591911316,
      "learning_rate": 1.0097634601932136e-06,
      "loss": 2.0949,
      "step": 224384
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.4904639720916748,
      "learning_rate": 1.0094281510137247e-06,
      "loss": 2.0661,
      "step": 224385
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.4330540895462036,
      "learning_rate": 1.0090928974222035e-06,
      "loss": 2.1153,
      "step": 224386
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.4832594394683838,
      "learning_rate": 1.0087576994187495e-06,
      "loss": 1.9909,
      "step": 224387
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.5367192029953003,
      "learning_rate": 1.0084225570033965e-06,
      "loss": 2.1701,
      "step": 224388
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.3833688497543335,
      "learning_rate": 1.008087470176211e-06,
      "loss": 2.0077,
      "step": 224389
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.4381941556930542,
      "learning_rate": 1.0077524389372925e-06,
      "loss": 2.0325,
      "step": 224390
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.495016098022461,
      "learning_rate": 1.007417463286675e-06,
      "loss": 1.8504,
      "step": 224391
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.3419785499572754,
      "learning_rate": 1.0070825432243913e-06,
      "loss": 1.9347,
      "step": 224392
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.5629873275756836,
      "learning_rate": 1.0067476787505745e-06,
      "loss": 1.8744,
      "step": 224393
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.6106492280960083,
      "learning_rate": 1.006412869865225e-06,
      "loss": 1.9741,
      "step": 224394
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.6575781106948853,
      "learning_rate": 1.0060781165684428e-06,
      "loss": 2.1319,
      "step": 224395
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.4463567733764648,
      "learning_rate": 1.005743418860261e-06,
      "loss": 2.1075,
      "step": 224396
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.4084261655807495,
      "learning_rate": 1.005408776740746e-06,
      "loss": 1.9983,
      "step": 224397
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.4781508445739746,
      "learning_rate": 1.0050741902099645e-06,
      "loss": 2.1617,
      "step": 224398
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.3844891786575317,
      "learning_rate": 1.0047396592680168e-06,
      "loss": 1.8424,
      "step": 224399
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.4614630937576294,
      "learning_rate": 1.0044051839149024e-06,
      "loss": 1.9289,
      "step": 224400
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.4991815090179443,
      "learning_rate": 1.004070764150755e-06,
      "loss": 1.8473,
      "step": 224401
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.488549828529358,
      "learning_rate": 1.003736399975541e-06,
      "loss": 1.9213,
      "step": 224402
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.4364498853683472,
      "learning_rate": 1.0034020913894269e-06,
      "loss": 1.957,
      "step": 224403
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.428385615348816,
      "learning_rate": 1.0030678383923795e-06,
      "loss": 1.7125,
      "step": 224404
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.3189339637756348,
      "learning_rate": 1.0027336409845322e-06,
      "loss": 1.826,
      "step": 224405
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.4790960550308228,
      "learning_rate": 1.002399499165918e-06,
      "loss": 2.007,
      "step": 224406
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.4370548725128174,
      "learning_rate": 1.0020654129366035e-06,
      "loss": 1.8467,
      "step": 224407
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.3962881565093994,
      "learning_rate": 1.0017313822966223e-06,
      "loss": 1.8439,
      "step": 224408
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.4840201139450073,
      "learning_rate": 1.0013974072461072e-06,
      "loss": 1.9378,
      "step": 224409
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.5020140409469604,
      "learning_rate": 1.0010634877850255e-06,
      "loss": 2.1957,
      "step": 224410
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.5700711011886597,
      "learning_rate": 1.0007296239135432e-06,
      "loss": 2.2863,
      "step": 224411
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.4502360820770264,
      "learning_rate": 1.0003958156316271e-06,
      "loss": 1.8801,
      "step": 224412
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.3791931867599487,
      "learning_rate": 1.0000620629394108e-06,
      "loss": 1.8072,
      "step": 224413
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.5151808261871338,
      "learning_rate": 9.99728365836927e-07,
      "loss": 2.1386,
      "step": 224414
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.4834563732147217,
      "learning_rate": 9.993947243242095e-07,
      "loss": 2.074,
      "step": 224415
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.5214157104492188,
      "learning_rate": 9.99061138401358e-07,
      "loss": 2.0787,
      "step": 224416
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.4068719148635864,
      "learning_rate": 9.987276080684393e-07,
      "loss": 1.8165,
      "step": 224417
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.4319580793380737,
      "learning_rate": 9.983941333254864e-07,
      "loss": 1.9737,
      "step": 224418
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.4256137609481812,
      "learning_rate": 9.980607141725993e-07,
      "loss": 1.9415,
      "step": 224419
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.3863645792007446,
      "learning_rate": 9.977273506098115e-07,
      "loss": 1.8251,
      "step": 224420
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.3698793649673462,
      "learning_rate": 9.973940426371564e-07,
      "loss": 1.8673,
      "step": 224421
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.3899248838424683,
      "learning_rate": 9.970607902547668e-07,
      "loss": 2.026,
      "step": 224422
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.3678442239761353,
      "learning_rate": 9.967275934626428e-07,
      "loss": 1.8384,
      "step": 224423
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.3532320261001587,
      "learning_rate": 9.963944522608846e-07,
      "loss": 2.0679,
      "step": 224424
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.4657351970672607,
      "learning_rate": 9.960613666495588e-07,
      "loss": 2.1013,
      "step": 224425
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.4164915084838867,
      "learning_rate": 9.95728336628665e-07,
      "loss": 2.0283,
      "step": 224426
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.3307018280029297,
      "learning_rate": 9.953953621983367e-07,
      "loss": 2.0313,
      "step": 224427
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.4197397232055664,
      "learning_rate": 9.950624433586073e-07,
      "loss": 1.7761,
      "step": 224428
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.5191407203674316,
      "learning_rate": 9.947295801095433e-07,
      "loss": 1.7535,
      "step": 224429
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.4972375631332397,
      "learning_rate": 9.943967724511781e-07,
      "loss": 1.9639,
      "step": 224430
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.5183700323104858,
      "learning_rate": 9.940640203836448e-07,
      "loss": 2.1443,
      "step": 224431
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.3675146102905273,
      "learning_rate": 9.937313239069433e-07,
      "loss": 1.9129,
      "step": 224432
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.4678785800933838,
      "learning_rate": 9.933986830211405e-07,
      "loss": 2.0454,
      "step": 224433
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.4690288305282593,
      "learning_rate": 9.930660977263028e-07,
      "loss": 1.7161,
      "step": 224434
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.4957451820373535,
      "learning_rate": 9.927335680224968e-07,
      "loss": 1.8802,
      "step": 224435
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.5742143392562866,
      "learning_rate": 9.924010939097893e-07,
      "loss": 2.0126,
      "step": 224436
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.4261044263839722,
      "learning_rate": 9.920686753882468e-07,
      "loss": 1.9231,
      "step": 224437
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.5214967727661133,
      "learning_rate": 9.917363124579358e-07,
      "loss": 1.9543,
      "step": 224438
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.4643949270248413,
      "learning_rate": 9.9140400511889e-07,
      "loss": 2.0609,
      "step": 224439
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.5139286518096924,
      "learning_rate": 9.910717533711754e-07,
      "loss": 1.8048,
      "step": 224440
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.4869381189346313,
      "learning_rate": 9.907395572148924e-07,
      "loss": 1.9986,
      "step": 224441
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.6686559915542603,
      "learning_rate": 9.90407416650041e-07,
      "loss": 2.2529,
      "step": 224442
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.3738641738891602,
      "learning_rate": 9.90075331676754e-07,
      "loss": 1.852,
      "step": 224443
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.4610263109207153,
      "learning_rate": 9.89743302295032e-07,
      "loss": 1.7689,
      "step": 224444
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.468595027923584,
      "learning_rate": 9.89411328504941e-07,
      "loss": 1.8784,
      "step": 224445
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.4023654460906982,
      "learning_rate": 9.890794103065813e-07,
      "loss": 2.0206,
      "step": 224446
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.43533194065094,
      "learning_rate": 9.887475477000195e-07,
      "loss": 1.9683,
      "step": 224447
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.442939281463623,
      "learning_rate": 9.884157406852554e-07,
      "loss": 1.866,
      "step": 224448
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.5171401500701904,
      "learning_rate": 9.880839892624226e-07,
      "loss": 2.1117,
      "step": 224449
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.5032973289489746,
      "learning_rate": 9.877522934315541e-07,
      "loss": 2.0156,
      "step": 224450
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.4604332447052002,
      "learning_rate": 9.874206531926832e-07,
      "loss": 2.0162,
      "step": 224451
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.3645610809326172,
      "learning_rate": 9.870890685458765e-07,
      "loss": 2.0243,
      "step": 224452
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.538649320602417,
      "learning_rate": 9.867575394912675e-07,
      "loss": 2.0958,
      "step": 224453
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.6681779623031616,
      "learning_rate": 9.864260660288225e-07,
      "loss": 1.8476,
      "step": 224454
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.48416006565094,
      "learning_rate": 9.86094648158642e-07,
      "loss": 1.7986,
      "step": 224455
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.5798721313476562,
      "learning_rate": 9.857632858808252e-07,
      "loss": 2.0464,
      "step": 224456
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.4646302461624146,
      "learning_rate": 9.854319791953724e-07,
      "loss": 1.9783,
      "step": 224457
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.3583970069885254,
      "learning_rate": 9.851007281023836e-07,
      "loss": 1.8526,
      "step": 224458
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.509855031967163,
      "learning_rate": 9.847695326019256e-07,
      "loss": 2.07,
      "step": 224459
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.6214619874954224,
      "learning_rate": 9.84438392693998e-07,
      "loss": 1.9616,
      "step": 224460
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.4445780515670776,
      "learning_rate": 9.841073083787343e-07,
      "loss": 1.6777,
      "step": 224461
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.3865947723388672,
      "learning_rate": 9.837762796561677e-07,
      "loss": 1.9684,
      "step": 224462
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.366466760635376,
      "learning_rate": 9.834453065263647e-07,
      "loss": 2.0302,
      "step": 224463
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.4153811931610107,
      "learning_rate": 9.831143889893921e-07,
      "loss": 2.1275,
      "step": 224464
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.406237006187439,
      "learning_rate": 9.827835270452832e-07,
      "loss": 1.921,
      "step": 224465
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.425288200378418,
      "learning_rate": 9.82452720694138e-07,
      "loss": 2.2496,
      "step": 224466
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.415138602256775,
      "learning_rate": 9.821219699359895e-07,
      "loss": 1.8381,
      "step": 224467
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.4402300119400024,
      "learning_rate": 9.817912747709046e-07,
      "loss": 2.0465,
      "step": 224468
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.4511574506759644,
      "learning_rate": 9.814606351989495e-07,
      "loss": 1.8047,
      "step": 224469
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.4012166261672974,
      "learning_rate": 9.811300512201914e-07,
      "loss": 2.0662,
      "step": 224470
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.5514817237854004,
      "learning_rate": 9.807995228346965e-07,
      "loss": 1.8669,
      "step": 224471
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.579580307006836,
      "learning_rate": 9.804690500424983e-07,
      "loss": 1.9507,
      "step": 224472
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.4047881364822388,
      "learning_rate": 9.801386328436633e-07,
      "loss": 1.9655,
      "step": 224473
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.345845103263855,
      "learning_rate": 9.798082712382915e-07,
      "loss": 1.6744,
      "step": 224474
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.4834074974060059,
      "learning_rate": 9.794779652264163e-07,
      "loss": 1.8713,
      "step": 224475
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.5078413486480713,
      "learning_rate": 9.791477148081039e-07,
      "loss": 1.9396,
      "step": 224476
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.394574522972107,
      "learning_rate": 9.78817519983388e-07,
      "loss": 2.1045,
      "step": 224477
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.5108548402786255,
      "learning_rate": 9.784873807523685e-07,
      "loss": 2.1336,
      "step": 224478
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.5187183618545532,
      "learning_rate": 9.781572971150785e-07,
      "loss": 2.148,
      "step": 224479
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.4651325941085815,
      "learning_rate": 9.778272690716182e-07,
      "loss": 2.0849,
      "step": 224480
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.5414000749588013,
      "learning_rate": 9.774972966219874e-07,
      "loss": 1.7363,
      "step": 224481
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.4653749465942383,
      "learning_rate": 9.771673797663193e-07,
      "loss": 2.0479,
      "step": 224482
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.490846872329712,
      "learning_rate": 9.768375185046472e-07,
      "loss": 1.932,
      "step": 224483
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.482077717781067,
      "learning_rate": 9.765077128370046e-07,
      "loss": 2.0982,
      "step": 224484
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.479533314704895,
      "learning_rate": 9.76177962763458e-07,
      "loss": 2.0806,
      "step": 224485
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.5424785614013672,
      "learning_rate": 9.758482682841074e-07,
      "loss": 1.8005,
      "step": 224486
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.50770902633667,
      "learning_rate": 9.75518629398986e-07,
      "loss": 2.0045,
      "step": 224487
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.5381252765655518,
      "learning_rate": 9.751890461081603e-07,
      "loss": 1.9723,
      "step": 224488
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.5002201795578003,
      "learning_rate": 9.74859518411697e-07,
      "loss": 1.7452,
      "step": 224489
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.4776431322097778,
      "learning_rate": 9.745300463096628e-07,
      "loss": 2.125,
      "step": 224490
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.4431555271148682,
      "learning_rate": 9.74200629802091e-07,
      "loss": 1.8962,
      "step": 224491
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.4721639156341553,
      "learning_rate": 9.738712688890482e-07,
      "loss": 1.9985,
      "step": 224492
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.5442144870758057,
      "learning_rate": 9.735419635706344e-07,
      "loss": 2.0418,
      "step": 224493
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.5589436292648315,
      "learning_rate": 9.732127138468825e-07,
      "loss": 1.994,
      "step": 224494
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.6191767454147339,
      "learning_rate": 9.728835197178264e-07,
      "loss": 1.9079,
      "step": 224495
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.4342916011810303,
      "learning_rate": 9.725543811835656e-07,
      "loss": 2.0318,
      "step": 224496
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.473142147064209,
      "learning_rate": 9.72225298244167e-07,
      "loss": 1.9461,
      "step": 224497
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.5295865535736084,
      "learning_rate": 9.718962708996635e-07,
      "loss": 1.9017,
      "step": 224498
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.5539517402648926,
      "learning_rate": 9.715672991501222e-07,
      "loss": 2.2637,
      "step": 224499
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.4894479513168335,
      "learning_rate": 9.712383829956428e-07,
      "loss": 2.0239,
      "step": 224500
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.4971901178359985,
      "learning_rate": 9.70909522436225e-07,
      "loss": 2.2233,
      "step": 224501
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.4258302450180054,
      "learning_rate": 9.705807174719692e-07,
      "loss": 1.8726,
      "step": 224502
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.584646463394165,
      "learning_rate": 9.702519681029087e-07,
      "loss": 1.8705,
      "step": 224503
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.458938479423523,
      "learning_rate": 9.699232743291762e-07,
      "loss": 2.1881,
      "step": 224504
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.5074517726898193,
      "learning_rate": 9.695946361507056e-07,
      "loss": 1.99,
      "step": 224505
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.5916329622268677,
      "learning_rate": 9.692660535676967e-07,
      "loss": 1.9479,
      "step": 224506
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.4482725858688354,
      "learning_rate": 9.689375265801158e-07,
      "loss": 2.1924,
      "step": 224507
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.4527535438537598,
      "learning_rate": 9.686090551880632e-07,
      "loss": 1.8793,
      "step": 224508
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.418029546737671,
      "learning_rate": 9.68280639391572e-07,
      "loss": 1.9397,
      "step": 224509
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.6657335758209229,
      "learning_rate": 9.679522791907424e-07,
      "loss": 2.0457,
      "step": 224510
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.4934090375900269,
      "learning_rate": 9.676239745856074e-07,
      "loss": 2.1648,
      "step": 224511
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.4937649965286255,
      "learning_rate": 9.672957255762336e-07,
      "loss": 1.9884,
      "step": 224512
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.3713996410369873,
      "learning_rate": 9.66967532162688e-07,
      "loss": 1.808,
      "step": 224513
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.5927921533584595,
      "learning_rate": 9.666393943450367e-07,
      "loss": 1.8881,
      "step": 224514
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.4386905431747437,
      "learning_rate": 9.663113121233135e-07,
      "loss": 1.8121,
      "step": 224515
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.4692442417144775,
      "learning_rate": 9.659832854976179e-07,
      "loss": 2.1055,
      "step": 224516
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.3800792694091797,
      "learning_rate": 9.656553144679502e-07,
      "loss": 2.0676,
      "step": 224517
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.528551697731018,
      "learning_rate": 9.653273990344434e-07,
      "loss": 1.9244,
      "step": 224518
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.3628207445144653,
      "learning_rate": 9.64999539197131e-07,
      "loss": 2.0109,
      "step": 224519
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.3617541790008545,
      "learning_rate": 9.646717349560796e-07,
      "loss": 2.0175,
      "step": 224520
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.4471065998077393,
      "learning_rate": 9.643439863113222e-07,
      "loss": 1.8515,
      "step": 224521
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.5825473070144653,
      "learning_rate": 9.64016293262926e-07,
      "loss": 2.0086,
      "step": 224522
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.4856079816818237,
      "learning_rate": 9.636886558109902e-07,
      "loss": 1.9041,
      "step": 224523
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.4944348335266113,
      "learning_rate": 9.633610739555486e-07,
      "loss": 2.0037,
      "step": 224524
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.4511361122131348,
      "learning_rate": 9.630335476966344e-07,
      "loss": 1.9862,
      "step": 224525
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.5034615993499756,
      "learning_rate": 9.627060770343476e-07,
      "loss": 2.0976,
      "step": 224526
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.5421537160873413,
      "learning_rate": 9.623786619687546e-07,
      "loss": 1.8325,
      "step": 224527
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.5239448547363281,
      "learning_rate": 9.620513024998888e-07,
      "loss": 2.0005,
      "step": 224528
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.4911770820617676,
      "learning_rate": 9.617239986278169e-07,
      "loss": 2.0402,
      "step": 224529
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.4814999103546143,
      "learning_rate": 9.613967503526388e-07,
      "loss": 2.0071,
      "step": 224530
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.3691670894622803,
      "learning_rate": 9.61069557674321e-07,
      "loss": 2.0331,
      "step": 224531
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.4209152460098267,
      "learning_rate": 9.607424205930303e-07,
      "loss": 2.0965,
      "step": 224532
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.4574756622314453,
      "learning_rate": 9.604153391087667e-07,
      "loss": 1.8923,
      "step": 224533
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.3737890720367432,
      "learning_rate": 9.600883132216298e-07,
      "loss": 1.9139,
      "step": 224534
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.4668742418289185,
      "learning_rate": 9.597613429316198e-07,
      "loss": 1.992,
      "step": 224535
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.3962589502334595,
      "learning_rate": 9.594344282388699e-07,
      "loss": 1.8915,
      "step": 224536
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.566193699836731,
      "learning_rate": 9.591075691433803e-07,
      "loss": 2.1112,
      "step": 224537
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.65061616897583,
      "learning_rate": 9.587807656452507e-07,
      "loss": 1.9486,
      "step": 224538
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.5535967350006104,
      "learning_rate": 9.584540177445144e-07,
      "loss": 2.0608,
      "step": 224539
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.466646671295166,
      "learning_rate": 9.581273254412715e-07,
      "loss": 2.1,
      "step": 224540
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.432482361793518,
      "learning_rate": 9.578006887355549e-07,
      "loss": 2.1308,
      "step": 224541
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.4008325338363647,
      "learning_rate": 9.574741076273985e-07,
      "loss": 2.0471,
      "step": 224542
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.4194066524505615,
      "learning_rate": 9.571475821169017e-07,
      "loss": 1.9765,
      "step": 224543
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.4595558643341064,
      "learning_rate": 9.568211122041314e-07,
      "loss": 2.0348,
      "step": 224544
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.457322359085083,
      "learning_rate": 9.564946978891207e-07,
      "loss": 2.0026,
      "step": 224545
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.5741162300109863,
      "learning_rate": 9.561683391719698e-07,
      "loss": 1.8043,
      "step": 224546
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.394205927848816,
      "learning_rate": 9.558420360526787e-07,
      "loss": 1.6844,
      "step": 224547
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.503597617149353,
      "learning_rate": 9.555157885313803e-07,
      "loss": 2.037,
      "step": 224548
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.543172001838684,
      "learning_rate": 9.551895966080415e-07,
      "loss": 1.9898,
      "step": 224549
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.8636608123779297,
      "learning_rate": 9.548634602828287e-07,
      "loss": 1.9648,
      "step": 224550
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.582988977432251,
      "learning_rate": 9.545373795557088e-07,
      "loss": 1.9104,
      "step": 224551
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.4433506727218628,
      "learning_rate": 9.542113544268148e-07,
      "loss": 1.9138,
      "step": 224552
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.4420361518859863,
      "learning_rate": 9.538853848961804e-07,
      "loss": 1.806,
      "step": 224553
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.4761569499969482,
      "learning_rate": 9.535594709638384e-07,
      "loss": 1.8457,
      "step": 224554
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.3729469776153564,
      "learning_rate": 9.53233612629889e-07,
      "loss": 1.8354,
      "step": 224555
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.4238721132278442,
      "learning_rate": 9.529078098943654e-07,
      "loss": 1.8679,
      "step": 224556
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.6930439472198486,
      "learning_rate": 9.525820627573677e-07,
      "loss": 1.9461,
      "step": 224557
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.4587849378585815,
      "learning_rate": 9.522563712188958e-07,
      "loss": 1.7257,
      "step": 224558
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.5131666660308838,
      "learning_rate": 9.519307352790828e-07,
      "loss": 1.9623,
      "step": 224559
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.471591830253601,
      "learning_rate": 9.516051549379289e-07,
      "loss": 2.1527,
      "step": 224560
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.4179418087005615,
      "learning_rate": 9.512796301955339e-07,
      "loss": 1.717,
      "step": 224561
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.4062249660491943,
      "learning_rate": 9.509541610518978e-07,
      "loss": 1.8988,
      "step": 224562
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.4519520998001099,
      "learning_rate": 9.506287475071539e-07,
      "loss": 1.7921,
      "step": 224563
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.662007212638855,
      "learning_rate": 9.503033895613354e-07,
      "loss": 1.9493,
      "step": 224564
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.6233234405517578,
      "learning_rate": 9.499780872145091e-07,
      "loss": 2.0192,
      "step": 224565
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.4193439483642578,
      "learning_rate": 9.496528404667081e-07,
      "loss": 2.0548,
      "step": 224566
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.463046908378601,
      "learning_rate": 9.493276493180324e-07,
      "loss": 2.0339,
      "step": 224567
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.4471380710601807,
      "learning_rate": 9.49002513768482e-07,
      "loss": 1.8387,
      "step": 224568
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.5051329135894775,
      "learning_rate": 9.486774338181902e-07,
      "loss": 1.9699,
      "step": 224569
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.4608933925628662,
      "learning_rate": 9.483524094671902e-07,
      "loss": 1.8566,
      "step": 224570
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.5573492050170898,
      "learning_rate": 9.480274407155153e-07,
      "loss": 1.8428,
      "step": 224571
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.5109238624572754,
      "learning_rate": 9.477025275632655e-07,
      "loss": 1.9002,
      "step": 224572
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.546654462814331,
      "learning_rate": 9.47377670010474e-07,
      "loss": 1.9002,
      "step": 224573
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.5965213775634766,
      "learning_rate": 9.470528680572076e-07,
      "loss": 2.0836,
      "step": 224574
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.3934130668640137,
      "learning_rate": 9.467281217035328e-07,
      "loss": 1.7769,
      "step": 224575
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.482523798942566,
      "learning_rate": 9.464034309494828e-07,
      "loss": 1.8263,
      "step": 224576
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.6864548921585083,
      "learning_rate": 9.46078795795191e-07,
      "loss": 2.1823,
      "step": 224577
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.4927189350128174,
      "learning_rate": 9.45754216240624e-07,
      "loss": 1.9628,
      "step": 224578
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.4973678588867188,
      "learning_rate": 9.45429692285915e-07,
      "loss": 1.949,
      "step": 224579
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.5269373655319214,
      "learning_rate": 9.451052239310642e-07,
      "loss": 2.106,
      "step": 224580
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.5763996839523315,
      "learning_rate": 9.447808111762046e-07,
      "loss": 1.8771,
      "step": 224581
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.3648607730865479,
      "learning_rate": 9.444564540213029e-07,
      "loss": 1.9525,
      "step": 224582
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.628969669342041,
      "learning_rate": 9.441321524665258e-07,
      "loss": 2.1125,
      "step": 224583
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.4244285821914673,
      "learning_rate": 9.438079065118398e-07,
      "loss": 2.1607,
      "step": 224584
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.6556392908096313,
      "learning_rate": 9.434837161573783e-07,
      "loss": 2.059,
      "step": 224585
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.3912395238876343,
      "learning_rate": 9.431595814031411e-07,
      "loss": 1.951,
      "step": 224586
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.4337648153305054,
      "learning_rate": 9.428355022492285e-07,
      "loss": 1.7578,
      "step": 224587
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.645864725112915,
      "learning_rate": 9.425114786956733e-07,
      "loss": 2.1071,
      "step": 224588
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.4380824565887451,
      "learning_rate": 9.421875107425759e-07,
      "loss": 1.7672,
      "step": 224589
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.5304077863693237,
      "learning_rate": 9.418635983899691e-07,
      "loss": 2.0031,
      "step": 224590
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.4121983051300049,
      "learning_rate": 9.415397416378867e-07,
      "loss": 1.8366,
      "step": 224591
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.4586836099624634,
      "learning_rate": 9.412159404864284e-07,
      "loss": 1.9742,
      "step": 224592
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.5182766914367676,
      "learning_rate": 9.408921949356607e-07,
      "loss": 1.8287,
      "step": 224593
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.3686984777450562,
      "learning_rate": 9.405685049856171e-07,
      "loss": 2.1443,
      "step": 224594
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.4552159309387207,
      "learning_rate": 9.402448706363641e-07,
      "loss": 2.16,
      "step": 224595
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.4817533493041992,
      "learning_rate": 9.399212918879684e-07,
      "loss": 1.8873,
      "step": 224596
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.494601845741272,
      "learning_rate": 9.395977687404966e-07,
      "loss": 1.8962,
      "step": 224597
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.6726738214492798,
      "learning_rate": 9.392743011939485e-07,
      "loss": 2.0574,
      "step": 224598
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.5370668172836304,
      "learning_rate": 9.38950889248491e-07,
      "loss": 1.9312,
      "step": 224599
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.7292054891586304,
      "learning_rate": 9.386275329040904e-07,
      "loss": 1.8377,
      "step": 224600
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.4572914838790894,
      "learning_rate": 9.383042321608802e-07,
      "loss": 2.1243,
      "step": 224601
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.4185380935668945,
      "learning_rate": 9.379809870188604e-07,
      "loss": 1.9899,
      "step": 224602
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.357415795326233,
      "learning_rate": 9.376577974781308e-07,
      "loss": 2.0556,
      "step": 224603
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.4044477939605713,
      "learning_rate": 9.373346635387247e-07,
      "loss": 2.0746,
      "step": 224604
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.4667285680770874,
      "learning_rate": 9.370115852007088e-07,
      "loss": 1.8848,
      "step": 224605
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.358573317527771,
      "learning_rate": 9.366885624641496e-07,
      "loss": 2.0509,
      "step": 224606
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.4670456647872925,
      "learning_rate": 9.363655953291138e-07,
      "loss": 2.1049,
      "step": 224607
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.4574085474014282,
      "learning_rate": 9.360426837956347e-07,
      "loss": 2.0725,
      "step": 224608
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.7186120748519897,
      "learning_rate": 9.357198278638123e-07,
      "loss": 2.0366,
      "step": 224609
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.4863053560256958,
      "learning_rate": 9.353970275336798e-07,
      "loss": 2.1492,
      "step": 224610
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.5223592519760132,
      "learning_rate": 9.350742828053037e-07,
      "loss": 2.1144,
      "step": 224611
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.400797724723816,
      "learning_rate": 9.347515936787175e-07,
      "loss": 1.7949,
      "step": 224612
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.4393302202224731,
      "learning_rate": 9.344289601540545e-07,
      "loss": 1.8886,
      "step": 224613
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.4579005241394043,
      "learning_rate": 9.341063822312811e-07,
      "loss": 1.9937,
      "step": 224614
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.5345948934555054,
      "learning_rate": 9.337838599105307e-07,
      "loss": 2.039,
      "step": 224615
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.3754112720489502,
      "learning_rate": 9.334613931918034e-07,
      "loss": 2.1165,
      "step": 224616
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.389253854751587,
      "learning_rate": 9.331389820752322e-07,
      "loss": 1.9836,
      "step": 224617
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.4388560056686401,
      "learning_rate": 9.328166265608173e-07,
      "loss": 2.0086,
      "step": 224618
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.4623581171035767,
      "learning_rate": 9.324943266486251e-07,
      "loss": 1.8341,
      "step": 224619
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.467095136642456,
      "learning_rate": 9.321720823387557e-07,
      "loss": 1.8,
      "step": 224620
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.39407217502594,
      "learning_rate": 9.318498936312091e-07,
      "loss": 2.0403,
      "step": 224621
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.4343122243881226,
      "learning_rate": 9.315277605260852e-07,
      "loss": 1.6771,
      "step": 224622
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.3899734020233154,
      "learning_rate": 9.312056830234505e-07,
      "loss": 1.9869,
      "step": 224623
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.4996962547302246,
      "learning_rate": 9.308836611233383e-07,
      "loss": 2.0175,
      "step": 224624
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.4432793855667114,
      "learning_rate": 9.305616948258154e-07,
      "loss": 1.9671,
      "step": 224625
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.4006762504577637,
      "learning_rate": 9.302397841309483e-07,
      "loss": 2.0236,
      "step": 224626
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.4971388578414917,
      "learning_rate": 9.299179290388036e-07,
      "loss": 2.0708,
      "step": 224627
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.4446626901626587,
      "learning_rate": 9.29596129549448e-07,
      "loss": 1.8956,
      "step": 224628
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.632489800453186,
      "learning_rate": 9.292743856629148e-07,
      "loss": 2.0543,
      "step": 224629
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.5027862787246704,
      "learning_rate": 9.289526973792371e-07,
      "loss": 1.9756,
      "step": 224630
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.4407882690429688,
      "learning_rate": 9.286310646985484e-07,
      "loss": 1.6933,
      "step": 224631
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.8335261344909668,
      "learning_rate": 9.283094876208819e-07,
      "loss": 1.8049,
      "step": 224632
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.5989596843719482,
      "learning_rate": 9.279879661462708e-07,
      "loss": 2.2349,
      "step": 224633
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.400139331817627,
      "learning_rate": 9.276665002747819e-07,
      "loss": 2.1103,
      "step": 224634
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.6094130277633667,
      "learning_rate": 9.273450900065149e-07,
      "loss": 1.9357,
      "step": 224635
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.4129784107208252,
      "learning_rate": 9.2702373534147e-07,
      "loss": 1.8864,
      "step": 224636
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.5145998001098633,
      "learning_rate": 9.26702436279747e-07,
      "loss": 1.8242,
      "step": 224637
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.4093564748764038,
      "learning_rate": 9.263811928214126e-07,
      "loss": 1.8741,
      "step": 224638
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.6134477853775024,
      "learning_rate": 9.260600049664668e-07,
      "loss": 2.082,
      "step": 224639
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.5288493633270264,
      "learning_rate": 9.257388727150427e-07,
      "loss": 2.0582,
      "step": 224640
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.456153392791748,
      "learning_rate": 9.254177960671738e-07,
      "loss": 2.1784,
      "step": 224641
    },
    {
      "epoch": 2.92,
      "grad_norm": 1.5127426385879517,
      "learning_rate": 9.250967750229265e-07,
      "loss": 1.9302,
      "step": 224642
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.5536611080169678,
      "learning_rate": 9.247758095823343e-07,
      "loss": 2.1407,
      "step": 224643
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.526919960975647,
      "learning_rate": 9.244548997454637e-07,
      "loss": 1.99,
      "step": 224644
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.5499367713928223,
      "learning_rate": 9.241340455123814e-07,
      "loss": 1.8594,
      "step": 224645
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.3567954301834106,
      "learning_rate": 9.238132468831539e-07,
      "loss": 1.8618,
      "step": 224646
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.4381120204925537,
      "learning_rate": 9.234925038578145e-07,
      "loss": 1.9995,
      "step": 224647
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.422571063041687,
      "learning_rate": 9.231718164364632e-07,
      "loss": 1.9538,
      "step": 224648
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.447119116783142,
      "learning_rate": 9.228511846191666e-07,
      "loss": 2.0123,
      "step": 224649
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.4236375093460083,
      "learning_rate": 9.225306084059248e-07,
      "loss": 1.8806,
      "step": 224650
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.3828474283218384,
      "learning_rate": 9.222100877968375e-07,
      "loss": 1.8916,
      "step": 224651
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.5129461288452148,
      "learning_rate": 9.218896227919381e-07,
      "loss": 2.1127,
      "step": 224652
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.5876507759094238,
      "learning_rate": 9.215692133913266e-07,
      "loss": 2.0834,
      "step": 224653
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.4837300777435303,
      "learning_rate": 9.212488595950363e-07,
      "loss": 2.1498,
      "step": 224654
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.3622440099716187,
      "learning_rate": 9.209285614031337e-07,
      "loss": 1.9718,
      "step": 224655
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.414648413658142,
      "learning_rate": 9.206083188156854e-07,
      "loss": 1.899,
      "step": 224656
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.5193973779678345,
      "learning_rate": 9.202881318327249e-07,
      "loss": 2.0589,
      "step": 224657
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.8814548254013062,
      "learning_rate": 9.199680004543187e-07,
      "loss": 2.0462,
      "step": 224658
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.6835929155349731,
      "learning_rate": 9.196479246805666e-07,
      "loss": 2.1305,
      "step": 224659
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.3682247400283813,
      "learning_rate": 9.193279045114688e-07,
      "loss": 1.8684,
      "step": 224660
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.4698588848114014,
      "learning_rate": 9.19007939947125e-07,
      "loss": 1.907,
      "step": 224661
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.41641366481781,
      "learning_rate": 9.186880309876022e-07,
      "loss": 1.8715,
      "step": 224662
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.4892401695251465,
      "learning_rate": 9.183681776329332e-07,
      "loss": 1.865,
      "step": 224663
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.4985400438308716,
      "learning_rate": 9.180483798831517e-07,
      "loss": 1.8703,
      "step": 224664
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.8798776865005493,
      "learning_rate": 9.177286377383907e-07,
      "loss": 2.09,
      "step": 224665
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.385049819946289,
      "learning_rate": 9.174089511986505e-07,
      "loss": 2.228,
      "step": 224666
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.8250051736831665,
      "learning_rate": 9.170893202639973e-07,
      "loss": 1.9785,
      "step": 224667
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.5904591083526611,
      "learning_rate": 9.167697449345312e-07,
      "loss": 2.1185,
      "step": 224668
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.635483980178833,
      "learning_rate": 9.164502252102523e-07,
      "loss": 1.9823,
      "step": 224669
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.5081672668457031,
      "learning_rate": 9.161307610912938e-07,
      "loss": 2.0859,
      "step": 224670
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.4667699337005615,
      "learning_rate": 9.158113525776222e-07,
      "loss": 1.8276,
      "step": 224671
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.431047797203064,
      "learning_rate": 9.154919996694044e-07,
      "loss": 2.0269,
      "step": 224672
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.6265689134597778,
      "learning_rate": 9.151727023665733e-07,
      "loss": 2.0402,
      "step": 224673
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.6036326885223389,
      "learning_rate": 9.148534606692959e-07,
      "loss": 2.1098,
      "step": 224674
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.4264194965362549,
      "learning_rate": 9.145342745776052e-07,
      "loss": 1.9239,
      "step": 224675
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.399367332458496,
      "learning_rate": 9.142151440915347e-07,
      "loss": 1.9328,
      "step": 224676
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.4659266471862793,
      "learning_rate": 9.138960692111508e-07,
      "loss": 2.0795,
      "step": 224677
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.4475115537643433,
      "learning_rate": 9.135770499365203e-07,
      "loss": 2.0373,
      "step": 224678
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.4237380027770996,
      "learning_rate": 9.132580862677096e-07,
      "loss": 2.0725,
      "step": 224679
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.496800184249878,
      "learning_rate": 9.129391782047857e-07,
      "loss": 1.6928,
      "step": 224680
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.4798479080200195,
      "learning_rate": 9.126203257477483e-07,
      "loss": 2.0,
      "step": 224681
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.4695225954055786,
      "learning_rate": 9.123015288967306e-07,
      "loss": 1.7659,
      "step": 224682
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.4219214916229248,
      "learning_rate": 9.119827876517661e-07,
      "loss": 1.9567,
      "step": 224683
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.3647083044052124,
      "learning_rate": 9.116641020128878e-07,
      "loss": 1.9758,
      "step": 224684
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.4438121318817139,
      "learning_rate": 9.11345471980196e-07,
      "loss": 1.9653,
      "step": 224685
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.5024298429489136,
      "learning_rate": 9.110268975537239e-07,
      "loss": 1.7291,
      "step": 224686
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.4115737676620483,
      "learning_rate": 9.107083787335379e-07,
      "loss": 2.1319,
      "step": 224687
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.3637558221817017,
      "learning_rate": 9.103899155197047e-07,
      "loss": 1.837,
      "step": 224688
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.5993543863296509,
      "learning_rate": 9.100715079122578e-07,
      "loss": 1.9493,
      "step": 224689
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.4957503080368042,
      "learning_rate": 9.09753155911297e-07,
      "loss": 2.0293,
      "step": 224690
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.5534998178482056,
      "learning_rate": 9.094348595168222e-07,
      "loss": 1.7979,
      "step": 224691
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.3501439094543457,
      "learning_rate": 9.091166187289667e-07,
      "loss": 1.8946,
      "step": 224692
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.4207496643066406,
      "learning_rate": 9.087984335477305e-07,
      "loss": 1.8524,
      "step": 224693
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.5017627477645874,
      "learning_rate": 9.084803039732136e-07,
      "loss": 2.2295,
      "step": 224694
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.6723766326904297,
      "learning_rate": 9.081622300054492e-07,
      "loss": 1.9031,
      "step": 224695
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.6749956607818604,
      "learning_rate": 9.07844211644504e-07,
      "loss": 2.0088,
      "step": 224696
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.470323920249939,
      "learning_rate": 9.075262488904111e-07,
      "loss": 1.8823,
      "step": 224697
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.4188730716705322,
      "learning_rate": 9.072083417432707e-07,
      "loss": 1.7965,
      "step": 224698
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.5744028091430664,
      "learning_rate": 9.068904902031493e-07,
      "loss": 2.1913,
      "step": 224699
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.3396564722061157,
      "learning_rate": 9.065726942700801e-07,
      "loss": 1.7379,
      "step": 224700
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.3256735801696777,
      "learning_rate": 9.062549539440966e-07,
      "loss": 1.8912,
      "step": 224701
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.5272190570831299,
      "learning_rate": 9.059372692252987e-07,
      "loss": 2.0917,
      "step": 224702
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.4418272972106934,
      "learning_rate": 9.056196401137194e-07,
      "loss": 2.0128,
      "step": 224703
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.476011037826538,
      "learning_rate": 9.053020666094591e-07,
      "loss": 1.9181,
      "step": 224704
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.596275806427002,
      "learning_rate": 9.049845487125173e-07,
      "loss": 1.9099,
      "step": 224705
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.3967574834823608,
      "learning_rate": 9.046670864230277e-07,
      "loss": 1.9781,
      "step": 224706
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.479474663734436,
      "learning_rate": 9.043496797409567e-07,
      "loss": 1.8825,
      "step": 224707
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.5664145946502686,
      "learning_rate": 9.040323286664375e-07,
      "loss": 1.9779,
      "step": 224708
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.4626553058624268,
      "learning_rate": 9.037150331995036e-07,
      "loss": 1.8036,
      "step": 224709
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.6528346538543701,
      "learning_rate": 9.033977933402214e-07,
      "loss": 1.9135,
      "step": 224710
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.4945547580718994,
      "learning_rate": 9.030806090886245e-07,
      "loss": 1.8333,
      "step": 224711
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.4083374738693237,
      "learning_rate": 9.027634804448125e-07,
      "loss": 2.1855,
      "step": 224712
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.4835399389266968,
      "learning_rate": 9.024464074088189e-07,
      "loss": 1.901,
      "step": 224713
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.4768352508544922,
      "learning_rate": 9.021293899807103e-07,
      "loss": 1.7867,
      "step": 224714
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.5297054052352905,
      "learning_rate": 9.018124281605199e-07,
      "loss": 2.1582,
      "step": 224715
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.4803571701049805,
      "learning_rate": 9.014955219483477e-07,
      "loss": 1.8112,
      "step": 224716
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.543489694595337,
      "learning_rate": 9.011786713442271e-07,
      "loss": 2.0957,
      "step": 224717
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.5241631269454956,
      "learning_rate": 9.008618763482245e-07,
      "loss": 2.0868,
      "step": 224718
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.4632307291030884,
      "learning_rate": 9.005451369603733e-07,
      "loss": 2.0212,
      "step": 224719
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.4117802381515503,
      "learning_rate": 9.002284531808069e-07,
      "loss": 1.9067,
      "step": 224720
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.5408480167388916,
      "learning_rate": 8.999118250094917e-07,
      "loss": 2.202,
      "step": 224721
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.3048323392868042,
      "learning_rate": 8.995952524465278e-07,
      "loss": 1.7608,
      "step": 224722
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.434054970741272,
      "learning_rate": 8.99278735492015e-07,
      "loss": 1.9647,
      "step": 224723
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.325922966003418,
      "learning_rate": 8.989622741459201e-07,
      "loss": 2.0265,
      "step": 224724
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.4505410194396973,
      "learning_rate": 8.986458684084097e-07,
      "loss": 2.063,
      "step": 224725
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.4419467449188232,
      "learning_rate": 8.983295182794503e-07,
      "loss": 1.9322,
      "step": 224726
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.4214001893997192,
      "learning_rate": 8.98013223759142e-07,
      "loss": 2.0091,
      "step": 224727
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.7014954090118408,
      "learning_rate": 8.976969848475513e-07,
      "loss": 2.0203,
      "step": 224728
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.4034076929092407,
      "learning_rate": 8.973808015447114e-07,
      "loss": 2.0449,
      "step": 224729
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.3531558513641357,
      "learning_rate": 8.970646738506893e-07,
      "loss": 1.8951,
      "step": 224730
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.346035361289978,
      "learning_rate": 8.967486017655845e-07,
      "loss": 1.9188,
      "step": 224731
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.5494154691696167,
      "learning_rate": 8.964325852893972e-07,
      "loss": 2.0499,
      "step": 224732
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.4785577058792114,
      "learning_rate": 8.96116624422194e-07,
      "loss": 2.0412,
      "step": 224733
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.6828360557556152,
      "learning_rate": 8.958007191640748e-07,
      "loss": 2.0739,
      "step": 224734
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.3696742057800293,
      "learning_rate": 8.954848695150397e-07,
      "loss": 2.0428,
      "step": 224735
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.5106555223464966,
      "learning_rate": 8.951690754751883e-07,
      "loss": 2.0189,
      "step": 224736
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.561840534210205,
      "learning_rate": 8.948533370445876e-07,
      "loss": 1.9017,
      "step": 224737
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.462345004081726,
      "learning_rate": 8.945376542232708e-07,
      "loss": 1.7975,
      "step": 224738
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.6195485591888428,
      "learning_rate": 8.942220270113376e-07,
      "loss": 1.9338,
      "step": 224739
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.3680499792099,
      "learning_rate": 8.939064554087882e-07,
      "loss": 1.9913,
      "step": 224740
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.4325599670410156,
      "learning_rate": 8.935909394156892e-07,
      "loss": 1.9167,
      "step": 224741
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.406125545501709,
      "learning_rate": 8.932754790321405e-07,
      "loss": 1.8809,
      "step": 224742
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.5172240734100342,
      "learning_rate": 8.929600742581755e-07,
      "loss": 1.9496,
      "step": 224743
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.5439003705978394,
      "learning_rate": 8.926447250938273e-07,
      "loss": 2.0617,
      "step": 224744
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.4597218036651611,
      "learning_rate": 8.923294315392293e-07,
      "loss": 1.8343,
      "step": 224745
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.4580285549163818,
      "learning_rate": 8.920141935943481e-07,
      "loss": 1.9177,
      "step": 224746
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.4083201885223389,
      "learning_rate": 8.91699011259317e-07,
      "loss": 1.9927,
      "step": 224747
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.3964539766311646,
      "learning_rate": 8.913838845341692e-07,
      "loss": 1.9755,
      "step": 224748
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.5560778379440308,
      "learning_rate": 8.910688134189381e-07,
      "loss": 1.9591,
      "step": 224749
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.4254781007766724,
      "learning_rate": 8.907537979137236e-07,
      "loss": 2.0312,
      "step": 224750
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.4679783582687378,
      "learning_rate": 8.90438838018559e-07,
      "loss": 1.788,
      "step": 224751
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.3496880531311035,
      "learning_rate": 8.901239337334775e-07,
      "loss": 2.2057,
      "step": 224752
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.5424760580062866,
      "learning_rate": 8.898090850586126e-07,
      "loss": 1.9232,
      "step": 224753
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.4341833591461182,
      "learning_rate": 8.89494291993964e-07,
      "loss": 1.9032,
      "step": 224754
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.335132122039795,
      "learning_rate": 8.891795545395986e-07,
      "loss": 1.9191,
      "step": 224755
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.4657374620437622,
      "learning_rate": 8.888648726955827e-07,
      "loss": 1.9752,
      "step": 224756
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.4933643341064453,
      "learning_rate": 8.885502464619831e-07,
      "loss": 2.1891,
      "step": 224757
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.3961181640625,
      "learning_rate": 8.882356758387998e-07,
      "loss": 1.9011,
      "step": 224758
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.4521914720535278,
      "learning_rate": 8.879211608261993e-07,
      "loss": 1.9677,
      "step": 224759
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.504205584526062,
      "learning_rate": 8.876067014241484e-07,
      "loss": 1.8755,
      "step": 224760
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.5970311164855957,
      "learning_rate": 8.872922976327468e-07,
      "loss": 1.828,
      "step": 224761
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.5147249698638916,
      "learning_rate": 8.869779494520613e-07,
      "loss": 1.797,
      "step": 224762
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.4324954748153687,
      "learning_rate": 8.86663656882125e-07,
      "loss": 1.8671,
      "step": 224763
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.4638515710830688,
      "learning_rate": 8.863494199229715e-07,
      "loss": 1.8513,
      "step": 224764
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.3767781257629395,
      "learning_rate": 8.860352385747338e-07,
      "loss": 1.9163,
      "step": 224765
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.4337321519851685,
      "learning_rate": 8.85721112837412e-07,
      "loss": 1.782,
      "step": 224766
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.4602028131484985,
      "learning_rate": 8.854070427110727e-07,
      "loss": 2.1292,
      "step": 224767
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.4347457885742188,
      "learning_rate": 8.850930281957825e-07,
      "loss": 1.9886,
      "step": 224768
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.4115369319915771,
      "learning_rate": 8.847790692916079e-07,
      "loss": 1.9829,
      "step": 224769
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.5182702541351318,
      "learning_rate": 8.844651659985824e-07,
      "loss": 2.0301,
      "step": 224770
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.5955368280410767,
      "learning_rate": 8.841513183168058e-07,
      "loss": 2.0445,
      "step": 224771
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.343295693397522,
      "learning_rate": 8.838375262463115e-07,
      "loss": 1.6932,
      "step": 224772
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.6331530809402466,
      "learning_rate": 8.835237897871661e-07,
      "loss": 2.1214,
      "step": 224773
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.5036661624908447,
      "learning_rate": 8.832101089393695e-07,
      "loss": 2.241,
      "step": 224774
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.3917068243026733,
      "learning_rate": 8.828964837030883e-07,
      "loss": 2.1159,
      "step": 224775
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.4626785516738892,
      "learning_rate": 8.825829140782892e-07,
      "loss": 2.0317,
      "step": 224776
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.5318294763565063,
      "learning_rate": 8.822694000650721e-07,
      "loss": 1.8544,
      "step": 224777
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.4489470720291138,
      "learning_rate": 8.819559416634703e-07,
      "loss": 1.8463,
      "step": 224778
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.4528347253799438,
      "learning_rate": 8.816425388735837e-07,
      "loss": 2.1046,
      "step": 224779
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.5532703399658203,
      "learning_rate": 8.813291916954456e-07,
      "loss": 2.2561,
      "step": 224780
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.581510066986084,
      "learning_rate": 8.810159001291229e-07,
      "loss": 2.2148,
      "step": 224781
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.4499279260635376,
      "learning_rate": 8.807026641746484e-07,
      "loss": 2.1215,
      "step": 224782
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.4303680658340454,
      "learning_rate": 8.803894838320891e-07,
      "loss": 2.0402,
      "step": 224783
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.4221246242523193,
      "learning_rate": 8.800763591015114e-07,
      "loss": 1.9023,
      "step": 224784
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.474278450012207,
      "learning_rate": 8.797632899830154e-07,
      "loss": 1.9817,
      "step": 224785
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.462334156036377,
      "learning_rate": 8.794502764765676e-07,
      "loss": 1.9943,
      "step": 224786
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.5088344812393188,
      "learning_rate": 8.791373185823014e-07,
      "loss": 2.0529,
      "step": 224787
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.486099123954773,
      "learning_rate": 8.7882441630025e-07,
      "loss": 1.987,
      "step": 224788
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.4181277751922607,
      "learning_rate": 8.785115696304468e-07,
      "loss": 1.814,
      "step": 224789
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.4869574308395386,
      "learning_rate": 8.781987785730249e-07,
      "loss": 1.9233,
      "step": 224790
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.517231822013855,
      "learning_rate": 8.778860431279511e-07,
      "loss": 1.8002,
      "step": 224791
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.4918023347854614,
      "learning_rate": 8.775733632953252e-07,
      "loss": 2.0147,
      "step": 224792
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.4591152667999268,
      "learning_rate": 8.77260739075214e-07,
      "loss": 1.9964,
      "step": 224793
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.5645747184753418,
      "learning_rate": 8.769481704676839e-07,
      "loss": 1.9431,
      "step": 224794
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.4809983968734741,
      "learning_rate": 8.766356574727351e-07,
      "loss": 1.9722,
      "step": 224795
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.401826024055481,
      "learning_rate": 8.763232000905007e-07,
      "loss": 2.1226,
      "step": 224796
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.5359984636306763,
      "learning_rate": 8.760107983210141e-07,
      "loss": 1.7447,
      "step": 224797
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.3472353219985962,
      "learning_rate": 8.756984521642752e-07,
      "loss": 1.7859,
      "step": 224798
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.4439494609832764,
      "learning_rate": 8.753861616204172e-07,
      "loss": 1.9777,
      "step": 224799
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.3389286994934082,
      "learning_rate": 8.750739266894736e-07,
      "loss": 1.7347,
      "step": 224800
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.3719124794006348,
      "learning_rate": 8.747617473715107e-07,
      "loss": 2.0277,
      "step": 224801
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.4503093957901,
      "learning_rate": 8.744496236665622e-07,
      "loss": 1.9164,
      "step": 224802
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.4510746002197266,
      "learning_rate": 8.741375555746944e-07,
      "loss": 1.7462,
      "step": 224803
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.4641293287277222,
      "learning_rate": 8.738255430959739e-07,
      "loss": 1.8481,
      "step": 224804
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.4339414834976196,
      "learning_rate": 8.735135862304676e-07,
      "loss": 2.1145,
      "step": 224805
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.4311643838882446,
      "learning_rate": 8.732016849782086e-07,
      "loss": 1.9069,
      "step": 224806
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.4701571464538574,
      "learning_rate": 8.728898393392636e-07,
      "loss": 1.9857,
      "step": 224807
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.4685847759246826,
      "learning_rate": 8.725780493136991e-07,
      "loss": 2.0209,
      "step": 224808
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.5149312019348145,
      "learning_rate": 8.722663149015819e-07,
      "loss": 1.8633,
      "step": 224809
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.466674566268921,
      "learning_rate": 8.719546361029783e-07,
      "loss": 1.7876,
      "step": 224810
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.4956575632095337,
      "learning_rate": 8.716430129178886e-07,
      "loss": 1.9979,
      "step": 224811
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.446767807006836,
      "learning_rate": 8.713314453464126e-07,
      "loss": 2.0292,
      "step": 224812
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.6355366706848145,
      "learning_rate": 8.710199333885837e-07,
      "loss": 1.9474,
      "step": 224813
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.390615701675415,
      "learning_rate": 8.707084770445017e-07,
      "loss": 2.3251,
      "step": 224814
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.5238239765167236,
      "learning_rate": 8.703970763141998e-07,
      "loss": 2.0119,
      "step": 224815
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.4439733028411865,
      "learning_rate": 8.70085731197745e-07,
      "loss": 1.913,
      "step": 224816
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.3658455610275269,
      "learning_rate": 8.697744416951702e-07,
      "loss": 2.1287,
      "step": 224817
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.6299841403961182,
      "learning_rate": 8.694632078065755e-07,
      "loss": 2.0748,
      "step": 224818
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.4695380926132202,
      "learning_rate": 8.69152029531961e-07,
      "loss": 1.8092,
      "step": 224819
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.5024962425231934,
      "learning_rate": 8.688409068714264e-07,
      "loss": 1.9935,
      "step": 224820
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.6374859809875488,
      "learning_rate": 8.685298398250384e-07,
      "loss": 1.9555,
      "step": 224821
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.5264592170715332,
      "learning_rate": 8.682188283928304e-07,
      "loss": 1.8694,
      "step": 224822
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.4734307527542114,
      "learning_rate": 8.679078725748689e-07,
      "loss": 1.826,
      "step": 224823
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.5482841730117798,
      "learning_rate": 8.675969723712206e-07,
      "loss": 2.0159,
      "step": 224824
    },
    {
      "epoch": 2.93,
      "grad_norm": 2.512176990509033,
      "learning_rate": 8.672861277819188e-07,
      "loss": 1.8746,
      "step": 224825
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.331793189048767,
      "learning_rate": 8.669753388070299e-07,
      "loss": 1.9758,
      "step": 224826
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.6127389669418335,
      "learning_rate": 8.666646054466208e-07,
      "loss": 1.9424,
      "step": 224827
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.4328186511993408,
      "learning_rate": 8.663539277007581e-07,
      "loss": 2.0,
      "step": 224828
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.6251405477523804,
      "learning_rate": 8.660433055694415e-07,
      "loss": 2.0468,
      "step": 224829
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.4558123350143433,
      "learning_rate": 8.657327390528379e-07,
      "loss": 1.8574,
      "step": 224830
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.4407172203063965,
      "learning_rate": 8.654222281509138e-07,
      "loss": 1.6906,
      "step": 224831
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.493316411972046,
      "learning_rate": 8.651117728637358e-07,
      "loss": 2.1919,
      "step": 224832
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.479189395904541,
      "learning_rate": 8.648013731914039e-07,
      "loss": 2.1246,
      "step": 224833
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.7569382190704346,
      "learning_rate": 8.64491029133918e-07,
      "loss": 2.1125,
      "step": 224834
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.3689792156219482,
      "learning_rate": 8.641807406914114e-07,
      "loss": 1.7698,
      "step": 224835
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.423037052154541,
      "learning_rate": 8.638705078638841e-07,
      "loss": 1.978,
      "step": 224836
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.4571171998977661,
      "learning_rate": 8.635603306514027e-07,
      "loss": 1.901,
      "step": 224837
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.4486151933670044,
      "learning_rate": 8.632502090540339e-07,
      "loss": 1.8867,
      "step": 224838
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.6962676048278809,
      "learning_rate": 8.629401430718441e-07,
      "loss": 1.7662,
      "step": 224839
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.4414011240005493,
      "learning_rate": 8.626301327048668e-07,
      "loss": 1.9364,
      "step": 224840
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.5109590291976929,
      "learning_rate": 8.623201779532019e-07,
      "loss": 2.0265,
      "step": 224841
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.4497793912887573,
      "learning_rate": 8.620102788168492e-07,
      "loss": 1.937,
      "step": 224842
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.4497747421264648,
      "learning_rate": 8.617004352959089e-07,
      "loss": 1.9629,
      "step": 224843
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.4451658725738525,
      "learning_rate": 8.613906473904142e-07,
      "loss": 1.9768,
      "step": 224844
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.3664829730987549,
      "learning_rate": 8.610809151004317e-07,
      "loss": 1.7179,
      "step": 224845
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.630732774734497,
      "learning_rate": 8.607712384260279e-07,
      "loss": 1.9026,
      "step": 224846
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.4100043773651123,
      "learning_rate": 8.604616173672696e-07,
      "loss": 2.1599,
      "step": 224847
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.4848791360855103,
      "learning_rate": 8.6015205192419e-07,
      "loss": 1.8839,
      "step": 224848
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.3826686143875122,
      "learning_rate": 8.598425420968225e-07,
      "loss": 2.0229,
      "step": 224849
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.6739389896392822,
      "learning_rate": 8.595330878853001e-07,
      "loss": 2.011,
      "step": 224850
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.447631597518921,
      "learning_rate": 8.592236892896231e-07,
      "loss": 1.9697,
      "step": 224851
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.5657752752304077,
      "learning_rate": 8.58914346309858e-07,
      "loss": 1.9408,
      "step": 224852
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.4233816862106323,
      "learning_rate": 8.586050589460714e-07,
      "loss": 1.8054,
      "step": 224853
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.4040944576263428,
      "learning_rate": 8.582958271982965e-07,
      "loss": 2.0344,
      "step": 224854
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.5217084884643555,
      "learning_rate": 8.579866510666334e-07,
      "loss": 1.9603,
      "step": 224855
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.417201280593872,
      "learning_rate": 8.576775305511152e-07,
      "loss": 1.7733,
      "step": 224856
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.4780772924423218,
      "learning_rate": 8.573684656518087e-07,
      "loss": 2.0509,
      "step": 224857
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.5186666250228882,
      "learning_rate": 8.570594563687472e-07,
      "loss": 2.2767,
      "step": 224858
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.4926388263702393,
      "learning_rate": 8.567505027020304e-07,
      "loss": 1.9249,
      "step": 224859
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.5686503648757935,
      "learning_rate": 8.564416046516587e-07,
      "loss": 2.0064,
      "step": 224860
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.413848876953125,
      "learning_rate": 8.561327622177316e-07,
      "loss": 1.9691,
      "step": 224861
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.6131857633590698,
      "learning_rate": 8.55823975400316e-07,
      "loss": 2.0119,
      "step": 224862
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.5509676933288574,
      "learning_rate": 8.555152441994117e-07,
      "loss": 1.9692,
      "step": 224863
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.5762557983398438,
      "learning_rate": 8.552065686151522e-07,
      "loss": 1.8701,
      "step": 224864
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.4958813190460205,
      "learning_rate": 8.548979486475371e-07,
      "loss": 1.8841,
      "step": 224865
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.4360573291778564,
      "learning_rate": 8.545893842966333e-07,
      "loss": 1.7684,
      "step": 224866
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.6089633703231812,
      "learning_rate": 8.542808755625407e-07,
      "loss": 2.1871,
      "step": 224867
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.5115464925765991,
      "learning_rate": 8.539724224452593e-07,
      "loss": 1.9033,
      "step": 224868
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.480318307876587,
      "learning_rate": 8.536640249448556e-07,
      "loss": 1.9353,
      "step": 224869
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.4564454555511475,
      "learning_rate": 8.53355683061463e-07,
      "loss": 2.0103,
      "step": 224870
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.399426817893982,
      "learning_rate": 8.530473967950147e-07,
      "loss": 2.018,
      "step": 224871
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.4149574041366577,
      "learning_rate": 8.527391661456773e-07,
      "loss": 1.8347,
      "step": 224872
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.4301868677139282,
      "learning_rate": 8.524309911134175e-07,
      "loss": 2.0038,
      "step": 224873
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.3736708164215088,
      "learning_rate": 8.521228716983685e-07,
      "loss": 1.9175,
      "step": 224874
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.4978262186050415,
      "learning_rate": 8.518148079005305e-07,
      "loss": 2.1816,
      "step": 224875
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.401196837425232,
      "learning_rate": 8.515067997200365e-07,
      "loss": 2.0839,
      "step": 224876
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.484505534172058,
      "learning_rate": 8.511988471568532e-07,
      "loss": 1.7861,
      "step": 224877
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.5075329542160034,
      "learning_rate": 8.508909502110806e-07,
      "loss": 1.9421,
      "step": 224878
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.5795726776123047,
      "learning_rate": 8.505831088827852e-07,
      "loss": 2.2317,
      "step": 224879
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.281985878944397,
      "learning_rate": 8.502753231720005e-07,
      "loss": 1.6355,
      "step": 224880
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.6027888059616089,
      "learning_rate": 8.499675930788264e-07,
      "loss": 2.0068,
      "step": 224881
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.340518593788147,
      "learning_rate": 8.496599186032627e-07,
      "loss": 2.0162,
      "step": 224882
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.4858101606369019,
      "learning_rate": 8.493522997454094e-07,
      "loss": 1.8205,
      "step": 224883
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.37202787399292,
      "learning_rate": 8.490447365053e-07,
      "loss": 1.7704,
      "step": 224884
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.551061987876892,
      "learning_rate": 8.487372288830008e-07,
      "loss": 2.086,
      "step": 224885
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.4262009859085083,
      "learning_rate": 8.484297768785786e-07,
      "loss": 1.9403,
      "step": 224886
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.4048365354537964,
      "learning_rate": 8.481223804920667e-07,
      "loss": 1.8297,
      "step": 224887
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.3582817316055298,
      "learning_rate": 8.478150397235317e-07,
      "loss": 1.9483,
      "step": 224888
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.4857521057128906,
      "learning_rate": 8.475077545730402e-07,
      "loss": 1.9897,
      "step": 224889
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.3738951683044434,
      "learning_rate": 8.472005250406589e-07,
      "loss": 2.0704,
      "step": 224890
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.4104515314102173,
      "learning_rate": 8.468933511264208e-07,
      "loss": 2.0969,
      "step": 224891
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.4643257856369019,
      "learning_rate": 8.46586232830393e-07,
      "loss": 1.8787,
      "step": 224892
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.5042780637741089,
      "learning_rate": 8.462791701526084e-07,
      "loss": 2.0246,
      "step": 224893
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.5379602909088135,
      "learning_rate": 8.45972163093167e-07,
      "loss": 1.8963,
      "step": 224894
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.6469978094100952,
      "learning_rate": 8.456652116521023e-07,
      "loss": 1.9809,
      "step": 224895
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.4173812866210938,
      "learning_rate": 8.453583158294808e-07,
      "loss": 1.7577,
      "step": 224896
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.5418277978897095,
      "learning_rate": 8.45051475625369e-07,
      "loss": 2.1156,
      "step": 224897
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.7057279348373413,
      "learning_rate": 8.447446910398003e-07,
      "loss": 2.018,
      "step": 224898
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.4101638793945312,
      "learning_rate": 8.444379620728081e-07,
      "loss": 1.8709,
      "step": 224899
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.5857279300689697,
      "learning_rate": 8.441312887245255e-07,
      "loss": 1.9714,
      "step": 224900
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.4170763492584229,
      "learning_rate": 8.438246709949525e-07,
      "loss": 1.7764,
      "step": 224901
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.3875417709350586,
      "learning_rate": 8.435181088841559e-07,
      "loss": 2.0082,
      "step": 224902
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.3889975547790527,
      "learning_rate": 8.432116023921687e-07,
      "loss": 1.8406,
      "step": 224903
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.4905492067337036,
      "learning_rate": 8.429051515191243e-07,
      "loss": 1.9389,
      "step": 224904
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.4589985609054565,
      "learning_rate": 8.425987562649894e-07,
      "loss": 1.8066,
      "step": 224905
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.4598690271377563,
      "learning_rate": 8.422924166298972e-07,
      "loss": 1.8795,
      "step": 224906
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.2891336679458618,
      "learning_rate": 8.419861326138478e-07,
      "loss": 1.7681,
      "step": 224907
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.4108189344406128,
      "learning_rate": 8.416799042169075e-07,
      "loss": 2.0203,
      "step": 224908
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.5890448093414307,
      "learning_rate": 8.413737314391766e-07,
      "loss": 2.0822,
      "step": 224909
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.430999517440796,
      "learning_rate": 8.41067614280655e-07,
      "loss": 1.8377,
      "step": 224910
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.4644818305969238,
      "learning_rate": 8.407615527414424e-07,
      "loss": 1.9317,
      "step": 224911
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.4582239389419556,
      "learning_rate": 8.404555468215723e-07,
      "loss": 1.8604,
      "step": 224912
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.4015288352966309,
      "learning_rate": 8.401495965211447e-07,
      "loss": 2.1783,
      "step": 224913
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.6303009986877441,
      "learning_rate": 8.39843701840126e-07,
      "loss": 1.8013,
      "step": 224914
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.60796320438385,
      "learning_rate": 8.395378627786497e-07,
      "loss": 1.8131,
      "step": 224915
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.4718397855758667,
      "learning_rate": 8.392320793367491e-07,
      "loss": 2.1477,
      "step": 224916
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.5735573768615723,
      "learning_rate": 8.389263515144906e-07,
      "loss": 1.9251,
      "step": 224917
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.507128357887268,
      "learning_rate": 8.386206793119077e-07,
      "loss": 2.0448,
      "step": 224918
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.5494816303253174,
      "learning_rate": 8.383150627291002e-07,
      "loss": 1.927,
      "step": 224919
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.3648772239685059,
      "learning_rate": 8.380095017660682e-07,
      "loss": 2.0142,
      "step": 224920
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.3049895763397217,
      "learning_rate": 8.377039964229115e-07,
      "loss": 1.6527,
      "step": 224921
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.5414296388626099,
      "learning_rate": 8.373985466996635e-07,
      "loss": 2.0145,
      "step": 224922
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.4556952714920044,
      "learning_rate": 8.370931525963909e-07,
      "loss": 1.9638,
      "step": 224923
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.536933422088623,
      "learning_rate": 8.367878141131601e-07,
      "loss": 1.8513,
      "step": 224924
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.4160733222961426,
      "learning_rate": 8.364825312500045e-07,
      "loss": 1.973,
      "step": 224925
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.5311473608016968,
      "learning_rate": 8.361773040070241e-07,
      "loss": 2.0076,
      "step": 224926
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.4447603225708008,
      "learning_rate": 8.358721323842188e-07,
      "loss": 2.0477,
      "step": 224927
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.4009798765182495,
      "learning_rate": 8.355670163816552e-07,
      "loss": 1.7072,
      "step": 224928
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.6276096105575562,
      "learning_rate": 8.352619559994334e-07,
      "loss": 2.0196,
      "step": 224929
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.553566575050354,
      "learning_rate": 8.349569512375865e-07,
      "loss": 1.9321,
      "step": 224930
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.4541980028152466,
      "learning_rate": 8.346520020961478e-07,
      "loss": 1.988,
      "step": 224931
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.5563383102416992,
      "learning_rate": 8.343471085752173e-07,
      "loss": 2.0769,
      "step": 224932
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.3809638023376465,
      "learning_rate": 8.340422706748284e-07,
      "loss": 1.8672,
      "step": 224933
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.3530796766281128,
      "learning_rate": 8.337374883950143e-07,
      "loss": 2.0282,
      "step": 224934
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.5434783697128296,
      "learning_rate": 8.334327617358749e-07,
      "loss": 1.9984,
      "step": 224935
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.4262654781341553,
      "learning_rate": 8.331280906974436e-07,
      "loss": 2.2466,
      "step": 224936
    },
    {
      "epoch": 2.93,
      "grad_norm": 2.0587778091430664,
      "learning_rate": 8.328234752797869e-07,
      "loss": 1.9367,
      "step": 224937
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.639776587486267,
      "learning_rate": 8.325189154829381e-07,
      "loss": 2.0191,
      "step": 224938
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.4015655517578125,
      "learning_rate": 8.32214411306964e-07,
      "loss": 2.0459,
      "step": 224939
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.3902355432510376,
      "learning_rate": 8.319099627519644e-07,
      "loss": 2.0235,
      "step": 224940
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.521592140197754,
      "learning_rate": 8.316055698179392e-07,
      "loss": 1.7935,
      "step": 224941
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.440516710281372,
      "learning_rate": 8.313012325049551e-07,
      "loss": 1.9045,
      "step": 224942
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.4672530889511108,
      "learning_rate": 8.30996950813112e-07,
      "loss": 1.9284,
      "step": 224943
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.5000524520874023,
      "learning_rate": 8.3069272474241e-07,
      "loss": 1.9937,
      "step": 224944
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.4024096727371216,
      "learning_rate": 8.303885542929156e-07,
      "loss": 2.0233,
      "step": 224945
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.6794859170913696,
      "learning_rate": 8.300844394647288e-07,
      "loss": 2.0709,
      "step": 224946
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.4498540163040161,
      "learning_rate": 8.297803802578495e-07,
      "loss": 1.9267,
      "step": 224947
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.5332145690917969,
      "learning_rate": 8.294763766723777e-07,
      "loss": 1.7632,
      "step": 224948
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.3988698720932007,
      "learning_rate": 8.291724287083801e-07,
      "loss": 1.8813,
      "step": 224949
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.5377919673919678,
      "learning_rate": 8.288685363658232e-07,
      "loss": 2.0416,
      "step": 224950
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.557393193244934,
      "learning_rate": 8.285646996448736e-07,
      "loss": 1.9348,
      "step": 224951
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.4354771375656128,
      "learning_rate": 8.282609185455313e-07,
      "loss": 1.7396,
      "step": 224952
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.6023832559585571,
      "learning_rate": 8.27957193067863e-07,
      "loss": 2.0475,
      "step": 224953
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.6982935667037964,
      "learning_rate": 8.276535232119019e-07,
      "loss": 1.8491,
      "step": 224954
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.5341994762420654,
      "learning_rate": 8.273499089777813e-07,
      "loss": 2.1909,
      "step": 224955
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.4449098110198975,
      "learning_rate": 8.270463503654678e-07,
      "loss": 1.9256,
      "step": 224956
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.4015547037124634,
      "learning_rate": 8.267428473750615e-07,
      "loss": 2.0535,
      "step": 224957
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.4845589399337769,
      "learning_rate": 8.264394000065954e-07,
      "loss": 1.9416,
      "step": 224958
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.4336000680923462,
      "learning_rate": 8.261360082601365e-07,
      "loss": 1.9505,
      "step": 224959
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.5363490581512451,
      "learning_rate": 8.258326721357511e-07,
      "loss": 2.0166,
      "step": 224960
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.5088274478912354,
      "learning_rate": 8.255293916335392e-07,
      "loss": 1.9532,
      "step": 224961
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.4523756504058838,
      "learning_rate": 8.252261667534677e-07,
      "loss": 1.789,
      "step": 224962
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.488450050354004,
      "learning_rate": 8.249229974956362e-07,
      "loss": 2.1014,
      "step": 224963
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.5054091215133667,
      "learning_rate": 8.246198838600781e-07,
      "loss": 1.9276,
      "step": 224964
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.5427032709121704,
      "learning_rate": 8.243168258468935e-07,
      "loss": 2.1516,
      "step": 224965
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.5657124519348145,
      "learning_rate": 8.240138234561155e-07,
      "loss": 2.1906,
      "step": 224966
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.6521354913711548,
      "learning_rate": 8.237108766878109e-07,
      "loss": 1.8617,
      "step": 224967
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.466791033744812,
      "learning_rate": 8.234079855420128e-07,
      "loss": 1.9544,
      "step": 224968
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.5474876165390015,
      "learning_rate": 8.23105150018788e-07,
      "loss": 2.0541,
      "step": 224969
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.493187427520752,
      "learning_rate": 8.228023701182029e-07,
      "loss": 2.0998,
      "step": 224970
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.5047178268432617,
      "learning_rate": 8.22499645840291e-07,
      "loss": 1.941,
      "step": 224971
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.3595446348190308,
      "learning_rate": 8.221969771851522e-07,
      "loss": 1.8549,
      "step": 224972
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.5139448642730713,
      "learning_rate": 8.218943641528197e-07,
      "loss": 1.869,
      "step": 224973
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.4834667444229126,
      "learning_rate": 8.215918067432936e-07,
      "loss": 1.8624,
      "step": 224974
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.4765236377716064,
      "learning_rate": 8.212893049567071e-07,
      "loss": 2.0734,
      "step": 224975
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.4134312868118286,
      "learning_rate": 8.209868587930935e-07,
      "loss": 2.0505,
      "step": 224976
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.4075275659561157,
      "learning_rate": 8.206844682525193e-07,
      "loss": 1.9681,
      "step": 224977
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.4285550117492676,
      "learning_rate": 8.20382133335018e-07,
      "loss": 1.8427,
      "step": 224978
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.4474329948425293,
      "learning_rate": 8.200798540406562e-07,
      "loss": 1.9714,
      "step": 224979
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.6011468172073364,
      "learning_rate": 8.19777630369467e-07,
      "loss": 1.7417,
      "step": 224980
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.4511383771896362,
      "learning_rate": 8.194754623215505e-07,
      "loss": 1.8707,
      "step": 224981
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.496206521987915,
      "learning_rate": 8.191733498969399e-07,
      "loss": 2.1356,
      "step": 224982
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.391758918762207,
      "learning_rate": 8.18871293095702e-07,
      "loss": 1.8654,
      "step": 224983
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.4906630516052246,
      "learning_rate": 8.185692919178366e-07,
      "loss": 1.7572,
      "step": 224984
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.5257093906402588,
      "learning_rate": 8.182673463635103e-07,
      "loss": 1.9614,
      "step": 224985
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.503746509552002,
      "learning_rate": 8.179654564326898e-07,
      "loss": 1.8901,
      "step": 224986
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.6110492944717407,
      "learning_rate": 8.176636221254418e-07,
      "loss": 1.8907,
      "step": 224987
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.3846732378005981,
      "learning_rate": 8.173618434418328e-07,
      "loss": 1.7428,
      "step": 224988
    },
    {
      "epoch": 2.93,
      "grad_norm": 2.953420877456665,
      "learning_rate": 8.170601203819627e-07,
      "loss": 2.0283,
      "step": 224989
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.5162988901138306,
      "learning_rate": 8.167584529457983e-07,
      "loss": 1.9381,
      "step": 224990
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.3966435194015503,
      "learning_rate": 8.164568411334727e-07,
      "loss": 1.804,
      "step": 224991
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.6148958206176758,
      "learning_rate": 8.161552849450192e-07,
      "loss": 2.1434,
      "step": 224992
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.6810476779937744,
      "learning_rate": 8.158537843805046e-07,
      "loss": 2.0136,
      "step": 224993
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.492795467376709,
      "learning_rate": 8.155523394399288e-07,
      "loss": 2.1373,
      "step": 224994
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.4911527633666992,
      "learning_rate": 8.15250950123425e-07,
      "loss": 2.07,
      "step": 224995
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.5938425064086914,
      "learning_rate": 8.149496164309932e-07,
      "loss": 1.8551,
      "step": 224996
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.487923264503479,
      "learning_rate": 8.146483383627334e-07,
      "loss": 2.1274,
      "step": 224997
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.420434594154358,
      "learning_rate": 8.143471159186787e-07,
      "loss": 2.1082,
      "step": 224998
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.5187795162200928,
      "learning_rate": 8.140459490988627e-07,
      "loss": 2.2926,
      "step": 224999
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.5819299221038818,
      "learning_rate": 8.137448379033518e-07,
      "loss": 2.0383,
      "step": 225000
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.385455846786499,
      "learning_rate": 8.134437823322459e-07,
      "loss": 1.8902,
      "step": 225001
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.5146538019180298,
      "learning_rate": 8.131427823855452e-07,
      "loss": 1.8578,
      "step": 225002
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.5431280136108398,
      "learning_rate": 8.128418380633162e-07,
      "loss": 2.0214,
      "step": 225003
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.4512768983840942,
      "learning_rate": 8.125409493656588e-07,
      "loss": 1.92,
      "step": 225004
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.4132707118988037,
      "learning_rate": 8.122401162925729e-07,
      "loss": 2.0593,
      "step": 225005
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.473747730255127,
      "learning_rate": 8.119393388441586e-07,
      "loss": 1.9444,
      "step": 225006
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.443813681602478,
      "learning_rate": 8.116386170204492e-07,
      "loss": 1.9862,
      "step": 225007
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.67349374294281,
      "learning_rate": 8.113379508215112e-07,
      "loss": 2.1102,
      "step": 225008
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.438902735710144,
      "learning_rate": 8.11037340247378e-07,
      "loss": 2.1436,
      "step": 225009
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.3871434926986694,
      "learning_rate": 8.107367852981494e-07,
      "loss": 1.9186,
      "step": 225010
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.4989114999771118,
      "learning_rate": 8.104362859737923e-07,
      "loss": 1.8664,
      "step": 225011
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.4623284339904785,
      "learning_rate": 8.101358422744731e-07,
      "loss": 2.0912,
      "step": 225012
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.5290387868881226,
      "learning_rate": 8.098354542001917e-07,
      "loss": 2.1696,
      "step": 225013
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.478872299194336,
      "learning_rate": 8.095351217509816e-07,
      "loss": 1.932,
      "step": 225014
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.4713910818099976,
      "learning_rate": 8.092348449269759e-07,
      "loss": 2.0627,
      "step": 225015
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.436742901802063,
      "learning_rate": 8.089346237281413e-07,
      "loss": 1.9709,
      "step": 225016
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.4758208990097046,
      "learning_rate": 8.086344581545778e-07,
      "loss": 1.9871,
      "step": 225017
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.4776113033294678,
      "learning_rate": 8.08334348206352e-07,
      "loss": 1.9193,
      "step": 225018
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.6027867794036865,
      "learning_rate": 8.080342938834972e-07,
      "loss": 1.993,
      "step": 225019
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.485310435295105,
      "learning_rate": 8.077342951860799e-07,
      "loss": 1.7295,
      "step": 225020
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.5163899660110474,
      "learning_rate": 8.074343521141335e-07,
      "loss": 1.8966,
      "step": 225021
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.5008171796798706,
      "learning_rate": 8.07134464667758e-07,
      "loss": 1.826,
      "step": 225022
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.5827536582946777,
      "learning_rate": 8.068346328469532e-07,
      "loss": 2.0006,
      "step": 225023
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.4194811582565308,
      "learning_rate": 8.065348566518193e-07,
      "loss": 1.8708,
      "step": 225024
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.5027121305465698,
      "learning_rate": 8.062351360824226e-07,
      "loss": 1.9214,
      "step": 225025
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.4384440183639526,
      "learning_rate": 8.059354711387634e-07,
      "loss": 1.8914,
      "step": 225026
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.4154475927352905,
      "learning_rate": 8.056358618209413e-07,
      "loss": 1.9523,
      "step": 225027
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.5771669149398804,
      "learning_rate": 8.053363081290232e-07,
      "loss": 1.9967,
      "step": 225028
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.549750566482544,
      "learning_rate": 8.050368100630089e-07,
      "loss": 1.9679,
      "step": 225029
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.5413256883621216,
      "learning_rate": 8.047373676229985e-07,
      "loss": 1.9297,
      "step": 225030
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.536717176437378,
      "learning_rate": 8.044379808089919e-07,
      "loss": 2.0593,
      "step": 225031
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.3500072956085205,
      "learning_rate": 8.041386496211555e-07,
      "loss": 1.8259,
      "step": 225032
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.561782956123352,
      "learning_rate": 8.038393740594229e-07,
      "loss": 2.08,
      "step": 225033
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.560171127319336,
      "learning_rate": 8.035401541239605e-07,
      "loss": 2.1885,
      "step": 225034
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.6257915496826172,
      "learning_rate": 8.032409898147352e-07,
      "loss": 1.8745,
      "step": 225035
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.7666974067687988,
      "learning_rate": 8.029418811318133e-07,
      "loss": 1.9216,
      "step": 225036
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.5336629152297974,
      "learning_rate": 8.02642828075295e-07,
      "loss": 2.0327,
      "step": 225037
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.4266666173934937,
      "learning_rate": 8.023438306452135e-07,
      "loss": 2.1116,
      "step": 225038
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.6580580472946167,
      "learning_rate": 8.020448888416353e-07,
      "loss": 2.1766,
      "step": 225039
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.6617200374603271,
      "learning_rate": 8.017460026645939e-07,
      "loss": 2.2507,
      "step": 225040
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.548005223274231,
      "learning_rate": 8.014471721141558e-07,
      "loss": 1.952,
      "step": 225041
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.4768697023391724,
      "learning_rate": 8.011483971903875e-07,
      "loss": 2.0216,
      "step": 225042
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.446879506111145,
      "learning_rate": 8.008496778933227e-07,
      "loss": 2.0138,
      "step": 225043
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.439431071281433,
      "learning_rate": 8.005510142230276e-07,
      "loss": 1.8945,
      "step": 225044
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.5405669212341309,
      "learning_rate": 8.002524061795356e-07,
      "loss": 1.985,
      "step": 225045
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.3857512474060059,
      "learning_rate": 7.999538537629469e-07,
      "loss": 1.9343,
      "step": 225046
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.5659104585647583,
      "learning_rate": 7.996553569732944e-07,
      "loss": 1.9716,
      "step": 225047
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.4723310470581055,
      "learning_rate": 7.99356915810645e-07,
      "loss": 2.0029,
      "step": 225048
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.4112651348114014,
      "learning_rate": 7.990585302750318e-07,
      "loss": 2.0575,
      "step": 225049
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.5441969633102417,
      "learning_rate": 7.987602003665217e-07,
      "loss": 2.1561,
      "step": 225050
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.4018778800964355,
      "learning_rate": 7.984619260851477e-07,
      "loss": 2.2499,
      "step": 225051
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.3949393033981323,
      "learning_rate": 7.981637074310432e-07,
      "loss": 1.8778,
      "step": 225052
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.3979007005691528,
      "learning_rate": 7.978655444041749e-07,
      "loss": 2.0462,
      "step": 225053
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.54806387424469,
      "learning_rate": 7.975674370046093e-07,
      "loss": 1.8938,
      "step": 225054
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.3599916696548462,
      "learning_rate": 7.972693852324464e-07,
      "loss": 2.0031,
      "step": 225055
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.6262303590774536,
      "learning_rate": 7.969713890877194e-07,
      "loss": 2.0505,
      "step": 225056
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.3197981119155884,
      "learning_rate": 7.966734485704951e-07,
      "loss": 1.623,
      "step": 225057
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.4119917154312134,
      "learning_rate": 7.963755636808066e-07,
      "loss": 1.8955,
      "step": 225058
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.565802812576294,
      "learning_rate": 7.960777344187208e-07,
      "loss": 1.8987,
      "step": 225059
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.457598328590393,
      "learning_rate": 7.957799607842708e-07,
      "loss": 1.9111,
      "step": 225060
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.533200979232788,
      "learning_rate": 7.954822427775565e-07,
      "loss": 1.9751,
      "step": 225061
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.4753608703613281,
      "learning_rate": 7.951845803986113e-07,
      "loss": 1.9976,
      "step": 225062
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.3692988157272339,
      "learning_rate": 7.948869736474684e-07,
      "loss": 1.638,
      "step": 225063
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.3098622560501099,
      "learning_rate": 7.945894225242278e-07,
      "loss": 1.9264,
      "step": 225064
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.5881294012069702,
      "learning_rate": 7.942919270288894e-07,
      "loss": 1.8715,
      "step": 225065
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.4710595607757568,
      "learning_rate": 7.939944871615866e-07,
      "loss": 1.9955,
      "step": 225066
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.475313663482666,
      "learning_rate": 7.936971029222861e-07,
      "loss": 1.9879,
      "step": 225067
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.417288064956665,
      "learning_rate": 7.933997743111209e-07,
      "loss": 2.0827,
      "step": 225068
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.5023763179779053,
      "learning_rate": 7.931025013280578e-07,
      "loss": 2.0718,
      "step": 225069
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.436987042427063,
      "learning_rate": 7.928052839732635e-07,
      "loss": 1.8288,
      "step": 225070
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.4765993356704712,
      "learning_rate": 7.925081222467044e-07,
      "loss": 2.0216,
      "step": 225071
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.3760660886764526,
      "learning_rate": 7.922110161484807e-07,
      "loss": 2.0133,
      "step": 225072
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.3902363777160645,
      "learning_rate": 7.919139656786256e-07,
      "loss": 1.8649,
      "step": 225073
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.530715823173523,
      "learning_rate": 7.916169708372056e-07,
      "loss": 1.8531,
      "step": 225074
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.599380612373352,
      "learning_rate": 7.91320031624254e-07,
      "loss": 1.9637,
      "step": 225075
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.5903271436691284,
      "learning_rate": 7.910231480398377e-07,
      "loss": 1.9335,
      "step": 225076
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.565758228302002,
      "learning_rate": 7.907263200840564e-07,
      "loss": 2.1841,
      "step": 225077
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.3357934951782227,
      "learning_rate": 7.9042954775691e-07,
      "loss": 1.8812,
      "step": 225078
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.56902015209198,
      "learning_rate": 7.901328310584653e-07,
      "loss": 2.0211,
      "step": 225079
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.6072683334350586,
      "learning_rate": 7.89836169988789e-07,
      "loss": 2.1056,
      "step": 225080
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.3621604442596436,
      "learning_rate": 7.895395645479141e-07,
      "loss": 1.8477,
      "step": 225081
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.4953023195266724,
      "learning_rate": 7.892430147359407e-07,
      "loss": 2.0926,
      "step": 225082
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.6463578939437866,
      "learning_rate": 7.889465205528688e-07,
      "loss": 2.1352,
      "step": 225083
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.5045756101608276,
      "learning_rate": 7.886500819987984e-07,
      "loss": 2.0456,
      "step": 225084
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.3134924173355103,
      "learning_rate": 7.883536990737627e-07,
      "loss": 1.9467,
      "step": 225085
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.3846848011016846,
      "learning_rate": 7.88057371777795e-07,
      "loss": 2.059,
      "step": 225086
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.6303236484527588,
      "learning_rate": 7.877611001109951e-07,
      "loss": 1.9,
      "step": 225087
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.4837251901626587,
      "learning_rate": 7.874648840733965e-07,
      "loss": 2.2316,
      "step": 225088
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.474445104598999,
      "learning_rate": 7.871687236650659e-07,
      "loss": 1.9745,
      "step": 225089
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.522987723350525,
      "learning_rate": 7.868726188860364e-07,
      "loss": 1.9425,
      "step": 225090
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.6358065605163574,
      "learning_rate": 7.865765697363746e-07,
      "loss": 2.0216,
      "step": 225091
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.3811956644058228,
      "learning_rate": 7.862805762161473e-07,
      "loss": 1.8733,
      "step": 225092
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.468738079071045,
      "learning_rate": 7.859846383253876e-07,
      "loss": 1.7691,
      "step": 225093
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.5820989608764648,
      "learning_rate": 7.856887560641956e-07,
      "loss": 1.9421,
      "step": 225094
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.41434907913208,
      "learning_rate": 7.853929294325712e-07,
      "loss": 1.9909,
      "step": 225095
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.3479586839675903,
      "learning_rate": 7.850971584305477e-07,
      "loss": 1.7339,
      "step": 225096
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.3924052715301514,
      "learning_rate": 7.848014430582583e-07,
      "loss": 1.8533,
      "step": 225097
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.611151933670044,
      "learning_rate": 7.845057833157365e-07,
      "loss": 2.0688,
      "step": 225098
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.5224413871765137,
      "learning_rate": 7.842101792030153e-07,
      "loss": 1.8875,
      "step": 225099
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.5031970739364624,
      "learning_rate": 7.839146307201615e-07,
      "loss": 1.973,
      "step": 225100
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.4777491092681885,
      "learning_rate": 7.836191378672085e-07,
      "loss": 1.838,
      "step": 225101
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.4957482814788818,
      "learning_rate": 7.833237006442227e-07,
      "loss": 1.7961,
      "step": 225102
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.5216014385223389,
      "learning_rate": 7.83028319051271e-07,
      "loss": 1.8923,
      "step": 225103
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.5619182586669922,
      "learning_rate": 7.827329930884196e-07,
      "loss": 2.0689,
      "step": 225104
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.3589134216308594,
      "learning_rate": 7.824377227557022e-07,
      "loss": 2.1462,
      "step": 225105
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.362863540649414,
      "learning_rate": 7.821425080531518e-07,
      "loss": 1.9264,
      "step": 225106
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.5789482593536377,
      "learning_rate": 7.818473489809018e-07,
      "loss": 2.109,
      "step": 225107
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.698567271232605,
      "learning_rate": 7.815522455389189e-07,
      "loss": 1.9445,
      "step": 225108
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.467174768447876,
      "learning_rate": 7.81257197727303e-07,
      "loss": 1.924,
      "step": 225109
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.4588428735733032,
      "learning_rate": 7.809622055460874e-07,
      "loss": 1.9587,
      "step": 225110
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.5037087202072144,
      "learning_rate": 7.806672689953386e-07,
      "loss": 2.3475,
      "step": 225111
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.4682519435882568,
      "learning_rate": 7.803723880751233e-07,
      "loss": 2.208,
      "step": 225112
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.421891450881958,
      "learning_rate": 7.800775627854749e-07,
      "loss": 2.0396,
      "step": 225113
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.4018514156341553,
      "learning_rate": 7.7978279312646e-07,
      "loss": 2.0968,
      "step": 225114
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.7039461135864258,
      "learning_rate": 7.79488079098145e-07,
      "loss": 2.1829,
      "step": 225115
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.3765555620193481,
      "learning_rate": 7.791934207005634e-07,
      "loss": 1.9637,
      "step": 225116
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.4530668258666992,
      "learning_rate": 7.788988179337818e-07,
      "loss": 2.1209,
      "step": 225117
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.5685138702392578,
      "learning_rate": 7.786042707978335e-07,
      "loss": 1.9599,
      "step": 225118
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.4650285243988037,
      "learning_rate": 7.783097792928184e-07,
      "loss": 2.3764,
      "step": 225119
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.4518868923187256,
      "learning_rate": 7.780153434187364e-07,
      "loss": 1.9725,
      "step": 225120
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.348874807357788,
      "learning_rate": 7.777209631756876e-07,
      "loss": 1.747,
      "step": 225121
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.4951163530349731,
      "learning_rate": 7.774266385637051e-07,
      "loss": 2.1288,
      "step": 225122
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.5027034282684326,
      "learning_rate": 7.771323695828557e-07,
      "loss": 2.0014,
      "step": 225123
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.4685834646224976,
      "learning_rate": 7.768381562331394e-07,
      "loss": 2.2256,
      "step": 225124
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.5390042066574097,
      "learning_rate": 7.765439985147226e-07,
      "loss": 2.118,
      "step": 225125
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.524918556213379,
      "learning_rate": 7.762498964275721e-07,
      "loss": 1.9417,
      "step": 225126
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.4159919023513794,
      "learning_rate": 7.759558499717544e-07,
      "loss": 1.8891,
      "step": 225127
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.5237120389938354,
      "learning_rate": 7.756618591473363e-07,
      "loss": 2.1316,
      "step": 225128
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.563857913017273,
      "learning_rate": 7.753679239543509e-07,
      "loss": 2.0118,
      "step": 225129
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.35957932472229,
      "learning_rate": 7.750740443928982e-07,
      "loss": 1.9568,
      "step": 225130
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.586437463760376,
      "learning_rate": 7.747802204630116e-07,
      "loss": 1.8853,
      "step": 225131
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.5046032667160034,
      "learning_rate": 7.744864521647242e-07,
      "loss": 1.9818,
      "step": 225132
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.4130083322525024,
      "learning_rate": 7.741927394981362e-07,
      "loss": 2.1214,
      "step": 225133
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.4320836067199707,
      "learning_rate": 7.738990824632474e-07,
      "loss": 2.011,
      "step": 225134
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.4666467905044556,
      "learning_rate": 7.736054810601577e-07,
      "loss": 2.044,
      "step": 225135
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.5734171867370605,
      "learning_rate": 7.733119352888672e-07,
      "loss": 1.8618,
      "step": 225136
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.441103219985962,
      "learning_rate": 7.730184451495092e-07,
      "loss": 1.7699,
      "step": 225137
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.5308184623718262,
      "learning_rate": 7.727250106420835e-07,
      "loss": 1.9272,
      "step": 225138
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.5686616897583008,
      "learning_rate": 7.724316317666568e-07,
      "loss": 1.917,
      "step": 225139
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.5519988536834717,
      "learning_rate": 7.721383085232624e-07,
      "loss": 2.0279,
      "step": 225140
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.5180649757385254,
      "learning_rate": 7.718450409120002e-07,
      "loss": 2.1561,
      "step": 225141
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.4637295007705688,
      "learning_rate": 7.715518289329037e-07,
      "loss": 1.895,
      "step": 225142
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.369106411933899,
      "learning_rate": 7.712586725860059e-07,
      "loss": 1.9323,
      "step": 225143
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.3712040185928345,
      "learning_rate": 7.709655718713737e-07,
      "loss": 1.7842,
      "step": 225144
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.5141069889068604,
      "learning_rate": 7.706725267891068e-07,
      "loss": 1.9364,
      "step": 225145
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.3768649101257324,
      "learning_rate": 7.703795373391719e-07,
      "loss": 1.8688,
      "step": 225146
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.431030511856079,
      "learning_rate": 7.700866035217024e-07,
      "loss": 1.7677,
      "step": 225147
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.5199558734893799,
      "learning_rate": 7.697937253366982e-07,
      "loss": 1.8935,
      "step": 225148
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.5506073236465454,
      "learning_rate": 7.695009027842591e-07,
      "loss": 2.1187,
      "step": 225149
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.4624605178833008,
      "learning_rate": 7.692081358644187e-07,
      "loss": 1.8798,
      "step": 225150
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.4749488830566406,
      "learning_rate": 7.6891542457721e-07,
      "loss": 1.8193,
      "step": 225151
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.6381890773773193,
      "learning_rate": 7.686227689227331e-07,
      "loss": 1.959,
      "step": 225152
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.4468916654586792,
      "learning_rate": 7.683301689010213e-07,
      "loss": 1.9451,
      "step": 225153
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.5808385610580444,
      "learning_rate": 7.680376245120745e-07,
      "loss": 2.0268,
      "step": 225154
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.5067616701126099,
      "learning_rate": 7.677451357560593e-07,
      "loss": 2.1282,
      "step": 225155
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.513018250465393,
      "learning_rate": 7.674527026329424e-07,
      "loss": 1.811,
      "step": 225156
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.5617587566375732,
      "learning_rate": 7.671603251428237e-07,
      "loss": 1.9476,
      "step": 225157
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.6152786016464233,
      "learning_rate": 7.668680032857033e-07,
      "loss": 2.1786,
      "step": 225158
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.494593858718872,
      "learning_rate": 7.66575737061681e-07,
      "loss": 1.958,
      "step": 225159
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.373422622680664,
      "learning_rate": 7.662835264707901e-07,
      "loss": 1.9629,
      "step": 225160
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.7657195329666138,
      "learning_rate": 7.659913715131305e-07,
      "loss": 1.9797,
      "step": 225161
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.431767463684082,
      "learning_rate": 7.656992721887024e-07,
      "loss": 1.991,
      "step": 225162
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.371200442314148,
      "learning_rate": 7.654072284976054e-07,
      "loss": 1.8796,
      "step": 225163
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.5509274005889893,
      "learning_rate": 7.651152404398064e-07,
      "loss": 1.8649,
      "step": 225164
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.3412196636199951,
      "learning_rate": 7.648233080154719e-07,
      "loss": 1.9855,
      "step": 225165
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.3867806196212769,
      "learning_rate": 7.64531431224602e-07,
      "loss": 1.9173,
      "step": 225166
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.5076539516448975,
      "learning_rate": 7.642396100672299e-07,
      "loss": 2.0093,
      "step": 225167
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.3999601602554321,
      "learning_rate": 7.639478445434555e-07,
      "loss": 1.9971,
      "step": 225168
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.4479504823684692,
      "learning_rate": 7.636561346533121e-07,
      "loss": 1.9409,
      "step": 225169
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.371316909790039,
      "learning_rate": 7.633644803968331e-07,
      "loss": 1.9749,
      "step": 225170
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.3921915292739868,
      "learning_rate": 7.630728817741183e-07,
      "loss": 1.9157,
      "step": 225171
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.4887661933898926,
      "learning_rate": 7.627813387851678e-07,
      "loss": 1.9209,
      "step": 225172
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.5099884271621704,
      "learning_rate": 7.624898514300814e-07,
      "loss": 1.9567,
      "step": 225173
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.5218241214752197,
      "learning_rate": 7.621984197088926e-07,
      "loss": 2.2196,
      "step": 225174
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.4581952095031738,
      "learning_rate": 7.619070436216679e-07,
      "loss": 1.9467,
      "step": 225175
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.575216293334961,
      "learning_rate": 7.616157231684406e-07,
      "loss": 2.0985,
      "step": 225176
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.4015843868255615,
      "learning_rate": 7.613244583493106e-07,
      "loss": 2.0011,
      "step": 225177
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.341265320777893,
      "learning_rate": 7.610332491642446e-07,
      "loss": 2.0682,
      "step": 225178
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.4113271236419678,
      "learning_rate": 7.607420956134092e-07,
      "loss": 1.895,
      "step": 225179
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.573945164680481,
      "learning_rate": 7.60450997696771e-07,
      "loss": 2.2469,
      "step": 225180
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.7705683708190918,
      "learning_rate": 7.601599554143966e-07,
      "loss": 1.9485,
      "step": 225181
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.4341509342193604,
      "learning_rate": 7.598689687663861e-07,
      "loss": 2.1549,
      "step": 225182
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.4727082252502441,
      "learning_rate": 7.595780377527394e-07,
      "loss": 1.9966,
      "step": 225183
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.4828083515167236,
      "learning_rate": 7.592871623735563e-07,
      "loss": 2.0603,
      "step": 225184
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.4062045812606812,
      "learning_rate": 7.589963426288703e-07,
      "loss": 1.7134,
      "step": 225185
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.4919759035110474,
      "learning_rate": 7.587055785187146e-07,
      "loss": 1.9512,
      "step": 225186
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.3558692932128906,
      "learning_rate": 7.584148700431891e-07,
      "loss": 1.7729,
      "step": 225187
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.929200291633606,
      "learning_rate": 7.581242172022939e-07,
      "loss": 2.1285,
      "step": 225188
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.4333903789520264,
      "learning_rate": 7.578336199961288e-07,
      "loss": 1.8682,
      "step": 225189
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.4585952758789062,
      "learning_rate": 7.575430784247271e-07,
      "loss": 2.1393,
      "step": 225190
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.4859614372253418,
      "learning_rate": 7.572525924881556e-07,
      "loss": 1.969,
      "step": 225191
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.4362258911132812,
      "learning_rate": 7.569621621864807e-07,
      "loss": 2.2072,
      "step": 225192
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.4377591609954834,
      "learning_rate": 7.566717875197026e-07,
      "loss": 1.996,
      "step": 225193
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.5113883018493652,
      "learning_rate": 7.56381468487921e-07,
      "loss": 2.2642,
      "step": 225194
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.2990819215774536,
      "learning_rate": 7.560912050911694e-07,
      "loss": 1.7977,
      "step": 225195
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.4138870239257812,
      "learning_rate": 7.558009973295143e-07,
      "loss": 2.0272,
      "step": 225196
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.507709264755249,
      "learning_rate": 7.55510845202989e-07,
      "loss": 1.9364,
      "step": 225197
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.49539053440094,
      "learning_rate": 7.552207487116935e-07,
      "loss": 1.8781,
      "step": 225198
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.7486546039581299,
      "learning_rate": 7.549307078556277e-07,
      "loss": 1.9288,
      "step": 225199
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.547270655632019,
      "learning_rate": 7.546407226348584e-07,
      "loss": 2.1034,
      "step": 225200
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.4256101846694946,
      "learning_rate": 7.54350793049452e-07,
      "loss": 1.8864,
      "step": 225201
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.4549801349639893,
      "learning_rate": 7.54060919099475e-07,
      "loss": 1.8283,
      "step": 225202
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.451780915260315,
      "learning_rate": 7.537711007849612e-07,
      "loss": 2.1377,
      "step": 225203
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.7148066759109497,
      "learning_rate": 7.534813381059768e-07,
      "loss": 1.9348,
      "step": 225204
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.6544363498687744,
      "learning_rate": 7.531916310625885e-07,
      "loss": 2.1162,
      "step": 225205
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.6568875312805176,
      "learning_rate": 7.529019796547964e-07,
      "loss": 1.8295,
      "step": 225206
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.4799233675003052,
      "learning_rate": 7.526123838827003e-07,
      "loss": 1.9159,
      "step": 225207
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.3898844718933105,
      "learning_rate": 7.523228437463668e-07,
      "loss": 1.924,
      "step": 225208
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.5127391815185547,
      "learning_rate": 7.520333592457961e-07,
      "loss": 1.9483,
      "step": 225209
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.406654715538025,
      "learning_rate": 7.517439303810547e-07,
      "loss": 1.8621,
      "step": 225210
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.4379137754440308,
      "learning_rate": 7.514545571522423e-07,
      "loss": 1.9352,
      "step": 225211
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.4960248470306396,
      "learning_rate": 7.511652395593925e-07,
      "loss": 2.0091,
      "step": 225212
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.3926798105239868,
      "learning_rate": 7.508759776025053e-07,
      "loss": 1.8463,
      "step": 225213
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.4575142860412598,
      "learning_rate": 7.505867712817138e-07,
      "loss": 1.8406,
      "step": 225214
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.4159859418869019,
      "learning_rate": 7.502976205970179e-07,
      "loss": 1.9234,
      "step": 225215
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.5294469594955444,
      "learning_rate": 7.500085255485177e-07,
      "loss": 1.8736,
      "step": 225216
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.5777390003204346,
      "learning_rate": 7.497194861362466e-07,
      "loss": 1.9706,
      "step": 225217
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.4039052724838257,
      "learning_rate": 7.494305023602043e-07,
      "loss": 1.953,
      "step": 225218
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.516404151916504,
      "learning_rate": 7.491415742205243e-07,
      "loss": 1.9302,
      "step": 225219
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.6152762174606323,
      "learning_rate": 7.488527017172064e-07,
      "loss": 2.0496,
      "step": 225220
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.5855529308319092,
      "learning_rate": 7.485638848503505e-07,
      "loss": 2.2327,
      "step": 225221
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.3452006578445435,
      "learning_rate": 7.482751236199902e-07,
      "loss": 2.0309,
      "step": 225222
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.3989052772521973,
      "learning_rate": 7.479864180261585e-07,
      "loss": 1.8768,
      "step": 225223
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.4337371587753296,
      "learning_rate": 7.476977680689222e-07,
      "loss": 1.8601,
      "step": 225224
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.4618326425552368,
      "learning_rate": 7.474091737483145e-07,
      "loss": 2.0144,
      "step": 225225
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.5549646615982056,
      "learning_rate": 7.471206350644687e-07,
      "loss": 1.9559,
      "step": 225226
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.4755712747573853,
      "learning_rate": 7.468321520173182e-07,
      "loss": 2.0652,
      "step": 225227
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.445529818534851,
      "learning_rate": 7.465437246070294e-07,
      "loss": 1.9438,
      "step": 225228
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.4370733499526978,
      "learning_rate": 7.462553528336024e-07,
      "loss": 1.7341,
      "step": 225229
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.6455788612365723,
      "learning_rate": 7.459670366970705e-07,
      "loss": 2.0338,
      "step": 225230
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.4542748928070068,
      "learning_rate": 7.456787761975336e-07,
      "loss": 1.937,
      "step": 225231
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.6665515899658203,
      "learning_rate": 7.45390571335025e-07,
      "loss": 1.8693,
      "step": 225232
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.5950469970703125,
      "learning_rate": 7.451024221095781e-07,
      "loss": 1.9082,
      "step": 225233
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.4375479221343994,
      "learning_rate": 7.448143285212594e-07,
      "loss": 1.9007,
      "step": 225234
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.3345063924789429,
      "learning_rate": 7.445262905701688e-07,
      "loss": 2.11,
      "step": 225235
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.4834301471710205,
      "learning_rate": 7.442383082563064e-07,
      "loss": 1.8885,
      "step": 225236
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.48310124874115,
      "learning_rate": 7.439503815797054e-07,
      "loss": 2.0033,
      "step": 225237
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.4150792360305786,
      "learning_rate": 7.436625105404659e-07,
      "loss": 2.0453,
      "step": 225238
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.5491732358932495,
      "learning_rate": 7.433746951386544e-07,
      "loss": 2.0416,
      "step": 225239
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.480798363685608,
      "learning_rate": 7.430869353742708e-07,
      "loss": 1.9358,
      "step": 225240
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.4576866626739502,
      "learning_rate": 7.427992312474151e-07,
      "loss": 1.7757,
      "step": 225241
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.5867749452590942,
      "learning_rate": 7.425115827581207e-07,
      "loss": 1.9994,
      "step": 225242
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.5036181211471558,
      "learning_rate": 7.422239899064208e-07,
      "loss": 1.8766,
      "step": 225243
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.4759788513183594,
      "learning_rate": 7.419364526924154e-07,
      "loss": 1.94,
      "step": 225244
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.336575984954834,
      "learning_rate": 7.416489711161377e-07,
      "loss": 2.039,
      "step": 225245
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.4210137128829956,
      "learning_rate": 7.413615451776212e-07,
      "loss": 1.9101,
      "step": 225246
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.4876331090927124,
      "learning_rate": 7.410741748769323e-07,
      "loss": 2.1641,
      "step": 225247
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.4393646717071533,
      "learning_rate": 7.407868602141376e-07,
      "loss": 1.8304,
      "step": 225248
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.4229243993759155,
      "learning_rate": 7.404996011892705e-07,
      "loss": 1.9276,
      "step": 225249
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.4277503490447998,
      "learning_rate": 7.402123978023977e-07,
      "loss": 2.0428,
      "step": 225250
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.4824857711791992,
      "learning_rate": 7.399252500535524e-07,
      "loss": 2.0169,
      "step": 225251
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.4618583917617798,
      "learning_rate": 7.396381579428013e-07,
      "loss": 1.9357,
      "step": 225252
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.6270700693130493,
      "learning_rate": 7.393511214702441e-07,
      "loss": 2.0875,
      "step": 225253
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.3073679208755493,
      "learning_rate": 7.390641406358477e-07,
      "loss": 2.034,
      "step": 225254
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.332916498184204,
      "learning_rate": 7.38777215439712e-07,
      "loss": 1.9322,
      "step": 225255
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.3121545314788818,
      "learning_rate": 7.384903458819036e-07,
      "loss": 2.048,
      "step": 225256
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.486258864402771,
      "learning_rate": 7.382035319624558e-07,
      "loss": 2.0286,
      "step": 225257
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.4335377216339111,
      "learning_rate": 7.379167736814351e-07,
      "loss": 2.067,
      "step": 225258
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.860167145729065,
      "learning_rate": 7.37630071038875e-07,
      "loss": 2.1202,
      "step": 225259
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.4878939390182495,
      "learning_rate": 7.373434240348086e-07,
      "loss": 2.0353,
      "step": 225260
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.5757250785827637,
      "learning_rate": 7.370568326693693e-07,
      "loss": 2.1049,
      "step": 225261
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.5630178451538086,
      "learning_rate": 7.367702969425238e-07,
      "loss": 1.8471,
      "step": 225262
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.4231913089752197,
      "learning_rate": 7.364838168543719e-07,
      "loss": 1.8341,
      "step": 225263
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.6588177680969238,
      "learning_rate": 7.361973924049469e-07,
      "loss": 2.0729,
      "step": 225264
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.3672040700912476,
      "learning_rate": 7.359110235943488e-07,
      "loss": 1.9843,
      "step": 225265
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.4991357326507568,
      "learning_rate": 7.356247104225443e-07,
      "loss": 1.8816,
      "step": 225266
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.4084941148757935,
      "learning_rate": 7.353384528896667e-07,
      "loss": 1.9527,
      "step": 225267
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.3696119785308838,
      "learning_rate": 7.35052250995749e-07,
      "loss": 2.083,
      "step": 225268
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.330397367477417,
      "learning_rate": 7.347661047408249e-07,
      "loss": 2.1934,
      "step": 225269
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.447776436805725,
      "learning_rate": 7.344800141249275e-07,
      "loss": 1.8382,
      "step": 225270
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.3826833963394165,
      "learning_rate": 7.341939791481566e-07,
      "loss": 1.8005,
      "step": 225271
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.504150152206421,
      "learning_rate": 7.339079998105791e-07,
      "loss": 2.0038,
      "step": 225272
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.4301468133926392,
      "learning_rate": 7.336220761121947e-07,
      "loss": 1.8625,
      "step": 225273
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.4233424663543701,
      "learning_rate": 7.333362080530703e-07,
      "loss": 1.9304,
      "step": 225274
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.4577728509902954,
      "learning_rate": 7.330503956332723e-07,
      "loss": 1.8806,
      "step": 225275
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.5008854866027832,
      "learning_rate": 7.327646388528341e-07,
      "loss": 2.2033,
      "step": 225276
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.5259777307510376,
      "learning_rate": 7.324789377118556e-07,
      "loss": 1.9907,
      "step": 225277
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.4420019388198853,
      "learning_rate": 7.321932922103369e-07,
      "loss": 1.8059,
      "step": 225278
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.5233066082000732,
      "learning_rate": 7.319077023483777e-07,
      "loss": 2.0801,
      "step": 225279
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.3593915700912476,
      "learning_rate": 7.316221681259782e-07,
      "loss": 1.921,
      "step": 225280
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.50101637840271,
      "learning_rate": 7.313366895432049e-07,
      "loss": 2.0181,
      "step": 225281
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.4369263648986816,
      "learning_rate": 7.310512666001578e-07,
      "loss": 1.756,
      "step": 225282
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.4791988134384155,
      "learning_rate": 7.307658992968368e-07,
      "loss": 1.9362,
      "step": 225283
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.4509445428848267,
      "learning_rate": 7.304805876333419e-07,
      "loss": 1.9737,
      "step": 225284
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.4724576473236084,
      "learning_rate": 7.301953316097064e-07,
      "loss": 1.8117,
      "step": 225285
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.5627448558807373,
      "learning_rate": 7.299101312259304e-07,
      "loss": 2.1663,
      "step": 225286
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.5262044668197632,
      "learning_rate": 7.296249864821468e-07,
      "loss": 1.9551,
      "step": 225287
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.5732084512710571,
      "learning_rate": 7.29339897378356e-07,
      "loss": 2.1485,
      "step": 225288
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.6146568059921265,
      "learning_rate": 7.290548639146576e-07,
      "loss": 1.9679,
      "step": 225289
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.4887230396270752,
      "learning_rate": 7.287698860910519e-07,
      "loss": 1.9434,
      "step": 225290
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.3990758657455444,
      "learning_rate": 7.284849639076384e-07,
      "loss": 1.7843,
      "step": 225291
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.5234516859054565,
      "learning_rate": 7.282000973644175e-07,
      "loss": 1.9496,
      "step": 225292
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.5085557699203491,
      "learning_rate": 7.27915286461489e-07,
      "loss": 1.8627,
      "step": 225293
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.492932677268982,
      "learning_rate": 7.276305311988861e-07,
      "loss": 2.2074,
      "step": 225294
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.442869782447815,
      "learning_rate": 7.273458315766756e-07,
      "loss": 1.8529,
      "step": 225295
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.5130054950714111,
      "learning_rate": 7.27061187594924e-07,
      "loss": 2.0192,
      "step": 225296
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.3750462532043457,
      "learning_rate": 7.267765992536312e-07,
      "loss": 2.0862,
      "step": 225297
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.5008665323257446,
      "learning_rate": 7.26492066552864e-07,
      "loss": 2.0284,
      "step": 225298
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.5627001523971558,
      "learning_rate": 7.262075894927222e-07,
      "loss": 1.8473,
      "step": 225299
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.5180163383483887,
      "learning_rate": 7.259231680732059e-07,
      "loss": 2.0069,
      "step": 225300
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.5043402910232544,
      "learning_rate": 7.256388022944148e-07,
      "loss": 1.8572,
      "step": 225301
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.5103764533996582,
      "learning_rate": 7.253544921563492e-07,
      "loss": 1.9181,
      "step": 225302
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.416323184967041,
      "learning_rate": 7.250702376591089e-07,
      "loss": 1.9997,
      "step": 225303
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.4826078414916992,
      "learning_rate": 7.247860388027271e-07,
      "loss": 1.9481,
      "step": 225304
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.5969600677490234,
      "learning_rate": 7.245018955872372e-07,
      "loss": 2.1632,
      "step": 225305
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.4106223583221436,
      "learning_rate": 7.242178080127392e-07,
      "loss": 1.8712,
      "step": 225306
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.5433778762817383,
      "learning_rate": 7.239337760792329e-07,
      "loss": 2.0545,
      "step": 225307
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.5319747924804688,
      "learning_rate": 7.236497997868185e-07,
      "loss": 2.0742,
      "step": 225308
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.5372705459594727,
      "learning_rate": 7.233658791354957e-07,
      "loss": 1.99,
      "step": 225309
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.5410571098327637,
      "learning_rate": 7.230820141253646e-07,
      "loss": 1.8911,
      "step": 225310
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.510716199874878,
      "learning_rate": 7.227982047564917e-07,
      "loss": 2.2377,
      "step": 225311
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.483943223953247,
      "learning_rate": 7.225144510288439e-07,
      "loss": 2.0004,
      "step": 225312
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.429046869277954,
      "learning_rate": 7.222307529425875e-07,
      "loss": 1.8332,
      "step": 225313
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.595632553100586,
      "learning_rate": 7.219471104976893e-07,
      "loss": 2.0215,
      "step": 225314
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.4042418003082275,
      "learning_rate": 7.216635236942159e-07,
      "loss": 2.0312,
      "step": 225315
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.5124410390853882,
      "learning_rate": 7.213799925322672e-07,
      "loss": 1.8596,
      "step": 225316
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.4771078824996948,
      "learning_rate": 7.210965170118433e-07,
      "loss": 2.1377,
      "step": 225317
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.5444473028182983,
      "learning_rate": 7.208130971330107e-07,
      "loss": 2.0273,
      "step": 225318
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.4441208839416504,
      "learning_rate": 7.205297328958359e-07,
      "loss": 1.8936,
      "step": 225319
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.471899151802063,
      "learning_rate": 7.202464243003525e-07,
      "loss": 2.0011,
      "step": 225320
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.4643714427947998,
      "learning_rate": 7.199631713466602e-07,
      "loss": 2.0293,
      "step": 225321
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.5985937118530273,
      "learning_rate": 7.196799740347259e-07,
      "loss": 1.9577,
      "step": 225322
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.4882714748382568,
      "learning_rate": 7.193968323647159e-07,
      "loss": 1.9606,
      "step": 225323
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.4555473327636719,
      "learning_rate": 7.191137463365637e-07,
      "loss": 1.9288,
      "step": 225324
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.4444948434829712,
      "learning_rate": 7.188307159504026e-07,
      "loss": 1.9999,
      "step": 225325
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.5239614248275757,
      "learning_rate": 7.185477412062657e-07,
      "loss": 1.9229,
      "step": 225326
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.4198291301727295,
      "learning_rate": 7.182648221041865e-07,
      "loss": 2.1008,
      "step": 225327
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.578796148300171,
      "learning_rate": 7.179819586442315e-07,
      "loss": 1.9351,
      "step": 225328
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.4338635206222534,
      "learning_rate": 7.176991508264341e-07,
      "loss": 1.8599,
      "step": 225329
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.4210574626922607,
      "learning_rate": 7.174163986508941e-07,
      "loss": 1.6947,
      "step": 225330
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.4776742458343506,
      "learning_rate": 7.171337021176449e-07,
      "loss": 2.0273,
      "step": 225331
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.5590626001358032,
      "learning_rate": 7.168510612266865e-07,
      "loss": 1.9149,
      "step": 225332
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.3744196891784668,
      "learning_rate": 7.16568475978152e-07,
      "loss": 2.0749,
      "step": 225333
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.3845943212509155,
      "learning_rate": 7.162859463720416e-07,
      "loss": 2.0158,
      "step": 225334
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.5053398609161377,
      "learning_rate": 7.160034724084552e-07,
      "loss": 2.078,
      "step": 225335
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.5084679126739502,
      "learning_rate": 7.157210540873593e-07,
      "loss": 2.0028,
      "step": 225336
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.429701328277588,
      "learning_rate": 7.154386914088872e-07,
      "loss": 2.1501,
      "step": 225337
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.3590501546859741,
      "learning_rate": 7.151563843730723e-07,
      "loss": 2.036,
      "step": 225338
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.66060209274292,
      "learning_rate": 7.148741329799812e-07,
      "loss": 2.236,
      "step": 225339
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.4766016006469727,
      "learning_rate": 7.145919372295805e-07,
      "loss": 1.9908,
      "step": 225340
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.4398212432861328,
      "learning_rate": 7.143097971220369e-07,
      "loss": 2.0949,
      "step": 225341
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.4906233549118042,
      "learning_rate": 7.140277126573168e-07,
      "loss": 1.8679,
      "step": 225342
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.3964325189590454,
      "learning_rate": 7.137456838355537e-07,
      "loss": 1.8583,
      "step": 225343
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.484801173210144,
      "learning_rate": 7.134637106567143e-07,
      "loss": 1.9093,
      "step": 225344
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.458765983581543,
      "learning_rate": 7.131817931208983e-07,
      "loss": 1.8952,
      "step": 225345
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.4752711057662964,
      "learning_rate": 7.128999312281724e-07,
      "loss": 2.1441,
      "step": 225346
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.492266297340393,
      "learning_rate": 7.1261812497857e-07,
      "loss": 1.8932,
      "step": 225347
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.6107581853866577,
      "learning_rate": 7.123363743721244e-07,
      "loss": 2.035,
      "step": 225348
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.4731587171554565,
      "learning_rate": 7.120546794089021e-07,
      "loss": 2.2167,
      "step": 225349
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.5428680181503296,
      "learning_rate": 7.117730400889365e-07,
      "loss": 2.0417,
      "step": 225350
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.4183144569396973,
      "learning_rate": 7.114914564123609e-07,
      "loss": 2.1448,
      "step": 225351
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.5096497535705566,
      "learning_rate": 7.112099283791417e-07,
      "loss": 1.92,
      "step": 225352
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.6901466846466064,
      "learning_rate": 7.109284559893457e-07,
      "loss": 1.9522,
      "step": 225353
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.5310711860656738,
      "learning_rate": 7.106470392430397e-07,
      "loss": 1.9614,
      "step": 225354
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.4298542737960815,
      "learning_rate": 7.1036567814029e-07,
      "loss": 2.1977,
      "step": 225355
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.4964263439178467,
      "learning_rate": 7.100843726810967e-07,
      "loss": 2.0836,
      "step": 225356
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.4218553304672241,
      "learning_rate": 7.098031228655932e-07,
      "loss": 2.1748,
      "step": 225357
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.4489895105361938,
      "learning_rate": 7.095219286937459e-07,
      "loss": 1.7486,
      "step": 225358
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.4477803707122803,
      "learning_rate": 7.092407901656882e-07,
      "loss": 2.1268,
      "step": 225359
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.5190366506576538,
      "learning_rate": 7.089597072813868e-07,
      "loss": 2.2365,
      "step": 225360
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.5484217405319214,
      "learning_rate": 7.08678680040975e-07,
      "loss": 1.8592,
      "step": 225361
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.4370156526565552,
      "learning_rate": 7.083977084444525e-07,
      "loss": 1.8847,
      "step": 225362
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.567515254020691,
      "learning_rate": 7.081167924919195e-07,
      "loss": 1.8358,
      "step": 225363
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.5705065727233887,
      "learning_rate": 7.07835932183376e-07,
      "loss": 1.8623,
      "step": 225364
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.4454070329666138,
      "learning_rate": 7.075551275188884e-07,
      "loss": 2.0716,
      "step": 225365
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.410996675491333,
      "learning_rate": 7.072743784985236e-07,
      "loss": 2.1671,
      "step": 225366
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.5764228105545044,
      "learning_rate": 7.069936851223146e-07,
      "loss": 1.9846,
      "step": 225367
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.3078482151031494,
      "learning_rate": 7.067130473903282e-07,
      "loss": 1.9762,
      "step": 225368
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.4126482009887695,
      "learning_rate": 7.06432465302631e-07,
      "loss": 2.0683,
      "step": 225369
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.4983248710632324,
      "learning_rate": 7.061519388592562e-07,
      "loss": 2.0776,
      "step": 225370
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.434901237487793,
      "learning_rate": 7.058714680602373e-07,
      "loss": 1.8642,
      "step": 225371
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.5413923263549805,
      "learning_rate": 7.055910529056408e-07,
      "loss": 2.0204,
      "step": 225372
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.511036992073059,
      "learning_rate": 7.053106933955332e-07,
      "loss": 2.0198,
      "step": 225373
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.4921165704727173,
      "learning_rate": 7.050303895299814e-07,
      "loss": 1.9754,
      "step": 225374
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.4351823329925537,
      "learning_rate": 7.047501413089851e-07,
      "loss": 1.9805,
      "step": 225375
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.5678313970565796,
      "learning_rate": 7.044699487326444e-07,
      "loss": 1.9311,
      "step": 225376
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.3809493780136108,
      "learning_rate": 7.041898118009925e-07,
      "loss": 1.8164,
      "step": 225377
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.4293196201324463,
      "learning_rate": 7.039097305140629e-07,
      "loss": 2.0358,
      "step": 225378
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.3608911037445068,
      "learning_rate": 7.036297048719219e-07,
      "loss": 1.8062,
      "step": 225379
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.634751796722412,
      "learning_rate": 7.033497348746697e-07,
      "loss": 1.9774,
      "step": 225380
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.4299424886703491,
      "learning_rate": 7.030698205222729e-07,
      "loss": 1.6297,
      "step": 225381
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.3991363048553467,
      "learning_rate": 7.027899618148314e-07,
      "loss": 1.811,
      "step": 225382
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.448493242263794,
      "learning_rate": 7.025101587524118e-07,
      "loss": 2.0603,
      "step": 225383
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.551684856414795,
      "learning_rate": 7.022304113350141e-07,
      "loss": 2.0251,
      "step": 225384
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.5045028924942017,
      "learning_rate": 7.019507195627383e-07,
      "loss": 1.9703,
      "step": 225385
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.361992359161377,
      "learning_rate": 7.016710834356176e-07,
      "loss": 1.8415,
      "step": 225386
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.5063503980636597,
      "learning_rate": 7.013915029537187e-07,
      "loss": 2.0838,
      "step": 225387
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.4913396835327148,
      "learning_rate": 7.011119781170749e-07,
      "loss": 1.963,
      "step": 225388
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.3883702754974365,
      "learning_rate": 7.008325089257195e-07,
      "loss": 1.8961,
      "step": 225389
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.420457363128662,
      "learning_rate": 7.005530953797522e-07,
      "loss": 1.7797,
      "step": 225390
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.5118160247802734,
      "learning_rate": 7.002737374792067e-07,
      "loss": 2.0233,
      "step": 225391
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.4292361736297607,
      "learning_rate": 6.99994435224116e-07,
      "loss": 1.8619,
      "step": 225392
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.4884579181671143,
      "learning_rate": 6.997151886145468e-07,
      "loss": 2.0249,
      "step": 225393
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.4600398540496826,
      "learning_rate": 6.994359976505659e-07,
      "loss": 1.7865,
      "step": 225394
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.4795422554016113,
      "learning_rate": 6.991568623322063e-07,
      "loss": 1.9344,
      "step": 225395
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.4677878618240356,
      "learning_rate": 6.988777826595015e-07,
      "loss": 2.0571,
      "step": 225396
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.476900577545166,
      "learning_rate": 6.985987586325514e-07,
      "loss": 1.9101,
      "step": 225397
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.484337568283081,
      "learning_rate": 6.98319790251356e-07,
      "loss": 2.0032,
      "step": 225398
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.4717521667480469,
      "learning_rate": 6.980408775160151e-07,
      "loss": 2.1453,
      "step": 225399
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.441440224647522,
      "learning_rate": 6.977620204265622e-07,
      "loss": 1.9509,
      "step": 225400
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.430943489074707,
      "learning_rate": 6.974832189830304e-07,
      "loss": 1.855,
      "step": 225401
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.3758482933044434,
      "learning_rate": 6.972044731854864e-07,
      "loss": 1.7839,
      "step": 225402
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.470115065574646,
      "learning_rate": 6.969257830339637e-07,
      "loss": 1.9868,
      "step": 225403
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.2823072671890259,
      "learning_rate": 6.966471485285619e-07,
      "loss": 2.0042,
      "step": 225404
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.49308443069458,
      "learning_rate": 6.963685696692811e-07,
      "loss": 1.895,
      "step": 225405
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.3672510385513306,
      "learning_rate": 6.960900464562214e-07,
      "loss": 1.8311,
      "step": 225406
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.4941433668136597,
      "learning_rate": 6.958115788893825e-07,
      "loss": 1.9083,
      "step": 225407
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.3021007776260376,
      "learning_rate": 6.955331669688646e-07,
      "loss": 1.936,
      "step": 225408
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.550508975982666,
      "learning_rate": 6.952548106946676e-07,
      "loss": 2.08,
      "step": 225409
    },
    {
      "epoch": 2.93,
      "grad_norm": 1.6152311563491821,
      "learning_rate": 6.949765100669246e-07,
      "loss": 1.9503,
      "step": 225410
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.6256107091903687,
      "learning_rate": 6.946982650855692e-07,
      "loss": 2.0676,
      "step": 225411
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.4955008029937744,
      "learning_rate": 6.944200757507678e-07,
      "loss": 2.063,
      "step": 225412
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.6077927350997925,
      "learning_rate": 6.941419420625205e-07,
      "loss": 2.0415,
      "step": 225413
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.4152379035949707,
      "learning_rate": 6.938638640208604e-07,
      "loss": 1.7626,
      "step": 225414
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.5155401229858398,
      "learning_rate": 6.935858416258544e-07,
      "loss": 1.9598,
      "step": 225415
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.4581794738769531,
      "learning_rate": 6.933078748776022e-07,
      "loss": 1.8482,
      "step": 225416
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.333939552307129,
      "learning_rate": 6.930299637760706e-07,
      "loss": 1.7865,
      "step": 225417
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.4693291187286377,
      "learning_rate": 6.927521083213928e-07,
      "loss": 1.9699,
      "step": 225418
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.3504000902175903,
      "learning_rate": 6.924743085135353e-07,
      "loss": 1.8961,
      "step": 225419
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.366894245147705,
      "learning_rate": 6.921965643526317e-07,
      "loss": 1.9754,
      "step": 225420
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.7581586837768555,
      "learning_rate": 6.919188758386818e-07,
      "loss": 1.9063,
      "step": 225421
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.5932880640029907,
      "learning_rate": 6.916412429717522e-07,
      "loss": 1.9467,
      "step": 225422
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.4876364469528198,
      "learning_rate": 6.913636657519095e-07,
      "loss": 2.0209,
      "step": 225423
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.4791908264160156,
      "learning_rate": 6.910861441792204e-07,
      "loss": 1.9563,
      "step": 225424
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.4942659139633179,
      "learning_rate": 6.908086782536515e-07,
      "loss": 1.906,
      "step": 225425
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.4814308881759644,
      "learning_rate": 6.905312679753361e-07,
      "loss": 1.9867,
      "step": 225426
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.483912467956543,
      "learning_rate": 6.902539133443074e-07,
      "loss": 1.7825,
      "step": 225427
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.5328984260559082,
      "learning_rate": 6.899766143605989e-07,
      "loss": 1.9889,
      "step": 225428
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.4655789136886597,
      "learning_rate": 6.896993710242771e-07,
      "loss": 2.0265,
      "step": 225429
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.7369472980499268,
      "learning_rate": 6.894221833353752e-07,
      "loss": 1.872,
      "step": 225430
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.3241572380065918,
      "learning_rate": 6.891450512939933e-07,
      "loss": 1.7058,
      "step": 225431
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.566791296005249,
      "learning_rate": 6.888679749001314e-07,
      "loss": 2.0884,
      "step": 225432
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.4655122756958008,
      "learning_rate": 6.885909541538559e-07,
      "loss": 1.8327,
      "step": 225433
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.489836573600769,
      "learning_rate": 6.883139890552336e-07,
      "loss": 1.9786,
      "step": 225434
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.485864520072937,
      "learning_rate": 6.880370796042978e-07,
      "loss": 1.9843,
      "step": 225435
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.4621714353561401,
      "learning_rate": 6.877602258010817e-07,
      "loss": 1.9204,
      "step": 225436
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.4582228660583496,
      "learning_rate": 6.874834276456853e-07,
      "loss": 1.9161,
      "step": 225437
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.524627685546875,
      "learning_rate": 6.872066851381419e-07,
      "loss": 1.9964,
      "step": 225438
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.4912930727005005,
      "learning_rate": 6.869299982784849e-07,
      "loss": 2.1925,
      "step": 225439
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.648527979850769,
      "learning_rate": 6.866533670667807e-07,
      "loss": 1.9675,
      "step": 225440
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.4166724681854248,
      "learning_rate": 6.863767915030627e-07,
      "loss": 1.8238,
      "step": 225441
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.446706771850586,
      "learning_rate": 6.861002715874308e-07,
      "loss": 1.9319,
      "step": 225442
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.404236078262329,
      "learning_rate": 6.858238073198852e-07,
      "loss": 1.8486,
      "step": 225443
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.4676311016082764,
      "learning_rate": 6.855473987004923e-07,
      "loss": 1.9453,
      "step": 225444
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.493600845336914,
      "learning_rate": 6.852710457293187e-07,
      "loss": 1.791,
      "step": 225445
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.5254567861557007,
      "learning_rate": 6.849947484063645e-07,
      "loss": 2.0334,
      "step": 225446
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.807647943496704,
      "learning_rate": 6.847185067317629e-07,
      "loss": 2.0918,
      "step": 225447
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.391631841659546,
      "learning_rate": 6.844423207055138e-07,
      "loss": 1.8977,
      "step": 225448
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.5551137924194336,
      "learning_rate": 6.841661903276508e-07,
      "loss": 1.8098,
      "step": 225449
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.3383885622024536,
      "learning_rate": 6.838901155982734e-07,
      "loss": 2.0252,
      "step": 225450
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.4891512393951416,
      "learning_rate": 6.836140965174153e-07,
      "loss": 1.9262,
      "step": 225451
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.6018470525741577,
      "learning_rate": 6.833381330851096e-07,
      "loss": 2.0888,
      "step": 225452
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.4991140365600586,
      "learning_rate": 6.830622253014229e-07,
      "loss": 2.103,
      "step": 225453
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.644314169883728,
      "learning_rate": 6.827863731664218e-07,
      "loss": 1.8107,
      "step": 225454
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.6293061971664429,
      "learning_rate": 6.825105766801064e-07,
      "loss": 1.8628,
      "step": 225455
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.4406434297561646,
      "learning_rate": 6.8223483584261e-07,
      "loss": 2.0194,
      "step": 225456
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.5289424657821655,
      "learning_rate": 6.819591506538991e-07,
      "loss": 1.8928,
      "step": 225457
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.535495400428772,
      "learning_rate": 6.816835211140736e-07,
      "loss": 1.921,
      "step": 225458
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.397039771080017,
      "learning_rate": 6.81407947223167e-07,
      "loss": 1.6958,
      "step": 225459
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.4719034433364868,
      "learning_rate": 6.811324289812459e-07,
      "loss": 1.9465,
      "step": 225460
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.4156306982040405,
      "learning_rate": 6.808569663883434e-07,
      "loss": 2.1088,
      "step": 225461
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.5482842922210693,
      "learning_rate": 6.805815594445263e-07,
      "loss": 1.9456,
      "step": 225462
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.4926859140396118,
      "learning_rate": 6.803062081498278e-07,
      "loss": 1.9162,
      "step": 225463
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.4873861074447632,
      "learning_rate": 6.800309125043146e-07,
      "loss": 1.9554,
      "step": 225464
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.5234227180480957,
      "learning_rate": 6.797556725080533e-07,
      "loss": 1.902,
      "step": 225465
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.5519664287567139,
      "learning_rate": 6.794804881610438e-07,
      "loss": 2.2622,
      "step": 225466
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.4805535078048706,
      "learning_rate": 6.792053594634195e-07,
      "loss": 2.071,
      "step": 225467
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.5363128185272217,
      "learning_rate": 6.789302864151469e-07,
      "loss": 2.036,
      "step": 225468
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.6289342641830444,
      "learning_rate": 6.786552690162928e-07,
      "loss": 2.0712,
      "step": 225469
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.5124022960662842,
      "learning_rate": 6.783803072669569e-07,
      "loss": 1.9233,
      "step": 225470
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.3621234893798828,
      "learning_rate": 6.781054011671394e-07,
      "loss": 2.0352,
      "step": 225471
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.392476201057434,
      "learning_rate": 6.778305507169402e-07,
      "loss": 2.0884,
      "step": 225472
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.6228231191635132,
      "learning_rate": 6.775557559163591e-07,
      "loss": 2.0098,
      "step": 225473
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.5983037948608398,
      "learning_rate": 6.772810167654963e-07,
      "loss": 1.987,
      "step": 225474
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.4841846227645874,
      "learning_rate": 6.770063332643516e-07,
      "loss": 1.8033,
      "step": 225475
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.515824556350708,
      "learning_rate": 6.767317054129917e-07,
      "loss": 2.1562,
      "step": 225476
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.4521819353103638,
      "learning_rate": 6.764571332115166e-07,
      "loss": 1.9659,
      "step": 225477
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.445988655090332,
      "learning_rate": 6.761826166598927e-07,
      "loss": 1.9952,
      "step": 225478
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.4712895154953003,
      "learning_rate": 6.759081557582535e-07,
      "loss": 2.0403,
      "step": 225479
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.5070096254348755,
      "learning_rate": 6.756337505065989e-07,
      "loss": 1.8922,
      "step": 225480
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.4066294431686401,
      "learning_rate": 6.753594009049957e-07,
      "loss": 1.9456,
      "step": 225481
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.5276941061019897,
      "learning_rate": 6.750851069534768e-07,
      "loss": 1.9429,
      "step": 225482
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.4212052822113037,
      "learning_rate": 6.748108686521425e-07,
      "loss": 1.7806,
      "step": 225483
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.5169533491134644,
      "learning_rate": 6.745366860009927e-07,
      "loss": 2.144,
      "step": 225484
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.5496615171432495,
      "learning_rate": 6.742625590000939e-07,
      "loss": 2.0453,
      "step": 225485
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.5872780084609985,
      "learning_rate": 6.739884876495128e-07,
      "loss": 1.8637,
      "step": 225486
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.3662782907485962,
      "learning_rate": 6.737144719492826e-07,
      "loss": 1.7549,
      "step": 225487
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.4238513708114624,
      "learning_rate": 6.734405118994368e-07,
      "loss": 1.9026,
      "step": 225488
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.5808134078979492,
      "learning_rate": 6.731666075000752e-07,
      "loss": 1.9585,
      "step": 225489
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.5172743797302246,
      "learning_rate": 6.728927587511979e-07,
      "loss": 2.1279,
      "step": 225490
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.3838746547698975,
      "learning_rate": 6.726189656529046e-07,
      "loss": 1.7961,
      "step": 225491
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.4861726760864258,
      "learning_rate": 6.723452282051955e-07,
      "loss": 1.9111,
      "step": 225492
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.527747392654419,
      "learning_rate": 6.720715464081705e-07,
      "loss": 2.068,
      "step": 225493
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.4514533281326294,
      "learning_rate": 6.717979202618628e-07,
      "loss": 1.9655,
      "step": 225494
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.4414979219436646,
      "learning_rate": 6.715243497663059e-07,
      "loss": 1.9363,
      "step": 225495
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.4695459604263306,
      "learning_rate": 6.712508349215662e-07,
      "loss": 1.9884,
      "step": 225496
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.6073487997055054,
      "learning_rate": 6.709773757276771e-07,
      "loss": 1.9909,
      "step": 225497
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.6042792797088623,
      "learning_rate": 6.707039721847385e-07,
      "loss": 1.8523,
      "step": 225498
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.535890817642212,
      "learning_rate": 6.704306242927505e-07,
      "loss": 1.9963,
      "step": 225499
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.4629569053649902,
      "learning_rate": 6.701573320517462e-07,
      "loss": 1.7781,
      "step": 225500
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.5568811893463135,
      "learning_rate": 6.69884095461859e-07,
      "loss": 1.9487,
      "step": 225501
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.520453691482544,
      "learning_rate": 6.696109145230554e-07,
      "loss": 1.9122,
      "step": 225502
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.4757850170135498,
      "learning_rate": 6.693377892354357e-07,
      "loss": 1.9393,
      "step": 225503
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.4407696723937988,
      "learning_rate": 6.690647195990328e-07,
      "loss": 1.8327,
      "step": 225504
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.897910237312317,
      "learning_rate": 6.687917056139136e-07,
      "loss": 2.1846,
      "step": 225505
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.6265701055526733,
      "learning_rate": 6.68518747280078e-07,
      "loss": 1.9575,
      "step": 225506
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.5907176733016968,
      "learning_rate": 6.68245844597659e-07,
      "loss": 1.8593,
      "step": 225507
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.5236080884933472,
      "learning_rate": 6.67972997566657e-07,
      "loss": 2.0994,
      "step": 225508
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.5162602663040161,
      "learning_rate": 6.677002061871051e-07,
      "loss": 1.9436,
      "step": 225509
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.4787745475769043,
      "learning_rate": 6.674274704591032e-07,
      "loss": 1.8644,
      "step": 225510
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.471101999282837,
      "learning_rate": 6.671547903826513e-07,
      "loss": 1.9413,
      "step": 225511
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.458147406578064,
      "learning_rate": 6.668821659578493e-07,
      "loss": 2.1597,
      "step": 225512
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.3537414073944092,
      "learning_rate": 6.666095971847308e-07,
      "loss": 1.6951,
      "step": 225513
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.5204951763153076,
      "learning_rate": 6.663370840632953e-07,
      "loss": 2.2428,
      "step": 225514
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.3259978294372559,
      "learning_rate": 6.660646265936765e-07,
      "loss": 2.1095,
      "step": 225515
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.535199761390686,
      "learning_rate": 6.65792224775874e-07,
      "loss": 2.0508,
      "step": 225516
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.5270121097564697,
      "learning_rate": 6.655198786099547e-07,
      "loss": 1.9678,
      "step": 225517
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.372196078300476,
      "learning_rate": 6.652475880959851e-07,
      "loss": 1.7269,
      "step": 225518
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.6447038650512695,
      "learning_rate": 6.649753532339653e-07,
      "loss": 1.9731,
      "step": 225519
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.4775389432907104,
      "learning_rate": 6.64703174023995e-07,
      "loss": 1.8027,
      "step": 225520
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.5023454427719116,
      "learning_rate": 6.644310504661076e-07,
      "loss": 1.9469,
      "step": 225521
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.3870223760604858,
      "learning_rate": 6.641589825603366e-07,
      "loss": 2.005,
      "step": 225522
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.4870405197143555,
      "learning_rate": 6.638869703067484e-07,
      "loss": 1.9543,
      "step": 225523
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.3206554651260376,
      "learning_rate": 6.636150137054097e-07,
      "loss": 1.9189,
      "step": 225524
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.6488717794418335,
      "learning_rate": 6.633431127563538e-07,
      "loss": 1.8349,
      "step": 225525
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.3950196504592896,
      "learning_rate": 6.63071267459614e-07,
      "loss": 1.9418,
      "step": 225526
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.5501139163970947,
      "learning_rate": 6.627994778152901e-07,
      "loss": 1.82,
      "step": 225527
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.499659776687622,
      "learning_rate": 6.625277438233823e-07,
      "loss": 1.9945,
      "step": 225528
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.4157027006149292,
      "learning_rate": 6.62256065483957e-07,
      "loss": 1.9649,
      "step": 225529
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.6151797771453857,
      "learning_rate": 6.619844427970811e-07,
      "loss": 2.0004,
      "step": 225530
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.4313886165618896,
      "learning_rate": 6.617128757627876e-07,
      "loss": 1.8689,
      "step": 225531
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.4313832521438599,
      "learning_rate": 6.614413643811434e-07,
      "loss": 2.0296,
      "step": 225532
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.3083302974700928,
      "learning_rate": 6.611699086521815e-07,
      "loss": 2.0353,
      "step": 225533
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.4732365608215332,
      "learning_rate": 6.608985085759355e-07,
      "loss": 2.0248,
      "step": 225534
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.5312262773513794,
      "learning_rate": 6.606271641525051e-07,
      "loss": 2.026,
      "step": 225535
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.3268382549285889,
      "learning_rate": 6.603558753818905e-07,
      "loss": 2.1548,
      "step": 225536
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.5633642673492432,
      "learning_rate": 6.600846422641914e-07,
      "loss": 1.8215,
      "step": 225537
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.344442367553711,
      "learning_rate": 6.598134647994413e-07,
      "loss": 2.084,
      "step": 225538
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.5355905294418335,
      "learning_rate": 6.595423429876401e-07,
      "loss": 1.9795,
      "step": 225539
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.396844506263733,
      "learning_rate": 6.59271276828921e-07,
      "loss": 1.6844,
      "step": 225540
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.3171701431274414,
      "learning_rate": 6.590002663232841e-07,
      "loss": 1.8719,
      "step": 225541
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.4244227409362793,
      "learning_rate": 6.587293114707626e-07,
      "loss": 2.1203,
      "step": 225542
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.533097505569458,
      "learning_rate": 6.584584122714564e-07,
      "loss": 1.9602,
      "step": 225543
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.5123354196548462,
      "learning_rate": 6.58187568725399e-07,
      "loss": 2.1631,
      "step": 225544
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.5542539358139038,
      "learning_rate": 6.579167808326235e-07,
      "loss": 2.0596,
      "step": 225545
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.4389551877975464,
      "learning_rate": 6.576460485931967e-07,
      "loss": 2.1362,
      "step": 225546
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.6110366582870483,
      "learning_rate": 6.57375372007185e-07,
      "loss": 2.0843,
      "step": 225547
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.4015694856643677,
      "learning_rate": 6.571047510745886e-07,
      "loss": 1.9492,
      "step": 225548
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.738324522972107,
      "learning_rate": 6.56834185795474e-07,
      "loss": 2.0009,
      "step": 225549
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.490401268005371,
      "learning_rate": 6.565636761699411e-07,
      "loss": 2.1118,
      "step": 225550
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.4439575672149658,
      "learning_rate": 6.5629322219799e-07,
      "loss": 1.9462,
      "step": 225551
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.5991028547286987,
      "learning_rate": 6.560228238796872e-07,
      "loss": 1.9956,
      "step": 225552
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.4924854040145874,
      "learning_rate": 6.55752481215066e-07,
      "loss": 1.9084,
      "step": 225553
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.3671753406524658,
      "learning_rate": 6.554821942041932e-07,
      "loss": 1.8959,
      "step": 225554
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.6484425067901611,
      "learning_rate": 6.552119628471353e-07,
      "loss": 1.97,
      "step": 225555
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.4920631647109985,
      "learning_rate": 6.549417871439255e-07,
      "loss": 2.0681,
      "step": 225556
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.5504169464111328,
      "learning_rate": 6.546716670945973e-07,
      "loss": 2.0758,
      "step": 225557
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.583953857421875,
      "learning_rate": 6.544016026992171e-07,
      "loss": 2.1183,
      "step": 225558
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.5951213836669922,
      "learning_rate": 6.541315939578185e-07,
      "loss": 1.9618,
      "step": 225559
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.3246761560440063,
      "learning_rate": 6.53861640870501e-07,
      "loss": 1.9271,
      "step": 225560
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.590800404548645,
      "learning_rate": 6.535917434372651e-07,
      "loss": 2.178,
      "step": 225561
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.4798235893249512,
      "learning_rate": 6.53321901658177e-07,
      "loss": 1.7873,
      "step": 225562
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.4460467100143433,
      "learning_rate": 6.530521155332702e-07,
      "loss": 1.8974,
      "step": 225563
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.5703476667404175,
      "learning_rate": 6.527823850626446e-07,
      "loss": 2.1408,
      "step": 225564
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.388910174369812,
      "learning_rate": 6.525127102463002e-07,
      "loss": 1.8259,
      "step": 225565
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.74524986743927,
      "learning_rate": 6.522430910843035e-07,
      "loss": 2.1104,
      "step": 225566
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.698325276374817,
      "learning_rate": 6.519735275766879e-07,
      "loss": 1.9177,
      "step": 225567
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.4482065439224243,
      "learning_rate": 6.517040197235535e-07,
      "loss": 2.0734,
      "step": 225568
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.414811611175537,
      "learning_rate": 6.514345675249e-07,
      "loss": 2.0237,
      "step": 225569
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.3330717086791992,
      "learning_rate": 6.511651709808274e-07,
      "loss": 2.0635,
      "step": 225570
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.4756276607513428,
      "learning_rate": 6.508958300913025e-07,
      "loss": 2.2408,
      "step": 225571
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.5399187803268433,
      "learning_rate": 6.506265448564585e-07,
      "loss": 1.974,
      "step": 225572
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.557661771774292,
      "learning_rate": 6.503573152762953e-07,
      "loss": 1.824,
      "step": 225573
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.538455605506897,
      "learning_rate": 6.500881413509129e-07,
      "loss": 2.0215,
      "step": 225574
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.4265196323394775,
      "learning_rate": 6.49819023080278e-07,
      "loss": 2.0465,
      "step": 225575
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.706648588180542,
      "learning_rate": 6.495499604645571e-07,
      "loss": 1.7877,
      "step": 225576
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.5695120096206665,
      "learning_rate": 6.492809535036836e-07,
      "loss": 2.1453,
      "step": 225577
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.4788039922714233,
      "learning_rate": 6.490120021977907e-07,
      "loss": 1.7778,
      "step": 225578
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.4240316152572632,
      "learning_rate": 6.487431065468785e-07,
      "loss": 1.9963,
      "step": 225579
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.423659086227417,
      "learning_rate": 6.484742665510134e-07,
      "loss": 1.8266,
      "step": 225580
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.4602962732315063,
      "learning_rate": 6.482054822102622e-07,
      "loss": 2.0005,
      "step": 225581
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.4735596179962158,
      "learning_rate": 6.479367535246582e-07,
      "loss": 1.8286,
      "step": 225582
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.3692208528518677,
      "learning_rate": 6.476680804942347e-07,
      "loss": 1.9428,
      "step": 225583
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.4311084747314453,
      "learning_rate": 6.473994631190916e-07,
      "loss": 1.8873,
      "step": 225584
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.4853389263153076,
      "learning_rate": 6.471309013992287e-07,
      "loss": 2.1801,
      "step": 225585
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.4235866069793701,
      "learning_rate": 6.46862395334713e-07,
      "loss": 1.7151,
      "step": 225586
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.4521234035491943,
      "learning_rate": 6.465939449256107e-07,
      "loss": 1.8287,
      "step": 225587
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.4945272207260132,
      "learning_rate": 6.463255501719555e-07,
      "loss": 2.1066,
      "step": 225588
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.6041450500488281,
      "learning_rate": 6.460572110737805e-07,
      "loss": 1.9069,
      "step": 225589
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.398077130317688,
      "learning_rate": 6.457889276311523e-07,
      "loss": 2.0449,
      "step": 225590
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.4720319509506226,
      "learning_rate": 6.455206998441376e-07,
      "loss": 2.1093,
      "step": 225591
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.6186976432800293,
      "learning_rate": 6.452525277127696e-07,
      "loss": 2.0992,
      "step": 225592
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.296615719795227,
      "learning_rate": 6.449844112370816e-07,
      "loss": 2.2491,
      "step": 225593
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.4797159433364868,
      "learning_rate": 6.447163504171737e-07,
      "loss": 2.1027,
      "step": 225594
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.5644092559814453,
      "learning_rate": 6.444483452530458e-07,
      "loss": 2.1599,
      "step": 225595
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.574007511138916,
      "learning_rate": 6.441803957447645e-07,
      "loss": 1.9986,
      "step": 225596
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.5071256160736084,
      "learning_rate": 6.439125018923963e-07,
      "loss": 1.9842,
      "step": 225597
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.4421027898788452,
      "learning_rate": 6.436446636959747e-07,
      "loss": 2.045,
      "step": 225598
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.593045949935913,
      "learning_rate": 6.433768811555329e-07,
      "loss": 2.0316,
      "step": 225599
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.4040794372558594,
      "learning_rate": 6.431091542711375e-07,
      "loss": 1.9047,
      "step": 225600
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.4533026218414307,
      "learning_rate": 6.428414830428552e-07,
      "loss": 2.0131,
      "step": 225601
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.5441441535949707,
      "learning_rate": 6.425738674707193e-07,
      "loss": 2.137,
      "step": 225602
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.5304286479949951,
      "learning_rate": 6.423063075547629e-07,
      "loss": 1.8979,
      "step": 225603
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.5249130725860596,
      "learning_rate": 6.42038803295053e-07,
      "loss": 1.8599,
      "step": 225604
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.4833652973175049,
      "learning_rate": 6.417713546916559e-07,
      "loss": 2.0183,
      "step": 225605
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.3594443798065186,
      "learning_rate": 6.41503961744605e-07,
      "loss": 1.8159,
      "step": 225606
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.4733139276504517,
      "learning_rate": 6.412366244539335e-07,
      "loss": 2.1087,
      "step": 225607
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.5379992723464966,
      "learning_rate": 6.409693428197082e-07,
      "loss": 2.0626,
      "step": 225608
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.3686586618423462,
      "learning_rate": 6.407021168419624e-07,
      "loss": 2.0243,
      "step": 225609
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.454426646232605,
      "learning_rate": 6.404349465207958e-07,
      "loss": 1.8174,
      "step": 225610
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.4319040775299072,
      "learning_rate": 6.401678318562087e-07,
      "loss": 1.8339,
      "step": 225611
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.4917739629745483,
      "learning_rate": 6.399007728482674e-07,
      "loss": 1.9417,
      "step": 225612
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.3927208185195923,
      "learning_rate": 6.396337694970056e-07,
      "loss": 1.8942,
      "step": 225613
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.4457006454467773,
      "learning_rate": 6.393668218024894e-07,
      "loss": 2.0758,
      "step": 225614
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.536470651626587,
      "learning_rate": 6.390999297647858e-07,
      "loss": 1.8884,
      "step": 225615
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.358751654624939,
      "learning_rate": 6.388330933839281e-07,
      "loss": 1.8169,
      "step": 225616
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.845494031906128,
      "learning_rate": 6.385663126599494e-07,
      "loss": 1.8602,
      "step": 225617
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.4886194467544556,
      "learning_rate": 6.382995875929165e-07,
      "loss": 1.9129,
      "step": 225618
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.5858358144760132,
      "learning_rate": 6.380329181828625e-07,
      "loss": 2.0915,
      "step": 225619
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.4326876401901245,
      "learning_rate": 6.377663044298542e-07,
      "loss": 1.8512,
      "step": 225620
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.5407501459121704,
      "learning_rate": 6.374997463339249e-07,
      "loss": 1.8738,
      "step": 225621
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.3247795104980469,
      "learning_rate": 6.37233243895141e-07,
      "loss": 1.9689,
      "step": 225622
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.4176690578460693,
      "learning_rate": 6.369667971135695e-07,
      "loss": 1.8025,
      "step": 225623
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.5591130256652832,
      "learning_rate": 6.3670040598921e-07,
      "loss": 1.9137,
      "step": 225624
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.3798214197158813,
      "learning_rate": 6.364340705221627e-07,
      "loss": 1.9251,
      "step": 225625
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.5873898267745972,
      "learning_rate": 6.361677907124274e-07,
      "loss": 2.0806,
      "step": 225626
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.4158248901367188,
      "learning_rate": 6.359015665601041e-07,
      "loss": 1.9507,
      "step": 225627
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.6068012714385986,
      "learning_rate": 6.356353980651929e-07,
      "loss": 2.013,
      "step": 225628
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.5267016887664795,
      "learning_rate": 6.353692852277603e-07,
      "loss": 2.0337,
      "step": 225629
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.4687117338180542,
      "learning_rate": 6.351032280479063e-07,
      "loss": 2.1944,
      "step": 225630
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.5298248529434204,
      "learning_rate": 6.348372265255975e-07,
      "loss": 2.0241,
      "step": 225631
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.398252248764038,
      "learning_rate": 6.345712806609338e-07,
      "loss": 1.7777,
      "step": 225632
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.4291719198226929,
      "learning_rate": 6.343053904539819e-07,
      "loss": 1.786,
      "step": 225633
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.5642424821853638,
      "learning_rate": 6.340395559047417e-07,
      "loss": 1.7837,
      "step": 225634
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.457363247871399,
      "learning_rate": 6.3377377701328e-07,
      "loss": 1.9252,
      "step": 225635
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.3870409727096558,
      "learning_rate": 6.335080537796633e-07,
      "loss": 1.944,
      "step": 225636
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.484902024269104,
      "learning_rate": 6.332423862039248e-07,
      "loss": 2.0973,
      "step": 225637
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.5253764390945435,
      "learning_rate": 6.32976774286098e-07,
      "loss": 2.3157,
      "step": 225638
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.3886663913726807,
      "learning_rate": 6.327112180262827e-07,
      "loss": 1.8759,
      "step": 225639
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.4922435283660889,
      "learning_rate": 6.32445717424479e-07,
      "loss": 1.9738,
      "step": 225640
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.6285433769226074,
      "learning_rate": 6.321802724807534e-07,
      "loss": 1.882,
      "step": 225641
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.3647944927215576,
      "learning_rate": 6.319148831951393e-07,
      "loss": 2.0451,
      "step": 225642
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.4695043563842773,
      "learning_rate": 6.316495495677365e-07,
      "loss": 1.9656,
      "step": 225643
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.5772651433944702,
      "learning_rate": 6.31384271598545e-07,
      "loss": 1.8488,
      "step": 225644
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.5366557836532593,
      "learning_rate": 6.311190492876317e-07,
      "loss": 1.9679,
      "step": 225645
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.5594617128372192,
      "learning_rate": 6.308538826350629e-07,
      "loss": 2.0161,
      "step": 225646
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.576576828956604,
      "learning_rate": 6.305887716408386e-07,
      "loss": 1.888,
      "step": 225647
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.4923919439315796,
      "learning_rate": 6.30323716305059e-07,
      "loss": 2.0338,
      "step": 225648
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.4755171537399292,
      "learning_rate": 6.30058716627757e-07,
      "loss": 2.0078,
      "step": 225649
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.3938840627670288,
      "learning_rate": 6.297937726089663e-07,
      "loss": 1.9263,
      "step": 225650
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.363286018371582,
      "learning_rate": 6.295288842487534e-07,
      "loss": 1.8522,
      "step": 225651
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.3650357723236084,
      "learning_rate": 6.292640515471514e-07,
      "loss": 2.015,
      "step": 225652
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.5272157192230225,
      "learning_rate": 6.289992745042271e-07,
      "loss": 1.8994,
      "step": 225653
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.3367578983306885,
      "learning_rate": 6.287345531200139e-07,
      "loss": 1.8019,
      "step": 225654
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.6427351236343384,
      "learning_rate": 6.284698873946115e-07,
      "loss": 1.9014,
      "step": 225655
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.467952847480774,
      "learning_rate": 6.282052773279867e-07,
      "loss": 1.8658,
      "step": 225656
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.429444670677185,
      "learning_rate": 6.279407229202393e-07,
      "loss": 1.9299,
      "step": 225657
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.4381093978881836,
      "learning_rate": 6.276762241714028e-07,
      "loss": 1.8775,
      "step": 225658
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.4926735162734985,
      "learning_rate": 6.274117810815437e-07,
      "loss": 2.0418,
      "step": 225659
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.4741548299789429,
      "learning_rate": 6.271473936506954e-07,
      "loss": 1.96,
      "step": 225660
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.5290977954864502,
      "learning_rate": 6.268830618789244e-07,
      "loss": 1.9621,
      "step": 225661
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.6666752099990845,
      "learning_rate": 6.266187857662308e-07,
      "loss": 1.8809,
      "step": 225662
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.480732798576355,
      "learning_rate": 6.263545653127145e-07,
      "loss": 1.8708,
      "step": 225663
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.4900349378585815,
      "learning_rate": 6.260904005184419e-07,
      "loss": 1.7942,
      "step": 225664
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.485001802444458,
      "learning_rate": 6.2582629138338e-07,
      "loss": 1.8943,
      "step": 225665
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.4579275846481323,
      "learning_rate": 6.255622379076619e-07,
      "loss": 1.8625,
      "step": 225666
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.5401947498321533,
      "learning_rate": 6.252982400912876e-07,
      "loss": 2.0634,
      "step": 225667
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.4352703094482422,
      "learning_rate": 6.25034297934357e-07,
      "loss": 1.9631,
      "step": 225668
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.3880226612091064,
      "learning_rate": 6.247704114368368e-07,
      "loss": 2.1952,
      "step": 225669
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.5237846374511719,
      "learning_rate": 6.245065805988603e-07,
      "loss": 1.8933,
      "step": 225670
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.3524283170700073,
      "learning_rate": 6.242428054204274e-07,
      "loss": 1.9634,
      "step": 225671
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.4815739393234253,
      "learning_rate": 6.239790859015714e-07,
      "loss": 1.943,
      "step": 225672
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.3838928937911987,
      "learning_rate": 6.237154220423923e-07,
      "loss": 2.0675,
      "step": 225673
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.5267105102539062,
      "learning_rate": 6.234518138429234e-07,
      "loss": 1.8367,
      "step": 225674
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.4043211936950684,
      "learning_rate": 6.23188261303198e-07,
      "loss": 1.9238,
      "step": 225675
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.442051887512207,
      "learning_rate": 6.229247644232494e-07,
      "loss": 1.9279,
      "step": 225676
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.6672782897949219,
      "learning_rate": 6.226613232031774e-07,
      "loss": 1.844,
      "step": 225677
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.5798128843307495,
      "learning_rate": 6.223979376430155e-07,
      "loss": 2.2188,
      "step": 225678
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.5125458240509033,
      "learning_rate": 6.221346077427636e-07,
      "loss": 1.9906,
      "step": 225679
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.3858503103256226,
      "learning_rate": 6.218713335025216e-07,
      "loss": 2.029,
      "step": 225680
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.4609965085983276,
      "learning_rate": 6.216081149223562e-07,
      "loss": 1.9849,
      "step": 225681
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.6060737371444702,
      "learning_rate": 6.21344952002234e-07,
      "loss": 1.8326,
      "step": 225682
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.574462890625,
      "learning_rate": 6.210818447422883e-07,
      "loss": 1.9161,
      "step": 225683
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.623958945274353,
      "learning_rate": 6.20818793142519e-07,
      "loss": 2.3341,
      "step": 225684
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.4613581895828247,
      "learning_rate": 6.205557972030262e-07,
      "loss": 1.936,
      "step": 225685
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.5488396883010864,
      "learning_rate": 6.202928569237764e-07,
      "loss": 2.0311,
      "step": 225686
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.401089072227478,
      "learning_rate": 6.200299723049029e-07,
      "loss": 1.9861,
      "step": 225687
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.5752408504486084,
      "learning_rate": 6.197671433464058e-07,
      "loss": 1.9195,
      "step": 225688
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.5653870105743408,
      "learning_rate": 6.195043700483182e-07,
      "loss": 2.0853,
      "step": 225689
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.6685901880264282,
      "learning_rate": 6.192416524107402e-07,
      "loss": 2.0834,
      "step": 225690
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.6300476789474487,
      "learning_rate": 6.18978990433705e-07,
      "loss": 2.0092,
      "step": 225691
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.4758670330047607,
      "learning_rate": 6.18716384117246e-07,
      "loss": 1.7925,
      "step": 225692
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.370927095413208,
      "learning_rate": 6.184538334613964e-07,
      "loss": 2.208,
      "step": 225693
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.4802783727645874,
      "learning_rate": 6.181913384662563e-07,
      "loss": 2.0527,
      "step": 225694
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.6192964315414429,
      "learning_rate": 6.179288991318254e-07,
      "loss": 2.0635,
      "step": 225695
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.4992505311965942,
      "learning_rate": 6.176665154581706e-07,
      "loss": 2.0805,
      "step": 225696
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.600867509841919,
      "learning_rate": 6.174041874453583e-07,
      "loss": 2.1412,
      "step": 225697
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.5531643629074097,
      "learning_rate": 6.17141915093422e-07,
      "loss": 2.0578,
      "step": 225698
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.485356330871582,
      "learning_rate": 6.168796984023949e-07,
      "loss": 2.0924,
      "step": 225699
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.5851049423217773,
      "learning_rate": 6.16617537372377e-07,
      "loss": 1.86,
      "step": 225700
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.3713786602020264,
      "learning_rate": 6.163554320033682e-07,
      "loss": 1.8567,
      "step": 225701
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.485935926437378,
      "learning_rate": 6.160933822954017e-07,
      "loss": 1.9496,
      "step": 225702
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.521485447883606,
      "learning_rate": 6.158313882485777e-07,
      "loss": 1.9902,
      "step": 225703
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.545118808746338,
      "learning_rate": 6.15569449862896e-07,
      "loss": 1.9599,
      "step": 225704
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.4055355787277222,
      "learning_rate": 6.153075671384566e-07,
      "loss": 2.0221,
      "step": 225705
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.4507964849472046,
      "learning_rate": 6.150457400752928e-07,
      "loss": 1.9372,
      "step": 225706
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.4032723903656006,
      "learning_rate": 6.147839686734379e-07,
      "loss": 2.067,
      "step": 225707
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.449367642402649,
      "learning_rate": 6.145222529329252e-07,
      "loss": 2.1167,
      "step": 225708
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.5211979150772095,
      "learning_rate": 6.142605928538214e-07,
      "loss": 1.8116,
      "step": 225709
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.6577847003936768,
      "learning_rate": 6.139989884361929e-07,
      "loss": 2.0328,
      "step": 225710
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.5199397802352905,
      "learning_rate": 6.137374396800731e-07,
      "loss": 2.1276,
      "step": 225711
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.5234451293945312,
      "learning_rate": 6.134759465855288e-07,
      "loss": 2.0523,
      "step": 225712
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.4991685152053833,
      "learning_rate": 6.132145091525598e-07,
      "loss": 1.909,
      "step": 225713
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.5116318464279175,
      "learning_rate": 6.129531273812661e-07,
      "loss": 2.0348,
      "step": 225714
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.425697684288025,
      "learning_rate": 6.126918012716475e-07,
      "loss": 1.9816,
      "step": 225715
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.359739065170288,
      "learning_rate": 6.124305308238042e-07,
      "loss": 1.7222,
      "step": 225716
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.5528767108917236,
      "learning_rate": 6.121693160377694e-07,
      "loss": 1.8089,
      "step": 225717
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.370085597038269,
      "learning_rate": 6.119081569135765e-07,
      "loss": 1.8438,
      "step": 225718
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.541386604309082,
      "learning_rate": 6.116470534512585e-07,
      "loss": 1.8985,
      "step": 225719
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.3855416774749756,
      "learning_rate": 6.113860056509156e-07,
      "loss": 1.8995,
      "step": 225720
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.5624445676803589,
      "learning_rate": 6.111250135125479e-07,
      "loss": 2.011,
      "step": 225721
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.5222803354263306,
      "learning_rate": 6.108640770362216e-07,
      "loss": 1.8504,
      "step": 225722
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.4528942108154297,
      "learning_rate": 6.106031962220037e-07,
      "loss": 1.8405,
      "step": 225723
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.6059526205062866,
      "learning_rate": 6.103423710699273e-07,
      "loss": 1.9733,
      "step": 225724
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.6168617010116577,
      "learning_rate": 6.100816015800259e-07,
      "loss": 2.0037,
      "step": 225725
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.5184518098831177,
      "learning_rate": 6.098208877523658e-07,
      "loss": 2.1079,
      "step": 225726
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.48967707157135,
      "learning_rate": 6.095602295869806e-07,
      "loss": 1.7483,
      "step": 225727
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.5367997884750366,
      "learning_rate": 6.0929962708397e-07,
      "loss": 2.0617,
      "step": 225728
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.5259547233581543,
      "learning_rate": 6.090390802433009e-07,
      "loss": 2.0458,
      "step": 225729
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.341991901397705,
      "learning_rate": 6.087785890650731e-07,
      "loss": 1.9411,
      "step": 225730
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.4676032066345215,
      "learning_rate": 6.0851815354932e-07,
      "loss": 2.0922,
      "step": 225731
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.414255142211914,
      "learning_rate": 6.082577736961081e-07,
      "loss": 1.9048,
      "step": 225732
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.714808702468872,
      "learning_rate": 6.079974495054707e-07,
      "loss": 2.0038,
      "step": 225733
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.5708650350570679,
      "learning_rate": 6.077371809774744e-07,
      "loss": 1.9179,
      "step": 225734
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.487013816833496,
      "learning_rate": 6.074769681121194e-07,
      "loss": 2.0524,
      "step": 225735
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.400551676750183,
      "learning_rate": 6.072168109095055e-07,
      "loss": 1.7144,
      "step": 225736
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.4081114530563354,
      "learning_rate": 6.069567093696326e-07,
      "loss": 1.8733,
      "step": 225737
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.3922137022018433,
      "learning_rate": 6.066966634926341e-07,
      "loss": 1.9195,
      "step": 225738
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.4947654008865356,
      "learning_rate": 6.064366732784431e-07,
      "loss": 2.0145,
      "step": 225739
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.4510345458984375,
      "learning_rate": 6.061767387271932e-07,
      "loss": 1.8669,
      "step": 225740
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.394532322883606,
      "learning_rate": 6.059168598389175e-07,
      "loss": 1.9062,
      "step": 225741
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.4057683944702148,
      "learning_rate": 6.056570366136493e-07,
      "loss": 1.9891,
      "step": 225742
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.525139331817627,
      "learning_rate": 6.053972690514219e-07,
      "loss": 2.0619,
      "step": 225743
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.346584677696228,
      "learning_rate": 6.05137557152302e-07,
      "loss": 1.9527,
      "step": 225744
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.4414708614349365,
      "learning_rate": 6.048779009163562e-07,
      "loss": 1.948,
      "step": 225745
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.6297893524169922,
      "learning_rate": 6.046183003436178e-07,
      "loss": 1.7974,
      "step": 225746
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.5422061681747437,
      "learning_rate": 6.0435875543412e-07,
      "loss": 1.9026,
      "step": 225747
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.3602949380874634,
      "learning_rate": 6.040992661879296e-07,
      "loss": 2.1527,
      "step": 225748
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.430667519569397,
      "learning_rate": 6.038398326050797e-07,
      "loss": 2.091,
      "step": 225749
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.3790123462677002,
      "learning_rate": 6.035804546856371e-07,
      "loss": 1.7836,
      "step": 225750
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.4218883514404297,
      "learning_rate": 6.03321132429635e-07,
      "loss": 2.0693,
      "step": 225751
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.4175671339035034,
      "learning_rate": 6.0306186583714e-07,
      "loss": 2.1218,
      "step": 225752
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.4378180503845215,
      "learning_rate": 6.028026549081522e-07,
      "loss": 1.9426,
      "step": 225753
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.4048818349838257,
      "learning_rate": 6.025434996428047e-07,
      "loss": 2.0242,
      "step": 225754
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.5427101850509644,
      "learning_rate": 6.022844000410643e-07,
      "loss": 1.967,
      "step": 225755
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.5203590393066406,
      "learning_rate": 6.020253561029975e-07,
      "loss": 1.8987,
      "step": 225756
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.4018232822418213,
      "learning_rate": 6.017663678287044e-07,
      "loss": 1.8529,
      "step": 225757
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.5699671506881714,
      "learning_rate": 6.015074352181515e-07,
      "loss": 1.8208,
      "step": 225758
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.5521302223205566,
      "learning_rate": 6.012485582714721e-07,
      "loss": 1.932,
      "step": 225759
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.4229605197906494,
      "learning_rate": 6.009897369886663e-07,
      "loss": 1.9111,
      "step": 225760
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.4747960567474365,
      "learning_rate": 6.007309713697672e-07,
      "loss": 2.1423,
      "step": 225761
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.431778907775879,
      "learning_rate": 6.004722614148417e-07,
      "loss": 1.9869,
      "step": 225762
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.5502773523330688,
      "learning_rate": 6.002136071239561e-07,
      "loss": 1.7868,
      "step": 225763
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.4262651205062866,
      "learning_rate": 5.999550084971439e-07,
      "loss": 1.8275,
      "step": 225764
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.590140700340271,
      "learning_rate": 5.996964655344382e-07,
      "loss": 2.1483,
      "step": 225765
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.3244496583938599,
      "learning_rate": 5.99437978235906e-07,
      "loss": 1.9804,
      "step": 225766
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.3888171911239624,
      "learning_rate": 5.991795466015803e-07,
      "loss": 1.8686,
      "step": 225767
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.4608094692230225,
      "learning_rate": 5.989211706315278e-07,
      "loss": 2.0843,
      "step": 225768
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.4580398797988892,
      "learning_rate": 5.98662850325815e-07,
      "loss": 2.2778,
      "step": 225769
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.4700521230697632,
      "learning_rate": 5.984045856844088e-07,
      "loss": 1.8911,
      "step": 225770
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.552761435508728,
      "learning_rate": 5.981463767074424e-07,
      "loss": 2.1262,
      "step": 225771
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.441925048828125,
      "learning_rate": 5.978882233949489e-07,
      "loss": 1.7819,
      "step": 225772
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.5828704833984375,
      "learning_rate": 5.976301257469285e-07,
      "loss": 2.0349,
      "step": 225773
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.461527943611145,
      "learning_rate": 5.97372083763481e-07,
      "loss": 1.8119,
      "step": 225774
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.393999695777893,
      "learning_rate": 5.971140974446065e-07,
      "loss": 1.7504,
      "step": 225775
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.3807451725006104,
      "learning_rate": 5.968561667904049e-07,
      "loss": 1.6819,
      "step": 225776
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.4285743236541748,
      "learning_rate": 5.965982918009093e-07,
      "loss": 1.7486,
      "step": 225777
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.3078343868255615,
      "learning_rate": 5.9634047247612e-07,
      "loss": 1.8762,
      "step": 225778
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.4940450191497803,
      "learning_rate": 5.960827088161368e-07,
      "loss": 1.9487,
      "step": 225779
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.5411367416381836,
      "learning_rate": 5.958250008210263e-07,
      "loss": 2.0734,
      "step": 225780
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.4877880811691284,
      "learning_rate": 5.955673484907553e-07,
      "loss": 1.8284,
      "step": 225781
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.4995092153549194,
      "learning_rate": 5.953097518254568e-07,
      "loss": 2.2534,
      "step": 225782
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.5952917337417603,
      "learning_rate": 5.95052210825131e-07,
      "loss": 1.9837,
      "step": 225783
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.4657377004623413,
      "learning_rate": 5.947947254898111e-07,
      "loss": 1.8586,
      "step": 225784
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.7058779001235962,
      "learning_rate": 5.945372958195971e-07,
      "loss": 2.074,
      "step": 225785
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.467969298362732,
      "learning_rate": 5.942799218145222e-07,
      "loss": 1.983,
      "step": 225786
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.5286580324172974,
      "learning_rate": 5.940226034745865e-07,
      "loss": 2.0665,
      "step": 225787
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.487816572189331,
      "learning_rate": 5.937653407998899e-07,
      "loss": 1.9904,
      "step": 225788
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.4991613626480103,
      "learning_rate": 5.935081337904657e-07,
      "loss": 1.7695,
      "step": 225789
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.5833884477615356,
      "learning_rate": 5.932509824463805e-07,
      "loss": 1.7985,
      "step": 225790
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.652672529220581,
      "learning_rate": 5.929938867676342e-07,
      "loss": 1.8538,
      "step": 225791
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.404698133468628,
      "learning_rate": 5.927368467542936e-07,
      "loss": 2.0063,
      "step": 225792
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.550206184387207,
      "learning_rate": 5.924798624064253e-07,
      "loss": 2.1447,
      "step": 225793
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.5334709882736206,
      "learning_rate": 5.922229337240958e-07,
      "loss": 1.9948,
      "step": 225794
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.5011171102523804,
      "learning_rate": 5.919660607073051e-07,
      "loss": 2.0552,
      "step": 225795
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.4102507829666138,
      "learning_rate": 5.917092433561199e-07,
      "loss": 2.0236,
      "step": 225796
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.5543403625488281,
      "learning_rate": 5.914524816705735e-07,
      "loss": 1.8447,
      "step": 225797
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.4970452785491943,
      "learning_rate": 5.911957756507324e-07,
      "loss": 1.9314,
      "step": 225798
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.5022953748703003,
      "learning_rate": 5.909391252966301e-07,
      "loss": 1.8725,
      "step": 225799
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.4896183013916016,
      "learning_rate": 5.906825306083662e-07,
      "loss": 1.9499,
      "step": 225800
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.5545986890792847,
      "learning_rate": 5.904259915859078e-07,
      "loss": 2.0212,
      "step": 225801
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.3662019968032837,
      "learning_rate": 5.901695082293545e-07,
      "loss": 2.1124,
      "step": 225802
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.5511212348937988,
      "learning_rate": 5.899130805387397e-07,
      "loss": 2.1147,
      "step": 225803
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.420268177986145,
      "learning_rate": 5.896567085140968e-07,
      "loss": 2.1679,
      "step": 225804
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.5347936153411865,
      "learning_rate": 5.894003921555257e-07,
      "loss": 1.9485,
      "step": 225805
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.438412070274353,
      "learning_rate": 5.89144131462993e-07,
      "loss": 2.2476,
      "step": 225806
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.6176689863204956,
      "learning_rate": 5.888879264365987e-07,
      "loss": 1.7833,
      "step": 225807
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.4202717542648315,
      "learning_rate": 5.886317770764093e-07,
      "loss": 1.9351,
      "step": 225808
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.4509477615356445,
      "learning_rate": 5.883756833824249e-07,
      "loss": 2.0711,
      "step": 225809
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.5011250972747803,
      "learning_rate": 5.881196453547121e-07,
      "loss": 1.7454,
      "step": 225810
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.501031517982483,
      "learning_rate": 5.878636629933376e-07,
      "loss": 2.0936,
      "step": 225811
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.4557064771652222,
      "learning_rate": 5.876077362983012e-07,
      "loss": 1.7853,
      "step": 225812
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.5471876859664917,
      "learning_rate": 5.873518652697029e-07,
      "loss": 1.9248,
      "step": 225813
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.5359047651290894,
      "learning_rate": 5.870960499075761e-07,
      "loss": 2.2043,
      "step": 225814
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.422687292098999,
      "learning_rate": 5.86840290211954e-07,
      "loss": 1.7601,
      "step": 225815
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.5359152555465698,
      "learning_rate": 5.8658458618287e-07,
      "loss": 1.9548,
      "step": 225816
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.6102412939071655,
      "learning_rate": 5.863289378204239e-07,
      "loss": 2.0465,
      "step": 225817
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.465240240097046,
      "learning_rate": 5.860733451245825e-07,
      "loss": 2.1595,
      "step": 225818
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.4564480781555176,
      "learning_rate": 5.858178080954789e-07,
      "loss": 2.0323,
      "step": 225819
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.5165153741836548,
      "learning_rate": 5.855623267331133e-07,
      "loss": 2.0847,
      "step": 225820
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.645002007484436,
      "learning_rate": 5.853069010375522e-07,
      "loss": 1.9399,
      "step": 225821
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.3713594675064087,
      "learning_rate": 5.850515310088288e-07,
      "loss": 1.9928,
      "step": 225822
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.534090280532837,
      "learning_rate": 5.8479621664701e-07,
      "loss": 1.9029,
      "step": 225823
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.3962182998657227,
      "learning_rate": 5.845409579520954e-07,
      "loss": 2.0512,
      "step": 225824
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.5219717025756836,
      "learning_rate": 5.842857549241853e-07,
      "loss": 1.9311,
      "step": 225825
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.3922563791275024,
      "learning_rate": 5.840306075633128e-07,
      "loss": 1.8968,
      "step": 225826
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.4730113744735718,
      "learning_rate": 5.837755158695445e-07,
      "loss": 1.9142,
      "step": 225827
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.5065666437149048,
      "learning_rate": 5.835204798428472e-07,
      "loss": 2.2202,
      "step": 225828
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.4333058595657349,
      "learning_rate": 5.832654994833874e-07,
      "loss": 1.9337,
      "step": 225829
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.4897257089614868,
      "learning_rate": 5.830105747910985e-07,
      "loss": 1.8902,
      "step": 225830
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.542404055595398,
      "learning_rate": 5.827557057661137e-07,
      "loss": 1.9507,
      "step": 225831
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.3735756874084473,
      "learning_rate": 5.825008924084329e-07,
      "loss": 1.7557,
      "step": 225832
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.4154632091522217,
      "learning_rate": 5.822461347181229e-07,
      "loss": 1.8197,
      "step": 225833
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.2757269144058228,
      "learning_rate": 5.819914326952168e-07,
      "loss": 1.7324,
      "step": 225834
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.5220922231674194,
      "learning_rate": 5.817367863397482e-07,
      "loss": 1.9742,
      "step": 225835
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.810813546180725,
      "learning_rate": 5.814821956518168e-07,
      "loss": 2.1211,
      "step": 225836
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.5142024755477905,
      "learning_rate": 5.812276606314559e-07,
      "loss": 2.1891,
      "step": 225837
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.5656473636627197,
      "learning_rate": 5.809731812786655e-07,
      "loss": 1.8647,
      "step": 225838
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.4569367170333862,
      "learning_rate": 5.807187575935457e-07,
      "loss": 1.6938,
      "step": 225839
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.5702025890350342,
      "learning_rate": 5.804643895760963e-07,
      "loss": 1.9532,
      "step": 225840
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.560561180114746,
      "learning_rate": 5.802100772264173e-07,
      "loss": 1.8549,
      "step": 225841
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.4596915245056152,
      "learning_rate": 5.799558205445088e-07,
      "loss": 1.9861,
      "step": 225842
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.5783741474151611,
      "learning_rate": 5.797016195304371e-07,
      "loss": 2.1823,
      "step": 225843
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.5217598676681519,
      "learning_rate": 5.794474741842692e-07,
      "loss": 1.8779,
      "step": 225844
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.6607156991958618,
      "learning_rate": 5.791933845060382e-07,
      "loss": 1.995,
      "step": 225845
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.4290440082550049,
      "learning_rate": 5.789393504957774e-07,
      "loss": 2.1468,
      "step": 225846
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.490355134010315,
      "learning_rate": 5.786853721535201e-07,
      "loss": 1.8965,
      "step": 225847
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.4954441785812378,
      "learning_rate": 5.784314494793662e-07,
      "loss": 2.0542,
      "step": 225848
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.3533905744552612,
      "learning_rate": 5.781775824733159e-07,
      "loss": 1.8093,
      "step": 225849
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.3802525997161865,
      "learning_rate": 5.77923771135469e-07,
      "loss": 1.7612,
      "step": 225850
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.4997483491897583,
      "learning_rate": 5.776700154657921e-07,
      "loss": 2.12,
      "step": 225851
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.3962806463241577,
      "learning_rate": 5.774163154644185e-07,
      "loss": 2.0679,
      "step": 225852
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.5876319408416748,
      "learning_rate": 5.771626711313148e-07,
      "loss": 1.8293,
      "step": 225853
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.4886940717697144,
      "learning_rate": 5.769090824665812e-07,
      "loss": 2.1269,
      "step": 225854
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.508400321006775,
      "learning_rate": 5.76655549470284e-07,
      "loss": 1.9917,
      "step": 225855
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.6603459119796753,
      "learning_rate": 5.764020721424234e-07,
      "loss": 2.0355,
      "step": 225856
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.431499719619751,
      "learning_rate": 5.761486504830326e-07,
      "loss": 2.1138,
      "step": 225857
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.556277871131897,
      "learning_rate": 5.758952844922116e-07,
      "loss": 1.963,
      "step": 225858
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.351920485496521,
      "learning_rate": 5.756419741699935e-07,
      "loss": 1.8519,
      "step": 225859
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.4201213121414185,
      "learning_rate": 5.753887195163787e-07,
      "loss": 2.066,
      "step": 225860
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.4783344268798828,
      "learning_rate": 5.751355205315e-07,
      "loss": 2.0593,
      "step": 225861
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.4680222272872925,
      "learning_rate": 5.748823772153244e-07,
      "loss": 2.1152,
      "step": 225862
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.439705491065979,
      "learning_rate": 5.746292895679516e-07,
      "loss": 1.9799,
      "step": 225863
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.4854021072387695,
      "learning_rate": 5.743762575893817e-07,
      "loss": 1.9966,
      "step": 225864
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.3876761198043823,
      "learning_rate": 5.741232812797147e-07,
      "loss": 2.1485,
      "step": 225865
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.3885334730148315,
      "learning_rate": 5.738703606389172e-07,
      "loss": 1.9114,
      "step": 225866
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.3671704530715942,
      "learning_rate": 5.736174956671558e-07,
      "loss": 1.952,
      "step": 225867
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.4782785177230835,
      "learning_rate": 5.733646863643637e-07,
      "loss": 1.9181,
      "step": 225868
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.3956784009933472,
      "learning_rate": 5.731119327306744e-07,
      "loss": 2.0933,
      "step": 225869
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.4523680210113525,
      "learning_rate": 5.728592347660544e-07,
      "loss": 2.0102,
      "step": 225870
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.4088318347930908,
      "learning_rate": 5.726065924706369e-07,
      "loss": 2.0069,
      "step": 225871
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.386113166809082,
      "learning_rate": 5.723540058443888e-07,
      "loss": 1.8479,
      "step": 225872
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.460423231124878,
      "learning_rate": 5.721014748874098e-07,
      "loss": 1.9374,
      "step": 225873
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.4901983737945557,
      "learning_rate": 5.718489995997e-07,
      "loss": 1.7689,
      "step": 225874
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.4125248193740845,
      "learning_rate": 5.715965799813593e-07,
      "loss": 2.0169,
      "step": 225875
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.497854471206665,
      "learning_rate": 5.713442160324211e-07,
      "loss": 1.9679,
      "step": 225876
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.519515872001648,
      "learning_rate": 5.710919077529186e-07,
      "loss": 2.1023,
      "step": 225877
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.4802823066711426,
      "learning_rate": 5.708396551428851e-07,
      "loss": 2.0019,
      "step": 225878
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.558160662651062,
      "learning_rate": 5.705874582023873e-07,
      "loss": 1.8734,
      "step": 225879
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.5094152688980103,
      "learning_rate": 5.703353169314584e-07,
      "loss": 2.375,
      "step": 225880
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.4764246940612793,
      "learning_rate": 5.700832313301984e-07,
      "loss": 1.9783,
      "step": 225881
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.6011241674423218,
      "learning_rate": 5.698312013985739e-07,
      "loss": 1.9577,
      "step": 225882
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.4713575839996338,
      "learning_rate": 5.69579227136685e-07,
      "loss": 1.8236,
      "step": 225883
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.5009020566940308,
      "learning_rate": 5.693273085445648e-07,
      "loss": 1.9744,
      "step": 225884
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.4320580959320068,
      "learning_rate": 5.690754456222469e-07,
      "loss": 1.9134,
      "step": 225885
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.584147572517395,
      "learning_rate": 5.688236383697975e-07,
      "loss": 2.0072,
      "step": 225886
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.344222903251648,
      "learning_rate": 5.685718867872835e-07,
      "loss": 2.0406,
      "step": 225887
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.4575086832046509,
      "learning_rate": 5.683201908747048e-07,
      "loss": 2.1566,
      "step": 225888
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.431320309638977,
      "learning_rate": 5.680685506320948e-07,
      "loss": 1.9274,
      "step": 225889
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.3816479444503784,
      "learning_rate": 5.678169660595533e-07,
      "loss": 1.802,
      "step": 225890
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.4532986879348755,
      "learning_rate": 5.67565437157147e-07,
      "loss": 2.1737,
      "step": 225891
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.4221150875091553,
      "learning_rate": 5.673139639248426e-07,
      "loss": 1.8492,
      "step": 225892
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.3588428497314453,
      "learning_rate": 5.670625463627065e-07,
      "loss": 1.8071,
      "step": 225893
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.526353120803833,
      "learning_rate": 5.66811184470839e-07,
      "loss": 2.0113,
      "step": 225894
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.4620674848556519,
      "learning_rate": 5.665598782492398e-07,
      "loss": 2.055,
      "step": 225895
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.3499248027801514,
      "learning_rate": 5.663086276979756e-07,
      "loss": 2.0493,
      "step": 225896
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.432071328163147,
      "learning_rate": 5.66057432817113e-07,
      "loss": 2.1513,
      "step": 225897
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.4252804517745972,
      "learning_rate": 5.658062936066187e-07,
      "loss": 1.9301,
      "step": 225898
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.501405954360962,
      "learning_rate": 5.655552100666261e-07,
      "loss": 1.9333,
      "step": 225899
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.571702003479004,
      "learning_rate": 5.653041821971682e-07,
      "loss": 2.1063,
      "step": 225900
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.4543668031692505,
      "learning_rate": 5.650532099982453e-07,
      "loss": 2.0048,
      "step": 225901
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.4532679319381714,
      "learning_rate": 5.648022934699236e-07,
      "loss": 1.8341,
      "step": 225902
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.4933955669403076,
      "learning_rate": 5.645514326122702e-07,
      "loss": 2.0048,
      "step": 225903
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.3699921369552612,
      "learning_rate": 5.643006274253181e-07,
      "loss": 2.082,
      "step": 225904
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.411909580230713,
      "learning_rate": 5.64049877909134e-07,
      "loss": 1.8889,
      "step": 225905
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.498724341392517,
      "learning_rate": 5.63799184063718e-07,
      "loss": 1.9198,
      "step": 225906
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.47469162940979,
      "learning_rate": 5.635485458891364e-07,
      "loss": 2.0904,
      "step": 225907
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.4627599716186523,
      "learning_rate": 5.632979633854562e-07,
      "loss": 1.9542,
      "step": 225908
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.611040711402893,
      "learning_rate": 5.630474365527105e-07,
      "loss": 1.9723,
      "step": 225909
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.5335592031478882,
      "learning_rate": 5.627969653909659e-07,
      "loss": 2.1001,
      "step": 225910
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.4105883836746216,
      "learning_rate": 5.625465499002224e-07,
      "loss": 1.7629,
      "step": 225911
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.4836616516113281,
      "learning_rate": 5.6229619008058e-07,
      "loss": 2.0143,
      "step": 225912
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.6469836235046387,
      "learning_rate": 5.620458859320387e-07,
      "loss": 1.8648,
      "step": 225913
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.4934989213943481,
      "learning_rate": 5.617956374546651e-07,
      "loss": 1.9706,
      "step": 225914
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.4639028310775757,
      "learning_rate": 5.615454446484924e-07,
      "loss": 2.028,
      "step": 225915
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.5092524290084839,
      "learning_rate": 5.612953075136206e-07,
      "loss": 1.9878,
      "step": 225916
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.3455910682678223,
      "learning_rate": 5.610452260500498e-07,
      "loss": 1.9521,
      "step": 225917
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.4062339067459106,
      "learning_rate": 5.607952002578131e-07,
      "loss": 2.0203,
      "step": 225918
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.5372960567474365,
      "learning_rate": 5.605452301369772e-07,
      "loss": 1.9977,
      "step": 225919
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.4140937328338623,
      "learning_rate": 5.602953156875756e-07,
      "loss": 2.2938,
      "step": 225920
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.5100451707839966,
      "learning_rate": 5.60045456909708e-07,
      "loss": 1.9606,
      "step": 225921
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.37886643409729,
      "learning_rate": 5.59795653803341e-07,
      "loss": 1.9732,
      "step": 225922
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.3872883319854736,
      "learning_rate": 5.595459063685748e-07,
      "loss": 1.8533,
      "step": 225923
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.5744712352752686,
      "learning_rate": 5.592962146054425e-07,
      "loss": 1.9911,
      "step": 225924
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.4207721948623657,
      "learning_rate": 5.590465785140108e-07,
      "loss": 1.9531,
      "step": 225925
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.3687726259231567,
      "learning_rate": 5.587969980942797e-07,
      "loss": 2.0268,
      "step": 225926
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.484352946281433,
      "learning_rate": 5.585474733463157e-07,
      "loss": 1.9368,
      "step": 225927
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.471675992012024,
      "learning_rate": 5.582980042701857e-07,
      "loss": 2.0587,
      "step": 225928
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.5145230293273926,
      "learning_rate": 5.580485908659227e-07,
      "loss": 2.0121,
      "step": 225929
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.441550850868225,
      "learning_rate": 5.577992331335601e-07,
      "loss": 1.7918,
      "step": 225930
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.59205961227417,
      "learning_rate": 5.575499310731646e-07,
      "loss": 1.8418,
      "step": 225931
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.4596279859542847,
      "learning_rate": 5.573006846847694e-07,
      "loss": 1.9203,
      "step": 225932
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.439724326133728,
      "learning_rate": 5.570514939684078e-07,
      "loss": 1.9761,
      "step": 225933
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.5295391082763672,
      "learning_rate": 5.568023589241466e-07,
      "loss": 1.9376,
      "step": 225934
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.4639816284179688,
      "learning_rate": 5.565532795520522e-07,
      "loss": 1.8243,
      "step": 225935
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.6057329177856445,
      "learning_rate": 5.563042558521247e-07,
      "loss": 2.0308,
      "step": 225936
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.4940887689590454,
      "learning_rate": 5.560552878244307e-07,
      "loss": 1.9734,
      "step": 225937
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.6470948457717896,
      "learning_rate": 5.558063754690367e-07,
      "loss": 2.1799,
      "step": 225938
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.5333285331726074,
      "learning_rate": 5.555575187859762e-07,
      "loss": 2.0801,
      "step": 225939
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.5510634183883667,
      "learning_rate": 5.553087177752824e-07,
      "loss": 2.0232,
      "step": 225940
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.3939543962478638,
      "learning_rate": 5.550599724369886e-07,
      "loss": 2.1698,
      "step": 225941
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.4396940469741821,
      "learning_rate": 5.548112827711948e-07,
      "loss": 2.0843,
      "step": 225942
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.5324848890304565,
      "learning_rate": 5.545626487778676e-07,
      "loss": 2.0778,
      "step": 225943
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.3975741863250732,
      "learning_rate": 5.543140704571402e-07,
      "loss": 2.0577,
      "step": 225944
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.485654592514038,
      "learning_rate": 5.540655478090128e-07,
      "loss": 2.2503,
      "step": 225945
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.4220384359359741,
      "learning_rate": 5.538170808335186e-07,
      "loss": 1.8824,
      "step": 225946
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.6137663125991821,
      "learning_rate": 5.53568669530724e-07,
      "loss": 2.097,
      "step": 225947
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.5647823810577393,
      "learning_rate": 5.533203139006958e-07,
      "loss": 1.9676,
      "step": 225948
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.4241563081741333,
      "learning_rate": 5.530720139434341e-07,
      "loss": 2.0149,
      "step": 225949
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.5552712678909302,
      "learning_rate": 5.528237696590387e-07,
      "loss": 2.1144,
      "step": 225950
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.3632075786590576,
      "learning_rate": 5.525755810474764e-07,
      "loss": 2.1198,
      "step": 225951
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.5455876588821411,
      "learning_rate": 5.523274481088802e-07,
      "loss": 1.8805,
      "step": 225952
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.4525104761123657,
      "learning_rate": 5.520793708432503e-07,
      "loss": 1.9285,
      "step": 225953
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.3582843542099,
      "learning_rate": 5.518313492506532e-07,
      "loss": 1.8451,
      "step": 225954
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.49418306350708,
      "learning_rate": 5.515833833311222e-07,
      "loss": 1.9988,
      "step": 225955
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.451773762702942,
      "learning_rate": 5.513354730846909e-07,
      "loss": 1.9862,
      "step": 225956
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.4734854698181152,
      "learning_rate": 5.510876185114255e-07,
      "loss": 1.9113,
      "step": 225957
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.5491713285446167,
      "learning_rate": 5.508398196113595e-07,
      "loss": 1.8725,
      "step": 225958
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.437921404838562,
      "learning_rate": 5.505920763845262e-07,
      "loss": 1.919,
      "step": 225959
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.4232165813446045,
      "learning_rate": 5.503443888310255e-07,
      "loss": 2.1439,
      "step": 225960
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.4197561740875244,
      "learning_rate": 5.500967569508574e-07,
      "loss": 1.8522,
      "step": 225961
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.4848953485488892,
      "learning_rate": 5.498491807440886e-07,
      "loss": 1.9671,
      "step": 225962
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.383528470993042,
      "learning_rate": 5.496016602107523e-07,
      "loss": 1.9213,
      "step": 225963
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.485714316368103,
      "learning_rate": 5.493541953508818e-07,
      "loss": 2.0177,
      "step": 225964
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.6162288188934326,
      "learning_rate": 5.491067861645437e-07,
      "loss": 2.2459,
      "step": 225965
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.482940912246704,
      "learning_rate": 5.488594326517715e-07,
      "loss": 2.0558,
      "step": 225966
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.514829397201538,
      "learning_rate": 5.486121348126316e-07,
      "loss": 1.9182,
      "step": 225967
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.4924365282058716,
      "learning_rate": 5.483648926471573e-07,
      "loss": 2.0528,
      "step": 225968
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.6674264669418335,
      "learning_rate": 5.481177061553821e-07,
      "loss": 2.0971,
      "step": 225969
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.605863332748413,
      "learning_rate": 5.478705753373724e-07,
      "loss": 2.0521,
      "step": 225970
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.4402295351028442,
      "learning_rate": 5.476235001931617e-07,
      "loss": 1.906,
      "step": 225971
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.4933968782424927,
      "learning_rate": 5.473764807228165e-07,
      "loss": 2.0604,
      "step": 225972
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.4402529001235962,
      "learning_rate": 5.471295169263701e-07,
      "loss": 1.9872,
      "step": 225973
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.5351353883743286,
      "learning_rate": 5.468826088038558e-07,
      "loss": 1.8393,
      "step": 225974
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.5120794773101807,
      "learning_rate": 5.46635756355307e-07,
      "loss": 2.0318,
      "step": 225975
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.6691020727157593,
      "learning_rate": 5.463889595808236e-07,
      "loss": 1.8689,
      "step": 225976
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.3528352975845337,
      "learning_rate": 5.461422184804054e-07,
      "loss": 1.788,
      "step": 225977
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.423046588897705,
      "learning_rate": 5.458955330541193e-07,
      "loss": 2.1513,
      "step": 225978
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.4906165599822998,
      "learning_rate": 5.456489033019984e-07,
      "loss": 2.1123,
      "step": 225979
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.41973078250885,
      "learning_rate": 5.454023292240761e-07,
      "loss": 2.0689,
      "step": 225980
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.457097053527832,
      "learning_rate": 5.451558108204191e-07,
      "loss": 2.0609,
      "step": 225981
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.4263638257980347,
      "learning_rate": 5.449093480910937e-07,
      "loss": 2.027,
      "step": 225982
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.4421899318695068,
      "learning_rate": 5.446629410361003e-07,
      "loss": 2.2497,
      "step": 225983
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.3721866607666016,
      "learning_rate": 5.444165896555386e-07,
      "loss": 1.7595,
      "step": 225984
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.5163664817810059,
      "learning_rate": 5.441702939493753e-07,
      "loss": 1.9809,
      "step": 225985
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.491640567779541,
      "learning_rate": 5.439240539177437e-07,
      "loss": 1.9594,
      "step": 225986
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.4120255708694458,
      "learning_rate": 5.436778695606437e-07,
      "loss": 2.0671,
      "step": 225987
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.7707573175430298,
      "learning_rate": 5.434317408781086e-07,
      "loss": 1.8118,
      "step": 225988
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.4961977005004883,
      "learning_rate": 5.431856678702052e-07,
      "loss": 1.9099,
      "step": 225989
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.570521354675293,
      "learning_rate": 5.429396505369998e-07,
      "loss": 1.9796,
      "step": 225990
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.5332701206207275,
      "learning_rate": 5.426936888784927e-07,
      "loss": 2.1368,
      "step": 225991
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.3892959356307983,
      "learning_rate": 5.424477828947504e-07,
      "loss": 1.8543,
      "step": 225992
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.412738561630249,
      "learning_rate": 5.422019325858395e-07,
      "loss": 2.0656,
      "step": 225993
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.3958168029785156,
      "learning_rate": 5.419561379517601e-07,
      "loss": 1.8942,
      "step": 225994
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.563995599746704,
      "learning_rate": 5.417103989925786e-07,
      "loss": 1.8644,
      "step": 225995
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.492706060409546,
      "learning_rate": 5.414647157083618e-07,
      "loss": 1.8614,
      "step": 225996
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.4165732860565186,
      "learning_rate": 5.41219088099143e-07,
      "loss": 1.9397,
      "step": 225997
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.5001543760299683,
      "learning_rate": 5.409735161649553e-07,
      "loss": 1.9298,
      "step": 225998
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.4449840784072876,
      "learning_rate": 5.407279999058655e-07,
      "loss": 2.1697,
      "step": 225999
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.6274099349975586,
      "learning_rate": 5.40482539321907e-07,
      "loss": 1.953,
      "step": 226000
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.4882389307022095,
      "learning_rate": 5.402371344131128e-07,
      "loss": 2.072,
      "step": 226001
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.4058303833007812,
      "learning_rate": 5.399917851795499e-07,
      "loss": 1.9859,
      "step": 226002
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.5481624603271484,
      "learning_rate": 5.397464916212846e-07,
      "loss": 2.0281,
      "step": 226003
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.5597728490829468,
      "learning_rate": 5.395012537382836e-07,
      "loss": 1.9918,
      "step": 226004
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.552578330039978,
      "learning_rate": 5.39256071530647e-07,
      "loss": 2.1109,
      "step": 226005
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.400267481803894,
      "learning_rate": 5.390109449984414e-07,
      "loss": 2.0925,
      "step": 226006
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.3656991720199585,
      "learning_rate": 5.387658741416667e-07,
      "loss": 1.7457,
      "step": 226007
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.4480962753295898,
      "learning_rate": 5.385208589603895e-07,
      "loss": 1.9606,
      "step": 226008
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.352919101715088,
      "learning_rate": 5.382758994546765e-07,
      "loss": 1.7657,
      "step": 226009
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.4580085277557373,
      "learning_rate": 5.380309956245276e-07,
      "loss": 1.8558,
      "step": 226010
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.3943586349487305,
      "learning_rate": 5.377861474700429e-07,
      "loss": 2.0103,
      "step": 226011
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.2735977172851562,
      "learning_rate": 5.375413549911889e-07,
      "loss": 1.898,
      "step": 226012
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.5270311832427979,
      "learning_rate": 5.372966181880988e-07,
      "loss": 2.098,
      "step": 226013
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.4776217937469482,
      "learning_rate": 5.370519370607396e-07,
      "loss": 2.1553,
      "step": 226014
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.5845261812210083,
      "learning_rate": 5.368073116092442e-07,
      "loss": 1.8927,
      "step": 226015
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.5212702751159668,
      "learning_rate": 5.365627418335794e-07,
      "loss": 2.085,
      "step": 226016
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.46273934841156,
      "learning_rate": 5.363182277338119e-07,
      "loss": 2.0178,
      "step": 226017
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.4596971273422241,
      "learning_rate": 5.360737693100081e-07,
      "loss": 2.018,
      "step": 226018
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.3039287328720093,
      "learning_rate": 5.358293665622015e-07,
      "loss": 1.8163,
      "step": 226019
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.54097580909729,
      "learning_rate": 5.355850194904254e-07,
      "loss": 2.0851,
      "step": 226020
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.5628949403762817,
      "learning_rate": 5.353407280947464e-07,
      "loss": 1.9855,
      "step": 226021
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.41927170753479,
      "learning_rate": 5.350964923751977e-07,
      "loss": 1.9927,
      "step": 226022
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.4163233041763306,
      "learning_rate": 5.34852312331846e-07,
      "loss": 2.0546,
      "step": 226023
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.6033891439437866,
      "learning_rate": 5.346081879646913e-07,
      "loss": 2.0065,
      "step": 226024
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.3347887992858887,
      "learning_rate": 5.343641192738335e-07,
      "loss": 1.838,
      "step": 226025
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.637147307395935,
      "learning_rate": 5.341201062592393e-07,
      "loss": 1.9557,
      "step": 226026
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.3258916139602661,
      "learning_rate": 5.338761489210752e-07,
      "loss": 1.8758,
      "step": 226027
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.4088571071624756,
      "learning_rate": 5.336322472592747e-07,
      "loss": 2.0784,
      "step": 226028
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.4845707416534424,
      "learning_rate": 5.333884012739376e-07,
      "loss": 2.1334,
      "step": 226029
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.5428200960159302,
      "learning_rate": 5.331446109650639e-07,
      "loss": 1.8311,
      "step": 226030
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.4724273681640625,
      "learning_rate": 5.329008763327536e-07,
      "loss": 2.1985,
      "step": 226031
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.5305482149124146,
      "learning_rate": 5.326571973770399e-07,
      "loss": 2.032,
      "step": 226032
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.5612074136734009,
      "learning_rate": 5.324135740979563e-07,
      "loss": 1.9274,
      "step": 226033
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.439103126525879,
      "learning_rate": 5.321700064955359e-07,
      "loss": 2.2538,
      "step": 226034
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.4302647113800049,
      "learning_rate": 5.319264945698787e-07,
      "loss": 1.9809,
      "step": 226035
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.421940565109253,
      "learning_rate": 5.316830383209514e-07,
      "loss": 2.0859,
      "step": 226036
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.4586647748947144,
      "learning_rate": 5.314396377488205e-07,
      "loss": 2.011,
      "step": 226037
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.5268657207489014,
      "learning_rate": 5.311962928535862e-07,
      "loss": 1.7837,
      "step": 226038
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.3495293855667114,
      "learning_rate": 5.309530036352483e-07,
      "loss": 2.0144,
      "step": 226039
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.5458499193191528,
      "learning_rate": 5.307097700938401e-07,
      "loss": 2.0592,
      "step": 226040
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.5560362339019775,
      "learning_rate": 5.304665922294282e-07,
      "loss": 1.8487,
      "step": 226041
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.5372035503387451,
      "learning_rate": 5.302234700420794e-07,
      "loss": 1.9619,
      "step": 226042
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.5046234130859375,
      "learning_rate": 5.299804035317934e-07,
      "loss": 1.7861,
      "step": 226043
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.4258768558502197,
      "learning_rate": 5.29737392698637e-07,
      "loss": 1.9743,
      "step": 226044
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.4962396621704102,
      "learning_rate": 5.294944375426769e-07,
      "loss": 1.8149,
      "step": 226045
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.501184105873108,
      "learning_rate": 5.292515380639129e-07,
      "loss": 1.9498,
      "step": 226046
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.5292199850082397,
      "learning_rate": 5.290086942624117e-07,
      "loss": 2.1841,
      "step": 226047
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.3876049518585205,
      "learning_rate": 5.287659061382399e-07,
      "loss": 1.7995,
      "step": 226048
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.8340507745742798,
      "learning_rate": 5.285231736914309e-07,
      "loss": 2.0286,
      "step": 226049
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.4501268863677979,
      "learning_rate": 5.282804969219845e-07,
      "loss": 1.8532,
      "step": 226050
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.5222759246826172,
      "learning_rate": 5.280378758300008e-07,
      "loss": 1.9906,
      "step": 226051
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.5535112619400024,
      "learning_rate": 5.277953104155131e-07,
      "loss": 1.76,
      "step": 226052
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.50673246383667,
      "learning_rate": 5.275528006785879e-07,
      "loss": 2.0795,
      "step": 226053
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.542542815208435,
      "learning_rate": 5.27310346619192e-07,
      "loss": 2.0288,
      "step": 226054
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.4438647031784058,
      "learning_rate": 5.270679482374585e-07,
      "loss": 1.8104,
      "step": 226055
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.939003825187683,
      "learning_rate": 5.268256055333875e-07,
      "loss": 2.0435,
      "step": 226056
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.3597350120544434,
      "learning_rate": 5.265833185070123e-07,
      "loss": 2.0829,
      "step": 226057
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.4548460245132446,
      "learning_rate": 5.263410871584329e-07,
      "loss": 1.866,
      "step": 226058
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.4594496488571167,
      "learning_rate": 5.26098911487649e-07,
      "loss": 1.9608,
      "step": 226059
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.417203664779663,
      "learning_rate": 5.258567914946943e-07,
      "loss": 2.0866,
      "step": 226060
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.3653866052627563,
      "learning_rate": 5.256147271796684e-07,
      "loss": 2.0466,
      "step": 226061
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.4245998859405518,
      "learning_rate": 5.253727185425383e-07,
      "loss": 2.1419,
      "step": 226062
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.43441903591156,
      "learning_rate": 5.251307655834369e-07,
      "loss": 2.0414,
      "step": 226063
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.4940181970596313,
      "learning_rate": 5.248888683023644e-07,
      "loss": 1.7639,
      "step": 226064
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.4896267652511597,
      "learning_rate": 5.246470266993541e-07,
      "loss": 2.0071,
      "step": 226065
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.4859983921051025,
      "learning_rate": 5.244052407744726e-07,
      "loss": 1.9729,
      "step": 226066
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.4434138536453247,
      "learning_rate": 5.241635105277531e-07,
      "loss": 2.1445,
      "step": 226067
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.5224930047988892,
      "learning_rate": 5.23921835959229e-07,
      "loss": 2.2109,
      "step": 226068
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.469130516052246,
      "learning_rate": 5.236802170689669e-07,
      "loss": 1.9163,
      "step": 226069
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.4302330017089844,
      "learning_rate": 5.234386538570334e-07,
      "loss": 1.8951,
      "step": 226070
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.4252911806106567,
      "learning_rate": 5.231971463233952e-07,
      "loss": 2.0911,
      "step": 226071
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.4017930030822754,
      "learning_rate": 5.229556944681856e-07,
      "loss": 1.9627,
      "step": 226072
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.6615530252456665,
      "learning_rate": 5.227142982914045e-07,
      "loss": 2.1902,
      "step": 226073
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.40251624584198,
      "learning_rate": 5.224729577931186e-07,
      "loss": 2.0431,
      "step": 226074
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.4343894720077515,
      "learning_rate": 5.222316729733277e-07,
      "loss": 1.8703,
      "step": 226075
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.5893704891204834,
      "learning_rate": 5.21990443832132e-07,
      "loss": 1.9359,
      "step": 226076
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.449252963066101,
      "learning_rate": 5.217492703695314e-07,
      "loss": 1.9322,
      "step": 226077
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.5029823780059814,
      "learning_rate": 5.215081525856257e-07,
      "loss": 1.6723,
      "step": 226078
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.3911396265029907,
      "learning_rate": 5.212670904803818e-07,
      "loss": 2.0766,
      "step": 226079
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.5324463844299316,
      "learning_rate": 5.210260840538994e-07,
      "loss": 1.9444,
      "step": 226080
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.4610559940338135,
      "learning_rate": 5.207851333062119e-07,
      "loss": 2.2444,
      "step": 226081
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.48759126663208,
      "learning_rate": 5.205442382373858e-07,
      "loss": 1.7739,
      "step": 226082
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.5424250364303589,
      "learning_rate": 5.203033988474215e-07,
      "loss": 1.9462,
      "step": 226083
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.5178879499435425,
      "learning_rate": 5.200626151364185e-07,
      "loss": 1.8747,
      "step": 226084
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.444105863571167,
      "learning_rate": 5.198218871043436e-07,
      "loss": 1.8086,
      "step": 226085
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.5279099941253662,
      "learning_rate": 5.195812147512967e-07,
      "loss": 2.0492,
      "step": 226086
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.6952660083770752,
      "learning_rate": 5.193405980773446e-07,
      "loss": 1.9514,
      "step": 226087
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.444380283355713,
      "learning_rate": 5.191000370824872e-07,
      "loss": 2.0236,
      "step": 226088
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.3808093070983887,
      "learning_rate": 5.188595317667577e-07,
      "loss": 2.1643,
      "step": 226089
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.4896084070205688,
      "learning_rate": 5.186190821302227e-07,
      "loss": 1.8585,
      "step": 226090
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.634325385093689,
      "learning_rate": 5.183786881729491e-07,
      "loss": 1.9497,
      "step": 226091
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.5175890922546387,
      "learning_rate": 5.181383498949699e-07,
      "loss": 1.9462,
      "step": 226092
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.5254065990447998,
      "learning_rate": 5.178980672963184e-07,
      "loss": 2.2064,
      "step": 226093
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.6092214584350586,
      "learning_rate": 5.176578403770282e-07,
      "loss": 1.9463,
      "step": 226094
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.3589158058166504,
      "learning_rate": 5.174176691371656e-07,
      "loss": 2.094,
      "step": 226095
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.4809982776641846,
      "learning_rate": 5.171775535767642e-07,
      "loss": 1.7858,
      "step": 226096
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.5248746871948242,
      "learning_rate": 5.169374936958903e-07,
      "loss": 2.1304,
      "step": 226097
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.4853936433792114,
      "learning_rate": 5.166974894945442e-07,
      "loss": 1.869,
      "step": 226098
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.470510482788086,
      "learning_rate": 5.164575409728256e-07,
      "loss": 1.9154,
      "step": 226099
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.491588830947876,
      "learning_rate": 5.162176481307346e-07,
      "loss": 1.8556,
      "step": 226100
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.395979881286621,
      "learning_rate": 5.159778109683377e-07,
      "loss": 2.0888,
      "step": 226101
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.4973535537719727,
      "learning_rate": 5.157380294856683e-07,
      "loss": 2.0117,
      "step": 226102
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.5184880495071411,
      "learning_rate": 5.154983036827598e-07,
      "loss": 2.1376,
      "step": 226103
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.422207236289978,
      "learning_rate": 5.15258633559712e-07,
      "loss": 1.7748,
      "step": 226104
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.3858331441879272,
      "learning_rate": 5.150190191164916e-07,
      "loss": 2.1649,
      "step": 226105
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.516014575958252,
      "learning_rate": 5.147794603532319e-07,
      "loss": 2.0003,
      "step": 226106
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.423300862312317,
      "learning_rate": 5.145399572698661e-07,
      "loss": 1.9572,
      "step": 226107
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.4280818700790405,
      "learning_rate": 5.143005098665609e-07,
      "loss": 2.0596,
      "step": 226108
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.3576980829238892,
      "learning_rate": 5.140611181432496e-07,
      "loss": 1.738,
      "step": 226109
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.45811128616333,
      "learning_rate": 5.138217821000656e-07,
      "loss": 2.0295,
      "step": 226110
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.4210302829742432,
      "learning_rate": 5.135825017369755e-07,
      "loss": 1.9515,
      "step": 226111
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.511093258857727,
      "learning_rate": 5.133432770541124e-07,
      "loss": 1.9779,
      "step": 226112
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.4931411743164062,
      "learning_rate": 5.13104108051443e-07,
      "loss": 2.156,
      "step": 226113
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.4153584241867065,
      "learning_rate": 5.128649947290342e-07,
      "loss": 1.9117,
      "step": 226114
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.4181891679763794,
      "learning_rate": 5.126259370869523e-07,
      "loss": 1.8931,
      "step": 226115
    },
    {
      "epoch": 2.94,
      "grad_norm": 2.5086417198181152,
      "learning_rate": 5.123869351252307e-07,
      "loss": 1.9501,
      "step": 226116
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.4199916124343872,
      "learning_rate": 5.121479888439029e-07,
      "loss": 1.827,
      "step": 226117
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.4415085315704346,
      "learning_rate": 5.119090982430019e-07,
      "loss": 1.923,
      "step": 226118
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.4956589937210083,
      "learning_rate": 5.116702633226277e-07,
      "loss": 1.7508,
      "step": 226119
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.447381854057312,
      "learning_rate": 5.114314840827471e-07,
      "loss": 1.8293,
      "step": 226120
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.5145727396011353,
      "learning_rate": 5.111927605234601e-07,
      "loss": 1.9407,
      "step": 226121
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.393262267112732,
      "learning_rate": 5.109540926447997e-07,
      "loss": 1.864,
      "step": 226122
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.3770345449447632,
      "learning_rate": 5.107154804467995e-07,
      "loss": 2.0577,
      "step": 226123
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.5087807178497314,
      "learning_rate": 5.104769239295259e-07,
      "loss": 1.8923,
      "step": 226124
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.3743621110916138,
      "learning_rate": 5.102384230930123e-07,
      "loss": 2.0937,
      "step": 226125
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.426265835762024,
      "learning_rate": 5.09999977937292e-07,
      "loss": 1.9677,
      "step": 226126
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.3831952810287476,
      "learning_rate": 5.097615884623984e-07,
      "loss": 2.0458,
      "step": 226127
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.5038535594940186,
      "learning_rate": 5.095232546683981e-07,
      "loss": 1.903,
      "step": 226128
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.5468528270721436,
      "learning_rate": 5.092849765553242e-07,
      "loss": 2.1145,
      "step": 226129
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.5007734298706055,
      "learning_rate": 5.090467541232435e-07,
      "loss": 2.1382,
      "step": 226130
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.5114452838897705,
      "learning_rate": 5.088085873721893e-07,
      "loss": 2.0265,
      "step": 226131
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.2813661098480225,
      "learning_rate": 5.085704763021947e-07,
      "loss": 1.6144,
      "step": 226132
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.494290828704834,
      "learning_rate": 5.083324209132933e-07,
      "loss": 2.0258,
      "step": 226133
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.5856245756149292,
      "learning_rate": 5.080944212055849e-07,
      "loss": 2.0765,
      "step": 226134
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.3698152303695679,
      "learning_rate": 5.078564771790361e-07,
      "loss": 2.0522,
      "step": 226135
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.5475172996520996,
      "learning_rate": 5.076185888337469e-07,
      "loss": 2.0565,
      "step": 226136
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.5872242450714111,
      "learning_rate": 5.073807561697507e-07,
      "loss": 2.2481,
      "step": 226137
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.6258111000061035,
      "learning_rate": 5.071429791870807e-07,
      "loss": 2.0835,
      "step": 226138
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.4912323951721191,
      "learning_rate": 5.069052578857702e-07,
      "loss": 2.0352,
      "step": 226139
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.4722906351089478,
      "learning_rate": 5.066675922658858e-07,
      "loss": 2.1341,
      "step": 226140
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.4806522130966187,
      "learning_rate": 5.064299823274942e-07,
      "loss": 2.0416,
      "step": 226141
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.293993353843689,
      "learning_rate": 5.061924280705621e-07,
      "loss": 1.7363,
      "step": 226142
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.3708462715148926,
      "learning_rate": 5.059549294952226e-07,
      "loss": 2.1824,
      "step": 226143
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.4582500457763672,
      "learning_rate": 5.057174866014758e-07,
      "loss": 1.9469,
      "step": 226144
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.4023797512054443,
      "learning_rate": 5.054800993893549e-07,
      "loss": 1.9796,
      "step": 226145
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.5263011455535889,
      "learning_rate": 5.052427678589266e-07,
      "loss": 1.9337,
      "step": 226146
    },
    {
      "epoch": 2.94,
      "grad_norm": 2.0633883476257324,
      "learning_rate": 5.050054920102243e-07,
      "loss": 1.9946,
      "step": 226147
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.4486886262893677,
      "learning_rate": 5.04768271843281e-07,
      "loss": 2.0178,
      "step": 226148
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.472208023071289,
      "learning_rate": 5.045311073581637e-07,
      "loss": 1.95,
      "step": 226149
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.8117117881774902,
      "learning_rate": 5.042939985549055e-07,
      "loss": 2.2384,
      "step": 226150
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.5200583934783936,
      "learning_rate": 5.04056945433573e-07,
      "loss": 2.1063,
      "step": 226151
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.5931141376495361,
      "learning_rate": 5.038199479941663e-07,
      "loss": 2.1271,
      "step": 226152
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.4666671752929688,
      "learning_rate": 5.035830062367519e-07,
      "loss": 1.9097,
      "step": 226153
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.5677868127822876,
      "learning_rate": 5.033461201613964e-07,
      "loss": 2.1589,
      "step": 226154
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.5718886852264404,
      "learning_rate": 5.031092897681332e-07,
      "loss": 1.865,
      "step": 226155
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.4782124757766724,
      "learning_rate": 5.028725150569624e-07,
      "loss": 1.9344,
      "step": 226156
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.4448727369308472,
      "learning_rate": 5.026357960279504e-07,
      "loss": 2.0249,
      "step": 226157
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.6212153434753418,
      "learning_rate": 5.023991326811971e-07,
      "loss": 2.2027,
      "step": 226158
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.3740997314453125,
      "learning_rate": 5.021625250166694e-07,
      "loss": 2.0975,
      "step": 226159
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.5225788354873657,
      "learning_rate": 5.019259730344671e-07,
      "loss": 2.0541,
      "step": 226160
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.540053129196167,
      "learning_rate": 5.016894767345902e-07,
      "loss": 1.9596,
      "step": 226161
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.4711840152740479,
      "learning_rate": 5.014530361171055e-07,
      "loss": 1.8211,
      "step": 226162
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.3708916902542114,
      "learning_rate": 5.012166511820459e-07,
      "loss": 1.8678,
      "step": 226163
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.4704757928848267,
      "learning_rate": 5.009803219294783e-07,
      "loss": 1.7804,
      "step": 226164
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.5280512571334839,
      "learning_rate": 5.007440483594027e-07,
      "loss": 2.0787,
      "step": 226165
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.33469820022583,
      "learning_rate": 5.00507830471919e-07,
      "loss": 1.9137,
      "step": 226166
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.5323060750961304,
      "learning_rate": 5.002716682670604e-07,
      "loss": 1.8115,
      "step": 226167
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.6458622217178345,
      "learning_rate": 5.000355617448271e-07,
      "loss": 1.9288,
      "step": 226168
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.4584051370620728,
      "learning_rate": 4.997995109053188e-07,
      "loss": 1.9752,
      "step": 226169
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.5056053400039673,
      "learning_rate": 4.995635157485023e-07,
      "loss": 2.0138,
      "step": 226170
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.4280028343200684,
      "learning_rate": 4.993275762745108e-07,
      "loss": 2.2904,
      "step": 226171
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.5428612232208252,
      "learning_rate": 4.990916924833444e-07,
      "loss": 1.915,
      "step": 226172
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.9527947902679443,
      "learning_rate": 4.988558643750362e-07,
      "loss": 1.9336,
      "step": 226173
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.5152949094772339,
      "learning_rate": 4.98620091949653e-07,
      "loss": 1.9353,
      "step": 226174
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.442149043083191,
      "learning_rate": 4.98384375207228e-07,
      "loss": 1.8743,
      "step": 226175
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.4630154371261597,
      "learning_rate": 4.981487141477947e-07,
      "loss": 1.8309,
      "step": 226176
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.4131498336791992,
      "learning_rate": 4.979131087714194e-07,
      "loss": 1.8224,
      "step": 226177
    },
    {
      "epoch": 2.94,
      "grad_norm": 1.56865394115448,
      "learning_rate": 4.976775590781356e-07,
      "loss": 2.1306,
      "step": 226178
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.4226107597351074,
      "learning_rate": 4.9744206506801e-07,
      "loss": 1.9446,
      "step": 226179
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.379984974861145,
      "learning_rate": 4.972066267410424e-07,
      "loss": 2.0799,
      "step": 226180
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.4980735778808594,
      "learning_rate": 4.969712440972995e-07,
      "loss": 1.8304,
      "step": 226181
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.5716911554336548,
      "learning_rate": 4.967359171368146e-07,
      "loss": 1.6514,
      "step": 226182
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.381853461265564,
      "learning_rate": 4.965006458596543e-07,
      "loss": 1.7951,
      "step": 226183
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.4974623918533325,
      "learning_rate": 4.962654302658187e-07,
      "loss": 1.9955,
      "step": 226184
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.5306252241134644,
      "learning_rate": 4.960302703554075e-07,
      "loss": 1.8549,
      "step": 226185
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.5133651494979858,
      "learning_rate": 4.957951661284543e-07,
      "loss": 1.9278,
      "step": 226186
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.4966007471084595,
      "learning_rate": 4.955601175849589e-07,
      "loss": 1.9487,
      "step": 226187
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.4552193880081177,
      "learning_rate": 4.953251247249879e-07,
      "loss": 2.0063,
      "step": 226188
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.5479260683059692,
      "learning_rate": 4.95090187548608e-07,
      "loss": 2.0819,
      "step": 226189
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.4995239973068237,
      "learning_rate": 4.948553060558191e-07,
      "loss": 1.9517,
      "step": 226190
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.4947383403778076,
      "learning_rate": 4.946204802467213e-07,
      "loss": 2.0528,
      "step": 226191
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.378674864768982,
      "learning_rate": 4.943857101213144e-07,
      "loss": 1.9574,
      "step": 226192
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.6461353302001953,
      "learning_rate": 4.941509956796652e-07,
      "loss": 1.7392,
      "step": 226193
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.5858309268951416,
      "learning_rate": 4.939163369217736e-07,
      "loss": 2.097,
      "step": 226194
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.6660915613174438,
      "learning_rate": 4.936817338477394e-07,
      "loss": 2.0447,
      "step": 226195
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.412069320678711,
      "learning_rate": 4.934471864575962e-07,
      "loss": 1.7775,
      "step": 226196
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.5088433027267456,
      "learning_rate": 4.932126947513771e-07,
      "loss": 2.0246,
      "step": 226197
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.3957664966583252,
      "learning_rate": 4.92978258729082e-07,
      "loss": 1.9215,
      "step": 226198
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.4859482049942017,
      "learning_rate": 4.927438783908444e-07,
      "loss": 2.2201,
      "step": 226199
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.4204449653625488,
      "learning_rate": 4.925095537366309e-07,
      "loss": 2.0654,
      "step": 226200
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.4953091144561768,
      "learning_rate": 4.922752847665079e-07,
      "loss": 1.9494,
      "step": 226201
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.658820390701294,
      "learning_rate": 4.920410714805423e-07,
      "loss": 2.0088,
      "step": 226202
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.4420702457427979,
      "learning_rate": 4.91806913878734e-07,
      "loss": 1.9585,
      "step": 226203
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.499048113822937,
      "learning_rate": 4.915728119611828e-07,
      "loss": 1.9971,
      "step": 226204
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.3987987041473389,
      "learning_rate": 4.913387657278888e-07,
      "loss": 1.9821,
      "step": 226205
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.476716160774231,
      "learning_rate": 4.911047751789188e-07,
      "loss": 1.9773,
      "step": 226206
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.4131665229797363,
      "learning_rate": 4.908708403143058e-07,
      "loss": 1.9944,
      "step": 226207
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.3610641956329346,
      "learning_rate": 4.906369611340832e-07,
      "loss": 1.9905,
      "step": 226208
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.6591869592666626,
      "learning_rate": 4.904031376383177e-07,
      "loss": 1.9775,
      "step": 226209
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.5045214891433716,
      "learning_rate": 4.901693698270092e-07,
      "loss": 1.9455,
      "step": 226210
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.5818761587142944,
      "learning_rate": 4.899356577002577e-07,
      "loss": 2.1206,
      "step": 226211
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.5268306732177734,
      "learning_rate": 4.897020012580632e-07,
      "loss": 2.0721,
      "step": 226212
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.3980603218078613,
      "learning_rate": 4.894684005005256e-07,
      "loss": 2.1638,
      "step": 226213
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.4087821245193481,
      "learning_rate": 4.892348554276116e-07,
      "loss": 2.1589,
      "step": 226214
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.5624092817306519,
      "learning_rate": 4.89001366039421e-07,
      "loss": 1.9177,
      "step": 226215
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.5367975234985352,
      "learning_rate": 4.887679323359539e-07,
      "loss": 1.858,
      "step": 226216
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.4854642152786255,
      "learning_rate": 4.885345543173102e-07,
      "loss": 1.8831,
      "step": 226217
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.459694504737854,
      "learning_rate": 4.8830123198349e-07,
      "loss": 2.033,
      "step": 226218
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.6245027780532837,
      "learning_rate": 4.880679653345265e-07,
      "loss": 2.0627,
      "step": 226219
    },
    {
      "epoch": 2.95,
      "grad_norm": 2.449052572250366,
      "learning_rate": 4.878347543705196e-07,
      "loss": 1.908,
      "step": 226220
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.5913618803024292,
      "learning_rate": 4.876015990914695e-07,
      "loss": 2.0258,
      "step": 226221
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.3896552324295044,
      "learning_rate": 4.873684994974092e-07,
      "loss": 1.9063,
      "step": 226222
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.533759593963623,
      "learning_rate": 4.871354555884388e-07,
      "loss": 1.98,
      "step": 226223
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.6952790021896362,
      "learning_rate": 4.86902467364525e-07,
      "loss": 2.0222,
      "step": 226224
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.8120832443237305,
      "learning_rate": 4.866695348257676e-07,
      "loss": 1.901,
      "step": 226225
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.469312310218811,
      "learning_rate": 4.864366579721668e-07,
      "loss": 1.8522,
      "step": 226226
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.5339524745941162,
      "learning_rate": 4.862038368038223e-07,
      "loss": 2.2109,
      "step": 226227
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.7243132591247559,
      "learning_rate": 4.859710713207343e-07,
      "loss": 2.1594,
      "step": 226228
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.4286612272262573,
      "learning_rate": 4.857383615229692e-07,
      "loss": 1.8414,
      "step": 226229
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.4307403564453125,
      "learning_rate": 4.855057074105273e-07,
      "loss": 1.9106,
      "step": 226230
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.4434300661087036,
      "learning_rate": 4.852731089835083e-07,
      "loss": 1.9093,
      "step": 226231
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.3663009405136108,
      "learning_rate": 4.850405662419454e-07,
      "loss": 1.8366,
      "step": 226232
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.5885406732559204,
      "learning_rate": 4.848080791858389e-07,
      "loss": 2.0226,
      "step": 226233
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.4299793243408203,
      "learning_rate": 4.845756478152884e-07,
      "loss": 2.0735,
      "step": 226234
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.554510474205017,
      "learning_rate": 4.843432721302609e-07,
      "loss": 1.8399,
      "step": 226235
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.3778069019317627,
      "learning_rate": 4.841109521308894e-07,
      "loss": 2.137,
      "step": 226236
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.5199013948440552,
      "learning_rate": 4.838786878171741e-07,
      "loss": 2.2297,
      "step": 226237
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.4263674020767212,
      "learning_rate": 4.836464791891481e-07,
      "loss": 1.9019,
      "step": 226238
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.5104866027832031,
      "learning_rate": 4.834143262468781e-07,
      "loss": 2.1301,
      "step": 226239
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.5459305047988892,
      "learning_rate": 4.831822289903642e-07,
      "loss": 2.0332,
      "step": 226240
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.5108617544174194,
      "learning_rate": 4.829501874197061e-07,
      "loss": 1.9587,
      "step": 226241
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.4870541095733643,
      "learning_rate": 4.827182015349373e-07,
      "loss": 2.1005,
      "step": 226242
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.4726765155792236,
      "learning_rate": 4.824862713360578e-07,
      "loss": 1.7001,
      "step": 226243
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.435990810394287,
      "learning_rate": 4.82254396823134e-07,
      "loss": 1.9113,
      "step": 226244
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.416591763496399,
      "learning_rate": 4.820225779962328e-07,
      "loss": 2.0577,
      "step": 226245
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.3544071912765503,
      "learning_rate": 4.81790814855354e-07,
      "loss": 1.7673,
      "step": 226246
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.6046382188796997,
      "learning_rate": 4.815591074005975e-07,
      "loss": 2.0245,
      "step": 226247
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.3573507070541382,
      "learning_rate": 4.813274556319635e-07,
      "loss": 1.8915,
      "step": 226248
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.4720476865768433,
      "learning_rate": 4.810958595494851e-07,
      "loss": 1.9967,
      "step": 226249
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.4115644693374634,
      "learning_rate": 4.808643191532291e-07,
      "loss": 1.9644,
      "step": 226250
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.4010982513427734,
      "learning_rate": 4.806328344432286e-07,
      "loss": 1.92,
      "step": 226251
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.4373668432235718,
      "learning_rate": 4.804014054195505e-07,
      "loss": 1.7481,
      "step": 226252
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.3382821083068848,
      "learning_rate": 4.801700320822277e-07,
      "loss": 2.011,
      "step": 226253
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.3363523483276367,
      "learning_rate": 4.799387144312606e-07,
      "loss": 2.0547,
      "step": 226254
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.4409736394882202,
      "learning_rate": 4.797074524667488e-07,
      "loss": 2.0323,
      "step": 226255
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.5110116004943848,
      "learning_rate": 4.794762461887259e-07,
      "loss": 2.1139,
      "step": 226256
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.5833063125610352,
      "learning_rate": 4.792450955971916e-07,
      "loss": 2.0236,
      "step": 226257
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.4233077764511108,
      "learning_rate": 4.790140006922461e-07,
      "loss": 2.0041,
      "step": 226258
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.5544580221176147,
      "learning_rate": 4.787829614738892e-07,
      "loss": 2.1796,
      "step": 226259
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.4389591217041016,
      "learning_rate": 4.785519779421876e-07,
      "loss": 1.9128,
      "step": 226260
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.464874267578125,
      "learning_rate": 4.783210500971745e-07,
      "loss": 2.1505,
      "step": 226261
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.3835601806640625,
      "learning_rate": 4.780901779388835e-07,
      "loss": 2.1584,
      "step": 226262
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.5755391120910645,
      "learning_rate": 4.778593614673809e-07,
      "loss": 1.9382,
      "step": 226263
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.531285285949707,
      "learning_rate": 4.776286006827334e-07,
      "loss": 1.8442,
      "step": 226264
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.559231162071228,
      "learning_rate": 4.773978955849079e-07,
      "loss": 2.1959,
      "step": 226265
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.3522895574569702,
      "learning_rate": 4.771672461740039e-07,
      "loss": 1.984,
      "step": 226266
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.4706674814224243,
      "learning_rate": 4.769366524500217e-07,
      "loss": 2.2,
      "step": 226267
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.4315401315689087,
      "learning_rate": 4.767061144130613e-07,
      "loss": 1.9901,
      "step": 226268
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.372371792793274,
      "learning_rate": 4.7647563206312246e-07,
      "loss": 1.8166,
      "step": 226269
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.6109540462493896,
      "learning_rate": 4.762452054002386e-07,
      "loss": 2.1375,
      "step": 226270
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.6328574419021606,
      "learning_rate": 4.760148344245096e-07,
      "loss": 1.9161,
      "step": 226271
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.5682355165481567,
      "learning_rate": 4.757845191359355e-07,
      "loss": 2.0851,
      "step": 226272
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.414050579071045,
      "learning_rate": 4.755542595345496e-07,
      "loss": 1.8129,
      "step": 226273
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.5915846824645996,
      "learning_rate": 4.7532405562045184e-07,
      "loss": 2.0578,
      "step": 226274
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.4840244054794312,
      "learning_rate": 4.7509390739360884e-07,
      "loss": 2.046,
      "step": 226275
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.3679760694503784,
      "learning_rate": 4.748638148540873e-07,
      "loss": 1.9807,
      "step": 226276
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.5634031295776367,
      "learning_rate": 4.746337780019871e-07,
      "loss": 1.8641,
      "step": 226277
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.4745979309082031,
      "learning_rate": 4.7440379683727493e-07,
      "loss": 1.9929,
      "step": 226278
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.5295652151107788,
      "learning_rate": 4.741738713600507e-07,
      "loss": 1.799,
      "step": 226279
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.4720852375030518,
      "learning_rate": 4.739440015703144e-07,
      "loss": 2.027,
      "step": 226280
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.3303284645080566,
      "learning_rate": 4.7371418746809944e-07,
      "loss": 1.8585,
      "step": 226281
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.5890142917633057,
      "learning_rate": 4.734844290535056e-07,
      "loss": 1.7724,
      "step": 226282
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.3172041177749634,
      "learning_rate": 4.7325472632656627e-07,
      "loss": 1.8211,
      "step": 226283
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.4029675722122192,
      "learning_rate": 4.7302507928728137e-07,
      "loss": 1.8909,
      "step": 226284
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.4908123016357422,
      "learning_rate": 4.7279548793568435e-07,
      "loss": 2.0295,
      "step": 226285
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.7322734594345093,
      "learning_rate": 4.725659522719083e-07,
      "loss": 1.9639,
      "step": 226286
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.4109954833984375,
      "learning_rate": 4.723364722958867e-07,
      "loss": 2.0462,
      "step": 226287
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.582184076309204,
      "learning_rate": 4.721070480077194e-07,
      "loss": 1.9164,
      "step": 226288
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.5594770908355713,
      "learning_rate": 4.71877679407473e-07,
      "loss": 1.9233,
      "step": 226289
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.3643378019332886,
      "learning_rate": 4.716483664951143e-07,
      "loss": 2.1253,
      "step": 226290
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.4628010988235474,
      "learning_rate": 4.7141910927077644e-07,
      "loss": 1.9287,
      "step": 226291
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.6147669553756714,
      "learning_rate": 4.711899077344261e-07,
      "loss": 2.0293,
      "step": 226292
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.4870195388793945,
      "learning_rate": 4.7096076188612996e-07,
      "loss": 2.0964,
      "step": 226293
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.433035135269165,
      "learning_rate": 4.707316717259546e-07,
      "loss": 2.1362,
      "step": 226294
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.3775324821472168,
      "learning_rate": 4.7050263725393335e-07,
      "loss": 1.9992,
      "step": 226295
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.5401382446289062,
      "learning_rate": 4.7027365847006615e-07,
      "loss": 1.9429,
      "step": 226296
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.4274930953979492,
      "learning_rate": 4.7004473537445295e-07,
      "loss": 1.9991,
      "step": 226297
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.3271477222442627,
      "learning_rate": 4.6981586796709383e-07,
      "loss": 1.8858,
      "step": 226298
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.4156889915466309,
      "learning_rate": 4.6958705624805527e-07,
      "loss": 1.9571,
      "step": 226299
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.4822288751602173,
      "learning_rate": 4.6935830021740397e-07,
      "loss": 1.808,
      "step": 226300
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.4587137699127197,
      "learning_rate": 4.691295998751399e-07,
      "loss": 1.9881,
      "step": 226301
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.3835549354553223,
      "learning_rate": 4.6890095522129635e-07,
      "loss": 1.8097,
      "step": 226302
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.4993672370910645,
      "learning_rate": 4.6867236625593993e-07,
      "loss": 1.9063,
      "step": 226303
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.4986287355422974,
      "learning_rate": 4.684438329791373e-07,
      "loss": 2.0561,
      "step": 226304
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.4324527978897095,
      "learning_rate": 4.682153553908885e-07,
      "loss": 1.8863,
      "step": 226305
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.5085608959197998,
      "learning_rate": 4.6798693349126004e-07,
      "loss": 1.8381,
      "step": 226306
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.4881772994995117,
      "learning_rate": 4.6775856728028526e-07,
      "loss": 2.0038,
      "step": 226307
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.3713746070861816,
      "learning_rate": 4.675302567579975e-07,
      "loss": 2.057,
      "step": 226308
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.399857997894287,
      "learning_rate": 4.673020019244633e-07,
      "loss": 1.8238,
      "step": 226309
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.3703631162643433,
      "learning_rate": 4.6707380277971605e-07,
      "loss": 1.779,
      "step": 226310
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.6420038938522339,
      "learning_rate": 4.6684565932375574e-07,
      "loss": 2.0531,
      "step": 226311
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.529802918434143,
      "learning_rate": 4.6661757155671554e-07,
      "loss": 1.82,
      "step": 226312
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.4366363286972046,
      "learning_rate": 4.663895394785622e-07,
      "loss": 1.7262,
      "step": 226313
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.3507170677185059,
      "learning_rate": 4.6616156308936223e-07,
      "loss": 1.8304,
      "step": 226314
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.3764797449111938,
      "learning_rate": 4.659336423891491e-07,
      "loss": 1.8936,
      "step": 226315
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.5107873678207397,
      "learning_rate": 4.657057773779893e-07,
      "loss": 2.167,
      "step": 226316
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.3887711763381958,
      "learning_rate": 4.654779680558829e-07,
      "loss": 1.9734,
      "step": 226317
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.5044981241226196,
      "learning_rate": 4.6525021442289644e-07,
      "loss": 1.9508,
      "step": 226318
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.3868725299835205,
      "learning_rate": 4.650225164790966e-07,
      "loss": 2.0205,
      "step": 226319
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.5597409009933472,
      "learning_rate": 4.647948742245167e-07,
      "loss": 1.9481,
      "step": 226320
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.567319393157959,
      "learning_rate": 4.6456728765915663e-07,
      "loss": 1.895,
      "step": 226321
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.5272332429885864,
      "learning_rate": 4.643397567831164e-07,
      "loss": 1.7612,
      "step": 226322
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.4387868642807007,
      "learning_rate": 4.641122815963627e-07,
      "loss": 1.8917,
      "step": 226323
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.5065336227416992,
      "learning_rate": 4.6388486209902874e-07,
      "loss": 1.9489,
      "step": 226324
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.4842699766159058,
      "learning_rate": 4.636574982910812e-07,
      "loss": 1.8696,
      "step": 226325
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.4609318971633911,
      "learning_rate": 4.634301901726201e-07,
      "loss": 1.9016,
      "step": 226326
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.5638859272003174,
      "learning_rate": 4.6320293774364525e-07,
      "loss": 1.986,
      "step": 226327
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.4028042554855347,
      "learning_rate": 4.629757410042234e-07,
      "loss": 1.9647,
      "step": 226328
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.4832110404968262,
      "learning_rate": 4.6274859995438785e-07,
      "loss": 1.8133,
      "step": 226329
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.443238615989685,
      "learning_rate": 4.625215145941719e-07,
      "loss": 1.9578,
      "step": 226330
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.465510368347168,
      "learning_rate": 4.6229448492364205e-07,
      "loss": 1.9989,
      "step": 226331
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.5305778980255127,
      "learning_rate": 4.6206751094283177e-07,
      "loss": 1.8664,
      "step": 226332
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.4045135974884033,
      "learning_rate": 4.618405926517743e-07,
      "loss": 1.7994,
      "step": 226333
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.563502550125122,
      "learning_rate": 4.6161373005046966e-07,
      "loss": 1.9031,
      "step": 226334
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.543272614479065,
      "learning_rate": 4.6138692313905103e-07,
      "loss": 1.9581,
      "step": 226335
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.455138087272644,
      "learning_rate": 4.6116017191751844e-07,
      "loss": 1.9652,
      "step": 226336
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.507697343826294,
      "learning_rate": 4.609334763859052e-07,
      "loss": 1.9305,
      "step": 226337
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.656697392463684,
      "learning_rate": 4.6070683654424457e-07,
      "loss": 1.9394,
      "step": 226338
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.6289607286453247,
      "learning_rate": 4.6048025239260323e-07,
      "loss": 1.9331,
      "step": 226339
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.5721467733383179,
      "learning_rate": 4.602537239309811e-07,
      "loss": 1.8459,
      "step": 226340
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.556822419166565,
      "learning_rate": 4.6002725115947826e-07,
      "loss": 1.9467,
      "step": 226341
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.476982831954956,
      "learning_rate": 4.5980083407812787e-07,
      "loss": 2.1362,
      "step": 226342
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.5061700344085693,
      "learning_rate": 4.595744726869299e-07,
      "loss": 1.9832,
      "step": 226343
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.4536570310592651,
      "learning_rate": 4.5934816698595113e-07,
      "loss": 2.0064,
      "step": 226344
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.3588024377822876,
      "learning_rate": 4.59121916975258e-07,
      "loss": 1.8786,
      "step": 226345
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.4910224676132202,
      "learning_rate": 4.5889572265485065e-07,
      "loss": 2.002,
      "step": 226346
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.4378647804260254,
      "learning_rate": 4.586695840247956e-07,
      "loss": 2.1614,
      "step": 226347
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.5509129762649536,
      "learning_rate": 4.584435010851262e-07,
      "loss": 1.8885,
      "step": 226348
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.4576112031936646,
      "learning_rate": 4.582174738358757e-07,
      "loss": 2.0239,
      "step": 226349
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.4264371395111084,
      "learning_rate": 4.579915022771108e-07,
      "loss": 2.0008,
      "step": 226350
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.4639798402786255,
      "learning_rate": 4.5776558640886474e-07,
      "loss": 2.1016,
      "step": 226351
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.3740472793579102,
      "learning_rate": 4.575397262311376e-07,
      "loss": 1.8912,
      "step": 226352
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.422588586807251,
      "learning_rate": 4.5731392174406243e-07,
      "loss": 1.8883,
      "step": 226353
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.3906357288360596,
      "learning_rate": 4.570881729475728e-07,
      "loss": 1.8289,
      "step": 226354
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.4160724878311157,
      "learning_rate": 4.568624798418019e-07,
      "loss": 2.0461,
      "step": 226355
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.560412883758545,
      "learning_rate": 4.566368424267497e-07,
      "loss": 1.9242,
      "step": 226356
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.5412259101867676,
      "learning_rate": 4.5641126070244947e-07,
      "loss": 1.6039,
      "step": 226357
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.4285480976104736,
      "learning_rate": 4.5618573466896793e-07,
      "loss": 1.8124,
      "step": 226358
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.469489574432373,
      "learning_rate": 4.5596026432633826e-07,
      "loss": 1.9894,
      "step": 226359
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.50011146068573,
      "learning_rate": 4.5573484967456055e-07,
      "loss": 2.0874,
      "step": 226360
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.500053882598877,
      "learning_rate": 4.5550949071376797e-07,
      "loss": 2.0852,
      "step": 226361
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.4759316444396973,
      "learning_rate": 4.5528418744392725e-07,
      "loss": 1.9688,
      "step": 226362
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.5062968730926514,
      "learning_rate": 4.55058939865105e-07,
      "loss": 1.9412,
      "step": 226363
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.5075945854187012,
      "learning_rate": 4.548337479773345e-07,
      "loss": 1.9122,
      "step": 226364
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.4118112325668335,
      "learning_rate": 4.5460861178068243e-07,
      "loss": 2.0389,
      "step": 226365
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.5272048711776733,
      "learning_rate": 4.54383531275182e-07,
      "loss": 1.9203,
      "step": 226366
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.658311128616333,
      "learning_rate": 4.5415850646083327e-07,
      "loss": 2.1535,
      "step": 226367
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.5979759693145752,
      "learning_rate": 4.539335373377362e-07,
      "loss": 2.0047,
      "step": 226368
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.4388184547424316,
      "learning_rate": 4.537086239058907e-07,
      "loss": 2.0145,
      "step": 226369
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.6639270782470703,
      "learning_rate": 4.534837661653634e-07,
      "loss": 2.0536,
      "step": 226370
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.6787503957748413,
      "learning_rate": 4.532589641161877e-07,
      "loss": 2.0883,
      "step": 226371
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.6425055265426636,
      "learning_rate": 4.5303421775839674e-07,
      "loss": 2.1614,
      "step": 226372
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.5199739933013916,
      "learning_rate": 4.528095270920573e-07,
      "loss": 1.9618,
      "step": 226373
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.526199221611023,
      "learning_rate": 4.5258489211720263e-07,
      "loss": 1.9751,
      "step": 226374
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.4512050151824951,
      "learning_rate": 4.5236031283383266e-07,
      "loss": 1.9993,
      "step": 226375
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.425954818725586,
      "learning_rate": 4.521357892420474e-07,
      "loss": 1.7878,
      "step": 226376
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.478808879852295,
      "learning_rate": 4.5191132134184683e-07,
      "loss": 1.9259,
      "step": 226377
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.3727922439575195,
      "learning_rate": 4.516869091332975e-07,
      "loss": 2.0084,
      "step": 226378
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.35905921459198,
      "learning_rate": 4.514625526164328e-07,
      "loss": 1.7746,
      "step": 226379
    },
    {
      "epoch": 2.95,
      "grad_norm": 2.170804500579834,
      "learning_rate": 4.512382517913193e-07,
      "loss": 1.9305,
      "step": 226380
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.6048930883407593,
      "learning_rate": 4.510140066579571e-07,
      "loss": 1.9594,
      "step": 226381
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.4424490928649902,
      "learning_rate": 4.5078981721641263e-07,
      "loss": 1.8713,
      "step": 226382
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.4600107669830322,
      "learning_rate": 4.505656834667193e-07,
      "loss": 1.9102,
      "step": 226383
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.527969241142273,
      "learning_rate": 4.503416054089104e-07,
      "loss": 2.0859,
      "step": 226384
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.5220000743865967,
      "learning_rate": 4.5011758304301925e-07,
      "loss": 1.9963,
      "step": 226385
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.40054452419281,
      "learning_rate": 4.498936163691458e-07,
      "loss": 2.0543,
      "step": 226386
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.6590616703033447,
      "learning_rate": 4.4966970538725665e-07,
      "loss": 1.9754,
      "step": 226387
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.6093857288360596,
      "learning_rate": 4.494458500974518e-07,
      "loss": 2.1312,
      "step": 226388
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.4916926622390747,
      "learning_rate": 4.4922205049976455e-07,
      "loss": 2.3022,
      "step": 226389
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.42093825340271,
      "learning_rate": 4.489983065941949e-07,
      "loss": 1.9392,
      "step": 226390
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.4629123210906982,
      "learning_rate": 4.4877461838080943e-07,
      "loss": 1.973,
      "step": 226391
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.537379503250122,
      "learning_rate": 4.4855098585964146e-07,
      "loss": 1.9339,
      "step": 226392
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.359610676765442,
      "learning_rate": 4.483274090307576e-07,
      "loss": 1.7778,
      "step": 226393
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.5034986734390259,
      "learning_rate": 4.4810388789419115e-07,
      "loss": 1.9859,
      "step": 226394
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.5932081937789917,
      "learning_rate": 4.4788042244997546e-07,
      "loss": 2.0175,
      "step": 226395
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.5284018516540527,
      "learning_rate": 4.476570126981438e-07,
      "loss": 2.0872,
      "step": 226396
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.474534273147583,
      "learning_rate": 4.474336586387628e-07,
      "loss": 1.8934,
      "step": 226397
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.4441331624984741,
      "learning_rate": 4.4721036027183244e-07,
      "loss": 1.9191,
      "step": 226398
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.4555720090866089,
      "learning_rate": 4.4698711759745266e-07,
      "loss": 2.0432,
      "step": 226399
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.4078952074050903,
      "learning_rate": 4.467639306156234e-07,
      "loss": 1.9483,
      "step": 226400
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.5141496658325195,
      "learning_rate": 4.465407993263781e-07,
      "loss": 1.8735,
      "step": 226401
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.4622423648834229,
      "learning_rate": 4.4631772372978325e-07,
      "loss": 1.7169,
      "step": 226402
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.4589053392410278,
      "learning_rate": 4.460947038258722e-07,
      "loss": 2.077,
      "step": 226403
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.478092074394226,
      "learning_rate": 4.458717396146783e-07,
      "loss": 2.0377,
      "step": 226404
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.578680157661438,
      "learning_rate": 4.45648831096268e-07,
      "loss": 2.0104,
      "step": 226405
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.6114994287490845,
      "learning_rate": 4.4542597827064154e-07,
      "loss": 1.7623,
      "step": 226406
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.4684957265853882,
      "learning_rate": 4.4520318113789865e-07,
      "loss": 2.022,
      "step": 226407
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.509301781654358,
      "learning_rate": 4.4498043969803944e-07,
      "loss": 1.9666,
      "step": 226408
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.4990283250808716,
      "learning_rate": 4.4475775395109716e-07,
      "loss": 1.9254,
      "step": 226409
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.428023099899292,
      "learning_rate": 4.445351238971384e-07,
      "loss": 2.0431,
      "step": 226410
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.4255560636520386,
      "learning_rate": 4.4431254953619655e-07,
      "loss": 1.8114,
      "step": 226411
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.4115557670593262,
      "learning_rate": 4.4409003086830486e-07,
      "loss": 1.8898,
      "step": 226412
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.4841581583023071,
      "learning_rate": 4.4386756789349665e-07,
      "loss": 2.0154,
      "step": 226413
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.3769735097885132,
      "learning_rate": 4.436451606118718e-07,
      "loss": 1.9632,
      "step": 226414
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.3170140981674194,
      "learning_rate": 4.434228090233971e-07,
      "loss": 2.0572,
      "step": 226415
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.4720630645751953,
      "learning_rate": 4.432005131281391e-07,
      "loss": 1.9834,
      "step": 226416
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.5662068128585815,
      "learning_rate": 4.429782729261644e-07,
      "loss": 1.964,
      "step": 226417
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.458734154701233,
      "learning_rate": 4.42756088417473e-07,
      "loss": 1.951,
      "step": 226418
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.5788946151733398,
      "learning_rate": 4.425339596021316e-07,
      "loss": 1.8658,
      "step": 226419
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.4989278316497803,
      "learning_rate": 4.4231188648020665e-07,
      "loss": 1.8522,
      "step": 226420
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.5822042226791382,
      "learning_rate": 4.42089869051665e-07,
      "loss": 1.8695,
      "step": 226421
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.3293278217315674,
      "learning_rate": 4.4186790731663977e-07,
      "loss": 1.9026,
      "step": 226422
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.4341007471084595,
      "learning_rate": 4.4164600127509776e-07,
      "loss": 1.996,
      "step": 226423
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.3921072483062744,
      "learning_rate": 4.4142415092710547e-07,
      "loss": 2.0124,
      "step": 226424
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.30536949634552,
      "learning_rate": 4.412023562727296e-07,
      "loss": 1.9951,
      "step": 226425
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.498050332069397,
      "learning_rate": 4.409806173119701e-07,
      "loss": 1.9964,
      "step": 226426
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.412550449371338,
      "learning_rate": 4.407589340448936e-07,
      "loss": 2.0785,
      "step": 226427
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.465091586112976,
      "learning_rate": 4.405373064715334e-07,
      "loss": 2.0762,
      "step": 226428
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.4269044399261475,
      "learning_rate": 4.403157345919228e-07,
      "loss": 2.0595,
      "step": 226429
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.5143836736679077,
      "learning_rate": 4.4009421840612844e-07,
      "loss": 1.8124,
      "step": 226430
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.5266629457473755,
      "learning_rate": 4.3987275791418363e-07,
      "loss": 1.9867,
      "step": 226431
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.3899962902069092,
      "learning_rate": 4.3965135311608833e-07,
      "loss": 2.0471,
      "step": 226432
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.5667082071304321,
      "learning_rate": 4.394300040119425e-07,
      "loss": 1.9131,
      "step": 226433
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.4450781345367432,
      "learning_rate": 4.3920871060174613e-07,
      "loss": 1.9409,
      "step": 226434
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.4404163360595703,
      "learning_rate": 4.389874728855658e-07,
      "loss": 1.859,
      "step": 226435
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.4254310131072998,
      "learning_rate": 4.3876629086343483e-07,
      "loss": 1.9082,
      "step": 226436
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.415534257888794,
      "learning_rate": 4.3854516453538655e-07,
      "loss": 1.6796,
      "step": 226437
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.462490200996399,
      "learning_rate": 4.383240939014543e-07,
      "loss": 1.843,
      "step": 226438
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.541800618171692,
      "learning_rate": 4.381030789617046e-07,
      "loss": 1.9951,
      "step": 226439
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.4473358392715454,
      "learning_rate": 4.378821197161708e-07,
      "loss": 2.1538,
      "step": 226440
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.3998641967773438,
      "learning_rate": 4.3766121616488627e-07,
      "loss": 1.8613,
      "step": 226441
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.5784640312194824,
      "learning_rate": 4.374403683078842e-07,
      "loss": 1.9544,
      "step": 226442
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.5130114555358887,
      "learning_rate": 4.372195761452313e-07,
      "loss": 1.9713,
      "step": 226443
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.4591522216796875,
      "learning_rate": 4.3699883967696083e-07,
      "loss": 1.8631,
      "step": 226444
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.5798618793487549,
      "learning_rate": 4.3677815890307277e-07,
      "loss": 2.0596,
      "step": 226445
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.4296711683273315,
      "learning_rate": 4.365575338236671e-07,
      "loss": 2.0537,
      "step": 226446
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.3400039672851562,
      "learning_rate": 4.363369644387771e-07,
      "loss": 1.7248,
      "step": 226447
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.65919029712677,
      "learning_rate": 4.3611645074840274e-07,
      "loss": 1.9991,
      "step": 226448
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.554917573928833,
      "learning_rate": 4.358959927526107e-07,
      "loss": 2.0223,
      "step": 226449
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.4637470245361328,
      "learning_rate": 4.356755904514342e-07,
      "loss": 2.1639,
      "step": 226450
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.4387238025665283,
      "learning_rate": 4.354552438449399e-07,
      "loss": 1.911,
      "step": 226451
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.458302617073059,
      "learning_rate": 4.3523495293312784e-07,
      "loss": 2.1737,
      "step": 226452
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.4434226751327515,
      "learning_rate": 4.3501471771606454e-07,
      "loss": 2.0267,
      "step": 226453
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.4190064668655396,
      "learning_rate": 4.347945381938167e-07,
      "loss": 1.8316,
      "step": 226454
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.5703586339950562,
      "learning_rate": 4.34574414366351e-07,
      "loss": 1.8673,
      "step": 226455
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.5679410696029663,
      "learning_rate": 4.343543462337673e-07,
      "loss": 1.8102,
      "step": 226456
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.346604585647583,
      "learning_rate": 4.34134333796099e-07,
      "loss": 1.9536,
      "step": 226457
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.5043351650238037,
      "learning_rate": 4.3391437705337927e-07,
      "loss": 1.9707,
      "step": 226458
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.3912931680679321,
      "learning_rate": 4.3369447600567485e-07,
      "loss": 2.1658,
      "step": 226459
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.4157557487487793,
      "learning_rate": 4.334746306529524e-07,
      "loss": 1.9302,
      "step": 226460
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.5078574419021606,
      "learning_rate": 4.332548409953451e-07,
      "loss": 1.8194,
      "step": 226461
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.424817442893982,
      "learning_rate": 4.330351070328197e-07,
      "loss": 1.9704,
      "step": 226462
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.4671047925949097,
      "learning_rate": 4.3281542876544284e-07,
      "loss": 1.7995,
      "step": 226463
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.5073927640914917,
      "learning_rate": 4.3259580619328104e-07,
      "loss": 1.7987,
      "step": 226464
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.4572572708129883,
      "learning_rate": 4.323762393163677e-07,
      "loss": 2.0587,
      "step": 226465
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.431392788887024,
      "learning_rate": 4.321567281347027e-07,
      "loss": 1.7333,
      "step": 226466
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.4576417207717896,
      "learning_rate": 4.319372726483528e-07,
      "loss": 1.911,
      "step": 226467
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.4722410440444946,
      "learning_rate": 4.3171787285738444e-07,
      "loss": 2.0381,
      "step": 226468
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.5164092779159546,
      "learning_rate": 4.314985287617978e-07,
      "loss": 2.0698,
      "step": 226469
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.5673214197158813,
      "learning_rate": 4.312792403616594e-07,
      "loss": 2.1625,
      "step": 226470
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.5013865232467651,
      "learning_rate": 4.3106000765700256e-07,
      "loss": 2.027,
      "step": 226471
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.5547027587890625,
      "learning_rate": 4.3084083064786057e-07,
      "loss": 1.92,
      "step": 226472
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.523309350013733,
      "learning_rate": 4.3062170933426676e-07,
      "loss": 2.0719,
      "step": 226473
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.447433590888977,
      "learning_rate": 4.3040264371628774e-07,
      "loss": 2.0672,
      "step": 226474
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.4577442407608032,
      "learning_rate": 4.301836337939568e-07,
      "loss": 1.8849,
      "step": 226475
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.3436024188995361,
      "learning_rate": 4.299646795673073e-07,
      "loss": 1.7168,
      "step": 226476
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.4216939210891724,
      "learning_rate": 4.297457810363725e-07,
      "loss": 2.0479,
      "step": 226477
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.4163622856140137,
      "learning_rate": 4.29526938201219e-07,
      "loss": 1.8487,
      "step": 226478
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.4474788904190063,
      "learning_rate": 4.293081510618468e-07,
      "loss": 2.1088,
      "step": 226479
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.566483736038208,
      "learning_rate": 4.2908941961835586e-07,
      "loss": 2.1603,
      "step": 226480
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.3367292881011963,
      "learning_rate": 4.2887074387071286e-07,
      "loss": 1.738,
      "step": 226481
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.388413667678833,
      "learning_rate": 4.286521238190177e-07,
      "loss": 1.8958,
      "step": 226482
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.3452409505844116,
      "learning_rate": 4.284335594633037e-07,
      "loss": 2.0567,
      "step": 226483
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.608124017715454,
      "learning_rate": 4.282150508035709e-07,
      "loss": 2.0116,
      "step": 226484
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.4125726222991943,
      "learning_rate": 4.2799659783991915e-07,
      "loss": 2.0167,
      "step": 226485
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.5603504180908203,
      "learning_rate": 4.2777820057234844e-07,
      "loss": 1.9681,
      "step": 226486
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.4524099826812744,
      "learning_rate": 4.2755985900089215e-07,
      "loss": 1.947,
      "step": 226487
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.5686076879501343,
      "learning_rate": 4.273415731256169e-07,
      "loss": 2.1029,
      "step": 226488
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.5140464305877686,
      "learning_rate": 4.271233429465559e-07,
      "loss": 2.0455,
      "step": 226489
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.4787201881408691,
      "learning_rate": 4.2690516846374254e-07,
      "loss": 1.8408,
      "step": 226490
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.5299303531646729,
      "learning_rate": 4.266870496772101e-07,
      "loss": 2.125,
      "step": 226491
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.371767282485962,
      "learning_rate": 4.2646898658702523e-07,
      "loss": 2.0991,
      "step": 226492
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.447874665260315,
      "learning_rate": 4.2625097919322113e-07,
      "loss": 1.9449,
      "step": 226493
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.5150880813598633,
      "learning_rate": 4.2603302749583126e-07,
      "loss": 1.9222,
      "step": 226494
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.435262680053711,
      "learning_rate": 4.258151314948888e-07,
      "loss": 1.9727,
      "step": 226495
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.437188982963562,
      "learning_rate": 4.2559729119042705e-07,
      "loss": 1.718,
      "step": 226496
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.4255651235580444,
      "learning_rate": 4.2537950658254603e-07,
      "loss": 1.9821,
      "step": 226497
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.6038354635238647,
      "learning_rate": 4.251617776712124e-07,
      "loss": 2.0432,
      "step": 226498
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.6602048873901367,
      "learning_rate": 4.2494410445649274e-07,
      "loss": 2.1127,
      "step": 226499
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.5827453136444092,
      "learning_rate": 4.2472648693845367e-07,
      "loss": 2.2021,
      "step": 226500
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.427621603012085,
      "learning_rate": 4.245089251170619e-07,
      "loss": 1.8645,
      "step": 226501
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.4845340251922607,
      "learning_rate": 4.2429141899245067e-07,
      "loss": 1.7987,
      "step": 226502
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.441663146018982,
      "learning_rate": 4.2407396856461994e-07,
      "loss": 1.7453,
      "step": 226503
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.362791895866394,
      "learning_rate": 4.238565738336031e-07,
      "loss": 1.8567,
      "step": 226504
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.4420181512832642,
      "learning_rate": 4.2363923479943327e-07,
      "loss": 1.8927,
      "step": 226505
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.4747132062911987,
      "learning_rate": 4.234219514621773e-07,
      "loss": 1.8539,
      "step": 226506
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.5490188598632812,
      "learning_rate": 4.232047238218683e-07,
      "loss": 2.0131,
      "step": 226507
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.5068808794021606,
      "learning_rate": 4.2298755187853973e-07,
      "loss": 2.1228,
      "step": 226508
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.5092685222625732,
      "learning_rate": 4.2277043563222477e-07,
      "loss": 1.7523,
      "step": 226509
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.3528608083724976,
      "learning_rate": 4.225533750829568e-07,
      "loss": 1.8062,
      "step": 226510
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.563344120979309,
      "learning_rate": 4.2233637023080247e-07,
      "loss": 2.1692,
      "step": 226511
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.5150662660598755,
      "learning_rate": 4.22119421075795e-07,
      "loss": 1.9458,
      "step": 226512
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.7032970190048218,
      "learning_rate": 4.219025276179677e-07,
      "loss": 2.1509,
      "step": 226513
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.433811068534851,
      "learning_rate": 4.21685689857354e-07,
      "loss": 1.905,
      "step": 226514
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.3830828666687012,
      "learning_rate": 4.214689077940203e-07,
      "loss": 2.1186,
      "step": 226515
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.4685412645339966,
      "learning_rate": 4.212521814280001e-07,
      "loss": 2.0836,
      "step": 226516
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.4723036289215088,
      "learning_rate": 4.2103551075929334e-07,
      "loss": 1.7653,
      "step": 226517
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.4351047277450562,
      "learning_rate": 4.208188957879999e-07,
      "loss": 1.8769,
      "step": 226518
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.4888370037078857,
      "learning_rate": 4.206023365141198e-07,
      "loss": 1.9162,
      "step": 226519
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.6056793928146362,
      "learning_rate": 4.2038583293768637e-07,
      "loss": 2.2761,
      "step": 226520
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.4237616062164307,
      "learning_rate": 4.2016938505879947e-07,
      "loss": 2.0412,
      "step": 226521
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.5100200176239014,
      "learning_rate": 4.199529928774259e-07,
      "loss": 2.003,
      "step": 226522
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.4656274318695068,
      "learning_rate": 4.1973665639363216e-07,
      "loss": 1.8964,
      "step": 226523
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.6255502700805664,
      "learning_rate": 4.1952037560748496e-07,
      "loss": 2.0047,
      "step": 226524
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.4041476249694824,
      "learning_rate": 4.1930415051898426e-07,
      "loss": 1.85,
      "step": 226525
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.44468092918396,
      "learning_rate": 4.1908798112822995e-07,
      "loss": 2.1343,
      "step": 226526
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.4577677249908447,
      "learning_rate": 4.1887186743518873e-07,
      "loss": 1.864,
      "step": 226527
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.4801832437515259,
      "learning_rate": 4.1865580943992727e-07,
      "loss": 1.9519,
      "step": 226528
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.4112101793289185,
      "learning_rate": 4.1843980714251215e-07,
      "loss": 1.8784,
      "step": 226529
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.421380877494812,
      "learning_rate": 4.182238605429433e-07,
      "loss": 2.0905,
      "step": 226530
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.4628338813781738,
      "learning_rate": 4.180079696413208e-07,
      "loss": 2.095,
      "step": 226531
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.4463286399841309,
      "learning_rate": 4.1779213443761115e-07,
      "loss": 2.1003,
      "step": 226532
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.497552752494812,
      "learning_rate": 4.1757635493191446e-07,
      "loss": 2.0037,
      "step": 226533
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.4725219011306763,
      "learning_rate": 4.1736063112419725e-07,
      "loss": 1.9093,
      "step": 226534
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.38963782787323,
      "learning_rate": 4.171449630145929e-07,
      "loss": 1.8553,
      "step": 226535
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.337186574935913,
      "learning_rate": 4.1692935060306796e-07,
      "loss": 1.8018,
      "step": 226536
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.4169116020202637,
      "learning_rate": 4.167137938897225e-07,
      "loss": 2.0759,
      "step": 226537
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.3965657949447632,
      "learning_rate": 4.164982928745564e-07,
      "loss": 1.9926,
      "step": 226538
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.4128919839859009,
      "learning_rate": 4.16282847557603e-07,
      "loss": 1.9488,
      "step": 226539
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.459431529045105,
      "learning_rate": 4.160674579389289e-07,
      "loss": 1.6536,
      "step": 226540
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.5406919717788696,
      "learning_rate": 4.158521240185675e-07,
      "loss": 1.8081,
      "step": 226541
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.529450535774231,
      "learning_rate": 4.156368457965187e-07,
      "loss": 1.8646,
      "step": 226542
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.7446627616882324,
      "learning_rate": 4.154216232729157e-07,
      "loss": 1.9428,
      "step": 226543
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.5245236158370972,
      "learning_rate": 4.15206456447692e-07,
      "loss": 2.1097,
      "step": 226544
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.4622066020965576,
      "learning_rate": 4.1499134532098075e-07,
      "loss": 2.0844,
      "step": 226545
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.36818265914917,
      "learning_rate": 4.147762898927487e-07,
      "loss": 1.8098,
      "step": 226546
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.3918505907058716,
      "learning_rate": 4.1456129016306236e-07,
      "loss": 1.8235,
      "step": 226547
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.5156548023223877,
      "learning_rate": 4.1434634613195517e-07,
      "loss": 2.0412,
      "step": 226548
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.4674720764160156,
      "learning_rate": 4.1413145779949365e-07,
      "loss": 1.9453,
      "step": 226549
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.4273321628570557,
      "learning_rate": 4.139166251657111e-07,
      "loss": 1.9247,
      "step": 226550
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.5325928926467896,
      "learning_rate": 4.137018482306076e-07,
      "loss": 2.054,
      "step": 226551
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.5916980504989624,
      "learning_rate": 4.13487126994283e-07,
      "loss": 1.9616,
      "step": 226552
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.596916913986206,
      "learning_rate": 4.132724614567373e-07,
      "loss": 1.9824,
      "step": 226553
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.4449050426483154,
      "learning_rate": 4.130578516180039e-07,
      "loss": 1.9521,
      "step": 226554
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.4827592372894287,
      "learning_rate": 4.128432974781493e-07,
      "loss": 2.0295,
      "step": 226555
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.5116974115371704,
      "learning_rate": 4.126287990372068e-07,
      "loss": 1.9268,
      "step": 226556
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.500090479850769,
      "learning_rate": 4.124143562952098e-07,
      "loss": 2.1419,
      "step": 226557
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.4561163187026978,
      "learning_rate": 4.1219996925219156e-07,
      "loss": 2.0236,
      "step": 226558
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.6804661750793457,
      "learning_rate": 4.119856379081854e-07,
      "loss": 1.8213,
      "step": 226559
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.7054108381271362,
      "learning_rate": 4.117713622632912e-07,
      "loss": 1.9987,
      "step": 226560
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.3492511510849,
      "learning_rate": 4.115571423174757e-07,
      "loss": 1.8568,
      "step": 226561
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.548348069190979,
      "learning_rate": 4.1134297807080553e-07,
      "loss": 2.0275,
      "step": 226562
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.4555522203445435,
      "learning_rate": 4.111288695233139e-07,
      "loss": 1.8728,
      "step": 226563
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.6385194063186646,
      "learning_rate": 4.1091481667506754e-07,
      "loss": 2.0732,
      "step": 226564
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.4084104299545288,
      "learning_rate": 4.107008195260997e-07,
      "loss": 1.9344,
      "step": 226565
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.3682197332382202,
      "learning_rate": 4.1048687807641034e-07,
      "loss": 1.847,
      "step": 226566
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.496196985244751,
      "learning_rate": 4.1027299232606616e-07,
      "loss": 1.7351,
      "step": 226567
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.4920183420181274,
      "learning_rate": 4.100591622751004e-07,
      "loss": 1.923,
      "step": 226568
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.362499713897705,
      "learning_rate": 4.09845387923613e-07,
      "loss": 1.966,
      "step": 226569
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.4031903743743896,
      "learning_rate": 4.0963166927153734e-07,
      "loss": 1.9117,
      "step": 226570
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.5120487213134766,
      "learning_rate": 4.094180063189734e-07,
      "loss": 1.9096,
      "step": 226571
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.5408036708831787,
      "learning_rate": 4.092043990659544e-07,
      "loss": 2.0722,
      "step": 226572
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.3611886501312256,
      "learning_rate": 4.0899084751254696e-07,
      "loss": 2.0133,
      "step": 226573
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.4746025800704956,
      "learning_rate": 4.0877735165871784e-07,
      "loss": 1.8483,
      "step": 226574
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.3846263885498047,
      "learning_rate": 4.0856391150460024e-07,
      "loss": 1.8867,
      "step": 226575
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.4457515478134155,
      "learning_rate": 4.083505270501608e-07,
      "loss": 2.0615,
      "step": 226576
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.4433298110961914,
      "learning_rate": 4.081371982954662e-07,
      "loss": 2.0408,
      "step": 226577
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.3972749710083008,
      "learning_rate": 4.0792392524054975e-07,
      "loss": 1.9608,
      "step": 226578
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.4061126708984375,
      "learning_rate": 4.07710707885478e-07,
      "loss": 2.0038,
      "step": 226579
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.509856939315796,
      "learning_rate": 4.07497546230251e-07,
      "loss": 2.2555,
      "step": 226580
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.5120210647583008,
      "learning_rate": 4.0728444027490205e-07,
      "loss": 1.8222,
      "step": 226581
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.4278533458709717,
      "learning_rate": 4.0707139001953106e-07,
      "loss": 1.807,
      "step": 226582
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.3706549406051636,
      "learning_rate": 4.0685839546413804e-07,
      "loss": 1.9899,
      "step": 226583
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.5351133346557617,
      "learning_rate": 4.0664545660875627e-07,
      "loss": 2.066,
      "step": 226584
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.442236304283142,
      "learning_rate": 4.0643257345341905e-07,
      "loss": 1.7785,
      "step": 226585
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.3873153924942017,
      "learning_rate": 4.062197459981931e-07,
      "loss": 1.877,
      "step": 226586
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.4593005180358887,
      "learning_rate": 4.060069742431116e-07,
      "loss": 2.1107,
      "step": 226587
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.43571937084198,
      "learning_rate": 4.057942581882079e-07,
      "loss": 1.8695,
      "step": 226588
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.56697416305542,
      "learning_rate": 4.055815978335153e-07,
      "loss": 1.8988,
      "step": 226589
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.494721531867981,
      "learning_rate": 4.0536899317910044e-07,
      "loss": 1.9845,
      "step": 226590
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.536084771156311,
      "learning_rate": 4.051564442249633e-07,
      "loss": 2.1188,
      "step": 226591
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.4012730121612549,
      "learning_rate": 4.049439509711705e-07,
      "loss": 1.7708,
      "step": 226592
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.5690265893936157,
      "learning_rate": 4.0473151341775536e-07,
      "loss": 1.8298,
      "step": 226593
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.4134294986724854,
      "learning_rate": 4.045191315647511e-07,
      "loss": 1.9244,
      "step": 226594
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.400497555732727,
      "learning_rate": 4.0430680541219117e-07,
      "loss": 2.0489,
      "step": 226595
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.5558223724365234,
      "learning_rate": 4.040945349601421e-07,
      "loss": 1.9264,
      "step": 226596
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.4680755138397217,
      "learning_rate": 4.0388232020860387e-07,
      "loss": 2.0667,
      "step": 226597
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.5148537158966064,
      "learning_rate": 4.0367016115767647e-07,
      "loss": 1.8933,
      "step": 226598
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.5057339668273926,
      "learning_rate": 4.0345805780732656e-07,
      "loss": 2.1046,
      "step": 226599
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.403374433517456,
      "learning_rate": 4.03246010157654e-07,
      "loss": 1.937,
      "step": 226600
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.4710663557052612,
      "learning_rate": 4.030340182086589e-07,
      "loss": 2.0892,
      "step": 226601
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.4153876304626465,
      "learning_rate": 4.0282208196040776e-07,
      "loss": 1.857,
      "step": 226602
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.4526888132095337,
      "learning_rate": 4.026102014129007e-07,
      "loss": 2.1217,
      "step": 226603
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.583315134048462,
      "learning_rate": 4.0239837656623754e-07,
      "loss": 2.0664,
      "step": 226604
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.4632099866867065,
      "learning_rate": 4.021866074204183e-07,
      "loss": 1.9458,
      "step": 226605
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.383136510848999,
      "learning_rate": 4.0197489397547634e-07,
      "loss": 1.9169,
      "step": 226606
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.6314419507980347,
      "learning_rate": 4.0176323623144493e-07,
      "loss": 1.966,
      "step": 226607
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.3891010284423828,
      "learning_rate": 4.0155163418842395e-07,
      "loss": 1.7861,
      "step": 226608
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.472819447517395,
      "learning_rate": 4.0134008784638017e-07,
      "loss": 1.9435,
      "step": 226609
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.6343261003494263,
      "learning_rate": 4.0112859720538017e-07,
      "loss": 1.9322,
      "step": 226610
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.5165479183197021,
      "learning_rate": 4.009171622654905e-07,
      "loss": 2.0775,
      "step": 226611
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.4228696823120117,
      "learning_rate": 4.007057830267113e-07,
      "loss": 2.2519,
      "step": 226612
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.5847853422164917,
      "learning_rate": 4.0049445948907575e-07,
      "loss": 2.0992,
      "step": 226613
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.5931237936019897,
      "learning_rate": 4.0028319165268385e-07,
      "loss": 2.0334,
      "step": 226614
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.6308995485305786,
      "learning_rate": 4.0007197951750223e-07,
      "loss": 2.0774,
      "step": 226615
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.4549797773361206,
      "learning_rate": 3.998608230836309e-07,
      "loss": 1.9649,
      "step": 226616
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.3948768377304077,
      "learning_rate": 3.9964972235103645e-07,
      "loss": 1.9771,
      "step": 226617
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.5841355323791504,
      "learning_rate": 3.9943867731985214e-07,
      "loss": 2.1804,
      "step": 226618
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.5732349157333374,
      "learning_rate": 3.9922768799004465e-07,
      "loss": 2.1014,
      "step": 226619
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.553602695465088,
      "learning_rate": 3.990167543616807e-07,
      "loss": 2.0157,
      "step": 226620
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.5186506509780884,
      "learning_rate": 3.988058764347934e-07,
      "loss": 2.2002,
      "step": 226621
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.4482923746109009,
      "learning_rate": 3.985950542094163e-07,
      "loss": 1.8441,
      "step": 226622
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.9318938255310059,
      "learning_rate": 3.983842876855825e-07,
      "loss": 1.8813,
      "step": 226623
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.6646125316619873,
      "learning_rate": 3.9817357686339204e-07,
      "loss": 1.9828,
      "step": 226624
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.4007631540298462,
      "learning_rate": 3.9796292174277823e-07,
      "loss": 1.9742,
      "step": 226625
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.4965269565582275,
      "learning_rate": 3.9775232232387435e-07,
      "loss": 1.8208,
      "step": 226626
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.4763532876968384,
      "learning_rate": 3.9754177860668035e-07,
      "loss": 1.9897,
      "step": 226627
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.4346542358398438,
      "learning_rate": 3.9733129059122957e-07,
      "loss": 1.9486,
      "step": 226628
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.601786494255066,
      "learning_rate": 3.9712085827758867e-07,
      "loss": 1.8815,
      "step": 226629
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.4543646574020386,
      "learning_rate": 3.9691048166575754e-07,
      "loss": 2.0417,
      "step": 226630
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.4614620208740234,
      "learning_rate": 3.967001607558029e-07,
      "loss": 2.1269,
      "step": 226631
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.4329354763031006,
      "learning_rate": 3.96489895547758e-07,
      "loss": 1.9323,
      "step": 226632
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.3785959482192993,
      "learning_rate": 3.9627968604168945e-07,
      "loss": 1.8047,
      "step": 226633
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.4566093683242798,
      "learning_rate": 3.96069532237564e-07,
      "loss": 1.92,
      "step": 226634
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.4107733964920044,
      "learning_rate": 3.958594341354815e-07,
      "loss": 1.8179,
      "step": 226635
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.4682124853134155,
      "learning_rate": 3.9564939173547525e-07,
      "loss": 2.0409,
      "step": 226636
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.4302781820297241,
      "learning_rate": 3.9543940503754533e-07,
      "loss": 2.0167,
      "step": 226637
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.337116003036499,
      "learning_rate": 3.952294740417916e-07,
      "loss": 1.9145,
      "step": 226638
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.4363818168640137,
      "learning_rate": 3.950195987482141e-07,
      "loss": 2.0872,
      "step": 226639
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.5974127054214478,
      "learning_rate": 3.9480977915684606e-07,
      "loss": 2.0479,
      "step": 226640
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.541945219039917,
      "learning_rate": 3.946000152677209e-07,
      "loss": 2.0359,
      "step": 226641
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.356984257698059,
      "learning_rate": 3.943903070809384e-07,
      "loss": 2.1135,
      "step": 226642
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.4508640766143799,
      "learning_rate": 3.941806545964654e-07,
      "loss": 1.9537,
      "step": 226643
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.4966951608657837,
      "learning_rate": 3.9397105781436844e-07,
      "loss": 2.135,
      "step": 226644
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.5091458559036255,
      "learning_rate": 3.937615167346808e-07,
      "loss": 2.0255,
      "step": 226645
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.4644814729690552,
      "learning_rate": 3.9355203135746917e-07,
      "loss": 2.0587,
      "step": 226646
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.3818527460098267,
      "learning_rate": 3.9334260168273345e-07,
      "loss": 1.7521,
      "step": 226647
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.4010647535324097,
      "learning_rate": 3.9313322771057364e-07,
      "loss": 1.9848,
      "step": 226648
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.4416319131851196,
      "learning_rate": 3.929239094409231e-07,
      "loss": 1.8591,
      "step": 226649
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.4846223592758179,
      "learning_rate": 3.9271464687391507e-07,
      "loss": 2.0454,
      "step": 226650
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.5166863203048706,
      "learning_rate": 3.9250544000958284e-07,
      "loss": 2.0741,
      "step": 226651
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.5000529289245605,
      "learning_rate": 3.9229628884789313e-07,
      "loss": 2.0027,
      "step": 226652
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.429582953453064,
      "learning_rate": 3.9208719338894577e-07,
      "loss": 1.9439,
      "step": 226653
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.7101824283599854,
      "learning_rate": 3.9187815363277417e-07,
      "loss": 1.9535,
      "step": 226654
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.5463277101516724,
      "learning_rate": 3.9166916957941163e-07,
      "loss": 1.9788,
      "step": 226655
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.3857626914978027,
      "learning_rate": 3.914602412288914e-07,
      "loss": 1.8927,
      "step": 226656
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.4235265254974365,
      "learning_rate": 3.912513685812468e-07,
      "loss": 2.0405,
      "step": 226657
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.3827368021011353,
      "learning_rate": 3.910425516365112e-07,
      "loss": 2.0598,
      "step": 226658
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.4316504001617432,
      "learning_rate": 3.908337903947512e-07,
      "loss": 2.1795,
      "step": 226659
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.5059250593185425,
      "learning_rate": 3.9062508485599996e-07,
      "loss": 1.8337,
      "step": 226660
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.5518712997436523,
      "learning_rate": 3.9041643502029095e-07,
      "loss": 1.9219,
      "step": 226661
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.5386714935302734,
      "learning_rate": 3.9020784088762414e-07,
      "loss": 1.8348,
      "step": 226662
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.4513760805130005,
      "learning_rate": 3.899993024580994e-07,
      "loss": 2.0605,
      "step": 226663
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.5195544958114624,
      "learning_rate": 3.897908197317168e-07,
      "loss": 2.0098,
      "step": 226664
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.666243553161621,
      "learning_rate": 3.8958239270854287e-07,
      "loss": 1.8707,
      "step": 226665
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.4685161113739014,
      "learning_rate": 3.8937402138861096e-07,
      "loss": 1.8927,
      "step": 226666
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.385025143623352,
      "learning_rate": 3.8916570577192105e-07,
      "loss": 1.7317,
      "step": 226667
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.4077376127243042,
      "learning_rate": 3.8895744585853984e-07,
      "loss": 1.8904,
      "step": 226668
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.4433398246765137,
      "learning_rate": 3.8874924164853383e-07,
      "loss": 1.8187,
      "step": 226669
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.6306710243225098,
      "learning_rate": 3.8854109314190306e-07,
      "loss": 2.0103,
      "step": 226670
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.6107615232467651,
      "learning_rate": 3.8833300033868084e-07,
      "loss": 2.1818,
      "step": 226671
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.466913104057312,
      "learning_rate": 3.881249632389338e-07,
      "loss": 2.0123,
      "step": 226672
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.4237010478973389,
      "learning_rate": 3.8791698184269524e-07,
      "loss": 2.0663,
      "step": 226673
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.4268304109573364,
      "learning_rate": 3.877090561500318e-07,
      "loss": 1.8132,
      "step": 226674
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.493234634399414,
      "learning_rate": 3.8750118616091006e-07,
      "loss": 1.8842,
      "step": 226675
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.423709511756897,
      "learning_rate": 3.8729337187539677e-07,
      "loss": 2.1207,
      "step": 226676
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.550065517425537,
      "learning_rate": 3.8708561329355846e-07,
      "loss": 1.9089,
      "step": 226677
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.6019619703292847,
      "learning_rate": 3.868779104154285e-07,
      "loss": 2.1743,
      "step": 226678
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.4699569940567017,
      "learning_rate": 3.866702632410068e-07,
      "loss": 2.0358,
      "step": 226679
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.4996740818023682,
      "learning_rate": 3.8646267177036004e-07,
      "loss": 1.8146,
      "step": 226680
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.4222033023834229,
      "learning_rate": 3.8625513600355484e-07,
      "loss": 1.8184,
      "step": 226681
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.6217912435531616,
      "learning_rate": 3.860476559405912e-07,
      "loss": 2.1946,
      "step": 226682
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.474876880645752,
      "learning_rate": 3.8584023158150237e-07,
      "loss": 1.7595,
      "step": 226683
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.4707870483398438,
      "learning_rate": 3.85632862926355e-07,
      "loss": 1.8244,
      "step": 226684
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.6009702682495117,
      "learning_rate": 3.854255499751491e-07,
      "loss": 1.9738,
      "step": 226685
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.4731409549713135,
      "learning_rate": 3.8521829272795123e-07,
      "loss": 1.9181,
      "step": 226686
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.4598655700683594,
      "learning_rate": 3.850110911848281e-07,
      "loss": 2.0593,
      "step": 226687
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.471361517906189,
      "learning_rate": 3.848039453457796e-07,
      "loss": 1.9047,
      "step": 226688
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.635271430015564,
      "learning_rate": 3.8459685521083917e-07,
      "loss": 1.8758,
      "step": 226689
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.4592493772506714,
      "learning_rate": 3.843898207800733e-07,
      "loss": 1.7688,
      "step": 226690
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.4836337566375732,
      "learning_rate": 3.8418284205348206e-07,
      "loss": 1.7235,
      "step": 226691
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.5089248418807983,
      "learning_rate": 3.83975919031132e-07,
      "loss": 1.8975,
      "step": 226692
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.405463695526123,
      "learning_rate": 3.837690517130565e-07,
      "loss": 1.8175,
      "step": 226693
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.6084179878234863,
      "learning_rate": 3.8356224009928883e-07,
      "loss": 1.9624,
      "step": 226694
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.4480254650115967,
      "learning_rate": 3.8335548418989556e-07,
      "loss": 1.9214,
      "step": 226695
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.4012258052825928,
      "learning_rate": 3.8314878398487683e-07,
      "loss": 1.9662,
      "step": 226696
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.4762253761291504,
      "learning_rate": 3.8294213948429907e-07,
      "loss": 2.0412,
      "step": 226697
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.5148168802261353,
      "learning_rate": 3.8273555068819574e-07,
      "loss": 1.9869,
      "step": 226698
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.469129204750061,
      "learning_rate": 3.8252901759656674e-07,
      "loss": 1.8502,
      "step": 226699
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.6396151781082153,
      "learning_rate": 3.823225402094787e-07,
      "loss": 1.9539,
      "step": 226700
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.4913291931152344,
      "learning_rate": 3.821161185269983e-07,
      "loss": 1.8503,
      "step": 226701
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.579376459121704,
      "learning_rate": 3.819097525491255e-07,
      "loss": 2.1258,
      "step": 226702
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.3509635925292969,
      "learning_rate": 3.817034422759269e-07,
      "loss": 2.0004,
      "step": 226703
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.4013975858688354,
      "learning_rate": 3.8149718770740246e-07,
      "loss": 1.7403,
      "step": 226704
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.4838693141937256,
      "learning_rate": 3.8129098884365215e-07,
      "loss": 2.1872,
      "step": 226705
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.3525726795196533,
      "learning_rate": 3.810848456846427e-07,
      "loss": 1.8068,
      "step": 226706
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.4890674352645874,
      "learning_rate": 3.8087875823044066e-07,
      "loss": 2.2129,
      "step": 226707
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.514818549156189,
      "learning_rate": 3.8067272648111267e-07,
      "loss": 1.8842,
      "step": 226708
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.4405171871185303,
      "learning_rate": 3.8046675043665875e-07,
      "loss": 1.8045,
      "step": 226709
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.515162467956543,
      "learning_rate": 3.8026083009711215e-07,
      "loss": 1.7273,
      "step": 226710
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.5376272201538086,
      "learning_rate": 3.800549654625729e-07,
      "loss": 2.0146,
      "step": 226711
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.4485975503921509,
      "learning_rate": 3.798491565330075e-07,
      "loss": 1.9725,
      "step": 226712
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.554565191268921,
      "learning_rate": 3.7964340330848277e-07,
      "loss": 2.1139,
      "step": 226713
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.4475584030151367,
      "learning_rate": 3.794377057890652e-07,
      "loss": 2.1461,
      "step": 226714
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.425799012184143,
      "learning_rate": 3.7923206397475483e-07,
      "loss": 1.8105,
      "step": 226715
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.5638854503631592,
      "learning_rate": 3.7902647786558493e-07,
      "loss": 2.0292,
      "step": 226716
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.4571871757507324,
      "learning_rate": 3.7882094746162215e-07,
      "loss": 1.9729,
      "step": 226717
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.4300981760025024,
      "learning_rate": 3.7861547276286654e-07,
      "loss": 2.0379,
      "step": 226718
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.4610435962677002,
      "learning_rate": 3.7841005376941793e-07,
      "loss": 1.9311,
      "step": 226719
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.4290202856063843,
      "learning_rate": 3.7820469048127634e-07,
      "loss": 2.1394,
      "step": 226720
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.5463989973068237,
      "learning_rate": 3.779993828984751e-07,
      "loss": 2.0964,
      "step": 226721
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.428276777267456,
      "learning_rate": 3.777941310210475e-07,
      "loss": 1.8694,
      "step": 226722
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.5604257583618164,
      "learning_rate": 3.7758893484906014e-07,
      "loss": 2.1902,
      "step": 226723
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.376013994216919,
      "learning_rate": 3.773837943825131e-07,
      "loss": 1.8627,
      "step": 226724
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.3508285284042358,
      "learning_rate": 3.771787096214729e-07,
      "loss": 1.9205,
      "step": 226725
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.4812906980514526,
      "learning_rate": 3.769736805659729e-07,
      "loss": 1.793,
      "step": 226726
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.442122459411621,
      "learning_rate": 3.7676870721607967e-07,
      "loss": 1.9347,
      "step": 226727
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.508895754814148,
      "learning_rate": 3.765637895717599e-07,
      "loss": 1.9314,
      "step": 226728
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.6286954879760742,
      "learning_rate": 3.763589276331136e-07,
      "loss": 2.0011,
      "step": 226729
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.5139250755310059,
      "learning_rate": 3.76154121400174e-07,
      "loss": 1.8869,
      "step": 226730
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.5812264680862427,
      "learning_rate": 3.7594937087294106e-07,
      "loss": 1.956,
      "step": 226731
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.4493416547775269,
      "learning_rate": 3.757446760514482e-07,
      "loss": 2.0092,
      "step": 226732
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.4107457399368286,
      "learning_rate": 3.755400369357953e-07,
      "loss": 2.1301,
      "step": 226733
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.462307333946228,
      "learning_rate": 3.7533545352598226e-07,
      "loss": 1.9569,
      "step": 226734
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.4198213815689087,
      "learning_rate": 3.7513092582204253e-07,
      "loss": 1.6857,
      "step": 226735
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.5461822748184204,
      "learning_rate": 3.7492645382404265e-07,
      "loss": 1.8433,
      "step": 226736
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.4830894470214844,
      "learning_rate": 3.747220375319826e-07,
      "loss": 1.7873,
      "step": 226737
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.3524127006530762,
      "learning_rate": 3.7451767694589576e-07,
      "loss": 1.9648,
      "step": 226738
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.8049677610397339,
      "learning_rate": 3.74313372065882e-07,
      "loss": 2.0257,
      "step": 226739
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.4961192607879639,
      "learning_rate": 3.7410912289190796e-07,
      "loss": 2.1828,
      "step": 226740
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.6071767807006836,
      "learning_rate": 3.739049294240737e-07,
      "loss": 2.1767,
      "step": 226741
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.5049282312393188,
      "learning_rate": 3.737007916623791e-07,
      "loss": 1.9563,
      "step": 226742
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.3836510181427002,
      "learning_rate": 3.734967096068575e-07,
      "loss": 2.31,
      "step": 226743
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.3641436100006104,
      "learning_rate": 3.7329268325754224e-07,
      "loss": 1.9887,
      "step": 226744
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.5446451902389526,
      "learning_rate": 3.730887126144999e-07,
      "loss": 1.998,
      "step": 226745
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.4188352823257446,
      "learning_rate": 3.728847976777638e-07,
      "loss": 2.022,
      "step": 226746
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.490640640258789,
      "learning_rate": 3.726809384473672e-07,
      "loss": 2.0326,
      "step": 226747
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.4672601222991943,
      "learning_rate": 3.724771349233435e-07,
      "loss": 1.9181,
      "step": 226748
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.515802025794983,
      "learning_rate": 3.7227338710572595e-07,
      "loss": 1.9948,
      "step": 226749
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.4985467195510864,
      "learning_rate": 3.720696949945812e-07,
      "loss": 1.9668,
      "step": 226750
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.5272704362869263,
      "learning_rate": 3.7186605858990916e-07,
      "loss": 2.0758,
      "step": 226751
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.4363222122192383,
      "learning_rate": 3.716624778917432e-07,
      "loss": 1.8532,
      "step": 226752
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.7083539962768555,
      "learning_rate": 3.7145895290018323e-07,
      "loss": 2.225,
      "step": 226753
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.42526113986969,
      "learning_rate": 3.71255483615196e-07,
      "loss": 1.9292,
      "step": 226754
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.506925344467163,
      "learning_rate": 3.71052070036848e-07,
      "loss": 2.1238,
      "step": 226755
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.4965121746063232,
      "learning_rate": 3.7084871216517264e-07,
      "loss": 2.0976,
      "step": 226756
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.4901604652404785,
      "learning_rate": 3.706454100002365e-07,
      "loss": 2.2138,
      "step": 226757
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.589719295501709,
      "learning_rate": 3.704421635420396e-07,
      "loss": 1.8944,
      "step": 226758
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.4960126876831055,
      "learning_rate": 3.702389727906152e-07,
      "loss": 1.7636,
      "step": 226759
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.4532853364944458,
      "learning_rate": 3.7003583774602997e-07,
      "loss": 2.3006,
      "step": 226760
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.491133213043213,
      "learning_rate": 3.698327584083172e-07,
      "loss": 1.9555,
      "step": 226761
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.5597895383834839,
      "learning_rate": 3.696297347775101e-07,
      "loss": 1.9316,
      "step": 226762
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.5776013135910034,
      "learning_rate": 3.6942676685364215e-07,
      "loss": 2.0336,
      "step": 226763
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.5507116317749023,
      "learning_rate": 3.692238546367465e-07,
      "loss": 2.0725,
      "step": 226764
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.3435065746307373,
      "learning_rate": 3.690209981268899e-07,
      "loss": 2.2023,
      "step": 226765
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.4601960182189941,
      "learning_rate": 3.688181973240722e-07,
      "loss": 2.0465,
      "step": 226766
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.4659814834594727,
      "learning_rate": 3.686154522283602e-07,
      "loss": 1.9232,
      "step": 226767
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.6416250467300415,
      "learning_rate": 3.684127628397537e-07,
      "loss": 2.1122,
      "step": 226768
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.3409574031829834,
      "learning_rate": 3.682101291583195e-07,
      "loss": 2.154,
      "step": 226769
    },
    {
      "epoch": 2.95,
      "grad_norm": 2.543198347091675,
      "learning_rate": 3.680075511841241e-07,
      "loss": 1.9575,
      "step": 226770
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.5860706567764282,
      "learning_rate": 3.6780502891713416e-07,
      "loss": 1.745,
      "step": 226771
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.3913602828979492,
      "learning_rate": 3.6760256235744966e-07,
      "loss": 2.0741,
      "step": 226772
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.5340896844863892,
      "learning_rate": 3.6740015150507064e-07,
      "loss": 2.0924,
      "step": 226773
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.53336501121521,
      "learning_rate": 3.6719779636006365e-07,
      "loss": 2.1264,
      "step": 226774
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.3660186529159546,
      "learning_rate": 3.669954969224287e-07,
      "loss": 1.8458,
      "step": 226775
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.6221544742584229,
      "learning_rate": 3.667932531922324e-07,
      "loss": 2.0845,
      "step": 226776
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.3974138498306274,
      "learning_rate": 3.665910651695081e-07,
      "loss": 1.9586,
      "step": 226777
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.4443795680999756,
      "learning_rate": 3.6638893285428903e-07,
      "loss": 1.8451,
      "step": 226778
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.4021018743515015,
      "learning_rate": 3.661868562466086e-07,
      "loss": 1.9771,
      "step": 226779
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.546984076499939,
      "learning_rate": 3.659848353465333e-07,
      "loss": 1.9122,
      "step": 226780
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.4935202598571777,
      "learning_rate": 3.6578287015406324e-07,
      "loss": 1.9144,
      "step": 226781
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.6103698015213013,
      "learning_rate": 3.655809606692317e-07,
      "loss": 2.089,
      "step": 226782
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.4915664196014404,
      "learning_rate": 3.6537910689210526e-07,
      "loss": 2.1592,
      "step": 226783
    },
    {
      "epoch": 2.95,
      "grad_norm": 2.6080336570739746,
      "learning_rate": 3.6517730882271725e-07,
      "loss": 2.0087,
      "step": 226784
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.4740171432495117,
      "learning_rate": 3.6497556646110095e-07,
      "loss": 2.0056,
      "step": 226785
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.4926161766052246,
      "learning_rate": 3.647738798072897e-07,
      "loss": 2.0542,
      "step": 226786
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.5527666807174683,
      "learning_rate": 3.6457224886131677e-07,
      "loss": 1.893,
      "step": 226787
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.3848841190338135,
      "learning_rate": 3.643706736232488e-07,
      "loss": 2.0554,
      "step": 226788
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.417634129524231,
      "learning_rate": 3.6416915409308576e-07,
      "loss": 2.1302,
      "step": 226789
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.5126330852508545,
      "learning_rate": 3.6396769027086105e-07,
      "loss": 2.3446,
      "step": 226790
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.4612470865249634,
      "learning_rate": 3.6376628215664116e-07,
      "loss": 1.9602,
      "step": 226791
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.4897927045822144,
      "learning_rate": 3.635649297504595e-07,
      "loss": 2.0148,
      "step": 226792
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.5939631462097168,
      "learning_rate": 3.6336363305234927e-07,
      "loss": 2.0307,
      "step": 226793
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.4615509510040283,
      "learning_rate": 3.6316239206237717e-07,
      "loss": 2.126,
      "step": 226794
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.4418361186981201,
      "learning_rate": 3.629612067805099e-07,
      "loss": 2.1247,
      "step": 226795
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.4000906944274902,
      "learning_rate": 3.62760077206814e-07,
      "loss": 1.6664,
      "step": 226796
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.420313835144043,
      "learning_rate": 3.625590033413561e-07,
      "loss": 1.8496,
      "step": 226797
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.4216653108596802,
      "learning_rate": 3.623579851841363e-07,
      "loss": 1.9045,
      "step": 226798
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.4635260105133057,
      "learning_rate": 3.6215702273522106e-07,
      "loss": 2.1236,
      "step": 226799
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.535951852798462,
      "learning_rate": 3.619561159946438e-07,
      "loss": 2.004,
      "step": 226800
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.5354069471359253,
      "learning_rate": 3.617552649624378e-07,
      "loss": 1.9375,
      "step": 226801
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.5893343687057495,
      "learning_rate": 3.6155446963863634e-07,
      "loss": 1.9223,
      "step": 226802
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.6063960790634155,
      "learning_rate": 3.6135373002327273e-07,
      "loss": 2.1155,
      "step": 226803
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.4850436449050903,
      "learning_rate": 3.611530461163803e-07,
      "loss": 2.197,
      "step": 226804
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.5112513303756714,
      "learning_rate": 3.6095241791802563e-07,
      "loss": 1.6889,
      "step": 226805
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.4590204954147339,
      "learning_rate": 3.607518454282088e-07,
      "loss": 2.0815,
      "step": 226806
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.4009864330291748,
      "learning_rate": 3.605513286469963e-07,
      "loss": 2.002,
      "step": 226807
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.4409130811691284,
      "learning_rate": 3.603508675744216e-07,
      "loss": 2.2283,
      "step": 226808
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.5450711250305176,
      "learning_rate": 3.601504622104845e-07,
      "loss": 1.9724,
      "step": 226809
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.3840956687927246,
      "learning_rate": 3.599501125552851e-07,
      "loss": 1.9435,
      "step": 226810
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.5633549690246582,
      "learning_rate": 3.5974981860878995e-07,
      "loss": 1.8944,
      "step": 226811
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.447371006011963,
      "learning_rate": 3.5954958037109903e-07,
      "loss": 1.9519,
      "step": 226812
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.3602157831192017,
      "learning_rate": 3.593493978422457e-07,
      "loss": 1.7352,
      "step": 226813
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.7054637670516968,
      "learning_rate": 3.591492710221966e-07,
      "loss": 2.0015,
      "step": 226814
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.3981937170028687,
      "learning_rate": 3.589491999110516e-07,
      "loss": 2.0601,
      "step": 226815
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.5704041719436646,
      "learning_rate": 3.5874918450884415e-07,
      "loss": 2.0428,
      "step": 226816
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.4149199724197388,
      "learning_rate": 3.585492248156074e-07,
      "loss": 1.8985,
      "step": 226817
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.4571740627288818,
      "learning_rate": 3.583493208313748e-07,
      "loss": 1.8089,
      "step": 226818
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.520490288734436,
      "learning_rate": 3.581494725561795e-07,
      "loss": 2.0391,
      "step": 226819
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.4978454113006592,
      "learning_rate": 3.57949679990055e-07,
      "loss": 1.832,
      "step": 226820
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.3551362752914429,
      "learning_rate": 3.577499431330344e-07,
      "loss": 1.8954,
      "step": 226821
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.5326056480407715,
      "learning_rate": 3.575502619851511e-07,
      "loss": 1.9263,
      "step": 226822
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.4232662916183472,
      "learning_rate": 3.5735063654647176e-07,
      "loss": 1.9332,
      "step": 226823
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.5086960792541504,
      "learning_rate": 3.5715106681702966e-07,
      "loss": 1.899,
      "step": 226824
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.371684193611145,
      "learning_rate": 3.5695155279685805e-07,
      "loss": 1.9074,
      "step": 226825
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.4873441457748413,
      "learning_rate": 3.56752094485957e-07,
      "loss": 1.8807,
      "step": 226826
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.5669053792953491,
      "learning_rate": 3.5655269188439305e-07,
      "loss": 1.8365,
      "step": 226827
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.4509515762329102,
      "learning_rate": 3.563533449922329e-07,
      "loss": 2.1308,
      "step": 226828
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.3161633014678955,
      "learning_rate": 3.5615405380944317e-07,
      "loss": 1.9083,
      "step": 226829
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.4990445375442505,
      "learning_rate": 3.559548183361238e-07,
      "loss": 1.9697,
      "step": 226830
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.4012497663497925,
      "learning_rate": 3.5575563857227484e-07,
      "loss": 2.0263,
      "step": 226831
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.35099196434021,
      "learning_rate": 3.5555651451796287e-07,
      "loss": 2.0328,
      "step": 226832
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.4900912046432495,
      "learning_rate": 3.553574461731878e-07,
      "loss": 2.025,
      "step": 226833
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.36371910572052,
      "learning_rate": 3.551584335380497e-07,
      "loss": 2.1061,
      "step": 226834
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.5504162311553955,
      "learning_rate": 3.549594766124819e-07,
      "loss": 1.878,
      "step": 226835
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.5717768669128418,
      "learning_rate": 3.547605753966509e-07,
      "loss": 2.2106,
      "step": 226836
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.6078814268112183,
      "learning_rate": 3.545617298904901e-07,
      "loss": 1.9223,
      "step": 226837
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.3946515321731567,
      "learning_rate": 3.5436294009406616e-07,
      "loss": 1.9262,
      "step": 226838
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.614009976387024,
      "learning_rate": 3.541642060074457e-07,
      "loss": 2.3103,
      "step": 226839
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.4727203845977783,
      "learning_rate": 3.539655276306286e-07,
      "loss": 1.9524,
      "step": 226840
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.4860265254974365,
      "learning_rate": 3.537669049636482e-07,
      "loss": 1.9128,
      "step": 226841
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.4938799142837524,
      "learning_rate": 3.5356833800657124e-07,
      "loss": 2.1956,
      "step": 226842
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.3587557077407837,
      "learning_rate": 3.5336982675943094e-07,
      "loss": 1.8734,
      "step": 226843
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.7334496974945068,
      "learning_rate": 3.531713712222606e-07,
      "loss": 2.1944,
      "step": 226844
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.797762155532837,
      "learning_rate": 3.529729713950935e-07,
      "loss": 1.9959,
      "step": 226845
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.3134478330612183,
      "learning_rate": 3.527746272779297e-07,
      "loss": 1.9664,
      "step": 226846
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.5781946182250977,
      "learning_rate": 3.525763388708691e-07,
      "loss": 2.1046,
      "step": 226847
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.477830410003662,
      "learning_rate": 3.523781061739117e-07,
      "loss": 1.9595,
      "step": 226848
    },
    {
      "epoch": 2.95,
      "grad_norm": 2.0469443798065186,
      "learning_rate": 3.521799291871241e-07,
      "loss": 2.0362,
      "step": 226849
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.508182406425476,
      "learning_rate": 3.5198180791050636e-07,
      "loss": 2.0057,
      "step": 226850
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.495489478111267,
      "learning_rate": 3.51783742344125e-07,
      "loss": 1.9343,
      "step": 226851
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.6676502227783203,
      "learning_rate": 3.5158573248798004e-07,
      "loss": 2.1308,
      "step": 226852
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.4177566766738892,
      "learning_rate": 3.5138777834213817e-07,
      "loss": 1.9342,
      "step": 226853
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.789179801940918,
      "learning_rate": 3.511898799066326e-07,
      "loss": 1.8313,
      "step": 226854
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.4012348651885986,
      "learning_rate": 3.509920371814967e-07,
      "loss": 1.971,
      "step": 226855
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.5709011554718018,
      "learning_rate": 3.507942501667638e-07,
      "loss": 2.0474,
      "step": 226856
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.5882829427719116,
      "learning_rate": 3.505965188625004e-07,
      "loss": 2.0543,
      "step": 226857
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.6121102571487427,
      "learning_rate": 3.503988432687066e-07,
      "loss": 1.9344,
      "step": 226858
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.4904530048370361,
      "learning_rate": 3.5020122338541565e-07,
      "loss": 1.858,
      "step": 226859
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.5212523937225342,
      "learning_rate": 3.5000365921269423e-07,
      "loss": 1.9429,
      "step": 226860
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.6231684684753418,
      "learning_rate": 3.498061507505423e-07,
      "loss": 2.1395,
      "step": 226861
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.4490593671798706,
      "learning_rate": 3.4960869799902645e-07,
      "loss": 2.102,
      "step": 226862
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.5379834175109863,
      "learning_rate": 3.4941130095818004e-07,
      "loss": 1.9934,
      "step": 226863
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.4766621589660645,
      "learning_rate": 3.4921395962803634e-07,
      "loss": 1.9361,
      "step": 226864
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.5411776304244995,
      "learning_rate": 3.4901667400862866e-07,
      "loss": 1.9279,
      "step": 226865
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.412225604057312,
      "learning_rate": 3.488194440999903e-07,
      "loss": 1.993,
      "step": 226866
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.4488470554351807,
      "learning_rate": 3.486222699021546e-07,
      "loss": 1.7769,
      "step": 226867
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.4964162111282349,
      "learning_rate": 3.484251514151881e-07,
      "loss": 1.9872,
      "step": 226868
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.4045765399932861,
      "learning_rate": 3.482280886390909e-07,
      "loss": 1.926,
      "step": 226869
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.5099343061447144,
      "learning_rate": 3.4803108157389625e-07,
      "loss": 2.0261,
      "step": 226870
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.4464784860610962,
      "learning_rate": 3.4783413021970405e-07,
      "loss": 1.7885,
      "step": 226871
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.411682367324829,
      "learning_rate": 3.476372345764478e-07,
      "loss": 2.0165,
      "step": 226872
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.6073323488235474,
      "learning_rate": 3.4744039464426056e-07,
      "loss": 2.0688,
      "step": 226873
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.5761185884475708,
      "learning_rate": 3.472436104231424e-07,
      "loss": 2.0098,
      "step": 226874
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.336077094078064,
      "learning_rate": 3.470468819131267e-07,
      "loss": 1.8302,
      "step": 226875
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.4867266416549683,
      "learning_rate": 3.4685020911424667e-07,
      "loss": 1.8728,
      "step": 226876
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.4705098867416382,
      "learning_rate": 3.4665359202653564e-07,
      "loss": 2.2206,
      "step": 226877
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.3907772302627563,
      "learning_rate": 3.4645703065006027e-07,
      "loss": 1.8623,
      "step": 226878
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.3794364929199219,
      "learning_rate": 3.462605249848205e-07,
      "loss": 1.8364,
      "step": 226879
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.486477017402649,
      "learning_rate": 3.4606407503084966e-07,
      "loss": 1.8388,
      "step": 226880
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.4891448020935059,
      "learning_rate": 3.4586768078824766e-07,
      "loss": 1.9078,
      "step": 226881
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.6500651836395264,
      "learning_rate": 3.456713422569479e-07,
      "loss": 2.027,
      "step": 226882
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.4978100061416626,
      "learning_rate": 3.4547505943708363e-07,
      "loss": 1.8632,
      "step": 226883
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.4167295694351196,
      "learning_rate": 3.4527883232865485e-07,
      "loss": 2.115,
      "step": 226884
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.4521807432174683,
      "learning_rate": 3.4508266093166147e-07,
      "loss": 1.7645,
      "step": 226885
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.4346330165863037,
      "learning_rate": 3.448865452462035e-07,
      "loss": 1.8793,
      "step": 226886
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.4225367307662964,
      "learning_rate": 3.446904852722809e-07,
      "loss": 1.8881,
      "step": 226887
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.4365142583847046,
      "learning_rate": 3.4449448100992705e-07,
      "loss": 1.8574,
      "step": 226888
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.5615978240966797,
      "learning_rate": 3.442985324592085e-07,
      "loss": 2.2406,
      "step": 226889
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.3964749574661255,
      "learning_rate": 3.4410263962012516e-07,
      "loss": 2.0654,
      "step": 226890
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.3930219411849976,
      "learning_rate": 3.4390680249274385e-07,
      "loss": 1.955,
      "step": 226891
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.8044512271881104,
      "learning_rate": 3.437110210770644e-07,
      "loss": 1.854,
      "step": 226892
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.593820571899414,
      "learning_rate": 3.4351529537315346e-07,
      "loss": 1.9055,
      "step": 226893
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.4281965494155884,
      "learning_rate": 3.433196253810444e-07,
      "loss": 2.0075,
      "step": 226894
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.4947779178619385,
      "learning_rate": 3.4312401110077046e-07,
      "loss": 1.9412,
      "step": 226895
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.4854846000671387,
      "learning_rate": 3.42928452532365e-07,
      "loss": 1.8279,
      "step": 226896
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.4495853185653687,
      "learning_rate": 3.427329496758613e-07,
      "loss": 1.8844,
      "step": 226897
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.428317666053772,
      "learning_rate": 3.425375025312926e-07,
      "loss": 1.9974,
      "step": 226898
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.4317032098770142,
      "learning_rate": 3.423421110987257e-07,
      "loss": 2.0457,
      "step": 226899
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.6407263278961182,
      "learning_rate": 3.421467753781604e-07,
      "loss": 1.9427,
      "step": 226900
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.4276015758514404,
      "learning_rate": 3.419514953696301e-07,
      "loss": 2.0549,
      "step": 226901
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.6101667881011963,
      "learning_rate": 3.4175627107320137e-07,
      "loss": 1.9648,
      "step": 226902
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.4455174207687378,
      "learning_rate": 3.4156110248890755e-07,
      "loss": 1.9837,
      "step": 226903
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.4458836317062378,
      "learning_rate": 3.4136598961678197e-07,
      "loss": 1.9355,
      "step": 226904
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.478087306022644,
      "learning_rate": 3.411709324568246e-07,
      "loss": 1.9064,
      "step": 226905
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.56392240524292,
      "learning_rate": 3.4097593100910204e-07,
      "loss": 1.986,
      "step": 226906
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.4873757362365723,
      "learning_rate": 3.4078098527368094e-07,
      "loss": 2.1396,
      "step": 226907
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.4330174922943115,
      "learning_rate": 3.405860952505279e-07,
      "loss": 1.9299,
      "step": 226908
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.5478746891021729,
      "learning_rate": 3.40391260939743e-07,
      "loss": 2.0239,
      "step": 226909
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.432969093322754,
      "learning_rate": 3.4019648234132615e-07,
      "loss": 1.7478,
      "step": 226910
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.4367519617080688,
      "learning_rate": 3.400017594553106e-07,
      "loss": 1.9233,
      "step": 226911
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.4119393825531006,
      "learning_rate": 3.398070922817631e-07,
      "loss": 2.0217,
      "step": 226912
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.5360666513442993,
      "learning_rate": 3.396124808206835e-07,
      "loss": 2.198,
      "step": 226913
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.4864873886108398,
      "learning_rate": 3.394179250721385e-07,
      "loss": 2.1104,
      "step": 226914
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.4151426553726196,
      "learning_rate": 3.392234250361614e-07,
      "loss": 1.9621,
      "step": 226915
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.4003900289535522,
      "learning_rate": 3.3902898071275217e-07,
      "loss": 1.7845,
      "step": 226916
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.4246259927749634,
      "learning_rate": 3.3883459210201083e-07,
      "loss": 1.9543,
      "step": 226917
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.395174503326416,
      "learning_rate": 3.386402592039039e-07,
      "loss": 1.9831,
      "step": 226918
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.5057998895645142,
      "learning_rate": 3.3844598201849814e-07,
      "loss": 1.8266,
      "step": 226919
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.395731806755066,
      "learning_rate": 3.382517605458601e-07,
      "loss": 2.1181,
      "step": 226920
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.4337173700332642,
      "learning_rate": 3.3805759478598983e-07,
      "loss": 1.9017,
      "step": 226921
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.3184813261032104,
      "learning_rate": 3.3786348473892054e-07,
      "loss": 1.7678,
      "step": 226922
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.4586540460586548,
      "learning_rate": 3.3766943040471894e-07,
      "loss": 1.882,
      "step": 226923
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.5272536277770996,
      "learning_rate": 3.37475431783385e-07,
      "loss": 2.0749,
      "step": 226924
    },
    {
      "epoch": 2.95,
      "grad_norm": 10.227420806884766,
      "learning_rate": 3.37281488874952e-07,
      "loss": 2.0819,
      "step": 226925
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.4434155225753784,
      "learning_rate": 3.3708760167948654e-07,
      "loss": 1.9416,
      "step": 226926
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.5499125719070435,
      "learning_rate": 3.36893770197022e-07,
      "loss": 1.9277,
      "step": 226927
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.4617626667022705,
      "learning_rate": 3.366999944275917e-07,
      "loss": 1.9096,
      "step": 226928
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.4935024976730347,
      "learning_rate": 3.365062743712288e-07,
      "loss": 2.0272,
      "step": 226929
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.3756605386734009,
      "learning_rate": 3.363126100279667e-07,
      "loss": 1.9203,
      "step": 226930
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.4599123001098633,
      "learning_rate": 3.361190013978055e-07,
      "loss": 1.8173,
      "step": 226931
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.5223819017410278,
      "learning_rate": 3.359254484808449e-07,
      "loss": 1.9082,
      "step": 226932
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.4962822198867798,
      "learning_rate": 3.3573195127708506e-07,
      "loss": 1.8116,
      "step": 226933
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.598282814025879,
      "learning_rate": 3.355385097865926e-07,
      "loss": 1.8649,
      "step": 226934
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.4052399396896362,
      "learning_rate": 3.3534512400936743e-07,
      "loss": 2.0312,
      "step": 226935
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.5035392045974731,
      "learning_rate": 3.351517939454429e-07,
      "loss": 1.9201,
      "step": 226936
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.5257530212402344,
      "learning_rate": 3.3495851959488564e-07,
      "loss": 2.1647,
      "step": 226937
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.5827934741973877,
      "learning_rate": 3.347653009576956e-07,
      "loss": 2.1826,
      "step": 226938
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.5601375102996826,
      "learning_rate": 3.3457213803393943e-07,
      "loss": 2.1414,
      "step": 226939
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.4448480606079102,
      "learning_rate": 3.343790308236505e-07,
      "loss": 1.7393,
      "step": 226940
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.5815821886062622,
      "learning_rate": 3.34185979326862e-07,
      "loss": 1.9311,
      "step": 226941
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.4778366088867188,
      "learning_rate": 3.33992983543574e-07,
      "loss": 2.1238,
      "step": 226942
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.4582525491714478,
      "learning_rate": 3.3380004347388633e-07,
      "loss": 2.1939,
      "step": 226943
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.363070011138916,
      "learning_rate": 3.336071591177991e-07,
      "loss": 1.9908,
      "step": 226944
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.4464372396469116,
      "learning_rate": 3.3341433047534563e-07,
      "loss": 2.1359,
      "step": 226945
    },
    {
      "epoch": 2.95,
      "grad_norm": 1.572874665260315,
      "learning_rate": 3.332215575465591e-07,
      "loss": 1.8698,
      "step": 226946
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.4275422096252441,
      "learning_rate": 3.3302884033147294e-07,
      "loss": 2.0597,
      "step": 226947
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.4236161708831787,
      "learning_rate": 3.3283617883015366e-07,
      "loss": 1.9591,
      "step": 226948
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.2959190607070923,
      "learning_rate": 3.3264357304260135e-07,
      "loss": 2.0826,
      "step": 226949
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.4594188928604126,
      "learning_rate": 3.324510229688826e-07,
      "loss": 1.7806,
      "step": 226950
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.4239659309387207,
      "learning_rate": 3.3225852860903066e-07,
      "loss": 1.9308,
      "step": 226951
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.4627584218978882,
      "learning_rate": 3.3206608996304563e-07,
      "loss": 2.0003,
      "step": 226952
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.460707426071167,
      "learning_rate": 3.31873707030994e-07,
      "loss": 2.0461,
      "step": 226953
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.5007305145263672,
      "learning_rate": 3.316813798129092e-07,
      "loss": 2.1666,
      "step": 226954
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.3890424966812134,
      "learning_rate": 3.3148910830879116e-07,
      "loss": 1.9869,
      "step": 226955
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.4593510627746582,
      "learning_rate": 3.312968925187398e-07,
      "loss": 2.1125,
      "step": 226956
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.4458445310592651,
      "learning_rate": 3.3110473244275514e-07,
      "loss": 2.0731,
      "step": 226957
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.5347979068756104,
      "learning_rate": 3.3091262808083717e-07,
      "loss": 1.9429,
      "step": 226958
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.418292760848999,
      "learning_rate": 3.3072057943308585e-07,
      "loss": 1.9177,
      "step": 226959
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.5253890752792358,
      "learning_rate": 3.305285864995344e-07,
      "loss": 1.8484,
      "step": 226960
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.5468138456344604,
      "learning_rate": 3.303366492801496e-07,
      "loss": 2.1644,
      "step": 226961
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.4855763912200928,
      "learning_rate": 3.301447677750313e-07,
      "loss": 1.8318,
      "step": 226962
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.40133535861969,
      "learning_rate": 3.2995294198421283e-07,
      "loss": 1.9822,
      "step": 226963
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.4616281986236572,
      "learning_rate": 3.2976117190769424e-07,
      "loss": 2.005,
      "step": 226964
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.3940192461013794,
      "learning_rate": 3.295694575455421e-07,
      "loss": 1.9114,
      "step": 226965
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.3985846042633057,
      "learning_rate": 3.293777988977564e-07,
      "loss": 1.7239,
      "step": 226966
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.7683732509613037,
      "learning_rate": 3.291861959644371e-07,
      "loss": 1.8348,
      "step": 226967
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.4819542169570923,
      "learning_rate": 3.2899464874555084e-07,
      "loss": 2.0473,
      "step": 226968
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.5681946277618408,
      "learning_rate": 3.2880315724113095e-07,
      "loss": 2.1736,
      "step": 226969
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.4331659078598022,
      "learning_rate": 3.2861172145127737e-07,
      "loss": 1.9061,
      "step": 226970
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.6216778755187988,
      "learning_rate": 3.284203413760234e-07,
      "loss": 2.0455,
      "step": 226971
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.3470319509506226,
      "learning_rate": 3.2822901701533564e-07,
      "loss": 1.8072,
      "step": 226972
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.4927804470062256,
      "learning_rate": 3.2803774836928085e-07,
      "loss": 1.9555,
      "step": 226973
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.5102245807647705,
      "learning_rate": 3.278465354378923e-07,
      "loss": 1.8263,
      "step": 226974
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.4017289876937866,
      "learning_rate": 3.2765537822123656e-07,
      "loss": 2.0914,
      "step": 226975
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.5373119115829468,
      "learning_rate": 3.2746427671934696e-07,
      "loss": 2.1431,
      "step": 226976
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.3222606182098389,
      "learning_rate": 3.2727323093219015e-07,
      "loss": 2.0693,
      "step": 226977
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.490250587463379,
      "learning_rate": 3.2708224085989944e-07,
      "loss": 2.1033,
      "step": 226978
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.6625124216079712,
      "learning_rate": 3.2689130650240813e-07,
      "loss": 1.923,
      "step": 226979
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.5626848936080933,
      "learning_rate": 3.267004278598495e-07,
      "loss": 1.9598,
      "step": 226980
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.3870866298675537,
      "learning_rate": 3.265096049321903e-07,
      "loss": 2.1004,
      "step": 226981
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.437563419342041,
      "learning_rate": 3.26318837719497e-07,
      "loss": 1.9778,
      "step": 226982
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.382156491279602,
      "learning_rate": 3.2612812622180294e-07,
      "loss": 1.9124,
      "step": 226983
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.441676139831543,
      "learning_rate": 3.2593747043910823e-07,
      "loss": 1.906,
      "step": 226984
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.4192270040512085,
      "learning_rate": 3.257468703715127e-07,
      "loss": 1.9871,
      "step": 226985
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.4519939422607422,
      "learning_rate": 3.255563260190164e-07,
      "loss": 1.9676,
      "step": 226986
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.5305516719818115,
      "learning_rate": 3.2536583738161925e-07,
      "loss": 1.9346,
      "step": 226987
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.4319324493408203,
      "learning_rate": 3.2517540445942126e-07,
      "loss": 1.8591,
      "step": 226988
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.655842900276184,
      "learning_rate": 3.249850272524557e-07,
      "loss": 1.9586,
      "step": 226989
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.4922772645950317,
      "learning_rate": 3.247947057606892e-07,
      "loss": 2.0002,
      "step": 226990
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.5034136772155762,
      "learning_rate": 3.2460443998422183e-07,
      "loss": 2.2282,
      "step": 226991
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.4356027841567993,
      "learning_rate": 3.2441422992305343e-07,
      "loss": 1.6567,
      "step": 226992
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.3815706968307495,
      "learning_rate": 3.242240755772507e-07,
      "loss": 2.1232,
      "step": 226993
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.5973212718963623,
      "learning_rate": 3.2403397694681363e-07,
      "loss": 1.929,
      "step": 226994
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.3839062452316284,
      "learning_rate": 3.2384393403180884e-07,
      "loss": 1.7268,
      "step": 226995
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.5026112794876099,
      "learning_rate": 3.236539468322696e-07,
      "loss": 2.031,
      "step": 226996
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.4917454719543457,
      "learning_rate": 3.23464015348196e-07,
      "loss": 2.0726,
      "step": 226997
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.3745474815368652,
      "learning_rate": 3.2327413957965454e-07,
      "loss": 2.0753,
      "step": 226998
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.567858099937439,
      "learning_rate": 3.230843195266453e-07,
      "loss": 1.9477,
      "step": 226999
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.4575161933898926,
      "learning_rate": 3.228945551892681e-07,
      "loss": 2.1687,
      "step": 227000
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.317115306854248,
      "learning_rate": 3.2270484656752303e-07,
      "loss": 2.0119,
      "step": 227001
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.39323890209198,
      "learning_rate": 3.225151936614434e-07,
      "loss": 2.2311,
      "step": 227002
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.4412546157836914,
      "learning_rate": 3.223255964710625e-07,
      "loss": 2.1434,
      "step": 227003
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.4311012029647827,
      "learning_rate": 3.221360549964136e-07,
      "loss": 1.7869,
      "step": 227004
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.4466357231140137,
      "learning_rate": 3.2194656923753005e-07,
      "loss": 1.9714,
      "step": 227005
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.4616872072219849,
      "learning_rate": 3.2175713919444514e-07,
      "loss": 2.1502,
      "step": 227006
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.6281158924102783,
      "learning_rate": 3.2156776486722545e-07,
      "loss": 2.166,
      "step": 227007
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.4434597492218018,
      "learning_rate": 3.2137844625587104e-07,
      "loss": 2.0252,
      "step": 227008
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.684230923652649,
      "learning_rate": 3.211891833604152e-07,
      "loss": 2.0294,
      "step": 227009
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.4250589609146118,
      "learning_rate": 3.2099997618092454e-07,
      "loss": 1.7797,
      "step": 227010
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.6519901752471924,
      "learning_rate": 3.2081082471743235e-07,
      "loss": 1.9425,
      "step": 227011
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.4292937517166138,
      "learning_rate": 3.2062172896993864e-07,
      "loss": 1.9896,
      "step": 227012
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.4334219694137573,
      "learning_rate": 3.2043268893847675e-07,
      "loss": 1.8451,
      "step": 227013
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.574087142944336,
      "learning_rate": 3.202437046231465e-07,
      "loss": 1.951,
      "step": 227014
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.559340476989746,
      "learning_rate": 3.2005477602391473e-07,
      "loss": 2.0936,
      "step": 227015
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.4982078075408936,
      "learning_rate": 3.198659031408479e-07,
      "loss": 2.1254,
      "step": 227016
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.5930191278457642,
      "learning_rate": 3.1967708597397944e-07,
      "loss": 1.8556,
      "step": 227017
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.6587212085723877,
      "learning_rate": 3.194883245233426e-07,
      "loss": 2.0058,
      "step": 227018
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.4747880697250366,
      "learning_rate": 3.192996187889707e-07,
      "loss": 1.8545,
      "step": 227019
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.4641815423965454,
      "learning_rate": 3.1911096877089704e-07,
      "loss": 1.8461,
      "step": 227020
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.6473572254180908,
      "learning_rate": 3.189223744691216e-07,
      "loss": 1.9912,
      "step": 227021
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.4139487743377686,
      "learning_rate": 3.1873383588374435e-07,
      "loss": 1.9821,
      "step": 227022
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.436918020248413,
      "learning_rate": 3.1854535301476525e-07,
      "loss": 2.1766,
      "step": 227023
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.3420236110687256,
      "learning_rate": 3.183569258622509e-07,
      "loss": 1.9616,
      "step": 227024
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.4354000091552734,
      "learning_rate": 3.1816855442620135e-07,
      "loss": 1.7761,
      "step": 227025
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.4403071403503418,
      "learning_rate": 3.179802387066499e-07,
      "loss": 2.0035,
      "step": 227026
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.4615179300308228,
      "learning_rate": 3.177919787036298e-07,
      "loss": 1.9564,
      "step": 227027
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.6114249229431152,
      "learning_rate": 3.1760377441720774e-07,
      "loss": 1.8213,
      "step": 227028
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.4292691946029663,
      "learning_rate": 3.174156258473837e-07,
      "loss": 2.0307,
      "step": 227029
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.5351635217666626,
      "learning_rate": 3.1722753299422423e-07,
      "loss": 2.026,
      "step": 227030
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.33692467212677,
      "learning_rate": 3.170394958577627e-07,
      "loss": 1.978,
      "step": 227031
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.5370126962661743,
      "learning_rate": 3.1685151443799906e-07,
      "loss": 1.9257,
      "step": 227032
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.6130117177963257,
      "learning_rate": 3.16663588735e-07,
      "loss": 2.0497,
      "step": 227033
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.504897952079773,
      "learning_rate": 3.1647571874876544e-07,
      "loss": 1.8339,
      "step": 227034
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.6015751361846924,
      "learning_rate": 3.1628790447939535e-07,
      "loss": 1.7064,
      "step": 227035
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.3369522094726562,
      "learning_rate": 3.161001459268564e-07,
      "loss": 1.7101,
      "step": 227036
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.4783458709716797,
      "learning_rate": 3.1591244309121523e-07,
      "loss": 2.0258,
      "step": 227037
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.4447256326675415,
      "learning_rate": 3.157247959725051e-07,
      "loss": 1.8096,
      "step": 227038
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.5309747457504272,
      "learning_rate": 3.155372045707594e-07,
      "loss": 1.8088,
      "step": 227039
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.534676432609558,
      "learning_rate": 3.1534966888601133e-07,
      "loss": 1.9936,
      "step": 227040
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.575116753578186,
      "learning_rate": 3.1516218891829427e-07,
      "loss": 2.122,
      "step": 227041
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.4016644954681396,
      "learning_rate": 3.1497476466767477e-07,
      "loss": 1.7477,
      "step": 227042
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.5455411672592163,
      "learning_rate": 3.147873961341196e-07,
      "loss": 1.7393,
      "step": 227043
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.5010080337524414,
      "learning_rate": 3.146000833176954e-07,
      "loss": 2.1585,
      "step": 227044
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.4256203174591064,
      "learning_rate": 3.1441282621846865e-07,
      "loss": 1.761,
      "step": 227045
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.5130276679992676,
      "learning_rate": 3.142256248364394e-07,
      "loss": 1.9625,
      "step": 227046
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.4441802501678467,
      "learning_rate": 3.140384791716744e-07,
      "loss": 1.8634,
      "step": 227047
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.4121874570846558,
      "learning_rate": 3.138513892241734e-07,
      "loss": 1.8688,
      "step": 227048
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.6430450677871704,
      "learning_rate": 3.136643549939699e-07,
      "loss": 1.5864,
      "step": 227049
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.5498982667922974,
      "learning_rate": 3.1347737648113047e-07,
      "loss": 1.9345,
      "step": 227050
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.3529067039489746,
      "learning_rate": 3.132904536856884e-07,
      "loss": 2.1133,
      "step": 227051
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.4491709470748901,
      "learning_rate": 3.1310358660761035e-07,
      "loss": 2.1459,
      "step": 227052
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.4682154655456543,
      "learning_rate": 3.129167752470296e-07,
      "loss": 1.7252,
      "step": 227053
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.4944499731063843,
      "learning_rate": 3.127300196039129e-07,
      "loss": 1.795,
      "step": 227054
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.4322574138641357,
      "learning_rate": 3.125433196783267e-07,
      "loss": 2.0124,
      "step": 227055
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.4995306730270386,
      "learning_rate": 3.123566754703044e-07,
      "loss": 1.9719,
      "step": 227056
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.470008134841919,
      "learning_rate": 3.1217008697984604e-07,
      "loss": 2.2785,
      "step": 227057
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.4272297620773315,
      "learning_rate": 3.1198355420701813e-07,
      "loss": 2.015,
      "step": 227058
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.5152727365493774,
      "learning_rate": 3.117970771518874e-07,
      "loss": 2.0203,
      "step": 227059
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.4121241569519043,
      "learning_rate": 3.1161065581438715e-07,
      "loss": 2.0242,
      "step": 227060
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.5415065288543701,
      "learning_rate": 3.11424290194684e-07,
      "loss": 1.8928,
      "step": 227061
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.5257114171981812,
      "learning_rate": 3.112379802927112e-07,
      "loss": 1.8698,
      "step": 227062
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.382454752922058,
      "learning_rate": 3.1105172610853545e-07,
      "loss": 1.9585,
      "step": 227063
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.5210185050964355,
      "learning_rate": 3.1086552764215675e-07,
      "loss": 2.0845,
      "step": 227064
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.5377306938171387,
      "learning_rate": 3.1067938489370834e-07,
      "loss": 1.9188,
      "step": 227065
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.4976519346237183,
      "learning_rate": 3.1049329786312363e-07,
      "loss": 2.0616,
      "step": 227066
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.5448148250579834,
      "learning_rate": 3.103072665504691e-07,
      "loss": 1.9107,
      "step": 227067
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.3924179077148438,
      "learning_rate": 3.101212909558115e-07,
      "loss": 1.8693,
      "step": 227068
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.5264090299606323,
      "learning_rate": 3.099353710791508e-07,
      "loss": 1.758,
      "step": 227069
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.4573029279708862,
      "learning_rate": 3.0974950692052024e-07,
      "loss": 2.1268,
      "step": 227070
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.4991453886032104,
      "learning_rate": 3.095636984799865e-07,
      "loss": 1.7106,
      "step": 227071
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.4562022686004639,
      "learning_rate": 3.0937794575754957e-07,
      "loss": 1.844,
      "step": 227072
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.5294253826141357,
      "learning_rate": 3.0919224875324277e-07,
      "loss": 2.0801,
      "step": 227073
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.3384956121444702,
      "learning_rate": 3.090066074671327e-07,
      "loss": 1.9986,
      "step": 227074
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.6882131099700928,
      "learning_rate": 3.088210218992193e-07,
      "loss": 1.842,
      "step": 227075
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.435545802116394,
      "learning_rate": 3.0863549204956926e-07,
      "loss": 2.0598,
      "step": 227076
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.5140920877456665,
      "learning_rate": 3.084500179182159e-07,
      "loss": 2.025,
      "step": 227077
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.3907374143600464,
      "learning_rate": 3.082645995051258e-07,
      "loss": 1.857,
      "step": 227078
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.5448763370513916,
      "learning_rate": 3.0807923681043236e-07,
      "loss": 1.8685,
      "step": 227079
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.307328224182129,
      "learning_rate": 3.078939298341021e-07,
      "loss": 1.8664,
      "step": 227080
    },
    {
      "epoch": 2.96,
      "grad_norm": 2.467310905456543,
      "learning_rate": 3.0770867857620175e-07,
      "loss": 1.8668,
      "step": 227081
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.4294226169586182,
      "learning_rate": 3.0752348303676454e-07,
      "loss": 1.8838,
      "step": 227082
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.4359755516052246,
      "learning_rate": 3.0733834321579056e-07,
      "loss": 1.7766,
      "step": 227083
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.4957424402236938,
      "learning_rate": 3.071532591133796e-07,
      "loss": 1.9189,
      "step": 227084
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.3582978248596191,
      "learning_rate": 3.069682307294985e-07,
      "loss": 1.74,
      "step": 227085
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.452498197555542,
      "learning_rate": 3.067832580641805e-07,
      "loss": 2.258,
      "step": 227086
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.5547490119934082,
      "learning_rate": 3.065983411175255e-07,
      "loss": 2.235,
      "step": 227087
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.4043655395507812,
      "learning_rate": 3.064134798895335e-07,
      "loss": 1.8026,
      "step": 227088
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.4188579320907593,
      "learning_rate": 3.062286743802378e-07,
      "loss": 1.9249,
      "step": 227089
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.5214779376983643,
      "learning_rate": 3.0604392458963843e-07,
      "loss": 1.9088,
      "step": 227090
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.482061505317688,
      "learning_rate": 3.0585923051783537e-07,
      "loss": 1.9297,
      "step": 227091
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.4620717763900757,
      "learning_rate": 3.056745921648285e-07,
      "loss": 1.9306,
      "step": 227092
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.485335350036621,
      "learning_rate": 3.0549000953065116e-07,
      "loss": 2.1411,
      "step": 227093
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.5089517831802368,
      "learning_rate": 3.053054826153367e-07,
      "loss": 2.0397,
      "step": 227094
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.450871229171753,
      "learning_rate": 3.0512101141891844e-07,
      "loss": 2.0828,
      "step": 227095
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.679064154624939,
      "learning_rate": 3.049365959414296e-07,
      "loss": 2.0564,
      "step": 227096
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.427019715309143,
      "learning_rate": 3.0475223618290357e-07,
      "loss": 1.991,
      "step": 227097
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.4510163068771362,
      "learning_rate": 3.045679321434069e-07,
      "loss": 2.0278,
      "step": 227098
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.461335301399231,
      "learning_rate": 3.0438368382293967e-07,
      "loss": 2.0258,
      "step": 227099
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.440250277519226,
      "learning_rate": 3.041994912215351e-07,
      "loss": 2.0282,
      "step": 227100
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.452386736869812,
      "learning_rate": 3.0401535433925986e-07,
      "loss": 2.0227,
      "step": 227101
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.510246753692627,
      "learning_rate": 3.0383127317611387e-07,
      "loss": 1.8011,
      "step": 227102
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.4800994396209717,
      "learning_rate": 3.0364724773213056e-07,
      "loss": 1.884,
      "step": 227103
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.8191204071044922,
      "learning_rate": 3.0346327800734317e-07,
      "loss": 2.0648,
      "step": 227104
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.4705994129180908,
      "learning_rate": 3.032793640018516e-07,
      "loss": 2.2006,
      "step": 227105
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.562354564666748,
      "learning_rate": 3.0309550571558927e-07,
      "loss": 2.1674,
      "step": 227106
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.416924238204956,
      "learning_rate": 3.0291170314865607e-07,
      "loss": 2.0939,
      "step": 227107
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.5352078676223755,
      "learning_rate": 3.027279563010854e-07,
      "loss": 2.0781,
      "step": 227108
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.5286091566085815,
      "learning_rate": 3.025442651728771e-07,
      "loss": 2.1781,
      "step": 227109
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.4880214929580688,
      "learning_rate": 3.023606297640646e-07,
      "loss": 2.0741,
      "step": 227110
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.418267846107483,
      "learning_rate": 3.0217705007471446e-07,
      "loss": 2.3046,
      "step": 227111
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.4206185340881348,
      "learning_rate": 3.0199352610486003e-07,
      "loss": 1.9816,
      "step": 227112
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.452693223953247,
      "learning_rate": 3.018100578545346e-07,
      "loss": 2.0891,
      "step": 227113
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.3017053604125977,
      "learning_rate": 3.0162664532373815e-07,
      "loss": 1.9512,
      "step": 227114
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.4640846252441406,
      "learning_rate": 3.01443288512504e-07,
      "loss": 1.8697,
      "step": 227115
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.5600357055664062,
      "learning_rate": 3.0125998742093206e-07,
      "loss": 1.8972,
      "step": 227116
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.360573649406433,
      "learning_rate": 3.0107674204898903e-07,
      "loss": 1.7571,
      "step": 227117
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.5059099197387695,
      "learning_rate": 3.008935523967415e-07,
      "loss": 1.9288,
      "step": 227118
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.3965660333633423,
      "learning_rate": 3.0071041846422284e-07,
      "loss": 1.8907,
      "step": 227119
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.4570177793502808,
      "learning_rate": 3.00527340251433e-07,
      "loss": 1.9699,
      "step": 227120
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.5768719911575317,
      "learning_rate": 3.003443177584386e-07,
      "loss": 1.8475,
      "step": 227121
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.5603172779083252,
      "learning_rate": 3.0016135098527294e-07,
      "loss": 2.0016,
      "step": 227122
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.4190109968185425,
      "learning_rate": 2.9997843993196935e-07,
      "loss": 2.0203,
      "step": 227123
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.4591785669326782,
      "learning_rate": 2.997955845985611e-07,
      "loss": 1.7994,
      "step": 227124
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.3988693952560425,
      "learning_rate": 2.996127849850816e-07,
      "loss": 1.9059,
      "step": 227125
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.5699892044067383,
      "learning_rate": 2.9943004109156396e-07,
      "loss": 2.0946,
      "step": 227126
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.5091776847839355,
      "learning_rate": 2.992473529180084e-07,
      "loss": 2.0805,
      "step": 227127
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.5888642072677612,
      "learning_rate": 2.9906472046451467e-07,
      "loss": 2.1874,
      "step": 227128
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.6311038732528687,
      "learning_rate": 2.9888214373104955e-07,
      "loss": 2.0292,
      "step": 227129
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.5273438692092896,
      "learning_rate": 2.98699622717713e-07,
      "loss": 1.9299,
      "step": 227130
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.4323740005493164,
      "learning_rate": 2.9851715742447157e-07,
      "loss": 2.0609,
      "step": 227131
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.3129781484603882,
      "learning_rate": 2.983347478514253e-07,
      "loss": 1.8423,
      "step": 227132
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.591506838798523,
      "learning_rate": 2.9815239399857413e-07,
      "loss": 1.891,
      "step": 227133
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.352716326713562,
      "learning_rate": 2.979700958659181e-07,
      "loss": 1.9953,
      "step": 227134
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.4168665409088135,
      "learning_rate": 2.9778785345355715e-07,
      "loss": 1.9499,
      "step": 227135
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.4723423719406128,
      "learning_rate": 2.9760566676149123e-07,
      "loss": 1.8854,
      "step": 227136
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.4340362548828125,
      "learning_rate": 2.9742353578975365e-07,
      "loss": 2.1015,
      "step": 227137
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.4494695663452148,
      "learning_rate": 2.972414605383777e-07,
      "loss": 1.8669,
      "step": 227138
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.3704674243927002,
      "learning_rate": 2.970594410074301e-07,
      "loss": 1.8493,
      "step": 227139
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.3932433128356934,
      "learning_rate": 2.9687747719687735e-07,
      "loss": 1.9305,
      "step": 227140
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.348246455192566,
      "learning_rate": 2.9669556910681956e-07,
      "loss": 1.9844,
      "step": 227141
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.425995111465454,
      "learning_rate": 2.9651371673725666e-07,
      "loss": 1.9558,
      "step": 227142
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.4498302936553955,
      "learning_rate": 2.963319200882219e-07,
      "loss": 1.9472,
      "step": 227143
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.52176034450531,
      "learning_rate": 2.961501791597487e-07,
      "loss": 1.9644,
      "step": 227144
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.5117684602737427,
      "learning_rate": 2.959684939519036e-07,
      "loss": 2.1105,
      "step": 227145
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.4617033004760742,
      "learning_rate": 2.957868644646866e-07,
      "loss": 1.9124,
      "step": 227146
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.4984323978424072,
      "learning_rate": 2.95605290698131e-07,
      "loss": 2.4508,
      "step": 227147
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.4406721591949463,
      "learning_rate": 2.954237726522701e-07,
      "loss": 2.0635,
      "step": 227148
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.5332292318344116,
      "learning_rate": 2.952423103271706e-07,
      "loss": 1.8813,
      "step": 227149
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.3953595161437988,
      "learning_rate": 2.9506090372283244e-07,
      "loss": 1.772,
      "step": 227150
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.629634141921997,
      "learning_rate": 2.9487955283928886e-07,
      "loss": 1.7149,
      "step": 227151
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.594051480293274,
      "learning_rate": 2.9469825767660657e-07,
      "loss": 2.076,
      "step": 227152
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.733657717704773,
      "learning_rate": 2.9451701823478557e-07,
      "loss": 1.8997,
      "step": 227153
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.5408815145492554,
      "learning_rate": 2.943358345138924e-07,
      "loss": 1.9914,
      "step": 227154
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.3609991073608398,
      "learning_rate": 2.941547065138938e-07,
      "loss": 1.7985,
      "step": 227155
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.592007040977478,
      "learning_rate": 2.9397363423488973e-07,
      "loss": 2.1128,
      "step": 227156
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.4386742115020752,
      "learning_rate": 2.937926176769134e-07,
      "loss": 1.9508,
      "step": 227157
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.5159651041030884,
      "learning_rate": 2.936116568399649e-07,
      "loss": 2.0116,
      "step": 227158
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.4036492109298706,
      "learning_rate": 2.9343075172407746e-07,
      "loss": 1.9952,
      "step": 227159
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.4182851314544678,
      "learning_rate": 2.9324990232931775e-07,
      "loss": 1.9887,
      "step": 227160
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.4317376613616943,
      "learning_rate": 2.930691086556858e-07,
      "loss": 1.9193,
      "step": 227161
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.4756728410720825,
      "learning_rate": 2.928883707032481e-07,
      "loss": 1.8379,
      "step": 227162
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.384426236152649,
      "learning_rate": 2.927076884720048e-07,
      "loss": 1.9912,
      "step": 227163
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.6047011613845825,
      "learning_rate": 2.9252706196202235e-07,
      "loss": 1.8075,
      "step": 227164
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.5940654277801514,
      "learning_rate": 2.923464911733009e-07,
      "loss": 1.9595,
      "step": 227165
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.5211858749389648,
      "learning_rate": 2.92165976105907e-07,
      "loss": 1.9277,
      "step": 227166
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.5084081888198853,
      "learning_rate": 2.919855167598073e-07,
      "loss": 2.0163,
      "step": 227167
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.4878932237625122,
      "learning_rate": 2.918051131351351e-07,
      "loss": 2.1061,
      "step": 227168
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.434593915939331,
      "learning_rate": 2.9162476523185705e-07,
      "loss": 1.9731,
      "step": 227169
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.3785927295684814,
      "learning_rate": 2.9144447305003984e-07,
      "loss": 2.1731,
      "step": 227170
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.4666134119033813,
      "learning_rate": 2.9126423658968336e-07,
      "loss": 1.9484,
      "step": 227171
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.5326340198516846,
      "learning_rate": 2.9108405585082096e-07,
      "loss": 2.0615,
      "step": 227172
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.437523365020752,
      "learning_rate": 2.909039308335193e-07,
      "loss": 1.978,
      "step": 227173
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.3947227001190186,
      "learning_rate": 2.907238615378116e-07,
      "loss": 2.0029,
      "step": 227174
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.5232257843017578,
      "learning_rate": 2.905438479636979e-07,
      "loss": 1.978,
      "step": 227175
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.6211131811141968,
      "learning_rate": 2.9036389011124485e-07,
      "loss": 1.9652,
      "step": 227176
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.6115293502807617,
      "learning_rate": 2.9018398798045237e-07,
      "loss": 1.9629,
      "step": 227177
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.5005241632461548,
      "learning_rate": 2.9000414157138716e-07,
      "loss": 2.1748,
      "step": 227178
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.4410741329193115,
      "learning_rate": 2.8982435088404923e-07,
      "loss": 2.0414,
      "step": 227179
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.485502004623413,
      "learning_rate": 2.8964461591850506e-07,
      "loss": 2.064,
      "step": 227180
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.4467393159866333,
      "learning_rate": 2.894649366747881e-07,
      "loss": 2.0745,
      "step": 227181
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.40873384475708,
      "learning_rate": 2.8928531315289824e-07,
      "loss": 1.8622,
      "step": 227182
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.4220799207687378,
      "learning_rate": 2.891057453528689e-07,
      "loss": 1.8715,
      "step": 227183
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.5358757972717285,
      "learning_rate": 2.889262332747999e-07,
      "loss": 1.9933,
      "step": 227184
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.4529303312301636,
      "learning_rate": 2.88746776918658e-07,
      "loss": 2.0241,
      "step": 227185
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.5368455648422241,
      "learning_rate": 2.8856737628447645e-07,
      "loss": 2.0111,
      "step": 227186
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.5589085817337036,
      "learning_rate": 2.883880313723219e-07,
      "loss": 1.6713,
      "step": 227187
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.6078894138336182,
      "learning_rate": 2.8820874218219435e-07,
      "loss": 1.9846,
      "step": 227188
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.6498769521713257,
      "learning_rate": 2.880295087141604e-07,
      "loss": 2.053,
      "step": 227189
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.4588546752929688,
      "learning_rate": 2.878503309682534e-07,
      "loss": 1.8411,
      "step": 227190
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.4388760328292847,
      "learning_rate": 2.8767120894447326e-07,
      "loss": 1.8438,
      "step": 227191
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.4258307218551636,
      "learning_rate": 2.8749214264288665e-07,
      "loss": 1.9674,
      "step": 227192
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.4687949419021606,
      "learning_rate": 2.873131320634936e-07,
      "loss": 2.0383,
      "step": 227193
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.558013916015625,
      "learning_rate": 2.871341772063607e-07,
      "loss": 1.9431,
      "step": 227194
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.6015112400054932,
      "learning_rate": 2.869552780715212e-07,
      "loss": 1.9341,
      "step": 227195
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.551911473274231,
      "learning_rate": 2.8677643465897513e-07,
      "loss": 1.9179,
      "step": 227196
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.320388674736023,
      "learning_rate": 2.865976469687892e-07,
      "loss": 1.7342,
      "step": 227197
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.4338855743408203,
      "learning_rate": 2.8641891500096326e-07,
      "loss": 1.8584,
      "step": 227198
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.3966913223266602,
      "learning_rate": 2.86240238755564e-07,
      "loss": 1.8421,
      "step": 227199
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.5320985317230225,
      "learning_rate": 2.8606161823259144e-07,
      "loss": 1.8169,
      "step": 227200
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.4702695608139038,
      "learning_rate": 2.8588305343211214e-07,
      "loss": 1.9684,
      "step": 227201
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.5427547693252563,
      "learning_rate": 2.8570454435412615e-07,
      "loss": 1.945,
      "step": 227202
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.3140819072723389,
      "learning_rate": 2.8552609099870007e-07,
      "loss": 1.817,
      "step": 227203
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.4347095489501953,
      "learning_rate": 2.8534769336586714e-07,
      "loss": 1.822,
      "step": 227204
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.3508367538452148,
      "learning_rate": 2.8516935145562747e-07,
      "loss": 2.0307,
      "step": 227205
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.401654601097107,
      "learning_rate": 2.849910652680476e-07,
      "loss": 2.0544,
      "step": 227206
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.5440152883529663,
      "learning_rate": 2.8481283480312755e-07,
      "loss": 2.0564,
      "step": 227207
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.4884307384490967,
      "learning_rate": 2.8463466006093395e-07,
      "loss": 1.9766,
      "step": 227208
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.451402187347412,
      "learning_rate": 2.8445654104146676e-07,
      "loss": 2.0274,
      "step": 227209
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.3837932348251343,
      "learning_rate": 2.842784777447926e-07,
      "loss": 1.9863,
      "step": 227210
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.5463917255401611,
      "learning_rate": 2.841004701709115e-07,
      "loss": 1.7935,
      "step": 227211
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.6619492769241333,
      "learning_rate": 2.839225183198901e-07,
      "loss": 2.0763,
      "step": 227212
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.4494956731796265,
      "learning_rate": 2.8374462219172835e-07,
      "loss": 1.7084,
      "step": 227213
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.4988658428192139,
      "learning_rate": 2.835667817864928e-07,
      "loss": 2.0355,
      "step": 227214
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.4385558366775513,
      "learning_rate": 2.8338899710421694e-07,
      "loss": 2.0735,
      "step": 227215
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.4258310794830322,
      "learning_rate": 2.8321126814486725e-07,
      "loss": 1.8726,
      "step": 227216
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.3938485383987427,
      "learning_rate": 2.830335949085771e-07,
      "loss": 2.0126,
      "step": 227217
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.456367015838623,
      "learning_rate": 2.828559773952799e-07,
      "loss": 1.8325,
      "step": 227218
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.6168010234832764,
      "learning_rate": 2.826784156050754e-07,
      "loss": 1.8749,
      "step": 227219
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.484573245048523,
      "learning_rate": 2.8250090953799706e-07,
      "loss": 1.9664,
      "step": 227220
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.4979939460754395,
      "learning_rate": 2.823234591940449e-07,
      "loss": 1.8154,
      "step": 227221
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.4810912609100342,
      "learning_rate": 2.821460645732854e-07,
      "loss": 2.0089,
      "step": 227222
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.6701465845108032,
      "learning_rate": 2.819687256757186e-07,
      "loss": 2.0554,
      "step": 227223
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.4844127893447876,
      "learning_rate": 2.817914425013779e-07,
      "loss": 1.8367,
      "step": 227224
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.5596429109573364,
      "learning_rate": 2.816142150502965e-07,
      "loss": 2.0806,
      "step": 227225
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.4281857013702393,
      "learning_rate": 2.814370433225743e-07,
      "loss": 1.9772,
      "step": 227226
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.541204571723938,
      "learning_rate": 2.8125992731814485e-07,
      "loss": 2.0401,
      "step": 227227
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.4694124460220337,
      "learning_rate": 2.8108286703710793e-07,
      "loss": 1.8513,
      "step": 227228
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.4852794408798218,
      "learning_rate": 2.8090586247946355e-07,
      "loss": 1.963,
      "step": 227229
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.6669834852218628,
      "learning_rate": 2.807289136452784e-07,
      "loss": 1.9975,
      "step": 227230
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.4455056190490723,
      "learning_rate": 2.805520205345524e-07,
      "loss": 1.6483,
      "step": 227231
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.4588816165924072,
      "learning_rate": 2.803751831473189e-07,
      "loss": 2.0163,
      "step": 227232
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.6036856174468994,
      "learning_rate": 2.801984014836112e-07,
      "loss": 1.9737,
      "step": 227233
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.4130213260650635,
      "learning_rate": 2.8002167554349586e-07,
      "loss": 1.9436,
      "step": 227234
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.5307252407073975,
      "learning_rate": 2.79845005326973e-07,
      "loss": 1.9618,
      "step": 227235
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.487057089805603,
      "learning_rate": 2.796683908340758e-07,
      "loss": 1.9981,
      "step": 227236
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.5325191020965576,
      "learning_rate": 2.794918320648709e-07,
      "loss": 2.3371,
      "step": 227237
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.4382915496826172,
      "learning_rate": 2.793153290193584e-07,
      "loss": 2.0092,
      "step": 227238
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.762560486793518,
      "learning_rate": 2.7913888169757146e-07,
      "loss": 2.0228,
      "step": 227239
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.4585192203521729,
      "learning_rate": 2.789624900995435e-07,
      "loss": 1.8248,
      "step": 227240
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.3063467741012573,
      "learning_rate": 2.787861542253078e-07,
      "loss": 1.9937,
      "step": 227241
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.5604870319366455,
      "learning_rate": 2.7860987407493097e-07,
      "loss": 2.1511,
      "step": 227242
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.4492906332015991,
      "learning_rate": 2.7843364964841297e-07,
      "loss": 2.1273,
      "step": 227243
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.4213056564331055,
      "learning_rate": 2.7825748094575387e-07,
      "loss": 1.9482,
      "step": 227244
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.5613847970962524,
      "learning_rate": 2.780813679670868e-07,
      "loss": 2.1317,
      "step": 227245
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.56093168258667,
      "learning_rate": 2.7790531071234524e-07,
      "loss": 2.0627,
      "step": 227246
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.647338628768921,
      "learning_rate": 2.7772930918159573e-07,
      "loss": 1.9887,
      "step": 227247
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.5175198316574097,
      "learning_rate": 2.7755336337487165e-07,
      "loss": 2.0096,
      "step": 227248
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.5249751806259155,
      "learning_rate": 2.773774732922396e-07,
      "loss": 1.9154,
      "step": 227249
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.5530959367752075,
      "learning_rate": 2.772016389336662e-07,
      "loss": 1.9941,
      "step": 227250
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.5363149642944336,
      "learning_rate": 2.7702586029921815e-07,
      "loss": 1.9354,
      "step": 227251
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.4638779163360596,
      "learning_rate": 2.76850137388962e-07,
      "loss": 2.0666,
      "step": 227252
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.4104177951812744,
      "learning_rate": 2.7667447020286446e-07,
      "loss": 1.824,
      "step": 227253
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.446346402168274,
      "learning_rate": 2.764988587409922e-07,
      "loss": 1.9073,
      "step": 227254
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.4201343059539795,
      "learning_rate": 2.763233030034118e-07,
      "loss": 1.8283,
      "step": 227255
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.432727575302124,
      "learning_rate": 2.761478029900899e-07,
      "loss": 1.9215,
      "step": 227256
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.5062862634658813,
      "learning_rate": 2.759723587010931e-07,
      "loss": 2.0251,
      "step": 227257
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.4892266988754272,
      "learning_rate": 2.757969701364549e-07,
      "loss": 2.0765,
      "step": 227258
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.6172020435333252,
      "learning_rate": 2.7562163729620836e-07,
      "loss": 1.8507,
      "step": 227259
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.4553656578063965,
      "learning_rate": 2.754463601803536e-07,
      "loss": 1.6817,
      "step": 227260
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.5405869483947754,
      "learning_rate": 2.7527113878899055e-07,
      "loss": 2.0308,
      "step": 227261
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.402489423751831,
      "learning_rate": 2.7509597312208585e-07,
      "loss": 1.8583,
      "step": 227262
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.3910002708435059,
      "learning_rate": 2.749208631797062e-07,
      "loss": 2.1028,
      "step": 227263
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.4963045120239258,
      "learning_rate": 2.7474580896188483e-07,
      "loss": 2.0799,
      "step": 227264
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.4468168020248413,
      "learning_rate": 2.7457081046865505e-07,
      "loss": 2.0085,
      "step": 227265
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.4937353134155273,
      "learning_rate": 2.743958677000169e-07,
      "loss": 1.8671,
      "step": 227266
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.539003849029541,
      "learning_rate": 2.742209806560369e-07,
      "loss": 2.0281,
      "step": 227267
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.454067587852478,
      "learning_rate": 2.740461493367152e-07,
      "loss": 1.8681,
      "step": 227268
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.4134948253631592,
      "learning_rate": 2.7387137374211834e-07,
      "loss": 1.9588,
      "step": 227269
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.407082200050354,
      "learning_rate": 2.7369665387227956e-07,
      "loss": 2.1044,
      "step": 227270
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.5039609670639038,
      "learning_rate": 2.7352198972723225e-07,
      "loss": 2.4762,
      "step": 227271
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.552567720413208,
      "learning_rate": 2.733473813069764e-07,
      "loss": 2.2133,
      "step": 227272
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.4128133058547974,
      "learning_rate": 2.7317282861154534e-07,
      "loss": 1.9061,
      "step": 227273
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.5285881757736206,
      "learning_rate": 2.729983316410056e-07,
      "loss": 2.1332,
      "step": 227274
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.393099069595337,
      "learning_rate": 2.7282389039539054e-07,
      "loss": 2.0959,
      "step": 227275
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.401463270187378,
      "learning_rate": 2.7264950487470016e-07,
      "loss": 2.1299,
      "step": 227276
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.4951952695846558,
      "learning_rate": 2.7247517507896775e-07,
      "loss": 1.886,
      "step": 227277
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.45551598072052,
      "learning_rate": 2.7230090100825996e-07,
      "loss": 1.9968,
      "step": 227278
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.5678855180740356,
      "learning_rate": 2.721266826625768e-07,
      "loss": 2.0149,
      "step": 227279
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.4832135438919067,
      "learning_rate": 2.7195252004195146e-07,
      "loss": 2.0636,
      "step": 227280
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.5459930896759033,
      "learning_rate": 2.7177841314645065e-07,
      "loss": 2.2007,
      "step": 227281
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.4275997877120972,
      "learning_rate": 2.7160436197607435e-07,
      "loss": 2.0221,
      "step": 227282
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.3918954133987427,
      "learning_rate": 2.714303665308892e-07,
      "loss": 2.0606,
      "step": 227283
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.312584638595581,
      "learning_rate": 2.712564268108619e-07,
      "loss": 1.9166,
      "step": 227284
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.4343129396438599,
      "learning_rate": 2.7108254281609234e-07,
      "loss": 1.8769,
      "step": 227285
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.4369148015975952,
      "learning_rate": 2.709087145465805e-07,
      "loss": 2.0501,
      "step": 227286
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.462525725364685,
      "learning_rate": 2.707349420023597e-07,
      "loss": 2.2135,
      "step": 227287
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.510335922241211,
      "learning_rate": 2.705612251834632e-07,
      "loss": 1.9087,
      "step": 227288
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.4518764019012451,
      "learning_rate": 2.703875640899578e-07,
      "loss": 1.9646,
      "step": 227289
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.450054407119751,
      "learning_rate": 2.7021395872180997e-07,
      "loss": 1.7817,
      "step": 227290
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.641847014427185,
      "learning_rate": 2.700404090790864e-07,
      "loss": 2.2553,
      "step": 227291
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.4358214139938354,
      "learning_rate": 2.6986691516185376e-07,
      "loss": 1.8556,
      "step": 227292
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.3352950811386108,
      "learning_rate": 2.696934769700787e-07,
      "loss": 1.8393,
      "step": 227293
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.3862524032592773,
      "learning_rate": 2.695200945038278e-07,
      "loss": 1.7469,
      "step": 227294
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.5002442598342896,
      "learning_rate": 2.693467677631345e-07,
      "loss": 2.1782,
      "step": 227295
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.4589271545410156,
      "learning_rate": 2.691734967480319e-07,
      "loss": 1.9443,
      "step": 227296
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.6577353477478027,
      "learning_rate": 2.6900028145855346e-07,
      "loss": 1.9303,
      "step": 227297
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.5381932258605957,
      "learning_rate": 2.6882712189473243e-07,
      "loss": 2.0595,
      "step": 227298
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.4770376682281494,
      "learning_rate": 2.686540180565688e-07,
      "loss": 2.0345,
      "step": 227299
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.7315104007720947,
      "learning_rate": 2.684809699441626e-07,
      "loss": 1.7799,
      "step": 227300
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.6557152271270752,
      "learning_rate": 2.6830797755744707e-07,
      "loss": 2.0943,
      "step": 227301
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.4921085834503174,
      "learning_rate": 2.6813504089655544e-07,
      "loss": 1.8909,
      "step": 227302
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.5109069347381592,
      "learning_rate": 2.679621599614545e-07,
      "loss": 1.894,
      "step": 227303
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.4351547956466675,
      "learning_rate": 2.6778933475221085e-07,
      "loss": 1.9711,
      "step": 227304
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.4358609914779663,
      "learning_rate": 2.676165652688245e-07,
      "loss": 2.11,
      "step": 227305
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.584140658378601,
      "learning_rate": 2.6744385151136195e-07,
      "loss": 2.1143,
      "step": 227306
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.422020673751831,
      "learning_rate": 2.672711934798233e-07,
      "loss": 1.7129,
      "step": 227307
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.4007805585861206,
      "learning_rate": 2.670985911742751e-07,
      "loss": 1.9097,
      "step": 227308
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.4276478290557861,
      "learning_rate": 2.6692604459471747e-07,
      "loss": 1.9281,
      "step": 227309
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.4827609062194824,
      "learning_rate": 2.667535537412169e-07,
      "loss": 2.1802,
      "step": 227310
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.3738635778427124,
      "learning_rate": 2.665811186137734e-07,
      "loss": 1.9746,
      "step": 227311
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.5048526525497437,
      "learning_rate": 2.6640873921242034e-07,
      "loss": 2.0395,
      "step": 227312
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.3994145393371582,
      "learning_rate": 2.66236415537191e-07,
      "loss": 1.8908,
      "step": 227313
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.4792568683624268,
      "learning_rate": 2.66064147588152e-07,
      "loss": 2.0538,
      "step": 227314
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.4329415559768677,
      "learning_rate": 2.6589193536530327e-07,
      "loss": 1.9319,
      "step": 227315
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.443665623664856,
      "learning_rate": 2.6571977886867825e-07,
      "loss": 1.8089,
      "step": 227316
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.4912328720092773,
      "learning_rate": 2.655476780983101e-07,
      "loss": 1.8684,
      "step": 227317
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.3810110092163086,
      "learning_rate": 2.6537563305423224e-07,
      "loss": 2.0116,
      "step": 227318
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.3992866277694702,
      "learning_rate": 2.6520364373647794e-07,
      "loss": 1.9844,
      "step": 227319
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.430920958518982,
      "learning_rate": 2.6503171014511384e-07,
      "loss": 1.8535,
      "step": 227320
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.4706847667694092,
      "learning_rate": 2.648598322801065e-07,
      "loss": 2.1175,
      "step": 227321
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.493621587753296,
      "learning_rate": 2.6468801014152275e-07,
      "loss": 1.9488,
      "step": 227322
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.6037927865982056,
      "learning_rate": 2.645162437293957e-07,
      "loss": 1.9575,
      "step": 227323
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.464391827583313,
      "learning_rate": 2.643445330437588e-07,
      "loss": 1.8818,
      "step": 227324
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.481913447380066,
      "learning_rate": 2.6417287808461195e-07,
      "loss": 1.8487,
      "step": 227325
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.4591339826583862,
      "learning_rate": 2.640012788520218e-07,
      "loss": 1.8933,
      "step": 227326
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.4123815298080444,
      "learning_rate": 2.638297353460217e-07,
      "loss": 1.9999,
      "step": 227327
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.4572482109069824,
      "learning_rate": 2.636582475666449e-07,
      "loss": 2.0438,
      "step": 227328
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.6758464574813843,
      "learning_rate": 2.6348681551389137e-07,
      "loss": 2.2888,
      "step": 227329
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.5228590965270996,
      "learning_rate": 2.6331543918782784e-07,
      "loss": 2.2005,
      "step": 227330
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.5330750942230225,
      "learning_rate": 2.631441185884875e-07,
      "loss": 1.9111,
      "step": 227331
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.5262739658355713,
      "learning_rate": 2.629728537158704e-07,
      "loss": 2.091,
      "step": 227332
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.4047188758850098,
      "learning_rate": 2.6280164457000985e-07,
      "loss": 2.0062,
      "step": 227333
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.4268758296966553,
      "learning_rate": 2.6263049115097245e-07,
      "loss": 1.8285,
      "step": 227334
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.4949851036071777,
      "learning_rate": 2.6245939345875824e-07,
      "loss": 2.0515,
      "step": 227335
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.5248910188674927,
      "learning_rate": 2.6228835149340045e-07,
      "loss": 2.0405,
      "step": 227336
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.4545317888259888,
      "learning_rate": 2.6211736525496575e-07,
      "loss": 1.8212,
      "step": 227337
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.5402082204818726,
      "learning_rate": 2.6194643474345413e-07,
      "loss": 1.9927,
      "step": 227338
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.3709481954574585,
      "learning_rate": 2.617755599588989e-07,
      "loss": 1.8863,
      "step": 227339
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.5587981939315796,
      "learning_rate": 2.616047409013333e-07,
      "loss": 2.216,
      "step": 227340
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.4371700286865234,
      "learning_rate": 2.614339775707908e-07,
      "loss": 2.1252,
      "step": 227341
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.4749757051467896,
      "learning_rate": 2.6126326996730453e-07,
      "loss": 2.1402,
      "step": 227342
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.5291748046875,
      "learning_rate": 2.610926180909412e-07,
      "loss": 2.237,
      "step": 227343
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.482897400856018,
      "learning_rate": 2.6092202194166745e-07,
      "loss": 2.061,
      "step": 227344
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.4767632484436035,
      "learning_rate": 2.607514815195499e-07,
      "loss": 2.1035,
      "step": 227345
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.4095872640609741,
      "learning_rate": 2.6058099682462197e-07,
      "loss": 1.874,
      "step": 227346
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.442031979560852,
      "learning_rate": 2.6041056785691685e-07,
      "loss": 2.145,
      "step": 227347
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.5283313989639282,
      "learning_rate": 2.6024019461643455e-07,
      "loss": 1.8664,
      "step": 227348
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.4973682165145874,
      "learning_rate": 2.6006987710324167e-07,
      "loss": 1.9745,
      "step": 227349
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.6273655891418457,
      "learning_rate": 2.5989961531737157e-07,
      "loss": 2.1344,
      "step": 227350
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.5097121000289917,
      "learning_rate": 2.5972940925885754e-07,
      "loss": 1.8218,
      "step": 227351
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.463358759880066,
      "learning_rate": 2.595592589276996e-07,
      "loss": 2.0767,
      "step": 227352
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.8930920362472534,
      "learning_rate": 2.59389164323931e-07,
      "loss": 1.9484,
      "step": 227353
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.295112133026123,
      "learning_rate": 2.5921912544761834e-07,
      "loss": 1.9112,
      "step": 227354
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.4493768215179443,
      "learning_rate": 2.5904914229876175e-07,
      "loss": 1.7535,
      "step": 227355
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.4854779243469238,
      "learning_rate": 2.5887921487742767e-07,
      "loss": 2.1221,
      "step": 227356
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.374840259552002,
      "learning_rate": 2.5870934318361627e-07,
      "loss": 1.8952,
      "step": 227357
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.391451358795166,
      "learning_rate": 2.585395272173607e-07,
      "loss": 1.8723,
      "step": 227358
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.3789907693862915,
      "learning_rate": 2.5836976697872767e-07,
      "loss": 1.9663,
      "step": 227359
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.5018278360366821,
      "learning_rate": 2.582000624676839e-07,
      "loss": 1.8776,
      "step": 227360
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.5939934253692627,
      "learning_rate": 2.580304136843292e-07,
      "loss": 1.9929,
      "step": 227361
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.429606556892395,
      "learning_rate": 2.5786082062866364e-07,
      "loss": 2.0319,
      "step": 227362
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.5166797637939453,
      "learning_rate": 2.5769128330068724e-07,
      "loss": 1.8677,
      "step": 227363
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.7271407842636108,
      "learning_rate": 2.5752180170049986e-07,
      "loss": 1.8984,
      "step": 227364
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.4580142498016357,
      "learning_rate": 2.5735237582806825e-07,
      "loss": 2.0647,
      "step": 227365
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.4380236864089966,
      "learning_rate": 2.5718300568345895e-07,
      "loss": 2.0682,
      "step": 227366
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.6970998048782349,
      "learning_rate": 2.5701369126670536e-07,
      "loss": 2.0703,
      "step": 227367
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.4250099658966064,
      "learning_rate": 2.568444325778407e-07,
      "loss": 1.8667,
      "step": 227368
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.4459552764892578,
      "learning_rate": 2.5667522961689833e-07,
      "loss": 1.8383,
      "step": 227369
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.5010740756988525,
      "learning_rate": 2.565060823838783e-07,
      "loss": 2.0217,
      "step": 227370
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.5329735279083252,
      "learning_rate": 2.5633699087881375e-07,
      "loss": 1.9659,
      "step": 227371
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.3650622367858887,
      "learning_rate": 2.561679551017715e-07,
      "loss": 1.7393,
      "step": 227372
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.44346022605896,
      "learning_rate": 2.5599897505278466e-07,
      "loss": 1.7476,
      "step": 227373
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.4606789350509644,
      "learning_rate": 2.5583005073182006e-07,
      "loss": 1.9405,
      "step": 227374
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.5384464263916016,
      "learning_rate": 2.5566118213897757e-07,
      "loss": 2.1992,
      "step": 227375
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.4358271360397339,
      "learning_rate": 2.5549236927429053e-07,
      "loss": 1.8124,
      "step": 227376
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.4569380283355713,
      "learning_rate": 2.5532361213772556e-07,
      "loss": 1.9661,
      "step": 227377
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.3979318141937256,
      "learning_rate": 2.5515491072938264e-07,
      "loss": 1.887,
      "step": 227378
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.5483709573745728,
      "learning_rate": 2.549862650492618e-07,
      "loss": 1.7038,
      "step": 227379
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.4464595317840576,
      "learning_rate": 2.548176750973963e-07,
      "loss": 2.0986,
      "step": 227380
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.474003553390503,
      "learning_rate": 2.546491408737861e-07,
      "loss": 1.8852,
      "step": 227381
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.4309483766555786,
      "learning_rate": 2.5448066237853114e-07,
      "loss": 2.0113,
      "step": 227382
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.5119168758392334,
      "learning_rate": 2.543122396116315e-07,
      "loss": 2.0313,
      "step": 227383
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.4976617097854614,
      "learning_rate": 2.5414387257308713e-07,
      "loss": 1.91,
      "step": 227384
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.4799500703811646,
      "learning_rate": 2.539755612629646e-07,
      "loss": 1.8613,
      "step": 227385
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.5009078979492188,
      "learning_rate": 2.538073056812973e-07,
      "loss": 2.0446,
      "step": 227386
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.3244235515594482,
      "learning_rate": 2.536391058280851e-07,
      "loss": 2.1955,
      "step": 227387
    },
    {
      "epoch": 2.96,
      "grad_norm": 2.1251795291900635,
      "learning_rate": 2.5347096170339474e-07,
      "loss": 1.9532,
      "step": 227388
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.4291387796401978,
      "learning_rate": 2.533028733072262e-07,
      "loss": 2.0746,
      "step": 227389
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.4690510034561157,
      "learning_rate": 2.5313484063964606e-07,
      "loss": 1.9091,
      "step": 227390
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.589797854423523,
      "learning_rate": 2.5296686370065435e-07,
      "loss": 2.0248,
      "step": 227391
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.4002764225006104,
      "learning_rate": 2.5279894249031764e-07,
      "loss": 1.9951,
      "step": 227392
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.336707353591919,
      "learning_rate": 2.526310770086026e-07,
      "loss": 1.8215,
      "step": 227393
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.4773671627044678,
      "learning_rate": 2.5246326725560926e-07,
      "loss": 1.9605,
      "step": 227394
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.7331961393356323,
      "learning_rate": 2.5229551323133754e-07,
      "loss": 2.0099,
      "step": 227395
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.5193963050842285,
      "learning_rate": 2.521278149358208e-07,
      "loss": 1.9292,
      "step": 227396
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.4086110591888428,
      "learning_rate": 2.5196017236909225e-07,
      "loss": 1.9611,
      "step": 227397
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.526116132736206,
      "learning_rate": 2.5179258553118533e-07,
      "loss": 1.9003,
      "step": 227398
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.5401437282562256,
      "learning_rate": 2.516250544221332e-07,
      "loss": 2.0575,
      "step": 227399
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.434728980064392,
      "learning_rate": 2.51457579041936e-07,
      "loss": 1.8718,
      "step": 227400
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.4432411193847656,
      "learning_rate": 2.5129015939069354e-07,
      "loss": 1.897,
      "step": 227401
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.3718647956848145,
      "learning_rate": 2.5112279546837255e-07,
      "loss": 1.8577,
      "step": 227402
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.5241798162460327,
      "learning_rate": 2.5095548727500635e-07,
      "loss": 2.1758,
      "step": 227403
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.4263263940811157,
      "learning_rate": 2.5078823481066155e-07,
      "loss": 1.9488,
      "step": 227404
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.3986364603042603,
      "learning_rate": 2.506210380753715e-07,
      "loss": 1.9136,
      "step": 227405
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.4673539400100708,
      "learning_rate": 2.5045389706910277e-07,
      "loss": 2.1689,
      "step": 227406
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.592124581336975,
      "learning_rate": 2.5028681179198875e-07,
      "loss": 1.7973,
      "step": 227407
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.4439884424209595,
      "learning_rate": 2.5011978224396266e-07,
      "loss": 1.8774,
      "step": 227408
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.3880720138549805,
      "learning_rate": 2.4995280842512457e-07,
      "loss": 1.8757,
      "step": 227409
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.3112545013427734,
      "learning_rate": 2.4978589033547437e-07,
      "loss": 1.8929,
      "step": 227410
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.4157460927963257,
      "learning_rate": 2.496190279750454e-07,
      "loss": 1.9194,
      "step": 227411
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.4404211044311523,
      "learning_rate": 2.494522213438377e-07,
      "loss": 2.0203,
      "step": 227412
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.5386309623718262,
      "learning_rate": 2.4928547044195115e-07,
      "loss": 1.9259,
      "step": 227413
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.4724823236465454,
      "learning_rate": 2.4911877526938575e-07,
      "loss": 1.8501,
      "step": 227414
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.466416597366333,
      "learning_rate": 2.489521358261415e-07,
      "loss": 2.0622,
      "step": 227415
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.4184167385101318,
      "learning_rate": 2.48785552112285e-07,
      "loss": 2.0945,
      "step": 227416
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.5652943849563599,
      "learning_rate": 2.4861902412784964e-07,
      "loss": 1.8674,
      "step": 227417
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.4651983976364136,
      "learning_rate": 2.484525518728353e-07,
      "loss": 2.0749,
      "step": 227418
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.5304958820343018,
      "learning_rate": 2.4828613534730867e-07,
      "loss": 2.0713,
      "step": 227419
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.6932088136672974,
      "learning_rate": 2.4811977455126974e-07,
      "loss": 1.8762,
      "step": 227420
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.3688267469406128,
      "learning_rate": 2.479534694847518e-07,
      "loss": 1.9478,
      "step": 227421
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.4387879371643066,
      "learning_rate": 2.477872201478215e-07,
      "loss": 1.9755,
      "step": 227422
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.5670937299728394,
      "learning_rate": 2.4762102654047876e-07,
      "loss": 2.0888,
      "step": 227423
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.3755947351455688,
      "learning_rate": 2.4745488866279026e-07,
      "loss": 2.0012,
      "step": 227424
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.4734947681427002,
      "learning_rate": 2.472888065147227e-07,
      "loss": 1.8353,
      "step": 227425
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.4265296459197998,
      "learning_rate": 2.4712278009634267e-07,
      "loss": 2.022,
      "step": 227426
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.3464572429656982,
      "learning_rate": 2.4695680940768347e-07,
      "loss": 1.947,
      "step": 227427
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.4613003730773926,
      "learning_rate": 2.467908944487784e-07,
      "loss": 2.0188,
      "step": 227428
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.4820876121520996,
      "learning_rate": 2.466250352196608e-07,
      "loss": 2.0824,
      "step": 227429
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.6766340732574463,
      "learning_rate": 2.4645923172033065e-07,
      "loss": 1.7789,
      "step": 227430
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.4238063097000122,
      "learning_rate": 2.4629348395088787e-07,
      "loss": 2.0382,
      "step": 227431
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.5078983306884766,
      "learning_rate": 2.4612779191126587e-07,
      "loss": 1.9803,
      "step": 227432
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.6442447900772095,
      "learning_rate": 2.459621556015645e-07,
      "loss": 1.7069,
      "step": 227433
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.5373883247375488,
      "learning_rate": 2.4579657502181714e-07,
      "loss": 1.9075,
      "step": 227434
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.4874184131622314,
      "learning_rate": 2.4563105017202377e-07,
      "loss": 1.8498,
      "step": 227435
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.4734091758728027,
      "learning_rate": 2.4546558105221767e-07,
      "loss": 2.0962,
      "step": 227436
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.4573562145233154,
      "learning_rate": 2.453001676624322e-07,
      "loss": 2.1226,
      "step": 227437
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.4111610651016235,
      "learning_rate": 2.451348100027006e-07,
      "loss": 1.9529,
      "step": 227438
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.5242063999176025,
      "learning_rate": 2.449695080730896e-07,
      "loss": 2.0094,
      "step": 227439
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.4430831670761108,
      "learning_rate": 2.4480426187356574e-07,
      "loss": 1.9925,
      "step": 227440
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.3971335887908936,
      "learning_rate": 2.446390714041957e-07,
      "loss": 2.0831,
      "step": 227441
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.5229483842849731,
      "learning_rate": 2.444739366650128e-07,
      "loss": 2.0164,
      "step": 227442
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.4950898885726929,
      "learning_rate": 2.4430885765605034e-07,
      "loss": 2.0259,
      "step": 227443
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.556142807006836,
      "learning_rate": 2.4414383437730834e-07,
      "loss": 2.0411,
      "step": 227444
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.5512793064117432,
      "learning_rate": 2.4397886682885335e-07,
      "loss": 2.0514,
      "step": 227445
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.6157411336898804,
      "learning_rate": 2.4381395501068544e-07,
      "loss": 2.0257,
      "step": 227446
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.458594560623169,
      "learning_rate": 2.436490989228712e-07,
      "loss": 2.0117,
      "step": 227447
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.328121542930603,
      "learning_rate": 2.434842985654106e-07,
      "loss": 1.8201,
      "step": 227448
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.4348427057266235,
      "learning_rate": 2.43319553938337e-07,
      "loss": 2.0422,
      "step": 227449
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.5328631401062012,
      "learning_rate": 2.431548650416837e-07,
      "loss": 2.0797,
      "step": 227450
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.3620866537094116,
      "learning_rate": 2.4299023187551726e-07,
      "loss": 1.9801,
      "step": 227451
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.3444644212722778,
      "learning_rate": 2.428256544398044e-07,
      "loss": 1.9333,
      "step": 227452
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.44998037815094,
      "learning_rate": 2.42661132734645e-07,
      "loss": 1.7733,
      "step": 227453
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.492107629776001,
      "learning_rate": 2.424966667600059e-07,
      "loss": 1.9255,
      "step": 227454
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.4212623834609985,
      "learning_rate": 2.4233225651595355e-07,
      "loss": 2.1464,
      "step": 227455
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.5739370584487915,
      "learning_rate": 2.42167902002488e-07,
      "loss": 1.9384,
      "step": 227456
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.4214216470718384,
      "learning_rate": 2.4200360321967595e-07,
      "loss": 1.8854,
      "step": 227457
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.4310978651046753,
      "learning_rate": 2.418393601675506e-07,
      "loss": 1.9895,
      "step": 227458
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.5246119499206543,
      "learning_rate": 2.4167517284611194e-07,
      "loss": 1.9413,
      "step": 227459
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.613643765449524,
      "learning_rate": 2.415110412553933e-07,
      "loss": 2.1201,
      "step": 227460
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.4714099168777466,
      "learning_rate": 2.413469653954614e-07,
      "loss": 2.0426,
      "step": 227461
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.469812273979187,
      "learning_rate": 2.411829452663161e-07,
      "loss": 1.8438,
      "step": 227462
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.5453206300735474,
      "learning_rate": 2.4101898086799074e-07,
      "loss": 2.0804,
      "step": 227463
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.3913826942443848,
      "learning_rate": 2.408550722005187e-07,
      "loss": 1.9616,
      "step": 227464
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.4438748359680176,
      "learning_rate": 2.406912192638999e-07,
      "loss": 2.0461,
      "step": 227465
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.4222544431686401,
      "learning_rate": 2.4052742205823427e-07,
      "loss": 1.9401,
      "step": 227466
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.4616888761520386,
      "learning_rate": 2.403636805834885e-07,
      "loss": 2.0223,
      "step": 227467
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.492728590965271,
      "learning_rate": 2.4019999483972927e-07,
      "loss": 1.975,
      "step": 227468
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.4380543231964111,
      "learning_rate": 2.4003636482698983e-07,
      "loss": 1.9584,
      "step": 227469
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.4339332580566406,
      "learning_rate": 2.398727905452369e-07,
      "loss": 1.9227,
      "step": 227470
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.590816617012024,
      "learning_rate": 2.3970927199460363e-07,
      "loss": 2.0107,
      "step": 227471
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.7022308111190796,
      "learning_rate": 2.395458091750568e-07,
      "loss": 1.7792,
      "step": 227472
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.5628491640090942,
      "learning_rate": 2.393824020866297e-07,
      "loss": 1.9337,
      "step": 227473
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.5229486227035522,
      "learning_rate": 2.3921905072932236e-07,
      "loss": 1.9917,
      "step": 227474
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.4586175680160522,
      "learning_rate": 2.390557551032679e-07,
      "loss": 2.0175,
      "step": 227475
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.6458241939544678,
      "learning_rate": 2.388925152083665e-07,
      "loss": 1.9884,
      "step": 227476
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.4935396909713745,
      "learning_rate": 2.387293310447513e-07,
      "loss": 2.0753,
      "step": 227477
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.3970885276794434,
      "learning_rate": 2.3856620261242243e-07,
      "loss": 2.0159,
      "step": 227478
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.6216816902160645,
      "learning_rate": 2.3840312991137978e-07,
      "loss": 1.9056,
      "step": 227479
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.4712879657745361,
      "learning_rate": 2.3824011294169e-07,
      "loss": 1.9207,
      "step": 227480
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.458767294883728,
      "learning_rate": 2.3807715170338636e-07,
      "loss": 1.8529,
      "step": 227481
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.6494581699371338,
      "learning_rate": 2.379142461964356e-07,
      "loss": 1.9591,
      "step": 227482
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.5140132904052734,
      "learning_rate": 2.3775139642097096e-07,
      "loss": 2.1387,
      "step": 227483
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.5662739276885986,
      "learning_rate": 2.3758860237692578e-07,
      "loss": 1.9602,
      "step": 227484
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.5267537832260132,
      "learning_rate": 2.374258640643667e-07,
      "loss": 2.0287,
      "step": 227485
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.3656928539276123,
      "learning_rate": 2.3726318148336032e-07,
      "loss": 1.8659,
      "step": 227486
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.5037237405776978,
      "learning_rate": 2.3710055463390664e-07,
      "loss": 2.0505,
      "step": 227487
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.5054059028625488,
      "learning_rate": 2.369379835160057e-07,
      "loss": 1.9892,
      "step": 227488
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.449997067451477,
      "learning_rate": 2.3677546812972403e-07,
      "loss": 1.9493,
      "step": 227489
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.3578622341156006,
      "learning_rate": 2.3661300847509501e-07,
      "loss": 1.7715,
      "step": 227490
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.4198508262634277,
      "learning_rate": 2.3645060455211862e-07,
      "loss": 2.0531,
      "step": 227491
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.468027949333191,
      "learning_rate": 2.3628825636086145e-07,
      "loss": 2.0457,
      "step": 227492
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.4235767126083374,
      "learning_rate": 2.3612596390135685e-07,
      "loss": 1.954,
      "step": 227493
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.4744222164154053,
      "learning_rate": 2.3596372717357147e-07,
      "loss": 2.0572,
      "step": 227494
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.3055585622787476,
      "learning_rate": 2.3580154617760526e-07,
      "loss": 1.9143,
      "step": 227495
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.3846958875656128,
      "learning_rate": 2.3563942091345822e-07,
      "loss": 1.9188,
      "step": 227496
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.567142367362976,
      "learning_rate": 2.3547735138116365e-07,
      "loss": 1.7893,
      "step": 227497
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.4541759490966797,
      "learning_rate": 2.3531533758075482e-07,
      "loss": 2.2865,
      "step": 227498
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.47309410572052,
      "learning_rate": 2.3515337951223178e-07,
      "loss": 1.9257,
      "step": 227499
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.4818800687789917,
      "learning_rate": 2.3499147717569443e-07,
      "loss": 2.0491,
      "step": 227500
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.6953519582748413,
      "learning_rate": 2.3482963057110948e-07,
      "loss": 2.0809,
      "step": 227501
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.549218773841858,
      "learning_rate": 2.3466783969854354e-07,
      "loss": 2.0609,
      "step": 227502
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.4922045469284058,
      "learning_rate": 2.3450610455799656e-07,
      "loss": 1.9629,
      "step": 227503
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.5177947282791138,
      "learning_rate": 2.343444251495019e-07,
      "loss": 2.0824,
      "step": 227504
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.5300886631011963,
      "learning_rate": 2.3418280147312618e-07,
      "loss": 2.1732,
      "step": 227505
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.5017555952072144,
      "learning_rate": 2.3402123352886937e-07,
      "loss": 2.0954,
      "step": 227506
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.4117014408111572,
      "learning_rate": 2.3385972131676478e-07,
      "loss": 1.8891,
      "step": 227507
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.618587851524353,
      "learning_rate": 2.3369826483681242e-07,
      "loss": 1.6969,
      "step": 227508
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.546586275100708,
      "learning_rate": 2.335368640891122e-07,
      "loss": 1.8962,
      "step": 227509
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.5540989637374878,
      "learning_rate": 2.3337551907363084e-07,
      "loss": 1.843,
      "step": 227510
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.6502336263656616,
      "learning_rate": 2.3321422979043492e-07,
      "loss": 2.051,
      "step": 227511
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.3629417419433594,
      "learning_rate": 2.3305299623955775e-07,
      "loss": 1.9519,
      "step": 227512
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.3993066549301147,
      "learning_rate": 2.3289181842099935e-07,
      "loss": 1.8283,
      "step": 227513
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.4932427406311035,
      "learning_rate": 2.3273069633482633e-07,
      "loss": 1.8769,
      "step": 227514
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.4896748065948486,
      "learning_rate": 2.3256962998100537e-07,
      "loss": 1.7893,
      "step": 227515
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.4324207305908203,
      "learning_rate": 2.324086193596364e-07,
      "loss": 2.0483,
      "step": 227516
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.47247314453125,
      "learning_rate": 2.322476644707194e-07,
      "loss": 2.0277,
      "step": 227517
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.6663546562194824,
      "learning_rate": 2.320867653142877e-07,
      "loss": 1.9427,
      "step": 227518
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.5641380548477173,
      "learning_rate": 2.3192592189037463e-07,
      "loss": 2.1708,
      "step": 227519
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.5065314769744873,
      "learning_rate": 2.3176513419898013e-07,
      "loss": 2.0258,
      "step": 227520
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.6006594896316528,
      "learning_rate": 2.3160440224017086e-07,
      "loss": 2.0853,
      "step": 227521
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.436134696006775,
      "learning_rate": 2.314437260139801e-07,
      "loss": 2.1082,
      "step": 227522
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.4419970512390137,
      "learning_rate": 2.3128310552040786e-07,
      "loss": 1.9355,
      "step": 227523
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.3019325733184814,
      "learning_rate": 2.3112254075952076e-07,
      "loss": 1.6728,
      "step": 227524
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.4505869150161743,
      "learning_rate": 2.3096203173128548e-07,
      "loss": 2.2057,
      "step": 227525
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.3412977457046509,
      "learning_rate": 2.3080157843580193e-07,
      "loss": 1.9659,
      "step": 227526
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.4615328311920166,
      "learning_rate": 2.3064118087307016e-07,
      "loss": 1.7949,
      "step": 227527
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.4391748905181885,
      "learning_rate": 2.304808390431234e-07,
      "loss": 2.0104,
      "step": 227528
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.4164097309112549,
      "learning_rate": 2.3032055294596174e-07,
      "loss": 1.9977,
      "step": 227529
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.477160096168518,
      "learning_rate": 2.3016032258165172e-07,
      "loss": 1.9246,
      "step": 227530
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.5827747583389282,
      "learning_rate": 2.300001479502267e-07,
      "loss": 2.0816,
      "step": 227531
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.3622937202453613,
      "learning_rate": 2.2984002905171994e-07,
      "loss": 1.9415,
      "step": 227532
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.40946626663208,
      "learning_rate": 2.2967996588609817e-07,
      "loss": 1.9491,
      "step": 227533
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.688678503036499,
      "learning_rate": 2.2951995845349457e-07,
      "loss": 2.1709,
      "step": 227534
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.4879696369171143,
      "learning_rate": 2.2936000675384257e-07,
      "loss": 1.8733,
      "step": 227535
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.452080488204956,
      "learning_rate": 2.292001107872088e-07,
      "loss": 2.0169,
      "step": 227536
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.4881294965744019,
      "learning_rate": 2.290402705536265e-07,
      "loss": 2.1416,
      "step": 227537
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.4323772192001343,
      "learning_rate": 2.2888048605316234e-07,
      "loss": 1.7955,
      "step": 227538
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.7210732698440552,
      "learning_rate": 2.2872075728574968e-07,
      "loss": 1.9777,
      "step": 227539
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.423349142074585,
      "learning_rate": 2.2856108425152177e-07,
      "loss": 2.0205,
      "step": 227540
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.398402452468872,
      "learning_rate": 2.2840146695044525e-07,
      "loss": 1.8744,
      "step": 227541
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.4287638664245605,
      "learning_rate": 2.282419053825535e-07,
      "loss": 1.9203,
      "step": 227542
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.5164546966552734,
      "learning_rate": 2.2808239954791306e-07,
      "loss": 1.967,
      "step": 227543
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.460120677947998,
      "learning_rate": 2.27922949446524e-07,
      "loss": 1.8049,
      "step": 227544
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.5178039073944092,
      "learning_rate": 2.2776355507841958e-07,
      "loss": 2.1097,
      "step": 227545
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.4366363286972046,
      "learning_rate": 2.2760421644363313e-07,
      "loss": 1.8925,
      "step": 227546
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.5310853719711304,
      "learning_rate": 2.2744493354219794e-07,
      "loss": 2.0088,
      "step": 227547
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.547191858291626,
      "learning_rate": 2.2728570637414732e-07,
      "loss": 2.2211,
      "step": 227548
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.4745572805404663,
      "learning_rate": 2.271265349394813e-07,
      "loss": 1.9477,
      "step": 227549
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.3926868438720703,
      "learning_rate": 2.2696741923826646e-07,
      "loss": 2.0568,
      "step": 227550
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.3358943462371826,
      "learning_rate": 2.2680835927050278e-07,
      "loss": 2.1555,
      "step": 227551
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.5172295570373535,
      "learning_rate": 2.2664935503625693e-07,
      "loss": 1.875,
      "step": 227552
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.5079307556152344,
      "learning_rate": 2.2649040653552885e-07,
      "loss": 1.9056,
      "step": 227553
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.3894628286361694,
      "learning_rate": 2.263315137683519e-07,
      "loss": 2.2687,
      "step": 227554
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.4860162734985352,
      "learning_rate": 2.2617267673472605e-07,
      "loss": 2.18,
      "step": 227555
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.3342299461364746,
      "learning_rate": 2.2601389543475125e-07,
      "loss": 1.8753,
      "step": 227556
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.555140733718872,
      "learning_rate": 2.2585516986842745e-07,
      "loss": 2.2385,
      "step": 227557
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.4527511596679688,
      "learning_rate": 2.256965000357547e-07,
      "loss": 2.1021,
      "step": 227558
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.5012662410736084,
      "learning_rate": 2.2553788593679956e-07,
      "loss": 2.1947,
      "step": 227559
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.431046962738037,
      "learning_rate": 2.2537932757156207e-07,
      "loss": 1.9998,
      "step": 227560
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.4725991487503052,
      "learning_rate": 2.2522082494007555e-07,
      "loss": 1.9159,
      "step": 227561
    },
    {
      "epoch": 2.96,
      "grad_norm": 2.5312511920928955,
      "learning_rate": 2.2506237804237325e-07,
      "loss": 2.1068,
      "step": 227562
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.3828558921813965,
      "learning_rate": 2.2490398687852184e-07,
      "loss": 1.9841,
      "step": 227563
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.61487877368927,
      "learning_rate": 2.247456514485213e-07,
      "loss": 1.9991,
      "step": 227564
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.4384989738464355,
      "learning_rate": 2.2458737175237164e-07,
      "loss": 2.1951,
      "step": 227565
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.5497924089431763,
      "learning_rate": 2.2442914779013942e-07,
      "loss": 1.9969,
      "step": 227566
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.6393743753433228,
      "learning_rate": 2.24270979561858e-07,
      "loss": 1.9793,
      "step": 227567
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.587989091873169,
      "learning_rate": 2.241128670675274e-07,
      "loss": 1.973,
      "step": 227568
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.4591805934906006,
      "learning_rate": 2.239548103071809e-07,
      "loss": 1.9397,
      "step": 227569
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.5287444591522217,
      "learning_rate": 2.2379680928088507e-07,
      "loss": 1.8618,
      "step": 227570
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.6258776187896729,
      "learning_rate": 2.2363886398863996e-07,
      "loss": 2.2087,
      "step": 227571
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.382748007774353,
      "learning_rate": 2.2348097443047885e-07,
      "loss": 2.0278,
      "step": 227572
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.6025946140289307,
      "learning_rate": 2.233231406064351e-07,
      "loss": 1.9497,
      "step": 227573
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.4524072408676147,
      "learning_rate": 2.2316536251650864e-07,
      "loss": 1.9994,
      "step": 227574
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.4846315383911133,
      "learning_rate": 2.230076401607994e-07,
      "loss": 1.945,
      "step": 227575
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.6129885911941528,
      "learning_rate": 2.2284997353924084e-07,
      "loss": 1.8739,
      "step": 227576
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.3388341665267944,
      "learning_rate": 2.226923626519328e-07,
      "loss": 1.7361,
      "step": 227577
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.5104209184646606,
      "learning_rate": 2.2253480749890861e-07,
      "loss": 1.7994,
      "step": 227578
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.514230489730835,
      "learning_rate": 2.22377308080135e-07,
      "loss": 2.2346,
      "step": 227579
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.4804747104644775,
      "learning_rate": 2.2221986439571182e-07,
      "loss": 1.953,
      "step": 227580
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.4107403755187988,
      "learning_rate": 2.2206247644563912e-07,
      "loss": 1.7852,
      "step": 227581
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.5427439212799072,
      "learning_rate": 2.219051442299169e-07,
      "loss": 2.0244,
      "step": 227582
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.624890685081482,
      "learning_rate": 2.2174786774861174e-07,
      "loss": 2.1558,
      "step": 227583
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.4586281776428223,
      "learning_rate": 2.21590647001757e-07,
      "loss": 1.8904,
      "step": 227584
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.3934199810028076,
      "learning_rate": 2.2143348198935263e-07,
      "loss": 2.2756,
      "step": 227585
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.5186517238616943,
      "learning_rate": 2.2127637271143195e-07,
      "loss": 2.066,
      "step": 227586
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.5756076574325562,
      "learning_rate": 2.2111931916806158e-07,
      "loss": 2.2268,
      "step": 227587
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.5156232118606567,
      "learning_rate": 2.2096232135920821e-07,
      "loss": 1.9868,
      "step": 227588
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.4968034029006958,
      "learning_rate": 2.2080537928497176e-07,
      "loss": 2.0357,
      "step": 227589
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.5631979703903198,
      "learning_rate": 2.2064849294531893e-07,
      "loss": 1.7583,
      "step": 227590
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.3237173557281494,
      "learning_rate": 2.2049166234034965e-07,
      "loss": 1.988,
      "step": 227591
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.4041097164154053,
      "learning_rate": 2.2033488746999728e-07,
      "loss": 1.9153,
      "step": 227592
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.3984098434448242,
      "learning_rate": 2.2017816833436176e-07,
      "loss": 1.7084,
      "step": 227593
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.396378993988037,
      "learning_rate": 2.2002150493347638e-07,
      "loss": 1.8683,
      "step": 227594
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.4041122198104858,
      "learning_rate": 2.1986489726734114e-07,
      "loss": 1.671,
      "step": 227595
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.4486316442489624,
      "learning_rate": 2.1970834533595604e-07,
      "loss": 2.0336,
      "step": 227596
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.4233553409576416,
      "learning_rate": 2.1955184913942103e-07,
      "loss": 2.0915,
      "step": 227597
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.3688384294509888,
      "learning_rate": 2.1939540867773609e-07,
      "loss": 2.0644,
      "step": 227598
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.343337893486023,
      "learning_rate": 2.192390239509012e-07,
      "loss": 2.0727,
      "step": 227599
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.413033127784729,
      "learning_rate": 2.190826949589497e-07,
      "loss": 1.9638,
      "step": 227600
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.4536170959472656,
      "learning_rate": 2.189264217019815e-07,
      "loss": 2.0606,
      "step": 227601
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.4444431066513062,
      "learning_rate": 2.1877020417992996e-07,
      "loss": 2.1176,
      "step": 227602
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.4280922412872314,
      "learning_rate": 2.1861404239289504e-07,
      "loss": 2.1239,
      "step": 227603
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.6522281169891357,
      "learning_rate": 2.1845793634087673e-07,
      "loss": 2.1612,
      "step": 227604
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.466151475906372,
      "learning_rate": 2.183018860239083e-07,
      "loss": 1.9947,
      "step": 227605
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.59730863571167,
      "learning_rate": 2.181458914419898e-07,
      "loss": 2.0177,
      "step": 227606
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.4522972106933594,
      "learning_rate": 2.179899525951878e-07,
      "loss": 2.0087,
      "step": 227607
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.391052484512329,
      "learning_rate": 2.1783406948353567e-07,
      "loss": 1.8489,
      "step": 227608
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.393790602684021,
      "learning_rate": 2.1767824210703334e-07,
      "loss": 1.9351,
      "step": 227609
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.4236608743667603,
      "learning_rate": 2.1752247046571414e-07,
      "loss": 2.1448,
      "step": 227610
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.4906705617904663,
      "learning_rate": 2.173667545596447e-07,
      "loss": 1.9992,
      "step": 227611
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.5659582614898682,
      "learning_rate": 2.172110943887917e-07,
      "loss": 1.8698,
      "step": 227612
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.451006293296814,
      "learning_rate": 2.1705548995325507e-07,
      "loss": 1.9426,
      "step": 227613
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.6610136032104492,
      "learning_rate": 2.1689994125300148e-07,
      "loss": 1.9656,
      "step": 227614
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.7343083620071411,
      "learning_rate": 2.1674444828809757e-07,
      "loss": 2.2668,
      "step": 227615
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.4047293663024902,
      "learning_rate": 2.165890110585433e-07,
      "loss": 1.9738,
      "step": 227616
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.4317197799682617,
      "learning_rate": 2.1643362956440536e-07,
      "loss": 1.7494,
      "step": 227617
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.4831123352050781,
      "learning_rate": 2.1627830380568367e-07,
      "loss": 1.9639,
      "step": 227618
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.5376598834991455,
      "learning_rate": 2.1612303378241158e-07,
      "loss": 1.9925,
      "step": 227619
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.2873201370239258,
      "learning_rate": 2.1596781949462237e-07,
      "loss": 1.7706,
      "step": 227620
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.4533536434173584,
      "learning_rate": 2.158126609423494e-07,
      "loss": 1.8664,
      "step": 227621
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.4264391660690308,
      "learning_rate": 2.156575581255926e-07,
      "loss": 2.1288,
      "step": 227622
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.4793171882629395,
      "learning_rate": 2.1550251104441861e-07,
      "loss": 2.1435,
      "step": 227623
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.4536572694778442,
      "learning_rate": 2.1534751969886076e-07,
      "loss": 2.107,
      "step": 227624
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.5831786394119263,
      "learning_rate": 2.1519258408888574e-07,
      "loss": 2.0254,
      "step": 227625
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.4623132944107056,
      "learning_rate": 2.1503770421459342e-07,
      "loss": 1.9502,
      "step": 227626
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.5250756740570068,
      "learning_rate": 2.1488288007598385e-07,
      "loss": 1.9339,
      "step": 227627
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.54133939743042,
      "learning_rate": 2.1472811167309033e-07,
      "loss": 2.0723,
      "step": 227628
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.4912467002868652,
      "learning_rate": 2.1457339900594616e-07,
      "loss": 2.1899,
      "step": 227629
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.3726160526275635,
      "learning_rate": 2.1441874207455134e-07,
      "loss": 1.8393,
      "step": 227630
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.5306018590927124,
      "learning_rate": 2.1426414087893917e-07,
      "loss": 1.9681,
      "step": 227631
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.3523370027542114,
      "learning_rate": 2.1410959541917627e-07,
      "loss": 2.4154,
      "step": 227632
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.4613615274429321,
      "learning_rate": 2.1395510569526264e-07,
      "loss": 2.0889,
      "step": 227633
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.489044189453125,
      "learning_rate": 2.1380067170726488e-07,
      "loss": 2.0128,
      "step": 227634
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.4548431634902954,
      "learning_rate": 2.1364629345514972e-07,
      "loss": 1.8727,
      "step": 227635
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.4427772760391235,
      "learning_rate": 2.1349197093898373e-07,
      "loss": 1.9082,
      "step": 227636
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.4967103004455566,
      "learning_rate": 2.1333770415880024e-07,
      "loss": 2.0599,
      "step": 227637
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.553873896598816,
      "learning_rate": 2.1318349311459927e-07,
      "loss": 2.0476,
      "step": 227638
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.3507648706436157,
      "learning_rate": 2.1302933780641406e-07,
      "loss": 1.8421,
      "step": 227639
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.6223392486572266,
      "learning_rate": 2.128752382343113e-07,
      "loss": 1.8415,
      "step": 227640
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.4302061796188354,
      "learning_rate": 2.1272119439825764e-07,
      "loss": 2.0716,
      "step": 227641
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.4035483598709106,
      "learning_rate": 2.12567206298353e-07,
      "loss": 1.8797,
      "step": 227642
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.5271704196929932,
      "learning_rate": 2.1241327393459738e-07,
      "loss": 2.1331,
      "step": 227643
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.5452321767807007,
      "learning_rate": 2.1225939730699082e-07,
      "loss": 1.9912,
      "step": 227644
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.5084630250930786,
      "learning_rate": 2.121055764155999e-07,
      "loss": 2.0006,
      "step": 227645
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.5497339963912964,
      "learning_rate": 2.119518112604246e-07,
      "loss": 1.9157,
      "step": 227646
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.4354982376098633,
      "learning_rate": 2.1179810184149825e-07,
      "loss": 1.8428,
      "step": 227647
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.4878215789794922,
      "learning_rate": 2.1164444815888747e-07,
      "loss": 1.8105,
      "step": 227648
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.5451728105545044,
      "learning_rate": 2.1149085021255896e-07,
      "loss": 2.1252,
      "step": 227649
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.474437952041626,
      "learning_rate": 2.113373080026126e-07,
      "loss": 1.7626,
      "step": 227650
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.463037371635437,
      "learning_rate": 2.1118382152898182e-07,
      "loss": 2.0358,
      "step": 227651
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.4039032459259033,
      "learning_rate": 2.1103039079179985e-07,
      "loss": 1.8459,
      "step": 227652
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.5195258855819702,
      "learning_rate": 2.1087701579103334e-07,
      "loss": 1.9313,
      "step": 227653
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.350013256072998,
      "learning_rate": 2.1072369652671562e-07,
      "loss": 1.9619,
      "step": 227654
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.551584005355835,
      "learning_rate": 2.1057043299887998e-07,
      "loss": 1.8169,
      "step": 227655
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.5349889993667603,
      "learning_rate": 2.1041722520755978e-07,
      "loss": 1.9527,
      "step": 227656
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.397271752357483,
      "learning_rate": 2.1026407315278826e-07,
      "loss": 2.0756,
      "step": 227657
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.4374258518218994,
      "learning_rate": 2.1011097683459876e-07,
      "loss": 1.8741,
      "step": 227658
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.4886008501052856,
      "learning_rate": 2.0995793625299128e-07,
      "loss": 1.8374,
      "step": 227659
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.384229063987732,
      "learning_rate": 2.0980495140803245e-07,
      "loss": 2.0597,
      "step": 227660
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.662451982498169,
      "learning_rate": 2.0965202229968892e-07,
      "loss": 2.0482,
      "step": 227661
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.4280970096588135,
      "learning_rate": 2.0949914892806064e-07,
      "loss": 1.8098,
      "step": 227662
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.4624971151351929,
      "learning_rate": 2.093463312931476e-07,
      "loss": 1.8362,
      "step": 227663
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.4810163974761963,
      "learning_rate": 2.091935693949498e-07,
      "loss": 1.9777,
      "step": 227664
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.4761662483215332,
      "learning_rate": 2.0904086323356717e-07,
      "loss": 1.9749,
      "step": 227665
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.5582515001296997,
      "learning_rate": 2.0888821280893307e-07,
      "loss": 2.1538,
      "step": 227666
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.447508454322815,
      "learning_rate": 2.0873561812114748e-07,
      "loss": 1.9636,
      "step": 227667
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.503934621810913,
      "learning_rate": 2.0858307917024365e-07,
      "loss": 2.026,
      "step": 227668
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.468874216079712,
      "learning_rate": 2.0843059595618827e-07,
      "loss": 1.9564,
      "step": 227669
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.4286476373672485,
      "learning_rate": 2.08278168479048e-07,
      "loss": 2.03,
      "step": 227670
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.4654089212417603,
      "learning_rate": 2.081257967388561e-07,
      "loss": 2.007,
      "step": 227671
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.4347103834152222,
      "learning_rate": 2.079734807356459e-07,
      "loss": 1.998,
      "step": 227672
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.377095341682434,
      "learning_rate": 2.078212204694174e-07,
      "loss": 1.9828,
      "step": 227673
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.6426737308502197,
      "learning_rate": 2.076690159402039e-07,
      "loss": 1.8691,
      "step": 227674
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.4746047258377075,
      "learning_rate": 2.075168671480387e-07,
      "loss": 1.9954,
      "step": 227675
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.2851407527923584,
      "learning_rate": 2.0736477409298847e-07,
      "loss": 1.8438,
      "step": 227676
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.32257878780365,
      "learning_rate": 2.072127367750198e-07,
      "loss": 1.7769,
      "step": 227677
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.497165322303772,
      "learning_rate": 2.070607551941994e-07,
      "loss": 2.0195,
      "step": 227678
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.4662284851074219,
      "learning_rate": 2.0690882935056053e-07,
      "loss": 1.954,
      "step": 227679
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.394567608833313,
      "learning_rate": 2.0675695924410318e-07,
      "loss": 1.7592,
      "step": 227680
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.4086192846298218,
      "learning_rate": 2.066051448748607e-07,
      "loss": 1.9275,
      "step": 227681
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.4898358583450317,
      "learning_rate": 2.0645338624286634e-07,
      "loss": 2.1797,
      "step": 227682
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.8759912252426147,
      "learning_rate": 2.0630168334815344e-07,
      "loss": 1.9737,
      "step": 227683
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.5189450979232788,
      "learning_rate": 2.0615003619075531e-07,
      "loss": 1.9755,
      "step": 227684
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.539759635925293,
      "learning_rate": 2.0599844477070526e-07,
      "loss": 2.0121,
      "step": 227685
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.4910999536514282,
      "learning_rate": 2.0584690908800327e-07,
      "loss": 1.9844,
      "step": 227686
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.4842499494552612,
      "learning_rate": 2.0569542914268266e-07,
      "loss": 2.011,
      "step": 227687
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.5351943969726562,
      "learning_rate": 2.0554400493481004e-07,
      "loss": 1.868,
      "step": 227688
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.3473247289657593,
      "learning_rate": 2.053926364643521e-07,
      "loss": 1.9134,
      "step": 227689
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.4817663431167603,
      "learning_rate": 2.0524132373140878e-07,
      "loss": 2.1174,
      "step": 227690
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.472341775894165,
      "learning_rate": 2.0509006673594674e-07,
      "loss": 2.1524,
      "step": 227691
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.4806740283966064,
      "learning_rate": 2.0493886547803262e-07,
      "loss": 2.0829,
      "step": 227692
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.6010169982910156,
      "learning_rate": 2.0478771995766642e-07,
      "loss": 1.9722,
      "step": 227693
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.505181074142456,
      "learning_rate": 2.046366301748814e-07,
      "loss": 2.2071,
      "step": 227694
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.462278962135315,
      "learning_rate": 2.0448559612974426e-07,
      "loss": 1.8756,
      "step": 227695
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.4185795783996582,
      "learning_rate": 2.043346178222216e-07,
      "loss": 1.8199,
      "step": 227696
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.4845880270004272,
      "learning_rate": 2.0418369525238009e-07,
      "loss": 2.144,
      "step": 227697
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.4253865480422974,
      "learning_rate": 2.0403282842025304e-07,
      "loss": 2.1642,
      "step": 227698
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.6449205875396729,
      "learning_rate": 2.0388201732584043e-07,
      "loss": 1.9034,
      "step": 227699
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.66339910030365,
      "learning_rate": 2.0373126196917556e-07,
      "loss": 1.9675,
      "step": 227700
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.5821574926376343,
      "learning_rate": 2.0358056235032505e-07,
      "loss": 1.9382,
      "step": 227701
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.6353867053985596,
      "learning_rate": 2.034299184692889e-07,
      "loss": 2.0887,
      "step": 227702
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.4878026247024536,
      "learning_rate": 2.0327933032606714e-07,
      "loss": 1.8588,
      "step": 227703
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.493203043937683,
      "learning_rate": 2.0312879792072635e-07,
      "loss": 2.0076,
      "step": 227704
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.4691137075424194,
      "learning_rate": 2.029783212532665e-07,
      "loss": 1.9556,
      "step": 227705
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.4476884603500366,
      "learning_rate": 2.0282790032375429e-07,
      "loss": 2.0788,
      "step": 227706
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.4767918586730957,
      "learning_rate": 2.0267753513218965e-07,
      "loss": 1.9594,
      "step": 227707
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.4377564191818237,
      "learning_rate": 2.0252722567860592e-07,
      "loss": 2.0829,
      "step": 227708
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.45927894115448,
      "learning_rate": 2.0237697196303637e-07,
      "loss": 2.0292,
      "step": 227709
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.4846111536026,
      "learning_rate": 2.0222677398551434e-07,
      "loss": 1.8905,
      "step": 227710
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.4356473684310913,
      "learning_rate": 2.0207663174603983e-07,
      "loss": 2.008,
      "step": 227711
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.502570629119873,
      "learning_rate": 2.0192654524464614e-07,
      "loss": 2.0706,
      "step": 227712
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.4150750637054443,
      "learning_rate": 2.0177651448139988e-07,
      "loss": 2.0778,
      "step": 227713
    },
    {
      "epoch": 2.96,
      "grad_norm": 1.5536062717437744,
      "learning_rate": 2.0162653945630102e-07,
      "loss": 2.1874,
      "step": 227714
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.467636227607727,
      "learning_rate": 2.0147662016934962e-07,
      "loss": 2.0693,
      "step": 227715
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.483656644821167,
      "learning_rate": 2.0132675662064558e-07,
      "loss": 1.8952,
      "step": 227716
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.553297758102417,
      "learning_rate": 2.0117694881015557e-07,
      "loss": 1.9699,
      "step": 227717
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.543219804763794,
      "learning_rate": 2.010271967379129e-07,
      "loss": 1.8764,
      "step": 227718
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.3610646724700928,
      "learning_rate": 2.008775004039509e-07,
      "loss": 1.9427,
      "step": 227719
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.5699740648269653,
      "learning_rate": 2.0072785980833615e-07,
      "loss": 1.8431,
      "step": 227720
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.419411301612854,
      "learning_rate": 2.005782749510354e-07,
      "loss": 2.0488,
      "step": 227721
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.4310580492019653,
      "learning_rate": 2.0042874583214852e-07,
      "loss": 1.8283,
      "step": 227722
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.441504955291748,
      "learning_rate": 2.002792724516089e-07,
      "loss": 2.0179,
      "step": 227723
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.4173446893692017,
      "learning_rate": 2.001298548095165e-07,
      "loss": 1.9381,
      "step": 227724
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.4469623565673828,
      "learning_rate": 1.9998049290590458e-07,
      "loss": 1.9241,
      "step": 227725
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.4864213466644287,
      "learning_rate": 1.9983118674073984e-07,
      "loss": 2.0397,
      "step": 227726
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.5940178632736206,
      "learning_rate": 1.996819363141222e-07,
      "loss": 1.9809,
      "step": 227727
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.4157196283340454,
      "learning_rate": 1.995327416260184e-07,
      "loss": 1.9922,
      "step": 227728
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.5228768587112427,
      "learning_rate": 1.993836026764617e-07,
      "loss": 1.9379,
      "step": 227729
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.4285147190093994,
      "learning_rate": 1.9923451946551872e-07,
      "loss": 1.6589,
      "step": 227730
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.4386345148086548,
      "learning_rate": 1.9908549199322278e-07,
      "loss": 1.8552,
      "step": 227731
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.5888649225234985,
      "learning_rate": 1.9893652025954054e-07,
      "loss": 1.9328,
      "step": 227732
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.5345338582992554,
      "learning_rate": 1.9878760426453865e-07,
      "loss": 2.0171,
      "step": 227733
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.3986116647720337,
      "learning_rate": 1.9863874400825043e-07,
      "loss": 2.1605,
      "step": 227734
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.4845118522644043,
      "learning_rate": 1.9848993949070913e-07,
      "loss": 1.994,
      "step": 227735
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.5323865413665771,
      "learning_rate": 1.983411907119148e-07,
      "loss": 2.0117,
      "step": 227736
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.437759518623352,
      "learning_rate": 1.9819249767190071e-07,
      "loss": 1.814,
      "step": 227737
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.3114019632339478,
      "learning_rate": 1.980438603707002e-07,
      "loss": 2.1186,
      "step": 227738
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.4856674671173096,
      "learning_rate": 1.9789527880834656e-07,
      "loss": 2.0932,
      "step": 227739
    },
    {
      "epoch": 2.97,
      "grad_norm": 2.220778226852417,
      "learning_rate": 1.9774675298487308e-07,
      "loss": 2.0821,
      "step": 227740
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.4025670289993286,
      "learning_rate": 1.975982829002798e-07,
      "loss": 1.9963,
      "step": 227741
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.4160107374191284,
      "learning_rate": 1.974498685546333e-07,
      "loss": 2.1923,
      "step": 227742
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.499398946762085,
      "learning_rate": 1.973015099479336e-07,
      "loss": 2.0621,
      "step": 227743
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.5080901384353638,
      "learning_rate": 1.9715320708021397e-07,
      "loss": 1.9804,
      "step": 227744
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.5100197792053223,
      "learning_rate": 1.9700495995150778e-07,
      "loss": 2.0158,
      "step": 227745
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.3866099119186401,
      "learning_rate": 1.96856768561815e-07,
      "loss": 1.8295,
      "step": 227746
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.393998384475708,
      "learning_rate": 1.967086329112022e-07,
      "loss": 1.6547,
      "step": 227747
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.5324722528457642,
      "learning_rate": 1.9656055299966945e-07,
      "loss": 1.9713,
      "step": 227748
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.5597138404846191,
      "learning_rate": 1.9641252882728332e-07,
      "loss": 1.8665,
      "step": 227749
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.425050973892212,
      "learning_rate": 1.9626456039401051e-07,
      "loss": 2.0156,
      "step": 227750
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.4968746900558472,
      "learning_rate": 1.9611664769991764e-07,
      "loss": 2.0465,
      "step": 227751
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.4092483520507812,
      "learning_rate": 1.95968790745038e-07,
      "loss": 2.0265,
      "step": 227752
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.4313321113586426,
      "learning_rate": 1.9582098952940494e-07,
      "loss": 2.1069,
      "step": 227753
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.414263367652893,
      "learning_rate": 1.9567324405298513e-07,
      "loss": 1.8897,
      "step": 227754
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.37077796459198,
      "learning_rate": 1.9552555431587846e-07,
      "loss": 2.0415,
      "step": 227755
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.6139403581619263,
      "learning_rate": 1.9537792031808498e-07,
      "loss": 1.6781,
      "step": 227756
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.5219357013702393,
      "learning_rate": 1.9523034205960463e-07,
      "loss": 2.0533,
      "step": 227757
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.462712287902832,
      "learning_rate": 1.950828195405041e-07,
      "loss": 1.8482,
      "step": 227758
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.4194873571395874,
      "learning_rate": 1.949353527607833e-07,
      "loss": 2.0718,
      "step": 227759
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.5202453136444092,
      "learning_rate": 1.9478794172050894e-07,
      "loss": 2.0189,
      "step": 227760
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.3984700441360474,
      "learning_rate": 1.9464058641968093e-07,
      "loss": 1.9042,
      "step": 227761
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.5391924381256104,
      "learning_rate": 1.9449328685829936e-07,
      "loss": 1.9966,
      "step": 227762
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.405081033706665,
      "learning_rate": 1.9434604303646406e-07,
      "loss": 1.9571,
      "step": 227763
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.5678431987762451,
      "learning_rate": 1.9419885495410848e-07,
      "loss": 1.9797,
      "step": 227764
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.3834483623504639,
      "learning_rate": 1.9405172261136582e-07,
      "loss": 2.0066,
      "step": 227765
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.518179178237915,
      "learning_rate": 1.9390464600816948e-07,
      "loss": 1.8243,
      "step": 227766
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.4565726518630981,
      "learning_rate": 1.9375762514458603e-07,
      "loss": 1.8899,
      "step": 227767
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.5458979606628418,
      "learning_rate": 1.9361066002064885e-07,
      "loss": 1.7815,
      "step": 227768
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.4028669595718384,
      "learning_rate": 1.934637506363912e-07,
      "loss": 2.1753,
      "step": 227769
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.4161289930343628,
      "learning_rate": 1.9331689699181306e-07,
      "loss": 1.8642,
      "step": 227770
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.389194369316101,
      "learning_rate": 1.9317009908698112e-07,
      "loss": 1.9292,
      "step": 227771
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.4292851686477661,
      "learning_rate": 1.93023356921862e-07,
      "loss": 1.9755,
      "step": 227772
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.4121099710464478,
      "learning_rate": 1.9287667049652233e-07,
      "loss": 1.9695,
      "step": 227773
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.3788529634475708,
      "learning_rate": 1.9273003981099544e-07,
      "loss": 1.9887,
      "step": 227774
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.3572473526000977,
      "learning_rate": 1.925834648653146e-07,
      "loss": 1.9103,
      "step": 227775
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.4429582357406616,
      "learning_rate": 1.9243694565947987e-07,
      "loss": 2.1167,
      "step": 227776
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.56513249874115,
      "learning_rate": 1.922904821935578e-07,
      "loss": 1.9906,
      "step": 227777
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.5381755828857422,
      "learning_rate": 1.9214407446751513e-07,
      "loss": 2.0351,
      "step": 227778
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.4461852312088013,
      "learning_rate": 1.9199772248141842e-07,
      "loss": 1.691,
      "step": 227779
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.4467569589614868,
      "learning_rate": 1.9185142623526771e-07,
      "loss": 2.0188,
      "step": 227780
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.504469871520996,
      "learning_rate": 1.9170518572916293e-07,
      "loss": 2.1144,
      "step": 227781
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.4801082611083984,
      "learning_rate": 1.9155900096303745e-07,
      "loss": 1.9638,
      "step": 227782
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.5067909955978394,
      "learning_rate": 1.9141287193695786e-07,
      "loss": 2.0223,
      "step": 227783
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.327338695526123,
      "learning_rate": 1.9126679865099082e-07,
      "loss": 1.9546,
      "step": 227784
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.4277852773666382,
      "learning_rate": 1.9112078110510298e-07,
      "loss": 1.8421,
      "step": 227785
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.5338243246078491,
      "learning_rate": 1.9097481929932769e-07,
      "loss": 1.969,
      "step": 227786
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.480778694152832,
      "learning_rate": 1.908289132337315e-07,
      "loss": 2.2006,
      "step": 227787
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.6691116094589233,
      "learning_rate": 1.9068306290831448e-07,
      "loss": 1.8987,
      "step": 227788
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.549648642539978,
      "learning_rate": 1.905372683231099e-07,
      "loss": 1.8052,
      "step": 227789
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.5003807544708252,
      "learning_rate": 1.9039152947815107e-07,
      "loss": 2.0152,
      "step": 227790
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.5201033353805542,
      "learning_rate": 1.9024584637343797e-07,
      "loss": 1.8652,
      "step": 227791
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.428948998451233,
      "learning_rate": 1.9010021900903727e-07,
      "loss": 1.9604,
      "step": 227792
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.3264080286026,
      "learning_rate": 1.8995464738494892e-07,
      "loss": 1.9201,
      "step": 227793
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.4843251705169678,
      "learning_rate": 1.8980913150120626e-07,
      "loss": 1.9909,
      "step": 227794
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.5197545289993286,
      "learning_rate": 1.8966367135784256e-07,
      "loss": 2.0059,
      "step": 227795
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.6872507333755493,
      "learning_rate": 1.8951826695489113e-07,
      "loss": 1.7932,
      "step": 227796
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.4419987201690674,
      "learning_rate": 1.8937291829235202e-07,
      "loss": 1.9937,
      "step": 227797
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.3824280500411987,
      "learning_rate": 1.8922762537025848e-07,
      "loss": 2.1276,
      "step": 227798
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.482747197151184,
      "learning_rate": 1.8908238818867715e-07,
      "loss": 1.8492,
      "step": 227799
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.398524522781372,
      "learning_rate": 1.8893720674757473e-07,
      "loss": 1.9023,
      "step": 227800
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.3704595565795898,
      "learning_rate": 1.887920810470178e-07,
      "loss": 1.818,
      "step": 227801
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.3179824352264404,
      "learning_rate": 1.8864701108700643e-07,
      "loss": 2.0033,
      "step": 227802
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.4190709590911865,
      "learning_rate": 1.8850199686760715e-07,
      "loss": 1.9568,
      "step": 227803
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.392390251159668,
      "learning_rate": 1.8835703838882e-07,
      "loss": 1.9992,
      "step": 227804
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.5047473907470703,
      "learning_rate": 1.8821213565067827e-07,
      "loss": 1.8575,
      "step": 227805
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.5212576389312744,
      "learning_rate": 1.880672886532153e-07,
      "loss": 2.0077,
      "step": 227806
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.4635593891143799,
      "learning_rate": 1.8792249739643107e-07,
      "loss": 1.9539,
      "step": 227807
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.6013741493225098,
      "learning_rate": 1.8777776188039218e-07,
      "loss": 2.0048,
      "step": 227808
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.3032182455062866,
      "learning_rate": 1.8763308210509863e-07,
      "loss": 1.9033,
      "step": 227809
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.4708473682403564,
      "learning_rate": 1.8748845807058377e-07,
      "loss": 2.0409,
      "step": 227810
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.5371534824371338,
      "learning_rate": 1.8734388977684757e-07,
      "loss": 2.0283,
      "step": 227811
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.3623944520950317,
      "learning_rate": 1.8719937722398992e-07,
      "loss": 1.718,
      "step": 227812
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.4412353038787842,
      "learning_rate": 1.8705492041194426e-07,
      "loss": 2.0962,
      "step": 227813
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.4645354747772217,
      "learning_rate": 1.8691051934081048e-07,
      "loss": 2.0126,
      "step": 227814
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.51007080078125,
      "learning_rate": 1.867661740105886e-07,
      "loss": 2.0576,
      "step": 227815
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.4614274501800537,
      "learning_rate": 1.8662188442131188e-07,
      "loss": 2.1146,
      "step": 227816
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.5337550640106201,
      "learning_rate": 1.8647765057298037e-07,
      "loss": 1.9534,
      "step": 227817
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.557395100593567,
      "learning_rate": 1.8633347246566065e-07,
      "loss": 2.2888,
      "step": 227818
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.3546559810638428,
      "learning_rate": 1.8618935009935276e-07,
      "loss": 2.1484,
      "step": 227819
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.5385336875915527,
      "learning_rate": 1.8604528347408997e-07,
      "loss": 2.1334,
      "step": 227820
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.4317798614501953,
      "learning_rate": 1.859012725899056e-07,
      "loss": 1.9564,
      "step": 227821
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.430970311164856,
      "learning_rate": 1.8575731744683299e-07,
      "loss": 1.79,
      "step": 227822
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.3953393697738647,
      "learning_rate": 1.8561341804487206e-07,
      "loss": 1.8955,
      "step": 227823
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.5622925758361816,
      "learning_rate": 1.854695743840562e-07,
      "loss": 1.8099,
      "step": 227824
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.4216514825820923,
      "learning_rate": 1.8532578646445195e-07,
      "loss": 1.8964,
      "step": 227825
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.6333932876586914,
      "learning_rate": 1.8518205428602607e-07,
      "loss": 1.9877,
      "step": 227826
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.4077765941619873,
      "learning_rate": 1.8503837784884513e-07,
      "loss": 2.0681,
      "step": 227827
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.3769989013671875,
      "learning_rate": 1.8489475715294245e-07,
      "loss": 1.8329,
      "step": 227828
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.4643319845199585,
      "learning_rate": 1.8475119219828471e-07,
      "loss": 1.6757,
      "step": 227829
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.520950198173523,
      "learning_rate": 1.8460768298497186e-07,
      "loss": 1.766,
      "step": 227830
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.4176455736160278,
      "learning_rate": 1.8446422951300388e-07,
      "loss": 2.1247,
      "step": 227831
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.506312370300293,
      "learning_rate": 1.8432083178238077e-07,
      "loss": 2.0175,
      "step": 227832
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.4444669485092163,
      "learning_rate": 1.8417748979316917e-07,
      "loss": 1.9537,
      "step": 227833
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.6645913124084473,
      "learning_rate": 1.84034203545369e-07,
      "loss": 2.0906,
      "step": 227834
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.4943815469741821,
      "learning_rate": 1.8389097303901367e-07,
      "loss": 1.8733,
      "step": 227835
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.4293181896209717,
      "learning_rate": 1.8374779827416974e-07,
      "loss": 1.9487,
      "step": 227836
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.493255853652954,
      "learning_rate": 1.836046792507706e-07,
      "loss": 1.8331,
      "step": 227837
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.4576045274734497,
      "learning_rate": 1.8346161596894948e-07,
      "loss": 2.0187,
      "step": 227838
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.4239484071731567,
      "learning_rate": 1.8331860842863976e-07,
      "loss": 2.0289,
      "step": 227839
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.452890396118164,
      "learning_rate": 1.831756566299414e-07,
      "loss": 2.0091,
      "step": 227840
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.4106851816177368,
      "learning_rate": 1.83032760572821e-07,
      "loss": 2.1405,
      "step": 227841
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.52848219871521,
      "learning_rate": 1.8288992025734528e-07,
      "loss": 1.972,
      "step": 227842
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.485146403312683,
      "learning_rate": 1.827471356835475e-07,
      "loss": 2.2314,
      "step": 227843
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.658401608467102,
      "learning_rate": 1.8260440685142762e-07,
      "loss": 1.9487,
      "step": 227844
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.4553736448287964,
      "learning_rate": 1.8246173376101902e-07,
      "loss": 2.1074,
      "step": 227845
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.3672198057174683,
      "learning_rate": 1.8231911641235497e-07,
      "loss": 1.902,
      "step": 227846
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.4124579429626465,
      "learning_rate": 1.8217655480543548e-07,
      "loss": 2.0812,
      "step": 227847
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.5139079093933105,
      "learning_rate": 1.8203404894032713e-07,
      "loss": 1.8865,
      "step": 227848
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.5270074605941772,
      "learning_rate": 1.8189159881702998e-07,
      "loss": 1.9625,
      "step": 227849
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.3847631216049194,
      "learning_rate": 1.817492044356106e-07,
      "loss": 2.0763,
      "step": 227850
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.6095986366271973,
      "learning_rate": 1.8160686579600237e-07,
      "loss": 2.0716,
      "step": 227851
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.3451967239379883,
      "learning_rate": 1.8146458289833854e-07,
      "loss": 1.7478,
      "step": 227852
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.5232974290847778,
      "learning_rate": 1.813223557425858e-07,
      "loss": 2.2902,
      "step": 227853
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.402554988861084,
      "learning_rate": 1.8118018432881077e-07,
      "loss": 1.8844,
      "step": 227854
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.6573303937911987,
      "learning_rate": 1.810380686569801e-07,
      "loss": 1.9993,
      "step": 227855
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.4333012104034424,
      "learning_rate": 1.8089600872716048e-07,
      "loss": 1.8114,
      "step": 227856
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.4183021783828735,
      "learning_rate": 1.8075400453938515e-07,
      "loss": 1.9847,
      "step": 227857
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.5610642433166504,
      "learning_rate": 1.8061205609365415e-07,
      "loss": 1.8838,
      "step": 227858
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.3708276748657227,
      "learning_rate": 1.8047016339000077e-07,
      "loss": 1.9045,
      "step": 227859
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.3921173810958862,
      "learning_rate": 1.803283264284916e-07,
      "loss": 1.9129,
      "step": 227860
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.791170597076416,
      "learning_rate": 1.8018654520909338e-07,
      "loss": 1.8975,
      "step": 227861
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.5388118028640747,
      "learning_rate": 1.8004481973183938e-07,
      "loss": 2.1068,
      "step": 227862
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.4073978662490845,
      "learning_rate": 1.7990314999679624e-07,
      "loss": 1.8478,
      "step": 227863
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.3760935068130493,
      "learning_rate": 1.7976153600396393e-07,
      "loss": 1.9468,
      "step": 227864
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.485660433769226,
      "learning_rate": 1.7961997775337579e-07,
      "loss": 2.005,
      "step": 227865
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.7330694198608398,
      "learning_rate": 1.794784752450651e-07,
      "loss": 1.9928,
      "step": 227866
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.3643301725387573,
      "learning_rate": 1.7933702847903187e-07,
      "loss": 1.767,
      "step": 227867
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.5120404958724976,
      "learning_rate": 1.7919563745534272e-07,
      "loss": 1.9764,
      "step": 227868
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.450452208518982,
      "learning_rate": 1.7905430217399763e-07,
      "loss": 1.8078,
      "step": 227869
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.4958657026290894,
      "learning_rate": 1.7891302263499663e-07,
      "loss": 1.6745,
      "step": 227870
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.6065844297409058,
      "learning_rate": 1.787717988384063e-07,
      "loss": 1.9809,
      "step": 227871
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.3592162132263184,
      "learning_rate": 1.7863063078425999e-07,
      "loss": 2.0769,
      "step": 227872
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.6206692457199097,
      "learning_rate": 1.7848951847255765e-07,
      "loss": 1.9869,
      "step": 227873
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.3898266553878784,
      "learning_rate": 1.7834846190333264e-07,
      "loss": 1.9179,
      "step": 227874
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.4197510480880737,
      "learning_rate": 1.7820746107661821e-07,
      "loss": 1.9847,
      "step": 227875
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.4990065097808838,
      "learning_rate": 1.780665159924144e-07,
      "loss": 1.9076,
      "step": 227876
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.3484333753585815,
      "learning_rate": 1.779256266507878e-07,
      "loss": 1.926,
      "step": 227877
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.9009346961975098,
      "learning_rate": 1.7778479305173843e-07,
      "loss": 2.1054,
      "step": 227878
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.7649041414260864,
      "learning_rate": 1.776440151952996e-07,
      "loss": 1.8416,
      "step": 227879
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.4995359182357788,
      "learning_rate": 1.7750329308150458e-07,
      "loss": 2.1029,
      "step": 227880
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.5075137615203857,
      "learning_rate": 1.773626267103534e-07,
      "loss": 2.0471,
      "step": 227881
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.4566975831985474,
      "learning_rate": 1.7722201608187937e-07,
      "loss": 2.0529,
      "step": 227882
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.4519370794296265,
      "learning_rate": 1.770814611961491e-07,
      "loss": 1.9699,
      "step": 227883
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.4153579473495483,
      "learning_rate": 1.7694096205312923e-07,
      "loss": 1.9688,
      "step": 227884
    },
    {
      "epoch": 2.97,
      "grad_norm": 2.0989646911621094,
      "learning_rate": 1.7680051865291978e-07,
      "loss": 1.9769,
      "step": 227885
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.5237423181533813,
      "learning_rate": 1.7666013099545407e-07,
      "loss": 1.8336,
      "step": 227886
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.563377022743225,
      "learning_rate": 1.7651979908083202e-07,
      "loss": 2.1223,
      "step": 227887
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.3574970960617065,
      "learning_rate": 1.7637952290905366e-07,
      "loss": 1.6355,
      "step": 227888
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.5634803771972656,
      "learning_rate": 1.7623930248015228e-07,
      "loss": 1.9035,
      "step": 227889
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.3650386333465576,
      "learning_rate": 1.7609913779412787e-07,
      "loss": 1.7997,
      "step": 227890
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.538310170173645,
      "learning_rate": 1.7595902885104708e-07,
      "loss": 2.1352,
      "step": 227891
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.5612553358078003,
      "learning_rate": 1.7581897565090986e-07,
      "loss": 1.9635,
      "step": 227892
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.480214238166809,
      "learning_rate": 1.7567897819374954e-07,
      "loss": 1.8882,
      "step": 227893
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.380372166633606,
      "learning_rate": 1.7553903647956612e-07,
      "loss": 1.7633,
      "step": 227894
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.4080445766448975,
      "learning_rate": 1.7539915050842624e-07,
      "loss": 2.0445,
      "step": 227895
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.6142176389694214,
      "learning_rate": 1.7525932028032986e-07,
      "loss": 1.9339,
      "step": 227896
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.4782661199569702,
      "learning_rate": 1.751195457953436e-07,
      "loss": 2.0279,
      "step": 227897
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.3284822702407837,
      "learning_rate": 1.7497982705343415e-07,
      "loss": 2.1757,
      "step": 227898
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.6632506847381592,
      "learning_rate": 1.7484016405466818e-07,
      "loss": 1.9864,
      "step": 227899
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.4340566396713257,
      "learning_rate": 1.747005567990456e-07,
      "loss": 1.819,
      "step": 227900
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.4578498601913452,
      "learning_rate": 1.7456100528659978e-07,
      "loss": 2.0971,
      "step": 227901
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.5238709449768066,
      "learning_rate": 1.744215095173973e-07,
      "loss": 2.0781,
      "step": 227902
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.4394793510437012,
      "learning_rate": 1.7428206949140488e-07,
      "loss": 1.8527,
      "step": 227903
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.5685091018676758,
      "learning_rate": 1.7414268520868911e-07,
      "loss": 2.1285,
      "step": 227904
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.392014980316162,
      "learning_rate": 1.7400335666925002e-07,
      "loss": 1.9199,
      "step": 227905
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.5923056602478027,
      "learning_rate": 1.7386408387312088e-07,
      "loss": 2.1446,
      "step": 227906
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.5272443294525146,
      "learning_rate": 1.7372486682030174e-07,
      "loss": 2.1504,
      "step": 227907
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.6748862266540527,
      "learning_rate": 1.7358570551089247e-07,
      "loss": 2.1007,
      "step": 227908
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.327789306640625,
      "learning_rate": 1.7344659994485975e-07,
      "loss": 2.0834,
      "step": 227909
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.5579047203063965,
      "learning_rate": 1.7330755012223695e-07,
      "loss": 1.9949,
      "step": 227910
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.475982427597046,
      "learning_rate": 1.7316855604305734e-07,
      "loss": 1.8478,
      "step": 227911
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.6038806438446045,
      "learning_rate": 1.7302961770735424e-07,
      "loss": 1.7709,
      "step": 227912
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.531036615371704,
      "learning_rate": 1.7289073511512763e-07,
      "loss": 2.1173,
      "step": 227913
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.5228124856948853,
      "learning_rate": 1.7275190826644413e-07,
      "loss": 1.8912,
      "step": 227914
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.6870378255844116,
      "learning_rate": 1.7261313716127045e-07,
      "loss": 1.9711,
      "step": 227915
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.553602695465088,
      "learning_rate": 1.724744217997065e-07,
      "loss": 2.2355,
      "step": 227916
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.5972453355789185,
      "learning_rate": 1.7233576218171895e-07,
      "loss": 2.0402,
      "step": 227917
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.507788896560669,
      "learning_rate": 1.7219715830734115e-07,
      "loss": 1.9932,
      "step": 227918
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.396092176437378,
      "learning_rate": 1.7205861017663968e-07,
      "loss": 1.9545,
      "step": 227919
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.6122750043869019,
      "learning_rate": 1.7192011778958126e-07,
      "loss": 2.0646,
      "step": 227920
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.5550752878189087,
      "learning_rate": 1.7178168114626579e-07,
      "loss": 2.06,
      "step": 227921
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.3502156734466553,
      "learning_rate": 1.7164330024662664e-07,
      "loss": 1.9248,
      "step": 227922
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.4720977544784546,
      "learning_rate": 1.7150497509076378e-07,
      "loss": 2.0084,
      "step": 227923
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.3930765390396118,
      "learning_rate": 1.7136670567867717e-07,
      "loss": 1.8733,
      "step": 227924
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.505683183670044,
      "learning_rate": 1.7122849201040012e-07,
      "loss": 1.9571,
      "step": 227925
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.6001354455947876,
      "learning_rate": 1.7109033408593266e-07,
      "loss": 1.8081,
      "step": 227926
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.509494662284851,
      "learning_rate": 1.7095223190534134e-07,
      "loss": 1.9063,
      "step": 227927
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.4584441184997559,
      "learning_rate": 1.7081418546862624e-07,
      "loss": 1.9188,
      "step": 227928
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.5125290155410767,
      "learning_rate": 1.7067619477578731e-07,
      "loss": 1.9334,
      "step": 227929
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.51368248462677,
      "learning_rate": 1.705382598269245e-07,
      "loss": 2.0372,
      "step": 227930
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.3722615242004395,
      "learning_rate": 1.7040038062197115e-07,
      "loss": 1.8164,
      "step": 227931
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.4510743618011475,
      "learning_rate": 1.702625571610272e-07,
      "loss": 1.8207,
      "step": 227932
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.569532036781311,
      "learning_rate": 1.701247894440927e-07,
      "loss": 1.7633,
      "step": 227933
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.376318097114563,
      "learning_rate": 1.699870774712009e-07,
      "loss": 1.7114,
      "step": 227934
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.4899433851242065,
      "learning_rate": 1.698494212423518e-07,
      "loss": 1.8643,
      "step": 227935
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.4735740423202515,
      "learning_rate": 1.697118207575787e-07,
      "loss": 2.0115,
      "step": 227936
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.3963369131088257,
      "learning_rate": 1.6957427601694828e-07,
      "loss": 1.9512,
      "step": 227937
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.5048941373825073,
      "learning_rate": 1.6943678702042718e-07,
      "loss": 1.9051,
      "step": 227938
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.4776537418365479,
      "learning_rate": 1.692993537680487e-07,
      "loss": 1.8615,
      "step": 227939
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.4962481260299683,
      "learning_rate": 1.6916197625991278e-07,
      "loss": 1.9689,
      "step": 227940
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.3829680681228638,
      "learning_rate": 1.690246544959528e-07,
      "loss": 2.0673,
      "step": 227941
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.353552222251892,
      "learning_rate": 1.688873884762354e-07,
      "loss": 1.831,
      "step": 227942
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.3952258825302124,
      "learning_rate": 1.6875017820076053e-07,
      "loss": 2.0082,
      "step": 227943
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.4596691131591797,
      "learning_rate": 1.6861302366959485e-07,
      "loss": 1.8021,
      "step": 227944
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.674530029296875,
      "learning_rate": 1.6847592488277162e-07,
      "loss": 2.0208,
      "step": 227945
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.4037790298461914,
      "learning_rate": 1.6833888184025756e-07,
      "loss": 2.1701,
      "step": 227946
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.487252950668335,
      "learning_rate": 1.682018945421193e-07,
      "loss": 2.098,
      "step": 227947
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.5155963897705078,
      "learning_rate": 1.680649629883568e-07,
      "loss": 1.973,
      "step": 227948
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.4269484281539917,
      "learning_rate": 1.679280871790034e-07,
      "loss": 1.8733,
      "step": 227949
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.526138186454773,
      "learning_rate": 1.677912671141257e-07,
      "loss": 1.9509,
      "step": 227950
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.3921176195144653,
      "learning_rate": 1.676545027936904e-07,
      "loss": 1.9253,
      "step": 227951
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.5382155179977417,
      "learning_rate": 1.6751779421776412e-07,
      "loss": 1.9686,
      "step": 227952
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.488741159439087,
      "learning_rate": 1.6738114138634683e-07,
      "loss": 2.14,
      "step": 227953
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.3442554473876953,
      "learning_rate": 1.6724454429947186e-07,
      "loss": 1.9774,
      "step": 227954
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.3959214687347412,
      "learning_rate": 1.6710800295717253e-07,
      "loss": 1.6993,
      "step": 227955
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.445792555809021,
      "learning_rate": 1.669715173594488e-07,
      "loss": 2.0231,
      "step": 227956
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.5494962930679321,
      "learning_rate": 1.668350875063673e-07,
      "loss": 2.0463,
      "step": 227957
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.4199590682983398,
      "learning_rate": 1.6669871339792805e-07,
      "loss": 2.0675,
      "step": 227958
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.5670959949493408,
      "learning_rate": 1.6656239503416434e-07,
      "loss": 1.902,
      "step": 227959
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.3863626718521118,
      "learning_rate": 1.6642613241507617e-07,
      "loss": 2.0093,
      "step": 227960
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.4444019794464111,
      "learning_rate": 1.6628992554073017e-07,
      "loss": 2.0049,
      "step": 227961
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.5297654867172241,
      "learning_rate": 1.66153774411093e-07,
      "loss": 2.2176,
      "step": 227962
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.5087133646011353,
      "learning_rate": 1.660176790262646e-07,
      "loss": 1.9319,
      "step": 227963
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.609994649887085,
      "learning_rate": 1.65881639386245e-07,
      "loss": 1.9104,
      "step": 227964
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.4358441829681396,
      "learning_rate": 1.6574565549100083e-07,
      "loss": 1.8917,
      "step": 227965
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.3600709438323975,
      "learning_rate": 1.6560972734063205e-07,
      "loss": 1.9357,
      "step": 227966
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.532900094985962,
      "learning_rate": 1.6547385493513866e-07,
      "loss": 2.0705,
      "step": 227967
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.6351275444030762,
      "learning_rate": 1.6533803827455395e-07,
      "loss": 2.0215,
      "step": 227968
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.4837071895599365,
      "learning_rate": 1.6520227735887793e-07,
      "loss": 2.0038,
      "step": 227969
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.4699722528457642,
      "learning_rate": 1.650665721881439e-07,
      "loss": 1.7816,
      "step": 227970
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.4307774305343628,
      "learning_rate": 1.6493092276238518e-07,
      "loss": 2.2304,
      "step": 227971
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.4839304685592651,
      "learning_rate": 1.6479532908163507e-07,
      "loss": 2.0554,
      "step": 227972
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.4503867626190186,
      "learning_rate": 1.6465979114589357e-07,
      "loss": 1.8111,
      "step": 227973
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.3619564771652222,
      "learning_rate": 1.6452430895522728e-07,
      "loss": 1.7092,
      "step": 227974
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.3998037576675415,
      "learning_rate": 1.6438888250960292e-07,
      "loss": 2.0086,
      "step": 227975
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.4440481662750244,
      "learning_rate": 1.642535118091204e-07,
      "loss": 2.2092,
      "step": 227976
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.4559175968170166,
      "learning_rate": 1.6411819685374638e-07,
      "loss": 1.7794,
      "step": 227977
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.4160109758377075,
      "learning_rate": 1.6398293764351423e-07,
      "loss": 1.7448,
      "step": 227978
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.4877893924713135,
      "learning_rate": 1.638477341784572e-07,
      "loss": 2.0837,
      "step": 227979
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.4345766305923462,
      "learning_rate": 1.637125864585753e-07,
      "loss": 1.8964,
      "step": 227980
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.5158601999282837,
      "learning_rate": 1.635774944839685e-07,
      "loss": 1.9977,
      "step": 227981
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.5301377773284912,
      "learning_rate": 1.6344245825457015e-07,
      "loss": 2.0372,
      "step": 227982
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.4518622159957886,
      "learning_rate": 1.6330747777048015e-07,
      "loss": 1.909,
      "step": 227983
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.4890695810317993,
      "learning_rate": 1.631725530316652e-07,
      "loss": 2.2206,
      "step": 227984
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.5847041606903076,
      "learning_rate": 1.6303768403819195e-07,
      "loss": 1.9953,
      "step": 227985
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.5308020114898682,
      "learning_rate": 1.6290287079009366e-07,
      "loss": 2.0295,
      "step": 227986
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.479871392250061,
      "learning_rate": 1.6276811328733709e-07,
      "loss": 1.8437,
      "step": 227987
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.4680845737457275,
      "learning_rate": 1.6263341152998876e-07,
      "loss": 1.8134,
      "step": 227988
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.6203508377075195,
      "learning_rate": 1.6249876551808206e-07,
      "loss": 1.9604,
      "step": 227989
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.7258083820343018,
      "learning_rate": 1.6236417525158362e-07,
      "loss": 1.7553,
      "step": 227990
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.4770240783691406,
      "learning_rate": 1.622296407305934e-07,
      "loss": 2.0132,
      "step": 227991
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.5416496992111206,
      "learning_rate": 1.620951619551114e-07,
      "loss": 1.9125,
      "step": 227992
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.4954081773757935,
      "learning_rate": 1.6196073892513762e-07,
      "loss": 2.0951,
      "step": 227993
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.476133108139038,
      "learning_rate": 1.6182637164073863e-07,
      "loss": 2.1258,
      "step": 227994
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.5562113523483276,
      "learning_rate": 1.616920601019145e-07,
      "loss": 1.9182,
      "step": 227995
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.3743153810501099,
      "learning_rate": 1.6155780430866516e-07,
      "loss": 2.0102,
      "step": 227996
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.6265102624893188,
      "learning_rate": 1.6142360426105727e-07,
      "loss": 1.9279,
      "step": 227997
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.5532758235931396,
      "learning_rate": 1.612894599590908e-07,
      "loss": 1.7348,
      "step": 227998
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.5925246477127075,
      "learning_rate": 1.6115537140283242e-07,
      "loss": 1.8372,
      "step": 227999
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.5058492422103882,
      "learning_rate": 1.6102133859224875e-07,
      "loss": 2.0684,
      "step": 228000
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.6843184232711792,
      "learning_rate": 1.6088736152737313e-07,
      "loss": 2.1407,
      "step": 228001
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.4929755926132202,
      "learning_rate": 1.6075344020827218e-07,
      "loss": 2.0428,
      "step": 228002
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.4866943359375,
      "learning_rate": 1.606195746349459e-07,
      "loss": 2.0245,
      "step": 228003
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.4181987047195435,
      "learning_rate": 1.6048576480742758e-07,
      "loss": 1.9402,
      "step": 228004
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.4560458660125732,
      "learning_rate": 1.6035201072571725e-07,
      "loss": 2.1015,
      "step": 228005
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.6523597240447998,
      "learning_rate": 1.6021831238988146e-07,
      "loss": 2.0066,
      "step": 228006
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.393723726272583,
      "learning_rate": 1.60084669799887e-07,
      "loss": 1.958,
      "step": 228007
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.406252384185791,
      "learning_rate": 1.599510829558337e-07,
      "loss": 1.8587,
      "step": 228008
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.5609958171844482,
      "learning_rate": 1.59817551857655e-07,
      "loss": 2.0647,
      "step": 228009
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.5129660367965698,
      "learning_rate": 1.596840765054508e-07,
      "loss": 2.1304,
      "step": 228010
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.3427932262420654,
      "learning_rate": 1.595506568992544e-07,
      "loss": 1.6246,
      "step": 228011
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.6052833795547485,
      "learning_rate": 1.594172930389992e-07,
      "loss": 2.087,
      "step": 228012
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.3879756927490234,
      "learning_rate": 1.5928398492481842e-07,
      "loss": 1.9476,
      "step": 228013
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.5047825574874878,
      "learning_rate": 1.5915073255664546e-07,
      "loss": 2.2664,
      "step": 228014
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.4100368022918701,
      "learning_rate": 1.5901753593458022e-07,
      "loss": 1.816,
      "step": 228015
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.4752813577651978,
      "learning_rate": 1.5888439505858941e-07,
      "loss": 1.9201,
      "step": 228016
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.4364233016967773,
      "learning_rate": 1.5875130992873962e-07,
      "loss": 2.0198,
      "step": 228017
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.573846459388733,
      "learning_rate": 1.586182805450309e-07,
      "loss": 1.8284,
      "step": 228018
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.4664753675460815,
      "learning_rate": 1.584853069074965e-07,
      "loss": 2.1231,
      "step": 228019
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.4132070541381836,
      "learning_rate": 1.5835238901616975e-07,
      "loss": 2.0373,
      "step": 228020
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.7409483194351196,
      "learning_rate": 1.5821952687105067e-07,
      "loss": 1.8637,
      "step": 228021
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.553828477859497,
      "learning_rate": 1.5808672047217253e-07,
      "loss": 2.0189,
      "step": 228022
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.4717586040496826,
      "learning_rate": 1.5795396981956865e-07,
      "loss": 2.0603,
      "step": 228023
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.4172632694244385,
      "learning_rate": 1.5782127491327234e-07,
      "loss": 2.1079,
      "step": 228024
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.4739065170288086,
      "learning_rate": 1.576886357532836e-07,
      "loss": 1.8894,
      "step": 228025
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.5236488580703735,
      "learning_rate": 1.5755605233966907e-07,
      "loss": 1.9123,
      "step": 228026
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.6336638927459717,
      "learning_rate": 1.574235246723954e-07,
      "loss": 1.9526,
      "step": 228027
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.4166319370269775,
      "learning_rate": 1.572910527515292e-07,
      "loss": 1.8771,
      "step": 228028
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.480068564414978,
      "learning_rate": 1.571586365770705e-07,
      "loss": 2.0136,
      "step": 228029
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.4355061054229736,
      "learning_rate": 1.5702627614905261e-07,
      "loss": 1.9421,
      "step": 228030
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.543005108833313,
      "learning_rate": 1.5689397146750883e-07,
      "loss": 1.8836,
      "step": 228031
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.5441209077835083,
      "learning_rate": 1.5676172253247242e-07,
      "loss": 1.9318,
      "step": 228032
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.4769489765167236,
      "learning_rate": 1.5662952934394345e-07,
      "loss": 1.8803,
      "step": 228033
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.352480173110962,
      "learning_rate": 1.5649739190195522e-07,
      "loss": 1.9699,
      "step": 228034
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.333012342453003,
      "learning_rate": 1.5636531020654096e-07,
      "loss": 2.1408,
      "step": 228035
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.4424939155578613,
      "learning_rate": 1.5623328425773406e-07,
      "loss": 2.2357,
      "step": 228036
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.5422890186309814,
      "learning_rate": 1.561013140555345e-07,
      "loss": 1.904,
      "step": 228037
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.5452191829681396,
      "learning_rate": 1.5596939959994227e-07,
      "loss": 1.8998,
      "step": 228038
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.4793510437011719,
      "learning_rate": 1.5583754089105727e-07,
      "loss": 1.9921,
      "step": 228039
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.4477108716964722,
      "learning_rate": 1.5570573792884622e-07,
      "loss": 2.1426,
      "step": 228040
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.4098552465438843,
      "learning_rate": 1.5557399071334243e-07,
      "loss": 2.1357,
      "step": 228041
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.499985694885254,
      "learning_rate": 1.5544229924457917e-07,
      "loss": 2.0013,
      "step": 228042
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.5047345161437988,
      "learning_rate": 1.5531066352258982e-07,
      "loss": 1.9905,
      "step": 228043
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.4805760383605957,
      "learning_rate": 1.551790835473743e-07,
      "loss": 1.9446,
      "step": 228044
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.4156837463378906,
      "learning_rate": 1.5504755931896594e-07,
      "loss": 1.7727,
      "step": 228045
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.3943679332733154,
      "learning_rate": 1.549160908374314e-07,
      "loss": 2.1022,
      "step": 228046
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.4165747165679932,
      "learning_rate": 1.54784678102704e-07,
      "loss": 1.9353,
      "step": 228047
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.4665935039520264,
      "learning_rate": 1.5465332111491701e-07,
      "loss": 1.8913,
      "step": 228048
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.4212629795074463,
      "learning_rate": 1.545220198740038e-07,
      "loss": 1.9632,
      "step": 228049
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.6003568172454834,
      "learning_rate": 1.54390774380031e-07,
      "loss": 1.9424,
      "step": 228050
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.5215502977371216,
      "learning_rate": 1.542595846330319e-07,
      "loss": 1.9527,
      "step": 228051
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.4636856317520142,
      "learning_rate": 1.541284506330065e-07,
      "loss": 2.047,
      "step": 228052
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.4909851551055908,
      "learning_rate": 1.539973723799881e-07,
      "loss": 2.0473,
      "step": 228053
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.5126500129699707,
      "learning_rate": 1.5386634987397673e-07,
      "loss": 1.8767,
      "step": 228054
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.4030122756958008,
      "learning_rate": 1.5373538311507226e-07,
      "loss": 2.0587,
      "step": 228055
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.5054842233657837,
      "learning_rate": 1.5360447210320815e-07,
      "loss": 1.9383,
      "step": 228056
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.4029368162155151,
      "learning_rate": 1.5347361683845094e-07,
      "loss": 1.9842,
      "step": 228057
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.519004225730896,
      "learning_rate": 1.5334281732083397e-07,
      "loss": 2.1891,
      "step": 228058
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.580536961555481,
      "learning_rate": 1.5321207355039056e-07,
      "loss": 2.1868,
      "step": 228059
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.4642339944839478,
      "learning_rate": 1.530813855270874e-07,
      "loss": 1.8318,
      "step": 228060
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.4595253467559814,
      "learning_rate": 1.5295075325102434e-07,
      "loss": 1.901,
      "step": 228061
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.4478815793991089,
      "learning_rate": 1.5282017672216819e-07,
      "loss": 2.0491,
      "step": 228062
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.5388340950012207,
      "learning_rate": 1.5268965594055215e-07,
      "loss": 1.9794,
      "step": 228063
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.5747919082641602,
      "learning_rate": 1.5255919090620962e-07,
      "loss": 1.8772,
      "step": 228064
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.5332484245300293,
      "learning_rate": 1.5242878161920713e-07,
      "loss": 2.0532,
      "step": 228065
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.4850414991378784,
      "learning_rate": 1.5229842807947813e-07,
      "loss": 2.0041,
      "step": 228066
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.461517572402954,
      "learning_rate": 1.521681302871225e-07,
      "loss": 1.7425,
      "step": 228067
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.4816418886184692,
      "learning_rate": 1.5203788824214024e-07,
      "loss": 2.153,
      "step": 228068
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.4263122081756592,
      "learning_rate": 1.519077019445647e-07,
      "loss": 2.0727,
      "step": 228069
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.7016866207122803,
      "learning_rate": 1.5177757139436253e-07,
      "loss": 1.8725,
      "step": 228070
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.4250617027282715,
      "learning_rate": 1.5164749659163367e-07,
      "loss": 2.025,
      "step": 228071
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.4803050756454468,
      "learning_rate": 1.515174775363781e-07,
      "loss": 1.8279,
      "step": 228072
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.5299537181854248,
      "learning_rate": 1.5138751422862916e-07,
      "loss": 1.9334,
      "step": 228073
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.3740429878234863,
      "learning_rate": 1.5125760666838683e-07,
      "loss": 1.7964,
      "step": 228074
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.4942799806594849,
      "learning_rate": 1.511277548556844e-07,
      "loss": 2.1061,
      "step": 228075
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.4606168270111084,
      "learning_rate": 1.509979587905219e-07,
      "loss": 2.1005,
      "step": 228076
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.4493772983551025,
      "learning_rate": 1.5086821847299923e-07,
      "loss": 2.0857,
      "step": 228077
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.6042729616165161,
      "learning_rate": 1.507385339030498e-07,
      "loss": 1.9137,
      "step": 228078
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.5159845352172852,
      "learning_rate": 1.5060890508074018e-07,
      "loss": 2.0295,
      "step": 228079
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.4480986595153809,
      "learning_rate": 1.5047933200610374e-07,
      "loss": 2.2966,
      "step": 228080
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.4318925142288208,
      "learning_rate": 1.5034981467917373e-07,
      "loss": 1.9626,
      "step": 228081
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.4057031869888306,
      "learning_rate": 1.5022035309991687e-07,
      "loss": 2.1269,
      "step": 228082
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.5131893157958984,
      "learning_rate": 1.5009094726839976e-07,
      "loss": 2.136,
      "step": 228083
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.5245568752288818,
      "learning_rate": 1.4996159718465572e-07,
      "loss": 1.7858,
      "step": 228084
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.5958348512649536,
      "learning_rate": 1.4983230284868474e-07,
      "loss": 2.0008,
      "step": 228085
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.3976471424102783,
      "learning_rate": 1.4970306426052014e-07,
      "loss": 1.9276,
      "step": 228086
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.5536385774612427,
      "learning_rate": 1.495738814201952e-07,
      "loss": 1.8414,
      "step": 228087
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.521331787109375,
      "learning_rate": 1.4944475432774329e-07,
      "loss": 2.0547,
      "step": 228088
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.6258478164672852,
      "learning_rate": 1.4931568298313102e-07,
      "loss": 1.9709,
      "step": 228089
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.3595386743545532,
      "learning_rate": 1.4918666738642504e-07,
      "loss": 2.0069,
      "step": 228090
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.491919994354248,
      "learning_rate": 1.4905770753765867e-07,
      "loss": 1.9162,
      "step": 228091
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.488356351852417,
      "learning_rate": 1.4892880343683188e-07,
      "loss": 1.9854,
      "step": 228092
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.4321849346160889,
      "learning_rate": 1.4879995508401133e-07,
      "loss": 2.147,
      "step": 228093
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.3611868619918823,
      "learning_rate": 1.4867116247916365e-07,
      "loss": 1.8526,
      "step": 228094
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.4201844930648804,
      "learning_rate": 1.4854242562232221e-07,
      "loss": 1.8675,
      "step": 228095
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.496109127998352,
      "learning_rate": 1.4841374451355358e-07,
      "loss": 1.8008,
      "step": 228096
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.510359764099121,
      "learning_rate": 1.4828511915285778e-07,
      "loss": 2.0192,
      "step": 228097
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.4425984621047974,
      "learning_rate": 1.481565495402348e-07,
      "loss": 2.0335,
      "step": 228098
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.4580579996109009,
      "learning_rate": 1.4802803567575127e-07,
      "loss": 2.0839,
      "step": 228099
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.339991569519043,
      "learning_rate": 1.4789957755937388e-07,
      "loss": 2.0129,
      "step": 228100
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.560262680053711,
      "learning_rate": 1.4777117519120253e-07,
      "loss": 2.0558,
      "step": 228101
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.5046616792678833,
      "learning_rate": 1.476428285712039e-07,
      "loss": 2.179,
      "step": 228102
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.5067412853240967,
      "learning_rate": 1.4751453769941135e-07,
      "loss": 2.1072,
      "step": 228103
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.4854151010513306,
      "learning_rate": 1.4738630257585816e-07,
      "loss": 1.9023,
      "step": 228104
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.424666404724121,
      "learning_rate": 1.4725812320057762e-07,
      "loss": 2.0696,
      "step": 228105
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.4795312881469727,
      "learning_rate": 1.4712999957356975e-07,
      "loss": 2.0662,
      "step": 228106
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.3901816606521606,
      "learning_rate": 1.4700193169486784e-07,
      "loss": 1.8706,
      "step": 228107
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.3879313468933105,
      "learning_rate": 1.4687391956450523e-07,
      "loss": 2.107,
      "step": 228108
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.451024055480957,
      "learning_rate": 1.467459631825152e-07,
      "loss": 1.9352,
      "step": 228109
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.4343719482421875,
      "learning_rate": 1.4661806254886444e-07,
      "loss": 1.9969,
      "step": 228110
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.4662846326828003,
      "learning_rate": 1.4649021766365288e-07,
      "loss": 1.9875,
      "step": 228111
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.4398373365402222,
      "learning_rate": 1.463624285268472e-07,
      "loss": 1.7888,
      "step": 228112
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.475664496421814,
      "learning_rate": 1.4623469513851404e-07,
      "loss": 2.0502,
      "step": 228113
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.3652033805847168,
      "learning_rate": 1.4610701749865339e-07,
      "loss": 1.9695,
      "step": 228114
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.5044605731964111,
      "learning_rate": 1.4597939560729855e-07,
      "loss": 2.0479,
      "step": 228115
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.7028840780258179,
      "learning_rate": 1.4585182946444952e-07,
      "loss": 1.8902,
      "step": 228116
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.4756522178649902,
      "learning_rate": 1.457243190701396e-07,
      "loss": 1.9613,
      "step": 228117
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.5062034130096436,
      "learning_rate": 1.455968644244021e-07,
      "loss": 1.9582,
      "step": 228118
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.4173524379730225,
      "learning_rate": 1.4546946552727034e-07,
      "loss": 1.8129,
      "step": 228119
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.5700773000717163,
      "learning_rate": 1.4534212237874432e-07,
      "loss": 2.0672,
      "step": 228120
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.4894016981124878,
      "learning_rate": 1.4521483497889065e-07,
      "loss": 1.9219,
      "step": 228121
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.511861801147461,
      "learning_rate": 1.45087603327676e-07,
      "loss": 2.3227,
      "step": 228122
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.491559386253357,
      "learning_rate": 1.449604274251337e-07,
      "loss": 1.8588,
      "step": 228123
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.5048129558563232,
      "learning_rate": 1.4483330727133037e-07,
      "loss": 2.0362,
      "step": 228124
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.4681885242462158,
      "learning_rate": 1.4470624286626597e-07,
      "loss": 1.9528,
      "step": 228125
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.438391923904419,
      "learning_rate": 1.4457923420994055e-07,
      "loss": 2.1423,
      "step": 228126
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.4539132118225098,
      "learning_rate": 1.4445228130242071e-07,
      "loss": 1.9895,
      "step": 228127
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.3990193605422974,
      "learning_rate": 1.4432538414367313e-07,
      "loss": 2.1088,
      "step": 228128
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.6147866249084473,
      "learning_rate": 1.441985427337977e-07,
      "loss": 2.0987,
      "step": 228129
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.566683053970337,
      "learning_rate": 1.4407175707272788e-07,
      "loss": 1.9877,
      "step": 228130
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.443840503692627,
      "learning_rate": 1.4394502716056354e-07,
      "loss": 1.926,
      "step": 228131
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.4645075798034668,
      "learning_rate": 1.438183529973047e-07,
      "loss": 2.0729,
      "step": 228132
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.4722661972045898,
      "learning_rate": 1.4369173458295137e-07,
      "loss": 1.9788,
      "step": 228133
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.446039080619812,
      "learning_rate": 1.435651719175701e-07,
      "loss": 1.979,
      "step": 228134
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.4616199731826782,
      "learning_rate": 1.4343866500116096e-07,
      "loss": 1.8416,
      "step": 228135
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.4133728742599487,
      "learning_rate": 1.4331221383372393e-07,
      "loss": 1.8944,
      "step": 228136
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.5903440713882446,
      "learning_rate": 1.431858184153256e-07,
      "loss": 2.0709,
      "step": 228137
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.489147663116455,
      "learning_rate": 1.430594787459327e-07,
      "loss": 2.1212,
      "step": 228138
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.4538049697875977,
      "learning_rate": 1.4293319482564514e-07,
      "loss": 1.883,
      "step": 228139
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.5565431118011475,
      "learning_rate": 1.428069666544296e-07,
      "loss": 1.9527,
      "step": 228140
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.4401695728302002,
      "learning_rate": 1.4268079423235268e-07,
      "loss": 2.0358,
      "step": 228141
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.523438811302185,
      "learning_rate": 1.425546775593811e-07,
      "loss": 2.2619,
      "step": 228142
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.463185429573059,
      "learning_rate": 1.4242861663558146e-07,
      "loss": 1.9671,
      "step": 228143
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.4353572130203247,
      "learning_rate": 1.423026114609538e-07,
      "loss": 1.7651,
      "step": 228144
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.5690226554870605,
      "learning_rate": 1.4217666203553135e-07,
      "loss": 2.0319,
      "step": 228145
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.3052524328231812,
      "learning_rate": 1.4205076835934747e-07,
      "loss": 1.8082,
      "step": 228146
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.3789299726486206,
      "learning_rate": 1.4192493043240217e-07,
      "loss": 1.851,
      "step": 228147
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.391372561454773,
      "learning_rate": 1.41799148254762e-07,
      "loss": 2.03,
      "step": 228148
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.4339503049850464,
      "learning_rate": 1.4167342182639374e-07,
      "loss": 1.827,
      "step": 228149
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.4474327564239502,
      "learning_rate": 1.4154775114736395e-07,
      "loss": 2.0133,
      "step": 228150
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.3760558366775513,
      "learning_rate": 1.4142213621767262e-07,
      "loss": 1.7552,
      "step": 228151
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.8056522607803345,
      "learning_rate": 1.412965770373531e-07,
      "loss": 2.1343,
      "step": 228152
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.557185173034668,
      "learning_rate": 1.4117107360643865e-07,
      "loss": 2.0129,
      "step": 228153
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.4613063335418701,
      "learning_rate": 1.410456259249293e-07,
      "loss": 1.9014,
      "step": 228154
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.3884228467941284,
      "learning_rate": 1.4092023399285834e-07,
      "loss": 1.9626,
      "step": 228155
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.4541045427322388,
      "learning_rate": 1.407948978102591e-07,
      "loss": 2.1629,
      "step": 228156
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.4494949579238892,
      "learning_rate": 1.4066961737713156e-07,
      "loss": 2.0666,
      "step": 228157
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.5481235980987549,
      "learning_rate": 1.4054439269354233e-07,
      "loss": 1.9594,
      "step": 228158
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.38387930393219,
      "learning_rate": 1.4041922375945813e-07,
      "loss": 1.8777,
      "step": 228159
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.5647491216659546,
      "learning_rate": 1.4029411057494556e-07,
      "loss": 1.9013,
      "step": 228160
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.418899416923523,
      "learning_rate": 1.401690531400046e-07,
      "loss": 1.9205,
      "step": 228161
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.4619449377059937,
      "learning_rate": 1.4004405145466858e-07,
      "loss": 2.0454,
      "step": 228162
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.4224298000335693,
      "learning_rate": 1.399191055189708e-07,
      "loss": 1.9689,
      "step": 228163
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.599623203277588,
      "learning_rate": 1.3979421533291124e-07,
      "loss": 1.9703,
      "step": 228164
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.469067931175232,
      "learning_rate": 1.3966938089652324e-07,
      "loss": 1.7973,
      "step": 228165
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.5528082847595215,
      "learning_rate": 1.395446022098068e-07,
      "loss": 2.0272,
      "step": 228166
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.4204387664794922,
      "learning_rate": 1.394198792728618e-07,
      "loss": 1.8959,
      "step": 228167
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.416825294494629,
      "learning_rate": 1.3929521208562167e-07,
      "loss": 1.7568,
      "step": 228168
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.379160761833191,
      "learning_rate": 1.39170600648153e-07,
      "loss": 1.7519,
      "step": 228169
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.560469150543213,
      "learning_rate": 1.390460449604891e-07,
      "loss": 2.0857,
      "step": 228170
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.4537934064865112,
      "learning_rate": 1.3892154502263e-07,
      "loss": 1.8526,
      "step": 228171
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.5418106317520142,
      "learning_rate": 1.3879710083457562e-07,
      "loss": 2.1026,
      "step": 228172
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.594491958618164,
      "learning_rate": 1.3867271239642598e-07,
      "loss": 1.9825,
      "step": 228173
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.4816739559173584,
      "learning_rate": 1.3854837970811438e-07,
      "loss": 1.7557,
      "step": 228174
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.4685345888137817,
      "learning_rate": 1.3842410276974082e-07,
      "loss": 2.2033,
      "step": 228175
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.5873311758041382,
      "learning_rate": 1.3829988158127192e-07,
      "loss": 1.8619,
      "step": 228176
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.3518394231796265,
      "learning_rate": 1.3817571614277435e-07,
      "loss": 1.8966,
      "step": 228177
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.4959590435028076,
      "learning_rate": 1.3805160645421475e-07,
      "loss": 1.832,
      "step": 228178
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.447134256362915,
      "learning_rate": 1.379275525156931e-07,
      "loss": 1.9314,
      "step": 228179
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.442626953125,
      "learning_rate": 1.3780355432717604e-07,
      "loss": 2.4015,
      "step": 228180
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.503621220588684,
      "learning_rate": 1.3767961188869693e-07,
      "loss": 2.1597,
      "step": 228181
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.391817331314087,
      "learning_rate": 1.37555725200289e-07,
      "loss": 1.8292,
      "step": 228182
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.3809237480163574,
      "learning_rate": 1.3743189426195235e-07,
      "loss": 1.9529,
      "step": 228183
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.454227328300476,
      "learning_rate": 1.373081190737535e-07,
      "loss": 2.1012,
      "step": 228184
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.598540186882019,
      "learning_rate": 1.371843996356592e-07,
      "loss": 2.0151,
      "step": 228185
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.4752458333969116,
      "learning_rate": 1.3706073594773604e-07,
      "loss": 1.94,
      "step": 228186
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.3549524545669556,
      "learning_rate": 1.3693712801001732e-07,
      "loss": 1.9533,
      "step": 228187
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.51990807056427,
      "learning_rate": 1.3681357582246978e-07,
      "loss": 1.9218,
      "step": 228188
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.544853925704956,
      "learning_rate": 1.3669007938515997e-07,
      "loss": 2.0899,
      "step": 228189
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.4392372369766235,
      "learning_rate": 1.3656663869812124e-07,
      "loss": 2.1356,
      "step": 228190
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.5151565074920654,
      "learning_rate": 1.3644325376132027e-07,
      "loss": 1.9727,
      "step": 228191
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.4426047801971436,
      "learning_rate": 1.3631992457482367e-07,
      "loss": 1.9053,
      "step": 228192
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.5200128555297852,
      "learning_rate": 1.3619665113866474e-07,
      "loss": 1.9453,
      "step": 228193
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.439758062362671,
      "learning_rate": 1.360734334528435e-07,
      "loss": 1.9536,
      "step": 228194
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.4812675714492798,
      "learning_rate": 1.3595027151735994e-07,
      "loss": 2.0628,
      "step": 228195
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.5551286935806274,
      "learning_rate": 1.3582716533228065e-07,
      "loss": 1.9948,
      "step": 228196
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.397756576538086,
      "learning_rate": 1.3570411489760568e-07,
      "loss": 2.0663,
      "step": 228197
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.4858912229537964,
      "learning_rate": 1.3558112021340163e-07,
      "loss": 2.0868,
      "step": 228198
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.698264241218567,
      "learning_rate": 1.3545818127960183e-07,
      "loss": 2.1236,
      "step": 228199
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.3752821683883667,
      "learning_rate": 1.3533529809630628e-07,
      "loss": 2.0508,
      "step": 228200
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.498077154159546,
      "learning_rate": 1.3521247066351494e-07,
      "loss": 1.948,
      "step": 228201
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.468421220779419,
      "learning_rate": 1.3508969898122779e-07,
      "loss": 2.0152,
      "step": 228202
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.617294430732727,
      "learning_rate": 1.3496698304951147e-07,
      "loss": 2.0846,
      "step": 228203
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.6507368087768555,
      "learning_rate": 1.34844322868366e-07,
      "loss": 1.9073,
      "step": 228204
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.4020029306411743,
      "learning_rate": 1.3472171843779132e-07,
      "loss": 2.0214,
      "step": 228205
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.3373521566390991,
      "learning_rate": 1.3459916975782082e-07,
      "loss": 1.8772,
      "step": 228206
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.4253188371658325,
      "learning_rate": 1.3447667682852103e-07,
      "loss": 1.9683,
      "step": 228207
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.4953278303146362,
      "learning_rate": 1.34354239649892e-07,
      "loss": 1.8584,
      "step": 228208
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.5114842653274536,
      "learning_rate": 1.3423185822190042e-07,
      "loss": 2.0768,
      "step": 228209
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.540099024772644,
      "learning_rate": 1.3410953254464618e-07,
      "loss": 1.9299,
      "step": 228210
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.4900411367416382,
      "learning_rate": 1.339872626181293e-07,
      "loss": 1.9389,
      "step": 228211
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.3958871364593506,
      "learning_rate": 1.338650484423498e-07,
      "loss": 2.0933,
      "step": 228212
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.3877942562103271,
      "learning_rate": 1.3374289001734094e-07,
      "loss": 2.1096,
      "step": 228213
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.6030336618423462,
      "learning_rate": 1.336207873431361e-07,
      "loss": 2.2248,
      "step": 228214
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.4800554513931274,
      "learning_rate": 1.334987404197685e-07,
      "loss": 1.8591,
      "step": 228215
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.5668200254440308,
      "learning_rate": 1.3337674924720488e-07,
      "loss": 1.9325,
      "step": 228216
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.5485990047454834,
      "learning_rate": 1.3325481382554515e-07,
      "loss": 2.1057,
      "step": 228217
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.6527897119522095,
      "learning_rate": 1.3313293415475602e-07,
      "loss": 1.8871,
      "step": 228218
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.4628106355667114,
      "learning_rate": 1.3301111023487078e-07,
      "loss": 1.9846,
      "step": 228219
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.4196560382843018,
      "learning_rate": 1.328893420659227e-07,
      "loss": 1.9118,
      "step": 228220
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.5679117441177368,
      "learning_rate": 1.3276762964794518e-07,
      "loss": 2.0059,
      "step": 228221
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.453792929649353,
      "learning_rate": 1.3264597298093815e-07,
      "loss": 1.9621,
      "step": 228222
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.4611685276031494,
      "learning_rate": 1.3252437206493492e-07,
      "loss": 2.0321,
      "step": 228223
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.3923490047454834,
      "learning_rate": 1.3240282689993553e-07,
      "loss": 1.883,
      "step": 228224
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.5288147926330566,
      "learning_rate": 1.3228133748600656e-07,
      "loss": 2.0286,
      "step": 228225
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.340309739112854,
      "learning_rate": 1.32159903823148e-07,
      "loss": 1.9449,
      "step": 228226
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.6953294277191162,
      "learning_rate": 1.320385259113932e-07,
      "loss": 2.118,
      "step": 228227
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.5054994821548462,
      "learning_rate": 1.319172037507088e-07,
      "loss": 2.0374,
      "step": 228228
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.4826747179031372,
      "learning_rate": 1.3179593734119475e-07,
      "loss": 1.7428,
      "step": 228229
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.4491965770721436,
      "learning_rate": 1.3167472668285107e-07,
      "loss": 1.9232,
      "step": 228230
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.3390804529190063,
      "learning_rate": 1.315535717756444e-07,
      "loss": 1.7424,
      "step": 228231
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.5289475917816162,
      "learning_rate": 1.3143247261967472e-07,
      "loss": 2.2079,
      "step": 228232
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.4866613149642944,
      "learning_rate": 1.3131142921494197e-07,
      "loss": 2.1825,
      "step": 228233
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.4655201435089111,
      "learning_rate": 1.311904415614462e-07,
      "loss": 2.1147,
      "step": 228234
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.2872376441955566,
      "learning_rate": 1.310695096592207e-07,
      "loss": 2.1296,
      "step": 228235
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.4457416534423828,
      "learning_rate": 1.3094863350826546e-07,
      "loss": 2.0867,
      "step": 228236
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.405233383178711,
      "learning_rate": 1.3082781310864708e-07,
      "loss": 1.9506,
      "step": 228237
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.6263227462768555,
      "learning_rate": 1.3070704846039893e-07,
      "loss": 2.0982,
      "step": 228238
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.3717986345291138,
      "learning_rate": 1.3058633956345432e-07,
      "loss": 2.0887,
      "step": 228239
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.5368950366973877,
      "learning_rate": 1.3046568641794651e-07,
      "loss": 2.0239,
      "step": 228240
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.694381833076477,
      "learning_rate": 1.303450890238089e-07,
      "loss": 1.9105,
      "step": 228241
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.6325212717056274,
      "learning_rate": 1.3022454738114141e-07,
      "loss": 2.016,
      "step": 228242
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.533226728439331,
      "learning_rate": 1.301040614898774e-07,
      "loss": 1.8237,
      "step": 228243
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.4765020608901978,
      "learning_rate": 1.299836313501168e-07,
      "loss": 2.0133,
      "step": 228244
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.5892547369003296,
      "learning_rate": 1.2986325696185962e-07,
      "loss": 2.0208,
      "step": 228245
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.4305827617645264,
      "learning_rate": 1.2974293832510584e-07,
      "loss": 1.8609,
      "step": 228246
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.4349150657653809,
      "learning_rate": 1.296226754398888e-07,
      "loss": 1.9434,
      "step": 228247
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.5294102430343628,
      "learning_rate": 1.295024683062418e-07,
      "loss": 1.8502,
      "step": 228248
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.4527106285095215,
      "learning_rate": 1.293823169241648e-07,
      "loss": 1.9217,
      "step": 228249
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.4298640489578247,
      "learning_rate": 1.2926222129372444e-07,
      "loss": 2.0549,
      "step": 228250
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.6141904592514038,
      "learning_rate": 1.2914218141492073e-07,
      "loss": 2.0602,
      "step": 228251
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.5737049579620361,
      "learning_rate": 1.2902219728775364e-07,
      "loss": 2.0192,
      "step": 228252
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.3930847644805908,
      "learning_rate": 1.2890226891225651e-07,
      "loss": 2.0708,
      "step": 228253
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.52252995967865,
      "learning_rate": 1.2878239628846266e-07,
      "loss": 1.8727,
      "step": 228254
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.6028302907943726,
      "learning_rate": 1.2866257941637204e-07,
      "loss": 1.9561,
      "step": 228255
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.3362561464309692,
      "learning_rate": 1.285428182960513e-07,
      "loss": 1.7969,
      "step": 228256
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.5416301488876343,
      "learning_rate": 1.284231129274671e-07,
      "loss": 1.9558,
      "step": 228257
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.4880563020706177,
      "learning_rate": 1.283034633106861e-07,
      "loss": 1.8989,
      "step": 228258
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.4845975637435913,
      "learning_rate": 1.2818386944570824e-07,
      "loss": 2.1,
      "step": 228259
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.3401137590408325,
      "learning_rate": 1.2806433133256688e-07,
      "loss": 1.7885,
      "step": 228260
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.5089960098266602,
      "learning_rate": 1.279448489712953e-07,
      "loss": 1.9049,
      "step": 228261
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.4020717144012451,
      "learning_rate": 1.2782542236186022e-07,
      "loss": 1.8537,
      "step": 228262
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.3561341762542725,
      "learning_rate": 1.2770605150436153e-07,
      "loss": 1.5179,
      "step": 228263
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.561155080795288,
      "learning_rate": 1.2758673639876594e-07,
      "loss": 2.0738,
      "step": 228264
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.4903932809829712,
      "learning_rate": 1.2746747704510672e-07,
      "loss": 2.024,
      "step": 228265
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.4299591779708862,
      "learning_rate": 1.2734827344341724e-07,
      "loss": 2.0098,
      "step": 228266
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.4506700038909912,
      "learning_rate": 1.2722912559369747e-07,
      "loss": 1.9876,
      "step": 228267
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.6169673204421997,
      "learning_rate": 1.2711003349601402e-07,
      "loss": 2.0567,
      "step": 228268
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.6495234966278076,
      "learning_rate": 1.2699099715036687e-07,
      "loss": 2.2383,
      "step": 228269
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.4468203783035278,
      "learning_rate": 1.2687201655675606e-07,
      "loss": 2.0667,
      "step": 228270
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.2935633659362793,
      "learning_rate": 1.2675309171521486e-07,
      "loss": 2.1134,
      "step": 228271
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.5561062097549438,
      "learning_rate": 1.266342226257766e-07,
      "loss": 1.9009,
      "step": 228272
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.5437825918197632,
      "learning_rate": 1.265154092884413e-07,
      "loss": 1.9241,
      "step": 228273
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.472362995147705,
      "learning_rate": 1.2639665170327552e-07,
      "loss": 1.9296,
      "step": 228274
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.43074369430542,
      "learning_rate": 1.2627794987024598e-07,
      "loss": 2.0552,
      "step": 228275
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.488553762435913,
      "learning_rate": 1.261593037894193e-07,
      "loss": 1.9234,
      "step": 228276
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.4526331424713135,
      "learning_rate": 1.2604071346079548e-07,
      "loss": 1.8777,
      "step": 228277
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.571264624595642,
      "learning_rate": 1.2592217888440782e-07,
      "loss": 2.0861,
      "step": 228278
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.5143306255340576,
      "learning_rate": 1.258037000602896e-07,
      "loss": 2.0826,
      "step": 228279
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.4464932680130005,
      "learning_rate": 1.256852769884076e-07,
      "loss": 1.9389,
      "step": 228280
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.6538197994232178,
      "learning_rate": 1.2556690966882832e-07,
      "loss": 1.9642,
      "step": 228281
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.5107817649841309,
      "learning_rate": 1.2544859810158514e-07,
      "loss": 2.0522,
      "step": 228282
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.374112844467163,
      "learning_rate": 1.2533034228667803e-07,
      "loss": 1.8858,
      "step": 228283
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.4530744552612305,
      "learning_rate": 1.2521214222410703e-07,
      "loss": 1.9608,
      "step": 228284
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.4663151502609253,
      "learning_rate": 1.2509399791393871e-07,
      "loss": 1.8237,
      "step": 228285
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.472193956375122,
      "learning_rate": 1.2497590935620637e-07,
      "loss": 1.9382,
      "step": 228286
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.4751654863357544,
      "learning_rate": 1.2485787655084344e-07,
      "loss": 1.9896,
      "step": 228287
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.412610411643982,
      "learning_rate": 1.2473989949798313e-07,
      "loss": 1.9737,
      "step": 228288
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.437264323234558,
      "learning_rate": 1.246219781975588e-07,
      "loss": 1.8695,
      "step": 228289
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.425087332725525,
      "learning_rate": 1.245041126496371e-07,
      "loss": 2.2317,
      "step": 228290
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.5182007551193237,
      "learning_rate": 1.2438630285421803e-07,
      "loss": 2.1278,
      "step": 228291
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.5602695941925049,
      "learning_rate": 1.2426854881136816e-07,
      "loss": 2.0779,
      "step": 228292
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.5449817180633545,
      "learning_rate": 1.2415085052105422e-07,
      "loss": 1.9445,
      "step": 228293
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.351155400276184,
      "learning_rate": 1.240332079833095e-07,
      "loss": 2.0127,
      "step": 228294
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.4747315645217896,
      "learning_rate": 1.2391562119820064e-07,
      "loss": 2.0235,
      "step": 228295
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.486618161201477,
      "learning_rate": 1.237980901656943e-07,
      "loss": 2.0418,
      "step": 228296
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.370069980621338,
      "learning_rate": 1.236806148858571e-07,
      "loss": 2.0009,
      "step": 228297
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.4522727727890015,
      "learning_rate": 1.2356319535865577e-07,
      "loss": 1.8466,
      "step": 228298
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.5107661485671997,
      "learning_rate": 1.2344583158415688e-07,
      "loss": 1.7619,
      "step": 228299
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.608650803565979,
      "learning_rate": 1.2332852356239377e-07,
      "loss": 1.9507,
      "step": 228300
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.5644084215164185,
      "learning_rate": 1.2321127129333308e-07,
      "loss": 2.0166,
      "step": 228301
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.4572288990020752,
      "learning_rate": 1.2309407477704148e-07,
      "loss": 2.3963,
      "step": 228302
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.4512687921524048,
      "learning_rate": 1.2297693401351894e-07,
      "loss": 1.8808,
      "step": 228303
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.5529509782791138,
      "learning_rate": 1.2285984900279877e-07,
      "loss": 2.0754,
      "step": 228304
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.533349871635437,
      "learning_rate": 1.227428197449143e-07,
      "loss": 1.921,
      "step": 228305
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.4026165008544922,
      "learning_rate": 1.226258462398655e-07,
      "loss": 1.9865,
      "step": 228306
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.4469163417816162,
      "learning_rate": 1.2250892848765236e-07,
      "loss": 1.8522,
      "step": 228307
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.4049534797668457,
      "learning_rate": 1.2239206648837485e-07,
      "loss": 1.979,
      "step": 228308
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.4291099309921265,
      "learning_rate": 1.2227526024196634e-07,
      "loss": 1.9282,
      "step": 228309
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.5479103326797485,
      "learning_rate": 1.221585097484934e-07,
      "loss": 2.1507,
      "step": 228310
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.3540523052215576,
      "learning_rate": 1.2204181500798937e-07,
      "loss": 1.8218,
      "step": 228311
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.5076261758804321,
      "learning_rate": 1.2192517602045426e-07,
      "loss": 1.9508,
      "step": 228312
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.4941120147705078,
      "learning_rate": 1.2180859278588805e-07,
      "loss": 2.032,
      "step": 228313
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.4591140747070312,
      "learning_rate": 1.216920653043574e-07,
      "loss": 1.7666,
      "step": 228314
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.3822342157363892,
      "learning_rate": 1.2157559357586221e-07,
      "loss": 1.9137,
      "step": 228315
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.5045808553695679,
      "learning_rate": 1.2145917760043588e-07,
      "loss": 1.9105,
      "step": 228316
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.5679699182510376,
      "learning_rate": 1.213428173781117e-07,
      "loss": 1.9744,
      "step": 228317
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.5312473773956299,
      "learning_rate": 1.2122651290885633e-07,
      "loss": 1.9791,
      "step": 228318
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.4973093271255493,
      "learning_rate": 1.211102641927364e-07,
      "loss": 2.045,
      "step": 228319
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.3234026432037354,
      "learning_rate": 1.2099407122975191e-07,
      "loss": 1.9087,
      "step": 228320
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.4330800771713257,
      "learning_rate": 1.2087793401993617e-07,
      "loss": 2.1574,
      "step": 228321
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.31568443775177,
      "learning_rate": 1.207618525633225e-07,
      "loss": 2.0515,
      "step": 228322
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.4996328353881836,
      "learning_rate": 1.2064582685991088e-07,
      "loss": 1.9208,
      "step": 228323
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.4804013967514038,
      "learning_rate": 1.205298569097346e-07,
      "loss": 1.9517,
      "step": 228324
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.385364055633545,
      "learning_rate": 1.20413942712827e-07,
      "loss": 1.8755,
      "step": 228325
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.5069371461868286,
      "learning_rate": 1.2029808426915477e-07,
      "loss": 2.0348,
      "step": 228326
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.6638764142990112,
      "learning_rate": 1.2018228157881782e-07,
      "loss": 1.8891,
      "step": 228327
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.4367222785949707,
      "learning_rate": 1.2006653464178285e-07,
      "loss": 1.9238,
      "step": 228328
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.5214061737060547,
      "learning_rate": 1.1995084345811645e-07,
      "loss": 1.7864,
      "step": 228329
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.4723201990127563,
      "learning_rate": 1.1983520802778535e-07,
      "loss": 1.9674,
      "step": 228330
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.6881054639816284,
      "learning_rate": 1.1971962835082283e-07,
      "loss": 2.1981,
      "step": 228331
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.3624075651168823,
      "learning_rate": 1.196041044272955e-07,
      "loss": 2.0675,
      "step": 228332
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.4869775772094727,
      "learning_rate": 1.1948863625717008e-07,
      "loss": 2.1388,
      "step": 228333
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.441879153251648,
      "learning_rate": 1.1937322384051318e-07,
      "loss": 2.1848,
      "step": 228334
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.491834282875061,
      "learning_rate": 1.1925786717732477e-07,
      "loss": 2.0834,
      "step": 228335
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.6106566190719604,
      "learning_rate": 1.1914256626760488e-07,
      "loss": 1.8286,
      "step": 228336
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.395027995109558,
      "learning_rate": 1.190273211114201e-07,
      "loss": 1.922,
      "step": 228337
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.6377893686294556,
      "learning_rate": 1.1891213170873715e-07,
      "loss": 2.0022,
      "step": 228338
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.4637192487716675,
      "learning_rate": 1.1879699805965593e-07,
      "loss": 1.8365,
      "step": 228339
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.5617622137069702,
      "learning_rate": 1.1868192016410982e-07,
      "loss": 1.8802,
      "step": 228340
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.569443702697754,
      "learning_rate": 1.1856689802219877e-07,
      "loss": 1.8818,
      "step": 228341
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.3063362836837769,
      "learning_rate": 1.1845193163388944e-07,
      "loss": 2.0116,
      "step": 228342
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.4340574741363525,
      "learning_rate": 1.1833702099921516e-07,
      "loss": 1.8025,
      "step": 228343
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.466597557067871,
      "learning_rate": 1.1822216611820922e-07,
      "loss": 1.9442,
      "step": 228344
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.4615036249160767,
      "learning_rate": 1.1810736699087164e-07,
      "loss": 1.979,
      "step": 228345
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.519513487815857,
      "learning_rate": 1.1799262361723571e-07,
      "loss": 2.0543,
      "step": 228346
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.3706399202346802,
      "learning_rate": 1.1787793599733476e-07,
      "loss": 1.9666,
      "step": 228347
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.5452401638031006,
      "learning_rate": 1.1776330413120205e-07,
      "loss": 2.3343,
      "step": 228348
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.7016043663024902,
      "learning_rate": 1.1764872801883762e-07,
      "loss": 2.1901,
      "step": 228349
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.5160455703735352,
      "learning_rate": 1.1753420766024147e-07,
      "loss": 1.9419,
      "step": 228350
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.355945110321045,
      "learning_rate": 1.1741974305544688e-07,
      "loss": 1.8962,
      "step": 228351
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.401053547859192,
      "learning_rate": 1.1730533420448717e-07,
      "loss": 2.0125,
      "step": 228352
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.3960591554641724,
      "learning_rate": 1.1719098110739567e-07,
      "loss": 2.0109,
      "step": 228353
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.4918020963668823,
      "learning_rate": 1.1707668376420564e-07,
      "loss": 1.9146,
      "step": 228354
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.5101046562194824,
      "learning_rate": 1.1696244217488382e-07,
      "loss": 1.9254,
      "step": 228355
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.6371064186096191,
      "learning_rate": 1.1684825633946349e-07,
      "loss": 1.9643,
      "step": 228356
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.660943865776062,
      "learning_rate": 1.1673412625801126e-07,
      "loss": 1.9113,
      "step": 228357
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.4842885732650757,
      "learning_rate": 1.1662005193049384e-07,
      "loss": 2.0796,
      "step": 228358
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.3637642860412598,
      "learning_rate": 1.1650603335697784e-07,
      "loss": 1.9547,
      "step": 228359
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.4915063381195068,
      "learning_rate": 1.1639207053742994e-07,
      "loss": 1.9039,
      "step": 228360
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.479849100112915,
      "learning_rate": 1.1627816347195007e-07,
      "loss": 1.7083,
      "step": 228361
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.4434337615966797,
      "learning_rate": 1.1616431216050492e-07,
      "loss": 1.817,
      "step": 228362
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.568512201309204,
      "learning_rate": 1.160505166030945e-07,
      "loss": 1.9662,
      "step": 228363
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.56609308719635,
      "learning_rate": 1.1593677679978542e-07,
      "loss": 1.7616,
      "step": 228364
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.5425550937652588,
      "learning_rate": 1.1582309275061097e-07,
      "loss": 1.8419,
      "step": 228365
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.4077869653701782,
      "learning_rate": 1.1570946445553786e-07,
      "loss": 1.9006,
      "step": 228366
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.3333600759506226,
      "learning_rate": 1.155958919146327e-07,
      "loss": 1.8961,
      "step": 228367
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.558258056640625,
      "learning_rate": 1.1548237512789549e-07,
      "loss": 1.9821,
      "step": 228368
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.5669752359390259,
      "learning_rate": 1.1536891409532622e-07,
      "loss": 2.0012,
      "step": 228369
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.4316771030426025,
      "learning_rate": 1.1525550881699153e-07,
      "loss": 2.0804,
      "step": 228370
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.360081434249878,
      "learning_rate": 1.1514215929289139e-07,
      "loss": 2.1584,
      "step": 228371
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.5281561613082886,
      "learning_rate": 1.1502886552305913e-07,
      "loss": 1.9886,
      "step": 228372
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.4485327005386353,
      "learning_rate": 1.1491562750746141e-07,
      "loss": 2.1368,
      "step": 228373
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.4843167066574097,
      "learning_rate": 1.148024452461982e-07,
      "loss": 2.2042,
      "step": 228374
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.4185383319854736,
      "learning_rate": 1.1468931873923615e-07,
      "loss": 1.9714,
      "step": 228375
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.2989585399627686,
      "learning_rate": 1.1457624798664189e-07,
      "loss": 1.9396,
      "step": 228376
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.4624505043029785,
      "learning_rate": 1.1446323298838212e-07,
      "loss": 1.8463,
      "step": 228377
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.4354140758514404,
      "learning_rate": 1.1435027374452343e-07,
      "loss": 1.8374,
      "step": 228378
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.4931288957595825,
      "learning_rate": 1.1423737025506585e-07,
      "loss": 1.9059,
      "step": 228379
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.4318523406982422,
      "learning_rate": 1.1412452252000937e-07,
      "loss": 2.1042,
      "step": 228380
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.4268642663955688,
      "learning_rate": 1.1401173053942059e-07,
      "loss": 2.2276,
      "step": 228381
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.5654983520507812,
      "learning_rate": 1.1389899431329953e-07,
      "loss": 1.9462,
      "step": 228382
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.4633805751800537,
      "learning_rate": 1.1378631384164616e-07,
      "loss": 1.9862,
      "step": 228383
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.4120495319366455,
      "learning_rate": 1.1367368912449382e-07,
      "loss": 1.9523,
      "step": 228384
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.3490352630615234,
      "learning_rate": 1.1356112016190911e-07,
      "loss": 1.8837,
      "step": 228385
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.372365117073059,
      "learning_rate": 1.134486069538254e-07,
      "loss": 1.8339,
      "step": 228386
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.4616178274154663,
      "learning_rate": 1.1333614950034264e-07,
      "loss": 2.0736,
      "step": 228387
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.4118587970733643,
      "learning_rate": 1.1322374780146082e-07,
      "loss": 1.9206,
      "step": 228388
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.379062533378601,
      "learning_rate": 1.1311140185717992e-07,
      "loss": 2.0416,
      "step": 228389
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.4471477270126343,
      "learning_rate": 1.1299911166753328e-07,
      "loss": 1.9114,
      "step": 228390
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.6788451671600342,
      "learning_rate": 1.1288687723252088e-07,
      "loss": 2.0646,
      "step": 228391
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.5305404663085938,
      "learning_rate": 1.1277469855220933e-07,
      "loss": 2.0058,
      "step": 228392
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.4998096227645874,
      "learning_rate": 1.1266257562659863e-07,
      "loss": 1.9942,
      "step": 228393
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.4181863069534302,
      "learning_rate": 1.125505084556888e-07,
      "loss": 1.8674,
      "step": 228394
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.4441072940826416,
      "learning_rate": 1.1243849703951313e-07,
      "loss": 2.0492,
      "step": 228395
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.4110177755355835,
      "learning_rate": 1.1232654137810493e-07,
      "loss": 1.874,
      "step": 228396
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.5019950866699219,
      "learning_rate": 1.122146414714642e-07,
      "loss": 1.6649,
      "step": 228397
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.4911091327667236,
      "learning_rate": 1.1210279731962424e-07,
      "loss": 2.0354,
      "step": 228398
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.5316407680511475,
      "learning_rate": 1.1199100892261836e-07,
      "loss": 1.9053,
      "step": 228399
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.474623203277588,
      "learning_rate": 1.1187927628044657e-07,
      "loss": 2.1759,
      "step": 228400
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.6468076705932617,
      "learning_rate": 1.1176759939314218e-07,
      "loss": 1.8675,
      "step": 228401
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.4570752382278442,
      "learning_rate": 1.1165597826073846e-07,
      "loss": 1.8876,
      "step": 228402
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.4893237352371216,
      "learning_rate": 1.1154441288320215e-07,
      "loss": 2.1326,
      "step": 228403
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.4638537168502808,
      "learning_rate": 1.1143290326063314e-07,
      "loss": 1.7582,
      "step": 228404
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.4527217149734497,
      "learning_rate": 1.1132144939296484e-07,
      "loss": 1.9504,
      "step": 228405
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.4958446025848389,
      "learning_rate": 1.1121005128029714e-07,
      "loss": 1.9507,
      "step": 228406
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.581591010093689,
      "learning_rate": 1.1109870892259676e-07,
      "loss": 1.9318,
      "step": 228407
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.5294510126113892,
      "learning_rate": 1.10987422319897e-07,
      "loss": 1.9976,
      "step": 228408
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.5164852142333984,
      "learning_rate": 1.1087619147223115e-07,
      "loss": 2.0412,
      "step": 228409
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.5986146926879883,
      "learning_rate": 1.1076501637963254e-07,
      "loss": 1.7541,
      "step": 228410
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.58610999584198,
      "learning_rate": 1.1065389704210115e-07,
      "loss": 2.063,
      "step": 228411
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.5077555179595947,
      "learning_rate": 1.1054283345963699e-07,
      "loss": 1.9529,
      "step": 228412
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.59858238697052,
      "learning_rate": 1.1043182563230669e-07,
      "loss": 1.8787,
      "step": 228413
    },
    {
      "epoch": 2.97,
      "grad_norm": 2.954150438308716,
      "learning_rate": 1.103208735600769e-07,
      "loss": 2.1225,
      "step": 228414
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.5502560138702393,
      "learning_rate": 1.1020997724304758e-07,
      "loss": 1.8844,
      "step": 228415
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.5237345695495605,
      "learning_rate": 1.1009913668115211e-07,
      "loss": 2.0078,
      "step": 228416
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.4869675636291504,
      "learning_rate": 1.0998835187445709e-07,
      "loss": 1.8977,
      "step": 228417
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.45628821849823,
      "learning_rate": 1.0987762282299584e-07,
      "loss": 2.0879,
      "step": 228418
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.553156852722168,
      "learning_rate": 1.0976694952673504e-07,
      "loss": 2.1234,
      "step": 228419
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.6083606481552124,
      "learning_rate": 1.0965633198574131e-07,
      "loss": 2.0551,
      "step": 228420
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.5366863012313843,
      "learning_rate": 1.0954577020004796e-07,
      "loss": 2.0175,
      "step": 228421
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.4497063159942627,
      "learning_rate": 1.0943526416962167e-07,
      "loss": 2.1019,
      "step": 228422
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.3292491436004639,
      "learning_rate": 1.0932481389452907e-07,
      "loss": 1.9924,
      "step": 228423
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.495087742805481,
      "learning_rate": 1.0921441937477016e-07,
      "loss": 2.0576,
      "step": 228424
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.4687564373016357,
      "learning_rate": 1.0910408061037824e-07,
      "loss": 2.107,
      "step": 228425
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.44564950466156,
      "learning_rate": 1.089937976013533e-07,
      "loss": 1.8741,
      "step": 228426
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.3587554693222046,
      "learning_rate": 1.0888357034772865e-07,
      "loss": 1.9865,
      "step": 228427
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.6387089490890503,
      "learning_rate": 1.0877339884950432e-07,
      "loss": 1.9763,
      "step": 228428
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.5152983665466309,
      "learning_rate": 1.0866328310674688e-07,
      "loss": 1.955,
      "step": 228429
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.5321074724197388,
      "learning_rate": 1.0855322311945636e-07,
      "loss": 2.0439,
      "step": 228430
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.5286855697631836,
      "learning_rate": 1.0844321888763274e-07,
      "loss": 2.1282,
      "step": 228431
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.47780442237854,
      "learning_rate": 1.0833327041130935e-07,
      "loss": 1.9897,
      "step": 228432
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.3706574440002441,
      "learning_rate": 1.0822337769048616e-07,
      "loss": 1.8936,
      "step": 228433
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.4638488292694092,
      "learning_rate": 1.0811354072522982e-07,
      "loss": 1.7804,
      "step": 228434
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.5860650539398193,
      "learning_rate": 1.080037595155403e-07,
      "loss": 1.8393,
      "step": 228435
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.4798709154129028,
      "learning_rate": 1.0789403406141761e-07,
      "loss": 2.1131,
      "step": 228436
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.5100959539413452,
      "learning_rate": 1.0778436436292837e-07,
      "loss": 1.9192,
      "step": 228437
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.4757529497146606,
      "learning_rate": 1.0767475042003925e-07,
      "loss": 1.9446,
      "step": 228438
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.402825951576233,
      "learning_rate": 1.075651922327836e-07,
      "loss": 1.9918,
      "step": 228439
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.3762705326080322,
      "learning_rate": 1.0745568980122799e-07,
      "loss": 1.8785,
      "step": 228440
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.4647866487503052,
      "learning_rate": 1.0734624312530581e-07,
      "loss": 2.0082,
      "step": 228441
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.424686312675476,
      "learning_rate": 1.0723685220511702e-07,
      "loss": 1.8799,
      "step": 228442
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.4406524896621704,
      "learning_rate": 1.0712751704066158e-07,
      "loss": 2.1073,
      "step": 228443
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.5507850646972656,
      "learning_rate": 1.070182376319395e-07,
      "loss": 2.0736,
      "step": 228444
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.7408416271209717,
      "learning_rate": 1.0690901397901741e-07,
      "loss": 2.2128,
      "step": 228445
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.4740514755249023,
      "learning_rate": 1.0679984608182868e-07,
      "loss": 2.1409,
      "step": 228446
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.536268711090088,
      "learning_rate": 1.0669073394047322e-07,
      "loss": 2.0329,
      "step": 228447
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.5875879526138306,
      "learning_rate": 1.0658167755495107e-07,
      "loss": 1.9714,
      "step": 228448
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.6417617797851562,
      "learning_rate": 1.0647267692526218e-07,
      "loss": 2.0788,
      "step": 228449
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.6045080423355103,
      "learning_rate": 1.0636373205147319e-07,
      "loss": 1.9808,
      "step": 228450
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.5516371726989746,
      "learning_rate": 1.062548429335508e-07,
      "loss": 1.8912,
      "step": 228451
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.5748056173324585,
      "learning_rate": 1.061460095715283e-07,
      "loss": 2.023,
      "step": 228452
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.4878332614898682,
      "learning_rate": 1.0603723196543901e-07,
      "loss": 2.159,
      "step": 228453
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.4645957946777344,
      "learning_rate": 1.0592851011528292e-07,
      "loss": 1.8442,
      "step": 228454
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.4332963228225708,
      "learning_rate": 1.0581984402112664e-07,
      "loss": 1.961,
      "step": 228455
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.3386062383651733,
      "learning_rate": 1.0571123368293688e-07,
      "loss": 2.1433,
      "step": 228456
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.4369330406188965,
      "learning_rate": 1.0560267910074693e-07,
      "loss": 1.7816,
      "step": 228457
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.5341424942016602,
      "learning_rate": 1.0549418027462342e-07,
      "loss": 1.9847,
      "step": 228458
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.4827759265899658,
      "learning_rate": 1.0538573720453303e-07,
      "loss": 2.0527,
      "step": 228459
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.5586947202682495,
      "learning_rate": 1.0527734989047576e-07,
      "loss": 1.8857,
      "step": 228460
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.5324807167053223,
      "learning_rate": 1.0516901833255153e-07,
      "loss": 2.0033,
      "step": 228461
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.7889665365219116,
      "learning_rate": 1.0506074253072704e-07,
      "loss": 2.0159,
      "step": 228462
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.4984703063964844,
      "learning_rate": 1.049525224850356e-07,
      "loss": 1.99,
      "step": 228463
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.5288175344467163,
      "learning_rate": 1.048443581954772e-07,
      "loss": 2.0521,
      "step": 228464
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.4508984088897705,
      "learning_rate": 1.0473624966211846e-07,
      "loss": 2.145,
      "step": 228465
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.4031788110733032,
      "learning_rate": 1.0462819688492608e-07,
      "loss": 1.6802,
      "step": 228466
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.359145998954773,
      "learning_rate": 1.0452019986396665e-07,
      "loss": 1.9138,
      "step": 228467
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.5278983116149902,
      "learning_rate": 1.0441225859920687e-07,
      "loss": 2.0309,
      "step": 228468
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.461082100868225,
      "learning_rate": 1.0430437309071337e-07,
      "loss": 2.0791,
      "step": 228469
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.4622843265533447,
      "learning_rate": 1.0419654333851946e-07,
      "loss": 1.9311,
      "step": 228470
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.4696240425109863,
      "learning_rate": 1.040887693425585e-07,
      "loss": 2.1028,
      "step": 228471
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.6204818487167358,
      "learning_rate": 1.0398105110296373e-07,
      "loss": 1.987,
      "step": 228472
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.370627522468567,
      "learning_rate": 1.0387338861966854e-07,
      "loss": 2.0662,
      "step": 228473
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.3948450088500977,
      "learning_rate": 1.0376578189273954e-07,
      "loss": 1.9282,
      "step": 228474
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.5040546655654907,
      "learning_rate": 1.0365823092214342e-07,
      "loss": 1.9448,
      "step": 228475
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.5084621906280518,
      "learning_rate": 1.0355073570798012e-07,
      "loss": 2.1798,
      "step": 228476
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.385101556777954,
      "learning_rate": 1.034432962502163e-07,
      "loss": 2.1812,
      "step": 228477
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.4869868755340576,
      "learning_rate": 1.033359125488853e-07,
      "loss": 2.0233,
      "step": 228478
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.5235087871551514,
      "learning_rate": 1.0322858460398709e-07,
      "loss": 2.1574,
      "step": 228479
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.5308306217193604,
      "learning_rate": 1.0312131241558829e-07,
      "loss": 1.9828,
      "step": 228480
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.4806104898452759,
      "learning_rate": 1.0301409598365562e-07,
      "loss": 1.8643,
      "step": 228481
    },
    {
      "epoch": 2.97,
      "grad_norm": 1.697054386138916,
      "learning_rate": 1.0290693530825567e-07,
      "loss": 2.1158,
      "step": 228482
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.4141428470611572,
      "learning_rate": 1.0279983038938844e-07,
      "loss": 1.9405,
      "step": 228483
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.3814806938171387,
      "learning_rate": 1.0269278122705394e-07,
      "loss": 1.9733,
      "step": 228484
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.6938579082489014,
      "learning_rate": 1.0258578782128546e-07,
      "loss": 2.089,
      "step": 228485
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.448838472366333,
      "learning_rate": 1.0247885017211633e-07,
      "loss": 1.8008,
      "step": 228486
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.5162674188613892,
      "learning_rate": 1.0237196827954652e-07,
      "loss": 1.8411,
      "step": 228487
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.4138892889022827,
      "learning_rate": 1.0226514214364267e-07,
      "loss": 1.8688,
      "step": 228488
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.4556121826171875,
      "learning_rate": 1.0215837176433817e-07,
      "loss": 2.0013,
      "step": 228489
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.3711153268814087,
      "learning_rate": 1.0205165714173292e-07,
      "loss": 2.0662,
      "step": 228490
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.4089176654815674,
      "learning_rate": 1.0194499827582692e-07,
      "loss": 2.0336,
      "step": 228491
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.4915846586227417,
      "learning_rate": 1.0183839516662017e-07,
      "loss": 1.9777,
      "step": 228492
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.357801079750061,
      "learning_rate": 1.0173184781414601e-07,
      "loss": 1.8585,
      "step": 228493
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.4670888185501099,
      "learning_rate": 1.016253562184044e-07,
      "loss": 1.9985,
      "step": 228494
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.468643069267273,
      "learning_rate": 1.0151892037942866e-07,
      "loss": 1.9069,
      "step": 228495
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.5368685722351074,
      "learning_rate": 1.014125402972521e-07,
      "loss": 1.9043,
      "step": 228496
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.3979047536849976,
      "learning_rate": 1.0130621597190802e-07,
      "loss": 1.7622,
      "step": 228497
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.618330717086792,
      "learning_rate": 1.0119994740336312e-07,
      "loss": 1.969,
      "step": 228498
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.5204088687896729,
      "learning_rate": 1.0109373459168402e-07,
      "loss": 2.163,
      "step": 228499
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.51864492893219,
      "learning_rate": 1.0098757753683739e-07,
      "loss": 1.798,
      "step": 228500
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.4877675771713257,
      "learning_rate": 1.0088147623892317e-07,
      "loss": 2.1302,
      "step": 228501
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.5015548467636108,
      "learning_rate": 1.0077543069787475e-07,
      "loss": 2.0539,
      "step": 228502
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.402241826057434,
      "learning_rate": 1.0066944091379203e-07,
      "loss": 1.7872,
      "step": 228503
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.3789443969726562,
      "learning_rate": 1.0056350688664172e-07,
      "loss": 1.7243,
      "step": 228504
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.5044277906417847,
      "learning_rate": 1.0045762861645712e-07,
      "loss": 1.8869,
      "step": 228505
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.4390959739685059,
      "learning_rate": 1.0035180610323822e-07,
      "loss": 1.8642,
      "step": 228506
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.4404937028884888,
      "learning_rate": 1.0024603934705167e-07,
      "loss": 2.1068,
      "step": 228507
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.3123323917388916,
      "learning_rate": 1.0014032834786412e-07,
      "loss": 1.7812,
      "step": 228508
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.5081416368484497,
      "learning_rate": 1.000346731057422e-07,
      "loss": 2.0994,
      "step": 228509
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.6730413436889648,
      "learning_rate": 9.992907362068591e-08,
      "loss": 2.1185,
      "step": 228510
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.514827847480774,
      "learning_rate": 9.982352989269527e-08,
      "loss": 1.9361,
      "step": 228511
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.518423080444336,
      "learning_rate": 9.971804192183686e-08,
      "loss": 1.919,
      "step": 228512
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.4638410806655884,
      "learning_rate": 9.961260970807738e-08,
      "loss": 1.8672,
      "step": 228513
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.5052497386932373,
      "learning_rate": 9.950723325148346e-08,
      "loss": 1.9786,
      "step": 228514
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.6067591905593872,
      "learning_rate": 9.940191255202179e-08,
      "loss": 1.9703,
      "step": 228515
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.5031497478485107,
      "learning_rate": 9.929664760975898e-08,
      "loss": 1.9717,
      "step": 228516
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.5012240409851074,
      "learning_rate": 9.919143842472832e-08,
      "loss": 1.9206,
      "step": 228517
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.4983501434326172,
      "learning_rate": 9.908628499686322e-08,
      "loss": 1.9111,
      "step": 228518
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.449999451637268,
      "learning_rate": 9.89811873262969e-08,
      "loss": 2.0499,
      "step": 228519
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.4065372943878174,
      "learning_rate": 9.887614541292944e-08,
      "loss": 2.0704,
      "step": 228520
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.429598331451416,
      "learning_rate": 9.877115925689405e-08,
      "loss": 1.9649,
      "step": 228521
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.5070948600769043,
      "learning_rate": 9.866622885812414e-08,
      "loss": 2.2581,
      "step": 228522
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.4162205457687378,
      "learning_rate": 9.856135421668632e-08,
      "loss": 2.1089,
      "step": 228523
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.4225867986679077,
      "learning_rate": 9.845653533258058e-08,
      "loss": 2.0857,
      "step": 228524
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.5634946823120117,
      "learning_rate": 9.835177220584023e-08,
      "loss": 1.9329,
      "step": 228525
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.4678696393966675,
      "learning_rate": 9.824706483649858e-08,
      "loss": 1.8991,
      "step": 228526
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.3986800909042358,
      "learning_rate": 9.814241322452232e-08,
      "loss": 1.9682,
      "step": 228527
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.4627346992492676,
      "learning_rate": 9.803781736997807e-08,
      "loss": 1.9876,
      "step": 228528
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.3370603322982788,
      "learning_rate": 9.793327727283251e-08,
      "loss": 1.9148,
      "step": 228529
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.6826071739196777,
      "learning_rate": 9.782879293318558e-08,
      "loss": 2.0501,
      "step": 228530
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.5478562116622925,
      "learning_rate": 9.772436435100395e-08,
      "loss": 1.9104,
      "step": 228531
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.4645726680755615,
      "learning_rate": 9.761999152632095e-08,
      "loss": 2.0894,
      "step": 228532
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.3876922130584717,
      "learning_rate": 9.751567445913654e-08,
      "loss": 2.0181,
      "step": 228533
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.3223083019256592,
      "learning_rate": 9.741141314948408e-08,
      "loss": 2.1389,
      "step": 228534
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.527570366859436,
      "learning_rate": 9.730720759739685e-08,
      "loss": 2.1616,
      "step": 228535
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.5360705852508545,
      "learning_rate": 9.720305780287485e-08,
      "loss": 2.0157,
      "step": 228536
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.429185152053833,
      "learning_rate": 9.709896376595139e-08,
      "loss": 1.7446,
      "step": 228537
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.4071903228759766,
      "learning_rate": 9.699492548662646e-08,
      "loss": 2.0931,
      "step": 228538
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.4053517580032349,
      "learning_rate": 9.689094296493339e-08,
      "loss": 2.0969,
      "step": 228539
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.4269522428512573,
      "learning_rate": 9.678701620090545e-08,
      "loss": 2.0814,
      "step": 228540
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.4781471490859985,
      "learning_rate": 9.668314519454268e-08,
      "loss": 1.778,
      "step": 228541
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.362988829612732,
      "learning_rate": 9.657932994584505e-08,
      "loss": 2.008,
      "step": 228542
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.4534929990768433,
      "learning_rate": 9.647557045487919e-08,
      "loss": 2.181,
      "step": 228543
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.4260400533676147,
      "learning_rate": 9.637186672161179e-08,
      "loss": 2.0434,
      "step": 228544
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.5026296377182007,
      "learning_rate": 9.626821874610946e-08,
      "loss": 1.9314,
      "step": 228545
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.4261910915374756,
      "learning_rate": 9.61646265283722e-08,
      "loss": 1.8662,
      "step": 228546
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.3673900365829468,
      "learning_rate": 9.606109006840001e-08,
      "loss": 1.6098,
      "step": 228547
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.5425316095352173,
      "learning_rate": 9.595760936625951e-08,
      "loss": 1.9876,
      "step": 228548
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.6809852123260498,
      "learning_rate": 9.585418442191739e-08,
      "loss": 1.9919,
      "step": 228549
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.42586350440979,
      "learning_rate": 9.575081523544026e-08,
      "loss": 2.1027,
      "step": 228550
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.5665998458862305,
      "learning_rate": 9.564750180679482e-08,
      "loss": 1.9338,
      "step": 228551
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.449126124382019,
      "learning_rate": 9.554424413608097e-08,
      "loss": 1.9892,
      "step": 228552
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.5881251096725464,
      "learning_rate": 9.544104222323213e-08,
      "loss": 2.0011,
      "step": 228553
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.3972525596618652,
      "learning_rate": 9.533789606828157e-08,
      "loss": 1.8584,
      "step": 228554
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.5016933679580688,
      "learning_rate": 9.523480567129594e-08,
      "loss": 2.0994,
      "step": 228555
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.4475839138031006,
      "learning_rate": 9.513177103227521e-08,
      "loss": 1.9977,
      "step": 228556
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.4710742235183716,
      "learning_rate": 9.502879215121939e-08,
      "loss": 1.8364,
      "step": 228557
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.4626989364624023,
      "learning_rate": 9.492586902812848e-08,
      "loss": 1.8615,
      "step": 228558
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.4510953426361084,
      "learning_rate": 9.482300166310242e-08,
      "loss": 1.968,
      "step": 228559
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.5832854509353638,
      "learning_rate": 9.472019005607457e-08,
      "loss": 1.9991,
      "step": 228560
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.4545555114746094,
      "learning_rate": 9.461743420711154e-08,
      "loss": 1.9326,
      "step": 228561
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.4815256595611572,
      "learning_rate": 9.451473411621335e-08,
      "loss": 2.0328,
      "step": 228562
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.3446190357208252,
      "learning_rate": 9.44120897834133e-08,
      "loss": 1.9846,
      "step": 228563
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.4744243621826172,
      "learning_rate": 9.43095012087447e-08,
      "loss": 1.7667,
      "step": 228564
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.392871379852295,
      "learning_rate": 9.420696839217424e-08,
      "loss": 1.8113,
      "step": 228565
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.6661797761917114,
      "learning_rate": 9.410449133376852e-08,
      "loss": 2.182,
      "step": 228566
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.5066958665847778,
      "learning_rate": 9.400207003352755e-08,
      "loss": 1.9637,
      "step": 228567
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.4301190376281738,
      "learning_rate": 9.389970449145135e-08,
      "loss": 2.0143,
      "step": 228568
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.4191241264343262,
      "learning_rate": 9.37973947076065e-08,
      "loss": 1.8641,
      "step": 228569
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.4694949388504028,
      "learning_rate": 9.369514068195971e-08,
      "loss": 2.1505,
      "step": 228570
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.4341574907302856,
      "learning_rate": 9.35929424145776e-08,
      "loss": 1.9855,
      "step": 228571
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.4445759057998657,
      "learning_rate": 9.349079990546015e-08,
      "loss": 2.0103,
      "step": 228572
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.4989792108535767,
      "learning_rate": 9.338871315460738e-08,
      "loss": 1.9517,
      "step": 228573
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.363363265991211,
      "learning_rate": 9.32866821620859e-08,
      "loss": 1.9847,
      "step": 228574
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.5047050714492798,
      "learning_rate": 9.318470692786239e-08,
      "loss": 2.1652,
      "step": 228575
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.4772469997406006,
      "learning_rate": 9.308278745197017e-08,
      "loss": 1.9078,
      "step": 228576
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.4501668214797974,
      "learning_rate": 9.298092373444255e-08,
      "loss": 2.0964,
      "step": 228577
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.4854992628097534,
      "learning_rate": 9.287911577527952e-08,
      "loss": 2.01,
      "step": 228578
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.6095805168151855,
      "learning_rate": 9.277736357451437e-08,
      "loss": 2.0587,
      "step": 228579
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.4165526628494263,
      "learning_rate": 9.267566713218044e-08,
      "loss": 1.905,
      "step": 228580
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.4285355806350708,
      "learning_rate": 9.257402644827772e-08,
      "loss": 2.0056,
      "step": 228581
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.4264416694641113,
      "learning_rate": 9.24724415228062e-08,
      "loss": 1.8064,
      "step": 228582
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.5071077346801758,
      "learning_rate": 9.237091235583249e-08,
      "loss": 1.7312,
      "step": 228583
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.5152626037597656,
      "learning_rate": 9.226943894732331e-08,
      "loss": 2.0563,
      "step": 228584
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.4804775714874268,
      "learning_rate": 9.216802129734524e-08,
      "loss": 1.9048,
      "step": 228585
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.5068557262420654,
      "learning_rate": 9.206665940586499e-08,
      "loss": 2.1294,
      "step": 228586
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.574407935142517,
      "learning_rate": 9.196535327294918e-08,
      "loss": 1.8776,
      "step": 228587
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.6258634328842163,
      "learning_rate": 9.18641028985978e-08,
      "loss": 1.9064,
      "step": 228588
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.4172767400741577,
      "learning_rate": 9.176290828284416e-08,
      "loss": 2.1417,
      "step": 228589
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.5959171056747437,
      "learning_rate": 9.166176942565495e-08,
      "loss": 1.809,
      "step": 228590
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.490071177482605,
      "learning_rate": 9.15606863271301e-08,
      "loss": 2.0138,
      "step": 228591
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.5357905626296997,
      "learning_rate": 9.145965898720298e-08,
      "loss": 2.2492,
      "step": 228592
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.4696075916290283,
      "learning_rate": 9.135868740597352e-08,
      "loss": 2.0746,
      "step": 228593
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.487894058227539,
      "learning_rate": 9.125777158340841e-08,
      "loss": 2.0336,
      "step": 228594
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.3806469440460205,
      "learning_rate": 9.115691151954097e-08,
      "loss": 2.1369,
      "step": 228595
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.5277155637741089,
      "learning_rate": 9.105610721437117e-08,
      "loss": 1.9329,
      "step": 228596
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.4090044498443604,
      "learning_rate": 9.095535866796567e-08,
      "loss": 2.0096,
      "step": 228597
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.500229001045227,
      "learning_rate": 9.08546658802911e-08,
      "loss": 1.9925,
      "step": 228598
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.4841917753219604,
      "learning_rate": 9.075402885138084e-08,
      "loss": 1.9484,
      "step": 228599
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.4086098670959473,
      "learning_rate": 9.065344758126813e-08,
      "loss": 2.059,
      "step": 228600
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.392074704170227,
      "learning_rate": 9.055292206995302e-08,
      "loss": 1.9354,
      "step": 228601
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.3388112783432007,
      "learning_rate": 9.045245231750209e-08,
      "loss": 1.8015,
      "step": 228602
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.4736902713775635,
      "learning_rate": 9.035203832388205e-08,
      "loss": 1.8064,
      "step": 228603
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.4441816806793213,
      "learning_rate": 9.02516800890929e-08,
      "loss": 2.0498,
      "step": 228604
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.524060845375061,
      "learning_rate": 9.015137761323454e-08,
      "loss": 2.0512,
      "step": 228605
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.496686339378357,
      "learning_rate": 9.005113089624039e-08,
      "loss": 1.9458,
      "step": 228606
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.4722867012023926,
      "learning_rate": 8.995093993821034e-08,
      "loss": 1.8913,
      "step": 228607
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.4109792709350586,
      "learning_rate": 8.98508047390778e-08,
      "loss": 2.1307,
      "step": 228608
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.65657639503479,
      "learning_rate": 8.975072529890937e-08,
      "loss": 1.7983,
      "step": 228609
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.5348308086395264,
      "learning_rate": 8.965070161773835e-08,
      "loss": 2.0777,
      "step": 228610
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.5749973058700562,
      "learning_rate": 8.955073369556476e-08,
      "loss": 2.1521,
      "step": 228611
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.3270025253295898,
      "learning_rate": 8.945082153238858e-08,
      "loss": 2.0137,
      "step": 228612
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.5432524681091309,
      "learning_rate": 8.935096512824314e-08,
      "loss": 2.0473,
      "step": 228613
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.826150894165039,
      "learning_rate": 8.925116448316172e-08,
      "loss": 2.0652,
      "step": 228614
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.4393161535263062,
      "learning_rate": 8.915141959711103e-08,
      "loss": 1.9224,
      "step": 228615
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.3717186450958252,
      "learning_rate": 8.905173047019099e-08,
      "loss": 1.9639,
      "step": 228616
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.530432939529419,
      "learning_rate": 8.895209710236828e-08,
      "loss": 2.121,
      "step": 228617
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.4978128671646118,
      "learning_rate": 8.885251949367622e-08,
      "loss": 2.0981,
      "step": 228618
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.4876136779785156,
      "learning_rate": 8.875299764411481e-08,
      "loss": 2.0281,
      "step": 228619
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.413839340209961,
      "learning_rate": 8.865353155371735e-08,
      "loss": 1.9712,
      "step": 228620
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.4070146083831787,
      "learning_rate": 8.855412122248385e-08,
      "loss": 2.0119,
      "step": 228621
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.3602617979049683,
      "learning_rate": 8.84547666504809e-08,
      "loss": 1.9643,
      "step": 228622
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.4668995141983032,
      "learning_rate": 8.835546783767522e-08,
      "loss": 1.9114,
      "step": 228623
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.4168293476104736,
      "learning_rate": 8.825622478413342e-08,
      "loss": 1.772,
      "step": 228624
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.6790165901184082,
      "learning_rate": 8.815703748982217e-08,
      "loss": 2.0267,
      "step": 228625
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.5894793272018433,
      "learning_rate": 8.80579059547748e-08,
      "loss": 1.9944,
      "step": 228626
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.4076870679855347,
      "learning_rate": 8.795883017902461e-08,
      "loss": 2.0614,
      "step": 228627
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.523482084274292,
      "learning_rate": 8.785981016260491e-08,
      "loss": 1.91,
      "step": 228628
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.4480574131011963,
      "learning_rate": 8.776084590551568e-08,
      "loss": 1.9504,
      "step": 228629
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.3077374696731567,
      "learning_rate": 8.766193740772365e-08,
      "loss": 1.9478,
      "step": 228630
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.5409713983535767,
      "learning_rate": 8.756308466936202e-08,
      "loss": 2.1061,
      "step": 228631
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.4693151712417603,
      "learning_rate": 8.746428769033088e-08,
      "loss": 2.0349,
      "step": 228632
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.4215189218521118,
      "learning_rate": 8.736554647073013e-08,
      "loss": 1.9609,
      "step": 228633
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.4751383066177368,
      "learning_rate": 8.726686101052648e-08,
      "loss": 1.9615,
      "step": 228634
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.3797580003738403,
      "learning_rate": 8.716823130978656e-08,
      "loss": 2.0088,
      "step": 228635
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.337166428565979,
      "learning_rate": 8.706965736847704e-08,
      "loss": 2.1182,
      "step": 228636
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.4333271980285645,
      "learning_rate": 8.697113918666453e-08,
      "loss": 1.7812,
      "step": 228637
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.4732394218444824,
      "learning_rate": 8.687267676434906e-08,
      "loss": 2.3525,
      "step": 228638
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.4105236530303955,
      "learning_rate": 8.677427010153059e-08,
      "loss": 1.9311,
      "step": 228639
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.4146842956542969,
      "learning_rate": 8.667591919824246e-08,
      "loss": 1.9208,
      "step": 228640
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.3567310571670532,
      "learning_rate": 8.657762405451796e-08,
      "loss": 1.8926,
      "step": 228641
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.493528127670288,
      "learning_rate": 8.647938467035709e-08,
      "loss": 1.774,
      "step": 228642
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.4663712978363037,
      "learning_rate": 8.638120104575986e-08,
      "loss": 1.8966,
      "step": 228643
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.6062945127487183,
      "learning_rate": 8.628307318079286e-08,
      "loss": 2.0591,
      "step": 228644
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.7344591617584229,
      "learning_rate": 8.618500107542281e-08,
      "loss": 2.2455,
      "step": 228645
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.6852487325668335,
      "learning_rate": 8.608698472971631e-08,
      "loss": 2.0221,
      "step": 228646
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.4483999013900757,
      "learning_rate": 8.598902414364006e-08,
      "loss": 1.8793,
      "step": 228647
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.4202548265457153,
      "learning_rate": 8.589111931726067e-08,
      "loss": 1.9329,
      "step": 228648
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.4068958759307861,
      "learning_rate": 8.579327025054483e-08,
      "loss": 2.0048,
      "step": 228649
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.5436594486236572,
      "learning_rate": 8.569547694359247e-08,
      "loss": 1.8608,
      "step": 228650
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.5184694528579712,
      "learning_rate": 8.559773939633696e-08,
      "loss": 2.0598,
      "step": 228651
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.4945889711380005,
      "learning_rate": 8.550005760884493e-08,
      "loss": 2.1722,
      "step": 228652
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.5667533874511719,
      "learning_rate": 8.540243158108307e-08,
      "loss": 1.889,
      "step": 228653
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.4561315774917603,
      "learning_rate": 8.530486131315129e-08,
      "loss": 1.9326,
      "step": 228654
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.3993829488754272,
      "learning_rate": 8.520734680498298e-08,
      "loss": 1.8127,
      "step": 228655
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.6486057043075562,
      "learning_rate": 8.510988805667806e-08,
      "loss": 2.074,
      "step": 228656
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.4137269258499146,
      "learning_rate": 8.501248506820324e-08,
      "loss": 2.0179,
      "step": 228657
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.530419111251831,
      "learning_rate": 8.49151378395585e-08,
      "loss": 2.0381,
      "step": 228658
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.3802684545516968,
      "learning_rate": 8.481784637081047e-08,
      "loss": 1.9048,
      "step": 228659
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.5016193389892578,
      "learning_rate": 8.472061066195912e-08,
      "loss": 2.1559,
      "step": 228660
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.412112832069397,
      "learning_rate": 8.462343071300448e-08,
      "loss": 2.0455,
      "step": 228661
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.5516422986984253,
      "learning_rate": 8.452630652397985e-08,
      "loss": 2.0156,
      "step": 228662
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.684398889541626,
      "learning_rate": 8.442923809491852e-08,
      "loss": 2.0106,
      "step": 228663
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.3332477807998657,
      "learning_rate": 8.433222542582052e-08,
      "loss": 1.7343,
      "step": 228664
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.3573333024978638,
      "learning_rate": 8.423526851668583e-08,
      "loss": 2.1384,
      "step": 228665
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.5095272064208984,
      "learning_rate": 8.413836736758105e-08,
      "loss": 1.8497,
      "step": 228666
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.5017503499984741,
      "learning_rate": 8.404152197847291e-08,
      "loss": 2.2542,
      "step": 228667
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.378389596939087,
      "learning_rate": 8.3944732349428e-08,
      "loss": 2.0657,
      "step": 228668
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.7041682004928589,
      "learning_rate": 8.384799848041302e-08,
      "loss": 2.2498,
      "step": 228669
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.3808321952819824,
      "learning_rate": 8.375132037149457e-08,
      "loss": 2.083,
      "step": 228670
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.5083436965942383,
      "learning_rate": 8.365469802267266e-08,
      "loss": 2.2051,
      "step": 228671
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.4476079940795898,
      "learning_rate": 8.355813143391399e-08,
      "loss": 2.0294,
      "step": 228672
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.452793002128601,
      "learning_rate": 8.346162060531847e-08,
      "loss": 1.8917,
      "step": 228673
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.4837727546691895,
      "learning_rate": 8.33651655368861e-08,
      "loss": 1.9187,
      "step": 228674
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.4294006824493408,
      "learning_rate": 8.326876622858358e-08,
      "loss": 1.7991,
      "step": 228675
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.5554519891738892,
      "learning_rate": 8.317242268047752e-08,
      "loss": 2.0842,
      "step": 228676
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.4356778860092163,
      "learning_rate": 8.307613489256792e-08,
      "loss": 1.9485,
      "step": 228677
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.4787588119506836,
      "learning_rate": 8.297990286488809e-08,
      "loss": 1.9415,
      "step": 228678
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.4269520044326782,
      "learning_rate": 8.288372659743803e-08,
      "loss": 1.9968,
      "step": 228679
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.5156267881393433,
      "learning_rate": 8.278760609021773e-08,
      "loss": 2.1057,
      "step": 228680
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.5056211948394775,
      "learning_rate": 8.269154134329381e-08,
      "loss": 1.9277,
      "step": 228681
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.4251748323440552,
      "learning_rate": 8.259553235663297e-08,
      "loss": 2.0126,
      "step": 228682
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.6560068130493164,
      "learning_rate": 8.24995791303018e-08,
      "loss": 2.2369,
      "step": 228683
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.3976715803146362,
      "learning_rate": 8.240368166430034e-08,
      "loss": 2.0357,
      "step": 228684
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.4588465690612793,
      "learning_rate": 8.230783995862856e-08,
      "loss": 1.9901,
      "step": 228685
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.649646282196045,
      "learning_rate": 8.221205401331976e-08,
      "loss": 2.0546,
      "step": 228686
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.6027405261993408,
      "learning_rate": 8.211632382840727e-08,
      "loss": 1.8445,
      "step": 228687
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.5005850791931152,
      "learning_rate": 8.202064940385778e-08,
      "loss": 1.9746,
      "step": 228688
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.458685278892517,
      "learning_rate": 8.192503073973788e-08,
      "loss": 1.9441,
      "step": 228689
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.598305344581604,
      "learning_rate": 8.18294678360476e-08,
      "loss": 1.9131,
      "step": 228690
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.4998050928115845,
      "learning_rate": 8.173396069282023e-08,
      "loss": 2.0305,
      "step": 228691
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.4817581176757812,
      "learning_rate": 8.163850931005577e-08,
      "loss": 2.0864,
      "step": 228692
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.4496089220046997,
      "learning_rate": 8.154311368775424e-08,
      "loss": 2.1971,
      "step": 228693
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.546137809753418,
      "learning_rate": 8.144777382598222e-08,
      "loss": 1.846,
      "step": 228694
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.443642497062683,
      "learning_rate": 8.135248972470642e-08,
      "loss": 1.8333,
      "step": 228695
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.4530072212219238,
      "learning_rate": 8.125726138399347e-08,
      "loss": 1.9825,
      "step": 228696
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.3793811798095703,
      "learning_rate": 8.116208880381003e-08,
      "loss": 2.0675,
      "step": 228697
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.4233802556991577,
      "learning_rate": 8.106697198422274e-08,
      "loss": 1.7925,
      "step": 228698
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.5385873317718506,
      "learning_rate": 8.097191092523158e-08,
      "loss": 2.1043,
      "step": 228699
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.4885715246200562,
      "learning_rate": 8.087690562680327e-08,
      "loss": 2.0731,
      "step": 228700
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.6172549724578857,
      "learning_rate": 8.078195608903771e-08,
      "loss": 2.1526,
      "step": 228701
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.4247264862060547,
      "learning_rate": 8.068706231193489e-08,
      "loss": 1.9882,
      "step": 228702
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.5017292499542236,
      "learning_rate": 8.059222429546153e-08,
      "loss": 2.1435,
      "step": 228703
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.5023497343063354,
      "learning_rate": 8.049744203968422e-08,
      "loss": 2.0351,
      "step": 228704
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.4528579711914062,
      "learning_rate": 8.040271554456968e-08,
      "loss": 1.9659,
      "step": 228705
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.6020950078964233,
      "learning_rate": 8.030804481021779e-08,
      "loss": 2.1051,
      "step": 228706
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.4553018808364868,
      "learning_rate": 8.021342983656198e-08,
      "loss": 1.9924,
      "step": 228707
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.6401430368423462,
      "learning_rate": 8.011887062366885e-08,
      "loss": 1.9273,
      "step": 228708
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.5284746885299683,
      "learning_rate": 8.002436717153837e-08,
      "loss": 2.1433,
      "step": 228709
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.4410470724105835,
      "learning_rate": 7.992991948020389e-08,
      "loss": 2.1652,
      "step": 228710
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.4642137289047241,
      "learning_rate": 7.983552754966537e-08,
      "loss": 1.9029,
      "step": 228711
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.587122917175293,
      "learning_rate": 7.974119137995615e-08,
      "loss": 2.0731,
      "step": 228712
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.4147710800170898,
      "learning_rate": 7.964691097107623e-08,
      "loss": 1.9708,
      "step": 228713
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.422963261604309,
      "learning_rate": 7.955268632305889e-08,
      "loss": 1.9165,
      "step": 228714
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.5659252405166626,
      "learning_rate": 7.945851743590415e-08,
      "loss": 2.058,
      "step": 228715
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.5280303955078125,
      "learning_rate": 7.93644043096453e-08,
      "loss": 2.2245,
      "step": 228716
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.6053861379623413,
      "learning_rate": 7.927034694428235e-08,
      "loss": 2.2661,
      "step": 228717
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.3880712985992432,
      "learning_rate": 7.917634533984862e-08,
      "loss": 2.0111,
      "step": 228718
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.5638539791107178,
      "learning_rate": 7.90823994963774e-08,
      "loss": 2.0629,
      "step": 228719
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.3421567678451538,
      "learning_rate": 7.898850941383539e-08,
      "loss": 2.0201,
      "step": 228720
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.4498800039291382,
      "learning_rate": 7.88946750922892e-08,
      "loss": 1.9277,
      "step": 228721
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.394662618637085,
      "learning_rate": 7.880089653173883e-08,
      "loss": 1.9209,
      "step": 228722
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.428011178970337,
      "learning_rate": 7.870717373218428e-08,
      "loss": 2.2563,
      "step": 228723
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.4136558771133423,
      "learning_rate": 7.861350669369215e-08,
      "loss": 1.8952,
      "step": 228724
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.3351373672485352,
      "learning_rate": 7.851989541622917e-08,
      "loss": 1.937,
      "step": 228725
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.4498273134231567,
      "learning_rate": 7.84263398998286e-08,
      "loss": 1.985,
      "step": 228726
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.4037039279937744,
      "learning_rate": 7.833284014449049e-08,
      "loss": 2.0164,
      "step": 228727
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.4779354333877563,
      "learning_rate": 7.82393961502814e-08,
      "loss": 2.0315,
      "step": 228728
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.3787891864776611,
      "learning_rate": 7.814600791716807e-08,
      "loss": 1.975,
      "step": 228729
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.5763795375823975,
      "learning_rate": 7.805267544521709e-08,
      "loss": 2.149,
      "step": 228730
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.5218756198883057,
      "learning_rate": 7.795939873439516e-08,
      "loss": 1.9211,
      "step": 228731
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.6391321420669556,
      "learning_rate": 7.786617778473559e-08,
      "loss": 2.1105,
      "step": 228732
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.4661364555358887,
      "learning_rate": 7.777301259627167e-08,
      "loss": 1.9342,
      "step": 228733
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.5060577392578125,
      "learning_rate": 7.767990316900341e-08,
      "loss": 2.0195,
      "step": 228734
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.586387276649475,
      "learning_rate": 7.758684950296413e-08,
      "loss": 1.9686,
      "step": 228735
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.490912914276123,
      "learning_rate": 7.74938515981871e-08,
      "loss": 1.9589,
      "step": 228736
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.4848395586013794,
      "learning_rate": 7.740090945463906e-08,
      "loss": 2.0167,
      "step": 228737
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.637575626373291,
      "learning_rate": 7.730802307235329e-08,
      "loss": 2.0683,
      "step": 228738
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.4866870641708374,
      "learning_rate": 7.72151924513964e-08,
      "loss": 1.9591,
      "step": 228739
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.3602532148361206,
      "learning_rate": 7.712241759170179e-08,
      "loss": 1.9095,
      "step": 228740
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.543587327003479,
      "learning_rate": 7.702969849336937e-08,
      "loss": 2.2854,
      "step": 228741
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.5544724464416504,
      "learning_rate": 7.693703515636585e-08,
      "loss": 1.918,
      "step": 228742
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.4753100872039795,
      "learning_rate": 7.68444275806912e-08,
      "loss": 1.8934,
      "step": 228743
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.3617383241653442,
      "learning_rate": 7.675187576644537e-08,
      "loss": 2.0146,
      "step": 228744
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.6966885328292847,
      "learning_rate": 7.665937971356173e-08,
      "loss": 1.8217,
      "step": 228745
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.3959065675735474,
      "learning_rate": 7.65669394220736e-08,
      "loss": 2.3856,
      "step": 228746
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.5129129886627197,
      "learning_rate": 7.647455489204757e-08,
      "loss": 1.8573,
      "step": 228747
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.5669150352478027,
      "learning_rate": 7.638222612345035e-08,
      "loss": 1.7556,
      "step": 228748
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.5296818017959595,
      "learning_rate": 7.628995311631524e-08,
      "loss": 2.045,
      "step": 228749
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.4843075275421143,
      "learning_rate": 7.619773587064227e-08,
      "loss": 1.9852,
      "step": 228750
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.8888012170791626,
      "learning_rate": 7.610557438649801e-08,
      "loss": 1.9093,
      "step": 228751
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.3662012815475464,
      "learning_rate": 7.601346866384917e-08,
      "loss": 1.7327,
      "step": 228752
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.521464467048645,
      "learning_rate": 7.592141870272906e-08,
      "loss": 2.0296,
      "step": 228753
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.478018045425415,
      "learning_rate": 7.582942450317098e-08,
      "loss": 2.0052,
      "step": 228754
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.3951181173324585,
      "learning_rate": 7.573748606514163e-08,
      "loss": 1.969,
      "step": 228755
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.4024786949157715,
      "learning_rate": 7.564560338874093e-08,
      "loss": 1.7044,
      "step": 228756
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.290047287940979,
      "learning_rate": 7.555377647390227e-08,
      "loss": 1.8835,
      "step": 228757
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.535661220550537,
      "learning_rate": 7.546200532072555e-08,
      "loss": 2.0436,
      "step": 228758
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.4258368015289307,
      "learning_rate": 7.537028992914418e-08,
      "loss": 2.0444,
      "step": 228759
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.513863444328308,
      "learning_rate": 7.527863029922477e-08,
      "loss": 2.0703,
      "step": 228760
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.506091833114624,
      "learning_rate": 7.51870264309673e-08,
      "loss": 2.0735,
      "step": 228761
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.4743404388427734,
      "learning_rate": 7.50954783244051e-08,
      "loss": 1.8868,
      "step": 228762
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.426429271697998,
      "learning_rate": 7.500398597950485e-08,
      "loss": 2.0432,
      "step": 228763
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.4134461879730225,
      "learning_rate": 7.491254939636648e-08,
      "loss": 1.8451,
      "step": 228764
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.4320353269577026,
      "learning_rate": 7.482116857495668e-08,
      "loss": 1.8831,
      "step": 228765
    },
    {
      "epoch": 2.98,
      "grad_norm": 3.33764386177063,
      "learning_rate": 7.472984351527544e-08,
      "loss": 1.978,
      "step": 228766
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.4621379375457764,
      "learning_rate": 7.463857421738939e-08,
      "loss": 1.9845,
      "step": 228767
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.5807876586914062,
      "learning_rate": 7.454736068129852e-08,
      "loss": 1.9158,
      "step": 228768
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.4582446813583374,
      "learning_rate": 7.445620290696952e-08,
      "loss": 1.9942,
      "step": 228769
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.6630092859268188,
      "learning_rate": 7.436510089450231e-08,
      "loss": 2.1333,
      "step": 228770
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.5389174222946167,
      "learning_rate": 7.427405464383029e-08,
      "loss": 1.994,
      "step": 228771
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.4502147436141968,
      "learning_rate": 7.418306415505337e-08,
      "loss": 2.2253,
      "step": 228772
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.416072964668274,
      "learning_rate": 7.409212942813825e-08,
      "loss": 1.8334,
      "step": 228773
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.534227728843689,
      "learning_rate": 7.400125046308491e-08,
      "loss": 2.0698,
      "step": 228774
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.5219939947128296,
      "learning_rate": 7.391042725995999e-08,
      "loss": 2.0766,
      "step": 228775
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.342028260231018,
      "learning_rate": 7.381965981876348e-08,
      "loss": 1.8136,
      "step": 228776
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.4605236053466797,
      "learning_rate": 7.372894813949538e-08,
      "loss": 1.8399,
      "step": 228777
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.4072619676589966,
      "learning_rate": 7.363829222215567e-08,
      "loss": 2.1073,
      "step": 228778
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.5789451599121094,
      "learning_rate": 7.354769206681099e-08,
      "loss": 1.9326,
      "step": 228779
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.4884430170059204,
      "learning_rate": 7.345714767346134e-08,
      "loss": 1.8957,
      "step": 228780
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.5491007566452026,
      "learning_rate": 7.33666590421067e-08,
      "loss": 2.1783,
      "step": 228781
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.8535493612289429,
      "learning_rate": 7.32762261727804e-08,
      "loss": 1.8841,
      "step": 228782
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.4796149730682373,
      "learning_rate": 7.318584906551572e-08,
      "loss": 2.1114,
      "step": 228783
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.495563268661499,
      "learning_rate": 7.309552772027938e-08,
      "loss": 2.3027,
      "step": 228784
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.535990834236145,
      "learning_rate": 7.300526213710467e-08,
      "loss": 1.8725,
      "step": 228785
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.437637209892273,
      "learning_rate": 7.291505231605821e-08,
      "loss": 1.9487,
      "step": 228786
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.4955224990844727,
      "learning_rate": 7.282489825710669e-08,
      "loss": 1.9084,
      "step": 228787
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.6389762163162231,
      "learning_rate": 7.273479996025012e-08,
      "loss": 2.1048,
      "step": 228788
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.5760481357574463,
      "learning_rate": 7.264475742555509e-08,
      "loss": 1.8766,
      "step": 228789
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.3260811567306519,
      "learning_rate": 7.255477065302162e-08,
      "loss": 1.7526,
      "step": 228790
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.5248936414718628,
      "learning_rate": 7.246483964264971e-08,
      "loss": 1.9875,
      "step": 228791
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.587815284729004,
      "learning_rate": 7.237496439447265e-08,
      "loss": 2.0643,
      "step": 228792
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.4851417541503906,
      "learning_rate": 7.228514490852378e-08,
      "loss": 1.9924,
      "step": 228793
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.4466534852981567,
      "learning_rate": 7.219538118476975e-08,
      "loss": 1.8915,
      "step": 228794
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.4160819053649902,
      "learning_rate": 7.21056732232772e-08,
      "loss": 1.9446,
      "step": 228795
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.4383907318115234,
      "learning_rate": 7.201602102401283e-08,
      "loss": 1.9913,
      "step": 228796
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.5490050315856934,
      "learning_rate": 7.192642458704323e-08,
      "loss": 2.1409,
      "step": 228797
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.5888360738754272,
      "learning_rate": 7.183688391236842e-08,
      "loss": 2.2662,
      "step": 228798
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.4630974531173706,
      "learning_rate": 7.174739899998839e-08,
      "loss": 1.9324,
      "step": 228799
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.3924888372421265,
      "learning_rate": 7.165796984993644e-08,
      "loss": 1.9233,
      "step": 228800
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.6759417057037354,
      "learning_rate": 7.156859646221258e-08,
      "loss": 2.0266,
      "step": 228801
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.4932641983032227,
      "learning_rate": 7.147927883685012e-08,
      "loss": 2.093,
      "step": 228802
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.5550055503845215,
      "learning_rate": 7.139001697388236e-08,
      "loss": 1.9906,
      "step": 228803
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.50795578956604,
      "learning_rate": 7.13008108733093e-08,
      "loss": 1.994,
      "step": 228804
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.5691866874694824,
      "learning_rate": 7.121166053509763e-08,
      "loss": 1.8394,
      "step": 228805
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.4762204885482788,
      "learning_rate": 7.112256595934728e-08,
      "loss": 1.7916,
      "step": 228806
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.515041708946228,
      "learning_rate": 7.103352714602495e-08,
      "loss": 2.0633,
      "step": 228807
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.4669216871261597,
      "learning_rate": 7.094454409516392e-08,
      "loss": 1.9764,
      "step": 228808
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.4675979614257812,
      "learning_rate": 7.085561680676421e-08,
      "loss": 2.0921,
      "step": 228809
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.6493265628814697,
      "learning_rate": 7.076674528085912e-08,
      "loss": 2.1182,
      "step": 228810
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.4015010595321655,
      "learning_rate": 7.067792951748197e-08,
      "loss": 1.8773,
      "step": 228811
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.6342523097991943,
      "learning_rate": 7.058916951659943e-08,
      "loss": 1.9915,
      "step": 228812
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.3497973680496216,
      "learning_rate": 7.050046527827813e-08,
      "loss": 2.1172,
      "step": 228813
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.6396198272705078,
      "learning_rate": 7.041181680248476e-08,
      "loss": 1.958,
      "step": 228814
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.5528488159179688,
      "learning_rate": 7.032322408928592e-08,
      "loss": 2.0353,
      "step": 228815
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.4803173542022705,
      "learning_rate": 7.023468713864833e-08,
      "loss": 1.7734,
      "step": 228816
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.4384280443191528,
      "learning_rate": 7.014620595063858e-08,
      "loss": 2.0203,
      "step": 228817
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.526989221572876,
      "learning_rate": 7.005778052525669e-08,
      "loss": 1.6326,
      "step": 228818
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.38433837890625,
      "learning_rate": 6.996941086250263e-08,
      "loss": 1.9924,
      "step": 228819
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.3623476028442383,
      "learning_rate": 6.988109696240973e-08,
      "loss": 2.0087,
      "step": 228820
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.5302908420562744,
      "learning_rate": 6.9792838824978e-08,
      "loss": 2.0041,
      "step": 228821
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.5575014352798462,
      "learning_rate": 6.970463645024071e-08,
      "loss": 1.9,
      "step": 228822
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.3805373907089233,
      "learning_rate": 6.96164898381979e-08,
      "loss": 1.8916,
      "step": 228823
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.4164988994598389,
      "learning_rate": 6.952839898888286e-08,
      "loss": 1.8903,
      "step": 228824
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.5180177688598633,
      "learning_rate": 6.944036390229557e-08,
      "loss": 2.0251,
      "step": 228825
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.4517818689346313,
      "learning_rate": 6.935238457846936e-08,
      "loss": 1.9365,
      "step": 228826
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.4513616561889648,
      "learning_rate": 6.926446101740423e-08,
      "loss": 2.1394,
      "step": 228827
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.4742388725280762,
      "learning_rate": 6.917659321913349e-08,
      "loss": 2.1604,
      "step": 228828
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.5423086881637573,
      "learning_rate": 6.908878118365713e-08,
      "loss": 1.9728,
      "step": 228829
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.4997010231018066,
      "learning_rate": 6.900102491100845e-08,
      "loss": 1.9119,
      "step": 228830
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.3441944122314453,
      "learning_rate": 6.891332440118747e-08,
      "loss": 1.7883,
      "step": 228831
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.4954458475112915,
      "learning_rate": 6.882567965422747e-08,
      "loss": 2.0044,
      "step": 228832
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.4821873903274536,
      "learning_rate": 6.873809067009517e-08,
      "loss": 1.8449,
      "step": 228833
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.5982530117034912,
      "learning_rate": 6.865055744889048e-08,
      "loss": 1.792,
      "step": 228834
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.3856761455535889,
      "learning_rate": 6.856307999058008e-08,
      "loss": 1.9068,
      "step": 228835
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.5050609111785889,
      "learning_rate": 6.8475658295164e-08,
      "loss": 1.9155,
      "step": 228836
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.406532645225525,
      "learning_rate": 6.838829236270882e-08,
      "loss": 2.0401,
      "step": 228837
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.4248830080032349,
      "learning_rate": 6.830098219318125e-08,
      "loss": 1.9673,
      "step": 228838
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.3625272512435913,
      "learning_rate": 6.82137277866146e-08,
      "loss": 1.9084,
      "step": 228839
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.5508856773376465,
      "learning_rate": 6.812652914304217e-08,
      "loss": 1.9522,
      "step": 228840
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.4233942031860352,
      "learning_rate": 6.803938626246397e-08,
      "loss": 2.1402,
      "step": 228841
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.3504294157028198,
      "learning_rate": 6.795229914487998e-08,
      "loss": 2.0429,
      "step": 228842
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.6595767736434937,
      "learning_rate": 6.786526779032353e-08,
      "loss": 2.292,
      "step": 228843
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.3719191551208496,
      "learning_rate": 6.777829219882791e-08,
      "loss": 1.7821,
      "step": 228844
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.4627964496612549,
      "learning_rate": 6.769137237042643e-08,
      "loss": 1.9992,
      "step": 228845
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.3189139366149902,
      "learning_rate": 6.760450830505249e-08,
      "loss": 1.9481,
      "step": 228846
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.7248620986938477,
      "learning_rate": 6.751770000280599e-08,
      "loss": 1.9889,
      "step": 228847
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.510559320449829,
      "learning_rate": 6.743094746365363e-08,
      "loss": 1.9885,
      "step": 228848
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.3932788372039795,
      "learning_rate": 6.734425068762872e-08,
      "loss": 1.8212,
      "step": 228849
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.515735387802124,
      "learning_rate": 6.725760967473126e-08,
      "loss": 1.8668,
      "step": 228850
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.6154959201812744,
      "learning_rate": 6.717102442502787e-08,
      "loss": 2.2229,
      "step": 228851
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.5119520425796509,
      "learning_rate": 6.708449493845192e-08,
      "loss": 2.0651,
      "step": 228852
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.3404693603515625,
      "learning_rate": 6.699802121510334e-08,
      "loss": 1.7083,
      "step": 228853
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.4024722576141357,
      "learning_rate": 6.691160325494882e-08,
      "loss": 2.0854,
      "step": 228854
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.4588638544082642,
      "learning_rate": 6.682524105802167e-08,
      "loss": 2.2291,
      "step": 228855
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.4407387971878052,
      "learning_rate": 6.67389346243219e-08,
      "loss": 1.9104,
      "step": 228856
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.4605438709259033,
      "learning_rate": 6.66526839539161e-08,
      "loss": 2.0029,
      "step": 228857
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.6526705026626587,
      "learning_rate": 6.656648904673767e-08,
      "loss": 1.8599,
      "step": 228858
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.5611183643341064,
      "learning_rate": 6.648034990285322e-08,
      "loss": 1.9928,
      "step": 228859
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.2666347026824951,
      "learning_rate": 6.639426652229607e-08,
      "loss": 1.8786,
      "step": 228860
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.4944653511047363,
      "learning_rate": 6.63082389050329e-08,
      "loss": 2.1071,
      "step": 228861
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.5075219869613647,
      "learning_rate": 6.622226705109701e-08,
      "loss": 2.0197,
      "step": 228862
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.6070643663406372,
      "learning_rate": 6.613635096055503e-08,
      "loss": 2.1067,
      "step": 228863
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.4837853908538818,
      "learning_rate": 6.605049063334034e-08,
      "loss": 1.9663,
      "step": 228864
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.4712940454483032,
      "learning_rate": 6.596468606951955e-08,
      "loss": 2.0531,
      "step": 228865
    },
    {
      "epoch": 2.98,
      "grad_norm": 2.518587589263916,
      "learning_rate": 6.587893726909265e-08,
      "loss": 1.9231,
      "step": 228866
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.487351655960083,
      "learning_rate": 6.579324423209298e-08,
      "loss": 1.9601,
      "step": 228867
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.5083831548690796,
      "learning_rate": 6.57076069585205e-08,
      "loss": 1.9895,
      "step": 228868
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.769148349761963,
      "learning_rate": 6.562202544840855e-08,
      "loss": 2.2816,
      "step": 228869
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.4059536457061768,
      "learning_rate": 6.55364997017238e-08,
      "loss": 1.8712,
      "step": 228870
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.4780117273330688,
      "learning_rate": 6.545102971853288e-08,
      "loss": 1.9301,
      "step": 228871
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.5067174434661865,
      "learning_rate": 6.536561549883579e-08,
      "loss": 2.0544,
      "step": 228872
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.5257123708724976,
      "learning_rate": 6.528025704266582e-08,
      "loss": 2.0799,
      "step": 228873
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.539581537246704,
      "learning_rate": 6.519495435002297e-08,
      "loss": 2.0142,
      "step": 228874
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.8144221305847168,
      "learning_rate": 6.510970742090727e-08,
      "loss": 2.0114,
      "step": 228875
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.4509532451629639,
      "learning_rate": 6.502451625535198e-08,
      "loss": 1.8505,
      "step": 228876
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.5410202741622925,
      "learning_rate": 6.493938085335715e-08,
      "loss": 2.1058,
      "step": 228877
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.4733072519302368,
      "learning_rate": 6.485430121495605e-08,
      "loss": 1.9278,
      "step": 228878
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.4518001079559326,
      "learning_rate": 6.4769277340182e-08,
      "loss": 1.9705,
      "step": 228879
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.582518458366394,
      "learning_rate": 6.468430922900169e-08,
      "loss": 2.0219,
      "step": 228880
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.4130693674087524,
      "learning_rate": 6.459939688148175e-08,
      "loss": 2.1364,
      "step": 228881
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.6743240356445312,
      "learning_rate": 6.451454029758884e-08,
      "loss": 2.0351,
      "step": 228882
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.4546891450881958,
      "learning_rate": 6.44297394773896e-08,
      "loss": 2.0063,
      "step": 228883
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.4434057474136353,
      "learning_rate": 6.434499442085073e-08,
      "loss": 1.8529,
      "step": 228884
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.4916473627090454,
      "learning_rate": 6.426030512803881e-08,
      "loss": 1.9892,
      "step": 228885
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.636407732963562,
      "learning_rate": 6.417567159892056e-08,
      "loss": 2.0613,
      "step": 228886
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.3628348112106323,
      "learning_rate": 6.409109383356259e-08,
      "loss": 1.8101,
      "step": 228887
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.4572309255599976,
      "learning_rate": 6.400657183193158e-08,
      "loss": 1.7241,
      "step": 228888
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.479162335395813,
      "learning_rate": 6.392210559406086e-08,
      "loss": 1.962,
      "step": 228889
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.4802528619766235,
      "learning_rate": 6.383769511995041e-08,
      "loss": 2.2323,
      "step": 228890
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.4139267206192017,
      "learning_rate": 6.375334040966684e-08,
      "loss": 2.0043,
      "step": 228891
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.433626651763916,
      "learning_rate": 6.366904146317686e-08,
      "loss": 1.8092,
      "step": 228892
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.4034947156906128,
      "learning_rate": 6.358479828051377e-08,
      "loss": 2.0273,
      "step": 228893
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.5460933446884155,
      "learning_rate": 6.350061086171087e-08,
      "loss": 2.2569,
      "step": 228894
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.5535463094711304,
      "learning_rate": 6.341647920673488e-08,
      "loss": 2.0204,
      "step": 228895
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.4144796133041382,
      "learning_rate": 6.333240331565237e-08,
      "loss": 2.0649,
      "step": 228896
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.4695713520050049,
      "learning_rate": 6.324838318846337e-08,
      "loss": 1.8633,
      "step": 228897
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.3716950416564941,
      "learning_rate": 6.316441882516787e-08,
      "loss": 2.1679,
      "step": 228898
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.3090323209762573,
      "learning_rate": 6.308051022579919e-08,
      "loss": 2.0447,
      "step": 228899
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.5643081665039062,
      "learning_rate": 6.299665739035731e-08,
      "loss": 1.8981,
      "step": 228900
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.4882091283798218,
      "learning_rate": 6.291286031884224e-08,
      "loss": 1.9789,
      "step": 228901
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.4340791702270508,
      "learning_rate": 6.282911901132059e-08,
      "loss": 2.1228,
      "step": 228902
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.4659976959228516,
      "learning_rate": 6.274543346779237e-08,
      "loss": 2.1025,
      "step": 228903
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.5313912630081177,
      "learning_rate": 6.266180368825757e-08,
      "loss": 2.0711,
      "step": 228904
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.39651620388031,
      "learning_rate": 6.257822967271619e-08,
      "loss": 1.9593,
      "step": 228905
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.3964637517929077,
      "learning_rate": 6.249471142120155e-08,
      "loss": 2.0414,
      "step": 228906
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.4763808250427246,
      "learning_rate": 6.241124893374694e-08,
      "loss": 1.9642,
      "step": 228907
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.4959805011749268,
      "learning_rate": 6.232784221035236e-08,
      "loss": 2.0329,
      "step": 228908
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.4151428937911987,
      "learning_rate": 6.224449125101782e-08,
      "loss": 2.0105,
      "step": 228909
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.5558958053588867,
      "learning_rate": 6.216119605577664e-08,
      "loss": 2.1014,
      "step": 228910
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.4986993074417114,
      "learning_rate": 6.207795662466208e-08,
      "loss": 1.9397,
      "step": 228911
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.3791706562042236,
      "learning_rate": 6.199477295764088e-08,
      "loss": 1.7934,
      "step": 228912
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.6653228998184204,
      "learning_rate": 6.191164505477963e-08,
      "loss": 2.0505,
      "step": 228913
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.4727141857147217,
      "learning_rate": 6.182857291607835e-08,
      "loss": 1.8808,
      "step": 228914
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.4717293977737427,
      "learning_rate": 6.174555654150371e-08,
      "loss": 2.0169,
      "step": 228915
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.671491265296936,
      "learning_rate": 6.166259593115563e-08,
      "loss": 1.9896,
      "step": 228916
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.4374791383743286,
      "learning_rate": 6.157969108496752e-08,
      "loss": 1.8967,
      "step": 228917
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.3619508743286133,
      "learning_rate": 6.149684200300598e-08,
      "loss": 1.933,
      "step": 228918
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.5246251821517944,
      "learning_rate": 6.141404868530431e-08,
      "loss": 1.9331,
      "step": 228919
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.6830781698226929,
      "learning_rate": 6.13313111318292e-08,
      "loss": 1.9275,
      "step": 228920
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.3663822412490845,
      "learning_rate": 6.124862934258067e-08,
      "loss": 1.9329,
      "step": 228921
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.4882630109786987,
      "learning_rate": 6.116600331765864e-08,
      "loss": 2.1044,
      "step": 228922
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.4455881118774414,
      "learning_rate": 6.108343305699648e-08,
      "loss": 1.9015,
      "step": 228923
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.5475852489471436,
      "learning_rate": 6.100091856066081e-08,
      "loss": 1.8748,
      "step": 228924
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.4450687170028687,
      "learning_rate": 6.091845982861832e-08,
      "loss": 1.8042,
      "step": 228925
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.407025694847107,
      "learning_rate": 6.083605686093562e-08,
      "loss": 1.9259,
      "step": 228926
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.3942371606826782,
      "learning_rate": 6.075370965761272e-08,
      "loss": 1.9046,
      "step": 228927
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.4260002374649048,
      "learning_rate": 6.067141821864963e-08,
      "loss": 1.9102,
      "step": 228928
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.6041879653930664,
      "learning_rate": 6.058918254404632e-08,
      "loss": 2.1188,
      "step": 228929
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.4252625703811646,
      "learning_rate": 6.050700263386942e-08,
      "loss": 1.9274,
      "step": 228930
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.409850001335144,
      "learning_rate": 6.042487848808564e-08,
      "loss": 1.8557,
      "step": 228931
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.3733872175216675,
      "learning_rate": 6.034281010676157e-08,
      "loss": 1.9963,
      "step": 228932
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.6940687894821167,
      "learning_rate": 6.02607974898639e-08,
      "loss": 1.8987,
      "step": 228933
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.4863100051879883,
      "learning_rate": 6.017884063742595e-08,
      "loss": 2.0393,
      "step": 228934
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.5207408666610718,
      "learning_rate": 6.009693954944772e-08,
      "loss": 2.1051,
      "step": 228935
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.5147182941436768,
      "learning_rate": 6.001509422596251e-08,
      "loss": 1.9655,
      "step": 228936
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.4736177921295166,
      "learning_rate": 5.993330466700364e-08,
      "loss": 1.834,
      "step": 228937
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.5041042566299438,
      "learning_rate": 5.985157087253779e-08,
      "loss": 2.0694,
      "step": 228938
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.4816937446594238,
      "learning_rate": 5.976989284263156e-08,
      "loss": 1.7065,
      "step": 228939
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.4744552373886108,
      "learning_rate": 5.968827057725167e-08,
      "loss": 1.946,
      "step": 228940
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.5392976999282837,
      "learning_rate": 5.960670407646473e-08,
      "loss": 2.0994,
      "step": 228941
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.4329496622085571,
      "learning_rate": 5.952519334023742e-08,
      "loss": 2.0371,
      "step": 228942
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.5694472789764404,
      "learning_rate": 5.9443738368636364e-08,
      "loss": 1.7679,
      "step": 228943
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.626173973083496,
      "learning_rate": 5.9362339161628244e-08,
      "loss": 1.8933,
      "step": 228944
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.5425434112548828,
      "learning_rate": 5.928099571924638e-08,
      "loss": 1.9564,
      "step": 228945
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.5390228033065796,
      "learning_rate": 5.9199708041490764e-08,
      "loss": 2.1686,
      "step": 228946
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.4985721111297607,
      "learning_rate": 5.911847612839471e-08,
      "loss": 1.8394,
      "step": 228947
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.4612534046173096,
      "learning_rate": 5.903729997999152e-08,
      "loss": 1.9789,
      "step": 228948
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.489273190498352,
      "learning_rate": 5.8956179596247875e-08,
      "loss": 1.8942,
      "step": 228949
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.5242836475372314,
      "learning_rate": 5.887511497723041e-08,
      "loss": 1.7421,
      "step": 228950
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.5371960401535034,
      "learning_rate": 5.87941061229058e-08,
      "loss": 2.2104,
      "step": 228951
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.3952239751815796,
      "learning_rate": 5.8713153033340676e-08,
      "loss": 1.8799,
      "step": 228952
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.3951603174209595,
      "learning_rate": 5.863225570850172e-08,
      "loss": 2.069,
      "step": 228953
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.5748605728149414,
      "learning_rate": 5.8551414148455546e-08,
      "loss": 2.065,
      "step": 228954
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.4577808380126953,
      "learning_rate": 5.847062835316885e-08,
      "loss": 1.7186,
      "step": 228955
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.4397671222686768,
      "learning_rate": 5.838989832264163e-08,
      "loss": 1.9533,
      "step": 228956
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.3999265432357788,
      "learning_rate": 5.83092240569738e-08,
      "loss": 1.9661,
      "step": 228957
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.5705389976501465,
      "learning_rate": 5.8228605556098764e-08,
      "loss": 1.9192,
      "step": 228958
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.6582473516464233,
      "learning_rate": 5.814804282004981e-08,
      "loss": 1.9971,
      "step": 228959
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.4249069690704346,
      "learning_rate": 5.806753584889356e-08,
      "loss": 2.1225,
      "step": 228960
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.4714959859848022,
      "learning_rate": 5.798708464256341e-08,
      "loss": 1.9468,
      "step": 228961
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.5052154064178467,
      "learning_rate": 5.7906689201125955e-08,
      "loss": 1.912,
      "step": 228962
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.57143235206604,
      "learning_rate": 5.7826349524614514e-08,
      "loss": 1.9971,
      "step": 228963
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.4315340518951416,
      "learning_rate": 5.7746065612995776e-08,
      "loss": 2.0652,
      "step": 228964
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.502166509628296,
      "learning_rate": 5.766583746626974e-08,
      "loss": 2.0916,
      "step": 228965
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.6379239559173584,
      "learning_rate": 5.758566508453632e-08,
      "loss": 2.0559,
      "step": 228966
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.347619891166687,
      "learning_rate": 5.750554846772892e-08,
      "loss": 2.2056,
      "step": 228967
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.5260752439498901,
      "learning_rate": 5.7425487615914144e-08,
      "loss": 1.8132,
      "step": 228968
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.6282830238342285,
      "learning_rate": 5.734548252905868e-08,
      "loss": 2.0246,
      "step": 228969
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.483177900314331,
      "learning_rate": 5.726553320722915e-08,
      "loss": 1.7306,
      "step": 228970
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.7113850116729736,
      "learning_rate": 5.718563965042555e-08,
      "loss": 2.1649,
      "step": 228971
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.5785973072052002,
      "learning_rate": 5.7105801858614576e-08,
      "loss": 1.9089,
      "step": 228972
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.4449390172958374,
      "learning_rate": 5.702601983186283e-08,
      "loss": 2.0161,
      "step": 228973
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.4203791618347168,
      "learning_rate": 5.694629357020364e-08,
      "loss": 2.1594,
      "step": 228974
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.5406168699264526,
      "learning_rate": 5.6866623073570375e-08,
      "loss": 2.1599,
      "step": 228975
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.4309974908828735,
      "learning_rate": 5.678700834206295e-08,
      "loss": 2.0735,
      "step": 228976
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.4619446992874146,
      "learning_rate": 5.670744937564808e-08,
      "loss": 1.8659,
      "step": 228977
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.4482951164245605,
      "learning_rate": 5.6627946174359064e-08,
      "loss": 1.9694,
      "step": 228978
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.4362534284591675,
      "learning_rate": 5.654849873819589e-08,
      "loss": 2.0554,
      "step": 228979
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.4505692720413208,
      "learning_rate": 5.646910706719188e-08,
      "loss": 1.9634,
      "step": 228980
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.4494725465774536,
      "learning_rate": 5.638977116134702e-08,
      "loss": 1.8226,
      "step": 228981
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.5684148073196411,
      "learning_rate": 5.631049102069462e-08,
      "loss": 1.9777,
      "step": 228982
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.5971803665161133,
      "learning_rate": 5.623126664520139e-08,
      "loss": 2.135,
      "step": 228983
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.4756312370300293,
      "learning_rate": 5.6152098034933925e-08,
      "loss": 1.972,
      "step": 228984
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.3903456926345825,
      "learning_rate": 5.607298518992553e-08,
      "loss": 1.9965,
      "step": 228985
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.512493371963501,
      "learning_rate": 5.5993928110109606e-08,
      "loss": 1.7803,
      "step": 228986
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.4385045766830444,
      "learning_rate": 5.5914926795552764e-08,
      "loss": 2.1965,
      "step": 228987
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.4988847970962524,
      "learning_rate": 5.58359812462883e-08,
      "loss": 2.009,
      "step": 228988
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.4980539083480835,
      "learning_rate": 5.575709146228291e-08,
      "loss": 2.0302,
      "step": 228989
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.3982664346694946,
      "learning_rate": 5.567825744356991e-08,
      "loss": 1.8498,
      "step": 228990
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.3794201612472534,
      "learning_rate": 5.559947919018259e-08,
      "loss": 2.212,
      "step": 228991
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.4725250005722046,
      "learning_rate": 5.5520756702120975e-08,
      "loss": 1.8552,
      "step": 228992
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.4567402601242065,
      "learning_rate": 5.544208997938504e-08,
      "loss": 1.9319,
      "step": 228993
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.4540956020355225,
      "learning_rate": 5.536347902200811e-08,
      "loss": 1.9078,
      "step": 228994
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.505709171295166,
      "learning_rate": 5.5284923830023474e-08,
      "loss": 1.97,
      "step": 228995
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.4167375564575195,
      "learning_rate": 5.520642440339784e-08,
      "loss": 1.8736,
      "step": 228996
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.4576728343963623,
      "learning_rate": 5.512798074219782e-08,
      "loss": 1.8971,
      "step": 228997
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.5306016206741333,
      "learning_rate": 5.50495928463901e-08,
      "loss": 2.0573,
      "step": 228998
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.4404973983764648,
      "learning_rate": 5.4971260716007995e-08,
      "loss": 1.8166,
      "step": 228999
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.419399619102478,
      "learning_rate": 5.48929843510848e-08,
      "loss": 1.8362,
      "step": 229000
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.5664232969284058,
      "learning_rate": 5.481476375158722e-08,
      "loss": 1.8466,
      "step": 229001
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.4061088562011719,
      "learning_rate": 5.4736598917581865e-08,
      "loss": 1.7874,
      "step": 229002
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.559468388557434,
      "learning_rate": 5.465848984906873e-08,
      "loss": 1.8802,
      "step": 229003
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.4491091966629028,
      "learning_rate": 5.4580436546047826e-08,
      "loss": 1.9713,
      "step": 229004
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.6176961660385132,
      "learning_rate": 5.4502439008552444e-08,
      "loss": 1.8676,
      "step": 229005
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.5322535037994385,
      "learning_rate": 5.4424497236582596e-08,
      "loss": 2.0564,
      "step": 229006
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.4840993881225586,
      "learning_rate": 5.434661123017159e-08,
      "loss": 2.1058,
      "step": 229007
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.5254426002502441,
      "learning_rate": 5.4268780989286106e-08,
      "loss": 1.966,
      "step": 229008
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.519963026046753,
      "learning_rate": 5.4191006513992775e-08,
      "loss": 1.9938,
      "step": 229009
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.5656452178955078,
      "learning_rate": 5.4113287804291584e-08,
      "loss": 1.8504,
      "step": 229010
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.4810304641723633,
      "learning_rate": 5.403562486018254e-08,
      "loss": 1.7611,
      "step": 229011
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.535537838935852,
      "learning_rate": 5.395801768169894e-08,
      "loss": 1.9441,
      "step": 229012
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.4719043970108032,
      "learning_rate": 5.388046626884079e-08,
      "loss": 1.8821,
      "step": 229013
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.4167522192001343,
      "learning_rate": 5.38029706216414e-08,
      "loss": 1.9763,
      "step": 229014
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.5174473524093628,
      "learning_rate": 5.372553074010078e-08,
      "loss": 2.0185,
      "step": 229015
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.4671330451965332,
      "learning_rate": 5.3648146624218904e-08,
      "loss": 2.0467,
      "step": 229016
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.3931244611740112,
      "learning_rate": 5.3570818274029093e-08,
      "loss": 2.0844,
      "step": 229017
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.5776488780975342,
      "learning_rate": 5.3493545689531346e-08,
      "loss": 2.0375,
      "step": 229018
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.4685057401657104,
      "learning_rate": 5.3416328870758974e-08,
      "loss": 1.88,
      "step": 229019
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.5538774728775024,
      "learning_rate": 5.3339167817745277e-08,
      "loss": 1.7953,
      "step": 229020
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.3739136457443237,
      "learning_rate": 5.326206253045695e-08,
      "loss": 2.0988,
      "step": 229021
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.5163894891738892,
      "learning_rate": 5.31850130089273e-08,
      "loss": 2.0686,
      "step": 229022
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.5280818939208984,
      "learning_rate": 5.310801925315633e-08,
      "loss": 2.0078,
      "step": 229023
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.4543522596359253,
      "learning_rate": 5.3031081263210654e-08,
      "loss": 2.0756,
      "step": 229024
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.3951431512832642,
      "learning_rate": 5.295419903902365e-08,
      "loss": 1.823,
      "step": 229025
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.475759744644165,
      "learning_rate": 5.2877372580695246e-08,
      "loss": 1.9653,
      "step": 229026
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.4908384084701538,
      "learning_rate": 5.280060188819213e-08,
      "loss": 2.0043,
      "step": 229027
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.533910870552063,
      "learning_rate": 5.272388696151431e-08,
      "loss": 2.0788,
      "step": 229028
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.377130389213562,
      "learning_rate": 5.264722780072839e-08,
      "loss": 1.9568,
      "step": 229029
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.552281141281128,
      "learning_rate": 5.257062440576776e-08,
      "loss": 1.9271,
      "step": 229030
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.3684957027435303,
      "learning_rate": 5.249407677673234e-08,
      "loss": 1.9601,
      "step": 229031
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.4000928401947021,
      "learning_rate": 5.2417584913588825e-08,
      "loss": 2.0585,
      "step": 229032
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.4607229232788086,
      "learning_rate": 5.2341148816370526e-08,
      "loss": 1.865,
      "step": 229033
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.544646978378296,
      "learning_rate": 5.226476848511074e-08,
      "loss": 2.2287,
      "step": 229034
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.5280829668045044,
      "learning_rate": 5.218844391974286e-08,
      "loss": 2.1339,
      "step": 229035
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.4338785409927368,
      "learning_rate": 5.2112175120366806e-08,
      "loss": 1.9489,
      "step": 229036
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.5292197465896606,
      "learning_rate": 5.203596208698257e-08,
      "loss": 1.9085,
      "step": 229037
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.5483157634735107,
      "learning_rate": 5.195980481955686e-08,
      "loss": 1.9699,
      "step": 229038
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.496317982673645,
      "learning_rate": 5.1883703318122974e-08,
      "loss": 1.9436,
      "step": 229039
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.4231879711151123,
      "learning_rate": 5.1807657582714215e-08,
      "loss": 1.9844,
      "step": 229040
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.495577335357666,
      "learning_rate": 5.173166761336389e-08,
      "loss": 2.2336,
      "step": 229041
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.5994818210601807,
      "learning_rate": 5.1655733410038704e-08,
      "loss": 2.0549,
      "step": 229042
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.5360662937164307,
      "learning_rate": 5.157985497277195e-08,
      "loss": 1.9977,
      "step": 229043
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.5145717859268188,
      "learning_rate": 5.150403230159694e-08,
      "loss": 2.0184,
      "step": 229044
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.3383526802062988,
      "learning_rate": 5.142826539648037e-08,
      "loss": 2.002,
      "step": 229045
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.3488624095916748,
      "learning_rate": 5.135255425748885e-08,
      "loss": 2.1164,
      "step": 229046
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.41025972366333,
      "learning_rate": 5.127689888458908e-08,
      "loss": 2.0275,
      "step": 229047
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.4193083047866821,
      "learning_rate": 5.1201299277814355e-08,
      "loss": 1.9229,
      "step": 229048
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.5528992414474487,
      "learning_rate": 5.1125755437231296e-08,
      "loss": 1.9558,
      "step": 229049
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.4418582916259766,
      "learning_rate": 5.105026736277329e-08,
      "loss": 2.0043,
      "step": 229050
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.487274408340454,
      "learning_rate": 5.097483505447364e-08,
      "loss": 1.9777,
      "step": 229051
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.3929266929626465,
      "learning_rate": 5.0899458512365656e-08,
      "loss": 2.0463,
      "step": 229052
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.3964262008666992,
      "learning_rate": 5.082413773648264e-08,
      "loss": 1.9753,
      "step": 229053
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.4554189443588257,
      "learning_rate": 5.074887272679129e-08,
      "loss": 2.1225,
      "step": 229054
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.6278125047683716,
      "learning_rate": 5.067366348332491e-08,
      "loss": 1.9954,
      "step": 229055
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.5211158990859985,
      "learning_rate": 5.0598510006116813e-08,
      "loss": 2.0604,
      "step": 229056
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.5112292766571045,
      "learning_rate": 5.052341229516699e-08,
      "loss": 2.1574,
      "step": 229057
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.4434399604797363,
      "learning_rate": 5.044837035047544e-08,
      "loss": 1.9904,
      "step": 229058
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.4335176944732666,
      "learning_rate": 5.0373384172075485e-08,
      "loss": 1.9796,
      "step": 229059
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.5169140100479126,
      "learning_rate": 5.0298453759967106e-08,
      "loss": 1.9011,
      "step": 229060
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.5648283958435059,
      "learning_rate": 5.022357911415031e-08,
      "loss": 2.0761,
      "step": 229061
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.6191619634628296,
      "learning_rate": 5.014876023469172e-08,
      "loss": 2.1072,
      "step": 229062
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.4397189617156982,
      "learning_rate": 5.007399712155802e-08,
      "loss": 2.0888,
      "step": 229063
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.4307491779327393,
      "learning_rate": 4.999928977478251e-08,
      "loss": 2.1563,
      "step": 229064
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.4570808410644531,
      "learning_rate": 4.9924638194365205e-08,
      "loss": 1.9167,
      "step": 229065
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.4555448293685913,
      "learning_rate": 4.9850042380339406e-08,
      "loss": 2.0663,
      "step": 229066
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.4641655683517456,
      "learning_rate": 4.977550233270511e-08,
      "loss": 1.987,
      "step": 229067
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.3810582160949707,
      "learning_rate": 4.9701018051462316e-08,
      "loss": 1.9743,
      "step": 229068
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.547778606414795,
      "learning_rate": 4.962658953664434e-08,
      "loss": 2.1749,
      "step": 229069
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.674839735031128,
      "learning_rate": 4.955221678828447e-08,
      "loss": 1.8322,
      "step": 229070
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.4494884014129639,
      "learning_rate": 4.9477899806382723e-08,
      "loss": 1.9083,
      "step": 229071
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.5699354410171509,
      "learning_rate": 4.9403638590905795e-08,
      "loss": 2.0297,
      "step": 229072
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.498475193977356,
      "learning_rate": 4.932943314195359e-08,
      "loss": 2.0256,
      "step": 229073
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.3998430967330933,
      "learning_rate": 4.9255283459459504e-08,
      "loss": 1.8816,
      "step": 229074
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.534483790397644,
      "learning_rate": 4.9181189543490154e-08,
      "loss": 2.2852,
      "step": 229075
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.5049715042114258,
      "learning_rate": 4.910715139401222e-08,
      "loss": 2.2032,
      "step": 229076
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.362735629081726,
      "learning_rate": 4.903316901109233e-08,
      "loss": 1.9616,
      "step": 229077
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.3820006847381592,
      "learning_rate": 4.8959242394697176e-08,
      "loss": 1.8666,
      "step": 229078
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.402491807937622,
      "learning_rate": 4.8885371544893357e-08,
      "loss": 1.9389,
      "step": 229079
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.40819251537323,
      "learning_rate": 4.8811556461647585e-08,
      "loss": 2.021,
      "step": 229080
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.528127908706665,
      "learning_rate": 4.873779714499315e-08,
      "loss": 2.0813,
      "step": 229081
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.4042826890945435,
      "learning_rate": 4.8664093594930066e-08,
      "loss": 1.8959,
      "step": 229082
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.4053009748458862,
      "learning_rate": 4.859044581149163e-08,
      "loss": 1.8633,
      "step": 229083
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.4598193168640137,
      "learning_rate": 4.851685379467785e-08,
      "loss": 2.0072,
      "step": 229084
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.509492039680481,
      "learning_rate": 4.844331754452202e-08,
      "loss": 2.0669,
      "step": 229085
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.404168725013733,
      "learning_rate": 4.8369837060990846e-08,
      "loss": 1.9469,
      "step": 229086
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.3694846630096436,
      "learning_rate": 4.829641234415094e-08,
      "loss": 2.1478,
      "step": 229087
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.3697646856307983,
      "learning_rate": 4.822304339400229e-08,
      "loss": 1.8092,
      "step": 229088
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.6089377403259277,
      "learning_rate": 4.814973021054491e-08,
      "loss": 2.1434,
      "step": 229089
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.4440383911132812,
      "learning_rate": 4.80764727937788e-08,
      "loss": 1.8226,
      "step": 229090
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.3961576223373413,
      "learning_rate": 4.8003271143737256e-08,
      "loss": 2.0379,
      "step": 229091
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.4464564323425293,
      "learning_rate": 4.793012526045359e-08,
      "loss": 2.067,
      "step": 229092
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.5774770975112915,
      "learning_rate": 4.78570351439278e-08,
      "loss": 1.9629,
      "step": 229093
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.6923836469650269,
      "learning_rate": 4.778400079415989e-08,
      "loss": 1.8013,
      "step": 229094
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.4569882154464722,
      "learning_rate": 4.771102221118317e-08,
      "loss": 1.9528,
      "step": 229095
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.3674808740615845,
      "learning_rate": 4.7638099394964324e-08,
      "loss": 2.1012,
      "step": 229096
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.4821847677230835,
      "learning_rate": 4.756523234556997e-08,
      "loss": 1.9845,
      "step": 229097
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.3867816925048828,
      "learning_rate": 4.7492421063000105e-08,
      "loss": 2.3015,
      "step": 229098
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.592397689819336,
      "learning_rate": 4.741966554725474e-08,
      "loss": 2.1016,
      "step": 229099
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.4865816831588745,
      "learning_rate": 4.7346965798367165e-08,
      "loss": 2.0366,
      "step": 229100
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.4825705289840698,
      "learning_rate": 4.727432181633739e-08,
      "loss": 1.8498,
      "step": 229101
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.4158132076263428,
      "learning_rate": 4.7201733601165414e-08,
      "loss": 1.9421,
      "step": 229102
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.367275357246399,
      "learning_rate": 4.712920115291785e-08,
      "loss": 1.9966,
      "step": 229103
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.4909993410110474,
      "learning_rate": 4.705672447152808e-08,
      "loss": 2.0985,
      "step": 229104
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.4323197603225708,
      "learning_rate": 4.698430355706273e-08,
      "loss": 1.9835,
      "step": 229105
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.4360030889511108,
      "learning_rate": 4.691193840955509e-08,
      "loss": 2.1194,
      "step": 229106
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.4273278713226318,
      "learning_rate": 4.683962902897187e-08,
      "loss": 1.9095,
      "step": 229107
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.4319024085998535,
      "learning_rate": 4.676737541531306e-08,
      "loss": 2.0416,
      "step": 229108
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.9153904914855957,
      "learning_rate": 4.6695177568678574e-08,
      "loss": 2.1005,
      "step": 229109
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.7019678354263306,
      "learning_rate": 4.662303548896851e-08,
      "loss": 2.3141,
      "step": 229110
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.4346638917922974,
      "learning_rate": 4.655094917628277e-08,
      "loss": 1.962,
      "step": 229111
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.4623037576675415,
      "learning_rate": 4.6478918630621365e-08,
      "loss": 2.1145,
      "step": 229112
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.7416774034500122,
      "learning_rate": 4.6406943851950984e-08,
      "loss": 2.1399,
      "step": 229113
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.4621233940124512,
      "learning_rate": 4.633502484033824e-08,
      "loss": 1.7387,
      "step": 229114
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.5593645572662354,
      "learning_rate": 4.6263161595749834e-08,
      "loss": 1.9179,
      "step": 229115
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.5972139835357666,
      "learning_rate": 4.619135411821906e-08,
      "loss": 2.0719,
      "step": 229116
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.6054309606552124,
      "learning_rate": 4.6119602407779234e-08,
      "loss": 2.0069,
      "step": 229117
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.5807039737701416,
      "learning_rate": 4.604790646443035e-08,
      "loss": 1.8667,
      "step": 229118
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.4811898469924927,
      "learning_rate": 4.597626628820572e-08,
      "loss": 1.8444,
      "step": 229119
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.4804185628890991,
      "learning_rate": 4.5904681879072035e-08,
      "loss": 1.9095,
      "step": 229120
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.3991268873214722,
      "learning_rate": 4.58331532370626e-08,
      "loss": 1.6488,
      "step": 229121
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.5610257387161255,
      "learning_rate": 4.5761680362177425e-08,
      "loss": 1.9176,
      "step": 229122
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.5038472414016724,
      "learning_rate": 4.5690263254483103e-08,
      "loss": 2.0969,
      "step": 229123
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.4672510623931885,
      "learning_rate": 4.561890191394635e-08,
      "loss": 1.8396,
      "step": 229124
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.5378546714782715,
      "learning_rate": 4.554759634056715e-08,
      "loss": 2.0179,
      "step": 229125
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.4440271854400635,
      "learning_rate": 4.5476346534412126e-08,
      "loss": 1.879,
      "step": 229126
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.4847283363342285,
      "learning_rate": 4.540515249544796e-08,
      "loss": 1.9045,
      "step": 229127
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.551283597946167,
      "learning_rate": 4.533401422374128e-08,
      "loss": 2.0918,
      "step": 229128
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.5694409608840942,
      "learning_rate": 4.526293171922546e-08,
      "loss": 2.0785,
      "step": 229129
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.372243046760559,
      "learning_rate": 4.5191904981967117e-08,
      "loss": 1.9308,
      "step": 229130
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.4159849882125854,
      "learning_rate": 4.5120934011966256e-08,
      "loss": 1.5949,
      "step": 229131
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.5456860065460205,
      "learning_rate": 4.505001880925618e-08,
      "loss": 1.8297,
      "step": 229132
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.461931824684143,
      "learning_rate": 4.4979159373836896e-08,
      "loss": 1.8407,
      "step": 229133
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.4556041955947876,
      "learning_rate": 4.490835570570839e-08,
      "loss": 2.012,
      "step": 229134
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.3609049320220947,
      "learning_rate": 4.483760780487067e-08,
      "loss": 2.0572,
      "step": 229135
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.6396727561950684,
      "learning_rate": 4.4766915671390345e-08,
      "loss": 1.8634,
      "step": 229136
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.4005697965621948,
      "learning_rate": 4.469627930523412e-08,
      "loss": 1.8957,
      "step": 229137
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.6648825407028198,
      "learning_rate": 4.462569870643529e-08,
      "loss": 1.7791,
      "step": 229138
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.4785736799240112,
      "learning_rate": 4.4555173875027164e-08,
      "loss": 2.0732,
      "step": 229139
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.624497413635254,
      "learning_rate": 4.448470481097644e-08,
      "loss": 1.8383,
      "step": 229140
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.3839898109436035,
      "learning_rate": 4.4414291514316414e-08,
      "loss": 2.0817,
      "step": 229141
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.491947054862976,
      "learning_rate": 4.43439339850471e-08,
      "loss": 1.6074,
      "step": 229142
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.4156913757324219,
      "learning_rate": 4.42736322232351e-08,
      "loss": 1.8671,
      "step": 229143
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.422600507736206,
      "learning_rate": 4.420338622881381e-08,
      "loss": 1.7251,
      "step": 229144
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.3342865705490112,
      "learning_rate": 4.413319600184983e-08,
      "loss": 1.9533,
      "step": 229145
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.5044885873794556,
      "learning_rate": 4.4063061542343184e-08,
      "loss": 2.3104,
      "step": 229146
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.4419876337051392,
      "learning_rate": 4.3992982850327154e-08,
      "loss": 2.1814,
      "step": 229147
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.4963270425796509,
      "learning_rate": 4.3922959925768444e-08,
      "loss": 2.1951,
      "step": 229148
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.3729889392852783,
      "learning_rate": 4.3852992768733665e-08,
      "loss": 1.7751,
      "step": 229149
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.4671778678894043,
      "learning_rate": 4.378308137918951e-08,
      "loss": 2.0996,
      "step": 229150
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.549137830734253,
      "learning_rate": 4.371322575716929e-08,
      "loss": 2.2796,
      "step": 229151
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.4758436679840088,
      "learning_rate": 4.364342590270631e-08,
      "loss": 1.9839,
      "step": 229152
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.3667287826538086,
      "learning_rate": 4.357368181576726e-08,
      "loss": 1.9067,
      "step": 229153
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.462964415550232,
      "learning_rate": 4.3503993496385446e-08,
      "loss": 2.2327,
      "step": 229154
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.6894687414169312,
      "learning_rate": 4.343436094459418e-08,
      "loss": 2.103,
      "step": 229155
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.4766238927841187,
      "learning_rate": 4.336478416039346e-08,
      "loss": 1.7111,
      "step": 229156
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.4082640409469604,
      "learning_rate": 4.3295263143783286e-08,
      "loss": 1.8311,
      "step": 229157
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.5597604513168335,
      "learning_rate": 4.322579789479697e-08,
      "loss": 2.009,
      "step": 229158
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.6522983312606812,
      "learning_rate": 4.3156388413434496e-08,
      "loss": 1.9751,
      "step": 229159
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.4286348819732666,
      "learning_rate": 4.308703469969588e-08,
      "loss": 1.989,
      "step": 229160
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.5820978879928589,
      "learning_rate": 4.3017736753614415e-08,
      "loss": 2.0008,
      "step": 229161
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.424017071723938,
      "learning_rate": 4.2948494575223423e-08,
      "loss": 2.0,
      "step": 229162
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.543145775794983,
      "learning_rate": 4.287930816448959e-08,
      "loss": 1.8327,
      "step": 229163
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.5620205402374268,
      "learning_rate": 4.281017752144622e-08,
      "loss": 1.9806,
      "step": 229164
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.3669536113739014,
      "learning_rate": 4.274110264609332e-08,
      "loss": 1.8213,
      "step": 229165
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.4889510869979858,
      "learning_rate": 4.267208353846419e-08,
      "loss": 2.0019,
      "step": 229166
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.4012254476547241,
      "learning_rate": 4.2603120198592135e-08,
      "loss": 1.9774,
      "step": 229167
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.3920648097991943,
      "learning_rate": 4.253421262644385e-08,
      "loss": 1.9186,
      "step": 229168
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.5805045366287231,
      "learning_rate": 4.246536082201934e-08,
      "loss": 2.0361,
      "step": 229169
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.4940223693847656,
      "learning_rate": 4.239656478541853e-08,
      "loss": 1.9054,
      "step": 229170
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.4599614143371582,
      "learning_rate": 4.232782451657479e-08,
      "loss": 1.965,
      "step": 229171
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.4072654247283936,
      "learning_rate": 4.225914001552144e-08,
      "loss": 1.8488,
      "step": 229172
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.4512511491775513,
      "learning_rate": 4.219051128225848e-08,
      "loss": 1.6998,
      "step": 229173
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.4964890480041504,
      "learning_rate": 4.2121938316819206e-08,
      "loss": 1.987,
      "step": 229174
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.4877357482910156,
      "learning_rate": 4.205342111923693e-08,
      "loss": 2.3037,
      "step": 229175
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.3474525213241577,
      "learning_rate": 4.1984959689478346e-08,
      "loss": 1.9605,
      "step": 229176
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.441426396369934,
      "learning_rate": 4.191655402761007e-08,
      "loss": 1.8717,
      "step": 229177
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.5941184759140015,
      "learning_rate": 4.1848204133565486e-08,
      "loss": 1.9373,
      "step": 229178
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.4700427055358887,
      "learning_rate": 4.177991000741121e-08,
      "loss": 2.1299,
      "step": 229179
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.4542239904403687,
      "learning_rate": 4.1711671649180544e-08,
      "loss": 1.8878,
      "step": 229180
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.607527494430542,
      "learning_rate": 4.164348905884019e-08,
      "loss": 1.9777,
      "step": 229181
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.420113205909729,
      "learning_rate": 4.1575362236423436e-08,
      "loss": 2.0762,
      "step": 229182
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.6438535451889038,
      "learning_rate": 4.15072911819303e-08,
      "loss": 2.0693,
      "step": 229183
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.368971824645996,
      "learning_rate": 4.1439275895394085e-08,
      "loss": 1.7957,
      "step": 229184
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.413651466369629,
      "learning_rate": 4.137131637681479e-08,
      "loss": 2.0663,
      "step": 229185
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.484032392501831,
      "learning_rate": 4.1303412626192415e-08,
      "loss": 2.1975,
      "step": 229186
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.6530442237854004,
      "learning_rate": 4.1235564643593566e-08,
      "loss": 2.0758,
      "step": 229187
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.3315621614456177,
      "learning_rate": 4.116777242895164e-08,
      "loss": 1.9592,
      "step": 229188
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.4159446954727173,
      "learning_rate": 4.1100035982333244e-08,
      "loss": 2.0071,
      "step": 229189
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.4328055381774902,
      "learning_rate": 4.103235530373838e-08,
      "loss": 1.8935,
      "step": 229190
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.391550064086914,
      "learning_rate": 4.0964730393167055e-08,
      "loss": 1.926,
      "step": 229191
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.6036356687545776,
      "learning_rate": 4.089716125065257e-08,
      "loss": 1.9421,
      "step": 229192
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.46465265750885,
      "learning_rate": 4.0829647876194914e-08,
      "loss": 2.1759,
      "step": 229193
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.4105632305145264,
      "learning_rate": 4.0762190269827413e-08,
      "loss": 1.9075,
      "step": 229194
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.627859354019165,
      "learning_rate": 4.0694788431516746e-08,
      "loss": 2.0331,
      "step": 229195
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.3471161127090454,
      "learning_rate": 4.0627442361296225e-08,
      "loss": 2.1769,
      "step": 229196
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.578567624092102,
      "learning_rate": 4.0560152059199155e-08,
      "loss": 2.0421,
      "step": 229197
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.5396645069122314,
      "learning_rate": 4.0492917525225544e-08,
      "loss": 1.8922,
      "step": 229198
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.6455037593841553,
      "learning_rate": 4.042573875940869e-08,
      "loss": 2.0904,
      "step": 229199
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.5334357023239136,
      "learning_rate": 4.0358615761715286e-08,
      "loss": 2.0476,
      "step": 229200
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.484525442123413,
      "learning_rate": 4.029154853217864e-08,
      "loss": 1.821,
      "step": 229201
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.590375304222107,
      "learning_rate": 4.0224537070832064e-08,
      "loss": 1.8025,
      "step": 229202
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.3977783918380737,
      "learning_rate": 4.0157581377642245e-08,
      "loss": 1.9103,
      "step": 229203
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.590314269065857,
      "learning_rate": 4.00906814526758e-08,
      "loss": 2.1524,
      "step": 229204
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.355178713798523,
      "learning_rate": 4.002383729589942e-08,
      "loss": 1.9937,
      "step": 229205
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.4485055208206177,
      "learning_rate": 3.995704890734641e-08,
      "loss": 1.7969,
      "step": 229206
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.5589213371276855,
      "learning_rate": 3.9890316287050086e-08,
      "loss": 2.0595,
      "step": 229207
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.3797599077224731,
      "learning_rate": 3.982363943497713e-08,
      "loss": 1.8661,
      "step": 229208
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.4972953796386719,
      "learning_rate": 3.975701835116085e-08,
      "loss": 2.0157,
      "step": 229209
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.4964865446090698,
      "learning_rate": 3.969045303563456e-08,
      "loss": 2.2027,
      "step": 229210
    },
    {
      "epoch": 2.98,
      "grad_norm": 2.5632758140563965,
      "learning_rate": 3.962394348836495e-08,
      "loss": 2.018,
      "step": 229211
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.3626551628112793,
      "learning_rate": 3.955748970941863e-08,
      "loss": 1.9121,
      "step": 229212
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.5458718538284302,
      "learning_rate": 3.9491091698762303e-08,
      "loss": 1.9272,
      "step": 229213
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.5112417936325073,
      "learning_rate": 3.942474945642926e-08,
      "loss": 1.8506,
      "step": 229214
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.4504419565200806,
      "learning_rate": 3.935846298245282e-08,
      "loss": 2.0077,
      "step": 229215
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.6401889324188232,
      "learning_rate": 3.929223227679967e-08,
      "loss": 2.298,
      "step": 229216
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.4914898872375488,
      "learning_rate": 3.9226057339503125e-08,
      "loss": 1.9145,
      "step": 229217
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.4325861930847168,
      "learning_rate": 3.915993817059648e-08,
      "loss": 1.804,
      "step": 229218
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.4533286094665527,
      "learning_rate": 3.909387477004644e-08,
      "loss": 1.9028,
      "step": 229219
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.3914852142333984,
      "learning_rate": 3.90278671378863e-08,
      "loss": 2.0444,
      "step": 229220
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.5127332210540771,
      "learning_rate": 3.8961915274182683e-08,
      "loss": 2.1235,
      "step": 229221
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.4107402563095093,
      "learning_rate": 3.889601917883567e-08,
      "loss": 2.0073,
      "step": 229222
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.5888373851776123,
      "learning_rate": 3.8830178851978476e-08,
      "loss": 2.0871,
      "step": 229223
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.4684255123138428,
      "learning_rate": 3.876439429351119e-08,
      "loss": 2.1094,
      "step": 229224
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.4680145978927612,
      "learning_rate": 3.8698665503533734e-08,
      "loss": 1.9832,
      "step": 229225
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.375963568687439,
      "learning_rate": 3.863299248201279e-08,
      "loss": 1.7989,
      "step": 229226
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.6397150754928589,
      "learning_rate": 3.856737522898168e-08,
      "loss": 1.9499,
      "step": 229227
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.4274975061416626,
      "learning_rate": 3.850181374444039e-08,
      "loss": 1.8589,
      "step": 229228
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.468976378440857,
      "learning_rate": 3.843630802838893e-08,
      "loss": 2.0368,
      "step": 229229
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.5697888135910034,
      "learning_rate": 3.83708580808606e-08,
      "loss": 2.0846,
      "step": 229230
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.6914892196655273,
      "learning_rate": 3.830546390188871e-08,
      "loss": 1.9147,
      "step": 229231
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.308174729347229,
      "learning_rate": 3.824012549140665e-08,
      "loss": 1.8771,
      "step": 229232
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.5953072309494019,
      "learning_rate": 3.817484284951433e-08,
      "loss": 1.9208,
      "step": 229233
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.4520658254623413,
      "learning_rate": 3.8109615976178454e-08,
      "loss": 1.9943,
      "step": 229234
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.580870270729065,
      "learning_rate": 3.804444487143232e-08,
      "loss": 2.0506,
      "step": 229235
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.4002633094787598,
      "learning_rate": 3.797932953524263e-08,
      "loss": 2.0202,
      "step": 229236
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.5239099264144897,
      "learning_rate": 3.791426996767599e-08,
      "loss": 2.009,
      "step": 229237
    },
    {
      "epoch": 2.98,
      "grad_norm": 2.382002353668213,
      "learning_rate": 3.7849266168732407e-08,
      "loss": 2.3742,
      "step": 229238
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.4452627897262573,
      "learning_rate": 3.778431813841187e-08,
      "loss": 1.863,
      "step": 229239
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.5369032621383667,
      "learning_rate": 3.771942587671439e-08,
      "loss": 2.0857,
      "step": 229240
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.5025689601898193,
      "learning_rate": 3.765458938370658e-08,
      "loss": 1.9348,
      "step": 229241
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.5557568073272705,
      "learning_rate": 3.7589808659321816e-08,
      "loss": 2.2617,
      "step": 229242
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.4689571857452393,
      "learning_rate": 3.7525083703626726e-08,
      "loss": 2.1458,
      "step": 229243
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.4157371520996094,
      "learning_rate": 3.746041451658799e-08,
      "loss": 2.0359,
      "step": 229244
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.257266879081726,
      "learning_rate": 3.7395801098272225e-08,
      "loss": 1.7437,
      "step": 229245
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.4017407894134521,
      "learning_rate": 3.733124344867944e-08,
      "loss": 1.8322,
      "step": 229246
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.5303248167037964,
      "learning_rate": 3.726674156780962e-08,
      "loss": 2.0775,
      "step": 229247
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.587969183921814,
      "learning_rate": 3.7202295455629474e-08,
      "loss": 2.3592,
      "step": 229248
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.4464669227600098,
      "learning_rate": 3.713790511223891e-08,
      "loss": 2.0826,
      "step": 229249
    },
    {
      "epoch": 2.98,
      "grad_norm": 1.480444073677063,
      "learning_rate": 3.7073570537604625e-08,
      "loss": 2.0289,
      "step": 229250
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.477946162223816,
      "learning_rate": 3.700929173172662e-08,
      "loss": 1.863,
      "step": 229251
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.4051755666732788,
      "learning_rate": 3.69450686946382e-08,
      "loss": 1.8028,
      "step": 229252
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.464890956878662,
      "learning_rate": 3.688090142633937e-08,
      "loss": 1.8953,
      "step": 229253
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.7218811511993408,
      "learning_rate": 3.681678992683013e-08,
      "loss": 1.8813,
      "step": 229254
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.375797986984253,
      "learning_rate": 3.675273419617708e-08,
      "loss": 1.7731,
      "step": 229255
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.3556923866271973,
      "learning_rate": 3.668873423431362e-08,
      "loss": 1.8161,
      "step": 229256
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.4055067300796509,
      "learning_rate": 3.6624790041339666e-08,
      "loss": 1.787,
      "step": 229257
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.4811338186264038,
      "learning_rate": 3.656090161718861e-08,
      "loss": 1.8529,
      "step": 229258
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.517690658569336,
      "learning_rate": 3.649706896189375e-08,
      "loss": 1.812,
      "step": 229259
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.6383020877838135,
      "learning_rate": 3.6433292075488397e-08,
      "loss": 1.9773,
      "step": 229260
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.6189922094345093,
      "learning_rate": 3.636957095797255e-08,
      "loss": 2.1063,
      "step": 229261
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.4918794631958008,
      "learning_rate": 3.630590560934621e-08,
      "loss": 1.8203,
      "step": 229262
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.6091057062149048,
      "learning_rate": 3.624229602964268e-08,
      "loss": 1.9603,
      "step": 229263
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.4784525632858276,
      "learning_rate": 3.6178742218895277e-08,
      "loss": 1.9099,
      "step": 229264
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.481339931488037,
      "learning_rate": 3.611524417703737e-08,
      "loss": 2.2638,
      "step": 229265
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.4342461824417114,
      "learning_rate": 3.605180190413559e-08,
      "loss": 2.0466,
      "step": 229266
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.5926326513290405,
      "learning_rate": 3.5988415400223234e-08,
      "loss": 2.0552,
      "step": 229267
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.5487446784973145,
      "learning_rate": 3.5925084665267e-08,
      "loss": 1.9067,
      "step": 229268
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.4383705854415894,
      "learning_rate": 3.586180969926689e-08,
      "loss": 2.1794,
      "step": 229269
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.398208737373352,
      "learning_rate": 3.57985905022895e-08,
      "loss": 1.7673,
      "step": 229270
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.6734203100204468,
      "learning_rate": 3.573542707433485e-08,
      "loss": 2.2672,
      "step": 229271
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.5959478616714478,
      "learning_rate": 3.5672319415369634e-08,
      "loss": 1.7731,
      "step": 229272
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.6641439199447632,
      "learning_rate": 3.560926752546045e-08,
      "loss": 2.0914,
      "step": 229273
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.4170583486557007,
      "learning_rate": 3.5546271404574e-08,
      "loss": 1.9806,
      "step": 229274
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.7528069019317627,
      "learning_rate": 3.5483331052743594e-08,
      "loss": 1.8254,
      "step": 229275
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.5426502227783203,
      "learning_rate": 3.542044646996922e-08,
      "loss": 1.8129,
      "step": 229276
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.4463297128677368,
      "learning_rate": 3.5357617656284196e-08,
      "loss": 2.0026,
      "step": 229277
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.5301626920700073,
      "learning_rate": 3.5294844611688525e-08,
      "loss": 2.0253,
      "step": 229278
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.6551810503005981,
      "learning_rate": 3.52321273362155e-08,
      "loss": 2.1448,
      "step": 229279
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.5069539546966553,
      "learning_rate": 3.516946582983182e-08,
      "loss": 2.2735,
      "step": 229280
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.378159761428833,
      "learning_rate": 3.51068600925708e-08,
      "loss": 2.0768,
      "step": 229281
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.5155118703842163,
      "learning_rate": 3.5044310124465733e-08,
      "loss": 2.0541,
      "step": 229282
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.5970097780227661,
      "learning_rate": 3.4981815925483325e-08,
      "loss": 2.0655,
      "step": 229283
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.4602406024932861,
      "learning_rate": 3.4919377495656874e-08,
      "loss": 1.9766,
      "step": 229284
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.4714982509613037,
      "learning_rate": 3.4856994835019694e-08,
      "loss": 1.9275,
      "step": 229285
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.4088506698608398,
      "learning_rate": 3.479466794357177e-08,
      "loss": 2.1006,
      "step": 229286
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.472266674041748,
      "learning_rate": 3.4732396821313126e-08,
      "loss": 1.9453,
      "step": 229287
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.3518720865249634,
      "learning_rate": 3.467018146824374e-08,
      "loss": 1.8557,
      "step": 229288
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.383208155632019,
      "learning_rate": 3.460802188439693e-08,
      "loss": 1.9549,
      "step": 229289
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.527808666229248,
      "learning_rate": 3.4545918069772695e-08,
      "loss": 1.9653,
      "step": 229290
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.3333532810211182,
      "learning_rate": 3.4483870024404334e-08,
      "loss": 2.1207,
      "step": 229291
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.522814393043518,
      "learning_rate": 3.442187774829186e-08,
      "loss": 2.141,
      "step": 229292
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.3423808813095093,
      "learning_rate": 3.435994124143526e-08,
      "loss": 1.8463,
      "step": 229293
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.542482614517212,
      "learning_rate": 3.4298060503834546e-08,
      "loss": 1.9119,
      "step": 229294
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.4498800039291382,
      "learning_rate": 3.423623553555632e-08,
      "loss": 2.0015,
      "step": 229295
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.5456366539001465,
      "learning_rate": 3.417446633653398e-08,
      "loss": 2.0468,
      "step": 229296
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.3547097444534302,
      "learning_rate": 3.411275290683413e-08,
      "loss": 2.0651,
      "step": 229297
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.548737645149231,
      "learning_rate": 3.405109524649008e-08,
      "loss": 2.2841,
      "step": 229298
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.303315281867981,
      "learning_rate": 3.398949335543522e-08,
      "loss": 1.9421,
      "step": 229299
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.4560210704803467,
      "learning_rate": 3.392794723373615e-08,
      "loss": 1.8447,
      "step": 229300
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.443603277206421,
      "learning_rate": 3.386645688139289e-08,
      "loss": 1.8486,
      "step": 229301
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.4719641208648682,
      "learning_rate": 3.3805022298438736e-08,
      "loss": 1.8209,
      "step": 229302
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.8084901571273804,
      "learning_rate": 3.374364348484038e-08,
      "loss": 2.103,
      "step": 229303
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.3851521015167236,
      "learning_rate": 3.3682320440631125e-08,
      "loss": 1.7829,
      "step": 229304
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.4178720712661743,
      "learning_rate": 3.362105316581098e-08,
      "loss": 2.0583,
      "step": 229305
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.410345435142517,
      "learning_rate": 3.355984166044656e-08,
      "loss": 1.7648,
      "step": 229306
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.446300983428955,
      "learning_rate": 3.3498685924471244e-08,
      "loss": 1.7234,
      "step": 229307
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.6024370193481445,
      "learning_rate": 3.343758595791834e-08,
      "loss": 2.2216,
      "step": 229308
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.5044867992401123,
      "learning_rate": 3.3376541760821164e-08,
      "loss": 1.7023,
      "step": 229309
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.5372337102890015,
      "learning_rate": 3.3315553333213006e-08,
      "loss": 1.8505,
      "step": 229310
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.5061150789260864,
      "learning_rate": 3.325462067506057e-08,
      "loss": 2.0896,
      "step": 229311
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.499794363975525,
      "learning_rate": 3.3193743786363855e-08,
      "loss": 1.9444,
      "step": 229312
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.4724618196487427,
      "learning_rate": 3.3132922667189474e-08,
      "loss": 1.8545,
      "step": 229313
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.6732871532440186,
      "learning_rate": 3.307215731750412e-08,
      "loss": 2.1356,
      "step": 229314
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.3277769088745117,
      "learning_rate": 3.3011447737307794e-08,
      "loss": 1.9015,
      "step": 229315
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.387881875038147,
      "learning_rate": 3.2950793926667106e-08,
      "loss": 1.8415,
      "step": 229316
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.4103941917419434,
      "learning_rate": 3.289019588554875e-08,
      "loss": 2.064,
      "step": 229317
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.4153116941452026,
      "learning_rate": 3.282965361398604e-08,
      "loss": 1.8367,
      "step": 229318
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.50606107711792,
      "learning_rate": 3.2769167111978964e-08,
      "loss": 1.8906,
      "step": 229319
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.436058759689331,
      "learning_rate": 3.270873637956084e-08,
      "loss": 1.9755,
      "step": 229320
    },
    {
      "epoch": 2.99,
      "grad_norm": 2.4009687900543213,
      "learning_rate": 3.2648361416698356e-08,
      "loss": 2.1477,
      "step": 229321
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.5364713668823242,
      "learning_rate": 3.2588042223458125e-08,
      "loss": 2.1579,
      "step": 229322
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.571359634399414,
      "learning_rate": 3.252777879980684e-08,
      "loss": 1.9229,
      "step": 229323
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.436771035194397,
      "learning_rate": 3.2467571145744495e-08,
      "loss": 1.9602,
      "step": 229324
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.434622049331665,
      "learning_rate": 3.240741926133772e-08,
      "loss": 2.0957,
      "step": 229325
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.4465463161468506,
      "learning_rate": 3.23473231465865e-08,
      "loss": 1.9261,
      "step": 229326
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.6769291162490845,
      "learning_rate": 3.228728280145754e-08,
      "loss": 2.1677,
      "step": 229327
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.5153875350952148,
      "learning_rate": 3.222729822598413e-08,
      "loss": 1.9943,
      "step": 229328
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.6444512605667114,
      "learning_rate": 3.216736942019959e-08,
      "loss": 1.827,
      "step": 229329
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.5249112844467163,
      "learning_rate": 3.2107496384070616e-08,
      "loss": 1.8667,
      "step": 229330
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.4372185468673706,
      "learning_rate": 3.204767911766382e-08,
      "loss": 1.8583,
      "step": 229331
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.4731134176254272,
      "learning_rate": 3.198791762094588e-08,
      "loss": 1.8121,
      "step": 229332
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.415123462677002,
      "learning_rate": 3.192821189395012e-08,
      "loss": 2.0689,
      "step": 229333
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.3361972570419312,
      "learning_rate": 3.186856193667653e-08,
      "loss": 1.8595,
      "step": 229334
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.382722020149231,
      "learning_rate": 3.180896774915842e-08,
      "loss": 1.8094,
      "step": 229335
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.5286691188812256,
      "learning_rate": 3.174942933136249e-08,
      "loss": 1.9584,
      "step": 229336
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.489694595336914,
      "learning_rate": 3.168994668332203e-08,
      "loss": 1.8967,
      "step": 229337
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.5154900550842285,
      "learning_rate": 3.163051980507036e-08,
      "loss": 1.9686,
      "step": 229338
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.562247395515442,
      "learning_rate": 3.1571148696607486e-08,
      "loss": 1.9461,
      "step": 229339
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.4311656951904297,
      "learning_rate": 3.1511833357933393e-08,
      "loss": 1.8058,
      "step": 229340
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.4857884645462036,
      "learning_rate": 3.145257378908139e-08,
      "loss": 1.8869,
      "step": 229341
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.6673449277877808,
      "learning_rate": 3.139336999001818e-08,
      "loss": 2.0277,
      "step": 229342
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.6472771167755127,
      "learning_rate": 3.133422196077706e-08,
      "loss": 2.1697,
      "step": 229343
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.3943427801132202,
      "learning_rate": 3.127512970139134e-08,
      "loss": 2.0983,
      "step": 229344
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.5102717876434326,
      "learning_rate": 3.121609321182772e-08,
      "loss": 2.0372,
      "step": 229345
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.3170876502990723,
      "learning_rate": 3.1157112492152806e-08,
      "loss": 1.8302,
      "step": 229346
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.5228157043457031,
      "learning_rate": 3.109818754233329e-08,
      "loss": 1.9221,
      "step": 229347
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.5208427906036377,
      "learning_rate": 3.103931836240248e-08,
      "loss": 1.9408,
      "step": 229348
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.4337549209594727,
      "learning_rate": 3.0980504952360375e-08,
      "loss": 2.0065,
      "step": 229349
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.3522298336029053,
      "learning_rate": 3.092174731220698e-08,
      "loss": 1.8338,
      "step": 229350
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.401682734489441,
      "learning_rate": 3.0863045442008904e-08,
      "loss": 1.7602,
      "step": 229351
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.5771435499191284,
      "learning_rate": 3.080439934169954e-08,
      "loss": 1.8456,
      "step": 229352
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.451869249343872,
      "learning_rate": 3.074580901134549e-08,
      "loss": 1.8907,
      "step": 229353
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.5969092845916748,
      "learning_rate": 3.068727445091346e-08,
      "loss": 1.9937,
      "step": 229354
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.5730620622634888,
      "learning_rate": 3.062879566047005e-08,
      "loss": 2.0093,
      "step": 229355
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.4333447217941284,
      "learning_rate": 3.057037263994866e-08,
      "loss": 1.865,
      "step": 229356
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.5929712057113647,
      "learning_rate": 3.05120053894492e-08,
      "loss": 2.022,
      "step": 229357
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.5299915075302124,
      "learning_rate": 3.045369390893837e-08,
      "loss": 2.1018,
      "step": 229358
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.3830015659332275,
      "learning_rate": 3.039543819841617e-08,
      "loss": 2.001,
      "step": 229359
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.4249227046966553,
      "learning_rate": 3.033723825788259e-08,
      "loss": 1.9619,
      "step": 229360
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.436956524848938,
      "learning_rate": 3.027909408740426e-08,
      "loss": 1.79,
      "step": 229361
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.4777934551239014,
      "learning_rate": 3.022100568694785e-08,
      "loss": 1.8165,
      "step": 229362
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.48539137840271,
      "learning_rate": 3.0162973056546694e-08,
      "loss": 2.2487,
      "step": 229363
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.4901912212371826,
      "learning_rate": 3.010499619620077e-08,
      "loss": 2.0618,
      "step": 229364
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.3965330123901367,
      "learning_rate": 3.004707510591009e-08,
      "loss": 2.1426,
      "step": 229365
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.4270037412643433,
      "learning_rate": 2.9989209785707956e-08,
      "loss": 1.9051,
      "step": 229366
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.4389621019363403,
      "learning_rate": 2.993140023556107e-08,
      "loss": 1.8403,
      "step": 229367
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.6040897369384766,
      "learning_rate": 2.9873646455536026e-08,
      "loss": 2.0219,
      "step": 229368
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.3969486951828003,
      "learning_rate": 2.9815948445632845e-08,
      "loss": 2.1526,
      "step": 229369
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.4099160432815552,
      "learning_rate": 2.9758306205818206e-08,
      "loss": 1.9393,
      "step": 229370
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.3813412189483643,
      "learning_rate": 2.970071973615873e-08,
      "loss": 1.8577,
      "step": 229371
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.3662019968032837,
      "learning_rate": 2.9643189036654415e-08,
      "loss": 1.9036,
      "step": 229372
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.5616614818572998,
      "learning_rate": 2.958571410727195e-08,
      "loss": 1.9573,
      "step": 229373
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.4634134769439697,
      "learning_rate": 2.952829494804465e-08,
      "loss": 1.9649,
      "step": 229374
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.533881425857544,
      "learning_rate": 2.9470931559005817e-08,
      "loss": 2.0299,
      "step": 229375
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.4516557455062866,
      "learning_rate": 2.9413623940155452e-08,
      "loss": 2.0786,
      "step": 229376
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.3835563659667969,
      "learning_rate": 2.9356372091493552e-08,
      "loss": 2.0794,
      "step": 229377
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.70637845993042,
      "learning_rate": 2.9299176013053426e-08,
      "loss": 2.0999,
      "step": 229378
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.4639360904693604,
      "learning_rate": 2.9242035704801768e-08,
      "loss": 2.1337,
      "step": 229379
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.5163006782531738,
      "learning_rate": 2.9184951166805193e-08,
      "loss": 1.7737,
      "step": 229380
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.5251084566116333,
      "learning_rate": 2.912792239903039e-08,
      "loss": 1.9599,
      "step": 229381
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.468978762626648,
      "learning_rate": 2.907094940147736e-08,
      "loss": 1.9248,
      "step": 229382
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.5190707445144653,
      "learning_rate": 2.9014032174212722e-08,
      "loss": 2.0098,
      "step": 229383
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.5186541080474854,
      "learning_rate": 2.895717071720316e-08,
      "loss": 1.8024,
      "step": 229384
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.438987135887146,
      "learning_rate": 2.8900365030481987e-08,
      "loss": 1.9939,
      "step": 229385
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.4519784450531006,
      "learning_rate": 2.88436151140492e-08,
      "loss": 1.7819,
      "step": 229386
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.5268903970718384,
      "learning_rate": 2.878692096793811e-08,
      "loss": 1.8153,
      "step": 229387
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.621323585510254,
      "learning_rate": 2.87302825920821e-08,
      "loss": 1.9092,
      "step": 229388
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.7736941576004028,
      "learning_rate": 2.8673699986614394e-08,
      "loss": 1.9563,
      "step": 229389
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.4873169660568237,
      "learning_rate": 2.861717315143508e-08,
      "loss": 2.006,
      "step": 229390
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.4530723094940186,
      "learning_rate": 2.8560702086610764e-08,
      "loss": 1.9369,
      "step": 229391
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.438059687614441,
      "learning_rate": 2.850428679214145e-08,
      "loss": 1.925,
      "step": 229392
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.516349196434021,
      "learning_rate": 2.8447927268027136e-08,
      "loss": 1.7569,
      "step": 229393
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.5399531126022339,
      "learning_rate": 2.8391623514267824e-08,
      "loss": 2.1627,
      "step": 229394
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.5808900594711304,
      "learning_rate": 2.8335375530930127e-08,
      "loss": 1.9421,
      "step": 229395
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.4290273189544678,
      "learning_rate": 2.8279183317980735e-08,
      "loss": 1.962,
      "step": 229396
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.3799808025360107,
      "learning_rate": 2.8223046875419654e-08,
      "loss": 2.0468,
      "step": 229397
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.5446423292160034,
      "learning_rate": 2.8166966203280184e-08,
      "loss": 2.0494,
      "step": 229398
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.4444555044174194,
      "learning_rate": 2.8110941301595637e-08,
      "loss": 2.157,
      "step": 229399
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.542609453201294,
      "learning_rate": 2.8054972170299395e-08,
      "loss": 1.8439,
      "step": 229400
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.431954264640808,
      "learning_rate": 2.7999058809491382e-08,
      "loss": 2.0078,
      "step": 229401
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.3154693841934204,
      "learning_rate": 2.7943201219104982e-08,
      "loss": 1.8506,
      "step": 229402
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.4227104187011719,
      "learning_rate": 2.788739939920681e-08,
      "loss": 1.9641,
      "step": 229403
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.3519009351730347,
      "learning_rate": 2.7831653349796866e-08,
      "loss": 1.8468,
      "step": 229404
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.5384819507598877,
      "learning_rate": 2.7775963070841844e-08,
      "loss": 2.1192,
      "step": 229405
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.485817790031433,
      "learning_rate": 2.7720328562408355e-08,
      "loss": 1.9997,
      "step": 229406
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.3571828603744507,
      "learning_rate": 2.76647498244964e-08,
      "loss": 1.8439,
      "step": 229407
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.5310330390930176,
      "learning_rate": 2.7609226857072675e-08,
      "loss": 2.1399,
      "step": 229408
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.4870883226394653,
      "learning_rate": 2.755375966020379e-08,
      "loss": 2.2736,
      "step": 229409
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.498145580291748,
      "learning_rate": 2.7498348233889743e-08,
      "loss": 1.9407,
      "step": 229410
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.3988242149353027,
      "learning_rate": 2.7442992578097234e-08,
      "loss": 2.0328,
      "step": 229411
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.517716646194458,
      "learning_rate": 2.7387692692859565e-08,
      "loss": 2.1582,
      "step": 229412
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.3308902978897095,
      "learning_rate": 2.733244857821004e-08,
      "loss": 1.6403,
      "step": 229413
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.5519050359725952,
      "learning_rate": 2.727726023414867e-08,
      "loss": 2.0737,
      "step": 229414
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.4371793270111084,
      "learning_rate": 2.7222127660642134e-08,
      "loss": 2.1201,
      "step": 229415
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.6275279521942139,
      "learning_rate": 2.7167050857790362e-08,
      "loss": 1.8295,
      "step": 229416
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.4879437685012817,
      "learning_rate": 2.711202982549343e-08,
      "loss": 1.986,
      "step": 229417
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.499442458152771,
      "learning_rate": 2.705706456385126e-08,
      "loss": 1.9496,
      "step": 229418
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.4635101556777954,
      "learning_rate": 2.700215507286385e-08,
      "loss": 1.8924,
      "step": 229419
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.4087861776351929,
      "learning_rate": 2.694730135246459e-08,
      "loss": 1.922,
      "step": 229420
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.4524086713790894,
      "learning_rate": 2.6892503402753395e-08,
      "loss": 1.984,
      "step": 229421
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.628173828125,
      "learning_rate": 2.6837761223696963e-08,
      "loss": 1.9772,
      "step": 229422
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.4581571817398071,
      "learning_rate": 2.67830748153286e-08,
      "loss": 2.0452,
      "step": 229423
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.4142606258392334,
      "learning_rate": 2.6728444177614993e-08,
      "loss": 2.0948,
      "step": 229424
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.4186540842056274,
      "learning_rate": 2.6673869310622763e-08,
      "loss": 2.0313,
      "step": 229425
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.3406803607940674,
      "learning_rate": 2.6619350214318603e-08,
      "loss": 1.6544,
      "step": 229426
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.382523536682129,
      "learning_rate": 2.6564886888735814e-08,
      "loss": 1.9468,
      "step": 229427
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.5098931789398193,
      "learning_rate": 2.6510479333874402e-08,
      "loss": 2.223,
      "step": 229428
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.4694788455963135,
      "learning_rate": 2.645612754976767e-08,
      "loss": 1.8814,
      "step": 229429
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.455346941947937,
      "learning_rate": 2.6401831536382313e-08,
      "loss": 1.8989,
      "step": 229430
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.5537596940994263,
      "learning_rate": 2.6347591293751636e-08,
      "loss": 2.0202,
      "step": 229431
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.502179503440857,
      "learning_rate": 2.629340682187564e-08,
      "loss": 2.0519,
      "step": 229432
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.5648409128189087,
      "learning_rate": 2.6239278120787633e-08,
      "loss": 1.8139,
      "step": 229433
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.6822282075881958,
      "learning_rate": 2.6185205190487613e-08,
      "loss": 2.1453,
      "step": 229434
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.4257944822311401,
      "learning_rate": 2.613118803097558e-08,
      "loss": 2.0154,
      "step": 229435
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.6438060998916626,
      "learning_rate": 2.6077226642251537e-08,
      "loss": 1.9911,
      "step": 229436
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.6291401386260986,
      "learning_rate": 2.6023321024382092e-08,
      "loss": 1.9382,
      "step": 229437
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.4668835401535034,
      "learning_rate": 2.596947117730064e-08,
      "loss": 2.1838,
      "step": 229438
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.436582326889038,
      "learning_rate": 2.5915677101073783e-08,
      "loss": 2.0358,
      "step": 229439
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.5589067935943604,
      "learning_rate": 2.586193879570153e-08,
      "loss": 2.0875,
      "step": 229440
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.544205665588379,
      "learning_rate": 2.5808256261150572e-08,
      "loss": 2.026,
      "step": 229441
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.5764508247375488,
      "learning_rate": 2.5754629497487522e-08,
      "loss": 2.032,
      "step": 229442
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.544768214225769,
      "learning_rate": 2.5701058504679073e-08,
      "loss": 1.7925,
      "step": 229443
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.497646689414978,
      "learning_rate": 2.5647543282791838e-08,
      "loss": 1.821,
      "step": 229444
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.578908920288086,
      "learning_rate": 2.5594083831759204e-08,
      "loss": 1.763,
      "step": 229445
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.4692524671554565,
      "learning_rate": 2.5540680151647787e-08,
      "loss": 1.8461,
      "step": 229446
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.698804497718811,
      "learning_rate": 2.548733224245758e-08,
      "loss": 2.0399,
      "step": 229447
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.3519426584243774,
      "learning_rate": 2.543404010418859e-08,
      "loss": 1.8127,
      "step": 229448
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.456148624420166,
      "learning_rate": 2.5380803736840817e-08,
      "loss": 2.0852,
      "step": 229449
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.489365816116333,
      "learning_rate": 2.5327623140480867e-08,
      "loss": 2.0105,
      "step": 229450
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.445258378982544,
      "learning_rate": 2.5274498315042134e-08,
      "loss": 1.9195,
      "step": 229451
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.4170103073120117,
      "learning_rate": 2.5221429260557925e-08,
      "loss": 1.7281,
      "step": 229452
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.3842183351516724,
      "learning_rate": 2.516841597706154e-08,
      "loss": 2.0508,
      "step": 229453
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.4061450958251953,
      "learning_rate": 2.5115458464519678e-08,
      "loss": 1.8144,
      "step": 229454
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.4406803846359253,
      "learning_rate": 2.506255672299895e-08,
      "loss": 1.801,
      "step": 229455
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.5096348524093628,
      "learning_rate": 2.5009710752499356e-08,
      "loss": 2.0243,
      "step": 229456
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.400973916053772,
      "learning_rate": 2.495692055298759e-08,
      "loss": 1.7558,
      "step": 229457
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.4696029424667358,
      "learning_rate": 2.490418612449696e-08,
      "loss": 2.0929,
      "step": 229458
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.4349745512008667,
      "learning_rate": 2.485150746706077e-08,
      "loss": 1.8736,
      "step": 229459
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.556344985961914,
      "learning_rate": 2.4798884580645716e-08,
      "loss": 1.6759,
      "step": 229460
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.4584559202194214,
      "learning_rate": 2.4746317465285105e-08,
      "loss": 2.1464,
      "step": 229461
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.4343682527542114,
      "learning_rate": 2.469380612097893e-08,
      "loss": 1.9796,
      "step": 229462
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.5318338871002197,
      "learning_rate": 2.4641350547760508e-08,
      "loss": 2.0738,
      "step": 229463
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.449170470237732,
      "learning_rate": 2.4588950745629832e-08,
      "loss": 2.0374,
      "step": 229464
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.5181080102920532,
      "learning_rate": 2.4536606714586905e-08,
      "loss": 1.9555,
      "step": 229465
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.532899260520935,
      "learning_rate": 2.4484318454631724e-08,
      "loss": 2.0121,
      "step": 229466
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.6438794136047363,
      "learning_rate": 2.4432085965797598e-08,
      "loss": 1.9403,
      "step": 229467
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.5470495223999023,
      "learning_rate": 2.4379909248084527e-08,
      "loss": 2.2724,
      "step": 229468
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.4503813982009888,
      "learning_rate": 2.4327788301492513e-08,
      "loss": 1.9609,
      "step": 229469
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.3901323080062866,
      "learning_rate": 2.4275723126054857e-08,
      "loss": 1.8278,
      "step": 229470
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.5913630723953247,
      "learning_rate": 2.422371372173826e-08,
      "loss": 1.8644,
      "step": 229471
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.5364527702331543,
      "learning_rate": 2.4171760088609327e-08,
      "loss": 2.0717,
      "step": 229472
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.6283016204833984,
      "learning_rate": 2.4119862226634755e-08,
      "loss": 2.1249,
      "step": 229473
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.4023407697677612,
      "learning_rate": 2.4068020135847853e-08,
      "loss": 1.8292,
      "step": 229474
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.570387840270996,
      "learning_rate": 2.401623381624862e-08,
      "loss": 1.788,
      "step": 229475
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.6180760860443115,
      "learning_rate": 2.3964503267837053e-08,
      "loss": 1.9993,
      "step": 229476
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.5187876224517822,
      "learning_rate": 2.3912828490646464e-08,
      "loss": 2.0262,
      "step": 229477
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.4676905870437622,
      "learning_rate": 2.3861209484676846e-08,
      "loss": 2.0978,
      "step": 229478
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.4791539907455444,
      "learning_rate": 2.3809646249928204e-08,
      "loss": 2.2017,
      "step": 229479
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.464254379272461,
      "learning_rate": 2.3758138786400537e-08,
      "loss": 1.9193,
      "step": 229480
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.704070806503296,
      "learning_rate": 2.3706687094127153e-08,
      "loss": 1.9931,
      "step": 229481
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.471106767654419,
      "learning_rate": 2.365529117310805e-08,
      "loss": 1.8666,
      "step": 229482
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.4717036485671997,
      "learning_rate": 2.3603951023376533e-08,
      "loss": 1.9961,
      "step": 229483
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.2741752862930298,
      "learning_rate": 2.35526666448993e-08,
      "loss": 1.949,
      "step": 229484
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.460780143737793,
      "learning_rate": 2.3501438037709653e-08,
      "loss": 2.0534,
      "step": 229485
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.4955706596374512,
      "learning_rate": 2.3450265201807595e-08,
      "loss": 2.1083,
      "step": 229486
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.4058071374893188,
      "learning_rate": 2.3399148137193123e-08,
      "loss": 1.9151,
      "step": 229487
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.3991538286209106,
      "learning_rate": 2.3348086843932856e-08,
      "loss": 1.9279,
      "step": 229488
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.434053659439087,
      "learning_rate": 2.3297081321960177e-08,
      "loss": 2.0098,
      "step": 229489
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.5135403871536255,
      "learning_rate": 2.3246131571341696e-08,
      "loss": 2.1117,
      "step": 229490
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.3873462677001953,
      "learning_rate": 2.3195237592044114e-08,
      "loss": 2.0469,
      "step": 229491
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.5108698606491089,
      "learning_rate": 2.3144399384100732e-08,
      "loss": 2.0223,
      "step": 229492
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.56776762008667,
      "learning_rate": 2.3093616947544856e-08,
      "loss": 2.0524,
      "step": 229493
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.3916996717453003,
      "learning_rate": 2.304289028230988e-08,
      "loss": 1.9738,
      "step": 229494
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.4674805402755737,
      "learning_rate": 2.2992219388495713e-08,
      "loss": 1.9543,
      "step": 229495
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.451538324356079,
      "learning_rate": 2.2941604266069057e-08,
      "loss": 1.9582,
      "step": 229496
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.5110687017440796,
      "learning_rate": 2.2891044914996605e-08,
      "loss": 1.9279,
      "step": 229497
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.425087332725525,
      "learning_rate": 2.284054133537827e-08,
      "loss": 2.0151,
      "step": 229498
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.410531997680664,
      "learning_rate": 2.2790093527147445e-08,
      "loss": 1.9635,
      "step": 229499
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.4699571132659912,
      "learning_rate": 2.2739701490370744e-08,
      "loss": 1.9054,
      "step": 229500
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.5368810892105103,
      "learning_rate": 2.2689365225014854e-08,
      "loss": 1.9239,
      "step": 229501
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.5973161458969116,
      "learning_rate": 2.263908473107978e-08,
      "loss": 2.0436,
      "step": 229502
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.4424721002578735,
      "learning_rate": 2.2588860008632136e-08,
      "loss": 2.1463,
      "step": 229503
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.474707841873169,
      "learning_rate": 2.2538691057638615e-08,
      "loss": 2.1981,
      "step": 229504
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.462350606918335,
      "learning_rate": 2.2488577878099212e-08,
      "loss": 1.8776,
      "step": 229505
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.4481738805770874,
      "learning_rate": 2.2438520470080544e-08,
      "loss": 2.0084,
      "step": 229506
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.559924840927124,
      "learning_rate": 2.2388518833516e-08,
      "loss": 1.9548,
      "step": 229507
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.4653267860412598,
      "learning_rate": 2.233857296847219e-08,
      "loss": 2.1207,
      "step": 229508
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.470301866531372,
      "learning_rate": 2.228868287491581e-08,
      "loss": 2.0695,
      "step": 229509
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.3313050270080566,
      "learning_rate": 2.2238848552913468e-08,
      "loss": 2.0076,
      "step": 229510
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.462007761001587,
      "learning_rate": 2.2189070002398556e-08,
      "loss": 1.955,
      "step": 229511
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.535045862197876,
      "learning_rate": 2.2139347223437688e-08,
      "loss": 2.0362,
      "step": 229512
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.5577784776687622,
      "learning_rate": 2.2089680216030858e-08,
      "loss": 2.0461,
      "step": 229513
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.4799703359603882,
      "learning_rate": 2.2040068980178072e-08,
      "loss": 1.9483,
      "step": 229514
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.444860816001892,
      "learning_rate": 2.1990513515912633e-08,
      "loss": 2.0572,
      "step": 229515
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.4820029735565186,
      "learning_rate": 2.1941013823201237e-08,
      "loss": 1.8425,
      "step": 229516
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.545606017112732,
      "learning_rate": 2.1891569902043883e-08,
      "loss": 1.9526,
      "step": 229517
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.3072383403778076,
      "learning_rate": 2.1842181752507183e-08,
      "loss": 1.8827,
      "step": 229518
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.5937769412994385,
      "learning_rate": 2.1792849374591136e-08,
      "loss": 1.833,
      "step": 229519
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.3702306747436523,
      "learning_rate": 2.1743572768262442e-08,
      "loss": 1.8996,
      "step": 229520
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.4256343841552734,
      "learning_rate": 2.16943519335544e-08,
      "loss": 1.9466,
      "step": 229521
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.4163938760757446,
      "learning_rate": 2.164518687050032e-08,
      "loss": 1.7513,
      "step": 229522
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.3701525926589966,
      "learning_rate": 2.1596077579066896e-08,
      "loss": 2.012,
      "step": 229523
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.5555036067962646,
      "learning_rate": 2.1547024059254126e-08,
      "loss": 2.1182,
      "step": 229524
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.5697475671768188,
      "learning_rate": 2.1498026311128625e-08,
      "loss": 1.9117,
      "step": 229525
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.4320694208145142,
      "learning_rate": 2.1449084334657085e-08,
      "loss": 1.9231,
      "step": 229526
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.56967031955719,
      "learning_rate": 2.1400198129872816e-08,
      "loss": 1.8994,
      "step": 229527
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.5089977979660034,
      "learning_rate": 2.1351367696775812e-08,
      "loss": 2.01,
      "step": 229528
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.414813756942749,
      "learning_rate": 2.130259303536608e-08,
      "loss": 1.7982,
      "step": 229529
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.3636642694473267,
      "learning_rate": 2.1253874145643612e-08,
      "loss": 1.9281,
      "step": 229530
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.5286762714385986,
      "learning_rate": 2.120521102767503e-08,
      "loss": 2.014,
      "step": 229531
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.4581738710403442,
      "learning_rate": 2.1156603681393713e-08,
      "loss": 2.0047,
      "step": 229532
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.3533014059066772,
      "learning_rate": 2.1108052106832973e-08,
      "loss": 1.8181,
      "step": 229533
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.4142355918884277,
      "learning_rate": 2.1059556304026114e-08,
      "loss": 1.7054,
      "step": 229534
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.6312006711959839,
      "learning_rate": 2.101111627297314e-08,
      "loss": 1.826,
      "step": 229535
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.667102575302124,
      "learning_rate": 2.0962732013674044e-08,
      "loss": 1.9521,
      "step": 229536
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.3644248247146606,
      "learning_rate": 2.091440352612883e-08,
      "loss": 1.7767,
      "step": 229537
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.6069350242614746,
      "learning_rate": 2.086613081037081e-08,
      "loss": 1.9375,
      "step": 229538
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.5234216451644897,
      "learning_rate": 2.0817913866366663e-08,
      "loss": 1.9771,
      "step": 229539
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.502748966217041,
      "learning_rate": 2.0769752694183017e-08,
      "loss": 1.9033,
      "step": 229540
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.581864833831787,
      "learning_rate": 2.0721647293786557e-08,
      "loss": 2.309,
      "step": 229541
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.4664334058761597,
      "learning_rate": 2.0673597665210596e-08,
      "loss": 2.1467,
      "step": 229542
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.3943126201629639,
      "learning_rate": 2.0625603808455128e-08,
      "loss": 1.976,
      "step": 229543
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.3238391876220703,
      "learning_rate": 2.0577665723520154e-08,
      "loss": 2.0675,
      "step": 229544
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.4563300609588623,
      "learning_rate": 2.0529783410438983e-08,
      "loss": 2.0552,
      "step": 229545
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.3984293937683105,
      "learning_rate": 2.0481956869211612e-08,
      "loss": 1.863,
      "step": 229546
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.4346483945846558,
      "learning_rate": 2.043418609980474e-08,
      "loss": 1.9706,
      "step": 229547
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.3505502939224243,
      "learning_rate": 2.0386471102284975e-08,
      "loss": 1.9351,
      "step": 229548
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.567324161529541,
      "learning_rate": 2.033881187661901e-08,
      "loss": 1.9833,
      "step": 229549
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.35764479637146,
      "learning_rate": 2.0291208422840154e-08,
      "loss": 2.066,
      "step": 229550
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.5985826253890991,
      "learning_rate": 2.0243660740981716e-08,
      "loss": 2.2001,
      "step": 229551
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.5127965211868286,
      "learning_rate": 2.019616883097708e-08,
      "loss": 1.8944,
      "step": 229552
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.4892929792404175,
      "learning_rate": 2.0148732692926163e-08,
      "loss": 1.9882,
      "step": 229553
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.4961049556732178,
      "learning_rate": 2.0101352326762354e-08,
      "loss": 1.8541,
      "step": 229554
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.4336873292922974,
      "learning_rate": 2.0054027732518964e-08,
      "loss": 1.9852,
      "step": 229555
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.4247629642486572,
      "learning_rate": 2.0006758910229292e-08,
      "loss": 1.7934,
      "step": 229556
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.415708303451538,
      "learning_rate": 1.995954585986004e-08,
      "loss": 1.9085,
      "step": 229557
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.4689675569534302,
      "learning_rate": 1.9912388581477813e-08,
      "loss": 1.9279,
      "step": 229558
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.4641525745391846,
      "learning_rate": 1.9865287075016002e-08,
      "loss": 2.0093,
      "step": 229559
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.4365445375442505,
      "learning_rate": 1.981824134054122e-08,
      "loss": 1.9468,
      "step": 229560
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.5885241031646729,
      "learning_rate": 1.9771251378053465e-08,
      "loss": 2.0922,
      "step": 229561
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.6300779581069946,
      "learning_rate": 1.9724317187552742e-08,
      "loss": 2.0884,
      "step": 229562
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.4316811561584473,
      "learning_rate": 1.9677438769039046e-08,
      "loss": 1.7213,
      "step": 229563
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.4368947744369507,
      "learning_rate": 1.9630616122545685e-08,
      "loss": 1.7309,
      "step": 229564
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.5250617265701294,
      "learning_rate": 1.958384924803935e-08,
      "loss": 1.9707,
      "step": 229565
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.539979338645935,
      "learning_rate": 1.9537138145586662e-08,
      "loss": 2.0707,
      "step": 229566
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.6554492712020874,
      "learning_rate": 1.9490482815121e-08,
      "loss": 2.003,
      "step": 229567
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.587915301322937,
      "learning_rate": 1.9443883256742288e-08,
      "loss": 1.9396,
      "step": 229568
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.580272912979126,
      "learning_rate": 1.939733947038391e-08,
      "loss": 1.9368,
      "step": 229569
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.4228453636169434,
      "learning_rate": 1.9350851456079174e-08,
      "loss": 1.864,
      "step": 229570
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.3992083072662354,
      "learning_rate": 1.930441921386139e-08,
      "loss": 2.0409,
      "step": 229571
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.402750849723816,
      "learning_rate": 1.9258042743697243e-08,
      "loss": 1.9981,
      "step": 229572
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.511212944984436,
      "learning_rate": 1.9211722045620047e-08,
      "loss": 1.9863,
      "step": 229573
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.511879801750183,
      "learning_rate": 1.91654571196298e-08,
      "loss": 1.7408,
      "step": 229574
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.5336326360702515,
      "learning_rate": 1.911924796575981e-08,
      "loss": 1.8365,
      "step": 229575
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.4326757192611694,
      "learning_rate": 1.9073094583976766e-08,
      "loss": 2.0105,
      "step": 229576
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.5612908601760864,
      "learning_rate": 1.902699697431398e-08,
      "loss": 2.0737,
      "step": 229577
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.7323336601257324,
      "learning_rate": 1.8980955136771448e-08,
      "loss": 1.9334,
      "step": 229578
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.404862642288208,
      "learning_rate": 1.8934969071382478e-08,
      "loss": 1.6788,
      "step": 229579
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.4727184772491455,
      "learning_rate": 1.8889038778113764e-08,
      "loss": 1.8896,
      "step": 229580
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.382374882698059,
      "learning_rate": 1.8843164257031918e-08,
      "loss": 1.9619,
      "step": 229581
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.4652538299560547,
      "learning_rate": 1.8797345508070327e-08,
      "loss": 1.964,
      "step": 229582
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.4235692024230957,
      "learning_rate": 1.8751582531295607e-08,
      "loss": 1.9995,
      "step": 229583
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.3344963788986206,
      "learning_rate": 1.8705875326707753e-08,
      "loss": 1.8902,
      "step": 229584
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.4004907608032227,
      "learning_rate": 1.866022389427346e-08,
      "loss": 1.9799,
      "step": 229585
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.4901554584503174,
      "learning_rate": 1.861462823405935e-08,
      "loss": 2.3387,
      "step": 229586
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.5181503295898438,
      "learning_rate": 1.85690883460321e-08,
      "loss": 1.9046,
      "step": 229587
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.3232898712158203,
      "learning_rate": 1.852360423022503e-08,
      "loss": 1.7056,
      "step": 229588
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.447684645652771,
      "learning_rate": 1.8478175886638135e-08,
      "loss": 1.7634,
      "step": 229589
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.3493813276290894,
      "learning_rate": 1.8432803315271417e-08,
      "loss": 1.9759,
      "step": 229590
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.4534506797790527,
      "learning_rate": 1.8387486516124873e-08,
      "loss": 2.153,
      "step": 229591
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.5800628662109375,
      "learning_rate": 1.8342225489265115e-08,
      "loss": 2.0152,
      "step": 229592
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.4522242546081543,
      "learning_rate": 1.8297020234625537e-08,
      "loss": 2.0676,
      "step": 229593
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.5113860368728638,
      "learning_rate": 1.8251870752272746e-08,
      "loss": 2.1849,
      "step": 229594
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.577064871788025,
      "learning_rate": 1.8206777042173437e-08,
      "loss": 2.1067,
      "step": 229595
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.5290573835372925,
      "learning_rate": 1.816173910432761e-08,
      "loss": 1.8955,
      "step": 229596
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.5359989404678345,
      "learning_rate": 1.811675693880188e-08,
      "loss": 2.2442,
      "step": 229597
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.5036673545837402,
      "learning_rate": 1.8071830545562936e-08,
      "loss": 1.8205,
      "step": 229598
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.5049490928649902,
      "learning_rate": 1.8026959924610786e-08,
      "loss": 1.9547,
      "step": 229599
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.4513336420059204,
      "learning_rate": 1.7982145076012034e-08,
      "loss": 2.0374,
      "step": 229600
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.4753490686416626,
      "learning_rate": 1.7937385999700072e-08,
      "loss": 2.0839,
      "step": 229601
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.4426226615905762,
      "learning_rate": 1.7892682695708206e-08,
      "loss": 1.9721,
      "step": 229602
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.4863417148590088,
      "learning_rate": 1.784803516406974e-08,
      "loss": 2.0268,
      "step": 229603
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.4131416082382202,
      "learning_rate": 1.7803443404784677e-08,
      "loss": 1.7067,
      "step": 229604
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.4003407955169678,
      "learning_rate": 1.775890741781971e-08,
      "loss": 2.0938,
      "step": 229605
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.514391303062439,
      "learning_rate": 1.7714427203241456e-08,
      "loss": 1.9218,
      "step": 229606
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.4134751558303833,
      "learning_rate": 1.76700027610166e-08,
      "loss": 2.0441,
      "step": 229607
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.425276279449463,
      "learning_rate": 1.7625634091178452e-08,
      "loss": 2.0257,
      "step": 229608
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.6990444660186768,
      "learning_rate": 1.7581321193760322e-08,
      "loss": 1.8285,
      "step": 229609
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.3602609634399414,
      "learning_rate": 1.7537064068695594e-08,
      "loss": 1.9515,
      "step": 229610
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.470760703086853,
      "learning_rate": 1.749286271605088e-08,
      "loss": 2.0871,
      "step": 229611
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.4274660348892212,
      "learning_rate": 1.7448717135792876e-08,
      "loss": 2.054,
      "step": 229612
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.3380751609802246,
      "learning_rate": 1.740462732795489e-08,
      "loss": 2.0086,
      "step": 229613
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.7480796575546265,
      "learning_rate": 1.7360593292570224e-08,
      "loss": 2.2853,
      "step": 229614
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.4295729398727417,
      "learning_rate": 1.7316615029605574e-08,
      "loss": 2.0247,
      "step": 229615
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.6550099849700928,
      "learning_rate": 1.7272692539094245e-08,
      "loss": 2.0874,
      "step": 229616
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.4598692655563354,
      "learning_rate": 1.722882582103624e-08,
      "loss": 1.9266,
      "step": 229617
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.5429792404174805,
      "learning_rate": 1.718501487543156e-08,
      "loss": 2.0489,
      "step": 229618
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.672760009765625,
      "learning_rate": 1.7141259702280196e-08,
      "loss": 1.9539,
      "step": 229619
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.3107701539993286,
      "learning_rate": 1.7097560301615466e-08,
      "loss": 1.7712,
      "step": 229620
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.4758933782577515,
      "learning_rate": 1.705391667343736e-08,
      "loss": 2.0804,
      "step": 229621
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.502549648284912,
      "learning_rate": 1.701032881774589e-08,
      "loss": 1.9537,
      "step": 229622
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.5166552066802979,
      "learning_rate": 1.6966796734574352e-08,
      "loss": 2.128,
      "step": 229623
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.5508276224136353,
      "learning_rate": 1.692332042388944e-08,
      "loss": 1.9216,
      "step": 229624
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.5607516765594482,
      "learning_rate": 1.687989988572447e-08,
      "loss": 1.9773,
      "step": 229625
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.3597190380096436,
      "learning_rate": 1.6836535120112737e-08,
      "loss": 1.9811,
      "step": 229626
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.4171520471572876,
      "learning_rate": 1.6793226126987636e-08,
      "loss": 1.8543,
      "step": 229627
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.4680331945419312,
      "learning_rate": 1.6749972906415776e-08,
      "loss": 2.0593,
      "step": 229628
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.4971394538879395,
      "learning_rate": 1.6706775458430467e-08,
      "loss": 2.1713,
      "step": 229629
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.4074569940567017,
      "learning_rate": 1.666363378296509e-08,
      "loss": 1.8611,
      "step": 229630
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.4984118938446045,
      "learning_rate": 1.662054788005296e-08,
      "loss": 2.0173,
      "step": 229631
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.558839201927185,
      "learning_rate": 1.6577517749760682e-08,
      "loss": 2.1763,
      "step": 229632
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.4643678665161133,
      "learning_rate": 1.653454339202165e-08,
      "loss": 2.02,
      "step": 229633
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.3986127376556396,
      "learning_rate": 1.6491624806869165e-08,
      "loss": 2.0956,
      "step": 229634
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.4930784702301025,
      "learning_rate": 1.6448761994303228e-08,
      "loss": 1.7979,
      "step": 229635
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.492781162261963,
      "learning_rate": 1.6405954954357147e-08,
      "loss": 1.9919,
      "step": 229636
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.6455291509628296,
      "learning_rate": 1.6363203687030922e-08,
      "loss": 1.787,
      "step": 229637
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.5444554090499878,
      "learning_rate": 1.6320508192324554e-08,
      "loss": 2.0643,
      "step": 229638
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.5623273849487305,
      "learning_rate": 1.6277868470238043e-08,
      "loss": 2.0472,
      "step": 229639
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.6815913915634155,
      "learning_rate": 1.6235284520771384e-08,
      "loss": 1.7663,
      "step": 229640
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.4598180055618286,
      "learning_rate": 1.619275634395789e-08,
      "loss": 1.809,
      "step": 229641
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.4546988010406494,
      "learning_rate": 1.6150283939830866e-08,
      "loss": 2.0439,
      "step": 229642
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.5643572807312012,
      "learning_rate": 1.6107867308323697e-08,
      "loss": 2.1674,
      "step": 229643
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.5135608911514282,
      "learning_rate": 1.6065506449502996e-08,
      "loss": 2.0646,
      "step": 229644
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.440545916557312,
      "learning_rate": 1.6023201363368765e-08,
      "loss": 2.0097,
      "step": 229645
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.4957315921783447,
      "learning_rate": 1.5980952049887697e-08,
      "loss": 1.7359,
      "step": 229646
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.4517669677734375,
      "learning_rate": 1.5938758509126405e-08,
      "loss": 1.8552,
      "step": 229647
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.4361190795898438,
      "learning_rate": 1.589662074105158e-08,
      "loss": 2.022,
      "step": 229648
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.4530807733535767,
      "learning_rate": 1.5854538745696533e-08,
      "loss": 1.7329,
      "step": 229649
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.377963662147522,
      "learning_rate": 1.5812512523027953e-08,
      "loss": 2.3443,
      "step": 229650
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.5025795698165894,
      "learning_rate": 1.577054207311246e-08,
      "loss": 2.0735,
      "step": 229651
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.6298408508300781,
      "learning_rate": 1.5728627395916737e-08,
      "loss": 1.8346,
      "step": 229652
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.4262588024139404,
      "learning_rate": 1.5686768491440792e-08,
      "loss": 2.0063,
      "step": 229653
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.6508886814117432,
      "learning_rate": 1.564496535975124e-08,
      "loss": 2.1342,
      "step": 229654
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.4587292671203613,
      "learning_rate": 1.5603218000781458e-08,
      "loss": 1.7981,
      "step": 229655
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.5085173845291138,
      "learning_rate": 1.556152641459807e-08,
      "loss": 1.702,
      "step": 229656
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.490602970123291,
      "learning_rate": 1.551989060116776e-08,
      "loss": 2.0405,
      "step": 229657
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.5341423749923706,
      "learning_rate": 1.5478310560523842e-08,
      "loss": 1.9084,
      "step": 229658
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.5516258478164673,
      "learning_rate": 1.5436786292666314e-08,
      "loss": 2.0387,
      "step": 229659
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.5417221784591675,
      "learning_rate": 1.5395317797595176e-08,
      "loss": 1.8742,
      "step": 229660
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.4015471935272217,
      "learning_rate": 1.5353905075310423e-08,
      "loss": 1.9788,
      "step": 229661
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.5548770427703857,
      "learning_rate": 1.5312548125878676e-08,
      "loss": 1.9513,
      "step": 229662
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.4194453954696655,
      "learning_rate": 1.527124694923332e-08,
      "loss": 1.899,
      "step": 229663
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.4343695640563965,
      "learning_rate": 1.5230001545407656e-08,
      "loss": 1.9104,
      "step": 229664
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.4965521097183228,
      "learning_rate": 1.518881191440169e-08,
      "loss": 2.0238,
      "step": 229665
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.5357586145401,
      "learning_rate": 1.5147678056248723e-08,
      "loss": 1.8757,
      "step": 229666
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.4049992561340332,
      "learning_rate": 1.5106599970948764e-08,
      "loss": 1.8504,
      "step": 229667
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.4812105894088745,
      "learning_rate": 1.5065577658501805e-08,
      "loss": 1.8894,
      "step": 229668
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.5162780284881592,
      "learning_rate": 1.5024611118941153e-08,
      "loss": 2.0952,
      "step": 229669
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.4329453706741333,
      "learning_rate": 1.49837003522002e-08,
      "loss": 1.8431,
      "step": 229670
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.6385926008224487,
      "learning_rate": 1.4942845358378863e-08,
      "loss": 2.1103,
      "step": 229671
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.4161344766616821,
      "learning_rate": 1.4902046137410527e-08,
      "loss": 1.9828,
      "step": 229672
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.3448313474655151,
      "learning_rate": 1.486130268936181e-08,
      "loss": 1.7621,
      "step": 229673
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.4951661825180054,
      "learning_rate": 1.48206150141994e-08,
      "loss": 1.9888,
      "step": 229674
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.4365487098693848,
      "learning_rate": 1.47799831119233e-08,
      "loss": 2.0862,
      "step": 229675
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.522062063217163,
      "learning_rate": 1.4739406982600121e-08,
      "loss": 2.0272,
      "step": 229676
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.48392653465271,
      "learning_rate": 1.4698886626163253e-08,
      "loss": 1.9998,
      "step": 229677
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.3863145112991333,
      "learning_rate": 1.4658422042679307e-08,
      "loss": 1.8461,
      "step": 229678
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.5605733394622803,
      "learning_rate": 1.4618013232148285e-08,
      "loss": 1.941,
      "step": 229679
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.4926947355270386,
      "learning_rate": 1.4577660194536878e-08,
      "loss": 1.8106,
      "step": 229680
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.4794379472732544,
      "learning_rate": 1.4537362929878394e-08,
      "loss": 2.1282,
      "step": 229681
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.5088430643081665,
      "learning_rate": 1.4497121438206139e-08,
      "loss": 1.8694,
      "step": 229682
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.4166088104248047,
      "learning_rate": 1.4456935719453499e-08,
      "loss": 1.8424,
      "step": 229683
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.5431979894638062,
      "learning_rate": 1.4416805773720396e-08,
      "loss": 1.8815,
      "step": 229684
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.5612508058547974,
      "learning_rate": 1.4376731600940217e-08,
      "loss": 1.9531,
      "step": 229685
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.541159987449646,
      "learning_rate": 1.4336713201179572e-08,
      "loss": 1.8662,
      "step": 229686
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.5428434610366821,
      "learning_rate": 1.4296750574405158e-08,
      "loss": 1.9607,
      "step": 229687
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.5209170579910278,
      "learning_rate": 1.4256843720616974e-08,
      "loss": 2.1622,
      "step": 229688
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.4672296047210693,
      "learning_rate": 1.4216992639848324e-08,
      "loss": 1.8898,
      "step": 229689
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.3696324825286865,
      "learning_rate": 1.4177197332132518e-08,
      "loss": 1.9043,
      "step": 229690
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.655748724937439,
      "learning_rate": 1.413745779740294e-08,
      "loss": 1.9427,
      "step": 229691
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.4857369661331177,
      "learning_rate": 1.4097774035726206e-08,
      "loss": 2.2906,
      "step": 229692
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.4862489700317383,
      "learning_rate": 1.4058146047102314e-08,
      "loss": 2.0646,
      "step": 229693
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.4789330959320068,
      "learning_rate": 1.4018573831497959e-08,
      "loss": 2.117,
      "step": 229694
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.4431484937667847,
      "learning_rate": 1.3979057388979753e-08,
      "loss": 2.0171,
      "step": 229695
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.617526650428772,
      "learning_rate": 1.3939596719481083e-08,
      "loss": 2.1097,
      "step": 229696
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.2535971403121948,
      "learning_rate": 1.390019182310187e-08,
      "loss": 1.9311,
      "step": 229697
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.3655261993408203,
      "learning_rate": 1.3860842699775499e-08,
      "loss": 2.01,
      "step": 229698
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.5234203338623047,
      "learning_rate": 1.3821549349535277e-08,
      "loss": 1.8031,
      "step": 229699
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.4668546915054321,
      "learning_rate": 1.3782311772414512e-08,
      "loss": 1.7631,
      "step": 229700
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.452852725982666,
      "learning_rate": 1.3743129968379896e-08,
      "loss": 1.8902,
      "step": 229701
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.494244933128357,
      "learning_rate": 1.370400393743143e-08,
      "loss": 1.996,
      "step": 229702
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.4674168825149536,
      "learning_rate": 1.3664933679635726e-08,
      "loss": 2.005,
      "step": 229703
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.5957436561584473,
      "learning_rate": 1.3625919194926172e-08,
      "loss": 1.9605,
      "step": 229704
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.4798035621643066,
      "learning_rate": 1.358696048336938e-08,
      "loss": 1.9569,
      "step": 229705
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.4494892358779907,
      "learning_rate": 1.3548057544932046e-08,
      "loss": 2.1653,
      "step": 229706
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.4783738851547241,
      "learning_rate": 1.3509210379647472e-08,
      "loss": 1.9664,
      "step": 229707
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.4163929224014282,
      "learning_rate": 1.3470418987515663e-08,
      "loss": 1.9021,
      "step": 229708
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.4218246936798096,
      "learning_rate": 1.3431683368536616e-08,
      "loss": 2.0072,
      "step": 229709
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.5003769397735596,
      "learning_rate": 1.339300352274364e-08,
      "loss": 1.8361,
      "step": 229710
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.6803679466247559,
      "learning_rate": 1.3354379450103425e-08,
      "loss": 2.0262,
      "step": 229711
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.450218677520752,
      "learning_rate": 1.331581115064928e-08,
      "loss": 2.0747,
      "step": 229712
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.5989829301834106,
      "learning_rate": 1.3277298624381204e-08,
      "loss": 2.01,
      "step": 229713
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.6370996236801147,
      "learning_rate": 1.3238841871332507e-08,
      "loss": 2.2002,
      "step": 229714
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.4336611032485962,
      "learning_rate": 1.320044089143657e-08,
      "loss": 1.8149,
      "step": 229715
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.4480319023132324,
      "learning_rate": 1.3162095684793316e-08,
      "loss": 1.9025,
      "step": 229716
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.4604027271270752,
      "learning_rate": 1.3123806251336132e-08,
      "loss": 1.9796,
      "step": 229717
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.5432212352752686,
      "learning_rate": 1.3085572591098326e-08,
      "loss": 1.9911,
      "step": 229718
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.447404146194458,
      "learning_rate": 1.3047394704113201e-08,
      "loss": 2.0558,
      "step": 229719
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.354607343673706,
      "learning_rate": 1.3009272590347453e-08,
      "loss": 1.8428,
      "step": 229720
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.537914752960205,
      "learning_rate": 1.2971206249867694e-08,
      "loss": 2.2266,
      "step": 229721
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.3536828756332397,
      "learning_rate": 1.2933195682574005e-08,
      "loss": 2.039,
      "step": 229722
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.4523454904556274,
      "learning_rate": 1.2895240888599611e-08,
      "loss": 1.9471,
      "step": 229723
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.4896135330200195,
      "learning_rate": 1.2857341867844595e-08,
      "loss": 2.0443,
      "step": 229724
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.3935930728912354,
      "learning_rate": 1.2819498620375568e-08,
      "loss": 2.1057,
      "step": 229725
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.6669622659683228,
      "learning_rate": 1.278171114619253e-08,
      "loss": 2.0983,
      "step": 229726
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.4968104362487793,
      "learning_rate": 1.2743979445295482e-08,
      "loss": 2.14,
      "step": 229727
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.3568099737167358,
      "learning_rate": 1.2706303517684424e-08,
      "loss": 1.7781,
      "step": 229728
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.5115495920181274,
      "learning_rate": 1.2668683363392663e-08,
      "loss": 2.0934,
      "step": 229729
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.4508445262908936,
      "learning_rate": 1.263111898238689e-08,
      "loss": 1.8999,
      "step": 229730
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.4868525266647339,
      "learning_rate": 1.259361037473372e-08,
      "loss": 1.9335,
      "step": 229731
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.4723336696624756,
      "learning_rate": 1.255615754036654e-08,
      "loss": 2.0807,
      "step": 229732
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.4968886375427246,
      "learning_rate": 1.2518760479318657e-08,
      "loss": 2.1768,
      "step": 229733
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.4945117235183716,
      "learning_rate": 1.2481419191623376e-08,
      "loss": 2.0535,
      "step": 229734
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.5072340965270996,
      "learning_rate": 1.2444133677280699e-08,
      "loss": 2.0197,
      "step": 229735
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.6690783500671387,
      "learning_rate": 1.2406903936257317e-08,
      "loss": 2.0871,
      "step": 229736
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.5013160705566406,
      "learning_rate": 1.2369729968619846e-08,
      "loss": 1.9806,
      "step": 229737
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.3362956047058105,
      "learning_rate": 1.2332611774334977e-08,
      "loss": 1.9665,
      "step": 229738
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.5843594074249268,
      "learning_rate": 1.2295549353436018e-08,
      "loss": 2.0032,
      "step": 229739
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.5938302278518677,
      "learning_rate": 1.2258542705889662e-08,
      "loss": 1.8863,
      "step": 229740
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.5030509233474731,
      "learning_rate": 1.2221591831729216e-08,
      "loss": 2.1368,
      "step": 229741
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.502780556678772,
      "learning_rate": 1.2184696730954679e-08,
      "loss": 2.1012,
      "step": 229742
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.4365345239639282,
      "learning_rate": 1.2147857403599359e-08,
      "loss": 2.0428,
      "step": 229743
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.4900095462799072,
      "learning_rate": 1.2111073849629949e-08,
      "loss": 2.043,
      "step": 229744
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.421466588973999,
      "learning_rate": 1.2074346069079754e-08,
      "loss": 2.0287,
      "step": 229745
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.5603309869766235,
      "learning_rate": 1.2037674061948777e-08,
      "loss": 1.8813,
      "step": 229746
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.4027845859527588,
      "learning_rate": 1.2001057828237016e-08,
      "loss": 1.8988,
      "step": 229747
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.3783105611801147,
      "learning_rate": 1.1964497367977777e-08,
      "loss": 1.8089,
      "step": 229748
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.4076199531555176,
      "learning_rate": 1.1927992681137754e-08,
      "loss": 2.0432,
      "step": 229749
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.5627175569534302,
      "learning_rate": 1.1891543767750256e-08,
      "loss": 1.8037,
      "step": 229750
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.6118662357330322,
      "learning_rate": 1.1855150627815279e-08,
      "loss": 1.793,
      "step": 229751
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.5273373126983643,
      "learning_rate": 1.1818813261332827e-08,
      "loss": 2.0345,
      "step": 229752
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.4668818712234497,
      "learning_rate": 1.1782531668302897e-08,
      "loss": 1.9745,
      "step": 229753
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.6082483530044556,
      "learning_rate": 1.1746305848758796e-08,
      "loss": 2.2205,
      "step": 229754
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.5468592643737793,
      "learning_rate": 1.1710135802700527e-08,
      "loss": 1.7418,
      "step": 229755
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.462009310722351,
      "learning_rate": 1.1674021530128086e-08,
      "loss": 2.002,
      "step": 229756
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.5838409662246704,
      "learning_rate": 1.1637963031041476e-08,
      "loss": 1.8999,
      "step": 229757
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.4380500316619873,
      "learning_rate": 1.1601960305474001e-08,
      "loss": 1.9073,
      "step": 229758
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.4052588939666748,
      "learning_rate": 1.1566013353392356e-08,
      "loss": 1.9981,
      "step": 229759
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.4476008415222168,
      "learning_rate": 1.1530122174796542e-08,
      "loss": 1.9482,
      "step": 229760
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.3829433917999268,
      "learning_rate": 1.1494286769753169e-08,
      "loss": 2.0406,
      "step": 229761
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.3463268280029297,
      "learning_rate": 1.1458507138228933e-08,
      "loss": 1.8014,
      "step": 229762
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.4955368041992188,
      "learning_rate": 1.1422783280257142e-08,
      "loss": 1.9477,
      "step": 229763
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.4043409824371338,
      "learning_rate": 1.1387115195804485e-08,
      "loss": 1.9829,
      "step": 229764
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.448595404624939,
      "learning_rate": 1.1351502884904273e-08,
      "loss": 2.2418,
      "step": 229765
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.5128388404846191,
      "learning_rate": 1.1315946347556504e-08,
      "loss": 1.8351,
      "step": 229766
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.4798442125320435,
      "learning_rate": 1.1280445583761178e-08,
      "loss": 1.9462,
      "step": 229767
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.4528553485870361,
      "learning_rate": 1.1245000593518295e-08,
      "loss": 2.005,
      "step": 229768
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.5352075099945068,
      "learning_rate": 1.120961137686116e-08,
      "loss": 1.9797,
      "step": 229769
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.4370050430297852,
      "learning_rate": 1.1174277933789778e-08,
      "loss": 1.8993,
      "step": 229770
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.4588223695755005,
      "learning_rate": 1.1139000264304144e-08,
      "loss": 1.9422,
      "step": 229771
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.6293669939041138,
      "learning_rate": 1.1103778368404258e-08,
      "loss": 1.9299,
      "step": 229772
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.402344822883606,
      "learning_rate": 1.1068612246090125e-08,
      "loss": 1.8261,
      "step": 229773
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.5535893440246582,
      "learning_rate": 1.1033501897395048e-08,
      "loss": 1.8288,
      "step": 229774
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.6756064891815186,
      "learning_rate": 1.0998447322319026e-08,
      "loss": 2.0669,
      "step": 229775
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.4239197969436646,
      "learning_rate": 1.0963448520862062e-08,
      "loss": 1.7596,
      "step": 229776
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.4409903287887573,
      "learning_rate": 1.0928505493024154e-08,
      "loss": 2.1001,
      "step": 229777
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.707220196723938,
      "learning_rate": 1.0893618238805301e-08,
      "loss": 1.937,
      "step": 229778
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.475156545639038,
      "learning_rate": 1.0858786758238814e-08,
      "loss": 2.0384,
      "step": 229779
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.4090467691421509,
      "learning_rate": 1.0824011051291382e-08,
      "loss": 2.127,
      "step": 229780
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.518153190612793,
      "learning_rate": 1.0789291118029619e-08,
      "loss": 1.9241,
      "step": 229781
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.440102219581604,
      "learning_rate": 1.0754626958420221e-08,
      "loss": 1.7868,
      "step": 229782
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.5483946800231934,
      "learning_rate": 1.0720018572463184e-08,
      "loss": 1.7425,
      "step": 229783
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.4339489936828613,
      "learning_rate": 1.0685465960158512e-08,
      "loss": 2.0048,
      "step": 229784
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.4643791913986206,
      "learning_rate": 1.0650969121539509e-08,
      "loss": 1.8226,
      "step": 229785
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.411746859550476,
      "learning_rate": 1.0616528056639484e-08,
      "loss": 1.7666,
      "step": 229786
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.5591634511947632,
      "learning_rate": 1.0582142765391821e-08,
      "loss": 2.2666,
      "step": 229787
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.441479206085205,
      "learning_rate": 1.0547813247829828e-08,
      "loss": 2.1943,
      "step": 229788
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.5449659824371338,
      "learning_rate": 1.0513539503986812e-08,
      "loss": 2.1982,
      "step": 229789
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.4403467178344727,
      "learning_rate": 1.0479321533862771e-08,
      "loss": 2.0395,
      "step": 229790
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.4364765882492065,
      "learning_rate": 1.0445159337457709e-08,
      "loss": 1.8107,
      "step": 229791
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.4500244855880737,
      "learning_rate": 1.0411052914738315e-08,
      "loss": 1.9154,
      "step": 229792
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.359268307685852,
      "learning_rate": 1.0377002265771207e-08,
      "loss": 1.7558,
      "step": 229793
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.4320250749588013,
      "learning_rate": 1.034300739055638e-08,
      "loss": 1.8419,
      "step": 229794
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.4540653228759766,
      "learning_rate": 1.0309068289027223e-08,
      "loss": 1.7363,
      "step": 229795
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.5242812633514404,
      "learning_rate": 1.0275184961283656e-08,
      "loss": 1.839,
      "step": 229796
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.5601012706756592,
      "learning_rate": 1.0241357407292372e-08,
      "loss": 2.1407,
      "step": 229797
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.5225169658660889,
      "learning_rate": 1.0207585627053372e-08,
      "loss": 1.9053,
      "step": 229798
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.5332058668136597,
      "learning_rate": 1.0173869620566655e-08,
      "loss": 2.1964,
      "step": 229799
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.4336695671081543,
      "learning_rate": 1.0140209387865527e-08,
      "loss": 1.9589,
      "step": 229800
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.634955644607544,
      "learning_rate": 1.010660492894999e-08,
      "loss": 1.8273,
      "step": 229801
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.5181303024291992,
      "learning_rate": 1.0073056243820043e-08,
      "loss": 1.9091,
      "step": 229802
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.6964149475097656,
      "learning_rate": 1.0039563332475686e-08,
      "loss": 2.1647,
      "step": 229803
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.5046422481536865,
      "learning_rate": 1.000612619491692e-08,
      "loss": 1.9742,
      "step": 229804
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.4262324571609497,
      "learning_rate": 9.972744831143741e-09,
      "loss": 2.2322,
      "step": 229805
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.4441120624542236,
      "learning_rate": 9.939419241222768e-09,
      "loss": 1.7714,
      "step": 229806
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.450821042060852,
      "learning_rate": 9.906149425087385e-09,
      "loss": 1.9574,
      "step": 229807
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.4734399318695068,
      "learning_rate": 9.872935382770898e-09,
      "loss": 1.8832,
      "step": 229808
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.5828256607055664,
      "learning_rate": 9.839777114306612e-09,
      "loss": 1.9712,
      "step": 229809
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.5899349451065063,
      "learning_rate": 9.806674619661225e-09,
      "loss": 2.0681,
      "step": 229810
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.763412356376648,
      "learning_rate": 9.773627898834735e-09,
      "loss": 1.8872,
      "step": 229811
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.4807695150375366,
      "learning_rate": 9.740636951893754e-09,
      "loss": 2.057,
      "step": 229812
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.4634981155395508,
      "learning_rate": 9.707701778771671e-09,
      "loss": 2.1032,
      "step": 229813
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.479184627532959,
      "learning_rate": 9.674822379535097e-09,
      "loss": 2.0308,
      "step": 229814
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.6162408590316772,
      "learning_rate": 9.641998754150726e-09,
      "loss": 2.1461,
      "step": 229815
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.4335585832595825,
      "learning_rate": 9.609230902618559e-09,
      "loss": 1.944,
      "step": 229816
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.4444825649261475,
      "learning_rate": 9.576518825005207e-09,
      "loss": 2.259,
      "step": 229817
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.5188114643096924,
      "learning_rate": 9.543862521244061e-09,
      "loss": 2.0191,
      "step": 229818
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.7131280899047852,
      "learning_rate": 9.511261991368424e-09,
      "loss": 1.9889,
      "step": 229819
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.4292421340942383,
      "learning_rate": 9.478717235378297e-09,
      "loss": 1.7793,
      "step": 229820
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.4199528694152832,
      "learning_rate": 9.446228253306987e-09,
      "loss": 1.8996,
      "step": 229821
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.537164568901062,
      "learning_rate": 9.413795045154493e-09,
      "loss": 2.0224,
      "step": 229822
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.5846813917160034,
      "learning_rate": 9.381417610887509e-09,
      "loss": 2.0433,
      "step": 229823
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.5541638135910034,
      "learning_rate": 9.349095950572649e-09,
      "loss": 2.0088,
      "step": 229824
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.4960306882858276,
      "learning_rate": 9.316830064143299e-09,
      "loss": 1.8902,
      "step": 229825
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.6164301633834839,
      "learning_rate": 9.284619951666073e-09,
      "loss": 2.1805,
      "step": 229826
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.3609287738800049,
      "learning_rate": 9.252465613140969e-09,
      "loss": 1.9523,
      "step": 229827
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.578456997871399,
      "learning_rate": 9.220367048534682e-09,
      "loss": 2.2776,
      "step": 229828
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.423604965209961,
      "learning_rate": 9.188324257880519e-09,
      "loss": 1.8228,
      "step": 229829
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.5041307210922241,
      "learning_rate": 9.15633724117848e-09,
      "loss": 2.1702,
      "step": 229830
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.3240370750427246,
      "learning_rate": 9.12440599846187e-09,
      "loss": 1.9637,
      "step": 229831
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.5729979276657104,
      "learning_rate": 9.092530529664076e-09,
      "loss": 2.0889,
      "step": 229832
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.5421680212020874,
      "learning_rate": 9.060710834885022e-09,
      "loss": 2.0055,
      "step": 229833
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.4588711261749268,
      "learning_rate": 9.028946914058089e-09,
      "loss": 1.8462,
      "step": 229834
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.380663275718689,
      "learning_rate": 8.997238767216585e-09,
      "loss": 2.0736,
      "step": 229835
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.477301836013794,
      "learning_rate": 8.96558639439382e-09,
      "loss": 1.8912,
      "step": 229836
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.354904294013977,
      "learning_rate": 8.933989795523177e-09,
      "loss": 1.8281,
      "step": 229837
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.429133415222168,
      "learning_rate": 8.902448970704578e-09,
      "loss": 2.1378,
      "step": 229838
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.4168113470077515,
      "learning_rate": 8.870963919838103e-09,
      "loss": 1.8574,
      "step": 229839
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.6223317384719849,
      "learning_rate": 8.839534643023671e-09,
      "loss": 2.1861,
      "step": 229840
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.4176664352416992,
      "learning_rate": 8.808161140227976e-09,
      "loss": 2.0466,
      "step": 229841
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.4902679920196533,
      "learning_rate": 8.776843411451018e-09,
      "loss": 1.9605,
      "step": 229842
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.398909091949463,
      "learning_rate": 8.745581456692797e-09,
      "loss": 1.8746,
      "step": 229843
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.4879733324050903,
      "learning_rate": 8.714375275986618e-09,
      "loss": 2.1223,
      "step": 229844
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.3321961164474487,
      "learning_rate": 8.683224869299176e-09,
      "loss": 1.8478,
      "step": 229845
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.375884771347046,
      "learning_rate": 8.652130236697086e-09,
      "loss": 1.9394,
      "step": 229846
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.465880274772644,
      "learning_rate": 8.621091378113732e-09,
      "loss": 1.9473,
      "step": 229847
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.443526268005371,
      "learning_rate": 8.590108293615727e-09,
      "loss": 1.9142,
      "step": 229848
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.6550633907318115,
      "learning_rate": 8.559180983169767e-09,
      "loss": 1.9401,
      "step": 229849
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.5014601945877075,
      "learning_rate": 8.528309446809156e-09,
      "loss": 1.9632,
      "step": 229850
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.5610268115997314,
      "learning_rate": 8.497493684533897e-09,
      "loss": 2.2092,
      "step": 229851
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.4932739734649658,
      "learning_rate": 8.466733696310679e-09,
      "loss": 1.8811,
      "step": 229852
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.5263326168060303,
      "learning_rate": 8.436029482206119e-09,
      "loss": 2.0289,
      "step": 229853
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.3834271430969238,
      "learning_rate": 8.40538104218691e-09,
      "loss": 2.0954,
      "step": 229854
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.4952080249786377,
      "learning_rate": 8.37478837625305e-09,
      "loss": 1.9322,
      "step": 229855
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.4528095722198486,
      "learning_rate": 8.344251484437847e-09,
      "loss": 1.9085,
      "step": 229856
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.379123568534851,
      "learning_rate": 8.313770366707994e-09,
      "loss": 1.8777,
      "step": 229857
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.4683821201324463,
      "learning_rate": 8.283345023130105e-09,
      "loss": 1.9766,
      "step": 229858
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.4540380239486694,
      "learning_rate": 8.252975453670873e-09,
      "loss": 1.968,
      "step": 229859
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.2839797735214233,
      "learning_rate": 8.222661658330298e-09,
      "loss": 2.1506,
      "step": 229860
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.4415578842163086,
      "learning_rate": 8.192403637141686e-09,
      "loss": 2.1494,
      "step": 229861
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.5621061325073242,
      "learning_rate": 8.162201390071732e-09,
      "loss": 2.0564,
      "step": 229862
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.4816372394561768,
      "learning_rate": 8.132054917153741e-09,
      "loss": 2.0591,
      "step": 229863
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.597102165222168,
      "learning_rate": 8.101964218387713e-09,
      "loss": 2.0969,
      "step": 229864
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.6151355504989624,
      "learning_rate": 8.07192929377365e-09,
      "loss": 1.956,
      "step": 229865
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.524821162223816,
      "learning_rate": 8.041950143311548e-09,
      "loss": 2.1749,
      "step": 229866
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.523630976676941,
      "learning_rate": 8.012026767034718e-09,
      "loss": 2.0039,
      "step": 229867
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.5937975645065308,
      "learning_rate": 7.982159164943159e-09,
      "loss": 2.0777,
      "step": 229868
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.4561628103256226,
      "learning_rate": 7.952347337003561e-09,
      "loss": 2.1005,
      "step": 229869
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.4648431539535522,
      "learning_rate": 7.922591283249235e-09,
      "loss": 1.979,
      "step": 229870
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.7335667610168457,
      "learning_rate": 7.892891003713487e-09,
      "loss": 1.8494,
      "step": 229871
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.4413856267929077,
      "learning_rate": 7.8632464983297e-09,
      "loss": 2.0882,
      "step": 229872
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.461049199104309,
      "learning_rate": 7.833657767197798e-09,
      "loss": 1.7947,
      "step": 229873
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.395884394645691,
      "learning_rate": 7.80412481021786e-09,
      "loss": 1.9544,
      "step": 229874
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.46610426902771,
      "learning_rate": 7.774647627489804e-09,
      "loss": 1.9084,
      "step": 229875
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.5762032270431519,
      "learning_rate": 7.745226218980328e-09,
      "loss": 2.1439,
      "step": 229876
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.3860225677490234,
      "learning_rate": 7.715860584689426e-09,
      "loss": 2.049,
      "step": 229877
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.3283699750900269,
      "learning_rate": 7.686550724617102e-09,
      "loss": 1.8626,
      "step": 229878
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.4191343784332275,
      "learning_rate": 7.657296638763355e-09,
      "loss": 2.0157,
      "step": 229879
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.5183708667755127,
      "learning_rate": 7.628098327161492e-09,
      "loss": 2.176,
      "step": 229880
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.4486702680587769,
      "learning_rate": 7.598955789811512e-09,
      "loss": 2.0612,
      "step": 229881
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.3679341077804565,
      "learning_rate": 7.569869026713415e-09,
      "loss": 1.7256,
      "step": 229882
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.6315137147903442,
      "learning_rate": 7.540838037867203e-09,
      "loss": 1.9561,
      "step": 229883
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.5886214971542358,
      "learning_rate": 7.511862823272874e-09,
      "loss": 2.0642,
      "step": 229884
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.4273346662521362,
      "learning_rate": 7.482943382963736e-09,
      "loss": 1.8932,
      "step": 229885
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.4028183221817017,
      "learning_rate": 7.45407971690648e-09,
      "loss": 1.9703,
      "step": 229886
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.5465303659439087,
      "learning_rate": 7.425271825134416e-09,
      "loss": 1.822,
      "step": 229887
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.5214786529541016,
      "learning_rate": 7.396519707647541e-09,
      "loss": 1.7911,
      "step": 229888
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.4688966274261475,
      "learning_rate": 7.3678233644458575e-09,
      "loss": 2.2674,
      "step": 229889
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.5258989334106445,
      "learning_rate": 7.339182795562671e-09,
      "loss": 1.9645,
      "step": 229890
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.418123722076416,
      "learning_rate": 7.310598000931367e-09,
      "loss": 2.3135,
      "step": 229891
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.3983123302459717,
      "learning_rate": 7.282068980618561e-09,
      "loss": 1.9111,
      "step": 229892
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.4130104780197144,
      "learning_rate": 7.253595734624251e-09,
      "loss": 1.718,
      "step": 229893
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.5960693359375,
      "learning_rate": 7.225178262948439e-09,
      "loss": 1.9282,
      "step": 229894
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.7628898620605469,
      "learning_rate": 7.196816565591124e-09,
      "loss": 1.9806,
      "step": 229895
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.4948194026947021,
      "learning_rate": 7.1685106425523055e-09,
      "loss": 2.1364,
      "step": 229896
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.6647857427597046,
      "learning_rate": 7.140260493865291e-09,
      "loss": 1.9578,
      "step": 229897
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.4259486198425293,
      "learning_rate": 7.112066119463466e-09,
      "loss": 2.0036,
      "step": 229898
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.4975721836090088,
      "learning_rate": 7.083927519446752e-09,
      "loss": 1.9887,
      "step": 229899
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.420656681060791,
      "learning_rate": 7.055844693748536e-09,
      "loss": 2.0029,
      "step": 229900
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.5212149620056152,
      "learning_rate": 7.027817642435429e-09,
      "loss": 1.9376,
      "step": 229901
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.4632477760314941,
      "learning_rate": 6.99984636544082e-09,
      "loss": 2.0293,
      "step": 229902
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.3469738960266113,
      "learning_rate": 6.9719308628313206e-09,
      "loss": 1.8502,
      "step": 229903
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.6408390998840332,
      "learning_rate": 6.944071134573625e-09,
      "loss": 2.1464,
      "step": 229904
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.4479076862335205,
      "learning_rate": 6.9162671807010405e-09,
      "loss": 2.0693,
      "step": 229905
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.511392593383789,
      "learning_rate": 6.888519001213566e-09,
      "loss": 2.0254,
      "step": 229906
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.609894871711731,
      "learning_rate": 6.860826596077895e-09,
      "loss": 1.9396,
      "step": 229907
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.3706105947494507,
      "learning_rate": 6.8331899653606414e-09,
      "loss": 2.0859,
      "step": 229908
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.6252636909484863,
      "learning_rate": 6.805609109028498e-09,
      "loss": 2.0289,
      "step": 229909
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.423717975616455,
      "learning_rate": 6.7780840270814655e-09,
      "loss": 2.0485,
      "step": 229910
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.5165437459945679,
      "learning_rate": 6.750614719552849e-09,
      "loss": 2.0399,
      "step": 229911
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.4378564357757568,
      "learning_rate": 6.723201186409344e-09,
      "loss": 2.1512,
      "step": 229912
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.5010017156600952,
      "learning_rate": 6.695843427684255e-09,
      "loss": 1.9408,
      "step": 229913
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.5350431203842163,
      "learning_rate": 6.6685414434108904e-09,
      "loss": 1.9636,
      "step": 229914
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.4230674505233765,
      "learning_rate": 6.641295233522637e-09,
      "loss": 1.9268,
      "step": 229915
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.6164835691452026,
      "learning_rate": 6.614104798086106e-09,
      "loss": 2.1834,
      "step": 229916
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.4703195095062256,
      "learning_rate": 6.586970137067992e-09,
      "loss": 1.7931,
      "step": 229917
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.6040763854980469,
      "learning_rate": 6.5598912504682965e-09,
      "loss": 1.8071,
      "step": 229918
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.4972771406173706,
      "learning_rate": 6.532868138353631e-09,
      "loss": 2.1633,
      "step": 229919
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.4672763347625732,
      "learning_rate": 6.505900800657382e-09,
      "loss": 1.7373,
      "step": 229920
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.3967729806900024,
      "learning_rate": 6.478989237446164e-09,
      "loss": 2.0303,
      "step": 229921
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.3923296928405762,
      "learning_rate": 6.452133448653362e-09,
      "loss": 1.9162,
      "step": 229922
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.5284907817840576,
      "learning_rate": 6.425333434345592e-09,
      "loss": 2.0263,
      "step": 229923
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.4573460817337036,
      "learning_rate": 6.398589194522852e-09,
      "loss": 2.0975,
      "step": 229924
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.5088298320770264,
      "learning_rate": 6.3719007291518355e-09,
      "loss": 1.8795,
      "step": 229925
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.4235721826553345,
      "learning_rate": 6.345268038265849e-09,
      "loss": 2.0713,
      "step": 229926
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.4418739080429077,
      "learning_rate": 6.318691121864894e-09,
      "loss": 1.8693,
      "step": 229927
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.62972092628479,
      "learning_rate": 6.292169979948969e-09,
      "loss": 2.0641,
      "step": 229928
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.5307064056396484,
      "learning_rate": 6.265704612518075e-09,
      "loss": 2.2327,
      "step": 229929
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.5339014530181885,
      "learning_rate": 6.23929501957221e-09,
      "loss": 1.9524,
      "step": 229930
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.5380700826644897,
      "learning_rate": 6.212941201144683e-09,
      "loss": 2.1695,
      "step": 229931
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.4052610397338867,
      "learning_rate": 6.186643157235493e-09,
      "loss": 1.9878,
      "step": 229932
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.4546778202056885,
      "learning_rate": 6.160400887844641e-09,
      "loss": 2.1258,
      "step": 229933
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.3998944759368896,
      "learning_rate": 6.1342143929388184e-09,
      "loss": 1.8159,
      "step": 229934
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.475732684135437,
      "learning_rate": 6.108083672584641e-09,
      "loss": 1.8638,
      "step": 229935
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.463273525238037,
      "learning_rate": 6.082008726748799e-09,
      "loss": 2.0484,
      "step": 229936
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.3062447309494019,
      "learning_rate": 6.055989555464602e-09,
      "loss": 1.9551,
      "step": 229937
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.4874502420425415,
      "learning_rate": 6.030026158698742e-09,
      "loss": 1.9958,
      "step": 229938
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.4620168209075928,
      "learning_rate": 6.004118536451219e-09,
      "loss": 1.8166,
      "step": 229939
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.5064243078231812,
      "learning_rate": 5.978266688788647e-09,
      "loss": 1.9298,
      "step": 229940
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.4747997522354126,
      "learning_rate": 5.952470615644411e-09,
      "loss": 2.1853,
      "step": 229941
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.4704750776290894,
      "learning_rate": 5.926730317085127e-09,
      "loss": 1.7692,
      "step": 229942
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.5028736591339111,
      "learning_rate": 5.901045793077486e-09,
      "loss": 2.0581,
      "step": 229943
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.446946382522583,
      "learning_rate": 5.875417043621489e-09,
      "loss": 1.9794,
      "step": 229944
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.505084753036499,
      "learning_rate": 5.849844068750442e-09,
      "loss": 2.0526,
      "step": 229945
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.451208233833313,
      "learning_rate": 5.824326868464346e-09,
      "loss": 1.9288,
      "step": 229946
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.470159649848938,
      "learning_rate": 5.798865442729894e-09,
      "loss": 2.066,
      "step": 229947
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.4015300273895264,
      "learning_rate": 5.7734597916137e-09,
      "loss": 1.8814,
      "step": 229948
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.4944134950637817,
      "learning_rate": 5.748109915082455e-09,
      "loss": 2.0539,
      "step": 229949
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.4512887001037598,
      "learning_rate": 5.722815813136161e-09,
      "loss": 1.8891,
      "step": 229950
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.4508496522903442,
      "learning_rate": 5.697577485774818e-09,
      "loss": 2.3893,
      "step": 229951
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.4226982593536377,
      "learning_rate": 5.672394933031732e-09,
      "loss": 2.0523,
      "step": 229952
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.3345521688461304,
      "learning_rate": 5.647268154873596e-09,
      "loss": 2.0395,
      "step": 229953
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.7439786195755005,
      "learning_rate": 5.622197151367025e-09,
      "loss": 2.1731,
      "step": 229954
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.4628006219863892,
      "learning_rate": 5.597181922445404e-09,
      "loss": 1.8417,
      "step": 229955
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.3977502584457397,
      "learning_rate": 5.572222468175347e-09,
      "loss": 1.9112,
      "step": 229956
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.613585114479065,
      "learning_rate": 5.547318788523547e-09,
      "loss": 2.0344,
      "step": 229957
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.3426837921142578,
      "learning_rate": 5.5224708834900045e-09,
      "loss": 2.0554,
      "step": 229958
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.4593944549560547,
      "learning_rate": 5.497678753108026e-09,
      "loss": 2.0032,
      "step": 229959
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.5238125324249268,
      "learning_rate": 5.472942397344305e-09,
      "loss": 1.9378,
      "step": 229960
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.5463331937789917,
      "learning_rate": 5.448261816232147e-09,
      "loss": 2.1178,
      "step": 229961
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.4839829206466675,
      "learning_rate": 5.423637009771553e-09,
      "loss": 1.8563,
      "step": 229962
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.6567127704620361,
      "learning_rate": 5.3990679779958304e-09,
      "loss": 1.6702,
      "step": 229963
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.505548357963562,
      "learning_rate": 5.374554720838364e-09,
      "loss": 1.9695,
      "step": 229964
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.3706527948379517,
      "learning_rate": 5.3500972383657695e-09,
      "loss": 2.0072,
      "step": 229965
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.4783180952072144,
      "learning_rate": 5.325695530578045e-09,
      "loss": 1.9866,
      "step": 229966
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.3282663822174072,
      "learning_rate": 5.301349597441884e-09,
      "loss": 1.9574,
      "step": 229967
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.4424140453338623,
      "learning_rate": 5.2770594389572875e-09,
      "loss": 1.9004,
      "step": 229968
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.501151204109192,
      "learning_rate": 5.252825055190868e-09,
      "loss": 1.9397,
      "step": 229969
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.4340991973876953,
      "learning_rate": 5.2286464461093194e-09,
      "loss": 1.9708,
      "step": 229970
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.35291588306427,
      "learning_rate": 5.204523611712641e-09,
      "loss": 1.8468,
      "step": 229971
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.3785958290100098,
      "learning_rate": 5.180456552000833e-09,
      "loss": 1.8841,
      "step": 229972
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.4548161029815674,
      "learning_rate": 5.156445267007203e-09,
      "loss": 1.8699,
      "step": 229973
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.3964306116104126,
      "learning_rate": 5.132489756698444e-09,
      "loss": 1.9097,
      "step": 229974
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.538927435874939,
      "learning_rate": 5.108590021107861e-09,
      "loss": 1.9596,
      "step": 229975
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.496282935142517,
      "learning_rate": 5.084746060235456e-09,
      "loss": 1.8431,
      "step": 229976
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.6102585792541504,
      "learning_rate": 5.060957874081228e-09,
      "loss": 2.0061,
      "step": 229977
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.5465303659439087,
      "learning_rate": 5.037225462645178e-09,
      "loss": 2.1438,
      "step": 229978
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.4854364395141602,
      "learning_rate": 5.013548825927305e-09,
      "loss": 2.1982,
      "step": 229979
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.610068917274475,
      "learning_rate": 4.9899279639276094e-09,
      "loss": 1.869,
      "step": 229980
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.4378104209899902,
      "learning_rate": 4.966362876679397e-09,
      "loss": 1.9885,
      "step": 229981
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.4372013807296753,
      "learning_rate": 4.94285356418267e-09,
      "loss": 2.0263,
      "step": 229982
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.4060574769973755,
      "learning_rate": 4.9194000264374255e-09,
      "loss": 1.786,
      "step": 229983
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.8577885627746582,
      "learning_rate": 4.896002263410359e-09,
      "loss": 1.8061,
      "step": 229984
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.4636476039886475,
      "learning_rate": 4.872660275134776e-09,
      "loss": 1.8855,
      "step": 229985
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.374314546585083,
      "learning_rate": 4.849374061643984e-09,
      "loss": 1.7343,
      "step": 229986
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.6153849363327026,
      "learning_rate": 4.826143622904677e-09,
      "loss": 2.2079,
      "step": 229987
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.4205920696258545,
      "learning_rate": 4.802968958916853e-09,
      "loss": 1.7622,
      "step": 229988
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.6475733518600464,
      "learning_rate": 4.77985006971382e-09,
      "loss": 2.0581,
      "step": 229989
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.4159401655197144,
      "learning_rate": 4.756786955295577e-09,
      "loss": 2.0076,
      "step": 229990
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.5762519836425781,
      "learning_rate": 4.733779615628819e-09,
      "loss": 2.0385,
      "step": 229991
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.519742727279663,
      "learning_rate": 4.710828050746851e-09,
      "loss": 1.8606,
      "step": 229992
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.4403280019760132,
      "learning_rate": 4.687932260682981e-09,
      "loss": 2.0082,
      "step": 229993
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.5309710502624512,
      "learning_rate": 4.665092245370594e-09,
      "loss": 1.9906,
      "step": 229994
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.3670752048492432,
      "learning_rate": 4.642308004876305e-09,
      "loss": 2.0233,
      "step": 229995
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.6053441762924194,
      "learning_rate": 4.619579539166807e-09,
      "loss": 2.1148,
      "step": 229996
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.4548472166061401,
      "learning_rate": 4.596906848275406e-09,
      "loss": 1.8573,
      "step": 229997
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.4533268213272095,
      "learning_rate": 4.574289932202102e-09,
      "loss": 2.1294,
      "step": 229998
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.4363348484039307,
      "learning_rate": 4.551728790946896e-09,
      "loss": 2.0441,
      "step": 229999
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.425492763519287,
      "learning_rate": 4.52922342447648e-09,
      "loss": 2.0874,
      "step": 230000
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.5721906423568726,
      "learning_rate": 4.506773832824162e-09,
      "loss": 2.0689,
      "step": 230001
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.4323784112930298,
      "learning_rate": 4.484380016023248e-09,
      "loss": 2.0931,
      "step": 230002
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.4517861604690552,
      "learning_rate": 4.462041974040431e-09,
      "loss": 2.2348,
      "step": 230003
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.5201717615127563,
      "learning_rate": 4.4397597069090185e-09,
      "loss": 2.0578,
      "step": 230004
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.4116899967193604,
      "learning_rate": 4.417533214595703e-09,
      "loss": 2.2618,
      "step": 230005
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.3498371839523315,
      "learning_rate": 4.395362497133792e-09,
      "loss": 2.0105,
      "step": 230006
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.5138847827911377,
      "learning_rate": 4.373247554489978e-09,
      "loss": 2.1417,
      "step": 230007
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.4801777601242065,
      "learning_rate": 4.351188386730875e-09,
      "loss": 1.8624,
      "step": 230008
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.4905586242675781,
      "learning_rate": 4.329184993823176e-09,
      "loss": 1.9802,
      "step": 230009
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.3976651430130005,
      "learning_rate": 4.3072373757335745e-09,
      "loss": 2.0505,
      "step": 230010
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.5112718343734741,
      "learning_rate": 4.28534553256199e-09,
      "loss": 2.0872,
      "step": 230011
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.4253318309783936,
      "learning_rate": 4.263509464208503e-09,
      "loss": 2.1284,
      "step": 230012
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.4968347549438477,
      "learning_rate": 4.241729170739727e-09,
      "loss": 2.0592,
      "step": 230013
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.398384928703308,
      "learning_rate": 4.220004652155662e-09,
      "loss": 1.9007,
      "step": 230014
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.3402446508407593,
      "learning_rate": 4.198335908456307e-09,
      "loss": 1.7543,
      "step": 230015
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.5937258005142212,
      "learning_rate": 4.176722939641663e-09,
      "loss": 1.9667,
      "step": 230016
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.5475177764892578,
      "learning_rate": 4.1551657456784236e-09,
      "loss": 2.1752,
      "step": 230017
    },
    {
      "epoch": 2.99,
      "grad_norm": 1.3187059164047241,
      "learning_rate": 4.133664326633201e-09,
      "loss": 1.8903,
      "step": 230018
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.5072078704833984,
      "learning_rate": 4.11221868247269e-09,
      "loss": 1.8258,
      "step": 230019
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.6267668008804321,
      "learning_rate": 4.090828813230196e-09,
      "loss": 2.0946,
      "step": 230020
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.4868371486663818,
      "learning_rate": 4.069494718872412e-09,
      "loss": 1.8208,
      "step": 230021
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.4520907402038574,
      "learning_rate": 4.048216399432646e-09,
      "loss": 1.8654,
      "step": 230022
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.6502869129180908,
      "learning_rate": 4.026993854877591e-09,
      "loss": 1.9758,
      "step": 230023
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.4487037658691406,
      "learning_rate": 4.005827085273861e-09,
      "loss": 2.0837,
      "step": 230024
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.677709937095642,
      "learning_rate": 3.98471609055484e-09,
      "loss": 2.071,
      "step": 230025
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.3995819091796875,
      "learning_rate": 3.963660870753838e-09,
      "loss": 1.7483,
      "step": 230026
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.4262120723724365,
      "learning_rate": 3.942661425904159e-09,
      "loss": 1.8414,
      "step": 230027
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.3328135013580322,
      "learning_rate": 3.921717755972498e-09,
      "loss": 1.8754,
      "step": 230028
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.5259590148925781,
      "learning_rate": 3.900829860992161e-09,
      "loss": 2.0691,
      "step": 230029
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.498583197593689,
      "learning_rate": 3.879997740929841e-09,
      "loss": 1.9442,
      "step": 230030
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.4611256122589111,
      "learning_rate": 3.8592213958188455e-09,
      "loss": 2.0312,
      "step": 230031
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.4870046377182007,
      "learning_rate": 3.838500825659174e-09,
      "loss": 2.1568,
      "step": 230032
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.5195574760437012,
      "learning_rate": 3.81783603041752e-09,
      "loss": 2.1018,
      "step": 230033
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.509365439414978,
      "learning_rate": 3.797227010160497e-09,
      "loss": 2.0836,
      "step": 230034
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.35837984085083,
      "learning_rate": 3.776673764854799e-09,
      "loss": 1.7662,
      "step": 230035
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.3664849996566772,
      "learning_rate": 3.756176294500423e-09,
      "loss": 1.9509,
      "step": 230036
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.5235644578933716,
      "learning_rate": 3.73573459913068e-09,
      "loss": 2.0528,
      "step": 230037
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.5240519046783447,
      "learning_rate": 3.71534867871226e-09,
      "loss": 1.8641,
      "step": 230038
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.3968435525894165,
      "learning_rate": 3.695018533245164e-09,
      "loss": 1.8336,
      "step": 230039
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.4365142583847046,
      "learning_rate": 3.674744162796006e-09,
      "loss": 2.1767,
      "step": 230040
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.4608081579208374,
      "learning_rate": 3.6545255672981722e-09,
      "loss": 2.0213,
      "step": 230041
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.4598429203033447,
      "learning_rate": 3.634362746784969e-09,
      "loss": 1.8862,
      "step": 230042
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.4533705711364746,
      "learning_rate": 3.6142557012897034e-09,
      "loss": 2.0531,
      "step": 230043
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.4657609462738037,
      "learning_rate": 3.594204430745762e-09,
      "loss": 1.9995,
      "step": 230044
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.4944250583648682,
      "learning_rate": 3.574208935219758e-09,
      "loss": 2.0862,
      "step": 230045
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.464146614074707,
      "learning_rate": 3.5542692146783846e-09,
      "loss": 1.9495,
      "step": 230046
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.3997037410736084,
      "learning_rate": 3.534385269154949e-09,
      "loss": 2.4971,
      "step": 230047
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.5300415754318237,
      "learning_rate": 3.514557098616144e-09,
      "loss": 1.915,
      "step": 230048
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.4543330669403076,
      "learning_rate": 3.494784703095277e-09,
      "loss": 2.1354,
      "step": 230049
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.6524549722671509,
      "learning_rate": 3.475068082592347e-09,
      "loss": 2.1317,
      "step": 230050
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.4412767887115479,
      "learning_rate": 3.455407237107355e-09,
      "loss": 1.8099,
      "step": 230051
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.7333569526672363,
      "learning_rate": 3.4358021666403e-09,
      "loss": 2.1854,
      "step": 230052
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.4540802240371704,
      "learning_rate": 3.4162528712244898e-09,
      "loss": 2.0532,
      "step": 230053
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.7134523391723633,
      "learning_rate": 3.39675935079331e-09,
      "loss": 1.972,
      "step": 230054
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.3570551872253418,
      "learning_rate": 3.3773216054133745e-09,
      "loss": 1.8374,
      "step": 230055
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.4992091655731201,
      "learning_rate": 3.3579396350846833e-09,
      "loss": 2.1489,
      "step": 230056
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.468074917793274,
      "learning_rate": 3.3386134397739294e-09,
      "loss": 1.77,
      "step": 230057
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.6365162134170532,
      "learning_rate": 3.3193430194811134e-09,
      "loss": 1.8757,
      "step": 230058
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.4333491325378418,
      "learning_rate": 3.300128374272848e-09,
      "loss": 2.035,
      "step": 230059
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.5802035331726074,
      "learning_rate": 3.280969504115827e-09,
      "loss": 2.16,
      "step": 230060
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.4268479347229004,
      "learning_rate": 3.2618664089767432e-09,
      "loss": 2.1209,
      "step": 230061
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.4333299398422241,
      "learning_rate": 3.2428190889222106e-09,
      "loss": 1.9432,
      "step": 230062
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.5075973272323608,
      "learning_rate": 3.2238275439189222e-09,
      "loss": 1.9395,
      "step": 230063
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.5813757181167603,
      "learning_rate": 3.2048917739668777e-09,
      "loss": 2.0175,
      "step": 230064
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.5886982679367065,
      "learning_rate": 3.1860117790993843e-09,
      "loss": 1.9171,
      "step": 230065
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.5814926624298096,
      "learning_rate": 3.167187559316442e-09,
      "loss": 2.0477,
      "step": 230066
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.3199678659439087,
      "learning_rate": 3.148419114584744e-09,
      "loss": 1.991,
      "step": 230067
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.4399291276931763,
      "learning_rate": 3.1297064449375964e-09,
      "loss": 1.9234,
      "step": 230068
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.4236671924591064,
      "learning_rate": 3.111049550375e-09,
      "loss": 1.9019,
      "step": 230069
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.4872016906738281,
      "learning_rate": 3.0924484308969544e-09,
      "loss": 1.8069,
      "step": 230070
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.5573042631149292,
      "learning_rate": 3.07390308650346e-09,
      "loss": 1.9322,
      "step": 230071
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.6158242225646973,
      "learning_rate": 3.055413517194516e-09,
      "loss": 2.0265,
      "step": 230072
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.565122127532959,
      "learning_rate": 3.036979722970123e-09,
      "loss": 1.9656,
      "step": 230073
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.3878651857376099,
      "learning_rate": 3.0186017038635878e-09,
      "loss": 2.0838,
      "step": 230074
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.4127646684646606,
      "learning_rate": 3.0002794598416036e-09,
      "loss": 2.0495,
      "step": 230075
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.4747978448867798,
      "learning_rate": 2.9820129909707833e-09,
      "loss": 1.8806,
      "step": 230076
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.3748465776443481,
      "learning_rate": 2.9638022971512077e-09,
      "loss": 2.0033,
      "step": 230077
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.5246342420578003,
      "learning_rate": 2.945647378482796e-09,
      "loss": 1.8742,
      "step": 230078
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.439497470855713,
      "learning_rate": 2.9275482348989355e-09,
      "loss": 1.9233,
      "step": 230079
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.3023384809494019,
      "learning_rate": 2.909504866466239e-09,
      "loss": 1.8438,
      "step": 230080
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.409536361694336,
      "learning_rate": 2.8915172731180937e-09,
      "loss": 1.8662,
      "step": 230081
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.5774822235107422,
      "learning_rate": 2.8735854549211125e-09,
      "loss": 2.1952,
      "step": 230082
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.6038265228271484,
      "learning_rate": 2.855709411841989e-09,
      "loss": 1.9827,
      "step": 230083
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.470339059829712,
      "learning_rate": 2.8378891439140295e-09,
      "loss": 1.8686,
      "step": 230084
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.4943922758102417,
      "learning_rate": 2.8201246511039278e-09,
      "loss": 1.9076,
      "step": 230085
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.4545502662658691,
      "learning_rate": 2.8024159334449903e-09,
      "loss": 1.9582,
      "step": 230086
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.4923338890075684,
      "learning_rate": 2.7847629909039104e-09,
      "loss": 1.862,
      "step": 230087
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.5444161891937256,
      "learning_rate": 2.7671658235473017e-09,
      "loss": 1.7319,
      "step": 230088
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.4849218130111694,
      "learning_rate": 2.7496244313085505e-09,
      "loss": 1.9929,
      "step": 230089
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.4827346801757812,
      "learning_rate": 2.732138814220963e-09,
      "loss": 1.7987,
      "step": 230090
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.350292682647705,
      "learning_rate": 2.7147089722845407e-09,
      "loss": 1.8443,
      "step": 230091
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.3214614391326904,
      "learning_rate": 2.697334905532589e-09,
      "loss": 2.2159,
      "step": 230092
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.475776195526123,
      "learning_rate": 2.6800166139318012e-09,
      "loss": 1.7656,
      "step": 230093
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.4596160650253296,
      "learning_rate": 2.6627540974821782e-09,
      "loss": 1.7925,
      "step": 230094
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.29436194896698,
      "learning_rate": 2.645547356217026e-09,
      "loss": 2.0783,
      "step": 230095
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.4498573541641235,
      "learning_rate": 2.628396390103038e-09,
      "loss": 1.9213,
      "step": 230096
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.481981873512268,
      "learning_rate": 2.611301199173521e-09,
      "loss": 2.0189,
      "step": 230097
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.3817901611328125,
      "learning_rate": 2.5942617834284752e-09,
      "loss": 2.1848,
      "step": 230098
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.3119664192199707,
      "learning_rate": 2.5772781428679e-09,
      "loss": 1.8651,
      "step": 230099
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.5172065496444702,
      "learning_rate": 2.5603502774917962e-09,
      "loss": 1.9149,
      "step": 230100
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.414476752281189,
      "learning_rate": 2.5434781872668563e-09,
      "loss": 1.8352,
      "step": 230101
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.4346567392349243,
      "learning_rate": 2.5266618722596944e-09,
      "loss": 1.8929,
      "step": 230102
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.4947088956832886,
      "learning_rate": 2.50990133247031e-09,
      "loss": 1.9236,
      "step": 230103
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.5372484922409058,
      "learning_rate": 2.4931965678320897e-09,
      "loss": 1.9152,
      "step": 230104
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.5783765316009521,
      "learning_rate": 2.4765475784116474e-09,
      "loss": 2.0029,
      "step": 230105
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.578202724456787,
      "learning_rate": 2.4599543642089824e-09,
      "loss": 1.999,
      "step": 230106
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.4792463779449463,
      "learning_rate": 2.443416925190789e-09,
      "loss": 2.1712,
      "step": 230107
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.4727262258529663,
      "learning_rate": 2.4269352613903726e-09,
      "loss": 2.2389,
      "step": 230108
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.4701820611953735,
      "learning_rate": 2.410509372807734e-09,
      "loss": 1.8085,
      "step": 230109
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.6438974142074585,
      "learning_rate": 2.3941392594095665e-09,
      "loss": 2.0856,
      "step": 230110
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.4459987878799438,
      "learning_rate": 2.3778249212624835e-09,
      "loss": 1.9235,
      "step": 230111
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.5275388956069946,
      "learning_rate": 2.3615663583331778e-09,
      "loss": 1.9794,
      "step": 230112
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.478708028793335,
      "learning_rate": 2.34536357062165e-09,
      "loss": 1.9589,
      "step": 230113
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.4914734363555908,
      "learning_rate": 2.3292165581279e-09,
      "loss": 2.1441,
      "step": 230114
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.343829870223999,
      "learning_rate": 2.3131253208519275e-09,
      "loss": 1.8732,
      "step": 230115
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.3868831396102905,
      "learning_rate": 2.297089858827039e-09,
      "loss": 2.0716,
      "step": 230116
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.4550806283950806,
      "learning_rate": 2.2811101720532355e-09,
      "loss": 1.8411,
      "step": 230117
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.5279698371887207,
      "learning_rate": 2.265186260497209e-09,
      "loss": 1.9968,
      "step": 230118
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.4491057395935059,
      "learning_rate": 2.2493181241922677e-09,
      "loss": 1.876,
      "step": 230119
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.400649905204773,
      "learning_rate": 2.2335057631384103e-09,
      "loss": 2.0551,
      "step": 230120
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.4267724752426147,
      "learning_rate": 2.2177491773023305e-09,
      "loss": 1.9304,
      "step": 230121
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.4842491149902344,
      "learning_rate": 2.2020483667506417e-09,
      "loss": 2.1176,
      "step": 230122
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.7291165590286255,
      "learning_rate": 2.1864033314500372e-09,
      "loss": 2.3512,
      "step": 230123
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.4813876152038574,
      "learning_rate": 2.1708140714005173e-09,
      "loss": 1.6819,
      "step": 230124
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.3222724199295044,
      "learning_rate": 2.155280586602082e-09,
      "loss": 2.0065,
      "step": 230125
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.5641148090362549,
      "learning_rate": 2.1398028770547304e-09,
      "loss": 1.9812,
      "step": 230126
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.4961180686950684,
      "learning_rate": 2.1243809427917702e-09,
      "loss": 1.9438,
      "step": 230127
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.4879621267318726,
      "learning_rate": 2.109014783779894e-09,
      "loss": 1.9698,
      "step": 230128
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.4171704053878784,
      "learning_rate": 2.0937044000524094e-09,
      "loss": 1.8061,
      "step": 230129
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.387516975402832,
      "learning_rate": 2.0784497916093155e-09,
      "loss": 2.1765,
      "step": 230130
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.3505321741104126,
      "learning_rate": 2.063250958417306e-09,
      "loss": 2.0549,
      "step": 230131
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.386719822883606,
      "learning_rate": 2.048107900509688e-09,
      "loss": 1.8747,
      "step": 230132
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.3545019626617432,
      "learning_rate": 2.0330206179197673e-09,
      "loss": 1.8637,
      "step": 230133
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.374559998512268,
      "learning_rate": 2.017989110580931e-09,
      "loss": 1.97,
      "step": 230134
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.4236549139022827,
      "learning_rate": 2.0030133785264855e-09,
      "loss": 1.8372,
      "step": 230135
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.5749590396881104,
      "learning_rate": 1.9880934217897383e-09,
      "loss": 2.1277,
      "step": 230136
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.6513309478759766,
      "learning_rate": 1.973229240304075e-09,
      "loss": 2.1366,
      "step": 230137
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.4606364965438843,
      "learning_rate": 1.9584208341361096e-09,
      "loss": 2.1527,
      "step": 230138
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.418311595916748,
      "learning_rate": 1.943668203285842e-09,
      "loss": 2.1331,
      "step": 230139
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.418869137763977,
      "learning_rate": 1.9289713477199654e-09,
      "loss": 2.247,
      "step": 230140
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.6787266731262207,
      "learning_rate": 1.9143302674717863e-09,
      "loss": 2.1418,
      "step": 230141
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.3947521448135376,
      "learning_rate": 1.8997449625079985e-09,
      "loss": 1.8387,
      "step": 230142
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.6401318311691284,
      "learning_rate": 1.8852154328619084e-09,
      "loss": 2.1318,
      "step": 230143
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.4841082096099854,
      "learning_rate": 1.870741678533516e-09,
      "loss": 1.9976,
      "step": 230144
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.4801126718521118,
      "learning_rate": 1.8563236995228214e-09,
      "loss": 1.971,
      "step": 230145
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.5666098594665527,
      "learning_rate": 1.8419614958298245e-09,
      "loss": 1.9963,
      "step": 230146
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.599204421043396,
      "learning_rate": 1.8276550674545254e-09,
      "loss": 1.9839,
      "step": 230147
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.3914389610290527,
      "learning_rate": 1.813404414396924e-09,
      "loss": 1.8537,
      "step": 230148
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.4413588047027588,
      "learning_rate": 1.7992095366570203e-09,
      "loss": 2.0104,
      "step": 230149
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.4899277687072754,
      "learning_rate": 1.785070434268121e-09,
      "loss": 2.1124,
      "step": 230150
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.3853199481964111,
      "learning_rate": 1.7709871071969195e-09,
      "loss": 2.0745,
      "step": 230151
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.5225659608840942,
      "learning_rate": 1.7569595554434156e-09,
      "loss": 1.851,
      "step": 230152
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.5241420269012451,
      "learning_rate": 1.7429877790409164e-09,
      "loss": 1.9296,
      "step": 230153
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.4659804105758667,
      "learning_rate": 1.7290717779894215e-09,
      "loss": 1.8124,
      "step": 230154
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.4704192876815796,
      "learning_rate": 1.715211552288931e-09,
      "loss": 1.7105,
      "step": 230155
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.4010158777236938,
      "learning_rate": 1.7014071019061382e-09,
      "loss": 1.8929,
      "step": 230156
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.4397690296173096,
      "learning_rate": 1.6876584268743498e-09,
      "loss": 1.824,
      "step": 230157
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.4583330154418945,
      "learning_rate": 1.673965527193566e-09,
      "loss": 2.0763,
      "step": 230158
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.5221538543701172,
      "learning_rate": 1.6603284028637864e-09,
      "loss": 2.0303,
      "step": 230159
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.4033920764923096,
      "learning_rate": 1.6467470538850113e-09,
      "loss": 2.0074,
      "step": 230160
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.6331067085266113,
      "learning_rate": 1.6332214802905474e-09,
      "loss": 1.9169,
      "step": 230161
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.6647337675094604,
      "learning_rate": 1.6197516820137813e-09,
      "loss": 2.1016,
      "step": 230162
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.592160940170288,
      "learning_rate": 1.6063376591213261e-09,
      "loss": 1.8368,
      "step": 230163
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.4273111820220947,
      "learning_rate": 1.5929794115798755e-09,
      "loss": 1.8128,
      "step": 230164
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.8794829845428467,
      "learning_rate": 1.579676939422736e-09,
      "loss": 2.036,
      "step": 230165
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.4671207666397095,
      "learning_rate": 1.5664302426499076e-09,
      "loss": 2.0218,
      "step": 230166
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.6290327310562134,
      "learning_rate": 1.553239321194777e-09,
      "loss": 1.9876,
      "step": 230167
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.465922236442566,
      "learning_rate": 1.540104175157264e-09,
      "loss": 2.0049,
      "step": 230168
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.4658806324005127,
      "learning_rate": 1.5270248045040624e-09,
      "loss": 1.9975,
      "step": 230169
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.458998680114746,
      "learning_rate": 1.514001209201865e-09,
      "loss": 2.0175,
      "step": 230170
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.45783531665802,
      "learning_rate": 1.5010333892839788e-09,
      "loss": 2.0033,
      "step": 230171
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.4403626918792725,
      "learning_rate": 1.4881213447504038e-09,
      "loss": 2.0741,
      "step": 230172
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.4150584936141968,
      "learning_rate": 1.4752650756011398e-09,
      "loss": 1.8196,
      "step": 230173
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.3832286596298218,
      "learning_rate": 1.4624645818694936e-09,
      "loss": 1.986,
      "step": 230174
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.6770068407058716,
      "learning_rate": 1.4497198634888518e-09,
      "loss": 1.9735,
      "step": 230175
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.4370872974395752,
      "learning_rate": 1.4370309205258279e-09,
      "loss": 1.9603,
      "step": 230176
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.45756196975708,
      "learning_rate": 1.4243977529471151e-09,
      "loss": 1.9178,
      "step": 230177
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.4673696756362915,
      "learning_rate": 1.41182036078602e-09,
      "loss": 2.1896,
      "step": 230178
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.5148123502731323,
      "learning_rate": 1.3992987440092362e-09,
      "loss": 1.9584,
      "step": 230179
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.4500885009765625,
      "learning_rate": 1.3868329026500701e-09,
      "loss": 2.067,
      "step": 230180
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.359655499458313,
      "learning_rate": 1.3744228366752153e-09,
      "loss": 2.0293,
      "step": 230181
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.5255987644195557,
      "learning_rate": 1.362068546117978e-09,
      "loss": 2.042,
      "step": 230182
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.3896647691726685,
      "learning_rate": 1.3497700309783587e-09,
      "loss": 1.797,
      "step": 230183
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.492758870124817,
      "learning_rate": 1.3375272912230506e-09,
      "loss": 2.2449,
      "step": 230184
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.4138426780700684,
      "learning_rate": 1.32534032688536e-09,
      "loss": 1.921,
      "step": 230185
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.4560399055480957,
      "learning_rate": 1.3132091379985943e-09,
      "loss": 2.0961,
      "step": 230186
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.7563962936401367,
      "learning_rate": 1.3011337244961395e-09,
      "loss": 2.0342,
      "step": 230187
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.6196794509887695,
      "learning_rate": 1.2891140864446093e-09,
      "loss": 2.0638,
      "step": 230188
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.5509268045425415,
      "learning_rate": 1.2771502238106968e-09,
      "loss": 2.0624,
      "step": 230189
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.4416922330856323,
      "learning_rate": 1.2652421365944022e-09,
      "loss": 1.9542,
      "step": 230190
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.3693528175354004,
      "learning_rate": 1.2533898247957252e-09,
      "loss": 1.8496,
      "step": 230191
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.4492194652557373,
      "learning_rate": 1.2415932884146662e-09,
      "loss": 1.9857,
      "step": 230192
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.486204743385315,
      "learning_rate": 1.2298525274845317e-09,
      "loss": 2.2377,
      "step": 230193
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.452928900718689,
      "learning_rate": 1.2181675420053217e-09,
      "loss": 1.8211,
      "step": 230194
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.5431830883026123,
      "learning_rate": 1.2065383319437293e-09,
      "loss": 1.8132,
      "step": 230195
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.5115352869033813,
      "learning_rate": 1.1949648973330616e-09,
      "loss": 1.9888,
      "step": 230196
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.5180290937423706,
      "learning_rate": 1.1834472381400118e-09,
      "loss": 2.071,
      "step": 230197
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.4229938983917236,
      "learning_rate": 1.171985354431193e-09,
      "loss": 1.9463,
      "step": 230198
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.49434232711792,
      "learning_rate": 1.1605792461399922e-09,
      "loss": 1.9613,
      "step": 230199
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.4044811725616455,
      "learning_rate": 1.1492289132997158e-09,
      "loss": 1.9458,
      "step": 230200
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.4307125806808472,
      "learning_rate": 1.137934355877057e-09,
      "loss": 1.9191,
      "step": 230201
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.5469565391540527,
      "learning_rate": 1.1266955739386295e-09,
      "loss": 1.8305,
      "step": 230202
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.7515450716018677,
      "learning_rate": 1.1155125674511268e-09,
      "loss": 1.9232,
      "step": 230203
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.5184763669967651,
      "learning_rate": 1.1043853364145483e-09,
      "loss": 1.9953,
      "step": 230204
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.4715031385421753,
      "learning_rate": 1.093313880862201e-09,
      "loss": 2.1924,
      "step": 230205
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.523078441619873,
      "learning_rate": 1.0822982007274716e-09,
      "loss": 1.9448,
      "step": 230206
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.5083599090576172,
      "learning_rate": 1.0713382960769733e-09,
      "loss": 2.0604,
      "step": 230207
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.4097950458526611,
      "learning_rate": 1.0604341668773996e-09,
      "loss": 1.9994,
      "step": 230208
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.498669147491455,
      "learning_rate": 1.0495858131620572e-09,
      "loss": 1.9028,
      "step": 230209
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.6205356121063232,
      "learning_rate": 1.038793234897639e-09,
      "loss": 1.9977,
      "step": 230210
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.5132101774215698,
      "learning_rate": 1.0280564321174522e-09,
      "loss": 2.0509,
      "step": 230211
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.4748259782791138,
      "learning_rate": 1.0173754048214967e-09,
      "loss": 2.0462,
      "step": 230212
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.6359648704528809,
      "learning_rate": 1.0067501529764654e-09,
      "loss": 1.7837,
      "step": 230213
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.663998007774353,
      "learning_rate": 9.961806765823588e-10,
      "loss": 1.8546,
      "step": 230214
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.4158785343170166,
      "learning_rate": 9.8566697570579e-10,
      "loss": 2.2944,
      "step": 230215
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.3678889274597168,
      "learning_rate": 9.752090503134523e-10,
      "loss": 1.8955,
      "step": 230216
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.4680525064468384,
      "learning_rate": 9.648069003720393e-10,
      "loss": 1.9504,
      "step": 230217
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.4746705293655396,
      "learning_rate": 9.544605259148574e-10,
      "loss": 1.9076,
      "step": 230218
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.4481472969055176,
      "learning_rate": 9.441699269752135e-10,
      "loss": 1.9177,
      "step": 230219
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.5005443096160889,
      "learning_rate": 9.339351035198006e-10,
      "loss": 2.0857,
      "step": 230220
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.4646565914154053,
      "learning_rate": 9.237560555153122e-10,
      "loss": 1.848,
      "step": 230221
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.3375872373580933,
      "learning_rate": 9.136327830283618e-10,
      "loss": 2.038,
      "step": 230222
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.4771943092346191,
      "learning_rate": 9.035652860256426e-10,
      "loss": 1.9808,
      "step": 230223
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.3978912830352783,
      "learning_rate": 8.935535645404612e-10,
      "loss": 1.679,
      "step": 230224
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.623518466949463,
      "learning_rate": 8.835976185395111e-10,
      "loss": 2.1153,
      "step": 230225
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.4598500728607178,
      "learning_rate": 8.73697448022792e-10,
      "loss": 2.004,
      "step": 230226
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.4482643604278564,
      "learning_rate": 8.63853053023611e-10,
      "loss": 1.8528,
      "step": 230227
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.567581295967102,
      "learning_rate": 8.54064433508661e-10,
      "loss": 2.0242,
      "step": 230228
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.4949190616607666,
      "learning_rate": 8.44331589511249e-10,
      "loss": 2.0949,
      "step": 230229
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.4863667488098145,
      "learning_rate": 8.346545210313749e-10,
      "loss": 1.9249,
      "step": 230230
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.4042341709136963,
      "learning_rate": 8.25033228035732e-10,
      "loss": 1.983,
      "step": 230231
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.5680179595947266,
      "learning_rate": 8.154677105576269e-10,
      "loss": 1.8733,
      "step": 230232
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.470832347869873,
      "learning_rate": 8.059579685970596e-10,
      "loss": 2.0096,
      "step": 230233
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.5566222667694092,
      "learning_rate": 7.965040021540303e-10,
      "loss": 1.9121,
      "step": 230234
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.3829675912857056,
      "learning_rate": 7.871058111952322e-10,
      "loss": 1.9193,
      "step": 230235
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.33917236328125,
      "learning_rate": 7.777633957872786e-10,
      "loss": 2.0428,
      "step": 230236
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.4687572717666626,
      "learning_rate": 7.68476755896863e-10,
      "loss": 2.0525,
      "step": 230237
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.4952073097229004,
      "learning_rate": 7.592458914906784e-10,
      "loss": 2.0429,
      "step": 230238
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.4249441623687744,
      "learning_rate": 7.500708026353386e-10,
      "loss": 2.1539,
      "step": 230239
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.4190034866333008,
      "learning_rate": 7.409514892975365e-10,
      "loss": 1.8272,
      "step": 230240
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.5391151905059814,
      "learning_rate": 7.318879514772724e-10,
      "loss": 1.7647,
      "step": 230241
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.4567581415176392,
      "learning_rate": 7.228801892078529e-10,
      "loss": 2.0983,
      "step": 230242
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.5561708211898804,
      "learning_rate": 7.139282024559712e-10,
      "loss": 2.1115,
      "step": 230243
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.5048027038574219,
      "learning_rate": 7.050319912216273e-10,
      "loss": 2.0341,
      "step": 230244
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.5014625787734985,
      "learning_rate": 6.961915555381281e-10,
      "loss": 2.3588,
      "step": 230245
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.598516821861267,
      "learning_rate": 6.874068953721668e-10,
      "loss": 1.9661,
      "step": 230246
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.4350917339324951,
      "learning_rate": 6.7867801075705e-10,
      "loss": 2.0599,
      "step": 230247
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.4105851650238037,
      "learning_rate": 6.700049016594711e-10,
      "loss": 1.9069,
      "step": 230248
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.472715139389038,
      "learning_rate": 6.613875681127367e-10,
      "loss": 1.7986,
      "step": 230249
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.4956594705581665,
      "learning_rate": 6.528260100835403e-10,
      "loss": 2.0814,
      "step": 230250
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.5149917602539062,
      "learning_rate": 6.443202276384951e-10,
      "loss": 1.9334,
      "step": 230251
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.425687551498413,
      "learning_rate": 6.358702207109878e-10,
      "loss": 2.0237,
      "step": 230252
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.4819384813308716,
      "learning_rate": 6.274759893343251e-10,
      "loss": 1.9867,
      "step": 230253
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.5866297483444214,
      "learning_rate": 6.191375334752003e-10,
      "loss": 2.0386,
      "step": 230254
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.4910569190979004,
      "learning_rate": 6.108548532002267e-10,
      "loss": 2.0287,
      "step": 230255
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.5327122211456299,
      "learning_rate": 6.026279484760977e-10,
      "loss": 1.9075,
      "step": 230256
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.6528695821762085,
      "learning_rate": 5.944568192695065e-10,
      "loss": 2.1247,
      "step": 230257
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.5411347150802612,
      "learning_rate": 5.863414656470667e-10,
      "loss": 1.8174,
      "step": 230258
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.4371269941329956,
      "learning_rate": 5.782818875754713e-10,
      "loss": 1.908,
      "step": 230259
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.4054036140441895,
      "learning_rate": 5.702780850547207e-10,
      "loss": 1.9539,
      "step": 230260
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.3179792165756226,
      "learning_rate": 5.623300580848145e-10,
      "loss": 2.1283,
      "step": 230261
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.3334119319915771,
      "learning_rate": 5.544378066657529e-10,
      "loss": 2.0897,
      "step": 230262
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.5128084421157837,
      "learning_rate": 5.466013308308425e-10,
      "loss": 2.0586,
      "step": 230263
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.5112882852554321,
      "learning_rate": 5.388206305134701e-10,
      "loss": 2.0017,
      "step": 230264
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.4803529977798462,
      "learning_rate": 5.310957058135557e-10,
      "loss": 1.9282,
      "step": 230265
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.514849305152893,
      "learning_rate": 5.234265566311791e-10,
      "loss": 1.8189,
      "step": 230266
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.4457815885543823,
      "learning_rate": 5.158131830329537e-10,
      "loss": 2.0613,
      "step": 230267
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.4610052108764648,
      "learning_rate": 5.082555850188796e-10,
      "loss": 1.8258,
      "step": 230268
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.5679110288619995,
      "learning_rate": 5.007537625556501e-10,
      "loss": 2.0582,
      "step": 230269
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.3660839796066284,
      "learning_rate": 4.933077156765719e-10,
      "loss": 1.9036,
      "step": 230270
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.3674417734146118,
      "learning_rate": 4.859174443483382e-10,
      "loss": 1.9121,
      "step": 230271
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.469967007637024,
      "learning_rate": 4.785829486042558e-10,
      "loss": 1.8094,
      "step": 230272
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.462638258934021,
      "learning_rate": 4.713042284443247e-10,
      "loss": 2.1265,
      "step": 230273
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.3682752847671509,
      "learning_rate": 4.6408128383523814e-10,
      "loss": 2.0159,
      "step": 230274
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.4240540266036987,
      "learning_rate": 4.5691411481030283e-10,
      "loss": 1.9265,
      "step": 230275
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.5115562677383423,
      "learning_rate": 4.498027213695188e-10,
      "loss": 1.691,
      "step": 230276
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.558397650718689,
      "learning_rate": 4.42747103512886e-10,
      "loss": 2.2476,
      "step": 230277
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.5510870218276978,
      "learning_rate": 4.357472612070978e-10,
      "loss": 1.7903,
      "step": 230278
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.5790194272994995,
      "learning_rate": 4.2880319451876754e-10,
      "loss": 1.9465,
      "step": 230279
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.4791353940963745,
      "learning_rate": 4.2191490341458855e-10,
      "loss": 1.8622,
      "step": 230280
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.4787002801895142,
      "learning_rate": 4.1508238786125414e-10,
      "loss": 1.874,
      "step": 230281
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.454330325126648,
      "learning_rate": 4.083056479253777e-10,
      "loss": 2.0982,
      "step": 230282
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.3558768033981323,
      "learning_rate": 4.0158468357365246e-10,
      "loss": 1.798,
      "step": 230283
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.5362545251846313,
      "learning_rate": 3.9491949480607856e-10,
      "loss": 1.8099,
      "step": 230284
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.5948443412780762,
      "learning_rate": 3.883100816226559e-10,
      "loss": 2.0624,
      "step": 230285
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.5402532815933228,
      "learning_rate": 3.8175644402338447e-10,
      "loss": 2.1199,
      "step": 230286
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.4288181066513062,
      "learning_rate": 3.752585820082643e-10,
      "loss": 2.0444,
      "step": 230287
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.4982290267944336,
      "learning_rate": 3.688164956106021e-10,
      "loss": 2.035,
      "step": 230288
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.698404312133789,
      "learning_rate": 3.624301847970912e-10,
      "loss": 1.9224,
      "step": 230289
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.4037895202636719,
      "learning_rate": 3.560996495677315e-10,
      "loss": 2.0934,
      "step": 230290
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.4497853517532349,
      "learning_rate": 3.498248899558298e-10,
      "loss": 2.0155,
      "step": 230291
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.4611551761627197,
      "learning_rate": 3.4360590592807935e-10,
      "loss": 1.865,
      "step": 230292
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.493336796760559,
      "learning_rate": 3.3744269751778686e-10,
      "loss": 2.1597,
      "step": 230293
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.3812810182571411,
      "learning_rate": 3.3133526469164564e-10,
      "loss": 1.9209,
      "step": 230294
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.5674352645874023,
      "learning_rate": 3.2528360748296233e-10,
      "loss": 2.129,
      "step": 230295
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.48927640914917,
      "learning_rate": 3.192877258584303e-10,
      "loss": 1.9728,
      "step": 230296
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.5016511678695679,
      "learning_rate": 3.1334761985135627e-10,
      "loss": 1.906,
      "step": 230297
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.555346131324768,
      "learning_rate": 3.0746328942843346e-10,
      "loss": 1.8602,
      "step": 230298
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.4116185903549194,
      "learning_rate": 3.016347346229686e-10,
      "loss": 1.7704,
      "step": 230299
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.5183733701705933,
      "learning_rate": 2.9586195543496173e-10,
      "loss": 2.0662,
      "step": 230300
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.4591139554977417,
      "learning_rate": 2.9014495183110606e-10,
      "loss": 1.9479,
      "step": 230301
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.4948192834854126,
      "learning_rate": 2.844837238447084e-10,
      "loss": 1.9324,
      "step": 230302
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.4216781854629517,
      "learning_rate": 2.788782714757687e-10,
      "loss": 2.0702,
      "step": 230303
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.4694464206695557,
      "learning_rate": 2.733285947242869e-10,
      "loss": 2.002,
      "step": 230304
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.4563467502593994,
      "learning_rate": 2.678346935902631e-10,
      "loss": 1.904,
      "step": 230305
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.5046433210372925,
      "learning_rate": 2.6239656804039053e-10,
      "loss": 2.0365,
      "step": 230306
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.4103153944015503,
      "learning_rate": 2.5701421810797594e-10,
      "loss": 1.9328,
      "step": 230307
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.4036802053451538,
      "learning_rate": 2.51687643826326e-10,
      "loss": 1.794,
      "step": 230308
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.4135292768478394,
      "learning_rate": 2.464168451288273e-10,
      "loss": 2.0315,
      "step": 230309
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.4604992866516113,
      "learning_rate": 2.412018220487866e-10,
      "loss": 1.9731,
      "step": 230310
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.5719592571258545,
      "learning_rate": 2.360425746195105e-10,
      "loss": 1.996,
      "step": 230311
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.5240217447280884,
      "learning_rate": 2.3093910277438565e-10,
      "loss": 1.7704,
      "step": 230312
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.3938257694244385,
      "learning_rate": 2.2589140654671878e-10,
      "loss": 1.7572,
      "step": 230313
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.6423065662384033,
      "learning_rate": 2.2089948596981655e-10,
      "loss": 1.9024,
      "step": 230314
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.3799222707748413,
      "learning_rate": 2.1596334097706557e-10,
      "loss": 2.1509,
      "step": 230315
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.3777046203613281,
      "learning_rate": 2.1108297163507926e-10,
      "loss": 2.0806,
      "step": 230316
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.3719562292099,
      "learning_rate": 2.0625837791055089e-10,
      "loss": 1.8385,
      "step": 230317
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.509786605834961,
      "learning_rate": 2.0148955980348047e-10,
      "loss": 2.0063,
      "step": 230318
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.5029925107955933,
      "learning_rate": 1.967765173471747e-10,
      "loss": 1.8975,
      "step": 230319
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.3931993246078491,
      "learning_rate": 1.921192504750202e-10,
      "loss": 2.0516,
      "step": 230320
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.4577953815460205,
      "learning_rate": 1.8751775925363032e-10,
      "loss": 1.9753,
      "step": 230321
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.5460233688354492,
      "learning_rate": 1.829720436830051e-10,
      "loss": 2.1703,
      "step": 230322
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.462038516998291,
      "learning_rate": 1.7848210369653115e-10,
      "loss": 1.863,
      "step": 230323
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.3959101438522339,
      "learning_rate": 1.7404793936082183e-10,
      "loss": 1.9336,
      "step": 230324
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.5926058292388916,
      "learning_rate": 1.6966955067587719e-10,
      "loss": 1.9996,
      "step": 230325
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.3922964334487915,
      "learning_rate": 1.6534693760839047e-10,
      "loss": 1.8486,
      "step": 230326
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.4093056917190552,
      "learning_rate": 1.610801001583617e-10,
      "loss": 1.9308,
      "step": 230327
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.562242865562439,
      "learning_rate": 1.5686903835909758e-10,
      "loss": 2.1714,
      "step": 230328
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.4906659126281738,
      "learning_rate": 1.5271375217729143e-10,
      "loss": 1.6103,
      "step": 230329
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.4334481954574585,
      "learning_rate": 1.486142416462499e-10,
      "loss": 1.8163,
      "step": 230330
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.7797355651855469,
      "learning_rate": 1.4457050673266635e-10,
      "loss": 2.0258,
      "step": 230331
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.494010090827942,
      "learning_rate": 1.4058254746984743e-10,
      "loss": 1.9395,
      "step": 230332
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.4956871271133423,
      "learning_rate": 1.3665036382448646e-10,
      "loss": 1.8753,
      "step": 230333
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.4594306945800781,
      "learning_rate": 1.3277395582989013e-10,
      "loss": 2.0934,
      "step": 230334
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.5672498941421509,
      "learning_rate": 1.2895332348605848e-10,
      "loss": 1.9598,
      "step": 230335
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.6674067974090576,
      "learning_rate": 1.2518846675968476e-10,
      "loss": 2.1267,
      "step": 230336
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.4613271951675415,
      "learning_rate": 1.2147938568407568e-10,
      "loss": 1.7543,
      "step": 230337
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.4266570806503296,
      "learning_rate": 1.1782608025923123e-10,
      "loss": 2.1621,
      "step": 230338
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.4033746719360352,
      "learning_rate": 1.1422855045184475e-10,
      "loss": 1.9357,
      "step": 230339
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.5169241428375244,
      "learning_rate": 1.1068679629522293e-10,
      "loss": 2.0123,
      "step": 230340
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.470644474029541,
      "learning_rate": 1.0720081778936573e-10,
      "loss": 1.9697,
      "step": 230341
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.5254924297332764,
      "learning_rate": 1.0377061490096649e-10,
      "loss": 2.0332,
      "step": 230342
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.4142298698425293,
      "learning_rate": 1.003961876966386e-10,
      "loss": 1.9146,
      "step": 230343
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.365295648574829,
      "learning_rate": 9.707753610976865e-11,
      "loss": 2.1942,
      "step": 230344
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.4613990783691406,
      "learning_rate": 9.381466017366335e-11,
      "loss": 2.0212,
      "step": 230345
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.4490805864334106,
      "learning_rate": 9.060755988832269e-11,
      "loss": 1.963,
      "step": 230346
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.4553461074829102,
      "learning_rate": 8.745623525374668e-11,
      "loss": 1.9529,
      "step": 230347
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.4433764219284058,
      "learning_rate": 8.436068623662861e-11,
      "loss": 1.9811,
      "step": 230348
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.6326318979263306,
      "learning_rate": 8.132091290358189e-11,
      "loss": 2.0784,
      "step": 230349
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.5634655952453613,
      "learning_rate": 7.833691518799313e-11,
      "loss": 2.1138,
      "step": 230350
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.4147882461547852,
      "learning_rate": 7.540869312316899e-11,
      "loss": 2.0349,
      "step": 230351
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.4405003786087036,
      "learning_rate": 7.25362467424162e-11,
      "loss": 1.9685,
      "step": 230352
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.473637342453003,
      "learning_rate": 6.971957597912137e-11,
      "loss": 1.8473,
      "step": 230353
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.595641851425171,
      "learning_rate": 6.695868086659118e-11,
      "loss": 2.0572,
      "step": 230354
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.4867507219314575,
      "learning_rate": 6.425356140482562e-11,
      "loss": 2.1817,
      "step": 230355
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.5517098903656006,
      "learning_rate": 6.160421759382473e-11,
      "loss": 2.0692,
      "step": 230356
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.4750312566757202,
      "learning_rate": 5.901064946689516e-11,
      "loss": 2.0076,
      "step": 230357
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.3875325918197632,
      "learning_rate": 5.647285695742354e-11,
      "loss": 1.8734,
      "step": 230358
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.435524344444275,
      "learning_rate": 5.399084009871657e-11,
      "loss": 1.9904,
      "step": 230359
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.398927092552185,
      "learning_rate": 5.1564598924080937e-11,
      "loss": 1.9353,
      "step": 230360
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.4874485731124878,
      "learning_rate": 4.919413336690325e-11,
      "loss": 1.7443,
      "step": 230361
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.404636263847351,
      "learning_rate": 4.687944349379691e-11,
      "loss": 1.8107,
      "step": 230362
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.470137119293213,
      "learning_rate": 4.4620529238148515e-11,
      "loss": 2.0058,
      "step": 230363
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.5543161630630493,
      "learning_rate": 4.2417390666571463e-11,
      "loss": 2.0521,
      "step": 230364
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.6439775228500366,
      "learning_rate": 4.027002774575905e-11,
      "loss": 1.9381,
      "step": 230365
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.5961002111434937,
      "learning_rate": 3.817844047571128e-11,
      "loss": 1.8805,
      "step": 230366
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.4022215604782104,
      "learning_rate": 3.614262885642816e-11,
      "loss": 1.984,
      "step": 230367
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.5085599422454834,
      "learning_rate": 3.416259288790968e-11,
      "loss": 1.9906,
      "step": 230368
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.4603550434112549,
      "learning_rate": 3.2238332603462536e-11,
      "loss": 1.9675,
      "step": 230369
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.4777053594589233,
      "learning_rate": 3.036984796978004e-11,
      "loss": 2.0241,
      "step": 230370
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.4162193536758423,
      "learning_rate": 2.8557138986862183e-11,
      "loss": 1.9343,
      "step": 230371
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.584547519683838,
      "learning_rate": 2.6800205654708972e-11,
      "loss": 1.8685,
      "step": 230372
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.5211070775985718,
      "learning_rate": 2.5099047973320406e-11,
      "loss": 1.9522,
      "step": 230373
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.6117055416107178,
      "learning_rate": 2.3453665976003177e-11,
      "loss": 1.9302,
      "step": 230374
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.4494450092315674,
      "learning_rate": 2.18640595961439e-11,
      "loss": 1.9251,
      "step": 230375
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.4630184173583984,
      "learning_rate": 2.033022890035596e-11,
      "loss": 2.2259,
      "step": 230376
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.4353363513946533,
      "learning_rate": 1.8852173888639356e-11,
      "loss": 2.0086,
      "step": 230377
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.6369162797927856,
      "learning_rate": 1.7429894494380702e-11,
      "loss": 2.2608,
      "step": 230378
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.5123409032821655,
      "learning_rate": 1.6063390784193386e-11,
      "loss": 2.0754,
      "step": 230379
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.519881248474121,
      "learning_rate": 1.4752662724770715e-11,
      "loss": 1.8723,
      "step": 230380
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.4276105165481567,
      "learning_rate": 1.3497710316112687e-11,
      "loss": 1.8968,
      "step": 230381
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.510252833366394,
      "learning_rate": 1.2298533591525994e-11,
      "loss": 1.8278,
      "step": 230382
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.5303270816802979,
      "learning_rate": 1.1155132517703946e-11,
      "loss": 2.0159,
      "step": 230383
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.4263423681259155,
      "learning_rate": 1.0067507094646543e-11,
      "loss": 1.8515,
      "step": 230384
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.457606315612793,
      "learning_rate": 9.035657355660475e-12,
      "loss": 2.2217,
      "step": 230385
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.4487693309783936,
      "learning_rate": 8.05958326743905e-12,
      "loss": 2.0629,
      "step": 230386
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.4565653800964355,
      "learning_rate": 7.139284829982273e-12,
      "loss": 1.8895,
      "step": 230387
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.4681507349014282,
      "learning_rate": 6.274762043290138e-12,
      "loss": 1.9035,
      "step": 230388
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.6348810195922852,
      "learning_rate": 5.466014940669339e-12,
      "loss": 1.9922,
      "step": 230389
    },
    {
      "epoch": 3.0,
      "grad_norm": 2.3466577529907227,
      "learning_rate": 4.713043522119875e-12,
      "loss": 1.7947,
      "step": 230390
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.429481863975525,
      "learning_rate": 4.015847721028365e-12,
      "loss": 2.1445,
      "step": 230391
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.3407831192016602,
      "learning_rate": 3.3744276040081897e-12,
      "loss": 1.9267,
      "step": 230392
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.535589337348938,
      "learning_rate": 2.788783137752659e-12,
      "loss": 2.0024,
      "step": 230393
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.513513207435608,
      "learning_rate": 2.258914355568464e-12,
      "loss": 1.9094,
      "step": 230394
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.3581141233444214,
      "learning_rate": 1.7848212241489135e-12,
      "loss": 1.8309,
      "step": 230395
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.5283277034759521,
      "learning_rate": 1.3665037434940074e-12,
      "loss": 2.0235,
      "step": 230396
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.427774429321289,
      "learning_rate": 1.0039619469104366e-12,
      "loss": 2.0748,
      "step": 230397
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.4740822315216064,
      "learning_rate": 6.971958010915102e-13,
      "loss": 2.2339,
      "step": 230398
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.2813552618026733,
      "learning_rate": 4.462053060372284e-13,
      "loss": 2.0184,
      "step": 230399
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.3621835708618164,
      "learning_rate": 2.5099049505428183e-13,
      "loss": 1.7727,
      "step": 230400
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.361789345741272,
      "learning_rate": 1.1155133483597978e-13,
      "loss": 1.8687,
      "step": 230401
    },
    {
      "epoch": 3.0,
      "grad_norm": 1.4413223266601562,
      "learning_rate": 2.788782538232226e-14,
      "loss": 2.0124,
      "step": 230402
    },
    {
      "epoch": 3.0,
      "grad_norm": 2.095079183578491,
      "learning_rate": 0.0,
      "loss": 1.9259,
      "step": 230403
    },
    {
      "epoch": 3.0,
      "step": 230403,
      "total_flos": 4.245591655097303e+17,
      "train_loss": 2.164494913556694,
      "train_runtime": 21319.4605,
      "train_samples_per_second": 691.655,
      "train_steps_per_second": 10.807
    }
  ],
  "logging_steps": 1.0,
  "max_steps": 230403,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 3,
  "save_steps": 5000,
  "total_flos": 4.245591655097303e+17,
  "train_batch_size": 64,
  "trial_name": null,
  "trial_params": null
}